{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 3540,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000847457627118644,
      "grad_norm": 4.541811466217041,
      "learning_rate": 4.998587570621469e-05,
      "loss": 4.6204,
      "step": 1
    },
    {
      "epoch": 0.001694915254237288,
      "grad_norm": 2.0906004905700684,
      "learning_rate": 4.997175141242938e-05,
      "loss": 4.1783,
      "step": 2
    },
    {
      "epoch": 0.002542372881355932,
      "grad_norm": 1.9184880256652832,
      "learning_rate": 4.9957627118644066e-05,
      "loss": 4.1249,
      "step": 3
    },
    {
      "epoch": 0.003389830508474576,
      "grad_norm": 1.4485046863555908,
      "learning_rate": 4.994350282485876e-05,
      "loss": 3.9969,
      "step": 4
    },
    {
      "epoch": 0.00423728813559322,
      "grad_norm": 1.5436034202575684,
      "learning_rate": 4.992937853107345e-05,
      "loss": 4.022,
      "step": 5
    },
    {
      "epoch": 0.005084745762711864,
      "grad_norm": 1.273412823677063,
      "learning_rate": 4.991525423728814e-05,
      "loss": 3.8472,
      "step": 6
    },
    {
      "epoch": 0.005932203389830509,
      "grad_norm": 1.2345166206359863,
      "learning_rate": 4.9901129943502826e-05,
      "loss": 3.6756,
      "step": 7
    },
    {
      "epoch": 0.006779661016949152,
      "grad_norm": 1.2115682363510132,
      "learning_rate": 4.9887005649717516e-05,
      "loss": 3.5986,
      "step": 8
    },
    {
      "epoch": 0.007627118644067797,
      "grad_norm": 1.0976130962371826,
      "learning_rate": 4.9872881355932206e-05,
      "loss": 3.6337,
      "step": 9
    },
    {
      "epoch": 0.00847457627118644,
      "grad_norm": 1.1125215291976929,
      "learning_rate": 4.9858757062146896e-05,
      "loss": 3.7499,
      "step": 10
    },
    {
      "epoch": 0.009322033898305085,
      "grad_norm": 2.036208152770996,
      "learning_rate": 4.984463276836158e-05,
      "loss": 3.803,
      "step": 11
    },
    {
      "epoch": 0.010169491525423728,
      "grad_norm": 1.2083711624145508,
      "learning_rate": 4.9830508474576276e-05,
      "loss": 3.8121,
      "step": 12
    },
    {
      "epoch": 0.011016949152542373,
      "grad_norm": 1.387211561203003,
      "learning_rate": 4.9816384180790966e-05,
      "loss": 3.7796,
      "step": 13
    },
    {
      "epoch": 0.011864406779661017,
      "grad_norm": 1.193283200263977,
      "learning_rate": 4.9802259887005656e-05,
      "loss": 3.6786,
      "step": 14
    },
    {
      "epoch": 0.012711864406779662,
      "grad_norm": 1.308258295059204,
      "learning_rate": 4.978813559322034e-05,
      "loss": 3.7025,
      "step": 15
    },
    {
      "epoch": 0.013559322033898305,
      "grad_norm": 1.2283083200454712,
      "learning_rate": 4.977401129943503e-05,
      "loss": 3.6833,
      "step": 16
    },
    {
      "epoch": 0.01440677966101695,
      "grad_norm": 1.0624451637268066,
      "learning_rate": 4.975988700564972e-05,
      "loss": 3.7537,
      "step": 17
    },
    {
      "epoch": 0.015254237288135594,
      "grad_norm": 1.1808664798736572,
      "learning_rate": 4.974576271186441e-05,
      "loss": 3.6077,
      "step": 18
    },
    {
      "epoch": 0.016101694915254237,
      "grad_norm": 1.0545834302902222,
      "learning_rate": 4.97316384180791e-05,
      "loss": 3.6229,
      "step": 19
    },
    {
      "epoch": 0.01694915254237288,
      "grad_norm": 1.1053547859191895,
      "learning_rate": 4.971751412429379e-05,
      "loss": 3.6968,
      "step": 20
    },
    {
      "epoch": 0.017796610169491526,
      "grad_norm": 1.1348943710327148,
      "learning_rate": 4.970338983050848e-05,
      "loss": 3.5653,
      "step": 21
    },
    {
      "epoch": 0.01864406779661017,
      "grad_norm": 1.1071490049362183,
      "learning_rate": 4.968926553672317e-05,
      "loss": 3.8216,
      "step": 22
    },
    {
      "epoch": 0.019491525423728815,
      "grad_norm": 1.0050177574157715,
      "learning_rate": 4.967514124293786e-05,
      "loss": 3.6065,
      "step": 23
    },
    {
      "epoch": 0.020338983050847456,
      "grad_norm": 1.2216566801071167,
      "learning_rate": 4.966101694915254e-05,
      "loss": 3.7189,
      "step": 24
    },
    {
      "epoch": 0.0211864406779661,
      "grad_norm": 1.062896966934204,
      "learning_rate": 4.964689265536723e-05,
      "loss": 3.672,
      "step": 25
    },
    {
      "epoch": 0.022033898305084745,
      "grad_norm": 1.2944012880325317,
      "learning_rate": 4.963276836158192e-05,
      "loss": 3.6715,
      "step": 26
    },
    {
      "epoch": 0.02288135593220339,
      "grad_norm": 1.076984167098999,
      "learning_rate": 4.961864406779661e-05,
      "loss": 3.7157,
      "step": 27
    },
    {
      "epoch": 0.023728813559322035,
      "grad_norm": 1.0346524715423584,
      "learning_rate": 4.96045197740113e-05,
      "loss": 3.4722,
      "step": 28
    },
    {
      "epoch": 0.02457627118644068,
      "grad_norm": 1.1048908233642578,
      "learning_rate": 4.959039548022599e-05,
      "loss": 3.6395,
      "step": 29
    },
    {
      "epoch": 0.025423728813559324,
      "grad_norm": 1.0698761940002441,
      "learning_rate": 4.957627118644068e-05,
      "loss": 3.6145,
      "step": 30
    },
    {
      "epoch": 0.026271186440677965,
      "grad_norm": 1.1007237434387207,
      "learning_rate": 4.956214689265537e-05,
      "loss": 3.5166,
      "step": 31
    },
    {
      "epoch": 0.02711864406779661,
      "grad_norm": 1.0510921478271484,
      "learning_rate": 4.954802259887006e-05,
      "loss": 3.4679,
      "step": 32
    },
    {
      "epoch": 0.027966101694915254,
      "grad_norm": 0.966667652130127,
      "learning_rate": 4.9533898305084745e-05,
      "loss": 3.4887,
      "step": 33
    },
    {
      "epoch": 0.0288135593220339,
      "grad_norm": 1.2693874835968018,
      "learning_rate": 4.9519774011299435e-05,
      "loss": 3.6128,
      "step": 34
    },
    {
      "epoch": 0.029661016949152543,
      "grad_norm": 0.9852153062820435,
      "learning_rate": 4.9505649717514125e-05,
      "loss": 3.723,
      "step": 35
    },
    {
      "epoch": 0.030508474576271188,
      "grad_norm": 1.13206946849823,
      "learning_rate": 4.9491525423728815e-05,
      "loss": 3.6117,
      "step": 36
    },
    {
      "epoch": 0.03135593220338983,
      "grad_norm": 0.8744503855705261,
      "learning_rate": 4.9477401129943505e-05,
      "loss": 3.5283,
      "step": 37
    },
    {
      "epoch": 0.03220338983050847,
      "grad_norm": 1.1233553886413574,
      "learning_rate": 4.9463276836158195e-05,
      "loss": 3.5572,
      "step": 38
    },
    {
      "epoch": 0.03305084745762712,
      "grad_norm": 1.0636460781097412,
      "learning_rate": 4.9449152542372885e-05,
      "loss": 3.6352,
      "step": 39
    },
    {
      "epoch": 0.03389830508474576,
      "grad_norm": 1.1380923986434937,
      "learning_rate": 4.9435028248587575e-05,
      "loss": 3.573,
      "step": 40
    },
    {
      "epoch": 0.03474576271186441,
      "grad_norm": 1.0083677768707275,
      "learning_rate": 4.942090395480226e-05,
      "loss": 3.4665,
      "step": 41
    },
    {
      "epoch": 0.03559322033898305,
      "grad_norm": 1.374564528465271,
      "learning_rate": 4.940677966101695e-05,
      "loss": 3.6156,
      "step": 42
    },
    {
      "epoch": 0.036440677966101696,
      "grad_norm": 1.0445719957351685,
      "learning_rate": 4.939265536723164e-05,
      "loss": 3.618,
      "step": 43
    },
    {
      "epoch": 0.03728813559322034,
      "grad_norm": 0.9282047748565674,
      "learning_rate": 4.9378531073446335e-05,
      "loss": 3.5182,
      "step": 44
    },
    {
      "epoch": 0.038135593220338986,
      "grad_norm": 1.0341825485229492,
      "learning_rate": 4.936440677966102e-05,
      "loss": 3.6193,
      "step": 45
    },
    {
      "epoch": 0.03898305084745763,
      "grad_norm": 1.0397495031356812,
      "learning_rate": 4.935028248587571e-05,
      "loss": 3.573,
      "step": 46
    },
    {
      "epoch": 0.03983050847457627,
      "grad_norm": 1.06525719165802,
      "learning_rate": 4.93361581920904e-05,
      "loss": 3.6207,
      "step": 47
    },
    {
      "epoch": 0.04067796610169491,
      "grad_norm": 1.061848759651184,
      "learning_rate": 4.932203389830509e-05,
      "loss": 3.5083,
      "step": 48
    },
    {
      "epoch": 0.04152542372881356,
      "grad_norm": 0.9978002309799194,
      "learning_rate": 4.930790960451978e-05,
      "loss": 3.4151,
      "step": 49
    },
    {
      "epoch": 0.0423728813559322,
      "grad_norm": 1.023680567741394,
      "learning_rate": 4.929378531073446e-05,
      "loss": 3.5786,
      "step": 50
    },
    {
      "epoch": 0.043220338983050846,
      "grad_norm": 1.1530768871307373,
      "learning_rate": 4.927966101694915e-05,
      "loss": 3.5118,
      "step": 51
    },
    {
      "epoch": 0.04406779661016949,
      "grad_norm": 1.0338066816329956,
      "learning_rate": 4.926553672316385e-05,
      "loss": 3.7083,
      "step": 52
    },
    {
      "epoch": 0.044915254237288135,
      "grad_norm": 1.0216397047042847,
      "learning_rate": 4.925141242937854e-05,
      "loss": 3.598,
      "step": 53
    },
    {
      "epoch": 0.04576271186440678,
      "grad_norm": 0.9304412007331848,
      "learning_rate": 4.923728813559322e-05,
      "loss": 3.5175,
      "step": 54
    },
    {
      "epoch": 0.046610169491525424,
      "grad_norm": 1.0519485473632812,
      "learning_rate": 4.922316384180791e-05,
      "loss": 3.5904,
      "step": 55
    },
    {
      "epoch": 0.04745762711864407,
      "grad_norm": 0.9387502670288086,
      "learning_rate": 4.92090395480226e-05,
      "loss": 3.6034,
      "step": 56
    },
    {
      "epoch": 0.048305084745762714,
      "grad_norm": 1.0415656566619873,
      "learning_rate": 4.919491525423729e-05,
      "loss": 3.627,
      "step": 57
    },
    {
      "epoch": 0.04915254237288136,
      "grad_norm": 1.013018012046814,
      "learning_rate": 4.9180790960451975e-05,
      "loss": 3.5533,
      "step": 58
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0008492469787598,
      "learning_rate": 4.9166666666666665e-05,
      "loss": 3.5489,
      "step": 59
    },
    {
      "epoch": 0.05084745762711865,
      "grad_norm": 1.020042896270752,
      "learning_rate": 4.915254237288136e-05,
      "loss": 3.5427,
      "step": 60
    },
    {
      "epoch": 0.051694915254237285,
      "grad_norm": 1.1085624694824219,
      "learning_rate": 4.913841807909605e-05,
      "loss": 3.4387,
      "step": 61
    },
    {
      "epoch": 0.05254237288135593,
      "grad_norm": 0.9071241021156311,
      "learning_rate": 4.9124293785310735e-05,
      "loss": 3.6151,
      "step": 62
    },
    {
      "epoch": 0.053389830508474574,
      "grad_norm": 0.971471905708313,
      "learning_rate": 4.9110169491525425e-05,
      "loss": 3.4926,
      "step": 63
    },
    {
      "epoch": 0.05423728813559322,
      "grad_norm": 1.0597212314605713,
      "learning_rate": 4.9096045197740115e-05,
      "loss": 3.6694,
      "step": 64
    },
    {
      "epoch": 0.05508474576271186,
      "grad_norm": 0.8976388573646545,
      "learning_rate": 4.9081920903954805e-05,
      "loss": 3.5233,
      "step": 65
    },
    {
      "epoch": 0.05593220338983051,
      "grad_norm": 0.9068442583084106,
      "learning_rate": 4.9067796610169495e-05,
      "loss": 3.5472,
      "step": 66
    },
    {
      "epoch": 0.05677966101694915,
      "grad_norm": 0.9367517232894897,
      "learning_rate": 4.905367231638418e-05,
      "loss": 3.5259,
      "step": 67
    },
    {
      "epoch": 0.0576271186440678,
      "grad_norm": 0.8741654753684998,
      "learning_rate": 4.9039548022598875e-05,
      "loss": 3.4881,
      "step": 68
    },
    {
      "epoch": 0.05847457627118644,
      "grad_norm": 0.8584069609642029,
      "learning_rate": 4.9025423728813565e-05,
      "loss": 3.4114,
      "step": 69
    },
    {
      "epoch": 0.059322033898305086,
      "grad_norm": 0.9734240174293518,
      "learning_rate": 4.9011299435028255e-05,
      "loss": 3.5721,
      "step": 70
    },
    {
      "epoch": 0.06016949152542373,
      "grad_norm": 0.9685958027839661,
      "learning_rate": 4.899717514124294e-05,
      "loss": 3.3634,
      "step": 71
    },
    {
      "epoch": 0.061016949152542375,
      "grad_norm": 1.123534083366394,
      "learning_rate": 4.898305084745763e-05,
      "loss": 3.4039,
      "step": 72
    },
    {
      "epoch": 0.06186440677966102,
      "grad_norm": 1.012391448020935,
      "learning_rate": 4.896892655367232e-05,
      "loss": 3.439,
      "step": 73
    },
    {
      "epoch": 0.06271186440677966,
      "grad_norm": 1.3435492515563965,
      "learning_rate": 4.895480225988701e-05,
      "loss": 3.4419,
      "step": 74
    },
    {
      "epoch": 0.0635593220338983,
      "grad_norm": 1.04447603225708,
      "learning_rate": 4.89406779661017e-05,
      "loss": 3.5461,
      "step": 75
    },
    {
      "epoch": 0.06440677966101695,
      "grad_norm": 0.9301888346672058,
      "learning_rate": 4.892655367231639e-05,
      "loss": 3.5858,
      "step": 76
    },
    {
      "epoch": 0.06525423728813559,
      "grad_norm": 0.9731817841529846,
      "learning_rate": 4.891242937853108e-05,
      "loss": 3.5402,
      "step": 77
    },
    {
      "epoch": 0.06610169491525424,
      "grad_norm": 0.8086643218994141,
      "learning_rate": 4.889830508474577e-05,
      "loss": 3.4813,
      "step": 78
    },
    {
      "epoch": 0.06694915254237288,
      "grad_norm": 1.0235638618469238,
      "learning_rate": 4.888418079096045e-05,
      "loss": 3.29,
      "step": 79
    },
    {
      "epoch": 0.06779661016949153,
      "grad_norm": 0.9478564858436584,
      "learning_rate": 4.887005649717514e-05,
      "loss": 3.5304,
      "step": 80
    },
    {
      "epoch": 0.06864406779661017,
      "grad_norm": 0.9608615040779114,
      "learning_rate": 4.885593220338983e-05,
      "loss": 3.5029,
      "step": 81
    },
    {
      "epoch": 0.06949152542372881,
      "grad_norm": 0.8642784357070923,
      "learning_rate": 4.884180790960452e-05,
      "loss": 3.4545,
      "step": 82
    },
    {
      "epoch": 0.07033898305084746,
      "grad_norm": 1.0307790040969849,
      "learning_rate": 4.882768361581921e-05,
      "loss": 3.4939,
      "step": 83
    },
    {
      "epoch": 0.0711864406779661,
      "grad_norm": 1.0859419107437134,
      "learning_rate": 4.88135593220339e-05,
      "loss": 3.4883,
      "step": 84
    },
    {
      "epoch": 0.07203389830508475,
      "grad_norm": 0.946661651134491,
      "learning_rate": 4.879943502824859e-05,
      "loss": 3.4903,
      "step": 85
    },
    {
      "epoch": 0.07288135593220339,
      "grad_norm": 1.1954424381256104,
      "learning_rate": 4.878531073446328e-05,
      "loss": 3.4469,
      "step": 86
    },
    {
      "epoch": 0.07372881355932204,
      "grad_norm": 1.3650797605514526,
      "learning_rate": 4.877118644067797e-05,
      "loss": 3.5309,
      "step": 87
    },
    {
      "epoch": 0.07457627118644068,
      "grad_norm": 0.9673838019371033,
      "learning_rate": 4.8757062146892654e-05,
      "loss": 3.455,
      "step": 88
    },
    {
      "epoch": 0.07542372881355933,
      "grad_norm": 0.9562941789627075,
      "learning_rate": 4.8742937853107344e-05,
      "loss": 3.4427,
      "step": 89
    },
    {
      "epoch": 0.07627118644067797,
      "grad_norm": 1.061118721961975,
      "learning_rate": 4.8728813559322034e-05,
      "loss": 3.5553,
      "step": 90
    },
    {
      "epoch": 0.07711864406779662,
      "grad_norm": 0.9655681848526001,
      "learning_rate": 4.871468926553673e-05,
      "loss": 3.3637,
      "step": 91
    },
    {
      "epoch": 0.07796610169491526,
      "grad_norm": 1.1086546182632446,
      "learning_rate": 4.8700564971751414e-05,
      "loss": 3.5719,
      "step": 92
    },
    {
      "epoch": 0.0788135593220339,
      "grad_norm": 1.1144481897354126,
      "learning_rate": 4.8686440677966104e-05,
      "loss": 3.4608,
      "step": 93
    },
    {
      "epoch": 0.07966101694915254,
      "grad_norm": 0.9629172682762146,
      "learning_rate": 4.8672316384180794e-05,
      "loss": 3.4206,
      "step": 94
    },
    {
      "epoch": 0.08050847457627118,
      "grad_norm": 0.9343329071998596,
      "learning_rate": 4.8658192090395484e-05,
      "loss": 3.4169,
      "step": 95
    },
    {
      "epoch": 0.08135593220338982,
      "grad_norm": 1.0698659420013428,
      "learning_rate": 4.8644067796610174e-05,
      "loss": 3.359,
      "step": 96
    },
    {
      "epoch": 0.08220338983050847,
      "grad_norm": 0.9566194415092468,
      "learning_rate": 4.862994350282486e-05,
      "loss": 3.5309,
      "step": 97
    },
    {
      "epoch": 0.08305084745762711,
      "grad_norm": 1.3343651294708252,
      "learning_rate": 4.861581920903955e-05,
      "loss": 3.4451,
      "step": 98
    },
    {
      "epoch": 0.08389830508474576,
      "grad_norm": 1.4111084938049316,
      "learning_rate": 4.8601694915254244e-05,
      "loss": 3.4957,
      "step": 99
    },
    {
      "epoch": 0.0847457627118644,
      "grad_norm": 1.1460424661636353,
      "learning_rate": 4.8587570621468934e-05,
      "loss": 3.3921,
      "step": 100
    },
    {
      "epoch": 0.08559322033898305,
      "grad_norm": 1.136348843574524,
      "learning_rate": 4.857344632768362e-05,
      "loss": 3.3772,
      "step": 101
    },
    {
      "epoch": 0.08644067796610169,
      "grad_norm": 1.1128734350204468,
      "learning_rate": 4.855932203389831e-05,
      "loss": 3.5162,
      "step": 102
    },
    {
      "epoch": 0.08728813559322034,
      "grad_norm": 1.1538329124450684,
      "learning_rate": 4.8545197740113e-05,
      "loss": 3.5819,
      "step": 103
    },
    {
      "epoch": 0.08813559322033898,
      "grad_norm": 1.1180751323699951,
      "learning_rate": 4.853107344632769e-05,
      "loss": 3.5386,
      "step": 104
    },
    {
      "epoch": 0.08898305084745763,
      "grad_norm": 0.8497837781906128,
      "learning_rate": 4.851694915254237e-05,
      "loss": 3.4281,
      "step": 105
    },
    {
      "epoch": 0.08983050847457627,
      "grad_norm": 1.0639275312423706,
      "learning_rate": 4.850282485875706e-05,
      "loss": 3.4381,
      "step": 106
    },
    {
      "epoch": 0.09067796610169492,
      "grad_norm": 0.8994787335395813,
      "learning_rate": 4.848870056497176e-05,
      "loss": 3.4649,
      "step": 107
    },
    {
      "epoch": 0.09152542372881356,
      "grad_norm": 0.9433245062828064,
      "learning_rate": 4.847457627118645e-05,
      "loss": 3.4298,
      "step": 108
    },
    {
      "epoch": 0.0923728813559322,
      "grad_norm": 1.1026298999786377,
      "learning_rate": 4.846045197740113e-05,
      "loss": 3.2809,
      "step": 109
    },
    {
      "epoch": 0.09322033898305085,
      "grad_norm": 0.9991617202758789,
      "learning_rate": 4.844632768361582e-05,
      "loss": 3.4374,
      "step": 110
    },
    {
      "epoch": 0.0940677966101695,
      "grad_norm": 1.0328609943389893,
      "learning_rate": 4.843220338983051e-05,
      "loss": 3.3927,
      "step": 111
    },
    {
      "epoch": 0.09491525423728814,
      "grad_norm": 0.962786853313446,
      "learning_rate": 4.84180790960452e-05,
      "loss": 3.4113,
      "step": 112
    },
    {
      "epoch": 0.09576271186440678,
      "grad_norm": 0.8685274720191956,
      "learning_rate": 4.840395480225989e-05,
      "loss": 3.4442,
      "step": 113
    },
    {
      "epoch": 0.09661016949152543,
      "grad_norm": 1.1098908185958862,
      "learning_rate": 4.8389830508474574e-05,
      "loss": 3.2912,
      "step": 114
    },
    {
      "epoch": 0.09745762711864407,
      "grad_norm": 1.128380298614502,
      "learning_rate": 4.837570621468927e-05,
      "loss": 3.4944,
      "step": 115
    },
    {
      "epoch": 0.09830508474576272,
      "grad_norm": 1.1417268514633179,
      "learning_rate": 4.836158192090396e-05,
      "loss": 3.4952,
      "step": 116
    },
    {
      "epoch": 0.09915254237288136,
      "grad_norm": 0.9682890772819519,
      "learning_rate": 4.834745762711865e-05,
      "loss": 3.4218,
      "step": 117
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0429868698120117,
      "learning_rate": 4.8333333333333334e-05,
      "loss": 3.4454,
      "step": 118
    },
    {
      "epoch": 0.10084745762711865,
      "grad_norm": 1.1318089962005615,
      "learning_rate": 4.8319209039548024e-05,
      "loss": 3.4676,
      "step": 119
    },
    {
      "epoch": 0.1016949152542373,
      "grad_norm": 0.9262925386428833,
      "learning_rate": 4.8305084745762714e-05,
      "loss": 3.4663,
      "step": 120
    },
    {
      "epoch": 0.10254237288135593,
      "grad_norm": 0.9783544540405273,
      "learning_rate": 4.8290960451977404e-05,
      "loss": 3.4829,
      "step": 121
    },
    {
      "epoch": 0.10338983050847457,
      "grad_norm": 0.8752434253692627,
      "learning_rate": 4.8276836158192094e-05,
      "loss": 3.4232,
      "step": 122
    },
    {
      "epoch": 0.10423728813559321,
      "grad_norm": 1.132392406463623,
      "learning_rate": 4.8262711864406784e-05,
      "loss": 3.3249,
      "step": 123
    },
    {
      "epoch": 0.10508474576271186,
      "grad_norm": 1.0966801643371582,
      "learning_rate": 4.8248587570621474e-05,
      "loss": 3.5047,
      "step": 124
    },
    {
      "epoch": 0.1059322033898305,
      "grad_norm": 0.9103270173072815,
      "learning_rate": 4.8234463276836164e-05,
      "loss": 3.5133,
      "step": 125
    },
    {
      "epoch": 0.10677966101694915,
      "grad_norm": 1.3025434017181396,
      "learning_rate": 4.822033898305085e-05,
      "loss": 3.4794,
      "step": 126
    },
    {
      "epoch": 0.10762711864406779,
      "grad_norm": 0.9362355470657349,
      "learning_rate": 4.820621468926554e-05,
      "loss": 3.4954,
      "step": 127
    },
    {
      "epoch": 0.10847457627118644,
      "grad_norm": 1.0637141466140747,
      "learning_rate": 4.819209039548023e-05,
      "loss": 3.4143,
      "step": 128
    },
    {
      "epoch": 0.10932203389830508,
      "grad_norm": 1.0837198495864868,
      "learning_rate": 4.817796610169492e-05,
      "loss": 3.3429,
      "step": 129
    },
    {
      "epoch": 0.11016949152542373,
      "grad_norm": 0.9153018593788147,
      "learning_rate": 4.816384180790961e-05,
      "loss": 3.3445,
      "step": 130
    },
    {
      "epoch": 0.11101694915254237,
      "grad_norm": 1.0642338991165161,
      "learning_rate": 4.81497175141243e-05,
      "loss": 3.4297,
      "step": 131
    },
    {
      "epoch": 0.11186440677966102,
      "grad_norm": 1.0573382377624512,
      "learning_rate": 4.813559322033899e-05,
      "loss": 3.5539,
      "step": 132
    },
    {
      "epoch": 0.11271186440677966,
      "grad_norm": 0.9512803554534912,
      "learning_rate": 4.812146892655368e-05,
      "loss": 3.398,
      "step": 133
    },
    {
      "epoch": 0.1135593220338983,
      "grad_norm": 0.9490872621536255,
      "learning_rate": 4.810734463276837e-05,
      "loss": 3.3706,
      "step": 134
    },
    {
      "epoch": 0.11440677966101695,
      "grad_norm": 1.0442562103271484,
      "learning_rate": 4.809322033898305e-05,
      "loss": 3.4986,
      "step": 135
    },
    {
      "epoch": 0.1152542372881356,
      "grad_norm": 0.7907192707061768,
      "learning_rate": 4.807909604519774e-05,
      "loss": 3.4127,
      "step": 136
    },
    {
      "epoch": 0.11610169491525424,
      "grad_norm": 1.0272952318191528,
      "learning_rate": 4.806497175141243e-05,
      "loss": 3.4172,
      "step": 137
    },
    {
      "epoch": 0.11694915254237288,
      "grad_norm": 1.170906662940979,
      "learning_rate": 4.805084745762712e-05,
      "loss": 3.4011,
      "step": 138
    },
    {
      "epoch": 0.11779661016949153,
      "grad_norm": 1.0225719213485718,
      "learning_rate": 4.803672316384181e-05,
      "loss": 3.4664,
      "step": 139
    },
    {
      "epoch": 0.11864406779661017,
      "grad_norm": 0.9054561853408813,
      "learning_rate": 4.80225988700565e-05,
      "loss": 3.3756,
      "step": 140
    },
    {
      "epoch": 0.11949152542372882,
      "grad_norm": 1.0109195709228516,
      "learning_rate": 4.800847457627119e-05,
      "loss": 3.506,
      "step": 141
    },
    {
      "epoch": 0.12033898305084746,
      "grad_norm": 0.9443098902702332,
      "learning_rate": 4.799435028248588e-05,
      "loss": 3.4363,
      "step": 142
    },
    {
      "epoch": 0.1211864406779661,
      "grad_norm": 0.9755582809448242,
      "learning_rate": 4.798022598870057e-05,
      "loss": 3.3118,
      "step": 143
    },
    {
      "epoch": 0.12203389830508475,
      "grad_norm": 0.9171461462974548,
      "learning_rate": 4.796610169491525e-05,
      "loss": 3.4458,
      "step": 144
    },
    {
      "epoch": 0.1228813559322034,
      "grad_norm": 1.032393217086792,
      "learning_rate": 4.795197740112994e-05,
      "loss": 3.4287,
      "step": 145
    },
    {
      "epoch": 0.12372881355932204,
      "grad_norm": 0.9760985970497131,
      "learning_rate": 4.793785310734463e-05,
      "loss": 3.4285,
      "step": 146
    },
    {
      "epoch": 0.12457627118644068,
      "grad_norm": 0.9190145134925842,
      "learning_rate": 4.792372881355933e-05,
      "loss": 3.2856,
      "step": 147
    },
    {
      "epoch": 0.12542372881355932,
      "grad_norm": 1.0193102359771729,
      "learning_rate": 4.790960451977401e-05,
      "loss": 3.4372,
      "step": 148
    },
    {
      "epoch": 0.12627118644067797,
      "grad_norm": 1.2727913856506348,
      "learning_rate": 4.78954802259887e-05,
      "loss": 3.3763,
      "step": 149
    },
    {
      "epoch": 0.1271186440677966,
      "grad_norm": 0.9420662522315979,
      "learning_rate": 4.788135593220339e-05,
      "loss": 3.4134,
      "step": 150
    },
    {
      "epoch": 0.12796610169491526,
      "grad_norm": 1.1431869268417358,
      "learning_rate": 4.786723163841808e-05,
      "loss": 3.343,
      "step": 151
    },
    {
      "epoch": 0.1288135593220339,
      "grad_norm": 1.1734342575073242,
      "learning_rate": 4.7853107344632766e-05,
      "loss": 3.4031,
      "step": 152
    },
    {
      "epoch": 0.12966101694915255,
      "grad_norm": 0.9540769457817078,
      "learning_rate": 4.7838983050847456e-05,
      "loss": 3.4251,
      "step": 153
    },
    {
      "epoch": 0.13050847457627118,
      "grad_norm": 0.9957358241081238,
      "learning_rate": 4.7824858757062146e-05,
      "loss": 3.2844,
      "step": 154
    },
    {
      "epoch": 0.13135593220338984,
      "grad_norm": 1.0116667747497559,
      "learning_rate": 4.781073446327684e-05,
      "loss": 3.3582,
      "step": 155
    },
    {
      "epoch": 0.13220338983050847,
      "grad_norm": 0.9623947143554688,
      "learning_rate": 4.7796610169491526e-05,
      "loss": 3.3734,
      "step": 156
    },
    {
      "epoch": 0.13305084745762713,
      "grad_norm": 1.0396393537521362,
      "learning_rate": 4.7782485875706216e-05,
      "loss": 3.3273,
      "step": 157
    },
    {
      "epoch": 0.13389830508474576,
      "grad_norm": 1.0981794595718384,
      "learning_rate": 4.7768361581920906e-05,
      "loss": 3.2942,
      "step": 158
    },
    {
      "epoch": 0.13474576271186442,
      "grad_norm": 0.9261422157287598,
      "learning_rate": 4.7754237288135596e-05,
      "loss": 3.3209,
      "step": 159
    },
    {
      "epoch": 0.13559322033898305,
      "grad_norm": 0.9253855347633362,
      "learning_rate": 4.7740112994350286e-05,
      "loss": 3.3043,
      "step": 160
    },
    {
      "epoch": 0.13644067796610168,
      "grad_norm": 1.132527232170105,
      "learning_rate": 4.772598870056497e-05,
      "loss": 3.5229,
      "step": 161
    },
    {
      "epoch": 0.13728813559322034,
      "grad_norm": 0.9555791020393372,
      "learning_rate": 4.7711864406779666e-05,
      "loss": 3.5308,
      "step": 162
    },
    {
      "epoch": 0.13813559322033897,
      "grad_norm": 0.8020421266555786,
      "learning_rate": 4.7697740112994356e-05,
      "loss": 3.4104,
      "step": 163
    },
    {
      "epoch": 0.13898305084745763,
      "grad_norm": 1.0284488201141357,
      "learning_rate": 4.7683615819209046e-05,
      "loss": 3.451,
      "step": 164
    },
    {
      "epoch": 0.13983050847457626,
      "grad_norm": 1.0072463750839233,
      "learning_rate": 4.766949152542373e-05,
      "loss": 3.3837,
      "step": 165
    },
    {
      "epoch": 0.14067796610169492,
      "grad_norm": 1.0214042663574219,
      "learning_rate": 4.765536723163842e-05,
      "loss": 3.416,
      "step": 166
    },
    {
      "epoch": 0.14152542372881355,
      "grad_norm": 0.9888172745704651,
      "learning_rate": 4.764124293785311e-05,
      "loss": 3.3122,
      "step": 167
    },
    {
      "epoch": 0.1423728813559322,
      "grad_norm": 0.8845030069351196,
      "learning_rate": 4.76271186440678e-05,
      "loss": 3.4606,
      "step": 168
    },
    {
      "epoch": 0.14322033898305084,
      "grad_norm": 1.0305500030517578,
      "learning_rate": 4.761299435028248e-05,
      "loss": 3.3604,
      "step": 169
    },
    {
      "epoch": 0.1440677966101695,
      "grad_norm": 0.9892067313194275,
      "learning_rate": 4.759887005649718e-05,
      "loss": 3.3878,
      "step": 170
    },
    {
      "epoch": 0.14491525423728813,
      "grad_norm": 1.0221586227416992,
      "learning_rate": 4.758474576271187e-05,
      "loss": 3.3996,
      "step": 171
    },
    {
      "epoch": 0.14576271186440679,
      "grad_norm": 0.8211665749549866,
      "learning_rate": 4.757062146892656e-05,
      "loss": 3.3762,
      "step": 172
    },
    {
      "epoch": 0.14661016949152542,
      "grad_norm": 1.1288832426071167,
      "learning_rate": 4.755649717514124e-05,
      "loss": 3.3915,
      "step": 173
    },
    {
      "epoch": 0.14745762711864407,
      "grad_norm": 1.1231085062026978,
      "learning_rate": 4.754237288135593e-05,
      "loss": 3.4621,
      "step": 174
    },
    {
      "epoch": 0.1483050847457627,
      "grad_norm": 0.9494627714157104,
      "learning_rate": 4.752824858757062e-05,
      "loss": 3.4505,
      "step": 175
    },
    {
      "epoch": 0.14915254237288136,
      "grad_norm": 1.1615875959396362,
      "learning_rate": 4.751412429378531e-05,
      "loss": 3.4558,
      "step": 176
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9336468577384949,
      "learning_rate": 4.75e-05,
      "loss": 3.3102,
      "step": 177
    },
    {
      "epoch": 0.15084745762711865,
      "grad_norm": 0.8708992004394531,
      "learning_rate": 4.748587570621469e-05,
      "loss": 3.4432,
      "step": 178
    },
    {
      "epoch": 0.15169491525423728,
      "grad_norm": 1.1003191471099854,
      "learning_rate": 4.747175141242938e-05,
      "loss": 3.4165,
      "step": 179
    },
    {
      "epoch": 0.15254237288135594,
      "grad_norm": 0.8817538619041443,
      "learning_rate": 4.745762711864407e-05,
      "loss": 3.4338,
      "step": 180
    },
    {
      "epoch": 0.15338983050847457,
      "grad_norm": 0.9520441889762878,
      "learning_rate": 4.744350282485876e-05,
      "loss": 3.3009,
      "step": 181
    },
    {
      "epoch": 0.15423728813559323,
      "grad_norm": 1.0844974517822266,
      "learning_rate": 4.7429378531073446e-05,
      "loss": 3.2901,
      "step": 182
    },
    {
      "epoch": 0.15508474576271186,
      "grad_norm": 1.1425660848617554,
      "learning_rate": 4.7415254237288136e-05,
      "loss": 3.2906,
      "step": 183
    },
    {
      "epoch": 0.15593220338983052,
      "grad_norm": 0.9530520439147949,
      "learning_rate": 4.7401129943502826e-05,
      "loss": 3.4059,
      "step": 184
    },
    {
      "epoch": 0.15677966101694915,
      "grad_norm": 1.011949896812439,
      "learning_rate": 4.7387005649717516e-05,
      "loss": 3.2765,
      "step": 185
    },
    {
      "epoch": 0.1576271186440678,
      "grad_norm": 1.089328646659851,
      "learning_rate": 4.7372881355932206e-05,
      "loss": 3.3774,
      "step": 186
    },
    {
      "epoch": 0.15847457627118644,
      "grad_norm": 0.9004995226860046,
      "learning_rate": 4.7358757062146896e-05,
      "loss": 3.4686,
      "step": 187
    },
    {
      "epoch": 0.15932203389830507,
      "grad_norm": 0.9860526323318481,
      "learning_rate": 4.7344632768361586e-05,
      "loss": 3.3794,
      "step": 188
    },
    {
      "epoch": 0.16016949152542373,
      "grad_norm": 1.0845123529434204,
      "learning_rate": 4.7330508474576276e-05,
      "loss": 3.4398,
      "step": 189
    },
    {
      "epoch": 0.16101694915254236,
      "grad_norm": 1.10084068775177,
      "learning_rate": 4.7316384180790966e-05,
      "loss": 3.2941,
      "step": 190
    },
    {
      "epoch": 0.16186440677966102,
      "grad_norm": 1.0290236473083496,
      "learning_rate": 4.730225988700565e-05,
      "loss": 3.4847,
      "step": 191
    },
    {
      "epoch": 0.16271186440677965,
      "grad_norm": 1.2007007598876953,
      "learning_rate": 4.728813559322034e-05,
      "loss": 3.3852,
      "step": 192
    },
    {
      "epoch": 0.1635593220338983,
      "grad_norm": 0.9030916094779968,
      "learning_rate": 4.727401129943503e-05,
      "loss": 3.4596,
      "step": 193
    },
    {
      "epoch": 0.16440677966101694,
      "grad_norm": 0.9876952171325684,
      "learning_rate": 4.725988700564972e-05,
      "loss": 3.4634,
      "step": 194
    },
    {
      "epoch": 0.1652542372881356,
      "grad_norm": 1.025385856628418,
      "learning_rate": 4.724576271186441e-05,
      "loss": 3.4471,
      "step": 195
    },
    {
      "epoch": 0.16610169491525423,
      "grad_norm": 1.0539445877075195,
      "learning_rate": 4.72316384180791e-05,
      "loss": 3.4172,
      "step": 196
    },
    {
      "epoch": 0.1669491525423729,
      "grad_norm": 0.973169207572937,
      "learning_rate": 4.721751412429379e-05,
      "loss": 3.374,
      "step": 197
    },
    {
      "epoch": 0.16779661016949152,
      "grad_norm": 1.0636146068572998,
      "learning_rate": 4.720338983050848e-05,
      "loss": 3.404,
      "step": 198
    },
    {
      "epoch": 0.16864406779661018,
      "grad_norm": 0.9954953193664551,
      "learning_rate": 4.718926553672316e-05,
      "loss": 3.501,
      "step": 199
    },
    {
      "epoch": 0.1694915254237288,
      "grad_norm": 0.8904051184654236,
      "learning_rate": 4.717514124293785e-05,
      "loss": 3.4237,
      "step": 200
    },
    {
      "epoch": 0.17033898305084746,
      "grad_norm": 0.9949321746826172,
      "learning_rate": 4.716101694915254e-05,
      "loss": 3.3651,
      "step": 201
    },
    {
      "epoch": 0.1711864406779661,
      "grad_norm": 1.3412538766860962,
      "learning_rate": 4.714689265536724e-05,
      "loss": 3.2532,
      "step": 202
    },
    {
      "epoch": 0.17203389830508475,
      "grad_norm": 1.0155951976776123,
      "learning_rate": 4.713276836158192e-05,
      "loss": 3.3058,
      "step": 203
    },
    {
      "epoch": 0.17288135593220338,
      "grad_norm": 1.0480566024780273,
      "learning_rate": 4.711864406779661e-05,
      "loss": 3.3365,
      "step": 204
    },
    {
      "epoch": 0.17372881355932204,
      "grad_norm": 1.1093454360961914,
      "learning_rate": 4.71045197740113e-05,
      "loss": 3.369,
      "step": 205
    },
    {
      "epoch": 0.17457627118644067,
      "grad_norm": 1.3034790754318237,
      "learning_rate": 4.709039548022599e-05,
      "loss": 3.3571,
      "step": 206
    },
    {
      "epoch": 0.17542372881355933,
      "grad_norm": 1.0214184522628784,
      "learning_rate": 4.707627118644068e-05,
      "loss": 3.2449,
      "step": 207
    },
    {
      "epoch": 0.17627118644067796,
      "grad_norm": 1.0114665031433105,
      "learning_rate": 4.7062146892655365e-05,
      "loss": 3.2185,
      "step": 208
    },
    {
      "epoch": 0.17711864406779662,
      "grad_norm": 0.9668043851852417,
      "learning_rate": 4.7048022598870055e-05,
      "loss": 3.3632,
      "step": 209
    },
    {
      "epoch": 0.17796610169491525,
      "grad_norm": 1.1336166858673096,
      "learning_rate": 4.703389830508475e-05,
      "loss": 3.4155,
      "step": 210
    },
    {
      "epoch": 0.1788135593220339,
      "grad_norm": 1.0119339227676392,
      "learning_rate": 4.701977401129944e-05,
      "loss": 3.3101,
      "step": 211
    },
    {
      "epoch": 0.17966101694915254,
      "grad_norm": 1.012792706489563,
      "learning_rate": 4.7005649717514125e-05,
      "loss": 3.3279,
      "step": 212
    },
    {
      "epoch": 0.1805084745762712,
      "grad_norm": 0.8829046487808228,
      "learning_rate": 4.6991525423728815e-05,
      "loss": 3.2263,
      "step": 213
    },
    {
      "epoch": 0.18135593220338983,
      "grad_norm": 0.903784453868866,
      "learning_rate": 4.6977401129943505e-05,
      "loss": 3.4108,
      "step": 214
    },
    {
      "epoch": 0.18220338983050846,
      "grad_norm": 0.9798447489738464,
      "learning_rate": 4.6963276836158195e-05,
      "loss": 3.4257,
      "step": 215
    },
    {
      "epoch": 0.18305084745762712,
      "grad_norm": 1.0526540279388428,
      "learning_rate": 4.694915254237288e-05,
      "loss": 3.3412,
      "step": 216
    },
    {
      "epoch": 0.18389830508474575,
      "grad_norm": 0.9264103174209595,
      "learning_rate": 4.693502824858757e-05,
      "loss": 3.2474,
      "step": 217
    },
    {
      "epoch": 0.1847457627118644,
      "grad_norm": 1.113366961479187,
      "learning_rate": 4.6920903954802265e-05,
      "loss": 3.4456,
      "step": 218
    },
    {
      "epoch": 0.18559322033898304,
      "grad_norm": 0.9500327706336975,
      "learning_rate": 4.6906779661016955e-05,
      "loss": 3.2991,
      "step": 219
    },
    {
      "epoch": 0.1864406779661017,
      "grad_norm": 0.9833083152770996,
      "learning_rate": 4.689265536723164e-05,
      "loss": 3.3296,
      "step": 220
    },
    {
      "epoch": 0.18728813559322033,
      "grad_norm": 1.0229041576385498,
      "learning_rate": 4.687853107344633e-05,
      "loss": 3.2996,
      "step": 221
    },
    {
      "epoch": 0.188135593220339,
      "grad_norm": 1.103664517402649,
      "learning_rate": 4.686440677966102e-05,
      "loss": 3.4222,
      "step": 222
    },
    {
      "epoch": 0.18898305084745762,
      "grad_norm": 1.1817638874053955,
      "learning_rate": 4.685028248587571e-05,
      "loss": 3.5137,
      "step": 223
    },
    {
      "epoch": 0.18983050847457628,
      "grad_norm": 1.0594677925109863,
      "learning_rate": 4.68361581920904e-05,
      "loss": 3.4079,
      "step": 224
    },
    {
      "epoch": 0.1906779661016949,
      "grad_norm": 1.0142221450805664,
      "learning_rate": 4.682203389830508e-05,
      "loss": 3.2882,
      "step": 225
    },
    {
      "epoch": 0.19152542372881357,
      "grad_norm": 0.9378176927566528,
      "learning_rate": 4.680790960451978e-05,
      "loss": 3.4805,
      "step": 226
    },
    {
      "epoch": 0.1923728813559322,
      "grad_norm": 0.9688156843185425,
      "learning_rate": 4.679378531073447e-05,
      "loss": 3.3128,
      "step": 227
    },
    {
      "epoch": 0.19322033898305085,
      "grad_norm": 1.0353584289550781,
      "learning_rate": 4.677966101694916e-05,
      "loss": 3.4662,
      "step": 228
    },
    {
      "epoch": 0.19406779661016949,
      "grad_norm": 1.0622531175613403,
      "learning_rate": 4.676553672316384e-05,
      "loss": 3.3556,
      "step": 229
    },
    {
      "epoch": 0.19491525423728814,
      "grad_norm": 1.037964940071106,
      "learning_rate": 4.675141242937853e-05,
      "loss": 3.3904,
      "step": 230
    },
    {
      "epoch": 0.19576271186440677,
      "grad_norm": 1.0185190439224243,
      "learning_rate": 4.673728813559322e-05,
      "loss": 3.3667,
      "step": 231
    },
    {
      "epoch": 0.19661016949152543,
      "grad_norm": 1.0384317636489868,
      "learning_rate": 4.672316384180791e-05,
      "loss": 3.2698,
      "step": 232
    },
    {
      "epoch": 0.19745762711864406,
      "grad_norm": 0.911548912525177,
      "learning_rate": 4.67090395480226e-05,
      "loss": 3.2601,
      "step": 233
    },
    {
      "epoch": 0.19830508474576272,
      "grad_norm": 0.9575884938240051,
      "learning_rate": 4.669491525423729e-05,
      "loss": 3.3901,
      "step": 234
    },
    {
      "epoch": 0.19915254237288135,
      "grad_norm": 0.8838641047477722,
      "learning_rate": 4.668079096045198e-05,
      "loss": 3.2924,
      "step": 235
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0221738815307617,
      "learning_rate": 4.666666666666667e-05,
      "loss": 3.5573,
      "step": 236
    },
    {
      "epoch": 0.20084745762711864,
      "grad_norm": 1.0286304950714111,
      "learning_rate": 4.6652542372881355e-05,
      "loss": 3.2207,
      "step": 237
    },
    {
      "epoch": 0.2016949152542373,
      "grad_norm": 0.8968472480773926,
      "learning_rate": 4.6638418079096045e-05,
      "loss": 3.3057,
      "step": 238
    },
    {
      "epoch": 0.20254237288135593,
      "grad_norm": 0.9834404587745667,
      "learning_rate": 4.6624293785310735e-05,
      "loss": 3.3126,
      "step": 239
    },
    {
      "epoch": 0.2033898305084746,
      "grad_norm": 1.0731356143951416,
      "learning_rate": 4.6610169491525425e-05,
      "loss": 3.4379,
      "step": 240
    },
    {
      "epoch": 0.20423728813559322,
      "grad_norm": 1.158944010734558,
      "learning_rate": 4.6596045197740115e-05,
      "loss": 3.2545,
      "step": 241
    },
    {
      "epoch": 0.20508474576271185,
      "grad_norm": 0.9405023455619812,
      "learning_rate": 4.6581920903954805e-05,
      "loss": 3.4378,
      "step": 242
    },
    {
      "epoch": 0.2059322033898305,
      "grad_norm": 1.1383235454559326,
      "learning_rate": 4.6567796610169495e-05,
      "loss": 3.3256,
      "step": 243
    },
    {
      "epoch": 0.20677966101694914,
      "grad_norm": 0.8874810934066772,
      "learning_rate": 4.6553672316384185e-05,
      "loss": 3.3622,
      "step": 244
    },
    {
      "epoch": 0.2076271186440678,
      "grad_norm": 1.118872880935669,
      "learning_rate": 4.6539548022598875e-05,
      "loss": 3.3155,
      "step": 245
    },
    {
      "epoch": 0.20847457627118643,
      "grad_norm": 1.102102518081665,
      "learning_rate": 4.652542372881356e-05,
      "loss": 3.1913,
      "step": 246
    },
    {
      "epoch": 0.2093220338983051,
      "grad_norm": 1.002294659614563,
      "learning_rate": 4.651129943502825e-05,
      "loss": 3.3824,
      "step": 247
    },
    {
      "epoch": 0.21016949152542372,
      "grad_norm": 1.0566807985305786,
      "learning_rate": 4.649717514124294e-05,
      "loss": 3.4923,
      "step": 248
    },
    {
      "epoch": 0.21101694915254238,
      "grad_norm": 0.828079879283905,
      "learning_rate": 4.6483050847457635e-05,
      "loss": 3.3245,
      "step": 249
    },
    {
      "epoch": 0.211864406779661,
      "grad_norm": 1.1394438743591309,
      "learning_rate": 4.646892655367232e-05,
      "loss": 3.4789,
      "step": 250
    },
    {
      "epoch": 0.21271186440677967,
      "grad_norm": 1.1321762800216675,
      "learning_rate": 4.645480225988701e-05,
      "loss": 3.3182,
      "step": 251
    },
    {
      "epoch": 0.2135593220338983,
      "grad_norm": 1.0497584342956543,
      "learning_rate": 4.64406779661017e-05,
      "loss": 3.3439,
      "step": 252
    },
    {
      "epoch": 0.21440677966101696,
      "grad_norm": 0.9015762209892273,
      "learning_rate": 4.642655367231639e-05,
      "loss": 3.3564,
      "step": 253
    },
    {
      "epoch": 0.21525423728813559,
      "grad_norm": 0.9954678416252136,
      "learning_rate": 4.641242937853108e-05,
      "loss": 3.2391,
      "step": 254
    },
    {
      "epoch": 0.21610169491525424,
      "grad_norm": 1.0133748054504395,
      "learning_rate": 4.639830508474576e-05,
      "loss": 3.3173,
      "step": 255
    },
    {
      "epoch": 0.21694915254237288,
      "grad_norm": 1.0119907855987549,
      "learning_rate": 4.638418079096045e-05,
      "loss": 3.5523,
      "step": 256
    },
    {
      "epoch": 0.21779661016949153,
      "grad_norm": 0.9730232954025269,
      "learning_rate": 4.637005649717515e-05,
      "loss": 3.2612,
      "step": 257
    },
    {
      "epoch": 0.21864406779661016,
      "grad_norm": 1.0322071313858032,
      "learning_rate": 4.635593220338984e-05,
      "loss": 3.3631,
      "step": 258
    },
    {
      "epoch": 0.21949152542372882,
      "grad_norm": 1.0476667881011963,
      "learning_rate": 4.634180790960452e-05,
      "loss": 3.3455,
      "step": 259
    },
    {
      "epoch": 0.22033898305084745,
      "grad_norm": 1.0775266885757446,
      "learning_rate": 4.632768361581921e-05,
      "loss": 3.3354,
      "step": 260
    },
    {
      "epoch": 0.2211864406779661,
      "grad_norm": 1.1111152172088623,
      "learning_rate": 4.63135593220339e-05,
      "loss": 3.3548,
      "step": 261
    },
    {
      "epoch": 0.22203389830508474,
      "grad_norm": 0.9489743113517761,
      "learning_rate": 4.629943502824859e-05,
      "loss": 3.3328,
      "step": 262
    },
    {
      "epoch": 0.2228813559322034,
      "grad_norm": 0.9627196192741394,
      "learning_rate": 4.6285310734463274e-05,
      "loss": 3.285,
      "step": 263
    },
    {
      "epoch": 0.22372881355932203,
      "grad_norm": 1.1259305477142334,
      "learning_rate": 4.6271186440677964e-05,
      "loss": 3.434,
      "step": 264
    },
    {
      "epoch": 0.2245762711864407,
      "grad_norm": 0.8832256197929382,
      "learning_rate": 4.625706214689266e-05,
      "loss": 3.2969,
      "step": 265
    },
    {
      "epoch": 0.22542372881355932,
      "grad_norm": 0.8672728538513184,
      "learning_rate": 4.624293785310735e-05,
      "loss": 3.3588,
      "step": 266
    },
    {
      "epoch": 0.22627118644067798,
      "grad_norm": 1.0554966926574707,
      "learning_rate": 4.6228813559322034e-05,
      "loss": 3.2845,
      "step": 267
    },
    {
      "epoch": 0.2271186440677966,
      "grad_norm": 0.9938945770263672,
      "learning_rate": 4.6214689265536724e-05,
      "loss": 3.2831,
      "step": 268
    },
    {
      "epoch": 0.22796610169491524,
      "grad_norm": 1.081074833869934,
      "learning_rate": 4.6200564971751414e-05,
      "loss": 3.2885,
      "step": 269
    },
    {
      "epoch": 0.2288135593220339,
      "grad_norm": 0.8921335935592651,
      "learning_rate": 4.6186440677966104e-05,
      "loss": 3.3689,
      "step": 270
    },
    {
      "epoch": 0.22966101694915253,
      "grad_norm": 1.042938232421875,
      "learning_rate": 4.6172316384180794e-05,
      "loss": 3.2974,
      "step": 271
    },
    {
      "epoch": 0.2305084745762712,
      "grad_norm": 1.040932059288025,
      "learning_rate": 4.615819209039548e-05,
      "loss": 3.2452,
      "step": 272
    },
    {
      "epoch": 0.23135593220338982,
      "grad_norm": 1.1497408151626587,
      "learning_rate": 4.6144067796610174e-05,
      "loss": 3.4384,
      "step": 273
    },
    {
      "epoch": 0.23220338983050848,
      "grad_norm": 0.8547231554985046,
      "learning_rate": 4.6129943502824864e-05,
      "loss": 3.4982,
      "step": 274
    },
    {
      "epoch": 0.2330508474576271,
      "grad_norm": 1.0105772018432617,
      "learning_rate": 4.6115819209039554e-05,
      "loss": 3.3598,
      "step": 275
    },
    {
      "epoch": 0.23389830508474577,
      "grad_norm": 1.1239712238311768,
      "learning_rate": 4.610169491525424e-05,
      "loss": 3.2822,
      "step": 276
    },
    {
      "epoch": 0.2347457627118644,
      "grad_norm": 1.0043582916259766,
      "learning_rate": 4.608757062146893e-05,
      "loss": 3.3077,
      "step": 277
    },
    {
      "epoch": 0.23559322033898306,
      "grad_norm": 1.0133074522018433,
      "learning_rate": 4.607344632768362e-05,
      "loss": 3.3506,
      "step": 278
    },
    {
      "epoch": 0.2364406779661017,
      "grad_norm": 0.9234474301338196,
      "learning_rate": 4.605932203389831e-05,
      "loss": 3.3618,
      "step": 279
    },
    {
      "epoch": 0.23728813559322035,
      "grad_norm": 0.92128586769104,
      "learning_rate": 4.6045197740113e-05,
      "loss": 3.3541,
      "step": 280
    },
    {
      "epoch": 0.23813559322033898,
      "grad_norm": 0.9465177059173584,
      "learning_rate": 4.603107344632769e-05,
      "loss": 3.266,
      "step": 281
    },
    {
      "epoch": 0.23898305084745763,
      "grad_norm": 0.9086260199546814,
      "learning_rate": 4.601694915254238e-05,
      "loss": 3.2646,
      "step": 282
    },
    {
      "epoch": 0.23983050847457626,
      "grad_norm": 1.1903716325759888,
      "learning_rate": 4.600282485875707e-05,
      "loss": 3.2163,
      "step": 283
    },
    {
      "epoch": 0.24067796610169492,
      "grad_norm": 1.3813388347625732,
      "learning_rate": 4.598870056497175e-05,
      "loss": 3.3081,
      "step": 284
    },
    {
      "epoch": 0.24152542372881355,
      "grad_norm": 1.0255000591278076,
      "learning_rate": 4.597457627118644e-05,
      "loss": 3.2165,
      "step": 285
    },
    {
      "epoch": 0.2423728813559322,
      "grad_norm": 0.9308796525001526,
      "learning_rate": 4.596045197740113e-05,
      "loss": 3.4846,
      "step": 286
    },
    {
      "epoch": 0.24322033898305084,
      "grad_norm": 1.1603049039840698,
      "learning_rate": 4.594632768361582e-05,
      "loss": 3.4762,
      "step": 287
    },
    {
      "epoch": 0.2440677966101695,
      "grad_norm": 0.9982358813285828,
      "learning_rate": 4.593220338983051e-05,
      "loss": 3.279,
      "step": 288
    },
    {
      "epoch": 0.24491525423728813,
      "grad_norm": 0.989406406879425,
      "learning_rate": 4.59180790960452e-05,
      "loss": 3.3495,
      "step": 289
    },
    {
      "epoch": 0.2457627118644068,
      "grad_norm": 0.9485755562782288,
      "learning_rate": 4.590395480225989e-05,
      "loss": 3.4254,
      "step": 290
    },
    {
      "epoch": 0.24661016949152542,
      "grad_norm": 1.067507028579712,
      "learning_rate": 4.588983050847458e-05,
      "loss": 3.3039,
      "step": 291
    },
    {
      "epoch": 0.24745762711864408,
      "grad_norm": 0.9350253939628601,
      "learning_rate": 4.587570621468927e-05,
      "loss": 3.3643,
      "step": 292
    },
    {
      "epoch": 0.2483050847457627,
      "grad_norm": 0.9265627264976501,
      "learning_rate": 4.5861581920903954e-05,
      "loss": 3.4288,
      "step": 293
    },
    {
      "epoch": 0.24915254237288137,
      "grad_norm": 1.0432260036468506,
      "learning_rate": 4.5847457627118644e-05,
      "loss": 3.3999,
      "step": 294
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.937401294708252,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 3.287,
      "step": 295
    },
    {
      "epoch": 0.25084745762711863,
      "grad_norm": 1.0307767391204834,
      "learning_rate": 4.5819209039548024e-05,
      "loss": 3.3331,
      "step": 296
    },
    {
      "epoch": 0.25169491525423726,
      "grad_norm": 1.1131483316421509,
      "learning_rate": 4.5805084745762714e-05,
      "loss": 3.3136,
      "step": 297
    },
    {
      "epoch": 0.25254237288135595,
      "grad_norm": 0.8680022954940796,
      "learning_rate": 4.5790960451977404e-05,
      "loss": 3.3641,
      "step": 298
    },
    {
      "epoch": 0.2533898305084746,
      "grad_norm": 0.8451167345046997,
      "learning_rate": 4.5776836158192094e-05,
      "loss": 3.2599,
      "step": 299
    },
    {
      "epoch": 0.2542372881355932,
      "grad_norm": 0.9475318789482117,
      "learning_rate": 4.5762711864406784e-05,
      "loss": 3.3414,
      "step": 300
    },
    {
      "epoch": 0.25508474576271184,
      "grad_norm": 1.108223557472229,
      "learning_rate": 4.5748587570621474e-05,
      "loss": 3.4209,
      "step": 301
    },
    {
      "epoch": 0.2559322033898305,
      "grad_norm": 0.922511100769043,
      "learning_rate": 4.573446327683616e-05,
      "loss": 3.3206,
      "step": 302
    },
    {
      "epoch": 0.25677966101694916,
      "grad_norm": 1.1367084980010986,
      "learning_rate": 4.572033898305085e-05,
      "loss": 3.5687,
      "step": 303
    },
    {
      "epoch": 0.2576271186440678,
      "grad_norm": 0.91807621717453,
      "learning_rate": 4.570621468926554e-05,
      "loss": 3.3019,
      "step": 304
    },
    {
      "epoch": 0.2584745762711864,
      "grad_norm": 0.9386091828346252,
      "learning_rate": 4.5692090395480234e-05,
      "loss": 3.4212,
      "step": 305
    },
    {
      "epoch": 0.2593220338983051,
      "grad_norm": 0.9776807427406311,
      "learning_rate": 4.567796610169492e-05,
      "loss": 3.419,
      "step": 306
    },
    {
      "epoch": 0.26016949152542374,
      "grad_norm": 1.0071264505386353,
      "learning_rate": 4.566384180790961e-05,
      "loss": 3.2786,
      "step": 307
    },
    {
      "epoch": 0.26101694915254237,
      "grad_norm": 0.9951607584953308,
      "learning_rate": 4.56497175141243e-05,
      "loss": 3.2884,
      "step": 308
    },
    {
      "epoch": 0.261864406779661,
      "grad_norm": 1.0642110109329224,
      "learning_rate": 4.563559322033899e-05,
      "loss": 3.3133,
      "step": 309
    },
    {
      "epoch": 0.2627118644067797,
      "grad_norm": 0.8972973227500916,
      "learning_rate": 4.562146892655367e-05,
      "loss": 3.4772,
      "step": 310
    },
    {
      "epoch": 0.2635593220338983,
      "grad_norm": 0.9095409512519836,
      "learning_rate": 4.560734463276836e-05,
      "loss": 3.2671,
      "step": 311
    },
    {
      "epoch": 0.26440677966101694,
      "grad_norm": 0.9951375126838684,
      "learning_rate": 4.559322033898305e-05,
      "loss": 3.4345,
      "step": 312
    },
    {
      "epoch": 0.2652542372881356,
      "grad_norm": 1.0811796188354492,
      "learning_rate": 4.557909604519775e-05,
      "loss": 3.4065,
      "step": 313
    },
    {
      "epoch": 0.26610169491525426,
      "grad_norm": 0.9475049376487732,
      "learning_rate": 4.556497175141243e-05,
      "loss": 3.2354,
      "step": 314
    },
    {
      "epoch": 0.2669491525423729,
      "grad_norm": 1.0567224025726318,
      "learning_rate": 4.555084745762712e-05,
      "loss": 3.3456,
      "step": 315
    },
    {
      "epoch": 0.2677966101694915,
      "grad_norm": 1.1524204015731812,
      "learning_rate": 4.553672316384181e-05,
      "loss": 3.4103,
      "step": 316
    },
    {
      "epoch": 0.26864406779661015,
      "grad_norm": 0.9683026075363159,
      "learning_rate": 4.55225988700565e-05,
      "loss": 3.4152,
      "step": 317
    },
    {
      "epoch": 0.26949152542372884,
      "grad_norm": 1.0460152626037598,
      "learning_rate": 4.550847457627119e-05,
      "loss": 3.2732,
      "step": 318
    },
    {
      "epoch": 0.27033898305084747,
      "grad_norm": 1.1625036001205444,
      "learning_rate": 4.549435028248587e-05,
      "loss": 3.2985,
      "step": 319
    },
    {
      "epoch": 0.2711864406779661,
      "grad_norm": 1.077868938446045,
      "learning_rate": 4.548022598870056e-05,
      "loss": 3.4228,
      "step": 320
    },
    {
      "epoch": 0.27203389830508473,
      "grad_norm": 0.8281847238540649,
      "learning_rate": 4.546610169491526e-05,
      "loss": 3.3651,
      "step": 321
    },
    {
      "epoch": 0.27288135593220336,
      "grad_norm": 0.8520678281784058,
      "learning_rate": 4.545197740112995e-05,
      "loss": 3.2273,
      "step": 322
    },
    {
      "epoch": 0.27372881355932205,
      "grad_norm": 1.0431276559829712,
      "learning_rate": 4.543785310734463e-05,
      "loss": 3.3444,
      "step": 323
    },
    {
      "epoch": 0.2745762711864407,
      "grad_norm": 1.1178992986679077,
      "learning_rate": 4.542372881355932e-05,
      "loss": 3.3786,
      "step": 324
    },
    {
      "epoch": 0.2754237288135593,
      "grad_norm": 1.0613808631896973,
      "learning_rate": 4.540960451977401e-05,
      "loss": 3.3488,
      "step": 325
    },
    {
      "epoch": 0.27627118644067794,
      "grad_norm": 0.9420961737632751,
      "learning_rate": 4.53954802259887e-05,
      "loss": 3.5248,
      "step": 326
    },
    {
      "epoch": 0.2771186440677966,
      "grad_norm": 1.008429765701294,
      "learning_rate": 4.5381355932203387e-05,
      "loss": 3.3873,
      "step": 327
    },
    {
      "epoch": 0.27796610169491526,
      "grad_norm": 0.9365347623825073,
      "learning_rate": 4.536723163841808e-05,
      "loss": 3.2143,
      "step": 328
    },
    {
      "epoch": 0.2788135593220339,
      "grad_norm": 1.0193514823913574,
      "learning_rate": 4.535310734463277e-05,
      "loss": 3.2791,
      "step": 329
    },
    {
      "epoch": 0.2796610169491525,
      "grad_norm": 1.1028565168380737,
      "learning_rate": 4.533898305084746e-05,
      "loss": 3.2877,
      "step": 330
    },
    {
      "epoch": 0.2805084745762712,
      "grad_norm": 1.04025137424469,
      "learning_rate": 4.5324858757062147e-05,
      "loss": 3.2707,
      "step": 331
    },
    {
      "epoch": 0.28135593220338984,
      "grad_norm": 0.877678632736206,
      "learning_rate": 4.5310734463276837e-05,
      "loss": 3.4001,
      "step": 332
    },
    {
      "epoch": 0.28220338983050847,
      "grad_norm": 0.9502778649330139,
      "learning_rate": 4.5296610169491527e-05,
      "loss": 3.284,
      "step": 333
    },
    {
      "epoch": 0.2830508474576271,
      "grad_norm": 0.9594213962554932,
      "learning_rate": 4.5282485875706217e-05,
      "loss": 3.2292,
      "step": 334
    },
    {
      "epoch": 0.2838983050847458,
      "grad_norm": 1.030950903892517,
      "learning_rate": 4.5268361581920906e-05,
      "loss": 3.3599,
      "step": 335
    },
    {
      "epoch": 0.2847457627118644,
      "grad_norm": 0.912626326084137,
      "learning_rate": 4.5254237288135596e-05,
      "loss": 3.2828,
      "step": 336
    },
    {
      "epoch": 0.28559322033898304,
      "grad_norm": 0.9484438896179199,
      "learning_rate": 4.5240112994350286e-05,
      "loss": 3.3328,
      "step": 337
    },
    {
      "epoch": 0.2864406779661017,
      "grad_norm": 1.0654056072235107,
      "learning_rate": 4.5225988700564976e-05,
      "loss": 3.3364,
      "step": 338
    },
    {
      "epoch": 0.28728813559322036,
      "grad_norm": 1.1941118240356445,
      "learning_rate": 4.5211864406779666e-05,
      "loss": 3.4278,
      "step": 339
    },
    {
      "epoch": 0.288135593220339,
      "grad_norm": 1.1450444459915161,
      "learning_rate": 4.519774011299435e-05,
      "loss": 3.3119,
      "step": 340
    },
    {
      "epoch": 0.2889830508474576,
      "grad_norm": 0.9154060482978821,
      "learning_rate": 4.518361581920904e-05,
      "loss": 3.3388,
      "step": 341
    },
    {
      "epoch": 0.28983050847457625,
      "grad_norm": 1.1089015007019043,
      "learning_rate": 4.516949152542373e-05,
      "loss": 3.3443,
      "step": 342
    },
    {
      "epoch": 0.29067796610169494,
      "grad_norm": 1.036319375038147,
      "learning_rate": 4.515536723163842e-05,
      "loss": 3.2356,
      "step": 343
    },
    {
      "epoch": 0.29152542372881357,
      "grad_norm": 0.9845830798149109,
      "learning_rate": 4.514124293785311e-05,
      "loss": 3.4297,
      "step": 344
    },
    {
      "epoch": 0.2923728813559322,
      "grad_norm": 1.0907063484191895,
      "learning_rate": 4.51271186440678e-05,
      "loss": 3.3689,
      "step": 345
    },
    {
      "epoch": 0.29322033898305083,
      "grad_norm": 1.1132543087005615,
      "learning_rate": 4.511299435028249e-05,
      "loss": 3.3098,
      "step": 346
    },
    {
      "epoch": 0.2940677966101695,
      "grad_norm": 1.1934894323349,
      "learning_rate": 4.509887005649718e-05,
      "loss": 3.2462,
      "step": 347
    },
    {
      "epoch": 0.29491525423728815,
      "grad_norm": 1.2050788402557373,
      "learning_rate": 4.508474576271187e-05,
      "loss": 3.2446,
      "step": 348
    },
    {
      "epoch": 0.2957627118644068,
      "grad_norm": 0.9746072292327881,
      "learning_rate": 4.507062146892655e-05,
      "loss": 3.4158,
      "step": 349
    },
    {
      "epoch": 0.2966101694915254,
      "grad_norm": 1.038040041923523,
      "learning_rate": 4.505649717514124e-05,
      "loss": 3.2165,
      "step": 350
    },
    {
      "epoch": 0.29745762711864404,
      "grad_norm": 0.9746460914611816,
      "learning_rate": 4.504237288135593e-05,
      "loss": 3.3609,
      "step": 351
    },
    {
      "epoch": 0.2983050847457627,
      "grad_norm": 0.9296185970306396,
      "learning_rate": 4.502824858757062e-05,
      "loss": 3.3355,
      "step": 352
    },
    {
      "epoch": 0.29915254237288136,
      "grad_norm": 0.8227651119232178,
      "learning_rate": 4.501412429378531e-05,
      "loss": 3.442,
      "step": 353
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9187232851982117,
      "learning_rate": 4.5e-05,
      "loss": 3.3707,
      "step": 354
    },
    {
      "epoch": 0.3008474576271186,
      "grad_norm": 0.9959421157836914,
      "learning_rate": 4.498587570621469e-05,
      "loss": 3.4917,
      "step": 355
    },
    {
      "epoch": 0.3016949152542373,
      "grad_norm": 0.9714428782463074,
      "learning_rate": 4.497175141242938e-05,
      "loss": 3.325,
      "step": 356
    },
    {
      "epoch": 0.30254237288135594,
      "grad_norm": 0.9690954089164734,
      "learning_rate": 4.4957627118644066e-05,
      "loss": 3.3142,
      "step": 357
    },
    {
      "epoch": 0.30338983050847457,
      "grad_norm": 1.0296752452850342,
      "learning_rate": 4.4943502824858756e-05,
      "loss": 3.2786,
      "step": 358
    },
    {
      "epoch": 0.3042372881355932,
      "grad_norm": 1.152315616607666,
      "learning_rate": 4.4929378531073446e-05,
      "loss": 3.2337,
      "step": 359
    },
    {
      "epoch": 0.3050847457627119,
      "grad_norm": 1.0662038326263428,
      "learning_rate": 4.491525423728814e-05,
      "loss": 3.2721,
      "step": 360
    },
    {
      "epoch": 0.3059322033898305,
      "grad_norm": 0.8738830089569092,
      "learning_rate": 4.4901129943502826e-05,
      "loss": 3.3993,
      "step": 361
    },
    {
      "epoch": 0.30677966101694915,
      "grad_norm": 0.9107250571250916,
      "learning_rate": 4.4887005649717516e-05,
      "loss": 3.2791,
      "step": 362
    },
    {
      "epoch": 0.3076271186440678,
      "grad_norm": 1.0070526599884033,
      "learning_rate": 4.4872881355932206e-05,
      "loss": 3.3507,
      "step": 363
    },
    {
      "epoch": 0.30847457627118646,
      "grad_norm": 0.9358483552932739,
      "learning_rate": 4.4858757062146896e-05,
      "loss": 3.3353,
      "step": 364
    },
    {
      "epoch": 0.3093220338983051,
      "grad_norm": 0.78831547498703,
      "learning_rate": 4.4844632768361586e-05,
      "loss": 3.3122,
      "step": 365
    },
    {
      "epoch": 0.3101694915254237,
      "grad_norm": 0.9441731572151184,
      "learning_rate": 4.483050847457627e-05,
      "loss": 3.3249,
      "step": 366
    },
    {
      "epoch": 0.31101694915254235,
      "grad_norm": 0.9365278482437134,
      "learning_rate": 4.481638418079096e-05,
      "loss": 3.4133,
      "step": 367
    },
    {
      "epoch": 0.31186440677966104,
      "grad_norm": 1.0537331104278564,
      "learning_rate": 4.4802259887005656e-05,
      "loss": 3.3266,
      "step": 368
    },
    {
      "epoch": 0.31271186440677967,
      "grad_norm": 1.143284559249878,
      "learning_rate": 4.4788135593220346e-05,
      "loss": 3.3562,
      "step": 369
    },
    {
      "epoch": 0.3135593220338983,
      "grad_norm": 1.0316330194473267,
      "learning_rate": 4.477401129943503e-05,
      "loss": 3.2217,
      "step": 370
    },
    {
      "epoch": 0.31440677966101693,
      "grad_norm": 1.0107542276382446,
      "learning_rate": 4.475988700564972e-05,
      "loss": 3.3568,
      "step": 371
    },
    {
      "epoch": 0.3152542372881356,
      "grad_norm": 1.0114811658859253,
      "learning_rate": 4.474576271186441e-05,
      "loss": 3.3989,
      "step": 372
    },
    {
      "epoch": 0.31610169491525425,
      "grad_norm": 0.9530246257781982,
      "learning_rate": 4.47316384180791e-05,
      "loss": 3.3506,
      "step": 373
    },
    {
      "epoch": 0.3169491525423729,
      "grad_norm": 1.0929839611053467,
      "learning_rate": 4.471751412429378e-05,
      "loss": 3.3431,
      "step": 374
    },
    {
      "epoch": 0.3177966101694915,
      "grad_norm": 0.99641352891922,
      "learning_rate": 4.470338983050847e-05,
      "loss": 3.4992,
      "step": 375
    },
    {
      "epoch": 0.31864406779661014,
      "grad_norm": 0.8388562202453613,
      "learning_rate": 4.468926553672317e-05,
      "loss": 3.2914,
      "step": 376
    },
    {
      "epoch": 0.31949152542372883,
      "grad_norm": 0.9754483699798584,
      "learning_rate": 4.467514124293786e-05,
      "loss": 3.2311,
      "step": 377
    },
    {
      "epoch": 0.32033898305084746,
      "grad_norm": 1.01131010055542,
      "learning_rate": 4.466101694915254e-05,
      "loss": 3.3946,
      "step": 378
    },
    {
      "epoch": 0.3211864406779661,
      "grad_norm": 0.9653787016868591,
      "learning_rate": 4.464689265536723e-05,
      "loss": 3.2293,
      "step": 379
    },
    {
      "epoch": 0.3220338983050847,
      "grad_norm": 0.9971035122871399,
      "learning_rate": 4.463276836158192e-05,
      "loss": 3.3691,
      "step": 380
    },
    {
      "epoch": 0.3228813559322034,
      "grad_norm": 1.0390263795852661,
      "learning_rate": 4.461864406779661e-05,
      "loss": 3.3818,
      "step": 381
    },
    {
      "epoch": 0.32372881355932204,
      "grad_norm": 1.0146158933639526,
      "learning_rate": 4.46045197740113e-05,
      "loss": 3.285,
      "step": 382
    },
    {
      "epoch": 0.32457627118644067,
      "grad_norm": 1.0889482498168945,
      "learning_rate": 4.4590395480225986e-05,
      "loss": 3.2663,
      "step": 383
    },
    {
      "epoch": 0.3254237288135593,
      "grad_norm": 1.1688122749328613,
      "learning_rate": 4.457627118644068e-05,
      "loss": 3.2271,
      "step": 384
    },
    {
      "epoch": 0.326271186440678,
      "grad_norm": 1.1634156703948975,
      "learning_rate": 4.456214689265537e-05,
      "loss": 3.1355,
      "step": 385
    },
    {
      "epoch": 0.3271186440677966,
      "grad_norm": 0.9405133128166199,
      "learning_rate": 4.454802259887006e-05,
      "loss": 3.2457,
      "step": 386
    },
    {
      "epoch": 0.32796610169491525,
      "grad_norm": 1.0555479526519775,
      "learning_rate": 4.4533898305084746e-05,
      "loss": 3.3926,
      "step": 387
    },
    {
      "epoch": 0.3288135593220339,
      "grad_norm": 1.1605401039123535,
      "learning_rate": 4.4519774011299436e-05,
      "loss": 3.2686,
      "step": 388
    },
    {
      "epoch": 0.32966101694915256,
      "grad_norm": 1.0401630401611328,
      "learning_rate": 4.4505649717514126e-05,
      "loss": 3.1957,
      "step": 389
    },
    {
      "epoch": 0.3305084745762712,
      "grad_norm": 1.1166563034057617,
      "learning_rate": 4.4491525423728816e-05,
      "loss": 3.1175,
      "step": 390
    },
    {
      "epoch": 0.3313559322033898,
      "grad_norm": 0.9442133903503418,
      "learning_rate": 4.4477401129943506e-05,
      "loss": 3.3294,
      "step": 391
    },
    {
      "epoch": 0.33220338983050846,
      "grad_norm": 1.0875482559204102,
      "learning_rate": 4.4463276836158196e-05,
      "loss": 3.1947,
      "step": 392
    },
    {
      "epoch": 0.33305084745762714,
      "grad_norm": 1.0665291547775269,
      "learning_rate": 4.4449152542372886e-05,
      "loss": 3.3995,
      "step": 393
    },
    {
      "epoch": 0.3338983050847458,
      "grad_norm": 1.101614236831665,
      "learning_rate": 4.4435028248587575e-05,
      "loss": 3.3054,
      "step": 394
    },
    {
      "epoch": 0.3347457627118644,
      "grad_norm": 1.0335354804992676,
      "learning_rate": 4.442090395480226e-05,
      "loss": 3.3855,
      "step": 395
    },
    {
      "epoch": 0.33559322033898303,
      "grad_norm": 1.0372169017791748,
      "learning_rate": 4.440677966101695e-05,
      "loss": 3.2569,
      "step": 396
    },
    {
      "epoch": 0.3364406779661017,
      "grad_norm": 0.8847305774688721,
      "learning_rate": 4.439265536723164e-05,
      "loss": 3.3062,
      "step": 397
    },
    {
      "epoch": 0.33728813559322035,
      "grad_norm": 1.005831003189087,
      "learning_rate": 4.437853107344633e-05,
      "loss": 3.3711,
      "step": 398
    },
    {
      "epoch": 0.338135593220339,
      "grad_norm": 1.0427862405776978,
      "learning_rate": 4.436440677966102e-05,
      "loss": 3.3021,
      "step": 399
    },
    {
      "epoch": 0.3389830508474576,
      "grad_norm": 1.1072407960891724,
      "learning_rate": 4.435028248587571e-05,
      "loss": 3.2586,
      "step": 400
    },
    {
      "epoch": 0.3398305084745763,
      "grad_norm": 0.8681941032409668,
      "learning_rate": 4.43361581920904e-05,
      "loss": 3.3237,
      "step": 401
    },
    {
      "epoch": 0.34067796610169493,
      "grad_norm": 0.9611847996711731,
      "learning_rate": 4.432203389830509e-05,
      "loss": 3.3863,
      "step": 402
    },
    {
      "epoch": 0.34152542372881356,
      "grad_norm": 1.0184085369110107,
      "learning_rate": 4.430790960451978e-05,
      "loss": 3.3933,
      "step": 403
    },
    {
      "epoch": 0.3423728813559322,
      "grad_norm": 1.0703850984573364,
      "learning_rate": 4.429378531073446e-05,
      "loss": 3.2485,
      "step": 404
    },
    {
      "epoch": 0.3432203389830508,
      "grad_norm": 0.9682871103286743,
      "learning_rate": 4.427966101694915e-05,
      "loss": 3.3,
      "step": 405
    },
    {
      "epoch": 0.3440677966101695,
      "grad_norm": 1.0103580951690674,
      "learning_rate": 4.426553672316384e-05,
      "loss": 3.2532,
      "step": 406
    },
    {
      "epoch": 0.34491525423728814,
      "grad_norm": 1.1308388710021973,
      "learning_rate": 4.425141242937854e-05,
      "loss": 3.362,
      "step": 407
    },
    {
      "epoch": 0.34576271186440677,
      "grad_norm": 0.9657171368598938,
      "learning_rate": 4.423728813559322e-05,
      "loss": 3.3707,
      "step": 408
    },
    {
      "epoch": 0.3466101694915254,
      "grad_norm": 0.8440027236938477,
      "learning_rate": 4.422316384180791e-05,
      "loss": 3.3622,
      "step": 409
    },
    {
      "epoch": 0.3474576271186441,
      "grad_norm": 1.0896556377410889,
      "learning_rate": 4.42090395480226e-05,
      "loss": 3.1594,
      "step": 410
    },
    {
      "epoch": 0.3483050847457627,
      "grad_norm": 0.8792701363563538,
      "learning_rate": 4.419491525423729e-05,
      "loss": 3.4112,
      "step": 411
    },
    {
      "epoch": 0.34915254237288135,
      "grad_norm": 0.9398267269134521,
      "learning_rate": 4.418079096045198e-05,
      "loss": 3.2665,
      "step": 412
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0859825611114502,
      "learning_rate": 4.4166666666666665e-05,
      "loss": 3.2293,
      "step": 413
    },
    {
      "epoch": 0.35084745762711866,
      "grad_norm": 0.935768187046051,
      "learning_rate": 4.4152542372881355e-05,
      "loss": 3.4043,
      "step": 414
    },
    {
      "epoch": 0.3516949152542373,
      "grad_norm": 0.9599078893661499,
      "learning_rate": 4.413841807909605e-05,
      "loss": 3.2861,
      "step": 415
    },
    {
      "epoch": 0.3525423728813559,
      "grad_norm": 0.9147136211395264,
      "learning_rate": 4.412429378531074e-05,
      "loss": 3.3048,
      "step": 416
    },
    {
      "epoch": 0.35338983050847456,
      "grad_norm": 1.0079538822174072,
      "learning_rate": 4.4110169491525425e-05,
      "loss": 3.4239,
      "step": 417
    },
    {
      "epoch": 0.35423728813559324,
      "grad_norm": 0.9813336730003357,
      "learning_rate": 4.4096045197740115e-05,
      "loss": 3.1808,
      "step": 418
    },
    {
      "epoch": 0.3550847457627119,
      "grad_norm": 0.937896728515625,
      "learning_rate": 4.4081920903954805e-05,
      "loss": 3.3465,
      "step": 419
    },
    {
      "epoch": 0.3559322033898305,
      "grad_norm": 0.9621577858924866,
      "learning_rate": 4.4067796610169495e-05,
      "loss": 3.2737,
      "step": 420
    },
    {
      "epoch": 0.35677966101694913,
      "grad_norm": 1.0671390295028687,
      "learning_rate": 4.405367231638418e-05,
      "loss": 3.2426,
      "step": 421
    },
    {
      "epoch": 0.3576271186440678,
      "grad_norm": 0.9483035802841187,
      "learning_rate": 4.403954802259887e-05,
      "loss": 3.358,
      "step": 422
    },
    {
      "epoch": 0.35847457627118645,
      "grad_norm": 0.9741415977478027,
      "learning_rate": 4.4025423728813565e-05,
      "loss": 3.2336,
      "step": 423
    },
    {
      "epoch": 0.3593220338983051,
      "grad_norm": 1.000356674194336,
      "learning_rate": 4.4011299435028255e-05,
      "loss": 3.1706,
      "step": 424
    },
    {
      "epoch": 0.3601694915254237,
      "grad_norm": 0.9553226828575134,
      "learning_rate": 4.399717514124294e-05,
      "loss": 3.3855,
      "step": 425
    },
    {
      "epoch": 0.3610169491525424,
      "grad_norm": 1.0244450569152832,
      "learning_rate": 4.398305084745763e-05,
      "loss": 3.4004,
      "step": 426
    },
    {
      "epoch": 0.36186440677966103,
      "grad_norm": 0.9900097250938416,
      "learning_rate": 4.396892655367232e-05,
      "loss": 3.2182,
      "step": 427
    },
    {
      "epoch": 0.36271186440677966,
      "grad_norm": 0.9080206751823425,
      "learning_rate": 4.395480225988701e-05,
      "loss": 3.3002,
      "step": 428
    },
    {
      "epoch": 0.3635593220338983,
      "grad_norm": 1.0378522872924805,
      "learning_rate": 4.39406779661017e-05,
      "loss": 3.3241,
      "step": 429
    },
    {
      "epoch": 0.3644067796610169,
      "grad_norm": 1.0043469667434692,
      "learning_rate": 4.392655367231638e-05,
      "loss": 3.1954,
      "step": 430
    },
    {
      "epoch": 0.3652542372881356,
      "grad_norm": 0.9944254159927368,
      "learning_rate": 4.391242937853108e-05,
      "loss": 3.4255,
      "step": 431
    },
    {
      "epoch": 0.36610169491525424,
      "grad_norm": 1.0727494955062866,
      "learning_rate": 4.389830508474577e-05,
      "loss": 3.3453,
      "step": 432
    },
    {
      "epoch": 0.36694915254237287,
      "grad_norm": 0.900799036026001,
      "learning_rate": 4.388418079096046e-05,
      "loss": 3.2814,
      "step": 433
    },
    {
      "epoch": 0.3677966101694915,
      "grad_norm": 1.000331163406372,
      "learning_rate": 4.387005649717514e-05,
      "loss": 3.2186,
      "step": 434
    },
    {
      "epoch": 0.3686440677966102,
      "grad_norm": 0.9366008639335632,
      "learning_rate": 4.385593220338983e-05,
      "loss": 3.4577,
      "step": 435
    },
    {
      "epoch": 0.3694915254237288,
      "grad_norm": 1.0296732187271118,
      "learning_rate": 4.384180790960452e-05,
      "loss": 3.409,
      "step": 436
    },
    {
      "epoch": 0.37033898305084745,
      "grad_norm": 0.9627364873886108,
      "learning_rate": 4.382768361581921e-05,
      "loss": 3.2705,
      "step": 437
    },
    {
      "epoch": 0.3711864406779661,
      "grad_norm": 0.8561643362045288,
      "learning_rate": 4.38135593220339e-05,
      "loss": 3.3395,
      "step": 438
    },
    {
      "epoch": 0.37203389830508476,
      "grad_norm": 1.1192781925201416,
      "learning_rate": 4.379943502824859e-05,
      "loss": 3.2083,
      "step": 439
    },
    {
      "epoch": 0.3728813559322034,
      "grad_norm": 1.0735692977905273,
      "learning_rate": 4.378531073446328e-05,
      "loss": 3.4037,
      "step": 440
    },
    {
      "epoch": 0.373728813559322,
      "grad_norm": 0.8644734025001526,
      "learning_rate": 4.377118644067797e-05,
      "loss": 3.3224,
      "step": 441
    },
    {
      "epoch": 0.37457627118644066,
      "grad_norm": 1.0343711376190186,
      "learning_rate": 4.3757062146892655e-05,
      "loss": 3.1807,
      "step": 442
    },
    {
      "epoch": 0.37542372881355934,
      "grad_norm": 1.017646312713623,
      "learning_rate": 4.3742937853107345e-05,
      "loss": 3.2531,
      "step": 443
    },
    {
      "epoch": 0.376271186440678,
      "grad_norm": 0.9342203736305237,
      "learning_rate": 4.3728813559322035e-05,
      "loss": 3.3607,
      "step": 444
    },
    {
      "epoch": 0.3771186440677966,
      "grad_norm": 0.9267883896827698,
      "learning_rate": 4.3714689265536725e-05,
      "loss": 3.3541,
      "step": 445
    },
    {
      "epoch": 0.37796610169491524,
      "grad_norm": 0.9999634623527527,
      "learning_rate": 4.3700564971751415e-05,
      "loss": 3.229,
      "step": 446
    },
    {
      "epoch": 0.3788135593220339,
      "grad_norm": 1.0142055749893188,
      "learning_rate": 4.3686440677966105e-05,
      "loss": 3.3592,
      "step": 447
    },
    {
      "epoch": 0.37966101694915255,
      "grad_norm": 1.1882275342941284,
      "learning_rate": 4.3672316384180795e-05,
      "loss": 3.3692,
      "step": 448
    },
    {
      "epoch": 0.3805084745762712,
      "grad_norm": 1.0376838445663452,
      "learning_rate": 4.3658192090395485e-05,
      "loss": 3.3467,
      "step": 449
    },
    {
      "epoch": 0.3813559322033898,
      "grad_norm": 1.064318299293518,
      "learning_rate": 4.3644067796610175e-05,
      "loss": 3.2491,
      "step": 450
    },
    {
      "epoch": 0.3822033898305085,
      "grad_norm": 0.9614359736442566,
      "learning_rate": 4.362994350282486e-05,
      "loss": 3.3057,
      "step": 451
    },
    {
      "epoch": 0.38305084745762713,
      "grad_norm": 1.059322476387024,
      "learning_rate": 4.361581920903955e-05,
      "loss": 3.1946,
      "step": 452
    },
    {
      "epoch": 0.38389830508474576,
      "grad_norm": 0.9306525588035583,
      "learning_rate": 4.360169491525424e-05,
      "loss": 3.337,
      "step": 453
    },
    {
      "epoch": 0.3847457627118644,
      "grad_norm": 0.9853308200836182,
      "learning_rate": 4.358757062146893e-05,
      "loss": 3.2687,
      "step": 454
    },
    {
      "epoch": 0.3855932203389831,
      "grad_norm": 1.0130183696746826,
      "learning_rate": 4.357344632768362e-05,
      "loss": 3.2698,
      "step": 455
    },
    {
      "epoch": 0.3864406779661017,
      "grad_norm": 0.9569457173347473,
      "learning_rate": 4.355932203389831e-05,
      "loss": 3.2626,
      "step": 456
    },
    {
      "epoch": 0.38728813559322034,
      "grad_norm": 1.0113606452941895,
      "learning_rate": 4.3545197740113e-05,
      "loss": 3.3712,
      "step": 457
    },
    {
      "epoch": 0.38813559322033897,
      "grad_norm": 0.8372867703437805,
      "learning_rate": 4.353107344632769e-05,
      "loss": 3.2705,
      "step": 458
    },
    {
      "epoch": 0.3889830508474576,
      "grad_norm": 0.9295400977134705,
      "learning_rate": 4.351694915254238e-05,
      "loss": 3.3907,
      "step": 459
    },
    {
      "epoch": 0.3898305084745763,
      "grad_norm": 0.9511747360229492,
      "learning_rate": 4.350282485875706e-05,
      "loss": 3.2118,
      "step": 460
    },
    {
      "epoch": 0.3906779661016949,
      "grad_norm": 1.056266188621521,
      "learning_rate": 4.348870056497175e-05,
      "loss": 3.2752,
      "step": 461
    },
    {
      "epoch": 0.39152542372881355,
      "grad_norm": 0.9616141319274902,
      "learning_rate": 4.347457627118644e-05,
      "loss": 3.3691,
      "step": 462
    },
    {
      "epoch": 0.3923728813559322,
      "grad_norm": 1.032304048538208,
      "learning_rate": 4.346045197740114e-05,
      "loss": 3.2426,
      "step": 463
    },
    {
      "epoch": 0.39322033898305087,
      "grad_norm": 0.9025688767433167,
      "learning_rate": 4.344632768361582e-05,
      "loss": 3.3979,
      "step": 464
    },
    {
      "epoch": 0.3940677966101695,
      "grad_norm": 1.0538108348846436,
      "learning_rate": 4.343220338983051e-05,
      "loss": 3.3375,
      "step": 465
    },
    {
      "epoch": 0.3949152542372881,
      "grad_norm": 0.8587263226509094,
      "learning_rate": 4.34180790960452e-05,
      "loss": 3.3093,
      "step": 466
    },
    {
      "epoch": 0.39576271186440676,
      "grad_norm": 1.357358455657959,
      "learning_rate": 4.340395480225989e-05,
      "loss": 3.3067,
      "step": 467
    },
    {
      "epoch": 0.39661016949152544,
      "grad_norm": 0.884049117565155,
      "learning_rate": 4.3389830508474574e-05,
      "loss": 3.2978,
      "step": 468
    },
    {
      "epoch": 0.3974576271186441,
      "grad_norm": 1.1013493537902832,
      "learning_rate": 4.3375706214689264e-05,
      "loss": 3.3654,
      "step": 469
    },
    {
      "epoch": 0.3983050847457627,
      "grad_norm": 0.9923128485679626,
      "learning_rate": 4.3361581920903954e-05,
      "loss": 3.2856,
      "step": 470
    },
    {
      "epoch": 0.39915254237288134,
      "grad_norm": 0.9877704977989197,
      "learning_rate": 4.334745762711865e-05,
      "loss": 3.2825,
      "step": 471
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8816186189651489,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 3.3138,
      "step": 472
    },
    {
      "epoch": 0.40084745762711865,
      "grad_norm": 0.9489175081253052,
      "learning_rate": 4.3319209039548024e-05,
      "loss": 3.3202,
      "step": 473
    },
    {
      "epoch": 0.4016949152542373,
      "grad_norm": 0.9873949289321899,
      "learning_rate": 4.3305084745762714e-05,
      "loss": 3.2659,
      "step": 474
    },
    {
      "epoch": 0.4025423728813559,
      "grad_norm": 0.9659498333930969,
      "learning_rate": 4.3290960451977404e-05,
      "loss": 3.2711,
      "step": 475
    },
    {
      "epoch": 0.4033898305084746,
      "grad_norm": 1.061615228652954,
      "learning_rate": 4.3276836158192094e-05,
      "loss": 3.2285,
      "step": 476
    },
    {
      "epoch": 0.40423728813559323,
      "grad_norm": 1.028267502784729,
      "learning_rate": 4.326271186440678e-05,
      "loss": 3.3729,
      "step": 477
    },
    {
      "epoch": 0.40508474576271186,
      "grad_norm": 1.2434405088424683,
      "learning_rate": 4.324858757062147e-05,
      "loss": 3.2294,
      "step": 478
    },
    {
      "epoch": 0.4059322033898305,
      "grad_norm": 1.103103518486023,
      "learning_rate": 4.3234463276836164e-05,
      "loss": 3.3127,
      "step": 479
    },
    {
      "epoch": 0.4067796610169492,
      "grad_norm": 0.9126176834106445,
      "learning_rate": 4.3220338983050854e-05,
      "loss": 3.2937,
      "step": 480
    },
    {
      "epoch": 0.4076271186440678,
      "grad_norm": 0.9698008298873901,
      "learning_rate": 4.320621468926554e-05,
      "loss": 3.1964,
      "step": 481
    },
    {
      "epoch": 0.40847457627118644,
      "grad_norm": 0.9518531560897827,
      "learning_rate": 4.319209039548023e-05,
      "loss": 3.3216,
      "step": 482
    },
    {
      "epoch": 0.40932203389830507,
      "grad_norm": 0.9920042157173157,
      "learning_rate": 4.317796610169492e-05,
      "loss": 3.166,
      "step": 483
    },
    {
      "epoch": 0.4101694915254237,
      "grad_norm": 1.0381064414978027,
      "learning_rate": 4.316384180790961e-05,
      "loss": 3.2306,
      "step": 484
    },
    {
      "epoch": 0.4110169491525424,
      "grad_norm": 1.0730125904083252,
      "learning_rate": 4.314971751412429e-05,
      "loss": 3.2474,
      "step": 485
    },
    {
      "epoch": 0.411864406779661,
      "grad_norm": 1.008436918258667,
      "learning_rate": 4.313559322033899e-05,
      "loss": 3.4149,
      "step": 486
    },
    {
      "epoch": 0.41271186440677965,
      "grad_norm": 0.8438705205917358,
      "learning_rate": 4.312146892655368e-05,
      "loss": 3.3449,
      "step": 487
    },
    {
      "epoch": 0.4135593220338983,
      "grad_norm": 0.9985308647155762,
      "learning_rate": 4.310734463276837e-05,
      "loss": 3.2744,
      "step": 488
    },
    {
      "epoch": 0.41440677966101697,
      "grad_norm": 1.0134936571121216,
      "learning_rate": 4.309322033898305e-05,
      "loss": 3.2965,
      "step": 489
    },
    {
      "epoch": 0.4152542372881356,
      "grad_norm": 1.0140883922576904,
      "learning_rate": 4.307909604519774e-05,
      "loss": 3.4264,
      "step": 490
    },
    {
      "epoch": 0.4161016949152542,
      "grad_norm": 1.141499400138855,
      "learning_rate": 4.306497175141243e-05,
      "loss": 3.2636,
      "step": 491
    },
    {
      "epoch": 0.41694915254237286,
      "grad_norm": 1.0653141736984253,
      "learning_rate": 4.305084745762712e-05,
      "loss": 3.3115,
      "step": 492
    },
    {
      "epoch": 0.41779661016949154,
      "grad_norm": 1.101950764656067,
      "learning_rate": 4.303672316384181e-05,
      "loss": 3.3546,
      "step": 493
    },
    {
      "epoch": 0.4186440677966102,
      "grad_norm": 0.9042879939079285,
      "learning_rate": 4.30225988700565e-05,
      "loss": 3.1467,
      "step": 494
    },
    {
      "epoch": 0.4194915254237288,
      "grad_norm": 1.0648889541625977,
      "learning_rate": 4.300847457627119e-05,
      "loss": 3.2356,
      "step": 495
    },
    {
      "epoch": 0.42033898305084744,
      "grad_norm": 0.8660322427749634,
      "learning_rate": 4.299435028248588e-05,
      "loss": 3.3347,
      "step": 496
    },
    {
      "epoch": 0.4211864406779661,
      "grad_norm": 0.9183448553085327,
      "learning_rate": 4.298022598870057e-05,
      "loss": 3.2862,
      "step": 497
    },
    {
      "epoch": 0.42203389830508475,
      "grad_norm": 0.9873458743095398,
      "learning_rate": 4.2966101694915254e-05,
      "loss": 3.2141,
      "step": 498
    },
    {
      "epoch": 0.4228813559322034,
      "grad_norm": 1.1227437257766724,
      "learning_rate": 4.2951977401129944e-05,
      "loss": 3.3546,
      "step": 499
    },
    {
      "epoch": 0.423728813559322,
      "grad_norm": 0.9175024032592773,
      "learning_rate": 4.2937853107344634e-05,
      "loss": 3.3137,
      "step": 500
    },
    {
      "epoch": 0.4245762711864407,
      "grad_norm": 0.9533364772796631,
      "learning_rate": 4.2923728813559324e-05,
      "loss": 3.2952,
      "step": 501
    },
    {
      "epoch": 0.42542372881355933,
      "grad_norm": 0.9788798689842224,
      "learning_rate": 4.2909604519774014e-05,
      "loss": 3.3925,
      "step": 502
    },
    {
      "epoch": 0.42627118644067796,
      "grad_norm": 0.9586952328681946,
      "learning_rate": 4.2895480225988704e-05,
      "loss": 3.3547,
      "step": 503
    },
    {
      "epoch": 0.4271186440677966,
      "grad_norm": 1.0977470874786377,
      "learning_rate": 4.2881355932203394e-05,
      "loss": 3.4164,
      "step": 504
    },
    {
      "epoch": 0.4279661016949153,
      "grad_norm": 1.0293604135513306,
      "learning_rate": 4.2867231638418084e-05,
      "loss": 3.1522,
      "step": 505
    },
    {
      "epoch": 0.4288135593220339,
      "grad_norm": 0.9427899718284607,
      "learning_rate": 4.2853107344632774e-05,
      "loss": 3.3939,
      "step": 506
    },
    {
      "epoch": 0.42966101694915254,
      "grad_norm": 1.0391638278961182,
      "learning_rate": 4.283898305084746e-05,
      "loss": 3.3508,
      "step": 507
    },
    {
      "epoch": 0.43050847457627117,
      "grad_norm": 0.9802543520927429,
      "learning_rate": 4.282485875706215e-05,
      "loss": 3.2025,
      "step": 508
    },
    {
      "epoch": 0.43135593220338986,
      "grad_norm": 1.0067499876022339,
      "learning_rate": 4.281073446327684e-05,
      "loss": 3.2871,
      "step": 509
    },
    {
      "epoch": 0.4322033898305085,
      "grad_norm": 1.0879120826721191,
      "learning_rate": 4.279661016949153e-05,
      "loss": 3.2955,
      "step": 510
    },
    {
      "epoch": 0.4330508474576271,
      "grad_norm": 0.9421293139457703,
      "learning_rate": 4.278248587570622e-05,
      "loss": 3.3885,
      "step": 511
    },
    {
      "epoch": 0.43389830508474575,
      "grad_norm": 0.9865682721138,
      "learning_rate": 4.276836158192091e-05,
      "loss": 3.4184,
      "step": 512
    },
    {
      "epoch": 0.4347457627118644,
      "grad_norm": 1.0247126817703247,
      "learning_rate": 4.27542372881356e-05,
      "loss": 3.2416,
      "step": 513
    },
    {
      "epoch": 0.43559322033898307,
      "grad_norm": 1.006000280380249,
      "learning_rate": 4.274011299435029e-05,
      "loss": 3.3106,
      "step": 514
    },
    {
      "epoch": 0.4364406779661017,
      "grad_norm": 0.927374541759491,
      "learning_rate": 4.272598870056497e-05,
      "loss": 3.2072,
      "step": 515
    },
    {
      "epoch": 0.43728813559322033,
      "grad_norm": 0.9651941061019897,
      "learning_rate": 4.271186440677966e-05,
      "loss": 3.2353,
      "step": 516
    },
    {
      "epoch": 0.43813559322033896,
      "grad_norm": 1.0576109886169434,
      "learning_rate": 4.269774011299435e-05,
      "loss": 3.2792,
      "step": 517
    },
    {
      "epoch": 0.43898305084745765,
      "grad_norm": 0.9735087752342224,
      "learning_rate": 4.268361581920905e-05,
      "loss": 3.1281,
      "step": 518
    },
    {
      "epoch": 0.4398305084745763,
      "grad_norm": 0.8727237582206726,
      "learning_rate": 4.266949152542373e-05,
      "loss": 3.3176,
      "step": 519
    },
    {
      "epoch": 0.4406779661016949,
      "grad_norm": 1.028702735900879,
      "learning_rate": 4.265536723163842e-05,
      "loss": 3.2833,
      "step": 520
    },
    {
      "epoch": 0.44152542372881354,
      "grad_norm": 0.9425292015075684,
      "learning_rate": 4.264124293785311e-05,
      "loss": 3.2493,
      "step": 521
    },
    {
      "epoch": 0.4423728813559322,
      "grad_norm": 0.9147586226463318,
      "learning_rate": 4.26271186440678e-05,
      "loss": 3.299,
      "step": 522
    },
    {
      "epoch": 0.44322033898305085,
      "grad_norm": 0.9517231583595276,
      "learning_rate": 4.261299435028249e-05,
      "loss": 3.2144,
      "step": 523
    },
    {
      "epoch": 0.4440677966101695,
      "grad_norm": 0.8731418251991272,
      "learning_rate": 4.259887005649717e-05,
      "loss": 3.2729,
      "step": 524
    },
    {
      "epoch": 0.4449152542372881,
      "grad_norm": 1.0718739032745361,
      "learning_rate": 4.258474576271186e-05,
      "loss": 3.274,
      "step": 525
    },
    {
      "epoch": 0.4457627118644068,
      "grad_norm": 0.9271964430809021,
      "learning_rate": 4.257062146892656e-05,
      "loss": 3.1512,
      "step": 526
    },
    {
      "epoch": 0.44661016949152543,
      "grad_norm": 0.9362923502922058,
      "learning_rate": 4.255649717514125e-05,
      "loss": 3.2176,
      "step": 527
    },
    {
      "epoch": 0.44745762711864406,
      "grad_norm": 0.9278737902641296,
      "learning_rate": 4.254237288135593e-05,
      "loss": 3.3155,
      "step": 528
    },
    {
      "epoch": 0.4483050847457627,
      "grad_norm": 1.1741576194763184,
      "learning_rate": 4.252824858757062e-05,
      "loss": 3.1685,
      "step": 529
    },
    {
      "epoch": 0.4491525423728814,
      "grad_norm": 1.1526938676834106,
      "learning_rate": 4.251412429378531e-05,
      "loss": 3.2377,
      "step": 530
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8986443281173706,
      "learning_rate": 4.25e-05,
      "loss": 3.2809,
      "step": 531
    },
    {
      "epoch": 0.45084745762711864,
      "grad_norm": 1.035595417022705,
      "learning_rate": 4.2485875706214686e-05,
      "loss": 3.2069,
      "step": 532
    },
    {
      "epoch": 0.4516949152542373,
      "grad_norm": 0.9578921794891357,
      "learning_rate": 4.2471751412429376e-05,
      "loss": 3.1584,
      "step": 533
    },
    {
      "epoch": 0.45254237288135596,
      "grad_norm": 0.9431161284446716,
      "learning_rate": 4.245762711864407e-05,
      "loss": 3.3695,
      "step": 534
    },
    {
      "epoch": 0.4533898305084746,
      "grad_norm": 0.9524796605110168,
      "learning_rate": 4.244350282485876e-05,
      "loss": 3.3014,
      "step": 535
    },
    {
      "epoch": 0.4542372881355932,
      "grad_norm": 0.9229897260665894,
      "learning_rate": 4.2429378531073446e-05,
      "loss": 3.3035,
      "step": 536
    },
    {
      "epoch": 0.45508474576271185,
      "grad_norm": 1.1222385168075562,
      "learning_rate": 4.2415254237288136e-05,
      "loss": 3.1968,
      "step": 537
    },
    {
      "epoch": 0.4559322033898305,
      "grad_norm": 0.8892698884010315,
      "learning_rate": 4.2401129943502826e-05,
      "loss": 3.2539,
      "step": 538
    },
    {
      "epoch": 0.45677966101694917,
      "grad_norm": 1.1344577074050903,
      "learning_rate": 4.2387005649717516e-05,
      "loss": 3.297,
      "step": 539
    },
    {
      "epoch": 0.4576271186440678,
      "grad_norm": 1.0784862041473389,
      "learning_rate": 4.2372881355932206e-05,
      "loss": 3.3599,
      "step": 540
    },
    {
      "epoch": 0.45847457627118643,
      "grad_norm": 1.1405490636825562,
      "learning_rate": 4.235875706214689e-05,
      "loss": 3.2566,
      "step": 541
    },
    {
      "epoch": 0.45932203389830506,
      "grad_norm": 1.0826579332351685,
      "learning_rate": 4.2344632768361586e-05,
      "loss": 3.1976,
      "step": 542
    },
    {
      "epoch": 0.46016949152542375,
      "grad_norm": 0.9129642248153687,
      "learning_rate": 4.2330508474576276e-05,
      "loss": 3.1628,
      "step": 543
    },
    {
      "epoch": 0.4610169491525424,
      "grad_norm": 0.9974708557128906,
      "learning_rate": 4.2316384180790966e-05,
      "loss": 3.1971,
      "step": 544
    },
    {
      "epoch": 0.461864406779661,
      "grad_norm": 0.8488754630088806,
      "learning_rate": 4.230225988700565e-05,
      "loss": 3.2751,
      "step": 545
    },
    {
      "epoch": 0.46271186440677964,
      "grad_norm": 1.0173834562301636,
      "learning_rate": 4.228813559322034e-05,
      "loss": 3.3439,
      "step": 546
    },
    {
      "epoch": 0.4635593220338983,
      "grad_norm": 1.0113790035247803,
      "learning_rate": 4.227401129943503e-05,
      "loss": 3.2798,
      "step": 547
    },
    {
      "epoch": 0.46440677966101696,
      "grad_norm": 0.9715742468833923,
      "learning_rate": 4.225988700564972e-05,
      "loss": 3.2373,
      "step": 548
    },
    {
      "epoch": 0.4652542372881356,
      "grad_norm": 0.9641973376274109,
      "learning_rate": 4.224576271186441e-05,
      "loss": 3.2877,
      "step": 549
    },
    {
      "epoch": 0.4661016949152542,
      "grad_norm": 1.0681318044662476,
      "learning_rate": 4.22316384180791e-05,
      "loss": 3.3821,
      "step": 550
    },
    {
      "epoch": 0.4669491525423729,
      "grad_norm": 1.1120551824569702,
      "learning_rate": 4.221751412429379e-05,
      "loss": 3.2773,
      "step": 551
    },
    {
      "epoch": 0.46779661016949153,
      "grad_norm": 1.0814745426177979,
      "learning_rate": 4.220338983050848e-05,
      "loss": 3.2962,
      "step": 552
    },
    {
      "epoch": 0.46864406779661016,
      "grad_norm": 0.9989905953407288,
      "learning_rate": 4.218926553672316e-05,
      "loss": 3.2294,
      "step": 553
    },
    {
      "epoch": 0.4694915254237288,
      "grad_norm": 1.1746211051940918,
      "learning_rate": 4.217514124293785e-05,
      "loss": 3.2401,
      "step": 554
    },
    {
      "epoch": 0.4703389830508475,
      "grad_norm": 1.0154939889907837,
      "learning_rate": 4.216101694915254e-05,
      "loss": 3.1437,
      "step": 555
    },
    {
      "epoch": 0.4711864406779661,
      "grad_norm": 0.8692964315414429,
      "learning_rate": 4.214689265536723e-05,
      "loss": 3.1535,
      "step": 556
    },
    {
      "epoch": 0.47203389830508474,
      "grad_norm": 1.010939121246338,
      "learning_rate": 4.213276836158192e-05,
      "loss": 3.2304,
      "step": 557
    },
    {
      "epoch": 0.4728813559322034,
      "grad_norm": 1.1787054538726807,
      "learning_rate": 4.211864406779661e-05,
      "loss": 3.1956,
      "step": 558
    },
    {
      "epoch": 0.47372881355932206,
      "grad_norm": 1.1601383686065674,
      "learning_rate": 4.21045197740113e-05,
      "loss": 3.378,
      "step": 559
    },
    {
      "epoch": 0.4745762711864407,
      "grad_norm": 0.9629536271095276,
      "learning_rate": 4.209039548022599e-05,
      "loss": 3.1456,
      "step": 560
    },
    {
      "epoch": 0.4754237288135593,
      "grad_norm": 1.0416926145553589,
      "learning_rate": 4.207627118644068e-05,
      "loss": 3.3116,
      "step": 561
    },
    {
      "epoch": 0.47627118644067795,
      "grad_norm": 1.0895572900772095,
      "learning_rate": 4.2062146892655366e-05,
      "loss": 3.2127,
      "step": 562
    },
    {
      "epoch": 0.47711864406779664,
      "grad_norm": 0.8969883322715759,
      "learning_rate": 4.2048022598870056e-05,
      "loss": 3.2026,
      "step": 563
    },
    {
      "epoch": 0.47796610169491527,
      "grad_norm": 0.9508427381515503,
      "learning_rate": 4.2033898305084746e-05,
      "loss": 3.3177,
      "step": 564
    },
    {
      "epoch": 0.4788135593220339,
      "grad_norm": 1.079058051109314,
      "learning_rate": 4.201977401129944e-05,
      "loss": 3.2039,
      "step": 565
    },
    {
      "epoch": 0.47966101694915253,
      "grad_norm": 1.0017307996749878,
      "learning_rate": 4.2005649717514126e-05,
      "loss": 3.3683,
      "step": 566
    },
    {
      "epoch": 0.48050847457627116,
      "grad_norm": 0.9946544766426086,
      "learning_rate": 4.1991525423728816e-05,
      "loss": 3.2457,
      "step": 567
    },
    {
      "epoch": 0.48135593220338985,
      "grad_norm": 0.951226532459259,
      "learning_rate": 4.1977401129943506e-05,
      "loss": 3.2903,
      "step": 568
    },
    {
      "epoch": 0.4822033898305085,
      "grad_norm": 0.922289252281189,
      "learning_rate": 4.1963276836158196e-05,
      "loss": 3.1778,
      "step": 569
    },
    {
      "epoch": 0.4830508474576271,
      "grad_norm": 0.9599305391311646,
      "learning_rate": 4.1949152542372886e-05,
      "loss": 3.268,
      "step": 570
    },
    {
      "epoch": 0.48389830508474574,
      "grad_norm": 0.998159646987915,
      "learning_rate": 4.193502824858757e-05,
      "loss": 3.2417,
      "step": 571
    },
    {
      "epoch": 0.4847457627118644,
      "grad_norm": 1.0418500900268555,
      "learning_rate": 4.192090395480226e-05,
      "loss": 3.2949,
      "step": 572
    },
    {
      "epoch": 0.48559322033898306,
      "grad_norm": 0.8767018914222717,
      "learning_rate": 4.1906779661016956e-05,
      "loss": 3.1952,
      "step": 573
    },
    {
      "epoch": 0.4864406779661017,
      "grad_norm": 0.9565122723579407,
      "learning_rate": 4.1892655367231646e-05,
      "loss": 3.1557,
      "step": 574
    },
    {
      "epoch": 0.4872881355932203,
      "grad_norm": 0.9719352126121521,
      "learning_rate": 4.187853107344633e-05,
      "loss": 3.2688,
      "step": 575
    },
    {
      "epoch": 0.488135593220339,
      "grad_norm": 1.0732676982879639,
      "learning_rate": 4.186440677966102e-05,
      "loss": 3.1759,
      "step": 576
    },
    {
      "epoch": 0.48898305084745763,
      "grad_norm": 0.9919960498809814,
      "learning_rate": 4.185028248587571e-05,
      "loss": 3.4158,
      "step": 577
    },
    {
      "epoch": 0.48983050847457626,
      "grad_norm": 1.181933045387268,
      "learning_rate": 4.18361581920904e-05,
      "loss": 3.3148,
      "step": 578
    },
    {
      "epoch": 0.4906779661016949,
      "grad_norm": 0.9915511608123779,
      "learning_rate": 4.182203389830508e-05,
      "loss": 3.2224,
      "step": 579
    },
    {
      "epoch": 0.4915254237288136,
      "grad_norm": 1.0402354001998901,
      "learning_rate": 4.180790960451977e-05,
      "loss": 3.2054,
      "step": 580
    },
    {
      "epoch": 0.4923728813559322,
      "grad_norm": 1.0277245044708252,
      "learning_rate": 4.179378531073447e-05,
      "loss": 3.2746,
      "step": 581
    },
    {
      "epoch": 0.49322033898305084,
      "grad_norm": 0.9708535075187683,
      "learning_rate": 4.177966101694916e-05,
      "loss": 3.4521,
      "step": 582
    },
    {
      "epoch": 0.4940677966101695,
      "grad_norm": 0.9602912068367004,
      "learning_rate": 4.176553672316384e-05,
      "loss": 3.3488,
      "step": 583
    },
    {
      "epoch": 0.49491525423728816,
      "grad_norm": 0.9112564921379089,
      "learning_rate": 4.175141242937853e-05,
      "loss": 3.2176,
      "step": 584
    },
    {
      "epoch": 0.4957627118644068,
      "grad_norm": 0.8513944745063782,
      "learning_rate": 4.173728813559322e-05,
      "loss": 3.2106,
      "step": 585
    },
    {
      "epoch": 0.4966101694915254,
      "grad_norm": 1.0845856666564941,
      "learning_rate": 4.172316384180791e-05,
      "loss": 3.2922,
      "step": 586
    },
    {
      "epoch": 0.49745762711864405,
      "grad_norm": 1.0840615034103394,
      "learning_rate": 4.17090395480226e-05,
      "loss": 3.1675,
      "step": 587
    },
    {
      "epoch": 0.49830508474576274,
      "grad_norm": 1.1470903158187866,
      "learning_rate": 4.1694915254237285e-05,
      "loss": 3.289,
      "step": 588
    },
    {
      "epoch": 0.49915254237288137,
      "grad_norm": 0.9398099780082703,
      "learning_rate": 4.168079096045198e-05,
      "loss": 3.1895,
      "step": 589
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9933624863624573,
      "learning_rate": 4.166666666666667e-05,
      "loss": 3.2252,
      "step": 590
    },
    {
      "epoch": 0.5008474576271187,
      "grad_norm": 0.8540239334106445,
      "learning_rate": 4.165254237288136e-05,
      "loss": 3.2701,
      "step": 591
    },
    {
      "epoch": 0.5016949152542373,
      "grad_norm": 0.9668716192245483,
      "learning_rate": 4.1638418079096045e-05,
      "loss": 3.3911,
      "step": 592
    },
    {
      "epoch": 0.502542372881356,
      "grad_norm": 0.9678142070770264,
      "learning_rate": 4.1624293785310735e-05,
      "loss": 3.3385,
      "step": 593
    },
    {
      "epoch": 0.5033898305084745,
      "grad_norm": 1.1182767152786255,
      "learning_rate": 4.1610169491525425e-05,
      "loss": 3.2207,
      "step": 594
    },
    {
      "epoch": 0.5042372881355932,
      "grad_norm": 1.115086317062378,
      "learning_rate": 4.1596045197740115e-05,
      "loss": 3.1893,
      "step": 595
    },
    {
      "epoch": 0.5050847457627119,
      "grad_norm": 0.9647793173789978,
      "learning_rate": 4.1581920903954805e-05,
      "loss": 3.1498,
      "step": 596
    },
    {
      "epoch": 0.5059322033898305,
      "grad_norm": 1.0119106769561768,
      "learning_rate": 4.1567796610169495e-05,
      "loss": 3.1322,
      "step": 597
    },
    {
      "epoch": 0.5067796610169492,
      "grad_norm": 1.070473551750183,
      "learning_rate": 4.1553672316384185e-05,
      "loss": 3.1809,
      "step": 598
    },
    {
      "epoch": 0.5076271186440678,
      "grad_norm": 0.979231595993042,
      "learning_rate": 4.1539548022598875e-05,
      "loss": 3.294,
      "step": 599
    },
    {
      "epoch": 0.5084745762711864,
      "grad_norm": 1.0784424543380737,
      "learning_rate": 4.152542372881356e-05,
      "loss": 3.2778,
      "step": 600
    },
    {
      "epoch": 0.5093220338983051,
      "grad_norm": 0.9782651662826538,
      "learning_rate": 4.151129943502825e-05,
      "loss": 3.1707,
      "step": 601
    },
    {
      "epoch": 0.5101694915254237,
      "grad_norm": 0.9835995435714722,
      "learning_rate": 4.149717514124294e-05,
      "loss": 3.2508,
      "step": 602
    },
    {
      "epoch": 0.5110169491525424,
      "grad_norm": 1.139223337173462,
      "learning_rate": 4.148305084745763e-05,
      "loss": 3.1667,
      "step": 603
    },
    {
      "epoch": 0.511864406779661,
      "grad_norm": 0.8076462149620056,
      "learning_rate": 4.146892655367232e-05,
      "loss": 3.3504,
      "step": 604
    },
    {
      "epoch": 0.5127118644067796,
      "grad_norm": 1.087288737297058,
      "learning_rate": 4.145480225988701e-05,
      "loss": 3.2342,
      "step": 605
    },
    {
      "epoch": 0.5135593220338983,
      "grad_norm": 1.18161141872406,
      "learning_rate": 4.14406779661017e-05,
      "loss": 3.3338,
      "step": 606
    },
    {
      "epoch": 0.514406779661017,
      "grad_norm": 1.0188791751861572,
      "learning_rate": 4.142655367231639e-05,
      "loss": 3.2616,
      "step": 607
    },
    {
      "epoch": 0.5152542372881356,
      "grad_norm": 1.0371781587600708,
      "learning_rate": 4.141242937853108e-05,
      "loss": 3.3656,
      "step": 608
    },
    {
      "epoch": 0.5161016949152543,
      "grad_norm": 0.9674088954925537,
      "learning_rate": 4.139830508474576e-05,
      "loss": 3.2331,
      "step": 609
    },
    {
      "epoch": 0.5169491525423728,
      "grad_norm": 1.136169672012329,
      "learning_rate": 4.138418079096045e-05,
      "loss": 3.2559,
      "step": 610
    },
    {
      "epoch": 0.5177966101694915,
      "grad_norm": 1.034875512123108,
      "learning_rate": 4.137005649717514e-05,
      "loss": 3.2006,
      "step": 611
    },
    {
      "epoch": 0.5186440677966102,
      "grad_norm": 1.0347411632537842,
      "learning_rate": 4.135593220338983e-05,
      "loss": 3.246,
      "step": 612
    },
    {
      "epoch": 0.5194915254237288,
      "grad_norm": 1.0293282270431519,
      "learning_rate": 4.134180790960452e-05,
      "loss": 3.2248,
      "step": 613
    },
    {
      "epoch": 0.5203389830508475,
      "grad_norm": 1.0923638343811035,
      "learning_rate": 4.132768361581921e-05,
      "loss": 3.208,
      "step": 614
    },
    {
      "epoch": 0.5211864406779662,
      "grad_norm": 1.2328356504440308,
      "learning_rate": 4.13135593220339e-05,
      "loss": 3.3027,
      "step": 615
    },
    {
      "epoch": 0.5220338983050847,
      "grad_norm": 0.9866951704025269,
      "learning_rate": 4.129943502824859e-05,
      "loss": 3.1971,
      "step": 616
    },
    {
      "epoch": 0.5228813559322034,
      "grad_norm": 1.0059102773666382,
      "learning_rate": 4.128531073446328e-05,
      "loss": 3.2596,
      "step": 617
    },
    {
      "epoch": 0.523728813559322,
      "grad_norm": 1.1470650434494019,
      "learning_rate": 4.1271186440677965e-05,
      "loss": 3.3411,
      "step": 618
    },
    {
      "epoch": 0.5245762711864407,
      "grad_norm": 0.9078112244606018,
      "learning_rate": 4.1257062146892655e-05,
      "loss": 3.3085,
      "step": 619
    },
    {
      "epoch": 0.5254237288135594,
      "grad_norm": 0.8482179045677185,
      "learning_rate": 4.1242937853107345e-05,
      "loss": 3.1034,
      "step": 620
    },
    {
      "epoch": 0.5262711864406779,
      "grad_norm": 1.1044117212295532,
      "learning_rate": 4.1228813559322035e-05,
      "loss": 3.309,
      "step": 621
    },
    {
      "epoch": 0.5271186440677966,
      "grad_norm": 1.0806647539138794,
      "learning_rate": 4.1214689265536725e-05,
      "loss": 3.3362,
      "step": 622
    },
    {
      "epoch": 0.5279661016949152,
      "grad_norm": 1.0033314228057861,
      "learning_rate": 4.1200564971751415e-05,
      "loss": 3.2762,
      "step": 623
    },
    {
      "epoch": 0.5288135593220339,
      "grad_norm": 1.0104668140411377,
      "learning_rate": 4.1186440677966105e-05,
      "loss": 3.1519,
      "step": 624
    },
    {
      "epoch": 0.5296610169491526,
      "grad_norm": 0.9980913996696472,
      "learning_rate": 4.1172316384180795e-05,
      "loss": 3.2228,
      "step": 625
    },
    {
      "epoch": 0.5305084745762711,
      "grad_norm": 1.044937014579773,
      "learning_rate": 4.115819209039548e-05,
      "loss": 3.3799,
      "step": 626
    },
    {
      "epoch": 0.5313559322033898,
      "grad_norm": 0.9099481105804443,
      "learning_rate": 4.114406779661017e-05,
      "loss": 3.224,
      "step": 627
    },
    {
      "epoch": 0.5322033898305085,
      "grad_norm": 0.968677282333374,
      "learning_rate": 4.112994350282486e-05,
      "loss": 3.2132,
      "step": 628
    },
    {
      "epoch": 0.5330508474576271,
      "grad_norm": 1.0089943408966064,
      "learning_rate": 4.1115819209039555e-05,
      "loss": 3.289,
      "step": 629
    },
    {
      "epoch": 0.5338983050847458,
      "grad_norm": 1.0764185190200806,
      "learning_rate": 4.110169491525424e-05,
      "loss": 3.1789,
      "step": 630
    },
    {
      "epoch": 0.5347457627118644,
      "grad_norm": 0.9846460223197937,
      "learning_rate": 4.108757062146893e-05,
      "loss": 3.3573,
      "step": 631
    },
    {
      "epoch": 0.535593220338983,
      "grad_norm": 1.070899486541748,
      "learning_rate": 4.107344632768362e-05,
      "loss": 3.1844,
      "step": 632
    },
    {
      "epoch": 0.5364406779661017,
      "grad_norm": 0.881467878818512,
      "learning_rate": 4.105932203389831e-05,
      "loss": 3.2178,
      "step": 633
    },
    {
      "epoch": 0.5372881355932203,
      "grad_norm": 1.0086942911148071,
      "learning_rate": 4.1045197740113e-05,
      "loss": 3.2735,
      "step": 634
    },
    {
      "epoch": 0.538135593220339,
      "grad_norm": 0.9170603156089783,
      "learning_rate": 4.103107344632768e-05,
      "loss": 3.2565,
      "step": 635
    },
    {
      "epoch": 0.5389830508474577,
      "grad_norm": 0.9414057731628418,
      "learning_rate": 4.101694915254237e-05,
      "loss": 3.1911,
      "step": 636
    },
    {
      "epoch": 0.5398305084745763,
      "grad_norm": 1.075670599937439,
      "learning_rate": 4.100282485875707e-05,
      "loss": 3.3587,
      "step": 637
    },
    {
      "epoch": 0.5406779661016949,
      "grad_norm": 0.8787429928779602,
      "learning_rate": 4.098870056497176e-05,
      "loss": 3.1534,
      "step": 638
    },
    {
      "epoch": 0.5415254237288135,
      "grad_norm": 1.183467984199524,
      "learning_rate": 4.097457627118644e-05,
      "loss": 3.1506,
      "step": 639
    },
    {
      "epoch": 0.5423728813559322,
      "grad_norm": 1.0731979608535767,
      "learning_rate": 4.096045197740113e-05,
      "loss": 3.2767,
      "step": 640
    },
    {
      "epoch": 0.5432203389830509,
      "grad_norm": 0.9497907757759094,
      "learning_rate": 4.094632768361582e-05,
      "loss": 3.3184,
      "step": 641
    },
    {
      "epoch": 0.5440677966101695,
      "grad_norm": 1.0031533241271973,
      "learning_rate": 4.093220338983051e-05,
      "loss": 3.1646,
      "step": 642
    },
    {
      "epoch": 0.5449152542372881,
      "grad_norm": 0.9335317611694336,
      "learning_rate": 4.0918079096045194e-05,
      "loss": 3.3837,
      "step": 643
    },
    {
      "epoch": 0.5457627118644067,
      "grad_norm": 0.9828890562057495,
      "learning_rate": 4.090395480225989e-05,
      "loss": 3.2533,
      "step": 644
    },
    {
      "epoch": 0.5466101694915254,
      "grad_norm": 0.7858693599700928,
      "learning_rate": 4.088983050847458e-05,
      "loss": 3.2746,
      "step": 645
    },
    {
      "epoch": 0.5474576271186441,
      "grad_norm": 0.878791332244873,
      "learning_rate": 4.087570621468927e-05,
      "loss": 3.1824,
      "step": 646
    },
    {
      "epoch": 0.5483050847457627,
      "grad_norm": 0.9138060212135315,
      "learning_rate": 4.0861581920903954e-05,
      "loss": 3.2223,
      "step": 647
    },
    {
      "epoch": 0.5491525423728814,
      "grad_norm": 1.0286030769348145,
      "learning_rate": 4.0847457627118644e-05,
      "loss": 3.2213,
      "step": 648
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8754992485046387,
      "learning_rate": 4.0833333333333334e-05,
      "loss": 3.333,
      "step": 649
    },
    {
      "epoch": 0.5508474576271186,
      "grad_norm": 0.7981268167495728,
      "learning_rate": 4.0819209039548024e-05,
      "loss": 3.2108,
      "step": 650
    },
    {
      "epoch": 0.5516949152542373,
      "grad_norm": 0.8459306955337524,
      "learning_rate": 4.0805084745762714e-05,
      "loss": 3.2302,
      "step": 651
    },
    {
      "epoch": 0.5525423728813559,
      "grad_norm": 0.9788030385971069,
      "learning_rate": 4.0790960451977404e-05,
      "loss": 3.3078,
      "step": 652
    },
    {
      "epoch": 0.5533898305084746,
      "grad_norm": 0.974795937538147,
      "learning_rate": 4.0776836158192094e-05,
      "loss": 3.1427,
      "step": 653
    },
    {
      "epoch": 0.5542372881355933,
      "grad_norm": 1.1200718879699707,
      "learning_rate": 4.0762711864406784e-05,
      "loss": 3.1824,
      "step": 654
    },
    {
      "epoch": 0.5550847457627118,
      "grad_norm": 0.8478380441665649,
      "learning_rate": 4.0748587570621474e-05,
      "loss": 3.2875,
      "step": 655
    },
    {
      "epoch": 0.5559322033898305,
      "grad_norm": 0.8825022578239441,
      "learning_rate": 4.073446327683616e-05,
      "loss": 3.4266,
      "step": 656
    },
    {
      "epoch": 0.5567796610169492,
      "grad_norm": 0.8521758317947388,
      "learning_rate": 4.072033898305085e-05,
      "loss": 3.1245,
      "step": 657
    },
    {
      "epoch": 0.5576271186440678,
      "grad_norm": 1.152536392211914,
      "learning_rate": 4.070621468926554e-05,
      "loss": 3.2448,
      "step": 658
    },
    {
      "epoch": 0.5584745762711865,
      "grad_norm": 0.9221817851066589,
      "learning_rate": 4.069209039548023e-05,
      "loss": 3.3025,
      "step": 659
    },
    {
      "epoch": 0.559322033898305,
      "grad_norm": 1.1099809408187866,
      "learning_rate": 4.067796610169492e-05,
      "loss": 3.0932,
      "step": 660
    },
    {
      "epoch": 0.5601694915254237,
      "grad_norm": 0.9236830472946167,
      "learning_rate": 4.066384180790961e-05,
      "loss": 3.1852,
      "step": 661
    },
    {
      "epoch": 0.5610169491525424,
      "grad_norm": 0.9270802736282349,
      "learning_rate": 4.06497175141243e-05,
      "loss": 3.2191,
      "step": 662
    },
    {
      "epoch": 0.561864406779661,
      "grad_norm": 1.103100061416626,
      "learning_rate": 4.063559322033899e-05,
      "loss": 3.2248,
      "step": 663
    },
    {
      "epoch": 0.5627118644067797,
      "grad_norm": 0.904633641242981,
      "learning_rate": 4.062146892655368e-05,
      "loss": 3.1827,
      "step": 664
    },
    {
      "epoch": 0.5635593220338984,
      "grad_norm": 0.9881935119628906,
      "learning_rate": 4.060734463276836e-05,
      "loss": 3.359,
      "step": 665
    },
    {
      "epoch": 0.5644067796610169,
      "grad_norm": 0.8849520683288574,
      "learning_rate": 4.059322033898305e-05,
      "loss": 3.2448,
      "step": 666
    },
    {
      "epoch": 0.5652542372881356,
      "grad_norm": 1.035140037536621,
      "learning_rate": 4.057909604519774e-05,
      "loss": 3.1842,
      "step": 667
    },
    {
      "epoch": 0.5661016949152542,
      "grad_norm": 1.0417625904083252,
      "learning_rate": 4.056497175141243e-05,
      "loss": 3.2512,
      "step": 668
    },
    {
      "epoch": 0.5669491525423729,
      "grad_norm": 1.0602601766586304,
      "learning_rate": 4.055084745762712e-05,
      "loss": 3.257,
      "step": 669
    },
    {
      "epoch": 0.5677966101694916,
      "grad_norm": 1.0768911838531494,
      "learning_rate": 4.053672316384181e-05,
      "loss": 3.2235,
      "step": 670
    },
    {
      "epoch": 0.5686440677966101,
      "grad_norm": 1.012581467628479,
      "learning_rate": 4.05225988700565e-05,
      "loss": 3.1992,
      "step": 671
    },
    {
      "epoch": 0.5694915254237288,
      "grad_norm": 1.0086499452590942,
      "learning_rate": 4.050847457627119e-05,
      "loss": 3.1611,
      "step": 672
    },
    {
      "epoch": 0.5703389830508474,
      "grad_norm": 1.0682578086853027,
      "learning_rate": 4.0494350282485874e-05,
      "loss": 3.1736,
      "step": 673
    },
    {
      "epoch": 0.5711864406779661,
      "grad_norm": 0.9920092821121216,
      "learning_rate": 4.0480225988700564e-05,
      "loss": 3.3064,
      "step": 674
    },
    {
      "epoch": 0.5720338983050848,
      "grad_norm": 1.1007444858551025,
      "learning_rate": 4.0466101694915254e-05,
      "loss": 3.3154,
      "step": 675
    },
    {
      "epoch": 0.5728813559322034,
      "grad_norm": 0.971848726272583,
      "learning_rate": 4.045197740112995e-05,
      "loss": 3.2014,
      "step": 676
    },
    {
      "epoch": 0.573728813559322,
      "grad_norm": 1.0632452964782715,
      "learning_rate": 4.0437853107344634e-05,
      "loss": 3.2349,
      "step": 677
    },
    {
      "epoch": 0.5745762711864407,
      "grad_norm": 0.9419097304344177,
      "learning_rate": 4.0423728813559324e-05,
      "loss": 3.1495,
      "step": 678
    },
    {
      "epoch": 0.5754237288135593,
      "grad_norm": 0.7477332353591919,
      "learning_rate": 4.0409604519774014e-05,
      "loss": 3.2508,
      "step": 679
    },
    {
      "epoch": 0.576271186440678,
      "grad_norm": 1.1096488237380981,
      "learning_rate": 4.0395480225988704e-05,
      "loss": 3.1411,
      "step": 680
    },
    {
      "epoch": 0.5771186440677966,
      "grad_norm": 0.9235982894897461,
      "learning_rate": 4.0381355932203394e-05,
      "loss": 3.1801,
      "step": 681
    },
    {
      "epoch": 0.5779661016949152,
      "grad_norm": 1.0497263669967651,
      "learning_rate": 4.036723163841808e-05,
      "loss": 3.2461,
      "step": 682
    },
    {
      "epoch": 0.5788135593220339,
      "grad_norm": 0.974819540977478,
      "learning_rate": 4.035310734463277e-05,
      "loss": 3.1853,
      "step": 683
    },
    {
      "epoch": 0.5796610169491525,
      "grad_norm": 1.018683910369873,
      "learning_rate": 4.0338983050847464e-05,
      "loss": 3.2662,
      "step": 684
    },
    {
      "epoch": 0.5805084745762712,
      "grad_norm": 0.819183886051178,
      "learning_rate": 4.0324858757062154e-05,
      "loss": 3.2036,
      "step": 685
    },
    {
      "epoch": 0.5813559322033899,
      "grad_norm": 1.2149080038070679,
      "learning_rate": 4.031073446327684e-05,
      "loss": 3.1061,
      "step": 686
    },
    {
      "epoch": 0.5822033898305085,
      "grad_norm": 0.9971520304679871,
      "learning_rate": 4.029661016949153e-05,
      "loss": 3.2792,
      "step": 687
    },
    {
      "epoch": 0.5830508474576271,
      "grad_norm": 1.1348650455474854,
      "learning_rate": 4.028248587570622e-05,
      "loss": 3.0649,
      "step": 688
    },
    {
      "epoch": 0.5838983050847457,
      "grad_norm": 0.966973066329956,
      "learning_rate": 4.026836158192091e-05,
      "loss": 3.2352,
      "step": 689
    },
    {
      "epoch": 0.5847457627118644,
      "grad_norm": 1.069762945175171,
      "learning_rate": 4.025423728813559e-05,
      "loss": 3.2476,
      "step": 690
    },
    {
      "epoch": 0.5855932203389831,
      "grad_norm": 1.018039345741272,
      "learning_rate": 4.024011299435028e-05,
      "loss": 3.3637,
      "step": 691
    },
    {
      "epoch": 0.5864406779661017,
      "grad_norm": 1.0465275049209595,
      "learning_rate": 4.022598870056498e-05,
      "loss": 3.2668,
      "step": 692
    },
    {
      "epoch": 0.5872881355932204,
      "grad_norm": 0.8877243995666504,
      "learning_rate": 4.021186440677967e-05,
      "loss": 3.1696,
      "step": 693
    },
    {
      "epoch": 0.588135593220339,
      "grad_norm": 0.8796149492263794,
      "learning_rate": 4.019774011299435e-05,
      "loss": 3.28,
      "step": 694
    },
    {
      "epoch": 0.5889830508474576,
      "grad_norm": 1.0082428455352783,
      "learning_rate": 4.018361581920904e-05,
      "loss": 3.2442,
      "step": 695
    },
    {
      "epoch": 0.5898305084745763,
      "grad_norm": 0.9520376324653625,
      "learning_rate": 4.016949152542373e-05,
      "loss": 3.2093,
      "step": 696
    },
    {
      "epoch": 0.5906779661016949,
      "grad_norm": 1.0042046308517456,
      "learning_rate": 4.015536723163842e-05,
      "loss": 3.196,
      "step": 697
    },
    {
      "epoch": 0.5915254237288136,
      "grad_norm": 0.9781447649002075,
      "learning_rate": 4.014124293785311e-05,
      "loss": 3.2572,
      "step": 698
    },
    {
      "epoch": 0.5923728813559322,
      "grad_norm": 1.0865386724472046,
      "learning_rate": 4.012711864406779e-05,
      "loss": 3.3572,
      "step": 699
    },
    {
      "epoch": 0.5932203389830508,
      "grad_norm": 0.9630225896835327,
      "learning_rate": 4.011299435028249e-05,
      "loss": 3.2831,
      "step": 700
    },
    {
      "epoch": 0.5940677966101695,
      "grad_norm": 1.0484930276870728,
      "learning_rate": 4.009887005649718e-05,
      "loss": 3.2598,
      "step": 701
    },
    {
      "epoch": 0.5949152542372881,
      "grad_norm": 0.8625562191009521,
      "learning_rate": 4.008474576271187e-05,
      "loss": 3.1784,
      "step": 702
    },
    {
      "epoch": 0.5957627118644068,
      "grad_norm": 1.0610886812210083,
      "learning_rate": 4.007062146892655e-05,
      "loss": 3.115,
      "step": 703
    },
    {
      "epoch": 0.5966101694915255,
      "grad_norm": 1.0813184976577759,
      "learning_rate": 4.005649717514124e-05,
      "loss": 3.2275,
      "step": 704
    },
    {
      "epoch": 0.597457627118644,
      "grad_norm": 0.9595462083816528,
      "learning_rate": 4.004237288135593e-05,
      "loss": 3.2497,
      "step": 705
    },
    {
      "epoch": 0.5983050847457627,
      "grad_norm": 0.8793260455131531,
      "learning_rate": 4.002824858757062e-05,
      "loss": 3.1899,
      "step": 706
    },
    {
      "epoch": 0.5991525423728814,
      "grad_norm": 0.9518945813179016,
      "learning_rate": 4.001412429378531e-05,
      "loss": 3.2939,
      "step": 707
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9193763136863708,
      "learning_rate": 4e-05,
      "loss": 3.2363,
      "step": 708
    },
    {
      "epoch": 0.6008474576271187,
      "grad_norm": 0.970078706741333,
      "learning_rate": 3.998587570621469e-05,
      "loss": 3.166,
      "step": 709
    },
    {
      "epoch": 0.6016949152542372,
      "grad_norm": 1.0736448764801025,
      "learning_rate": 3.997175141242938e-05,
      "loss": 3.2552,
      "step": 710
    },
    {
      "epoch": 0.6025423728813559,
      "grad_norm": 1.0440733432769775,
      "learning_rate": 3.9957627118644066e-05,
      "loss": 3.0961,
      "step": 711
    },
    {
      "epoch": 0.6033898305084746,
      "grad_norm": 0.8885401487350464,
      "learning_rate": 3.9943502824858756e-05,
      "loss": 3.1327,
      "step": 712
    },
    {
      "epoch": 0.6042372881355932,
      "grad_norm": 1.0647501945495605,
      "learning_rate": 3.9929378531073446e-05,
      "loss": 3.1719,
      "step": 713
    },
    {
      "epoch": 0.6050847457627119,
      "grad_norm": 0.8892009854316711,
      "learning_rate": 3.9915254237288136e-05,
      "loss": 3.2557,
      "step": 714
    },
    {
      "epoch": 0.6059322033898306,
      "grad_norm": 1.0955719947814941,
      "learning_rate": 3.9901129943502826e-05,
      "loss": 3.2905,
      "step": 715
    },
    {
      "epoch": 0.6067796610169491,
      "grad_norm": 0.9575005769729614,
      "learning_rate": 3.9887005649717516e-05,
      "loss": 3.1437,
      "step": 716
    },
    {
      "epoch": 0.6076271186440678,
      "grad_norm": 0.8964182138442993,
      "learning_rate": 3.9872881355932206e-05,
      "loss": 3.1041,
      "step": 717
    },
    {
      "epoch": 0.6084745762711864,
      "grad_norm": 1.0153088569641113,
      "learning_rate": 3.9858757062146896e-05,
      "loss": 3.2048,
      "step": 718
    },
    {
      "epoch": 0.6093220338983051,
      "grad_norm": 1.0228848457336426,
      "learning_rate": 3.9844632768361586e-05,
      "loss": 3.2921,
      "step": 719
    },
    {
      "epoch": 0.6101694915254238,
      "grad_norm": 0.8641826510429382,
      "learning_rate": 3.983050847457627e-05,
      "loss": 3.238,
      "step": 720
    },
    {
      "epoch": 0.6110169491525423,
      "grad_norm": 1.1096068620681763,
      "learning_rate": 3.981638418079096e-05,
      "loss": 3.2023,
      "step": 721
    },
    {
      "epoch": 0.611864406779661,
      "grad_norm": 1.0792356729507446,
      "learning_rate": 3.980225988700565e-05,
      "loss": 3.1953,
      "step": 722
    },
    {
      "epoch": 0.6127118644067797,
      "grad_norm": 1.0577806234359741,
      "learning_rate": 3.978813559322034e-05,
      "loss": 3.2092,
      "step": 723
    },
    {
      "epoch": 0.6135593220338983,
      "grad_norm": 0.9147716760635376,
      "learning_rate": 3.977401129943503e-05,
      "loss": 3.2009,
      "step": 724
    },
    {
      "epoch": 0.614406779661017,
      "grad_norm": 0.8805369734764099,
      "learning_rate": 3.975988700564972e-05,
      "loss": 3.4287,
      "step": 725
    },
    {
      "epoch": 0.6152542372881356,
      "grad_norm": 1.0842047929763794,
      "learning_rate": 3.974576271186441e-05,
      "loss": 3.2973,
      "step": 726
    },
    {
      "epoch": 0.6161016949152542,
      "grad_norm": 1.020552158355713,
      "learning_rate": 3.97316384180791e-05,
      "loss": 3.2854,
      "step": 727
    },
    {
      "epoch": 0.6169491525423729,
      "grad_norm": 0.9600501656532288,
      "learning_rate": 3.971751412429379e-05,
      "loss": 3.2243,
      "step": 728
    },
    {
      "epoch": 0.6177966101694915,
      "grad_norm": 1.0542972087860107,
      "learning_rate": 3.970338983050847e-05,
      "loss": 3.2449,
      "step": 729
    },
    {
      "epoch": 0.6186440677966102,
      "grad_norm": 0.8552963137626648,
      "learning_rate": 3.968926553672316e-05,
      "loss": 3.2027,
      "step": 730
    },
    {
      "epoch": 0.6194915254237288,
      "grad_norm": 0.9919419288635254,
      "learning_rate": 3.967514124293786e-05,
      "loss": 3.211,
      "step": 731
    },
    {
      "epoch": 0.6203389830508474,
      "grad_norm": 0.928841233253479,
      "learning_rate": 3.966101694915255e-05,
      "loss": 3.4355,
      "step": 732
    },
    {
      "epoch": 0.6211864406779661,
      "grad_norm": 0.8842092156410217,
      "learning_rate": 3.964689265536723e-05,
      "loss": 3.1617,
      "step": 733
    },
    {
      "epoch": 0.6220338983050847,
      "grad_norm": 0.9310187697410583,
      "learning_rate": 3.963276836158192e-05,
      "loss": 3.2295,
      "step": 734
    },
    {
      "epoch": 0.6228813559322034,
      "grad_norm": 0.9824132919311523,
      "learning_rate": 3.961864406779661e-05,
      "loss": 3.3125,
      "step": 735
    },
    {
      "epoch": 0.6237288135593221,
      "grad_norm": 1.0622237920761108,
      "learning_rate": 3.96045197740113e-05,
      "loss": 3.2571,
      "step": 736
    },
    {
      "epoch": 0.6245762711864407,
      "grad_norm": 0.8373996615409851,
      "learning_rate": 3.9590395480225986e-05,
      "loss": 3.1563,
      "step": 737
    },
    {
      "epoch": 0.6254237288135593,
      "grad_norm": 0.9568261504173279,
      "learning_rate": 3.9576271186440676e-05,
      "loss": 3.1749,
      "step": 738
    },
    {
      "epoch": 0.6262711864406779,
      "grad_norm": 1.0485754013061523,
      "learning_rate": 3.956214689265537e-05,
      "loss": 3.248,
      "step": 739
    },
    {
      "epoch": 0.6271186440677966,
      "grad_norm": 0.9002594947814941,
      "learning_rate": 3.954802259887006e-05,
      "loss": 3.1241,
      "step": 740
    },
    {
      "epoch": 0.6279661016949153,
      "grad_norm": 1.010948896408081,
      "learning_rate": 3.9533898305084746e-05,
      "loss": 3.2764,
      "step": 741
    },
    {
      "epoch": 0.6288135593220339,
      "grad_norm": 1.0623823404312134,
      "learning_rate": 3.9519774011299436e-05,
      "loss": 3.1771,
      "step": 742
    },
    {
      "epoch": 0.6296610169491526,
      "grad_norm": 0.951085090637207,
      "learning_rate": 3.9505649717514126e-05,
      "loss": 3.245,
      "step": 743
    },
    {
      "epoch": 0.6305084745762712,
      "grad_norm": 0.8962848782539368,
      "learning_rate": 3.9491525423728816e-05,
      "loss": 3.1905,
      "step": 744
    },
    {
      "epoch": 0.6313559322033898,
      "grad_norm": 1.0807331800460815,
      "learning_rate": 3.9477401129943506e-05,
      "loss": 3.1325,
      "step": 745
    },
    {
      "epoch": 0.6322033898305085,
      "grad_norm": 0.9225153923034668,
      "learning_rate": 3.946327683615819e-05,
      "loss": 3.1945,
      "step": 746
    },
    {
      "epoch": 0.6330508474576271,
      "grad_norm": 0.8945325613021851,
      "learning_rate": 3.9449152542372886e-05,
      "loss": 3.2607,
      "step": 747
    },
    {
      "epoch": 0.6338983050847458,
      "grad_norm": 1.0851283073425293,
      "learning_rate": 3.9435028248587576e-05,
      "loss": 3.3143,
      "step": 748
    },
    {
      "epoch": 0.6347457627118644,
      "grad_norm": 1.0192757844924927,
      "learning_rate": 3.9420903954802266e-05,
      "loss": 3.119,
      "step": 749
    },
    {
      "epoch": 0.635593220338983,
      "grad_norm": 0.9469757080078125,
      "learning_rate": 3.940677966101695e-05,
      "loss": 3.2111,
      "step": 750
    },
    {
      "epoch": 0.6364406779661017,
      "grad_norm": 0.9430133700370789,
      "learning_rate": 3.939265536723164e-05,
      "loss": 3.1526,
      "step": 751
    },
    {
      "epoch": 0.6372881355932203,
      "grad_norm": 1.0730900764465332,
      "learning_rate": 3.937853107344633e-05,
      "loss": 3.0441,
      "step": 752
    },
    {
      "epoch": 0.638135593220339,
      "grad_norm": 1.1505954265594482,
      "learning_rate": 3.936440677966102e-05,
      "loss": 3.2502,
      "step": 753
    },
    {
      "epoch": 0.6389830508474577,
      "grad_norm": 1.0467617511749268,
      "learning_rate": 3.935028248587571e-05,
      "loss": 3.191,
      "step": 754
    },
    {
      "epoch": 0.6398305084745762,
      "grad_norm": 0.9028640389442444,
      "learning_rate": 3.93361581920904e-05,
      "loss": 3.2017,
      "step": 755
    },
    {
      "epoch": 0.6406779661016949,
      "grad_norm": 1.0674667358398438,
      "learning_rate": 3.932203389830509e-05,
      "loss": 3.1055,
      "step": 756
    },
    {
      "epoch": 0.6415254237288136,
      "grad_norm": 1.0378832817077637,
      "learning_rate": 3.930790960451978e-05,
      "loss": 3.3165,
      "step": 757
    },
    {
      "epoch": 0.6423728813559322,
      "grad_norm": 1.0089573860168457,
      "learning_rate": 3.929378531073446e-05,
      "loss": 3.2966,
      "step": 758
    },
    {
      "epoch": 0.6432203389830509,
      "grad_norm": 1.0097712278366089,
      "learning_rate": 3.927966101694915e-05,
      "loss": 3.2123,
      "step": 759
    },
    {
      "epoch": 0.6440677966101694,
      "grad_norm": 0.9445637464523315,
      "learning_rate": 3.926553672316384e-05,
      "loss": 3.0963,
      "step": 760
    },
    {
      "epoch": 0.6449152542372881,
      "grad_norm": 0.9790430068969727,
      "learning_rate": 3.925141242937853e-05,
      "loss": 3.2946,
      "step": 761
    },
    {
      "epoch": 0.6457627118644068,
      "grad_norm": 0.962350070476532,
      "learning_rate": 3.923728813559322e-05,
      "loss": 3.1931,
      "step": 762
    },
    {
      "epoch": 0.6466101694915254,
      "grad_norm": 0.918471097946167,
      "learning_rate": 3.922316384180791e-05,
      "loss": 3.2357,
      "step": 763
    },
    {
      "epoch": 0.6474576271186441,
      "grad_norm": 0.9634029269218445,
      "learning_rate": 3.92090395480226e-05,
      "loss": 3.2918,
      "step": 764
    },
    {
      "epoch": 0.6483050847457628,
      "grad_norm": 1.0664550065994263,
      "learning_rate": 3.919491525423729e-05,
      "loss": 3.3134,
      "step": 765
    },
    {
      "epoch": 0.6491525423728813,
      "grad_norm": 1.2073235511779785,
      "learning_rate": 3.918079096045198e-05,
      "loss": 3.1881,
      "step": 766
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9471457600593567,
      "learning_rate": 3.9166666666666665e-05,
      "loss": 3.229,
      "step": 767
    },
    {
      "epoch": 0.6508474576271186,
      "grad_norm": 1.0511412620544434,
      "learning_rate": 3.9152542372881355e-05,
      "loss": 3.188,
      "step": 768
    },
    {
      "epoch": 0.6516949152542373,
      "grad_norm": 0.982347846031189,
      "learning_rate": 3.9138418079096045e-05,
      "loss": 3.3035,
      "step": 769
    },
    {
      "epoch": 0.652542372881356,
      "grad_norm": 1.1548480987548828,
      "learning_rate": 3.9124293785310735e-05,
      "loss": 3.1049,
      "step": 770
    },
    {
      "epoch": 0.6533898305084745,
      "grad_norm": 1.029438853263855,
      "learning_rate": 3.9110169491525425e-05,
      "loss": 3.2432,
      "step": 771
    },
    {
      "epoch": 0.6542372881355932,
      "grad_norm": 1.0703421831130981,
      "learning_rate": 3.9096045197740115e-05,
      "loss": 3.151,
      "step": 772
    },
    {
      "epoch": 0.6550847457627119,
      "grad_norm": 0.9013370275497437,
      "learning_rate": 3.9081920903954805e-05,
      "loss": 3.235,
      "step": 773
    },
    {
      "epoch": 0.6559322033898305,
      "grad_norm": 1.1812381744384766,
      "learning_rate": 3.9067796610169495e-05,
      "loss": 3.0823,
      "step": 774
    },
    {
      "epoch": 0.6567796610169492,
      "grad_norm": 0.9711376428604126,
      "learning_rate": 3.9053672316384185e-05,
      "loss": 3.1236,
      "step": 775
    },
    {
      "epoch": 0.6576271186440678,
      "grad_norm": 1.2097504138946533,
      "learning_rate": 3.903954802259887e-05,
      "loss": 3.1695,
      "step": 776
    },
    {
      "epoch": 0.6584745762711864,
      "grad_norm": 1.0563030242919922,
      "learning_rate": 3.902542372881356e-05,
      "loss": 3.2828,
      "step": 777
    },
    {
      "epoch": 0.6593220338983051,
      "grad_norm": 0.8016531467437744,
      "learning_rate": 3.901129943502825e-05,
      "loss": 3.1842,
      "step": 778
    },
    {
      "epoch": 0.6601694915254237,
      "grad_norm": 0.9715229868888855,
      "learning_rate": 3.899717514124294e-05,
      "loss": 3.1828,
      "step": 779
    },
    {
      "epoch": 0.6610169491525424,
      "grad_norm": 0.970073938369751,
      "learning_rate": 3.898305084745763e-05,
      "loss": 3.1113,
      "step": 780
    },
    {
      "epoch": 0.661864406779661,
      "grad_norm": 1.078091025352478,
      "learning_rate": 3.896892655367232e-05,
      "loss": 3.2263,
      "step": 781
    },
    {
      "epoch": 0.6627118644067796,
      "grad_norm": 1.0375956296920776,
      "learning_rate": 3.895480225988701e-05,
      "loss": 3.2353,
      "step": 782
    },
    {
      "epoch": 0.6635593220338983,
      "grad_norm": 1.000220775604248,
      "learning_rate": 3.89406779661017e-05,
      "loss": 3.1787,
      "step": 783
    },
    {
      "epoch": 0.6644067796610169,
      "grad_norm": 0.9153347015380859,
      "learning_rate": 3.892655367231638e-05,
      "loss": 3.3877,
      "step": 784
    },
    {
      "epoch": 0.6652542372881356,
      "grad_norm": 0.882677435874939,
      "learning_rate": 3.891242937853107e-05,
      "loss": 3.3769,
      "step": 785
    },
    {
      "epoch": 0.6661016949152543,
      "grad_norm": 0.8701127171516418,
      "learning_rate": 3.889830508474576e-05,
      "loss": 3.1109,
      "step": 786
    },
    {
      "epoch": 0.6669491525423729,
      "grad_norm": 0.9817793369293213,
      "learning_rate": 3.888418079096046e-05,
      "loss": 3.204,
      "step": 787
    },
    {
      "epoch": 0.6677966101694915,
      "grad_norm": 1.0561681985855103,
      "learning_rate": 3.887005649717514e-05,
      "loss": 3.1629,
      "step": 788
    },
    {
      "epoch": 0.6686440677966101,
      "grad_norm": 0.9255909323692322,
      "learning_rate": 3.885593220338983e-05,
      "loss": 3.2657,
      "step": 789
    },
    {
      "epoch": 0.6694915254237288,
      "grad_norm": 0.9756750464439392,
      "learning_rate": 3.884180790960452e-05,
      "loss": 3.2273,
      "step": 790
    },
    {
      "epoch": 0.6703389830508475,
      "grad_norm": 0.8761763572692871,
      "learning_rate": 3.882768361581921e-05,
      "loss": 3.2865,
      "step": 791
    },
    {
      "epoch": 0.6711864406779661,
      "grad_norm": 0.9409726858139038,
      "learning_rate": 3.88135593220339e-05,
      "loss": 3.14,
      "step": 792
    },
    {
      "epoch": 0.6720338983050848,
      "grad_norm": 1.0637263059616089,
      "learning_rate": 3.8799435028248585e-05,
      "loss": 3.303,
      "step": 793
    },
    {
      "epoch": 0.6728813559322034,
      "grad_norm": 1.0434190034866333,
      "learning_rate": 3.8785310734463275e-05,
      "loss": 3.1735,
      "step": 794
    },
    {
      "epoch": 0.673728813559322,
      "grad_norm": 1.0193620920181274,
      "learning_rate": 3.877118644067797e-05,
      "loss": 3.2918,
      "step": 795
    },
    {
      "epoch": 0.6745762711864407,
      "grad_norm": 0.8655635714530945,
      "learning_rate": 3.875706214689266e-05,
      "loss": 3.1329,
      "step": 796
    },
    {
      "epoch": 0.6754237288135593,
      "grad_norm": 1.1758955717086792,
      "learning_rate": 3.8742937853107345e-05,
      "loss": 3.1198,
      "step": 797
    },
    {
      "epoch": 0.676271186440678,
      "grad_norm": 0.9966703057289124,
      "learning_rate": 3.8728813559322035e-05,
      "loss": 3.2267,
      "step": 798
    },
    {
      "epoch": 0.6771186440677966,
      "grad_norm": 1.0191624164581299,
      "learning_rate": 3.8714689265536725e-05,
      "loss": 3.2686,
      "step": 799
    },
    {
      "epoch": 0.6779661016949152,
      "grad_norm": 0.9812909960746765,
      "learning_rate": 3.8700564971751415e-05,
      "loss": 3.1289,
      "step": 800
    },
    {
      "epoch": 0.6788135593220339,
      "grad_norm": 1.0124516487121582,
      "learning_rate": 3.86864406779661e-05,
      "loss": 3.354,
      "step": 801
    },
    {
      "epoch": 0.6796610169491526,
      "grad_norm": 0.9470294117927551,
      "learning_rate": 3.867231638418079e-05,
      "loss": 3.3026,
      "step": 802
    },
    {
      "epoch": 0.6805084745762712,
      "grad_norm": 1.2277476787567139,
      "learning_rate": 3.8658192090395485e-05,
      "loss": 3.2692,
      "step": 803
    },
    {
      "epoch": 0.6813559322033899,
      "grad_norm": 1.003859281539917,
      "learning_rate": 3.8644067796610175e-05,
      "loss": 3.2283,
      "step": 804
    },
    {
      "epoch": 0.6822033898305084,
      "grad_norm": 0.9394976496696472,
      "learning_rate": 3.862994350282486e-05,
      "loss": 3.1652,
      "step": 805
    },
    {
      "epoch": 0.6830508474576271,
      "grad_norm": 0.9242647290229797,
      "learning_rate": 3.861581920903955e-05,
      "loss": 3.2833,
      "step": 806
    },
    {
      "epoch": 0.6838983050847458,
      "grad_norm": 1.0409823656082153,
      "learning_rate": 3.860169491525424e-05,
      "loss": 3.276,
      "step": 807
    },
    {
      "epoch": 0.6847457627118644,
      "grad_norm": 1.079538106918335,
      "learning_rate": 3.858757062146893e-05,
      "loss": 3.13,
      "step": 808
    },
    {
      "epoch": 0.6855932203389831,
      "grad_norm": 1.1553713083267212,
      "learning_rate": 3.857344632768362e-05,
      "loss": 3.1682,
      "step": 809
    },
    {
      "epoch": 0.6864406779661016,
      "grad_norm": 1.3005847930908203,
      "learning_rate": 3.855932203389831e-05,
      "loss": 3.2126,
      "step": 810
    },
    {
      "epoch": 0.6872881355932203,
      "grad_norm": 0.918170690536499,
      "learning_rate": 3.8545197740113e-05,
      "loss": 3.2297,
      "step": 811
    },
    {
      "epoch": 0.688135593220339,
      "grad_norm": 0.9381653666496277,
      "learning_rate": 3.853107344632769e-05,
      "loss": 3.1483,
      "step": 812
    },
    {
      "epoch": 0.6889830508474576,
      "grad_norm": 0.9094195365905762,
      "learning_rate": 3.851694915254238e-05,
      "loss": 3.116,
      "step": 813
    },
    {
      "epoch": 0.6898305084745763,
      "grad_norm": 0.9763381481170654,
      "learning_rate": 3.850282485875706e-05,
      "loss": 3.2376,
      "step": 814
    },
    {
      "epoch": 0.690677966101695,
      "grad_norm": 1.1527831554412842,
      "learning_rate": 3.848870056497175e-05,
      "loss": 3.1901,
      "step": 815
    },
    {
      "epoch": 0.6915254237288135,
      "grad_norm": 1.1849908828735352,
      "learning_rate": 3.847457627118644e-05,
      "loss": 3.3913,
      "step": 816
    },
    {
      "epoch": 0.6923728813559322,
      "grad_norm": 1.0601989030838013,
      "learning_rate": 3.846045197740113e-05,
      "loss": 3.2095,
      "step": 817
    },
    {
      "epoch": 0.6932203389830508,
      "grad_norm": 1.2081618309020996,
      "learning_rate": 3.844632768361582e-05,
      "loss": 3.2408,
      "step": 818
    },
    {
      "epoch": 0.6940677966101695,
      "grad_norm": 1.1345655918121338,
      "learning_rate": 3.843220338983051e-05,
      "loss": 3.2974,
      "step": 819
    },
    {
      "epoch": 0.6949152542372882,
      "grad_norm": 1.0576690435409546,
      "learning_rate": 3.84180790960452e-05,
      "loss": 3.3008,
      "step": 820
    },
    {
      "epoch": 0.6957627118644067,
      "grad_norm": 0.9421098232269287,
      "learning_rate": 3.840395480225989e-05,
      "loss": 3.2526,
      "step": 821
    },
    {
      "epoch": 0.6966101694915254,
      "grad_norm": 0.8413639068603516,
      "learning_rate": 3.838983050847458e-05,
      "loss": 3.216,
      "step": 822
    },
    {
      "epoch": 0.6974576271186441,
      "grad_norm": 1.0307563543319702,
      "learning_rate": 3.8375706214689265e-05,
      "loss": 3.1639,
      "step": 823
    },
    {
      "epoch": 0.6983050847457627,
      "grad_norm": 1.0490492582321167,
      "learning_rate": 3.8361581920903955e-05,
      "loss": 3.0989,
      "step": 824
    },
    {
      "epoch": 0.6991525423728814,
      "grad_norm": 1.03816819190979,
      "learning_rate": 3.8347457627118644e-05,
      "loss": 3.1146,
      "step": 825
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.902752697467804,
      "learning_rate": 3.8333333333333334e-05,
      "loss": 3.3416,
      "step": 826
    },
    {
      "epoch": 0.7008474576271186,
      "grad_norm": 1.0437109470367432,
      "learning_rate": 3.8319209039548024e-05,
      "loss": 3.1637,
      "step": 827
    },
    {
      "epoch": 0.7016949152542373,
      "grad_norm": 1.011850357055664,
      "learning_rate": 3.8305084745762714e-05,
      "loss": 3.2652,
      "step": 828
    },
    {
      "epoch": 0.7025423728813559,
      "grad_norm": 1.0259604454040527,
      "learning_rate": 3.8290960451977404e-05,
      "loss": 3.183,
      "step": 829
    },
    {
      "epoch": 0.7033898305084746,
      "grad_norm": 0.9570820927619934,
      "learning_rate": 3.8276836158192094e-05,
      "loss": 3.2709,
      "step": 830
    },
    {
      "epoch": 0.7042372881355933,
      "grad_norm": 1.1066950559616089,
      "learning_rate": 3.826271186440678e-05,
      "loss": 3.1859,
      "step": 831
    },
    {
      "epoch": 0.7050847457627119,
      "grad_norm": 0.9420603513717651,
      "learning_rate": 3.824858757062147e-05,
      "loss": 3.2022,
      "step": 832
    },
    {
      "epoch": 0.7059322033898305,
      "grad_norm": 1.1082286834716797,
      "learning_rate": 3.823446327683616e-05,
      "loss": 3.2539,
      "step": 833
    },
    {
      "epoch": 0.7067796610169491,
      "grad_norm": 0.8663553595542908,
      "learning_rate": 3.8220338983050854e-05,
      "loss": 3.1253,
      "step": 834
    },
    {
      "epoch": 0.7076271186440678,
      "grad_norm": 0.9714896082878113,
      "learning_rate": 3.820621468926554e-05,
      "loss": 3.4205,
      "step": 835
    },
    {
      "epoch": 0.7084745762711865,
      "grad_norm": 1.0602551698684692,
      "learning_rate": 3.819209039548023e-05,
      "loss": 3.1955,
      "step": 836
    },
    {
      "epoch": 0.7093220338983051,
      "grad_norm": 1.0083402395248413,
      "learning_rate": 3.817796610169492e-05,
      "loss": 3.1873,
      "step": 837
    },
    {
      "epoch": 0.7101694915254237,
      "grad_norm": 1.052625298500061,
      "learning_rate": 3.816384180790961e-05,
      "loss": 3.2745,
      "step": 838
    },
    {
      "epoch": 0.7110169491525423,
      "grad_norm": 0.9955534338951111,
      "learning_rate": 3.81497175141243e-05,
      "loss": 3.1355,
      "step": 839
    },
    {
      "epoch": 0.711864406779661,
      "grad_norm": 1.1159393787384033,
      "learning_rate": 3.813559322033898e-05,
      "loss": 3.0875,
      "step": 840
    },
    {
      "epoch": 0.7127118644067797,
      "grad_norm": 0.9464603662490845,
      "learning_rate": 3.812146892655367e-05,
      "loss": 3.3046,
      "step": 841
    },
    {
      "epoch": 0.7135593220338983,
      "grad_norm": 0.8679315447807312,
      "learning_rate": 3.810734463276837e-05,
      "loss": 3.2797,
      "step": 842
    },
    {
      "epoch": 0.714406779661017,
      "grad_norm": 1.0267614126205444,
      "learning_rate": 3.809322033898306e-05,
      "loss": 3.1855,
      "step": 843
    },
    {
      "epoch": 0.7152542372881356,
      "grad_norm": 1.043196201324463,
      "learning_rate": 3.807909604519774e-05,
      "loss": 3.1491,
      "step": 844
    },
    {
      "epoch": 0.7161016949152542,
      "grad_norm": 0.9654357433319092,
      "learning_rate": 3.806497175141243e-05,
      "loss": 3.1582,
      "step": 845
    },
    {
      "epoch": 0.7169491525423729,
      "grad_norm": 1.2324398756027222,
      "learning_rate": 3.805084745762712e-05,
      "loss": 3.1493,
      "step": 846
    },
    {
      "epoch": 0.7177966101694915,
      "grad_norm": 1.07389497756958,
      "learning_rate": 3.803672316384181e-05,
      "loss": 3.22,
      "step": 847
    },
    {
      "epoch": 0.7186440677966102,
      "grad_norm": 1.0601229667663574,
      "learning_rate": 3.8022598870056494e-05,
      "loss": 3.3015,
      "step": 848
    },
    {
      "epoch": 0.7194915254237289,
      "grad_norm": 0.9054369330406189,
      "learning_rate": 3.8008474576271184e-05,
      "loss": 3.1114,
      "step": 849
    },
    {
      "epoch": 0.7203389830508474,
      "grad_norm": 0.9484003186225891,
      "learning_rate": 3.799435028248588e-05,
      "loss": 3.2363,
      "step": 850
    },
    {
      "epoch": 0.7211864406779661,
      "grad_norm": 0.9740283489227295,
      "learning_rate": 3.798022598870057e-05,
      "loss": 3.1581,
      "step": 851
    },
    {
      "epoch": 0.7220338983050848,
      "grad_norm": 1.0103195905685425,
      "learning_rate": 3.7966101694915254e-05,
      "loss": 3.2511,
      "step": 852
    },
    {
      "epoch": 0.7228813559322034,
      "grad_norm": 0.8766213655471802,
      "learning_rate": 3.7951977401129944e-05,
      "loss": 3.2887,
      "step": 853
    },
    {
      "epoch": 0.7237288135593221,
      "grad_norm": 0.9263383746147156,
      "learning_rate": 3.7937853107344634e-05,
      "loss": 3.144,
      "step": 854
    },
    {
      "epoch": 0.7245762711864406,
      "grad_norm": 1.0999993085861206,
      "learning_rate": 3.7923728813559324e-05,
      "loss": 3.2315,
      "step": 855
    },
    {
      "epoch": 0.7254237288135593,
      "grad_norm": 1.10708487033844,
      "learning_rate": 3.7909604519774014e-05,
      "loss": 3.2733,
      "step": 856
    },
    {
      "epoch": 0.726271186440678,
      "grad_norm": 0.9970806837081909,
      "learning_rate": 3.78954802259887e-05,
      "loss": 3.1575,
      "step": 857
    },
    {
      "epoch": 0.7271186440677966,
      "grad_norm": 0.8489723205566406,
      "learning_rate": 3.7881355932203394e-05,
      "loss": 3.1487,
      "step": 858
    },
    {
      "epoch": 0.7279661016949153,
      "grad_norm": 0.9401115775108337,
      "learning_rate": 3.7867231638418084e-05,
      "loss": 3.2257,
      "step": 859
    },
    {
      "epoch": 0.7288135593220338,
      "grad_norm": 0.8342302441596985,
      "learning_rate": 3.7853107344632774e-05,
      "loss": 3.2797,
      "step": 860
    },
    {
      "epoch": 0.7296610169491525,
      "grad_norm": 1.0335946083068848,
      "learning_rate": 3.783898305084746e-05,
      "loss": 3.1556,
      "step": 861
    },
    {
      "epoch": 0.7305084745762712,
      "grad_norm": 1.0974116325378418,
      "learning_rate": 3.782485875706215e-05,
      "loss": 3.1745,
      "step": 862
    },
    {
      "epoch": 0.7313559322033898,
      "grad_norm": 1.2478196620941162,
      "learning_rate": 3.781073446327684e-05,
      "loss": 3.1794,
      "step": 863
    },
    {
      "epoch": 0.7322033898305085,
      "grad_norm": 1.2136197090148926,
      "learning_rate": 3.779661016949153e-05,
      "loss": 3.2302,
      "step": 864
    },
    {
      "epoch": 0.7330508474576272,
      "grad_norm": 0.876383900642395,
      "learning_rate": 3.778248587570622e-05,
      "loss": 3.1867,
      "step": 865
    },
    {
      "epoch": 0.7338983050847457,
      "grad_norm": 1.1100138425827026,
      "learning_rate": 3.776836158192091e-05,
      "loss": 3.1793,
      "step": 866
    },
    {
      "epoch": 0.7347457627118644,
      "grad_norm": 1.0313066244125366,
      "learning_rate": 3.77542372881356e-05,
      "loss": 3.1828,
      "step": 867
    },
    {
      "epoch": 0.735593220338983,
      "grad_norm": 1.0279622077941895,
      "learning_rate": 3.774011299435029e-05,
      "loss": 3.1729,
      "step": 868
    },
    {
      "epoch": 0.7364406779661017,
      "grad_norm": 1.026025652885437,
      "learning_rate": 3.772598870056497e-05,
      "loss": 3.1684,
      "step": 869
    },
    {
      "epoch": 0.7372881355932204,
      "grad_norm": 0.9214180707931519,
      "learning_rate": 3.771186440677966e-05,
      "loss": 3.2182,
      "step": 870
    },
    {
      "epoch": 0.738135593220339,
      "grad_norm": 1.0707991123199463,
      "learning_rate": 3.769774011299435e-05,
      "loss": 3.2536,
      "step": 871
    },
    {
      "epoch": 0.7389830508474576,
      "grad_norm": 0.9317803978919983,
      "learning_rate": 3.768361581920904e-05,
      "loss": 3.1933,
      "step": 872
    },
    {
      "epoch": 0.7398305084745763,
      "grad_norm": 1.0014886856079102,
      "learning_rate": 3.766949152542373e-05,
      "loss": 3.2591,
      "step": 873
    },
    {
      "epoch": 0.7406779661016949,
      "grad_norm": 1.0117686986923218,
      "learning_rate": 3.765536723163842e-05,
      "loss": 3.1762,
      "step": 874
    },
    {
      "epoch": 0.7415254237288136,
      "grad_norm": 1.034946084022522,
      "learning_rate": 3.764124293785311e-05,
      "loss": 3.3375,
      "step": 875
    },
    {
      "epoch": 0.7423728813559322,
      "grad_norm": 0.8465149402618408,
      "learning_rate": 3.76271186440678e-05,
      "loss": 3.157,
      "step": 876
    },
    {
      "epoch": 0.7432203389830508,
      "grad_norm": 1.0686304569244385,
      "learning_rate": 3.761299435028249e-05,
      "loss": 3.132,
      "step": 877
    },
    {
      "epoch": 0.7440677966101695,
      "grad_norm": 1.0762747526168823,
      "learning_rate": 3.7598870056497174e-05,
      "loss": 3.11,
      "step": 878
    },
    {
      "epoch": 0.7449152542372881,
      "grad_norm": 0.9002189636230469,
      "learning_rate": 3.7584745762711864e-05,
      "loss": 3.2446,
      "step": 879
    },
    {
      "epoch": 0.7457627118644068,
      "grad_norm": 0.9305406808853149,
      "learning_rate": 3.7570621468926554e-05,
      "loss": 3.202,
      "step": 880
    },
    {
      "epoch": 0.7466101694915255,
      "grad_norm": 1.0967098474502563,
      "learning_rate": 3.7556497175141244e-05,
      "loss": 3.1081,
      "step": 881
    },
    {
      "epoch": 0.747457627118644,
      "grad_norm": 1.1288423538208008,
      "learning_rate": 3.7542372881355934e-05,
      "loss": 3.1961,
      "step": 882
    },
    {
      "epoch": 0.7483050847457627,
      "grad_norm": 1.2286369800567627,
      "learning_rate": 3.7528248587570624e-05,
      "loss": 3.2565,
      "step": 883
    },
    {
      "epoch": 0.7491525423728813,
      "grad_norm": 1.069288730621338,
      "learning_rate": 3.7514124293785313e-05,
      "loss": 3.1073,
      "step": 884
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1600650548934937,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 3.1716,
      "step": 885
    },
    {
      "epoch": 0.7508474576271187,
      "grad_norm": 0.9592218399047852,
      "learning_rate": 3.7485875706214693e-05,
      "loss": 3.1922,
      "step": 886
    },
    {
      "epoch": 0.7516949152542373,
      "grad_norm": 1.0506439208984375,
      "learning_rate": 3.747175141242938e-05,
      "loss": 3.0984,
      "step": 887
    },
    {
      "epoch": 0.752542372881356,
      "grad_norm": 0.9353660941123962,
      "learning_rate": 3.745762711864407e-05,
      "loss": 3.0849,
      "step": 888
    },
    {
      "epoch": 0.7533898305084745,
      "grad_norm": 1.062563180923462,
      "learning_rate": 3.7443502824858763e-05,
      "loss": 3.0448,
      "step": 889
    },
    {
      "epoch": 0.7542372881355932,
      "grad_norm": 0.9308637380599976,
      "learning_rate": 3.7429378531073453e-05,
      "loss": 3.1779,
      "step": 890
    },
    {
      "epoch": 0.7550847457627119,
      "grad_norm": 0.8553966283798218,
      "learning_rate": 3.741525423728814e-05,
      "loss": 3.2715,
      "step": 891
    },
    {
      "epoch": 0.7559322033898305,
      "grad_norm": 0.9313588738441467,
      "learning_rate": 3.740112994350283e-05,
      "loss": 3.2929,
      "step": 892
    },
    {
      "epoch": 0.7567796610169492,
      "grad_norm": 0.8892399072647095,
      "learning_rate": 3.738700564971752e-05,
      "loss": 3.1983,
      "step": 893
    },
    {
      "epoch": 0.7576271186440678,
      "grad_norm": 0.9385632872581482,
      "learning_rate": 3.737288135593221e-05,
      "loss": 3.1614,
      "step": 894
    },
    {
      "epoch": 0.7584745762711864,
      "grad_norm": 1.4118468761444092,
      "learning_rate": 3.735875706214689e-05,
      "loss": 3.1691,
      "step": 895
    },
    {
      "epoch": 0.7593220338983051,
      "grad_norm": 1.0137449502944946,
      "learning_rate": 3.734463276836158e-05,
      "loss": 3.2526,
      "step": 896
    },
    {
      "epoch": 0.7601694915254237,
      "grad_norm": 1.022806167602539,
      "learning_rate": 3.733050847457628e-05,
      "loss": 3.0853,
      "step": 897
    },
    {
      "epoch": 0.7610169491525424,
      "grad_norm": 0.96624755859375,
      "learning_rate": 3.731638418079097e-05,
      "loss": 3.2513,
      "step": 898
    },
    {
      "epoch": 0.761864406779661,
      "grad_norm": 1.2545897960662842,
      "learning_rate": 3.730225988700565e-05,
      "loss": 3.0895,
      "step": 899
    },
    {
      "epoch": 0.7627118644067796,
      "grad_norm": 1.1567143201828003,
      "learning_rate": 3.728813559322034e-05,
      "loss": 3.128,
      "step": 900
    },
    {
      "epoch": 0.7635593220338983,
      "grad_norm": 0.9578268527984619,
      "learning_rate": 3.727401129943503e-05,
      "loss": 3.2373,
      "step": 901
    },
    {
      "epoch": 0.764406779661017,
      "grad_norm": 0.8743712902069092,
      "learning_rate": 3.725988700564972e-05,
      "loss": 3.0874,
      "step": 902
    },
    {
      "epoch": 0.7652542372881356,
      "grad_norm": 1.1642415523529053,
      "learning_rate": 3.724576271186441e-05,
      "loss": 3.1992,
      "step": 903
    },
    {
      "epoch": 0.7661016949152543,
      "grad_norm": 1.0963908433914185,
      "learning_rate": 3.723163841807909e-05,
      "loss": 3.2434,
      "step": 904
    },
    {
      "epoch": 0.7669491525423728,
      "grad_norm": 0.9827107787132263,
      "learning_rate": 3.721751412429379e-05,
      "loss": 3.0884,
      "step": 905
    },
    {
      "epoch": 0.7677966101694915,
      "grad_norm": 0.9908869862556458,
      "learning_rate": 3.720338983050848e-05,
      "loss": 3.0993,
      "step": 906
    },
    {
      "epoch": 0.7686440677966102,
      "grad_norm": 0.939150333404541,
      "learning_rate": 3.718926553672317e-05,
      "loss": 3.2156,
      "step": 907
    },
    {
      "epoch": 0.7694915254237288,
      "grad_norm": 1.029697060585022,
      "learning_rate": 3.717514124293785e-05,
      "loss": 3.1514,
      "step": 908
    },
    {
      "epoch": 0.7703389830508475,
      "grad_norm": 0.9674649238586426,
      "learning_rate": 3.716101694915254e-05,
      "loss": 3.1725,
      "step": 909
    },
    {
      "epoch": 0.7711864406779662,
      "grad_norm": 1.0553406476974487,
      "learning_rate": 3.714689265536723e-05,
      "loss": 3.1553,
      "step": 910
    },
    {
      "epoch": 0.7720338983050847,
      "grad_norm": 0.8645443916320801,
      "learning_rate": 3.713276836158192e-05,
      "loss": 3.0972,
      "step": 911
    },
    {
      "epoch": 0.7728813559322034,
      "grad_norm": 0.909220814704895,
      "learning_rate": 3.711864406779661e-05,
      "loss": 3.1069,
      "step": 912
    },
    {
      "epoch": 0.773728813559322,
      "grad_norm": 1.0346428155899048,
      "learning_rate": 3.71045197740113e-05,
      "loss": 3.1356,
      "step": 913
    },
    {
      "epoch": 0.7745762711864407,
      "grad_norm": 1.0702482461929321,
      "learning_rate": 3.709039548022599e-05,
      "loss": 3.2995,
      "step": 914
    },
    {
      "epoch": 0.7754237288135594,
      "grad_norm": 0.93990159034729,
      "learning_rate": 3.707627118644068e-05,
      "loss": 3.1958,
      "step": 915
    },
    {
      "epoch": 0.7762711864406779,
      "grad_norm": 1.1232110261917114,
      "learning_rate": 3.7062146892655366e-05,
      "loss": 3.1094,
      "step": 916
    },
    {
      "epoch": 0.7771186440677966,
      "grad_norm": 1.079178810119629,
      "learning_rate": 3.7048022598870056e-05,
      "loss": 3.3298,
      "step": 917
    },
    {
      "epoch": 0.7779661016949152,
      "grad_norm": 1.0218323469161987,
      "learning_rate": 3.7033898305084746e-05,
      "loss": 3.1555,
      "step": 918
    },
    {
      "epoch": 0.7788135593220339,
      "grad_norm": 1.0812876224517822,
      "learning_rate": 3.7019774011299436e-05,
      "loss": 3.1834,
      "step": 919
    },
    {
      "epoch": 0.7796610169491526,
      "grad_norm": 0.9746326804161072,
      "learning_rate": 3.7005649717514126e-05,
      "loss": 3.1584,
      "step": 920
    },
    {
      "epoch": 0.7805084745762711,
      "grad_norm": 1.3261560201644897,
      "learning_rate": 3.6991525423728816e-05,
      "loss": 3.1947,
      "step": 921
    },
    {
      "epoch": 0.7813559322033898,
      "grad_norm": 1.0622180700302124,
      "learning_rate": 3.6977401129943506e-05,
      "loss": 3.1363,
      "step": 922
    },
    {
      "epoch": 0.7822033898305085,
      "grad_norm": 0.9775424003601074,
      "learning_rate": 3.6963276836158196e-05,
      "loss": 3.1012,
      "step": 923
    },
    {
      "epoch": 0.7830508474576271,
      "grad_norm": 1.0168304443359375,
      "learning_rate": 3.6949152542372886e-05,
      "loss": 3.1112,
      "step": 924
    },
    {
      "epoch": 0.7838983050847458,
      "grad_norm": 1.0288655757904053,
      "learning_rate": 3.693502824858757e-05,
      "loss": 3.2457,
      "step": 925
    },
    {
      "epoch": 0.7847457627118644,
      "grad_norm": 0.9620063304901123,
      "learning_rate": 3.692090395480226e-05,
      "loss": 3.2217,
      "step": 926
    },
    {
      "epoch": 0.785593220338983,
      "grad_norm": 1.0269447565078735,
      "learning_rate": 3.690677966101695e-05,
      "loss": 3.1024,
      "step": 927
    },
    {
      "epoch": 0.7864406779661017,
      "grad_norm": 0.8785094022750854,
      "learning_rate": 3.689265536723164e-05,
      "loss": 3.1708,
      "step": 928
    },
    {
      "epoch": 0.7872881355932203,
      "grad_norm": 0.9808406829833984,
      "learning_rate": 3.687853107344633e-05,
      "loss": 3.202,
      "step": 929
    },
    {
      "epoch": 0.788135593220339,
      "grad_norm": 1.1924328804016113,
      "learning_rate": 3.686440677966102e-05,
      "loss": 3.2412,
      "step": 930
    },
    {
      "epoch": 0.7889830508474577,
      "grad_norm": 1.0383801460266113,
      "learning_rate": 3.685028248587571e-05,
      "loss": 3.1894,
      "step": 931
    },
    {
      "epoch": 0.7898305084745763,
      "grad_norm": 1.1522718667984009,
      "learning_rate": 3.68361581920904e-05,
      "loss": 3.0404,
      "step": 932
    },
    {
      "epoch": 0.7906779661016949,
      "grad_norm": 0.9969879388809204,
      "learning_rate": 3.682203389830509e-05,
      "loss": 3.1535,
      "step": 933
    },
    {
      "epoch": 0.7915254237288135,
      "grad_norm": 1.0482536554336548,
      "learning_rate": 3.680790960451977e-05,
      "loss": 3.194,
      "step": 934
    },
    {
      "epoch": 0.7923728813559322,
      "grad_norm": 0.9472736120223999,
      "learning_rate": 3.679378531073446e-05,
      "loss": 3.242,
      "step": 935
    },
    {
      "epoch": 0.7932203389830509,
      "grad_norm": 1.1946145296096802,
      "learning_rate": 3.677966101694915e-05,
      "loss": 3.0908,
      "step": 936
    },
    {
      "epoch": 0.7940677966101695,
      "grad_norm": 0.8134632706642151,
      "learning_rate": 3.676553672316384e-05,
      "loss": 3.0917,
      "step": 937
    },
    {
      "epoch": 0.7949152542372881,
      "grad_norm": 1.1249165534973145,
      "learning_rate": 3.675141242937853e-05,
      "loss": 3.189,
      "step": 938
    },
    {
      "epoch": 0.7957627118644067,
      "grad_norm": 0.9513387084007263,
      "learning_rate": 3.673728813559322e-05,
      "loss": 3.1798,
      "step": 939
    },
    {
      "epoch": 0.7966101694915254,
      "grad_norm": 0.9367225766181946,
      "learning_rate": 3.672316384180791e-05,
      "loss": 3.2219,
      "step": 940
    },
    {
      "epoch": 0.7974576271186441,
      "grad_norm": 0.8651477694511414,
      "learning_rate": 3.67090395480226e-05,
      "loss": 3.0195,
      "step": 941
    },
    {
      "epoch": 0.7983050847457627,
      "grad_norm": 1.0951869487762451,
      "learning_rate": 3.6694915254237286e-05,
      "loss": 3.1604,
      "step": 942
    },
    {
      "epoch": 0.7991525423728814,
      "grad_norm": 0.9521233439445496,
      "learning_rate": 3.6680790960451976e-05,
      "loss": 3.3187,
      "step": 943
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0337250232696533,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 3.2058,
      "step": 944
    },
    {
      "epoch": 0.8008474576271186,
      "grad_norm": 0.9719616174697876,
      "learning_rate": 3.665254237288136e-05,
      "loss": 3.179,
      "step": 945
    },
    {
      "epoch": 0.8016949152542373,
      "grad_norm": 0.9687196016311646,
      "learning_rate": 3.6638418079096046e-05,
      "loss": 3.2478,
      "step": 946
    },
    {
      "epoch": 0.8025423728813559,
      "grad_norm": 0.9144794344902039,
      "learning_rate": 3.6624293785310736e-05,
      "loss": 3.2144,
      "step": 947
    },
    {
      "epoch": 0.8033898305084746,
      "grad_norm": 1.1403071880340576,
      "learning_rate": 3.6610169491525426e-05,
      "loss": 3.1879,
      "step": 948
    },
    {
      "epoch": 0.8042372881355933,
      "grad_norm": 0.9317966103553772,
      "learning_rate": 3.6596045197740116e-05,
      "loss": 3.189,
      "step": 949
    },
    {
      "epoch": 0.8050847457627118,
      "grad_norm": 0.9150249361991882,
      "learning_rate": 3.6581920903954806e-05,
      "loss": 3.2399,
      "step": 950
    },
    {
      "epoch": 0.8059322033898305,
      "grad_norm": 1.1650607585906982,
      "learning_rate": 3.656779661016949e-05,
      "loss": 3.1,
      "step": 951
    },
    {
      "epoch": 0.8067796610169492,
      "grad_norm": 1.0316554307937622,
      "learning_rate": 3.655367231638418e-05,
      "loss": 3.1944,
      "step": 952
    },
    {
      "epoch": 0.8076271186440678,
      "grad_norm": 1.0725008249282837,
      "learning_rate": 3.6539548022598876e-05,
      "loss": 3.1793,
      "step": 953
    },
    {
      "epoch": 0.8084745762711865,
      "grad_norm": 0.9841150641441345,
      "learning_rate": 3.6525423728813566e-05,
      "loss": 3.2736,
      "step": 954
    },
    {
      "epoch": 0.809322033898305,
      "grad_norm": 0.9306738972663879,
      "learning_rate": 3.651129943502825e-05,
      "loss": 3.1939,
      "step": 955
    },
    {
      "epoch": 0.8101694915254237,
      "grad_norm": 0.7785322070121765,
      "learning_rate": 3.649717514124294e-05,
      "loss": 3.2341,
      "step": 956
    },
    {
      "epoch": 0.8110169491525424,
      "grad_norm": 0.9500228762626648,
      "learning_rate": 3.648305084745763e-05,
      "loss": 3.1834,
      "step": 957
    },
    {
      "epoch": 0.811864406779661,
      "grad_norm": 0.9458290338516235,
      "learning_rate": 3.646892655367232e-05,
      "loss": 3.2147,
      "step": 958
    },
    {
      "epoch": 0.8127118644067797,
      "grad_norm": 1.0137789249420166,
      "learning_rate": 3.6454802259887e-05,
      "loss": 3.0907,
      "step": 959
    },
    {
      "epoch": 0.8135593220338984,
      "grad_norm": 1.0603535175323486,
      "learning_rate": 3.644067796610169e-05,
      "loss": 3.288,
      "step": 960
    },
    {
      "epoch": 0.8144067796610169,
      "grad_norm": 0.9757066369056702,
      "learning_rate": 3.642655367231639e-05,
      "loss": 3.1827,
      "step": 961
    },
    {
      "epoch": 0.8152542372881356,
      "grad_norm": 0.9474855661392212,
      "learning_rate": 3.641242937853108e-05,
      "loss": 3.1803,
      "step": 962
    },
    {
      "epoch": 0.8161016949152542,
      "grad_norm": 0.9666436314582825,
      "learning_rate": 3.639830508474576e-05,
      "loss": 3.202,
      "step": 963
    },
    {
      "epoch": 0.8169491525423729,
      "grad_norm": 0.9885278940200806,
      "learning_rate": 3.638418079096045e-05,
      "loss": 3.3161,
      "step": 964
    },
    {
      "epoch": 0.8177966101694916,
      "grad_norm": 1.1008918285369873,
      "learning_rate": 3.637005649717514e-05,
      "loss": 3.2544,
      "step": 965
    },
    {
      "epoch": 0.8186440677966101,
      "grad_norm": 1.0333728790283203,
      "learning_rate": 3.635593220338983e-05,
      "loss": 3.1205,
      "step": 966
    },
    {
      "epoch": 0.8194915254237288,
      "grad_norm": 1.096467137336731,
      "learning_rate": 3.634180790960452e-05,
      "loss": 3.0647,
      "step": 967
    },
    {
      "epoch": 0.8203389830508474,
      "grad_norm": 0.9635850787162781,
      "learning_rate": 3.632768361581921e-05,
      "loss": 3.1874,
      "step": 968
    },
    {
      "epoch": 0.8211864406779661,
      "grad_norm": 1.0629239082336426,
      "learning_rate": 3.63135593220339e-05,
      "loss": 3.1279,
      "step": 969
    },
    {
      "epoch": 0.8220338983050848,
      "grad_norm": 0.9065790772438049,
      "learning_rate": 3.629943502824859e-05,
      "loss": 3.2068,
      "step": 970
    },
    {
      "epoch": 0.8228813559322034,
      "grad_norm": 1.0202919244766235,
      "learning_rate": 3.628531073446328e-05,
      "loss": 2.9332,
      "step": 971
    },
    {
      "epoch": 0.823728813559322,
      "grad_norm": 1.0279127359390259,
      "learning_rate": 3.6271186440677965e-05,
      "loss": 3.1212,
      "step": 972
    },
    {
      "epoch": 0.8245762711864407,
      "grad_norm": 1.0611159801483154,
      "learning_rate": 3.6257062146892655e-05,
      "loss": 3.2993,
      "step": 973
    },
    {
      "epoch": 0.8254237288135593,
      "grad_norm": 0.8197402954101562,
      "learning_rate": 3.6242937853107345e-05,
      "loss": 3.1063,
      "step": 974
    },
    {
      "epoch": 0.826271186440678,
      "grad_norm": 0.9545575976371765,
      "learning_rate": 3.6228813559322035e-05,
      "loss": 3.0439,
      "step": 975
    },
    {
      "epoch": 0.8271186440677966,
      "grad_norm": 0.8873161673545837,
      "learning_rate": 3.6214689265536725e-05,
      "loss": 3.1075,
      "step": 976
    },
    {
      "epoch": 0.8279661016949152,
      "grad_norm": 0.981919527053833,
      "learning_rate": 3.6200564971751415e-05,
      "loss": 3.151,
      "step": 977
    },
    {
      "epoch": 0.8288135593220339,
      "grad_norm": 1.072557806968689,
      "learning_rate": 3.6186440677966105e-05,
      "loss": 3.0936,
      "step": 978
    },
    {
      "epoch": 0.8296610169491525,
      "grad_norm": 1.063246726989746,
      "learning_rate": 3.6172316384180795e-05,
      "loss": 3.2732,
      "step": 979
    },
    {
      "epoch": 0.8305084745762712,
      "grad_norm": 1.1005663871765137,
      "learning_rate": 3.6158192090395485e-05,
      "loss": 3.1267,
      "step": 980
    },
    {
      "epoch": 0.8313559322033899,
      "grad_norm": 0.9517815113067627,
      "learning_rate": 3.614406779661017e-05,
      "loss": 3.1137,
      "step": 981
    },
    {
      "epoch": 0.8322033898305085,
      "grad_norm": 1.1257226467132568,
      "learning_rate": 3.612994350282486e-05,
      "loss": 3.1086,
      "step": 982
    },
    {
      "epoch": 0.8330508474576271,
      "grad_norm": 1.0828503370285034,
      "learning_rate": 3.611581920903955e-05,
      "loss": 3.1462,
      "step": 983
    },
    {
      "epoch": 0.8338983050847457,
      "grad_norm": 0.8862940669059753,
      "learning_rate": 3.610169491525424e-05,
      "loss": 3.1509,
      "step": 984
    },
    {
      "epoch": 0.8347457627118644,
      "grad_norm": 0.9474267363548279,
      "learning_rate": 3.608757062146893e-05,
      "loss": 3.0484,
      "step": 985
    },
    {
      "epoch": 0.8355932203389831,
      "grad_norm": 1.0636128187179565,
      "learning_rate": 3.607344632768362e-05,
      "loss": 3.0442,
      "step": 986
    },
    {
      "epoch": 0.8364406779661017,
      "grad_norm": 1.1512815952301025,
      "learning_rate": 3.605932203389831e-05,
      "loss": 3.1838,
      "step": 987
    },
    {
      "epoch": 0.8372881355932204,
      "grad_norm": 0.9204326868057251,
      "learning_rate": 3.6045197740113e-05,
      "loss": 3.1685,
      "step": 988
    },
    {
      "epoch": 0.838135593220339,
      "grad_norm": 0.978344738483429,
      "learning_rate": 3.603107344632768e-05,
      "loss": 3.0575,
      "step": 989
    },
    {
      "epoch": 0.8389830508474576,
      "grad_norm": 0.8899116516113281,
      "learning_rate": 3.601694915254237e-05,
      "loss": 3.2671,
      "step": 990
    },
    {
      "epoch": 0.8398305084745763,
      "grad_norm": 1.0270024538040161,
      "learning_rate": 3.600282485875706e-05,
      "loss": 3.0934,
      "step": 991
    },
    {
      "epoch": 0.8406779661016949,
      "grad_norm": 1.0981197357177734,
      "learning_rate": 3.598870056497176e-05,
      "loss": 3.2038,
      "step": 992
    },
    {
      "epoch": 0.8415254237288136,
      "grad_norm": 1.0259603261947632,
      "learning_rate": 3.597457627118644e-05,
      "loss": 3.1694,
      "step": 993
    },
    {
      "epoch": 0.8423728813559322,
      "grad_norm": 1.0101300477981567,
      "learning_rate": 3.596045197740113e-05,
      "loss": 3.2253,
      "step": 994
    },
    {
      "epoch": 0.8432203389830508,
      "grad_norm": 1.0292799472808838,
      "learning_rate": 3.594632768361582e-05,
      "loss": 3.27,
      "step": 995
    },
    {
      "epoch": 0.8440677966101695,
      "grad_norm": 0.9315595030784607,
      "learning_rate": 3.593220338983051e-05,
      "loss": 3.1596,
      "step": 996
    },
    {
      "epoch": 0.8449152542372881,
      "grad_norm": 1.0345228910446167,
      "learning_rate": 3.59180790960452e-05,
      "loss": 3.0738,
      "step": 997
    },
    {
      "epoch": 0.8457627118644068,
      "grad_norm": 0.887259304523468,
      "learning_rate": 3.5903954802259885e-05,
      "loss": 3.152,
      "step": 998
    },
    {
      "epoch": 0.8466101694915255,
      "grad_norm": 0.9285693168640137,
      "learning_rate": 3.5889830508474575e-05,
      "loss": 3.0995,
      "step": 999
    },
    {
      "epoch": 0.847457627118644,
      "grad_norm": 1.0636188983917236,
      "learning_rate": 3.587570621468927e-05,
      "loss": 3.0885,
      "step": 1000
    },
    {
      "epoch": 0.8483050847457627,
      "grad_norm": 0.9358274936676025,
      "learning_rate": 3.586158192090396e-05,
      "loss": 3.1655,
      "step": 1001
    },
    {
      "epoch": 0.8491525423728814,
      "grad_norm": 1.0933887958526611,
      "learning_rate": 3.5847457627118645e-05,
      "loss": 3.123,
      "step": 1002
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8956212401390076,
      "learning_rate": 3.5833333333333335e-05,
      "loss": 3.158,
      "step": 1003
    },
    {
      "epoch": 0.8508474576271187,
      "grad_norm": 1.0863122940063477,
      "learning_rate": 3.5819209039548025e-05,
      "loss": 3.2314,
      "step": 1004
    },
    {
      "epoch": 0.8516949152542372,
      "grad_norm": 0.9890239834785461,
      "learning_rate": 3.5805084745762715e-05,
      "loss": 3.0972,
      "step": 1005
    },
    {
      "epoch": 0.8525423728813559,
      "grad_norm": 1.0596765279769897,
      "learning_rate": 3.57909604519774e-05,
      "loss": 3.074,
      "step": 1006
    },
    {
      "epoch": 0.8533898305084746,
      "grad_norm": 0.9362082481384277,
      "learning_rate": 3.577683615819209e-05,
      "loss": 3.0835,
      "step": 1007
    },
    {
      "epoch": 0.8542372881355932,
      "grad_norm": 0.9448023438453674,
      "learning_rate": 3.5762711864406785e-05,
      "loss": 3.2233,
      "step": 1008
    },
    {
      "epoch": 0.8550847457627119,
      "grad_norm": 0.9858917593955994,
      "learning_rate": 3.5748587570621475e-05,
      "loss": 3.0609,
      "step": 1009
    },
    {
      "epoch": 0.8559322033898306,
      "grad_norm": 0.9291073679924011,
      "learning_rate": 3.573446327683616e-05,
      "loss": 3.1418,
      "step": 1010
    },
    {
      "epoch": 0.8567796610169491,
      "grad_norm": 1.019467830657959,
      "learning_rate": 3.572033898305085e-05,
      "loss": 3.1534,
      "step": 1011
    },
    {
      "epoch": 0.8576271186440678,
      "grad_norm": 0.8311463594436646,
      "learning_rate": 3.570621468926554e-05,
      "loss": 3.18,
      "step": 1012
    },
    {
      "epoch": 0.8584745762711864,
      "grad_norm": 1.0131714344024658,
      "learning_rate": 3.569209039548023e-05,
      "loss": 2.9819,
      "step": 1013
    },
    {
      "epoch": 0.8593220338983051,
      "grad_norm": 1.0068503618240356,
      "learning_rate": 3.567796610169492e-05,
      "loss": 3.1073,
      "step": 1014
    },
    {
      "epoch": 0.8601694915254238,
      "grad_norm": 0.8969554901123047,
      "learning_rate": 3.56638418079096e-05,
      "loss": 3.1638,
      "step": 1015
    },
    {
      "epoch": 0.8610169491525423,
      "grad_norm": 0.8899940252304077,
      "learning_rate": 3.56497175141243e-05,
      "loss": 3.2716,
      "step": 1016
    },
    {
      "epoch": 0.861864406779661,
      "grad_norm": 0.9648430943489075,
      "learning_rate": 3.563559322033899e-05,
      "loss": 3.2465,
      "step": 1017
    },
    {
      "epoch": 0.8627118644067797,
      "grad_norm": 1.0650931596755981,
      "learning_rate": 3.562146892655368e-05,
      "loss": 3.2254,
      "step": 1018
    },
    {
      "epoch": 0.8635593220338983,
      "grad_norm": 1.2342017889022827,
      "learning_rate": 3.560734463276836e-05,
      "loss": 2.9903,
      "step": 1019
    },
    {
      "epoch": 0.864406779661017,
      "grad_norm": 1.0534242391586304,
      "learning_rate": 3.559322033898305e-05,
      "loss": 3.0908,
      "step": 1020
    },
    {
      "epoch": 0.8652542372881356,
      "grad_norm": 1.1241739988327026,
      "learning_rate": 3.557909604519774e-05,
      "loss": 3.1064,
      "step": 1021
    },
    {
      "epoch": 0.8661016949152542,
      "grad_norm": 0.9351906776428223,
      "learning_rate": 3.556497175141243e-05,
      "loss": 3.2339,
      "step": 1022
    },
    {
      "epoch": 0.8669491525423729,
      "grad_norm": 1.0255619287490845,
      "learning_rate": 3.555084745762712e-05,
      "loss": 3.1819,
      "step": 1023
    },
    {
      "epoch": 0.8677966101694915,
      "grad_norm": 1.1390897035598755,
      "learning_rate": 3.553672316384181e-05,
      "loss": 2.9748,
      "step": 1024
    },
    {
      "epoch": 0.8686440677966102,
      "grad_norm": 0.9607123732566833,
      "learning_rate": 3.55225988700565e-05,
      "loss": 3.229,
      "step": 1025
    },
    {
      "epoch": 0.8694915254237288,
      "grad_norm": 0.9447827935218811,
      "learning_rate": 3.550847457627119e-05,
      "loss": 3.1474,
      "step": 1026
    },
    {
      "epoch": 0.8703389830508474,
      "grad_norm": 0.8953350782394409,
      "learning_rate": 3.5494350282485874e-05,
      "loss": 3.1226,
      "step": 1027
    },
    {
      "epoch": 0.8711864406779661,
      "grad_norm": 0.8721854090690613,
      "learning_rate": 3.5480225988700564e-05,
      "loss": 3.1238,
      "step": 1028
    },
    {
      "epoch": 0.8720338983050847,
      "grad_norm": 0.9612311124801636,
      "learning_rate": 3.5466101694915254e-05,
      "loss": 3.1679,
      "step": 1029
    },
    {
      "epoch": 0.8728813559322034,
      "grad_norm": 0.9968669414520264,
      "learning_rate": 3.5451977401129944e-05,
      "loss": 3.092,
      "step": 1030
    },
    {
      "epoch": 0.8737288135593221,
      "grad_norm": 1.1011210680007935,
      "learning_rate": 3.5437853107344634e-05,
      "loss": 3.1723,
      "step": 1031
    },
    {
      "epoch": 0.8745762711864407,
      "grad_norm": 1.0265681743621826,
      "learning_rate": 3.5423728813559324e-05,
      "loss": 3.2485,
      "step": 1032
    },
    {
      "epoch": 0.8754237288135593,
      "grad_norm": 1.0001736879348755,
      "learning_rate": 3.5409604519774014e-05,
      "loss": 3.1407,
      "step": 1033
    },
    {
      "epoch": 0.8762711864406779,
      "grad_norm": 0.862217903137207,
      "learning_rate": 3.5395480225988704e-05,
      "loss": 3.1386,
      "step": 1034
    },
    {
      "epoch": 0.8771186440677966,
      "grad_norm": 1.0363800525665283,
      "learning_rate": 3.5381355932203394e-05,
      "loss": 3.1702,
      "step": 1035
    },
    {
      "epoch": 0.8779661016949153,
      "grad_norm": 1.1347532272338867,
      "learning_rate": 3.536723163841808e-05,
      "loss": 3.0559,
      "step": 1036
    },
    {
      "epoch": 0.8788135593220339,
      "grad_norm": 1.0662977695465088,
      "learning_rate": 3.535310734463277e-05,
      "loss": 3.0345,
      "step": 1037
    },
    {
      "epoch": 0.8796610169491526,
      "grad_norm": 0.9417952299118042,
      "learning_rate": 3.533898305084746e-05,
      "loss": 3.1034,
      "step": 1038
    },
    {
      "epoch": 0.8805084745762712,
      "grad_norm": 1.018358588218689,
      "learning_rate": 3.532485875706215e-05,
      "loss": 3.1722,
      "step": 1039
    },
    {
      "epoch": 0.8813559322033898,
      "grad_norm": 0.9088732004165649,
      "learning_rate": 3.531073446327684e-05,
      "loss": 3.2741,
      "step": 1040
    },
    {
      "epoch": 0.8822033898305085,
      "grad_norm": 1.2275269031524658,
      "learning_rate": 3.529661016949153e-05,
      "loss": 3.1578,
      "step": 1041
    },
    {
      "epoch": 0.8830508474576271,
      "grad_norm": 0.9241889715194702,
      "learning_rate": 3.528248587570622e-05,
      "loss": 3.1585,
      "step": 1042
    },
    {
      "epoch": 0.8838983050847458,
      "grad_norm": 1.006468415260315,
      "learning_rate": 3.526836158192091e-05,
      "loss": 3.1019,
      "step": 1043
    },
    {
      "epoch": 0.8847457627118644,
      "grad_norm": 1.0921082496643066,
      "learning_rate": 3.52542372881356e-05,
      "loss": 3.1618,
      "step": 1044
    },
    {
      "epoch": 0.885593220338983,
      "grad_norm": 0.9923046231269836,
      "learning_rate": 3.524011299435028e-05,
      "loss": 3.2301,
      "step": 1045
    },
    {
      "epoch": 0.8864406779661017,
      "grad_norm": 1.018099308013916,
      "learning_rate": 3.522598870056497e-05,
      "loss": 3.042,
      "step": 1046
    },
    {
      "epoch": 0.8872881355932203,
      "grad_norm": 0.9938775300979614,
      "learning_rate": 3.521186440677967e-05,
      "loss": 3.0783,
      "step": 1047
    },
    {
      "epoch": 0.888135593220339,
      "grad_norm": 1.0113307237625122,
      "learning_rate": 3.519774011299436e-05,
      "loss": 3.1987,
      "step": 1048
    },
    {
      "epoch": 0.8889830508474577,
      "grad_norm": 0.9744778871536255,
      "learning_rate": 3.518361581920904e-05,
      "loss": 3.2,
      "step": 1049
    },
    {
      "epoch": 0.8898305084745762,
      "grad_norm": 1.1762267351150513,
      "learning_rate": 3.516949152542373e-05,
      "loss": 3.0518,
      "step": 1050
    },
    {
      "epoch": 0.8906779661016949,
      "grad_norm": 0.9027320146560669,
      "learning_rate": 3.515536723163842e-05,
      "loss": 3.1483,
      "step": 1051
    },
    {
      "epoch": 0.8915254237288136,
      "grad_norm": 1.0681418180465698,
      "learning_rate": 3.514124293785311e-05,
      "loss": 3.2282,
      "step": 1052
    },
    {
      "epoch": 0.8923728813559322,
      "grad_norm": 1.1005299091339111,
      "learning_rate": 3.5127118644067794e-05,
      "loss": 3.152,
      "step": 1053
    },
    {
      "epoch": 0.8932203389830509,
      "grad_norm": 0.8355681300163269,
      "learning_rate": 3.5112994350282484e-05,
      "loss": 3.0815,
      "step": 1054
    },
    {
      "epoch": 0.8940677966101694,
      "grad_norm": 0.8652706146240234,
      "learning_rate": 3.509887005649718e-05,
      "loss": 3.1695,
      "step": 1055
    },
    {
      "epoch": 0.8949152542372881,
      "grad_norm": 1.1734929084777832,
      "learning_rate": 3.508474576271187e-05,
      "loss": 3.1214,
      "step": 1056
    },
    {
      "epoch": 0.8957627118644068,
      "grad_norm": 0.866345226764679,
      "learning_rate": 3.5070621468926554e-05,
      "loss": 3.098,
      "step": 1057
    },
    {
      "epoch": 0.8966101694915254,
      "grad_norm": 1.004052758216858,
      "learning_rate": 3.5056497175141244e-05,
      "loss": 3.2785,
      "step": 1058
    },
    {
      "epoch": 0.8974576271186441,
      "grad_norm": 1.025038719177246,
      "learning_rate": 3.5042372881355934e-05,
      "loss": 3.2701,
      "step": 1059
    },
    {
      "epoch": 0.8983050847457628,
      "grad_norm": 0.9300771355628967,
      "learning_rate": 3.5028248587570624e-05,
      "loss": 3.0623,
      "step": 1060
    },
    {
      "epoch": 0.8991525423728813,
      "grad_norm": 0.8885724544525146,
      "learning_rate": 3.5014124293785314e-05,
      "loss": 3.1737,
      "step": 1061
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9810854196548462,
      "learning_rate": 3.5e-05,
      "loss": 3.2598,
      "step": 1062
    },
    {
      "epoch": 0.9008474576271186,
      "grad_norm": 1.1284035444259644,
      "learning_rate": 3.4985875706214694e-05,
      "loss": 3.1814,
      "step": 1063
    },
    {
      "epoch": 0.9016949152542373,
      "grad_norm": 0.9558525085449219,
      "learning_rate": 3.4971751412429384e-05,
      "loss": 3.1885,
      "step": 1064
    },
    {
      "epoch": 0.902542372881356,
      "grad_norm": 0.8473979234695435,
      "learning_rate": 3.4957627118644074e-05,
      "loss": 3.137,
      "step": 1065
    },
    {
      "epoch": 0.9033898305084745,
      "grad_norm": 0.8437260389328003,
      "learning_rate": 3.494350282485876e-05,
      "loss": 3.389,
      "step": 1066
    },
    {
      "epoch": 0.9042372881355932,
      "grad_norm": 1.0226668119430542,
      "learning_rate": 3.492937853107345e-05,
      "loss": 3.0994,
      "step": 1067
    },
    {
      "epoch": 0.9050847457627119,
      "grad_norm": 0.822215735912323,
      "learning_rate": 3.491525423728814e-05,
      "loss": 3.2503,
      "step": 1068
    },
    {
      "epoch": 0.9059322033898305,
      "grad_norm": 1.1045360565185547,
      "learning_rate": 3.490112994350283e-05,
      "loss": 3.2006,
      "step": 1069
    },
    {
      "epoch": 0.9067796610169492,
      "grad_norm": 0.9761415123939514,
      "learning_rate": 3.488700564971752e-05,
      "loss": 2.9618,
      "step": 1070
    },
    {
      "epoch": 0.9076271186440678,
      "grad_norm": 0.955146849155426,
      "learning_rate": 3.487288135593221e-05,
      "loss": 3.2325,
      "step": 1071
    },
    {
      "epoch": 0.9084745762711864,
      "grad_norm": 1.0589054822921753,
      "learning_rate": 3.48587570621469e-05,
      "loss": 3.1792,
      "step": 1072
    },
    {
      "epoch": 0.9093220338983051,
      "grad_norm": 0.9603967070579529,
      "learning_rate": 3.484463276836159e-05,
      "loss": 3.0278,
      "step": 1073
    },
    {
      "epoch": 0.9101694915254237,
      "grad_norm": 1.0430519580841064,
      "learning_rate": 3.483050847457627e-05,
      "loss": 3.2564,
      "step": 1074
    },
    {
      "epoch": 0.9110169491525424,
      "grad_norm": 0.9699019193649292,
      "learning_rate": 3.481638418079096e-05,
      "loss": 3.1516,
      "step": 1075
    },
    {
      "epoch": 0.911864406779661,
      "grad_norm": 0.9822274446487427,
      "learning_rate": 3.480225988700565e-05,
      "loss": 3.1477,
      "step": 1076
    },
    {
      "epoch": 0.9127118644067796,
      "grad_norm": 0.8602921962738037,
      "learning_rate": 3.478813559322034e-05,
      "loss": 3.2671,
      "step": 1077
    },
    {
      "epoch": 0.9135593220338983,
      "grad_norm": 1.0001736879348755,
      "learning_rate": 3.477401129943503e-05,
      "loss": 3.056,
      "step": 1078
    },
    {
      "epoch": 0.9144067796610169,
      "grad_norm": 0.9457423686981201,
      "learning_rate": 3.475988700564972e-05,
      "loss": 3.2139,
      "step": 1079
    },
    {
      "epoch": 0.9152542372881356,
      "grad_norm": 1.0191426277160645,
      "learning_rate": 3.474576271186441e-05,
      "loss": 3.231,
      "step": 1080
    },
    {
      "epoch": 0.9161016949152543,
      "grad_norm": 0.9214416146278381,
      "learning_rate": 3.47316384180791e-05,
      "loss": 3.1936,
      "step": 1081
    },
    {
      "epoch": 0.9169491525423729,
      "grad_norm": 1.0589238405227661,
      "learning_rate": 3.471751412429379e-05,
      "loss": 3.0493,
      "step": 1082
    },
    {
      "epoch": 0.9177966101694915,
      "grad_norm": 1.1473020315170288,
      "learning_rate": 3.470338983050847e-05,
      "loss": 3.1314,
      "step": 1083
    },
    {
      "epoch": 0.9186440677966101,
      "grad_norm": 1.0457464456558228,
      "learning_rate": 3.468926553672316e-05,
      "loss": 3.1452,
      "step": 1084
    },
    {
      "epoch": 0.9194915254237288,
      "grad_norm": 0.9404765963554382,
      "learning_rate": 3.467514124293785e-05,
      "loss": 3.1878,
      "step": 1085
    },
    {
      "epoch": 0.9203389830508475,
      "grad_norm": 1.1287540197372437,
      "learning_rate": 3.466101694915254e-05,
      "loss": 3.1953,
      "step": 1086
    },
    {
      "epoch": 0.9211864406779661,
      "grad_norm": 1.0317624807357788,
      "learning_rate": 3.464689265536723e-05,
      "loss": 3.0901,
      "step": 1087
    },
    {
      "epoch": 0.9220338983050848,
      "grad_norm": 1.1479357481002808,
      "learning_rate": 3.463276836158192e-05,
      "loss": 3.2511,
      "step": 1088
    },
    {
      "epoch": 0.9228813559322034,
      "grad_norm": 1.1169977188110352,
      "learning_rate": 3.461864406779661e-05,
      "loss": 3.176,
      "step": 1089
    },
    {
      "epoch": 0.923728813559322,
      "grad_norm": 0.8637784719467163,
      "learning_rate": 3.46045197740113e-05,
      "loss": 3.2857,
      "step": 1090
    },
    {
      "epoch": 0.9245762711864407,
      "grad_norm": 0.860206127166748,
      "learning_rate": 3.459039548022599e-05,
      "loss": 3.3376,
      "step": 1091
    },
    {
      "epoch": 0.9254237288135593,
      "grad_norm": 1.0485258102416992,
      "learning_rate": 3.4576271186440676e-05,
      "loss": 3.155,
      "step": 1092
    },
    {
      "epoch": 0.926271186440678,
      "grad_norm": 0.9788015484809875,
      "learning_rate": 3.4562146892655366e-05,
      "loss": 3.3039,
      "step": 1093
    },
    {
      "epoch": 0.9271186440677966,
      "grad_norm": 0.9749478101730347,
      "learning_rate": 3.4548022598870056e-05,
      "loss": 3.1296,
      "step": 1094
    },
    {
      "epoch": 0.9279661016949152,
      "grad_norm": 1.0621051788330078,
      "learning_rate": 3.4533898305084746e-05,
      "loss": 3.2054,
      "step": 1095
    },
    {
      "epoch": 0.9288135593220339,
      "grad_norm": 0.7976939082145691,
      "learning_rate": 3.4519774011299436e-05,
      "loss": 3.1385,
      "step": 1096
    },
    {
      "epoch": 0.9296610169491526,
      "grad_norm": 1.0727747678756714,
      "learning_rate": 3.4505649717514126e-05,
      "loss": 3.1828,
      "step": 1097
    },
    {
      "epoch": 0.9305084745762712,
      "grad_norm": 0.9779229164123535,
      "learning_rate": 3.4491525423728816e-05,
      "loss": 3.0821,
      "step": 1098
    },
    {
      "epoch": 0.9313559322033899,
      "grad_norm": 0.9867045879364014,
      "learning_rate": 3.4477401129943506e-05,
      "loss": 3.1804,
      "step": 1099
    },
    {
      "epoch": 0.9322033898305084,
      "grad_norm": 0.9286884665489197,
      "learning_rate": 3.446327683615819e-05,
      "loss": 3.2151,
      "step": 1100
    },
    {
      "epoch": 0.9330508474576271,
      "grad_norm": 0.9230116009712219,
      "learning_rate": 3.444915254237288e-05,
      "loss": 3.2612,
      "step": 1101
    },
    {
      "epoch": 0.9338983050847458,
      "grad_norm": 0.9774742126464844,
      "learning_rate": 3.443502824858757e-05,
      "loss": 3.086,
      "step": 1102
    },
    {
      "epoch": 0.9347457627118644,
      "grad_norm": 1.0071637630462646,
      "learning_rate": 3.4420903954802266e-05,
      "loss": 3.1048,
      "step": 1103
    },
    {
      "epoch": 0.9355932203389831,
      "grad_norm": 0.956024706363678,
      "learning_rate": 3.440677966101695e-05,
      "loss": 3.0709,
      "step": 1104
    },
    {
      "epoch": 0.9364406779661016,
      "grad_norm": 0.9846750497817993,
      "learning_rate": 3.439265536723164e-05,
      "loss": 3.2082,
      "step": 1105
    },
    {
      "epoch": 0.9372881355932203,
      "grad_norm": 1.0164185762405396,
      "learning_rate": 3.437853107344633e-05,
      "loss": 3.0972,
      "step": 1106
    },
    {
      "epoch": 0.938135593220339,
      "grad_norm": 1.0777028799057007,
      "learning_rate": 3.436440677966102e-05,
      "loss": 2.9985,
      "step": 1107
    },
    {
      "epoch": 0.9389830508474576,
      "grad_norm": 0.9713720679283142,
      "learning_rate": 3.435028248587571e-05,
      "loss": 3.17,
      "step": 1108
    },
    {
      "epoch": 0.9398305084745763,
      "grad_norm": 0.8038375377655029,
      "learning_rate": 3.433615819209039e-05,
      "loss": 3.2833,
      "step": 1109
    },
    {
      "epoch": 0.940677966101695,
      "grad_norm": 1.069003939628601,
      "learning_rate": 3.432203389830508e-05,
      "loss": 3.1023,
      "step": 1110
    },
    {
      "epoch": 0.9415254237288135,
      "grad_norm": 2.074564218521118,
      "learning_rate": 3.430790960451978e-05,
      "loss": 3.1168,
      "step": 1111
    },
    {
      "epoch": 0.9423728813559322,
      "grad_norm": 1.077620506286621,
      "learning_rate": 3.429378531073447e-05,
      "loss": 3.0685,
      "step": 1112
    },
    {
      "epoch": 0.9432203389830508,
      "grad_norm": 0.9343460202217102,
      "learning_rate": 3.427966101694915e-05,
      "loss": 3.0626,
      "step": 1113
    },
    {
      "epoch": 0.9440677966101695,
      "grad_norm": 1.0174064636230469,
      "learning_rate": 3.426553672316384e-05,
      "loss": 3.1679,
      "step": 1114
    },
    {
      "epoch": 0.9449152542372882,
      "grad_norm": 0.9161520600318909,
      "learning_rate": 3.425141242937853e-05,
      "loss": 3.0145,
      "step": 1115
    },
    {
      "epoch": 0.9457627118644067,
      "grad_norm": 0.9859434962272644,
      "learning_rate": 3.423728813559322e-05,
      "loss": 3.2604,
      "step": 1116
    },
    {
      "epoch": 0.9466101694915254,
      "grad_norm": 1.0903204679489136,
      "learning_rate": 3.4223163841807906e-05,
      "loss": 3.2061,
      "step": 1117
    },
    {
      "epoch": 0.9474576271186441,
      "grad_norm": 0.9489259719848633,
      "learning_rate": 3.4209039548022596e-05,
      "loss": 3.0373,
      "step": 1118
    },
    {
      "epoch": 0.9483050847457627,
      "grad_norm": 1.042888879776001,
      "learning_rate": 3.419491525423729e-05,
      "loss": 3.0331,
      "step": 1119
    },
    {
      "epoch": 0.9491525423728814,
      "grad_norm": 0.9367261528968811,
      "learning_rate": 3.418079096045198e-05,
      "loss": 3.0994,
      "step": 1120
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9779276847839355,
      "learning_rate": 3.4166666666666666e-05,
      "loss": 3.0605,
      "step": 1121
    },
    {
      "epoch": 0.9508474576271186,
      "grad_norm": 0.9787055850028992,
      "learning_rate": 3.4152542372881356e-05,
      "loss": 3.1164,
      "step": 1122
    },
    {
      "epoch": 0.9516949152542373,
      "grad_norm": 0.926442563533783,
      "learning_rate": 3.4138418079096046e-05,
      "loss": 3.1982,
      "step": 1123
    },
    {
      "epoch": 0.9525423728813559,
      "grad_norm": 1.0305582284927368,
      "learning_rate": 3.4124293785310736e-05,
      "loss": 3.1462,
      "step": 1124
    },
    {
      "epoch": 0.9533898305084746,
      "grad_norm": 0.895269513130188,
      "learning_rate": 3.4110169491525426e-05,
      "loss": 3.2023,
      "step": 1125
    },
    {
      "epoch": 0.9542372881355933,
      "grad_norm": 0.8479098081588745,
      "learning_rate": 3.409604519774011e-05,
      "loss": 3.0351,
      "step": 1126
    },
    {
      "epoch": 0.9550847457627119,
      "grad_norm": 0.9858337044715881,
      "learning_rate": 3.4081920903954806e-05,
      "loss": 3.2064,
      "step": 1127
    },
    {
      "epoch": 0.9559322033898305,
      "grad_norm": 0.9453044533729553,
      "learning_rate": 3.4067796610169496e-05,
      "loss": 3.0836,
      "step": 1128
    },
    {
      "epoch": 0.9567796610169491,
      "grad_norm": 0.9639103412628174,
      "learning_rate": 3.4053672316384186e-05,
      "loss": 3.1083,
      "step": 1129
    },
    {
      "epoch": 0.9576271186440678,
      "grad_norm": 0.955803632736206,
      "learning_rate": 3.403954802259887e-05,
      "loss": 3.1033,
      "step": 1130
    },
    {
      "epoch": 0.9584745762711865,
      "grad_norm": 1.1019872426986694,
      "learning_rate": 3.402542372881356e-05,
      "loss": 3.1655,
      "step": 1131
    },
    {
      "epoch": 0.9593220338983051,
      "grad_norm": 0.8813576698303223,
      "learning_rate": 3.401129943502825e-05,
      "loss": 3.1289,
      "step": 1132
    },
    {
      "epoch": 0.9601694915254237,
      "grad_norm": 0.8251473307609558,
      "learning_rate": 3.399717514124294e-05,
      "loss": 3.1034,
      "step": 1133
    },
    {
      "epoch": 0.9610169491525423,
      "grad_norm": 0.9033800363540649,
      "learning_rate": 3.398305084745763e-05,
      "loss": 3.1821,
      "step": 1134
    },
    {
      "epoch": 0.961864406779661,
      "grad_norm": 1.3056793212890625,
      "learning_rate": 3.396892655367232e-05,
      "loss": 3.163,
      "step": 1135
    },
    {
      "epoch": 0.9627118644067797,
      "grad_norm": 0.9215332865715027,
      "learning_rate": 3.395480225988701e-05,
      "loss": 3.1409,
      "step": 1136
    },
    {
      "epoch": 0.9635593220338983,
      "grad_norm": 0.9346233606338501,
      "learning_rate": 3.39406779661017e-05,
      "loss": 3.1261,
      "step": 1137
    },
    {
      "epoch": 0.964406779661017,
      "grad_norm": 0.980396568775177,
      "learning_rate": 3.392655367231639e-05,
      "loss": 3.0834,
      "step": 1138
    },
    {
      "epoch": 0.9652542372881356,
      "grad_norm": 0.9811868071556091,
      "learning_rate": 3.391242937853107e-05,
      "loss": 3.2059,
      "step": 1139
    },
    {
      "epoch": 0.9661016949152542,
      "grad_norm": 0.9104767441749573,
      "learning_rate": 3.389830508474576e-05,
      "loss": 3.0948,
      "step": 1140
    },
    {
      "epoch": 0.9669491525423729,
      "grad_norm": 0.9422304630279541,
      "learning_rate": 3.388418079096045e-05,
      "loss": 3.3007,
      "step": 1141
    },
    {
      "epoch": 0.9677966101694915,
      "grad_norm": 1.0825037956237793,
      "learning_rate": 3.387005649717514e-05,
      "loss": 3.2037,
      "step": 1142
    },
    {
      "epoch": 0.9686440677966102,
      "grad_norm": 0.8510132431983948,
      "learning_rate": 3.385593220338983e-05,
      "loss": 3.1312,
      "step": 1143
    },
    {
      "epoch": 0.9694915254237289,
      "grad_norm": 0.8678553700447083,
      "learning_rate": 3.384180790960452e-05,
      "loss": 3.0714,
      "step": 1144
    },
    {
      "epoch": 0.9703389830508474,
      "grad_norm": 0.9524672627449036,
      "learning_rate": 3.382768361581921e-05,
      "loss": 3.1557,
      "step": 1145
    },
    {
      "epoch": 0.9711864406779661,
      "grad_norm": 0.8765469193458557,
      "learning_rate": 3.38135593220339e-05,
      "loss": 3.2833,
      "step": 1146
    },
    {
      "epoch": 0.9720338983050848,
      "grad_norm": 1.1473183631896973,
      "learning_rate": 3.3799435028248585e-05,
      "loss": 3.1759,
      "step": 1147
    },
    {
      "epoch": 0.9728813559322034,
      "grad_norm": 1.138814926147461,
      "learning_rate": 3.3785310734463275e-05,
      "loss": 3.1647,
      "step": 1148
    },
    {
      "epoch": 0.9737288135593221,
      "grad_norm": 1.0071004629135132,
      "learning_rate": 3.3771186440677965e-05,
      "loss": 3.1534,
      "step": 1149
    },
    {
      "epoch": 0.9745762711864406,
      "grad_norm": 1.166270136833191,
      "learning_rate": 3.375706214689266e-05,
      "loss": 3.1155,
      "step": 1150
    },
    {
      "epoch": 0.9754237288135593,
      "grad_norm": 0.9369944930076599,
      "learning_rate": 3.3742937853107345e-05,
      "loss": 3.0024,
      "step": 1151
    },
    {
      "epoch": 0.976271186440678,
      "grad_norm": 0.9275339245796204,
      "learning_rate": 3.3728813559322035e-05,
      "loss": 3.0509,
      "step": 1152
    },
    {
      "epoch": 0.9771186440677966,
      "grad_norm": 0.8760161399841309,
      "learning_rate": 3.3714689265536725e-05,
      "loss": 3.2487,
      "step": 1153
    },
    {
      "epoch": 0.9779661016949153,
      "grad_norm": 1.0563088655471802,
      "learning_rate": 3.3700564971751415e-05,
      "loss": 3.1959,
      "step": 1154
    },
    {
      "epoch": 0.9788135593220338,
      "grad_norm": 0.9117673635482788,
      "learning_rate": 3.3686440677966105e-05,
      "loss": 3.1583,
      "step": 1155
    },
    {
      "epoch": 0.9796610169491525,
      "grad_norm": 1.1466143131256104,
      "learning_rate": 3.367231638418079e-05,
      "loss": 3.1284,
      "step": 1156
    },
    {
      "epoch": 0.9805084745762712,
      "grad_norm": 0.8947937488555908,
      "learning_rate": 3.365819209039548e-05,
      "loss": 3.0446,
      "step": 1157
    },
    {
      "epoch": 0.9813559322033898,
      "grad_norm": 0.9606156945228577,
      "learning_rate": 3.3644067796610175e-05,
      "loss": 3.1659,
      "step": 1158
    },
    {
      "epoch": 0.9822033898305085,
      "grad_norm": 1.0145713090896606,
      "learning_rate": 3.3629943502824865e-05,
      "loss": 3.2073,
      "step": 1159
    },
    {
      "epoch": 0.9830508474576272,
      "grad_norm": 0.8798396587371826,
      "learning_rate": 3.361581920903955e-05,
      "loss": 3.06,
      "step": 1160
    },
    {
      "epoch": 0.9838983050847457,
      "grad_norm": 0.9526357054710388,
      "learning_rate": 3.360169491525424e-05,
      "loss": 3.1306,
      "step": 1161
    },
    {
      "epoch": 0.9847457627118644,
      "grad_norm": 0.967297375202179,
      "learning_rate": 3.358757062146893e-05,
      "loss": 3.0771,
      "step": 1162
    },
    {
      "epoch": 0.985593220338983,
      "grad_norm": 1.0536571741104126,
      "learning_rate": 3.357344632768362e-05,
      "loss": 3.1413,
      "step": 1163
    },
    {
      "epoch": 0.9864406779661017,
      "grad_norm": 0.9622642397880554,
      "learning_rate": 3.35593220338983e-05,
      "loss": 3.0232,
      "step": 1164
    },
    {
      "epoch": 0.9872881355932204,
      "grad_norm": 1.0252420902252197,
      "learning_rate": 3.354519774011299e-05,
      "loss": 3.1725,
      "step": 1165
    },
    {
      "epoch": 0.988135593220339,
      "grad_norm": 1.169731616973877,
      "learning_rate": 3.353107344632769e-05,
      "loss": 3.1866,
      "step": 1166
    },
    {
      "epoch": 0.9889830508474576,
      "grad_norm": 1.0889506340026855,
      "learning_rate": 3.351694915254238e-05,
      "loss": 3.0281,
      "step": 1167
    },
    {
      "epoch": 0.9898305084745763,
      "grad_norm": 1.040248990058899,
      "learning_rate": 3.350282485875706e-05,
      "loss": 3.1935,
      "step": 1168
    },
    {
      "epoch": 0.9906779661016949,
      "grad_norm": 0.9672927260398865,
      "learning_rate": 3.348870056497175e-05,
      "loss": 3.2079,
      "step": 1169
    },
    {
      "epoch": 0.9915254237288136,
      "grad_norm": 0.9489609599113464,
      "learning_rate": 3.347457627118644e-05,
      "loss": 3.1211,
      "step": 1170
    },
    {
      "epoch": 0.9923728813559322,
      "grad_norm": 1.0373070240020752,
      "learning_rate": 3.346045197740113e-05,
      "loss": 3.1851,
      "step": 1171
    },
    {
      "epoch": 0.9932203389830508,
      "grad_norm": 0.9559980630874634,
      "learning_rate": 3.344632768361582e-05,
      "loss": 3.1407,
      "step": 1172
    },
    {
      "epoch": 0.9940677966101695,
      "grad_norm": 1.0616077184677124,
      "learning_rate": 3.3432203389830505e-05,
      "loss": 3.0577,
      "step": 1173
    },
    {
      "epoch": 0.9949152542372881,
      "grad_norm": 1.0753858089447021,
      "learning_rate": 3.34180790960452e-05,
      "loss": 3.2391,
      "step": 1174
    },
    {
      "epoch": 0.9957627118644068,
      "grad_norm": 0.9057319164276123,
      "learning_rate": 3.340395480225989e-05,
      "loss": 3.1968,
      "step": 1175
    },
    {
      "epoch": 0.9966101694915255,
      "grad_norm": 0.9730505347251892,
      "learning_rate": 3.338983050847458e-05,
      "loss": 3.1794,
      "step": 1176
    },
    {
      "epoch": 0.997457627118644,
      "grad_norm": 1.2745962142944336,
      "learning_rate": 3.3375706214689265e-05,
      "loss": 3.0877,
      "step": 1177
    },
    {
      "epoch": 0.9983050847457627,
      "grad_norm": 1.0337282419204712,
      "learning_rate": 3.3361581920903955e-05,
      "loss": 3.1717,
      "step": 1178
    },
    {
      "epoch": 0.9991525423728813,
      "grad_norm": 0.8962303996086121,
      "learning_rate": 3.3347457627118645e-05,
      "loss": 3.089,
      "step": 1179
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8768669962882996,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 3.1633,
      "step": 1180
    },
    {
      "epoch": 1.0008474576271187,
      "grad_norm": 1.1547901630401611,
      "learning_rate": 3.3319209039548025e-05,
      "loss": 2.9119,
      "step": 1181
    },
    {
      "epoch": 1.0016949152542374,
      "grad_norm": 1.2839479446411133,
      "learning_rate": 3.3305084745762715e-05,
      "loss": 2.7222,
      "step": 1182
    },
    {
      "epoch": 1.0025423728813558,
      "grad_norm": 1.108107566833496,
      "learning_rate": 3.3290960451977405e-05,
      "loss": 2.8815,
      "step": 1183
    },
    {
      "epoch": 1.0033898305084745,
      "grad_norm": 1.4743744134902954,
      "learning_rate": 3.3276836158192095e-05,
      "loss": 2.7498,
      "step": 1184
    },
    {
      "epoch": 1.0042372881355932,
      "grad_norm": 1.104684829711914,
      "learning_rate": 3.326271186440678e-05,
      "loss": 2.6521,
      "step": 1185
    },
    {
      "epoch": 1.005084745762712,
      "grad_norm": 1.1616811752319336,
      "learning_rate": 3.324858757062147e-05,
      "loss": 2.7501,
      "step": 1186
    },
    {
      "epoch": 1.0059322033898306,
      "grad_norm": 0.9652979373931885,
      "learning_rate": 3.323446327683616e-05,
      "loss": 2.8519,
      "step": 1187
    },
    {
      "epoch": 1.006779661016949,
      "grad_norm": 1.0420432090759277,
      "learning_rate": 3.322033898305085e-05,
      "loss": 2.6377,
      "step": 1188
    },
    {
      "epoch": 1.0076271186440677,
      "grad_norm": 1.0684734582901,
      "learning_rate": 3.320621468926554e-05,
      "loss": 2.7969,
      "step": 1189
    },
    {
      "epoch": 1.0084745762711864,
      "grad_norm": 1.0408910512924194,
      "learning_rate": 3.319209039548023e-05,
      "loss": 2.7661,
      "step": 1190
    },
    {
      "epoch": 1.009322033898305,
      "grad_norm": 0.9833167195320129,
      "learning_rate": 3.317796610169492e-05,
      "loss": 2.8351,
      "step": 1191
    },
    {
      "epoch": 1.0101694915254238,
      "grad_norm": 1.2136975526809692,
      "learning_rate": 3.316384180790961e-05,
      "loss": 2.7654,
      "step": 1192
    },
    {
      "epoch": 1.0110169491525425,
      "grad_norm": 1.0686248540878296,
      "learning_rate": 3.31497175141243e-05,
      "loss": 2.7673,
      "step": 1193
    },
    {
      "epoch": 1.011864406779661,
      "grad_norm": 1.084439992904663,
      "learning_rate": 3.313559322033898e-05,
      "loss": 2.8005,
      "step": 1194
    },
    {
      "epoch": 1.0127118644067796,
      "grad_norm": 1.1534408330917358,
      "learning_rate": 3.312146892655367e-05,
      "loss": 2.8348,
      "step": 1195
    },
    {
      "epoch": 1.0135593220338983,
      "grad_norm": 1.188952922821045,
      "learning_rate": 3.310734463276836e-05,
      "loss": 2.7189,
      "step": 1196
    },
    {
      "epoch": 1.014406779661017,
      "grad_norm": 1.1142390966415405,
      "learning_rate": 3.309322033898305e-05,
      "loss": 2.842,
      "step": 1197
    },
    {
      "epoch": 1.0152542372881357,
      "grad_norm": 1.0218876600265503,
      "learning_rate": 3.307909604519774e-05,
      "loss": 2.985,
      "step": 1198
    },
    {
      "epoch": 1.0161016949152541,
      "grad_norm": 1.0653412342071533,
      "learning_rate": 3.306497175141243e-05,
      "loss": 2.7049,
      "step": 1199
    },
    {
      "epoch": 1.0169491525423728,
      "grad_norm": 1.1319935321807861,
      "learning_rate": 3.305084745762712e-05,
      "loss": 2.594,
      "step": 1200
    },
    {
      "epoch": 1.0177966101694915,
      "grad_norm": 1.013126015663147,
      "learning_rate": 3.303672316384181e-05,
      "loss": 2.8042,
      "step": 1201
    },
    {
      "epoch": 1.0186440677966102,
      "grad_norm": 0.9670597314834595,
      "learning_rate": 3.30225988700565e-05,
      "loss": 2.7315,
      "step": 1202
    },
    {
      "epoch": 1.019491525423729,
      "grad_norm": 0.9469462633132935,
      "learning_rate": 3.3008474576271184e-05,
      "loss": 2.8303,
      "step": 1203
    },
    {
      "epoch": 1.0203389830508474,
      "grad_norm": 1.1374952793121338,
      "learning_rate": 3.2994350282485874e-05,
      "loss": 2.8758,
      "step": 1204
    },
    {
      "epoch": 1.021186440677966,
      "grad_norm": 1.1895190477371216,
      "learning_rate": 3.2980225988700564e-05,
      "loss": 2.7217,
      "step": 1205
    },
    {
      "epoch": 1.0220338983050847,
      "grad_norm": 1.0638797283172607,
      "learning_rate": 3.296610169491526e-05,
      "loss": 2.7883,
      "step": 1206
    },
    {
      "epoch": 1.0228813559322034,
      "grad_norm": 1.0590007305145264,
      "learning_rate": 3.2951977401129944e-05,
      "loss": 2.7737,
      "step": 1207
    },
    {
      "epoch": 1.023728813559322,
      "grad_norm": 1.0253334045410156,
      "learning_rate": 3.2937853107344634e-05,
      "loss": 2.7251,
      "step": 1208
    },
    {
      "epoch": 1.0245762711864406,
      "grad_norm": 1.1091796159744263,
      "learning_rate": 3.2923728813559324e-05,
      "loss": 2.723,
      "step": 1209
    },
    {
      "epoch": 1.0254237288135593,
      "grad_norm": 1.2076777219772339,
      "learning_rate": 3.2909604519774014e-05,
      "loss": 2.628,
      "step": 1210
    },
    {
      "epoch": 1.026271186440678,
      "grad_norm": 1.097025990486145,
      "learning_rate": 3.28954802259887e-05,
      "loss": 2.7423,
      "step": 1211
    },
    {
      "epoch": 1.0271186440677966,
      "grad_norm": 0.9871625304222107,
      "learning_rate": 3.288135593220339e-05,
      "loss": 2.9542,
      "step": 1212
    },
    {
      "epoch": 1.0279661016949153,
      "grad_norm": 1.4033690690994263,
      "learning_rate": 3.2867231638418084e-05,
      "loss": 2.408,
      "step": 1213
    },
    {
      "epoch": 1.028813559322034,
      "grad_norm": 1.1936814785003662,
      "learning_rate": 3.2853107344632774e-05,
      "loss": 2.6728,
      "step": 1214
    },
    {
      "epoch": 1.0296610169491525,
      "grad_norm": 1.0251595973968506,
      "learning_rate": 3.283898305084746e-05,
      "loss": 2.8915,
      "step": 1215
    },
    {
      "epoch": 1.0305084745762711,
      "grad_norm": 1.038507342338562,
      "learning_rate": 3.282485875706215e-05,
      "loss": 2.6211,
      "step": 1216
    },
    {
      "epoch": 1.0313559322033898,
      "grad_norm": 1.195397138595581,
      "learning_rate": 3.281073446327684e-05,
      "loss": 2.691,
      "step": 1217
    },
    {
      "epoch": 1.0322033898305085,
      "grad_norm": 1.0375072956085205,
      "learning_rate": 3.279661016949153e-05,
      "loss": 2.7848,
      "step": 1218
    },
    {
      "epoch": 1.0330508474576272,
      "grad_norm": 1.1907418966293335,
      "learning_rate": 3.278248587570622e-05,
      "loss": 2.6764,
      "step": 1219
    },
    {
      "epoch": 1.0338983050847457,
      "grad_norm": 1.1549733877182007,
      "learning_rate": 3.27683615819209e-05,
      "loss": 2.7625,
      "step": 1220
    },
    {
      "epoch": 1.0347457627118644,
      "grad_norm": 1.0455302000045776,
      "learning_rate": 3.27542372881356e-05,
      "loss": 2.7691,
      "step": 1221
    },
    {
      "epoch": 1.035593220338983,
      "grad_norm": 1.2212104797363281,
      "learning_rate": 3.274011299435029e-05,
      "loss": 2.7455,
      "step": 1222
    },
    {
      "epoch": 1.0364406779661017,
      "grad_norm": 1.0414098501205444,
      "learning_rate": 3.272598870056498e-05,
      "loss": 2.7957,
      "step": 1223
    },
    {
      "epoch": 1.0372881355932204,
      "grad_norm": 0.9203545451164246,
      "learning_rate": 3.271186440677966e-05,
      "loss": 2.8824,
      "step": 1224
    },
    {
      "epoch": 1.0381355932203389,
      "grad_norm": 1.130113124847412,
      "learning_rate": 3.269774011299435e-05,
      "loss": 2.7725,
      "step": 1225
    },
    {
      "epoch": 1.0389830508474576,
      "grad_norm": 0.9704325199127197,
      "learning_rate": 3.268361581920904e-05,
      "loss": 2.7973,
      "step": 1226
    },
    {
      "epoch": 1.0398305084745763,
      "grad_norm": 0.9468993544578552,
      "learning_rate": 3.266949152542373e-05,
      "loss": 2.8226,
      "step": 1227
    },
    {
      "epoch": 1.040677966101695,
      "grad_norm": 1.0267146825790405,
      "learning_rate": 3.2655367231638414e-05,
      "loss": 2.7649,
      "step": 1228
    },
    {
      "epoch": 1.0415254237288136,
      "grad_norm": 1.0473774671554565,
      "learning_rate": 3.264124293785311e-05,
      "loss": 2.708,
      "step": 1229
    },
    {
      "epoch": 1.042372881355932,
      "grad_norm": 0.9830741286277771,
      "learning_rate": 3.26271186440678e-05,
      "loss": 2.7738,
      "step": 1230
    },
    {
      "epoch": 1.0432203389830508,
      "grad_norm": 1.0539050102233887,
      "learning_rate": 3.261299435028249e-05,
      "loss": 2.7942,
      "step": 1231
    },
    {
      "epoch": 1.0440677966101695,
      "grad_norm": 1.2233527898788452,
      "learning_rate": 3.2598870056497174e-05,
      "loss": 2.6881,
      "step": 1232
    },
    {
      "epoch": 1.0449152542372881,
      "grad_norm": 1.1111575365066528,
      "learning_rate": 3.2584745762711864e-05,
      "loss": 2.7387,
      "step": 1233
    },
    {
      "epoch": 1.0457627118644068,
      "grad_norm": 1.2477937936782837,
      "learning_rate": 3.2570621468926554e-05,
      "loss": 2.6672,
      "step": 1234
    },
    {
      "epoch": 1.0466101694915255,
      "grad_norm": 1.075348138809204,
      "learning_rate": 3.2556497175141244e-05,
      "loss": 2.81,
      "step": 1235
    },
    {
      "epoch": 1.047457627118644,
      "grad_norm": 1.1108568906784058,
      "learning_rate": 3.2542372881355934e-05,
      "loss": 2.635,
      "step": 1236
    },
    {
      "epoch": 1.0483050847457627,
      "grad_norm": 1.04779851436615,
      "learning_rate": 3.2528248587570624e-05,
      "loss": 2.9208,
      "step": 1237
    },
    {
      "epoch": 1.0491525423728814,
      "grad_norm": 1.3098849058151245,
      "learning_rate": 3.2514124293785314e-05,
      "loss": 2.6156,
      "step": 1238
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.1287405490875244,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 2.6735,
      "step": 1239
    },
    {
      "epoch": 1.0508474576271187,
      "grad_norm": 1.1208323240280151,
      "learning_rate": 3.2485875706214694e-05,
      "loss": 2.7031,
      "step": 1240
    },
    {
      "epoch": 1.0516949152542372,
      "grad_norm": 1.1564394235610962,
      "learning_rate": 3.247175141242938e-05,
      "loss": 2.7012,
      "step": 1241
    },
    {
      "epoch": 1.0525423728813559,
      "grad_norm": 1.0639265775680542,
      "learning_rate": 3.245762711864407e-05,
      "loss": 2.7558,
      "step": 1242
    },
    {
      "epoch": 1.0533898305084746,
      "grad_norm": 1.3013960123062134,
      "learning_rate": 3.244350282485876e-05,
      "loss": 2.7392,
      "step": 1243
    },
    {
      "epoch": 1.0542372881355933,
      "grad_norm": 1.2004514932632446,
      "learning_rate": 3.242937853107345e-05,
      "loss": 2.6263,
      "step": 1244
    },
    {
      "epoch": 1.055084745762712,
      "grad_norm": 1.1411895751953125,
      "learning_rate": 3.241525423728814e-05,
      "loss": 2.7157,
      "step": 1245
    },
    {
      "epoch": 1.0559322033898304,
      "grad_norm": 0.9765958189964294,
      "learning_rate": 3.240112994350283e-05,
      "loss": 2.7072,
      "step": 1246
    },
    {
      "epoch": 1.056779661016949,
      "grad_norm": 1.0957763195037842,
      "learning_rate": 3.238700564971752e-05,
      "loss": 2.8281,
      "step": 1247
    },
    {
      "epoch": 1.0576271186440678,
      "grad_norm": 0.9992038607597351,
      "learning_rate": 3.237288135593221e-05,
      "loss": 2.8777,
      "step": 1248
    },
    {
      "epoch": 1.0584745762711865,
      "grad_norm": 1.152011513710022,
      "learning_rate": 3.23587570621469e-05,
      "loss": 2.6419,
      "step": 1249
    },
    {
      "epoch": 1.0593220338983051,
      "grad_norm": 1.0962355136871338,
      "learning_rate": 3.234463276836158e-05,
      "loss": 2.7468,
      "step": 1250
    },
    {
      "epoch": 1.0601694915254238,
      "grad_norm": 1.2691516876220703,
      "learning_rate": 3.233050847457627e-05,
      "loss": 2.7081,
      "step": 1251
    },
    {
      "epoch": 1.0610169491525423,
      "grad_norm": 0.8663666844367981,
      "learning_rate": 3.231638418079096e-05,
      "loss": 2.8186,
      "step": 1252
    },
    {
      "epoch": 1.061864406779661,
      "grad_norm": 1.0920077562332153,
      "learning_rate": 3.230225988700565e-05,
      "loss": 2.9159,
      "step": 1253
    },
    {
      "epoch": 1.0627118644067797,
      "grad_norm": 1.0791915655136108,
      "learning_rate": 3.228813559322034e-05,
      "loss": 2.8422,
      "step": 1254
    },
    {
      "epoch": 1.0635593220338984,
      "grad_norm": 1.2146872282028198,
      "learning_rate": 3.227401129943503e-05,
      "loss": 2.5906,
      "step": 1255
    },
    {
      "epoch": 1.064406779661017,
      "grad_norm": 1.0364774465560913,
      "learning_rate": 3.225988700564972e-05,
      "loss": 2.6943,
      "step": 1256
    },
    {
      "epoch": 1.0652542372881355,
      "grad_norm": 1.1046572923660278,
      "learning_rate": 3.224576271186441e-05,
      "loss": 2.7248,
      "step": 1257
    },
    {
      "epoch": 1.0661016949152542,
      "grad_norm": 1.0412293672561646,
      "learning_rate": 3.2231638418079093e-05,
      "loss": 2.8076,
      "step": 1258
    },
    {
      "epoch": 1.0669491525423729,
      "grad_norm": 1.0001367330551147,
      "learning_rate": 3.2217514124293783e-05,
      "loss": 2.8023,
      "step": 1259
    },
    {
      "epoch": 1.0677966101694916,
      "grad_norm": 1.0104426145553589,
      "learning_rate": 3.2203389830508473e-05,
      "loss": 2.8011,
      "step": 1260
    },
    {
      "epoch": 1.0686440677966103,
      "grad_norm": 1.2797577381134033,
      "learning_rate": 3.218926553672317e-05,
      "loss": 2.6239,
      "step": 1261
    },
    {
      "epoch": 1.0694915254237287,
      "grad_norm": 0.9057015776634216,
      "learning_rate": 3.2175141242937853e-05,
      "loss": 2.9027,
      "step": 1262
    },
    {
      "epoch": 1.0703389830508474,
      "grad_norm": 1.0806961059570312,
      "learning_rate": 3.2161016949152543e-05,
      "loss": 2.7245,
      "step": 1263
    },
    {
      "epoch": 1.071186440677966,
      "grad_norm": 0.9371135830879211,
      "learning_rate": 3.2146892655367233e-05,
      "loss": 2.8056,
      "step": 1264
    },
    {
      "epoch": 1.0720338983050848,
      "grad_norm": 1.0706758499145508,
      "learning_rate": 3.2132768361581923e-05,
      "loss": 2.7724,
      "step": 1265
    },
    {
      "epoch": 1.0728813559322035,
      "grad_norm": 1.069041132926941,
      "learning_rate": 3.2118644067796613e-05,
      "loss": 2.7142,
      "step": 1266
    },
    {
      "epoch": 1.0737288135593221,
      "grad_norm": 1.1043695211410522,
      "learning_rate": 3.21045197740113e-05,
      "loss": 2.6181,
      "step": 1267
    },
    {
      "epoch": 1.0745762711864406,
      "grad_norm": 0.9835469722747803,
      "learning_rate": 3.209039548022599e-05,
      "loss": 2.781,
      "step": 1268
    },
    {
      "epoch": 1.0754237288135593,
      "grad_norm": 1.2907835245132446,
      "learning_rate": 3.2076271186440683e-05,
      "loss": 2.501,
      "step": 1269
    },
    {
      "epoch": 1.076271186440678,
      "grad_norm": 1.1371930837631226,
      "learning_rate": 3.2062146892655373e-05,
      "loss": 2.6317,
      "step": 1270
    },
    {
      "epoch": 1.0771186440677967,
      "grad_norm": 0.9557111263275146,
      "learning_rate": 3.204802259887006e-05,
      "loss": 2.9196,
      "step": 1271
    },
    {
      "epoch": 1.0779661016949154,
      "grad_norm": 1.3277744054794312,
      "learning_rate": 3.203389830508475e-05,
      "loss": 2.6195,
      "step": 1272
    },
    {
      "epoch": 1.0788135593220338,
      "grad_norm": 1.0682483911514282,
      "learning_rate": 3.201977401129944e-05,
      "loss": 2.6922,
      "step": 1273
    },
    {
      "epoch": 1.0796610169491525,
      "grad_norm": 1.1835170984268188,
      "learning_rate": 3.200564971751413e-05,
      "loss": 2.6221,
      "step": 1274
    },
    {
      "epoch": 1.0805084745762712,
      "grad_norm": 1.1254576444625854,
      "learning_rate": 3.199152542372881e-05,
      "loss": 2.7208,
      "step": 1275
    },
    {
      "epoch": 1.0813559322033899,
      "grad_norm": 1.0872687101364136,
      "learning_rate": 3.19774011299435e-05,
      "loss": 2.7985,
      "step": 1276
    },
    {
      "epoch": 1.0822033898305086,
      "grad_norm": 1.0942344665527344,
      "learning_rate": 3.1963276836158197e-05,
      "loss": 2.6473,
      "step": 1277
    },
    {
      "epoch": 1.083050847457627,
      "grad_norm": 1.2729380130767822,
      "learning_rate": 3.1949152542372887e-05,
      "loss": 2.6103,
      "step": 1278
    },
    {
      "epoch": 1.0838983050847457,
      "grad_norm": 1.2173162698745728,
      "learning_rate": 3.193502824858757e-05,
      "loss": 2.573,
      "step": 1279
    },
    {
      "epoch": 1.0847457627118644,
      "grad_norm": 1.0390675067901611,
      "learning_rate": 3.192090395480226e-05,
      "loss": 2.7393,
      "step": 1280
    },
    {
      "epoch": 1.085593220338983,
      "grad_norm": 0.9719063639640808,
      "learning_rate": 3.190677966101695e-05,
      "loss": 2.793,
      "step": 1281
    },
    {
      "epoch": 1.0864406779661018,
      "grad_norm": 1.2492557764053345,
      "learning_rate": 3.189265536723164e-05,
      "loss": 2.5754,
      "step": 1282
    },
    {
      "epoch": 1.0872881355932202,
      "grad_norm": 1.089656949043274,
      "learning_rate": 3.187853107344633e-05,
      "loss": 2.8917,
      "step": 1283
    },
    {
      "epoch": 1.088135593220339,
      "grad_norm": 1.071498155593872,
      "learning_rate": 3.186440677966101e-05,
      "loss": 2.8634,
      "step": 1284
    },
    {
      "epoch": 1.0889830508474576,
      "grad_norm": 1.124239444732666,
      "learning_rate": 3.185028248587571e-05,
      "loss": 2.7158,
      "step": 1285
    },
    {
      "epoch": 1.0898305084745763,
      "grad_norm": 1.1056195497512817,
      "learning_rate": 3.18361581920904e-05,
      "loss": 2.7763,
      "step": 1286
    },
    {
      "epoch": 1.090677966101695,
      "grad_norm": 0.9646786451339722,
      "learning_rate": 3.182203389830509e-05,
      "loss": 2.9494,
      "step": 1287
    },
    {
      "epoch": 1.0915254237288137,
      "grad_norm": 1.1487090587615967,
      "learning_rate": 3.180790960451977e-05,
      "loss": 2.82,
      "step": 1288
    },
    {
      "epoch": 1.0923728813559321,
      "grad_norm": 1.1627110242843628,
      "learning_rate": 3.179378531073446e-05,
      "loss": 2.6772,
      "step": 1289
    },
    {
      "epoch": 1.0932203389830508,
      "grad_norm": 0.9802050590515137,
      "learning_rate": 3.177966101694915e-05,
      "loss": 2.7999,
      "step": 1290
    },
    {
      "epoch": 1.0940677966101695,
      "grad_norm": 1.009687066078186,
      "learning_rate": 3.176553672316384e-05,
      "loss": 2.9195,
      "step": 1291
    },
    {
      "epoch": 1.0949152542372882,
      "grad_norm": 1.2086440324783325,
      "learning_rate": 3.175141242937853e-05,
      "loss": 2.6826,
      "step": 1292
    },
    {
      "epoch": 1.0957627118644069,
      "grad_norm": 0.9848881363868713,
      "learning_rate": 3.173728813559322e-05,
      "loss": 2.8132,
      "step": 1293
    },
    {
      "epoch": 1.0966101694915253,
      "grad_norm": 1.3762964010238647,
      "learning_rate": 3.172316384180791e-05,
      "loss": 2.532,
      "step": 1294
    },
    {
      "epoch": 1.097457627118644,
      "grad_norm": 1.1501647233963013,
      "learning_rate": 3.17090395480226e-05,
      "loss": 2.6281,
      "step": 1295
    },
    {
      "epoch": 1.0983050847457627,
      "grad_norm": 1.080271601676941,
      "learning_rate": 3.169491525423729e-05,
      "loss": 2.8009,
      "step": 1296
    },
    {
      "epoch": 1.0991525423728814,
      "grad_norm": 1.1762102842330933,
      "learning_rate": 3.1680790960451976e-05,
      "loss": 2.6938,
      "step": 1297
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.2100443840026855,
      "learning_rate": 3.1666666666666666e-05,
      "loss": 2.5648,
      "step": 1298
    },
    {
      "epoch": 1.1008474576271186,
      "grad_norm": 0.9676806330680847,
      "learning_rate": 3.1652542372881356e-05,
      "loss": 2.8492,
      "step": 1299
    },
    {
      "epoch": 1.1016949152542372,
      "grad_norm": 1.1340441703796387,
      "learning_rate": 3.1638418079096046e-05,
      "loss": 2.6525,
      "step": 1300
    },
    {
      "epoch": 1.102542372881356,
      "grad_norm": 1.207497239112854,
      "learning_rate": 3.1624293785310736e-05,
      "loss": 2.6985,
      "step": 1301
    },
    {
      "epoch": 1.1033898305084746,
      "grad_norm": 1.3901035785675049,
      "learning_rate": 3.1610169491525426e-05,
      "loss": 2.5859,
      "step": 1302
    },
    {
      "epoch": 1.1042372881355933,
      "grad_norm": 1.3611477613449097,
      "learning_rate": 3.1596045197740116e-05,
      "loss": 2.7371,
      "step": 1303
    },
    {
      "epoch": 1.1050847457627118,
      "grad_norm": 1.0604912042617798,
      "learning_rate": 3.1581920903954806e-05,
      "loss": 2.7643,
      "step": 1304
    },
    {
      "epoch": 1.1059322033898304,
      "grad_norm": 1.2928720712661743,
      "learning_rate": 3.156779661016949e-05,
      "loss": 2.5703,
      "step": 1305
    },
    {
      "epoch": 1.1067796610169491,
      "grad_norm": 0.9556746482849121,
      "learning_rate": 3.155367231638418e-05,
      "loss": 2.7694,
      "step": 1306
    },
    {
      "epoch": 1.1076271186440678,
      "grad_norm": 1.0268397331237793,
      "learning_rate": 3.153954802259887e-05,
      "loss": 2.7441,
      "step": 1307
    },
    {
      "epoch": 1.1084745762711865,
      "grad_norm": 1.4118244647979736,
      "learning_rate": 3.1525423728813566e-05,
      "loss": 2.5304,
      "step": 1308
    },
    {
      "epoch": 1.1093220338983052,
      "grad_norm": 1.2121537923812866,
      "learning_rate": 3.151129943502825e-05,
      "loss": 2.5853,
      "step": 1309
    },
    {
      "epoch": 1.1101694915254237,
      "grad_norm": 1.0660877227783203,
      "learning_rate": 3.149717514124294e-05,
      "loss": 2.8251,
      "step": 1310
    },
    {
      "epoch": 1.1110169491525423,
      "grad_norm": 1.2280398607254028,
      "learning_rate": 3.148305084745763e-05,
      "loss": 2.7137,
      "step": 1311
    },
    {
      "epoch": 1.111864406779661,
      "grad_norm": 1.3553054332733154,
      "learning_rate": 3.146892655367232e-05,
      "loss": 2.5988,
      "step": 1312
    },
    {
      "epoch": 1.1127118644067797,
      "grad_norm": 1.094398021697998,
      "learning_rate": 3.145480225988701e-05,
      "loss": 2.7889,
      "step": 1313
    },
    {
      "epoch": 1.1135593220338984,
      "grad_norm": 1.17271888256073,
      "learning_rate": 3.144067796610169e-05,
      "loss": 2.753,
      "step": 1314
    },
    {
      "epoch": 1.1144067796610169,
      "grad_norm": 1.3248339891433716,
      "learning_rate": 3.142655367231638e-05,
      "loss": 2.5908,
      "step": 1315
    },
    {
      "epoch": 1.1152542372881356,
      "grad_norm": 1.353985071182251,
      "learning_rate": 3.141242937853108e-05,
      "loss": 2.8259,
      "step": 1316
    },
    {
      "epoch": 1.1161016949152542,
      "grad_norm": 1.0256413221359253,
      "learning_rate": 3.139830508474577e-05,
      "loss": 2.9938,
      "step": 1317
    },
    {
      "epoch": 1.116949152542373,
      "grad_norm": 1.1994404792785645,
      "learning_rate": 3.138418079096045e-05,
      "loss": 2.7358,
      "step": 1318
    },
    {
      "epoch": 1.1177966101694916,
      "grad_norm": 1.0351284742355347,
      "learning_rate": 3.137005649717514e-05,
      "loss": 2.7658,
      "step": 1319
    },
    {
      "epoch": 1.11864406779661,
      "grad_norm": 0.8627428412437439,
      "learning_rate": 3.135593220338983e-05,
      "loss": 3.0224,
      "step": 1320
    },
    {
      "epoch": 1.1194915254237288,
      "grad_norm": 0.9812117218971252,
      "learning_rate": 3.134180790960452e-05,
      "loss": 2.879,
      "step": 1321
    },
    {
      "epoch": 1.1203389830508474,
      "grad_norm": 1.0016770362854004,
      "learning_rate": 3.1327683615819206e-05,
      "loss": 2.7623,
      "step": 1322
    },
    {
      "epoch": 1.1211864406779661,
      "grad_norm": 1.1582958698272705,
      "learning_rate": 3.1313559322033896e-05,
      "loss": 2.7582,
      "step": 1323
    },
    {
      "epoch": 1.1220338983050848,
      "grad_norm": 1.0659432411193848,
      "learning_rate": 3.129943502824859e-05,
      "loss": 2.7713,
      "step": 1324
    },
    {
      "epoch": 1.1228813559322033,
      "grad_norm": 1.0960432291030884,
      "learning_rate": 3.128531073446328e-05,
      "loss": 2.6518,
      "step": 1325
    },
    {
      "epoch": 1.123728813559322,
      "grad_norm": 1.0338033437728882,
      "learning_rate": 3.1271186440677966e-05,
      "loss": 2.7533,
      "step": 1326
    },
    {
      "epoch": 1.1245762711864407,
      "grad_norm": 1.039915919303894,
      "learning_rate": 3.1257062146892656e-05,
      "loss": 2.7186,
      "step": 1327
    },
    {
      "epoch": 1.1254237288135593,
      "grad_norm": 1.2192764282226562,
      "learning_rate": 3.1242937853107346e-05,
      "loss": 2.6632,
      "step": 1328
    },
    {
      "epoch": 1.126271186440678,
      "grad_norm": 1.0014623403549194,
      "learning_rate": 3.1228813559322036e-05,
      "loss": 2.7623,
      "step": 1329
    },
    {
      "epoch": 1.1271186440677967,
      "grad_norm": 1.0116573572158813,
      "learning_rate": 3.1214689265536726e-05,
      "loss": 2.894,
      "step": 1330
    },
    {
      "epoch": 1.1279661016949152,
      "grad_norm": 1.1956560611724854,
      "learning_rate": 3.120056497175141e-05,
      "loss": 2.7316,
      "step": 1331
    },
    {
      "epoch": 1.1288135593220339,
      "grad_norm": 1.0047454833984375,
      "learning_rate": 3.1186440677966106e-05,
      "loss": 2.7688,
      "step": 1332
    },
    {
      "epoch": 1.1296610169491526,
      "grad_norm": 0.9235908389091492,
      "learning_rate": 3.1172316384180796e-05,
      "loss": 2.8439,
      "step": 1333
    },
    {
      "epoch": 1.1305084745762712,
      "grad_norm": 1.4579118490219116,
      "learning_rate": 3.1158192090395486e-05,
      "loss": 2.5944,
      "step": 1334
    },
    {
      "epoch": 1.13135593220339,
      "grad_norm": 1.1531609296798706,
      "learning_rate": 3.114406779661017e-05,
      "loss": 2.7467,
      "step": 1335
    },
    {
      "epoch": 1.1322033898305084,
      "grad_norm": 0.9288375973701477,
      "learning_rate": 3.112994350282486e-05,
      "loss": 2.8698,
      "step": 1336
    },
    {
      "epoch": 1.133050847457627,
      "grad_norm": 1.1345102787017822,
      "learning_rate": 3.111581920903955e-05,
      "loss": 2.8097,
      "step": 1337
    },
    {
      "epoch": 1.1338983050847458,
      "grad_norm": 1.0098881721496582,
      "learning_rate": 3.110169491525424e-05,
      "loss": 2.8774,
      "step": 1338
    },
    {
      "epoch": 1.1347457627118644,
      "grad_norm": 1.323324203491211,
      "learning_rate": 3.108757062146893e-05,
      "loss": 2.5576,
      "step": 1339
    },
    {
      "epoch": 1.1355932203389831,
      "grad_norm": 0.9318826198577881,
      "learning_rate": 3.107344632768362e-05,
      "loss": 2.9073,
      "step": 1340
    },
    {
      "epoch": 1.1364406779661016,
      "grad_norm": 1.0271952152252197,
      "learning_rate": 3.105932203389831e-05,
      "loss": 2.8181,
      "step": 1341
    },
    {
      "epoch": 1.1372881355932203,
      "grad_norm": 0.9295307993888855,
      "learning_rate": 3.1045197740113e-05,
      "loss": 2.7162,
      "step": 1342
    },
    {
      "epoch": 1.138135593220339,
      "grad_norm": 1.0659383535385132,
      "learning_rate": 3.103107344632768e-05,
      "loss": 2.694,
      "step": 1343
    },
    {
      "epoch": 1.1389830508474577,
      "grad_norm": 1.2376900911331177,
      "learning_rate": 3.101694915254237e-05,
      "loss": 2.6704,
      "step": 1344
    },
    {
      "epoch": 1.1398305084745763,
      "grad_norm": 1.0314176082611084,
      "learning_rate": 3.100282485875706e-05,
      "loss": 2.9519,
      "step": 1345
    },
    {
      "epoch": 1.1406779661016948,
      "grad_norm": 1.1144027709960938,
      "learning_rate": 3.098870056497175e-05,
      "loss": 2.8029,
      "step": 1346
    },
    {
      "epoch": 1.1415254237288135,
      "grad_norm": 0.9675883650779724,
      "learning_rate": 3.097457627118644e-05,
      "loss": 2.7469,
      "step": 1347
    },
    {
      "epoch": 1.1423728813559322,
      "grad_norm": 1.004947304725647,
      "learning_rate": 3.096045197740113e-05,
      "loss": 2.7452,
      "step": 1348
    },
    {
      "epoch": 1.1432203389830509,
      "grad_norm": 1.0772300958633423,
      "learning_rate": 3.094632768361582e-05,
      "loss": 2.7596,
      "step": 1349
    },
    {
      "epoch": 1.1440677966101696,
      "grad_norm": 1.249557375907898,
      "learning_rate": 3.093220338983051e-05,
      "loss": 2.7725,
      "step": 1350
    },
    {
      "epoch": 1.1449152542372882,
      "grad_norm": 1.1400595903396606,
      "learning_rate": 3.09180790960452e-05,
      "loss": 2.7257,
      "step": 1351
    },
    {
      "epoch": 1.1457627118644067,
      "grad_norm": 1.0549755096435547,
      "learning_rate": 3.0903954802259885e-05,
      "loss": 2.7247,
      "step": 1352
    },
    {
      "epoch": 1.1466101694915254,
      "grad_norm": 1.2909915447235107,
      "learning_rate": 3.0889830508474575e-05,
      "loss": 2.5551,
      "step": 1353
    },
    {
      "epoch": 1.147457627118644,
      "grad_norm": 1.232528567314148,
      "learning_rate": 3.0875706214689265e-05,
      "loss": 2.8768,
      "step": 1354
    },
    {
      "epoch": 1.1483050847457628,
      "grad_norm": 1.3124966621398926,
      "learning_rate": 3.0861581920903955e-05,
      "loss": 2.8048,
      "step": 1355
    },
    {
      "epoch": 1.1491525423728814,
      "grad_norm": 1.0502287149429321,
      "learning_rate": 3.0847457627118645e-05,
      "loss": 2.7559,
      "step": 1356
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.252584457397461,
      "learning_rate": 3.0833333333333335e-05,
      "loss": 2.7825,
      "step": 1357
    },
    {
      "epoch": 1.1508474576271186,
      "grad_norm": 1.24968683719635,
      "learning_rate": 3.0819209039548025e-05,
      "loss": 2.6221,
      "step": 1358
    },
    {
      "epoch": 1.1516949152542373,
      "grad_norm": 1.22371244430542,
      "learning_rate": 3.0805084745762715e-05,
      "loss": 2.7645,
      "step": 1359
    },
    {
      "epoch": 1.152542372881356,
      "grad_norm": 1.1917370557785034,
      "learning_rate": 3.0790960451977405e-05,
      "loss": 2.6992,
      "step": 1360
    },
    {
      "epoch": 1.1533898305084747,
      "grad_norm": 1.5677720308303833,
      "learning_rate": 3.077683615819209e-05,
      "loss": 2.4938,
      "step": 1361
    },
    {
      "epoch": 1.1542372881355933,
      "grad_norm": 0.9529935717582703,
      "learning_rate": 3.076271186440678e-05,
      "loss": 2.8425,
      "step": 1362
    },
    {
      "epoch": 1.1550847457627118,
      "grad_norm": 1.0626323223114014,
      "learning_rate": 3.074858757062147e-05,
      "loss": 2.7787,
      "step": 1363
    },
    {
      "epoch": 1.1559322033898305,
      "grad_norm": 0.9936074018478394,
      "learning_rate": 3.0734463276836165e-05,
      "loss": 2.7484,
      "step": 1364
    },
    {
      "epoch": 1.1567796610169492,
      "grad_norm": 1.3875172138214111,
      "learning_rate": 3.072033898305085e-05,
      "loss": 2.6042,
      "step": 1365
    },
    {
      "epoch": 1.1576271186440679,
      "grad_norm": 1.0043777227401733,
      "learning_rate": 3.070621468926554e-05,
      "loss": 2.7155,
      "step": 1366
    },
    {
      "epoch": 1.1584745762711863,
      "grad_norm": 1.1337268352508545,
      "learning_rate": 3.069209039548023e-05,
      "loss": 2.8497,
      "step": 1367
    },
    {
      "epoch": 1.159322033898305,
      "grad_norm": 1.079054832458496,
      "learning_rate": 3.067796610169492e-05,
      "loss": 2.7388,
      "step": 1368
    },
    {
      "epoch": 1.1601694915254237,
      "grad_norm": 1.1742485761642456,
      "learning_rate": 3.06638418079096e-05,
      "loss": 2.7066,
      "step": 1369
    },
    {
      "epoch": 1.1610169491525424,
      "grad_norm": 1.1586884260177612,
      "learning_rate": 3.064971751412429e-05,
      "loss": 2.7078,
      "step": 1370
    },
    {
      "epoch": 1.161864406779661,
      "grad_norm": 1.1820478439331055,
      "learning_rate": 3.063559322033899e-05,
      "loss": 2.7279,
      "step": 1371
    },
    {
      "epoch": 1.1627118644067798,
      "grad_norm": 1.2828441858291626,
      "learning_rate": 3.062146892655368e-05,
      "loss": 2.6271,
      "step": 1372
    },
    {
      "epoch": 1.1635593220338982,
      "grad_norm": 1.115065336227417,
      "learning_rate": 3.060734463276836e-05,
      "loss": 2.877,
      "step": 1373
    },
    {
      "epoch": 1.164406779661017,
      "grad_norm": 1.353116750717163,
      "learning_rate": 3.059322033898305e-05,
      "loss": 2.6163,
      "step": 1374
    },
    {
      "epoch": 1.1652542372881356,
      "grad_norm": 1.2320899963378906,
      "learning_rate": 3.057909604519774e-05,
      "loss": 2.7729,
      "step": 1375
    },
    {
      "epoch": 1.1661016949152543,
      "grad_norm": 1.1651616096496582,
      "learning_rate": 3.056497175141243e-05,
      "loss": 2.8106,
      "step": 1376
    },
    {
      "epoch": 1.166949152542373,
      "grad_norm": 1.0272785425186157,
      "learning_rate": 3.055084745762712e-05,
      "loss": 2.7614,
      "step": 1377
    },
    {
      "epoch": 1.1677966101694914,
      "grad_norm": 1.002989649772644,
      "learning_rate": 3.0536723163841805e-05,
      "loss": 2.7554,
      "step": 1378
    },
    {
      "epoch": 1.1686440677966101,
      "grad_norm": 1.0328543186187744,
      "learning_rate": 3.05225988700565e-05,
      "loss": 2.7662,
      "step": 1379
    },
    {
      "epoch": 1.1694915254237288,
      "grad_norm": 1.0183919668197632,
      "learning_rate": 3.050847457627119e-05,
      "loss": 2.8341,
      "step": 1380
    },
    {
      "epoch": 1.1703389830508475,
      "grad_norm": 1.1252543926239014,
      "learning_rate": 3.0494350282485878e-05,
      "loss": 2.5955,
      "step": 1381
    },
    {
      "epoch": 1.1711864406779662,
      "grad_norm": 0.9931685924530029,
      "learning_rate": 3.0480225988700568e-05,
      "loss": 2.65,
      "step": 1382
    },
    {
      "epoch": 1.1720338983050849,
      "grad_norm": 1.1650798320770264,
      "learning_rate": 3.0466101694915255e-05,
      "loss": 2.7463,
      "step": 1383
    },
    {
      "epoch": 1.1728813559322033,
      "grad_norm": 0.9627938270568848,
      "learning_rate": 3.0451977401129945e-05,
      "loss": 2.9753,
      "step": 1384
    },
    {
      "epoch": 1.173728813559322,
      "grad_norm": 1.136277437210083,
      "learning_rate": 3.043785310734463e-05,
      "loss": 2.7037,
      "step": 1385
    },
    {
      "epoch": 1.1745762711864407,
      "grad_norm": 1.0859447717666626,
      "learning_rate": 3.042372881355932e-05,
      "loss": 2.9777,
      "step": 1386
    },
    {
      "epoch": 1.1754237288135594,
      "grad_norm": 1.218129277229309,
      "learning_rate": 3.0409604519774015e-05,
      "loss": 2.7412,
      "step": 1387
    },
    {
      "epoch": 1.1762711864406779,
      "grad_norm": 0.9532760977745056,
      "learning_rate": 3.0395480225988705e-05,
      "loss": 2.8729,
      "step": 1388
    },
    {
      "epoch": 1.1771186440677965,
      "grad_norm": 1.3282712697982788,
      "learning_rate": 3.038135593220339e-05,
      "loss": 2.6393,
      "step": 1389
    },
    {
      "epoch": 1.1779661016949152,
      "grad_norm": 1.4244226217269897,
      "learning_rate": 3.036723163841808e-05,
      "loss": 2.7061,
      "step": 1390
    },
    {
      "epoch": 1.178813559322034,
      "grad_norm": 1.3524574041366577,
      "learning_rate": 3.035310734463277e-05,
      "loss": 2.7684,
      "step": 1391
    },
    {
      "epoch": 1.1796610169491526,
      "grad_norm": 1.028135895729065,
      "learning_rate": 3.0338983050847458e-05,
      "loss": 2.9048,
      "step": 1392
    },
    {
      "epoch": 1.1805084745762713,
      "grad_norm": 1.0645534992218018,
      "learning_rate": 3.0324858757062148e-05,
      "loss": 2.704,
      "step": 1393
    },
    {
      "epoch": 1.1813559322033897,
      "grad_norm": 1.1310391426086426,
      "learning_rate": 3.0310734463276834e-05,
      "loss": 2.7615,
      "step": 1394
    },
    {
      "epoch": 1.1822033898305084,
      "grad_norm": 0.9573028087615967,
      "learning_rate": 3.0296610169491528e-05,
      "loss": 2.8263,
      "step": 1395
    },
    {
      "epoch": 1.1830508474576271,
      "grad_norm": 1.1354314088821411,
      "learning_rate": 3.0282485875706218e-05,
      "loss": 2.6444,
      "step": 1396
    },
    {
      "epoch": 1.1838983050847458,
      "grad_norm": 0.9704809188842773,
      "learning_rate": 3.0268361581920908e-05,
      "loss": 2.9929,
      "step": 1397
    },
    {
      "epoch": 1.1847457627118645,
      "grad_norm": 1.1086372137069702,
      "learning_rate": 3.0254237288135594e-05,
      "loss": 2.6978,
      "step": 1398
    },
    {
      "epoch": 1.185593220338983,
      "grad_norm": 1.0527740716934204,
      "learning_rate": 3.0240112994350284e-05,
      "loss": 2.7222,
      "step": 1399
    },
    {
      "epoch": 1.1864406779661016,
      "grad_norm": 0.9511035084724426,
      "learning_rate": 3.022598870056497e-05,
      "loss": 2.8779,
      "step": 1400
    },
    {
      "epoch": 1.1872881355932203,
      "grad_norm": 0.953075110912323,
      "learning_rate": 3.021186440677966e-05,
      "loss": 2.8414,
      "step": 1401
    },
    {
      "epoch": 1.188135593220339,
      "grad_norm": 1.228217363357544,
      "learning_rate": 3.019774011299435e-05,
      "loss": 2.7516,
      "step": 1402
    },
    {
      "epoch": 1.1889830508474577,
      "grad_norm": 1.2182135581970215,
      "learning_rate": 3.0183615819209044e-05,
      "loss": 2.6564,
      "step": 1403
    },
    {
      "epoch": 1.1898305084745764,
      "grad_norm": 0.9400214552879333,
      "learning_rate": 3.016949152542373e-05,
      "loss": 2.7884,
      "step": 1404
    },
    {
      "epoch": 1.1906779661016949,
      "grad_norm": 1.2233799695968628,
      "learning_rate": 3.015536723163842e-05,
      "loss": 2.6467,
      "step": 1405
    },
    {
      "epoch": 1.1915254237288135,
      "grad_norm": 0.9913449883460999,
      "learning_rate": 3.0141242937853108e-05,
      "loss": 2.8703,
      "step": 1406
    },
    {
      "epoch": 1.1923728813559322,
      "grad_norm": 1.144369125366211,
      "learning_rate": 3.0127118644067798e-05,
      "loss": 2.6984,
      "step": 1407
    },
    {
      "epoch": 1.193220338983051,
      "grad_norm": 1.1918572187423706,
      "learning_rate": 3.0112994350282488e-05,
      "loss": 2.7068,
      "step": 1408
    },
    {
      "epoch": 1.1940677966101694,
      "grad_norm": 1.0861234664916992,
      "learning_rate": 3.0098870056497174e-05,
      "loss": 2.7732,
      "step": 1409
    },
    {
      "epoch": 1.194915254237288,
      "grad_norm": 0.9997703433036804,
      "learning_rate": 3.0084745762711864e-05,
      "loss": 2.8182,
      "step": 1410
    },
    {
      "epoch": 1.1957627118644067,
      "grad_norm": 1.2805825471878052,
      "learning_rate": 3.0070621468926558e-05,
      "loss": 2.6062,
      "step": 1411
    },
    {
      "epoch": 1.1966101694915254,
      "grad_norm": 1.077366590499878,
      "learning_rate": 3.0056497175141248e-05,
      "loss": 2.9351,
      "step": 1412
    },
    {
      "epoch": 1.1974576271186441,
      "grad_norm": 1.4923194646835327,
      "learning_rate": 3.0042372881355934e-05,
      "loss": 2.661,
      "step": 1413
    },
    {
      "epoch": 1.1983050847457628,
      "grad_norm": 1.1752442121505737,
      "learning_rate": 3.0028248587570624e-05,
      "loss": 2.6618,
      "step": 1414
    },
    {
      "epoch": 1.1991525423728813,
      "grad_norm": 1.1268277168273926,
      "learning_rate": 3.001412429378531e-05,
      "loss": 2.6549,
      "step": 1415
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1235272884368896,
      "learning_rate": 3e-05,
      "loss": 2.8397,
      "step": 1416
    },
    {
      "epoch": 1.2008474576271186,
      "grad_norm": 1.0348951816558838,
      "learning_rate": 2.9985875706214687e-05,
      "loss": 2.8437,
      "step": 1417
    },
    {
      "epoch": 1.2016949152542373,
      "grad_norm": 1.1336616277694702,
      "learning_rate": 2.9971751412429377e-05,
      "loss": 2.8073,
      "step": 1418
    },
    {
      "epoch": 1.202542372881356,
      "grad_norm": 1.147831916809082,
      "learning_rate": 2.995762711864407e-05,
      "loss": 2.6553,
      "step": 1419
    },
    {
      "epoch": 1.2033898305084745,
      "grad_norm": 1.1440467834472656,
      "learning_rate": 2.994350282485876e-05,
      "loss": 2.6407,
      "step": 1420
    },
    {
      "epoch": 1.2042372881355932,
      "grad_norm": 1.1262454986572266,
      "learning_rate": 2.9929378531073447e-05,
      "loss": 2.7065,
      "step": 1421
    },
    {
      "epoch": 1.2050847457627119,
      "grad_norm": 1.222533106803894,
      "learning_rate": 2.9915254237288137e-05,
      "loss": 2.8257,
      "step": 1422
    },
    {
      "epoch": 1.2059322033898305,
      "grad_norm": 1.0046762228012085,
      "learning_rate": 2.9901129943502827e-05,
      "loss": 2.7413,
      "step": 1423
    },
    {
      "epoch": 1.2067796610169492,
      "grad_norm": 0.9929599761962891,
      "learning_rate": 2.9887005649717514e-05,
      "loss": 2.9109,
      "step": 1424
    },
    {
      "epoch": 1.207627118644068,
      "grad_norm": 1.1053084135055542,
      "learning_rate": 2.9872881355932204e-05,
      "loss": 2.6662,
      "step": 1425
    },
    {
      "epoch": 1.2084745762711864,
      "grad_norm": 1.1630088090896606,
      "learning_rate": 2.985875706214689e-05,
      "loss": 2.7099,
      "step": 1426
    },
    {
      "epoch": 1.209322033898305,
      "grad_norm": 1.0490261316299438,
      "learning_rate": 2.9844632768361587e-05,
      "loss": 2.6698,
      "step": 1427
    },
    {
      "epoch": 1.2101694915254237,
      "grad_norm": 1.0740617513656616,
      "learning_rate": 2.9830508474576274e-05,
      "loss": 2.6712,
      "step": 1428
    },
    {
      "epoch": 1.2110169491525424,
      "grad_norm": 1.0406426191329956,
      "learning_rate": 2.9816384180790964e-05,
      "loss": 2.8546,
      "step": 1429
    },
    {
      "epoch": 1.211864406779661,
      "grad_norm": 1.1298630237579346,
      "learning_rate": 2.980225988700565e-05,
      "loss": 2.7401,
      "step": 1430
    },
    {
      "epoch": 1.2127118644067796,
      "grad_norm": 1.1068373918533325,
      "learning_rate": 2.978813559322034e-05,
      "loss": 2.8314,
      "step": 1431
    },
    {
      "epoch": 1.2135593220338983,
      "grad_norm": 1.4264651536941528,
      "learning_rate": 2.9774011299435027e-05,
      "loss": 2.575,
      "step": 1432
    },
    {
      "epoch": 1.214406779661017,
      "grad_norm": 1.017459750175476,
      "learning_rate": 2.9759887005649717e-05,
      "loss": 2.9322,
      "step": 1433
    },
    {
      "epoch": 1.2152542372881356,
      "grad_norm": 1.1577345132827759,
      "learning_rate": 2.9745762711864407e-05,
      "loss": 2.6906,
      "step": 1434
    },
    {
      "epoch": 1.2161016949152543,
      "grad_norm": 1.0599602460861206,
      "learning_rate": 2.97316384180791e-05,
      "loss": 2.9082,
      "step": 1435
    },
    {
      "epoch": 1.2169491525423728,
      "grad_norm": 1.1662206649780273,
      "learning_rate": 2.9717514124293787e-05,
      "loss": 2.5966,
      "step": 1436
    },
    {
      "epoch": 1.2177966101694915,
      "grad_norm": 1.1501646041870117,
      "learning_rate": 2.9703389830508477e-05,
      "loss": 2.7142,
      "step": 1437
    },
    {
      "epoch": 1.2186440677966102,
      "grad_norm": 1.2464590072631836,
      "learning_rate": 2.9689265536723164e-05,
      "loss": 2.7361,
      "step": 1438
    },
    {
      "epoch": 1.2194915254237289,
      "grad_norm": 1.0664721727371216,
      "learning_rate": 2.9675141242937854e-05,
      "loss": 2.8738,
      "step": 1439
    },
    {
      "epoch": 1.2203389830508475,
      "grad_norm": 1.086744785308838,
      "learning_rate": 2.9661016949152544e-05,
      "loss": 2.7444,
      "step": 1440
    },
    {
      "epoch": 1.221186440677966,
      "grad_norm": 1.0933836698532104,
      "learning_rate": 2.964689265536723e-05,
      "loss": 2.6999,
      "step": 1441
    },
    {
      "epoch": 1.2220338983050847,
      "grad_norm": 0.9780147075653076,
      "learning_rate": 2.963276836158192e-05,
      "loss": 2.8808,
      "step": 1442
    },
    {
      "epoch": 1.2228813559322034,
      "grad_norm": 1.0730159282684326,
      "learning_rate": 2.9618644067796614e-05,
      "loss": 2.9024,
      "step": 1443
    },
    {
      "epoch": 1.223728813559322,
      "grad_norm": 0.9952771663665771,
      "learning_rate": 2.9604519774011304e-05,
      "loss": 2.8256,
      "step": 1444
    },
    {
      "epoch": 1.2245762711864407,
      "grad_norm": 1.0768945217132568,
      "learning_rate": 2.959039548022599e-05,
      "loss": 2.7651,
      "step": 1445
    },
    {
      "epoch": 1.2254237288135594,
      "grad_norm": 0.9231528639793396,
      "learning_rate": 2.957627118644068e-05,
      "loss": 2.9463,
      "step": 1446
    },
    {
      "epoch": 1.226271186440678,
      "grad_norm": 1.3907428979873657,
      "learning_rate": 2.9562146892655367e-05,
      "loss": 2.6094,
      "step": 1447
    },
    {
      "epoch": 1.2271186440677966,
      "grad_norm": 1.030638575553894,
      "learning_rate": 2.9548022598870057e-05,
      "loss": 3.032,
      "step": 1448
    },
    {
      "epoch": 1.2279661016949153,
      "grad_norm": 1.017385721206665,
      "learning_rate": 2.9533898305084743e-05,
      "loss": 2.8845,
      "step": 1449
    },
    {
      "epoch": 1.228813559322034,
      "grad_norm": 1.0024408102035522,
      "learning_rate": 2.951977401129944e-05,
      "loss": 3.0054,
      "step": 1450
    },
    {
      "epoch": 1.2296610169491524,
      "grad_norm": 1.0980383157730103,
      "learning_rate": 2.9505649717514127e-05,
      "loss": 2.8692,
      "step": 1451
    },
    {
      "epoch": 1.230508474576271,
      "grad_norm": 1.1808464527130127,
      "learning_rate": 2.9491525423728817e-05,
      "loss": 2.8447,
      "step": 1452
    },
    {
      "epoch": 1.2313559322033898,
      "grad_norm": 0.9598093628883362,
      "learning_rate": 2.9477401129943503e-05,
      "loss": 2.7909,
      "step": 1453
    },
    {
      "epoch": 1.2322033898305085,
      "grad_norm": 1.0515165328979492,
      "learning_rate": 2.9463276836158193e-05,
      "loss": 2.8163,
      "step": 1454
    },
    {
      "epoch": 1.2330508474576272,
      "grad_norm": 1.0463999509811401,
      "learning_rate": 2.9449152542372883e-05,
      "loss": 2.8476,
      "step": 1455
    },
    {
      "epoch": 1.2338983050847459,
      "grad_norm": 0.9163907170295715,
      "learning_rate": 2.943502824858757e-05,
      "loss": 2.7353,
      "step": 1456
    },
    {
      "epoch": 1.2347457627118643,
      "grad_norm": 0.9019412398338318,
      "learning_rate": 2.942090395480226e-05,
      "loss": 2.7821,
      "step": 1457
    },
    {
      "epoch": 1.235593220338983,
      "grad_norm": 1.192559003829956,
      "learning_rate": 2.9406779661016953e-05,
      "loss": 2.6194,
      "step": 1458
    },
    {
      "epoch": 1.2364406779661017,
      "grad_norm": 1.111675500869751,
      "learning_rate": 2.9392655367231643e-05,
      "loss": 2.7556,
      "step": 1459
    },
    {
      "epoch": 1.2372881355932204,
      "grad_norm": 0.9991422891616821,
      "learning_rate": 2.937853107344633e-05,
      "loss": 2.8622,
      "step": 1460
    },
    {
      "epoch": 1.238135593220339,
      "grad_norm": 0.9809384346008301,
      "learning_rate": 2.936440677966102e-05,
      "loss": 2.742,
      "step": 1461
    },
    {
      "epoch": 1.2389830508474575,
      "grad_norm": 1.0809677839279175,
      "learning_rate": 2.9350282485875707e-05,
      "loss": 2.7372,
      "step": 1462
    },
    {
      "epoch": 1.2398305084745762,
      "grad_norm": 1.1370301246643066,
      "learning_rate": 2.9336158192090397e-05,
      "loss": 2.8335,
      "step": 1463
    },
    {
      "epoch": 1.240677966101695,
      "grad_norm": 1.2629517316818237,
      "learning_rate": 2.9322033898305083e-05,
      "loss": 2.6227,
      "step": 1464
    },
    {
      "epoch": 1.2415254237288136,
      "grad_norm": 1.0647246837615967,
      "learning_rate": 2.9307909604519773e-05,
      "loss": 2.7986,
      "step": 1465
    },
    {
      "epoch": 1.2423728813559323,
      "grad_norm": 1.108402132987976,
      "learning_rate": 2.9293785310734467e-05,
      "loss": 2.7282,
      "step": 1466
    },
    {
      "epoch": 1.243220338983051,
      "grad_norm": 1.4929739236831665,
      "learning_rate": 2.9279661016949157e-05,
      "loss": 2.5842,
      "step": 1467
    },
    {
      "epoch": 1.2440677966101694,
      "grad_norm": 0.9982267618179321,
      "learning_rate": 2.9265536723163843e-05,
      "loss": 2.7141,
      "step": 1468
    },
    {
      "epoch": 1.244915254237288,
      "grad_norm": 1.1600935459136963,
      "learning_rate": 2.9251412429378533e-05,
      "loss": 2.6322,
      "step": 1469
    },
    {
      "epoch": 1.2457627118644068,
      "grad_norm": 1.0451264381408691,
      "learning_rate": 2.9237288135593223e-05,
      "loss": 2.823,
      "step": 1470
    },
    {
      "epoch": 1.2466101694915255,
      "grad_norm": 1.1497570276260376,
      "learning_rate": 2.922316384180791e-05,
      "loss": 2.7282,
      "step": 1471
    },
    {
      "epoch": 1.2474576271186442,
      "grad_norm": 1.070095181465149,
      "learning_rate": 2.92090395480226e-05,
      "loss": 2.8049,
      "step": 1472
    },
    {
      "epoch": 1.2483050847457626,
      "grad_norm": 0.9942064881324768,
      "learning_rate": 2.9194915254237286e-05,
      "loss": 2.8992,
      "step": 1473
    },
    {
      "epoch": 1.2491525423728813,
      "grad_norm": 1.1197259426116943,
      "learning_rate": 2.918079096045198e-05,
      "loss": 2.7402,
      "step": 1474
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0887670516967773,
      "learning_rate": 2.916666666666667e-05,
      "loss": 2.7317,
      "step": 1475
    },
    {
      "epoch": 1.2508474576271187,
      "grad_norm": 1.061383605003357,
      "learning_rate": 2.915254237288136e-05,
      "loss": 2.7887,
      "step": 1476
    },
    {
      "epoch": 1.2516949152542374,
      "grad_norm": 1.5140694379806519,
      "learning_rate": 2.9138418079096046e-05,
      "loss": 2.5884,
      "step": 1477
    },
    {
      "epoch": 1.252542372881356,
      "grad_norm": 1.155662178993225,
      "learning_rate": 2.9124293785310736e-05,
      "loss": 2.6443,
      "step": 1478
    },
    {
      "epoch": 1.2533898305084745,
      "grad_norm": 1.0662916898727417,
      "learning_rate": 2.9110169491525423e-05,
      "loss": 2.8724,
      "step": 1479
    },
    {
      "epoch": 1.2542372881355932,
      "grad_norm": 1.0730879306793213,
      "learning_rate": 2.9096045197740113e-05,
      "loss": 2.7966,
      "step": 1480
    },
    {
      "epoch": 1.255084745762712,
      "grad_norm": 1.324304223060608,
      "learning_rate": 2.9081920903954803e-05,
      "loss": 2.5592,
      "step": 1481
    },
    {
      "epoch": 1.2559322033898306,
      "grad_norm": 1.211549162864685,
      "learning_rate": 2.9067796610169496e-05,
      "loss": 2.6806,
      "step": 1482
    },
    {
      "epoch": 1.256779661016949,
      "grad_norm": 1.0320687294006348,
      "learning_rate": 2.9053672316384183e-05,
      "loss": 2.7471,
      "step": 1483
    },
    {
      "epoch": 1.2576271186440677,
      "grad_norm": 1.028415322303772,
      "learning_rate": 2.9039548022598873e-05,
      "loss": 2.7035,
      "step": 1484
    },
    {
      "epoch": 1.2584745762711864,
      "grad_norm": 1.4593757390975952,
      "learning_rate": 2.902542372881356e-05,
      "loss": 2.7799,
      "step": 1485
    },
    {
      "epoch": 1.259322033898305,
      "grad_norm": 0.9076970219612122,
      "learning_rate": 2.901129943502825e-05,
      "loss": 2.8766,
      "step": 1486
    },
    {
      "epoch": 1.2601694915254238,
      "grad_norm": 1.0136754512786865,
      "learning_rate": 2.899717514124294e-05,
      "loss": 2.9435,
      "step": 1487
    },
    {
      "epoch": 1.2610169491525425,
      "grad_norm": 1.1704306602478027,
      "learning_rate": 2.8983050847457626e-05,
      "loss": 2.7245,
      "step": 1488
    },
    {
      "epoch": 1.261864406779661,
      "grad_norm": 1.155551552772522,
      "learning_rate": 2.8968926553672316e-05,
      "loss": 2.5759,
      "step": 1489
    },
    {
      "epoch": 1.2627118644067796,
      "grad_norm": 1.2033894062042236,
      "learning_rate": 2.895480225988701e-05,
      "loss": 2.6947,
      "step": 1490
    },
    {
      "epoch": 1.2635593220338983,
      "grad_norm": 1.052081823348999,
      "learning_rate": 2.89406779661017e-05,
      "loss": 2.6866,
      "step": 1491
    },
    {
      "epoch": 1.264406779661017,
      "grad_norm": 1.2990190982818604,
      "learning_rate": 2.8926553672316386e-05,
      "loss": 2.6354,
      "step": 1492
    },
    {
      "epoch": 1.2652542372881355,
      "grad_norm": 0.9295713901519775,
      "learning_rate": 2.8912429378531076e-05,
      "loss": 2.7963,
      "step": 1493
    },
    {
      "epoch": 1.2661016949152541,
      "grad_norm": 1.1894383430480957,
      "learning_rate": 2.8898305084745763e-05,
      "loss": 2.5809,
      "step": 1494
    },
    {
      "epoch": 1.2669491525423728,
      "grad_norm": 1.1730421781539917,
      "learning_rate": 2.8884180790960453e-05,
      "loss": 2.5862,
      "step": 1495
    },
    {
      "epoch": 1.2677966101694915,
      "grad_norm": 1.0560240745544434,
      "learning_rate": 2.887005649717514e-05,
      "loss": 2.8433,
      "step": 1496
    },
    {
      "epoch": 1.2686440677966102,
      "grad_norm": 1.52442467212677,
      "learning_rate": 2.885593220338983e-05,
      "loss": 2.4808,
      "step": 1497
    },
    {
      "epoch": 1.269491525423729,
      "grad_norm": 1.186542272567749,
      "learning_rate": 2.8841807909604523e-05,
      "loss": 2.7571,
      "step": 1498
    },
    {
      "epoch": 1.2703389830508476,
      "grad_norm": 1.104839563369751,
      "learning_rate": 2.8827683615819213e-05,
      "loss": 2.879,
      "step": 1499
    },
    {
      "epoch": 1.271186440677966,
      "grad_norm": 1.1865556240081787,
      "learning_rate": 2.88135593220339e-05,
      "loss": 2.806,
      "step": 1500
    },
    {
      "epoch": 1.2720338983050847,
      "grad_norm": 1.2355942726135254,
      "learning_rate": 2.879943502824859e-05,
      "loss": 2.7078,
      "step": 1501
    },
    {
      "epoch": 1.2728813559322034,
      "grad_norm": 1.0432177782058716,
      "learning_rate": 2.878531073446328e-05,
      "loss": 2.7354,
      "step": 1502
    },
    {
      "epoch": 1.273728813559322,
      "grad_norm": 0.9389800429344177,
      "learning_rate": 2.8771186440677966e-05,
      "loss": 2.9527,
      "step": 1503
    },
    {
      "epoch": 1.2745762711864406,
      "grad_norm": 1.133212685585022,
      "learning_rate": 2.8757062146892656e-05,
      "loss": 2.935,
      "step": 1504
    },
    {
      "epoch": 1.2754237288135593,
      "grad_norm": 1.0598336458206177,
      "learning_rate": 2.8742937853107342e-05,
      "loss": 2.7303,
      "step": 1505
    },
    {
      "epoch": 1.276271186440678,
      "grad_norm": 1.124864935874939,
      "learning_rate": 2.8728813559322036e-05,
      "loss": 2.849,
      "step": 1506
    },
    {
      "epoch": 1.2771186440677966,
      "grad_norm": 0.9836654663085938,
      "learning_rate": 2.8714689265536726e-05,
      "loss": 2.9168,
      "step": 1507
    },
    {
      "epoch": 1.2779661016949153,
      "grad_norm": 1.2238267660140991,
      "learning_rate": 2.8700564971751416e-05,
      "loss": 2.7461,
      "step": 1508
    },
    {
      "epoch": 1.278813559322034,
      "grad_norm": 1.1149492263793945,
      "learning_rate": 2.8686440677966102e-05,
      "loss": 2.7516,
      "step": 1509
    },
    {
      "epoch": 1.2796610169491525,
      "grad_norm": 1.0936237573623657,
      "learning_rate": 2.8672316384180792e-05,
      "loss": 2.7713,
      "step": 1510
    },
    {
      "epoch": 1.2805084745762711,
      "grad_norm": 1.239094614982605,
      "learning_rate": 2.865819209039548e-05,
      "loss": 2.6343,
      "step": 1511
    },
    {
      "epoch": 1.2813559322033898,
      "grad_norm": 1.0122066736221313,
      "learning_rate": 2.864406779661017e-05,
      "loss": 2.7883,
      "step": 1512
    },
    {
      "epoch": 1.2822033898305085,
      "grad_norm": 1.1487418413162231,
      "learning_rate": 2.862994350282486e-05,
      "loss": 2.5364,
      "step": 1513
    },
    {
      "epoch": 1.283050847457627,
      "grad_norm": 1.1750707626342773,
      "learning_rate": 2.8615819209039552e-05,
      "loss": 2.8201,
      "step": 1514
    },
    {
      "epoch": 1.2838983050847457,
      "grad_norm": 0.9801908731460571,
      "learning_rate": 2.860169491525424e-05,
      "loss": 2.6655,
      "step": 1515
    },
    {
      "epoch": 1.2847457627118644,
      "grad_norm": 1.0138574838638306,
      "learning_rate": 2.858757062146893e-05,
      "loss": 2.8752,
      "step": 1516
    },
    {
      "epoch": 1.285593220338983,
      "grad_norm": 1.3303834199905396,
      "learning_rate": 2.8573446327683616e-05,
      "loss": 2.8182,
      "step": 1517
    },
    {
      "epoch": 1.2864406779661017,
      "grad_norm": 1.0959980487823486,
      "learning_rate": 2.8559322033898306e-05,
      "loss": 2.7536,
      "step": 1518
    },
    {
      "epoch": 1.2872881355932204,
      "grad_norm": 1.2038679122924805,
      "learning_rate": 2.8545197740112996e-05,
      "loss": 2.7121,
      "step": 1519
    },
    {
      "epoch": 1.288135593220339,
      "grad_norm": 0.9292485117912292,
      "learning_rate": 2.8531073446327682e-05,
      "loss": 2.9412,
      "step": 1520
    },
    {
      "epoch": 1.2889830508474576,
      "grad_norm": 1.2763807773590088,
      "learning_rate": 2.8516949152542372e-05,
      "loss": 2.8264,
      "step": 1521
    },
    {
      "epoch": 1.2898305084745763,
      "grad_norm": 1.2033941745758057,
      "learning_rate": 2.8502824858757066e-05,
      "loss": 2.5891,
      "step": 1522
    },
    {
      "epoch": 1.290677966101695,
      "grad_norm": 0.9282636046409607,
      "learning_rate": 2.8488700564971756e-05,
      "loss": 2.8097,
      "step": 1523
    },
    {
      "epoch": 1.2915254237288136,
      "grad_norm": 1.1813933849334717,
      "learning_rate": 2.8474576271186442e-05,
      "loss": 2.7252,
      "step": 1524
    },
    {
      "epoch": 1.292372881355932,
      "grad_norm": 1.190480351448059,
      "learning_rate": 2.8460451977401132e-05,
      "loss": 2.5932,
      "step": 1525
    },
    {
      "epoch": 1.2932203389830508,
      "grad_norm": 1.1157951354980469,
      "learning_rate": 2.844632768361582e-05,
      "loss": 2.679,
      "step": 1526
    },
    {
      "epoch": 1.2940677966101695,
      "grad_norm": 1.2422996759414673,
      "learning_rate": 2.843220338983051e-05,
      "loss": 2.7436,
      "step": 1527
    },
    {
      "epoch": 1.2949152542372881,
      "grad_norm": 1.383557677268982,
      "learning_rate": 2.8418079096045195e-05,
      "loss": 2.4242,
      "step": 1528
    },
    {
      "epoch": 1.2957627118644068,
      "grad_norm": 1.2432738542556763,
      "learning_rate": 2.8403954802259892e-05,
      "loss": 2.6603,
      "step": 1529
    },
    {
      "epoch": 1.2966101694915255,
      "grad_norm": 1.155629277229309,
      "learning_rate": 2.838983050847458e-05,
      "loss": 2.8567,
      "step": 1530
    },
    {
      "epoch": 1.297457627118644,
      "grad_norm": 1.200217604637146,
      "learning_rate": 2.837570621468927e-05,
      "loss": 2.6833,
      "step": 1531
    },
    {
      "epoch": 1.2983050847457627,
      "grad_norm": 1.031434178352356,
      "learning_rate": 2.8361581920903955e-05,
      "loss": 2.7,
      "step": 1532
    },
    {
      "epoch": 1.2991525423728814,
      "grad_norm": 1.168958306312561,
      "learning_rate": 2.8347457627118645e-05,
      "loss": 2.6607,
      "step": 1533
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9971181750297546,
      "learning_rate": 2.8333333333333335e-05,
      "loss": 2.7787,
      "step": 1534
    },
    {
      "epoch": 1.3008474576271185,
      "grad_norm": 1.00819993019104,
      "learning_rate": 2.8319209039548022e-05,
      "loss": 2.8582,
      "step": 1535
    },
    {
      "epoch": 1.3016949152542372,
      "grad_norm": 1.047484040260315,
      "learning_rate": 2.8305084745762712e-05,
      "loss": 2.6924,
      "step": 1536
    },
    {
      "epoch": 1.3025423728813559,
      "grad_norm": 1.1539669036865234,
      "learning_rate": 2.8290960451977405e-05,
      "loss": 2.8083,
      "step": 1537
    },
    {
      "epoch": 1.3033898305084746,
      "grad_norm": 1.2319265604019165,
      "learning_rate": 2.8276836158192095e-05,
      "loss": 2.8651,
      "step": 1538
    },
    {
      "epoch": 1.3042372881355933,
      "grad_norm": 0.9315810203552246,
      "learning_rate": 2.8262711864406782e-05,
      "loss": 2.8533,
      "step": 1539
    },
    {
      "epoch": 1.305084745762712,
      "grad_norm": 1.0935174226760864,
      "learning_rate": 2.8248587570621472e-05,
      "loss": 2.6608,
      "step": 1540
    },
    {
      "epoch": 1.3059322033898306,
      "grad_norm": 1.512353539466858,
      "learning_rate": 2.823446327683616e-05,
      "loss": 2.4215,
      "step": 1541
    },
    {
      "epoch": 1.306779661016949,
      "grad_norm": 1.205745816230774,
      "learning_rate": 2.822033898305085e-05,
      "loss": 2.7931,
      "step": 1542
    },
    {
      "epoch": 1.3076271186440678,
      "grad_norm": 1.1302180290222168,
      "learning_rate": 2.8206214689265535e-05,
      "loss": 2.7272,
      "step": 1543
    },
    {
      "epoch": 1.3084745762711865,
      "grad_norm": 1.0707296133041382,
      "learning_rate": 2.8192090395480225e-05,
      "loss": 2.8232,
      "step": 1544
    },
    {
      "epoch": 1.3093220338983051,
      "grad_norm": 1.346828818321228,
      "learning_rate": 2.817796610169492e-05,
      "loss": 2.5666,
      "step": 1545
    },
    {
      "epoch": 1.3101694915254236,
      "grad_norm": 1.23905611038208,
      "learning_rate": 2.816384180790961e-05,
      "loss": 2.6975,
      "step": 1546
    },
    {
      "epoch": 1.3110169491525423,
      "grad_norm": 0.95879065990448,
      "learning_rate": 2.8149717514124295e-05,
      "loss": 2.9588,
      "step": 1547
    },
    {
      "epoch": 1.311864406779661,
      "grad_norm": 1.173724889755249,
      "learning_rate": 2.8135593220338985e-05,
      "loss": 2.7457,
      "step": 1548
    },
    {
      "epoch": 1.3127118644067797,
      "grad_norm": 1.2177459001541138,
      "learning_rate": 2.8121468926553675e-05,
      "loss": 2.5301,
      "step": 1549
    },
    {
      "epoch": 1.3135593220338984,
      "grad_norm": 1.3209285736083984,
      "learning_rate": 2.8107344632768362e-05,
      "loss": 2.6206,
      "step": 1550
    },
    {
      "epoch": 1.314406779661017,
      "grad_norm": 1.108282446861267,
      "learning_rate": 2.8093220338983052e-05,
      "loss": 2.6941,
      "step": 1551
    },
    {
      "epoch": 1.3152542372881357,
      "grad_norm": 1.2630671262741089,
      "learning_rate": 2.807909604519774e-05,
      "loss": 2.6299,
      "step": 1552
    },
    {
      "epoch": 1.3161016949152542,
      "grad_norm": 1.1710407733917236,
      "learning_rate": 2.8064971751412432e-05,
      "loss": 2.6817,
      "step": 1553
    },
    {
      "epoch": 1.3169491525423729,
      "grad_norm": 1.0206531286239624,
      "learning_rate": 2.8050847457627122e-05,
      "loss": 2.7454,
      "step": 1554
    },
    {
      "epoch": 1.3177966101694916,
      "grad_norm": 1.2598350048065186,
      "learning_rate": 2.803672316384181e-05,
      "loss": 2.7229,
      "step": 1555
    },
    {
      "epoch": 1.31864406779661,
      "grad_norm": 1.4363096952438354,
      "learning_rate": 2.8022598870056498e-05,
      "loss": 2.5569,
      "step": 1556
    },
    {
      "epoch": 1.3194915254237287,
      "grad_norm": 1.116281509399414,
      "learning_rate": 2.8008474576271188e-05,
      "loss": 2.6604,
      "step": 1557
    },
    {
      "epoch": 1.3203389830508474,
      "grad_norm": 1.04258131980896,
      "learning_rate": 2.7994350282485875e-05,
      "loss": 2.7203,
      "step": 1558
    },
    {
      "epoch": 1.321186440677966,
      "grad_norm": 1.1648573875427246,
      "learning_rate": 2.7980225988700565e-05,
      "loss": 2.5303,
      "step": 1559
    },
    {
      "epoch": 1.3220338983050848,
      "grad_norm": 1.2743134498596191,
      "learning_rate": 2.7966101694915255e-05,
      "loss": 2.5215,
      "step": 1560
    },
    {
      "epoch": 1.3228813559322035,
      "grad_norm": 0.9022263884544373,
      "learning_rate": 2.7951977401129948e-05,
      "loss": 2.8265,
      "step": 1561
    },
    {
      "epoch": 1.3237288135593221,
      "grad_norm": 0.9590513110160828,
      "learning_rate": 2.7937853107344635e-05,
      "loss": 2.7795,
      "step": 1562
    },
    {
      "epoch": 1.3245762711864406,
      "grad_norm": 1.5466468334197998,
      "learning_rate": 2.7923728813559325e-05,
      "loss": 2.6883,
      "step": 1563
    },
    {
      "epoch": 1.3254237288135593,
      "grad_norm": 0.9928246140480042,
      "learning_rate": 2.790960451977401e-05,
      "loss": 2.9273,
      "step": 1564
    },
    {
      "epoch": 1.326271186440678,
      "grad_norm": 1.2357574701309204,
      "learning_rate": 2.78954802259887e-05,
      "loss": 2.893,
      "step": 1565
    },
    {
      "epoch": 1.3271186440677967,
      "grad_norm": 0.9751428365707397,
      "learning_rate": 2.788135593220339e-05,
      "loss": 2.885,
      "step": 1566
    },
    {
      "epoch": 1.3279661016949151,
      "grad_norm": 1.1124868392944336,
      "learning_rate": 2.7867231638418078e-05,
      "loss": 2.8267,
      "step": 1567
    },
    {
      "epoch": 1.3288135593220338,
      "grad_norm": 1.1400896310806274,
      "learning_rate": 2.7853107344632768e-05,
      "loss": 2.7506,
      "step": 1568
    },
    {
      "epoch": 1.3296610169491525,
      "grad_norm": 1.4547759294509888,
      "learning_rate": 2.783898305084746e-05,
      "loss": 2.5548,
      "step": 1569
    },
    {
      "epoch": 1.3305084745762712,
      "grad_norm": 1.1902233362197876,
      "learning_rate": 2.782485875706215e-05,
      "loss": 2.5925,
      "step": 1570
    },
    {
      "epoch": 1.3313559322033899,
      "grad_norm": 1.2671072483062744,
      "learning_rate": 2.7810734463276838e-05,
      "loss": 2.6583,
      "step": 1571
    },
    {
      "epoch": 1.3322033898305086,
      "grad_norm": 1.1155524253845215,
      "learning_rate": 2.7796610169491528e-05,
      "loss": 2.7445,
      "step": 1572
    },
    {
      "epoch": 1.3330508474576273,
      "grad_norm": 1.1112772226333618,
      "learning_rate": 2.7782485875706215e-05,
      "loss": 2.7136,
      "step": 1573
    },
    {
      "epoch": 1.3338983050847457,
      "grad_norm": 1.033073902130127,
      "learning_rate": 2.7768361581920905e-05,
      "loss": 2.7376,
      "step": 1574
    },
    {
      "epoch": 1.3347457627118644,
      "grad_norm": 1.1341297626495361,
      "learning_rate": 2.775423728813559e-05,
      "loss": 2.6522,
      "step": 1575
    },
    {
      "epoch": 1.335593220338983,
      "grad_norm": 1.0817641019821167,
      "learning_rate": 2.774011299435028e-05,
      "loss": 2.7918,
      "step": 1576
    },
    {
      "epoch": 1.3364406779661018,
      "grad_norm": 1.2469141483306885,
      "learning_rate": 2.7725988700564975e-05,
      "loss": 2.7048,
      "step": 1577
    },
    {
      "epoch": 1.3372881355932202,
      "grad_norm": 1.068687915802002,
      "learning_rate": 2.7711864406779665e-05,
      "loss": 2.54,
      "step": 1578
    },
    {
      "epoch": 1.338135593220339,
      "grad_norm": 1.1469836235046387,
      "learning_rate": 2.769774011299435e-05,
      "loss": 2.5623,
      "step": 1579
    },
    {
      "epoch": 1.3389830508474576,
      "grad_norm": 0.9954297542572021,
      "learning_rate": 2.768361581920904e-05,
      "loss": 2.8458,
      "step": 1580
    },
    {
      "epoch": 1.3398305084745763,
      "grad_norm": 1.0572340488433838,
      "learning_rate": 2.766949152542373e-05,
      "loss": 2.7111,
      "step": 1581
    },
    {
      "epoch": 1.340677966101695,
      "grad_norm": 1.21023690700531,
      "learning_rate": 2.7655367231638418e-05,
      "loss": 2.62,
      "step": 1582
    },
    {
      "epoch": 1.3415254237288137,
      "grad_norm": 1.0304431915283203,
      "learning_rate": 2.7641242937853108e-05,
      "loss": 2.8045,
      "step": 1583
    },
    {
      "epoch": 1.3423728813559321,
      "grad_norm": 1.6247771978378296,
      "learning_rate": 2.7627118644067794e-05,
      "loss": 2.6552,
      "step": 1584
    },
    {
      "epoch": 1.3432203389830508,
      "grad_norm": 1.2078450918197632,
      "learning_rate": 2.7612994350282488e-05,
      "loss": 2.8752,
      "step": 1585
    },
    {
      "epoch": 1.3440677966101695,
      "grad_norm": 1.1666221618652344,
      "learning_rate": 2.7598870056497178e-05,
      "loss": 2.7792,
      "step": 1586
    },
    {
      "epoch": 1.3449152542372882,
      "grad_norm": 1.1695728302001953,
      "learning_rate": 2.7584745762711868e-05,
      "loss": 2.8091,
      "step": 1587
    },
    {
      "epoch": 1.3457627118644067,
      "grad_norm": 1.0968023538589478,
      "learning_rate": 2.7570621468926554e-05,
      "loss": 2.8666,
      "step": 1588
    },
    {
      "epoch": 1.3466101694915253,
      "grad_norm": 1.167147159576416,
      "learning_rate": 2.7556497175141244e-05,
      "loss": 2.7643,
      "step": 1589
    },
    {
      "epoch": 1.347457627118644,
      "grad_norm": 1.1259609460830688,
      "learning_rate": 2.754237288135593e-05,
      "loss": 2.8463,
      "step": 1590
    },
    {
      "epoch": 1.3483050847457627,
      "grad_norm": 1.1308674812316895,
      "learning_rate": 2.752824858757062e-05,
      "loss": 2.788,
      "step": 1591
    },
    {
      "epoch": 1.3491525423728814,
      "grad_norm": 1.2447667121887207,
      "learning_rate": 2.751412429378531e-05,
      "loss": 2.6729,
      "step": 1592
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9429620504379272,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 2.8734,
      "step": 1593
    },
    {
      "epoch": 1.3508474576271188,
      "grad_norm": 1.1183481216430664,
      "learning_rate": 2.748587570621469e-05,
      "loss": 2.7005,
      "step": 1594
    },
    {
      "epoch": 1.3516949152542372,
      "grad_norm": 1.1970775127410889,
      "learning_rate": 2.747175141242938e-05,
      "loss": 2.7509,
      "step": 1595
    },
    {
      "epoch": 1.352542372881356,
      "grad_norm": 1.1498631238937378,
      "learning_rate": 2.7457627118644068e-05,
      "loss": 2.7412,
      "step": 1596
    },
    {
      "epoch": 1.3533898305084746,
      "grad_norm": 1.047776222229004,
      "learning_rate": 2.7443502824858758e-05,
      "loss": 2.8804,
      "step": 1597
    },
    {
      "epoch": 1.3542372881355933,
      "grad_norm": 1.227165699005127,
      "learning_rate": 2.7429378531073448e-05,
      "loss": 2.7615,
      "step": 1598
    },
    {
      "epoch": 1.3550847457627118,
      "grad_norm": 1.1898428201675415,
      "learning_rate": 2.7415254237288134e-05,
      "loss": 2.8409,
      "step": 1599
    },
    {
      "epoch": 1.3559322033898304,
      "grad_norm": 1.0390897989273071,
      "learning_rate": 2.7401129943502824e-05,
      "loss": 2.687,
      "step": 1600
    },
    {
      "epoch": 1.3567796610169491,
      "grad_norm": 1.0258177518844604,
      "learning_rate": 2.7387005649717518e-05,
      "loss": 2.7485,
      "step": 1601
    },
    {
      "epoch": 1.3576271186440678,
      "grad_norm": 1.0979821681976318,
      "learning_rate": 2.7372881355932208e-05,
      "loss": 2.901,
      "step": 1602
    },
    {
      "epoch": 1.3584745762711865,
      "grad_norm": 1.1121689081192017,
      "learning_rate": 2.7358757062146894e-05,
      "loss": 2.8018,
      "step": 1603
    },
    {
      "epoch": 1.3593220338983052,
      "grad_norm": 1.0678071975708008,
      "learning_rate": 2.7344632768361584e-05,
      "loss": 2.8211,
      "step": 1604
    },
    {
      "epoch": 1.3601694915254237,
      "grad_norm": 0.9185408353805542,
      "learning_rate": 2.733050847457627e-05,
      "loss": 2.9341,
      "step": 1605
    },
    {
      "epoch": 1.3610169491525423,
      "grad_norm": 1.3406561613082886,
      "learning_rate": 2.731638418079096e-05,
      "loss": 2.6606,
      "step": 1606
    },
    {
      "epoch": 1.361864406779661,
      "grad_norm": 1.0097061395645142,
      "learning_rate": 2.7302259887005647e-05,
      "loss": 2.8991,
      "step": 1607
    },
    {
      "epoch": 1.3627118644067797,
      "grad_norm": 1.1009202003479004,
      "learning_rate": 2.7288135593220337e-05,
      "loss": 2.6984,
      "step": 1608
    },
    {
      "epoch": 1.3635593220338982,
      "grad_norm": 1.0728604793548584,
      "learning_rate": 2.727401129943503e-05,
      "loss": 2.8817,
      "step": 1609
    },
    {
      "epoch": 1.3644067796610169,
      "grad_norm": 1.459185004234314,
      "learning_rate": 2.725988700564972e-05,
      "loss": 2.5627,
      "step": 1610
    },
    {
      "epoch": 1.3652542372881356,
      "grad_norm": 1.506783366203308,
      "learning_rate": 2.7245762711864407e-05,
      "loss": 2.4316,
      "step": 1611
    },
    {
      "epoch": 1.3661016949152542,
      "grad_norm": 1.1802157163619995,
      "learning_rate": 2.7231638418079097e-05,
      "loss": 2.6748,
      "step": 1612
    },
    {
      "epoch": 1.366949152542373,
      "grad_norm": 0.9531373381614685,
      "learning_rate": 2.7217514124293787e-05,
      "loss": 2.8042,
      "step": 1613
    },
    {
      "epoch": 1.3677966101694916,
      "grad_norm": 1.0940016508102417,
      "learning_rate": 2.7203389830508474e-05,
      "loss": 2.7417,
      "step": 1614
    },
    {
      "epoch": 1.3686440677966103,
      "grad_norm": 1.1369091272354126,
      "learning_rate": 2.7189265536723164e-05,
      "loss": 2.742,
      "step": 1615
    },
    {
      "epoch": 1.3694915254237288,
      "grad_norm": 1.4163222312927246,
      "learning_rate": 2.7175141242937857e-05,
      "loss": 2.6235,
      "step": 1616
    },
    {
      "epoch": 1.3703389830508474,
      "grad_norm": 1.1588315963745117,
      "learning_rate": 2.7161016949152547e-05,
      "loss": 2.6688,
      "step": 1617
    },
    {
      "epoch": 1.3711864406779661,
      "grad_norm": 1.1556146144866943,
      "learning_rate": 2.7146892655367234e-05,
      "loss": 2.594,
      "step": 1618
    },
    {
      "epoch": 1.3720338983050848,
      "grad_norm": 1.055390477180481,
      "learning_rate": 2.7132768361581924e-05,
      "loss": 2.7697,
      "step": 1619
    },
    {
      "epoch": 1.3728813559322033,
      "grad_norm": 0.9598971009254456,
      "learning_rate": 2.711864406779661e-05,
      "loss": 2.7616,
      "step": 1620
    },
    {
      "epoch": 1.373728813559322,
      "grad_norm": 1.347302794456482,
      "learning_rate": 2.71045197740113e-05,
      "loss": 2.5287,
      "step": 1621
    },
    {
      "epoch": 1.3745762711864407,
      "grad_norm": 1.150273084640503,
      "learning_rate": 2.7090395480225987e-05,
      "loss": 2.6823,
      "step": 1622
    },
    {
      "epoch": 1.3754237288135593,
      "grad_norm": 1.1114012002944946,
      "learning_rate": 2.7076271186440677e-05,
      "loss": 2.708,
      "step": 1623
    },
    {
      "epoch": 1.376271186440678,
      "grad_norm": 1.2295079231262207,
      "learning_rate": 2.706214689265537e-05,
      "loss": 2.7309,
      "step": 1624
    },
    {
      "epoch": 1.3771186440677967,
      "grad_norm": 1.1405549049377441,
      "learning_rate": 2.704802259887006e-05,
      "loss": 2.7787,
      "step": 1625
    },
    {
      "epoch": 1.3779661016949152,
      "grad_norm": 0.976355254650116,
      "learning_rate": 2.7033898305084747e-05,
      "loss": 2.7419,
      "step": 1626
    },
    {
      "epoch": 1.3788135593220339,
      "grad_norm": 1.122742772102356,
      "learning_rate": 2.7019774011299437e-05,
      "loss": 2.7403,
      "step": 1627
    },
    {
      "epoch": 1.3796610169491526,
      "grad_norm": 1.099894404411316,
      "learning_rate": 2.7005649717514127e-05,
      "loss": 2.7016,
      "step": 1628
    },
    {
      "epoch": 1.3805084745762712,
      "grad_norm": 1.1797363758087158,
      "learning_rate": 2.6991525423728814e-05,
      "loss": 2.6439,
      "step": 1629
    },
    {
      "epoch": 1.3813559322033897,
      "grad_norm": 1.2248269319534302,
      "learning_rate": 2.6977401129943504e-05,
      "loss": 2.8427,
      "step": 1630
    },
    {
      "epoch": 1.3822033898305084,
      "grad_norm": 1.2509466409683228,
      "learning_rate": 2.696327683615819e-05,
      "loss": 2.6564,
      "step": 1631
    },
    {
      "epoch": 1.383050847457627,
      "grad_norm": 1.3079769611358643,
      "learning_rate": 2.6949152542372884e-05,
      "loss": 2.6483,
      "step": 1632
    },
    {
      "epoch": 1.3838983050847458,
      "grad_norm": 1.0489200353622437,
      "learning_rate": 2.6935028248587574e-05,
      "loss": 2.7253,
      "step": 1633
    },
    {
      "epoch": 1.3847457627118644,
      "grad_norm": 1.2310895919799805,
      "learning_rate": 2.6920903954802264e-05,
      "loss": 2.793,
      "step": 1634
    },
    {
      "epoch": 1.3855932203389831,
      "grad_norm": 1.0037175416946411,
      "learning_rate": 2.690677966101695e-05,
      "loss": 3.0587,
      "step": 1635
    },
    {
      "epoch": 1.3864406779661018,
      "grad_norm": 1.1418050527572632,
      "learning_rate": 2.689265536723164e-05,
      "loss": 2.7489,
      "step": 1636
    },
    {
      "epoch": 1.3872881355932203,
      "grad_norm": 0.8957945108413696,
      "learning_rate": 2.6878531073446327e-05,
      "loss": 3.0793,
      "step": 1637
    },
    {
      "epoch": 1.388135593220339,
      "grad_norm": 1.1641825437545776,
      "learning_rate": 2.6864406779661017e-05,
      "loss": 2.8238,
      "step": 1638
    },
    {
      "epoch": 1.3889830508474577,
      "grad_norm": 1.217188835144043,
      "learning_rate": 2.6850282485875707e-05,
      "loss": 2.6698,
      "step": 1639
    },
    {
      "epoch": 1.3898305084745763,
      "grad_norm": 1.331202507019043,
      "learning_rate": 2.68361581920904e-05,
      "loss": 2.7459,
      "step": 1640
    },
    {
      "epoch": 1.3906779661016948,
      "grad_norm": 0.9791933298110962,
      "learning_rate": 2.6822033898305087e-05,
      "loss": 2.7336,
      "step": 1641
    },
    {
      "epoch": 1.3915254237288135,
      "grad_norm": 1.183793306350708,
      "learning_rate": 2.6807909604519777e-05,
      "loss": 2.7254,
      "step": 1642
    },
    {
      "epoch": 1.3923728813559322,
      "grad_norm": 1.0322171449661255,
      "learning_rate": 2.6793785310734463e-05,
      "loss": 2.8907,
      "step": 1643
    },
    {
      "epoch": 1.3932203389830509,
      "grad_norm": 1.5175987482070923,
      "learning_rate": 2.6779661016949153e-05,
      "loss": 2.6094,
      "step": 1644
    },
    {
      "epoch": 1.3940677966101696,
      "grad_norm": 1.1093040704727173,
      "learning_rate": 2.6765536723163843e-05,
      "loss": 2.7547,
      "step": 1645
    },
    {
      "epoch": 1.3949152542372882,
      "grad_norm": 1.0322200059890747,
      "learning_rate": 2.675141242937853e-05,
      "loss": 2.8324,
      "step": 1646
    },
    {
      "epoch": 1.3957627118644067,
      "grad_norm": 1.0330455303192139,
      "learning_rate": 2.673728813559322e-05,
      "loss": 2.7325,
      "step": 1647
    },
    {
      "epoch": 1.3966101694915254,
      "grad_norm": 1.1793253421783447,
      "learning_rate": 2.6723163841807913e-05,
      "loss": 2.6088,
      "step": 1648
    },
    {
      "epoch": 1.397457627118644,
      "grad_norm": 1.4393837451934814,
      "learning_rate": 2.6709039548022603e-05,
      "loss": 2.5526,
      "step": 1649
    },
    {
      "epoch": 1.3983050847457628,
      "grad_norm": 1.3199998140335083,
      "learning_rate": 2.669491525423729e-05,
      "loss": 2.7828,
      "step": 1650
    },
    {
      "epoch": 1.3991525423728812,
      "grad_norm": 1.1920833587646484,
      "learning_rate": 2.668079096045198e-05,
      "loss": 2.7317,
      "step": 1651
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0866401195526123,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 2.7917,
      "step": 1652
    },
    {
      "epoch": 1.4008474576271186,
      "grad_norm": 1.4187414646148682,
      "learning_rate": 2.6652542372881357e-05,
      "loss": 2.5495,
      "step": 1653
    },
    {
      "epoch": 1.4016949152542373,
      "grad_norm": 1.0413742065429688,
      "learning_rate": 2.6638418079096043e-05,
      "loss": 2.6181,
      "step": 1654
    },
    {
      "epoch": 1.402542372881356,
      "grad_norm": 1.2924762964248657,
      "learning_rate": 2.6624293785310733e-05,
      "loss": 2.6635,
      "step": 1655
    },
    {
      "epoch": 1.4033898305084747,
      "grad_norm": 1.1318178176879883,
      "learning_rate": 2.6610169491525427e-05,
      "loss": 2.6875,
      "step": 1656
    },
    {
      "epoch": 1.4042372881355933,
      "grad_norm": 1.1205334663391113,
      "learning_rate": 2.6596045197740117e-05,
      "loss": 2.6214,
      "step": 1657
    },
    {
      "epoch": 1.4050847457627118,
      "grad_norm": 0.9691821336746216,
      "learning_rate": 2.6581920903954803e-05,
      "loss": 2.7269,
      "step": 1658
    },
    {
      "epoch": 1.4059322033898305,
      "grad_norm": 1.0846081972122192,
      "learning_rate": 2.6567796610169493e-05,
      "loss": 2.8449,
      "step": 1659
    },
    {
      "epoch": 1.4067796610169492,
      "grad_norm": 1.0578593015670776,
      "learning_rate": 2.6553672316384183e-05,
      "loss": 2.7101,
      "step": 1660
    },
    {
      "epoch": 1.4076271186440679,
      "grad_norm": 1.1737051010131836,
      "learning_rate": 2.653954802259887e-05,
      "loss": 2.6119,
      "step": 1661
    },
    {
      "epoch": 1.4084745762711863,
      "grad_norm": 1.1304816007614136,
      "learning_rate": 2.652542372881356e-05,
      "loss": 2.9181,
      "step": 1662
    },
    {
      "epoch": 1.409322033898305,
      "grad_norm": 1.335292100906372,
      "learning_rate": 2.6511299435028246e-05,
      "loss": 2.603,
      "step": 1663
    },
    {
      "epoch": 1.4101694915254237,
      "grad_norm": 1.1524286270141602,
      "learning_rate": 2.649717514124294e-05,
      "loss": 2.6836,
      "step": 1664
    },
    {
      "epoch": 1.4110169491525424,
      "grad_norm": 1.1399136781692505,
      "learning_rate": 2.648305084745763e-05,
      "loss": 2.693,
      "step": 1665
    },
    {
      "epoch": 1.411864406779661,
      "grad_norm": 1.0053728818893433,
      "learning_rate": 2.646892655367232e-05,
      "loss": 2.8955,
      "step": 1666
    },
    {
      "epoch": 1.4127118644067798,
      "grad_norm": 0.9324097037315369,
      "learning_rate": 2.6454802259887006e-05,
      "loss": 2.7561,
      "step": 1667
    },
    {
      "epoch": 1.4135593220338982,
      "grad_norm": 1.0216845273971558,
      "learning_rate": 2.6440677966101696e-05,
      "loss": 2.8979,
      "step": 1668
    },
    {
      "epoch": 1.414406779661017,
      "grad_norm": 1.094473958015442,
      "learning_rate": 2.6426553672316383e-05,
      "loss": 2.6899,
      "step": 1669
    },
    {
      "epoch": 1.4152542372881356,
      "grad_norm": 1.1630312204360962,
      "learning_rate": 2.6412429378531073e-05,
      "loss": 2.7004,
      "step": 1670
    },
    {
      "epoch": 1.4161016949152543,
      "grad_norm": 1.5137243270874023,
      "learning_rate": 2.6398305084745763e-05,
      "loss": 2.6018,
      "step": 1671
    },
    {
      "epoch": 1.4169491525423727,
      "grad_norm": 1.1621006727218628,
      "learning_rate": 2.6384180790960456e-05,
      "loss": 2.8431,
      "step": 1672
    },
    {
      "epoch": 1.4177966101694914,
      "grad_norm": 1.2741423845291138,
      "learning_rate": 2.6370056497175143e-05,
      "loss": 2.601,
      "step": 1673
    },
    {
      "epoch": 1.4186440677966101,
      "grad_norm": 1.2445948123931885,
      "learning_rate": 2.6355932203389833e-05,
      "loss": 2.6987,
      "step": 1674
    },
    {
      "epoch": 1.4194915254237288,
      "grad_norm": 1.028989553451538,
      "learning_rate": 2.634180790960452e-05,
      "loss": 2.7849,
      "step": 1675
    },
    {
      "epoch": 1.4203389830508475,
      "grad_norm": 0.9762065410614014,
      "learning_rate": 2.632768361581921e-05,
      "loss": 2.8959,
      "step": 1676
    },
    {
      "epoch": 1.4211864406779662,
      "grad_norm": 1.280434489250183,
      "learning_rate": 2.63135593220339e-05,
      "loss": 2.6658,
      "step": 1677
    },
    {
      "epoch": 1.4220338983050849,
      "grad_norm": 1.094059944152832,
      "learning_rate": 2.6299435028248586e-05,
      "loss": 2.757,
      "step": 1678
    },
    {
      "epoch": 1.4228813559322033,
      "grad_norm": 1.388394832611084,
      "learning_rate": 2.6285310734463276e-05,
      "loss": 2.7182,
      "step": 1679
    },
    {
      "epoch": 1.423728813559322,
      "grad_norm": 1.4729382991790771,
      "learning_rate": 2.627118644067797e-05,
      "loss": 2.5601,
      "step": 1680
    },
    {
      "epoch": 1.4245762711864407,
      "grad_norm": 1.4540657997131348,
      "learning_rate": 2.625706214689266e-05,
      "loss": 2.6347,
      "step": 1681
    },
    {
      "epoch": 1.4254237288135594,
      "grad_norm": 0.8756323456764221,
      "learning_rate": 2.6242937853107346e-05,
      "loss": 2.964,
      "step": 1682
    },
    {
      "epoch": 1.4262711864406779,
      "grad_norm": 1.0743720531463623,
      "learning_rate": 2.6228813559322036e-05,
      "loss": 2.8669,
      "step": 1683
    },
    {
      "epoch": 1.4271186440677965,
      "grad_norm": 1.1583024263381958,
      "learning_rate": 2.6214689265536723e-05,
      "loss": 2.7134,
      "step": 1684
    },
    {
      "epoch": 1.4279661016949152,
      "grad_norm": 1.476781964302063,
      "learning_rate": 2.6200564971751413e-05,
      "loss": 2.6654,
      "step": 1685
    },
    {
      "epoch": 1.428813559322034,
      "grad_norm": 1.3158514499664307,
      "learning_rate": 2.61864406779661e-05,
      "loss": 2.727,
      "step": 1686
    },
    {
      "epoch": 1.4296610169491526,
      "grad_norm": 1.1468459367752075,
      "learning_rate": 2.617231638418079e-05,
      "loss": 2.7465,
      "step": 1687
    },
    {
      "epoch": 1.4305084745762713,
      "grad_norm": 1.385338306427002,
      "learning_rate": 2.6158192090395483e-05,
      "loss": 2.4701,
      "step": 1688
    },
    {
      "epoch": 1.43135593220339,
      "grad_norm": 1.102578043937683,
      "learning_rate": 2.6144067796610173e-05,
      "loss": 2.9113,
      "step": 1689
    },
    {
      "epoch": 1.4322033898305084,
      "grad_norm": 1.049554705619812,
      "learning_rate": 2.612994350282486e-05,
      "loss": 2.6923,
      "step": 1690
    },
    {
      "epoch": 1.4330508474576271,
      "grad_norm": 1.1550787687301636,
      "learning_rate": 2.611581920903955e-05,
      "loss": 2.8105,
      "step": 1691
    },
    {
      "epoch": 1.4338983050847458,
      "grad_norm": 1.4350584745407104,
      "learning_rate": 2.610169491525424e-05,
      "loss": 2.5275,
      "step": 1692
    },
    {
      "epoch": 1.4347457627118643,
      "grad_norm": 0.9529697895050049,
      "learning_rate": 2.6087570621468926e-05,
      "loss": 2.7931,
      "step": 1693
    },
    {
      "epoch": 1.435593220338983,
      "grad_norm": 1.3449734449386597,
      "learning_rate": 2.6073446327683616e-05,
      "loss": 2.6613,
      "step": 1694
    },
    {
      "epoch": 1.4364406779661016,
      "grad_norm": 0.968408465385437,
      "learning_rate": 2.605932203389831e-05,
      "loss": 2.8287,
      "step": 1695
    },
    {
      "epoch": 1.4372881355932203,
      "grad_norm": 1.1594336032867432,
      "learning_rate": 2.6045197740113e-05,
      "loss": 2.8094,
      "step": 1696
    },
    {
      "epoch": 1.438135593220339,
      "grad_norm": 1.3010283708572388,
      "learning_rate": 2.6031073446327686e-05,
      "loss": 2.635,
      "step": 1697
    },
    {
      "epoch": 1.4389830508474577,
      "grad_norm": 1.1900665760040283,
      "learning_rate": 2.6016949152542376e-05,
      "loss": 2.6159,
      "step": 1698
    },
    {
      "epoch": 1.4398305084745764,
      "grad_norm": 1.013306975364685,
      "learning_rate": 2.6002824858757062e-05,
      "loss": 2.8351,
      "step": 1699
    },
    {
      "epoch": 1.4406779661016949,
      "grad_norm": 1.3891487121582031,
      "learning_rate": 2.5988700564971752e-05,
      "loss": 2.4998,
      "step": 1700
    },
    {
      "epoch": 1.4415254237288135,
      "grad_norm": 1.34544038772583,
      "learning_rate": 2.597457627118644e-05,
      "loss": 2.7262,
      "step": 1701
    },
    {
      "epoch": 1.4423728813559322,
      "grad_norm": 1.1992326974868774,
      "learning_rate": 2.596045197740113e-05,
      "loss": 2.6216,
      "step": 1702
    },
    {
      "epoch": 1.443220338983051,
      "grad_norm": 1.0708142518997192,
      "learning_rate": 2.5946327683615822e-05,
      "loss": 2.8212,
      "step": 1703
    },
    {
      "epoch": 1.4440677966101694,
      "grad_norm": 1.231167197227478,
      "learning_rate": 2.5932203389830512e-05,
      "loss": 2.6428,
      "step": 1704
    },
    {
      "epoch": 1.444915254237288,
      "grad_norm": 0.9052745699882507,
      "learning_rate": 2.59180790960452e-05,
      "loss": 2.9066,
      "step": 1705
    },
    {
      "epoch": 1.4457627118644067,
      "grad_norm": 1.2641510963439941,
      "learning_rate": 2.590395480225989e-05,
      "loss": 2.6419,
      "step": 1706
    },
    {
      "epoch": 1.4466101694915254,
      "grad_norm": 1.1792833805084229,
      "learning_rate": 2.588983050847458e-05,
      "loss": 2.7039,
      "step": 1707
    },
    {
      "epoch": 1.4474576271186441,
      "grad_norm": 1.0563932657241821,
      "learning_rate": 2.5875706214689266e-05,
      "loss": 2.7077,
      "step": 1708
    },
    {
      "epoch": 1.4483050847457628,
      "grad_norm": 1.3587576150894165,
      "learning_rate": 2.5861581920903956e-05,
      "loss": 2.8453,
      "step": 1709
    },
    {
      "epoch": 1.4491525423728815,
      "grad_norm": 1.1781967878341675,
      "learning_rate": 2.5847457627118642e-05,
      "loss": 2.7997,
      "step": 1710
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.46746027469635,
      "learning_rate": 2.5833333333333336e-05,
      "loss": 2.6531,
      "step": 1711
    },
    {
      "epoch": 1.4508474576271186,
      "grad_norm": 0.918735682964325,
      "learning_rate": 2.5819209039548026e-05,
      "loss": 2.9465,
      "step": 1712
    },
    {
      "epoch": 1.4516949152542373,
      "grad_norm": 1.0160126686096191,
      "learning_rate": 2.5805084745762716e-05,
      "loss": 2.9582,
      "step": 1713
    },
    {
      "epoch": 1.452542372881356,
      "grad_norm": 1.0358372926712036,
      "learning_rate": 2.5790960451977402e-05,
      "loss": 2.6894,
      "step": 1714
    },
    {
      "epoch": 1.4533898305084745,
      "grad_norm": 1.0059192180633545,
      "learning_rate": 2.5776836158192092e-05,
      "loss": 2.8616,
      "step": 1715
    },
    {
      "epoch": 1.4542372881355932,
      "grad_norm": 1.1379948854446411,
      "learning_rate": 2.576271186440678e-05,
      "loss": 2.6607,
      "step": 1716
    },
    {
      "epoch": 1.4550847457627119,
      "grad_norm": 1.1622596979141235,
      "learning_rate": 2.574858757062147e-05,
      "loss": 2.8035,
      "step": 1717
    },
    {
      "epoch": 1.4559322033898305,
      "grad_norm": 1.3360217809677124,
      "learning_rate": 2.573446327683616e-05,
      "loss": 2.6771,
      "step": 1718
    },
    {
      "epoch": 1.4567796610169492,
      "grad_norm": 1.2945500612258911,
      "learning_rate": 2.5720338983050852e-05,
      "loss": 2.5487,
      "step": 1719
    },
    {
      "epoch": 1.457627118644068,
      "grad_norm": 1.3775050640106201,
      "learning_rate": 2.570621468926554e-05,
      "loss": 2.5943,
      "step": 1720
    },
    {
      "epoch": 1.4584745762711864,
      "grad_norm": 1.2074862718582153,
      "learning_rate": 2.569209039548023e-05,
      "loss": 2.7334,
      "step": 1721
    },
    {
      "epoch": 1.459322033898305,
      "grad_norm": 0.8728470802307129,
      "learning_rate": 2.5677966101694915e-05,
      "loss": 2.9732,
      "step": 1722
    },
    {
      "epoch": 1.4601694915254237,
      "grad_norm": 1.1561737060546875,
      "learning_rate": 2.5663841807909605e-05,
      "loss": 2.7425,
      "step": 1723
    },
    {
      "epoch": 1.4610169491525424,
      "grad_norm": 1.5228084325790405,
      "learning_rate": 2.5649717514124295e-05,
      "loss": 2.4591,
      "step": 1724
    },
    {
      "epoch": 1.461864406779661,
      "grad_norm": 1.0325641632080078,
      "learning_rate": 2.5635593220338982e-05,
      "loss": 2.7793,
      "step": 1725
    },
    {
      "epoch": 1.4627118644067796,
      "grad_norm": 1.1751588582992554,
      "learning_rate": 2.5621468926553672e-05,
      "loss": 2.8034,
      "step": 1726
    },
    {
      "epoch": 1.4635593220338983,
      "grad_norm": 1.2335675954818726,
      "learning_rate": 2.5607344632768365e-05,
      "loss": 2.6746,
      "step": 1727
    },
    {
      "epoch": 1.464406779661017,
      "grad_norm": 1.31641685962677,
      "learning_rate": 2.5593220338983055e-05,
      "loss": 2.678,
      "step": 1728
    },
    {
      "epoch": 1.4652542372881356,
      "grad_norm": 1.3009672164916992,
      "learning_rate": 2.5579096045197742e-05,
      "loss": 2.8398,
      "step": 1729
    },
    {
      "epoch": 1.4661016949152543,
      "grad_norm": 1.1604074239730835,
      "learning_rate": 2.5564971751412432e-05,
      "loss": 2.8324,
      "step": 1730
    },
    {
      "epoch": 1.466949152542373,
      "grad_norm": 1.3685365915298462,
      "learning_rate": 2.555084745762712e-05,
      "loss": 2.5721,
      "step": 1731
    },
    {
      "epoch": 1.4677966101694915,
      "grad_norm": 1.187212347984314,
      "learning_rate": 2.553672316384181e-05,
      "loss": 2.7589,
      "step": 1732
    },
    {
      "epoch": 1.4686440677966102,
      "grad_norm": 1.1539112329483032,
      "learning_rate": 2.5522598870056495e-05,
      "loss": 2.6395,
      "step": 1733
    },
    {
      "epoch": 1.4694915254237289,
      "grad_norm": 1.0945439338684082,
      "learning_rate": 2.5508474576271185e-05,
      "loss": 2.8949,
      "step": 1734
    },
    {
      "epoch": 1.4703389830508475,
      "grad_norm": 1.189335584640503,
      "learning_rate": 2.549435028248588e-05,
      "loss": 2.7979,
      "step": 1735
    },
    {
      "epoch": 1.471186440677966,
      "grad_norm": 1.138032078742981,
      "learning_rate": 2.548022598870057e-05,
      "loss": 2.7786,
      "step": 1736
    },
    {
      "epoch": 1.4720338983050847,
      "grad_norm": 1.6101855039596558,
      "learning_rate": 2.5466101694915255e-05,
      "loss": 2.3606,
      "step": 1737
    },
    {
      "epoch": 1.4728813559322034,
      "grad_norm": 0.9621807336807251,
      "learning_rate": 2.5451977401129945e-05,
      "loss": 2.8372,
      "step": 1738
    },
    {
      "epoch": 1.473728813559322,
      "grad_norm": 1.0987484455108643,
      "learning_rate": 2.5437853107344635e-05,
      "loss": 2.7639,
      "step": 1739
    },
    {
      "epoch": 1.4745762711864407,
      "grad_norm": 1.103590488433838,
      "learning_rate": 2.5423728813559322e-05,
      "loss": 2.6284,
      "step": 1740
    },
    {
      "epoch": 1.4754237288135594,
      "grad_norm": 0.938614547252655,
      "learning_rate": 2.5409604519774012e-05,
      "loss": 2.7863,
      "step": 1741
    },
    {
      "epoch": 1.476271186440678,
      "grad_norm": 1.0708339214324951,
      "learning_rate": 2.5395480225988698e-05,
      "loss": 2.7349,
      "step": 1742
    },
    {
      "epoch": 1.4771186440677966,
      "grad_norm": 0.8444576859474182,
      "learning_rate": 2.538135593220339e-05,
      "loss": 2.8595,
      "step": 1743
    },
    {
      "epoch": 1.4779661016949153,
      "grad_norm": 1.3227760791778564,
      "learning_rate": 2.536723163841808e-05,
      "loss": 2.7629,
      "step": 1744
    },
    {
      "epoch": 1.478813559322034,
      "grad_norm": 1.4015511274337769,
      "learning_rate": 2.535310734463277e-05,
      "loss": 2.6918,
      "step": 1745
    },
    {
      "epoch": 1.4796610169491524,
      "grad_norm": 0.9624196887016296,
      "learning_rate": 2.5338983050847458e-05,
      "loss": 2.8514,
      "step": 1746
    },
    {
      "epoch": 1.480508474576271,
      "grad_norm": 1.2324362993240356,
      "learning_rate": 2.5324858757062148e-05,
      "loss": 2.6816,
      "step": 1747
    },
    {
      "epoch": 1.4813559322033898,
      "grad_norm": 1.64078688621521,
      "learning_rate": 2.5310734463276835e-05,
      "loss": 2.4035,
      "step": 1748
    },
    {
      "epoch": 1.4822033898305085,
      "grad_norm": 1.0446299314498901,
      "learning_rate": 2.5296610169491525e-05,
      "loss": 2.824,
      "step": 1749
    },
    {
      "epoch": 1.4830508474576272,
      "grad_norm": 0.9898091554641724,
      "learning_rate": 2.5282485875706215e-05,
      "loss": 2.8308,
      "step": 1750
    },
    {
      "epoch": 1.4838983050847459,
      "grad_norm": 1.2462795972824097,
      "learning_rate": 2.5268361581920908e-05,
      "loss": 2.6622,
      "step": 1751
    },
    {
      "epoch": 1.4847457627118645,
      "grad_norm": 1.0705769062042236,
      "learning_rate": 2.5254237288135595e-05,
      "loss": 2.7527,
      "step": 1752
    },
    {
      "epoch": 1.485593220338983,
      "grad_norm": 1.0593948364257812,
      "learning_rate": 2.5240112994350285e-05,
      "loss": 2.8198,
      "step": 1753
    },
    {
      "epoch": 1.4864406779661017,
      "grad_norm": 1.1521574258804321,
      "learning_rate": 2.522598870056497e-05,
      "loss": 2.8464,
      "step": 1754
    },
    {
      "epoch": 1.4872881355932204,
      "grad_norm": 1.0825778245925903,
      "learning_rate": 2.521186440677966e-05,
      "loss": 2.6956,
      "step": 1755
    },
    {
      "epoch": 1.488135593220339,
      "grad_norm": 1.1135597229003906,
      "learning_rate": 2.519774011299435e-05,
      "loss": 2.7401,
      "step": 1756
    },
    {
      "epoch": 1.4889830508474575,
      "grad_norm": 1.4296479225158691,
      "learning_rate": 2.5183615819209038e-05,
      "loss": 2.5709,
      "step": 1757
    },
    {
      "epoch": 1.4898305084745762,
      "grad_norm": 1.0948504209518433,
      "learning_rate": 2.5169491525423728e-05,
      "loss": 2.5154,
      "step": 1758
    },
    {
      "epoch": 1.490677966101695,
      "grad_norm": 1.2337549924850464,
      "learning_rate": 2.515536723163842e-05,
      "loss": 2.5797,
      "step": 1759
    },
    {
      "epoch": 1.4915254237288136,
      "grad_norm": 1.142493724822998,
      "learning_rate": 2.514124293785311e-05,
      "loss": 2.7528,
      "step": 1760
    },
    {
      "epoch": 1.4923728813559323,
      "grad_norm": 1.2989643812179565,
      "learning_rate": 2.5127118644067798e-05,
      "loss": 2.4817,
      "step": 1761
    },
    {
      "epoch": 1.493220338983051,
      "grad_norm": 1.1945594549179077,
      "learning_rate": 2.5112994350282488e-05,
      "loss": 2.8188,
      "step": 1762
    },
    {
      "epoch": 1.4940677966101694,
      "grad_norm": 1.234764575958252,
      "learning_rate": 2.5098870056497175e-05,
      "loss": 2.703,
      "step": 1763
    },
    {
      "epoch": 1.494915254237288,
      "grad_norm": 1.1107516288757324,
      "learning_rate": 2.5084745762711865e-05,
      "loss": 2.649,
      "step": 1764
    },
    {
      "epoch": 1.4957627118644068,
      "grad_norm": 1.1896271705627441,
      "learning_rate": 2.507062146892655e-05,
      "loss": 2.7029,
      "step": 1765
    },
    {
      "epoch": 1.4966101694915255,
      "grad_norm": 0.9180924296379089,
      "learning_rate": 2.505649717514124e-05,
      "loss": 2.8742,
      "step": 1766
    },
    {
      "epoch": 1.497457627118644,
      "grad_norm": 0.9900988340377808,
      "learning_rate": 2.5042372881355935e-05,
      "loss": 2.7208,
      "step": 1767
    },
    {
      "epoch": 1.4983050847457626,
      "grad_norm": 1.1403770446777344,
      "learning_rate": 2.5028248587570625e-05,
      "loss": 2.6409,
      "step": 1768
    },
    {
      "epoch": 1.4991525423728813,
      "grad_norm": 1.050744652748108,
      "learning_rate": 2.501412429378531e-05,
      "loss": 2.6425,
      "step": 1769
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0845805406570435,
      "learning_rate": 2.5e-05,
      "loss": 2.7535,
      "step": 1770
    },
    {
      "epoch": 1.5008474576271187,
      "grad_norm": 1.1701406240463257,
      "learning_rate": 2.498587570621469e-05,
      "loss": 2.7587,
      "step": 1771
    },
    {
      "epoch": 1.5016949152542374,
      "grad_norm": 1.1284900903701782,
      "learning_rate": 2.497175141242938e-05,
      "loss": 2.6616,
      "step": 1772
    },
    {
      "epoch": 1.502542372881356,
      "grad_norm": 1.1275113821029663,
      "learning_rate": 2.495762711864407e-05,
      "loss": 2.7371,
      "step": 1773
    },
    {
      "epoch": 1.5033898305084745,
      "grad_norm": 1.18697190284729,
      "learning_rate": 2.4943502824858758e-05,
      "loss": 2.7345,
      "step": 1774
    },
    {
      "epoch": 1.5042372881355932,
      "grad_norm": 1.1543711423873901,
      "learning_rate": 2.4929378531073448e-05,
      "loss": 2.7861,
      "step": 1775
    },
    {
      "epoch": 1.505084745762712,
      "grad_norm": 0.9525389671325684,
      "learning_rate": 2.4915254237288138e-05,
      "loss": 2.7745,
      "step": 1776
    },
    {
      "epoch": 1.5059322033898304,
      "grad_norm": 1.243618369102478,
      "learning_rate": 2.4901129943502828e-05,
      "loss": 2.5957,
      "step": 1777
    },
    {
      "epoch": 1.506779661016949,
      "grad_norm": 0.9921051263809204,
      "learning_rate": 2.4887005649717514e-05,
      "loss": 2.6621,
      "step": 1778
    },
    {
      "epoch": 1.5076271186440677,
      "grad_norm": 0.9193978309631348,
      "learning_rate": 2.4872881355932204e-05,
      "loss": 2.8639,
      "step": 1779
    },
    {
      "epoch": 1.5084745762711864,
      "grad_norm": 1.0668909549713135,
      "learning_rate": 2.4858757062146894e-05,
      "loss": 2.8757,
      "step": 1780
    },
    {
      "epoch": 1.509322033898305,
      "grad_norm": 1.579140543937683,
      "learning_rate": 2.4844632768361584e-05,
      "loss": 2.4371,
      "step": 1781
    },
    {
      "epoch": 1.5101694915254238,
      "grad_norm": 1.1261061429977417,
      "learning_rate": 2.483050847457627e-05,
      "loss": 2.6986,
      "step": 1782
    },
    {
      "epoch": 1.5110169491525425,
      "grad_norm": 1.0194988250732422,
      "learning_rate": 2.481638418079096e-05,
      "loss": 2.749,
      "step": 1783
    },
    {
      "epoch": 1.5118644067796612,
      "grad_norm": 1.1523332595825195,
      "learning_rate": 2.480225988700565e-05,
      "loss": 2.7165,
      "step": 1784
    },
    {
      "epoch": 1.5127118644067796,
      "grad_norm": 1.1937497854232788,
      "learning_rate": 2.478813559322034e-05,
      "loss": 2.7766,
      "step": 1785
    },
    {
      "epoch": 1.5135593220338983,
      "grad_norm": 1.128448486328125,
      "learning_rate": 2.477401129943503e-05,
      "loss": 2.7316,
      "step": 1786
    },
    {
      "epoch": 1.514406779661017,
      "grad_norm": 1.3518177270889282,
      "learning_rate": 2.4759887005649718e-05,
      "loss": 2.6125,
      "step": 1787
    },
    {
      "epoch": 1.5152542372881355,
      "grad_norm": 1.0886684656143188,
      "learning_rate": 2.4745762711864408e-05,
      "loss": 2.8088,
      "step": 1788
    },
    {
      "epoch": 1.5161016949152541,
      "grad_norm": 1.0553864240646362,
      "learning_rate": 2.4731638418079098e-05,
      "loss": 2.7269,
      "step": 1789
    },
    {
      "epoch": 1.5169491525423728,
      "grad_norm": 0.9939340949058533,
      "learning_rate": 2.4717514124293788e-05,
      "loss": 2.8575,
      "step": 1790
    },
    {
      "epoch": 1.5177966101694915,
      "grad_norm": 1.4051233530044556,
      "learning_rate": 2.4703389830508474e-05,
      "loss": 2.6995,
      "step": 1791
    },
    {
      "epoch": 1.5186440677966102,
      "grad_norm": 1.1315674781799316,
      "learning_rate": 2.4689265536723168e-05,
      "loss": 2.8004,
      "step": 1792
    },
    {
      "epoch": 1.519491525423729,
      "grad_norm": 1.2620928287506104,
      "learning_rate": 2.4675141242937854e-05,
      "loss": 2.6974,
      "step": 1793
    },
    {
      "epoch": 1.5203389830508476,
      "grad_norm": 1.2280446290969849,
      "learning_rate": 2.4661016949152544e-05,
      "loss": 2.5931,
      "step": 1794
    },
    {
      "epoch": 1.5211864406779663,
      "grad_norm": 1.1947920322418213,
      "learning_rate": 2.464689265536723e-05,
      "loss": 2.6941,
      "step": 1795
    },
    {
      "epoch": 1.5220338983050847,
      "grad_norm": 1.2421876192092896,
      "learning_rate": 2.4632768361581924e-05,
      "loss": 2.7424,
      "step": 1796
    },
    {
      "epoch": 1.5228813559322034,
      "grad_norm": 1.032297134399414,
      "learning_rate": 2.461864406779661e-05,
      "loss": 2.8071,
      "step": 1797
    },
    {
      "epoch": 1.5237288135593219,
      "grad_norm": 1.0700403451919556,
      "learning_rate": 2.46045197740113e-05,
      "loss": 2.9141,
      "step": 1798
    },
    {
      "epoch": 1.5245762711864406,
      "grad_norm": 1.2481400966644287,
      "learning_rate": 2.4590395480225987e-05,
      "loss": 2.7076,
      "step": 1799
    },
    {
      "epoch": 1.5254237288135593,
      "grad_norm": 1.1966779232025146,
      "learning_rate": 2.457627118644068e-05,
      "loss": 2.6864,
      "step": 1800
    },
    {
      "epoch": 1.526271186440678,
      "grad_norm": 1.036336898803711,
      "learning_rate": 2.4562146892655367e-05,
      "loss": 2.7467,
      "step": 1801
    },
    {
      "epoch": 1.5271186440677966,
      "grad_norm": 0.9020054936408997,
      "learning_rate": 2.4548022598870057e-05,
      "loss": 3.0466,
      "step": 1802
    },
    {
      "epoch": 1.5279661016949153,
      "grad_norm": 1.2337982654571533,
      "learning_rate": 2.4533898305084747e-05,
      "loss": 2.6878,
      "step": 1803
    },
    {
      "epoch": 1.528813559322034,
      "grad_norm": 1.003456473350525,
      "learning_rate": 2.4519774011299437e-05,
      "loss": 2.9421,
      "step": 1804
    },
    {
      "epoch": 1.5296610169491527,
      "grad_norm": 0.945570170879364,
      "learning_rate": 2.4505649717514127e-05,
      "loss": 2.7682,
      "step": 1805
    },
    {
      "epoch": 1.5305084745762711,
      "grad_norm": 1.0575337409973145,
      "learning_rate": 2.4491525423728814e-05,
      "loss": 2.7781,
      "step": 1806
    },
    {
      "epoch": 1.5313559322033898,
      "grad_norm": 1.2243188619613647,
      "learning_rate": 2.4477401129943504e-05,
      "loss": 2.8299,
      "step": 1807
    },
    {
      "epoch": 1.5322033898305085,
      "grad_norm": 1.0221068859100342,
      "learning_rate": 2.4463276836158194e-05,
      "loss": 2.863,
      "step": 1808
    },
    {
      "epoch": 1.533050847457627,
      "grad_norm": 0.961390495300293,
      "learning_rate": 2.4449152542372884e-05,
      "loss": 2.7633,
      "step": 1809
    },
    {
      "epoch": 1.5338983050847457,
      "grad_norm": 1.3998870849609375,
      "learning_rate": 2.443502824858757e-05,
      "loss": 2.5782,
      "step": 1810
    },
    {
      "epoch": 1.5347457627118644,
      "grad_norm": 1.0910758972167969,
      "learning_rate": 2.442090395480226e-05,
      "loss": 2.8715,
      "step": 1811
    },
    {
      "epoch": 1.535593220338983,
      "grad_norm": 1.138889193534851,
      "learning_rate": 2.440677966101695e-05,
      "loss": 2.7039,
      "step": 1812
    },
    {
      "epoch": 1.5364406779661017,
      "grad_norm": 1.3160468339920044,
      "learning_rate": 2.439265536723164e-05,
      "loss": 2.6346,
      "step": 1813
    },
    {
      "epoch": 1.5372881355932204,
      "grad_norm": 1.3582723140716553,
      "learning_rate": 2.4378531073446327e-05,
      "loss": 2.5785,
      "step": 1814
    },
    {
      "epoch": 1.538135593220339,
      "grad_norm": 1.2774215936660767,
      "learning_rate": 2.4364406779661017e-05,
      "loss": 2.7209,
      "step": 1815
    },
    {
      "epoch": 1.5389830508474578,
      "grad_norm": 1.0066934823989868,
      "learning_rate": 2.4350282485875707e-05,
      "loss": 2.8582,
      "step": 1816
    },
    {
      "epoch": 1.5398305084745763,
      "grad_norm": 1.0705580711364746,
      "learning_rate": 2.4336158192090397e-05,
      "loss": 2.7616,
      "step": 1817
    },
    {
      "epoch": 1.540677966101695,
      "grad_norm": 1.2924797534942627,
      "learning_rate": 2.4322033898305087e-05,
      "loss": 2.7503,
      "step": 1818
    },
    {
      "epoch": 1.5415254237288134,
      "grad_norm": 1.266080617904663,
      "learning_rate": 2.4307909604519774e-05,
      "loss": 2.6332,
      "step": 1819
    },
    {
      "epoch": 1.542372881355932,
      "grad_norm": 0.9904401302337646,
      "learning_rate": 2.4293785310734467e-05,
      "loss": 2.7892,
      "step": 1820
    },
    {
      "epoch": 1.5432203389830508,
      "grad_norm": 1.0994033813476562,
      "learning_rate": 2.4279661016949154e-05,
      "loss": 2.8478,
      "step": 1821
    },
    {
      "epoch": 1.5440677966101695,
      "grad_norm": 1.2663304805755615,
      "learning_rate": 2.4265536723163844e-05,
      "loss": 2.888,
      "step": 1822
    },
    {
      "epoch": 1.5449152542372881,
      "grad_norm": 1.1716861724853516,
      "learning_rate": 2.425141242937853e-05,
      "loss": 2.6604,
      "step": 1823
    },
    {
      "epoch": 1.5457627118644068,
      "grad_norm": 1.1331392526626587,
      "learning_rate": 2.4237288135593224e-05,
      "loss": 2.7479,
      "step": 1824
    },
    {
      "epoch": 1.5466101694915255,
      "grad_norm": 1.0596132278442383,
      "learning_rate": 2.422316384180791e-05,
      "loss": 2.6982,
      "step": 1825
    },
    {
      "epoch": 1.5474576271186442,
      "grad_norm": 1.2000031471252441,
      "learning_rate": 2.42090395480226e-05,
      "loss": 2.745,
      "step": 1826
    },
    {
      "epoch": 1.5483050847457627,
      "grad_norm": 1.1302003860473633,
      "learning_rate": 2.4194915254237287e-05,
      "loss": 2.7681,
      "step": 1827
    },
    {
      "epoch": 1.5491525423728814,
      "grad_norm": 1.099095106124878,
      "learning_rate": 2.418079096045198e-05,
      "loss": 2.5677,
      "step": 1828
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.074489951133728,
      "learning_rate": 2.4166666666666667e-05,
      "loss": 2.7839,
      "step": 1829
    },
    {
      "epoch": 1.5508474576271185,
      "grad_norm": 1.2375173568725586,
      "learning_rate": 2.4152542372881357e-05,
      "loss": 2.7313,
      "step": 1830
    },
    {
      "epoch": 1.5516949152542372,
      "grad_norm": 1.0323246717453003,
      "learning_rate": 2.4138418079096047e-05,
      "loss": 2.8609,
      "step": 1831
    },
    {
      "epoch": 1.5525423728813559,
      "grad_norm": 1.2047739028930664,
      "learning_rate": 2.4124293785310737e-05,
      "loss": 2.6048,
      "step": 1832
    },
    {
      "epoch": 1.5533898305084746,
      "grad_norm": 1.1812658309936523,
      "learning_rate": 2.4110169491525423e-05,
      "loss": 2.7495,
      "step": 1833
    },
    {
      "epoch": 1.5542372881355933,
      "grad_norm": 1.3200953006744385,
      "learning_rate": 2.4096045197740113e-05,
      "loss": 2.671,
      "step": 1834
    },
    {
      "epoch": 1.555084745762712,
      "grad_norm": 1.1331515312194824,
      "learning_rate": 2.4081920903954803e-05,
      "loss": 2.7263,
      "step": 1835
    },
    {
      "epoch": 1.5559322033898306,
      "grad_norm": 0.9608669281005859,
      "learning_rate": 2.4067796610169493e-05,
      "loss": 2.8821,
      "step": 1836
    },
    {
      "epoch": 1.5567796610169493,
      "grad_norm": 0.9678069949150085,
      "learning_rate": 2.4053672316384183e-05,
      "loss": 2.857,
      "step": 1837
    },
    {
      "epoch": 1.5576271186440678,
      "grad_norm": 1.0414766073226929,
      "learning_rate": 2.403954802259887e-05,
      "loss": 2.7889,
      "step": 1838
    },
    {
      "epoch": 1.5584745762711865,
      "grad_norm": 1.003260612487793,
      "learning_rate": 2.402542372881356e-05,
      "loss": 2.7885,
      "step": 1839
    },
    {
      "epoch": 1.559322033898305,
      "grad_norm": 1.2616666555404663,
      "learning_rate": 2.401129943502825e-05,
      "loss": 2.8512,
      "step": 1840
    },
    {
      "epoch": 1.5601694915254236,
      "grad_norm": 1.6481845378875732,
      "learning_rate": 2.399717514124294e-05,
      "loss": 2.5663,
      "step": 1841
    },
    {
      "epoch": 1.5610169491525423,
      "grad_norm": 1.0453968048095703,
      "learning_rate": 2.3983050847457627e-05,
      "loss": 2.7644,
      "step": 1842
    },
    {
      "epoch": 1.561864406779661,
      "grad_norm": 0.992255687713623,
      "learning_rate": 2.3968926553672317e-05,
      "loss": 2.8203,
      "step": 1843
    },
    {
      "epoch": 1.5627118644067797,
      "grad_norm": 1.561972737312317,
      "learning_rate": 2.3954802259887007e-05,
      "loss": 2.581,
      "step": 1844
    },
    {
      "epoch": 1.5635593220338984,
      "grad_norm": 1.0985513925552368,
      "learning_rate": 2.3940677966101697e-05,
      "loss": 2.7722,
      "step": 1845
    },
    {
      "epoch": 1.564406779661017,
      "grad_norm": 1.2070634365081787,
      "learning_rate": 2.3926553672316383e-05,
      "loss": 2.6985,
      "step": 1846
    },
    {
      "epoch": 1.5652542372881357,
      "grad_norm": 1.043664574623108,
      "learning_rate": 2.3912429378531073e-05,
      "loss": 2.715,
      "step": 1847
    },
    {
      "epoch": 1.5661016949152542,
      "grad_norm": 1.4287388324737549,
      "learning_rate": 2.3898305084745763e-05,
      "loss": 2.6028,
      "step": 1848
    },
    {
      "epoch": 1.5669491525423729,
      "grad_norm": 1.2643784284591675,
      "learning_rate": 2.3884180790960453e-05,
      "loss": 2.6177,
      "step": 1849
    },
    {
      "epoch": 1.5677966101694916,
      "grad_norm": 1.0906355381011963,
      "learning_rate": 2.3870056497175143e-05,
      "loss": 2.8016,
      "step": 1850
    },
    {
      "epoch": 1.56864406779661,
      "grad_norm": 1.302488088607788,
      "learning_rate": 2.3855932203389833e-05,
      "loss": 2.683,
      "step": 1851
    },
    {
      "epoch": 1.5694915254237287,
      "grad_norm": 1.0560115575790405,
      "learning_rate": 2.3841807909604523e-05,
      "loss": 2.8361,
      "step": 1852
    },
    {
      "epoch": 1.5703389830508474,
      "grad_norm": 1.3614486455917358,
      "learning_rate": 2.382768361581921e-05,
      "loss": 2.5801,
      "step": 1853
    },
    {
      "epoch": 1.571186440677966,
      "grad_norm": 1.3357983827590942,
      "learning_rate": 2.38135593220339e-05,
      "loss": 2.6701,
      "step": 1854
    },
    {
      "epoch": 1.5720338983050848,
      "grad_norm": 1.0064375400543213,
      "learning_rate": 2.379943502824859e-05,
      "loss": 2.8113,
      "step": 1855
    },
    {
      "epoch": 1.5728813559322035,
      "grad_norm": 1.1035053730010986,
      "learning_rate": 2.378531073446328e-05,
      "loss": 2.822,
      "step": 1856
    },
    {
      "epoch": 1.5737288135593221,
      "grad_norm": 0.9912019371986389,
      "learning_rate": 2.3771186440677966e-05,
      "loss": 2.8875,
      "step": 1857
    },
    {
      "epoch": 1.5745762711864408,
      "grad_norm": 1.3179409503936768,
      "learning_rate": 2.3757062146892656e-05,
      "loss": 2.6668,
      "step": 1858
    },
    {
      "epoch": 1.5754237288135593,
      "grad_norm": 1.052337884902954,
      "learning_rate": 2.3742937853107346e-05,
      "loss": 2.7392,
      "step": 1859
    },
    {
      "epoch": 1.576271186440678,
      "grad_norm": 1.140894889831543,
      "learning_rate": 2.3728813559322036e-05,
      "loss": 2.8434,
      "step": 1860
    },
    {
      "epoch": 1.5771186440677964,
      "grad_norm": 1.0413471460342407,
      "learning_rate": 2.3714689265536723e-05,
      "loss": 2.7602,
      "step": 1861
    },
    {
      "epoch": 1.5779661016949151,
      "grad_norm": 1.1304707527160645,
      "learning_rate": 2.3700564971751413e-05,
      "loss": 2.5664,
      "step": 1862
    },
    {
      "epoch": 1.5788135593220338,
      "grad_norm": 0.958276629447937,
      "learning_rate": 2.3686440677966103e-05,
      "loss": 2.684,
      "step": 1863
    },
    {
      "epoch": 1.5796610169491525,
      "grad_norm": 1.195338249206543,
      "learning_rate": 2.3672316384180793e-05,
      "loss": 2.7652,
      "step": 1864
    },
    {
      "epoch": 1.5805084745762712,
      "grad_norm": 1.096615195274353,
      "learning_rate": 2.3658192090395483e-05,
      "loss": 2.6835,
      "step": 1865
    },
    {
      "epoch": 1.5813559322033899,
      "grad_norm": 1.0772604942321777,
      "learning_rate": 2.364406779661017e-05,
      "loss": 2.7362,
      "step": 1866
    },
    {
      "epoch": 1.5822033898305086,
      "grad_norm": 1.1231951713562012,
      "learning_rate": 2.362994350282486e-05,
      "loss": 2.7992,
      "step": 1867
    },
    {
      "epoch": 1.5830508474576273,
      "grad_norm": 1.3755862712860107,
      "learning_rate": 2.361581920903955e-05,
      "loss": 2.6187,
      "step": 1868
    },
    {
      "epoch": 1.5838983050847457,
      "grad_norm": 0.9679206013679504,
      "learning_rate": 2.360169491525424e-05,
      "loss": 2.7601,
      "step": 1869
    },
    {
      "epoch": 1.5847457627118644,
      "grad_norm": 1.0926215648651123,
      "learning_rate": 2.3587570621468926e-05,
      "loss": 2.6911,
      "step": 1870
    },
    {
      "epoch": 1.585593220338983,
      "grad_norm": 1.1125919818878174,
      "learning_rate": 2.357344632768362e-05,
      "loss": 2.7169,
      "step": 1871
    },
    {
      "epoch": 1.5864406779661016,
      "grad_norm": 1.0439825057983398,
      "learning_rate": 2.3559322033898306e-05,
      "loss": 2.8413,
      "step": 1872
    },
    {
      "epoch": 1.5872881355932202,
      "grad_norm": 1.411345362663269,
      "learning_rate": 2.3545197740112996e-05,
      "loss": 2.7078,
      "step": 1873
    },
    {
      "epoch": 1.588135593220339,
      "grad_norm": 1.0051039457321167,
      "learning_rate": 2.3531073446327683e-05,
      "loss": 2.8511,
      "step": 1874
    },
    {
      "epoch": 1.5889830508474576,
      "grad_norm": 1.382293939590454,
      "learning_rate": 2.3516949152542376e-05,
      "loss": 2.5916,
      "step": 1875
    },
    {
      "epoch": 1.5898305084745763,
      "grad_norm": 1.1161446571350098,
      "learning_rate": 2.3502824858757063e-05,
      "loss": 2.6336,
      "step": 1876
    },
    {
      "epoch": 1.590677966101695,
      "grad_norm": 1.3635808229446411,
      "learning_rate": 2.3488700564971753e-05,
      "loss": 2.5219,
      "step": 1877
    },
    {
      "epoch": 1.5915254237288137,
      "grad_norm": 1.280559778213501,
      "learning_rate": 2.347457627118644e-05,
      "loss": 2.6639,
      "step": 1878
    },
    {
      "epoch": 1.5923728813559324,
      "grad_norm": 1.0561965703964233,
      "learning_rate": 2.3460451977401133e-05,
      "loss": 2.7664,
      "step": 1879
    },
    {
      "epoch": 1.5932203389830508,
      "grad_norm": 0.8967818021774292,
      "learning_rate": 2.344632768361582e-05,
      "loss": 2.9752,
      "step": 1880
    },
    {
      "epoch": 1.5940677966101695,
      "grad_norm": 1.0559184551239014,
      "learning_rate": 2.343220338983051e-05,
      "loss": 2.9621,
      "step": 1881
    },
    {
      "epoch": 1.594915254237288,
      "grad_norm": 1.0838136672973633,
      "learning_rate": 2.34180790960452e-05,
      "loss": 2.8132,
      "step": 1882
    },
    {
      "epoch": 1.5957627118644067,
      "grad_norm": 1.212471604347229,
      "learning_rate": 2.340395480225989e-05,
      "loss": 2.8631,
      "step": 1883
    },
    {
      "epoch": 1.5966101694915253,
      "grad_norm": 0.9554041028022766,
      "learning_rate": 2.338983050847458e-05,
      "loss": 2.7652,
      "step": 1884
    },
    {
      "epoch": 1.597457627118644,
      "grad_norm": 0.9795320630073547,
      "learning_rate": 2.3375706214689266e-05,
      "loss": 2.8476,
      "step": 1885
    },
    {
      "epoch": 1.5983050847457627,
      "grad_norm": 0.8778656125068665,
      "learning_rate": 2.3361581920903956e-05,
      "loss": 2.8676,
      "step": 1886
    },
    {
      "epoch": 1.5991525423728814,
      "grad_norm": 0.999576985836029,
      "learning_rate": 2.3347457627118646e-05,
      "loss": 2.6703,
      "step": 1887
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9874566197395325,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 2.8561,
      "step": 1888
    },
    {
      "epoch": 1.6008474576271188,
      "grad_norm": 1.0946693420410156,
      "learning_rate": 2.3319209039548022e-05,
      "loss": 2.5874,
      "step": 1889
    },
    {
      "epoch": 1.6016949152542372,
      "grad_norm": 1.1346005201339722,
      "learning_rate": 2.3305084745762712e-05,
      "loss": 2.7149,
      "step": 1890
    },
    {
      "epoch": 1.602542372881356,
      "grad_norm": 1.1960369348526,
      "learning_rate": 2.3290960451977402e-05,
      "loss": 2.5591,
      "step": 1891
    },
    {
      "epoch": 1.6033898305084746,
      "grad_norm": 1.0531506538391113,
      "learning_rate": 2.3276836158192092e-05,
      "loss": 2.7084,
      "step": 1892
    },
    {
      "epoch": 1.604237288135593,
      "grad_norm": 1.2548158168792725,
      "learning_rate": 2.326271186440678e-05,
      "loss": 2.6188,
      "step": 1893
    },
    {
      "epoch": 1.6050847457627118,
      "grad_norm": 0.9160987734794617,
      "learning_rate": 2.324858757062147e-05,
      "loss": 2.9216,
      "step": 1894
    },
    {
      "epoch": 1.6059322033898304,
      "grad_norm": 1.2223246097564697,
      "learning_rate": 2.323446327683616e-05,
      "loss": 2.5757,
      "step": 1895
    },
    {
      "epoch": 1.6067796610169491,
      "grad_norm": 1.0778521299362183,
      "learning_rate": 2.322033898305085e-05,
      "loss": 2.6405,
      "step": 1896
    },
    {
      "epoch": 1.6076271186440678,
      "grad_norm": 0.9888900518417358,
      "learning_rate": 2.320621468926554e-05,
      "loss": 2.8904,
      "step": 1897
    },
    {
      "epoch": 1.6084745762711865,
      "grad_norm": 1.130634069442749,
      "learning_rate": 2.3192090395480226e-05,
      "loss": 2.8235,
      "step": 1898
    },
    {
      "epoch": 1.6093220338983052,
      "grad_norm": 1.1311981678009033,
      "learning_rate": 2.317796610169492e-05,
      "loss": 2.7509,
      "step": 1899
    },
    {
      "epoch": 1.6101694915254239,
      "grad_norm": 1.1725093126296997,
      "learning_rate": 2.3163841807909606e-05,
      "loss": 2.6908,
      "step": 1900
    },
    {
      "epoch": 1.6110169491525423,
      "grad_norm": 1.0086802244186401,
      "learning_rate": 2.3149717514124296e-05,
      "loss": 2.8708,
      "step": 1901
    },
    {
      "epoch": 1.611864406779661,
      "grad_norm": 1.1591293811798096,
      "learning_rate": 2.3135593220338982e-05,
      "loss": 2.8952,
      "step": 1902
    },
    {
      "epoch": 1.6127118644067797,
      "grad_norm": 0.9742602705955505,
      "learning_rate": 2.3121468926553676e-05,
      "loss": 2.725,
      "step": 1903
    },
    {
      "epoch": 1.6135593220338982,
      "grad_norm": 1.1805888414382935,
      "learning_rate": 2.3107344632768362e-05,
      "loss": 2.953,
      "step": 1904
    },
    {
      "epoch": 1.6144067796610169,
      "grad_norm": 0.9819029569625854,
      "learning_rate": 2.3093220338983052e-05,
      "loss": 2.7734,
      "step": 1905
    },
    {
      "epoch": 1.6152542372881356,
      "grad_norm": 1.1077427864074707,
      "learning_rate": 2.307909604519774e-05,
      "loss": 2.76,
      "step": 1906
    },
    {
      "epoch": 1.6161016949152542,
      "grad_norm": 1.243615984916687,
      "learning_rate": 2.3064971751412432e-05,
      "loss": 2.6716,
      "step": 1907
    },
    {
      "epoch": 1.616949152542373,
      "grad_norm": 1.054997205734253,
      "learning_rate": 2.305084745762712e-05,
      "loss": 2.8281,
      "step": 1908
    },
    {
      "epoch": 1.6177966101694916,
      "grad_norm": 1.0687192678451538,
      "learning_rate": 2.303672316384181e-05,
      "loss": 2.8526,
      "step": 1909
    },
    {
      "epoch": 1.6186440677966103,
      "grad_norm": 1.1847962141036987,
      "learning_rate": 2.30225988700565e-05,
      "loss": 2.6144,
      "step": 1910
    },
    {
      "epoch": 1.6194915254237288,
      "grad_norm": 0.9783480763435364,
      "learning_rate": 2.300847457627119e-05,
      "loss": 2.7837,
      "step": 1911
    },
    {
      "epoch": 1.6203389830508474,
      "grad_norm": 1.0385940074920654,
      "learning_rate": 2.2994350282485875e-05,
      "loss": 2.781,
      "step": 1912
    },
    {
      "epoch": 1.6211864406779661,
      "grad_norm": 1.1184855699539185,
      "learning_rate": 2.2980225988700565e-05,
      "loss": 2.6853,
      "step": 1913
    },
    {
      "epoch": 1.6220338983050846,
      "grad_norm": 0.9464384913444519,
      "learning_rate": 2.2966101694915255e-05,
      "loss": 2.8047,
      "step": 1914
    },
    {
      "epoch": 1.6228813559322033,
      "grad_norm": 1.2419618368148804,
      "learning_rate": 2.2951977401129945e-05,
      "loss": 2.7639,
      "step": 1915
    },
    {
      "epoch": 1.623728813559322,
      "grad_norm": 1.2323758602142334,
      "learning_rate": 2.2937853107344635e-05,
      "loss": 2.776,
      "step": 1916
    },
    {
      "epoch": 1.6245762711864407,
      "grad_norm": 1.1942468881607056,
      "learning_rate": 2.2923728813559322e-05,
      "loss": 2.7054,
      "step": 1917
    },
    {
      "epoch": 1.6254237288135593,
      "grad_norm": 1.1644326448440552,
      "learning_rate": 2.2909604519774012e-05,
      "loss": 2.6464,
      "step": 1918
    },
    {
      "epoch": 1.626271186440678,
      "grad_norm": 1.204388976097107,
      "learning_rate": 2.2895480225988702e-05,
      "loss": 2.6597,
      "step": 1919
    },
    {
      "epoch": 1.6271186440677967,
      "grad_norm": 1.2430897951126099,
      "learning_rate": 2.2881355932203392e-05,
      "loss": 2.7475,
      "step": 1920
    },
    {
      "epoch": 1.6279661016949154,
      "grad_norm": 1.0028111934661865,
      "learning_rate": 2.286723163841808e-05,
      "loss": 2.9939,
      "step": 1921
    },
    {
      "epoch": 1.6288135593220339,
      "grad_norm": 1.1967658996582031,
      "learning_rate": 2.285310734463277e-05,
      "loss": 2.7748,
      "step": 1922
    },
    {
      "epoch": 1.6296610169491526,
      "grad_norm": 1.1778464317321777,
      "learning_rate": 2.283898305084746e-05,
      "loss": 2.7629,
      "step": 1923
    },
    {
      "epoch": 1.6305084745762712,
      "grad_norm": 1.6117842197418213,
      "learning_rate": 2.282485875706215e-05,
      "loss": 2.4678,
      "step": 1924
    },
    {
      "epoch": 1.6313559322033897,
      "grad_norm": 0.9721297025680542,
      "learning_rate": 2.2810734463276835e-05,
      "loss": 2.8654,
      "step": 1925
    },
    {
      "epoch": 1.6322033898305084,
      "grad_norm": 1.3772358894348145,
      "learning_rate": 2.2796610169491525e-05,
      "loss": 2.7953,
      "step": 1926
    },
    {
      "epoch": 1.633050847457627,
      "grad_norm": 1.2268047332763672,
      "learning_rate": 2.2782485875706215e-05,
      "loss": 2.5029,
      "step": 1927
    },
    {
      "epoch": 1.6338983050847458,
      "grad_norm": 1.0063567161560059,
      "learning_rate": 2.2768361581920905e-05,
      "loss": 2.7561,
      "step": 1928
    },
    {
      "epoch": 1.6347457627118644,
      "grad_norm": 1.155717372894287,
      "learning_rate": 2.2754237288135595e-05,
      "loss": 2.7237,
      "step": 1929
    },
    {
      "epoch": 1.6355932203389831,
      "grad_norm": 1.1854345798492432,
      "learning_rate": 2.274011299435028e-05,
      "loss": 2.6825,
      "step": 1930
    },
    {
      "epoch": 1.6364406779661018,
      "grad_norm": 1.2681466341018677,
      "learning_rate": 2.2725988700564975e-05,
      "loss": 2.5796,
      "step": 1931
    },
    {
      "epoch": 1.6372881355932203,
      "grad_norm": 1.0469571352005005,
      "learning_rate": 2.271186440677966e-05,
      "loss": 2.7957,
      "step": 1932
    },
    {
      "epoch": 1.638135593220339,
      "grad_norm": 1.0084184408187866,
      "learning_rate": 2.269774011299435e-05,
      "loss": 2.8405,
      "step": 1933
    },
    {
      "epoch": 1.6389830508474577,
      "grad_norm": 1.0356253385543823,
      "learning_rate": 2.268361581920904e-05,
      "loss": 2.694,
      "step": 1934
    },
    {
      "epoch": 1.6398305084745761,
      "grad_norm": 1.150457739830017,
      "learning_rate": 2.266949152542373e-05,
      "loss": 2.7612,
      "step": 1935
    },
    {
      "epoch": 1.6406779661016948,
      "grad_norm": 1.0183244943618774,
      "learning_rate": 2.2655367231638418e-05,
      "loss": 2.745,
      "step": 1936
    },
    {
      "epoch": 1.6415254237288135,
      "grad_norm": 1.1340312957763672,
      "learning_rate": 2.2641242937853108e-05,
      "loss": 2.6348,
      "step": 1937
    },
    {
      "epoch": 1.6423728813559322,
      "grad_norm": 1.0995330810546875,
      "learning_rate": 2.2627118644067798e-05,
      "loss": 2.8379,
      "step": 1938
    },
    {
      "epoch": 1.6432203389830509,
      "grad_norm": 0.9740652441978455,
      "learning_rate": 2.2612994350282488e-05,
      "loss": 2.7789,
      "step": 1939
    },
    {
      "epoch": 1.6440677966101696,
      "grad_norm": 1.1527608633041382,
      "learning_rate": 2.2598870056497175e-05,
      "loss": 2.8493,
      "step": 1940
    },
    {
      "epoch": 1.6449152542372882,
      "grad_norm": 1.5391933917999268,
      "learning_rate": 2.2584745762711865e-05,
      "loss": 2.3918,
      "step": 1941
    },
    {
      "epoch": 1.645762711864407,
      "grad_norm": 1.2131741046905518,
      "learning_rate": 2.2570621468926555e-05,
      "loss": 2.6372,
      "step": 1942
    },
    {
      "epoch": 1.6466101694915254,
      "grad_norm": 0.8697400689125061,
      "learning_rate": 2.2556497175141245e-05,
      "loss": 2.9724,
      "step": 1943
    },
    {
      "epoch": 1.647457627118644,
      "grad_norm": 1.125433325767517,
      "learning_rate": 2.2542372881355935e-05,
      "loss": 2.7292,
      "step": 1944
    },
    {
      "epoch": 1.6483050847457628,
      "grad_norm": 1.0400315523147583,
      "learning_rate": 2.252824858757062e-05,
      "loss": 2.8846,
      "step": 1945
    },
    {
      "epoch": 1.6491525423728812,
      "grad_norm": 1.170027732849121,
      "learning_rate": 2.251412429378531e-05,
      "loss": 2.6606,
      "step": 1946
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.337228536605835,
      "learning_rate": 2.25e-05,
      "loss": 2.6804,
      "step": 1947
    },
    {
      "epoch": 1.6508474576271186,
      "grad_norm": 1.1318726539611816,
      "learning_rate": 2.248587570621469e-05,
      "loss": 2.7268,
      "step": 1948
    },
    {
      "epoch": 1.6516949152542373,
      "grad_norm": 0.984640896320343,
      "learning_rate": 2.2471751412429378e-05,
      "loss": 2.6941,
      "step": 1949
    },
    {
      "epoch": 1.652542372881356,
      "grad_norm": 0.9486579895019531,
      "learning_rate": 2.245762711864407e-05,
      "loss": 2.7897,
      "step": 1950
    },
    {
      "epoch": 1.6533898305084747,
      "grad_norm": 1.0196901559829712,
      "learning_rate": 2.2443502824858758e-05,
      "loss": 2.7002,
      "step": 1951
    },
    {
      "epoch": 1.6542372881355933,
      "grad_norm": 1.0792779922485352,
      "learning_rate": 2.2429378531073448e-05,
      "loss": 2.647,
      "step": 1952
    },
    {
      "epoch": 1.655084745762712,
      "grad_norm": 1.0463929176330566,
      "learning_rate": 2.2415254237288135e-05,
      "loss": 2.7542,
      "step": 1953
    },
    {
      "epoch": 1.6559322033898305,
      "grad_norm": 1.1446179151535034,
      "learning_rate": 2.2401129943502828e-05,
      "loss": 2.812,
      "step": 1954
    },
    {
      "epoch": 1.6567796610169492,
      "grad_norm": 1.073575735092163,
      "learning_rate": 2.2387005649717515e-05,
      "loss": 2.7027,
      "step": 1955
    },
    {
      "epoch": 1.6576271186440676,
      "grad_norm": 1.2444988489151,
      "learning_rate": 2.2372881355932205e-05,
      "loss": 2.8578,
      "step": 1956
    },
    {
      "epoch": 1.6584745762711863,
      "grad_norm": 1.077351450920105,
      "learning_rate": 2.235875706214689e-05,
      "loss": 2.7909,
      "step": 1957
    },
    {
      "epoch": 1.659322033898305,
      "grad_norm": 1.1523077487945557,
      "learning_rate": 2.2344632768361585e-05,
      "loss": 2.7268,
      "step": 1958
    },
    {
      "epoch": 1.6601694915254237,
      "grad_norm": 0.9423948526382446,
      "learning_rate": 2.233050847457627e-05,
      "loss": 2.6836,
      "step": 1959
    },
    {
      "epoch": 1.6610169491525424,
      "grad_norm": 1.1524324417114258,
      "learning_rate": 2.231638418079096e-05,
      "loss": 2.6782,
      "step": 1960
    },
    {
      "epoch": 1.661864406779661,
      "grad_norm": 1.0088781118392944,
      "learning_rate": 2.230225988700565e-05,
      "loss": 2.7087,
      "step": 1961
    },
    {
      "epoch": 1.6627118644067798,
      "grad_norm": 1.4018473625183105,
      "learning_rate": 2.228813559322034e-05,
      "loss": 2.567,
      "step": 1962
    },
    {
      "epoch": 1.6635593220338984,
      "grad_norm": 0.9550151228904724,
      "learning_rate": 2.227401129943503e-05,
      "loss": 2.8346,
      "step": 1963
    },
    {
      "epoch": 1.664406779661017,
      "grad_norm": 1.198732614517212,
      "learning_rate": 2.2259887005649718e-05,
      "loss": 2.6654,
      "step": 1964
    },
    {
      "epoch": 1.6652542372881356,
      "grad_norm": 1.1893459558486938,
      "learning_rate": 2.2245762711864408e-05,
      "loss": 2.7053,
      "step": 1965
    },
    {
      "epoch": 1.6661016949152543,
      "grad_norm": 1.078377604484558,
      "learning_rate": 2.2231638418079098e-05,
      "loss": 2.7485,
      "step": 1966
    },
    {
      "epoch": 1.6669491525423727,
      "grad_norm": 1.2355225086212158,
      "learning_rate": 2.2217514124293788e-05,
      "loss": 2.6898,
      "step": 1967
    },
    {
      "epoch": 1.6677966101694914,
      "grad_norm": 1.1316275596618652,
      "learning_rate": 2.2203389830508474e-05,
      "loss": 2.737,
      "step": 1968
    },
    {
      "epoch": 1.6686440677966101,
      "grad_norm": 1.065366506576538,
      "learning_rate": 2.2189265536723164e-05,
      "loss": 2.7938,
      "step": 1969
    },
    {
      "epoch": 1.6694915254237288,
      "grad_norm": 1.3895978927612305,
      "learning_rate": 2.2175141242937854e-05,
      "loss": 2.5969,
      "step": 1970
    },
    {
      "epoch": 1.6703389830508475,
      "grad_norm": 1.1037791967391968,
      "learning_rate": 2.2161016949152544e-05,
      "loss": 2.8264,
      "step": 1971
    },
    {
      "epoch": 1.6711864406779662,
      "grad_norm": 1.2196025848388672,
      "learning_rate": 2.214689265536723e-05,
      "loss": 2.8404,
      "step": 1972
    },
    {
      "epoch": 1.6720338983050849,
      "grad_norm": 1.1980524063110352,
      "learning_rate": 2.213276836158192e-05,
      "loss": 2.7032,
      "step": 1973
    },
    {
      "epoch": 1.6728813559322036,
      "grad_norm": 1.2860914468765259,
      "learning_rate": 2.211864406779661e-05,
      "loss": 2.7491,
      "step": 1974
    },
    {
      "epoch": 1.673728813559322,
      "grad_norm": 1.0300694704055786,
      "learning_rate": 2.21045197740113e-05,
      "loss": 2.679,
      "step": 1975
    },
    {
      "epoch": 1.6745762711864407,
      "grad_norm": 1.0046870708465576,
      "learning_rate": 2.209039548022599e-05,
      "loss": 2.6776,
      "step": 1976
    },
    {
      "epoch": 1.6754237288135592,
      "grad_norm": 1.3938623666763306,
      "learning_rate": 2.2076271186440678e-05,
      "loss": 2.6404,
      "step": 1977
    },
    {
      "epoch": 1.6762711864406779,
      "grad_norm": 1.0191386938095093,
      "learning_rate": 2.206214689265537e-05,
      "loss": 2.8735,
      "step": 1978
    },
    {
      "epoch": 1.6771186440677965,
      "grad_norm": 1.0004791021347046,
      "learning_rate": 2.2048022598870058e-05,
      "loss": 2.7898,
      "step": 1979
    },
    {
      "epoch": 1.6779661016949152,
      "grad_norm": 1.1002367734909058,
      "learning_rate": 2.2033898305084748e-05,
      "loss": 2.6758,
      "step": 1980
    },
    {
      "epoch": 1.678813559322034,
      "grad_norm": 1.1558040380477905,
      "learning_rate": 2.2019774011299434e-05,
      "loss": 2.689,
      "step": 1981
    },
    {
      "epoch": 1.6796610169491526,
      "grad_norm": 1.2505487203598022,
      "learning_rate": 2.2005649717514127e-05,
      "loss": 2.6016,
      "step": 1982
    },
    {
      "epoch": 1.6805084745762713,
      "grad_norm": 1.2562954425811768,
      "learning_rate": 2.1991525423728814e-05,
      "loss": 2.6221,
      "step": 1983
    },
    {
      "epoch": 1.68135593220339,
      "grad_norm": 1.0955479145050049,
      "learning_rate": 2.1977401129943504e-05,
      "loss": 2.7393,
      "step": 1984
    },
    {
      "epoch": 1.6822033898305084,
      "grad_norm": 1.326519250869751,
      "learning_rate": 2.196327683615819e-05,
      "loss": 2.6772,
      "step": 1985
    },
    {
      "epoch": 1.6830508474576271,
      "grad_norm": 1.0585700273513794,
      "learning_rate": 2.1949152542372884e-05,
      "loss": 2.7725,
      "step": 1986
    },
    {
      "epoch": 1.6838983050847458,
      "grad_norm": 0.9212808609008789,
      "learning_rate": 2.193502824858757e-05,
      "loss": 2.7984,
      "step": 1987
    },
    {
      "epoch": 1.6847457627118643,
      "grad_norm": 1.2387858629226685,
      "learning_rate": 2.192090395480226e-05,
      "loss": 2.7233,
      "step": 1988
    },
    {
      "epoch": 1.685593220338983,
      "grad_norm": 1.1236997842788696,
      "learning_rate": 2.190677966101695e-05,
      "loss": 2.5904,
      "step": 1989
    },
    {
      "epoch": 1.6864406779661016,
      "grad_norm": 1.1906774044036865,
      "learning_rate": 2.189265536723164e-05,
      "loss": 2.7381,
      "step": 1990
    },
    {
      "epoch": 1.6872881355932203,
      "grad_norm": 1.161232352256775,
      "learning_rate": 2.1878531073446327e-05,
      "loss": 2.5432,
      "step": 1991
    },
    {
      "epoch": 1.688135593220339,
      "grad_norm": 0.8911912441253662,
      "learning_rate": 2.1864406779661017e-05,
      "loss": 2.8894,
      "step": 1992
    },
    {
      "epoch": 1.6889830508474577,
      "grad_norm": 0.9659197330474854,
      "learning_rate": 2.1850282485875707e-05,
      "loss": 2.7261,
      "step": 1993
    },
    {
      "epoch": 1.6898305084745764,
      "grad_norm": 1.0060009956359863,
      "learning_rate": 2.1836158192090397e-05,
      "loss": 2.7793,
      "step": 1994
    },
    {
      "epoch": 1.690677966101695,
      "grad_norm": 1.2250174283981323,
      "learning_rate": 2.1822033898305087e-05,
      "loss": 2.6318,
      "step": 1995
    },
    {
      "epoch": 1.6915254237288135,
      "grad_norm": 1.129069209098816,
      "learning_rate": 2.1807909604519774e-05,
      "loss": 2.7596,
      "step": 1996
    },
    {
      "epoch": 1.6923728813559322,
      "grad_norm": 1.1996171474456787,
      "learning_rate": 2.1793785310734464e-05,
      "loss": 2.6167,
      "step": 1997
    },
    {
      "epoch": 1.6932203389830507,
      "grad_norm": 1.1588618755340576,
      "learning_rate": 2.1779661016949154e-05,
      "loss": 2.7102,
      "step": 1998
    },
    {
      "epoch": 1.6940677966101694,
      "grad_norm": 1.0075221061706543,
      "learning_rate": 2.1765536723163844e-05,
      "loss": 2.7758,
      "step": 1999
    },
    {
      "epoch": 1.694915254237288,
      "grad_norm": 1.0586878061294556,
      "learning_rate": 2.175141242937853e-05,
      "loss": 2.8316,
      "step": 2000
    },
    {
      "epoch": 1.6957627118644067,
      "grad_norm": 1.0339900255203247,
      "learning_rate": 2.173728813559322e-05,
      "loss": 2.6571,
      "step": 2001
    },
    {
      "epoch": 1.6966101694915254,
      "grad_norm": 0.967481791973114,
      "learning_rate": 2.172316384180791e-05,
      "loss": 2.869,
      "step": 2002
    },
    {
      "epoch": 1.6974576271186441,
      "grad_norm": 0.8604980707168579,
      "learning_rate": 2.17090395480226e-05,
      "loss": 2.8952,
      "step": 2003
    },
    {
      "epoch": 1.6983050847457628,
      "grad_norm": 1.0698426961898804,
      "learning_rate": 2.1694915254237287e-05,
      "loss": 2.6932,
      "step": 2004
    },
    {
      "epoch": 1.6991525423728815,
      "grad_norm": 1.254738211631775,
      "learning_rate": 2.1680790960451977e-05,
      "loss": 2.8027,
      "step": 2005
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1405632495880127,
      "learning_rate": 2.1666666666666667e-05,
      "loss": 2.6379,
      "step": 2006
    },
    {
      "epoch": 1.7008474576271186,
      "grad_norm": 1.1888066530227661,
      "learning_rate": 2.1652542372881357e-05,
      "loss": 2.5186,
      "step": 2007
    },
    {
      "epoch": 1.7016949152542373,
      "grad_norm": 1.0663483142852783,
      "learning_rate": 2.1638418079096047e-05,
      "loss": 2.6905,
      "step": 2008
    },
    {
      "epoch": 1.7025423728813558,
      "grad_norm": 1.0556362867355347,
      "learning_rate": 2.1624293785310734e-05,
      "loss": 2.8582,
      "step": 2009
    },
    {
      "epoch": 1.7033898305084745,
      "grad_norm": 0.9792917966842651,
      "learning_rate": 2.1610169491525427e-05,
      "loss": 2.7174,
      "step": 2010
    },
    {
      "epoch": 1.7042372881355932,
      "grad_norm": 0.9167950749397278,
      "learning_rate": 2.1596045197740114e-05,
      "loss": 2.7645,
      "step": 2011
    },
    {
      "epoch": 1.7050847457627119,
      "grad_norm": 1.2853854894638062,
      "learning_rate": 2.1581920903954804e-05,
      "loss": 2.6393,
      "step": 2012
    },
    {
      "epoch": 1.7059322033898305,
      "grad_norm": 1.1802425384521484,
      "learning_rate": 2.1567796610169494e-05,
      "loss": 2.8005,
      "step": 2013
    },
    {
      "epoch": 1.7067796610169492,
      "grad_norm": 1.1032609939575195,
      "learning_rate": 2.1553672316384184e-05,
      "loss": 2.7769,
      "step": 2014
    },
    {
      "epoch": 1.707627118644068,
      "grad_norm": 1.0899887084960938,
      "learning_rate": 2.153954802259887e-05,
      "loss": 2.7425,
      "step": 2015
    },
    {
      "epoch": 1.7084745762711866,
      "grad_norm": 1.0336703062057495,
      "learning_rate": 2.152542372881356e-05,
      "loss": 2.9216,
      "step": 2016
    },
    {
      "epoch": 1.709322033898305,
      "grad_norm": 1.4915704727172852,
      "learning_rate": 2.151129943502825e-05,
      "loss": 2.5933,
      "step": 2017
    },
    {
      "epoch": 1.7101694915254237,
      "grad_norm": 1.213881015777588,
      "learning_rate": 2.149717514124294e-05,
      "loss": 2.6736,
      "step": 2018
    },
    {
      "epoch": 1.7110169491525422,
      "grad_norm": 1.058539628982544,
      "learning_rate": 2.1483050847457627e-05,
      "loss": 2.9701,
      "step": 2019
    },
    {
      "epoch": 1.711864406779661,
      "grad_norm": 0.9809156060218811,
      "learning_rate": 2.1468926553672317e-05,
      "loss": 2.7857,
      "step": 2020
    },
    {
      "epoch": 1.7127118644067796,
      "grad_norm": 0.9568430781364441,
      "learning_rate": 2.1454802259887007e-05,
      "loss": 2.7869,
      "step": 2021
    },
    {
      "epoch": 1.7135593220338983,
      "grad_norm": 1.3682050704956055,
      "learning_rate": 2.1440677966101697e-05,
      "loss": 2.6752,
      "step": 2022
    },
    {
      "epoch": 1.714406779661017,
      "grad_norm": 0.9975570440292358,
      "learning_rate": 2.1426553672316387e-05,
      "loss": 2.8756,
      "step": 2023
    },
    {
      "epoch": 1.7152542372881356,
      "grad_norm": 1.389093279838562,
      "learning_rate": 2.1412429378531073e-05,
      "loss": 2.7175,
      "step": 2024
    },
    {
      "epoch": 1.7161016949152543,
      "grad_norm": 1.0701448917388916,
      "learning_rate": 2.1398305084745763e-05,
      "loss": 2.7378,
      "step": 2025
    },
    {
      "epoch": 1.716949152542373,
      "grad_norm": 0.9431941509246826,
      "learning_rate": 2.1384180790960453e-05,
      "loss": 2.8699,
      "step": 2026
    },
    {
      "epoch": 1.7177966101694915,
      "grad_norm": 1.179017186164856,
      "learning_rate": 2.1370056497175143e-05,
      "loss": 2.6515,
      "step": 2027
    },
    {
      "epoch": 1.7186440677966102,
      "grad_norm": 1.029389500617981,
      "learning_rate": 2.135593220338983e-05,
      "loss": 2.6199,
      "step": 2028
    },
    {
      "epoch": 1.7194915254237289,
      "grad_norm": 1.1984156370162964,
      "learning_rate": 2.1341807909604523e-05,
      "loss": 2.7407,
      "step": 2029
    },
    {
      "epoch": 1.7203389830508473,
      "grad_norm": 1.0471163988113403,
      "learning_rate": 2.132768361581921e-05,
      "loss": 2.9038,
      "step": 2030
    },
    {
      "epoch": 1.721186440677966,
      "grad_norm": 1.1695588827133179,
      "learning_rate": 2.13135593220339e-05,
      "loss": 2.6905,
      "step": 2031
    },
    {
      "epoch": 1.7220338983050847,
      "grad_norm": 1.046454668045044,
      "learning_rate": 2.1299435028248587e-05,
      "loss": 2.7893,
      "step": 2032
    },
    {
      "epoch": 1.7228813559322034,
      "grad_norm": 1.100285530090332,
      "learning_rate": 2.128531073446328e-05,
      "loss": 2.7836,
      "step": 2033
    },
    {
      "epoch": 1.723728813559322,
      "grad_norm": 1.6512371301651,
      "learning_rate": 2.1271186440677967e-05,
      "loss": 2.5651,
      "step": 2034
    },
    {
      "epoch": 1.7245762711864407,
      "grad_norm": 1.2220083475112915,
      "learning_rate": 2.1257062146892657e-05,
      "loss": 2.7178,
      "step": 2035
    },
    {
      "epoch": 1.7254237288135594,
      "grad_norm": 0.9805908799171448,
      "learning_rate": 2.1242937853107343e-05,
      "loss": 2.7406,
      "step": 2036
    },
    {
      "epoch": 1.7262711864406781,
      "grad_norm": 0.9179597496986389,
      "learning_rate": 2.1228813559322037e-05,
      "loss": 2.7452,
      "step": 2037
    },
    {
      "epoch": 1.7271186440677966,
      "grad_norm": 0.8249093890190125,
      "learning_rate": 2.1214689265536723e-05,
      "loss": 2.9393,
      "step": 2038
    },
    {
      "epoch": 1.7279661016949153,
      "grad_norm": 1.2761414051055908,
      "learning_rate": 2.1200564971751413e-05,
      "loss": 2.6703,
      "step": 2039
    },
    {
      "epoch": 1.7288135593220337,
      "grad_norm": 1.178770899772644,
      "learning_rate": 2.1186440677966103e-05,
      "loss": 2.6835,
      "step": 2040
    },
    {
      "epoch": 1.7296610169491524,
      "grad_norm": 1.1519194841384888,
      "learning_rate": 2.1172316384180793e-05,
      "loss": 2.8016,
      "step": 2041
    },
    {
      "epoch": 1.730508474576271,
      "grad_norm": 1.166398525238037,
      "learning_rate": 2.1158192090395483e-05,
      "loss": 2.7879,
      "step": 2042
    },
    {
      "epoch": 1.7313559322033898,
      "grad_norm": 0.9436035752296448,
      "learning_rate": 2.114406779661017e-05,
      "loss": 2.7919,
      "step": 2043
    },
    {
      "epoch": 1.7322033898305085,
      "grad_norm": 1.119163155555725,
      "learning_rate": 2.112994350282486e-05,
      "loss": 2.6872,
      "step": 2044
    },
    {
      "epoch": 1.7330508474576272,
      "grad_norm": 1.446716070175171,
      "learning_rate": 2.111581920903955e-05,
      "loss": 2.5478,
      "step": 2045
    },
    {
      "epoch": 1.7338983050847459,
      "grad_norm": 1.1846516132354736,
      "learning_rate": 2.110169491525424e-05,
      "loss": 2.768,
      "step": 2046
    },
    {
      "epoch": 1.7347457627118645,
      "grad_norm": 0.8903865218162537,
      "learning_rate": 2.1087570621468926e-05,
      "loss": 2.803,
      "step": 2047
    },
    {
      "epoch": 1.735593220338983,
      "grad_norm": 1.2745734453201294,
      "learning_rate": 2.1073446327683616e-05,
      "loss": 2.6938,
      "step": 2048
    },
    {
      "epoch": 1.7364406779661017,
      "grad_norm": 1.1986397504806519,
      "learning_rate": 2.1059322033898306e-05,
      "loss": 2.7291,
      "step": 2049
    },
    {
      "epoch": 1.7372881355932204,
      "grad_norm": 1.3172969818115234,
      "learning_rate": 2.1045197740112996e-05,
      "loss": 2.5729,
      "step": 2050
    },
    {
      "epoch": 1.7381355932203388,
      "grad_norm": 1.0908544063568115,
      "learning_rate": 2.1031073446327683e-05,
      "loss": 2.8598,
      "step": 2051
    },
    {
      "epoch": 1.7389830508474575,
      "grad_norm": 1.2555841207504272,
      "learning_rate": 2.1016949152542373e-05,
      "loss": 2.6042,
      "step": 2052
    },
    {
      "epoch": 1.7398305084745762,
      "grad_norm": 1.2310426235198975,
      "learning_rate": 2.1002824858757063e-05,
      "loss": 2.5916,
      "step": 2053
    },
    {
      "epoch": 1.740677966101695,
      "grad_norm": 0.9163360595703125,
      "learning_rate": 2.0988700564971753e-05,
      "loss": 2.8336,
      "step": 2054
    },
    {
      "epoch": 1.7415254237288136,
      "grad_norm": 1.045538306236267,
      "learning_rate": 2.0974576271186443e-05,
      "loss": 2.8838,
      "step": 2055
    },
    {
      "epoch": 1.7423728813559323,
      "grad_norm": 1.1291589736938477,
      "learning_rate": 2.096045197740113e-05,
      "loss": 2.7363,
      "step": 2056
    },
    {
      "epoch": 1.743220338983051,
      "grad_norm": 1.0940430164337158,
      "learning_rate": 2.0946327683615823e-05,
      "loss": 2.7447,
      "step": 2057
    },
    {
      "epoch": 1.7440677966101696,
      "grad_norm": 1.12651789188385,
      "learning_rate": 2.093220338983051e-05,
      "loss": 2.7276,
      "step": 2058
    },
    {
      "epoch": 1.744915254237288,
      "grad_norm": 1.0112786293029785,
      "learning_rate": 2.09180790960452e-05,
      "loss": 2.8141,
      "step": 2059
    },
    {
      "epoch": 1.7457627118644068,
      "grad_norm": 1.320303201675415,
      "learning_rate": 2.0903954802259886e-05,
      "loss": 2.5977,
      "step": 2060
    },
    {
      "epoch": 1.7466101694915255,
      "grad_norm": 1.361250638961792,
      "learning_rate": 2.088983050847458e-05,
      "loss": 2.6541,
      "step": 2061
    },
    {
      "epoch": 1.747457627118644,
      "grad_norm": 1.0819406509399414,
      "learning_rate": 2.0875706214689266e-05,
      "loss": 2.83,
      "step": 2062
    },
    {
      "epoch": 1.7483050847457626,
      "grad_norm": 1.2149043083190918,
      "learning_rate": 2.0861581920903956e-05,
      "loss": 2.6023,
      "step": 2063
    },
    {
      "epoch": 1.7491525423728813,
      "grad_norm": 0.9299839735031128,
      "learning_rate": 2.0847457627118643e-05,
      "loss": 2.8601,
      "step": 2064
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8682264685630798,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 2.8056,
      "step": 2065
    },
    {
      "epoch": 1.7508474576271187,
      "grad_norm": 1.0110892057418823,
      "learning_rate": 2.0819209039548023e-05,
      "loss": 2.7593,
      "step": 2066
    },
    {
      "epoch": 1.7516949152542374,
      "grad_norm": 1.448298454284668,
      "learning_rate": 2.0805084745762713e-05,
      "loss": 2.341,
      "step": 2067
    },
    {
      "epoch": 1.752542372881356,
      "grad_norm": 0.9779382944107056,
      "learning_rate": 2.0790960451977403e-05,
      "loss": 2.7048,
      "step": 2068
    },
    {
      "epoch": 1.7533898305084745,
      "grad_norm": 1.1361010074615479,
      "learning_rate": 2.0776836158192093e-05,
      "loss": 2.6464,
      "step": 2069
    },
    {
      "epoch": 1.7542372881355932,
      "grad_norm": 1.1303348541259766,
      "learning_rate": 2.076271186440678e-05,
      "loss": 2.7549,
      "step": 2070
    },
    {
      "epoch": 1.755084745762712,
      "grad_norm": 1.1424024105072021,
      "learning_rate": 2.074858757062147e-05,
      "loss": 2.7068,
      "step": 2071
    },
    {
      "epoch": 1.7559322033898304,
      "grad_norm": 0.9543996453285217,
      "learning_rate": 2.073446327683616e-05,
      "loss": 2.7092,
      "step": 2072
    },
    {
      "epoch": 1.756779661016949,
      "grad_norm": 1.3549798727035522,
      "learning_rate": 2.072033898305085e-05,
      "loss": 2.6977,
      "step": 2073
    },
    {
      "epoch": 1.7576271186440677,
      "grad_norm": 0.9426900744438171,
      "learning_rate": 2.070621468926554e-05,
      "loss": 2.8266,
      "step": 2074
    },
    {
      "epoch": 1.7584745762711864,
      "grad_norm": 0.9650610089302063,
      "learning_rate": 2.0692090395480226e-05,
      "loss": 2.7062,
      "step": 2075
    },
    {
      "epoch": 1.759322033898305,
      "grad_norm": 1.1414823532104492,
      "learning_rate": 2.0677966101694916e-05,
      "loss": 2.692,
      "step": 2076
    },
    {
      "epoch": 1.7601694915254238,
      "grad_norm": 1.1584335565567017,
      "learning_rate": 2.0663841807909606e-05,
      "loss": 2.7301,
      "step": 2077
    },
    {
      "epoch": 1.7610169491525425,
      "grad_norm": 1.273834466934204,
      "learning_rate": 2.0649717514124296e-05,
      "loss": 2.6977,
      "step": 2078
    },
    {
      "epoch": 1.7618644067796612,
      "grad_norm": 1.0295066833496094,
      "learning_rate": 2.0635593220338982e-05,
      "loss": 2.7839,
      "step": 2079
    },
    {
      "epoch": 1.7627118644067796,
      "grad_norm": 1.2004121541976929,
      "learning_rate": 2.0621468926553672e-05,
      "loss": 2.7101,
      "step": 2080
    },
    {
      "epoch": 1.7635593220338983,
      "grad_norm": 1.094724416732788,
      "learning_rate": 2.0607344632768362e-05,
      "loss": 2.823,
      "step": 2081
    },
    {
      "epoch": 1.764406779661017,
      "grad_norm": 1.1883249282836914,
      "learning_rate": 2.0593220338983052e-05,
      "loss": 2.7372,
      "step": 2082
    },
    {
      "epoch": 1.7652542372881355,
      "grad_norm": 1.0656414031982422,
      "learning_rate": 2.057909604519774e-05,
      "loss": 2.9454,
      "step": 2083
    },
    {
      "epoch": 1.7661016949152541,
      "grad_norm": 0.9597123265266418,
      "learning_rate": 2.056497175141243e-05,
      "loss": 2.8202,
      "step": 2084
    },
    {
      "epoch": 1.7669491525423728,
      "grad_norm": 1.1121854782104492,
      "learning_rate": 2.055084745762712e-05,
      "loss": 2.8834,
      "step": 2085
    },
    {
      "epoch": 1.7677966101694915,
      "grad_norm": 1.2816334962844849,
      "learning_rate": 2.053672316384181e-05,
      "loss": 2.6536,
      "step": 2086
    },
    {
      "epoch": 1.7686440677966102,
      "grad_norm": 0.9627737402915955,
      "learning_rate": 2.05225988700565e-05,
      "loss": 2.8593,
      "step": 2087
    },
    {
      "epoch": 1.769491525423729,
      "grad_norm": 1.2878440618515015,
      "learning_rate": 2.0508474576271186e-05,
      "loss": 2.6157,
      "step": 2088
    },
    {
      "epoch": 1.7703389830508476,
      "grad_norm": 0.9968247413635254,
      "learning_rate": 2.049435028248588e-05,
      "loss": 3.0156,
      "step": 2089
    },
    {
      "epoch": 1.7711864406779663,
      "grad_norm": 1.215289831161499,
      "learning_rate": 2.0480225988700566e-05,
      "loss": 2.833,
      "step": 2090
    },
    {
      "epoch": 1.7720338983050847,
      "grad_norm": 1.0854917764663696,
      "learning_rate": 2.0466101694915256e-05,
      "loss": 2.8132,
      "step": 2091
    },
    {
      "epoch": 1.7728813559322034,
      "grad_norm": 0.9769462943077087,
      "learning_rate": 2.0451977401129946e-05,
      "loss": 2.7863,
      "step": 2092
    },
    {
      "epoch": 1.7737288135593219,
      "grad_norm": 1.1860283613204956,
      "learning_rate": 2.0437853107344636e-05,
      "loss": 2.6353,
      "step": 2093
    },
    {
      "epoch": 1.7745762711864406,
      "grad_norm": 1.1719470024108887,
      "learning_rate": 2.0423728813559322e-05,
      "loss": 2.6009,
      "step": 2094
    },
    {
      "epoch": 1.7754237288135593,
      "grad_norm": 1.058423638343811,
      "learning_rate": 2.0409604519774012e-05,
      "loss": 2.6743,
      "step": 2095
    },
    {
      "epoch": 1.776271186440678,
      "grad_norm": 0.9297478795051575,
      "learning_rate": 2.0395480225988702e-05,
      "loss": 2.7433,
      "step": 2096
    },
    {
      "epoch": 1.7771186440677966,
      "grad_norm": 0.9432790875434875,
      "learning_rate": 2.0381355932203392e-05,
      "loss": 2.8175,
      "step": 2097
    },
    {
      "epoch": 1.7779661016949153,
      "grad_norm": 1.2309918403625488,
      "learning_rate": 2.036723163841808e-05,
      "loss": 2.5988,
      "step": 2098
    },
    {
      "epoch": 1.778813559322034,
      "grad_norm": 1.0979061126708984,
      "learning_rate": 2.035310734463277e-05,
      "loss": 2.7199,
      "step": 2099
    },
    {
      "epoch": 1.7796610169491527,
      "grad_norm": 1.209404706954956,
      "learning_rate": 2.033898305084746e-05,
      "loss": 2.774,
      "step": 2100
    },
    {
      "epoch": 1.7805084745762711,
      "grad_norm": 1.371813178062439,
      "learning_rate": 2.032485875706215e-05,
      "loss": 2.6455,
      "step": 2101
    },
    {
      "epoch": 1.7813559322033898,
      "grad_norm": 1.0844600200653076,
      "learning_rate": 2.031073446327684e-05,
      "loss": 2.6517,
      "step": 2102
    },
    {
      "epoch": 1.7822033898305085,
      "grad_norm": 1.1611357927322388,
      "learning_rate": 2.0296610169491525e-05,
      "loss": 2.7386,
      "step": 2103
    },
    {
      "epoch": 1.783050847457627,
      "grad_norm": 1.02379310131073,
      "learning_rate": 2.0282485875706215e-05,
      "loss": 2.8435,
      "step": 2104
    },
    {
      "epoch": 1.7838983050847457,
      "grad_norm": 0.9997214674949646,
      "learning_rate": 2.0268361581920905e-05,
      "loss": 2.7857,
      "step": 2105
    },
    {
      "epoch": 1.7847457627118644,
      "grad_norm": 1.022742748260498,
      "learning_rate": 2.0254237288135595e-05,
      "loss": 2.7433,
      "step": 2106
    },
    {
      "epoch": 1.785593220338983,
      "grad_norm": 1.0071974992752075,
      "learning_rate": 2.0240112994350282e-05,
      "loss": 2.8102,
      "step": 2107
    },
    {
      "epoch": 1.7864406779661017,
      "grad_norm": 1.0485122203826904,
      "learning_rate": 2.0225988700564975e-05,
      "loss": 2.748,
      "step": 2108
    },
    {
      "epoch": 1.7872881355932204,
      "grad_norm": 1.1566766500473022,
      "learning_rate": 2.0211864406779662e-05,
      "loss": 2.7775,
      "step": 2109
    },
    {
      "epoch": 1.788135593220339,
      "grad_norm": 1.1086699962615967,
      "learning_rate": 2.0197740112994352e-05,
      "loss": 2.6061,
      "step": 2110
    },
    {
      "epoch": 1.7889830508474578,
      "grad_norm": 1.3776793479919434,
      "learning_rate": 2.018361581920904e-05,
      "loss": 2.5477,
      "step": 2111
    },
    {
      "epoch": 1.7898305084745763,
      "grad_norm": 1.1087639331817627,
      "learning_rate": 2.0169491525423732e-05,
      "loss": 2.6624,
      "step": 2112
    },
    {
      "epoch": 1.790677966101695,
      "grad_norm": 1.152855396270752,
      "learning_rate": 2.015536723163842e-05,
      "loss": 2.7034,
      "step": 2113
    },
    {
      "epoch": 1.7915254237288134,
      "grad_norm": 1.210541844367981,
      "learning_rate": 2.014124293785311e-05,
      "loss": 2.6426,
      "step": 2114
    },
    {
      "epoch": 1.792372881355932,
      "grad_norm": 1.0268137454986572,
      "learning_rate": 2.0127118644067795e-05,
      "loss": 2.7474,
      "step": 2115
    },
    {
      "epoch": 1.7932203389830508,
      "grad_norm": 0.939582884311676,
      "learning_rate": 2.011299435028249e-05,
      "loss": 2.9162,
      "step": 2116
    },
    {
      "epoch": 1.7940677966101695,
      "grad_norm": 1.0411386489868164,
      "learning_rate": 2.0098870056497175e-05,
      "loss": 2.8323,
      "step": 2117
    },
    {
      "epoch": 1.7949152542372881,
      "grad_norm": 0.9271771907806396,
      "learning_rate": 2.0084745762711865e-05,
      "loss": 2.8062,
      "step": 2118
    },
    {
      "epoch": 1.7957627118644068,
      "grad_norm": 1.2471201419830322,
      "learning_rate": 2.0070621468926555e-05,
      "loss": 2.5155,
      "step": 2119
    },
    {
      "epoch": 1.7966101694915255,
      "grad_norm": 1.2951605319976807,
      "learning_rate": 2.0056497175141245e-05,
      "loss": 2.728,
      "step": 2120
    },
    {
      "epoch": 1.7974576271186442,
      "grad_norm": 1.287815809249878,
      "learning_rate": 2.0042372881355935e-05,
      "loss": 2.6545,
      "step": 2121
    },
    {
      "epoch": 1.7983050847457627,
      "grad_norm": 1.2197237014770508,
      "learning_rate": 2.002824858757062e-05,
      "loss": 2.7637,
      "step": 2122
    },
    {
      "epoch": 1.7991525423728814,
      "grad_norm": 1.4295421838760376,
      "learning_rate": 2.001412429378531e-05,
      "loss": 2.403,
      "step": 2123
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1763114929199219,
      "learning_rate": 2e-05,
      "loss": 2.7786,
      "step": 2124
    },
    {
      "epoch": 1.8008474576271185,
      "grad_norm": 1.3243646621704102,
      "learning_rate": 1.998587570621469e-05,
      "loss": 2.5181,
      "step": 2125
    },
    {
      "epoch": 1.8016949152542372,
      "grad_norm": 1.274304986000061,
      "learning_rate": 1.9971751412429378e-05,
      "loss": 2.694,
      "step": 2126
    },
    {
      "epoch": 1.8025423728813559,
      "grad_norm": 0.9818578958511353,
      "learning_rate": 1.9957627118644068e-05,
      "loss": 2.8525,
      "step": 2127
    },
    {
      "epoch": 1.8033898305084746,
      "grad_norm": 0.9407678842544556,
      "learning_rate": 1.9943502824858758e-05,
      "loss": 2.8587,
      "step": 2128
    },
    {
      "epoch": 1.8042372881355933,
      "grad_norm": 1.1443729400634766,
      "learning_rate": 1.9929378531073448e-05,
      "loss": 2.7655,
      "step": 2129
    },
    {
      "epoch": 1.805084745762712,
      "grad_norm": 1.6161749362945557,
      "learning_rate": 1.9915254237288135e-05,
      "loss": 2.4994,
      "step": 2130
    },
    {
      "epoch": 1.8059322033898306,
      "grad_norm": 0.8980959057807922,
      "learning_rate": 1.9901129943502825e-05,
      "loss": 2.9398,
      "step": 2131
    },
    {
      "epoch": 1.8067796610169493,
      "grad_norm": 1.040073037147522,
      "learning_rate": 1.9887005649717515e-05,
      "loss": 2.7628,
      "step": 2132
    },
    {
      "epoch": 1.8076271186440678,
      "grad_norm": 0.876245379447937,
      "learning_rate": 1.9872881355932205e-05,
      "loss": 2.8709,
      "step": 2133
    },
    {
      "epoch": 1.8084745762711865,
      "grad_norm": 1.258423089981079,
      "learning_rate": 1.9858757062146895e-05,
      "loss": 2.6829,
      "step": 2134
    },
    {
      "epoch": 1.809322033898305,
      "grad_norm": 1.1346708536148071,
      "learning_rate": 1.984463276836158e-05,
      "loss": 2.6062,
      "step": 2135
    },
    {
      "epoch": 1.8101694915254236,
      "grad_norm": 1.0326768159866333,
      "learning_rate": 1.9830508474576275e-05,
      "loss": 2.8353,
      "step": 2136
    },
    {
      "epoch": 1.8110169491525423,
      "grad_norm": 1.1588972806930542,
      "learning_rate": 1.981638418079096e-05,
      "loss": 2.7704,
      "step": 2137
    },
    {
      "epoch": 1.811864406779661,
      "grad_norm": 1.1882299184799194,
      "learning_rate": 1.980225988700565e-05,
      "loss": 2.756,
      "step": 2138
    },
    {
      "epoch": 1.8127118644067797,
      "grad_norm": 1.1389881372451782,
      "learning_rate": 1.9788135593220338e-05,
      "loss": 2.7587,
      "step": 2139
    },
    {
      "epoch": 1.8135593220338984,
      "grad_norm": 1.2022271156311035,
      "learning_rate": 1.977401129943503e-05,
      "loss": 2.7051,
      "step": 2140
    },
    {
      "epoch": 1.814406779661017,
      "grad_norm": 0.979529857635498,
      "learning_rate": 1.9759887005649718e-05,
      "loss": 2.7192,
      "step": 2141
    },
    {
      "epoch": 1.8152542372881357,
      "grad_norm": 1.2489396333694458,
      "learning_rate": 1.9745762711864408e-05,
      "loss": 2.6634,
      "step": 2142
    },
    {
      "epoch": 1.8161016949152542,
      "grad_norm": 1.022739291191101,
      "learning_rate": 1.9731638418079095e-05,
      "loss": 2.8749,
      "step": 2143
    },
    {
      "epoch": 1.8169491525423729,
      "grad_norm": 1.1680999994277954,
      "learning_rate": 1.9717514124293788e-05,
      "loss": 2.7571,
      "step": 2144
    },
    {
      "epoch": 1.8177966101694916,
      "grad_norm": 0.9415079355239868,
      "learning_rate": 1.9703389830508475e-05,
      "loss": 2.9925,
      "step": 2145
    },
    {
      "epoch": 1.81864406779661,
      "grad_norm": 1.0888499021530151,
      "learning_rate": 1.9689265536723165e-05,
      "loss": 2.8189,
      "step": 2146
    },
    {
      "epoch": 1.8194915254237287,
      "grad_norm": 1.1665993928909302,
      "learning_rate": 1.9675141242937855e-05,
      "loss": 2.7571,
      "step": 2147
    },
    {
      "epoch": 1.8203389830508474,
      "grad_norm": 0.8712019920349121,
      "learning_rate": 1.9661016949152545e-05,
      "loss": 2.8911,
      "step": 2148
    },
    {
      "epoch": 1.821186440677966,
      "grad_norm": 1.4699656963348389,
      "learning_rate": 1.964689265536723e-05,
      "loss": 2.4671,
      "step": 2149
    },
    {
      "epoch": 1.8220338983050848,
      "grad_norm": 1.2239631414413452,
      "learning_rate": 1.963276836158192e-05,
      "loss": 2.7121,
      "step": 2150
    },
    {
      "epoch": 1.8228813559322035,
      "grad_norm": 0.9783174395561218,
      "learning_rate": 1.961864406779661e-05,
      "loss": 2.7338,
      "step": 2151
    },
    {
      "epoch": 1.8237288135593221,
      "grad_norm": 1.070614218711853,
      "learning_rate": 1.96045197740113e-05,
      "loss": 2.7555,
      "step": 2152
    },
    {
      "epoch": 1.8245762711864408,
      "grad_norm": 1.1637415885925293,
      "learning_rate": 1.959039548022599e-05,
      "loss": 2.5712,
      "step": 2153
    },
    {
      "epoch": 1.8254237288135593,
      "grad_norm": 1.3556163311004639,
      "learning_rate": 1.9576271186440678e-05,
      "loss": 2.5231,
      "step": 2154
    },
    {
      "epoch": 1.826271186440678,
      "grad_norm": 1.3749444484710693,
      "learning_rate": 1.9562146892655368e-05,
      "loss": 2.6462,
      "step": 2155
    },
    {
      "epoch": 1.8271186440677964,
      "grad_norm": 0.9771831631660461,
      "learning_rate": 1.9548022598870058e-05,
      "loss": 2.7385,
      "step": 2156
    },
    {
      "epoch": 1.8279661016949151,
      "grad_norm": 1.2296011447906494,
      "learning_rate": 1.9533898305084748e-05,
      "loss": 2.6671,
      "step": 2157
    },
    {
      "epoch": 1.8288135593220338,
      "grad_norm": 1.0204658508300781,
      "learning_rate": 1.9519774011299434e-05,
      "loss": 2.7002,
      "step": 2158
    },
    {
      "epoch": 1.8296610169491525,
      "grad_norm": 1.3096648454666138,
      "learning_rate": 1.9505649717514124e-05,
      "loss": 2.6166,
      "step": 2159
    },
    {
      "epoch": 1.8305084745762712,
      "grad_norm": 1.095273733139038,
      "learning_rate": 1.9491525423728814e-05,
      "loss": 2.6907,
      "step": 2160
    },
    {
      "epoch": 1.8313559322033899,
      "grad_norm": 1.1508970260620117,
      "learning_rate": 1.9477401129943504e-05,
      "loss": 2.584,
      "step": 2161
    },
    {
      "epoch": 1.8322033898305086,
      "grad_norm": 1.021013617515564,
      "learning_rate": 1.946327683615819e-05,
      "loss": 2.7443,
      "step": 2162
    },
    {
      "epoch": 1.8330508474576273,
      "grad_norm": 1.1462894678115845,
      "learning_rate": 1.944915254237288e-05,
      "loss": 2.7603,
      "step": 2163
    },
    {
      "epoch": 1.8338983050847457,
      "grad_norm": 1.1708370447158813,
      "learning_rate": 1.943502824858757e-05,
      "loss": 2.7284,
      "step": 2164
    },
    {
      "epoch": 1.8347457627118644,
      "grad_norm": 1.0491888523101807,
      "learning_rate": 1.942090395480226e-05,
      "loss": 2.7925,
      "step": 2165
    },
    {
      "epoch": 1.835593220338983,
      "grad_norm": 1.0713671445846558,
      "learning_rate": 1.940677966101695e-05,
      "loss": 2.8049,
      "step": 2166
    },
    {
      "epoch": 1.8364406779661016,
      "grad_norm": 0.9071455001831055,
      "learning_rate": 1.9392655367231638e-05,
      "loss": 2.8179,
      "step": 2167
    },
    {
      "epoch": 1.8372881355932202,
      "grad_norm": 1.0441315174102783,
      "learning_rate": 1.937853107344633e-05,
      "loss": 2.7795,
      "step": 2168
    },
    {
      "epoch": 1.838135593220339,
      "grad_norm": 1.1267039775848389,
      "learning_rate": 1.9364406779661017e-05,
      "loss": 2.8345,
      "step": 2169
    },
    {
      "epoch": 1.8389830508474576,
      "grad_norm": 0.9966780543327332,
      "learning_rate": 1.9350282485875707e-05,
      "loss": 2.8733,
      "step": 2170
    },
    {
      "epoch": 1.8398305084745763,
      "grad_norm": 0.9944255352020264,
      "learning_rate": 1.9336158192090394e-05,
      "loss": 2.8219,
      "step": 2171
    },
    {
      "epoch": 1.840677966101695,
      "grad_norm": 1.210039734840393,
      "learning_rate": 1.9322033898305087e-05,
      "loss": 2.6556,
      "step": 2172
    },
    {
      "epoch": 1.8415254237288137,
      "grad_norm": 0.9125044941902161,
      "learning_rate": 1.9307909604519774e-05,
      "loss": 2.7901,
      "step": 2173
    },
    {
      "epoch": 1.8423728813559324,
      "grad_norm": 1.077428936958313,
      "learning_rate": 1.9293785310734464e-05,
      "loss": 2.6814,
      "step": 2174
    },
    {
      "epoch": 1.8432203389830508,
      "grad_norm": 1.1277340650558472,
      "learning_rate": 1.9279661016949154e-05,
      "loss": 2.7702,
      "step": 2175
    },
    {
      "epoch": 1.8440677966101695,
      "grad_norm": 0.9860967397689819,
      "learning_rate": 1.9265536723163844e-05,
      "loss": 2.8478,
      "step": 2176
    },
    {
      "epoch": 1.844915254237288,
      "grad_norm": 1.0616719722747803,
      "learning_rate": 1.925141242937853e-05,
      "loss": 2.6662,
      "step": 2177
    },
    {
      "epoch": 1.8457627118644067,
      "grad_norm": 1.067347526550293,
      "learning_rate": 1.923728813559322e-05,
      "loss": 2.6692,
      "step": 2178
    },
    {
      "epoch": 1.8466101694915253,
      "grad_norm": 1.049932837486267,
      "learning_rate": 1.922316384180791e-05,
      "loss": 2.9301,
      "step": 2179
    },
    {
      "epoch": 1.847457627118644,
      "grad_norm": 1.3768372535705566,
      "learning_rate": 1.92090395480226e-05,
      "loss": 2.6821,
      "step": 2180
    },
    {
      "epoch": 1.8483050847457627,
      "grad_norm": 1.0082629919052124,
      "learning_rate": 1.919491525423729e-05,
      "loss": 2.6566,
      "step": 2181
    },
    {
      "epoch": 1.8491525423728814,
      "grad_norm": 1.349716305732727,
      "learning_rate": 1.9180790960451977e-05,
      "loss": 2.5341,
      "step": 2182
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0393669605255127,
      "learning_rate": 1.9166666666666667e-05,
      "loss": 2.7813,
      "step": 2183
    },
    {
      "epoch": 1.8508474576271188,
      "grad_norm": 1.1654995679855347,
      "learning_rate": 1.9152542372881357e-05,
      "loss": 2.7206,
      "step": 2184
    },
    {
      "epoch": 1.8516949152542372,
      "grad_norm": 1.2510303258895874,
      "learning_rate": 1.9138418079096047e-05,
      "loss": 2.7678,
      "step": 2185
    },
    {
      "epoch": 1.852542372881356,
      "grad_norm": 1.1300528049468994,
      "learning_rate": 1.9124293785310734e-05,
      "loss": 2.8084,
      "step": 2186
    },
    {
      "epoch": 1.8533898305084746,
      "grad_norm": 1.061294436454773,
      "learning_rate": 1.9110169491525427e-05,
      "loss": 2.7987,
      "step": 2187
    },
    {
      "epoch": 1.854237288135593,
      "grad_norm": 1.3102290630340576,
      "learning_rate": 1.9096045197740114e-05,
      "loss": 2.717,
      "step": 2188
    },
    {
      "epoch": 1.8550847457627118,
      "grad_norm": 1.0602400302886963,
      "learning_rate": 1.9081920903954804e-05,
      "loss": 2.778,
      "step": 2189
    },
    {
      "epoch": 1.8559322033898304,
      "grad_norm": 1.071914553642273,
      "learning_rate": 1.906779661016949e-05,
      "loss": 2.6677,
      "step": 2190
    },
    {
      "epoch": 1.8567796610169491,
      "grad_norm": 1.200467824935913,
      "learning_rate": 1.9053672316384184e-05,
      "loss": 2.765,
      "step": 2191
    },
    {
      "epoch": 1.8576271186440678,
      "grad_norm": 1.106851577758789,
      "learning_rate": 1.903954802259887e-05,
      "loss": 2.7109,
      "step": 2192
    },
    {
      "epoch": 1.8584745762711865,
      "grad_norm": 1.20026433467865,
      "learning_rate": 1.902542372881356e-05,
      "loss": 2.6831,
      "step": 2193
    },
    {
      "epoch": 1.8593220338983052,
      "grad_norm": 1.107622742652893,
      "learning_rate": 1.9011299435028247e-05,
      "loss": 2.6599,
      "step": 2194
    },
    {
      "epoch": 1.8601694915254239,
      "grad_norm": 0.9286526441574097,
      "learning_rate": 1.899717514124294e-05,
      "loss": 2.8641,
      "step": 2195
    },
    {
      "epoch": 1.8610169491525423,
      "grad_norm": 1.001364827156067,
      "learning_rate": 1.8983050847457627e-05,
      "loss": 2.7642,
      "step": 2196
    },
    {
      "epoch": 1.861864406779661,
      "grad_norm": 1.1795566082000732,
      "learning_rate": 1.8968926553672317e-05,
      "loss": 2.7738,
      "step": 2197
    },
    {
      "epoch": 1.8627118644067797,
      "grad_norm": 0.9288800954818726,
      "learning_rate": 1.8954802259887007e-05,
      "loss": 2.7021,
      "step": 2198
    },
    {
      "epoch": 1.8635593220338982,
      "grad_norm": 1.2071678638458252,
      "learning_rate": 1.8940677966101697e-05,
      "loss": 2.7749,
      "step": 2199
    },
    {
      "epoch": 1.8644067796610169,
      "grad_norm": 1.0884045362472534,
      "learning_rate": 1.8926553672316387e-05,
      "loss": 2.7885,
      "step": 2200
    },
    {
      "epoch": 1.8652542372881356,
      "grad_norm": 1.0357964038848877,
      "learning_rate": 1.8912429378531074e-05,
      "loss": 2.7344,
      "step": 2201
    },
    {
      "epoch": 1.8661016949152542,
      "grad_norm": 1.1365445852279663,
      "learning_rate": 1.8898305084745764e-05,
      "loss": 2.7749,
      "step": 2202
    },
    {
      "epoch": 1.866949152542373,
      "grad_norm": 1.066388487815857,
      "learning_rate": 1.8884180790960454e-05,
      "loss": 2.7136,
      "step": 2203
    },
    {
      "epoch": 1.8677966101694916,
      "grad_norm": 1.2417033910751343,
      "learning_rate": 1.8870056497175144e-05,
      "loss": 2.6089,
      "step": 2204
    },
    {
      "epoch": 1.8686440677966103,
      "grad_norm": 1.0588102340698242,
      "learning_rate": 1.885593220338983e-05,
      "loss": 2.737,
      "step": 2205
    },
    {
      "epoch": 1.8694915254237288,
      "grad_norm": 1.0809686183929443,
      "learning_rate": 1.884180790960452e-05,
      "loss": 2.6101,
      "step": 2206
    },
    {
      "epoch": 1.8703389830508474,
      "grad_norm": 0.9519467353820801,
      "learning_rate": 1.882768361581921e-05,
      "loss": 2.8298,
      "step": 2207
    },
    {
      "epoch": 1.8711864406779661,
      "grad_norm": 0.9441993236541748,
      "learning_rate": 1.88135593220339e-05,
      "loss": 2.7523,
      "step": 2208
    },
    {
      "epoch": 1.8720338983050846,
      "grad_norm": 0.996285080909729,
      "learning_rate": 1.8799435028248587e-05,
      "loss": 2.8619,
      "step": 2209
    },
    {
      "epoch": 1.8728813559322033,
      "grad_norm": 1.322708010673523,
      "learning_rate": 1.8785310734463277e-05,
      "loss": 2.6528,
      "step": 2210
    },
    {
      "epoch": 1.873728813559322,
      "grad_norm": 1.1378732919692993,
      "learning_rate": 1.8771186440677967e-05,
      "loss": 2.6988,
      "step": 2211
    },
    {
      "epoch": 1.8745762711864407,
      "grad_norm": 1.2078667879104614,
      "learning_rate": 1.8757062146892657e-05,
      "loss": 2.6782,
      "step": 2212
    },
    {
      "epoch": 1.8754237288135593,
      "grad_norm": 1.134391188621521,
      "learning_rate": 1.8742937853107347e-05,
      "loss": 2.701,
      "step": 2213
    },
    {
      "epoch": 1.876271186440678,
      "grad_norm": 0.9816274642944336,
      "learning_rate": 1.8728813559322033e-05,
      "loss": 2.6723,
      "step": 2214
    },
    {
      "epoch": 1.8771186440677967,
      "grad_norm": 1.3717323541641235,
      "learning_rate": 1.8714689265536727e-05,
      "loss": 2.4816,
      "step": 2215
    },
    {
      "epoch": 1.8779661016949154,
      "grad_norm": 1.190843105316162,
      "learning_rate": 1.8700564971751413e-05,
      "loss": 2.6468,
      "step": 2216
    },
    {
      "epoch": 1.8788135593220339,
      "grad_norm": 0.9955922961235046,
      "learning_rate": 1.8686440677966103e-05,
      "loss": 2.7169,
      "step": 2217
    },
    {
      "epoch": 1.8796610169491526,
      "grad_norm": 1.3320982456207275,
      "learning_rate": 1.867231638418079e-05,
      "loss": 2.6252,
      "step": 2218
    },
    {
      "epoch": 1.8805084745762712,
      "grad_norm": 1.1454169750213623,
      "learning_rate": 1.8658192090395483e-05,
      "loss": 2.7791,
      "step": 2219
    },
    {
      "epoch": 1.8813559322033897,
      "grad_norm": 1.174757480621338,
      "learning_rate": 1.864406779661017e-05,
      "loss": 2.7574,
      "step": 2220
    },
    {
      "epoch": 1.8822033898305084,
      "grad_norm": 1.0740658044815063,
      "learning_rate": 1.862994350282486e-05,
      "loss": 2.6172,
      "step": 2221
    },
    {
      "epoch": 1.883050847457627,
      "grad_norm": 1.2753480672836304,
      "learning_rate": 1.8615819209039547e-05,
      "loss": 2.8159,
      "step": 2222
    },
    {
      "epoch": 1.8838983050847458,
      "grad_norm": 1.1446774005889893,
      "learning_rate": 1.860169491525424e-05,
      "loss": 2.7301,
      "step": 2223
    },
    {
      "epoch": 1.8847457627118644,
      "grad_norm": 1.1188921928405762,
      "learning_rate": 1.8587570621468927e-05,
      "loss": 2.6202,
      "step": 2224
    },
    {
      "epoch": 1.8855932203389831,
      "grad_norm": 0.9338502883911133,
      "learning_rate": 1.8573446327683617e-05,
      "loss": 2.7956,
      "step": 2225
    },
    {
      "epoch": 1.8864406779661018,
      "grad_norm": 1.2575122117996216,
      "learning_rate": 1.8559322033898307e-05,
      "loss": 2.613,
      "step": 2226
    },
    {
      "epoch": 1.8872881355932203,
      "grad_norm": 1.1470165252685547,
      "learning_rate": 1.8545197740112996e-05,
      "loss": 2.6774,
      "step": 2227
    },
    {
      "epoch": 1.888135593220339,
      "grad_norm": 1.2760436534881592,
      "learning_rate": 1.8531073446327683e-05,
      "loss": 2.5741,
      "step": 2228
    },
    {
      "epoch": 1.8889830508474577,
      "grad_norm": 1.1932239532470703,
      "learning_rate": 1.8516949152542373e-05,
      "loss": 2.5973,
      "step": 2229
    },
    {
      "epoch": 1.8898305084745761,
      "grad_norm": 1.0808064937591553,
      "learning_rate": 1.8502824858757063e-05,
      "loss": 2.7211,
      "step": 2230
    },
    {
      "epoch": 1.8906779661016948,
      "grad_norm": 1.3394954204559326,
      "learning_rate": 1.8488700564971753e-05,
      "loss": 2.6208,
      "step": 2231
    },
    {
      "epoch": 1.8915254237288135,
      "grad_norm": 1.260796070098877,
      "learning_rate": 1.8474576271186443e-05,
      "loss": 2.675,
      "step": 2232
    },
    {
      "epoch": 1.8923728813559322,
      "grad_norm": 0.9751023054122925,
      "learning_rate": 1.846045197740113e-05,
      "loss": 2.8524,
      "step": 2233
    },
    {
      "epoch": 1.8932203389830509,
      "grad_norm": 0.94771409034729,
      "learning_rate": 1.844632768361582e-05,
      "loss": 2.7193,
      "step": 2234
    },
    {
      "epoch": 1.8940677966101696,
      "grad_norm": 1.106333613395691,
      "learning_rate": 1.843220338983051e-05,
      "loss": 2.736,
      "step": 2235
    },
    {
      "epoch": 1.8949152542372882,
      "grad_norm": 1.285568356513977,
      "learning_rate": 1.84180790960452e-05,
      "loss": 2.6821,
      "step": 2236
    },
    {
      "epoch": 1.895762711864407,
      "grad_norm": 1.1260167360305786,
      "learning_rate": 1.8403954802259886e-05,
      "loss": 2.7211,
      "step": 2237
    },
    {
      "epoch": 1.8966101694915254,
      "grad_norm": 1.0271403789520264,
      "learning_rate": 1.8389830508474576e-05,
      "loss": 2.723,
      "step": 2238
    },
    {
      "epoch": 1.897457627118644,
      "grad_norm": 1.2635908126831055,
      "learning_rate": 1.8375706214689266e-05,
      "loss": 2.6671,
      "step": 2239
    },
    {
      "epoch": 1.8983050847457628,
      "grad_norm": 1.036975622177124,
      "learning_rate": 1.8361581920903956e-05,
      "loss": 2.7593,
      "step": 2240
    },
    {
      "epoch": 1.8991525423728812,
      "grad_norm": 1.1501283645629883,
      "learning_rate": 1.8347457627118643e-05,
      "loss": 2.6194,
      "step": 2241
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1000744104385376,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 2.742,
      "step": 2242
    },
    {
      "epoch": 1.9008474576271186,
      "grad_norm": 1.2108416557312012,
      "learning_rate": 1.8319209039548023e-05,
      "loss": 2.5673,
      "step": 2243
    },
    {
      "epoch": 1.9016949152542373,
      "grad_norm": 1.5413299798965454,
      "learning_rate": 1.8305084745762713e-05,
      "loss": 2.6373,
      "step": 2244
    },
    {
      "epoch": 1.902542372881356,
      "grad_norm": 1.1783409118652344,
      "learning_rate": 1.8290960451977403e-05,
      "loss": 2.7444,
      "step": 2245
    },
    {
      "epoch": 1.9033898305084747,
      "grad_norm": 1.1007132530212402,
      "learning_rate": 1.827683615819209e-05,
      "loss": 2.6802,
      "step": 2246
    },
    {
      "epoch": 1.9042372881355933,
      "grad_norm": 1.174351692199707,
      "learning_rate": 1.8262711864406783e-05,
      "loss": 2.5913,
      "step": 2247
    },
    {
      "epoch": 1.905084745762712,
      "grad_norm": 1.317480444908142,
      "learning_rate": 1.824858757062147e-05,
      "loss": 2.7326,
      "step": 2248
    },
    {
      "epoch": 1.9059322033898305,
      "grad_norm": 1.2611827850341797,
      "learning_rate": 1.823446327683616e-05,
      "loss": 2.5288,
      "step": 2249
    },
    {
      "epoch": 1.9067796610169492,
      "grad_norm": 1.1809849739074707,
      "learning_rate": 1.8220338983050846e-05,
      "loss": 2.7133,
      "step": 2250
    },
    {
      "epoch": 1.9076271186440676,
      "grad_norm": 1.1002089977264404,
      "learning_rate": 1.820621468926554e-05,
      "loss": 2.7633,
      "step": 2251
    },
    {
      "epoch": 1.9084745762711863,
      "grad_norm": 0.9564800262451172,
      "learning_rate": 1.8192090395480226e-05,
      "loss": 2.7799,
      "step": 2252
    },
    {
      "epoch": 1.909322033898305,
      "grad_norm": 1.0163698196411133,
      "learning_rate": 1.8177966101694916e-05,
      "loss": 2.8996,
      "step": 2253
    },
    {
      "epoch": 1.9101694915254237,
      "grad_norm": 1.151449203491211,
      "learning_rate": 1.8163841807909606e-05,
      "loss": 2.6482,
      "step": 2254
    },
    {
      "epoch": 1.9110169491525424,
      "grad_norm": 1.3174258470535278,
      "learning_rate": 1.8149717514124296e-05,
      "loss": 2.6334,
      "step": 2255
    },
    {
      "epoch": 1.911864406779661,
      "grad_norm": 1.4039971828460693,
      "learning_rate": 1.8135593220338983e-05,
      "loss": 2.6215,
      "step": 2256
    },
    {
      "epoch": 1.9127118644067798,
      "grad_norm": 1.1137800216674805,
      "learning_rate": 1.8121468926553673e-05,
      "loss": 2.653,
      "step": 2257
    },
    {
      "epoch": 1.9135593220338984,
      "grad_norm": 1.1515967845916748,
      "learning_rate": 1.8107344632768363e-05,
      "loss": 2.6147,
      "step": 2258
    },
    {
      "epoch": 1.914406779661017,
      "grad_norm": 1.1500734090805054,
      "learning_rate": 1.8093220338983053e-05,
      "loss": 2.6794,
      "step": 2259
    },
    {
      "epoch": 1.9152542372881356,
      "grad_norm": 1.1761561632156372,
      "learning_rate": 1.8079096045197743e-05,
      "loss": 2.6194,
      "step": 2260
    },
    {
      "epoch": 1.9161016949152543,
      "grad_norm": 1.298223853111267,
      "learning_rate": 1.806497175141243e-05,
      "loss": 2.5401,
      "step": 2261
    },
    {
      "epoch": 1.9169491525423727,
      "grad_norm": 1.3742104768753052,
      "learning_rate": 1.805084745762712e-05,
      "loss": 2.4979,
      "step": 2262
    },
    {
      "epoch": 1.9177966101694914,
      "grad_norm": 1.0494269132614136,
      "learning_rate": 1.803672316384181e-05,
      "loss": 2.692,
      "step": 2263
    },
    {
      "epoch": 1.9186440677966101,
      "grad_norm": 1.443848967552185,
      "learning_rate": 1.80225988700565e-05,
      "loss": 2.5217,
      "step": 2264
    },
    {
      "epoch": 1.9194915254237288,
      "grad_norm": 0.9262023568153381,
      "learning_rate": 1.8008474576271186e-05,
      "loss": 2.8329,
      "step": 2265
    },
    {
      "epoch": 1.9203389830508475,
      "grad_norm": 1.0921679735183716,
      "learning_rate": 1.799435028248588e-05,
      "loss": 2.7401,
      "step": 2266
    },
    {
      "epoch": 1.9211864406779662,
      "grad_norm": 0.9958825707435608,
      "learning_rate": 1.7980225988700566e-05,
      "loss": 2.7963,
      "step": 2267
    },
    {
      "epoch": 1.9220338983050849,
      "grad_norm": 1.287933349609375,
      "learning_rate": 1.7966101694915256e-05,
      "loss": 2.6219,
      "step": 2268
    },
    {
      "epoch": 1.9228813559322036,
      "grad_norm": 1.3278027772903442,
      "learning_rate": 1.7951977401129942e-05,
      "loss": 2.6627,
      "step": 2269
    },
    {
      "epoch": 1.923728813559322,
      "grad_norm": 1.1513268947601318,
      "learning_rate": 1.7937853107344636e-05,
      "loss": 2.6723,
      "step": 2270
    },
    {
      "epoch": 1.9245762711864407,
      "grad_norm": 1.1693884134292603,
      "learning_rate": 1.7923728813559322e-05,
      "loss": 2.7443,
      "step": 2271
    },
    {
      "epoch": 1.9254237288135592,
      "grad_norm": 1.3294950723648071,
      "learning_rate": 1.7909604519774012e-05,
      "loss": 2.7371,
      "step": 2272
    },
    {
      "epoch": 1.9262711864406779,
      "grad_norm": 1.154387354850769,
      "learning_rate": 1.78954802259887e-05,
      "loss": 2.644,
      "step": 2273
    },
    {
      "epoch": 1.9271186440677965,
      "grad_norm": 1.3097997903823853,
      "learning_rate": 1.7881355932203392e-05,
      "loss": 2.4605,
      "step": 2274
    },
    {
      "epoch": 1.9279661016949152,
      "grad_norm": 1.0979993343353271,
      "learning_rate": 1.786723163841808e-05,
      "loss": 2.823,
      "step": 2275
    },
    {
      "epoch": 1.928813559322034,
      "grad_norm": 1.0449799299240112,
      "learning_rate": 1.785310734463277e-05,
      "loss": 2.8048,
      "step": 2276
    },
    {
      "epoch": 1.9296610169491526,
      "grad_norm": 1.0796915292739868,
      "learning_rate": 1.783898305084746e-05,
      "loss": 2.6155,
      "step": 2277
    },
    {
      "epoch": 1.9305084745762713,
      "grad_norm": 1.142608880996704,
      "learning_rate": 1.782485875706215e-05,
      "loss": 2.7279,
      "step": 2278
    },
    {
      "epoch": 1.93135593220339,
      "grad_norm": 1.0652730464935303,
      "learning_rate": 1.781073446327684e-05,
      "loss": 2.7479,
      "step": 2279
    },
    {
      "epoch": 1.9322033898305084,
      "grad_norm": 1.1445574760437012,
      "learning_rate": 1.7796610169491526e-05,
      "loss": 2.7113,
      "step": 2280
    },
    {
      "epoch": 1.9330508474576271,
      "grad_norm": 1.0645530223846436,
      "learning_rate": 1.7782485875706216e-05,
      "loss": 2.6893,
      "step": 2281
    },
    {
      "epoch": 1.9338983050847458,
      "grad_norm": 1.1419038772583008,
      "learning_rate": 1.7768361581920906e-05,
      "loss": 2.9107,
      "step": 2282
    },
    {
      "epoch": 1.9347457627118643,
      "grad_norm": 1.1077309846878052,
      "learning_rate": 1.7754237288135596e-05,
      "loss": 2.6608,
      "step": 2283
    },
    {
      "epoch": 1.935593220338983,
      "grad_norm": 1.1196407079696655,
      "learning_rate": 1.7740112994350282e-05,
      "loss": 2.6925,
      "step": 2284
    },
    {
      "epoch": 1.9364406779661016,
      "grad_norm": 0.935629665851593,
      "learning_rate": 1.7725988700564972e-05,
      "loss": 2.8368,
      "step": 2285
    },
    {
      "epoch": 1.9372881355932203,
      "grad_norm": 1.159534215927124,
      "learning_rate": 1.7711864406779662e-05,
      "loss": 2.7368,
      "step": 2286
    },
    {
      "epoch": 1.938135593220339,
      "grad_norm": 1.1814593076705933,
      "learning_rate": 1.7697740112994352e-05,
      "loss": 2.6718,
      "step": 2287
    },
    {
      "epoch": 1.9389830508474577,
      "grad_norm": 1.2709976434707642,
      "learning_rate": 1.768361581920904e-05,
      "loss": 2.6549,
      "step": 2288
    },
    {
      "epoch": 1.9398305084745764,
      "grad_norm": 1.1238071918487549,
      "learning_rate": 1.766949152542373e-05,
      "loss": 2.7141,
      "step": 2289
    },
    {
      "epoch": 1.940677966101695,
      "grad_norm": 1.2803257703781128,
      "learning_rate": 1.765536723163842e-05,
      "loss": 2.629,
      "step": 2290
    },
    {
      "epoch": 1.9415254237288135,
      "grad_norm": 1.1455533504486084,
      "learning_rate": 1.764124293785311e-05,
      "loss": 2.8567,
      "step": 2291
    },
    {
      "epoch": 1.9423728813559322,
      "grad_norm": 0.9419334530830383,
      "learning_rate": 1.76271186440678e-05,
      "loss": 2.9034,
      "step": 2292
    },
    {
      "epoch": 1.9432203389830507,
      "grad_norm": 1.222684383392334,
      "learning_rate": 1.7612994350282485e-05,
      "loss": 2.7279,
      "step": 2293
    },
    {
      "epoch": 1.9440677966101694,
      "grad_norm": 1.0229870080947876,
      "learning_rate": 1.759887005649718e-05,
      "loss": 2.8089,
      "step": 2294
    },
    {
      "epoch": 1.944915254237288,
      "grad_norm": 0.9789454340934753,
      "learning_rate": 1.7584745762711865e-05,
      "loss": 2.7545,
      "step": 2295
    },
    {
      "epoch": 1.9457627118644067,
      "grad_norm": 1.1572662591934204,
      "learning_rate": 1.7570621468926555e-05,
      "loss": 2.7593,
      "step": 2296
    },
    {
      "epoch": 1.9466101694915254,
      "grad_norm": 1.4241892099380493,
      "learning_rate": 1.7556497175141242e-05,
      "loss": 2.6931,
      "step": 2297
    },
    {
      "epoch": 1.9474576271186441,
      "grad_norm": 1.0359416007995605,
      "learning_rate": 1.7542372881355935e-05,
      "loss": 2.7762,
      "step": 2298
    },
    {
      "epoch": 1.9483050847457628,
      "grad_norm": 1.3145811557769775,
      "learning_rate": 1.7528248587570622e-05,
      "loss": 2.5772,
      "step": 2299
    },
    {
      "epoch": 1.9491525423728815,
      "grad_norm": 1.0483804941177368,
      "learning_rate": 1.7514124293785312e-05,
      "loss": 2.8207,
      "step": 2300
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9640621542930603,
      "learning_rate": 1.75e-05,
      "loss": 2.8818,
      "step": 2301
    },
    {
      "epoch": 1.9508474576271186,
      "grad_norm": 1.356916069984436,
      "learning_rate": 1.7485875706214692e-05,
      "loss": 2.6609,
      "step": 2302
    },
    {
      "epoch": 1.9516949152542373,
      "grad_norm": 1.158837080001831,
      "learning_rate": 1.747175141242938e-05,
      "loss": 2.8008,
      "step": 2303
    },
    {
      "epoch": 1.9525423728813558,
      "grad_norm": 1.094359278678894,
      "learning_rate": 1.745762711864407e-05,
      "loss": 2.6449,
      "step": 2304
    },
    {
      "epoch": 1.9533898305084745,
      "grad_norm": 1.1630890369415283,
      "learning_rate": 1.744350282485876e-05,
      "loss": 2.7461,
      "step": 2305
    },
    {
      "epoch": 1.9542372881355932,
      "grad_norm": 1.2461740970611572,
      "learning_rate": 1.742937853107345e-05,
      "loss": 2.7222,
      "step": 2306
    },
    {
      "epoch": 1.9550847457627119,
      "grad_norm": 0.9834041595458984,
      "learning_rate": 1.7415254237288135e-05,
      "loss": 2.8727,
      "step": 2307
    },
    {
      "epoch": 1.9559322033898305,
      "grad_norm": 1.1653308868408203,
      "learning_rate": 1.7401129943502825e-05,
      "loss": 2.7133,
      "step": 2308
    },
    {
      "epoch": 1.9567796610169492,
      "grad_norm": 1.253213882446289,
      "learning_rate": 1.7387005649717515e-05,
      "loss": 2.614,
      "step": 2309
    },
    {
      "epoch": 1.957627118644068,
      "grad_norm": 1.2413033246994019,
      "learning_rate": 1.7372881355932205e-05,
      "loss": 2.793,
      "step": 2310
    },
    {
      "epoch": 1.9584745762711866,
      "grad_norm": 1.64323890209198,
      "learning_rate": 1.7358757062146895e-05,
      "loss": 2.3969,
      "step": 2311
    },
    {
      "epoch": 1.959322033898305,
      "grad_norm": 1.1061500310897827,
      "learning_rate": 1.734463276836158e-05,
      "loss": 2.6983,
      "step": 2312
    },
    {
      "epoch": 1.9601694915254237,
      "grad_norm": 1.1904529333114624,
      "learning_rate": 1.733050847457627e-05,
      "loss": 2.599,
      "step": 2313
    },
    {
      "epoch": 1.9610169491525422,
      "grad_norm": 0.9217993021011353,
      "learning_rate": 1.731638418079096e-05,
      "loss": 2.8319,
      "step": 2314
    },
    {
      "epoch": 1.961864406779661,
      "grad_norm": 1.1838812828063965,
      "learning_rate": 1.730225988700565e-05,
      "loss": 2.6076,
      "step": 2315
    },
    {
      "epoch": 1.9627118644067796,
      "grad_norm": 1.109630823135376,
      "learning_rate": 1.7288135593220338e-05,
      "loss": 2.7433,
      "step": 2316
    },
    {
      "epoch": 1.9635593220338983,
      "grad_norm": 1.0150429010391235,
      "learning_rate": 1.7274011299435028e-05,
      "loss": 2.8878,
      "step": 2317
    },
    {
      "epoch": 1.964406779661017,
      "grad_norm": 1.0494391918182373,
      "learning_rate": 1.7259887005649718e-05,
      "loss": 2.7438,
      "step": 2318
    },
    {
      "epoch": 1.9652542372881356,
      "grad_norm": 0.9451582431793213,
      "learning_rate": 1.7245762711864408e-05,
      "loss": 2.9528,
      "step": 2319
    },
    {
      "epoch": 1.9661016949152543,
      "grad_norm": 1.250891089439392,
      "learning_rate": 1.7231638418079095e-05,
      "loss": 2.7219,
      "step": 2320
    },
    {
      "epoch": 1.966949152542373,
      "grad_norm": 1.2979968786239624,
      "learning_rate": 1.7217514124293785e-05,
      "loss": 2.7372,
      "step": 2321
    },
    {
      "epoch": 1.9677966101694915,
      "grad_norm": 1.022067666053772,
      "learning_rate": 1.7203389830508475e-05,
      "loss": 2.7496,
      "step": 2322
    },
    {
      "epoch": 1.9686440677966102,
      "grad_norm": 1.2830743789672852,
      "learning_rate": 1.7189265536723165e-05,
      "loss": 2.6263,
      "step": 2323
    },
    {
      "epoch": 1.9694915254237289,
      "grad_norm": 0.9140318036079407,
      "learning_rate": 1.7175141242937855e-05,
      "loss": 2.8402,
      "step": 2324
    },
    {
      "epoch": 1.9703389830508473,
      "grad_norm": 0.993951141834259,
      "learning_rate": 1.716101694915254e-05,
      "loss": 2.82,
      "step": 2325
    },
    {
      "epoch": 1.971186440677966,
      "grad_norm": 1.0804479122161865,
      "learning_rate": 1.7146892655367235e-05,
      "loss": 2.7957,
      "step": 2326
    },
    {
      "epoch": 1.9720338983050847,
      "grad_norm": 1.2359389066696167,
      "learning_rate": 1.713276836158192e-05,
      "loss": 2.5425,
      "step": 2327
    },
    {
      "epoch": 1.9728813559322034,
      "grad_norm": 1.4272359609603882,
      "learning_rate": 1.711864406779661e-05,
      "loss": 2.4573,
      "step": 2328
    },
    {
      "epoch": 1.973728813559322,
      "grad_norm": 1.2127587795257568,
      "learning_rate": 1.7104519774011298e-05,
      "loss": 2.615,
      "step": 2329
    },
    {
      "epoch": 1.9745762711864407,
      "grad_norm": 1.5587515830993652,
      "learning_rate": 1.709039548022599e-05,
      "loss": 2.364,
      "step": 2330
    },
    {
      "epoch": 1.9754237288135594,
      "grad_norm": 1.2130502462387085,
      "learning_rate": 1.7076271186440678e-05,
      "loss": 2.6527,
      "step": 2331
    },
    {
      "epoch": 1.9762711864406781,
      "grad_norm": 0.9383020997047424,
      "learning_rate": 1.7062146892655368e-05,
      "loss": 2.9776,
      "step": 2332
    },
    {
      "epoch": 1.9771186440677966,
      "grad_norm": 1.0962392091751099,
      "learning_rate": 1.7048022598870055e-05,
      "loss": 2.6739,
      "step": 2333
    },
    {
      "epoch": 1.9779661016949153,
      "grad_norm": 1.1072427034378052,
      "learning_rate": 1.7033898305084748e-05,
      "loss": 2.5939,
      "step": 2334
    },
    {
      "epoch": 1.9788135593220337,
      "grad_norm": 1.2364929914474487,
      "learning_rate": 1.7019774011299435e-05,
      "loss": 2.6278,
      "step": 2335
    },
    {
      "epoch": 1.9796610169491524,
      "grad_norm": 1.187670350074768,
      "learning_rate": 1.7005649717514125e-05,
      "loss": 2.6159,
      "step": 2336
    },
    {
      "epoch": 1.980508474576271,
      "grad_norm": 1.0247775316238403,
      "learning_rate": 1.6991525423728815e-05,
      "loss": 2.6774,
      "step": 2337
    },
    {
      "epoch": 1.9813559322033898,
      "grad_norm": 0.7927355766296387,
      "learning_rate": 1.6977401129943505e-05,
      "loss": 2.8451,
      "step": 2338
    },
    {
      "epoch": 1.9822033898305085,
      "grad_norm": 1.3290356397628784,
      "learning_rate": 1.6963276836158195e-05,
      "loss": 2.7093,
      "step": 2339
    },
    {
      "epoch": 1.9830508474576272,
      "grad_norm": 1.2458208799362183,
      "learning_rate": 1.694915254237288e-05,
      "loss": 2.6228,
      "step": 2340
    },
    {
      "epoch": 1.9838983050847459,
      "grad_norm": 0.9700512290000916,
      "learning_rate": 1.693502824858757e-05,
      "loss": 2.7374,
      "step": 2341
    },
    {
      "epoch": 1.9847457627118645,
      "grad_norm": 0.9352066516876221,
      "learning_rate": 1.692090395480226e-05,
      "loss": 3.0912,
      "step": 2342
    },
    {
      "epoch": 1.985593220338983,
      "grad_norm": 1.1421597003936768,
      "learning_rate": 1.690677966101695e-05,
      "loss": 2.6697,
      "step": 2343
    },
    {
      "epoch": 1.9864406779661017,
      "grad_norm": 1.1418193578720093,
      "learning_rate": 1.6892655367231638e-05,
      "loss": 2.8203,
      "step": 2344
    },
    {
      "epoch": 1.9872881355932204,
      "grad_norm": 1.2710899114608765,
      "learning_rate": 1.687853107344633e-05,
      "loss": 2.7604,
      "step": 2345
    },
    {
      "epoch": 1.9881355932203388,
      "grad_norm": 1.2632701396942139,
      "learning_rate": 1.6864406779661018e-05,
      "loss": 2.7564,
      "step": 2346
    },
    {
      "epoch": 1.9889830508474575,
      "grad_norm": 0.9911966323852539,
      "learning_rate": 1.6850282485875708e-05,
      "loss": 2.9501,
      "step": 2347
    },
    {
      "epoch": 1.9898305084745762,
      "grad_norm": 1.1990536451339722,
      "learning_rate": 1.6836158192090394e-05,
      "loss": 2.6537,
      "step": 2348
    },
    {
      "epoch": 1.990677966101695,
      "grad_norm": 1.1346081495285034,
      "learning_rate": 1.6822033898305088e-05,
      "loss": 2.7558,
      "step": 2349
    },
    {
      "epoch": 1.9915254237288136,
      "grad_norm": 1.1950312852859497,
      "learning_rate": 1.6807909604519774e-05,
      "loss": 2.6231,
      "step": 2350
    },
    {
      "epoch": 1.9923728813559323,
      "grad_norm": 1.1291497945785522,
      "learning_rate": 1.6793785310734464e-05,
      "loss": 2.6095,
      "step": 2351
    },
    {
      "epoch": 1.993220338983051,
      "grad_norm": 1.2912720441818237,
      "learning_rate": 1.677966101694915e-05,
      "loss": 2.5782,
      "step": 2352
    },
    {
      "epoch": 1.9940677966101696,
      "grad_norm": 1.1213643550872803,
      "learning_rate": 1.6765536723163844e-05,
      "loss": 2.6206,
      "step": 2353
    },
    {
      "epoch": 1.994915254237288,
      "grad_norm": 1.0405361652374268,
      "learning_rate": 1.675141242937853e-05,
      "loss": 2.8375,
      "step": 2354
    },
    {
      "epoch": 1.9957627118644068,
      "grad_norm": 1.1870661973953247,
      "learning_rate": 1.673728813559322e-05,
      "loss": 2.6768,
      "step": 2355
    },
    {
      "epoch": 1.9966101694915255,
      "grad_norm": 1.1844818592071533,
      "learning_rate": 1.672316384180791e-05,
      "loss": 2.7363,
      "step": 2356
    },
    {
      "epoch": 1.997457627118644,
      "grad_norm": 1.2449629306793213,
      "learning_rate": 1.67090395480226e-05,
      "loss": 2.7136,
      "step": 2357
    },
    {
      "epoch": 1.9983050847457626,
      "grad_norm": 0.9491821527481079,
      "learning_rate": 1.669491525423729e-05,
      "loss": 2.8445,
      "step": 2358
    },
    {
      "epoch": 1.9991525423728813,
      "grad_norm": 1.2417043447494507,
      "learning_rate": 1.6680790960451977e-05,
      "loss": 2.7803,
      "step": 2359
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.399558663368225,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 2.5864,
      "step": 2360
    },
    {
      "epoch": 2.0008474576271187,
      "grad_norm": 1.954841136932373,
      "learning_rate": 1.6652542372881357e-05,
      "loss": 2.4323,
      "step": 2361
    },
    {
      "epoch": 2.0016949152542374,
      "grad_norm": 1.9575591087341309,
      "learning_rate": 1.6638418079096047e-05,
      "loss": 2.2182,
      "step": 2362
    },
    {
      "epoch": 2.002542372881356,
      "grad_norm": 1.880661964416504,
      "learning_rate": 1.6624293785310734e-05,
      "loss": 2.3444,
      "step": 2363
    },
    {
      "epoch": 2.0033898305084747,
      "grad_norm": 2.620041847229004,
      "learning_rate": 1.6610169491525424e-05,
      "loss": 2.0165,
      "step": 2364
    },
    {
      "epoch": 2.0042372881355934,
      "grad_norm": 1.9008338451385498,
      "learning_rate": 1.6596045197740114e-05,
      "loss": 2.3014,
      "step": 2365
    },
    {
      "epoch": 2.0050847457627117,
      "grad_norm": 1.9222924709320068,
      "learning_rate": 1.6581920903954804e-05,
      "loss": 2.4128,
      "step": 2366
    },
    {
      "epoch": 2.0059322033898304,
      "grad_norm": 1.6511833667755127,
      "learning_rate": 1.656779661016949e-05,
      "loss": 2.3422,
      "step": 2367
    },
    {
      "epoch": 2.006779661016949,
      "grad_norm": 1.7319014072418213,
      "learning_rate": 1.655367231638418e-05,
      "loss": 2.0119,
      "step": 2368
    },
    {
      "epoch": 2.0076271186440677,
      "grad_norm": 1.5663323402404785,
      "learning_rate": 1.653954802259887e-05,
      "loss": 2.111,
      "step": 2369
    },
    {
      "epoch": 2.0084745762711864,
      "grad_norm": 2.0640721321105957,
      "learning_rate": 1.652542372881356e-05,
      "loss": 1.8947,
      "step": 2370
    },
    {
      "epoch": 2.009322033898305,
      "grad_norm": 1.609630823135376,
      "learning_rate": 1.651129943502825e-05,
      "loss": 2.1651,
      "step": 2371
    },
    {
      "epoch": 2.010169491525424,
      "grad_norm": 1.3497323989868164,
      "learning_rate": 1.6497175141242937e-05,
      "loss": 2.6179,
      "step": 2372
    },
    {
      "epoch": 2.0110169491525425,
      "grad_norm": 2.0327656269073486,
      "learning_rate": 1.648305084745763e-05,
      "loss": 1.9496,
      "step": 2373
    },
    {
      "epoch": 2.011864406779661,
      "grad_norm": 1.9299832582473755,
      "learning_rate": 1.6468926553672317e-05,
      "loss": 2.1082,
      "step": 2374
    },
    {
      "epoch": 2.01271186440678,
      "grad_norm": 2.3567018508911133,
      "learning_rate": 1.6454802259887007e-05,
      "loss": 1.81,
      "step": 2375
    },
    {
      "epoch": 2.013559322033898,
      "grad_norm": 1.9467577934265137,
      "learning_rate": 1.6440677966101694e-05,
      "loss": 2.2086,
      "step": 2376
    },
    {
      "epoch": 2.0144067796610168,
      "grad_norm": 2.413874864578247,
      "learning_rate": 1.6426553672316387e-05,
      "loss": 2.0452,
      "step": 2377
    },
    {
      "epoch": 2.0152542372881355,
      "grad_norm": 2.1449880599975586,
      "learning_rate": 1.6412429378531074e-05,
      "loss": 2.3787,
      "step": 2378
    },
    {
      "epoch": 2.016101694915254,
      "grad_norm": 2.0846853256225586,
      "learning_rate": 1.6398305084745764e-05,
      "loss": 1.9526,
      "step": 2379
    },
    {
      "epoch": 2.016949152542373,
      "grad_norm": 2.0826375484466553,
      "learning_rate": 1.638418079096045e-05,
      "loss": 2.0344,
      "step": 2380
    },
    {
      "epoch": 2.0177966101694915,
      "grad_norm": 2.189852237701416,
      "learning_rate": 1.6370056497175144e-05,
      "loss": 1.6708,
      "step": 2381
    },
    {
      "epoch": 2.01864406779661,
      "grad_norm": 1.6226427555084229,
      "learning_rate": 1.635593220338983e-05,
      "loss": 2.3535,
      "step": 2382
    },
    {
      "epoch": 2.019491525423729,
      "grad_norm": 1.3864531517028809,
      "learning_rate": 1.634180790960452e-05,
      "loss": 2.4154,
      "step": 2383
    },
    {
      "epoch": 2.0203389830508476,
      "grad_norm": 1.4276227951049805,
      "learning_rate": 1.6327683615819207e-05,
      "loss": 2.2698,
      "step": 2384
    },
    {
      "epoch": 2.0211864406779663,
      "grad_norm": 1.8648180961608887,
      "learning_rate": 1.63135593220339e-05,
      "loss": 1.7603,
      "step": 2385
    },
    {
      "epoch": 2.022033898305085,
      "grad_norm": 1.316807746887207,
      "learning_rate": 1.6299435028248587e-05,
      "loss": 2.3367,
      "step": 2386
    },
    {
      "epoch": 2.022881355932203,
      "grad_norm": 2.0323164463043213,
      "learning_rate": 1.6285310734463277e-05,
      "loss": 1.7487,
      "step": 2387
    },
    {
      "epoch": 2.023728813559322,
      "grad_norm": 1.241205096244812,
      "learning_rate": 1.6271186440677967e-05,
      "loss": 2.426,
      "step": 2388
    },
    {
      "epoch": 2.0245762711864406,
      "grad_norm": 1.3961604833602905,
      "learning_rate": 1.6257062146892657e-05,
      "loss": 2.1473,
      "step": 2389
    },
    {
      "epoch": 2.0254237288135593,
      "grad_norm": 1.5464637279510498,
      "learning_rate": 1.6242937853107347e-05,
      "loss": 2.1547,
      "step": 2390
    },
    {
      "epoch": 2.026271186440678,
      "grad_norm": 1.2724565267562866,
      "learning_rate": 1.6228813559322034e-05,
      "loss": 2.3671,
      "step": 2391
    },
    {
      "epoch": 2.0271186440677966,
      "grad_norm": 1.15665602684021,
      "learning_rate": 1.6214689265536724e-05,
      "loss": 2.3736,
      "step": 2392
    },
    {
      "epoch": 2.0279661016949153,
      "grad_norm": 1.5079132318496704,
      "learning_rate": 1.6200564971751414e-05,
      "loss": 2.0668,
      "step": 2393
    },
    {
      "epoch": 2.028813559322034,
      "grad_norm": 1.650124430656433,
      "learning_rate": 1.6186440677966104e-05,
      "loss": 1.9998,
      "step": 2394
    },
    {
      "epoch": 2.0296610169491527,
      "grad_norm": 1.4562724828720093,
      "learning_rate": 1.617231638418079e-05,
      "loss": 2.3713,
      "step": 2395
    },
    {
      "epoch": 2.0305084745762714,
      "grad_norm": 1.5359783172607422,
      "learning_rate": 1.615819209039548e-05,
      "loss": 2.2703,
      "step": 2396
    },
    {
      "epoch": 2.0313559322033896,
      "grad_norm": 1.863041877746582,
      "learning_rate": 1.614406779661017e-05,
      "loss": 2.0951,
      "step": 2397
    },
    {
      "epoch": 2.0322033898305083,
      "grad_norm": 1.913000226020813,
      "learning_rate": 1.612994350282486e-05,
      "loss": 1.8643,
      "step": 2398
    },
    {
      "epoch": 2.033050847457627,
      "grad_norm": 1.6459336280822754,
      "learning_rate": 1.6115819209039547e-05,
      "loss": 2.2146,
      "step": 2399
    },
    {
      "epoch": 2.0338983050847457,
      "grad_norm": 1.598587989807129,
      "learning_rate": 1.6101694915254237e-05,
      "loss": 2.2268,
      "step": 2400
    },
    {
      "epoch": 2.0347457627118644,
      "grad_norm": 1.9659417867660522,
      "learning_rate": 1.6087570621468927e-05,
      "loss": 2.0437,
      "step": 2401
    },
    {
      "epoch": 2.035593220338983,
      "grad_norm": 1.3990250825881958,
      "learning_rate": 1.6073446327683617e-05,
      "loss": 2.1897,
      "step": 2402
    },
    {
      "epoch": 2.0364406779661017,
      "grad_norm": 1.0709125995635986,
      "learning_rate": 1.6059322033898307e-05,
      "loss": 2.4127,
      "step": 2403
    },
    {
      "epoch": 2.0372881355932204,
      "grad_norm": 1.5688316822052002,
      "learning_rate": 1.6045197740112993e-05,
      "loss": 2.1076,
      "step": 2404
    },
    {
      "epoch": 2.038135593220339,
      "grad_norm": 1.3257402181625366,
      "learning_rate": 1.6031073446327687e-05,
      "loss": 2.3743,
      "step": 2405
    },
    {
      "epoch": 2.038983050847458,
      "grad_norm": 1.7270565032958984,
      "learning_rate": 1.6016949152542373e-05,
      "loss": 2.0426,
      "step": 2406
    },
    {
      "epoch": 2.0398305084745765,
      "grad_norm": 1.4624197483062744,
      "learning_rate": 1.6002824858757063e-05,
      "loss": 2.0831,
      "step": 2407
    },
    {
      "epoch": 2.0406779661016947,
      "grad_norm": 2.149177312850952,
      "learning_rate": 1.598870056497175e-05,
      "loss": 1.8567,
      "step": 2408
    },
    {
      "epoch": 2.0415254237288134,
      "grad_norm": 1.4796191453933716,
      "learning_rate": 1.5974576271186443e-05,
      "loss": 2.3493,
      "step": 2409
    },
    {
      "epoch": 2.042372881355932,
      "grad_norm": 1.7318049669265747,
      "learning_rate": 1.596045197740113e-05,
      "loss": 2.0822,
      "step": 2410
    },
    {
      "epoch": 2.0432203389830508,
      "grad_norm": 1.522325038909912,
      "learning_rate": 1.594632768361582e-05,
      "loss": 2.1485,
      "step": 2411
    },
    {
      "epoch": 2.0440677966101695,
      "grad_norm": 1.7488744258880615,
      "learning_rate": 1.5932203389830507e-05,
      "loss": 2.0617,
      "step": 2412
    },
    {
      "epoch": 2.044915254237288,
      "grad_norm": 1.9676945209503174,
      "learning_rate": 1.59180790960452e-05,
      "loss": 1.8596,
      "step": 2413
    },
    {
      "epoch": 2.045762711864407,
      "grad_norm": 1.7921208143234253,
      "learning_rate": 1.5903954802259886e-05,
      "loss": 2.1513,
      "step": 2414
    },
    {
      "epoch": 2.0466101694915255,
      "grad_norm": 1.465053915977478,
      "learning_rate": 1.5889830508474576e-05,
      "loss": 2.0812,
      "step": 2415
    },
    {
      "epoch": 2.047457627118644,
      "grad_norm": 2.2215895652770996,
      "learning_rate": 1.5875706214689266e-05,
      "loss": 2.2899,
      "step": 2416
    },
    {
      "epoch": 2.048305084745763,
      "grad_norm": 1.323302984237671,
      "learning_rate": 1.5861581920903956e-05,
      "loss": 2.3655,
      "step": 2417
    },
    {
      "epoch": 2.049152542372881,
      "grad_norm": 1.572412371635437,
      "learning_rate": 1.5847457627118646e-05,
      "loss": 2.0031,
      "step": 2418
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5636740922927856,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 2.2966,
      "step": 2419
    },
    {
      "epoch": 2.0508474576271185,
      "grad_norm": 1.3185337781906128,
      "learning_rate": 1.5819209039548023e-05,
      "loss": 2.2967,
      "step": 2420
    },
    {
      "epoch": 2.051694915254237,
      "grad_norm": 1.592160940170288,
      "learning_rate": 1.5805084745762713e-05,
      "loss": 2.2037,
      "step": 2421
    },
    {
      "epoch": 2.052542372881356,
      "grad_norm": 2.0628747940063477,
      "learning_rate": 1.5790960451977403e-05,
      "loss": 1.652,
      "step": 2422
    },
    {
      "epoch": 2.0533898305084746,
      "grad_norm": 2.0035552978515625,
      "learning_rate": 1.577683615819209e-05,
      "loss": 2.0703,
      "step": 2423
    },
    {
      "epoch": 2.0542372881355933,
      "grad_norm": 1.6242327690124512,
      "learning_rate": 1.5762711864406783e-05,
      "loss": 2.1453,
      "step": 2424
    },
    {
      "epoch": 2.055084745762712,
      "grad_norm": 1.9367742538452148,
      "learning_rate": 1.574858757062147e-05,
      "loss": 2.0133,
      "step": 2425
    },
    {
      "epoch": 2.0559322033898306,
      "grad_norm": 1.5298945903778076,
      "learning_rate": 1.573446327683616e-05,
      "loss": 2.0661,
      "step": 2426
    },
    {
      "epoch": 2.0567796610169493,
      "grad_norm": 1.9443947076797485,
      "learning_rate": 1.5720338983050846e-05,
      "loss": 1.7775,
      "step": 2427
    },
    {
      "epoch": 2.057627118644068,
      "grad_norm": 1.8264436721801758,
      "learning_rate": 1.570621468926554e-05,
      "loss": 1.694,
      "step": 2428
    },
    {
      "epoch": 2.0584745762711862,
      "grad_norm": 1.9748023748397827,
      "learning_rate": 1.5692090395480226e-05,
      "loss": 1.6182,
      "step": 2429
    },
    {
      "epoch": 2.059322033898305,
      "grad_norm": 1.5430885553359985,
      "learning_rate": 1.5677966101694916e-05,
      "loss": 2.3806,
      "step": 2430
    },
    {
      "epoch": 2.0601694915254236,
      "grad_norm": 1.8126107454299927,
      "learning_rate": 1.5663841807909603e-05,
      "loss": 1.9787,
      "step": 2431
    },
    {
      "epoch": 2.0610169491525423,
      "grad_norm": 2.0328750610351562,
      "learning_rate": 1.5649717514124296e-05,
      "loss": 1.864,
      "step": 2432
    },
    {
      "epoch": 2.061864406779661,
      "grad_norm": 1.8950706720352173,
      "learning_rate": 1.5635593220338983e-05,
      "loss": 1.7348,
      "step": 2433
    },
    {
      "epoch": 2.0627118644067797,
      "grad_norm": 1.7175761461257935,
      "learning_rate": 1.5621468926553673e-05,
      "loss": 2.2092,
      "step": 2434
    },
    {
      "epoch": 2.0635593220338984,
      "grad_norm": 1.7328435182571411,
      "learning_rate": 1.5607344632768363e-05,
      "loss": 1.9453,
      "step": 2435
    },
    {
      "epoch": 2.064406779661017,
      "grad_norm": 1.4919472932815552,
      "learning_rate": 1.5593220338983053e-05,
      "loss": 2.2208,
      "step": 2436
    },
    {
      "epoch": 2.0652542372881357,
      "grad_norm": 1.7060024738311768,
      "learning_rate": 1.5579096045197743e-05,
      "loss": 2.1301,
      "step": 2437
    },
    {
      "epoch": 2.0661016949152544,
      "grad_norm": 1.7842141389846802,
      "learning_rate": 1.556497175141243e-05,
      "loss": 2.0364,
      "step": 2438
    },
    {
      "epoch": 2.0669491525423727,
      "grad_norm": 1.7225085496902466,
      "learning_rate": 1.555084745762712e-05,
      "loss": 2.2352,
      "step": 2439
    },
    {
      "epoch": 2.0677966101694913,
      "grad_norm": 1.8794629573822021,
      "learning_rate": 1.553672316384181e-05,
      "loss": 2.077,
      "step": 2440
    },
    {
      "epoch": 2.06864406779661,
      "grad_norm": 2.137139320373535,
      "learning_rate": 1.55225988700565e-05,
      "loss": 1.9319,
      "step": 2441
    },
    {
      "epoch": 2.0694915254237287,
      "grad_norm": 1.8608564138412476,
      "learning_rate": 1.5508474576271186e-05,
      "loss": 1.8859,
      "step": 2442
    },
    {
      "epoch": 2.0703389830508474,
      "grad_norm": 1.515536904335022,
      "learning_rate": 1.5494350282485876e-05,
      "loss": 2.2575,
      "step": 2443
    },
    {
      "epoch": 2.071186440677966,
      "grad_norm": 1.615857481956482,
      "learning_rate": 1.5480225988700566e-05,
      "loss": 2.242,
      "step": 2444
    },
    {
      "epoch": 2.0720338983050848,
      "grad_norm": 1.7035619020462036,
      "learning_rate": 1.5466101694915256e-05,
      "loss": 2.0534,
      "step": 2445
    },
    {
      "epoch": 2.0728813559322035,
      "grad_norm": 1.9742575883865356,
      "learning_rate": 1.5451977401129943e-05,
      "loss": 1.8712,
      "step": 2446
    },
    {
      "epoch": 2.073728813559322,
      "grad_norm": 1.5173736810684204,
      "learning_rate": 1.5437853107344633e-05,
      "loss": 2.2006,
      "step": 2447
    },
    {
      "epoch": 2.074576271186441,
      "grad_norm": 1.8541594743728638,
      "learning_rate": 1.5423728813559323e-05,
      "loss": 2.1028,
      "step": 2448
    },
    {
      "epoch": 2.0754237288135595,
      "grad_norm": 1.481216549873352,
      "learning_rate": 1.5409604519774013e-05,
      "loss": 2.3778,
      "step": 2449
    },
    {
      "epoch": 2.0762711864406778,
      "grad_norm": 1.506359577178955,
      "learning_rate": 1.5395480225988703e-05,
      "loss": 2.2367,
      "step": 2450
    },
    {
      "epoch": 2.0771186440677964,
      "grad_norm": 1.9489935636520386,
      "learning_rate": 1.538135593220339e-05,
      "loss": 2.0066,
      "step": 2451
    },
    {
      "epoch": 2.077966101694915,
      "grad_norm": 1.7360435724258423,
      "learning_rate": 1.5367231638418083e-05,
      "loss": 2.1559,
      "step": 2452
    },
    {
      "epoch": 2.078813559322034,
      "grad_norm": 1.4112789630889893,
      "learning_rate": 1.535310734463277e-05,
      "loss": 2.2551,
      "step": 2453
    },
    {
      "epoch": 2.0796610169491525,
      "grad_norm": 1.9627602100372314,
      "learning_rate": 1.533898305084746e-05,
      "loss": 2.0402,
      "step": 2454
    },
    {
      "epoch": 2.080508474576271,
      "grad_norm": 1.8383172750473022,
      "learning_rate": 1.5324858757062146e-05,
      "loss": 2.1307,
      "step": 2455
    },
    {
      "epoch": 2.08135593220339,
      "grad_norm": 1.5599380731582642,
      "learning_rate": 1.531073446327684e-05,
      "loss": 2.3239,
      "step": 2456
    },
    {
      "epoch": 2.0822033898305086,
      "grad_norm": 2.331843852996826,
      "learning_rate": 1.5296610169491526e-05,
      "loss": 1.7808,
      "step": 2457
    },
    {
      "epoch": 2.0830508474576273,
      "grad_norm": 2.7220191955566406,
      "learning_rate": 1.5282485875706216e-05,
      "loss": 1.7593,
      "step": 2458
    },
    {
      "epoch": 2.083898305084746,
      "grad_norm": 1.6436541080474854,
      "learning_rate": 1.5268361581920902e-05,
      "loss": 2.1357,
      "step": 2459
    },
    {
      "epoch": 2.084745762711864,
      "grad_norm": 1.0616300106048584,
      "learning_rate": 1.5254237288135596e-05,
      "loss": 2.6704,
      "step": 2460
    },
    {
      "epoch": 2.085593220338983,
      "grad_norm": 1.5112767219543457,
      "learning_rate": 1.5240112994350284e-05,
      "loss": 2.3113,
      "step": 2461
    },
    {
      "epoch": 2.0864406779661016,
      "grad_norm": 1.6891629695892334,
      "learning_rate": 1.5225988700564972e-05,
      "loss": 1.9751,
      "step": 2462
    },
    {
      "epoch": 2.0872881355932202,
      "grad_norm": 1.5620174407958984,
      "learning_rate": 1.521186440677966e-05,
      "loss": 2.2542,
      "step": 2463
    },
    {
      "epoch": 2.088135593220339,
      "grad_norm": 1.851977825164795,
      "learning_rate": 1.5197740112994352e-05,
      "loss": 1.9648,
      "step": 2464
    },
    {
      "epoch": 2.0889830508474576,
      "grad_norm": 1.4661450386047363,
      "learning_rate": 1.518361581920904e-05,
      "loss": 2.1972,
      "step": 2465
    },
    {
      "epoch": 2.0898305084745763,
      "grad_norm": 1.6566978693008423,
      "learning_rate": 1.5169491525423729e-05,
      "loss": 2.2925,
      "step": 2466
    },
    {
      "epoch": 2.090677966101695,
      "grad_norm": 1.2233691215515137,
      "learning_rate": 1.5155367231638417e-05,
      "loss": 2.4515,
      "step": 2467
    },
    {
      "epoch": 2.0915254237288137,
      "grad_norm": 2.050008773803711,
      "learning_rate": 1.5141242937853109e-05,
      "loss": 1.6345,
      "step": 2468
    },
    {
      "epoch": 2.0923728813559324,
      "grad_norm": 1.3644006252288818,
      "learning_rate": 1.5127118644067797e-05,
      "loss": 2.2742,
      "step": 2469
    },
    {
      "epoch": 2.093220338983051,
      "grad_norm": 1.8843778371810913,
      "learning_rate": 1.5112994350282486e-05,
      "loss": 2.1363,
      "step": 2470
    },
    {
      "epoch": 2.0940677966101693,
      "grad_norm": 1.8734453916549683,
      "learning_rate": 1.5098870056497176e-05,
      "loss": 1.9785,
      "step": 2471
    },
    {
      "epoch": 2.094915254237288,
      "grad_norm": 1.614042043685913,
      "learning_rate": 1.5084745762711865e-05,
      "loss": 2.2247,
      "step": 2472
    },
    {
      "epoch": 2.0957627118644067,
      "grad_norm": 1.8866091966629028,
      "learning_rate": 1.5070621468926554e-05,
      "loss": 2.0338,
      "step": 2473
    },
    {
      "epoch": 2.0966101694915253,
      "grad_norm": 1.4197208881378174,
      "learning_rate": 1.5056497175141244e-05,
      "loss": 2.3567,
      "step": 2474
    },
    {
      "epoch": 2.097457627118644,
      "grad_norm": 1.4004616737365723,
      "learning_rate": 1.5042372881355932e-05,
      "loss": 2.2448,
      "step": 2475
    },
    {
      "epoch": 2.0983050847457627,
      "grad_norm": 1.7236913442611694,
      "learning_rate": 1.5028248587570624e-05,
      "loss": 1.9488,
      "step": 2476
    },
    {
      "epoch": 2.0991525423728814,
      "grad_norm": 1.5777957439422607,
      "learning_rate": 1.5014124293785312e-05,
      "loss": 2.1542,
      "step": 2477
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8451133966445923,
      "learning_rate": 1.5e-05,
      "loss": 2.1358,
      "step": 2478
    },
    {
      "epoch": 2.1008474576271188,
      "grad_norm": 3.029209852218628,
      "learning_rate": 1.4985875706214689e-05,
      "loss": 2.0149,
      "step": 2479
    },
    {
      "epoch": 2.1016949152542375,
      "grad_norm": 1.345157265663147,
      "learning_rate": 1.497175141242938e-05,
      "loss": 2.4342,
      "step": 2480
    },
    {
      "epoch": 2.1025423728813557,
      "grad_norm": 1.496124267578125,
      "learning_rate": 1.4957627118644069e-05,
      "loss": 2.1712,
      "step": 2481
    },
    {
      "epoch": 2.1033898305084744,
      "grad_norm": 1.9682401418685913,
      "learning_rate": 1.4943502824858757e-05,
      "loss": 1.7816,
      "step": 2482
    },
    {
      "epoch": 2.104237288135593,
      "grad_norm": 1.799791693687439,
      "learning_rate": 1.4929378531073445e-05,
      "loss": 1.8135,
      "step": 2483
    },
    {
      "epoch": 2.1050847457627118,
      "grad_norm": 1.8387713432312012,
      "learning_rate": 1.4915254237288137e-05,
      "loss": 1.9054,
      "step": 2484
    },
    {
      "epoch": 2.1059322033898304,
      "grad_norm": 1.6758382320404053,
      "learning_rate": 1.4901129943502825e-05,
      "loss": 2.3163,
      "step": 2485
    },
    {
      "epoch": 2.106779661016949,
      "grad_norm": 1.6890454292297363,
      "learning_rate": 1.4887005649717514e-05,
      "loss": 2.1173,
      "step": 2486
    },
    {
      "epoch": 2.107627118644068,
      "grad_norm": 1.4125128984451294,
      "learning_rate": 1.4872881355932204e-05,
      "loss": 2.2514,
      "step": 2487
    },
    {
      "epoch": 2.1084745762711865,
      "grad_norm": 1.3438292741775513,
      "learning_rate": 1.4858757062146894e-05,
      "loss": 2.4917,
      "step": 2488
    },
    {
      "epoch": 2.109322033898305,
      "grad_norm": 2.0984585285186768,
      "learning_rate": 1.4844632768361582e-05,
      "loss": 1.6638,
      "step": 2489
    },
    {
      "epoch": 2.110169491525424,
      "grad_norm": 1.9447909593582153,
      "learning_rate": 1.4830508474576272e-05,
      "loss": 1.9094,
      "step": 2490
    },
    {
      "epoch": 2.1110169491525426,
      "grad_norm": 1.397506833076477,
      "learning_rate": 1.481638418079096e-05,
      "loss": 2.385,
      "step": 2491
    },
    {
      "epoch": 2.111864406779661,
      "grad_norm": 1.4305623769760132,
      "learning_rate": 1.4802259887005652e-05,
      "loss": 2.5347,
      "step": 2492
    },
    {
      "epoch": 2.1127118644067795,
      "grad_norm": 1.444412350654602,
      "learning_rate": 1.478813559322034e-05,
      "loss": 2.2904,
      "step": 2493
    },
    {
      "epoch": 2.113559322033898,
      "grad_norm": 1.458781361579895,
      "learning_rate": 1.4774011299435028e-05,
      "loss": 2.3122,
      "step": 2494
    },
    {
      "epoch": 2.114406779661017,
      "grad_norm": 1.4855881929397583,
      "learning_rate": 1.475988700564972e-05,
      "loss": 2.3321,
      "step": 2495
    },
    {
      "epoch": 2.1152542372881356,
      "grad_norm": 1.5247066020965576,
      "learning_rate": 1.4745762711864408e-05,
      "loss": 2.1949,
      "step": 2496
    },
    {
      "epoch": 2.1161016949152542,
      "grad_norm": 1.168670415878296,
      "learning_rate": 1.4731638418079097e-05,
      "loss": 2.6419,
      "step": 2497
    },
    {
      "epoch": 2.116949152542373,
      "grad_norm": 1.9894770383834839,
      "learning_rate": 1.4717514124293785e-05,
      "loss": 1.7941,
      "step": 2498
    },
    {
      "epoch": 2.1177966101694916,
      "grad_norm": 1.8084489107131958,
      "learning_rate": 1.4703389830508477e-05,
      "loss": 1.9211,
      "step": 2499
    },
    {
      "epoch": 2.1186440677966103,
      "grad_norm": 1.9926161766052246,
      "learning_rate": 1.4689265536723165e-05,
      "loss": 1.9123,
      "step": 2500
    },
    {
      "epoch": 2.119491525423729,
      "grad_norm": 1.8883634805679321,
      "learning_rate": 1.4675141242937853e-05,
      "loss": 1.905,
      "step": 2501
    },
    {
      "epoch": 2.1203389830508477,
      "grad_norm": 2.07058048248291,
      "learning_rate": 1.4661016949152542e-05,
      "loss": 1.9945,
      "step": 2502
    },
    {
      "epoch": 2.121186440677966,
      "grad_norm": 1.820389747619629,
      "learning_rate": 1.4646892655367233e-05,
      "loss": 2.0952,
      "step": 2503
    },
    {
      "epoch": 2.1220338983050846,
      "grad_norm": 1.6159000396728516,
      "learning_rate": 1.4632768361581922e-05,
      "loss": 2.0177,
      "step": 2504
    },
    {
      "epoch": 2.1228813559322033,
      "grad_norm": 1.8930219411849976,
      "learning_rate": 1.4618644067796612e-05,
      "loss": 2.0866,
      "step": 2505
    },
    {
      "epoch": 2.123728813559322,
      "grad_norm": 1.8374148607254028,
      "learning_rate": 1.46045197740113e-05,
      "loss": 1.938,
      "step": 2506
    },
    {
      "epoch": 2.1245762711864407,
      "grad_norm": 1.598009467124939,
      "learning_rate": 1.459039548022599e-05,
      "loss": 2.299,
      "step": 2507
    },
    {
      "epoch": 2.1254237288135593,
      "grad_norm": 1.8127394914627075,
      "learning_rate": 1.457627118644068e-05,
      "loss": 2.1198,
      "step": 2508
    },
    {
      "epoch": 2.126271186440678,
      "grad_norm": 1.3069536685943604,
      "learning_rate": 1.4562146892655368e-05,
      "loss": 2.4244,
      "step": 2509
    },
    {
      "epoch": 2.1271186440677967,
      "grad_norm": 1.4885691404342651,
      "learning_rate": 1.4548022598870056e-05,
      "loss": 2.1462,
      "step": 2510
    },
    {
      "epoch": 2.1279661016949154,
      "grad_norm": 1.642301321029663,
      "learning_rate": 1.4533898305084748e-05,
      "loss": 2.1435,
      "step": 2511
    },
    {
      "epoch": 2.128813559322034,
      "grad_norm": 1.5486087799072266,
      "learning_rate": 1.4519774011299436e-05,
      "loss": 2.2388,
      "step": 2512
    },
    {
      "epoch": 2.1296610169491528,
      "grad_norm": 1.9132736921310425,
      "learning_rate": 1.4505649717514125e-05,
      "loss": 1.9723,
      "step": 2513
    },
    {
      "epoch": 2.130508474576271,
      "grad_norm": 1.6700730323791504,
      "learning_rate": 1.4491525423728813e-05,
      "loss": 2.1479,
      "step": 2514
    },
    {
      "epoch": 2.1313559322033897,
      "grad_norm": 1.5021833181381226,
      "learning_rate": 1.4477401129943505e-05,
      "loss": 2.3397,
      "step": 2515
    },
    {
      "epoch": 2.1322033898305084,
      "grad_norm": 2.378293991088867,
      "learning_rate": 1.4463276836158193e-05,
      "loss": 2.3639,
      "step": 2516
    },
    {
      "epoch": 2.133050847457627,
      "grad_norm": 1.5428694486618042,
      "learning_rate": 1.4449152542372881e-05,
      "loss": 2.2182,
      "step": 2517
    },
    {
      "epoch": 2.1338983050847458,
      "grad_norm": 1.633458137512207,
      "learning_rate": 1.443502824858757e-05,
      "loss": 2.3229,
      "step": 2518
    },
    {
      "epoch": 2.1347457627118644,
      "grad_norm": 1.5148998498916626,
      "learning_rate": 1.4420903954802261e-05,
      "loss": 2.2358,
      "step": 2519
    },
    {
      "epoch": 2.135593220338983,
      "grad_norm": 1.5872907638549805,
      "learning_rate": 1.440677966101695e-05,
      "loss": 2.2053,
      "step": 2520
    },
    {
      "epoch": 2.136440677966102,
      "grad_norm": 1.6889498233795166,
      "learning_rate": 1.439265536723164e-05,
      "loss": 2.2384,
      "step": 2521
    },
    {
      "epoch": 2.1372881355932205,
      "grad_norm": 1.6237796545028687,
      "learning_rate": 1.4378531073446328e-05,
      "loss": 2.2532,
      "step": 2522
    },
    {
      "epoch": 2.1381355932203387,
      "grad_norm": 1.9023722410202026,
      "learning_rate": 1.4364406779661018e-05,
      "loss": 1.9404,
      "step": 2523
    },
    {
      "epoch": 2.1389830508474574,
      "grad_norm": 1.763069987297058,
      "learning_rate": 1.4350282485875708e-05,
      "loss": 2.0033,
      "step": 2524
    },
    {
      "epoch": 2.139830508474576,
      "grad_norm": 1.5526543855667114,
      "learning_rate": 1.4336158192090396e-05,
      "loss": 2.0892,
      "step": 2525
    },
    {
      "epoch": 2.140677966101695,
      "grad_norm": 1.6335585117340088,
      "learning_rate": 1.4322033898305085e-05,
      "loss": 1.9705,
      "step": 2526
    },
    {
      "epoch": 2.1415254237288135,
      "grad_norm": 1.6628258228302002,
      "learning_rate": 1.4307909604519776e-05,
      "loss": 2.1448,
      "step": 2527
    },
    {
      "epoch": 2.142372881355932,
      "grad_norm": 1.4387049674987793,
      "learning_rate": 1.4293785310734465e-05,
      "loss": 2.2944,
      "step": 2528
    },
    {
      "epoch": 2.143220338983051,
      "grad_norm": 1.481052279472351,
      "learning_rate": 1.4279661016949153e-05,
      "loss": 2.2219,
      "step": 2529
    },
    {
      "epoch": 2.1440677966101696,
      "grad_norm": 1.803296446800232,
      "learning_rate": 1.4265536723163841e-05,
      "loss": 2.1046,
      "step": 2530
    },
    {
      "epoch": 2.1449152542372882,
      "grad_norm": 1.8609411716461182,
      "learning_rate": 1.4251412429378533e-05,
      "loss": 2.0763,
      "step": 2531
    },
    {
      "epoch": 2.145762711864407,
      "grad_norm": 1.9128785133361816,
      "learning_rate": 1.4237288135593221e-05,
      "loss": 1.8958,
      "step": 2532
    },
    {
      "epoch": 2.1466101694915256,
      "grad_norm": 1.9034333229064941,
      "learning_rate": 1.422316384180791e-05,
      "loss": 1.8461,
      "step": 2533
    },
    {
      "epoch": 2.1474576271186443,
      "grad_norm": 2.5008606910705566,
      "learning_rate": 1.4209039548022598e-05,
      "loss": 1.91,
      "step": 2534
    },
    {
      "epoch": 2.1483050847457625,
      "grad_norm": 1.623769760131836,
      "learning_rate": 1.419491525423729e-05,
      "loss": 2.2586,
      "step": 2535
    },
    {
      "epoch": 2.1491525423728812,
      "grad_norm": 2.1108202934265137,
      "learning_rate": 1.4180790960451978e-05,
      "loss": 1.9027,
      "step": 2536
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9891767501831055,
      "learning_rate": 1.4166666666666668e-05,
      "loss": 1.9846,
      "step": 2537
    },
    {
      "epoch": 2.1508474576271186,
      "grad_norm": 2.1394753456115723,
      "learning_rate": 1.4152542372881356e-05,
      "loss": 1.9395,
      "step": 2538
    },
    {
      "epoch": 2.1516949152542373,
      "grad_norm": 1.662566900253296,
      "learning_rate": 1.4138418079096048e-05,
      "loss": 2.3018,
      "step": 2539
    },
    {
      "epoch": 2.152542372881356,
      "grad_norm": 2.159883737564087,
      "learning_rate": 1.4124293785310736e-05,
      "loss": 1.9256,
      "step": 2540
    },
    {
      "epoch": 2.1533898305084747,
      "grad_norm": 2.0187926292419434,
      "learning_rate": 1.4110169491525424e-05,
      "loss": 1.7855,
      "step": 2541
    },
    {
      "epoch": 2.1542372881355933,
      "grad_norm": 1.3911402225494385,
      "learning_rate": 1.4096045197740113e-05,
      "loss": 2.3972,
      "step": 2542
    },
    {
      "epoch": 2.155084745762712,
      "grad_norm": 2.866255283355713,
      "learning_rate": 1.4081920903954804e-05,
      "loss": 2.0278,
      "step": 2543
    },
    {
      "epoch": 2.1559322033898307,
      "grad_norm": 1.668233036994934,
      "learning_rate": 1.4067796610169493e-05,
      "loss": 2.1187,
      "step": 2544
    },
    {
      "epoch": 2.156779661016949,
      "grad_norm": 1.8213480710983276,
      "learning_rate": 1.4053672316384181e-05,
      "loss": 1.9344,
      "step": 2545
    },
    {
      "epoch": 2.1576271186440676,
      "grad_norm": 1.5447547435760498,
      "learning_rate": 1.403954802259887e-05,
      "loss": 2.2768,
      "step": 2546
    },
    {
      "epoch": 2.1584745762711863,
      "grad_norm": 1.3066322803497314,
      "learning_rate": 1.4025423728813561e-05,
      "loss": 2.427,
      "step": 2547
    },
    {
      "epoch": 2.159322033898305,
      "grad_norm": 1.8616938591003418,
      "learning_rate": 1.4011299435028249e-05,
      "loss": 1.9183,
      "step": 2548
    },
    {
      "epoch": 2.1601694915254237,
      "grad_norm": 1.8417625427246094,
      "learning_rate": 1.3997175141242937e-05,
      "loss": 1.982,
      "step": 2549
    },
    {
      "epoch": 2.1610169491525424,
      "grad_norm": 1.9708733558654785,
      "learning_rate": 1.3983050847457627e-05,
      "loss": 1.935,
      "step": 2550
    },
    {
      "epoch": 2.161864406779661,
      "grad_norm": 1.737199068069458,
      "learning_rate": 1.3968926553672317e-05,
      "loss": 2.0792,
      "step": 2551
    },
    {
      "epoch": 2.1627118644067798,
      "grad_norm": 1.598211646080017,
      "learning_rate": 1.3954802259887006e-05,
      "loss": 2.2252,
      "step": 2552
    },
    {
      "epoch": 2.1635593220338984,
      "grad_norm": 1.5978015661239624,
      "learning_rate": 1.3940677966101696e-05,
      "loss": 2.2673,
      "step": 2553
    },
    {
      "epoch": 2.164406779661017,
      "grad_norm": 1.3255168199539185,
      "learning_rate": 1.3926553672316384e-05,
      "loss": 2.3453,
      "step": 2554
    },
    {
      "epoch": 2.165254237288136,
      "grad_norm": 1.6760737895965576,
      "learning_rate": 1.3912429378531076e-05,
      "loss": 2.2258,
      "step": 2555
    },
    {
      "epoch": 2.166101694915254,
      "grad_norm": 1.4257606267929077,
      "learning_rate": 1.3898305084745764e-05,
      "loss": 2.2405,
      "step": 2556
    },
    {
      "epoch": 2.1669491525423727,
      "grad_norm": 1.9469870328903198,
      "learning_rate": 1.3884180790960452e-05,
      "loss": 1.8691,
      "step": 2557
    },
    {
      "epoch": 2.1677966101694914,
      "grad_norm": 1.497591495513916,
      "learning_rate": 1.387005649717514e-05,
      "loss": 2.2402,
      "step": 2558
    },
    {
      "epoch": 2.16864406779661,
      "grad_norm": 1.9147684574127197,
      "learning_rate": 1.3855932203389832e-05,
      "loss": 1.7802,
      "step": 2559
    },
    {
      "epoch": 2.169491525423729,
      "grad_norm": 1.671188235282898,
      "learning_rate": 1.384180790960452e-05,
      "loss": 2.2023,
      "step": 2560
    },
    {
      "epoch": 2.1703389830508475,
      "grad_norm": 2.085939407348633,
      "learning_rate": 1.3827683615819209e-05,
      "loss": 1.6894,
      "step": 2561
    },
    {
      "epoch": 2.171186440677966,
      "grad_norm": 2.1118597984313965,
      "learning_rate": 1.3813559322033897e-05,
      "loss": 1.8384,
      "step": 2562
    },
    {
      "epoch": 2.172033898305085,
      "grad_norm": 1.3560526371002197,
      "learning_rate": 1.3799435028248589e-05,
      "loss": 2.4099,
      "step": 2563
    },
    {
      "epoch": 2.1728813559322036,
      "grad_norm": 1.557410478591919,
      "learning_rate": 1.3785310734463277e-05,
      "loss": 2.3363,
      "step": 2564
    },
    {
      "epoch": 2.1737288135593222,
      "grad_norm": 2.623718738555908,
      "learning_rate": 1.3771186440677965e-05,
      "loss": 1.5711,
      "step": 2565
    },
    {
      "epoch": 2.1745762711864405,
      "grad_norm": 1.4253687858581543,
      "learning_rate": 1.3757062146892655e-05,
      "loss": 2.3093,
      "step": 2566
    },
    {
      "epoch": 2.175423728813559,
      "grad_norm": 1.5429052114486694,
      "learning_rate": 1.3742937853107345e-05,
      "loss": 2.3963,
      "step": 2567
    },
    {
      "epoch": 2.176271186440678,
      "grad_norm": 1.3952724933624268,
      "learning_rate": 1.3728813559322034e-05,
      "loss": 2.3189,
      "step": 2568
    },
    {
      "epoch": 2.1771186440677965,
      "grad_norm": 2.0680289268493652,
      "learning_rate": 1.3714689265536724e-05,
      "loss": 1.8086,
      "step": 2569
    },
    {
      "epoch": 2.1779661016949152,
      "grad_norm": 2.042335271835327,
      "learning_rate": 1.3700564971751412e-05,
      "loss": 1.8987,
      "step": 2570
    },
    {
      "epoch": 2.178813559322034,
      "grad_norm": 1.3315010070800781,
      "learning_rate": 1.3686440677966104e-05,
      "loss": 2.3551,
      "step": 2571
    },
    {
      "epoch": 2.1796610169491526,
      "grad_norm": 1.4842523336410522,
      "learning_rate": 1.3672316384180792e-05,
      "loss": 2.42,
      "step": 2572
    },
    {
      "epoch": 2.1805084745762713,
      "grad_norm": 2.355437994003296,
      "learning_rate": 1.365819209039548e-05,
      "loss": 1.3587,
      "step": 2573
    },
    {
      "epoch": 2.18135593220339,
      "grad_norm": 1.5268384218215942,
      "learning_rate": 1.3644067796610169e-05,
      "loss": 2.135,
      "step": 2574
    },
    {
      "epoch": 2.1822033898305087,
      "grad_norm": 1.2764055728912354,
      "learning_rate": 1.362994350282486e-05,
      "loss": 2.2816,
      "step": 2575
    },
    {
      "epoch": 2.1830508474576273,
      "grad_norm": 1.74439537525177,
      "learning_rate": 1.3615819209039549e-05,
      "loss": 2.0551,
      "step": 2576
    },
    {
      "epoch": 2.1838983050847456,
      "grad_norm": 1.5508873462677002,
      "learning_rate": 1.3601694915254237e-05,
      "loss": 2.0705,
      "step": 2577
    },
    {
      "epoch": 2.1847457627118643,
      "grad_norm": 1.7426412105560303,
      "learning_rate": 1.3587570621468929e-05,
      "loss": 2.1942,
      "step": 2578
    },
    {
      "epoch": 2.185593220338983,
      "grad_norm": 1.6260688304901123,
      "learning_rate": 1.3573446327683617e-05,
      "loss": 2.1463,
      "step": 2579
    },
    {
      "epoch": 2.1864406779661016,
      "grad_norm": 2.0990731716156006,
      "learning_rate": 1.3559322033898305e-05,
      "loss": 1.9374,
      "step": 2580
    },
    {
      "epoch": 2.1872881355932203,
      "grad_norm": 1.7750979661941528,
      "learning_rate": 1.3545197740112994e-05,
      "loss": 2.0562,
      "step": 2581
    },
    {
      "epoch": 2.188135593220339,
      "grad_norm": 1.866346001625061,
      "learning_rate": 1.3531073446327685e-05,
      "loss": 2.0816,
      "step": 2582
    },
    {
      "epoch": 2.1889830508474577,
      "grad_norm": 1.5054261684417725,
      "learning_rate": 1.3516949152542374e-05,
      "loss": 2.4755,
      "step": 2583
    },
    {
      "epoch": 2.1898305084745764,
      "grad_norm": 1.2178317308425903,
      "learning_rate": 1.3502824858757064e-05,
      "loss": 2.6567,
      "step": 2584
    },
    {
      "epoch": 2.190677966101695,
      "grad_norm": 1.5665068626403809,
      "learning_rate": 1.3488700564971752e-05,
      "loss": 2.2267,
      "step": 2585
    },
    {
      "epoch": 2.1915254237288138,
      "grad_norm": 2.063912868499756,
      "learning_rate": 1.3474576271186442e-05,
      "loss": 1.9713,
      "step": 2586
    },
    {
      "epoch": 2.192372881355932,
      "grad_norm": 1.6485923528671265,
      "learning_rate": 1.3460451977401132e-05,
      "loss": 2.1568,
      "step": 2587
    },
    {
      "epoch": 2.1932203389830507,
      "grad_norm": 1.222883939743042,
      "learning_rate": 1.344632768361582e-05,
      "loss": 2.5162,
      "step": 2588
    },
    {
      "epoch": 2.1940677966101694,
      "grad_norm": 1.8396320343017578,
      "learning_rate": 1.3432203389830508e-05,
      "loss": 1.8913,
      "step": 2589
    },
    {
      "epoch": 2.194915254237288,
      "grad_norm": 1.6970138549804688,
      "learning_rate": 1.34180790960452e-05,
      "loss": 1.767,
      "step": 2590
    },
    {
      "epoch": 2.1957627118644067,
      "grad_norm": 1.7239935398101807,
      "learning_rate": 1.3403954802259888e-05,
      "loss": 1.7974,
      "step": 2591
    },
    {
      "epoch": 2.1966101694915254,
      "grad_norm": 1.249467134475708,
      "learning_rate": 1.3389830508474577e-05,
      "loss": 2.3644,
      "step": 2592
    },
    {
      "epoch": 2.197457627118644,
      "grad_norm": 1.3214976787567139,
      "learning_rate": 1.3375706214689265e-05,
      "loss": 2.3804,
      "step": 2593
    },
    {
      "epoch": 2.198305084745763,
      "grad_norm": 1.844478964805603,
      "learning_rate": 1.3361581920903957e-05,
      "loss": 2.0592,
      "step": 2594
    },
    {
      "epoch": 2.1991525423728815,
      "grad_norm": 1.8242712020874023,
      "learning_rate": 1.3347457627118645e-05,
      "loss": 1.9523,
      "step": 2595
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3069279193878174,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 2.3387,
      "step": 2596
    },
    {
      "epoch": 2.200847457627119,
      "grad_norm": 1.7718732357025146,
      "learning_rate": 1.3319209039548022e-05,
      "loss": 2.3433,
      "step": 2597
    },
    {
      "epoch": 2.201694915254237,
      "grad_norm": 1.8796429634094238,
      "learning_rate": 1.3305084745762713e-05,
      "loss": 1.9381,
      "step": 2598
    },
    {
      "epoch": 2.202542372881356,
      "grad_norm": 1.8052655458450317,
      "learning_rate": 1.3290960451977402e-05,
      "loss": 2.1447,
      "step": 2599
    },
    {
      "epoch": 2.2033898305084745,
      "grad_norm": 2.3024721145629883,
      "learning_rate": 1.3276836158192092e-05,
      "loss": 1.614,
      "step": 2600
    },
    {
      "epoch": 2.204237288135593,
      "grad_norm": 1.120464563369751,
      "learning_rate": 1.326271186440678e-05,
      "loss": 2.7688,
      "step": 2601
    },
    {
      "epoch": 2.205084745762712,
      "grad_norm": 1.4297536611557007,
      "learning_rate": 1.324858757062147e-05,
      "loss": 2.2969,
      "step": 2602
    },
    {
      "epoch": 2.2059322033898305,
      "grad_norm": 1.6033844947814941,
      "learning_rate": 1.323446327683616e-05,
      "loss": 2.3372,
      "step": 2603
    },
    {
      "epoch": 2.2067796610169492,
      "grad_norm": 1.608870506286621,
      "learning_rate": 1.3220338983050848e-05,
      "loss": 2.1432,
      "step": 2604
    },
    {
      "epoch": 2.207627118644068,
      "grad_norm": 1.7524429559707642,
      "learning_rate": 1.3206214689265536e-05,
      "loss": 1.9324,
      "step": 2605
    },
    {
      "epoch": 2.2084745762711866,
      "grad_norm": 1.6078535318374634,
      "learning_rate": 1.3192090395480228e-05,
      "loss": 2.2075,
      "step": 2606
    },
    {
      "epoch": 2.2093220338983053,
      "grad_norm": 2.2441558837890625,
      "learning_rate": 1.3177966101694916e-05,
      "loss": 1.624,
      "step": 2607
    },
    {
      "epoch": 2.2101694915254235,
      "grad_norm": 1.5701043605804443,
      "learning_rate": 1.3163841807909605e-05,
      "loss": 2.3632,
      "step": 2608
    },
    {
      "epoch": 2.211016949152542,
      "grad_norm": 2.2938971519470215,
      "learning_rate": 1.3149717514124293e-05,
      "loss": 1.5606,
      "step": 2609
    },
    {
      "epoch": 2.211864406779661,
      "grad_norm": 1.898794174194336,
      "learning_rate": 1.3135593220338985e-05,
      "loss": 2.3264,
      "step": 2610
    },
    {
      "epoch": 2.2127118644067796,
      "grad_norm": 2.166476011276245,
      "learning_rate": 1.3121468926553673e-05,
      "loss": 1.6981,
      "step": 2611
    },
    {
      "epoch": 2.2135593220338983,
      "grad_norm": 2.0929105281829834,
      "learning_rate": 1.3107344632768361e-05,
      "loss": 1.7719,
      "step": 2612
    },
    {
      "epoch": 2.214406779661017,
      "grad_norm": 2.0681040287017822,
      "learning_rate": 1.309322033898305e-05,
      "loss": 1.7477,
      "step": 2613
    },
    {
      "epoch": 2.2152542372881356,
      "grad_norm": 1.704654574394226,
      "learning_rate": 1.3079096045197741e-05,
      "loss": 2.1921,
      "step": 2614
    },
    {
      "epoch": 2.2161016949152543,
      "grad_norm": 1.507063388824463,
      "learning_rate": 1.306497175141243e-05,
      "loss": 2.3369,
      "step": 2615
    },
    {
      "epoch": 2.216949152542373,
      "grad_norm": 1.3990044593811035,
      "learning_rate": 1.305084745762712e-05,
      "loss": 2.3672,
      "step": 2616
    },
    {
      "epoch": 2.2177966101694917,
      "grad_norm": 1.8505538702011108,
      "learning_rate": 1.3036723163841808e-05,
      "loss": 1.9728,
      "step": 2617
    },
    {
      "epoch": 2.2186440677966104,
      "grad_norm": 1.4994605779647827,
      "learning_rate": 1.30225988700565e-05,
      "loss": 2.3302,
      "step": 2618
    },
    {
      "epoch": 2.2194915254237286,
      "grad_norm": 1.9457956552505493,
      "learning_rate": 1.3008474576271188e-05,
      "loss": 2.0578,
      "step": 2619
    },
    {
      "epoch": 2.2203389830508473,
      "grad_norm": 2.009333848953247,
      "learning_rate": 1.2994350282485876e-05,
      "loss": 2.0096,
      "step": 2620
    },
    {
      "epoch": 2.221186440677966,
      "grad_norm": 1.5474363565444946,
      "learning_rate": 1.2980225988700565e-05,
      "loss": 2.3008,
      "step": 2621
    },
    {
      "epoch": 2.2220338983050847,
      "grad_norm": 1.6145862340927124,
      "learning_rate": 1.2966101694915256e-05,
      "loss": 2.144,
      "step": 2622
    },
    {
      "epoch": 2.2228813559322034,
      "grad_norm": 1.2498451471328735,
      "learning_rate": 1.2951977401129945e-05,
      "loss": 2.4407,
      "step": 2623
    },
    {
      "epoch": 2.223728813559322,
      "grad_norm": 1.3638949394226074,
      "learning_rate": 1.2937853107344633e-05,
      "loss": 2.2856,
      "step": 2624
    },
    {
      "epoch": 2.2245762711864407,
      "grad_norm": 1.553649663925171,
      "learning_rate": 1.2923728813559321e-05,
      "loss": 2.2411,
      "step": 2625
    },
    {
      "epoch": 2.2254237288135594,
      "grad_norm": 2.9169318675994873,
      "learning_rate": 1.2909604519774013e-05,
      "loss": 1.9925,
      "step": 2626
    },
    {
      "epoch": 2.226271186440678,
      "grad_norm": 1.6467723846435547,
      "learning_rate": 1.2895480225988701e-05,
      "loss": 2.073,
      "step": 2627
    },
    {
      "epoch": 2.227118644067797,
      "grad_norm": 1.6251808404922485,
      "learning_rate": 1.288135593220339e-05,
      "loss": 2.2559,
      "step": 2628
    },
    {
      "epoch": 2.227966101694915,
      "grad_norm": 1.448697566986084,
      "learning_rate": 1.286723163841808e-05,
      "loss": 2.4973,
      "step": 2629
    },
    {
      "epoch": 2.2288135593220337,
      "grad_norm": 1.5851848125457764,
      "learning_rate": 1.285310734463277e-05,
      "loss": 2.1272,
      "step": 2630
    },
    {
      "epoch": 2.2296610169491524,
      "grad_norm": 1.842235803604126,
      "learning_rate": 1.2838983050847458e-05,
      "loss": 1.9558,
      "step": 2631
    },
    {
      "epoch": 2.230508474576271,
      "grad_norm": 1.7956777811050415,
      "learning_rate": 1.2824858757062148e-05,
      "loss": 2.1902,
      "step": 2632
    },
    {
      "epoch": 2.23135593220339,
      "grad_norm": 1.7201855182647705,
      "learning_rate": 1.2810734463276836e-05,
      "loss": 2.1372,
      "step": 2633
    },
    {
      "epoch": 2.2322033898305085,
      "grad_norm": 2.0029139518737793,
      "learning_rate": 1.2796610169491528e-05,
      "loss": 1.9169,
      "step": 2634
    },
    {
      "epoch": 2.233050847457627,
      "grad_norm": 1.797213077545166,
      "learning_rate": 1.2782485875706216e-05,
      "loss": 2.0318,
      "step": 2635
    },
    {
      "epoch": 2.233898305084746,
      "grad_norm": 1.9128228425979614,
      "learning_rate": 1.2768361581920904e-05,
      "loss": 2.0825,
      "step": 2636
    },
    {
      "epoch": 2.2347457627118645,
      "grad_norm": 1.6019635200500488,
      "learning_rate": 1.2754237288135593e-05,
      "loss": 2.1612,
      "step": 2637
    },
    {
      "epoch": 2.2355932203389832,
      "grad_norm": 1.6599159240722656,
      "learning_rate": 1.2740112994350284e-05,
      "loss": 2.2443,
      "step": 2638
    },
    {
      "epoch": 2.236440677966102,
      "grad_norm": 1.4098831415176392,
      "learning_rate": 1.2725988700564973e-05,
      "loss": 2.2044,
      "step": 2639
    },
    {
      "epoch": 2.23728813559322,
      "grad_norm": 1.9883822202682495,
      "learning_rate": 1.2711864406779661e-05,
      "loss": 1.9197,
      "step": 2640
    },
    {
      "epoch": 2.238135593220339,
      "grad_norm": 1.4971897602081299,
      "learning_rate": 1.2697740112994349e-05,
      "loss": 2.3918,
      "step": 2641
    },
    {
      "epoch": 2.2389830508474575,
      "grad_norm": 1.145992636680603,
      "learning_rate": 1.268361581920904e-05,
      "loss": 2.6599,
      "step": 2642
    },
    {
      "epoch": 2.239830508474576,
      "grad_norm": 1.6472599506378174,
      "learning_rate": 1.2669491525423729e-05,
      "loss": 2.2362,
      "step": 2643
    },
    {
      "epoch": 2.240677966101695,
      "grad_norm": 1.640352487564087,
      "learning_rate": 1.2655367231638417e-05,
      "loss": 1.9791,
      "step": 2644
    },
    {
      "epoch": 2.2415254237288136,
      "grad_norm": 1.8979905843734741,
      "learning_rate": 1.2641242937853107e-05,
      "loss": 1.9645,
      "step": 2645
    },
    {
      "epoch": 2.2423728813559323,
      "grad_norm": 1.440248966217041,
      "learning_rate": 1.2627118644067797e-05,
      "loss": 2.3772,
      "step": 2646
    },
    {
      "epoch": 2.243220338983051,
      "grad_norm": 1.2969114780426025,
      "learning_rate": 1.2612994350282486e-05,
      "loss": 2.4062,
      "step": 2647
    },
    {
      "epoch": 2.2440677966101696,
      "grad_norm": 1.9726459980010986,
      "learning_rate": 1.2598870056497176e-05,
      "loss": 1.7231,
      "step": 2648
    },
    {
      "epoch": 2.2449152542372883,
      "grad_norm": 1.451680064201355,
      "learning_rate": 1.2584745762711864e-05,
      "loss": 2.4602,
      "step": 2649
    },
    {
      "epoch": 2.2457627118644066,
      "grad_norm": 2.0853891372680664,
      "learning_rate": 1.2570621468926556e-05,
      "loss": 1.9044,
      "step": 2650
    },
    {
      "epoch": 2.2466101694915253,
      "grad_norm": 1.7201420068740845,
      "learning_rate": 1.2556497175141244e-05,
      "loss": 1.9594,
      "step": 2651
    },
    {
      "epoch": 2.247457627118644,
      "grad_norm": 1.182131290435791,
      "learning_rate": 1.2542372881355932e-05,
      "loss": 2.5289,
      "step": 2652
    },
    {
      "epoch": 2.2483050847457626,
      "grad_norm": 1.6462275981903076,
      "learning_rate": 1.252824858757062e-05,
      "loss": 2.2147,
      "step": 2653
    },
    {
      "epoch": 2.2491525423728813,
      "grad_norm": 1.8890339136123657,
      "learning_rate": 1.2514124293785312e-05,
      "loss": 1.7884,
      "step": 2654
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6973693370819092,
      "learning_rate": 1.25e-05,
      "loss": 2.2188,
      "step": 2655
    },
    {
      "epoch": 2.2508474576271187,
      "grad_norm": 1.3676884174346924,
      "learning_rate": 1.248587570621469e-05,
      "loss": 2.3783,
      "step": 2656
    },
    {
      "epoch": 2.2516949152542374,
      "grad_norm": 1.6827466487884521,
      "learning_rate": 1.2471751412429379e-05,
      "loss": 2.2626,
      "step": 2657
    },
    {
      "epoch": 2.252542372881356,
      "grad_norm": 2.5459835529327393,
      "learning_rate": 1.2457627118644069e-05,
      "loss": 1.4011,
      "step": 2658
    },
    {
      "epoch": 2.2533898305084747,
      "grad_norm": 1.1796575784683228,
      "learning_rate": 1.2443502824858757e-05,
      "loss": 2.6082,
      "step": 2659
    },
    {
      "epoch": 2.2542372881355934,
      "grad_norm": 2.302685499191284,
      "learning_rate": 1.2429378531073447e-05,
      "loss": 1.6164,
      "step": 2660
    },
    {
      "epoch": 2.2550847457627117,
      "grad_norm": 1.9911892414093018,
      "learning_rate": 1.2415254237288135e-05,
      "loss": 2.0376,
      "step": 2661
    },
    {
      "epoch": 2.2559322033898304,
      "grad_norm": 1.6822550296783447,
      "learning_rate": 1.2401129943502825e-05,
      "loss": 2.1804,
      "step": 2662
    },
    {
      "epoch": 2.256779661016949,
      "grad_norm": 1.8066636323928833,
      "learning_rate": 1.2387005649717515e-05,
      "loss": 2.1502,
      "step": 2663
    },
    {
      "epoch": 2.2576271186440677,
      "grad_norm": 1.7069318294525146,
      "learning_rate": 1.2372881355932204e-05,
      "loss": 2.0484,
      "step": 2664
    },
    {
      "epoch": 2.2584745762711864,
      "grad_norm": 1.8507767915725708,
      "learning_rate": 1.2358757062146894e-05,
      "loss": 2.0893,
      "step": 2665
    },
    {
      "epoch": 2.259322033898305,
      "grad_norm": 1.5114821195602417,
      "learning_rate": 1.2344632768361584e-05,
      "loss": 2.3222,
      "step": 2666
    },
    {
      "epoch": 2.260169491525424,
      "grad_norm": 1.617348074913025,
      "learning_rate": 1.2330508474576272e-05,
      "loss": 2.2411,
      "step": 2667
    },
    {
      "epoch": 2.2610169491525425,
      "grad_norm": 1.4618602991104126,
      "learning_rate": 1.2316384180790962e-05,
      "loss": 2.4252,
      "step": 2668
    },
    {
      "epoch": 2.261864406779661,
      "grad_norm": 1.623011589050293,
      "learning_rate": 1.230225988700565e-05,
      "loss": 2.1134,
      "step": 2669
    },
    {
      "epoch": 2.26271186440678,
      "grad_norm": 1.7612273693084717,
      "learning_rate": 1.228813559322034e-05,
      "loss": 2.0647,
      "step": 2670
    },
    {
      "epoch": 2.263559322033898,
      "grad_norm": 1.4703688621520996,
      "learning_rate": 1.2274011299435029e-05,
      "loss": 2.2724,
      "step": 2671
    },
    {
      "epoch": 2.2644067796610168,
      "grad_norm": 1.3753740787506104,
      "learning_rate": 1.2259887005649719e-05,
      "loss": 2.3529,
      "step": 2672
    },
    {
      "epoch": 2.2652542372881355,
      "grad_norm": 1.1475927829742432,
      "learning_rate": 1.2245762711864407e-05,
      "loss": 2.5087,
      "step": 2673
    },
    {
      "epoch": 2.266101694915254,
      "grad_norm": 1.6536250114440918,
      "learning_rate": 1.2231638418079097e-05,
      "loss": 2.065,
      "step": 2674
    },
    {
      "epoch": 2.266949152542373,
      "grad_norm": 1.6411614418029785,
      "learning_rate": 1.2217514124293785e-05,
      "loss": 2.2422,
      "step": 2675
    },
    {
      "epoch": 2.2677966101694915,
      "grad_norm": 1.7272388935089111,
      "learning_rate": 1.2203389830508475e-05,
      "loss": 2.053,
      "step": 2676
    },
    {
      "epoch": 2.26864406779661,
      "grad_norm": 1.8547290563583374,
      "learning_rate": 1.2189265536723164e-05,
      "loss": 1.9623,
      "step": 2677
    },
    {
      "epoch": 2.269491525423729,
      "grad_norm": 1.6145331859588623,
      "learning_rate": 1.2175141242937854e-05,
      "loss": 2.1809,
      "step": 2678
    },
    {
      "epoch": 2.2703389830508476,
      "grad_norm": 1.6128720045089722,
      "learning_rate": 1.2161016949152544e-05,
      "loss": 2.0409,
      "step": 2679
    },
    {
      "epoch": 2.2711864406779663,
      "grad_norm": 1.833178162574768,
      "learning_rate": 1.2146892655367234e-05,
      "loss": 2.043,
      "step": 2680
    },
    {
      "epoch": 2.272033898305085,
      "grad_norm": 1.2633790969848633,
      "learning_rate": 1.2132768361581922e-05,
      "loss": 2.5096,
      "step": 2681
    },
    {
      "epoch": 2.272881355932203,
      "grad_norm": 1.8555970191955566,
      "learning_rate": 1.2118644067796612e-05,
      "loss": 2.0319,
      "step": 2682
    },
    {
      "epoch": 2.273728813559322,
      "grad_norm": 1.6284433603286743,
      "learning_rate": 1.21045197740113e-05,
      "loss": 2.2383,
      "step": 2683
    },
    {
      "epoch": 2.2745762711864406,
      "grad_norm": 1.8158788681030273,
      "learning_rate": 1.209039548022599e-05,
      "loss": 2.0611,
      "step": 2684
    },
    {
      "epoch": 2.2754237288135593,
      "grad_norm": 1.6966010332107544,
      "learning_rate": 1.2076271186440678e-05,
      "loss": 2.0186,
      "step": 2685
    },
    {
      "epoch": 2.276271186440678,
      "grad_norm": 1.9948965311050415,
      "learning_rate": 1.2062146892655368e-05,
      "loss": 1.7529,
      "step": 2686
    },
    {
      "epoch": 2.2771186440677966,
      "grad_norm": 2.175189733505249,
      "learning_rate": 1.2048022598870057e-05,
      "loss": 1.7595,
      "step": 2687
    },
    {
      "epoch": 2.2779661016949153,
      "grad_norm": 1.519124150276184,
      "learning_rate": 1.2033898305084747e-05,
      "loss": 2.2935,
      "step": 2688
    },
    {
      "epoch": 2.278813559322034,
      "grad_norm": 1.118815541267395,
      "learning_rate": 1.2019774011299435e-05,
      "loss": 2.5004,
      "step": 2689
    },
    {
      "epoch": 2.2796610169491527,
      "grad_norm": 1.5908997058868408,
      "learning_rate": 1.2005649717514125e-05,
      "loss": 2.2099,
      "step": 2690
    },
    {
      "epoch": 2.2805084745762714,
      "grad_norm": 1.539771318435669,
      "learning_rate": 1.1991525423728813e-05,
      "loss": 2.1345,
      "step": 2691
    },
    {
      "epoch": 2.2813559322033896,
      "grad_norm": 1.9502480030059814,
      "learning_rate": 1.1977401129943503e-05,
      "loss": 1.8979,
      "step": 2692
    },
    {
      "epoch": 2.2822033898305083,
      "grad_norm": 1.8579846620559692,
      "learning_rate": 1.1963276836158192e-05,
      "loss": 2.2162,
      "step": 2693
    },
    {
      "epoch": 2.283050847457627,
      "grad_norm": 1.2290531396865845,
      "learning_rate": 1.1949152542372882e-05,
      "loss": 2.5044,
      "step": 2694
    },
    {
      "epoch": 2.2838983050847457,
      "grad_norm": 1.667528510093689,
      "learning_rate": 1.1935028248587572e-05,
      "loss": 2.0949,
      "step": 2695
    },
    {
      "epoch": 2.2847457627118644,
      "grad_norm": 1.9315160512924194,
      "learning_rate": 1.1920903954802262e-05,
      "loss": 1.864,
      "step": 2696
    },
    {
      "epoch": 2.285593220338983,
      "grad_norm": 1.8570200204849243,
      "learning_rate": 1.190677966101695e-05,
      "loss": 2.2182,
      "step": 2697
    },
    {
      "epoch": 2.2864406779661017,
      "grad_norm": 2.0234036445617676,
      "learning_rate": 1.189265536723164e-05,
      "loss": 1.8277,
      "step": 2698
    },
    {
      "epoch": 2.2872881355932204,
      "grad_norm": 1.6580219268798828,
      "learning_rate": 1.1878531073446328e-05,
      "loss": 2.0986,
      "step": 2699
    },
    {
      "epoch": 2.288135593220339,
      "grad_norm": 1.8003672361373901,
      "learning_rate": 1.1864406779661018e-05,
      "loss": 1.9952,
      "step": 2700
    },
    {
      "epoch": 2.288983050847458,
      "grad_norm": 2.0319578647613525,
      "learning_rate": 1.1850282485875706e-05,
      "loss": 1.8889,
      "step": 2701
    },
    {
      "epoch": 2.2898305084745765,
      "grad_norm": 2.218423843383789,
      "learning_rate": 1.1836158192090396e-05,
      "loss": 1.6844,
      "step": 2702
    },
    {
      "epoch": 2.290677966101695,
      "grad_norm": 1.7018412351608276,
      "learning_rate": 1.1822033898305085e-05,
      "loss": 2.0824,
      "step": 2703
    },
    {
      "epoch": 2.2915254237288134,
      "grad_norm": 1.8618487119674683,
      "learning_rate": 1.1807909604519775e-05,
      "loss": 1.9784,
      "step": 2704
    },
    {
      "epoch": 2.292372881355932,
      "grad_norm": 1.6080292463302612,
      "learning_rate": 1.1793785310734463e-05,
      "loss": 2.2731,
      "step": 2705
    },
    {
      "epoch": 2.2932203389830508,
      "grad_norm": 2.1574764251708984,
      "learning_rate": 1.1779661016949153e-05,
      "loss": 2.0009,
      "step": 2706
    },
    {
      "epoch": 2.2940677966101695,
      "grad_norm": 1.6515130996704102,
      "learning_rate": 1.1765536723163841e-05,
      "loss": 2.1144,
      "step": 2707
    },
    {
      "epoch": 2.294915254237288,
      "grad_norm": 1.5661877393722534,
      "learning_rate": 1.1751412429378531e-05,
      "loss": 2.2562,
      "step": 2708
    },
    {
      "epoch": 2.295762711864407,
      "grad_norm": 2.0387375354766846,
      "learning_rate": 1.173728813559322e-05,
      "loss": 1.8794,
      "step": 2709
    },
    {
      "epoch": 2.2966101694915255,
      "grad_norm": 1.8044418096542358,
      "learning_rate": 1.172316384180791e-05,
      "loss": 2.1209,
      "step": 2710
    },
    {
      "epoch": 2.297457627118644,
      "grad_norm": 1.8447290658950806,
      "learning_rate": 1.17090395480226e-05,
      "loss": 2.0123,
      "step": 2711
    },
    {
      "epoch": 2.298305084745763,
      "grad_norm": 1.7493408918380737,
      "learning_rate": 1.169491525423729e-05,
      "loss": 2.1515,
      "step": 2712
    },
    {
      "epoch": 2.299152542372881,
      "grad_norm": 1.6475424766540527,
      "learning_rate": 1.1680790960451978e-05,
      "loss": 2.2462,
      "step": 2713
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6466798782348633,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 2.2663,
      "step": 2714
    },
    {
      "epoch": 2.3008474576271185,
      "grad_norm": 1.8530153036117554,
      "learning_rate": 1.1652542372881356e-05,
      "loss": 2.0214,
      "step": 2715
    },
    {
      "epoch": 2.301694915254237,
      "grad_norm": 1.6622998714447021,
      "learning_rate": 1.1638418079096046e-05,
      "loss": 2.1939,
      "step": 2716
    },
    {
      "epoch": 2.302542372881356,
      "grad_norm": 1.883543610572815,
      "learning_rate": 1.1624293785310734e-05,
      "loss": 1.9494,
      "step": 2717
    },
    {
      "epoch": 2.3033898305084746,
      "grad_norm": 1.7924774885177612,
      "learning_rate": 1.1610169491525424e-05,
      "loss": 2.0853,
      "step": 2718
    },
    {
      "epoch": 2.3042372881355933,
      "grad_norm": 1.6783621311187744,
      "learning_rate": 1.1596045197740113e-05,
      "loss": 2.1701,
      "step": 2719
    },
    {
      "epoch": 2.305084745762712,
      "grad_norm": 1.4182416200637817,
      "learning_rate": 1.1581920903954803e-05,
      "loss": 2.4056,
      "step": 2720
    },
    {
      "epoch": 2.3059322033898306,
      "grad_norm": 2.2613394260406494,
      "learning_rate": 1.1567796610169491e-05,
      "loss": 1.9061,
      "step": 2721
    },
    {
      "epoch": 2.3067796610169493,
      "grad_norm": 2.1956305503845215,
      "learning_rate": 1.1553672316384181e-05,
      "loss": 1.7216,
      "step": 2722
    },
    {
      "epoch": 2.307627118644068,
      "grad_norm": 1.333406686782837,
      "learning_rate": 1.153954802259887e-05,
      "loss": 2.3645,
      "step": 2723
    },
    {
      "epoch": 2.3084745762711867,
      "grad_norm": 1.9453949928283691,
      "learning_rate": 1.152542372881356e-05,
      "loss": 1.7782,
      "step": 2724
    },
    {
      "epoch": 2.309322033898305,
      "grad_norm": 1.9100425243377686,
      "learning_rate": 1.151129943502825e-05,
      "loss": 1.8895,
      "step": 2725
    },
    {
      "epoch": 2.3101694915254236,
      "grad_norm": 1.7991353273391724,
      "learning_rate": 1.1497175141242938e-05,
      "loss": 2.125,
      "step": 2726
    },
    {
      "epoch": 2.3110169491525423,
      "grad_norm": 1.6553651094436646,
      "learning_rate": 1.1483050847457628e-05,
      "loss": 2.0838,
      "step": 2727
    },
    {
      "epoch": 2.311864406779661,
      "grad_norm": 1.5747385025024414,
      "learning_rate": 1.1468926553672318e-05,
      "loss": 1.9827,
      "step": 2728
    },
    {
      "epoch": 2.3127118644067797,
      "grad_norm": 1.8182562589645386,
      "learning_rate": 1.1454802259887006e-05,
      "loss": 2.1982,
      "step": 2729
    },
    {
      "epoch": 2.3135593220338984,
      "grad_norm": 2.0031235218048096,
      "learning_rate": 1.1440677966101696e-05,
      "loss": 1.7508,
      "step": 2730
    },
    {
      "epoch": 2.314406779661017,
      "grad_norm": 1.7654558420181274,
      "learning_rate": 1.1426553672316384e-05,
      "loss": 2.1223,
      "step": 2731
    },
    {
      "epoch": 2.3152542372881357,
      "grad_norm": 1.3843965530395508,
      "learning_rate": 1.1412429378531074e-05,
      "loss": 2.3811,
      "step": 2732
    },
    {
      "epoch": 2.3161016949152544,
      "grad_norm": 2.099733352661133,
      "learning_rate": 1.1398305084745763e-05,
      "loss": 1.8498,
      "step": 2733
    },
    {
      "epoch": 2.3169491525423727,
      "grad_norm": 2.0038676261901855,
      "learning_rate": 1.1384180790960453e-05,
      "loss": 2.2145,
      "step": 2734
    },
    {
      "epoch": 2.3177966101694913,
      "grad_norm": 2.0445220470428467,
      "learning_rate": 1.137005649717514e-05,
      "loss": 1.9238,
      "step": 2735
    },
    {
      "epoch": 2.31864406779661,
      "grad_norm": 1.8142948150634766,
      "learning_rate": 1.135593220338983e-05,
      "loss": 2.0539,
      "step": 2736
    },
    {
      "epoch": 2.3194915254237287,
      "grad_norm": 2.1312167644500732,
      "learning_rate": 1.134180790960452e-05,
      "loss": 1.8662,
      "step": 2737
    },
    {
      "epoch": 2.3203389830508474,
      "grad_norm": 1.7986578941345215,
      "learning_rate": 1.1327683615819209e-05,
      "loss": 2.1375,
      "step": 2738
    },
    {
      "epoch": 2.321186440677966,
      "grad_norm": 1.3996113538742065,
      "learning_rate": 1.1313559322033899e-05,
      "loss": 2.3352,
      "step": 2739
    },
    {
      "epoch": 2.3220338983050848,
      "grad_norm": 1.4943941831588745,
      "learning_rate": 1.1299435028248587e-05,
      "loss": 2.3732,
      "step": 2740
    },
    {
      "epoch": 2.3228813559322035,
      "grad_norm": 1.417456865310669,
      "learning_rate": 1.1285310734463277e-05,
      "loss": 2.3092,
      "step": 2741
    },
    {
      "epoch": 2.323728813559322,
      "grad_norm": 1.2574368715286255,
      "learning_rate": 1.1271186440677967e-05,
      "loss": 2.6752,
      "step": 2742
    },
    {
      "epoch": 2.324576271186441,
      "grad_norm": 1.65830659866333,
      "learning_rate": 1.1257062146892656e-05,
      "loss": 2.2794,
      "step": 2743
    },
    {
      "epoch": 2.3254237288135595,
      "grad_norm": 1.9377236366271973,
      "learning_rate": 1.1242937853107346e-05,
      "loss": 1.8714,
      "step": 2744
    },
    {
      "epoch": 2.326271186440678,
      "grad_norm": 1.7150442600250244,
      "learning_rate": 1.1228813559322036e-05,
      "loss": 2.3143,
      "step": 2745
    },
    {
      "epoch": 2.3271186440677964,
      "grad_norm": 1.9324195384979248,
      "learning_rate": 1.1214689265536724e-05,
      "loss": 2.1327,
      "step": 2746
    },
    {
      "epoch": 2.327966101694915,
      "grad_norm": 1.2602370977401733,
      "learning_rate": 1.1200564971751414e-05,
      "loss": 2.4212,
      "step": 2747
    },
    {
      "epoch": 2.328813559322034,
      "grad_norm": 1.5296436548233032,
      "learning_rate": 1.1186440677966102e-05,
      "loss": 2.2718,
      "step": 2748
    },
    {
      "epoch": 2.3296610169491525,
      "grad_norm": 1.7456475496292114,
      "learning_rate": 1.1172316384180792e-05,
      "loss": 2.1383,
      "step": 2749
    },
    {
      "epoch": 2.330508474576271,
      "grad_norm": 1.623611330986023,
      "learning_rate": 1.115819209039548e-05,
      "loss": 2.1878,
      "step": 2750
    },
    {
      "epoch": 2.33135593220339,
      "grad_norm": 1.6439287662506104,
      "learning_rate": 1.114406779661017e-05,
      "loss": 2.1031,
      "step": 2751
    },
    {
      "epoch": 2.3322033898305086,
      "grad_norm": 1.5060378313064575,
      "learning_rate": 1.1129943502824859e-05,
      "loss": 2.2501,
      "step": 2752
    },
    {
      "epoch": 2.3330508474576273,
      "grad_norm": 1.630726933479309,
      "learning_rate": 1.1115819209039549e-05,
      "loss": 2.0935,
      "step": 2753
    },
    {
      "epoch": 2.333898305084746,
      "grad_norm": 1.8235399723052979,
      "learning_rate": 1.1101694915254237e-05,
      "loss": 1.8763,
      "step": 2754
    },
    {
      "epoch": 2.334745762711864,
      "grad_norm": 1.6734403371810913,
      "learning_rate": 1.1087570621468927e-05,
      "loss": 2.1593,
      "step": 2755
    },
    {
      "epoch": 2.335593220338983,
      "grad_norm": 1.6418523788452148,
      "learning_rate": 1.1073446327683615e-05,
      "loss": 2.3486,
      "step": 2756
    },
    {
      "epoch": 2.3364406779661016,
      "grad_norm": 1.5744705200195312,
      "learning_rate": 1.1059322033898305e-05,
      "loss": 2.3483,
      "step": 2757
    },
    {
      "epoch": 2.3372881355932202,
      "grad_norm": 2.012573719024658,
      "learning_rate": 1.1045197740112995e-05,
      "loss": 2.081,
      "step": 2758
    },
    {
      "epoch": 2.338135593220339,
      "grad_norm": 1.7633767127990723,
      "learning_rate": 1.1031073446327685e-05,
      "loss": 2.0997,
      "step": 2759
    },
    {
      "epoch": 2.3389830508474576,
      "grad_norm": 1.5577518939971924,
      "learning_rate": 1.1016949152542374e-05,
      "loss": 2.1127,
      "step": 2760
    },
    {
      "epoch": 2.3398305084745763,
      "grad_norm": 1.6724456548690796,
      "learning_rate": 1.1002824858757064e-05,
      "loss": 2.2547,
      "step": 2761
    },
    {
      "epoch": 2.340677966101695,
      "grad_norm": 2.2773401737213135,
      "learning_rate": 1.0988700564971752e-05,
      "loss": 1.7376,
      "step": 2762
    },
    {
      "epoch": 2.3415254237288137,
      "grad_norm": 1.601205825805664,
      "learning_rate": 1.0974576271186442e-05,
      "loss": 2.3381,
      "step": 2763
    },
    {
      "epoch": 2.3423728813559324,
      "grad_norm": 1.7957279682159424,
      "learning_rate": 1.096045197740113e-05,
      "loss": 2.246,
      "step": 2764
    },
    {
      "epoch": 2.343220338983051,
      "grad_norm": 2.417149305343628,
      "learning_rate": 1.094632768361582e-05,
      "loss": 1.99,
      "step": 2765
    },
    {
      "epoch": 2.3440677966101697,
      "grad_norm": 1.5688859224319458,
      "learning_rate": 1.0932203389830509e-05,
      "loss": 2.2738,
      "step": 2766
    },
    {
      "epoch": 2.344915254237288,
      "grad_norm": 1.453046441078186,
      "learning_rate": 1.0918079096045199e-05,
      "loss": 2.3192,
      "step": 2767
    },
    {
      "epoch": 2.3457627118644067,
      "grad_norm": 2.468358039855957,
      "learning_rate": 1.0903954802259887e-05,
      "loss": 1.6678,
      "step": 2768
    },
    {
      "epoch": 2.3466101694915253,
      "grad_norm": 1.8789864778518677,
      "learning_rate": 1.0889830508474577e-05,
      "loss": 2.1603,
      "step": 2769
    },
    {
      "epoch": 2.347457627118644,
      "grad_norm": 1.5327998399734497,
      "learning_rate": 1.0875706214689265e-05,
      "loss": 2.262,
      "step": 2770
    },
    {
      "epoch": 2.3483050847457627,
      "grad_norm": 1.5479819774627686,
      "learning_rate": 1.0861581920903955e-05,
      "loss": 2.2451,
      "step": 2771
    },
    {
      "epoch": 2.3491525423728814,
      "grad_norm": 1.2894599437713623,
      "learning_rate": 1.0847457627118644e-05,
      "loss": 2.4146,
      "step": 2772
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3935586214065552,
      "learning_rate": 1.0833333333333334e-05,
      "loss": 2.4342,
      "step": 2773
    },
    {
      "epoch": 2.3508474576271188,
      "grad_norm": 1.552455186843872,
      "learning_rate": 1.0819209039548024e-05,
      "loss": 2.2017,
      "step": 2774
    },
    {
      "epoch": 2.3516949152542375,
      "grad_norm": 1.7251968383789062,
      "learning_rate": 1.0805084745762714e-05,
      "loss": 2.0838,
      "step": 2775
    },
    {
      "epoch": 2.3525423728813557,
      "grad_norm": 1.8735079765319824,
      "learning_rate": 1.0790960451977402e-05,
      "loss": 1.9269,
      "step": 2776
    },
    {
      "epoch": 2.3533898305084744,
      "grad_norm": 1.9194589853286743,
      "learning_rate": 1.0776836158192092e-05,
      "loss": 1.9188,
      "step": 2777
    },
    {
      "epoch": 2.354237288135593,
      "grad_norm": 1.724419116973877,
      "learning_rate": 1.076271186440678e-05,
      "loss": 1.8482,
      "step": 2778
    },
    {
      "epoch": 2.3550847457627118,
      "grad_norm": 1.979636549949646,
      "learning_rate": 1.074858757062147e-05,
      "loss": 2.1036,
      "step": 2779
    },
    {
      "epoch": 2.3559322033898304,
      "grad_norm": 2.013507843017578,
      "learning_rate": 1.0734463276836158e-05,
      "loss": 1.9052,
      "step": 2780
    },
    {
      "epoch": 2.356779661016949,
      "grad_norm": 1.5404797792434692,
      "learning_rate": 1.0720338983050848e-05,
      "loss": 2.2025,
      "step": 2781
    },
    {
      "epoch": 2.357627118644068,
      "grad_norm": 1.9583626985549927,
      "learning_rate": 1.0706214689265537e-05,
      "loss": 2.0116,
      "step": 2782
    },
    {
      "epoch": 2.3584745762711865,
      "grad_norm": 1.8386528491973877,
      "learning_rate": 1.0692090395480227e-05,
      "loss": 2.1348,
      "step": 2783
    },
    {
      "epoch": 2.359322033898305,
      "grad_norm": 1.9649231433868408,
      "learning_rate": 1.0677966101694915e-05,
      "loss": 1.7567,
      "step": 2784
    },
    {
      "epoch": 2.360169491525424,
      "grad_norm": 1.9654440879821777,
      "learning_rate": 1.0663841807909605e-05,
      "loss": 2.0183,
      "step": 2785
    },
    {
      "epoch": 2.3610169491525426,
      "grad_norm": 1.9445511102676392,
      "learning_rate": 1.0649717514124293e-05,
      "loss": 2.1337,
      "step": 2786
    },
    {
      "epoch": 2.3618644067796613,
      "grad_norm": 1.9277061223983765,
      "learning_rate": 1.0635593220338983e-05,
      "loss": 1.8659,
      "step": 2787
    },
    {
      "epoch": 2.3627118644067795,
      "grad_norm": 1.9481409788131714,
      "learning_rate": 1.0621468926553672e-05,
      "loss": 1.9901,
      "step": 2788
    },
    {
      "epoch": 2.363559322033898,
      "grad_norm": 1.097215175628662,
      "learning_rate": 1.0607344632768362e-05,
      "loss": 2.5807,
      "step": 2789
    },
    {
      "epoch": 2.364406779661017,
      "grad_norm": 2.107255220413208,
      "learning_rate": 1.0593220338983052e-05,
      "loss": 1.6506,
      "step": 2790
    },
    {
      "epoch": 2.3652542372881356,
      "grad_norm": 1.51083242893219,
      "learning_rate": 1.0579096045197742e-05,
      "loss": 2.2698,
      "step": 2791
    },
    {
      "epoch": 2.3661016949152542,
      "grad_norm": 2.2006630897521973,
      "learning_rate": 1.056497175141243e-05,
      "loss": 1.6796,
      "step": 2792
    },
    {
      "epoch": 2.366949152542373,
      "grad_norm": 1.544604778289795,
      "learning_rate": 1.055084745762712e-05,
      "loss": 2.2162,
      "step": 2793
    },
    {
      "epoch": 2.3677966101694916,
      "grad_norm": 1.856799840927124,
      "learning_rate": 1.0536723163841808e-05,
      "loss": 1.8844,
      "step": 2794
    },
    {
      "epoch": 2.3686440677966103,
      "grad_norm": 1.3587238788604736,
      "learning_rate": 1.0522598870056498e-05,
      "loss": 2.2896,
      "step": 2795
    },
    {
      "epoch": 2.369491525423729,
      "grad_norm": 1.6245845556259155,
      "learning_rate": 1.0508474576271186e-05,
      "loss": 2.2257,
      "step": 2796
    },
    {
      "epoch": 2.3703389830508472,
      "grad_norm": 1.266310453414917,
      "learning_rate": 1.0494350282485876e-05,
      "loss": 2.4608,
      "step": 2797
    },
    {
      "epoch": 2.371186440677966,
      "grad_norm": 2.4574804306030273,
      "learning_rate": 1.0480225988700565e-05,
      "loss": 1.6334,
      "step": 2798
    },
    {
      "epoch": 2.3720338983050846,
      "grad_norm": 1.696324110031128,
      "learning_rate": 1.0466101694915255e-05,
      "loss": 2.1779,
      "step": 2799
    },
    {
      "epoch": 2.3728813559322033,
      "grad_norm": 1.362596035003662,
      "learning_rate": 1.0451977401129943e-05,
      "loss": 2.5589,
      "step": 2800
    },
    {
      "epoch": 2.373728813559322,
      "grad_norm": 1.9849655628204346,
      "learning_rate": 1.0437853107344633e-05,
      "loss": 2.1233,
      "step": 2801
    },
    {
      "epoch": 2.3745762711864407,
      "grad_norm": 2.01254940032959,
      "learning_rate": 1.0423728813559321e-05,
      "loss": 2.1137,
      "step": 2802
    },
    {
      "epoch": 2.3754237288135593,
      "grad_norm": 1.8358657360076904,
      "learning_rate": 1.0409604519774011e-05,
      "loss": 2.0911,
      "step": 2803
    },
    {
      "epoch": 2.376271186440678,
      "grad_norm": 1.8649876117706299,
      "learning_rate": 1.0395480225988701e-05,
      "loss": 1.9854,
      "step": 2804
    },
    {
      "epoch": 2.3771186440677967,
      "grad_norm": 2.0961883068084717,
      "learning_rate": 1.038135593220339e-05,
      "loss": 2.0988,
      "step": 2805
    },
    {
      "epoch": 2.3779661016949154,
      "grad_norm": 1.7328708171844482,
      "learning_rate": 1.036723163841808e-05,
      "loss": 2.2116,
      "step": 2806
    },
    {
      "epoch": 2.378813559322034,
      "grad_norm": 1.34184730052948,
      "learning_rate": 1.035310734463277e-05,
      "loss": 2.3053,
      "step": 2807
    },
    {
      "epoch": 2.3796610169491528,
      "grad_norm": 1.623054027557373,
      "learning_rate": 1.0338983050847458e-05,
      "loss": 2.3248,
      "step": 2808
    },
    {
      "epoch": 2.380508474576271,
      "grad_norm": 1.5115116834640503,
      "learning_rate": 1.0324858757062148e-05,
      "loss": 2.3151,
      "step": 2809
    },
    {
      "epoch": 2.3813559322033897,
      "grad_norm": 2.490647792816162,
      "learning_rate": 1.0310734463276836e-05,
      "loss": 1.6813,
      "step": 2810
    },
    {
      "epoch": 2.3822033898305084,
      "grad_norm": 1.9092079401016235,
      "learning_rate": 1.0296610169491526e-05,
      "loss": 1.9132,
      "step": 2811
    },
    {
      "epoch": 2.383050847457627,
      "grad_norm": 1.6100598573684692,
      "learning_rate": 1.0282485875706214e-05,
      "loss": 2.1598,
      "step": 2812
    },
    {
      "epoch": 2.3838983050847458,
      "grad_norm": 1.8987376689910889,
      "learning_rate": 1.0268361581920904e-05,
      "loss": 1.9817,
      "step": 2813
    },
    {
      "epoch": 2.3847457627118644,
      "grad_norm": 1.4695061445236206,
      "learning_rate": 1.0254237288135593e-05,
      "loss": 2.328,
      "step": 2814
    },
    {
      "epoch": 2.385593220338983,
      "grad_norm": 2.067765712738037,
      "learning_rate": 1.0240112994350283e-05,
      "loss": 1.8811,
      "step": 2815
    },
    {
      "epoch": 2.386440677966102,
      "grad_norm": 1.7795380353927612,
      "learning_rate": 1.0225988700564973e-05,
      "loss": 2.1139,
      "step": 2816
    },
    {
      "epoch": 2.3872881355932205,
      "grad_norm": 1.671406865119934,
      "learning_rate": 1.0211864406779661e-05,
      "loss": 2.1007,
      "step": 2817
    },
    {
      "epoch": 2.3881355932203387,
      "grad_norm": 2.126786947250366,
      "learning_rate": 1.0197740112994351e-05,
      "loss": 1.8009,
      "step": 2818
    },
    {
      "epoch": 2.3889830508474574,
      "grad_norm": 1.7504265308380127,
      "learning_rate": 1.018361581920904e-05,
      "loss": 2.0789,
      "step": 2819
    },
    {
      "epoch": 2.389830508474576,
      "grad_norm": 1.5832493305206299,
      "learning_rate": 1.016949152542373e-05,
      "loss": 2.2497,
      "step": 2820
    },
    {
      "epoch": 2.390677966101695,
      "grad_norm": 1.560479760169983,
      "learning_rate": 1.015536723163842e-05,
      "loss": 2.3253,
      "step": 2821
    },
    {
      "epoch": 2.3915254237288135,
      "grad_norm": 1.9991607666015625,
      "learning_rate": 1.0141242937853108e-05,
      "loss": 1.9794,
      "step": 2822
    },
    {
      "epoch": 2.392372881355932,
      "grad_norm": 1.3287773132324219,
      "learning_rate": 1.0127118644067798e-05,
      "loss": 2.4257,
      "step": 2823
    },
    {
      "epoch": 2.393220338983051,
      "grad_norm": 1.7449588775634766,
      "learning_rate": 1.0112994350282488e-05,
      "loss": 2.2996,
      "step": 2824
    },
    {
      "epoch": 2.3940677966101696,
      "grad_norm": 1.979357361793518,
      "learning_rate": 1.0098870056497176e-05,
      "loss": 2.0135,
      "step": 2825
    },
    {
      "epoch": 2.3949152542372882,
      "grad_norm": 1.8858957290649414,
      "learning_rate": 1.0084745762711866e-05,
      "loss": 1.8583,
      "step": 2826
    },
    {
      "epoch": 2.395762711864407,
      "grad_norm": 1.7009750604629517,
      "learning_rate": 1.0070621468926554e-05,
      "loss": 2.0953,
      "step": 2827
    },
    {
      "epoch": 2.3966101694915256,
      "grad_norm": 1.5228315591812134,
      "learning_rate": 1.0056497175141244e-05,
      "loss": 2.2899,
      "step": 2828
    },
    {
      "epoch": 2.3974576271186443,
      "grad_norm": 1.3745592832565308,
      "learning_rate": 1.0042372881355933e-05,
      "loss": 2.454,
      "step": 2829
    },
    {
      "epoch": 2.3983050847457625,
      "grad_norm": 2.0505104064941406,
      "learning_rate": 1.0028248587570623e-05,
      "loss": 1.6602,
      "step": 2830
    },
    {
      "epoch": 2.3991525423728812,
      "grad_norm": 2.1742746829986572,
      "learning_rate": 1.001412429378531e-05,
      "loss": 1.5815,
      "step": 2831
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5642485618591309,
      "learning_rate": 1e-05,
      "loss": 2.2674,
      "step": 2832
    },
    {
      "epoch": 2.4008474576271186,
      "grad_norm": 1.7752679586410522,
      "learning_rate": 9.985875706214689e-06,
      "loss": 2.2217,
      "step": 2833
    },
    {
      "epoch": 2.4016949152542373,
      "grad_norm": 1.7923104763031006,
      "learning_rate": 9.971751412429379e-06,
      "loss": 2.1723,
      "step": 2834
    },
    {
      "epoch": 2.402542372881356,
      "grad_norm": 1.6144061088562012,
      "learning_rate": 9.957627118644067e-06,
      "loss": 2.09,
      "step": 2835
    },
    {
      "epoch": 2.4033898305084747,
      "grad_norm": 1.3840607404708862,
      "learning_rate": 9.943502824858757e-06,
      "loss": 2.3489,
      "step": 2836
    },
    {
      "epoch": 2.4042372881355933,
      "grad_norm": 1.6972010135650635,
      "learning_rate": 9.929378531073447e-06,
      "loss": 2.1179,
      "step": 2837
    },
    {
      "epoch": 2.405084745762712,
      "grad_norm": 1.6566342115402222,
      "learning_rate": 9.915254237288137e-06,
      "loss": 2.3024,
      "step": 2838
    },
    {
      "epoch": 2.4059322033898303,
      "grad_norm": 1.9090768098831177,
      "learning_rate": 9.901129943502826e-06,
      "loss": 1.861,
      "step": 2839
    },
    {
      "epoch": 2.406779661016949,
      "grad_norm": 3.16331148147583,
      "learning_rate": 9.887005649717516e-06,
      "loss": 2.2437,
      "step": 2840
    },
    {
      "epoch": 2.4076271186440676,
      "grad_norm": 1.7605934143066406,
      "learning_rate": 9.872881355932204e-06,
      "loss": 1.996,
      "step": 2841
    },
    {
      "epoch": 2.4084745762711863,
      "grad_norm": 1.681734561920166,
      "learning_rate": 9.858757062146894e-06,
      "loss": 2.1829,
      "step": 2842
    },
    {
      "epoch": 2.409322033898305,
      "grad_norm": 1.2623953819274902,
      "learning_rate": 9.844632768361582e-06,
      "loss": 2.5751,
      "step": 2843
    },
    {
      "epoch": 2.4101694915254237,
      "grad_norm": 1.6833974123001099,
      "learning_rate": 9.830508474576272e-06,
      "loss": 2.3035,
      "step": 2844
    },
    {
      "epoch": 2.4110169491525424,
      "grad_norm": 1.344804048538208,
      "learning_rate": 9.81638418079096e-06,
      "loss": 2.5304,
      "step": 2845
    },
    {
      "epoch": 2.411864406779661,
      "grad_norm": 2.0134329795837402,
      "learning_rate": 9.80225988700565e-06,
      "loss": 1.9028,
      "step": 2846
    },
    {
      "epoch": 2.4127118644067798,
      "grad_norm": 1.6806460618972778,
      "learning_rate": 9.788135593220339e-06,
      "loss": 2.2782,
      "step": 2847
    },
    {
      "epoch": 2.4135593220338984,
      "grad_norm": 1.6235787868499756,
      "learning_rate": 9.774011299435029e-06,
      "loss": 2.3137,
      "step": 2848
    },
    {
      "epoch": 2.414406779661017,
      "grad_norm": 1.7612007856369019,
      "learning_rate": 9.759887005649717e-06,
      "loss": 1.9919,
      "step": 2849
    },
    {
      "epoch": 2.415254237288136,
      "grad_norm": 2.030646800994873,
      "learning_rate": 9.745762711864407e-06,
      "loss": 1.812,
      "step": 2850
    },
    {
      "epoch": 2.416101694915254,
      "grad_norm": 1.82063627243042,
      "learning_rate": 9.731638418079095e-06,
      "loss": 2.0613,
      "step": 2851
    },
    {
      "epoch": 2.4169491525423727,
      "grad_norm": 1.7056388854980469,
      "learning_rate": 9.717514124293785e-06,
      "loss": 2.1565,
      "step": 2852
    },
    {
      "epoch": 2.4177966101694914,
      "grad_norm": 1.4293837547302246,
      "learning_rate": 9.703389830508475e-06,
      "loss": 2.2925,
      "step": 2853
    },
    {
      "epoch": 2.41864406779661,
      "grad_norm": 2.1872732639312744,
      "learning_rate": 9.689265536723165e-06,
      "loss": 1.9586,
      "step": 2854
    },
    {
      "epoch": 2.419491525423729,
      "grad_norm": 1.6308726072311401,
      "learning_rate": 9.675141242937854e-06,
      "loss": 2.1272,
      "step": 2855
    },
    {
      "epoch": 2.4203389830508475,
      "grad_norm": 2.1899256706237793,
      "learning_rate": 9.661016949152544e-06,
      "loss": 2.0633,
      "step": 2856
    },
    {
      "epoch": 2.421186440677966,
      "grad_norm": 1.827673077583313,
      "learning_rate": 9.646892655367232e-06,
      "loss": 2.0225,
      "step": 2857
    },
    {
      "epoch": 2.422033898305085,
      "grad_norm": 1.8376131057739258,
      "learning_rate": 9.632768361581922e-06,
      "loss": 2.1096,
      "step": 2858
    },
    {
      "epoch": 2.4228813559322036,
      "grad_norm": 1.9163520336151123,
      "learning_rate": 9.61864406779661e-06,
      "loss": 1.8362,
      "step": 2859
    },
    {
      "epoch": 2.423728813559322,
      "grad_norm": 2.1687750816345215,
      "learning_rate": 9.6045197740113e-06,
      "loss": 1.6314,
      "step": 2860
    },
    {
      "epoch": 2.4245762711864405,
      "grad_norm": 1.5438897609710693,
      "learning_rate": 9.590395480225989e-06,
      "loss": 2.3105,
      "step": 2861
    },
    {
      "epoch": 2.425423728813559,
      "grad_norm": 2.1108450889587402,
      "learning_rate": 9.576271186440679e-06,
      "loss": 1.9747,
      "step": 2862
    },
    {
      "epoch": 2.426271186440678,
      "grad_norm": 1.5420048236846924,
      "learning_rate": 9.562146892655367e-06,
      "loss": 2.2647,
      "step": 2863
    },
    {
      "epoch": 2.4271186440677965,
      "grad_norm": 1.6626832485198975,
      "learning_rate": 9.548022598870057e-06,
      "loss": 2.0965,
      "step": 2864
    },
    {
      "epoch": 2.4279661016949152,
      "grad_norm": 1.7532156705856323,
      "learning_rate": 9.533898305084745e-06,
      "loss": 2.1138,
      "step": 2865
    },
    {
      "epoch": 2.428813559322034,
      "grad_norm": 1.538637638092041,
      "learning_rate": 9.519774011299435e-06,
      "loss": 2.376,
      "step": 2866
    },
    {
      "epoch": 2.4296610169491526,
      "grad_norm": 1.618196964263916,
      "learning_rate": 9.505649717514124e-06,
      "loss": 2.1637,
      "step": 2867
    },
    {
      "epoch": 2.4305084745762713,
      "grad_norm": 1.9505409002304077,
      "learning_rate": 9.491525423728814e-06,
      "loss": 1.932,
      "step": 2868
    },
    {
      "epoch": 2.43135593220339,
      "grad_norm": 2.0378029346466064,
      "learning_rate": 9.477401129943503e-06,
      "loss": 1.9573,
      "step": 2869
    },
    {
      "epoch": 2.4322033898305087,
      "grad_norm": 2.2740654945373535,
      "learning_rate": 9.463276836158193e-06,
      "loss": 1.6023,
      "step": 2870
    },
    {
      "epoch": 2.4330508474576273,
      "grad_norm": 2.0650203227996826,
      "learning_rate": 9.449152542372882e-06,
      "loss": 1.7876,
      "step": 2871
    },
    {
      "epoch": 2.4338983050847456,
      "grad_norm": 1.5308619737625122,
      "learning_rate": 9.435028248587572e-06,
      "loss": 2.3185,
      "step": 2872
    },
    {
      "epoch": 2.4347457627118643,
      "grad_norm": 1.7073314189910889,
      "learning_rate": 9.42090395480226e-06,
      "loss": 2.2802,
      "step": 2873
    },
    {
      "epoch": 2.435593220338983,
      "grad_norm": 1.7173891067504883,
      "learning_rate": 9.40677966101695e-06,
      "loss": 2.1891,
      "step": 2874
    },
    {
      "epoch": 2.4364406779661016,
      "grad_norm": 1.8193533420562744,
      "learning_rate": 9.392655367231638e-06,
      "loss": 2.2416,
      "step": 2875
    },
    {
      "epoch": 2.4372881355932203,
      "grad_norm": 1.5303488969802856,
      "learning_rate": 9.378531073446328e-06,
      "loss": 2.2405,
      "step": 2876
    },
    {
      "epoch": 2.438135593220339,
      "grad_norm": 1.632622480392456,
      "learning_rate": 9.364406779661017e-06,
      "loss": 2.1976,
      "step": 2877
    },
    {
      "epoch": 2.4389830508474577,
      "grad_norm": 1.5927083492279053,
      "learning_rate": 9.350282485875707e-06,
      "loss": 2.012,
      "step": 2878
    },
    {
      "epoch": 2.4398305084745764,
      "grad_norm": 1.881761074066162,
      "learning_rate": 9.336158192090395e-06,
      "loss": 1.9449,
      "step": 2879
    },
    {
      "epoch": 2.440677966101695,
      "grad_norm": 1.309574842453003,
      "learning_rate": 9.322033898305085e-06,
      "loss": 2.5524,
      "step": 2880
    },
    {
      "epoch": 2.4415254237288133,
      "grad_norm": 1.5436817407608032,
      "learning_rate": 9.307909604519773e-06,
      "loss": 2.2504,
      "step": 2881
    },
    {
      "epoch": 2.442372881355932,
      "grad_norm": 1.769839882850647,
      "learning_rate": 9.293785310734463e-06,
      "loss": 2.0125,
      "step": 2882
    },
    {
      "epoch": 2.4432203389830507,
      "grad_norm": 1.6599061489105225,
      "learning_rate": 9.279661016949153e-06,
      "loss": 2.25,
      "step": 2883
    },
    {
      "epoch": 2.4440677966101694,
      "grad_norm": 1.8840241432189941,
      "learning_rate": 9.265536723163842e-06,
      "loss": 2.2019,
      "step": 2884
    },
    {
      "epoch": 2.444915254237288,
      "grad_norm": 1.6114009618759155,
      "learning_rate": 9.251412429378532e-06,
      "loss": 2.1258,
      "step": 2885
    },
    {
      "epoch": 2.4457627118644067,
      "grad_norm": 1.8707669973373413,
      "learning_rate": 9.237288135593222e-06,
      "loss": 2.0681,
      "step": 2886
    },
    {
      "epoch": 2.4466101694915254,
      "grad_norm": 2.1916449069976807,
      "learning_rate": 9.22316384180791e-06,
      "loss": 1.9212,
      "step": 2887
    },
    {
      "epoch": 2.447457627118644,
      "grad_norm": 1.7218174934387207,
      "learning_rate": 9.2090395480226e-06,
      "loss": 2.1402,
      "step": 2888
    },
    {
      "epoch": 2.448305084745763,
      "grad_norm": 1.666730284690857,
      "learning_rate": 9.194915254237288e-06,
      "loss": 2.2446,
      "step": 2889
    },
    {
      "epoch": 2.4491525423728815,
      "grad_norm": 2.093885660171509,
      "learning_rate": 9.180790960451978e-06,
      "loss": 1.6503,
      "step": 2890
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.076740026473999,
      "learning_rate": 9.166666666666666e-06,
      "loss": 1.7412,
      "step": 2891
    },
    {
      "epoch": 2.450847457627119,
      "grad_norm": 1.4801702499389648,
      "learning_rate": 9.152542372881356e-06,
      "loss": 2.4669,
      "step": 2892
    },
    {
      "epoch": 2.451694915254237,
      "grad_norm": 1.8929673433303833,
      "learning_rate": 9.138418079096045e-06,
      "loss": 1.9661,
      "step": 2893
    },
    {
      "epoch": 2.452542372881356,
      "grad_norm": 1.5265833139419556,
      "learning_rate": 9.124293785310735e-06,
      "loss": 2.2726,
      "step": 2894
    },
    {
      "epoch": 2.4533898305084745,
      "grad_norm": 1.1610336303710938,
      "learning_rate": 9.110169491525423e-06,
      "loss": 2.4884,
      "step": 2895
    },
    {
      "epoch": 2.454237288135593,
      "grad_norm": 1.2153184413909912,
      "learning_rate": 9.096045197740113e-06,
      "loss": 2.4264,
      "step": 2896
    },
    {
      "epoch": 2.455084745762712,
      "grad_norm": 1.9768568277359009,
      "learning_rate": 9.081920903954803e-06,
      "loss": 2.0703,
      "step": 2897
    },
    {
      "epoch": 2.4559322033898305,
      "grad_norm": 1.6819117069244385,
      "learning_rate": 9.067796610169491e-06,
      "loss": 2.1022,
      "step": 2898
    },
    {
      "epoch": 2.4567796610169492,
      "grad_norm": 1.7616136074066162,
      "learning_rate": 9.053672316384181e-06,
      "loss": 2.1575,
      "step": 2899
    },
    {
      "epoch": 2.457627118644068,
      "grad_norm": 1.7328168153762817,
      "learning_rate": 9.039548022598871e-06,
      "loss": 2.0504,
      "step": 2900
    },
    {
      "epoch": 2.4584745762711866,
      "grad_norm": 1.7471535205841064,
      "learning_rate": 9.02542372881356e-06,
      "loss": 2.061,
      "step": 2901
    },
    {
      "epoch": 2.459322033898305,
      "grad_norm": 1.6638541221618652,
      "learning_rate": 9.01129943502825e-06,
      "loss": 2.3405,
      "step": 2902
    },
    {
      "epoch": 2.4601694915254235,
      "grad_norm": 1.5824464559555054,
      "learning_rate": 8.99717514124294e-06,
      "loss": 2.1088,
      "step": 2903
    },
    {
      "epoch": 2.461016949152542,
      "grad_norm": 1.5538772344589233,
      "learning_rate": 8.983050847457628e-06,
      "loss": 2.1619,
      "step": 2904
    },
    {
      "epoch": 2.461864406779661,
      "grad_norm": 2.0436947345733643,
      "learning_rate": 8.968926553672318e-06,
      "loss": 1.7712,
      "step": 2905
    },
    {
      "epoch": 2.4627118644067796,
      "grad_norm": 1.3412665128707886,
      "learning_rate": 8.954802259887006e-06,
      "loss": 2.4668,
      "step": 2906
    },
    {
      "epoch": 2.4635593220338983,
      "grad_norm": 1.483189344406128,
      "learning_rate": 8.940677966101696e-06,
      "loss": 2.3362,
      "step": 2907
    },
    {
      "epoch": 2.464406779661017,
      "grad_norm": 1.8103327751159668,
      "learning_rate": 8.926553672316384e-06,
      "loss": 2.0242,
      "step": 2908
    },
    {
      "epoch": 2.4652542372881356,
      "grad_norm": 2.065345048904419,
      "learning_rate": 8.912429378531074e-06,
      "loss": 2.0116,
      "step": 2909
    },
    {
      "epoch": 2.4661016949152543,
      "grad_norm": 1.9503544569015503,
      "learning_rate": 8.898305084745763e-06,
      "loss": 2.0927,
      "step": 2910
    },
    {
      "epoch": 2.466949152542373,
      "grad_norm": 1.9558987617492676,
      "learning_rate": 8.884180790960453e-06,
      "loss": 2.0382,
      "step": 2911
    },
    {
      "epoch": 2.4677966101694917,
      "grad_norm": 1.5871424674987793,
      "learning_rate": 8.870056497175141e-06,
      "loss": 2.2568,
      "step": 2912
    },
    {
      "epoch": 2.4686440677966104,
      "grad_norm": 1.7679028511047363,
      "learning_rate": 8.855932203389831e-06,
      "loss": 1.9557,
      "step": 2913
    },
    {
      "epoch": 2.4694915254237286,
      "grad_norm": 1.8938539028167725,
      "learning_rate": 8.84180790960452e-06,
      "loss": 2.0015,
      "step": 2914
    },
    {
      "epoch": 2.4703389830508473,
      "grad_norm": 1.2811024188995361,
      "learning_rate": 8.82768361581921e-06,
      "loss": 2.4041,
      "step": 2915
    },
    {
      "epoch": 2.471186440677966,
      "grad_norm": 1.781759262084961,
      "learning_rate": 8.8135593220339e-06,
      "loss": 2.0379,
      "step": 2916
    },
    {
      "epoch": 2.4720338983050847,
      "grad_norm": 1.2422384023666382,
      "learning_rate": 8.79943502824859e-06,
      "loss": 2.4718,
      "step": 2917
    },
    {
      "epoch": 2.4728813559322034,
      "grad_norm": 1.4767169952392578,
      "learning_rate": 8.785310734463278e-06,
      "loss": 2.4039,
      "step": 2918
    },
    {
      "epoch": 2.473728813559322,
      "grad_norm": 1.2458562850952148,
      "learning_rate": 8.771186440677968e-06,
      "loss": 2.5647,
      "step": 2919
    },
    {
      "epoch": 2.4745762711864407,
      "grad_norm": 2.1160736083984375,
      "learning_rate": 8.757062146892656e-06,
      "loss": 1.7902,
      "step": 2920
    },
    {
      "epoch": 2.4754237288135594,
      "grad_norm": 1.473446011543274,
      "learning_rate": 8.742937853107346e-06,
      "loss": 2.2131,
      "step": 2921
    },
    {
      "epoch": 2.476271186440678,
      "grad_norm": 1.4707233905792236,
      "learning_rate": 8.728813559322034e-06,
      "loss": 2.1728,
      "step": 2922
    },
    {
      "epoch": 2.477118644067797,
      "grad_norm": 1.497360348701477,
      "learning_rate": 8.714689265536724e-06,
      "loss": 2.2662,
      "step": 2923
    },
    {
      "epoch": 2.477966101694915,
      "grad_norm": 1.9382723569869995,
      "learning_rate": 8.700564971751413e-06,
      "loss": 1.8265,
      "step": 2924
    },
    {
      "epoch": 2.4788135593220337,
      "grad_norm": 1.7174965143203735,
      "learning_rate": 8.686440677966103e-06,
      "loss": 2.1343,
      "step": 2925
    },
    {
      "epoch": 2.4796610169491524,
      "grad_norm": 1.9615626335144043,
      "learning_rate": 8.67231638418079e-06,
      "loss": 2.1523,
      "step": 2926
    },
    {
      "epoch": 2.480508474576271,
      "grad_norm": 2.1843740940093994,
      "learning_rate": 8.65819209039548e-06,
      "loss": 1.8332,
      "step": 2927
    },
    {
      "epoch": 2.48135593220339,
      "grad_norm": 1.793603539466858,
      "learning_rate": 8.644067796610169e-06,
      "loss": 2.2019,
      "step": 2928
    },
    {
      "epoch": 2.4822033898305085,
      "grad_norm": 2.0399248600006104,
      "learning_rate": 8.629943502824859e-06,
      "loss": 2.0296,
      "step": 2929
    },
    {
      "epoch": 2.483050847457627,
      "grad_norm": 2.1520917415618896,
      "learning_rate": 8.615819209039547e-06,
      "loss": 2.2958,
      "step": 2930
    },
    {
      "epoch": 2.483898305084746,
      "grad_norm": 1.677255630493164,
      "learning_rate": 8.601694915254237e-06,
      "loss": 2.2038,
      "step": 2931
    },
    {
      "epoch": 2.4847457627118645,
      "grad_norm": 1.4223687648773193,
      "learning_rate": 8.587570621468927e-06,
      "loss": 2.4918,
      "step": 2932
    },
    {
      "epoch": 2.4855932203389832,
      "grad_norm": 1.4818404912948608,
      "learning_rate": 8.573446327683617e-06,
      "loss": 2.4961,
      "step": 2933
    },
    {
      "epoch": 2.486440677966102,
      "grad_norm": 1.8745479583740234,
      "learning_rate": 8.559322033898306e-06,
      "loss": 2.1015,
      "step": 2934
    },
    {
      "epoch": 2.48728813559322,
      "grad_norm": 2.2350783348083496,
      "learning_rate": 8.545197740112996e-06,
      "loss": 2.0184,
      "step": 2935
    },
    {
      "epoch": 2.488135593220339,
      "grad_norm": 1.8060007095336914,
      "learning_rate": 8.531073446327684e-06,
      "loss": 2.095,
      "step": 2936
    },
    {
      "epoch": 2.4889830508474575,
      "grad_norm": 1.8137600421905518,
      "learning_rate": 8.516949152542374e-06,
      "loss": 2.0185,
      "step": 2937
    },
    {
      "epoch": 2.489830508474576,
      "grad_norm": 1.374155879020691,
      "learning_rate": 8.502824858757062e-06,
      "loss": 2.3764,
      "step": 2938
    },
    {
      "epoch": 2.490677966101695,
      "grad_norm": 1.761371374130249,
      "learning_rate": 8.488700564971752e-06,
      "loss": 2.0937,
      "step": 2939
    },
    {
      "epoch": 2.4915254237288136,
      "grad_norm": 2.1312267780303955,
      "learning_rate": 8.47457627118644e-06,
      "loss": 1.8166,
      "step": 2940
    },
    {
      "epoch": 2.4923728813559323,
      "grad_norm": 1.6262218952178955,
      "learning_rate": 8.46045197740113e-06,
      "loss": 2.2838,
      "step": 2941
    },
    {
      "epoch": 2.493220338983051,
      "grad_norm": 1.7206265926361084,
      "learning_rate": 8.446327683615819e-06,
      "loss": 2.0711,
      "step": 2942
    },
    {
      "epoch": 2.4940677966101696,
      "grad_norm": 1.7607817649841309,
      "learning_rate": 8.432203389830509e-06,
      "loss": 2.0993,
      "step": 2943
    },
    {
      "epoch": 2.4949152542372883,
      "grad_norm": 2.1898818016052246,
      "learning_rate": 8.418079096045197e-06,
      "loss": 1.6373,
      "step": 2944
    },
    {
      "epoch": 2.4957627118644066,
      "grad_norm": 1.7678654193878174,
      "learning_rate": 8.403954802259887e-06,
      "loss": 2.1646,
      "step": 2945
    },
    {
      "epoch": 2.4966101694915253,
      "grad_norm": 1.371290922164917,
      "learning_rate": 8.389830508474575e-06,
      "loss": 2.3973,
      "step": 2946
    },
    {
      "epoch": 2.497457627118644,
      "grad_norm": 1.9622776508331299,
      "learning_rate": 8.375706214689265e-06,
      "loss": 1.8268,
      "step": 2947
    },
    {
      "epoch": 2.4983050847457626,
      "grad_norm": 1.8486266136169434,
      "learning_rate": 8.361581920903955e-06,
      "loss": 2.1427,
      "step": 2948
    },
    {
      "epoch": 2.4991525423728813,
      "grad_norm": 1.2044062614440918,
      "learning_rate": 8.347457627118645e-06,
      "loss": 2.6267,
      "step": 2949
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.120523452758789,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.887,
      "step": 2950
    },
    {
      "epoch": 2.5008474576271187,
      "grad_norm": 1.5874050855636597,
      "learning_rate": 8.319209039548024e-06,
      "loss": 2.303,
      "step": 2951
    },
    {
      "epoch": 2.5016949152542374,
      "grad_norm": 1.2719577550888062,
      "learning_rate": 8.305084745762712e-06,
      "loss": 2.3387,
      "step": 2952
    },
    {
      "epoch": 2.502542372881356,
      "grad_norm": 1.9110260009765625,
      "learning_rate": 8.290960451977402e-06,
      "loss": 2.3022,
      "step": 2953
    },
    {
      "epoch": 2.5033898305084747,
      "grad_norm": 1.998567819595337,
      "learning_rate": 8.27683615819209e-06,
      "loss": 2.0504,
      "step": 2954
    },
    {
      "epoch": 2.5042372881355934,
      "grad_norm": 1.2988436222076416,
      "learning_rate": 8.26271186440678e-06,
      "loss": 2.4504,
      "step": 2955
    },
    {
      "epoch": 2.505084745762712,
      "grad_norm": 1.398335576057434,
      "learning_rate": 8.248587570621469e-06,
      "loss": 2.273,
      "step": 2956
    },
    {
      "epoch": 2.5059322033898304,
      "grad_norm": 1.8977516889572144,
      "learning_rate": 8.234463276836159e-06,
      "loss": 2.1891,
      "step": 2957
    },
    {
      "epoch": 2.506779661016949,
      "grad_norm": 1.8704180717468262,
      "learning_rate": 8.220338983050847e-06,
      "loss": 2.0068,
      "step": 2958
    },
    {
      "epoch": 2.5076271186440677,
      "grad_norm": 1.4644842147827148,
      "learning_rate": 8.206214689265537e-06,
      "loss": 2.2268,
      "step": 2959
    },
    {
      "epoch": 2.5084745762711864,
      "grad_norm": 1.4621210098266602,
      "learning_rate": 8.192090395480225e-06,
      "loss": 2.3169,
      "step": 2960
    },
    {
      "epoch": 2.509322033898305,
      "grad_norm": 1.5093555450439453,
      "learning_rate": 8.177966101694915e-06,
      "loss": 2.3677,
      "step": 2961
    },
    {
      "epoch": 2.510169491525424,
      "grad_norm": 1.8150832653045654,
      "learning_rate": 8.163841807909603e-06,
      "loss": 1.9276,
      "step": 2962
    },
    {
      "epoch": 2.5110169491525425,
      "grad_norm": 1.834709644317627,
      "learning_rate": 8.149717514124293e-06,
      "loss": 2.046,
      "step": 2963
    },
    {
      "epoch": 2.511864406779661,
      "grad_norm": 2.189105749130249,
      "learning_rate": 8.135593220338983e-06,
      "loss": 1.4947,
      "step": 2964
    },
    {
      "epoch": 2.5127118644067794,
      "grad_norm": 1.4535638093948364,
      "learning_rate": 8.121468926553673e-06,
      "loss": 2.1777,
      "step": 2965
    },
    {
      "epoch": 2.513559322033898,
      "grad_norm": 1.4188885688781738,
      "learning_rate": 8.107344632768362e-06,
      "loss": 2.4833,
      "step": 2966
    },
    {
      "epoch": 2.5144067796610168,
      "grad_norm": 1.7854934930801392,
      "learning_rate": 8.093220338983052e-06,
      "loss": 2.098,
      "step": 2967
    },
    {
      "epoch": 2.5152542372881355,
      "grad_norm": 1.513467788696289,
      "learning_rate": 8.07909604519774e-06,
      "loss": 2.2624,
      "step": 2968
    },
    {
      "epoch": 2.516101694915254,
      "grad_norm": 1.996180534362793,
      "learning_rate": 8.06497175141243e-06,
      "loss": 1.7446,
      "step": 2969
    },
    {
      "epoch": 2.516949152542373,
      "grad_norm": 2.0471272468566895,
      "learning_rate": 8.050847457627118e-06,
      "loss": 1.782,
      "step": 2970
    },
    {
      "epoch": 2.5177966101694915,
      "grad_norm": 1.7185624837875366,
      "learning_rate": 8.036723163841808e-06,
      "loss": 2.1194,
      "step": 2971
    },
    {
      "epoch": 2.51864406779661,
      "grad_norm": 1.852493166923523,
      "learning_rate": 8.022598870056497e-06,
      "loss": 1.9063,
      "step": 2972
    },
    {
      "epoch": 2.519491525423729,
      "grad_norm": 1.630000114440918,
      "learning_rate": 8.008474576271187e-06,
      "loss": 2.2022,
      "step": 2973
    },
    {
      "epoch": 2.5203389830508476,
      "grad_norm": 1.9185245037078857,
      "learning_rate": 7.994350282485875e-06,
      "loss": 2.1009,
      "step": 2974
    },
    {
      "epoch": 2.5211864406779663,
      "grad_norm": 2.0377321243286133,
      "learning_rate": 7.980225988700565e-06,
      "loss": 1.938,
      "step": 2975
    },
    {
      "epoch": 2.522033898305085,
      "grad_norm": 1.6007628440856934,
      "learning_rate": 7.966101694915253e-06,
      "loss": 2.1577,
      "step": 2976
    },
    {
      "epoch": 2.5228813559322036,
      "grad_norm": 2.1180145740509033,
      "learning_rate": 7.951977401129943e-06,
      "loss": 1.9347,
      "step": 2977
    },
    {
      "epoch": 2.523728813559322,
      "grad_norm": 1.706475853919983,
      "learning_rate": 7.937853107344633e-06,
      "loss": 2.1324,
      "step": 2978
    },
    {
      "epoch": 2.5245762711864406,
      "grad_norm": 1.9764870405197144,
      "learning_rate": 7.923728813559323e-06,
      "loss": 1.9918,
      "step": 2979
    },
    {
      "epoch": 2.5254237288135593,
      "grad_norm": 1.5780428647994995,
      "learning_rate": 7.909604519774012e-06,
      "loss": 2.238,
      "step": 2980
    },
    {
      "epoch": 2.526271186440678,
      "grad_norm": 2.1836647987365723,
      "learning_rate": 7.895480225988702e-06,
      "loss": 1.6425,
      "step": 2981
    },
    {
      "epoch": 2.5271186440677966,
      "grad_norm": 1.736869215965271,
      "learning_rate": 7.881355932203392e-06,
      "loss": 2.056,
      "step": 2982
    },
    {
      "epoch": 2.5279661016949153,
      "grad_norm": 1.3528568744659424,
      "learning_rate": 7.86723163841808e-06,
      "loss": 2.3249,
      "step": 2983
    },
    {
      "epoch": 2.528813559322034,
      "grad_norm": 1.4063446521759033,
      "learning_rate": 7.85310734463277e-06,
      "loss": 2.3622,
      "step": 2984
    },
    {
      "epoch": 2.5296610169491527,
      "grad_norm": 1.8569039106369019,
      "learning_rate": 7.838983050847458e-06,
      "loss": 2.0365,
      "step": 2985
    },
    {
      "epoch": 2.530508474576271,
      "grad_norm": 1.8357794284820557,
      "learning_rate": 7.824858757062148e-06,
      "loss": 1.8323,
      "step": 2986
    },
    {
      "epoch": 2.5313559322033896,
      "grad_norm": 1.867386817932129,
      "learning_rate": 7.810734463276836e-06,
      "loss": 1.8655,
      "step": 2987
    },
    {
      "epoch": 2.5322033898305083,
      "grad_norm": 1.9853345155715942,
      "learning_rate": 7.796610169491526e-06,
      "loss": 1.9758,
      "step": 2988
    },
    {
      "epoch": 2.533050847457627,
      "grad_norm": 1.244191288948059,
      "learning_rate": 7.782485875706215e-06,
      "loss": 2.3999,
      "step": 2989
    },
    {
      "epoch": 2.5338983050847457,
      "grad_norm": 1.669678807258606,
      "learning_rate": 7.768361581920905e-06,
      "loss": 1.9253,
      "step": 2990
    },
    {
      "epoch": 2.5347457627118644,
      "grad_norm": 2.0847461223602295,
      "learning_rate": 7.754237288135593e-06,
      "loss": 2.0371,
      "step": 2991
    },
    {
      "epoch": 2.535593220338983,
      "grad_norm": 1.7110786437988281,
      "learning_rate": 7.740112994350283e-06,
      "loss": 2.207,
      "step": 2992
    },
    {
      "epoch": 2.5364406779661017,
      "grad_norm": 1.4931682348251343,
      "learning_rate": 7.725988700564971e-06,
      "loss": 2.2115,
      "step": 2993
    },
    {
      "epoch": 2.5372881355932204,
      "grad_norm": 1.9188610315322876,
      "learning_rate": 7.711864406779661e-06,
      "loss": 1.7046,
      "step": 2994
    },
    {
      "epoch": 2.538135593220339,
      "grad_norm": 2.1624057292938232,
      "learning_rate": 7.697740112994351e-06,
      "loss": 1.8313,
      "step": 2995
    },
    {
      "epoch": 2.538983050847458,
      "grad_norm": 1.8246220350265503,
      "learning_rate": 7.683615819209041e-06,
      "loss": 1.9271,
      "step": 2996
    },
    {
      "epoch": 2.5398305084745765,
      "grad_norm": 2.056644916534424,
      "learning_rate": 7.66949152542373e-06,
      "loss": 1.677,
      "step": 2997
    },
    {
      "epoch": 2.540677966101695,
      "grad_norm": 2.052074670791626,
      "learning_rate": 7.65536723163842e-06,
      "loss": 1.76,
      "step": 2998
    },
    {
      "epoch": 2.5415254237288134,
      "grad_norm": 1.3644156455993652,
      "learning_rate": 7.641242937853108e-06,
      "loss": 2.3393,
      "step": 2999
    },
    {
      "epoch": 2.542372881355932,
      "grad_norm": 1.7320497035980225,
      "learning_rate": 7.627118644067798e-06,
      "loss": 2.1087,
      "step": 3000
    },
    {
      "epoch": 2.5432203389830508,
      "grad_norm": 1.5821408033370972,
      "learning_rate": 7.612994350282486e-06,
      "loss": 2.1904,
      "step": 3001
    },
    {
      "epoch": 2.5440677966101695,
      "grad_norm": 1.8494126796722412,
      "learning_rate": 7.598870056497176e-06,
      "loss": 2.0945,
      "step": 3002
    },
    {
      "epoch": 2.544915254237288,
      "grad_norm": 1.8187628984451294,
      "learning_rate": 7.5847457627118645e-06,
      "loss": 2.1857,
      "step": 3003
    },
    {
      "epoch": 2.545762711864407,
      "grad_norm": 1.6754740476608276,
      "learning_rate": 7.5706214689265545e-06,
      "loss": 1.7984,
      "step": 3004
    },
    {
      "epoch": 2.5466101694915255,
      "grad_norm": 1.7630802392959595,
      "learning_rate": 7.556497175141243e-06,
      "loss": 2.0151,
      "step": 3005
    },
    {
      "epoch": 2.547457627118644,
      "grad_norm": 2.1192195415496826,
      "learning_rate": 7.542372881355933e-06,
      "loss": 2.0071,
      "step": 3006
    },
    {
      "epoch": 2.5483050847457624,
      "grad_norm": 1.8960492610931396,
      "learning_rate": 7.528248587570622e-06,
      "loss": 2.0779,
      "step": 3007
    },
    {
      "epoch": 2.549152542372881,
      "grad_norm": 1.5422511100769043,
      "learning_rate": 7.514124293785312e-06,
      "loss": 2.2483,
      "step": 3008
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4536371231079102,
      "learning_rate": 7.5e-06,
      "loss": 2.2602,
      "step": 3009
    },
    {
      "epoch": 2.5508474576271185,
      "grad_norm": 1.515049695968628,
      "learning_rate": 7.48587570621469e-06,
      "loss": 2.3488,
      "step": 3010
    },
    {
      "epoch": 2.551694915254237,
      "grad_norm": 1.879209280014038,
      "learning_rate": 7.4717514124293785e-06,
      "loss": 1.8461,
      "step": 3011
    },
    {
      "epoch": 2.552542372881356,
      "grad_norm": 1.7145729064941406,
      "learning_rate": 7.4576271186440685e-06,
      "loss": 2.1602,
      "step": 3012
    },
    {
      "epoch": 2.5533898305084746,
      "grad_norm": 1.601664662361145,
      "learning_rate": 7.443502824858757e-06,
      "loss": 2.106,
      "step": 3013
    },
    {
      "epoch": 2.5542372881355933,
      "grad_norm": 1.810798168182373,
      "learning_rate": 7.429378531073447e-06,
      "loss": 2.2859,
      "step": 3014
    },
    {
      "epoch": 2.555084745762712,
      "grad_norm": 1.6008425951004028,
      "learning_rate": 7.415254237288136e-06,
      "loss": 2.1538,
      "step": 3015
    },
    {
      "epoch": 2.5559322033898306,
      "grad_norm": 1.68952476978302,
      "learning_rate": 7.401129943502826e-06,
      "loss": 2.0608,
      "step": 3016
    },
    {
      "epoch": 2.5567796610169493,
      "grad_norm": 1.606894850730896,
      "learning_rate": 7.387005649717514e-06,
      "loss": 2.2093,
      "step": 3017
    },
    {
      "epoch": 2.557627118644068,
      "grad_norm": 1.3230990171432495,
      "learning_rate": 7.372881355932204e-06,
      "loss": 2.3442,
      "step": 3018
    },
    {
      "epoch": 2.5584745762711867,
      "grad_norm": 1.8088946342468262,
      "learning_rate": 7.3587570621468925e-06,
      "loss": 2.0649,
      "step": 3019
    },
    {
      "epoch": 2.559322033898305,
      "grad_norm": 1.5568079948425293,
      "learning_rate": 7.3446327683615825e-06,
      "loss": 2.2807,
      "step": 3020
    },
    {
      "epoch": 2.5601694915254236,
      "grad_norm": 2.1737327575683594,
      "learning_rate": 7.330508474576271e-06,
      "loss": 1.6519,
      "step": 3021
    },
    {
      "epoch": 2.5610169491525423,
      "grad_norm": 1.812609076499939,
      "learning_rate": 7.316384180790961e-06,
      "loss": 2.1693,
      "step": 3022
    },
    {
      "epoch": 2.561864406779661,
      "grad_norm": 1.5055807828903198,
      "learning_rate": 7.30225988700565e-06,
      "loss": 2.3576,
      "step": 3023
    },
    {
      "epoch": 2.5627118644067797,
      "grad_norm": 2.2145583629608154,
      "learning_rate": 7.28813559322034e-06,
      "loss": 1.9627,
      "step": 3024
    },
    {
      "epoch": 2.5635593220338984,
      "grad_norm": 1.6984370946884155,
      "learning_rate": 7.274011299435028e-06,
      "loss": 2.0201,
      "step": 3025
    },
    {
      "epoch": 2.564406779661017,
      "grad_norm": 1.5419265031814575,
      "learning_rate": 7.259887005649718e-06,
      "loss": 2.2826,
      "step": 3026
    },
    {
      "epoch": 2.5652542372881357,
      "grad_norm": 2.3023569583892822,
      "learning_rate": 7.2457627118644065e-06,
      "loss": 1.592,
      "step": 3027
    },
    {
      "epoch": 2.566101694915254,
      "grad_norm": 1.7549352645874023,
      "learning_rate": 7.2316384180790965e-06,
      "loss": 2.058,
      "step": 3028
    },
    {
      "epoch": 2.5669491525423727,
      "grad_norm": 1.3474128246307373,
      "learning_rate": 7.217514124293785e-06,
      "loss": 2.3685,
      "step": 3029
    },
    {
      "epoch": 2.5677966101694913,
      "grad_norm": 1.5920982360839844,
      "learning_rate": 7.203389830508475e-06,
      "loss": 2.1354,
      "step": 3030
    },
    {
      "epoch": 2.56864406779661,
      "grad_norm": 1.834998607635498,
      "learning_rate": 7.189265536723164e-06,
      "loss": 2.0803,
      "step": 3031
    },
    {
      "epoch": 2.5694915254237287,
      "grad_norm": 1.6628954410552979,
      "learning_rate": 7.175141242937854e-06,
      "loss": 2.0229,
      "step": 3032
    },
    {
      "epoch": 2.5703389830508474,
      "grad_norm": 1.7421000003814697,
      "learning_rate": 7.161016949152542e-06,
      "loss": 2.2294,
      "step": 3033
    },
    {
      "epoch": 2.571186440677966,
      "grad_norm": 2.0664875507354736,
      "learning_rate": 7.146892655367232e-06,
      "loss": 1.8082,
      "step": 3034
    },
    {
      "epoch": 2.5720338983050848,
      "grad_norm": 1.7396173477172852,
      "learning_rate": 7.1327683615819206e-06,
      "loss": 2.1463,
      "step": 3035
    },
    {
      "epoch": 2.5728813559322035,
      "grad_norm": 1.789170503616333,
      "learning_rate": 7.1186440677966106e-06,
      "loss": 2.019,
      "step": 3036
    },
    {
      "epoch": 2.573728813559322,
      "grad_norm": 1.9018986225128174,
      "learning_rate": 7.104519774011299e-06,
      "loss": 2.0897,
      "step": 3037
    },
    {
      "epoch": 2.574576271186441,
      "grad_norm": 1.6295945644378662,
      "learning_rate": 7.090395480225989e-06,
      "loss": 1.898,
      "step": 3038
    },
    {
      "epoch": 2.5754237288135595,
      "grad_norm": 1.4587124586105347,
      "learning_rate": 7.076271186440678e-06,
      "loss": 2.4802,
      "step": 3039
    },
    {
      "epoch": 2.576271186440678,
      "grad_norm": 2.1292927265167236,
      "learning_rate": 7.062146892655368e-06,
      "loss": 1.8714,
      "step": 3040
    },
    {
      "epoch": 2.5771186440677964,
      "grad_norm": 1.8581408262252808,
      "learning_rate": 7.048022598870056e-06,
      "loss": 1.9625,
      "step": 3041
    },
    {
      "epoch": 2.577966101694915,
      "grad_norm": 1.1935030221939087,
      "learning_rate": 7.033898305084746e-06,
      "loss": 2.4388,
      "step": 3042
    },
    {
      "epoch": 2.578813559322034,
      "grad_norm": 1.8859221935272217,
      "learning_rate": 7.019774011299435e-06,
      "loss": 1.939,
      "step": 3043
    },
    {
      "epoch": 2.5796610169491525,
      "grad_norm": 1.7914434671401978,
      "learning_rate": 7.0056497175141246e-06,
      "loss": 2.0808,
      "step": 3044
    },
    {
      "epoch": 2.580508474576271,
      "grad_norm": 1.882596731185913,
      "learning_rate": 6.991525423728814e-06,
      "loss": 1.8391,
      "step": 3045
    },
    {
      "epoch": 2.58135593220339,
      "grad_norm": 2.0972578525543213,
      "learning_rate": 6.977401129943503e-06,
      "loss": 1.9076,
      "step": 3046
    },
    {
      "epoch": 2.5822033898305086,
      "grad_norm": 1.6736787557601929,
      "learning_rate": 6.963276836158192e-06,
      "loss": 2.3895,
      "step": 3047
    },
    {
      "epoch": 2.5830508474576273,
      "grad_norm": 2.118112802505493,
      "learning_rate": 6.949152542372882e-06,
      "loss": 1.8,
      "step": 3048
    },
    {
      "epoch": 2.5838983050847455,
      "grad_norm": 1.4006154537200928,
      "learning_rate": 6.93502824858757e-06,
      "loss": 2.2656,
      "step": 3049
    },
    {
      "epoch": 2.584745762711864,
      "grad_norm": 1.7507838010787964,
      "learning_rate": 6.92090395480226e-06,
      "loss": 1.9411,
      "step": 3050
    },
    {
      "epoch": 2.585593220338983,
      "grad_norm": 1.6378151178359985,
      "learning_rate": 6.906779661016949e-06,
      "loss": 2.1043,
      "step": 3051
    },
    {
      "epoch": 2.5864406779661016,
      "grad_norm": 2.0987699031829834,
      "learning_rate": 6.892655367231639e-06,
      "loss": 1.7536,
      "step": 3052
    },
    {
      "epoch": 2.5872881355932202,
      "grad_norm": 1.8500317335128784,
      "learning_rate": 6.878531073446328e-06,
      "loss": 1.9234,
      "step": 3053
    },
    {
      "epoch": 2.588135593220339,
      "grad_norm": 1.954512119293213,
      "learning_rate": 6.864406779661017e-06,
      "loss": 2.1282,
      "step": 3054
    },
    {
      "epoch": 2.5889830508474576,
      "grad_norm": 1.7337831258773804,
      "learning_rate": 6.850282485875706e-06,
      "loss": 2.1146,
      "step": 3055
    },
    {
      "epoch": 2.5898305084745763,
      "grad_norm": 1.5660083293914795,
      "learning_rate": 6.836158192090396e-06,
      "loss": 2.3264,
      "step": 3056
    },
    {
      "epoch": 2.590677966101695,
      "grad_norm": 1.435356855392456,
      "learning_rate": 6.822033898305084e-06,
      "loss": 2.457,
      "step": 3057
    },
    {
      "epoch": 2.5915254237288137,
      "grad_norm": 2.0490822792053223,
      "learning_rate": 6.807909604519774e-06,
      "loss": 1.7802,
      "step": 3058
    },
    {
      "epoch": 2.5923728813559324,
      "grad_norm": 1.5111957788467407,
      "learning_rate": 6.793785310734464e-06,
      "loss": 2.2187,
      "step": 3059
    },
    {
      "epoch": 2.593220338983051,
      "grad_norm": 1.6086337566375732,
      "learning_rate": 6.779661016949153e-06,
      "loss": 2.1684,
      "step": 3060
    },
    {
      "epoch": 2.5940677966101697,
      "grad_norm": 1.8838337659835815,
      "learning_rate": 6.765536723163843e-06,
      "loss": 2.1587,
      "step": 3061
    },
    {
      "epoch": 2.594915254237288,
      "grad_norm": 1.6110546588897705,
      "learning_rate": 6.751412429378532e-06,
      "loss": 2.225,
      "step": 3062
    },
    {
      "epoch": 2.5957627118644067,
      "grad_norm": 2.178858995437622,
      "learning_rate": 6.737288135593221e-06,
      "loss": 1.9288,
      "step": 3063
    },
    {
      "epoch": 2.5966101694915253,
      "grad_norm": 1.4706746339797974,
      "learning_rate": 6.72316384180791e-06,
      "loss": 2.3295,
      "step": 3064
    },
    {
      "epoch": 2.597457627118644,
      "grad_norm": 1.2675209045410156,
      "learning_rate": 6.7090395480226e-06,
      "loss": 2.4013,
      "step": 3065
    },
    {
      "epoch": 2.5983050847457627,
      "grad_norm": 1.82991361618042,
      "learning_rate": 6.694915254237288e-06,
      "loss": 2.0283,
      "step": 3066
    },
    {
      "epoch": 2.5991525423728814,
      "grad_norm": 1.9969542026519775,
      "learning_rate": 6.680790960451978e-06,
      "loss": 1.9261,
      "step": 3067
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5021088123321533,
      "learning_rate": 6.666666666666667e-06,
      "loss": 2.3465,
      "step": 3068
    },
    {
      "epoch": 2.6008474576271188,
      "grad_norm": 1.1992968320846558,
      "learning_rate": 6.652542372881357e-06,
      "loss": 2.4956,
      "step": 3069
    },
    {
      "epoch": 2.601694915254237,
      "grad_norm": 1.1849461793899536,
      "learning_rate": 6.638418079096046e-06,
      "loss": 2.6337,
      "step": 3070
    },
    {
      "epoch": 2.6025423728813557,
      "grad_norm": 1.8810172080993652,
      "learning_rate": 6.624293785310735e-06,
      "loss": 2.0795,
      "step": 3071
    },
    {
      "epoch": 2.6033898305084744,
      "grad_norm": 1.9608169794082642,
      "learning_rate": 6.610169491525424e-06,
      "loss": 2.0597,
      "step": 3072
    },
    {
      "epoch": 2.604237288135593,
      "grad_norm": 1.8074970245361328,
      "learning_rate": 6.596045197740114e-06,
      "loss": 1.9833,
      "step": 3073
    },
    {
      "epoch": 2.6050847457627118,
      "grad_norm": 1.6962586641311646,
      "learning_rate": 6.581920903954802e-06,
      "loss": 2.2267,
      "step": 3074
    },
    {
      "epoch": 2.6059322033898304,
      "grad_norm": 1.7375556230545044,
      "learning_rate": 6.567796610169492e-06,
      "loss": 2.101,
      "step": 3075
    },
    {
      "epoch": 2.606779661016949,
      "grad_norm": 1.7727997303009033,
      "learning_rate": 6.553672316384181e-06,
      "loss": 2.036,
      "step": 3076
    },
    {
      "epoch": 2.607627118644068,
      "grad_norm": 1.2793000936508179,
      "learning_rate": 6.539548022598871e-06,
      "loss": 2.4457,
      "step": 3077
    },
    {
      "epoch": 2.6084745762711865,
      "grad_norm": 1.711143136024475,
      "learning_rate": 6.52542372881356e-06,
      "loss": 2.1449,
      "step": 3078
    },
    {
      "epoch": 2.609322033898305,
      "grad_norm": 2.098222255706787,
      "learning_rate": 6.51129943502825e-06,
      "loss": 1.7625,
      "step": 3079
    },
    {
      "epoch": 2.610169491525424,
      "grad_norm": 2.062216281890869,
      "learning_rate": 6.497175141242938e-06,
      "loss": 1.8931,
      "step": 3080
    },
    {
      "epoch": 2.6110169491525426,
      "grad_norm": 2.228379487991333,
      "learning_rate": 6.483050847457628e-06,
      "loss": 1.8649,
      "step": 3081
    },
    {
      "epoch": 2.6118644067796613,
      "grad_norm": 1.9092179536819458,
      "learning_rate": 6.468926553672316e-06,
      "loss": 1.985,
      "step": 3082
    },
    {
      "epoch": 2.61271186440678,
      "grad_norm": 1.9925532341003418,
      "learning_rate": 6.454802259887006e-06,
      "loss": 2.0447,
      "step": 3083
    },
    {
      "epoch": 2.613559322033898,
      "grad_norm": 1.8609484434127808,
      "learning_rate": 6.440677966101695e-06,
      "loss": 2.1578,
      "step": 3084
    },
    {
      "epoch": 2.614406779661017,
      "grad_norm": 1.516598105430603,
      "learning_rate": 6.426553672316385e-06,
      "loss": 2.2194,
      "step": 3085
    },
    {
      "epoch": 2.6152542372881356,
      "grad_norm": 1.7448123693466187,
      "learning_rate": 6.412429378531074e-06,
      "loss": 2.1428,
      "step": 3086
    },
    {
      "epoch": 2.6161016949152542,
      "grad_norm": 2.107908248901367,
      "learning_rate": 6.398305084745764e-06,
      "loss": 1.7867,
      "step": 3087
    },
    {
      "epoch": 2.616949152542373,
      "grad_norm": 1.9045662879943848,
      "learning_rate": 6.384180790960452e-06,
      "loss": 2.0377,
      "step": 3088
    },
    {
      "epoch": 2.6177966101694916,
      "grad_norm": 1.5822416543960571,
      "learning_rate": 6.370056497175142e-06,
      "loss": 2.3111,
      "step": 3089
    },
    {
      "epoch": 2.6186440677966103,
      "grad_norm": 1.8819859027862549,
      "learning_rate": 6.3559322033898304e-06,
      "loss": 2.1015,
      "step": 3090
    },
    {
      "epoch": 2.6194915254237285,
      "grad_norm": 2.2771382331848145,
      "learning_rate": 6.34180790960452e-06,
      "loss": 1.7219,
      "step": 3091
    },
    {
      "epoch": 2.6203389830508472,
      "grad_norm": 1.9226919412612915,
      "learning_rate": 6.327683615819209e-06,
      "loss": 1.934,
      "step": 3092
    },
    {
      "epoch": 2.621186440677966,
      "grad_norm": 2.1976590156555176,
      "learning_rate": 6.313559322033899e-06,
      "loss": 1.7882,
      "step": 3093
    },
    {
      "epoch": 2.6220338983050846,
      "grad_norm": 1.9256106615066528,
      "learning_rate": 6.299435028248588e-06,
      "loss": 1.9117,
      "step": 3094
    },
    {
      "epoch": 2.6228813559322033,
      "grad_norm": 2.0661087036132812,
      "learning_rate": 6.285310734463278e-06,
      "loss": 1.9503,
      "step": 3095
    },
    {
      "epoch": 2.623728813559322,
      "grad_norm": 2.0507121086120605,
      "learning_rate": 6.271186440677966e-06,
      "loss": 2.1173,
      "step": 3096
    },
    {
      "epoch": 2.6245762711864407,
      "grad_norm": 1.5374890565872192,
      "learning_rate": 6.257062146892656e-06,
      "loss": 2.2263,
      "step": 3097
    },
    {
      "epoch": 2.6254237288135593,
      "grad_norm": 1.3915599584579468,
      "learning_rate": 6.242937853107345e-06,
      "loss": 2.2953,
      "step": 3098
    },
    {
      "epoch": 2.626271186440678,
      "grad_norm": 1.79976487159729,
      "learning_rate": 6.2288135593220344e-06,
      "loss": 2.0337,
      "step": 3099
    },
    {
      "epoch": 2.6271186440677967,
      "grad_norm": 2.077894926071167,
      "learning_rate": 6.214689265536724e-06,
      "loss": 2.0656,
      "step": 3100
    },
    {
      "epoch": 2.6279661016949154,
      "grad_norm": 1.538857340812683,
      "learning_rate": 6.200564971751413e-06,
      "loss": 2.3387,
      "step": 3101
    },
    {
      "epoch": 2.628813559322034,
      "grad_norm": 1.6980963945388794,
      "learning_rate": 6.186440677966102e-06,
      "loss": 2.1492,
      "step": 3102
    },
    {
      "epoch": 2.6296610169491528,
      "grad_norm": 1.9201914072036743,
      "learning_rate": 6.172316384180792e-06,
      "loss": 2.0327,
      "step": 3103
    },
    {
      "epoch": 2.6305084745762715,
      "grad_norm": 1.8236585855484009,
      "learning_rate": 6.158192090395481e-06,
      "loss": 2.121,
      "step": 3104
    },
    {
      "epoch": 2.6313559322033897,
      "grad_norm": 2.30562162399292,
      "learning_rate": 6.14406779661017e-06,
      "loss": 2.1442,
      "step": 3105
    },
    {
      "epoch": 2.6322033898305084,
      "grad_norm": 1.5762290954589844,
      "learning_rate": 6.129943502824859e-06,
      "loss": 2.27,
      "step": 3106
    },
    {
      "epoch": 2.633050847457627,
      "grad_norm": 2.0478971004486084,
      "learning_rate": 6.1158192090395485e-06,
      "loss": 1.5608,
      "step": 3107
    },
    {
      "epoch": 2.6338983050847458,
      "grad_norm": 2.074697494506836,
      "learning_rate": 6.101694915254238e-06,
      "loss": 1.807,
      "step": 3108
    },
    {
      "epoch": 2.6347457627118644,
      "grad_norm": 1.7831599712371826,
      "learning_rate": 6.087570621468927e-06,
      "loss": 2.147,
      "step": 3109
    },
    {
      "epoch": 2.635593220338983,
      "grad_norm": 1.6036802530288696,
      "learning_rate": 6.073446327683617e-06,
      "loss": 2.1833,
      "step": 3110
    },
    {
      "epoch": 2.636440677966102,
      "grad_norm": 1.7700417041778564,
      "learning_rate": 6.059322033898306e-06,
      "loss": 2.1094,
      "step": 3111
    },
    {
      "epoch": 2.63728813559322,
      "grad_norm": 1.8165521621704102,
      "learning_rate": 6.045197740112995e-06,
      "loss": 1.9511,
      "step": 3112
    },
    {
      "epoch": 2.6381355932203387,
      "grad_norm": 1.8041558265686035,
      "learning_rate": 6.031073446327684e-06,
      "loss": 2.0213,
      "step": 3113
    },
    {
      "epoch": 2.6389830508474574,
      "grad_norm": 1.56833016872406,
      "learning_rate": 6.016949152542373e-06,
      "loss": 2.1927,
      "step": 3114
    },
    {
      "epoch": 2.639830508474576,
      "grad_norm": 1.8998544216156006,
      "learning_rate": 6.0028248587570625e-06,
      "loss": 2.1135,
      "step": 3115
    },
    {
      "epoch": 2.640677966101695,
      "grad_norm": 1.405078649520874,
      "learning_rate": 5.988700564971752e-06,
      "loss": 2.4591,
      "step": 3116
    },
    {
      "epoch": 2.6415254237288135,
      "grad_norm": 1.8880126476287842,
      "learning_rate": 5.974576271186441e-06,
      "loss": 1.8116,
      "step": 3117
    },
    {
      "epoch": 2.642372881355932,
      "grad_norm": 1.917741060256958,
      "learning_rate": 5.960451977401131e-06,
      "loss": 2.0983,
      "step": 3118
    },
    {
      "epoch": 2.643220338983051,
      "grad_norm": 1.492091178894043,
      "learning_rate": 5.94632768361582e-06,
      "loss": 2.3454,
      "step": 3119
    },
    {
      "epoch": 2.6440677966101696,
      "grad_norm": 1.7252168655395508,
      "learning_rate": 5.932203389830509e-06,
      "loss": 2.3544,
      "step": 3120
    },
    {
      "epoch": 2.6449152542372882,
      "grad_norm": 2.009967803955078,
      "learning_rate": 5.918079096045198e-06,
      "loss": 2.1103,
      "step": 3121
    },
    {
      "epoch": 2.645762711864407,
      "grad_norm": 1.4831916093826294,
      "learning_rate": 5.903954802259887e-06,
      "loss": 2.4132,
      "step": 3122
    },
    {
      "epoch": 2.6466101694915256,
      "grad_norm": 1.9462958574295044,
      "learning_rate": 5.8898305084745765e-06,
      "loss": 1.9792,
      "step": 3123
    },
    {
      "epoch": 2.6474576271186443,
      "grad_norm": 1.7218736410140991,
      "learning_rate": 5.875706214689266e-06,
      "loss": 2.1313,
      "step": 3124
    },
    {
      "epoch": 2.648305084745763,
      "grad_norm": 1.5944900512695312,
      "learning_rate": 5.861581920903955e-06,
      "loss": 2.3315,
      "step": 3125
    },
    {
      "epoch": 2.6491525423728812,
      "grad_norm": 1.5159193277359009,
      "learning_rate": 5.847457627118645e-06,
      "loss": 2.2296,
      "step": 3126
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9195917844772339,
      "learning_rate": 5.833333333333334e-06,
      "loss": 1.8245,
      "step": 3127
    },
    {
      "epoch": 2.6508474576271186,
      "grad_norm": 1.477134108543396,
      "learning_rate": 5.819209039548023e-06,
      "loss": 2.3509,
      "step": 3128
    },
    {
      "epoch": 2.6516949152542373,
      "grad_norm": 2.3269686698913574,
      "learning_rate": 5.805084745762712e-06,
      "loss": 1.9118,
      "step": 3129
    },
    {
      "epoch": 2.652542372881356,
      "grad_norm": 1.980559229850769,
      "learning_rate": 5.790960451977401e-06,
      "loss": 1.7757,
      "step": 3130
    },
    {
      "epoch": 2.6533898305084747,
      "grad_norm": 1.5287688970565796,
      "learning_rate": 5.7768361581920905e-06,
      "loss": 2.1286,
      "step": 3131
    },
    {
      "epoch": 2.6542372881355933,
      "grad_norm": 1.9623382091522217,
      "learning_rate": 5.76271186440678e-06,
      "loss": 1.7988,
      "step": 3132
    },
    {
      "epoch": 2.655084745762712,
      "grad_norm": 1.6550776958465576,
      "learning_rate": 5.748587570621469e-06,
      "loss": 2.1933,
      "step": 3133
    },
    {
      "epoch": 2.6559322033898303,
      "grad_norm": 1.4529967308044434,
      "learning_rate": 5.734463276836159e-06,
      "loss": 2.4504,
      "step": 3134
    },
    {
      "epoch": 2.656779661016949,
      "grad_norm": 1.5573678016662598,
      "learning_rate": 5.720338983050848e-06,
      "loss": 1.8805,
      "step": 3135
    },
    {
      "epoch": 2.6576271186440676,
      "grad_norm": 1.2603721618652344,
      "learning_rate": 5.706214689265537e-06,
      "loss": 2.4311,
      "step": 3136
    },
    {
      "epoch": 2.6584745762711863,
      "grad_norm": 1.6075152158737183,
      "learning_rate": 5.692090395480226e-06,
      "loss": 2.1516,
      "step": 3137
    },
    {
      "epoch": 2.659322033898305,
      "grad_norm": 1.4322408437728882,
      "learning_rate": 5.677966101694915e-06,
      "loss": 2.2252,
      "step": 3138
    },
    {
      "epoch": 2.6601694915254237,
      "grad_norm": 1.7860020399093628,
      "learning_rate": 5.6638418079096046e-06,
      "loss": 1.9393,
      "step": 3139
    },
    {
      "epoch": 2.6610169491525424,
      "grad_norm": 1.843631386756897,
      "learning_rate": 5.649717514124294e-06,
      "loss": 2.0904,
      "step": 3140
    },
    {
      "epoch": 2.661864406779661,
      "grad_norm": 2.062152147293091,
      "learning_rate": 5.635593220338984e-06,
      "loss": 2.1247,
      "step": 3141
    },
    {
      "epoch": 2.6627118644067798,
      "grad_norm": 1.5441206693649292,
      "learning_rate": 5.621468926553673e-06,
      "loss": 2.2471,
      "step": 3142
    },
    {
      "epoch": 2.6635593220338984,
      "grad_norm": 1.609926462173462,
      "learning_rate": 5.607344632768362e-06,
      "loss": 2.4538,
      "step": 3143
    },
    {
      "epoch": 2.664406779661017,
      "grad_norm": 1.4819270372390747,
      "learning_rate": 5.593220338983051e-06,
      "loss": 2.3786,
      "step": 3144
    },
    {
      "epoch": 2.665254237288136,
      "grad_norm": 1.875760555267334,
      "learning_rate": 5.57909604519774e-06,
      "loss": 2.1711,
      "step": 3145
    },
    {
      "epoch": 2.6661016949152545,
      "grad_norm": 2.2035133838653564,
      "learning_rate": 5.5649717514124294e-06,
      "loss": 2.0346,
      "step": 3146
    },
    {
      "epoch": 2.6669491525423727,
      "grad_norm": 1.886523723602295,
      "learning_rate": 5.550847457627119e-06,
      "loss": 2.1007,
      "step": 3147
    },
    {
      "epoch": 2.6677966101694914,
      "grad_norm": 2.027946949005127,
      "learning_rate": 5.536723163841808e-06,
      "loss": 2.0433,
      "step": 3148
    },
    {
      "epoch": 2.66864406779661,
      "grad_norm": 1.673337459564209,
      "learning_rate": 5.522598870056498e-06,
      "loss": 2.2676,
      "step": 3149
    },
    {
      "epoch": 2.669491525423729,
      "grad_norm": 1.5989969968795776,
      "learning_rate": 5.508474576271187e-06,
      "loss": 2.0769,
      "step": 3150
    },
    {
      "epoch": 2.6703389830508475,
      "grad_norm": 1.9226703643798828,
      "learning_rate": 5.494350282485876e-06,
      "loss": 2.0417,
      "step": 3151
    },
    {
      "epoch": 2.671186440677966,
      "grad_norm": 1.5872721672058105,
      "learning_rate": 5.480225988700565e-06,
      "loss": 2.3024,
      "step": 3152
    },
    {
      "epoch": 2.672033898305085,
      "grad_norm": 1.6903926134109497,
      "learning_rate": 5.466101694915254e-06,
      "loss": 2.015,
      "step": 3153
    },
    {
      "epoch": 2.6728813559322036,
      "grad_norm": 1.5938266515731812,
      "learning_rate": 5.4519774011299435e-06,
      "loss": 1.9506,
      "step": 3154
    },
    {
      "epoch": 2.673728813559322,
      "grad_norm": 1.930178165435791,
      "learning_rate": 5.437853107344633e-06,
      "loss": 2.0749,
      "step": 3155
    },
    {
      "epoch": 2.6745762711864405,
      "grad_norm": 1.697782039642334,
      "learning_rate": 5.423728813559322e-06,
      "loss": 1.8642,
      "step": 3156
    },
    {
      "epoch": 2.675423728813559,
      "grad_norm": 2.4836606979370117,
      "learning_rate": 5.409604519774012e-06,
      "loss": 1.6492,
      "step": 3157
    },
    {
      "epoch": 2.676271186440678,
      "grad_norm": 1.9653449058532715,
      "learning_rate": 5.395480225988701e-06,
      "loss": 1.9011,
      "step": 3158
    },
    {
      "epoch": 2.6771186440677965,
      "grad_norm": 1.3556913137435913,
      "learning_rate": 5.38135593220339e-06,
      "loss": 2.4513,
      "step": 3159
    },
    {
      "epoch": 2.6779661016949152,
      "grad_norm": 1.405932068824768,
      "learning_rate": 5.367231638418079e-06,
      "loss": 2.3086,
      "step": 3160
    },
    {
      "epoch": 2.678813559322034,
      "grad_norm": 1.8774460554122925,
      "learning_rate": 5.353107344632768e-06,
      "loss": 2.0499,
      "step": 3161
    },
    {
      "epoch": 2.6796610169491526,
      "grad_norm": 1.4726383686065674,
      "learning_rate": 5.3389830508474575e-06,
      "loss": 2.2896,
      "step": 3162
    },
    {
      "epoch": 2.6805084745762713,
      "grad_norm": 2.240968704223633,
      "learning_rate": 5.324858757062147e-06,
      "loss": 1.809,
      "step": 3163
    },
    {
      "epoch": 2.68135593220339,
      "grad_norm": 1.4824795722961426,
      "learning_rate": 5.310734463276836e-06,
      "loss": 2.391,
      "step": 3164
    },
    {
      "epoch": 2.6822033898305087,
      "grad_norm": 1.6574009656906128,
      "learning_rate": 5.296610169491526e-06,
      "loss": 2.0352,
      "step": 3165
    },
    {
      "epoch": 2.6830508474576273,
      "grad_norm": 1.8091789484024048,
      "learning_rate": 5.282485875706215e-06,
      "loss": 2.1136,
      "step": 3166
    },
    {
      "epoch": 2.683898305084746,
      "grad_norm": 1.7391529083251953,
      "learning_rate": 5.268361581920904e-06,
      "loss": 2.1191,
      "step": 3167
    },
    {
      "epoch": 2.6847457627118643,
      "grad_norm": 1.2865487337112427,
      "learning_rate": 5.254237288135593e-06,
      "loss": 2.4338,
      "step": 3168
    },
    {
      "epoch": 2.685593220338983,
      "grad_norm": 1.9282454252243042,
      "learning_rate": 5.240112994350282e-06,
      "loss": 2.0678,
      "step": 3169
    },
    {
      "epoch": 2.6864406779661016,
      "grad_norm": 1.9156838655471802,
      "learning_rate": 5.2259887005649715e-06,
      "loss": 2.0578,
      "step": 3170
    },
    {
      "epoch": 2.6872881355932203,
      "grad_norm": 1.5116440057754517,
      "learning_rate": 5.211864406779661e-06,
      "loss": 2.1673,
      "step": 3171
    },
    {
      "epoch": 2.688135593220339,
      "grad_norm": 1.5679614543914795,
      "learning_rate": 5.197740112994351e-06,
      "loss": 2.2779,
      "step": 3172
    },
    {
      "epoch": 2.6889830508474577,
      "grad_norm": 1.3827556371688843,
      "learning_rate": 5.18361581920904e-06,
      "loss": 2.4234,
      "step": 3173
    },
    {
      "epoch": 2.6898305084745764,
      "grad_norm": 2.056248426437378,
      "learning_rate": 5.169491525423729e-06,
      "loss": 1.9765,
      "step": 3174
    },
    {
      "epoch": 2.690677966101695,
      "grad_norm": 1.9912073612213135,
      "learning_rate": 5.155367231638418e-06,
      "loss": 1.8635,
      "step": 3175
    },
    {
      "epoch": 2.6915254237288133,
      "grad_norm": 2.0789661407470703,
      "learning_rate": 5.141242937853107e-06,
      "loss": 1.8654,
      "step": 3176
    },
    {
      "epoch": 2.692372881355932,
      "grad_norm": 1.617958903312683,
      "learning_rate": 5.127118644067796e-06,
      "loss": 2.3341,
      "step": 3177
    },
    {
      "epoch": 2.6932203389830507,
      "grad_norm": 1.8623449802398682,
      "learning_rate": 5.112994350282486e-06,
      "loss": 2.0931,
      "step": 3178
    },
    {
      "epoch": 2.6940677966101694,
      "grad_norm": 1.7776962518692017,
      "learning_rate": 5.0988700564971755e-06,
      "loss": 1.9738,
      "step": 3179
    },
    {
      "epoch": 2.694915254237288,
      "grad_norm": 1.5862172842025757,
      "learning_rate": 5.084745762711865e-06,
      "loss": 2.1484,
      "step": 3180
    },
    {
      "epoch": 2.6957627118644067,
      "grad_norm": 1.8234288692474365,
      "learning_rate": 5.070621468926554e-06,
      "loss": 1.9512,
      "step": 3181
    },
    {
      "epoch": 2.6966101694915254,
      "grad_norm": 1.7137696743011475,
      "learning_rate": 5.056497175141244e-06,
      "loss": 2.1188,
      "step": 3182
    },
    {
      "epoch": 2.697457627118644,
      "grad_norm": 1.5230895280838013,
      "learning_rate": 5.042372881355933e-06,
      "loss": 2.3191,
      "step": 3183
    },
    {
      "epoch": 2.698305084745763,
      "grad_norm": 1.7367206811904907,
      "learning_rate": 5.028248587570622e-06,
      "loss": 2.0591,
      "step": 3184
    },
    {
      "epoch": 2.6991525423728815,
      "grad_norm": 1.7793805599212646,
      "learning_rate": 5.014124293785311e-06,
      "loss": 2.2335,
      "step": 3185
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5189614295959473,
      "learning_rate": 5e-06,
      "loss": 2.4363,
      "step": 3186
    },
    {
      "epoch": 2.700847457627119,
      "grad_norm": 1.3897454738616943,
      "learning_rate": 4.9858757062146896e-06,
      "loss": 2.4523,
      "step": 3187
    },
    {
      "epoch": 2.7016949152542376,
      "grad_norm": 1.6624294519424438,
      "learning_rate": 4.971751412429379e-06,
      "loss": 1.985,
      "step": 3188
    },
    {
      "epoch": 2.702542372881356,
      "grad_norm": 1.8261654376983643,
      "learning_rate": 4.957627118644069e-06,
      "loss": 2.2007,
      "step": 3189
    },
    {
      "epoch": 2.7033898305084745,
      "grad_norm": 1.8060328960418701,
      "learning_rate": 4.943502824858758e-06,
      "loss": 2.1156,
      "step": 3190
    },
    {
      "epoch": 2.704237288135593,
      "grad_norm": 1.6717157363891602,
      "learning_rate": 4.929378531073447e-06,
      "loss": 2.1968,
      "step": 3191
    },
    {
      "epoch": 2.705084745762712,
      "grad_norm": 1.9698718786239624,
      "learning_rate": 4.915254237288136e-06,
      "loss": 1.9488,
      "step": 3192
    },
    {
      "epoch": 2.7059322033898305,
      "grad_norm": 1.6100242137908936,
      "learning_rate": 4.901129943502825e-06,
      "loss": 2.2336,
      "step": 3193
    },
    {
      "epoch": 2.7067796610169492,
      "grad_norm": 1.6457633972167969,
      "learning_rate": 4.8870056497175144e-06,
      "loss": 2.0917,
      "step": 3194
    },
    {
      "epoch": 2.707627118644068,
      "grad_norm": 1.5893149375915527,
      "learning_rate": 4.872881355932204e-06,
      "loss": 2.3023,
      "step": 3195
    },
    {
      "epoch": 2.7084745762711866,
      "grad_norm": 1.736437201499939,
      "learning_rate": 4.858757062146893e-06,
      "loss": 2.226,
      "step": 3196
    },
    {
      "epoch": 2.709322033898305,
      "grad_norm": 2.0657811164855957,
      "learning_rate": 4.844632768361583e-06,
      "loss": 2.1206,
      "step": 3197
    },
    {
      "epoch": 2.7101694915254235,
      "grad_norm": 1.9966922998428345,
      "learning_rate": 4.830508474576272e-06,
      "loss": 1.9236,
      "step": 3198
    },
    {
      "epoch": 2.711016949152542,
      "grad_norm": 1.7514450550079346,
      "learning_rate": 4.816384180790961e-06,
      "loss": 2.1602,
      "step": 3199
    },
    {
      "epoch": 2.711864406779661,
      "grad_norm": 1.5922361612319946,
      "learning_rate": 4.80225988700565e-06,
      "loss": 2.2008,
      "step": 3200
    },
    {
      "epoch": 2.7127118644067796,
      "grad_norm": 1.90059494972229,
      "learning_rate": 4.788135593220339e-06,
      "loss": 1.925,
      "step": 3201
    },
    {
      "epoch": 2.7135593220338983,
      "grad_norm": 1.9019501209259033,
      "learning_rate": 4.7740112994350285e-06,
      "loss": 1.959,
      "step": 3202
    },
    {
      "epoch": 2.714406779661017,
      "grad_norm": 1.727725863456726,
      "learning_rate": 4.759887005649718e-06,
      "loss": 2.4346,
      "step": 3203
    },
    {
      "epoch": 2.7152542372881356,
      "grad_norm": 1.9527908563613892,
      "learning_rate": 4.745762711864407e-06,
      "loss": 1.798,
      "step": 3204
    },
    {
      "epoch": 2.7161016949152543,
      "grad_norm": 1.6068388223648071,
      "learning_rate": 4.731638418079097e-06,
      "loss": 2.1341,
      "step": 3205
    },
    {
      "epoch": 2.716949152542373,
      "grad_norm": 1.9126149415969849,
      "learning_rate": 4.717514124293786e-06,
      "loss": 1.899,
      "step": 3206
    },
    {
      "epoch": 2.7177966101694917,
      "grad_norm": 1.710034966468811,
      "learning_rate": 4.703389830508475e-06,
      "loss": 2.0316,
      "step": 3207
    },
    {
      "epoch": 2.7186440677966104,
      "grad_norm": 1.5138301849365234,
      "learning_rate": 4.689265536723164e-06,
      "loss": 2.3638,
      "step": 3208
    },
    {
      "epoch": 2.719491525423729,
      "grad_norm": 2.125113010406494,
      "learning_rate": 4.675141242937853e-06,
      "loss": 1.8224,
      "step": 3209
    },
    {
      "epoch": 2.7203389830508473,
      "grad_norm": 1.3093822002410889,
      "learning_rate": 4.6610169491525425e-06,
      "loss": 2.4209,
      "step": 3210
    },
    {
      "epoch": 2.721186440677966,
      "grad_norm": 1.7390575408935547,
      "learning_rate": 4.646892655367232e-06,
      "loss": 2.1579,
      "step": 3211
    },
    {
      "epoch": 2.7220338983050847,
      "grad_norm": 1.7444908618927002,
      "learning_rate": 4.632768361581921e-06,
      "loss": 2.22,
      "step": 3212
    },
    {
      "epoch": 2.7228813559322034,
      "grad_norm": 1.6553432941436768,
      "learning_rate": 4.618644067796611e-06,
      "loss": 2.0794,
      "step": 3213
    },
    {
      "epoch": 2.723728813559322,
      "grad_norm": 1.4758368730545044,
      "learning_rate": 4.6045197740113e-06,
      "loss": 2.2501,
      "step": 3214
    },
    {
      "epoch": 2.7245762711864407,
      "grad_norm": 1.5900572538375854,
      "learning_rate": 4.590395480225989e-06,
      "loss": 2.1464,
      "step": 3215
    },
    {
      "epoch": 2.7254237288135594,
      "grad_norm": 2.284975290298462,
      "learning_rate": 4.576271186440678e-06,
      "loss": 1.6978,
      "step": 3216
    },
    {
      "epoch": 2.726271186440678,
      "grad_norm": 1.8749831914901733,
      "learning_rate": 4.562146892655367e-06,
      "loss": 2.1577,
      "step": 3217
    },
    {
      "epoch": 2.7271186440677964,
      "grad_norm": 1.7108556032180786,
      "learning_rate": 4.5480225988700565e-06,
      "loss": 2.1418,
      "step": 3218
    },
    {
      "epoch": 2.727966101694915,
      "grad_norm": 1.5814942121505737,
      "learning_rate": 4.533898305084746e-06,
      "loss": 2.2746,
      "step": 3219
    },
    {
      "epoch": 2.7288135593220337,
      "grad_norm": 1.5617300271987915,
      "learning_rate": 4.519774011299436e-06,
      "loss": 2.4608,
      "step": 3220
    },
    {
      "epoch": 2.7296610169491524,
      "grad_norm": 1.6508846282958984,
      "learning_rate": 4.505649717514125e-06,
      "loss": 2.21,
      "step": 3221
    },
    {
      "epoch": 2.730508474576271,
      "grad_norm": 1.6110411882400513,
      "learning_rate": 4.491525423728814e-06,
      "loss": 2.3215,
      "step": 3222
    },
    {
      "epoch": 2.73135593220339,
      "grad_norm": 1.7519724369049072,
      "learning_rate": 4.477401129943503e-06,
      "loss": 2.2066,
      "step": 3223
    },
    {
      "epoch": 2.7322033898305085,
      "grad_norm": 1.8455437421798706,
      "learning_rate": 4.463276836158192e-06,
      "loss": 2.2371,
      "step": 3224
    },
    {
      "epoch": 2.733050847457627,
      "grad_norm": 2.0526866912841797,
      "learning_rate": 4.449152542372881e-06,
      "loss": 1.9398,
      "step": 3225
    },
    {
      "epoch": 2.733898305084746,
      "grad_norm": 1.8807004690170288,
      "learning_rate": 4.4350282485875705e-06,
      "loss": 2.1996,
      "step": 3226
    },
    {
      "epoch": 2.7347457627118645,
      "grad_norm": 1.7172704935073853,
      "learning_rate": 4.42090395480226e-06,
      "loss": 1.9544,
      "step": 3227
    },
    {
      "epoch": 2.7355932203389832,
      "grad_norm": 2.149353265762329,
      "learning_rate": 4.40677966101695e-06,
      "loss": 1.5812,
      "step": 3228
    },
    {
      "epoch": 2.736440677966102,
      "grad_norm": 1.7039453983306885,
      "learning_rate": 4.392655367231639e-06,
      "loss": 2.0823,
      "step": 3229
    },
    {
      "epoch": 2.7372881355932206,
      "grad_norm": 1.71449875831604,
      "learning_rate": 4.378531073446328e-06,
      "loss": 2.23,
      "step": 3230
    },
    {
      "epoch": 2.738135593220339,
      "grad_norm": 1.9371012449264526,
      "learning_rate": 4.364406779661017e-06,
      "loss": 1.8363,
      "step": 3231
    },
    {
      "epoch": 2.7389830508474575,
      "grad_norm": 1.851549506187439,
      "learning_rate": 4.350282485875706e-06,
      "loss": 2.0805,
      "step": 3232
    },
    {
      "epoch": 2.739830508474576,
      "grad_norm": 1.7192928791046143,
      "learning_rate": 4.336158192090395e-06,
      "loss": 2.1875,
      "step": 3233
    },
    {
      "epoch": 2.740677966101695,
      "grad_norm": 1.9296584129333496,
      "learning_rate": 4.3220338983050846e-06,
      "loss": 1.938,
      "step": 3234
    },
    {
      "epoch": 2.7415254237288136,
      "grad_norm": 1.6388558149337769,
      "learning_rate": 4.307909604519774e-06,
      "loss": 2.335,
      "step": 3235
    },
    {
      "epoch": 2.7423728813559323,
      "grad_norm": 1.5845305919647217,
      "learning_rate": 4.293785310734464e-06,
      "loss": 2.19,
      "step": 3236
    },
    {
      "epoch": 2.743220338983051,
      "grad_norm": 2.0693283081054688,
      "learning_rate": 4.279661016949153e-06,
      "loss": 1.8567,
      "step": 3237
    },
    {
      "epoch": 2.7440677966101696,
      "grad_norm": 1.8741554021835327,
      "learning_rate": 4.265536723163842e-06,
      "loss": 2.0643,
      "step": 3238
    },
    {
      "epoch": 2.744915254237288,
      "grad_norm": 1.7262877225875854,
      "learning_rate": 4.251412429378531e-06,
      "loss": 2.2332,
      "step": 3239
    },
    {
      "epoch": 2.7457627118644066,
      "grad_norm": 1.424884557723999,
      "learning_rate": 4.23728813559322e-06,
      "loss": 2.2643,
      "step": 3240
    },
    {
      "epoch": 2.7466101694915253,
      "grad_norm": 1.7748403549194336,
      "learning_rate": 4.2231638418079094e-06,
      "loss": 1.8507,
      "step": 3241
    },
    {
      "epoch": 2.747457627118644,
      "grad_norm": 2.2172834873199463,
      "learning_rate": 4.209039548022599e-06,
      "loss": 1.6501,
      "step": 3242
    },
    {
      "epoch": 2.7483050847457626,
      "grad_norm": 1.8971115350723267,
      "learning_rate": 4.194915254237288e-06,
      "loss": 1.7921,
      "step": 3243
    },
    {
      "epoch": 2.7491525423728813,
      "grad_norm": 2.0673277378082275,
      "learning_rate": 4.180790960451978e-06,
      "loss": 1.891,
      "step": 3244
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.614029049873352,
      "learning_rate": 4.166666666666667e-06,
      "loss": 2.3923,
      "step": 3245
    },
    {
      "epoch": 2.7508474576271187,
      "grad_norm": 1.8733458518981934,
      "learning_rate": 4.152542372881356e-06,
      "loss": 1.9396,
      "step": 3246
    },
    {
      "epoch": 2.7516949152542374,
      "grad_norm": 1.8691768646240234,
      "learning_rate": 4.138418079096045e-06,
      "loss": 1.9589,
      "step": 3247
    },
    {
      "epoch": 2.752542372881356,
      "grad_norm": 1.7177436351776123,
      "learning_rate": 4.124293785310734e-06,
      "loss": 2.2102,
      "step": 3248
    },
    {
      "epoch": 2.7533898305084747,
      "grad_norm": 1.5605498552322388,
      "learning_rate": 4.1101694915254235e-06,
      "loss": 2.2768,
      "step": 3249
    },
    {
      "epoch": 2.7542372881355934,
      "grad_norm": 1.9510782957077026,
      "learning_rate": 4.096045197740113e-06,
      "loss": 2.0669,
      "step": 3250
    },
    {
      "epoch": 2.755084745762712,
      "grad_norm": 1.723276138305664,
      "learning_rate": 4.081920903954802e-06,
      "loss": 1.8774,
      "step": 3251
    },
    {
      "epoch": 2.7559322033898304,
      "grad_norm": 1.8737914562225342,
      "learning_rate": 4.067796610169492e-06,
      "loss": 1.9715,
      "step": 3252
    },
    {
      "epoch": 2.756779661016949,
      "grad_norm": 1.9561488628387451,
      "learning_rate": 4.053672316384181e-06,
      "loss": 2.0171,
      "step": 3253
    },
    {
      "epoch": 2.7576271186440677,
      "grad_norm": 1.9660571813583374,
      "learning_rate": 4.03954802259887e-06,
      "loss": 2.0788,
      "step": 3254
    },
    {
      "epoch": 2.7584745762711864,
      "grad_norm": 1.5801583528518677,
      "learning_rate": 4.025423728813559e-06,
      "loss": 2.2551,
      "step": 3255
    },
    {
      "epoch": 2.759322033898305,
      "grad_norm": 1.958560585975647,
      "learning_rate": 4.011299435028248e-06,
      "loss": 2.0017,
      "step": 3256
    },
    {
      "epoch": 2.760169491525424,
      "grad_norm": 2.0370123386383057,
      "learning_rate": 3.9971751412429375e-06,
      "loss": 1.9183,
      "step": 3257
    },
    {
      "epoch": 2.7610169491525425,
      "grad_norm": 1.924447774887085,
      "learning_rate": 3.983050847457627e-06,
      "loss": 1.9518,
      "step": 3258
    },
    {
      "epoch": 2.761864406779661,
      "grad_norm": 1.909886121749878,
      "learning_rate": 3.968926553672317e-06,
      "loss": 1.9164,
      "step": 3259
    },
    {
      "epoch": 2.7627118644067794,
      "grad_norm": 1.5260083675384521,
      "learning_rate": 3.954802259887006e-06,
      "loss": 2.2727,
      "step": 3260
    },
    {
      "epoch": 2.763559322033898,
      "grad_norm": 1.8427681922912598,
      "learning_rate": 3.940677966101696e-06,
      "loss": 1.8079,
      "step": 3261
    },
    {
      "epoch": 2.7644067796610168,
      "grad_norm": 1.7439347505569458,
      "learning_rate": 3.926553672316385e-06,
      "loss": 2.2202,
      "step": 3262
    },
    {
      "epoch": 2.7652542372881355,
      "grad_norm": 1.720460295677185,
      "learning_rate": 3.912429378531074e-06,
      "loss": 2.0204,
      "step": 3263
    },
    {
      "epoch": 2.766101694915254,
      "grad_norm": 1.6500192880630493,
      "learning_rate": 3.898305084745763e-06,
      "loss": 2.5506,
      "step": 3264
    },
    {
      "epoch": 2.766949152542373,
      "grad_norm": 1.5738345384597778,
      "learning_rate": 3.884180790960452e-06,
      "loss": 2.217,
      "step": 3265
    },
    {
      "epoch": 2.7677966101694915,
      "grad_norm": 1.842882752418518,
      "learning_rate": 3.8700564971751415e-06,
      "loss": 2.0424,
      "step": 3266
    },
    {
      "epoch": 2.76864406779661,
      "grad_norm": 1.4322773218154907,
      "learning_rate": 3.855932203389831e-06,
      "loss": 2.4257,
      "step": 3267
    },
    {
      "epoch": 2.769491525423729,
      "grad_norm": 1.6227883100509644,
      "learning_rate": 3.841807909604521e-06,
      "loss": 2.3145,
      "step": 3268
    },
    {
      "epoch": 2.7703389830508476,
      "grad_norm": 1.8622524738311768,
      "learning_rate": 3.82768361581921e-06,
      "loss": 2.1128,
      "step": 3269
    },
    {
      "epoch": 2.7711864406779663,
      "grad_norm": 1.8271149396896362,
      "learning_rate": 3.813559322033899e-06,
      "loss": 2.1978,
      "step": 3270
    },
    {
      "epoch": 2.772033898305085,
      "grad_norm": 1.6806551218032837,
      "learning_rate": 3.799435028248588e-06,
      "loss": 2.25,
      "step": 3271
    },
    {
      "epoch": 2.7728813559322036,
      "grad_norm": 2.082566976547241,
      "learning_rate": 3.7853107344632772e-06,
      "loss": 1.7762,
      "step": 3272
    },
    {
      "epoch": 2.773728813559322,
      "grad_norm": 2.108517646789551,
      "learning_rate": 3.7711864406779664e-06,
      "loss": 1.7451,
      "step": 3273
    },
    {
      "epoch": 2.7745762711864406,
      "grad_norm": 1.8154159784317017,
      "learning_rate": 3.757062146892656e-06,
      "loss": 2.2876,
      "step": 3274
    },
    {
      "epoch": 2.7754237288135593,
      "grad_norm": 1.8693562746047974,
      "learning_rate": 3.742937853107345e-06,
      "loss": 1.9695,
      "step": 3275
    },
    {
      "epoch": 2.776271186440678,
      "grad_norm": 1.78440523147583,
      "learning_rate": 3.7288135593220342e-06,
      "loss": 2.0366,
      "step": 3276
    },
    {
      "epoch": 2.7771186440677966,
      "grad_norm": 1.734572410583496,
      "learning_rate": 3.7146892655367234e-06,
      "loss": 2.0626,
      "step": 3277
    },
    {
      "epoch": 2.7779661016949153,
      "grad_norm": 2.122291326522827,
      "learning_rate": 3.700564971751413e-06,
      "loss": 1.8098,
      "step": 3278
    },
    {
      "epoch": 2.778813559322034,
      "grad_norm": 1.1030203104019165,
      "learning_rate": 3.686440677966102e-06,
      "loss": 2.5436,
      "step": 3279
    },
    {
      "epoch": 2.7796610169491527,
      "grad_norm": 1.6981253623962402,
      "learning_rate": 3.6723163841807913e-06,
      "loss": 2.0083,
      "step": 3280
    },
    {
      "epoch": 2.780508474576271,
      "grad_norm": 1.5894949436187744,
      "learning_rate": 3.6581920903954804e-06,
      "loss": 2.238,
      "step": 3281
    },
    {
      "epoch": 2.7813559322033896,
      "grad_norm": 1.549978256225586,
      "learning_rate": 3.64406779661017e-06,
      "loss": 2.2281,
      "step": 3282
    },
    {
      "epoch": 2.7822033898305083,
      "grad_norm": 1.5143518447875977,
      "learning_rate": 3.629943502824859e-06,
      "loss": 2.3051,
      "step": 3283
    },
    {
      "epoch": 2.783050847457627,
      "grad_norm": 1.7828015089035034,
      "learning_rate": 3.6158192090395483e-06,
      "loss": 2.1792,
      "step": 3284
    },
    {
      "epoch": 2.7838983050847457,
      "grad_norm": 1.6845375299453735,
      "learning_rate": 3.6016949152542374e-06,
      "loss": 2.1513,
      "step": 3285
    },
    {
      "epoch": 2.7847457627118644,
      "grad_norm": 1.6669672727584839,
      "learning_rate": 3.587570621468927e-06,
      "loss": 2.2889,
      "step": 3286
    },
    {
      "epoch": 2.785593220338983,
      "grad_norm": 1.5005353689193726,
      "learning_rate": 3.573446327683616e-06,
      "loss": 2.3482,
      "step": 3287
    },
    {
      "epoch": 2.7864406779661017,
      "grad_norm": 1.8809428215026855,
      "learning_rate": 3.5593220338983053e-06,
      "loss": 1.9747,
      "step": 3288
    },
    {
      "epoch": 2.7872881355932204,
      "grad_norm": 1.5482172966003418,
      "learning_rate": 3.5451977401129944e-06,
      "loss": 2.5255,
      "step": 3289
    },
    {
      "epoch": 2.788135593220339,
      "grad_norm": 1.8034367561340332,
      "learning_rate": 3.531073446327684e-06,
      "loss": 2.1528,
      "step": 3290
    },
    {
      "epoch": 2.788983050847458,
      "grad_norm": 1.7874549627304077,
      "learning_rate": 3.516949152542373e-06,
      "loss": 2.2353,
      "step": 3291
    },
    {
      "epoch": 2.7898305084745765,
      "grad_norm": 1.96170175075531,
      "learning_rate": 3.5028248587570623e-06,
      "loss": 1.8544,
      "step": 3292
    },
    {
      "epoch": 2.790677966101695,
      "grad_norm": 1.7292556762695312,
      "learning_rate": 3.4887005649717514e-06,
      "loss": 2.2137,
      "step": 3293
    },
    {
      "epoch": 2.7915254237288134,
      "grad_norm": 1.6710186004638672,
      "learning_rate": 3.474576271186441e-06,
      "loss": 2.413,
      "step": 3294
    },
    {
      "epoch": 2.792372881355932,
      "grad_norm": 1.7541210651397705,
      "learning_rate": 3.46045197740113e-06,
      "loss": 2.1582,
      "step": 3295
    },
    {
      "epoch": 2.7932203389830508,
      "grad_norm": 1.61122727394104,
      "learning_rate": 3.4463276836158193e-06,
      "loss": 2.1661,
      "step": 3296
    },
    {
      "epoch": 2.7940677966101695,
      "grad_norm": 1.8008424043655396,
      "learning_rate": 3.4322033898305084e-06,
      "loss": 2.1737,
      "step": 3297
    },
    {
      "epoch": 2.794915254237288,
      "grad_norm": 1.6333184242248535,
      "learning_rate": 3.418079096045198e-06,
      "loss": 1.9677,
      "step": 3298
    },
    {
      "epoch": 2.795762711864407,
      "grad_norm": 1.569275975227356,
      "learning_rate": 3.403954802259887e-06,
      "loss": 2.3139,
      "step": 3299
    },
    {
      "epoch": 2.7966101694915255,
      "grad_norm": 1.9034026861190796,
      "learning_rate": 3.3898305084745763e-06,
      "loss": 1.8646,
      "step": 3300
    },
    {
      "epoch": 2.797457627118644,
      "grad_norm": 1.9896222352981567,
      "learning_rate": 3.375706214689266e-06,
      "loss": 1.6079,
      "step": 3301
    },
    {
      "epoch": 2.7983050847457624,
      "grad_norm": 1.4076114892959595,
      "learning_rate": 3.361581920903955e-06,
      "loss": 2.3119,
      "step": 3302
    },
    {
      "epoch": 2.799152542372881,
      "grad_norm": 1.6879991292953491,
      "learning_rate": 3.347457627118644e-06,
      "loss": 2.2824,
      "step": 3303
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.23570716381073,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 2.5017,
      "step": 3304
    },
    {
      "epoch": 2.8008474576271185,
      "grad_norm": 1.734196662902832,
      "learning_rate": 3.319209039548023e-06,
      "loss": 2.0933,
      "step": 3305
    },
    {
      "epoch": 2.801694915254237,
      "grad_norm": 1.900134563446045,
      "learning_rate": 3.305084745762712e-06,
      "loss": 2.155,
      "step": 3306
    },
    {
      "epoch": 2.802542372881356,
      "grad_norm": 2.1223700046539307,
      "learning_rate": 3.290960451977401e-06,
      "loss": 1.8647,
      "step": 3307
    },
    {
      "epoch": 2.8033898305084746,
      "grad_norm": 1.3897490501403809,
      "learning_rate": 3.2768361581920903e-06,
      "loss": 2.425,
      "step": 3308
    },
    {
      "epoch": 2.8042372881355933,
      "grad_norm": 1.7049990892410278,
      "learning_rate": 3.26271186440678e-06,
      "loss": 1.8893,
      "step": 3309
    },
    {
      "epoch": 2.805084745762712,
      "grad_norm": 1.7432835102081299,
      "learning_rate": 3.248587570621469e-06,
      "loss": 2.0095,
      "step": 3310
    },
    {
      "epoch": 2.8059322033898306,
      "grad_norm": 1.9104880094528198,
      "learning_rate": 3.234463276836158e-06,
      "loss": 1.9359,
      "step": 3311
    },
    {
      "epoch": 2.8067796610169493,
      "grad_norm": 1.3656202554702759,
      "learning_rate": 3.2203389830508473e-06,
      "loss": 2.3345,
      "step": 3312
    },
    {
      "epoch": 2.807627118644068,
      "grad_norm": 1.4636744260787964,
      "learning_rate": 3.206214689265537e-06,
      "loss": 2.2708,
      "step": 3313
    },
    {
      "epoch": 2.8084745762711867,
      "grad_norm": 1.8310048580169678,
      "learning_rate": 3.192090395480226e-06,
      "loss": 2.1519,
      "step": 3314
    },
    {
      "epoch": 2.809322033898305,
      "grad_norm": 1.9034868478775024,
      "learning_rate": 3.1779661016949152e-06,
      "loss": 2.0292,
      "step": 3315
    },
    {
      "epoch": 2.8101694915254236,
      "grad_norm": 1.5066553354263306,
      "learning_rate": 3.1638418079096044e-06,
      "loss": 2.4282,
      "step": 3316
    },
    {
      "epoch": 2.8110169491525423,
      "grad_norm": 1.8293631076812744,
      "learning_rate": 3.149717514124294e-06,
      "loss": 1.9231,
      "step": 3317
    },
    {
      "epoch": 2.811864406779661,
      "grad_norm": 1.886911153793335,
      "learning_rate": 3.135593220338983e-06,
      "loss": 2.0372,
      "step": 3318
    },
    {
      "epoch": 2.8127118644067797,
      "grad_norm": 1.9897105693817139,
      "learning_rate": 3.1214689265536726e-06,
      "loss": 2.034,
      "step": 3319
    },
    {
      "epoch": 2.8135593220338984,
      "grad_norm": 1.4622457027435303,
      "learning_rate": 3.107344632768362e-06,
      "loss": 2.3314,
      "step": 3320
    },
    {
      "epoch": 2.814406779661017,
      "grad_norm": 1.5068597793579102,
      "learning_rate": 3.093220338983051e-06,
      "loss": 2.3699,
      "step": 3321
    },
    {
      "epoch": 2.8152542372881357,
      "grad_norm": 1.9084815979003906,
      "learning_rate": 3.0790960451977405e-06,
      "loss": 1.9478,
      "step": 3322
    },
    {
      "epoch": 2.816101694915254,
      "grad_norm": 1.5619968175888062,
      "learning_rate": 3.0649717514124297e-06,
      "loss": 2.1414,
      "step": 3323
    },
    {
      "epoch": 2.8169491525423727,
      "grad_norm": 1.6689423322677612,
      "learning_rate": 3.050847457627119e-06,
      "loss": 2.2752,
      "step": 3324
    },
    {
      "epoch": 2.8177966101694913,
      "grad_norm": 1.325331449508667,
      "learning_rate": 3.0367231638418084e-06,
      "loss": 2.4943,
      "step": 3325
    },
    {
      "epoch": 2.81864406779661,
      "grad_norm": 1.5005288124084473,
      "learning_rate": 3.0225988700564975e-06,
      "loss": 2.3628,
      "step": 3326
    },
    {
      "epoch": 2.8194915254237287,
      "grad_norm": 1.8637874126434326,
      "learning_rate": 3.0084745762711867e-06,
      "loss": 2.0922,
      "step": 3327
    },
    {
      "epoch": 2.8203389830508474,
      "grad_norm": 2.4057538509368896,
      "learning_rate": 2.994350282485876e-06,
      "loss": 1.6564,
      "step": 3328
    },
    {
      "epoch": 2.821186440677966,
      "grad_norm": 2.0353689193725586,
      "learning_rate": 2.9802259887005654e-06,
      "loss": 1.9043,
      "step": 3329
    },
    {
      "epoch": 2.8220338983050848,
      "grad_norm": 1.7443592548370361,
      "learning_rate": 2.9661016949152545e-06,
      "loss": 2.1515,
      "step": 3330
    },
    {
      "epoch": 2.8228813559322035,
      "grad_norm": 1.5564725399017334,
      "learning_rate": 2.9519774011299437e-06,
      "loss": 2.3661,
      "step": 3331
    },
    {
      "epoch": 2.823728813559322,
      "grad_norm": 1.7410897016525269,
      "learning_rate": 2.937853107344633e-06,
      "loss": 2.1678,
      "step": 3332
    },
    {
      "epoch": 2.824576271186441,
      "grad_norm": 1.715358018875122,
      "learning_rate": 2.9237288135593224e-06,
      "loss": 2.0724,
      "step": 3333
    },
    {
      "epoch": 2.8254237288135595,
      "grad_norm": 2.0631182193756104,
      "learning_rate": 2.9096045197740115e-06,
      "loss": 1.8538,
      "step": 3334
    },
    {
      "epoch": 2.826271186440678,
      "grad_norm": 1.677146077156067,
      "learning_rate": 2.8954802259887007e-06,
      "loss": 2.1019,
      "step": 3335
    },
    {
      "epoch": 2.8271186440677964,
      "grad_norm": 2.1181676387786865,
      "learning_rate": 2.88135593220339e-06,
      "loss": 2.0057,
      "step": 3336
    },
    {
      "epoch": 2.827966101694915,
      "grad_norm": 1.9416948556900024,
      "learning_rate": 2.8672316384180794e-06,
      "loss": 1.775,
      "step": 3337
    },
    {
      "epoch": 2.828813559322034,
      "grad_norm": 1.7427901029586792,
      "learning_rate": 2.8531073446327686e-06,
      "loss": 2.0498,
      "step": 3338
    },
    {
      "epoch": 2.8296610169491525,
      "grad_norm": 1.6878767013549805,
      "learning_rate": 2.8389830508474577e-06,
      "loss": 2.1399,
      "step": 3339
    },
    {
      "epoch": 2.830508474576271,
      "grad_norm": 1.4075371026992798,
      "learning_rate": 2.824858757062147e-06,
      "loss": 2.2619,
      "step": 3340
    },
    {
      "epoch": 2.83135593220339,
      "grad_norm": 1.383636236190796,
      "learning_rate": 2.8107344632768364e-06,
      "loss": 2.3308,
      "step": 3341
    },
    {
      "epoch": 2.8322033898305086,
      "grad_norm": 1.1737382411956787,
      "learning_rate": 2.7966101694915256e-06,
      "loss": 2.4857,
      "step": 3342
    },
    {
      "epoch": 2.8330508474576273,
      "grad_norm": 1.6975308656692505,
      "learning_rate": 2.7824858757062147e-06,
      "loss": 2.1746,
      "step": 3343
    },
    {
      "epoch": 2.8338983050847455,
      "grad_norm": 1.8578035831451416,
      "learning_rate": 2.768361581920904e-06,
      "loss": 2.061,
      "step": 3344
    },
    {
      "epoch": 2.834745762711864,
      "grad_norm": 1.2902839183807373,
      "learning_rate": 2.7542372881355934e-06,
      "loss": 2.4812,
      "step": 3345
    },
    {
      "epoch": 2.835593220338983,
      "grad_norm": 1.8486239910125732,
      "learning_rate": 2.7401129943502826e-06,
      "loss": 2.0483,
      "step": 3346
    },
    {
      "epoch": 2.8364406779661016,
      "grad_norm": 1.5760711431503296,
      "learning_rate": 2.7259887005649717e-06,
      "loss": 2.201,
      "step": 3347
    },
    {
      "epoch": 2.8372881355932202,
      "grad_norm": 1.4381455183029175,
      "learning_rate": 2.711864406779661e-06,
      "loss": 2.2938,
      "step": 3348
    },
    {
      "epoch": 2.838135593220339,
      "grad_norm": 2.1512436866760254,
      "learning_rate": 2.6977401129943505e-06,
      "loss": 1.9279,
      "step": 3349
    },
    {
      "epoch": 2.8389830508474576,
      "grad_norm": 1.9421781301498413,
      "learning_rate": 2.6836158192090396e-06,
      "loss": 2.1418,
      "step": 3350
    },
    {
      "epoch": 2.8398305084745763,
      "grad_norm": 1.4164178371429443,
      "learning_rate": 2.6694915254237287e-06,
      "loss": 2.3653,
      "step": 3351
    },
    {
      "epoch": 2.840677966101695,
      "grad_norm": 1.6631990671157837,
      "learning_rate": 2.655367231638418e-06,
      "loss": 2.1167,
      "step": 3352
    },
    {
      "epoch": 2.8415254237288137,
      "grad_norm": 1.8184740543365479,
      "learning_rate": 2.6412429378531075e-06,
      "loss": 2.0879,
      "step": 3353
    },
    {
      "epoch": 2.8423728813559324,
      "grad_norm": 2.1014161109924316,
      "learning_rate": 2.6271186440677966e-06,
      "loss": 1.8246,
      "step": 3354
    },
    {
      "epoch": 2.843220338983051,
      "grad_norm": 1.7352596521377563,
      "learning_rate": 2.6129943502824858e-06,
      "loss": 2.1501,
      "step": 3355
    },
    {
      "epoch": 2.8440677966101697,
      "grad_norm": 1.4966508150100708,
      "learning_rate": 2.5988700564971753e-06,
      "loss": 2.3759,
      "step": 3356
    },
    {
      "epoch": 2.844915254237288,
      "grad_norm": 1.745072364807129,
      "learning_rate": 2.5847457627118645e-06,
      "loss": 2.1665,
      "step": 3357
    },
    {
      "epoch": 2.8457627118644067,
      "grad_norm": 1.8252897262573242,
      "learning_rate": 2.5706214689265536e-06,
      "loss": 2.1921,
      "step": 3358
    },
    {
      "epoch": 2.8466101694915253,
      "grad_norm": 1.5177760124206543,
      "learning_rate": 2.556497175141243e-06,
      "loss": 1.9854,
      "step": 3359
    },
    {
      "epoch": 2.847457627118644,
      "grad_norm": 1.8378403186798096,
      "learning_rate": 2.5423728813559323e-06,
      "loss": 2.0397,
      "step": 3360
    },
    {
      "epoch": 2.8483050847457627,
      "grad_norm": 1.78495192527771,
      "learning_rate": 2.528248587570622e-06,
      "loss": 2.1185,
      "step": 3361
    },
    {
      "epoch": 2.8491525423728814,
      "grad_norm": 1.9980145692825317,
      "learning_rate": 2.514124293785311e-06,
      "loss": 1.9541,
      "step": 3362
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.014122486114502,
      "learning_rate": 2.5e-06,
      "loss": 1.9235,
      "step": 3363
    },
    {
      "epoch": 2.8508474576271188,
      "grad_norm": 2.524773597717285,
      "learning_rate": 2.4858757062146894e-06,
      "loss": 1.8301,
      "step": 3364
    },
    {
      "epoch": 2.851694915254237,
      "grad_norm": 2.225480556488037,
      "learning_rate": 2.471751412429379e-06,
      "loss": 1.8999,
      "step": 3365
    },
    {
      "epoch": 2.8525423728813557,
      "grad_norm": 1.6353254318237305,
      "learning_rate": 2.457627118644068e-06,
      "loss": 2.1385,
      "step": 3366
    },
    {
      "epoch": 2.8533898305084744,
      "grad_norm": 1.9169930219650269,
      "learning_rate": 2.4435028248587572e-06,
      "loss": 1.8262,
      "step": 3367
    },
    {
      "epoch": 2.854237288135593,
      "grad_norm": 1.9980542659759521,
      "learning_rate": 2.4293785310734464e-06,
      "loss": 2.0419,
      "step": 3368
    },
    {
      "epoch": 2.8550847457627118,
      "grad_norm": 2.3818628787994385,
      "learning_rate": 2.415254237288136e-06,
      "loss": 1.9207,
      "step": 3369
    },
    {
      "epoch": 2.8559322033898304,
      "grad_norm": 1.5910136699676514,
      "learning_rate": 2.401129943502825e-06,
      "loss": 2.1945,
      "step": 3370
    },
    {
      "epoch": 2.856779661016949,
      "grad_norm": 1.8092448711395264,
      "learning_rate": 2.3870056497175142e-06,
      "loss": 1.9613,
      "step": 3371
    },
    {
      "epoch": 2.857627118644068,
      "grad_norm": 1.812135100364685,
      "learning_rate": 2.3728813559322034e-06,
      "loss": 1.8699,
      "step": 3372
    },
    {
      "epoch": 2.8584745762711865,
      "grad_norm": 1.7698718309402466,
      "learning_rate": 2.358757062146893e-06,
      "loss": 1.976,
      "step": 3373
    },
    {
      "epoch": 2.859322033898305,
      "grad_norm": 1.7062208652496338,
      "learning_rate": 2.344632768361582e-06,
      "loss": 2.0149,
      "step": 3374
    },
    {
      "epoch": 2.860169491525424,
      "grad_norm": 1.8272556066513062,
      "learning_rate": 2.3305084745762712e-06,
      "loss": 2.0581,
      "step": 3375
    },
    {
      "epoch": 2.8610169491525426,
      "grad_norm": 1.8334457874298096,
      "learning_rate": 2.3163841807909604e-06,
      "loss": 2.0563,
      "step": 3376
    },
    {
      "epoch": 2.8618644067796613,
      "grad_norm": 2.0398409366607666,
      "learning_rate": 2.30225988700565e-06,
      "loss": 1.87,
      "step": 3377
    },
    {
      "epoch": 2.86271186440678,
      "grad_norm": 1.729124665260315,
      "learning_rate": 2.288135593220339e-06,
      "loss": 2.0625,
      "step": 3378
    },
    {
      "epoch": 2.863559322033898,
      "grad_norm": 1.687659502029419,
      "learning_rate": 2.2740112994350283e-06,
      "loss": 2.1105,
      "step": 3379
    },
    {
      "epoch": 2.864406779661017,
      "grad_norm": 1.8048326969146729,
      "learning_rate": 2.259887005649718e-06,
      "loss": 2.1501,
      "step": 3380
    },
    {
      "epoch": 2.8652542372881356,
      "grad_norm": 1.6243799924850464,
      "learning_rate": 2.245762711864407e-06,
      "loss": 2.2105,
      "step": 3381
    },
    {
      "epoch": 2.8661016949152542,
      "grad_norm": 1.5646708011627197,
      "learning_rate": 2.231638418079096e-06,
      "loss": 2.1571,
      "step": 3382
    },
    {
      "epoch": 2.866949152542373,
      "grad_norm": 1.6267356872558594,
      "learning_rate": 2.2175141242937853e-06,
      "loss": 2.2915,
      "step": 3383
    },
    {
      "epoch": 2.8677966101694916,
      "grad_norm": 1.7456969022750854,
      "learning_rate": 2.203389830508475e-06,
      "loss": 1.9753,
      "step": 3384
    },
    {
      "epoch": 2.8686440677966103,
      "grad_norm": 1.92332923412323,
      "learning_rate": 2.189265536723164e-06,
      "loss": 1.8815,
      "step": 3385
    },
    {
      "epoch": 2.8694915254237285,
      "grad_norm": 1.9790289402008057,
      "learning_rate": 2.175141242937853e-06,
      "loss": 2.0115,
      "step": 3386
    },
    {
      "epoch": 2.8703389830508472,
      "grad_norm": 2.064730405807495,
      "learning_rate": 2.1610169491525423e-06,
      "loss": 1.8712,
      "step": 3387
    },
    {
      "epoch": 2.871186440677966,
      "grad_norm": 2.201131820678711,
      "learning_rate": 2.146892655367232e-06,
      "loss": 1.639,
      "step": 3388
    },
    {
      "epoch": 2.8720338983050846,
      "grad_norm": 1.7507131099700928,
      "learning_rate": 2.132768361581921e-06,
      "loss": 2.2192,
      "step": 3389
    },
    {
      "epoch": 2.8728813559322033,
      "grad_norm": 1.9839781522750854,
      "learning_rate": 2.11864406779661e-06,
      "loss": 1.8978,
      "step": 3390
    },
    {
      "epoch": 2.873728813559322,
      "grad_norm": 2.417419195175171,
      "learning_rate": 2.1045197740112993e-06,
      "loss": 1.7828,
      "step": 3391
    },
    {
      "epoch": 2.8745762711864407,
      "grad_norm": 1.8423184156417847,
      "learning_rate": 2.090395480225989e-06,
      "loss": 2.0322,
      "step": 3392
    },
    {
      "epoch": 2.8754237288135593,
      "grad_norm": 1.6614102125167847,
      "learning_rate": 2.076271186440678e-06,
      "loss": 2.2679,
      "step": 3393
    },
    {
      "epoch": 2.876271186440678,
      "grad_norm": 1.0237455368041992,
      "learning_rate": 2.062146892655367e-06,
      "loss": 2.5817,
      "step": 3394
    },
    {
      "epoch": 2.8771186440677967,
      "grad_norm": 1.5811541080474854,
      "learning_rate": 2.0480225988700563e-06,
      "loss": 2.3303,
      "step": 3395
    },
    {
      "epoch": 2.8779661016949154,
      "grad_norm": 1.284185528755188,
      "learning_rate": 2.033898305084746e-06,
      "loss": 2.3703,
      "step": 3396
    },
    {
      "epoch": 2.878813559322034,
      "grad_norm": 1.6183527708053589,
      "learning_rate": 2.019774011299435e-06,
      "loss": 2.1484,
      "step": 3397
    },
    {
      "epoch": 2.8796610169491528,
      "grad_norm": 1.596364140510559,
      "learning_rate": 2.005649717514124e-06,
      "loss": 2.2813,
      "step": 3398
    },
    {
      "epoch": 2.8805084745762715,
      "grad_norm": 2.3689513206481934,
      "learning_rate": 1.9915254237288133e-06,
      "loss": 1.4751,
      "step": 3399
    },
    {
      "epoch": 2.8813559322033897,
      "grad_norm": 2.1542131900787354,
      "learning_rate": 1.977401129943503e-06,
      "loss": 1.779,
      "step": 3400
    },
    {
      "epoch": 2.8822033898305084,
      "grad_norm": 1.5996615886688232,
      "learning_rate": 1.9632768361581925e-06,
      "loss": 2.4124,
      "step": 3401
    },
    {
      "epoch": 2.883050847457627,
      "grad_norm": 1.4909052848815918,
      "learning_rate": 1.9491525423728816e-06,
      "loss": 2.3465,
      "step": 3402
    },
    {
      "epoch": 2.8838983050847458,
      "grad_norm": 1.9433684349060059,
      "learning_rate": 1.9350282485875707e-06,
      "loss": 1.7697,
      "step": 3403
    },
    {
      "epoch": 2.8847457627118644,
      "grad_norm": 1.9747514724731445,
      "learning_rate": 1.9209039548022603e-06,
      "loss": 1.7446,
      "step": 3404
    },
    {
      "epoch": 2.885593220338983,
      "grad_norm": 1.4945168495178223,
      "learning_rate": 1.9067796610169495e-06,
      "loss": 2.222,
      "step": 3405
    },
    {
      "epoch": 2.886440677966102,
      "grad_norm": 2.045989751815796,
      "learning_rate": 1.8926553672316386e-06,
      "loss": 1.8782,
      "step": 3406
    },
    {
      "epoch": 2.88728813559322,
      "grad_norm": 1.7009637355804443,
      "learning_rate": 1.878531073446328e-06,
      "loss": 1.9011,
      "step": 3407
    },
    {
      "epoch": 2.8881355932203387,
      "grad_norm": 2.1265628337860107,
      "learning_rate": 1.8644067796610171e-06,
      "loss": 1.9526,
      "step": 3408
    },
    {
      "epoch": 2.8889830508474574,
      "grad_norm": 1.696357250213623,
      "learning_rate": 1.8502824858757065e-06,
      "loss": 2.2116,
      "step": 3409
    },
    {
      "epoch": 2.889830508474576,
      "grad_norm": 2.1881537437438965,
      "learning_rate": 1.8361581920903956e-06,
      "loss": 1.7084,
      "step": 3410
    },
    {
      "epoch": 2.890677966101695,
      "grad_norm": 1.791032075881958,
      "learning_rate": 1.822033898305085e-06,
      "loss": 2.0799,
      "step": 3411
    },
    {
      "epoch": 2.8915254237288135,
      "grad_norm": 1.15604829788208,
      "learning_rate": 1.8079096045197741e-06,
      "loss": 2.5477,
      "step": 3412
    },
    {
      "epoch": 2.892372881355932,
      "grad_norm": 1.626738429069519,
      "learning_rate": 1.7937853107344635e-06,
      "loss": 2.4368,
      "step": 3413
    },
    {
      "epoch": 2.893220338983051,
      "grad_norm": 1.5530431270599365,
      "learning_rate": 1.7796610169491526e-06,
      "loss": 2.2614,
      "step": 3414
    },
    {
      "epoch": 2.8940677966101696,
      "grad_norm": 1.7293918132781982,
      "learning_rate": 1.765536723163842e-06,
      "loss": 2.1684,
      "step": 3415
    },
    {
      "epoch": 2.8949152542372882,
      "grad_norm": 2.015207052230835,
      "learning_rate": 1.7514124293785311e-06,
      "loss": 1.8092,
      "step": 3416
    },
    {
      "epoch": 2.895762711864407,
      "grad_norm": 1.7043472528457642,
      "learning_rate": 1.7372881355932205e-06,
      "loss": 2.151,
      "step": 3417
    },
    {
      "epoch": 2.8966101694915256,
      "grad_norm": 1.7575970888137817,
      "learning_rate": 1.7231638418079096e-06,
      "loss": 2.0345,
      "step": 3418
    },
    {
      "epoch": 2.8974576271186443,
      "grad_norm": 1.8352203369140625,
      "learning_rate": 1.709039548022599e-06,
      "loss": 2.0827,
      "step": 3419
    },
    {
      "epoch": 2.898305084745763,
      "grad_norm": 2.0024681091308594,
      "learning_rate": 1.6949152542372882e-06,
      "loss": 1.8409,
      "step": 3420
    },
    {
      "epoch": 2.8991525423728812,
      "grad_norm": 2.071882963180542,
      "learning_rate": 1.6807909604519775e-06,
      "loss": 1.6405,
      "step": 3421
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.923279881477356,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.9334,
      "step": 3422
    },
    {
      "epoch": 2.9008474576271186,
      "grad_norm": 1.7485185861587524,
      "learning_rate": 1.652542372881356e-06,
      "loss": 2.0783,
      "step": 3423
    },
    {
      "epoch": 2.9016949152542373,
      "grad_norm": 1.6141704320907593,
      "learning_rate": 1.6384180790960452e-06,
      "loss": 2.199,
      "step": 3424
    },
    {
      "epoch": 2.902542372881356,
      "grad_norm": 1.840831995010376,
      "learning_rate": 1.6242937853107345e-06,
      "loss": 1.988,
      "step": 3425
    },
    {
      "epoch": 2.9033898305084747,
      "grad_norm": 1.810667634010315,
      "learning_rate": 1.6101694915254237e-06,
      "loss": 2.0097,
      "step": 3426
    },
    {
      "epoch": 2.9042372881355933,
      "grad_norm": 1.6117627620697021,
      "learning_rate": 1.596045197740113e-06,
      "loss": 2.1645,
      "step": 3427
    },
    {
      "epoch": 2.905084745762712,
      "grad_norm": 1.802703619003296,
      "learning_rate": 1.5819209039548022e-06,
      "loss": 2.1695,
      "step": 3428
    },
    {
      "epoch": 2.9059322033898303,
      "grad_norm": 1.5631777048110962,
      "learning_rate": 1.5677966101694915e-06,
      "loss": 2.1949,
      "step": 3429
    },
    {
      "epoch": 2.906779661016949,
      "grad_norm": 1.8716254234313965,
      "learning_rate": 1.553672316384181e-06,
      "loss": 1.8599,
      "step": 3430
    },
    {
      "epoch": 2.9076271186440676,
      "grad_norm": 2.2305080890655518,
      "learning_rate": 1.5395480225988703e-06,
      "loss": 1.929,
      "step": 3431
    },
    {
      "epoch": 2.9084745762711863,
      "grad_norm": 1.560698390007019,
      "learning_rate": 1.5254237288135594e-06,
      "loss": 2.2672,
      "step": 3432
    },
    {
      "epoch": 2.909322033898305,
      "grad_norm": 2.0044262409210205,
      "learning_rate": 1.5112994350282488e-06,
      "loss": 1.7978,
      "step": 3433
    },
    {
      "epoch": 2.9101694915254237,
      "grad_norm": 1.956240177154541,
      "learning_rate": 1.497175141242938e-06,
      "loss": 1.8694,
      "step": 3434
    },
    {
      "epoch": 2.9110169491525424,
      "grad_norm": 2.3589026927948,
      "learning_rate": 1.4830508474576273e-06,
      "loss": 1.6085,
      "step": 3435
    },
    {
      "epoch": 2.911864406779661,
      "grad_norm": 1.6329890489578247,
      "learning_rate": 1.4689265536723164e-06,
      "loss": 2.2799,
      "step": 3436
    },
    {
      "epoch": 2.9127118644067798,
      "grad_norm": 1.5386956930160522,
      "learning_rate": 1.4548022598870058e-06,
      "loss": 2.1172,
      "step": 3437
    },
    {
      "epoch": 2.9135593220338984,
      "grad_norm": 1.8822194337844849,
      "learning_rate": 1.440677966101695e-06,
      "loss": 1.9766,
      "step": 3438
    },
    {
      "epoch": 2.914406779661017,
      "grad_norm": 1.6315912008285522,
      "learning_rate": 1.4265536723163843e-06,
      "loss": 2.2342,
      "step": 3439
    },
    {
      "epoch": 2.915254237288136,
      "grad_norm": 1.6176600456237793,
      "learning_rate": 1.4124293785310734e-06,
      "loss": 2.0053,
      "step": 3440
    },
    {
      "epoch": 2.9161016949152545,
      "grad_norm": 1.7778948545455933,
      "learning_rate": 1.3983050847457628e-06,
      "loss": 2.0083,
      "step": 3441
    },
    {
      "epoch": 2.9169491525423727,
      "grad_norm": 2.0902791023254395,
      "learning_rate": 1.384180790960452e-06,
      "loss": 1.7752,
      "step": 3442
    },
    {
      "epoch": 2.9177966101694914,
      "grad_norm": 1.661112666130066,
      "learning_rate": 1.3700564971751413e-06,
      "loss": 2.5118,
      "step": 3443
    },
    {
      "epoch": 2.91864406779661,
      "grad_norm": 1.8177666664123535,
      "learning_rate": 1.3559322033898304e-06,
      "loss": 2.035,
      "step": 3444
    },
    {
      "epoch": 2.919491525423729,
      "grad_norm": 2.120927095413208,
      "learning_rate": 1.3418079096045198e-06,
      "loss": 1.7663,
      "step": 3445
    },
    {
      "epoch": 2.9203389830508475,
      "grad_norm": 1.778826355934143,
      "learning_rate": 1.327683615819209e-06,
      "loss": 1.8262,
      "step": 3446
    },
    {
      "epoch": 2.921186440677966,
      "grad_norm": 1.8968206644058228,
      "learning_rate": 1.3135593220338983e-06,
      "loss": 2.0749,
      "step": 3447
    },
    {
      "epoch": 2.922033898305085,
      "grad_norm": 1.3669257164001465,
      "learning_rate": 1.2994350282485877e-06,
      "loss": 2.4951,
      "step": 3448
    },
    {
      "epoch": 2.9228813559322036,
      "grad_norm": 1.7350246906280518,
      "learning_rate": 1.2853107344632768e-06,
      "loss": 2.0928,
      "step": 3449
    },
    {
      "epoch": 2.923728813559322,
      "grad_norm": 1.2821245193481445,
      "learning_rate": 1.2711864406779662e-06,
      "loss": 2.5033,
      "step": 3450
    },
    {
      "epoch": 2.9245762711864405,
      "grad_norm": 2.0260376930236816,
      "learning_rate": 1.2570621468926555e-06,
      "loss": 1.828,
      "step": 3451
    },
    {
      "epoch": 2.925423728813559,
      "grad_norm": 1.8080556392669678,
      "learning_rate": 1.2429378531073447e-06,
      "loss": 1.8051,
      "step": 3452
    },
    {
      "epoch": 2.926271186440678,
      "grad_norm": 1.7276920080184937,
      "learning_rate": 1.228813559322034e-06,
      "loss": 2.0536,
      "step": 3453
    },
    {
      "epoch": 2.9271186440677965,
      "grad_norm": 2.0028326511383057,
      "learning_rate": 1.2146892655367232e-06,
      "loss": 1.8631,
      "step": 3454
    },
    {
      "epoch": 2.9279661016949152,
      "grad_norm": 2.219172954559326,
      "learning_rate": 1.2005649717514125e-06,
      "loss": 1.676,
      "step": 3455
    },
    {
      "epoch": 2.928813559322034,
      "grad_norm": 1.5696964263916016,
      "learning_rate": 1.1864406779661017e-06,
      "loss": 2.2682,
      "step": 3456
    },
    {
      "epoch": 2.9296610169491526,
      "grad_norm": 1.4314230680465698,
      "learning_rate": 1.172316384180791e-06,
      "loss": 2.2949,
      "step": 3457
    },
    {
      "epoch": 2.9305084745762713,
      "grad_norm": 1.8091028928756714,
      "learning_rate": 1.1581920903954802e-06,
      "loss": 2.2635,
      "step": 3458
    },
    {
      "epoch": 2.93135593220339,
      "grad_norm": 1.8473447561264038,
      "learning_rate": 1.1440677966101696e-06,
      "loss": 2.2436,
      "step": 3459
    },
    {
      "epoch": 2.9322033898305087,
      "grad_norm": 1.2700364589691162,
      "learning_rate": 1.129943502824859e-06,
      "loss": 2.318,
      "step": 3460
    },
    {
      "epoch": 2.9330508474576273,
      "grad_norm": 1.5851829051971436,
      "learning_rate": 1.115819209039548e-06,
      "loss": 2.3973,
      "step": 3461
    },
    {
      "epoch": 2.933898305084746,
      "grad_norm": 1.7963261604309082,
      "learning_rate": 1.1016949152542374e-06,
      "loss": 2.0373,
      "step": 3462
    },
    {
      "epoch": 2.9347457627118643,
      "grad_norm": 1.9901875257492065,
      "learning_rate": 1.0875706214689266e-06,
      "loss": 2.0528,
      "step": 3463
    },
    {
      "epoch": 2.935593220338983,
      "grad_norm": 1.9313286542892456,
      "learning_rate": 1.073446327683616e-06,
      "loss": 1.9245,
      "step": 3464
    },
    {
      "epoch": 2.9364406779661016,
      "grad_norm": 1.8070708513259888,
      "learning_rate": 1.059322033898305e-06,
      "loss": 1.998,
      "step": 3465
    },
    {
      "epoch": 2.9372881355932203,
      "grad_norm": 2.0432677268981934,
      "learning_rate": 1.0451977401129944e-06,
      "loss": 1.8153,
      "step": 3466
    },
    {
      "epoch": 2.938135593220339,
      "grad_norm": 1.7318568229675293,
      "learning_rate": 1.0310734463276836e-06,
      "loss": 2.1531,
      "step": 3467
    },
    {
      "epoch": 2.9389830508474577,
      "grad_norm": 1.4131430387496948,
      "learning_rate": 1.016949152542373e-06,
      "loss": 2.1936,
      "step": 3468
    },
    {
      "epoch": 2.9398305084745764,
      "grad_norm": 2.067652463912964,
      "learning_rate": 1.002824858757062e-06,
      "loss": 1.7605,
      "step": 3469
    },
    {
      "epoch": 2.940677966101695,
      "grad_norm": 1.9416172504425049,
      "learning_rate": 9.887005649717514e-07,
      "loss": 1.9297,
      "step": 3470
    },
    {
      "epoch": 2.9415254237288133,
      "grad_norm": 1.131363868713379,
      "learning_rate": 9.745762711864408e-07,
      "loss": 2.5132,
      "step": 3471
    },
    {
      "epoch": 2.942372881355932,
      "grad_norm": 1.8346997499465942,
      "learning_rate": 9.604519774011302e-07,
      "loss": 2.0331,
      "step": 3472
    },
    {
      "epoch": 2.9432203389830507,
      "grad_norm": 1.6108843088150024,
      "learning_rate": 9.463276836158193e-07,
      "loss": 2.3032,
      "step": 3473
    },
    {
      "epoch": 2.9440677966101694,
      "grad_norm": 1.7487404346466064,
      "learning_rate": 9.322033898305086e-07,
      "loss": 2.134,
      "step": 3474
    },
    {
      "epoch": 2.944915254237288,
      "grad_norm": 1.2577048540115356,
      "learning_rate": 9.180790960451978e-07,
      "loss": 2.3789,
      "step": 3475
    },
    {
      "epoch": 2.9457627118644067,
      "grad_norm": 1.4981341361999512,
      "learning_rate": 9.039548022598871e-07,
      "loss": 2.3104,
      "step": 3476
    },
    {
      "epoch": 2.9466101694915254,
      "grad_norm": 1.8683017492294312,
      "learning_rate": 8.898305084745763e-07,
      "loss": 1.9977,
      "step": 3477
    },
    {
      "epoch": 2.947457627118644,
      "grad_norm": 1.9750769138336182,
      "learning_rate": 8.757062146892656e-07,
      "loss": 1.8513,
      "step": 3478
    },
    {
      "epoch": 2.948305084745763,
      "grad_norm": 1.6585434675216675,
      "learning_rate": 8.615819209039548e-07,
      "loss": 2.1858,
      "step": 3479
    },
    {
      "epoch": 2.9491525423728815,
      "grad_norm": 1.730651617050171,
      "learning_rate": 8.474576271186441e-07,
      "loss": 2.0628,
      "step": 3480
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.089017152786255,
      "learning_rate": 8.333333333333333e-07,
      "loss": 1.9277,
      "step": 3481
    },
    {
      "epoch": 2.950847457627119,
      "grad_norm": 1.5568169355392456,
      "learning_rate": 8.192090395480226e-07,
      "loss": 2.1834,
      "step": 3482
    },
    {
      "epoch": 2.9516949152542376,
      "grad_norm": 1.3532565832138062,
      "learning_rate": 8.050847457627118e-07,
      "loss": 2.3501,
      "step": 3483
    },
    {
      "epoch": 2.952542372881356,
      "grad_norm": 1.320961356163025,
      "learning_rate": 7.909604519774011e-07,
      "loss": 2.4191,
      "step": 3484
    },
    {
      "epoch": 2.9533898305084745,
      "grad_norm": 1.6994813680648804,
      "learning_rate": 7.768361581920904e-07,
      "loss": 2.1174,
      "step": 3485
    },
    {
      "epoch": 2.954237288135593,
      "grad_norm": 2.273296594619751,
      "learning_rate": 7.627118644067797e-07,
      "loss": 1.5144,
      "step": 3486
    },
    {
      "epoch": 2.955084745762712,
      "grad_norm": 1.3097214698791504,
      "learning_rate": 7.48587570621469e-07,
      "loss": 2.4397,
      "step": 3487
    },
    {
      "epoch": 2.9559322033898305,
      "grad_norm": 1.9971638917922974,
      "learning_rate": 7.344632768361582e-07,
      "loss": 1.8863,
      "step": 3488
    },
    {
      "epoch": 2.9567796610169492,
      "grad_norm": 1.8968480825424194,
      "learning_rate": 7.203389830508475e-07,
      "loss": 1.8625,
      "step": 3489
    },
    {
      "epoch": 2.957627118644068,
      "grad_norm": 1.6351325511932373,
      "learning_rate": 7.062146892655367e-07,
      "loss": 2.2286,
      "step": 3490
    },
    {
      "epoch": 2.9584745762711866,
      "grad_norm": 1.5790823698043823,
      "learning_rate": 6.92090395480226e-07,
      "loss": 2.2996,
      "step": 3491
    },
    {
      "epoch": 2.959322033898305,
      "grad_norm": 1.9345752000808716,
      "learning_rate": 6.779661016949152e-07,
      "loss": 1.9687,
      "step": 3492
    },
    {
      "epoch": 2.9601694915254235,
      "grad_norm": 2.0262906551361084,
      "learning_rate": 6.638418079096045e-07,
      "loss": 1.9205,
      "step": 3493
    },
    {
      "epoch": 2.961016949152542,
      "grad_norm": 1.7300212383270264,
      "learning_rate": 6.497175141242938e-07,
      "loss": 2.1565,
      "step": 3494
    },
    {
      "epoch": 2.961864406779661,
      "grad_norm": 1.7060924768447876,
      "learning_rate": 6.355932203389831e-07,
      "loss": 2.153,
      "step": 3495
    },
    {
      "epoch": 2.9627118644067796,
      "grad_norm": 1.7925012111663818,
      "learning_rate": 6.214689265536723e-07,
      "loss": 2.0491,
      "step": 3496
    },
    {
      "epoch": 2.9635593220338983,
      "grad_norm": 1.2668328285217285,
      "learning_rate": 6.073446327683616e-07,
      "loss": 2.4809,
      "step": 3497
    },
    {
      "epoch": 2.964406779661017,
      "grad_norm": 2.553030014038086,
      "learning_rate": 5.932203389830508e-07,
      "loss": 1.9113,
      "step": 3498
    },
    {
      "epoch": 2.9652542372881356,
      "grad_norm": 1.9871301651000977,
      "learning_rate": 5.790960451977401e-07,
      "loss": 1.7621,
      "step": 3499
    },
    {
      "epoch": 2.9661016949152543,
      "grad_norm": 1.802756428718567,
      "learning_rate": 5.649717514124295e-07,
      "loss": 1.9916,
      "step": 3500
    },
    {
      "epoch": 2.966949152542373,
      "grad_norm": 1.663110613822937,
      "learning_rate": 5.508474576271187e-07,
      "loss": 2.0954,
      "step": 3501
    },
    {
      "epoch": 2.9677966101694917,
      "grad_norm": 1.2281522750854492,
      "learning_rate": 5.36723163841808e-07,
      "loss": 2.4056,
      "step": 3502
    },
    {
      "epoch": 2.9686440677966104,
      "grad_norm": 1.5557782649993896,
      "learning_rate": 5.225988700564972e-07,
      "loss": 2.2795,
      "step": 3503
    },
    {
      "epoch": 2.969491525423729,
      "grad_norm": 1.5493392944335938,
      "learning_rate": 5.084745762711865e-07,
      "loss": 2.1856,
      "step": 3504
    },
    {
      "epoch": 2.9703389830508473,
      "grad_norm": 1.5563039779663086,
      "learning_rate": 4.943502824858757e-07,
      "loss": 2.155,
      "step": 3505
    },
    {
      "epoch": 2.971186440677966,
      "grad_norm": 1.797594666481018,
      "learning_rate": 4.802259887005651e-07,
      "loss": 1.9369,
      "step": 3506
    },
    {
      "epoch": 2.9720338983050847,
      "grad_norm": 1.5994278192520142,
      "learning_rate": 4.661016949152543e-07,
      "loss": 2.2201,
      "step": 3507
    },
    {
      "epoch": 2.9728813559322034,
      "grad_norm": 1.510952353477478,
      "learning_rate": 4.5197740112994353e-07,
      "loss": 2.1432,
      "step": 3508
    },
    {
      "epoch": 2.973728813559322,
      "grad_norm": 1.8224469423294067,
      "learning_rate": 4.378531073446328e-07,
      "loss": 1.9827,
      "step": 3509
    },
    {
      "epoch": 2.9745762711864407,
      "grad_norm": 1.4395874738693237,
      "learning_rate": 4.2372881355932204e-07,
      "loss": 2.4096,
      "step": 3510
    },
    {
      "epoch": 2.9754237288135594,
      "grad_norm": 1.8219739198684692,
      "learning_rate": 4.096045197740113e-07,
      "loss": 2.1854,
      "step": 3511
    },
    {
      "epoch": 2.976271186440678,
      "grad_norm": 1.499512791633606,
      "learning_rate": 3.9548022598870054e-07,
      "loss": 2.3892,
      "step": 3512
    },
    {
      "epoch": 2.9771186440677964,
      "grad_norm": 2.0295488834381104,
      "learning_rate": 3.8135593220338985e-07,
      "loss": 2.0847,
      "step": 3513
    },
    {
      "epoch": 2.977966101694915,
      "grad_norm": 1.4956918954849243,
      "learning_rate": 3.672316384180791e-07,
      "loss": 2.0691,
      "step": 3514
    },
    {
      "epoch": 2.9788135593220337,
      "grad_norm": 1.4348477125167847,
      "learning_rate": 3.5310734463276836e-07,
      "loss": 2.3293,
      "step": 3515
    },
    {
      "epoch": 2.9796610169491524,
      "grad_norm": 1.8539484739303589,
      "learning_rate": 3.389830508474576e-07,
      "loss": 2.0574,
      "step": 3516
    },
    {
      "epoch": 2.980508474576271,
      "grad_norm": 1.7122975587844849,
      "learning_rate": 3.248587570621469e-07,
      "loss": 2.0708,
      "step": 3517
    },
    {
      "epoch": 2.98135593220339,
      "grad_norm": 1.9447230100631714,
      "learning_rate": 3.1073446327683617e-07,
      "loss": 2.2801,
      "step": 3518
    },
    {
      "epoch": 2.9822033898305085,
      "grad_norm": 1.7502176761627197,
      "learning_rate": 2.966101694915254e-07,
      "loss": 2.0411,
      "step": 3519
    },
    {
      "epoch": 2.983050847457627,
      "grad_norm": 1.6554374694824219,
      "learning_rate": 2.8248587570621473e-07,
      "loss": 2.0652,
      "step": 3520
    },
    {
      "epoch": 2.983898305084746,
      "grad_norm": 1.910003662109375,
      "learning_rate": 2.68361581920904e-07,
      "loss": 1.9196,
      "step": 3521
    },
    {
      "epoch": 2.9847457627118645,
      "grad_norm": 1.9911201000213623,
      "learning_rate": 2.5423728813559323e-07,
      "loss": 1.9887,
      "step": 3522
    },
    {
      "epoch": 2.9855932203389832,
      "grad_norm": 1.9181311130523682,
      "learning_rate": 2.4011299435028254e-07,
      "loss": 2.6016,
      "step": 3523
    },
    {
      "epoch": 2.986440677966102,
      "grad_norm": 1.4985933303833008,
      "learning_rate": 2.2598870056497177e-07,
      "loss": 2.3922,
      "step": 3524
    },
    {
      "epoch": 2.9872881355932206,
      "grad_norm": 2.1219472885131836,
      "learning_rate": 2.1186440677966102e-07,
      "loss": 2.3057,
      "step": 3525
    },
    {
      "epoch": 2.988135593220339,
      "grad_norm": 1.5086705684661865,
      "learning_rate": 1.9774011299435027e-07,
      "loss": 2.345,
      "step": 3526
    },
    {
      "epoch": 2.9889830508474575,
      "grad_norm": 1.9622693061828613,
      "learning_rate": 1.8361581920903955e-07,
      "loss": 1.8735,
      "step": 3527
    },
    {
      "epoch": 2.989830508474576,
      "grad_norm": 1.5369393825531006,
      "learning_rate": 1.694915254237288e-07,
      "loss": 2.4739,
      "step": 3528
    },
    {
      "epoch": 2.990677966101695,
      "grad_norm": 1.6584341526031494,
      "learning_rate": 1.5536723163841808e-07,
      "loss": 2.2392,
      "step": 3529
    },
    {
      "epoch": 2.9915254237288136,
      "grad_norm": 1.76746666431427,
      "learning_rate": 1.4124293785310736e-07,
      "loss": 2.1265,
      "step": 3530
    },
    {
      "epoch": 2.9923728813559323,
      "grad_norm": 1.4161677360534668,
      "learning_rate": 1.2711864406779662e-07,
      "loss": 2.4357,
      "step": 3531
    },
    {
      "epoch": 2.993220338983051,
      "grad_norm": 1.9487510919570923,
      "learning_rate": 1.1299435028248588e-07,
      "loss": 2.1245,
      "step": 3532
    },
    {
      "epoch": 2.9940677966101696,
      "grad_norm": 1.567739725112915,
      "learning_rate": 9.887005649717514e-08,
      "loss": 2.2204,
      "step": 3533
    },
    {
      "epoch": 2.994915254237288,
      "grad_norm": 1.8734211921691895,
      "learning_rate": 8.47457627118644e-08,
      "loss": 2.0297,
      "step": 3534
    },
    {
      "epoch": 2.9957627118644066,
      "grad_norm": 1.9217379093170166,
      "learning_rate": 7.062146892655368e-08,
      "loss": 1.801,
      "step": 3535
    },
    {
      "epoch": 2.9966101694915253,
      "grad_norm": 1.9299284219741821,
      "learning_rate": 5.649717514124294e-08,
      "loss": 2.1399,
      "step": 3536
    },
    {
      "epoch": 2.997457627118644,
      "grad_norm": 2.1523752212524414,
      "learning_rate": 4.23728813559322e-08,
      "loss": 1.5545,
      "step": 3537
    },
    {
      "epoch": 2.9983050847457626,
      "grad_norm": 2.013671636581421,
      "learning_rate": 2.824858757062147e-08,
      "loss": 1.7243,
      "step": 3538
    },
    {
      "epoch": 2.9991525423728813,
      "grad_norm": 1.8303087949752808,
      "learning_rate": 1.4124293785310735e-08,
      "loss": 2.0243,
      "step": 3539
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8203142881393433,
      "learning_rate": 0.0,
      "loss": 2.1589,
      "step": 3540
    }
  ],
  "logging_steps": 1,
  "max_steps": 3540,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.65500133392384e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}