| { | |
| "best_metric": 2.0652668476104736, | |
| "best_model_checkpoint": "/home/sunggeunan/data/ICL/outputs/lora/SKIML-ICL_mrqa_nq_v3/Meta-Llama-3-8B-Instruct-unanswerable-3Q-0U-0C-qa_first/checkpoint-1215", | |
| "epoch": 2.99537750385208, | |
| "eval_steps": 500, | |
| "global_step": 1215, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.002465331278890601, | |
| "grad_norm": 0.37433016300201416, | |
| "learning_rate": 8.19672131147541e-09, | |
| "loss": 2.1235, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004930662557781202, | |
| "grad_norm": 0.36386838555336, | |
| "learning_rate": 1.639344262295082e-08, | |
| "loss": 2.1448, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.007395993836671803, | |
| "grad_norm": 0.37453725934028625, | |
| "learning_rate": 2.459016393442623e-08, | |
| "loss": 2.1666, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.009861325115562404, | |
| "grad_norm": 0.35461676120758057, | |
| "learning_rate": 3.278688524590164e-08, | |
| "loss": 2.1625, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.012326656394453005, | |
| "grad_norm": 0.37718096375465393, | |
| "learning_rate": 4.0983606557377046e-08, | |
| "loss": 2.0905, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.014791987673343606, | |
| "grad_norm": 0.3686259090900421, | |
| "learning_rate": 4.918032786885246e-08, | |
| "loss": 2.1175, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.017257318952234205, | |
| "grad_norm": 0.3776250183582306, | |
| "learning_rate": 5.7377049180327866e-08, | |
| "loss": 2.0978, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.019722650231124808, | |
| "grad_norm": 0.37626877427101135, | |
| "learning_rate": 6.557377049180328e-08, | |
| "loss": 2.0964, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.022187981510015407, | |
| "grad_norm": 0.39006781578063965, | |
| "learning_rate": 7.377049180327868e-08, | |
| "loss": 2.1656, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02465331278890601, | |
| "grad_norm": 0.3790552616119385, | |
| "learning_rate": 8.196721311475409e-08, | |
| "loss": 2.2033, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02711864406779661, | |
| "grad_norm": 0.3637031018733978, | |
| "learning_rate": 9.01639344262295e-08, | |
| "loss": 2.0818, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.029583975346687212, | |
| "grad_norm": 0.3852219581604004, | |
| "learning_rate": 9.836065573770492e-08, | |
| "loss": 2.1287, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03204930662557781, | |
| "grad_norm": 0.3391871154308319, | |
| "learning_rate": 1.0655737704918032e-07, | |
| "loss": 2.0188, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03451463790446841, | |
| "grad_norm": 0.33681735396385193, | |
| "learning_rate": 1.1475409836065573e-07, | |
| "loss": 2.0317, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03697996918335902, | |
| "grad_norm": 0.3790912330150604, | |
| "learning_rate": 1.2295081967213113e-07, | |
| "loss": 2.0958, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.039445300462249616, | |
| "grad_norm": 0.3681642711162567, | |
| "learning_rate": 1.3114754098360656e-07, | |
| "loss": 2.1073, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.041910631741140215, | |
| "grad_norm": 0.3428540825843811, | |
| "learning_rate": 1.3934426229508196e-07, | |
| "loss": 2.0882, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.044375963020030815, | |
| "grad_norm": 0.36673665046691895, | |
| "learning_rate": 1.4754098360655736e-07, | |
| "loss": 2.1117, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04684129429892142, | |
| "grad_norm": 0.3691534399986267, | |
| "learning_rate": 1.5573770491803278e-07, | |
| "loss": 2.1221, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.04930662557781202, | |
| "grad_norm": 0.3561345636844635, | |
| "learning_rate": 1.6393442622950818e-07, | |
| "loss": 2.0993, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05177195685670262, | |
| "grad_norm": 0.36137452721595764, | |
| "learning_rate": 1.7213114754098358e-07, | |
| "loss": 2.116, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.05423728813559322, | |
| "grad_norm": 0.37735700607299805, | |
| "learning_rate": 1.80327868852459e-07, | |
| "loss": 2.1092, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.05670261941448382, | |
| "grad_norm": 0.40573787689208984, | |
| "learning_rate": 1.885245901639344e-07, | |
| "loss": 2.1597, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.059167950693374424, | |
| "grad_norm": 0.3804572820663452, | |
| "learning_rate": 1.9672131147540984e-07, | |
| "loss": 2.1873, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.061633281972265024, | |
| "grad_norm": 0.35102495551109314, | |
| "learning_rate": 2.0491803278688524e-07, | |
| "loss": 2.0974, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.06409861325115562, | |
| "grad_norm": 0.39554598927497864, | |
| "learning_rate": 2.1311475409836064e-07, | |
| "loss": 2.0785, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.06656394453004623, | |
| "grad_norm": 0.3689061105251312, | |
| "learning_rate": 2.2131147540983606e-07, | |
| "loss": 2.1201, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.06902927580893682, | |
| "grad_norm": 0.36995020508766174, | |
| "learning_rate": 2.2950819672131146e-07, | |
| "loss": 2.134, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.07149460708782743, | |
| "grad_norm": 0.3756244480609894, | |
| "learning_rate": 2.3770491803278686e-07, | |
| "loss": 2.1297, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.07395993836671803, | |
| "grad_norm": 0.3423691987991333, | |
| "learning_rate": 2.4590163934426226e-07, | |
| "loss": 2.0344, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07642526964560863, | |
| "grad_norm": 0.3934348225593567, | |
| "learning_rate": 2.540983606557377e-07, | |
| "loss": 2.1493, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07889060092449923, | |
| "grad_norm": 0.3401438891887665, | |
| "learning_rate": 2.622950819672131e-07, | |
| "loss": 2.0759, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.08135593220338982, | |
| "grad_norm": 0.36425745487213135, | |
| "learning_rate": 2.704918032786885e-07, | |
| "loss": 2.0684, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.08382126348228043, | |
| "grad_norm": 0.369589626789093, | |
| "learning_rate": 2.786885245901639e-07, | |
| "loss": 2.1358, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.08628659476117104, | |
| "grad_norm": 0.3736041486263275, | |
| "learning_rate": 2.868852459016393e-07, | |
| "loss": 2.1095, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08875192604006163, | |
| "grad_norm": 0.3737315833568573, | |
| "learning_rate": 2.950819672131147e-07, | |
| "loss": 2.0754, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.09121725731895224, | |
| "grad_norm": 0.3834841251373291, | |
| "learning_rate": 3.0327868852459017e-07, | |
| "loss": 2.1772, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.09368258859784284, | |
| "grad_norm": 0.3807787001132965, | |
| "learning_rate": 3.1147540983606557e-07, | |
| "loss": 2.1241, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.09614791987673343, | |
| "grad_norm": 0.3752988576889038, | |
| "learning_rate": 3.1967213114754097e-07, | |
| "loss": 2.1427, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.09861325115562404, | |
| "grad_norm": 0.35566121339797974, | |
| "learning_rate": 3.2786885245901637e-07, | |
| "loss": 2.0672, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.10107858243451463, | |
| "grad_norm": 0.36831313371658325, | |
| "learning_rate": 3.3606557377049177e-07, | |
| "loss": 2.1271, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.10354391371340524, | |
| "grad_norm": 0.37055161595344543, | |
| "learning_rate": 3.4426229508196717e-07, | |
| "loss": 2.1146, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.10600924499229585, | |
| "grad_norm": 0.38187873363494873, | |
| "learning_rate": 3.524590163934426e-07, | |
| "loss": 2.1615, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.10847457627118644, | |
| "grad_norm": 0.3832800090312958, | |
| "learning_rate": 3.60655737704918e-07, | |
| "loss": 2.1479, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.11093990755007704, | |
| "grad_norm": 0.3809047043323517, | |
| "learning_rate": 3.6885245901639347e-07, | |
| "loss": 2.1329, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.11340523882896764, | |
| "grad_norm": 0.3516607880592346, | |
| "learning_rate": 3.770491803278688e-07, | |
| "loss": 2.104, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.11587057010785824, | |
| "grad_norm": 0.38060668110847473, | |
| "learning_rate": 3.852459016393442e-07, | |
| "loss": 2.0961, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.11833590138674885, | |
| "grad_norm": 0.3907550573348999, | |
| "learning_rate": 3.9344262295081967e-07, | |
| "loss": 2.0938, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.12080123266563944, | |
| "grad_norm": 0.35364964604377747, | |
| "learning_rate": 4.0163934426229507e-07, | |
| "loss": 2.0671, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.12326656394453005, | |
| "grad_norm": 0.3716958165168762, | |
| "learning_rate": 4.0983606557377047e-07, | |
| "loss": 2.1482, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12573189522342065, | |
| "grad_norm": 0.3951459228992462, | |
| "learning_rate": 4.180327868852459e-07, | |
| "loss": 2.0902, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.12819722650231125, | |
| "grad_norm": 0.4165395498275757, | |
| "learning_rate": 4.2622950819672127e-07, | |
| "loss": 2.1433, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.13066255778120184, | |
| "grad_norm": 0.38022342324256897, | |
| "learning_rate": 4.3442622950819667e-07, | |
| "loss": 2.0527, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.13312788906009246, | |
| "grad_norm": 0.3521603047847748, | |
| "learning_rate": 4.426229508196721e-07, | |
| "loss": 2.0865, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.13559322033898305, | |
| "grad_norm": 0.3616274297237396, | |
| "learning_rate": 4.508196721311475e-07, | |
| "loss": 2.1095, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.13805855161787364, | |
| "grad_norm": 0.3837212324142456, | |
| "learning_rate": 4.590163934426229e-07, | |
| "loss": 2.0693, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.14052388289676426, | |
| "grad_norm": 0.3840532600879669, | |
| "learning_rate": 4.672131147540984e-07, | |
| "loss": 2.0463, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.14298921417565486, | |
| "grad_norm": 0.3691370189189911, | |
| "learning_rate": 4.754098360655737e-07, | |
| "loss": 2.1356, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.14545454545454545, | |
| "grad_norm": 0.3823927342891693, | |
| "learning_rate": 4.836065573770492e-07, | |
| "loss": 2.1318, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.14791987673343607, | |
| "grad_norm": 0.39722442626953125, | |
| "learning_rate": 4.918032786885245e-07, | |
| "loss": 2.1507, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.15038520801232666, | |
| "grad_norm": 0.406395822763443, | |
| "learning_rate": 5e-07, | |
| "loss": 2.1622, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.15285053929121725, | |
| "grad_norm": 0.3791063129901886, | |
| "learning_rate": 4.995667244367418e-07, | |
| "loss": 2.022, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.15531587057010784, | |
| "grad_norm": 0.3658573627471924, | |
| "learning_rate": 4.991334488734835e-07, | |
| "loss": 2.1009, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.15778120184899846, | |
| "grad_norm": 0.3828792870044708, | |
| "learning_rate": 4.987001733102253e-07, | |
| "loss": 2.1396, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.16024653312788906, | |
| "grad_norm": 0.3802504539489746, | |
| "learning_rate": 4.982668977469671e-07, | |
| "loss": 2.0973, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.16271186440677965, | |
| "grad_norm": 0.3841285705566406, | |
| "learning_rate": 4.978336221837089e-07, | |
| "loss": 2.1172, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.16517719568567027, | |
| "grad_norm": 0.392628937959671, | |
| "learning_rate": 4.974003466204506e-07, | |
| "loss": 2.1292, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.16764252696456086, | |
| "grad_norm": 0.3984483480453491, | |
| "learning_rate": 4.969670710571924e-07, | |
| "loss": 2.1439, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.17010785824345145, | |
| "grad_norm": 0.38579827547073364, | |
| "learning_rate": 4.96533795493934e-07, | |
| "loss": 2.0981, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.17257318952234207, | |
| "grad_norm": 0.3777226507663727, | |
| "learning_rate": 4.961005199306758e-07, | |
| "loss": 2.0915, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17503852080123267, | |
| "grad_norm": 0.3962363302707672, | |
| "learning_rate": 4.956672443674176e-07, | |
| "loss": 2.0782, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.17750385208012326, | |
| "grad_norm": 0.3745022416114807, | |
| "learning_rate": 4.952339688041594e-07, | |
| "loss": 2.0846, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.17996918335901388, | |
| "grad_norm": 0.3679901361465454, | |
| "learning_rate": 4.948006932409011e-07, | |
| "loss": 2.0548, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.18243451463790447, | |
| "grad_norm": 0.3869926929473877, | |
| "learning_rate": 4.943674176776429e-07, | |
| "loss": 2.1415, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.18489984591679506, | |
| "grad_norm": 0.40461650490760803, | |
| "learning_rate": 4.939341421143847e-07, | |
| "loss": 2.0748, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.18736517719568568, | |
| "grad_norm": 0.39631369709968567, | |
| "learning_rate": 4.935008665511265e-07, | |
| "loss": 2.1064, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.18983050847457628, | |
| "grad_norm": 0.3528262674808502, | |
| "learning_rate": 4.930675909878682e-07, | |
| "loss": 2.0709, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.19229583975346687, | |
| "grad_norm": 0.40359705686569214, | |
| "learning_rate": 4.9263431542461e-07, | |
| "loss": 2.1111, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.19476117103235746, | |
| "grad_norm": 0.39586734771728516, | |
| "learning_rate": 4.922010398613518e-07, | |
| "loss": 2.1154, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.19722650231124808, | |
| "grad_norm": 0.4013460576534271, | |
| "learning_rate": 4.917677642980936e-07, | |
| "loss": 2.1619, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19969183359013867, | |
| "grad_norm": 0.4224221110343933, | |
| "learning_rate": 4.913344887348353e-07, | |
| "loss": 2.199, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.20215716486902927, | |
| "grad_norm": 0.40476611256599426, | |
| "learning_rate": 4.909012131715771e-07, | |
| "loss": 2.0757, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.20462249614791989, | |
| "grad_norm": 0.3916946351528168, | |
| "learning_rate": 4.904679376083189e-07, | |
| "loss": 2.0717, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.20708782742681048, | |
| "grad_norm": 0.4132992923259735, | |
| "learning_rate": 4.900346620450606e-07, | |
| "loss": 2.0941, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.20955315870570107, | |
| "grad_norm": 0.42981430888175964, | |
| "learning_rate": 4.896013864818024e-07, | |
| "loss": 2.0938, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.2120184899845917, | |
| "grad_norm": 0.3837836980819702, | |
| "learning_rate": 4.891681109185442e-07, | |
| "loss": 2.0832, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.21448382126348228, | |
| "grad_norm": 0.4250790476799011, | |
| "learning_rate": 4.887348353552859e-07, | |
| "loss": 2.1589, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.21694915254237288, | |
| "grad_norm": 0.3854515552520752, | |
| "learning_rate": 4.883015597920277e-07, | |
| "loss": 2.0966, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.2194144838212635, | |
| "grad_norm": 0.3787481486797333, | |
| "learning_rate": 4.878682842287695e-07, | |
| "loss": 2.0971, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2218798151001541, | |
| "grad_norm": 0.40984106063842773, | |
| "learning_rate": 4.874350086655113e-07, | |
| "loss": 2.1721, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.22434514637904468, | |
| "grad_norm": 0.4011061191558838, | |
| "learning_rate": 4.87001733102253e-07, | |
| "loss": 2.1267, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.22681047765793527, | |
| "grad_norm": 0.4035578668117523, | |
| "learning_rate": 4.865684575389948e-07, | |
| "loss": 2.1888, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2292758089368259, | |
| "grad_norm": 0.38300806283950806, | |
| "learning_rate": 4.861351819757366e-07, | |
| "loss": 2.0984, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.23174114021571648, | |
| "grad_norm": 0.39732295274734497, | |
| "learning_rate": 4.857019064124784e-07, | |
| "loss": 2.0648, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.23420647149460708, | |
| "grad_norm": 0.4041942059993744, | |
| "learning_rate": 4.8526863084922e-07, | |
| "loss": 2.1297, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2366718027734977, | |
| "grad_norm": 0.4050208032131195, | |
| "learning_rate": 4.848353552859618e-07, | |
| "loss": 2.0902, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.2391371340523883, | |
| "grad_norm": 0.3880615532398224, | |
| "learning_rate": 4.844020797227036e-07, | |
| "loss": 2.0922, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.24160246533127888, | |
| "grad_norm": 0.43308451771736145, | |
| "learning_rate": 4.839688041594454e-07, | |
| "loss": 2.1079, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.2440677966101695, | |
| "grad_norm": 0.416716068983078, | |
| "learning_rate": 4.835355285961871e-07, | |
| "loss": 2.1053, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.2465331278890601, | |
| "grad_norm": 0.38583382964134216, | |
| "learning_rate": 4.831022530329289e-07, | |
| "loss": 2.0907, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2489984591679507, | |
| "grad_norm": 0.4200878143310547, | |
| "learning_rate": 4.826689774696706e-07, | |
| "loss": 2.0973, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.2514637904468413, | |
| "grad_norm": 0.41719701886177063, | |
| "learning_rate": 4.822357019064124e-07, | |
| "loss": 2.0836, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.2539291217257319, | |
| "grad_norm": 0.40001362562179565, | |
| "learning_rate": 4.818024263431542e-07, | |
| "loss": 2.1162, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.2563944530046225, | |
| "grad_norm": 0.40380966663360596, | |
| "learning_rate": 4.81369150779896e-07, | |
| "loss": 2.1471, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.2588597842835131, | |
| "grad_norm": 0.3940064311027527, | |
| "learning_rate": 4.809358752166377e-07, | |
| "loss": 2.0997, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.2613251155624037, | |
| "grad_norm": 0.41269171237945557, | |
| "learning_rate": 4.805025996533795e-07, | |
| "loss": 2.1218, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.2637904468412943, | |
| "grad_norm": 0.422402024269104, | |
| "learning_rate": 4.800693240901213e-07, | |
| "loss": 2.1381, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.2662557781201849, | |
| "grad_norm": 0.40187913179397583, | |
| "learning_rate": 4.796360485268631e-07, | |
| "loss": 2.0733, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.2687211093990755, | |
| "grad_norm": 0.43270590901374817, | |
| "learning_rate": 4.792027729636048e-07, | |
| "loss": 2.1131, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.2711864406779661, | |
| "grad_norm": 0.42297425866127014, | |
| "learning_rate": 4.787694974003466e-07, | |
| "loss": 2.1376, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2736517719568567, | |
| "grad_norm": 0.4329090118408203, | |
| "learning_rate": 4.783362218370884e-07, | |
| "loss": 2.1567, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.2761171032357473, | |
| "grad_norm": 0.4003148674964905, | |
| "learning_rate": 4.779029462738302e-07, | |
| "loss": 2.0797, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.2785824345146379, | |
| "grad_norm": 0.4078727960586548, | |
| "learning_rate": 4.774696707105719e-07, | |
| "loss": 2.1282, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.2810477657935285, | |
| "grad_norm": 0.41799014806747437, | |
| "learning_rate": 4.770363951473137e-07, | |
| "loss": 2.1333, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.2835130970724191, | |
| "grad_norm": 0.40705522894859314, | |
| "learning_rate": 4.7660311958405543e-07, | |
| "loss": 2.0916, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2859784283513097, | |
| "grad_norm": 0.392926961183548, | |
| "learning_rate": 4.7616984402079717e-07, | |
| "loss": 2.0652, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.2884437596302003, | |
| "grad_norm": 0.4167540669441223, | |
| "learning_rate": 4.7573656845753897e-07, | |
| "loss": 2.1324, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.2909090909090909, | |
| "grad_norm": 0.4212091267108917, | |
| "learning_rate": 4.753032928942807e-07, | |
| "loss": 2.0937, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2933744221879815, | |
| "grad_norm": 0.41585585474967957, | |
| "learning_rate": 4.748700173310225e-07, | |
| "loss": 2.1307, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.29583975346687214, | |
| "grad_norm": 0.382997065782547, | |
| "learning_rate": 4.7443674176776426e-07, | |
| "loss": 2.0076, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2983050847457627, | |
| "grad_norm": 0.44930651783943176, | |
| "learning_rate": 4.7400346620450606e-07, | |
| "loss": 2.1763, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.3007704160246533, | |
| "grad_norm": 0.41596314311027527, | |
| "learning_rate": 4.735701906412478e-07, | |
| "loss": 2.1102, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.3032357473035439, | |
| "grad_norm": 0.44001415371894836, | |
| "learning_rate": 4.731369150779896e-07, | |
| "loss": 2.1505, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.3057010785824345, | |
| "grad_norm": 0.3971948027610779, | |
| "learning_rate": 4.7270363951473134e-07, | |
| "loss": 2.0609, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.3081664098613251, | |
| "grad_norm": 0.4413090944290161, | |
| "learning_rate": 4.7227036395147314e-07, | |
| "loss": 2.0882, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3106317411402157, | |
| "grad_norm": 0.43514329195022583, | |
| "learning_rate": 4.718370883882149e-07, | |
| "loss": 2.1229, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.31309707241910634, | |
| "grad_norm": 0.40400221943855286, | |
| "learning_rate": 4.714038128249567e-07, | |
| "loss": 2.1171, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.31556240369799693, | |
| "grad_norm": 0.3838646411895752, | |
| "learning_rate": 4.7097053726169843e-07, | |
| "loss": 2.0802, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.3180277349768875, | |
| "grad_norm": 0.4001474678516388, | |
| "learning_rate": 4.7053726169844023e-07, | |
| "loss": 2.0363, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.3204930662557781, | |
| "grad_norm": 0.4248114824295044, | |
| "learning_rate": 4.701039861351819e-07, | |
| "loss": 2.1385, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3229583975346687, | |
| "grad_norm": 0.4382472336292267, | |
| "learning_rate": 4.696707105719237e-07, | |
| "loss": 2.1422, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.3254237288135593, | |
| "grad_norm": 0.4617295265197754, | |
| "learning_rate": 4.6923743500866546e-07, | |
| "loss": 2.0857, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.32788906009244995, | |
| "grad_norm": 0.451002836227417, | |
| "learning_rate": 4.6880415944540726e-07, | |
| "loss": 2.1353, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.33035439137134054, | |
| "grad_norm": 0.4292704164981842, | |
| "learning_rate": 4.68370883882149e-07, | |
| "loss": 2.0632, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.33281972265023113, | |
| "grad_norm": 0.4048818349838257, | |
| "learning_rate": 4.679376083188908e-07, | |
| "loss": 2.1017, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3352850539291217, | |
| "grad_norm": 0.4577648639678955, | |
| "learning_rate": 4.6750433275563255e-07, | |
| "loss": 2.1834, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.3377503852080123, | |
| "grad_norm": 0.42989662289619446, | |
| "learning_rate": 4.6707105719237435e-07, | |
| "loss": 2.0785, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.3402157164869029, | |
| "grad_norm": 0.41798529028892517, | |
| "learning_rate": 4.666377816291161e-07, | |
| "loss": 2.0326, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.34268104776579356, | |
| "grad_norm": 0.4211689233779907, | |
| "learning_rate": 4.662045060658579e-07, | |
| "loss": 2.0613, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.34514637904468415, | |
| "grad_norm": 0.4358021020889282, | |
| "learning_rate": 4.6577123050259963e-07, | |
| "loss": 2.0944, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.34761171032357474, | |
| "grad_norm": 0.44546744227409363, | |
| "learning_rate": 4.6533795493934143e-07, | |
| "loss": 2.1474, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.35007704160246533, | |
| "grad_norm": 0.4270431399345398, | |
| "learning_rate": 4.649046793760832e-07, | |
| "loss": 2.1295, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.3525423728813559, | |
| "grad_norm": 0.4490653872489929, | |
| "learning_rate": 4.64471403812825e-07, | |
| "loss": 2.0991, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.3550077041602465, | |
| "grad_norm": 0.431923508644104, | |
| "learning_rate": 4.640381282495667e-07, | |
| "loss": 2.1416, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.3574730354391371, | |
| "grad_norm": 0.4275124967098236, | |
| "learning_rate": 4.6360485268630847e-07, | |
| "loss": 2.1094, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.35993836671802776, | |
| "grad_norm": 0.4312582015991211, | |
| "learning_rate": 4.631715771230502e-07, | |
| "loss": 2.1345, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.36240369799691835, | |
| "grad_norm": 0.4102493226528168, | |
| "learning_rate": 4.62738301559792e-07, | |
| "loss": 2.1765, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.36486902927580894, | |
| "grad_norm": 0.4305970072746277, | |
| "learning_rate": 4.6230502599653375e-07, | |
| "loss": 2.0682, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.36733436055469953, | |
| "grad_norm": 0.4102775752544403, | |
| "learning_rate": 4.6187175043327555e-07, | |
| "loss": 2.0501, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.3697996918335901, | |
| "grad_norm": 0.41851964592933655, | |
| "learning_rate": 4.614384748700173e-07, | |
| "loss": 2.0649, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3722650231124807, | |
| "grad_norm": 0.43330028653144836, | |
| "learning_rate": 4.610051993067591e-07, | |
| "loss": 2.1693, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.37473035439137137, | |
| "grad_norm": 0.4430111050605774, | |
| "learning_rate": 4.6057192374350084e-07, | |
| "loss": 2.0567, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.37719568567026196, | |
| "grad_norm": 0.42955276370048523, | |
| "learning_rate": 4.6013864818024264e-07, | |
| "loss": 2.0635, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.37966101694915255, | |
| "grad_norm": 0.4428994357585907, | |
| "learning_rate": 4.597053726169844e-07, | |
| "loss": 2.1036, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.38212634822804314, | |
| "grad_norm": 0.4334391951560974, | |
| "learning_rate": 4.592720970537262e-07, | |
| "loss": 2.1253, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.38459167950693374, | |
| "grad_norm": 0.42987290024757385, | |
| "learning_rate": 4.588388214904679e-07, | |
| "loss": 2.1372, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.38705701078582433, | |
| "grad_norm": 0.4356411397457123, | |
| "learning_rate": 4.584055459272097e-07, | |
| "loss": 2.1321, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.3895223420647149, | |
| "grad_norm": 0.419073224067688, | |
| "learning_rate": 4.5797227036395147e-07, | |
| "loss": 2.0334, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.39198767334360557, | |
| "grad_norm": 0.4288201928138733, | |
| "learning_rate": 4.5753899480069327e-07, | |
| "loss": 2.1144, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.39445300462249616, | |
| "grad_norm": 0.4334442615509033, | |
| "learning_rate": 4.5710571923743496e-07, | |
| "loss": 2.1439, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.39691833590138675, | |
| "grad_norm": 0.442800372838974, | |
| "learning_rate": 4.5667244367417676e-07, | |
| "loss": 2.1069, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.39938366718027735, | |
| "grad_norm": 0.4469603896141052, | |
| "learning_rate": 4.562391681109185e-07, | |
| "loss": 2.0845, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.40184899845916794, | |
| "grad_norm": 0.4325008988380432, | |
| "learning_rate": 4.558058925476603e-07, | |
| "loss": 2.1111, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.40431432973805853, | |
| "grad_norm": 0.4402778744697571, | |
| "learning_rate": 4.5537261698440204e-07, | |
| "loss": 2.088, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.4067796610169492, | |
| "grad_norm": 0.42320767045021057, | |
| "learning_rate": 4.5493934142114384e-07, | |
| "loss": 2.0505, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.40924499229583977, | |
| "grad_norm": 0.4501608610153198, | |
| "learning_rate": 4.545060658578856e-07, | |
| "loss": 2.1184, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.41171032357473036, | |
| "grad_norm": 0.42072442173957825, | |
| "learning_rate": 4.540727902946274e-07, | |
| "loss": 2.0049, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.41417565485362096, | |
| "grad_norm": 0.47302156686782837, | |
| "learning_rate": 4.5363951473136913e-07, | |
| "loss": 2.1615, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.41664098613251155, | |
| "grad_norm": 0.43148499727249146, | |
| "learning_rate": 4.5320623916811093e-07, | |
| "loss": 2.1874, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.41910631741140214, | |
| "grad_norm": 0.43476980924606323, | |
| "learning_rate": 4.527729636048527e-07, | |
| "loss": 2.0885, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.42157164869029273, | |
| "grad_norm": 0.4497949779033661, | |
| "learning_rate": 4.5233968804159447e-07, | |
| "loss": 2.055, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.4240369799691834, | |
| "grad_norm": 0.443490594625473, | |
| "learning_rate": 4.519064124783362e-07, | |
| "loss": 2.1033, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.426502311248074, | |
| "grad_norm": 0.44823795557022095, | |
| "learning_rate": 4.51473136915078e-07, | |
| "loss": 2.0738, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.42896764252696457, | |
| "grad_norm": 0.45069214701652527, | |
| "learning_rate": 4.5103986135181976e-07, | |
| "loss": 2.103, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.43143297380585516, | |
| "grad_norm": 0.4625852406024933, | |
| "learning_rate": 4.5060658578856145e-07, | |
| "loss": 2.1873, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.43389830508474575, | |
| "grad_norm": 0.43068787455558777, | |
| "learning_rate": 4.5017331022530325e-07, | |
| "loss": 1.9754, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.43636363636363634, | |
| "grad_norm": 0.4397013783454895, | |
| "learning_rate": 4.49740034662045e-07, | |
| "loss": 2.1289, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.438828967642527, | |
| "grad_norm": 0.4777672588825226, | |
| "learning_rate": 4.493067590987868e-07, | |
| "loss": 2.0847, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.4412942989214176, | |
| "grad_norm": 0.4455231726169586, | |
| "learning_rate": 4.4887348353552854e-07, | |
| "loss": 2.1268, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.4437596302003082, | |
| "grad_norm": 0.43128734827041626, | |
| "learning_rate": 4.4844020797227034e-07, | |
| "loss": 2.0711, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.44622496147919877, | |
| "grad_norm": 0.44358691573143005, | |
| "learning_rate": 4.480069324090121e-07, | |
| "loss": 2.1138, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.44869029275808936, | |
| "grad_norm": 0.43691667914390564, | |
| "learning_rate": 4.475736568457539e-07, | |
| "loss": 2.0992, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.45115562403697995, | |
| "grad_norm": 0.47155120968818665, | |
| "learning_rate": 4.471403812824956e-07, | |
| "loss": 2.1712, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.45362095531587054, | |
| "grad_norm": 0.43848317861557007, | |
| "learning_rate": 4.467071057192374e-07, | |
| "loss": 2.0718, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.4560862865947612, | |
| "grad_norm": 0.4509446918964386, | |
| "learning_rate": 4.4627383015597917e-07, | |
| "loss": 2.0888, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.4585516178736518, | |
| "grad_norm": 0.4379338324069977, | |
| "learning_rate": 4.4584055459272096e-07, | |
| "loss": 2.1421, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.4610169491525424, | |
| "grad_norm": 0.43303945660591125, | |
| "learning_rate": 4.454072790294627e-07, | |
| "loss": 2.1286, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.46348228043143297, | |
| "grad_norm": 0.44023025035858154, | |
| "learning_rate": 4.449740034662045e-07, | |
| "loss": 2.1176, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.46594761171032356, | |
| "grad_norm": 0.46459197998046875, | |
| "learning_rate": 4.4454072790294625e-07, | |
| "loss": 2.1093, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.46841294298921415, | |
| "grad_norm": 0.45678266882896423, | |
| "learning_rate": 4.4410745233968805e-07, | |
| "loss": 2.1429, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.4708782742681048, | |
| "grad_norm": 0.47240617871284485, | |
| "learning_rate": 4.4367417677642974e-07, | |
| "loss": 2.119, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.4733436055469954, | |
| "grad_norm": 0.4273033142089844, | |
| "learning_rate": 4.4324090121317154e-07, | |
| "loss": 2.0333, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.475808936825886, | |
| "grad_norm": 0.4396463632583618, | |
| "learning_rate": 4.428076256499133e-07, | |
| "loss": 2.1181, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.4782742681047766, | |
| "grad_norm": 0.4450843334197998, | |
| "learning_rate": 4.423743500866551e-07, | |
| "loss": 2.1118, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.48073959938366717, | |
| "grad_norm": 0.4202272295951843, | |
| "learning_rate": 4.4194107452339683e-07, | |
| "loss": 2.0234, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.48320493066255776, | |
| "grad_norm": 0.48180142045021057, | |
| "learning_rate": 4.4150779896013863e-07, | |
| "loss": 2.1127, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.48567026194144836, | |
| "grad_norm": 0.4243899881839752, | |
| "learning_rate": 4.4107452339688037e-07, | |
| "loss": 2.0899, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.488135593220339, | |
| "grad_norm": 0.47803816199302673, | |
| "learning_rate": 4.4064124783362217e-07, | |
| "loss": 2.1038, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.4906009244992296, | |
| "grad_norm": 0.4795818030834198, | |
| "learning_rate": 4.402079722703639e-07, | |
| "loss": 2.1029, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.4930662557781202, | |
| "grad_norm": 0.46659576892852783, | |
| "learning_rate": 4.397746967071057e-07, | |
| "loss": 2.1318, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4955315870570108, | |
| "grad_norm": 0.46830591559410095, | |
| "learning_rate": 4.3934142114384746e-07, | |
| "loss": 2.1385, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.4979969183359014, | |
| "grad_norm": 0.4268909692764282, | |
| "learning_rate": 4.3890814558058926e-07, | |
| "loss": 2.0493, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.500462249614792, | |
| "grad_norm": 0.4640822112560272, | |
| "learning_rate": 4.38474870017331e-07, | |
| "loss": 2.1036, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.5029275808936826, | |
| "grad_norm": 0.46667805314064026, | |
| "learning_rate": 4.380415944540728e-07, | |
| "loss": 2.0978, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.5053929121725732, | |
| "grad_norm": 0.4559430778026581, | |
| "learning_rate": 4.3760831889081454e-07, | |
| "loss": 2.1081, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.5078582434514638, | |
| "grad_norm": 0.46032530069351196, | |
| "learning_rate": 4.371750433275563e-07, | |
| "loss": 2.1576, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.5103235747303544, | |
| "grad_norm": 0.46327170729637146, | |
| "learning_rate": 4.3674176776429803e-07, | |
| "loss": 2.1022, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.512788906009245, | |
| "grad_norm": 0.4393162429332733, | |
| "learning_rate": 4.3630849220103983e-07, | |
| "loss": 2.056, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.5152542372881356, | |
| "grad_norm": 0.4648692011833191, | |
| "learning_rate": 4.358752166377816e-07, | |
| "loss": 2.1021, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.5177195685670262, | |
| "grad_norm": 0.4504803419113159, | |
| "learning_rate": 4.354419410745234e-07, | |
| "loss": 2.1482, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.5201848998459168, | |
| "grad_norm": 0.46734288334846497, | |
| "learning_rate": 4.350086655112651e-07, | |
| "loss": 2.0916, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.5226502311248074, | |
| "grad_norm": 0.4660179018974304, | |
| "learning_rate": 4.345753899480069e-07, | |
| "loss": 2.1399, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.5251155624036979, | |
| "grad_norm": 0.4509792625904083, | |
| "learning_rate": 4.3414211438474866e-07, | |
| "loss": 2.0583, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.5275808936825886, | |
| "grad_norm": 0.4666799306869507, | |
| "learning_rate": 4.3370883882149046e-07, | |
| "loss": 2.0465, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.5300462249614792, | |
| "grad_norm": 0.4598578214645386, | |
| "learning_rate": 4.332755632582322e-07, | |
| "loss": 2.0943, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.5325115562403698, | |
| "grad_norm": 0.45499753952026367, | |
| "learning_rate": 4.32842287694974e-07, | |
| "loss": 2.1261, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.5349768875192604, | |
| "grad_norm": 0.4656490683555603, | |
| "learning_rate": 4.3240901213171575e-07, | |
| "loss": 2.0735, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.537442218798151, | |
| "grad_norm": 0.46361130475997925, | |
| "learning_rate": 4.3197573656845755e-07, | |
| "loss": 2.0971, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.5399075500770416, | |
| "grad_norm": 0.45737481117248535, | |
| "learning_rate": 4.315424610051993e-07, | |
| "loss": 2.1334, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.5423728813559322, | |
| "grad_norm": 0.4745784401893616, | |
| "learning_rate": 4.311091854419411e-07, | |
| "loss": 2.0847, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.5448382126348228, | |
| "grad_norm": 0.4402310848236084, | |
| "learning_rate": 4.306759098786828e-07, | |
| "loss": 2.0876, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.5473035439137134, | |
| "grad_norm": 0.49000170826911926, | |
| "learning_rate": 4.302426343154246e-07, | |
| "loss": 2.1435, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.549768875192604, | |
| "grad_norm": 0.46280747652053833, | |
| "learning_rate": 4.298093587521663e-07, | |
| "loss": 2.1426, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.5522342064714946, | |
| "grad_norm": 0.45368483662605286, | |
| "learning_rate": 4.293760831889081e-07, | |
| "loss": 2.0494, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.5546995377503852, | |
| "grad_norm": 0.4530903995037079, | |
| "learning_rate": 4.2894280762564987e-07, | |
| "loss": 2.037, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5571648690292758, | |
| "grad_norm": 0.48510506749153137, | |
| "learning_rate": 4.2850953206239167e-07, | |
| "loss": 2.0678, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.5596302003081665, | |
| "grad_norm": 0.465676873922348, | |
| "learning_rate": 4.280762564991334e-07, | |
| "loss": 2.1838, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.562095531587057, | |
| "grad_norm": 0.4515366852283478, | |
| "learning_rate": 4.276429809358752e-07, | |
| "loss": 2.1176, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.5645608628659476, | |
| "grad_norm": 0.49126410484313965, | |
| "learning_rate": 4.2720970537261695e-07, | |
| "loss": 2.1086, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.5670261941448382, | |
| "grad_norm": 0.5021019577980042, | |
| "learning_rate": 4.2677642980935875e-07, | |
| "loss": 2.0984, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5694915254237288, | |
| "grad_norm": 0.48141247034072876, | |
| "learning_rate": 4.263431542461005e-07, | |
| "loss": 2.1249, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.5719568567026194, | |
| "grad_norm": 0.45555227994918823, | |
| "learning_rate": 4.259098786828423e-07, | |
| "loss": 2.1029, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.57442218798151, | |
| "grad_norm": 0.46432173252105713, | |
| "learning_rate": 4.2547660311958404e-07, | |
| "loss": 2.099, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.5768875192604006, | |
| "grad_norm": 0.4571111798286438, | |
| "learning_rate": 4.2504332755632584e-07, | |
| "loss": 2.0792, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.5793528505392912, | |
| "grad_norm": 0.47881045937538147, | |
| "learning_rate": 4.246100519930676e-07, | |
| "loss": 2.091, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.5818181818181818, | |
| "grad_norm": 0.47099265456199646, | |
| "learning_rate": 4.2417677642980933e-07, | |
| "loss": 2.101, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.5842835130970724, | |
| "grad_norm": 0.49213889241218567, | |
| "learning_rate": 4.2374350086655107e-07, | |
| "loss": 2.1439, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.586748844375963, | |
| "grad_norm": 0.45399606227874756, | |
| "learning_rate": 4.2331022530329287e-07, | |
| "loss": 2.099, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.5892141756548536, | |
| "grad_norm": 0.48907190561294556, | |
| "learning_rate": 4.228769497400346e-07, | |
| "loss": 2.173, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.5916795069337443, | |
| "grad_norm": 0.47644051909446716, | |
| "learning_rate": 4.224436741767764e-07, | |
| "loss": 2.0897, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5941448382126349, | |
| "grad_norm": 0.4855840802192688, | |
| "learning_rate": 4.2201039861351816e-07, | |
| "loss": 2.1058, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.5966101694915255, | |
| "grad_norm": 0.4714296758174896, | |
| "learning_rate": 4.2157712305025996e-07, | |
| "loss": 2.1442, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.599075500770416, | |
| "grad_norm": 0.478901207447052, | |
| "learning_rate": 4.211438474870017e-07, | |
| "loss": 2.0779, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.6015408320493066, | |
| "grad_norm": 0.4762595295906067, | |
| "learning_rate": 4.207105719237435e-07, | |
| "loss": 2.1094, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.6040061633281972, | |
| "grad_norm": 0.4861314594745636, | |
| "learning_rate": 4.2027729636048525e-07, | |
| "loss": 2.1391, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.6064714946070878, | |
| "grad_norm": 0.46600624918937683, | |
| "learning_rate": 4.1984402079722704e-07, | |
| "loss": 2.0362, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.6089368258859784, | |
| "grad_norm": 0.46670466661453247, | |
| "learning_rate": 4.194107452339688e-07, | |
| "loss": 2.11, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.611402157164869, | |
| "grad_norm": 0.45662227272987366, | |
| "learning_rate": 4.189774696707106e-07, | |
| "loss": 2.0349, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.6138674884437596, | |
| "grad_norm": 0.4674631357192993, | |
| "learning_rate": 4.1854419410745233e-07, | |
| "loss": 2.1652, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.6163328197226502, | |
| "grad_norm": 0.4665353000164032, | |
| "learning_rate": 4.1811091854419413e-07, | |
| "loss": 2.0729, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.6187981510015408, | |
| "grad_norm": 0.48149293661117554, | |
| "learning_rate": 4.176776429809358e-07, | |
| "loss": 2.176, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.6212634822804314, | |
| "grad_norm": 0.48049747943878174, | |
| "learning_rate": 4.172443674176776e-07, | |
| "loss": 2.0839, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.6237288135593221, | |
| "grad_norm": 0.47829097509384155, | |
| "learning_rate": 4.1681109185441936e-07, | |
| "loss": 2.1457, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.6261941448382127, | |
| "grad_norm": 0.493964284658432, | |
| "learning_rate": 4.1637781629116116e-07, | |
| "loss": 2.085, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.6286594761171033, | |
| "grad_norm": 0.462367981672287, | |
| "learning_rate": 4.159445407279029e-07, | |
| "loss": 2.1344, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.6311248073959939, | |
| "grad_norm": 0.5168716907501221, | |
| "learning_rate": 4.155112651646447e-07, | |
| "loss": 2.1085, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.6335901386748845, | |
| "grad_norm": 0.456626832485199, | |
| "learning_rate": 4.1507798960138645e-07, | |
| "loss": 2.0953, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.636055469953775, | |
| "grad_norm": 0.5292502045631409, | |
| "learning_rate": 4.1464471403812825e-07, | |
| "loss": 2.0891, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.6385208012326656, | |
| "grad_norm": 0.4995521903038025, | |
| "learning_rate": 4.1421143847487e-07, | |
| "loss": 2.1328, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.6409861325115562, | |
| "grad_norm": 0.4320158362388611, | |
| "learning_rate": 4.137781629116118e-07, | |
| "loss": 1.9853, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.6434514637904468, | |
| "grad_norm": 0.47814425826072693, | |
| "learning_rate": 4.1334488734835354e-07, | |
| "loss": 2.0682, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.6459167950693374, | |
| "grad_norm": 0.4769778251647949, | |
| "learning_rate": 4.1291161178509533e-07, | |
| "loss": 2.0443, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.648382126348228, | |
| "grad_norm": 0.4876897633075714, | |
| "learning_rate": 4.124783362218371e-07, | |
| "loss": 2.0205, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.6508474576271186, | |
| "grad_norm": 0.4950641989707947, | |
| "learning_rate": 4.120450606585789e-07, | |
| "loss": 2.0841, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.6533127889060092, | |
| "grad_norm": 0.49192503094673157, | |
| "learning_rate": 4.116117850953206e-07, | |
| "loss": 2.0997, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.6557781201848999, | |
| "grad_norm": 0.49646303057670593, | |
| "learning_rate": 4.1117850953206237e-07, | |
| "loss": 2.14, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.6582434514637905, | |
| "grad_norm": 0.45144787430763245, | |
| "learning_rate": 4.107452339688041e-07, | |
| "loss": 2.0162, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.6607087827426811, | |
| "grad_norm": 0.4697396457195282, | |
| "learning_rate": 4.103119584055459e-07, | |
| "loss": 2.0876, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.6631741140215717, | |
| "grad_norm": 0.4968138337135315, | |
| "learning_rate": 4.0987868284228766e-07, | |
| "loss": 2.1535, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.6656394453004623, | |
| "grad_norm": 0.5264018177986145, | |
| "learning_rate": 4.0944540727902945e-07, | |
| "loss": 2.1838, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6681047765793529, | |
| "grad_norm": 0.505137026309967, | |
| "learning_rate": 4.090121317157712e-07, | |
| "loss": 2.1095, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.6705701078582434, | |
| "grad_norm": 0.4924645721912384, | |
| "learning_rate": 4.08578856152513e-07, | |
| "loss": 2.1105, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.673035439137134, | |
| "grad_norm": 0.4794199764728546, | |
| "learning_rate": 4.0814558058925474e-07, | |
| "loss": 2.1246, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.6755007704160246, | |
| "grad_norm": 0.4818281829357147, | |
| "learning_rate": 4.0771230502599654e-07, | |
| "loss": 2.069, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.6779661016949152, | |
| "grad_norm": 0.5034011602401733, | |
| "learning_rate": 4.072790294627383e-07, | |
| "loss": 2.1568, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6804314329738058, | |
| "grad_norm": 0.49432459473609924, | |
| "learning_rate": 4.068457538994801e-07, | |
| "loss": 2.1297, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.6828967642526964, | |
| "grad_norm": 0.4923751652240753, | |
| "learning_rate": 4.0641247833622183e-07, | |
| "loss": 2.1336, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.6853620955315871, | |
| "grad_norm": 0.48838868737220764, | |
| "learning_rate": 4.059792027729636e-07, | |
| "loss": 2.128, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.6878274268104777, | |
| "grad_norm": 0.5089288949966431, | |
| "learning_rate": 4.0554592720970537e-07, | |
| "loss": 2.1061, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.6902927580893683, | |
| "grad_norm": 0.49236613512039185, | |
| "learning_rate": 4.0511265164644717e-07, | |
| "loss": 2.0789, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.6927580893682589, | |
| "grad_norm": 0.4705197513103485, | |
| "learning_rate": 4.046793760831889e-07, | |
| "loss": 2.0775, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.6952234206471495, | |
| "grad_norm": 0.48192402720451355, | |
| "learning_rate": 4.0424610051993066e-07, | |
| "loss": 2.0509, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.6976887519260401, | |
| "grad_norm": 0.4875544607639313, | |
| "learning_rate": 4.038128249566724e-07, | |
| "loss": 2.0831, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.7001540832049307, | |
| "grad_norm": 0.4907492995262146, | |
| "learning_rate": 4.033795493934142e-07, | |
| "loss": 2.1026, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.7026194144838213, | |
| "grad_norm": 0.49712225794792175, | |
| "learning_rate": 4.0294627383015595e-07, | |
| "loss": 2.0875, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.7050847457627119, | |
| "grad_norm": 0.4676704406738281, | |
| "learning_rate": 4.0251299826689774e-07, | |
| "loss": 2.0936, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.7075500770416024, | |
| "grad_norm": 0.4734593629837036, | |
| "learning_rate": 4.020797227036395e-07, | |
| "loss": 2.0837, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.710015408320493, | |
| "grad_norm": 0.465678334236145, | |
| "learning_rate": 4.016464471403813e-07, | |
| "loss": 1.9965, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.7124807395993836, | |
| "grad_norm": 0.5036219954490662, | |
| "learning_rate": 4.0121317157712303e-07, | |
| "loss": 2.0894, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.7149460708782742, | |
| "grad_norm": 0.49854403734207153, | |
| "learning_rate": 4.0077989601386483e-07, | |
| "loss": 2.094, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.7174114021571649, | |
| "grad_norm": 0.5132552981376648, | |
| "learning_rate": 4.003466204506066e-07, | |
| "loss": 2.0846, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.7198767334360555, | |
| "grad_norm": 0.48033636808395386, | |
| "learning_rate": 3.9991334488734837e-07, | |
| "loss": 2.0818, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.7223420647149461, | |
| "grad_norm": 0.5240620374679565, | |
| "learning_rate": 3.994800693240901e-07, | |
| "loss": 2.1014, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.7248073959938367, | |
| "grad_norm": 0.5264602303504944, | |
| "learning_rate": 3.990467937608319e-07, | |
| "loss": 2.063, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 0.49397385120391846, | |
| "learning_rate": 3.9861351819757366e-07, | |
| "loss": 2.1056, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.7297380585516179, | |
| "grad_norm": 0.4983115792274475, | |
| "learning_rate": 3.9818024263431546e-07, | |
| "loss": 2.0637, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.7322033898305085, | |
| "grad_norm": 0.5232862234115601, | |
| "learning_rate": 3.9774696707105715e-07, | |
| "loss": 2.1352, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.7346687211093991, | |
| "grad_norm": 0.47278425097465515, | |
| "learning_rate": 3.973136915077989e-07, | |
| "loss": 2.1005, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.7371340523882897, | |
| "grad_norm": 0.5086476802825928, | |
| "learning_rate": 3.968804159445407e-07, | |
| "loss": 2.0621, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.7395993836671803, | |
| "grad_norm": 0.4940975308418274, | |
| "learning_rate": 3.9644714038128244e-07, | |
| "loss": 2.0924, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7420647149460708, | |
| "grad_norm": 0.49190765619277954, | |
| "learning_rate": 3.9601386481802424e-07, | |
| "loss": 2.0956, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.7445300462249614, | |
| "grad_norm": 0.5044582486152649, | |
| "learning_rate": 3.95580589254766e-07, | |
| "loss": 2.0571, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.746995377503852, | |
| "grad_norm": 0.4926859438419342, | |
| "learning_rate": 3.951473136915078e-07, | |
| "loss": 2.079, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.7494607087827427, | |
| "grad_norm": 0.5044658780097961, | |
| "learning_rate": 3.947140381282495e-07, | |
| "loss": 2.1056, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.7519260400616333, | |
| "grad_norm": 0.4952108561992645, | |
| "learning_rate": 3.942807625649913e-07, | |
| "loss": 2.0946, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.7543913713405239, | |
| "grad_norm": 0.5132980346679688, | |
| "learning_rate": 3.9384748700173307e-07, | |
| "loss": 2.0693, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.7568567026194145, | |
| "grad_norm": 0.5124617218971252, | |
| "learning_rate": 3.9341421143847487e-07, | |
| "loss": 2.1272, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.7593220338983051, | |
| "grad_norm": 0.4977681338787079, | |
| "learning_rate": 3.929809358752166e-07, | |
| "loss": 2.0998, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.7617873651771957, | |
| "grad_norm": 0.4836325943470001, | |
| "learning_rate": 3.925476603119584e-07, | |
| "loss": 2.1336, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.7642526964560863, | |
| "grad_norm": 0.5036113262176514, | |
| "learning_rate": 3.9211438474870015e-07, | |
| "loss": 2.1222, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.7667180277349769, | |
| "grad_norm": 0.5216833353042603, | |
| "learning_rate": 3.9168110918544195e-07, | |
| "loss": 2.1161, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.7691833590138675, | |
| "grad_norm": 0.48103034496307373, | |
| "learning_rate": 3.9124783362218364e-07, | |
| "loss": 2.1047, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.7716486902927581, | |
| "grad_norm": 0.4988188147544861, | |
| "learning_rate": 3.9081455805892544e-07, | |
| "loss": 2.1449, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.7741140215716487, | |
| "grad_norm": 0.524972140789032, | |
| "learning_rate": 3.903812824956672e-07, | |
| "loss": 2.0457, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.7765793528505393, | |
| "grad_norm": 0.472748339176178, | |
| "learning_rate": 3.89948006932409e-07, | |
| "loss": 2.0993, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7790446841294298, | |
| "grad_norm": 0.5132001638412476, | |
| "learning_rate": 3.8951473136915073e-07, | |
| "loss": 2.1254, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.7815100154083205, | |
| "grad_norm": 0.4903256893157959, | |
| "learning_rate": 3.8908145580589253e-07, | |
| "loss": 2.1643, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.7839753466872111, | |
| "grad_norm": 0.5069398283958435, | |
| "learning_rate": 3.8864818024263427e-07, | |
| "loss": 2.115, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.7864406779661017, | |
| "grad_norm": 0.5095073580741882, | |
| "learning_rate": 3.8821490467937607e-07, | |
| "loss": 2.1492, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.7889060092449923, | |
| "grad_norm": 0.5011394619941711, | |
| "learning_rate": 3.877816291161178e-07, | |
| "loss": 2.0831, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.7913713405238829, | |
| "grad_norm": 0.502396821975708, | |
| "learning_rate": 3.873483535528596e-07, | |
| "loss": 2.098, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.7938366718027735, | |
| "grad_norm": 0.52918541431427, | |
| "learning_rate": 3.8691507798960136e-07, | |
| "loss": 2.1014, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.7963020030816641, | |
| "grad_norm": 0.48291951417922974, | |
| "learning_rate": 3.8648180242634316e-07, | |
| "loss": 2.0585, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.7987673343605547, | |
| "grad_norm": 0.5417861342430115, | |
| "learning_rate": 3.860485268630849e-07, | |
| "loss": 2.083, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.8012326656394453, | |
| "grad_norm": 0.51700758934021, | |
| "learning_rate": 3.856152512998267e-07, | |
| "loss": 2.1483, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.8036979969183359, | |
| "grad_norm": 0.536354124546051, | |
| "learning_rate": 3.8518197573656845e-07, | |
| "loss": 2.0772, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.8061633281972265, | |
| "grad_norm": 0.48870107531547546, | |
| "learning_rate": 3.847487001733102e-07, | |
| "loss": 2.1208, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.8086286594761171, | |
| "grad_norm": 0.5053774118423462, | |
| "learning_rate": 3.8431542461005194e-07, | |
| "loss": 2.0871, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.8110939907550077, | |
| "grad_norm": 0.5325949788093567, | |
| "learning_rate": 3.8388214904679373e-07, | |
| "loss": 2.0905, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.8135593220338984, | |
| "grad_norm": 0.5083361268043518, | |
| "learning_rate": 3.834488734835355e-07, | |
| "loss": 2.1987, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.816024653312789, | |
| "grad_norm": 0.5428608655929565, | |
| "learning_rate": 3.830155979202773e-07, | |
| "loss": 2.2072, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.8184899845916795, | |
| "grad_norm": 0.5175209641456604, | |
| "learning_rate": 3.82582322357019e-07, | |
| "loss": 2.0833, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.8209553158705701, | |
| "grad_norm": 0.5032325387001038, | |
| "learning_rate": 3.821490467937608e-07, | |
| "loss": 2.1651, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.8234206471494607, | |
| "grad_norm": 0.5360395312309265, | |
| "learning_rate": 3.8171577123050256e-07, | |
| "loss": 2.082, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.8258859784283513, | |
| "grad_norm": 0.505496084690094, | |
| "learning_rate": 3.8128249566724436e-07, | |
| "loss": 2.1208, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.8283513097072419, | |
| "grad_norm": 0.524784505367279, | |
| "learning_rate": 3.808492201039861e-07, | |
| "loss": 2.0893, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.8308166409861325, | |
| "grad_norm": 0.5291479825973511, | |
| "learning_rate": 3.804159445407279e-07, | |
| "loss": 2.1181, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.8332819722650231, | |
| "grad_norm": 0.5064086318016052, | |
| "learning_rate": 3.7998266897746965e-07, | |
| "loss": 2.1009, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.8357473035439137, | |
| "grad_norm": 0.49233660101890564, | |
| "learning_rate": 3.7954939341421145e-07, | |
| "loss": 2.1292, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.8382126348228043, | |
| "grad_norm": 0.5401001572608948, | |
| "learning_rate": 3.791161178509532e-07, | |
| "loss": 2.0972, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.8406779661016949, | |
| "grad_norm": 0.48516029119491577, | |
| "learning_rate": 3.78682842287695e-07, | |
| "loss": 2.0895, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.8431432973805855, | |
| "grad_norm": 0.5405505299568176, | |
| "learning_rate": 3.782495667244367e-07, | |
| "loss": 2.1756, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.8456086286594762, | |
| "grad_norm": 0.5255026817321777, | |
| "learning_rate": 3.778162911611785e-07, | |
| "loss": 2.1275, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.8480739599383668, | |
| "grad_norm": 0.520824670791626, | |
| "learning_rate": 3.7738301559792023e-07, | |
| "loss": 2.1276, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.8505392912172574, | |
| "grad_norm": 0.5046127438545227, | |
| "learning_rate": 3.76949740034662e-07, | |
| "loss": 2.0993, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.853004622496148, | |
| "grad_norm": 0.5003817677497864, | |
| "learning_rate": 3.7651646447140377e-07, | |
| "loss": 2.0911, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.8554699537750385, | |
| "grad_norm": 0.5272133350372314, | |
| "learning_rate": 3.7608318890814557e-07, | |
| "loss": 2.1541, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.8579352850539291, | |
| "grad_norm": 0.5194313526153564, | |
| "learning_rate": 3.756499133448873e-07, | |
| "loss": 2.0722, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.8604006163328197, | |
| "grad_norm": 0.5442456603050232, | |
| "learning_rate": 3.752166377816291e-07, | |
| "loss": 2.133, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.8628659476117103, | |
| "grad_norm": 0.4979715645313263, | |
| "learning_rate": 3.7478336221837086e-07, | |
| "loss": 2.0577, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.8653312788906009, | |
| "grad_norm": 0.5156003832817078, | |
| "learning_rate": 3.7435008665511265e-07, | |
| "loss": 2.1104, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.8677966101694915, | |
| "grad_norm": 0.5033104419708252, | |
| "learning_rate": 3.739168110918544e-07, | |
| "loss": 2.0654, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.8702619414483821, | |
| "grad_norm": 0.48328933119773865, | |
| "learning_rate": 3.734835355285962e-07, | |
| "loss": 2.0754, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.8727272727272727, | |
| "grad_norm": 0.5282177925109863, | |
| "learning_rate": 3.7305025996533794e-07, | |
| "loss": 2.0852, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.8751926040061633, | |
| "grad_norm": 0.4942963719367981, | |
| "learning_rate": 3.7261698440207974e-07, | |
| "loss": 2.0664, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.877657935285054, | |
| "grad_norm": 0.5125203728675842, | |
| "learning_rate": 3.721837088388215e-07, | |
| "loss": 2.1084, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.8801232665639446, | |
| "grad_norm": 0.5272080898284912, | |
| "learning_rate": 3.717504332755633e-07, | |
| "loss": 2.093, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.8825885978428352, | |
| "grad_norm": 0.529028058052063, | |
| "learning_rate": 3.71317157712305e-07, | |
| "loss": 2.0856, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.8850539291217258, | |
| "grad_norm": 0.5381813049316406, | |
| "learning_rate": 3.7088388214904677e-07, | |
| "loss": 2.1001, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.8875192604006163, | |
| "grad_norm": 0.5188086628913879, | |
| "learning_rate": 3.704506065857885e-07, | |
| "loss": 2.0755, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8899845916795069, | |
| "grad_norm": 0.5289164781570435, | |
| "learning_rate": 3.700173310225303e-07, | |
| "loss": 2.1329, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.8924499229583975, | |
| "grad_norm": 0.5325300097465515, | |
| "learning_rate": 3.6958405545927206e-07, | |
| "loss": 2.0798, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.8949152542372881, | |
| "grad_norm": 0.4899941384792328, | |
| "learning_rate": 3.6915077989601386e-07, | |
| "loss": 2.0779, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.8973805855161787, | |
| "grad_norm": 0.5234933495521545, | |
| "learning_rate": 3.687175043327556e-07, | |
| "loss": 2.127, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.8998459167950693, | |
| "grad_norm": 0.5290928483009338, | |
| "learning_rate": 3.682842287694974e-07, | |
| "loss": 2.0466, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.9023112480739599, | |
| "grad_norm": 0.5081377029418945, | |
| "learning_rate": 3.6785095320623915e-07, | |
| "loss": 2.0474, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.9047765793528505, | |
| "grad_norm": 0.5121719241142273, | |
| "learning_rate": 3.6741767764298094e-07, | |
| "loss": 2.0462, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.9072419106317411, | |
| "grad_norm": 0.5094989538192749, | |
| "learning_rate": 3.669844020797227e-07, | |
| "loss": 2.0313, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.9097072419106318, | |
| "grad_norm": 0.5027374029159546, | |
| "learning_rate": 3.665511265164645e-07, | |
| "loss": 2.1838, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.9121725731895224, | |
| "grad_norm": 0.500373363494873, | |
| "learning_rate": 3.6611785095320623e-07, | |
| "loss": 2.0903, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.914637904468413, | |
| "grad_norm": 0.4968167543411255, | |
| "learning_rate": 3.6568457538994803e-07, | |
| "loss": 2.1124, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.9171032357473036, | |
| "grad_norm": 0.5415815114974976, | |
| "learning_rate": 3.652512998266898e-07, | |
| "loss": 2.1302, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.9195685670261942, | |
| "grad_norm": 0.5093196630477905, | |
| "learning_rate": 3.648180242634315e-07, | |
| "loss": 2.0565, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.9220338983050848, | |
| "grad_norm": 0.5081534385681152, | |
| "learning_rate": 3.6438474870017327e-07, | |
| "loss": 2.0346, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.9244992295839753, | |
| "grad_norm": 0.49750521779060364, | |
| "learning_rate": 3.6395147313691506e-07, | |
| "loss": 2.0603, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.9269645608628659, | |
| "grad_norm": 0.5009690523147583, | |
| "learning_rate": 3.635181975736568e-07, | |
| "loss": 2.0378, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.9294298921417565, | |
| "grad_norm": 0.4784706234931946, | |
| "learning_rate": 3.630849220103986e-07, | |
| "loss": 2.057, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.9318952234206471, | |
| "grad_norm": 0.5433520078659058, | |
| "learning_rate": 3.6265164644714035e-07, | |
| "loss": 2.0579, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.9343605546995377, | |
| "grad_norm": 0.5368413329124451, | |
| "learning_rate": 3.6221837088388215e-07, | |
| "loss": 2.1095, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.9368258859784283, | |
| "grad_norm": 0.516017496585846, | |
| "learning_rate": 3.617850953206239e-07, | |
| "loss": 2.0946, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.9392912172573189, | |
| "grad_norm": 0.491853803396225, | |
| "learning_rate": 3.613518197573657e-07, | |
| "loss": 2.0712, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.9417565485362096, | |
| "grad_norm": 0.5004855990409851, | |
| "learning_rate": 3.6091854419410744e-07, | |
| "loss": 2.0958, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.9442218798151002, | |
| "grad_norm": 0.509719967842102, | |
| "learning_rate": 3.6048526863084924e-07, | |
| "loss": 2.0751, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.9466872110939908, | |
| "grad_norm": 0.5413833856582642, | |
| "learning_rate": 3.60051993067591e-07, | |
| "loss": 2.1404, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.9491525423728814, | |
| "grad_norm": 0.5230715870857239, | |
| "learning_rate": 3.596187175043328e-07, | |
| "loss": 2.0349, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.951617873651772, | |
| "grad_norm": 0.5308024883270264, | |
| "learning_rate": 3.591854419410745e-07, | |
| "loss": 2.1079, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.9540832049306626, | |
| "grad_norm": 0.5121743679046631, | |
| "learning_rate": 3.587521663778163e-07, | |
| "loss": 2.0772, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.9565485362095532, | |
| "grad_norm": 0.49954754114151, | |
| "learning_rate": 3.58318890814558e-07, | |
| "loss": 2.1119, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.9590138674884437, | |
| "grad_norm": 0.5327329039573669, | |
| "learning_rate": 3.578856152512998e-07, | |
| "loss": 2.1093, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.9614791987673343, | |
| "grad_norm": 0.5051793456077576, | |
| "learning_rate": 3.5745233968804156e-07, | |
| "loss": 2.0992, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.9639445300462249, | |
| "grad_norm": 0.5336674451828003, | |
| "learning_rate": 3.5701906412478335e-07, | |
| "loss": 2.0602, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.9664098613251155, | |
| "grad_norm": 0.5329372882843018, | |
| "learning_rate": 3.565857885615251e-07, | |
| "loss": 2.1052, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.9688751926040061, | |
| "grad_norm": 0.5276802182197571, | |
| "learning_rate": 3.561525129982669e-07, | |
| "loss": 2.0498, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.9713405238828967, | |
| "grad_norm": 0.49508848786354065, | |
| "learning_rate": 3.5571923743500864e-07, | |
| "loss": 2.0771, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.9738058551617874, | |
| "grad_norm": 0.5108320713043213, | |
| "learning_rate": 3.5528596187175044e-07, | |
| "loss": 2.0973, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.976271186440678, | |
| "grad_norm": 0.5265666246414185, | |
| "learning_rate": 3.548526863084922e-07, | |
| "loss": 2.1155, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.9787365177195686, | |
| "grad_norm": 0.5228252410888672, | |
| "learning_rate": 3.54419410745234e-07, | |
| "loss": 2.0856, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.9812018489984592, | |
| "grad_norm": 0.5155977010726929, | |
| "learning_rate": 3.5398613518197573e-07, | |
| "loss": 2.1268, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.9836671802773498, | |
| "grad_norm": 0.5661581754684448, | |
| "learning_rate": 3.5355285961871753e-07, | |
| "loss": 2.1504, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.9861325115562404, | |
| "grad_norm": 0.5168757438659668, | |
| "learning_rate": 3.5311958405545927e-07, | |
| "loss": 2.0874, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.988597842835131, | |
| "grad_norm": 0.5343320965766907, | |
| "learning_rate": 3.5268630849220107e-07, | |
| "loss": 2.0831, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.9910631741140216, | |
| "grad_norm": 0.5256153345108032, | |
| "learning_rate": 3.522530329289428e-07, | |
| "loss": 2.0848, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.9935285053929122, | |
| "grad_norm": 0.531794548034668, | |
| "learning_rate": 3.518197573656845e-07, | |
| "loss": 2.0557, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.9959938366718027, | |
| "grad_norm": 0.5358244180679321, | |
| "learning_rate": 3.513864818024263e-07, | |
| "loss": 2.0252, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.9984591679506933, | |
| "grad_norm": 0.5200023651123047, | |
| "learning_rate": 3.5095320623916805e-07, | |
| "loss": 2.1135, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9984591679506933, | |
| "eval_loss": 2.093395709991455, | |
| "eval_runtime": 398.8526, | |
| "eval_samples_per_second": 1.234, | |
| "eval_steps_per_second": 0.308, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.000924499229584, | |
| "grad_norm": 0.5109534859657288, | |
| "learning_rate": 3.5051993067590985e-07, | |
| "loss": 2.0975, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.0033898305084745, | |
| "grad_norm": 0.5240735411643982, | |
| "learning_rate": 3.500866551126516e-07, | |
| "loss": 2.1655, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.0058551617873652, | |
| "grad_norm": 0.5256450772285461, | |
| "learning_rate": 3.496533795493934e-07, | |
| "loss": 2.1388, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.0083204930662557, | |
| "grad_norm": 0.5329455733299255, | |
| "learning_rate": 3.4922010398613514e-07, | |
| "loss": 2.089, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.0107858243451464, | |
| "grad_norm": 0.5201774835586548, | |
| "learning_rate": 3.4878682842287693e-07, | |
| "loss": 2.0812, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.013251155624037, | |
| "grad_norm": 0.482138454914093, | |
| "learning_rate": 3.483535528596187e-07, | |
| "loss": 2.1011, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.0157164869029276, | |
| "grad_norm": 0.5111163258552551, | |
| "learning_rate": 3.479202772963605e-07, | |
| "loss": 2.1351, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.018181818181818, | |
| "grad_norm": 0.5070258975028992, | |
| "learning_rate": 3.474870017331022e-07, | |
| "loss": 2.0726, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.0206471494607088, | |
| "grad_norm": 0.5213172435760498, | |
| "learning_rate": 3.47053726169844e-07, | |
| "loss": 2.0709, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.0231124807395995, | |
| "grad_norm": 0.5116674304008484, | |
| "learning_rate": 3.4662045060658576e-07, | |
| "loss": 2.1098, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.02557781201849, | |
| "grad_norm": 0.5342626571655273, | |
| "learning_rate": 3.4618717504332756e-07, | |
| "loss": 2.0572, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.0280431432973807, | |
| "grad_norm": 0.5366671085357666, | |
| "learning_rate": 3.457538994800693e-07, | |
| "loss": 2.1119, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.0305084745762711, | |
| "grad_norm": 0.5127258896827698, | |
| "learning_rate": 3.4532062391681105e-07, | |
| "loss": 2.1222, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.0329738058551619, | |
| "grad_norm": 0.5240333080291748, | |
| "learning_rate": 3.448873483535528e-07, | |
| "loss": 2.0789, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.0354391371340523, | |
| "grad_norm": 0.529655933380127, | |
| "learning_rate": 3.444540727902946e-07, | |
| "loss": 2.1118, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.037904468412943, | |
| "grad_norm": 0.5909092426300049, | |
| "learning_rate": 3.4402079722703634e-07, | |
| "loss": 2.0552, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.0403697996918335, | |
| "grad_norm": 0.5189386010169983, | |
| "learning_rate": 3.4358752166377814e-07, | |
| "loss": 2.0904, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.0428351309707242, | |
| "grad_norm": 0.5365529656410217, | |
| "learning_rate": 3.431542461005199e-07, | |
| "loss": 2.0861, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.0453004622496147, | |
| "grad_norm": 0.5217022895812988, | |
| "learning_rate": 3.427209705372617e-07, | |
| "loss": 2.0698, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.0477657935285054, | |
| "grad_norm": 0.5230218172073364, | |
| "learning_rate": 3.4228769497400343e-07, | |
| "loss": 2.0571, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.0502311248073959, | |
| "grad_norm": 0.5243125557899475, | |
| "learning_rate": 3.418544194107452e-07, | |
| "loss": 2.1333, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.0526964560862866, | |
| "grad_norm": 0.5268332958221436, | |
| "learning_rate": 3.4142114384748697e-07, | |
| "loss": 2.0533, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.0551617873651773, | |
| "grad_norm": 0.5369420051574707, | |
| "learning_rate": 3.4098786828422877e-07, | |
| "loss": 2.0523, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.0576271186440678, | |
| "grad_norm": 0.5415328145027161, | |
| "learning_rate": 3.405545927209705e-07, | |
| "loss": 2.1297, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.0600924499229585, | |
| "grad_norm": 0.5068016648292542, | |
| "learning_rate": 3.401213171577123e-07, | |
| "loss": 2.1055, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.062557781201849, | |
| "grad_norm": 0.5332845449447632, | |
| "learning_rate": 3.3968804159445406e-07, | |
| "loss": 2.0956, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.0650231124807397, | |
| "grad_norm": 0.5464420318603516, | |
| "learning_rate": 3.3925476603119585e-07, | |
| "loss": 2.1394, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.0674884437596301, | |
| "grad_norm": 0.5103365778923035, | |
| "learning_rate": 3.388214904679376e-07, | |
| "loss": 2.0665, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.0699537750385208, | |
| "grad_norm": 0.516782820224762, | |
| "learning_rate": 3.3838821490467934e-07, | |
| "loss": 2.0944, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.0724191063174113, | |
| "grad_norm": 0.518387496471405, | |
| "learning_rate": 3.379549393414211e-07, | |
| "loss": 2.071, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.074884437596302, | |
| "grad_norm": 0.5234736204147339, | |
| "learning_rate": 3.375216637781629e-07, | |
| "loss": 2.1083, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.0773497688751925, | |
| "grad_norm": 0.5194358229637146, | |
| "learning_rate": 3.3708838821490463e-07, | |
| "loss": 2.0615, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.0798151001540832, | |
| "grad_norm": 0.4968649744987488, | |
| "learning_rate": 3.3665511265164643e-07, | |
| "loss": 2.0345, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.0822804314329737, | |
| "grad_norm": 0.5469282865524292, | |
| "learning_rate": 3.362218370883882e-07, | |
| "loss": 2.1193, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.0847457627118644, | |
| "grad_norm": 0.509821891784668, | |
| "learning_rate": 3.3578856152512997e-07, | |
| "loss": 2.092, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.087211093990755, | |
| "grad_norm": 0.5226166248321533, | |
| "learning_rate": 3.353552859618717e-07, | |
| "loss": 2.0753, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.0896764252696456, | |
| "grad_norm": 0.5321833491325378, | |
| "learning_rate": 3.349220103986135e-07, | |
| "loss": 2.0423, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.0921417565485363, | |
| "grad_norm": 0.5013507604598999, | |
| "learning_rate": 3.3448873483535526e-07, | |
| "loss": 2.0561, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.0946070878274268, | |
| "grad_norm": 0.5189869999885559, | |
| "learning_rate": 3.3405545927209706e-07, | |
| "loss": 2.074, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.0970724191063175, | |
| "grad_norm": 0.5598823428153992, | |
| "learning_rate": 3.336221837088388e-07, | |
| "loss": 2.082, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.099537750385208, | |
| "grad_norm": 0.553289532661438, | |
| "learning_rate": 3.331889081455806e-07, | |
| "loss": 2.1035, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.1020030816640987, | |
| "grad_norm": 0.5163353681564331, | |
| "learning_rate": 3.3275563258232235e-07, | |
| "loss": 2.0894, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.1044684129429891, | |
| "grad_norm": 0.5245928168296814, | |
| "learning_rate": 3.3232235701906414e-07, | |
| "loss": 2.0705, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.1069337442218798, | |
| "grad_norm": 0.5174081921577454, | |
| "learning_rate": 3.3188908145580584e-07, | |
| "loss": 2.1132, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.1093990755007703, | |
| "grad_norm": 0.5218523740768433, | |
| "learning_rate": 3.3145580589254763e-07, | |
| "loss": 2.0536, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.111864406779661, | |
| "grad_norm": 0.525867760181427, | |
| "learning_rate": 3.310225303292894e-07, | |
| "loss": 2.1543, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.1143297380585517, | |
| "grad_norm": 0.5667258501052856, | |
| "learning_rate": 3.305892547660312e-07, | |
| "loss": 2.0583, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.1167950693374422, | |
| "grad_norm": 0.541770875453949, | |
| "learning_rate": 3.301559792027729e-07, | |
| "loss": 2.1327, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.119260400616333, | |
| "grad_norm": 0.5466485023498535, | |
| "learning_rate": 3.297227036395147e-07, | |
| "loss": 2.0528, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.1217257318952234, | |
| "grad_norm": 0.5483329892158508, | |
| "learning_rate": 3.2928942807625647e-07, | |
| "loss": 2.1536, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.124191063174114, | |
| "grad_norm": 0.5235666632652283, | |
| "learning_rate": 3.2885615251299826e-07, | |
| "loss": 2.0838, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.1266563944530046, | |
| "grad_norm": 0.5424934029579163, | |
| "learning_rate": 3.2842287694974e-07, | |
| "loss": 2.113, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.1291217257318953, | |
| "grad_norm": 0.5532853007316589, | |
| "learning_rate": 3.279896013864818e-07, | |
| "loss": 2.0799, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.1315870570107858, | |
| "grad_norm": 0.5158779621124268, | |
| "learning_rate": 3.2755632582322355e-07, | |
| "loss": 2.0674, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.1340523882896765, | |
| "grad_norm": 0.5670775175094604, | |
| "learning_rate": 3.2712305025996535e-07, | |
| "loss": 2.0898, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.136517719568567, | |
| "grad_norm": 0.5700893402099609, | |
| "learning_rate": 3.266897746967071e-07, | |
| "loss": 2.0833, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.1389830508474577, | |
| "grad_norm": 0.5165070295333862, | |
| "learning_rate": 3.262564991334489e-07, | |
| "loss": 2.1227, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.1414483821263481, | |
| "grad_norm": 0.5242537260055542, | |
| "learning_rate": 3.2582322357019064e-07, | |
| "loss": 2.0674, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.1439137134052388, | |
| "grad_norm": 0.5270352363586426, | |
| "learning_rate": 3.253899480069324e-07, | |
| "loss": 2.0815, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.1463790446841293, | |
| "grad_norm": 0.5376419425010681, | |
| "learning_rate": 3.2495667244367413e-07, | |
| "loss": 2.0699, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.14884437596302, | |
| "grad_norm": 0.5110929012298584, | |
| "learning_rate": 3.245233968804159e-07, | |
| "loss": 2.0393, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.1513097072419107, | |
| "grad_norm": 0.5514217019081116, | |
| "learning_rate": 3.2409012131715767e-07, | |
| "loss": 2.0484, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.1537750385208012, | |
| "grad_norm": 0.5122043490409851, | |
| "learning_rate": 3.2365684575389947e-07, | |
| "loss": 2.0707, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.156240369799692, | |
| "grad_norm": 0.5329337120056152, | |
| "learning_rate": 3.232235701906412e-07, | |
| "loss": 2.1176, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.1587057010785824, | |
| "grad_norm": 0.546440064907074, | |
| "learning_rate": 3.22790294627383e-07, | |
| "loss": 2.0814, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.161171032357473, | |
| "grad_norm": 0.5156552791595459, | |
| "learning_rate": 3.2235701906412476e-07, | |
| "loss": 2.0764, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.1636363636363636, | |
| "grad_norm": 0.5265877842903137, | |
| "learning_rate": 3.2192374350086656e-07, | |
| "loss": 2.0558, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.1661016949152543, | |
| "grad_norm": 0.5469325184822083, | |
| "learning_rate": 3.214904679376083e-07, | |
| "loss": 2.1109, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.1685670261941448, | |
| "grad_norm": 0.5088624358177185, | |
| "learning_rate": 3.210571923743501e-07, | |
| "loss": 2.1014, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.1710323574730355, | |
| "grad_norm": 0.5592899322509766, | |
| "learning_rate": 3.2062391681109184e-07, | |
| "loss": 2.0925, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.173497688751926, | |
| "grad_norm": 0.5526501536369324, | |
| "learning_rate": 3.2019064124783364e-07, | |
| "loss": 2.0594, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.1759630200308167, | |
| "grad_norm": 0.5420464277267456, | |
| "learning_rate": 3.197573656845754e-07, | |
| "loss": 2.1364, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.1784283513097074, | |
| "grad_norm": 0.53159099817276, | |
| "learning_rate": 3.193240901213172e-07, | |
| "loss": 2.0358, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.1808936825885978, | |
| "grad_norm": 0.5268188118934631, | |
| "learning_rate": 3.188908145580589e-07, | |
| "loss": 2.0514, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.1833590138674885, | |
| "grad_norm": 0.5499210953712463, | |
| "learning_rate": 3.184575389948007e-07, | |
| "loss": 2.0831, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.185824345146379, | |
| "grad_norm": 0.5099960565567017, | |
| "learning_rate": 3.180242634315424e-07, | |
| "loss": 2.0724, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.1882896764252697, | |
| "grad_norm": 0.5238215327262878, | |
| "learning_rate": 3.175909878682842e-07, | |
| "loss": 2.0665, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.1907550077041602, | |
| "grad_norm": 0.5423519611358643, | |
| "learning_rate": 3.1715771230502596e-07, | |
| "loss": 2.1059, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.193220338983051, | |
| "grad_norm": 0.5435311198234558, | |
| "learning_rate": 3.1672443674176776e-07, | |
| "loss": 2.1405, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.1956856702619414, | |
| "grad_norm": 0.5252048373222351, | |
| "learning_rate": 3.162911611785095e-07, | |
| "loss": 2.0534, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.198151001540832, | |
| "grad_norm": 0.5369661450386047, | |
| "learning_rate": 3.158578856152513e-07, | |
| "loss": 2.1207, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.2006163328197226, | |
| "grad_norm": 0.5454829931259155, | |
| "learning_rate": 3.1542461005199305e-07, | |
| "loss": 2.086, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.2030816640986133, | |
| "grad_norm": 0.5264859199523926, | |
| "learning_rate": 3.1499133448873485e-07, | |
| "loss": 2.0628, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.2055469953775038, | |
| "grad_norm": 0.5218833684921265, | |
| "learning_rate": 3.145580589254766e-07, | |
| "loss": 2.0481, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.2080123266563945, | |
| "grad_norm": 0.552793025970459, | |
| "learning_rate": 3.141247833622184e-07, | |
| "loss": 2.0154, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.210477657935285, | |
| "grad_norm": 0.5695677995681763, | |
| "learning_rate": 3.1369150779896013e-07, | |
| "loss": 2.0963, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.2129429892141756, | |
| "grad_norm": 0.5433192849159241, | |
| "learning_rate": 3.1325823223570193e-07, | |
| "loss": 2.1519, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.2154083204930664, | |
| "grad_norm": 0.5215097069740295, | |
| "learning_rate": 3.128249566724437e-07, | |
| "loss": 2.0963, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.2178736517719568, | |
| "grad_norm": 0.5091007351875305, | |
| "learning_rate": 3.123916811091854e-07, | |
| "loss": 2.1167, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.2203389830508475, | |
| "grad_norm": 0.500039279460907, | |
| "learning_rate": 3.1195840554592717e-07, | |
| "loss": 2.0676, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.222804314329738, | |
| "grad_norm": 0.550313413143158, | |
| "learning_rate": 3.1152512998266897e-07, | |
| "loss": 2.141, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.2252696456086287, | |
| "grad_norm": 0.5415179133415222, | |
| "learning_rate": 3.110918544194107e-07, | |
| "loss": 2.0414, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.2277349768875192, | |
| "grad_norm": 0.5273166298866272, | |
| "learning_rate": 3.106585788561525e-07, | |
| "loss": 2.0847, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.23020030816641, | |
| "grad_norm": 0.5227318406105042, | |
| "learning_rate": 3.1022530329289425e-07, | |
| "loss": 2.074, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.2326656394453004, | |
| "grad_norm": 0.5437159538269043, | |
| "learning_rate": 3.0979202772963605e-07, | |
| "loss": 2.072, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.235130970724191, | |
| "grad_norm": 0.5285058617591858, | |
| "learning_rate": 3.093587521663778e-07, | |
| "loss": 2.0242, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.2375963020030816, | |
| "grad_norm": 0.5310484766960144, | |
| "learning_rate": 3.089254766031196e-07, | |
| "loss": 2.0797, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.2400616332819723, | |
| "grad_norm": 0.5042904019355774, | |
| "learning_rate": 3.0849220103986134e-07, | |
| "loss": 2.0703, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.242526964560863, | |
| "grad_norm": 0.5337966084480286, | |
| "learning_rate": 3.0805892547660314e-07, | |
| "loss": 2.0958, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.2449922958397535, | |
| "grad_norm": 0.5376964211463928, | |
| "learning_rate": 3.076256499133449e-07, | |
| "loss": 2.0587, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.2474576271186442, | |
| "grad_norm": 0.5448833703994751, | |
| "learning_rate": 3.071923743500867e-07, | |
| "loss": 2.0371, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.2499229583975346, | |
| "grad_norm": 0.5545293688774109, | |
| "learning_rate": 3.067590987868284e-07, | |
| "loss": 2.1334, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.2523882896764253, | |
| "grad_norm": 0.5596605539321899, | |
| "learning_rate": 3.063258232235702e-07, | |
| "loss": 2.1101, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.2548536209553158, | |
| "grad_norm": 0.5680539011955261, | |
| "learning_rate": 3.058925476603119e-07, | |
| "loss": 2.1224, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.2573189522342065, | |
| "grad_norm": 0.5574333071708679, | |
| "learning_rate": 3.054592720970537e-07, | |
| "loss": 2.0774, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.259784283513097, | |
| "grad_norm": 0.5248346924781799, | |
| "learning_rate": 3.0502599653379546e-07, | |
| "loss": 2.0657, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.2622496147919877, | |
| "grad_norm": 0.5491170883178711, | |
| "learning_rate": 3.0459272097053726e-07, | |
| "loss": 2.0696, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.2647149460708782, | |
| "grad_norm": 0.5188306570053101, | |
| "learning_rate": 3.04159445407279e-07, | |
| "loss": 2.0528, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.267180277349769, | |
| "grad_norm": 0.5485676527023315, | |
| "learning_rate": 3.037261698440208e-07, | |
| "loss": 2.0722, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.2696456086286596, | |
| "grad_norm": 0.5163741707801819, | |
| "learning_rate": 3.0329289428076254e-07, | |
| "loss": 2.0316, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.27211093990755, | |
| "grad_norm": 0.5422912836074829, | |
| "learning_rate": 3.0285961871750434e-07, | |
| "loss": 2.131, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.2745762711864406, | |
| "grad_norm": 0.5590441823005676, | |
| "learning_rate": 3.024263431542461e-07, | |
| "loss": 2.136, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.2770416024653313, | |
| "grad_norm": 0.5339657068252563, | |
| "learning_rate": 3.019930675909879e-07, | |
| "loss": 2.0866, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.279506933744222, | |
| "grad_norm": 0.5249587893486023, | |
| "learning_rate": 3.0155979202772963e-07, | |
| "loss": 2.0505, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.2819722650231125, | |
| "grad_norm": 0.5352839827537537, | |
| "learning_rate": 3.0112651646447143e-07, | |
| "loss": 2.0719, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.2844375963020032, | |
| "grad_norm": 0.521338939666748, | |
| "learning_rate": 3.0069324090121317e-07, | |
| "loss": 2.1264, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.2869029275808936, | |
| "grad_norm": 0.5433731079101562, | |
| "learning_rate": 3.0025996533795497e-07, | |
| "loss": 2.1272, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.2893682588597843, | |
| "grad_norm": 0.5479230880737305, | |
| "learning_rate": 2.998266897746967e-07, | |
| "loss": 2.0996, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.2918335901386748, | |
| "grad_norm": 0.5389401912689209, | |
| "learning_rate": 2.993934142114385e-07, | |
| "loss": 2.1212, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.2942989214175655, | |
| "grad_norm": 0.5670454502105713, | |
| "learning_rate": 2.989601386481802e-07, | |
| "loss": 2.0662, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.296764252696456, | |
| "grad_norm": 0.5215103030204773, | |
| "learning_rate": 2.9852686308492195e-07, | |
| "loss": 2.0437, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.2992295839753467, | |
| "grad_norm": 0.5562092065811157, | |
| "learning_rate": 2.9809358752166375e-07, | |
| "loss": 2.0364, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.3016949152542372, | |
| "grad_norm": 0.5172645449638367, | |
| "learning_rate": 2.976603119584055e-07, | |
| "loss": 2.0302, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.304160246533128, | |
| "grad_norm": 0.5652945041656494, | |
| "learning_rate": 2.972270363951473e-07, | |
| "loss": 2.1963, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.3066255778120186, | |
| "grad_norm": 0.5477254986763, | |
| "learning_rate": 2.9679376083188904e-07, | |
| "loss": 2.0815, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.309090909090909, | |
| "grad_norm": 0.5283427834510803, | |
| "learning_rate": 2.9636048526863084e-07, | |
| "loss": 2.0962, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.3115562403697996, | |
| "grad_norm": 0.5537866950035095, | |
| "learning_rate": 2.959272097053726e-07, | |
| "loss": 2.0885, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.3140215716486903, | |
| "grad_norm": 0.5142833590507507, | |
| "learning_rate": 2.954939341421144e-07, | |
| "loss": 2.0664, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.316486902927581, | |
| "grad_norm": 0.5292583107948303, | |
| "learning_rate": 2.950606585788561e-07, | |
| "loss": 2.0065, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.3189522342064715, | |
| "grad_norm": 0.5204624533653259, | |
| "learning_rate": 2.946273830155979e-07, | |
| "loss": 2.0522, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.3214175654853622, | |
| "grad_norm": 0.534831702709198, | |
| "learning_rate": 2.9419410745233967e-07, | |
| "loss": 2.1378, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.3238828967642526, | |
| "grad_norm": 0.5467716455459595, | |
| "learning_rate": 2.9376083188908146e-07, | |
| "loss": 2.0894, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.3263482280431433, | |
| "grad_norm": 0.536545991897583, | |
| "learning_rate": 2.933275563258232e-07, | |
| "loss": 2.0737, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.3288135593220338, | |
| "grad_norm": 0.5458997488021851, | |
| "learning_rate": 2.92894280762565e-07, | |
| "loss": 2.0601, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.3312788906009245, | |
| "grad_norm": 0.5358441472053528, | |
| "learning_rate": 2.924610051993067e-07, | |
| "loss": 2.0644, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.3337442218798152, | |
| "grad_norm": 0.5691521763801575, | |
| "learning_rate": 2.920277296360485e-07, | |
| "loss": 2.1359, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.3362095531587057, | |
| "grad_norm": 0.5218629837036133, | |
| "learning_rate": 2.9159445407279024e-07, | |
| "loss": 2.058, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.3386748844375962, | |
| "grad_norm": 0.5232541561126709, | |
| "learning_rate": 2.9116117850953204e-07, | |
| "loss": 2.0103, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.341140215716487, | |
| "grad_norm": 0.4909007251262665, | |
| "learning_rate": 2.907279029462738e-07, | |
| "loss": 2.0264, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.3436055469953776, | |
| "grad_norm": 0.5056614875793457, | |
| "learning_rate": 2.902946273830156e-07, | |
| "loss": 2.0628, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.346070878274268, | |
| "grad_norm": 0.5265542268753052, | |
| "learning_rate": 2.8986135181975733e-07, | |
| "loss": 2.098, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.3485362095531588, | |
| "grad_norm": 0.5243731141090393, | |
| "learning_rate": 2.894280762564991e-07, | |
| "loss": 2.0943, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.3510015408320493, | |
| "grad_norm": 0.5294694304466248, | |
| "learning_rate": 2.8899480069324087e-07, | |
| "loss": 2.0928, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.35346687211094, | |
| "grad_norm": 0.5630142688751221, | |
| "learning_rate": 2.8856152512998267e-07, | |
| "loss": 2.0451, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.3559322033898304, | |
| "grad_norm": 0.517575740814209, | |
| "learning_rate": 2.881282495667244e-07, | |
| "loss": 2.0556, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.3583975346687212, | |
| "grad_norm": 0.5253962874412537, | |
| "learning_rate": 2.876949740034662e-07, | |
| "loss": 2.1088, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.3608628659476116, | |
| "grad_norm": 0.5190132260322571, | |
| "learning_rate": 2.8726169844020796e-07, | |
| "loss": 2.0459, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.3633281972265023, | |
| "grad_norm": 0.5542247891426086, | |
| "learning_rate": 2.8682842287694976e-07, | |
| "loss": 2.0736, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.3657935285053928, | |
| "grad_norm": 0.5446171164512634, | |
| "learning_rate": 2.863951473136915e-07, | |
| "loss": 2.0366, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.3682588597842835, | |
| "grad_norm": 0.5620147585868835, | |
| "learning_rate": 2.8596187175043325e-07, | |
| "loss": 2.0593, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.3707241910631742, | |
| "grad_norm": 0.5434351563453674, | |
| "learning_rate": 2.85528596187175e-07, | |
| "loss": 2.0697, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.3731895223420647, | |
| "grad_norm": 0.5291680693626404, | |
| "learning_rate": 2.850953206239168e-07, | |
| "loss": 2.117, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.3756548536209552, | |
| "grad_norm": 0.5382573008537292, | |
| "learning_rate": 2.8466204506065853e-07, | |
| "loss": 2.0755, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.378120184899846, | |
| "grad_norm": 0.5260967016220093, | |
| "learning_rate": 2.8422876949740033e-07, | |
| "loss": 2.0305, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.3805855161787366, | |
| "grad_norm": 0.5255964994430542, | |
| "learning_rate": 2.837954939341421e-07, | |
| "loss": 2.0419, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.383050847457627, | |
| "grad_norm": 0.541347086429596, | |
| "learning_rate": 2.833622183708839e-07, | |
| "loss": 2.0988, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.3855161787365178, | |
| "grad_norm": 0.5098375082015991, | |
| "learning_rate": 2.829289428076256e-07, | |
| "loss": 2.0677, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.3879815100154083, | |
| "grad_norm": 0.5496364235877991, | |
| "learning_rate": 2.824956672443674e-07, | |
| "loss": 2.0748, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.390446841294299, | |
| "grad_norm": 0.5202908515930176, | |
| "learning_rate": 2.8206239168110916e-07, | |
| "loss": 2.1031, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.3929121725731894, | |
| "grad_norm": 0.5658861398696899, | |
| "learning_rate": 2.8162911611785096e-07, | |
| "loss": 2.1063, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.3953775038520801, | |
| "grad_norm": 0.5813626646995544, | |
| "learning_rate": 2.811958405545927e-07, | |
| "loss": 2.0727, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.3978428351309709, | |
| "grad_norm": 0.5407326817512512, | |
| "learning_rate": 2.807625649913345e-07, | |
| "loss": 2.1281, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.4003081664098613, | |
| "grad_norm": 0.5366054177284241, | |
| "learning_rate": 2.8032928942807625e-07, | |
| "loss": 2.0793, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.4027734976887518, | |
| "grad_norm": 0.5454252362251282, | |
| "learning_rate": 2.7989601386481805e-07, | |
| "loss": 2.0573, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.4052388289676425, | |
| "grad_norm": 0.5705124735832214, | |
| "learning_rate": 2.7946273830155974e-07, | |
| "loss": 2.0594, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.4077041602465332, | |
| "grad_norm": 0.5513876676559448, | |
| "learning_rate": 2.7902946273830154e-07, | |
| "loss": 2.0789, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.4101694915254237, | |
| "grad_norm": 0.5202826261520386, | |
| "learning_rate": 2.785961871750433e-07, | |
| "loss": 2.0809, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.4126348228043144, | |
| "grad_norm": 0.5438618659973145, | |
| "learning_rate": 2.781629116117851e-07, | |
| "loss": 2.1247, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.4151001540832049, | |
| "grad_norm": 0.5352728962898254, | |
| "learning_rate": 2.777296360485268e-07, | |
| "loss": 2.0723, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.4175654853620956, | |
| "grad_norm": 0.5342223048210144, | |
| "learning_rate": 2.772963604852686e-07, | |
| "loss": 2.0088, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.420030816640986, | |
| "grad_norm": 0.556289553642273, | |
| "learning_rate": 2.7686308492201037e-07, | |
| "loss": 2.088, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.4224961479198768, | |
| "grad_norm": 0.5240088105201721, | |
| "learning_rate": 2.7642980935875217e-07, | |
| "loss": 2.05, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.4249614791987673, | |
| "grad_norm": 0.5449855327606201, | |
| "learning_rate": 2.759965337954939e-07, | |
| "loss": 2.0436, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.427426810477658, | |
| "grad_norm": 0.5392140746116638, | |
| "learning_rate": 2.755632582322357e-07, | |
| "loss": 2.0478, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.4298921417565484, | |
| "grad_norm": 0.5651275515556335, | |
| "learning_rate": 2.7512998266897745e-07, | |
| "loss": 2.0421, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.4323574730354391, | |
| "grad_norm": 0.5078974366188049, | |
| "learning_rate": 2.7469670710571925e-07, | |
| "loss": 2.0267, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.4348228043143298, | |
| "grad_norm": 0.5486698150634766, | |
| "learning_rate": 2.74263431542461e-07, | |
| "loss": 2.0553, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.4372881355932203, | |
| "grad_norm": 0.5565054416656494, | |
| "learning_rate": 2.738301559792028e-07, | |
| "loss": 2.1163, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.4397534668721108, | |
| "grad_norm": 0.5276550650596619, | |
| "learning_rate": 2.7339688041594454e-07, | |
| "loss": 2.0427, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.4422187981510015, | |
| "grad_norm": 0.5744592547416687, | |
| "learning_rate": 2.729636048526863e-07, | |
| "loss": 2.1292, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.4446841294298922, | |
| "grad_norm": 0.5316488146781921, | |
| "learning_rate": 2.7253032928942803e-07, | |
| "loss": 2.0663, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.4471494607087827, | |
| "grad_norm": 0.562899649143219, | |
| "learning_rate": 2.7209705372616983e-07, | |
| "loss": 2.1515, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.4496147919876734, | |
| "grad_norm": 0.5735856294631958, | |
| "learning_rate": 2.7166377816291157e-07, | |
| "loss": 2.0883, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.4520801232665639, | |
| "grad_norm": 0.5496561527252197, | |
| "learning_rate": 2.7123050259965337e-07, | |
| "loss": 2.069, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.4545454545454546, | |
| "grad_norm": 0.5434293150901794, | |
| "learning_rate": 2.707972270363951e-07, | |
| "loss": 2.0468, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.457010785824345, | |
| "grad_norm": 0.5591951012611389, | |
| "learning_rate": 2.703639514731369e-07, | |
| "loss": 2.0979, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.4594761171032358, | |
| "grad_norm": 0.556535542011261, | |
| "learning_rate": 2.6993067590987866e-07, | |
| "loss": 2.104, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.4619414483821265, | |
| "grad_norm": 0.5265719890594482, | |
| "learning_rate": 2.6949740034662046e-07, | |
| "loss": 2.0565, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.464406779661017, | |
| "grad_norm": 0.5520715117454529, | |
| "learning_rate": 2.690641247833622e-07, | |
| "loss": 2.1049, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.4668721109399074, | |
| "grad_norm": 0.5084496736526489, | |
| "learning_rate": 2.68630849220104e-07, | |
| "loss": 2.0027, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.4693374422187981, | |
| "grad_norm": 0.5238492488861084, | |
| "learning_rate": 2.6819757365684574e-07, | |
| "loss": 2.019, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.4718027734976888, | |
| "grad_norm": 0.5607398748397827, | |
| "learning_rate": 2.6776429809358754e-07, | |
| "loss": 2.097, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.4742681047765793, | |
| "grad_norm": 0.5568619966506958, | |
| "learning_rate": 2.673310225303293e-07, | |
| "loss": 2.0772, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.47673343605547, | |
| "grad_norm": 0.553952157497406, | |
| "learning_rate": 2.668977469670711e-07, | |
| "loss": 2.0929, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.4791987673343605, | |
| "grad_norm": 0.5402290225028992, | |
| "learning_rate": 2.6646447140381283e-07, | |
| "loss": 2.062, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.4816640986132512, | |
| "grad_norm": 0.5399553179740906, | |
| "learning_rate": 2.660311958405546e-07, | |
| "loss": 2.0673, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.4841294298921417, | |
| "grad_norm": 0.5668056011199951, | |
| "learning_rate": 2.655979202772963e-07, | |
| "loss": 2.0596, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.4865947611710324, | |
| "grad_norm": 0.5426621437072754, | |
| "learning_rate": 2.651646447140381e-07, | |
| "loss": 2.0718, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.4890600924499229, | |
| "grad_norm": 0.5283502340316772, | |
| "learning_rate": 2.6473136915077986e-07, | |
| "loss": 2.0544, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.4915254237288136, | |
| "grad_norm": 0.5753063559532166, | |
| "learning_rate": 2.6429809358752166e-07, | |
| "loss": 2.1082, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.493990755007704, | |
| "grad_norm": 0.5240095257759094, | |
| "learning_rate": 2.638648180242634e-07, | |
| "loss": 2.054, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.4964560862865948, | |
| "grad_norm": 0.5336340069770813, | |
| "learning_rate": 2.634315424610052e-07, | |
| "loss": 2.0585, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.4989214175654855, | |
| "grad_norm": 0.5449371337890625, | |
| "learning_rate": 2.6299826689774695e-07, | |
| "loss": 2.0645, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.501386748844376, | |
| "grad_norm": 0.553382396697998, | |
| "learning_rate": 2.6256499133448875e-07, | |
| "loss": 2.0974, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.5038520801232664, | |
| "grad_norm": 0.5372682809829712, | |
| "learning_rate": 2.621317157712305e-07, | |
| "loss": 2.1287, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.5063174114021571, | |
| "grad_norm": 0.5454490780830383, | |
| "learning_rate": 2.616984402079723e-07, | |
| "loss": 2.0699, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.5087827426810478, | |
| "grad_norm": 0.5598608255386353, | |
| "learning_rate": 2.6126516464471404e-07, | |
| "loss": 2.1242, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.5112480739599383, | |
| "grad_norm": 0.5377222895622253, | |
| "learning_rate": 2.6083188908145583e-07, | |
| "loss": 2.078, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.5137134052388288, | |
| "grad_norm": 0.549659013748169, | |
| "learning_rate": 2.603986135181976e-07, | |
| "loss": 2.077, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.5161787365177197, | |
| "grad_norm": 0.5235285758972168, | |
| "learning_rate": 2.599653379549394e-07, | |
| "loss": 2.0384, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.5186440677966102, | |
| "grad_norm": 0.5353394746780396, | |
| "learning_rate": 2.5953206239168107e-07, | |
| "loss": 2.084, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.5211093990755007, | |
| "grad_norm": 0.5501331686973572, | |
| "learning_rate": 2.5909878682842287e-07, | |
| "loss": 2.0653, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.5235747303543914, | |
| "grad_norm": 0.5552558898925781, | |
| "learning_rate": 2.586655112651646e-07, | |
| "loss": 2.0984, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.526040061633282, | |
| "grad_norm": 0.5290137529373169, | |
| "learning_rate": 2.582322357019064e-07, | |
| "loss": 2.101, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.5285053929121726, | |
| "grad_norm": 0.5513224005699158, | |
| "learning_rate": 2.5779896013864815e-07, | |
| "loss": 2.0511, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.530970724191063, | |
| "grad_norm": 0.525702714920044, | |
| "learning_rate": 2.5736568457538995e-07, | |
| "loss": 2.1566, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.5334360554699538, | |
| "grad_norm": 0.5237429738044739, | |
| "learning_rate": 2.569324090121317e-07, | |
| "loss": 2.0632, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.5359013867488445, | |
| "grad_norm": 0.5647119879722595, | |
| "learning_rate": 2.564991334488735e-07, | |
| "loss": 2.0885, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.538366718027735, | |
| "grad_norm": 0.5435053706169128, | |
| "learning_rate": 2.5606585788561524e-07, | |
| "loss": 2.063, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.5408320493066254, | |
| "grad_norm": 0.5810684561729431, | |
| "learning_rate": 2.5563258232235704e-07, | |
| "loss": 2.1319, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.5432973805855161, | |
| "grad_norm": 0.5526551008224487, | |
| "learning_rate": 2.551993067590988e-07, | |
| "loss": 2.1514, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.5457627118644068, | |
| "grad_norm": 0.5331485271453857, | |
| "learning_rate": 2.547660311958406e-07, | |
| "loss": 1.982, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.5482280431432973, | |
| "grad_norm": 0.5447163581848145, | |
| "learning_rate": 2.5433275563258233e-07, | |
| "loss": 2.0662, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.550693374422188, | |
| "grad_norm": 0.5696806311607361, | |
| "learning_rate": 2.538994800693241e-07, | |
| "loss": 2.0573, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.5531587057010787, | |
| "grad_norm": 0.5400473475456238, | |
| "learning_rate": 2.5346620450606587e-07, | |
| "loss": 2.0772, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.5556240369799692, | |
| "grad_norm": 0.543637752532959, | |
| "learning_rate": 2.530329289428076e-07, | |
| "loss": 2.0869, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.5580893682588597, | |
| "grad_norm": 0.5363517999649048, | |
| "learning_rate": 2.5259965337954936e-07, | |
| "loss": 2.106, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.5605546995377504, | |
| "grad_norm": 0.5055550932884216, | |
| "learning_rate": 2.5216637781629116e-07, | |
| "loss": 2.0274, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.563020030816641, | |
| "grad_norm": 0.5467242002487183, | |
| "learning_rate": 2.517331022530329e-07, | |
| "loss": 2.0927, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.5654853620955316, | |
| "grad_norm": 0.5361837148666382, | |
| "learning_rate": 2.512998266897747e-07, | |
| "loss": 2.1079, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.567950693374422, | |
| "grad_norm": 0.5386531949043274, | |
| "learning_rate": 2.5086655112651645e-07, | |
| "loss": 2.0499, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.5704160246533128, | |
| "grad_norm": 0.5682772994041443, | |
| "learning_rate": 2.5043327556325824e-07, | |
| "loss": 2.1246, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.5728813559322035, | |
| "grad_norm": 0.5458372831344604, | |
| "learning_rate": 2.5e-07, | |
| "loss": 2.1237, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.575346687211094, | |
| "grad_norm": 0.5346107482910156, | |
| "learning_rate": 2.4956672443674173e-07, | |
| "loss": 2.0898, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.5778120184899846, | |
| "grad_norm": 0.5748855471611023, | |
| "learning_rate": 2.4913344887348353e-07, | |
| "loss": 2.1254, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.5802773497688754, | |
| "grad_norm": 0.5122947692871094, | |
| "learning_rate": 2.487001733102253e-07, | |
| "loss": 2.057, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.5827426810477658, | |
| "grad_norm": 0.5241718888282776, | |
| "learning_rate": 2.48266897746967e-07, | |
| "loss": 2.0911, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.5852080123266563, | |
| "grad_norm": 0.5585358142852783, | |
| "learning_rate": 2.478336221837088e-07, | |
| "loss": 2.0615, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.587673343605547, | |
| "grad_norm": 0.6165497899055481, | |
| "learning_rate": 2.4740034662045056e-07, | |
| "loss": 2.179, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.5901386748844377, | |
| "grad_norm": 0.5242385864257812, | |
| "learning_rate": 2.4696707105719236e-07, | |
| "loss": 2.0643, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.5926040061633282, | |
| "grad_norm": 0.5357149243354797, | |
| "learning_rate": 2.465337954939341e-07, | |
| "loss": 2.0407, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.5950693374422187, | |
| "grad_norm": 0.525974452495575, | |
| "learning_rate": 2.461005199306759e-07, | |
| "loss": 2.0345, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.5975346687211094, | |
| "grad_norm": 0.5614073872566223, | |
| "learning_rate": 2.4566724436741765e-07, | |
| "loss": 2.0699, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 0.526637613773346, | |
| "learning_rate": 2.4523396880415945e-07, | |
| "loss": 2.0783, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.6024653312788906, | |
| "grad_norm": 0.5309383869171143, | |
| "learning_rate": 2.448006932409012e-07, | |
| "loss": 2.0257, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.604930662557781, | |
| "grad_norm": 0.5437653660774231, | |
| "learning_rate": 2.4436741767764294e-07, | |
| "loss": 2.073, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.6073959938366718, | |
| "grad_norm": 0.5697052478790283, | |
| "learning_rate": 2.4393414211438474e-07, | |
| "loss": 2.0774, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.6098613251155625, | |
| "grad_norm": 0.5350382328033447, | |
| "learning_rate": 2.435008665511265e-07, | |
| "loss": 2.0817, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.612326656394453, | |
| "grad_norm": 0.5554140210151672, | |
| "learning_rate": 2.430675909878683e-07, | |
| "loss": 2.0842, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.6147919876733436, | |
| "grad_norm": 0.5539381504058838, | |
| "learning_rate": 2.4263431542461e-07, | |
| "loss": 2.0684, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.6172573189522343, | |
| "grad_norm": 0.5573647618293762, | |
| "learning_rate": 2.422010398613518e-07, | |
| "loss": 2.0848, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.6197226502311248, | |
| "grad_norm": 0.5368602871894836, | |
| "learning_rate": 2.4176776429809357e-07, | |
| "loss": 2.0463, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.6221879815100153, | |
| "grad_norm": 0.5693660974502563, | |
| "learning_rate": 2.413344887348353e-07, | |
| "loss": 2.1469, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.624653312788906, | |
| "grad_norm": 0.5438942909240723, | |
| "learning_rate": 2.409012131715771e-07, | |
| "loss": 2.0725, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.6271186440677967, | |
| "grad_norm": 0.5400587916374207, | |
| "learning_rate": 2.4046793760831886e-07, | |
| "loss": 2.1452, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.6295839753466872, | |
| "grad_norm": 0.5040330290794373, | |
| "learning_rate": 2.4003466204506065e-07, | |
| "loss": 2.0522, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.6320493066255777, | |
| "grad_norm": 0.5188309550285339, | |
| "learning_rate": 2.396013864818024e-07, | |
| "loss": 2.0538, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.6345146379044684, | |
| "grad_norm": 0.5574179887771606, | |
| "learning_rate": 2.391681109185442e-07, | |
| "loss": 2.1391, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.636979969183359, | |
| "grad_norm": 0.5268549919128418, | |
| "learning_rate": 2.3873483535528594e-07, | |
| "loss": 2.0742, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.6394453004622496, | |
| "grad_norm": 0.5314316749572754, | |
| "learning_rate": 2.3830155979202771e-07, | |
| "loss": 2.0721, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.6419106317411403, | |
| "grad_norm": 0.5841025710105896, | |
| "learning_rate": 2.3786828422876948e-07, | |
| "loss": 2.124, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.644375963020031, | |
| "grad_norm": 0.5572120547294617, | |
| "learning_rate": 2.3743500866551126e-07, | |
| "loss": 2.0926, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.6468412942989215, | |
| "grad_norm": 0.5612335801124573, | |
| "learning_rate": 2.3700173310225303e-07, | |
| "loss": 2.07, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.649306625577812, | |
| "grad_norm": 0.545322835445404, | |
| "learning_rate": 2.365684575389948e-07, | |
| "loss": 2.0675, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.6517719568567026, | |
| "grad_norm": 0.5538606643676758, | |
| "learning_rate": 2.3613518197573657e-07, | |
| "loss": 2.0667, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.6542372881355933, | |
| "grad_norm": 0.5166656374931335, | |
| "learning_rate": 2.3570190641247834e-07, | |
| "loss": 2.0332, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.6567026194144838, | |
| "grad_norm": 0.5593602061271667, | |
| "learning_rate": 2.3526863084922011e-07, | |
| "loss": 2.1419, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.6591679506933743, | |
| "grad_norm": 0.5563687086105347, | |
| "learning_rate": 2.3483535528596186e-07, | |
| "loss": 2.0222, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.661633281972265, | |
| "grad_norm": 0.5505043864250183, | |
| "learning_rate": 2.3440207972270363e-07, | |
| "loss": 2.0949, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.6640986132511557, | |
| "grad_norm": 0.49905431270599365, | |
| "learning_rate": 2.339688041594454e-07, | |
| "loss": 2.0702, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.6665639445300462, | |
| "grad_norm": 0.5658527612686157, | |
| "learning_rate": 2.3353552859618717e-07, | |
| "loss": 2.0925, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.6690292758089367, | |
| "grad_norm": 0.5240568518638611, | |
| "learning_rate": 2.3310225303292894e-07, | |
| "loss": 2.0719, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.6714946070878274, | |
| "grad_norm": 0.5115277767181396, | |
| "learning_rate": 2.3266897746967072e-07, | |
| "loss": 2.0976, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.673959938366718, | |
| "grad_norm": 0.5566647052764893, | |
| "learning_rate": 2.322357019064125e-07, | |
| "loss": 2.0161, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.6764252696456086, | |
| "grad_norm": 0.5448424816131592, | |
| "learning_rate": 2.3180242634315423e-07, | |
| "loss": 2.1174, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.6788906009244993, | |
| "grad_norm": 0.5539455413818359, | |
| "learning_rate": 2.31369150779896e-07, | |
| "loss": 2.0338, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.68135593220339, | |
| "grad_norm": 0.5492231845855713, | |
| "learning_rate": 2.3093587521663778e-07, | |
| "loss": 2.1217, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.6838212634822805, | |
| "grad_norm": 0.5186160206794739, | |
| "learning_rate": 2.3050259965337955e-07, | |
| "loss": 2.0538, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.686286594761171, | |
| "grad_norm": 0.5424460768699646, | |
| "learning_rate": 2.3006932409012132e-07, | |
| "loss": 2.0764, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.6887519260400616, | |
| "grad_norm": 0.5673890709877014, | |
| "learning_rate": 2.296360485268631e-07, | |
| "loss": 2.1029, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.6912172573189523, | |
| "grad_norm": 0.5408136248588562, | |
| "learning_rate": 2.2920277296360486e-07, | |
| "loss": 2.1321, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.6936825885978428, | |
| "grad_norm": 0.5323325991630554, | |
| "learning_rate": 2.2876949740034663e-07, | |
| "loss": 2.1003, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.6961479198767333, | |
| "grad_norm": 0.5330997109413147, | |
| "learning_rate": 2.2833622183708838e-07, | |
| "loss": 2.0747, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.698613251155624, | |
| "grad_norm": 0.5504052042961121, | |
| "learning_rate": 2.2790294627383015e-07, | |
| "loss": 2.1067, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.7010785824345147, | |
| "grad_norm": 0.5399577617645264, | |
| "learning_rate": 2.2746967071057192e-07, | |
| "loss": 2.0032, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.7035439137134052, | |
| "grad_norm": 0.5435107350349426, | |
| "learning_rate": 2.270363951473137e-07, | |
| "loss": 2.0389, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.706009244992296, | |
| "grad_norm": 0.5398039221763611, | |
| "learning_rate": 2.2660311958405546e-07, | |
| "loss": 2.0745, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.7084745762711866, | |
| "grad_norm": 0.5262869000434875, | |
| "learning_rate": 2.2616984402079724e-07, | |
| "loss": 2.0557, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.710939907550077, | |
| "grad_norm": 0.5872994661331177, | |
| "learning_rate": 2.25736568457539e-07, | |
| "loss": 2.1679, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.7134052388289676, | |
| "grad_norm": 0.526508092880249, | |
| "learning_rate": 2.2530329289428073e-07, | |
| "loss": 2.088, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.7158705701078583, | |
| "grad_norm": 0.5324400663375854, | |
| "learning_rate": 2.248700173310225e-07, | |
| "loss": 2.0343, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.718335901386749, | |
| "grad_norm": 0.5571790933609009, | |
| "learning_rate": 2.2443674176776427e-07, | |
| "loss": 2.1619, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.7208012326656394, | |
| "grad_norm": 0.5518919825553894, | |
| "learning_rate": 2.2400346620450604e-07, | |
| "loss": 2.1053, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.72326656394453, | |
| "grad_norm": 0.5263517498970032, | |
| "learning_rate": 2.235701906412478e-07, | |
| "loss": 2.1048, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.7257318952234206, | |
| "grad_norm": 0.5402522087097168, | |
| "learning_rate": 2.2313691507798958e-07, | |
| "loss": 2.1042, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.7281972265023113, | |
| "grad_norm": 0.527553141117096, | |
| "learning_rate": 2.2270363951473135e-07, | |
| "loss": 2.0133, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.7306625577812018, | |
| "grad_norm": 0.5525938868522644, | |
| "learning_rate": 2.2227036395147313e-07, | |
| "loss": 2.0717, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.7331278890600923, | |
| "grad_norm": 0.49240729212760925, | |
| "learning_rate": 2.2183708838821487e-07, | |
| "loss": 2.0334, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.735593220338983, | |
| "grad_norm": 0.5105845928192139, | |
| "learning_rate": 2.2140381282495664e-07, | |
| "loss": 2.0768, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.7380585516178737, | |
| "grad_norm": 0.5725539326667786, | |
| "learning_rate": 2.2097053726169841e-07, | |
| "loss": 2.0955, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.7405238828967642, | |
| "grad_norm": 0.5486234426498413, | |
| "learning_rate": 2.2053726169844019e-07, | |
| "loss": 2.0214, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.7429892141756549, | |
| "grad_norm": 0.5501362681388855, | |
| "learning_rate": 2.2010398613518196e-07, | |
| "loss": 2.1141, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.7454545454545456, | |
| "grad_norm": 0.543329656124115, | |
| "learning_rate": 2.1967071057192373e-07, | |
| "loss": 2.0873, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.747919876733436, | |
| "grad_norm": 0.5453404188156128, | |
| "learning_rate": 2.192374350086655e-07, | |
| "loss": 2.1165, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.7503852080123266, | |
| "grad_norm": 0.5702621936798096, | |
| "learning_rate": 2.1880415944540727e-07, | |
| "loss": 2.1072, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.7528505392912173, | |
| "grad_norm": 0.5359793305397034, | |
| "learning_rate": 2.1837088388214902e-07, | |
| "loss": 2.0051, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 1.755315870570108, | |
| "grad_norm": 0.5496872663497925, | |
| "learning_rate": 2.179376083188908e-07, | |
| "loss": 2.0871, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.7577812018489984, | |
| "grad_norm": 0.5678308606147766, | |
| "learning_rate": 2.1750433275563256e-07, | |
| "loss": 2.0548, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 1.760246533127889, | |
| "grad_norm": 0.546625554561615, | |
| "learning_rate": 2.1707105719237433e-07, | |
| "loss": 2.1175, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.7627118644067796, | |
| "grad_norm": 0.5344644784927368, | |
| "learning_rate": 2.166377816291161e-07, | |
| "loss": 2.0468, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.7651771956856703, | |
| "grad_norm": 0.5369846820831299, | |
| "learning_rate": 2.1620450606585787e-07, | |
| "loss": 2.0596, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.7676425269645608, | |
| "grad_norm": 0.5267930626869202, | |
| "learning_rate": 2.1577123050259965e-07, | |
| "loss": 2.0495, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.7701078582434515, | |
| "grad_norm": 0.5174152255058289, | |
| "learning_rate": 2.153379549393414e-07, | |
| "loss": 2.0451, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.7725731895223422, | |
| "grad_norm": 0.5533501505851746, | |
| "learning_rate": 2.1490467937608316e-07, | |
| "loss": 2.0884, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 1.7750385208012327, | |
| "grad_norm": 0.5463818907737732, | |
| "learning_rate": 2.1447140381282493e-07, | |
| "loss": 2.1242, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.7775038520801232, | |
| "grad_norm": 0.5650577545166016, | |
| "learning_rate": 2.140381282495667e-07, | |
| "loss": 2.0692, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 1.7799691833590139, | |
| "grad_norm": 0.55766361951828, | |
| "learning_rate": 2.1360485268630848e-07, | |
| "loss": 2.0454, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.7824345146379046, | |
| "grad_norm": 0.5192793607711792, | |
| "learning_rate": 2.1317157712305025e-07, | |
| "loss": 2.0684, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 1.784899845916795, | |
| "grad_norm": 0.5581333041191101, | |
| "learning_rate": 2.1273830155979202e-07, | |
| "loss": 2.0862, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.7873651771956856, | |
| "grad_norm": 0.5046393275260925, | |
| "learning_rate": 2.123050259965338e-07, | |
| "loss": 2.0813, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.7898305084745763, | |
| "grad_norm": 0.537695586681366, | |
| "learning_rate": 2.1187175043327554e-07, | |
| "loss": 2.0809, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.792295839753467, | |
| "grad_norm": 0.5130316615104675, | |
| "learning_rate": 2.114384748700173e-07, | |
| "loss": 2.0937, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 1.7947611710323574, | |
| "grad_norm": 0.535754382610321, | |
| "learning_rate": 2.1100519930675908e-07, | |
| "loss": 2.0515, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.797226502311248, | |
| "grad_norm": 0.5362329483032227, | |
| "learning_rate": 2.1057192374350085e-07, | |
| "loss": 2.0982, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 1.7996918335901386, | |
| "grad_norm": 0.4987218379974365, | |
| "learning_rate": 2.1013864818024262e-07, | |
| "loss": 1.9768, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.8021571648690293, | |
| "grad_norm": 0.5529645085334778, | |
| "learning_rate": 2.097053726169844e-07, | |
| "loss": 2.0765, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 1.8046224961479198, | |
| "grad_norm": 0.5982069373130798, | |
| "learning_rate": 2.0927209705372617e-07, | |
| "loss": 2.1271, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.8070878274268105, | |
| "grad_norm": 0.5242738723754883, | |
| "learning_rate": 2.088388214904679e-07, | |
| "loss": 2.0285, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.8095531587057012, | |
| "grad_norm": 0.522243857383728, | |
| "learning_rate": 2.0840554592720968e-07, | |
| "loss": 2.0701, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.8120184899845917, | |
| "grad_norm": 0.5549618601799011, | |
| "learning_rate": 2.0797227036395145e-07, | |
| "loss": 2.0853, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.8144838212634822, | |
| "grad_norm": 0.5378853678703308, | |
| "learning_rate": 2.0753899480069323e-07, | |
| "loss": 2.129, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.8169491525423729, | |
| "grad_norm": 0.5620313286781311, | |
| "learning_rate": 2.07105719237435e-07, | |
| "loss": 2.0955, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 1.8194144838212636, | |
| "grad_norm": 0.5175241231918335, | |
| "learning_rate": 2.0667244367417677e-07, | |
| "loss": 2.0136, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.821879815100154, | |
| "grad_norm": 0.5567770600318909, | |
| "learning_rate": 2.0623916811091854e-07, | |
| "loss": 2.0785, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 1.8243451463790445, | |
| "grad_norm": 0.5436593294143677, | |
| "learning_rate": 2.058058925476603e-07, | |
| "loss": 2.0805, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.8268104776579352, | |
| "grad_norm": 0.5625580549240112, | |
| "learning_rate": 2.0537261698440206e-07, | |
| "loss": 2.0666, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 1.829275808936826, | |
| "grad_norm": 0.5570465922355652, | |
| "learning_rate": 2.0493934142114383e-07, | |
| "loss": 2.0822, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.8317411402157164, | |
| "grad_norm": 0.5406922698020935, | |
| "learning_rate": 2.045060658578856e-07, | |
| "loss": 2.08, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 1.8342064714946071, | |
| "grad_norm": 0.5050183534622192, | |
| "learning_rate": 2.0407279029462737e-07, | |
| "loss": 2.0316, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.8366718027734978, | |
| "grad_norm": 0.5195122361183167, | |
| "learning_rate": 2.0363951473136914e-07, | |
| "loss": 2.0036, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.8391371340523883, | |
| "grad_norm": 0.5114386677742004, | |
| "learning_rate": 2.0320623916811091e-07, | |
| "loss": 2.0347, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.8416024653312788, | |
| "grad_norm": 0.556669294834137, | |
| "learning_rate": 2.0277296360485269e-07, | |
| "loss": 2.0139, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.8440677966101695, | |
| "grad_norm": 0.5159513354301453, | |
| "learning_rate": 2.0233968804159446e-07, | |
| "loss": 2.0301, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.8465331278890602, | |
| "grad_norm": 0.5236509442329407, | |
| "learning_rate": 2.019064124783362e-07, | |
| "loss": 1.995, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.8489984591679507, | |
| "grad_norm": 0.5232506990432739, | |
| "learning_rate": 2.0147313691507797e-07, | |
| "loss": 1.9718, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.8514637904468412, | |
| "grad_norm": 0.5529718399047852, | |
| "learning_rate": 2.0103986135181974e-07, | |
| "loss": 2.0509, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.8539291217257319, | |
| "grad_norm": 0.5564405918121338, | |
| "learning_rate": 2.0060658578856152e-07, | |
| "loss": 2.1508, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.8563944530046226, | |
| "grad_norm": 0.5209116339683533, | |
| "learning_rate": 2.001733102253033e-07, | |
| "loss": 2.0108, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.858859784283513, | |
| "grad_norm": 0.5505017638206482, | |
| "learning_rate": 1.9974003466204506e-07, | |
| "loss": 2.0589, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 1.8613251155624035, | |
| "grad_norm": 0.5664639472961426, | |
| "learning_rate": 1.9930675909878683e-07, | |
| "loss": 2.1018, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.8637904468412942, | |
| "grad_norm": 0.5276853442192078, | |
| "learning_rate": 1.9887348353552858e-07, | |
| "loss": 2.095, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 1.866255778120185, | |
| "grad_norm": 0.5520254373550415, | |
| "learning_rate": 1.9844020797227035e-07, | |
| "loss": 2.1249, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 1.8687211093990754, | |
| "grad_norm": 0.5317251086235046, | |
| "learning_rate": 1.9800693240901212e-07, | |
| "loss": 2.0812, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 1.8711864406779661, | |
| "grad_norm": 0.5425392985343933, | |
| "learning_rate": 1.975736568457539e-07, | |
| "loss": 2.0785, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 1.8736517719568568, | |
| "grad_norm": 0.5492266416549683, | |
| "learning_rate": 1.9714038128249566e-07, | |
| "loss": 2.0745, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.8761171032357473, | |
| "grad_norm": 0.5728384256362915, | |
| "learning_rate": 1.9670710571923743e-07, | |
| "loss": 2.0847, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 1.8785824345146378, | |
| "grad_norm": 0.5570703744888306, | |
| "learning_rate": 1.962738301559792e-07, | |
| "loss": 2.0432, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 1.8810477657935285, | |
| "grad_norm": 0.5224375128746033, | |
| "learning_rate": 1.9584055459272098e-07, | |
| "loss": 2.0734, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 1.8835130970724192, | |
| "grad_norm": 0.5276843905448914, | |
| "learning_rate": 1.9540727902946272e-07, | |
| "loss": 2.0405, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 1.8859784283513097, | |
| "grad_norm": 0.521706223487854, | |
| "learning_rate": 1.949740034662045e-07, | |
| "loss": 2.0807, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.8884437596302002, | |
| "grad_norm": 0.5321924090385437, | |
| "learning_rate": 1.9454072790294626e-07, | |
| "loss": 1.9975, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 1.8909090909090909, | |
| "grad_norm": 0.5504716038703918, | |
| "learning_rate": 1.9410745233968804e-07, | |
| "loss": 2.1148, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 1.8933744221879816, | |
| "grad_norm": 0.5735941529273987, | |
| "learning_rate": 1.936741767764298e-07, | |
| "loss": 2.0855, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 1.895839753466872, | |
| "grad_norm": 0.5594768524169922, | |
| "learning_rate": 1.9324090121317158e-07, | |
| "loss": 2.0675, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 1.8983050847457628, | |
| "grad_norm": 0.5534265041351318, | |
| "learning_rate": 1.9280762564991335e-07, | |
| "loss": 2.059, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.9007704160246535, | |
| "grad_norm": 0.5681132674217224, | |
| "learning_rate": 1.923743500866551e-07, | |
| "loss": 2.1262, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 1.903235747303544, | |
| "grad_norm": 0.52858567237854, | |
| "learning_rate": 1.9194107452339687e-07, | |
| "loss": 2.1065, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 1.9057010785824344, | |
| "grad_norm": 0.5450841784477234, | |
| "learning_rate": 1.9150779896013864e-07, | |
| "loss": 2.0728, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 1.9081664098613251, | |
| "grad_norm": 0.558519184589386, | |
| "learning_rate": 1.910745233968804e-07, | |
| "loss": 2.0356, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 1.9106317411402158, | |
| "grad_norm": 0.5373818874359131, | |
| "learning_rate": 1.9064124783362218e-07, | |
| "loss": 2.1255, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.9130970724191063, | |
| "grad_norm": 0.5443577170372009, | |
| "learning_rate": 1.9020797227036395e-07, | |
| "loss": 2.0232, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 1.9155624036979968, | |
| "grad_norm": 0.5216531157493591, | |
| "learning_rate": 1.8977469670710572e-07, | |
| "loss": 2.091, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 1.9180277349768875, | |
| "grad_norm": 0.5133727788925171, | |
| "learning_rate": 1.893414211438475e-07, | |
| "loss": 2.0208, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 1.9204930662557782, | |
| "grad_norm": 0.5339122414588928, | |
| "learning_rate": 1.8890814558058924e-07, | |
| "loss": 2.085, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 1.9229583975346687, | |
| "grad_norm": 0.529255211353302, | |
| "learning_rate": 1.88474870017331e-07, | |
| "loss": 2.0794, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.9254237288135592, | |
| "grad_norm": 0.54831463098526, | |
| "learning_rate": 1.8804159445407278e-07, | |
| "loss": 2.1017, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 1.9278890600924499, | |
| "grad_norm": 0.5309552550315857, | |
| "learning_rate": 1.8760831889081456e-07, | |
| "loss": 2.1126, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 1.9303543913713406, | |
| "grad_norm": 0.5368801355361938, | |
| "learning_rate": 1.8717504332755633e-07, | |
| "loss": 2.0153, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 1.932819722650231, | |
| "grad_norm": 0.5303763747215271, | |
| "learning_rate": 1.867417677642981e-07, | |
| "loss": 2.0798, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 1.9352850539291218, | |
| "grad_norm": 0.5320292115211487, | |
| "learning_rate": 1.8630849220103987e-07, | |
| "loss": 2.0749, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.9377503852080125, | |
| "grad_norm": 0.5434033274650574, | |
| "learning_rate": 1.8587521663778164e-07, | |
| "loss": 1.9961, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.940215716486903, | |
| "grad_norm": 0.5612481236457825, | |
| "learning_rate": 1.8544194107452339e-07, | |
| "loss": 2.0702, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 1.9426810477657934, | |
| "grad_norm": 0.5725658535957336, | |
| "learning_rate": 1.8500866551126516e-07, | |
| "loss": 2.106, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 1.9451463790446841, | |
| "grad_norm": 0.5459932088851929, | |
| "learning_rate": 1.8457538994800693e-07, | |
| "loss": 2.1029, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 1.9476117103235748, | |
| "grad_norm": 0.5336212515830994, | |
| "learning_rate": 1.841421143847487e-07, | |
| "loss": 2.1044, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.9500770416024653, | |
| "grad_norm": 0.5421571135520935, | |
| "learning_rate": 1.8370883882149047e-07, | |
| "loss": 2.111, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 1.9525423728813558, | |
| "grad_norm": 0.5272592306137085, | |
| "learning_rate": 1.8327556325823224e-07, | |
| "loss": 2.0597, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 1.9550077041602465, | |
| "grad_norm": 0.5724270939826965, | |
| "learning_rate": 1.8284228769497402e-07, | |
| "loss": 2.1377, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 1.9574730354391372, | |
| "grad_norm": 0.544668972492218, | |
| "learning_rate": 1.8240901213171576e-07, | |
| "loss": 2.0321, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.9599383667180277, | |
| "grad_norm": 0.5739486217498779, | |
| "learning_rate": 1.8197573656845753e-07, | |
| "loss": 2.1109, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.9624036979969184, | |
| "grad_norm": 0.5566874742507935, | |
| "learning_rate": 1.815424610051993e-07, | |
| "loss": 2.0582, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 1.964869029275809, | |
| "grad_norm": 0.5381664633750916, | |
| "learning_rate": 1.8110918544194107e-07, | |
| "loss": 2.0734, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 1.9673343605546996, | |
| "grad_norm": 0.5530892014503479, | |
| "learning_rate": 1.8067590987868285e-07, | |
| "loss": 2.0903, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 1.96979969183359, | |
| "grad_norm": 0.5485751628875732, | |
| "learning_rate": 1.8024263431542462e-07, | |
| "loss": 2.0376, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 1.9722650231124808, | |
| "grad_norm": 0.5459527373313904, | |
| "learning_rate": 1.798093587521664e-07, | |
| "loss": 2.0328, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.9747303543913715, | |
| "grad_norm": 0.5666062235832214, | |
| "learning_rate": 1.7937608318890816e-07, | |
| "loss": 2.0966, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 1.977195685670262, | |
| "grad_norm": 0.5256234407424927, | |
| "learning_rate": 1.789428076256499e-07, | |
| "loss": 2.0458, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 1.9796610169491524, | |
| "grad_norm": 0.567815899848938, | |
| "learning_rate": 1.7850953206239168e-07, | |
| "loss": 2.0793, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 1.9821263482280431, | |
| "grad_norm": 0.5290260910987854, | |
| "learning_rate": 1.7807625649913345e-07, | |
| "loss": 2.0759, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 1.9845916795069338, | |
| "grad_norm": 0.5459809303283691, | |
| "learning_rate": 1.7764298093587522e-07, | |
| "loss": 2.0634, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.9870570107858243, | |
| "grad_norm": 0.5483869910240173, | |
| "learning_rate": 1.77209705372617e-07, | |
| "loss": 2.1077, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 1.9895223420647148, | |
| "grad_norm": 0.5340948700904846, | |
| "learning_rate": 1.7677642980935876e-07, | |
| "loss": 2.0584, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 1.9919876733436055, | |
| "grad_norm": 0.5514506101608276, | |
| "learning_rate": 1.7634315424610053e-07, | |
| "loss": 2.1299, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 1.9944530046224962, | |
| "grad_norm": 0.5255659222602844, | |
| "learning_rate": 1.7590987868284225e-07, | |
| "loss": 2.0081, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 1.9969183359013867, | |
| "grad_norm": 0.5237001776695251, | |
| "learning_rate": 1.7547660311958402e-07, | |
| "loss": 1.9692, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.9993836671802774, | |
| "grad_norm": 0.51847243309021, | |
| "learning_rate": 1.750433275563258e-07, | |
| "loss": 2.0306, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 1.9993836671802774, | |
| "eval_loss": 2.072087287902832, | |
| "eval_runtime": 399.1878, | |
| "eval_samples_per_second": 1.233, | |
| "eval_steps_per_second": 0.308, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 2.001848998459168, | |
| "grad_norm": 0.5584355592727661, | |
| "learning_rate": 1.7461005199306757e-07, | |
| "loss": 2.0642, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 2.0043143297380586, | |
| "grad_norm": 0.5292902588844299, | |
| "learning_rate": 1.7417677642980934e-07, | |
| "loss": 2.0819, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 2.006779661016949, | |
| "grad_norm": 0.5554391145706177, | |
| "learning_rate": 1.737435008665511e-07, | |
| "loss": 2.1057, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 2.00924499229584, | |
| "grad_norm": 0.5533407926559448, | |
| "learning_rate": 1.7331022530329288e-07, | |
| "loss": 2.1317, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.0117103235747305, | |
| "grad_norm": 0.5749047994613647, | |
| "learning_rate": 1.7287694974003465e-07, | |
| "loss": 2.0898, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 2.014175654853621, | |
| "grad_norm": 0.5400955080986023, | |
| "learning_rate": 1.724436741767764e-07, | |
| "loss": 2.0524, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 2.0166409861325114, | |
| "grad_norm": 0.563098669052124, | |
| "learning_rate": 1.7201039861351817e-07, | |
| "loss": 2.1349, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 2.0191063174114023, | |
| "grad_norm": 0.5513168573379517, | |
| "learning_rate": 1.7157712305025994e-07, | |
| "loss": 2.0715, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 2.021571648690293, | |
| "grad_norm": 0.530206561088562, | |
| "learning_rate": 1.7114384748700171e-07, | |
| "loss": 2.0893, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.0240369799691833, | |
| "grad_norm": 0.5274333953857422, | |
| "learning_rate": 1.7071057192374348e-07, | |
| "loss": 2.0455, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 2.026502311248074, | |
| "grad_norm": 0.5695887804031372, | |
| "learning_rate": 1.7027729636048526e-07, | |
| "loss": 2.095, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 2.0289676425269647, | |
| "grad_norm": 0.5287963151931763, | |
| "learning_rate": 1.6984402079722703e-07, | |
| "loss": 2.0476, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 2.031432973805855, | |
| "grad_norm": 0.5647676587104797, | |
| "learning_rate": 1.694107452339688e-07, | |
| "loss": 2.1102, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 2.0338983050847457, | |
| "grad_norm": 0.547614336013794, | |
| "learning_rate": 1.6897746967071054e-07, | |
| "loss": 2.079, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.036363636363636, | |
| "grad_norm": 0.5482441782951355, | |
| "learning_rate": 1.6854419410745232e-07, | |
| "loss": 2.0889, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 2.038828967642527, | |
| "grad_norm": 0.5546557903289795, | |
| "learning_rate": 1.681109185441941e-07, | |
| "loss": 2.0866, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 2.0412942989214176, | |
| "grad_norm": 0.5520861744880676, | |
| "learning_rate": 1.6767764298093586e-07, | |
| "loss": 2.0439, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 2.043759630200308, | |
| "grad_norm": 0.5318979024887085, | |
| "learning_rate": 1.6724436741767763e-07, | |
| "loss": 2.044, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 2.046224961479199, | |
| "grad_norm": 0.5348273515701294, | |
| "learning_rate": 1.668110918544194e-07, | |
| "loss": 2.0408, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.0486902927580894, | |
| "grad_norm": 0.5456458330154419, | |
| "learning_rate": 1.6637781629116117e-07, | |
| "loss": 2.0729, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 2.05115562403698, | |
| "grad_norm": 0.514026939868927, | |
| "learning_rate": 1.6594454072790292e-07, | |
| "loss": 2.0506, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 2.0536209553158704, | |
| "grad_norm": 0.5415788888931274, | |
| "learning_rate": 1.655112651646447e-07, | |
| "loss": 2.0678, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 2.0560862865947613, | |
| "grad_norm": 0.5444661378860474, | |
| "learning_rate": 1.6507798960138646e-07, | |
| "loss": 2.0503, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 2.058551617873652, | |
| "grad_norm": 0.5593656897544861, | |
| "learning_rate": 1.6464471403812823e-07, | |
| "loss": 2.1189, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 2.0610169491525423, | |
| "grad_norm": 0.537968635559082, | |
| "learning_rate": 1.6421143847487e-07, | |
| "loss": 2.0548, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 2.063482280431433, | |
| "grad_norm": 0.5174167156219482, | |
| "learning_rate": 1.6377816291161178e-07, | |
| "loss": 2.0028, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 2.0659476117103237, | |
| "grad_norm": 0.5231641530990601, | |
| "learning_rate": 1.6334488734835355e-07, | |
| "loss": 2.0717, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 2.068412942989214, | |
| "grad_norm": 0.4953687787055969, | |
| "learning_rate": 1.6291161178509532e-07, | |
| "loss": 2.0127, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 2.0708782742681047, | |
| "grad_norm": 0.5350403785705566, | |
| "learning_rate": 1.6247833622183706e-07, | |
| "loss": 2.0704, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.073343605546995, | |
| "grad_norm": 0.5553934574127197, | |
| "learning_rate": 1.6204506065857884e-07, | |
| "loss": 2.0524, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 2.075808936825886, | |
| "grad_norm": 0.5565380454063416, | |
| "learning_rate": 1.616117850953206e-07, | |
| "loss": 2.0233, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 2.0782742681047766, | |
| "grad_norm": 0.536406934261322, | |
| "learning_rate": 1.6117850953206238e-07, | |
| "loss": 2.0322, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 2.080739599383667, | |
| "grad_norm": 0.5308685898780823, | |
| "learning_rate": 1.6074523396880415e-07, | |
| "loss": 2.077, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 2.083204930662558, | |
| "grad_norm": 0.5718575716018677, | |
| "learning_rate": 1.6031195840554592e-07, | |
| "loss": 2.0756, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 2.0856702619414484, | |
| "grad_norm": 0.5534864068031311, | |
| "learning_rate": 1.598786828422877e-07, | |
| "loss": 2.05, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 2.088135593220339, | |
| "grad_norm": 0.5546628832817078, | |
| "learning_rate": 1.5944540727902944e-07, | |
| "loss": 2.1312, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 2.0906009244992294, | |
| "grad_norm": 0.5383484959602356, | |
| "learning_rate": 1.590121317157712e-07, | |
| "loss": 2.1304, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 2.0930662557781203, | |
| "grad_norm": 0.533428430557251, | |
| "learning_rate": 1.5857885615251298e-07, | |
| "loss": 2.0438, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 2.095531587057011, | |
| "grad_norm": 0.5307345390319824, | |
| "learning_rate": 1.5814558058925475e-07, | |
| "loss": 2.0567, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.0979969183359013, | |
| "grad_norm": 0.5388327836990356, | |
| "learning_rate": 1.5771230502599652e-07, | |
| "loss": 2.0334, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 2.1004622496147918, | |
| "grad_norm": 0.5538581609725952, | |
| "learning_rate": 1.572790294627383e-07, | |
| "loss": 2.0543, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 2.1029275808936827, | |
| "grad_norm": 0.5587063431739807, | |
| "learning_rate": 1.5684575389948007e-07, | |
| "loss": 2.0625, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 2.105392912172573, | |
| "grad_norm": 0.5408331155776978, | |
| "learning_rate": 1.5641247833622184e-07, | |
| "loss": 2.0825, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 2.1078582434514637, | |
| "grad_norm": 0.5650213956832886, | |
| "learning_rate": 1.5597920277296358e-07, | |
| "loss": 2.0556, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 2.1103235747303546, | |
| "grad_norm": 0.521017849445343, | |
| "learning_rate": 1.5554592720970536e-07, | |
| "loss": 2.1193, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 2.112788906009245, | |
| "grad_norm": 0.557650625705719, | |
| "learning_rate": 1.5511265164644713e-07, | |
| "loss": 2.0413, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 2.1152542372881356, | |
| "grad_norm": 0.5632362961769104, | |
| "learning_rate": 1.546793760831889e-07, | |
| "loss": 2.0961, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 2.117719568567026, | |
| "grad_norm": 0.5181437730789185, | |
| "learning_rate": 1.5424610051993067e-07, | |
| "loss": 2.0451, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 2.120184899845917, | |
| "grad_norm": 0.533606231212616, | |
| "learning_rate": 1.5381282495667244e-07, | |
| "loss": 2.073, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.1226502311248074, | |
| "grad_norm": 0.5333230495452881, | |
| "learning_rate": 1.533795493934142e-07, | |
| "loss": 2.1046, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 2.125115562403698, | |
| "grad_norm": 0.5309494137763977, | |
| "learning_rate": 1.5294627383015596e-07, | |
| "loss": 2.0936, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 2.1275808936825884, | |
| "grad_norm": 0.5571375489234924, | |
| "learning_rate": 1.5251299826689773e-07, | |
| "loss": 2.0501, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 2.1300462249614793, | |
| "grad_norm": 0.5348679423332214, | |
| "learning_rate": 1.520797227036395e-07, | |
| "loss": 2.1188, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 2.13251155624037, | |
| "grad_norm": 0.5523291230201721, | |
| "learning_rate": 1.5164644714038127e-07, | |
| "loss": 2.0758, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 2.1349768875192603, | |
| "grad_norm": 0.5347540974617004, | |
| "learning_rate": 1.5121317157712304e-07, | |
| "loss": 2.0445, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 2.137442218798151, | |
| "grad_norm": 0.5445932745933533, | |
| "learning_rate": 1.5077989601386482e-07, | |
| "loss": 2.0976, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 2.1399075500770417, | |
| "grad_norm": 0.5866304636001587, | |
| "learning_rate": 1.5034662045060659e-07, | |
| "loss": 2.1779, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 2.142372881355932, | |
| "grad_norm": 0.5338853597640991, | |
| "learning_rate": 1.4991334488734836e-07, | |
| "loss": 2.0782, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 2.1448382126348227, | |
| "grad_norm": 0.5243597626686096, | |
| "learning_rate": 1.494800693240901e-07, | |
| "loss": 2.0461, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.1473035439137136, | |
| "grad_norm": 0.49124395847320557, | |
| "learning_rate": 1.4904679376083187e-07, | |
| "loss": 1.9764, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 2.149768875192604, | |
| "grad_norm": 0.5705875158309937, | |
| "learning_rate": 1.4861351819757365e-07, | |
| "loss": 2.0567, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 2.1522342064714945, | |
| "grad_norm": 0.5213779807090759, | |
| "learning_rate": 1.4818024263431542e-07, | |
| "loss": 2.0596, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 2.154699537750385, | |
| "grad_norm": 0.5367861390113831, | |
| "learning_rate": 1.477469670710572e-07, | |
| "loss": 2.0842, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 2.157164869029276, | |
| "grad_norm": 0.5243573188781738, | |
| "learning_rate": 1.4731369150779896e-07, | |
| "loss": 2.0585, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.1596302003081664, | |
| "grad_norm": 0.4946637451648712, | |
| "learning_rate": 1.4688041594454073e-07, | |
| "loss": 2.0466, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 2.162095531587057, | |
| "grad_norm": 0.521358072757721, | |
| "learning_rate": 1.464471403812825e-07, | |
| "loss": 2.0255, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 2.1645608628659474, | |
| "grad_norm": 0.5509891510009766, | |
| "learning_rate": 1.4601386481802425e-07, | |
| "loss": 2.0372, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 2.1670261941448383, | |
| "grad_norm": 0.5519523024559021, | |
| "learning_rate": 1.4558058925476602e-07, | |
| "loss": 2.0703, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 2.169491525423729, | |
| "grad_norm": 0.5824438333511353, | |
| "learning_rate": 1.451473136915078e-07, | |
| "loss": 2.1207, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.1719568567026193, | |
| "grad_norm": 0.547703742980957, | |
| "learning_rate": 1.4471403812824956e-07, | |
| "loss": 2.0514, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 2.17442218798151, | |
| "grad_norm": 0.5056921243667603, | |
| "learning_rate": 1.4428076256499133e-07, | |
| "loss": 2.0716, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 2.1768875192604007, | |
| "grad_norm": 0.5273824334144592, | |
| "learning_rate": 1.438474870017331e-07, | |
| "loss": 2.0256, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 2.179352850539291, | |
| "grad_norm": 0.509907066822052, | |
| "learning_rate": 1.4341421143847488e-07, | |
| "loss": 2.0043, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 2.1818181818181817, | |
| "grad_norm": 0.5353569388389587, | |
| "learning_rate": 1.4298093587521662e-07, | |
| "loss": 2.0209, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 2.1842835130970726, | |
| "grad_norm": 0.5298518538475037, | |
| "learning_rate": 1.425476603119584e-07, | |
| "loss": 2.0981, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 2.186748844375963, | |
| "grad_norm": 0.5307156443595886, | |
| "learning_rate": 1.4211438474870017e-07, | |
| "loss": 2.0828, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 2.1892141756548535, | |
| "grad_norm": 0.5242281556129456, | |
| "learning_rate": 1.4168110918544194e-07, | |
| "loss": 2.0023, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 2.191679506933744, | |
| "grad_norm": 0.5324615240097046, | |
| "learning_rate": 1.412478336221837e-07, | |
| "loss": 2.0425, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 2.194144838212635, | |
| "grad_norm": 0.5723989009857178, | |
| "learning_rate": 1.4081455805892548e-07, | |
| "loss": 2.1017, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.1966101694915254, | |
| "grad_norm": 0.5204815864562988, | |
| "learning_rate": 1.4038128249566725e-07, | |
| "loss": 2.0347, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 2.199075500770416, | |
| "grad_norm": 0.5336599349975586, | |
| "learning_rate": 1.3994800693240902e-07, | |
| "loss": 2.0568, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 2.2015408320493064, | |
| "grad_norm": 0.5549258589744568, | |
| "learning_rate": 1.3951473136915077e-07, | |
| "loss": 2.1108, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 2.2040061633281973, | |
| "grad_norm": 0.536512017250061, | |
| "learning_rate": 1.3908145580589254e-07, | |
| "loss": 1.9747, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 2.206471494607088, | |
| "grad_norm": 0.575552761554718, | |
| "learning_rate": 1.386481802426343e-07, | |
| "loss": 2.1284, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 2.2089368258859783, | |
| "grad_norm": 0.5624683499336243, | |
| "learning_rate": 1.3821490467937608e-07, | |
| "loss": 2.0404, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 2.211402157164869, | |
| "grad_norm": 0.5736910700798035, | |
| "learning_rate": 1.3778162911611785e-07, | |
| "loss": 2.1057, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 2.2138674884437597, | |
| "grad_norm": 0.5336840152740479, | |
| "learning_rate": 1.3734835355285963e-07, | |
| "loss": 2.0677, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 2.21633281972265, | |
| "grad_norm": 0.538602888584137, | |
| "learning_rate": 1.369150779896014e-07, | |
| "loss": 2.0613, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 2.2187981510015407, | |
| "grad_norm": 0.5597540736198425, | |
| "learning_rate": 1.3648180242634314e-07, | |
| "loss": 2.038, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.2212634822804316, | |
| "grad_norm": 0.5410885810852051, | |
| "learning_rate": 1.3604852686308491e-07, | |
| "loss": 2.0088, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 2.223728813559322, | |
| "grad_norm": 0.5312333106994629, | |
| "learning_rate": 1.3561525129982669e-07, | |
| "loss": 2.1265, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 2.2261941448382125, | |
| "grad_norm": 0.5065499544143677, | |
| "learning_rate": 1.3518197573656846e-07, | |
| "loss": 2.0552, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 2.2286594761171035, | |
| "grad_norm": 0.6002041101455688, | |
| "learning_rate": 1.3474870017331023e-07, | |
| "loss": 2.1298, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.231124807395994, | |
| "grad_norm": 0.5347397327423096, | |
| "learning_rate": 1.34315424610052e-07, | |
| "loss": 2.0224, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.2335901386748844, | |
| "grad_norm": 0.5485448241233826, | |
| "learning_rate": 1.3388214904679377e-07, | |
| "loss": 2.0542, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 2.236055469953775, | |
| "grad_norm": 0.5377957820892334, | |
| "learning_rate": 1.3344887348353554e-07, | |
| "loss": 2.0108, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 2.238520801232666, | |
| "grad_norm": 0.5628427863121033, | |
| "learning_rate": 1.330155979202773e-07, | |
| "loss": 2.0757, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.2409861325115563, | |
| "grad_norm": 0.5469453930854797, | |
| "learning_rate": 1.3258232235701906e-07, | |
| "loss": 2.0858, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 2.243451463790447, | |
| "grad_norm": 0.5658044219017029, | |
| "learning_rate": 1.3214904679376083e-07, | |
| "loss": 2.0153, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.2459167950693373, | |
| "grad_norm": 0.53436279296875, | |
| "learning_rate": 1.317157712305026e-07, | |
| "loss": 2.0632, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 2.248382126348228, | |
| "grad_norm": 0.5214031934738159, | |
| "learning_rate": 1.3128249566724437e-07, | |
| "loss": 2.041, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.2508474576271187, | |
| "grad_norm": 0.562078595161438, | |
| "learning_rate": 1.3084922010398615e-07, | |
| "loss": 2.0901, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 2.253312788906009, | |
| "grad_norm": 0.5474571585655212, | |
| "learning_rate": 1.3041594454072792e-07, | |
| "loss": 2.1005, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.2557781201848996, | |
| "grad_norm": 0.4903823733329773, | |
| "learning_rate": 1.299826689774697e-07, | |
| "loss": 2.0976, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.2582434514637906, | |
| "grad_norm": 0.5153361558914185, | |
| "learning_rate": 1.2954939341421143e-07, | |
| "loss": 2.0719, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.260708782742681, | |
| "grad_norm": 0.5511119365692139, | |
| "learning_rate": 1.291161178509532e-07, | |
| "loss": 2.1, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 2.2631741140215715, | |
| "grad_norm": 0.5670679211616516, | |
| "learning_rate": 1.2868284228769498e-07, | |
| "loss": 2.0562, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.2656394453004625, | |
| "grad_norm": 0.5923585891723633, | |
| "learning_rate": 1.2824956672443675e-07, | |
| "loss": 2.089, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 2.268104776579353, | |
| "grad_norm": 0.5289289951324463, | |
| "learning_rate": 1.2781629116117852e-07, | |
| "loss": 2.0465, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.2705701078582434, | |
| "grad_norm": 0.5217342972755432, | |
| "learning_rate": 1.273830155979203e-07, | |
| "loss": 2.0802, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 2.273035439137134, | |
| "grad_norm": 0.49211281538009644, | |
| "learning_rate": 1.2694974003466206e-07, | |
| "loss": 1.9795, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.275500770416025, | |
| "grad_norm": 0.5618743300437927, | |
| "learning_rate": 1.265164644714038e-07, | |
| "loss": 2.0689, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 2.2779661016949153, | |
| "grad_norm": 0.5565395951271057, | |
| "learning_rate": 1.2608318890814558e-07, | |
| "loss": 2.0646, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.280431432973806, | |
| "grad_norm": 0.5676069855690002, | |
| "learning_rate": 1.2564991334488735e-07, | |
| "loss": 2.1177, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.2828967642526963, | |
| "grad_norm": 0.5341387391090393, | |
| "learning_rate": 1.2521663778162912e-07, | |
| "loss": 2.0936, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 2.285362095531587, | |
| "grad_norm": 0.5444186925888062, | |
| "learning_rate": 1.2478336221837087e-07, | |
| "loss": 2.0803, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 2.2878274268104777, | |
| "grad_norm": 0.5417422652244568, | |
| "learning_rate": 1.2435008665511264e-07, | |
| "loss": 2.0709, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 2.290292758089368, | |
| "grad_norm": 0.5582102537155151, | |
| "learning_rate": 1.239168110918544e-07, | |
| "loss": 2.0996, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 2.2927580893682586, | |
| "grad_norm": 0.5045194625854492, | |
| "learning_rate": 1.2348353552859618e-07, | |
| "loss": 2.0541, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.2952234206471496, | |
| "grad_norm": 0.5552394986152649, | |
| "learning_rate": 1.2305025996533795e-07, | |
| "loss": 2.0941, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 2.29768875192604, | |
| "grad_norm": 0.5538477897644043, | |
| "learning_rate": 1.2261698440207972e-07, | |
| "loss": 2.0667, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 2.3001540832049305, | |
| "grad_norm": 0.5248857140541077, | |
| "learning_rate": 1.2218370883882147e-07, | |
| "loss": 2.0399, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 2.3026194144838215, | |
| "grad_norm": 0.5399145483970642, | |
| "learning_rate": 1.2175043327556324e-07, | |
| "loss": 2.0839, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 2.305084745762712, | |
| "grad_norm": 0.4992510676383972, | |
| "learning_rate": 1.21317157712305e-07, | |
| "loss": 2.066, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.3075500770416024, | |
| "grad_norm": 0.5216765999794006, | |
| "learning_rate": 1.2088388214904678e-07, | |
| "loss": 2.0596, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 2.310015408320493, | |
| "grad_norm": 0.5164439082145691, | |
| "learning_rate": 1.2045060658578856e-07, | |
| "loss": 2.0475, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 2.312480739599384, | |
| "grad_norm": 0.5475905537605286, | |
| "learning_rate": 1.2001733102253033e-07, | |
| "loss": 2.1321, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 2.3149460708782743, | |
| "grad_norm": 0.5179426670074463, | |
| "learning_rate": 1.195840554592721e-07, | |
| "loss": 2.0291, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 2.317411402157165, | |
| "grad_norm": 0.5330396890640259, | |
| "learning_rate": 1.1915077989601386e-07, | |
| "loss": 2.0818, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.3198767334360557, | |
| "grad_norm": 0.5324074029922485, | |
| "learning_rate": 1.1871750433275563e-07, | |
| "loss": 2.0444, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 2.322342064714946, | |
| "grad_norm": 0.5341172814369202, | |
| "learning_rate": 1.182842287694974e-07, | |
| "loss": 2.1091, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 2.3248073959938367, | |
| "grad_norm": 0.5381863713264465, | |
| "learning_rate": 1.1785095320623917e-07, | |
| "loss": 2.0696, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 2.327272727272727, | |
| "grad_norm": 0.5387257933616638, | |
| "learning_rate": 1.1741767764298093e-07, | |
| "loss": 2.0165, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 2.3297380585516176, | |
| "grad_norm": 0.5189254283905029, | |
| "learning_rate": 1.169844020797227e-07, | |
| "loss": 2.007, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.3322033898305086, | |
| "grad_norm": 0.5254266858100891, | |
| "learning_rate": 1.1655112651646447e-07, | |
| "loss": 2.0533, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 2.334668721109399, | |
| "grad_norm": 0.5086634159088135, | |
| "learning_rate": 1.1611785095320624e-07, | |
| "loss": 2.0516, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 2.3371340523882895, | |
| "grad_norm": 0.5261170268058777, | |
| "learning_rate": 1.15684575389948e-07, | |
| "loss": 2.0425, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 2.3395993836671805, | |
| "grad_norm": 0.5267928242683411, | |
| "learning_rate": 1.1525129982668977e-07, | |
| "loss": 2.0764, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 2.342064714946071, | |
| "grad_norm": 0.5495226979255676, | |
| "learning_rate": 1.1481802426343155e-07, | |
| "loss": 2.0627, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.3445300462249614, | |
| "grad_norm": 0.5452602505683899, | |
| "learning_rate": 1.1438474870017332e-07, | |
| "loss": 2.0614, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 2.346995377503852, | |
| "grad_norm": 0.5248115658760071, | |
| "learning_rate": 1.1395147313691507e-07, | |
| "loss": 2.0621, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 2.349460708782743, | |
| "grad_norm": 0.5369260907173157, | |
| "learning_rate": 1.1351819757365685e-07, | |
| "loss": 2.0759, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 2.3519260400616333, | |
| "grad_norm": 0.5348695516586304, | |
| "learning_rate": 1.1308492201039862e-07, | |
| "loss": 2.0857, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 2.354391371340524, | |
| "grad_norm": 0.5731531381607056, | |
| "learning_rate": 1.1265164644714036e-07, | |
| "loss": 2.1256, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.3568567026194147, | |
| "grad_norm": 0.5018579959869385, | |
| "learning_rate": 1.1221837088388213e-07, | |
| "loss": 2.094, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 2.359322033898305, | |
| "grad_norm": 0.5458043217658997, | |
| "learning_rate": 1.117850953206239e-07, | |
| "loss": 2.1194, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 2.3617873651771957, | |
| "grad_norm": 0.5173928141593933, | |
| "learning_rate": 1.1135181975736568e-07, | |
| "loss": 2.0397, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 2.364252696456086, | |
| "grad_norm": 0.5401055216789246, | |
| "learning_rate": 1.1091854419410744e-07, | |
| "loss": 2.0689, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 2.366718027734977, | |
| "grad_norm": 0.5073379278182983, | |
| "learning_rate": 1.1048526863084921e-07, | |
| "loss": 2.0451, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.3691833590138676, | |
| "grad_norm": 0.5283588767051697, | |
| "learning_rate": 1.1005199306759098e-07, | |
| "loss": 2.0587, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 2.371648690292758, | |
| "grad_norm": 0.5442807674407959, | |
| "learning_rate": 1.0961871750433275e-07, | |
| "loss": 2.0932, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 2.3741140215716485, | |
| "grad_norm": 0.5510744452476501, | |
| "learning_rate": 1.0918544194107451e-07, | |
| "loss": 2.0002, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 2.3765793528505395, | |
| "grad_norm": 0.5534915328025818, | |
| "learning_rate": 1.0875216637781628e-07, | |
| "loss": 2.0747, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 2.37904468412943, | |
| "grad_norm": 0.5089408159255981, | |
| "learning_rate": 1.0831889081455805e-07, | |
| "loss": 2.0468, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.3815100154083204, | |
| "grad_norm": 0.5470302104949951, | |
| "learning_rate": 1.0788561525129982e-07, | |
| "loss": 2.0852, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 2.383975346687211, | |
| "grad_norm": 0.537598729133606, | |
| "learning_rate": 1.0745233968804158e-07, | |
| "loss": 2.052, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 2.386440677966102, | |
| "grad_norm": 0.5340083837509155, | |
| "learning_rate": 1.0701906412478335e-07, | |
| "loss": 2.0654, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 2.3889060092449923, | |
| "grad_norm": 0.5371851325035095, | |
| "learning_rate": 1.0658578856152512e-07, | |
| "loss": 2.0654, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 2.391371340523883, | |
| "grad_norm": 0.523451566696167, | |
| "learning_rate": 1.061525129982669e-07, | |
| "loss": 2.0399, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.3938366718027737, | |
| "grad_norm": 0.5346859693527222, | |
| "learning_rate": 1.0571923743500865e-07, | |
| "loss": 2.0665, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 2.396302003081664, | |
| "grad_norm": 0.5562816858291626, | |
| "learning_rate": 1.0528596187175043e-07, | |
| "loss": 2.0161, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.3987673343605547, | |
| "grad_norm": 0.5285778641700745, | |
| "learning_rate": 1.048526863084922e-07, | |
| "loss": 1.9842, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 2.401232665639445, | |
| "grad_norm": 0.5487334728240967, | |
| "learning_rate": 1.0441941074523396e-07, | |
| "loss": 2.1344, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 2.403697996918336, | |
| "grad_norm": 0.5225200057029724, | |
| "learning_rate": 1.0398613518197573e-07, | |
| "loss": 2.0685, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.4061633281972266, | |
| "grad_norm": 0.5379974842071533, | |
| "learning_rate": 1.035528596187175e-07, | |
| "loss": 2.1233, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 2.408628659476117, | |
| "grad_norm": 0.5409395098686218, | |
| "learning_rate": 1.0311958405545927e-07, | |
| "loss": 2.0772, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 2.4110939907550075, | |
| "grad_norm": 0.5314688086509705, | |
| "learning_rate": 1.0268630849220103e-07, | |
| "loss": 2.1255, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 2.4135593220338984, | |
| "grad_norm": 0.5340871214866638, | |
| "learning_rate": 1.022530329289428e-07, | |
| "loss": 2.0418, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 2.416024653312789, | |
| "grad_norm": 0.5267408490180969, | |
| "learning_rate": 1.0181975736568457e-07, | |
| "loss": 2.0437, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.4184899845916794, | |
| "grad_norm": 0.5541632771492004, | |
| "learning_rate": 1.0138648180242634e-07, | |
| "loss": 2.1056, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 2.42095531587057, | |
| "grad_norm": 0.5380145907402039, | |
| "learning_rate": 1.009532062391681e-07, | |
| "loss": 2.0414, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 2.423420647149461, | |
| "grad_norm": 0.5519285202026367, | |
| "learning_rate": 1.0051993067590987e-07, | |
| "loss": 2.1093, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 2.4258859784283513, | |
| "grad_norm": 0.5127108097076416, | |
| "learning_rate": 1.0008665511265164e-07, | |
| "loss": 2.0603, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 2.428351309707242, | |
| "grad_norm": 0.5526323914527893, | |
| "learning_rate": 9.965337954939342e-08, | |
| "loss": 1.9958, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.4308166409861327, | |
| "grad_norm": 0.5529227256774902, | |
| "learning_rate": 9.922010398613517e-08, | |
| "loss": 2.0501, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 2.433281972265023, | |
| "grad_norm": 0.5273423790931702, | |
| "learning_rate": 9.878682842287695e-08, | |
| "loss": 2.0101, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 2.4357473035439137, | |
| "grad_norm": 0.5523142218589783, | |
| "learning_rate": 9.835355285961872e-08, | |
| "loss": 2.0865, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 2.438212634822804, | |
| "grad_norm": 0.5643287897109985, | |
| "learning_rate": 9.792027729636049e-08, | |
| "loss": 2.0684, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 2.440677966101695, | |
| "grad_norm": 0.557341992855072, | |
| "learning_rate": 9.748700173310225e-08, | |
| "loss": 2.081, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.4431432973805856, | |
| "grad_norm": 0.5340744853019714, | |
| "learning_rate": 9.705372616984402e-08, | |
| "loss": 2.055, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 2.445608628659476, | |
| "grad_norm": 0.5190538167953491, | |
| "learning_rate": 9.662045060658579e-08, | |
| "loss": 2.0683, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 2.448073959938367, | |
| "grad_norm": 0.5445495247840881, | |
| "learning_rate": 9.618717504332755e-08, | |
| "loss": 2.0722, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 2.4505392912172574, | |
| "grad_norm": 0.5343917608261108, | |
| "learning_rate": 9.575389948006932e-08, | |
| "loss": 2.0433, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 2.453004622496148, | |
| "grad_norm": 0.5210959911346436, | |
| "learning_rate": 9.532062391681109e-08, | |
| "loss": 1.9952, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.4554699537750384, | |
| "grad_norm": 0.5104400515556335, | |
| "learning_rate": 9.488734835355286e-08, | |
| "loss": 2.0422, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 2.457935285053929, | |
| "grad_norm": 0.497904509305954, | |
| "learning_rate": 9.445407279029462e-08, | |
| "loss": 2.0064, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 2.46040061633282, | |
| "grad_norm": 0.5744480490684509, | |
| "learning_rate": 9.402079722703639e-08, | |
| "loss": 2.0706, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 2.4628659476117103, | |
| "grad_norm": 0.5227028727531433, | |
| "learning_rate": 9.358752166377816e-08, | |
| "loss": 2.0436, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 2.4653312788906008, | |
| "grad_norm": 0.5533729791641235, | |
| "learning_rate": 9.315424610051993e-08, | |
| "loss": 2.099, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.4677966101694917, | |
| "grad_norm": 0.5527722239494324, | |
| "learning_rate": 9.272097053726169e-08, | |
| "loss": 2.1031, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 2.470261941448382, | |
| "grad_norm": 0.5120221376419067, | |
| "learning_rate": 9.228769497400346e-08, | |
| "loss": 1.9816, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 2.4727272727272727, | |
| "grad_norm": 0.543556809425354, | |
| "learning_rate": 9.185441941074524e-08, | |
| "loss": 2.11, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 2.475192604006163, | |
| "grad_norm": 0.5218470096588135, | |
| "learning_rate": 9.142114384748701e-08, | |
| "loss": 2.0315, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 2.477657935285054, | |
| "grad_norm": 0.5415855050086975, | |
| "learning_rate": 9.098786828422877e-08, | |
| "loss": 2.0113, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 2.4801232665639446, | |
| "grad_norm": 0.5189818739891052, | |
| "learning_rate": 9.055459272097054e-08, | |
| "loss": 2.0254, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 2.482588597842835, | |
| "grad_norm": 0.5290147662162781, | |
| "learning_rate": 9.012131715771231e-08, | |
| "loss": 2.0068, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 2.485053929121726, | |
| "grad_norm": 0.5759127736091614, | |
| "learning_rate": 8.968804159445408e-08, | |
| "loss": 2.0765, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 2.4875192604006164, | |
| "grad_norm": 0.5441100001335144, | |
| "learning_rate": 8.925476603119584e-08, | |
| "loss": 2.0505, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 2.489984591679507, | |
| "grad_norm": 0.5378677845001221, | |
| "learning_rate": 8.882149046793761e-08, | |
| "loss": 2.015, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.4924499229583974, | |
| "grad_norm": 0.5395601987838745, | |
| "learning_rate": 8.838821490467938e-08, | |
| "loss": 2.0722, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 2.4949152542372883, | |
| "grad_norm": 0.5622943043708801, | |
| "learning_rate": 8.795493934142113e-08, | |
| "loss": 2.0416, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 2.497380585516179, | |
| "grad_norm": 0.5506992936134338, | |
| "learning_rate": 8.75216637781629e-08, | |
| "loss": 2.0613, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 2.4998459167950693, | |
| "grad_norm": 0.5459305644035339, | |
| "learning_rate": 8.708838821490467e-08, | |
| "loss": 2.1208, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 2.50231124807396, | |
| "grad_norm": 0.5379814505577087, | |
| "learning_rate": 8.665511265164644e-08, | |
| "loss": 2.1047, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 2.5047765793528507, | |
| "grad_norm": 0.5360638499259949, | |
| "learning_rate": 8.62218370883882e-08, | |
| "loss": 2.1164, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 2.507241910631741, | |
| "grad_norm": 0.5440200567245483, | |
| "learning_rate": 8.578856152512997e-08, | |
| "loss": 2.0739, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 2.5097072419106317, | |
| "grad_norm": 0.515691339969635, | |
| "learning_rate": 8.535528596187174e-08, | |
| "loss": 2.0349, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 2.512172573189522, | |
| "grad_norm": 0.5369467735290527, | |
| "learning_rate": 8.492201039861351e-08, | |
| "loss": 2.0522, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 2.514637904468413, | |
| "grad_norm": 0.5383486747741699, | |
| "learning_rate": 8.448873483535527e-08, | |
| "loss": 2.1116, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.5171032357473035, | |
| "grad_norm": 0.5523207187652588, | |
| "learning_rate": 8.405545927209704e-08, | |
| "loss": 2.1088, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 2.519568567026194, | |
| "grad_norm": 0.5444223284721375, | |
| "learning_rate": 8.362218370883882e-08, | |
| "loss": 2.0104, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 2.522033898305085, | |
| "grad_norm": 0.5488723516464233, | |
| "learning_rate": 8.318890814558059e-08, | |
| "loss": 2.1172, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 2.5244992295839754, | |
| "grad_norm": 0.5439826250076294, | |
| "learning_rate": 8.275563258232234e-08, | |
| "loss": 2.0759, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 2.526964560862866, | |
| "grad_norm": 0.5681473016738892, | |
| "learning_rate": 8.232235701906412e-08, | |
| "loss": 2.1139, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.5294298921417564, | |
| "grad_norm": 0.5635061860084534, | |
| "learning_rate": 8.188908145580589e-08, | |
| "loss": 2.1519, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 2.531895223420647, | |
| "grad_norm": 0.5256964564323425, | |
| "learning_rate": 8.145580589254766e-08, | |
| "loss": 1.9879, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 2.534360554699538, | |
| "grad_norm": 0.5846258401870728, | |
| "learning_rate": 8.102253032928942e-08, | |
| "loss": 2.1386, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 2.5368258859784283, | |
| "grad_norm": 0.5462486147880554, | |
| "learning_rate": 8.058925476603119e-08, | |
| "loss": 2.1044, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 2.539291217257319, | |
| "grad_norm": 0.5283377766609192, | |
| "learning_rate": 8.015597920277296e-08, | |
| "loss": 2.012, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.5417565485362097, | |
| "grad_norm": 0.5379184484481812, | |
| "learning_rate": 7.972270363951472e-08, | |
| "loss": 2.0931, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 2.5442218798151, | |
| "grad_norm": 0.5492013692855835, | |
| "learning_rate": 7.928942807625649e-08, | |
| "loss": 2.0455, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 2.5466872110939907, | |
| "grad_norm": 0.5451326966285706, | |
| "learning_rate": 7.885615251299826e-08, | |
| "loss": 2.1058, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 2.549152542372881, | |
| "grad_norm": 0.575563371181488, | |
| "learning_rate": 7.842287694974003e-08, | |
| "loss": 2.1257, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 2.551617873651772, | |
| "grad_norm": 0.5399583578109741, | |
| "learning_rate": 7.798960138648179e-08, | |
| "loss": 2.0936, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 2.5540832049306625, | |
| "grad_norm": 0.5116003751754761, | |
| "learning_rate": 7.755632582322356e-08, | |
| "loss": 2.0322, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 2.556548536209553, | |
| "grad_norm": 0.5290835499763489, | |
| "learning_rate": 7.712305025996533e-08, | |
| "loss": 1.9983, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 2.559013867488444, | |
| "grad_norm": 0.5547816753387451, | |
| "learning_rate": 7.66897746967071e-08, | |
| "loss": 2.056, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 2.5614791987673344, | |
| "grad_norm": 0.5498971939086914, | |
| "learning_rate": 7.625649913344886e-08, | |
| "loss": 2.073, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 2.563944530046225, | |
| "grad_norm": 0.5650001764297485, | |
| "learning_rate": 7.582322357019064e-08, | |
| "loss": 2.043, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.5664098613251154, | |
| "grad_norm": 0.5324546098709106, | |
| "learning_rate": 7.538994800693241e-08, | |
| "loss": 2.0302, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 2.5688751926040063, | |
| "grad_norm": 0.5463674664497375, | |
| "learning_rate": 7.495667244367418e-08, | |
| "loss": 2.1044, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 2.571340523882897, | |
| "grad_norm": 0.5481089949607849, | |
| "learning_rate": 7.452339688041594e-08, | |
| "loss": 2.1399, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 2.5738058551617873, | |
| "grad_norm": 0.5360924005508423, | |
| "learning_rate": 7.409012131715771e-08, | |
| "loss": 2.0823, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 2.576271186440678, | |
| "grad_norm": 0.5051180124282837, | |
| "learning_rate": 7.365684575389948e-08, | |
| "loss": 2.0697, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 2.5787365177195687, | |
| "grad_norm": 0.5565017461776733, | |
| "learning_rate": 7.322357019064125e-08, | |
| "loss": 2.119, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 2.581201848998459, | |
| "grad_norm": 0.5469699501991272, | |
| "learning_rate": 7.279029462738301e-08, | |
| "loss": 2.0709, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 2.5836671802773497, | |
| "grad_norm": 0.5414313673973083, | |
| "learning_rate": 7.235701906412478e-08, | |
| "loss": 2.0261, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 2.58613251155624, | |
| "grad_norm": 0.5786354541778564, | |
| "learning_rate": 7.192374350086655e-08, | |
| "loss": 2.1286, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 2.588597842835131, | |
| "grad_norm": 0.5345943570137024, | |
| "learning_rate": 7.149046793760831e-08, | |
| "loss": 2.0257, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.5910631741140215, | |
| "grad_norm": 0.5214563608169556, | |
| "learning_rate": 7.105719237435008e-08, | |
| "loss": 2.087, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 2.593528505392912, | |
| "grad_norm": 0.5570570230484009, | |
| "learning_rate": 7.062391681109185e-08, | |
| "loss": 2.0851, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 2.595993836671803, | |
| "grad_norm": 0.5345001220703125, | |
| "learning_rate": 7.019064124783363e-08, | |
| "loss": 2.0493, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 2.5984591679506934, | |
| "grad_norm": 0.5382468700408936, | |
| "learning_rate": 6.975736568457538e-08, | |
| "loss": 2.0143, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 2.600924499229584, | |
| "grad_norm": 0.5520262122154236, | |
| "learning_rate": 6.932409012131716e-08, | |
| "loss": 2.0448, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 2.6033898305084744, | |
| "grad_norm": 0.5316290259361267, | |
| "learning_rate": 6.889081455805893e-08, | |
| "loss": 2.0786, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 2.6058551617873653, | |
| "grad_norm": 0.5460601449012756, | |
| "learning_rate": 6.84575389948007e-08, | |
| "loss": 2.0344, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 2.608320493066256, | |
| "grad_norm": 0.5313700437545776, | |
| "learning_rate": 6.802426343154246e-08, | |
| "loss": 2.0203, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 2.6107858243451463, | |
| "grad_norm": 0.5356596112251282, | |
| "learning_rate": 6.759098786828423e-08, | |
| "loss": 2.049, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 2.613251155624037, | |
| "grad_norm": 0.5224314332008362, | |
| "learning_rate": 6.7157712305026e-08, | |
| "loss": 2.1187, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.6157164869029277, | |
| "grad_norm": 0.5021442770957947, | |
| "learning_rate": 6.672443674176777e-08, | |
| "loss": 2.05, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 2.618181818181818, | |
| "grad_norm": 0.548633873462677, | |
| "learning_rate": 6.629116117850953e-08, | |
| "loss": 2.0263, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 2.6206471494607086, | |
| "grad_norm": 0.5166215896606445, | |
| "learning_rate": 6.58578856152513e-08, | |
| "loss": 2.0944, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 2.623112480739599, | |
| "grad_norm": 0.5464893579483032, | |
| "learning_rate": 6.542461005199307e-08, | |
| "loss": 2.0461, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 2.62557781201849, | |
| "grad_norm": 0.5411100387573242, | |
| "learning_rate": 6.499133448873484e-08, | |
| "loss": 2.063, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 2.6280431432973805, | |
| "grad_norm": 0.5743029713630676, | |
| "learning_rate": 6.45580589254766e-08, | |
| "loss": 2.0907, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 2.6305084745762715, | |
| "grad_norm": 0.5295523405075073, | |
| "learning_rate": 6.412478336221837e-08, | |
| "loss": 2.0439, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 2.632973805855162, | |
| "grad_norm": 0.5270727276802063, | |
| "learning_rate": 6.369150779896015e-08, | |
| "loss": 2.1029, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 2.6354391371340524, | |
| "grad_norm": 0.5252944827079773, | |
| "learning_rate": 6.32582322357019e-08, | |
| "loss": 2.128, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 2.637904468412943, | |
| "grad_norm": 0.5068302750587463, | |
| "learning_rate": 6.282495667244368e-08, | |
| "loss": 2.0755, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.6403697996918334, | |
| "grad_norm": 0.5070674419403076, | |
| "learning_rate": 6.239168110918543e-08, | |
| "loss": 2.0293, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 2.6428351309707243, | |
| "grad_norm": 0.5849376320838928, | |
| "learning_rate": 6.19584055459272e-08, | |
| "loss": 1.9944, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 2.645300462249615, | |
| "grad_norm": 0.5567883253097534, | |
| "learning_rate": 6.152512998266898e-08, | |
| "loss": 2.0614, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 2.6477657935285053, | |
| "grad_norm": 0.5402822494506836, | |
| "learning_rate": 6.109185441941073e-08, | |
| "loss": 2.0492, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 2.650231124807396, | |
| "grad_norm": 0.5675923228263855, | |
| "learning_rate": 6.06585788561525e-08, | |
| "loss": 2.0722, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.6526964560862867, | |
| "grad_norm": 0.5432523488998413, | |
| "learning_rate": 6.022530329289428e-08, | |
| "loss": 2.1229, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 2.655161787365177, | |
| "grad_norm": 0.5045982003211975, | |
| "learning_rate": 5.979202772963605e-08, | |
| "loss": 2.0172, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 2.6576271186440676, | |
| "grad_norm": 0.55848628282547, | |
| "learning_rate": 5.9358752166377814e-08, | |
| "loss": 2.0945, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 2.660092449922958, | |
| "grad_norm": 0.5616159439086914, | |
| "learning_rate": 5.8925476603119586e-08, | |
| "loss": 2.0604, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 2.662557781201849, | |
| "grad_norm": 0.5720440149307251, | |
| "learning_rate": 5.849220103986135e-08, | |
| "loss": 2.1264, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.6650231124807395, | |
| "grad_norm": 0.5143904685974121, | |
| "learning_rate": 5.805892547660312e-08, | |
| "loss": 2.0427, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 2.6674884437596305, | |
| "grad_norm": 0.5810486078262329, | |
| "learning_rate": 5.762564991334489e-08, | |
| "loss": 2.07, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 2.669953775038521, | |
| "grad_norm": 0.5719928741455078, | |
| "learning_rate": 5.719237435008666e-08, | |
| "loss": 2.0964, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 2.6724191063174114, | |
| "grad_norm": 0.5077625513076782, | |
| "learning_rate": 5.675909878682842e-08, | |
| "loss": 2.0299, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 2.674884437596302, | |
| "grad_norm": 0.5371079444885254, | |
| "learning_rate": 5.632582322357018e-08, | |
| "loss": 2.0597, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 2.6773497688751924, | |
| "grad_norm": 0.5217646360397339, | |
| "learning_rate": 5.589254766031195e-08, | |
| "loss": 2.0428, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 2.6798151001540833, | |
| "grad_norm": 0.5251481533050537, | |
| "learning_rate": 5.545927209705372e-08, | |
| "loss": 2.0108, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 2.682280431432974, | |
| "grad_norm": 0.5702562928199768, | |
| "learning_rate": 5.502599653379549e-08, | |
| "loss": 2.0502, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 2.6847457627118643, | |
| "grad_norm": 0.5626732110977173, | |
| "learning_rate": 5.4592720970537254e-08, | |
| "loss": 2.117, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 2.687211093990755, | |
| "grad_norm": 0.5354193449020386, | |
| "learning_rate": 5.4159445407279026e-08, | |
| "loss": 2.0782, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.6896764252696457, | |
| "grad_norm": 0.5256460309028625, | |
| "learning_rate": 5.372616984402079e-08, | |
| "loss": 2.0282, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 2.692141756548536, | |
| "grad_norm": 0.5491920709609985, | |
| "learning_rate": 5.329289428076256e-08, | |
| "loss": 2.0504, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 2.6946070878274266, | |
| "grad_norm": 0.5547864437103271, | |
| "learning_rate": 5.285961871750433e-08, | |
| "loss": 2.0578, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 2.6970724191063176, | |
| "grad_norm": 0.5350878834724426, | |
| "learning_rate": 5.24263431542461e-08, | |
| "loss": 2.1029, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 2.699537750385208, | |
| "grad_norm": 0.5535500049591064, | |
| "learning_rate": 5.1993067590987863e-08, | |
| "loss": 2.0707, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 2.7020030816640985, | |
| "grad_norm": 0.5403065085411072, | |
| "learning_rate": 5.1559792027729635e-08, | |
| "loss": 1.998, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 2.7044684129429895, | |
| "grad_norm": 0.5719413161277771, | |
| "learning_rate": 5.11265164644714e-08, | |
| "loss": 2.1054, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 2.70693374422188, | |
| "grad_norm": 0.5244591236114502, | |
| "learning_rate": 5.069324090121317e-08, | |
| "loss": 2.049, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 2.7093990755007704, | |
| "grad_norm": 0.5488945245742798, | |
| "learning_rate": 5.0259965337954936e-08, | |
| "loss": 2.0355, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 2.711864406779661, | |
| "grad_norm": 0.5236762166023254, | |
| "learning_rate": 4.982668977469671e-08, | |
| "loss": 2.0486, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.7143297380585514, | |
| "grad_norm": 0.5210961699485779, | |
| "learning_rate": 4.939341421143847e-08, | |
| "loss": 2.0529, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 2.7167950693374423, | |
| "grad_norm": 0.5268608927726746, | |
| "learning_rate": 4.8960138648180244e-08, | |
| "loss": 2.0309, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 2.719260400616333, | |
| "grad_norm": 0.5346582531929016, | |
| "learning_rate": 4.852686308492201e-08, | |
| "loss": 2.0741, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 2.7217257318952233, | |
| "grad_norm": 0.5198339223861694, | |
| "learning_rate": 4.8093587521663774e-08, | |
| "loss": 2.0415, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 2.724191063174114, | |
| "grad_norm": 0.5542382597923279, | |
| "learning_rate": 4.7660311958405545e-08, | |
| "loss": 1.9782, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 2.7266563944530047, | |
| "grad_norm": 0.5626508593559265, | |
| "learning_rate": 4.722703639514731e-08, | |
| "loss": 2.0709, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 2.729121725731895, | |
| "grad_norm": 0.5496556758880615, | |
| "learning_rate": 4.679376083188908e-08, | |
| "loss": 2.0471, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 2.7315870570107856, | |
| "grad_norm": 0.5292820930480957, | |
| "learning_rate": 4.6360485268630847e-08, | |
| "loss": 2.0754, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 2.7340523882896766, | |
| "grad_norm": 0.5677548050880432, | |
| "learning_rate": 4.592720970537262e-08, | |
| "loss": 2.0872, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 2.736517719568567, | |
| "grad_norm": 0.540000855922699, | |
| "learning_rate": 4.549393414211438e-08, | |
| "loss": 2.0581, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.7389830508474575, | |
| "grad_norm": 0.5193243026733398, | |
| "learning_rate": 4.5060658578856154e-08, | |
| "loss": 2.0456, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 2.7414483821263484, | |
| "grad_norm": 0.46619105339050293, | |
| "learning_rate": 4.462738301559792e-08, | |
| "loss": 2.071, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 2.743913713405239, | |
| "grad_norm": 0.5568689107894897, | |
| "learning_rate": 4.419410745233969e-08, | |
| "loss": 2.0484, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 2.7463790446841294, | |
| "grad_norm": 0.5671171545982361, | |
| "learning_rate": 4.376083188908145e-08, | |
| "loss": 2.118, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 2.74884437596302, | |
| "grad_norm": 0.5143083333969116, | |
| "learning_rate": 4.332755632582322e-08, | |
| "loss": 2.0614, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 2.7513097072419104, | |
| "grad_norm": 0.5241749286651611, | |
| "learning_rate": 4.2894280762564985e-08, | |
| "loss": 2.0289, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 2.7537750385208013, | |
| "grad_norm": 0.5477257370948792, | |
| "learning_rate": 4.246100519930676e-08, | |
| "loss": 2.0791, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 2.756240369799692, | |
| "grad_norm": 0.5546357035636902, | |
| "learning_rate": 4.202772963604852e-08, | |
| "loss": 2.1143, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 2.7587057010785827, | |
| "grad_norm": 0.5520354509353638, | |
| "learning_rate": 4.1594454072790293e-08, | |
| "loss": 2.0603, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 2.761171032357473, | |
| "grad_norm": 0.5563856959342957, | |
| "learning_rate": 4.116117850953206e-08, | |
| "loss": 2.1351, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.7636363636363637, | |
| "grad_norm": 0.525393009185791, | |
| "learning_rate": 4.072790294627383e-08, | |
| "loss": 2.0263, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 2.766101694915254, | |
| "grad_norm": 0.5529180765151978, | |
| "learning_rate": 4.0294627383015595e-08, | |
| "loss": 2.0467, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 2.7685670261941446, | |
| "grad_norm": 0.5137497186660767, | |
| "learning_rate": 3.986135181975736e-08, | |
| "loss": 2.0483, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 2.7710323574730356, | |
| "grad_norm": 0.5601691603660583, | |
| "learning_rate": 3.942807625649913e-08, | |
| "loss": 2.086, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 2.773497688751926, | |
| "grad_norm": 0.5232656598091125, | |
| "learning_rate": 3.8994800693240896e-08, | |
| "loss": 2.0691, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.7759630200308165, | |
| "grad_norm": 0.5236607789993286, | |
| "learning_rate": 3.856152512998267e-08, | |
| "loss": 2.0526, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 2.7784283513097074, | |
| "grad_norm": 0.5435487627983093, | |
| "learning_rate": 3.812824956672443e-08, | |
| "loss": 2.0556, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 2.780893682588598, | |
| "grad_norm": 0.5397012233734131, | |
| "learning_rate": 3.7694974003466204e-08, | |
| "loss": 2.0034, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 2.7833590138674884, | |
| "grad_norm": 0.5322771072387695, | |
| "learning_rate": 3.726169844020797e-08, | |
| "loss": 1.9779, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 2.785824345146379, | |
| "grad_norm": 0.5031158328056335, | |
| "learning_rate": 3.682842287694974e-08, | |
| "loss": 2.0684, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.7882896764252694, | |
| "grad_norm": 0.5290920734405518, | |
| "learning_rate": 3.6395147313691505e-08, | |
| "loss": 2.0269, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 2.7907550077041603, | |
| "grad_norm": 0.5569365620613098, | |
| "learning_rate": 3.5961871750433277e-08, | |
| "loss": 2.0999, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 2.7932203389830508, | |
| "grad_norm": 0.5367685556411743, | |
| "learning_rate": 3.552859618717504e-08, | |
| "loss": 2.1059, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 2.7956856702619417, | |
| "grad_norm": 0.5496487021446228, | |
| "learning_rate": 3.509532062391681e-08, | |
| "loss": 2.0743, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 2.798151001540832, | |
| "grad_norm": 0.5711979269981384, | |
| "learning_rate": 3.466204506065858e-08, | |
| "loss": 2.1088, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 2.8006163328197227, | |
| "grad_norm": 0.5262277126312256, | |
| "learning_rate": 3.422876949740035e-08, | |
| "loss": 2.0074, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 2.803081664098613, | |
| "grad_norm": 0.5461674928665161, | |
| "learning_rate": 3.3795493934142114e-08, | |
| "loss": 2.1001, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 2.8055469953775036, | |
| "grad_norm": 0.531196653842926, | |
| "learning_rate": 3.3362218370883886e-08, | |
| "loss": 2.0391, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 2.8080123266563946, | |
| "grad_norm": 0.5338148474693298, | |
| "learning_rate": 3.292894280762565e-08, | |
| "loss": 2.0516, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 2.810477657935285, | |
| "grad_norm": 0.5303349494934082, | |
| "learning_rate": 3.249566724436742e-08, | |
| "loss": 2.1145, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.8129429892141755, | |
| "grad_norm": 0.5264041423797607, | |
| "learning_rate": 3.206239168110919e-08, | |
| "loss": 2.0457, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 2.8154083204930664, | |
| "grad_norm": 0.5118086338043213, | |
| "learning_rate": 3.162911611785095e-08, | |
| "loss": 2.0861, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 2.817873651771957, | |
| "grad_norm": 0.5181803703308105, | |
| "learning_rate": 3.119584055459272e-08, | |
| "loss": 2.1112, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 2.8203389830508474, | |
| "grad_norm": 0.5603974461555481, | |
| "learning_rate": 3.076256499133449e-08, | |
| "loss": 2.0837, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 2.822804314329738, | |
| "grad_norm": 0.5450738072395325, | |
| "learning_rate": 3.032928942807625e-08, | |
| "loss": 2.0777, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 2.825269645608629, | |
| "grad_norm": 0.5328848361968994, | |
| "learning_rate": 2.9896013864818025e-08, | |
| "loss": 2.0869, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 2.8277349768875193, | |
| "grad_norm": 0.5186157822608948, | |
| "learning_rate": 2.9462738301559793e-08, | |
| "loss": 2.0064, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 2.8302003081664098, | |
| "grad_norm": 0.5262656807899475, | |
| "learning_rate": 2.902946273830156e-08, | |
| "loss": 2.0448, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 2.8326656394453007, | |
| "grad_norm": 0.5398064851760864, | |
| "learning_rate": 2.859618717504333e-08, | |
| "loss": 2.0676, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 2.835130970724191, | |
| "grad_norm": 0.5422700643539429, | |
| "learning_rate": 2.816291161178509e-08, | |
| "loss": 2.0679, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.8375963020030817, | |
| "grad_norm": 0.5508624911308289, | |
| "learning_rate": 2.772963604852686e-08, | |
| "loss": 2.0311, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 2.840061633281972, | |
| "grad_norm": 0.531106173992157, | |
| "learning_rate": 2.7296360485268627e-08, | |
| "loss": 2.0733, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 2.8425269645608626, | |
| "grad_norm": 0.5154179930686951, | |
| "learning_rate": 2.6863084922010395e-08, | |
| "loss": 1.9951, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 2.8449922958397535, | |
| "grad_norm": 0.5272330045700073, | |
| "learning_rate": 2.6429809358752164e-08, | |
| "loss": 2.0557, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 2.847457627118644, | |
| "grad_norm": 0.5329026579856873, | |
| "learning_rate": 2.5996533795493932e-08, | |
| "loss": 2.0669, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.8499229583975345, | |
| "grad_norm": 0.513034999370575, | |
| "learning_rate": 2.55632582322357e-08, | |
| "loss": 1.9837, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 2.8523882896764254, | |
| "grad_norm": 0.5385428071022034, | |
| "learning_rate": 2.5129982668977468e-08, | |
| "loss": 2.062, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 2.854853620955316, | |
| "grad_norm": 0.5449069738388062, | |
| "learning_rate": 2.4696707105719236e-08, | |
| "loss": 2.1095, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 2.8573189522342064, | |
| "grad_norm": 0.551082193851471, | |
| "learning_rate": 2.4263431542461004e-08, | |
| "loss": 2.0575, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 2.859784283513097, | |
| "grad_norm": 0.5356119275093079, | |
| "learning_rate": 2.3830155979202773e-08, | |
| "loss": 1.9981, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.862249614791988, | |
| "grad_norm": 0.5164713263511658, | |
| "learning_rate": 2.339688041594454e-08, | |
| "loss": 2.0084, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 2.8647149460708783, | |
| "grad_norm": 0.5730435848236084, | |
| "learning_rate": 2.296360485268631e-08, | |
| "loss": 2.0758, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 2.8671802773497688, | |
| "grad_norm": 0.5603390336036682, | |
| "learning_rate": 2.2530329289428077e-08, | |
| "loss": 2.044, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 2.8696456086286597, | |
| "grad_norm": 0.5511407852172852, | |
| "learning_rate": 2.2097053726169845e-08, | |
| "loss": 2.0807, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 2.87211093990755, | |
| "grad_norm": 0.5323342084884644, | |
| "learning_rate": 2.166377816291161e-08, | |
| "loss": 2.0432, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.8745762711864407, | |
| "grad_norm": 0.5549432039260864, | |
| "learning_rate": 2.123050259965338e-08, | |
| "loss": 2.0042, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 2.877041602465331, | |
| "grad_norm": 0.506240963935852, | |
| "learning_rate": 2.0797227036395147e-08, | |
| "loss": 2.0495, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 2.8795069337442216, | |
| "grad_norm": 0.5654304027557373, | |
| "learning_rate": 2.0363951473136915e-08, | |
| "loss": 2.0668, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 2.8819722650231125, | |
| "grad_norm": 0.5406729578971863, | |
| "learning_rate": 1.993067590987868e-08, | |
| "loss": 2.1237, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 2.884437596302003, | |
| "grad_norm": 0.5315223932266235, | |
| "learning_rate": 1.9497400346620448e-08, | |
| "loss": 2.0395, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.886902927580894, | |
| "grad_norm": 0.5458704233169556, | |
| "learning_rate": 1.9064124783362216e-08, | |
| "loss": 2.0812, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 2.8893682588597844, | |
| "grad_norm": 0.5505434274673462, | |
| "learning_rate": 1.8630849220103984e-08, | |
| "loss": 2.1015, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 2.891833590138675, | |
| "grad_norm": 0.511069655418396, | |
| "learning_rate": 1.8197573656845753e-08, | |
| "loss": 2.0394, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 2.8942989214175654, | |
| "grad_norm": 0.5475332140922546, | |
| "learning_rate": 1.776429809358752e-08, | |
| "loss": 2.0576, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 2.896764252696456, | |
| "grad_norm": 0.5790749788284302, | |
| "learning_rate": 1.733102253032929e-08, | |
| "loss": 2.0159, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.899229583975347, | |
| "grad_norm": 0.5202337503433228, | |
| "learning_rate": 1.6897746967071057e-08, | |
| "loss": 2.0111, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 2.9016949152542373, | |
| "grad_norm": 0.5084658861160278, | |
| "learning_rate": 1.6464471403812825e-08, | |
| "loss": 2.0603, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 2.9041602465331278, | |
| "grad_norm": 0.5587444305419922, | |
| "learning_rate": 1.6031195840554593e-08, | |
| "loss": 2.1044, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 2.9066255778120187, | |
| "grad_norm": 0.5491884350776672, | |
| "learning_rate": 1.559792027729636e-08, | |
| "loss": 2.051, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 2.909090909090909, | |
| "grad_norm": 0.5718016624450684, | |
| "learning_rate": 1.5164644714038127e-08, | |
| "loss": 2.1022, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.9115562403697997, | |
| "grad_norm": 0.5275129079818726, | |
| "learning_rate": 1.4731369150779896e-08, | |
| "loss": 2.0739, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 2.91402157164869, | |
| "grad_norm": 0.5488988757133484, | |
| "learning_rate": 1.4298093587521665e-08, | |
| "loss": 2.0944, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 2.916486902927581, | |
| "grad_norm": 0.5559231638908386, | |
| "learning_rate": 1.386481802426343e-08, | |
| "loss": 2.1181, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 2.9189522342064715, | |
| "grad_norm": 0.5378817319869995, | |
| "learning_rate": 1.3431542461005198e-08, | |
| "loss": 2.0507, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 2.921417565485362, | |
| "grad_norm": 0.51389080286026, | |
| "learning_rate": 1.2998266897746966e-08, | |
| "loss": 2.0445, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.923882896764253, | |
| "grad_norm": 0.5524910688400269, | |
| "learning_rate": 1.2564991334488734e-08, | |
| "loss": 2.0815, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 2.9263482280431434, | |
| "grad_norm": 0.5438218116760254, | |
| "learning_rate": 1.2131715771230502e-08, | |
| "loss": 1.9915, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 2.928813559322034, | |
| "grad_norm": 0.5185765624046326, | |
| "learning_rate": 1.169844020797227e-08, | |
| "loss": 2.0385, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 2.9312788906009244, | |
| "grad_norm": 0.526657223701477, | |
| "learning_rate": 1.1265164644714039e-08, | |
| "loss": 2.0811, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 2.933744221879815, | |
| "grad_norm": 0.531923234462738, | |
| "learning_rate": 1.0831889081455805e-08, | |
| "loss": 2.0466, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.936209553158706, | |
| "grad_norm": 0.5281239151954651, | |
| "learning_rate": 1.0398613518197573e-08, | |
| "loss": 2.0746, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 2.9386748844375963, | |
| "grad_norm": 0.5412416458129883, | |
| "learning_rate": 9.96533795493934e-09, | |
| "loss": 2.043, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 2.9411402157164868, | |
| "grad_norm": 0.5383726358413696, | |
| "learning_rate": 9.532062391681108e-09, | |
| "loss": 2.026, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 2.9436055469953777, | |
| "grad_norm": 0.5251293182373047, | |
| "learning_rate": 9.098786828422876e-09, | |
| "loss": 2.015, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 2.946070878274268, | |
| "grad_norm": 0.5310044288635254, | |
| "learning_rate": 8.665511265164644e-09, | |
| "loss": 2.0473, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.9485362095531586, | |
| "grad_norm": 0.5628396272659302, | |
| "learning_rate": 8.232235701906413e-09, | |
| "loss": 2.0538, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 2.951001540832049, | |
| "grad_norm": 0.5359899997711182, | |
| "learning_rate": 7.79896013864818e-09, | |
| "loss": 2.0877, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 2.95346687211094, | |
| "grad_norm": 0.5459215641021729, | |
| "learning_rate": 7.365684575389948e-09, | |
| "loss": 2.0309, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 2.9559322033898305, | |
| "grad_norm": 0.5681634545326233, | |
| "learning_rate": 6.932409012131715e-09, | |
| "loss": 2.1033, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 2.958397534668721, | |
| "grad_norm": 0.5362127423286438, | |
| "learning_rate": 6.499133448873483e-09, | |
| "loss": 2.0645, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.960862865947612, | |
| "grad_norm": 0.5522863864898682, | |
| "learning_rate": 6.065857885615251e-09, | |
| "loss": 2.0663, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 2.9633281972265024, | |
| "grad_norm": 0.558259129524231, | |
| "learning_rate": 5.632582322357019e-09, | |
| "loss": 2.0855, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 2.965793528505393, | |
| "grad_norm": 0.5171055793762207, | |
| "learning_rate": 5.199306759098787e-09, | |
| "loss": 2.0014, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 2.9682588597842834, | |
| "grad_norm": 0.5289451479911804, | |
| "learning_rate": 4.766031195840554e-09, | |
| "loss": 1.9726, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 2.970724191063174, | |
| "grad_norm": 0.5061673521995544, | |
| "learning_rate": 4.332755632582322e-09, | |
| "loss": 2.0545, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.973189522342065, | |
| "grad_norm": 0.5542516708374023, | |
| "learning_rate": 3.89948006932409e-09, | |
| "loss": 2.0868, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 2.9756548536209553, | |
| "grad_norm": 0.5415639281272888, | |
| "learning_rate": 3.4662045060658574e-09, | |
| "loss": 2.1442, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 2.9781201848998458, | |
| "grad_norm": 0.5381219387054443, | |
| "learning_rate": 3.0329289428076256e-09, | |
| "loss": 2.0866, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 2.9805855161787367, | |
| "grad_norm": 0.5186250805854797, | |
| "learning_rate": 2.5996533795493933e-09, | |
| "loss": 2.0547, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 2.983050847457627, | |
| "grad_norm": 0.5454862713813782, | |
| "learning_rate": 2.166377816291161e-09, | |
| "loss": 2.0, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.9855161787365176, | |
| "grad_norm": 0.5138260126113892, | |
| "learning_rate": 1.7331022530329287e-09, | |
| "loss": 2.0755, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 2.987981510015408, | |
| "grad_norm": 0.5278797745704651, | |
| "learning_rate": 1.2998266897746967e-09, | |
| "loss": 2.0513, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 2.990446841294299, | |
| "grad_norm": 0.5508170127868652, | |
| "learning_rate": 8.665511265164643e-10, | |
| "loss": 2.0949, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 2.9929121725731895, | |
| "grad_norm": 0.5479553937911987, | |
| "learning_rate": 4.3327556325823217e-10, | |
| "loss": 2.0683, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 2.99537750385208, | |
| "grad_norm": 0.5673417448997498, | |
| "learning_rate": 0.0, | |
| "loss": 2.1179, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.99537750385208, | |
| "eval_loss": 2.0652668476104736, | |
| "eval_runtime": 398.8201, | |
| "eval_samples_per_second": 1.234, | |
| "eval_steps_per_second": 0.308, | |
| "step": 1215 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1215, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.5867926853289247e+18, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |