| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.038632, | |
| "global_step": 650000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 5.2706, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 3.8416, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6e-06, | |
| "loss": 3.0552, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 3.013, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1e-05, | |
| "loss": 2.8291, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.6151, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 2.5429, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.4003, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.8e-05, | |
| "loss": 2.1582, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-05, | |
| "loss": 2.1312, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 2.1117, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.9489, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 1.9571, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 1.9172, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3e-05, | |
| "loss": 1.9242, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 1.8219, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 1.7352, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6e-05, | |
| "loss": 1.7633, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8e-05, | |
| "loss": 1.757, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4e-05, | |
| "loss": 1.5583, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2e-05, | |
| "loss": 1.6072, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 1.6661, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 1.5409, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8e-05, | |
| "loss": 1.6312, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5866, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 1.2261, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 1.0432, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 1.1146, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.8e-05, | |
| "loss": 1.1212, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6e-05, | |
| "loss": 1.1243, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.2e-05, | |
| "loss": 0.938, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 0.8937, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.6e-05, | |
| "loss": 0.9787, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 0.9757, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7e-05, | |
| "loss": 1.0489, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.2e-05, | |
| "loss": 0.953, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.4e-05, | |
| "loss": 0.8359, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.6e-05, | |
| "loss": 0.7929, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.800000000000001e-05, | |
| "loss": 0.9099, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8e-05, | |
| "loss": 0.9498, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.2e-05, | |
| "loss": 0.9639, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.4e-05, | |
| "loss": 0.816, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.6e-05, | |
| "loss": 0.789, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 0.8444, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9e-05, | |
| "loss": 0.8851, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 0.9563, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.4e-05, | |
| "loss": 0.8531, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.6e-05, | |
| "loss": 0.7637, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.8e-05, | |
| "loss": 0.7266, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8467, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_loss": 0.5956527590751648, | |
| "eval_runtime": 27.6358, | |
| "eval_samples_per_second": 180.925, | |
| "eval_steps_per_second": 2.859, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.999972660400536e-05, | |
| "loss": 0.8917, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.999890641901125e-05, | |
| "loss": 0.8964, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.999753945398704e-05, | |
| "loss": 0.7736, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.99956257238817e-05, | |
| "loss": 0.7534, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.7739, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.999015805811965e-05, | |
| "loss": 0.8433, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.998660418225645e-05, | |
| "loss": 0.9162, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.998250366089848e-05, | |
| "loss": 0.8019, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.997785653888835e-05, | |
| "loss": 0.7206, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.6863, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.996692270216947e-05, | |
| "loss": 0.7978, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.996063610703137e-05, | |
| "loss": 0.8457, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.995380315038119e-05, | |
| "loss": 0.8531, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.994642390694308e-05, | |
| "loss": 0.7452, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.7258, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.993002688846913e-05, | |
| "loss": 0.7228, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.992100929274846e-05, | |
| "loss": 0.8107, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.991144576886823e-05, | |
| "loss": 0.885, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.990133642141359e-05, | |
| "loss": 0.7697, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.6873, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.987948070396571e-05, | |
| "loss": 0.6637, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.986773457298311e-05, | |
| "loss": 0.7624, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.985544309644475e-05, | |
| "loss": 0.817, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.984260640876821e-05, | |
| "loss": 0.823, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.7292, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.981529796748134e-05, | |
| "loss": 0.7083, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.980082651251175e-05, | |
| "loss": 0.6811, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.97858104436822e-05, | |
| "loss": 0.797, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.977024992520602e-05, | |
| "loss": 0.8565, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.7605, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.973749622593534e-05, | |
| "loss": 0.6574, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.972030340333001e-05, | |
| "loss": 0.6495, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.970256684745258e-05, | |
| "loss": 0.7299, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.968428675226714e-05, | |
| "loss": 0.7921, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.8014, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.964609674954696e-05, | |
| "loss": 0.721, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.962618725965196e-05, | |
| "loss": 0.6983, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.96057350657239e-05, | |
| "loss": 0.6478, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.95847403914247e-05, | |
| "loss": 0.7873, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.8259, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.954112452602045e-05, | |
| "loss": 0.7583, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.95185038118915e-05, | |
| "loss": 0.6389, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.949534157133844e-05, | |
| "loss": 0.6414, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.94716380576598e-05, | |
| "loss": 0.7065, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.7707, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.942260825371358e-05, | |
| "loss": 0.7862, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.939728249962807e-05, | |
| "loss": 0.7196, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.937141654477528e-05, | |
| "loss": 0.6862, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.934501067202117e-05, | |
| "loss": 0.6338, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.7777, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_loss": 0.454129159450531, | |
| "eval_runtime": 19.9122, | |
| "eval_samples_per_second": 251.102, | |
| "eval_steps_per_second": 3.967, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.929058033379181e-05, | |
| "loss": 0.7962, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.926255646355804e-05, | |
| "loss": 0.7458, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.923399386589933e-05, | |
| "loss": 0.6392, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.92048928531717e-05, | |
| "loss": 0.631, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.6928, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.914507686137019e-05, | |
| "loss": 0.7553, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.911436253643445e-05, | |
| "loss": 0.7672, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.90831111046988e-05, | |
| "loss": 0.7193, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.905132290792394e-05, | |
| "loss": 0.6786, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.6201, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.89861376156452e-05, | |
| "loss": 0.7655, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.895274123299723e-05, | |
| "loss": 0.7712, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.891880951101407e-05, | |
| "loss": 0.7369, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.888434282076758e-05, | |
| "loss": 0.6439, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.6237, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.881380604901964e-05, | |
| "loss": 0.6758, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.877773673889701e-05, | |
| "loss": 0.7365, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.87411340032603e-05, | |
| "loss": 0.7579, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.870399824239117e-05, | |
| "loss": 0.7161, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.668, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.862812927522309e-05, | |
| "loss": 0.6153, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.858939689861506e-05, | |
| "loss": 0.7529, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.855013315614725e-05, | |
| "loss": 0.7428, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.851033847720166e-05, | |
| "loss": 0.7285, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.6505, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.842915805643155e-05, | |
| "loss": 0.6175, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.838777320238312e-05, | |
| "loss": 0.6625, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.834585918739936e-05, | |
| "loss": 0.721, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.830341646984521e-05, | |
| "loss": 0.749, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.7176, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.821694678938953e-05, | |
| "loss": 0.6573, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.817292077210659e-05, | |
| "loss": 0.6116, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.812836794348004e-05, | |
| "loss": 0.7463, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.808328879073251e-05, | |
| "loss": 0.7162, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.7043, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.799155349053851e-05, | |
| "loss": 0.7667, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.794489834629455e-05, | |
| "loss": 0.7286, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.789771888432375e-05, | |
| "loss": 0.5967, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.785001562057309e-05, | |
| "loss": 0.5752, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.7106, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.775303978015585e-05, | |
| "loss": 0.7391, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.77037682640015e-05, | |
| "loss": 0.768, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.765397506708023e-05, | |
| "loss": 0.6382, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.760366073392246e-05, | |
| "loss": 0.5991, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.6929, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.750147086550844e-05, | |
| "loss": 0.7047, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.744959644778422e-05, | |
| "loss": 0.7684, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.739720312887535e-05, | |
| "loss": 0.7285, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.734429148174675e-05, | |
| "loss": 0.5971, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.5639, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "eval_loss": 0.4348195195198059, | |
| "eval_runtime": 18.8713, | |
| "eval_samples_per_second": 264.953, | |
| "eval_steps_per_second": 4.186, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.723691552302562e-05, | |
| "loss": 0.7032, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.718245238567939e-05, | |
| "loss": 0.7287, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.712747326859315e-05, | |
| "loss": 0.7653, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.707197877300974e-05, | |
| "loss": 0.6367, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.5988, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.695944607949649e-05, | |
| "loss": 0.6748, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.690240911220618e-05, | |
| "loss": 0.696, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.684485922768422e-05, | |
| "loss": 0.7725, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.6786797055287e-05, | |
| "loss": 0.7177, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.5959, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.66691383922964e-05, | |
| "loss": 0.5597, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.660954318839933e-05, | |
| "loss": 0.6957, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.654943827000548e-05, | |
| "loss": 0.7231, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.648882429441257e-05, | |
| "loss": 0.7555, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.6363, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.636607182864827e-05, | |
| "loss": 0.5993, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.630393468087818e-05, | |
| "loss": 0.6546, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.624129116069694e-05, | |
| "loss": 0.6919, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.617814195316411e-05, | |
| "loss": 0.7686, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.7068, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.605032924392457e-05, | |
| "loss": 0.5922, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.598566713995718e-05, | |
| "loss": 0.5558, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.59205021441015e-05, | |
| "loss": 0.6863, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.58548349689915e-05, | |
| "loss": 0.7117, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.7473, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.572199695899522e-05, | |
| "loss": 0.6402, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.565482757680415e-05, | |
| "loss": 0.6006, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.558715892073323e-05, | |
| "loss": 0.638, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.551899173079607e-05, | |
| "loss": 0.6898, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.7611, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.538116473662861e-05, | |
| "loss": 0.7005, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.531150643965223e-05, | |
| "loss": 0.5858, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.524135262330098e-05, | |
| "loss": 0.5554, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.517070405476575e-05, | |
| "loss": 0.6768, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.7114, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.502792575695112e-05, | |
| "loss": 0.7331, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.49557975890723e-05, | |
| "loss": 0.6434, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.488317779179361e-05, | |
| "loss": 0.6053, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.481006715927351e-05, | |
| "loss": 0.6135, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.6929, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.46623765919727e-05, | |
| "loss": 0.7579, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.458779827231237e-05, | |
| "loss": 0.6972, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.451273234763371e-05, | |
| "loss": 0.5783, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.443717963884569e-05, | |
| "loss": 0.5633, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.6577, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.428461717918511e-05, | |
| "loss": 0.7068, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.420760909671118e-05, | |
| "loss": 0.7249, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.413011756690685e-05, | |
| "loss": 0.6484, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.405214343720707e-05, | |
| "loss": 0.6115, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.5929, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "eval_loss": 0.40294867753982544, | |
| "eval_runtime": 28.21, | |
| "eval_samples_per_second": 177.242, | |
| "eval_steps_per_second": 2.8, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.389475079423988e-05, | |
| "loss": 0.6945, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.381533400219318e-05, | |
| "loss": 0.7569, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.373543805267368e-05, | |
| "loss": 0.6929, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.365506381941066e-05, | |
| "loss": 0.5678, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.5675, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.349288402271388e-05, | |
| "loss": 0.6459, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.341108023285238e-05, | |
| "loss": 0.7009, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.332880170637252e-05, | |
| "loss": 0.7201, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.32460493430591e-05, | |
| "loss": 0.6509, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.6173, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.30791267309698e-05, | |
| "loss": 0.5708, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.299495830763286e-05, | |
| "loss": 0.6987, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.291031969832026e-05, | |
| "loss": 0.7496, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.282521182862629e-05, | |
| "loss": 0.6845, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.5681, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.265359203611987e-05, | |
| "loss": 0.5686, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.256708199011401e-05, | |
| "loss": 0.6305, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.248010643731935e-05, | |
| "loss": 0.6971, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.239266632888659e-05, | |
| "loss": 0.7083, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.6505, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.221639627510076e-05, | |
| "loss": 0.6241, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.212756825740873e-05, | |
| "loss": 0.5542, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.20382795393797e-05, | |
| "loss": 0.6974, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.194853109746074e-05, | |
| "loss": 0.7402, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.678, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.176765897286813e-05, | |
| "loss": 0.5751, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.167653726818305e-05, | |
| "loss": 0.5735, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.158495979556358e-05, | |
| "loss": 0.6172, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.14929275564863e-05, | |
| "loss": 0.6927, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.7046, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.130750280971978e-05, | |
| "loss": 0.654, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.121411232980588e-05, | |
| "loss": 0.6226, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.112027113896262e-05, | |
| "loss": 0.5528, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.102598026342222e-05, | |
| "loss": 0.7002, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.7249, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.083605358775612e-05, | |
| "loss": 0.676, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.074041986463808e-05, | |
| "loss": 0.5795, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.064434061081562e-05, | |
| "loss": 0.5726, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.0547816876996e-05, | |
| "loss": 0.6078, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.6875, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.035344019648702e-05, | |
| "loss": 0.6969, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.025558937546988e-05, | |
| "loss": 0.6565, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.015729832577681e-05, | |
| "loss": 0.6206, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.005856812230304e-05, | |
| "loss": 0.548, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.6972, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.98597945775948e-05, | |
| "loss": 0.704, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.975975341011596e-05, | |
| "loss": 0.6694, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.965927743634391e-05, | |
| "loss": 0.5906, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.955836775506776e-05, | |
| "loss": 0.5725, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.5976, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "eval_loss": 0.4060193598270416, | |
| "eval_runtime": 19.405, | |
| "eval_samples_per_second": 257.665, | |
| "eval_steps_per_second": 4.071, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.935525168886262e-05, | |
| "loss": 0.6768, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.92530475251784e-05, | |
| "loss": 0.6926, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.91504140964553e-05, | |
| "loss": 0.667, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.90473525250761e-05, | |
| "loss": 0.6088, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.5454, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.883994946727849e-05, | |
| "loss": 0.6887, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.873561024898668e-05, | |
| "loss": 0.6843, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.863084742426719e-05, | |
| "loss": 0.6654, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.852566213878947e-05, | |
| "loss": 0.6013, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.5689, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.831402879132446e-05, | |
| "loss": 0.5905, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.820758304372557e-05, | |
| "loss": 0.6676, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.810071946411989e-05, | |
| "loss": 0.6896, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.799343922115044e-05, | |
| "loss": 0.6735, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.6028, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.77776334424621e-05, | |
| "loss": 0.5534, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.766911026676064e-05, | |
| "loss": 0.678, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.756017514770443e-05, | |
| "loss": 0.6658, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.745082927659047e-05, | |
| "loss": 0.6495, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.7026, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.723091006582389e-05, | |
| "loss": 0.6788, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.71203391311725e-05, | |
| "loss": 0.5613, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.700936225443959e-05, | |
| "loss": 0.5202, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.689798064925049e-05, | |
| "loss": 0.6501, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.6806, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.6674008130122e-05, | |
| "loss": 0.7155, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.656141966551019e-05, | |
| "loss": 0.6027, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.644843137107059e-05, | |
| "loss": 0.5595, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.633504448242505e-05, | |
| "loss": 0.6221, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.6514, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.610707988678503e-05, | |
| "loss": 0.7072, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.599250467277483e-05, | |
| "loss": 0.6836, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.587753585050004e-05, | |
| "loss": 0.5585, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.576217467724128e-05, | |
| "loss": 0.5101, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.6471, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.553028032833397e-05, | |
| "loss": 0.671, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.541374968864487e-05, | |
| "loss": 0.7087, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.529683176986295e-05, | |
| "loss": 0.6022, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.517952785058385e-05, | |
| "loss": 0.5593, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.6112, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.494376714600878e-05, | |
| "loss": 0.639, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.482531293895412e-05, | |
| "loss": 0.7099, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.470647788785665e-05, | |
| "loss": 0.6897, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.458726329227747e-05, | |
| "loss": 0.5486, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.5073, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.434770068665723e-05, | |
| "loss": 0.6397, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.422735529643444e-05, | |
| "loss": 0.6651, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.410663560133784e-05, | |
| "loss": 0.7039, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.398554292153866e-05, | |
| "loss": 0.605, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.5598, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "eval_loss": 0.3902786374092102, | |
| "eval_runtime": 18.9457, | |
| "eval_samples_per_second": 263.912, | |
| "eval_steps_per_second": 4.17, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.37422439088976e-05, | |
| "loss": 0.5953, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.362004023673474e-05, | |
| "loss": 0.6344, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.349746890119826e-05, | |
| "loss": 0.7151, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.337453124270863e-05, | |
| "loss": 0.6878, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.5403, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.31275623385675e-05, | |
| "loss": 0.5117, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.300353379372834e-05, | |
| "loss": 0.6272, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.287914432753123e-05, | |
| "loss": 0.6585, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.275439530027948e-05, | |
| "loss": 0.6988, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.6127, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.250382402347065e-05, | |
| "loss": 0.561, | |
| "step": 311000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.237800451412095e-05, | |
| "loss": 0.5771, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.225183092410128e-05, | |
| "loss": 0.6372, | |
| "step": 313000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.212530463322583e-05, | |
| "loss": 0.7122, | |
| "step": 314000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.6862, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.18711994874345e-05, | |
| "loss": 0.5318, | |
| "step": 316000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.174362341137177e-05, | |
| "loss": 0.5188, | |
| "step": 317000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.161570019212921e-05, | |
| "loss": 0.6145, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.148743122865463e-05, | |
| "loss": 0.6539, | |
| "step": 319000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.6925, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.12298616836904e-05, | |
| "loss": 0.6152, | |
| "step": 321000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.110056391894005e-05, | |
| "loss": 0.5647, | |
| "step": 322000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.097092604340542e-05, | |
| "loss": 0.5581, | |
| "step": 323000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.084094947478556e-05, | |
| "loss": 0.6422, | |
| "step": 324000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.7059, | |
| "step": 325000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.057998594759022e-05, | |
| "loss": 0.679, | |
| "step": 326000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.044900184287007e-05, | |
| "loss": 0.5344, | |
| "step": 327000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.031768475274413e-05, | |
| "loss": 0.5249, | |
| "step": 328000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.018603611327504e-05, | |
| "loss": 0.6019, | |
| "step": 329000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.653, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.992174994867123e-05, | |
| "loss": 0.6829, | |
| "step": 331000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.978911531372765e-05, | |
| "loss": 0.6183, | |
| "step": 332000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.965615490979163e-05, | |
| "loss": 0.5724, | |
| "step": 333000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.952287019089685e-05, | |
| "loss": 0.5446, | |
| "step": 334000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.6442, | |
| "step": 335000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.925533364208309e-05, | |
| "loss": 0.7007, | |
| "step": 336000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.912108473790092e-05, | |
| "loss": 0.665, | |
| "step": 337000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.898651737020166e-05, | |
| "loss": 0.5385, | |
| "step": 338000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.88516330105925e-05, | |
| "loss": 0.5304, | |
| "step": 339000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.5917, | |
| "step": 340000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.858091921938988e-05, | |
| "loss": 0.6529, | |
| "step": 341000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.844509274827907e-05, | |
| "loss": 0.6754, | |
| "step": 342000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.830895520619128e-05, | |
| "loss": 0.62, | |
| "step": 343000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.817250808190483e-05, | |
| "loss": 0.629, | |
| "step": 344000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.6933, | |
| "step": 345000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.789869105876083e-05, | |
| "loss": 0.6513, | |
| "step": 346000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.776132415432234e-05, | |
| "loss": 0.5257, | |
| "step": 347000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.762365365649067e-05, | |
| "loss": 0.5081, | |
| "step": 348000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.748568107080832e-05, | |
| "loss": 0.6351, | |
| "step": 349000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.6629, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "eval_loss": 0.37512102723121643, | |
| "eval_runtime": 22.7092, | |
| "eval_samples_per_second": 220.175, | |
| "eval_steps_per_second": 3.479, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.720883567456298e-05, | |
| "loss": 0.6878, | |
| "step": 351000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.70699658915369e-05, | |
| "loss": 0.5621, | |
| "step": 352000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.693080007570084e-05, | |
| "loss": 0.5316, | |
| "step": 353000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.679133974894983e-05, | |
| "loss": 0.6281, | |
| "step": 354000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.6288, | |
| "step": 355000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.651154166637025e-05, | |
| "loss": 0.6956, | |
| "step": 356000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.637120697036866e-05, | |
| "loss": 0.6504, | |
| "step": 357000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.623058388307269e-05, | |
| "loss": 0.5237, | |
| "step": 358000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.608967394231387e-05, | |
| "loss": 0.5017, | |
| "step": 359000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.6306, | |
| "step": 360000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.580699966740201e-05, | |
| "loss": 0.6555, | |
| "step": 361000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.566523842452958e-05, | |
| "loss": 0.6857, | |
| "step": 362000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.552319651072164e-05, | |
| "loss": 0.5646, | |
| "step": 363000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.538087547932585e-05, | |
| "loss": 0.5331, | |
| "step": 364000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.6099, | |
| "step": 365000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.509540229240601e-05, | |
| "loss": 0.624, | |
| "step": 366000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.495225325877103e-05, | |
| "loss": 0.6964, | |
| "step": 367000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.480883135129211e-05, | |
| "loss": 0.6405, | |
| "step": 368000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.466513813840825e-05, | |
| "loss": 0.5255, | |
| "step": 369000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.5008, | |
| "step": 370000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.437694408499933e-05, | |
| "loss": 0.6269, | |
| "step": 371000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.423244639611826e-05, | |
| "loss": 0.6535, | |
| "step": 372000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.408768370508576e-05, | |
| "loss": 0.6812, | |
| "step": 373000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.394265759500348e-05, | |
| "loss": 0.5654, | |
| "step": 374000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.537, | |
| "step": 375000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.365182146448205e-05, | |
| "loss": 0.5914, | |
| "step": 376000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.350601462458024e-05, | |
| "loss": 0.6249, | |
| "step": 377000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.335995072666848e-05, | |
| "loss": 0.6973, | |
| "step": 378000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.32136313680782e-05, | |
| "loss": 0.629, | |
| "step": 379000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.5266, | |
| "step": 380000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.292023267213835e-05, | |
| "loss": 0.5005, | |
| "step": 381000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.277315654334997e-05, | |
| "loss": 0.6212, | |
| "step": 382000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.262583137097018e-05, | |
| "loss": 0.6483, | |
| "step": 383000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.247825876612353e-05, | |
| "loss": 0.6752, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.5732, | |
| "step": 385000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.218237771703921e-05, | |
| "loss": 0.5404, | |
| "step": 386000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.203407250850928e-05, | |
| "loss": 0.5763, | |
| "step": 387000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.188552633889259e-05, | |
| "loss": 0.6268, | |
| "step": 388000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.173674083266624e-05, | |
| "loss": 0.692, | |
| "step": 389000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.6264, | |
| "step": 390000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.143845832136188e-05, | |
| "loss": 0.5213, | |
| "step": 391000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.128896457825364e-05, | |
| "loss": 0.499, | |
| "step": 392000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.113923802243957e-05, | |
| "loss": 0.6167, | |
| "step": 393000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.09892802913053e-05, | |
| "loss": 0.6491, | |
| "step": 394000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.6636, | |
| "step": 395000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.068867786524116e-05, | |
| "loss": 0.5769, | |
| "step": 396000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.053803645765128e-05, | |
| "loss": 0.5475, | |
| "step": 397000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.038717044938519e-05, | |
| "loss": 0.5534, | |
| "step": 398000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.023608149028937e-05, | |
| "loss": 0.6323, | |
| "step": 399000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.6908, | |
| "step": 400000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "eval_loss": 0.35064515471458435, | |
| "eval_runtime": 19.4337, | |
| "eval_samples_per_second": 257.285, | |
| "eval_steps_per_second": 4.065, | |
| "step": 400000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.993324133116726e-05, | |
| "loss": 0.6153, | |
| "step": 401000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.978149344295242e-05, | |
| "loss": 0.5254, | |
| "step": 402000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.962952922749457e-05, | |
| "loss": 0.5045, | |
| "step": 403000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.947735034665002e-05, | |
| "loss": 0.601, | |
| "step": 404000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.6463, | |
| "step": 405000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.917235524794558e-05, | |
| "loss": 0.6586, | |
| "step": 406000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.901954236546323e-05, | |
| "loss": 0.5835, | |
| "step": 407000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.886652148831279e-05, | |
| "loss": 0.5539, | |
| "step": 408000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.871329428990602e-05, | |
| "loss": 0.5354, | |
| "step": 409000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.6341, | |
| "step": 410000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.840622763423391e-05, | |
| "loss": 0.6923, | |
| "step": 411000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.825239153500029e-05, | |
| "loss": 0.6159, | |
| "step": 412000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.809835583053715e-05, | |
| "loss": 0.5116, | |
| "step": 413000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.794412220535426e-05, | |
| "loss": 0.5119, | |
| "step": 414000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.5906, | |
| "step": 415000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.763506794167208e-05, | |
| "loss": 0.6402, | |
| "step": 416000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.748025068294067e-05, | |
| "loss": 0.6549, | |
| "step": 417000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.732524226298841e-05, | |
| "loss": 0.5867, | |
| "step": 418000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.71700443769625e-05, | |
| "loss": 0.559, | |
| "step": 419000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.5155, | |
| "step": 420000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.685908699762002e-05, | |
| "loss": 0.6399, | |
| "step": 421000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.670333090488356e-05, | |
| "loss": 0.6847, | |
| "step": 422000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.654739214719641e-05, | |
| "loss": 0.6183, | |
| "step": 423000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.639127242987988e-05, | |
| "loss": 0.5072, | |
| "step": 424000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.513, | |
| "step": 425000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.607849694751977e-05, | |
| "loss": 0.5793, | |
| "step": 426000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.592184460293877e-05, | |
| "loss": 0.6379, | |
| "step": 427000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.576501813961609e-05, | |
| "loss": 0.6484, | |
| "step": 428000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.56080192725808e-05, | |
| "loss": 0.5893, | |
| "step": 429000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.5657, | |
| "step": 430000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.529351119689688e-05, | |
| "loss": 0.5003, | |
| "step": 431000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.513600542765817e-05, | |
| "loss": 0.6413, | |
| "step": 432000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.497833413348909e-05, | |
| "loss": 0.6753, | |
| "step": 433000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.48204990386577e-05, | |
| "loss": 0.6138, | |
| "step": 434000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.5166, | |
| "step": 435000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.450434435301751e-05, | |
| "loss": 0.5174, | |
| "step": 436000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.43460282196257e-05, | |
| "loss": 0.5669, | |
| "step": 437000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.418755520036775e-05, | |
| "loss": 0.634, | |
| "step": 438000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.402892702827916e-05, | |
| "loss": 0.6451, | |
| "step": 439000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.596, | |
| "step": 440000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.371121216621698e-05, | |
| "loss": 0.5627, | |
| "step": 441000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.355212895072223e-05, | |
| "loss": 0.5008, | |
| "step": 442000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.339289753131649e-05, | |
| "loss": 0.6418, | |
| "step": 443000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.323351964932908e-05, | |
| "loss": 0.6592, | |
| "step": 444000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.6128, | |
| "step": 445000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.291433147091583e-05, | |
| "loss": 0.521, | |
| "step": 446000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.275452466508077e-05, | |
| "loss": 0.5179, | |
| "step": 447000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.259457837780742e-05, | |
| "loss": 0.5598, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.243449435824276e-05, | |
| "loss": 0.631, | |
| "step": 449000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.6381, | |
| "step": 450000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "eval_loss": 0.3511766195297241, | |
| "eval_runtime": 18.2763, | |
| "eval_samples_per_second": 273.578, | |
| "eval_steps_per_second": 4.323, | |
| "step": 450000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.211392012633932e-05, | |
| "loss": 0.5995, | |
| "step": 451000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.195343341974899e-05, | |
| "loss": 0.5591, | |
| "step": 452000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.179281599232591e-05, | |
| "loss": 0.4996, | |
| "step": 453000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.163206960055651e-05, | |
| "loss": 0.6401, | |
| "step": 454000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.6407, | |
| "step": 455000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.131019695695702e-05, | |
| "loss": 0.6093, | |
| "step": 456000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.11490742250746e-05, | |
| "loss": 0.5331, | |
| "step": 457000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.0987829568702656e-05, | |
| "loss": 0.5195, | |
| "step": 458000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.0826464751186994e-05, | |
| "loss": 0.5519, | |
| "step": 459000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.6198, | |
| "step": 460000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.05033816926583e-05, | |
| "loss": 0.6362, | |
| "step": 461000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.034166698482984e-05, | |
| "loss": 0.6054, | |
| "step": 462000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.017983918218812e-05, | |
| "loss": 0.5557, | |
| "step": 463000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.001790005445607e-05, | |
| "loss": 0.4995, | |
| "step": 464000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.6346, | |
| "step": 465000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.969369490868042e-05, | |
| "loss": 0.6212, | |
| "step": 466000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.953143243609235e-05, | |
| "loss": 0.6074, | |
| "step": 467000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.9369065729286245e-05, | |
| "loss": 0.5449, | |
| "step": 468000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.9206596563878357e-05, | |
| "loss": 0.5182, | |
| "step": 469000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.5475, | |
| "step": 470000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.888135796530544e-05, | |
| "loss": 0.6102, | |
| "step": 471000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.871859208889759e-05, | |
| "loss": 0.6328, | |
| "step": 472000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.85557308673635e-05, | |
| "loss": 0.6164, | |
| "step": 473000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.8392776081727385e-05, | |
| "loss": 0.5453, | |
| "step": 474000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.5044, | |
| "step": 475000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.8066592947342555e-05, | |
| "loss": 0.6267, | |
| "step": 476000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.7903368165680327e-05, | |
| "loss": 0.6045, | |
| "step": 477000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.7740056954050084e-05, | |
| "loss": 0.5922, | |
| "step": 478000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.757666109839702e-05, | |
| "loss": 0.6466, | |
| "step": 479000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.6223, | |
| "step": 480000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.72496226034123e-05, | |
| "loss": 0.5029, | |
| "step": 481000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.7085983540521216e-05, | |
| "loss": 0.4718, | |
| "step": 482000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.692226698644938e-05, | |
| "loss": 0.596, | |
| "step": 483000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.675847473157485e-05, | |
| "loss": 0.6254, | |
| "step": 484000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.657, | |
| "step": 485000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.6430670285049314e-05, | |
| "loss": 0.5453, | |
| "step": 486000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.6266661678215216e-05, | |
| "loss": 0.5089, | |
| "step": 487000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.6102584540173006e-05, | |
| "loss": 0.5764, | |
| "step": 488000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.5938440665244006e-05, | |
| "loss": 0.5935, | |
| "step": 489000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.6528, | |
| "step": 490000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.560995988564023e-05, | |
| "loss": 0.6246, | |
| "step": 491000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.544562657317863e-05, | |
| "loss": 0.5041, | |
| "step": 492000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.52812337082173e-05, | |
| "loss": 0.4668, | |
| "step": 493000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.511678308853026e-05, | |
| "loss": 0.5947, | |
| "step": 494000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.6195, | |
| "step": 495000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.478771577921351e-05, | |
| "loss": 0.6533, | |
| "step": 496000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.462310268821118e-05, | |
| "loss": 0.548, | |
| "step": 497000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.445843903969854e-05, | |
| "loss": 0.5116, | |
| "step": 498000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.4293726634410855e-05, | |
| "loss": 0.565, | |
| "step": 499000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.5847, | |
| "step": 500000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "eval_loss": 0.3354858458042145, | |
| "eval_runtime": 18.2066, | |
| "eval_samples_per_second": 274.625, | |
| "eval_steps_per_second": 4.339, | |
| "step": 500000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.396416275909779e-05, | |
| "loss": 0.6564, | |
| "step": 501000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.379931489313016e-05, | |
| "loss": 0.6283, | |
| "step": 502000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.363442547846356e-05, | |
| "loss": 0.4985, | |
| "step": 503000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.3469496318302204e-05, | |
| "loss": 0.4633, | |
| "step": 504000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.5911, | |
| "step": 505000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.313952597646568e-05, | |
| "loss": 0.6127, | |
| "step": 506000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.297448840329329e-05, | |
| "loss": 0.6492, | |
| "step": 507000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.280941830159227e-05, | |
| "loss": 0.5514, | |
| "step": 508000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.264431747654284e-05, | |
| "loss": 0.5129, | |
| "step": 509000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.5505, | |
| "step": 510000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.231403087877955e-05, | |
| "loss": 0.5818, | |
| "step": 511000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.214884871802703e-05, | |
| "loss": 0.6606, | |
| "step": 512000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.198364305780922e-05, | |
| "loss": 0.6196, | |
| "step": 513000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.1818415704788725e-05, | |
| "loss": 0.497, | |
| "step": 514000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.4664, | |
| "step": 515000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.148790314815663e-05, | |
| "loss": 0.581, | |
| "step": 516000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.132262155897739e-05, | |
| "loss": 0.6087, | |
| "step": 517000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.1157325505820694e-05, | |
| "loss": 0.6447, | |
| "step": 518000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.0992016796337686e-05, | |
| "loss": 0.5588, | |
| "step": 519000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.5145, | |
| "step": 520000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.066136863966963e-05, | |
| "loss": 0.5328, | |
| "step": 521000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.0496032808399815e-05, | |
| "loss": 0.5876, | |
| "step": 522000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 5.033069155259471e-05, | |
| "loss": 0.6572, | |
| "step": 523000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 5.016534668039976e-05, | |
| "loss": 0.6204, | |
| "step": 524000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 5e-05, | |
| "loss": 0.487, | |
| "step": 525000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.9834653319600246e-05, | |
| "loss": 0.4748, | |
| "step": 526000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.96693084474053e-05, | |
| "loss": 0.5704, | |
| "step": 527000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.950396719160018e-05, | |
| "loss": 0.6067, | |
| "step": 528000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.93386313603304e-05, | |
| "loss": 0.6362, | |
| "step": 529000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.563, | |
| "step": 530000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.9007983203662326e-05, | |
| "loss": 0.5207, | |
| "step": 531000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.884267449417931e-05, | |
| "loss": 0.5126, | |
| "step": 532000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.867737844102261e-05, | |
| "loss": 0.5924, | |
| "step": 533000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.851209685184338e-05, | |
| "loss": 0.6539, | |
| "step": 534000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.6201, | |
| "step": 535000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.818158429521129e-05, | |
| "loss": 0.4841, | |
| "step": 536000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.801635694219079e-05, | |
| "loss": 0.4804, | |
| "step": 537000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.785115128197298e-05, | |
| "loss": 0.5584, | |
| "step": 538000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.7685969121220456e-05, | |
| "loss": 0.6045, | |
| "step": 539000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.6299, | |
| "step": 540000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.735568252345718e-05, | |
| "loss": 0.5677, | |
| "step": 541000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.7190581698407725e-05, | |
| "loss": 0.5271, | |
| "step": 542000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.702551159670672e-05, | |
| "loss": 0.4975, | |
| "step": 543000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.6860474023534335e-05, | |
| "loss": 0.5969, | |
| "step": 544000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.648, | |
| "step": 545000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.65305036816978e-05, | |
| "loss": 0.6045, | |
| "step": 546000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.6365574521536445e-05, | |
| "loss": 0.4908, | |
| "step": 547000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.620068510686985e-05, | |
| "loss": 0.4847, | |
| "step": 548000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.60358372409022e-05, | |
| "loss": 0.5487, | |
| "step": 549000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.605, | |
| "step": 550000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "eval_loss": 0.3317408561706543, | |
| "eval_runtime": 18.5965, | |
| "eval_samples_per_second": 268.868, | |
| "eval_steps_per_second": 4.248, | |
| "step": 550000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.570627336558915e-05, | |
| "loss": 0.6229, | |
| "step": 551000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.554156096030149e-05, | |
| "loss": 0.57, | |
| "step": 552000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.537689731178883e-05, | |
| "loss": 0.5323, | |
| "step": 553000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.5212284220786494e-05, | |
| "loss": 0.4832, | |
| "step": 554000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.6006, | |
| "step": 555000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.488321691146975e-05, | |
| "loss": 0.6477, | |
| "step": 556000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.471876629178273e-05, | |
| "loss": 0.5941, | |
| "step": 557000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.4554373426821374e-05, | |
| "loss": 0.4898, | |
| "step": 558000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.439004011435979e-05, | |
| "loss": 0.4889, | |
| "step": 559000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.537, | |
| "step": 560000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.406155933475599e-05, | |
| "loss": 0.6031, | |
| "step": 561000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.3897415459827e-05, | |
| "loss": 0.6173, | |
| "step": 562000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.373333832178478e-05, | |
| "loss": 0.5678, | |
| "step": 563000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.3569329714950704e-05, | |
| "loss": 0.5419, | |
| "step": 564000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.4704, | |
| "step": 565000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.324152526842517e-05, | |
| "loss": 0.5991, | |
| "step": 566000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.307773301355062e-05, | |
| "loss": 0.6427, | |
| "step": 567000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.291401645947879e-05, | |
| "loss": 0.5833, | |
| "step": 568000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.275037739658771e-05, | |
| "loss": 0.4972, | |
| "step": 569000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.4953, | |
| "step": 570000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.2423338901602985e-05, | |
| "loss": 0.5242, | |
| "step": 571000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.2259943045949934e-05, | |
| "loss": 0.5994, | |
| "step": 572000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.209663183431969e-05, | |
| "loss": 0.6135, | |
| "step": 573000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.1933407052657456e-05, | |
| "loss": 0.573, | |
| "step": 574000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.5454, | |
| "step": 575000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.160722391827262e-05, | |
| "loss": 0.4636, | |
| "step": 576000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.14442691326365e-05, | |
| "loss": 0.6024, | |
| "step": 577000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.1281407911102425e-05, | |
| "loss": 0.6273, | |
| "step": 578000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.111864203469457e-05, | |
| "loss": 0.5809, | |
| "step": 579000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.5038, | |
| "step": 580000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.079340343612165e-05, | |
| "loss": 0.4953, | |
| "step": 581000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.063093427071376e-05, | |
| "loss": 0.5168, | |
| "step": 582000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.046856756390767e-05, | |
| "loss": 0.5971, | |
| "step": 583000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.0306305091319595e-05, | |
| "loss": 0.6083, | |
| "step": 584000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.5835, | |
| "step": 585000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.9982099945543945e-05, | |
| "loss": 0.5346, | |
| "step": 586000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.982016081781189e-05, | |
| "loss": 0.4621, | |
| "step": 587000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.965833301517017e-05, | |
| "loss": 0.6009, | |
| "step": 588000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.949661830734172e-05, | |
| "loss": 0.613, | |
| "step": 589000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.5751, | |
| "step": 590000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.917353524881302e-05, | |
| "loss": 0.5148, | |
| "step": 591000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.901217043129735e-05, | |
| "loss": 0.4966, | |
| "step": 592000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.8850925774925425e-05, | |
| "loss": 0.5079, | |
| "step": 593000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.8689803043043e-05, | |
| "loss": 0.5916, | |
| "step": 594000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.6047, | |
| "step": 595000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.836793039944349e-05, | |
| "loss": 0.5939, | |
| "step": 596000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.820718400767409e-05, | |
| "loss": 0.523, | |
| "step": 597000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.8046566580251e-05, | |
| "loss": 0.4689, | |
| "step": 598000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.788607987366069e-05, | |
| "loss": 0.592, | |
| "step": 599000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.5945, | |
| "step": 600000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "eval_loss": 0.32049721479415894, | |
| "eval_runtime": 18.2725, | |
| "eval_samples_per_second": 273.635, | |
| "eval_steps_per_second": 4.323, | |
| "step": 600000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.756550564175727e-05, | |
| "loss": 0.5732, | |
| "step": 601000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.74054216221926e-05, | |
| "loss": 0.5276, | |
| "step": 602000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.7245475334919246e-05, | |
| "loss": 0.4961, | |
| "step": 603000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.7085668529084184e-05, | |
| "loss": 0.4999, | |
| "step": 604000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.5845, | |
| "step": 605000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.676648035067093e-05, | |
| "loss": 0.6021, | |
| "step": 606000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.6607102468683526e-05, | |
| "loss": 0.6086, | |
| "step": 607000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.6447871049277796e-05, | |
| "loss": 0.5087, | |
| "step": 608000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.628878783378302e-05, | |
| "loss": 0.4775, | |
| "step": 609000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.582, | |
| "step": 610000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.597107297172084e-05, | |
| "loss": 0.5757, | |
| "step": 611000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.581244479963225e-05, | |
| "loss": 0.564, | |
| "step": 612000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.5653971780374295e-05, | |
| "loss": 0.6098, | |
| "step": 613000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.5495655646982505e-05, | |
| "loss": 0.6033, | |
| "step": 614000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.4888, | |
| "step": 615000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.517950096134232e-05, | |
| "loss": 0.4419, | |
| "step": 616000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.5021665866510925e-05, | |
| "loss": 0.56, | |
| "step": 617000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.4863994572341843e-05, | |
| "loss": 0.5894, | |
| "step": 618000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.470648880310313e-05, | |
| "loss": 0.6236, | |
| "step": 619000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.5347, | |
| "step": 620000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.439198072741921e-05, | |
| "loss": 0.4889, | |
| "step": 621000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.423498186038393e-05, | |
| "loss": 0.5303, | |
| "step": 622000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.407815539706124e-05, | |
| "loss": 0.5674, | |
| "step": 623000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.392150305248024e-05, | |
| "loss": 0.6141, | |
| "step": 624000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.6134, | |
| "step": 625000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.360872757012011e-05, | |
| "loss": 0.4802, | |
| "step": 626000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.3452607852803584e-05, | |
| "loss": 0.436, | |
| "step": 627000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.329666909511645e-05, | |
| "loss": 0.557, | |
| "step": 628000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.3140913002379995e-05, | |
| "loss": 0.5836, | |
| "step": 629000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.6191, | |
| "step": 630000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.282995562303754e-05, | |
| "loss": 0.537, | |
| "step": 631000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.267475773701161e-05, | |
| "loss": 0.4898, | |
| "step": 632000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.251974931705933e-05, | |
| "loss": 0.5207, | |
| "step": 633000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.236493205832795e-05, | |
| "loss": 0.5578, | |
| "step": 634000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.618, | |
| "step": 635000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.205587779464576e-05, | |
| "loss": 0.6211, | |
| "step": 636000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.190164416946285e-05, | |
| "loss": 0.4672, | |
| "step": 637000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.1747608464999725e-05, | |
| "loss": 0.4402, | |
| "step": 638000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.1593772365766105e-05, | |
| "loss": 0.5467, | |
| "step": 639000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.5789, | |
| "step": 640000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.128670571009399e-05, | |
| "loss": 0.6173, | |
| "step": 641000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.113347851168721e-05, | |
| "loss": 0.5404, | |
| "step": 642000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.098045763453678e-05, | |
| "loss": 0.4921, | |
| "step": 643000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.082764475205442e-05, | |
| "loss": 0.5099, | |
| "step": 644000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.5509, | |
| "step": 645000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.052264965335e-05, | |
| "loss": 0.6214, | |
| "step": 646000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.0370470772505433e-05, | |
| "loss": 0.6277, | |
| "step": 647000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.0218506557047598e-05, | |
| "loss": 0.4556, | |
| "step": 648000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.006675866883275e-05, | |
| "loss": 0.4449, | |
| "step": 649000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.5551, | |
| "step": 650000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "eval_loss": 0.3212037682533264, | |
| "eval_runtime": 18.8472, | |
| "eval_samples_per_second": 265.291, | |
| "eval_steps_per_second": 4.192, | |
| "step": 650000 | |
| } | |
| ], | |
| "max_steps": 1000000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 3.877246373020356e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |