{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.98971898560658,
  "eval_steps": 500,
  "global_step": 910,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0054832076764907475,
      "grad_norm": 604.3109811491834,
      "learning_rate": 8.791208791208792e-07,
      "loss": 3.4565,
      "step": 1
    },
    {
      "epoch": 0.010966415352981495,
      "grad_norm": 602.1736348957847,
      "learning_rate": 1.7582417582417585e-06,
      "loss": 3.445,
      "step": 2
    },
    {
      "epoch": 0.01644962302947224,
      "grad_norm": 572.1582926794415,
      "learning_rate": 2.6373626373626375e-06,
      "loss": 3.426,
      "step": 3
    },
    {
      "epoch": 0.02193283070596299,
      "grad_norm": 301.9672730610099,
      "learning_rate": 3.516483516483517e-06,
      "loss": 3.2554,
      "step": 4
    },
    {
      "epoch": 0.027416038382453736,
      "grad_norm": 157.4890153555457,
      "learning_rate": 4.395604395604396e-06,
      "loss": 2.9123,
      "step": 5
    },
    {
      "epoch": 0.03289924605894448,
      "grad_norm": 93.35477967830393,
      "learning_rate": 5.274725274725275e-06,
      "loss": 2.6655,
      "step": 6
    },
    {
      "epoch": 0.03838245373543523,
      "grad_norm": 23.960002925273315,
      "learning_rate": 6.153846153846155e-06,
      "loss": 2.526,
      "step": 7
    },
    {
      "epoch": 0.04386566141192598,
      "grad_norm": 16.617925535315912,
      "learning_rate": 7.032967032967034e-06,
      "loss": 2.3244,
      "step": 8
    },
    {
      "epoch": 0.04934886908841672,
      "grad_norm": 13.65445534016627,
      "learning_rate": 7.912087912087913e-06,
      "loss": 2.2167,
      "step": 9
    },
    {
      "epoch": 0.05483207676490747,
      "grad_norm": 23.60213319182947,
      "learning_rate": 8.791208791208792e-06,
      "loss": 2.1214,
      "step": 10
    },
    {
      "epoch": 0.06031528444139822,
      "grad_norm": 9.750741286283983,
      "learning_rate": 9.670329670329671e-06,
      "loss": 2.0631,
      "step": 11
    },
    {
      "epoch": 0.06579849211788896,
      "grad_norm": 13.69478328276252,
      "learning_rate": 1.054945054945055e-05,
      "loss": 2.0035,
      "step": 12
    },
    {
      "epoch": 0.07128169979437972,
      "grad_norm": 8.863133767549522,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 1.9018,
      "step": 13
    },
    {
      "epoch": 0.07676490747087046,
      "grad_norm": 6.184273756096322,
      "learning_rate": 1.230769230769231e-05,
      "loss": 1.8409,
      "step": 14
    },
    {
      "epoch": 0.0822481151473612,
      "grad_norm": 8.463726907785716,
      "learning_rate": 1.3186813186813187e-05,
      "loss": 1.7862,
      "step": 15
    },
    {
      "epoch": 0.08773132282385196,
      "grad_norm": 5.9606779223424455,
      "learning_rate": 1.4065934065934068e-05,
      "loss": 1.7449,
      "step": 16
    },
    {
      "epoch": 0.0932145305003427,
      "grad_norm": 18.546012018929744,
      "learning_rate": 1.4945054945054947e-05,
      "loss": 1.7071,
      "step": 17
    },
    {
      "epoch": 0.09869773817683344,
      "grad_norm": 9.23610239091234,
      "learning_rate": 1.5824175824175826e-05,
      "loss": 1.7254,
      "step": 18
    },
    {
      "epoch": 0.1041809458533242,
      "grad_norm": 44.47081850643636,
      "learning_rate": 1.6703296703296707e-05,
      "loss": 1.6708,
      "step": 19
    },
    {
      "epoch": 0.10966415352981494,
      "grad_norm": 29.350706185079904,
      "learning_rate": 1.7582417582417584e-05,
      "loss": 1.7401,
      "step": 20
    },
    {
      "epoch": 0.11514736120630568,
      "grad_norm": 7.168404792591547,
      "learning_rate": 1.8461538461538465e-05,
      "loss": 1.6881,
      "step": 21
    },
    {
      "epoch": 0.12063056888279644,
      "grad_norm": 28.75909337885957,
      "learning_rate": 1.9340659340659342e-05,
      "loss": 1.5999,
      "step": 22
    },
    {
      "epoch": 0.12611377655928718,
      "grad_norm": 12.270722231807452,
      "learning_rate": 2.021978021978022e-05,
      "loss": 1.7174,
      "step": 23
    },
    {
      "epoch": 0.13159698423577793,
      "grad_norm": 51.182199832000634,
      "learning_rate": 2.10989010989011e-05,
      "loss": 1.6164,
      "step": 24
    },
    {
      "epoch": 0.13708019191226867,
      "grad_norm": 5.983294276939527,
      "learning_rate": 2.197802197802198e-05,
      "loss": 1.6279,
      "step": 25
    },
    {
      "epoch": 0.14256339958875944,
      "grad_norm": 12.36006380527371,
      "learning_rate": 2.2857142857142858e-05,
      "loss": 1.5335,
      "step": 26
    },
    {
      "epoch": 0.14804660726525018,
      "grad_norm": 3.912477135556689,
      "learning_rate": 2.373626373626374e-05,
      "loss": 1.5637,
      "step": 27
    },
    {
      "epoch": 0.15352981494174092,
      "grad_norm": 8.798850308211108,
      "learning_rate": 2.461538461538462e-05,
      "loss": 1.546,
      "step": 28
    },
    {
      "epoch": 0.15901302261823166,
      "grad_norm": 6.133566343370423,
      "learning_rate": 2.5494505494505493e-05,
      "loss": 1.5493,
      "step": 29
    },
    {
      "epoch": 0.1644962302947224,
      "grad_norm": 3.0393729351601753,
      "learning_rate": 2.6373626373626374e-05,
      "loss": 1.4988,
      "step": 30
    },
    {
      "epoch": 0.16997943797121315,
      "grad_norm": 28.575328415461772,
      "learning_rate": 2.7252747252747255e-05,
      "loss": 1.5026,
      "step": 31
    },
    {
      "epoch": 0.17546264564770392,
      "grad_norm": 27.101171796307156,
      "learning_rate": 2.8131868131868136e-05,
      "loss": 1.6522,
      "step": 32
    },
    {
      "epoch": 0.18094585332419466,
      "grad_norm": 7.286329527124996,
      "learning_rate": 2.9010989010989013e-05,
      "loss": 1.6616,
      "step": 33
    },
    {
      "epoch": 0.1864290610006854,
      "grad_norm": 4.786509337900977,
      "learning_rate": 2.9890109890109894e-05,
      "loss": 1.495,
      "step": 34
    },
    {
      "epoch": 0.19191226867717615,
      "grad_norm": 2.804143471750863,
      "learning_rate": 3.0769230769230774e-05,
      "loss": 1.4495,
      "step": 35
    },
    {
      "epoch": 0.1973954763536669,
      "grad_norm": 9.304401113296121,
      "learning_rate": 3.164835164835165e-05,
      "loss": 1.4403,
      "step": 36
    },
    {
      "epoch": 0.20287868403015763,
      "grad_norm": 5.457009814887685,
      "learning_rate": 3.252747252747253e-05,
      "loss": 1.4336,
      "step": 37
    },
    {
      "epoch": 0.2083618917066484,
      "grad_norm": 3.7957565393477437,
      "learning_rate": 3.340659340659341e-05,
      "loss": 1.4243,
      "step": 38
    },
    {
      "epoch": 0.21384509938313914,
      "grad_norm": 4.9220861710974075,
      "learning_rate": 3.4285714285714284e-05,
      "loss": 1.3984,
      "step": 39
    },
    {
      "epoch": 0.21932830705962988,
      "grad_norm": 4.736412206172175,
      "learning_rate": 3.516483516483517e-05,
      "loss": 1.3668,
      "step": 40
    },
    {
      "epoch": 0.22481151473612063,
      "grad_norm": 5.581594123475227,
      "learning_rate": 3.6043956043956045e-05,
      "loss": 1.3513,
      "step": 41
    },
    {
      "epoch": 0.23029472241261137,
      "grad_norm": 3.5585171099348427,
      "learning_rate": 3.692307692307693e-05,
      "loss": 1.3557,
      "step": 42
    },
    {
      "epoch": 0.2357779300891021,
      "grad_norm": 5.217567824661255,
      "learning_rate": 3.7802197802197807e-05,
      "loss": 1.3368,
      "step": 43
    },
    {
      "epoch": 0.24126113776559288,
      "grad_norm": 3.8941164699166784,
      "learning_rate": 3.8681318681318684e-05,
      "loss": 1.3192,
      "step": 44
    },
    {
      "epoch": 0.24674434544208362,
      "grad_norm": 3.3715823234615345,
      "learning_rate": 3.956043956043957e-05,
      "loss": 1.3194,
      "step": 45
    },
    {
      "epoch": 0.25222755311857437,
      "grad_norm": 4.071045479574351,
      "learning_rate": 4.043956043956044e-05,
      "loss": 1.3,
      "step": 46
    },
    {
      "epoch": 0.2577107607950651,
      "grad_norm": 4.790439556324944,
      "learning_rate": 4.131868131868133e-05,
      "loss": 1.3051,
      "step": 47
    },
    {
      "epoch": 0.26319396847155585,
      "grad_norm": 3.6760783714364793,
      "learning_rate": 4.21978021978022e-05,
      "loss": 1.2885,
      "step": 48
    },
    {
      "epoch": 0.2686771761480466,
      "grad_norm": 3.19497049785778,
      "learning_rate": 4.307692307692308e-05,
      "loss": 1.2962,
      "step": 49
    },
    {
      "epoch": 0.27416038382453733,
      "grad_norm": 2.620333577151409,
      "learning_rate": 4.395604395604396e-05,
      "loss": 1.2452,
      "step": 50
    },
    {
      "epoch": 0.2796435915010281,
      "grad_norm": 3.573140694736915,
      "learning_rate": 4.483516483516484e-05,
      "loss": 1.2912,
      "step": 51
    },
    {
      "epoch": 0.2851267991775189,
      "grad_norm": 2.3723854257821513,
      "learning_rate": 4.5714285714285716e-05,
      "loss": 1.2468,
      "step": 52
    },
    {
      "epoch": 0.2906100068540096,
      "grad_norm": 2.5628938638391277,
      "learning_rate": 4.65934065934066e-05,
      "loss": 1.2883,
      "step": 53
    },
    {
      "epoch": 0.29609321453050036,
      "grad_norm": 3.2477927500052184,
      "learning_rate": 4.747252747252748e-05,
      "loss": 1.2864,
      "step": 54
    },
    {
      "epoch": 0.3015764222069911,
      "grad_norm": 2.4110262583540716,
      "learning_rate": 4.8351648351648355e-05,
      "loss": 1.2886,
      "step": 55
    },
    {
      "epoch": 0.30705962988348184,
      "grad_norm": 3.9054257622250614,
      "learning_rate": 4.923076923076924e-05,
      "loss": 1.2544,
      "step": 56
    },
    {
      "epoch": 0.3125428375599726,
      "grad_norm": 2.684900989864461,
      "learning_rate": 5.0109890109890116e-05,
      "loss": 1.2506,
      "step": 57
    },
    {
      "epoch": 0.31802604523646333,
      "grad_norm": 3.1294292463297673,
      "learning_rate": 5.098901098901099e-05,
      "loss": 1.2619,
      "step": 58
    },
    {
      "epoch": 0.32350925291295407,
      "grad_norm": 4.302163409683541,
      "learning_rate": 5.186813186813188e-05,
      "loss": 1.2531,
      "step": 59
    },
    {
      "epoch": 0.3289924605894448,
      "grad_norm": 2.3198593094902566,
      "learning_rate": 5.274725274725275e-05,
      "loss": 1.2444,
      "step": 60
    },
    {
      "epoch": 0.33447566826593556,
      "grad_norm": 3.075723813655852,
      "learning_rate": 5.3626373626373626e-05,
      "loss": 1.2318,
      "step": 61
    },
    {
      "epoch": 0.3399588759424263,
      "grad_norm": 2.751318001428128,
      "learning_rate": 5.450549450549451e-05,
      "loss": 1.2283,
      "step": 62
    },
    {
      "epoch": 0.34544208361891704,
      "grad_norm": 3.792517967186425,
      "learning_rate": 5.538461538461539e-05,
      "loss": 1.2347,
      "step": 63
    },
    {
      "epoch": 0.35092529129540784,
      "grad_norm": 2.502677006785022,
      "learning_rate": 5.626373626373627e-05,
      "loss": 1.2154,
      "step": 64
    },
    {
      "epoch": 0.3564084989718986,
      "grad_norm": 2.4447089296161955,
      "learning_rate": 5.714285714285715e-05,
      "loss": 1.2062,
      "step": 65
    },
    {
      "epoch": 0.3618917066483893,
      "grad_norm": 2.7612204760550183,
      "learning_rate": 5.8021978021978026e-05,
      "loss": 1.2105,
      "step": 66
    },
    {
      "epoch": 0.36737491432488006,
      "grad_norm": 3.3730773809799968,
      "learning_rate": 5.890109890109891e-05,
      "loss": 1.2129,
      "step": 67
    },
    {
      "epoch": 0.3728581220013708,
      "grad_norm": 2.067462539301751,
      "learning_rate": 5.978021978021979e-05,
      "loss": 1.2021,
      "step": 68
    },
    {
      "epoch": 0.37834132967786155,
      "grad_norm": 3.367733224494582,
      "learning_rate": 6.0659340659340665e-05,
      "loss": 1.2207,
      "step": 69
    },
    {
      "epoch": 0.3838245373543523,
      "grad_norm": 2.1443388252000335,
      "learning_rate": 6.153846153846155e-05,
      "loss": 1.2025,
      "step": 70
    },
    {
      "epoch": 0.38930774503084303,
      "grad_norm": 3.4158140446082994,
      "learning_rate": 6.241758241758242e-05,
      "loss": 1.2132,
      "step": 71
    },
    {
      "epoch": 0.3947909527073338,
      "grad_norm": 2.58088978525386,
      "learning_rate": 6.32967032967033e-05,
      "loss": 1.1843,
      "step": 72
    },
    {
      "epoch": 0.4002741603838245,
      "grad_norm": 2.897001887626188,
      "learning_rate": 6.417582417582419e-05,
      "loss": 1.1996,
      "step": 73
    },
    {
      "epoch": 0.40575736806031526,
      "grad_norm": 3.0910986197243573,
      "learning_rate": 6.505494505494506e-05,
      "loss": 1.215,
      "step": 74
    },
    {
      "epoch": 0.41124057573680606,
      "grad_norm": 2.36902708485275,
      "learning_rate": 6.593406593406594e-05,
      "loss": 1.172,
      "step": 75
    },
    {
      "epoch": 0.4167237834132968,
      "grad_norm": 2.9090585544359366,
      "learning_rate": 6.681318681318683e-05,
      "loss": 1.192,
      "step": 76
    },
    {
      "epoch": 0.42220699108978754,
      "grad_norm": 2.682374090006958,
      "learning_rate": 6.76923076923077e-05,
      "loss": 1.1929,
      "step": 77
    },
    {
      "epoch": 0.4276901987662783,
      "grad_norm": 2.6472629365834037,
      "learning_rate": 6.857142857142857e-05,
      "loss": 1.1578,
      "step": 78
    },
    {
      "epoch": 0.433173406442769,
      "grad_norm": 2.58566062639449,
      "learning_rate": 6.945054945054945e-05,
      "loss": 1.153,
      "step": 79
    },
    {
      "epoch": 0.43865661411925977,
      "grad_norm": 2.7406033885199848,
      "learning_rate": 7.032967032967034e-05,
      "loss": 1.1697,
      "step": 80
    },
    {
      "epoch": 0.4441398217957505,
      "grad_norm": 2.137577114283556,
      "learning_rate": 7.12087912087912e-05,
      "loss": 1.1492,
      "step": 81
    },
    {
      "epoch": 0.44962302947224125,
      "grad_norm": 2.9855017798950976,
      "learning_rate": 7.208791208791209e-05,
      "loss": 1.1584,
      "step": 82
    },
    {
      "epoch": 0.455106237148732,
      "grad_norm": 2.220583427738456,
      "learning_rate": 7.296703296703297e-05,
      "loss": 1.1437,
      "step": 83
    },
    {
      "epoch": 0.46058944482522274,
      "grad_norm": 2.706030331520761,
      "learning_rate": 7.384615384615386e-05,
      "loss": 1.1673,
      "step": 84
    },
    {
      "epoch": 0.4660726525017135,
      "grad_norm": 3.190829729449872,
      "learning_rate": 7.472527472527473e-05,
      "loss": 1.1413,
      "step": 85
    },
    {
      "epoch": 0.4715558601782042,
      "grad_norm": 1.9236806571413316,
      "learning_rate": 7.560439560439561e-05,
      "loss": 1.1546,
      "step": 86
    },
    {
      "epoch": 0.477039067854695,
      "grad_norm": 2.9656722488411,
      "learning_rate": 7.64835164835165e-05,
      "loss": 1.1705,
      "step": 87
    },
    {
      "epoch": 0.48252227553118576,
      "grad_norm": 1.7873438452480972,
      "learning_rate": 7.736263736263737e-05,
      "loss": 1.1512,
      "step": 88
    },
    {
      "epoch": 0.4880054832076765,
      "grad_norm": 3.366153588378385,
      "learning_rate": 7.824175824175825e-05,
      "loss": 1.1854,
      "step": 89
    },
    {
      "epoch": 0.49348869088416725,
      "grad_norm": 2.3979761080289306,
      "learning_rate": 7.912087912087914e-05,
      "loss": 1.1785,
      "step": 90
    },
    {
      "epoch": 0.498971898560658,
      "grad_norm": 2.68813176726075,
      "learning_rate": 8e-05,
      "loss": 1.1689,
      "step": 91
    },
    {
      "epoch": 0.5044551062371487,
      "grad_norm": 3.634421050069424,
      "learning_rate": 7.999970571955439e-05,
      "loss": 1.1534,
      "step": 92
    },
    {
      "epoch": 0.5099383139136395,
      "grad_norm": 2.110037855631259,
      "learning_rate": 7.999882288254757e-05,
      "loss": 1.1398,
      "step": 93
    },
    {
      "epoch": 0.5154215215901302,
      "grad_norm": 3.029801321990504,
      "learning_rate": 7.999735150196965e-05,
      "loss": 1.15,
      "step": 94
    },
    {
      "epoch": 0.520904729266621,
      "grad_norm": 1.9878631276509169,
      "learning_rate": 7.999529159947053e-05,
      "loss": 1.1524,
      "step": 95
    },
    {
      "epoch": 0.5263879369431117,
      "grad_norm": 2.997439251632094,
      "learning_rate": 7.999264320535968e-05,
      "loss": 1.161,
      "step": 96
    },
    {
      "epoch": 0.5318711446196025,
      "grad_norm": 1.7997765525760119,
      "learning_rate": 7.998940635860564e-05,
      "loss": 1.1392,
      "step": 97
    },
    {
      "epoch": 0.5373543522960932,
      "grad_norm": 2.3934821316457353,
      "learning_rate": 7.998558110683542e-05,
      "loss": 1.1511,
      "step": 98
    },
    {
      "epoch": 0.542837559972584,
      "grad_norm": 2.394286604934712,
      "learning_rate": 7.998116750633388e-05,
      "loss": 1.1482,
      "step": 99
    },
    {
      "epoch": 0.5483207676490747,
      "grad_norm": 2.5638379332850447,
      "learning_rate": 7.997616562204282e-05,
      "loss": 1.132,
      "step": 100
    },
    {
      "epoch": 0.5538039753255655,
      "grad_norm": 2.4073701211461715,
      "learning_rate": 7.99705755275601e-05,
      "loss": 1.1653,
      "step": 101
    },
    {
      "epoch": 0.5592871830020562,
      "grad_norm": 2.9908052478558127,
      "learning_rate": 7.996439730513846e-05,
      "loss": 1.1437,
      "step": 102
    },
    {
      "epoch": 0.564770390678547,
      "grad_norm": 2.338649947963062,
      "learning_rate": 7.995763104568444e-05,
      "loss": 1.1257,
      "step": 103
    },
    {
      "epoch": 0.5702535983550377,
      "grad_norm": 2.1324859396744853,
      "learning_rate": 7.99502768487569e-05,
      "loss": 1.1082,
      "step": 104
    },
    {
      "epoch": 0.5757368060315284,
      "grad_norm": 1.8207695969892468,
      "learning_rate": 7.994233482256567e-05,
      "loss": 1.1324,
      "step": 105
    },
    {
      "epoch": 0.5812200137080192,
      "grad_norm": 3.0671728842532793,
      "learning_rate": 7.993380508396992e-05,
      "loss": 1.1673,
      "step": 106
    },
    {
      "epoch": 0.5867032213845099,
      "grad_norm": 1.2132079987970015,
      "learning_rate": 7.992468775847638e-05,
      "loss": 1.1332,
      "step": 107
    },
    {
      "epoch": 0.5921864290610007,
      "grad_norm": 2.8722682327215647,
      "learning_rate": 7.99149829802376e-05,
      "loss": 1.1937,
      "step": 108
    },
    {
      "epoch": 0.5976696367374914,
      "grad_norm": 1.578063060667415,
      "learning_rate": 7.990469089204992e-05,
      "loss": 1.1516,
      "step": 109
    },
    {
      "epoch": 0.6031528444139822,
      "grad_norm": 2.4024809989588336,
      "learning_rate": 7.989381164535131e-05,
      "loss": 1.1776,
      "step": 110
    },
    {
      "epoch": 0.6086360520904729,
      "grad_norm": 1.7795121351579433,
      "learning_rate": 7.988234540021928e-05,
      "loss": 1.153,
      "step": 111
    },
    {
      "epoch": 0.6141192597669637,
      "grad_norm": 2.5507155124877032,
      "learning_rate": 7.987029232536841e-05,
      "loss": 1.1546,
      "step": 112
    },
    {
      "epoch": 0.6196024674434544,
      "grad_norm": 2.7262135259336095,
      "learning_rate": 7.98576525981479e-05,
      "loss": 1.1681,
      "step": 113
    },
    {
      "epoch": 0.6250856751199452,
      "grad_norm": 1.756472159467694,
      "learning_rate": 7.9844426404539e-05,
      "loss": 1.1337,
      "step": 114
    },
    {
      "epoch": 0.630568882796436,
      "grad_norm": 3.664337440475356,
      "learning_rate": 7.983061393915222e-05,
      "loss": 1.1458,
      "step": 115
    },
    {
      "epoch": 0.6360520904729267,
      "grad_norm": 1.951846935673031,
      "learning_rate": 7.981621540522444e-05,
      "loss": 1.1322,
      "step": 116
    },
    {
      "epoch": 0.6415352981494175,
      "grad_norm": 1.7976750357904085,
      "learning_rate": 7.980123101461606e-05,
      "loss": 1.1263,
      "step": 117
    },
    {
      "epoch": 0.6470185058259081,
      "grad_norm": 2.300277427581849,
      "learning_rate": 7.978566098780771e-05,
      "loss": 1.1413,
      "step": 118
    },
    {
      "epoch": 0.6525017135023989,
      "grad_norm": 2.360648922071168,
      "learning_rate": 7.976950555389713e-05,
      "loss": 1.1274,
      "step": 119
    },
    {
      "epoch": 0.6579849211788896,
      "grad_norm": 2.0913575786408147,
      "learning_rate": 7.97527649505957e-05,
      "loss": 1.1213,
      "step": 120
    },
    {
      "epoch": 0.6634681288553804,
      "grad_norm": 2.022249517249403,
      "learning_rate": 7.973543942422506e-05,
      "loss": 1.1175,
      "step": 121
    },
    {
      "epoch": 0.6689513365318711,
      "grad_norm": 2.2294222010405362,
      "learning_rate": 7.97175292297134e-05,
      "loss": 1.1281,
      "step": 122
    },
    {
      "epoch": 0.6744345442083619,
      "grad_norm": 1.7133244378182648,
      "learning_rate": 7.969903463059169e-05,
      "loss": 1.1099,
      "step": 123
    },
    {
      "epoch": 0.6799177518848526,
      "grad_norm": 2.0157523028333775,
      "learning_rate": 7.96799558989899e-05,
      "loss": 1.1175,
      "step": 124
    },
    {
      "epoch": 0.6854009595613434,
      "grad_norm": 1.7297097607076724,
      "learning_rate": 7.966029331563287e-05,
      "loss": 1.1131,
      "step": 125
    },
    {
      "epoch": 0.6908841672378341,
      "grad_norm": 2.2317660437246034,
      "learning_rate": 7.964004716983635e-05,
      "loss": 1.1066,
      "step": 126
    },
    {
      "epoch": 0.6963673749143249,
      "grad_norm": 1.9857286499887847,
      "learning_rate": 7.961921775950254e-05,
      "loss": 1.1083,
      "step": 127
    },
    {
      "epoch": 0.7018505825908157,
      "grad_norm": 1.6946992419248856,
      "learning_rate": 7.959780539111585e-05,
      "loss": 1.0985,
      "step": 128
    },
    {
      "epoch": 0.7073337902673064,
      "grad_norm": 1.949379922441364,
      "learning_rate": 7.957581037973835e-05,
      "loss": 1.0792,
      "step": 129
    },
    {
      "epoch": 0.7128169979437972,
      "grad_norm": 2.0374474232968454,
      "learning_rate": 7.955323304900514e-05,
      "loss": 1.1081,
      "step": 130
    },
    {
      "epoch": 0.7183002056202878,
      "grad_norm": 1.5382297067387796,
      "learning_rate": 7.953007373111956e-05,
      "loss": 1.0894,
      "step": 131
    },
    {
      "epoch": 0.7237834132967786,
      "grad_norm": 1.994423662926792,
      "learning_rate": 7.950633276684833e-05,
      "loss": 1.0971,
      "step": 132
    },
    {
      "epoch": 0.7292666209732693,
      "grad_norm": 1.6956272845939566,
      "learning_rate": 7.948201050551651e-05,
      "loss": 1.078,
      "step": 133
    },
    {
      "epoch": 0.7347498286497601,
      "grad_norm": 1.808121530050818,
      "learning_rate": 7.945710730500243e-05,
      "loss": 1.087,
      "step": 134
    },
    {
      "epoch": 0.7402330363262508,
      "grad_norm": 1.6031163127588859,
      "learning_rate": 7.943162353173232e-05,
      "loss": 1.0878,
      "step": 135
    },
    {
      "epoch": 0.7457162440027416,
      "grad_norm": 2.8069443037445057,
      "learning_rate": 7.940555956067495e-05,
      "loss": 1.0852,
      "step": 136
    },
    {
      "epoch": 0.7511994516792323,
      "grad_norm": 1.023257465419051,
      "learning_rate": 7.937891577533624e-05,
      "loss": 1.0864,
      "step": 137
    },
    {
      "epoch": 0.7566826593557231,
      "grad_norm": 3.7981840568943266,
      "learning_rate": 7.93516925677534e-05,
      "loss": 1.1577,
      "step": 138
    },
    {
      "epoch": 0.7621658670322139,
      "grad_norm": 1.957367420792266,
      "learning_rate": 7.932389033848931e-05,
      "loss": 1.1428,
      "step": 139
    },
    {
      "epoch": 0.7676490747087046,
      "grad_norm": 2.620391650383465,
      "learning_rate": 7.929550949662659e-05,
      "loss": 1.1262,
      "step": 140
    },
    {
      "epoch": 0.7731322823851954,
      "grad_norm": 1.8055917441902372,
      "learning_rate": 7.92665504597616e-05,
      "loss": 1.1087,
      "step": 141
    },
    {
      "epoch": 0.7786154900616861,
      "grad_norm": 1.5140320658198094,
      "learning_rate": 7.923701365399826e-05,
      "loss": 1.1206,
      "step": 142
    },
    {
      "epoch": 0.7840986977381769,
      "grad_norm": 2.3311158871467983,
      "learning_rate": 7.920689951394175e-05,
      "loss": 1.1354,
      "step": 143
    },
    {
      "epoch": 0.7895819054146676,
      "grad_norm": 2.871009250239358,
      "learning_rate": 7.917620848269224e-05,
      "loss": 1.1159,
      "step": 144
    },
    {
      "epoch": 0.7950651130911583,
      "grad_norm": 1.174246832683915,
      "learning_rate": 7.914494101183822e-05,
      "loss": 1.112,
      "step": 145
    },
    {
      "epoch": 0.800548320767649,
      "grad_norm": 1.9715656445082748,
      "learning_rate": 7.911309756144995e-05,
      "loss": 1.1233,
      "step": 146
    },
    {
      "epoch": 0.8060315284441398,
      "grad_norm": 1.3667047276351785,
      "learning_rate": 7.908067860007268e-05,
      "loss": 1.0966,
      "step": 147
    },
    {
      "epoch": 0.8115147361206305,
      "grad_norm": 1.8908076306855728,
      "learning_rate": 7.904768460471975e-05,
      "loss": 1.118,
      "step": 148
    },
    {
      "epoch": 0.8169979437971213,
      "grad_norm": 1.6160590718366654,
      "learning_rate": 7.90141160608655e-05,
      "loss": 1.0836,
      "step": 149
    },
    {
      "epoch": 0.8224811514736121,
      "grad_norm": 2.257542807740706,
      "learning_rate": 7.897997346243825e-05,
      "loss": 1.0985,
      "step": 150
    },
    {
      "epoch": 0.8279643591501028,
      "grad_norm": 1.2533944841043951,
      "learning_rate": 7.894525731181297e-05,
      "loss": 1.0678,
      "step": 151
    },
    {
      "epoch": 0.8334475668265936,
      "grad_norm": 2.4996679708615988,
      "learning_rate": 7.890996811980386e-05,
      "loss": 1.1143,
      "step": 152
    },
    {
      "epoch": 0.8389307745030843,
      "grad_norm": 1.6215671222990462,
      "learning_rate": 7.887410640565689e-05,
      "loss": 1.0947,
      "step": 153
    },
    {
      "epoch": 0.8444139821795751,
      "grad_norm": 1.4102568455851006,
      "learning_rate": 7.883767269704209e-05,
      "loss": 1.0635,
      "step": 154
    },
    {
      "epoch": 0.8498971898560658,
      "grad_norm": 2.0199914067223763,
      "learning_rate": 7.880066753004588e-05,
      "loss": 1.1007,
      "step": 155
    },
    {
      "epoch": 0.8553803975325566,
      "grad_norm": 1.4233619206252466,
      "learning_rate": 7.876309144916312e-05,
      "loss": 1.0964,
      "step": 156
    },
    {
      "epoch": 0.8608636052090473,
      "grad_norm": 1.5315953952088068,
      "learning_rate": 7.87249450072891e-05,
      "loss": 1.1091,
      "step": 157
    },
    {
      "epoch": 0.866346812885538,
      "grad_norm": 1.5672306259553663,
      "learning_rate": 7.86862287657114e-05,
      "loss": 1.0961,
      "step": 158
    },
    {
      "epoch": 0.8718300205620287,
      "grad_norm": 1.722337979582118,
      "learning_rate": 7.864694329410168e-05,
      "loss": 1.1032,
      "step": 159
    },
    {
      "epoch": 0.8773132282385195,
      "grad_norm": 1.1707259582896943,
      "learning_rate": 7.860708917050722e-05,
      "loss": 1.083,
      "step": 160
    },
    {
      "epoch": 0.8827964359150103,
      "grad_norm": 2.1260596185722216,
      "learning_rate": 7.85666669813425e-05,
      "loss": 1.0775,
      "step": 161
    },
    {
      "epoch": 0.888279643591501,
      "grad_norm": 1.1049918316810494,
      "learning_rate": 7.852567732138051e-05,
      "loss": 1.0804,
      "step": 162
    },
    {
      "epoch": 0.8937628512679918,
      "grad_norm": 1.5314067038496302,
      "learning_rate": 7.848412079374403e-05,
      "loss": 1.0931,
      "step": 163
    },
    {
      "epoch": 0.8992460589444825,
      "grad_norm": 1.526207291286965,
      "learning_rate": 7.844199800989672e-05,
      "loss": 1.126,
      "step": 164
    },
    {
      "epoch": 0.9047292666209733,
      "grad_norm": 1.4889589302339163,
      "learning_rate": 7.839930958963415e-05,
      "loss": 1.0688,
      "step": 165
    },
    {
      "epoch": 0.910212474297464,
      "grad_norm": 1.8224369715346593,
      "learning_rate": 7.835605616107471e-05,
      "loss": 1.0981,
      "step": 166
    },
    {
      "epoch": 0.9156956819739548,
      "grad_norm": 1.9809507356149727,
      "learning_rate": 7.83122383606503e-05,
      "loss": 1.0716,
      "step": 167
    },
    {
      "epoch": 0.9211788896504455,
      "grad_norm": 1.319128673805323,
      "learning_rate": 7.826785683309702e-05,
      "loss": 1.0511,
      "step": 168
    },
    {
      "epoch": 0.9266620973269363,
      "grad_norm": 1.7628250921107829,
      "learning_rate": 7.822291223144564e-05,
      "loss": 1.0815,
      "step": 169
    },
    {
      "epoch": 0.932145305003427,
      "grad_norm": 1.5619673215880712,
      "learning_rate": 7.817740521701204e-05,
      "loss": 1.0851,
      "step": 170
    },
    {
      "epoch": 0.9376285126799178,
      "grad_norm": 1.6877201996768514,
      "learning_rate": 7.813133645938744e-05,
      "loss": 1.0655,
      "step": 171
    },
    {
      "epoch": 0.9431117203564084,
      "grad_norm": 1.4453595064062965,
      "learning_rate": 7.808470663642856e-05,
      "loss": 1.0468,
      "step": 172
    },
    {
      "epoch": 0.9485949280328992,
      "grad_norm": 1.3988144950022152,
      "learning_rate": 7.803751643424769e-05,
      "loss": 1.0684,
      "step": 173
    },
    {
      "epoch": 0.95407813570939,
      "grad_norm": 1.829632986890131,
      "learning_rate": 7.798976654720248e-05,
      "loss": 1.0658,
      "step": 174
    },
    {
      "epoch": 0.9595613433858807,
      "grad_norm": 1.160599126459967,
      "learning_rate": 7.794145767788582e-05,
      "loss": 1.0666,
      "step": 175
    },
    {
      "epoch": 0.9650445510623715,
      "grad_norm": 1.6311368539718085,
      "learning_rate": 7.789259053711554e-05,
      "loss": 1.0732,
      "step": 176
    },
    {
      "epoch": 0.9705277587388622,
      "grad_norm": 1.3145190591456402,
      "learning_rate": 7.784316584392379e-05,
      "loss": 1.0564,
      "step": 177
    },
    {
      "epoch": 0.976010966415353,
      "grad_norm": 1.434756052069605,
      "learning_rate": 7.779318432554663e-05,
      "loss": 1.0624,
      "step": 178
    },
    {
      "epoch": 0.9814941740918437,
      "grad_norm": 1.3078999315572475,
      "learning_rate": 7.774264671741324e-05,
      "loss": 1.0486,
      "step": 179
    },
    {
      "epoch": 0.9869773817683345,
      "grad_norm": 1.4474824177372518,
      "learning_rate": 7.769155376313509e-05,
      "loss": 1.0511,
      "step": 180
    },
    {
      "epoch": 0.9924605894448252,
      "grad_norm": 1.4306506181803345,
      "learning_rate": 7.763990621449507e-05,
      "loss": 1.0657,
      "step": 181
    },
    {
      "epoch": 0.997943797121316,
      "grad_norm": 1.5854524936397603,
      "learning_rate": 7.758770483143634e-05,
      "loss": 1.0638,
      "step": 182
    },
    {
      "epoch": 1.0034270047978067,
      "grad_norm": 2.818624037527729,
      "learning_rate": 7.753495038205123e-05,
      "loss": 1.7465,
      "step": 183
    },
    {
      "epoch": 1.0089102124742975,
      "grad_norm": 1.2531970095723952,
      "learning_rate": 7.748164364256989e-05,
      "loss": 1.0425,
      "step": 184
    },
    {
      "epoch": 1.0143934201507883,
      "grad_norm": 1.8473887590213065,
      "learning_rate": 7.742778539734884e-05,
      "loss": 1.0309,
      "step": 185
    },
    {
      "epoch": 1.019876627827279,
      "grad_norm": 1.1978766650574866,
      "learning_rate": 7.737337643885956e-05,
      "loss": 1.0365,
      "step": 186
    },
    {
      "epoch": 1.0253598355037696,
      "grad_norm": 1.6327228122841784,
      "learning_rate": 7.73184175676766e-05,
      "loss": 1.0412,
      "step": 187
    },
    {
      "epoch": 1.0308430431802604,
      "grad_norm": 1.038869594675331,
      "learning_rate": 7.726290959246606e-05,
      "loss": 1.0325,
      "step": 188
    },
    {
      "epoch": 1.0363262508567512,
      "grad_norm": 1.8931974706378887,
      "learning_rate": 7.720685332997354e-05,
      "loss": 1.0575,
      "step": 189
    },
    {
      "epoch": 1.041809458533242,
      "grad_norm": 1.0099409227594682,
      "learning_rate": 7.715024960501209e-05,
      "loss": 1.0356,
      "step": 190
    },
    {
      "epoch": 1.0472926662097326,
      "grad_norm": 1.7017523773128445,
      "learning_rate": 7.709309925045023e-05,
      "loss": 1.048,
      "step": 191
    },
    {
      "epoch": 1.0527758738862234,
      "grad_norm": 1.0632647956537205,
      "learning_rate": 7.70354031071995e-05,
      "loss": 1.0619,
      "step": 192
    },
    {
      "epoch": 1.0582590815627142,
      "grad_norm": 1.6480122219648587,
      "learning_rate": 7.697716202420227e-05,
      "loss": 1.0626,
      "step": 193
    },
    {
      "epoch": 1.063742289239205,
      "grad_norm": 1.337533514643958,
      "learning_rate": 7.691837685841913e-05,
      "loss": 1.0521,
      "step": 194
    },
    {
      "epoch": 1.0692254969156956,
      "grad_norm": 1.405204005470527,
      "learning_rate": 7.685904847481631e-05,
      "loss": 1.0488,
      "step": 195
    },
    {
      "epoch": 1.0747087045921864,
      "grad_norm": 1.0789567262015418,
      "learning_rate": 7.679917774635298e-05,
      "loss": 1.0354,
      "step": 196
    },
    {
      "epoch": 1.0801919122686772,
      "grad_norm": 1.5059550846436562,
      "learning_rate": 7.673876555396835e-05,
      "loss": 1.0488,
      "step": 197
    },
    {
      "epoch": 1.085675119945168,
      "grad_norm": 1.285703383014824,
      "learning_rate": 7.667781278656879e-05,
      "loss": 1.0326,
      "step": 198
    },
    {
      "epoch": 1.0911583276216588,
      "grad_norm": 1.166732355574816,
      "learning_rate": 7.661632034101466e-05,
      "loss": 1.0322,
      "step": 199
    },
    {
      "epoch": 1.0966415352981493,
      "grad_norm": 1.6867809480621192,
      "learning_rate": 7.655428912210718e-05,
      "loss": 1.0364,
      "step": 200
    },
    {
      "epoch": 1.1021247429746401,
      "grad_norm": 1.2531266932796477,
      "learning_rate": 7.64917200425751e-05,
      "loss": 1.0374,
      "step": 201
    },
    {
      "epoch": 1.107607950651131,
      "grad_norm": 1.6183973189746483,
      "learning_rate": 7.642861402306123e-05,
      "loss": 1.0302,
      "step": 202
    },
    {
      "epoch": 1.1130911583276217,
      "grad_norm": 1.5885719958570816,
      "learning_rate": 7.636497199210895e-05,
      "loss": 1.0312,
      "step": 203
    },
    {
      "epoch": 1.1185743660041123,
      "grad_norm": 1.271489486550103,
      "learning_rate": 7.630079488614853e-05,
      "loss": 1.0439,
      "step": 204
    },
    {
      "epoch": 1.124057573680603,
      "grad_norm": 1.6077174464582225,
      "learning_rate": 7.623608364948334e-05,
      "loss": 1.0369,
      "step": 205
    },
    {
      "epoch": 1.129540781357094,
      "grad_norm": 1.0526829284737067,
      "learning_rate": 7.617083923427596e-05,
      "loss": 1.0294,
      "step": 206
    },
    {
      "epoch": 1.1350239890335847,
      "grad_norm": 1.8026775192610152,
      "learning_rate": 7.610506260053415e-05,
      "loss": 1.0249,
      "step": 207
    },
    {
      "epoch": 1.1405071967100753,
      "grad_norm": 1.1092234118793582,
      "learning_rate": 7.603875471609677e-05,
      "loss": 1.0479,
      "step": 208
    },
    {
      "epoch": 1.145990404386566,
      "grad_norm": 1.252308944775191,
      "learning_rate": 7.597191655661952e-05,
      "loss": 1.0377,
      "step": 209
    },
    {
      "epoch": 1.1514736120630569,
      "grad_norm": 1.481646296662436,
      "learning_rate": 7.590454910556058e-05,
      "loss": 1.0475,
      "step": 210
    },
    {
      "epoch": 1.1569568197395477,
      "grad_norm": 1.1007433182898057,
      "learning_rate": 7.583665335416608e-05,
      "loss": 1.0105,
      "step": 211
    },
    {
      "epoch": 1.1624400274160385,
      "grad_norm": 1.2445540839929121,
      "learning_rate": 7.576823030145566e-05,
      "loss": 1.0236,
      "step": 212
    },
    {
      "epoch": 1.167923235092529,
      "grad_norm": 1.4019899879808024,
      "learning_rate": 7.569928095420762e-05,
      "loss": 1.0437,
      "step": 213
    },
    {
      "epoch": 1.1734064427690198,
      "grad_norm": 1.293921037672153,
      "learning_rate": 7.562980632694421e-05,
      "loss": 1.0078,
      "step": 214
    },
    {
      "epoch": 1.1788896504455106,
      "grad_norm": 1.574451864443491,
      "learning_rate": 7.555980744191666e-05,
      "loss": 1.0183,
      "step": 215
    },
    {
      "epoch": 1.1843728581220014,
      "grad_norm": 1.0261703383565228,
      "learning_rate": 7.548928532909006e-05,
      "loss": 1.0243,
      "step": 216
    },
    {
      "epoch": 1.1898560657984922,
      "grad_norm": 1.3888687863743303,
      "learning_rate": 7.541824102612839e-05,
      "loss": 1.0241,
      "step": 217
    },
    {
      "epoch": 1.1953392734749828,
      "grad_norm": 1.435498169074072,
      "learning_rate": 7.534667557837912e-05,
      "loss": 1.0348,
      "step": 218
    },
    {
      "epoch": 1.2008224811514736,
      "grad_norm": 1.1812741919729888,
      "learning_rate": 7.527459003885783e-05,
      "loss": 1.013,
      "step": 219
    },
    {
      "epoch": 1.2063056888279644,
      "grad_norm": 1.209488409322031,
      "learning_rate": 7.520198546823275e-05,
      "loss": 1.0279,
      "step": 220
    },
    {
      "epoch": 1.211788896504455,
      "grad_norm": 1.5861672644314255,
      "learning_rate": 7.512886293480914e-05,
      "loss": 1.0217,
      "step": 221
    },
    {
      "epoch": 1.2172721041809458,
      "grad_norm": 1.0784984727297116,
      "learning_rate": 7.505522351451363e-05,
      "loss": 1.0308,
      "step": 222
    },
    {
      "epoch": 1.2227553118574366,
      "grad_norm": 1.4013682384096804,
      "learning_rate": 7.498106829087822e-05,
      "loss": 1.0209,
      "step": 223
    },
    {
      "epoch": 1.2282385195339274,
      "grad_norm": 1.1069329108834502,
      "learning_rate": 7.490639835502458e-05,
      "loss": 1.0184,
      "step": 224
    },
    {
      "epoch": 1.2337217272104182,
      "grad_norm": 1.2292930770165162,
      "learning_rate": 7.483121480564779e-05,
      "loss": 1.0192,
      "step": 225
    },
    {
      "epoch": 1.2392049348869087,
      "grad_norm": 1.1091280093145537,
      "learning_rate": 7.475551874900027e-05,
      "loss": 1.0227,
      "step": 226
    },
    {
      "epoch": 1.2446881425633995,
      "grad_norm": 1.3883445894297355,
      "learning_rate": 7.467931129887548e-05,
      "loss": 1.0317,
      "step": 227
    },
    {
      "epoch": 1.2501713502398903,
      "grad_norm": 1.3258593612132867,
      "learning_rate": 7.460259357659155e-05,
      "loss": 1.0276,
      "step": 228
    },
    {
      "epoch": 1.2556545579163811,
      "grad_norm": 1.2962983275547162,
      "learning_rate": 7.452536671097476e-05,
      "loss": 1.0155,
      "step": 229
    },
    {
      "epoch": 1.261137765592872,
      "grad_norm": 1.2345976366071654,
      "learning_rate": 7.444763183834292e-05,
      "loss": 0.9999,
      "step": 230
    },
    {
      "epoch": 1.2666209732693625,
      "grad_norm": 1.371675205771468,
      "learning_rate": 7.436939010248868e-05,
      "loss": 1.0242,
      "step": 231
    },
    {
      "epoch": 1.2721041809458533,
      "grad_norm": 1.043563875055389,
      "learning_rate": 7.429064265466269e-05,
      "loss": 0.9978,
      "step": 232
    },
    {
      "epoch": 1.2775873886223441,
      "grad_norm": 1.4841522976495107,
      "learning_rate": 7.421139065355663e-05,
      "loss": 1.0152,
      "step": 233
    },
    {
      "epoch": 1.2830705962988347,
      "grad_norm": 0.9605681175633184,
      "learning_rate": 7.413163526528623e-05,
      "loss": 1.0162,
      "step": 234
    },
    {
      "epoch": 1.2885538039753255,
      "grad_norm": 1.2067100095706547,
      "learning_rate": 7.405137766337406e-05,
      "loss": 1.0565,
      "step": 235
    },
    {
      "epoch": 1.2940370116518163,
      "grad_norm": 1.5489835530200247,
      "learning_rate": 7.397061902873223e-05,
      "loss": 1.0316,
      "step": 236
    },
    {
      "epoch": 1.299520219328307,
      "grad_norm": 0.936867837906044,
      "learning_rate": 7.388936054964512e-05,
      "loss": 1.0218,
      "step": 237
    },
    {
      "epoch": 1.3050034270047979,
      "grad_norm": 1.3113159040022304,
      "learning_rate": 7.38076034217518e-05,
      "loss": 1.0305,
      "step": 238
    },
    {
      "epoch": 1.3104866346812885,
      "grad_norm": 0.9712447789123708,
      "learning_rate": 7.372534884802844e-05,
      "loss": 1.0403,
      "step": 239
    },
    {
      "epoch": 1.3159698423577793,
      "grad_norm": 1.2727545703286138,
      "learning_rate": 7.364259803877072e-05,
      "loss": 1.0366,
      "step": 240
    },
    {
      "epoch": 1.32145305003427,
      "grad_norm": 1.1239260262608586,
      "learning_rate": 7.355935221157584e-05,
      "loss": 1.0031,
      "step": 241
    },
    {
      "epoch": 1.3269362577107608,
      "grad_norm": 1.1122104721303026,
      "learning_rate": 7.347561259132479e-05,
      "loss": 1.0358,
      "step": 242
    },
    {
      "epoch": 1.3324194653872516,
      "grad_norm": 1.4676521237806195,
      "learning_rate": 7.33913804101642e-05,
      "loss": 1.0362,
      "step": 243
    },
    {
      "epoch": 1.3379026730637422,
      "grad_norm": 1.2984708218697247,
      "learning_rate": 7.330665690748825e-05,
      "loss": 1.0095,
      "step": 244
    },
    {
      "epoch": 1.343385880740233,
      "grad_norm": 1.2692918018410921,
      "learning_rate": 7.322144332992047e-05,
      "loss": 1.0126,
      "step": 245
    },
    {
      "epoch": 1.3488690884167238,
      "grad_norm": 0.9963650435822852,
      "learning_rate": 7.313574093129532e-05,
      "loss": 0.9937,
      "step": 246
    },
    {
      "epoch": 1.3543522960932146,
      "grad_norm": 1.45458036698363,
      "learning_rate": 7.30495509726398e-05,
      "loss": 1.0389,
      "step": 247
    },
    {
      "epoch": 1.3598355037697054,
      "grad_norm": 1.0844293357139385,
      "learning_rate": 7.29628747221549e-05,
      "loss": 1.042,
      "step": 248
    },
    {
      "epoch": 1.365318711446196,
      "grad_norm": 1.3150162241275731,
      "learning_rate": 7.287571345519688e-05,
      "loss": 1.0076,
      "step": 249
    },
    {
      "epoch": 1.3708019191226868,
      "grad_norm": 1.170486608498072,
      "learning_rate": 7.278806845425856e-05,
      "loss": 1.0411,
      "step": 250
    },
    {
      "epoch": 1.3762851267991776,
      "grad_norm": 1.0369874340952356,
      "learning_rate": 7.269994100895047e-05,
      "loss": 1.0137,
      "step": 251
    },
    {
      "epoch": 1.3817683344756682,
      "grad_norm": 1.3592340170974726,
      "learning_rate": 7.261133241598177e-05,
      "loss": 1.0104,
      "step": 252
    },
    {
      "epoch": 1.387251542152159,
      "grad_norm": 1.0387909603628305,
      "learning_rate": 7.25222439791413e-05,
      "loss": 1.0039,
      "step": 253
    },
    {
      "epoch": 1.3927347498286498,
      "grad_norm": 1.2165525508974753,
      "learning_rate": 7.24326770092783e-05,
      "loss": 1.0117,
      "step": 254
    },
    {
      "epoch": 1.3982179575051406,
      "grad_norm": 1.0298641450176182,
      "learning_rate": 7.234263282428312e-05,
      "loss": 1.0087,
      "step": 255
    },
    {
      "epoch": 1.4037011651816313,
      "grad_norm": 2.3062469579312768,
      "learning_rate": 7.225211274906795e-05,
      "loss": 1.0376,
      "step": 256
    },
    {
      "epoch": 1.409184372858122,
      "grad_norm": 1.0448171443413676,
      "learning_rate": 7.216111811554718e-05,
      "loss": 1.015,
      "step": 257
    },
    {
      "epoch": 1.4146675805346127,
      "grad_norm": 2.282441256400205,
      "learning_rate": 7.206965026261787e-05,
      "loss": 1.0506,
      "step": 258
    },
    {
      "epoch": 1.4201507882111035,
      "grad_norm": 1.5228239109337287,
      "learning_rate": 7.197771053614006e-05,
      "loss": 1.0488,
      "step": 259
    },
    {
      "epoch": 1.4256339958875943,
      "grad_norm": 7.1797647736282535,
      "learning_rate": 7.188530028891691e-05,
      "loss": 1.0477,
      "step": 260
    },
    {
      "epoch": 1.4311172035640851,
      "grad_norm": 19.858381633007593,
      "learning_rate": 7.179242088067487e-05,
      "loss": 1.2469,
      "step": 261
    },
    {
      "epoch": 1.4366004112405757,
      "grad_norm": 12.439075720330774,
      "learning_rate": 7.169907367804363e-05,
      "loss": 1.3376,
      "step": 262
    },
    {
      "epoch": 1.4420836189170665,
      "grad_norm": 10.336294598744168,
      "learning_rate": 7.160526005453599e-05,
      "loss": 1.2434,
      "step": 263
    },
    {
      "epoch": 1.4475668265935573,
      "grad_norm": 3325.4750490493866,
      "learning_rate": 7.151098139052772e-05,
      "loss": 4.0042,
      "step": 264
    },
    {
      "epoch": 1.4530500342700479,
      "grad_norm": 131.55901739711575,
      "learning_rate": 7.141623907323717e-05,
      "loss": 1.8859,
      "step": 265
    },
    {
      "epoch": 1.4585332419465387,
      "grad_norm": 11.388495314344222,
      "learning_rate": 7.13210344967049e-05,
      "loss": 1.3033,
      "step": 266
    },
    {
      "epoch": 1.4640164496230295,
      "grad_norm": 5.045627021720784,
      "learning_rate": 7.122536906177318e-05,
      "loss": 1.1984,
      "step": 267
    },
    {
      "epoch": 1.4694996572995203,
      "grad_norm": 63.13120968720244,
      "learning_rate": 7.112924417606536e-05,
      "loss": 1.2658,
      "step": 268
    },
    {
      "epoch": 1.474982864976011,
      "grad_norm": 78.12090408575494,
      "learning_rate": 7.103266125396512e-05,
      "loss": 1.6972,
      "step": 269
    },
    {
      "epoch": 1.4804660726525016,
      "grad_norm": 40.95778397870689,
      "learning_rate": 7.093562171659577e-05,
      "loss": 1.4509,
      "step": 270
    },
    {
      "epoch": 1.4859492803289924,
      "grad_norm": 37.962763812559935,
      "learning_rate": 7.083812699179919e-05,
      "loss": 1.34,
      "step": 271
    },
    {
      "epoch": 1.4914324880054832,
      "grad_norm": 26.82750605185721,
      "learning_rate": 7.074017851411495e-05,
      "loss": 1.2961,
      "step": 272
    },
    {
      "epoch": 1.496915695681974,
      "grad_norm": 4.438206627978028,
      "learning_rate": 7.064177772475912e-05,
      "loss": 1.2112,
      "step": 273
    },
    {
      "epoch": 1.5023989033584648,
      "grad_norm": 8.269192562168978,
      "learning_rate": 7.054292607160313e-05,
      "loss": 1.1848,
      "step": 274
    },
    {
      "epoch": 1.5078821110349554,
      "grad_norm": 3.4078464376801048,
      "learning_rate": 7.044362500915239e-05,
      "loss": 1.1018,
      "step": 275
    },
    {
      "epoch": 1.5133653187114462,
      "grad_norm": 3.8991572015827836,
      "learning_rate": 7.034387599852494e-05,
      "loss": 1.0906,
      "step": 276
    },
    {
      "epoch": 1.518848526387937,
      "grad_norm": 2.2792452805163577,
      "learning_rate": 7.024368050742996e-05,
      "loss": 1.0535,
      "step": 277
    },
    {
      "epoch": 1.5243317340644276,
      "grad_norm": 7.92182168324662,
      "learning_rate": 7.014304001014614e-05,
      "loss": 1.09,
      "step": 278
    },
    {
      "epoch": 1.5298149417409186,
      "grad_norm": 4.604463247427479,
      "learning_rate": 7.004195598749997e-05,
      "loss": 1.0929,
      "step": 279
    },
    {
      "epoch": 1.5352981494174092,
      "grad_norm": 1.6269169421948018,
      "learning_rate": 6.994042992684406e-05,
      "loss": 1.0721,
      "step": 280
    },
    {
      "epoch": 1.5407813570939,
      "grad_norm": 3.1522500098672723,
      "learning_rate": 6.983846332203508e-05,
      "loss": 1.052,
      "step": 281
    },
    {
      "epoch": 1.5462645647703908,
      "grad_norm": 1.744104230884076,
      "learning_rate": 6.973605767341194e-05,
      "loss": 1.0536,
      "step": 282
    },
    {
      "epoch": 1.5517477724468813,
      "grad_norm": 1.7435972526503143,
      "learning_rate": 6.963321448777367e-05,
      "loss": 1.0562,
      "step": 283
    },
    {
      "epoch": 1.5572309801233721,
      "grad_norm": 2.0699287078075783,
      "learning_rate": 6.952993527835714e-05,
      "loss": 1.0508,
      "step": 284
    },
    {
      "epoch": 1.562714187799863,
      "grad_norm": 2.5216940307065374,
      "learning_rate": 6.942622156481498e-05,
      "loss": 1.0696,
      "step": 285
    },
    {
      "epoch": 1.5681973954763535,
      "grad_norm": 1.5542171373751579,
      "learning_rate": 6.932207487319305e-05,
      "loss": 1.0351,
      "step": 286
    },
    {
      "epoch": 1.5736806031528445,
      "grad_norm": 1.1081954487560843,
      "learning_rate": 6.921749673590813e-05,
      "loss": 1.0399,
      "step": 287
    },
    {
      "epoch": 1.579163810829335,
      "grad_norm": 1.5993253594110808,
      "learning_rate": 6.911248869172523e-05,
      "loss": 1.0364,
      "step": 288
    },
    {
      "epoch": 1.584647018505826,
      "grad_norm": 1.077571235080621,
      "learning_rate": 6.900705228573507e-05,
      "loss": 1.0142,
      "step": 289
    },
    {
      "epoch": 1.5901302261823167,
      "grad_norm": 1.2982202905714664,
      "learning_rate": 6.890118906933126e-05,
      "loss": 1.0182,
      "step": 290
    },
    {
      "epoch": 1.5956134338588073,
      "grad_norm": 1.4040374561281856,
      "learning_rate": 6.879490060018754e-05,
      "loss": 1.0215,
      "step": 291
    },
    {
      "epoch": 1.6010966415352983,
      "grad_norm": 1.6011611520244904,
      "learning_rate": 6.86881884422348e-05,
      "loss": 0.9977,
      "step": 292
    },
    {
      "epoch": 1.6065798492117889,
      "grad_norm": 1.7528977017352316,
      "learning_rate": 6.858105416563812e-05,
      "loss": 0.9914,
      "step": 293
    },
    {
      "epoch": 1.6120630568882797,
      "grad_norm": 1.0189454039968566,
      "learning_rate": 6.847349934677363e-05,
      "loss": 1.0233,
      "step": 294
    },
    {
      "epoch": 1.6175462645647705,
      "grad_norm": 1.6918231992095927,
      "learning_rate": 6.836552556820533e-05,
      "loss": 1.0325,
      "step": 295
    },
    {
      "epoch": 1.623029472241261,
      "grad_norm": 0.9564755595265833,
      "learning_rate": 6.82571344186618e-05,
      "loss": 1.0198,
      "step": 296
    },
    {
      "epoch": 1.6285126799177518,
      "grad_norm": 1.2847377343488389,
      "learning_rate": 6.814832749301285e-05,
      "loss": 1.0148,
      "step": 297
    },
    {
      "epoch": 1.6339958875942426,
      "grad_norm": 0.9873795646279533,
      "learning_rate": 6.803910639224598e-05,
      "loss": 1.0357,
      "step": 298
    },
    {
      "epoch": 1.6394790952707334,
      "grad_norm": 1.5820758914116304,
      "learning_rate": 6.792947272344292e-05,
      "loss": 1.0376,
      "step": 299
    },
    {
      "epoch": 1.6449623029472242,
      "grad_norm": 0.901645544400314,
      "learning_rate": 6.78194280997559e-05,
      "loss": 1.0506,
      "step": 300
    },
    {
      "epoch": 1.6504455106237148,
      "grad_norm": 1.4932631104259941,
      "learning_rate": 6.770897414038398e-05,
      "loss": 1.0178,
      "step": 301
    },
    {
      "epoch": 1.6559287183002056,
      "grad_norm": 0.7587099965919871,
      "learning_rate": 6.759811247054918e-05,
      "loss": 1.0304,
      "step": 302
    },
    {
      "epoch": 1.6614119259766964,
      "grad_norm": 1.6220429069155198,
      "learning_rate": 6.748684472147255e-05,
      "loss": 1.0042,
      "step": 303
    },
    {
      "epoch": 1.666895133653187,
      "grad_norm": 0.9060198165841729,
      "learning_rate": 6.737517253035027e-05,
      "loss": 1.0264,
      "step": 304
    },
    {
      "epoch": 1.672378341329678,
      "grad_norm": 0.7810840322821543,
      "learning_rate": 6.726309754032942e-05,
      "loss": 1.0194,
      "step": 305
    },
    {
      "epoch": 1.6778615490061686,
      "grad_norm": 0.8389807329289671,
      "learning_rate": 6.715062140048392e-05,
      "loss": 1.0336,
      "step": 306
    },
    {
      "epoch": 1.6833447566826594,
      "grad_norm": 0.9811139261812258,
      "learning_rate": 6.703774576579018e-05,
      "loss": 0.9961,
      "step": 307
    },
    {
      "epoch": 1.6888279643591502,
      "grad_norm": 1.693203875036964,
      "learning_rate": 6.69244722971028e-05,
      "loss": 1.0054,
      "step": 308
    },
    {
      "epoch": 1.6943111720356407,
      "grad_norm": 0.6313152362555375,
      "learning_rate": 6.681080266113017e-05,
      "loss": 1.0138,
      "step": 309
    },
    {
      "epoch": 1.6997943797121315,
      "grad_norm": 1.2050041258375503,
      "learning_rate": 6.669673853040979e-05,
      "loss": 1.0181,
      "step": 310
    },
    {
      "epoch": 1.7052775873886223,
      "grad_norm": 1.0692574241468813,
      "learning_rate": 6.658228158328384e-05,
      "loss": 1.0196,
      "step": 311
    },
    {
      "epoch": 1.7107607950651131,
      "grad_norm": 1.3523681454597505,
      "learning_rate": 6.646743350387438e-05,
      "loss": 1.0405,
      "step": 312
    },
    {
      "epoch": 1.716244002741604,
      "grad_norm": 0.9778959208393927,
      "learning_rate": 6.635219598205863e-05,
      "loss": 1.0308,
      "step": 313
    },
    {
      "epoch": 1.7217272104180945,
      "grad_norm": 0.9710639823435132,
      "learning_rate": 6.623657071344407e-05,
      "loss": 0.9845,
      "step": 314
    },
    {
      "epoch": 1.7272104180945853,
      "grad_norm": 1.1795695782951767,
      "learning_rate": 6.61205593993434e-05,
      "loss": 1.0354,
      "step": 315
    },
    {
      "epoch": 1.732693625771076,
      "grad_norm": 1.256640921349798,
      "learning_rate": 6.600416374674978e-05,
      "loss": 1.0074,
      "step": 316
    },
    {
      "epoch": 1.7381768334475667,
      "grad_norm": 0.989814843263136,
      "learning_rate": 6.588738546831136e-05,
      "loss": 1.025,
      "step": 317
    },
    {
      "epoch": 1.7436600411240577,
      "grad_norm": 0.9259711398370861,
      "learning_rate": 6.577022628230638e-05,
      "loss": 1.0132,
      "step": 318
    },
    {
      "epoch": 1.7491432488005483,
      "grad_norm": 0.7510586657590674,
      "learning_rate": 6.565268791261769e-05,
      "loss": 1.0092,
      "step": 319
    },
    {
      "epoch": 1.754626456477039,
      "grad_norm": 0.6249763060539171,
      "learning_rate": 6.553477208870748e-05,
      "loss": 1.0198,
      "step": 320
    },
    {
      "epoch": 1.7601096641535299,
      "grad_norm": 0.6961806925064483,
      "learning_rate": 6.541648054559182e-05,
      "loss": 1.0265,
      "step": 321
    },
    {
      "epoch": 1.7655928718300204,
      "grad_norm": 0.8447275384338139,
      "learning_rate": 6.529781502381509e-05,
      "loss": 1.014,
      "step": 322
    },
    {
      "epoch": 1.7710760795065115,
      "grad_norm": 1.052987904165586,
      "learning_rate": 6.517877726942445e-05,
      "loss": 0.9936,
      "step": 323
    },
    {
      "epoch": 1.776559287183002,
      "grad_norm": 1.2555546445227173,
      "learning_rate": 6.505936903394406e-05,
      "loss": 0.9993,
      "step": 324
    },
    {
      "epoch": 1.7820424948594928,
      "grad_norm": 0.7871410118515071,
      "learning_rate": 6.493959207434934e-05,
      "loss": 1.0246,
      "step": 325
    },
    {
      "epoch": 1.7875257025359836,
      "grad_norm": 0.8254368841706734,
      "learning_rate": 6.481944815304117e-05,
      "loss": 0.9871,
      "step": 326
    },
    {
      "epoch": 1.7930089102124742,
      "grad_norm": 2.389893141482841,
      "learning_rate": 6.469893903781987e-05,
      "loss": 1.0288,
      "step": 327
    },
    {
      "epoch": 1.798492117888965,
      "grad_norm": 0.8529982927275416,
      "learning_rate": 6.457806650185925e-05,
      "loss": 1.0209,
      "step": 328
    },
    {
      "epoch": 1.8039753255654558,
      "grad_norm": 1.9419731761625634,
      "learning_rate": 6.44568323236805e-05,
      "loss": 0.9977,
      "step": 329
    },
    {
      "epoch": 1.8094585332419464,
      "grad_norm": 0.8112015130720743,
      "learning_rate": 6.433523828712599e-05,
      "loss": 1.0182,
      "step": 330
    },
    {
      "epoch": 1.8149417409184374,
      "grad_norm": 1.8647552723000032,
      "learning_rate": 6.421328618133312e-05,
      "loss": 1.0059,
      "step": 331
    },
    {
      "epoch": 1.820424948594928,
      "grad_norm": 1.1202805462423913,
      "learning_rate": 6.409097780070789e-05,
      "loss": 1.0315,
      "step": 332
    },
    {
      "epoch": 1.8259081562714188,
      "grad_norm": 1.4717552883952516,
      "learning_rate": 6.396831494489852e-05,
      "loss": 1.0178,
      "step": 333
    },
    {
      "epoch": 1.8313913639479096,
      "grad_norm": 1.247198837427445,
      "learning_rate": 6.384529941876902e-05,
      "loss": 1.0244,
      "step": 334
    },
    {
      "epoch": 1.8368745716244002,
      "grad_norm": 1.165713928714055,
      "learning_rate": 6.372193303237258e-05,
      "loss": 1.0234,
      "step": 335
    },
    {
      "epoch": 1.8423577793008912,
      "grad_norm": 0.819719891951629,
      "learning_rate": 6.359821760092493e-05,
      "loss": 1.0252,
      "step": 336
    },
    {
      "epoch": 1.8478409869773817,
      "grad_norm": 0.9694295545160206,
      "learning_rate": 6.347415494477771e-05,
      "loss": 1.0005,
      "step": 337
    },
    {
      "epoch": 1.8533241946538725,
      "grad_norm": 0.833369832433467,
      "learning_rate": 6.334974688939161e-05,
      "loss": 1.0005,
      "step": 338
    },
    {
      "epoch": 1.8588074023303633,
      "grad_norm": 1.051580411769736,
      "learning_rate": 6.322499526530951e-05,
      "loss": 0.994,
      "step": 339
    },
    {
      "epoch": 1.864290610006854,
      "grad_norm": 1.3409908447267063,
      "learning_rate": 6.30999019081296e-05,
      "loss": 1.0051,
      "step": 340
    },
    {
      "epoch": 1.8697738176833447,
      "grad_norm": 0.7200296682178966,
      "learning_rate": 6.297446865847833e-05,
      "loss": 0.9916,
      "step": 341
    },
    {
      "epoch": 1.8752570253598355,
      "grad_norm": 0.5703748532613969,
      "learning_rate": 6.284869736198332e-05,
      "loss": 1.0117,
      "step": 342
    },
    {
      "epoch": 1.880740233036326,
      "grad_norm": 0.656233302341983,
      "learning_rate": 6.272258986924624e-05,
      "loss": 0.9995,
      "step": 343
    },
    {
      "epoch": 1.8862234407128171,
      "grad_norm": 0.9959957194372924,
      "learning_rate": 6.259614803581553e-05,
      "loss": 0.9863,
      "step": 344
    },
    {
      "epoch": 1.8917066483893077,
      "grad_norm": 1.2637430599411104,
      "learning_rate": 6.246937372215916e-05,
      "loss": 1.0044,
      "step": 345
    },
    {
      "epoch": 1.8971898560657985,
      "grad_norm": 0.6655587539599807,
      "learning_rate": 6.23422687936372e-05,
      "loss": 1.004,
      "step": 346
    },
    {
      "epoch": 1.9026730637422893,
      "grad_norm": 0.6292486765776564,
      "learning_rate": 6.22148351204744e-05,
      "loss": 1.0071,
      "step": 347
    },
    {
      "epoch": 1.9081562714187799,
      "grad_norm": 0.7336634351921288,
      "learning_rate": 6.208707457773267e-05,
      "loss": 0.9866,
      "step": 348
    },
    {
      "epoch": 1.9136394790952709,
      "grad_norm": 0.6842175616230739,
      "learning_rate": 6.195898904528346e-05,
      "loss": 0.9911,
      "step": 349
    },
    {
      "epoch": 1.9191226867717615,
      "grad_norm": 0.6937670660831978,
      "learning_rate": 6.183058040778018e-05,
      "loss": 0.9927,
      "step": 350
    },
    {
      "epoch": 1.9246058944482523,
      "grad_norm": 0.6943683013295557,
      "learning_rate": 6.170185055463039e-05,
      "loss": 0.9883,
      "step": 351
    },
    {
      "epoch": 1.930089102124743,
      "grad_norm": 0.6759661412127859,
      "learning_rate": 6.157280137996797e-05,
      "loss": 0.9792,
      "step": 352
    },
    {
      "epoch": 1.9355723098012336,
      "grad_norm": 0.7487530675075756,
      "learning_rate": 6.14434347826254e-05,
      "loss": 0.9917,
      "step": 353
    },
    {
      "epoch": 1.9410555174777244,
      "grad_norm": 0.8480627285748816,
      "learning_rate": 6.131375266610564e-05,
      "loss": 1.0116,
      "step": 354
    },
    {
      "epoch": 1.9465387251542152,
      "grad_norm": 1.026881712032747,
      "learning_rate": 6.118375693855426e-05,
      "loss": 0.997,
      "step": 355
    },
    {
      "epoch": 1.9520219328307058,
      "grad_norm": 1.1819824734836144,
      "learning_rate": 6.10534495127313e-05,
      "loss": 0.9866,
      "step": 356
    },
    {
      "epoch": 1.9575051405071968,
      "grad_norm": 0.7488114847350166,
      "learning_rate": 6.092283230598311e-05,
      "loss": 0.9992,
      "step": 357
    },
    {
      "epoch": 1.9629883481836874,
      "grad_norm": 0.5942125372742829,
      "learning_rate": 6.079190724021418e-05,
      "loss": 1.0019,
      "step": 358
    },
    {
      "epoch": 1.9684715558601782,
      "grad_norm": 0.6416782063906716,
      "learning_rate": 6.066067624185886e-05,
      "loss": 0.9851,
      "step": 359
    },
    {
      "epoch": 1.973954763536669,
      "grad_norm": 0.6541261024351206,
      "learning_rate": 6.0529141241852974e-05,
      "loss": 0.9967,
      "step": 360
    },
    {
      "epoch": 1.9794379712131596,
      "grad_norm": 0.7497124499642812,
      "learning_rate": 6.0397304175605444e-05,
      "loss": 0.9809,
      "step": 361
    },
    {
      "epoch": 1.9849211788896506,
      "grad_norm": 0.868567720778104,
      "learning_rate": 6.026516698296979e-05,
      "loss": 1.0105,
      "step": 362
    },
    {
      "epoch": 1.9904043865661412,
      "grad_norm": 0.9302778924884637,
      "learning_rate": 6.0132731608215626e-05,
      "loss": 0.9922,
      "step": 363
    },
    {
      "epoch": 1.995887594242632,
      "grad_norm": 0.9976983084275789,
      "learning_rate": 6.000000000000001e-05,
      "loss": 1.0146,
      "step": 364
    },
    {
      "epoch": 2.0013708019191228,
      "grad_norm": 1.7195064119299488,
      "learning_rate": 5.9866974111338764e-05,
      "loss": 1.5667,
      "step": 365
    },
    {
      "epoch": 2.0068540095956133,
      "grad_norm": 0.8460708264395937,
      "learning_rate": 5.973365589957777e-05,
      "loss": 0.9585,
      "step": 366
    },
    {
      "epoch": 2.0123372172721044,
      "grad_norm": 0.6860754084451314,
      "learning_rate": 5.96000473263642e-05,
      "loss": 0.9479,
      "step": 367
    },
    {
      "epoch": 2.017820424948595,
      "grad_norm": 0.569858067536389,
      "learning_rate": 5.946615035761756e-05,
      "loss": 0.9485,
      "step": 368
    },
    {
      "epoch": 2.0233036326250855,
      "grad_norm": 0.4334842237308715,
      "learning_rate": 5.9331966963500825e-05,
      "loss": 0.9368,
      "step": 369
    },
    {
      "epoch": 2.0287868403015765,
      "grad_norm": 0.5181677147256706,
      "learning_rate": 5.919749911839146e-05,
      "loss": 0.9804,
      "step": 370
    },
    {
      "epoch": 2.034270047978067,
      "grad_norm": 0.6881500556087109,
      "learning_rate": 5.9062748800852315e-05,
      "loss": 0.9499,
      "step": 371
    },
    {
      "epoch": 2.039753255654558,
      "grad_norm": 0.7541551892158753,
      "learning_rate": 5.892771799360258e-05,
      "loss": 0.9832,
      "step": 372
    },
    {
      "epoch": 2.0452364633310487,
      "grad_norm": 0.7791303072747863,
      "learning_rate": 5.879240868348857e-05,
      "loss": 0.9449,
      "step": 373
    },
    {
      "epoch": 2.0507196710075393,
      "grad_norm": 0.7144846911633607,
      "learning_rate": 5.865682286145446e-05,
      "loss": 0.9534,
      "step": 374
    },
    {
      "epoch": 2.0562028786840303,
      "grad_norm": 0.677470076470832,
      "learning_rate": 5.852096252251308e-05,
      "loss": 0.9579,
      "step": 375
    },
    {
      "epoch": 2.061686086360521,
      "grad_norm": 0.7427233443080802,
      "learning_rate": 5.8384829665716475e-05,
      "loss": 0.9233,
      "step": 376
    },
    {
      "epoch": 2.0671692940370114,
      "grad_norm": 0.7777873567557757,
      "learning_rate": 5.824842629412653e-05,
      "loss": 0.9706,
      "step": 377
    },
    {
      "epoch": 2.0726525017135025,
      "grad_norm": 0.7135923513575179,
      "learning_rate": 5.8111754414785504e-05,
      "loss": 0.9232,
      "step": 378
    },
    {
      "epoch": 2.078135709389993,
      "grad_norm": 0.7125391347465736,
      "learning_rate": 5.797481603868646e-05,
      "loss": 0.9525,
      "step": 379
    },
    {
      "epoch": 2.083618917066484,
      "grad_norm": 0.7238619031777882,
      "learning_rate": 5.783761318074373e-05,
      "loss": 0.9525,
      "step": 380
    },
    {
      "epoch": 2.0891021247429746,
      "grad_norm": 0.7128949358918412,
      "learning_rate": 5.770014785976322e-05,
      "loss": 0.9511,
      "step": 381
    },
    {
      "epoch": 2.094585332419465,
      "grad_norm": 0.7209215696044494,
      "learning_rate": 5.756242209841272e-05,
      "loss": 0.9636,
      "step": 382
    },
    {
      "epoch": 2.1000685400959562,
      "grad_norm": 0.6327662060979778,
      "learning_rate": 5.742443792319216e-05,
      "loss": 0.968,
      "step": 383
    },
    {
      "epoch": 2.105551747772447,
      "grad_norm": 0.4985051187489137,
      "learning_rate": 5.728619736440375e-05,
      "loss": 0.953,
      "step": 384
    },
    {
      "epoch": 2.111034955448938,
      "grad_norm": 0.3861018767101939,
      "learning_rate": 5.714770245612217e-05,
      "loss": 0.967,
      "step": 385
    },
    {
      "epoch": 2.1165181631254284,
      "grad_norm": 0.32959266195178644,
      "learning_rate": 5.700895523616459e-05,
      "loss": 0.9281,
      "step": 386
    },
    {
      "epoch": 2.122001370801919,
      "grad_norm": 0.34591568155580654,
      "learning_rate": 5.6869957746060675e-05,
      "loss": 0.9521,
      "step": 387
    },
    {
      "epoch": 2.12748457847841,
      "grad_norm": 0.4625069922403467,
      "learning_rate": 5.673071203102261e-05,
      "loss": 0.9628,
      "step": 388
    },
    {
      "epoch": 2.1329677861549006,
      "grad_norm": 0.4988336639530923,
      "learning_rate": 5.6591220139914945e-05,
      "loss": 0.9568,
      "step": 389
    },
    {
      "epoch": 2.138450993831391,
      "grad_norm": 0.49194383493690574,
      "learning_rate": 5.645148412522447e-05,
      "loss": 0.9524,
      "step": 390
    },
    {
      "epoch": 2.143934201507882,
      "grad_norm": 0.43352665389513567,
      "learning_rate": 5.6311506043030006e-05,
      "loss": 0.9469,
      "step": 391
    },
    {
      "epoch": 2.1494174091843727,
      "grad_norm": 0.3306646386411896,
      "learning_rate": 5.61712879529722e-05,
      "loss": 0.9417,
      "step": 392
    },
    {
      "epoch": 2.1549006168608638,
      "grad_norm": 0.29441855117758814,
      "learning_rate": 5.6030831918223136e-05,
      "loss": 0.9579,
      "step": 393
    },
    {
      "epoch": 2.1603838245373543,
      "grad_norm": 0.3597929142259359,
      "learning_rate": 5.5890140005456056e-05,
      "loss": 0.9299,
      "step": 394
    },
    {
      "epoch": 2.165867032213845,
      "grad_norm": 0.4844488175844035,
      "learning_rate": 5.574921428481487e-05,
      "loss": 0.9401,
      "step": 395
    },
    {
      "epoch": 2.171350239890336,
      "grad_norm": 0.6802142906239217,
      "learning_rate": 5.5608056829883796e-05,
      "loss": 0.9357,
      "step": 396
    },
    {
      "epoch": 2.1768334475668265,
      "grad_norm": 1.2106002196981427,
      "learning_rate": 5.546666971765675e-05,
      "loss": 0.9651,
      "step": 397
    },
    {
      "epoch": 2.1823166552433175,
      "grad_norm": 0.6157717006727198,
      "learning_rate": 5.532505502850688e-05,
      "loss": 0.9528,
      "step": 398
    },
    {
      "epoch": 2.187799862919808,
      "grad_norm": 0.641804078912436,
      "learning_rate": 5.5183214846155864e-05,
      "loss": 0.9718,
      "step": 399
    },
    {
      "epoch": 2.1932830705962987,
      "grad_norm": 0.6070401599275227,
      "learning_rate": 5.504115125764329e-05,
      "loss": 0.9572,
      "step": 400
    },
    {
      "epoch": 2.1987662782727897,
      "grad_norm": 0.7100305243791368,
      "learning_rate": 5.489886635329598e-05,
      "loss": 0.9417,
      "step": 401
    },
    {
      "epoch": 2.2042494859492803,
      "grad_norm": 0.9164307400605365,
      "learning_rate": 5.4756362226697193e-05,
      "loss": 0.9449,
      "step": 402
    },
    {
      "epoch": 2.209732693625771,
      "grad_norm": 1.3336341868468473,
      "learning_rate": 5.461364097465581e-05,
      "loss": 0.9758,
      "step": 403
    },
    {
      "epoch": 2.215215901302262,
      "grad_norm": 0.5707410680573715,
      "learning_rate": 5.447070469717552e-05,
      "loss": 0.9509,
      "step": 404
    },
    {
      "epoch": 2.2206991089787524,
      "grad_norm": 0.5349427819988657,
      "learning_rate": 5.4327555497423874e-05,
      "loss": 0.977,
      "step": 405
    },
    {
      "epoch": 2.2261823166552435,
      "grad_norm": 0.6665137379275449,
      "learning_rate": 5.4184195481701425e-05,
      "loss": 0.9718,
      "step": 406
    },
    {
      "epoch": 2.231665524331734,
      "grad_norm": 0.6788016902611529,
      "learning_rate": 5.4040626759410625e-05,
      "loss": 0.9559,
      "step": 407
    },
    {
      "epoch": 2.2371487320082246,
      "grad_norm": 0.7062646461019303,
      "learning_rate": 5.3896851443024837e-05,
      "loss": 0.9663,
      "step": 408
    },
    {
      "epoch": 2.2426319396847156,
      "grad_norm": 0.6871839246478715,
      "learning_rate": 5.375287164805727e-05,
      "loss": 0.9814,
      "step": 409
    },
    {
      "epoch": 2.248115147361206,
      "grad_norm": 0.5968021041080723,
      "learning_rate": 5.360868949302986e-05,
      "loss": 0.937,
      "step": 410
    },
    {
      "epoch": 2.2535983550376972,
      "grad_norm": 0.5805892891972629,
      "learning_rate": 5.3464307099442035e-05,
      "loss": 0.9479,
      "step": 411
    },
    {
      "epoch": 2.259081562714188,
      "grad_norm": 0.4860199686870337,
      "learning_rate": 5.3319726591739536e-05,
      "loss": 0.9541,
      "step": 412
    },
    {
      "epoch": 2.2645647703906784,
      "grad_norm": 0.5697676557756951,
      "learning_rate": 5.317495009728319e-05,
      "loss": 0.9471,
      "step": 413
    },
    {
      "epoch": 2.2700479780671694,
      "grad_norm": 0.3474403252554382,
      "learning_rate": 5.302997974631757e-05,
      "loss": 0.9535,
      "step": 414
    },
    {
      "epoch": 2.27553118574366,
      "grad_norm": 0.3892967631879122,
      "learning_rate": 5.288481767193963e-05,
      "loss": 0.9573,
      "step": 415
    },
    {
      "epoch": 2.2810143934201506,
      "grad_norm": 0.43748961608691317,
      "learning_rate": 5.2739466010067385e-05,
      "loss": 0.9377,
      "step": 416
    },
    {
      "epoch": 2.2864976010966416,
      "grad_norm": 0.410672984619657,
      "learning_rate": 5.259392689940841e-05,
      "loss": 0.9313,
      "step": 417
    },
    {
      "epoch": 2.291980808773132,
      "grad_norm": 0.44449306463157623,
      "learning_rate": 5.244820248142844e-05,
      "loss": 0.9373,
      "step": 418
    },
    {
      "epoch": 2.297464016449623,
      "grad_norm": 0.457420584226533,
      "learning_rate": 5.2302294900319796e-05,
      "loss": 0.9627,
      "step": 419
    },
    {
      "epoch": 2.3029472241261137,
      "grad_norm": 0.49606181089497686,
      "learning_rate": 5.215620630296988e-05,
      "loss": 0.9491,
      "step": 420
    },
    {
      "epoch": 2.3084304318026048,
      "grad_norm": 0.5557714109621562,
      "learning_rate": 5.200993883892956e-05,
      "loss": 0.9538,
      "step": 421
    },
    {
      "epoch": 2.3139136394790953,
      "grad_norm": 0.5848899773483123,
      "learning_rate": 5.1863494660381586e-05,
      "loss": 0.9589,
      "step": 422
    },
    {
      "epoch": 2.319396847155586,
      "grad_norm": 0.5756814498770187,
      "learning_rate": 5.1716875922108836e-05,
      "loss": 0.9437,
      "step": 423
    },
    {
      "epoch": 2.324880054832077,
      "grad_norm": 0.5975395685498862,
      "learning_rate": 5.1570084781462716e-05,
      "loss": 0.9752,
      "step": 424
    },
    {
      "epoch": 2.3303632625085675,
      "grad_norm": 0.6741079069902028,
      "learning_rate": 5.142312339833131e-05,
      "loss": 0.9695,
      "step": 425
    },
    {
      "epoch": 2.335846470185058,
      "grad_norm": 0.47322117724121776,
      "learning_rate": 5.1275993935107714e-05,
      "loss": 0.9376,
      "step": 426
    },
    {
      "epoch": 2.341329677861549,
      "grad_norm": 0.3713624782507764,
      "learning_rate": 5.112869855665811e-05,
      "loss": 0.9587,
      "step": 427
    },
    {
      "epoch": 2.3468128855380397,
      "grad_norm": 0.3297040646564409,
      "learning_rate": 5.098123943028999e-05,
      "loss": 0.95,
      "step": 428
    },
    {
      "epoch": 2.3522960932145303,
      "grad_norm": 0.3283051265027244,
      "learning_rate": 5.0833618725720214e-05,
      "loss": 0.9585,
      "step": 429
    },
    {
      "epoch": 2.3577793008910213,
      "grad_norm": 0.33355324970835326,
      "learning_rate": 5.0685838615043124e-05,
      "loss": 0.9373,
      "step": 430
    },
    {
      "epoch": 2.363262508567512,
      "grad_norm": 0.32077316347401524,
      "learning_rate": 5.053790127269855e-05,
      "loss": 0.9281,
      "step": 431
    },
    {
      "epoch": 2.368745716244003,
      "grad_norm": 0.29194427533135364,
      "learning_rate": 5.038980887543987e-05,
      "loss": 0.9565,
      "step": 432
    },
    {
      "epoch": 2.3742289239204935,
      "grad_norm": 0.2613277032435807,
      "learning_rate": 5.024156360230189e-05,
      "loss": 0.9654,
      "step": 433
    },
    {
      "epoch": 2.3797121315969845,
      "grad_norm": 0.30838651942670436,
      "learning_rate": 5.0093167634568874e-05,
      "loss": 0.9555,
      "step": 434
    },
    {
      "epoch": 2.385195339273475,
      "grad_norm": 0.3084305013600498,
      "learning_rate": 4.9944623155742395e-05,
      "loss": 0.9301,
      "step": 435
    },
    {
      "epoch": 2.3906785469499656,
      "grad_norm": 0.3387529105331699,
      "learning_rate": 4.979593235150924e-05,
      "loss": 0.9428,
      "step": 436
    },
    {
      "epoch": 2.3961617546264566,
      "grad_norm": 0.3995434200366217,
      "learning_rate": 4.9647097409709186e-05,
      "loss": 0.9405,
      "step": 437
    },
    {
      "epoch": 2.401644962302947,
      "grad_norm": 0.41290791299928126,
      "learning_rate": 4.94981205203029e-05,
      "loss": 0.9483,
      "step": 438
    },
    {
      "epoch": 2.407128169979438,
      "grad_norm": 0.4272879341834311,
      "learning_rate": 4.934900387533965e-05,
      "loss": 0.9434,
      "step": 439
    },
    {
      "epoch": 2.412611377655929,
      "grad_norm": 0.46378214773916265,
      "learning_rate": 4.9199749668925076e-05,
      "loss": 0.9551,
      "step": 440
    },
    {
      "epoch": 2.4180945853324194,
      "grad_norm": 0.46144056548958406,
      "learning_rate": 4.9050360097188904e-05,
      "loss": 0.952,
      "step": 441
    },
    {
      "epoch": 2.42357779300891,
      "grad_norm": 0.4496399315688108,
      "learning_rate": 4.890083735825258e-05,
      "loss": 0.9414,
      "step": 442
    },
    {
      "epoch": 2.429061000685401,
      "grad_norm": 0.5404123880896423,
      "learning_rate": 4.875118365219706e-05,
      "loss": 0.976,
      "step": 443
    },
    {
      "epoch": 2.4345442083618916,
      "grad_norm": 0.6694234081039024,
      "learning_rate": 4.86014011810303e-05,
      "loss": 0.938,
      "step": 444
    },
    {
      "epoch": 2.4400274160383826,
      "grad_norm": 0.7976721444374203,
      "learning_rate": 4.845149214865491e-05,
      "loss": 0.9428,
      "step": 445
    },
    {
      "epoch": 2.445510623714873,
      "grad_norm": 0.8514351115579054,
      "learning_rate": 4.830145876083575e-05,
      "loss": 0.9716,
      "step": 446
    },
    {
      "epoch": 2.450993831391364,
      "grad_norm": 0.8419233920727014,
      "learning_rate": 4.81513032251674e-05,
      "loss": 0.9375,
      "step": 447
    },
    {
      "epoch": 2.4564770390678548,
      "grad_norm": 0.7939704221285054,
      "learning_rate": 4.8001027751041784e-05,
      "loss": 0.9539,
      "step": 448
    },
    {
      "epoch": 2.4619602467443453,
      "grad_norm": 0.6506959170383838,
      "learning_rate": 4.785063454961557e-05,
      "loss": 0.974,
      "step": 449
    },
    {
      "epoch": 2.4674434544208363,
      "grad_norm": 0.4039955860724156,
      "learning_rate": 4.7700125833777664e-05,
      "loss": 0.9409,
      "step": 450
    },
    {
      "epoch": 2.472926662097327,
      "grad_norm": 0.33076049398826807,
      "learning_rate": 4.754950381811667e-05,
      "loss": 0.9561,
      "step": 451
    },
    {
      "epoch": 2.4784098697738175,
      "grad_norm": 0.4841447525816379,
      "learning_rate": 4.7398770718888296e-05,
      "loss": 0.9557,
      "step": 452
    },
    {
      "epoch": 2.4838930774503085,
      "grad_norm": 0.6029861941124044,
      "learning_rate": 4.724792875398271e-05,
      "loss": 0.9254,
      "step": 453
    },
    {
      "epoch": 2.489376285126799,
      "grad_norm": 0.6018267599600162,
      "learning_rate": 4.7096980142891936e-05,
      "loss": 0.9336,
      "step": 454
    },
    {
      "epoch": 2.49485949280329,
      "grad_norm": 0.4547177146031252,
      "learning_rate": 4.694592710667723e-05,
      "loss": 0.941,
      "step": 455
    },
    {
      "epoch": 2.5003427004797807,
      "grad_norm": 0.2560704544963682,
      "learning_rate": 4.6794771867936286e-05,
      "loss": 0.9413,
      "step": 456
    },
    {
      "epoch": 2.5058259081562713,
      "grad_norm": 0.30599482861554783,
      "learning_rate": 4.66435166507707e-05,
      "loss": 0.9605,
      "step": 457
    },
    {
      "epoch": 2.5113091158327623,
      "grad_norm": 0.42429824249004056,
      "learning_rate": 4.6492163680753096e-05,
      "loss": 0.9403,
      "step": 458
    },
    {
      "epoch": 2.516792323509253,
      "grad_norm": 0.476594932388952,
      "learning_rate": 4.634071518489443e-05,
      "loss": 0.9354,
      "step": 459
    },
    {
      "epoch": 2.522275531185744,
      "grad_norm": 0.46789821860796127,
      "learning_rate": 4.618917339161125e-05,
      "loss": 0.9495,
      "step": 460
    },
    {
      "epoch": 2.5277587388622345,
      "grad_norm": 0.3784567986912495,
      "learning_rate": 4.6037540530692905e-05,
      "loss": 0.958,
      "step": 461
    },
    {
      "epoch": 2.533241946538725,
      "grad_norm": 0.3589933646618808,
      "learning_rate": 4.588581883326865e-05,
      "loss": 0.9449,
      "step": 462
    },
    {
      "epoch": 2.538725154215216,
      "grad_norm": 0.4378215691524641,
      "learning_rate": 4.573401053177494e-05,
      "loss": 0.949,
      "step": 463
    },
    {
      "epoch": 2.5442083618917066,
      "grad_norm": 0.4673331119883799,
      "learning_rate": 4.558211785992251e-05,
      "loss": 0.9555,
      "step": 464
    },
    {
      "epoch": 2.549691569568197,
      "grad_norm": 0.4792027471063911,
      "learning_rate": 4.543014305266352e-05,
      "loss": 0.981,
      "step": 465
    },
    {
      "epoch": 2.5551747772446882,
      "grad_norm": 0.4774897666673642,
      "learning_rate": 4.5278088346158665e-05,
      "loss": 0.9456,
      "step": 466
    },
    {
      "epoch": 2.560657984921179,
      "grad_norm": 0.4425042364504943,
      "learning_rate": 4.512595597774427e-05,
      "loss": 0.9429,
      "step": 467
    },
    {
      "epoch": 2.5661411925976694,
      "grad_norm": 0.39911527326216606,
      "learning_rate": 4.4973748185899416e-05,
      "loss": 0.9413,
      "step": 468
    },
    {
      "epoch": 2.5716244002741604,
      "grad_norm": 0.29210171265761314,
      "learning_rate": 4.4821467210212924e-05,
      "loss": 0.9297,
      "step": 469
    },
    {
      "epoch": 2.577107607950651,
      "grad_norm": 0.25565384930514384,
      "learning_rate": 4.4669115291350484e-05,
      "loss": 0.9454,
      "step": 470
    },
    {
      "epoch": 2.582590815627142,
      "grad_norm": 0.2581461861152204,
      "learning_rate": 4.451669467102162e-05,
      "loss": 0.948,
      "step": 471
    },
    {
      "epoch": 2.5880740233036326,
      "grad_norm": 0.29491880993583836,
      "learning_rate": 4.436420759194671e-05,
      "loss": 0.9381,
      "step": 472
    },
    {
      "epoch": 2.5935572309801236,
      "grad_norm": 0.31658803555973,
      "learning_rate": 4.4211656297824064e-05,
      "loss": 0.9618,
      "step": 473
    },
    {
      "epoch": 2.599040438656614,
      "grad_norm": 0.3245542687115596,
      "learning_rate": 4.4059043033296815e-05,
      "loss": 0.9488,
      "step": 474
    },
    {
      "epoch": 2.6045236463331047,
      "grad_norm": 0.2905774382505926,
      "learning_rate": 4.390637004391993e-05,
      "loss": 0.9323,
      "step": 475
    },
    {
      "epoch": 2.6100068540095958,
      "grad_norm": 0.2428082128722324,
      "learning_rate": 4.375363957612717e-05,
      "loss": 0.9549,
      "step": 476
    },
    {
      "epoch": 2.6154900616860863,
      "grad_norm": 0.24550826505798423,
      "learning_rate": 4.360085387719806e-05,
      "loss": 0.9493,
      "step": 477
    },
    {
      "epoch": 2.620973269362577,
      "grad_norm": 0.30222793473570203,
      "learning_rate": 4.344801519522478e-05,
      "loss": 0.9317,
      "step": 478
    },
    {
      "epoch": 2.626456477039068,
      "grad_norm": 0.3497976704406349,
      "learning_rate": 4.32951257790791e-05,
      "loss": 0.9377,
      "step": 479
    },
    {
      "epoch": 2.6319396847155585,
      "grad_norm": 1.2304346438090321,
      "learning_rate": 4.314218787837925e-05,
      "loss": 0.9442,
      "step": 480
    },
    {
      "epoch": 2.637422892392049,
      "grad_norm": 0.2466343919140661,
      "learning_rate": 4.298920374345698e-05,
      "loss": 0.944,
      "step": 481
    },
    {
      "epoch": 2.64290610006854,
      "grad_norm": 1.8185754976050628,
      "learning_rate": 4.283617562532421e-05,
      "loss": 0.9825,
      "step": 482
    },
    {
      "epoch": 2.648389307745031,
      "grad_norm": 0.5797415336422694,
      "learning_rate": 4.2683105775640096e-05,
      "loss": 0.9565,
      "step": 483
    },
    {
      "epoch": 2.6538725154215217,
      "grad_norm": 0.6585844779419097,
      "learning_rate": 4.2529996446677814e-05,
      "loss": 0.9349,
      "step": 484
    },
    {
      "epoch": 2.6593557230980123,
      "grad_norm": 0.6364339320143122,
      "learning_rate": 4.237684989129146e-05,
      "loss": 0.9407,
      "step": 485
    },
    {
      "epoch": 2.6648389307745033,
      "grad_norm": 0.7003736753475072,
      "learning_rate": 4.2223668362882846e-05,
      "loss": 0.952,
      "step": 486
    },
    {
      "epoch": 2.670322138450994,
      "grad_norm": 0.6868570924103546,
      "learning_rate": 4.2070454115368385e-05,
      "loss": 0.9465,
      "step": 487
    },
    {
      "epoch": 2.6758053461274844,
      "grad_norm": 0.8377123241663084,
      "learning_rate": 4.191720940314593e-05,
      "loss": 0.9365,
      "step": 488
    },
    {
      "epoch": 2.6812885538039755,
      "grad_norm": 1.6613962498279427,
      "learning_rate": 4.176393648106161e-05,
      "loss": 0.9592,
      "step": 489
    },
    {
      "epoch": 2.686771761480466,
      "grad_norm": 0.9696773125080363,
      "learning_rate": 4.1610637604376614e-05,
      "loss": 0.9526,
      "step": 490
    },
    {
      "epoch": 2.6922549691569566,
      "grad_norm": 1.0222168588069447,
      "learning_rate": 4.1457315028734015e-05,
      "loss": 0.9736,
      "step": 491
    },
    {
      "epoch": 2.6977381768334476,
      "grad_norm": 0.9553354559606164,
      "learning_rate": 4.13039710101256e-05,
      "loss": 0.9668,
      "step": 492
    },
    {
      "epoch": 2.703221384509938,
      "grad_norm": 0.8745999789478393,
      "learning_rate": 4.11506078048587e-05,
      "loss": 0.9716,
      "step": 493
    },
    {
      "epoch": 2.7087045921864292,
      "grad_norm": 0.7411939891535938,
      "learning_rate": 4.0997227669522924e-05,
      "loss": 0.9602,
      "step": 494
    },
    {
      "epoch": 2.71418779986292,
      "grad_norm": 0.6053293148082218,
      "learning_rate": 4.0843832860956994e-05,
      "loss": 0.9321,
      "step": 495
    },
    {
      "epoch": 2.719671007539411,
      "grad_norm": 0.6015174283144377,
      "learning_rate": 4.069042563621555e-05,
      "loss": 0.9556,
      "step": 496
    },
    {
      "epoch": 2.7251542152159014,
      "grad_norm": 0.5341988781830062,
      "learning_rate": 4.0537008252535904e-05,
      "loss": 0.9502,
      "step": 497
    },
    {
      "epoch": 2.730637422892392,
      "grad_norm": 0.3986261784888644,
      "learning_rate": 4.0383582967304865e-05,
      "loss": 0.9498,
      "step": 498
    },
    {
      "epoch": 2.736120630568883,
      "grad_norm": 0.4491497759193806,
      "learning_rate": 4.023015203802551e-05,
      "loss": 0.9578,
      "step": 499
    },
    {
      "epoch": 2.7416038382453736,
      "grad_norm": 0.42597854684305547,
      "learning_rate": 4.0076717722283936e-05,
      "loss": 0.9258,
      "step": 500
    },
    {
      "epoch": 2.747087045921864,
      "grad_norm": 0.3674427892141133,
      "learning_rate": 3.992328227771608e-05,
      "loss": 0.956,
      "step": 501
    },
    {
      "epoch": 2.752570253598355,
      "grad_norm": 0.3441073888754271,
      "learning_rate": 3.976984796197451e-05,
      "loss": 0.9347,
      "step": 502
    },
    {
      "epoch": 2.7580534612748457,
      "grad_norm": 0.5779083792402964,
      "learning_rate": 3.961641703269514e-05,
      "loss": 0.9591,
      "step": 503
    },
    {
      "epoch": 2.7635366689513363,
      "grad_norm": 0.45842637741331516,
      "learning_rate": 3.946299174746411e-05,
      "loss": 0.9497,
      "step": 504
    },
    {
      "epoch": 2.7690198766278273,
      "grad_norm": 0.42003328606156265,
      "learning_rate": 3.9309574363784465e-05,
      "loss": 0.9365,
      "step": 505
    },
    {
      "epoch": 2.774503084304318,
      "grad_norm": 0.3971675276536041,
      "learning_rate": 3.915616713904302e-05,
      "loss": 0.9532,
      "step": 506
    },
    {
      "epoch": 2.779986291980809,
      "grad_norm": 0.3738039726781753,
      "learning_rate": 3.9002772330477096e-05,
      "loss": 0.9505,
      "step": 507
    },
    {
      "epoch": 2.7854694996572995,
      "grad_norm": 0.3490205849659376,
      "learning_rate": 3.884939219514132e-05,
      "loss": 0.9383,
      "step": 508
    },
    {
      "epoch": 2.7909527073337905,
      "grad_norm": 0.6333313596692628,
      "learning_rate": 3.869602898987441e-05,
      "loss": 0.9528,
      "step": 509
    },
    {
      "epoch": 2.796435915010281,
      "grad_norm": 0.3242106302112962,
      "learning_rate": 3.854268497126601e-05,
      "loss": 0.9554,
      "step": 510
    },
    {
      "epoch": 2.8019191226867717,
      "grad_norm": 0.2849035104268222,
      "learning_rate": 3.8389362395623406e-05,
      "loss": 0.9477,
      "step": 511
    },
    {
      "epoch": 2.8074023303632627,
      "grad_norm": 0.23339126446276168,
      "learning_rate": 3.8236063518938405e-05,
      "loss": 0.9401,
      "step": 512
    },
    {
      "epoch": 2.8128855380397533,
      "grad_norm": 0.2658559032559636,
      "learning_rate": 3.8082790596854075e-05,
      "loss": 0.9322,
      "step": 513
    },
    {
      "epoch": 2.818368745716244,
      "grad_norm": 0.28632019300770967,
      "learning_rate": 3.792954588463162e-05,
      "loss": 0.9647,
      "step": 514
    },
    {
      "epoch": 2.823851953392735,
      "grad_norm": 0.2585763044971737,
      "learning_rate": 3.777633163711716e-05,
      "loss": 0.9505,
      "step": 515
    },
    {
      "epoch": 2.8293351610692254,
      "grad_norm": 0.2742118891600389,
      "learning_rate": 3.7623150108708546e-05,
      "loss": 0.9446,
      "step": 516
    },
    {
      "epoch": 2.834818368745716,
      "grad_norm": 0.2344118819440086,
      "learning_rate": 3.7470003553322186e-05,
      "loss": 0.9309,
      "step": 517
    },
    {
      "epoch": 2.840301576422207,
      "grad_norm": 0.5837521053187771,
      "learning_rate": 3.7316894224359904e-05,
      "loss": 0.9394,
      "step": 518
    },
    {
      "epoch": 2.8457847840986976,
      "grad_norm": 0.31498022974880685,
      "learning_rate": 3.71638243746758e-05,
      "loss": 0.9321,
      "step": 519
    },
    {
      "epoch": 2.8512679917751886,
      "grad_norm": 0.5111390953286222,
      "learning_rate": 3.7010796256543034e-05,
      "loss": 0.9565,
      "step": 520
    },
    {
      "epoch": 2.856751199451679,
      "grad_norm": 0.24452242688766815,
      "learning_rate": 3.6857812121620756e-05,
      "loss": 0.938,
      "step": 521
    },
    {
      "epoch": 2.8622344071281702,
      "grad_norm": 0.263705639317996,
      "learning_rate": 3.670487422092092e-05,
      "loss": 0.9434,
      "step": 522
    },
    {
      "epoch": 2.867717614804661,
      "grad_norm": 0.25074905428863853,
      "learning_rate": 3.655198480477523e-05,
      "loss": 0.9514,
      "step": 523
    },
    {
      "epoch": 2.8732008224811514,
      "grad_norm": 0.23134392999771497,
      "learning_rate": 3.639914612280194e-05,
      "loss": 0.9437,
      "step": 524
    },
    {
      "epoch": 2.8786840301576424,
      "grad_norm": 0.34358764407986175,
      "learning_rate": 3.6246360423872834e-05,
      "loss": 0.9518,
      "step": 525
    },
    {
      "epoch": 2.884167237834133,
      "grad_norm": 0.21629316436514154,
      "learning_rate": 3.609362995608008e-05,
      "loss": 0.9572,
      "step": 526
    },
    {
      "epoch": 2.8896504455106236,
      "grad_norm": 0.22317832298096357,
      "learning_rate": 3.59409569667032e-05,
      "loss": 0.9492,
      "step": 527
    },
    {
      "epoch": 2.8951336531871146,
      "grad_norm": 0.20759859863486543,
      "learning_rate": 3.578834370217595e-05,
      "loss": 0.9355,
      "step": 528
    },
    {
      "epoch": 2.900616860863605,
      "grad_norm": 0.2138741670103374,
      "learning_rate": 3.5635792408053304e-05,
      "loss": 0.9526,
      "step": 529
    },
    {
      "epoch": 2.9061000685400957,
      "grad_norm": 0.2071241548771815,
      "learning_rate": 3.54833053289784e-05,
      "loss": 0.9459,
      "step": 530
    },
    {
      "epoch": 2.9115832762165867,
      "grad_norm": 0.24989924945583794,
      "learning_rate": 3.533088470864953e-05,
      "loss": 0.9402,
      "step": 531
    },
    {
      "epoch": 2.9170664838930773,
      "grad_norm": 0.27509189913976284,
      "learning_rate": 3.517853278978708e-05,
      "loss": 0.9554,
      "step": 532
    },
    {
      "epoch": 2.9225496915695683,
      "grad_norm": 0.23063790000408171,
      "learning_rate": 3.5026251814100604e-05,
      "loss": 0.9457,
      "step": 533
    },
    {
      "epoch": 2.928032899246059,
      "grad_norm": 0.2166555423682174,
      "learning_rate": 3.487404402225574e-05,
      "loss": 0.9272,
      "step": 534
    },
    {
      "epoch": 2.93351610692255,
      "grad_norm": 0.21899933270897742,
      "learning_rate": 3.4721911653841355e-05,
      "loss": 0.9507,
      "step": 535
    },
    {
      "epoch": 2.9389993145990405,
      "grad_norm": 0.19354444875830706,
      "learning_rate": 3.45698569473365e-05,
      "loss": 0.9682,
      "step": 536
    },
    {
      "epoch": 2.944482522275531,
      "grad_norm": 0.17844494801443178,
      "learning_rate": 3.44178821400775e-05,
      "loss": 0.9473,
      "step": 537
    },
    {
      "epoch": 2.949965729952022,
      "grad_norm": 1.3083812330864948,
      "learning_rate": 3.426598946822507e-05,
      "loss": 0.9461,
      "step": 538
    },
    {
      "epoch": 2.9554489376285127,
      "grad_norm": 0.21349382766535052,
      "learning_rate": 3.4114181166731355e-05,
      "loss": 0.9238,
      "step": 539
    },
    {
      "epoch": 2.9609321453050033,
      "grad_norm": 0.2740853283765663,
      "learning_rate": 3.39624594693071e-05,
      "loss": 0.9483,
      "step": 540
    },
    {
      "epoch": 2.9664153529814943,
      "grad_norm": 0.2604547593626456,
      "learning_rate": 3.381082660838875e-05,
      "loss": 0.9566,
      "step": 541
    },
    {
      "epoch": 2.971898560657985,
      "grad_norm": 0.2636302620609088,
      "learning_rate": 3.365928481510558e-05,
      "loss": 0.9529,
      "step": 542
    },
    {
      "epoch": 2.9773817683344754,
      "grad_norm": 0.24591140742789205,
      "learning_rate": 3.3507836319246924e-05,
      "loss": 0.9327,
      "step": 543
    },
    {
      "epoch": 2.9828649760109665,
      "grad_norm": 0.23336910029530064,
      "learning_rate": 3.33564833492293e-05,
      "loss": 0.9606,
      "step": 544
    },
    {
      "epoch": 2.988348183687457,
      "grad_norm": 0.215970712690598,
      "learning_rate": 3.3205228132063714e-05,
      "loss": 0.9495,
      "step": 545
    },
    {
      "epoch": 2.993831391363948,
      "grad_norm": 0.23530537682003938,
      "learning_rate": 3.305407289332279e-05,
      "loss": 0.9533,
      "step": 546
    },
    {
      "epoch": 2.9993145990404386,
      "grad_norm": 0.35138852092483364,
      "learning_rate": 3.290301985710807e-05,
      "loss": 1.3456,
      "step": 547
    },
    {
      "epoch": 3.004797806716929,
      "grad_norm": 0.5481962224464548,
      "learning_rate": 3.27520712460173e-05,
      "loss": 1.1027,
      "step": 548
    },
    {
      "epoch": 3.01028101439342,
      "grad_norm": 0.7662140587544488,
      "learning_rate": 3.260122928111172e-05,
      "loss": 0.8979,
      "step": 549
    },
    {
      "epoch": 3.015764222069911,
      "grad_norm": 0.9454349543994065,
      "learning_rate": 3.245049618188334e-05,
      "loss": 0.9039,
      "step": 550
    },
    {
      "epoch": 3.021247429746402,
      "grad_norm": 0.7559182562175768,
      "learning_rate": 3.229987416622235e-05,
      "loss": 0.9046,
      "step": 551
    },
    {
      "epoch": 3.0267306374228924,
      "grad_norm": 0.6019789035769177,
      "learning_rate": 3.2149365450384445e-05,
      "loss": 0.8982,
      "step": 552
    },
    {
      "epoch": 3.032213845099383,
      "grad_norm": 0.608624397941904,
      "learning_rate": 3.199897224895823e-05,
      "loss": 0.9024,
      "step": 553
    },
    {
      "epoch": 3.037697052775874,
      "grad_norm": 1.3331460907440373,
      "learning_rate": 3.184869677483261e-05,
      "loss": 0.9183,
      "step": 554
    },
    {
      "epoch": 3.0431802604523646,
      "grad_norm": 0.5617254798828216,
      "learning_rate": 3.169854123916426e-05,
      "loss": 0.9044,
      "step": 555
    },
    {
      "epoch": 3.0486634681288556,
      "grad_norm": 0.6117567607126662,
      "learning_rate": 3.1548507851345094e-05,
      "loss": 0.9084,
      "step": 556
    },
    {
      "epoch": 3.054146675805346,
      "grad_norm": 0.5486288335887587,
      "learning_rate": 3.139859881896971e-05,
      "loss": 0.9086,
      "step": 557
    },
    {
      "epoch": 3.0596298834818367,
      "grad_norm": 0.43883951251804787,
      "learning_rate": 3.124881634780295e-05,
      "loss": 0.8822,
      "step": 558
    },
    {
      "epoch": 3.0651130911583278,
      "grad_norm": 0.375243221483072,
      "learning_rate": 3.109916264174743e-05,
      "loss": 0.9089,
      "step": 559
    },
    {
      "epoch": 3.0705962988348183,
      "grad_norm": 0.45192811672936717,
      "learning_rate": 3.094963990281112e-05,
      "loss": 0.8979,
      "step": 560
    },
    {
      "epoch": 3.076079506511309,
      "grad_norm": 0.3691771282212795,
      "learning_rate": 3.080025033107494e-05,
      "loss": 0.8973,
      "step": 561
    },
    {
      "epoch": 3.0815627141878,
      "grad_norm": 0.28371809189511954,
      "learning_rate": 3.065099612466037e-05,
      "loss": 0.887,
      "step": 562
    },
    {
      "epoch": 3.0870459218642905,
      "grad_norm": 0.33397747859909344,
      "learning_rate": 3.0501879479697112e-05,
      "loss": 0.9324,
      "step": 563
    },
    {
      "epoch": 3.0925291295407815,
      "grad_norm": 8.830372813312158,
      "learning_rate": 3.035290259029083e-05,
      "loss": 0.9137,
      "step": 564
    },
    {
      "epoch": 3.098012337217272,
      "grad_norm": 0.6793506372960832,
      "learning_rate": 3.0204067648490766e-05,
      "loss": 0.9178,
      "step": 565
    },
    {
      "epoch": 3.1034955448937627,
      "grad_norm": 0.8864777488999107,
      "learning_rate": 3.00553768442576e-05,
      "loss": 0.8926,
      "step": 566
    },
    {
      "epoch": 3.1089787525702537,
      "grad_norm": 0.6288925135464869,
      "learning_rate": 2.9906832365431132e-05,
      "loss": 0.9198,
      "step": 567
    },
    {
      "epoch": 3.1144619602467443,
      "grad_norm": 0.40885944277256037,
      "learning_rate": 2.9758436397698118e-05,
      "loss": 0.9127,
      "step": 568
    },
    {
      "epoch": 3.1199451679232353,
      "grad_norm": 0.7284002718913494,
      "learning_rate": 2.961019112456014e-05,
      "loss": 0.882,
      "step": 569
    },
    {
      "epoch": 3.125428375599726,
      "grad_norm": 0.5837455202754323,
      "learning_rate": 2.946209872730145e-05,
      "loss": 0.9078,
      "step": 570
    },
    {
      "epoch": 3.1309115832762164,
      "grad_norm": 0.33186964600527347,
      "learning_rate": 2.931416138495689e-05,
      "loss": 0.8925,
      "step": 571
    },
    {
      "epoch": 3.1363947909527075,
      "grad_norm": 0.53882269291948,
      "learning_rate": 2.9166381274279803e-05,
      "loss": 0.9042,
      "step": 572
    },
    {
      "epoch": 3.141877998629198,
      "grad_norm": 0.46331612230401065,
      "learning_rate": 2.901876056971002e-05,
      "loss": 0.9217,
      "step": 573
    },
    {
      "epoch": 3.147361206305689,
      "grad_norm": 0.31771843320448484,
      "learning_rate": 2.88713014433419e-05,
      "loss": 0.9018,
      "step": 574
    },
    {
      "epoch": 3.1528444139821796,
      "grad_norm": 0.49472378335259276,
      "learning_rate": 2.8724006064892296e-05,
      "loss": 0.8748,
      "step": 575
    },
    {
      "epoch": 3.15832762165867,
      "grad_norm": 0.3530540111863458,
      "learning_rate": 2.85768766016687e-05,
      "loss": 0.8957,
      "step": 576
    },
    {
      "epoch": 3.1638108293351612,
      "grad_norm": 0.3023194345015207,
      "learning_rate": 2.8429915218537297e-05,
      "loss": 0.9023,
      "step": 577
    },
    {
      "epoch": 3.169294037011652,
      "grad_norm": 0.39598120224098793,
      "learning_rate": 2.8283124077891167e-05,
      "loss": 0.8928,
      "step": 578
    },
    {
      "epoch": 3.1747772446881424,
      "grad_norm": 0.2762440885300842,
      "learning_rate": 2.813650533961843e-05,
      "loss": 0.8986,
      "step": 579
    },
    {
      "epoch": 3.1802604523646334,
      "grad_norm": 0.2731980966351029,
      "learning_rate": 2.7990061161070445e-05,
      "loss": 0.9012,
      "step": 580
    },
    {
      "epoch": 3.185743660041124,
      "grad_norm": 0.41348684607047675,
      "learning_rate": 2.7843793697030128e-05,
      "loss": 0.9069,
      "step": 581
    },
    {
      "epoch": 3.191226867717615,
      "grad_norm": 0.25162826940534694,
      "learning_rate": 2.7697705099680217e-05,
      "loss": 0.9061,
      "step": 582
    },
    {
      "epoch": 3.1967100753941056,
      "grad_norm": 0.2506514671680193,
      "learning_rate": 2.7551797518571573e-05,
      "loss": 0.8851,
      "step": 583
    },
    {
      "epoch": 3.202193283070596,
      "grad_norm": 0.2659325922311813,
      "learning_rate": 2.7406073100591605e-05,
      "loss": 0.8999,
      "step": 584
    },
    {
      "epoch": 3.207676490747087,
      "grad_norm": 0.22833189604550308,
      "learning_rate": 2.7260533989932628e-05,
      "loss": 0.8804,
      "step": 585
    },
    {
      "epoch": 3.2131596984235777,
      "grad_norm": 0.26002377936612986,
      "learning_rate": 2.7115182328060385e-05,
      "loss": 0.8987,
      "step": 586
    },
    {
      "epoch": 3.2186429061000688,
      "grad_norm": 0.24416881973039506,
      "learning_rate": 2.697002025368245e-05,
      "loss": 0.8982,
      "step": 587
    },
    {
      "epoch": 3.2241261137765593,
      "grad_norm": 0.22174095398609206,
      "learning_rate": 2.682504990271682e-05,
      "loss": 0.8962,
      "step": 588
    },
    {
      "epoch": 3.22960932145305,
      "grad_norm": 0.2715414538357121,
      "learning_rate": 2.668027340826048e-05,
      "loss": 0.8896,
      "step": 589
    },
    {
      "epoch": 3.235092529129541,
      "grad_norm": 0.2765507918752038,
      "learning_rate": 2.653569290055799e-05,
      "loss": 0.8938,
      "step": 590
    },
    {
      "epoch": 3.2405757368060315,
      "grad_norm": 0.20005217231999928,
      "learning_rate": 2.6391310506970147e-05,
      "loss": 0.8827,
      "step": 591
    },
    {
      "epoch": 3.246058944482522,
      "grad_norm": 0.8704396300357945,
      "learning_rate": 2.6247128351942726e-05,
      "loss": 0.9075,
      "step": 592
    },
    {
      "epoch": 3.251542152159013,
      "grad_norm": 0.3586781432686027,
      "learning_rate": 2.6103148556975173e-05,
      "loss": 0.8919,
      "step": 593
    },
    {
      "epoch": 3.2570253598355037,
      "grad_norm": 0.3879500657373189,
      "learning_rate": 2.5959373240589382e-05,
      "loss": 0.906,
      "step": 594
    },
    {
      "epoch": 3.2625085675119947,
      "grad_norm": 0.3206624742458832,
      "learning_rate": 2.5815804518298575e-05,
      "loss": 0.9046,
      "step": 595
    },
    {
      "epoch": 3.2679917751884853,
      "grad_norm": 0.5655322768937269,
      "learning_rate": 2.5672444502576122e-05,
      "loss": 0.8792,
      "step": 596
    },
    {
      "epoch": 3.273474982864976,
      "grad_norm": 0.26423994402378853,
      "learning_rate": 2.55292953028245e-05,
      "loss": 0.9029,
      "step": 597
    },
    {
      "epoch": 3.278958190541467,
      "grad_norm": 0.3568680800314573,
      "learning_rate": 2.53863590253442e-05,
      "loss": 0.9097,
      "step": 598
    },
    {
      "epoch": 3.2844413982179574,
      "grad_norm": 0.2876917661506257,
      "learning_rate": 2.5243637773302817e-05,
      "loss": 0.9039,
      "step": 599
    },
    {
      "epoch": 3.2899246058944485,
      "grad_norm": 0.2543858700320745,
      "learning_rate": 2.510113364670403e-05,
      "loss": 0.8974,
      "step": 600
    },
    {
      "epoch": 3.295407813570939,
      "grad_norm": 0.29660460583308934,
      "learning_rate": 2.4958848742356724e-05,
      "loss": 0.9085,
      "step": 601
    },
    {
      "epoch": 3.3008910212474296,
      "grad_norm": 0.21363483198577174,
      "learning_rate": 2.481678515384415e-05,
      "loss": 0.9022,
      "step": 602
    },
    {
      "epoch": 3.3063742289239206,
      "grad_norm": 0.2406946488569973,
      "learning_rate": 2.4674944971493123e-05,
      "loss": 0.9017,
      "step": 603
    },
    {
      "epoch": 3.311857436600411,
      "grad_norm": 0.25257504290451166,
      "learning_rate": 2.453333028234325e-05,
      "loss": 0.913,
      "step": 604
    },
    {
      "epoch": 3.317340644276902,
      "grad_norm": 0.17923139676357774,
      "learning_rate": 2.439194317011622e-05,
      "loss": 0.9097,
      "step": 605
    },
    {
      "epoch": 3.322823851953393,
      "grad_norm": 0.2282134057524544,
      "learning_rate": 2.4250785715185138e-05,
      "loss": 0.8902,
      "step": 606
    },
    {
      "epoch": 3.3283070596298834,
      "grad_norm": 0.19939767203864542,
      "learning_rate": 2.410985999454396e-05,
      "loss": 0.8929,
      "step": 607
    },
    {
      "epoch": 3.3337902673063744,
      "grad_norm": 0.1962097138596388,
      "learning_rate": 2.3969168081776867e-05,
      "loss": 0.8816,
      "step": 608
    },
    {
      "epoch": 3.339273474982865,
      "grad_norm": 0.2408654729001506,
      "learning_rate": 2.382871204702781e-05,
      "loss": 0.8904,
      "step": 609
    },
    {
      "epoch": 3.3447566826593556,
      "grad_norm": 0.1857502171735821,
      "learning_rate": 2.3688493956969997e-05,
      "loss": 0.9253,
      "step": 610
    },
    {
      "epoch": 3.3502398903358466,
      "grad_norm": 0.1967645774870265,
      "learning_rate": 2.3548515874775547e-05,
      "loss": 0.8996,
      "step": 611
    },
    {
      "epoch": 3.355723098012337,
      "grad_norm": 0.19126170344434232,
      "learning_rate": 2.340877986008507e-05,
      "loss": 0.9036,
      "step": 612
    },
    {
      "epoch": 3.361206305688828,
      "grad_norm": 0.17261773740991576,
      "learning_rate": 2.3269287968977406e-05,
      "loss": 0.8974,
      "step": 613
    },
    {
      "epoch": 3.3666895133653187,
      "grad_norm": 0.1828461955951697,
      "learning_rate": 2.3130042253939334e-05,
      "loss": 0.9053,
      "step": 614
    },
    {
      "epoch": 3.3721727210418093,
      "grad_norm": 0.1787029837346015,
      "learning_rate": 2.2991044763835438e-05,
      "loss": 0.8796,
      "step": 615
    },
    {
      "epoch": 3.3776559287183003,
      "grad_norm": 0.1782795515564827,
      "learning_rate": 2.285229754387783e-05,
      "loss": 0.9047,
      "step": 616
    },
    {
      "epoch": 3.383139136394791,
      "grad_norm": 0.1808046435484797,
      "learning_rate": 2.2713802635596246e-05,
      "loss": 0.9014,
      "step": 617
    },
    {
      "epoch": 3.3886223440712815,
      "grad_norm": 0.16986006910284632,
      "learning_rate": 2.2575562076807857e-05,
      "loss": 0.9182,
      "step": 618
    },
    {
      "epoch": 3.3941055517477725,
      "grad_norm": 0.17065356826960065,
      "learning_rate": 2.2437577901587284e-05,
      "loss": 0.9148,
      "step": 619
    },
    {
      "epoch": 3.399588759424263,
      "grad_norm": 0.1821028903512155,
      "learning_rate": 2.22998521402368e-05,
      "loss": 0.892,
      "step": 620
    },
    {
      "epoch": 3.405071967100754,
      "grad_norm": 0.16756736884692283,
      "learning_rate": 2.216238681925628e-05,
      "loss": 0.9055,
      "step": 621
    },
    {
      "epoch": 3.4105551747772447,
      "grad_norm": 0.19298599105907382,
      "learning_rate": 2.2025183961313542e-05,
      "loss": 0.905,
      "step": 622
    },
    {
      "epoch": 3.4160383824537353,
      "grad_norm": 0.17599992995668898,
      "learning_rate": 2.188824558521452e-05,
      "loss": 0.9076,
      "step": 623
    },
    {
      "epoch": 3.4215215901302263,
      "grad_norm": 0.17170320698537844,
      "learning_rate": 2.175157370587348e-05,
      "loss": 0.9036,
      "step": 624
    },
    {
      "epoch": 3.427004797806717,
      "grad_norm": 0.19564360580394577,
      "learning_rate": 2.1615170334283535e-05,
      "loss": 0.8988,
      "step": 625
    },
    {
      "epoch": 3.432488005483208,
      "grad_norm": 0.15688766150962508,
      "learning_rate": 2.1479037477486936e-05,
      "loss": 0.8866,
      "step": 626
    },
    {
      "epoch": 3.4379712131596984,
      "grad_norm": 0.1814220481133191,
      "learning_rate": 2.1343177138545547e-05,
      "loss": 0.8957,
      "step": 627
    },
    {
      "epoch": 3.443454420836189,
      "grad_norm": 0.19751191092231474,
      "learning_rate": 2.1207591316511454e-05,
      "loss": 0.8952,
      "step": 628
    },
    {
      "epoch": 3.44893762851268,
      "grad_norm": 0.1646360237105222,
      "learning_rate": 2.1072282006397425e-05,
      "loss": 0.8953,
      "step": 629
    },
    {
      "epoch": 3.4544208361891706,
      "grad_norm": 0.18843073086840514,
      "learning_rate": 2.0937251199147684e-05,
      "loss": 0.9038,
      "step": 630
    },
    {
      "epoch": 3.459904043865661,
      "grad_norm": 0.17451033699060078,
      "learning_rate": 2.0802500881608557e-05,
      "loss": 0.9059,
      "step": 631
    },
    {
      "epoch": 3.465387251542152,
      "grad_norm": 0.15864010191777436,
      "learning_rate": 2.066803303649918e-05,
      "loss": 0.8736,
      "step": 632
    },
    {
      "epoch": 3.470870459218643,
      "grad_norm": 0.1745241432795982,
      "learning_rate": 2.0533849642382446e-05,
      "loss": 0.8966,
      "step": 633
    },
    {
      "epoch": 3.476353666895134,
      "grad_norm": 0.15563829064977594,
      "learning_rate": 2.039995267363581e-05,
      "loss": 0.9013,
      "step": 634
    },
    {
      "epoch": 3.4818368745716244,
      "grad_norm": 0.18758674942103779,
      "learning_rate": 2.026634410042223e-05,
      "loss": 0.8947,
      "step": 635
    },
    {
      "epoch": 3.487320082248115,
      "grad_norm": 0.15887193712810688,
      "learning_rate": 2.0133025888661263e-05,
      "loss": 0.9054,
      "step": 636
    },
    {
      "epoch": 3.492803289924606,
      "grad_norm": 0.17452023322549837,
      "learning_rate": 2.0000000000000012e-05,
      "loss": 0.9076,
      "step": 637
    },
    {
      "epoch": 3.4982864976010966,
      "grad_norm": 0.18330071982251422,
      "learning_rate": 1.986726839178438e-05,
      "loss": 0.9015,
      "step": 638
    },
    {
      "epoch": 3.5037697052775876,
      "grad_norm": 0.1575715551896094,
      "learning_rate": 1.9734833017030227e-05,
      "loss": 0.8813,
      "step": 639
    },
    {
      "epoch": 3.509252912954078,
      "grad_norm": 0.18878562836114246,
      "learning_rate": 1.9602695824394576e-05,
      "loss": 0.9005,
      "step": 640
    },
    {
      "epoch": 3.5147361206305687,
      "grad_norm": 0.15533387161156822,
      "learning_rate": 1.9470858758147036e-05,
      "loss": 0.8757,
      "step": 641
    },
    {
      "epoch": 3.5202193283070597,
      "grad_norm": 0.1630547237218842,
      "learning_rate": 1.933932375814114e-05,
      "loss": 0.9038,
      "step": 642
    },
    {
      "epoch": 3.5257025359835503,
      "grad_norm": 0.1712109811813745,
      "learning_rate": 1.9208092759785818e-05,
      "loss": 0.9049,
      "step": 643
    },
    {
      "epoch": 3.531185743660041,
      "grad_norm": 0.14587673483344632,
      "learning_rate": 1.9077167694016903e-05,
      "loss": 0.8985,
      "step": 644
    },
    {
      "epoch": 3.536668951336532,
      "grad_norm": 0.16414938500317564,
      "learning_rate": 1.8946550487268706e-05,
      "loss": 0.9068,
      "step": 645
    },
    {
      "epoch": 3.5421521590130225,
      "grad_norm": 0.15376745325776353,
      "learning_rate": 1.8816243061445734e-05,
      "loss": 0.8955,
      "step": 646
    },
    {
      "epoch": 3.5476353666895135,
      "grad_norm": 0.15960811991170182,
      "learning_rate": 1.8686247333894366e-05,
      "loss": 0.8967,
      "step": 647
    },
    {
      "epoch": 3.553118574366004,
      "grad_norm": 0.15855046926600352,
      "learning_rate": 1.8556565217374606e-05,
      "loss": 0.8951,
      "step": 648
    },
    {
      "epoch": 3.558601782042495,
      "grad_norm": 0.15624502487015215,
      "learning_rate": 1.8427198620032037e-05,
      "loss": 0.9061,
      "step": 649
    },
    {
      "epoch": 3.5640849897189857,
      "grad_norm": 0.1679249396514824,
      "learning_rate": 1.829814944536963e-05,
      "loss": 0.9094,
      "step": 650
    },
    {
      "epoch": 3.5695681973954763,
      "grad_norm": 0.15999412608591182,
      "learning_rate": 1.8169419592219813e-05,
      "loss": 0.8991,
      "step": 651
    },
    {
      "epoch": 3.5750514050719673,
      "grad_norm": 0.17340179382194568,
      "learning_rate": 1.8041010954716544e-05,
      "loss": 0.9103,
      "step": 652
    },
    {
      "epoch": 3.580534612748458,
      "grad_norm": 0.15237872236931496,
      "learning_rate": 1.7912925422267345e-05,
      "loss": 0.8999,
      "step": 653
    },
    {
      "epoch": 3.5860178204249484,
      "grad_norm": 1.2272413852202702,
      "learning_rate": 1.7785164879525604e-05,
      "loss": 0.9349,
      "step": 654
    },
    {
      "epoch": 3.5915010281014395,
      "grad_norm": 0.17476206012460227,
      "learning_rate": 1.7657731206362813e-05,
      "loss": 0.9132,
      "step": 655
    },
    {
      "epoch": 3.59698423577793,
      "grad_norm": 0.1762962782430207,
      "learning_rate": 1.7530626277840846e-05,
      "loss": 0.8938,
      "step": 656
    },
    {
      "epoch": 3.6024674434544206,
      "grad_norm": 0.181015770218541,
      "learning_rate": 1.7403851964184486e-05,
      "loss": 0.9069,
      "step": 657
    },
    {
      "epoch": 3.6079506511309116,
      "grad_norm": 0.16366717982789378,
      "learning_rate": 1.7277410130753775e-05,
      "loss": 0.8954,
      "step": 658
    },
    {
      "epoch": 3.613433858807402,
      "grad_norm": 0.1612334600367671,
      "learning_rate": 1.7151302638016683e-05,
      "loss": 0.884,
      "step": 659
    },
    {
      "epoch": 3.618917066483893,
      "grad_norm": 0.16685277983540153,
      "learning_rate": 1.7025531341521685e-05,
      "loss": 0.894,
      "step": 660
    },
    {
      "epoch": 3.624400274160384,
      "grad_norm": 0.1669827992942319,
      "learning_rate": 1.690009809187041e-05,
      "loss": 0.8952,
      "step": 661
    },
    {
      "epoch": 3.629883481836875,
      "grad_norm": 0.15253380478976128,
      "learning_rate": 1.6775004734690495e-05,
      "loss": 0.8864,
      "step": 662
    },
    {
      "epoch": 3.6353666895133654,
      "grad_norm": 0.15609996007733176,
      "learning_rate": 1.6650253110608415e-05,
      "loss": 0.9006,
      "step": 663
    },
    {
      "epoch": 3.640849897189856,
      "grad_norm": 0.1461455903080232,
      "learning_rate": 1.6525845055222306e-05,
      "loss": 0.917,
      "step": 664
    },
    {
      "epoch": 3.646333104866347,
      "grad_norm": 0.15716311122738527,
      "learning_rate": 1.6401782399075098e-05,
      "loss": 0.9149,
      "step": 665
    },
    {
      "epoch": 3.6518163125428376,
      "grad_norm": 0.32627646467352905,
      "learning_rate": 1.627806696762745e-05,
      "loss": 0.9054,
      "step": 666
    },
    {
      "epoch": 3.657299520219328,
      "grad_norm": 0.16854342857121438,
      "learning_rate": 1.615470058123099e-05,
      "loss": 0.8805,
      "step": 667
    },
    {
      "epoch": 3.662782727895819,
      "grad_norm": 0.1747226962814483,
      "learning_rate": 1.603168505510148e-05,
      "loss": 0.9052,
      "step": 668
    },
    {
      "epoch": 3.6682659355723097,
      "grad_norm": 0.16903191589677927,
      "learning_rate": 1.5909022199292104e-05,
      "loss": 0.8961,
      "step": 669
    },
    {
      "epoch": 3.6737491432488003,
      "grad_norm": 8.87107113952348,
      "learning_rate": 1.5786713818666876e-05,
      "loss": 0.9752,
      "step": 670
    },
    {
      "epoch": 3.6792323509252913,
      "grad_norm": 0.20674605140421184,
      "learning_rate": 1.566476171287401e-05,
      "loss": 0.8898,
      "step": 671
    },
    {
      "epoch": 3.684715558601782,
      "grad_norm": 0.19629370706239815,
      "learning_rate": 1.554316767631951e-05,
      "loss": 0.9059,
      "step": 672
    },
    {
      "epoch": 3.690198766278273,
      "grad_norm": 0.21259237974634135,
      "learning_rate": 1.5421933498140763e-05,
      "loss": 0.8958,
      "step": 673
    },
    {
      "epoch": 3.6956819739547635,
      "grad_norm": 0.18310426924162343,
      "learning_rate": 1.5301060962180133e-05,
      "loss": 0.8853,
      "step": 674
    },
    {
      "epoch": 3.7011651816312545,
      "grad_norm": 0.1751171847378099,
      "learning_rate": 1.518055184695884e-05,
      "loss": 0.8813,
      "step": 675
    },
    {
      "epoch": 3.706648389307745,
      "grad_norm": 0.16886538069007115,
      "learning_rate": 1.5060407925650662e-05,
      "loss": 0.8881,
      "step": 676
    },
    {
      "epoch": 3.7121315969842357,
      "grad_norm": 0.17046906947879148,
      "learning_rate": 1.494063096605595e-05,
      "loss": 0.8937,
      "step": 677
    },
    {
      "epoch": 3.7176148046607267,
      "grad_norm": 0.1581379052592717,
      "learning_rate": 1.4821222730575561e-05,
      "loss": 0.8972,
      "step": 678
    },
    {
      "epoch": 3.7230980123372173,
      "grad_norm": 0.6969777085406519,
      "learning_rate": 1.4702184976184915e-05,
      "loss": 0.9041,
      "step": 679
    },
    {
      "epoch": 3.728581220013708,
      "grad_norm": 0.2182887056206488,
      "learning_rate": 1.4583519454408191e-05,
      "loss": 0.9039,
      "step": 680
    },
    {
      "epoch": 3.734064427690199,
      "grad_norm": 0.22945666556054492,
      "learning_rate": 1.4465227911292537e-05,
      "loss": 0.8887,
      "step": 681
    },
    {
      "epoch": 3.7395476353666894,
      "grad_norm": 0.22736819225186297,
      "learning_rate": 1.434731208738232e-05,
      "loss": 0.9121,
      "step": 682
    },
    {
      "epoch": 3.74503084304318,
      "grad_norm": 0.20165874554977742,
      "learning_rate": 1.4229773717693625e-05,
      "loss": 0.8965,
      "step": 683
    },
    {
      "epoch": 3.750514050719671,
      "grad_norm": 0.184903868108441,
      "learning_rate": 1.4112614531688645e-05,
      "loss": 0.9045,
      "step": 684
    },
    {
      "epoch": 3.7559972583961616,
      "grad_norm": 0.18326386101695807,
      "learning_rate": 1.3995836253250233e-05,
      "loss": 0.8995,
      "step": 685
    },
    {
      "epoch": 3.7614804660726526,
      "grad_norm": 0.20107358682198542,
      "learning_rate": 1.3879440600656607e-05,
      "loss": 0.9072,
      "step": 686
    },
    {
      "epoch": 3.766963673749143,
      "grad_norm": 0.3569275389027143,
      "learning_rate": 1.3763429286555963e-05,
      "loss": 0.9048,
      "step": 687
    },
    {
      "epoch": 3.7724468814256342,
      "grad_norm": 0.19199246127012237,
      "learning_rate": 1.3647804017941373e-05,
      "loss": 0.8941,
      "step": 688
    },
    {
      "epoch": 3.777930089102125,
      "grad_norm": 0.19373559370826268,
      "learning_rate": 1.3532566496125634e-05,
      "loss": 0.9014,
      "step": 689
    },
    {
      "epoch": 3.7834132967786154,
      "grad_norm": 0.180911982737044,
      "learning_rate": 1.3417718416716183e-05,
      "loss": 0.8911,
      "step": 690
    },
    {
      "epoch": 3.7888965044551064,
      "grad_norm": 0.17759995896183428,
      "learning_rate": 1.3303261469590228e-05,
      "loss": 0.8932,
      "step": 691
    },
    {
      "epoch": 3.794379712131597,
      "grad_norm": 0.159075207881893,
      "learning_rate": 1.3189197338869853e-05,
      "loss": 0.882,
      "step": 692
    },
    {
      "epoch": 3.7998629198080875,
      "grad_norm": 0.18045574791107008,
      "learning_rate": 1.3075527702897185e-05,
      "loss": 0.9228,
      "step": 693
    },
    {
      "epoch": 3.8053461274845786,
      "grad_norm": 0.16316813300832433,
      "learning_rate": 1.2962254234209826e-05,
      "loss": 0.8894,
      "step": 694
    },
    {
      "epoch": 3.810829335161069,
      "grad_norm": 0.17660313697972807,
      "learning_rate": 1.2849378599516085e-05,
      "loss": 0.8997,
      "step": 695
    },
    {
      "epoch": 3.8163125428375597,
      "grad_norm": 0.1679766974660261,
      "learning_rate": 1.273690245967059e-05,
      "loss": 0.9042,
      "step": 696
    },
    {
      "epoch": 3.8217957505140507,
      "grad_norm": 0.15705812295975527,
      "learning_rate": 1.2624827469649739e-05,
      "loss": 0.8959,
      "step": 697
    },
    {
      "epoch": 3.8272789581905413,
      "grad_norm": 0.1600279121047265,
      "learning_rate": 1.2513155278527446e-05,
      "loss": 0.8942,
      "step": 698
    },
    {
      "epoch": 3.8327621658670323,
      "grad_norm": 0.14891745897684322,
      "learning_rate": 1.240188752945084e-05,
      "loss": 0.9133,
      "step": 699
    },
    {
      "epoch": 3.838245373543523,
      "grad_norm": 1.0164234645051264,
      "learning_rate": 1.2291025859616026e-05,
      "loss": 0.8963,
      "step": 700
    },
    {
      "epoch": 3.843728581220014,
      "grad_norm": 0.15287661592525834,
      "learning_rate": 1.21805719002441e-05,
      "loss": 0.896,
      "step": 701
    },
    {
      "epoch": 3.8492117888965045,
      "grad_norm": 0.18515627168326385,
      "learning_rate": 1.2070527276557092e-05,
      "loss": 0.8907,
      "step": 702
    },
    {
      "epoch": 3.854694996572995,
      "grad_norm": 0.19538289491293945,
      "learning_rate": 1.1960893607754022e-05,
      "loss": 0.8866,
      "step": 703
    },
    {
      "epoch": 3.860178204249486,
      "grad_norm": 0.1803420043271255,
      "learning_rate": 1.1851672506987165e-05,
      "loss": 0.9027,
      "step": 704
    },
    {
      "epoch": 3.8656614119259767,
      "grad_norm": 0.1728624061069401,
      "learning_rate": 1.17428655813382e-05,
      "loss": 0.9088,
      "step": 705
    },
    {
      "epoch": 3.8711446196024673,
      "grad_norm": 0.18740842748319184,
      "learning_rate": 1.1634474431794676e-05,
      "loss": 0.9012,
      "step": 706
    },
    {
      "epoch": 3.8766278272789583,
      "grad_norm": 0.16138769768933767,
      "learning_rate": 1.1526500653226385e-05,
      "loss": 0.9168,
      "step": 707
    },
    {
      "epoch": 3.882111034955449,
      "grad_norm": 0.14601780097035175,
      "learning_rate": 1.141894583436189e-05,
      "loss": 0.8923,
      "step": 708
    },
    {
      "epoch": 3.8875942426319394,
      "grad_norm": 0.16268490177886352,
      "learning_rate": 1.1311811557765208e-05,
      "loss": 0.9011,
      "step": 709
    },
    {
      "epoch": 3.8930774503084304,
      "grad_norm": 0.14407863627045317,
      "learning_rate": 1.1205099399812478e-05,
      "loss": 0.8937,
      "step": 710
    },
    {
      "epoch": 3.898560657984921,
      "grad_norm": 0.14837780514070958,
      "learning_rate": 1.1098810930668754e-05,
      "loss": 0.9035,
      "step": 711
    },
    {
      "epoch": 3.904043865661412,
      "grad_norm": 0.13890190728316182,
      "learning_rate": 1.0992947714264952e-05,
      "loss": 0.8925,
      "step": 712
    },
    {
      "epoch": 3.9095270733379026,
      "grad_norm": 0.1482520122684546,
      "learning_rate": 1.088751130827478e-05,
      "loss": 0.8864,
      "step": 713
    },
    {
      "epoch": 3.9150102810143936,
      "grad_norm": 0.19175152200646928,
      "learning_rate": 1.078250326409188e-05,
      "loss": 0.9022,
      "step": 714
    },
    {
      "epoch": 3.920493488690884,
      "grad_norm": 0.14176488398671347,
      "learning_rate": 1.0677925126806956e-05,
      "loss": 0.9017,
      "step": 715
    },
    {
      "epoch": 3.925976696367375,
      "grad_norm": 0.14891595552628026,
      "learning_rate": 1.0573778435185039e-05,
      "loss": 0.9038,
      "step": 716
    },
    {
      "epoch": 3.931459904043866,
      "grad_norm": 0.13751484111663212,
      "learning_rate": 1.047006472164287e-05,
      "loss": 0.8925,
      "step": 717
    },
    {
      "epoch": 3.9369431117203564,
      "grad_norm": 0.1431600770457188,
      "learning_rate": 1.0366785512226359e-05,
      "loss": 0.903,
      "step": 718
    },
    {
      "epoch": 3.942426319396847,
      "grad_norm": 0.7396147350976154,
      "learning_rate": 1.0263942326588054e-05,
      "loss": 0.9243,
      "step": 719
    },
    {
      "epoch": 3.947909527073338,
      "grad_norm": 0.18213078561266646,
      "learning_rate": 1.0161536677964933e-05,
      "loss": 0.8895,
      "step": 720
    },
    {
      "epoch": 3.9533927347498286,
      "grad_norm": 0.19698575071196872,
      "learning_rate": 1.0059570073155953e-05,
      "loss": 0.8988,
      "step": 721
    },
    {
      "epoch": 3.958875942426319,
      "grad_norm": 0.16625311791486364,
      "learning_rate": 9.958044012500023e-06,
      "loss": 0.8841,
      "step": 722
    },
    {
      "epoch": 3.96435915010281,
      "grad_norm": 0.18406574667380657,
      "learning_rate": 9.856959989853876e-06,
      "loss": 0.8959,
      "step": 723
    },
    {
      "epoch": 3.969842357779301,
      "grad_norm": 0.16041228708375482,
      "learning_rate": 9.75631949257004e-06,
      "loss": 0.8885,
      "step": 724
    },
    {
      "epoch": 3.9753255654557917,
      "grad_norm": 0.16790760471515281,
      "learning_rate": 9.656124001475068e-06,
      "loss": 0.8832,
      "step": 725
    },
    {
      "epoch": 3.9808087731322823,
      "grad_norm": 0.1643989573541531,
      "learning_rate": 9.556374990847618e-06,
      "loss": 0.9077,
      "step": 726
    },
    {
      "epoch": 3.9862919808087733,
      "grad_norm": 0.15089157844878578,
      "learning_rate": 9.457073928396871e-06,
      "loss": 0.8924,
      "step": 727
    },
    {
      "epoch": 3.991775188485264,
      "grad_norm": 0.15938422497231847,
      "learning_rate": 9.358222275240884e-06,
      "loss": 0.891,
      "step": 728
    },
    {
      "epoch": 3.9972583961617545,
      "grad_norm": 0.1386488462476611,
      "learning_rate": 9.25982148588506e-06,
      "loss": 0.8969,
      "step": 729
    },
    {
      "epoch": 4.0027416038382455,
      "grad_norm": 0.34700978060414556,
      "learning_rate": 9.161873008200816e-06,
      "loss": 1.4657,
      "step": 730
    },
    {
      "epoch": 4.0082248115147365,
      "grad_norm": 0.2465097801021323,
      "learning_rate": 9.064378283404247e-06,
      "loss": 0.8725,
      "step": 731
    },
    {
      "epoch": 4.013708019191227,
      "grad_norm": 0.26497344033852505,
      "learning_rate": 8.967338746034882e-06,
      "loss": 0.8759,
      "step": 732
    },
    {
      "epoch": 4.019191226867718,
      "grad_norm": 0.21838213857553232,
      "learning_rate": 8.870755823934662e-06,
      "loss": 0.8627,
      "step": 733
    },
    {
      "epoch": 4.024674434544209,
      "grad_norm": 0.20277610227352777,
      "learning_rate": 8.774630938226831e-06,
      "loss": 0.886,
      "step": 734
    },
    {
      "epoch": 4.030157642220699,
      "grad_norm": 0.24713492980404156,
      "learning_rate": 8.678965503295114e-06,
      "loss": 0.8692,
      "step": 735
    },
    {
      "epoch": 4.03564084989719,
      "grad_norm": 0.18596227337602494,
      "learning_rate": 8.583760926762852e-06,
      "loss": 0.8702,
      "step": 736
    },
    {
      "epoch": 4.041124057573681,
      "grad_norm": 0.2255598033578029,
      "learning_rate": 8.489018609472297e-06,
      "loss": 0.859,
      "step": 737
    },
    {
      "epoch": 4.046607265250171,
      "grad_norm": 0.19671824377342131,
      "learning_rate": 8.394739945464016e-06,
      "loss": 0.8825,
      "step": 738
    },
    {
      "epoch": 4.052090472926662,
      "grad_norm": 0.19536276470724584,
      "learning_rate": 8.300926321956391e-06,
      "loss": 0.8655,
      "step": 739
    },
    {
      "epoch": 4.057573680603153,
      "grad_norm": 0.20553044507457272,
      "learning_rate": 8.207579119325145e-06,
      "loss": 0.8581,
      "step": 740
    },
    {
      "epoch": 4.063056888279643,
      "grad_norm": 0.1745254415584832,
      "learning_rate": 8.114699711083113e-06,
      "loss": 0.8559,
      "step": 741
    },
    {
      "epoch": 4.068540095956134,
      "grad_norm": 0.19419389496880718,
      "learning_rate": 8.022289463859963e-06,
      "loss": 0.8684,
      "step": 742
    },
    {
      "epoch": 4.074023303632625,
      "grad_norm": 0.16866208377226335,
      "learning_rate": 7.930349737382137e-06,
      "loss": 0.8603,
      "step": 743
    },
    {
      "epoch": 4.079506511309116,
      "grad_norm": 0.17045009966171426,
      "learning_rate": 7.838881884452827e-06,
      "loss": 0.8775,
      "step": 744
    },
    {
      "epoch": 4.084989718985606,
      "grad_norm": 0.1672289163103852,
      "learning_rate": 7.747887250932047e-06,
      "loss": 0.8577,
      "step": 745
    },
    {
      "epoch": 4.090472926662097,
      "grad_norm": 0.16957620944180635,
      "learning_rate": 7.657367175716884e-06,
      "loss": 0.8573,
      "step": 746
    },
    {
      "epoch": 4.095956134338588,
      "grad_norm": 0.15814438359723273,
      "learning_rate": 7.5673229907217146e-06,
      "loss": 0.8519,
      "step": 747
    },
    {
      "epoch": 4.1014393420150785,
      "grad_norm": 0.1554296876299363,
      "learning_rate": 7.477756020858695e-06,
      "loss": 0.8635,
      "step": 748
    },
    {
      "epoch": 4.10692254969157,
      "grad_norm": 0.16052588316393465,
      "learning_rate": 7.38866758401823e-06,
      "loss": 0.8723,
      "step": 749
    },
    {
      "epoch": 4.112405757368061,
      "grad_norm": 0.15330364083531656,
      "learning_rate": 7.300058991049534e-06,
      "loss": 0.8828,
      "step": 750
    },
    {
      "epoch": 4.117888965044551,
      "grad_norm": 0.1495227789255928,
      "learning_rate": 7.211931545741433e-06,
      "loss": 0.8506,
      "step": 751
    },
    {
      "epoch": 4.123372172721042,
      "grad_norm": 0.15408046491976884,
      "learning_rate": 7.124286544803136e-06,
      "loss": 0.8668,
      "step": 752
    },
    {
      "epoch": 4.128855380397533,
      "grad_norm": 0.14881498638247898,
      "learning_rate": 7.037125277845112e-06,
      "loss": 0.8801,
      "step": 753
    },
    {
      "epoch": 4.134338588074023,
      "grad_norm": 0.1464102070171832,
      "learning_rate": 6.950449027360213e-06,
      "loss": 0.8726,
      "step": 754
    },
    {
      "epoch": 4.139821795750514,
      "grad_norm": 0.1453166333634366,
      "learning_rate": 6.864259068704688e-06,
      "loss": 0.8607,
      "step": 755
    },
    {
      "epoch": 4.145305003427005,
      "grad_norm": 0.13625555326942757,
      "learning_rate": 6.778556670079535e-06,
      "loss": 0.8792,
      "step": 756
    },
    {
      "epoch": 4.150788211103496,
      "grad_norm": 0.13807498863680387,
      "learning_rate": 6.69334309251175e-06,
      "loss": 0.8474,
      "step": 757
    },
    {
      "epoch": 4.156271418779986,
      "grad_norm": 0.14308904016946167,
      "learning_rate": 6.608619589835803e-06,
      "loss": 0.8683,
      "step": 758
    },
    {
      "epoch": 4.161754626456477,
      "grad_norm": 0.1440963907671081,
      "learning_rate": 6.524387408675208e-06,
      "loss": 0.8633,
      "step": 759
    },
    {
      "epoch": 4.167237834132968,
      "grad_norm": 0.14254526065135373,
      "learning_rate": 6.440647788424166e-06,
      "loss": 0.8714,
      "step": 760
    },
    {
      "epoch": 4.172721041809458,
      "grad_norm": 0.15400971348680637,
      "learning_rate": 6.357401961229293e-06,
      "loss": 0.8719,
      "step": 761
    },
    {
      "epoch": 4.178204249485949,
      "grad_norm": 0.1334821564613061,
      "learning_rate": 6.274651151971567e-06,
      "loss": 0.8604,
      "step": 762
    },
    {
      "epoch": 4.18368745716244,
      "grad_norm": 0.15458395585147408,
      "learning_rate": 6.1923965782482165e-06,
      "loss": 0.8855,
      "step": 763
    },
    {
      "epoch": 4.18917066483893,
      "grad_norm": 0.14422168418866493,
      "learning_rate": 6.110639450354882e-06,
      "loss": 0.8709,
      "step": 764
    },
    {
      "epoch": 4.194653872515421,
      "grad_norm": 0.13640204014715013,
      "learning_rate": 6.0293809712677775e-06,
      "loss": 0.8779,
      "step": 765
    },
    {
      "epoch": 4.2001370801919125,
      "grad_norm": 0.13728381458929773,
      "learning_rate": 5.9486223366259555e-06,
      "loss": 0.863,
      "step": 766
    },
    {
      "epoch": 4.205620287868403,
      "grad_norm": 1.0957146408038854,
      "learning_rate": 5.868364734713776e-06,
      "loss": 0.8706,
      "step": 767
    },
    {
      "epoch": 4.211103495544894,
      "grad_norm": 0.14870058965626423,
      "learning_rate": 5.788609346443386e-06,
      "loss": 0.8808,
      "step": 768
    },
    {
      "epoch": 4.216586703221385,
      "grad_norm": 0.16703761517821936,
      "learning_rate": 5.70935734533733e-06,
      "loss": 0.852,
      "step": 769
    },
    {
      "epoch": 4.222069910897876,
      "grad_norm": 0.1844114299522098,
      "learning_rate": 5.630609897511328e-06,
      "loss": 0.8613,
      "step": 770
    },
    {
      "epoch": 4.227553118574366,
      "grad_norm": 0.17398006404372696,
      "learning_rate": 5.552368161657082e-06,
      "loss": 0.8556,
      "step": 771
    },
    {
      "epoch": 4.233036326250857,
      "grad_norm": 0.16676512516292483,
      "learning_rate": 5.474633289025244e-06,
      "loss": 0.8425,
      "step": 772
    },
    {
      "epoch": 4.238519533927348,
      "grad_norm": 0.15690331804763263,
      "learning_rate": 5.397406423408446e-06,
      "loss": 0.874,
      "step": 773
    },
    {
      "epoch": 4.244002741603838,
      "grad_norm": 0.15751054116710525,
      "learning_rate": 5.3206887011245165e-06,
      "loss": 0.8871,
      "step": 774
    },
    {
      "epoch": 4.249485949280329,
      "grad_norm": 0.1535880170728185,
      "learning_rate": 5.24448125099974e-06,
      "loss": 0.8681,
      "step": 775
    },
    {
      "epoch": 4.25496915695682,
      "grad_norm": 0.15063251461992208,
      "learning_rate": 5.1687851943522215e-06,
      "loss": 0.8744,
      "step": 776
    },
    {
      "epoch": 4.26045236463331,
      "grad_norm": 0.1520419945091867,
      "learning_rate": 5.093601644975428e-06,
      "loss": 0.8681,
      "step": 777
    },
    {
      "epoch": 4.265935572309801,
      "grad_norm": 0.14153569721366746,
      "learning_rate": 5.018931709121791e-06,
      "loss": 0.8638,
      "step": 778
    },
    {
      "epoch": 4.271418779986292,
      "grad_norm": 0.14202440325296312,
      "learning_rate": 4.9447764854863915e-06,
      "loss": 0.8595,
      "step": 779
    },
    {
      "epoch": 4.276901987662782,
      "grad_norm": 0.13789016773017412,
      "learning_rate": 4.871137065190854e-06,
      "loss": 0.8738,
      "step": 780
    },
    {
      "epoch": 4.282385195339273,
      "grad_norm": 0.13587353525443027,
      "learning_rate": 4.798014531767261e-06,
      "loss": 0.8665,
      "step": 781
    },
    {
      "epoch": 4.287868403015764,
      "grad_norm": 0.13265718560401082,
      "learning_rate": 4.725409961142173e-06,
      "loss": 0.874,
      "step": 782
    },
    {
      "epoch": 4.293351610692255,
      "grad_norm": 0.13693830134507576,
      "learning_rate": 4.653324421620884e-06,
      "loss": 0.8619,
      "step": 783
    },
    {
      "epoch": 4.2988348183687455,
      "grad_norm": 0.13146748198275093,
      "learning_rate": 4.581758973871609e-06,
      "loss": 0.8825,
      "step": 784
    },
    {
      "epoch": 4.3043180260452365,
      "grad_norm": 0.13405773370704815,
      "learning_rate": 4.510714670909946e-06,
      "loss": 0.8536,
      "step": 785
    },
    {
      "epoch": 4.3098012337217275,
      "grad_norm": 0.14574117129879566,
      "learning_rate": 4.440192558083367e-06,
      "loss": 0.8506,
      "step": 786
    },
    {
      "epoch": 4.315284441398218,
      "grad_norm": 0.13101655607771853,
      "learning_rate": 4.370193673055787e-06,
      "loss": 0.8636,
      "step": 787
    },
    {
      "epoch": 4.320767649074709,
      "grad_norm": 0.13191753249596408,
      "learning_rate": 4.300719045792376e-06,
      "loss": 0.8671,
      "step": 788
    },
    {
      "epoch": 4.3262508567512,
      "grad_norm": 0.1310909838553508,
      "learning_rate": 4.231769698544352e-06,
      "loss": 0.8924,
      "step": 789
    },
    {
      "epoch": 4.33173406442769,
      "grad_norm": 0.13210093237615378,
      "learning_rate": 4.163346645833928e-06,
      "loss": 0.8605,
      "step": 790
    },
    {
      "epoch": 4.337217272104181,
      "grad_norm": 0.128792479044049,
      "learning_rate": 4.0954508944394474e-06,
      "loss": 0.8695,
      "step": 791
    },
    {
      "epoch": 4.342700479780672,
      "grad_norm": 0.22455677717587358,
      "learning_rate": 4.028083443380486e-06,
      "loss": 0.8691,
      "step": 792
    },
    {
      "epoch": 4.348183687457162,
      "grad_norm": 0.40527297890394276,
      "learning_rate": 3.961245283903239e-06,
      "loss": 0.8805,
      "step": 793
    },
    {
      "epoch": 4.353666895133653,
      "grad_norm": 0.2888577519432354,
      "learning_rate": 3.89493739946587e-06,
      "loss": 0.8803,
      "step": 794
    },
    {
      "epoch": 4.359150102810144,
      "grad_norm": 0.1350998279076715,
      "learning_rate": 3.829160765724052e-06,
      "loss": 0.8513,
      "step": 795
    },
    {
      "epoch": 4.364633310486635,
      "grad_norm": 0.1309943366405807,
      "learning_rate": 3.7639163505166633e-06,
      "loss": 0.8668,
      "step": 796
    },
    {
      "epoch": 4.370116518163125,
      "grad_norm": 0.12953103571452862,
      "learning_rate": 3.6992051138514717e-06,
      "loss": 0.8712,
      "step": 797
    },
    {
      "epoch": 4.375599725839616,
      "grad_norm": 0.13057599590166327,
      "learning_rate": 3.635028007891048e-06,
      "loss": 0.8686,
      "step": 798
    },
    {
      "epoch": 4.381082933516107,
      "grad_norm": 0.12973353273093458,
      "learning_rate": 3.5713859769387795e-06,
      "loss": 0.8831,
      "step": 799
    },
    {
      "epoch": 4.386566141192597,
      "grad_norm": 0.13131456969773786,
      "learning_rate": 3.5082799574249094e-06,
      "loss": 0.8639,
      "step": 800
    },
    {
      "epoch": 4.392049348869088,
      "grad_norm": 0.12595490195849035,
      "learning_rate": 3.4457108778928272e-06,
      "loss": 0.8631,
      "step": 801
    },
    {
      "epoch": 4.397532556545579,
      "grad_norm": 0.12813268032502836,
      "learning_rate": 3.3836796589853484e-06,
      "loss": 0.8663,
      "step": 802
    },
    {
      "epoch": 4.4030157642220695,
      "grad_norm": 0.12716051374477083,
      "learning_rate": 3.3221872134312184e-06,
      "loss": 0.8525,
      "step": 803
    },
    {
      "epoch": 4.4084989718985605,
      "grad_norm": 0.12371044909708838,
      "learning_rate": 3.261234446031658e-06,
      "loss": 0.8679,
      "step": 804
    },
    {
      "epoch": 4.413982179575052,
      "grad_norm": 0.12478585102448334,
      "learning_rate": 3.200822253647031e-06,
      "loss": 0.8889,
      "step": 805
    },
    {
      "epoch": 4.419465387251542,
      "grad_norm": 0.1293948457179513,
      "learning_rate": 3.140951525183691e-06,
      "loss": 0.8629,
      "step": 806
    },
    {
      "epoch": 4.424948594928033,
      "grad_norm": 0.127763766689417,
      "learning_rate": 3.0816231415808785e-06,
      "loss": 0.8545,
      "step": 807
    },
    {
      "epoch": 4.430431802604524,
      "grad_norm": 0.12507739335573223,
      "learning_rate": 3.02283797579773e-06,
      "loss": 0.8719,
      "step": 808
    },
    {
      "epoch": 4.435915010281015,
      "grad_norm": 0.12408679274948532,
      "learning_rate": 2.9645968928005085e-06,
      "loss": 0.8643,
      "step": 809
    },
    {
      "epoch": 4.441398217957505,
      "grad_norm": 0.14843016184470606,
      "learning_rate": 2.906900749549784e-06,
      "loss": 0.8993,
      "step": 810
    },
    {
      "epoch": 4.446881425633996,
      "grad_norm": 0.12190268058344725,
      "learning_rate": 2.849750394987907e-06,
      "loss": 0.8632,
      "step": 811
    },
    {
      "epoch": 4.452364633310487,
      "grad_norm": 0.12079449899424143,
      "learning_rate": 2.793146670026472e-06,
      "loss": 0.8792,
      "step": 812
    },
    {
      "epoch": 4.457847840986977,
      "grad_norm": 0.12301325559585843,
      "learning_rate": 2.737090407533938e-06,
      "loss": 0.8561,
      "step": 813
    },
    {
      "epoch": 4.463331048663468,
      "grad_norm": 0.12435308059508353,
      "learning_rate": 2.681582432323406e-06,
      "loss": 0.867,
      "step": 814
    },
    {
      "epoch": 4.468814256339959,
      "grad_norm": 0.12288901600568765,
      "learning_rate": 2.6266235611404645e-06,
      "loss": 0.8742,
      "step": 815
    },
    {
      "epoch": 4.474297464016449,
      "grad_norm": 0.12223916369585316,
      "learning_rate": 2.5722146026511574e-06,
      "loss": 0.8566,
      "step": 816
    },
    {
      "epoch": 4.47978067169294,
      "grad_norm": 0.12065119171288469,
      "learning_rate": 2.5183563574301185e-06,
      "loss": 0.869,
      "step": 817
    },
    {
      "epoch": 4.485263879369431,
      "grad_norm": 0.12212414616793427,
      "learning_rate": 2.465049617948778e-06,
      "loss": 0.8776,
      "step": 818
    },
    {
      "epoch": 4.490747087045921,
      "grad_norm": 0.12172913149868783,
      "learning_rate": 2.4122951685636674e-06,
      "loss": 0.8778,
      "step": 819
    },
    {
      "epoch": 4.496230294722412,
      "grad_norm": 0.12399214672845997,
      "learning_rate": 2.3600937855049467e-06,
      "loss": 0.872,
      "step": 820
    },
    {
      "epoch": 4.501713502398903,
      "grad_norm": 0.1209018345998114,
      "learning_rate": 2.308446236864916e-06,
      "loss": 0.8586,
      "step": 821
    },
    {
      "epoch": 4.5071967100753945,
      "grad_norm": 0.1177513204148733,
      "learning_rate": 2.257353282586774e-06,
      "loss": 0.8404,
      "step": 822
    },
    {
      "epoch": 4.512679917751885,
      "grad_norm": 0.12365836386703247,
      "learning_rate": 2.206815674453373e-06,
      "loss": 0.876,
      "step": 823
    },
    {
      "epoch": 4.518163125428376,
      "grad_norm": 0.12060944467881936,
      "learning_rate": 2.1568341560762152e-06,
      "loss": 0.8756,
      "step": 824
    },
    {
      "epoch": 4.523646333104867,
      "grad_norm": 0.1205949737711182,
      "learning_rate": 2.1074094628844754e-06,
      "loss": 0.8777,
      "step": 825
    },
    {
      "epoch": 4.529129540781357,
      "grad_norm": 0.12079575476686041,
      "learning_rate": 2.0585423221141807e-06,
      "loss": 0.8736,
      "step": 826
    },
    {
      "epoch": 4.534612748457848,
      "grad_norm": 0.1218689347455744,
      "learning_rate": 2.010233452797534e-06,
      "loss": 0.8667,
      "step": 827
    },
    {
      "epoch": 4.540095956134339,
      "grad_norm": 0.12283214110808495,
      "learning_rate": 1.9624835657523222e-06,
      "loss": 0.8666,
      "step": 828
    },
    {
      "epoch": 4.54557916381083,
      "grad_norm": 0.11913304460847574,
      "learning_rate": 1.9152933635714354e-06,
      "loss": 0.8667,
      "step": 829
    },
    {
      "epoch": 4.55106237148732,
      "grad_norm": 0.12377547685145791,
      "learning_rate": 1.8686635406125697e-06,
      "loss": 0.8898,
      "step": 830
    },
    {
      "epoch": 4.556545579163811,
      "grad_norm": 0.11633012806752614,
      "learning_rate": 1.822594782987972e-06,
      "loss": 0.8411,
      "step": 831
    },
    {
      "epoch": 4.562028786840301,
      "grad_norm": 0.11864592533450119,
      "learning_rate": 1.7770877685543687e-06,
      "loss": 0.8679,
      "step": 832
    },
    {
      "epoch": 4.567511994516792,
      "grad_norm": 0.12190861480595079,
      "learning_rate": 1.7321431669029953e-06,
      "loss": 0.8708,
      "step": 833
    },
    {
      "epoch": 4.572995202193283,
      "grad_norm": 0.12195430153778734,
      "learning_rate": 1.6877616393497075e-06,
      "loss": 0.8762,
      "step": 834
    },
    {
      "epoch": 4.578478409869774,
      "grad_norm": 0.12318104357609574,
      "learning_rate": 1.6439438389252948e-06,
      "loss": 0.8817,
      "step": 835
    },
    {
      "epoch": 4.583961617546264,
      "grad_norm": 0.1252397305163344,
      "learning_rate": 1.6006904103658572e-06,
      "loss": 0.8713,
      "step": 836
    },
    {
      "epoch": 4.589444825222755,
      "grad_norm": 0.11863485627738708,
      "learning_rate": 1.5580019901032929e-06,
      "loss": 0.8585,
      "step": 837
    },
    {
      "epoch": 4.594928032899246,
      "grad_norm": 0.11926883785074256,
      "learning_rate": 1.5158792062559813e-06,
      "loss": 0.8553,
      "step": 838
    },
    {
      "epoch": 4.6004112405757365,
      "grad_norm": 0.12010806317107317,
      "learning_rate": 1.4743226786194931e-06,
      "loss": 0.8632,
      "step": 839
    },
    {
      "epoch": 4.6058944482522275,
      "grad_norm": 0.1209497918934812,
      "learning_rate": 1.4333330186575079e-06,
      "loss": 0.8661,
      "step": 840
    },
    {
      "epoch": 4.6113776559287185,
      "grad_norm": 0.12015700387003853,
      "learning_rate": 1.3929108294927951e-06,
      "loss": 0.8672,
      "step": 841
    },
    {
      "epoch": 4.6168608636052095,
      "grad_norm": 0.12422998319235647,
      "learning_rate": 1.3530567058983369e-06,
      "loss": 0.8732,
      "step": 842
    },
    {
      "epoch": 4.6223440712817,
      "grad_norm": 0.11888832603251932,
      "learning_rate": 1.31377123428861e-06,
      "loss": 0.8723,
      "step": 843
    },
    {
      "epoch": 4.627827278958191,
      "grad_norm": 0.11996366586250991,
      "learning_rate": 1.2750549927109136e-06,
      "loss": 0.8574,
      "step": 844
    },
    {
      "epoch": 4.633310486634681,
      "grad_norm": 0.11745160261296496,
      "learning_rate": 1.2369085508368862e-06,
      "loss": 0.8604,
      "step": 845
    },
    {
      "epoch": 4.638793694311172,
      "grad_norm": 0.12200784232729492,
      "learning_rate": 1.1993324699541265e-06,
      "loss": 0.8517,
      "step": 846
    },
    {
      "epoch": 4.644276901987663,
      "grad_norm": 0.11775847148513344,
      "learning_rate": 1.1623273029579195e-06,
      "loss": 0.8695,
      "step": 847
    },
    {
      "epoch": 4.649760109664154,
      "grad_norm": 0.11881330238355213,
      "learning_rate": 1.1258935943431237e-06,
      "loss": 0.8667,
      "step": 848
    },
    {
      "epoch": 4.655243317340644,
      "grad_norm": 0.11930205382890313,
      "learning_rate": 1.090031880196145e-06,
      "loss": 0.8669,
      "step": 849
    },
    {
      "epoch": 4.660726525017135,
      "grad_norm": 0.12062456998925507,
      "learning_rate": 1.0547426881870292e-06,
      "loss": 0.8706,
      "step": 850
    },
    {
      "epoch": 4.666209732693626,
      "grad_norm": 0.119609080366811,
      "learning_rate": 1.0200265375617514e-06,
      "loss": 0.8735,
      "step": 851
    },
    {
      "epoch": 4.671692940370116,
      "grad_norm": 0.11883822127861333,
      "learning_rate": 9.858839391345065e-07,
      "loss": 0.8857,
      "step": 852
    },
    {
      "epoch": 4.677176148046607,
      "grad_norm": 0.11869096130082298,
      "learning_rate": 9.523153952802633e-07,
      "loss": 0.8567,
      "step": 853
    },
    {
      "epoch": 4.682659355723098,
      "grad_norm": 0.11660180847816304,
      "learning_rate": 9.193213999273199e-07,
      "loss": 0.8732,
      "step": 854
    },
    {
      "epoch": 4.688142563399589,
      "grad_norm": 0.1156650459781897,
      "learning_rate": 8.869024385500524e-07,
      "loss": 0.8694,
      "step": 855
    },
    {
      "epoch": 4.693625771076079,
      "grad_norm": 0.11849635720653122,
      "learning_rate": 8.550589881617877e-07,
      "loss": 0.8666,
      "step": 856
    },
    {
      "epoch": 4.69910897875257,
      "grad_norm": 0.12360260427554048,
      "learning_rate": 8.237915173077681e-07,
      "loss": 0.8503,
      "step": 857
    },
    {
      "epoch": 4.7045921864290605,
      "grad_norm": 0.11834338075827068,
      "learning_rate": 7.93100486058247e-07,
      "loss": 0.8824,
      "step": 858
    },
    {
      "epoch": 4.7100753941055515,
      "grad_norm": 0.12094889766224019,
      "learning_rate": 7.629863460017506e-07,
      "loss": 0.8686,
      "step": 859
    },
    {
      "epoch": 4.715558601782043,
      "grad_norm": 0.12037069037280847,
      "learning_rate": 7.334495402383957e-07,
      "loss": 0.8668,
      "step": 860
    },
    {
      "epoch": 4.721041809458534,
      "grad_norm": 0.1165471688624166,
      "learning_rate": 7.044905033734096e-07,
      "loss": 0.878,
      "step": 861
    },
    {
      "epoch": 4.726525017135024,
      "grad_norm": 0.11868226704814998,
      "learning_rate": 6.761096615107043e-07,
      "loss": 0.8786,
      "step": 862
    },
    {
      "epoch": 4.732008224811515,
      "grad_norm": 0.11761938779683033,
      "learning_rate": 6.483074322466154e-07,
      "loss": 0.8801,
      "step": 863
    },
    {
      "epoch": 4.737491432488006,
      "grad_norm": 0.119444219753523,
      "learning_rate": 6.210842246637683e-07,
      "loss": 0.8734,
      "step": 864
    },
    {
      "epoch": 4.742974640164496,
      "grad_norm": 0.11552293652107705,
      "learning_rate": 5.944404393250481e-07,
      "loss": 0.8443,
      "step": 865
    },
    {
      "epoch": 4.748457847840987,
      "grad_norm": 0.11624067828548895,
      "learning_rate": 5.683764682677018e-07,
      "loss": 0.8826,
      "step": 866
    },
    {
      "epoch": 4.753941055517478,
      "grad_norm": 0.11747660823290842,
      "learning_rate": 5.428926949975788e-07,
      "loss": 0.871,
      "step": 867
    },
    {
      "epoch": 4.759424263193969,
      "grad_norm": 0.11800804895284038,
      "learning_rate": 5.179894944834863e-07,
      "loss": 0.8808,
      "step": 868
    },
    {
      "epoch": 4.764907470870459,
      "grad_norm": 0.11426381832520421,
      "learning_rate": 4.936672331516778e-07,
      "loss": 0.865,
      "step": 869
    },
    {
      "epoch": 4.77039067854695,
      "grad_norm": 0.11934453977172532,
      "learning_rate": 4.699262688804451e-07,
      "loss": 0.8476,
      "step": 870
    },
    {
      "epoch": 4.77587388622344,
      "grad_norm": 0.11932772884879854,
      "learning_rate": 4.467669509948591e-07,
      "loss": 0.8644,
      "step": 871
    },
    {
      "epoch": 4.781357093899931,
      "grad_norm": 0.11533502002121204,
      "learning_rate": 4.241896202616502e-07,
      "loss": 0.8659,
      "step": 872
    },
    {
      "epoch": 4.786840301576422,
      "grad_norm": 0.11707563398942186,
      "learning_rate": 4.0219460888415884e-07,
      "loss": 0.8687,
      "step": 873
    },
    {
      "epoch": 4.792323509252913,
      "grad_norm": 0.11572299884307491,
      "learning_rate": 3.807822404974726e-07,
      "loss": 0.8619,
      "step": 874
    },
    {
      "epoch": 4.797806716929403,
      "grad_norm": 0.11414580388031612,
      "learning_rate": 3.599528301636612e-07,
      "loss": 0.8773,
      "step": 875
    },
    {
      "epoch": 4.803289924605894,
      "grad_norm": 0.1139811437838818,
      "learning_rate": 3.397066843671315e-07,
      "loss": 0.8614,
      "step": 876
    },
    {
      "epoch": 4.8087731322823855,
      "grad_norm": 0.11506534996205621,
      "learning_rate": 3.200441010101196e-07,
      "loss": 0.851,
      "step": 877
    },
    {
      "epoch": 4.814256339958876,
      "grad_norm": 0.11489034470022166,
      "learning_rate": 3.0096536940832145e-07,
      "loss": 0.8847,
      "step": 878
    },
    {
      "epoch": 4.819739547635367,
      "grad_norm": 0.11675777776548621,
      "learning_rate": 2.824707702866114e-07,
      "loss": 0.8569,
      "step": 879
    },
    {
      "epoch": 4.825222755311858,
      "grad_norm": 0.11463623776794303,
      "learning_rate": 2.645605757749392e-07,
      "loss": 0.8607,
      "step": 880
    },
    {
      "epoch": 4.830705962988349,
      "grad_norm": 0.11596024436355572,
      "learning_rate": 2.4723504940430187e-07,
      "loss": 0.8815,
      "step": 881
    },
    {
      "epoch": 4.836189170664839,
      "grad_norm": 0.11832403238695531,
      "learning_rate": 2.3049444610288462e-07,
      "loss": 0.8729,
      "step": 882
    },
    {
      "epoch": 4.84167237834133,
      "grad_norm": 0.1163599327714919,
      "learning_rate": 2.1433901219229502e-07,
      "loss": 0.8607,
      "step": 883
    },
    {
      "epoch": 4.84715558601782,
      "grad_norm": 0.11506741204644842,
      "learning_rate": 1.9876898538394362e-07,
      "loss": 0.875,
      "step": 884
    },
    {
      "epoch": 4.852638793694311,
      "grad_norm": 0.11571697024128855,
      "learning_rate": 1.8378459477555788e-07,
      "loss": 0.876,
      "step": 885
    },
    {
      "epoch": 4.858122001370802,
      "grad_norm": 0.1233092898569507,
      "learning_rate": 1.6938606084779375e-07,
      "loss": 0.8612,
      "step": 886
    },
    {
      "epoch": 4.863605209047293,
      "grad_norm": 0.11387876698076485,
      "learning_rate": 1.555735954610027e-07,
      "loss": 0.8499,
      "step": 887
    },
    {
      "epoch": 4.869088416723783,
      "grad_norm": 0.11498145411665506,
      "learning_rate": 1.4234740185210095e-07,
      "loss": 0.859,
      "step": 888
    },
    {
      "epoch": 4.874571624400274,
      "grad_norm": 0.11650609203788179,
      "learning_rate": 1.2970767463160284e-07,
      "loss": 0.8602,
      "step": 889
    },
    {
      "epoch": 4.880054832076765,
      "grad_norm": 0.11664236567167152,
      "learning_rate": 1.176545997807299e-07,
      "loss": 0.8746,
      "step": 890
    },
    {
      "epoch": 4.885538039753255,
      "grad_norm": 0.11702468705401428,
      "learning_rate": 1.0618835464870191e-07,
      "loss": 0.8525,
      "step": 891
    },
    {
      "epoch": 4.891021247429746,
      "grad_norm": 0.11451498374169765,
      "learning_rate": 9.530910795009895e-08,
      "loss": 0.8599,
      "step": 892
    },
    {
      "epoch": 4.896504455106237,
      "grad_norm": 0.11475845584891285,
      "learning_rate": 8.501701976239673e-08,
      "loss": 0.8583,
      "step": 893
    },
    {
      "epoch": 4.901987662782728,
      "grad_norm": 0.11547791831787842,
      "learning_rate": 7.531224152362183e-08,
      "loss": 0.8655,
      "step": 894
    },
    {
      "epoch": 4.9074708704592185,
      "grad_norm": 0.11917262127195702,
      "learning_rate": 6.619491603008676e-08,
      "loss": 0.8724,
      "step": 895
    },
    {
      "epoch": 4.9129540781357095,
      "grad_norm": 0.11681023683693044,
      "learning_rate": 5.766517743432953e-08,
      "loss": 0.8638,
      "step": 896
    },
    {
      "epoch": 4.9184372858122005,
      "grad_norm": 0.11279645464293689,
      "learning_rate": 4.9723151243106225e-08,
      "loss": 0.8528,
      "step": 897
    },
    {
      "epoch": 4.923920493488691,
      "grad_norm": 0.1138256632587849,
      "learning_rate": 4.236895431557031e-08,
      "loss": 0.8572,
      "step": 898
    },
    {
      "epoch": 4.929403701165182,
      "grad_norm": 0.11375846263511914,
      "learning_rate": 3.560269486154066e-08,
      "loss": 0.8391,
      "step": 899
    },
    {
      "epoch": 4.934886908841673,
      "grad_norm": 0.11717016335228325,
      "learning_rate": 2.9424472439911757e-08,
      "loss": 0.8752,
      "step": 900
    },
    {
      "epoch": 4.940370116518163,
      "grad_norm": 0.11269596219934125,
      "learning_rate": 2.3834377957183684e-08,
      "loss": 0.8491,
      "step": 901
    },
    {
      "epoch": 4.945853324194654,
      "grad_norm": 0.1158772425437692,
      "learning_rate": 1.8832493666125494e-08,
      "loss": 0.872,
      "step": 902
    },
    {
      "epoch": 4.951336531871145,
      "grad_norm": 0.9980363287282,
      "learning_rate": 1.4418893164585002e-08,
      "loss": 0.9242,
      "step": 903
    },
    {
      "epoch": 4.956819739547635,
      "grad_norm": 0.11610912640066366,
      "learning_rate": 1.0593641394369691e-08,
      "loss": 0.8648,
      "step": 904
    },
    {
      "epoch": 4.962302947224126,
      "grad_norm": 0.11625433625052747,
      "learning_rate": 7.356794640318576e-09,
      "loss": 0.857,
      "step": 905
    },
    {
      "epoch": 4.967786154900617,
      "grad_norm": 0.11721935619429466,
      "learning_rate": 4.708400529476187e-09,
      "loss": 0.85,
      "step": 906
    },
    {
      "epoch": 4.973269362577108,
      "grad_norm": 0.11503586503967776,
      "learning_rate": 2.648498030364266e-09,
      "loss": 0.8709,
      "step": 907
    },
    {
      "epoch": 4.978752570253598,
      "grad_norm": 0.11671324103566125,
      "learning_rate": 1.1771174524355388e-09,
      "loss": 0.8843,
      "step": 908
    },
    {
      "epoch": 4.984235777930089,
      "grad_norm": 0.11519765971232111,
      "learning_rate": 2.9428044562074265e-10,
      "loss": 0.862,
      "step": 909
    },
    {
      "epoch": 4.98971898560658,
      "grad_norm": 0.11510258089520117,
      "learning_rate": 0.0,
      "loss": 0.8606,
      "step": 910
    },
    {
      "epoch": 4.98971898560658,
      "step": 910,
      "total_flos": 2.1065883137322516e+19,
      "train_loss": 1.0237522615836216,
      "train_runtime": 109255.8386,
      "train_samples_per_second": 4.272,
      "train_steps_per_second": 0.008
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 910,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.1065883137322516e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}