diff --git "a/checkpoints/Qwen2.5-7B/babylm_reverse_control_10M_seed0/runs/checkpoint-1382/trainer_state.json" "b/checkpoints/Qwen2.5-7B/babylm_reverse_control_10M_seed0/runs/checkpoint-1382/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoints/Qwen2.5-7B/babylm_reverse_control_10M_seed0/runs/checkpoint-1382/trainer_state.json" @@ -0,0 +1,10811 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.0, + "eval_steps": 10, + "global_step": 1382, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.001447178002894356, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.5701, + "step": 1 + }, + { + "epoch": 0.002894356005788712, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.5998, + "step": 2 + }, + { + "epoch": 0.004341534008683068, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6083, + "step": 3 + }, + { + "epoch": 0.005788712011577424, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.5925, + "step": 4 + }, + { + "epoch": 0.00723589001447178, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6063, + "step": 5 + }, + { + "epoch": 0.008683068017366137, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6258, + "step": 6 + }, + { + "epoch": 0.010130246020260492, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6144, + "step": 7 + }, + { + "epoch": 0.011577424023154847, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6105, + "step": 8 + }, + { + "epoch": 0.013024602026049204, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6112, + "step": 9 + }, + { + "epoch": 0.01447178002894356, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.5906, + "step": 10 + }, + { + "epoch": 0.01447178002894356, + "eval_loss": 1.6181164979934692, + "eval_runtime": 44.1203, + "eval_samples_per_second": 22.665, + "eval_steps_per_second": 1.428, + "step": 10 + }, + { + "epoch": 0.015918958031837915, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6301, + "step": 11 + }, + { + "epoch": 0.017366136034732273, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6266, + "step": 12 + }, + { + "epoch": 0.01881331403762663, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6151, + "step": 13 + }, + { + "epoch": 0.020260492040520984, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.5979, + "step": 14 + }, + { + "epoch": 0.02170767004341534, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6384, + "step": 15 + }, + { + "epoch": 0.023154848046309694, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6084, + "step": 16 + }, + { + "epoch": 0.024602026049204053, + "grad_norm": 3.8417842388153076, + "learning_rate": 3.597122302158274e-08, + "loss": 1.6086, + "step": 17 + }, + { + "epoch": 0.02604920405209841, + "grad_norm": 4.410459041595459, + "learning_rate": 7.194244604316547e-08, + "loss": 1.6346, + "step": 18 + }, + { + "epoch": 0.027496382054992764, + "grad_norm": 4.410459041595459, + "learning_rate": 7.194244604316547e-08, + "loss": 1.6004, + "step": 19 + }, + { + "epoch": 0.02894356005788712, + "grad_norm": 4.134227752685547, + "learning_rate": 1.0791366906474822e-07, + "loss": 1.6269, + "step": 20 + }, + { + "epoch": 0.02894356005788712, + "eval_loss": 1.6180763244628906, + "eval_runtime": 43.9435, + "eval_samples_per_second": 22.757, + "eval_steps_per_second": 1.434, + "step": 20 + }, + { + "epoch": 0.030390738060781478, + "grad_norm": 4.5793657302856445, + "learning_rate": 1.4388489208633095e-07, + "loss": 1.6202, + "step": 21 + }, + { + "epoch": 0.03183791606367583, + "grad_norm": 4.321710586547852, + "learning_rate": 1.7985611510791368e-07, + "loss": 1.5832, + "step": 22 + }, + { + "epoch": 0.03328509406657019, + "grad_norm": 4.097082614898682, + "learning_rate": 2.1582733812949643e-07, + "loss": 1.5958, + "step": 23 + }, + { + "epoch": 0.03473227206946455, + "grad_norm": 3.6840298175811768, + "learning_rate": 2.5179856115107916e-07, + "loss": 1.6351, + "step": 24 + }, + { + "epoch": 0.0361794500723589, + "grad_norm": 4.154601097106934, + "learning_rate": 2.877697841726619e-07, + "loss": 1.625, + "step": 25 + }, + { + "epoch": 0.03762662807525326, + "grad_norm": 4.378636837005615, + "learning_rate": 3.237410071942446e-07, + "loss": 1.5812, + "step": 26 + }, + { + "epoch": 0.03907380607814761, + "grad_norm": 3.616579532623291, + "learning_rate": 3.5971223021582736e-07, + "loss": 1.5943, + "step": 27 + }, + { + "epoch": 0.04052098408104197, + "grad_norm": 3.8743185997009277, + "learning_rate": 3.956834532374101e-07, + "loss": 1.6289, + "step": 28 + }, + { + "epoch": 0.041968162083936326, + "grad_norm": 2.9247794151306152, + "learning_rate": 4.3165467625899287e-07, + "loss": 1.5669, + "step": 29 + }, + { + "epoch": 0.04341534008683068, + "grad_norm": 3.3514983654022217, + "learning_rate": 4.676258992805756e-07, + "loss": 1.5788, + "step": 30 + }, + { + "epoch": 0.04341534008683068, + "eval_loss": 1.5967938899993896, + "eval_runtime": 45.2286, + "eval_samples_per_second": 22.11, + "eval_steps_per_second": 1.393, + "step": 30 + }, + { + "epoch": 0.04486251808972504, + "grad_norm": 2.745694160461426, + "learning_rate": 5.035971223021583e-07, + "loss": 1.5663, + "step": 31 + }, + { + "epoch": 0.04630969609261939, + "grad_norm": 3.0562522411346436, + "learning_rate": 5.39568345323741e-07, + "loss": 1.5915, + "step": 32 + }, + { + "epoch": 0.04775687409551375, + "grad_norm": 2.5593438148498535, + "learning_rate": 5.755395683453238e-07, + "loss": 1.592, + "step": 33 + }, + { + "epoch": 0.049204052098408106, + "grad_norm": 1.8786447048187256, + "learning_rate": 6.115107913669066e-07, + "loss": 1.5743, + "step": 34 + }, + { + "epoch": 0.05065123010130246, + "grad_norm": 2.607635021209717, + "learning_rate": 6.474820143884893e-07, + "loss": 1.5429, + "step": 35 + }, + { + "epoch": 0.05209840810419682, + "grad_norm": 1.9432525634765625, + "learning_rate": 6.83453237410072e-07, + "loss": 1.5403, + "step": 36 + }, + { + "epoch": 0.053545586107091175, + "grad_norm": 2.1000545024871826, + "learning_rate": 7.194244604316547e-07, + "loss": 1.5746, + "step": 37 + }, + { + "epoch": 0.05499276410998553, + "grad_norm": 2.4229369163513184, + "learning_rate": 7.553956834532375e-07, + "loss": 1.5387, + "step": 38 + }, + { + "epoch": 0.056439942112879886, + "grad_norm": 2.3055121898651123, + "learning_rate": 7.913669064748202e-07, + "loss": 1.5532, + "step": 39 + }, + { + "epoch": 0.05788712011577424, + "grad_norm": 2.142834424972534, + "learning_rate": 8.27338129496403e-07, + "loss": 1.5351, + "step": 40 + }, + { + "epoch": 0.05788712011577424, + "eval_loss": 1.551975965499878, + "eval_runtime": 46.1843, + "eval_samples_per_second": 21.652, + "eval_steps_per_second": 1.364, + "step": 40 + }, + { + "epoch": 0.059334298118668596, + "grad_norm": 1.9450421333312988, + "learning_rate": 8.633093525179857e-07, + "loss": 1.531, + "step": 41 + }, + { + "epoch": 0.060781476121562955, + "grad_norm": 1.9668039083480835, + "learning_rate": 8.992805755395684e-07, + "loss": 1.5469, + "step": 42 + }, + { + "epoch": 0.06222865412445731, + "grad_norm": 1.598451018333435, + "learning_rate": 9.352517985611512e-07, + "loss": 1.5648, + "step": 43 + }, + { + "epoch": 0.06367583212735166, + "grad_norm": 2.1381263732910156, + "learning_rate": 9.71223021582734e-07, + "loss": 1.5438, + "step": 44 + }, + { + "epoch": 0.06512301013024602, + "grad_norm": 2.2356314659118652, + "learning_rate": 1.0071942446043167e-06, + "loss": 1.5343, + "step": 45 + }, + { + "epoch": 0.06657018813314038, + "grad_norm": 1.8879215717315674, + "learning_rate": 1.0431654676258993e-06, + "loss": 1.5145, + "step": 46 + }, + { + "epoch": 0.06801736613603473, + "grad_norm": 2.1284284591674805, + "learning_rate": 1.079136690647482e-06, + "loss": 1.5312, + "step": 47 + }, + { + "epoch": 0.0694645441389291, + "grad_norm": 1.5030293464660645, + "learning_rate": 1.115107913669065e-06, + "loss": 1.5105, + "step": 48 + }, + { + "epoch": 0.07091172214182344, + "grad_norm": 1.7201606035232544, + "learning_rate": 1.1510791366906476e-06, + "loss": 1.4986, + "step": 49 + }, + { + "epoch": 0.0723589001447178, + "grad_norm": 1.6409934759140015, + "learning_rate": 1.1870503597122303e-06, + "loss": 1.4671, + "step": 50 + }, + { + "epoch": 0.0723589001447178, + "eval_loss": 1.4901193380355835, + "eval_runtime": 46.4275, + "eval_samples_per_second": 21.539, + "eval_steps_per_second": 1.357, + "step": 50 + }, + { + "epoch": 0.07380607814761216, + "grad_norm": 1.4075820446014404, + "learning_rate": 1.2230215827338131e-06, + "loss": 1.4506, + "step": 51 + }, + { + "epoch": 0.07525325615050651, + "grad_norm": 2.0347142219543457, + "learning_rate": 1.2589928057553958e-06, + "loss": 1.465, + "step": 52 + }, + { + "epoch": 0.07670043415340087, + "grad_norm": 1.4260163307189941, + "learning_rate": 1.2949640287769785e-06, + "loss": 1.4381, + "step": 53 + }, + { + "epoch": 0.07814761215629522, + "grad_norm": 1.604688048362732, + "learning_rate": 1.3309352517985614e-06, + "loss": 1.4311, + "step": 54 + }, + { + "epoch": 0.07959479015918958, + "grad_norm": 2.3518781661987305, + "learning_rate": 1.366906474820144e-06, + "loss": 1.4383, + "step": 55 + }, + { + "epoch": 0.08104196816208394, + "grad_norm": 2.327068567276001, + "learning_rate": 1.4028776978417265e-06, + "loss": 1.3908, + "step": 56 + }, + { + "epoch": 0.0824891461649783, + "grad_norm": 2.4417808055877686, + "learning_rate": 1.4388489208633094e-06, + "loss": 1.3343, + "step": 57 + }, + { + "epoch": 0.08393632416787265, + "grad_norm": 1.875885009765625, + "learning_rate": 1.474820143884892e-06, + "loss": 1.3723, + "step": 58 + }, + { + "epoch": 0.085383502170767, + "grad_norm": 1.987950325012207, + "learning_rate": 1.510791366906475e-06, + "loss": 1.3837, + "step": 59 + }, + { + "epoch": 0.08683068017366136, + "grad_norm": 1.5646475553512573, + "learning_rate": 1.5467625899280579e-06, + "loss": 1.3243, + "step": 60 + }, + { + "epoch": 0.08683068017366136, + "eval_loss": 1.3704127073287964, + "eval_runtime": 46.5809, + "eval_samples_per_second": 21.468, + "eval_steps_per_second": 1.352, + "step": 60 + }, + { + "epoch": 0.08827785817655572, + "grad_norm": 2.0767595767974854, + "learning_rate": 1.5827338129496403e-06, + "loss": 1.3551, + "step": 61 + }, + { + "epoch": 0.08972503617945007, + "grad_norm": 1.3927065134048462, + "learning_rate": 1.618705035971223e-06, + "loss": 1.3531, + "step": 62 + }, + { + "epoch": 0.09117221418234443, + "grad_norm": 1.6953964233398438, + "learning_rate": 1.654676258992806e-06, + "loss": 1.3299, + "step": 63 + }, + { + "epoch": 0.09261939218523878, + "grad_norm": 1.5323591232299805, + "learning_rate": 1.6906474820143886e-06, + "loss": 1.3066, + "step": 64 + }, + { + "epoch": 0.09406657018813314, + "grad_norm": 2.1093645095825195, + "learning_rate": 1.7266187050359715e-06, + "loss": 1.298, + "step": 65 + }, + { + "epoch": 0.0955137481910275, + "grad_norm": 1.2841155529022217, + "learning_rate": 1.762589928057554e-06, + "loss": 1.2961, + "step": 66 + }, + { + "epoch": 0.09696092619392185, + "grad_norm": 1.851452350616455, + "learning_rate": 1.7985611510791368e-06, + "loss": 1.2741, + "step": 67 + }, + { + "epoch": 0.09840810419681621, + "grad_norm": 1.7876909971237183, + "learning_rate": 1.8345323741007195e-06, + "loss": 1.2702, + "step": 68 + }, + { + "epoch": 0.09985528219971057, + "grad_norm": 1.3599499464035034, + "learning_rate": 1.8705035971223024e-06, + "loss": 1.2968, + "step": 69 + }, + { + "epoch": 0.10130246020260492, + "grad_norm": 1.6877838373184204, + "learning_rate": 1.906474820143885e-06, + "loss": 1.2724, + "step": 70 + }, + { + "epoch": 0.10130246020260492, + "eval_loss": 1.2967768907546997, + "eval_runtime": 46.568, + "eval_samples_per_second": 21.474, + "eval_steps_per_second": 1.353, + "step": 70 + }, + { + "epoch": 0.10274963820549927, + "grad_norm": 1.7636157274246216, + "learning_rate": 1.942446043165468e-06, + "loss": 1.289, + "step": 71 + }, + { + "epoch": 0.10419681620839363, + "grad_norm": 2.3618972301483154, + "learning_rate": 1.9784172661870504e-06, + "loss": 1.2349, + "step": 72 + }, + { + "epoch": 0.10564399421128799, + "grad_norm": 1.9659570455551147, + "learning_rate": 2.0143884892086333e-06, + "loss": 1.2354, + "step": 73 + }, + { + "epoch": 0.10709117221418235, + "grad_norm": 1.390474796295166, + "learning_rate": 2.050359712230216e-06, + "loss": 1.2192, + "step": 74 + }, + { + "epoch": 0.1085383502170767, + "grad_norm": 1.752890706062317, + "learning_rate": 2.0863309352517987e-06, + "loss": 1.1888, + "step": 75 + }, + { + "epoch": 0.10998552821997105, + "grad_norm": 1.396960735321045, + "learning_rate": 2.1223021582733816e-06, + "loss": 1.2397, + "step": 76 + }, + { + "epoch": 0.11143270622286541, + "grad_norm": 2.1841070652008057, + "learning_rate": 2.158273381294964e-06, + "loss": 1.2354, + "step": 77 + }, + { + "epoch": 0.11287988422575977, + "grad_norm": 2.0736773014068604, + "learning_rate": 2.194244604316547e-06, + "loss": 1.242, + "step": 78 + }, + { + "epoch": 0.11432706222865413, + "grad_norm": 1.753383994102478, + "learning_rate": 2.23021582733813e-06, + "loss": 1.2465, + "step": 79 + }, + { + "epoch": 0.11577424023154848, + "grad_norm": 1.7572365999221802, + "learning_rate": 2.2661870503597123e-06, + "loss": 1.217, + "step": 80 + }, + { + "epoch": 0.11577424023154848, + "eval_loss": 1.2516987323760986, + "eval_runtime": 46.6247, + "eval_samples_per_second": 21.448, + "eval_steps_per_second": 1.351, + "step": 80 + }, + { + "epoch": 0.11722141823444283, + "grad_norm": 2.70068097114563, + "learning_rate": 2.302158273381295e-06, + "loss": 1.1746, + "step": 81 + }, + { + "epoch": 0.11866859623733719, + "grad_norm": 1.100277304649353, + "learning_rate": 2.3381294964028776e-06, + "loss": 1.1929, + "step": 82 + }, + { + "epoch": 0.12011577424023155, + "grad_norm": 1.406956672668457, + "learning_rate": 2.3741007194244605e-06, + "loss": 1.2003, + "step": 83 + }, + { + "epoch": 0.12156295224312591, + "grad_norm": 1.885254144668579, + "learning_rate": 2.4100719424460434e-06, + "loss": 1.1919, + "step": 84 + }, + { + "epoch": 0.12301013024602026, + "grad_norm": 1.995215654373169, + "learning_rate": 2.4460431654676263e-06, + "loss": 1.1618, + "step": 85 + }, + { + "epoch": 0.12445730824891461, + "grad_norm": 1.5441099405288696, + "learning_rate": 2.4820143884892088e-06, + "loss": 1.1932, + "step": 86 + }, + { + "epoch": 0.12590448625180897, + "grad_norm": 1.6617332696914673, + "learning_rate": 2.5179856115107916e-06, + "loss": 1.139, + "step": 87 + }, + { + "epoch": 0.12735166425470332, + "grad_norm": 1.7822656631469727, + "learning_rate": 2.5539568345323745e-06, + "loss": 1.163, + "step": 88 + }, + { + "epoch": 0.1287988422575977, + "grad_norm": 1.4938169717788696, + "learning_rate": 2.589928057553957e-06, + "loss": 1.1928, + "step": 89 + }, + { + "epoch": 0.13024602026049203, + "grad_norm": 1.1358822584152222, + "learning_rate": 2.6258992805755395e-06, + "loss": 1.1756, + "step": 90 + }, + { + "epoch": 0.13024602026049203, + "eval_loss": 1.221023440361023, + "eval_runtime": 46.6384, + "eval_samples_per_second": 21.442, + "eval_steps_per_second": 1.351, + "step": 90 + }, + { + "epoch": 0.1316931982633864, + "grad_norm": 1.6468782424926758, + "learning_rate": 2.6618705035971228e-06, + "loss": 1.1887, + "step": 91 + }, + { + "epoch": 0.13314037626628075, + "grad_norm": 1.4682163000106812, + "learning_rate": 2.6978417266187052e-06, + "loss": 1.1933, + "step": 92 + }, + { + "epoch": 0.1345875542691751, + "grad_norm": 2.167820453643799, + "learning_rate": 2.733812949640288e-06, + "loss": 1.1512, + "step": 93 + }, + { + "epoch": 0.13603473227206947, + "grad_norm": 1.514905333518982, + "learning_rate": 2.7697841726618706e-06, + "loss": 1.1596, + "step": 94 + }, + { + "epoch": 0.13748191027496381, + "grad_norm": 1.6251145601272583, + "learning_rate": 2.805755395683453e-06, + "loss": 1.1566, + "step": 95 + }, + { + "epoch": 0.1389290882778582, + "grad_norm": 1.5806950330734253, + "learning_rate": 2.8417266187050364e-06, + "loss": 1.1304, + "step": 96 + }, + { + "epoch": 0.14037626628075253, + "grad_norm": 1.9903422594070435, + "learning_rate": 2.877697841726619e-06, + "loss": 1.149, + "step": 97 + }, + { + "epoch": 0.14182344428364688, + "grad_norm": 2.0324182510375977, + "learning_rate": 2.9136690647482017e-06, + "loss": 1.1265, + "step": 98 + }, + { + "epoch": 0.14327062228654125, + "grad_norm": 1.4268184900283813, + "learning_rate": 2.949640287769784e-06, + "loss": 1.1679, + "step": 99 + }, + { + "epoch": 0.1447178002894356, + "grad_norm": 1.503604769706726, + "learning_rate": 2.985611510791367e-06, + "loss": 1.1713, + "step": 100 + }, + { + "epoch": 0.1447178002894356, + "eval_loss": 1.1963330507278442, + "eval_runtime": 46.6254, + "eval_samples_per_second": 21.448, + "eval_steps_per_second": 1.351, + "step": 100 + }, + { + "epoch": 0.14616497829232997, + "grad_norm": 1.7602556943893433, + "learning_rate": 3.02158273381295e-06, + "loss": 1.1161, + "step": 101 + }, + { + "epoch": 0.1476121562952243, + "grad_norm": 1.799340844154358, + "learning_rate": 3.0575539568345324e-06, + "loss": 1.1287, + "step": 102 + }, + { + "epoch": 0.14905933429811866, + "grad_norm": 1.734973669052124, + "learning_rate": 3.0935251798561158e-06, + "loss": 1.0941, + "step": 103 + }, + { + "epoch": 0.15050651230101303, + "grad_norm": 1.3803553581237793, + "learning_rate": 3.1294964028776982e-06, + "loss": 1.1175, + "step": 104 + }, + { + "epoch": 0.15195369030390737, + "grad_norm": 1.438547134399414, + "learning_rate": 3.1654676258992807e-06, + "loss": 1.1093, + "step": 105 + }, + { + "epoch": 0.15340086830680175, + "grad_norm": 1.7081354856491089, + "learning_rate": 3.2014388489208636e-06, + "loss": 1.1018, + "step": 106 + }, + { + "epoch": 0.1548480463096961, + "grad_norm": 1.9293290376663208, + "learning_rate": 3.237410071942446e-06, + "loss": 1.1169, + "step": 107 + }, + { + "epoch": 0.15629522431259044, + "grad_norm": 1.5548726320266724, + "learning_rate": 3.2733812949640294e-06, + "loss": 1.1053, + "step": 108 + }, + { + "epoch": 0.1577424023154848, + "grad_norm": 1.692173957824707, + "learning_rate": 3.309352517985612e-06, + "loss": 1.1262, + "step": 109 + }, + { + "epoch": 0.15918958031837915, + "grad_norm": 1.740042805671692, + "learning_rate": 3.3453237410071943e-06, + "loss": 1.0643, + "step": 110 + }, + { + "epoch": 0.15918958031837915, + "eval_loss": 1.1800063848495483, + "eval_runtime": 46.5606, + "eval_samples_per_second": 21.477, + "eval_steps_per_second": 1.353, + "step": 110 + }, + { + "epoch": 0.16063675832127353, + "grad_norm": 1.8296616077423096, + "learning_rate": 3.381294964028777e-06, + "loss": 1.1386, + "step": 111 + }, + { + "epoch": 0.16208393632416787, + "grad_norm": 1.8983434438705444, + "learning_rate": 3.4172661870503596e-06, + "loss": 1.1297, + "step": 112 + }, + { + "epoch": 0.16353111432706222, + "grad_norm": 2.53128719329834, + "learning_rate": 3.453237410071943e-06, + "loss": 1.1082, + "step": 113 + }, + { + "epoch": 0.1649782923299566, + "grad_norm": 2.0817222595214844, + "learning_rate": 3.4892086330935254e-06, + "loss": 1.1049, + "step": 114 + }, + { + "epoch": 0.16642547033285093, + "grad_norm": 1.7140759229660034, + "learning_rate": 3.525179856115108e-06, + "loss": 1.0806, + "step": 115 + }, + { + "epoch": 0.1678726483357453, + "grad_norm": 2.2356998920440674, + "learning_rate": 3.561151079136691e-06, + "loss": 1.0658, + "step": 116 + }, + { + "epoch": 0.16931982633863965, + "grad_norm": 1.7578060626983643, + "learning_rate": 3.5971223021582737e-06, + "loss": 1.1169, + "step": 117 + }, + { + "epoch": 0.170767004341534, + "grad_norm": 2.370487928390503, + "learning_rate": 3.6330935251798566e-06, + "loss": 1.0744, + "step": 118 + }, + { + "epoch": 0.17221418234442837, + "grad_norm": 1.9071413278579712, + "learning_rate": 3.669064748201439e-06, + "loss": 1.1055, + "step": 119 + }, + { + "epoch": 0.1736613603473227, + "grad_norm": 2.035879611968994, + "learning_rate": 3.7050359712230215e-06, + "loss": 1.0888, + "step": 120 + }, + { + "epoch": 0.1736613603473227, + "eval_loss": 1.158413052558899, + "eval_runtime": 46.6367, + "eval_samples_per_second": 21.442, + "eval_steps_per_second": 1.351, + "step": 120 + }, + { + "epoch": 0.17510853835021709, + "grad_norm": 1.5385394096374512, + "learning_rate": 3.741007194244605e-06, + "loss": 1.0862, + "step": 121 + }, + { + "epoch": 0.17655571635311143, + "grad_norm": 3.185743808746338, + "learning_rate": 3.7769784172661873e-06, + "loss": 1.079, + "step": 122 + }, + { + "epoch": 0.17800289435600578, + "grad_norm": 1.78337562084198, + "learning_rate": 3.81294964028777e-06, + "loss": 1.0481, + "step": 123 + }, + { + "epoch": 0.17945007235890015, + "grad_norm": 1.8080698251724243, + "learning_rate": 3.848920863309353e-06, + "loss": 1.08, + "step": 124 + }, + { + "epoch": 0.1808972503617945, + "grad_norm": 2.705418109893799, + "learning_rate": 3.884892086330936e-06, + "loss": 1.1043, + "step": 125 + }, + { + "epoch": 0.18234442836468887, + "grad_norm": 1.6752021312713623, + "learning_rate": 3.920863309352518e-06, + "loss": 1.0931, + "step": 126 + }, + { + "epoch": 0.1837916063675832, + "grad_norm": 2.0246944427490234, + "learning_rate": 3.956834532374101e-06, + "loss": 1.1056, + "step": 127 + }, + { + "epoch": 0.18523878437047755, + "grad_norm": 2.5710747241973877, + "learning_rate": 3.992805755395684e-06, + "loss": 1.0724, + "step": 128 + }, + { + "epoch": 0.18668596237337193, + "grad_norm": 1.8522069454193115, + "learning_rate": 4.028776978417267e-06, + "loss": 1.0777, + "step": 129 + }, + { + "epoch": 0.18813314037626627, + "grad_norm": 1.9311445951461792, + "learning_rate": 4.0647482014388495e-06, + "loss": 1.0667, + "step": 130 + }, + { + "epoch": 0.18813314037626627, + "eval_loss": 1.1432955265045166, + "eval_runtime": 46.5745, + "eval_samples_per_second": 21.471, + "eval_steps_per_second": 1.353, + "step": 130 + }, + { + "epoch": 0.18958031837916064, + "grad_norm": 1.5454154014587402, + "learning_rate": 4.100719424460432e-06, + "loss": 1.1082, + "step": 131 + }, + { + "epoch": 0.191027496382055, + "grad_norm": 1.5837955474853516, + "learning_rate": 4.1366906474820145e-06, + "loss": 1.061, + "step": 132 + }, + { + "epoch": 0.19247467438494936, + "grad_norm": 1.5605251789093018, + "learning_rate": 4.172661870503597e-06, + "loss": 1.1014, + "step": 133 + }, + { + "epoch": 0.1939218523878437, + "grad_norm": 1.5702067613601685, + "learning_rate": 4.20863309352518e-06, + "loss": 1.0939, + "step": 134 + }, + { + "epoch": 0.19536903039073805, + "grad_norm": 2.58363938331604, + "learning_rate": 4.244604316546763e-06, + "loss": 1.0712, + "step": 135 + }, + { + "epoch": 0.19681620839363242, + "grad_norm": 1.5577141046524048, + "learning_rate": 4.280575539568346e-06, + "loss": 1.0589, + "step": 136 + }, + { + "epoch": 0.19826338639652677, + "grad_norm": 2.346186876296997, + "learning_rate": 4.316546762589928e-06, + "loss": 1.042, + "step": 137 + }, + { + "epoch": 0.19971056439942114, + "grad_norm": 1.7034313678741455, + "learning_rate": 4.352517985611511e-06, + "loss": 1.0271, + "step": 138 + }, + { + "epoch": 0.2011577424023155, + "grad_norm": 2.714397668838501, + "learning_rate": 4.388489208633094e-06, + "loss": 1.0722, + "step": 139 + }, + { + "epoch": 0.20260492040520983, + "grad_norm": 1.451943278312683, + "learning_rate": 4.424460431654677e-06, + "loss": 1.0593, + "step": 140 + }, + { + "epoch": 0.20260492040520983, + "eval_loss": 1.132778525352478, + "eval_runtime": 46.6861, + "eval_samples_per_second": 21.42, + "eval_steps_per_second": 1.349, + "step": 140 + }, + { + "epoch": 0.2040520984081042, + "grad_norm": 2.1872525215148926, + "learning_rate": 4.46043165467626e-06, + "loss": 1.0364, + "step": 141 + }, + { + "epoch": 0.20549927641099855, + "grad_norm": 2.934126138687134, + "learning_rate": 4.496402877697842e-06, + "loss": 1.0503, + "step": 142 + }, + { + "epoch": 0.20694645441389292, + "grad_norm": 1.9569369554519653, + "learning_rate": 4.5323741007194245e-06, + "loss": 1.043, + "step": 143 + }, + { + "epoch": 0.20839363241678727, + "grad_norm": 2.095224618911743, + "learning_rate": 4.5683453237410074e-06, + "loss": 1.0404, + "step": 144 + }, + { + "epoch": 0.2098408104196816, + "grad_norm": 2.44909930229187, + "learning_rate": 4.60431654676259e-06, + "loss": 1.0725, + "step": 145 + }, + { + "epoch": 0.21128798842257598, + "grad_norm": 1.914057731628418, + "learning_rate": 4.640287769784173e-06, + "loss": 1.0639, + "step": 146 + }, + { + "epoch": 0.21273516642547033, + "grad_norm": 1.4629244804382324, + "learning_rate": 4.676258992805755e-06, + "loss": 1.0035, + "step": 147 + }, + { + "epoch": 0.2141823444283647, + "grad_norm": 2.2034757137298584, + "learning_rate": 4.712230215827339e-06, + "loss": 1.0548, + "step": 148 + }, + { + "epoch": 0.21562952243125905, + "grad_norm": 2.0527920722961426, + "learning_rate": 4.748201438848921e-06, + "loss": 1.0393, + "step": 149 + }, + { + "epoch": 0.2170767004341534, + "grad_norm": 2.0599184036254883, + "learning_rate": 4.784172661870504e-06, + "loss": 1.0556, + "step": 150 + }, + { + "epoch": 0.2170767004341534, + "eval_loss": 1.1204142570495605, + "eval_runtime": 46.5995, + "eval_samples_per_second": 21.459, + "eval_steps_per_second": 1.352, + "step": 150 + }, + { + "epoch": 0.21852387843704776, + "grad_norm": 1.6190012693405151, + "learning_rate": 4.820143884892087e-06, + "loss": 1.0486, + "step": 151 + }, + { + "epoch": 0.2199710564399421, + "grad_norm": 1.9229748249053955, + "learning_rate": 4.856115107913669e-06, + "loss": 0.9995, + "step": 152 + }, + { + "epoch": 0.22141823444283648, + "grad_norm": 2.443791627883911, + "learning_rate": 4.892086330935253e-06, + "loss": 1.024, + "step": 153 + }, + { + "epoch": 0.22286541244573083, + "grad_norm": 2.2911691665649414, + "learning_rate": 4.928057553956835e-06, + "loss": 1.0328, + "step": 154 + }, + { + "epoch": 0.22431259044862517, + "grad_norm": 2.4395852088928223, + "learning_rate": 4.9640287769784175e-06, + "loss": 1.0205, + "step": 155 + }, + { + "epoch": 0.22575976845151954, + "grad_norm": 2.147939920425415, + "learning_rate": 5e-06, + "loss": 1.041, + "step": 156 + }, + { + "epoch": 0.2272069464544139, + "grad_norm": 1.873216986656189, + "learning_rate": 4.9959774738535805e-06, + "loss": 1.0538, + "step": 157 + }, + { + "epoch": 0.22865412445730826, + "grad_norm": 1.6969950199127197, + "learning_rate": 4.991954947707161e-06, + "loss": 1.006, + "step": 158 + }, + { + "epoch": 0.2301013024602026, + "grad_norm": 2.0071520805358887, + "learning_rate": 4.987932421560741e-06, + "loss": 1.0242, + "step": 159 + }, + { + "epoch": 0.23154848046309695, + "grad_norm": 2.0993785858154297, + "learning_rate": 4.983909895414321e-06, + "loss": 1.0513, + "step": 160 + }, + { + "epoch": 0.23154848046309695, + "eval_loss": 1.1105519533157349, + "eval_runtime": 46.5817, + "eval_samples_per_second": 21.468, + "eval_steps_per_second": 1.352, + "step": 160 + }, + { + "epoch": 0.23299565846599132, + "grad_norm": 1.6196647882461548, + "learning_rate": 4.9798873692679e-06, + "loss": 1.0322, + "step": 161 + }, + { + "epoch": 0.23444283646888567, + "grad_norm": 1.8116074800491333, + "learning_rate": 4.97586484312148e-06, + "loss": 1.0009, + "step": 162 + }, + { + "epoch": 0.23589001447178004, + "grad_norm": 2.247441530227661, + "learning_rate": 4.9718423169750605e-06, + "loss": 1.0238, + "step": 163 + }, + { + "epoch": 0.23733719247467439, + "grad_norm": 1.5156170129776, + "learning_rate": 4.967819790828641e-06, + "loss": 1.0271, + "step": 164 + }, + { + "epoch": 0.23878437047756873, + "grad_norm": 2.117612361907959, + "learning_rate": 4.963797264682221e-06, + "loss": 0.9838, + "step": 165 + }, + { + "epoch": 0.2402315484804631, + "grad_norm": 2.2577571868896484, + "learning_rate": 4.959774738535801e-06, + "loss": 1.033, + "step": 166 + }, + { + "epoch": 0.24167872648335745, + "grad_norm": 1.7815032005310059, + "learning_rate": 4.955752212389381e-06, + "loss": 1.0035, + "step": 167 + }, + { + "epoch": 0.24312590448625182, + "grad_norm": 2.131831169128418, + "learning_rate": 4.951729686242961e-06, + "loss": 1.0305, + "step": 168 + }, + { + "epoch": 0.24457308248914617, + "grad_norm": 1.940930724143982, + "learning_rate": 4.947707160096541e-06, + "loss": 1.002, + "step": 169 + }, + { + "epoch": 0.2460202604920405, + "grad_norm": 2.0040535926818848, + "learning_rate": 4.943684633950121e-06, + "loss": 1.0343, + "step": 170 + }, + { + "epoch": 0.2460202604920405, + "eval_loss": 1.1006911993026733, + "eval_runtime": 46.615, + "eval_samples_per_second": 21.452, + "eval_steps_per_second": 1.351, + "step": 170 + }, + { + "epoch": 0.24746743849493488, + "grad_norm": 2.0221688747406006, + "learning_rate": 4.939662107803701e-06, + "loss": 1.048, + "step": 171 + }, + { + "epoch": 0.24891461649782923, + "grad_norm": 1.4428244829177856, + "learning_rate": 4.935639581657281e-06, + "loss": 1.0083, + "step": 172 + }, + { + "epoch": 0.2503617945007236, + "grad_norm": 2.2255241870880127, + "learning_rate": 4.931617055510861e-06, + "loss": 0.9821, + "step": 173 + }, + { + "epoch": 0.25180897250361794, + "grad_norm": 1.340147614479065, + "learning_rate": 4.927594529364441e-06, + "loss": 0.999, + "step": 174 + }, + { + "epoch": 0.2532561505065123, + "grad_norm": 2.459054946899414, + "learning_rate": 4.923572003218021e-06, + "loss": 1.0233, + "step": 175 + }, + { + "epoch": 0.25470332850940663, + "grad_norm": 2.362109899520874, + "learning_rate": 4.919549477071601e-06, + "loss": 0.9918, + "step": 176 + }, + { + "epoch": 0.25615050651230103, + "grad_norm": 2.200514554977417, + "learning_rate": 4.915526950925181e-06, + "loss": 1.044, + "step": 177 + }, + { + "epoch": 0.2575976845151954, + "grad_norm": 1.9201098680496216, + "learning_rate": 4.9115044247787615e-06, + "loss": 1.0, + "step": 178 + }, + { + "epoch": 0.2590448625180897, + "grad_norm": 2.2396862506866455, + "learning_rate": 4.907481898632342e-06, + "loss": 1.0066, + "step": 179 + }, + { + "epoch": 0.26049204052098407, + "grad_norm": 2.5237581729888916, + "learning_rate": 4.903459372485922e-06, + "loss": 1.0279, + "step": 180 + }, + { + "epoch": 0.26049204052098407, + "eval_loss": 1.0948052406311035, + "eval_runtime": 46.6051, + "eval_samples_per_second": 21.457, + "eval_steps_per_second": 1.352, + "step": 180 + }, + { + "epoch": 0.2619392185238784, + "grad_norm": 1.8590710163116455, + "learning_rate": 4.899436846339501e-06, + "loss": 1.0285, + "step": 181 + }, + { + "epoch": 0.2633863965267728, + "grad_norm": 2.299046039581299, + "learning_rate": 4.895414320193081e-06, + "loss": 1.0054, + "step": 182 + }, + { + "epoch": 0.26483357452966716, + "grad_norm": 1.7754614353179932, + "learning_rate": 4.891391794046661e-06, + "loss": 1.0105, + "step": 183 + }, + { + "epoch": 0.2662807525325615, + "grad_norm": 2.2448294162750244, + "learning_rate": 4.8873692679002414e-06, + "loss": 0.9932, + "step": 184 + }, + { + "epoch": 0.26772793053545585, + "grad_norm": 1.6432521343231201, + "learning_rate": 4.8833467417538216e-06, + "loss": 1.0193, + "step": 185 + }, + { + "epoch": 0.2691751085383502, + "grad_norm": 2.1414456367492676, + "learning_rate": 4.879324215607402e-06, + "loss": 0.9996, + "step": 186 + }, + { + "epoch": 0.2706222865412446, + "grad_norm": 2.040170431137085, + "learning_rate": 4.875301689460982e-06, + "loss": 0.9835, + "step": 187 + }, + { + "epoch": 0.27206946454413894, + "grad_norm": 1.591558814048767, + "learning_rate": 4.871279163314562e-06, + "loss": 0.9914, + "step": 188 + }, + { + "epoch": 0.2735166425470333, + "grad_norm": 1.534763216972351, + "learning_rate": 4.867256637168142e-06, + "loss": 0.9985, + "step": 189 + }, + { + "epoch": 0.27496382054992763, + "grad_norm": 1.9314086437225342, + "learning_rate": 4.863234111021722e-06, + "loss": 0.9794, + "step": 190 + }, + { + "epoch": 0.27496382054992763, + "eval_loss": 1.088218092918396, + "eval_runtime": 46.7319, + "eval_samples_per_second": 21.399, + "eval_steps_per_second": 1.348, + "step": 190 + }, + { + "epoch": 0.276410998552822, + "grad_norm": 1.6735827922821045, + "learning_rate": 4.8592115848753015e-06, + "loss": 1.0204, + "step": 191 + }, + { + "epoch": 0.2778581765557164, + "grad_norm": 3.009899854660034, + "learning_rate": 4.855189058728882e-06, + "loss": 0.952, + "step": 192 + }, + { + "epoch": 0.2793053545586107, + "grad_norm": 1.2619800567626953, + "learning_rate": 4.851166532582462e-06, + "loss": 0.965, + "step": 193 + }, + { + "epoch": 0.28075253256150506, + "grad_norm": 1.5649442672729492, + "learning_rate": 4.847144006436042e-06, + "loss": 0.9812, + "step": 194 + }, + { + "epoch": 0.2821997105643994, + "grad_norm": 1.4692801237106323, + "learning_rate": 4.843121480289622e-06, + "loss": 1.0315, + "step": 195 + }, + { + "epoch": 0.28364688856729375, + "grad_norm": 1.460066795349121, + "learning_rate": 4.839098954143202e-06, + "loss": 1.0022, + "step": 196 + }, + { + "epoch": 0.28509406657018815, + "grad_norm": 1.6650729179382324, + "learning_rate": 4.835076427996782e-06, + "loss": 0.9826, + "step": 197 + }, + { + "epoch": 0.2865412445730825, + "grad_norm": 1.723167896270752, + "learning_rate": 4.831053901850362e-06, + "loss": 0.9784, + "step": 198 + }, + { + "epoch": 0.28798842257597684, + "grad_norm": 1.7653883695602417, + "learning_rate": 4.8270313757039425e-06, + "loss": 0.9841, + "step": 199 + }, + { + "epoch": 0.2894356005788712, + "grad_norm": 1.8667902946472168, + "learning_rate": 4.823008849557523e-06, + "loss": 0.9776, + "step": 200 + }, + { + "epoch": 0.2894356005788712, + "eval_loss": 1.0795806646347046, + "eval_runtime": 46.7257, + "eval_samples_per_second": 21.402, + "eval_steps_per_second": 1.348, + "step": 200 + }, + { + "epoch": 0.29088277858176553, + "grad_norm": 2.379058361053467, + "learning_rate": 4.818986323411103e-06, + "loss": 0.9774, + "step": 201 + }, + { + "epoch": 0.29232995658465993, + "grad_norm": 1.3199034929275513, + "learning_rate": 4.814963797264683e-06, + "loss": 0.9716, + "step": 202 + }, + { + "epoch": 0.2937771345875543, + "grad_norm": 1.762894630432129, + "learning_rate": 4.810941271118263e-06, + "loss": 1.0146, + "step": 203 + }, + { + "epoch": 0.2952243125904486, + "grad_norm": 1.5601073503494263, + "learning_rate": 4.806918744971843e-06, + "loss": 1.0461, + "step": 204 + }, + { + "epoch": 0.29667149059334297, + "grad_norm": 1.6448826789855957, + "learning_rate": 4.802896218825423e-06, + "loss": 1.035, + "step": 205 + }, + { + "epoch": 0.2981186685962373, + "grad_norm": 1.6448826789855957, + "learning_rate": 4.802896218825423e-06, + "loss": 0.993, + "step": 206 + }, + { + "epoch": 0.2995658465991317, + "grad_norm": 1.8260093927383423, + "learning_rate": 4.798873692679003e-06, + "loss": 1.0174, + "step": 207 + }, + { + "epoch": 0.30101302460202606, + "grad_norm": 1.286902904510498, + "learning_rate": 4.794851166532583e-06, + "loss": 0.9611, + "step": 208 + }, + { + "epoch": 0.3024602026049204, + "grad_norm": 1.580316185951233, + "learning_rate": 4.790828640386163e-06, + "loss": 1.003, + "step": 209 + }, + { + "epoch": 0.30390738060781475, + "grad_norm": 1.3181246519088745, + "learning_rate": 4.786806114239743e-06, + "loss": 0.9678, + "step": 210 + }, + { + "epoch": 0.30390738060781475, + "eval_loss": 1.0733227729797363, + "eval_runtime": 46.7793, + "eval_samples_per_second": 21.377, + "eval_steps_per_second": 1.347, + "step": 210 + }, + { + "epoch": 0.3053545586107091, + "grad_norm": 1.6349185705184937, + "learning_rate": 4.782783588093323e-06, + "loss": 0.9872, + "step": 211 + }, + { + "epoch": 0.3068017366136035, + "grad_norm": 1.9016426801681519, + "learning_rate": 4.778761061946903e-06, + "loss": 1.004, + "step": 212 + }, + { + "epoch": 0.30824891461649784, + "grad_norm": 1.7413405179977417, + "learning_rate": 4.774738535800483e-06, + "loss": 0.9871, + "step": 213 + }, + { + "epoch": 0.3096960926193922, + "grad_norm": 1.6280605792999268, + "learning_rate": 4.7707160096540635e-06, + "loss": 1.0236, + "step": 214 + }, + { + "epoch": 0.3111432706222865, + "grad_norm": 2.00734806060791, + "learning_rate": 4.766693483507644e-06, + "loss": 0.9983, + "step": 215 + }, + { + "epoch": 0.3125904486251809, + "grad_norm": 1.3498228788375854, + "learning_rate": 4.762670957361224e-06, + "loss": 0.9957, + "step": 216 + }, + { + "epoch": 0.3140376266280753, + "grad_norm": 1.3405708074569702, + "learning_rate": 4.758648431214804e-06, + "loss": 1.0132, + "step": 217 + }, + { + "epoch": 0.3154848046309696, + "grad_norm": 1.6332919597625732, + "learning_rate": 4.754625905068383e-06, + "loss": 0.993, + "step": 218 + }, + { + "epoch": 0.31693198263386396, + "grad_norm": 1.54900062084198, + "learning_rate": 4.750603378921963e-06, + "loss": 0.9913, + "step": 219 + }, + { + "epoch": 0.3183791606367583, + "grad_norm": 1.4074807167053223, + "learning_rate": 4.746580852775543e-06, + "loss": 0.9712, + "step": 220 + }, + { + "epoch": 0.3183791606367583, + "eval_loss": 1.068241000175476, + "eval_runtime": 46.6828, + "eval_samples_per_second": 21.421, + "eval_steps_per_second": 1.35, + "step": 220 + }, + { + "epoch": 0.31982633863965265, + "grad_norm": 1.8566498756408691, + "learning_rate": 4.7425583266291235e-06, + "loss": 0.9793, + "step": 221 + }, + { + "epoch": 0.32127351664254705, + "grad_norm": 1.5620661973953247, + "learning_rate": 4.738535800482704e-06, + "loss": 0.9886, + "step": 222 + }, + { + "epoch": 0.3227206946454414, + "grad_norm": 2.01657772064209, + "learning_rate": 4.734513274336284e-06, + "loss": 1.0042, + "step": 223 + }, + { + "epoch": 0.32416787264833574, + "grad_norm": 1.6317390203475952, + "learning_rate": 4.730490748189864e-06, + "loss": 0.9357, + "step": 224 + }, + { + "epoch": 0.3256150506512301, + "grad_norm": 1.8639886379241943, + "learning_rate": 4.726468222043444e-06, + "loss": 0.9417, + "step": 225 + }, + { + "epoch": 0.32706222865412443, + "grad_norm": 1.4710664749145508, + "learning_rate": 4.722445695897024e-06, + "loss": 0.9659, + "step": 226 + }, + { + "epoch": 0.32850940665701883, + "grad_norm": 1.487237572669983, + "learning_rate": 4.718423169750604e-06, + "loss": 0.9776, + "step": 227 + }, + { + "epoch": 0.3299565846599132, + "grad_norm": 1.6453754901885986, + "learning_rate": 4.7144006436041835e-06, + "loss": 0.987, + "step": 228 + }, + { + "epoch": 0.3314037626628075, + "grad_norm": 1.7862600088119507, + "learning_rate": 4.710378117457764e-06, + "loss": 0.9512, + "step": 229 + }, + { + "epoch": 0.33285094066570187, + "grad_norm": 1.447259545326233, + "learning_rate": 4.706355591311344e-06, + "loss": 0.9957, + "step": 230 + }, + { + "epoch": 0.33285094066570187, + "eval_loss": 1.0609079599380493, + "eval_runtime": 46.7045, + "eval_samples_per_second": 21.411, + "eval_steps_per_second": 1.349, + "step": 230 + }, + { + "epoch": 0.3342981186685962, + "grad_norm": 1.5151357650756836, + "learning_rate": 4.702333065164924e-06, + "loss": 0.9792, + "step": 231 + }, + { + "epoch": 0.3357452966714906, + "grad_norm": 1.4104821681976318, + "learning_rate": 4.698310539018504e-06, + "loss": 0.9679, + "step": 232 + }, + { + "epoch": 0.33719247467438496, + "grad_norm": 1.466752529144287, + "learning_rate": 4.694288012872084e-06, + "loss": 0.989, + "step": 233 + }, + { + "epoch": 0.3386396526772793, + "grad_norm": 1.631919026374817, + "learning_rate": 4.690265486725664e-06, + "loss": 0.9843, + "step": 234 + }, + { + "epoch": 0.34008683068017365, + "grad_norm": 1.282708764076233, + "learning_rate": 4.6862429605792444e-06, + "loss": 1.0267, + "step": 235 + }, + { + "epoch": 0.341534008683068, + "grad_norm": 1.5720704793930054, + "learning_rate": 4.6822204344328246e-06, + "loss": 1.0031, + "step": 236 + }, + { + "epoch": 0.3429811866859624, + "grad_norm": 1.8414279222488403, + "learning_rate": 4.678197908286405e-06, + "loss": 0.9839, + "step": 237 + }, + { + "epoch": 0.34442836468885674, + "grad_norm": 1.5520700216293335, + "learning_rate": 4.674175382139984e-06, + "loss": 1.0047, + "step": 238 + }, + { + "epoch": 0.3458755426917511, + "grad_norm": 1.536651849746704, + "learning_rate": 4.670152855993564e-06, + "loss": 0.9896, + "step": 239 + }, + { + "epoch": 0.3473227206946454, + "grad_norm": 2.2550888061523438, + "learning_rate": 4.666130329847144e-06, + "loss": 0.9622, + "step": 240 + }, + { + "epoch": 0.3473227206946454, + "eval_loss": 1.055155873298645, + "eval_runtime": 46.752, + "eval_samples_per_second": 21.389, + "eval_steps_per_second": 1.348, + "step": 240 + }, + { + "epoch": 0.34876989869753977, + "grad_norm": 1.4783450365066528, + "learning_rate": 4.662107803700724e-06, + "loss": 0.9668, + "step": 241 + }, + { + "epoch": 0.35021707670043417, + "grad_norm": 1.802058219909668, + "learning_rate": 4.6580852775543045e-06, + "loss": 0.9375, + "step": 242 + }, + { + "epoch": 0.3516642547033285, + "grad_norm": 1.6570472717285156, + "learning_rate": 4.654062751407885e-06, + "loss": 0.9825, + "step": 243 + }, + { + "epoch": 0.35311143270622286, + "grad_norm": 1.7852758169174194, + "learning_rate": 4.650040225261465e-06, + "loss": 0.9729, + "step": 244 + }, + { + "epoch": 0.3545586107091172, + "grad_norm": 1.4200557470321655, + "learning_rate": 4.646017699115045e-06, + "loss": 0.9882, + "step": 245 + }, + { + "epoch": 0.35600578871201155, + "grad_norm": 1.8521738052368164, + "learning_rate": 4.641995172968625e-06, + "loss": 0.9715, + "step": 246 + }, + { + "epoch": 0.35745296671490595, + "grad_norm": 1.8887485265731812, + "learning_rate": 4.637972646822205e-06, + "loss": 0.98, + "step": 247 + }, + { + "epoch": 0.3589001447178003, + "grad_norm": 1.3600974082946777, + "learning_rate": 4.633950120675784e-06, + "loss": 0.9725, + "step": 248 + }, + { + "epoch": 0.36034732272069464, + "grad_norm": 1.7848236560821533, + "learning_rate": 4.6299275945293645e-06, + "loss": 0.9901, + "step": 249 + }, + { + "epoch": 0.361794500723589, + "grad_norm": 1.6750437021255493, + "learning_rate": 4.625905068382945e-06, + "loss": 0.9529, + "step": 250 + }, + { + "epoch": 0.361794500723589, + "eval_loss": 1.0548979043960571, + "eval_runtime": 46.6832, + "eval_samples_per_second": 21.421, + "eval_steps_per_second": 1.35, + "step": 250 + }, + { + "epoch": 0.36324167872648333, + "grad_norm": 1.7013113498687744, + "learning_rate": 4.621882542236525e-06, + "loss": 0.9791, + "step": 251 + }, + { + "epoch": 0.36468885672937773, + "grad_norm": 1.581060767173767, + "learning_rate": 4.617860016090105e-06, + "loss": 0.9535, + "step": 252 + }, + { + "epoch": 0.3661360347322721, + "grad_norm": 1.2800977230072021, + "learning_rate": 4.613837489943685e-06, + "loss": 0.9485, + "step": 253 + }, + { + "epoch": 0.3675832127351664, + "grad_norm": 1.5921649932861328, + "learning_rate": 4.609814963797265e-06, + "loss": 0.9725, + "step": 254 + }, + { + "epoch": 0.36903039073806077, + "grad_norm": 1.8379957675933838, + "learning_rate": 4.605792437650845e-06, + "loss": 0.9743, + "step": 255 + }, + { + "epoch": 0.3704775687409551, + "grad_norm": 1.478106141090393, + "learning_rate": 4.6017699115044254e-06, + "loss": 0.9476, + "step": 256 + }, + { + "epoch": 0.3719247467438495, + "grad_norm": 1.6843070983886719, + "learning_rate": 4.5977473853580056e-06, + "loss": 0.9566, + "step": 257 + }, + { + "epoch": 0.37337192474674386, + "grad_norm": 1.4570088386535645, + "learning_rate": 4.593724859211585e-06, + "loss": 0.977, + "step": 258 + }, + { + "epoch": 0.3748191027496382, + "grad_norm": 1.4170172214508057, + "learning_rate": 4.589702333065165e-06, + "loss": 0.9732, + "step": 259 + }, + { + "epoch": 0.37626628075253254, + "grad_norm": 1.8241543769836426, + "learning_rate": 4.585679806918745e-06, + "loss": 0.9836, + "step": 260 + }, + { + "epoch": 0.37626628075253254, + "eval_loss": 1.049314022064209, + "eval_runtime": 46.6418, + "eval_samples_per_second": 21.44, + "eval_steps_per_second": 1.351, + "step": 260 + }, + { + "epoch": 0.37771345875542695, + "grad_norm": 1.5743048191070557, + "learning_rate": 4.581657280772325e-06, + "loss": 0.9179, + "step": 261 + }, + { + "epoch": 0.3791606367583213, + "grad_norm": 1.4050718545913696, + "learning_rate": 4.577634754625905e-06, + "loss": 0.9611, + "step": 262 + }, + { + "epoch": 0.38060781476121563, + "grad_norm": 1.4073106050491333, + "learning_rate": 4.5736122284794855e-06, + "loss": 0.982, + "step": 263 + }, + { + "epoch": 0.38205499276411, + "grad_norm": 1.33458411693573, + "learning_rate": 4.569589702333066e-06, + "loss": 0.9933, + "step": 264 + }, + { + "epoch": 0.3835021707670043, + "grad_norm": 1.7036134004592896, + "learning_rate": 4.565567176186646e-06, + "loss": 0.95, + "step": 265 + }, + { + "epoch": 0.3849493487698987, + "grad_norm": 1.8228670358657837, + "learning_rate": 4.561544650040226e-06, + "loss": 0.9631, + "step": 266 + }, + { + "epoch": 0.38639652677279307, + "grad_norm": 1.5331554412841797, + "learning_rate": 4.557522123893805e-06, + "loss": 0.9615, + "step": 267 + }, + { + "epoch": 0.3878437047756874, + "grad_norm": 2.1214566230773926, + "learning_rate": 4.553499597747385e-06, + "loss": 0.9874, + "step": 268 + }, + { + "epoch": 0.38929088277858176, + "grad_norm": 1.894487738609314, + "learning_rate": 4.549477071600965e-06, + "loss": 0.9565, + "step": 269 + }, + { + "epoch": 0.3907380607814761, + "grad_norm": 1.4514497518539429, + "learning_rate": 4.5454545454545455e-06, + "loss": 0.9506, + "step": 270 + }, + { + "epoch": 0.3907380607814761, + "eval_loss": 1.047164797782898, + "eval_runtime": 46.8372, + "eval_samples_per_second": 21.351, + "eval_steps_per_second": 1.345, + "step": 270 + }, + { + "epoch": 0.3921852387843705, + "grad_norm": 1.2923952341079712, + "learning_rate": 4.541432019308126e-06, + "loss": 0.9179, + "step": 271 + }, + { + "epoch": 0.39363241678726485, + "grad_norm": 1.5526196956634521, + "learning_rate": 4.537409493161706e-06, + "loss": 1.0126, + "step": 272 + }, + { + "epoch": 0.3950795947901592, + "grad_norm": 1.480729341506958, + "learning_rate": 4.533386967015286e-06, + "loss": 0.9222, + "step": 273 + }, + { + "epoch": 0.39652677279305354, + "grad_norm": 1.3874094486236572, + "learning_rate": 4.529364440868866e-06, + "loss": 0.9565, + "step": 274 + }, + { + "epoch": 0.3979739507959479, + "grad_norm": 1.574267029762268, + "learning_rate": 4.525341914722446e-06, + "loss": 0.9387, + "step": 275 + }, + { + "epoch": 0.3994211287988423, + "grad_norm": 1.2167718410491943, + "learning_rate": 4.521319388576026e-06, + "loss": 0.9617, + "step": 276 + }, + { + "epoch": 0.40086830680173663, + "grad_norm": 1.8016854524612427, + "learning_rate": 4.5172968624296056e-06, + "loss": 0.9661, + "step": 277 + }, + { + "epoch": 0.402315484804631, + "grad_norm": 2.013470411300659, + "learning_rate": 4.513274336283186e-06, + "loss": 0.9411, + "step": 278 + }, + { + "epoch": 0.4037626628075253, + "grad_norm": 1.4565365314483643, + "learning_rate": 4.509251810136766e-06, + "loss": 0.9547, + "step": 279 + }, + { + "epoch": 0.40520984081041966, + "grad_norm": 1.730705738067627, + "learning_rate": 4.505229283990346e-06, + "loss": 0.9474, + "step": 280 + }, + { + "epoch": 0.40520984081041966, + "eval_loss": 1.0404577255249023, + "eval_runtime": 46.8657, + "eval_samples_per_second": 21.338, + "eval_steps_per_second": 1.344, + "step": 280 + }, + { + "epoch": 0.40665701881331406, + "grad_norm": 2.0525460243225098, + "learning_rate": 4.501206757843926e-06, + "loss": 0.9767, + "step": 281 + }, + { + "epoch": 0.4081041968162084, + "grad_norm": 1.8083363771438599, + "learning_rate": 4.497184231697506e-06, + "loss": 0.99, + "step": 282 + }, + { + "epoch": 0.40955137481910275, + "grad_norm": 1.6748610734939575, + "learning_rate": 4.493161705551086e-06, + "loss": 0.9385, + "step": 283 + }, + { + "epoch": 0.4109985528219971, + "grad_norm": 2.5204918384552, + "learning_rate": 4.4891391794046665e-06, + "loss": 0.9656, + "step": 284 + }, + { + "epoch": 0.41244573082489144, + "grad_norm": 1.5199946165084839, + "learning_rate": 4.485116653258247e-06, + "loss": 0.9709, + "step": 285 + }, + { + "epoch": 0.41389290882778584, + "grad_norm": 1.4089913368225098, + "learning_rate": 4.481094127111827e-06, + "loss": 0.9441, + "step": 286 + }, + { + "epoch": 0.4153400868306802, + "grad_norm": 2.1064705848693848, + "learning_rate": 4.477071600965406e-06, + "loss": 0.9482, + "step": 287 + }, + { + "epoch": 0.41678726483357453, + "grad_norm": 1.3072782754898071, + "learning_rate": 4.473049074818986e-06, + "loss": 0.9496, + "step": 288 + }, + { + "epoch": 0.4182344428364689, + "grad_norm": 1.4177823066711426, + "learning_rate": 4.469026548672566e-06, + "loss": 0.9685, + "step": 289 + }, + { + "epoch": 0.4196816208393632, + "grad_norm": 1.5199861526489258, + "learning_rate": 4.465004022526146e-06, + "loss": 0.9796, + "step": 290 + }, + { + "epoch": 0.4196816208393632, + "eval_loss": 1.038499355316162, + "eval_runtime": 46.7882, + "eval_samples_per_second": 21.373, + "eval_steps_per_second": 1.346, + "step": 290 + }, + { + "epoch": 0.4211287988422576, + "grad_norm": 1.5769327878952026, + "learning_rate": 4.460981496379727e-06, + "loss": 0.9377, + "step": 291 + }, + { + "epoch": 0.42257597684515197, + "grad_norm": 1.4754797220230103, + "learning_rate": 4.4569589702333075e-06, + "loss": 0.9603, + "step": 292 + }, + { + "epoch": 0.4240231548480463, + "grad_norm": 1.4994487762451172, + "learning_rate": 4.452936444086888e-06, + "loss": 0.945, + "step": 293 + }, + { + "epoch": 0.42547033285094066, + "grad_norm": 1.676659107208252, + "learning_rate": 4.448913917940467e-06, + "loss": 0.9646, + "step": 294 + }, + { + "epoch": 0.426917510853835, + "grad_norm": 2.0028061866760254, + "learning_rate": 4.444891391794047e-06, + "loss": 0.9298, + "step": 295 + }, + { + "epoch": 0.4283646888567294, + "grad_norm": 1.5318928956985474, + "learning_rate": 4.440868865647627e-06, + "loss": 0.9421, + "step": 296 + }, + { + "epoch": 0.42981186685962375, + "grad_norm": 1.3500826358795166, + "learning_rate": 4.436846339501207e-06, + "loss": 0.9246, + "step": 297 + }, + { + "epoch": 0.4312590448625181, + "grad_norm": 1.6406341791152954, + "learning_rate": 4.432823813354787e-06, + "loss": 0.9454, + "step": 298 + }, + { + "epoch": 0.43270622286541244, + "grad_norm": 1.6433024406433105, + "learning_rate": 4.4288012872083675e-06, + "loss": 0.9714, + "step": 299 + }, + { + "epoch": 0.4341534008683068, + "grad_norm": 1.6767746210098267, + "learning_rate": 4.424778761061948e-06, + "loss": 0.9127, + "step": 300 + }, + { + "epoch": 0.4341534008683068, + "eval_loss": 1.037190318107605, + "eval_runtime": 46.758, + "eval_samples_per_second": 21.387, + "eval_steps_per_second": 1.347, + "step": 300 + }, + { + "epoch": 0.4356005788712012, + "grad_norm": 1.8116039037704468, + "learning_rate": 4.420756234915528e-06, + "loss": 0.9161, + "step": 301 + }, + { + "epoch": 0.4370477568740955, + "grad_norm": 1.711745262145996, + "learning_rate": 4.416733708769108e-06, + "loss": 0.9784, + "step": 302 + }, + { + "epoch": 0.4384949348769899, + "grad_norm": 1.477962613105774, + "learning_rate": 4.412711182622687e-06, + "loss": 0.9286, + "step": 303 + }, + { + "epoch": 0.4399421128798842, + "grad_norm": 1.450057029724121, + "learning_rate": 4.408688656476267e-06, + "loss": 0.9404, + "step": 304 + }, + { + "epoch": 0.44138929088277856, + "grad_norm": 2.032090187072754, + "learning_rate": 4.4046661303298474e-06, + "loss": 0.9256, + "step": 305 + }, + { + "epoch": 0.44283646888567296, + "grad_norm": 1.6227837800979614, + "learning_rate": 4.4006436041834276e-06, + "loss": 0.9297, + "step": 306 + }, + { + "epoch": 0.4442836468885673, + "grad_norm": 1.5233511924743652, + "learning_rate": 4.396621078037008e-06, + "loss": 0.923, + "step": 307 + }, + { + "epoch": 0.44573082489146165, + "grad_norm": 1.7230010032653809, + "learning_rate": 4.392598551890588e-06, + "loss": 1.0137, + "step": 308 + }, + { + "epoch": 0.447178002894356, + "grad_norm": 1.4706658124923706, + "learning_rate": 4.388576025744168e-06, + "loss": 0.9381, + "step": 309 + }, + { + "epoch": 0.44862518089725034, + "grad_norm": 1.7938367128372192, + "learning_rate": 4.384553499597748e-06, + "loss": 0.9219, + "step": 310 + }, + { + "epoch": 0.44862518089725034, + "eval_loss": 1.0325793027877808, + "eval_runtime": 46.7452, + "eval_samples_per_second": 21.393, + "eval_steps_per_second": 1.348, + "step": 310 + }, + { + "epoch": 0.45007235890014474, + "grad_norm": 1.9137241840362549, + "learning_rate": 4.380530973451328e-06, + "loss": 0.9754, + "step": 311 + }, + { + "epoch": 0.4515195369030391, + "grad_norm": 1.4783958196640015, + "learning_rate": 4.376508447304908e-06, + "loss": 0.9018, + "step": 312 + }, + { + "epoch": 0.45296671490593343, + "grad_norm": 1.8495306968688965, + "learning_rate": 4.372485921158488e-06, + "loss": 0.9476, + "step": 313 + }, + { + "epoch": 0.4544138929088278, + "grad_norm": 1.4381948709487915, + "learning_rate": 4.368463395012068e-06, + "loss": 0.9116, + "step": 314 + }, + { + "epoch": 0.4558610709117221, + "grad_norm": 1.5488134622573853, + "learning_rate": 4.364440868865648e-06, + "loss": 0.9692, + "step": 315 + }, + { + "epoch": 0.4573082489146165, + "grad_norm": 1.573097586631775, + "learning_rate": 4.360418342719228e-06, + "loss": 0.9141, + "step": 316 + }, + { + "epoch": 0.45875542691751087, + "grad_norm": 1.4818278551101685, + "learning_rate": 4.356395816572808e-06, + "loss": 0.9891, + "step": 317 + }, + { + "epoch": 0.4602026049204052, + "grad_norm": 1.728715419769287, + "learning_rate": 4.352373290426388e-06, + "loss": 0.9093, + "step": 318 + }, + { + "epoch": 0.46164978292329956, + "grad_norm": 1.3753663301467896, + "learning_rate": 4.348350764279968e-06, + "loss": 0.9503, + "step": 319 + }, + { + "epoch": 0.4630969609261939, + "grad_norm": 1.6115820407867432, + "learning_rate": 4.3443282381335485e-06, + "loss": 0.9535, + "step": 320 + }, + { + "epoch": 0.4630969609261939, + "eval_loss": 1.0294053554534912, + "eval_runtime": 46.7255, + "eval_samples_per_second": 21.402, + "eval_steps_per_second": 1.348, + "step": 320 + }, + { + "epoch": 0.4645441389290883, + "grad_norm": 1.6108647584915161, + "learning_rate": 4.340305711987129e-06, + "loss": 0.9484, + "step": 321 + }, + { + "epoch": 0.46599131693198265, + "grad_norm": 1.3854362964630127, + "learning_rate": 4.336283185840709e-06, + "loss": 0.9258, + "step": 322 + }, + { + "epoch": 0.467438494934877, + "grad_norm": 1.9534512758255005, + "learning_rate": 4.332260659694288e-06, + "loss": 0.9542, + "step": 323 + }, + { + "epoch": 0.46888567293777134, + "grad_norm": 1.6645305156707764, + "learning_rate": 4.328238133547868e-06, + "loss": 0.9463, + "step": 324 + }, + { + "epoch": 0.4703328509406657, + "grad_norm": 1.3589664697647095, + "learning_rate": 4.324215607401448e-06, + "loss": 0.9405, + "step": 325 + }, + { + "epoch": 0.4717800289435601, + "grad_norm": 1.6425889730453491, + "learning_rate": 4.3201930812550284e-06, + "loss": 0.9246, + "step": 326 + }, + { + "epoch": 0.4732272069464544, + "grad_norm": 1.3623247146606445, + "learning_rate": 4.3161705551086086e-06, + "loss": 0.9245, + "step": 327 + }, + { + "epoch": 0.47467438494934877, + "grad_norm": 1.5490796566009521, + "learning_rate": 4.312148028962189e-06, + "loss": 0.9963, + "step": 328 + }, + { + "epoch": 0.4761215629522431, + "grad_norm": 1.1860958337783813, + "learning_rate": 4.308125502815769e-06, + "loss": 0.9623, + "step": 329 + }, + { + "epoch": 0.47756874095513746, + "grad_norm": 1.5183347463607788, + "learning_rate": 4.304102976669349e-06, + "loss": 0.9522, + "step": 330 + }, + { + "epoch": 0.47756874095513746, + "eval_loss": 1.0273957252502441, + "eval_runtime": 46.8055, + "eval_samples_per_second": 21.365, + "eval_steps_per_second": 1.346, + "step": 330 + }, + { + "epoch": 0.47901591895803186, + "grad_norm": 1.7083364725112915, + "learning_rate": 4.300080450522929e-06, + "loss": 0.9069, + "step": 331 + }, + { + "epoch": 0.4804630969609262, + "grad_norm": 1.5451003313064575, + "learning_rate": 4.296057924376509e-06, + "loss": 0.9376, + "step": 332 + }, + { + "epoch": 0.48191027496382055, + "grad_norm": 1.4622794389724731, + "learning_rate": 4.2920353982300885e-06, + "loss": 0.9496, + "step": 333 + }, + { + "epoch": 0.4833574529667149, + "grad_norm": 1.382709264755249, + "learning_rate": 4.288012872083669e-06, + "loss": 0.951, + "step": 334 + }, + { + "epoch": 0.48480463096960924, + "grad_norm": 1.581541657447815, + "learning_rate": 4.283990345937249e-06, + "loss": 0.9342, + "step": 335 + }, + { + "epoch": 0.48625180897250364, + "grad_norm": 1.6656814813613892, + "learning_rate": 4.279967819790829e-06, + "loss": 0.9339, + "step": 336 + }, + { + "epoch": 0.487698986975398, + "grad_norm": 1.4430766105651855, + "learning_rate": 4.275945293644409e-06, + "loss": 0.9353, + "step": 337 + }, + { + "epoch": 0.48914616497829233, + "grad_norm": 1.8380608558654785, + "learning_rate": 4.271922767497989e-06, + "loss": 0.9364, + "step": 338 + }, + { + "epoch": 0.4905933429811867, + "grad_norm": 1.269301176071167, + "learning_rate": 4.267900241351569e-06, + "loss": 0.9413, + "step": 339 + }, + { + "epoch": 0.492040520984081, + "grad_norm": 1.1995859146118164, + "learning_rate": 4.263877715205149e-06, + "loss": 0.9048, + "step": 340 + }, + { + "epoch": 0.492040520984081, + "eval_loss": 1.0264354944229126, + "eval_runtime": 46.7401, + "eval_samples_per_second": 21.395, + "eval_steps_per_second": 1.348, + "step": 340 + }, + { + "epoch": 0.4934876989869754, + "grad_norm": 1.7994416952133179, + "learning_rate": 4.2598551890587295e-06, + "loss": 0.9449, + "step": 341 + }, + { + "epoch": 0.49493487698986977, + "grad_norm": 1.6128180027008057, + "learning_rate": 4.25583266291231e-06, + "loss": 0.9432, + "step": 342 + }, + { + "epoch": 0.4963820549927641, + "grad_norm": 1.3202834129333496, + "learning_rate": 4.251810136765889e-06, + "loss": 0.8938, + "step": 343 + }, + { + "epoch": 0.49782923299565845, + "grad_norm": 1.461272120475769, + "learning_rate": 4.247787610619469e-06, + "loss": 0.9336, + "step": 344 + }, + { + "epoch": 0.4992764109985528, + "grad_norm": 1.2114381790161133, + "learning_rate": 4.243765084473049e-06, + "loss": 0.9331, + "step": 345 + }, + { + "epoch": 0.5007235890014472, + "grad_norm": 1.4170477390289307, + "learning_rate": 4.239742558326629e-06, + "loss": 0.9238, + "step": 346 + }, + { + "epoch": 0.5021707670043415, + "grad_norm": 1.5438958406448364, + "learning_rate": 4.235720032180209e-06, + "loss": 0.9017, + "step": 347 + }, + { + "epoch": 0.5036179450072359, + "grad_norm": 1.3673092126846313, + "learning_rate": 4.2316975060337895e-06, + "loss": 0.9413, + "step": 348 + }, + { + "epoch": 0.5050651230101303, + "grad_norm": 1.8526067733764648, + "learning_rate": 4.22767497988737e-06, + "loss": 0.9117, + "step": 349 + }, + { + "epoch": 0.5065123010130246, + "grad_norm": 1.341046929359436, + "learning_rate": 4.22365245374095e-06, + "loss": 0.91, + "step": 350 + }, + { + "epoch": 0.5065123010130246, + "eval_loss": 1.0264149904251099, + "eval_runtime": 46.7, + "eval_samples_per_second": 21.413, + "eval_steps_per_second": 1.349, + "step": 350 + }, + { + "epoch": 0.507959479015919, + "grad_norm": 1.5988823175430298, + "learning_rate": 4.21962992759453e-06, + "loss": 0.952, + "step": 351 + }, + { + "epoch": 0.5094066570188133, + "grad_norm": 1.3648533821105957, + "learning_rate": 4.21560740144811e-06, + "loss": 0.9571, + "step": 352 + }, + { + "epoch": 0.5108538350217077, + "grad_norm": 1.495408535003662, + "learning_rate": 4.211584875301689e-06, + "loss": 0.9223, + "step": 353 + }, + { + "epoch": 0.5123010130246021, + "grad_norm": 1.183349370956421, + "learning_rate": 4.2075623491552695e-06, + "loss": 0.9632, + "step": 354 + }, + { + "epoch": 0.5137481910274964, + "grad_norm": 1.5074958801269531, + "learning_rate": 4.20353982300885e-06, + "loss": 0.893, + "step": 355 + }, + { + "epoch": 0.5151953690303908, + "grad_norm": 1.3132048845291138, + "learning_rate": 4.19951729686243e-06, + "loss": 0.9823, + "step": 356 + }, + { + "epoch": 0.516642547033285, + "grad_norm": 1.4051076173782349, + "learning_rate": 4.19549477071601e-06, + "loss": 0.9296, + "step": 357 + }, + { + "epoch": 0.5180897250361794, + "grad_norm": 1.1534206867218018, + "learning_rate": 4.19147224456959e-06, + "loss": 0.9253, + "step": 358 + }, + { + "epoch": 0.5195369030390738, + "grad_norm": 1.5908139944076538, + "learning_rate": 4.18744971842317e-06, + "loss": 0.9521, + "step": 359 + }, + { + "epoch": 0.5209840810419681, + "grad_norm": 1.4979522228240967, + "learning_rate": 4.18342719227675e-06, + "loss": 0.9159, + "step": 360 + }, + { + "epoch": 0.5209840810419681, + "eval_loss": 1.017462134361267, + "eval_runtime": 46.6372, + "eval_samples_per_second": 21.442, + "eval_steps_per_second": 1.351, + "step": 360 + }, + { + "epoch": 0.5224312590448625, + "grad_norm": 1.4574133157730103, + "learning_rate": 4.17940466613033e-06, + "loss": 0.956, + "step": 361 + }, + { + "epoch": 0.5238784370477568, + "grad_norm": 1.6397000551223755, + "learning_rate": 4.1753821399839105e-06, + "loss": 0.9207, + "step": 362 + }, + { + "epoch": 0.5253256150506512, + "grad_norm": 1.3943746089935303, + "learning_rate": 4.17135961383749e-06, + "loss": 0.9208, + "step": 363 + }, + { + "epoch": 0.5267727930535456, + "grad_norm": 1.3155155181884766, + "learning_rate": 4.16733708769107e-06, + "loss": 0.9258, + "step": 364 + }, + { + "epoch": 0.5282199710564399, + "grad_norm": 2.077277660369873, + "learning_rate": 4.16331456154465e-06, + "loss": 0.9329, + "step": 365 + }, + { + "epoch": 0.5296671490593343, + "grad_norm": 1.9682797193527222, + "learning_rate": 4.15929203539823e-06, + "loss": 0.9274, + "step": 366 + }, + { + "epoch": 0.5311143270622286, + "grad_norm": 1.5062534809112549, + "learning_rate": 4.15526950925181e-06, + "loss": 0.9011, + "step": 367 + }, + { + "epoch": 0.532561505065123, + "grad_norm": 1.434861183166504, + "learning_rate": 4.15124698310539e-06, + "loss": 0.9287, + "step": 368 + }, + { + "epoch": 0.5340086830680174, + "grad_norm": 1.4730939865112305, + "learning_rate": 4.1472244569589705e-06, + "loss": 0.9263, + "step": 369 + }, + { + "epoch": 0.5354558610709117, + "grad_norm": 1.8028590679168701, + "learning_rate": 4.143201930812551e-06, + "loss": 0.9197, + "step": 370 + }, + { + "epoch": 0.5354558610709117, + "eval_loss": 1.0177210569381714, + "eval_runtime": 46.6339, + "eval_samples_per_second": 21.444, + "eval_steps_per_second": 1.351, + "step": 370 + }, + { + "epoch": 0.5369030390738061, + "grad_norm": 1.3988326787948608, + "learning_rate": 4.139179404666131e-06, + "loss": 0.9226, + "step": 371 + }, + { + "epoch": 0.5383502170767004, + "grad_norm": 1.7191375494003296, + "learning_rate": 4.13515687851971e-06, + "loss": 0.886, + "step": 372 + }, + { + "epoch": 0.5397973950795948, + "grad_norm": 1.411466360092163, + "learning_rate": 4.13113435237329e-06, + "loss": 0.9184, + "step": 373 + }, + { + "epoch": 0.5412445730824892, + "grad_norm": 1.5268676280975342, + "learning_rate": 4.12711182622687e-06, + "loss": 0.939, + "step": 374 + }, + { + "epoch": 0.5426917510853835, + "grad_norm": 1.6145097017288208, + "learning_rate": 4.1230893000804505e-06, + "loss": 0.9287, + "step": 375 + }, + { + "epoch": 0.5441389290882779, + "grad_norm": 1.3369029760360718, + "learning_rate": 4.119066773934031e-06, + "loss": 0.9015, + "step": 376 + }, + { + "epoch": 0.5455861070911722, + "grad_norm": 1.5564725399017334, + "learning_rate": 4.115044247787611e-06, + "loss": 0.8884, + "step": 377 + }, + { + "epoch": 0.5470332850940666, + "grad_norm": 1.7139111757278442, + "learning_rate": 4.111021721641191e-06, + "loss": 0.9149, + "step": 378 + }, + { + "epoch": 0.548480463096961, + "grad_norm": 1.6800168752670288, + "learning_rate": 4.106999195494771e-06, + "loss": 0.9365, + "step": 379 + }, + { + "epoch": 0.5499276410998553, + "grad_norm": 1.871535301208496, + "learning_rate": 4.102976669348351e-06, + "loss": 0.9562, + "step": 380 + }, + { + "epoch": 0.5499276410998553, + "eval_loss": 1.0147827863693237, + "eval_runtime": 46.7265, + "eval_samples_per_second": 21.401, + "eval_steps_per_second": 1.348, + "step": 380 + }, + { + "epoch": 0.5513748191027497, + "grad_norm": 1.3376110792160034, + "learning_rate": 4.098954143201931e-06, + "loss": 0.8866, + "step": 381 + }, + { + "epoch": 0.552821997105644, + "grad_norm": 2.097120523452759, + "learning_rate": 4.094931617055511e-06, + "loss": 0.9263, + "step": 382 + }, + { + "epoch": 0.5542691751085383, + "grad_norm": 1.973817229270935, + "learning_rate": 4.0909090909090915e-06, + "loss": 0.9177, + "step": 383 + }, + { + "epoch": 0.5557163531114327, + "grad_norm": 1.4708714485168457, + "learning_rate": 4.086886564762672e-06, + "loss": 0.9523, + "step": 384 + }, + { + "epoch": 0.557163531114327, + "grad_norm": 1.4091403484344482, + "learning_rate": 4.082864038616252e-06, + "loss": 0.9543, + "step": 385 + }, + { + "epoch": 0.5586107091172214, + "grad_norm": 1.2289339303970337, + "learning_rate": 4.078841512469832e-06, + "loss": 0.9354, + "step": 386 + }, + { + "epoch": 0.5600578871201157, + "grad_norm": 1.5708125829696655, + "learning_rate": 4.074818986323412e-06, + "loss": 0.8794, + "step": 387 + }, + { + "epoch": 0.5615050651230101, + "grad_norm": 1.6371783018112183, + "learning_rate": 4.070796460176992e-06, + "loss": 0.8795, + "step": 388 + }, + { + "epoch": 0.5629522431259045, + "grad_norm": 1.4529517889022827, + "learning_rate": 4.066773934030571e-06, + "loss": 0.927, + "step": 389 + }, + { + "epoch": 0.5643994211287988, + "grad_norm": 1.421964406967163, + "learning_rate": 4.0627514078841515e-06, + "loss": 0.9169, + "step": 390 + }, + { + "epoch": 0.5643994211287988, + "eval_loss": 1.0115739107131958, + "eval_runtime": 46.5942, + "eval_samples_per_second": 21.462, + "eval_steps_per_second": 1.352, + "step": 390 + }, + { + "epoch": 0.5658465991316932, + "grad_norm": 1.6639692783355713, + "learning_rate": 4.058728881737732e-06, + "loss": 0.9186, + "step": 391 + }, + { + "epoch": 0.5672937771345875, + "grad_norm": 1.9193339347839355, + "learning_rate": 4.054706355591312e-06, + "loss": 0.9343, + "step": 392 + }, + { + "epoch": 0.5687409551374819, + "grad_norm": 1.3492581844329834, + "learning_rate": 4.050683829444892e-06, + "loss": 0.9261, + "step": 393 + }, + { + "epoch": 0.5701881331403763, + "grad_norm": 1.3705259561538696, + "learning_rate": 4.046661303298472e-06, + "loss": 0.9083, + "step": 394 + }, + { + "epoch": 0.5716353111432706, + "grad_norm": 1.5951439142227173, + "learning_rate": 4.042638777152052e-06, + "loss": 0.9239, + "step": 395 + }, + { + "epoch": 0.573082489146165, + "grad_norm": 1.4445085525512695, + "learning_rate": 4.038616251005632e-06, + "loss": 0.9345, + "step": 396 + }, + { + "epoch": 0.5745296671490593, + "grad_norm": 1.725517988204956, + "learning_rate": 4.034593724859212e-06, + "loss": 0.9149, + "step": 397 + }, + { + "epoch": 0.5759768451519537, + "grad_norm": 1.5287132263183594, + "learning_rate": 4.0305711987127925e-06, + "loss": 0.9307, + "step": 398 + }, + { + "epoch": 0.5774240231548481, + "grad_norm": 1.3868067264556885, + "learning_rate": 4.026548672566372e-06, + "loss": 0.9235, + "step": 399 + }, + { + "epoch": 0.5788712011577424, + "grad_norm": 1.611224889755249, + "learning_rate": 4.022526146419952e-06, + "loss": 0.9272, + "step": 400 + }, + { + "epoch": 0.5788712011577424, + "eval_loss": 1.0099736452102661, + "eval_runtime": 46.6983, + "eval_samples_per_second": 21.414, + "eval_steps_per_second": 1.349, + "step": 400 + }, + { + "epoch": 0.5803183791606368, + "grad_norm": 1.4565112590789795, + "learning_rate": 4.018503620273532e-06, + "loss": 0.9137, + "step": 401 + }, + { + "epoch": 0.5817655571635311, + "grad_norm": 1.5255826711654663, + "learning_rate": 4.014481094127112e-06, + "loss": 0.9168, + "step": 402 + }, + { + "epoch": 0.5832127351664255, + "grad_norm": 1.4553810358047485, + "learning_rate": 4.010458567980692e-06, + "loss": 0.9042, + "step": 403 + }, + { + "epoch": 0.5846599131693199, + "grad_norm": 1.4039092063903809, + "learning_rate": 4.0064360418342725e-06, + "loss": 0.9443, + "step": 404 + }, + { + "epoch": 0.5861070911722142, + "grad_norm": 1.5143613815307617, + "learning_rate": 4.002413515687853e-06, + "loss": 0.9446, + "step": 405 + }, + { + "epoch": 0.5875542691751086, + "grad_norm": 1.2412786483764648, + "learning_rate": 3.998390989541433e-06, + "loss": 0.9574, + "step": 406 + }, + { + "epoch": 0.5890014471780028, + "grad_norm": 1.2274982929229736, + "learning_rate": 3.994368463395013e-06, + "loss": 0.9213, + "step": 407 + }, + { + "epoch": 0.5904486251808972, + "grad_norm": 1.3927079439163208, + "learning_rate": 3.990345937248592e-06, + "loss": 0.9046, + "step": 408 + }, + { + "epoch": 0.5918958031837916, + "grad_norm": 1.5312466621398926, + "learning_rate": 3.986323411102172e-06, + "loss": 0.9039, + "step": 409 + }, + { + "epoch": 0.5933429811866859, + "grad_norm": 1.5107614994049072, + "learning_rate": 3.982300884955752e-06, + "loss": 0.8883, + "step": 410 + }, + { + "epoch": 0.5933429811866859, + "eval_loss": 1.0098820924758911, + "eval_runtime": 46.5971, + "eval_samples_per_second": 21.461, + "eval_steps_per_second": 1.352, + "step": 410 + }, + { + "epoch": 0.5947901591895803, + "grad_norm": 1.9479697942733765, + "learning_rate": 3.9782783588093325e-06, + "loss": 0.915, + "step": 411 + }, + { + "epoch": 0.5962373371924746, + "grad_norm": 1.3607192039489746, + "learning_rate": 3.974255832662913e-06, + "loss": 0.8986, + "step": 412 + }, + { + "epoch": 0.597684515195369, + "grad_norm": 1.6084372997283936, + "learning_rate": 3.970233306516493e-06, + "loss": 0.8857, + "step": 413 + }, + { + "epoch": 0.5991316931982634, + "grad_norm": 1.6653928756713867, + "learning_rate": 3.966210780370073e-06, + "loss": 0.9346, + "step": 414 + }, + { + "epoch": 0.6005788712011577, + "grad_norm": 1.4960678815841675, + "learning_rate": 3.962188254223653e-06, + "loss": 0.9392, + "step": 415 + }, + { + "epoch": 0.6020260492040521, + "grad_norm": 1.4856607913970947, + "learning_rate": 3.958165728077233e-06, + "loss": 0.8949, + "step": 416 + }, + { + "epoch": 0.6034732272069464, + "grad_norm": 1.3842711448669434, + "learning_rate": 3.954143201930813e-06, + "loss": 0.8853, + "step": 417 + }, + { + "epoch": 0.6049204052098408, + "grad_norm": 1.319642186164856, + "learning_rate": 3.9501206757843926e-06, + "loss": 0.9459, + "step": 418 + }, + { + "epoch": 0.6063675832127352, + "grad_norm": 1.5316814184188843, + "learning_rate": 3.946098149637973e-06, + "loss": 0.9227, + "step": 419 + }, + { + "epoch": 0.6078147612156295, + "grad_norm": 1.1355427503585815, + "learning_rate": 3.942075623491553e-06, + "loss": 0.8777, + "step": 420 + }, + { + "epoch": 0.6078147612156295, + "eval_loss": 1.006725788116455, + "eval_runtime": 46.695, + "eval_samples_per_second": 21.416, + "eval_steps_per_second": 1.349, + "step": 420 + }, + { + "epoch": 0.6092619392185239, + "grad_norm": 1.938143014907837, + "learning_rate": 3.938053097345133e-06, + "loss": 0.8956, + "step": 421 + }, + { + "epoch": 0.6107091172214182, + "grad_norm": 1.4127334356307983, + "learning_rate": 3.934030571198713e-06, + "loss": 0.9204, + "step": 422 + }, + { + "epoch": 0.6121562952243126, + "grad_norm": 1.1328880786895752, + "learning_rate": 3.930008045052293e-06, + "loss": 0.8967, + "step": 423 + }, + { + "epoch": 0.613603473227207, + "grad_norm": 1.3078372478485107, + "learning_rate": 3.925985518905873e-06, + "loss": 0.9533, + "step": 424 + }, + { + "epoch": 0.6150506512301013, + "grad_norm": 1.4185065031051636, + "learning_rate": 3.9219629927594534e-06, + "loss": 0.9157, + "step": 425 + }, + { + "epoch": 0.6164978292329957, + "grad_norm": 1.4606282711029053, + "learning_rate": 3.9179404666130336e-06, + "loss": 0.9015, + "step": 426 + }, + { + "epoch": 0.61794500723589, + "grad_norm": 1.3323071002960205, + "learning_rate": 3.913917940466614e-06, + "loss": 0.8871, + "step": 427 + }, + { + "epoch": 0.6193921852387844, + "grad_norm": 1.0887928009033203, + "learning_rate": 3.909895414320193e-06, + "loss": 0.8892, + "step": 428 + }, + { + "epoch": 0.6208393632416788, + "grad_norm": 1.3578531742095947, + "learning_rate": 3.905872888173773e-06, + "loss": 0.8988, + "step": 429 + }, + { + "epoch": 0.622286541244573, + "grad_norm": 1.2270028591156006, + "learning_rate": 3.901850362027353e-06, + "loss": 0.901, + "step": 430 + }, + { + "epoch": 0.622286541244573, + "eval_loss": 1.0063830614089966, + "eval_runtime": 46.6354, + "eval_samples_per_second": 21.443, + "eval_steps_per_second": 1.351, + "step": 430 + }, + { + "epoch": 0.6237337192474675, + "grad_norm": 1.1941803693771362, + "learning_rate": 3.897827835880933e-06, + "loss": 0.8984, + "step": 431 + }, + { + "epoch": 0.6251808972503617, + "grad_norm": 1.2198870182037354, + "learning_rate": 3.8938053097345135e-06, + "loss": 0.9103, + "step": 432 + }, + { + "epoch": 0.6266280752532561, + "grad_norm": 1.2568440437316895, + "learning_rate": 3.889782783588094e-06, + "loss": 0.9201, + "step": 433 + }, + { + "epoch": 0.6280752532561505, + "grad_norm": 1.345555067062378, + "learning_rate": 3.885760257441674e-06, + "loss": 0.9466, + "step": 434 + }, + { + "epoch": 0.6295224312590448, + "grad_norm": 1.2755234241485596, + "learning_rate": 3.881737731295254e-06, + "loss": 0.8775, + "step": 435 + }, + { + "epoch": 0.6309696092619392, + "grad_norm": 1.2727750539779663, + "learning_rate": 3.877715205148834e-06, + "loss": 0.8926, + "step": 436 + }, + { + "epoch": 0.6324167872648335, + "grad_norm": 1.5701411962509155, + "learning_rate": 3.873692679002414e-06, + "loss": 0.9386, + "step": 437 + }, + { + "epoch": 0.6338639652677279, + "grad_norm": 1.8770571947097778, + "learning_rate": 3.869670152855993e-06, + "loss": 0.9532, + "step": 438 + }, + { + "epoch": 0.6353111432706223, + "grad_norm": 1.4552773237228394, + "learning_rate": 3.8656476267095735e-06, + "loss": 0.9218, + "step": 439 + }, + { + "epoch": 0.6367583212735166, + "grad_norm": 1.5343029499053955, + "learning_rate": 3.861625100563154e-06, + "loss": 0.8937, + "step": 440 + }, + { + "epoch": 0.6367583212735166, + "eval_loss": 1.0005669593811035, + "eval_runtime": 46.6277, + "eval_samples_per_second": 21.446, + "eval_steps_per_second": 1.351, + "step": 440 + }, + { + "epoch": 0.638205499276411, + "grad_norm": 1.2671089172363281, + "learning_rate": 3.857602574416734e-06, + "loss": 0.8781, + "step": 441 + }, + { + "epoch": 0.6396526772793053, + "grad_norm": 1.6595746278762817, + "learning_rate": 3.853580048270314e-06, + "loss": 0.9227, + "step": 442 + }, + { + "epoch": 0.6410998552821997, + "grad_norm": 1.29404878616333, + "learning_rate": 3.849557522123894e-06, + "loss": 0.9106, + "step": 443 + }, + { + "epoch": 0.6425470332850941, + "grad_norm": 1.3093692064285278, + "learning_rate": 3.845534995977474e-06, + "loss": 0.8871, + "step": 444 + }, + { + "epoch": 0.6439942112879884, + "grad_norm": 1.3726588487625122, + "learning_rate": 3.841512469831054e-06, + "loss": 0.8914, + "step": 445 + }, + { + "epoch": 0.6454413892908828, + "grad_norm": 1.6478410959243774, + "learning_rate": 3.8374899436846344e-06, + "loss": 0.8751, + "step": 446 + }, + { + "epoch": 0.6468885672937771, + "grad_norm": 1.2585914134979248, + "learning_rate": 3.8334674175382146e-06, + "loss": 0.9428, + "step": 447 + }, + { + "epoch": 0.6483357452966715, + "grad_norm": 1.5675022602081299, + "learning_rate": 3.829444891391794e-06, + "loss": 0.896, + "step": 448 + }, + { + "epoch": 0.6497829232995659, + "grad_norm": 1.545560598373413, + "learning_rate": 3.825422365245374e-06, + "loss": 0.9417, + "step": 449 + }, + { + "epoch": 0.6512301013024602, + "grad_norm": 1.4318615198135376, + "learning_rate": 3.821399839098954e-06, + "loss": 0.9243, + "step": 450 + }, + { + "epoch": 0.6512301013024602, + "eval_loss": 1.0001108646392822, + "eval_runtime": 46.6085, + "eval_samples_per_second": 21.455, + "eval_steps_per_second": 1.352, + "step": 450 + }, + { + "epoch": 0.6526772793053546, + "grad_norm": 1.8627666234970093, + "learning_rate": 3.817377312952534e-06, + "loss": 0.9075, + "step": 451 + }, + { + "epoch": 0.6541244573082489, + "grad_norm": 1.4701703786849976, + "learning_rate": 3.8133547868061144e-06, + "loss": 0.901, + "step": 452 + }, + { + "epoch": 0.6555716353111433, + "grad_norm": 1.4974210262298584, + "learning_rate": 3.8093322606596945e-06, + "loss": 0.8967, + "step": 453 + }, + { + "epoch": 0.6570188133140377, + "grad_norm": 1.4413353204727173, + "learning_rate": 3.8053097345132746e-06, + "loss": 0.8832, + "step": 454 + }, + { + "epoch": 0.658465991316932, + "grad_norm": 1.5375796556472778, + "learning_rate": 3.8012872083668543e-06, + "loss": 0.8656, + "step": 455 + }, + { + "epoch": 0.6599131693198264, + "grad_norm": 1.5743008852005005, + "learning_rate": 3.7972646822204344e-06, + "loss": 0.9229, + "step": 456 + }, + { + "epoch": 0.6613603473227206, + "grad_norm": 4.570529937744141, + "learning_rate": 3.7932421560740146e-06, + "loss": 0.9145, + "step": 457 + }, + { + "epoch": 0.662807525325615, + "grad_norm": 1.2687451839447021, + "learning_rate": 3.7892196299275947e-06, + "loss": 0.9156, + "step": 458 + }, + { + "epoch": 0.6642547033285094, + "grad_norm": 1.436997413635254, + "learning_rate": 3.785197103781175e-06, + "loss": 0.8993, + "step": 459 + }, + { + "epoch": 0.6657018813314037, + "grad_norm": 1.5583665370941162, + "learning_rate": 3.7811745776347545e-06, + "loss": 0.9125, + "step": 460 + }, + { + "epoch": 0.6657018813314037, + "eval_loss": 0.999251127243042, + "eval_runtime": 46.6557, + "eval_samples_per_second": 21.434, + "eval_steps_per_second": 1.35, + "step": 460 + }, + { + "epoch": 0.6671490593342981, + "grad_norm": 1.3632088899612427, + "learning_rate": 3.7771520514883347e-06, + "loss": 0.9232, + "step": 461 + }, + { + "epoch": 0.6685962373371924, + "grad_norm": 1.4672008752822876, + "learning_rate": 3.7731295253419148e-06, + "loss": 0.9287, + "step": 462 + }, + { + "epoch": 0.6700434153400868, + "grad_norm": 1.3250764608383179, + "learning_rate": 3.769106999195495e-06, + "loss": 0.8967, + "step": 463 + }, + { + "epoch": 0.6714905933429812, + "grad_norm": 1.344111442565918, + "learning_rate": 3.765084473049075e-06, + "loss": 0.9347, + "step": 464 + }, + { + "epoch": 0.6729377713458755, + "grad_norm": 1.389032006263733, + "learning_rate": 3.7610619469026547e-06, + "loss": 0.9324, + "step": 465 + }, + { + "epoch": 0.6743849493487699, + "grad_norm": 1.383352518081665, + "learning_rate": 3.757039420756235e-06, + "loss": 0.8871, + "step": 466 + }, + { + "epoch": 0.6758321273516642, + "grad_norm": 1.3043408393859863, + "learning_rate": 3.753016894609815e-06, + "loss": 0.8968, + "step": 467 + }, + { + "epoch": 0.6772793053545586, + "grad_norm": 1.2764406204223633, + "learning_rate": 3.7489943684633956e-06, + "loss": 0.8922, + "step": 468 + }, + { + "epoch": 0.678726483357453, + "grad_norm": 1.1833993196487427, + "learning_rate": 3.7449718423169757e-06, + "loss": 0.8933, + "step": 469 + }, + { + "epoch": 0.6801736613603473, + "grad_norm": 1.4321240186691284, + "learning_rate": 3.740949316170556e-06, + "loss": 0.9263, + "step": 470 + }, + { + "epoch": 0.6801736613603473, + "eval_loss": 0.9993090629577637, + "eval_runtime": 46.5834, + "eval_samples_per_second": 21.467, + "eval_steps_per_second": 1.352, + "step": 470 + }, + { + "epoch": 0.6816208393632417, + "grad_norm": 1.4045946598052979, + "learning_rate": 3.7369267900241355e-06, + "loss": 0.9396, + "step": 471 + }, + { + "epoch": 0.683068017366136, + "grad_norm": 1.3511790037155151, + "learning_rate": 3.7329042638777156e-06, + "loss": 0.9089, + "step": 472 + }, + { + "epoch": 0.6845151953690304, + "grad_norm": 1.4561907052993774, + "learning_rate": 3.7288817377312958e-06, + "loss": 0.8948, + "step": 473 + }, + { + "epoch": 0.6859623733719248, + "grad_norm": 1.2890348434448242, + "learning_rate": 3.724859211584876e-06, + "loss": 0.9276, + "step": 474 + }, + { + "epoch": 0.6874095513748191, + "grad_norm": 1.3402453660964966, + "learning_rate": 3.720836685438456e-06, + "loss": 0.9147, + "step": 475 + }, + { + "epoch": 0.6888567293777135, + "grad_norm": 1.5784218311309814, + "learning_rate": 3.7168141592920357e-06, + "loss": 0.8824, + "step": 476 + }, + { + "epoch": 0.6903039073806078, + "grad_norm": 1.5682567358016968, + "learning_rate": 3.712791633145616e-06, + "loss": 0.905, + "step": 477 + }, + { + "epoch": 0.6917510853835022, + "grad_norm": 1.2175792455673218, + "learning_rate": 3.708769106999196e-06, + "loss": 0.8704, + "step": 478 + }, + { + "epoch": 0.6931982633863966, + "grad_norm": 1.451760172843933, + "learning_rate": 3.704746580852776e-06, + "loss": 0.892, + "step": 479 + }, + { + "epoch": 0.6946454413892909, + "grad_norm": 1.422426462173462, + "learning_rate": 3.7007240547063562e-06, + "loss": 0.9205, + "step": 480 + }, + { + "epoch": 0.6946454413892909, + "eval_loss": 0.9962843656539917, + "eval_runtime": 46.6209, + "eval_samples_per_second": 21.45, + "eval_steps_per_second": 1.351, + "step": 480 + }, + { + "epoch": 0.6960926193921853, + "grad_norm": 1.2854764461517334, + "learning_rate": 3.696701528559936e-06, + "loss": 0.846, + "step": 481 + }, + { + "epoch": 0.6975397973950795, + "grad_norm": 1.509216547012329, + "learning_rate": 3.692679002413516e-06, + "loss": 0.8877, + "step": 482 + }, + { + "epoch": 0.6989869753979739, + "grad_norm": 1.3205519914627075, + "learning_rate": 3.688656476267096e-06, + "loss": 0.8869, + "step": 483 + }, + { + "epoch": 0.7004341534008683, + "grad_norm": 1.4966216087341309, + "learning_rate": 3.6846339501206763e-06, + "loss": 0.9168, + "step": 484 + }, + { + "epoch": 0.7018813314037626, + "grad_norm": 1.158457636833191, + "learning_rate": 3.6806114239742565e-06, + "loss": 0.8933, + "step": 485 + }, + { + "epoch": 0.703328509406657, + "grad_norm": 1.2933553457260132, + "learning_rate": 3.676588897827836e-06, + "loss": 0.926, + "step": 486 + }, + { + "epoch": 0.7047756874095513, + "grad_norm": 1.3354488611221313, + "learning_rate": 3.6725663716814163e-06, + "loss": 0.8915, + "step": 487 + }, + { + "epoch": 0.7062228654124457, + "grad_norm": 1.40239679813385, + "learning_rate": 3.6685438455349964e-06, + "loss": 0.8865, + "step": 488 + }, + { + "epoch": 0.7076700434153401, + "grad_norm": 1.4706095457077026, + "learning_rate": 3.6645213193885765e-06, + "loss": 0.8804, + "step": 489 + }, + { + "epoch": 0.7091172214182344, + "grad_norm": 1.2073744535446167, + "learning_rate": 3.6604987932421567e-06, + "loss": 0.9049, + "step": 490 + }, + { + "epoch": 0.7091172214182344, + "eval_loss": 0.9954089522361755, + "eval_runtime": 46.6761, + "eval_samples_per_second": 21.424, + "eval_steps_per_second": 1.35, + "step": 490 + }, + { + "epoch": 0.7105643994211288, + "grad_norm": 1.5668460130691528, + "learning_rate": 3.6564762670957364e-06, + "loss": 0.9251, + "step": 491 + }, + { + "epoch": 0.7120115774240231, + "grad_norm": 1.446851134300232, + "learning_rate": 3.6524537409493165e-06, + "loss": 0.9503, + "step": 492 + }, + { + "epoch": 0.7134587554269175, + "grad_norm": 1.515219807624817, + "learning_rate": 3.6484312148028966e-06, + "loss": 0.914, + "step": 493 + }, + { + "epoch": 0.7149059334298119, + "grad_norm": 1.3325178623199463, + "learning_rate": 3.6444086886564768e-06, + "loss": 0.899, + "step": 494 + }, + { + "epoch": 0.7163531114327062, + "grad_norm": 1.225527048110962, + "learning_rate": 3.640386162510057e-06, + "loss": 0.8814, + "step": 495 + }, + { + "epoch": 0.7178002894356006, + "grad_norm": 1.7016558647155762, + "learning_rate": 3.6363636363636366e-06, + "loss": 0.8743, + "step": 496 + }, + { + "epoch": 0.7192474674384949, + "grad_norm": 1.2231067419052124, + "learning_rate": 3.6323411102172167e-06, + "loss": 0.8708, + "step": 497 + }, + { + "epoch": 0.7206946454413893, + "grad_norm": 1.417515754699707, + "learning_rate": 3.628318584070797e-06, + "loss": 0.8725, + "step": 498 + }, + { + "epoch": 0.7221418234442837, + "grad_norm": 1.4536945819854736, + "learning_rate": 3.624296057924377e-06, + "loss": 0.9165, + "step": 499 + }, + { + "epoch": 0.723589001447178, + "grad_norm": 1.4294936656951904, + "learning_rate": 3.620273531777957e-06, + "loss": 0.8928, + "step": 500 + }, + { + "epoch": 0.723589001447178, + "eval_loss": 0.9940905570983887, + "eval_runtime": 46.6641, + "eval_samples_per_second": 21.43, + "eval_steps_per_second": 1.35, + "step": 500 + }, + { + "epoch": 0.7250361794500724, + "grad_norm": 1.5148828029632568, + "learning_rate": 3.616251005631537e-06, + "loss": 0.8738, + "step": 501 + }, + { + "epoch": 0.7264833574529667, + "grad_norm": 1.629025936126709, + "learning_rate": 3.612228479485117e-06, + "loss": 0.9254, + "step": 502 + }, + { + "epoch": 0.7279305354558611, + "grad_norm": 1.6170278787612915, + "learning_rate": 3.608205953338697e-06, + "loss": 0.8918, + "step": 503 + }, + { + "epoch": 0.7293777134587555, + "grad_norm": 1.3651961088180542, + "learning_rate": 3.604183427192277e-06, + "loss": 0.9093, + "step": 504 + }, + { + "epoch": 0.7308248914616498, + "grad_norm": 1.2708286046981812, + "learning_rate": 3.6001609010458573e-06, + "loss": 0.9002, + "step": 505 + }, + { + "epoch": 0.7322720694645442, + "grad_norm": 1.7087815999984741, + "learning_rate": 3.596138374899437e-06, + "loss": 0.882, + "step": 506 + }, + { + "epoch": 0.7337192474674384, + "grad_norm": 1.7053250074386597, + "learning_rate": 3.592115848753017e-06, + "loss": 0.8824, + "step": 507 + }, + { + "epoch": 0.7351664254703328, + "grad_norm": 1.339426875114441, + "learning_rate": 3.5880933226065973e-06, + "loss": 0.8872, + "step": 508 + }, + { + "epoch": 0.7366136034732272, + "grad_norm": 1.355403184890747, + "learning_rate": 3.5840707964601774e-06, + "loss": 0.8513, + "step": 509 + }, + { + "epoch": 0.7380607814761215, + "grad_norm": 1.6746940612792969, + "learning_rate": 3.5800482703137575e-06, + "loss": 0.9208, + "step": 510 + }, + { + "epoch": 0.7380607814761215, + "eval_loss": 0.9942624568939209, + "eval_runtime": 46.6951, + "eval_samples_per_second": 21.416, + "eval_steps_per_second": 1.349, + "step": 510 + }, + { + "epoch": 0.7395079594790159, + "grad_norm": 1.2381854057312012, + "learning_rate": 3.5760257441673372e-06, + "loss": 0.8953, + "step": 511 + }, + { + "epoch": 0.7409551374819102, + "grad_norm": 1.2436381578445435, + "learning_rate": 3.5720032180209174e-06, + "loss": 0.8881, + "step": 512 + }, + { + "epoch": 0.7424023154848046, + "grad_norm": 1.5799764394760132, + "learning_rate": 3.5679806918744975e-06, + "loss": 0.9215, + "step": 513 + }, + { + "epoch": 0.743849493487699, + "grad_norm": 1.5527855157852173, + "learning_rate": 3.5639581657280776e-06, + "loss": 0.887, + "step": 514 + }, + { + "epoch": 0.7452966714905933, + "grad_norm": 1.244868516921997, + "learning_rate": 3.5599356395816577e-06, + "loss": 0.8324, + "step": 515 + }, + { + "epoch": 0.7467438494934877, + "grad_norm": 1.3446401357650757, + "learning_rate": 3.5559131134352374e-06, + "loss": 0.8897, + "step": 516 + }, + { + "epoch": 0.748191027496382, + "grad_norm": 1.2377643585205078, + "learning_rate": 3.5518905872888176e-06, + "loss": 0.9177, + "step": 517 + }, + { + "epoch": 0.7496382054992764, + "grad_norm": 1.5362792015075684, + "learning_rate": 3.5478680611423977e-06, + "loss": 0.9297, + "step": 518 + }, + { + "epoch": 0.7510853835021708, + "grad_norm": 1.4541444778442383, + "learning_rate": 3.543845534995978e-06, + "loss": 0.8564, + "step": 519 + }, + { + "epoch": 0.7525325615050651, + "grad_norm": 1.3017665147781372, + "learning_rate": 3.539823008849558e-06, + "loss": 0.8807, + "step": 520 + }, + { + "epoch": 0.7525325615050651, + "eval_loss": 0.9944441914558411, + "eval_runtime": 46.7284, + "eval_samples_per_second": 21.4, + "eval_steps_per_second": 1.348, + "step": 520 + }, + { + "epoch": 0.7539797395079595, + "grad_norm": 1.4199774265289307, + "learning_rate": 3.5358004827031377e-06, + "loss": 0.9077, + "step": 521 + }, + { + "epoch": 0.7554269175108539, + "grad_norm": 1.7387546300888062, + "learning_rate": 3.5317779565567178e-06, + "loss": 0.8636, + "step": 522 + }, + { + "epoch": 0.7568740955137482, + "grad_norm": 1.4066709280014038, + "learning_rate": 3.527755430410298e-06, + "loss": 0.9112, + "step": 523 + }, + { + "epoch": 0.7583212735166426, + "grad_norm": 1.3991035223007202, + "learning_rate": 3.523732904263878e-06, + "loss": 0.8988, + "step": 524 + }, + { + "epoch": 0.7597684515195369, + "grad_norm": 1.6597838401794434, + "learning_rate": 3.519710378117458e-06, + "loss": 0.8897, + "step": 525 + }, + { + "epoch": 0.7612156295224313, + "grad_norm": 1.4131901264190674, + "learning_rate": 3.515687851971038e-06, + "loss": 0.8963, + "step": 526 + }, + { + "epoch": 0.7626628075253257, + "grad_norm": 1.355067491531372, + "learning_rate": 3.511665325824618e-06, + "loss": 0.9076, + "step": 527 + }, + { + "epoch": 0.76410998552822, + "grad_norm": 1.4299192428588867, + "learning_rate": 3.507642799678198e-06, + "loss": 0.8883, + "step": 528 + }, + { + "epoch": 0.7655571635311144, + "grad_norm": 1.5582046508789062, + "learning_rate": 3.5036202735317783e-06, + "loss": 0.8954, + "step": 529 + }, + { + "epoch": 0.7670043415340086, + "grad_norm": 1.9302220344543457, + "learning_rate": 3.4995977473853584e-06, + "loss": 0.8994, + "step": 530 + }, + { + "epoch": 0.7670043415340086, + "eval_loss": 0.9905989170074463, + "eval_runtime": 46.7553, + "eval_samples_per_second": 21.388, + "eval_steps_per_second": 1.347, + "step": 530 + }, + { + "epoch": 0.768451519536903, + "grad_norm": 1.3474477529525757, + "learning_rate": 3.495575221238938e-06, + "loss": 0.8819, + "step": 531 + }, + { + "epoch": 0.7698986975397974, + "grad_norm": 1.4210630655288696, + "learning_rate": 3.4915526950925182e-06, + "loss": 0.8689, + "step": 532 + }, + { + "epoch": 0.7713458755426917, + "grad_norm": 1.4004331827163696, + "learning_rate": 3.4875301689460983e-06, + "loss": 0.901, + "step": 533 + }, + { + "epoch": 0.7727930535455861, + "grad_norm": 1.6812330484390259, + "learning_rate": 3.4835076427996785e-06, + "loss": 0.8872, + "step": 534 + }, + { + "epoch": 0.7742402315484804, + "grad_norm": 1.190755844116211, + "learning_rate": 3.4794851166532586e-06, + "loss": 0.9033, + "step": 535 + }, + { + "epoch": 0.7756874095513748, + "grad_norm": 1.4481284618377686, + "learning_rate": 3.4754625905068383e-06, + "loss": 0.9047, + "step": 536 + }, + { + "epoch": 0.7771345875542692, + "grad_norm": 1.313199758529663, + "learning_rate": 3.4714400643604184e-06, + "loss": 0.8901, + "step": 537 + }, + { + "epoch": 0.7785817655571635, + "grad_norm": 1.495705485343933, + "learning_rate": 3.4674175382139986e-06, + "loss": 0.8924, + "step": 538 + }, + { + "epoch": 0.7800289435600579, + "grad_norm": 1.28499174118042, + "learning_rate": 3.4633950120675787e-06, + "loss": 0.8791, + "step": 539 + }, + { + "epoch": 0.7814761215629522, + "grad_norm": 1.6623579263687134, + "learning_rate": 3.4593724859211584e-06, + "loss": 0.9076, + "step": 540 + }, + { + "epoch": 0.7814761215629522, + "eval_loss": 0.9871039986610413, + "eval_runtime": 46.4345, + "eval_samples_per_second": 21.536, + "eval_steps_per_second": 1.357, + "step": 540 + }, + { + "epoch": 0.7829232995658466, + "grad_norm": 1.4080674648284912, + "learning_rate": 3.4553499597747385e-06, + "loss": 0.9175, + "step": 541 + }, + { + "epoch": 0.784370477568741, + "grad_norm": 1.557834506034851, + "learning_rate": 3.4513274336283186e-06, + "loss": 0.8892, + "step": 542 + }, + { + "epoch": 0.7858176555716353, + "grad_norm": 1.423572301864624, + "learning_rate": 3.4473049074818988e-06, + "loss": 0.8809, + "step": 543 + }, + { + "epoch": 0.7872648335745297, + "grad_norm": 1.7118240594863892, + "learning_rate": 3.443282381335479e-06, + "loss": 0.9049, + "step": 544 + }, + { + "epoch": 0.788712011577424, + "grad_norm": 1.235611915588379, + "learning_rate": 3.4392598551890586e-06, + "loss": 0.8655, + "step": 545 + }, + { + "epoch": 0.7901591895803184, + "grad_norm": 1.4629002809524536, + "learning_rate": 3.4352373290426387e-06, + "loss": 0.908, + "step": 546 + }, + { + "epoch": 0.7916063675832128, + "grad_norm": 1.289513111114502, + "learning_rate": 3.431214802896219e-06, + "loss": 0.8949, + "step": 547 + }, + { + "epoch": 0.7930535455861071, + "grad_norm": 1.6830641031265259, + "learning_rate": 3.427192276749799e-06, + "loss": 0.8826, + "step": 548 + }, + { + "epoch": 0.7945007235890015, + "grad_norm": 1.1464828252792358, + "learning_rate": 3.423169750603379e-06, + "loss": 0.9107, + "step": 549 + }, + { + "epoch": 0.7959479015918958, + "grad_norm": 1.4295430183410645, + "learning_rate": 3.419147224456959e-06, + "loss": 0.893, + "step": 550 + }, + { + "epoch": 0.7959479015918958, + "eval_loss": 0.9852504730224609, + "eval_runtime": 46.372, + "eval_samples_per_second": 21.565, + "eval_steps_per_second": 1.359, + "step": 550 + }, + { + "epoch": 0.7973950795947902, + "grad_norm": 1.4433420896530151, + "learning_rate": 3.415124698310539e-06, + "loss": 0.8684, + "step": 551 + }, + { + "epoch": 0.7988422575976846, + "grad_norm": 1.117049217224121, + "learning_rate": 3.411102172164119e-06, + "loss": 0.9086, + "step": 552 + }, + { + "epoch": 0.8002894356005789, + "grad_norm": 1.2112184762954712, + "learning_rate": 3.407079646017699e-06, + "loss": 0.8873, + "step": 553 + }, + { + "epoch": 0.8017366136034733, + "grad_norm": 1.454324722290039, + "learning_rate": 3.4030571198712793e-06, + "loss": 0.8721, + "step": 554 + }, + { + "epoch": 0.8031837916063675, + "grad_norm": 1.1535415649414062, + "learning_rate": 3.399034593724859e-06, + "loss": 0.8715, + "step": 555 + }, + { + "epoch": 0.804630969609262, + "grad_norm": 1.5421648025512695, + "learning_rate": 3.395012067578439e-06, + "loss": 0.876, + "step": 556 + }, + { + "epoch": 0.8060781476121563, + "grad_norm": 1.1396689414978027, + "learning_rate": 3.3909895414320197e-06, + "loss": 0.8689, + "step": 557 + }, + { + "epoch": 0.8075253256150506, + "grad_norm": 1.276900053024292, + "learning_rate": 3.3869670152856e-06, + "loss": 0.8774, + "step": 558 + }, + { + "epoch": 0.808972503617945, + "grad_norm": 1.2561426162719727, + "learning_rate": 3.38294448913918e-06, + "loss": 0.9241, + "step": 559 + }, + { + "epoch": 0.8104196816208393, + "grad_norm": 1.4037206172943115, + "learning_rate": 3.37892196299276e-06, + "loss": 0.9015, + "step": 560 + }, + { + "epoch": 0.8104196816208393, + "eval_loss": 0.9847711324691772, + "eval_runtime": 46.3811, + "eval_samples_per_second": 21.561, + "eval_steps_per_second": 1.358, + "step": 560 + }, + { + "epoch": 0.8118668596237337, + "grad_norm": 1.4091521501541138, + "learning_rate": 3.3748994368463402e-06, + "loss": 0.9033, + "step": 561 + }, + { + "epoch": 0.8133140376266281, + "grad_norm": 1.1932907104492188, + "learning_rate": 3.37087691069992e-06, + "loss": 0.8653, + "step": 562 + }, + { + "epoch": 0.8147612156295224, + "grad_norm": 1.7423676252365112, + "learning_rate": 3.3668543845535e-06, + "loss": 0.8832, + "step": 563 + }, + { + "epoch": 0.8162083936324168, + "grad_norm": 1.396618366241455, + "learning_rate": 3.36283185840708e-06, + "loss": 0.8814, + "step": 564 + }, + { + "epoch": 0.8176555716353111, + "grad_norm": 1.4076389074325562, + "learning_rate": 3.3588093322606603e-06, + "loss": 0.8804, + "step": 565 + }, + { + "epoch": 0.8191027496382055, + "grad_norm": 1.3262157440185547, + "learning_rate": 3.3547868061142404e-06, + "loss": 0.8663, + "step": 566 + }, + { + "epoch": 0.8205499276410999, + "grad_norm": 1.4094736576080322, + "learning_rate": 3.35076427996782e-06, + "loss": 0.8764, + "step": 567 + }, + { + "epoch": 0.8219971056439942, + "grad_norm": 1.579625129699707, + "learning_rate": 3.3467417538214003e-06, + "loss": 0.8833, + "step": 568 + }, + { + "epoch": 0.8234442836468886, + "grad_norm": 1.4347237348556519, + "learning_rate": 3.3427192276749804e-06, + "loss": 0.8946, + "step": 569 + }, + { + "epoch": 0.8248914616497829, + "grad_norm": 1.2382946014404297, + "learning_rate": 3.3386967015285605e-06, + "loss": 0.8988, + "step": 570 + }, + { + "epoch": 0.8248914616497829, + "eval_loss": 0.9849082827568054, + "eval_runtime": 46.3879, + "eval_samples_per_second": 21.557, + "eval_steps_per_second": 1.358, + "step": 570 + }, + { + "epoch": 0.8263386396526773, + "grad_norm": 1.2935632467269897, + "learning_rate": 3.3346741753821407e-06, + "loss": 0.9002, + "step": 571 + }, + { + "epoch": 0.8277858176555717, + "grad_norm": 1.3754832744598389, + "learning_rate": 3.3306516492357204e-06, + "loss": 0.9302, + "step": 572 + }, + { + "epoch": 0.829232995658466, + "grad_norm": 1.5214083194732666, + "learning_rate": 3.3266291230893005e-06, + "loss": 0.8799, + "step": 573 + }, + { + "epoch": 0.8306801736613604, + "grad_norm": 1.3328135013580322, + "learning_rate": 3.3226065969428806e-06, + "loss": 0.8967, + "step": 574 + }, + { + "epoch": 0.8321273516642547, + "grad_norm": 1.3815909624099731, + "learning_rate": 3.3185840707964607e-06, + "loss": 0.8667, + "step": 575 + }, + { + "epoch": 0.8335745296671491, + "grad_norm": 1.3776534795761108, + "learning_rate": 3.3145615446500404e-06, + "loss": 0.8812, + "step": 576 + }, + { + "epoch": 0.8350217076700435, + "grad_norm": 1.169659972190857, + "learning_rate": 3.3105390185036206e-06, + "loss": 0.8388, + "step": 577 + }, + { + "epoch": 0.8364688856729378, + "grad_norm": 1.181516170501709, + "learning_rate": 3.3065164923572007e-06, + "loss": 0.874, + "step": 578 + }, + { + "epoch": 0.8379160636758322, + "grad_norm": 1.2207849025726318, + "learning_rate": 3.302493966210781e-06, + "loss": 0.8901, + "step": 579 + }, + { + "epoch": 0.8393632416787264, + "grad_norm": 1.3702313899993896, + "learning_rate": 3.298471440064361e-06, + "loss": 0.9063, + "step": 580 + }, + { + "epoch": 0.8393632416787264, + "eval_loss": 0.985295295715332, + "eval_runtime": 46.4449, + "eval_samples_per_second": 21.531, + "eval_steps_per_second": 1.356, + "step": 580 + }, + { + "epoch": 0.8408104196816208, + "grad_norm": 1.5399174690246582, + "learning_rate": 3.2944489139179407e-06, + "loss": 0.8648, + "step": 581 + }, + { + "epoch": 0.8422575976845152, + "grad_norm": 1.5093787908554077, + "learning_rate": 3.2904263877715208e-06, + "loss": 0.8604, + "step": 582 + }, + { + "epoch": 0.8437047756874095, + "grad_norm": 1.253140926361084, + "learning_rate": 3.286403861625101e-06, + "loss": 0.9052, + "step": 583 + }, + { + "epoch": 0.8451519536903039, + "grad_norm": 1.2082233428955078, + "learning_rate": 3.282381335478681e-06, + "loss": 0.9137, + "step": 584 + }, + { + "epoch": 0.8465991316931982, + "grad_norm": 1.3378984928131104, + "learning_rate": 3.278358809332261e-06, + "loss": 0.8604, + "step": 585 + }, + { + "epoch": 0.8480463096960926, + "grad_norm": 1.3688184022903442, + "learning_rate": 3.274336283185841e-06, + "loss": 0.891, + "step": 586 + }, + { + "epoch": 0.849493487698987, + "grad_norm": 1.265341877937317, + "learning_rate": 3.270313757039421e-06, + "loss": 0.8675, + "step": 587 + }, + { + "epoch": 0.8509406657018813, + "grad_norm": 1.5762468576431274, + "learning_rate": 3.266291230893001e-06, + "loss": 0.9008, + "step": 588 + }, + { + "epoch": 0.8523878437047757, + "grad_norm": 1.4196147918701172, + "learning_rate": 3.2622687047465813e-06, + "loss": 0.8938, + "step": 589 + }, + { + "epoch": 0.85383502170767, + "grad_norm": 1.437941551208496, + "learning_rate": 3.2582461786001614e-06, + "loss": 0.9164, + "step": 590 + }, + { + "epoch": 0.85383502170767, + "eval_loss": 0.98630291223526, + "eval_runtime": 46.4002, + "eval_samples_per_second": 21.552, + "eval_steps_per_second": 1.358, + "step": 590 + }, + { + "epoch": 0.8552821997105644, + "grad_norm": 1.8454606533050537, + "learning_rate": 3.254223652453741e-06, + "loss": 0.8681, + "step": 591 + }, + { + "epoch": 0.8567293777134588, + "grad_norm": 1.508359670639038, + "learning_rate": 3.2502011263073212e-06, + "loss": 0.8834, + "step": 592 + }, + { + "epoch": 0.8581765557163531, + "grad_norm": 1.6999026536941528, + "learning_rate": 3.2461786001609013e-06, + "loss": 0.9007, + "step": 593 + }, + { + "epoch": 0.8596237337192475, + "grad_norm": 1.1893867254257202, + "learning_rate": 3.2421560740144815e-06, + "loss": 0.8616, + "step": 594 + }, + { + "epoch": 0.8610709117221418, + "grad_norm": 1.5587775707244873, + "learning_rate": 3.2381335478680616e-06, + "loss": 0.8681, + "step": 595 + }, + { + "epoch": 0.8625180897250362, + "grad_norm": 1.4295573234558105, + "learning_rate": 3.2341110217216413e-06, + "loss": 0.8854, + "step": 596 + }, + { + "epoch": 0.8639652677279306, + "grad_norm": 1.1666234731674194, + "learning_rate": 3.2300884955752214e-06, + "loss": 0.8934, + "step": 597 + }, + { + "epoch": 0.8654124457308249, + "grad_norm": 1.317785620689392, + "learning_rate": 3.2260659694288016e-06, + "loss": 0.8827, + "step": 598 + }, + { + "epoch": 0.8668596237337193, + "grad_norm": 1.7967166900634766, + "learning_rate": 3.2220434432823817e-06, + "loss": 0.8659, + "step": 599 + }, + { + "epoch": 0.8683068017366136, + "grad_norm": 1.506920337677002, + "learning_rate": 3.218020917135962e-06, + "loss": 0.8722, + "step": 600 + }, + { + "epoch": 0.8683068017366136, + "eval_loss": 0.9865502715110779, + "eval_runtime": 46.3828, + "eval_samples_per_second": 21.56, + "eval_steps_per_second": 1.358, + "step": 600 + }, + { + "epoch": 0.869753979739508, + "grad_norm": 1.5308632850646973, + "learning_rate": 3.2139983909895415e-06, + "loss": 0.8912, + "step": 601 + }, + { + "epoch": 0.8712011577424024, + "grad_norm": 1.5166269540786743, + "learning_rate": 3.2099758648431216e-06, + "loss": 0.8565, + "step": 602 + }, + { + "epoch": 0.8726483357452967, + "grad_norm": 1.5561326742172241, + "learning_rate": 3.2059533386967018e-06, + "loss": 0.8938, + "step": 603 + }, + { + "epoch": 0.874095513748191, + "grad_norm": 1.3041249513626099, + "learning_rate": 3.201930812550282e-06, + "loss": 0.8656, + "step": 604 + }, + { + "epoch": 0.8755426917510853, + "grad_norm": 1.4048686027526855, + "learning_rate": 3.197908286403862e-06, + "loss": 0.8771, + "step": 605 + }, + { + "epoch": 0.8769898697539797, + "grad_norm": 1.3990355730056763, + "learning_rate": 3.1938857602574417e-06, + "loss": 0.8698, + "step": 606 + }, + { + "epoch": 0.8784370477568741, + "grad_norm": 1.594112515449524, + "learning_rate": 3.189863234111022e-06, + "loss": 0.8461, + "step": 607 + }, + { + "epoch": 0.8798842257597684, + "grad_norm": 1.1919422149658203, + "learning_rate": 3.185840707964602e-06, + "loss": 0.8573, + "step": 608 + }, + { + "epoch": 0.8813314037626628, + "grad_norm": 1.3675607442855835, + "learning_rate": 3.181818181818182e-06, + "loss": 0.8746, + "step": 609 + }, + { + "epoch": 0.8827785817655571, + "grad_norm": 1.187645435333252, + "learning_rate": 3.1777956556717622e-06, + "loss": 0.8367, + "step": 610 + }, + { + "epoch": 0.8827785817655571, + "eval_loss": 0.9847010970115662, + "eval_runtime": 46.4036, + "eval_samples_per_second": 21.55, + "eval_steps_per_second": 1.358, + "step": 610 + }, + { + "epoch": 0.8842257597684515, + "grad_norm": 1.405061960220337, + "learning_rate": 3.173773129525342e-06, + "loss": 0.8668, + "step": 611 + }, + { + "epoch": 0.8856729377713459, + "grad_norm": 1.469196081161499, + "learning_rate": 3.169750603378922e-06, + "loss": 0.8807, + "step": 612 + }, + { + "epoch": 0.8871201157742402, + "grad_norm": 1.2013216018676758, + "learning_rate": 3.165728077232502e-06, + "loss": 0.8923, + "step": 613 + }, + { + "epoch": 0.8885672937771346, + "grad_norm": 1.2316696643829346, + "learning_rate": 3.1617055510860823e-06, + "loss": 0.8456, + "step": 614 + }, + { + "epoch": 0.8900144717800289, + "grad_norm": 1.7709813117980957, + "learning_rate": 3.1576830249396625e-06, + "loss": 0.8706, + "step": 615 + }, + { + "epoch": 0.8914616497829233, + "grad_norm": 1.2025177478790283, + "learning_rate": 3.153660498793242e-06, + "loss": 0.8911, + "step": 616 + }, + { + "epoch": 0.8929088277858177, + "grad_norm": 1.3831732273101807, + "learning_rate": 3.1496379726468223e-06, + "loss": 0.8716, + "step": 617 + }, + { + "epoch": 0.894356005788712, + "grad_norm": 1.50855553150177, + "learning_rate": 3.1456154465004024e-06, + "loss": 0.8758, + "step": 618 + }, + { + "epoch": 0.8958031837916064, + "grad_norm": 1.6425405740737915, + "learning_rate": 3.1415929203539825e-06, + "loss": 0.901, + "step": 619 + }, + { + "epoch": 0.8972503617945007, + "grad_norm": 1.6914435625076294, + "learning_rate": 3.1375703942075627e-06, + "loss": 0.8955, + "step": 620 + }, + { + "epoch": 0.8972503617945007, + "eval_loss": 0.9835866689682007, + "eval_runtime": 46.3664, + "eval_samples_per_second": 21.567, + "eval_steps_per_second": 1.359, + "step": 620 + }, + { + "epoch": 0.8986975397973951, + "grad_norm": 1.2449274063110352, + "learning_rate": 3.1335478680611424e-06, + "loss": 0.9048, + "step": 621 + }, + { + "epoch": 0.9001447178002895, + "grad_norm": 1.5092636346817017, + "learning_rate": 3.1295253419147225e-06, + "loss": 0.8935, + "step": 622 + }, + { + "epoch": 0.9015918958031838, + "grad_norm": 1.299573540687561, + "learning_rate": 3.1255028157683026e-06, + "loss": 0.9109, + "step": 623 + }, + { + "epoch": 0.9030390738060782, + "grad_norm": 1.3478418588638306, + "learning_rate": 3.1214802896218828e-06, + "loss": 0.8634, + "step": 624 + }, + { + "epoch": 0.9044862518089725, + "grad_norm": 1.2084031105041504, + "learning_rate": 3.117457763475463e-06, + "loss": 0.8478, + "step": 625 + }, + { + "epoch": 0.9059334298118669, + "grad_norm": 1.474709153175354, + "learning_rate": 3.1134352373290426e-06, + "loss": 0.8904, + "step": 626 + }, + { + "epoch": 0.9073806078147613, + "grad_norm": 1.4116274118423462, + "learning_rate": 3.1094127111826227e-06, + "loss": 0.9116, + "step": 627 + }, + { + "epoch": 0.9088277858176556, + "grad_norm": 1.4768096208572388, + "learning_rate": 3.105390185036203e-06, + "loss": 0.8672, + "step": 628 + }, + { + "epoch": 0.91027496382055, + "grad_norm": 1.2628055810928345, + "learning_rate": 3.101367658889783e-06, + "loss": 0.8758, + "step": 629 + }, + { + "epoch": 0.9117221418234442, + "grad_norm": 1.4072173833847046, + "learning_rate": 3.097345132743363e-06, + "loss": 0.8315, + "step": 630 + }, + { + "epoch": 0.9117221418234442, + "eval_loss": 0.9802346229553223, + "eval_runtime": 46.4529, + "eval_samples_per_second": 21.527, + "eval_steps_per_second": 1.356, + "step": 630 + }, + { + "epoch": 0.9131693198263386, + "grad_norm": 1.2694050073623657, + "learning_rate": 3.093322606596943e-06, + "loss": 0.9121, + "step": 631 + }, + { + "epoch": 0.914616497829233, + "grad_norm": 1.1973212957382202, + "learning_rate": 3.089300080450523e-06, + "loss": 0.8961, + "step": 632 + }, + { + "epoch": 0.9160636758321273, + "grad_norm": 1.311354637145996, + "learning_rate": 3.085277554304103e-06, + "loss": 0.8789, + "step": 633 + }, + { + "epoch": 0.9175108538350217, + "grad_norm": 1.2725342512130737, + "learning_rate": 3.081255028157683e-06, + "loss": 0.859, + "step": 634 + }, + { + "epoch": 0.918958031837916, + "grad_norm": 1.327165126800537, + "learning_rate": 3.0772325020112633e-06, + "loss": 0.8804, + "step": 635 + }, + { + "epoch": 0.9204052098408104, + "grad_norm": 1.243758201599121, + "learning_rate": 3.073209975864843e-06, + "loss": 0.8651, + "step": 636 + }, + { + "epoch": 0.9218523878437048, + "grad_norm": 1.2164416313171387, + "learning_rate": 3.069187449718423e-06, + "loss": 0.8704, + "step": 637 + }, + { + "epoch": 0.9232995658465991, + "grad_norm": 1.2358460426330566, + "learning_rate": 3.0651649235720033e-06, + "loss": 0.9298, + "step": 638 + }, + { + "epoch": 0.9247467438494935, + "grad_norm": 1.238119125366211, + "learning_rate": 3.0611423974255834e-06, + "loss": 0.8943, + "step": 639 + }, + { + "epoch": 0.9261939218523878, + "grad_norm": 1.2019529342651367, + "learning_rate": 3.057119871279163e-06, + "loss": 0.8882, + "step": 640 + }, + { + "epoch": 0.9261939218523878, + "eval_loss": 0.980927586555481, + "eval_runtime": 46.3967, + "eval_samples_per_second": 21.553, + "eval_steps_per_second": 1.358, + "step": 640 + }, + { + "epoch": 0.9276410998552822, + "grad_norm": 1.3751319646835327, + "learning_rate": 3.0530973451327432e-06, + "loss": 0.8616, + "step": 641 + }, + { + "epoch": 0.9290882778581766, + "grad_norm": 1.4114649295806885, + "learning_rate": 3.0490748189863234e-06, + "loss": 0.8921, + "step": 642 + }, + { + "epoch": 0.9305354558610709, + "grad_norm": 1.1888560056686401, + "learning_rate": 3.0450522928399035e-06, + "loss": 0.884, + "step": 643 + }, + { + "epoch": 0.9319826338639653, + "grad_norm": 1.4184575080871582, + "learning_rate": 3.0410297666934836e-06, + "loss": 0.8668, + "step": 644 + }, + { + "epoch": 0.9334298118668596, + "grad_norm": 1.4550480842590332, + "learning_rate": 3.0370072405470633e-06, + "loss": 0.8765, + "step": 645 + }, + { + "epoch": 0.934876989869754, + "grad_norm": 1.264320969581604, + "learning_rate": 3.0329847144006443e-06, + "loss": 0.8515, + "step": 646 + }, + { + "epoch": 0.9363241678726484, + "grad_norm": 1.3218379020690918, + "learning_rate": 3.028962188254224e-06, + "loss": 0.8785, + "step": 647 + }, + { + "epoch": 0.9377713458755427, + "grad_norm": 1.3922557830810547, + "learning_rate": 3.024939662107804e-06, + "loss": 0.8629, + "step": 648 + }, + { + "epoch": 0.9392185238784371, + "grad_norm": 1.3875412940979004, + "learning_rate": 3.0209171359613843e-06, + "loss": 0.8862, + "step": 649 + }, + { + "epoch": 0.9406657018813314, + "grad_norm": 1.3726916313171387, + "learning_rate": 3.0168946098149644e-06, + "loss": 0.8642, + "step": 650 + }, + { + "epoch": 0.9406657018813314, + "eval_loss": 0.980999767780304, + "eval_runtime": 46.4623, + "eval_samples_per_second": 21.523, + "eval_steps_per_second": 1.356, + "step": 650 + }, + { + "epoch": 0.9421128798842258, + "grad_norm": 1.5294082164764404, + "learning_rate": 3.0128720836685445e-06, + "loss": 0.8672, + "step": 651 + }, + { + "epoch": 0.9435600578871202, + "grad_norm": 1.1220685243606567, + "learning_rate": 3.0088495575221242e-06, + "loss": 0.8557, + "step": 652 + }, + { + "epoch": 0.9450072358900145, + "grad_norm": 1.4353463649749756, + "learning_rate": 3.0048270313757043e-06, + "loss": 0.8753, + "step": 653 + }, + { + "epoch": 0.9464544138929089, + "grad_norm": 1.184134840965271, + "learning_rate": 3.0008045052292845e-06, + "loss": 0.8765, + "step": 654 + }, + { + "epoch": 0.9479015918958031, + "grad_norm": 1.5620888471603394, + "learning_rate": 2.9967819790828646e-06, + "loss": 0.9008, + "step": 655 + }, + { + "epoch": 0.9493487698986975, + "grad_norm": 1.5790199041366577, + "learning_rate": 2.9927594529364447e-06, + "loss": 0.9159, + "step": 656 + }, + { + "epoch": 0.9507959479015919, + "grad_norm": 1.4135411977767944, + "learning_rate": 2.9887369267900244e-06, + "loss": 0.8516, + "step": 657 + }, + { + "epoch": 0.9522431259044862, + "grad_norm": 1.8455939292907715, + "learning_rate": 2.9847144006436046e-06, + "loss": 0.8566, + "step": 658 + }, + { + "epoch": 0.9536903039073806, + "grad_norm": 1.258618950843811, + "learning_rate": 2.9806918744971847e-06, + "loss": 0.8689, + "step": 659 + }, + { + "epoch": 0.9551374819102749, + "grad_norm": 1.5818113088607788, + "learning_rate": 2.976669348350765e-06, + "loss": 0.8631, + "step": 660 + }, + { + "epoch": 0.9551374819102749, + "eval_loss": 0.9798663258552551, + "eval_runtime": 46.3795, + "eval_samples_per_second": 21.561, + "eval_steps_per_second": 1.358, + "step": 660 + }, + { + "epoch": 0.9565846599131693, + "grad_norm": 1.892830729484558, + "learning_rate": 2.972646822204345e-06, + "loss": 0.8978, + "step": 661 + }, + { + "epoch": 0.9580318379160637, + "grad_norm": 1.4997377395629883, + "learning_rate": 2.9686242960579246e-06, + "loss": 0.858, + "step": 662 + }, + { + "epoch": 0.959479015918958, + "grad_norm": 1.526278018951416, + "learning_rate": 2.9646017699115048e-06, + "loss": 0.8675, + "step": 663 + }, + { + "epoch": 0.9609261939218524, + "grad_norm": 1.4683629274368286, + "learning_rate": 2.960579243765085e-06, + "loss": 0.8915, + "step": 664 + }, + { + "epoch": 0.9623733719247467, + "grad_norm": 1.3979508876800537, + "learning_rate": 2.956556717618665e-06, + "loss": 0.815, + "step": 665 + }, + { + "epoch": 0.9638205499276411, + "grad_norm": 1.293160319328308, + "learning_rate": 2.952534191472245e-06, + "loss": 0.8908, + "step": 666 + }, + { + "epoch": 0.9652677279305355, + "grad_norm": 1.2902677059173584, + "learning_rate": 2.948511665325825e-06, + "loss": 0.8732, + "step": 667 + }, + { + "epoch": 0.9667149059334298, + "grad_norm": 1.4202075004577637, + "learning_rate": 2.944489139179405e-06, + "loss": 0.893, + "step": 668 + }, + { + "epoch": 0.9681620839363242, + "grad_norm": 1.3067445755004883, + "learning_rate": 2.940466613032985e-06, + "loss": 0.8803, + "step": 669 + }, + { + "epoch": 0.9696092619392185, + "grad_norm": 1.1895015239715576, + "learning_rate": 2.9364440868865652e-06, + "loss": 0.8268, + "step": 670 + }, + { + "epoch": 0.9696092619392185, + "eval_loss": 0.9786641597747803, + "eval_runtime": 46.4887, + "eval_samples_per_second": 21.511, + "eval_steps_per_second": 1.355, + "step": 670 + }, + { + "epoch": 0.9710564399421129, + "grad_norm": 1.555291771888733, + "learning_rate": 2.9324215607401454e-06, + "loss": 0.8767, + "step": 671 + }, + { + "epoch": 0.9725036179450073, + "grad_norm": 1.308671474456787, + "learning_rate": 2.928399034593725e-06, + "loss": 0.8552, + "step": 672 + }, + { + "epoch": 0.9739507959479016, + "grad_norm": 1.2239017486572266, + "learning_rate": 2.924376508447305e-06, + "loss": 0.8719, + "step": 673 + }, + { + "epoch": 0.975397973950796, + "grad_norm": 1.3224045038223267, + "learning_rate": 2.9203539823008853e-06, + "loss": 0.8389, + "step": 674 + }, + { + "epoch": 0.9768451519536903, + "grad_norm": 1.2512754201889038, + "learning_rate": 2.9163314561544655e-06, + "loss": 0.8555, + "step": 675 + }, + { + "epoch": 0.9782923299565847, + "grad_norm": 1.2753958702087402, + "learning_rate": 2.912308930008045e-06, + "loss": 0.8741, + "step": 676 + }, + { + "epoch": 0.9797395079594791, + "grad_norm": 1.3567620515823364, + "learning_rate": 2.9082864038616253e-06, + "loss": 0.8345, + "step": 677 + }, + { + "epoch": 0.9811866859623734, + "grad_norm": 1.2484797239303589, + "learning_rate": 2.9042638777152054e-06, + "loss": 0.8842, + "step": 678 + }, + { + "epoch": 0.9826338639652678, + "grad_norm": 1.3281735181808472, + "learning_rate": 2.9002413515687855e-06, + "loss": 0.8822, + "step": 679 + }, + { + "epoch": 0.984081041968162, + "grad_norm": 1.6872611045837402, + "learning_rate": 2.8962188254223657e-06, + "loss": 0.9065, + "step": 680 + }, + { + "epoch": 0.984081041968162, + "eval_loss": 0.974894642829895, + "eval_runtime": 46.5499, + "eval_samples_per_second": 21.482, + "eval_steps_per_second": 1.353, + "step": 680 + }, + { + "epoch": 0.9855282199710564, + "grad_norm": 1.2515989542007446, + "learning_rate": 2.8921962992759454e-06, + "loss": 0.8801, + "step": 681 + }, + { + "epoch": 0.9869753979739508, + "grad_norm": 1.4953069686889648, + "learning_rate": 2.8881737731295255e-06, + "loss": 0.835, + "step": 682 + }, + { + "epoch": 0.9884225759768451, + "grad_norm": 1.306508183479309, + "learning_rate": 2.8841512469831056e-06, + "loss": 0.8829, + "step": 683 + }, + { + "epoch": 0.9898697539797395, + "grad_norm": 1.1294792890548706, + "learning_rate": 2.8801287208366858e-06, + "loss": 0.8601, + "step": 684 + }, + { + "epoch": 0.9913169319826338, + "grad_norm": 1.4681851863861084, + "learning_rate": 2.876106194690266e-06, + "loss": 0.8655, + "step": 685 + }, + { + "epoch": 0.9927641099855282, + "grad_norm": 1.1896895170211792, + "learning_rate": 2.8720836685438456e-06, + "loss": 0.888, + "step": 686 + }, + { + "epoch": 0.9942112879884226, + "grad_norm": 1.2866474390029907, + "learning_rate": 2.8680611423974257e-06, + "loss": 0.8641, + "step": 687 + }, + { + "epoch": 0.9956584659913169, + "grad_norm": 1.401280403137207, + "learning_rate": 2.864038616251006e-06, + "loss": 0.8526, + "step": 688 + }, + { + "epoch": 0.9971056439942113, + "grad_norm": 1.2278425693511963, + "learning_rate": 2.860016090104586e-06, + "loss": 0.8694, + "step": 689 + }, + { + "epoch": 0.9985528219971056, + "grad_norm": 1.395071029663086, + "learning_rate": 2.855993563958166e-06, + "loss": 0.8899, + "step": 690 + }, + { + "epoch": 0.9985528219971056, + "eval_loss": 0.975599467754364, + "eval_runtime": 46.5485, + "eval_samples_per_second": 21.483, + "eval_steps_per_second": 1.353, + "step": 690 + }, + { + "epoch": 1.0, + "grad_norm": 1.1318234205245972, + "learning_rate": 2.851971037811746e-06, + "loss": 0.8448, + "step": 691 + }, + { + "epoch": 1.0014471780028944, + "grad_norm": 1.4713577032089233, + "learning_rate": 2.847948511665326e-06, + "loss": 0.8283, + "step": 692 + }, + { + "epoch": 1.0028943560057888, + "grad_norm": 1.3575835227966309, + "learning_rate": 2.843925985518906e-06, + "loss": 0.8235, + "step": 693 + }, + { + "epoch": 1.004341534008683, + "grad_norm": 1.2872260808944702, + "learning_rate": 2.839903459372486e-06, + "loss": 0.8664, + "step": 694 + }, + { + "epoch": 1.0057887120115774, + "grad_norm": 1.30717933177948, + "learning_rate": 2.8358809332260663e-06, + "loss": 0.8769, + "step": 695 + }, + { + "epoch": 1.0072358900144718, + "grad_norm": 1.5510741472244263, + "learning_rate": 2.831858407079646e-06, + "loss": 0.8594, + "step": 696 + }, + { + "epoch": 1.0086830680173662, + "grad_norm": 1.4470067024230957, + "learning_rate": 2.827835880933226e-06, + "loss": 0.8413, + "step": 697 + }, + { + "epoch": 1.0101302460202606, + "grad_norm": 1.3502882719039917, + "learning_rate": 2.8238133547868063e-06, + "loss": 0.816, + "step": 698 + }, + { + "epoch": 1.0115774240231548, + "grad_norm": 1.388826608657837, + "learning_rate": 2.8197908286403864e-06, + "loss": 0.832, + "step": 699 + }, + { + "epoch": 1.0130246020260492, + "grad_norm": 1.4521276950836182, + "learning_rate": 2.8157683024939665e-06, + "loss": 0.8991, + "step": 700 + }, + { + "epoch": 1.0130246020260492, + "eval_loss": 0.976570188999176, + "eval_runtime": 46.4378, + "eval_samples_per_second": 21.534, + "eval_steps_per_second": 1.357, + "step": 700 + }, + { + "epoch": 1.0144717800289436, + "grad_norm": 1.2110936641693115, + "learning_rate": 2.8117457763475462e-06, + "loss": 0.8805, + "step": 701 + }, + { + "epoch": 1.015918958031838, + "grad_norm": 1.3216700553894043, + "learning_rate": 2.8077232502011264e-06, + "loss": 0.8834, + "step": 702 + }, + { + "epoch": 1.0173661360347324, + "grad_norm": 1.5154712200164795, + "learning_rate": 2.8037007240547065e-06, + "loss": 0.8654, + "step": 703 + }, + { + "epoch": 1.0188133140376265, + "grad_norm": 1.4227033853530884, + "learning_rate": 2.7996781979082866e-06, + "loss": 0.8539, + "step": 704 + }, + { + "epoch": 1.020260492040521, + "grad_norm": 1.4331364631652832, + "learning_rate": 2.7956556717618667e-06, + "loss": 0.89, + "step": 705 + }, + { + "epoch": 1.0217076700434153, + "grad_norm": 1.3642888069152832, + "learning_rate": 2.7916331456154465e-06, + "loss": 0.814, + "step": 706 + }, + { + "epoch": 1.0231548480463097, + "grad_norm": 1.2116179466247559, + "learning_rate": 2.7876106194690266e-06, + "loss": 0.859, + "step": 707 + }, + { + "epoch": 1.0246020260492041, + "grad_norm": 1.4762978553771973, + "learning_rate": 2.7835880933226067e-06, + "loss": 0.8935, + "step": 708 + }, + { + "epoch": 1.0260492040520983, + "grad_norm": 1.4162617921829224, + "learning_rate": 2.779565567176187e-06, + "loss": 0.8128, + "step": 709 + }, + { + "epoch": 1.0274963820549927, + "grad_norm": 1.5273042917251587, + "learning_rate": 2.775543041029767e-06, + "loss": 0.8342, + "step": 710 + }, + { + "epoch": 1.0274963820549927, + "eval_loss": 0.9771099090576172, + "eval_runtime": 46.5601, + "eval_samples_per_second": 21.478, + "eval_steps_per_second": 1.353, + "step": 710 + }, + { + "epoch": 1.0289435600578871, + "grad_norm": 1.284264087677002, + "learning_rate": 2.7715205148833467e-06, + "loss": 0.8574, + "step": 711 + }, + { + "epoch": 1.0303907380607815, + "grad_norm": 1.3443101644515991, + "learning_rate": 2.767497988736927e-06, + "loss": 0.8495, + "step": 712 + }, + { + "epoch": 1.031837916063676, + "grad_norm": 1.3903629779815674, + "learning_rate": 2.763475462590507e-06, + "loss": 0.8685, + "step": 713 + }, + { + "epoch": 1.03328509406657, + "grad_norm": 1.723893404006958, + "learning_rate": 2.759452936444087e-06, + "loss": 0.8465, + "step": 714 + }, + { + "epoch": 1.0347322720694645, + "grad_norm": 1.487080454826355, + "learning_rate": 2.755430410297667e-06, + "loss": 0.8587, + "step": 715 + }, + { + "epoch": 1.036179450072359, + "grad_norm": 1.2809351682662964, + "learning_rate": 2.751407884151247e-06, + "loss": 0.854, + "step": 716 + }, + { + "epoch": 1.0376266280752533, + "grad_norm": 1.335334062576294, + "learning_rate": 2.747385358004827e-06, + "loss": 0.8412, + "step": 717 + }, + { + "epoch": 1.0390738060781477, + "grad_norm": 1.66199791431427, + "learning_rate": 2.743362831858407e-06, + "loss": 0.8249, + "step": 718 + }, + { + "epoch": 1.0405209840810419, + "grad_norm": 1.449202299118042, + "learning_rate": 2.7393403057119873e-06, + "loss": 0.8391, + "step": 719 + }, + { + "epoch": 1.0419681620839363, + "grad_norm": 1.3165470361709595, + "learning_rate": 2.7353177795655674e-06, + "loss": 0.8626, + "step": 720 + }, + { + "epoch": 1.0419681620839363, + "eval_loss": 0.9783347249031067, + "eval_runtime": 46.5241, + "eval_samples_per_second": 21.494, + "eval_steps_per_second": 1.354, + "step": 720 + }, + { + "epoch": 1.0434153400868307, + "grad_norm": 1.2465099096298218, + "learning_rate": 2.731295253419147e-06, + "loss": 0.8348, + "step": 721 + }, + { + "epoch": 1.044862518089725, + "grad_norm": 1.456369400024414, + "learning_rate": 2.7272727272727272e-06, + "loss": 0.8176, + "step": 722 + }, + { + "epoch": 1.0463096960926195, + "grad_norm": 1.386748194694519, + "learning_rate": 2.7232502011263074e-06, + "loss": 0.8523, + "step": 723 + }, + { + "epoch": 1.0477568740955137, + "grad_norm": 1.2743839025497437, + "learning_rate": 2.7192276749798875e-06, + "loss": 0.8466, + "step": 724 + }, + { + "epoch": 1.049204052098408, + "grad_norm": 1.3785805702209473, + "learning_rate": 2.7152051488334676e-06, + "loss": 0.8046, + "step": 725 + }, + { + "epoch": 1.0506512301013025, + "grad_norm": 1.4485764503479004, + "learning_rate": 2.7111826226870473e-06, + "loss": 0.7988, + "step": 726 + }, + { + "epoch": 1.0520984081041969, + "grad_norm": 1.2922154664993286, + "learning_rate": 2.7071600965406274e-06, + "loss": 0.8462, + "step": 727 + }, + { + "epoch": 1.0535455861070913, + "grad_norm": 1.5016216039657593, + "learning_rate": 2.7031375703942076e-06, + "loss": 0.8487, + "step": 728 + }, + { + "epoch": 1.0549927641099854, + "grad_norm": 1.5889242887496948, + "learning_rate": 2.6991150442477877e-06, + "loss": 0.9077, + "step": 729 + }, + { + "epoch": 1.0564399421128798, + "grad_norm": 1.568780541419983, + "learning_rate": 2.695092518101368e-06, + "loss": 0.8442, + "step": 730 + }, + { + "epoch": 1.0564399421128798, + "eval_loss": 0.9772480726242065, + "eval_runtime": 46.448, + "eval_samples_per_second": 21.529, + "eval_steps_per_second": 1.356, + "step": 730 + }, + { + "epoch": 1.0578871201157742, + "grad_norm": 1.3580796718597412, + "learning_rate": 2.6910699919549475e-06, + "loss": 0.8311, + "step": 731 + }, + { + "epoch": 1.0593342981186686, + "grad_norm": 1.3131071329116821, + "learning_rate": 2.6870474658085277e-06, + "loss": 0.8508, + "step": 732 + }, + { + "epoch": 1.060781476121563, + "grad_norm": 1.3442649841308594, + "learning_rate": 2.6830249396621078e-06, + "loss": 0.8578, + "step": 733 + }, + { + "epoch": 1.0622286541244572, + "grad_norm": 1.3269685506820679, + "learning_rate": 2.679002413515688e-06, + "loss": 0.8452, + "step": 734 + }, + { + "epoch": 1.0636758321273516, + "grad_norm": 1.315401554107666, + "learning_rate": 2.6749798873692685e-06, + "loss": 0.7997, + "step": 735 + }, + { + "epoch": 1.065123010130246, + "grad_norm": 1.261452317237854, + "learning_rate": 2.6709573612228486e-06, + "loss": 0.8417, + "step": 736 + }, + { + "epoch": 1.0665701881331404, + "grad_norm": 1.4281047582626343, + "learning_rate": 2.6669348350764283e-06, + "loss": 0.8391, + "step": 737 + }, + { + "epoch": 1.0680173661360348, + "grad_norm": 1.4431053400039673, + "learning_rate": 2.6629123089300084e-06, + "loss": 0.854, + "step": 738 + }, + { + "epoch": 1.069464544138929, + "grad_norm": 1.356166124343872, + "learning_rate": 2.6588897827835885e-06, + "loss": 0.8834, + "step": 739 + }, + { + "epoch": 1.0709117221418234, + "grad_norm": 1.294845700263977, + "learning_rate": 2.6548672566371687e-06, + "loss": 0.8569, + "step": 740 + }, + { + "epoch": 1.0709117221418234, + "eval_loss": 0.9747171401977539, + "eval_runtime": 46.6243, + "eval_samples_per_second": 21.448, + "eval_steps_per_second": 1.351, + "step": 740 + }, + { + "epoch": 1.0723589001447178, + "grad_norm": 1.2511882781982422, + "learning_rate": 2.650844730490749e-06, + "loss": 0.8102, + "step": 741 + }, + { + "epoch": 1.0738060781476122, + "grad_norm": 1.099092960357666, + "learning_rate": 2.6468222043443285e-06, + "loss": 0.8354, + "step": 742 + }, + { + "epoch": 1.0752532561505066, + "grad_norm": 1.4110586643218994, + "learning_rate": 2.6427996781979086e-06, + "loss": 0.8442, + "step": 743 + }, + { + "epoch": 1.0767004341534008, + "grad_norm": 1.3858177661895752, + "learning_rate": 2.6387771520514888e-06, + "loss": 0.8484, + "step": 744 + }, + { + "epoch": 1.0781476121562952, + "grad_norm": 1.2831997871398926, + "learning_rate": 2.634754625905069e-06, + "loss": 0.8323, + "step": 745 + }, + { + "epoch": 1.0795947901591896, + "grad_norm": 1.1923599243164062, + "learning_rate": 2.630732099758649e-06, + "loss": 0.8472, + "step": 746 + }, + { + "epoch": 1.081041968162084, + "grad_norm": 1.127179503440857, + "learning_rate": 2.6267095736122287e-06, + "loss": 0.8584, + "step": 747 + }, + { + "epoch": 1.0824891461649784, + "grad_norm": 1.1873705387115479, + "learning_rate": 2.622687047465809e-06, + "loss": 0.8868, + "step": 748 + }, + { + "epoch": 1.0839363241678726, + "grad_norm": 1.2759733200073242, + "learning_rate": 2.618664521319389e-06, + "loss": 0.8216, + "step": 749 + }, + { + "epoch": 1.085383502170767, + "grad_norm": 1.4137777090072632, + "learning_rate": 2.614641995172969e-06, + "loss": 0.8477, + "step": 750 + }, + { + "epoch": 1.085383502170767, + "eval_loss": 0.9748939275741577, + "eval_runtime": 46.5367, + "eval_samples_per_second": 21.488, + "eval_steps_per_second": 1.354, + "step": 750 + }, + { + "epoch": 1.0868306801736614, + "grad_norm": 1.246415376663208, + "learning_rate": 2.6106194690265492e-06, + "loss": 0.854, + "step": 751 + }, + { + "epoch": 1.0882778581765558, + "grad_norm": 1.3193167448043823, + "learning_rate": 2.606596942880129e-06, + "loss": 0.8436, + "step": 752 + }, + { + "epoch": 1.0897250361794502, + "grad_norm": 1.3617298603057861, + "learning_rate": 2.602574416733709e-06, + "loss": 0.8634, + "step": 753 + }, + { + "epoch": 1.0911722141823443, + "grad_norm": 1.4259262084960938, + "learning_rate": 2.598551890587289e-06, + "loss": 0.8557, + "step": 754 + }, + { + "epoch": 1.0926193921852387, + "grad_norm": 1.3111982345581055, + "learning_rate": 2.5945293644408693e-06, + "loss": 0.8717, + "step": 755 + }, + { + "epoch": 1.0940665701881331, + "grad_norm": 1.287125587463379, + "learning_rate": 2.5905068382944494e-06, + "loss": 0.8114, + "step": 756 + }, + { + "epoch": 1.0955137481910275, + "grad_norm": 1.201884150505066, + "learning_rate": 2.586484312148029e-06, + "loss": 0.8589, + "step": 757 + }, + { + "epoch": 1.096960926193922, + "grad_norm": 1.319590449333191, + "learning_rate": 2.5824617860016093e-06, + "loss": 0.8415, + "step": 758 + }, + { + "epoch": 1.0984081041968161, + "grad_norm": 1.473556399345398, + "learning_rate": 2.5784392598551894e-06, + "loss": 0.8245, + "step": 759 + }, + { + "epoch": 1.0998552821997105, + "grad_norm": 1.4037073850631714, + "learning_rate": 2.5744167337087695e-06, + "loss": 0.812, + "step": 760 + }, + { + "epoch": 1.0998552821997105, + "eval_loss": 0.9768498539924622, + "eval_runtime": 46.6407, + "eval_samples_per_second": 21.44, + "eval_steps_per_second": 1.351, + "step": 760 + }, + { + "epoch": 1.101302460202605, + "grad_norm": 1.3993818759918213, + "learning_rate": 2.5703942075623497e-06, + "loss": 0.8355, + "step": 761 + }, + { + "epoch": 1.1027496382054993, + "grad_norm": 1.4144906997680664, + "learning_rate": 2.5663716814159294e-06, + "loss": 0.8164, + "step": 762 + }, + { + "epoch": 1.1041968162083937, + "grad_norm": 1.3686332702636719, + "learning_rate": 2.5623491552695095e-06, + "loss": 0.8152, + "step": 763 + }, + { + "epoch": 1.105643994211288, + "grad_norm": 1.2269115447998047, + "learning_rate": 2.5583266291230896e-06, + "loss": 0.8612, + "step": 764 + }, + { + "epoch": 1.1070911722141823, + "grad_norm": 1.2964756488800049, + "learning_rate": 2.5543041029766698e-06, + "loss": 0.8383, + "step": 765 + }, + { + "epoch": 1.1085383502170767, + "grad_norm": 1.1819671392440796, + "learning_rate": 2.55028157683025e-06, + "loss": 0.8381, + "step": 766 + }, + { + "epoch": 1.109985528219971, + "grad_norm": 1.5450025796890259, + "learning_rate": 2.5462590506838296e-06, + "loss": 0.8357, + "step": 767 + }, + { + "epoch": 1.1114327062228655, + "grad_norm": 1.6499382257461548, + "learning_rate": 2.5422365245374097e-06, + "loss": 0.8519, + "step": 768 + }, + { + "epoch": 1.1128798842257597, + "grad_norm": 1.3141034841537476, + "learning_rate": 2.53821399839099e-06, + "loss": 0.8328, + "step": 769 + }, + { + "epoch": 1.114327062228654, + "grad_norm": 1.387905478477478, + "learning_rate": 2.53419147224457e-06, + "loss": 0.8925, + "step": 770 + }, + { + "epoch": 1.114327062228654, + "eval_loss": 0.9749211668968201, + "eval_runtime": 46.588, + "eval_samples_per_second": 21.465, + "eval_steps_per_second": 1.352, + "step": 770 + }, + { + "epoch": 1.1157742402315485, + "grad_norm": 1.2313570976257324, + "learning_rate": 2.53016894609815e-06, + "loss": 0.8468, + "step": 771 + }, + { + "epoch": 1.1172214182344429, + "grad_norm": 1.2164853811264038, + "learning_rate": 2.52614641995173e-06, + "loss": 0.8346, + "step": 772 + }, + { + "epoch": 1.1186685962373373, + "grad_norm": 1.3034911155700684, + "learning_rate": 2.52212389380531e-06, + "loss": 0.889, + "step": 773 + }, + { + "epoch": 1.1201157742402315, + "grad_norm": 1.516965627670288, + "learning_rate": 2.51810136765889e-06, + "loss": 0.8722, + "step": 774 + }, + { + "epoch": 1.1215629522431259, + "grad_norm": 1.3549655675888062, + "learning_rate": 2.51407884151247e-06, + "loss": 0.8353, + "step": 775 + }, + { + "epoch": 1.1230101302460203, + "grad_norm": 1.561373233795166, + "learning_rate": 2.5100563153660503e-06, + "loss": 0.8442, + "step": 776 + }, + { + "epoch": 1.1244573082489147, + "grad_norm": 1.3647520542144775, + "learning_rate": 2.50603378921963e-06, + "loss": 0.858, + "step": 777 + }, + { + "epoch": 1.125904486251809, + "grad_norm": 1.4156564474105835, + "learning_rate": 2.50201126307321e-06, + "loss": 0.8322, + "step": 778 + }, + { + "epoch": 1.1273516642547032, + "grad_norm": 1.6450928449630737, + "learning_rate": 2.4979887369267903e-06, + "loss": 0.8732, + "step": 779 + }, + { + "epoch": 1.1287988422575976, + "grad_norm": 1.2066187858581543, + "learning_rate": 2.4939662107803704e-06, + "loss": 0.8527, + "step": 780 + }, + { + "epoch": 1.1287988422575976, + "eval_loss": 0.9730677604675293, + "eval_runtime": 46.5733, + "eval_samples_per_second": 21.472, + "eval_steps_per_second": 1.353, + "step": 780 + }, + { + "epoch": 1.130246020260492, + "grad_norm": 1.4132040739059448, + "learning_rate": 2.48994368463395e-06, + "loss": 0.8155, + "step": 781 + }, + { + "epoch": 1.1316931982633864, + "grad_norm": 1.3123786449432373, + "learning_rate": 2.4859211584875302e-06, + "loss": 0.8611, + "step": 782 + }, + { + "epoch": 1.1331403762662808, + "grad_norm": 1.4299067258834839, + "learning_rate": 2.4818986323411104e-06, + "loss": 0.8589, + "step": 783 + }, + { + "epoch": 1.134587554269175, + "grad_norm": 1.534983515739441, + "learning_rate": 2.4778761061946905e-06, + "loss": 0.8286, + "step": 784 + }, + { + "epoch": 1.1360347322720694, + "grad_norm": 1.4856404066085815, + "learning_rate": 2.4738535800482706e-06, + "loss": 0.8692, + "step": 785 + }, + { + "epoch": 1.1374819102749638, + "grad_norm": 1.4667112827301025, + "learning_rate": 2.4698310539018503e-06, + "loss": 0.8339, + "step": 786 + }, + { + "epoch": 1.1389290882778582, + "grad_norm": 1.5479187965393066, + "learning_rate": 2.4658085277554304e-06, + "loss": 0.8565, + "step": 787 + }, + { + "epoch": 1.1403762662807526, + "grad_norm": 1.5787947177886963, + "learning_rate": 2.4617860016090106e-06, + "loss": 0.8365, + "step": 788 + }, + { + "epoch": 1.1418234442836468, + "grad_norm": 1.363026738166809, + "learning_rate": 2.4577634754625907e-06, + "loss": 0.83, + "step": 789 + }, + { + "epoch": 1.1432706222865412, + "grad_norm": 1.3956077098846436, + "learning_rate": 2.453740949316171e-06, + "loss": 0.8361, + "step": 790 + }, + { + "epoch": 1.1432706222865412, + "eval_loss": 0.9747700691223145, + "eval_runtime": 46.5301, + "eval_samples_per_second": 21.491, + "eval_steps_per_second": 1.354, + "step": 790 + }, + { + "epoch": 1.1447178002894356, + "grad_norm": 1.3515560626983643, + "learning_rate": 2.4497184231697505e-06, + "loss": 0.862, + "step": 791 + }, + { + "epoch": 1.14616497829233, + "grad_norm": 1.613192081451416, + "learning_rate": 2.4456958970233307e-06, + "loss": 0.8396, + "step": 792 + }, + { + "epoch": 1.1476121562952244, + "grad_norm": 1.552998661994934, + "learning_rate": 2.4416733708769108e-06, + "loss": 0.8412, + "step": 793 + }, + { + "epoch": 1.1490593342981186, + "grad_norm": 1.589155673980713, + "learning_rate": 2.437650844730491e-06, + "loss": 0.8875, + "step": 794 + }, + { + "epoch": 1.150506512301013, + "grad_norm": 1.458111047744751, + "learning_rate": 2.433628318584071e-06, + "loss": 0.821, + "step": 795 + }, + { + "epoch": 1.1519536903039074, + "grad_norm": 1.2502684593200684, + "learning_rate": 2.4296057924376507e-06, + "loss": 0.8402, + "step": 796 + }, + { + "epoch": 1.1534008683068018, + "grad_norm": 1.2446728944778442, + "learning_rate": 2.425583266291231e-06, + "loss": 0.8208, + "step": 797 + }, + { + "epoch": 1.1548480463096962, + "grad_norm": 1.6090346574783325, + "learning_rate": 2.421560740144811e-06, + "loss": 0.7941, + "step": 798 + }, + { + "epoch": 1.1562952243125904, + "grad_norm": 1.4145108461380005, + "learning_rate": 2.417538213998391e-06, + "loss": 0.8342, + "step": 799 + }, + { + "epoch": 1.1577424023154848, + "grad_norm": 1.4553568363189697, + "learning_rate": 2.4135156878519713e-06, + "loss": 0.8412, + "step": 800 + }, + { + "epoch": 1.1577424023154848, + "eval_loss": 0.9750587344169617, + "eval_runtime": 46.569, + "eval_samples_per_second": 21.474, + "eval_steps_per_second": 1.353, + "step": 800 + }, + { + "epoch": 1.1591895803183792, + "grad_norm": 1.4231923818588257, + "learning_rate": 2.4094931617055514e-06, + "loss": 0.8861, + "step": 801 + }, + { + "epoch": 1.1606367583212736, + "grad_norm": 1.4850999116897583, + "learning_rate": 2.4054706355591315e-06, + "loss": 0.8071, + "step": 802 + }, + { + "epoch": 1.162083936324168, + "grad_norm": 1.5963687896728516, + "learning_rate": 2.4014481094127116e-06, + "loss": 0.8722, + "step": 803 + }, + { + "epoch": 1.1635311143270621, + "grad_norm": 1.185127854347229, + "learning_rate": 2.3974255832662913e-06, + "loss": 0.8571, + "step": 804 + }, + { + "epoch": 1.1649782923299565, + "grad_norm": 1.2226200103759766, + "learning_rate": 2.3934030571198715e-06, + "loss": 0.8429, + "step": 805 + }, + { + "epoch": 1.166425470332851, + "grad_norm": 1.2654438018798828, + "learning_rate": 2.3893805309734516e-06, + "loss": 0.8387, + "step": 806 + }, + { + "epoch": 1.1678726483357453, + "grad_norm": 1.512937307357788, + "learning_rate": 2.3853580048270317e-06, + "loss": 0.8689, + "step": 807 + }, + { + "epoch": 1.1693198263386397, + "grad_norm": 1.445814847946167, + "learning_rate": 2.381335478680612e-06, + "loss": 0.8348, + "step": 808 + }, + { + "epoch": 1.170767004341534, + "grad_norm": 1.2088735103607178, + "learning_rate": 2.3773129525341916e-06, + "loss": 0.8265, + "step": 809 + }, + { + "epoch": 1.1722141823444283, + "grad_norm": 1.451804518699646, + "learning_rate": 2.3732904263877717e-06, + "loss": 0.863, + "step": 810 + }, + { + "epoch": 1.1722141823444283, + "eval_loss": 0.9725402593612671, + "eval_runtime": 46.6254, + "eval_samples_per_second": 21.448, + "eval_steps_per_second": 1.351, + "step": 810 + }, + { + "epoch": 1.1736613603473227, + "grad_norm": 1.3164602518081665, + "learning_rate": 2.369267900241352e-06, + "loss": 0.8364, + "step": 811 + }, + { + "epoch": 1.1751085383502171, + "grad_norm": 1.3769503831863403, + "learning_rate": 2.365245374094932e-06, + "loss": 0.8432, + "step": 812 + }, + { + "epoch": 1.1765557163531115, + "grad_norm": 1.397869348526001, + "learning_rate": 2.361222847948512e-06, + "loss": 0.8469, + "step": 813 + }, + { + "epoch": 1.1780028943560057, + "grad_norm": 1.4825493097305298, + "learning_rate": 2.3572003218020918e-06, + "loss": 0.865, + "step": 814 + }, + { + "epoch": 1.1794500723589, + "grad_norm": 1.4249802827835083, + "learning_rate": 2.353177795655672e-06, + "loss": 0.8579, + "step": 815 + }, + { + "epoch": 1.1808972503617945, + "grad_norm": 1.308760404586792, + "learning_rate": 2.349155269509252e-06, + "loss": 0.8628, + "step": 816 + }, + { + "epoch": 1.182344428364689, + "grad_norm": 1.3602763414382935, + "learning_rate": 2.345132743362832e-06, + "loss": 0.8568, + "step": 817 + }, + { + "epoch": 1.1837916063675833, + "grad_norm": 1.2844862937927246, + "learning_rate": 2.3411102172164123e-06, + "loss": 0.8363, + "step": 818 + }, + { + "epoch": 1.1852387843704775, + "grad_norm": 1.2435146570205688, + "learning_rate": 2.337087691069992e-06, + "loss": 0.851, + "step": 819 + }, + { + "epoch": 1.1866859623733719, + "grad_norm": 1.3820791244506836, + "learning_rate": 2.333065164923572e-06, + "loss": 0.8145, + "step": 820 + }, + { + "epoch": 1.1866859623733719, + "eval_loss": 0.9713840484619141, + "eval_runtime": 46.6395, + "eval_samples_per_second": 21.441, + "eval_steps_per_second": 1.351, + "step": 820 + }, + { + "epoch": 1.1881331403762663, + "grad_norm": 1.377058506011963, + "learning_rate": 2.3290426387771522e-06, + "loss": 0.8543, + "step": 821 + }, + { + "epoch": 1.1895803183791607, + "grad_norm": 1.3572325706481934, + "learning_rate": 2.3250201126307324e-06, + "loss": 0.8623, + "step": 822 + }, + { + "epoch": 1.191027496382055, + "grad_norm": 1.4975099563598633, + "learning_rate": 2.3209975864843125e-06, + "loss": 0.8393, + "step": 823 + }, + { + "epoch": 1.1924746743849495, + "grad_norm": 1.6872055530548096, + "learning_rate": 2.316975060337892e-06, + "loss": 0.8394, + "step": 824 + }, + { + "epoch": 1.1939218523878437, + "grad_norm": 1.5407776832580566, + "learning_rate": 2.3129525341914723e-06, + "loss": 0.8713, + "step": 825 + }, + { + "epoch": 1.195369030390738, + "grad_norm": 1.32473623752594, + "learning_rate": 2.3089300080450525e-06, + "loss": 0.8488, + "step": 826 + }, + { + "epoch": 1.1968162083936325, + "grad_norm": 1.4274264574050903, + "learning_rate": 2.3049074818986326e-06, + "loss": 0.8687, + "step": 827 + }, + { + "epoch": 1.1982633863965269, + "grad_norm": 1.1923948526382446, + "learning_rate": 2.3008849557522127e-06, + "loss": 0.8383, + "step": 828 + }, + { + "epoch": 1.199710564399421, + "grad_norm": 1.6807974576950073, + "learning_rate": 2.2968624296057924e-06, + "loss": 0.872, + "step": 829 + }, + { + "epoch": 1.2011577424023154, + "grad_norm": 1.5730746984481812, + "learning_rate": 2.2928399034593725e-06, + "loss": 0.8358, + "step": 830 + }, + { + "epoch": 1.2011577424023154, + "eval_loss": 0.9676039814949036, + "eval_runtime": 46.6143, + "eval_samples_per_second": 21.453, + "eval_steps_per_second": 1.352, + "step": 830 + }, + { + "epoch": 1.2026049204052098, + "grad_norm": 1.3702462911605835, + "learning_rate": 2.2888173773129527e-06, + "loss": 0.8294, + "step": 831 + }, + { + "epoch": 1.2040520984081042, + "grad_norm": 1.5075457096099854, + "learning_rate": 2.284794851166533e-06, + "loss": 0.846, + "step": 832 + }, + { + "epoch": 1.2054992764109986, + "grad_norm": 1.3182711601257324, + "learning_rate": 2.280772325020113e-06, + "loss": 0.8361, + "step": 833 + }, + { + "epoch": 1.206946454413893, + "grad_norm": 1.4154385328292847, + "learning_rate": 2.2767497988736926e-06, + "loss": 0.8275, + "step": 834 + }, + { + "epoch": 1.2083936324167872, + "grad_norm": 1.8060089349746704, + "learning_rate": 2.2727272727272728e-06, + "loss": 0.8144, + "step": 835 + }, + { + "epoch": 1.2098408104196816, + "grad_norm": 1.587913990020752, + "learning_rate": 2.268704746580853e-06, + "loss": 0.8326, + "step": 836 + }, + { + "epoch": 1.211287988422576, + "grad_norm": 1.2587682008743286, + "learning_rate": 2.264682220434433e-06, + "loss": 0.8363, + "step": 837 + }, + { + "epoch": 1.2127351664254704, + "grad_norm": 1.4036239385604858, + "learning_rate": 2.260659694288013e-06, + "loss": 0.8452, + "step": 838 + }, + { + "epoch": 1.2141823444283646, + "grad_norm": 1.2955446243286133, + "learning_rate": 2.256637168141593e-06, + "loss": 0.8131, + "step": 839 + }, + { + "epoch": 1.215629522431259, + "grad_norm": 1.6513280868530273, + "learning_rate": 2.252614641995173e-06, + "loss": 0.8561, + "step": 840 + }, + { + "epoch": 1.215629522431259, + "eval_loss": 0.9668388962745667, + "eval_runtime": 46.6529, + "eval_samples_per_second": 21.435, + "eval_steps_per_second": 1.35, + "step": 840 + }, + { + "epoch": 1.2170767004341534, + "grad_norm": 1.236353874206543, + "learning_rate": 2.248592115848753e-06, + "loss": 0.8112, + "step": 841 + }, + { + "epoch": 1.2185238784370478, + "grad_norm": 1.3581727743148804, + "learning_rate": 2.2445695897023332e-06, + "loss": 0.8398, + "step": 842 + }, + { + "epoch": 1.2199710564399422, + "grad_norm": 1.5839052200317383, + "learning_rate": 2.2405470635559134e-06, + "loss": 0.8132, + "step": 843 + }, + { + "epoch": 1.2214182344428366, + "grad_norm": 1.3417881727218628, + "learning_rate": 2.236524537409493e-06, + "loss": 0.8316, + "step": 844 + }, + { + "epoch": 1.2228654124457308, + "grad_norm": 1.3200281858444214, + "learning_rate": 2.232502011263073e-06, + "loss": 0.8438, + "step": 845 + }, + { + "epoch": 1.2243125904486252, + "grad_norm": 1.3113948106765747, + "learning_rate": 2.2284794851166537e-06, + "loss": 0.8538, + "step": 846 + }, + { + "epoch": 1.2257597684515196, + "grad_norm": 1.2885632514953613, + "learning_rate": 2.2244569589702334e-06, + "loss": 0.8521, + "step": 847 + }, + { + "epoch": 1.227206946454414, + "grad_norm": 1.3389625549316406, + "learning_rate": 2.2204344328238136e-06, + "loss": 0.8645, + "step": 848 + }, + { + "epoch": 1.2286541244573081, + "grad_norm": 1.2929267883300781, + "learning_rate": 2.2164119066773937e-06, + "loss": 0.8362, + "step": 849 + }, + { + "epoch": 1.2301013024602026, + "grad_norm": 1.3463146686553955, + "learning_rate": 2.212389380530974e-06, + "loss": 0.8376, + "step": 850 + }, + { + "epoch": 1.2301013024602026, + "eval_loss": 0.9657988548278809, + "eval_runtime": 46.6383, + "eval_samples_per_second": 21.442, + "eval_steps_per_second": 1.351, + "step": 850 + }, + { + "epoch": 1.231548480463097, + "grad_norm": 1.291881799697876, + "learning_rate": 2.208366854384554e-06, + "loss": 0.8496, + "step": 851 + }, + { + "epoch": 1.2329956584659914, + "grad_norm": 1.4908794164657593, + "learning_rate": 2.2043443282381337e-06, + "loss": 0.8533, + "step": 852 + }, + { + "epoch": 1.2344428364688858, + "grad_norm": 1.4043347835540771, + "learning_rate": 2.2003218020917138e-06, + "loss": 0.8174, + "step": 853 + }, + { + "epoch": 1.2358900144717802, + "grad_norm": 1.1834262609481812, + "learning_rate": 2.196299275945294e-06, + "loss": 0.8298, + "step": 854 + }, + { + "epoch": 1.2373371924746743, + "grad_norm": 1.2109973430633545, + "learning_rate": 2.192276749798874e-06, + "loss": 0.8484, + "step": 855 + }, + { + "epoch": 1.2387843704775687, + "grad_norm": 1.2916066646575928, + "learning_rate": 2.188254223652454e-06, + "loss": 0.8156, + "step": 856 + }, + { + "epoch": 1.2402315484804631, + "grad_norm": 1.3536677360534668, + "learning_rate": 2.184231697506034e-06, + "loss": 0.8305, + "step": 857 + }, + { + "epoch": 1.2416787264833575, + "grad_norm": 1.3534505367279053, + "learning_rate": 2.180209171359614e-06, + "loss": 0.8461, + "step": 858 + }, + { + "epoch": 1.2431259044862517, + "grad_norm": 1.4003204107284546, + "learning_rate": 2.176186645213194e-06, + "loss": 0.8271, + "step": 859 + }, + { + "epoch": 1.244573082489146, + "grad_norm": 1.3095545768737793, + "learning_rate": 2.1721641190667743e-06, + "loss": 0.8708, + "step": 860 + }, + { + "epoch": 1.244573082489146, + "eval_loss": 0.9674656391143799, + "eval_runtime": 46.6752, + "eval_samples_per_second": 21.425, + "eval_steps_per_second": 1.35, + "step": 860 + }, + { + "epoch": 1.2460202604920405, + "grad_norm": 1.2316744327545166, + "learning_rate": 2.1681415929203544e-06, + "loss": 0.8269, + "step": 861 + }, + { + "epoch": 1.247467438494935, + "grad_norm": 1.4316420555114746, + "learning_rate": 2.164119066773934e-06, + "loss": 0.846, + "step": 862 + }, + { + "epoch": 1.2489146164978293, + "grad_norm": 1.3739399909973145, + "learning_rate": 2.1600965406275142e-06, + "loss": 0.823, + "step": 863 + }, + { + "epoch": 1.2503617945007237, + "grad_norm": 1.1691535711288452, + "learning_rate": 2.1560740144810943e-06, + "loss": 0.8, + "step": 864 + }, + { + "epoch": 1.251808972503618, + "grad_norm": 1.1465041637420654, + "learning_rate": 2.1520514883346745e-06, + "loss": 0.8387, + "step": 865 + }, + { + "epoch": 1.2532561505065123, + "grad_norm": 1.459706425666809, + "learning_rate": 2.1480289621882546e-06, + "loss": 0.8472, + "step": 866 + }, + { + "epoch": 1.2547033285094067, + "grad_norm": 1.5499004125595093, + "learning_rate": 2.1440064360418343e-06, + "loss": 0.8539, + "step": 867 + }, + { + "epoch": 1.256150506512301, + "grad_norm": 1.2544631958007812, + "learning_rate": 2.1399839098954144e-06, + "loss": 0.8304, + "step": 868 + }, + { + "epoch": 1.2575976845151953, + "grad_norm": 1.2187174558639526, + "learning_rate": 2.1359613837489946e-06, + "loss": 0.8231, + "step": 869 + }, + { + "epoch": 1.2590448625180897, + "grad_norm": 1.4403445720672607, + "learning_rate": 2.1319388576025747e-06, + "loss": 0.8652, + "step": 870 + }, + { + "epoch": 1.2590448625180897, + "eval_loss": 0.9671112298965454, + "eval_runtime": 46.7368, + "eval_samples_per_second": 21.396, + "eval_steps_per_second": 1.348, + "step": 870 + }, + { + "epoch": 1.260492040520984, + "grad_norm": 1.4051568508148193, + "learning_rate": 2.127916331456155e-06, + "loss": 0.834, + "step": 871 + }, + { + "epoch": 1.2619392185238785, + "grad_norm": 1.430464267730713, + "learning_rate": 2.1238938053097345e-06, + "loss": 0.8395, + "step": 872 + }, + { + "epoch": 1.2633863965267729, + "grad_norm": 1.2451918125152588, + "learning_rate": 2.1198712791633146e-06, + "loss": 0.8011, + "step": 873 + }, + { + "epoch": 1.2648335745296673, + "grad_norm": 1.2514066696166992, + "learning_rate": 2.1158487530168948e-06, + "loss": 0.8256, + "step": 874 + }, + { + "epoch": 1.2662807525325614, + "grad_norm": 1.4800642728805542, + "learning_rate": 2.111826226870475e-06, + "loss": 0.8339, + "step": 875 + }, + { + "epoch": 1.2677279305354558, + "grad_norm": 1.3095451593399048, + "learning_rate": 2.107803700724055e-06, + "loss": 0.7934, + "step": 876 + }, + { + "epoch": 1.2691751085383502, + "grad_norm": 1.2315187454223633, + "learning_rate": 2.1037811745776347e-06, + "loss": 0.8285, + "step": 877 + }, + { + "epoch": 1.2706222865412446, + "grad_norm": 1.4502155780792236, + "learning_rate": 2.099758648431215e-06, + "loss": 0.8402, + "step": 878 + }, + { + "epoch": 1.2720694645441388, + "grad_norm": 1.5042566061019897, + "learning_rate": 2.095736122284795e-06, + "loss": 0.8448, + "step": 879 + }, + { + "epoch": 1.2735166425470332, + "grad_norm": 1.384658694267273, + "learning_rate": 2.091713596138375e-06, + "loss": 0.8432, + "step": 880 + }, + { + "epoch": 1.2735166425470332, + "eval_loss": 0.9652736783027649, + "eval_runtime": 46.7468, + "eval_samples_per_second": 21.392, + "eval_steps_per_second": 1.348, + "step": 880 + }, + { + "epoch": 1.2749638205499276, + "grad_norm": 1.82254159450531, + "learning_rate": 2.0876910699919552e-06, + "loss": 0.8448, + "step": 881 + }, + { + "epoch": 1.276410998552822, + "grad_norm": 1.4482412338256836, + "learning_rate": 2.083668543845535e-06, + "loss": 0.8581, + "step": 882 + }, + { + "epoch": 1.2778581765557164, + "grad_norm": 1.3943830728530884, + "learning_rate": 2.079646017699115e-06, + "loss": 0.8108, + "step": 883 + }, + { + "epoch": 1.2793053545586108, + "grad_norm": 1.3625071048736572, + "learning_rate": 2.075623491552695e-06, + "loss": 0.8333, + "step": 884 + }, + { + "epoch": 1.280752532561505, + "grad_norm": 1.3175891637802124, + "learning_rate": 2.0716009654062753e-06, + "loss": 0.841, + "step": 885 + }, + { + "epoch": 1.2821997105643994, + "grad_norm": 1.391445279121399, + "learning_rate": 2.067578439259855e-06, + "loss": 0.8328, + "step": 886 + }, + { + "epoch": 1.2836468885672938, + "grad_norm": 1.3408946990966797, + "learning_rate": 2.063555913113435e-06, + "loss": 0.8334, + "step": 887 + }, + { + "epoch": 1.2850940665701882, + "grad_norm": 1.5718055963516235, + "learning_rate": 2.0595333869670153e-06, + "loss": 0.8301, + "step": 888 + }, + { + "epoch": 1.2865412445730824, + "grad_norm": 1.506117343902588, + "learning_rate": 2.0555108608205954e-06, + "loss": 0.849, + "step": 889 + }, + { + "epoch": 1.2879884225759768, + "grad_norm": 1.287121057510376, + "learning_rate": 2.0514883346741755e-06, + "loss": 0.8355, + "step": 890 + }, + { + "epoch": 1.2879884225759768, + "eval_loss": 0.9627444744110107, + "eval_runtime": 46.7591, + "eval_samples_per_second": 21.386, + "eval_steps_per_second": 1.347, + "step": 890 + }, + { + "epoch": 1.2894356005788712, + "grad_norm": 1.1907145977020264, + "learning_rate": 2.0474658085277557e-06, + "loss": 0.8001, + "step": 891 + }, + { + "epoch": 1.2908827785817656, + "grad_norm": 1.4546834230422974, + "learning_rate": 2.043443282381336e-06, + "loss": 0.8456, + "step": 892 + }, + { + "epoch": 1.29232995658466, + "grad_norm": 1.6136828660964966, + "learning_rate": 2.039420756234916e-06, + "loss": 0.8443, + "step": 893 + }, + { + "epoch": 1.2937771345875544, + "grad_norm": 1.4644947052001953, + "learning_rate": 2.035398230088496e-06, + "loss": 0.8493, + "step": 894 + }, + { + "epoch": 1.2952243125904486, + "grad_norm": 1.3102108240127563, + "learning_rate": 2.0313757039420758e-06, + "loss": 0.8311, + "step": 895 + }, + { + "epoch": 1.296671490593343, + "grad_norm": 1.4119598865509033, + "learning_rate": 2.027353177795656e-06, + "loss": 0.8081, + "step": 896 + }, + { + "epoch": 1.2981186685962374, + "grad_norm": 1.2437739372253418, + "learning_rate": 2.023330651649236e-06, + "loss": 0.8403, + "step": 897 + }, + { + "epoch": 1.2995658465991318, + "grad_norm": 1.4662320613861084, + "learning_rate": 2.019308125502816e-06, + "loss": 0.8489, + "step": 898 + }, + { + "epoch": 1.301013024602026, + "grad_norm": 1.3092684745788574, + "learning_rate": 2.0152855993563963e-06, + "loss": 0.84, + "step": 899 + }, + { + "epoch": 1.3024602026049203, + "grad_norm": 1.2626603841781616, + "learning_rate": 2.011263073209976e-06, + "loss": 0.8317, + "step": 900 + }, + { + "epoch": 1.3024602026049203, + "eval_loss": 0.9606959223747253, + "eval_runtime": 46.7384, + "eval_samples_per_second": 21.396, + "eval_steps_per_second": 1.348, + "step": 900 + }, + { + "epoch": 1.3039073806078147, + "grad_norm": 1.457918405532837, + "learning_rate": 2.007240547063556e-06, + "loss": 0.8581, + "step": 901 + }, + { + "epoch": 1.3053545586107091, + "grad_norm": 1.3452205657958984, + "learning_rate": 2.0032180209171362e-06, + "loss": 0.8441, + "step": 902 + }, + { + "epoch": 1.3068017366136035, + "grad_norm": 1.3265753984451294, + "learning_rate": 1.9991954947707164e-06, + "loss": 0.8233, + "step": 903 + }, + { + "epoch": 1.308248914616498, + "grad_norm": 1.531461477279663, + "learning_rate": 1.995172968624296e-06, + "loss": 0.8266, + "step": 904 + }, + { + "epoch": 1.3096960926193921, + "grad_norm": 1.3354597091674805, + "learning_rate": 1.991150442477876e-06, + "loss": 0.8551, + "step": 905 + }, + { + "epoch": 1.3111432706222865, + "grad_norm": 1.472241997718811, + "learning_rate": 1.9871279163314563e-06, + "loss": 0.8508, + "step": 906 + }, + { + "epoch": 1.312590448625181, + "grad_norm": 1.1910359859466553, + "learning_rate": 1.9831053901850364e-06, + "loss": 0.8392, + "step": 907 + }, + { + "epoch": 1.3140376266280753, + "grad_norm": 1.4266308546066284, + "learning_rate": 1.9790828640386166e-06, + "loss": 0.8394, + "step": 908 + }, + { + "epoch": 1.3154848046309695, + "grad_norm": 1.241333246231079, + "learning_rate": 1.9750603378921963e-06, + "loss": 0.8318, + "step": 909 + }, + { + "epoch": 1.316931982633864, + "grad_norm": 1.2622087001800537, + "learning_rate": 1.9710378117457764e-06, + "loss": 0.8047, + "step": 910 + }, + { + "epoch": 1.316931982633864, + "eval_loss": 0.9627918601036072, + "eval_runtime": 46.7549, + "eval_samples_per_second": 21.388, + "eval_steps_per_second": 1.347, + "step": 910 + }, + { + "epoch": 1.3183791606367583, + "grad_norm": 1.3607842922210693, + "learning_rate": 1.9670152855993565e-06, + "loss": 0.8679, + "step": 911 + }, + { + "epoch": 1.3198263386396527, + "grad_norm": 1.2438714504241943, + "learning_rate": 1.9629927594529367e-06, + "loss": 0.8109, + "step": 912 + }, + { + "epoch": 1.321273516642547, + "grad_norm": 1.3873783349990845, + "learning_rate": 1.9589702333065168e-06, + "loss": 0.8448, + "step": 913 + }, + { + "epoch": 1.3227206946454415, + "grad_norm": 1.413836121559143, + "learning_rate": 1.9549477071600965e-06, + "loss": 0.8661, + "step": 914 + }, + { + "epoch": 1.3241678726483357, + "grad_norm": 1.2227388620376587, + "learning_rate": 1.9509251810136766e-06, + "loss": 0.8607, + "step": 915 + }, + { + "epoch": 1.32561505065123, + "grad_norm": 1.546280860900879, + "learning_rate": 1.9469026548672567e-06, + "loss": 0.8442, + "step": 916 + }, + { + "epoch": 1.3270622286541245, + "grad_norm": 1.2610927820205688, + "learning_rate": 1.942880128720837e-06, + "loss": 0.8216, + "step": 917 + }, + { + "epoch": 1.3285094066570189, + "grad_norm": 1.1880555152893066, + "learning_rate": 1.938857602574417e-06, + "loss": 0.8189, + "step": 918 + }, + { + "epoch": 1.329956584659913, + "grad_norm": 1.4394183158874512, + "learning_rate": 1.9348350764279967e-06, + "loss": 0.84, + "step": 919 + }, + { + "epoch": 1.3314037626628075, + "grad_norm": 1.2461533546447754, + "learning_rate": 1.930812550281577e-06, + "loss": 0.8568, + "step": 920 + }, + { + "epoch": 1.3314037626628075, + "eval_loss": 0.9645136594772339, + "eval_runtime": 46.6835, + "eval_samples_per_second": 21.421, + "eval_steps_per_second": 1.35, + "step": 920 + }, + { + "epoch": 1.3328509406657019, + "grad_norm": 1.293158769607544, + "learning_rate": 1.926790024135157e-06, + "loss": 0.8614, + "step": 921 + }, + { + "epoch": 1.3342981186685963, + "grad_norm": 1.4125447273254395, + "learning_rate": 1.922767497988737e-06, + "loss": 0.846, + "step": 922 + }, + { + "epoch": 1.3357452966714907, + "grad_norm": 1.456660509109497, + "learning_rate": 1.9187449718423172e-06, + "loss": 0.8363, + "step": 923 + }, + { + "epoch": 1.337192474674385, + "grad_norm": 1.342479944229126, + "learning_rate": 1.914722445695897e-06, + "loss": 0.8312, + "step": 924 + }, + { + "epoch": 1.3386396526772792, + "grad_norm": 1.305435061454773, + "learning_rate": 1.910699919549477e-06, + "loss": 0.8172, + "step": 925 + }, + { + "epoch": 1.3400868306801736, + "grad_norm": 1.3013609647750854, + "learning_rate": 1.9066773934030572e-06, + "loss": 0.83, + "step": 926 + }, + { + "epoch": 1.341534008683068, + "grad_norm": 1.3227487802505493, + "learning_rate": 1.9026548672566373e-06, + "loss": 0.8238, + "step": 927 + }, + { + "epoch": 1.3429811866859624, + "grad_norm": 1.4733039140701294, + "learning_rate": 1.8986323411102172e-06, + "loss": 0.8131, + "step": 928 + }, + { + "epoch": 1.3444283646888566, + "grad_norm": 1.3531107902526855, + "learning_rate": 1.8946098149637973e-06, + "loss": 0.8099, + "step": 929 + }, + { + "epoch": 1.345875542691751, + "grad_norm": 1.4294116497039795, + "learning_rate": 1.8905872888173773e-06, + "loss": 0.8367, + "step": 930 + }, + { + "epoch": 1.345875542691751, + "eval_loss": 0.9635813236236572, + "eval_runtime": 46.8133, + "eval_samples_per_second": 21.361, + "eval_steps_per_second": 1.346, + "step": 930 + }, + { + "epoch": 1.3473227206946454, + "grad_norm": 1.491234540939331, + "learning_rate": 1.8865647626709574e-06, + "loss": 0.8162, + "step": 931 + }, + { + "epoch": 1.3487698986975398, + "grad_norm": 1.279494047164917, + "learning_rate": 1.8825422365245375e-06, + "loss": 0.8429, + "step": 932 + }, + { + "epoch": 1.3502170767004342, + "grad_norm": 1.2704001665115356, + "learning_rate": 1.8785197103781174e-06, + "loss": 0.8192, + "step": 933 + }, + { + "epoch": 1.3516642547033286, + "grad_norm": 1.1714613437652588, + "learning_rate": 1.8744971842316978e-06, + "loss": 0.824, + "step": 934 + }, + { + "epoch": 1.3531114327062228, + "grad_norm": 1.4071584939956665, + "learning_rate": 1.870474658085278e-06, + "loss": 0.8601, + "step": 935 + }, + { + "epoch": 1.3545586107091172, + "grad_norm": 1.3441849946975708, + "learning_rate": 1.8664521319388578e-06, + "loss": 0.8439, + "step": 936 + }, + { + "epoch": 1.3560057887120116, + "grad_norm": 1.2592124938964844, + "learning_rate": 1.862429605792438e-06, + "loss": 0.8396, + "step": 937 + }, + { + "epoch": 1.357452966714906, + "grad_norm": 1.201258659362793, + "learning_rate": 1.8584070796460179e-06, + "loss": 0.8096, + "step": 938 + }, + { + "epoch": 1.3589001447178002, + "grad_norm": 1.3822835683822632, + "learning_rate": 1.854384553499598e-06, + "loss": 0.8245, + "step": 939 + }, + { + "epoch": 1.3603473227206946, + "grad_norm": 1.2722387313842773, + "learning_rate": 1.8503620273531781e-06, + "loss": 0.8193, + "step": 940 + }, + { + "epoch": 1.3603473227206946, + "eval_loss": 0.9659318923950195, + "eval_runtime": 46.8079, + "eval_samples_per_second": 21.364, + "eval_steps_per_second": 1.346, + "step": 940 + }, + { + "epoch": 1.361794500723589, + "grad_norm": 1.2419379949569702, + "learning_rate": 1.846339501206758e-06, + "loss": 0.8391, + "step": 941 + }, + { + "epoch": 1.3632416787264834, + "grad_norm": 1.4300531148910522, + "learning_rate": 1.8423169750603382e-06, + "loss": 0.8314, + "step": 942 + }, + { + "epoch": 1.3646888567293778, + "grad_norm": 1.363715648651123, + "learning_rate": 1.838294448913918e-06, + "loss": 0.8361, + "step": 943 + }, + { + "epoch": 1.3661360347322722, + "grad_norm": 1.2163121700286865, + "learning_rate": 1.8342719227674982e-06, + "loss": 0.8243, + "step": 944 + }, + { + "epoch": 1.3675832127351664, + "grad_norm": 1.5672646760940552, + "learning_rate": 1.8302493966210783e-06, + "loss": 0.8221, + "step": 945 + }, + { + "epoch": 1.3690303907380608, + "grad_norm": 1.145054578781128, + "learning_rate": 1.8262268704746582e-06, + "loss": 0.8355, + "step": 946 + }, + { + "epoch": 1.3704775687409552, + "grad_norm": 1.3501542806625366, + "learning_rate": 1.8222043443282384e-06, + "loss": 0.8449, + "step": 947 + }, + { + "epoch": 1.3719247467438496, + "grad_norm": 1.431017279624939, + "learning_rate": 1.8181818181818183e-06, + "loss": 0.866, + "step": 948 + }, + { + "epoch": 1.3733719247467437, + "grad_norm": 1.3053781986236572, + "learning_rate": 1.8141592920353984e-06, + "loss": 0.838, + "step": 949 + }, + { + "epoch": 1.3748191027496381, + "grad_norm": 1.1893788576126099, + "learning_rate": 1.8101367658889785e-06, + "loss": 0.7909, + "step": 950 + }, + { + "epoch": 1.3748191027496381, + "eval_loss": 0.9653781652450562, + "eval_runtime": 46.868, + "eval_samples_per_second": 21.336, + "eval_steps_per_second": 1.344, + "step": 950 + }, + { + "epoch": 1.3762662807525325, + "grad_norm": 1.6269181966781616, + "learning_rate": 1.8061142397425585e-06, + "loss": 0.7886, + "step": 951 + }, + { + "epoch": 1.377713458755427, + "grad_norm": 1.1881003379821777, + "learning_rate": 1.8020917135961386e-06, + "loss": 0.8137, + "step": 952 + }, + { + "epoch": 1.3791606367583213, + "grad_norm": 1.270447015762329, + "learning_rate": 1.7980691874497185e-06, + "loss": 0.845, + "step": 953 + }, + { + "epoch": 1.3806078147612157, + "grad_norm": 1.209436058998108, + "learning_rate": 1.7940466613032986e-06, + "loss": 0.7992, + "step": 954 + }, + { + "epoch": 1.38205499276411, + "grad_norm": 1.3471684455871582, + "learning_rate": 1.7900241351568788e-06, + "loss": 0.8067, + "step": 955 + }, + { + "epoch": 1.3835021707670043, + "grad_norm": 1.3163692951202393, + "learning_rate": 1.7860016090104587e-06, + "loss": 0.8307, + "step": 956 + }, + { + "epoch": 1.3849493487698987, + "grad_norm": 1.2121058702468872, + "learning_rate": 1.7819790828640388e-06, + "loss": 0.8076, + "step": 957 + }, + { + "epoch": 1.3863965267727931, + "grad_norm": 1.4720529317855835, + "learning_rate": 1.7779565567176187e-06, + "loss": 0.8409, + "step": 958 + }, + { + "epoch": 1.3878437047756873, + "grad_norm": 1.3038276433944702, + "learning_rate": 1.7739340305711988e-06, + "loss": 0.8273, + "step": 959 + }, + { + "epoch": 1.3892908827785817, + "grad_norm": 1.3459664583206177, + "learning_rate": 1.769911504424779e-06, + "loss": 0.8478, + "step": 960 + }, + { + "epoch": 1.3892908827785817, + "eval_loss": 0.9639009237289429, + "eval_runtime": 46.8386, + "eval_samples_per_second": 21.35, + "eval_steps_per_second": 1.345, + "step": 960 + }, + { + "epoch": 1.390738060781476, + "grad_norm": 1.3280129432678223, + "learning_rate": 1.7658889782783589e-06, + "loss": 0.8466, + "step": 961 + }, + { + "epoch": 1.3921852387843705, + "grad_norm": 1.230965256690979, + "learning_rate": 1.761866452131939e-06, + "loss": 0.8204, + "step": 962 + }, + { + "epoch": 1.393632416787265, + "grad_norm": 1.1879146099090576, + "learning_rate": 1.757843925985519e-06, + "loss": 0.848, + "step": 963 + }, + { + "epoch": 1.3950795947901593, + "grad_norm": 1.3148863315582275, + "learning_rate": 1.753821399839099e-06, + "loss": 0.8688, + "step": 964 + }, + { + "epoch": 1.3965267727930535, + "grad_norm": 1.4079407453536987, + "learning_rate": 1.7497988736926792e-06, + "loss": 0.8385, + "step": 965 + }, + { + "epoch": 1.3979739507959479, + "grad_norm": 1.4647493362426758, + "learning_rate": 1.7457763475462591e-06, + "loss": 0.8372, + "step": 966 + }, + { + "epoch": 1.3994211287988423, + "grad_norm": 1.4081465005874634, + "learning_rate": 1.7417538213998392e-06, + "loss": 0.8263, + "step": 967 + }, + { + "epoch": 1.4008683068017367, + "grad_norm": 1.2983832359313965, + "learning_rate": 1.7377312952534192e-06, + "loss": 0.7842, + "step": 968 + }, + { + "epoch": 1.4023154848046309, + "grad_norm": 1.1634629964828491, + "learning_rate": 1.7337087691069993e-06, + "loss": 0.8195, + "step": 969 + }, + { + "epoch": 1.4037626628075253, + "grad_norm": 1.3308466672897339, + "learning_rate": 1.7296862429605792e-06, + "loss": 0.8617, + "step": 970 + }, + { + "epoch": 1.4037626628075253, + "eval_loss": 0.9622061848640442, + "eval_runtime": 46.7347, + "eval_samples_per_second": 21.397, + "eval_steps_per_second": 1.348, + "step": 970 + }, + { + "epoch": 1.4052098408104197, + "grad_norm": 1.1847211122512817, + "learning_rate": 1.7256637168141593e-06, + "loss": 0.8233, + "step": 971 + }, + { + "epoch": 1.406657018813314, + "grad_norm": 1.2786881923675537, + "learning_rate": 1.7216411906677395e-06, + "loss": 0.7962, + "step": 972 + }, + { + "epoch": 1.4081041968162085, + "grad_norm": 1.4232426881790161, + "learning_rate": 1.7176186645213194e-06, + "loss": 0.8394, + "step": 973 + }, + { + "epoch": 1.4095513748191029, + "grad_norm": 1.2933361530303955, + "learning_rate": 1.7135961383748995e-06, + "loss": 0.8678, + "step": 974 + }, + { + "epoch": 1.410998552821997, + "grad_norm": 1.3775824308395386, + "learning_rate": 1.7095736122284794e-06, + "loss": 0.8173, + "step": 975 + }, + { + "epoch": 1.4124457308248914, + "grad_norm": 1.4065430164337158, + "learning_rate": 1.7055510860820595e-06, + "loss": 0.8549, + "step": 976 + }, + { + "epoch": 1.4138929088277858, + "grad_norm": 1.2125884294509888, + "learning_rate": 1.7015285599356397e-06, + "loss": 0.8404, + "step": 977 + }, + { + "epoch": 1.4153400868306802, + "grad_norm": 1.291913628578186, + "learning_rate": 1.6975060337892196e-06, + "loss": 0.8422, + "step": 978 + }, + { + "epoch": 1.4167872648335744, + "grad_norm": 1.2509552240371704, + "learning_rate": 1.6934835076428e-06, + "loss": 0.8288, + "step": 979 + }, + { + "epoch": 1.4182344428364688, + "grad_norm": 1.2863197326660156, + "learning_rate": 1.68946098149638e-06, + "loss": 0.8643, + "step": 980 + }, + { + "epoch": 1.4182344428364688, + "eval_loss": 0.957442045211792, + "eval_runtime": 46.8775, + "eval_samples_per_second": 21.332, + "eval_steps_per_second": 1.344, + "step": 980 + }, + { + "epoch": 1.4196816208393632, + "grad_norm": 1.3801816701889038, + "learning_rate": 1.68543845534996e-06, + "loss": 0.7916, + "step": 981 + }, + { + "epoch": 1.4211287988422576, + "grad_norm": 1.1788544654846191, + "learning_rate": 1.68141592920354e-06, + "loss": 0.8256, + "step": 982 + }, + { + "epoch": 1.422575976845152, + "grad_norm": 1.309241771697998, + "learning_rate": 1.6773934030571202e-06, + "loss": 0.8141, + "step": 983 + }, + { + "epoch": 1.4240231548480464, + "grad_norm": 1.2282994985580444, + "learning_rate": 1.6733708769107001e-06, + "loss": 0.851, + "step": 984 + }, + { + "epoch": 1.4254703328509406, + "grad_norm": 1.3392151594161987, + "learning_rate": 1.6693483507642803e-06, + "loss": 0.801, + "step": 985 + }, + { + "epoch": 1.426917510853835, + "grad_norm": 1.513327717781067, + "learning_rate": 1.6653258246178602e-06, + "loss": 0.8377, + "step": 986 + }, + { + "epoch": 1.4283646888567294, + "grad_norm": 1.6124207973480225, + "learning_rate": 1.6613032984714403e-06, + "loss": 0.8421, + "step": 987 + }, + { + "epoch": 1.4298118668596238, + "grad_norm": 1.4307218790054321, + "learning_rate": 1.6572807723250202e-06, + "loss": 0.8408, + "step": 988 + }, + { + "epoch": 1.431259044862518, + "grad_norm": 1.3090101480484009, + "learning_rate": 1.6532582461786004e-06, + "loss": 0.8232, + "step": 989 + }, + { + "epoch": 1.4327062228654124, + "grad_norm": 1.5111008882522583, + "learning_rate": 1.6492357200321805e-06, + "loss": 0.869, + "step": 990 + }, + { + "epoch": 1.4327062228654124, + "eval_loss": 0.9584060907363892, + "eval_runtime": 46.8679, + "eval_samples_per_second": 21.337, + "eval_steps_per_second": 1.344, + "step": 990 + }, + { + "epoch": 1.4341534008683068, + "grad_norm": 1.4899122714996338, + "learning_rate": 1.6452131938857604e-06, + "loss": 0.814, + "step": 991 + }, + { + "epoch": 1.4356005788712012, + "grad_norm": 1.3443936109542847, + "learning_rate": 1.6411906677393405e-06, + "loss": 0.8229, + "step": 992 + }, + { + "epoch": 1.4370477568740956, + "grad_norm": 1.3372870683670044, + "learning_rate": 1.6371681415929204e-06, + "loss": 0.8336, + "step": 993 + }, + { + "epoch": 1.43849493487699, + "grad_norm": 1.423226237297058, + "learning_rate": 1.6331456154465006e-06, + "loss": 0.8408, + "step": 994 + }, + { + "epoch": 1.4399421128798842, + "grad_norm": 1.2469552755355835, + "learning_rate": 1.6291230893000807e-06, + "loss": 0.8537, + "step": 995 + }, + { + "epoch": 1.4413892908827786, + "grad_norm": 1.4579977989196777, + "learning_rate": 1.6251005631536606e-06, + "loss": 0.8633, + "step": 996 + }, + { + "epoch": 1.442836468885673, + "grad_norm": 1.3303523063659668, + "learning_rate": 1.6210780370072407e-06, + "loss": 0.816, + "step": 997 + }, + { + "epoch": 1.4442836468885674, + "grad_norm": 1.3875349760055542, + "learning_rate": 1.6170555108608207e-06, + "loss": 0.7846, + "step": 998 + }, + { + "epoch": 1.4457308248914615, + "grad_norm": 1.2573555707931519, + "learning_rate": 1.6130329847144008e-06, + "loss": 0.8322, + "step": 999 + }, + { + "epoch": 1.447178002894356, + "grad_norm": 1.3047151565551758, + "learning_rate": 1.609010458567981e-06, + "loss": 0.8285, + "step": 1000 + }, + { + "epoch": 1.447178002894356, + "eval_loss": 0.9623283743858337, + "eval_runtime": 46.9131, + "eval_samples_per_second": 21.316, + "eval_steps_per_second": 1.343, + "step": 1000 + }, + { + "epoch": 1.4486251808972503, + "grad_norm": 1.1986051797866821, + "learning_rate": 1.6049879324215608e-06, + "loss": 0.7907, + "step": 1001 + }, + { + "epoch": 1.4500723589001447, + "grad_norm": 1.3615407943725586, + "learning_rate": 1.600965406275141e-06, + "loss": 0.8309, + "step": 1002 + }, + { + "epoch": 1.4515195369030391, + "grad_norm": 1.2729979753494263, + "learning_rate": 1.5969428801287209e-06, + "loss": 0.808, + "step": 1003 + }, + { + "epoch": 1.4529667149059335, + "grad_norm": 1.4709190130233765, + "learning_rate": 1.592920353982301e-06, + "loss": 0.8625, + "step": 1004 + }, + { + "epoch": 1.4544138929088277, + "grad_norm": 1.479375958442688, + "learning_rate": 1.5888978278358811e-06, + "loss": 0.8198, + "step": 1005 + }, + { + "epoch": 1.4558610709117221, + "grad_norm": 1.2876075506210327, + "learning_rate": 1.584875301689461e-06, + "loss": 0.8429, + "step": 1006 + }, + { + "epoch": 1.4573082489146165, + "grad_norm": 1.3595634698867798, + "learning_rate": 1.5808527755430412e-06, + "loss": 0.7825, + "step": 1007 + }, + { + "epoch": 1.458755426917511, + "grad_norm": 1.3171416521072388, + "learning_rate": 1.576830249396621e-06, + "loss": 0.846, + "step": 1008 + }, + { + "epoch": 1.460202604920405, + "grad_norm": 1.4200496673583984, + "learning_rate": 1.5728077232502012e-06, + "loss": 0.7955, + "step": 1009 + }, + { + "epoch": 1.4616497829232995, + "grad_norm": 1.2488915920257568, + "learning_rate": 1.5687851971037813e-06, + "loss": 0.827, + "step": 1010 + }, + { + "epoch": 1.4616497829232995, + "eval_loss": 0.9625693559646606, + "eval_runtime": 46.741, + "eval_samples_per_second": 21.394, + "eval_steps_per_second": 1.348, + "step": 1010 + }, + { + "epoch": 1.463096960926194, + "grad_norm": 1.3633371591567993, + "learning_rate": 1.5647626709573613e-06, + "loss": 0.8445, + "step": 1011 + }, + { + "epoch": 1.4645441389290883, + "grad_norm": 1.24705171585083, + "learning_rate": 1.5607401448109414e-06, + "loss": 0.8095, + "step": 1012 + }, + { + "epoch": 1.4659913169319827, + "grad_norm": 1.3340017795562744, + "learning_rate": 1.5567176186645213e-06, + "loss": 0.8419, + "step": 1013 + }, + { + "epoch": 1.467438494934877, + "grad_norm": 1.1927664279937744, + "learning_rate": 1.5526950925181014e-06, + "loss": 0.8123, + "step": 1014 + }, + { + "epoch": 1.4688856729377713, + "grad_norm": 1.2698441743850708, + "learning_rate": 1.5486725663716816e-06, + "loss": 0.8214, + "step": 1015 + }, + { + "epoch": 1.4703328509406657, + "grad_norm": 1.3115718364715576, + "learning_rate": 1.5446500402252615e-06, + "loss": 0.8562, + "step": 1016 + }, + { + "epoch": 1.47178002894356, + "grad_norm": 1.4729806184768677, + "learning_rate": 1.5406275140788416e-06, + "loss": 0.8591, + "step": 1017 + }, + { + "epoch": 1.4732272069464545, + "grad_norm": 1.2308181524276733, + "learning_rate": 1.5366049879324215e-06, + "loss": 0.8004, + "step": 1018 + }, + { + "epoch": 1.4746743849493487, + "grad_norm": 1.2574098110198975, + "learning_rate": 1.5325824617860016e-06, + "loss": 0.7936, + "step": 1019 + }, + { + "epoch": 1.476121562952243, + "grad_norm": 1.355643630027771, + "learning_rate": 1.5285599356395816e-06, + "loss": 0.8736, + "step": 1020 + }, + { + "epoch": 1.476121562952243, + "eval_loss": 0.9595701694488525, + "eval_runtime": 46.8363, + "eval_samples_per_second": 21.351, + "eval_steps_per_second": 1.345, + "step": 1020 + }, + { + "epoch": 1.4775687409551375, + "grad_norm": 1.4033701419830322, + "learning_rate": 1.5245374094931617e-06, + "loss": 0.825, + "step": 1021 + }, + { + "epoch": 1.4790159189580319, + "grad_norm": 1.4104188680648804, + "learning_rate": 1.5205148833467418e-06, + "loss": 0.8346, + "step": 1022 + }, + { + "epoch": 1.4804630969609263, + "grad_norm": 1.3565980195999146, + "learning_rate": 1.5164923572003221e-06, + "loss": 0.824, + "step": 1023 + }, + { + "epoch": 1.4819102749638207, + "grad_norm": 1.369945764541626, + "learning_rate": 1.512469831053902e-06, + "loss": 0.8679, + "step": 1024 + }, + { + "epoch": 1.4833574529667148, + "grad_norm": 1.2620166540145874, + "learning_rate": 1.5084473049074822e-06, + "loss": 0.8212, + "step": 1025 + }, + { + "epoch": 1.4848046309696092, + "grad_norm": 1.2675236463546753, + "learning_rate": 1.5044247787610621e-06, + "loss": 0.8234, + "step": 1026 + }, + { + "epoch": 1.4862518089725036, + "grad_norm": 1.2762507200241089, + "learning_rate": 1.5004022526146422e-06, + "loss": 0.8323, + "step": 1027 + }, + { + "epoch": 1.487698986975398, + "grad_norm": 1.3624025583267212, + "learning_rate": 1.4963797264682224e-06, + "loss": 0.8327, + "step": 1028 + }, + { + "epoch": 1.4891461649782922, + "grad_norm": 1.2440052032470703, + "learning_rate": 1.4923572003218023e-06, + "loss": 0.8095, + "step": 1029 + }, + { + "epoch": 1.4905933429811866, + "grad_norm": 1.4403339624404907, + "learning_rate": 1.4883346741753824e-06, + "loss": 0.8157, + "step": 1030 + }, + { + "epoch": 1.4905933429811866, + "eval_loss": 0.9604761004447937, + "eval_runtime": 46.7951, + "eval_samples_per_second": 21.37, + "eval_steps_per_second": 1.346, + "step": 1030 + }, + { + "epoch": 1.492040520984081, + "grad_norm": 1.425316333770752, + "learning_rate": 1.4843121480289623e-06, + "loss": 0.8635, + "step": 1031 + }, + { + "epoch": 1.4934876989869754, + "grad_norm": 1.2514164447784424, + "learning_rate": 1.4802896218825425e-06, + "loss": 0.8226, + "step": 1032 + }, + { + "epoch": 1.4949348769898698, + "grad_norm": 1.3030881881713867, + "learning_rate": 1.4762670957361226e-06, + "loss": 0.8359, + "step": 1033 + }, + { + "epoch": 1.4963820549927642, + "grad_norm": 1.3198504447937012, + "learning_rate": 1.4722445695897025e-06, + "loss": 0.8012, + "step": 1034 + }, + { + "epoch": 1.4978292329956584, + "grad_norm": 1.4206291437149048, + "learning_rate": 1.4682220434432826e-06, + "loss": 0.8375, + "step": 1035 + }, + { + "epoch": 1.4992764109985528, + "grad_norm": 1.3416086435317993, + "learning_rate": 1.4641995172968625e-06, + "loss": 0.8125, + "step": 1036 + }, + { + "epoch": 1.5007235890014472, + "grad_norm": 1.3290832042694092, + "learning_rate": 1.4601769911504427e-06, + "loss": 0.8376, + "step": 1037 + }, + { + "epoch": 1.5021707670043414, + "grad_norm": 1.4313510656356812, + "learning_rate": 1.4561544650040226e-06, + "loss": 0.7909, + "step": 1038 + }, + { + "epoch": 1.5036179450072358, + "grad_norm": 1.2440143823623657, + "learning_rate": 1.4521319388576027e-06, + "loss": 0.8142, + "step": 1039 + }, + { + "epoch": 1.5050651230101302, + "grad_norm": 1.278433084487915, + "learning_rate": 1.4481094127111828e-06, + "loss": 0.8203, + "step": 1040 + }, + { + "epoch": 1.5050651230101302, + "eval_loss": 0.9619273543357849, + "eval_runtime": 46.7733, + "eval_samples_per_second": 21.38, + "eval_steps_per_second": 1.347, + "step": 1040 + }, + { + "epoch": 1.5065123010130246, + "grad_norm": 1.566141963005066, + "learning_rate": 1.4440868865647628e-06, + "loss": 0.8472, + "step": 1041 + }, + { + "epoch": 1.507959479015919, + "grad_norm": 1.4744622707366943, + "learning_rate": 1.4400643604183429e-06, + "loss": 0.8334, + "step": 1042 + }, + { + "epoch": 1.5094066570188134, + "grad_norm": 1.2336266040802002, + "learning_rate": 1.4360418342719228e-06, + "loss": 0.835, + "step": 1043 + }, + { + "epoch": 1.5108538350217078, + "grad_norm": 1.499137043952942, + "learning_rate": 1.432019308125503e-06, + "loss": 0.8394, + "step": 1044 + }, + { + "epoch": 1.5123010130246022, + "grad_norm": 1.3021162748336792, + "learning_rate": 1.427996781979083e-06, + "loss": 0.8428, + "step": 1045 + }, + { + "epoch": 1.5137481910274964, + "grad_norm": 1.1404552459716797, + "learning_rate": 1.423974255832663e-06, + "loss": 0.8222, + "step": 1046 + }, + { + "epoch": 1.5151953690303908, + "grad_norm": 1.488775610923767, + "learning_rate": 1.419951729686243e-06, + "loss": 0.7986, + "step": 1047 + }, + { + "epoch": 1.516642547033285, + "grad_norm": 1.5495660305023193, + "learning_rate": 1.415929203539823e-06, + "loss": 0.8408, + "step": 1048 + }, + { + "epoch": 1.5180897250361793, + "grad_norm": 1.3491184711456299, + "learning_rate": 1.4119066773934031e-06, + "loss": 0.8572, + "step": 1049 + }, + { + "epoch": 1.5195369030390737, + "grad_norm": 1.2956827878952026, + "learning_rate": 1.4078841512469833e-06, + "loss": 0.8254, + "step": 1050 + }, + { + "epoch": 1.5195369030390737, + "eval_loss": 0.9604234099388123, + "eval_runtime": 46.7565, + "eval_samples_per_second": 21.387, + "eval_steps_per_second": 1.347, + "step": 1050 + }, + { + "epoch": 1.5209840810419681, + "grad_norm": 1.1695504188537598, + "learning_rate": 1.4038616251005632e-06, + "loss": 0.8062, + "step": 1051 + }, + { + "epoch": 1.5224312590448625, + "grad_norm": 1.3935517072677612, + "learning_rate": 1.3998390989541433e-06, + "loss": 0.8388, + "step": 1052 + }, + { + "epoch": 1.523878437047757, + "grad_norm": 1.2945525646209717, + "learning_rate": 1.3958165728077232e-06, + "loss": 0.8401, + "step": 1053 + }, + { + "epoch": 1.5253256150506513, + "grad_norm": 1.3963762521743774, + "learning_rate": 1.3917940466613034e-06, + "loss": 0.8526, + "step": 1054 + }, + { + "epoch": 1.5267727930535457, + "grad_norm": 1.416558861732483, + "learning_rate": 1.3877715205148835e-06, + "loss": 0.8258, + "step": 1055 + }, + { + "epoch": 1.52821997105644, + "grad_norm": 1.3979424238204956, + "learning_rate": 1.3837489943684634e-06, + "loss": 0.8246, + "step": 1056 + }, + { + "epoch": 1.5296671490593343, + "grad_norm": 1.2990747690200806, + "learning_rate": 1.3797264682220435e-06, + "loss": 0.8097, + "step": 1057 + }, + { + "epoch": 1.5311143270622285, + "grad_norm": 1.2977523803710938, + "learning_rate": 1.3757039420756234e-06, + "loss": 0.8106, + "step": 1058 + }, + { + "epoch": 1.532561505065123, + "grad_norm": 1.384505271911621, + "learning_rate": 1.3716814159292036e-06, + "loss": 0.8211, + "step": 1059 + }, + { + "epoch": 1.5340086830680173, + "grad_norm": 1.341659665107727, + "learning_rate": 1.3676588897827837e-06, + "loss": 0.8157, + "step": 1060 + }, + { + "epoch": 1.5340086830680173, + "eval_loss": 0.9590486884117126, + "eval_runtime": 46.7344, + "eval_samples_per_second": 21.398, + "eval_steps_per_second": 1.348, + "step": 1060 + }, + { + "epoch": 1.5354558610709117, + "grad_norm": 1.348020076751709, + "learning_rate": 1.3636363636363636e-06, + "loss": 0.82, + "step": 1061 + }, + { + "epoch": 1.536903039073806, + "grad_norm": 1.2610324621200562, + "learning_rate": 1.3596138374899437e-06, + "loss": 0.8462, + "step": 1062 + }, + { + "epoch": 1.5383502170767005, + "grad_norm": 1.4453649520874023, + "learning_rate": 1.3555913113435237e-06, + "loss": 0.8116, + "step": 1063 + }, + { + "epoch": 1.539797395079595, + "grad_norm": 1.236574649810791, + "learning_rate": 1.3515687851971038e-06, + "loss": 0.8167, + "step": 1064 + }, + { + "epoch": 1.5412445730824893, + "grad_norm": 1.3207247257232666, + "learning_rate": 1.347546259050684e-06, + "loss": 0.8224, + "step": 1065 + }, + { + "epoch": 1.5426917510853835, + "grad_norm": 1.4886301755905151, + "learning_rate": 1.3435237329042638e-06, + "loss": 0.8466, + "step": 1066 + }, + { + "epoch": 1.5441389290882779, + "grad_norm": 1.2021862268447876, + "learning_rate": 1.339501206757844e-06, + "loss": 0.818, + "step": 1067 + }, + { + "epoch": 1.545586107091172, + "grad_norm": 1.4846407175064087, + "learning_rate": 1.3354786806114243e-06, + "loss": 0.8336, + "step": 1068 + }, + { + "epoch": 1.5470332850940665, + "grad_norm": 1.4045193195343018, + "learning_rate": 1.3314561544650042e-06, + "loss": 0.8219, + "step": 1069 + }, + { + "epoch": 1.5484804630969609, + "grad_norm": 1.333292007446289, + "learning_rate": 1.3274336283185843e-06, + "loss": 0.8381, + "step": 1070 + }, + { + "epoch": 1.5484804630969609, + "eval_loss": 0.9582304954528809, + "eval_runtime": 46.7993, + "eval_samples_per_second": 21.368, + "eval_steps_per_second": 1.346, + "step": 1070 + }, + { + "epoch": 1.5499276410998553, + "grad_norm": 1.4133455753326416, + "learning_rate": 1.3234111021721643e-06, + "loss": 0.8439, + "step": 1071 + }, + { + "epoch": 1.5513748191027497, + "grad_norm": 1.2504298686981201, + "learning_rate": 1.3193885760257444e-06, + "loss": 0.808, + "step": 1072 + }, + { + "epoch": 1.552821997105644, + "grad_norm": 1.279789924621582, + "learning_rate": 1.3153660498793245e-06, + "loss": 0.8184, + "step": 1073 + }, + { + "epoch": 1.5542691751085385, + "grad_norm": 1.3571714162826538, + "learning_rate": 1.3113435237329044e-06, + "loss": 0.8313, + "step": 1074 + }, + { + "epoch": 1.5557163531114329, + "grad_norm": 1.301203966140747, + "learning_rate": 1.3073209975864846e-06, + "loss": 0.825, + "step": 1075 + }, + { + "epoch": 1.557163531114327, + "grad_norm": 1.4094393253326416, + "learning_rate": 1.3032984714400645e-06, + "loss": 0.801, + "step": 1076 + }, + { + "epoch": 1.5586107091172214, + "grad_norm": 1.350214958190918, + "learning_rate": 1.2992759452936446e-06, + "loss": 0.792, + "step": 1077 + }, + { + "epoch": 1.5600578871201156, + "grad_norm": 1.3494690656661987, + "learning_rate": 1.2952534191472247e-06, + "loss": 0.8407, + "step": 1078 + }, + { + "epoch": 1.56150506512301, + "grad_norm": 1.416135549545288, + "learning_rate": 1.2912308930008046e-06, + "loss": 0.7907, + "step": 1079 + }, + { + "epoch": 1.5629522431259044, + "grad_norm": 1.2209076881408691, + "learning_rate": 1.2872083668543848e-06, + "loss": 0.789, + "step": 1080 + }, + { + "epoch": 1.5629522431259044, + "eval_loss": 0.9584857225418091, + "eval_runtime": 46.8076, + "eval_samples_per_second": 21.364, + "eval_steps_per_second": 1.346, + "step": 1080 + }, + { + "epoch": 1.5643994211287988, + "grad_norm": 1.3523234128952026, + "learning_rate": 1.2831858407079647e-06, + "loss": 0.8593, + "step": 1081 + }, + { + "epoch": 1.5658465991316932, + "grad_norm": 1.2438052892684937, + "learning_rate": 1.2791633145615448e-06, + "loss": 0.828, + "step": 1082 + }, + { + "epoch": 1.5672937771345876, + "grad_norm": 1.3710579872131348, + "learning_rate": 1.275140788415125e-06, + "loss": 0.8285, + "step": 1083 + }, + { + "epoch": 1.568740955137482, + "grad_norm": 1.2357524633407593, + "learning_rate": 1.2711182622687049e-06, + "loss": 0.8563, + "step": 1084 + }, + { + "epoch": 1.5701881331403764, + "grad_norm": 1.367722749710083, + "learning_rate": 1.267095736122285e-06, + "loss": 0.8038, + "step": 1085 + }, + { + "epoch": 1.5716353111432706, + "grad_norm": 1.1723629236221313, + "learning_rate": 1.263073209975865e-06, + "loss": 0.8186, + "step": 1086 + }, + { + "epoch": 1.573082489146165, + "grad_norm": 1.2173646688461304, + "learning_rate": 1.259050683829445e-06, + "loss": 0.8189, + "step": 1087 + }, + { + "epoch": 1.5745296671490592, + "grad_norm": 1.3641481399536133, + "learning_rate": 1.2550281576830252e-06, + "loss": 0.8389, + "step": 1088 + }, + { + "epoch": 1.5759768451519536, + "grad_norm": 1.2916154861450195, + "learning_rate": 1.251005631536605e-06, + "loss": 0.8744, + "step": 1089 + }, + { + "epoch": 1.577424023154848, + "grad_norm": 1.3278297185897827, + "learning_rate": 1.2469831053901852e-06, + "loss": 0.8277, + "step": 1090 + }, + { + "epoch": 1.577424023154848, + "eval_loss": 0.9573075771331787, + "eval_runtime": 46.7567, + "eval_samples_per_second": 21.387, + "eval_steps_per_second": 1.347, + "step": 1090 + }, + { + "epoch": 1.5788712011577424, + "grad_norm": 1.3388468027114868, + "learning_rate": 1.2429605792437651e-06, + "loss": 0.8165, + "step": 1091 + }, + { + "epoch": 1.5803183791606368, + "grad_norm": 1.2376985549926758, + "learning_rate": 1.2389380530973452e-06, + "loss": 0.8166, + "step": 1092 + }, + { + "epoch": 1.5817655571635312, + "grad_norm": 1.3295481204986572, + "learning_rate": 1.2349155269509252e-06, + "loss": 0.8088, + "step": 1093 + }, + { + "epoch": 1.5832127351664256, + "grad_norm": 1.2857273817062378, + "learning_rate": 1.2308930008045053e-06, + "loss": 0.8277, + "step": 1094 + }, + { + "epoch": 1.58465991316932, + "grad_norm": 1.2978695631027222, + "learning_rate": 1.2268704746580854e-06, + "loss": 0.8519, + "step": 1095 + }, + { + "epoch": 1.5861070911722142, + "grad_norm": 1.2598134279251099, + "learning_rate": 1.2228479485116653e-06, + "loss": 0.8047, + "step": 1096 + }, + { + "epoch": 1.5875542691751086, + "grad_norm": 1.307244896888733, + "learning_rate": 1.2188254223652455e-06, + "loss": 0.779, + "step": 1097 + }, + { + "epoch": 1.5890014471780027, + "grad_norm": 1.2409782409667969, + "learning_rate": 1.2148028962188254e-06, + "loss": 0.8069, + "step": 1098 + }, + { + "epoch": 1.5904486251808971, + "grad_norm": 1.305288314819336, + "learning_rate": 1.2107803700724055e-06, + "loss": 0.82, + "step": 1099 + }, + { + "epoch": 1.5918958031837915, + "grad_norm": 1.2519434690475464, + "learning_rate": 1.2067578439259856e-06, + "loss": 0.8289, + "step": 1100 + }, + { + "epoch": 1.5918958031837915, + "eval_loss": 0.9581379294395447, + "eval_runtime": 46.7839, + "eval_samples_per_second": 21.375, + "eval_steps_per_second": 1.347, + "step": 1100 + }, + { + "epoch": 1.593342981186686, + "grad_norm": 1.3531020879745483, + "learning_rate": 1.2027353177795658e-06, + "loss": 0.8365, + "step": 1101 + }, + { + "epoch": 1.5947901591895803, + "grad_norm": 1.3726714849472046, + "learning_rate": 1.1987127916331457e-06, + "loss": 0.8094, + "step": 1102 + }, + { + "epoch": 1.5962373371924747, + "grad_norm": 1.2212868928909302, + "learning_rate": 1.1946902654867258e-06, + "loss": 0.8275, + "step": 1103 + }, + { + "epoch": 1.5976845151953691, + "grad_norm": 1.2503913640975952, + "learning_rate": 1.190667739340306e-06, + "loss": 0.8031, + "step": 1104 + }, + { + "epoch": 1.5991316931982635, + "grad_norm": 1.2780979871749878, + "learning_rate": 1.1866452131938858e-06, + "loss": 0.8171, + "step": 1105 + }, + { + "epoch": 1.6005788712011577, + "grad_norm": 1.241014838218689, + "learning_rate": 1.182622687047466e-06, + "loss": 0.8121, + "step": 1106 + }, + { + "epoch": 1.6020260492040521, + "grad_norm": 1.1598156690597534, + "learning_rate": 1.1786001609010459e-06, + "loss": 0.8208, + "step": 1107 + }, + { + "epoch": 1.6034732272069463, + "grad_norm": 1.2159602642059326, + "learning_rate": 1.174577634754626e-06, + "loss": 0.7949, + "step": 1108 + }, + { + "epoch": 1.6049204052098407, + "grad_norm": 1.310302495956421, + "learning_rate": 1.1705551086082061e-06, + "loss": 0.8116, + "step": 1109 + }, + { + "epoch": 1.606367583212735, + "grad_norm": 1.29893159866333, + "learning_rate": 1.166532582461786e-06, + "loss": 0.8299, + "step": 1110 + }, + { + "epoch": 1.606367583212735, + "eval_loss": 0.9566926956176758, + "eval_runtime": 46.9035, + "eval_samples_per_second": 21.32, + "eval_steps_per_second": 1.343, + "step": 1110 + }, + { + "epoch": 1.6078147612156295, + "grad_norm": 1.3786530494689941, + "learning_rate": 1.1625100563153662e-06, + "loss": 0.822, + "step": 1111 + }, + { + "epoch": 1.609261939218524, + "grad_norm": 1.3869152069091797, + "learning_rate": 1.158487530168946e-06, + "loss": 0.837, + "step": 1112 + }, + { + "epoch": 1.6107091172214183, + "grad_norm": 1.6763609647750854, + "learning_rate": 1.1544650040225262e-06, + "loss": 0.8344, + "step": 1113 + }, + { + "epoch": 1.6121562952243127, + "grad_norm": 1.2853882312774658, + "learning_rate": 1.1504424778761064e-06, + "loss": 0.8225, + "step": 1114 + }, + { + "epoch": 1.613603473227207, + "grad_norm": 1.2562637329101562, + "learning_rate": 1.1464199517296863e-06, + "loss": 0.8095, + "step": 1115 + }, + { + "epoch": 1.6150506512301013, + "grad_norm": 1.4541890621185303, + "learning_rate": 1.1423974255832664e-06, + "loss": 0.8695, + "step": 1116 + }, + { + "epoch": 1.6164978292329957, + "grad_norm": 1.3095476627349854, + "learning_rate": 1.1383748994368463e-06, + "loss": 0.833, + "step": 1117 + }, + { + "epoch": 1.6179450072358899, + "grad_norm": 1.374765157699585, + "learning_rate": 1.1343523732904264e-06, + "loss": 0.8088, + "step": 1118 + }, + { + "epoch": 1.6193921852387843, + "grad_norm": 1.3418219089508057, + "learning_rate": 1.1303298471440066e-06, + "loss": 0.8448, + "step": 1119 + }, + { + "epoch": 1.6208393632416787, + "grad_norm": 1.287712812423706, + "learning_rate": 1.1263073209975865e-06, + "loss": 0.8302, + "step": 1120 + }, + { + "epoch": 1.6208393632416787, + "eval_loss": 0.9556211829185486, + "eval_runtime": 46.7798, + "eval_samples_per_second": 21.377, + "eval_steps_per_second": 1.347, + "step": 1120 + }, + { + "epoch": 1.622286541244573, + "grad_norm": 1.290711522102356, + "learning_rate": 1.1222847948511666e-06, + "loss": 0.8335, + "step": 1121 + }, + { + "epoch": 1.6237337192474675, + "grad_norm": 1.296345591545105, + "learning_rate": 1.1182622687047465e-06, + "loss": 0.8376, + "step": 1122 + }, + { + "epoch": 1.6251808972503619, + "grad_norm": 1.2819390296936035, + "learning_rate": 1.1142397425583269e-06, + "loss": 0.8084, + "step": 1123 + }, + { + "epoch": 1.6266280752532563, + "grad_norm": 1.3919808864593506, + "learning_rate": 1.1102172164119068e-06, + "loss": 0.825, + "step": 1124 + }, + { + "epoch": 1.6280752532561507, + "grad_norm": 1.2804007530212402, + "learning_rate": 1.106194690265487e-06, + "loss": 0.8151, + "step": 1125 + }, + { + "epoch": 1.6295224312590448, + "grad_norm": 1.3394403457641602, + "learning_rate": 1.1021721641190668e-06, + "loss": 0.8116, + "step": 1126 + }, + { + "epoch": 1.6309696092619392, + "grad_norm": 1.2278305292129517, + "learning_rate": 1.098149637972647e-06, + "loss": 0.8118, + "step": 1127 + }, + { + "epoch": 1.6324167872648334, + "grad_norm": 1.513374924659729, + "learning_rate": 1.094127111826227e-06, + "loss": 0.8271, + "step": 1128 + }, + { + "epoch": 1.6338639652677278, + "grad_norm": 1.3171628713607788, + "learning_rate": 1.090104585679807e-06, + "loss": 0.8375, + "step": 1129 + }, + { + "epoch": 1.6353111432706222, + "grad_norm": 1.3745139837265015, + "learning_rate": 1.0860820595333871e-06, + "loss": 0.8268, + "step": 1130 + }, + { + "epoch": 1.6353111432706222, + "eval_loss": 0.9548259377479553, + "eval_runtime": 46.7473, + "eval_samples_per_second": 21.392, + "eval_steps_per_second": 1.348, + "step": 1130 + }, + { + "epoch": 1.6367583212735166, + "grad_norm": 1.3768346309661865, + "learning_rate": 1.082059533386967e-06, + "loss": 0.8066, + "step": 1131 + }, + { + "epoch": 1.638205499276411, + "grad_norm": 1.2537806034088135, + "learning_rate": 1.0780370072405472e-06, + "loss": 0.8425, + "step": 1132 + }, + { + "epoch": 1.6396526772793054, + "grad_norm": 1.3096121549606323, + "learning_rate": 1.0740144810941273e-06, + "loss": 0.7963, + "step": 1133 + }, + { + "epoch": 1.6410998552821998, + "grad_norm": 1.3774876594543457, + "learning_rate": 1.0699919549477072e-06, + "loss": 0.8372, + "step": 1134 + }, + { + "epoch": 1.6425470332850942, + "grad_norm": 1.5008494853973389, + "learning_rate": 1.0659694288012873e-06, + "loss": 0.8579, + "step": 1135 + }, + { + "epoch": 1.6439942112879884, + "grad_norm": 1.390331745147705, + "learning_rate": 1.0619469026548673e-06, + "loss": 0.8015, + "step": 1136 + }, + { + "epoch": 1.6454413892908828, + "grad_norm": 1.2677420377731323, + "learning_rate": 1.0579243765084474e-06, + "loss": 0.8552, + "step": 1137 + }, + { + "epoch": 1.646888567293777, + "grad_norm": 1.4993586540222168, + "learning_rate": 1.0539018503620275e-06, + "loss": 0.8328, + "step": 1138 + }, + { + "epoch": 1.6483357452966714, + "grad_norm": 1.5009984970092773, + "learning_rate": 1.0498793242156074e-06, + "loss": 0.7938, + "step": 1139 + }, + { + "epoch": 1.6497829232995658, + "grad_norm": 1.2156932353973389, + "learning_rate": 1.0458567980691876e-06, + "loss": 0.8108, + "step": 1140 + }, + { + "epoch": 1.6497829232995658, + "eval_loss": 0.9526401162147522, + "eval_runtime": 46.7129, + "eval_samples_per_second": 21.407, + "eval_steps_per_second": 1.349, + "step": 1140 + }, + { + "epoch": 1.6512301013024602, + "grad_norm": 1.2374553680419922, + "learning_rate": 1.0418342719227675e-06, + "loss": 0.8333, + "step": 1141 + }, + { + "epoch": 1.6526772793053546, + "grad_norm": 1.3393139839172363, + "learning_rate": 1.0378117457763476e-06, + "loss": 0.8136, + "step": 1142 + }, + { + "epoch": 1.654124457308249, + "grad_norm": 1.4265186786651611, + "learning_rate": 1.0337892196299275e-06, + "loss": 0.842, + "step": 1143 + }, + { + "epoch": 1.6555716353111434, + "grad_norm": 1.3121803998947144, + "learning_rate": 1.0297666934835076e-06, + "loss": 0.8142, + "step": 1144 + }, + { + "epoch": 1.6570188133140378, + "grad_norm": 1.244606614112854, + "learning_rate": 1.0257441673370878e-06, + "loss": 0.8331, + "step": 1145 + }, + { + "epoch": 1.658465991316932, + "grad_norm": 1.4439839124679565, + "learning_rate": 1.021721641190668e-06, + "loss": 0.7917, + "step": 1146 + }, + { + "epoch": 1.6599131693198264, + "grad_norm": 1.297786831855774, + "learning_rate": 1.017699115044248e-06, + "loss": 0.8212, + "step": 1147 + }, + { + "epoch": 1.6613603473227205, + "grad_norm": 1.4428986310958862, + "learning_rate": 1.013676588897828e-06, + "loss": 0.8457, + "step": 1148 + }, + { + "epoch": 1.662807525325615, + "grad_norm": 1.3428438901901245, + "learning_rate": 1.009654062751408e-06, + "loss": 0.8067, + "step": 1149 + }, + { + "epoch": 1.6642547033285093, + "grad_norm": 1.3897366523742676, + "learning_rate": 1.005631536604988e-06, + "loss": 0.8096, + "step": 1150 + }, + { + "epoch": 1.6642547033285093, + "eval_loss": 0.9543867111206055, + "eval_runtime": 46.7995, + "eval_samples_per_second": 21.368, + "eval_steps_per_second": 1.346, + "step": 1150 + }, + { + "epoch": 1.6657018813314037, + "grad_norm": 1.3801957368850708, + "learning_rate": 1.0016090104585681e-06, + "loss": 0.825, + "step": 1151 + }, + { + "epoch": 1.6671490593342981, + "grad_norm": 1.254943609237671, + "learning_rate": 9.97586484312148e-07, + "loss": 0.8252, + "step": 1152 + }, + { + "epoch": 1.6685962373371925, + "grad_norm": 1.3884161710739136, + "learning_rate": 9.935639581657282e-07, + "loss": 0.8182, + "step": 1153 + }, + { + "epoch": 1.670043415340087, + "grad_norm": 1.2761136293411255, + "learning_rate": 9.895414320193083e-07, + "loss": 0.8114, + "step": 1154 + }, + { + "epoch": 1.6714905933429813, + "grad_norm": 1.242130994796753, + "learning_rate": 9.855189058728882e-07, + "loss": 0.8399, + "step": 1155 + }, + { + "epoch": 1.6729377713458755, + "grad_norm": 1.3890230655670166, + "learning_rate": 9.814963797264683e-07, + "loss": 0.8292, + "step": 1156 + }, + { + "epoch": 1.67438494934877, + "grad_norm": 1.2863956689834595, + "learning_rate": 9.774738535800482e-07, + "loss": 0.8586, + "step": 1157 + }, + { + "epoch": 1.675832127351664, + "grad_norm": 1.4454671144485474, + "learning_rate": 9.734513274336284e-07, + "loss": 0.8107, + "step": 1158 + }, + { + "epoch": 1.6772793053545585, + "grad_norm": 1.470796823501587, + "learning_rate": 9.694288012872085e-07, + "loss": 0.7897, + "step": 1159 + }, + { + "epoch": 1.678726483357453, + "grad_norm": 1.2682064771652222, + "learning_rate": 9.654062751407884e-07, + "loss": 0.7942, + "step": 1160 + }, + { + "epoch": 1.678726483357453, + "eval_loss": 0.9547902941703796, + "eval_runtime": 46.7057, + "eval_samples_per_second": 21.411, + "eval_steps_per_second": 1.349, + "step": 1160 + }, + { + "epoch": 1.6801736613603473, + "grad_norm": 1.2822381258010864, + "learning_rate": 9.613837489943685e-07, + "loss": 0.7833, + "step": 1161 + }, + { + "epoch": 1.6816208393632417, + "grad_norm": 1.3998022079467773, + "learning_rate": 9.573612228479485e-07, + "loss": 0.7946, + "step": 1162 + }, + { + "epoch": 1.683068017366136, + "grad_norm": 1.5702491998672485, + "learning_rate": 9.533386967015286e-07, + "loss": 0.7958, + "step": 1163 + }, + { + "epoch": 1.6845151953690305, + "grad_norm": 1.2740254402160645, + "learning_rate": 9.493161705551086e-07, + "loss": 0.8158, + "step": 1164 + }, + { + "epoch": 1.685962373371925, + "grad_norm": 1.4471359252929688, + "learning_rate": 9.452936444086886e-07, + "loss": 0.8156, + "step": 1165 + }, + { + "epoch": 1.687409551374819, + "grad_norm": 1.4359145164489746, + "learning_rate": 9.412711182622688e-07, + "loss": 0.8601, + "step": 1166 + }, + { + "epoch": 1.6888567293777135, + "grad_norm": 1.269058108329773, + "learning_rate": 9.372485921158489e-07, + "loss": 0.8091, + "step": 1167 + }, + { + "epoch": 1.6903039073806077, + "grad_norm": 1.4691559076309204, + "learning_rate": 9.332260659694289e-07, + "loss": 0.8549, + "step": 1168 + }, + { + "epoch": 1.691751085383502, + "grad_norm": 1.4363200664520264, + "learning_rate": 9.292035398230089e-07, + "loss": 0.7995, + "step": 1169 + }, + { + "epoch": 1.6931982633863965, + "grad_norm": 1.4528241157531738, + "learning_rate": 9.251810136765891e-07, + "loss": 0.8142, + "step": 1170 + }, + { + "epoch": 1.6931982633863965, + "eval_loss": 0.9541338086128235, + "eval_runtime": 46.7711, + "eval_samples_per_second": 21.381, + "eval_steps_per_second": 1.347, + "step": 1170 + }, + { + "epoch": 1.6946454413892909, + "grad_norm": 1.4791617393493652, + "learning_rate": 9.211584875301691e-07, + "loss": 0.8236, + "step": 1171 + }, + { + "epoch": 1.6960926193921853, + "grad_norm": 1.4114556312561035, + "learning_rate": 9.171359613837491e-07, + "loss": 0.815, + "step": 1172 + }, + { + "epoch": 1.6975397973950797, + "grad_norm": 1.3555424213409424, + "learning_rate": 9.131134352373291e-07, + "loss": 0.8267, + "step": 1173 + }, + { + "epoch": 1.698986975397974, + "grad_norm": 1.2887614965438843, + "learning_rate": 9.090909090909091e-07, + "loss": 0.8325, + "step": 1174 + }, + { + "epoch": 1.7004341534008685, + "grad_norm": 1.4596766233444214, + "learning_rate": 9.050683829444893e-07, + "loss": 0.8313, + "step": 1175 + }, + { + "epoch": 1.7018813314037626, + "grad_norm": 1.782693862915039, + "learning_rate": 9.010458567980693e-07, + "loss": 0.8206, + "step": 1176 + }, + { + "epoch": 1.703328509406657, + "grad_norm": 1.4860711097717285, + "learning_rate": 8.970233306516493e-07, + "loss": 0.8045, + "step": 1177 + }, + { + "epoch": 1.7047756874095512, + "grad_norm": 1.450358510017395, + "learning_rate": 8.930008045052293e-07, + "loss": 0.8315, + "step": 1178 + }, + { + "epoch": 1.7062228654124456, + "grad_norm": 1.2317721843719482, + "learning_rate": 8.889782783588094e-07, + "loss": 0.7923, + "step": 1179 + }, + { + "epoch": 1.70767004341534, + "grad_norm": 1.232118844985962, + "learning_rate": 8.849557522123895e-07, + "loss": 0.7882, + "step": 1180 + }, + { + "epoch": 1.70767004341534, + "eval_loss": 0.9545546174049377, + "eval_runtime": 46.6971, + "eval_samples_per_second": 21.415, + "eval_steps_per_second": 1.349, + "step": 1180 + }, + { + "epoch": 1.7091172214182344, + "grad_norm": 1.4816395044326782, + "learning_rate": 8.809332260659695e-07, + "loss": 0.8149, + "step": 1181 + }, + { + "epoch": 1.7105643994211288, + "grad_norm": 1.3291946649551392, + "learning_rate": 8.769106999195495e-07, + "loss": 0.8133, + "step": 1182 + }, + { + "epoch": 1.7120115774240232, + "grad_norm": 1.3862465620040894, + "learning_rate": 8.728881737731296e-07, + "loss": 0.8097, + "step": 1183 + }, + { + "epoch": 1.7134587554269176, + "grad_norm": 1.4611399173736572, + "learning_rate": 8.688656476267096e-07, + "loss": 0.7999, + "step": 1184 + }, + { + "epoch": 1.714905933429812, + "grad_norm": 1.3023141622543335, + "learning_rate": 8.648431214802896e-07, + "loss": 0.842, + "step": 1185 + }, + { + "epoch": 1.7163531114327062, + "grad_norm": 1.3368428945541382, + "learning_rate": 8.608205953338697e-07, + "loss": 0.8466, + "step": 1186 + }, + { + "epoch": 1.7178002894356006, + "grad_norm": 1.5064977407455444, + "learning_rate": 8.567980691874497e-07, + "loss": 0.8368, + "step": 1187 + }, + { + "epoch": 1.7192474674384948, + "grad_norm": 1.4122167825698853, + "learning_rate": 8.527755430410298e-07, + "loss": 0.825, + "step": 1188 + }, + { + "epoch": 1.7206946454413892, + "grad_norm": 1.2917824983596802, + "learning_rate": 8.487530168946098e-07, + "loss": 0.8353, + "step": 1189 + }, + { + "epoch": 1.7221418234442836, + "grad_norm": 1.2089941501617432, + "learning_rate": 8.4473049074819e-07, + "loss": 0.8291, + "step": 1190 + }, + { + "epoch": 1.7221418234442836, + "eval_loss": 0.9527277946472168, + "eval_runtime": 46.7861, + "eval_samples_per_second": 21.374, + "eval_steps_per_second": 1.347, + "step": 1190 + }, + { + "epoch": 1.723589001447178, + "grad_norm": 1.4512019157409668, + "learning_rate": 8.4070796460177e-07, + "loss": 0.8537, + "step": 1191 + }, + { + "epoch": 1.7250361794500724, + "grad_norm": 1.361953616142273, + "learning_rate": 8.366854384553501e-07, + "loss": 0.8195, + "step": 1192 + }, + { + "epoch": 1.7264833574529668, + "grad_norm": 1.3929336071014404, + "learning_rate": 8.326629123089301e-07, + "loss": 0.8209, + "step": 1193 + }, + { + "epoch": 1.7279305354558612, + "grad_norm": 1.2415398359298706, + "learning_rate": 8.286403861625101e-07, + "loss": 0.8171, + "step": 1194 + }, + { + "epoch": 1.7293777134587556, + "grad_norm": 1.4853307008743286, + "learning_rate": 8.246178600160902e-07, + "loss": 0.82, + "step": 1195 + }, + { + "epoch": 1.7308248914616498, + "grad_norm": 1.492444634437561, + "learning_rate": 8.205953338696703e-07, + "loss": 0.8445, + "step": 1196 + }, + { + "epoch": 1.7322720694645442, + "grad_norm": 1.2242611646652222, + "learning_rate": 8.165728077232503e-07, + "loss": 0.8173, + "step": 1197 + }, + { + "epoch": 1.7337192474674383, + "grad_norm": 1.9206209182739258, + "learning_rate": 8.125502815768303e-07, + "loss": 0.8503, + "step": 1198 + }, + { + "epoch": 1.7351664254703327, + "grad_norm": 1.289128303527832, + "learning_rate": 8.085277554304103e-07, + "loss": 0.8083, + "step": 1199 + }, + { + "epoch": 1.7366136034732271, + "grad_norm": 1.3578416109085083, + "learning_rate": 8.045052292839905e-07, + "loss": 0.8345, + "step": 1200 + }, + { + "epoch": 1.7366136034732271, + "eval_loss": 0.9532684087753296, + "eval_runtime": 46.7272, + "eval_samples_per_second": 21.401, + "eval_steps_per_second": 1.348, + "step": 1200 + }, + { + "epoch": 1.7380607814761215, + "grad_norm": 1.329463005065918, + "learning_rate": 8.004827031375705e-07, + "loss": 0.8034, + "step": 1201 + }, + { + "epoch": 1.739507959479016, + "grad_norm": 1.5638056993484497, + "learning_rate": 7.964601769911505e-07, + "loss": 0.8327, + "step": 1202 + }, + { + "epoch": 1.7409551374819103, + "grad_norm": 1.3703676462173462, + "learning_rate": 7.924376508447305e-07, + "loss": 0.8096, + "step": 1203 + }, + { + "epoch": 1.7424023154848047, + "grad_norm": 1.2262520790100098, + "learning_rate": 7.884151246983105e-07, + "loss": 0.8095, + "step": 1204 + }, + { + "epoch": 1.7438494934876991, + "grad_norm": 1.4088160991668701, + "learning_rate": 7.843925985518907e-07, + "loss": 0.8313, + "step": 1205 + }, + { + "epoch": 1.7452966714905933, + "grad_norm": 1.3341182470321655, + "learning_rate": 7.803700724054707e-07, + "loss": 0.8413, + "step": 1206 + }, + { + "epoch": 1.7467438494934877, + "grad_norm": 1.4329415559768677, + "learning_rate": 7.763475462590507e-07, + "loss": 0.8221, + "step": 1207 + }, + { + "epoch": 1.7481910274963819, + "grad_norm": 1.2600018978118896, + "learning_rate": 7.723250201126307e-07, + "loss": 0.812, + "step": 1208 + }, + { + "epoch": 1.7496382054992763, + "grad_norm": 1.294754147529602, + "learning_rate": 7.683024939662108e-07, + "loss": 0.8435, + "step": 1209 + }, + { + "epoch": 1.7510853835021707, + "grad_norm": 1.27560555934906, + "learning_rate": 7.642799678197908e-07, + "loss": 0.7811, + "step": 1210 + }, + { + "epoch": 1.7510853835021707, + "eval_loss": 0.9537524580955505, + "eval_runtime": 46.6127, + "eval_samples_per_second": 21.453, + "eval_steps_per_second": 1.352, + "step": 1210 + }, + { + "epoch": 1.752532561505065, + "grad_norm": 1.3399219512939453, + "learning_rate": 7.602574416733709e-07, + "loss": 0.8623, + "step": 1211 + }, + { + "epoch": 1.7539797395079595, + "grad_norm": 1.2824227809906006, + "learning_rate": 7.56234915526951e-07, + "loss": 0.8127, + "step": 1212 + }, + { + "epoch": 1.755426917510854, + "grad_norm": 1.4298566579818726, + "learning_rate": 7.522123893805311e-07, + "loss": 0.8048, + "step": 1213 + }, + { + "epoch": 1.7568740955137483, + "grad_norm": 1.3512744903564453, + "learning_rate": 7.481898632341112e-07, + "loss": 0.8151, + "step": 1214 + }, + { + "epoch": 1.7583212735166427, + "grad_norm": 1.392371416091919, + "learning_rate": 7.441673370876912e-07, + "loss": 0.8084, + "step": 1215 + }, + { + "epoch": 1.7597684515195369, + "grad_norm": 1.312528371810913, + "learning_rate": 7.401448109412712e-07, + "loss": 0.8208, + "step": 1216 + }, + { + "epoch": 1.7612156295224313, + "grad_norm": 1.2709057331085205, + "learning_rate": 7.361222847948512e-07, + "loss": 0.8358, + "step": 1217 + }, + { + "epoch": 1.7626628075253257, + "grad_norm": 1.2670097351074219, + "learning_rate": 7.320997586484313e-07, + "loss": 0.8175, + "step": 1218 + }, + { + "epoch": 1.7641099855282198, + "grad_norm": 1.3621159791946411, + "learning_rate": 7.280772325020113e-07, + "loss": 0.8162, + "step": 1219 + }, + { + "epoch": 1.7655571635311142, + "grad_norm": 1.2618505954742432, + "learning_rate": 7.240547063555914e-07, + "loss": 0.8284, + "step": 1220 + }, + { + "epoch": 1.7655571635311142, + "eval_loss": 0.9540891051292419, + "eval_runtime": 46.7409, + "eval_samples_per_second": 21.395, + "eval_steps_per_second": 1.348, + "step": 1220 + }, + { + "epoch": 1.7670043415340086, + "grad_norm": 1.3684877157211304, + "learning_rate": 7.200321802091714e-07, + "loss": 0.804, + "step": 1221 + }, + { + "epoch": 1.768451519536903, + "grad_norm": 1.4869263172149658, + "learning_rate": 7.160096540627515e-07, + "loss": 0.8131, + "step": 1222 + }, + { + "epoch": 1.7698986975397974, + "grad_norm": 1.2008804082870483, + "learning_rate": 7.119871279163315e-07, + "loss": 0.8358, + "step": 1223 + }, + { + "epoch": 1.7713458755426919, + "grad_norm": 1.2698451280593872, + "learning_rate": 7.079646017699115e-07, + "loss": 0.8199, + "step": 1224 + }, + { + "epoch": 1.7727930535455863, + "grad_norm": 1.33381187915802, + "learning_rate": 7.039420756234916e-07, + "loss": 0.8068, + "step": 1225 + }, + { + "epoch": 1.7742402315484804, + "grad_norm": 1.3051142692565918, + "learning_rate": 6.999195494770717e-07, + "loss": 0.8347, + "step": 1226 + }, + { + "epoch": 1.7756874095513748, + "grad_norm": 1.4349479675292969, + "learning_rate": 6.958970233306517e-07, + "loss": 0.783, + "step": 1227 + }, + { + "epoch": 1.7771345875542692, + "grad_norm": 1.2964993715286255, + "learning_rate": 6.918744971842317e-07, + "loss": 0.8032, + "step": 1228 + }, + { + "epoch": 1.7785817655571634, + "grad_norm": 1.282899022102356, + "learning_rate": 6.878519710378117e-07, + "loss": 0.7836, + "step": 1229 + }, + { + "epoch": 1.7800289435600578, + "grad_norm": 1.261360764503479, + "learning_rate": 6.838294448913918e-07, + "loss": 0.8272, + "step": 1230 + }, + { + "epoch": 1.7800289435600578, + "eval_loss": 0.9520903825759888, + "eval_runtime": 46.6839, + "eval_samples_per_second": 21.421, + "eval_steps_per_second": 1.35, + "step": 1230 + }, + { + "epoch": 1.7814761215629522, + "grad_norm": 1.251921534538269, + "learning_rate": 6.798069187449719e-07, + "loss": 0.8327, + "step": 1231 + }, + { + "epoch": 1.7829232995658466, + "grad_norm": 1.294122338294983, + "learning_rate": 6.757843925985519e-07, + "loss": 0.8111, + "step": 1232 + }, + { + "epoch": 1.784370477568741, + "grad_norm": 1.2234042882919312, + "learning_rate": 6.717618664521319e-07, + "loss": 0.8169, + "step": 1233 + }, + { + "epoch": 1.7858176555716354, + "grad_norm": 1.2441366910934448, + "learning_rate": 6.677393403057121e-07, + "loss": 0.8179, + "step": 1234 + }, + { + "epoch": 1.7872648335745298, + "grad_norm": 1.3525168895721436, + "learning_rate": 6.637168141592922e-07, + "loss": 0.8335, + "step": 1235 + }, + { + "epoch": 1.788712011577424, + "grad_norm": 1.3566505908966064, + "learning_rate": 6.596942880128722e-07, + "loss": 0.8163, + "step": 1236 + }, + { + "epoch": 1.7901591895803184, + "grad_norm": 1.252092957496643, + "learning_rate": 6.556717618664522e-07, + "loss": 0.8007, + "step": 1237 + }, + { + "epoch": 1.7916063675832128, + "grad_norm": 1.3040640354156494, + "learning_rate": 6.516492357200322e-07, + "loss": 0.8206, + "step": 1238 + }, + { + "epoch": 1.793053545586107, + "grad_norm": 1.3594927787780762, + "learning_rate": 6.476267095736124e-07, + "loss": 0.8271, + "step": 1239 + }, + { + "epoch": 1.7945007235890014, + "grad_norm": 1.2246612310409546, + "learning_rate": 6.436041834271924e-07, + "loss": 0.8103, + "step": 1240 + }, + { + "epoch": 1.7945007235890014, + "eval_loss": 0.9517695307731628, + "eval_runtime": 46.7422, + "eval_samples_per_second": 21.394, + "eval_steps_per_second": 1.348, + "step": 1240 + }, + { + "epoch": 1.7959479015918958, + "grad_norm": 1.2993959188461304, + "learning_rate": 6.395816572807724e-07, + "loss": 0.7967, + "step": 1241 + }, + { + "epoch": 1.7973950795947902, + "grad_norm": 1.4147443771362305, + "learning_rate": 6.355591311343524e-07, + "loss": 0.8372, + "step": 1242 + }, + { + "epoch": 1.7988422575976846, + "grad_norm": 1.3695588111877441, + "learning_rate": 6.315366049879324e-07, + "loss": 0.8345, + "step": 1243 + }, + { + "epoch": 1.800289435600579, + "grad_norm": 1.3577102422714233, + "learning_rate": 6.275140788415126e-07, + "loss": 0.8051, + "step": 1244 + }, + { + "epoch": 1.8017366136034734, + "grad_norm": 1.2265558242797852, + "learning_rate": 6.234915526950926e-07, + "loss": 0.8087, + "step": 1245 + }, + { + "epoch": 1.8031837916063675, + "grad_norm": 1.3047088384628296, + "learning_rate": 6.194690265486726e-07, + "loss": 0.8325, + "step": 1246 + }, + { + "epoch": 1.804630969609262, + "grad_norm": 1.303094506263733, + "learning_rate": 6.154465004022526e-07, + "loss": 0.8368, + "step": 1247 + }, + { + "epoch": 1.8060781476121563, + "grad_norm": 1.2603576183319092, + "learning_rate": 6.114239742558327e-07, + "loss": 0.8103, + "step": 1248 + }, + { + "epoch": 1.8075253256150505, + "grad_norm": 1.45889413356781, + "learning_rate": 6.074014481094127e-07, + "loss": 0.8358, + "step": 1249 + }, + { + "epoch": 1.808972503617945, + "grad_norm": 1.342335820198059, + "learning_rate": 6.033789219629928e-07, + "loss": 0.8255, + "step": 1250 + }, + { + "epoch": 1.808972503617945, + "eval_loss": 0.9513235688209534, + "eval_runtime": 46.6471, + "eval_samples_per_second": 21.438, + "eval_steps_per_second": 1.351, + "step": 1250 + }, + { + "epoch": 1.8104196816208393, + "grad_norm": 1.258847951889038, + "learning_rate": 5.993563958165728e-07, + "loss": 0.8142, + "step": 1251 + }, + { + "epoch": 1.8118668596237337, + "grad_norm": 1.2348954677581787, + "learning_rate": 5.95333869670153e-07, + "loss": 0.8151, + "step": 1252 + }, + { + "epoch": 1.8133140376266281, + "grad_norm": 1.2233668565750122, + "learning_rate": 5.91311343523733e-07, + "loss": 0.8124, + "step": 1253 + }, + { + "epoch": 1.8147612156295225, + "grad_norm": 1.358266830444336, + "learning_rate": 5.87288817377313e-07, + "loss": 0.8236, + "step": 1254 + }, + { + "epoch": 1.816208393632417, + "grad_norm": 1.1655139923095703, + "learning_rate": 5.83266291230893e-07, + "loss": 0.8451, + "step": 1255 + }, + { + "epoch": 1.817655571635311, + "grad_norm": 1.2099170684814453, + "learning_rate": 5.79243765084473e-07, + "loss": 0.8064, + "step": 1256 + }, + { + "epoch": 1.8191027496382055, + "grad_norm": 1.322885274887085, + "learning_rate": 5.752212389380532e-07, + "loss": 0.8133, + "step": 1257 + }, + { + "epoch": 1.8205499276411, + "grad_norm": 1.3825887441635132, + "learning_rate": 5.711987127916332e-07, + "loss": 0.8181, + "step": 1258 + }, + { + "epoch": 1.821997105643994, + "grad_norm": 1.2636319398880005, + "learning_rate": 5.671761866452132e-07, + "loss": 0.8099, + "step": 1259 + }, + { + "epoch": 1.8234442836468885, + "grad_norm": 1.297945261001587, + "learning_rate": 5.631536604987932e-07, + "loss": 0.8394, + "step": 1260 + }, + { + "epoch": 1.8234442836468885, + "eval_loss": 0.9511587023735046, + "eval_runtime": 46.7528, + "eval_samples_per_second": 21.389, + "eval_steps_per_second": 1.348, + "step": 1260 + }, + { + "epoch": 1.8248914616497829, + "grad_norm": 1.303533911705017, + "learning_rate": 5.591311343523733e-07, + "loss": 0.8241, + "step": 1261 + }, + { + "epoch": 1.8263386396526773, + "grad_norm": 1.3721277713775635, + "learning_rate": 5.551086082059534e-07, + "loss": 0.7987, + "step": 1262 + }, + { + "epoch": 1.8277858176555717, + "grad_norm": 1.2044222354888916, + "learning_rate": 5.510860820595334e-07, + "loss": 0.7958, + "step": 1263 + }, + { + "epoch": 1.829232995658466, + "grad_norm": 1.218172550201416, + "learning_rate": 5.470635559131135e-07, + "loss": 0.8303, + "step": 1264 + }, + { + "epoch": 1.8306801736613605, + "grad_norm": 1.320393681526184, + "learning_rate": 5.430410297666936e-07, + "loss": 0.8183, + "step": 1265 + }, + { + "epoch": 1.8321273516642547, + "grad_norm": 1.4130040407180786, + "learning_rate": 5.390185036202736e-07, + "loss": 0.7949, + "step": 1266 + }, + { + "epoch": 1.833574529667149, + "grad_norm": 1.290032148361206, + "learning_rate": 5.349959774738536e-07, + "loss": 0.8245, + "step": 1267 + }, + { + "epoch": 1.8350217076700435, + "grad_norm": 1.2038600444793701, + "learning_rate": 5.309734513274336e-07, + "loss": 0.7764, + "step": 1268 + }, + { + "epoch": 1.8364688856729376, + "grad_norm": 1.3405817747116089, + "learning_rate": 5.269509251810138e-07, + "loss": 0.8273, + "step": 1269 + }, + { + "epoch": 1.837916063675832, + "grad_norm": 1.392913579940796, + "learning_rate": 5.229283990345938e-07, + "loss": 0.8458, + "step": 1270 + }, + { + "epoch": 1.837916063675832, + "eval_loss": 0.9525765180587769, + "eval_runtime": 46.742, + "eval_samples_per_second": 21.394, + "eval_steps_per_second": 1.348, + "step": 1270 + }, + { + "epoch": 1.8393632416787264, + "grad_norm": 1.3695766925811768, + "learning_rate": 5.189058728881738e-07, + "loss": 0.8161, + "step": 1271 + }, + { + "epoch": 1.8408104196816208, + "grad_norm": 1.374796986579895, + "learning_rate": 5.148833467417538e-07, + "loss": 0.8347, + "step": 1272 + }, + { + "epoch": 1.8422575976845152, + "grad_norm": 1.2616554498672485, + "learning_rate": 5.10860820595334e-07, + "loss": 0.81, + "step": 1273 + }, + { + "epoch": 1.8437047756874096, + "grad_norm": 1.4010777473449707, + "learning_rate": 5.06838294448914e-07, + "loss": 0.8326, + "step": 1274 + }, + { + "epoch": 1.845151953690304, + "grad_norm": 1.379961371421814, + "learning_rate": 5.02815768302494e-07, + "loss": 0.8134, + "step": 1275 + }, + { + "epoch": 1.8465991316931982, + "grad_norm": 1.3502508401870728, + "learning_rate": 4.98793242156074e-07, + "loss": 0.8012, + "step": 1276 + }, + { + "epoch": 1.8480463096960926, + "grad_norm": 1.2990753650665283, + "learning_rate": 4.947707160096541e-07, + "loss": 0.8174, + "step": 1277 + }, + { + "epoch": 1.849493487698987, + "grad_norm": 1.2463006973266602, + "learning_rate": 4.907481898632342e-07, + "loss": 0.8276, + "step": 1278 + }, + { + "epoch": 1.8509406657018812, + "grad_norm": 1.28299880027771, + "learning_rate": 4.867256637168142e-07, + "loss": 0.8104, + "step": 1279 + }, + { + "epoch": 1.8523878437047756, + "grad_norm": 1.2849760055541992, + "learning_rate": 4.827031375703942e-07, + "loss": 0.8119, + "step": 1280 + }, + { + "epoch": 1.8523878437047756, + "eval_loss": 0.9507870674133301, + "eval_runtime": 46.698, + "eval_samples_per_second": 21.414, + "eval_steps_per_second": 1.349, + "step": 1280 + }, + { + "epoch": 1.85383502170767, + "grad_norm": 1.308908224105835, + "learning_rate": 4.786806114239742e-07, + "loss": 0.788, + "step": 1281 + }, + { + "epoch": 1.8552821997105644, + "grad_norm": 1.1943525075912476, + "learning_rate": 4.746580852775543e-07, + "loss": 0.7855, + "step": 1282 + }, + { + "epoch": 1.8567293777134588, + "grad_norm": 1.3147971630096436, + "learning_rate": 4.706355591311344e-07, + "loss": 0.8183, + "step": 1283 + }, + { + "epoch": 1.8581765557163532, + "grad_norm": 1.3953913450241089, + "learning_rate": 4.6661303298471445e-07, + "loss": 0.8408, + "step": 1284 + }, + { + "epoch": 1.8596237337192476, + "grad_norm": 1.6656665802001953, + "learning_rate": 4.6259050683829453e-07, + "loss": 0.8037, + "step": 1285 + }, + { + "epoch": 1.8610709117221418, + "grad_norm": 1.214592456817627, + "learning_rate": 4.5856798069187455e-07, + "loss": 0.83, + "step": 1286 + }, + { + "epoch": 1.8625180897250362, + "grad_norm": 1.220953345298767, + "learning_rate": 4.5454545454545457e-07, + "loss": 0.8003, + "step": 1287 + }, + { + "epoch": 1.8639652677279306, + "grad_norm": 1.275334358215332, + "learning_rate": 4.5052292839903465e-07, + "loss": 0.8123, + "step": 1288 + }, + { + "epoch": 1.8654124457308248, + "grad_norm": 1.2643712759017944, + "learning_rate": 4.4650040225261467e-07, + "loss": 0.8293, + "step": 1289 + }, + { + "epoch": 1.8668596237337192, + "grad_norm": 1.2619174718856812, + "learning_rate": 4.4247787610619474e-07, + "loss": 0.8413, + "step": 1290 + }, + { + "epoch": 1.8668596237337192, + "eval_loss": 0.9520008563995361, + "eval_runtime": 46.8042, + "eval_samples_per_second": 21.366, + "eval_steps_per_second": 1.346, + "step": 1290 + }, + { + "epoch": 1.8683068017366136, + "grad_norm": 1.2789667844772339, + "learning_rate": 4.3845534995977477e-07, + "loss": 0.7941, + "step": 1291 + }, + { + "epoch": 1.869753979739508, + "grad_norm": 1.3443622589111328, + "learning_rate": 4.344328238133548e-07, + "loss": 0.8075, + "step": 1292 + }, + { + "epoch": 1.8712011577424024, + "grad_norm": 1.3462733030319214, + "learning_rate": 4.3041029766693486e-07, + "loss": 0.8145, + "step": 1293 + }, + { + "epoch": 1.8726483357452968, + "grad_norm": 1.2373638153076172, + "learning_rate": 4.263877715205149e-07, + "loss": 0.7966, + "step": 1294 + }, + { + "epoch": 1.8740955137481912, + "grad_norm": 1.4029980897903442, + "learning_rate": 4.22365245374095e-07, + "loss": 0.8458, + "step": 1295 + }, + { + "epoch": 1.8755426917510853, + "grad_norm": 1.4634947776794434, + "learning_rate": 4.1834271922767503e-07, + "loss": 0.8247, + "step": 1296 + }, + { + "epoch": 1.8769898697539797, + "grad_norm": 1.2962074279785156, + "learning_rate": 4.1432019308125506e-07, + "loss": 0.8268, + "step": 1297 + }, + { + "epoch": 1.8784370477568741, + "grad_norm": 1.3578174114227295, + "learning_rate": 4.1029766693483513e-07, + "loss": 0.7959, + "step": 1298 + }, + { + "epoch": 1.8798842257597683, + "grad_norm": 1.3372273445129395, + "learning_rate": 4.0627514078841515e-07, + "loss": 0.8189, + "step": 1299 + }, + { + "epoch": 1.8813314037626627, + "grad_norm": 1.3099868297576904, + "learning_rate": 4.0225261464199523e-07, + "loss": 0.8227, + "step": 1300 + }, + { + "epoch": 1.8813314037626627, + "eval_loss": 0.9517223834991455, + "eval_runtime": 46.7156, + "eval_samples_per_second": 21.406, + "eval_steps_per_second": 1.349, + "step": 1300 + }, + { + "epoch": 1.8827785817655571, + "grad_norm": 1.242724895477295, + "learning_rate": 3.9823008849557525e-07, + "loss": 0.825, + "step": 1301 + }, + { + "epoch": 1.8842257597684515, + "grad_norm": 1.348751425743103, + "learning_rate": 3.9420756234915527e-07, + "loss": 0.8355, + "step": 1302 + }, + { + "epoch": 1.885672937771346, + "grad_norm": 1.3841049671173096, + "learning_rate": 3.9018503620273535e-07, + "loss": 0.8144, + "step": 1303 + }, + { + "epoch": 1.8871201157742403, + "grad_norm": 1.2702233791351318, + "learning_rate": 3.8616251005631537e-07, + "loss": 0.8004, + "step": 1304 + }, + { + "epoch": 1.8885672937771347, + "grad_norm": 1.2769310474395752, + "learning_rate": 3.821399839098954e-07, + "loss": 0.8009, + "step": 1305 + }, + { + "epoch": 1.890014471780029, + "grad_norm": 1.26711905002594, + "learning_rate": 3.781174577634755e-07, + "loss": 0.7912, + "step": 1306 + }, + { + "epoch": 1.8914616497829233, + "grad_norm": 1.2899887561798096, + "learning_rate": 3.740949316170556e-07, + "loss": 0.8151, + "step": 1307 + }, + { + "epoch": 1.8929088277858177, + "grad_norm": 1.2439205646514893, + "learning_rate": 3.700724054706356e-07, + "loss": 0.818, + "step": 1308 + }, + { + "epoch": 1.8943560057887119, + "grad_norm": 1.4273436069488525, + "learning_rate": 3.6604987932421563e-07, + "loss": 0.8261, + "step": 1309 + }, + { + "epoch": 1.8958031837916063, + "grad_norm": 1.219878077507019, + "learning_rate": 3.620273531777957e-07, + "loss": 0.8147, + "step": 1310 + }, + { + "epoch": 1.8958031837916063, + "eval_loss": 0.9509444832801819, + "eval_runtime": 46.6295, + "eval_samples_per_second": 21.446, + "eval_steps_per_second": 1.351, + "step": 1310 + }, + { + "epoch": 1.8972503617945007, + "grad_norm": 1.390021562576294, + "learning_rate": 3.5800482703137573e-07, + "loss": 0.8111, + "step": 1311 + }, + { + "epoch": 1.898697539797395, + "grad_norm": 1.3138254880905151, + "learning_rate": 3.5398230088495575e-07, + "loss": 0.7996, + "step": 1312 + }, + { + "epoch": 1.9001447178002895, + "grad_norm": 1.3018814325332642, + "learning_rate": 3.4995977473853583e-07, + "loss": 0.8291, + "step": 1313 + }, + { + "epoch": 1.9015918958031839, + "grad_norm": 1.4169855117797852, + "learning_rate": 3.4593724859211585e-07, + "loss": 0.8228, + "step": 1314 + }, + { + "epoch": 1.9030390738060783, + "grad_norm": 1.35712468624115, + "learning_rate": 3.419147224456959e-07, + "loss": 0.8285, + "step": 1315 + }, + { + "epoch": 1.9044862518089725, + "grad_norm": 1.3731104135513306, + "learning_rate": 3.3789219629927595e-07, + "loss": 0.8033, + "step": 1316 + }, + { + "epoch": 1.9059334298118669, + "grad_norm": 1.321549654006958, + "learning_rate": 3.338696701528561e-07, + "loss": 0.795, + "step": 1317 + }, + { + "epoch": 1.9073806078147613, + "grad_norm": 1.2659696340560913, + "learning_rate": 3.298471440064361e-07, + "loss": 0.8367, + "step": 1318 + }, + { + "epoch": 1.9088277858176554, + "grad_norm": 1.4103540182113647, + "learning_rate": 3.258246178600161e-07, + "loss": 0.8028, + "step": 1319 + }, + { + "epoch": 1.9102749638205498, + "grad_norm": 1.2264856100082397, + "learning_rate": 3.218020917135962e-07, + "loss": 0.798, + "step": 1320 + }, + { + "epoch": 1.9102749638205498, + "eval_loss": 0.9507466554641724, + "eval_runtime": 46.7481, + "eval_samples_per_second": 21.391, + "eval_steps_per_second": 1.348, + "step": 1320 + }, + { + "epoch": 1.9117221418234442, + "grad_norm": 1.3266470432281494, + "learning_rate": 3.177795655671762e-07, + "loss": 0.8336, + "step": 1321 + }, + { + "epoch": 1.9131693198263386, + "grad_norm": 1.2796952724456787, + "learning_rate": 3.137570394207563e-07, + "loss": 0.8345, + "step": 1322 + }, + { + "epoch": 1.914616497829233, + "grad_norm": 1.326030969619751, + "learning_rate": 3.097345132743363e-07, + "loss": 0.8423, + "step": 1323 + }, + { + "epoch": 1.9160636758321274, + "grad_norm": 1.2464720010757446, + "learning_rate": 3.0571198712791633e-07, + "loss": 0.8, + "step": 1324 + }, + { + "epoch": 1.9175108538350218, + "grad_norm": 1.3275500535964966, + "learning_rate": 3.016894609814964e-07, + "loss": 0.8162, + "step": 1325 + }, + { + "epoch": 1.918958031837916, + "grad_norm": 1.3022739887237549, + "learning_rate": 2.976669348350765e-07, + "loss": 0.8104, + "step": 1326 + }, + { + "epoch": 1.9204052098408104, + "grad_norm": 1.3301210403442383, + "learning_rate": 2.936444086886565e-07, + "loss": 0.8568, + "step": 1327 + }, + { + "epoch": 1.9218523878437048, + "grad_norm": 1.3180088996887207, + "learning_rate": 2.896218825422365e-07, + "loss": 0.7998, + "step": 1328 + }, + { + "epoch": 1.923299565846599, + "grad_norm": 1.2535759210586548, + "learning_rate": 2.855993563958166e-07, + "loss": 0.8115, + "step": 1329 + }, + { + "epoch": 1.9247467438494934, + "grad_norm": 1.3544580936431885, + "learning_rate": 2.815768302493966e-07, + "loss": 0.8204, + "step": 1330 + }, + { + "epoch": 1.9247467438494934, + "eval_loss": 0.9503014087677002, + "eval_runtime": 46.7246, + "eval_samples_per_second": 21.402, + "eval_steps_per_second": 1.348, + "step": 1330 + }, + { + "epoch": 1.9261939218523878, + "grad_norm": 1.331663966178894, + "learning_rate": 2.775543041029767e-07, + "loss": 0.8274, + "step": 1331 + }, + { + "epoch": 1.9276410998552822, + "grad_norm": 1.2822785377502441, + "learning_rate": 2.7353177795655677e-07, + "loss": 0.7886, + "step": 1332 + }, + { + "epoch": 1.9290882778581766, + "grad_norm": 1.3937132358551025, + "learning_rate": 2.695092518101368e-07, + "loss": 0.8033, + "step": 1333 + }, + { + "epoch": 1.930535455861071, + "grad_norm": 1.3891572952270508, + "learning_rate": 2.654867256637168e-07, + "loss": 0.8475, + "step": 1334 + }, + { + "epoch": 1.9319826338639654, + "grad_norm": 1.2993863821029663, + "learning_rate": 2.614641995172969e-07, + "loss": 0.804, + "step": 1335 + }, + { + "epoch": 1.9334298118668596, + "grad_norm": 1.326062798500061, + "learning_rate": 2.574416733708769e-07, + "loss": 0.7921, + "step": 1336 + }, + { + "epoch": 1.934876989869754, + "grad_norm": 1.2884583473205566, + "learning_rate": 2.53419147224457e-07, + "loss": 0.8197, + "step": 1337 + }, + { + "epoch": 1.9363241678726484, + "grad_norm": 1.3081976175308228, + "learning_rate": 2.49396621078037e-07, + "loss": 0.806, + "step": 1338 + }, + { + "epoch": 1.9377713458755426, + "grad_norm": 1.3629621267318726, + "learning_rate": 2.453740949316171e-07, + "loss": 0.8323, + "step": 1339 + }, + { + "epoch": 1.939218523878437, + "grad_norm": 1.3372217416763306, + "learning_rate": 2.413515687851971e-07, + "loss": 0.7967, + "step": 1340 + }, + { + "epoch": 1.939218523878437, + "eval_loss": 0.951086163520813, + "eval_runtime": 46.7232, + "eval_samples_per_second": 21.403, + "eval_steps_per_second": 1.348, + "step": 1340 + }, + { + "epoch": 1.9406657018813314, + "grad_norm": 1.2203961610794067, + "learning_rate": 2.3732904263877715e-07, + "loss": 0.8105, + "step": 1341 + }, + { + "epoch": 1.9421128798842258, + "grad_norm": 1.3470215797424316, + "learning_rate": 2.3330651649235723e-07, + "loss": 0.7911, + "step": 1342 + }, + { + "epoch": 1.9435600578871202, + "grad_norm": 1.3332464694976807, + "learning_rate": 2.2928399034593728e-07, + "loss": 0.8222, + "step": 1343 + }, + { + "epoch": 1.9450072358900146, + "grad_norm": 1.37167227268219, + "learning_rate": 2.2526146419951732e-07, + "loss": 0.8363, + "step": 1344 + }, + { + "epoch": 1.946454413892909, + "grad_norm": 1.4836008548736572, + "learning_rate": 2.2123893805309737e-07, + "loss": 0.8388, + "step": 1345 + }, + { + "epoch": 1.9479015918958031, + "grad_norm": 1.3758822679519653, + "learning_rate": 2.172164119066774e-07, + "loss": 0.8572, + "step": 1346 + }, + { + "epoch": 1.9493487698986975, + "grad_norm": 1.3247509002685547, + "learning_rate": 2.1319388576025744e-07, + "loss": 0.7994, + "step": 1347 + }, + { + "epoch": 1.950795947901592, + "grad_norm": 1.3479751348495483, + "learning_rate": 2.0917135961383752e-07, + "loss": 0.8151, + "step": 1348 + }, + { + "epoch": 1.9522431259044861, + "grad_norm": 1.2421225309371948, + "learning_rate": 2.0514883346741757e-07, + "loss": 0.7977, + "step": 1349 + }, + { + "epoch": 1.9536903039073805, + "grad_norm": 1.319460391998291, + "learning_rate": 2.0112630732099761e-07, + "loss": 0.841, + "step": 1350 + }, + { + "epoch": 1.9536903039073805, + "eval_loss": 0.9502463340759277, + "eval_runtime": 46.7042, + "eval_samples_per_second": 21.411, + "eval_steps_per_second": 1.349, + "step": 1350 + }, + { + "epoch": 1.955137481910275, + "grad_norm": 1.3078151941299438, + "learning_rate": 1.9710378117457764e-07, + "loss": 0.7984, + "step": 1351 + }, + { + "epoch": 1.9565846599131693, + "grad_norm": 1.2529306411743164, + "learning_rate": 1.9308125502815768e-07, + "loss": 0.8135, + "step": 1352 + }, + { + "epoch": 1.9580318379160637, + "grad_norm": 1.3414435386657715, + "learning_rate": 1.8905872888173776e-07, + "loss": 0.8328, + "step": 1353 + }, + { + "epoch": 1.9594790159189581, + "grad_norm": 1.2409676313400269, + "learning_rate": 1.850362027353178e-07, + "loss": 0.8379, + "step": 1354 + }, + { + "epoch": 1.9609261939218525, + "grad_norm": 1.3834282159805298, + "learning_rate": 1.8101367658889785e-07, + "loss": 0.8306, + "step": 1355 + }, + { + "epoch": 1.9623733719247467, + "grad_norm": 1.2632218599319458, + "learning_rate": 1.7699115044247788e-07, + "loss": 0.7922, + "step": 1356 + }, + { + "epoch": 1.963820549927641, + "grad_norm": 1.3124845027923584, + "learning_rate": 1.7296862429605792e-07, + "loss": 0.8162, + "step": 1357 + }, + { + "epoch": 1.9652677279305355, + "grad_norm": 1.3145887851715088, + "learning_rate": 1.6894609814963797e-07, + "loss": 0.7895, + "step": 1358 + }, + { + "epoch": 1.9667149059334297, + "grad_norm": 1.218945860862732, + "learning_rate": 1.6492357200321805e-07, + "loss": 0.805, + "step": 1359 + }, + { + "epoch": 1.968162083936324, + "grad_norm": 1.4872159957885742, + "learning_rate": 1.609010458567981e-07, + "loss": 0.8431, + "step": 1360 + }, + { + "epoch": 1.968162083936324, + "eval_loss": 0.9496836066246033, + "eval_runtime": 46.324, + "eval_samples_per_second": 21.587, + "eval_steps_per_second": 1.36, + "step": 1360 + }, + { + "epoch": 1.9696092619392185, + "grad_norm": 1.2907038927078247, + "learning_rate": 1.5687851971037814e-07, + "loss": 0.8007, + "step": 1361 + }, + { + "epoch": 1.9710564399421129, + "grad_norm": 1.3102412223815918, + "learning_rate": 1.5285599356395817e-07, + "loss": 0.8277, + "step": 1362 + }, + { + "epoch": 1.9725036179450073, + "grad_norm": 1.232476830482483, + "learning_rate": 1.4883346741753824e-07, + "loss": 0.8088, + "step": 1363 + }, + { + "epoch": 1.9739507959479017, + "grad_norm": 1.3577868938446045, + "learning_rate": 1.4481094127111826e-07, + "loss": 0.8358, + "step": 1364 + }, + { + "epoch": 1.975397973950796, + "grad_norm": 1.16902494430542, + "learning_rate": 1.407884151246983e-07, + "loss": 0.8082, + "step": 1365 + }, + { + "epoch": 1.9768451519536903, + "grad_norm": 1.3158315420150757, + "learning_rate": 1.3676588897827839e-07, + "loss": 0.8341, + "step": 1366 + }, + { + "epoch": 1.9782923299565847, + "grad_norm": 1.3739079236984253, + "learning_rate": 1.327433628318584e-07, + "loss": 0.8473, + "step": 1367 + }, + { + "epoch": 1.979739507959479, + "grad_norm": 1.3810678720474243, + "learning_rate": 1.2872083668543846e-07, + "loss": 0.7826, + "step": 1368 + }, + { + "epoch": 1.9811866859623732, + "grad_norm": 1.263561487197876, + "learning_rate": 1.246983105390185e-07, + "loss": 0.8185, + "step": 1369 + }, + { + "epoch": 1.9826338639652676, + "grad_norm": 1.3035837411880493, + "learning_rate": 1.2067578439259855e-07, + "loss": 0.8606, + "step": 1370 + }, + { + "epoch": 1.9826338639652676, + "eval_loss": 0.9501672387123108, + "eval_runtime": 46.4265, + "eval_samples_per_second": 21.539, + "eval_steps_per_second": 1.357, + "step": 1370 + }, + { + "epoch": 1.984081041968162, + "grad_norm": 1.4977774620056152, + "learning_rate": 1.1665325824617861e-07, + "loss": 0.8031, + "step": 1371 + }, + { + "epoch": 1.9855282199710564, + "grad_norm": 1.3124608993530273, + "learning_rate": 1.1263073209975866e-07, + "loss": 0.8127, + "step": 1372 + }, + { + "epoch": 1.9869753979739508, + "grad_norm": 1.4336574077606201, + "learning_rate": 1.086082059533387e-07, + "loss": 0.831, + "step": 1373 + }, + { + "epoch": 1.9884225759768452, + "grad_norm": 1.419453740119934, + "learning_rate": 1.0458567980691876e-07, + "loss": 0.8403, + "step": 1374 + }, + { + "epoch": 1.9898697539797396, + "grad_norm": 1.4168434143066406, + "learning_rate": 1.0056315366049881e-07, + "loss": 0.7944, + "step": 1375 + }, + { + "epoch": 1.9913169319826338, + "grad_norm": 1.2436405420303345, + "learning_rate": 9.654062751407884e-08, + "loss": 0.7984, + "step": 1376 + }, + { + "epoch": 1.9927641099855282, + "grad_norm": 1.2651194334030151, + "learning_rate": 9.25181013676589e-08, + "loss": 0.8164, + "step": 1377 + }, + { + "epoch": 1.9942112879884226, + "grad_norm": 1.2977267503738403, + "learning_rate": 8.849557522123894e-08, + "loss": 0.8259, + "step": 1378 + }, + { + "epoch": 1.9956584659913168, + "grad_norm": 1.2483677864074707, + "learning_rate": 8.447304907481899e-08, + "loss": 0.8038, + "step": 1379 + }, + { + "epoch": 1.9971056439942112, + "grad_norm": 1.225170612335205, + "learning_rate": 8.045052292839905e-08, + "loss": 0.8055, + "step": 1380 + }, + { + "epoch": 1.9971056439942112, + "eval_loss": 0.9500590562820435, + "eval_runtime": 46.4424, + "eval_samples_per_second": 21.532, + "eval_steps_per_second": 1.357, + "step": 1380 + }, + { + "epoch": 1.9985528219971056, + "grad_norm": 1.3751300573349, + "learning_rate": 7.642799678197908e-08, + "loss": 0.8315, + "step": 1381 + }, + { + "epoch": 2.0, + "grad_norm": 1.2663606405258179, + "learning_rate": 7.240547063555913e-08, + "loss": 0.7802, + "step": 1382 + } + ], + "logging_steps": 1, + "max_steps": 1382, + "num_input_tokens_seen": 0, + "num_train_epochs": 2, + "save_steps": 50, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 1.4408803727903293e+18, + "train_batch_size": 6, + "trial_name": null, + "trial_params": null +}