| { | |
| "best_metric": 5.766289234161377, | |
| "best_model_checkpoint": "HBERTv1_emb_compress_48_L10_H512_A8/checkpoint-320000", | |
| "epoch": 3.0586689096835245, | |
| "eval_steps": 10000, | |
| "global_step": 320001, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 10.2239, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 10.0965, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.5e-06, | |
| "loss": 10.0188, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 9.9086, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.5e-06, | |
| "loss": 9.7732, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3e-06, | |
| "loss": 9.613, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.5e-06, | |
| "loss": 9.4265, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 9.2229, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.5e-06, | |
| "loss": 8.9975, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5e-06, | |
| "loss": 8.7596, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 8.5149, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6e-06, | |
| "loss": 8.2694, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 8.0297, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7e-06, | |
| "loss": 7.799, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 7.6005, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 7.4328, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.5e-06, | |
| "loss": 7.3073, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9e-06, | |
| "loss": 7.2197, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.5e-06, | |
| "loss": 7.1479, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1e-05, | |
| "loss": 7.1035, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_accuracy": 0.08438017176765814, | |
| "eval_loss": 7.083711624145508, | |
| "eval_runtime": 1116.9766, | |
| "eval_samples_per_second": 276.063, | |
| "eval_steps_per_second": 4.93, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.990255405813626e-06, | |
| "loss": 7.0607, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.98051081162725e-06, | |
| "loss": 7.0203, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.970766217440876e-06, | |
| "loss": 6.991, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.9610216232545e-06, | |
| "loss": 6.9617, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.951277029068126e-06, | |
| "loss": 6.9303, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.94153243488175e-06, | |
| "loss": 6.9057, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.931787840695375e-06, | |
| "loss": 6.886, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.922043246509e-06, | |
| "loss": 6.8666, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.912298652322625e-06, | |
| "loss": 6.8387, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.90255405813625e-06, | |
| "loss": 6.8218, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.892809463949875e-06, | |
| "loss": 6.8029, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.883064869763498e-06, | |
| "loss": 6.7808, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.873320275577123e-06, | |
| "loss": 6.7734, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.863575681390748e-06, | |
| "loss": 6.7524, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.853831087204375e-06, | |
| "loss": 6.7417, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.844086493018e-06, | |
| "loss": 6.7248, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.834341898831625e-06, | |
| "loss": 6.7081, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.82459730464525e-06, | |
| "loss": 6.6962, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.814852710458875e-06, | |
| "loss": 6.6903, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.805108116272498e-06, | |
| "loss": 6.6799, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "eval_accuracy": 0.1072023665308681, | |
| "eval_loss": 6.673709392547607, | |
| "eval_runtime": 1119.1952, | |
| "eval_samples_per_second": 275.516, | |
| "eval_steps_per_second": 4.921, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.795363522086123e-06, | |
| "loss": 6.6699, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.785618927899748e-06, | |
| "loss": 6.6557, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.775874333713373e-06, | |
| "loss": 6.6521, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.766129739526998e-06, | |
| "loss": 6.6478, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.756385145340623e-06, | |
| "loss": 6.6351, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.74664055115425e-06, | |
| "loss": 6.6231, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.736895956967874e-06, | |
| "loss": 6.6133, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.727151362781497e-06, | |
| "loss": 6.6059, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.717406768595122e-06, | |
| "loss": 6.5988, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.707662174408747e-06, | |
| "loss": 6.5963, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.697917580222372e-06, | |
| "loss": 6.5914, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.688172986035997e-06, | |
| "loss": 6.5815, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.678428391849622e-06, | |
| "loss": 6.5713, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.668683797663247e-06, | |
| "loss": 6.5611, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.658939203476872e-06, | |
| "loss": 6.5659, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.649194609290497e-06, | |
| "loss": 6.5541, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.639450015104122e-06, | |
| "loss": 6.5434, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.629705420917747e-06, | |
| "loss": 6.5385, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.619960826731372e-06, | |
| "loss": 6.5316, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.610216232544997e-06, | |
| "loss": 6.5327, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "eval_accuracy": 0.11937985240747717, | |
| "eval_loss": 6.527903079986572, | |
| "eval_runtime": 1119.8133, | |
| "eval_samples_per_second": 275.364, | |
| "eval_steps_per_second": 4.918, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.600471638358622e-06, | |
| "loss": 6.534, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.590727044172246e-06, | |
| "loss": 6.5186, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.580982449985871e-06, | |
| "loss": 6.5171, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.571237855799496e-06, | |
| "loss": 6.5145, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.561493261613121e-06, | |
| "loss": 6.4994, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.551748667426746e-06, | |
| "loss": 6.5033, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.542004073240371e-06, | |
| "loss": 6.496, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.532259479053996e-06, | |
| "loss": 6.487, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.522514884867621e-06, | |
| "loss": 6.4817, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.512770290681246e-06, | |
| "loss": 6.4834, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.50302569649487e-06, | |
| "loss": 6.4797, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.493281102308494e-06, | |
| "loss": 6.4688, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.483536508122119e-06, | |
| "loss": 6.4729, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.473791913935744e-06, | |
| "loss": 6.4625, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.46404731974937e-06, | |
| "loss": 6.4552, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.454302725562996e-06, | |
| "loss": 6.4582, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.44455813137662e-06, | |
| "loss": 6.4545, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.434813537190245e-06, | |
| "loss": 6.4369, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.425068943003869e-06, | |
| "loss": 6.4419, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.415324348817494e-06, | |
| "loss": 6.4362, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "eval_accuracy": 0.12722390205607037, | |
| "eval_loss": 6.435755252838135, | |
| "eval_runtime": 1123.6528, | |
| "eval_samples_per_second": 274.423, | |
| "eval_steps_per_second": 4.901, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.405579754631119e-06, | |
| "loss": 6.4372, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.395835160444744e-06, | |
| "loss": 6.4384, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.386090566258368e-06, | |
| "loss": 6.4255, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.376345972071993e-06, | |
| "loss": 6.4213, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.366601377885618e-06, | |
| "loss": 6.4245, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.356856783699245e-06, | |
| "loss": 6.4169, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.347112189512868e-06, | |
| "loss": 6.4101, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.337367595326493e-06, | |
| "loss": 6.4103, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.327623001140118e-06, | |
| "loss": 6.4054, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.317878406953743e-06, | |
| "loss": 6.4031, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.308133812767368e-06, | |
| "loss": 6.3926, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.298389218580993e-06, | |
| "loss": 6.3954, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.288644624394618e-06, | |
| "loss": 6.3905, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.278900030208243e-06, | |
| "loss": 6.3873, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.269155436021868e-06, | |
| "loss": 6.3857, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.259410841835493e-06, | |
| "loss": 6.3854, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.249666247649118e-06, | |
| "loss": 6.3809, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.239921653462742e-06, | |
| "loss": 6.384, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.230177059276367e-06, | |
| "loss": 6.3736, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.220432465089992e-06, | |
| "loss": 6.3648, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "eval_accuracy": 0.1334894987305005, | |
| "eval_loss": 6.370010852813721, | |
| "eval_runtime": 1122.335, | |
| "eval_samples_per_second": 274.745, | |
| "eval_steps_per_second": 4.907, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.210687870903617e-06, | |
| "loss": 6.369, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.200943276717242e-06, | |
| "loss": 6.3654, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.191198682530867e-06, | |
| "loss": 6.3651, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.181454088344492e-06, | |
| "loss": 6.3613, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.171709494158117e-06, | |
| "loss": 6.3605, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.161964899971742e-06, | |
| "loss": 6.3499, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.152220305785367e-06, | |
| "loss": 6.3519, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.142475711598992e-06, | |
| "loss": 6.3462, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.132731117412617e-06, | |
| "loss": 6.3463, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.12298652322624e-06, | |
| "loss": 6.3449, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.113241929039865e-06, | |
| "loss": 6.3346, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.10349733485349e-06, | |
| "loss": 6.349, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.093752740667115e-06, | |
| "loss": 6.3484, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.08400814648074e-06, | |
| "loss": 6.3329, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.074263552294366e-06, | |
| "loss": 6.3322, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.064518958107991e-06, | |
| "loss": 6.3257, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.054774363921616e-06, | |
| "loss": 6.3223, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.04502976973524e-06, | |
| "loss": 6.3195, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.035285175548864e-06, | |
| "loss": 6.3194, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.02554058136249e-06, | |
| "loss": 6.3181, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "eval_accuracy": 0.135479747837452, | |
| "eval_loss": 6.315780162811279, | |
| "eval_runtime": 1122.1728, | |
| "eval_samples_per_second": 274.785, | |
| "eval_steps_per_second": 4.907, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.015795987176114e-06, | |
| "loss": 6.3168, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.00605139298974e-06, | |
| "loss": 6.3186, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.996306798803364e-06, | |
| "loss": 6.3137, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.986562204616989e-06, | |
| "loss": 6.3118, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.976817610430614e-06, | |
| "loss": 6.3099, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.967073016244239e-06, | |
| "loss": 6.3034, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.957328422057864e-06, | |
| "loss": 6.3057, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.947583827871489e-06, | |
| "loss": 6.3004, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.937839233685114e-06, | |
| "loss": 6.3004, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.928094639498739e-06, | |
| "loss": 6.2981, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.918350045312364e-06, | |
| "loss": 6.3065, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.908605451125989e-06, | |
| "loss": 6.2911, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.898860856939614e-06, | |
| "loss": 6.293, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.889116262753239e-06, | |
| "loss": 6.2875, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.879371668566863e-06, | |
| "loss": 6.2893, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.869627074380488e-06, | |
| "loss": 6.2864, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.859882480194113e-06, | |
| "loss": 6.2926, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 8.850137886007738e-06, | |
| "loss": 6.2865, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 8.840393291821363e-06, | |
| "loss": 6.2802, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.830648697634988e-06, | |
| "loss": 6.2776, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "eval_accuracy": 0.1380251816919266, | |
| "eval_loss": 6.276886940002441, | |
| "eval_runtime": 1121.4341, | |
| "eval_samples_per_second": 274.966, | |
| "eval_steps_per_second": 4.911, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.820904103448613e-06, | |
| "loss": 6.2763, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.811159509262236e-06, | |
| "loss": 6.2727, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.801414915075863e-06, | |
| "loss": 6.2776, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.791670320889488e-06, | |
| "loss": 6.2712, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.781925726703113e-06, | |
| "loss": 6.2672, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.772181132516738e-06, | |
| "loss": 6.2678, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.762436538330363e-06, | |
| "loss": 6.2683, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.752691944143988e-06, | |
| "loss": 6.2659, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.742947349957611e-06, | |
| "loss": 6.2562, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.733202755771236e-06, | |
| "loss": 6.261, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.72345816158486e-06, | |
| "loss": 6.2607, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.713713567398486e-06, | |
| "loss": 6.2614, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.70396897321211e-06, | |
| "loss": 6.257, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.694224379025736e-06, | |
| "loss": 6.2502, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.684479784839362e-06, | |
| "loss": 6.2522, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.674735190652987e-06, | |
| "loss": 6.2545, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.66499059646661e-06, | |
| "loss": 6.247, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.655246002280235e-06, | |
| "loss": 6.2441, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.64550140809386e-06, | |
| "loss": 6.2497, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.635756813907485e-06, | |
| "loss": 6.2469, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "eval_accuracy": 0.13998976820299053, | |
| "eval_loss": 6.243773937225342, | |
| "eval_runtime": 1120.4439, | |
| "eval_samples_per_second": 275.209, | |
| "eval_steps_per_second": 4.915, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 8.62601221972111e-06, | |
| "loss": 6.2464, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 8.616267625534735e-06, | |
| "loss": 6.2429, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.60652303134836e-06, | |
| "loss": 6.2412, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.596778437161985e-06, | |
| "loss": 6.2439, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.58703384297561e-06, | |
| "loss": 6.2316, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.577289248789235e-06, | |
| "loss": 6.2368, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 8.56754465460286e-06, | |
| "loss": 6.2322, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 8.557800060416485e-06, | |
| "loss": 6.2328, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.54805546623011e-06, | |
| "loss": 6.2351, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.538310872043735e-06, | |
| "loss": 6.2374, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.52856627785736e-06, | |
| "loss": 6.2251, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.518821683670984e-06, | |
| "loss": 6.2333, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.50907708948461e-06, | |
| "loss": 6.2238, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.499332495298234e-06, | |
| "loss": 6.225, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.48958790111186e-06, | |
| "loss": 6.2233, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.479843306925484e-06, | |
| "loss": 6.2257, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.470098712739109e-06, | |
| "loss": 6.2234, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.460354118552734e-06, | |
| "loss": 6.2213, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.450609524366359e-06, | |
| "loss": 6.2196, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.440864930179984e-06, | |
| "loss": 6.218, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "eval_accuracy": 0.14220289100842737, | |
| "eval_loss": 6.218698501586914, | |
| "eval_runtime": 1119.0187, | |
| "eval_samples_per_second": 275.559, | |
| "eval_steps_per_second": 4.921, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.431120335993607e-06, | |
| "loss": 6.2166, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.421375741807232e-06, | |
| "loss": 6.2215, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.411631147620857e-06, | |
| "loss": 6.2193, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.401886553434484e-06, | |
| "loss": 6.2249, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.392141959248109e-06, | |
| "loss": 6.2111, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.382397365061734e-06, | |
| "loss": 6.2112, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.372652770875358e-06, | |
| "loss": 6.2154, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.362908176688983e-06, | |
| "loss": 6.2095, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.353163582502607e-06, | |
| "loss": 6.2078, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.343418988316232e-06, | |
| "loss": 6.2035, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.333674394129857e-06, | |
| "loss": 6.2087, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.323929799943481e-06, | |
| "loss": 6.2106, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.314185205757106e-06, | |
| "loss": 6.2056, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.304440611570731e-06, | |
| "loss": 6.2083, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.294696017384358e-06, | |
| "loss": 6.2052, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.284951423197981e-06, | |
| "loss": 6.2057, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.275206829011606e-06, | |
| "loss": 6.205, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.265462234825231e-06, | |
| "loss": 6.1995, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.255717640638856e-06, | |
| "loss": 6.2068, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.245973046452481e-06, | |
| "loss": 6.2036, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "eval_accuracy": 0.14343872779894534, | |
| "eval_loss": 6.196343421936035, | |
| "eval_runtime": 1118.4907, | |
| "eval_samples_per_second": 275.689, | |
| "eval_steps_per_second": 4.924, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.236228452266106e-06, | |
| "loss": 6.1962, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.22648385807973e-06, | |
| "loss": 6.1984, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.216739263893356e-06, | |
| "loss": 6.1897, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.20699466970698e-06, | |
| "loss": 6.2037, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.197250075520606e-06, | |
| "loss": 6.1947, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.18750548133423e-06, | |
| "loss": 6.1908, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.177760887147855e-06, | |
| "loss": 6.1916, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.16801629296148e-06, | |
| "loss": 6.1831, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.158271698775105e-06, | |
| "loss": 6.1926, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.14852710458873e-06, | |
| "loss": 6.1817, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.138782510402355e-06, | |
| "loss": 6.1883, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.12903791621598e-06, | |
| "loss": 6.1804, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.119293322029605e-06, | |
| "loss": 6.19, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.10954872784323e-06, | |
| "loss": 6.1799, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.099804133656855e-06, | |
| "loss": 6.186, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.09005953947048e-06, | |
| "loss": 6.182, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.080314945284105e-06, | |
| "loss": 6.179, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.07057035109773e-06, | |
| "loss": 6.1745, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.060825756911355e-06, | |
| "loss": 6.1828, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.051081162724978e-06, | |
| "loss": 6.1806, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "eval_accuracy": 0.145112327113654, | |
| "eval_loss": 6.177550315856934, | |
| "eval_runtime": 1123.3425, | |
| "eval_samples_per_second": 274.499, | |
| "eval_steps_per_second": 4.902, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.041336568538603e-06, | |
| "loss": 6.1756, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.031591974352228e-06, | |
| "loss": 6.1708, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.021847380165853e-06, | |
| "loss": 6.178, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.01210278597948e-06, | |
| "loss": 6.1813, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.002358191793104e-06, | |
| "loss": 6.1757, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.99261359760673e-06, | |
| "loss": 6.1794, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.982869003420354e-06, | |
| "loss": 6.1678, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.973124409233977e-06, | |
| "loss": 6.1702, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.963379815047602e-06, | |
| "loss": 6.1749, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.953635220861227e-06, | |
| "loss": 6.1718, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.943890626674852e-06, | |
| "loss": 6.173, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.934146032488477e-06, | |
| "loss": 6.1734, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.924401438302102e-06, | |
| "loss": 6.1723, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.914656844115727e-06, | |
| "loss": 6.167, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.904912249929352e-06, | |
| "loss": 6.1694, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.895167655742977e-06, | |
| "loss": 6.1644, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.885423061556602e-06, | |
| "loss": 6.1727, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.875678467370227e-06, | |
| "loss": 6.1642, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.865933873183852e-06, | |
| "loss": 6.171, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.856189278997477e-06, | |
| "loss": 6.1591, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "eval_accuracy": 0.14557786013315643, | |
| "eval_loss": 6.162111759185791, | |
| "eval_runtime": 1121.0921, | |
| "eval_samples_per_second": 275.05, | |
| "eval_steps_per_second": 4.912, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.846444684811102e-06, | |
| "loss": 6.1621, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.836700090624727e-06, | |
| "loss": 6.1606, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.826955496438352e-06, | |
| "loss": 6.1624, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.817210902251976e-06, | |
| "loss": 6.156, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.807466308065601e-06, | |
| "loss": 6.1534, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.797721713879226e-06, | |
| "loss": 6.1578, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.787977119692851e-06, | |
| "loss": 6.1595, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.778232525506476e-06, | |
| "loss": 6.1571, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.768487931320101e-06, | |
| "loss": 6.1604, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.758743337133726e-06, | |
| "loss": 6.1552, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.74899874294735e-06, | |
| "loss": 6.1523, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.739254148760976e-06, | |
| "loss": 6.1566, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.729509554574601e-06, | |
| "loss": 6.1547, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.719764960388226e-06, | |
| "loss": 6.1543, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.71002036620185e-06, | |
| "loss": 6.1485, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.700275772015476e-06, | |
| "loss": 6.1511, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.6905311778291e-06, | |
| "loss": 6.1583, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.680786583642726e-06, | |
| "loss": 6.1506, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.671041989456349e-06, | |
| "loss": 6.1529, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.661297395269974e-06, | |
| "loss": 6.1503, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "eval_accuracy": 0.14678527431167004, | |
| "eval_loss": 6.147279262542725, | |
| "eval_runtime": 1119.734, | |
| "eval_samples_per_second": 275.383, | |
| "eval_steps_per_second": 4.918, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.651552801083599e-06, | |
| "loss": 6.151, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.641808206897224e-06, | |
| "loss": 6.152, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 7.632063612710849e-06, | |
| "loss": 6.1463, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 7.622319018524474e-06, | |
| "loss": 6.1449, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.612574424338099e-06, | |
| "loss": 6.1485, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.602829830151724e-06, | |
| "loss": 6.1336, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.593085235965348e-06, | |
| "loss": 6.1408, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.583340641778973e-06, | |
| "loss": 6.143, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.573596047592598e-06, | |
| "loss": 6.1486, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.563851453406223e-06, | |
| "loss": 6.1471, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.554106859219849e-06, | |
| "loss": 6.1414, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.544362265033474e-06, | |
| "loss": 6.1425, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.534617670847099e-06, | |
| "loss": 6.1412, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.524873076660723e-06, | |
| "loss": 6.1485, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.515128482474348e-06, | |
| "loss": 6.1361, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 7.505383888287973e-06, | |
| "loss": 6.1375, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 7.495639294101598e-06, | |
| "loss": 6.1327, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 7.485894699915223e-06, | |
| "loss": 6.13, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 7.4761501057288475e-06, | |
| "loss": 6.1369, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 7.4664055115424725e-06, | |
| "loss": 6.1391, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "eval_accuracy": 0.14658729459484762, | |
| "eval_loss": 6.135674953460693, | |
| "eval_runtime": 1118.1964, | |
| "eval_samples_per_second": 275.762, | |
| "eval_steps_per_second": 4.925, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 7.456660917356097e-06, | |
| "loss": 6.139, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 7.4469163231697215e-06, | |
| "loss": 6.1336, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 7.4371717289833464e-06, | |
| "loss": 6.1348, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 7.427427134796971e-06, | |
| "loss": 6.1345, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 7.417682540610597e-06, | |
| "loss": 6.1227, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 7.407937946424222e-06, | |
| "loss": 6.1309, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 7.398193352237847e-06, | |
| "loss": 6.1362, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 7.388448758051472e-06, | |
| "loss": 6.1308, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 7.378704163865097e-06, | |
| "loss": 6.1304, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 7.368959569678721e-06, | |
| "loss": 6.1361, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 7.359214975492346e-06, | |
| "loss": 6.134, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 7.349470381305971e-06, | |
| "loss": 6.1287, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 7.339725787119596e-06, | |
| "loss": 6.1296, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 7.329981192933221e-06, | |
| "loss": 6.1284, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 7.320236598746846e-06, | |
| "loss": 6.1313, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 7.310492004560471e-06, | |
| "loss": 6.1287, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 7.300747410374096e-06, | |
| "loss": 6.1276, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 7.29100281618772e-06, | |
| "loss": 6.1228, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.2812582220013454e-06, | |
| "loss": 6.13, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.27151362781497e-06, | |
| "loss": 6.126, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "eval_accuracy": 0.14766462524448853, | |
| "eval_loss": 6.123033046722412, | |
| "eval_runtime": 1119.531, | |
| "eval_samples_per_second": 275.433, | |
| "eval_steps_per_second": 4.919, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 7.261769033628595e-06, | |
| "loss": 6.1275, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 7.25202443944222e-06, | |
| "loss": 6.1248, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 7.242279845255845e-06, | |
| "loss": 6.1267, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 7.23253525106947e-06, | |
| "loss": 6.1238, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 7.222790656883095e-06, | |
| "loss": 6.1204, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 7.213046062696719e-06, | |
| "loss": 6.1209, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 7.203301468510344e-06, | |
| "loss": 6.1138, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 7.193556874323969e-06, | |
| "loss": 6.1214, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.183812280137594e-06, | |
| "loss": 6.1211, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.174067685951219e-06, | |
| "loss": 6.1188, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 7.164323091764845e-06, | |
| "loss": 6.1129, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 7.15457849757847e-06, | |
| "loss": 6.1207, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 7.1448339033920945e-06, | |
| "loss": 6.1185, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 7.135089309205719e-06, | |
| "loss": 6.1152, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.1253447150193436e-06, | |
| "loss": 6.1228, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.1156001208329685e-06, | |
| "loss": 6.1177, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.1058555266465934e-06, | |
| "loss": 6.1197, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.096110932460218e-06, | |
| "loss": 6.1096, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.086366338273843e-06, | |
| "loss": 6.1207, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.076621744087468e-06, | |
| "loss": 6.1145, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "eval_accuracy": 0.14792931680665203, | |
| "eval_loss": 6.113304138183594, | |
| "eval_runtime": 1126.1359, | |
| "eval_samples_per_second": 273.818, | |
| "eval_steps_per_second": 4.89, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.066877149901092e-06, | |
| "loss": 6.1111, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 7.057132555714717e-06, | |
| "loss": 6.1148, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 7.047387961528342e-06, | |
| "loss": 6.11, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.037643367341967e-06, | |
| "loss": 6.1149, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.027898773155593e-06, | |
| "loss": 6.1134, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 7.018154178969218e-06, | |
| "loss": 6.1138, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 7.008409584782843e-06, | |
| "loss": 6.1112, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.998664990596468e-06, | |
| "loss": 6.1089, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.988920396410092e-06, | |
| "loss": 6.1085, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 6.979175802223717e-06, | |
| "loss": 6.1106, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 6.969431208037342e-06, | |
| "loss": 6.1096, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.959686613850967e-06, | |
| "loss": 6.1108, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.949942019664592e-06, | |
| "loss": 6.1035, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 6.9401974254782165e-06, | |
| "loss": 6.1103, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 6.9304528312918415e-06, | |
| "loss": 6.1091, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 6.920708237105466e-06, | |
| "loss": 6.1116, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 6.9109636429190905e-06, | |
| "loss": 6.1144, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 6.901219048732715e-06, | |
| "loss": 6.1117, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 6.891474454546341e-06, | |
| "loss": 6.1043, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 6.881729860359966e-06, | |
| "loss": 6.1067, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "eval_accuracy": 0.1486117878075688, | |
| "eval_loss": 6.104012966156006, | |
| "eval_runtime": 1131.1157, | |
| "eval_samples_per_second": 272.612, | |
| "eval_steps_per_second": 4.869, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.871985266173591e-06, | |
| "loss": 6.1069, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.862240671987216e-06, | |
| "loss": 6.1015, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 6.852496077800841e-06, | |
| "loss": 6.1043, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 6.842751483614466e-06, | |
| "loss": 6.1055, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 6.83300688942809e-06, | |
| "loss": 6.1012, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 6.823262295241715e-06, | |
| "loss": 6.1038, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 6.81351770105534e-06, | |
| "loss": 6.1, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 6.803773106868965e-06, | |
| "loss": 6.1, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 6.79402851268259e-06, | |
| "loss": 6.1014, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 6.784283918496215e-06, | |
| "loss": 6.0982, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 6.7745393243098404e-06, | |
| "loss": 6.0992, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 6.764794730123465e-06, | |
| "loss": 6.0965, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 6.755050135937089e-06, | |
| "loss": 6.0998, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 6.745305541750714e-06, | |
| "loss": 6.0991, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 6.735560947564339e-06, | |
| "loss": 6.1015, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 6.725816353377964e-06, | |
| "loss": 6.0967, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 6.716071759191589e-06, | |
| "loss": 6.099, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 6.706327165005214e-06, | |
| "loss": 6.0939, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 6.696582570818839e-06, | |
| "loss": 6.0957, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 6.686837976632463e-06, | |
| "loss": 6.097, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "eval_accuracy": 0.14878856544457936, | |
| "eval_loss": 6.096564292907715, | |
| "eval_runtime": 1119.0649, | |
| "eval_samples_per_second": 275.548, | |
| "eval_steps_per_second": 4.921, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 6.677093382446088e-06, | |
| "loss": 6.093, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 6.667348788259713e-06, | |
| "loss": 6.0854, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 6.657604194073338e-06, | |
| "loss": 6.0906, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 6.647859599886963e-06, | |
| "loss": 6.0935, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 6.638115005700589e-06, | |
| "loss": 6.0892, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 6.628370411514214e-06, | |
| "loss": 6.0973, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 6.618625817327839e-06, | |
| "loss": 6.0858, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 6.608881223141463e-06, | |
| "loss": 6.0894, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 6.599136628955088e-06, | |
| "loss": 6.0949, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.5893920347687125e-06, | |
| "loss": 6.095, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.5796474405823375e-06, | |
| "loss": 6.0902, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.569902846395962e-06, | |
| "loss": 6.1008, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.560158252209587e-06, | |
| "loss": 6.092, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 6.550413658023212e-06, | |
| "loss": 6.0939, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 6.540669063836837e-06, | |
| "loss": 6.091, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 6.530924469650461e-06, | |
| "loss": 6.0882, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 6.521179875464086e-06, | |
| "loss": 6.0901, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 6.511435281277711e-06, | |
| "loss": 6.0976, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 6.501690687091336e-06, | |
| "loss": 6.0901, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.491946092904962e-06, | |
| "loss": 6.0825, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "eval_accuracy": 0.1491677593736343, | |
| "eval_loss": 6.0875244140625, | |
| "eval_runtime": 1122.9447, | |
| "eval_samples_per_second": 274.596, | |
| "eval_steps_per_second": 4.904, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.482201498718587e-06, | |
| "loss": 6.0913, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.472456904532212e-06, | |
| "loss": 6.0924, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.462712310345837e-06, | |
| "loss": 6.0843, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.452967716159461e-06, | |
| "loss": 6.0833, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.443223121973086e-06, | |
| "loss": 6.0886, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.433478527786711e-06, | |
| "loss": 6.0807, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.423733933600336e-06, | |
| "loss": 6.0886, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.4139893394139606e-06, | |
| "loss": 6.0811, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 6.4042447452275855e-06, | |
| "loss": 6.0872, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 6.3945001510412104e-06, | |
| "loss": 6.0853, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 6.384755556854836e-06, | |
| "loss": 6.087, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 6.3750109626684595e-06, | |
| "loss": 6.0862, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.365266368482084e-06, | |
| "loss": 6.0826, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.35552177429571e-06, | |
| "loss": 6.0849, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.345777180109335e-06, | |
| "loss": 6.0781, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.33603258592296e-06, | |
| "loss": 6.0844, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.326287991736585e-06, | |
| "loss": 6.0899, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.31654339755021e-06, | |
| "loss": 6.0802, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.306798803363834e-06, | |
| "loss": 6.0838, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.297054209177459e-06, | |
| "loss": 6.0783, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "eval_accuracy": 0.1494112558072836, | |
| "eval_loss": 6.079698085784912, | |
| "eval_runtime": 1121.774, | |
| "eval_samples_per_second": 274.882, | |
| "eval_steps_per_second": 4.909, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 6.287309614991084e-06, | |
| "loss": 6.0824, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 6.277565020804709e-06, | |
| "loss": 6.0704, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.267820426618334e-06, | |
| "loss": 6.0755, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.258075832431959e-06, | |
| "loss": 6.0833, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 6.2483312382455845e-06, | |
| "loss": 6.0793, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 6.238586644059209e-06, | |
| "loss": 6.0739, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 6.228842049872833e-06, | |
| "loss": 6.0775, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 6.2190974556864584e-06, | |
| "loss": 6.079, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 6.209352861500083e-06, | |
| "loss": 6.0721, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.199608267313708e-06, | |
| "loss": 6.0753, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.189863673127333e-06, | |
| "loss": 6.0752, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 6.180119078940958e-06, | |
| "loss": 6.0751, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 6.170374484754583e-06, | |
| "loss": 6.0729, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.160629890568208e-06, | |
| "loss": 6.0765, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.150885296381832e-06, | |
| "loss": 6.0759, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 6.141140702195457e-06, | |
| "loss": 6.0766, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 6.131396108009082e-06, | |
| "loss": 6.0795, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.121651513822707e-06, | |
| "loss": 6.0707, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.111906919636332e-06, | |
| "loss": 6.0719, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.102162325449958e-06, | |
| "loss": 6.0673, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "eval_accuracy": 0.14990068299638576, | |
| "eval_loss": 6.073047637939453, | |
| "eval_runtime": 1126.8386, | |
| "eval_samples_per_second": 273.647, | |
| "eval_steps_per_second": 4.887, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.092417731263583e-06, | |
| "loss": 6.0732, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.0826731370772076e-06, | |
| "loss": 6.0709, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.072928542890832e-06, | |
| "loss": 6.0785, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.063183948704457e-06, | |
| "loss": 6.0696, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.0534393545180815e-06, | |
| "loss": 6.0682, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.0436947603317065e-06, | |
| "loss": 6.0705, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.033950166145331e-06, | |
| "loss": 6.071, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.024205571958956e-06, | |
| "loss": 6.0672, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.014460977772581e-06, | |
| "loss": 6.073, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.004716383586206e-06, | |
| "loss": 6.0711, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.99497178939983e-06, | |
| "loss": 6.0709, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.985227195213455e-06, | |
| "loss": 6.0696, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.97548260102708e-06, | |
| "loss": 6.0663, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.965738006840706e-06, | |
| "loss": 6.0715, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.955993412654331e-06, | |
| "loss": 6.069, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.946248818467956e-06, | |
| "loss": 6.0692, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 5.936504224281581e-06, | |
| "loss": 6.0679, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 5.926759630095206e-06, | |
| "loss": 6.0674, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.91701503590883e-06, | |
| "loss": 6.0657, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.907270441722455e-06, | |
| "loss": 6.066, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "eval_accuracy": 0.15014050910513674, | |
| "eval_loss": 6.062347412109375, | |
| "eval_runtime": 1123.1639, | |
| "eval_samples_per_second": 274.542, | |
| "eval_steps_per_second": 4.903, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.89752584753608e-06, | |
| "loss": 6.0662, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.887781253349705e-06, | |
| "loss": 6.0633, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.8780366591633295e-06, | |
| "loss": 6.0601, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.8682920649769545e-06, | |
| "loss": 6.0644, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.85854747079058e-06, | |
| "loss": 6.0642, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.8488028766042035e-06, | |
| "loss": 6.0652, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.8390582824178284e-06, | |
| "loss": 6.066, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.829313688231454e-06, | |
| "loss": 6.0568, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.819569094045079e-06, | |
| "loss": 6.0566, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.809824499858704e-06, | |
| "loss": 6.0601, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.800079905672329e-06, | |
| "loss": 6.0632, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.790335311485954e-06, | |
| "loss": 6.0611, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.780590717299579e-06, | |
| "loss": 6.0593, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.770846123113203e-06, | |
| "loss": 6.054, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.761101528926828e-06, | |
| "loss": 6.0623, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.751356934740453e-06, | |
| "loss": 6.0555, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.741612340554078e-06, | |
| "loss": 6.0583, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.731867746367703e-06, | |
| "loss": 6.0551, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.722123152181328e-06, | |
| "loss": 6.0544, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.7123785579949535e-06, | |
| "loss": 6.0534, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "eval_accuracy": 0.15038287506981798, | |
| "eval_loss": 6.050975799560547, | |
| "eval_runtime": 1125.1787, | |
| "eval_samples_per_second": 274.051, | |
| "eval_steps_per_second": 4.894, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.702633963808578e-06, | |
| "loss": 6.0522, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5.6928893696222025e-06, | |
| "loss": 6.0537, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5.683144775435827e-06, | |
| "loss": 6.0455, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5.673400181249452e-06, | |
| "loss": 6.0462, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5.663655587063077e-06, | |
| "loss": 6.0416, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5.653910992876702e-06, | |
| "loss": 6.0387, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5.644166398690327e-06, | |
| "loss": 6.0381, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.634421804503952e-06, | |
| "loss": 6.033, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.624677210317577e-06, | |
| "loss": 6.0337, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5.614932616131201e-06, | |
| "loss": 6.032, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5.605188021944826e-06, | |
| "loss": 6.0216, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 5.595443427758451e-06, | |
| "loss": 6.0261, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 5.585698833572076e-06, | |
| "loss": 6.0216, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.575954239385702e-06, | |
| "loss": 6.0228, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.566209645199327e-06, | |
| "loss": 6.0163, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.556465051012952e-06, | |
| "loss": 6.016, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5.5467204568265765e-06, | |
| "loss": 6.0117, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5.536975862640201e-06, | |
| "loss": 6.0139, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5.5272312684538256e-06, | |
| "loss": 6.008, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5.5174866742674505e-06, | |
| "loss": 6.0004, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "eval_accuracy": 0.15174394220057463, | |
| "eval_loss": 5.997183799743652, | |
| "eval_runtime": 1122.2227, | |
| "eval_samples_per_second": 274.773, | |
| "eval_steps_per_second": 4.907, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.5077420800810754e-06, | |
| "loss": 6.0039, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.4979974858947e-06, | |
| "loss": 6.0028, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.488252891708325e-06, | |
| "loss": 6.001, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.47850829752195e-06, | |
| "loss": 5.9965, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5.468763703335574e-06, | |
| "loss": 5.9928, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5.459019109149199e-06, | |
| "loss": 5.9885, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5.449274514962824e-06, | |
| "loss": 5.9842, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5.43952992077645e-06, | |
| "loss": 5.9864, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5.429785326590075e-06, | |
| "loss": 5.9823, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5.4200407324037e-06, | |
| "loss": 5.9829, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5.410296138217325e-06, | |
| "loss": 5.9805, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5.40055154403095e-06, | |
| "loss": 5.982, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 5.390806949844574e-06, | |
| "loss": 5.97, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 5.381062355658199e-06, | |
| "loss": 5.9672, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5.371317761471824e-06, | |
| "loss": 5.9695, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5.361573167285449e-06, | |
| "loss": 5.9622, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5.3518285730990736e-06, | |
| "loss": 5.9708, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5.3420839789126985e-06, | |
| "loss": 5.9601, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5.3323393847263234e-06, | |
| "loss": 5.9565, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5.322594790539949e-06, | |
| "loss": 5.9609, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "eval_accuracy": 0.1529847674086314, | |
| "eval_loss": 5.949230670928955, | |
| "eval_runtime": 1122.3302, | |
| "eval_samples_per_second": 274.746, | |
| "eval_steps_per_second": 4.907, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5.3128501963535725e-06, | |
| "loss": 5.9569, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5.303105602167197e-06, | |
| "loss": 5.9549, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5.293361007980823e-06, | |
| "loss": 5.96, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5.283616413794448e-06, | |
| "loss": 5.9539, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5.273871819608073e-06, | |
| "loss": 5.954, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5.264127225421698e-06, | |
| "loss": 5.9519, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5.254382631235323e-06, | |
| "loss": 5.9491, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5.244638037048948e-06, | |
| "loss": 5.948, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5.234893442862572e-06, | |
| "loss": 5.9525, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5.225148848676197e-06, | |
| "loss": 5.9467, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5.215404254489822e-06, | |
| "loss": 5.9413, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5.205659660303447e-06, | |
| "loss": 5.9383, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5.195915066117072e-06, | |
| "loss": 5.9422, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 5.1861704719306975e-06, | |
| "loss": 5.9398, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 5.1764258777443224e-06, | |
| "loss": 5.9362, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 5.166681283557947e-06, | |
| "loss": 5.9386, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 5.1569366893715715e-06, | |
| "loss": 5.9373, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 5.147192095185196e-06, | |
| "loss": 5.934, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 5.137447500998821e-06, | |
| "loss": 5.9277, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.127702906812446e-06, | |
| "loss": 5.93, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "eval_accuracy": 0.15509840986445972, | |
| "eval_loss": 5.9169206619262695, | |
| "eval_runtime": 1121.667, | |
| "eval_samples_per_second": 274.909, | |
| "eval_steps_per_second": 4.91, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.117958312626071e-06, | |
| "loss": 5.9241, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.108213718439696e-06, | |
| "loss": 5.9302, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 5.098469124253321e-06, | |
| "loss": 5.9287, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 5.088724530066946e-06, | |
| "loss": 5.9299, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 5.07897993588057e-06, | |
| "loss": 5.9347, | |
| "step": 262500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 5.069235341694195e-06, | |
| "loss": 5.9261, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 5.05949074750782e-06, | |
| "loss": 5.9214, | |
| "step": 263500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 5.049746153321446e-06, | |
| "loss": 5.9228, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 5.040001559135071e-06, | |
| "loss": 5.9211, | |
| "step": 264500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 5.030256964948696e-06, | |
| "loss": 5.9167, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 5.0205123707623206e-06, | |
| "loss": 5.9101, | |
| "step": 265500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 5.010767776575945e-06, | |
| "loss": 5.9161, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 5.00102318238957e-06, | |
| "loss": 5.9155, | |
| "step": 266500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.9912785882031945e-06, | |
| "loss": 5.9161, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.9815339940168195e-06, | |
| "loss": 5.9094, | |
| "step": 267500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.971789399830444e-06, | |
| "loss": 5.9059, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.962044805644069e-06, | |
| "loss": 5.9083, | |
| "step": 268500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.952300211457694e-06, | |
| "loss": 5.9098, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.942555617271319e-06, | |
| "loss": 5.9074, | |
| "step": 269500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.932811023084944e-06, | |
| "loss": 5.9058, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "eval_accuracy": 0.15708356115844138, | |
| "eval_loss": 5.889527797698975, | |
| "eval_runtime": 1123.8107, | |
| "eval_samples_per_second": 274.384, | |
| "eval_steps_per_second": 4.9, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.923066428898569e-06, | |
| "loss": 5.9026, | |
| "step": 270500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.913321834712193e-06, | |
| "loss": 5.9064, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.903577240525819e-06, | |
| "loss": 5.9012, | |
| "step": 271500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.893832646339444e-06, | |
| "loss": 5.8949, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.884088052153069e-06, | |
| "loss": 5.8981, | |
| "step": 272500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.874343457966693e-06, | |
| "loss": 5.8959, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.864598863780318e-06, | |
| "loss": 5.8925, | |
| "step": 273500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.854854269593944e-06, | |
| "loss": 5.8929, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.845109675407569e-06, | |
| "loss": 5.8929, | |
| "step": 274500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.835365081221193e-06, | |
| "loss": 5.8911, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.825620487034818e-06, | |
| "loss": 5.8869, | |
| "step": 275500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.8158758928484425e-06, | |
| "loss": 5.8904, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.8061312986620675e-06, | |
| "loss": 5.889, | |
| "step": 276500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.796386704475692e-06, | |
| "loss": 5.8829, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.786642110289317e-06, | |
| "loss": 5.8918, | |
| "step": 277500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.776897516102942e-06, | |
| "loss": 5.8869, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.767152921916567e-06, | |
| "loss": 5.8838, | |
| "step": 278500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.757408327730192e-06, | |
| "loss": 5.8841, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.747663733543817e-06, | |
| "loss": 5.8807, | |
| "step": 279500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.737919139357442e-06, | |
| "loss": 5.8834, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "eval_accuracy": 0.1596531986258025, | |
| "eval_loss": 5.861761093139648, | |
| "eval_runtime": 1125.4567, | |
| "eval_samples_per_second": 273.983, | |
| "eval_steps_per_second": 4.893, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.728174545171067e-06, | |
| "loss": 5.8803, | |
| "step": 280500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.718429950984691e-06, | |
| "loss": 5.8787, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.708685356798317e-06, | |
| "loss": 5.8748, | |
| "step": 281500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.698940762611942e-06, | |
| "loss": 5.8728, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.689196168425567e-06, | |
| "loss": 5.8797, | |
| "step": 282500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.679451574239191e-06, | |
| "loss": 5.8743, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.669706980052816e-06, | |
| "loss": 5.8715, | |
| "step": 283500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.6599623858664415e-06, | |
| "loss": 5.8729, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.650217791680066e-06, | |
| "loss": 5.8769, | |
| "step": 284500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.6404731974936906e-06, | |
| "loss": 5.876, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.6307286033073155e-06, | |
| "loss": 5.8713, | |
| "step": 285500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.6209840091209404e-06, | |
| "loss": 5.8694, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.611239414934565e-06, | |
| "loss": 5.8645, | |
| "step": 286500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.60149482074819e-06, | |
| "loss": 5.8687, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.591750226561815e-06, | |
| "loss": 5.8613, | |
| "step": 287500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.58200563237544e-06, | |
| "loss": 5.8592, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.572261038189065e-06, | |
| "loss": 5.8613, | |
| "step": 288500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.56251644400269e-06, | |
| "loss": 5.8596, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.552771849816315e-06, | |
| "loss": 5.8587, | |
| "step": 289500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.54302725562994e-06, | |
| "loss": 5.8572, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "eval_accuracy": 0.1622768898618876, | |
| "eval_loss": 5.839352130889893, | |
| "eval_runtime": 1122.0351, | |
| "eval_samples_per_second": 274.819, | |
| "eval_steps_per_second": 4.908, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.533282661443564e-06, | |
| "loss": 5.861, | |
| "step": 290500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.523538067257189e-06, | |
| "loss": 5.8567, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.513793473070815e-06, | |
| "loss": 5.8518, | |
| "step": 291500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.50404887888444e-06, | |
| "loss": 5.849, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.494304284698064e-06, | |
| "loss": 5.8498, | |
| "step": 292500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.484559690511689e-06, | |
| "loss": 5.8525, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.474815096325314e-06, | |
| "loss": 5.8518, | |
| "step": 293500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.465070502138939e-06, | |
| "loss": 5.8466, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.4553259079525635e-06, | |
| "loss": 5.8489, | |
| "step": 294500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.4455813137661884e-06, | |
| "loss": 5.8446, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.435836719579813e-06, | |
| "loss": 5.8467, | |
| "step": 295500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.426092125393438e-06, | |
| "loss": 5.8446, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.416347531207063e-06, | |
| "loss": 5.8469, | |
| "step": 296500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.406602937020688e-06, | |
| "loss": 5.8382, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.396858342834313e-06, | |
| "loss": 5.8398, | |
| "step": 297500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.387113748647938e-06, | |
| "loss": 5.8417, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.377369154461563e-06, | |
| "loss": 5.8355, | |
| "step": 298500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.367624560275188e-06, | |
| "loss": 5.837, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.357879966088813e-06, | |
| "loss": 5.8373, | |
| "step": 299500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.348135371902438e-06, | |
| "loss": 5.8296, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "eval_accuracy": 0.16607837941755138, | |
| "eval_loss": 5.816832065582275, | |
| "eval_runtime": 1119.4995, | |
| "eval_samples_per_second": 275.441, | |
| "eval_steps_per_second": 4.919, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.338390777716062e-06, | |
| "loss": 5.8373, | |
| "step": 300500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.328646183529687e-06, | |
| "loss": 5.8259, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.318901589343313e-06, | |
| "loss": 5.8351, | |
| "step": 301500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.3091569951569376e-06, | |
| "loss": 5.8327, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.299412400970562e-06, | |
| "loss": 5.8277, | |
| "step": 302500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.289667806784187e-06, | |
| "loss": 5.8228, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.2799232125978115e-06, | |
| "loss": 5.8273, | |
| "step": 303500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.270178618411437e-06, | |
| "loss": 5.821, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.260434024225061e-06, | |
| "loss": 5.8269, | |
| "step": 304500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.250689430038686e-06, | |
| "loss": 5.8243, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.240944835852311e-06, | |
| "loss": 5.8233, | |
| "step": 305500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.231200241665936e-06, | |
| "loss": 5.8219, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.221455647479561e-06, | |
| "loss": 5.8268, | |
| "step": 306500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.211711053293186e-06, | |
| "loss": 5.8215, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.201966459106811e-06, | |
| "loss": 5.8189, | |
| "step": 307500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.192221864920435e-06, | |
| "loss": 5.8176, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.182477270734061e-06, | |
| "loss": 5.8206, | |
| "step": 308500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.172732676547686e-06, | |
| "loss": 5.8167, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.162988082361311e-06, | |
| "loss": 5.8195, | |
| "step": 309500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.153243488174935e-06, | |
| "loss": 5.8085, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "eval_accuracy": 0.17028402442503204, | |
| "eval_loss": 5.792611598968506, | |
| "eval_runtime": 1121.7676, | |
| "eval_samples_per_second": 274.884, | |
| "eval_steps_per_second": 4.909, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.14349889398856e-06, | |
| "loss": 5.8129, | |
| "step": 310500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.133754299802185e-06, | |
| "loss": 5.8112, | |
| "step": 311000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.1240097056158105e-06, | |
| "loss": 5.81, | |
| "step": 311500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.114265111429435e-06, | |
| "loss": 5.8106, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.1045205172430595e-06, | |
| "loss": 5.8094, | |
| "step": 312500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.0947759230566845e-06, | |
| "loss": 5.8068, | |
| "step": 313000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.085031328870309e-06, | |
| "loss": 5.8044, | |
| "step": 313500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.075286734683934e-06, | |
| "loss": 5.8048, | |
| "step": 314000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.065542140497559e-06, | |
| "loss": 5.8025, | |
| "step": 314500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.055797546311184e-06, | |
| "loss": 5.8, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.046052952124809e-06, | |
| "loss": 5.8019, | |
| "step": 315500 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.036308357938434e-06, | |
| "loss": 5.8017, | |
| "step": 316000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.026563763752059e-06, | |
| "loss": 5.7962, | |
| "step": 316500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.016819169565684e-06, | |
| "loss": 5.7985, | |
| "step": 317000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.007074575379309e-06, | |
| "loss": 5.7987, | |
| "step": 317500 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.997329981192933e-06, | |
| "loss": 5.7956, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.987585387006559e-06, | |
| "loss": 5.7915, | |
| "step": 318500 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.977840792820184e-06, | |
| "loss": 5.7986, | |
| "step": 319000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.968096198633809e-06, | |
| "loss": 5.7951, | |
| "step": 319500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.958351604447433e-06, | |
| "loss": 5.7873, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "eval_accuracy": 0.1738716824007192, | |
| "eval_loss": 5.766289234161377, | |
| "eval_runtime": 1123.9464, | |
| "eval_samples_per_second": 274.351, | |
| "eval_steps_per_second": 4.9, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "step": 320001, | |
| "total_flos": 1.9643483518701404e+18, | |
| "train_loss": 6.214869752465887, | |
| "train_runtime": 199028.524, | |
| "train_samples_per_second": 147.184, | |
| "train_steps_per_second": 2.628 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 523105, | |
| "num_train_epochs": 5, | |
| "save_steps": 10000, | |
| "total_flos": 1.9643483518701404e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |