diff --git "a/checkpoints/Qwen2.5-7B/babylm_reverse_partial_10M_seed0/runs/checkpoint-1382/trainer_state.json" "b/checkpoints/Qwen2.5-7B/babylm_reverse_partial_10M_seed0/runs/checkpoint-1382/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoints/Qwen2.5-7B/babylm_reverse_partial_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": 1.5805, + "step": 1 + }, + { + "epoch": 0.002894356005788712, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6142, + "step": 2 + }, + { + "epoch": 0.004341534008683068, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6162, + "step": 3 + }, + { + "epoch": 0.005788712011577424, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6065, + "step": 4 + }, + { + "epoch": 0.00723589001447178, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6162, + "step": 5 + }, + { + "epoch": 0.008683068017366137, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6361, + "step": 6 + }, + { + "epoch": 0.010130246020260492, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.622, + "step": 7 + }, + { + "epoch": 0.011577424023154847, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.617, + "step": 8 + }, + { + "epoch": 0.013024602026049204, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6224, + "step": 9 + }, + { + "epoch": 0.01447178002894356, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6018, + "step": 10 + }, + { + "epoch": 0.01447178002894356, + "eval_loss": 1.6283396482467651, + "eval_runtime": 48.6017, + "eval_samples_per_second": 20.575, + "eval_steps_per_second": 1.296, + "step": 10 + }, + { + "epoch": 0.015918958031837915, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.641, + "step": 11 + }, + { + "epoch": 0.017366136034732273, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6368, + "step": 12 + }, + { + "epoch": 0.01881331403762663, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.627, + "step": 13 + }, + { + "epoch": 0.020260492040520984, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.6061, + "step": 14 + }, + { + "epoch": 0.02170767004341534, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.642, + "step": 15 + }, + { + "epoch": 0.023154848046309694, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 1.618, + "step": 16 + }, + { + "epoch": 0.024602026049204053, + "grad_norm": 3.9453370571136475, + "learning_rate": 3.597122302158274e-08, + "loss": 1.6164, + "step": 17 + }, + { + "epoch": 0.02604920405209841, + "grad_norm": 3.9453370571136475, + "learning_rate": 3.597122302158274e-08, + "loss": 1.6484, + "step": 18 + }, + { + "epoch": 0.027496382054992764, + "grad_norm": 4.764343738555908, + "learning_rate": 7.194244604316547e-08, + "loss": 1.6118, + "step": 19 + }, + { + "epoch": 0.02894356005788712, + "grad_norm": 4.082616806030273, + "learning_rate": 1.0791366906474822e-07, + "loss": 1.6332, + "step": 20 + }, + { + "epoch": 0.02894356005788712, + "eval_loss": 1.6283005475997925, + "eval_runtime": 48.5075, + "eval_samples_per_second": 20.615, + "eval_steps_per_second": 1.299, + "step": 20 + }, + { + "epoch": 0.030390738060781478, + "grad_norm": 4.864103317260742, + "learning_rate": 1.4388489208633095e-07, + "loss": 1.633, + "step": 21 + }, + { + "epoch": 0.03183791606367583, + "grad_norm": 4.279126167297363, + "learning_rate": 1.7985611510791368e-07, + "loss": 1.5955, + "step": 22 + }, + { + "epoch": 0.03328509406657019, + "grad_norm": 3.9418704509735107, + "learning_rate": 2.1582733812949643e-07, + "loss": 1.605, + "step": 23 + }, + { + "epoch": 0.03473227206946455, + "grad_norm": 3.4064688682556152, + "learning_rate": 2.5179856115107916e-07, + "loss": 1.6444, + "step": 24 + }, + { + "epoch": 0.0361794500723589, + "grad_norm": 4.212948322296143, + "learning_rate": 2.877697841726619e-07, + "loss": 1.6298, + "step": 25 + }, + { + "epoch": 0.03762662807525326, + "grad_norm": 4.028613090515137, + "learning_rate": 3.237410071942446e-07, + "loss": 1.5825, + "step": 26 + }, + { + "epoch": 0.03907380607814761, + "grad_norm": 3.4630320072174072, + "learning_rate": 3.5971223021582736e-07, + "loss": 1.6059, + "step": 27 + }, + { + "epoch": 0.04052098408104197, + "grad_norm": 3.459155559539795, + "learning_rate": 3.956834532374101e-07, + "loss": 1.6385, + "step": 28 + }, + { + "epoch": 0.041968162083936326, + "grad_norm": 2.9293971061706543, + "learning_rate": 4.3165467625899287e-07, + "loss": 1.5803, + "step": 29 + }, + { + "epoch": 0.04341534008683068, + "grad_norm": 2.7857871055603027, + "learning_rate": 4.676258992805756e-07, + "loss": 1.5919, + "step": 30 + }, + { + "epoch": 0.04341534008683068, + "eval_loss": 1.6075999736785889, + "eval_runtime": 49.5849, + "eval_samples_per_second": 20.167, + "eval_steps_per_second": 1.271, + "step": 30 + }, + { + "epoch": 0.04486251808972504, + "grad_norm": 2.3637592792510986, + "learning_rate": 5.035971223021583e-07, + "loss": 1.5792, + "step": 31 + }, + { + "epoch": 0.04630969609261939, + "grad_norm": 2.50028920173645, + "learning_rate": 5.39568345323741e-07, + "loss": 1.6048, + "step": 32 + }, + { + "epoch": 0.04775687409551375, + "grad_norm": 2.166193723678589, + "learning_rate": 5.755395683453238e-07, + "loss": 1.6001, + "step": 33 + }, + { + "epoch": 0.049204052098408106, + "grad_norm": 1.9250812530517578, + "learning_rate": 6.115107913669066e-07, + "loss": 1.5832, + "step": 34 + }, + { + "epoch": 0.05065123010130246, + "grad_norm": 2.997576951980591, + "learning_rate": 6.474820143884893e-07, + "loss": 1.5618, + "step": 35 + }, + { + "epoch": 0.05209840810419682, + "grad_norm": 2.308159589767456, + "learning_rate": 6.83453237410072e-07, + "loss": 1.5537, + "step": 36 + }, + { + "epoch": 0.053545586107091175, + "grad_norm": 2.106954574584961, + "learning_rate": 7.194244604316547e-07, + "loss": 1.5833, + "step": 37 + }, + { + "epoch": 0.05499276410998553, + "grad_norm": 1.966857671737671, + "learning_rate": 7.553956834532375e-07, + "loss": 1.5522, + "step": 38 + }, + { + "epoch": 0.056439942112879886, + "grad_norm": 2.3216395378112793, + "learning_rate": 7.913669064748202e-07, + "loss": 1.5628, + "step": 39 + }, + { + "epoch": 0.05788712011577424, + "grad_norm": 1.6036194562911987, + "learning_rate": 8.27338129496403e-07, + "loss": 1.551, + "step": 40 + }, + { + "epoch": 0.05788712011577424, + "eval_loss": 1.5650317668914795, + "eval_runtime": 50.4427, + "eval_samples_per_second": 19.824, + "eval_steps_per_second": 1.249, + "step": 40 + }, + { + "epoch": 0.059334298118668596, + "grad_norm": 1.8111491203308105, + "learning_rate": 8.633093525179857e-07, + "loss": 1.5414, + "step": 41 + }, + { + "epoch": 0.060781476121562955, + "grad_norm": 1.8309638500213623, + "learning_rate": 8.992805755395684e-07, + "loss": 1.5589, + "step": 42 + }, + { + "epoch": 0.06222865412445731, + "grad_norm": 1.5761781930923462, + "learning_rate": 9.352517985611512e-07, + "loss": 1.576, + "step": 43 + }, + { + "epoch": 0.06367583212735166, + "grad_norm": 1.6372134685516357, + "learning_rate": 9.71223021582734e-07, + "loss": 1.5636, + "step": 44 + }, + { + "epoch": 0.06512301013024602, + "grad_norm": 1.9859188795089722, + "learning_rate": 1.0071942446043167e-06, + "loss": 1.5516, + "step": 45 + }, + { + "epoch": 0.06657018813314038, + "grad_norm": 1.9442024230957031, + "learning_rate": 1.0431654676258993e-06, + "loss": 1.5307, + "step": 46 + }, + { + "epoch": 0.06801736613603473, + "grad_norm": 2.0288305282592773, + "learning_rate": 1.079136690647482e-06, + "loss": 1.5497, + "step": 47 + }, + { + "epoch": 0.0694645441389291, + "grad_norm": 1.7322059869766235, + "learning_rate": 1.115107913669065e-06, + "loss": 1.5337, + "step": 48 + }, + { + "epoch": 0.07091172214182344, + "grad_norm": 1.7373536825180054, + "learning_rate": 1.1510791366906476e-06, + "loss": 1.5218, + "step": 49 + }, + { + "epoch": 0.0723589001447178, + "grad_norm": 1.796687364578247, + "learning_rate": 1.1870503597122303e-06, + "loss": 1.4902, + "step": 50 + }, + { + "epoch": 0.0723589001447178, + "eval_loss": 1.5142266750335693, + "eval_runtime": 51.1573, + "eval_samples_per_second": 19.548, + "eval_steps_per_second": 1.231, + "step": 50 + }, + { + "epoch": 0.07380607814761216, + "grad_norm": 1.504360318183899, + "learning_rate": 1.2230215827338131e-06, + "loss": 1.4747, + "step": 51 + }, + { + "epoch": 0.07525325615050651, + "grad_norm": 1.9087471961975098, + "learning_rate": 1.2589928057553958e-06, + "loss": 1.4906, + "step": 52 + }, + { + "epoch": 0.07670043415340087, + "grad_norm": 1.5625972747802734, + "learning_rate": 1.2949640287769785e-06, + "loss": 1.4646, + "step": 53 + }, + { + "epoch": 0.07814761215629522, + "grad_norm": 1.8286936283111572, + "learning_rate": 1.3309352517985614e-06, + "loss": 1.4606, + "step": 54 + }, + { + "epoch": 0.07959479015918958, + "grad_norm": 1.6221673488616943, + "learning_rate": 1.366906474820144e-06, + "loss": 1.471, + "step": 55 + }, + { + "epoch": 0.08104196816208394, + "grad_norm": 1.8522502183914185, + "learning_rate": 1.4028776978417265e-06, + "loss": 1.4263, + "step": 56 + }, + { + "epoch": 0.0824891461649783, + "grad_norm": 2.6883490085601807, + "learning_rate": 1.4388489208633094e-06, + "loss": 1.369, + "step": 57 + }, + { + "epoch": 0.08393632416787265, + "grad_norm": 1.628127932548523, + "learning_rate": 1.474820143884892e-06, + "loss": 1.403, + "step": 58 + }, + { + "epoch": 0.085383502170767, + "grad_norm": 2.652324676513672, + "learning_rate": 1.510791366906475e-06, + "loss": 1.4187, + "step": 59 + }, + { + "epoch": 0.08683068017366136, + "grad_norm": 2.0814850330352783, + "learning_rate": 1.5467625899280579e-06, + "loss": 1.3632, + "step": 60 + }, + { + "epoch": 0.08683068017366136, + "eval_loss": 1.407728910446167, + "eval_runtime": 51.2932, + "eval_samples_per_second": 19.496, + "eval_steps_per_second": 1.228, + "step": 60 + }, + { + "epoch": 0.08827785817655572, + "grad_norm": 2.366215467453003, + "learning_rate": 1.5827338129496403e-06, + "loss": 1.3938, + "step": 61 + }, + { + "epoch": 0.08972503617945007, + "grad_norm": 1.8177162408828735, + "learning_rate": 1.618705035971223e-06, + "loss": 1.3909, + "step": 62 + }, + { + "epoch": 0.09117221418234443, + "grad_norm": 1.7091832160949707, + "learning_rate": 1.654676258992806e-06, + "loss": 1.3621, + "step": 63 + }, + { + "epoch": 0.09261939218523878, + "grad_norm": 1.4615815877914429, + "learning_rate": 1.6906474820143886e-06, + "loss": 1.3418, + "step": 64 + }, + { + "epoch": 0.09406657018813314, + "grad_norm": 1.8411164283752441, + "learning_rate": 1.7266187050359715e-06, + "loss": 1.3343, + "step": 65 + }, + { + "epoch": 0.0955137481910275, + "grad_norm": 1.4318944215774536, + "learning_rate": 1.762589928057554e-06, + "loss": 1.3342, + "step": 66 + }, + { + "epoch": 0.09696092619392185, + "grad_norm": 2.1167235374450684, + "learning_rate": 1.7985611510791368e-06, + "loss": 1.3081, + "step": 67 + }, + { + "epoch": 0.09840810419681621, + "grad_norm": 1.8746788501739502, + "learning_rate": 1.8345323741007195e-06, + "loss": 1.3014, + "step": 68 + }, + { + "epoch": 0.09985528219971057, + "grad_norm": 1.3041801452636719, + "learning_rate": 1.8705035971223024e-06, + "loss": 1.3235, + "step": 69 + }, + { + "epoch": 0.10130246020260492, + "grad_norm": 1.509260892868042, + "learning_rate": 1.906474820143885e-06, + "loss": 1.3085, + "step": 70 + }, + { + "epoch": 0.10130246020260492, + "eval_loss": 1.3333908319473267, + "eval_runtime": 51.0598, + "eval_samples_per_second": 19.585, + "eval_steps_per_second": 1.234, + "step": 70 + }, + { + "epoch": 0.10274963820549927, + "grad_norm": 1.7147074937820435, + "learning_rate": 1.942446043165468e-06, + "loss": 1.3259, + "step": 71 + }, + { + "epoch": 0.10419681620839363, + "grad_norm": 2.9030086994171143, + "learning_rate": 1.9784172661870504e-06, + "loss": 1.2676, + "step": 72 + }, + { + "epoch": 0.10564399421128799, + "grad_norm": 1.926299810409546, + "learning_rate": 2.0143884892086333e-06, + "loss": 1.2709, + "step": 73 + }, + { + "epoch": 0.10709117221418235, + "grad_norm": 1.627013921737671, + "learning_rate": 2.050359712230216e-06, + "loss": 1.2565, + "step": 74 + }, + { + "epoch": 0.1085383502170767, + "grad_norm": 2.345850706100464, + "learning_rate": 2.0863309352517987e-06, + "loss": 1.2308, + "step": 75 + }, + { + "epoch": 0.10998552821997105, + "grad_norm": 1.3716319799423218, + "learning_rate": 2.1223021582733816e-06, + "loss": 1.2738, + "step": 76 + }, + { + "epoch": 0.11143270622286541, + "grad_norm": 1.8737510442733765, + "learning_rate": 2.158273381294964e-06, + "loss": 1.2741, + "step": 77 + }, + { + "epoch": 0.11287988422575977, + "grad_norm": 1.6224879026412964, + "learning_rate": 2.194244604316547e-06, + "loss": 1.2814, + "step": 78 + }, + { + "epoch": 0.11432706222865413, + "grad_norm": 1.877746820449829, + "learning_rate": 2.23021582733813e-06, + "loss": 1.2841, + "step": 79 + }, + { + "epoch": 0.11577424023154848, + "grad_norm": 1.8192110061645508, + "learning_rate": 2.2661870503597123e-06, + "loss": 1.2573, + "step": 80 + }, + { + "epoch": 0.11577424023154848, + "eval_loss": 1.2883232831954956, + "eval_runtime": 51.5109, + "eval_samples_per_second": 19.413, + "eval_steps_per_second": 1.223, + "step": 80 + }, + { + "epoch": 0.11722141823444283, + "grad_norm": 2.4764959812164307, + "learning_rate": 2.302158273381295e-06, + "loss": 1.2153, + "step": 81 + }, + { + "epoch": 0.11866859623733719, + "grad_norm": 1.4122470617294312, + "learning_rate": 2.3381294964028776e-06, + "loss": 1.229, + "step": 82 + }, + { + "epoch": 0.12011577424023155, + "grad_norm": 2.1700825691223145, + "learning_rate": 2.3741007194244605e-06, + "loss": 1.2399, + "step": 83 + }, + { + "epoch": 0.12156295224312591, + "grad_norm": 1.6023834943771362, + "learning_rate": 2.4100719424460434e-06, + "loss": 1.2302, + "step": 84 + }, + { + "epoch": 0.12301013024602026, + "grad_norm": 2.208343029022217, + "learning_rate": 2.4460431654676263e-06, + "loss": 1.1955, + "step": 85 + }, + { + "epoch": 0.12445730824891461, + "grad_norm": 1.5742015838623047, + "learning_rate": 2.4820143884892088e-06, + "loss": 1.2257, + "step": 86 + }, + { + "epoch": 0.12590448625180897, + "grad_norm": 1.672551155090332, + "learning_rate": 2.5179856115107916e-06, + "loss": 1.1778, + "step": 87 + }, + { + "epoch": 0.12735166425470332, + "grad_norm": 2.18996262550354, + "learning_rate": 2.5539568345323745e-06, + "loss": 1.2073, + "step": 88 + }, + { + "epoch": 0.1287988422575977, + "grad_norm": 1.8384442329406738, + "learning_rate": 2.589928057553957e-06, + "loss": 1.2317, + "step": 89 + }, + { + "epoch": 0.13024602026049203, + "grad_norm": 1.5735281705856323, + "learning_rate": 2.6258992805755395e-06, + "loss": 1.214, + "step": 90 + }, + { + "epoch": 0.13024602026049203, + "eval_loss": 1.2567816972732544, + "eval_runtime": 51.3172, + "eval_samples_per_second": 19.487, + "eval_steps_per_second": 1.228, + "step": 90 + }, + { + "epoch": 0.1316931982633864, + "grad_norm": 1.9919185638427734, + "learning_rate": 2.6618705035971228e-06, + "loss": 1.2259, + "step": 91 + }, + { + "epoch": 0.13314037626628075, + "grad_norm": 1.7588752508163452, + "learning_rate": 2.6978417266187052e-06, + "loss": 1.2358, + "step": 92 + }, + { + "epoch": 0.1345875542691751, + "grad_norm": 1.5497698783874512, + "learning_rate": 2.733812949640288e-06, + "loss": 1.1824, + "step": 93 + }, + { + "epoch": 0.13603473227206947, + "grad_norm": 1.9599865674972534, + "learning_rate": 2.7697841726618706e-06, + "loss": 1.2017, + "step": 94 + }, + { + "epoch": 0.13748191027496381, + "grad_norm": 1.3761709928512573, + "learning_rate": 2.805755395683453e-06, + "loss": 1.1997, + "step": 95 + }, + { + "epoch": 0.1389290882778582, + "grad_norm": 2.3903093338012695, + "learning_rate": 2.8417266187050364e-06, + "loss": 1.1731, + "step": 96 + }, + { + "epoch": 0.14037626628075253, + "grad_norm": 1.6955513954162598, + "learning_rate": 2.877697841726619e-06, + "loss": 1.1907, + "step": 97 + }, + { + "epoch": 0.14182344428364688, + "grad_norm": 1.8511075973510742, + "learning_rate": 2.9136690647482017e-06, + "loss": 1.1643, + "step": 98 + }, + { + "epoch": 0.14327062228654125, + "grad_norm": 1.6991047859191895, + "learning_rate": 2.949640287769784e-06, + "loss": 1.2103, + "step": 99 + }, + { + "epoch": 0.1447178002894356, + "grad_norm": 1.4709733724594116, + "learning_rate": 2.985611510791367e-06, + "loss": 1.209, + "step": 100 + }, + { + "epoch": 0.1447178002894356, + "eval_loss": 1.235381007194519, + "eval_runtime": 51.4015, + "eval_samples_per_second": 19.455, + "eval_steps_per_second": 1.226, + "step": 100 + }, + { + "epoch": 0.14616497829232997, + "grad_norm": 1.741533875465393, + "learning_rate": 3.02158273381295e-06, + "loss": 1.1591, + "step": 101 + }, + { + "epoch": 0.1476121562952243, + "grad_norm": 2.225661039352417, + "learning_rate": 3.0575539568345324e-06, + "loss": 1.1765, + "step": 102 + }, + { + "epoch": 0.14905933429811866, + "grad_norm": 1.5822570323944092, + "learning_rate": 3.0935251798561158e-06, + "loss": 1.1395, + "step": 103 + }, + { + "epoch": 0.15050651230101303, + "grad_norm": 1.9070250988006592, + "learning_rate": 3.1294964028776982e-06, + "loss": 1.1654, + "step": 104 + }, + { + "epoch": 0.15195369030390737, + "grad_norm": 1.9010212421417236, + "learning_rate": 3.1654676258992807e-06, + "loss": 1.1578, + "step": 105 + }, + { + "epoch": 0.15340086830680175, + "grad_norm": 1.9947483539581299, + "learning_rate": 3.2014388489208636e-06, + "loss": 1.1445, + "step": 106 + }, + { + "epoch": 0.1548480463096961, + "grad_norm": 1.6002594232559204, + "learning_rate": 3.237410071942446e-06, + "loss": 1.1612, + "step": 107 + }, + { + "epoch": 0.15629522431259044, + "grad_norm": 2.283268451690674, + "learning_rate": 3.2733812949640294e-06, + "loss": 1.152, + "step": 108 + }, + { + "epoch": 0.1577424023154848, + "grad_norm": 1.9182709455490112, + "learning_rate": 3.309352517985612e-06, + "loss": 1.1602, + "step": 109 + }, + { + "epoch": 0.15918958031837915, + "grad_norm": 1.9607409238815308, + "learning_rate": 3.3453237410071943e-06, + "loss": 1.1143, + "step": 110 + }, + { + "epoch": 0.15918958031837915, + "eval_loss": 1.2227240800857544, + "eval_runtime": 51.0273, + "eval_samples_per_second": 19.597, + "eval_steps_per_second": 1.235, + "step": 110 + }, + { + "epoch": 0.16063675832127353, + "grad_norm": 2.4197559356689453, + "learning_rate": 3.381294964028777e-06, + "loss": 1.1847, + "step": 111 + }, + { + "epoch": 0.16208393632416787, + "grad_norm": 1.7451276779174805, + "learning_rate": 3.4172661870503596e-06, + "loss": 1.1795, + "step": 112 + }, + { + "epoch": 0.16353111432706222, + "grad_norm": 3.4664721488952637, + "learning_rate": 3.453237410071943e-06, + "loss": 1.1557, + "step": 113 + }, + { + "epoch": 0.1649782923299566, + "grad_norm": 2.4312498569488525, + "learning_rate": 3.4892086330935254e-06, + "loss": 1.1506, + "step": 114 + }, + { + "epoch": 0.16642547033285093, + "grad_norm": 1.7869541645050049, + "learning_rate": 3.525179856115108e-06, + "loss": 1.1385, + "step": 115 + }, + { + "epoch": 0.1678726483357453, + "grad_norm": 1.6890935897827148, + "learning_rate": 3.561151079136691e-06, + "loss": 1.1133, + "step": 116 + }, + { + "epoch": 0.16931982633863965, + "grad_norm": 2.3489439487457275, + "learning_rate": 3.5971223021582737e-06, + "loss": 1.1582, + "step": 117 + }, + { + "epoch": 0.170767004341534, + "grad_norm": 1.8442103862762451, + "learning_rate": 3.6330935251798566e-06, + "loss": 1.1233, + "step": 118 + }, + { + "epoch": 0.17221418234442837, + "grad_norm": 1.9009796380996704, + "learning_rate": 3.669064748201439e-06, + "loss": 1.1465, + "step": 119 + }, + { + "epoch": 0.1736613603473227, + "grad_norm": 1.5274282693862915, + "learning_rate": 3.7050359712230215e-06, + "loss": 1.1335, + "step": 120 + }, + { + "epoch": 0.1736613603473227, + "eval_loss": 1.2064287662506104, + "eval_runtime": 51.1154, + "eval_samples_per_second": 19.564, + "eval_steps_per_second": 1.233, + "step": 120 + }, + { + "epoch": 0.17510853835021709, + "grad_norm": 1.2717360258102417, + "learning_rate": 3.741007194244605e-06, + "loss": 1.1293, + "step": 121 + }, + { + "epoch": 0.17655571635311143, + "grad_norm": 3.3121867179870605, + "learning_rate": 3.7769784172661873e-06, + "loss": 1.1304, + "step": 122 + }, + { + "epoch": 0.17800289435600578, + "grad_norm": 1.66675865650177, + "learning_rate": 3.81294964028777e-06, + "loss": 1.1022, + "step": 123 + }, + { + "epoch": 0.17945007235890015, + "grad_norm": 1.6999047994613647, + "learning_rate": 3.848920863309353e-06, + "loss": 1.1224, + "step": 124 + }, + { + "epoch": 0.1808972503617945, + "grad_norm": 2.650796890258789, + "learning_rate": 3.884892086330936e-06, + "loss": 1.1482, + "step": 125 + }, + { + "epoch": 0.18234442836468887, + "grad_norm": 1.5036591291427612, + "learning_rate": 3.920863309352518e-06, + "loss": 1.1434, + "step": 126 + }, + { + "epoch": 0.1837916063675832, + "grad_norm": 2.3914601802825928, + "learning_rate": 3.956834532374101e-06, + "loss": 1.1483, + "step": 127 + }, + { + "epoch": 0.18523878437047755, + "grad_norm": 2.5098259449005127, + "learning_rate": 3.992805755395684e-06, + "loss": 1.1243, + "step": 128 + }, + { + "epoch": 0.18668596237337193, + "grad_norm": 1.5939863920211792, + "learning_rate": 4.028776978417267e-06, + "loss": 1.1272, + "step": 129 + }, + { + "epoch": 0.18813314037626627, + "grad_norm": 2.2251949310302734, + "learning_rate": 4.0647482014388495e-06, + "loss": 1.1197, + "step": 130 + }, + { + "epoch": 0.18813314037626627, + "eval_loss": 1.1897306442260742, + "eval_runtime": 51.4505, + "eval_samples_per_second": 19.436, + "eval_steps_per_second": 1.224, + "step": 130 + }, + { + "epoch": 0.18958031837916064, + "grad_norm": 1.7876744270324707, + "learning_rate": 4.100719424460432e-06, + "loss": 1.1509, + "step": 131 + }, + { + "epoch": 0.191027496382055, + "grad_norm": 1.83745539188385, + "learning_rate": 4.1366906474820145e-06, + "loss": 1.1106, + "step": 132 + }, + { + "epoch": 0.19247467438494936, + "grad_norm": 1.8682845830917358, + "learning_rate": 4.172661870503597e-06, + "loss": 1.1607, + "step": 133 + }, + { + "epoch": 0.1939218523878437, + "grad_norm": 2.0207204818725586, + "learning_rate": 4.20863309352518e-06, + "loss": 1.1461, + "step": 134 + }, + { + "epoch": 0.19536903039073805, + "grad_norm": 1.4707207679748535, + "learning_rate": 4.244604316546763e-06, + "loss": 1.127, + "step": 135 + }, + { + "epoch": 0.19681620839363242, + "grad_norm": 2.134096622467041, + "learning_rate": 4.280575539568346e-06, + "loss": 1.1134, + "step": 136 + }, + { + "epoch": 0.19826338639652677, + "grad_norm": 2.499931812286377, + "learning_rate": 4.316546762589928e-06, + "loss": 1.1001, + "step": 137 + }, + { + "epoch": 0.19971056439942114, + "grad_norm": 1.3337165117263794, + "learning_rate": 4.352517985611511e-06, + "loss": 1.0808, + "step": 138 + }, + { + "epoch": 0.2011577424023155, + "grad_norm": 2.5206005573272705, + "learning_rate": 4.388489208633094e-06, + "loss": 1.1233, + "step": 139 + }, + { + "epoch": 0.20260492040520983, + "grad_norm": 1.4684406518936157, + "learning_rate": 4.424460431654677e-06, + "loss": 1.1139, + "step": 140 + }, + { + "epoch": 0.20260492040520983, + "eval_loss": 1.1761609315872192, + "eval_runtime": 51.3668, + "eval_samples_per_second": 19.468, + "eval_steps_per_second": 1.226, + "step": 140 + }, + { + "epoch": 0.2040520984081042, + "grad_norm": 2.175002098083496, + "learning_rate": 4.46043165467626e-06, + "loss": 1.0886, + "step": 141 + }, + { + "epoch": 0.20549927641099855, + "grad_norm": 1.9261544942855835, + "learning_rate": 4.496402877697842e-06, + "loss": 1.1043, + "step": 142 + }, + { + "epoch": 0.20694645441389292, + "grad_norm": 2.775707244873047, + "learning_rate": 4.5323741007194245e-06, + "loss": 1.1071, + "step": 143 + }, + { + "epoch": 0.20839363241678727, + "grad_norm": 1.6811009645462036, + "learning_rate": 4.5683453237410074e-06, + "loss": 1.0915, + "step": 144 + }, + { + "epoch": 0.2098408104196816, + "grad_norm": 2.1066107749938965, + "learning_rate": 4.60431654676259e-06, + "loss": 1.1171, + "step": 145 + }, + { + "epoch": 0.21128798842257598, + "grad_norm": 2.0910868644714355, + "learning_rate": 4.640287769784173e-06, + "loss": 1.1137, + "step": 146 + }, + { + "epoch": 0.21273516642547033, + "grad_norm": 1.6458780765533447, + "learning_rate": 4.676258992805755e-06, + "loss": 1.0581, + "step": 147 + }, + { + "epoch": 0.2141823444283647, + "grad_norm": 1.6126712560653687, + "learning_rate": 4.712230215827339e-06, + "loss": 1.1105, + "step": 148 + }, + { + "epoch": 0.21562952243125905, + "grad_norm": 1.593931794166565, + "learning_rate": 4.748201438848921e-06, + "loss": 1.0942, + "step": 149 + }, + { + "epoch": 0.2170767004341534, + "grad_norm": 1.7260593175888062, + "learning_rate": 4.784172661870504e-06, + "loss": 1.1124, + "step": 150 + }, + { + "epoch": 0.2170767004341534, + "eval_loss": 1.1670387983322144, + "eval_runtime": 51.0823, + "eval_samples_per_second": 19.576, + "eval_steps_per_second": 1.233, + "step": 150 + }, + { + "epoch": 0.21852387843704776, + "grad_norm": 1.7146104574203491, + "learning_rate": 4.820143884892087e-06, + "loss": 1.1044, + "step": 151 + }, + { + "epoch": 0.2199710564399421, + "grad_norm": 2.0193190574645996, + "learning_rate": 4.856115107913669e-06, + "loss": 1.0534, + "step": 152 + }, + { + "epoch": 0.22141823444283648, + "grad_norm": 1.5494120121002197, + "learning_rate": 4.892086330935253e-06, + "loss": 1.0743, + "step": 153 + }, + { + "epoch": 0.22286541244573083, + "grad_norm": 2.0492382049560547, + "learning_rate": 4.928057553956835e-06, + "loss": 1.087, + "step": 154 + }, + { + "epoch": 0.22431259044862517, + "grad_norm": 2.057176113128662, + "learning_rate": 4.9640287769784175e-06, + "loss": 1.0722, + "step": 155 + }, + { + "epoch": 0.22575976845151954, + "grad_norm": 2.225590705871582, + "learning_rate": 5e-06, + "loss": 1.1013, + "step": 156 + }, + { + "epoch": 0.2272069464544139, + "grad_norm": 2.0944740772247314, + "learning_rate": 4.9959774738535805e-06, + "loss": 1.1061, + "step": 157 + }, + { + "epoch": 0.22865412445730826, + "grad_norm": 1.6131840944290161, + "learning_rate": 4.991954947707161e-06, + "loss": 1.05, + "step": 158 + }, + { + "epoch": 0.2301013024602026, + "grad_norm": 1.7985981702804565, + "learning_rate": 4.987932421560741e-06, + "loss": 1.0747, + "step": 159 + }, + { + "epoch": 0.23154848046309695, + "grad_norm": 1.636303424835205, + "learning_rate": 4.983909895414321e-06, + "loss": 1.1063, + "step": 160 + }, + { + "epoch": 0.23154848046309695, + "eval_loss": 1.1593362092971802, + "eval_runtime": 51.2453, + "eval_samples_per_second": 19.514, + "eval_steps_per_second": 1.229, + "step": 160 + }, + { + "epoch": 0.23299565846599132, + "grad_norm": 1.2702714204788208, + "learning_rate": 4.9798873692679e-06, + "loss": 1.0806, + "step": 161 + }, + { + "epoch": 0.23444283646888567, + "grad_norm": 2.020653247833252, + "learning_rate": 4.97586484312148e-06, + "loss": 1.0527, + "step": 162 + }, + { + "epoch": 0.23589001447178004, + "grad_norm": 1.675683617591858, + "learning_rate": 4.9718423169750605e-06, + "loss": 1.0715, + "step": 163 + }, + { + "epoch": 0.23733719247467439, + "grad_norm": 1.4638187885284424, + "learning_rate": 4.967819790828641e-06, + "loss": 1.0826, + "step": 164 + }, + { + "epoch": 0.23878437047756873, + "grad_norm": 1.9109163284301758, + "learning_rate": 4.963797264682221e-06, + "loss": 1.0349, + "step": 165 + }, + { + "epoch": 0.2402315484804631, + "grad_norm": 2.1733484268188477, + "learning_rate": 4.959774738535801e-06, + "loss": 1.0854, + "step": 166 + }, + { + "epoch": 0.24167872648335745, + "grad_norm": 2.0975842475891113, + "learning_rate": 4.955752212389381e-06, + "loss": 1.0573, + "step": 167 + }, + { + "epoch": 0.24312590448625182, + "grad_norm": 2.32848858833313, + "learning_rate": 4.951729686242961e-06, + "loss": 1.084, + "step": 168 + }, + { + "epoch": 0.24457308248914617, + "grad_norm": 1.9513778686523438, + "learning_rate": 4.947707160096541e-06, + "loss": 1.049, + "step": 169 + }, + { + "epoch": 0.2460202604920405, + "grad_norm": 2.0373921394348145, + "learning_rate": 4.943684633950121e-06, + "loss": 1.0898, + "step": 170 + }, + { + "epoch": 0.2460202604920405, + "eval_loss": 1.1524850130081177, + "eval_runtime": 51.2672, + "eval_samples_per_second": 19.506, + "eval_steps_per_second": 1.229, + "step": 170 + }, + { + "epoch": 0.24746743849493488, + "grad_norm": 2.1742377281188965, + "learning_rate": 4.939662107803701e-06, + "loss": 1.0951, + "step": 171 + }, + { + "epoch": 0.24891461649782923, + "grad_norm": 1.1019455194473267, + "learning_rate": 4.935639581657281e-06, + "loss": 1.0627, + "step": 172 + }, + { + "epoch": 0.2503617945007236, + "grad_norm": 2.151254892349243, + "learning_rate": 4.931617055510861e-06, + "loss": 1.0386, + "step": 173 + }, + { + "epoch": 0.25180897250361794, + "grad_norm": 1.6461738348007202, + "learning_rate": 4.927594529364441e-06, + "loss": 1.0585, + "step": 174 + }, + { + "epoch": 0.2532561505065123, + "grad_norm": 1.8385056257247925, + "learning_rate": 4.923572003218021e-06, + "loss": 1.0705, + "step": 175 + }, + { + "epoch": 0.25470332850940663, + "grad_norm": 1.6924220323562622, + "learning_rate": 4.919549477071601e-06, + "loss": 1.0483, + "step": 176 + }, + { + "epoch": 0.25615050651230103, + "grad_norm": 1.8686519861221313, + "learning_rate": 4.915526950925181e-06, + "loss": 1.1036, + "step": 177 + }, + { + "epoch": 0.2575976845151954, + "grad_norm": 1.2192116975784302, + "learning_rate": 4.9115044247787615e-06, + "loss": 1.0571, + "step": 178 + }, + { + "epoch": 0.2590448625180897, + "grad_norm": 1.6416335105895996, + "learning_rate": 4.907481898632342e-06, + "loss": 1.0595, + "step": 179 + }, + { + "epoch": 0.26049204052098407, + "grad_norm": 2.329639434814453, + "learning_rate": 4.903459372485922e-06, + "loss": 1.0788, + "step": 180 + }, + { + "epoch": 0.26049204052098407, + "eval_loss": 1.144730806350708, + "eval_runtime": 51.1483, + "eval_samples_per_second": 19.551, + "eval_steps_per_second": 1.232, + "step": 180 + }, + { + "epoch": 0.2619392185238784, + "grad_norm": 1.8451154232025146, + "learning_rate": 4.899436846339501e-06, + "loss": 1.0812, + "step": 181 + }, + { + "epoch": 0.2633863965267728, + "grad_norm": 1.6093605756759644, + "learning_rate": 4.895414320193081e-06, + "loss": 1.0629, + "step": 182 + }, + { + "epoch": 0.26483357452966716, + "grad_norm": 1.5421539545059204, + "learning_rate": 4.891391794046661e-06, + "loss": 1.067, + "step": 183 + }, + { + "epoch": 0.2662807525325615, + "grad_norm": 1.8255059719085693, + "learning_rate": 4.8873692679002414e-06, + "loss": 1.0453, + "step": 184 + }, + { + "epoch": 0.26772793053545585, + "grad_norm": 2.0148684978485107, + "learning_rate": 4.8833467417538216e-06, + "loss": 1.0714, + "step": 185 + }, + { + "epoch": 0.2691751085383502, + "grad_norm": 2.2409253120422363, + "learning_rate": 4.879324215607402e-06, + "loss": 1.0538, + "step": 186 + }, + { + "epoch": 0.2706222865412446, + "grad_norm": 1.7410125732421875, + "learning_rate": 4.875301689460982e-06, + "loss": 1.0389, + "step": 187 + }, + { + "epoch": 0.27206946454413894, + "grad_norm": 2.374063491821289, + "learning_rate": 4.871279163314562e-06, + "loss": 1.0511, + "step": 188 + }, + { + "epoch": 0.2735166425470333, + "grad_norm": 1.487918496131897, + "learning_rate": 4.867256637168142e-06, + "loss": 1.0551, + "step": 189 + }, + { + "epoch": 0.27496382054992763, + "grad_norm": 1.6984626054763794, + "learning_rate": 4.863234111021722e-06, + "loss": 1.0345, + "step": 190 + }, + { + "epoch": 0.27496382054992763, + "eval_loss": 1.139499545097351, + "eval_runtime": 51.4079, + "eval_samples_per_second": 19.452, + "eval_steps_per_second": 1.225, + "step": 190 + }, + { + "epoch": 0.276410998552822, + "grad_norm": 1.4627532958984375, + "learning_rate": 4.8592115848753015e-06, + "loss": 1.0797, + "step": 191 + }, + { + "epoch": 0.2778581765557164, + "grad_norm": 1.979169487953186, + "learning_rate": 4.855189058728882e-06, + "loss": 1.0075, + "step": 192 + }, + { + "epoch": 0.2793053545586107, + "grad_norm": 1.6630204916000366, + "learning_rate": 4.851166532582462e-06, + "loss": 1.0213, + "step": 193 + }, + { + "epoch": 0.28075253256150506, + "grad_norm": 2.0252039432525635, + "learning_rate": 4.847144006436042e-06, + "loss": 1.0398, + "step": 194 + }, + { + "epoch": 0.2821997105643994, + "grad_norm": 1.756808876991272, + "learning_rate": 4.843121480289622e-06, + "loss": 1.0852, + "step": 195 + }, + { + "epoch": 0.28364688856729375, + "grad_norm": 2.1497552394866943, + "learning_rate": 4.839098954143202e-06, + "loss": 1.0508, + "step": 196 + }, + { + "epoch": 0.28509406657018815, + "grad_norm": 1.7734496593475342, + "learning_rate": 4.835076427996782e-06, + "loss": 1.0364, + "step": 197 + }, + { + "epoch": 0.2865412445730825, + "grad_norm": 1.9345977306365967, + "learning_rate": 4.831053901850362e-06, + "loss": 1.0328, + "step": 198 + }, + { + "epoch": 0.28798842257597684, + "grad_norm": 1.5408843755722046, + "learning_rate": 4.8270313757039425e-06, + "loss": 1.0384, + "step": 199 + }, + { + "epoch": 0.2894356005788712, + "grad_norm": 1.9991062879562378, + "learning_rate": 4.823008849557523e-06, + "loss": 1.0282, + "step": 200 + }, + { + "epoch": 0.2894356005788712, + "eval_loss": 1.1295605897903442, + "eval_runtime": 51.0198, + "eval_samples_per_second": 19.6, + "eval_steps_per_second": 1.235, + "step": 200 + }, + { + "epoch": 0.29088277858176553, + "grad_norm": 2.3587474822998047, + "learning_rate": 4.818986323411103e-06, + "loss": 1.0307, + "step": 201 + }, + { + "epoch": 0.29232995658465993, + "grad_norm": 1.4203909635543823, + "learning_rate": 4.814963797264683e-06, + "loss": 1.0344, + "step": 202 + }, + { + "epoch": 0.2937771345875543, + "grad_norm": 1.9014217853546143, + "learning_rate": 4.810941271118263e-06, + "loss": 1.0662, + "step": 203 + }, + { + "epoch": 0.2952243125904486, + "grad_norm": 1.6116409301757812, + "learning_rate": 4.806918744971843e-06, + "loss": 1.1058, + "step": 204 + }, + { + "epoch": 0.29667149059334297, + "grad_norm": 2.118859052658081, + "learning_rate": 4.802896218825423e-06, + "loss": 1.0844, + "step": 205 + }, + { + "epoch": 0.2981186685962373, + "grad_norm": 1.594258189201355, + "learning_rate": 4.798873692679003e-06, + "loss": 1.052, + "step": 206 + }, + { + "epoch": 0.2995658465991317, + "grad_norm": 1.9107189178466797, + "learning_rate": 4.794851166532583e-06, + "loss": 1.0667, + "step": 207 + }, + { + "epoch": 0.30101302460202606, + "grad_norm": 1.3610974550247192, + "learning_rate": 4.790828640386163e-06, + "loss": 1.0141, + "step": 208 + }, + { + "epoch": 0.3024602026049204, + "grad_norm": 2.100587844848633, + "learning_rate": 4.786806114239743e-06, + "loss": 1.0519, + "step": 209 + }, + { + "epoch": 0.30390738060781475, + "grad_norm": 1.706264615058899, + "learning_rate": 4.782783588093323e-06, + "loss": 1.0242, + "step": 210 + }, + { + "epoch": 0.30390738060781475, + "eval_loss": 1.123977541923523, + "eval_runtime": 51.2141, + "eval_samples_per_second": 19.526, + "eval_steps_per_second": 1.23, + "step": 210 + }, + { + "epoch": 0.3053545586107091, + "grad_norm": 1.2387949228286743, + "learning_rate": 4.778761061946903e-06, + "loss": 1.0411, + "step": 211 + }, + { + "epoch": 0.3068017366136035, + "grad_norm": 2.0283620357513428, + "learning_rate": 4.774738535800483e-06, + "loss": 1.056, + "step": 212 + }, + { + "epoch": 0.30824891461649784, + "grad_norm": 1.4211299419403076, + "learning_rate": 4.7707160096540635e-06, + "loss": 1.0378, + "step": 213 + }, + { + "epoch": 0.3096960926193922, + "grad_norm": 1.6764143705368042, + "learning_rate": 4.766693483507644e-06, + "loss": 1.0767, + "step": 214 + }, + { + "epoch": 0.3111432706222865, + "grad_norm": 1.655656099319458, + "learning_rate": 4.762670957361224e-06, + "loss": 1.0482, + "step": 215 + }, + { + "epoch": 0.3125904486251809, + "grad_norm": 1.40765380859375, + "learning_rate": 4.758648431214804e-06, + "loss": 1.0476, + "step": 216 + }, + { + "epoch": 0.3140376266280753, + "grad_norm": 1.43539297580719, + "learning_rate": 4.754625905068383e-06, + "loss": 1.0669, + "step": 217 + }, + { + "epoch": 0.3154848046309696, + "grad_norm": 1.4077202081680298, + "learning_rate": 4.750603378921963e-06, + "loss": 1.0479, + "step": 218 + }, + { + "epoch": 0.31693198263386396, + "grad_norm": 1.3172467947006226, + "learning_rate": 4.746580852775543e-06, + "loss": 1.0378, + "step": 219 + }, + { + "epoch": 0.3183791606367583, + "grad_norm": 1.3114285469055176, + "learning_rate": 4.7425583266291235e-06, + "loss": 1.0188, + "step": 220 + }, + { + "epoch": 0.3183791606367583, + "eval_loss": 1.1186376810073853, + "eval_runtime": 51.1809, + "eval_samples_per_second": 19.539, + "eval_steps_per_second": 1.231, + "step": 220 + }, + { + "epoch": 0.31982633863965265, + "grad_norm": 1.5616333484649658, + "learning_rate": 4.738535800482704e-06, + "loss": 1.0264, + "step": 221 + }, + { + "epoch": 0.32127351664254705, + "grad_norm": 1.5578783750534058, + "learning_rate": 4.734513274336284e-06, + "loss": 1.0458, + "step": 222 + }, + { + "epoch": 0.3227206946454414, + "grad_norm": 1.9564948081970215, + "learning_rate": 4.730490748189864e-06, + "loss": 1.0605, + "step": 223 + }, + { + "epoch": 0.32416787264833574, + "grad_norm": 1.758457899093628, + "learning_rate": 4.726468222043444e-06, + "loss": 0.9944, + "step": 224 + }, + { + "epoch": 0.3256150506512301, + "grad_norm": 1.431219220161438, + "learning_rate": 4.722445695897024e-06, + "loss": 0.9916, + "step": 225 + }, + { + "epoch": 0.32706222865412443, + "grad_norm": 1.3209437131881714, + "learning_rate": 4.718423169750604e-06, + "loss": 1.0194, + "step": 226 + }, + { + "epoch": 0.32850940665701883, + "grad_norm": 1.387919306755066, + "learning_rate": 4.7144006436041835e-06, + "loss": 1.0269, + "step": 227 + }, + { + "epoch": 0.3299565846599132, + "grad_norm": 2.0873770713806152, + "learning_rate": 4.710378117457764e-06, + "loss": 1.0378, + "step": 228 + }, + { + "epoch": 0.3314037626628075, + "grad_norm": 1.8863862752914429, + "learning_rate": 4.706355591311344e-06, + "loss": 1.0064, + "step": 229 + }, + { + "epoch": 0.33285094066570187, + "grad_norm": 1.8914672136306763, + "learning_rate": 4.702333065164924e-06, + "loss": 1.0449, + "step": 230 + }, + { + "epoch": 0.33285094066570187, + "eval_loss": 1.1199406385421753, + "eval_runtime": 51.1539, + "eval_samples_per_second": 19.549, + "eval_steps_per_second": 1.232, + "step": 230 + }, + { + "epoch": 0.3342981186685962, + "grad_norm": 1.54817533493042, + "learning_rate": 4.698310539018504e-06, + "loss": 1.0248, + "step": 231 + }, + { + "epoch": 0.3357452966714906, + "grad_norm": 1.6530121564865112, + "learning_rate": 4.694288012872084e-06, + "loss": 1.0217, + "step": 232 + }, + { + "epoch": 0.33719247467438496, + "grad_norm": 1.4780431985855103, + "learning_rate": 4.690265486725664e-06, + "loss": 1.043, + "step": 233 + }, + { + "epoch": 0.3386396526772793, + "grad_norm": 1.9577306509017944, + "learning_rate": 4.6862429605792444e-06, + "loss": 1.0388, + "step": 234 + }, + { + "epoch": 0.34008683068017365, + "grad_norm": 1.519266963005066, + "learning_rate": 4.6822204344328246e-06, + "loss": 1.0834, + "step": 235 + }, + { + "epoch": 0.341534008683068, + "grad_norm": 1.4943430423736572, + "learning_rate": 4.678197908286405e-06, + "loss": 1.0554, + "step": 236 + }, + { + "epoch": 0.3429811866859624, + "grad_norm": 2.1030423641204834, + "learning_rate": 4.674175382139984e-06, + "loss": 1.0445, + "step": 237 + }, + { + "epoch": 0.34442836468885674, + "grad_norm": 1.4801278114318848, + "learning_rate": 4.670152855993564e-06, + "loss": 1.0598, + "step": 238 + }, + { + "epoch": 0.3458755426917511, + "grad_norm": 1.7392442226409912, + "learning_rate": 4.666130329847144e-06, + "loss": 1.0449, + "step": 239 + }, + { + "epoch": 0.3473227206946454, + "grad_norm": 2.3473947048187256, + "learning_rate": 4.662107803700724e-06, + "loss": 1.0171, + "step": 240 + }, + { + "epoch": 0.3473227206946454, + "eval_loss": 1.1113197803497314, + "eval_runtime": 51.3235, + "eval_samples_per_second": 19.484, + "eval_steps_per_second": 1.228, + "step": 240 + }, + { + "epoch": 0.34876989869753977, + "grad_norm": 1.618821382522583, + "learning_rate": 4.6580852775543045e-06, + "loss": 1.0208, + "step": 241 + }, + { + "epoch": 0.35021707670043417, + "grad_norm": 1.3195146322250366, + "learning_rate": 4.654062751407885e-06, + "loss": 0.9922, + "step": 242 + }, + { + "epoch": 0.3516642547033285, + "grad_norm": 1.5132856369018555, + "learning_rate": 4.650040225261465e-06, + "loss": 1.0393, + "step": 243 + }, + { + "epoch": 0.35311143270622286, + "grad_norm": 1.803024172782898, + "learning_rate": 4.646017699115045e-06, + "loss": 1.0283, + "step": 244 + }, + { + "epoch": 0.3545586107091172, + "grad_norm": 1.4926753044128418, + "learning_rate": 4.641995172968625e-06, + "loss": 1.0365, + "step": 245 + }, + { + "epoch": 0.35600578871201155, + "grad_norm": 1.2645905017852783, + "learning_rate": 4.637972646822205e-06, + "loss": 1.0238, + "step": 246 + }, + { + "epoch": 0.35745296671490595, + "grad_norm": 1.7088773250579834, + "learning_rate": 4.633950120675784e-06, + "loss": 1.0321, + "step": 247 + }, + { + "epoch": 0.3589001447178003, + "grad_norm": 1.703622579574585, + "learning_rate": 4.6299275945293645e-06, + "loss": 1.0243, + "step": 248 + }, + { + "epoch": 0.36034732272069464, + "grad_norm": 1.5209614038467407, + "learning_rate": 4.625905068382945e-06, + "loss": 1.0465, + "step": 249 + }, + { + "epoch": 0.361794500723589, + "grad_norm": 1.7122691869735718, + "learning_rate": 4.621882542236525e-06, + "loss": 1.0047, + "step": 250 + }, + { + "epoch": 0.361794500723589, + "eval_loss": 1.1057441234588623, + "eval_runtime": 50.948, + "eval_samples_per_second": 19.628, + "eval_steps_per_second": 1.237, + "step": 250 + }, + { + "epoch": 0.36324167872648333, + "grad_norm": 1.6284708976745605, + "learning_rate": 4.617860016090105e-06, + "loss": 1.039, + "step": 251 + }, + { + "epoch": 0.36468885672937773, + "grad_norm": 1.718626856803894, + "learning_rate": 4.613837489943685e-06, + "loss": 1.0102, + "step": 252 + }, + { + "epoch": 0.3661360347322721, + "grad_norm": 1.3475052118301392, + "learning_rate": 4.609814963797265e-06, + "loss": 1.0045, + "step": 253 + }, + { + "epoch": 0.3675832127351664, + "grad_norm": 1.542887806892395, + "learning_rate": 4.605792437650845e-06, + "loss": 1.0249, + "step": 254 + }, + { + "epoch": 0.36903039073806077, + "grad_norm": 1.7273961305618286, + "learning_rate": 4.6017699115044254e-06, + "loss": 1.036, + "step": 255 + }, + { + "epoch": 0.3704775687409551, + "grad_norm": 1.499443531036377, + "learning_rate": 4.5977473853580056e-06, + "loss": 1.0018, + "step": 256 + }, + { + "epoch": 0.3719247467438495, + "grad_norm": 1.7369823455810547, + "learning_rate": 4.593724859211585e-06, + "loss": 1.0142, + "step": 257 + }, + { + "epoch": 0.37337192474674386, + "grad_norm": 1.3566123247146606, + "learning_rate": 4.589702333065165e-06, + "loss": 1.032, + "step": 258 + }, + { + "epoch": 0.3748191027496382, + "grad_norm": 1.3981618881225586, + "learning_rate": 4.585679806918745e-06, + "loss": 1.0316, + "step": 259 + }, + { + "epoch": 0.37626628075253254, + "grad_norm": 1.5296891927719116, + "learning_rate": 4.581657280772325e-06, + "loss": 1.0337, + "step": 260 + }, + { + "epoch": 0.37626628075253254, + "eval_loss": 1.1034107208251953, + "eval_runtime": 51.1778, + "eval_samples_per_second": 19.54, + "eval_steps_per_second": 1.231, + "step": 260 + }, + { + "epoch": 0.37771345875542695, + "grad_norm": 1.5452585220336914, + "learning_rate": 4.577634754625905e-06, + "loss": 0.9799, + "step": 261 + }, + { + "epoch": 0.3791606367583213, + "grad_norm": 1.6030676364898682, + "learning_rate": 4.5736122284794855e-06, + "loss": 1.0208, + "step": 262 + }, + { + "epoch": 0.38060781476121563, + "grad_norm": 1.541081190109253, + "learning_rate": 4.569589702333066e-06, + "loss": 1.0368, + "step": 263 + }, + { + "epoch": 0.38205499276411, + "grad_norm": 1.183112382888794, + "learning_rate": 4.565567176186646e-06, + "loss": 1.0501, + "step": 264 + }, + { + "epoch": 0.3835021707670043, + "grad_norm": 1.4399793148040771, + "learning_rate": 4.561544650040226e-06, + "loss": 1.0029, + "step": 265 + }, + { + "epoch": 0.3849493487698987, + "grad_norm": 1.597119927406311, + "learning_rate": 4.557522123893805e-06, + "loss": 1.0131, + "step": 266 + }, + { + "epoch": 0.38639652677279307, + "grad_norm": 1.8481191396713257, + "learning_rate": 4.553499597747385e-06, + "loss": 1.0149, + "step": 267 + }, + { + "epoch": 0.3878437047756874, + "grad_norm": 1.5463130474090576, + "learning_rate": 4.549477071600965e-06, + "loss": 1.0411, + "step": 268 + }, + { + "epoch": 0.38929088277858176, + "grad_norm": 1.4717342853546143, + "learning_rate": 4.5454545454545455e-06, + "loss": 1.0136, + "step": 269 + }, + { + "epoch": 0.3907380607814761, + "grad_norm": 1.454348087310791, + "learning_rate": 4.541432019308126e-06, + "loss": 1.0146, + "step": 270 + }, + { + "epoch": 0.3907380607814761, + "eval_loss": 1.09866201877594, + "eval_runtime": 51.2982, + "eval_samples_per_second": 19.494, + "eval_steps_per_second": 1.228, + "step": 270 + }, + { + "epoch": 0.3921852387843705, + "grad_norm": 1.3492728471755981, + "learning_rate": 4.537409493161706e-06, + "loss": 0.9702, + "step": 271 + }, + { + "epoch": 0.39363241678726485, + "grad_norm": 1.5021007061004639, + "learning_rate": 4.533386967015286e-06, + "loss": 1.0747, + "step": 272 + }, + { + "epoch": 0.3950795947901592, + "grad_norm": 1.5908703804016113, + "learning_rate": 4.529364440868866e-06, + "loss": 0.978, + "step": 273 + }, + { + "epoch": 0.39652677279305354, + "grad_norm": 1.405413269996643, + "learning_rate": 4.525341914722446e-06, + "loss": 1.0163, + "step": 274 + }, + { + "epoch": 0.3979739507959479, + "grad_norm": 1.7771649360656738, + "learning_rate": 4.521319388576026e-06, + "loss": 1.0006, + "step": 275 + }, + { + "epoch": 0.3994211287988423, + "grad_norm": 1.3909844160079956, + "learning_rate": 4.5172968624296056e-06, + "loss": 1.0173, + "step": 276 + }, + { + "epoch": 0.40086830680173663, + "grad_norm": 1.487363576889038, + "learning_rate": 4.513274336283186e-06, + "loss": 1.0177, + "step": 277 + }, + { + "epoch": 0.402315484804631, + "grad_norm": 2.217724561691284, + "learning_rate": 4.509251810136766e-06, + "loss": 1.0022, + "step": 278 + }, + { + "epoch": 0.4037626628075253, + "grad_norm": 1.415589690208435, + "learning_rate": 4.505229283990346e-06, + "loss": 1.0008, + "step": 279 + }, + { + "epoch": 0.40520984081041966, + "grad_norm": 1.359315276145935, + "learning_rate": 4.501206757843926e-06, + "loss": 1.009, + "step": 280 + }, + { + "epoch": 0.40520984081041966, + "eval_loss": 1.0961823463439941, + "eval_runtime": 51.2706, + "eval_samples_per_second": 19.504, + "eval_steps_per_second": 1.229, + "step": 280 + }, + { + "epoch": 0.40665701881331406, + "grad_norm": 1.984439730644226, + "learning_rate": 4.497184231697506e-06, + "loss": 1.0327, + "step": 281 + }, + { + "epoch": 0.4081041968162084, + "grad_norm": 2.3341264724731445, + "learning_rate": 4.493161705551086e-06, + "loss": 1.0504, + "step": 282 + }, + { + "epoch": 0.40955137481910275, + "grad_norm": 1.2871556282043457, + "learning_rate": 4.4891391794046665e-06, + "loss": 0.9874, + "step": 283 + }, + { + "epoch": 0.4109985528219971, + "grad_norm": 2.0350260734558105, + "learning_rate": 4.485116653258247e-06, + "loss": 1.0211, + "step": 284 + }, + { + "epoch": 0.41244573082489144, + "grad_norm": 1.3546667098999023, + "learning_rate": 4.481094127111827e-06, + "loss": 1.028, + "step": 285 + }, + { + "epoch": 0.41389290882778584, + "grad_norm": 1.4391071796417236, + "learning_rate": 4.477071600965406e-06, + "loss": 0.9915, + "step": 286 + }, + { + "epoch": 0.4153400868306802, + "grad_norm": 1.695353627204895, + "learning_rate": 4.473049074818986e-06, + "loss": 0.9949, + "step": 287 + }, + { + "epoch": 0.41678726483357453, + "grad_norm": 1.5350335836410522, + "learning_rate": 4.469026548672566e-06, + "loss": 1.0005, + "step": 288 + }, + { + "epoch": 0.4182344428364689, + "grad_norm": 1.6493104696273804, + "learning_rate": 4.465004022526146e-06, + "loss": 1.0265, + "step": 289 + }, + { + "epoch": 0.4196816208393632, + "grad_norm": 1.4068589210510254, + "learning_rate": 4.460981496379727e-06, + "loss": 1.0337, + "step": 290 + }, + { + "epoch": 0.4196816208393632, + "eval_loss": 1.0912277698516846, + "eval_runtime": 50.8325, + "eval_samples_per_second": 19.672, + "eval_steps_per_second": 1.239, + "step": 290 + }, + { + "epoch": 0.4211287988422576, + "grad_norm": 1.4879275560379028, + "learning_rate": 4.4569589702333075e-06, + "loss": 0.9986, + "step": 291 + }, + { + "epoch": 0.42257597684515197, + "grad_norm": 1.399894118309021, + "learning_rate": 4.452936444086888e-06, + "loss": 1.0177, + "step": 292 + }, + { + "epoch": 0.4240231548480463, + "grad_norm": 1.2316914796829224, + "learning_rate": 4.448913917940467e-06, + "loss": 0.998, + "step": 293 + }, + { + "epoch": 0.42547033285094066, + "grad_norm": 1.7620831727981567, + "learning_rate": 4.444891391794047e-06, + "loss": 1.0189, + "step": 294 + }, + { + "epoch": 0.426917510853835, + "grad_norm": 1.546459436416626, + "learning_rate": 4.440868865647627e-06, + "loss": 0.9808, + "step": 295 + }, + { + "epoch": 0.4283646888567294, + "grad_norm": 1.628462791442871, + "learning_rate": 4.436846339501207e-06, + "loss": 0.9966, + "step": 296 + }, + { + "epoch": 0.42981186685962375, + "grad_norm": 1.3183311223983765, + "learning_rate": 4.432823813354787e-06, + "loss": 0.9735, + "step": 297 + }, + { + "epoch": 0.4312590448625181, + "grad_norm": 1.7016503810882568, + "learning_rate": 4.4288012872083675e-06, + "loss": 0.9928, + "step": 298 + }, + { + "epoch": 0.43270622286541244, + "grad_norm": 1.475518822669983, + "learning_rate": 4.424778761061948e-06, + "loss": 1.0277, + "step": 299 + }, + { + "epoch": 0.4341534008683068, + "grad_norm": 1.5372337102890015, + "learning_rate": 4.420756234915528e-06, + "loss": 0.9633, + "step": 300 + }, + { + "epoch": 0.4341534008683068, + "eval_loss": 1.0854488611221313, + "eval_runtime": 51.2714, + "eval_samples_per_second": 19.504, + "eval_steps_per_second": 1.229, + "step": 300 + }, + { + "epoch": 0.4356005788712012, + "grad_norm": 1.4601222276687622, + "learning_rate": 4.416733708769108e-06, + "loss": 0.9668, + "step": 301 + }, + { + "epoch": 0.4370477568740955, + "grad_norm": 1.6669672727584839, + "learning_rate": 4.412711182622687e-06, + "loss": 1.0328, + "step": 302 + }, + { + "epoch": 0.4384949348769899, + "grad_norm": 1.5886998176574707, + "learning_rate": 4.408688656476267e-06, + "loss": 0.9798, + "step": 303 + }, + { + "epoch": 0.4399421128798842, + "grad_norm": 1.9831432104110718, + "learning_rate": 4.4046661303298474e-06, + "loss": 0.9958, + "step": 304 + }, + { + "epoch": 0.44138929088277856, + "grad_norm": 2.3375403881073, + "learning_rate": 4.4006436041834276e-06, + "loss": 0.9818, + "step": 305 + }, + { + "epoch": 0.44283646888567296, + "grad_norm": 1.6058458089828491, + "learning_rate": 4.396621078037008e-06, + "loss": 0.9865, + "step": 306 + }, + { + "epoch": 0.4442836468885673, + "grad_norm": 1.7550805807113647, + "learning_rate": 4.392598551890588e-06, + "loss": 0.9766, + "step": 307 + }, + { + "epoch": 0.44573082489146165, + "grad_norm": 2.3233633041381836, + "learning_rate": 4.388576025744168e-06, + "loss": 1.0618, + "step": 308 + }, + { + "epoch": 0.447178002894356, + "grad_norm": 2.0034706592559814, + "learning_rate": 4.384553499597748e-06, + "loss": 0.9856, + "step": 309 + }, + { + "epoch": 0.44862518089725034, + "grad_norm": 1.6992923021316528, + "learning_rate": 4.380530973451328e-06, + "loss": 0.9806, + "step": 310 + }, + { + "epoch": 0.44862518089725034, + "eval_loss": 1.087030053138733, + "eval_runtime": 50.7931, + "eval_samples_per_second": 19.688, + "eval_steps_per_second": 1.24, + "step": 310 + }, + { + "epoch": 0.45007235890014474, + "grad_norm": 1.8932524919509888, + "learning_rate": 4.376508447304908e-06, + "loss": 1.0312, + "step": 311 + }, + { + "epoch": 0.4515195369030391, + "grad_norm": 1.7426635026931763, + "learning_rate": 4.372485921158488e-06, + "loss": 0.9611, + "step": 312 + }, + { + "epoch": 0.45296671490593343, + "grad_norm": 1.8954734802246094, + "learning_rate": 4.368463395012068e-06, + "loss": 1.0044, + "step": 313 + }, + { + "epoch": 0.4544138929088278, + "grad_norm": 1.2427709102630615, + "learning_rate": 4.364440868865648e-06, + "loss": 0.9616, + "step": 314 + }, + { + "epoch": 0.4558610709117221, + "grad_norm": 1.4823901653289795, + "learning_rate": 4.360418342719228e-06, + "loss": 1.0179, + "step": 315 + }, + { + "epoch": 0.4573082489146165, + "grad_norm": 1.750918984413147, + "learning_rate": 4.356395816572808e-06, + "loss": 0.9606, + "step": 316 + }, + { + "epoch": 0.45875542691751087, + "grad_norm": 1.6604682207107544, + "learning_rate": 4.352373290426388e-06, + "loss": 1.0494, + "step": 317 + }, + { + "epoch": 0.4602026049204052, + "grad_norm": 1.9531415700912476, + "learning_rate": 4.348350764279968e-06, + "loss": 0.9687, + "step": 318 + }, + { + "epoch": 0.46164978292329956, + "grad_norm": 1.4689624309539795, + "learning_rate": 4.3443282381335485e-06, + "loss": 1.0051, + "step": 319 + }, + { + "epoch": 0.4630969609261939, + "grad_norm": 1.5099252462387085, + "learning_rate": 4.340305711987129e-06, + "loss": 1.0057, + "step": 320 + }, + { + "epoch": 0.4630969609261939, + "eval_loss": 1.0815813541412354, + "eval_runtime": 50.7053, + "eval_samples_per_second": 19.722, + "eval_steps_per_second": 1.242, + "step": 320 + }, + { + "epoch": 0.4645441389290883, + "grad_norm": 1.741703987121582, + "learning_rate": 4.336283185840709e-06, + "loss": 1.0032, + "step": 321 + }, + { + "epoch": 0.46599131693198265, + "grad_norm": 1.3489866256713867, + "learning_rate": 4.332260659694288e-06, + "loss": 0.9822, + "step": 322 + }, + { + "epoch": 0.467438494934877, + "grad_norm": 1.9755520820617676, + "learning_rate": 4.328238133547868e-06, + "loss": 1.0098, + "step": 323 + }, + { + "epoch": 0.46888567293777134, + "grad_norm": 1.6864471435546875, + "learning_rate": 4.324215607401448e-06, + "loss": 0.9984, + "step": 324 + }, + { + "epoch": 0.4703328509406657, + "grad_norm": 1.256195306777954, + "learning_rate": 4.3201930812550284e-06, + "loss": 0.9878, + "step": 325 + }, + { + "epoch": 0.4717800289435601, + "grad_norm": 1.7909395694732666, + "learning_rate": 4.3161705551086086e-06, + "loss": 0.9835, + "step": 326 + }, + { + "epoch": 0.4732272069464544, + "grad_norm": 1.5978225469589233, + "learning_rate": 4.312148028962189e-06, + "loss": 0.9754, + "step": 327 + }, + { + "epoch": 0.47467438494934877, + "grad_norm": 1.8620542287826538, + "learning_rate": 4.308125502815769e-06, + "loss": 1.0505, + "step": 328 + }, + { + "epoch": 0.4761215629522431, + "grad_norm": 1.3078632354736328, + "learning_rate": 4.304102976669349e-06, + "loss": 1.0184, + "step": 329 + }, + { + "epoch": 0.47756874095513746, + "grad_norm": 1.2472238540649414, + "learning_rate": 4.300080450522929e-06, + "loss": 1.0047, + "step": 330 + }, + { + "epoch": 0.47756874095513746, + "eval_loss": 1.0812331438064575, + "eval_runtime": 50.8867, + "eval_samples_per_second": 19.651, + "eval_steps_per_second": 1.238, + "step": 330 + }, + { + "epoch": 0.47901591895803186, + "grad_norm": 1.9471399784088135, + "learning_rate": 4.296057924376509e-06, + "loss": 0.9594, + "step": 331 + }, + { + "epoch": 0.4804630969609262, + "grad_norm": 1.9978971481323242, + "learning_rate": 4.2920353982300885e-06, + "loss": 1.001, + "step": 332 + }, + { + "epoch": 0.48191027496382055, + "grad_norm": 1.4921774864196777, + "learning_rate": 4.288012872083669e-06, + "loss": 1.0068, + "step": 333 + }, + { + "epoch": 0.4833574529667149, + "grad_norm": 1.3463438749313354, + "learning_rate": 4.283990345937249e-06, + "loss": 1.0036, + "step": 334 + }, + { + "epoch": 0.48480463096960924, + "grad_norm": 1.5217732191085815, + "learning_rate": 4.279967819790829e-06, + "loss": 0.9829, + "step": 335 + }, + { + "epoch": 0.48625180897250364, + "grad_norm": 1.6425414085388184, + "learning_rate": 4.275945293644409e-06, + "loss": 0.9927, + "step": 336 + }, + { + "epoch": 0.487698986975398, + "grad_norm": 1.3807939291000366, + "learning_rate": 4.271922767497989e-06, + "loss": 0.9925, + "step": 337 + }, + { + "epoch": 0.48914616497829233, + "grad_norm": 1.6897764205932617, + "learning_rate": 4.267900241351569e-06, + "loss": 0.9941, + "step": 338 + }, + { + "epoch": 0.4905933429811867, + "grad_norm": 1.3020832538604736, + "learning_rate": 4.263877715205149e-06, + "loss": 0.9866, + "step": 339 + }, + { + "epoch": 0.492040520984081, + "grad_norm": 1.3091801404953003, + "learning_rate": 4.2598551890587295e-06, + "loss": 0.9604, + "step": 340 + }, + { + "epoch": 0.492040520984081, + "eval_loss": 1.0793449878692627, + "eval_runtime": 51.066, + "eval_samples_per_second": 19.583, + "eval_steps_per_second": 1.234, + "step": 340 + }, + { + "epoch": 0.4934876989869754, + "grad_norm": 1.5976084470748901, + "learning_rate": 4.25583266291231e-06, + "loss": 1.0037, + "step": 341 + }, + { + "epoch": 0.49493487698986977, + "grad_norm": 1.6975218057632446, + "learning_rate": 4.251810136765889e-06, + "loss": 1.0036, + "step": 342 + }, + { + "epoch": 0.4963820549927641, + "grad_norm": 1.2496190071105957, + "learning_rate": 4.247787610619469e-06, + "loss": 0.9458, + "step": 343 + }, + { + "epoch": 0.49782923299565845, + "grad_norm": 1.4577810764312744, + "learning_rate": 4.243765084473049e-06, + "loss": 0.988, + "step": 344 + }, + { + "epoch": 0.4992764109985528, + "grad_norm": 1.416592001914978, + "learning_rate": 4.239742558326629e-06, + "loss": 0.9901, + "step": 345 + }, + { + "epoch": 0.5007235890014472, + "grad_norm": 1.1948094367980957, + "learning_rate": 4.235720032180209e-06, + "loss": 0.9803, + "step": 346 + }, + { + "epoch": 0.5021707670043415, + "grad_norm": 1.2622143030166626, + "learning_rate": 4.2316975060337895e-06, + "loss": 0.956, + "step": 347 + }, + { + "epoch": 0.5036179450072359, + "grad_norm": 1.306982159614563, + "learning_rate": 4.22767497988737e-06, + "loss": 0.994, + "step": 348 + }, + { + "epoch": 0.5050651230101303, + "grad_norm": 1.956438660621643, + "learning_rate": 4.22365245374095e-06, + "loss": 0.9717, + "step": 349 + }, + { + "epoch": 0.5065123010130246, + "grad_norm": 1.4835199117660522, + "learning_rate": 4.21962992759453e-06, + "loss": 0.9692, + "step": 350 + }, + { + "epoch": 0.5065123010130246, + "eval_loss": 1.0755901336669922, + "eval_runtime": 50.685, + "eval_samples_per_second": 19.73, + "eval_steps_per_second": 1.243, + "step": 350 + }, + { + "epoch": 0.507959479015919, + "grad_norm": 1.4217889308929443, + "learning_rate": 4.21560740144811e-06, + "loss": 1.0041, + "step": 351 + }, + { + "epoch": 0.5094066570188133, + "grad_norm": 1.3517321348190308, + "learning_rate": 4.211584875301689e-06, + "loss": 1.0128, + "step": 352 + }, + { + "epoch": 0.5108538350217077, + "grad_norm": 1.454322099685669, + "learning_rate": 4.2075623491552695e-06, + "loss": 0.9741, + "step": 353 + }, + { + "epoch": 0.5123010130246021, + "grad_norm": 1.615440011024475, + "learning_rate": 4.20353982300885e-06, + "loss": 1.0165, + "step": 354 + }, + { + "epoch": 0.5137481910274964, + "grad_norm": 1.184549331665039, + "learning_rate": 4.19951729686243e-06, + "loss": 0.9467, + "step": 355 + }, + { + "epoch": 0.5151953690303908, + "grad_norm": 1.431377649307251, + "learning_rate": 4.19549477071601e-06, + "loss": 1.0366, + "step": 356 + }, + { + "epoch": 0.516642547033285, + "grad_norm": 1.4371246099472046, + "learning_rate": 4.19147224456959e-06, + "loss": 0.9886, + "step": 357 + }, + { + "epoch": 0.5180897250361794, + "grad_norm": 1.223812460899353, + "learning_rate": 4.18744971842317e-06, + "loss": 0.9792, + "step": 358 + }, + { + "epoch": 0.5195369030390738, + "grad_norm": 1.5297192335128784, + "learning_rate": 4.18342719227675e-06, + "loss": 1.0109, + "step": 359 + }, + { + "epoch": 0.5209840810419681, + "grad_norm": 1.4062352180480957, + "learning_rate": 4.17940466613033e-06, + "loss": 0.9646, + "step": 360 + }, + { + "epoch": 0.5209840810419681, + "eval_loss": 1.072110652923584, + "eval_runtime": 51.2482, + "eval_samples_per_second": 19.513, + "eval_steps_per_second": 1.229, + "step": 360 + }, + { + "epoch": 0.5224312590448625, + "grad_norm": 1.1919447183609009, + "learning_rate": 4.1753821399839105e-06, + "loss": 1.0076, + "step": 361 + }, + { + "epoch": 0.5238784370477568, + "grad_norm": 1.5904113054275513, + "learning_rate": 4.17135961383749e-06, + "loss": 0.9746, + "step": 362 + }, + { + "epoch": 0.5253256150506512, + "grad_norm": 1.379450798034668, + "learning_rate": 4.16733708769107e-06, + "loss": 0.9717, + "step": 363 + }, + { + "epoch": 0.5267727930535456, + "grad_norm": 1.4465031623840332, + "learning_rate": 4.16331456154465e-06, + "loss": 0.9791, + "step": 364 + }, + { + "epoch": 0.5282199710564399, + "grad_norm": 1.9270684719085693, + "learning_rate": 4.15929203539823e-06, + "loss": 0.9864, + "step": 365 + }, + { + "epoch": 0.5296671490593343, + "grad_norm": 1.644336223602295, + "learning_rate": 4.15526950925181e-06, + "loss": 0.9765, + "step": 366 + }, + { + "epoch": 0.5311143270622286, + "grad_norm": 1.5674631595611572, + "learning_rate": 4.15124698310539e-06, + "loss": 0.9555, + "step": 367 + }, + { + "epoch": 0.532561505065123, + "grad_norm": 1.5348894596099854, + "learning_rate": 4.1472244569589705e-06, + "loss": 0.9801, + "step": 368 + }, + { + "epoch": 0.5340086830680174, + "grad_norm": 1.5066866874694824, + "learning_rate": 4.143201930812551e-06, + "loss": 0.9779, + "step": 369 + }, + { + "epoch": 0.5354558610709117, + "grad_norm": 1.4870381355285645, + "learning_rate": 4.139179404666131e-06, + "loss": 0.972, + "step": 370 + }, + { + "epoch": 0.5354558610709117, + "eval_loss": 1.0730079412460327, + "eval_runtime": 51.2522, + "eval_samples_per_second": 19.511, + "eval_steps_per_second": 1.229, + "step": 370 + }, + { + "epoch": 0.5369030390738061, + "grad_norm": 1.3844788074493408, + "learning_rate": 4.13515687851971e-06, + "loss": 0.9709, + "step": 371 + }, + { + "epoch": 0.5383502170767004, + "grad_norm": 1.3862665891647339, + "learning_rate": 4.13113435237329e-06, + "loss": 0.9395, + "step": 372 + }, + { + "epoch": 0.5397973950795948, + "grad_norm": 1.5485719442367554, + "learning_rate": 4.12711182622687e-06, + "loss": 0.9718, + "step": 373 + }, + { + "epoch": 0.5412445730824892, + "grad_norm": 1.6063337326049805, + "learning_rate": 4.1230893000804505e-06, + "loss": 0.988, + "step": 374 + }, + { + "epoch": 0.5426917510853835, + "grad_norm": 1.5300590991973877, + "learning_rate": 4.119066773934031e-06, + "loss": 0.9862, + "step": 375 + }, + { + "epoch": 0.5441389290882779, + "grad_norm": 1.2461405992507935, + "learning_rate": 4.115044247787611e-06, + "loss": 0.9534, + "step": 376 + }, + { + "epoch": 0.5455861070911722, + "grad_norm": 1.5977176427841187, + "learning_rate": 4.111021721641191e-06, + "loss": 0.9454, + "step": 377 + }, + { + "epoch": 0.5470332850940666, + "grad_norm": 1.5744560956954956, + "learning_rate": 4.106999195494771e-06, + "loss": 0.9663, + "step": 378 + }, + { + "epoch": 0.548480463096961, + "grad_norm": 1.6001098155975342, + "learning_rate": 4.102976669348351e-06, + "loss": 0.9924, + "step": 379 + }, + { + "epoch": 0.5499276410998553, + "grad_norm": 1.6721731424331665, + "learning_rate": 4.098954143201931e-06, + "loss": 1.0145, + "step": 380 + }, + { + "epoch": 0.5499276410998553, + "eval_loss": 1.07347571849823, + "eval_runtime": 51.2174, + "eval_samples_per_second": 19.525, + "eval_steps_per_second": 1.23, + "step": 380 + }, + { + "epoch": 0.5513748191027497, + "grad_norm": 1.436753749847412, + "learning_rate": 4.094931617055511e-06, + "loss": 0.9442, + "step": 381 + }, + { + "epoch": 0.552821997105644, + "grad_norm": 1.6411333084106445, + "learning_rate": 4.0909090909090915e-06, + "loss": 0.9776, + "step": 382 + }, + { + "epoch": 0.5542691751085383, + "grad_norm": 1.8081423044204712, + "learning_rate": 4.086886564762672e-06, + "loss": 0.9738, + "step": 383 + }, + { + "epoch": 0.5557163531114327, + "grad_norm": 1.7487365007400513, + "learning_rate": 4.082864038616252e-06, + "loss": 1.0061, + "step": 384 + }, + { + "epoch": 0.557163531114327, + "grad_norm": 1.56580650806427, + "learning_rate": 4.078841512469832e-06, + "loss": 1.0139, + "step": 385 + }, + { + "epoch": 0.5586107091172214, + "grad_norm": 1.2649954557418823, + "learning_rate": 4.074818986323412e-06, + "loss": 0.9948, + "step": 386 + }, + { + "epoch": 0.5600578871201157, + "grad_norm": 2.0901412963867188, + "learning_rate": 4.070796460176992e-06, + "loss": 0.9325, + "step": 387 + }, + { + "epoch": 0.5615050651230101, + "grad_norm": 1.5386650562286377, + "learning_rate": 4.066773934030571e-06, + "loss": 0.9323, + "step": 388 + }, + { + "epoch": 0.5629522431259045, + "grad_norm": 1.3811073303222656, + "learning_rate": 4.0627514078841515e-06, + "loss": 0.9827, + "step": 389 + }, + { + "epoch": 0.5643994211287988, + "grad_norm": 1.4223031997680664, + "learning_rate": 4.058728881737732e-06, + "loss": 0.9727, + "step": 390 + }, + { + "epoch": 0.5643994211287988, + "eval_loss": 1.0686701536178589, + "eval_runtime": 51.2014, + "eval_samples_per_second": 19.531, + "eval_steps_per_second": 1.23, + "step": 390 + }, + { + "epoch": 0.5658465991316932, + "grad_norm": 1.455756425857544, + "learning_rate": 4.054706355591312e-06, + "loss": 0.9722, + "step": 391 + }, + { + "epoch": 0.5672937771345875, + "grad_norm": 1.876002550125122, + "learning_rate": 4.050683829444892e-06, + "loss": 0.9915, + "step": 392 + }, + { + "epoch": 0.5687409551374819, + "grad_norm": 1.491150140762329, + "learning_rate": 4.046661303298472e-06, + "loss": 0.9849, + "step": 393 + }, + { + "epoch": 0.5701881331403763, + "grad_norm": 1.3917310237884521, + "learning_rate": 4.042638777152052e-06, + "loss": 0.9677, + "step": 394 + }, + { + "epoch": 0.5716353111432706, + "grad_norm": 1.6309757232666016, + "learning_rate": 4.038616251005632e-06, + "loss": 0.9806, + "step": 395 + }, + { + "epoch": 0.573082489146165, + "grad_norm": 1.591487169265747, + "learning_rate": 4.034593724859212e-06, + "loss": 0.9924, + "step": 396 + }, + { + "epoch": 0.5745296671490593, + "grad_norm": 1.4842720031738281, + "learning_rate": 4.0305711987127925e-06, + "loss": 0.9706, + "step": 397 + }, + { + "epoch": 0.5759768451519537, + "grad_norm": 1.2435634136199951, + "learning_rate": 4.026548672566372e-06, + "loss": 0.9929, + "step": 398 + }, + { + "epoch": 0.5774240231548481, + "grad_norm": 1.4005391597747803, + "learning_rate": 4.022526146419952e-06, + "loss": 0.9782, + "step": 399 + }, + { + "epoch": 0.5788712011577424, + "grad_norm": 1.1938444375991821, + "learning_rate": 4.018503620273532e-06, + "loss": 0.9832, + "step": 400 + }, + { + "epoch": 0.5788712011577424, + "eval_loss": 1.0635030269622803, + "eval_runtime": 51.1261, + "eval_samples_per_second": 19.559, + "eval_steps_per_second": 1.232, + "step": 400 + }, + { + "epoch": 0.5803183791606368, + "grad_norm": 1.3605155944824219, + "learning_rate": 4.014481094127112e-06, + "loss": 0.9716, + "step": 401 + }, + { + "epoch": 0.5817655571635311, + "grad_norm": 1.1320856809616089, + "learning_rate": 4.010458567980692e-06, + "loss": 0.9674, + "step": 402 + }, + { + "epoch": 0.5832127351664255, + "grad_norm": 1.4511500597000122, + "learning_rate": 4.0064360418342725e-06, + "loss": 0.9628, + "step": 403 + }, + { + "epoch": 0.5846599131693199, + "grad_norm": 1.3960020542144775, + "learning_rate": 4.002413515687853e-06, + "loss": 1.0003, + "step": 404 + }, + { + "epoch": 0.5861070911722142, + "grad_norm": 1.4735814332962036, + "learning_rate": 3.998390989541433e-06, + "loss": 1.0052, + "step": 405 + }, + { + "epoch": 0.5875542691751086, + "grad_norm": 1.254914402961731, + "learning_rate": 3.994368463395013e-06, + "loss": 1.008, + "step": 406 + }, + { + "epoch": 0.5890014471780028, + "grad_norm": 1.3346195220947266, + "learning_rate": 3.990345937248592e-06, + "loss": 0.9802, + "step": 407 + }, + { + "epoch": 0.5904486251808972, + "grad_norm": 1.4377028942108154, + "learning_rate": 3.986323411102172e-06, + "loss": 0.9535, + "step": 408 + }, + { + "epoch": 0.5918958031837916, + "grad_norm": 1.4920870065689087, + "learning_rate": 3.982300884955752e-06, + "loss": 0.9556, + "step": 409 + }, + { + "epoch": 0.5933429811866859, + "grad_norm": 1.6939969062805176, + "learning_rate": 3.9782783588093325e-06, + "loss": 0.9476, + "step": 410 + }, + { + "epoch": 0.5933429811866859, + "eval_loss": 1.0602790117263794, + "eval_runtime": 51.5365, + "eval_samples_per_second": 19.404, + "eval_steps_per_second": 1.222, + "step": 410 + }, + { + "epoch": 0.5947901591895803, + "grad_norm": 1.638198971748352, + "learning_rate": 3.974255832662913e-06, + "loss": 0.9706, + "step": 411 + }, + { + "epoch": 0.5962373371924746, + "grad_norm": 1.4514776468276978, + "learning_rate": 3.970233306516493e-06, + "loss": 0.953, + "step": 412 + }, + { + "epoch": 0.597684515195369, + "grad_norm": 1.6052072048187256, + "learning_rate": 3.966210780370073e-06, + "loss": 0.9355, + "step": 413 + }, + { + "epoch": 0.5991316931982634, + "grad_norm": 1.8618943691253662, + "learning_rate": 3.962188254223653e-06, + "loss": 0.9863, + "step": 414 + }, + { + "epoch": 0.6005788712011577, + "grad_norm": 1.7248200178146362, + "learning_rate": 3.958165728077233e-06, + "loss": 0.9909, + "step": 415 + }, + { + "epoch": 0.6020260492040521, + "grad_norm": 1.452968955039978, + "learning_rate": 3.954143201930813e-06, + "loss": 0.9445, + "step": 416 + }, + { + "epoch": 0.6034732272069464, + "grad_norm": 1.4949851036071777, + "learning_rate": 3.9501206757843926e-06, + "loss": 0.9381, + "step": 417 + }, + { + "epoch": 0.6049204052098408, + "grad_norm": 1.5896854400634766, + "learning_rate": 3.946098149637973e-06, + "loss": 1.0008, + "step": 418 + }, + { + "epoch": 0.6063675832127352, + "grad_norm": 1.419547438621521, + "learning_rate": 3.942075623491553e-06, + "loss": 0.9675, + "step": 419 + }, + { + "epoch": 0.6078147612156295, + "grad_norm": 1.2070773839950562, + "learning_rate": 3.938053097345133e-06, + "loss": 0.9354, + "step": 420 + }, + { + "epoch": 0.6078147612156295, + "eval_loss": 1.0614653825759888, + "eval_runtime": 51.5384, + "eval_samples_per_second": 19.403, + "eval_steps_per_second": 1.222, + "step": 420 + }, + { + "epoch": 0.6092619392185239, + "grad_norm": 1.4175598621368408, + "learning_rate": 3.934030571198713e-06, + "loss": 0.9487, + "step": 421 + }, + { + "epoch": 0.6107091172214182, + "grad_norm": 1.390889048576355, + "learning_rate": 3.930008045052293e-06, + "loss": 0.9837, + "step": 422 + }, + { + "epoch": 0.6121562952243126, + "grad_norm": 1.1224489212036133, + "learning_rate": 3.925985518905873e-06, + "loss": 0.9492, + "step": 423 + }, + { + "epoch": 0.613603473227207, + "grad_norm": 1.5353734493255615, + "learning_rate": 3.9219629927594534e-06, + "loss": 1.0134, + "step": 424 + }, + { + "epoch": 0.6150506512301013, + "grad_norm": 1.1764028072357178, + "learning_rate": 3.9179404666130336e-06, + "loss": 0.972, + "step": 425 + }, + { + "epoch": 0.6164978292329957, + "grad_norm": 1.5113427639007568, + "learning_rate": 3.913917940466614e-06, + "loss": 0.9543, + "step": 426 + }, + { + "epoch": 0.61794500723589, + "grad_norm": 1.3129740953445435, + "learning_rate": 3.909895414320193e-06, + "loss": 0.9469, + "step": 427 + }, + { + "epoch": 0.6193921852387844, + "grad_norm": 1.0554115772247314, + "learning_rate": 3.905872888173773e-06, + "loss": 0.9378, + "step": 428 + }, + { + "epoch": 0.6208393632416788, + "grad_norm": 1.6029083728790283, + "learning_rate": 3.901850362027353e-06, + "loss": 0.9556, + "step": 429 + }, + { + "epoch": 0.622286541244573, + "grad_norm": 1.2108154296875, + "learning_rate": 3.897827835880933e-06, + "loss": 0.96, + "step": 430 + }, + { + "epoch": 0.622286541244573, + "eval_loss": 1.058322548866272, + "eval_runtime": 51.3654, + "eval_samples_per_second": 19.468, + "eval_steps_per_second": 1.227, + "step": 430 + }, + { + "epoch": 0.6237337192474675, + "grad_norm": 1.1471012830734253, + "learning_rate": 3.8938053097345135e-06, + "loss": 0.9542, + "step": 431 + }, + { + "epoch": 0.6251808972503617, + "grad_norm": 1.0930272340774536, + "learning_rate": 3.889782783588094e-06, + "loss": 0.9648, + "step": 432 + }, + { + "epoch": 0.6266280752532561, + "grad_norm": 1.1586384773254395, + "learning_rate": 3.885760257441674e-06, + "loss": 0.9686, + "step": 433 + }, + { + "epoch": 0.6280752532561505, + "grad_norm": 1.5384842157363892, + "learning_rate": 3.881737731295254e-06, + "loss": 1.0002, + "step": 434 + }, + { + "epoch": 0.6295224312590448, + "grad_norm": 1.2638671398162842, + "learning_rate": 3.877715205148834e-06, + "loss": 0.935, + "step": 435 + }, + { + "epoch": 0.6309696092619392, + "grad_norm": 1.6291896104812622, + "learning_rate": 3.873692679002414e-06, + "loss": 0.944, + "step": 436 + }, + { + "epoch": 0.6324167872648335, + "grad_norm": 1.8233437538146973, + "learning_rate": 3.869670152855993e-06, + "loss": 0.993, + "step": 437 + }, + { + "epoch": 0.6338639652677279, + "grad_norm": 1.2835270166397095, + "learning_rate": 3.8656476267095735e-06, + "loss": 1.0125, + "step": 438 + }, + { + "epoch": 0.6353111432706223, + "grad_norm": 1.2744511365890503, + "learning_rate": 3.861625100563154e-06, + "loss": 0.9781, + "step": 439 + }, + { + "epoch": 0.6367583212735166, + "grad_norm": 1.4396259784698486, + "learning_rate": 3.857602574416734e-06, + "loss": 0.9516, + "step": 440 + }, + { + "epoch": 0.6367583212735166, + "eval_loss": 1.0518107414245605, + "eval_runtime": 51.3856, + "eval_samples_per_second": 19.461, + "eval_steps_per_second": 1.226, + "step": 440 + }, + { + "epoch": 0.638205499276411, + "grad_norm": 1.5060687065124512, + "learning_rate": 3.853580048270314e-06, + "loss": 0.9322, + "step": 441 + }, + { + "epoch": 0.6396526772793053, + "grad_norm": 1.7582062482833862, + "learning_rate": 3.849557522123894e-06, + "loss": 0.9724, + "step": 442 + }, + { + "epoch": 0.6410998552821997, + "grad_norm": 1.402211308479309, + "learning_rate": 3.845534995977474e-06, + "loss": 0.972, + "step": 443 + }, + { + "epoch": 0.6425470332850941, + "grad_norm": 1.4355642795562744, + "learning_rate": 3.841512469831054e-06, + "loss": 0.9414, + "step": 444 + }, + { + "epoch": 0.6439942112879884, + "grad_norm": 1.4574646949768066, + "learning_rate": 3.8374899436846344e-06, + "loss": 0.9444, + "step": 445 + }, + { + "epoch": 0.6454413892908828, + "grad_norm": 1.2893948554992676, + "learning_rate": 3.8334674175382146e-06, + "loss": 0.9368, + "step": 446 + }, + { + "epoch": 0.6468885672937771, + "grad_norm": 1.0977437496185303, + "learning_rate": 3.829444891391794e-06, + "loss": 0.992, + "step": 447 + }, + { + "epoch": 0.6483357452966715, + "grad_norm": 1.7038030624389648, + "learning_rate": 3.825422365245374e-06, + "loss": 0.9569, + "step": 448 + }, + { + "epoch": 0.6497829232995659, + "grad_norm": 1.5461138486862183, + "learning_rate": 3.821399839098954e-06, + "loss": 0.9979, + "step": 449 + }, + { + "epoch": 0.6512301013024602, + "grad_norm": 1.4361249208450317, + "learning_rate": 3.817377312952534e-06, + "loss": 0.9817, + "step": 450 + }, + { + "epoch": 0.6512301013024602, + "eval_loss": 1.0517939329147339, + "eval_runtime": 51.384, + "eval_samples_per_second": 19.461, + "eval_steps_per_second": 1.226, + "step": 450 + }, + { + "epoch": 0.6526772793053546, + "grad_norm": 1.6348068714141846, + "learning_rate": 3.8133547868061144e-06, + "loss": 0.9509, + "step": 451 + }, + { + "epoch": 0.6541244573082489, + "grad_norm": 1.1530169248580933, + "learning_rate": 3.8093322606596945e-06, + "loss": 0.9555, + "step": 452 + }, + { + "epoch": 0.6555716353111433, + "grad_norm": 1.7867792844772339, + "learning_rate": 3.8053097345132746e-06, + "loss": 0.9541, + "step": 453 + }, + { + "epoch": 0.6570188133140377, + "grad_norm": 1.423865795135498, + "learning_rate": 3.8012872083668543e-06, + "loss": 0.9387, + "step": 454 + }, + { + "epoch": 0.658465991316932, + "grad_norm": 1.3634495735168457, + "learning_rate": 3.7972646822204344e-06, + "loss": 0.9174, + "step": 455 + }, + { + "epoch": 0.6599131693198264, + "grad_norm": 1.5767061710357666, + "learning_rate": 3.7932421560740146e-06, + "loss": 0.9809, + "step": 456 + }, + { + "epoch": 0.6613603473227206, + "grad_norm": 1.583375334739685, + "learning_rate": 3.7892196299275947e-06, + "loss": 0.9752, + "step": 457 + }, + { + "epoch": 0.662807525325615, + "grad_norm": 1.2754613161087036, + "learning_rate": 3.785197103781175e-06, + "loss": 0.9636, + "step": 458 + }, + { + "epoch": 0.6642547033285094, + "grad_norm": 1.3666307926177979, + "learning_rate": 3.7811745776347545e-06, + "loss": 0.9469, + "step": 459 + }, + { + "epoch": 0.6657018813314037, + "grad_norm": 1.3346625566482544, + "learning_rate": 3.7771520514883347e-06, + "loss": 0.9641, + "step": 460 + }, + { + "epoch": 0.6657018813314037, + "eval_loss": 1.0529979467391968, + "eval_runtime": 51.434, + "eval_samples_per_second": 19.442, + "eval_steps_per_second": 1.225, + "step": 460 + }, + { + "epoch": 0.6671490593342981, + "grad_norm": 1.5265439748764038, + "learning_rate": 3.7731295253419148e-06, + "loss": 0.9835, + "step": 461 + }, + { + "epoch": 0.6685962373371924, + "grad_norm": 1.306519865989685, + "learning_rate": 3.769106999195495e-06, + "loss": 0.9882, + "step": 462 + }, + { + "epoch": 0.6700434153400868, + "grad_norm": 1.1957255601882935, + "learning_rate": 3.765084473049075e-06, + "loss": 0.9494, + "step": 463 + }, + { + "epoch": 0.6714905933429812, + "grad_norm": 1.362594485282898, + "learning_rate": 3.7610619469026547e-06, + "loss": 0.9853, + "step": 464 + }, + { + "epoch": 0.6729377713458755, + "grad_norm": 1.2525136470794678, + "learning_rate": 3.757039420756235e-06, + "loss": 0.9877, + "step": 465 + }, + { + "epoch": 0.6743849493487699, + "grad_norm": 1.3395814895629883, + "learning_rate": 3.753016894609815e-06, + "loss": 0.9393, + "step": 466 + }, + { + "epoch": 0.6758321273516642, + "grad_norm": 1.2523000240325928, + "learning_rate": 3.7489943684633956e-06, + "loss": 0.945, + "step": 467 + }, + { + "epoch": 0.6772793053545586, + "grad_norm": 1.2548216581344604, + "learning_rate": 3.7449718423169757e-06, + "loss": 0.9501, + "step": 468 + }, + { + "epoch": 0.678726483357453, + "grad_norm": 1.494130253791809, + "learning_rate": 3.740949316170556e-06, + "loss": 0.9542, + "step": 469 + }, + { + "epoch": 0.6801736613603473, + "grad_norm": 1.3851969242095947, + "learning_rate": 3.7369267900241355e-06, + "loss": 0.9839, + "step": 470 + }, + { + "epoch": 0.6801736613603473, + "eval_loss": 1.0494792461395264, + "eval_runtime": 51.5082, + "eval_samples_per_second": 19.414, + "eval_steps_per_second": 1.223, + "step": 470 + }, + { + "epoch": 0.6816208393632417, + "grad_norm": 1.4386080503463745, + "learning_rate": 3.7329042638777156e-06, + "loss": 0.9956, + "step": 471 + }, + { + "epoch": 0.683068017366136, + "grad_norm": 1.5325937271118164, + "learning_rate": 3.7288817377312958e-06, + "loss": 0.9641, + "step": 472 + }, + { + "epoch": 0.6845151953690304, + "grad_norm": 1.5930521488189697, + "learning_rate": 3.724859211584876e-06, + "loss": 0.9535, + "step": 473 + }, + { + "epoch": 0.6859623733719248, + "grad_norm": 1.6067936420440674, + "learning_rate": 3.720836685438456e-06, + "loss": 0.9787, + "step": 474 + }, + { + "epoch": 0.6874095513748191, + "grad_norm": 1.3350180387496948, + "learning_rate": 3.7168141592920357e-06, + "loss": 0.9727, + "step": 475 + }, + { + "epoch": 0.6888567293777135, + "grad_norm": 1.4819083213806152, + "learning_rate": 3.712791633145616e-06, + "loss": 0.9427, + "step": 476 + }, + { + "epoch": 0.6903039073806078, + "grad_norm": 1.7812278270721436, + "learning_rate": 3.708769106999196e-06, + "loss": 0.9618, + "step": 477 + }, + { + "epoch": 0.6917510853835022, + "grad_norm": 1.4724676609039307, + "learning_rate": 3.704746580852776e-06, + "loss": 0.9289, + "step": 478 + }, + { + "epoch": 0.6931982633863966, + "grad_norm": 1.5866022109985352, + "learning_rate": 3.7007240547063562e-06, + "loss": 0.9469, + "step": 479 + }, + { + "epoch": 0.6946454413892909, + "grad_norm": 1.8160587549209595, + "learning_rate": 3.696701528559936e-06, + "loss": 0.963, + "step": 480 + }, + { + "epoch": 0.6946454413892909, + "eval_loss": 1.047990322113037, + "eval_runtime": 51.5253, + "eval_samples_per_second": 19.408, + "eval_steps_per_second": 1.223, + "step": 480 + }, + { + "epoch": 0.6960926193921853, + "grad_norm": 1.3968987464904785, + "learning_rate": 3.692679002413516e-06, + "loss": 0.8995, + "step": 481 + }, + { + "epoch": 0.6975397973950795, + "grad_norm": 1.8819745779037476, + "learning_rate": 3.688656476267096e-06, + "loss": 0.9435, + "step": 482 + }, + { + "epoch": 0.6989869753979739, + "grad_norm": 1.186019778251648, + "learning_rate": 3.6846339501206763e-06, + "loss": 0.9409, + "step": 483 + }, + { + "epoch": 0.7004341534008683, + "grad_norm": 1.4415274858474731, + "learning_rate": 3.6806114239742565e-06, + "loss": 0.9718, + "step": 484 + }, + { + "epoch": 0.7018813314037626, + "grad_norm": 1.6194242238998413, + "learning_rate": 3.676588897827836e-06, + "loss": 0.9484, + "step": 485 + }, + { + "epoch": 0.703328509406657, + "grad_norm": 1.5799955129623413, + "learning_rate": 3.6725663716814163e-06, + "loss": 0.9775, + "step": 486 + }, + { + "epoch": 0.7047756874095513, + "grad_norm": 1.3916490077972412, + "learning_rate": 3.6685438455349964e-06, + "loss": 0.9478, + "step": 487 + }, + { + "epoch": 0.7062228654124457, + "grad_norm": 1.2659951448440552, + "learning_rate": 3.6645213193885765e-06, + "loss": 0.9431, + "step": 488 + }, + { + "epoch": 0.7076700434153401, + "grad_norm": 1.0781667232513428, + "learning_rate": 3.6604987932421567e-06, + "loss": 0.9336, + "step": 489 + }, + { + "epoch": 0.7091172214182344, + "grad_norm": 1.1809048652648926, + "learning_rate": 3.6564762670957364e-06, + "loss": 0.9579, + "step": 490 + }, + { + "epoch": 0.7091172214182344, + "eval_loss": 1.0524089336395264, + "eval_runtime": 51.3652, + "eval_samples_per_second": 19.468, + "eval_steps_per_second": 1.227, + "step": 490 + }, + { + "epoch": 0.7105643994211288, + "grad_norm": 1.5771195888519287, + "learning_rate": 3.6524537409493165e-06, + "loss": 0.9768, + "step": 491 + }, + { + "epoch": 0.7120115774240231, + "grad_norm": 1.5392457246780396, + "learning_rate": 3.6484312148028966e-06, + "loss": 0.9991, + "step": 492 + }, + { + "epoch": 0.7134587554269175, + "grad_norm": 1.4841896295547485, + "learning_rate": 3.6444086886564768e-06, + "loss": 0.9724, + "step": 493 + }, + { + "epoch": 0.7149059334298119, + "grad_norm": 1.3620243072509766, + "learning_rate": 3.640386162510057e-06, + "loss": 0.9496, + "step": 494 + }, + { + "epoch": 0.7163531114327062, + "grad_norm": 1.4008426666259766, + "learning_rate": 3.6363636363636366e-06, + "loss": 0.9393, + "step": 495 + }, + { + "epoch": 0.7178002894356006, + "grad_norm": 1.6012742519378662, + "learning_rate": 3.6323411102172167e-06, + "loss": 0.9233, + "step": 496 + }, + { + "epoch": 0.7192474674384949, + "grad_norm": 1.3270018100738525, + "learning_rate": 3.628318584070797e-06, + "loss": 0.9255, + "step": 497 + }, + { + "epoch": 0.7206946454413893, + "grad_norm": 1.0990227460861206, + "learning_rate": 3.624296057924377e-06, + "loss": 0.9305, + "step": 498 + }, + { + "epoch": 0.7221418234442837, + "grad_norm": 1.4646888971328735, + "learning_rate": 3.620273531777957e-06, + "loss": 0.9713, + "step": 499 + }, + { + "epoch": 0.723589001447178, + "grad_norm": 1.4900317192077637, + "learning_rate": 3.616251005631537e-06, + "loss": 0.9446, + "step": 500 + }, + { + "epoch": 0.723589001447178, + "eval_loss": 1.0490972995758057, + "eval_runtime": 51.3055, + "eval_samples_per_second": 19.491, + "eval_steps_per_second": 1.228, + "step": 500 + }, + { + "epoch": 0.7250361794500724, + "grad_norm": 1.495969295501709, + "learning_rate": 3.612228479485117e-06, + "loss": 0.9303, + "step": 501 + }, + { + "epoch": 0.7264833574529667, + "grad_norm": 1.4521697759628296, + "learning_rate": 3.608205953338697e-06, + "loss": 0.9736, + "step": 502 + }, + { + "epoch": 0.7279305354558611, + "grad_norm": 1.3677924871444702, + "learning_rate": 3.604183427192277e-06, + "loss": 0.9483, + "step": 503 + }, + { + "epoch": 0.7293777134587555, + "grad_norm": 1.379379391670227, + "learning_rate": 3.6001609010458573e-06, + "loss": 0.96, + "step": 504 + }, + { + "epoch": 0.7308248914616498, + "grad_norm": 1.4630364179611206, + "learning_rate": 3.596138374899437e-06, + "loss": 0.9519, + "step": 505 + }, + { + "epoch": 0.7322720694645442, + "grad_norm": 1.492518424987793, + "learning_rate": 3.592115848753017e-06, + "loss": 0.9308, + "step": 506 + }, + { + "epoch": 0.7337192474674384, + "grad_norm": 1.4050403833389282, + "learning_rate": 3.5880933226065973e-06, + "loss": 0.9358, + "step": 507 + }, + { + "epoch": 0.7351664254703328, + "grad_norm": 1.3876930475234985, + "learning_rate": 3.5840707964601774e-06, + "loss": 0.9401, + "step": 508 + }, + { + "epoch": 0.7366136034732272, + "grad_norm": 1.792354702949524, + "learning_rate": 3.5800482703137575e-06, + "loss": 0.903, + "step": 509 + }, + { + "epoch": 0.7380607814761215, + "grad_norm": 1.7027020454406738, + "learning_rate": 3.5760257441673372e-06, + "loss": 0.9686, + "step": 510 + }, + { + "epoch": 0.7380607814761215, + "eval_loss": 1.0471789836883545, + "eval_runtime": 51.3169, + "eval_samples_per_second": 19.487, + "eval_steps_per_second": 1.228, + "step": 510 + }, + { + "epoch": 0.7395079594790159, + "grad_norm": 1.2350125312805176, + "learning_rate": 3.5720032180209174e-06, + "loss": 0.954, + "step": 511 + }, + { + "epoch": 0.7409551374819102, + "grad_norm": 1.1130231618881226, + "learning_rate": 3.5679806918744975e-06, + "loss": 0.9438, + "step": 512 + }, + { + "epoch": 0.7424023154848046, + "grad_norm": 1.4017415046691895, + "learning_rate": 3.5639581657280776e-06, + "loss": 0.9737, + "step": 513 + }, + { + "epoch": 0.743849493487699, + "grad_norm": 1.5187294483184814, + "learning_rate": 3.5599356395816577e-06, + "loss": 0.9416, + "step": 514 + }, + { + "epoch": 0.7452966714905933, + "grad_norm": 1.0262166261672974, + "learning_rate": 3.5559131134352374e-06, + "loss": 0.8832, + "step": 515 + }, + { + "epoch": 0.7467438494934877, + "grad_norm": 1.5807908773422241, + "learning_rate": 3.5518905872888176e-06, + "loss": 0.94, + "step": 516 + }, + { + "epoch": 0.748191027496382, + "grad_norm": 1.2302393913269043, + "learning_rate": 3.5478680611423977e-06, + "loss": 0.9787, + "step": 517 + }, + { + "epoch": 0.7496382054992764, + "grad_norm": 1.5715115070343018, + "learning_rate": 3.543845534995978e-06, + "loss": 0.9851, + "step": 518 + }, + { + "epoch": 0.7510853835021708, + "grad_norm": 1.4873887300491333, + "learning_rate": 3.539823008849558e-06, + "loss": 0.9165, + "step": 519 + }, + { + "epoch": 0.7525325615050651, + "grad_norm": 1.1715465784072876, + "learning_rate": 3.5358004827031377e-06, + "loss": 0.9355, + "step": 520 + }, + { + "epoch": 0.7525325615050651, + "eval_loss": 1.0428709983825684, + "eval_runtime": 51.4634, + "eval_samples_per_second": 19.431, + "eval_steps_per_second": 1.224, + "step": 520 + }, + { + "epoch": 0.7539797395079595, + "grad_norm": 1.5332242250442505, + "learning_rate": 3.5317779565567178e-06, + "loss": 0.9679, + "step": 521 + }, + { + "epoch": 0.7554269175108539, + "grad_norm": 1.6401145458221436, + "learning_rate": 3.527755430410298e-06, + "loss": 0.9243, + "step": 522 + }, + { + "epoch": 0.7568740955137482, + "grad_norm": 1.6344199180603027, + "learning_rate": 3.523732904263878e-06, + "loss": 0.9722, + "step": 523 + }, + { + "epoch": 0.7583212735166426, + "grad_norm": 1.3958275318145752, + "learning_rate": 3.519710378117458e-06, + "loss": 0.9471, + "step": 524 + }, + { + "epoch": 0.7597684515195369, + "grad_norm": 1.555349588394165, + "learning_rate": 3.515687851971038e-06, + "loss": 0.9448, + "step": 525 + }, + { + "epoch": 0.7612156295224313, + "grad_norm": 1.4171674251556396, + "learning_rate": 3.511665325824618e-06, + "loss": 0.9532, + "step": 526 + }, + { + "epoch": 0.7626628075253257, + "grad_norm": 1.2396488189697266, + "learning_rate": 3.507642799678198e-06, + "loss": 0.9638, + "step": 527 + }, + { + "epoch": 0.76410998552822, + "grad_norm": 1.359450340270996, + "learning_rate": 3.5036202735317783e-06, + "loss": 0.947, + "step": 528 + }, + { + "epoch": 0.7655571635311144, + "grad_norm": 1.3703235387802124, + "learning_rate": 3.4995977473853584e-06, + "loss": 0.9478, + "step": 529 + }, + { + "epoch": 0.7670043415340086, + "grad_norm": 1.7447525262832642, + "learning_rate": 3.495575221238938e-06, + "loss": 0.9548, + "step": 530 + }, + { + "epoch": 0.7670043415340086, + "eval_loss": 1.045773983001709, + "eval_runtime": 51.3493, + "eval_samples_per_second": 19.474, + "eval_steps_per_second": 1.227, + "step": 530 + }, + { + "epoch": 0.768451519536903, + "grad_norm": 1.1792054176330566, + "learning_rate": 3.4915526950925182e-06, + "loss": 0.9352, + "step": 531 + }, + { + "epoch": 0.7698986975397974, + "grad_norm": 1.4619675874710083, + "learning_rate": 3.4875301689460983e-06, + "loss": 0.9285, + "step": 532 + }, + { + "epoch": 0.7713458755426917, + "grad_norm": 1.2957642078399658, + "learning_rate": 3.4835076427996785e-06, + "loss": 0.9564, + "step": 533 + }, + { + "epoch": 0.7727930535455861, + "grad_norm": 1.3679780960083008, + "learning_rate": 3.4794851166532586e-06, + "loss": 0.9443, + "step": 534 + }, + { + "epoch": 0.7742402315484804, + "grad_norm": 1.1401643753051758, + "learning_rate": 3.4754625905068383e-06, + "loss": 0.9612, + "step": 535 + }, + { + "epoch": 0.7756874095513748, + "grad_norm": 1.5207276344299316, + "learning_rate": 3.4714400643604184e-06, + "loss": 0.9553, + "step": 536 + }, + { + "epoch": 0.7771345875542692, + "grad_norm": 1.4308780431747437, + "learning_rate": 3.4674175382139986e-06, + "loss": 0.9424, + "step": 537 + }, + { + "epoch": 0.7785817655571635, + "grad_norm": 1.5424914360046387, + "learning_rate": 3.4633950120675787e-06, + "loss": 0.9513, + "step": 538 + }, + { + "epoch": 0.7800289435600579, + "grad_norm": 1.1681467294692993, + "learning_rate": 3.4593724859211584e-06, + "loss": 0.9348, + "step": 539 + }, + { + "epoch": 0.7814761215629522, + "grad_norm": 1.4271103143692017, + "learning_rate": 3.4553499597747385e-06, + "loss": 0.9607, + "step": 540 + }, + { + "epoch": 0.7814761215629522, + "eval_loss": 1.044892430305481, + "eval_runtime": 51.2445, + "eval_samples_per_second": 19.514, + "eval_steps_per_second": 1.229, + "step": 540 + }, + { + "epoch": 0.7829232995658466, + "grad_norm": 1.4043258428573608, + "learning_rate": 3.4513274336283186e-06, + "loss": 0.9704, + "step": 541 + }, + { + "epoch": 0.784370477568741, + "grad_norm": 1.3699084520339966, + "learning_rate": 3.4473049074818988e-06, + "loss": 0.9479, + "step": 542 + }, + { + "epoch": 0.7858176555716353, + "grad_norm": 1.610023021697998, + "learning_rate": 3.443282381335479e-06, + "loss": 0.9381, + "step": 543 + }, + { + "epoch": 0.7872648335745297, + "grad_norm": 1.2331242561340332, + "learning_rate": 3.4392598551890586e-06, + "loss": 0.9572, + "step": 544 + }, + { + "epoch": 0.788712011577424, + "grad_norm": 1.2764670848846436, + "learning_rate": 3.4352373290426387e-06, + "loss": 0.9177, + "step": 545 + }, + { + "epoch": 0.7901591895803184, + "grad_norm": 1.3424965143203735, + "learning_rate": 3.431214802896219e-06, + "loss": 0.9609, + "step": 546 + }, + { + "epoch": 0.7916063675832128, + "grad_norm": 1.208187460899353, + "learning_rate": 3.427192276749799e-06, + "loss": 0.9455, + "step": 547 + }, + { + "epoch": 0.7930535455861071, + "grad_norm": 1.4419608116149902, + "learning_rate": 3.423169750603379e-06, + "loss": 0.9384, + "step": 548 + }, + { + "epoch": 0.7945007235890015, + "grad_norm": 1.4868179559707642, + "learning_rate": 3.419147224456959e-06, + "loss": 0.9657, + "step": 549 + }, + { + "epoch": 0.7959479015918958, + "grad_norm": 1.2070798873901367, + "learning_rate": 3.415124698310539e-06, + "loss": 0.9485, + "step": 550 + }, + { + "epoch": 0.7959479015918958, + "eval_loss": 1.0407907962799072, + "eval_runtime": 51.5302, + "eval_samples_per_second": 19.406, + "eval_steps_per_second": 1.223, + "step": 550 + }, + { + "epoch": 0.7973950795947902, + "grad_norm": 1.4873172044754028, + "learning_rate": 3.411102172164119e-06, + "loss": 0.9207, + "step": 551 + }, + { + "epoch": 0.7988422575976846, + "grad_norm": 1.2002183198928833, + "learning_rate": 3.407079646017699e-06, + "loss": 0.9714, + "step": 552 + }, + { + "epoch": 0.8002894356005789, + "grad_norm": 1.4901013374328613, + "learning_rate": 3.4030571198712793e-06, + "loss": 0.9359, + "step": 553 + }, + { + "epoch": 0.8017366136034733, + "grad_norm": 1.2517327070236206, + "learning_rate": 3.399034593724859e-06, + "loss": 0.9249, + "step": 554 + }, + { + "epoch": 0.8031837916063675, + "grad_norm": 1.2243083715438843, + "learning_rate": 3.395012067578439e-06, + "loss": 0.9192, + "step": 555 + }, + { + "epoch": 0.804630969609262, + "grad_norm": 1.4090490341186523, + "learning_rate": 3.3909895414320197e-06, + "loss": 0.9259, + "step": 556 + }, + { + "epoch": 0.8060781476121563, + "grad_norm": 1.2865501642227173, + "learning_rate": 3.3869670152856e-06, + "loss": 0.9251, + "step": 557 + }, + { + "epoch": 0.8075253256150506, + "grad_norm": 1.44110107421875, + "learning_rate": 3.38294448913918e-06, + "loss": 0.9311, + "step": 558 + }, + { + "epoch": 0.808972503617945, + "grad_norm": 1.1358684301376343, + "learning_rate": 3.37892196299276e-06, + "loss": 0.9794, + "step": 559 + }, + { + "epoch": 0.8104196816208393, + "grad_norm": 1.2640002965927124, + "learning_rate": 3.3748994368463402e-06, + "loss": 0.9531, + "step": 560 + }, + { + "epoch": 0.8104196816208393, + "eval_loss": 1.0371347665786743, + "eval_runtime": 51.3449, + "eval_samples_per_second": 19.476, + "eval_steps_per_second": 1.227, + "step": 560 + }, + { + "epoch": 0.8118668596237337, + "grad_norm": 1.5194166898727417, + "learning_rate": 3.37087691069992e-06, + "loss": 0.967, + "step": 561 + }, + { + "epoch": 0.8133140376266281, + "grad_norm": 1.3474825620651245, + "learning_rate": 3.3668543845535e-06, + "loss": 0.926, + "step": 562 + }, + { + "epoch": 0.8147612156295224, + "grad_norm": 1.5002163648605347, + "learning_rate": 3.36283185840708e-06, + "loss": 0.9371, + "step": 563 + }, + { + "epoch": 0.8162083936324168, + "grad_norm": 1.5828827619552612, + "learning_rate": 3.3588093322606603e-06, + "loss": 0.9405, + "step": 564 + }, + { + "epoch": 0.8176555716353111, + "grad_norm": 1.6439111232757568, + "learning_rate": 3.3547868061142404e-06, + "loss": 0.9386, + "step": 565 + }, + { + "epoch": 0.8191027496382055, + "grad_norm": 1.353527545928955, + "learning_rate": 3.35076427996782e-06, + "loss": 0.9271, + "step": 566 + }, + { + "epoch": 0.8205499276410999, + "grad_norm": 1.3500126600265503, + "learning_rate": 3.3467417538214003e-06, + "loss": 0.9317, + "step": 567 + }, + { + "epoch": 0.8219971056439942, + "grad_norm": 1.5200532674789429, + "learning_rate": 3.3427192276749804e-06, + "loss": 0.9398, + "step": 568 + }, + { + "epoch": 0.8234442836468886, + "grad_norm": 1.4447791576385498, + "learning_rate": 3.3386967015285605e-06, + "loss": 0.9573, + "step": 569 + }, + { + "epoch": 0.8248914616497829, + "grad_norm": 1.2587987184524536, + "learning_rate": 3.3346741753821407e-06, + "loss": 0.9587, + "step": 570 + }, + { + "epoch": 0.8248914616497829, + "eval_loss": 1.0374418497085571, + "eval_runtime": 51.653, + "eval_samples_per_second": 19.36, + "eval_steps_per_second": 1.22, + "step": 570 + }, + { + "epoch": 0.8263386396526773, + "grad_norm": 1.2351385354995728, + "learning_rate": 3.3306516492357204e-06, + "loss": 0.9528, + "step": 571 + }, + { + "epoch": 0.8277858176555717, + "grad_norm": 1.4653207063674927, + "learning_rate": 3.3266291230893005e-06, + "loss": 0.9933, + "step": 572 + }, + { + "epoch": 0.829232995658466, + "grad_norm": 1.5111782550811768, + "learning_rate": 3.3226065969428806e-06, + "loss": 0.936, + "step": 573 + }, + { + "epoch": 0.8306801736613604, + "grad_norm": 1.4335622787475586, + "learning_rate": 3.3185840707964607e-06, + "loss": 0.9449, + "step": 574 + }, + { + "epoch": 0.8321273516642547, + "grad_norm": 1.1462665796279907, + "learning_rate": 3.3145615446500404e-06, + "loss": 0.9183, + "step": 575 + }, + { + "epoch": 0.8335745296671491, + "grad_norm": 1.46926748752594, + "learning_rate": 3.3105390185036206e-06, + "loss": 0.9375, + "step": 576 + }, + { + "epoch": 0.8350217076700435, + "grad_norm": 1.273015022277832, + "learning_rate": 3.3065164923572007e-06, + "loss": 0.8901, + "step": 577 + }, + { + "epoch": 0.8364688856729378, + "grad_norm": 1.425917148590088, + "learning_rate": 3.302493966210781e-06, + "loss": 0.9279, + "step": 578 + }, + { + "epoch": 0.8379160636758322, + "grad_norm": 1.2700085639953613, + "learning_rate": 3.298471440064361e-06, + "loss": 0.9383, + "step": 579 + }, + { + "epoch": 0.8393632416787264, + "grad_norm": 1.2181962728500366, + "learning_rate": 3.2944489139179407e-06, + "loss": 0.9624, + "step": 580 + }, + { + "epoch": 0.8393632416787264, + "eval_loss": 1.039432168006897, + "eval_runtime": 51.5819, + "eval_samples_per_second": 19.387, + "eval_steps_per_second": 1.221, + "step": 580 + }, + { + "epoch": 0.8408104196816208, + "grad_norm": 1.769498348236084, + "learning_rate": 3.2904263877715208e-06, + "loss": 0.9183, + "step": 581 + }, + { + "epoch": 0.8422575976845152, + "grad_norm": 1.5170090198516846, + "learning_rate": 3.286403861625101e-06, + "loss": 0.9127, + "step": 582 + }, + { + "epoch": 0.8437047756874095, + "grad_norm": 1.3647818565368652, + "learning_rate": 3.282381335478681e-06, + "loss": 0.9591, + "step": 583 + }, + { + "epoch": 0.8451519536903039, + "grad_norm": 1.3169649839401245, + "learning_rate": 3.278358809332261e-06, + "loss": 0.9695, + "step": 584 + }, + { + "epoch": 0.8465991316931982, + "grad_norm": 1.4267555475234985, + "learning_rate": 3.274336283185841e-06, + "loss": 0.9189, + "step": 585 + }, + { + "epoch": 0.8480463096960926, + "grad_norm": 1.1928932666778564, + "learning_rate": 3.270313757039421e-06, + "loss": 0.9482, + "step": 586 + }, + { + "epoch": 0.849493487698987, + "grad_norm": 1.4360361099243164, + "learning_rate": 3.266291230893001e-06, + "loss": 0.9245, + "step": 587 + }, + { + "epoch": 0.8509406657018813, + "grad_norm": 1.7216849327087402, + "learning_rate": 3.2622687047465813e-06, + "loss": 0.9572, + "step": 588 + }, + { + "epoch": 0.8523878437047757, + "grad_norm": 1.4572755098342896, + "learning_rate": 3.2582461786001614e-06, + "loss": 0.9533, + "step": 589 + }, + { + "epoch": 0.85383502170767, + "grad_norm": 1.2555478811264038, + "learning_rate": 3.254223652453741e-06, + "loss": 0.9707, + "step": 590 + }, + { + "epoch": 0.85383502170767, + "eval_loss": 1.0382187366485596, + "eval_runtime": 51.6704, + "eval_samples_per_second": 19.353, + "eval_steps_per_second": 1.219, + "step": 590 + }, + { + "epoch": 0.8552821997105644, + "grad_norm": 1.3232746124267578, + "learning_rate": 3.2502011263073212e-06, + "loss": 0.9313, + "step": 591 + }, + { + "epoch": 0.8567293777134588, + "grad_norm": 1.3387691974639893, + "learning_rate": 3.2461786001609013e-06, + "loss": 0.9365, + "step": 592 + }, + { + "epoch": 0.8581765557163531, + "grad_norm": 1.468837857246399, + "learning_rate": 3.2421560740144815e-06, + "loss": 0.9615, + "step": 593 + }, + { + "epoch": 0.8596237337192475, + "grad_norm": 1.5204793214797974, + "learning_rate": 3.2381335478680616e-06, + "loss": 0.9134, + "step": 594 + }, + { + "epoch": 0.8610709117221418, + "grad_norm": 1.4535566568374634, + "learning_rate": 3.2341110217216413e-06, + "loss": 0.9232, + "step": 595 + }, + { + "epoch": 0.8625180897250362, + "grad_norm": 1.228965163230896, + "learning_rate": 3.2300884955752214e-06, + "loss": 0.9388, + "step": 596 + }, + { + "epoch": 0.8639652677279306, + "grad_norm": 1.3360660076141357, + "learning_rate": 3.2260659694288016e-06, + "loss": 0.9435, + "step": 597 + }, + { + "epoch": 0.8654124457308249, + "grad_norm": 1.2892252206802368, + "learning_rate": 3.2220434432823817e-06, + "loss": 0.9415, + "step": 598 + }, + { + "epoch": 0.8668596237337193, + "grad_norm": 2.1330487728118896, + "learning_rate": 3.218020917135962e-06, + "loss": 0.9222, + "step": 599 + }, + { + "epoch": 0.8683068017366136, + "grad_norm": 1.3030391931533813, + "learning_rate": 3.2139983909895415e-06, + "loss": 0.934, + "step": 600 + }, + { + "epoch": 0.8683068017366136, + "eval_loss": 1.0347357988357544, + "eval_runtime": 51.4839, + "eval_samples_per_second": 19.424, + "eval_steps_per_second": 1.224, + "step": 600 + }, + { + "epoch": 0.869753979739508, + "grad_norm": 1.4273481369018555, + "learning_rate": 3.2099758648431216e-06, + "loss": 0.9482, + "step": 601 + }, + { + "epoch": 0.8712011577424024, + "grad_norm": 1.436102271080017, + "learning_rate": 3.2059533386967018e-06, + "loss": 0.9113, + "step": 602 + }, + { + "epoch": 0.8726483357452967, + "grad_norm": 1.3415288925170898, + "learning_rate": 3.201930812550282e-06, + "loss": 0.9447, + "step": 603 + }, + { + "epoch": 0.874095513748191, + "grad_norm": 1.558142900466919, + "learning_rate": 3.197908286403862e-06, + "loss": 0.9226, + "step": 604 + }, + { + "epoch": 0.8755426917510853, + "grad_norm": 1.6435316801071167, + "learning_rate": 3.1938857602574417e-06, + "loss": 0.931, + "step": 605 + }, + { + "epoch": 0.8769898697539797, + "grad_norm": 1.5288655757904053, + "learning_rate": 3.189863234111022e-06, + "loss": 0.9244, + "step": 606 + }, + { + "epoch": 0.8784370477568741, + "grad_norm": 1.5140902996063232, + "learning_rate": 3.185840707964602e-06, + "loss": 0.8964, + "step": 607 + }, + { + "epoch": 0.8798842257597684, + "grad_norm": 1.2200645208358765, + "learning_rate": 3.181818181818182e-06, + "loss": 0.9112, + "step": 608 + }, + { + "epoch": 0.8813314037626628, + "grad_norm": 1.351054310798645, + "learning_rate": 3.1777956556717622e-06, + "loss": 0.9304, + "step": 609 + }, + { + "epoch": 0.8827785817655571, + "grad_norm": 1.2503132820129395, + "learning_rate": 3.173773129525342e-06, + "loss": 0.8902, + "step": 610 + }, + { + "epoch": 0.8827785817655571, + "eval_loss": 1.0326288938522339, + "eval_runtime": 51.0614, + "eval_samples_per_second": 19.584, + "eval_steps_per_second": 1.234, + "step": 610 + }, + { + "epoch": 0.8842257597684515, + "grad_norm": 1.5399298667907715, + "learning_rate": 3.169750603378922e-06, + "loss": 0.9271, + "step": 611 + }, + { + "epoch": 0.8856729377713459, + "grad_norm": 1.2259747982025146, + "learning_rate": 3.165728077232502e-06, + "loss": 0.9298, + "step": 612 + }, + { + "epoch": 0.8871201157742402, + "grad_norm": 1.3675987720489502, + "learning_rate": 3.1617055510860823e-06, + "loss": 0.9468, + "step": 613 + }, + { + "epoch": 0.8885672937771346, + "grad_norm": 1.186338186264038, + "learning_rate": 3.1576830249396625e-06, + "loss": 0.9016, + "step": 614 + }, + { + "epoch": 0.8900144717800289, + "grad_norm": 1.6312668323516846, + "learning_rate": 3.153660498793242e-06, + "loss": 0.9248, + "step": 615 + }, + { + "epoch": 0.8914616497829233, + "grad_norm": 1.3140789270401, + "learning_rate": 3.1496379726468223e-06, + "loss": 0.9425, + "step": 616 + }, + { + "epoch": 0.8929088277858177, + "grad_norm": 1.4555140733718872, + "learning_rate": 3.1456154465004024e-06, + "loss": 0.9309, + "step": 617 + }, + { + "epoch": 0.894356005788712, + "grad_norm": 1.4377708435058594, + "learning_rate": 3.1415929203539825e-06, + "loss": 0.9309, + "step": 618 + }, + { + "epoch": 0.8958031837916064, + "grad_norm": 1.6962463855743408, + "learning_rate": 3.1375703942075627e-06, + "loss": 0.9499, + "step": 619 + }, + { + "epoch": 0.8972503617945007, + "grad_norm": 1.630610466003418, + "learning_rate": 3.1335478680611424e-06, + "loss": 0.9546, + "step": 620 + }, + { + "epoch": 0.8972503617945007, + "eval_loss": 1.031374216079712, + "eval_runtime": 51.1344, + "eval_samples_per_second": 19.556, + "eval_steps_per_second": 1.232, + "step": 620 + }, + { + "epoch": 0.8986975397973951, + "grad_norm": 1.416258692741394, + "learning_rate": 3.1295253419147225e-06, + "loss": 0.9578, + "step": 621 + }, + { + "epoch": 0.9001447178002895, + "grad_norm": 1.5512739419937134, + "learning_rate": 3.1255028157683026e-06, + "loss": 0.9447, + "step": 622 + }, + { + "epoch": 0.9015918958031838, + "grad_norm": 1.2793455123901367, + "learning_rate": 3.1214802896218828e-06, + "loss": 0.9634, + "step": 623 + }, + { + "epoch": 0.9030390738060782, + "grad_norm": 1.2623417377471924, + "learning_rate": 3.117457763475463e-06, + "loss": 0.9194, + "step": 624 + }, + { + "epoch": 0.9044862518089725, + "grad_norm": 1.3823516368865967, + "learning_rate": 3.1134352373290426e-06, + "loss": 0.9044, + "step": 625 + }, + { + "epoch": 0.9059334298118669, + "grad_norm": 1.529886245727539, + "learning_rate": 3.1094127111826227e-06, + "loss": 0.9419, + "step": 626 + }, + { + "epoch": 0.9073806078147613, + "grad_norm": 1.3831008672714233, + "learning_rate": 3.105390185036203e-06, + "loss": 0.9629, + "step": 627 + }, + { + "epoch": 0.9088277858176556, + "grad_norm": 1.3207403421401978, + "learning_rate": 3.101367658889783e-06, + "loss": 0.9202, + "step": 628 + }, + { + "epoch": 0.91027496382055, + "grad_norm": 1.626394510269165, + "learning_rate": 3.097345132743363e-06, + "loss": 0.9343, + "step": 629 + }, + { + "epoch": 0.9117221418234442, + "grad_norm": 1.278717279434204, + "learning_rate": 3.093322606596943e-06, + "loss": 0.8809, + "step": 630 + }, + { + "epoch": 0.9117221418234442, + "eval_loss": 1.0292832851409912, + "eval_runtime": 51.0228, + "eval_samples_per_second": 19.599, + "eval_steps_per_second": 1.235, + "step": 630 + }, + { + "epoch": 0.9131693198263386, + "grad_norm": 1.3757354021072388, + "learning_rate": 3.089300080450523e-06, + "loss": 0.9657, + "step": 631 + }, + { + "epoch": 0.914616497829233, + "grad_norm": 1.2634453773498535, + "learning_rate": 3.085277554304103e-06, + "loss": 0.9464, + "step": 632 + }, + { + "epoch": 0.9160636758321273, + "grad_norm": 1.2393821477890015, + "learning_rate": 3.081255028157683e-06, + "loss": 0.9318, + "step": 633 + }, + { + "epoch": 0.9175108538350217, + "grad_norm": 1.5298211574554443, + "learning_rate": 3.0772325020112633e-06, + "loss": 0.9147, + "step": 634 + }, + { + "epoch": 0.918958031837916, + "grad_norm": 1.2504903078079224, + "learning_rate": 3.073209975864843e-06, + "loss": 0.9311, + "step": 635 + }, + { + "epoch": 0.9204052098408104, + "grad_norm": 1.2467671632766724, + "learning_rate": 3.069187449718423e-06, + "loss": 0.9233, + "step": 636 + }, + { + "epoch": 0.9218523878437048, + "grad_norm": 1.3215663433074951, + "learning_rate": 3.0651649235720033e-06, + "loss": 0.9209, + "step": 637 + }, + { + "epoch": 0.9232995658465991, + "grad_norm": 1.2857428789138794, + "learning_rate": 3.0611423974255834e-06, + "loss": 0.9831, + "step": 638 + }, + { + "epoch": 0.9247467438494935, + "grad_norm": 1.3851327896118164, + "learning_rate": 3.057119871279163e-06, + "loss": 0.9508, + "step": 639 + }, + { + "epoch": 0.9261939218523878, + "grad_norm": 1.480155110359192, + "learning_rate": 3.0530973451327432e-06, + "loss": 0.9438, + "step": 640 + }, + { + "epoch": 0.9261939218523878, + "eval_loss": 1.032254934310913, + "eval_runtime": 50.9549, + "eval_samples_per_second": 19.625, + "eval_steps_per_second": 1.236, + "step": 640 + }, + { + "epoch": 0.9276410998552822, + "grad_norm": 1.6129275560379028, + "learning_rate": 3.0490748189863234e-06, + "loss": 0.9202, + "step": 641 + }, + { + "epoch": 0.9290882778581766, + "grad_norm": 1.194135308265686, + "learning_rate": 3.0450522928399035e-06, + "loss": 0.9453, + "step": 642 + }, + { + "epoch": 0.9305354558610709, + "grad_norm": 1.219870924949646, + "learning_rate": 3.0410297666934836e-06, + "loss": 0.9376, + "step": 643 + }, + { + "epoch": 0.9319826338639653, + "grad_norm": 1.5214959383010864, + "learning_rate": 3.0370072405470633e-06, + "loss": 0.9255, + "step": 644 + }, + { + "epoch": 0.9334298118668596, + "grad_norm": 1.5212085247039795, + "learning_rate": 3.0329847144006443e-06, + "loss": 0.9327, + "step": 645 + }, + { + "epoch": 0.934876989869754, + "grad_norm": 1.4605064392089844, + "learning_rate": 3.028962188254224e-06, + "loss": 0.9072, + "step": 646 + }, + { + "epoch": 0.9363241678726484, + "grad_norm": 1.2423738241195679, + "learning_rate": 3.024939662107804e-06, + "loss": 0.928, + "step": 647 + }, + { + "epoch": 0.9377713458755427, + "grad_norm": 1.3490147590637207, + "learning_rate": 3.0209171359613843e-06, + "loss": 0.9151, + "step": 648 + }, + { + "epoch": 0.9392185238784371, + "grad_norm": 1.2052751779556274, + "learning_rate": 3.0168946098149644e-06, + "loss": 0.9486, + "step": 649 + }, + { + "epoch": 0.9406657018813314, + "grad_norm": 1.5702952146530151, + "learning_rate": 3.0128720836685445e-06, + "loss": 0.9189, + "step": 650 + }, + { + "epoch": 0.9406657018813314, + "eval_loss": 1.0342165231704712, + "eval_runtime": 50.6661, + "eval_samples_per_second": 19.737, + "eval_steps_per_second": 1.243, + "step": 650 + }, + { + "epoch": 0.9421128798842258, + "grad_norm": 1.4646233320236206, + "learning_rate": 3.0088495575221242e-06, + "loss": 0.9234, + "step": 651 + }, + { + "epoch": 0.9435600578871202, + "grad_norm": 1.3959081172943115, + "learning_rate": 3.0048270313757043e-06, + "loss": 0.9118, + "step": 652 + }, + { + "epoch": 0.9450072358900145, + "grad_norm": 1.226422905921936, + "learning_rate": 3.0008045052292845e-06, + "loss": 0.926, + "step": 653 + }, + { + "epoch": 0.9464544138929089, + "grad_norm": 1.1747907400131226, + "learning_rate": 2.9967819790828646e-06, + "loss": 0.9309, + "step": 654 + }, + { + "epoch": 0.9479015918958031, + "grad_norm": 1.7044570446014404, + "learning_rate": 2.9927594529364447e-06, + "loss": 0.9627, + "step": 655 + }, + { + "epoch": 0.9493487698986975, + "grad_norm": 1.4969197511672974, + "learning_rate": 2.9887369267900244e-06, + "loss": 0.962, + "step": 656 + }, + { + "epoch": 0.9507959479015919, + "grad_norm": 1.2375552654266357, + "learning_rate": 2.9847144006436046e-06, + "loss": 0.9144, + "step": 657 + }, + { + "epoch": 0.9522431259044862, + "grad_norm": 1.6275029182434082, + "learning_rate": 2.9806918744971847e-06, + "loss": 0.9073, + "step": 658 + }, + { + "epoch": 0.9536903039073806, + "grad_norm": 1.4387280941009521, + "learning_rate": 2.976669348350765e-06, + "loss": 0.9276, + "step": 659 + }, + { + "epoch": 0.9551374819102749, + "grad_norm": 1.4287992715835571, + "learning_rate": 2.972646822204345e-06, + "loss": 0.9149, + "step": 660 + }, + { + "epoch": 0.9551374819102749, + "eval_loss": 1.034416675567627, + "eval_runtime": 51.0425, + "eval_samples_per_second": 19.592, + "eval_steps_per_second": 1.234, + "step": 660 + }, + { + "epoch": 0.9565846599131693, + "grad_norm": 1.5335744619369507, + "learning_rate": 2.9686242960579246e-06, + "loss": 0.9543, + "step": 661 + }, + { + "epoch": 0.9580318379160637, + "grad_norm": 1.47525155544281, + "learning_rate": 2.9646017699115048e-06, + "loss": 0.9174, + "step": 662 + }, + { + "epoch": 0.959479015918958, + "grad_norm": 1.616828203201294, + "learning_rate": 2.960579243765085e-06, + "loss": 0.9237, + "step": 663 + }, + { + "epoch": 0.9609261939218524, + "grad_norm": 1.5514025688171387, + "learning_rate": 2.956556717618665e-06, + "loss": 0.9455, + "step": 664 + }, + { + "epoch": 0.9623733719247467, + "grad_norm": 1.3954055309295654, + "learning_rate": 2.952534191472245e-06, + "loss": 0.8665, + "step": 665 + }, + { + "epoch": 0.9638205499276411, + "grad_norm": 1.1854369640350342, + "learning_rate": 2.948511665325825e-06, + "loss": 0.9465, + "step": 666 + }, + { + "epoch": 0.9652677279305355, + "grad_norm": 1.1244102716445923, + "learning_rate": 2.944489139179405e-06, + "loss": 0.9264, + "step": 667 + }, + { + "epoch": 0.9667149059334298, + "grad_norm": 1.5507874488830566, + "learning_rate": 2.940466613032985e-06, + "loss": 0.9499, + "step": 668 + }, + { + "epoch": 0.9681620839363242, + "grad_norm": 1.6144473552703857, + "learning_rate": 2.9364440868865652e-06, + "loss": 0.9268, + "step": 669 + }, + { + "epoch": 0.9696092619392185, + "grad_norm": 1.2265530824661255, + "learning_rate": 2.9324215607401454e-06, + "loss": 0.8797, + "step": 670 + }, + { + "epoch": 0.9696092619392185, + "eval_loss": 1.028491735458374, + "eval_runtime": 50.8503, + "eval_samples_per_second": 19.666, + "eval_steps_per_second": 1.239, + "step": 670 + }, + { + "epoch": 0.9710564399421129, + "grad_norm": 1.5372741222381592, + "learning_rate": 2.928399034593725e-06, + "loss": 0.9284, + "step": 671 + }, + { + "epoch": 0.9725036179450073, + "grad_norm": 1.4664175510406494, + "learning_rate": 2.924376508447305e-06, + "loss": 0.9036, + "step": 672 + }, + { + "epoch": 0.9739507959479016, + "grad_norm": 1.3370870351791382, + "learning_rate": 2.9203539823008853e-06, + "loss": 0.9288, + "step": 673 + }, + { + "epoch": 0.975397973950796, + "grad_norm": 1.3292039632797241, + "learning_rate": 2.9163314561544655e-06, + "loss": 0.8878, + "step": 674 + }, + { + "epoch": 0.9768451519536903, + "grad_norm": 1.2751848697662354, + "learning_rate": 2.912308930008045e-06, + "loss": 0.9098, + "step": 675 + }, + { + "epoch": 0.9782923299565847, + "grad_norm": 1.3417860269546509, + "learning_rate": 2.9082864038616253e-06, + "loss": 0.9335, + "step": 676 + }, + { + "epoch": 0.9797395079594791, + "grad_norm": 1.3917615413665771, + "learning_rate": 2.9042638777152054e-06, + "loss": 0.88, + "step": 677 + }, + { + "epoch": 0.9811866859623734, + "grad_norm": 1.2170779705047607, + "learning_rate": 2.9002413515687855e-06, + "loss": 0.9364, + "step": 678 + }, + { + "epoch": 0.9826338639652678, + "grad_norm": 1.2658751010894775, + "learning_rate": 2.8962188254223657e-06, + "loss": 0.9349, + "step": 679 + }, + { + "epoch": 0.984081041968162, + "grad_norm": 1.94070565700531, + "learning_rate": 2.8921962992759454e-06, + "loss": 0.9631, + "step": 680 + }, + { + "epoch": 0.984081041968162, + "eval_loss": 1.0258020162582397, + "eval_runtime": 51.0515, + "eval_samples_per_second": 19.588, + "eval_steps_per_second": 1.234, + "step": 680 + }, + { + "epoch": 0.9855282199710564, + "grad_norm": 1.6800276041030884, + "learning_rate": 2.8881737731295255e-06, + "loss": 0.9327, + "step": 681 + }, + { + "epoch": 0.9869753979739508, + "grad_norm": 1.6828765869140625, + "learning_rate": 2.8841512469831056e-06, + "loss": 0.8903, + "step": 682 + }, + { + "epoch": 0.9884225759768451, + "grad_norm": 1.2908765077590942, + "learning_rate": 2.8801287208366858e-06, + "loss": 0.9422, + "step": 683 + }, + { + "epoch": 0.9898697539797395, + "grad_norm": 1.0831834077835083, + "learning_rate": 2.876106194690266e-06, + "loss": 0.9077, + "step": 684 + }, + { + "epoch": 0.9913169319826338, + "grad_norm": 1.6049082279205322, + "learning_rate": 2.8720836685438456e-06, + "loss": 0.9153, + "step": 685 + }, + { + "epoch": 0.9927641099855282, + "grad_norm": 1.0960749387741089, + "learning_rate": 2.8680611423974257e-06, + "loss": 0.9355, + "step": 686 + }, + { + "epoch": 0.9942112879884226, + "grad_norm": 1.262939453125, + "learning_rate": 2.864038616251006e-06, + "loss": 0.9179, + "step": 687 + }, + { + "epoch": 0.9956584659913169, + "grad_norm": 1.528802514076233, + "learning_rate": 2.860016090104586e-06, + "loss": 0.9073, + "step": 688 + }, + { + "epoch": 0.9971056439942113, + "grad_norm": 1.309669017791748, + "learning_rate": 2.855993563958166e-06, + "loss": 0.9222, + "step": 689 + }, + { + "epoch": 0.9985528219971056, + "grad_norm": 1.4222102165222168, + "learning_rate": 2.851971037811746e-06, + "loss": 0.9536, + "step": 690 + }, + { + "epoch": 0.9985528219971056, + "eval_loss": 1.0252593755722046, + "eval_runtime": 50.6979, + "eval_samples_per_second": 19.725, + "eval_steps_per_second": 1.243, + "step": 690 + }, + { + "epoch": 1.0, + "grad_norm": 1.2282742261886597, + "learning_rate": 2.847948511665326e-06, + "loss": 0.9037, + "step": 691 + }, + { + "epoch": 1.0014471780028944, + "grad_norm": 1.3866548538208008, + "learning_rate": 2.843925985518906e-06, + "loss": 0.8822, + "step": 692 + }, + { + "epoch": 1.0028943560057888, + "grad_norm": 1.294751524925232, + "learning_rate": 2.839903459372486e-06, + "loss": 0.8787, + "step": 693 + }, + { + "epoch": 1.004341534008683, + "grad_norm": 1.396641731262207, + "learning_rate": 2.8358809332260663e-06, + "loss": 0.9202, + "step": 694 + }, + { + "epoch": 1.0057887120115774, + "grad_norm": 1.4702874422073364, + "learning_rate": 2.831858407079646e-06, + "loss": 0.9312, + "step": 695 + }, + { + "epoch": 1.0072358900144718, + "grad_norm": 1.3640304803848267, + "learning_rate": 2.827835880933226e-06, + "loss": 0.9199, + "step": 696 + }, + { + "epoch": 1.0086830680173662, + "grad_norm": 1.4911202192306519, + "learning_rate": 2.8238133547868063e-06, + "loss": 0.8955, + "step": 697 + }, + { + "epoch": 1.0101302460202606, + "grad_norm": 1.6224459409713745, + "learning_rate": 2.8197908286403864e-06, + "loss": 0.8781, + "step": 698 + }, + { + "epoch": 1.0115774240231548, + "grad_norm": 1.5970016717910767, + "learning_rate": 2.8157683024939665e-06, + "loss": 0.8876, + "step": 699 + }, + { + "epoch": 1.0130246020260492, + "grad_norm": 1.4060670137405396, + "learning_rate": 2.8117457763475462e-06, + "loss": 0.9532, + "step": 700 + }, + { + "epoch": 1.0130246020260492, + "eval_loss": 1.0271044969558716, + "eval_runtime": 50.7538, + "eval_samples_per_second": 19.703, + "eval_steps_per_second": 1.241, + "step": 700 + }, + { + "epoch": 1.0144717800289436, + "grad_norm": 1.297188639640808, + "learning_rate": 2.8077232502011264e-06, + "loss": 0.9368, + "step": 701 + }, + { + "epoch": 1.015918958031838, + "grad_norm": 1.2966526746749878, + "learning_rate": 2.8037007240547065e-06, + "loss": 0.9351, + "step": 702 + }, + { + "epoch": 1.0173661360347324, + "grad_norm": 1.6314148902893066, + "learning_rate": 2.7996781979082866e-06, + "loss": 0.9237, + "step": 703 + }, + { + "epoch": 1.0188133140376265, + "grad_norm": 1.378373146057129, + "learning_rate": 2.7956556717618667e-06, + "loss": 0.9068, + "step": 704 + }, + { + "epoch": 1.020260492040521, + "grad_norm": 1.5105211734771729, + "learning_rate": 2.7916331456154465e-06, + "loss": 0.9472, + "step": 705 + }, + { + "epoch": 1.0217076700434153, + "grad_norm": 1.6486235857009888, + "learning_rate": 2.7876106194690266e-06, + "loss": 0.8742, + "step": 706 + }, + { + "epoch": 1.0231548480463097, + "grad_norm": 1.3056670427322388, + "learning_rate": 2.7835880933226067e-06, + "loss": 0.9177, + "step": 707 + }, + { + "epoch": 1.0246020260492041, + "grad_norm": 1.4832851886749268, + "learning_rate": 2.779565567176187e-06, + "loss": 0.9493, + "step": 708 + }, + { + "epoch": 1.0260492040520983, + "grad_norm": 1.4473241567611694, + "learning_rate": 2.775543041029767e-06, + "loss": 0.8587, + "step": 709 + }, + { + "epoch": 1.0274963820549927, + "grad_norm": 1.8095839023590088, + "learning_rate": 2.7715205148833467e-06, + "loss": 0.8908, + "step": 710 + }, + { + "epoch": 1.0274963820549927, + "eval_loss": 1.029747486114502, + "eval_runtime": 50.9502, + "eval_samples_per_second": 19.627, + "eval_steps_per_second": 1.237, + "step": 710 + }, + { + "epoch": 1.0289435600578871, + "grad_norm": 1.4973514080047607, + "learning_rate": 2.767497988736927e-06, + "loss": 0.9165, + "step": 711 + }, + { + "epoch": 1.0303907380607815, + "grad_norm": 1.1892437934875488, + "learning_rate": 2.763475462590507e-06, + "loss": 0.9055, + "step": 712 + }, + { + "epoch": 1.031837916063676, + "grad_norm": 1.341482162475586, + "learning_rate": 2.759452936444087e-06, + "loss": 0.9226, + "step": 713 + }, + { + "epoch": 1.03328509406657, + "grad_norm": 1.4629768133163452, + "learning_rate": 2.755430410297667e-06, + "loss": 0.8997, + "step": 714 + }, + { + "epoch": 1.0347322720694645, + "grad_norm": 1.4336752891540527, + "learning_rate": 2.751407884151247e-06, + "loss": 0.9162, + "step": 715 + }, + { + "epoch": 1.036179450072359, + "grad_norm": 1.4192440509796143, + "learning_rate": 2.747385358004827e-06, + "loss": 0.9088, + "step": 716 + }, + { + "epoch": 1.0376266280752533, + "grad_norm": 1.2410304546356201, + "learning_rate": 2.743362831858407e-06, + "loss": 0.8993, + "step": 717 + }, + { + "epoch": 1.0390738060781477, + "grad_norm": 1.373382329940796, + "learning_rate": 2.7393403057119873e-06, + "loss": 0.8774, + "step": 718 + }, + { + "epoch": 1.0405209840810419, + "grad_norm": 1.3006587028503418, + "learning_rate": 2.7353177795655674e-06, + "loss": 0.8876, + "step": 719 + }, + { + "epoch": 1.0419681620839363, + "grad_norm": 1.28960120677948, + "learning_rate": 2.731295253419147e-06, + "loss": 0.9222, + "step": 720 + }, + { + "epoch": 1.0419681620839363, + "eval_loss": 1.0302447080612183, + "eval_runtime": 51.7627, + "eval_samples_per_second": 19.319, + "eval_steps_per_second": 1.217, + "step": 720 + }, + { + "epoch": 1.0434153400868307, + "grad_norm": 1.3036158084869385, + "learning_rate": 2.7272727272727272e-06, + "loss": 0.8926, + "step": 721 + }, + { + "epoch": 1.044862518089725, + "grad_norm": 1.2900220155715942, + "learning_rate": 2.7232502011263074e-06, + "loss": 0.8762, + "step": 722 + }, + { + "epoch": 1.0463096960926195, + "grad_norm": 1.282186508178711, + "learning_rate": 2.7192276749798875e-06, + "loss": 0.9061, + "step": 723 + }, + { + "epoch": 1.0477568740955137, + "grad_norm": 1.2362816333770752, + "learning_rate": 2.7152051488334676e-06, + "loss": 0.9021, + "step": 724 + }, + { + "epoch": 1.049204052098408, + "grad_norm": 1.3351249694824219, + "learning_rate": 2.7111826226870473e-06, + "loss": 0.8621, + "step": 725 + }, + { + "epoch": 1.0506512301013025, + "grad_norm": 1.233289122581482, + "learning_rate": 2.7071600965406274e-06, + "loss": 0.8486, + "step": 726 + }, + { + "epoch": 1.0520984081041969, + "grad_norm": 1.3227379322052002, + "learning_rate": 2.7031375703942076e-06, + "loss": 0.8987, + "step": 727 + }, + { + "epoch": 1.0535455861070913, + "grad_norm": 1.3924241065979004, + "learning_rate": 2.6991150442477877e-06, + "loss": 0.9018, + "step": 728 + }, + { + "epoch": 1.0549927641099854, + "grad_norm": 1.756554126739502, + "learning_rate": 2.695092518101368e-06, + "loss": 0.9683, + "step": 729 + }, + { + "epoch": 1.0564399421128798, + "grad_norm": 1.2942228317260742, + "learning_rate": 2.6910699919549475e-06, + "loss": 0.9001, + "step": 730 + }, + { + "epoch": 1.0564399421128798, + "eval_loss": 1.0273805856704712, + "eval_runtime": 51.4178, + "eval_samples_per_second": 19.449, + "eval_steps_per_second": 1.225, + "step": 730 + }, + { + "epoch": 1.0578871201157742, + "grad_norm": 1.1950814723968506, + "learning_rate": 2.6870474658085277e-06, + "loss": 0.8785, + "step": 731 + }, + { + "epoch": 1.0593342981186686, + "grad_norm": 1.2649741172790527, + "learning_rate": 2.6830249396621078e-06, + "loss": 0.9063, + "step": 732 + }, + { + "epoch": 1.060781476121563, + "grad_norm": 1.3322865962982178, + "learning_rate": 2.679002413515688e-06, + "loss": 0.9102, + "step": 733 + }, + { + "epoch": 1.0622286541244572, + "grad_norm": 1.718910574913025, + "learning_rate": 2.6749798873692685e-06, + "loss": 0.8992, + "step": 734 + }, + { + "epoch": 1.0636758321273516, + "grad_norm": 1.1307506561279297, + "learning_rate": 2.6709573612228486e-06, + "loss": 0.8514, + "step": 735 + }, + { + "epoch": 1.065123010130246, + "grad_norm": 1.3008947372436523, + "learning_rate": 2.6669348350764283e-06, + "loss": 0.8939, + "step": 736 + }, + { + "epoch": 1.0665701881331404, + "grad_norm": 1.3788453340530396, + "learning_rate": 2.6629123089300084e-06, + "loss": 0.8995, + "step": 737 + }, + { + "epoch": 1.0680173661360348, + "grad_norm": 1.361738681793213, + "learning_rate": 2.6588897827835885e-06, + "loss": 0.9114, + "step": 738 + }, + { + "epoch": 1.069464544138929, + "grad_norm": 1.29777193069458, + "learning_rate": 2.6548672566371687e-06, + "loss": 0.9389, + "step": 739 + }, + { + "epoch": 1.0709117221418234, + "grad_norm": 1.2898824214935303, + "learning_rate": 2.650844730490749e-06, + "loss": 0.9169, + "step": 740 + }, + { + "epoch": 1.0709117221418234, + "eval_loss": 1.0288442373275757, + "eval_runtime": 51.5593, + "eval_samples_per_second": 19.395, + "eval_steps_per_second": 1.222, + "step": 740 + }, + { + "epoch": 1.0723589001447178, + "grad_norm": 1.3063137531280518, + "learning_rate": 2.6468222043443285e-06, + "loss": 0.8607, + "step": 741 + }, + { + "epoch": 1.0738060781476122, + "grad_norm": 1.2504301071166992, + "learning_rate": 2.6427996781979086e-06, + "loss": 0.891, + "step": 742 + }, + { + "epoch": 1.0752532561505066, + "grad_norm": 1.331756591796875, + "learning_rate": 2.6387771520514888e-06, + "loss": 0.9013, + "step": 743 + }, + { + "epoch": 1.0767004341534008, + "grad_norm": 1.4176838397979736, + "learning_rate": 2.634754625905069e-06, + "loss": 0.895, + "step": 744 + }, + { + "epoch": 1.0781476121562952, + "grad_norm": 1.3297080993652344, + "learning_rate": 2.630732099758649e-06, + "loss": 0.8828, + "step": 745 + }, + { + "epoch": 1.0795947901591896, + "grad_norm": 1.440510869026184, + "learning_rate": 2.6267095736122287e-06, + "loss": 0.9079, + "step": 746 + }, + { + "epoch": 1.081041968162084, + "grad_norm": 1.1796343326568604, + "learning_rate": 2.622687047465809e-06, + "loss": 0.9122, + "step": 747 + }, + { + "epoch": 1.0824891461649784, + "grad_norm": 1.266837239265442, + "learning_rate": 2.618664521319389e-06, + "loss": 0.9418, + "step": 748 + }, + { + "epoch": 1.0839363241678726, + "grad_norm": 1.4167532920837402, + "learning_rate": 2.614641995172969e-06, + "loss": 0.8711, + "step": 749 + }, + { + "epoch": 1.085383502170767, + "grad_norm": 1.4737548828125, + "learning_rate": 2.6106194690265492e-06, + "loss": 0.9022, + "step": 750 + }, + { + "epoch": 1.085383502170767, + "eval_loss": 1.030425786972046, + "eval_runtime": 51.5159, + "eval_samples_per_second": 19.411, + "eval_steps_per_second": 1.223, + "step": 750 + }, + { + "epoch": 1.0868306801736614, + "grad_norm": 1.3065539598464966, + "learning_rate": 2.606596942880129e-06, + "loss": 0.9066, + "step": 751 + }, + { + "epoch": 1.0882778581765558, + "grad_norm": 1.426745891571045, + "learning_rate": 2.602574416733709e-06, + "loss": 0.9001, + "step": 752 + }, + { + "epoch": 1.0897250361794502, + "grad_norm": 1.5242563486099243, + "learning_rate": 2.598551890587289e-06, + "loss": 0.9197, + "step": 753 + }, + { + "epoch": 1.0911722141823443, + "grad_norm": 1.354071021080017, + "learning_rate": 2.5945293644408693e-06, + "loss": 0.9194, + "step": 754 + }, + { + "epoch": 1.0926193921852387, + "grad_norm": 1.2317461967468262, + "learning_rate": 2.5905068382944494e-06, + "loss": 0.9241, + "step": 755 + }, + { + "epoch": 1.0940665701881331, + "grad_norm": 1.4435662031173706, + "learning_rate": 2.586484312148029e-06, + "loss": 0.8655, + "step": 756 + }, + { + "epoch": 1.0955137481910275, + "grad_norm": 1.1415653228759766, + "learning_rate": 2.5824617860016093e-06, + "loss": 0.9107, + "step": 757 + }, + { + "epoch": 1.096960926193922, + "grad_norm": 1.2376630306243896, + "learning_rate": 2.5784392598551894e-06, + "loss": 0.9006, + "step": 758 + }, + { + "epoch": 1.0984081041968161, + "grad_norm": 1.273107647895813, + "learning_rate": 2.5744167337087695e-06, + "loss": 0.8819, + "step": 759 + }, + { + "epoch": 1.0998552821997105, + "grad_norm": 1.5148357152938843, + "learning_rate": 2.5703942075623497e-06, + "loss": 0.8672, + "step": 760 + }, + { + "epoch": 1.0998552821997105, + "eval_loss": 1.0317097902297974, + "eval_runtime": 51.2206, + "eval_samples_per_second": 19.523, + "eval_steps_per_second": 1.23, + "step": 760 + }, + { + "epoch": 1.101302460202605, + "grad_norm": 1.4190945625305176, + "learning_rate": 2.5663716814159294e-06, + "loss": 0.8944, + "step": 761 + }, + { + "epoch": 1.1027496382054993, + "grad_norm": 1.567635416984558, + "learning_rate": 2.5623491552695095e-06, + "loss": 0.8696, + "step": 762 + }, + { + "epoch": 1.1041968162083937, + "grad_norm": 1.4464750289916992, + "learning_rate": 2.5583266291230896e-06, + "loss": 0.874, + "step": 763 + }, + { + "epoch": 1.105643994211288, + "grad_norm": 1.4979974031448364, + "learning_rate": 2.5543041029766698e-06, + "loss": 0.921, + "step": 764 + }, + { + "epoch": 1.1070911722141823, + "grad_norm": 1.2645759582519531, + "learning_rate": 2.55028157683025e-06, + "loss": 0.8969, + "step": 765 + }, + { + "epoch": 1.1085383502170767, + "grad_norm": 1.450600028038025, + "learning_rate": 2.5462590506838296e-06, + "loss": 0.8928, + "step": 766 + }, + { + "epoch": 1.109985528219971, + "grad_norm": 1.8721531629562378, + "learning_rate": 2.5422365245374097e-06, + "loss": 0.8923, + "step": 767 + }, + { + "epoch": 1.1114327062228655, + "grad_norm": 2.058394432067871, + "learning_rate": 2.53821399839099e-06, + "loss": 0.9044, + "step": 768 + }, + { + "epoch": 1.1128798842257597, + "grad_norm": 1.1461564302444458, + "learning_rate": 2.53419147224457e-06, + "loss": 0.8867, + "step": 769 + }, + { + "epoch": 1.114327062228654, + "grad_norm": 1.3547521829605103, + "learning_rate": 2.53016894609815e-06, + "loss": 0.9522, + "step": 770 + }, + { + "epoch": 1.114327062228654, + "eval_loss": 1.0291906595230103, + "eval_runtime": 51.6318, + "eval_samples_per_second": 19.368, + "eval_steps_per_second": 1.22, + "step": 770 + }, + { + "epoch": 1.1157742402315485, + "grad_norm": 1.2312697172164917, + "learning_rate": 2.52614641995173e-06, + "loss": 0.9044, + "step": 771 + }, + { + "epoch": 1.1172214182344429, + "grad_norm": 1.3786168098449707, + "learning_rate": 2.52212389380531e-06, + "loss": 0.8922, + "step": 772 + }, + { + "epoch": 1.1186685962373373, + "grad_norm": 1.369730830192566, + "learning_rate": 2.51810136765889e-06, + "loss": 0.9413, + "step": 773 + }, + { + "epoch": 1.1201157742402315, + "grad_norm": 1.7282310724258423, + "learning_rate": 2.51407884151247e-06, + "loss": 0.93, + "step": 774 + }, + { + "epoch": 1.1215629522431259, + "grad_norm": 1.4477009773254395, + "learning_rate": 2.5100563153660503e-06, + "loss": 0.8872, + "step": 775 + }, + { + "epoch": 1.1230101302460203, + "grad_norm": 1.3932946920394897, + "learning_rate": 2.50603378921963e-06, + "loss": 0.9014, + "step": 776 + }, + { + "epoch": 1.1244573082489147, + "grad_norm": 1.3780879974365234, + "learning_rate": 2.50201126307321e-06, + "loss": 0.9064, + "step": 777 + }, + { + "epoch": 1.125904486251809, + "grad_norm": 1.592652440071106, + "learning_rate": 2.4979887369267903e-06, + "loss": 0.8809, + "step": 778 + }, + { + "epoch": 1.1273516642547032, + "grad_norm": 1.5301320552825928, + "learning_rate": 2.4939662107803704e-06, + "loss": 0.9236, + "step": 779 + }, + { + "epoch": 1.1287988422575976, + "grad_norm": 1.4404382705688477, + "learning_rate": 2.48994368463395e-06, + "loss": 0.9064, + "step": 780 + }, + { + "epoch": 1.1287988422575976, + "eval_loss": 1.0307024717330933, + "eval_runtime": 51.3353, + "eval_samples_per_second": 19.48, + "eval_steps_per_second": 1.227, + "step": 780 + }, + { + "epoch": 1.130246020260492, + "grad_norm": 1.417981743812561, + "learning_rate": 2.4859211584875302e-06, + "loss": 0.8748, + "step": 781 + }, + { + "epoch": 1.1316931982633864, + "grad_norm": 1.1536470651626587, + "learning_rate": 2.4818986323411104e-06, + "loss": 0.9152, + "step": 782 + }, + { + "epoch": 1.1331403762662808, + "grad_norm": 1.1965649127960205, + "learning_rate": 2.4778761061946905e-06, + "loss": 0.9135, + "step": 783 + }, + { + "epoch": 1.134587554269175, + "grad_norm": 1.4656903743743896, + "learning_rate": 2.4738535800482706e-06, + "loss": 0.881, + "step": 784 + }, + { + "epoch": 1.1360347322720694, + "grad_norm": 1.3402633666992188, + "learning_rate": 2.4698310539018503e-06, + "loss": 0.9232, + "step": 785 + }, + { + "epoch": 1.1374819102749638, + "grad_norm": 1.399426817893982, + "learning_rate": 2.4658085277554304e-06, + "loss": 0.8891, + "step": 786 + }, + { + "epoch": 1.1389290882778582, + "grad_norm": 1.303605556488037, + "learning_rate": 2.4617860016090106e-06, + "loss": 0.9194, + "step": 787 + }, + { + "epoch": 1.1403762662807526, + "grad_norm": 1.5588994026184082, + "learning_rate": 2.4577634754625907e-06, + "loss": 0.8918, + "step": 788 + }, + { + "epoch": 1.1418234442836468, + "grad_norm": 1.1864101886749268, + "learning_rate": 2.453740949316171e-06, + "loss": 0.8915, + "step": 789 + }, + { + "epoch": 1.1432706222865412, + "grad_norm": 1.3448659181594849, + "learning_rate": 2.4497184231697505e-06, + "loss": 0.9057, + "step": 790 + }, + { + "epoch": 1.1432706222865412, + "eval_loss": 1.0278713703155518, + "eval_runtime": 51.4784, + "eval_samples_per_second": 19.426, + "eval_steps_per_second": 1.224, + "step": 790 + }, + { + "epoch": 1.1447178002894356, + "grad_norm": 1.362930417060852, + "learning_rate": 2.4456958970233307e-06, + "loss": 0.9101, + "step": 791 + }, + { + "epoch": 1.14616497829233, + "grad_norm": 1.5270448923110962, + "learning_rate": 2.4416733708769108e-06, + "loss": 0.8967, + "step": 792 + }, + { + "epoch": 1.1476121562952244, + "grad_norm": 1.2606215476989746, + "learning_rate": 2.437650844730491e-06, + "loss": 0.8947, + "step": 793 + }, + { + "epoch": 1.1490593342981186, + "grad_norm": 1.3619481325149536, + "learning_rate": 2.433628318584071e-06, + "loss": 0.9484, + "step": 794 + }, + { + "epoch": 1.150506512301013, + "grad_norm": 1.3002569675445557, + "learning_rate": 2.4296057924376507e-06, + "loss": 0.8761, + "step": 795 + }, + { + "epoch": 1.1519536903039074, + "grad_norm": 1.3039034605026245, + "learning_rate": 2.425583266291231e-06, + "loss": 0.9059, + "step": 796 + }, + { + "epoch": 1.1534008683068018, + "grad_norm": 1.1996506452560425, + "learning_rate": 2.421560740144811e-06, + "loss": 0.8775, + "step": 797 + }, + { + "epoch": 1.1548480463096962, + "grad_norm": 1.3600144386291504, + "learning_rate": 2.417538213998391e-06, + "loss": 0.8475, + "step": 798 + }, + { + "epoch": 1.1562952243125904, + "grad_norm": 1.2119777202606201, + "learning_rate": 2.4135156878519713e-06, + "loss": 0.89, + "step": 799 + }, + { + "epoch": 1.1577424023154848, + "grad_norm": 1.4600919485092163, + "learning_rate": 2.4094931617055514e-06, + "loss": 0.8972, + "step": 800 + }, + { + "epoch": 1.1577424023154848, + "eval_loss": 1.0256906747817993, + "eval_runtime": 51.3839, + "eval_samples_per_second": 19.461, + "eval_steps_per_second": 1.226, + "step": 800 + }, + { + "epoch": 1.1591895803183792, + "grad_norm": 1.5115501880645752, + "learning_rate": 2.4054706355591315e-06, + "loss": 0.9493, + "step": 801 + }, + { + "epoch": 1.1606367583212736, + "grad_norm": 1.4013158082962036, + "learning_rate": 2.4014481094127116e-06, + "loss": 0.8675, + "step": 802 + }, + { + "epoch": 1.162083936324168, + "grad_norm": 1.592124342918396, + "learning_rate": 2.3974255832662913e-06, + "loss": 0.9238, + "step": 803 + }, + { + "epoch": 1.1635311143270621, + "grad_norm": 1.2157760858535767, + "learning_rate": 2.3934030571198715e-06, + "loss": 0.9054, + "step": 804 + }, + { + "epoch": 1.1649782923299565, + "grad_norm": 1.1731619834899902, + "learning_rate": 2.3893805309734516e-06, + "loss": 0.8921, + "step": 805 + }, + { + "epoch": 1.166425470332851, + "grad_norm": 1.3047113418579102, + "learning_rate": 2.3853580048270317e-06, + "loss": 0.893, + "step": 806 + }, + { + "epoch": 1.1678726483357453, + "grad_norm": 1.4774771928787231, + "learning_rate": 2.381335478680612e-06, + "loss": 0.9316, + "step": 807 + }, + { + "epoch": 1.1693198263386397, + "grad_norm": 1.6613857746124268, + "learning_rate": 2.3773129525341916e-06, + "loss": 0.8829, + "step": 808 + }, + { + "epoch": 1.170767004341534, + "grad_norm": 1.2833667993545532, + "learning_rate": 2.3732904263877717e-06, + "loss": 0.8762, + "step": 809 + }, + { + "epoch": 1.1722141823444283, + "grad_norm": 1.3055139780044556, + "learning_rate": 2.369267900241352e-06, + "loss": 0.9184, + "step": 810 + }, + { + "epoch": 1.1722141823444283, + "eval_loss": 1.0228921175003052, + "eval_runtime": 51.2994, + "eval_samples_per_second": 19.493, + "eval_steps_per_second": 1.228, + "step": 810 + }, + { + "epoch": 1.1736613603473227, + "grad_norm": 1.2308969497680664, + "learning_rate": 2.365245374094932e-06, + "loss": 0.8929, + "step": 811 + }, + { + "epoch": 1.1751085383502171, + "grad_norm": 1.2995024919509888, + "learning_rate": 2.361222847948512e-06, + "loss": 0.8948, + "step": 812 + }, + { + "epoch": 1.1765557163531115, + "grad_norm": 1.27108895778656, + "learning_rate": 2.3572003218020918e-06, + "loss": 0.9054, + "step": 813 + }, + { + "epoch": 1.1780028943560057, + "grad_norm": 1.4717758893966675, + "learning_rate": 2.353177795655672e-06, + "loss": 0.9235, + "step": 814 + }, + { + "epoch": 1.1794500723589, + "grad_norm": 1.4134477376937866, + "learning_rate": 2.349155269509252e-06, + "loss": 0.9172, + "step": 815 + }, + { + "epoch": 1.1808972503617945, + "grad_norm": 1.5222864151000977, + "learning_rate": 2.345132743362832e-06, + "loss": 0.9185, + "step": 816 + }, + { + "epoch": 1.182344428364689, + "grad_norm": 1.409335970878601, + "learning_rate": 2.3411102172164123e-06, + "loss": 0.9111, + "step": 817 + }, + { + "epoch": 1.1837916063675833, + "grad_norm": 1.1633872985839844, + "learning_rate": 2.337087691069992e-06, + "loss": 0.8955, + "step": 818 + }, + { + "epoch": 1.1852387843704775, + "grad_norm": 1.2191036939620972, + "learning_rate": 2.333065164923572e-06, + "loss": 0.916, + "step": 819 + }, + { + "epoch": 1.1866859623733719, + "grad_norm": 1.638218641281128, + "learning_rate": 2.3290426387771522e-06, + "loss": 0.8686, + "step": 820 + }, + { + "epoch": 1.1866859623733719, + "eval_loss": 1.0238690376281738, + "eval_runtime": 51.2907, + "eval_samples_per_second": 19.497, + "eval_steps_per_second": 1.228, + "step": 820 + }, + { + "epoch": 1.1881331403762663, + "grad_norm": 1.203157901763916, + "learning_rate": 2.3250201126307324e-06, + "loss": 0.9122, + "step": 821 + }, + { + "epoch": 1.1895803183791607, + "grad_norm": 1.4143924713134766, + "learning_rate": 2.3209975864843125e-06, + "loss": 0.9211, + "step": 822 + }, + { + "epoch": 1.191027496382055, + "grad_norm": 1.4647315740585327, + "learning_rate": 2.316975060337892e-06, + "loss": 0.8965, + "step": 823 + }, + { + "epoch": 1.1924746743849495, + "grad_norm": 1.3775290250778198, + "learning_rate": 2.3129525341914723e-06, + "loss": 0.8912, + "step": 824 + }, + { + "epoch": 1.1939218523878437, + "grad_norm": 1.2931272983551025, + "learning_rate": 2.3089300080450525e-06, + "loss": 0.9263, + "step": 825 + }, + { + "epoch": 1.195369030390738, + "grad_norm": 1.298216700553894, + "learning_rate": 2.3049074818986326e-06, + "loss": 0.9101, + "step": 826 + }, + { + "epoch": 1.1968162083936325, + "grad_norm": 1.3943524360656738, + "learning_rate": 2.3008849557522127e-06, + "loss": 0.9233, + "step": 827 + }, + { + "epoch": 1.1982633863965269, + "grad_norm": 1.2961311340332031, + "learning_rate": 2.2968624296057924e-06, + "loss": 0.8971, + "step": 828 + }, + { + "epoch": 1.199710564399421, + "grad_norm": 1.5847829580307007, + "learning_rate": 2.2928399034593725e-06, + "loss": 0.9234, + "step": 829 + }, + { + "epoch": 1.2011577424023154, + "grad_norm": 1.3470019102096558, + "learning_rate": 2.2888173773129527e-06, + "loss": 0.8911, + "step": 830 + }, + { + "epoch": 1.2011577424023154, + "eval_loss": 1.024039626121521, + "eval_runtime": 51.5627, + "eval_samples_per_second": 19.394, + "eval_steps_per_second": 1.222, + "step": 830 + }, + { + "epoch": 1.2026049204052098, + "grad_norm": 1.459540605545044, + "learning_rate": 2.284794851166533e-06, + "loss": 0.8792, + "step": 831 + }, + { + "epoch": 1.2040520984081042, + "grad_norm": 1.4610881805419922, + "learning_rate": 2.280772325020113e-06, + "loss": 0.8985, + "step": 832 + }, + { + "epoch": 1.2054992764109986, + "grad_norm": 1.508454442024231, + "learning_rate": 2.2767497988736926e-06, + "loss": 0.887, + "step": 833 + }, + { + "epoch": 1.206946454413893, + "grad_norm": 1.3374409675598145, + "learning_rate": 2.2727272727272728e-06, + "loss": 0.8895, + "step": 834 + }, + { + "epoch": 1.2083936324167872, + "grad_norm": 1.6953614950180054, + "learning_rate": 2.268704746580853e-06, + "loss": 0.8656, + "step": 835 + }, + { + "epoch": 1.2098408104196816, + "grad_norm": 1.497324824333191, + "learning_rate": 2.264682220434433e-06, + "loss": 0.8877, + "step": 836 + }, + { + "epoch": 1.211287988422576, + "grad_norm": 1.2943347692489624, + "learning_rate": 2.260659694288013e-06, + "loss": 0.8944, + "step": 837 + }, + { + "epoch": 1.2127351664254704, + "grad_norm": 1.3988865613937378, + "learning_rate": 2.256637168141593e-06, + "loss": 0.8986, + "step": 838 + }, + { + "epoch": 1.2141823444283646, + "grad_norm": 1.2704005241394043, + "learning_rate": 2.252614641995173e-06, + "loss": 0.8733, + "step": 839 + }, + { + "epoch": 1.215629522431259, + "grad_norm": 1.561896800994873, + "learning_rate": 2.248592115848753e-06, + "loss": 0.907, + "step": 840 + }, + { + "epoch": 1.215629522431259, + "eval_loss": 1.0240271091461182, + "eval_runtime": 51.4382, + "eval_samples_per_second": 19.441, + "eval_steps_per_second": 1.225, + "step": 840 + }, + { + "epoch": 1.2170767004341534, + "grad_norm": 1.2712055444717407, + "learning_rate": 2.2445695897023332e-06, + "loss": 0.8663, + "step": 841 + }, + { + "epoch": 1.2185238784370478, + "grad_norm": 1.3718512058258057, + "learning_rate": 2.2405470635559134e-06, + "loss": 0.8961, + "step": 842 + }, + { + "epoch": 1.2199710564399422, + "grad_norm": 1.449492335319519, + "learning_rate": 2.236524537409493e-06, + "loss": 0.8669, + "step": 843 + }, + { + "epoch": 1.2214182344428366, + "grad_norm": 1.5076099634170532, + "learning_rate": 2.232502011263073e-06, + "loss": 0.8853, + "step": 844 + }, + { + "epoch": 1.2228654124457308, + "grad_norm": 1.3568211793899536, + "learning_rate": 2.2284794851166537e-06, + "loss": 0.8982, + "step": 845 + }, + { + "epoch": 1.2243125904486252, + "grad_norm": 1.34610116481781, + "learning_rate": 2.2244569589702334e-06, + "loss": 0.9104, + "step": 846 + }, + { + "epoch": 1.2257597684515196, + "grad_norm": 1.2830678224563599, + "learning_rate": 2.2204344328238136e-06, + "loss": 0.904, + "step": 847 + }, + { + "epoch": 1.227206946454414, + "grad_norm": 1.3711849451065063, + "learning_rate": 2.2164119066773937e-06, + "loss": 0.927, + "step": 848 + }, + { + "epoch": 1.2286541244573081, + "grad_norm": 1.373616337776184, + "learning_rate": 2.212389380530974e-06, + "loss": 0.8905, + "step": 849 + }, + { + "epoch": 1.2301013024602026, + "grad_norm": 1.2533118724822998, + "learning_rate": 2.208366854384554e-06, + "loss": 0.8912, + "step": 850 + }, + { + "epoch": 1.2301013024602026, + "eval_loss": 1.0211647748947144, + "eval_runtime": 51.2128, + "eval_samples_per_second": 19.526, + "eval_steps_per_second": 1.23, + "step": 850 + }, + { + "epoch": 1.231548480463097, + "grad_norm": 1.2218654155731201, + "learning_rate": 2.2043443282381337e-06, + "loss": 0.9031, + "step": 851 + }, + { + "epoch": 1.2329956584659914, + "grad_norm": 1.409911870956421, + "learning_rate": 2.2003218020917138e-06, + "loss": 0.9049, + "step": 852 + }, + { + "epoch": 1.2344428364688858, + "grad_norm": 1.412427544593811, + "learning_rate": 2.196299275945294e-06, + "loss": 0.8795, + "step": 853 + }, + { + "epoch": 1.2358900144717802, + "grad_norm": 1.4679683446884155, + "learning_rate": 2.192276749798874e-06, + "loss": 0.8913, + "step": 854 + }, + { + "epoch": 1.2373371924746743, + "grad_norm": 1.186155915260315, + "learning_rate": 2.188254223652454e-06, + "loss": 0.9007, + "step": 855 + }, + { + "epoch": 1.2387843704775687, + "grad_norm": 1.2910406589508057, + "learning_rate": 2.184231697506034e-06, + "loss": 0.867, + "step": 856 + }, + { + "epoch": 1.2402315484804631, + "grad_norm": 1.2544832229614258, + "learning_rate": 2.180209171359614e-06, + "loss": 0.8913, + "step": 857 + }, + { + "epoch": 1.2416787264833575, + "grad_norm": 1.4911911487579346, + "learning_rate": 2.176186645213194e-06, + "loss": 0.8981, + "step": 858 + }, + { + "epoch": 1.2431259044862517, + "grad_norm": 1.2359133958816528, + "learning_rate": 2.1721641190667743e-06, + "loss": 0.8786, + "step": 859 + }, + { + "epoch": 1.244573082489146, + "grad_norm": 1.3512214422225952, + "learning_rate": 2.1681415929203544e-06, + "loss": 0.9308, + "step": 860 + }, + { + "epoch": 1.244573082489146, + "eval_loss": 1.0206845998764038, + "eval_runtime": 51.1271, + "eval_samples_per_second": 19.559, + "eval_steps_per_second": 1.232, + "step": 860 + }, + { + "epoch": 1.2460202604920405, + "grad_norm": 1.419176459312439, + "learning_rate": 2.164119066773934e-06, + "loss": 0.8813, + "step": 861 + }, + { + "epoch": 1.247467438494935, + "grad_norm": 1.3000375032424927, + "learning_rate": 2.1600965406275142e-06, + "loss": 0.899, + "step": 862 + }, + { + "epoch": 1.2489146164978293, + "grad_norm": 1.3746064901351929, + "learning_rate": 2.1560740144810943e-06, + "loss": 0.8772, + "step": 863 + }, + { + "epoch": 1.2503617945007237, + "grad_norm": 1.162196397781372, + "learning_rate": 2.1520514883346745e-06, + "loss": 0.8573, + "step": 864 + }, + { + "epoch": 1.251808972503618, + "grad_norm": 1.1629972457885742, + "learning_rate": 2.1480289621882546e-06, + "loss": 0.8933, + "step": 865 + }, + { + "epoch": 1.2532561505065123, + "grad_norm": 1.3075627088546753, + "learning_rate": 2.1440064360418343e-06, + "loss": 0.9088, + "step": 866 + }, + { + "epoch": 1.2547033285094067, + "grad_norm": 1.386847972869873, + "learning_rate": 2.1399839098954144e-06, + "loss": 0.9041, + "step": 867 + }, + { + "epoch": 1.256150506512301, + "grad_norm": 1.3599576950073242, + "learning_rate": 2.1359613837489946e-06, + "loss": 0.8876, + "step": 868 + }, + { + "epoch": 1.2575976845151953, + "grad_norm": 1.5237191915512085, + "learning_rate": 2.1319388576025747e-06, + "loss": 0.8797, + "step": 869 + }, + { + "epoch": 1.2590448625180897, + "grad_norm": 1.4234733581542969, + "learning_rate": 2.127916331456155e-06, + "loss": 0.9145, + "step": 870 + }, + { + "epoch": 1.2590448625180897, + "eval_loss": 1.0202274322509766, + "eval_runtime": 51.2422, + "eval_samples_per_second": 19.515, + "eval_steps_per_second": 1.229, + "step": 870 + }, + { + "epoch": 1.260492040520984, + "grad_norm": 1.4617350101470947, + "learning_rate": 2.1238938053097345e-06, + "loss": 0.8853, + "step": 871 + }, + { + "epoch": 1.2619392185238785, + "grad_norm": 1.2346954345703125, + "learning_rate": 2.1198712791633146e-06, + "loss": 0.8928, + "step": 872 + }, + { + "epoch": 1.2633863965267729, + "grad_norm": 1.2907211780548096, + "learning_rate": 2.1158487530168948e-06, + "loss": 0.851, + "step": 873 + }, + { + "epoch": 1.2648335745296673, + "grad_norm": 1.2782104015350342, + "learning_rate": 2.111826226870475e-06, + "loss": 0.8848, + "step": 874 + }, + { + "epoch": 1.2662807525325614, + "grad_norm": 1.5206267833709717, + "learning_rate": 2.107803700724055e-06, + "loss": 0.8924, + "step": 875 + }, + { + "epoch": 1.2677279305354558, + "grad_norm": 1.294076919555664, + "learning_rate": 2.1037811745776347e-06, + "loss": 0.8455, + "step": 876 + }, + { + "epoch": 1.2691751085383502, + "grad_norm": 1.2310216426849365, + "learning_rate": 2.099758648431215e-06, + "loss": 0.8891, + "step": 877 + }, + { + "epoch": 1.2706222865412446, + "grad_norm": 1.4888914823532104, + "learning_rate": 2.095736122284795e-06, + "loss": 0.8986, + "step": 878 + }, + { + "epoch": 1.2720694645441388, + "grad_norm": 1.5908782482147217, + "learning_rate": 2.091713596138375e-06, + "loss": 0.9012, + "step": 879 + }, + { + "epoch": 1.2735166425470332, + "grad_norm": 1.3177837133407593, + "learning_rate": 2.0876910699919552e-06, + "loss": 0.8968, + "step": 880 + }, + { + "epoch": 1.2735166425470332, + "eval_loss": 1.0179506540298462, + "eval_runtime": 51.162, + "eval_samples_per_second": 19.546, + "eval_steps_per_second": 1.231, + "step": 880 + }, + { + "epoch": 1.2749638205499276, + "grad_norm": 1.3695924282073975, + "learning_rate": 2.083668543845535e-06, + "loss": 0.9013, + "step": 881 + }, + { + "epoch": 1.276410998552822, + "grad_norm": 1.2539571523666382, + "learning_rate": 2.079646017699115e-06, + "loss": 0.9086, + "step": 882 + }, + { + "epoch": 1.2778581765557164, + "grad_norm": 1.4029269218444824, + "learning_rate": 2.075623491552695e-06, + "loss": 0.87, + "step": 883 + }, + { + "epoch": 1.2793053545586108, + "grad_norm": 1.4106091260910034, + "learning_rate": 2.0716009654062753e-06, + "loss": 0.8922, + "step": 884 + }, + { + "epoch": 1.280752532561505, + "grad_norm": 1.2717335224151611, + "learning_rate": 2.067578439259855e-06, + "loss": 0.9014, + "step": 885 + }, + { + "epoch": 1.2821997105643994, + "grad_norm": 1.3195486068725586, + "learning_rate": 2.063555913113435e-06, + "loss": 0.89, + "step": 886 + }, + { + "epoch": 1.2836468885672938, + "grad_norm": 1.3389581441879272, + "learning_rate": 2.0595333869670153e-06, + "loss": 0.8904, + "step": 887 + }, + { + "epoch": 1.2850940665701882, + "grad_norm": 1.3154371976852417, + "learning_rate": 2.0555108608205954e-06, + "loss": 0.8815, + "step": 888 + }, + { + "epoch": 1.2865412445730824, + "grad_norm": 1.2228502035140991, + "learning_rate": 2.0514883346741755e-06, + "loss": 0.9022, + "step": 889 + }, + { + "epoch": 1.2879884225759768, + "grad_norm": 1.2586122751235962, + "learning_rate": 2.0474658085277557e-06, + "loss": 0.8907, + "step": 890 + }, + { + "epoch": 1.2879884225759768, + "eval_loss": 1.0168894529342651, + "eval_runtime": 51.3487, + "eval_samples_per_second": 19.475, + "eval_steps_per_second": 1.227, + "step": 890 + }, + { + "epoch": 1.2894356005788712, + "grad_norm": 1.1529226303100586, + "learning_rate": 2.043443282381336e-06, + "loss": 0.8583, + "step": 891 + }, + { + "epoch": 1.2908827785817656, + "grad_norm": 1.1946406364440918, + "learning_rate": 2.039420756234916e-06, + "loss": 0.8935, + "step": 892 + }, + { + "epoch": 1.29232995658466, + "grad_norm": 1.5191806554794312, + "learning_rate": 2.035398230088496e-06, + "loss": 0.9034, + "step": 893 + }, + { + "epoch": 1.2937771345875544, + "grad_norm": 1.3799525499343872, + "learning_rate": 2.0313757039420758e-06, + "loss": 0.9024, + "step": 894 + }, + { + "epoch": 1.2952243125904486, + "grad_norm": 1.268713355064392, + "learning_rate": 2.027353177795656e-06, + "loss": 0.8846, + "step": 895 + }, + { + "epoch": 1.296671490593343, + "grad_norm": 1.5482555627822876, + "learning_rate": 2.023330651649236e-06, + "loss": 0.8668, + "step": 896 + }, + { + "epoch": 1.2981186685962374, + "grad_norm": 1.1756871938705444, + "learning_rate": 2.019308125502816e-06, + "loss": 0.8943, + "step": 897 + }, + { + "epoch": 1.2995658465991318, + "grad_norm": 1.3052146434783936, + "learning_rate": 2.0152855993563963e-06, + "loss": 0.901, + "step": 898 + }, + { + "epoch": 1.301013024602026, + "grad_norm": 1.4517641067504883, + "learning_rate": 2.011263073209976e-06, + "loss": 0.8966, + "step": 899 + }, + { + "epoch": 1.3024602026049203, + "grad_norm": 1.357826590538025, + "learning_rate": 2.007240547063556e-06, + "loss": 0.8862, + "step": 900 + }, + { + "epoch": 1.3024602026049203, + "eval_loss": 1.0154352188110352, + "eval_runtime": 51.1543, + "eval_samples_per_second": 19.549, + "eval_steps_per_second": 1.232, + "step": 900 + }, + { + "epoch": 1.3039073806078147, + "grad_norm": 1.4572620391845703, + "learning_rate": 2.0032180209171362e-06, + "loss": 0.908, + "step": 901 + }, + { + "epoch": 1.3053545586107091, + "grad_norm": 1.3969900608062744, + "learning_rate": 1.9991954947707164e-06, + "loss": 0.894, + "step": 902 + }, + { + "epoch": 1.3068017366136035, + "grad_norm": 1.2790040969848633, + "learning_rate": 1.995172968624296e-06, + "loss": 0.8773, + "step": 903 + }, + { + "epoch": 1.308248914616498, + "grad_norm": 1.442952275276184, + "learning_rate": 1.991150442477876e-06, + "loss": 0.8784, + "step": 904 + }, + { + "epoch": 1.3096960926193921, + "grad_norm": 1.5723949670791626, + "learning_rate": 1.9871279163314563e-06, + "loss": 0.9124, + "step": 905 + }, + { + "epoch": 1.3111432706222865, + "grad_norm": 1.4733463525772095, + "learning_rate": 1.9831053901850364e-06, + "loss": 0.9128, + "step": 906 + }, + { + "epoch": 1.312590448625181, + "grad_norm": 1.2380741834640503, + "learning_rate": 1.9790828640386166e-06, + "loss": 0.9003, + "step": 907 + }, + { + "epoch": 1.3140376266280753, + "grad_norm": 1.36652672290802, + "learning_rate": 1.9750603378921963e-06, + "loss": 0.8944, + "step": 908 + }, + { + "epoch": 1.3154848046309695, + "grad_norm": 1.341328501701355, + "learning_rate": 1.9710378117457764e-06, + "loss": 0.8829, + "step": 909 + }, + { + "epoch": 1.316931982633864, + "grad_norm": 1.2410205602645874, + "learning_rate": 1.9670152855993565e-06, + "loss": 0.8549, + "step": 910 + }, + { + "epoch": 1.316931982633864, + "eval_loss": 1.015825629234314, + "eval_runtime": 50.9066, + "eval_samples_per_second": 19.644, + "eval_steps_per_second": 1.238, + "step": 910 + }, + { + "epoch": 1.3183791606367583, + "grad_norm": 1.3675426244735718, + "learning_rate": 1.9629927594529367e-06, + "loss": 0.923, + "step": 911 + }, + { + "epoch": 1.3198263386396527, + "grad_norm": 1.2662525177001953, + "learning_rate": 1.9589702333065168e-06, + "loss": 0.8704, + "step": 912 + }, + { + "epoch": 1.321273516642547, + "grad_norm": 1.3591573238372803, + "learning_rate": 1.9549477071600965e-06, + "loss": 0.8966, + "step": 913 + }, + { + "epoch": 1.3227206946454415, + "grad_norm": 1.308894157409668, + "learning_rate": 1.9509251810136766e-06, + "loss": 0.9214, + "step": 914 + }, + { + "epoch": 1.3241678726483357, + "grad_norm": 1.2246655225753784, + "learning_rate": 1.9469026548672567e-06, + "loss": 0.9103, + "step": 915 + }, + { + "epoch": 1.32561505065123, + "grad_norm": 1.4983863830566406, + "learning_rate": 1.942880128720837e-06, + "loss": 0.8988, + "step": 916 + }, + { + "epoch": 1.3270622286541245, + "grad_norm": 1.228445053100586, + "learning_rate": 1.938857602574417e-06, + "loss": 0.8758, + "step": 917 + }, + { + "epoch": 1.3285094066570189, + "grad_norm": 1.2386879920959473, + "learning_rate": 1.9348350764279967e-06, + "loss": 0.8696, + "step": 918 + }, + { + "epoch": 1.329956584659913, + "grad_norm": 1.5190383195877075, + "learning_rate": 1.930812550281577e-06, + "loss": 0.9013, + "step": 919 + }, + { + "epoch": 1.3314037626628075, + "grad_norm": 1.2595216035842896, + "learning_rate": 1.926790024135157e-06, + "loss": 0.9133, + "step": 920 + }, + { + "epoch": 1.3314037626628075, + "eval_loss": 1.0158839225769043, + "eval_runtime": 50.9635, + "eval_samples_per_second": 19.622, + "eval_steps_per_second": 1.236, + "step": 920 + }, + { + "epoch": 1.3328509406657019, + "grad_norm": 1.2638113498687744, + "learning_rate": 1.922767497988737e-06, + "loss": 0.9188, + "step": 921 + }, + { + "epoch": 1.3342981186685963, + "grad_norm": 1.2872328758239746, + "learning_rate": 1.9187449718423172e-06, + "loss": 0.9057, + "step": 922 + }, + { + "epoch": 1.3357452966714907, + "grad_norm": 1.2601577043533325, + "learning_rate": 1.914722445695897e-06, + "loss": 0.8928, + "step": 923 + }, + { + "epoch": 1.337192474674385, + "grad_norm": 1.4166321754455566, + "learning_rate": 1.910699919549477e-06, + "loss": 0.8845, + "step": 924 + }, + { + "epoch": 1.3386396526772792, + "grad_norm": 1.3288692235946655, + "learning_rate": 1.9066773934030572e-06, + "loss": 0.8759, + "step": 925 + }, + { + "epoch": 1.3400868306801736, + "grad_norm": 1.3668838739395142, + "learning_rate": 1.9026548672566373e-06, + "loss": 0.8795, + "step": 926 + }, + { + "epoch": 1.341534008683068, + "grad_norm": 1.3145331144332886, + "learning_rate": 1.8986323411102172e-06, + "loss": 0.8808, + "step": 927 + }, + { + "epoch": 1.3429811866859624, + "grad_norm": 1.5038890838623047, + "learning_rate": 1.8946098149637973e-06, + "loss": 0.8653, + "step": 928 + }, + { + "epoch": 1.3444283646888566, + "grad_norm": 1.3025553226470947, + "learning_rate": 1.8905872888173773e-06, + "loss": 0.8668, + "step": 929 + }, + { + "epoch": 1.345875542691751, + "grad_norm": 1.4719102382659912, + "learning_rate": 1.8865647626709574e-06, + "loss": 0.889, + "step": 930 + }, + { + "epoch": 1.345875542691751, + "eval_loss": 1.0172665119171143, + "eval_runtime": 51.0462, + "eval_samples_per_second": 19.59, + "eval_steps_per_second": 1.234, + "step": 930 + }, + { + "epoch": 1.3473227206946454, + "grad_norm": 1.3738595247268677, + "learning_rate": 1.8825422365245375e-06, + "loss": 0.8761, + "step": 931 + }, + { + "epoch": 1.3487698986975398, + "grad_norm": 1.2960922718048096, + "learning_rate": 1.8785197103781174e-06, + "loss": 0.891, + "step": 932 + }, + { + "epoch": 1.3502170767004342, + "grad_norm": 1.167519450187683, + "learning_rate": 1.8744971842316978e-06, + "loss": 0.8709, + "step": 933 + }, + { + "epoch": 1.3516642547033286, + "grad_norm": 1.1963785886764526, + "learning_rate": 1.870474658085278e-06, + "loss": 0.8768, + "step": 934 + }, + { + "epoch": 1.3531114327062228, + "grad_norm": 1.5814266204833984, + "learning_rate": 1.8664521319388578e-06, + "loss": 0.9167, + "step": 935 + }, + { + "epoch": 1.3545586107091172, + "grad_norm": 1.2720822095870972, + "learning_rate": 1.862429605792438e-06, + "loss": 0.9015, + "step": 936 + }, + { + "epoch": 1.3560057887120116, + "grad_norm": 1.3009334802627563, + "learning_rate": 1.8584070796460179e-06, + "loss": 0.8928, + "step": 937 + }, + { + "epoch": 1.357452966714906, + "grad_norm": 1.4435724020004272, + "learning_rate": 1.854384553499598e-06, + "loss": 0.863, + "step": 938 + }, + { + "epoch": 1.3589001447178002, + "grad_norm": 1.397314190864563, + "learning_rate": 1.8503620273531781e-06, + "loss": 0.8794, + "step": 939 + }, + { + "epoch": 1.3603473227206946, + "grad_norm": 1.3401625156402588, + "learning_rate": 1.846339501206758e-06, + "loss": 0.8747, + "step": 940 + }, + { + "epoch": 1.3603473227206946, + "eval_loss": 1.0189106464385986, + "eval_runtime": 51.0333, + "eval_samples_per_second": 19.595, + "eval_steps_per_second": 1.234, + "step": 940 + }, + { + "epoch": 1.361794500723589, + "grad_norm": 1.2408407926559448, + "learning_rate": 1.8423169750603382e-06, + "loss": 0.8973, + "step": 941 + }, + { + "epoch": 1.3632416787264834, + "grad_norm": 1.3709076642990112, + "learning_rate": 1.838294448913918e-06, + "loss": 0.8823, + "step": 942 + }, + { + "epoch": 1.3646888567293778, + "grad_norm": 1.4051848649978638, + "learning_rate": 1.8342719227674982e-06, + "loss": 0.8908, + "step": 943 + }, + { + "epoch": 1.3661360347322722, + "grad_norm": 1.22672700881958, + "learning_rate": 1.8302493966210783e-06, + "loss": 0.8803, + "step": 944 + }, + { + "epoch": 1.3675832127351664, + "grad_norm": 1.5199220180511475, + "learning_rate": 1.8262268704746582e-06, + "loss": 0.8757, + "step": 945 + }, + { + "epoch": 1.3690303907380608, + "grad_norm": 1.3716022968292236, + "learning_rate": 1.8222043443282384e-06, + "loss": 0.8937, + "step": 946 + }, + { + "epoch": 1.3704775687409552, + "grad_norm": 1.3780746459960938, + "learning_rate": 1.8181818181818183e-06, + "loss": 0.9002, + "step": 947 + }, + { + "epoch": 1.3719247467438496, + "grad_norm": 1.414121150970459, + "learning_rate": 1.8141592920353984e-06, + "loss": 0.9163, + "step": 948 + }, + { + "epoch": 1.3733719247467437, + "grad_norm": 1.2676604986190796, + "learning_rate": 1.8101367658889785e-06, + "loss": 0.8928, + "step": 949 + }, + { + "epoch": 1.3748191027496381, + "grad_norm": 1.3467508554458618, + "learning_rate": 1.8061142397425585e-06, + "loss": 0.8443, + "step": 950 + }, + { + "epoch": 1.3748191027496381, + "eval_loss": 1.015096664428711, + "eval_runtime": 51.0598, + "eval_samples_per_second": 19.585, + "eval_steps_per_second": 1.234, + "step": 950 + }, + { + "epoch": 1.3762662807525325, + "grad_norm": 1.4137110710144043, + "learning_rate": 1.8020917135961386e-06, + "loss": 0.8419, + "step": 951 + }, + { + "epoch": 1.377713458755427, + "grad_norm": 1.1944453716278076, + "learning_rate": 1.7980691874497185e-06, + "loss": 0.8634, + "step": 952 + }, + { + "epoch": 1.3791606367583213, + "grad_norm": 1.18637216091156, + "learning_rate": 1.7940466613032986e-06, + "loss": 0.8986, + "step": 953 + }, + { + "epoch": 1.3806078147612157, + "grad_norm": 1.3139936923980713, + "learning_rate": 1.7900241351568788e-06, + "loss": 0.8573, + "step": 954 + }, + { + "epoch": 1.38205499276411, + "grad_norm": 1.2180980443954468, + "learning_rate": 1.7860016090104587e-06, + "loss": 0.8634, + "step": 955 + }, + { + "epoch": 1.3835021707670043, + "grad_norm": 1.1703729629516602, + "learning_rate": 1.7819790828640388e-06, + "loss": 0.8807, + "step": 956 + }, + { + "epoch": 1.3849493487698987, + "grad_norm": 1.226999044418335, + "learning_rate": 1.7779565567176187e-06, + "loss": 0.8674, + "step": 957 + }, + { + "epoch": 1.3863965267727931, + "grad_norm": 1.3507004976272583, + "learning_rate": 1.7739340305711988e-06, + "loss": 0.8936, + "step": 958 + }, + { + "epoch": 1.3878437047756873, + "grad_norm": 1.2763739824295044, + "learning_rate": 1.769911504424779e-06, + "loss": 0.8797, + "step": 959 + }, + { + "epoch": 1.3892908827785817, + "grad_norm": 1.258967399597168, + "learning_rate": 1.7658889782783589e-06, + "loss": 0.9074, + "step": 960 + }, + { + "epoch": 1.3892908827785817, + "eval_loss": 1.0130211114883423, + "eval_runtime": 50.9335, + "eval_samples_per_second": 19.633, + "eval_steps_per_second": 1.237, + "step": 960 + }, + { + "epoch": 1.390738060781476, + "grad_norm": 1.2773218154907227, + "learning_rate": 1.761866452131939e-06, + "loss": 0.9065, + "step": 961 + }, + { + "epoch": 1.3921852387843705, + "grad_norm": 1.2194263935089111, + "learning_rate": 1.757843925985519e-06, + "loss": 0.8755, + "step": 962 + }, + { + "epoch": 1.393632416787265, + "grad_norm": 1.2352688312530518, + "learning_rate": 1.753821399839099e-06, + "loss": 0.9011, + "step": 963 + }, + { + "epoch": 1.3950795947901593, + "grad_norm": 1.2156544923782349, + "learning_rate": 1.7497988736926792e-06, + "loss": 0.9192, + "step": 964 + }, + { + "epoch": 1.3965267727930535, + "grad_norm": 1.279953122138977, + "learning_rate": 1.7457763475462591e-06, + "loss": 0.8963, + "step": 965 + }, + { + "epoch": 1.3979739507959479, + "grad_norm": 1.4173139333724976, + "learning_rate": 1.7417538213998392e-06, + "loss": 0.8955, + "step": 966 + }, + { + "epoch": 1.3994211287988423, + "grad_norm": 1.3765140771865845, + "learning_rate": 1.7377312952534192e-06, + "loss": 0.8824, + "step": 967 + }, + { + "epoch": 1.4008683068017367, + "grad_norm": 1.3438305854797363, + "learning_rate": 1.7337087691069993e-06, + "loss": 0.838, + "step": 968 + }, + { + "epoch": 1.4023154848046309, + "grad_norm": 1.2831956148147583, + "learning_rate": 1.7296862429605792e-06, + "loss": 0.8685, + "step": 969 + }, + { + "epoch": 1.4037626628075253, + "grad_norm": 1.3667343854904175, + "learning_rate": 1.7256637168141593e-06, + "loss": 0.9165, + "step": 970 + }, + { + "epoch": 1.4037626628075253, + "eval_loss": 1.011406421661377, + "eval_runtime": 51.1361, + "eval_samples_per_second": 19.556, + "eval_steps_per_second": 1.232, + "step": 970 + }, + { + "epoch": 1.4052098408104197, + "grad_norm": 1.2441140413284302, + "learning_rate": 1.7216411906677395e-06, + "loss": 0.878, + "step": 971 + }, + { + "epoch": 1.406657018813314, + "grad_norm": 1.2812925577163696, + "learning_rate": 1.7176186645213194e-06, + "loss": 0.8524, + "step": 972 + }, + { + "epoch": 1.4081041968162085, + "grad_norm": 1.4451730251312256, + "learning_rate": 1.7135961383748995e-06, + "loss": 0.8851, + "step": 973 + }, + { + "epoch": 1.4095513748191029, + "grad_norm": 1.387168288230896, + "learning_rate": 1.7095736122284794e-06, + "loss": 0.9299, + "step": 974 + }, + { + "epoch": 1.410998552821997, + "grad_norm": 1.359509825706482, + "learning_rate": 1.7055510860820595e-06, + "loss": 0.8729, + "step": 975 + }, + { + "epoch": 1.4124457308248914, + "grad_norm": 1.374069094657898, + "learning_rate": 1.7015285599356397e-06, + "loss": 0.9063, + "step": 976 + }, + { + "epoch": 1.4138929088277858, + "grad_norm": 1.2204853296279907, + "learning_rate": 1.6975060337892196e-06, + "loss": 0.8962, + "step": 977 + }, + { + "epoch": 1.4153400868306802, + "grad_norm": 1.2265360355377197, + "learning_rate": 1.6934835076428e-06, + "loss": 0.8977, + "step": 978 + }, + { + "epoch": 1.4167872648335744, + "grad_norm": 1.2486757040023804, + "learning_rate": 1.68946098149638e-06, + "loss": 0.8729, + "step": 979 + }, + { + "epoch": 1.4182344428364688, + "grad_norm": 1.2419143915176392, + "learning_rate": 1.68543845534996e-06, + "loss": 0.9203, + "step": 980 + }, + { + "epoch": 1.4182344428364688, + "eval_loss": 1.0111998319625854, + "eval_runtime": 50.9331, + "eval_samples_per_second": 19.634, + "eval_steps_per_second": 1.237, + "step": 980 + }, + { + "epoch": 1.4196816208393632, + "grad_norm": 1.435892105102539, + "learning_rate": 1.68141592920354e-06, + "loss": 0.8483, + "step": 981 + }, + { + "epoch": 1.4211287988422576, + "grad_norm": 1.1839962005615234, + "learning_rate": 1.6773934030571202e-06, + "loss": 0.872, + "step": 982 + }, + { + "epoch": 1.422575976845152, + "grad_norm": 1.3123773336410522, + "learning_rate": 1.6733708769107001e-06, + "loss": 0.8623, + "step": 983 + }, + { + "epoch": 1.4240231548480464, + "grad_norm": 1.3668432235717773, + "learning_rate": 1.6693483507642803e-06, + "loss": 0.9127, + "step": 984 + }, + { + "epoch": 1.4254703328509406, + "grad_norm": 1.2852669954299927, + "learning_rate": 1.6653258246178602e-06, + "loss": 0.8654, + "step": 985 + }, + { + "epoch": 1.426917510853835, + "grad_norm": 1.2866759300231934, + "learning_rate": 1.6613032984714403e-06, + "loss": 0.8925, + "step": 986 + }, + { + "epoch": 1.4283646888567294, + "grad_norm": 1.5316014289855957, + "learning_rate": 1.6572807723250202e-06, + "loss": 0.8906, + "step": 987 + }, + { + "epoch": 1.4298118668596238, + "grad_norm": 1.423528790473938, + "learning_rate": 1.6532582461786004e-06, + "loss": 0.8913, + "step": 988 + }, + { + "epoch": 1.431259044862518, + "grad_norm": 1.3927950859069824, + "learning_rate": 1.6492357200321805e-06, + "loss": 0.8746, + "step": 989 + }, + { + "epoch": 1.4327062228654124, + "grad_norm": 1.4410115480422974, + "learning_rate": 1.6452131938857604e-06, + "loss": 0.929, + "step": 990 + }, + { + "epoch": 1.4327062228654124, + "eval_loss": 1.0117580890655518, + "eval_runtime": 51.1077, + "eval_samples_per_second": 19.567, + "eval_steps_per_second": 1.233, + "step": 990 + }, + { + "epoch": 1.4341534008683068, + "grad_norm": 1.4615365266799927, + "learning_rate": 1.6411906677393405e-06, + "loss": 0.8728, + "step": 991 + }, + { + "epoch": 1.4356005788712012, + "grad_norm": 1.3534507751464844, + "learning_rate": 1.6371681415929204e-06, + "loss": 0.8816, + "step": 992 + }, + { + "epoch": 1.4370477568740956, + "grad_norm": 1.2888727188110352, + "learning_rate": 1.6331456154465006e-06, + "loss": 0.8929, + "step": 993 + }, + { + "epoch": 1.43849493487699, + "grad_norm": 1.2917914390563965, + "learning_rate": 1.6291230893000807e-06, + "loss": 0.8993, + "step": 994 + }, + { + "epoch": 1.4399421128798842, + "grad_norm": 1.2315443754196167, + "learning_rate": 1.6251005631536606e-06, + "loss": 0.9099, + "step": 995 + }, + { + "epoch": 1.4413892908827786, + "grad_norm": 1.3355698585510254, + "learning_rate": 1.6210780370072407e-06, + "loss": 0.9193, + "step": 996 + }, + { + "epoch": 1.442836468885673, + "grad_norm": 1.256848931312561, + "learning_rate": 1.6170555108608207e-06, + "loss": 0.8657, + "step": 997 + }, + { + "epoch": 1.4442836468885674, + "grad_norm": 1.2647980451583862, + "learning_rate": 1.6130329847144008e-06, + "loss": 0.8399, + "step": 998 + }, + { + "epoch": 1.4457308248914615, + "grad_norm": 1.1816374063491821, + "learning_rate": 1.609010458567981e-06, + "loss": 0.8896, + "step": 999 + }, + { + "epoch": 1.447178002894356, + "grad_norm": 1.2491004467010498, + "learning_rate": 1.6049879324215608e-06, + "loss": 0.8849, + "step": 1000 + }, + { + "epoch": 1.447178002894356, + "eval_loss": 1.01273512840271, + "eval_runtime": 50.9324, + "eval_samples_per_second": 19.634, + "eval_steps_per_second": 1.237, + "step": 1000 + }, + { + "epoch": 1.4486251808972503, + "grad_norm": 1.291909098625183, + "learning_rate": 1.600965406275141e-06, + "loss": 0.8433, + "step": 1001 + }, + { + "epoch": 1.4500723589001447, + "grad_norm": 1.398860216140747, + "learning_rate": 1.5969428801287209e-06, + "loss": 0.8882, + "step": 1002 + }, + { + "epoch": 1.4515195369030391, + "grad_norm": 1.3371323347091675, + "learning_rate": 1.592920353982301e-06, + "loss": 0.8715, + "step": 1003 + }, + { + "epoch": 1.4529667149059335, + "grad_norm": 1.5092949867248535, + "learning_rate": 1.5888978278358811e-06, + "loss": 0.9202, + "step": 1004 + }, + { + "epoch": 1.4544138929088277, + "grad_norm": 1.4176650047302246, + "learning_rate": 1.584875301689461e-06, + "loss": 0.8724, + "step": 1005 + }, + { + "epoch": 1.4558610709117221, + "grad_norm": 1.4517333507537842, + "learning_rate": 1.5808527755430412e-06, + "loss": 0.8975, + "step": 1006 + }, + { + "epoch": 1.4573082489146165, + "grad_norm": 1.385549545288086, + "learning_rate": 1.576830249396621e-06, + "loss": 0.8402, + "step": 1007 + }, + { + "epoch": 1.458755426917511, + "grad_norm": 1.2773293256759644, + "learning_rate": 1.5728077232502012e-06, + "loss": 0.9039, + "step": 1008 + }, + { + "epoch": 1.460202604920405, + "grad_norm": 1.4182472229003906, + "learning_rate": 1.5687851971037813e-06, + "loss": 0.8526, + "step": 1009 + }, + { + "epoch": 1.4616497829232995, + "grad_norm": 1.3395445346832275, + "learning_rate": 1.5647626709573613e-06, + "loss": 0.8842, + "step": 1010 + }, + { + "epoch": 1.4616497829232995, + "eval_loss": 1.01534903049469, + "eval_runtime": 51.1549, + "eval_samples_per_second": 19.548, + "eval_steps_per_second": 1.232, + "step": 1010 + }, + { + "epoch": 1.463096960926194, + "grad_norm": 1.3416283130645752, + "learning_rate": 1.5607401448109414e-06, + "loss": 0.8977, + "step": 1011 + }, + { + "epoch": 1.4645441389290883, + "grad_norm": 1.397812008857727, + "learning_rate": 1.5567176186645213e-06, + "loss": 0.8712, + "step": 1012 + }, + { + "epoch": 1.4659913169319827, + "grad_norm": 1.2280248403549194, + "learning_rate": 1.5526950925181014e-06, + "loss": 0.8935, + "step": 1013 + }, + { + "epoch": 1.467438494934877, + "grad_norm": 1.232506513595581, + "learning_rate": 1.5486725663716816e-06, + "loss": 0.8676, + "step": 1014 + }, + { + "epoch": 1.4688856729377713, + "grad_norm": 1.3824166059494019, + "learning_rate": 1.5446500402252615e-06, + "loss": 0.8805, + "step": 1015 + }, + { + "epoch": 1.4703328509406657, + "grad_norm": 1.3635704517364502, + "learning_rate": 1.5406275140788416e-06, + "loss": 0.912, + "step": 1016 + }, + { + "epoch": 1.47178002894356, + "grad_norm": 1.459098219871521, + "learning_rate": 1.5366049879324215e-06, + "loss": 0.9142, + "step": 1017 + }, + { + "epoch": 1.4732272069464545, + "grad_norm": 1.1681647300720215, + "learning_rate": 1.5325824617860016e-06, + "loss": 0.8562, + "step": 1018 + }, + { + "epoch": 1.4746743849493487, + "grad_norm": 1.4085551500320435, + "learning_rate": 1.5285599356395816e-06, + "loss": 0.8414, + "step": 1019 + }, + { + "epoch": 1.476121562952243, + "grad_norm": 1.343843698501587, + "learning_rate": 1.5245374094931617e-06, + "loss": 0.9363, + "step": 1020 + }, + { + "epoch": 1.476121562952243, + "eval_loss": 1.0108908414840698, + "eval_runtime": 51.2826, + "eval_samples_per_second": 19.5, + "eval_steps_per_second": 1.228, + "step": 1020 + }, + { + "epoch": 1.4775687409551375, + "grad_norm": 1.4455372095108032, + "learning_rate": 1.5205148833467418e-06, + "loss": 0.8803, + "step": 1021 + }, + { + "epoch": 1.4790159189580319, + "grad_norm": 1.3939288854599, + "learning_rate": 1.5164923572003221e-06, + "loss": 0.8962, + "step": 1022 + }, + { + "epoch": 1.4804630969609263, + "grad_norm": 1.3528865575790405, + "learning_rate": 1.512469831053902e-06, + "loss": 0.8779, + "step": 1023 + }, + { + "epoch": 1.4819102749638207, + "grad_norm": 1.2597918510437012, + "learning_rate": 1.5084473049074822e-06, + "loss": 0.9279, + "step": 1024 + }, + { + "epoch": 1.4833574529667148, + "grad_norm": 1.1785560846328735, + "learning_rate": 1.5044247787610621e-06, + "loss": 0.8705, + "step": 1025 + }, + { + "epoch": 1.4848046309696092, + "grad_norm": 1.330184817314148, + "learning_rate": 1.5004022526146422e-06, + "loss": 0.8776, + "step": 1026 + }, + { + "epoch": 1.4862518089725036, + "grad_norm": 1.4158875942230225, + "learning_rate": 1.4963797264682224e-06, + "loss": 0.8955, + "step": 1027 + }, + { + "epoch": 1.487698986975398, + "grad_norm": 1.3405498266220093, + "learning_rate": 1.4923572003218023e-06, + "loss": 0.8821, + "step": 1028 + }, + { + "epoch": 1.4891461649782922, + "grad_norm": 1.2876495122909546, + "learning_rate": 1.4883346741753824e-06, + "loss": 0.866, + "step": 1029 + }, + { + "epoch": 1.4905933429811866, + "grad_norm": 1.4403177499771118, + "learning_rate": 1.4843121480289623e-06, + "loss": 0.8674, + "step": 1030 + }, + { + "epoch": 1.4905933429811866, + "eval_loss": 1.0085880756378174, + "eval_runtime": 51.1661, + "eval_samples_per_second": 19.544, + "eval_steps_per_second": 1.231, + "step": 1030 + }, + { + "epoch": 1.492040520984081, + "grad_norm": 1.304076910018921, + "learning_rate": 1.4802896218825425e-06, + "loss": 0.9198, + "step": 1031 + }, + { + "epoch": 1.4934876989869754, + "grad_norm": 1.3198641538619995, + "learning_rate": 1.4762670957361226e-06, + "loss": 0.878, + "step": 1032 + }, + { + "epoch": 1.4949348769898698, + "grad_norm": 1.2858214378356934, + "learning_rate": 1.4722445695897025e-06, + "loss": 0.8968, + "step": 1033 + }, + { + "epoch": 1.4963820549927642, + "grad_norm": 1.3212248086929321, + "learning_rate": 1.4682220434432826e-06, + "loss": 0.863, + "step": 1034 + }, + { + "epoch": 1.4978292329956584, + "grad_norm": 1.351426124572754, + "learning_rate": 1.4641995172968625e-06, + "loss": 0.8923, + "step": 1035 + }, + { + "epoch": 1.4992764109985528, + "grad_norm": 1.227321743965149, + "learning_rate": 1.4601769911504427e-06, + "loss": 0.8619, + "step": 1036 + }, + { + "epoch": 1.5007235890014472, + "grad_norm": 1.390210509300232, + "learning_rate": 1.4561544650040226e-06, + "loss": 0.8937, + "step": 1037 + }, + { + "epoch": 1.5021707670043414, + "grad_norm": 1.3069322109222412, + "learning_rate": 1.4521319388576027e-06, + "loss": 0.8526, + "step": 1038 + }, + { + "epoch": 1.5036179450072358, + "grad_norm": 1.2899137735366821, + "learning_rate": 1.4481094127111828e-06, + "loss": 0.8707, + "step": 1039 + }, + { + "epoch": 1.5050651230101302, + "grad_norm": 1.2352845668792725, + "learning_rate": 1.4440868865647628e-06, + "loss": 0.8868, + "step": 1040 + }, + { + "epoch": 1.5050651230101302, + "eval_loss": 1.0064990520477295, + "eval_runtime": 51.255, + "eval_samples_per_second": 19.51, + "eval_steps_per_second": 1.229, + "step": 1040 + }, + { + "epoch": 1.5065123010130246, + "grad_norm": 1.510420322418213, + "learning_rate": 1.4400643604183429e-06, + "loss": 0.9012, + "step": 1041 + }, + { + "epoch": 1.507959479015919, + "grad_norm": 1.4931812286376953, + "learning_rate": 1.4360418342719228e-06, + "loss": 0.8866, + "step": 1042 + }, + { + "epoch": 1.5094066570188134, + "grad_norm": 1.3143318891525269, + "learning_rate": 1.432019308125503e-06, + "loss": 0.8842, + "step": 1043 + }, + { + "epoch": 1.5108538350217078, + "grad_norm": 1.41519296169281, + "learning_rate": 1.427996781979083e-06, + "loss": 0.892, + "step": 1044 + }, + { + "epoch": 1.5123010130246022, + "grad_norm": 1.2534558773040771, + "learning_rate": 1.423974255832663e-06, + "loss": 0.8936, + "step": 1045 + }, + { + "epoch": 1.5137481910274964, + "grad_norm": 1.2007780075073242, + "learning_rate": 1.419951729686243e-06, + "loss": 0.8742, + "step": 1046 + }, + { + "epoch": 1.5151953690303908, + "grad_norm": 1.461944818496704, + "learning_rate": 1.415929203539823e-06, + "loss": 0.8526, + "step": 1047 + }, + { + "epoch": 1.516642547033285, + "grad_norm": 1.4859479665756226, + "learning_rate": 1.4119066773934031e-06, + "loss": 0.8957, + "step": 1048 + }, + { + "epoch": 1.5180897250361793, + "grad_norm": 1.3720206022262573, + "learning_rate": 1.4078841512469833e-06, + "loss": 0.9112, + "step": 1049 + }, + { + "epoch": 1.5195369030390737, + "grad_norm": 1.3323490619659424, + "learning_rate": 1.4038616251005632e-06, + "loss": 0.8783, + "step": 1050 + }, + { + "epoch": 1.5195369030390737, + "eval_loss": 1.004501461982727, + "eval_runtime": 51.041, + "eval_samples_per_second": 19.592, + "eval_steps_per_second": 1.234, + "step": 1050 + }, + { + "epoch": 1.5209840810419681, + "grad_norm": 1.294517159461975, + "learning_rate": 1.3998390989541433e-06, + "loss": 0.8644, + "step": 1051 + }, + { + "epoch": 1.5224312590448625, + "grad_norm": 1.3740386962890625, + "learning_rate": 1.3958165728077232e-06, + "loss": 0.8915, + "step": 1052 + }, + { + "epoch": 1.523878437047757, + "grad_norm": 1.346398115158081, + "learning_rate": 1.3917940466613034e-06, + "loss": 0.9053, + "step": 1053 + }, + { + "epoch": 1.5253256150506513, + "grad_norm": 1.4877651929855347, + "learning_rate": 1.3877715205148835e-06, + "loss": 0.9102, + "step": 1054 + }, + { + "epoch": 1.5267727930535457, + "grad_norm": 1.4911222457885742, + "learning_rate": 1.3837489943684634e-06, + "loss": 0.8854, + "step": 1055 + }, + { + "epoch": 1.52821997105644, + "grad_norm": 1.2816040515899658, + "learning_rate": 1.3797264682220435e-06, + "loss": 0.885, + "step": 1056 + }, + { + "epoch": 1.5296671490593343, + "grad_norm": 1.2152899503707886, + "learning_rate": 1.3757039420756234e-06, + "loss": 0.863, + "step": 1057 + }, + { + "epoch": 1.5311143270622285, + "grad_norm": 1.3935914039611816, + "learning_rate": 1.3716814159292036e-06, + "loss": 0.8626, + "step": 1058 + }, + { + "epoch": 1.532561505065123, + "grad_norm": 1.3137140274047852, + "learning_rate": 1.3676588897827837e-06, + "loss": 0.8747, + "step": 1059 + }, + { + "epoch": 1.5340086830680173, + "grad_norm": 1.3641102313995361, + "learning_rate": 1.3636363636363636e-06, + "loss": 0.8728, + "step": 1060 + }, + { + "epoch": 1.5340086830680173, + "eval_loss": 1.0042489767074585, + "eval_runtime": 51.2861, + "eval_samples_per_second": 19.498, + "eval_steps_per_second": 1.228, + "step": 1060 + }, + { + "epoch": 1.5354558610709117, + "grad_norm": 1.3060004711151123, + "learning_rate": 1.3596138374899437e-06, + "loss": 0.8836, + "step": 1061 + }, + { + "epoch": 1.536903039073806, + "grad_norm": 1.180417537689209, + "learning_rate": 1.3555913113435237e-06, + "loss": 0.8941, + "step": 1062 + }, + { + "epoch": 1.5383502170767005, + "grad_norm": 1.2206364870071411, + "learning_rate": 1.3515687851971038e-06, + "loss": 0.8685, + "step": 1063 + }, + { + "epoch": 1.539797395079595, + "grad_norm": 1.3248261213302612, + "learning_rate": 1.347546259050684e-06, + "loss": 0.8743, + "step": 1064 + }, + { + "epoch": 1.5412445730824893, + "grad_norm": 1.4073798656463623, + "learning_rate": 1.3435237329042638e-06, + "loss": 0.878, + "step": 1065 + }, + { + "epoch": 1.5426917510853835, + "grad_norm": 1.4199151992797852, + "learning_rate": 1.339501206757844e-06, + "loss": 0.9022, + "step": 1066 + }, + { + "epoch": 1.5441389290882779, + "grad_norm": 1.2258391380310059, + "learning_rate": 1.3354786806114243e-06, + "loss": 0.8723, + "step": 1067 + }, + { + "epoch": 1.545586107091172, + "grad_norm": 1.45433509349823, + "learning_rate": 1.3314561544650042e-06, + "loss": 0.8901, + "step": 1068 + }, + { + "epoch": 1.5470332850940665, + "grad_norm": 1.288647174835205, + "learning_rate": 1.3274336283185843e-06, + "loss": 0.8771, + "step": 1069 + }, + { + "epoch": 1.5484804630969609, + "grad_norm": 1.4006913900375366, + "learning_rate": 1.3234111021721643e-06, + "loss": 0.8951, + "step": 1070 + }, + { + "epoch": 1.5484804630969609, + "eval_loss": 1.0051230192184448, + "eval_runtime": 51.406, + "eval_samples_per_second": 19.453, + "eval_steps_per_second": 1.226, + "step": 1070 + }, + { + "epoch": 1.5499276410998553, + "grad_norm": 1.487797737121582, + "learning_rate": 1.3193885760257444e-06, + "loss": 0.8981, + "step": 1071 + }, + { + "epoch": 1.5513748191027497, + "grad_norm": 1.2707769870758057, + "learning_rate": 1.3153660498793245e-06, + "loss": 0.8628, + "step": 1072 + }, + { + "epoch": 1.552821997105644, + "grad_norm": 1.37482750415802, + "learning_rate": 1.3113435237329044e-06, + "loss": 0.8722, + "step": 1073 + }, + { + "epoch": 1.5542691751085385, + "grad_norm": 1.260653018951416, + "learning_rate": 1.3073209975864846e-06, + "loss": 0.8872, + "step": 1074 + }, + { + "epoch": 1.5557163531114329, + "grad_norm": 1.2580828666687012, + "learning_rate": 1.3032984714400645e-06, + "loss": 0.8835, + "step": 1075 + }, + { + "epoch": 1.557163531114327, + "grad_norm": 1.2777159214019775, + "learning_rate": 1.2992759452936446e-06, + "loss": 0.8602, + "step": 1076 + }, + { + "epoch": 1.5586107091172214, + "grad_norm": 1.5712080001831055, + "learning_rate": 1.2952534191472247e-06, + "loss": 0.8439, + "step": 1077 + }, + { + "epoch": 1.5600578871201156, + "grad_norm": 1.4249807596206665, + "learning_rate": 1.2912308930008046e-06, + "loss": 0.8928, + "step": 1078 + }, + { + "epoch": 1.56150506512301, + "grad_norm": 1.4448374509811401, + "learning_rate": 1.2872083668543848e-06, + "loss": 0.8459, + "step": 1079 + }, + { + "epoch": 1.5629522431259044, + "grad_norm": 1.3210469484329224, + "learning_rate": 1.2831858407079647e-06, + "loss": 0.8522, + "step": 1080 + }, + { + "epoch": 1.5629522431259044, + "eval_loss": 1.0064622163772583, + "eval_runtime": 51.239, + "eval_samples_per_second": 19.516, + "eval_steps_per_second": 1.23, + "step": 1080 + }, + { + "epoch": 1.5643994211287988, + "grad_norm": 1.2785742282867432, + "learning_rate": 1.2791633145615448e-06, + "loss": 0.9148, + "step": 1081 + }, + { + "epoch": 1.5658465991316932, + "grad_norm": 1.3036953210830688, + "learning_rate": 1.275140788415125e-06, + "loss": 0.8843, + "step": 1082 + }, + { + "epoch": 1.5672937771345876, + "grad_norm": 1.395387887954712, + "learning_rate": 1.2711182622687049e-06, + "loss": 0.8865, + "step": 1083 + }, + { + "epoch": 1.568740955137482, + "grad_norm": 1.2966110706329346, + "learning_rate": 1.267095736122285e-06, + "loss": 0.903, + "step": 1084 + }, + { + "epoch": 1.5701881331403764, + "grad_norm": 1.26325261592865, + "learning_rate": 1.263073209975865e-06, + "loss": 0.8558, + "step": 1085 + }, + { + "epoch": 1.5716353111432706, + "grad_norm": 1.1640148162841797, + "learning_rate": 1.259050683829445e-06, + "loss": 0.8694, + "step": 1086 + }, + { + "epoch": 1.573082489146165, + "grad_norm": 1.1815216541290283, + "learning_rate": 1.2550281576830252e-06, + "loss": 0.8763, + "step": 1087 + }, + { + "epoch": 1.5745296671490592, + "grad_norm": 1.3960554599761963, + "learning_rate": 1.251005631536605e-06, + "loss": 0.8926, + "step": 1088 + }, + { + "epoch": 1.5759768451519536, + "grad_norm": 1.3517417907714844, + "learning_rate": 1.2469831053901852e-06, + "loss": 0.9278, + "step": 1089 + }, + { + "epoch": 1.577424023154848, + "grad_norm": 1.5481010675430298, + "learning_rate": 1.2429605792437651e-06, + "loss": 0.8816, + "step": 1090 + }, + { + "epoch": 1.577424023154848, + "eval_loss": 1.0066745281219482, + "eval_runtime": 51.7065, + "eval_samples_per_second": 19.34, + "eval_steps_per_second": 1.218, + "step": 1090 + }, + { + "epoch": 1.5788712011577424, + "grad_norm": 1.3831214904785156, + "learning_rate": 1.2389380530973452e-06, + "loss": 0.8805, + "step": 1091 + }, + { + "epoch": 1.5803183791606368, + "grad_norm": 1.2321022748947144, + "learning_rate": 1.2349155269509252e-06, + "loss": 0.8705, + "step": 1092 + }, + { + "epoch": 1.5817655571635312, + "grad_norm": 1.3171721696853638, + "learning_rate": 1.2308930008045053e-06, + "loss": 0.8658, + "step": 1093 + }, + { + "epoch": 1.5832127351664256, + "grad_norm": 1.4215795993804932, + "learning_rate": 1.2268704746580854e-06, + "loss": 0.8791, + "step": 1094 + }, + { + "epoch": 1.58465991316932, + "grad_norm": 1.2681177854537964, + "learning_rate": 1.2228479485116653e-06, + "loss": 0.9118, + "step": 1095 + }, + { + "epoch": 1.5861070911722142, + "grad_norm": 1.3912934064865112, + "learning_rate": 1.2188254223652455e-06, + "loss": 0.8654, + "step": 1096 + }, + { + "epoch": 1.5875542691751086, + "grad_norm": 1.2615245580673218, + "learning_rate": 1.2148028962188254e-06, + "loss": 0.8376, + "step": 1097 + }, + { + "epoch": 1.5890014471780027, + "grad_norm": 1.2494381666183472, + "learning_rate": 1.2107803700724055e-06, + "loss": 0.8593, + "step": 1098 + }, + { + "epoch": 1.5904486251808971, + "grad_norm": 1.4802052974700928, + "learning_rate": 1.2067578439259856e-06, + "loss": 0.869, + "step": 1099 + }, + { + "epoch": 1.5918958031837915, + "grad_norm": 1.345692753791809, + "learning_rate": 1.2027353177795658e-06, + "loss": 0.8809, + "step": 1100 + }, + { + "epoch": 1.5918958031837915, + "eval_loss": 1.0055845975875854, + "eval_runtime": 51.4568, + "eval_samples_per_second": 19.434, + "eval_steps_per_second": 1.224, + "step": 1100 + }, + { + "epoch": 1.593342981186686, + "grad_norm": 1.3866039514541626, + "learning_rate": 1.1987127916331457e-06, + "loss": 0.8952, + "step": 1101 + }, + { + "epoch": 1.5947901591895803, + "grad_norm": 1.2434381246566772, + "learning_rate": 1.1946902654867258e-06, + "loss": 0.8627, + "step": 1102 + }, + { + "epoch": 1.5962373371924747, + "grad_norm": 1.1858441829681396, + "learning_rate": 1.190667739340306e-06, + "loss": 0.8854, + "step": 1103 + }, + { + "epoch": 1.5976845151953691, + "grad_norm": 1.3802675008773804, + "learning_rate": 1.1866452131938858e-06, + "loss": 0.8595, + "step": 1104 + }, + { + "epoch": 1.5991316931982635, + "grad_norm": 1.3447083234786987, + "learning_rate": 1.182622687047466e-06, + "loss": 0.8786, + "step": 1105 + }, + { + "epoch": 1.6005788712011577, + "grad_norm": 1.199914813041687, + "learning_rate": 1.1786001609010459e-06, + "loss": 0.8649, + "step": 1106 + }, + { + "epoch": 1.6020260492040521, + "grad_norm": 1.234679937362671, + "learning_rate": 1.174577634754626e-06, + "loss": 0.8723, + "step": 1107 + }, + { + "epoch": 1.6034732272069463, + "grad_norm": 1.3795456886291504, + "learning_rate": 1.1705551086082061e-06, + "loss": 0.8527, + "step": 1108 + }, + { + "epoch": 1.6049204052098407, + "grad_norm": 1.4917933940887451, + "learning_rate": 1.166532582461786e-06, + "loss": 0.8703, + "step": 1109 + }, + { + "epoch": 1.606367583212735, + "grad_norm": 1.4692387580871582, + "learning_rate": 1.1625100563153662e-06, + "loss": 0.8901, + "step": 1110 + }, + { + "epoch": 1.606367583212735, + "eval_loss": 1.0043307542800903, + "eval_runtime": 51.2535, + "eval_samples_per_second": 19.511, + "eval_steps_per_second": 1.229, + "step": 1110 + }, + { + "epoch": 1.6078147612156295, + "grad_norm": 1.46327805519104, + "learning_rate": 1.158487530168946e-06, + "loss": 0.8776, + "step": 1111 + }, + { + "epoch": 1.609261939218524, + "grad_norm": 1.2555711269378662, + "learning_rate": 1.1544650040225262e-06, + "loss": 0.9015, + "step": 1112 + }, + { + "epoch": 1.6107091172214183, + "grad_norm": 1.5495578050613403, + "learning_rate": 1.1504424778761064e-06, + "loss": 0.8924, + "step": 1113 + }, + { + "epoch": 1.6121562952243127, + "grad_norm": 1.3383797407150269, + "learning_rate": 1.1464199517296863e-06, + "loss": 0.873, + "step": 1114 + }, + { + "epoch": 1.613603473227207, + "grad_norm": 1.517553687095642, + "learning_rate": 1.1423974255832664e-06, + "loss": 0.8696, + "step": 1115 + }, + { + "epoch": 1.6150506512301013, + "grad_norm": 1.3035321235656738, + "learning_rate": 1.1383748994368463e-06, + "loss": 0.9168, + "step": 1116 + }, + { + "epoch": 1.6164978292329957, + "grad_norm": 1.3796104192733765, + "learning_rate": 1.1343523732904264e-06, + "loss": 0.8848, + "step": 1117 + }, + { + "epoch": 1.6179450072358899, + "grad_norm": 1.5296746492385864, + "learning_rate": 1.1303298471440066e-06, + "loss": 0.8623, + "step": 1118 + }, + { + "epoch": 1.6193921852387843, + "grad_norm": 1.3489155769348145, + "learning_rate": 1.1263073209975865e-06, + "loss": 0.9029, + "step": 1119 + }, + { + "epoch": 1.6208393632416787, + "grad_norm": 1.1626771688461304, + "learning_rate": 1.1222847948511666e-06, + "loss": 0.8905, + "step": 1120 + }, + { + "epoch": 1.6208393632416787, + "eval_loss": 1.0036776065826416, + "eval_runtime": 51.1614, + "eval_samples_per_second": 19.546, + "eval_steps_per_second": 1.231, + "step": 1120 + }, + { + "epoch": 1.622286541244573, + "grad_norm": 1.251155138015747, + "learning_rate": 1.1182622687047465e-06, + "loss": 0.888, + "step": 1121 + }, + { + "epoch": 1.6237337192474675, + "grad_norm": 1.2730191946029663, + "learning_rate": 1.1142397425583269e-06, + "loss": 0.8995, + "step": 1122 + }, + { + "epoch": 1.6251808972503619, + "grad_norm": 1.3397986888885498, + "learning_rate": 1.1102172164119068e-06, + "loss": 0.8615, + "step": 1123 + }, + { + "epoch": 1.6266280752532563, + "grad_norm": 1.3800666332244873, + "learning_rate": 1.106194690265487e-06, + "loss": 0.883, + "step": 1124 + }, + { + "epoch": 1.6280752532561507, + "grad_norm": 1.3397117853164673, + "learning_rate": 1.1021721641190668e-06, + "loss": 0.8676, + "step": 1125 + }, + { + "epoch": 1.6295224312590448, + "grad_norm": 1.364835262298584, + "learning_rate": 1.098149637972647e-06, + "loss": 0.8644, + "step": 1126 + }, + { + "epoch": 1.6309696092619392, + "grad_norm": 1.4299700260162354, + "learning_rate": 1.094127111826227e-06, + "loss": 0.8592, + "step": 1127 + }, + { + "epoch": 1.6324167872648334, + "grad_norm": 1.4286457300186157, + "learning_rate": 1.090104585679807e-06, + "loss": 0.884, + "step": 1128 + }, + { + "epoch": 1.6338639652677278, + "grad_norm": 1.4902483224868774, + "learning_rate": 1.0860820595333871e-06, + "loss": 0.8908, + "step": 1129 + }, + { + "epoch": 1.6353111432706222, + "grad_norm": 1.4251055717468262, + "learning_rate": 1.082059533386967e-06, + "loss": 0.8789, + "step": 1130 + }, + { + "epoch": 1.6353111432706222, + "eval_loss": 1.002243995666504, + "eval_runtime": 51.3137, + "eval_samples_per_second": 19.488, + "eval_steps_per_second": 1.228, + "step": 1130 + }, + { + "epoch": 1.6367583212735166, + "grad_norm": 1.2832438945770264, + "learning_rate": 1.0780370072405472e-06, + "loss": 0.8602, + "step": 1131 + }, + { + "epoch": 1.638205499276411, + "grad_norm": 1.424117922782898, + "learning_rate": 1.0740144810941273e-06, + "loss": 0.8895, + "step": 1132 + }, + { + "epoch": 1.6396526772793054, + "grad_norm": 1.2426414489746094, + "learning_rate": 1.0699919549477072e-06, + "loss": 0.8486, + "step": 1133 + }, + { + "epoch": 1.6410998552821998, + "grad_norm": 1.3070167303085327, + "learning_rate": 1.0659694288012873e-06, + "loss": 0.8949, + "step": 1134 + }, + { + "epoch": 1.6425470332850942, + "grad_norm": 1.3324167728424072, + "learning_rate": 1.0619469026548673e-06, + "loss": 0.9082, + "step": 1135 + }, + { + "epoch": 1.6439942112879884, + "grad_norm": 1.3578763008117676, + "learning_rate": 1.0579243765084474e-06, + "loss": 0.85, + "step": 1136 + }, + { + "epoch": 1.6454413892908828, + "grad_norm": 1.510353684425354, + "learning_rate": 1.0539018503620275e-06, + "loss": 0.9176, + "step": 1137 + }, + { + "epoch": 1.646888567293777, + "grad_norm": 1.5371112823486328, + "learning_rate": 1.0498793242156074e-06, + "loss": 0.8811, + "step": 1138 + }, + { + "epoch": 1.6483357452966714, + "grad_norm": 1.5966984033584595, + "learning_rate": 1.0458567980691876e-06, + "loss": 0.8484, + "step": 1139 + }, + { + "epoch": 1.6497829232995658, + "grad_norm": 1.4401968717575073, + "learning_rate": 1.0418342719227675e-06, + "loss": 0.8679, + "step": 1140 + }, + { + "epoch": 1.6497829232995658, + "eval_loss": 1.0020740032196045, + "eval_runtime": 51.3426, + "eval_samples_per_second": 19.477, + "eval_steps_per_second": 1.227, + "step": 1140 + }, + { + "epoch": 1.6512301013024602, + "grad_norm": 1.2962098121643066, + "learning_rate": 1.0378117457763476e-06, + "loss": 0.8912, + "step": 1141 + }, + { + "epoch": 1.6526772793053546, + "grad_norm": 1.3592883348464966, + "learning_rate": 1.0337892196299275e-06, + "loss": 0.8635, + "step": 1142 + }, + { + "epoch": 1.654124457308249, + "grad_norm": 1.3698021173477173, + "learning_rate": 1.0297666934835076e-06, + "loss": 0.895, + "step": 1143 + }, + { + "epoch": 1.6555716353111434, + "grad_norm": 1.2848610877990723, + "learning_rate": 1.0257441673370878e-06, + "loss": 0.8634, + "step": 1144 + }, + { + "epoch": 1.6570188133140378, + "grad_norm": 1.3386603593826294, + "learning_rate": 1.021721641190668e-06, + "loss": 0.8908, + "step": 1145 + }, + { + "epoch": 1.658465991316932, + "grad_norm": 1.362449288368225, + "learning_rate": 1.017699115044248e-06, + "loss": 0.8521, + "step": 1146 + }, + { + "epoch": 1.6599131693198264, + "grad_norm": 1.3018198013305664, + "learning_rate": 1.013676588897828e-06, + "loss": 0.8772, + "step": 1147 + }, + { + "epoch": 1.6613603473227205, + "grad_norm": 1.5443693399429321, + "learning_rate": 1.009654062751408e-06, + "loss": 0.8963, + "step": 1148 + }, + { + "epoch": 1.662807525325615, + "grad_norm": 1.3997771739959717, + "learning_rate": 1.005631536604988e-06, + "loss": 0.8611, + "step": 1149 + }, + { + "epoch": 1.6642547033285093, + "grad_norm": 1.404136300086975, + "learning_rate": 1.0016090104585681e-06, + "loss": 0.8626, + "step": 1150 + }, + { + "epoch": 1.6642547033285093, + "eval_loss": 1.0032880306243896, + "eval_runtime": 51.4021, + "eval_samples_per_second": 19.454, + "eval_steps_per_second": 1.226, + "step": 1150 + }, + { + "epoch": 1.6657018813314037, + "grad_norm": 1.270905613899231, + "learning_rate": 9.97586484312148e-07, + "loss": 0.8747, + "step": 1151 + }, + { + "epoch": 1.6671490593342981, + "grad_norm": 1.2478210926055908, + "learning_rate": 9.935639581657282e-07, + "loss": 0.8836, + "step": 1152 + }, + { + "epoch": 1.6685962373371925, + "grad_norm": 1.4806969165802002, + "learning_rate": 9.895414320193083e-07, + "loss": 0.8704, + "step": 1153 + }, + { + "epoch": 1.670043415340087, + "grad_norm": 1.3371471166610718, + "learning_rate": 9.855189058728882e-07, + "loss": 0.8646, + "step": 1154 + }, + { + "epoch": 1.6714905933429813, + "grad_norm": 1.2173081636428833, + "learning_rate": 9.814963797264683e-07, + "loss": 0.9023, + "step": 1155 + }, + { + "epoch": 1.6729377713458755, + "grad_norm": 1.3810652494430542, + "learning_rate": 9.774738535800482e-07, + "loss": 0.8879, + "step": 1156 + }, + { + "epoch": 1.67438494934877, + "grad_norm": 1.2966829538345337, + "learning_rate": 9.734513274336284e-07, + "loss": 0.9112, + "step": 1157 + }, + { + "epoch": 1.675832127351664, + "grad_norm": 1.3627722263336182, + "learning_rate": 9.694288012872085e-07, + "loss": 0.8638, + "step": 1158 + }, + { + "epoch": 1.6772793053545585, + "grad_norm": 1.537874460220337, + "learning_rate": 9.654062751407884e-07, + "loss": 0.8467, + "step": 1159 + }, + { + "epoch": 1.678726483357453, + "grad_norm": 1.204199194908142, + "learning_rate": 9.613837489943685e-07, + "loss": 0.8511, + "step": 1160 + }, + { + "epoch": 1.678726483357453, + "eval_loss": 1.0015790462493896, + "eval_runtime": 51.1195, + "eval_samples_per_second": 19.562, + "eval_steps_per_second": 1.232, + "step": 1160 + }, + { + "epoch": 1.6801736613603473, + "grad_norm": 1.3930506706237793, + "learning_rate": 9.573612228479485e-07, + "loss": 0.8422, + "step": 1161 + }, + { + "epoch": 1.6816208393632417, + "grad_norm": 1.468660831451416, + "learning_rate": 9.533386967015286e-07, + "loss": 0.8434, + "step": 1162 + }, + { + "epoch": 1.683068017366136, + "grad_norm": 1.3240751028060913, + "learning_rate": 9.493161705551086e-07, + "loss": 0.8505, + "step": 1163 + }, + { + "epoch": 1.6845151953690305, + "grad_norm": 1.1470297574996948, + "learning_rate": 9.452936444086886e-07, + "loss": 0.8654, + "step": 1164 + }, + { + "epoch": 1.685962373371925, + "grad_norm": 1.3206348419189453, + "learning_rate": 9.412711182622688e-07, + "loss": 0.8718, + "step": 1165 + }, + { + "epoch": 1.687409551374819, + "grad_norm": 1.4581778049468994, + "learning_rate": 9.372485921158489e-07, + "loss": 0.9171, + "step": 1166 + }, + { + "epoch": 1.6888567293777135, + "grad_norm": 1.339200496673584, + "learning_rate": 9.332260659694289e-07, + "loss": 0.8616, + "step": 1167 + }, + { + "epoch": 1.6903039073806077, + "grad_norm": 1.268467903137207, + "learning_rate": 9.292035398230089e-07, + "loss": 0.9199, + "step": 1168 + }, + { + "epoch": 1.691751085383502, + "grad_norm": 1.230851173400879, + "learning_rate": 9.251810136765891e-07, + "loss": 0.8581, + "step": 1169 + }, + { + "epoch": 1.6931982633863965, + "grad_norm": 1.3200887441635132, + "learning_rate": 9.211584875301691e-07, + "loss": 0.8726, + "step": 1170 + }, + { + "epoch": 1.6931982633863965, + "eval_loss": 1.000694751739502, + "eval_runtime": 51.4929, + "eval_samples_per_second": 19.42, + "eval_steps_per_second": 1.223, + "step": 1170 + }, + { + "epoch": 1.6946454413892909, + "grad_norm": 1.3738995790481567, + "learning_rate": 9.171359613837491e-07, + "loss": 0.8735, + "step": 1171 + }, + { + "epoch": 1.6960926193921853, + "grad_norm": 1.3716143369674683, + "learning_rate": 9.131134352373291e-07, + "loss": 0.8647, + "step": 1172 + }, + { + "epoch": 1.6975397973950797, + "grad_norm": 1.395716667175293, + "learning_rate": 9.090909090909091e-07, + "loss": 0.8795, + "step": 1173 + }, + { + "epoch": 1.698986975397974, + "grad_norm": 1.2283461093902588, + "learning_rate": 9.050683829444893e-07, + "loss": 0.8893, + "step": 1174 + }, + { + "epoch": 1.7004341534008685, + "grad_norm": 1.4803199768066406, + "learning_rate": 9.010458567980693e-07, + "loss": 0.8822, + "step": 1175 + }, + { + "epoch": 1.7018813314037626, + "grad_norm": 1.5868210792541504, + "learning_rate": 8.970233306516493e-07, + "loss": 0.8738, + "step": 1176 + }, + { + "epoch": 1.703328509406657, + "grad_norm": 1.3454225063323975, + "learning_rate": 8.930008045052293e-07, + "loss": 0.8631, + "step": 1177 + }, + { + "epoch": 1.7047756874095512, + "grad_norm": 1.3840582370758057, + "learning_rate": 8.889782783588094e-07, + "loss": 0.8837, + "step": 1178 + }, + { + "epoch": 1.7062228654124456, + "grad_norm": 1.2975925207138062, + "learning_rate": 8.849557522123895e-07, + "loss": 0.8484, + "step": 1179 + }, + { + "epoch": 1.70767004341534, + "grad_norm": 1.2764484882354736, + "learning_rate": 8.809332260659695e-07, + "loss": 0.8459, + "step": 1180 + }, + { + "epoch": 1.70767004341534, + "eval_loss": 1.0018376111984253, + "eval_runtime": 51.146, + "eval_samples_per_second": 19.552, + "eval_steps_per_second": 1.232, + "step": 1180 + }, + { + "epoch": 1.7091172214182344, + "grad_norm": 1.4238110780715942, + "learning_rate": 8.769106999195495e-07, + "loss": 0.8752, + "step": 1181 + }, + { + "epoch": 1.7105643994211288, + "grad_norm": 1.359768033027649, + "learning_rate": 8.728881737731296e-07, + "loss": 0.8636, + "step": 1182 + }, + { + "epoch": 1.7120115774240232, + "grad_norm": 1.4667472839355469, + "learning_rate": 8.688656476267096e-07, + "loss": 0.8603, + "step": 1183 + }, + { + "epoch": 1.7134587554269176, + "grad_norm": 1.5296673774719238, + "learning_rate": 8.648431214802896e-07, + "loss": 0.8552, + "step": 1184 + }, + { + "epoch": 1.714905933429812, + "grad_norm": 1.3559892177581787, + "learning_rate": 8.608205953338697e-07, + "loss": 0.8925, + "step": 1185 + }, + { + "epoch": 1.7163531114327062, + "grad_norm": 1.2427695989608765, + "learning_rate": 8.567980691874497e-07, + "loss": 0.9044, + "step": 1186 + }, + { + "epoch": 1.7178002894356006, + "grad_norm": 1.2985711097717285, + "learning_rate": 8.527755430410298e-07, + "loss": 0.8931, + "step": 1187 + }, + { + "epoch": 1.7192474674384948, + "grad_norm": 1.3176608085632324, + "learning_rate": 8.487530168946098e-07, + "loss": 0.8803, + "step": 1188 + }, + { + "epoch": 1.7206946454413892, + "grad_norm": 1.2636421918869019, + "learning_rate": 8.4473049074819e-07, + "loss": 0.8897, + "step": 1189 + }, + { + "epoch": 1.7221418234442836, + "grad_norm": 1.2745264768600464, + "learning_rate": 8.4070796460177e-07, + "loss": 0.8828, + "step": 1190 + }, + { + "epoch": 1.7221418234442836, + "eval_loss": 1.000553846359253, + "eval_runtime": 51.5537, + "eval_samples_per_second": 19.397, + "eval_steps_per_second": 1.222, + "step": 1190 + }, + { + "epoch": 1.723589001447178, + "grad_norm": 1.4252241849899292, + "learning_rate": 8.366854384553501e-07, + "loss": 0.9115, + "step": 1191 + }, + { + "epoch": 1.7250361794500724, + "grad_norm": 1.4272230863571167, + "learning_rate": 8.326629123089301e-07, + "loss": 0.878, + "step": 1192 + }, + { + "epoch": 1.7264833574529668, + "grad_norm": 1.1996142864227295, + "learning_rate": 8.286403861625101e-07, + "loss": 0.871, + "step": 1193 + }, + { + "epoch": 1.7279305354558612, + "grad_norm": 1.219294548034668, + "learning_rate": 8.246178600160902e-07, + "loss": 0.87, + "step": 1194 + }, + { + "epoch": 1.7293777134587556, + "grad_norm": 1.4402002096176147, + "learning_rate": 8.205953338696703e-07, + "loss": 0.8683, + "step": 1195 + }, + { + "epoch": 1.7308248914616498, + "grad_norm": 1.5708014965057373, + "learning_rate": 8.165728077232503e-07, + "loss": 0.8931, + "step": 1196 + }, + { + "epoch": 1.7322720694645442, + "grad_norm": 1.4649317264556885, + "learning_rate": 8.125502815768303e-07, + "loss": 0.8771, + "step": 1197 + }, + { + "epoch": 1.7337192474674383, + "grad_norm": 1.5291906595230103, + "learning_rate": 8.085277554304103e-07, + "loss": 0.9068, + "step": 1198 + }, + { + "epoch": 1.7351664254703327, + "grad_norm": 1.3457047939300537, + "learning_rate": 8.045052292839905e-07, + "loss": 0.856, + "step": 1199 + }, + { + "epoch": 1.7366136034732271, + "grad_norm": 1.3352559804916382, + "learning_rate": 8.004827031375705e-07, + "loss": 0.8971, + "step": 1200 + }, + { + "epoch": 1.7366136034732271, + "eval_loss": 1.001030445098877, + "eval_runtime": 51.5318, + "eval_samples_per_second": 19.406, + "eval_steps_per_second": 1.223, + "step": 1200 + }, + { + "epoch": 1.7380607814761215, + "grad_norm": 1.3720887899398804, + "learning_rate": 7.964601769911505e-07, + "loss": 0.8571, + "step": 1201 + }, + { + "epoch": 1.739507959479016, + "grad_norm": 1.4130148887634277, + "learning_rate": 7.924376508447305e-07, + "loss": 0.8851, + "step": 1202 + }, + { + "epoch": 1.7409551374819103, + "grad_norm": 1.2895668745040894, + "learning_rate": 7.884151246983105e-07, + "loss": 0.8622, + "step": 1203 + }, + { + "epoch": 1.7424023154848047, + "grad_norm": 1.1915003061294556, + "learning_rate": 7.843925985518907e-07, + "loss": 0.8641, + "step": 1204 + }, + { + "epoch": 1.7438494934876991, + "grad_norm": 1.3241524696350098, + "learning_rate": 7.803700724054707e-07, + "loss": 0.8904, + "step": 1205 + }, + { + "epoch": 1.7452966714905933, + "grad_norm": 1.3196359872817993, + "learning_rate": 7.763475462590507e-07, + "loss": 0.8956, + "step": 1206 + }, + { + "epoch": 1.7467438494934877, + "grad_norm": 1.3720238208770752, + "learning_rate": 7.723250201126307e-07, + "loss": 0.8687, + "step": 1207 + }, + { + "epoch": 1.7481910274963819, + "grad_norm": 1.3211584091186523, + "learning_rate": 7.683024939662108e-07, + "loss": 0.8707, + "step": 1208 + }, + { + "epoch": 1.7496382054992763, + "grad_norm": 1.1938427686691284, + "learning_rate": 7.642799678197908e-07, + "loss": 0.8939, + "step": 1209 + }, + { + "epoch": 1.7510853835021707, + "grad_norm": 1.2782360315322876, + "learning_rate": 7.602574416733709e-07, + "loss": 0.8424, + "step": 1210 + }, + { + "epoch": 1.7510853835021707, + "eval_loss": 1.001528024673462, + "eval_runtime": 51.5327, + "eval_samples_per_second": 19.405, + "eval_steps_per_second": 1.223, + "step": 1210 + }, + { + "epoch": 1.752532561505065, + "grad_norm": 1.2574527263641357, + "learning_rate": 7.56234915526951e-07, + "loss": 0.9146, + "step": 1211 + }, + { + "epoch": 1.7539797395079595, + "grad_norm": 1.2645962238311768, + "learning_rate": 7.522123893805311e-07, + "loss": 0.8692, + "step": 1212 + }, + { + "epoch": 1.755426917510854, + "grad_norm": 1.3895294666290283, + "learning_rate": 7.481898632341112e-07, + "loss": 0.8605, + "step": 1213 + }, + { + "epoch": 1.7568740955137483, + "grad_norm": 1.3689818382263184, + "learning_rate": 7.441673370876912e-07, + "loss": 0.8743, + "step": 1214 + }, + { + "epoch": 1.7583212735166427, + "grad_norm": 1.3171066045761108, + "learning_rate": 7.401448109412712e-07, + "loss": 0.8639, + "step": 1215 + }, + { + "epoch": 1.7597684515195369, + "grad_norm": 1.3380731344223022, + "learning_rate": 7.361222847948512e-07, + "loss": 0.8743, + "step": 1216 + }, + { + "epoch": 1.7612156295224313, + "grad_norm": 1.2480511665344238, + "learning_rate": 7.320997586484313e-07, + "loss": 0.8863, + "step": 1217 + }, + { + "epoch": 1.7626628075253257, + "grad_norm": 1.3746633529663086, + "learning_rate": 7.280772325020113e-07, + "loss": 0.8666, + "step": 1218 + }, + { + "epoch": 1.7641099855282198, + "grad_norm": 1.3885321617126465, + "learning_rate": 7.240547063555914e-07, + "loss": 0.8706, + "step": 1219 + }, + { + "epoch": 1.7655571635311142, + "grad_norm": 1.319083333015442, + "learning_rate": 7.200321802091714e-07, + "loss": 0.8659, + "step": 1220 + }, + { + "epoch": 1.7655571635311142, + "eval_loss": 1.0013295412063599, + "eval_runtime": 51.4075, + "eval_samples_per_second": 19.452, + "eval_steps_per_second": 1.226, + "step": 1220 + }, + { + "epoch": 1.7670043415340086, + "grad_norm": 1.317706823348999, + "learning_rate": 7.160096540627515e-07, + "loss": 0.8615, + "step": 1221 + }, + { + "epoch": 1.768451519536903, + "grad_norm": 1.4504474401474, + "learning_rate": 7.119871279163315e-07, + "loss": 0.8755, + "step": 1222 + }, + { + "epoch": 1.7698986975397974, + "grad_norm": 1.2188657522201538, + "learning_rate": 7.079646017699115e-07, + "loss": 0.8887, + "step": 1223 + }, + { + "epoch": 1.7713458755426919, + "grad_norm": 1.2679225206375122, + "learning_rate": 7.039420756234916e-07, + "loss": 0.8682, + "step": 1224 + }, + { + "epoch": 1.7727930535455863, + "grad_norm": 1.3126505613327026, + "learning_rate": 6.999195494770717e-07, + "loss": 0.8606, + "step": 1225 + }, + { + "epoch": 1.7742402315484804, + "grad_norm": 1.338470458984375, + "learning_rate": 6.958970233306517e-07, + "loss": 0.89, + "step": 1226 + }, + { + "epoch": 1.7756874095513748, + "grad_norm": 1.378913402557373, + "learning_rate": 6.918744971842317e-07, + "loss": 0.8405, + "step": 1227 + }, + { + "epoch": 1.7771345875542692, + "grad_norm": 1.449596881866455, + "learning_rate": 6.878519710378117e-07, + "loss": 0.8627, + "step": 1228 + }, + { + "epoch": 1.7785817655571634, + "grad_norm": 1.242762565612793, + "learning_rate": 6.838294448913918e-07, + "loss": 0.841, + "step": 1229 + }, + { + "epoch": 1.7800289435600578, + "grad_norm": 1.3406858444213867, + "learning_rate": 6.798069187449719e-07, + "loss": 0.8773, + "step": 1230 + }, + { + "epoch": 1.7800289435600578, + "eval_loss": 0.9990753531455994, + "eval_runtime": 51.4208, + "eval_samples_per_second": 19.447, + "eval_steps_per_second": 1.225, + "step": 1230 + }, + { + "epoch": 1.7814761215629522, + "grad_norm": 1.2245969772338867, + "learning_rate": 6.757843925985519e-07, + "loss": 0.8812, + "step": 1231 + }, + { + "epoch": 1.7829232995658466, + "grad_norm": 1.4095789194107056, + "learning_rate": 6.717618664521319e-07, + "loss": 0.8696, + "step": 1232 + }, + { + "epoch": 1.784370477568741, + "grad_norm": 1.3338826894760132, + "learning_rate": 6.677393403057121e-07, + "loss": 0.8729, + "step": 1233 + }, + { + "epoch": 1.7858176555716354, + "grad_norm": 1.2761763334274292, + "learning_rate": 6.637168141592922e-07, + "loss": 0.8732, + "step": 1234 + }, + { + "epoch": 1.7872648335745298, + "grad_norm": 1.336746335029602, + "learning_rate": 6.596942880128722e-07, + "loss": 0.8831, + "step": 1235 + }, + { + "epoch": 1.788712011577424, + "grad_norm": 1.4172648191452026, + "learning_rate": 6.556717618664522e-07, + "loss": 0.8716, + "step": 1236 + }, + { + "epoch": 1.7901591895803184, + "grad_norm": 1.347998023033142, + "learning_rate": 6.516492357200322e-07, + "loss": 0.8519, + "step": 1237 + }, + { + "epoch": 1.7916063675832128, + "grad_norm": 1.283597469329834, + "learning_rate": 6.476267095736124e-07, + "loss": 0.8764, + "step": 1238 + }, + { + "epoch": 1.793053545586107, + "grad_norm": 1.365626573562622, + "learning_rate": 6.436041834271924e-07, + "loss": 0.8913, + "step": 1239 + }, + { + "epoch": 1.7945007235890014, + "grad_norm": 1.2543431520462036, + "learning_rate": 6.395816572807724e-07, + "loss": 0.8712, + "step": 1240 + }, + { + "epoch": 1.7945007235890014, + "eval_loss": 0.9990435838699341, + "eval_runtime": 51.3052, + "eval_samples_per_second": 19.491, + "eval_steps_per_second": 1.228, + "step": 1240 + }, + { + "epoch": 1.7959479015918958, + "grad_norm": 1.267150640487671, + "learning_rate": 6.355591311343524e-07, + "loss": 0.8466, + "step": 1241 + }, + { + "epoch": 1.7973950795947902, + "grad_norm": 1.3889726400375366, + "learning_rate": 6.315366049879324e-07, + "loss": 0.8945, + "step": 1242 + }, + { + "epoch": 1.7988422575976846, + "grad_norm": 1.3856347799301147, + "learning_rate": 6.275140788415126e-07, + "loss": 0.896, + "step": 1243 + }, + { + "epoch": 1.800289435600579, + "grad_norm": 1.2807971239089966, + "learning_rate": 6.234915526950926e-07, + "loss": 0.8622, + "step": 1244 + }, + { + "epoch": 1.8017366136034734, + "grad_norm": 1.1937495470046997, + "learning_rate": 6.194690265486726e-07, + "loss": 0.8679, + "step": 1245 + }, + { + "epoch": 1.8031837916063675, + "grad_norm": 1.3732362985610962, + "learning_rate": 6.154465004022526e-07, + "loss": 0.8929, + "step": 1246 + }, + { + "epoch": 1.804630969609262, + "grad_norm": 1.3486156463623047, + "learning_rate": 6.114239742558327e-07, + "loss": 0.8892, + "step": 1247 + }, + { + "epoch": 1.8060781476121563, + "grad_norm": 1.2842955589294434, + "learning_rate": 6.074014481094127e-07, + "loss": 0.8622, + "step": 1248 + }, + { + "epoch": 1.8075253256150505, + "grad_norm": 1.2810155153274536, + "learning_rate": 6.033789219629928e-07, + "loss": 0.8901, + "step": 1249 + }, + { + "epoch": 1.808972503617945, + "grad_norm": 1.2745157480239868, + "learning_rate": 5.993563958165728e-07, + "loss": 0.8867, + "step": 1250 + }, + { + "epoch": 1.808972503617945, + "eval_loss": 0.9986433982849121, + "eval_runtime": 51.2553, + "eval_samples_per_second": 19.51, + "eval_steps_per_second": 1.229, + "step": 1250 + }, + { + "epoch": 1.8104196816208393, + "grad_norm": 1.1747359037399292, + "learning_rate": 5.95333869670153e-07, + "loss": 0.8639, + "step": 1251 + }, + { + "epoch": 1.8118668596237337, + "grad_norm": 1.207029104232788, + "learning_rate": 5.91311343523733e-07, + "loss": 0.8668, + "step": 1252 + }, + { + "epoch": 1.8133140376266281, + "grad_norm": 1.2627661228179932, + "learning_rate": 5.87288817377313e-07, + "loss": 0.8697, + "step": 1253 + }, + { + "epoch": 1.8147612156295225, + "grad_norm": 1.3808412551879883, + "learning_rate": 5.83266291230893e-07, + "loss": 0.8753, + "step": 1254 + }, + { + "epoch": 1.816208393632417, + "grad_norm": 1.1532347202301025, + "learning_rate": 5.79243765084473e-07, + "loss": 0.9014, + "step": 1255 + }, + { + "epoch": 1.817655571635311, + "grad_norm": 1.3013627529144287, + "learning_rate": 5.752212389380532e-07, + "loss": 0.852, + "step": 1256 + }, + { + "epoch": 1.8191027496382055, + "grad_norm": 1.3484326601028442, + "learning_rate": 5.711987127916332e-07, + "loss": 0.866, + "step": 1257 + }, + { + "epoch": 1.8205499276411, + "grad_norm": 1.276702880859375, + "learning_rate": 5.671761866452132e-07, + "loss": 0.8636, + "step": 1258 + }, + { + "epoch": 1.821997105643994, + "grad_norm": 1.2385605573654175, + "learning_rate": 5.631536604987932e-07, + "loss": 0.8703, + "step": 1259 + }, + { + "epoch": 1.8234442836468885, + "grad_norm": 1.2592774629592896, + "learning_rate": 5.591311343523733e-07, + "loss": 0.8907, + "step": 1260 + }, + { + "epoch": 1.8234442836468885, + "eval_loss": 0.9987832307815552, + "eval_runtime": 51.1108, + "eval_samples_per_second": 19.565, + "eval_steps_per_second": 1.233, + "step": 1260 + }, + { + "epoch": 1.8248914616497829, + "grad_norm": 1.434237003326416, + "learning_rate": 5.551086082059534e-07, + "loss": 0.8721, + "step": 1261 + }, + { + "epoch": 1.8263386396526773, + "grad_norm": 1.3201887607574463, + "learning_rate": 5.510860820595334e-07, + "loss": 0.8474, + "step": 1262 + }, + { + "epoch": 1.8277858176555717, + "grad_norm": 1.2699350118637085, + "learning_rate": 5.470635559131135e-07, + "loss": 0.8452, + "step": 1263 + }, + { + "epoch": 1.829232995658466, + "grad_norm": 1.2369650602340698, + "learning_rate": 5.430410297666936e-07, + "loss": 0.8842, + "step": 1264 + }, + { + "epoch": 1.8306801736613605, + "grad_norm": 1.3479478359222412, + "learning_rate": 5.390185036202736e-07, + "loss": 0.8687, + "step": 1265 + }, + { + "epoch": 1.8321273516642547, + "grad_norm": 1.2551127672195435, + "learning_rate": 5.349959774738536e-07, + "loss": 0.8475, + "step": 1266 + }, + { + "epoch": 1.833574529667149, + "grad_norm": 1.2529149055480957, + "learning_rate": 5.309734513274336e-07, + "loss": 0.875, + "step": 1267 + }, + { + "epoch": 1.8350217076700435, + "grad_norm": 1.220129132270813, + "learning_rate": 5.269509251810138e-07, + "loss": 0.8264, + "step": 1268 + }, + { + "epoch": 1.8364688856729376, + "grad_norm": 1.347771406173706, + "learning_rate": 5.229283990345938e-07, + "loss": 0.8804, + "step": 1269 + }, + { + "epoch": 1.837916063675832, + "grad_norm": 1.2903997898101807, + "learning_rate": 5.189058728881738e-07, + "loss": 0.9003, + "step": 1270 + }, + { + "epoch": 1.837916063675832, + "eval_loss": 0.9989792108535767, + "eval_runtime": 50.9941, + "eval_samples_per_second": 19.61, + "eval_steps_per_second": 1.235, + "step": 1270 + }, + { + "epoch": 1.8393632416787264, + "grad_norm": 1.3161100149154663, + "learning_rate": 5.148833467417538e-07, + "loss": 0.8677, + "step": 1271 + }, + { + "epoch": 1.8408104196816208, + "grad_norm": 1.2684601545333862, + "learning_rate": 5.10860820595334e-07, + "loss": 0.8897, + "step": 1272 + }, + { + "epoch": 1.8422575976845152, + "grad_norm": 1.277329921722412, + "learning_rate": 5.06838294448914e-07, + "loss": 0.8548, + "step": 1273 + }, + { + "epoch": 1.8437047756874096, + "grad_norm": 1.339925765991211, + "learning_rate": 5.02815768302494e-07, + "loss": 0.8954, + "step": 1274 + }, + { + "epoch": 1.845151953690304, + "grad_norm": 1.3327856063842773, + "learning_rate": 4.98793242156074e-07, + "loss": 0.8702, + "step": 1275 + }, + { + "epoch": 1.8465991316931982, + "grad_norm": 1.28009831905365, + "learning_rate": 4.947707160096541e-07, + "loss": 0.8612, + "step": 1276 + }, + { + "epoch": 1.8480463096960926, + "grad_norm": 1.2893006801605225, + "learning_rate": 4.907481898632342e-07, + "loss": 0.8794, + "step": 1277 + }, + { + "epoch": 1.849493487698987, + "grad_norm": 1.3114680051803589, + "learning_rate": 4.867256637168142e-07, + "loss": 0.8801, + "step": 1278 + }, + { + "epoch": 1.8509406657018812, + "grad_norm": 1.3199737071990967, + "learning_rate": 4.827031375703942e-07, + "loss": 0.8585, + "step": 1279 + }, + { + "epoch": 1.8523878437047756, + "grad_norm": 1.2723737955093384, + "learning_rate": 4.786806114239742e-07, + "loss": 0.8702, + "step": 1280 + }, + { + "epoch": 1.8523878437047756, + "eval_loss": 0.9978076815605164, + "eval_runtime": 51.3095, + "eval_samples_per_second": 19.49, + "eval_steps_per_second": 1.228, + "step": 1280 + }, + { + "epoch": 1.85383502170767, + "grad_norm": 1.2504792213439941, + "learning_rate": 4.746580852775543e-07, + "loss": 0.8418, + "step": 1281 + }, + { + "epoch": 1.8552821997105644, + "grad_norm": 1.2581747770309448, + "learning_rate": 4.706355591311344e-07, + "loss": 0.8374, + "step": 1282 + }, + { + "epoch": 1.8567293777134588, + "grad_norm": 1.3575350046157837, + "learning_rate": 4.6661303298471445e-07, + "loss": 0.8678, + "step": 1283 + }, + { + "epoch": 1.8581765557163532, + "grad_norm": 1.3722816705703735, + "learning_rate": 4.6259050683829453e-07, + "loss": 0.9002, + "step": 1284 + }, + { + "epoch": 1.8596237337192476, + "grad_norm": 1.5536175966262817, + "learning_rate": 4.5856798069187455e-07, + "loss": 0.8591, + "step": 1285 + }, + { + "epoch": 1.8610709117221418, + "grad_norm": 1.2896004915237427, + "learning_rate": 4.5454545454545457e-07, + "loss": 0.8815, + "step": 1286 + }, + { + "epoch": 1.8625180897250362, + "grad_norm": 1.2189226150512695, + "learning_rate": 4.5052292839903465e-07, + "loss": 0.8527, + "step": 1287 + }, + { + "epoch": 1.8639652677279306, + "grad_norm": 1.33291757106781, + "learning_rate": 4.4650040225261467e-07, + "loss": 0.8695, + "step": 1288 + }, + { + "epoch": 1.8654124457308248, + "grad_norm": 1.2411049604415894, + "learning_rate": 4.4247787610619474e-07, + "loss": 0.8853, + "step": 1289 + }, + { + "epoch": 1.8668596237337192, + "grad_norm": 1.324524998664856, + "learning_rate": 4.3845534995977477e-07, + "loss": 0.8923, + "step": 1290 + }, + { + "epoch": 1.8668596237337192, + "eval_loss": 0.998564600944519, + "eval_runtime": 50.8858, + "eval_samples_per_second": 19.652, + "eval_steps_per_second": 1.238, + "step": 1290 + }, + { + "epoch": 1.8683068017366136, + "grad_norm": 1.2136908769607544, + "learning_rate": 4.344328238133548e-07, + "loss": 0.8463, + "step": 1291 + }, + { + "epoch": 1.869753979739508, + "grad_norm": 1.3158811330795288, + "learning_rate": 4.3041029766693486e-07, + "loss": 0.8626, + "step": 1292 + }, + { + "epoch": 1.8712011577424024, + "grad_norm": 1.4381555318832397, + "learning_rate": 4.263877715205149e-07, + "loss": 0.8752, + "step": 1293 + }, + { + "epoch": 1.8726483357452968, + "grad_norm": 1.1753792762756348, + "learning_rate": 4.22365245374095e-07, + "loss": 0.847, + "step": 1294 + }, + { + "epoch": 1.8740955137481912, + "grad_norm": 1.4084643125534058, + "learning_rate": 4.1834271922767503e-07, + "loss": 0.9012, + "step": 1295 + }, + { + "epoch": 1.8755426917510853, + "grad_norm": 1.400168538093567, + "learning_rate": 4.1432019308125506e-07, + "loss": 0.8833, + "step": 1296 + }, + { + "epoch": 1.8769898697539797, + "grad_norm": 1.2450108528137207, + "learning_rate": 4.1029766693483513e-07, + "loss": 0.887, + "step": 1297 + }, + { + "epoch": 1.8784370477568741, + "grad_norm": 1.3698848485946655, + "learning_rate": 4.0627514078841515e-07, + "loss": 0.8566, + "step": 1298 + }, + { + "epoch": 1.8798842257597683, + "grad_norm": 1.2968684434890747, + "learning_rate": 4.0225261464199523e-07, + "loss": 0.8704, + "step": 1299 + }, + { + "epoch": 1.8813314037626627, + "grad_norm": 1.3081411123275757, + "learning_rate": 3.9823008849557525e-07, + "loss": 0.8774, + "step": 1300 + }, + { + "epoch": 1.8813314037626627, + "eval_loss": 0.9973741769790649, + "eval_runtime": 50.7915, + "eval_samples_per_second": 19.688, + "eval_steps_per_second": 1.24, + "step": 1300 + }, + { + "epoch": 1.8827785817655571, + "grad_norm": 1.3137695789337158, + "learning_rate": 3.9420756234915527e-07, + "loss": 0.8791, + "step": 1301 + }, + { + "epoch": 1.8842257597684515, + "grad_norm": 1.2670238018035889, + "learning_rate": 3.9018503620273535e-07, + "loss": 0.8924, + "step": 1302 + }, + { + "epoch": 1.885672937771346, + "grad_norm": 1.411651849746704, + "learning_rate": 3.8616251005631537e-07, + "loss": 0.8718, + "step": 1303 + }, + { + "epoch": 1.8871201157742403, + "grad_norm": 1.2660640478134155, + "learning_rate": 3.821399839098954e-07, + "loss": 0.8592, + "step": 1304 + }, + { + "epoch": 1.8885672937771347, + "grad_norm": 1.3477122783660889, + "learning_rate": 3.781174577634755e-07, + "loss": 0.8536, + "step": 1305 + }, + { + "epoch": 1.890014471780029, + "grad_norm": 1.21705162525177, + "learning_rate": 3.740949316170556e-07, + "loss": 0.8391, + "step": 1306 + }, + { + "epoch": 1.8914616497829233, + "grad_norm": 1.2772414684295654, + "learning_rate": 3.700724054706356e-07, + "loss": 0.8673, + "step": 1307 + }, + { + "epoch": 1.8929088277858177, + "grad_norm": 1.277519702911377, + "learning_rate": 3.6604987932421563e-07, + "loss": 0.8722, + "step": 1308 + }, + { + "epoch": 1.8943560057887119, + "grad_norm": 1.391261100769043, + "learning_rate": 3.620273531777957e-07, + "loss": 0.8804, + "step": 1309 + }, + { + "epoch": 1.8958031837916063, + "grad_norm": 1.216473937034607, + "learning_rate": 3.5800482703137573e-07, + "loss": 0.8736, + "step": 1310 + }, + { + "epoch": 1.8958031837916063, + "eval_loss": 0.9964368939399719, + "eval_runtime": 50.7299, + "eval_samples_per_second": 19.712, + "eval_steps_per_second": 1.242, + "step": 1310 + }, + { + "epoch": 1.8972503617945007, + "grad_norm": 1.3251583576202393, + "learning_rate": 3.5398230088495575e-07, + "loss": 0.8681, + "step": 1311 + }, + { + "epoch": 1.898697539797395, + "grad_norm": 1.2805577516555786, + "learning_rate": 3.4995977473853583e-07, + "loss": 0.8548, + "step": 1312 + }, + { + "epoch": 1.9001447178002895, + "grad_norm": 1.263411521911621, + "learning_rate": 3.4593724859211585e-07, + "loss": 0.8808, + "step": 1313 + }, + { + "epoch": 1.9015918958031839, + "grad_norm": 1.388716459274292, + "learning_rate": 3.419147224456959e-07, + "loss": 0.8764, + "step": 1314 + }, + { + "epoch": 1.9030390738060783, + "grad_norm": 1.393967866897583, + "learning_rate": 3.3789219629927595e-07, + "loss": 0.8893, + "step": 1315 + }, + { + "epoch": 1.9044862518089725, + "grad_norm": 1.522119164466858, + "learning_rate": 3.338696701528561e-07, + "loss": 0.857, + "step": 1316 + }, + { + "epoch": 1.9059334298118669, + "grad_norm": 1.394858956336975, + "learning_rate": 3.298471440064361e-07, + "loss": 0.8507, + "step": 1317 + }, + { + "epoch": 1.9073806078147613, + "grad_norm": 1.2374058961868286, + "learning_rate": 3.258246178600161e-07, + "loss": 0.8904, + "step": 1318 + }, + { + "epoch": 1.9088277858176554, + "grad_norm": 1.3255163431167603, + "learning_rate": 3.218020917135962e-07, + "loss": 0.86, + "step": 1319 + }, + { + "epoch": 1.9102749638205498, + "grad_norm": 1.2023863792419434, + "learning_rate": 3.177795655671762e-07, + "loss": 0.8478, + "step": 1320 + }, + { + "epoch": 1.9102749638205498, + "eval_loss": 0.9957064390182495, + "eval_runtime": 51.1626, + "eval_samples_per_second": 19.546, + "eval_steps_per_second": 1.231, + "step": 1320 + }, + { + "epoch": 1.9117221418234442, + "grad_norm": 1.2819989919662476, + "learning_rate": 3.137570394207563e-07, + "loss": 0.8866, + "step": 1321 + }, + { + "epoch": 1.9131693198263386, + "grad_norm": 1.281292200088501, + "learning_rate": 3.097345132743363e-07, + "loss": 0.8854, + "step": 1322 + }, + { + "epoch": 1.914616497829233, + "grad_norm": 1.3323489427566528, + "learning_rate": 3.0571198712791633e-07, + "loss": 0.8912, + "step": 1323 + }, + { + "epoch": 1.9160636758321274, + "grad_norm": 1.2403019666671753, + "learning_rate": 3.016894609814964e-07, + "loss": 0.8594, + "step": 1324 + }, + { + "epoch": 1.9175108538350218, + "grad_norm": 1.2935993671417236, + "learning_rate": 2.976669348350765e-07, + "loss": 0.8704, + "step": 1325 + }, + { + "epoch": 1.918958031837916, + "grad_norm": 1.2857427597045898, + "learning_rate": 2.936444086886565e-07, + "loss": 0.8623, + "step": 1326 + }, + { + "epoch": 1.9204052098408104, + "grad_norm": 1.39542555809021, + "learning_rate": 2.896218825422365e-07, + "loss": 0.914, + "step": 1327 + }, + { + "epoch": 1.9218523878437048, + "grad_norm": 1.3790624141693115, + "learning_rate": 2.855993563958166e-07, + "loss": 0.8515, + "step": 1328 + }, + { + "epoch": 1.923299565846599, + "grad_norm": 1.3047775030136108, + "learning_rate": 2.815768302493966e-07, + "loss": 0.8645, + "step": 1329 + }, + { + "epoch": 1.9247467438494934, + "grad_norm": 1.2745611667633057, + "learning_rate": 2.775543041029767e-07, + "loss": 0.8789, + "step": 1330 + }, + { + "epoch": 1.9247467438494934, + "eval_loss": 0.9957833886146545, + "eval_runtime": 50.9757, + "eval_samples_per_second": 19.617, + "eval_steps_per_second": 1.236, + "step": 1330 + }, + { + "epoch": 1.9261939218523878, + "grad_norm": 1.2854071855545044, + "learning_rate": 2.7353177795655677e-07, + "loss": 0.8856, + "step": 1331 + }, + { + "epoch": 1.9276410998552822, + "grad_norm": 1.3202019929885864, + "learning_rate": 2.695092518101368e-07, + "loss": 0.8401, + "step": 1332 + }, + { + "epoch": 1.9290882778581766, + "grad_norm": 1.3794660568237305, + "learning_rate": 2.654867256637168e-07, + "loss": 0.8615, + "step": 1333 + }, + { + "epoch": 1.930535455861071, + "grad_norm": 1.39586341381073, + "learning_rate": 2.614641995172969e-07, + "loss": 0.8989, + "step": 1334 + }, + { + "epoch": 1.9319826338639654, + "grad_norm": 1.1921442747116089, + "learning_rate": 2.574416733708769e-07, + "loss": 0.8598, + "step": 1335 + }, + { + "epoch": 1.9334298118668596, + "grad_norm": 1.3120014667510986, + "learning_rate": 2.53419147224457e-07, + "loss": 0.8501, + "step": 1336 + }, + { + "epoch": 1.934876989869754, + "grad_norm": 1.2767196893692017, + "learning_rate": 2.49396621078037e-07, + "loss": 0.8781, + "step": 1337 + }, + { + "epoch": 1.9363241678726484, + "grad_norm": 1.1905773878097534, + "learning_rate": 2.453740949316171e-07, + "loss": 0.866, + "step": 1338 + }, + { + "epoch": 1.9377713458755426, + "grad_norm": 1.3939810991287231, + "learning_rate": 2.413515687851971e-07, + "loss": 0.8898, + "step": 1339 + }, + { + "epoch": 1.939218523878437, + "grad_norm": 1.3606054782867432, + "learning_rate": 2.3732904263877715e-07, + "loss": 0.8509, + "step": 1340 + }, + { + "epoch": 1.939218523878437, + "eval_loss": 0.9956860542297363, + "eval_runtime": 51.0549, + "eval_samples_per_second": 19.587, + "eval_steps_per_second": 1.234, + "step": 1340 + }, + { + "epoch": 1.9406657018813314, + "grad_norm": 1.226684331893921, + "learning_rate": 2.3330651649235723e-07, + "loss": 0.87, + "step": 1341 + }, + { + "epoch": 1.9421128798842258, + "grad_norm": 1.273861050605774, + "learning_rate": 2.2928399034593728e-07, + "loss": 0.8431, + "step": 1342 + }, + { + "epoch": 1.9435600578871202, + "grad_norm": 1.3127763271331787, + "learning_rate": 2.2526146419951732e-07, + "loss": 0.8779, + "step": 1343 + }, + { + "epoch": 1.9450072358900146, + "grad_norm": 1.2036153078079224, + "learning_rate": 2.2123893805309737e-07, + "loss": 0.886, + "step": 1344 + }, + { + "epoch": 1.946454413892909, + "grad_norm": 1.4593393802642822, + "learning_rate": 2.172164119066774e-07, + "loss": 0.9032, + "step": 1345 + }, + { + "epoch": 1.9479015918958031, + "grad_norm": 1.5014209747314453, + "learning_rate": 2.1319388576025744e-07, + "loss": 0.9178, + "step": 1346 + }, + { + "epoch": 1.9493487698986975, + "grad_norm": 1.3644002676010132, + "learning_rate": 2.0917135961383752e-07, + "loss": 0.8576, + "step": 1347 + }, + { + "epoch": 1.950795947901592, + "grad_norm": 1.3202813863754272, + "learning_rate": 2.0514883346741757e-07, + "loss": 0.87, + "step": 1348 + }, + { + "epoch": 1.9522431259044861, + "grad_norm": 1.20876944065094, + "learning_rate": 2.0112630732099761e-07, + "loss": 0.8557, + "step": 1349 + }, + { + "epoch": 1.9536903039073805, + "grad_norm": 1.2607148885726929, + "learning_rate": 1.9710378117457764e-07, + "loss": 0.898, + "step": 1350 + }, + { + "epoch": 1.9536903039073805, + "eval_loss": 0.9950266480445862, + "eval_runtime": 51.3011, + "eval_samples_per_second": 19.493, + "eval_steps_per_second": 1.228, + "step": 1350 + }, + { + "epoch": 1.955137481910275, + "grad_norm": 1.384392261505127, + "learning_rate": 1.9308125502815768e-07, + "loss": 0.8628, + "step": 1351 + }, + { + "epoch": 1.9565846599131693, + "grad_norm": 1.2599904537200928, + "learning_rate": 1.8905872888173776e-07, + "loss": 0.8724, + "step": 1352 + }, + { + "epoch": 1.9580318379160637, + "grad_norm": 1.3314640522003174, + "learning_rate": 1.850362027353178e-07, + "loss": 0.8852, + "step": 1353 + }, + { + "epoch": 1.9594790159189581, + "grad_norm": 1.1998828649520874, + "learning_rate": 1.8101367658889785e-07, + "loss": 0.8944, + "step": 1354 + }, + { + "epoch": 1.9609261939218525, + "grad_norm": 1.3198102712631226, + "learning_rate": 1.7699115044247788e-07, + "loss": 0.8903, + "step": 1355 + }, + { + "epoch": 1.9623733719247467, + "grad_norm": 1.3015177249908447, + "learning_rate": 1.7296862429605792e-07, + "loss": 0.8491, + "step": 1356 + }, + { + "epoch": 1.963820549927641, + "grad_norm": 1.2806864976882935, + "learning_rate": 1.6894609814963797e-07, + "loss": 0.8736, + "step": 1357 + }, + { + "epoch": 1.9652677279305355, + "grad_norm": 1.3791565895080566, + "learning_rate": 1.6492357200321805e-07, + "loss": 0.8431, + "step": 1358 + }, + { + "epoch": 1.9667149059334297, + "grad_norm": 1.2173056602478027, + "learning_rate": 1.609010458567981e-07, + "loss": 0.8605, + "step": 1359 + }, + { + "epoch": 1.968162083936324, + "grad_norm": 1.4914180040359497, + "learning_rate": 1.5687851971037814e-07, + "loss": 0.8946, + "step": 1360 + }, + { + "epoch": 1.968162083936324, + "eval_loss": 0.9946564435958862, + "eval_runtime": 51.0911, + "eval_samples_per_second": 19.573, + "eval_steps_per_second": 1.233, + "step": 1360 + }, + { + "epoch": 1.9696092619392185, + "grad_norm": 1.304384708404541, + "learning_rate": 1.5285599356395817e-07, + "loss": 0.8555, + "step": 1361 + }, + { + "epoch": 1.9710564399421129, + "grad_norm": 1.3001254796981812, + "learning_rate": 1.4883346741753824e-07, + "loss": 0.8835, + "step": 1362 + }, + { + "epoch": 1.9725036179450073, + "grad_norm": 1.2931289672851562, + "learning_rate": 1.4481094127111826e-07, + "loss": 0.8629, + "step": 1363 + }, + { + "epoch": 1.9739507959479017, + "grad_norm": 1.4861146211624146, + "learning_rate": 1.407884151246983e-07, + "loss": 0.8893, + "step": 1364 + }, + { + "epoch": 1.975397973950796, + "grad_norm": 1.25104558467865, + "learning_rate": 1.3676588897827839e-07, + "loss": 0.8667, + "step": 1365 + }, + { + "epoch": 1.9768451519536903, + "grad_norm": 1.3300350904464722, + "learning_rate": 1.327433628318584e-07, + "loss": 0.8876, + "step": 1366 + }, + { + "epoch": 1.9782923299565847, + "grad_norm": 1.3712011575698853, + "learning_rate": 1.2872083668543846e-07, + "loss": 0.901, + "step": 1367 + }, + { + "epoch": 1.979739507959479, + "grad_norm": 1.506166934967041, + "learning_rate": 1.246983105390185e-07, + "loss": 0.8367, + "step": 1368 + }, + { + "epoch": 1.9811866859623732, + "grad_norm": 1.2580986022949219, + "learning_rate": 1.2067578439259855e-07, + "loss": 0.8745, + "step": 1369 + }, + { + "epoch": 1.9826338639652676, + "grad_norm": 1.2753760814666748, + "learning_rate": 1.1665325824617861e-07, + "loss": 0.9054, + "step": 1370 + }, + { + "epoch": 1.9826338639652676, + "eval_loss": 0.9950919151306152, + "eval_runtime": 51.3018, + "eval_samples_per_second": 19.493, + "eval_steps_per_second": 1.228, + "step": 1370 + }, + { + "epoch": 1.984081041968162, + "grad_norm": 1.415486216545105, + "learning_rate": 1.1263073209975866e-07, + "loss": 0.8624, + "step": 1371 + }, + { + "epoch": 1.9855282199710564, + "grad_norm": 1.392276406288147, + "learning_rate": 1.086082059533387e-07, + "loss": 0.8555, + "step": 1372 + }, + { + "epoch": 1.9869753979739508, + "grad_norm": 1.4250545501708984, + "learning_rate": 1.0458567980691876e-07, + "loss": 0.8836, + "step": 1373 + }, + { + "epoch": 1.9884225759768452, + "grad_norm": 1.3518198728561401, + "learning_rate": 1.0056315366049881e-07, + "loss": 0.8898, + "step": 1374 + }, + { + "epoch": 1.9898697539797396, + "grad_norm": 1.4271644353866577, + "learning_rate": 9.654062751407884e-08, + "loss": 0.8519, + "step": 1375 + }, + { + "epoch": 1.9913169319826338, + "grad_norm": 1.184145450592041, + "learning_rate": 9.25181013676589e-08, + "loss": 0.8463, + "step": 1376 + }, + { + "epoch": 1.9927641099855282, + "grad_norm": 1.3909846544265747, + "learning_rate": 8.849557522123894e-08, + "loss": 0.8725, + "step": 1377 + }, + { + "epoch": 1.9942112879884226, + "grad_norm": 1.4419491291046143, + "learning_rate": 8.447304907481899e-08, + "loss": 0.8771, + "step": 1378 + }, + { + "epoch": 1.9956584659913168, + "grad_norm": 1.290542721748352, + "learning_rate": 8.045052292839905e-08, + "loss": 0.8617, + "step": 1379 + }, + { + "epoch": 1.9971056439942112, + "grad_norm": 1.235100269317627, + "learning_rate": 7.642799678197908e-08, + "loss": 0.8566, + "step": 1380 + }, + { + "epoch": 1.9971056439942112, + "eval_loss": 0.9954107403755188, + "eval_runtime": 51.199, + "eval_samples_per_second": 19.532, + "eval_steps_per_second": 1.23, + "step": 1380 + }, + { + "epoch": 1.9985528219971056, + "grad_norm": 1.296491265296936, + "learning_rate": 7.240547063555913e-08, + "loss": 0.8786, + "step": 1381 + }, + { + "epoch": 2.0, + "grad_norm": 1.266065239906311, + "learning_rate": 6.838294448913919e-08, + "loss": 0.836, + "step": 1382 + } + ], + "logging_steps": 1, + "max_steps": 1382, + "num_input_tokens_seen": 0, + "num_train_epochs": 2, + "save_steps": 50, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 1.4408803727903293e+18, + "train_batch_size": 6, + "trial_name": null, + "trial_params": null +}