|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.265921, |
|
"global_step": 800000, |
|
"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": 3.0064, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.000000000000001e-06, |
|
"loss": 2.5519, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 6e-06, |
|
"loss": 2.2584, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 8.000000000000001e-06, |
|
"loss": 2.1337, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1e-05, |
|
"loss": 2.0643, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.2e-05, |
|
"loss": 2.0175, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.4000000000000001e-05, |
|
"loss": 1.9603, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.6000000000000003e-05, |
|
"loss": 1.9016, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.8e-05, |
|
"loss": 1.8447, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 2e-05, |
|
"loss": 1.8228, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 2.2000000000000003e-05, |
|
"loss": 1.7864, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 2.4e-05, |
|
"loss": 1.7471, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 2.6000000000000002e-05, |
|
"loss": 1.7281, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 2.8000000000000003e-05, |
|
"loss": 1.6946, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 3e-05, |
|
"loss": 1.6671, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 3.2000000000000005e-05, |
|
"loss": 1.6402, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 3.4000000000000007e-05, |
|
"loss": 1.6367, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 3.6e-05, |
|
"loss": 1.6362, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 3.8e-05, |
|
"loss": 1.618, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4e-05, |
|
"loss": 1.601, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.2e-05, |
|
"loss": 1.5855, |
|
"step": 21000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.4000000000000006e-05, |
|
"loss": 1.5739, |
|
"step": 22000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.600000000000001e-05, |
|
"loss": 1.5626, |
|
"step": 23000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.8e-05, |
|
"loss": 1.5378, |
|
"step": 24000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 5e-05, |
|
"loss": 1.531, |
|
"step": 25000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 5.2000000000000004e-05, |
|
"loss": 1.4111, |
|
"step": 26000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 5.4000000000000005e-05, |
|
"loss": 1.3182, |
|
"step": 27000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 5.6000000000000006e-05, |
|
"loss": 1.2617, |
|
"step": 28000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 5.8e-05, |
|
"loss": 1.2387, |
|
"step": 29000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 6e-05, |
|
"loss": 1.2198, |
|
"step": 30000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 6.2e-05, |
|
"loss": 1.1973, |
|
"step": 31000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 6.400000000000001e-05, |
|
"loss": 1.1757, |
|
"step": 32000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 6.6e-05, |
|
"loss": 1.1629, |
|
"step": 33000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 6.800000000000001e-05, |
|
"loss": 1.1627, |
|
"step": 34000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 7e-05, |
|
"loss": 1.1379, |
|
"step": 35000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 7.2e-05, |
|
"loss": 1.117, |
|
"step": 36000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 7.4e-05, |
|
"loss": 1.1216, |
|
"step": 37000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 7.6e-05, |
|
"loss": 1.1075, |
|
"step": 38000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 7.800000000000001e-05, |
|
"loss": 1.1068, |
|
"step": 39000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 8e-05, |
|
"loss": 1.0974, |
|
"step": 40000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 8.2e-05, |
|
"loss": 1.0872, |
|
"step": 41000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 8.4e-05, |
|
"loss": 1.0596, |
|
"step": 42000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 8.6e-05, |
|
"loss": 1.085, |
|
"step": 43000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 8.800000000000001e-05, |
|
"loss": 1.0661, |
|
"step": 44000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 9e-05, |
|
"loss": 1.0733, |
|
"step": 45000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.200000000000001e-05, |
|
"loss": 1.0724, |
|
"step": 46000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.4e-05, |
|
"loss": 1.0617, |
|
"step": 47000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.6e-05, |
|
"loss": 1.0634, |
|
"step": 48000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.8e-05, |
|
"loss": 1.0536, |
|
"step": 49000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.0001, |
|
"loss": 1.0389, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"eval_loss": 0.9801571369171143, |
|
"eval_runtime": 29.3179, |
|
"eval_samples_per_second": 341.089, |
|
"eval_steps_per_second": 1.364, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.999972660400536e-05, |
|
"loss": 1.0358, |
|
"step": 51000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.999890641901125e-05, |
|
"loss": 1.0516, |
|
"step": 52000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.999753945398704e-05, |
|
"loss": 1.0663, |
|
"step": 53000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.99956257238817e-05, |
|
"loss": 1.0447, |
|
"step": 54000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.999316524962345e-05, |
|
"loss": 1.0509, |
|
"step": 55000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.999015805811965e-05, |
|
"loss": 1.0563, |
|
"step": 56000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.998660418225645e-05, |
|
"loss": 1.0647, |
|
"step": 57000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.998250366089848e-05, |
|
"loss": 1.0479, |
|
"step": 58000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.997785653888835e-05, |
|
"loss": 1.0505, |
|
"step": 59000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.997266286704631e-05, |
|
"loss": 1.0577, |
|
"step": 60000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.996692270216947e-05, |
|
"loss": 1.053, |
|
"step": 61000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.996063610703137e-05, |
|
"loss": 1.0215, |
|
"step": 62000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.995380315038119e-05, |
|
"loss": 1.0404, |
|
"step": 63000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.994642390694308e-05, |
|
"loss": 1.0197, |
|
"step": 64000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.993849845741524e-05, |
|
"loss": 1.007, |
|
"step": 65000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.993002688846913e-05, |
|
"loss": 1.0124, |
|
"step": 66000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.992100929274846e-05, |
|
"loss": 0.9988, |
|
"step": 67000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.991144576886823e-05, |
|
"loss": 0.9969, |
|
"step": 68000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.990133642141359e-05, |
|
"loss": 1.0032, |
|
"step": 69000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.989068136093873e-05, |
|
"loss": 1.0098, |
|
"step": 70000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.987948070396571e-05, |
|
"loss": 1.0193, |
|
"step": 71000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.986773457298311e-05, |
|
"loss": 1.0139, |
|
"step": 72000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.985544309644475e-05, |
|
"loss": 1.0117, |
|
"step": 73000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.984260640876821e-05, |
|
"loss": 1.0136, |
|
"step": 74000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.98292246503335e-05, |
|
"loss": 1.0054, |
|
"step": 75000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.981529796748134e-05, |
|
"loss": 1.003, |
|
"step": 76000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.980082651251175e-05, |
|
"loss": 1.0025, |
|
"step": 77000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.97858104436822e-05, |
|
"loss": 1.0047, |
|
"step": 78000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.977024992520602e-05, |
|
"loss": 1.009, |
|
"step": 79000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.975414512725057e-05, |
|
"loss": 0.9949, |
|
"step": 80000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.973749622593534e-05, |
|
"loss": 0.9953, |
|
"step": 81000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.972030340333001e-05, |
|
"loss": 1.0015, |
|
"step": 82000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.970256684745258e-05, |
|
"loss": 0.9973, |
|
"step": 83000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.968428675226714e-05, |
|
"loss": 0.9887, |
|
"step": 84000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.966546331768191e-05, |
|
"loss": 0.9926, |
|
"step": 85000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.964609674954696e-05, |
|
"loss": 0.9999, |
|
"step": 86000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.962618725965196e-05, |
|
"loss": 1.0005, |
|
"step": 87000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.96057350657239e-05, |
|
"loss": 0.9875, |
|
"step": 88000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.95847403914247e-05, |
|
"loss": 0.9856, |
|
"step": 89000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.956320346634876e-05, |
|
"loss": 0.992, |
|
"step": 90000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.954112452602045e-05, |
|
"loss": 0.9842, |
|
"step": 91000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.95185038118915e-05, |
|
"loss": 0.9902, |
|
"step": 92000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.949534157133844e-05, |
|
"loss": 0.9806, |
|
"step": 93000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.94716380576598e-05, |
|
"loss": 0.9719, |
|
"step": 94000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.944739353007344e-05, |
|
"loss": 0.9551, |
|
"step": 95000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.942260825371358e-05, |
|
"loss": 0.9713, |
|
"step": 96000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.939728249962807e-05, |
|
"loss": 0.9665, |
|
"step": 97000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.937141654477528e-05, |
|
"loss": 0.963, |
|
"step": 98000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.934501067202117e-05, |
|
"loss": 0.9627, |
|
"step": 99000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.931806517013612e-05, |
|
"loss": 0.9685, |
|
"step": 100000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"eval_loss": 0.9021177887916565, |
|
"eval_runtime": 18.972, |
|
"eval_samples_per_second": 527.092, |
|
"eval_steps_per_second": 2.108, |
|
"step": 100000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.929058033379181e-05, |
|
"loss": 0.9669, |
|
"step": 101000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.926255646355804e-05, |
|
"loss": 0.9662, |
|
"step": 102000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.923399386589933e-05, |
|
"loss": 0.9451, |
|
"step": 103000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.92048928531717e-05, |
|
"loss": 0.9497, |
|
"step": 104000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.917525374361912e-05, |
|
"loss": 0.9571, |
|
"step": 105000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.914507686137019e-05, |
|
"loss": 0.9429, |
|
"step": 106000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.911436253643445e-05, |
|
"loss": 0.9275, |
|
"step": 107000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.90831111046988e-05, |
|
"loss": 0.9376, |
|
"step": 108000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.905132290792394e-05, |
|
"loss": 0.938, |
|
"step": 109000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.901899829374047e-05, |
|
"loss": 0.9444, |
|
"step": 110000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.89861376156452e-05, |
|
"loss": 0.9425, |
|
"step": 111000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.895274123299723e-05, |
|
"loss": 0.9422, |
|
"step": 112000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.891880951101407e-05, |
|
"loss": 0.9552, |
|
"step": 113000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.888434282076758e-05, |
|
"loss": 0.9392, |
|
"step": 114000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.884934153917997e-05, |
|
"loss": 0.9262, |
|
"step": 115000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.881380604901964e-05, |
|
"loss": 0.9391, |
|
"step": 116000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.877773673889701e-05, |
|
"loss": 0.9313, |
|
"step": 117000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.87411340032603e-05, |
|
"loss": 0.9372, |
|
"step": 118000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.870399824239117e-05, |
|
"loss": 0.9495, |
|
"step": 119000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.86663298624003e-05, |
|
"loss": 0.9387, |
|
"step": 120000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.862812927522309e-05, |
|
"loss": 0.929, |
|
"step": 121000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.858939689861506e-05, |
|
"loss": 0.9516, |
|
"step": 122000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.855013315614725e-05, |
|
"loss": 0.942, |
|
"step": 123000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.851033847720166e-05, |
|
"loss": 0.9519, |
|
"step": 124000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.847001329696653e-05, |
|
"loss": 0.9536, |
|
"step": 125000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.842915805643155e-05, |
|
"loss": 0.949, |
|
"step": 126000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.838777320238312e-05, |
|
"loss": 0.9589, |
|
"step": 127000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.834585918739936e-05, |
|
"loss": 0.9503, |
|
"step": 128000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.830341646984521e-05, |
|
"loss": 0.9494, |
|
"step": 129000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.826044551386744e-05, |
|
"loss": 0.9461, |
|
"step": 130000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.821694678938953e-05, |
|
"loss": 0.9491, |
|
"step": 131000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.817292077210659e-05, |
|
"loss": 0.9504, |
|
"step": 132000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.812836794348004e-05, |
|
"loss": 0.9355, |
|
"step": 133000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.808328879073251e-05, |
|
"loss": 0.9451, |
|
"step": 134000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.803768380684242e-05, |
|
"loss": 0.941, |
|
"step": 135000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.799155349053851e-05, |
|
"loss": 0.944, |
|
"step": 136000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.794489834629455e-05, |
|
"loss": 0.9375, |
|
"step": 137000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.789771888432375e-05, |
|
"loss": 0.9461, |
|
"step": 138000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.785001562057309e-05, |
|
"loss": 0.9458, |
|
"step": 139000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.780178907671789e-05, |
|
"loss": 0.9489, |
|
"step": 140000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.775303978015585e-05, |
|
"loss": 0.9221, |
|
"step": 141000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.77037682640015e-05, |
|
"loss": 0.9333, |
|
"step": 142000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.765397506708023e-05, |
|
"loss": 0.9281, |
|
"step": 143000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.760366073392246e-05, |
|
"loss": 0.9258, |
|
"step": 144000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.755282581475769e-05, |
|
"loss": 0.9445, |
|
"step": 145000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.750147086550844e-05, |
|
"loss": 0.9277, |
|
"step": 146000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.744959644778422e-05, |
|
"loss": 0.9254, |
|
"step": 147000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.739720312887535e-05, |
|
"loss": 0.9234, |
|
"step": 148000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.734429148174675e-05, |
|
"loss": 0.9296, |
|
"step": 149000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.729086208503174e-05, |
|
"loss": 0.9337, |
|
"step": 150000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"eval_loss": 0.8752262592315674, |
|
"eval_runtime": 18.2792, |
|
"eval_samples_per_second": 547.07, |
|
"eval_steps_per_second": 2.188, |
|
"step": 150000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.723691552302562e-05, |
|
"loss": 0.9292, |
|
"step": 151000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.718245238567939e-05, |
|
"loss": 0.9244, |
|
"step": 152000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.712747326859315e-05, |
|
"loss": 0.9281, |
|
"step": 153000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.707197877300974e-05, |
|
"loss": 0.9301, |
|
"step": 154000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.701596950580806e-05, |
|
"loss": 0.9208, |
|
"step": 155000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.695944607949649e-05, |
|
"loss": 0.9191, |
|
"step": 156000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.690240911220618e-05, |
|
"loss": 0.926, |
|
"step": 157000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.684485922768422e-05, |
|
"loss": 0.9271, |
|
"step": 158000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.6786797055287e-05, |
|
"loss": 0.9205, |
|
"step": 159000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.672822322997305e-05, |
|
"loss": 0.9101, |
|
"step": 160000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.66691383922964e-05, |
|
"loss": 0.9243, |
|
"step": 161000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.660954318839933e-05, |
|
"loss": 0.9272, |
|
"step": 162000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.654943827000548e-05, |
|
"loss": 0.9264, |
|
"step": 163000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.648882429441257e-05, |
|
"loss": 0.9186, |
|
"step": 164000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.642770192448536e-05, |
|
"loss": 0.9291, |
|
"step": 165000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.636607182864827e-05, |
|
"loss": 0.9422, |
|
"step": 166000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.630393468087818e-05, |
|
"loss": 0.9203, |
|
"step": 167000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.624129116069694e-05, |
|
"loss": 0.9145, |
|
"step": 168000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.617814195316411e-05, |
|
"loss": 0.9202, |
|
"step": 169000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.611448774886924e-05, |
|
"loss": 0.9239, |
|
"step": 170000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.605032924392457e-05, |
|
"loss": 0.9308, |
|
"step": 171000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.598566713995718e-05, |
|
"loss": 0.9306, |
|
"step": 172000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.59205021441015e-05, |
|
"loss": 0.9246, |
|
"step": 173000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.58548349689915e-05, |
|
"loss": 0.9115, |
|
"step": 174000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.578866633275288e-05, |
|
"loss": 0.9368, |
|
"step": 175000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.572199695899522e-05, |
|
"loss": 0.9156, |
|
"step": 176000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.565482757680415e-05, |
|
"loss": 0.9237, |
|
"step": 177000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.558715892073323e-05, |
|
"loss": 0.9225, |
|
"step": 178000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.551899173079607e-05, |
|
"loss": 0.9257, |
|
"step": 179000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.545032675245813e-05, |
|
"loss": 0.9277, |
|
"step": 180000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.538116473662861e-05, |
|
"loss": 0.924, |
|
"step": 181000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.531150643965223e-05, |
|
"loss": 0.9107, |
|
"step": 182000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.524135262330098e-05, |
|
"loss": 0.9218, |
|
"step": 183000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.517070405476575e-05, |
|
"loss": 0.9278, |
|
"step": 184000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.509956150664796e-05, |
|
"loss": 0.9213, |
|
"step": 185000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.502792575695112e-05, |
|
"loss": 0.911, |
|
"step": 186000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.49557975890723e-05, |
|
"loss": 0.9217, |
|
"step": 187000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.488317779179361e-05, |
|
"loss": 0.9235, |
|
"step": 188000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.481006715927351e-05, |
|
"loss": 0.9318, |
|
"step": 189000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.473646649103818e-05, |
|
"loss": 0.9199, |
|
"step": 190000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.46623765919727e-05, |
|
"loss": 0.9251, |
|
"step": 191000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.458779827231237e-05, |
|
"loss": 0.9345, |
|
"step": 192000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.451273234763371e-05, |
|
"loss": 0.9227, |
|
"step": 193000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.443717963884569e-05, |
|
"loss": 0.9042, |
|
"step": 194000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.43611409721806e-05, |
|
"loss": 0.9233, |
|
"step": 195000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.428461717918511e-05, |
|
"loss": 0.9167, |
|
"step": 196000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.420760909671118e-05, |
|
"loss": 0.9174, |
|
"step": 197000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.413011756690685e-05, |
|
"loss": 0.9274, |
|
"step": 198000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.405214343720707e-05, |
|
"loss": 0.915, |
|
"step": 199000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.397368756032445e-05, |
|
"loss": 0.9106, |
|
"step": 200000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"eval_loss": 0.8557878732681274, |
|
"eval_runtime": 22.93, |
|
"eval_samples_per_second": 436.11, |
|
"eval_steps_per_second": 1.744, |
|
"step": 200000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.389475079423988e-05, |
|
"loss": 0.9179, |
|
"step": 201000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.381533400219318e-05, |
|
"loss": 0.9087, |
|
"step": 202000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.373543805267368e-05, |
|
"loss": 0.9254, |
|
"step": 203000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.365506381941066e-05, |
|
"loss": 0.9171, |
|
"step": 204000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.357421218136386e-05, |
|
"loss": 0.9152, |
|
"step": 205000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.349288402271388e-05, |
|
"loss": 0.9229, |
|
"step": 206000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.341108023285238e-05, |
|
"loss": 0.9087, |
|
"step": 207000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.332880170637252e-05, |
|
"loss": 0.9095, |
|
"step": 208000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.32460493430591e-05, |
|
"loss": 0.9072, |
|
"step": 209000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.316282404787871e-05, |
|
"loss": 0.9093, |
|
"step": 210000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.30791267309698e-05, |
|
"loss": 0.9148, |
|
"step": 211000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.299495830763286e-05, |
|
"loss": 0.9021, |
|
"step": 212000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.291031969832026e-05, |
|
"loss": 0.9069, |
|
"step": 213000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.282521182862629e-05, |
|
"loss": 0.9112, |
|
"step": 214000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.273963562927695e-05, |
|
"loss": 0.914, |
|
"step": 215000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.265359203611987e-05, |
|
"loss": 0.9049, |
|
"step": 216000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.256708199011401e-05, |
|
"loss": 0.9117, |
|
"step": 217000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.248010643731935e-05, |
|
"loss": 0.9158, |
|
"step": 218000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.239266632888659e-05, |
|
"loss": 0.9139, |
|
"step": 219000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.230476262104677e-05, |
|
"loss": 0.8969, |
|
"step": 220000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.221639627510076e-05, |
|
"loss": 0.8961, |
|
"step": 221000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.212756825740873e-05, |
|
"loss": 0.8985, |
|
"step": 222000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.20382795393797e-05, |
|
"loss": 0.8919, |
|
"step": 223000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.194853109746074e-05, |
|
"loss": 0.9034, |
|
"step": 224000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.185832391312644e-05, |
|
"loss": 0.8969, |
|
"step": 225000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.176765897286813e-05, |
|
"loss": 0.8996, |
|
"step": 226000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.167653726818305e-05, |
|
"loss": 0.8878, |
|
"step": 227000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.158495979556358e-05, |
|
"loss": 0.9062, |
|
"step": 228000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.14929275564863e-05, |
|
"loss": 0.9026, |
|
"step": 229000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.140044155740101e-05, |
|
"loss": 0.9016, |
|
"step": 230000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.130750280971978e-05, |
|
"loss": 0.9022, |
|
"step": 231000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.121411232980588e-05, |
|
"loss": 0.9027, |
|
"step": 232000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.112027113896262e-05, |
|
"loss": 0.9063, |
|
"step": 233000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.102598026342222e-05, |
|
"loss": 0.8991, |
|
"step": 234000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.093124073433463e-05, |
|
"loss": 0.886, |
|
"step": 235000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.083605358775612e-05, |
|
"loss": 0.8975, |
|
"step": 236000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.074041986463808e-05, |
|
"loss": 0.9021, |
|
"step": 237000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.064434061081562e-05, |
|
"loss": 0.8998, |
|
"step": 238000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.0547816876996e-05, |
|
"loss": 0.8875, |
|
"step": 239000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.045084971874738e-05, |
|
"loss": 0.902, |
|
"step": 240000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.035344019648702e-05, |
|
"loss": 0.9064, |
|
"step": 241000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.025558937546988e-05, |
|
"loss": 0.9044, |
|
"step": 242000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.015729832577681e-05, |
|
"loss": 0.9037, |
|
"step": 243000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.005856812230304e-05, |
|
"loss": 0.9025, |
|
"step": 244000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 8.995939984474624e-05, |
|
"loss": 0.9233, |
|
"step": 245000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.98597945775948e-05, |
|
"loss": 0.8967, |
|
"step": 246000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.975975341011596e-05, |
|
"loss": 0.8927, |
|
"step": 247000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.965927743634391e-05, |
|
"loss": 0.8981, |
|
"step": 248000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.955836775506776e-05, |
|
"loss": 0.8957, |
|
"step": 249000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.945702546981969e-05, |
|
"loss": 0.8981, |
|
"step": 250000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"eval_loss": 0.85121750831604, |
|
"eval_runtime": 20.5031, |
|
"eval_samples_per_second": 487.732, |
|
"eval_steps_per_second": 1.951, |
|
"step": 250000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.935525168886262e-05, |
|
"loss": 0.907, |
|
"step": 251000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.92530475251784e-05, |
|
"loss": 0.8985, |
|
"step": 252000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.91504140964553e-05, |
|
"loss": 0.8905, |
|
"step": 253000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.90473525250761e-05, |
|
"loss": 0.9049, |
|
"step": 254000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.894386393810563e-05, |
|
"loss": 0.8869, |
|
"step": 255000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.883994946727849e-05, |
|
"loss": 0.9039, |
|
"step": 256000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.873561024898668e-05, |
|
"loss": 0.9061, |
|
"step": 257000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.863084742426719e-05, |
|
"loss": 0.906, |
|
"step": 258000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.852566213878947e-05, |
|
"loss": 0.9074, |
|
"step": 259000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.842005554284296e-05, |
|
"loss": 0.8902, |
|
"step": 260000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.831402879132446e-05, |
|
"loss": 0.8941, |
|
"step": 261000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.820758304372557e-05, |
|
"loss": 0.8893, |
|
"step": 262000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.810071946411989e-05, |
|
"loss": 0.8922, |
|
"step": 263000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.799343922115044e-05, |
|
"loss": 0.8923, |
|
"step": 264000 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 8.788574348801675e-05, |
|
"loss": 0.8796, |
|
"step": 265000 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 8.77776334424621e-05, |
|
"loss": 0.8891, |
|
"step": 266000 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 8.766911026676064e-05, |
|
"loss": 0.8904, |
|
"step": 267000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.756017514770443e-05, |
|
"loss": 0.8777, |
|
"step": 268000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.745082927659047e-05, |
|
"loss": 0.8829, |
|
"step": 269000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.73410738492077e-05, |
|
"loss": 0.8968, |
|
"step": 270000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.723091006582389e-05, |
|
"loss": 0.9007, |
|
"step": 271000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.71203391311725e-05, |
|
"loss": 0.9072, |
|
"step": 272000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.700936225443959e-05, |
|
"loss": 0.9152, |
|
"step": 273000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.689798064925049e-05, |
|
"loss": 0.919, |
|
"step": 274000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.678619553365659e-05, |
|
"loss": 0.9189, |
|
"step": 275000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.6674008130122e-05, |
|
"loss": 0.9009, |
|
"step": 276000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.656141966551019e-05, |
|
"loss": 0.9005, |
|
"step": 277000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.644843137107059e-05, |
|
"loss": 0.9028, |
|
"step": 278000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.633504448242505e-05, |
|
"loss": 0.8946, |
|
"step": 279000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.622126023955446e-05, |
|
"loss": 0.9021, |
|
"step": 280000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.610707988678503e-05, |
|
"loss": 0.8995, |
|
"step": 281000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 8.599250467277483e-05, |
|
"loss": 1.1865, |
|
"step": 282000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.587753585050004e-05, |
|
"loss": 0.9683, |
|
"step": 283000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.576217467724128e-05, |
|
"loss": 0.9102, |
|
"step": 284000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.564642241456986e-05, |
|
"loss": 0.9166, |
|
"step": 285000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.553028032833397e-05, |
|
"loss": 0.9031, |
|
"step": 286000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.541374968864487e-05, |
|
"loss": 0.9031, |
|
"step": 287000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.529683176986295e-05, |
|
"loss": 0.9015, |
|
"step": 288000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.517952785058385e-05, |
|
"loss": 0.9023, |
|
"step": 289000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.506183921362443e-05, |
|
"loss": 0.8973, |
|
"step": 290000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.494376714600878e-05, |
|
"loss": 0.8804, |
|
"step": 291000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 8.482531293895412e-05, |
|
"loss": 0.8831, |
|
"step": 292000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.470647788785665e-05, |
|
"loss": 0.8951, |
|
"step": 293000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.458726329227747e-05, |
|
"loss": 0.8957, |
|
"step": 294000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.44676704559283e-05, |
|
"loss": 0.8782, |
|
"step": 295000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.434770068665723e-05, |
|
"loss": 0.8838, |
|
"step": 296000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.422735529643444e-05, |
|
"loss": 0.8902, |
|
"step": 297000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.410663560133784e-05, |
|
"loss": 0.8823, |
|
"step": 298000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.398554292153866e-05, |
|
"loss": 0.8785, |
|
"step": 299000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.386407858128706e-05, |
|
"loss": 0.8813, |
|
"step": 300000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"eval_loss": 0.8203372955322266, |
|
"eval_runtime": 20.3044, |
|
"eval_samples_per_second": 492.503, |
|
"eval_steps_per_second": 1.97, |
|
"step": 300000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.37422439088976e-05, |
|
"loss": 0.894, |
|
"step": 301000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.362004023673474e-05, |
|
"loss": 0.8804, |
|
"step": 302000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.349746890119826e-05, |
|
"loss": 0.8663, |
|
"step": 303000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.337453124270863e-05, |
|
"loss": 0.8767, |
|
"step": 304000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.32512286056924e-05, |
|
"loss": 0.8697, |
|
"step": 305000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.31275623385675e-05, |
|
"loss": 0.876, |
|
"step": 306000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.300353379372834e-05, |
|
"loss": 0.8752, |
|
"step": 307000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.287914432753123e-05, |
|
"loss": 0.8711, |
|
"step": 308000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.275439530027948e-05, |
|
"loss": 0.8526, |
|
"step": 309000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.262928807620843e-05, |
|
"loss": 0.8794, |
|
"step": 310000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.250382402347065e-05, |
|
"loss": 0.8671, |
|
"step": 311000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.237800451412095e-05, |
|
"loss": 0.8772, |
|
"step": 312000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.225183092410128e-05, |
|
"loss": 0.8786, |
|
"step": 313000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.212530463322583e-05, |
|
"loss": 0.8704, |
|
"step": 314000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.199842702516583e-05, |
|
"loss": 0.8801, |
|
"step": 315000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.18711994874345e-05, |
|
"loss": 0.8728, |
|
"step": 316000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.174362341137177e-05, |
|
"loss": 0.8628, |
|
"step": 317000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.161570019212921e-05, |
|
"loss": 0.8591, |
|
"step": 318000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.148743122865463e-05, |
|
"loss": 0.8793, |
|
"step": 319000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.135881792367686e-05, |
|
"loss": 0.8958, |
|
"step": 320000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.12298616836904e-05, |
|
"loss": 0.8768, |
|
"step": 321000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.110056391894005e-05, |
|
"loss": 0.8882, |
|
"step": 322000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.097092604340542e-05, |
|
"loss": 0.8954, |
|
"step": 323000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.084094947478556e-05, |
|
"loss": 0.9074, |
|
"step": 324000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.07106356344834e-05, |
|
"loss": 0.894, |
|
"step": 325000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.057998594759022e-05, |
|
"loss": 0.8978, |
|
"step": 326000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.044900184287007e-05, |
|
"loss": 0.9071, |
|
"step": 327000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.031768475274413e-05, |
|
"loss": 0.9107, |
|
"step": 328000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.018603611327504e-05, |
|
"loss": 0.8774, |
|
"step": 329000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.005405736415126e-05, |
|
"loss": 0.8975, |
|
"step": 330000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 7.992174994867123e-05, |
|
"loss": 0.8813, |
|
"step": 331000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 7.978911531372765e-05, |
|
"loss": 0.869, |
|
"step": 332000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.965615490979163e-05, |
|
"loss": 0.8809, |
|
"step": 333000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.952287019089685e-05, |
|
"loss": 0.8669, |
|
"step": 334000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.938926261462366e-05, |
|
"loss": 0.8675, |
|
"step": 335000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.925533364208309e-05, |
|
"loss": 0.8748, |
|
"step": 336000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.912108473790092e-05, |
|
"loss": 0.883, |
|
"step": 337000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.898651737020166e-05, |
|
"loss": 0.8941, |
|
"step": 338000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.88516330105925e-05, |
|
"loss": 0.8876, |
|
"step": 339000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.871643313414718e-05, |
|
"loss": 0.8878, |
|
"step": 340000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.858091921938988e-05, |
|
"loss": 0.8908, |
|
"step": 341000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 7.844509274827907e-05, |
|
"loss": 0.8878, |
|
"step": 342000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.830895520619128e-05, |
|
"loss": 0.8865, |
|
"step": 343000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.817250808190483e-05, |
|
"loss": 0.8842, |
|
"step": 344000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.803575286758364e-05, |
|
"loss": 0.8879, |
|
"step": 345000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.789869105876083e-05, |
|
"loss": 0.8964, |
|
"step": 346000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.776132415432234e-05, |
|
"loss": 0.8836, |
|
"step": 347000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.762365365649067e-05, |
|
"loss": 0.8834, |
|
"step": 348000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.748568107080832e-05, |
|
"loss": 0.8907, |
|
"step": 349000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.734740790612136e-05, |
|
"loss": 0.8899, |
|
"step": 350000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"eval_loss": 0.8285813331604004, |
|
"eval_runtime": 24.9856, |
|
"eval_samples_per_second": 400.23, |
|
"eval_steps_per_second": 1.601, |
|
"step": 350000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.720883567456298e-05, |
|
"loss": 0.8799, |
|
"step": 351000 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 7.70699658915369e-05, |
|
"loss": 0.8868, |
|
"step": 352000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.693080007570084e-05, |
|
"loss": 0.895, |
|
"step": 353000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.679133974894983e-05, |
|
"loss": 0.8987, |
|
"step": 354000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.66515864363997e-05, |
|
"loss": 0.8856, |
|
"step": 355000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.651154166637025e-05, |
|
"loss": 0.8851, |
|
"step": 356000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.637120697036866e-05, |
|
"loss": 0.8913, |
|
"step": 357000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.623058388307269e-05, |
|
"loss": 0.8832, |
|
"step": 358000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.608967394231387e-05, |
|
"loss": 0.8883, |
|
"step": 359000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.594847868906076e-05, |
|
"loss": 0.8796, |
|
"step": 360000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.580699966740201e-05, |
|
"loss": 0.8709, |
|
"step": 361000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 7.566523842452958e-05, |
|
"loss": 0.8559, |
|
"step": 362000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.552319651072164e-05, |
|
"loss": 0.8746, |
|
"step": 363000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.538087547932585e-05, |
|
"loss": 0.8677, |
|
"step": 364000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.52382768867422e-05, |
|
"loss": 0.8665, |
|
"step": 365000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.509540229240601e-05, |
|
"loss": 0.8651, |
|
"step": 366000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.495225325877103e-05, |
|
"loss": 0.8723, |
|
"step": 367000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.480883135129211e-05, |
|
"loss": 0.8734, |
|
"step": 368000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.466513813840825e-05, |
|
"loss": 0.8703, |
|
"step": 369000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.452117519152542e-05, |
|
"loss": 0.8547, |
|
"step": 370000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.437694408499933e-05, |
|
"loss": 0.8597, |
|
"step": 371000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 7.423244639611826e-05, |
|
"loss": 0.8681, |
|
"step": 372000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.408768370508576e-05, |
|
"loss": 0.8538, |
|
"step": 373000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.394265759500348e-05, |
|
"loss": 0.8409, |
|
"step": 374000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.379736965185368e-05, |
|
"loss": 0.8512, |
|
"step": 375000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.365182146448205e-05, |
|
"loss": 0.853, |
|
"step": 376000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.350601462458024e-05, |
|
"loss": 0.8569, |
|
"step": 377000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.335995072666848e-05, |
|
"loss": 0.8544, |
|
"step": 378000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.32136313680782e-05, |
|
"loss": 0.8558, |
|
"step": 379000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.30670581489344e-05, |
|
"loss": 0.869, |
|
"step": 380000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.292023267213835e-05, |
|
"loss": 0.8573, |
|
"step": 381000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 7.277315654334997e-05, |
|
"loss": 0.8425, |
|
"step": 382000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.262583137097018e-05, |
|
"loss": 0.8569, |
|
"step": 383000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.247825876612353e-05, |
|
"loss": 0.8471, |
|
"step": 384000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.233044034264034e-05, |
|
"loss": 0.8576, |
|
"step": 385000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.218237771703921e-05, |
|
"loss": 0.8656, |
|
"step": 386000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.203407250850928e-05, |
|
"loss": 0.8553, |
|
"step": 387000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.188552633889259e-05, |
|
"loss": 0.849, |
|
"step": 388000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.173674083266624e-05, |
|
"loss": 0.868, |
|
"step": 389000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.158771761692464e-05, |
|
"loss": 0.8618, |
|
"step": 390000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.143845832136188e-05, |
|
"loss": 0.8703, |
|
"step": 391000 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.128896457825364e-05, |
|
"loss": 0.8697, |
|
"step": 392000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.113923802243957e-05, |
|
"loss": 0.8709, |
|
"step": 393000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.09892802913053e-05, |
|
"loss": 0.8805, |
|
"step": 394000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.083909302476453e-05, |
|
"loss": 0.8711, |
|
"step": 395000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.068867786524116e-05, |
|
"loss": 0.8694, |
|
"step": 396000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.053803645765128e-05, |
|
"loss": 0.8638, |
|
"step": 397000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.038717044938519e-05, |
|
"loss": 0.8712, |
|
"step": 398000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.023608149028937e-05, |
|
"loss": 0.8715, |
|
"step": 399000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.008477123264848e-05, |
|
"loss": 0.8581, |
|
"step": 400000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"eval_loss": 0.814812421798706, |
|
"eval_runtime": 19.4033, |
|
"eval_samples_per_second": 515.376, |
|
"eval_steps_per_second": 2.062, |
|
"step": 400000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 6.993324133116726e-05, |
|
"loss": 0.8688, |
|
"step": 401000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 6.978149344295242e-05, |
|
"loss": 0.8657, |
|
"step": 402000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.962952922749457e-05, |
|
"loss": 0.8678, |
|
"step": 403000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.947735034665002e-05, |
|
"loss": 0.8624, |
|
"step": 404000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.932495846462261e-05, |
|
"loss": 0.8719, |
|
"step": 405000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.917235524794558e-05, |
|
"loss": 0.8697, |
|
"step": 406000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.901954236546323e-05, |
|
"loss": 0.8751, |
|
"step": 407000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.886652148831279e-05, |
|
"loss": 0.8507, |
|
"step": 408000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.871329428990602e-05, |
|
"loss": 0.8562, |
|
"step": 409000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.855986244591104e-05, |
|
"loss": 0.8542, |
|
"step": 410000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.840622763423391e-05, |
|
"loss": 0.8531, |
|
"step": 411000 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 6.825239153500029e-05, |
|
"loss": 0.871, |
|
"step": 412000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.809835583053715e-05, |
|
"loss": 0.8552, |
|
"step": 413000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.794412220535426e-05, |
|
"loss": 0.8555, |
|
"step": 414000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.778969234612584e-05, |
|
"loss": 0.8487, |
|
"step": 415000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.763506794167208e-05, |
|
"loss": 0.8603, |
|
"step": 416000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.748025068294067e-05, |
|
"loss": 0.8627, |
|
"step": 417000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.732524226298841e-05, |
|
"loss": 0.8568, |
|
"step": 418000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.71700443769625e-05, |
|
"loss": 0.8531, |
|
"step": 419000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.701465872208216e-05, |
|
"loss": 0.8598, |
|
"step": 420000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.685908699762002e-05, |
|
"loss": 0.8603, |
|
"step": 421000 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 6.670333090488356e-05, |
|
"loss": 0.854, |
|
"step": 422000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.654739214719641e-05, |
|
"loss": 0.8486, |
|
"step": 423000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.639127242987988e-05, |
|
"loss": 0.8588, |
|
"step": 424000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.623497346023418e-05, |
|
"loss": 0.8589, |
|
"step": 425000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.607849694751977e-05, |
|
"loss": 0.8547, |
|
"step": 426000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.592184460293877e-05, |
|
"loss": 0.8442, |
|
"step": 427000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.576501813961609e-05, |
|
"loss": 0.8554, |
|
"step": 428000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.56080192725808e-05, |
|
"loss": 0.8609, |
|
"step": 429000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.545084971874738e-05, |
|
"loss": 0.861, |
|
"step": 430000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.529351119689688e-05, |
|
"loss": 0.8529, |
|
"step": 431000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 6.513600542765817e-05, |
|
"loss": 0.8636, |
|
"step": 432000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.497833413348909e-05, |
|
"loss": 0.8775, |
|
"step": 433000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.48204990386577e-05, |
|
"loss": 0.8544, |
|
"step": 434000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.466250186922325e-05, |
|
"loss": 0.8497, |
|
"step": 435000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.450434435301751e-05, |
|
"loss": 0.8549, |
|
"step": 436000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.43460282196257e-05, |
|
"loss": 0.8598, |
|
"step": 437000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.418755520036775e-05, |
|
"loss": 0.8671, |
|
"step": 438000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.402892702827916e-05, |
|
"loss": 0.8656, |
|
"step": 439000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.387014543809223e-05, |
|
"loss": 0.8606, |
|
"step": 440000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.371121216621698e-05, |
|
"loss": 0.8489, |
|
"step": 441000 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 6.355212895072223e-05, |
|
"loss": 0.8729, |
|
"step": 442000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.339289753131649e-05, |
|
"loss": 0.8528, |
|
"step": 443000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.323351964932908e-05, |
|
"loss": 0.8608, |
|
"step": 444000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.307399704769099e-05, |
|
"loss": 0.8586, |
|
"step": 445000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.291433147091583e-05, |
|
"loss": 0.8612, |
|
"step": 446000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.275452466508077e-05, |
|
"loss": 0.8653, |
|
"step": 447000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.259457837780742e-05, |
|
"loss": 0.8611, |
|
"step": 448000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.243449435824276e-05, |
|
"loss": 0.8493, |
|
"step": 449000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.227427435703997e-05, |
|
"loss": 0.856, |
|
"step": 450000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"eval_loss": 0.8140913248062134, |
|
"eval_runtime": 19.7177, |
|
"eval_samples_per_second": 507.158, |
|
"eval_steps_per_second": 2.029, |
|
"step": 450000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.211392012633932e-05, |
|
"loss": 0.8644, |
|
"step": 451000 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 6.195343341974899e-05, |
|
"loss": 0.8581, |
|
"step": 452000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.179281599232591e-05, |
|
"loss": 0.8496, |
|
"step": 453000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.163206960055651e-05, |
|
"loss": 0.8606, |
|
"step": 454000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.147119600233758e-05, |
|
"loss": 0.8617, |
|
"step": 455000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.131019695695702e-05, |
|
"loss": 0.8672, |
|
"step": 456000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.11490742250746e-05, |
|
"loss": 0.8586, |
|
"step": 457000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.0987829568702656e-05, |
|
"loss": 0.8638, |
|
"step": 458000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.0826464751186994e-05, |
|
"loss": 0.8752, |
|
"step": 459000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.066498153718735e-05, |
|
"loss": 0.8642, |
|
"step": 460000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.05033816926583e-05, |
|
"loss": 0.8442, |
|
"step": 461000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 6.034166698482984e-05, |
|
"loss": 0.8631, |
|
"step": 462000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 6.017983918218812e-05, |
|
"loss": 0.8578, |
|
"step": 463000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 6.001790005445607e-05, |
|
"loss": 0.858, |
|
"step": 464000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 5.985585137257401e-05, |
|
"loss": 0.8688, |
|
"step": 465000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 5.969369490868042e-05, |
|
"loss": 0.8544, |
|
"step": 466000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 5.953143243609235e-05, |
|
"loss": 0.8526, |
|
"step": 467000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 5.9369065729286245e-05, |
|
"loss": 0.8589, |
|
"step": 468000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 5.9206596563878357e-05, |
|
"loss": 0.8495, |
|
"step": 469000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 5.90440267166055e-05, |
|
"loss": 0.8663, |
|
"step": 470000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 5.888135796530544e-05, |
|
"loss": 0.8584, |
|
"step": 471000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 5.871859208889759e-05, |
|
"loss": 0.8562, |
|
"step": 472000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.85557308673635e-05, |
|
"loss": 0.8633, |
|
"step": 473000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.8392776081727385e-05, |
|
"loss": 0.8548, |
|
"step": 474000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.8229729514036705e-05, |
|
"loss": 0.8517, |
|
"step": 475000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.8066592947342555e-05, |
|
"loss": 0.8483, |
|
"step": 476000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.7903368165680327e-05, |
|
"loss": 0.8502, |
|
"step": 477000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.7740056954050084e-05, |
|
"loss": 0.8586, |
|
"step": 478000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.757666109839702e-05, |
|
"loss": 0.8469, |
|
"step": 479000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.74131823855921e-05, |
|
"loss": 0.8479, |
|
"step": 480000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.72496226034123e-05, |
|
"loss": 0.8543, |
|
"step": 481000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 5.7085983540521216e-05, |
|
"loss": 0.8569, |
|
"step": 482000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.692226698644938e-05, |
|
"loss": 0.8494, |
|
"step": 483000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.675847473157485e-05, |
|
"loss": 0.8551, |
|
"step": 484000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.6594608567103456e-05, |
|
"loss": 0.8589, |
|
"step": 485000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.6430670285049314e-05, |
|
"loss": 0.8582, |
|
"step": 486000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.6266661678215216e-05, |
|
"loss": 0.8411, |
|
"step": 487000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.6102584540173006e-05, |
|
"loss": 0.8393, |
|
"step": 488000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.5938440665244006e-05, |
|
"loss": 0.8417, |
|
"step": 489000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.577423184847932e-05, |
|
"loss": 0.8366, |
|
"step": 490000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.560995988564023e-05, |
|
"loss": 0.8476, |
|
"step": 491000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 5.544562657317863e-05, |
|
"loss": 0.8414, |
|
"step": 492000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.52812337082173e-05, |
|
"loss": 0.844, |
|
"step": 493000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.511678308853026e-05, |
|
"loss": 0.8339, |
|
"step": 494000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.495227651252315e-05, |
|
"loss": 0.8524, |
|
"step": 495000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.478771577921351e-05, |
|
"loss": 0.845, |
|
"step": 496000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.462310268821118e-05, |
|
"loss": 0.8464, |
|
"step": 497000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.445843903969854e-05, |
|
"loss": 0.8471, |
|
"step": 498000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.4293726634410855e-05, |
|
"loss": 0.8498, |
|
"step": 499000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.4128967273616625e-05, |
|
"loss": 0.8527, |
|
"step": 500000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"eval_loss": 0.8033929467201233, |
|
"eval_runtime": 18.0845, |
|
"eval_samples_per_second": 552.961, |
|
"eval_steps_per_second": 2.212, |
|
"step": 500000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.396416275909779e-05, |
|
"loss": 0.8443, |
|
"step": 501000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 5.379931489313016e-05, |
|
"loss": 0.8343, |
|
"step": 502000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.363442547846356e-05, |
|
"loss": 0.8436, |
|
"step": 503000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.3469496318302204e-05, |
|
"loss": 0.8475, |
|
"step": 504000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.330452921628497e-05, |
|
"loss": 0.8455, |
|
"step": 505000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.313952597646568e-05, |
|
"loss": 0.8341, |
|
"step": 506000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.297448840329329e-05, |
|
"loss": 0.8485, |
|
"step": 507000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.280941830159227e-05, |
|
"loss": 0.8544, |
|
"step": 508000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.264431747654284e-05, |
|
"loss": 0.8527, |
|
"step": 509000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.247918773366112e-05, |
|
"loss": 0.851, |
|
"step": 510000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.231403087877955e-05, |
|
"loss": 0.8478, |
|
"step": 511000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 5.214884871802703e-05, |
|
"loss": 0.8693, |
|
"step": 512000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.198364305780922e-05, |
|
"loss": 0.8444, |
|
"step": 513000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.1818415704788725e-05, |
|
"loss": 0.8386, |
|
"step": 514000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.165316846586541e-05, |
|
"loss": 0.846, |
|
"step": 515000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.148790314815663e-05, |
|
"loss": 0.8417, |
|
"step": 516000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.132262155897739e-05, |
|
"loss": 0.8462, |
|
"step": 517000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.1157325505820694e-05, |
|
"loss": 0.8549, |
|
"step": 518000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.0992016796337686e-05, |
|
"loss": 0.8453, |
|
"step": 519000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.0826697238317935e-05, |
|
"loss": 0.8389, |
|
"step": 520000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.066136863966963e-05, |
|
"loss": 0.8524, |
|
"step": 521000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.0496032808399815e-05, |
|
"loss": 0.8352, |
|
"step": 522000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 5.033069155259471e-05, |
|
"loss": 0.8515, |
|
"step": 523000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 5.016534668039976e-05, |
|
"loss": 0.8509, |
|
"step": 524000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 5e-05, |
|
"loss": 0.854, |
|
"step": 525000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 4.9834653319600246e-05, |
|
"loss": 0.8569, |
|
"step": 526000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 4.96693084474053e-05, |
|
"loss": 0.8376, |
|
"step": 527000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 4.950396719160018e-05, |
|
"loss": 0.8403, |
|
"step": 528000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 4.93386313603304e-05, |
|
"loss": 0.838, |
|
"step": 529000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 4.917330276168208e-05, |
|
"loss": 0.8419, |
|
"step": 530000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 4.9007983203662326e-05, |
|
"loss": 0.8413, |
|
"step": 531000 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 4.884267449417931e-05, |
|
"loss": 0.8292, |
|
"step": 532000 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 4.867737844102261e-05, |
|
"loss": 0.8377, |
|
"step": 533000 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 4.851209685184338e-05, |
|
"loss": 0.8381, |
|
"step": 534000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 4.834683153413459e-05, |
|
"loss": 0.8306, |
|
"step": 535000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 4.818158429521129e-05, |
|
"loss": 0.8309, |
|
"step": 536000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 4.801635694219079e-05, |
|
"loss": 0.8476, |
|
"step": 537000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 4.785115128197298e-05, |
|
"loss": 0.8494, |
|
"step": 538000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 4.7685969121220456e-05, |
|
"loss": 0.856, |
|
"step": 539000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.7520812266338885e-05, |
|
"loss": 0.8639, |
|
"step": 540000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.735568252345718e-05, |
|
"loss": 0.8669, |
|
"step": 541000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.7190581698407725e-05, |
|
"loss": 0.8697, |
|
"step": 542000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.702551159670672e-05, |
|
"loss": 0.852, |
|
"step": 543000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.6860474023534335e-05, |
|
"loss": 0.848, |
|
"step": 544000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.669547078371504e-05, |
|
"loss": 0.8515, |
|
"step": 545000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.65305036816978e-05, |
|
"loss": 0.8409, |
|
"step": 546000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.6365574521536445e-05, |
|
"loss": 0.8486, |
|
"step": 547000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.620068510686985e-05, |
|
"loss": 0.8464, |
|
"step": 548000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 4.60358372409022e-05, |
|
"loss": 0.8405, |
|
"step": 549000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.5871032726383386e-05, |
|
"loss": 0.8345, |
|
"step": 550000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"eval_loss": 0.7762564420700073, |
|
"eval_runtime": 18.813, |
|
"eval_samples_per_second": 531.546, |
|
"eval_steps_per_second": 2.126, |
|
"step": 550000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.570627336558915e-05, |
|
"loss": 0.8509, |
|
"step": 551000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.554156096030149e-05, |
|
"loss": 0.8601, |
|
"step": 552000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.537689731178883e-05, |
|
"loss": 0.8469, |
|
"step": 553000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.5212284220786494e-05, |
|
"loss": 0.8494, |
|
"step": 554000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.504772348747687e-05, |
|
"loss": 0.8485, |
|
"step": 555000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.488321691146975e-05, |
|
"loss": 0.849, |
|
"step": 556000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.471876629178273e-05, |
|
"loss": 0.8428, |
|
"step": 557000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.4554373426821374e-05, |
|
"loss": 0.8278, |
|
"step": 558000 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 4.439004011435979e-05, |
|
"loss": 0.8304, |
|
"step": 559000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.4225768151520694e-05, |
|
"loss": 0.8436, |
|
"step": 560000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.406155933475599e-05, |
|
"loss": 0.8446, |
|
"step": 561000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.3897415459827e-05, |
|
"loss": 0.8262, |
|
"step": 562000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.373333832178478e-05, |
|
"loss": 0.8311, |
|
"step": 563000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.3569329714950704e-05, |
|
"loss": 0.8394, |
|
"step": 564000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.3405391432896555e-05, |
|
"loss": 0.8317, |
|
"step": 565000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.324152526842517e-05, |
|
"loss": 0.8282, |
|
"step": 566000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.307773301355062e-05, |
|
"loss": 0.8314, |
|
"step": 567000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.291401645947879e-05, |
|
"loss": 0.8431, |
|
"step": 568000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 4.275037739658771e-05, |
|
"loss": 0.8323, |
|
"step": 569000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.2586817614407895e-05, |
|
"loss": 0.8178, |
|
"step": 570000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.2423338901602985e-05, |
|
"loss": 0.8289, |
|
"step": 571000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.2259943045949934e-05, |
|
"loss": 0.8205, |
|
"step": 572000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.209663183431969e-05, |
|
"loss": 0.8272, |
|
"step": 573000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.1933407052657456e-05, |
|
"loss": 0.8261, |
|
"step": 574000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.17702704859633e-05, |
|
"loss": 0.8221, |
|
"step": 575000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.160722391827262e-05, |
|
"loss": 0.8054, |
|
"step": 576000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.14442691326365e-05, |
|
"loss": 0.8287, |
|
"step": 577000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.1281407911102425e-05, |
|
"loss": 0.8198, |
|
"step": 578000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 4.111864203469457e-05, |
|
"loss": 0.8288, |
|
"step": 579000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 4.095597328339452e-05, |
|
"loss": 0.8283, |
|
"step": 580000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 4.079340343612165e-05, |
|
"loss": 0.8207, |
|
"step": 581000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 4.063093427071376e-05, |
|
"loss": 0.8323, |
|
"step": 582000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 4.046856756390767e-05, |
|
"loss": 0.8222, |
|
"step": 583000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 4.0306305091319595e-05, |
|
"loss": 0.8173, |
|
"step": 584000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 4.0144148627425993e-05, |
|
"loss": 0.8107, |
|
"step": 585000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 3.9982099945543945e-05, |
|
"loss": 0.83, |
|
"step": 586000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 3.982016081781189e-05, |
|
"loss": 0.8479, |
|
"step": 587000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 3.965833301517017e-05, |
|
"loss": 0.8299, |
|
"step": 588000 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 3.949661830734172e-05, |
|
"loss": 0.841, |
|
"step": 589000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.933501846281267e-05, |
|
"loss": 0.8463, |
|
"step": 590000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.917353524881302e-05, |
|
"loss": 0.8593, |
|
"step": 591000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.901217043129735e-05, |
|
"loss": 0.8489, |
|
"step": 592000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.8850925774925425e-05, |
|
"loss": 0.851, |
|
"step": 593000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.8689803043043e-05, |
|
"loss": 0.8586, |
|
"step": 594000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.852880399766243e-05, |
|
"loss": 0.8641, |
|
"step": 595000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.836793039944349e-05, |
|
"loss": 0.8324, |
|
"step": 596000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.820718400767409e-05, |
|
"loss": 0.8486, |
|
"step": 597000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.8046566580251e-05, |
|
"loss": 0.8359, |
|
"step": 598000 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 3.788607987366069e-05, |
|
"loss": 0.8224, |
|
"step": 599000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.772572564296005e-05, |
|
"loss": 0.8342, |
|
"step": 600000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"eval_loss": 0.7862029075622559, |
|
"eval_runtime": 19.1548, |
|
"eval_samples_per_second": 522.064, |
|
"eval_steps_per_second": 2.088, |
|
"step": 600000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.756550564175727e-05, |
|
"loss": 0.8199, |
|
"step": 601000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.74054216221926e-05, |
|
"loss": 0.8218, |
|
"step": 602000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.7245475334919246e-05, |
|
"loss": 0.8264, |
|
"step": 603000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.7085668529084184e-05, |
|
"loss": 0.8367, |
|
"step": 604000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.6926002952309016e-05, |
|
"loss": 0.8472, |
|
"step": 605000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.676648035067093e-05, |
|
"loss": 0.8408, |
|
"step": 606000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.6607102468683526e-05, |
|
"loss": 0.8405, |
|
"step": 607000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.6447871049277796e-05, |
|
"loss": 0.8456, |
|
"step": 608000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 3.628878783378302e-05, |
|
"loss": 0.8433, |
|
"step": 609000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.612985456190778e-05, |
|
"loss": 0.8419, |
|
"step": 610000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.597107297172084e-05, |
|
"loss": 0.8392, |
|
"step": 611000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.581244479963225e-05, |
|
"loss": 0.8435, |
|
"step": 612000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.5653971780374295e-05, |
|
"loss": 0.8529, |
|
"step": 613000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.5495655646982505e-05, |
|
"loss": 0.8393, |
|
"step": 614000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.533749813077677e-05, |
|
"loss": 0.8389, |
|
"step": 615000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.517950096134232e-05, |
|
"loss": 0.8446, |
|
"step": 616000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.5021665866510925e-05, |
|
"loss": 0.8457, |
|
"step": 617000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.4863994572341843e-05, |
|
"loss": 0.8369, |
|
"step": 618000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 3.470648880310313e-05, |
|
"loss": 0.8423, |
|
"step": 619000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.4549150281252636e-05, |
|
"loss": 0.8498, |
|
"step": 620000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.439198072741921e-05, |
|
"loss": 0.8559, |
|
"step": 621000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.423498186038393e-05, |
|
"loss": 0.8429, |
|
"step": 622000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.407815539706124e-05, |
|
"loss": 0.8389, |
|
"step": 623000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.392150305248024e-05, |
|
"loss": 0.8478, |
|
"step": 624000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.3765026539765834e-05, |
|
"loss": 0.8383, |
|
"step": 625000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.360872757012011e-05, |
|
"loss": 0.842, |
|
"step": 626000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.3452607852803584e-05, |
|
"loss": 0.8347, |
|
"step": 627000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.329666909511645e-05, |
|
"loss": 0.8244, |
|
"step": 628000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 3.3140913002379995e-05, |
|
"loss": 0.8104, |
|
"step": 629000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.298534127791785e-05, |
|
"loss": 0.8305, |
|
"step": 630000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.282995562303754e-05, |
|
"loss": 0.8216, |
|
"step": 631000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.267475773701161e-05, |
|
"loss": 0.8278, |
|
"step": 632000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.251974931705933e-05, |
|
"loss": 0.8232, |
|
"step": 633000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.236493205832795e-05, |
|
"loss": 0.8284, |
|
"step": 634000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.221030765387417e-05, |
|
"loss": 0.8301, |
|
"step": 635000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.205587779464576e-05, |
|
"loss": 0.825, |
|
"step": 636000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.190164416946285e-05, |
|
"loss": 0.8115, |
|
"step": 637000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.1747608464999725e-05, |
|
"loss": 0.8152, |
|
"step": 638000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 3.1593772365766105e-05, |
|
"loss": 0.8229, |
|
"step": 639000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.144013755408895e-05, |
|
"loss": 0.8114, |
|
"step": 640000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.128670571009399e-05, |
|
"loss": 0.7985, |
|
"step": 641000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.113347851168721e-05, |
|
"loss": 0.808, |
|
"step": 642000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.098045763453678e-05, |
|
"loss": 0.8086, |
|
"step": 643000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.082764475205442e-05, |
|
"loss": 0.8143, |
|
"step": 644000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.0675041535377405e-05, |
|
"loss": 0.8113, |
|
"step": 645000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.052264965335e-05, |
|
"loss": 0.8124, |
|
"step": 646000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.0370470772505433e-05, |
|
"loss": 0.8249, |
|
"step": 647000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.0218506557047598e-05, |
|
"loss": 0.8151, |
|
"step": 648000 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 3.006675866883275e-05, |
|
"loss": 0.7981, |
|
"step": 649000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.991522876735154e-05, |
|
"loss": 0.8147, |
|
"step": 650000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"eval_loss": 0.7841782569885254, |
|
"eval_runtime": 16.8585, |
|
"eval_samples_per_second": 593.171, |
|
"eval_steps_per_second": 2.373, |
|
"step": 650000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.976391850971065e-05, |
|
"loss": 0.8032, |
|
"step": 651000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.9612829550614836e-05, |
|
"loss": 0.814, |
|
"step": 652000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.9461963542348737e-05, |
|
"loss": 0.8232, |
|
"step": 653000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.931132213475884e-05, |
|
"loss": 0.8142, |
|
"step": 654000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.916090697523549e-05, |
|
"loss": 0.8054, |
|
"step": 655000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.9010719708694722e-05, |
|
"loss": 0.8242, |
|
"step": 656000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.8860761977560436e-05, |
|
"loss": 0.8183, |
|
"step": 657000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.8711035421746367e-05, |
|
"loss": 0.8271, |
|
"step": 658000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.8561541678638142e-05, |
|
"loss": 0.8261, |
|
"step": 659000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.8412282383075363e-05, |
|
"loss": 0.8267, |
|
"step": 660000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.8263259167333777e-05, |
|
"loss": 0.8353, |
|
"step": 661000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.811447366110741e-05, |
|
"loss": 0.8273, |
|
"step": 662000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.7965927491490705e-05, |
|
"loss": 0.8259, |
|
"step": 663000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.7817622282960815e-05, |
|
"loss": 0.8215, |
|
"step": 664000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.766955965735968e-05, |
|
"loss": 0.8273, |
|
"step": 665000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.7521741233876496e-05, |
|
"loss": 0.8289, |
|
"step": 666000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.7374168629029813e-05, |
|
"loss": 0.8166, |
|
"step": 667000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.7226843456650037e-05, |
|
"loss": 0.8251, |
|
"step": 668000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.707976732786166e-05, |
|
"loss": 0.8231, |
|
"step": 669000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.693294185106562e-05, |
|
"loss": 0.8247, |
|
"step": 670000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.6786368631921836e-05, |
|
"loss": 0.8201, |
|
"step": 671000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.6640049273331515e-05, |
|
"loss": 0.8287, |
|
"step": 672000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.6493985375419778e-05, |
|
"loss": 0.828, |
|
"step": 673000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.6348178535517966e-05, |
|
"loss": 0.8333, |
|
"step": 674000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.6202630348146324e-05, |
|
"loss": 0.8111, |
|
"step": 675000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.6057342404996522e-05, |
|
"loss": 0.8129, |
|
"step": 676000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.591231629491423e-05, |
|
"loss": 0.8129, |
|
"step": 677000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.5767553603881767e-05, |
|
"loss": 0.8126, |
|
"step": 678000 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.562305591500069e-05, |
|
"loss": 0.8278, |
|
"step": 679000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.547882480847461e-05, |
|
"loss": 0.8143, |
|
"step": 680000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.5334861861591753e-05, |
|
"loss": 0.8143, |
|
"step": 681000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.5191168648707887e-05, |
|
"loss": 0.8076, |
|
"step": 682000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.5047746741228978e-05, |
|
"loss": 0.8209, |
|
"step": 683000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.490459770759398e-05, |
|
"loss": 0.8208, |
|
"step": 684000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.476172311325783e-05, |
|
"loss": 0.8155, |
|
"step": 685000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.4619124520674146e-05, |
|
"loss": 0.8124, |
|
"step": 686000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.447680348927837e-05, |
|
"loss": 0.8197, |
|
"step": 687000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.433476157547044e-05, |
|
"loss": 0.8197, |
|
"step": 688000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.419300033259798e-05, |
|
"loss": 0.8159, |
|
"step": 689000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.405152131093926e-05, |
|
"loss": 0.807, |
|
"step": 690000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.3910326057686127e-05, |
|
"loss": 0.8181, |
|
"step": 691000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.3769416116927335e-05, |
|
"loss": 0.819, |
|
"step": 692000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.362879302963135e-05, |
|
"loss": 0.8163, |
|
"step": 693000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.3488458333629777e-05, |
|
"loss": 0.8037, |
|
"step": 694000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.3348413563600325e-05, |
|
"loss": 0.8144, |
|
"step": 695000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.3208660251050158e-05, |
|
"loss": 0.821, |
|
"step": 696000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.3069199924299174e-05, |
|
"loss": 0.8205, |
|
"step": 697000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.29300341084631e-05, |
|
"loss": 0.8125, |
|
"step": 698000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.279116432543705e-05, |
|
"loss": 0.8239, |
|
"step": 699000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.2652592093878666e-05, |
|
"loss": 0.8369, |
|
"step": 700000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"eval_loss": 0.7765893340110779, |
|
"eval_runtime": 17.8732, |
|
"eval_samples_per_second": 559.497, |
|
"eval_steps_per_second": 2.238, |
|
"step": 700000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.251431892919171e-05, |
|
"loss": 0.8155, |
|
"step": 701000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.237634634350934e-05, |
|
"loss": 0.8093, |
|
"step": 702000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.2238675845677663e-05, |
|
"loss": 0.8162, |
|
"step": 703000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.2101308941239203e-05, |
|
"loss": 0.8201, |
|
"step": 704000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.196424713241637e-05, |
|
"loss": 0.8262, |
|
"step": 705000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.182749191809518e-05, |
|
"loss": 0.8267, |
|
"step": 706000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.1691044793808734e-05, |
|
"loss": 0.8208, |
|
"step": 707000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.1554907251720945e-05, |
|
"loss": 0.8085, |
|
"step": 708000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 2.1419080780610123e-05, |
|
"loss": 0.8328, |
|
"step": 709000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.128356686585282e-05, |
|
"loss": 0.8145, |
|
"step": 710000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.1148366989407496e-05, |
|
"loss": 0.8211, |
|
"step": 711000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.1013482629798333e-05, |
|
"loss": 0.8198, |
|
"step": 712000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.0878915262099098e-05, |
|
"loss": 0.8227, |
|
"step": 713000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.0744666357916925e-05, |
|
"loss": 0.827, |
|
"step": 714000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.061073738537635e-05, |
|
"loss": 0.8217, |
|
"step": 715000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.0477129809103147e-05, |
|
"loss": 0.8122, |
|
"step": 716000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.0343845090208368e-05, |
|
"loss": 0.8172, |
|
"step": 717000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.0210884686272368e-05, |
|
"loss": 0.8255, |
|
"step": 718000 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.0078250051328784e-05, |
|
"loss": 0.8193, |
|
"step": 719000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.9945942635848748e-05, |
|
"loss": 0.8105, |
|
"step": 720000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.981396388672496e-05, |
|
"loss": 0.8216, |
|
"step": 721000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.9682315247255894e-05, |
|
"loss": 0.8232, |
|
"step": 722000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.9550998157129946e-05, |
|
"loss": 0.8302, |
|
"step": 723000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.942001405240979e-05, |
|
"loss": 0.82, |
|
"step": 724000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.928936436551661e-05, |
|
"loss": 0.8264, |
|
"step": 725000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.9159050525214452e-05, |
|
"loss": 0.8356, |
|
"step": 726000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.9029073956594606e-05, |
|
"loss": 0.8271, |
|
"step": 727000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.8899436081059975e-05, |
|
"loss": 0.8051, |
|
"step": 728000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.877013831630961e-05, |
|
"loss": 0.8247, |
|
"step": 729000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.8641182076323148e-05, |
|
"loss": 0.8198, |
|
"step": 730000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.851256877134538e-05, |
|
"loss": 0.8189, |
|
"step": 731000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.838429980787081e-05, |
|
"loss": 0.8316, |
|
"step": 732000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.8256376588628238e-05, |
|
"loss": 0.817, |
|
"step": 733000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.8128800512565513e-05, |
|
"loss": 0.8169, |
|
"step": 734000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.800157297483417e-05, |
|
"loss": 0.821, |
|
"step": 735000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.787469536677419e-05, |
|
"loss": 0.8132, |
|
"step": 736000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.774816907589873e-05, |
|
"loss": 0.8282, |
|
"step": 737000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.7621995485879062e-05, |
|
"loss": 0.8211, |
|
"step": 738000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.749617597652934e-05, |
|
"loss": 0.8182, |
|
"step": 739000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.7370711923791567e-05, |
|
"loss": 0.8241, |
|
"step": 740000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.7245604699720535e-05, |
|
"loss": 0.8184, |
|
"step": 741000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.712085567246878e-05, |
|
"loss": 0.813, |
|
"step": 742000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.699646620627168e-05, |
|
"loss": 0.8109, |
|
"step": 743000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.6872437661432517e-05, |
|
"loss": 0.8137, |
|
"step": 744000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.6748771394307585e-05, |
|
"loss": 0.8216, |
|
"step": 745000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.662546875729138e-05, |
|
"loss": 0.81, |
|
"step": 746000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.6502531098801753e-05, |
|
"loss": 0.8109, |
|
"step": 747000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.637995976326527e-05, |
|
"loss": 0.8173, |
|
"step": 748000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.62577560911024e-05, |
|
"loss": 0.8197, |
|
"step": 749000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.6135921418712956e-05, |
|
"loss": 0.814, |
|
"step": 750000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"eval_loss": 0.773684024810791, |
|
"eval_runtime": 18.4635, |
|
"eval_samples_per_second": 541.609, |
|
"eval_steps_per_second": 2.166, |
|
"step": 750000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.6014457078461353e-05, |
|
"loss": 0.8177, |
|
"step": 751000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.5893364398662176e-05, |
|
"loss": 0.8223, |
|
"step": 752000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.5772644703565565e-05, |
|
"loss": 0.8236, |
|
"step": 753000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.5652299313342773e-05, |
|
"loss": 0.8056, |
|
"step": 754000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.553232954407171e-05, |
|
"loss": 0.802, |
|
"step": 755000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.5412736707722537e-05, |
|
"loss": 0.806, |
|
"step": 756000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.5293522112143373e-05, |
|
"loss": 0.8013, |
|
"step": 757000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.517468706104589e-05, |
|
"loss": 0.8098, |
|
"step": 758000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.5056232853991209e-05, |
|
"loss": 0.8062, |
|
"step": 759000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.4938160786375572e-05, |
|
"loss": 0.8078, |
|
"step": 760000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.4820472149416154e-05, |
|
"loss": 0.7976, |
|
"step": 761000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.470316823013707e-05, |
|
"loss": 0.8175, |
|
"step": 762000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.4586250311355132e-05, |
|
"loss": 0.8095, |
|
"step": 763000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.4469719671666043e-05, |
|
"loss": 0.8105, |
|
"step": 764000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.435357758543015e-05, |
|
"loss": 0.8103, |
|
"step": 765000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.4237825322758736e-05, |
|
"loss": 0.8153, |
|
"step": 766000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.412246414949997e-05, |
|
"loss": 0.8184, |
|
"step": 767000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.4007495327225162e-05, |
|
"loss": 0.8083, |
|
"step": 768000 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.389292011321498e-05, |
|
"loss": 0.7992, |
|
"step": 769000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.3778739760445552e-05, |
|
"loss": 0.8087, |
|
"step": 770000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.3664955517574968e-05, |
|
"loss": 0.8118, |
|
"step": 771000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.3551568628929434e-05, |
|
"loss": 0.8096, |
|
"step": 772000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.343858033448982e-05, |
|
"loss": 0.7987, |
|
"step": 773000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.3325991869878013e-05, |
|
"loss": 0.8141, |
|
"step": 774000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.3213804466343421e-05, |
|
"loss": 0.819, |
|
"step": 775000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.3102019350749528e-05, |
|
"loss": 0.8183, |
|
"step": 776000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.299063774556042e-05, |
|
"loss": 0.8153, |
|
"step": 777000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.2879660868827508e-05, |
|
"loss": 0.8135, |
|
"step": 778000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.2769089934176126e-05, |
|
"loss": 0.835, |
|
"step": 779000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.2658926150792322e-05, |
|
"loss": 0.8122, |
|
"step": 780000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.2549170723409549e-05, |
|
"loss": 0.8031, |
|
"step": 781000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.243982485229559e-05, |
|
"loss": 0.8134, |
|
"step": 782000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.233088973323937e-05, |
|
"loss": 0.8082, |
|
"step": 783000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.2222366557537911e-05, |
|
"loss": 0.8139, |
|
"step": 784000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.2114256511983274e-05, |
|
"loss": 0.8216, |
|
"step": 785000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.2006560778849578e-05, |
|
"loss": 0.8128, |
|
"step": 786000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.1899280535880119e-05, |
|
"loss": 0.8055, |
|
"step": 787000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.1792416956274444e-05, |
|
"loss": 0.8176, |
|
"step": 788000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.1685971208675539e-05, |
|
"loss": 0.8019, |
|
"step": 789000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.157994445715706e-05, |
|
"loss": 0.8173, |
|
"step": 790000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.1474337861210543e-05, |
|
"loss": 0.815, |
|
"step": 791000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.1369152575732822e-05, |
|
"loss": 0.8212, |
|
"step": 792000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.1264389751013326e-05, |
|
"loss": 0.8235, |
|
"step": 793000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.1160050532721528e-05, |
|
"loss": 0.8061, |
|
"step": 794000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.1056136061894384e-05, |
|
"loss": 0.8078, |
|
"step": 795000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.095264747492391e-05, |
|
"loss": 0.8045, |
|
"step": 796000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.0849585903544706e-05, |
|
"loss": 0.809, |
|
"step": 797000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.0746952474821614e-05, |
|
"loss": 0.808, |
|
"step": 798000 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.0644748311137376e-05, |
|
"loss": 0.7971, |
|
"step": 799000 |
|
}, |
|
{ |
|
"epoch": 2.27, |
|
"learning_rate": 1.0542974530180327e-05, |
|
"loss": 0.8046, |
|
"step": 800000 |
|
}, |
|
{ |
|
"epoch": 2.27, |
|
"eval_loss": 0.7692430019378662, |
|
"eval_runtime": 22.2242, |
|
"eval_samples_per_second": 449.959, |
|
"eval_steps_per_second": 1.8, |
|
"step": 800000 |
|
} |
|
], |
|
"max_steps": 1000000, |
|
"num_train_epochs": 9223372036854775807, |
|
"total_flos": 1.34791500398592e+19, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|