| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.194733, | |
| "global_step": 500000, | |
| "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": 5.9575, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 4.7583, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6e-06, | |
| "loss": 3.9123, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 3.5753, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1e-05, | |
| "loss": 3.4934, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.2e-05, | |
| "loss": 3.5832, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 3.4149, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 3.0618, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.8e-05, | |
| "loss": 2.9484, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-05, | |
| "loss": 2.9068, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 2.8138, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4e-05, | |
| "loss": 2.7495, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 2.6121, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 2.49, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3e-05, | |
| "loss": 2.4663, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 2.4551, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 2.2625, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.6e-05, | |
| "loss": 2.1167, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8e-05, | |
| "loss": 2.1513, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4e-05, | |
| "loss": 2.1183, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.2e-05, | |
| "loss": 2.0262, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 1.9458, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 1.9236, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.8e-05, | |
| "loss": 1.9408, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5e-05, | |
| "loss": 1.9222, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 1.6797, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 1.3323, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 1.3642, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.8e-05, | |
| "loss": 1.408, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6e-05, | |
| "loss": 1.4379, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.2e-05, | |
| "loss": 1.4542, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 1.2679, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.6e-05, | |
| "loss": 1.1874, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 1.2039, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7e-05, | |
| "loss": 1.2883, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.2e-05, | |
| "loss": 1.2982, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.4e-05, | |
| "loss": 0.9953, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.6e-05, | |
| "loss": 0.9074, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.800000000000001e-05, | |
| "loss": 1.1001, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8e-05, | |
| "loss": 1.1365, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.2e-05, | |
| "loss": 1.1886, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.4e-05, | |
| "loss": 1.007, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.6e-05, | |
| "loss": 0.9038, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 1.1163, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9e-05, | |
| "loss": 1.245, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 1.2137, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.4e-05, | |
| "loss": 1.0815, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.6e-05, | |
| "loss": 1.028, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.8e-05, | |
| "loss": 1.1969, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1796, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_loss": 0.6671047210693359, | |
| "eval_runtime": 27.8059, | |
| "eval_samples_per_second": 179.818, | |
| "eval_steps_per_second": 2.841, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.999972660400536e-05, | |
| "loss": 1.0995, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.999890641901125e-05, | |
| "loss": 0.9868, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.999753945398704e-05, | |
| "loss": 0.9061, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.99956257238817e-05, | |
| "loss": 0.9516, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.0541, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.999015805811965e-05, | |
| "loss": 0.9582, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.998660418225645e-05, | |
| "loss": 0.9137, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.998250366089848e-05, | |
| "loss": 0.9191, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.997785653888835e-05, | |
| "loss": 1.0307, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.0231, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.996692270216947e-05, | |
| "loss": 1.0593, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.996063610703137e-05, | |
| "loss": 0.9809, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.995380315038119e-05, | |
| "loss": 1.0332, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.994642390694308e-05, | |
| "loss": 1.1253, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.1675, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.993002688846913e-05, | |
| "loss": 1.0261, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.992100929274846e-05, | |
| "loss": 0.8793, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.991144576886823e-05, | |
| "loss": 0.9977, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.990133642141359e-05, | |
| "loss": 1.0335, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.963, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.987948070396571e-05, | |
| "loss": 0.827, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.986773457298311e-05, | |
| "loss": 0.7776, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.985544309644475e-05, | |
| "loss": 0.9454, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.984260640876821e-05, | |
| "loss": 0.9978, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.0376, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.981529796748134e-05, | |
| "loss": 0.8252, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.980082651251175e-05, | |
| "loss": 0.822, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.97858104436822e-05, | |
| "loss": 1.0793, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.977024992520602e-05, | |
| "loss": 1.096, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.1037, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.973749622593534e-05, | |
| "loss": 0.9581, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.972030340333001e-05, | |
| "loss": 0.9125, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.970256684745258e-05, | |
| "loss": 1.0138, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.968428675226714e-05, | |
| "loss": 1.0536, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.9778, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.964609674954696e-05, | |
| "loss": 0.8258, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.962618725965196e-05, | |
| "loss": 0.7035, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.96057350657239e-05, | |
| "loss": 0.9263, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.95847403914247e-05, | |
| "loss": 0.9476, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.9365, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.954112452602045e-05, | |
| "loss": 0.8336, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.95185038118915e-05, | |
| "loss": 0.8092, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.949534157133844e-05, | |
| "loss": 0.9216, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.94716380576598e-05, | |
| "loss": 1.0725, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.008, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.942260825371358e-05, | |
| "loss": 0.9075, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.939728249962807e-05, | |
| "loss": 0.9464, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.937141654477528e-05, | |
| "loss": 1.0647, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.934501067202117e-05, | |
| "loss": 1.0674, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.9174, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_loss": 0.561288058757782, | |
| "eval_runtime": 19.7846, | |
| "eval_samples_per_second": 252.722, | |
| "eval_steps_per_second": 3.993, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.929058033379181e-05, | |
| "loss": 0.837, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.926255646355804e-05, | |
| "loss": 0.8858, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.923399386589933e-05, | |
| "loss": 0.8565, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.92048928531717e-05, | |
| "loss": 0.918, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.7992, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.914507686137019e-05, | |
| "loss": 0.768, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.911436253643445e-05, | |
| "loss": 0.8874, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.90831111046988e-05, | |
| "loss": 0.9455, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.905132290792394e-05, | |
| "loss": 0.9339, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.8543, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.89861376156452e-05, | |
| "loss": 0.8612, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.895274123299723e-05, | |
| "loss": 1.0155, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.891880951101407e-05, | |
| "loss": 1.0372, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.888434282076758e-05, | |
| "loss": 1.0676, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.9111, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.881380604901964e-05, | |
| "loss": 0.7757, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.877773673889701e-05, | |
| "loss": 0.9877, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.87411340032603e-05, | |
| "loss": 0.953, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.870399824239117e-05, | |
| "loss": 0.8896, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.7065, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.862812927522309e-05, | |
| "loss": 0.7254, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.858939689861506e-05, | |
| "loss": 0.8743, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.855013315614725e-05, | |
| "loss": 0.9483, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.851033847720166e-05, | |
| "loss": 0.9044, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.7765, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.842915805643155e-05, | |
| "loss": 0.7459, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.838777320238312e-05, | |
| "loss": 1.0114, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.834585918739936e-05, | |
| "loss": 1.0475, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.830341646984521e-05, | |
| "loss": 0.9809, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.8786, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.821694678938953e-05, | |
| "loss": 0.9085, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.817292077210659e-05, | |
| "loss": 0.9959, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.812836794348004e-05, | |
| "loss": 0.9809, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.808328879073251e-05, | |
| "loss": 0.8956, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.7716, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.799155349053851e-05, | |
| "loss": 0.7292, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.794489834629455e-05, | |
| "loss": 0.8515, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.789771888432375e-05, | |
| "loss": 0.8932, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.785001562057309e-05, | |
| "loss": 0.8301, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.7513, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.775303978015585e-05, | |
| "loss": 0.8437, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.77037682640015e-05, | |
| "loss": 0.8646, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.765397506708023e-05, | |
| "loss": 0.9861, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.760366073392246e-05, | |
| "loss": 0.9019, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.8397, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.750147086550844e-05, | |
| "loss": 0.971, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.744959644778422e-05, | |
| "loss": 1.0129, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.739720312887535e-05, | |
| "loss": 1.0488, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.734429148174675e-05, | |
| "loss": 0.7958, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.7959, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_loss": 0.536599338054657, | |
| "eval_runtime": 19.5468, | |
| "eval_samples_per_second": 255.796, | |
| "eval_steps_per_second": 4.042, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.723691552302562e-05, | |
| "loss": 0.9099, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.718245238567939e-05, | |
| "loss": 0.8585, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.712747326859315e-05, | |
| "loss": 0.8351, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.707197877300974e-05, | |
| "loss": 0.7363, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.701, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.695944607949649e-05, | |
| "loss": 0.8942, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.690240911220618e-05, | |
| "loss": 0.928, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.684485922768422e-05, | |
| "loss": 0.8611, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.6786797055287e-05, | |
| "loss": 0.7645, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.8364, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.66691383922964e-05, | |
| "loss": 0.9962, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.660954318839933e-05, | |
| "loss": 1.0221, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.654943827000548e-05, | |
| "loss": 0.9659, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.648882429441257e-05, | |
| "loss": 0.8713, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.821, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.636607182864827e-05, | |
| "loss": 0.9402, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.630393468087818e-05, | |
| "loss": 0.9458, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.624129116069694e-05, | |
| "loss": 0.8473, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.617814195316411e-05, | |
| "loss": 0.6521, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.7333, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.605032924392457e-05, | |
| "loss": 0.8378, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.598566713995718e-05, | |
| "loss": 0.9204, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.59205021441015e-05, | |
| "loss": 0.8255, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.58548349689915e-05, | |
| "loss": 0.7353, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.7614, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.572199695899522e-05, | |
| "loss": 0.9466, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.565482757680415e-05, | |
| "loss": 1.0166, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.558715892073323e-05, | |
| "loss": 0.8979, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.551899173079607e-05, | |
| "loss": 0.8246, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.9446, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.538116473662861e-05, | |
| "loss": 0.9918, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.531150643965223e-05, | |
| "loss": 0.9424, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.524135262330098e-05, | |
| "loss": 0.8325, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.517070405476575e-05, | |
| "loss": 0.737, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.7896, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.502792575695112e-05, | |
| "loss": 0.8106, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.49557975890723e-05, | |
| "loss": 0.8712, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.488317779179361e-05, | |
| "loss": 0.732, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.481006715927351e-05, | |
| "loss": 0.708, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.8834, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.46623765919727e-05, | |
| "loss": 0.8634, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.458779827231237e-05, | |
| "loss": 0.8771, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.451273234763371e-05, | |
| "loss": 0.8505, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.443717963884569e-05, | |
| "loss": 0.8172, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.974, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.428461717918511e-05, | |
| "loss": 0.9992, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.420760909671118e-05, | |
| "loss": 0.9623, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.413011756690685e-05, | |
| "loss": 0.7999, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.405214343720707e-05, | |
| "loss": 0.7812, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.8985, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.5022637844085693, | |
| "eval_runtime": 30.1991, | |
| "eval_samples_per_second": 165.568, | |
| "eval_steps_per_second": 2.616, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.389475079423988e-05, | |
| "loss": 0.8904, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.381533400219318e-05, | |
| "loss": 0.7439, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.373543805267368e-05, | |
| "loss": 0.6965, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.365506381941066e-05, | |
| "loss": 0.704, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.8834, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.349288402271388e-05, | |
| "loss": 0.9006, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.341108023285238e-05, | |
| "loss": 0.8114, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.332880170637252e-05, | |
| "loss": 0.6866, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.32460493430591e-05, | |
| "loss": 0.8483, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.9693, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.30791267309698e-05, | |
| "loss": 0.995, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.299495830763286e-05, | |
| "loss": 0.8934, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.291031969832026e-05, | |
| "loss": 0.8249, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.282521182862629e-05, | |
| "loss": 0.8899, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.8975, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.265359203611987e-05, | |
| "loss": 0.9485, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.256708199011401e-05, | |
| "loss": 0.7919, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.248010643731935e-05, | |
| "loss": 0.6263, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.239266632888659e-05, | |
| "loss": 0.7575, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.8192, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.221639627510076e-05, | |
| "loss": 0.8844, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.212756825740873e-05, | |
| "loss": 0.7463, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.20382795393797e-05, | |
| "loss": 0.6978, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.194853109746074e-05, | |
| "loss": 0.8168, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.8836, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.176765897286813e-05, | |
| "loss": 0.9723, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.167653726818305e-05, | |
| "loss": 0.8346, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.158495979556358e-05, | |
| "loss": 0.793, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.14929275564863e-05, | |
| "loss": 0.9601, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.9825, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.130750280971978e-05, | |
| "loss": 0.8888, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.121411232980588e-05, | |
| "loss": 0.7961, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.112027113896262e-05, | |
| "loss": 0.7221, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.102598026342222e-05, | |
| "loss": 0.8318, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.7883, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.083605358775612e-05, | |
| "loss": 0.7952, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.074041986463808e-05, | |
| "loss": 0.6878, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.064434061081562e-05, | |
| "loss": 0.7299, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.0547816876996e-05, | |
| "loss": 0.8636, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.8684, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.035344019648702e-05, | |
| "loss": 0.7811, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.025558937546988e-05, | |
| "loss": 0.7957, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.015729832577681e-05, | |
| "loss": 0.8447, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.005856812230304e-05, | |
| "loss": 0.958, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.9778, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.98597945775948e-05, | |
| "loss": 0.9026, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.975975341011596e-05, | |
| "loss": 0.781, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.965927743634391e-05, | |
| "loss": 0.8074, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.955836775506776e-05, | |
| "loss": 0.8925, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.8989, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "eval_loss": 0.4966276288032532, | |
| "eval_runtime": 19.629, | |
| "eval_samples_per_second": 254.725, | |
| "eval_steps_per_second": 4.025, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.935525168886262e-05, | |
| "loss": 0.6879, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.92530475251784e-05, | |
| "loss": 0.6367, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.91504140964553e-05, | |
| "loss": 0.7463, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.90473525250761e-05, | |
| "loss": 0.8463, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.8833, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.883994946727849e-05, | |
| "loss": 0.7523, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.873561024898668e-05, | |
| "loss": 0.6475, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.863084742426719e-05, | |
| "loss": 0.858, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.852566213878947e-05, | |
| "loss": 0.9437, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.9718, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.831402879132446e-05, | |
| "loss": 0.8249, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.820758304372557e-05, | |
| "loss": 0.7836, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.810071946411989e-05, | |
| "loss": 0.9531, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.799343922115044e-05, | |
| "loss": 0.8801, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.913, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.77776334424621e-05, | |
| "loss": 0.7448, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.766911026676064e-05, | |
| "loss": 0.654, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.756017514770443e-05, | |
| "loss": 0.746, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.745082927659047e-05, | |
| "loss": 0.8248, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.7881, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.723091006582389e-05, | |
| "loss": 0.7319, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.71203391311725e-05, | |
| "loss": 0.7143, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.700936225443959e-05, | |
| "loss": 0.8258, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.689798064925049e-05, | |
| "loss": 0.8453, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.8883, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.6674008130122e-05, | |
| "loss": 0.8042, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.656141966551019e-05, | |
| "loss": 0.8175, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.644843137107059e-05, | |
| "loss": 0.945, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.633504448242505e-05, | |
| "loss": 0.9727, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.854, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.610707988678503e-05, | |
| "loss": 0.7374, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.599250467277483e-05, | |
| "loss": 0.7817, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.587753585050004e-05, | |
| "loss": 0.8499, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.576217467724128e-05, | |
| "loss": 0.7804, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.7193, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.553028032833397e-05, | |
| "loss": 0.6388, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.541374968864487e-05, | |
| "loss": 0.7719, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.529683176986295e-05, | |
| "loss": 0.8428, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.517952785058385e-05, | |
| "loss": 0.8685, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.7088, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.494376714600878e-05, | |
| "loss": 0.7253, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.482531293895412e-05, | |
| "loss": 0.8929, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.470647788785665e-05, | |
| "loss": 0.9336, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.458726329227747e-05, | |
| "loss": 0.9651, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.8287, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.434770068665723e-05, | |
| "loss": 0.7827, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.422735529643444e-05, | |
| "loss": 0.8367, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.410663560133784e-05, | |
| "loss": 0.9043, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.398554292153866e-05, | |
| "loss": 0.874, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.6627, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_loss": 0.48531636595726013, | |
| "eval_runtime": 19.6173, | |
| "eval_samples_per_second": 254.876, | |
| "eval_steps_per_second": 4.027, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.37422439088976e-05, | |
| "loss": 0.5916, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.362004023673474e-05, | |
| "loss": 0.7836, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.349746890119826e-05, | |
| "loss": 0.8181, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.337453124270863e-05, | |
| "loss": 0.8338, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.7231, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.31275623385675e-05, | |
| "loss": 0.7914, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.300353379372834e-05, | |
| "loss": 0.9144, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.287914432753123e-05, | |
| "loss": 0.7662, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.275439530027948e-05, | |
| "loss": 0.6571, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.6758, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.250382402347065e-05, | |
| "loss": 0.7863, | |
| "step": 311000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.237800451412095e-05, | |
| "loss": 0.8135, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.225183092410128e-05, | |
| "loss": 0.7382, | |
| "step": 313000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.212530463322583e-05, | |
| "loss": 0.6779, | |
| "step": 314000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.7889, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.18711994874345e-05, | |
| "loss": 0.7984, | |
| "step": 316000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.174362341137177e-05, | |
| "loss": 0.9374, | |
| "step": 317000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.161570019212921e-05, | |
| "loss": 0.8107, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.148743122865463e-05, | |
| "loss": 0.7603, | |
| "step": 319000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.9263, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.12298616836904e-05, | |
| "loss": 0.9464, | |
| "step": 321000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.110056391894005e-05, | |
| "loss": 0.9213, | |
| "step": 322000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.097092604340542e-05, | |
| "loss": 0.7353, | |
| "step": 323000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.084094947478556e-05, | |
| "loss": 0.7104, | |
| "step": 324000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.8383, | |
| "step": 325000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.057998594759022e-05, | |
| "loss": 0.7574, | |
| "step": 326000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.044900184287007e-05, | |
| "loss": 0.7436, | |
| "step": 327000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.031768475274413e-05, | |
| "loss": 0.6639, | |
| "step": 328000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.018603611327504e-05, | |
| "loss": 0.6737, | |
| "step": 329000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.8208, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.992174994867123e-05, | |
| "loss": 0.8542, | |
| "step": 331000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.978911531372765e-05, | |
| "loss": 0.7584, | |
| "step": 332000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.965615490979163e-05, | |
| "loss": 0.724, | |
| "step": 333000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.952287019089685e-05, | |
| "loss": 0.7922, | |
| "step": 334000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.9266, | |
| "step": 335000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.925533364208309e-05, | |
| "loss": 0.9447, | |
| "step": 336000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.912108473790092e-05, | |
| "loss": 0.8837, | |
| "step": 337000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.898651737020166e-05, | |
| "loss": 0.7977, | |
| "step": 338000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.88516330105925e-05, | |
| "loss": 0.7458, | |
| "step": 339000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.8868, | |
| "step": 340000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.858091921938988e-05, | |
| "loss": 0.8581, | |
| "step": 341000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.844509274827907e-05, | |
| "loss": 0.7276, | |
| "step": 342000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.830895520619128e-05, | |
| "loss": 0.6481, | |
| "step": 343000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.817250808190483e-05, | |
| "loss": 0.7676, | |
| "step": 344000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.8441, | |
| "step": 345000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.789869105876083e-05, | |
| "loss": 0.7821, | |
| "step": 346000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.776132415432234e-05, | |
| "loss": 0.9659, | |
| "step": 347000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.762365365649067e-05, | |
| "loss": 0.9989, | |
| "step": 348000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.748568107080832e-05, | |
| "loss": 1.0164, | |
| "step": 349000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.9084, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_loss": 0.4633873999118805, | |
| "eval_runtime": 19.1702, | |
| "eval_samples_per_second": 260.822, | |
| "eval_steps_per_second": 4.121, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.720883567456298e-05, | |
| "loss": 0.8651, | |
| "step": 351000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.70699658915369e-05, | |
| "loss": 0.677, | |
| "step": 352000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.693080007570084e-05, | |
| "loss": 0.7458, | |
| "step": 353000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.679133974894983e-05, | |
| "loss": 0.8246, | |
| "step": 354000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.7698, | |
| "step": 355000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.651154166637025e-05, | |
| "loss": 0.9427, | |
| "step": 356000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.637120697036866e-05, | |
| "loss": 0.9836, | |
| "step": 357000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.623058388307269e-05, | |
| "loss": 1.003, | |
| "step": 358000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.608967394231387e-05, | |
| "loss": 0.9023, | |
| "step": 359000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.8596, | |
| "step": 360000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.580699966740201e-05, | |
| "loss": 0.6692, | |
| "step": 361000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.566523842452958e-05, | |
| "loss": 0.7345, | |
| "step": 362000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.552319651072164e-05, | |
| "loss": 0.8172, | |
| "step": 363000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.538087547932585e-05, | |
| "loss": 0.7651, | |
| "step": 364000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.924, | |
| "step": 365000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.509540229240601e-05, | |
| "loss": 0.9742, | |
| "step": 366000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.495225325877103e-05, | |
| "loss": 0.9938, | |
| "step": 367000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.480883135129211e-05, | |
| "loss": 0.901, | |
| "step": 368000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.466513813840825e-05, | |
| "loss": 0.8604, | |
| "step": 369000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.6632, | |
| "step": 370000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.437694408499933e-05, | |
| "loss": 0.7256, | |
| "step": 371000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.423244639611826e-05, | |
| "loss": 0.8073, | |
| "step": 372000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.408768370508576e-05, | |
| "loss": 0.7628, | |
| "step": 373000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.394265759500348e-05, | |
| "loss": 0.9042, | |
| "step": 374000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.9672, | |
| "step": 375000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.365182146448205e-05, | |
| "loss": 0.9875, | |
| "step": 376000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.350601462458024e-05, | |
| "loss": 0.8989, | |
| "step": 377000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.335995072666848e-05, | |
| "loss": 0.8599, | |
| "step": 378000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.32136313680782e-05, | |
| "loss": 0.6627, | |
| "step": 379000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.7194, | |
| "step": 380000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.292023267213835e-05, | |
| "loss": 0.8005, | |
| "step": 381000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.277315654334997e-05, | |
| "loss": 0.7612, | |
| "step": 382000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.262583137097018e-05, | |
| "loss": 0.8897, | |
| "step": 383000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.247825876612353e-05, | |
| "loss": 0.9613, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.9756, | |
| "step": 385000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.218237771703921e-05, | |
| "loss": 0.8961, | |
| "step": 386000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.203407250850928e-05, | |
| "loss": 0.8547, | |
| "step": 387000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.188552633889259e-05, | |
| "loss": 0.6644, | |
| "step": 388000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.173674083266624e-05, | |
| "loss": 0.7123, | |
| "step": 389000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.7926, | |
| "step": 390000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.143845832136188e-05, | |
| "loss": 0.7619, | |
| "step": 391000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.128896457825364e-05, | |
| "loss": 0.8722, | |
| "step": 392000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.113923802243957e-05, | |
| "loss": 0.9563, | |
| "step": 393000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.09892802913053e-05, | |
| "loss": 0.9717, | |
| "step": 394000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.8937, | |
| "step": 395000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.068867786524116e-05, | |
| "loss": 0.8545, | |
| "step": 396000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.053803645765128e-05, | |
| "loss": 0.6729, | |
| "step": 397000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.038717044938519e-05, | |
| "loss": 0.7015, | |
| "step": 398000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.023608149028937e-05, | |
| "loss": 0.7826, | |
| "step": 399000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.7621, | |
| "step": 400000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "eval_loss": 0.5067209005355835, | |
| "eval_runtime": 18.1497, | |
| "eval_samples_per_second": 275.486, | |
| "eval_steps_per_second": 4.353, | |
| "step": 400000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.993324133116726e-05, | |
| "loss": 0.8315, | |
| "step": 401000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.978149344295242e-05, | |
| "loss": 0.786, | |
| "step": 402000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.962952922749457e-05, | |
| "loss": 0.9338, | |
| "step": 403000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.947735034665002e-05, | |
| "loss": 0.9648, | |
| "step": 404000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.8149, | |
| "step": 405000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.917235524794558e-05, | |
| "loss": 0.7549, | |
| "step": 406000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.901954236546323e-05, | |
| "loss": 0.7084, | |
| "step": 407000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.886652148831279e-05, | |
| "loss": 0.7488, | |
| "step": 408000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.871329428990602e-05, | |
| "loss": 0.7645, | |
| "step": 409000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.7303, | |
| "step": 410000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.840622763423391e-05, | |
| "loss": 0.6498, | |
| "step": 411000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.825239153500029e-05, | |
| "loss": 0.7127, | |
| "step": 412000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.809835583053715e-05, | |
| "loss": 0.8216, | |
| "step": 413000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.794412220535426e-05, | |
| "loss": 0.817, | |
| "step": 414000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.7474, | |
| "step": 415000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.763506794167208e-05, | |
| "loss": 0.7504, | |
| "step": 416000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.748025068294067e-05, | |
| "loss": 0.8277, | |
| "step": 417000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.732524226298841e-05, | |
| "loss": 0.9038, | |
| "step": 418000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.71700443769625e-05, | |
| "loss": 0.9412, | |
| "step": 419000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.8328, | |
| "step": 420000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.685908699762002e-05, | |
| "loss": 0.7013, | |
| "step": 421000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.670333090488356e-05, | |
| "loss": 0.8064, | |
| "step": 422000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.654739214719641e-05, | |
| "loss": 0.8228, | |
| "step": 423000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.639127242987988e-05, | |
| "loss": 0.831, | |
| "step": 424000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.606, | |
| "step": 425000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.607849694751977e-05, | |
| "loss": 0.596, | |
| "step": 426000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.592184460293877e-05, | |
| "loss": 0.7339, | |
| "step": 427000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.576501813961609e-05, | |
| "loss": 0.8095, | |
| "step": 428000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.56080192725808e-05, | |
| "loss": 0.8368, | |
| "step": 429000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.6778, | |
| "step": 430000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.529351119689688e-05, | |
| "loss": 0.602, | |
| "step": 431000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.513600542765817e-05, | |
| "loss": 0.8553, | |
| "step": 432000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.497833413348909e-05, | |
| "loss": 0.9017, | |
| "step": 433000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.48204990386577e-05, | |
| "loss": 0.8937, | |
| "step": 434000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.7829, | |
| "step": 435000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.450434435301751e-05, | |
| "loss": 0.7589, | |
| "step": 436000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.43460282196257e-05, | |
| "loss": 0.893, | |
| "step": 437000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.418755520036775e-05, | |
| "loss": 0.8418, | |
| "step": 438000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.402892702827916e-05, | |
| "loss": 0.8263, | |
| "step": 439000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.6969, | |
| "step": 440000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.371121216621698e-05, | |
| "loss": 0.5887, | |
| "step": 441000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.355212895072223e-05, | |
| "loss": 0.723, | |
| "step": 442000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.339289753131649e-05, | |
| "loss": 0.7757, | |
| "step": 443000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.323351964932908e-05, | |
| "loss": 0.7438, | |
| "step": 444000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.6629, | |
| "step": 445000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.291433147091583e-05, | |
| "loss": 0.6943, | |
| "step": 446000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.275452466508077e-05, | |
| "loss": 0.769, | |
| "step": 447000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.259457837780742e-05, | |
| "loss": 0.8287, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.243449435824276e-05, | |
| "loss": 0.8262, | |
| "step": 449000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.7464, | |
| "step": 450000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "eval_loss": 0.45768678188323975, | |
| "eval_runtime": 28.0661, | |
| "eval_samples_per_second": 178.151, | |
| "eval_steps_per_second": 2.815, | |
| "step": 450000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.211392012633932e-05, | |
| "loss": 0.8045, | |
| "step": 451000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.195343341974899e-05, | |
| "loss": 0.8956, | |
| "step": 452000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.179281599232591e-05, | |
| "loss": 0.9304, | |
| "step": 453000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.163206960055651e-05, | |
| "loss": 0.7553, | |
| "step": 454000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.7025, | |
| "step": 455000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.131019695695702e-05, | |
| "loss": 0.7472, | |
| "step": 456000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.11490742250746e-05, | |
| "loss": 0.7742, | |
| "step": 457000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.0987829568702656e-05, | |
| "loss": 0.733, | |
| "step": 458000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.0826464751186994e-05, | |
| "loss": 0.6584, | |
| "step": 459000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.5952, | |
| "step": 460000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.05033816926583e-05, | |
| "loss": 0.7624, | |
| "step": 461000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.034166698482984e-05, | |
| "loss": 0.7982, | |
| "step": 462000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.017983918218812e-05, | |
| "loss": 0.8103, | |
| "step": 463000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.001790005445607e-05, | |
| "loss": 0.6476, | |
| "step": 464000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.6961, | |
| "step": 465000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.969369490868042e-05, | |
| "loss": 0.8699, | |
| "step": 466000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.953143243609235e-05, | |
| "loss": 0.8909, | |
| "step": 467000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.9369065729286245e-05, | |
| "loss": 0.9026, | |
| "step": 468000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.9206596563878357e-05, | |
| "loss": 0.7783, | |
| "step": 469000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.712, | |
| "step": 470000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.888135796530544e-05, | |
| "loss": 0.8122, | |
| "step": 471000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.871859208889759e-05, | |
| "loss": 0.843, | |
| "step": 472000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.85557308673635e-05, | |
| "loss": 0.789, | |
| "step": 473000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.8392776081727385e-05, | |
| "loss": 0.5965, | |
| "step": 474000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.5814, | |
| "step": 475000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.8066592947342555e-05, | |
| "loss": 0.7358, | |
| "step": 476000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.7903368165680327e-05, | |
| "loss": 0.7996, | |
| "step": 477000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.7740056954050084e-05, | |
| "loss": 0.7557, | |
| "step": 478000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.757666109839702e-05, | |
| "loss": 0.6714, | |
| "step": 479000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.6314, | |
| "step": 480000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.72496226034123e-05, | |
| "loss": 0.8106, | |
| "step": 481000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.7085983540521216e-05, | |
| "loss": 0.8901, | |
| "step": 482000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.692226698644938e-05, | |
| "loss": 0.8339, | |
| "step": 483000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.675847473157485e-05, | |
| "loss": 0.7459, | |
| "step": 484000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.7894, | |
| "step": 485000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.6430670285049314e-05, | |
| "loss": 0.8981, | |
| "step": 486000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.6266661678215216e-05, | |
| "loss": 0.833, | |
| "step": 487000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.6102584540173006e-05, | |
| "loss": 0.7844, | |
| "step": 488000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.5938440665244006e-05, | |
| "loss": 0.6537, | |
| "step": 489000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.6783, | |
| "step": 490000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.560995988564023e-05, | |
| "loss": 0.6874, | |
| "step": 491000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.544562657317863e-05, | |
| "loss": 0.7752, | |
| "step": 492000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.52812337082173e-05, | |
| "loss": 0.6647, | |
| "step": 493000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.511678308853026e-05, | |
| "loss": 0.6382, | |
| "step": 494000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.7431, | |
| "step": 495000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.478771577921351e-05, | |
| "loss": 0.7675, | |
| "step": 496000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.462310268821118e-05, | |
| "loss": 0.787, | |
| "step": 497000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.445843903969854e-05, | |
| "loss": 0.7511, | |
| "step": 498000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.4293726634410855e-05, | |
| "loss": 0.7127, | |
| "step": 499000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.8539, | |
| "step": 500000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "eval_loss": 0.45055532455444336, | |
| "eval_runtime": 18.1381, | |
| "eval_samples_per_second": 275.663, | |
| "eval_steps_per_second": 4.355, | |
| "step": 500000 | |
| } | |
| ], | |
| "max_steps": 1000000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 2.982493424541342e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |