| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 0, | |
| "global_step": 365, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0027397260273972603, | |
| "grad_norm": 0.34919899702072144, | |
| "learning_rate": 1e-05, | |
| "loss": 1.6546, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005479452054794521, | |
| "grad_norm": 0.35103511810302734, | |
| "learning_rate": 9.972602739726028e-06, | |
| "loss": 1.6666, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00821917808219178, | |
| "grad_norm": 0.3742208182811737, | |
| "learning_rate": 9.945205479452056e-06, | |
| "loss": 1.735, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.010958904109589041, | |
| "grad_norm": 0.38575077056884766, | |
| "learning_rate": 9.917808219178083e-06, | |
| "loss": 1.7269, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0136986301369863, | |
| "grad_norm": 0.361535906791687, | |
| "learning_rate": 9.89041095890411e-06, | |
| "loss": 1.6679, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01643835616438356, | |
| "grad_norm": 0.3729570209980011, | |
| "learning_rate": 9.863013698630138e-06, | |
| "loss": 1.6534, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.019178082191780823, | |
| "grad_norm": 0.3666209876537323, | |
| "learning_rate": 9.835616438356166e-06, | |
| "loss": 1.6515, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.021917808219178082, | |
| "grad_norm": 0.3736323118209839, | |
| "learning_rate": 9.808219178082193e-06, | |
| "loss": 1.6758, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.024657534246575342, | |
| "grad_norm": 0.3758656084537506, | |
| "learning_rate": 9.78082191780822e-06, | |
| "loss": 1.7094, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0273972602739726, | |
| "grad_norm": 0.37960362434387207, | |
| "learning_rate": 9.753424657534248e-06, | |
| "loss": 1.6872, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.030136986301369864, | |
| "grad_norm": 0.39145174622535706, | |
| "learning_rate": 9.726027397260275e-06, | |
| "loss": 1.7096, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03287671232876712, | |
| "grad_norm": 0.36531391739845276, | |
| "learning_rate": 9.698630136986303e-06, | |
| "loss": 1.6133, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03561643835616438, | |
| "grad_norm": 0.35493168234825134, | |
| "learning_rate": 9.67123287671233e-06, | |
| "loss": 1.5616, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.038356164383561646, | |
| "grad_norm": 0.3559548258781433, | |
| "learning_rate": 9.643835616438358e-06, | |
| "loss": 1.6144, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0410958904109589, | |
| "grad_norm": 0.339250773191452, | |
| "learning_rate": 9.616438356164385e-06, | |
| "loss": 1.5508, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.043835616438356165, | |
| "grad_norm": 0.31973907351493835, | |
| "learning_rate": 9.589041095890411e-06, | |
| "loss": 1.5232, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04657534246575343, | |
| "grad_norm": 0.31599804759025574, | |
| "learning_rate": 9.561643835616438e-06, | |
| "loss": 1.5018, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.049315068493150684, | |
| "grad_norm": 0.3060247004032135, | |
| "learning_rate": 9.534246575342466e-06, | |
| "loss": 1.5041, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.052054794520547946, | |
| "grad_norm": 0.2997976541519165, | |
| "learning_rate": 9.506849315068493e-06, | |
| "loss": 1.4911, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0547945205479452, | |
| "grad_norm": 0.32057899236679077, | |
| "learning_rate": 9.47945205479452e-06, | |
| "loss": 1.5419, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.057534246575342465, | |
| "grad_norm": 0.27622660994529724, | |
| "learning_rate": 9.452054794520548e-06, | |
| "loss": 1.4222, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06027397260273973, | |
| "grad_norm": 0.29135027527809143, | |
| "learning_rate": 9.424657534246576e-06, | |
| "loss": 1.435, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06301369863013699, | |
| "grad_norm": 0.26389017701148987, | |
| "learning_rate": 9.397260273972603e-06, | |
| "loss": 1.4286, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.06575342465753424, | |
| "grad_norm": 0.2440488189458847, | |
| "learning_rate": 9.36986301369863e-06, | |
| "loss": 1.3593, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0684931506849315, | |
| "grad_norm": 0.25608792901039124, | |
| "learning_rate": 9.342465753424658e-06, | |
| "loss": 1.4187, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07123287671232877, | |
| "grad_norm": 0.2347661852836609, | |
| "learning_rate": 9.315068493150685e-06, | |
| "loss": 1.3816, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.07397260273972603, | |
| "grad_norm": 0.2517848610877991, | |
| "learning_rate": 9.287671232876713e-06, | |
| "loss": 1.2992, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.07671232876712329, | |
| "grad_norm": 0.24535977840423584, | |
| "learning_rate": 9.26027397260274e-06, | |
| "loss": 1.4509, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.07945205479452055, | |
| "grad_norm": 0.23485970497131348, | |
| "learning_rate": 9.232876712328768e-06, | |
| "loss": 1.4156, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0821917808219178, | |
| "grad_norm": 0.24302135407924652, | |
| "learning_rate": 9.205479452054795e-06, | |
| "loss": 1.4287, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08493150684931507, | |
| "grad_norm": 0.20637400448322296, | |
| "learning_rate": 9.178082191780823e-06, | |
| "loss": 1.3158, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.08767123287671233, | |
| "grad_norm": 0.20781908929347992, | |
| "learning_rate": 9.15068493150685e-06, | |
| "loss": 1.3538, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09041095890410959, | |
| "grad_norm": 0.2159215360879898, | |
| "learning_rate": 9.123287671232878e-06, | |
| "loss": 1.4117, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.09315068493150686, | |
| "grad_norm": 0.21819128096103668, | |
| "learning_rate": 9.095890410958905e-06, | |
| "loss": 1.3963, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0958904109589041, | |
| "grad_norm": 0.2067425698041916, | |
| "learning_rate": 9.068493150684932e-06, | |
| "loss": 1.3363, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.09863013698630137, | |
| "grad_norm": 0.1994429975748062, | |
| "learning_rate": 9.04109589041096e-06, | |
| "loss": 1.3555, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.10136986301369863, | |
| "grad_norm": 0.1868618130683899, | |
| "learning_rate": 9.013698630136987e-06, | |
| "loss": 1.3077, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.10410958904109589, | |
| "grad_norm": 0.20949655771255493, | |
| "learning_rate": 8.986301369863015e-06, | |
| "loss": 1.3717, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.10684931506849316, | |
| "grad_norm": 0.17898470163345337, | |
| "learning_rate": 8.958904109589042e-06, | |
| "loss": 1.26, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1095890410958904, | |
| "grad_norm": 0.1768495738506317, | |
| "learning_rate": 8.93150684931507e-06, | |
| "loss": 1.262, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11232876712328767, | |
| "grad_norm": 0.1851891726255417, | |
| "learning_rate": 8.904109589041097e-06, | |
| "loss": 1.2827, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.11506849315068493, | |
| "grad_norm": 0.18280373513698578, | |
| "learning_rate": 8.876712328767125e-06, | |
| "loss": 1.3053, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1178082191780822, | |
| "grad_norm": 0.17816248536109924, | |
| "learning_rate": 8.849315068493152e-06, | |
| "loss": 1.2808, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.12054794520547946, | |
| "grad_norm": 0.17815659940242767, | |
| "learning_rate": 8.82191780821918e-06, | |
| "loss": 1.28, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1232876712328767, | |
| "grad_norm": 0.17877663671970367, | |
| "learning_rate": 8.794520547945207e-06, | |
| "loss": 1.3015, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.12602739726027398, | |
| "grad_norm": 0.18253077566623688, | |
| "learning_rate": 8.767123287671233e-06, | |
| "loss": 1.2905, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.12876712328767123, | |
| "grad_norm": 0.16359081864356995, | |
| "learning_rate": 8.73972602739726e-06, | |
| "loss": 1.2685, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.13150684931506848, | |
| "grad_norm": 0.19661103188991547, | |
| "learning_rate": 8.712328767123288e-06, | |
| "loss": 1.2991, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.13424657534246576, | |
| "grad_norm": 0.1715298444032669, | |
| "learning_rate": 8.684931506849315e-06, | |
| "loss": 1.2592, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.136986301369863, | |
| "grad_norm": 0.1640940010547638, | |
| "learning_rate": 8.657534246575343e-06, | |
| "loss": 1.2517, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13972602739726028, | |
| "grad_norm": 0.22945547103881836, | |
| "learning_rate": 8.63013698630137e-06, | |
| "loss": 1.3143, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.14246575342465753, | |
| "grad_norm": 0.1699182689189911, | |
| "learning_rate": 8.602739726027397e-06, | |
| "loss": 1.2361, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.14520547945205478, | |
| "grad_norm": 0.17926287651062012, | |
| "learning_rate": 8.575342465753425e-06, | |
| "loss": 1.189, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.14794520547945206, | |
| "grad_norm": 0.1543978899717331, | |
| "learning_rate": 8.547945205479454e-06, | |
| "loss": 1.2138, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.1506849315068493, | |
| "grad_norm": 0.14743275940418243, | |
| "learning_rate": 8.520547945205481e-06, | |
| "loss": 1.1915, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.15342465753424658, | |
| "grad_norm": 0.1501009166240692, | |
| "learning_rate": 8.493150684931507e-06, | |
| "loss": 1.1943, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.15616438356164383, | |
| "grad_norm": 0.15394991636276245, | |
| "learning_rate": 8.465753424657535e-06, | |
| "loss": 1.2293, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1589041095890411, | |
| "grad_norm": 0.22368541359901428, | |
| "learning_rate": 8.438356164383562e-06, | |
| "loss": 1.2566, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.16164383561643836, | |
| "grad_norm": 0.15473651885986328, | |
| "learning_rate": 8.41095890410959e-06, | |
| "loss": 1.236, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1643835616438356, | |
| "grad_norm": 0.14430241286754608, | |
| "learning_rate": 8.383561643835617e-06, | |
| "loss": 1.1609, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.16712328767123288, | |
| "grad_norm": 0.1418728530406952, | |
| "learning_rate": 8.356164383561644e-06, | |
| "loss": 1.1768, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.16986301369863013, | |
| "grad_norm": 0.1435127854347229, | |
| "learning_rate": 8.328767123287672e-06, | |
| "loss": 1.1784, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.1726027397260274, | |
| "grad_norm": 0.14738471806049347, | |
| "learning_rate": 8.3013698630137e-06, | |
| "loss": 1.1672, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.17534246575342466, | |
| "grad_norm": 0.16719698905944824, | |
| "learning_rate": 8.273972602739727e-06, | |
| "loss": 1.1992, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.1780821917808219, | |
| "grad_norm": 0.1518981009721756, | |
| "learning_rate": 8.246575342465754e-06, | |
| "loss": 1.1971, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.18082191780821918, | |
| "grad_norm": 0.1392986923456192, | |
| "learning_rate": 8.219178082191782e-06, | |
| "loss": 1.1768, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.18356164383561643, | |
| "grad_norm": 0.13890501856803894, | |
| "learning_rate": 8.19178082191781e-06, | |
| "loss": 1.1672, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.1863013698630137, | |
| "grad_norm": 0.1505006104707718, | |
| "learning_rate": 8.164383561643837e-06, | |
| "loss": 1.1564, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.18904109589041096, | |
| "grad_norm": 0.13556960225105286, | |
| "learning_rate": 8.136986301369864e-06, | |
| "loss": 1.1818, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.1917808219178082, | |
| "grad_norm": 0.13326483964920044, | |
| "learning_rate": 8.109589041095892e-06, | |
| "loss": 1.1334, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.19452054794520549, | |
| "grad_norm": 0.15403442084789276, | |
| "learning_rate": 8.082191780821919e-06, | |
| "loss": 1.2299, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.19726027397260273, | |
| "grad_norm": 0.22601965069770813, | |
| "learning_rate": 8.054794520547946e-06, | |
| "loss": 1.1585, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.13847771286964417, | |
| "learning_rate": 8.027397260273974e-06, | |
| "loss": 1.1656, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.20273972602739726, | |
| "grad_norm": 0.14086662232875824, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.149, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.2054794520547945, | |
| "grad_norm": 0.14320115745067596, | |
| "learning_rate": 7.972602739726027e-06, | |
| "loss": 1.1623, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.20821917808219179, | |
| "grad_norm": 0.15054814517498016, | |
| "learning_rate": 7.945205479452055e-06, | |
| "loss": 1.148, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.21095890410958903, | |
| "grad_norm": 0.1444290280342102, | |
| "learning_rate": 7.917808219178082e-06, | |
| "loss": 1.1139, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.2136986301369863, | |
| "grad_norm": 0.13065987825393677, | |
| "learning_rate": 7.89041095890411e-06, | |
| "loss": 1.1368, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.21643835616438356, | |
| "grad_norm": 0.14221784472465515, | |
| "learning_rate": 7.863013698630137e-06, | |
| "loss": 1.1211, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.2191780821917808, | |
| "grad_norm": 0.139826700091362, | |
| "learning_rate": 7.835616438356164e-06, | |
| "loss": 1.1495, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2219178082191781, | |
| "grad_norm": 0.13123205304145813, | |
| "learning_rate": 7.808219178082192e-06, | |
| "loss": 1.1235, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.22465753424657534, | |
| "grad_norm": 0.1347399801015854, | |
| "learning_rate": 7.78082191780822e-06, | |
| "loss": 1.0951, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.2273972602739726, | |
| "grad_norm": 0.12708094716072083, | |
| "learning_rate": 7.753424657534248e-06, | |
| "loss": 1.1141, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.23013698630136986, | |
| "grad_norm": 0.134403258562088, | |
| "learning_rate": 7.726027397260276e-06, | |
| "loss": 1.1198, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.2328767123287671, | |
| "grad_norm": 0.1293765902519226, | |
| "learning_rate": 7.698630136986302e-06, | |
| "loss": 1.074, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.2356164383561644, | |
| "grad_norm": 0.1433599442243576, | |
| "learning_rate": 7.671232876712329e-06, | |
| "loss": 1.1273, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.23835616438356164, | |
| "grad_norm": 0.14464539289474487, | |
| "learning_rate": 7.643835616438356e-06, | |
| "loss": 1.1404, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.2410958904109589, | |
| "grad_norm": 0.15533122420310974, | |
| "learning_rate": 7.616438356164384e-06, | |
| "loss": 1.1486, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.24383561643835616, | |
| "grad_norm": 0.13592854142189026, | |
| "learning_rate": 7.589041095890411e-06, | |
| "loss": 1.1036, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2465753424657534, | |
| "grad_norm": 0.14032554626464844, | |
| "learning_rate": 7.561643835616439e-06, | |
| "loss": 1.1046, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2493150684931507, | |
| "grad_norm": 0.13799989223480225, | |
| "learning_rate": 7.534246575342466e-06, | |
| "loss": 1.0813, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.25205479452054796, | |
| "grad_norm": 0.1336989402770996, | |
| "learning_rate": 7.506849315068494e-06, | |
| "loss": 1.0856, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2547945205479452, | |
| "grad_norm": 0.14835570752620697, | |
| "learning_rate": 7.479452054794521e-06, | |
| "loss": 1.098, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.25753424657534246, | |
| "grad_norm": 0.14027932286262512, | |
| "learning_rate": 7.452054794520549e-06, | |
| "loss": 1.1069, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2602739726027397, | |
| "grad_norm": 0.13933329284191132, | |
| "learning_rate": 7.424657534246575e-06, | |
| "loss": 1.0936, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.26301369863013696, | |
| "grad_norm": 0.15654663741588593, | |
| "learning_rate": 7.397260273972603e-06, | |
| "loss": 1.0935, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.26575342465753427, | |
| "grad_norm": 0.17175213992595673, | |
| "learning_rate": 7.36986301369863e-06, | |
| "loss": 1.0783, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.2684931506849315, | |
| "grad_norm": 0.14236947894096375, | |
| "learning_rate": 7.342465753424658e-06, | |
| "loss": 1.081, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.27123287671232876, | |
| "grad_norm": 0.18027353286743164, | |
| "learning_rate": 7.315068493150685e-06, | |
| "loss": 1.1059, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.273972602739726, | |
| "grad_norm": 0.16158424317836761, | |
| "learning_rate": 7.287671232876713e-06, | |
| "loss": 1.0841, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.27671232876712326, | |
| "grad_norm": 0.15303635597229004, | |
| "learning_rate": 7.260273972602741e-06, | |
| "loss": 1.116, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.27945205479452057, | |
| "grad_norm": 0.15861615538597107, | |
| "learning_rate": 7.232876712328768e-06, | |
| "loss": 1.1038, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.2821917808219178, | |
| "grad_norm": 0.17147208750247955, | |
| "learning_rate": 7.205479452054796e-06, | |
| "loss": 1.115, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.28493150684931506, | |
| "grad_norm": 0.1569274365901947, | |
| "learning_rate": 7.178082191780823e-06, | |
| "loss": 1.0547, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.2876712328767123, | |
| "grad_norm": 0.153569296002388, | |
| "learning_rate": 7.15068493150685e-06, | |
| "loss": 1.0811, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.29041095890410956, | |
| "grad_norm": 0.14510026574134827, | |
| "learning_rate": 7.123287671232877e-06, | |
| "loss": 1.0283, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.29315068493150687, | |
| "grad_norm": 0.15047655999660492, | |
| "learning_rate": 7.095890410958905e-06, | |
| "loss": 1.0623, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.2958904109589041, | |
| "grad_norm": 0.15017667412757874, | |
| "learning_rate": 7.068493150684932e-06, | |
| "loss": 1.0856, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.29863013698630136, | |
| "grad_norm": 0.1588776707649231, | |
| "learning_rate": 7.0410958904109596e-06, | |
| "loss": 1.1063, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.3013698630136986, | |
| "grad_norm": 0.15009823441505432, | |
| "learning_rate": 7.013698630136987e-06, | |
| "loss": 1.063, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3041095890410959, | |
| "grad_norm": 0.15306375920772552, | |
| "learning_rate": 6.9863013698630145e-06, | |
| "loss": 1.0374, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.30684931506849317, | |
| "grad_norm": 0.15214037895202637, | |
| "learning_rate": 6.958904109589042e-06, | |
| "loss": 1.0603, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.3095890410958904, | |
| "grad_norm": 0.15497593581676483, | |
| "learning_rate": 6.931506849315069e-06, | |
| "loss": 1.0633, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.31232876712328766, | |
| "grad_norm": 0.1648060530424118, | |
| "learning_rate": 6.904109589041097e-06, | |
| "loss": 1.0393, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.3150684931506849, | |
| "grad_norm": 0.21841637790203094, | |
| "learning_rate": 6.876712328767123e-06, | |
| "loss": 1.0306, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.3178082191780822, | |
| "grad_norm": 0.1690291315317154, | |
| "learning_rate": 6.849315068493151e-06, | |
| "loss": 1.0928, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.32054794520547947, | |
| "grad_norm": 0.15783993899822235, | |
| "learning_rate": 6.821917808219178e-06, | |
| "loss": 1.0192, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.3232876712328767, | |
| "grad_norm": 0.15355956554412842, | |
| "learning_rate": 6.794520547945206e-06, | |
| "loss": 1.0457, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.32602739726027397, | |
| "grad_norm": 0.17985540628433228, | |
| "learning_rate": 6.767123287671233e-06, | |
| "loss": 1.0741, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.3287671232876712, | |
| "grad_norm": 0.1796933263540268, | |
| "learning_rate": 6.739726027397261e-06, | |
| "loss": 1.0805, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3315068493150685, | |
| "grad_norm": 0.17129400372505188, | |
| "learning_rate": 6.712328767123288e-06, | |
| "loss": 1.0443, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.33424657534246577, | |
| "grad_norm": 0.17514698207378387, | |
| "learning_rate": 6.684931506849316e-06, | |
| "loss": 1.0465, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.336986301369863, | |
| "grad_norm": 0.17114493250846863, | |
| "learning_rate": 6.657534246575343e-06, | |
| "loss": 1.0575, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.33972602739726027, | |
| "grad_norm": 0.1876526176929474, | |
| "learning_rate": 6.630136986301371e-06, | |
| "loss": 1.0671, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.3424657534246575, | |
| "grad_norm": 0.1693662852048874, | |
| "learning_rate": 6.602739726027397e-06, | |
| "loss": 1.0336, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3452054794520548, | |
| "grad_norm": 0.22197027504444122, | |
| "learning_rate": 6.5753424657534245e-06, | |
| "loss": 1.0381, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.34794520547945207, | |
| "grad_norm": 0.15992571413516998, | |
| "learning_rate": 6.547945205479452e-06, | |
| "loss": 1.0552, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.3506849315068493, | |
| "grad_norm": 0.19459573924541473, | |
| "learning_rate": 6.5205479452054794e-06, | |
| "loss": 1.0716, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.35342465753424657, | |
| "grad_norm": 0.16909852623939514, | |
| "learning_rate": 6.493150684931508e-06, | |
| "loss": 1.0082, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.3561643835616438, | |
| "grad_norm": 0.18583637475967407, | |
| "learning_rate": 6.465753424657535e-06, | |
| "loss": 1.0846, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3589041095890411, | |
| "grad_norm": 0.16531455516815186, | |
| "learning_rate": 6.438356164383563e-06, | |
| "loss": 1.0579, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.36164383561643837, | |
| "grad_norm": 0.15629485249519348, | |
| "learning_rate": 6.41095890410959e-06, | |
| "loss": 1.041, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.3643835616438356, | |
| "grad_norm": 0.15893372893333435, | |
| "learning_rate": 6.3835616438356175e-06, | |
| "loss": 1.0363, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.36712328767123287, | |
| "grad_norm": 0.17731238901615143, | |
| "learning_rate": 6.356164383561645e-06, | |
| "loss": 1.0379, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3698630136986301, | |
| "grad_norm": 0.20481380820274353, | |
| "learning_rate": 6.328767123287672e-06, | |
| "loss": 1.0506, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3726027397260274, | |
| "grad_norm": 0.15704141557216644, | |
| "learning_rate": 6.301369863013699e-06, | |
| "loss": 1.0326, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.37534246575342467, | |
| "grad_norm": 0.17464356124401093, | |
| "learning_rate": 6.2739726027397265e-06, | |
| "loss": 1.0506, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.3780821917808219, | |
| "grad_norm": 0.15756955742835999, | |
| "learning_rate": 6.246575342465754e-06, | |
| "loss": 1.0493, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.38082191780821917, | |
| "grad_norm": 0.14617951214313507, | |
| "learning_rate": 6.219178082191781e-06, | |
| "loss": 1.0398, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.3835616438356164, | |
| "grad_norm": 0.1518000364303589, | |
| "learning_rate": 6.191780821917809e-06, | |
| "loss": 1.029, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3863013698630137, | |
| "grad_norm": 0.15911860764026642, | |
| "learning_rate": 6.164383561643836e-06, | |
| "loss": 1.0632, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.38904109589041097, | |
| "grad_norm": 0.14329439401626587, | |
| "learning_rate": 6.136986301369864e-06, | |
| "loss": 0.9982, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.3917808219178082, | |
| "grad_norm": 0.14678213000297546, | |
| "learning_rate": 6.109589041095891e-06, | |
| "loss": 1.0269, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.39452054794520547, | |
| "grad_norm": 0.1700473427772522, | |
| "learning_rate": 6.082191780821919e-06, | |
| "loss": 1.0589, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.3972602739726027, | |
| "grad_norm": 0.15550148487091064, | |
| "learning_rate": 6.054794520547945e-06, | |
| "loss": 1.0175, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.1483745574951172, | |
| "learning_rate": 6.027397260273973e-06, | |
| "loss": 1.0354, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.40273972602739727, | |
| "grad_norm": 0.14025282859802246, | |
| "learning_rate": 6e-06, | |
| "loss": 0.9656, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.4054794520547945, | |
| "grad_norm": 0.15070085227489471, | |
| "learning_rate": 5.972602739726028e-06, | |
| "loss": 1.0103, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.40821917808219177, | |
| "grad_norm": 0.16396139562129974, | |
| "learning_rate": 5.945205479452055e-06, | |
| "loss": 1.0236, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.410958904109589, | |
| "grad_norm": 0.14801575243473053, | |
| "learning_rate": 5.9178082191780825e-06, | |
| "loss": 1.0818, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4136986301369863, | |
| "grad_norm": 0.15337041020393372, | |
| "learning_rate": 5.89041095890411e-06, | |
| "loss": 1.041, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.41643835616438357, | |
| "grad_norm": 0.1659669727087021, | |
| "learning_rate": 5.863013698630137e-06, | |
| "loss": 0.9632, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.4191780821917808, | |
| "grad_norm": 0.19434167444705963, | |
| "learning_rate": 5.835616438356166e-06, | |
| "loss": 1.0365, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.42191780821917807, | |
| "grad_norm": 0.1424913853406906, | |
| "learning_rate": 5.8082191780821915e-06, | |
| "loss": 1.0185, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.4246575342465753, | |
| "grad_norm": 0.159869983792305, | |
| "learning_rate": 5.780821917808219e-06, | |
| "loss": 1.0568, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.4273972602739726, | |
| "grad_norm": 0.14181528985500336, | |
| "learning_rate": 5.753424657534246e-06, | |
| "loss": 1.0098, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.4301369863013699, | |
| "grad_norm": 0.17228105664253235, | |
| "learning_rate": 5.726027397260274e-06, | |
| "loss": 1.0654, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.4328767123287671, | |
| "grad_norm": 0.15940870344638824, | |
| "learning_rate": 5.698630136986302e-06, | |
| "loss": 1.0354, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.43561643835616437, | |
| "grad_norm": 0.1820170134305954, | |
| "learning_rate": 5.6712328767123296e-06, | |
| "loss": 1.0184, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.4383561643835616, | |
| "grad_norm": 0.14352162182331085, | |
| "learning_rate": 5.643835616438357e-06, | |
| "loss": 1.0303, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.4410958904109589, | |
| "grad_norm": 0.14085696637630463, | |
| "learning_rate": 5.6164383561643845e-06, | |
| "loss": 1.0275, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.4438356164383562, | |
| "grad_norm": 0.14605316519737244, | |
| "learning_rate": 5.589041095890412e-06, | |
| "loss": 1.0231, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.4465753424657534, | |
| "grad_norm": 0.137510746717453, | |
| "learning_rate": 5.561643835616439e-06, | |
| "loss": 0.9733, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.44931506849315067, | |
| "grad_norm": 0.1410391479730606, | |
| "learning_rate": 5.534246575342466e-06, | |
| "loss": 1.0079, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.4520547945205479, | |
| "grad_norm": 0.17175598442554474, | |
| "learning_rate": 5.506849315068493e-06, | |
| "loss": 1.0035, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.4547945205479452, | |
| "grad_norm": 0.14537735283374786, | |
| "learning_rate": 5.479452054794521e-06, | |
| "loss": 1.0061, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.4575342465753425, | |
| "grad_norm": 0.2274583876132965, | |
| "learning_rate": 5.452054794520548e-06, | |
| "loss": 1.0426, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.4602739726027397, | |
| "grad_norm": 0.14900068938732147, | |
| "learning_rate": 5.424657534246576e-06, | |
| "loss": 1.0255, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.46301369863013697, | |
| "grad_norm": 0.14231647551059723, | |
| "learning_rate": 5.397260273972603e-06, | |
| "loss": 1.0428, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.4657534246575342, | |
| "grad_norm": 0.1493074893951416, | |
| "learning_rate": 5.369863013698631e-06, | |
| "loss": 0.9676, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.4684931506849315, | |
| "grad_norm": 0.18627113103866577, | |
| "learning_rate": 5.342465753424658e-06, | |
| "loss": 1.009, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.4712328767123288, | |
| "grad_norm": 0.16311538219451904, | |
| "learning_rate": 5.3150684931506856e-06, | |
| "loss": 1.0585, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.473972602739726, | |
| "grad_norm": 0.15902486443519592, | |
| "learning_rate": 5.287671232876713e-06, | |
| "loss": 0.987, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.4767123287671233, | |
| "grad_norm": 0.1851184368133545, | |
| "learning_rate": 5.26027397260274e-06, | |
| "loss": 1.0593, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.4794520547945205, | |
| "grad_norm": 0.14445023238658905, | |
| "learning_rate": 5.232876712328767e-06, | |
| "loss": 1.0439, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4821917808219178, | |
| "grad_norm": 0.16020196676254272, | |
| "learning_rate": 5.2054794520547945e-06, | |
| "loss": 1.0296, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.4849315068493151, | |
| "grad_norm": 0.13849924504756927, | |
| "learning_rate": 5.178082191780822e-06, | |
| "loss": 1.0116, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.4876712328767123, | |
| "grad_norm": 0.13803769648075104, | |
| "learning_rate": 5.1506849315068494e-06, | |
| "loss": 0.9989, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.4904109589041096, | |
| "grad_norm": 0.24443377554416656, | |
| "learning_rate": 5.123287671232877e-06, | |
| "loss": 0.9311, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.4931506849315068, | |
| "grad_norm": 0.18483811616897583, | |
| "learning_rate": 5.095890410958904e-06, | |
| "loss": 0.965, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4958904109589041, | |
| "grad_norm": 0.13710997998714447, | |
| "learning_rate": 5.068493150684932e-06, | |
| "loss": 0.9735, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.4986301369863014, | |
| "grad_norm": 0.1499159336090088, | |
| "learning_rate": 5.04109589041096e-06, | |
| "loss": 1.0204, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.5013698630136987, | |
| "grad_norm": 0.1443200260400772, | |
| "learning_rate": 5.0136986301369875e-06, | |
| "loss": 0.9589, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.5041095890410959, | |
| "grad_norm": 0.14708077907562256, | |
| "learning_rate": 4.986301369863014e-06, | |
| "loss": 0.9893, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5068493150684932, | |
| "grad_norm": 0.1446971595287323, | |
| "learning_rate": 4.958904109589042e-06, | |
| "loss": 1.0013, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5095890410958904, | |
| "grad_norm": 0.14119097590446472, | |
| "learning_rate": 4.931506849315069e-06, | |
| "loss": 1.0061, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.5123287671232877, | |
| "grad_norm": 0.14687605202198029, | |
| "learning_rate": 4.9041095890410965e-06, | |
| "loss": 1.0259, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.5150684931506849, | |
| "grad_norm": 0.14031365513801575, | |
| "learning_rate": 4.876712328767124e-06, | |
| "loss": 1.0262, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5178082191780822, | |
| "grad_norm": 0.1647661030292511, | |
| "learning_rate": 4.849315068493151e-06, | |
| "loss": 1.0225, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.5205479452054794, | |
| "grad_norm": 0.14606985449790955, | |
| "learning_rate": 4.821917808219179e-06, | |
| "loss": 0.9771, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5232876712328767, | |
| "grad_norm": 0.15172745287418365, | |
| "learning_rate": 4.7945205479452054e-06, | |
| "loss": 1.0252, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.5260273972602739, | |
| "grad_norm": 0.1375645399093628, | |
| "learning_rate": 4.767123287671233e-06, | |
| "loss": 1.0064, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.5287671232876713, | |
| "grad_norm": 0.1723964810371399, | |
| "learning_rate": 4.73972602739726e-06, | |
| "loss": 0.9935, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.5315068493150685, | |
| "grad_norm": 0.1510409563779831, | |
| "learning_rate": 4.712328767123288e-06, | |
| "loss": 1.0253, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.5342465753424658, | |
| "grad_norm": 0.14847567677497864, | |
| "learning_rate": 4.684931506849315e-06, | |
| "loss": 0.9796, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.536986301369863, | |
| "grad_norm": 0.14238575100898743, | |
| "learning_rate": 4.657534246575343e-06, | |
| "loss": 0.9835, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.5397260273972603, | |
| "grad_norm": 0.15265069901943207, | |
| "learning_rate": 4.63013698630137e-06, | |
| "loss": 1.0312, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.5424657534246575, | |
| "grad_norm": 0.13926750421524048, | |
| "learning_rate": 4.602739726027398e-06, | |
| "loss": 0.9606, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.5452054794520548, | |
| "grad_norm": 0.1447301059961319, | |
| "learning_rate": 4.575342465753425e-06, | |
| "loss": 0.9946, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.547945205479452, | |
| "grad_norm": 0.18391898274421692, | |
| "learning_rate": 4.5479452054794525e-06, | |
| "loss": 0.9846, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5506849315068493, | |
| "grad_norm": 0.13554824888706207, | |
| "learning_rate": 4.52054794520548e-06, | |
| "loss": 1.0005, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.5534246575342465, | |
| "grad_norm": 0.14264196157455444, | |
| "learning_rate": 4.493150684931507e-06, | |
| "loss": 0.9962, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.5561643835616439, | |
| "grad_norm": 0.14450880885124207, | |
| "learning_rate": 4.465753424657535e-06, | |
| "loss": 0.9783, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.5589041095890411, | |
| "grad_norm": 0.14610551297664642, | |
| "learning_rate": 4.438356164383562e-06, | |
| "loss": 1.0051, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.5616438356164384, | |
| "grad_norm": 0.1378694474697113, | |
| "learning_rate": 4.41095890410959e-06, | |
| "loss": 0.9696, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.5643835616438356, | |
| "grad_norm": 0.14404141902923584, | |
| "learning_rate": 4.383561643835616e-06, | |
| "loss": 0.9924, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.5671232876712329, | |
| "grad_norm": 0.2318737804889679, | |
| "learning_rate": 4.356164383561644e-06, | |
| "loss": 0.9768, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.5698630136986301, | |
| "grad_norm": 0.14497965574264526, | |
| "learning_rate": 4.328767123287671e-06, | |
| "loss": 0.9792, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.5726027397260274, | |
| "grad_norm": 0.2267143577337265, | |
| "learning_rate": 4.301369863013699e-06, | |
| "loss": 0.9884, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.5753424657534246, | |
| "grad_norm": 0.14338290691375732, | |
| "learning_rate": 4.273972602739727e-06, | |
| "loss": 0.9647, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.5780821917808219, | |
| "grad_norm": 0.1473742127418518, | |
| "learning_rate": 4.246575342465754e-06, | |
| "loss": 1.0206, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.5808219178082191, | |
| "grad_norm": 0.18932686746120453, | |
| "learning_rate": 4.219178082191781e-06, | |
| "loss": 0.9899, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.5835616438356165, | |
| "grad_norm": 0.179109126329422, | |
| "learning_rate": 4.1917808219178085e-06, | |
| "loss": 0.9942, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.5863013698630137, | |
| "grad_norm": 0.16374213993549347, | |
| "learning_rate": 4.164383561643836e-06, | |
| "loss": 0.987, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.589041095890411, | |
| "grad_norm": 0.15565191209316254, | |
| "learning_rate": 4.136986301369863e-06, | |
| "loss": 1.0136, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.5917808219178082, | |
| "grad_norm": 0.15738491714000702, | |
| "learning_rate": 4.109589041095891e-06, | |
| "loss": 0.9797, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.5945205479452055, | |
| "grad_norm": 0.1617341786623001, | |
| "learning_rate": 4.082191780821918e-06, | |
| "loss": 1.004, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.5972602739726027, | |
| "grad_norm": 0.15519091486930847, | |
| "learning_rate": 4.054794520547946e-06, | |
| "loss": 0.9476, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.14793817698955536, | |
| "learning_rate": 4.027397260273973e-06, | |
| "loss": 0.9594, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.6027397260273972, | |
| "grad_norm": 0.1568097174167633, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.9404, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6054794520547945, | |
| "grad_norm": 0.1501924991607666, | |
| "learning_rate": 3.972602739726027e-06, | |
| "loss": 0.9825, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.6082191780821918, | |
| "grad_norm": 0.2527293562889099, | |
| "learning_rate": 3.945205479452055e-06, | |
| "loss": 0.9469, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.6109589041095891, | |
| "grad_norm": 0.2237492948770523, | |
| "learning_rate": 3.917808219178082e-06, | |
| "loss": 0.9647, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.6136986301369863, | |
| "grad_norm": 0.1720942109823227, | |
| "learning_rate": 3.89041095890411e-06, | |
| "loss": 0.9707, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.6164383561643836, | |
| "grad_norm": 0.17925786972045898, | |
| "learning_rate": 3.863013698630138e-06, | |
| "loss": 0.9851, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6191780821917808, | |
| "grad_norm": 0.185493603348732, | |
| "learning_rate": 3.8356164383561645e-06, | |
| "loss": 0.9851, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.6219178082191781, | |
| "grad_norm": 0.15975800156593323, | |
| "learning_rate": 3.808219178082192e-06, | |
| "loss": 0.9983, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.6246575342465753, | |
| "grad_norm": 0.1643008291721344, | |
| "learning_rate": 3.7808219178082194e-06, | |
| "loss": 0.9997, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.6273972602739726, | |
| "grad_norm": 0.15068687498569489, | |
| "learning_rate": 3.753424657534247e-06, | |
| "loss": 0.9795, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.6301369863013698, | |
| "grad_norm": 0.15405616164207458, | |
| "learning_rate": 3.7260273972602743e-06, | |
| "loss": 0.9906, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6328767123287671, | |
| "grad_norm": 0.15264445543289185, | |
| "learning_rate": 3.6986301369863014e-06, | |
| "loss": 1.0125, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.6356164383561644, | |
| "grad_norm": 0.14779290556907654, | |
| "learning_rate": 3.671232876712329e-06, | |
| "loss": 0.9795, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.6383561643835617, | |
| "grad_norm": 0.16402508318424225, | |
| "learning_rate": 3.6438356164383567e-06, | |
| "loss": 1.002, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.6410958904109589, | |
| "grad_norm": 0.15044072270393372, | |
| "learning_rate": 3.616438356164384e-06, | |
| "loss": 0.9679, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.6438356164383562, | |
| "grad_norm": 0.1484067738056183, | |
| "learning_rate": 3.5890410958904116e-06, | |
| "loss": 1.0121, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.6465753424657534, | |
| "grad_norm": 0.14875975251197815, | |
| "learning_rate": 3.5616438356164386e-06, | |
| "loss": 0.9647, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.6493150684931507, | |
| "grad_norm": 0.1453394740819931, | |
| "learning_rate": 3.534246575342466e-06, | |
| "loss": 0.9703, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.6520547945205479, | |
| "grad_norm": 0.17434316873550415, | |
| "learning_rate": 3.5068493150684935e-06, | |
| "loss": 0.9591, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.6547945205479452, | |
| "grad_norm": 0.1531129777431488, | |
| "learning_rate": 3.479452054794521e-06, | |
| "loss": 0.9752, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.6575342465753424, | |
| "grad_norm": 0.16097253561019897, | |
| "learning_rate": 3.4520547945205484e-06, | |
| "loss": 0.9915, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.6602739726027397, | |
| "grad_norm": 0.14477920532226562, | |
| "learning_rate": 3.4246575342465754e-06, | |
| "loss": 0.995, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.663013698630137, | |
| "grad_norm": 0.1426738202571869, | |
| "learning_rate": 3.397260273972603e-06, | |
| "loss": 0.9429, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.6657534246575343, | |
| "grad_norm": 0.16615210473537445, | |
| "learning_rate": 3.3698630136986303e-06, | |
| "loss": 0.9951, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.6684931506849315, | |
| "grad_norm": 0.1474033147096634, | |
| "learning_rate": 3.342465753424658e-06, | |
| "loss": 0.958, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.6712328767123288, | |
| "grad_norm": 0.14540505409240723, | |
| "learning_rate": 3.3150684931506857e-06, | |
| "loss": 0.9452, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.673972602739726, | |
| "grad_norm": 0.22462385892868042, | |
| "learning_rate": 3.2876712328767123e-06, | |
| "loss": 0.9503, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.6767123287671233, | |
| "grad_norm": 0.14661099016666412, | |
| "learning_rate": 3.2602739726027397e-06, | |
| "loss": 0.9807, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.6794520547945205, | |
| "grad_norm": 0.14879795908927917, | |
| "learning_rate": 3.2328767123287676e-06, | |
| "loss": 0.9578, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.6821917808219178, | |
| "grad_norm": 0.16298916935920715, | |
| "learning_rate": 3.205479452054795e-06, | |
| "loss": 0.9525, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.684931506849315, | |
| "grad_norm": 0.1545659750699997, | |
| "learning_rate": 3.1780821917808225e-06, | |
| "loss": 0.9884, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.6876712328767123, | |
| "grad_norm": 0.14649316668510437, | |
| "learning_rate": 3.1506849315068495e-06, | |
| "loss": 0.9831, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.6904109589041096, | |
| "grad_norm": 0.25621578097343445, | |
| "learning_rate": 3.123287671232877e-06, | |
| "loss": 1.0186, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.6931506849315069, | |
| "grad_norm": 0.17166805267333984, | |
| "learning_rate": 3.0958904109589044e-06, | |
| "loss": 0.976, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.6958904109589041, | |
| "grad_norm": 0.2713611125946045, | |
| "learning_rate": 3.068493150684932e-06, | |
| "loss": 1.0367, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.6986301369863014, | |
| "grad_norm": 0.21252557635307312, | |
| "learning_rate": 3.0410958904109593e-06, | |
| "loss": 0.9695, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.7013698630136986, | |
| "grad_norm": 0.15245412290096283, | |
| "learning_rate": 3.0136986301369864e-06, | |
| "loss": 0.903, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.7041095890410959, | |
| "grad_norm": 0.16625380516052246, | |
| "learning_rate": 2.986301369863014e-06, | |
| "loss": 1.0032, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.7068493150684931, | |
| "grad_norm": 0.1509459912776947, | |
| "learning_rate": 2.9589041095890413e-06, | |
| "loss": 0.9269, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.7095890410958904, | |
| "grad_norm": 0.1466090828180313, | |
| "learning_rate": 2.9315068493150687e-06, | |
| "loss": 0.9565, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.7123287671232876, | |
| "grad_norm": 0.14883840084075928, | |
| "learning_rate": 2.9041095890410957e-06, | |
| "loss": 0.9848, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.7150684931506849, | |
| "grad_norm": 0.16292427480220795, | |
| "learning_rate": 2.876712328767123e-06, | |
| "loss": 1.0061, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.7178082191780822, | |
| "grad_norm": 0.2424314171075821, | |
| "learning_rate": 2.849315068493151e-06, | |
| "loss": 0.9698, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.7205479452054795, | |
| "grad_norm": 0.15062043070793152, | |
| "learning_rate": 2.8219178082191785e-06, | |
| "loss": 0.9648, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.7232876712328767, | |
| "grad_norm": 0.14946803450584412, | |
| "learning_rate": 2.794520547945206e-06, | |
| "loss": 0.9775, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.726027397260274, | |
| "grad_norm": 0.15413372218608856, | |
| "learning_rate": 2.767123287671233e-06, | |
| "loss": 0.9674, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.7287671232876712, | |
| "grad_norm": 0.15046992897987366, | |
| "learning_rate": 2.7397260273972604e-06, | |
| "loss": 0.9858, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.7315068493150685, | |
| "grad_norm": 0.14373600482940674, | |
| "learning_rate": 2.712328767123288e-06, | |
| "loss": 0.9536, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.7342465753424657, | |
| "grad_norm": 0.14666618406772614, | |
| "learning_rate": 2.6849315068493153e-06, | |
| "loss": 0.9525, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.736986301369863, | |
| "grad_norm": 0.21461984515190125, | |
| "learning_rate": 2.6575342465753428e-06, | |
| "loss": 0.9445, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.7397260273972602, | |
| "grad_norm": 0.1491764336824417, | |
| "learning_rate": 2.63013698630137e-06, | |
| "loss": 0.9414, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.7424657534246575, | |
| "grad_norm": 0.14540155231952667, | |
| "learning_rate": 2.6027397260273973e-06, | |
| "loss": 0.9715, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.7452054794520548, | |
| "grad_norm": 0.14945320785045624, | |
| "learning_rate": 2.5753424657534247e-06, | |
| "loss": 0.9837, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.7479452054794521, | |
| "grad_norm": 0.2210550606250763, | |
| "learning_rate": 2.547945205479452e-06, | |
| "loss": 0.9103, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.7506849315068493, | |
| "grad_norm": 0.15028400719165802, | |
| "learning_rate": 2.52054794520548e-06, | |
| "loss": 0.9913, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.7534246575342466, | |
| "grad_norm": 0.1455719918012619, | |
| "learning_rate": 2.493150684931507e-06, | |
| "loss": 0.9497, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.7561643835616438, | |
| "grad_norm": 0.22942738234996796, | |
| "learning_rate": 2.4657534246575345e-06, | |
| "loss": 0.9701, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.7589041095890411, | |
| "grad_norm": 0.1671881526708603, | |
| "learning_rate": 2.438356164383562e-06, | |
| "loss": 0.9739, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.7616438356164383, | |
| "grad_norm": 0.15146781504154205, | |
| "learning_rate": 2.4109589041095894e-06, | |
| "loss": 0.9913, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.7643835616438356, | |
| "grad_norm": 0.1616286039352417, | |
| "learning_rate": 2.3835616438356164e-06, | |
| "loss": 0.9819, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.7671232876712328, | |
| "grad_norm": 0.15679773688316345, | |
| "learning_rate": 2.356164383561644e-06, | |
| "loss": 0.9558, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.7698630136986301, | |
| "grad_norm": 0.14924940466880798, | |
| "learning_rate": 2.3287671232876713e-06, | |
| "loss": 0.9881, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.7726027397260274, | |
| "grad_norm": 0.16193532943725586, | |
| "learning_rate": 2.301369863013699e-06, | |
| "loss": 0.9702, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.7753424657534247, | |
| "grad_norm": 0.17239750921726227, | |
| "learning_rate": 2.2739726027397262e-06, | |
| "loss": 0.9642, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.7780821917808219, | |
| "grad_norm": 0.14394605159759521, | |
| "learning_rate": 2.2465753424657537e-06, | |
| "loss": 0.928, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.7808219178082192, | |
| "grad_norm": 0.22770152986049652, | |
| "learning_rate": 2.219178082191781e-06, | |
| "loss": 0.9784, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.7835616438356164, | |
| "grad_norm": 0.15883322060108185, | |
| "learning_rate": 2.191780821917808e-06, | |
| "loss": 0.9588, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.7863013698630137, | |
| "grad_norm": 0.17964458465576172, | |
| "learning_rate": 2.1643835616438356e-06, | |
| "loss": 0.9669, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.7890410958904109, | |
| "grad_norm": 0.15230049192905426, | |
| "learning_rate": 2.1369863013698635e-06, | |
| "loss": 0.9646, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.7917808219178082, | |
| "grad_norm": 0.16122160851955414, | |
| "learning_rate": 2.1095890410958905e-06, | |
| "loss": 0.9775, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.7945205479452054, | |
| "grad_norm": 0.1844584345817566, | |
| "learning_rate": 2.082191780821918e-06, | |
| "loss": 0.9625, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.7972602739726027, | |
| "grad_norm": 0.14859682321548462, | |
| "learning_rate": 2.0547945205479454e-06, | |
| "loss": 0.9514, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.22821080684661865, | |
| "learning_rate": 2.027397260273973e-06, | |
| "loss": 1.0101, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.8027397260273973, | |
| "grad_norm": 0.28670504689216614, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.0066, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.8054794520547945, | |
| "grad_norm": 0.18800397217273712, | |
| "learning_rate": 1.9726027397260274e-06, | |
| "loss": 0.9509, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.8082191780821918, | |
| "grad_norm": 0.17831219732761383, | |
| "learning_rate": 1.945205479452055e-06, | |
| "loss": 0.9519, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.810958904109589, | |
| "grad_norm": 0.16340646147727966, | |
| "learning_rate": 1.9178082191780823e-06, | |
| "loss": 0.9664, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.8136986301369863, | |
| "grad_norm": 0.25938671827316284, | |
| "learning_rate": 1.8904109589041097e-06, | |
| "loss": 0.9929, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.8164383561643835, | |
| "grad_norm": 0.20670604705810547, | |
| "learning_rate": 1.8630136986301372e-06, | |
| "loss": 0.955, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.8191780821917808, | |
| "grad_norm": 0.16003543138504028, | |
| "learning_rate": 1.8356164383561644e-06, | |
| "loss": 0.9041, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.821917808219178, | |
| "grad_norm": 0.17774829268455505, | |
| "learning_rate": 1.808219178082192e-06, | |
| "loss": 0.9766, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8246575342465754, | |
| "grad_norm": 0.1567939966917038, | |
| "learning_rate": 1.7808219178082193e-06, | |
| "loss": 0.9925, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.8273972602739726, | |
| "grad_norm": 0.17240066826343536, | |
| "learning_rate": 1.7534246575342468e-06, | |
| "loss": 0.9758, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.8301369863013699, | |
| "grad_norm": 0.19978557527065277, | |
| "learning_rate": 1.7260273972602742e-06, | |
| "loss": 1.028, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.8328767123287671, | |
| "grad_norm": 0.2522045075893402, | |
| "learning_rate": 1.6986301369863014e-06, | |
| "loss": 0.9918, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.8356164383561644, | |
| "grad_norm": 0.15379253029823303, | |
| "learning_rate": 1.671232876712329e-06, | |
| "loss": 0.9816, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.8383561643835616, | |
| "grad_norm": 0.1589875966310501, | |
| "learning_rate": 1.6438356164383561e-06, | |
| "loss": 0.9529, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.8410958904109589, | |
| "grad_norm": 0.1517769694328308, | |
| "learning_rate": 1.6164383561643838e-06, | |
| "loss": 0.9073, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.8438356164383561, | |
| "grad_norm": 0.14468395709991455, | |
| "learning_rate": 1.5890410958904112e-06, | |
| "loss": 0.9402, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.8465753424657534, | |
| "grad_norm": 0.2625545561313629, | |
| "learning_rate": 1.5616438356164385e-06, | |
| "loss": 1.0263, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.8493150684931506, | |
| "grad_norm": 0.1752869337797165, | |
| "learning_rate": 1.534246575342466e-06, | |
| "loss": 0.9552, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.852054794520548, | |
| "grad_norm": 0.2208409607410431, | |
| "learning_rate": 1.5068493150684932e-06, | |
| "loss": 0.963, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.8547945205479452, | |
| "grad_norm": 0.148488387465477, | |
| "learning_rate": 1.4794520547945206e-06, | |
| "loss": 0.9358, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.8575342465753425, | |
| "grad_norm": 0.1534607708454132, | |
| "learning_rate": 1.4520547945205479e-06, | |
| "loss": 0.9557, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.8602739726027397, | |
| "grad_norm": 0.24213606119155884, | |
| "learning_rate": 1.4246575342465755e-06, | |
| "loss": 0.9558, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.863013698630137, | |
| "grad_norm": 0.1569008082151413, | |
| "learning_rate": 1.397260273972603e-06, | |
| "loss": 0.9572, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.8657534246575342, | |
| "grad_norm": 0.14338001608848572, | |
| "learning_rate": 1.3698630136986302e-06, | |
| "loss": 0.9624, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.8684931506849315, | |
| "grad_norm": 0.15615947544574738, | |
| "learning_rate": 1.3424657534246577e-06, | |
| "loss": 0.9637, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.8712328767123287, | |
| "grad_norm": 0.1520787477493286, | |
| "learning_rate": 1.315068493150685e-06, | |
| "loss": 0.9364, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.873972602739726, | |
| "grad_norm": 0.15411341190338135, | |
| "learning_rate": 1.2876712328767124e-06, | |
| "loss": 0.9946, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.8767123287671232, | |
| "grad_norm": 0.1493544578552246, | |
| "learning_rate": 1.26027397260274e-06, | |
| "loss": 0.9843, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.8794520547945206, | |
| "grad_norm": 0.17391565442085266, | |
| "learning_rate": 1.2328767123287673e-06, | |
| "loss": 0.9841, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.8821917808219178, | |
| "grad_norm": 0.15456560254096985, | |
| "learning_rate": 1.2054794520547947e-06, | |
| "loss": 0.9773, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.8849315068493151, | |
| "grad_norm": 0.14900638163089752, | |
| "learning_rate": 1.178082191780822e-06, | |
| "loss": 0.9829, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.8876712328767123, | |
| "grad_norm": 0.1501474678516388, | |
| "learning_rate": 1.1506849315068494e-06, | |
| "loss": 0.9348, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.8904109589041096, | |
| "grad_norm": 0.14925050735473633, | |
| "learning_rate": 1.1232876712328769e-06, | |
| "loss": 0.9322, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.8931506849315068, | |
| "grad_norm": 0.16502335667610168, | |
| "learning_rate": 1.095890410958904e-06, | |
| "loss": 0.9881, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.8958904109589041, | |
| "grad_norm": 0.24135267734527588, | |
| "learning_rate": 1.0684931506849318e-06, | |
| "loss": 0.9633, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.8986301369863013, | |
| "grad_norm": 0.15312564373016357, | |
| "learning_rate": 1.041095890410959e-06, | |
| "loss": 0.97, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.9013698630136986, | |
| "grad_norm": 0.15663985908031464, | |
| "learning_rate": 1.0136986301369864e-06, | |
| "loss": 0.9347, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.9041095890410958, | |
| "grad_norm": 0.1708153784275055, | |
| "learning_rate": 9.863013698630137e-07, | |
| "loss": 0.9524, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.9068493150684932, | |
| "grad_norm": 0.2211901694536209, | |
| "learning_rate": 9.589041095890411e-07, | |
| "loss": 0.9804, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.9095890410958904, | |
| "grad_norm": 0.16123130917549133, | |
| "learning_rate": 9.315068493150686e-07, | |
| "loss": 0.9688, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.9123287671232877, | |
| "grad_norm": 0.17667633295059204, | |
| "learning_rate": 9.04109589041096e-07, | |
| "loss": 0.9707, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.915068493150685, | |
| "grad_norm": 0.15364082157611847, | |
| "learning_rate": 8.767123287671234e-07, | |
| "loss": 0.991, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.9178082191780822, | |
| "grad_norm": 0.15868520736694336, | |
| "learning_rate": 8.493150684931507e-07, | |
| "loss": 0.9807, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.9205479452054794, | |
| "grad_norm": 0.1566782146692276, | |
| "learning_rate": 8.219178082191781e-07, | |
| "loss": 0.9354, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.9232876712328767, | |
| "grad_norm": 0.17358021438121796, | |
| "learning_rate": 7.945205479452056e-07, | |
| "loss": 0.9728, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.9260273972602739, | |
| "grad_norm": 0.20244623720645905, | |
| "learning_rate": 7.67123287671233e-07, | |
| "loss": 0.9716, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.9287671232876712, | |
| "grad_norm": 0.15061792731285095, | |
| "learning_rate": 7.397260273972603e-07, | |
| "loss": 0.9489, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.9315068493150684, | |
| "grad_norm": 0.1528923660516739, | |
| "learning_rate": 7.123287671232878e-07, | |
| "loss": 0.9975, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.9342465753424658, | |
| "grad_norm": 0.15784519910812378, | |
| "learning_rate": 6.849315068493151e-07, | |
| "loss": 0.9525, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.936986301369863, | |
| "grad_norm": 0.15129871666431427, | |
| "learning_rate": 6.575342465753425e-07, | |
| "loss": 0.9815, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.9397260273972603, | |
| "grad_norm": 0.18728306889533997, | |
| "learning_rate": 6.3013698630137e-07, | |
| "loss": 0.9315, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.9424657534246575, | |
| "grad_norm": 0.15884990990161896, | |
| "learning_rate": 6.027397260273974e-07, | |
| "loss": 0.9915, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.9452054794520548, | |
| "grad_norm": 0.16598787903785706, | |
| "learning_rate": 5.753424657534247e-07, | |
| "loss": 0.9847, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.947945205479452, | |
| "grad_norm": 0.15402452647686005, | |
| "learning_rate": 5.47945205479452e-07, | |
| "loss": 0.9558, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.9506849315068493, | |
| "grad_norm": 0.1826355904340744, | |
| "learning_rate": 5.205479452054795e-07, | |
| "loss": 0.9863, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.9534246575342465, | |
| "grad_norm": 0.15361925959587097, | |
| "learning_rate": 4.931506849315068e-07, | |
| "loss": 0.9806, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.9561643835616438, | |
| "grad_norm": 0.1521327793598175, | |
| "learning_rate": 4.657534246575343e-07, | |
| "loss": 0.966, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.958904109589041, | |
| "grad_norm": 0.15801185369491577, | |
| "learning_rate": 4.383561643835617e-07, | |
| "loss": 0.95, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.9616438356164384, | |
| "grad_norm": 0.14759479463100433, | |
| "learning_rate": 4.1095890410958903e-07, | |
| "loss": 0.9835, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.9643835616438357, | |
| "grad_norm": 0.15526413917541504, | |
| "learning_rate": 3.835616438356165e-07, | |
| "loss": 0.9359, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.9671232876712329, | |
| "grad_norm": 0.152072012424469, | |
| "learning_rate": 3.561643835616439e-07, | |
| "loss": 0.943, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.9698630136986301, | |
| "grad_norm": 0.14854729175567627, | |
| "learning_rate": 3.2876712328767123e-07, | |
| "loss": 0.9297, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.9726027397260274, | |
| "grad_norm": 0.1651182621717453, | |
| "learning_rate": 3.013698630136987e-07, | |
| "loss": 0.9714, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.9753424657534246, | |
| "grad_norm": 0.15493687987327576, | |
| "learning_rate": 2.73972602739726e-07, | |
| "loss": 0.9792, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.9780821917808219, | |
| "grad_norm": 0.16965237259864807, | |
| "learning_rate": 2.465753424657534e-07, | |
| "loss": 0.9676, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.9808219178082191, | |
| "grad_norm": 0.2606389820575714, | |
| "learning_rate": 2.1917808219178084e-07, | |
| "loss": 0.9751, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.9835616438356164, | |
| "grad_norm": 0.17312797904014587, | |
| "learning_rate": 1.9178082191780824e-07, | |
| "loss": 1.0047, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.9863013698630136, | |
| "grad_norm": 0.16929611563682556, | |
| "learning_rate": 1.6438356164383561e-07, | |
| "loss": 0.9507, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.989041095890411, | |
| "grad_norm": 0.16281668841838837, | |
| "learning_rate": 1.36986301369863e-07, | |
| "loss": 0.9427, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.9917808219178083, | |
| "grad_norm": 0.15664711594581604, | |
| "learning_rate": 1.0958904109589042e-07, | |
| "loss": 0.9845, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.9945205479452055, | |
| "grad_norm": 0.14752177894115448, | |
| "learning_rate": 8.219178082191781e-08, | |
| "loss": 0.9271, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.9972602739726028, | |
| "grad_norm": 0.1772214025259018, | |
| "learning_rate": 5.479452054794521e-08, | |
| "loss": 1.0036, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.1503649801015854, | |
| "learning_rate": 2.7397260273972606e-08, | |
| "loss": 0.9636, | |
| "step": 365 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 365, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 0, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.375240306981601e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |