diff --git "a/checkpoints/Qwen2.5-3B/babylm_reverse_control_10M_seed0/runs/checkpoint-1382/trainer_state.json" "b/checkpoints/Qwen2.5-3B/babylm_reverse_control_10M_seed0/runs/checkpoint-1382/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoints/Qwen2.5-3B/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": 1.5826, + "step": 1 + }, + { + "epoch": 0.002894356005788712, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6119, + "step": 2 + }, + { + "epoch": 0.004341534008683068, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.62, + "step": 3 + }, + { + "epoch": 0.005788712011577424, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.604, + "step": 4 + }, + { + "epoch": 0.00723589001447178, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6207, + "step": 5 + }, + { + "epoch": 0.008683068017366137, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6348, + "step": 6 + }, + { + "epoch": 0.010130246020260492, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.627, + "step": 7 + }, + { + "epoch": 0.011577424023154847, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6195, + "step": 8 + }, + { + "epoch": 0.013024602026049204, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6251, + "step": 9 + }, + { + "epoch": 0.01447178002894356, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6017, + "step": 10 + }, + { + "epoch": 0.01447178002894356, + "eval_loss": 1.627951741218567, + "eval_runtime": 24.3872, + "eval_samples_per_second": 41.005, + "eval_steps_per_second": 2.583, + "step": 10 + }, + { + "epoch": 0.015918958031837915, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6406, + "step": 11 + }, + { + "epoch": 0.017366136034732273, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6387, + "step": 12 + }, + { + "epoch": 0.01881331403762663, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6292, + "step": 13 + }, + { + "epoch": 0.020260492040520984, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6097, + "step": 14 + }, + { + "epoch": 0.02170767004341534, + "grad_norm": 2.144015073776245, + "learning_rate": 3.597122302158274e-08, + "loss": 1.6483, + "step": 15 + }, + { + "epoch": 0.023154848046309694, + "grad_norm": 2.063476085662842, + "learning_rate": 7.194244604316547e-08, + "loss": 1.6203, + "step": 16 + }, + { + "epoch": 0.024602026049204053, + "grad_norm": 2.117913007736206, + "learning_rate": 1.0791366906474822e-07, + "loss": 1.6195, + "step": 17 + }, + { + "epoch": 0.02604920405209841, + "grad_norm": 2.244119167327881, + "learning_rate": 1.4388489208633095e-07, + "loss": 1.6481, + "step": 18 + }, + { + "epoch": 0.027496382054992764, + "grad_norm": 2.3340413570404053, + "learning_rate": 1.7985611510791368e-07, + "loss": 1.6166, + "step": 19 + }, + { + "epoch": 0.02894356005788712, + "grad_norm": 2.1988165378570557, + "learning_rate": 2.1582733812949643e-07, + "loss": 1.6399, + "step": 20 + }, + { + "epoch": 0.02894356005788712, + "eval_loss": 1.6270402669906616, + "eval_runtime": 24.6946, + "eval_samples_per_second": 40.495, + "eval_steps_per_second": 2.551, + "step": 20 + }, + { + "epoch": 0.030390738060781478, + "grad_norm": 2.1431221961975098, + "learning_rate": 2.5179856115107916e-07, + "loss": 1.6301, + "step": 21 + }, + { + "epoch": 0.03183791606367583, + "grad_norm": 2.168792247772217, + "learning_rate": 2.877697841726619e-07, + "loss": 1.5952, + "step": 22 + }, + { + "epoch": 0.03328509406657019, + "grad_norm": 2.015110731124878, + "learning_rate": 3.237410071942446e-07, + "loss": 1.6061, + "step": 23 + }, + { + "epoch": 0.03473227206946455, + "grad_norm": 2.050848960876465, + "learning_rate": 3.5971223021582736e-07, + "loss": 1.6472, + "step": 24 + }, + { + "epoch": 0.0361794500723589, + "grad_norm": 2.036330223083496, + "learning_rate": 3.956834532374101e-07, + "loss": 1.6324, + "step": 25 + }, + { + "epoch": 0.03762662807525326, + "grad_norm": 1.9470363855361938, + "learning_rate": 4.3165467625899287e-07, + "loss": 1.5941, + "step": 26 + }, + { + "epoch": 0.03907380607814761, + "grad_norm": 1.6043027639389038, + "learning_rate": 4.676258992805756e-07, + "loss": 1.6025, + "step": 27 + }, + { + "epoch": 0.04052098408104197, + "grad_norm": 1.7497000694274902, + "learning_rate": 5.035971223021583e-07, + "loss": 1.6381, + "step": 28 + }, + { + "epoch": 0.041968162083936326, + "grad_norm": 1.436023473739624, + "learning_rate": 5.39568345323741e-07, + "loss": 1.5767, + "step": 29 + }, + { + "epoch": 0.04341534008683068, + "grad_norm": 1.6225666999816895, + "learning_rate": 5.755395683453238e-07, + "loss": 1.5999, + "step": 30 + }, + { + "epoch": 0.04341534008683068, + "eval_loss": 1.600980520248413, + "eval_runtime": 25.1149, + "eval_samples_per_second": 39.817, + "eval_steps_per_second": 2.508, + "step": 30 + }, + { + "epoch": 0.04486251808972504, + "grad_norm": 1.3526175022125244, + "learning_rate": 6.115107913669066e-07, + "loss": 1.5752, + "step": 31 + }, + { + "epoch": 0.04630969609261939, + "grad_norm": 1.2586437463760376, + "learning_rate": 6.474820143884893e-07, + "loss": 1.6016, + "step": 32 + }, + { + "epoch": 0.04775687409551375, + "grad_norm": 1.216167688369751, + "learning_rate": 6.83453237410072e-07, + "loss": 1.6009, + "step": 33 + }, + { + "epoch": 0.049204052098408106, + "grad_norm": 1.1464942693710327, + "learning_rate": 7.194244604316547e-07, + "loss": 1.5922, + "step": 34 + }, + { + "epoch": 0.05065123010130246, + "grad_norm": 1.191906213760376, + "learning_rate": 7.553956834532375e-07, + "loss": 1.5621, + "step": 35 + }, + { + "epoch": 0.05209840810419682, + "grad_norm": 1.065446138381958, + "learning_rate": 7.913669064748202e-07, + "loss": 1.5611, + "step": 36 + }, + { + "epoch": 0.053545586107091175, + "grad_norm": 0.9210651516914368, + "learning_rate": 8.27338129496403e-07, + "loss": 1.592, + "step": 37 + }, + { + "epoch": 0.05499276410998553, + "grad_norm": 1.0145928859710693, + "learning_rate": 8.633093525179857e-07, + "loss": 1.551, + "step": 38 + }, + { + "epoch": 0.056439942112879886, + "grad_norm": 0.8538812398910522, + "learning_rate": 8.992805755395684e-07, + "loss": 1.564, + "step": 39 + }, + { + "epoch": 0.05788712011577424, + "grad_norm": 0.935691773891449, + "learning_rate": 9.352517985611512e-07, + "loss": 1.5504, + "step": 40 + }, + { + "epoch": 0.05788712011577424, + "eval_loss": 1.5676774978637695, + "eval_runtime": 25.5185, + "eval_samples_per_second": 39.187, + "eval_steps_per_second": 2.469, + "step": 40 + }, + { + "epoch": 0.059334298118668596, + "grad_norm": 0.9620401263237, + "learning_rate": 9.71223021582734e-07, + "loss": 1.5504, + "step": 41 + }, + { + "epoch": 0.060781476121562955, + "grad_norm": 0.9131060838699341, + "learning_rate": 1.0071942446043167e-06, + "loss": 1.5696, + "step": 42 + }, + { + "epoch": 0.06222865412445731, + "grad_norm": 0.875487744808197, + "learning_rate": 1.0431654676258993e-06, + "loss": 1.5916, + "step": 43 + }, + { + "epoch": 0.06367583212735166, + "grad_norm": 0.8755046725273132, + "learning_rate": 1.079136690647482e-06, + "loss": 1.5706, + "step": 44 + }, + { + "epoch": 0.06512301013024602, + "grad_norm": 0.8441756367683411, + "learning_rate": 1.115107913669065e-06, + "loss": 1.5616, + "step": 45 + }, + { + "epoch": 0.06657018813314038, + "grad_norm": 0.7749464511871338, + "learning_rate": 1.1510791366906476e-06, + "loss": 1.5414, + "step": 46 + }, + { + "epoch": 0.06801736613603473, + "grad_norm": 0.8531357645988464, + "learning_rate": 1.1870503597122303e-06, + "loss": 1.5627, + "step": 47 + }, + { + "epoch": 0.0694645441389291, + "grad_norm": 0.8716596364974976, + "learning_rate": 1.2230215827338131e-06, + "loss": 1.5464, + "step": 48 + }, + { + "epoch": 0.07091172214182344, + "grad_norm": 0.8619104027748108, + "learning_rate": 1.2589928057553958e-06, + "loss": 1.5329, + "step": 49 + }, + { + "epoch": 0.0723589001447178, + "grad_norm": 0.8491005897521973, + "learning_rate": 1.2949640287769785e-06, + "loss": 1.5075, + "step": 50 + }, + { + "epoch": 0.0723589001447178, + "eval_loss": 1.5282820463180542, + "eval_runtime": 25.6731, + "eval_samples_per_second": 38.951, + "eval_steps_per_second": 2.454, + "step": 50 + }, + { + "epoch": 0.07380607814761216, + "grad_norm": 0.8045928478240967, + "learning_rate": 1.3309352517985614e-06, + "loss": 1.4949, + "step": 51 + }, + { + "epoch": 0.07525325615050651, + "grad_norm": 0.9498971700668335, + "learning_rate": 1.366906474820144e-06, + "loss": 1.5126, + "step": 52 + }, + { + "epoch": 0.07670043415340087, + "grad_norm": 0.9595310091972351, + "learning_rate": 1.4028776978417265e-06, + "loss": 1.4954, + "step": 53 + }, + { + "epoch": 0.07814761215629522, + "grad_norm": 1.2015273571014404, + "learning_rate": 1.4388489208633094e-06, + "loss": 1.4952, + "step": 54 + }, + { + "epoch": 0.07959479015918958, + "grad_norm": 1.2412959337234497, + "learning_rate": 1.474820143884892e-06, + "loss": 1.5102, + "step": 55 + }, + { + "epoch": 0.08104196816208394, + "grad_norm": 1.0004088878631592, + "learning_rate": 1.510791366906475e-06, + "loss": 1.4709, + "step": 56 + }, + { + "epoch": 0.0824891461649783, + "grad_norm": 1.1461646556854248, + "learning_rate": 1.5467625899280579e-06, + "loss": 1.4214, + "step": 57 + }, + { + "epoch": 0.08393632416787265, + "grad_norm": 1.2485071420669556, + "learning_rate": 1.5827338129496403e-06, + "loss": 1.4698, + "step": 58 + }, + { + "epoch": 0.085383502170767, + "grad_norm": 1.2286319732666016, + "learning_rate": 1.618705035971223e-06, + "loss": 1.4802, + "step": 59 + }, + { + "epoch": 0.08683068017366136, + "grad_norm": 1.243710994720459, + "learning_rate": 1.654676258992806e-06, + "loss": 1.4204, + "step": 60 + }, + { + "epoch": 0.08683068017366136, + "eval_loss": 1.4558827877044678, + "eval_runtime": 25.629, + "eval_samples_per_second": 39.018, + "eval_steps_per_second": 2.458, + "step": 60 + }, + { + "epoch": 0.08827785817655572, + "grad_norm": 1.3729641437530518, + "learning_rate": 1.6906474820143886e-06, + "loss": 1.4566, + "step": 61 + }, + { + "epoch": 0.08972503617945007, + "grad_norm": 1.332112193107605, + "learning_rate": 1.7266187050359715e-06, + "loss": 1.453, + "step": 62 + }, + { + "epoch": 0.09117221418234443, + "grad_norm": 1.2128121852874756, + "learning_rate": 1.762589928057554e-06, + "loss": 1.4244, + "step": 63 + }, + { + "epoch": 0.09261939218523878, + "grad_norm": 1.3846666812896729, + "learning_rate": 1.7985611510791368e-06, + "loss": 1.4078, + "step": 64 + }, + { + "epoch": 0.09406657018813314, + "grad_norm": 1.2263684272766113, + "learning_rate": 1.8345323741007195e-06, + "loss": 1.3985, + "step": 65 + }, + { + "epoch": 0.0955137481910275, + "grad_norm": 1.3221999406814575, + "learning_rate": 1.8705035971223024e-06, + "loss": 1.3883, + "step": 66 + }, + { + "epoch": 0.09696092619392185, + "grad_norm": 1.3177974224090576, + "learning_rate": 1.906474820143885e-06, + "loss": 1.3672, + "step": 67 + }, + { + "epoch": 0.09840810419681621, + "grad_norm": 1.226685881614685, + "learning_rate": 1.942446043165468e-06, + "loss": 1.3625, + "step": 68 + }, + { + "epoch": 0.09985528219971057, + "grad_norm": 1.2232762575149536, + "learning_rate": 1.9784172661870504e-06, + "loss": 1.3844, + "step": 69 + }, + { + "epoch": 0.10130246020260492, + "grad_norm": 1.254563808441162, + "learning_rate": 2.0143884892086333e-06, + "loss": 1.3607, + "step": 70 + }, + { + "epoch": 0.10130246020260492, + "eval_loss": 1.3712953329086304, + "eval_runtime": 25.9893, + "eval_samples_per_second": 38.477, + "eval_steps_per_second": 2.424, + "step": 70 + }, + { + "epoch": 0.10274963820549927, + "grad_norm": 1.3770047426223755, + "learning_rate": 2.050359712230216e-06, + "loss": 1.3771, + "step": 71 + }, + { + "epoch": 0.10419681620839363, + "grad_norm": 1.3217036724090576, + "learning_rate": 2.0863309352517987e-06, + "loss": 1.3172, + "step": 72 + }, + { + "epoch": 0.10564399421128799, + "grad_norm": 1.3304638862609863, + "learning_rate": 2.1223021582733816e-06, + "loss": 1.318, + "step": 73 + }, + { + "epoch": 0.10709117221418235, + "grad_norm": 1.216414213180542, + "learning_rate": 2.158273381294964e-06, + "loss": 1.3, + "step": 74 + }, + { + "epoch": 0.1085383502170767, + "grad_norm": 1.1022595167160034, + "learning_rate": 2.194244604316547e-06, + "loss": 1.2698, + "step": 75 + }, + { + "epoch": 0.10998552821997105, + "grad_norm": 1.01797616481781, + "learning_rate": 2.23021582733813e-06, + "loss": 1.3161, + "step": 76 + }, + { + "epoch": 0.11143270622286541, + "grad_norm": 1.2778011560440063, + "learning_rate": 2.2661870503597123e-06, + "loss": 1.3176, + "step": 77 + }, + { + "epoch": 0.11287988422575977, + "grad_norm": 1.1276257038116455, + "learning_rate": 2.302158273381295e-06, + "loss": 1.3164, + "step": 78 + }, + { + "epoch": 0.11432706222865413, + "grad_norm": 1.0312668085098267, + "learning_rate": 2.3381294964028776e-06, + "loss": 1.3201, + "step": 79 + }, + { + "epoch": 0.11577424023154848, + "grad_norm": 1.1506502628326416, + "learning_rate": 2.3741007194244605e-06, + "loss": 1.2898, + "step": 80 + }, + { + "epoch": 0.11577424023154848, + "eval_loss": 1.3133100271224976, + "eval_runtime": 25.9424, + "eval_samples_per_second": 38.547, + "eval_steps_per_second": 2.428, + "step": 80 + }, + { + "epoch": 0.11722141823444283, + "grad_norm": 1.255002737045288, + "learning_rate": 2.4100719424460434e-06, + "loss": 1.247, + "step": 81 + }, + { + "epoch": 0.11866859623733719, + "grad_norm": 1.0065932273864746, + "learning_rate": 2.4460431654676263e-06, + "loss": 1.2602, + "step": 82 + }, + { + "epoch": 0.12011577424023155, + "grad_norm": 1.1586326360702515, + "learning_rate": 2.4820143884892088e-06, + "loss": 1.2704, + "step": 83 + }, + { + "epoch": 0.12156295224312591, + "grad_norm": 1.0571246147155762, + "learning_rate": 2.5179856115107916e-06, + "loss": 1.2633, + "step": 84 + }, + { + "epoch": 0.12301013024602026, + "grad_norm": 1.0377482175827026, + "learning_rate": 2.5539568345323745e-06, + "loss": 1.2238, + "step": 85 + }, + { + "epoch": 0.12445730824891461, + "grad_norm": 1.1468099355697632, + "learning_rate": 2.589928057553957e-06, + "loss": 1.2596, + "step": 86 + }, + { + "epoch": 0.12590448625180897, + "grad_norm": 1.0852339267730713, + "learning_rate": 2.6258992805755395e-06, + "loss": 1.2081, + "step": 87 + }, + { + "epoch": 0.12735166425470332, + "grad_norm": 1.1638356447219849, + "learning_rate": 2.6618705035971228e-06, + "loss": 1.2345, + "step": 88 + }, + { + "epoch": 0.1287988422575977, + "grad_norm": 1.0700938701629639, + "learning_rate": 2.6978417266187052e-06, + "loss": 1.257, + "step": 89 + }, + { + "epoch": 0.13024602026049203, + "grad_norm": 1.0385199785232544, + "learning_rate": 2.733812949640288e-06, + "loss": 1.2391, + "step": 90 + }, + { + "epoch": 0.13024602026049203, + "eval_loss": 1.2747834920883179, + "eval_runtime": 25.8114, + "eval_samples_per_second": 38.743, + "eval_steps_per_second": 2.441, + "step": 90 + }, + { + "epoch": 0.1316931982633864, + "grad_norm": 1.0480972528457642, + "learning_rate": 2.7697841726618706e-06, + "loss": 1.251, + "step": 91 + }, + { + "epoch": 0.13314037626628075, + "grad_norm": 1.0574753284454346, + "learning_rate": 2.805755395683453e-06, + "loss": 1.2553, + "step": 92 + }, + { + "epoch": 0.1345875542691751, + "grad_norm": 1.0661845207214355, + "learning_rate": 2.8417266187050364e-06, + "loss": 1.2123, + "step": 93 + }, + { + "epoch": 0.13603473227206947, + "grad_norm": 1.0629788637161255, + "learning_rate": 2.877697841726619e-06, + "loss": 1.2224, + "step": 94 + }, + { + "epoch": 0.13748191027496381, + "grad_norm": 1.0361292362213135, + "learning_rate": 2.9136690647482017e-06, + "loss": 1.2173, + "step": 95 + }, + { + "epoch": 0.1389290882778582, + "grad_norm": 0.9286133050918579, + "learning_rate": 2.949640287769784e-06, + "loss": 1.1927, + "step": 96 + }, + { + "epoch": 0.14037626628075253, + "grad_norm": 1.048107624053955, + "learning_rate": 2.985611510791367e-06, + "loss": 1.2088, + "step": 97 + }, + { + "epoch": 0.14182344428364688, + "grad_norm": 1.0357203483581543, + "learning_rate": 3.02158273381295e-06, + "loss": 1.1913, + "step": 98 + }, + { + "epoch": 0.14327062228654125, + "grad_norm": 0.9728686213493347, + "learning_rate": 3.0575539568345324e-06, + "loss": 1.2347, + "step": 99 + }, + { + "epoch": 0.1447178002894356, + "grad_norm": 1.0187407732009888, + "learning_rate": 3.0935251798561158e-06, + "loss": 1.2401, + "step": 100 + }, + { + "epoch": 0.1447178002894356, + "eval_loss": 1.246801495552063, + "eval_runtime": 26.0552, + "eval_samples_per_second": 38.38, + "eval_steps_per_second": 2.418, + "step": 100 + }, + { + "epoch": 0.14616497829232997, + "grad_norm": 1.071919322013855, + "learning_rate": 3.1294964028776982e-06, + "loss": 1.1819, + "step": 101 + }, + { + "epoch": 0.1476121562952243, + "grad_norm": 1.1320074796676636, + "learning_rate": 3.1654676258992807e-06, + "loss": 1.1978, + "step": 102 + }, + { + "epoch": 0.14905933429811866, + "grad_norm": 1.1210949420928955, + "learning_rate": 3.2014388489208636e-06, + "loss": 1.158, + "step": 103 + }, + { + "epoch": 0.15050651230101303, + "grad_norm": 1.0182701349258423, + "learning_rate": 3.237410071942446e-06, + "loss": 1.1825, + "step": 104 + }, + { + "epoch": 0.15195369030390737, + "grad_norm": 1.1875406503677368, + "learning_rate": 3.2733812949640294e-06, + "loss": 1.1763, + "step": 105 + }, + { + "epoch": 0.15340086830680175, + "grad_norm": 1.08661687374115, + "learning_rate": 3.309352517985612e-06, + "loss": 1.163, + "step": 106 + }, + { + "epoch": 0.1548480463096961, + "grad_norm": 1.0423978567123413, + "learning_rate": 3.3453237410071943e-06, + "loss": 1.1796, + "step": 107 + }, + { + "epoch": 0.15629522431259044, + "grad_norm": 1.0998430252075195, + "learning_rate": 3.381294964028777e-06, + "loss": 1.1686, + "step": 108 + }, + { + "epoch": 0.1577424023154848, + "grad_norm": 1.279999852180481, + "learning_rate": 3.4172661870503596e-06, + "loss": 1.1859, + "step": 109 + }, + { + "epoch": 0.15918958031837915, + "grad_norm": 1.2889167070388794, + "learning_rate": 3.453237410071943e-06, + "loss": 1.1262, + "step": 110 + }, + { + "epoch": 0.15918958031837915, + "eval_loss": 1.2298810482025146, + "eval_runtime": 25.854, + "eval_samples_per_second": 38.679, + "eval_steps_per_second": 2.437, + "step": 110 + }, + { + "epoch": 0.16063675832127353, + "grad_norm": 1.3285552263259888, + "learning_rate": 3.4892086330935254e-06, + "loss": 1.2033, + "step": 111 + }, + { + "epoch": 0.16208393632416787, + "grad_norm": 1.6612026691436768, + "learning_rate": 3.525179856115108e-06, + "loss": 1.1981, + "step": 112 + }, + { + "epoch": 0.16353111432706222, + "grad_norm": 1.6106890439987183, + "learning_rate": 3.561151079136691e-06, + "loss": 1.1716, + "step": 113 + }, + { + "epoch": 0.1649782923299566, + "grad_norm": 1.4226655960083008, + "learning_rate": 3.5971223021582737e-06, + "loss": 1.1666, + "step": 114 + }, + { + "epoch": 0.16642547033285093, + "grad_norm": 1.284498929977417, + "learning_rate": 3.6330935251798566e-06, + "loss": 1.1461, + "step": 115 + }, + { + "epoch": 0.1678726483357453, + "grad_norm": 1.3292125463485718, + "learning_rate": 3.669064748201439e-06, + "loss": 1.1256, + "step": 116 + }, + { + "epoch": 0.16931982633863965, + "grad_norm": 1.376024842262268, + "learning_rate": 3.7050359712230215e-06, + "loss": 1.1779, + "step": 117 + }, + { + "epoch": 0.170767004341534, + "grad_norm": 1.2837554216384888, + "learning_rate": 3.741007194244605e-06, + "loss": 1.1377, + "step": 118 + }, + { + "epoch": 0.17221418234442837, + "grad_norm": 1.2247073650360107, + "learning_rate": 3.7769784172661873e-06, + "loss": 1.1634, + "step": 119 + }, + { + "epoch": 0.1736613603473227, + "grad_norm": 1.1542692184448242, + "learning_rate": 3.81294964028777e-06, + "loss": 1.1496, + "step": 120 + }, + { + "epoch": 0.1736613603473227, + "eval_loss": 1.2104259729385376, + "eval_runtime": 25.8178, + "eval_samples_per_second": 38.733, + "eval_steps_per_second": 2.44, + "step": 120 + }, + { + "epoch": 0.17510853835021709, + "grad_norm": 1.136985421180725, + "learning_rate": 3.848920863309353e-06, + "loss": 1.1459, + "step": 121 + }, + { + "epoch": 0.17655571635311143, + "grad_norm": 1.451317548751831, + "learning_rate": 3.884892086330936e-06, + "loss": 1.1443, + "step": 122 + }, + { + "epoch": 0.17800289435600578, + "grad_norm": 1.2180333137512207, + "learning_rate": 3.920863309352518e-06, + "loss": 1.1108, + "step": 123 + }, + { + "epoch": 0.17945007235890015, + "grad_norm": 1.244494080543518, + "learning_rate": 3.956834532374101e-06, + "loss": 1.1416, + "step": 124 + }, + { + "epoch": 0.1808972503617945, + "grad_norm": 1.394103765487671, + "learning_rate": 3.992805755395684e-06, + "loss": 1.1612, + "step": 125 + }, + { + "epoch": 0.18234442836468887, + "grad_norm": 1.4224308729171753, + "learning_rate": 4.028776978417267e-06, + "loss": 1.1541, + "step": 126 + }, + { + "epoch": 0.1837916063675832, + "grad_norm": 1.1226342916488647, + "learning_rate": 4.0647482014388495e-06, + "loss": 1.158, + "step": 127 + }, + { + "epoch": 0.18523878437047755, + "grad_norm": 1.6892023086547852, + "learning_rate": 4.100719424460432e-06, + "loss": 1.1361, + "step": 128 + }, + { + "epoch": 0.18668596237337193, + "grad_norm": 1.2189652919769287, + "learning_rate": 4.1366906474820145e-06, + "loss": 1.1349, + "step": 129 + }, + { + "epoch": 0.18813314037626627, + "grad_norm": 1.3171944618225098, + "learning_rate": 4.172661870503597e-06, + "loss": 1.1279, + "step": 130 + }, + { + "epoch": 0.18813314037626627, + "eval_loss": 1.1947468519210815, + "eval_runtime": 25.8646, + "eval_samples_per_second": 38.663, + "eval_steps_per_second": 2.436, + "step": 130 + }, + { + "epoch": 0.18958031837916064, + "grad_norm": 1.3438791036605835, + "learning_rate": 4.20863309352518e-06, + "loss": 1.1665, + "step": 131 + }, + { + "epoch": 0.191027496382055, + "grad_norm": 1.237427830696106, + "learning_rate": 4.244604316546763e-06, + "loss": 1.1249, + "step": 132 + }, + { + "epoch": 0.19247467438494936, + "grad_norm": 1.2919330596923828, + "learning_rate": 4.280575539568346e-06, + "loss": 1.165, + "step": 133 + }, + { + "epoch": 0.1939218523878437, + "grad_norm": 1.0939977169036865, + "learning_rate": 4.316546762589928e-06, + "loss": 1.1565, + "step": 134 + }, + { + "epoch": 0.19536903039073805, + "grad_norm": 1.280110478401184, + "learning_rate": 4.352517985611511e-06, + "loss": 1.1298, + "step": 135 + }, + { + "epoch": 0.19681620839363242, + "grad_norm": 1.3320883512496948, + "learning_rate": 4.388489208633094e-06, + "loss": 1.1175, + "step": 136 + }, + { + "epoch": 0.19826338639652677, + "grad_norm": 1.280161738395691, + "learning_rate": 4.424460431654677e-06, + "loss": 1.1015, + "step": 137 + }, + { + "epoch": 0.19971056439942114, + "grad_norm": 1.1698580980300903, + "learning_rate": 4.46043165467626e-06, + "loss": 1.0857, + "step": 138 + }, + { + "epoch": 0.2011577424023155, + "grad_norm": 1.2695364952087402, + "learning_rate": 4.496402877697842e-06, + "loss": 1.1271, + "step": 139 + }, + { + "epoch": 0.20260492040520983, + "grad_norm": 1.2971086502075195, + "learning_rate": 4.5323741007194245e-06, + "loss": 1.1204, + "step": 140 + }, + { + "epoch": 0.20260492040520983, + "eval_loss": 1.1829454898834229, + "eval_runtime": 25.6947, + "eval_samples_per_second": 38.918, + "eval_steps_per_second": 2.452, + "step": 140 + }, + { + "epoch": 0.2040520984081042, + "grad_norm": 1.4637281894683838, + "learning_rate": 4.5683453237410074e-06, + "loss": 1.099, + "step": 141 + }, + { + "epoch": 0.20549927641099855, + "grad_norm": 1.3764904737472534, + "learning_rate": 4.60431654676259e-06, + "loss": 1.1115, + "step": 142 + }, + { + "epoch": 0.20694645441389292, + "grad_norm": 1.2485584020614624, + "learning_rate": 4.640287769784173e-06, + "loss": 1.1036, + "step": 143 + }, + { + "epoch": 0.20839363241678727, + "grad_norm": 1.5968283414840698, + "learning_rate": 4.676258992805755e-06, + "loss": 1.1008, + "step": 144 + }, + { + "epoch": 0.2098408104196816, + "grad_norm": 1.275233268737793, + "learning_rate": 4.712230215827339e-06, + "loss": 1.1326, + "step": 145 + }, + { + "epoch": 0.21128798842257598, + "grad_norm": 1.4760442972183228, + "learning_rate": 4.748201438848921e-06, + "loss": 1.1176, + "step": 146 + }, + { + "epoch": 0.21273516642547033, + "grad_norm": 1.381948709487915, + "learning_rate": 4.784172661870504e-06, + "loss": 1.0668, + "step": 147 + }, + { + "epoch": 0.2141823444283647, + "grad_norm": 1.8259927034378052, + "learning_rate": 4.820143884892087e-06, + "loss": 1.1167, + "step": 148 + }, + { + "epoch": 0.21562952243125905, + "grad_norm": 1.4848363399505615, + "learning_rate": 4.856115107913669e-06, + "loss": 1.1014, + "step": 149 + }, + { + "epoch": 0.2170767004341534, + "grad_norm": 1.5530890226364136, + "learning_rate": 4.892086330935253e-06, + "loss": 1.1171, + "step": 150 + }, + { + "epoch": 0.2170767004341534, + "eval_loss": 1.1686047315597534, + "eval_runtime": 25.7032, + "eval_samples_per_second": 38.906, + "eval_steps_per_second": 2.451, + "step": 150 + }, + { + "epoch": 0.21852387843704776, + "grad_norm": 1.445328712463379, + "learning_rate": 4.928057553956835e-06, + "loss": 1.112, + "step": 151 + }, + { + "epoch": 0.2199710564399421, + "grad_norm": 1.6887681484222412, + "learning_rate": 4.9640287769784175e-06, + "loss": 1.0563, + "step": 152 + }, + { + "epoch": 0.22141823444283648, + "grad_norm": 1.165201187133789, + "learning_rate": 5e-06, + "loss": 1.0794, + "step": 153 + }, + { + "epoch": 0.22286541244573083, + "grad_norm": 1.548592448234558, + "learning_rate": 4.9959774738535805e-06, + "loss": 1.0889, + "step": 154 + }, + { + "epoch": 0.22431259044862517, + "grad_norm": 1.2718911170959473, + "learning_rate": 4.991954947707161e-06, + "loss": 1.0748, + "step": 155 + }, + { + "epoch": 0.22575976845151954, + "grad_norm": 1.3692930936813354, + "learning_rate": 4.987932421560741e-06, + "loss": 1.103, + "step": 156 + }, + { + "epoch": 0.2272069464544139, + "grad_norm": 1.3592978715896606, + "learning_rate": 4.983909895414321e-06, + "loss": 1.1079, + "step": 157 + }, + { + "epoch": 0.22865412445730826, + "grad_norm": 1.5015676021575928, + "learning_rate": 4.9798873692679e-06, + "loss": 1.0584, + "step": 158 + }, + { + "epoch": 0.2301013024602026, + "grad_norm": 1.4109690189361572, + "learning_rate": 4.97586484312148e-06, + "loss": 1.0821, + "step": 159 + }, + { + "epoch": 0.23154848046309695, + "grad_norm": 1.3715922832489014, + "learning_rate": 4.9718423169750605e-06, + "loss": 1.1053, + "step": 160 + }, + { + "epoch": 0.23154848046309695, + "eval_loss": 1.1580647230148315, + "eval_runtime": 25.6393, + "eval_samples_per_second": 39.003, + "eval_steps_per_second": 2.457, + "step": 160 + }, + { + "epoch": 0.23299565846599132, + "grad_norm": 1.1541366577148438, + "learning_rate": 4.967819790828641e-06, + "loss": 1.0857, + "step": 161 + }, + { + "epoch": 0.23444283646888567, + "grad_norm": 1.5763776302337646, + "learning_rate": 4.963797264682221e-06, + "loss": 1.0539, + "step": 162 + }, + { + "epoch": 0.23589001447178004, + "grad_norm": 1.3444937467575073, + "learning_rate": 4.959774738535801e-06, + "loss": 1.0772, + "step": 163 + }, + { + "epoch": 0.23733719247467439, + "grad_norm": 1.4197810888290405, + "learning_rate": 4.955752212389381e-06, + "loss": 1.0827, + "step": 164 + }, + { + "epoch": 0.23878437047756873, + "grad_norm": 1.1938902139663696, + "learning_rate": 4.951729686242961e-06, + "loss": 1.034, + "step": 165 + }, + { + "epoch": 0.2402315484804631, + "grad_norm": 1.2534770965576172, + "learning_rate": 4.947707160096541e-06, + "loss": 1.0898, + "step": 166 + }, + { + "epoch": 0.24167872648335745, + "grad_norm": 1.3372267484664917, + "learning_rate": 4.943684633950121e-06, + "loss": 1.0581, + "step": 167 + }, + { + "epoch": 0.24312590448625182, + "grad_norm": 1.3008803129196167, + "learning_rate": 4.939662107803701e-06, + "loss": 1.085, + "step": 168 + }, + { + "epoch": 0.24457308248914617, + "grad_norm": 1.2555029392242432, + "learning_rate": 4.935639581657281e-06, + "loss": 1.0546, + "step": 169 + }, + { + "epoch": 0.2460202604920405, + "grad_norm": 1.3190017938613892, + "learning_rate": 4.931617055510861e-06, + "loss": 1.0895, + "step": 170 + }, + { + "epoch": 0.2460202604920405, + "eval_loss": 1.1438969373703003, + "eval_runtime": 25.6063, + "eval_samples_per_second": 39.053, + "eval_steps_per_second": 2.46, + "step": 170 + }, + { + "epoch": 0.24746743849493488, + "grad_norm": 1.3019165992736816, + "learning_rate": 4.927594529364441e-06, + "loss": 1.1047, + "step": 171 + }, + { + "epoch": 0.24891461649782923, + "grad_norm": 1.232087254524231, + "learning_rate": 4.923572003218021e-06, + "loss": 1.0703, + "step": 172 + }, + { + "epoch": 0.2503617945007236, + "grad_norm": 1.3068221807479858, + "learning_rate": 4.919549477071601e-06, + "loss": 1.0331, + "step": 173 + }, + { + "epoch": 0.25180897250361794, + "grad_norm": 1.521071434020996, + "learning_rate": 4.915526950925181e-06, + "loss": 1.0559, + "step": 174 + }, + { + "epoch": 0.2532561505065123, + "grad_norm": 1.411606788635254, + "learning_rate": 4.9115044247787615e-06, + "loss": 1.0705, + "step": 175 + }, + { + "epoch": 0.25470332850940663, + "grad_norm": 1.6073122024536133, + "learning_rate": 4.907481898632342e-06, + "loss": 1.0492, + "step": 176 + }, + { + "epoch": 0.25615050651230103, + "grad_norm": 1.4059184789657593, + "learning_rate": 4.903459372485922e-06, + "loss": 1.1006, + "step": 177 + }, + { + "epoch": 0.2575976845151954, + "grad_norm": 1.370344638824463, + "learning_rate": 4.899436846339501e-06, + "loss": 1.0533, + "step": 178 + }, + { + "epoch": 0.2590448625180897, + "grad_norm": 1.4646189212799072, + "learning_rate": 4.895414320193081e-06, + "loss": 1.0602, + "step": 179 + }, + { + "epoch": 0.26049204052098407, + "grad_norm": 1.5465567111968994, + "learning_rate": 4.891391794046661e-06, + "loss": 1.0773, + "step": 180 + }, + { + "epoch": 0.26049204052098407, + "eval_loss": 1.1373363733291626, + "eval_runtime": 25.632, + "eval_samples_per_second": 39.014, + "eval_steps_per_second": 2.458, + "step": 180 + }, + { + "epoch": 0.2619392185238784, + "grad_norm": 1.7964074611663818, + "learning_rate": 4.8873692679002414e-06, + "loss": 1.0827, + "step": 181 + }, + { + "epoch": 0.2633863965267728, + "grad_norm": 1.2767939567565918, + "learning_rate": 4.8833467417538216e-06, + "loss": 1.0588, + "step": 182 + }, + { + "epoch": 0.26483357452966716, + "grad_norm": 1.7200208902359009, + "learning_rate": 4.879324215607402e-06, + "loss": 1.069, + "step": 183 + }, + { + "epoch": 0.2662807525325615, + "grad_norm": 1.3738912343978882, + "learning_rate": 4.875301689460982e-06, + "loss": 1.0447, + "step": 184 + }, + { + "epoch": 0.26772793053545585, + "grad_norm": 1.4727184772491455, + "learning_rate": 4.871279163314562e-06, + "loss": 1.0738, + "step": 185 + }, + { + "epoch": 0.2691751085383502, + "grad_norm": 1.7510229349136353, + "learning_rate": 4.867256637168142e-06, + "loss": 1.0569, + "step": 186 + }, + { + "epoch": 0.2706222865412446, + "grad_norm": 1.5435395240783691, + "learning_rate": 4.863234111021722e-06, + "loss": 1.0365, + "step": 187 + }, + { + "epoch": 0.27206946454413894, + "grad_norm": 1.9311236143112183, + "learning_rate": 4.8592115848753015e-06, + "loss": 1.0527, + "step": 188 + }, + { + "epoch": 0.2735166425470333, + "grad_norm": 1.4670462608337402, + "learning_rate": 4.855189058728882e-06, + "loss": 1.0547, + "step": 189 + }, + { + "epoch": 0.27496382054992763, + "grad_norm": 1.773186445236206, + "learning_rate": 4.851166532582462e-06, + "loss": 1.0374, + "step": 190 + }, + { + "epoch": 0.27496382054992763, + "eval_loss": 1.1362701654434204, + "eval_runtime": 25.664, + "eval_samples_per_second": 38.965, + "eval_steps_per_second": 2.455, + "step": 190 + }, + { + "epoch": 0.276410998552822, + "grad_norm": 1.3181395530700684, + "learning_rate": 4.847144006436042e-06, + "loss": 1.0727, + "step": 191 + }, + { + "epoch": 0.2778581765557164, + "grad_norm": 1.30997633934021, + "learning_rate": 4.843121480289622e-06, + "loss": 1.0038, + "step": 192 + }, + { + "epoch": 0.2793053545586107, + "grad_norm": 1.6861798763275146, + "learning_rate": 4.839098954143202e-06, + "loss": 1.0219, + "step": 193 + }, + { + "epoch": 0.28075253256150506, + "grad_norm": 1.3235957622528076, + "learning_rate": 4.835076427996782e-06, + "loss": 1.0406, + "step": 194 + }, + { + "epoch": 0.2821997105643994, + "grad_norm": 1.4384372234344482, + "learning_rate": 4.831053901850362e-06, + "loss": 1.0828, + "step": 195 + }, + { + "epoch": 0.28364688856729375, + "grad_norm": 1.4204976558685303, + "learning_rate": 4.8270313757039425e-06, + "loss": 1.0542, + "step": 196 + }, + { + "epoch": 0.28509406657018815, + "grad_norm": 1.207572340965271, + "learning_rate": 4.823008849557523e-06, + "loss": 1.033, + "step": 197 + }, + { + "epoch": 0.2865412445730825, + "grad_norm": 1.4599734544754028, + "learning_rate": 4.818986323411103e-06, + "loss": 1.0303, + "step": 198 + }, + { + "epoch": 0.28798842257597684, + "grad_norm": 1.5304739475250244, + "learning_rate": 4.814963797264683e-06, + "loss": 1.0334, + "step": 199 + }, + { + "epoch": 0.2894356005788712, + "grad_norm": 1.3051408529281616, + "learning_rate": 4.810941271118263e-06, + "loss": 1.0265, + "step": 200 + }, + { + "epoch": 0.2894356005788712, + "eval_loss": 1.1237152814865112, + "eval_runtime": 25.6565, + "eval_samples_per_second": 38.977, + "eval_steps_per_second": 2.456, + "step": 200 + }, + { + "epoch": 0.29088277858176553, + "grad_norm": 1.3950132131576538, + "learning_rate": 4.806918744971843e-06, + "loss": 1.0247, + "step": 201 + }, + { + "epoch": 0.29232995658465993, + "grad_norm": 1.3100475072860718, + "learning_rate": 4.802896218825423e-06, + "loss": 1.0257, + "step": 202 + }, + { + "epoch": 0.2937771345875543, + "grad_norm": 1.1980586051940918, + "learning_rate": 4.798873692679003e-06, + "loss": 1.063, + "step": 203 + }, + { + "epoch": 0.2952243125904486, + "grad_norm": 1.4957659244537354, + "learning_rate": 4.794851166532583e-06, + "loss": 1.1039, + "step": 204 + }, + { + "epoch": 0.29667149059334297, + "grad_norm": 1.406442403793335, + "learning_rate": 4.790828640386163e-06, + "loss": 1.0844, + "step": 205 + }, + { + "epoch": 0.2981186685962373, + "grad_norm": 1.2532386779785156, + "learning_rate": 4.786806114239743e-06, + "loss": 1.0421, + "step": 206 + }, + { + "epoch": 0.2995658465991317, + "grad_norm": 1.311007022857666, + "learning_rate": 4.782783588093323e-06, + "loss": 1.0657, + "step": 207 + }, + { + "epoch": 0.30101302460202606, + "grad_norm": 1.427161693572998, + "learning_rate": 4.778761061946903e-06, + "loss": 1.0112, + "step": 208 + }, + { + "epoch": 0.3024602026049204, + "grad_norm": 1.6766445636749268, + "learning_rate": 4.774738535800483e-06, + "loss": 1.0514, + "step": 209 + }, + { + "epoch": 0.30390738060781475, + "grad_norm": 1.422170639038086, + "learning_rate": 4.7707160096540635e-06, + "loss": 1.0221, + "step": 210 + }, + { + "epoch": 0.30390738060781475, + "eval_loss": 1.1181718111038208, + "eval_runtime": 25.6967, + "eval_samples_per_second": 38.916, + "eval_steps_per_second": 2.452, + "step": 210 + }, + { + "epoch": 0.3053545586107091, + "grad_norm": 1.2965795993804932, + "learning_rate": 4.766693483507644e-06, + "loss": 1.0388, + "step": 211 + }, + { + "epoch": 0.3068017366136035, + "grad_norm": 1.3259484767913818, + "learning_rate": 4.762670957361224e-06, + "loss": 1.0565, + "step": 212 + }, + { + "epoch": 0.30824891461649784, + "grad_norm": 1.2306101322174072, + "learning_rate": 4.758648431214804e-06, + "loss": 1.0349, + "step": 213 + }, + { + "epoch": 0.3096960926193922, + "grad_norm": 1.4872595071792603, + "learning_rate": 4.754625905068383e-06, + "loss": 1.0754, + "step": 214 + }, + { + "epoch": 0.3111432706222865, + "grad_norm": 1.1391340494155884, + "learning_rate": 4.750603378921963e-06, + "loss": 1.0491, + "step": 215 + }, + { + "epoch": 0.3125904486251809, + "grad_norm": 1.254473328590393, + "learning_rate": 4.746580852775543e-06, + "loss": 1.0425, + "step": 216 + }, + { + "epoch": 0.3140376266280753, + "grad_norm": 1.3624285459518433, + "learning_rate": 4.7425583266291235e-06, + "loss": 1.0578, + "step": 217 + }, + { + "epoch": 0.3154848046309696, + "grad_norm": 1.3315480947494507, + "learning_rate": 4.738535800482704e-06, + "loss": 1.0402, + "step": 218 + }, + { + "epoch": 0.31693198263386396, + "grad_norm": 1.4668720960617065, + "learning_rate": 4.734513274336284e-06, + "loss": 1.0421, + "step": 219 + }, + { + "epoch": 0.3183791606367583, + "grad_norm": 1.246150255203247, + "learning_rate": 4.730490748189864e-06, + "loss": 1.0138, + "step": 220 + }, + { + "epoch": 0.3183791606367583, + "eval_loss": 1.1130249500274658, + "eval_runtime": 25.9195, + "eval_samples_per_second": 38.581, + "eval_steps_per_second": 2.431, + "step": 220 + }, + { + "epoch": 0.31982633863965265, + "grad_norm": 1.2191036939620972, + "learning_rate": 4.726468222043444e-06, + "loss": 1.0245, + "step": 221 + }, + { + "epoch": 0.32127351664254705, + "grad_norm": 1.3822863101959229, + "learning_rate": 4.722445695897024e-06, + "loss": 1.0409, + "step": 222 + }, + { + "epoch": 0.3227206946454414, + "grad_norm": 1.2923561334609985, + "learning_rate": 4.718423169750604e-06, + "loss": 1.0555, + "step": 223 + }, + { + "epoch": 0.32416787264833574, + "grad_norm": 1.3411794900894165, + "learning_rate": 4.7144006436041835e-06, + "loss": 0.9884, + "step": 224 + }, + { + "epoch": 0.3256150506512301, + "grad_norm": 1.4061198234558105, + "learning_rate": 4.710378117457764e-06, + "loss": 0.9896, + "step": 225 + }, + { + "epoch": 0.32706222865412443, + "grad_norm": 1.361370325088501, + "learning_rate": 4.706355591311344e-06, + "loss": 1.015, + "step": 226 + }, + { + "epoch": 0.32850940665701883, + "grad_norm": 1.2427265644073486, + "learning_rate": 4.702333065164924e-06, + "loss": 1.0235, + "step": 227 + }, + { + "epoch": 0.3299565846599132, + "grad_norm": 1.3974968194961548, + "learning_rate": 4.698310539018504e-06, + "loss": 1.0351, + "step": 228 + }, + { + "epoch": 0.3314037626628075, + "grad_norm": 1.4032317399978638, + "learning_rate": 4.694288012872084e-06, + "loss": 1.0012, + "step": 229 + }, + { + "epoch": 0.33285094066570187, + "grad_norm": 1.3513522148132324, + "learning_rate": 4.690265486725664e-06, + "loss": 1.043, + "step": 230 + }, + { + "epoch": 0.33285094066570187, + "eval_loss": 1.1030467748641968, + "eval_runtime": 25.7003, + "eval_samples_per_second": 38.91, + "eval_steps_per_second": 2.451, + "step": 230 + }, + { + "epoch": 0.3342981186685962, + "grad_norm": 1.430236577987671, + "learning_rate": 4.6862429605792444e-06, + "loss": 1.0229, + "step": 231 + }, + { + "epoch": 0.3357452966714906, + "grad_norm": 1.3285548686981201, + "learning_rate": 4.6822204344328246e-06, + "loss": 1.0177, + "step": 232 + }, + { + "epoch": 0.33719247467438496, + "grad_norm": 1.5301393270492554, + "learning_rate": 4.678197908286405e-06, + "loss": 1.0396, + "step": 233 + }, + { + "epoch": 0.3386396526772793, + "grad_norm": 1.2442700862884521, + "learning_rate": 4.674175382139984e-06, + "loss": 1.0346, + "step": 234 + }, + { + "epoch": 0.34008683068017365, + "grad_norm": 1.3574509620666504, + "learning_rate": 4.670152855993564e-06, + "loss": 1.0752, + "step": 235 + }, + { + "epoch": 0.341534008683068, + "grad_norm": 1.5170680284500122, + "learning_rate": 4.666130329847144e-06, + "loss": 1.0498, + "step": 236 + }, + { + "epoch": 0.3429811866859624, + "grad_norm": 1.5622068643569946, + "learning_rate": 4.662107803700724e-06, + "loss": 1.0397, + "step": 237 + }, + { + "epoch": 0.34442836468885674, + "grad_norm": 1.319341778755188, + "learning_rate": 4.6580852775543045e-06, + "loss": 1.0564, + "step": 238 + }, + { + "epoch": 0.3458755426917511, + "grad_norm": 1.6022669076919556, + "learning_rate": 4.654062751407885e-06, + "loss": 1.0339, + "step": 239 + }, + { + "epoch": 0.3473227206946454, + "grad_norm": 1.5898162126541138, + "learning_rate": 4.650040225261465e-06, + "loss": 1.0128, + "step": 240 + }, + { + "epoch": 0.3473227206946454, + "eval_loss": 1.0971758365631104, + "eval_runtime": 25.8681, + "eval_samples_per_second": 38.658, + "eval_steps_per_second": 2.435, + "step": 240 + }, + { + "epoch": 0.34876989869753977, + "grad_norm": 1.5174083709716797, + "learning_rate": 4.646017699115045e-06, + "loss": 1.0162, + "step": 241 + }, + { + "epoch": 0.35021707670043417, + "grad_norm": 1.569929599761963, + "learning_rate": 4.641995172968625e-06, + "loss": 0.9892, + "step": 242 + }, + { + "epoch": 0.3516642547033285, + "grad_norm": 1.3473092317581177, + "learning_rate": 4.637972646822205e-06, + "loss": 1.0343, + "step": 243 + }, + { + "epoch": 0.35311143270622286, + "grad_norm": 1.4948253631591797, + "learning_rate": 4.633950120675784e-06, + "loss": 1.0244, + "step": 244 + }, + { + "epoch": 0.3545586107091172, + "grad_norm": 1.4232202768325806, + "learning_rate": 4.6299275945293645e-06, + "loss": 1.0367, + "step": 245 + }, + { + "epoch": 0.35600578871201155, + "grad_norm": 1.5077722072601318, + "learning_rate": 4.625905068382945e-06, + "loss": 1.0208, + "step": 246 + }, + { + "epoch": 0.35745296671490595, + "grad_norm": 1.389813780784607, + "learning_rate": 4.621882542236525e-06, + "loss": 1.027, + "step": 247 + }, + { + "epoch": 0.3589001447178003, + "grad_norm": 1.3511253595352173, + "learning_rate": 4.617860016090105e-06, + "loss": 1.0247, + "step": 248 + }, + { + "epoch": 0.36034732272069464, + "grad_norm": 1.6211826801300049, + "learning_rate": 4.613837489943685e-06, + "loss": 1.0376, + "step": 249 + }, + { + "epoch": 0.361794500723589, + "grad_norm": 1.2799283266067505, + "learning_rate": 4.609814963797265e-06, + "loss": 0.9995, + "step": 250 + }, + { + "epoch": 0.361794500723589, + "eval_loss": 1.0912978649139404, + "eval_runtime": 25.8351, + "eval_samples_per_second": 38.707, + "eval_steps_per_second": 2.439, + "step": 250 + }, + { + "epoch": 0.36324167872648333, + "grad_norm": 1.6001930236816406, + "learning_rate": 4.605792437650845e-06, + "loss": 1.0318, + "step": 251 + }, + { + "epoch": 0.36468885672937773, + "grad_norm": 1.3637175559997559, + "learning_rate": 4.6017699115044254e-06, + "loss": 1.0027, + "step": 252 + }, + { + "epoch": 0.3661360347322721, + "grad_norm": 1.3025727272033691, + "learning_rate": 4.5977473853580056e-06, + "loss": 0.9971, + "step": 253 + }, + { + "epoch": 0.3675832127351664, + "grad_norm": 1.450335144996643, + "learning_rate": 4.593724859211585e-06, + "loss": 1.0201, + "step": 254 + }, + { + "epoch": 0.36903039073806077, + "grad_norm": 1.428523063659668, + "learning_rate": 4.589702333065165e-06, + "loss": 1.0261, + "step": 255 + }, + { + "epoch": 0.3704775687409551, + "grad_norm": 1.4156044721603394, + "learning_rate": 4.585679806918745e-06, + "loss": 0.9958, + "step": 256 + }, + { + "epoch": 0.3719247467438495, + "grad_norm": 1.3857672214508057, + "learning_rate": 4.581657280772325e-06, + "loss": 1.0098, + "step": 257 + }, + { + "epoch": 0.37337192474674386, + "grad_norm": 1.3765766620635986, + "learning_rate": 4.577634754625905e-06, + "loss": 1.0258, + "step": 258 + }, + { + "epoch": 0.3748191027496382, + "grad_norm": 1.3213120698928833, + "learning_rate": 4.5736122284794855e-06, + "loss": 1.0156, + "step": 259 + }, + { + "epoch": 0.37626628075253254, + "grad_norm": 1.3624275922775269, + "learning_rate": 4.569589702333066e-06, + "loss": 1.0266, + "step": 260 + }, + { + "epoch": 0.37626628075253254, + "eval_loss": 1.087217092514038, + "eval_runtime": 25.6623, + "eval_samples_per_second": 38.968, + "eval_steps_per_second": 2.455, + "step": 260 + }, + { + "epoch": 0.37771345875542695, + "grad_norm": 1.4796395301818848, + "learning_rate": 4.565567176186646e-06, + "loss": 0.969, + "step": 261 + }, + { + "epoch": 0.3791606367583213, + "grad_norm": 1.348137617111206, + "learning_rate": 4.561544650040226e-06, + "loss": 1.0073, + "step": 262 + }, + { + "epoch": 0.38060781476121563, + "grad_norm": 1.3586536645889282, + "learning_rate": 4.557522123893805e-06, + "loss": 1.0308, + "step": 263 + }, + { + "epoch": 0.38205499276411, + "grad_norm": 1.5610628128051758, + "learning_rate": 4.553499597747385e-06, + "loss": 1.0403, + "step": 264 + }, + { + "epoch": 0.3835021707670043, + "grad_norm": 1.4678453207015991, + "learning_rate": 4.549477071600965e-06, + "loss": 0.9937, + "step": 265 + }, + { + "epoch": 0.3849493487698987, + "grad_norm": 1.3599241971969604, + "learning_rate": 4.5454545454545455e-06, + "loss": 1.0112, + "step": 266 + }, + { + "epoch": 0.38639652677279307, + "grad_norm": 1.471745491027832, + "learning_rate": 4.541432019308126e-06, + "loss": 1.0145, + "step": 267 + }, + { + "epoch": 0.3878437047756874, + "grad_norm": 1.5481610298156738, + "learning_rate": 4.537409493161706e-06, + "loss": 1.0332, + "step": 268 + }, + { + "epoch": 0.38929088277858176, + "grad_norm": 1.475701928138733, + "learning_rate": 4.533386967015286e-06, + "loss": 1.0038, + "step": 269 + }, + { + "epoch": 0.3907380607814761, + "grad_norm": 1.6974904537200928, + "learning_rate": 4.529364440868866e-06, + "loss": 1.0033, + "step": 270 + }, + { + "epoch": 0.3907380607814761, + "eval_loss": 1.089996099472046, + "eval_runtime": 25.715, + "eval_samples_per_second": 38.888, + "eval_steps_per_second": 2.45, + "step": 270 + }, + { + "epoch": 0.3921852387843705, + "grad_norm": 1.5095347166061401, + "learning_rate": 4.525341914722446e-06, + "loss": 0.9633, + "step": 271 + }, + { + "epoch": 0.39363241678726485, + "grad_norm": 1.4550669193267822, + "learning_rate": 4.521319388576026e-06, + "loss": 1.065, + "step": 272 + }, + { + "epoch": 0.3950795947901592, + "grad_norm": 1.5253571271896362, + "learning_rate": 4.5172968624296056e-06, + "loss": 0.9701, + "step": 273 + }, + { + "epoch": 0.39652677279305354, + "grad_norm": 1.3699054718017578, + "learning_rate": 4.513274336283186e-06, + "loss": 1.0078, + "step": 274 + }, + { + "epoch": 0.3979739507959479, + "grad_norm": 1.3667832612991333, + "learning_rate": 4.509251810136766e-06, + "loss": 0.9924, + "step": 275 + }, + { + "epoch": 0.3994211287988423, + "grad_norm": 1.3792893886566162, + "learning_rate": 4.505229283990346e-06, + "loss": 1.011, + "step": 276 + }, + { + "epoch": 0.40086830680173663, + "grad_norm": 1.3925573825836182, + "learning_rate": 4.501206757843926e-06, + "loss": 1.0122, + "step": 277 + }, + { + "epoch": 0.402315484804631, + "grad_norm": 1.3556911945343018, + "learning_rate": 4.497184231697506e-06, + "loss": 0.9894, + "step": 278 + }, + { + "epoch": 0.4037626628075253, + "grad_norm": 1.3186146020889282, + "learning_rate": 4.493161705551086e-06, + "loss": 0.9996, + "step": 279 + }, + { + "epoch": 0.40520984081041966, + "grad_norm": 1.3015190362930298, + "learning_rate": 4.4891391794046665e-06, + "loss": 0.9923, + "step": 280 + }, + { + "epoch": 0.40520984081041966, + "eval_loss": 1.0799185037612915, + "eval_runtime": 25.8526, + "eval_samples_per_second": 38.681, + "eval_steps_per_second": 2.437, + "step": 280 + }, + { + "epoch": 0.40665701881331406, + "grad_norm": 1.5820448398590088, + "learning_rate": 4.485116653258247e-06, + "loss": 1.0292, + "step": 281 + }, + { + "epoch": 0.4081041968162084, + "grad_norm": 1.3347865343093872, + "learning_rate": 4.481094127111827e-06, + "loss": 1.0391, + "step": 282 + }, + { + "epoch": 0.40955137481910275, + "grad_norm": 1.5359543561935425, + "learning_rate": 4.477071600965406e-06, + "loss": 0.9842, + "step": 283 + }, + { + "epoch": 0.4109985528219971, + "grad_norm": 1.5072472095489502, + "learning_rate": 4.473049074818986e-06, + "loss": 1.0094, + "step": 284 + }, + { + "epoch": 0.41244573082489144, + "grad_norm": 1.3954992294311523, + "learning_rate": 4.469026548672566e-06, + "loss": 1.0169, + "step": 285 + }, + { + "epoch": 0.41389290882778584, + "grad_norm": 1.4730685949325562, + "learning_rate": 4.465004022526146e-06, + "loss": 0.981, + "step": 286 + }, + { + "epoch": 0.4153400868306802, + "grad_norm": 1.4737553596496582, + "learning_rate": 4.460981496379727e-06, + "loss": 0.9913, + "step": 287 + }, + { + "epoch": 0.41678726483357453, + "grad_norm": 1.3746957778930664, + "learning_rate": 4.4569589702333075e-06, + "loss": 0.9927, + "step": 288 + }, + { + "epoch": 0.4182344428364689, + "grad_norm": 1.3467106819152832, + "learning_rate": 4.452936444086888e-06, + "loss": 1.0173, + "step": 289 + }, + { + "epoch": 0.4196816208393632, + "grad_norm": 1.3768117427825928, + "learning_rate": 4.448913917940467e-06, + "loss": 1.0287, + "step": 290 + }, + { + "epoch": 0.4196816208393632, + "eval_loss": 1.0746866464614868, + "eval_runtime": 25.7053, + "eval_samples_per_second": 38.903, + "eval_steps_per_second": 2.451, + "step": 290 + }, + { + "epoch": 0.4211287988422576, + "grad_norm": 1.3297845125198364, + "learning_rate": 4.444891391794047e-06, + "loss": 0.9854, + "step": 291 + }, + { + "epoch": 0.42257597684515197, + "grad_norm": 1.3490155935287476, + "learning_rate": 4.440868865647627e-06, + "loss": 1.0071, + "step": 292 + }, + { + "epoch": 0.4240231548480463, + "grad_norm": 1.409170389175415, + "learning_rate": 4.436846339501207e-06, + "loss": 0.9944, + "step": 293 + }, + { + "epoch": 0.42547033285094066, + "grad_norm": 1.3009305000305176, + "learning_rate": 4.432823813354787e-06, + "loss": 1.0176, + "step": 294 + }, + { + "epoch": 0.426917510853835, + "grad_norm": 1.236854910850525, + "learning_rate": 4.4288012872083675e-06, + "loss": 0.9725, + "step": 295 + }, + { + "epoch": 0.4283646888567294, + "grad_norm": 1.3198227882385254, + "learning_rate": 4.424778761061948e-06, + "loss": 0.9878, + "step": 296 + }, + { + "epoch": 0.42981186685962375, + "grad_norm": 1.360911250114441, + "learning_rate": 4.420756234915528e-06, + "loss": 0.9698, + "step": 297 + }, + { + "epoch": 0.4312590448625181, + "grad_norm": 1.3463726043701172, + "learning_rate": 4.416733708769108e-06, + "loss": 0.9868, + "step": 298 + }, + { + "epoch": 0.43270622286541244, + "grad_norm": 1.409676432609558, + "learning_rate": 4.412711182622687e-06, + "loss": 1.0173, + "step": 299 + }, + { + "epoch": 0.4341534008683068, + "grad_norm": 1.3323174715042114, + "learning_rate": 4.408688656476267e-06, + "loss": 0.9568, + "step": 300 + }, + { + "epoch": 0.4341534008683068, + "eval_loss": 1.0745749473571777, + "eval_runtime": 25.7849, + "eval_samples_per_second": 38.782, + "eval_steps_per_second": 2.443, + "step": 300 + }, + { + "epoch": 0.4356005788712012, + "grad_norm": 1.4331218004226685, + "learning_rate": 4.4046661303298474e-06, + "loss": 0.9586, + "step": 301 + }, + { + "epoch": 0.4370477568740955, + "grad_norm": 1.5185577869415283, + "learning_rate": 4.4006436041834276e-06, + "loss": 1.0248, + "step": 302 + }, + { + "epoch": 0.4384949348769899, + "grad_norm": 1.7651474475860596, + "learning_rate": 4.396621078037008e-06, + "loss": 0.9775, + "step": 303 + }, + { + "epoch": 0.4399421128798842, + "grad_norm": 1.5057427883148193, + "learning_rate": 4.392598551890588e-06, + "loss": 0.9845, + "step": 304 + }, + { + "epoch": 0.44138929088277856, + "grad_norm": 1.4012314081192017, + "learning_rate": 4.388576025744168e-06, + "loss": 0.9677, + "step": 305 + }, + { + "epoch": 0.44283646888567296, + "grad_norm": 1.4347463846206665, + "learning_rate": 4.384553499597748e-06, + "loss": 0.9813, + "step": 306 + }, + { + "epoch": 0.4442836468885673, + "grad_norm": 1.3435741662979126, + "learning_rate": 4.380530973451328e-06, + "loss": 0.9734, + "step": 307 + }, + { + "epoch": 0.44573082489146165, + "grad_norm": 1.5648781061172485, + "learning_rate": 4.376508447304908e-06, + "loss": 1.0614, + "step": 308 + }, + { + "epoch": 0.447178002894356, + "grad_norm": 1.403222680091858, + "learning_rate": 4.372485921158488e-06, + "loss": 0.986, + "step": 309 + }, + { + "epoch": 0.44862518089725034, + "grad_norm": 1.4489191770553589, + "learning_rate": 4.368463395012068e-06, + "loss": 0.9709, + "step": 310 + }, + { + "epoch": 0.44862518089725034, + "eval_loss": 1.0693542957305908, + "eval_runtime": 25.7542, + "eval_samples_per_second": 38.829, + "eval_steps_per_second": 2.446, + "step": 310 + }, + { + "epoch": 0.45007235890014474, + "grad_norm": 1.5860564708709717, + "learning_rate": 4.364440868865648e-06, + "loss": 1.0255, + "step": 311 + }, + { + "epoch": 0.4515195369030391, + "grad_norm": 1.3616255521774292, + "learning_rate": 4.360418342719228e-06, + "loss": 0.9487, + "step": 312 + }, + { + "epoch": 0.45296671490593343, + "grad_norm": 1.4219884872436523, + "learning_rate": 4.356395816572808e-06, + "loss": 0.9859, + "step": 313 + }, + { + "epoch": 0.4544138929088278, + "grad_norm": 1.3668162822723389, + "learning_rate": 4.352373290426388e-06, + "loss": 0.9575, + "step": 314 + }, + { + "epoch": 0.4558610709117221, + "grad_norm": 1.4213571548461914, + "learning_rate": 4.348350764279968e-06, + "loss": 1.0129, + "step": 315 + }, + { + "epoch": 0.4573082489146165, + "grad_norm": 1.430222988128662, + "learning_rate": 4.3443282381335485e-06, + "loss": 0.9554, + "step": 316 + }, + { + "epoch": 0.45875542691751087, + "grad_norm": 1.3727890253067017, + "learning_rate": 4.340305711987129e-06, + "loss": 1.0396, + "step": 317 + }, + { + "epoch": 0.4602026049204052, + "grad_norm": 1.4125868082046509, + "learning_rate": 4.336283185840709e-06, + "loss": 0.9571, + "step": 318 + }, + { + "epoch": 0.46164978292329956, + "grad_norm": 1.4137847423553467, + "learning_rate": 4.332260659694288e-06, + "loss": 0.9992, + "step": 319 + }, + { + "epoch": 0.4630969609261939, + "grad_norm": 1.435394525527954, + "learning_rate": 4.328238133547868e-06, + "loss": 0.998, + "step": 320 + }, + { + "epoch": 0.4630969609261939, + "eval_loss": 1.0627951622009277, + "eval_runtime": 25.7658, + "eval_samples_per_second": 38.811, + "eval_steps_per_second": 2.445, + "step": 320 + }, + { + "epoch": 0.4645441389290883, + "grad_norm": 1.2891018390655518, + "learning_rate": 4.324215607401448e-06, + "loss": 0.9908, + "step": 321 + }, + { + "epoch": 0.46599131693198265, + "grad_norm": 1.3445963859558105, + "learning_rate": 4.3201930812550284e-06, + "loss": 0.9724, + "step": 322 + }, + { + "epoch": 0.467438494934877, + "grad_norm": 1.4441720247268677, + "learning_rate": 4.3161705551086086e-06, + "loss": 1.0035, + "step": 323 + }, + { + "epoch": 0.46888567293777134, + "grad_norm": 1.3721234798431396, + "learning_rate": 4.312148028962189e-06, + "loss": 0.9945, + "step": 324 + }, + { + "epoch": 0.4703328509406657, + "grad_norm": 1.387130856513977, + "learning_rate": 4.308125502815769e-06, + "loss": 0.9808, + "step": 325 + }, + { + "epoch": 0.4717800289435601, + "grad_norm": 1.467307686805725, + "learning_rate": 4.304102976669349e-06, + "loss": 0.9651, + "step": 326 + }, + { + "epoch": 0.4732272069464544, + "grad_norm": 1.3222352266311646, + "learning_rate": 4.300080450522929e-06, + "loss": 0.9618, + "step": 327 + }, + { + "epoch": 0.47467438494934877, + "grad_norm": 1.5600520372390747, + "learning_rate": 4.296057924376509e-06, + "loss": 1.0447, + "step": 328 + }, + { + "epoch": 0.4761215629522431, + "grad_norm": 1.3152457475662231, + "learning_rate": 4.2920353982300885e-06, + "loss": 1.0105, + "step": 329 + }, + { + "epoch": 0.47756874095513746, + "grad_norm": 1.3301295042037964, + "learning_rate": 4.288012872083669e-06, + "loss": 0.9997, + "step": 330 + }, + { + "epoch": 0.47756874095513746, + "eval_loss": 1.0656896829605103, + "eval_runtime": 25.7854, + "eval_samples_per_second": 38.782, + "eval_steps_per_second": 2.443, + "step": 330 + }, + { + "epoch": 0.47901591895803186, + "grad_norm": 1.537719964981079, + "learning_rate": 4.283990345937249e-06, + "loss": 0.9521, + "step": 331 + }, + { + "epoch": 0.4804630969609262, + "grad_norm": 1.3024014234542847, + "learning_rate": 4.279967819790829e-06, + "loss": 0.9855, + "step": 332 + }, + { + "epoch": 0.48191027496382055, + "grad_norm": 1.5578094720840454, + "learning_rate": 4.275945293644409e-06, + "loss": 0.9954, + "step": 333 + }, + { + "epoch": 0.4833574529667149, + "grad_norm": 1.4333091974258423, + "learning_rate": 4.271922767497989e-06, + "loss": 0.9975, + "step": 334 + }, + { + "epoch": 0.48480463096960924, + "grad_norm": 1.3617147207260132, + "learning_rate": 4.267900241351569e-06, + "loss": 0.9761, + "step": 335 + }, + { + "epoch": 0.48625180897250364, + "grad_norm": 1.4757050275802612, + "learning_rate": 4.263877715205149e-06, + "loss": 0.9852, + "step": 336 + }, + { + "epoch": 0.487698986975398, + "grad_norm": 1.4386905431747437, + "learning_rate": 4.2598551890587295e-06, + "loss": 0.983, + "step": 337 + }, + { + "epoch": 0.48914616497829233, + "grad_norm": 1.528428316116333, + "learning_rate": 4.25583266291231e-06, + "loss": 0.9801, + "step": 338 + }, + { + "epoch": 0.4905933429811867, + "grad_norm": 1.5482014417648315, + "learning_rate": 4.251810136765889e-06, + "loss": 0.9839, + "step": 339 + }, + { + "epoch": 0.492040520984081, + "grad_norm": 1.2495687007904053, + "learning_rate": 4.247787610619469e-06, + "loss": 0.9476, + "step": 340 + }, + { + "epoch": 0.492040520984081, + "eval_loss": 1.0619398355484009, + "eval_runtime": 25.8553, + "eval_samples_per_second": 38.677, + "eval_steps_per_second": 2.437, + "step": 340 + }, + { + "epoch": 0.4934876989869754, + "grad_norm": 1.3904533386230469, + "learning_rate": 4.243765084473049e-06, + "loss": 0.9935, + "step": 341 + }, + { + "epoch": 0.49493487698986977, + "grad_norm": 1.4647581577301025, + "learning_rate": 4.239742558326629e-06, + "loss": 0.9955, + "step": 342 + }, + { + "epoch": 0.4963820549927641, + "grad_norm": 1.3266704082489014, + "learning_rate": 4.235720032180209e-06, + "loss": 0.9304, + "step": 343 + }, + { + "epoch": 0.49782923299565845, + "grad_norm": 1.3789790868759155, + "learning_rate": 4.2316975060337895e-06, + "loss": 0.9834, + "step": 344 + }, + { + "epoch": 0.4992764109985528, + "grad_norm": 1.2976341247558594, + "learning_rate": 4.22767497988737e-06, + "loss": 0.9766, + "step": 345 + }, + { + "epoch": 0.5007235890014472, + "grad_norm": 1.3830623626708984, + "learning_rate": 4.22365245374095e-06, + "loss": 0.9684, + "step": 346 + }, + { + "epoch": 0.5021707670043415, + "grad_norm": 1.3655767440795898, + "learning_rate": 4.21962992759453e-06, + "loss": 0.9484, + "step": 347 + }, + { + "epoch": 0.5036179450072359, + "grad_norm": 1.4754517078399658, + "learning_rate": 4.21560740144811e-06, + "loss": 0.9792, + "step": 348 + }, + { + "epoch": 0.5050651230101303, + "grad_norm": 1.3112934827804565, + "learning_rate": 4.211584875301689e-06, + "loss": 0.9588, + "step": 349 + }, + { + "epoch": 0.5065123010130246, + "grad_norm": 1.4535300731658936, + "learning_rate": 4.2075623491552695e-06, + "loss": 0.9573, + "step": 350 + }, + { + "epoch": 0.5065123010130246, + "eval_loss": 1.0596165657043457, + "eval_runtime": 25.761, + "eval_samples_per_second": 38.818, + "eval_steps_per_second": 2.446, + "step": 350 + }, + { + "epoch": 0.507959479015919, + "grad_norm": 1.3574364185333252, + "learning_rate": 4.20353982300885e-06, + "loss": 0.9949, + "step": 351 + }, + { + "epoch": 0.5094066570188133, + "grad_norm": 1.2600637674331665, + "learning_rate": 4.19951729686243e-06, + "loss": 0.9972, + "step": 352 + }, + { + "epoch": 0.5108538350217077, + "grad_norm": 1.34312903881073, + "learning_rate": 4.19549477071601e-06, + "loss": 0.9695, + "step": 353 + }, + { + "epoch": 0.5123010130246021, + "grad_norm": 1.3600518703460693, + "learning_rate": 4.19147224456959e-06, + "loss": 1.0045, + "step": 354 + }, + { + "epoch": 0.5137481910274964, + "grad_norm": 1.3012819290161133, + "learning_rate": 4.18744971842317e-06, + "loss": 0.9377, + "step": 355 + }, + { + "epoch": 0.5151953690303908, + "grad_norm": 1.4134474992752075, + "learning_rate": 4.18342719227675e-06, + "loss": 1.0233, + "step": 356 + }, + { + "epoch": 0.516642547033285, + "grad_norm": 1.3824727535247803, + "learning_rate": 4.17940466613033e-06, + "loss": 0.979, + "step": 357 + }, + { + "epoch": 0.5180897250361794, + "grad_norm": 1.3735177516937256, + "learning_rate": 4.1753821399839105e-06, + "loss": 0.9657, + "step": 358 + }, + { + "epoch": 0.5195369030390738, + "grad_norm": 1.4538931846618652, + "learning_rate": 4.17135961383749e-06, + "loss": 0.9978, + "step": 359 + }, + { + "epoch": 0.5209840810419681, + "grad_norm": 1.3481597900390625, + "learning_rate": 4.16733708769107e-06, + "loss": 0.9582, + "step": 360 + }, + { + "epoch": 0.5209840810419681, + "eval_loss": 1.051351547241211, + "eval_runtime": 25.6222, + "eval_samples_per_second": 39.029, + "eval_steps_per_second": 2.459, + "step": 360 + }, + { + "epoch": 0.5224312590448625, + "grad_norm": 1.2694309949874878, + "learning_rate": 4.16331456154465e-06, + "loss": 1.0004, + "step": 361 + }, + { + "epoch": 0.5238784370477568, + "grad_norm": 1.461315631866455, + "learning_rate": 4.15929203539823e-06, + "loss": 0.9663, + "step": 362 + }, + { + "epoch": 0.5253256150506512, + "grad_norm": 1.3080368041992188, + "learning_rate": 4.15526950925181e-06, + "loss": 0.9654, + "step": 363 + }, + { + "epoch": 0.5267727930535456, + "grad_norm": 1.4705910682678223, + "learning_rate": 4.15124698310539e-06, + "loss": 0.9707, + "step": 364 + }, + { + "epoch": 0.5282199710564399, + "grad_norm": 1.4834836721420288, + "learning_rate": 4.1472244569589705e-06, + "loss": 0.9758, + "step": 365 + }, + { + "epoch": 0.5296671490593343, + "grad_norm": 1.5037809610366821, + "learning_rate": 4.143201930812551e-06, + "loss": 0.9678, + "step": 366 + }, + { + "epoch": 0.5311143270622286, + "grad_norm": 1.4555189609527588, + "learning_rate": 4.139179404666131e-06, + "loss": 0.9475, + "step": 367 + }, + { + "epoch": 0.532561505065123, + "grad_norm": 1.3858565092086792, + "learning_rate": 4.13515687851971e-06, + "loss": 0.975, + "step": 368 + }, + { + "epoch": 0.5340086830680174, + "grad_norm": 1.3702665567398071, + "learning_rate": 4.13113435237329e-06, + "loss": 0.966, + "step": 369 + }, + { + "epoch": 0.5354558610709117, + "grad_norm": 1.3536401987075806, + "learning_rate": 4.12711182622687e-06, + "loss": 0.9649, + "step": 370 + }, + { + "epoch": 0.5354558610709117, + "eval_loss": 1.0490535497665405, + "eval_runtime": 25.7917, + "eval_samples_per_second": 38.772, + "eval_steps_per_second": 2.443, + "step": 370 + }, + { + "epoch": 0.5369030390738061, + "grad_norm": 1.2890545129776, + "learning_rate": 4.1230893000804505e-06, + "loss": 0.9618, + "step": 371 + }, + { + "epoch": 0.5383502170767004, + "grad_norm": 1.4141520261764526, + "learning_rate": 4.119066773934031e-06, + "loss": 0.9292, + "step": 372 + }, + { + "epoch": 0.5397973950795948, + "grad_norm": 1.4172197580337524, + "learning_rate": 4.115044247787611e-06, + "loss": 0.9642, + "step": 373 + }, + { + "epoch": 0.5412445730824892, + "grad_norm": 1.4644924402236938, + "learning_rate": 4.111021721641191e-06, + "loss": 0.9842, + "step": 374 + }, + { + "epoch": 0.5426917510853835, + "grad_norm": 1.4348403215408325, + "learning_rate": 4.106999195494771e-06, + "loss": 0.9734, + "step": 375 + }, + { + "epoch": 0.5441389290882779, + "grad_norm": 1.311895728111267, + "learning_rate": 4.102976669348351e-06, + "loss": 0.9442, + "step": 376 + }, + { + "epoch": 0.5455861070911722, + "grad_norm": 1.3387318849563599, + "learning_rate": 4.098954143201931e-06, + "loss": 0.9323, + "step": 377 + }, + { + "epoch": 0.5470332850940666, + "grad_norm": 1.4854440689086914, + "learning_rate": 4.094931617055511e-06, + "loss": 0.9577, + "step": 378 + }, + { + "epoch": 0.548480463096961, + "grad_norm": 1.4293216466903687, + "learning_rate": 4.0909090909090915e-06, + "loss": 0.9795, + "step": 379 + }, + { + "epoch": 0.5499276410998553, + "grad_norm": 1.4908519983291626, + "learning_rate": 4.086886564762672e-06, + "loss": 1.0036, + "step": 380 + }, + { + "epoch": 0.5499276410998553, + "eval_loss": 1.0515390634536743, + "eval_runtime": 25.9468, + "eval_samples_per_second": 38.54, + "eval_steps_per_second": 2.428, + "step": 380 + }, + { + "epoch": 0.5513748191027497, + "grad_norm": 1.4864846467971802, + "learning_rate": 4.082864038616252e-06, + "loss": 0.9308, + "step": 381 + }, + { + "epoch": 0.552821997105644, + "grad_norm": 1.3892290592193604, + "learning_rate": 4.078841512469832e-06, + "loss": 0.9669, + "step": 382 + }, + { + "epoch": 0.5542691751085383, + "grad_norm": 1.3630082607269287, + "learning_rate": 4.074818986323412e-06, + "loss": 0.9657, + "step": 383 + }, + { + "epoch": 0.5557163531114327, + "grad_norm": 1.4945286512374878, + "learning_rate": 4.070796460176992e-06, + "loss": 0.9953, + "step": 384 + }, + { + "epoch": 0.557163531114327, + "grad_norm": 1.4368705749511719, + "learning_rate": 4.066773934030571e-06, + "loss": 1.0021, + "step": 385 + }, + { + "epoch": 0.5586107091172214, + "grad_norm": 1.4859812259674072, + "learning_rate": 4.0627514078841515e-06, + "loss": 0.983, + "step": 386 + }, + { + "epoch": 0.5600578871201157, + "grad_norm": 1.4304081201553345, + "learning_rate": 4.058728881737732e-06, + "loss": 0.9178, + "step": 387 + }, + { + "epoch": 0.5615050651230101, + "grad_norm": 1.3401498794555664, + "learning_rate": 4.054706355591312e-06, + "loss": 0.9209, + "step": 388 + }, + { + "epoch": 0.5629522431259045, + "grad_norm": 1.5096213817596436, + "learning_rate": 4.050683829444892e-06, + "loss": 0.9733, + "step": 389 + }, + { + "epoch": 0.5643994211287988, + "grad_norm": 1.5285199880599976, + "learning_rate": 4.046661303298472e-06, + "loss": 0.9649, + "step": 390 + }, + { + "epoch": 0.5643994211287988, + "eval_loss": 1.0477445125579834, + "eval_runtime": 26.0028, + "eval_samples_per_second": 38.457, + "eval_steps_per_second": 2.423, + "step": 390 + }, + { + "epoch": 0.5658465991316932, + "grad_norm": 1.3823034763336182, + "learning_rate": 4.042638777152052e-06, + "loss": 0.9635, + "step": 391 + }, + { + "epoch": 0.5672937771345875, + "grad_norm": 1.4934637546539307, + "learning_rate": 4.038616251005632e-06, + "loss": 0.9799, + "step": 392 + }, + { + "epoch": 0.5687409551374819, + "grad_norm": 1.3570342063903809, + "learning_rate": 4.034593724859212e-06, + "loss": 0.9736, + "step": 393 + }, + { + "epoch": 0.5701881331403763, + "grad_norm": 1.4477742910385132, + "learning_rate": 4.0305711987127925e-06, + "loss": 0.9539, + "step": 394 + }, + { + "epoch": 0.5716353111432706, + "grad_norm": 1.4653184413909912, + "learning_rate": 4.026548672566372e-06, + "loss": 0.9698, + "step": 395 + }, + { + "epoch": 0.573082489146165, + "grad_norm": 1.3791393041610718, + "learning_rate": 4.022526146419952e-06, + "loss": 0.9754, + "step": 396 + }, + { + "epoch": 0.5745296671490593, + "grad_norm": 1.5420602560043335, + "learning_rate": 4.018503620273532e-06, + "loss": 0.9592, + "step": 397 + }, + { + "epoch": 0.5759768451519537, + "grad_norm": 1.494424819946289, + "learning_rate": 4.014481094127112e-06, + "loss": 0.9783, + "step": 398 + }, + { + "epoch": 0.5774240231548481, + "grad_norm": 1.4302630424499512, + "learning_rate": 4.010458567980692e-06, + "loss": 0.9697, + "step": 399 + }, + { + "epoch": 0.5788712011577424, + "grad_norm": 1.4920209646224976, + "learning_rate": 4.0064360418342725e-06, + "loss": 0.9724, + "step": 400 + }, + { + "epoch": 0.5788712011577424, + "eval_loss": 1.0458935499191284, + "eval_runtime": 25.7502, + "eval_samples_per_second": 38.835, + "eval_steps_per_second": 2.447, + "step": 400 + }, + { + "epoch": 0.5803183791606368, + "grad_norm": 1.4532008171081543, + "learning_rate": 4.002413515687853e-06, + "loss": 0.9576, + "step": 401 + }, + { + "epoch": 0.5817655571635311, + "grad_norm": 1.4824233055114746, + "learning_rate": 3.998390989541433e-06, + "loss": 0.9631, + "step": 402 + }, + { + "epoch": 0.5832127351664255, + "grad_norm": 1.4938057661056519, + "learning_rate": 3.994368463395013e-06, + "loss": 0.9522, + "step": 403 + }, + { + "epoch": 0.5846599131693199, + "grad_norm": 1.5085760354995728, + "learning_rate": 3.990345937248592e-06, + "loss": 0.9873, + "step": 404 + }, + { + "epoch": 0.5861070911722142, + "grad_norm": 1.4096283912658691, + "learning_rate": 3.986323411102172e-06, + "loss": 0.9912, + "step": 405 + }, + { + "epoch": 0.5875542691751086, + "grad_norm": 1.4348584413528442, + "learning_rate": 3.982300884955752e-06, + "loss": 1.0022, + "step": 406 + }, + { + "epoch": 0.5890014471780028, + "grad_norm": 1.4397310018539429, + "learning_rate": 3.9782783588093325e-06, + "loss": 0.9702, + "step": 407 + }, + { + "epoch": 0.5904486251808972, + "grad_norm": 1.3248670101165771, + "learning_rate": 3.974255832662913e-06, + "loss": 0.9476, + "step": 408 + }, + { + "epoch": 0.5918958031837916, + "grad_norm": 1.3723477125167847, + "learning_rate": 3.970233306516493e-06, + "loss": 0.9463, + "step": 409 + }, + { + "epoch": 0.5933429811866859, + "grad_norm": 1.4383715391159058, + "learning_rate": 3.966210780370073e-06, + "loss": 0.9341, + "step": 410 + }, + { + "epoch": 0.5933429811866859, + "eval_loss": 1.04234778881073, + "eval_runtime": 25.7716, + "eval_samples_per_second": 38.802, + "eval_steps_per_second": 2.445, + "step": 410 + }, + { + "epoch": 0.5947901591895803, + "grad_norm": 1.4978934526443481, + "learning_rate": 3.962188254223653e-06, + "loss": 0.9593, + "step": 411 + }, + { + "epoch": 0.5962373371924746, + "grad_norm": 1.4144810438156128, + "learning_rate": 3.958165728077233e-06, + "loss": 0.9422, + "step": 412 + }, + { + "epoch": 0.597684515195369, + "grad_norm": 1.4646399021148682, + "learning_rate": 3.954143201930813e-06, + "loss": 0.9329, + "step": 413 + }, + { + "epoch": 0.5991316931982634, + "grad_norm": 1.5337008237838745, + "learning_rate": 3.9501206757843926e-06, + "loss": 0.9847, + "step": 414 + }, + { + "epoch": 0.6005788712011577, + "grad_norm": 1.342451810836792, + "learning_rate": 3.946098149637973e-06, + "loss": 0.9827, + "step": 415 + }, + { + "epoch": 0.6020260492040521, + "grad_norm": 1.4263315200805664, + "learning_rate": 3.942075623491553e-06, + "loss": 0.9404, + "step": 416 + }, + { + "epoch": 0.6034732272069464, + "grad_norm": 1.4434232711791992, + "learning_rate": 3.938053097345133e-06, + "loss": 0.9266, + "step": 417 + }, + { + "epoch": 0.6049204052098408, + "grad_norm": 1.4536890983581543, + "learning_rate": 3.934030571198713e-06, + "loss": 0.9889, + "step": 418 + }, + { + "epoch": 0.6063675832127352, + "grad_norm": 1.3431605100631714, + "learning_rate": 3.930008045052293e-06, + "loss": 0.9614, + "step": 419 + }, + { + "epoch": 0.6078147612156295, + "grad_norm": 1.3683048486709595, + "learning_rate": 3.925985518905873e-06, + "loss": 0.9207, + "step": 420 + }, + { + "epoch": 0.6078147612156295, + "eval_loss": 1.0399519205093384, + "eval_runtime": 25.9477, + "eval_samples_per_second": 38.539, + "eval_steps_per_second": 2.428, + "step": 420 + }, + { + "epoch": 0.6092619392185239, + "grad_norm": 1.5089398622512817, + "learning_rate": 3.9219629927594534e-06, + "loss": 0.9382, + "step": 421 + }, + { + "epoch": 0.6107091172214182, + "grad_norm": 1.5236350297927856, + "learning_rate": 3.9179404666130336e-06, + "loss": 0.9683, + "step": 422 + }, + { + "epoch": 0.6121562952243126, + "grad_norm": 1.3001353740692139, + "learning_rate": 3.913917940466614e-06, + "loss": 0.9385, + "step": 423 + }, + { + "epoch": 0.613603473227207, + "grad_norm": 1.6009608507156372, + "learning_rate": 3.909895414320193e-06, + "loss": 0.9988, + "step": 424 + }, + { + "epoch": 0.6150506512301013, + "grad_norm": 1.3586127758026123, + "learning_rate": 3.905872888173773e-06, + "loss": 0.9613, + "step": 425 + }, + { + "epoch": 0.6164978292329957, + "grad_norm": 1.4171210527420044, + "learning_rate": 3.901850362027353e-06, + "loss": 0.9485, + "step": 426 + }, + { + "epoch": 0.61794500723589, + "grad_norm": 1.4831761121749878, + "learning_rate": 3.897827835880933e-06, + "loss": 0.9337, + "step": 427 + }, + { + "epoch": 0.6193921852387844, + "grad_norm": 1.3427209854125977, + "learning_rate": 3.8938053097345135e-06, + "loss": 0.9319, + "step": 428 + }, + { + "epoch": 0.6208393632416788, + "grad_norm": 1.3631221055984497, + "learning_rate": 3.889782783588094e-06, + "loss": 0.9416, + "step": 429 + }, + { + "epoch": 0.622286541244573, + "grad_norm": 1.3836332559585571, + "learning_rate": 3.885760257441674e-06, + "loss": 0.9422, + "step": 430 + }, + { + "epoch": 0.622286541244573, + "eval_loss": 1.0372964143753052, + "eval_runtime": 26.054, + "eval_samples_per_second": 38.382, + "eval_steps_per_second": 2.418, + "step": 430 + }, + { + "epoch": 0.6237337192474675, + "grad_norm": 1.4307619333267212, + "learning_rate": 3.881737731295254e-06, + "loss": 0.9395, + "step": 431 + }, + { + "epoch": 0.6251808972503617, + "grad_norm": 1.338143229484558, + "learning_rate": 3.877715205148834e-06, + "loss": 0.9527, + "step": 432 + }, + { + "epoch": 0.6266280752532561, + "grad_norm": 1.3655433654785156, + "learning_rate": 3.873692679002414e-06, + "loss": 0.9633, + "step": 433 + }, + { + "epoch": 0.6280752532561505, + "grad_norm": 1.4386481046676636, + "learning_rate": 3.869670152855993e-06, + "loss": 0.9878, + "step": 434 + }, + { + "epoch": 0.6295224312590448, + "grad_norm": 1.4618841409683228, + "learning_rate": 3.8656476267095735e-06, + "loss": 0.922, + "step": 435 + }, + { + "epoch": 0.6309696092619392, + "grad_norm": 1.4335823059082031, + "learning_rate": 3.861625100563154e-06, + "loss": 0.9346, + "step": 436 + }, + { + "epoch": 0.6324167872648335, + "grad_norm": 1.4543538093566895, + "learning_rate": 3.857602574416734e-06, + "loss": 0.9847, + "step": 437 + }, + { + "epoch": 0.6338639652677279, + "grad_norm": 1.460121750831604, + "learning_rate": 3.853580048270314e-06, + "loss": 0.9954, + "step": 438 + }, + { + "epoch": 0.6353111432706223, + "grad_norm": 1.5101383924484253, + "learning_rate": 3.849557522123894e-06, + "loss": 0.963, + "step": 439 + }, + { + "epoch": 0.6367583212735166, + "grad_norm": 1.5034980773925781, + "learning_rate": 3.845534995977474e-06, + "loss": 0.9378, + "step": 440 + }, + { + "epoch": 0.6367583212735166, + "eval_loss": 1.034120798110962, + "eval_runtime": 26.0364, + "eval_samples_per_second": 38.408, + "eval_steps_per_second": 2.42, + "step": 440 + }, + { + "epoch": 0.638205499276411, + "grad_norm": 1.375977873802185, + "learning_rate": 3.841512469831054e-06, + "loss": 0.9232, + "step": 441 + }, + { + "epoch": 0.6396526772793053, + "grad_norm": 1.486662745475769, + "learning_rate": 3.8374899436846344e-06, + "loss": 0.9634, + "step": 442 + }, + { + "epoch": 0.6410998552821997, + "grad_norm": 1.4522157907485962, + "learning_rate": 3.8334674175382146e-06, + "loss": 0.9577, + "step": 443 + }, + { + "epoch": 0.6425470332850941, + "grad_norm": 1.3580085039138794, + "learning_rate": 3.829444891391794e-06, + "loss": 0.9283, + "step": 444 + }, + { + "epoch": 0.6439942112879884, + "grad_norm": 1.5201516151428223, + "learning_rate": 3.825422365245374e-06, + "loss": 0.9358, + "step": 445 + }, + { + "epoch": 0.6454413892908828, + "grad_norm": 1.5232244729995728, + "learning_rate": 3.821399839098954e-06, + "loss": 0.9197, + "step": 446 + }, + { + "epoch": 0.6468885672937771, + "grad_norm": 1.3573877811431885, + "learning_rate": 3.817377312952534e-06, + "loss": 0.9888, + "step": 447 + }, + { + "epoch": 0.6483357452966715, + "grad_norm": 1.3900681734085083, + "learning_rate": 3.8133547868061144e-06, + "loss": 0.9388, + "step": 448 + }, + { + "epoch": 0.6497829232995659, + "grad_norm": 1.4748214483261108, + "learning_rate": 3.8093322606596945e-06, + "loss": 0.9883, + "step": 449 + }, + { + "epoch": 0.6512301013024602, + "grad_norm": 1.3832060098648071, + "learning_rate": 3.8053097345132746e-06, + "loss": 0.9676, + "step": 450 + }, + { + "epoch": 0.6512301013024602, + "eval_loss": 1.031374454498291, + "eval_runtime": 25.8404, + "eval_samples_per_second": 38.699, + "eval_steps_per_second": 2.438, + "step": 450 + }, + { + "epoch": 0.6526772793053546, + "grad_norm": 1.4417644739151, + "learning_rate": 3.8012872083668543e-06, + "loss": 0.9424, + "step": 451 + }, + { + "epoch": 0.6541244573082489, + "grad_norm": 1.3401044607162476, + "learning_rate": 3.7972646822204344e-06, + "loss": 0.9427, + "step": 452 + }, + { + "epoch": 0.6555716353111433, + "grad_norm": 1.4024505615234375, + "learning_rate": 3.7932421560740146e-06, + "loss": 0.9422, + "step": 453 + }, + { + "epoch": 0.6570188133140377, + "grad_norm": 1.4496229887008667, + "learning_rate": 3.7892196299275947e-06, + "loss": 0.9249, + "step": 454 + }, + { + "epoch": 0.658465991316932, + "grad_norm": 1.3784229755401611, + "learning_rate": 3.785197103781175e-06, + "loss": 0.908, + "step": 455 + }, + { + "epoch": 0.6599131693198264, + "grad_norm": 1.413669228553772, + "learning_rate": 3.7811745776347545e-06, + "loss": 0.9697, + "step": 456 + }, + { + "epoch": 0.6613603473227206, + "grad_norm": 1.5418171882629395, + "learning_rate": 3.7771520514883347e-06, + "loss": 0.9636, + "step": 457 + }, + { + "epoch": 0.662807525325615, + "grad_norm": 1.3097503185272217, + "learning_rate": 3.7731295253419148e-06, + "loss": 0.9562, + "step": 458 + }, + { + "epoch": 0.6642547033285094, + "grad_norm": 1.3388592004776, + "learning_rate": 3.769106999195495e-06, + "loss": 0.9441, + "step": 459 + }, + { + "epoch": 0.6657018813314037, + "grad_norm": 1.3556655645370483, + "learning_rate": 3.765084473049075e-06, + "loss": 0.9527, + "step": 460 + }, + { + "epoch": 0.6657018813314037, + "eval_loss": 1.0326937437057495, + "eval_runtime": 25.7402, + "eval_samples_per_second": 38.85, + "eval_steps_per_second": 2.448, + "step": 460 + }, + { + "epoch": 0.6671490593342981, + "grad_norm": 1.3645198345184326, + "learning_rate": 3.7610619469026547e-06, + "loss": 0.972, + "step": 461 + }, + { + "epoch": 0.6685962373371924, + "grad_norm": 1.4470856189727783, + "learning_rate": 3.757039420756235e-06, + "loss": 0.9749, + "step": 462 + }, + { + "epoch": 0.6700434153400868, + "grad_norm": 1.3615493774414062, + "learning_rate": 3.753016894609815e-06, + "loss": 0.9386, + "step": 463 + }, + { + "epoch": 0.6714905933429812, + "grad_norm": 1.355056643486023, + "learning_rate": 3.7489943684633956e-06, + "loss": 0.9761, + "step": 464 + }, + { + "epoch": 0.6729377713458755, + "grad_norm": 1.4716227054595947, + "learning_rate": 3.7449718423169757e-06, + "loss": 0.983, + "step": 465 + }, + { + "epoch": 0.6743849493487699, + "grad_norm": 1.4398634433746338, + "learning_rate": 3.740949316170556e-06, + "loss": 0.9279, + "step": 466 + }, + { + "epoch": 0.6758321273516642, + "grad_norm": 1.469275712966919, + "learning_rate": 3.7369267900241355e-06, + "loss": 0.9343, + "step": 467 + }, + { + "epoch": 0.6772793053545586, + "grad_norm": 1.383954405784607, + "learning_rate": 3.7329042638777156e-06, + "loss": 0.9372, + "step": 468 + }, + { + "epoch": 0.678726483357453, + "grad_norm": 1.3994921445846558, + "learning_rate": 3.7288817377312958e-06, + "loss": 0.9399, + "step": 469 + }, + { + "epoch": 0.6801736613603473, + "grad_norm": 1.4469072818756104, + "learning_rate": 3.724859211584876e-06, + "loss": 0.9696, + "step": 470 + }, + { + "epoch": 0.6801736613603473, + "eval_loss": 1.0329785346984863, + "eval_runtime": 25.7443, + "eval_samples_per_second": 38.843, + "eval_steps_per_second": 2.447, + "step": 470 + }, + { + "epoch": 0.6816208393632417, + "grad_norm": 1.447624921798706, + "learning_rate": 3.720836685438456e-06, + "loss": 0.985, + "step": 471 + }, + { + "epoch": 0.683068017366136, + "grad_norm": 1.409177303314209, + "learning_rate": 3.7168141592920357e-06, + "loss": 0.948, + "step": 472 + }, + { + "epoch": 0.6845151953690304, + "grad_norm": 1.4697754383087158, + "learning_rate": 3.712791633145616e-06, + "loss": 0.9365, + "step": 473 + }, + { + "epoch": 0.6859623733719248, + "grad_norm": 1.4154843091964722, + "learning_rate": 3.708769106999196e-06, + "loss": 0.9701, + "step": 474 + }, + { + "epoch": 0.6874095513748191, + "grad_norm": 1.4701588153839111, + "learning_rate": 3.704746580852776e-06, + "loss": 0.9548, + "step": 475 + }, + { + "epoch": 0.6888567293777135, + "grad_norm": 1.4201622009277344, + "learning_rate": 3.7007240547063562e-06, + "loss": 0.9288, + "step": 476 + }, + { + "epoch": 0.6903039073806078, + "grad_norm": 1.365224003791809, + "learning_rate": 3.696701528559936e-06, + "loss": 0.9461, + "step": 477 + }, + { + "epoch": 0.6917510853835022, + "grad_norm": 1.3176798820495605, + "learning_rate": 3.692679002413516e-06, + "loss": 0.9193, + "step": 478 + }, + { + "epoch": 0.6931982633863966, + "grad_norm": 1.4290233850479126, + "learning_rate": 3.688656476267096e-06, + "loss": 0.9356, + "step": 479 + }, + { + "epoch": 0.6946454413892909, + "grad_norm": 1.4030050039291382, + "learning_rate": 3.6846339501206763e-06, + "loss": 0.9634, + "step": 480 + }, + { + "epoch": 0.6946454413892909, + "eval_loss": 1.0300215482711792, + "eval_runtime": 25.9854, + "eval_samples_per_second": 38.483, + "eval_steps_per_second": 2.424, + "step": 480 + }, + { + "epoch": 0.6960926193921853, + "grad_norm": 1.3863651752471924, + "learning_rate": 3.6806114239742565e-06, + "loss": 0.8867, + "step": 481 + }, + { + "epoch": 0.6975397973950795, + "grad_norm": 1.4828954935073853, + "learning_rate": 3.676588897827836e-06, + "loss": 0.9357, + "step": 482 + }, + { + "epoch": 0.6989869753979739, + "grad_norm": 1.4383491277694702, + "learning_rate": 3.6725663716814163e-06, + "loss": 0.9295, + "step": 483 + }, + { + "epoch": 0.7004341534008683, + "grad_norm": 1.446408748626709, + "learning_rate": 3.6685438455349964e-06, + "loss": 0.9609, + "step": 484 + }, + { + "epoch": 0.7018813314037626, + "grad_norm": 1.3710352182388306, + "learning_rate": 3.6645213193885765e-06, + "loss": 0.9365, + "step": 485 + }, + { + "epoch": 0.703328509406657, + "grad_norm": 1.4681317806243896, + "learning_rate": 3.6604987932421567e-06, + "loss": 0.9732, + "step": 486 + }, + { + "epoch": 0.7047756874095513, + "grad_norm": 1.3563289642333984, + "learning_rate": 3.6564762670957364e-06, + "loss": 0.9393, + "step": 487 + }, + { + "epoch": 0.7062228654124457, + "grad_norm": 1.3670072555541992, + "learning_rate": 3.6524537409493165e-06, + "loss": 0.9363, + "step": 488 + }, + { + "epoch": 0.7076700434153401, + "grad_norm": 1.4312621355056763, + "learning_rate": 3.6484312148028966e-06, + "loss": 0.9229, + "step": 489 + }, + { + "epoch": 0.7091172214182344, + "grad_norm": 1.4633783102035522, + "learning_rate": 3.6444086886564768e-06, + "loss": 0.9515, + "step": 490 + }, + { + "epoch": 0.7091172214182344, + "eval_loss": 1.0295219421386719, + "eval_runtime": 25.7043, + "eval_samples_per_second": 38.904, + "eval_steps_per_second": 2.451, + "step": 490 + }, + { + "epoch": 0.7105643994211288, + "grad_norm": 1.5504322052001953, + "learning_rate": 3.640386162510057e-06, + "loss": 0.9707, + "step": 491 + }, + { + "epoch": 0.7120115774240231, + "grad_norm": 1.445961594581604, + "learning_rate": 3.6363636363636366e-06, + "loss": 0.9924, + "step": 492 + }, + { + "epoch": 0.7134587554269175, + "grad_norm": 1.49595308303833, + "learning_rate": 3.6323411102172167e-06, + "loss": 0.9619, + "step": 493 + }, + { + "epoch": 0.7149059334298119, + "grad_norm": 1.5091534852981567, + "learning_rate": 3.628318584070797e-06, + "loss": 0.9393, + "step": 494 + }, + { + "epoch": 0.7163531114327062, + "grad_norm": 1.3934762477874756, + "learning_rate": 3.624296057924377e-06, + "loss": 0.9252, + "step": 495 + }, + { + "epoch": 0.7178002894356006, + "grad_norm": 1.4442094564437866, + "learning_rate": 3.620273531777957e-06, + "loss": 0.9149, + "step": 496 + }, + { + "epoch": 0.7192474674384949, + "grad_norm": 1.3727437257766724, + "learning_rate": 3.616251005631537e-06, + "loss": 0.9136, + "step": 497 + }, + { + "epoch": 0.7206946454413893, + "grad_norm": 1.3915483951568604, + "learning_rate": 3.612228479485117e-06, + "loss": 0.9132, + "step": 498 + }, + { + "epoch": 0.7221418234442837, + "grad_norm": 1.5220942497253418, + "learning_rate": 3.608205953338697e-06, + "loss": 0.9576, + "step": 499 + }, + { + "epoch": 0.723589001447178, + "grad_norm": 1.4072246551513672, + "learning_rate": 3.604183427192277e-06, + "loss": 0.9372, + "step": 500 + }, + { + "epoch": 0.723589001447178, + "eval_loss": 1.029038906097412, + "eval_runtime": 25.9369, + "eval_samples_per_second": 38.555, + "eval_steps_per_second": 2.429, + "step": 500 + }, + { + "epoch": 0.7250361794500724, + "grad_norm": 1.3465371131896973, + "learning_rate": 3.6001609010458573e-06, + "loss": 0.9192, + "step": 501 + }, + { + "epoch": 0.7264833574529667, + "grad_norm": 1.5331004858016968, + "learning_rate": 3.596138374899437e-06, + "loss": 0.9672, + "step": 502 + }, + { + "epoch": 0.7279305354558611, + "grad_norm": 1.4934667348861694, + "learning_rate": 3.592115848753017e-06, + "loss": 0.9349, + "step": 503 + }, + { + "epoch": 0.7293777134587555, + "grad_norm": 1.3691191673278809, + "learning_rate": 3.5880933226065973e-06, + "loss": 0.9471, + "step": 504 + }, + { + "epoch": 0.7308248914616498, + "grad_norm": 1.450124979019165, + "learning_rate": 3.5840707964601774e-06, + "loss": 0.9424, + "step": 505 + }, + { + "epoch": 0.7322720694645442, + "grad_norm": 1.5474168062210083, + "learning_rate": 3.5800482703137575e-06, + "loss": 0.9187, + "step": 506 + }, + { + "epoch": 0.7337192474674384, + "grad_norm": 1.4401590824127197, + "learning_rate": 3.5760257441673372e-06, + "loss": 0.9218, + "step": 507 + }, + { + "epoch": 0.7351664254703328, + "grad_norm": 1.5775281190872192, + "learning_rate": 3.5720032180209174e-06, + "loss": 0.93, + "step": 508 + }, + { + "epoch": 0.7366136034732272, + "grad_norm": 1.5710840225219727, + "learning_rate": 3.5679806918744975e-06, + "loss": 0.8954, + "step": 509 + }, + { + "epoch": 0.7380607814761215, + "grad_norm": 1.5885707139968872, + "learning_rate": 3.5639581657280776e-06, + "loss": 0.9682, + "step": 510 + }, + { + "epoch": 0.7380607814761215, + "eval_loss": 1.0261106491088867, + "eval_runtime": 25.8126, + "eval_samples_per_second": 38.741, + "eval_steps_per_second": 2.441, + "step": 510 + }, + { + "epoch": 0.7395079594790159, + "grad_norm": 1.4182794094085693, + "learning_rate": 3.5599356395816577e-06, + "loss": 0.9417, + "step": 511 + }, + { + "epoch": 0.7409551374819102, + "grad_norm": 1.416455626487732, + "learning_rate": 3.5559131134352374e-06, + "loss": 0.9325, + "step": 512 + }, + { + "epoch": 0.7424023154848046, + "grad_norm": 1.511069416999817, + "learning_rate": 3.5518905872888176e-06, + "loss": 0.9621, + "step": 513 + }, + { + "epoch": 0.743849493487699, + "grad_norm": 1.5709325075149536, + "learning_rate": 3.5478680611423977e-06, + "loss": 0.9302, + "step": 514 + }, + { + "epoch": 0.7452966714905933, + "grad_norm": 1.5268837213516235, + "learning_rate": 3.543845534995978e-06, + "loss": 0.8761, + "step": 515 + }, + { + "epoch": 0.7467438494934877, + "grad_norm": 1.6121459007263184, + "learning_rate": 3.539823008849558e-06, + "loss": 0.9322, + "step": 516 + }, + { + "epoch": 0.748191027496382, + "grad_norm": 1.479650616645813, + "learning_rate": 3.5358004827031377e-06, + "loss": 0.9654, + "step": 517 + }, + { + "epoch": 0.7496382054992764, + "grad_norm": 1.6954939365386963, + "learning_rate": 3.5317779565567178e-06, + "loss": 0.9719, + "step": 518 + }, + { + "epoch": 0.7510853835021708, + "grad_norm": 1.4463775157928467, + "learning_rate": 3.527755430410298e-06, + "loss": 0.8981, + "step": 519 + }, + { + "epoch": 0.7525325615050651, + "grad_norm": 1.4564443826675415, + "learning_rate": 3.523732904263878e-06, + "loss": 0.9273, + "step": 520 + }, + { + "epoch": 0.7525325615050651, + "eval_loss": 1.0260487794876099, + "eval_runtime": 25.8612, + "eval_samples_per_second": 38.668, + "eval_steps_per_second": 2.436, + "step": 520 + }, + { + "epoch": 0.7539797395079595, + "grad_norm": 1.436152696609497, + "learning_rate": 3.519710378117458e-06, + "loss": 0.9541, + "step": 521 + }, + { + "epoch": 0.7554269175108539, + "grad_norm": 1.596771478652954, + "learning_rate": 3.515687851971038e-06, + "loss": 0.9138, + "step": 522 + }, + { + "epoch": 0.7568740955137482, + "grad_norm": 1.4977961778640747, + "learning_rate": 3.511665325824618e-06, + "loss": 0.9557, + "step": 523 + }, + { + "epoch": 0.7583212735166426, + "grad_norm": 1.4800033569335938, + "learning_rate": 3.507642799678198e-06, + "loss": 0.9372, + "step": 524 + }, + { + "epoch": 0.7597684515195369, + "grad_norm": 1.4876062870025635, + "learning_rate": 3.5036202735317783e-06, + "loss": 0.9345, + "step": 525 + }, + { + "epoch": 0.7612156295224313, + "grad_norm": 1.4231287240982056, + "learning_rate": 3.4995977473853584e-06, + "loss": 0.9444, + "step": 526 + }, + { + "epoch": 0.7626628075253257, + "grad_norm": 1.5092722177505493, + "learning_rate": 3.495575221238938e-06, + "loss": 0.9536, + "step": 527 + }, + { + "epoch": 0.76410998552822, + "grad_norm": 1.465430498123169, + "learning_rate": 3.4915526950925182e-06, + "loss": 0.9378, + "step": 528 + }, + { + "epoch": 0.7655571635311144, + "grad_norm": 1.4292782545089722, + "learning_rate": 3.4875301689460983e-06, + "loss": 0.9345, + "step": 529 + }, + { + "epoch": 0.7670043415340086, + "grad_norm": 1.606329321861267, + "learning_rate": 3.4835076427996785e-06, + "loss": 0.9446, + "step": 530 + }, + { + "epoch": 0.7670043415340086, + "eval_loss": 1.0236235857009888, + "eval_runtime": 25.9147, + "eval_samples_per_second": 38.588, + "eval_steps_per_second": 2.431, + "step": 530 + }, + { + "epoch": 0.768451519536903, + "grad_norm": 1.3893780708312988, + "learning_rate": 3.4794851166532586e-06, + "loss": 0.9234, + "step": 531 + }, + { + "epoch": 0.7698986975397974, + "grad_norm": 1.4836314916610718, + "learning_rate": 3.4754625905068383e-06, + "loss": 0.9164, + "step": 532 + }, + { + "epoch": 0.7713458755426917, + "grad_norm": 1.3877429962158203, + "learning_rate": 3.4714400643604184e-06, + "loss": 0.9433, + "step": 533 + }, + { + "epoch": 0.7727930535455861, + "grad_norm": 1.486474871635437, + "learning_rate": 3.4674175382139986e-06, + "loss": 0.9303, + "step": 534 + }, + { + "epoch": 0.7742402315484804, + "grad_norm": 1.4549676179885864, + "learning_rate": 3.4633950120675787e-06, + "loss": 0.9438, + "step": 535 + }, + { + "epoch": 0.7756874095513748, + "grad_norm": 1.4801554679870605, + "learning_rate": 3.4593724859211584e-06, + "loss": 0.9441, + "step": 536 + }, + { + "epoch": 0.7771345875542692, + "grad_norm": 1.4578830003738403, + "learning_rate": 3.4553499597747385e-06, + "loss": 0.9301, + "step": 537 + }, + { + "epoch": 0.7785817655571635, + "grad_norm": 1.4650003910064697, + "learning_rate": 3.4513274336283186e-06, + "loss": 0.939, + "step": 538 + }, + { + "epoch": 0.7800289435600579, + "grad_norm": 1.416143774986267, + "learning_rate": 3.4473049074818988e-06, + "loss": 0.9211, + "step": 539 + }, + { + "epoch": 0.7814761215629522, + "grad_norm": 1.768223762512207, + "learning_rate": 3.443282381335479e-06, + "loss": 0.9502, + "step": 540 + }, + { + "epoch": 0.7814761215629522, + "eval_loss": 1.024090051651001, + "eval_runtime": 25.9198, + "eval_samples_per_second": 38.58, + "eval_steps_per_second": 2.431, + "step": 540 + }, + { + "epoch": 0.7829232995658466, + "grad_norm": 1.5100836753845215, + "learning_rate": 3.4392598551890586e-06, + "loss": 0.957, + "step": 541 + }, + { + "epoch": 0.784370477568741, + "grad_norm": 1.4793955087661743, + "learning_rate": 3.4352373290426387e-06, + "loss": 0.9341, + "step": 542 + }, + { + "epoch": 0.7858176555716353, + "grad_norm": 1.5424785614013672, + "learning_rate": 3.431214802896219e-06, + "loss": 0.9246, + "step": 543 + }, + { + "epoch": 0.7872648335745297, + "grad_norm": 1.5629793405532837, + "learning_rate": 3.427192276749799e-06, + "loss": 0.9494, + "step": 544 + }, + { + "epoch": 0.788712011577424, + "grad_norm": 1.5148011445999146, + "learning_rate": 3.423169750603379e-06, + "loss": 0.9063, + "step": 545 + }, + { + "epoch": 0.7901591895803184, + "grad_norm": 1.489189624786377, + "learning_rate": 3.419147224456959e-06, + "loss": 0.9498, + "step": 546 + }, + { + "epoch": 0.7916063675832128, + "grad_norm": 1.5156149864196777, + "learning_rate": 3.415124698310539e-06, + "loss": 0.9377, + "step": 547 + }, + { + "epoch": 0.7930535455861071, + "grad_norm": 1.5290278196334839, + "learning_rate": 3.411102172164119e-06, + "loss": 0.9235, + "step": 548 + }, + { + "epoch": 0.7945007235890015, + "grad_norm": 1.5882658958435059, + "learning_rate": 3.407079646017699e-06, + "loss": 0.9498, + "step": 549 + }, + { + "epoch": 0.7959479015918958, + "grad_norm": 1.4402974843978882, + "learning_rate": 3.4030571198712793e-06, + "loss": 0.9396, + "step": 550 + }, + { + "epoch": 0.7959479015918958, + "eval_loss": 1.020573616027832, + "eval_runtime": 25.7599, + "eval_samples_per_second": 38.82, + "eval_steps_per_second": 2.446, + "step": 550 + }, + { + "epoch": 0.7973950795947902, + "grad_norm": 1.4785488843917847, + "learning_rate": 3.399034593724859e-06, + "loss": 0.9143, + "step": 551 + }, + { + "epoch": 0.7988422575976846, + "grad_norm": 1.4422612190246582, + "learning_rate": 3.395012067578439e-06, + "loss": 0.9585, + "step": 552 + }, + { + "epoch": 0.8002894356005789, + "grad_norm": 1.3907090425491333, + "learning_rate": 3.3909895414320197e-06, + "loss": 0.9288, + "step": 553 + }, + { + "epoch": 0.8017366136034733, + "grad_norm": 1.4476579427719116, + "learning_rate": 3.3869670152856e-06, + "loss": 0.9135, + "step": 554 + }, + { + "epoch": 0.8031837916063675, + "grad_norm": 1.2672919034957886, + "learning_rate": 3.38294448913918e-06, + "loss": 0.9093, + "step": 555 + }, + { + "epoch": 0.804630969609262, + "grad_norm": 1.4998122453689575, + "learning_rate": 3.37892196299276e-06, + "loss": 0.9147, + "step": 556 + }, + { + "epoch": 0.8060781476121563, + "grad_norm": 1.3593398332595825, + "learning_rate": 3.3748994368463402e-06, + "loss": 0.914, + "step": 557 + }, + { + "epoch": 0.8075253256150506, + "grad_norm": 1.5175474882125854, + "learning_rate": 3.37087691069992e-06, + "loss": 0.9232, + "step": 558 + }, + { + "epoch": 0.808972503617945, + "grad_norm": 1.7521854639053345, + "learning_rate": 3.3668543845535e-06, + "loss": 0.9683, + "step": 559 + }, + { + "epoch": 0.8104196816208393, + "grad_norm": 1.451035499572754, + "learning_rate": 3.36283185840708e-06, + "loss": 0.9462, + "step": 560 + }, + { + "epoch": 0.8104196816208393, + "eval_loss": 1.0183910131454468, + "eval_runtime": 25.7176, + "eval_samples_per_second": 38.884, + "eval_steps_per_second": 2.45, + "step": 560 + }, + { + "epoch": 0.8118668596237337, + "grad_norm": 1.3751353025436401, + "learning_rate": 3.3588093322606603e-06, + "loss": 0.9493, + "step": 561 + }, + { + "epoch": 0.8133140376266281, + "grad_norm": 1.437719464302063, + "learning_rate": 3.3547868061142404e-06, + "loss": 0.915, + "step": 562 + }, + { + "epoch": 0.8147612156295224, + "grad_norm": 1.3803507089614868, + "learning_rate": 3.35076427996782e-06, + "loss": 0.9247, + "step": 563 + }, + { + "epoch": 0.8162083936324168, + "grad_norm": 1.4442408084869385, + "learning_rate": 3.3467417538214003e-06, + "loss": 0.9232, + "step": 564 + }, + { + "epoch": 0.8176555716353111, + "grad_norm": 1.419447422027588, + "learning_rate": 3.3427192276749804e-06, + "loss": 0.9242, + "step": 565 + }, + { + "epoch": 0.8191027496382055, + "grad_norm": 1.4496979713439941, + "learning_rate": 3.3386967015285605e-06, + "loss": 0.9079, + "step": 566 + }, + { + "epoch": 0.8205499276410999, + "grad_norm": 1.5532276630401611, + "learning_rate": 3.3346741753821407e-06, + "loss": 0.922, + "step": 567 + }, + { + "epoch": 0.8219971056439942, + "grad_norm": 1.5804080963134766, + "learning_rate": 3.3306516492357204e-06, + "loss": 0.9258, + "step": 568 + }, + { + "epoch": 0.8234442836468886, + "grad_norm": 1.5104866027832031, + "learning_rate": 3.3266291230893005e-06, + "loss": 0.9419, + "step": 569 + }, + { + "epoch": 0.8248914616497829, + "grad_norm": 1.4507907629013062, + "learning_rate": 3.3226065969428806e-06, + "loss": 0.9495, + "step": 570 + }, + { + "epoch": 0.8248914616497829, + "eval_loss": 1.018306016921997, + "eval_runtime": 25.653, + "eval_samples_per_second": 38.982, + "eval_steps_per_second": 2.456, + "step": 570 + }, + { + "epoch": 0.8263386396526773, + "grad_norm": 1.8267414569854736, + "learning_rate": 3.3185840707964607e-06, + "loss": 0.9442, + "step": 571 + }, + { + "epoch": 0.8277858176555717, + "grad_norm": 1.4675520658493042, + "learning_rate": 3.3145615446500404e-06, + "loss": 0.9802, + "step": 572 + }, + { + "epoch": 0.829232995658466, + "grad_norm": 1.506719946861267, + "learning_rate": 3.3105390185036206e-06, + "loss": 0.9248, + "step": 573 + }, + { + "epoch": 0.8306801736613604, + "grad_norm": 1.4712932109832764, + "learning_rate": 3.3065164923572007e-06, + "loss": 0.9366, + "step": 574 + }, + { + "epoch": 0.8321273516642547, + "grad_norm": 1.5222026109695435, + "learning_rate": 3.302493966210781e-06, + "loss": 0.913, + "step": 575 + }, + { + "epoch": 0.8335745296671491, + "grad_norm": 1.4533673524856567, + "learning_rate": 3.298471440064361e-06, + "loss": 0.9223, + "step": 576 + }, + { + "epoch": 0.8350217076700435, + "grad_norm": 1.446425437927246, + "learning_rate": 3.2944489139179407e-06, + "loss": 0.8802, + "step": 577 + }, + { + "epoch": 0.8364688856729378, + "grad_norm": 1.699572205543518, + "learning_rate": 3.2904263877715208e-06, + "loss": 0.9203, + "step": 578 + }, + { + "epoch": 0.8379160636758322, + "grad_norm": 1.5956206321716309, + "learning_rate": 3.286403861625101e-06, + "loss": 0.9334, + "step": 579 + }, + { + "epoch": 0.8393632416787264, + "grad_norm": 1.4471442699432373, + "learning_rate": 3.282381335478681e-06, + "loss": 0.9492, + "step": 580 + }, + { + "epoch": 0.8393632416787264, + "eval_loss": 1.0188478231430054, + "eval_runtime": 25.7874, + "eval_samples_per_second": 38.779, + "eval_steps_per_second": 2.443, + "step": 580 + }, + { + "epoch": 0.8408104196816208, + "grad_norm": 1.542809247970581, + "learning_rate": 3.278358809332261e-06, + "loss": 0.9042, + "step": 581 + }, + { + "epoch": 0.8422575976845152, + "grad_norm": 1.451400876045227, + "learning_rate": 3.274336283185841e-06, + "loss": 0.9005, + "step": 582 + }, + { + "epoch": 0.8437047756874095, + "grad_norm": 1.357790470123291, + "learning_rate": 3.270313757039421e-06, + "loss": 0.9463, + "step": 583 + }, + { + "epoch": 0.8451519536903039, + "grad_norm": 1.4770903587341309, + "learning_rate": 3.266291230893001e-06, + "loss": 0.9599, + "step": 584 + }, + { + "epoch": 0.8465991316931982, + "grad_norm": 1.3903440237045288, + "learning_rate": 3.2622687047465813e-06, + "loss": 0.9017, + "step": 585 + }, + { + "epoch": 0.8480463096960926, + "grad_norm": 1.508465051651001, + "learning_rate": 3.2582461786001614e-06, + "loss": 0.9376, + "step": 586 + }, + { + "epoch": 0.849493487698987, + "grad_norm": 1.352357029914856, + "learning_rate": 3.254223652453741e-06, + "loss": 0.9077, + "step": 587 + }, + { + "epoch": 0.8509406657018813, + "grad_norm": 1.53770112991333, + "learning_rate": 3.2502011263073212e-06, + "loss": 0.9418, + "step": 588 + }, + { + "epoch": 0.8523878437047757, + "grad_norm": 1.4406723976135254, + "learning_rate": 3.2461786001609013e-06, + "loss": 0.9408, + "step": 589 + }, + { + "epoch": 0.85383502170767, + "grad_norm": 1.3737328052520752, + "learning_rate": 3.2421560740144815e-06, + "loss": 0.9574, + "step": 590 + }, + { + "epoch": 0.85383502170767, + "eval_loss": 1.0164623260498047, + "eval_runtime": 25.6732, + "eval_samples_per_second": 38.951, + "eval_steps_per_second": 2.454, + "step": 590 + }, + { + "epoch": 0.8552821997105644, + "grad_norm": 1.4223228693008423, + "learning_rate": 3.2381335478680616e-06, + "loss": 0.9165, + "step": 591 + }, + { + "epoch": 0.8567293777134588, + "grad_norm": 1.372409462928772, + "learning_rate": 3.2341110217216413e-06, + "loss": 0.9262, + "step": 592 + }, + { + "epoch": 0.8581765557163531, + "grad_norm": 1.4255372285842896, + "learning_rate": 3.2300884955752214e-06, + "loss": 0.9518, + "step": 593 + }, + { + "epoch": 0.8596237337192475, + "grad_norm": 1.3796833753585815, + "learning_rate": 3.2260659694288016e-06, + "loss": 0.9034, + "step": 594 + }, + { + "epoch": 0.8610709117221418, + "grad_norm": 1.4612222909927368, + "learning_rate": 3.2220434432823817e-06, + "loss": 0.9126, + "step": 595 + }, + { + "epoch": 0.8625180897250362, + "grad_norm": 1.4215260744094849, + "learning_rate": 3.218020917135962e-06, + "loss": 0.929, + "step": 596 + }, + { + "epoch": 0.8639652677279306, + "grad_norm": 1.493041753768921, + "learning_rate": 3.2139983909895415e-06, + "loss": 0.9343, + "step": 597 + }, + { + "epoch": 0.8654124457308249, + "grad_norm": 1.4106526374816895, + "learning_rate": 3.2099758648431216e-06, + "loss": 0.9249, + "step": 598 + }, + { + "epoch": 0.8668596237337193, + "grad_norm": 1.5474896430969238, + "learning_rate": 3.2059533386967018e-06, + "loss": 0.9114, + "step": 599 + }, + { + "epoch": 0.8683068017366136, + "grad_norm": 1.4770991802215576, + "learning_rate": 3.201930812550282e-06, + "loss": 0.919, + "step": 600 + }, + { + "epoch": 0.8683068017366136, + "eval_loss": 1.0175906419754028, + "eval_runtime": 25.7685, + "eval_samples_per_second": 38.807, + "eval_steps_per_second": 2.445, + "step": 600 + }, + { + "epoch": 0.869753979739508, + "grad_norm": 1.4260352849960327, + "learning_rate": 3.197908286403862e-06, + "loss": 0.935, + "step": 601 + }, + { + "epoch": 0.8712011577424024, + "grad_norm": 1.4157609939575195, + "learning_rate": 3.1938857602574417e-06, + "loss": 0.9046, + "step": 602 + }, + { + "epoch": 0.8726483357452967, + "grad_norm": 1.4820977449417114, + "learning_rate": 3.189863234111022e-06, + "loss": 0.9363, + "step": 603 + }, + { + "epoch": 0.874095513748191, + "grad_norm": 1.437635064125061, + "learning_rate": 3.185840707964602e-06, + "loss": 0.906, + "step": 604 + }, + { + "epoch": 0.8755426917510853, + "grad_norm": 1.4991382360458374, + "learning_rate": 3.181818181818182e-06, + "loss": 0.921, + "step": 605 + }, + { + "epoch": 0.8769898697539797, + "grad_norm": 1.4861739873886108, + "learning_rate": 3.1777956556717622e-06, + "loss": 0.9135, + "step": 606 + }, + { + "epoch": 0.8784370477568741, + "grad_norm": 1.3911547660827637, + "learning_rate": 3.173773129525342e-06, + "loss": 0.8854, + "step": 607 + }, + { + "epoch": 0.8798842257597684, + "grad_norm": 1.3838648796081543, + "learning_rate": 3.169750603378922e-06, + "loss": 0.8957, + "step": 608 + }, + { + "epoch": 0.8813314037626628, + "grad_norm": 1.507276177406311, + "learning_rate": 3.165728077232502e-06, + "loss": 0.9171, + "step": 609 + }, + { + "epoch": 0.8827785817655571, + "grad_norm": 1.3060482740402222, + "learning_rate": 3.1617055510860823e-06, + "loss": 0.8788, + "step": 610 + }, + { + "epoch": 0.8827785817655571, + "eval_loss": 1.0178226232528687, + "eval_runtime": 26.0182, + "eval_samples_per_second": 38.435, + "eval_steps_per_second": 2.421, + "step": 610 + }, + { + "epoch": 0.8842257597684515, + "grad_norm": 1.5606573820114136, + "learning_rate": 3.1576830249396625e-06, + "loss": 0.909, + "step": 611 + }, + { + "epoch": 0.8856729377713459, + "grad_norm": 1.4704793691635132, + "learning_rate": 3.153660498793242e-06, + "loss": 0.9206, + "step": 612 + }, + { + "epoch": 0.8871201157742402, + "grad_norm": 1.2905973196029663, + "learning_rate": 3.1496379726468223e-06, + "loss": 0.9362, + "step": 613 + }, + { + "epoch": 0.8885672937771346, + "grad_norm": 1.4948153495788574, + "learning_rate": 3.1456154465004024e-06, + "loss": 0.8864, + "step": 614 + }, + { + "epoch": 0.8900144717800289, + "grad_norm": 1.450865626335144, + "learning_rate": 3.1415929203539825e-06, + "loss": 0.9162, + "step": 615 + }, + { + "epoch": 0.8914616497829233, + "grad_norm": 1.5478347539901733, + "learning_rate": 3.1375703942075627e-06, + "loss": 0.9319, + "step": 616 + }, + { + "epoch": 0.8929088277858177, + "grad_norm": 1.5064746141433716, + "learning_rate": 3.1335478680611424e-06, + "loss": 0.917, + "step": 617 + }, + { + "epoch": 0.894356005788712, + "grad_norm": 1.4333183765411377, + "learning_rate": 3.1295253419147225e-06, + "loss": 0.9227, + "step": 618 + }, + { + "epoch": 0.8958031837916064, + "grad_norm": 1.5096293687820435, + "learning_rate": 3.1255028157683026e-06, + "loss": 0.9406, + "step": 619 + }, + { + "epoch": 0.8972503617945007, + "grad_norm": 1.5495786666870117, + "learning_rate": 3.1214802896218828e-06, + "loss": 0.9406, + "step": 620 + }, + { + "epoch": 0.8972503617945007, + "eval_loss": 1.016027808189392, + "eval_runtime": 26.0019, + "eval_samples_per_second": 38.459, + "eval_steps_per_second": 2.423, + "step": 620 + }, + { + "epoch": 0.8986975397973951, + "grad_norm": 1.4580657482147217, + "learning_rate": 3.117457763475463e-06, + "loss": 0.9445, + "step": 621 + }, + { + "epoch": 0.9001447178002895, + "grad_norm": 1.4101994037628174, + "learning_rate": 3.1134352373290426e-06, + "loss": 0.9345, + "step": 622 + }, + { + "epoch": 0.9015918958031838, + "grad_norm": 1.4174447059631348, + "learning_rate": 3.1094127111826227e-06, + "loss": 0.9529, + "step": 623 + }, + { + "epoch": 0.9030390738060782, + "grad_norm": 1.399423360824585, + "learning_rate": 3.105390185036203e-06, + "loss": 0.9062, + "step": 624 + }, + { + "epoch": 0.9044862518089725, + "grad_norm": 1.332127571105957, + "learning_rate": 3.101367658889783e-06, + "loss": 0.887, + "step": 625 + }, + { + "epoch": 0.9059334298118669, + "grad_norm": 1.4557148218154907, + "learning_rate": 3.097345132743363e-06, + "loss": 0.928, + "step": 626 + }, + { + "epoch": 0.9073806078147613, + "grad_norm": 1.4364054203033447, + "learning_rate": 3.093322606596943e-06, + "loss": 0.952, + "step": 627 + }, + { + "epoch": 0.9088277858176556, + "grad_norm": 1.4176812171936035, + "learning_rate": 3.089300080450523e-06, + "loss": 0.9103, + "step": 628 + }, + { + "epoch": 0.91027496382055, + "grad_norm": 1.4091321229934692, + "learning_rate": 3.085277554304103e-06, + "loss": 0.9195, + "step": 629 + }, + { + "epoch": 0.9117221418234442, + "grad_norm": 1.4213095903396606, + "learning_rate": 3.081255028157683e-06, + "loss": 0.8743, + "step": 630 + }, + { + "epoch": 0.9117221418234442, + "eval_loss": 1.012543797492981, + "eval_runtime": 25.909, + "eval_samples_per_second": 38.597, + "eval_steps_per_second": 2.432, + "step": 630 + }, + { + "epoch": 0.9131693198263386, + "grad_norm": 1.4346204996109009, + "learning_rate": 3.0772325020112633e-06, + "loss": 0.954, + "step": 631 + }, + { + "epoch": 0.914616497829233, + "grad_norm": 1.3710834980010986, + "learning_rate": 3.073209975864843e-06, + "loss": 0.9346, + "step": 632 + }, + { + "epoch": 0.9160636758321273, + "grad_norm": 1.4266773462295532, + "learning_rate": 3.069187449718423e-06, + "loss": 0.9253, + "step": 633 + }, + { + "epoch": 0.9175108538350217, + "grad_norm": 1.4468302726745605, + "learning_rate": 3.0651649235720033e-06, + "loss": 0.905, + "step": 634 + }, + { + "epoch": 0.918958031837916, + "grad_norm": 1.3775814771652222, + "learning_rate": 3.0611423974255834e-06, + "loss": 0.9227, + "step": 635 + }, + { + "epoch": 0.9204052098408104, + "grad_norm": 1.4380732774734497, + "learning_rate": 3.057119871279163e-06, + "loss": 0.9128, + "step": 636 + }, + { + "epoch": 0.9218523878437048, + "grad_norm": 1.4668720960617065, + "learning_rate": 3.0530973451327432e-06, + "loss": 0.9111, + "step": 637 + }, + { + "epoch": 0.9232995658465991, + "grad_norm": 1.5741231441497803, + "learning_rate": 3.0490748189863234e-06, + "loss": 0.9715, + "step": 638 + }, + { + "epoch": 0.9247467438494935, + "grad_norm": 1.44276762008667, + "learning_rate": 3.0450522928399035e-06, + "loss": 0.9382, + "step": 639 + }, + { + "epoch": 0.9261939218523878, + "grad_norm": 1.5068806409835815, + "learning_rate": 3.0410297666934836e-06, + "loss": 0.9328, + "step": 640 + }, + { + "epoch": 0.9261939218523878, + "eval_loss": 1.0154249668121338, + "eval_runtime": 25.7492, + "eval_samples_per_second": 38.836, + "eval_steps_per_second": 2.447, + "step": 640 + }, + { + "epoch": 0.9276410998552822, + "grad_norm": 1.437995195388794, + "learning_rate": 3.0370072405470633e-06, + "loss": 0.9104, + "step": 641 + }, + { + "epoch": 0.9290882778581766, + "grad_norm": 1.6219338178634644, + "learning_rate": 3.0329847144006443e-06, + "loss": 0.9355, + "step": 642 + }, + { + "epoch": 0.9305354558610709, + "grad_norm": 1.4921026229858398, + "learning_rate": 3.028962188254224e-06, + "loss": 0.9298, + "step": 643 + }, + { + "epoch": 0.9319826338639653, + "grad_norm": 1.3929303884506226, + "learning_rate": 3.024939662107804e-06, + "loss": 0.9133, + "step": 644 + }, + { + "epoch": 0.9334298118668596, + "grad_norm": 1.4663861989974976, + "learning_rate": 3.0209171359613843e-06, + "loss": 0.9199, + "step": 645 + }, + { + "epoch": 0.934876989869754, + "grad_norm": 1.4335482120513916, + "learning_rate": 3.0168946098149644e-06, + "loss": 0.8951, + "step": 646 + }, + { + "epoch": 0.9363241678726484, + "grad_norm": 1.4543933868408203, + "learning_rate": 3.0128720836685445e-06, + "loss": 0.9221, + "step": 647 + }, + { + "epoch": 0.9377713458755427, + "grad_norm": 1.3568743467330933, + "learning_rate": 3.0088495575221242e-06, + "loss": 0.8988, + "step": 648 + }, + { + "epoch": 0.9392185238784371, + "grad_norm": 1.497255563735962, + "learning_rate": 3.0048270313757043e-06, + "loss": 0.9321, + "step": 649 + }, + { + "epoch": 0.9406657018813314, + "grad_norm": 1.3754018545150757, + "learning_rate": 3.0008045052292845e-06, + "loss": 0.906, + "step": 650 + }, + { + "epoch": 0.9406657018813314, + "eval_loss": 1.0140572786331177, + "eval_runtime": 26.0118, + "eval_samples_per_second": 38.444, + "eval_steps_per_second": 2.422, + "step": 650 + }, + { + "epoch": 0.9421128798842258, + "grad_norm": 1.3779268264770508, + "learning_rate": 2.9967819790828646e-06, + "loss": 0.9066, + "step": 651 + }, + { + "epoch": 0.9435600578871202, + "grad_norm": 1.4161144495010376, + "learning_rate": 2.9927594529364447e-06, + "loss": 0.8982, + "step": 652 + }, + { + "epoch": 0.9450072358900145, + "grad_norm": 1.4134407043457031, + "learning_rate": 2.9887369267900244e-06, + "loss": 0.9157, + "step": 653 + }, + { + "epoch": 0.9464544138929089, + "grad_norm": 1.4312589168548584, + "learning_rate": 2.9847144006436046e-06, + "loss": 0.9232, + "step": 654 + }, + { + "epoch": 0.9479015918958031, + "grad_norm": 1.521687388420105, + "learning_rate": 2.9806918744971847e-06, + "loss": 0.9489, + "step": 655 + }, + { + "epoch": 0.9493487698986975, + "grad_norm": 1.5733003616333008, + "learning_rate": 2.976669348350765e-06, + "loss": 0.9585, + "step": 656 + }, + { + "epoch": 0.9507959479015919, + "grad_norm": 1.4925999641418457, + "learning_rate": 2.972646822204345e-06, + "loss": 0.8928, + "step": 657 + }, + { + "epoch": 0.9522431259044862, + "grad_norm": 1.4285452365875244, + "learning_rate": 2.9686242960579246e-06, + "loss": 0.8946, + "step": 658 + }, + { + "epoch": 0.9536903039073806, + "grad_norm": 1.4919170141220093, + "learning_rate": 2.9646017699115048e-06, + "loss": 0.9162, + "step": 659 + }, + { + "epoch": 0.9551374819102749, + "grad_norm": 1.4638007879257202, + "learning_rate": 2.960579243765085e-06, + "loss": 0.9025, + "step": 660 + }, + { + "epoch": 0.9551374819102749, + "eval_loss": 1.0097233057022095, + "eval_runtime": 25.6834, + "eval_samples_per_second": 38.936, + "eval_steps_per_second": 2.453, + "step": 660 + }, + { + "epoch": 0.9565846599131693, + "grad_norm": 1.4907405376434326, + "learning_rate": 2.956556717618665e-06, + "loss": 0.9428, + "step": 661 + }, + { + "epoch": 0.9580318379160637, + "grad_norm": 1.4305118322372437, + "learning_rate": 2.952534191472245e-06, + "loss": 0.9007, + "step": 662 + }, + { + "epoch": 0.959479015918958, + "grad_norm": 1.401001214981079, + "learning_rate": 2.948511665325825e-06, + "loss": 0.908, + "step": 663 + }, + { + "epoch": 0.9609261939218524, + "grad_norm": 1.4340101480484009, + "learning_rate": 2.944489139179405e-06, + "loss": 0.9302, + "step": 664 + }, + { + "epoch": 0.9623733719247467, + "grad_norm": 1.4313371181488037, + "learning_rate": 2.940466613032985e-06, + "loss": 0.852, + "step": 665 + }, + { + "epoch": 0.9638205499276411, + "grad_norm": 1.6063755750656128, + "learning_rate": 2.9364440868865652e-06, + "loss": 0.9331, + "step": 666 + }, + { + "epoch": 0.9652677279305355, + "grad_norm": 1.48060142993927, + "learning_rate": 2.9324215607401454e-06, + "loss": 0.9181, + "step": 667 + }, + { + "epoch": 0.9667149059334298, + "grad_norm": 1.4865727424621582, + "learning_rate": 2.928399034593725e-06, + "loss": 0.9356, + "step": 668 + }, + { + "epoch": 0.9681620839363242, + "grad_norm": 1.409006953239441, + "learning_rate": 2.924376508447305e-06, + "loss": 0.9176, + "step": 669 + }, + { + "epoch": 0.9696092619392185, + "grad_norm": 1.411401629447937, + "learning_rate": 2.9203539823008853e-06, + "loss": 0.865, + "step": 670 + }, + { + "epoch": 0.9696092619392185, + "eval_loss": 1.007845401763916, + "eval_runtime": 25.8558, + "eval_samples_per_second": 38.676, + "eval_steps_per_second": 2.437, + "step": 670 + }, + { + "epoch": 0.9710564399421129, + "grad_norm": 1.4949239492416382, + "learning_rate": 2.9163314561544655e-06, + "loss": 0.9204, + "step": 671 + }, + { + "epoch": 0.9725036179450073, + "grad_norm": 1.4543801546096802, + "learning_rate": 2.912308930008045e-06, + "loss": 0.8932, + "step": 672 + }, + { + "epoch": 0.9739507959479016, + "grad_norm": 1.48538339138031, + "learning_rate": 2.9082864038616253e-06, + "loss": 0.9089, + "step": 673 + }, + { + "epoch": 0.975397973950796, + "grad_norm": 1.3665274381637573, + "learning_rate": 2.9042638777152054e-06, + "loss": 0.8793, + "step": 674 + }, + { + "epoch": 0.9768451519536903, + "grad_norm": 1.4270809888839722, + "learning_rate": 2.9002413515687855e-06, + "loss": 0.9028, + "step": 675 + }, + { + "epoch": 0.9782923299565847, + "grad_norm": 1.4889215230941772, + "learning_rate": 2.8962188254223657e-06, + "loss": 0.9161, + "step": 676 + }, + { + "epoch": 0.9797395079594791, + "grad_norm": 1.50188410282135, + "learning_rate": 2.8921962992759454e-06, + "loss": 0.871, + "step": 677 + }, + { + "epoch": 0.9811866859623734, + "grad_norm": 1.4788942337036133, + "learning_rate": 2.8881737731295255e-06, + "loss": 0.9241, + "step": 678 + }, + { + "epoch": 0.9826338639652678, + "grad_norm": 1.4699136018753052, + "learning_rate": 2.8841512469831056e-06, + "loss": 0.9242, + "step": 679 + }, + { + "epoch": 0.984081041968162, + "grad_norm": 1.4581975936889648, + "learning_rate": 2.8801287208366858e-06, + "loss": 0.9461, + "step": 680 + }, + { + "epoch": 0.984081041968162, + "eval_loss": 1.0047694444656372, + "eval_runtime": 25.6669, + "eval_samples_per_second": 38.961, + "eval_steps_per_second": 2.455, + "step": 680 + }, + { + "epoch": 0.9855282199710564, + "grad_norm": 1.3976376056671143, + "learning_rate": 2.876106194690266e-06, + "loss": 0.9247, + "step": 681 + }, + { + "epoch": 0.9869753979739508, + "grad_norm": 1.4753456115722656, + "learning_rate": 2.8720836685438456e-06, + "loss": 0.8744, + "step": 682 + }, + { + "epoch": 0.9884225759768451, + "grad_norm": 1.3396860361099243, + "learning_rate": 2.8680611423974257e-06, + "loss": 0.9268, + "step": 683 + }, + { + "epoch": 0.9898697539797395, + "grad_norm": 1.41839599609375, + "learning_rate": 2.864038616251006e-06, + "loss": 0.8972, + "step": 684 + }, + { + "epoch": 0.9913169319826338, + "grad_norm": 1.5397756099700928, + "learning_rate": 2.860016090104586e-06, + "loss": 0.9036, + "step": 685 + }, + { + "epoch": 0.9927641099855282, + "grad_norm": 1.458146333694458, + "learning_rate": 2.855993563958166e-06, + "loss": 0.924, + "step": 686 + }, + { + "epoch": 0.9942112879884226, + "grad_norm": 1.3859862089157104, + "learning_rate": 2.851971037811746e-06, + "loss": 0.9045, + "step": 687 + }, + { + "epoch": 0.9956584659913169, + "grad_norm": 1.5419162511825562, + "learning_rate": 2.847948511665326e-06, + "loss": 0.8943, + "step": 688 + }, + { + "epoch": 0.9971056439942113, + "grad_norm": 1.4252957105636597, + "learning_rate": 2.843925985518906e-06, + "loss": 0.9126, + "step": 689 + }, + { + "epoch": 0.9985528219971056, + "grad_norm": 1.5327415466308594, + "learning_rate": 2.839903459372486e-06, + "loss": 0.9356, + "step": 690 + }, + { + "epoch": 0.9985528219971056, + "eval_loss": 1.0072345733642578, + "eval_runtime": 25.8176, + "eval_samples_per_second": 38.733, + "eval_steps_per_second": 2.44, + "step": 690 + }, + { + "epoch": 1.0, + "grad_norm": 1.4523152112960815, + "learning_rate": 2.8358809332260663e-06, + "loss": 0.8922, + "step": 691 + }, + { + "epoch": 1.0014471780028944, + "grad_norm": 1.4414013624191284, + "learning_rate": 2.831858407079646e-06, + "loss": 0.8817, + "step": 692 + }, + { + "epoch": 1.0028943560057888, + "grad_norm": 1.4955841302871704, + "learning_rate": 2.827835880933226e-06, + "loss": 0.8755, + "step": 693 + }, + { + "epoch": 1.004341534008683, + "grad_norm": 1.4577631950378418, + "learning_rate": 2.8238133547868063e-06, + "loss": 0.916, + "step": 694 + }, + { + "epoch": 1.0057887120115774, + "grad_norm": 1.5309616327285767, + "learning_rate": 2.8197908286403864e-06, + "loss": 0.9334, + "step": 695 + }, + { + "epoch": 1.0072358900144718, + "grad_norm": 1.678214430809021, + "learning_rate": 2.8157683024939665e-06, + "loss": 0.9174, + "step": 696 + }, + { + "epoch": 1.0086830680173662, + "grad_norm": 1.4631013870239258, + "learning_rate": 2.8117457763475462e-06, + "loss": 0.8904, + "step": 697 + }, + { + "epoch": 1.0101302460202606, + "grad_norm": 1.5390706062316895, + "learning_rate": 2.8077232502011264e-06, + "loss": 0.8739, + "step": 698 + }, + { + "epoch": 1.0115774240231548, + "grad_norm": 1.4008797407150269, + "learning_rate": 2.8037007240547065e-06, + "loss": 0.8803, + "step": 699 + }, + { + "epoch": 1.0130246020260492, + "grad_norm": 1.5263433456420898, + "learning_rate": 2.7996781979082866e-06, + "loss": 0.9466, + "step": 700 + }, + { + "epoch": 1.0130246020260492, + "eval_loss": 1.0086634159088135, + "eval_runtime": 25.832, + "eval_samples_per_second": 38.712, + "eval_steps_per_second": 2.439, + "step": 700 + }, + { + "epoch": 1.0144717800289436, + "grad_norm": 1.528403639793396, + "learning_rate": 2.7956556717618667e-06, + "loss": 0.9317, + "step": 701 + }, + { + "epoch": 1.015918958031838, + "grad_norm": 1.6002906560897827, + "learning_rate": 2.7916331456154465e-06, + "loss": 0.9359, + "step": 702 + }, + { + "epoch": 1.0173661360347324, + "grad_norm": 1.6118007898330688, + "learning_rate": 2.7876106194690266e-06, + "loss": 0.9234, + "step": 703 + }, + { + "epoch": 1.0188133140376265, + "grad_norm": 1.5051100254058838, + "learning_rate": 2.7835880933226067e-06, + "loss": 0.9066, + "step": 704 + }, + { + "epoch": 1.020260492040521, + "grad_norm": 1.5129306316375732, + "learning_rate": 2.779565567176187e-06, + "loss": 0.9406, + "step": 705 + }, + { + "epoch": 1.0217076700434153, + "grad_norm": 1.5781571865081787, + "learning_rate": 2.775543041029767e-06, + "loss": 0.8667, + "step": 706 + }, + { + "epoch": 1.0231548480463097, + "grad_norm": 1.4130151271820068, + "learning_rate": 2.7715205148833467e-06, + "loss": 0.9081, + "step": 707 + }, + { + "epoch": 1.0246020260492041, + "grad_norm": 1.5489847660064697, + "learning_rate": 2.767497988736927e-06, + "loss": 0.9474, + "step": 708 + }, + { + "epoch": 1.0260492040520983, + "grad_norm": 1.490387201309204, + "learning_rate": 2.763475462590507e-06, + "loss": 0.8583, + "step": 709 + }, + { + "epoch": 1.0274963820549927, + "grad_norm": 1.6409109830856323, + "learning_rate": 2.759452936444087e-06, + "loss": 0.8878, + "step": 710 + }, + { + "epoch": 1.0274963820549927, + "eval_loss": 1.0066230297088623, + "eval_runtime": 25.8463, + "eval_samples_per_second": 38.69, + "eval_steps_per_second": 2.437, + "step": 710 + }, + { + "epoch": 1.0289435600578871, + "grad_norm": 1.537100076675415, + "learning_rate": 2.755430410297667e-06, + "loss": 0.9095, + "step": 711 + }, + { + "epoch": 1.0303907380607815, + "grad_norm": 1.5319218635559082, + "learning_rate": 2.751407884151247e-06, + "loss": 0.9023, + "step": 712 + }, + { + "epoch": 1.031837916063676, + "grad_norm": 1.5210882425308228, + "learning_rate": 2.747385358004827e-06, + "loss": 0.9183, + "step": 713 + }, + { + "epoch": 1.03328509406657, + "grad_norm": 1.5327351093292236, + "learning_rate": 2.743362831858407e-06, + "loss": 0.8974, + "step": 714 + }, + { + "epoch": 1.0347322720694645, + "grad_norm": 1.487595558166504, + "learning_rate": 2.7393403057119873e-06, + "loss": 0.9136, + "step": 715 + }, + { + "epoch": 1.036179450072359, + "grad_norm": 1.615623116493225, + "learning_rate": 2.7353177795655674e-06, + "loss": 0.9088, + "step": 716 + }, + { + "epoch": 1.0376266280752533, + "grad_norm": 1.4948785305023193, + "learning_rate": 2.731295253419147e-06, + "loss": 0.8943, + "step": 717 + }, + { + "epoch": 1.0390738060781477, + "grad_norm": 1.3898569345474243, + "learning_rate": 2.7272727272727272e-06, + "loss": 0.8781, + "step": 718 + }, + { + "epoch": 1.0405209840810419, + "grad_norm": 1.607021450996399, + "learning_rate": 2.7232502011263074e-06, + "loss": 0.8876, + "step": 719 + }, + { + "epoch": 1.0419681620839363, + "grad_norm": 1.6453396081924438, + "learning_rate": 2.7192276749798875e-06, + "loss": 0.919, + "step": 720 + }, + { + "epoch": 1.0419681620839363, + "eval_loss": 1.0065933465957642, + "eval_runtime": 25.7628, + "eval_samples_per_second": 38.816, + "eval_steps_per_second": 2.445, + "step": 720 + }, + { + "epoch": 1.0434153400868307, + "grad_norm": 1.5655184984207153, + "learning_rate": 2.7152051488334676e-06, + "loss": 0.885, + "step": 721 + }, + { + "epoch": 1.044862518089725, + "grad_norm": 1.4670000076293945, + "learning_rate": 2.7111826226870473e-06, + "loss": 0.8701, + "step": 722 + }, + { + "epoch": 1.0463096960926195, + "grad_norm": 1.4562489986419678, + "learning_rate": 2.7071600965406274e-06, + "loss": 0.9035, + "step": 723 + }, + { + "epoch": 1.0477568740955137, + "grad_norm": 1.5567082166671753, + "learning_rate": 2.7031375703942076e-06, + "loss": 0.8974, + "step": 724 + }, + { + "epoch": 1.049204052098408, + "grad_norm": 1.4280647039413452, + "learning_rate": 2.6991150442477877e-06, + "loss": 0.8588, + "step": 725 + }, + { + "epoch": 1.0506512301013025, + "grad_norm": 1.4315814971923828, + "learning_rate": 2.695092518101368e-06, + "loss": 0.8464, + "step": 726 + }, + { + "epoch": 1.0520984081041969, + "grad_norm": 1.5172886848449707, + "learning_rate": 2.6910699919549475e-06, + "loss": 0.9014, + "step": 727 + }, + { + "epoch": 1.0535455861070913, + "grad_norm": 1.635730266571045, + "learning_rate": 2.6870474658085277e-06, + "loss": 0.8978, + "step": 728 + }, + { + "epoch": 1.0549927641099854, + "grad_norm": 1.5512837171554565, + "learning_rate": 2.6830249396621078e-06, + "loss": 0.9593, + "step": 729 + }, + { + "epoch": 1.0564399421128798, + "grad_norm": 1.5531891584396362, + "learning_rate": 2.679002413515688e-06, + "loss": 0.897, + "step": 730 + }, + { + "epoch": 1.0564399421128798, + "eval_loss": 1.00581693649292, + "eval_runtime": 25.6366, + "eval_samples_per_second": 39.007, + "eval_steps_per_second": 2.457, + "step": 730 + }, + { + "epoch": 1.0578871201157742, + "grad_norm": 1.5719125270843506, + "learning_rate": 2.6749798873692685e-06, + "loss": 0.8763, + "step": 731 + }, + { + "epoch": 1.0593342981186686, + "grad_norm": 1.559281826019287, + "learning_rate": 2.6709573612228486e-06, + "loss": 0.9001, + "step": 732 + }, + { + "epoch": 1.060781476121563, + "grad_norm": 1.7323325872421265, + "learning_rate": 2.6669348350764283e-06, + "loss": 0.9084, + "step": 733 + }, + { + "epoch": 1.0622286541244572, + "grad_norm": 1.6019538640975952, + "learning_rate": 2.6629123089300084e-06, + "loss": 0.9016, + "step": 734 + }, + { + "epoch": 1.0636758321273516, + "grad_norm": 1.521162748336792, + "learning_rate": 2.6588897827835885e-06, + "loss": 0.8473, + "step": 735 + }, + { + "epoch": 1.065123010130246, + "grad_norm": 1.58328115940094, + "learning_rate": 2.6548672566371687e-06, + "loss": 0.8973, + "step": 736 + }, + { + "epoch": 1.0665701881331404, + "grad_norm": 1.6900322437286377, + "learning_rate": 2.650844730490749e-06, + "loss": 0.8908, + "step": 737 + }, + { + "epoch": 1.0680173661360348, + "grad_norm": 1.6272144317626953, + "learning_rate": 2.6468222043443285e-06, + "loss": 0.9089, + "step": 738 + }, + { + "epoch": 1.069464544138929, + "grad_norm": 1.6188445091247559, + "learning_rate": 2.6427996781979086e-06, + "loss": 0.938, + "step": 739 + }, + { + "epoch": 1.0709117221418234, + "grad_norm": 1.6111786365509033, + "learning_rate": 2.6387771520514888e-06, + "loss": 0.9106, + "step": 740 + }, + { + "epoch": 1.0709117221418234, + "eval_loss": 1.0052382946014404, + "eval_runtime": 25.7043, + "eval_samples_per_second": 38.904, + "eval_steps_per_second": 2.451, + "step": 740 + }, + { + "epoch": 1.0723589001447178, + "grad_norm": 1.5151225328445435, + "learning_rate": 2.634754625905069e-06, + "loss": 0.8542, + "step": 741 + }, + { + "epoch": 1.0738060781476122, + "grad_norm": 1.5061923265457153, + "learning_rate": 2.630732099758649e-06, + "loss": 0.883, + "step": 742 + }, + { + "epoch": 1.0752532561505066, + "grad_norm": 1.5270133018493652, + "learning_rate": 2.6267095736122287e-06, + "loss": 0.8941, + "step": 743 + }, + { + "epoch": 1.0767004341534008, + "grad_norm": 1.5916814804077148, + "learning_rate": 2.622687047465809e-06, + "loss": 0.8985, + "step": 744 + }, + { + "epoch": 1.0781476121562952, + "grad_norm": 1.4444690942764282, + "learning_rate": 2.618664521319389e-06, + "loss": 0.8818, + "step": 745 + }, + { + "epoch": 1.0795947901591896, + "grad_norm": 1.4805785417556763, + "learning_rate": 2.614641995172969e-06, + "loss": 0.9, + "step": 746 + }, + { + "epoch": 1.081041968162084, + "grad_norm": 1.4908889532089233, + "learning_rate": 2.6106194690265492e-06, + "loss": 0.912, + "step": 747 + }, + { + "epoch": 1.0824891461649784, + "grad_norm": 1.5267603397369385, + "learning_rate": 2.606596942880129e-06, + "loss": 0.938, + "step": 748 + }, + { + "epoch": 1.0839363241678726, + "grad_norm": 1.4756078720092773, + "learning_rate": 2.602574416733709e-06, + "loss": 0.8712, + "step": 749 + }, + { + "epoch": 1.085383502170767, + "grad_norm": 1.508716106414795, + "learning_rate": 2.598551890587289e-06, + "loss": 0.9001, + "step": 750 + }, + { + "epoch": 1.085383502170767, + "eval_loss": 1.0030596256256104, + "eval_runtime": 25.5438, + "eval_samples_per_second": 39.148, + "eval_steps_per_second": 2.466, + "step": 750 + }, + { + "epoch": 1.0868306801736614, + "grad_norm": 1.5817303657531738, + "learning_rate": 2.5945293644408693e-06, + "loss": 0.9058, + "step": 751 + }, + { + "epoch": 1.0882778581765558, + "grad_norm": 1.497726321220398, + "learning_rate": 2.5905068382944494e-06, + "loss": 0.8961, + "step": 752 + }, + { + "epoch": 1.0897250361794502, + "grad_norm": 1.5296815633773804, + "learning_rate": 2.586484312148029e-06, + "loss": 0.9155, + "step": 753 + }, + { + "epoch": 1.0911722141823443, + "grad_norm": 1.609724760055542, + "learning_rate": 2.5824617860016093e-06, + "loss": 0.9105, + "step": 754 + }, + { + "epoch": 1.0926193921852387, + "grad_norm": 1.4852569103240967, + "learning_rate": 2.5784392598551894e-06, + "loss": 0.923, + "step": 755 + }, + { + "epoch": 1.0940665701881331, + "grad_norm": 1.5415880680084229, + "learning_rate": 2.5744167337087695e-06, + "loss": 0.8643, + "step": 756 + }, + { + "epoch": 1.0955137481910275, + "grad_norm": 1.465867280960083, + "learning_rate": 2.5703942075623497e-06, + "loss": 0.913, + "step": 757 + }, + { + "epoch": 1.096960926193922, + "grad_norm": 1.576643943786621, + "learning_rate": 2.5663716814159294e-06, + "loss": 0.8944, + "step": 758 + }, + { + "epoch": 1.0984081041968161, + "grad_norm": 1.6592824459075928, + "learning_rate": 2.5623491552695095e-06, + "loss": 0.8829, + "step": 759 + }, + { + "epoch": 1.0998552821997105, + "grad_norm": 1.6082524061203003, + "learning_rate": 2.5583266291230896e-06, + "loss": 0.8653, + "step": 760 + }, + { + "epoch": 1.0998552821997105, + "eval_loss": 1.0026334524154663, + "eval_runtime": 25.607, + "eval_samples_per_second": 39.052, + "eval_steps_per_second": 2.46, + "step": 760 + }, + { + "epoch": 1.101302460202605, + "grad_norm": 1.52219557762146, + "learning_rate": 2.5543041029766698e-06, + "loss": 0.8905, + "step": 761 + }, + { + "epoch": 1.1027496382054993, + "grad_norm": 1.6422560214996338, + "learning_rate": 2.55028157683025e-06, + "loss": 0.8654, + "step": 762 + }, + { + "epoch": 1.1041968162083937, + "grad_norm": 1.6546605825424194, + "learning_rate": 2.5462590506838296e-06, + "loss": 0.8655, + "step": 763 + }, + { + "epoch": 1.105643994211288, + "grad_norm": 1.5878373384475708, + "learning_rate": 2.5422365245374097e-06, + "loss": 0.9194, + "step": 764 + }, + { + "epoch": 1.1070911722141823, + "grad_norm": 1.5256876945495605, + "learning_rate": 2.53821399839099e-06, + "loss": 0.891, + "step": 765 + }, + { + "epoch": 1.1085383502170767, + "grad_norm": 1.5687922239303589, + "learning_rate": 2.53419147224457e-06, + "loss": 0.8885, + "step": 766 + }, + { + "epoch": 1.109985528219971, + "grad_norm": 1.6002429723739624, + "learning_rate": 2.53016894609815e-06, + "loss": 0.8896, + "step": 767 + }, + { + "epoch": 1.1114327062228655, + "grad_norm": 1.5805329084396362, + "learning_rate": 2.52614641995173e-06, + "loss": 0.9043, + "step": 768 + }, + { + "epoch": 1.1128798842257597, + "grad_norm": 1.5537420511245728, + "learning_rate": 2.52212389380531e-06, + "loss": 0.8831, + "step": 769 + }, + { + "epoch": 1.114327062228654, + "grad_norm": 1.6533360481262207, + "learning_rate": 2.51810136765889e-06, + "loss": 0.9479, + "step": 770 + }, + { + "epoch": 1.114327062228654, + "eval_loss": 1.002786636352539, + "eval_runtime": 25.8282, + "eval_samples_per_second": 38.717, + "eval_steps_per_second": 2.439, + "step": 770 + }, + { + "epoch": 1.1157742402315485, + "grad_norm": 1.6494132280349731, + "learning_rate": 2.51407884151247e-06, + "loss": 0.895, + "step": 771 + }, + { + "epoch": 1.1172214182344429, + "grad_norm": 1.5635745525360107, + "learning_rate": 2.5100563153660503e-06, + "loss": 0.8859, + "step": 772 + }, + { + "epoch": 1.1186685962373373, + "grad_norm": 1.501560091972351, + "learning_rate": 2.50603378921963e-06, + "loss": 0.9423, + "step": 773 + }, + { + "epoch": 1.1201157742402315, + "grad_norm": 1.5016065835952759, + "learning_rate": 2.50201126307321e-06, + "loss": 0.9226, + "step": 774 + }, + { + "epoch": 1.1215629522431259, + "grad_norm": 1.5102709531784058, + "learning_rate": 2.4979887369267903e-06, + "loss": 0.889, + "step": 775 + }, + { + "epoch": 1.1230101302460203, + "grad_norm": 1.4280509948730469, + "learning_rate": 2.4939662107803704e-06, + "loss": 0.8936, + "step": 776 + }, + { + "epoch": 1.1244573082489147, + "grad_norm": 1.5048205852508545, + "learning_rate": 2.48994368463395e-06, + "loss": 0.9029, + "step": 777 + }, + { + "epoch": 1.125904486251809, + "grad_norm": 1.5334604978561401, + "learning_rate": 2.4859211584875302e-06, + "loss": 0.8864, + "step": 778 + }, + { + "epoch": 1.1273516642547032, + "grad_norm": 1.6037338972091675, + "learning_rate": 2.4818986323411104e-06, + "loss": 0.921, + "step": 779 + }, + { + "epoch": 1.1287988422575976, + "grad_norm": 1.6144342422485352, + "learning_rate": 2.4778761061946905e-06, + "loss": 0.9006, + "step": 780 + }, + { + "epoch": 1.1287988422575976, + "eval_loss": 1.0016281604766846, + "eval_runtime": 25.8012, + "eval_samples_per_second": 38.758, + "eval_steps_per_second": 2.442, + "step": 780 + }, + { + "epoch": 1.130246020260492, + "grad_norm": 1.593278408050537, + "learning_rate": 2.4738535800482706e-06, + "loss": 0.8724, + "step": 781 + }, + { + "epoch": 1.1316931982633864, + "grad_norm": 1.5138260126113892, + "learning_rate": 2.4698310539018503e-06, + "loss": 0.9093, + "step": 782 + }, + { + "epoch": 1.1331403762662808, + "grad_norm": 1.5158487558364868, + "learning_rate": 2.4658085277554304e-06, + "loss": 0.9086, + "step": 783 + }, + { + "epoch": 1.134587554269175, + "grad_norm": 1.5141456127166748, + "learning_rate": 2.4617860016090106e-06, + "loss": 0.8756, + "step": 784 + }, + { + "epoch": 1.1360347322720694, + "grad_norm": 1.5087177753448486, + "learning_rate": 2.4577634754625907e-06, + "loss": 0.9197, + "step": 785 + }, + { + "epoch": 1.1374819102749638, + "grad_norm": 1.5388543605804443, + "learning_rate": 2.453740949316171e-06, + "loss": 0.8808, + "step": 786 + }, + { + "epoch": 1.1389290882778582, + "grad_norm": 1.575091004371643, + "learning_rate": 2.4497184231697505e-06, + "loss": 0.9123, + "step": 787 + }, + { + "epoch": 1.1403762662807526, + "grad_norm": 1.5432230234146118, + "learning_rate": 2.4456958970233307e-06, + "loss": 0.8898, + "step": 788 + }, + { + "epoch": 1.1418234442836468, + "grad_norm": 1.5404096841812134, + "learning_rate": 2.4416733708769108e-06, + "loss": 0.8835, + "step": 789 + }, + { + "epoch": 1.1432706222865412, + "grad_norm": 1.7286585569381714, + "learning_rate": 2.437650844730491e-06, + "loss": 0.8938, + "step": 790 + }, + { + "epoch": 1.1432706222865412, + "eval_loss": 1.0016593933105469, + "eval_runtime": 25.7801, + "eval_samples_per_second": 38.79, + "eval_steps_per_second": 2.444, + "step": 790 + }, + { + "epoch": 1.1447178002894356, + "grad_norm": 1.693106770515442, + "learning_rate": 2.433628318584071e-06, + "loss": 0.9065, + "step": 791 + }, + { + "epoch": 1.14616497829233, + "grad_norm": 1.589040994644165, + "learning_rate": 2.4296057924376507e-06, + "loss": 0.8858, + "step": 792 + }, + { + "epoch": 1.1476121562952244, + "grad_norm": 1.5030606985092163, + "learning_rate": 2.425583266291231e-06, + "loss": 0.8924, + "step": 793 + }, + { + "epoch": 1.1490593342981186, + "grad_norm": 1.6685343980789185, + "learning_rate": 2.421560740144811e-06, + "loss": 0.9408, + "step": 794 + }, + { + "epoch": 1.150506512301013, + "grad_norm": 1.59011709690094, + "learning_rate": 2.417538213998391e-06, + "loss": 0.8711, + "step": 795 + }, + { + "epoch": 1.1519536903039074, + "grad_norm": 1.5271319150924683, + "learning_rate": 2.4135156878519713e-06, + "loss": 0.8951, + "step": 796 + }, + { + "epoch": 1.1534008683068018, + "grad_norm": 1.4878593683242798, + "learning_rate": 2.4094931617055514e-06, + "loss": 0.8683, + "step": 797 + }, + { + "epoch": 1.1548480463096962, + "grad_norm": 1.572616696357727, + "learning_rate": 2.4054706355591315e-06, + "loss": 0.8463, + "step": 798 + }, + { + "epoch": 1.1562952243125904, + "grad_norm": 1.5298666954040527, + "learning_rate": 2.4014481094127116e-06, + "loss": 0.891, + "step": 799 + }, + { + "epoch": 1.1577424023154848, + "grad_norm": 1.6578890085220337, + "learning_rate": 2.3974255832662913e-06, + "loss": 0.8952, + "step": 800 + }, + { + "epoch": 1.1577424023154848, + "eval_loss": 1.0011423826217651, + "eval_runtime": 25.786, + "eval_samples_per_second": 38.781, + "eval_steps_per_second": 2.443, + "step": 800 + }, + { + "epoch": 1.1591895803183792, + "grad_norm": 1.6832380294799805, + "learning_rate": 2.3934030571198715e-06, + "loss": 0.9434, + "step": 801 + }, + { + "epoch": 1.1606367583212736, + "grad_norm": 1.712808609008789, + "learning_rate": 2.3893805309734516e-06, + "loss": 0.8573, + "step": 802 + }, + { + "epoch": 1.162083936324168, + "grad_norm": 1.7583296298980713, + "learning_rate": 2.3853580048270317e-06, + "loss": 0.9205, + "step": 803 + }, + { + "epoch": 1.1635311143270621, + "grad_norm": 1.5997307300567627, + "learning_rate": 2.381335478680612e-06, + "loss": 0.9089, + "step": 804 + }, + { + "epoch": 1.1649782923299565, + "grad_norm": 1.4809633493423462, + "learning_rate": 2.3773129525341916e-06, + "loss": 0.8914, + "step": 805 + }, + { + "epoch": 1.166425470332851, + "grad_norm": 1.48288094997406, + "learning_rate": 2.3732904263877717e-06, + "loss": 0.8872, + "step": 806 + }, + { + "epoch": 1.1678726483357453, + "grad_norm": 1.5765438079833984, + "learning_rate": 2.369267900241352e-06, + "loss": 0.921, + "step": 807 + }, + { + "epoch": 1.1693198263386397, + "grad_norm": 1.5146821737289429, + "learning_rate": 2.365245374094932e-06, + "loss": 0.8814, + "step": 808 + }, + { + "epoch": 1.170767004341534, + "grad_norm": 1.4884650707244873, + "learning_rate": 2.361222847948512e-06, + "loss": 0.8718, + "step": 809 + }, + { + "epoch": 1.1722141823444283, + "grad_norm": 1.5190155506134033, + "learning_rate": 2.3572003218020918e-06, + "loss": 0.9138, + "step": 810 + }, + { + "epoch": 1.1722141823444283, + "eval_loss": 1.0010331869125366, + "eval_runtime": 25.9025, + "eval_samples_per_second": 38.606, + "eval_steps_per_second": 2.432, + "step": 810 + }, + { + "epoch": 1.1736613603473227, + "grad_norm": 1.619706392288208, + "learning_rate": 2.353177795655672e-06, + "loss": 0.8873, + "step": 811 + }, + { + "epoch": 1.1751085383502171, + "grad_norm": 1.6402854919433594, + "learning_rate": 2.349155269509252e-06, + "loss": 0.8954, + "step": 812 + }, + { + "epoch": 1.1765557163531115, + "grad_norm": 1.5414272546768188, + "learning_rate": 2.345132743362832e-06, + "loss": 0.899, + "step": 813 + }, + { + "epoch": 1.1780028943560057, + "grad_norm": 1.7477384805679321, + "learning_rate": 2.3411102172164123e-06, + "loss": 0.9155, + "step": 814 + }, + { + "epoch": 1.1794500723589, + "grad_norm": 1.5654603242874146, + "learning_rate": 2.337087691069992e-06, + "loss": 0.9058, + "step": 815 + }, + { + "epoch": 1.1808972503617945, + "grad_norm": 1.5773078203201294, + "learning_rate": 2.333065164923572e-06, + "loss": 0.909, + "step": 816 + }, + { + "epoch": 1.182344428364689, + "grad_norm": 1.5746979713439941, + "learning_rate": 2.3290426387771522e-06, + "loss": 0.9036, + "step": 817 + }, + { + "epoch": 1.1837916063675833, + "grad_norm": 1.5165321826934814, + "learning_rate": 2.3250201126307324e-06, + "loss": 0.8874, + "step": 818 + }, + { + "epoch": 1.1852387843704775, + "grad_norm": 1.4424513578414917, + "learning_rate": 2.3209975864843125e-06, + "loss": 0.9118, + "step": 819 + }, + { + "epoch": 1.1866859623733719, + "grad_norm": 1.494571328163147, + "learning_rate": 2.316975060337892e-06, + "loss": 0.8679, + "step": 820 + }, + { + "epoch": 1.1866859623733719, + "eval_loss": 0.9990912079811096, + "eval_runtime": 25.843, + "eval_samples_per_second": 38.695, + "eval_steps_per_second": 2.438, + "step": 820 + }, + { + "epoch": 1.1881331403762663, + "grad_norm": 1.480198860168457, + "learning_rate": 2.3129525341914723e-06, + "loss": 0.9037, + "step": 821 + }, + { + "epoch": 1.1895803183791607, + "grad_norm": 1.584207534790039, + "learning_rate": 2.3089300080450525e-06, + "loss": 0.9139, + "step": 822 + }, + { + "epoch": 1.191027496382055, + "grad_norm": 1.612007975578308, + "learning_rate": 2.3049074818986326e-06, + "loss": 0.8963, + "step": 823 + }, + { + "epoch": 1.1924746743849495, + "grad_norm": 1.6596332788467407, + "learning_rate": 2.3008849557522127e-06, + "loss": 0.8807, + "step": 824 + }, + { + "epoch": 1.1939218523878437, + "grad_norm": 1.6010795831680298, + "learning_rate": 2.2968624296057924e-06, + "loss": 0.9194, + "step": 825 + }, + { + "epoch": 1.195369030390738, + "grad_norm": 1.525209903717041, + "learning_rate": 2.2928399034593725e-06, + "loss": 0.9004, + "step": 826 + }, + { + "epoch": 1.1968162083936325, + "grad_norm": 1.5119304656982422, + "learning_rate": 2.2888173773129527e-06, + "loss": 0.9177, + "step": 827 + }, + { + "epoch": 1.1982633863965269, + "grad_norm": 1.5764222145080566, + "learning_rate": 2.284794851166533e-06, + "loss": 0.8945, + "step": 828 + }, + { + "epoch": 1.199710564399421, + "grad_norm": 1.632712960243225, + "learning_rate": 2.280772325020113e-06, + "loss": 0.9267, + "step": 829 + }, + { + "epoch": 1.2011577424023154, + "grad_norm": 1.6575121879577637, + "learning_rate": 2.2767497988736926e-06, + "loss": 0.8893, + "step": 830 + }, + { + "epoch": 1.2011577424023154, + "eval_loss": 0.9975345730781555, + "eval_runtime": 25.7991, + "eval_samples_per_second": 38.761, + "eval_steps_per_second": 2.442, + "step": 830 + }, + { + "epoch": 1.2026049204052098, + "grad_norm": 1.5680720806121826, + "learning_rate": 2.2727272727272728e-06, + "loss": 0.8726, + "step": 831 + }, + { + "epoch": 1.2040520984081042, + "grad_norm": 1.5080980062484741, + "learning_rate": 2.268704746580853e-06, + "loss": 0.8918, + "step": 832 + }, + { + "epoch": 1.2054992764109986, + "grad_norm": 1.4733283519744873, + "learning_rate": 2.264682220434433e-06, + "loss": 0.8833, + "step": 833 + }, + { + "epoch": 1.206946454413893, + "grad_norm": 1.526259183883667, + "learning_rate": 2.260659694288013e-06, + "loss": 0.8811, + "step": 834 + }, + { + "epoch": 1.2083936324167872, + "grad_norm": 1.5349605083465576, + "learning_rate": 2.256637168141593e-06, + "loss": 0.8577, + "step": 835 + }, + { + "epoch": 1.2098408104196816, + "grad_norm": 1.5627975463867188, + "learning_rate": 2.252614641995173e-06, + "loss": 0.884, + "step": 836 + }, + { + "epoch": 1.211287988422576, + "grad_norm": 1.6055257320404053, + "learning_rate": 2.248592115848753e-06, + "loss": 0.8896, + "step": 837 + }, + { + "epoch": 1.2127351664254704, + "grad_norm": 1.5470566749572754, + "learning_rate": 2.2445695897023332e-06, + "loss": 0.8953, + "step": 838 + }, + { + "epoch": 1.2141823444283646, + "grad_norm": 1.4979593753814697, + "learning_rate": 2.2405470635559134e-06, + "loss": 0.8676, + "step": 839 + }, + { + "epoch": 1.215629522431259, + "grad_norm": 1.5414305925369263, + "learning_rate": 2.236524537409493e-06, + "loss": 0.9068, + "step": 840 + }, + { + "epoch": 1.215629522431259, + "eval_loss": 0.9961005449295044, + "eval_runtime": 25.6585, + "eval_samples_per_second": 38.973, + "eval_steps_per_second": 2.455, + "step": 840 + }, + { + "epoch": 1.2170767004341534, + "grad_norm": 1.518509864807129, + "learning_rate": 2.232502011263073e-06, + "loss": 0.8575, + "step": 841 + }, + { + "epoch": 1.2185238784370478, + "grad_norm": 1.4536638259887695, + "learning_rate": 2.2284794851166537e-06, + "loss": 0.8858, + "step": 842 + }, + { + "epoch": 1.2199710564399422, + "grad_norm": 1.606372356414795, + "learning_rate": 2.2244569589702334e-06, + "loss": 0.8667, + "step": 843 + }, + { + "epoch": 1.2214182344428366, + "grad_norm": 1.526896357536316, + "learning_rate": 2.2204344328238136e-06, + "loss": 0.8832, + "step": 844 + }, + { + "epoch": 1.2228654124457308, + "grad_norm": 1.5027129650115967, + "learning_rate": 2.2164119066773937e-06, + "loss": 0.8948, + "step": 845 + }, + { + "epoch": 1.2243125904486252, + "grad_norm": 1.4722239971160889, + "learning_rate": 2.212389380530974e-06, + "loss": 0.9062, + "step": 846 + }, + { + "epoch": 1.2257597684515196, + "grad_norm": 1.4409259557724, + "learning_rate": 2.208366854384554e-06, + "loss": 0.9001, + "step": 847 + }, + { + "epoch": 1.227206946454414, + "grad_norm": 1.4950021505355835, + "learning_rate": 2.2043443282381337e-06, + "loss": 0.9215, + "step": 848 + }, + { + "epoch": 1.2286541244573081, + "grad_norm": 1.6024349927902222, + "learning_rate": 2.2003218020917138e-06, + "loss": 0.889, + "step": 849 + }, + { + "epoch": 1.2301013024602026, + "grad_norm": 1.4954161643981934, + "learning_rate": 2.196299275945294e-06, + "loss": 0.8865, + "step": 850 + }, + { + "epoch": 1.2301013024602026, + "eval_loss": 0.9965184330940247, + "eval_runtime": 26.0623, + "eval_samples_per_second": 38.37, + "eval_steps_per_second": 2.417, + "step": 850 + }, + { + "epoch": 1.231548480463097, + "grad_norm": 1.5020195245742798, + "learning_rate": 2.192276749798874e-06, + "loss": 0.9048, + "step": 851 + }, + { + "epoch": 1.2329956584659914, + "grad_norm": 1.5896894931793213, + "learning_rate": 2.188254223652454e-06, + "loss": 0.8976, + "step": 852 + }, + { + "epoch": 1.2344428364688858, + "grad_norm": 1.6095257997512817, + "learning_rate": 2.184231697506034e-06, + "loss": 0.8699, + "step": 853 + }, + { + "epoch": 1.2358900144717802, + "grad_norm": 1.5864946842193604, + "learning_rate": 2.180209171359614e-06, + "loss": 0.8791, + "step": 854 + }, + { + "epoch": 1.2373371924746743, + "grad_norm": 1.5503919124603271, + "learning_rate": 2.176186645213194e-06, + "loss": 0.8984, + "step": 855 + }, + { + "epoch": 1.2387843704775687, + "grad_norm": 1.5076204538345337, + "learning_rate": 2.1721641190667743e-06, + "loss": 0.8658, + "step": 856 + }, + { + "epoch": 1.2402315484804631, + "grad_norm": 1.5990691184997559, + "learning_rate": 2.1681415929203544e-06, + "loss": 0.8868, + "step": 857 + }, + { + "epoch": 1.2416787264833575, + "grad_norm": 1.5078307390213013, + "learning_rate": 2.164119066773934e-06, + "loss": 0.895, + "step": 858 + }, + { + "epoch": 1.2431259044862517, + "grad_norm": 1.5056289434432983, + "learning_rate": 2.1600965406275142e-06, + "loss": 0.8813, + "step": 859 + }, + { + "epoch": 1.244573082489146, + "grad_norm": 1.5606762170791626, + "learning_rate": 2.1560740144810943e-06, + "loss": 0.9292, + "step": 860 + }, + { + "epoch": 1.244573082489146, + "eval_loss": 0.9943708181381226, + "eval_runtime": 25.6768, + "eval_samples_per_second": 38.946, + "eval_steps_per_second": 2.454, + "step": 860 + }, + { + "epoch": 1.2460202604920405, + "grad_norm": 1.4824141263961792, + "learning_rate": 2.1520514883346745e-06, + "loss": 0.8732, + "step": 861 + }, + { + "epoch": 1.247467438494935, + "grad_norm": 1.5555789470672607, + "learning_rate": 2.1480289621882546e-06, + "loss": 0.8937, + "step": 862 + }, + { + "epoch": 1.2489146164978293, + "grad_norm": 1.5916050672531128, + "learning_rate": 2.1440064360418343e-06, + "loss": 0.8728, + "step": 863 + }, + { + "epoch": 1.2503617945007237, + "grad_norm": 1.6065150499343872, + "learning_rate": 2.1399839098954144e-06, + "loss": 0.8511, + "step": 864 + }, + { + "epoch": 1.251808972503618, + "grad_norm": 1.569021224975586, + "learning_rate": 2.1359613837489946e-06, + "loss": 0.8922, + "step": 865 + }, + { + "epoch": 1.2532561505065123, + "grad_norm": 1.5763746500015259, + "learning_rate": 2.1319388576025747e-06, + "loss": 0.9, + "step": 866 + }, + { + "epoch": 1.2547033285094067, + "grad_norm": 1.5964343547821045, + "learning_rate": 2.127916331456155e-06, + "loss": 0.9042, + "step": 867 + }, + { + "epoch": 1.256150506512301, + "grad_norm": 1.457443118095398, + "learning_rate": 2.1238938053097345e-06, + "loss": 0.8802, + "step": 868 + }, + { + "epoch": 1.2575976845151953, + "grad_norm": 1.4925570487976074, + "learning_rate": 2.1198712791633146e-06, + "loss": 0.8771, + "step": 869 + }, + { + "epoch": 1.2590448625180897, + "grad_norm": 1.543954849243164, + "learning_rate": 2.1158487530168948e-06, + "loss": 0.9085, + "step": 870 + }, + { + "epoch": 1.2590448625180897, + "eval_loss": 0.993406355381012, + "eval_runtime": 25.8004, + "eval_samples_per_second": 38.759, + "eval_steps_per_second": 2.442, + "step": 870 + }, + { + "epoch": 1.260492040520984, + "grad_norm": 1.5981473922729492, + "learning_rate": 2.111826226870475e-06, + "loss": 0.885, + "step": 871 + }, + { + "epoch": 1.2619392185238785, + "grad_norm": 1.5987436771392822, + "learning_rate": 2.107803700724055e-06, + "loss": 0.8883, + "step": 872 + }, + { + "epoch": 1.2633863965267729, + "grad_norm": 1.5223180055618286, + "learning_rate": 2.1037811745776347e-06, + "loss": 0.8445, + "step": 873 + }, + { + "epoch": 1.2648335745296673, + "grad_norm": 1.5502173900604248, + "learning_rate": 2.099758648431215e-06, + "loss": 0.8799, + "step": 874 + }, + { + "epoch": 1.2662807525325614, + "grad_norm": 1.574419379234314, + "learning_rate": 2.095736122284795e-06, + "loss": 0.8862, + "step": 875 + }, + { + "epoch": 1.2677279305354558, + "grad_norm": 1.5151711702346802, + "learning_rate": 2.091713596138375e-06, + "loss": 0.842, + "step": 876 + }, + { + "epoch": 1.2691751085383502, + "grad_norm": 1.5926389694213867, + "learning_rate": 2.0876910699919552e-06, + "loss": 0.8812, + "step": 877 + }, + { + "epoch": 1.2706222865412446, + "grad_norm": 1.5817290544509888, + "learning_rate": 2.083668543845535e-06, + "loss": 0.8927, + "step": 878 + }, + { + "epoch": 1.2720694645441388, + "grad_norm": 1.5659255981445312, + "learning_rate": 2.079646017699115e-06, + "loss": 0.8973, + "step": 879 + }, + { + "epoch": 1.2735166425470332, + "grad_norm": 1.5826165676116943, + "learning_rate": 2.075623491552695e-06, + "loss": 0.8934, + "step": 880 + }, + { + "epoch": 1.2735166425470332, + "eval_loss": 0.994766116142273, + "eval_runtime": 25.9544, + "eval_samples_per_second": 38.529, + "eval_steps_per_second": 2.427, + "step": 880 + }, + { + "epoch": 1.2749638205499276, + "grad_norm": 1.5688477754592896, + "learning_rate": 2.0716009654062753e-06, + "loss": 0.8909, + "step": 881 + }, + { + "epoch": 1.276410998552822, + "grad_norm": 1.6281142234802246, + "learning_rate": 2.067578439259855e-06, + "loss": 0.9109, + "step": 882 + }, + { + "epoch": 1.2778581765557164, + "grad_norm": 1.5646711587905884, + "learning_rate": 2.063555913113435e-06, + "loss": 0.8628, + "step": 883 + }, + { + "epoch": 1.2793053545586108, + "grad_norm": 1.5538504123687744, + "learning_rate": 2.0595333869670153e-06, + "loss": 0.8856, + "step": 884 + }, + { + "epoch": 1.280752532561505, + "grad_norm": 1.5453202724456787, + "learning_rate": 2.0555108608205954e-06, + "loss": 0.8892, + "step": 885 + }, + { + "epoch": 1.2821997105643994, + "grad_norm": 1.520643949508667, + "learning_rate": 2.0514883346741755e-06, + "loss": 0.8811, + "step": 886 + }, + { + "epoch": 1.2836468885672938, + "grad_norm": 1.4514366388320923, + "learning_rate": 2.0474658085277557e-06, + "loss": 0.8871, + "step": 887 + }, + { + "epoch": 1.2850940665701882, + "grad_norm": 1.5639816522598267, + "learning_rate": 2.043443282381336e-06, + "loss": 0.8786, + "step": 888 + }, + { + "epoch": 1.2865412445730824, + "grad_norm": 1.5956770181655884, + "learning_rate": 2.039420756234916e-06, + "loss": 0.8959, + "step": 889 + }, + { + "epoch": 1.2879884225759768, + "grad_norm": 1.6308395862579346, + "learning_rate": 2.035398230088496e-06, + "loss": 0.884, + "step": 890 + }, + { + "epoch": 1.2879884225759768, + "eval_loss": 0.9930154085159302, + "eval_runtime": 25.8341, + "eval_samples_per_second": 38.709, + "eval_steps_per_second": 2.439, + "step": 890 + }, + { + "epoch": 1.2894356005788712, + "grad_norm": 1.505920171737671, + "learning_rate": 2.0313757039420758e-06, + "loss": 0.85, + "step": 891 + }, + { + "epoch": 1.2908827785817656, + "grad_norm": 1.5757458209991455, + "learning_rate": 2.027353177795656e-06, + "loss": 0.8968, + "step": 892 + }, + { + "epoch": 1.29232995658466, + "grad_norm": 1.5903902053833008, + "learning_rate": 2.023330651649236e-06, + "loss": 0.8994, + "step": 893 + }, + { + "epoch": 1.2937771345875544, + "grad_norm": 1.4955521821975708, + "learning_rate": 2.019308125502816e-06, + "loss": 0.8982, + "step": 894 + }, + { + "epoch": 1.2952243125904486, + "grad_norm": 1.4922794103622437, + "learning_rate": 2.0152855993563963e-06, + "loss": 0.8778, + "step": 895 + }, + { + "epoch": 1.296671490593343, + "grad_norm": 1.5851147174835205, + "learning_rate": 2.011263073209976e-06, + "loss": 0.8602, + "step": 896 + }, + { + "epoch": 1.2981186685962374, + "grad_norm": 1.5002673864364624, + "learning_rate": 2.007240547063556e-06, + "loss": 0.8868, + "step": 897 + }, + { + "epoch": 1.2995658465991318, + "grad_norm": 1.5019080638885498, + "learning_rate": 2.0032180209171362e-06, + "loss": 0.8955, + "step": 898 + }, + { + "epoch": 1.301013024602026, + "grad_norm": 1.544146180152893, + "learning_rate": 1.9991954947707164e-06, + "loss": 0.8934, + "step": 899 + }, + { + "epoch": 1.3024602026049203, + "grad_norm": 1.5680269002914429, + "learning_rate": 1.995172968624296e-06, + "loss": 0.8784, + "step": 900 + }, + { + "epoch": 1.3024602026049203, + "eval_loss": 0.9916625618934631, + "eval_runtime": 25.7891, + "eval_samples_per_second": 38.776, + "eval_steps_per_second": 2.443, + "step": 900 + }, + { + "epoch": 1.3039073806078147, + "grad_norm": 1.6523807048797607, + "learning_rate": 1.991150442477876e-06, + "loss": 0.9072, + "step": 901 + }, + { + "epoch": 1.3053545586107091, + "grad_norm": 1.5159372091293335, + "learning_rate": 1.9871279163314563e-06, + "loss": 0.8983, + "step": 902 + }, + { + "epoch": 1.3068017366136035, + "grad_norm": 1.7033772468566895, + "learning_rate": 1.9831053901850364e-06, + "loss": 0.8733, + "step": 903 + }, + { + "epoch": 1.308248914616498, + "grad_norm": 1.5757745504379272, + "learning_rate": 1.9790828640386166e-06, + "loss": 0.8769, + "step": 904 + }, + { + "epoch": 1.3096960926193921, + "grad_norm": 1.5687916278839111, + "learning_rate": 1.9750603378921963e-06, + "loss": 0.9141, + "step": 905 + }, + { + "epoch": 1.3111432706222865, + "grad_norm": 1.6125038862228394, + "learning_rate": 1.9710378117457764e-06, + "loss": 0.9043, + "step": 906 + }, + { + "epoch": 1.312590448625181, + "grad_norm": 1.5606075525283813, + "learning_rate": 1.9670152855993565e-06, + "loss": 0.8921, + "step": 907 + }, + { + "epoch": 1.3140376266280753, + "grad_norm": 1.6325387954711914, + "learning_rate": 1.9629927594529367e-06, + "loss": 0.8908, + "step": 908 + }, + { + "epoch": 1.3154848046309695, + "grad_norm": 1.5046014785766602, + "learning_rate": 1.9589702333065168e-06, + "loss": 0.8781, + "step": 909 + }, + { + "epoch": 1.316931982633864, + "grad_norm": 1.5318217277526855, + "learning_rate": 1.9549477071600965e-06, + "loss": 0.8553, + "step": 910 + }, + { + "epoch": 1.316931982633864, + "eval_loss": 0.9922801852226257, + "eval_runtime": 25.6515, + "eval_samples_per_second": 38.984, + "eval_steps_per_second": 2.456, + "step": 910 + }, + { + "epoch": 1.3183791606367583, + "grad_norm": 1.5576587915420532, + "learning_rate": 1.9509251810136766e-06, + "loss": 0.9139, + "step": 911 + }, + { + "epoch": 1.3198263386396527, + "grad_norm": 1.6574680805206299, + "learning_rate": 1.9469026548672567e-06, + "loss": 0.8644, + "step": 912 + }, + { + "epoch": 1.321273516642547, + "grad_norm": 1.6218926906585693, + "learning_rate": 1.942880128720837e-06, + "loss": 0.8941, + "step": 913 + }, + { + "epoch": 1.3227206946454415, + "grad_norm": 1.6747978925704956, + "learning_rate": 1.938857602574417e-06, + "loss": 0.9131, + "step": 914 + }, + { + "epoch": 1.3241678726483357, + "grad_norm": 1.5592243671417236, + "learning_rate": 1.9348350764279967e-06, + "loss": 0.9109, + "step": 915 + }, + { + "epoch": 1.32561505065123, + "grad_norm": 1.5739071369171143, + "learning_rate": 1.930812550281577e-06, + "loss": 0.8956, + "step": 916 + }, + { + "epoch": 1.3270622286541245, + "grad_norm": 1.529422402381897, + "learning_rate": 1.926790024135157e-06, + "loss": 0.872, + "step": 917 + }, + { + "epoch": 1.3285094066570189, + "grad_norm": 1.4759328365325928, + "learning_rate": 1.922767497988737e-06, + "loss": 0.8646, + "step": 918 + }, + { + "epoch": 1.329956584659913, + "grad_norm": 1.5779008865356445, + "learning_rate": 1.9187449718423172e-06, + "loss": 0.8933, + "step": 919 + }, + { + "epoch": 1.3314037626628075, + "grad_norm": 1.5866317749023438, + "learning_rate": 1.914722445695897e-06, + "loss": 0.9112, + "step": 920 + }, + { + "epoch": 1.3314037626628075, + "eval_loss": 0.9918884634971619, + "eval_runtime": 26.0278, + "eval_samples_per_second": 38.42, + "eval_steps_per_second": 2.42, + "step": 920 + }, + { + "epoch": 1.3328509406657019, + "grad_norm": 1.5680437088012695, + "learning_rate": 1.910699919549477e-06, + "loss": 0.9152, + "step": 921 + }, + { + "epoch": 1.3342981186685963, + "grad_norm": 1.5756558179855347, + "learning_rate": 1.9066773934030572e-06, + "loss": 0.9017, + "step": 922 + }, + { + "epoch": 1.3357452966714907, + "grad_norm": 1.6703917980194092, + "learning_rate": 1.9026548672566373e-06, + "loss": 0.8885, + "step": 923 + }, + { + "epoch": 1.337192474674385, + "grad_norm": 1.567590355873108, + "learning_rate": 1.8986323411102172e-06, + "loss": 0.8838, + "step": 924 + }, + { + "epoch": 1.3386396526772792, + "grad_norm": 1.5318201780319214, + "learning_rate": 1.8946098149637973e-06, + "loss": 0.869, + "step": 925 + }, + { + "epoch": 1.3400868306801736, + "grad_norm": 1.553001880645752, + "learning_rate": 1.8905872888173773e-06, + "loss": 0.8781, + "step": 926 + }, + { + "epoch": 1.341534008683068, + "grad_norm": 1.5244364738464355, + "learning_rate": 1.8865647626709574e-06, + "loss": 0.8726, + "step": 927 + }, + { + "epoch": 1.3429811866859624, + "grad_norm": 1.4793615341186523, + "learning_rate": 1.8825422365245375e-06, + "loss": 0.8623, + "step": 928 + }, + { + "epoch": 1.3444283646888566, + "grad_norm": 1.5498313903808594, + "learning_rate": 1.8785197103781174e-06, + "loss": 0.8663, + "step": 929 + }, + { + "epoch": 1.345875542691751, + "grad_norm": 1.5433320999145508, + "learning_rate": 1.8744971842316978e-06, + "loss": 0.8864, + "step": 930 + }, + { + "epoch": 1.345875542691751, + "eval_loss": 0.9909474849700928, + "eval_runtime": 25.7846, + "eval_samples_per_second": 38.783, + "eval_steps_per_second": 2.443, + "step": 930 + }, + { + "epoch": 1.3473227206946454, + "grad_norm": 1.4593886137008667, + "learning_rate": 1.870474658085278e-06, + "loss": 0.8686, + "step": 931 + }, + { + "epoch": 1.3487698986975398, + "grad_norm": 1.5526374578475952, + "learning_rate": 1.8664521319388578e-06, + "loss": 0.8927, + "step": 932 + }, + { + "epoch": 1.3502170767004342, + "grad_norm": 1.6133079528808594, + "learning_rate": 1.862429605792438e-06, + "loss": 0.8653, + "step": 933 + }, + { + "epoch": 1.3516642547033286, + "grad_norm": 1.4973291158676147, + "learning_rate": 1.8584070796460179e-06, + "loss": 0.8763, + "step": 934 + }, + { + "epoch": 1.3531114327062228, + "grad_norm": 1.5784869194030762, + "learning_rate": 1.854384553499598e-06, + "loss": 0.9069, + "step": 935 + }, + { + "epoch": 1.3545586107091172, + "grad_norm": 1.5758600234985352, + "learning_rate": 1.8503620273531781e-06, + "loss": 0.9041, + "step": 936 + }, + { + "epoch": 1.3560057887120116, + "grad_norm": 1.4842194318771362, + "learning_rate": 1.846339501206758e-06, + "loss": 0.8912, + "step": 937 + }, + { + "epoch": 1.357452966714906, + "grad_norm": 1.522415280342102, + "learning_rate": 1.8423169750603382e-06, + "loss": 0.8582, + "step": 938 + }, + { + "epoch": 1.3589001447178002, + "grad_norm": 1.483718752861023, + "learning_rate": 1.838294448913918e-06, + "loss": 0.8774, + "step": 939 + }, + { + "epoch": 1.3603473227206946, + "grad_norm": 1.4885591268539429, + "learning_rate": 1.8342719227674982e-06, + "loss": 0.8778, + "step": 940 + }, + { + "epoch": 1.3603473227206946, + "eval_loss": 0.9908953905105591, + "eval_runtime": 25.6816, + "eval_samples_per_second": 38.938, + "eval_steps_per_second": 2.453, + "step": 940 + }, + { + "epoch": 1.361794500723589, + "grad_norm": 1.5625662803649902, + "learning_rate": 1.8302493966210783e-06, + "loss": 0.8922, + "step": 941 + }, + { + "epoch": 1.3632416787264834, + "grad_norm": 1.5827653408050537, + "learning_rate": 1.8262268704746582e-06, + "loss": 0.8762, + "step": 942 + }, + { + "epoch": 1.3646888567293778, + "grad_norm": 1.532116174697876, + "learning_rate": 1.8222043443282384e-06, + "loss": 0.8838, + "step": 943 + }, + { + "epoch": 1.3661360347322722, + "grad_norm": 1.5727930068969727, + "learning_rate": 1.8181818181818183e-06, + "loss": 0.8761, + "step": 944 + }, + { + "epoch": 1.3675832127351664, + "grad_norm": 1.5438214540481567, + "learning_rate": 1.8141592920353984e-06, + "loss": 0.8739, + "step": 945 + }, + { + "epoch": 1.3690303907380608, + "grad_norm": 1.5137232542037964, + "learning_rate": 1.8101367658889785e-06, + "loss": 0.885, + "step": 946 + }, + { + "epoch": 1.3704775687409552, + "grad_norm": 1.5985186100006104, + "learning_rate": 1.8061142397425585e-06, + "loss": 0.8967, + "step": 947 + }, + { + "epoch": 1.3719247467438496, + "grad_norm": 1.607809066772461, + "learning_rate": 1.8020917135961386e-06, + "loss": 0.9175, + "step": 948 + }, + { + "epoch": 1.3733719247467437, + "grad_norm": 1.5717999935150146, + "learning_rate": 1.7980691874497185e-06, + "loss": 0.8862, + "step": 949 + }, + { + "epoch": 1.3748191027496381, + "grad_norm": 1.490583062171936, + "learning_rate": 1.7940466613032986e-06, + "loss": 0.8425, + "step": 950 + }, + { + "epoch": 1.3748191027496381, + "eval_loss": 0.989203929901123, + "eval_runtime": 25.7453, + "eval_samples_per_second": 38.842, + "eval_steps_per_second": 2.447, + "step": 950 + }, + { + "epoch": 1.3762662807525325, + "grad_norm": 1.5401087999343872, + "learning_rate": 1.7900241351568788e-06, + "loss": 0.8418, + "step": 951 + }, + { + "epoch": 1.377713458755427, + "grad_norm": 1.527870774269104, + "learning_rate": 1.7860016090104587e-06, + "loss": 0.8591, + "step": 952 + }, + { + "epoch": 1.3791606367583213, + "grad_norm": 1.5778286457061768, + "learning_rate": 1.7819790828640388e-06, + "loss": 0.8959, + "step": 953 + }, + { + "epoch": 1.3806078147612157, + "grad_norm": 1.5266025066375732, + "learning_rate": 1.7779565567176187e-06, + "loss": 0.8449, + "step": 954 + }, + { + "epoch": 1.38205499276411, + "grad_norm": 1.508654236793518, + "learning_rate": 1.7739340305711988e-06, + "loss": 0.8578, + "step": 955 + }, + { + "epoch": 1.3835021707670043, + "grad_norm": 1.515903353691101, + "learning_rate": 1.769911504424779e-06, + "loss": 0.8801, + "step": 956 + }, + { + "epoch": 1.3849493487698987, + "grad_norm": 1.4178016185760498, + "learning_rate": 1.7658889782783589e-06, + "loss": 0.8601, + "step": 957 + }, + { + "epoch": 1.3863965267727931, + "grad_norm": 1.5548313856124878, + "learning_rate": 1.761866452131939e-06, + "loss": 0.8874, + "step": 958 + }, + { + "epoch": 1.3878437047756873, + "grad_norm": 1.6333560943603516, + "learning_rate": 1.757843925985519e-06, + "loss": 0.874, + "step": 959 + }, + { + "epoch": 1.3892908827785817, + "grad_norm": 1.6060621738433838, + "learning_rate": 1.753821399839099e-06, + "loss": 0.8987, + "step": 960 + }, + { + "epoch": 1.3892908827785817, + "eval_loss": 0.9903543591499329, + "eval_runtime": 25.6673, + "eval_samples_per_second": 38.96, + "eval_steps_per_second": 2.454, + "step": 960 + }, + { + "epoch": 1.390738060781476, + "grad_norm": 1.526719331741333, + "learning_rate": 1.7497988736926792e-06, + "loss": 0.9042, + "step": 961 + }, + { + "epoch": 1.3921852387843705, + "grad_norm": 1.654146671295166, + "learning_rate": 1.7457763475462591e-06, + "loss": 0.8697, + "step": 962 + }, + { + "epoch": 1.393632416787265, + "grad_norm": 1.5884320735931396, + "learning_rate": 1.7417538213998392e-06, + "loss": 0.8967, + "step": 963 + }, + { + "epoch": 1.3950795947901593, + "grad_norm": 1.5201061964035034, + "learning_rate": 1.7377312952534192e-06, + "loss": 0.9149, + "step": 964 + }, + { + "epoch": 1.3965267727930535, + "grad_norm": 1.4852148294448853, + "learning_rate": 1.7337087691069993e-06, + "loss": 0.8937, + "step": 965 + }, + { + "epoch": 1.3979739507959479, + "grad_norm": 1.4756934642791748, + "learning_rate": 1.7296862429605792e-06, + "loss": 0.8906, + "step": 966 + }, + { + "epoch": 1.3994211287988423, + "grad_norm": 1.4979703426361084, + "learning_rate": 1.7256637168141593e-06, + "loss": 0.8766, + "step": 967 + }, + { + "epoch": 1.4008683068017367, + "grad_norm": 1.4638196229934692, + "learning_rate": 1.7216411906677395e-06, + "loss": 0.8316, + "step": 968 + }, + { + "epoch": 1.4023154848046309, + "grad_norm": 1.4833625555038452, + "learning_rate": 1.7176186645213194e-06, + "loss": 0.8643, + "step": 969 + }, + { + "epoch": 1.4037626628075253, + "grad_norm": 1.6497318744659424, + "learning_rate": 1.7135961383748995e-06, + "loss": 0.9124, + "step": 970 + }, + { + "epoch": 1.4037626628075253, + "eval_loss": 0.9889197945594788, + "eval_runtime": 26.0218, + "eval_samples_per_second": 38.429, + "eval_steps_per_second": 2.421, + "step": 970 + }, + { + "epoch": 1.4052098408104197, + "grad_norm": 1.4444023370742798, + "learning_rate": 1.7095736122284794e-06, + "loss": 0.8733, + "step": 971 + }, + { + "epoch": 1.406657018813314, + "grad_norm": 1.5876818895339966, + "learning_rate": 1.7055510860820595e-06, + "loss": 0.8442, + "step": 972 + }, + { + "epoch": 1.4081041968162085, + "grad_norm": 1.6315959692001343, + "learning_rate": 1.7015285599356397e-06, + "loss": 0.8869, + "step": 973 + }, + { + "epoch": 1.4095513748191029, + "grad_norm": 1.6600921154022217, + "learning_rate": 1.6975060337892196e-06, + "loss": 0.9176, + "step": 974 + }, + { + "epoch": 1.410998552821997, + "grad_norm": 1.4976136684417725, + "learning_rate": 1.6934835076428e-06, + "loss": 0.8663, + "step": 975 + }, + { + "epoch": 1.4124457308248914, + "grad_norm": 1.5567373037338257, + "learning_rate": 1.68946098149638e-06, + "loss": 0.9061, + "step": 976 + }, + { + "epoch": 1.4138929088277858, + "grad_norm": 1.483586311340332, + "learning_rate": 1.68543845534996e-06, + "loss": 0.8904, + "step": 977 + }, + { + "epoch": 1.4153400868306802, + "grad_norm": 1.5179016590118408, + "learning_rate": 1.68141592920354e-06, + "loss": 0.8972, + "step": 978 + }, + { + "epoch": 1.4167872648335744, + "grad_norm": 1.4706259965896606, + "learning_rate": 1.6773934030571202e-06, + "loss": 0.8727, + "step": 979 + }, + { + "epoch": 1.4182344428364688, + "grad_norm": 1.6346557140350342, + "learning_rate": 1.6733708769107001e-06, + "loss": 0.9173, + "step": 980 + }, + { + "epoch": 1.4182344428364688, + "eval_loss": 0.9879488348960876, + "eval_runtime": 25.8173, + "eval_samples_per_second": 38.734, + "eval_steps_per_second": 2.44, + "step": 980 + }, + { + "epoch": 1.4196816208393632, + "grad_norm": 1.6798738241195679, + "learning_rate": 1.6693483507642803e-06, + "loss": 0.8445, + "step": 981 + }, + { + "epoch": 1.4211287988422576, + "grad_norm": 1.5508772134780884, + "learning_rate": 1.6653258246178602e-06, + "loss": 0.8713, + "step": 982 + }, + { + "epoch": 1.422575976845152, + "grad_norm": 1.557064414024353, + "learning_rate": 1.6613032984714403e-06, + "loss": 0.8596, + "step": 983 + }, + { + "epoch": 1.4240231548480464, + "grad_norm": 1.6628038883209229, + "learning_rate": 1.6572807723250202e-06, + "loss": 0.9002, + "step": 984 + }, + { + "epoch": 1.4254703328509406, + "grad_norm": 1.606233835220337, + "learning_rate": 1.6532582461786004e-06, + "loss": 0.8555, + "step": 985 + }, + { + "epoch": 1.426917510853835, + "grad_norm": 1.694331407546997, + "learning_rate": 1.6492357200321805e-06, + "loss": 0.8903, + "step": 986 + }, + { + "epoch": 1.4283646888567294, + "grad_norm": 1.5641584396362305, + "learning_rate": 1.6452131938857604e-06, + "loss": 0.8864, + "step": 987 + }, + { + "epoch": 1.4298118668596238, + "grad_norm": 1.524124264717102, + "learning_rate": 1.6411906677393405e-06, + "loss": 0.8899, + "step": 988 + }, + { + "epoch": 1.431259044862518, + "grad_norm": 1.5322781801223755, + "learning_rate": 1.6371681415929204e-06, + "loss": 0.8723, + "step": 989 + }, + { + "epoch": 1.4327062228654124, + "grad_norm": 1.5665675401687622, + "learning_rate": 1.6331456154465006e-06, + "loss": 0.9193, + "step": 990 + }, + { + "epoch": 1.4327062228654124, + "eval_loss": 0.9871683120727539, + "eval_runtime": 25.7573, + "eval_samples_per_second": 38.824, + "eval_steps_per_second": 2.446, + "step": 990 + }, + { + "epoch": 1.4341534008683068, + "grad_norm": 1.5333694219589233, + "learning_rate": 1.6291230893000807e-06, + "loss": 0.8669, + "step": 991 + }, + { + "epoch": 1.4356005788712012, + "grad_norm": 1.4930754899978638, + "learning_rate": 1.6251005631536606e-06, + "loss": 0.8744, + "step": 992 + }, + { + "epoch": 1.4370477568740956, + "grad_norm": 1.5360321998596191, + "learning_rate": 1.6210780370072407e-06, + "loss": 0.8883, + "step": 993 + }, + { + "epoch": 1.43849493487699, + "grad_norm": 1.568663477897644, + "learning_rate": 1.6170555108608207e-06, + "loss": 0.8927, + "step": 994 + }, + { + "epoch": 1.4399421128798842, + "grad_norm": 1.62112295627594, + "learning_rate": 1.6130329847144008e-06, + "loss": 0.9045, + "step": 995 + }, + { + "epoch": 1.4413892908827786, + "grad_norm": 1.5460090637207031, + "learning_rate": 1.609010458567981e-06, + "loss": 0.9154, + "step": 996 + }, + { + "epoch": 1.442836468885673, + "grad_norm": 1.470872402191162, + "learning_rate": 1.6049879324215608e-06, + "loss": 0.861, + "step": 997 + }, + { + "epoch": 1.4442836468885674, + "grad_norm": 1.552982211112976, + "learning_rate": 1.600965406275141e-06, + "loss": 0.8359, + "step": 998 + }, + { + "epoch": 1.4457308248914615, + "grad_norm": 1.5416020154953003, + "learning_rate": 1.5969428801287209e-06, + "loss": 0.885, + "step": 999 + }, + { + "epoch": 1.447178002894356, + "grad_norm": 1.5289608240127563, + "learning_rate": 1.592920353982301e-06, + "loss": 0.8767, + "step": 1000 + }, + { + "epoch": 1.447178002894356, + "eval_loss": 0.9877498745918274, + "eval_runtime": 26.065, + "eval_samples_per_second": 38.366, + "eval_steps_per_second": 2.417, + "step": 1000 + }, + { + "epoch": 1.4486251808972503, + "grad_norm": 1.5533311367034912, + "learning_rate": 1.5888978278358811e-06, + "loss": 0.837, + "step": 1001 + }, + { + "epoch": 1.4500723589001447, + "grad_norm": 1.605147361755371, + "learning_rate": 1.584875301689461e-06, + "loss": 0.8792, + "step": 1002 + }, + { + "epoch": 1.4515195369030391, + "grad_norm": 1.6621363162994385, + "learning_rate": 1.5808527755430412e-06, + "loss": 0.8586, + "step": 1003 + }, + { + "epoch": 1.4529667149059335, + "grad_norm": 1.5832258462905884, + "learning_rate": 1.576830249396621e-06, + "loss": 0.915, + "step": 1004 + }, + { + "epoch": 1.4544138929088277, + "grad_norm": 1.4973900318145752, + "learning_rate": 1.5728077232502012e-06, + "loss": 0.8703, + "step": 1005 + }, + { + "epoch": 1.4558610709117221, + "grad_norm": 1.5472540855407715, + "learning_rate": 1.5687851971037813e-06, + "loss": 0.8971, + "step": 1006 + }, + { + "epoch": 1.4573082489146165, + "grad_norm": 1.6117088794708252, + "learning_rate": 1.5647626709573613e-06, + "loss": 0.8401, + "step": 1007 + }, + { + "epoch": 1.458755426917511, + "grad_norm": 1.5230553150177002, + "learning_rate": 1.5607401448109414e-06, + "loss": 0.8998, + "step": 1008 + }, + { + "epoch": 1.460202604920405, + "grad_norm": 1.5158506631851196, + "learning_rate": 1.5567176186645213e-06, + "loss": 0.8469, + "step": 1009 + }, + { + "epoch": 1.4616497829232995, + "grad_norm": 1.6470974683761597, + "learning_rate": 1.5526950925181014e-06, + "loss": 0.8783, + "step": 1010 + }, + { + "epoch": 1.4616497829232995, + "eval_loss": 0.9876880049705505, + "eval_runtime": 25.7522, + "eval_samples_per_second": 38.832, + "eval_steps_per_second": 2.446, + "step": 1010 + }, + { + "epoch": 1.463096960926194, + "grad_norm": 1.6227943897247314, + "learning_rate": 1.5486725663716816e-06, + "loss": 0.8993, + "step": 1011 + }, + { + "epoch": 1.4645441389290883, + "grad_norm": 1.558890700340271, + "learning_rate": 1.5446500402252615e-06, + "loss": 0.8601, + "step": 1012 + }, + { + "epoch": 1.4659913169319827, + "grad_norm": 1.5884944200515747, + "learning_rate": 1.5406275140788416e-06, + "loss": 0.8886, + "step": 1013 + }, + { + "epoch": 1.467438494934877, + "grad_norm": 1.463099479675293, + "learning_rate": 1.5366049879324215e-06, + "loss": 0.8625, + "step": 1014 + }, + { + "epoch": 1.4688856729377713, + "grad_norm": 1.5330357551574707, + "learning_rate": 1.5325824617860016e-06, + "loss": 0.8759, + "step": 1015 + }, + { + "epoch": 1.4703328509406657, + "grad_norm": 1.5273404121398926, + "learning_rate": 1.5285599356395816e-06, + "loss": 0.9061, + "step": 1016 + }, + { + "epoch": 1.47178002894356, + "grad_norm": 1.5751070976257324, + "learning_rate": 1.5245374094931617e-06, + "loss": 0.9084, + "step": 1017 + }, + { + "epoch": 1.4732272069464545, + "grad_norm": 1.5140705108642578, + "learning_rate": 1.5205148833467418e-06, + "loss": 0.8522, + "step": 1018 + }, + { + "epoch": 1.4746743849493487, + "grad_norm": 1.5287972688674927, + "learning_rate": 1.5164923572003221e-06, + "loss": 0.8384, + "step": 1019 + }, + { + "epoch": 1.476121562952243, + "grad_norm": 1.5287972688674927, + "learning_rate": 1.5164923572003221e-06, + "loss": 0.929, + "step": 1020 + }, + { + "epoch": 1.476121562952243, + "eval_loss": 0.9861629605293274, + "eval_runtime": 25.9299, + "eval_samples_per_second": 38.566, + "eval_steps_per_second": 2.43, + "step": 1020 + }, + { + "epoch": 1.4775687409551375, + "grad_norm": 1.6659374237060547, + "learning_rate": 1.512469831053902e-06, + "loss": 0.8733, + "step": 1021 + }, + { + "epoch": 1.4790159189580319, + "grad_norm": 1.5888694524765015, + "learning_rate": 1.5084473049074822e-06, + "loss": 0.8927, + "step": 1022 + }, + { + "epoch": 1.4804630969609263, + "grad_norm": 1.5705657005310059, + "learning_rate": 1.5044247787610621e-06, + "loss": 0.8729, + "step": 1023 + }, + { + "epoch": 1.4819102749638207, + "grad_norm": 1.5564839839935303, + "learning_rate": 1.5004022526146422e-06, + "loss": 0.9164, + "step": 1024 + }, + { + "epoch": 1.4833574529667148, + "grad_norm": 1.5336191654205322, + "learning_rate": 1.4963797264682224e-06, + "loss": 0.8673, + "step": 1025 + }, + { + "epoch": 1.4848046309696092, + "grad_norm": 1.5255659818649292, + "learning_rate": 1.4923572003218023e-06, + "loss": 0.8689, + "step": 1026 + }, + { + "epoch": 1.4862518089725036, + "grad_norm": 1.5760184526443481, + "learning_rate": 1.4883346741753824e-06, + "loss": 0.8825, + "step": 1027 + }, + { + "epoch": 1.487698986975398, + "grad_norm": 1.5247989892959595, + "learning_rate": 1.4843121480289623e-06, + "loss": 0.8774, + "step": 1028 + }, + { + "epoch": 1.4891461649782922, + "grad_norm": 1.5857770442962646, + "learning_rate": 1.4802896218825425e-06, + "loss": 0.8552, + "step": 1029 + }, + { + "epoch": 1.4905933429811866, + "grad_norm": 1.5850846767425537, + "learning_rate": 1.4762670957361226e-06, + "loss": 0.8623, + "step": 1030 + }, + { + "epoch": 1.4905933429811866, + "eval_loss": 0.9863028526306152, + "eval_runtime": 25.8299, + "eval_samples_per_second": 38.715, + "eval_steps_per_second": 2.439, + "step": 1030 + }, + { + "epoch": 1.492040520984081, + "grad_norm": 1.5669705867767334, + "learning_rate": 1.4722445695897025e-06, + "loss": 0.915, + "step": 1031 + }, + { + "epoch": 1.4934876989869754, + "grad_norm": 1.5132821798324585, + "learning_rate": 1.4682220434432826e-06, + "loss": 0.8721, + "step": 1032 + }, + { + "epoch": 1.4949348769898698, + "grad_norm": 1.519844651222229, + "learning_rate": 1.4641995172968625e-06, + "loss": 0.8907, + "step": 1033 + }, + { + "epoch": 1.4963820549927642, + "grad_norm": 1.488808512687683, + "learning_rate": 1.4601769911504427e-06, + "loss": 0.854, + "step": 1034 + }, + { + "epoch": 1.4978292329956584, + "grad_norm": 1.5658503770828247, + "learning_rate": 1.4561544650040226e-06, + "loss": 0.8855, + "step": 1035 + }, + { + "epoch": 1.4992764109985528, + "grad_norm": 1.5756577253341675, + "learning_rate": 1.4521319388576027e-06, + "loss": 0.8593, + "step": 1036 + }, + { + "epoch": 1.5007235890014472, + "grad_norm": 1.6409657001495361, + "learning_rate": 1.4481094127111828e-06, + "loss": 0.887, + "step": 1037 + }, + { + "epoch": 1.5021707670043414, + "grad_norm": 1.5633593797683716, + "learning_rate": 1.4440868865647628e-06, + "loss": 0.8393, + "step": 1038 + }, + { + "epoch": 1.5036179450072358, + "grad_norm": 1.5313143730163574, + "learning_rate": 1.4400643604183429e-06, + "loss": 0.8633, + "step": 1039 + }, + { + "epoch": 1.5050651230101302, + "grad_norm": 1.6449614763259888, + "learning_rate": 1.4360418342719228e-06, + "loss": 0.8755, + "step": 1040 + }, + { + "epoch": 1.5050651230101302, + "eval_loss": 0.985752284526825, + "eval_runtime": 25.9811, + "eval_samples_per_second": 38.49, + "eval_steps_per_second": 2.425, + "step": 1040 + }, + { + "epoch": 1.5065123010130246, + "grad_norm": 1.5782673358917236, + "learning_rate": 1.432019308125503e-06, + "loss": 0.8965, + "step": 1041 + }, + { + "epoch": 1.507959479015919, + "grad_norm": 1.5749887228012085, + "learning_rate": 1.427996781979083e-06, + "loss": 0.8864, + "step": 1042 + }, + { + "epoch": 1.5094066570188134, + "grad_norm": 1.4836838245391846, + "learning_rate": 1.423974255832663e-06, + "loss": 0.8881, + "step": 1043 + }, + { + "epoch": 1.5108538350217078, + "grad_norm": 1.5988322496414185, + "learning_rate": 1.419951729686243e-06, + "loss": 0.8877, + "step": 1044 + }, + { + "epoch": 1.5123010130246022, + "grad_norm": 1.5654700994491577, + "learning_rate": 1.415929203539823e-06, + "loss": 0.8933, + "step": 1045 + }, + { + "epoch": 1.5137481910274964, + "grad_norm": 1.462059497833252, + "learning_rate": 1.4119066773934031e-06, + "loss": 0.8677, + "step": 1046 + }, + { + "epoch": 1.5151953690303908, + "grad_norm": 1.4891657829284668, + "learning_rate": 1.4078841512469833e-06, + "loss": 0.8486, + "step": 1047 + }, + { + "epoch": 1.516642547033285, + "grad_norm": 1.6066802740097046, + "learning_rate": 1.4038616251005632e-06, + "loss": 0.8887, + "step": 1048 + }, + { + "epoch": 1.5180897250361793, + "grad_norm": 1.5902212858200073, + "learning_rate": 1.3998390989541433e-06, + "loss": 0.9109, + "step": 1049 + }, + { + "epoch": 1.5195369030390737, + "grad_norm": 1.545209527015686, + "learning_rate": 1.3958165728077232e-06, + "loss": 0.8758, + "step": 1050 + }, + { + "epoch": 1.5195369030390737, + "eval_loss": 0.9853872060775757, + "eval_runtime": 25.8212, + "eval_samples_per_second": 38.728, + "eval_steps_per_second": 2.44, + "step": 1050 + }, + { + "epoch": 1.5209840810419681, + "grad_norm": 1.5264979600906372, + "learning_rate": 1.3917940466613034e-06, + "loss": 0.8628, + "step": 1051 + }, + { + "epoch": 1.5224312590448625, + "grad_norm": 1.5595197677612305, + "learning_rate": 1.3877715205148835e-06, + "loss": 0.885, + "step": 1052 + }, + { + "epoch": 1.523878437047757, + "grad_norm": 1.6068772077560425, + "learning_rate": 1.3837489943684634e-06, + "loss": 0.902, + "step": 1053 + }, + { + "epoch": 1.5253256150506513, + "grad_norm": 1.7956867218017578, + "learning_rate": 1.3797264682220435e-06, + "loss": 0.9056, + "step": 1054 + }, + { + "epoch": 1.5267727930535457, + "grad_norm": 1.6634644269943237, + "learning_rate": 1.3757039420756234e-06, + "loss": 0.8778, + "step": 1055 + }, + { + "epoch": 1.52821997105644, + "grad_norm": 1.535996437072754, + "learning_rate": 1.3716814159292036e-06, + "loss": 0.8736, + "step": 1056 + }, + { + "epoch": 1.5296671490593343, + "grad_norm": 1.5196404457092285, + "learning_rate": 1.3676588897827837e-06, + "loss": 0.8596, + "step": 1057 + }, + { + "epoch": 1.5311143270622285, + "grad_norm": 1.5545907020568848, + "learning_rate": 1.3636363636363636e-06, + "loss": 0.8605, + "step": 1058 + }, + { + "epoch": 1.532561505065123, + "grad_norm": 1.5114647150039673, + "learning_rate": 1.3596138374899437e-06, + "loss": 0.8693, + "step": 1059 + }, + { + "epoch": 1.5340086830680173, + "grad_norm": 1.587093710899353, + "learning_rate": 1.3555913113435237e-06, + "loss": 0.867, + "step": 1060 + }, + { + "epoch": 1.5340086830680173, + "eval_loss": 0.9839922189712524, + "eval_runtime": 25.7893, + "eval_samples_per_second": 38.776, + "eval_steps_per_second": 2.443, + "step": 1060 + }, + { + "epoch": 1.5354558610709117, + "grad_norm": 1.618294358253479, + "learning_rate": 1.3515687851971038e-06, + "loss": 0.8773, + "step": 1061 + }, + { + "epoch": 1.536903039073806, + "grad_norm": 1.6108827590942383, + "learning_rate": 1.347546259050684e-06, + "loss": 0.8939, + "step": 1062 + }, + { + "epoch": 1.5383502170767005, + "grad_norm": 1.555820345878601, + "learning_rate": 1.3435237329042638e-06, + "loss": 0.864, + "step": 1063 + }, + { + "epoch": 1.539797395079595, + "grad_norm": 1.5706232786178589, + "learning_rate": 1.339501206757844e-06, + "loss": 0.8698, + "step": 1064 + }, + { + "epoch": 1.5412445730824893, + "grad_norm": 1.6310842037200928, + "learning_rate": 1.3354786806114243e-06, + "loss": 0.8755, + "step": 1065 + }, + { + "epoch": 1.5426917510853835, + "grad_norm": 1.602475881576538, + "learning_rate": 1.3314561544650042e-06, + "loss": 0.8974, + "step": 1066 + }, + { + "epoch": 1.5441389290882779, + "grad_norm": 1.5371959209442139, + "learning_rate": 1.3274336283185843e-06, + "loss": 0.8691, + "step": 1067 + }, + { + "epoch": 1.545586107091172, + "grad_norm": 1.6384830474853516, + "learning_rate": 1.3234111021721643e-06, + "loss": 0.8818, + "step": 1068 + }, + { + "epoch": 1.5470332850940665, + "grad_norm": 1.5125739574432373, + "learning_rate": 1.3193885760257444e-06, + "loss": 0.8694, + "step": 1069 + }, + { + "epoch": 1.5484804630969609, + "grad_norm": 1.5838148593902588, + "learning_rate": 1.3153660498793245e-06, + "loss": 0.8898, + "step": 1070 + }, + { + "epoch": 1.5484804630969609, + "eval_loss": 0.9824967384338379, + "eval_runtime": 25.9096, + "eval_samples_per_second": 38.596, + "eval_steps_per_second": 2.432, + "step": 1070 + }, + { + "epoch": 1.5499276410998553, + "grad_norm": 1.6065926551818848, + "learning_rate": 1.3113435237329044e-06, + "loss": 0.8909, + "step": 1071 + }, + { + "epoch": 1.5513748191027497, + "grad_norm": 1.5557194948196411, + "learning_rate": 1.3073209975864846e-06, + "loss": 0.8545, + "step": 1072 + }, + { + "epoch": 1.552821997105644, + "grad_norm": 1.6023532152175903, + "learning_rate": 1.3032984714400645e-06, + "loss": 0.8724, + "step": 1073 + }, + { + "epoch": 1.5542691751085385, + "grad_norm": 1.5924384593963623, + "learning_rate": 1.2992759452936446e-06, + "loss": 0.8845, + "step": 1074 + }, + { + "epoch": 1.5557163531114329, + "grad_norm": 1.6290538311004639, + "learning_rate": 1.2952534191472247e-06, + "loss": 0.878, + "step": 1075 + }, + { + "epoch": 1.557163531114327, + "grad_norm": 1.6020493507385254, + "learning_rate": 1.2912308930008046e-06, + "loss": 0.8564, + "step": 1076 + }, + { + "epoch": 1.5586107091172214, + "grad_norm": 1.6875399351119995, + "learning_rate": 1.2872083668543848e-06, + "loss": 0.8443, + "step": 1077 + }, + { + "epoch": 1.5600578871201156, + "grad_norm": 1.662969708442688, + "learning_rate": 1.2831858407079647e-06, + "loss": 0.8879, + "step": 1078 + }, + { + "epoch": 1.56150506512301, + "grad_norm": 1.5883301496505737, + "learning_rate": 1.2791633145615448e-06, + "loss": 0.841, + "step": 1079 + }, + { + "epoch": 1.5629522431259044, + "grad_norm": 1.564860224723816, + "learning_rate": 1.275140788415125e-06, + "loss": 0.8378, + "step": 1080 + }, + { + "epoch": 1.5629522431259044, + "eval_loss": 0.9821707606315613, + "eval_runtime": 25.8456, + "eval_samples_per_second": 38.691, + "eval_steps_per_second": 2.438, + "step": 1080 + }, + { + "epoch": 1.5643994211287988, + "grad_norm": 1.690390944480896, + "learning_rate": 1.2711182622687049e-06, + "loss": 0.9134, + "step": 1081 + }, + { + "epoch": 1.5658465991316932, + "grad_norm": 1.5780271291732788, + "learning_rate": 1.267095736122285e-06, + "loss": 0.8815, + "step": 1082 + }, + { + "epoch": 1.5672937771345876, + "grad_norm": 1.6903564929962158, + "learning_rate": 1.263073209975865e-06, + "loss": 0.8806, + "step": 1083 + }, + { + "epoch": 1.568740955137482, + "grad_norm": 1.5610105991363525, + "learning_rate": 1.259050683829445e-06, + "loss": 0.9031, + "step": 1084 + }, + { + "epoch": 1.5701881331403764, + "grad_norm": 1.463962435722351, + "learning_rate": 1.2550281576830252e-06, + "loss": 0.8491, + "step": 1085 + }, + { + "epoch": 1.5716353111432706, + "grad_norm": 1.5153679847717285, + "learning_rate": 1.251005631536605e-06, + "loss": 0.866, + "step": 1086 + }, + { + "epoch": 1.573082489146165, + "grad_norm": 1.6320936679840088, + "learning_rate": 1.2469831053901852e-06, + "loss": 0.8712, + "step": 1087 + }, + { + "epoch": 1.5745296671490592, + "grad_norm": 1.5522252321243286, + "learning_rate": 1.2429605792437651e-06, + "loss": 0.8937, + "step": 1088 + }, + { + "epoch": 1.5759768451519536, + "grad_norm": 1.5487126111984253, + "learning_rate": 1.2389380530973452e-06, + "loss": 0.931, + "step": 1089 + }, + { + "epoch": 1.577424023154848, + "grad_norm": 1.602441430091858, + "learning_rate": 1.2349155269509252e-06, + "loss": 0.8811, + "step": 1090 + }, + { + "epoch": 1.577424023154848, + "eval_loss": 0.982520341873169, + "eval_runtime": 25.7693, + "eval_samples_per_second": 38.806, + "eval_steps_per_second": 2.445, + "step": 1090 + }, + { + "epoch": 1.5788712011577424, + "grad_norm": 1.633206844329834, + "learning_rate": 1.2308930008045053e-06, + "loss": 0.8697, + "step": 1091 + }, + { + "epoch": 1.5803183791606368, + "grad_norm": 1.5206350088119507, + "learning_rate": 1.2268704746580854e-06, + "loss": 0.8645, + "step": 1092 + }, + { + "epoch": 1.5817655571635312, + "grad_norm": 1.5659570693969727, + "learning_rate": 1.2228479485116653e-06, + "loss": 0.8569, + "step": 1093 + }, + { + "epoch": 1.5832127351664256, + "grad_norm": 1.5603435039520264, + "learning_rate": 1.2188254223652455e-06, + "loss": 0.8741, + "step": 1094 + }, + { + "epoch": 1.58465991316932, + "grad_norm": 1.5861403942108154, + "learning_rate": 1.2148028962188254e-06, + "loss": 0.9042, + "step": 1095 + }, + { + "epoch": 1.5861070911722142, + "grad_norm": 1.7639175653457642, + "learning_rate": 1.2107803700724055e-06, + "loss": 0.8556, + "step": 1096 + }, + { + "epoch": 1.5875542691751086, + "grad_norm": 1.5393552780151367, + "learning_rate": 1.2067578439259856e-06, + "loss": 0.8267, + "step": 1097 + }, + { + "epoch": 1.5890014471780027, + "grad_norm": 1.5538761615753174, + "learning_rate": 1.2027353177795658e-06, + "loss": 0.8574, + "step": 1098 + }, + { + "epoch": 1.5904486251808971, + "grad_norm": 1.6127047538757324, + "learning_rate": 1.1987127916331457e-06, + "loss": 0.8667, + "step": 1099 + }, + { + "epoch": 1.5918958031837915, + "grad_norm": 1.6033275127410889, + "learning_rate": 1.1946902654867258e-06, + "loss": 0.8788, + "step": 1100 + }, + { + "epoch": 1.5918958031837915, + "eval_loss": 0.9810674786567688, + "eval_runtime": 25.866, + "eval_samples_per_second": 38.661, + "eval_steps_per_second": 2.436, + "step": 1100 + }, + { + "epoch": 1.593342981186686, + "grad_norm": 1.6297693252563477, + "learning_rate": 1.190667739340306e-06, + "loss": 0.8847, + "step": 1101 + }, + { + "epoch": 1.5947901591895803, + "grad_norm": 1.6360310316085815, + "learning_rate": 1.1866452131938858e-06, + "loss": 0.8544, + "step": 1102 + }, + { + "epoch": 1.5962373371924747, + "grad_norm": 1.541109561920166, + "learning_rate": 1.182622687047466e-06, + "loss": 0.8759, + "step": 1103 + }, + { + "epoch": 1.5976845151953691, + "grad_norm": 1.5267829895019531, + "learning_rate": 1.1786001609010459e-06, + "loss": 0.8544, + "step": 1104 + }, + { + "epoch": 1.5991316931982635, + "grad_norm": 1.610019326210022, + "learning_rate": 1.174577634754626e-06, + "loss": 0.8711, + "step": 1105 + }, + { + "epoch": 1.6005788712011577, + "grad_norm": 1.5717980861663818, + "learning_rate": 1.1705551086082061e-06, + "loss": 0.8596, + "step": 1106 + }, + { + "epoch": 1.6020260492040521, + "grad_norm": 1.5795584917068481, + "learning_rate": 1.166532582461786e-06, + "loss": 0.8682, + "step": 1107 + }, + { + "epoch": 1.6034732272069463, + "grad_norm": 1.5268584489822388, + "learning_rate": 1.1625100563153662e-06, + "loss": 0.8469, + "step": 1108 + }, + { + "epoch": 1.6049204052098407, + "grad_norm": 1.5472923517227173, + "learning_rate": 1.158487530168946e-06, + "loss": 0.8614, + "step": 1109 + }, + { + "epoch": 1.606367583212735, + "grad_norm": 1.5849939584732056, + "learning_rate": 1.1544650040225262e-06, + "loss": 0.8891, + "step": 1110 + }, + { + "epoch": 1.606367583212735, + "eval_loss": 0.9799522161483765, + "eval_runtime": 25.7004, + "eval_samples_per_second": 38.91, + "eval_steps_per_second": 2.451, + "step": 1110 + }, + { + "epoch": 1.6078147612156295, + "grad_norm": 1.5831488370895386, + "learning_rate": 1.1504424778761064e-06, + "loss": 0.8719, + "step": 1111 + }, + { + "epoch": 1.609261939218524, + "grad_norm": 1.5813566446304321, + "learning_rate": 1.1464199517296863e-06, + "loss": 0.8889, + "step": 1112 + }, + { + "epoch": 1.6107091172214183, + "grad_norm": 1.7443983554840088, + "learning_rate": 1.1423974255832664e-06, + "loss": 0.8839, + "step": 1113 + }, + { + "epoch": 1.6121562952243127, + "grad_norm": 1.5268986225128174, + "learning_rate": 1.1383748994368463e-06, + "loss": 0.8705, + "step": 1114 + }, + { + "epoch": 1.613603473227207, + "grad_norm": 1.525639533996582, + "learning_rate": 1.1343523732904264e-06, + "loss": 0.8671, + "step": 1115 + }, + { + "epoch": 1.6150506512301013, + "grad_norm": 1.8092018365859985, + "learning_rate": 1.1303298471440066e-06, + "loss": 0.9264, + "step": 1116 + }, + { + "epoch": 1.6164978292329957, + "grad_norm": 1.5909322500228882, + "learning_rate": 1.1263073209975865e-06, + "loss": 0.8827, + "step": 1117 + }, + { + "epoch": 1.6179450072358899, + "grad_norm": 1.6932170391082764, + "learning_rate": 1.1222847948511666e-06, + "loss": 0.8555, + "step": 1118 + }, + { + "epoch": 1.6193921852387843, + "grad_norm": 1.5742934942245483, + "learning_rate": 1.1182622687047465e-06, + "loss": 0.8946, + "step": 1119 + }, + { + "epoch": 1.6208393632416787, + "grad_norm": 1.5503621101379395, + "learning_rate": 1.1142397425583269e-06, + "loss": 0.8803, + "step": 1120 + }, + { + "epoch": 1.6208393632416787, + "eval_loss": 0.9796671867370605, + "eval_runtime": 25.8219, + "eval_samples_per_second": 38.727, + "eval_steps_per_second": 2.44, + "step": 1120 + }, + { + "epoch": 1.622286541244573, + "grad_norm": 1.5717918872833252, + "learning_rate": 1.1102172164119068e-06, + "loss": 0.8869, + "step": 1121 + }, + { + "epoch": 1.6237337192474675, + "grad_norm": 1.5179367065429688, + "learning_rate": 1.106194690265487e-06, + "loss": 0.8859, + "step": 1122 + }, + { + "epoch": 1.6251808972503619, + "grad_norm": 1.5226870775222778, + "learning_rate": 1.1021721641190668e-06, + "loss": 0.8616, + "step": 1123 + }, + { + "epoch": 1.6266280752532563, + "grad_norm": 1.6265931129455566, + "learning_rate": 1.098149637972647e-06, + "loss": 0.8787, + "step": 1124 + }, + { + "epoch": 1.6280752532561507, + "grad_norm": 1.576897382736206, + "learning_rate": 1.094127111826227e-06, + "loss": 0.8592, + "step": 1125 + }, + { + "epoch": 1.6295224312590448, + "grad_norm": 1.5493236780166626, + "learning_rate": 1.090104585679807e-06, + "loss": 0.8601, + "step": 1126 + }, + { + "epoch": 1.6309696092619392, + "grad_norm": 1.5343093872070312, + "learning_rate": 1.0860820595333871e-06, + "loss": 0.8661, + "step": 1127 + }, + { + "epoch": 1.6324167872648334, + "grad_norm": 1.6531797647476196, + "learning_rate": 1.082059533386967e-06, + "loss": 0.8787, + "step": 1128 + }, + { + "epoch": 1.6338639652677278, + "grad_norm": 1.5565072298049927, + "learning_rate": 1.0780370072405472e-06, + "loss": 0.8843, + "step": 1129 + }, + { + "epoch": 1.6353111432706222, + "grad_norm": 1.5990854501724243, + "learning_rate": 1.0740144810941273e-06, + "loss": 0.8744, + "step": 1130 + }, + { + "epoch": 1.6353111432706222, + "eval_loss": 0.9809547066688538, + "eval_runtime": 26.0231, + "eval_samples_per_second": 38.427, + "eval_steps_per_second": 2.421, + "step": 1130 + }, + { + "epoch": 1.6367583212735166, + "grad_norm": 1.5343475341796875, + "learning_rate": 1.0699919549477072e-06, + "loss": 0.8587, + "step": 1131 + }, + { + "epoch": 1.638205499276411, + "grad_norm": 1.602108120918274, + "learning_rate": 1.0659694288012873e-06, + "loss": 0.889, + "step": 1132 + }, + { + "epoch": 1.6396526772793054, + "grad_norm": 1.6065183877944946, + "learning_rate": 1.0619469026548673e-06, + "loss": 0.8443, + "step": 1133 + }, + { + "epoch": 1.6410998552821998, + "grad_norm": 1.6848286390304565, + "learning_rate": 1.0579243765084474e-06, + "loss": 0.8857, + "step": 1134 + }, + { + "epoch": 1.6425470332850942, + "grad_norm": 1.5931259393692017, + "learning_rate": 1.0539018503620275e-06, + "loss": 0.9061, + "step": 1135 + }, + { + "epoch": 1.6439942112879884, + "grad_norm": 1.5638717412948608, + "learning_rate": 1.0498793242156074e-06, + "loss": 0.8512, + "step": 1136 + }, + { + "epoch": 1.6454413892908828, + "grad_norm": 1.6239174604415894, + "learning_rate": 1.0458567980691876e-06, + "loss": 0.9118, + "step": 1137 + }, + { + "epoch": 1.646888567293777, + "grad_norm": 1.6062713861465454, + "learning_rate": 1.0418342719227675e-06, + "loss": 0.8806, + "step": 1138 + }, + { + "epoch": 1.6483357452966714, + "grad_norm": 1.5737882852554321, + "learning_rate": 1.0378117457763476e-06, + "loss": 0.8444, + "step": 1139 + }, + { + "epoch": 1.6497829232995658, + "grad_norm": 1.5323673486709595, + "learning_rate": 1.0337892196299275e-06, + "loss": 0.8587, + "step": 1140 + }, + { + "epoch": 1.6497829232995658, + "eval_loss": 0.9802071452140808, + "eval_runtime": 25.8256, + "eval_samples_per_second": 38.721, + "eval_steps_per_second": 2.439, + "step": 1140 + }, + { + "epoch": 1.6512301013024602, + "grad_norm": 1.5475029945373535, + "learning_rate": 1.0297666934835076e-06, + "loss": 0.886, + "step": 1141 + }, + { + "epoch": 1.6526772793053546, + "grad_norm": 1.5495952367782593, + "learning_rate": 1.0257441673370878e-06, + "loss": 0.8586, + "step": 1142 + }, + { + "epoch": 1.654124457308249, + "grad_norm": 1.6458721160888672, + "learning_rate": 1.021721641190668e-06, + "loss": 0.8912, + "step": 1143 + }, + { + "epoch": 1.6555716353111434, + "grad_norm": 1.5719788074493408, + "learning_rate": 1.017699115044248e-06, + "loss": 0.86, + "step": 1144 + }, + { + "epoch": 1.6570188133140378, + "grad_norm": 1.574139952659607, + "learning_rate": 1.013676588897828e-06, + "loss": 0.8846, + "step": 1145 + }, + { + "epoch": 1.658465991316932, + "grad_norm": 1.6863696575164795, + "learning_rate": 1.009654062751408e-06, + "loss": 0.8461, + "step": 1146 + }, + { + "epoch": 1.6599131693198264, + "grad_norm": 1.5795451402664185, + "learning_rate": 1.005631536604988e-06, + "loss": 0.8725, + "step": 1147 + }, + { + "epoch": 1.6613603473227205, + "grad_norm": 1.6984606981277466, + "learning_rate": 1.0016090104585681e-06, + "loss": 0.8959, + "step": 1148 + }, + { + "epoch": 1.662807525325615, + "grad_norm": 1.682733416557312, + "learning_rate": 9.97586484312148e-07, + "loss": 0.8583, + "step": 1149 + }, + { + "epoch": 1.6642547033285093, + "grad_norm": 1.607202172279358, + "learning_rate": 9.935639581657282e-07, + "loss": 0.8601, + "step": 1150 + }, + { + "epoch": 1.6642547033285093, + "eval_loss": 0.9815489053726196, + "eval_runtime": 25.9889, + "eval_samples_per_second": 38.478, + "eval_steps_per_second": 2.424, + "step": 1150 + }, + { + "epoch": 1.6657018813314037, + "grad_norm": 1.6263314485549927, + "learning_rate": 9.895414320193083e-07, + "loss": 0.8728, + "step": 1151 + }, + { + "epoch": 1.6671490593342981, + "grad_norm": 1.5998060703277588, + "learning_rate": 9.855189058728882e-07, + "loss": 0.8781, + "step": 1152 + }, + { + "epoch": 1.6685962373371925, + "grad_norm": 1.6003531217575073, + "learning_rate": 9.814963797264683e-07, + "loss": 0.8686, + "step": 1153 + }, + { + "epoch": 1.670043415340087, + "grad_norm": 1.5383695363998413, + "learning_rate": 9.774738535800482e-07, + "loss": 0.8583, + "step": 1154 + }, + { + "epoch": 1.6714905933429813, + "grad_norm": 1.5972706079483032, + "learning_rate": 9.734513274336284e-07, + "loss": 0.8915, + "step": 1155 + }, + { + "epoch": 1.6729377713458755, + "grad_norm": 1.6024503707885742, + "learning_rate": 9.694288012872085e-07, + "loss": 0.8842, + "step": 1156 + }, + { + "epoch": 1.67438494934877, + "grad_norm": 1.5077908039093018, + "learning_rate": 9.654062751407884e-07, + "loss": 0.9109, + "step": 1157 + }, + { + "epoch": 1.675832127351664, + "grad_norm": 1.5892987251281738, + "learning_rate": 9.613837489943685e-07, + "loss": 0.8554, + "step": 1158 + }, + { + "epoch": 1.6772793053545585, + "grad_norm": 1.5132979154586792, + "learning_rate": 9.573612228479485e-07, + "loss": 0.8424, + "step": 1159 + }, + { + "epoch": 1.678726483357453, + "grad_norm": 1.559235692024231, + "learning_rate": 9.533386967015286e-07, + "loss": 0.8469, + "step": 1160 + }, + { + "epoch": 1.678726483357453, + "eval_loss": 0.9792452454566956, + "eval_runtime": 25.8401, + "eval_samples_per_second": 38.699, + "eval_steps_per_second": 2.438, + "step": 1160 + }, + { + "epoch": 1.6801736613603473, + "grad_norm": 1.5679746866226196, + "learning_rate": 9.493161705551086e-07, + "loss": 0.8348, + "step": 1161 + }, + { + "epoch": 1.6816208393632417, + "grad_norm": 1.5585919618606567, + "learning_rate": 9.452936444086886e-07, + "loss": 0.8399, + "step": 1162 + }, + { + "epoch": 1.683068017366136, + "grad_norm": 1.607049584388733, + "learning_rate": 9.412711182622688e-07, + "loss": 0.8437, + "step": 1163 + }, + { + "epoch": 1.6845151953690305, + "grad_norm": 1.5190579891204834, + "learning_rate": 9.372485921158489e-07, + "loss": 0.8597, + "step": 1164 + }, + { + "epoch": 1.685962373371925, + "grad_norm": 1.601736307144165, + "learning_rate": 9.332260659694289e-07, + "loss": 0.8707, + "step": 1165 + }, + { + "epoch": 1.687409551374819, + "grad_norm": 1.6469128131866455, + "learning_rate": 9.292035398230089e-07, + "loss": 0.9113, + "step": 1166 + }, + { + "epoch": 1.6888567293777135, + "grad_norm": 1.6424798965454102, + "learning_rate": 9.251810136765891e-07, + "loss": 0.8596, + "step": 1167 + }, + { + "epoch": 1.6903039073806077, + "grad_norm": 1.683808445930481, + "learning_rate": 9.211584875301691e-07, + "loss": 0.9157, + "step": 1168 + }, + { + "epoch": 1.691751085383502, + "grad_norm": 1.6408365964889526, + "learning_rate": 9.171359613837491e-07, + "loss": 0.853, + "step": 1169 + }, + { + "epoch": 1.6931982633863965, + "grad_norm": 1.7329471111297607, + "learning_rate": 9.131134352373291e-07, + "loss": 0.8696, + "step": 1170 + }, + { + "epoch": 1.6931982633863965, + "eval_loss": 0.9784539937973022, + "eval_runtime": 26.0423, + "eval_samples_per_second": 38.399, + "eval_steps_per_second": 2.419, + "step": 1170 + }, + { + "epoch": 1.6946454413892909, + "grad_norm": 1.599411964416504, + "learning_rate": 9.090909090909091e-07, + "loss": 0.8722, + "step": 1171 + }, + { + "epoch": 1.6960926193921853, + "grad_norm": 1.5941879749298096, + "learning_rate": 9.050683829444893e-07, + "loss": 0.8576, + "step": 1172 + }, + { + "epoch": 1.6975397973950797, + "grad_norm": 1.5630899667739868, + "learning_rate": 9.010458567980693e-07, + "loss": 0.8809, + "step": 1173 + }, + { + "epoch": 1.698986975397974, + "grad_norm": 1.5148115158081055, + "learning_rate": 8.970233306516493e-07, + "loss": 0.8841, + "step": 1174 + }, + { + "epoch": 1.7004341534008685, + "grad_norm": 1.5821913480758667, + "learning_rate": 8.930008045052293e-07, + "loss": 0.8772, + "step": 1175 + }, + { + "epoch": 1.7018813314037626, + "grad_norm": 1.7660754919052124, + "learning_rate": 8.889782783588094e-07, + "loss": 0.8731, + "step": 1176 + }, + { + "epoch": 1.703328509406657, + "grad_norm": 1.5370192527770996, + "learning_rate": 8.849557522123895e-07, + "loss": 0.8584, + "step": 1177 + }, + { + "epoch": 1.7047756874095512, + "grad_norm": 1.5817389488220215, + "learning_rate": 8.809332260659695e-07, + "loss": 0.8786, + "step": 1178 + }, + { + "epoch": 1.7062228654124456, + "grad_norm": 1.5897088050842285, + "learning_rate": 8.769106999195495e-07, + "loss": 0.8448, + "step": 1179 + }, + { + "epoch": 1.70767004341534, + "grad_norm": 1.5718865394592285, + "learning_rate": 8.728881737731296e-07, + "loss": 0.8432, + "step": 1180 + }, + { + "epoch": 1.70767004341534, + "eval_loss": 0.978662371635437, + "eval_runtime": 25.9113, + "eval_samples_per_second": 38.593, + "eval_steps_per_second": 2.431, + "step": 1180 + }, + { + "epoch": 1.7091172214182344, + "grad_norm": 1.6017844676971436, + "learning_rate": 8.688656476267096e-07, + "loss": 0.868, + "step": 1181 + }, + { + "epoch": 1.7105643994211288, + "grad_norm": 1.6823580265045166, + "learning_rate": 8.648431214802896e-07, + "loss": 0.8621, + "step": 1182 + }, + { + "epoch": 1.7120115774240232, + "grad_norm": 1.6109862327575684, + "learning_rate": 8.608205953338697e-07, + "loss": 0.8614, + "step": 1183 + }, + { + "epoch": 1.7134587554269176, + "grad_norm": 1.5950835943222046, + "learning_rate": 8.567980691874497e-07, + "loss": 0.8481, + "step": 1184 + }, + { + "epoch": 1.714905933429812, + "grad_norm": 1.7378734350204468, + "learning_rate": 8.527755430410298e-07, + "loss": 0.8863, + "step": 1185 + }, + { + "epoch": 1.7163531114327062, + "grad_norm": 1.6182748079299927, + "learning_rate": 8.487530168946098e-07, + "loss": 0.8982, + "step": 1186 + }, + { + "epoch": 1.7178002894356006, + "grad_norm": 1.61431884765625, + "learning_rate": 8.4473049074819e-07, + "loss": 0.8855, + "step": 1187 + }, + { + "epoch": 1.7192474674384948, + "grad_norm": 1.593196153640747, + "learning_rate": 8.4070796460177e-07, + "loss": 0.8745, + "step": 1188 + }, + { + "epoch": 1.7206946454413892, + "grad_norm": 1.5465645790100098, + "learning_rate": 8.366854384553501e-07, + "loss": 0.8863, + "step": 1189 + }, + { + "epoch": 1.7221418234442836, + "grad_norm": 1.6166539192199707, + "learning_rate": 8.326629123089301e-07, + "loss": 0.8826, + "step": 1190 + }, + { + "epoch": 1.7221418234442836, + "eval_loss": 0.977293848991394, + "eval_runtime": 25.8688, + "eval_samples_per_second": 38.657, + "eval_steps_per_second": 2.435, + "step": 1190 + }, + { + "epoch": 1.723589001447178, + "grad_norm": 1.5473971366882324, + "learning_rate": 8.286403861625101e-07, + "loss": 0.9087, + "step": 1191 + }, + { + "epoch": 1.7250361794500724, + "grad_norm": 1.557051420211792, + "learning_rate": 8.246178600160902e-07, + "loss": 0.8697, + "step": 1192 + }, + { + "epoch": 1.7264833574529668, + "grad_norm": 1.473290205001831, + "learning_rate": 8.205953338696703e-07, + "loss": 0.8659, + "step": 1193 + }, + { + "epoch": 1.7279305354558612, + "grad_norm": 1.5358150005340576, + "learning_rate": 8.165728077232503e-07, + "loss": 0.8668, + "step": 1194 + }, + { + "epoch": 1.7293777134587556, + "grad_norm": 1.5885599851608276, + "learning_rate": 8.125502815768303e-07, + "loss": 0.8705, + "step": 1195 + }, + { + "epoch": 1.7308248914616498, + "grad_norm": 1.5535321235656738, + "learning_rate": 8.085277554304103e-07, + "loss": 0.8895, + "step": 1196 + }, + { + "epoch": 1.7322720694645442, + "grad_norm": 1.5673221349716187, + "learning_rate": 8.045052292839905e-07, + "loss": 0.8685, + "step": 1197 + }, + { + "epoch": 1.7337192474674383, + "grad_norm": 1.768881916999817, + "learning_rate": 8.004827031375705e-07, + "loss": 0.9055, + "step": 1198 + }, + { + "epoch": 1.7351664254703327, + "grad_norm": 1.597657322883606, + "learning_rate": 7.964601769911505e-07, + "loss": 0.8558, + "step": 1199 + }, + { + "epoch": 1.7366136034732271, + "grad_norm": 1.5506863594055176, + "learning_rate": 7.924376508447305e-07, + "loss": 0.8852, + "step": 1200 + }, + { + "epoch": 1.7366136034732271, + "eval_loss": 0.9770252704620361, + "eval_runtime": 25.9917, + "eval_samples_per_second": 38.474, + "eval_steps_per_second": 2.424, + "step": 1200 + }, + { + "epoch": 1.7380607814761215, + "grad_norm": 1.6115679740905762, + "learning_rate": 7.884151246983105e-07, + "loss": 0.85, + "step": 1201 + }, + { + "epoch": 1.739507959479016, + "grad_norm": 1.662935733795166, + "learning_rate": 7.843925985518907e-07, + "loss": 0.8807, + "step": 1202 + }, + { + "epoch": 1.7409551374819103, + "grad_norm": 1.617238998413086, + "learning_rate": 7.803700724054707e-07, + "loss": 0.8544, + "step": 1203 + }, + { + "epoch": 1.7424023154848047, + "grad_norm": 1.5308291912078857, + "learning_rate": 7.763475462590507e-07, + "loss": 0.8601, + "step": 1204 + }, + { + "epoch": 1.7438494934876991, + "grad_norm": 1.603887915611267, + "learning_rate": 7.723250201126307e-07, + "loss": 0.889, + "step": 1205 + }, + { + "epoch": 1.7452966714905933, + "grad_norm": 1.546451449394226, + "learning_rate": 7.683024939662108e-07, + "loss": 0.8936, + "step": 1206 + }, + { + "epoch": 1.7467438494934877, + "grad_norm": 1.5847758054733276, + "learning_rate": 7.642799678197908e-07, + "loss": 0.866, + "step": 1207 + }, + { + "epoch": 1.7481910274963819, + "grad_norm": 1.5509288311004639, + "learning_rate": 7.602574416733709e-07, + "loss": 0.8593, + "step": 1208 + }, + { + "epoch": 1.7496382054992763, + "grad_norm": 1.615058183670044, + "learning_rate": 7.56234915526951e-07, + "loss": 0.8949, + "step": 1209 + }, + { + "epoch": 1.7510853835021707, + "grad_norm": 1.5887871980667114, + "learning_rate": 7.522123893805311e-07, + "loss": 0.8401, + "step": 1210 + }, + { + "epoch": 1.7510853835021707, + "eval_loss": 0.9778628349304199, + "eval_runtime": 25.8847, + "eval_samples_per_second": 38.633, + "eval_steps_per_second": 2.434, + "step": 1210 + }, + { + "epoch": 1.752532561505065, + "grad_norm": 1.772998332977295, + "learning_rate": 7.481898632341112e-07, + "loss": 0.9199, + "step": 1211 + }, + { + "epoch": 1.7539797395079595, + "grad_norm": 1.57089364528656, + "learning_rate": 7.441673370876912e-07, + "loss": 0.864, + "step": 1212 + }, + { + "epoch": 1.755426917510854, + "grad_norm": 1.603686809539795, + "learning_rate": 7.401448109412712e-07, + "loss": 0.8511, + "step": 1213 + }, + { + "epoch": 1.7568740955137483, + "grad_norm": 1.6855931282043457, + "learning_rate": 7.361222847948512e-07, + "loss": 0.8688, + "step": 1214 + }, + { + "epoch": 1.7583212735166427, + "grad_norm": 1.553842544555664, + "learning_rate": 7.320997586484313e-07, + "loss": 0.8604, + "step": 1215 + }, + { + "epoch": 1.7597684515195369, + "grad_norm": 1.6918758153915405, + "learning_rate": 7.280772325020113e-07, + "loss": 0.8767, + "step": 1216 + }, + { + "epoch": 1.7612156295224313, + "grad_norm": 1.5926964282989502, + "learning_rate": 7.240547063555914e-07, + "loss": 0.8871, + "step": 1217 + }, + { + "epoch": 1.7626628075253257, + "grad_norm": 1.5411027669906616, + "learning_rate": 7.200321802091714e-07, + "loss": 0.8663, + "step": 1218 + }, + { + "epoch": 1.7641099855282198, + "grad_norm": 1.595458745956421, + "learning_rate": 7.160096540627515e-07, + "loss": 0.8645, + "step": 1219 + }, + { + "epoch": 1.7655571635311142, + "grad_norm": 1.5208485126495361, + "learning_rate": 7.119871279163315e-07, + "loss": 0.8804, + "step": 1220 + }, + { + "epoch": 1.7655571635311142, + "eval_loss": 0.9778464436531067, + "eval_runtime": 25.9514, + "eval_samples_per_second": 38.534, + "eval_steps_per_second": 2.428, + "step": 1220 + }, + { + "epoch": 1.7670043415340086, + "grad_norm": 1.6658185720443726, + "learning_rate": 7.079646017699115e-07, + "loss": 0.8534, + "step": 1221 + }, + { + "epoch": 1.768451519536903, + "grad_norm": 1.6133363246917725, + "learning_rate": 7.039420756234916e-07, + "loss": 0.8685, + "step": 1222 + }, + { + "epoch": 1.7698986975397974, + "grad_norm": 1.5848267078399658, + "learning_rate": 6.999195494770717e-07, + "loss": 0.8888, + "step": 1223 + }, + { + "epoch": 1.7713458755426919, + "grad_norm": 1.6191439628601074, + "learning_rate": 6.958970233306517e-07, + "loss": 0.8707, + "step": 1224 + }, + { + "epoch": 1.7727930535455863, + "grad_norm": 1.5733712911605835, + "learning_rate": 6.918744971842317e-07, + "loss": 0.8581, + "step": 1225 + }, + { + "epoch": 1.7742402315484804, + "grad_norm": 1.6595110893249512, + "learning_rate": 6.878519710378117e-07, + "loss": 0.8882, + "step": 1226 + }, + { + "epoch": 1.7756874095513748, + "grad_norm": 1.6195080280303955, + "learning_rate": 6.838294448913918e-07, + "loss": 0.8386, + "step": 1227 + }, + { + "epoch": 1.7771345875542692, + "grad_norm": 1.535504937171936, + "learning_rate": 6.798069187449719e-07, + "loss": 0.8556, + "step": 1228 + }, + { + "epoch": 1.7785817655571634, + "grad_norm": 1.5953824520111084, + "learning_rate": 6.757843925985519e-07, + "loss": 0.8291, + "step": 1229 + }, + { + "epoch": 1.7800289435600578, + "grad_norm": 1.5881785154342651, + "learning_rate": 6.717618664521319e-07, + "loss": 0.8765, + "step": 1230 + }, + { + "epoch": 1.7800289435600578, + "eval_loss": 0.9774501919746399, + "eval_runtime": 26.0326, + "eval_samples_per_second": 38.413, + "eval_steps_per_second": 2.42, + "step": 1230 + }, + { + "epoch": 1.7814761215629522, + "grad_norm": 1.633536696434021, + "learning_rate": 6.677393403057121e-07, + "loss": 0.8795, + "step": 1231 + }, + { + "epoch": 1.7829232995658466, + "grad_norm": 1.5229909420013428, + "learning_rate": 6.637168141592922e-07, + "loss": 0.8595, + "step": 1232 + }, + { + "epoch": 1.784370477568741, + "grad_norm": 1.6129873991012573, + "learning_rate": 6.596942880128722e-07, + "loss": 0.8634, + "step": 1233 + }, + { + "epoch": 1.7858176555716354, + "grad_norm": 1.5728979110717773, + "learning_rate": 6.556717618664522e-07, + "loss": 0.8685, + "step": 1234 + }, + { + "epoch": 1.7872648335745298, + "grad_norm": 1.6559734344482422, + "learning_rate": 6.516492357200322e-07, + "loss": 0.8849, + "step": 1235 + }, + { + "epoch": 1.788712011577424, + "grad_norm": 1.5157029628753662, + "learning_rate": 6.476267095736124e-07, + "loss": 0.868, + "step": 1236 + }, + { + "epoch": 1.7901591895803184, + "grad_norm": 1.5743296146392822, + "learning_rate": 6.436041834271924e-07, + "loss": 0.8446, + "step": 1237 + }, + { + "epoch": 1.7916063675832128, + "grad_norm": 1.5727088451385498, + "learning_rate": 6.395816572807724e-07, + "loss": 0.8703, + "step": 1238 + }, + { + "epoch": 1.793053545586107, + "grad_norm": 1.6340490579605103, + "learning_rate": 6.355591311343524e-07, + "loss": 0.8852, + "step": 1239 + }, + { + "epoch": 1.7945007235890014, + "grad_norm": 1.592427134513855, + "learning_rate": 6.315366049879324e-07, + "loss": 0.8611, + "step": 1240 + }, + { + "epoch": 1.7945007235890014, + "eval_loss": 0.9771277904510498, + "eval_runtime": 25.8154, + "eval_samples_per_second": 38.737, + "eval_steps_per_second": 2.44, + "step": 1240 + }, + { + "epoch": 1.7959479015918958, + "grad_norm": 1.6300837993621826, + "learning_rate": 6.275140788415126e-07, + "loss": 0.8471, + "step": 1241 + }, + { + "epoch": 1.7973950795947902, + "grad_norm": 1.614335298538208, + "learning_rate": 6.234915526950926e-07, + "loss": 0.888, + "step": 1242 + }, + { + "epoch": 1.7988422575976846, + "grad_norm": 1.5964536666870117, + "learning_rate": 6.194690265486726e-07, + "loss": 0.8876, + "step": 1243 + }, + { + "epoch": 1.800289435600579, + "grad_norm": 1.601259708404541, + "learning_rate": 6.154465004022526e-07, + "loss": 0.8572, + "step": 1244 + }, + { + "epoch": 1.8017366136034734, + "grad_norm": 1.6280089616775513, + "learning_rate": 6.114239742558327e-07, + "loss": 0.8606, + "step": 1245 + }, + { + "epoch": 1.8031837916063675, + "grad_norm": 1.6801878213882446, + "learning_rate": 6.074014481094127e-07, + "loss": 0.8871, + "step": 1246 + }, + { + "epoch": 1.804630969609262, + "grad_norm": 1.5877913236618042, + "learning_rate": 6.033789219629928e-07, + "loss": 0.8912, + "step": 1247 + }, + { + "epoch": 1.8060781476121563, + "grad_norm": 1.6258810758590698, + "learning_rate": 5.993563958165728e-07, + "loss": 0.8619, + "step": 1248 + }, + { + "epoch": 1.8075253256150505, + "grad_norm": 1.5987861156463623, + "learning_rate": 5.95333869670153e-07, + "loss": 0.8819, + "step": 1249 + }, + { + "epoch": 1.808972503617945, + "grad_norm": 1.5958012342453003, + "learning_rate": 5.91311343523733e-07, + "loss": 0.8795, + "step": 1250 + }, + { + "epoch": 1.808972503617945, + "eval_loss": 0.9765346646308899, + "eval_runtime": 26.1073, + "eval_samples_per_second": 38.303, + "eval_steps_per_second": 2.413, + "step": 1250 + }, + { + "epoch": 1.8104196816208393, + "grad_norm": 1.5270453691482544, + "learning_rate": 5.87288817377313e-07, + "loss": 0.8617, + "step": 1251 + }, + { + "epoch": 1.8118668596237337, + "grad_norm": 1.575887680053711, + "learning_rate": 5.83266291230893e-07, + "loss": 0.8664, + "step": 1252 + }, + { + "epoch": 1.8133140376266281, + "grad_norm": 1.6458449363708496, + "learning_rate": 5.79243765084473e-07, + "loss": 0.8628, + "step": 1253 + }, + { + "epoch": 1.8147612156295225, + "grad_norm": 1.5487672090530396, + "learning_rate": 5.752212389380532e-07, + "loss": 0.871, + "step": 1254 + }, + { + "epoch": 1.816208393632417, + "grad_norm": 1.5608482360839844, + "learning_rate": 5.711987127916332e-07, + "loss": 0.898, + "step": 1255 + }, + { + "epoch": 1.817655571635311, + "grad_norm": 1.471292495727539, + "learning_rate": 5.671761866452132e-07, + "loss": 0.8563, + "step": 1256 + }, + { + "epoch": 1.8191027496382055, + "grad_norm": 1.6494553089141846, + "learning_rate": 5.631536604987932e-07, + "loss": 0.8673, + "step": 1257 + }, + { + "epoch": 1.8205499276411, + "grad_norm": 1.580310583114624, + "learning_rate": 5.591311343523733e-07, + "loss": 0.8655, + "step": 1258 + }, + { + "epoch": 1.821997105643994, + "grad_norm": 1.5952807664871216, + "learning_rate": 5.551086082059534e-07, + "loss": 0.8646, + "step": 1259 + }, + { + "epoch": 1.8234442836468885, + "grad_norm": 1.6021016836166382, + "learning_rate": 5.510860820595334e-07, + "loss": 0.8906, + "step": 1260 + }, + { + "epoch": 1.8234442836468885, + "eval_loss": 0.975847601890564, + "eval_runtime": 25.9152, + "eval_samples_per_second": 38.587, + "eval_steps_per_second": 2.431, + "step": 1260 + }, + { + "epoch": 1.8248914616497829, + "grad_norm": 1.6215472221374512, + "learning_rate": 5.470635559131135e-07, + "loss": 0.8775, + "step": 1261 + }, + { + "epoch": 1.8263386396526773, + "grad_norm": 1.6024051904678345, + "learning_rate": 5.430410297666936e-07, + "loss": 0.8467, + "step": 1262 + }, + { + "epoch": 1.8277858176555717, + "grad_norm": 1.5474594831466675, + "learning_rate": 5.390185036202736e-07, + "loss": 0.8411, + "step": 1263 + }, + { + "epoch": 1.829232995658466, + "grad_norm": 1.5789331197738647, + "learning_rate": 5.349959774738536e-07, + "loss": 0.8808, + "step": 1264 + }, + { + "epoch": 1.8306801736613605, + "grad_norm": 1.5437642335891724, + "learning_rate": 5.309734513274336e-07, + "loss": 0.8633, + "step": 1265 + }, + { + "epoch": 1.8321273516642547, + "grad_norm": 1.589099407196045, + "learning_rate": 5.269509251810138e-07, + "loss": 0.8466, + "step": 1266 + }, + { + "epoch": 1.833574529667149, + "grad_norm": 1.5559078454971313, + "learning_rate": 5.229283990345938e-07, + "loss": 0.8754, + "step": 1267 + }, + { + "epoch": 1.8350217076700435, + "grad_norm": 1.6219558715820312, + "learning_rate": 5.189058728881738e-07, + "loss": 0.8313, + "step": 1268 + }, + { + "epoch": 1.8364688856729376, + "grad_norm": 1.639571189880371, + "learning_rate": 5.148833467417538e-07, + "loss": 0.8791, + "step": 1269 + }, + { + "epoch": 1.837916063675832, + "grad_norm": 1.7509524822235107, + "learning_rate": 5.10860820595334e-07, + "loss": 0.8943, + "step": 1270 + }, + { + "epoch": 1.837916063675832, + "eval_loss": 0.9765909314155579, + "eval_runtime": 26.1083, + "eval_samples_per_second": 38.302, + "eval_steps_per_second": 2.413, + "step": 1270 + }, + { + "epoch": 1.8393632416787264, + "grad_norm": 1.5593583583831787, + "learning_rate": 5.06838294448914e-07, + "loss": 0.8662, + "step": 1271 + }, + { + "epoch": 1.8408104196816208, + "grad_norm": 1.6008491516113281, + "learning_rate": 5.02815768302494e-07, + "loss": 0.8895, + "step": 1272 + }, + { + "epoch": 1.8422575976845152, + "grad_norm": 1.5698717832565308, + "learning_rate": 4.98793242156074e-07, + "loss": 0.857, + "step": 1273 + }, + { + "epoch": 1.8437047756874096, + "grad_norm": 1.6328340768814087, + "learning_rate": 4.947707160096541e-07, + "loss": 0.8832, + "step": 1274 + }, + { + "epoch": 1.845151953690304, + "grad_norm": 1.5507820844650269, + "learning_rate": 4.907481898632342e-07, + "loss": 0.8703, + "step": 1275 + }, + { + "epoch": 1.8465991316931982, + "grad_norm": 1.6311931610107422, + "learning_rate": 4.867256637168142e-07, + "loss": 0.8491, + "step": 1276 + }, + { + "epoch": 1.8480463096960926, + "grad_norm": 1.5922417640686035, + "learning_rate": 4.827031375703942e-07, + "loss": 0.8691, + "step": 1277 + }, + { + "epoch": 1.849493487698987, + "grad_norm": 1.4700504541397095, + "learning_rate": 4.786806114239742e-07, + "loss": 0.8801, + "step": 1278 + }, + { + "epoch": 1.8509406657018812, + "grad_norm": 1.6556566953659058, + "learning_rate": 4.746580852775543e-07, + "loss": 0.8594, + "step": 1279 + }, + { + "epoch": 1.8523878437047756, + "grad_norm": 1.5696501731872559, + "learning_rate": 4.706355591311344e-07, + "loss": 0.8682, + "step": 1280 + }, + { + "epoch": 1.8523878437047756, + "eval_loss": 0.9761188626289368, + "eval_runtime": 25.9086, + "eval_samples_per_second": 38.597, + "eval_steps_per_second": 2.432, + "step": 1280 + }, + { + "epoch": 1.85383502170767, + "grad_norm": 1.5777618885040283, + "learning_rate": 4.6661303298471445e-07, + "loss": 0.8454, + "step": 1281 + }, + { + "epoch": 1.8552821997105644, + "grad_norm": 1.5324792861938477, + "learning_rate": 4.6259050683829453e-07, + "loss": 0.8363, + "step": 1282 + }, + { + "epoch": 1.8567293777134588, + "grad_norm": 1.6020021438598633, + "learning_rate": 4.5856798069187455e-07, + "loss": 0.8665, + "step": 1283 + }, + { + "epoch": 1.8581765557163532, + "grad_norm": 1.6646153926849365, + "learning_rate": 4.5454545454545457e-07, + "loss": 0.8926, + "step": 1284 + }, + { + "epoch": 1.8596237337192476, + "grad_norm": 1.72982919216156, + "learning_rate": 4.5052292839903465e-07, + "loss": 0.8532, + "step": 1285 + }, + { + "epoch": 1.8610709117221418, + "grad_norm": 1.6174458265304565, + "learning_rate": 4.4650040225261467e-07, + "loss": 0.8763, + "step": 1286 + }, + { + "epoch": 1.8625180897250362, + "grad_norm": 1.4950498342514038, + "learning_rate": 4.4247787610619474e-07, + "loss": 0.848, + "step": 1287 + }, + { + "epoch": 1.8639652677279306, + "grad_norm": 1.6399447917938232, + "learning_rate": 4.3845534995977477e-07, + "loss": 0.8663, + "step": 1288 + }, + { + "epoch": 1.8654124457308248, + "grad_norm": 1.5628941059112549, + "learning_rate": 4.344328238133548e-07, + "loss": 0.8834, + "step": 1289 + }, + { + "epoch": 1.8668596237337192, + "grad_norm": 1.5437206029891968, + "learning_rate": 4.3041029766693486e-07, + "loss": 0.8897, + "step": 1290 + }, + { + "epoch": 1.8668596237337192, + "eval_loss": 0.9763774871826172, + "eval_runtime": 26.0791, + "eval_samples_per_second": 38.345, + "eval_steps_per_second": 2.416, + "step": 1290 + }, + { + "epoch": 1.8683068017366136, + "grad_norm": 1.6498969793319702, + "learning_rate": 4.263877715205149e-07, + "loss": 0.841, + "step": 1291 + }, + { + "epoch": 1.869753979739508, + "grad_norm": 1.5987532138824463, + "learning_rate": 4.22365245374095e-07, + "loss": 0.8556, + "step": 1292 + }, + { + "epoch": 1.8712011577424024, + "grad_norm": 1.6004939079284668, + "learning_rate": 4.1834271922767503e-07, + "loss": 0.8762, + "step": 1293 + }, + { + "epoch": 1.8726483357452968, + "grad_norm": 1.5418647527694702, + "learning_rate": 4.1432019308125506e-07, + "loss": 0.8455, + "step": 1294 + }, + { + "epoch": 1.8740955137481912, + "grad_norm": 1.6261379718780518, + "learning_rate": 4.1029766693483513e-07, + "loss": 0.9002, + "step": 1295 + }, + { + "epoch": 1.8755426917510853, + "grad_norm": 1.6570844650268555, + "learning_rate": 4.0627514078841515e-07, + "loss": 0.8792, + "step": 1296 + }, + { + "epoch": 1.8769898697539797, + "grad_norm": 1.5834189653396606, + "learning_rate": 4.0225261464199523e-07, + "loss": 0.8761, + "step": 1297 + }, + { + "epoch": 1.8784370477568741, + "grad_norm": 1.6012641191482544, + "learning_rate": 3.9823008849557525e-07, + "loss": 0.8468, + "step": 1298 + }, + { + "epoch": 1.8798842257597683, + "grad_norm": 1.5698151588439941, + "learning_rate": 3.9420756234915527e-07, + "loss": 0.8669, + "step": 1299 + }, + { + "epoch": 1.8813314037626627, + "grad_norm": 1.5631901025772095, + "learning_rate": 3.9018503620273535e-07, + "loss": 0.8742, + "step": 1300 + }, + { + "epoch": 1.8813314037626627, + "eval_loss": 0.9766185283660889, + "eval_runtime": 25.8927, + "eval_samples_per_second": 38.621, + "eval_steps_per_second": 2.433, + "step": 1300 + }, + { + "epoch": 1.8827785817655571, + "grad_norm": 1.5822370052337646, + "learning_rate": 3.8616251005631537e-07, + "loss": 0.8775, + "step": 1301 + }, + { + "epoch": 1.8842257597684515, + "grad_norm": 1.739130973815918, + "learning_rate": 3.821399839098954e-07, + "loss": 0.8917, + "step": 1302 + }, + { + "epoch": 1.885672937771346, + "grad_norm": 1.565758466720581, + "learning_rate": 3.781174577634755e-07, + "loss": 0.8703, + "step": 1303 + }, + { + "epoch": 1.8871201157742403, + "grad_norm": 1.6006062030792236, + "learning_rate": 3.740949316170556e-07, + "loss": 0.8605, + "step": 1304 + }, + { + "epoch": 1.8885672937771347, + "grad_norm": 1.6097816228866577, + "learning_rate": 3.700724054706356e-07, + "loss": 0.8486, + "step": 1305 + }, + { + "epoch": 1.890014471780029, + "grad_norm": 1.5582165718078613, + "learning_rate": 3.6604987932421563e-07, + "loss": 0.8331, + "step": 1306 + }, + { + "epoch": 1.8914616497829233, + "grad_norm": 1.5906857252120972, + "learning_rate": 3.620273531777957e-07, + "loss": 0.8695, + "step": 1307 + }, + { + "epoch": 1.8929088277858177, + "grad_norm": 1.5649751424789429, + "learning_rate": 3.5800482703137573e-07, + "loss": 0.87, + "step": 1308 + }, + { + "epoch": 1.8943560057887119, + "grad_norm": 1.571973204612732, + "learning_rate": 3.5398230088495575e-07, + "loss": 0.8755, + "step": 1309 + }, + { + "epoch": 1.8958031837916063, + "grad_norm": 1.5199156999588013, + "learning_rate": 3.4995977473853583e-07, + "loss": 0.8709, + "step": 1310 + }, + { + "epoch": 1.8958031837916063, + "eval_loss": 0.9759008884429932, + "eval_runtime": 25.8688, + "eval_samples_per_second": 38.657, + "eval_steps_per_second": 2.435, + "step": 1310 + }, + { + "epoch": 1.8972503617945007, + "grad_norm": 1.60419499874115, + "learning_rate": 3.4593724859211585e-07, + "loss": 0.8648, + "step": 1311 + }, + { + "epoch": 1.898697539797395, + "grad_norm": 1.4729905128479004, + "learning_rate": 3.419147224456959e-07, + "loss": 0.8495, + "step": 1312 + }, + { + "epoch": 1.9001447178002895, + "grad_norm": 1.7213385105133057, + "learning_rate": 3.3789219629927595e-07, + "loss": 0.88, + "step": 1313 + }, + { + "epoch": 1.9015918958031839, + "grad_norm": 1.616490364074707, + "learning_rate": 3.338696701528561e-07, + "loss": 0.8703, + "step": 1314 + }, + { + "epoch": 1.9030390738060783, + "grad_norm": 1.6348494291305542, + "learning_rate": 3.298471440064361e-07, + "loss": 0.8844, + "step": 1315 + }, + { + "epoch": 1.9044862518089725, + "grad_norm": 1.6609773635864258, + "learning_rate": 3.258246178600161e-07, + "loss": 0.8496, + "step": 1316 + }, + { + "epoch": 1.9059334298118669, + "grad_norm": 1.5871094465255737, + "learning_rate": 3.218020917135962e-07, + "loss": 0.8451, + "step": 1317 + }, + { + "epoch": 1.9073806078147613, + "grad_norm": 1.6062517166137695, + "learning_rate": 3.177795655671762e-07, + "loss": 0.8885, + "step": 1318 + }, + { + "epoch": 1.9088277858176554, + "grad_norm": 1.636926293373108, + "learning_rate": 3.137570394207563e-07, + "loss": 0.8524, + "step": 1319 + }, + { + "epoch": 1.9102749638205498, + "grad_norm": 1.4566848278045654, + "learning_rate": 3.097345132743363e-07, + "loss": 0.8507, + "step": 1320 + }, + { + "epoch": 1.9102749638205498, + "eval_loss": 0.975534200668335, + "eval_runtime": 25.9357, + "eval_samples_per_second": 38.557, + "eval_steps_per_second": 2.429, + "step": 1320 + }, + { + "epoch": 1.9117221418234442, + "grad_norm": 1.643818736076355, + "learning_rate": 3.0571198712791633e-07, + "loss": 0.8868, + "step": 1321 + }, + { + "epoch": 1.9131693198263386, + "grad_norm": 1.5635424852371216, + "learning_rate": 3.016894609814964e-07, + "loss": 0.8856, + "step": 1322 + }, + { + "epoch": 1.914616497829233, + "grad_norm": 1.5588244199752808, + "learning_rate": 2.976669348350765e-07, + "loss": 0.8923, + "step": 1323 + }, + { + "epoch": 1.9160636758321274, + "grad_norm": 1.5307813882827759, + "learning_rate": 2.936444086886565e-07, + "loss": 0.8594, + "step": 1324 + }, + { + "epoch": 1.9175108538350218, + "grad_norm": 1.5048149824142456, + "learning_rate": 2.896218825422365e-07, + "loss": 0.8659, + "step": 1325 + }, + { + "epoch": 1.918958031837916, + "grad_norm": 1.5086400508880615, + "learning_rate": 2.855993563958166e-07, + "loss": 0.8608, + "step": 1326 + }, + { + "epoch": 1.9204052098408104, + "grad_norm": 1.659081220626831, + "learning_rate": 2.815768302493966e-07, + "loss": 0.9087, + "step": 1327 + }, + { + "epoch": 1.9218523878437048, + "grad_norm": 1.5549503564834595, + "learning_rate": 2.775543041029767e-07, + "loss": 0.8513, + "step": 1328 + }, + { + "epoch": 1.923299565846599, + "grad_norm": 1.531968116760254, + "learning_rate": 2.7353177795655677e-07, + "loss": 0.8641, + "step": 1329 + }, + { + "epoch": 1.9247467438494934, + "grad_norm": 1.5982201099395752, + "learning_rate": 2.695092518101368e-07, + "loss": 0.875, + "step": 1330 + }, + { + "epoch": 1.9247467438494934, + "eval_loss": 0.9751847982406616, + "eval_runtime": 25.9991, + "eval_samples_per_second": 38.463, + "eval_steps_per_second": 2.423, + "step": 1330 + }, + { + "epoch": 1.9261939218523878, + "grad_norm": 1.6116726398468018, + "learning_rate": 2.654867256637168e-07, + "loss": 0.8798, + "step": 1331 + }, + { + "epoch": 1.9276410998552822, + "grad_norm": 1.5425564050674438, + "learning_rate": 2.614641995172969e-07, + "loss": 0.8395, + "step": 1332 + }, + { + "epoch": 1.9290882778581766, + "grad_norm": 1.5614674091339111, + "learning_rate": 2.574416733708769e-07, + "loss": 0.8619, + "step": 1333 + }, + { + "epoch": 1.930535455861071, + "grad_norm": 1.579698085784912, + "learning_rate": 2.53419147224457e-07, + "loss": 0.8953, + "step": 1334 + }, + { + "epoch": 1.9319826338639654, + "grad_norm": 1.5420680046081543, + "learning_rate": 2.49396621078037e-07, + "loss": 0.8543, + "step": 1335 + }, + { + "epoch": 1.9334298118668596, + "grad_norm": 1.5527102947235107, + "learning_rate": 2.453740949316171e-07, + "loss": 0.8449, + "step": 1336 + }, + { + "epoch": 1.934876989869754, + "grad_norm": 1.6574015617370605, + "learning_rate": 2.413515687851971e-07, + "loss": 0.8719, + "step": 1337 + }, + { + "epoch": 1.9363241678726484, + "grad_norm": 1.5566723346710205, + "learning_rate": 2.3732904263877715e-07, + "loss": 0.8617, + "step": 1338 + }, + { + "epoch": 1.9377713458755426, + "grad_norm": 1.6653348207473755, + "learning_rate": 2.3330651649235723e-07, + "loss": 0.8858, + "step": 1339 + }, + { + "epoch": 1.939218523878437, + "grad_norm": 1.590993046760559, + "learning_rate": 2.2928399034593728e-07, + "loss": 0.8494, + "step": 1340 + }, + { + "epoch": 1.939218523878437, + "eval_loss": 0.9752265810966492, + "eval_runtime": 25.9274, + "eval_samples_per_second": 38.569, + "eval_steps_per_second": 2.43, + "step": 1340 + }, + { + "epoch": 1.9406657018813314, + "grad_norm": 1.5773200988769531, + "learning_rate": 2.2526146419951732e-07, + "loss": 0.8564, + "step": 1341 + }, + { + "epoch": 1.9421128798842258, + "grad_norm": 1.6304457187652588, + "learning_rate": 2.2123893805309737e-07, + "loss": 0.8367, + "step": 1342 + }, + { + "epoch": 1.9435600578871202, + "grad_norm": 1.6174850463867188, + "learning_rate": 2.172164119066774e-07, + "loss": 0.8704, + "step": 1343 + }, + { + "epoch": 1.9450072358900146, + "grad_norm": 1.6023101806640625, + "learning_rate": 2.1319388576025744e-07, + "loss": 0.8876, + "step": 1344 + }, + { + "epoch": 1.946454413892909, + "grad_norm": 1.6506774425506592, + "learning_rate": 2.0917135961383752e-07, + "loss": 0.8931, + "step": 1345 + }, + { + "epoch": 1.9479015918958031, + "grad_norm": 1.6398158073425293, + "learning_rate": 2.0514883346741757e-07, + "loss": 0.915, + "step": 1346 + }, + { + "epoch": 1.9493487698986975, + "grad_norm": 1.612850308418274, + "learning_rate": 2.0112630732099761e-07, + "loss": 0.8516, + "step": 1347 + }, + { + "epoch": 1.950795947901592, + "grad_norm": 1.5886873006820679, + "learning_rate": 1.9710378117457764e-07, + "loss": 0.8681, + "step": 1348 + }, + { + "epoch": 1.9522431259044861, + "grad_norm": 1.5209591388702393, + "learning_rate": 1.9308125502815768e-07, + "loss": 0.8493, + "step": 1349 + }, + { + "epoch": 1.9536903039073805, + "grad_norm": 1.6258975267410278, + "learning_rate": 1.8905872888173776e-07, + "loss": 0.8929, + "step": 1350 + }, + { + "epoch": 1.9536903039073805, + "eval_loss": 0.974741518497467, + "eval_runtime": 26.0872, + "eval_samples_per_second": 38.333, + "eval_steps_per_second": 2.415, + "step": 1350 + }, + { + "epoch": 1.955137481910275, + "grad_norm": 1.6310229301452637, + "learning_rate": 1.850362027353178e-07, + "loss": 0.8492, + "step": 1351 + }, + { + "epoch": 1.9565846599131693, + "grad_norm": 1.5790897607803345, + "learning_rate": 1.8101367658889785e-07, + "loss": 0.8644, + "step": 1352 + }, + { + "epoch": 1.9580318379160637, + "grad_norm": 1.6126723289489746, + "learning_rate": 1.7699115044247788e-07, + "loss": 0.8832, + "step": 1353 + }, + { + "epoch": 1.9594790159189581, + "grad_norm": 1.5507099628448486, + "learning_rate": 1.7296862429605792e-07, + "loss": 0.8901, + "step": 1354 + }, + { + "epoch": 1.9609261939218525, + "grad_norm": 1.5444624423980713, + "learning_rate": 1.6894609814963797e-07, + "loss": 0.8834, + "step": 1355 + }, + { + "epoch": 1.9623733719247467, + "grad_norm": 1.5959510803222656, + "learning_rate": 1.6492357200321805e-07, + "loss": 0.8432, + "step": 1356 + }, + { + "epoch": 1.963820549927641, + "grad_norm": 1.5451171398162842, + "learning_rate": 1.609010458567981e-07, + "loss": 0.8675, + "step": 1357 + }, + { + "epoch": 1.9652677279305355, + "grad_norm": 1.5004843473434448, + "learning_rate": 1.5687851971037814e-07, + "loss": 0.8437, + "step": 1358 + }, + { + "epoch": 1.9667149059334297, + "grad_norm": 1.4817488193511963, + "learning_rate": 1.5285599356395817e-07, + "loss": 0.8518, + "step": 1359 + }, + { + "epoch": 1.968162083936324, + "grad_norm": 1.6038738489151, + "learning_rate": 1.4883346741753824e-07, + "loss": 0.8968, + "step": 1360 + }, + { + "epoch": 1.968162083936324, + "eval_loss": 0.9743183255195618, + "eval_runtime": 26.0266, + "eval_samples_per_second": 38.422, + "eval_steps_per_second": 2.421, + "step": 1360 + }, + { + "epoch": 1.9696092619392185, + "grad_norm": 1.558771014213562, + "learning_rate": 1.4481094127111826e-07, + "loss": 0.8535, + "step": 1361 + }, + { + "epoch": 1.9710564399421129, + "grad_norm": 1.644518256187439, + "learning_rate": 1.407884151246983e-07, + "loss": 0.884, + "step": 1362 + }, + { + "epoch": 1.9725036179450073, + "grad_norm": 1.5933536291122437, + "learning_rate": 1.3676588897827839e-07, + "loss": 0.8643, + "step": 1363 + }, + { + "epoch": 1.9739507959479017, + "grad_norm": 1.6139038801193237, + "learning_rate": 1.327433628318584e-07, + "loss": 0.8814, + "step": 1364 + }, + { + "epoch": 1.975397973950796, + "grad_norm": 1.5435553789138794, + "learning_rate": 1.2872083668543846e-07, + "loss": 0.8626, + "step": 1365 + }, + { + "epoch": 1.9768451519536903, + "grad_norm": 1.5796884298324585, + "learning_rate": 1.246983105390185e-07, + "loss": 0.8887, + "step": 1366 + }, + { + "epoch": 1.9782923299565847, + "grad_norm": 1.560550332069397, + "learning_rate": 1.2067578439259855e-07, + "loss": 0.9024, + "step": 1367 + }, + { + "epoch": 1.979739507959479, + "grad_norm": 1.6081390380859375, + "learning_rate": 1.1665325824617861e-07, + "loss": 0.834, + "step": 1368 + }, + { + "epoch": 1.9811866859623732, + "grad_norm": 1.60971200466156, + "learning_rate": 1.1263073209975866e-07, + "loss": 0.8688, + "step": 1369 + }, + { + "epoch": 1.9826338639652676, + "grad_norm": 1.5743461847305298, + "learning_rate": 1.086082059533387e-07, + "loss": 0.9112, + "step": 1370 + }, + { + "epoch": 1.9826338639652676, + "eval_loss": 0.9745351672172546, + "eval_runtime": 26.141, + "eval_samples_per_second": 38.254, + "eval_steps_per_second": 2.41, + "step": 1370 + }, + { + "epoch": 1.984081041968162, + "grad_norm": 1.5900604724884033, + "learning_rate": 1.0458567980691876e-07, + "loss": 0.8531, + "step": 1371 + }, + { + "epoch": 1.9855282199710564, + "grad_norm": 1.6036739349365234, + "learning_rate": 1.0056315366049881e-07, + "loss": 0.8617, + "step": 1372 + }, + { + "epoch": 1.9869753979739508, + "grad_norm": 1.6389803886413574, + "learning_rate": 9.654062751407884e-08, + "loss": 0.8873, + "step": 1373 + }, + { + "epoch": 1.9884225759768452, + "grad_norm": 1.5734872817993164, + "learning_rate": 9.25181013676589e-08, + "loss": 0.8895, + "step": 1374 + }, + { + "epoch": 1.9898697539797396, + "grad_norm": 1.530853509902954, + "learning_rate": 8.849557522123894e-08, + "loss": 0.8509, + "step": 1375 + }, + { + "epoch": 1.9913169319826338, + "grad_norm": 1.5750211477279663, + "learning_rate": 8.447304907481899e-08, + "loss": 0.8455, + "step": 1376 + }, + { + "epoch": 1.9927641099855282, + "grad_norm": 1.5041974782943726, + "learning_rate": 8.045052292839905e-08, + "loss": 0.8677, + "step": 1377 + }, + { + "epoch": 1.9942112879884226, + "grad_norm": 1.6976056098937988, + "learning_rate": 7.642799678197908e-08, + "loss": 0.8759, + "step": 1378 + }, + { + "epoch": 1.9956584659913168, + "grad_norm": 1.535250186920166, + "learning_rate": 7.240547063555913e-08, + "loss": 0.8582, + "step": 1379 + }, + { + "epoch": 1.9971056439942112, + "grad_norm": 1.4982575178146362, + "learning_rate": 6.838294448913919e-08, + "loss": 0.8539, + "step": 1380 + }, + { + "epoch": 1.9971056439942112, + "eval_loss": 0.9745309352874756, + "eval_runtime": 26.0712, + "eval_samples_per_second": 38.356, + "eval_steps_per_second": 2.416, + "step": 1380 + }, + { + "epoch": 1.9985528219971056, + "grad_norm": 1.5955522060394287, + "learning_rate": 6.436041834271923e-08, + "loss": 0.881, + "step": 1381 + }, + { + "epoch": 2.0, + "grad_norm": 1.5880597829818726, + "learning_rate": 6.033789219629928e-08, + "loss": 0.8396, + "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": 5.654550286644019e+17, + "train_batch_size": 6, + "trial_name": null, + "trial_params": null +}