| { | |
| "best_metric": 0.907011091709137, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.007006358270130143, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 3.5031791350650716e-05, | |
| "grad_norm": 12.662116050720215, | |
| "learning_rate": 1.008e-05, | |
| "loss": 2.7705, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 3.5031791350650716e-05, | |
| "eval_loss": 3.4625937938690186, | |
| "eval_runtime": 2057.3933, | |
| "eval_samples_per_second": 5.842, | |
| "eval_steps_per_second": 1.461, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 7.006358270130143e-05, | |
| "grad_norm": 16.384368896484375, | |
| "learning_rate": 2.016e-05, | |
| "loss": 3.3597, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00010509537405195214, | |
| "grad_norm": 20.63599395751953, | |
| "learning_rate": 3.024e-05, | |
| "loss": 4.0404, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00014012716540260286, | |
| "grad_norm": 13.627049446105957, | |
| "learning_rate": 4.032e-05, | |
| "loss": 2.6582, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00017515895675325357, | |
| "grad_norm": 14.674355506896973, | |
| "learning_rate": 5.04e-05, | |
| "loss": 3.0364, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00021019074810390428, | |
| "grad_norm": 12.088839530944824, | |
| "learning_rate": 6.048e-05, | |
| "loss": 2.7996, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.000245222539454555, | |
| "grad_norm": 13.726351737976074, | |
| "learning_rate": 7.055999999999999e-05, | |
| "loss": 2.8547, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00028025433080520573, | |
| "grad_norm": 9.592828750610352, | |
| "learning_rate": 8.064e-05, | |
| "loss": 1.8008, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00031528612215585646, | |
| "grad_norm": 19.422744750976562, | |
| "learning_rate": 9.072e-05, | |
| "loss": 2.4474, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00035031791350650715, | |
| "grad_norm": 14.529265403747559, | |
| "learning_rate": 0.0001008, | |
| "loss": 2.1343, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0003853497048571579, | |
| "grad_norm": 12.239635467529297, | |
| "learning_rate": 0.00010026947368421052, | |
| "loss": 2.4732, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.00042038149620780857, | |
| "grad_norm": 9.195387840270996, | |
| "learning_rate": 9.973894736842104e-05, | |
| "loss": 1.5237, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0004554132875584593, | |
| "grad_norm": 9.380248069763184, | |
| "learning_rate": 9.920842105263157e-05, | |
| "loss": 1.9376, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.00049044507890911, | |
| "grad_norm": 10.060378074645996, | |
| "learning_rate": 9.86778947368421e-05, | |
| "loss": 2.0215, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0005254768702597607, | |
| "grad_norm": 9.508438110351562, | |
| "learning_rate": 9.814736842105264e-05, | |
| "loss": 1.7224, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0005605086616104115, | |
| "grad_norm": 10.150548934936523, | |
| "learning_rate": 9.761684210526316e-05, | |
| "loss": 1.8087, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0005955404529610622, | |
| "grad_norm": 11.015765190124512, | |
| "learning_rate": 9.708631578947368e-05, | |
| "loss": 1.4277, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0006305722443117129, | |
| "grad_norm": 10.192193031311035, | |
| "learning_rate": 9.655578947368421e-05, | |
| "loss": 1.3218, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0006656040356623636, | |
| "grad_norm": 9.296463012695312, | |
| "learning_rate": 9.602526315789473e-05, | |
| "loss": 1.6145, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0007006358270130143, | |
| "grad_norm": 10.011585235595703, | |
| "learning_rate": 9.549473684210525e-05, | |
| "loss": 1.5533, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.000735667618363665, | |
| "grad_norm": 10.020069122314453, | |
| "learning_rate": 9.496421052631579e-05, | |
| "loss": 1.6376, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0007706994097143158, | |
| "grad_norm": 7.785070896148682, | |
| "learning_rate": 9.443368421052631e-05, | |
| "loss": 1.3359, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0008057312010649665, | |
| "grad_norm": 6.168704509735107, | |
| "learning_rate": 9.390315789473683e-05, | |
| "loss": 1.0551, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0008407629924156171, | |
| "grad_norm": 5.922638416290283, | |
| "learning_rate": 9.337263157894737e-05, | |
| "loss": 0.9874, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0008757947837662679, | |
| "grad_norm": 5.987689018249512, | |
| "learning_rate": 9.28421052631579e-05, | |
| "loss": 0.6966, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0009108265751169186, | |
| "grad_norm": 6.146247386932373, | |
| "learning_rate": 9.231157894736842e-05, | |
| "loss": 1.022, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0009458583664675693, | |
| "grad_norm": 5.5729079246521, | |
| "learning_rate": 9.178105263157895e-05, | |
| "loss": 1.0427, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.00098089015781822, | |
| "grad_norm": 8.170185089111328, | |
| "learning_rate": 9.125052631578948e-05, | |
| "loss": 1.6254, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0010159219491688708, | |
| "grad_norm": 4.396035194396973, | |
| "learning_rate": 9.072e-05, | |
| "loss": 0.733, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0010509537405195214, | |
| "grad_norm": 6.016524791717529, | |
| "learning_rate": 9.018947368421052e-05, | |
| "loss": 1.0773, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0010859855318701723, | |
| "grad_norm": 5.453558444976807, | |
| "learning_rate": 8.965894736842104e-05, | |
| "loss": 0.8922, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.001121017323220823, | |
| "grad_norm": 6.520124435424805, | |
| "learning_rate": 8.912842105263157e-05, | |
| "loss": 0.7138, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0011560491145714735, | |
| "grad_norm": 5.270682334899902, | |
| "learning_rate": 8.85978947368421e-05, | |
| "loss": 0.705, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0011910809059221244, | |
| "grad_norm": 6.4010162353515625, | |
| "learning_rate": 8.806736842105264e-05, | |
| "loss": 0.9197, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.001226112697272775, | |
| "grad_norm": 6.343068599700928, | |
| "learning_rate": 8.753684210526316e-05, | |
| "loss": 0.9759, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0012611444886234259, | |
| "grad_norm": 7.611424446105957, | |
| "learning_rate": 8.700631578947369e-05, | |
| "loss": 0.9741, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0012961762799740765, | |
| "grad_norm": 6.170362949371338, | |
| "learning_rate": 8.647578947368421e-05, | |
| "loss": 1.2902, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0013312080713247271, | |
| "grad_norm": 6.117220401763916, | |
| "learning_rate": 8.594526315789473e-05, | |
| "loss": 0.7671, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.001366239862675378, | |
| "grad_norm": 7.04514217376709, | |
| "learning_rate": 8.541473684210525e-05, | |
| "loss": 1.036, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0014012716540260286, | |
| "grad_norm": 9.130364418029785, | |
| "learning_rate": 8.488421052631578e-05, | |
| "loss": 1.1909, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0014363034453766794, | |
| "grad_norm": 4.37771463394165, | |
| "learning_rate": 8.435368421052631e-05, | |
| "loss": 0.5097, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.00147133523672733, | |
| "grad_norm": 5.883099555969238, | |
| "learning_rate": 8.382315789473684e-05, | |
| "loss": 0.8581, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0015063670280779807, | |
| "grad_norm": 6.91945219039917, | |
| "learning_rate": 8.329263157894737e-05, | |
| "loss": 1.3828, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0015413988194286315, | |
| "grad_norm": 6.779603004455566, | |
| "learning_rate": 8.27621052631579e-05, | |
| "loss": 1.1814, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0015764306107792822, | |
| "grad_norm": 6.361364364624023, | |
| "learning_rate": 8.223157894736842e-05, | |
| "loss": 0.8581, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.001611462402129933, | |
| "grad_norm": 5.481888294219971, | |
| "learning_rate": 8.170105263157894e-05, | |
| "loss": 0.7128, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0016464941934805836, | |
| "grad_norm": 6.276697158813477, | |
| "learning_rate": 8.117052631578946e-05, | |
| "loss": 1.0992, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0016815259848312343, | |
| "grad_norm": 6.994898319244385, | |
| "learning_rate": 8.064e-05, | |
| "loss": 0.8142, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.001716557776181885, | |
| "grad_norm": 4.883788585662842, | |
| "learning_rate": 8.010947368421052e-05, | |
| "loss": 0.5797, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0017515895675325357, | |
| "grad_norm": 6.223549842834473, | |
| "learning_rate": 7.957894736842105e-05, | |
| "loss": 1.2047, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0017515895675325357, | |
| "eval_loss": 1.1290020942687988, | |
| "eval_runtime": 2061.5746, | |
| "eval_samples_per_second": 5.83, | |
| "eval_steps_per_second": 1.458, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0017866213588831866, | |
| "grad_norm": 5.159853458404541, | |
| "learning_rate": 7.904842105263158e-05, | |
| "loss": 1.027, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0018216531502338372, | |
| "grad_norm": 5.4184088706970215, | |
| "learning_rate": 7.85178947368421e-05, | |
| "loss": 1.286, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0018566849415844878, | |
| "grad_norm": 6.966073989868164, | |
| "learning_rate": 7.798736842105263e-05, | |
| "loss": 1.4248, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0018917167329351387, | |
| "grad_norm": 6.474468231201172, | |
| "learning_rate": 7.745684210526315e-05, | |
| "loss": 1.5506, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0019267485242857893, | |
| "grad_norm": 6.7876811027526855, | |
| "learning_rate": 7.692631578947369e-05, | |
| "loss": 1.6668, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.00196178031563644, | |
| "grad_norm": 6.088242530822754, | |
| "learning_rate": 7.639578947368421e-05, | |
| "loss": 1.4712, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0019968121069870908, | |
| "grad_norm": 4.957919597625732, | |
| "learning_rate": 7.586526315789473e-05, | |
| "loss": 1.2942, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0020318438983377416, | |
| "grad_norm": 5.453049659729004, | |
| "learning_rate": 7.533473684210526e-05, | |
| "loss": 1.1897, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.002066875689688392, | |
| "grad_norm": 8.739358901977539, | |
| "learning_rate": 7.480421052631578e-05, | |
| "loss": 1.2799, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.002101907481039043, | |
| "grad_norm": 8.237629890441895, | |
| "learning_rate": 7.427368421052632e-05, | |
| "loss": 1.5075, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0021369392723896937, | |
| "grad_norm": 6.064999580383301, | |
| "learning_rate": 7.374315789473685e-05, | |
| "loss": 1.4072, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0021719710637403446, | |
| "grad_norm": 7.591026306152344, | |
| "learning_rate": 7.321263157894737e-05, | |
| "loss": 1.4841, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.002207002855090995, | |
| "grad_norm": 6.479332447052002, | |
| "learning_rate": 7.26821052631579e-05, | |
| "loss": 1.3464, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.002242034646441646, | |
| "grad_norm": 7.660516738891602, | |
| "learning_rate": 7.215157894736842e-05, | |
| "loss": 1.3947, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0022770664377922967, | |
| "grad_norm": 6.516019821166992, | |
| "learning_rate": 7.162105263157894e-05, | |
| "loss": 0.9144, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.002312098229142947, | |
| "grad_norm": 8.34161376953125, | |
| "learning_rate": 7.109052631578947e-05, | |
| "loss": 1.6912, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.002347130020493598, | |
| "grad_norm": 10.69320297241211, | |
| "learning_rate": 7.055999999999999e-05, | |
| "loss": 1.9, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0023821618118442488, | |
| "grad_norm": 10.620810508728027, | |
| "learning_rate": 7.002947368421052e-05, | |
| "loss": 1.0454, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.002417193603194899, | |
| "grad_norm": 7.011707305908203, | |
| "learning_rate": 6.949894736842105e-05, | |
| "loss": 0.8123, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.00245222539454555, | |
| "grad_norm": 7.232404708862305, | |
| "learning_rate": 6.896842105263158e-05, | |
| "loss": 1.4846, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.002487257185896201, | |
| "grad_norm": 7.875359535217285, | |
| "learning_rate": 6.843789473684211e-05, | |
| "loss": 1.2222, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0025222889772468517, | |
| "grad_norm": 6.384848594665527, | |
| "learning_rate": 6.790736842105263e-05, | |
| "loss": 1.0321, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.002557320768597502, | |
| "grad_norm": 6.461179733276367, | |
| "learning_rate": 6.737684210526315e-05, | |
| "loss": 0.9518, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.002592352559948153, | |
| "grad_norm": 7.334846019744873, | |
| "learning_rate": 6.684631578947368e-05, | |
| "loss": 1.2871, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.002627384351298804, | |
| "grad_norm": 5.616960048675537, | |
| "learning_rate": 6.631578947368421e-05, | |
| "loss": 0.9077, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0026624161426494542, | |
| "grad_norm": 4.717360496520996, | |
| "learning_rate": 6.578526315789473e-05, | |
| "loss": 0.6344, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.002697447934000105, | |
| "grad_norm": 4.63266658782959, | |
| "learning_rate": 6.525473684210526e-05, | |
| "loss": 0.7442, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.002732479725350756, | |
| "grad_norm": 6.365645408630371, | |
| "learning_rate": 6.47242105263158e-05, | |
| "loss": 0.9088, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0027675115167014063, | |
| "grad_norm": 5.476707935333252, | |
| "learning_rate": 6.419368421052632e-05, | |
| "loss": 0.7585, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.002802543308052057, | |
| "grad_norm": 6.600567817687988, | |
| "learning_rate": 6.366315789473684e-05, | |
| "loss": 1.1681, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.002837575099402708, | |
| "grad_norm": 4.197580337524414, | |
| "learning_rate": 6.313263157894736e-05, | |
| "loss": 0.5965, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.002872606890753359, | |
| "grad_norm": 5.1230788230896, | |
| "learning_rate": 6.26021052631579e-05, | |
| "loss": 0.978, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0029076386821040093, | |
| "grad_norm": 5.09016752243042, | |
| "learning_rate": 6.207157894736842e-05, | |
| "loss": 0.663, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.00294267047345466, | |
| "grad_norm": 6.350630283355713, | |
| "learning_rate": 6.154105263157894e-05, | |
| "loss": 0.9641, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.002977702264805311, | |
| "grad_norm": 6.287975788116455, | |
| "learning_rate": 6.1010526315789474e-05, | |
| "loss": 1.2104, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0030127340561559614, | |
| "grad_norm": 5.212226867675781, | |
| "learning_rate": 6.048e-05, | |
| "loss": 0.9887, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.0030477658475066122, | |
| "grad_norm": 6.213005542755127, | |
| "learning_rate": 5.994947368421052e-05, | |
| "loss": 0.6417, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.003082797638857263, | |
| "grad_norm": 6.072321891784668, | |
| "learning_rate": 5.941894736842104e-05, | |
| "loss": 0.8963, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0031178294302079135, | |
| "grad_norm": 4.825734615325928, | |
| "learning_rate": 5.888842105263158e-05, | |
| "loss": 0.742, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0031528612215585643, | |
| "grad_norm": 5.300004959106445, | |
| "learning_rate": 5.835789473684211e-05, | |
| "loss": 0.7844, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.003187893012909215, | |
| "grad_norm": 8.157235145568848, | |
| "learning_rate": 5.782736842105263e-05, | |
| "loss": 1.2107, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.003222924804259866, | |
| "grad_norm": 4.026047706604004, | |
| "learning_rate": 5.7296842105263154e-05, | |
| "loss": 0.503, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0032579565956105164, | |
| "grad_norm": 5.493070125579834, | |
| "learning_rate": 5.676631578947368e-05, | |
| "loss": 0.8302, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0032929883869611673, | |
| "grad_norm": 5.728854179382324, | |
| "learning_rate": 5.623578947368421e-05, | |
| "loss": 0.7791, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.003328020178311818, | |
| "grad_norm": 5.390102863311768, | |
| "learning_rate": 5.570526315789474e-05, | |
| "loss": 0.9903, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0033630519696624685, | |
| "grad_norm": 6.115237712860107, | |
| "learning_rate": 5.5174736842105266e-05, | |
| "loss": 0.7596, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0033980837610131194, | |
| "grad_norm": 5.043397426605225, | |
| "learning_rate": 5.464421052631579e-05, | |
| "loss": 0.7616, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.00343311555236377, | |
| "grad_norm": 5.883362293243408, | |
| "learning_rate": 5.411368421052631e-05, | |
| "loss": 0.8057, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0034681473437144206, | |
| "grad_norm": 6.632864952087402, | |
| "learning_rate": 5.358315789473684e-05, | |
| "loss": 1.0609, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0035031791350650715, | |
| "grad_norm": 4.529953956604004, | |
| "learning_rate": 5.3052631578947364e-05, | |
| "loss": 0.6286, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0035031791350650715, | |
| "eval_loss": 1.029884696006775, | |
| "eval_runtime": 2061.1807, | |
| "eval_samples_per_second": 5.831, | |
| "eval_steps_per_second": 1.458, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0035382109264157223, | |
| "grad_norm": 4.1783599853515625, | |
| "learning_rate": 5.252210526315789e-05, | |
| "loss": 0.9496, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.003573242717766373, | |
| "grad_norm": 5.620020866394043, | |
| "learning_rate": 5.199157894736842e-05, | |
| "loss": 1.3367, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.0036082745091170236, | |
| "grad_norm": 5.590014457702637, | |
| "learning_rate": 5.1461052631578946e-05, | |
| "loss": 1.0153, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0036433063004676744, | |
| "grad_norm": 5.261481285095215, | |
| "learning_rate": 5.0930526315789476e-05, | |
| "loss": 0.9291, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.0036783380918183253, | |
| "grad_norm": 6.13337516784668, | |
| "learning_rate": 5.04e-05, | |
| "loss": 1.0299, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.0037133698831689757, | |
| "grad_norm": 4.944069862365723, | |
| "learning_rate": 4.986947368421052e-05, | |
| "loss": 0.9319, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.0037484016745196265, | |
| "grad_norm": 5.500107765197754, | |
| "learning_rate": 4.933894736842105e-05, | |
| "loss": 0.9859, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.0037834334658702774, | |
| "grad_norm": 4.677202224731445, | |
| "learning_rate": 4.880842105263158e-05, | |
| "loss": 0.9123, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.003818465257220928, | |
| "grad_norm": 6.256136417388916, | |
| "learning_rate": 4.8277894736842103e-05, | |
| "loss": 1.1807, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.0038534970485715786, | |
| "grad_norm": 8.950167655944824, | |
| "learning_rate": 4.7747368421052626e-05, | |
| "loss": 1.1984, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0038885288399222295, | |
| "grad_norm": 7.130908966064453, | |
| "learning_rate": 4.7216842105263156e-05, | |
| "loss": 1.4225, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.00392356063127288, | |
| "grad_norm": 7.4759931564331055, | |
| "learning_rate": 4.6686315789473686e-05, | |
| "loss": 1.3882, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.003958592422623531, | |
| "grad_norm": 10.55746078491211, | |
| "learning_rate": 4.615578947368421e-05, | |
| "loss": 1.5642, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0039936242139741816, | |
| "grad_norm": 7.839519500732422, | |
| "learning_rate": 4.562526315789474e-05, | |
| "loss": 1.2803, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.004028656005324832, | |
| "grad_norm": 7.067453861236572, | |
| "learning_rate": 4.509473684210526e-05, | |
| "loss": 1.0843, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.004063687796675483, | |
| "grad_norm": 7.410902500152588, | |
| "learning_rate": 4.4564210526315784e-05, | |
| "loss": 1.1034, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.004098719588026134, | |
| "grad_norm": 12.82276439666748, | |
| "learning_rate": 4.403368421052632e-05, | |
| "loss": 1.9388, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.004133751379376784, | |
| "grad_norm": 11.6984281539917, | |
| "learning_rate": 4.350315789473684e-05, | |
| "loss": 2.2554, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.004168783170727435, | |
| "grad_norm": 9.550557136535645, | |
| "learning_rate": 4.2972631578947366e-05, | |
| "loss": 1.2014, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.004203814962078086, | |
| "grad_norm": 6.161247730255127, | |
| "learning_rate": 4.244210526315789e-05, | |
| "loss": 0.8151, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.004238846753428736, | |
| "grad_norm": 7.186214447021484, | |
| "learning_rate": 4.191157894736842e-05, | |
| "loss": 0.9695, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.0042738785447793875, | |
| "grad_norm": 6.048009395599365, | |
| "learning_rate": 4.138105263157895e-05, | |
| "loss": 0.9508, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.004308910336130038, | |
| "grad_norm": 5.062438011169434, | |
| "learning_rate": 4.085052631578947e-05, | |
| "loss": 0.7397, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.004343942127480689, | |
| "grad_norm": 5.460752010345459, | |
| "learning_rate": 4.032e-05, | |
| "loss": 0.8562, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0043789739188313396, | |
| "grad_norm": 4.651021480560303, | |
| "learning_rate": 3.978947368421052e-05, | |
| "loss": 0.7181, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.00441400571018199, | |
| "grad_norm": 7.1053667068481445, | |
| "learning_rate": 3.925894736842105e-05, | |
| "loss": 0.9492, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.004449037501532641, | |
| "grad_norm": 4.558400630950928, | |
| "learning_rate": 3.8728421052631575e-05, | |
| "loss": 0.8316, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.004484069292883292, | |
| "grad_norm": 4.557405948638916, | |
| "learning_rate": 3.8197894736842105e-05, | |
| "loss": 0.6187, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.004519101084233942, | |
| "grad_norm": 6.3636298179626465, | |
| "learning_rate": 3.766736842105263e-05, | |
| "loss": 0.7996, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.004554132875584593, | |
| "grad_norm": 4.011357307434082, | |
| "learning_rate": 3.713684210526316e-05, | |
| "loss": 0.7043, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.004589164666935244, | |
| "grad_norm": 5.260727405548096, | |
| "learning_rate": 3.660631578947369e-05, | |
| "loss": 0.6685, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.004624196458285894, | |
| "grad_norm": 5.384383678436279, | |
| "learning_rate": 3.607578947368421e-05, | |
| "loss": 0.8229, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.0046592282496365454, | |
| "grad_norm": 5.44893217086792, | |
| "learning_rate": 3.554526315789473e-05, | |
| "loss": 0.6983, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.004694260040987196, | |
| "grad_norm": 5.031147003173828, | |
| "learning_rate": 3.501473684210526e-05, | |
| "loss": 0.6835, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.004729291832337846, | |
| "grad_norm": 5.930494785308838, | |
| "learning_rate": 3.448421052631579e-05, | |
| "loss": 0.9263, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.0047643236236884975, | |
| "grad_norm": 5.080932140350342, | |
| "learning_rate": 3.3953684210526315e-05, | |
| "loss": 0.7292, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.004799355415039148, | |
| "grad_norm": 5.154940128326416, | |
| "learning_rate": 3.342315789473684e-05, | |
| "loss": 0.6465, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.004834387206389798, | |
| "grad_norm": 5.2444939613342285, | |
| "learning_rate": 3.289263157894737e-05, | |
| "loss": 0.7776, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.00486941899774045, | |
| "grad_norm": 5.0661211013793945, | |
| "learning_rate": 3.23621052631579e-05, | |
| "loss": 0.7375, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.0049044507890911, | |
| "grad_norm": 4.658995628356934, | |
| "learning_rate": 3.183157894736842e-05, | |
| "loss": 0.5983, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0049394825804417505, | |
| "grad_norm": 5.3926920890808105, | |
| "learning_rate": 3.130105263157895e-05, | |
| "loss": 0.9688, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.004974514371792402, | |
| "grad_norm": 6.5495805740356445, | |
| "learning_rate": 3.077052631578947e-05, | |
| "loss": 0.912, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.005009546163143052, | |
| "grad_norm": 6.6911702156066895, | |
| "learning_rate": 3.024e-05, | |
| "loss": 1.257, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.0050445779544937034, | |
| "grad_norm": 5.760495662689209, | |
| "learning_rate": 2.970947368421052e-05, | |
| "loss": 0.8503, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.005079609745844354, | |
| "grad_norm": 3.4422309398651123, | |
| "learning_rate": 2.9178947368421054e-05, | |
| "loss": 0.4197, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.005114641537195004, | |
| "grad_norm": 4.122685432434082, | |
| "learning_rate": 2.8648421052631577e-05, | |
| "loss": 0.6266, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.0051496733285456555, | |
| "grad_norm": 4.601315975189209, | |
| "learning_rate": 2.8117894736842103e-05, | |
| "loss": 0.5214, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.005184705119896306, | |
| "grad_norm": 4.702164173126221, | |
| "learning_rate": 2.7587368421052633e-05, | |
| "loss": 0.6284, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.005219736911246956, | |
| "grad_norm": 5.033868789672852, | |
| "learning_rate": 2.7056842105263156e-05, | |
| "loss": 0.6065, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.005254768702597608, | |
| "grad_norm": 5.247815132141113, | |
| "learning_rate": 2.6526315789473682e-05, | |
| "loss": 0.887, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.005254768702597608, | |
| "eval_loss": 0.9501329064369202, | |
| "eval_runtime": 2058.4097, | |
| "eval_samples_per_second": 5.839, | |
| "eval_steps_per_second": 1.46, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.005289800493948258, | |
| "grad_norm": 4.7826056480407715, | |
| "learning_rate": 2.599578947368421e-05, | |
| "loss": 0.8471, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.0053248322852989085, | |
| "grad_norm": 4.294804573059082, | |
| "learning_rate": 2.5465263157894738e-05, | |
| "loss": 0.8585, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.00535986407664956, | |
| "grad_norm": 5.511264324188232, | |
| "learning_rate": 2.493473684210526e-05, | |
| "loss": 1.0528, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.00539489586800021, | |
| "grad_norm": 6.201894760131836, | |
| "learning_rate": 2.440421052631579e-05, | |
| "loss": 1.1332, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.0054299276593508606, | |
| "grad_norm": 4.838656425476074, | |
| "learning_rate": 2.3873684210526313e-05, | |
| "loss": 0.8937, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.005464959450701512, | |
| "grad_norm": 5.65955924987793, | |
| "learning_rate": 2.3343157894736843e-05, | |
| "loss": 1.0686, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.005499991242052162, | |
| "grad_norm": 6.030940055847168, | |
| "learning_rate": 2.281263157894737e-05, | |
| "loss": 1.0709, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.005535023033402813, | |
| "grad_norm": 5.607932090759277, | |
| "learning_rate": 2.2282105263157892e-05, | |
| "loss": 1.0595, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.005570054824753464, | |
| "grad_norm": 4.631815433502197, | |
| "learning_rate": 2.175157894736842e-05, | |
| "loss": 0.7808, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.005605086616104114, | |
| "grad_norm": 9.391894340515137, | |
| "learning_rate": 2.1221052631578944e-05, | |
| "loss": 1.6867, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.005640118407454766, | |
| "grad_norm": 6.0099616050720215, | |
| "learning_rate": 2.0690526315789474e-05, | |
| "loss": 0.9337, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.005675150198805416, | |
| "grad_norm": 7.833316802978516, | |
| "learning_rate": 2.016e-05, | |
| "loss": 1.6899, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.0057101819901560665, | |
| "grad_norm": 7.131737232208252, | |
| "learning_rate": 1.9629473684210526e-05, | |
| "loss": 1.3336, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.005745213781506718, | |
| "grad_norm": 7.738104343414307, | |
| "learning_rate": 1.9098947368421053e-05, | |
| "loss": 1.3881, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.005780245572857368, | |
| "grad_norm": 7.788103103637695, | |
| "learning_rate": 1.856842105263158e-05, | |
| "loss": 1.0198, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.0058152773642080186, | |
| "grad_norm": 7.89583158493042, | |
| "learning_rate": 1.8037894736842105e-05, | |
| "loss": 1.2767, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.00585030915555867, | |
| "grad_norm": 11.539169311523438, | |
| "learning_rate": 1.750736842105263e-05, | |
| "loss": 1.9096, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.00588534094690932, | |
| "grad_norm": 7.769778251647949, | |
| "learning_rate": 1.6976842105263157e-05, | |
| "loss": 1.4205, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.005920372738259971, | |
| "grad_norm": 4.522551536560059, | |
| "learning_rate": 1.6446315789473684e-05, | |
| "loss": 0.6861, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.005955404529610622, | |
| "grad_norm": 4.317171096801758, | |
| "learning_rate": 1.591578947368421e-05, | |
| "loss": 0.6416, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.005990436320961272, | |
| "grad_norm": 5.301867485046387, | |
| "learning_rate": 1.5385263157894736e-05, | |
| "loss": 0.8472, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.006025468112311923, | |
| "grad_norm": 3.946662425994873, | |
| "learning_rate": 1.485473684210526e-05, | |
| "loss": 0.505, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.006060499903662574, | |
| "grad_norm": 5.64309024810791, | |
| "learning_rate": 1.4324210526315789e-05, | |
| "loss": 0.9617, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.0060955316950132244, | |
| "grad_norm": 6.693427085876465, | |
| "learning_rate": 1.3793684210526316e-05, | |
| "loss": 0.8974, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.006130563486363875, | |
| "grad_norm": 4.707422733306885, | |
| "learning_rate": 1.3263157894736841e-05, | |
| "loss": 0.7757, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.006165595277714526, | |
| "grad_norm": 4.7554826736450195, | |
| "learning_rate": 1.2732631578947369e-05, | |
| "loss": 0.6396, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.0062006270690651765, | |
| "grad_norm": 4.768486499786377, | |
| "learning_rate": 1.2202105263157895e-05, | |
| "loss": 0.6989, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.006235658860415827, | |
| "grad_norm": 4.941887378692627, | |
| "learning_rate": 1.1671578947368421e-05, | |
| "loss": 0.6704, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.006270690651766478, | |
| "grad_norm": 5.827295780181885, | |
| "learning_rate": 1.1141052631578946e-05, | |
| "loss": 0.9484, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.006305722443117129, | |
| "grad_norm": 5.031542778015137, | |
| "learning_rate": 1.0610526315789472e-05, | |
| "loss": 0.7449, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.00634075423446778, | |
| "grad_norm": 5.917794704437256, | |
| "learning_rate": 1.008e-05, | |
| "loss": 1.0857, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.00637578602581843, | |
| "grad_norm": 4.596070766448975, | |
| "learning_rate": 9.549473684210526e-06, | |
| "loss": 0.6569, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.006410817817169081, | |
| "grad_norm": 4.4000325202941895, | |
| "learning_rate": 9.018947368421052e-06, | |
| "loss": 0.6524, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.006445849608519732, | |
| "grad_norm": 7.547094345092773, | |
| "learning_rate": 8.488421052631579e-06, | |
| "loss": 1.0409, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.0064808813998703824, | |
| "grad_norm": 5.421209335327148, | |
| "learning_rate": 7.957894736842105e-06, | |
| "loss": 0.8219, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.006515913191221033, | |
| "grad_norm": 6.741072177886963, | |
| "learning_rate": 7.42736842105263e-06, | |
| "loss": 0.8836, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.006550944982571684, | |
| "grad_norm": 6.431979656219482, | |
| "learning_rate": 6.896842105263158e-06, | |
| "loss": 0.9018, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.0065859767739223345, | |
| "grad_norm": 4.790152549743652, | |
| "learning_rate": 6.3663157894736845e-06, | |
| "loss": 0.5921, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.006621008565272985, | |
| "grad_norm": 5.673473358154297, | |
| "learning_rate": 5.835789473684211e-06, | |
| "loss": 0.6023, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.006656040356623636, | |
| "grad_norm": 4.413316249847412, | |
| "learning_rate": 5.305263157894736e-06, | |
| "loss": 0.6352, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.006691072147974287, | |
| "grad_norm": 5.780127048492432, | |
| "learning_rate": 4.774736842105263e-06, | |
| "loss": 0.9693, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.006726103939324937, | |
| "grad_norm": 4.126491069793701, | |
| "learning_rate": 4.244210526315789e-06, | |
| "loss": 0.5177, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.006761135730675588, | |
| "grad_norm": 8.283818244934082, | |
| "learning_rate": 3.713684210526315e-06, | |
| "loss": 0.874, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.006796167522026239, | |
| "grad_norm": 5.534949779510498, | |
| "learning_rate": 3.1831578947368422e-06, | |
| "loss": 0.8445, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.006831199313376889, | |
| "grad_norm": 6.650455951690674, | |
| "learning_rate": 2.652631578947368e-06, | |
| "loss": 1.0534, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.00686623110472754, | |
| "grad_norm": 6.208735942840576, | |
| "learning_rate": 2.1221052631578947e-06, | |
| "loss": 0.7896, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.006901262896078191, | |
| "grad_norm": 4.9419732093811035, | |
| "learning_rate": 1.5915789473684211e-06, | |
| "loss": 0.641, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.006936294687428841, | |
| "grad_norm": 5.195125579833984, | |
| "learning_rate": 1.0610526315789473e-06, | |
| "loss": 0.7821, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.0069713264787794925, | |
| "grad_norm": 5.101795196533203, | |
| "learning_rate": 5.305263157894737e-07, | |
| "loss": 0.5665, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.007006358270130143, | |
| "grad_norm": 4.9422173500061035, | |
| "learning_rate": 0.0, | |
| "loss": 0.5634, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.007006358270130143, | |
| "eval_loss": 0.907011091709137, | |
| "eval_runtime": 2060.6424, | |
| "eval_samples_per_second": 5.833, | |
| "eval_steps_per_second": 1.458, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.0571696306965709e+17, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |