|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 3.0, |
|
"global_step": 58011, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.956904724965955e-05, |
|
"loss": 1.7071, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.91380944993191e-05, |
|
"loss": 1.6086, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 4.8707141748978643e-05, |
|
"loss": 1.5292, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 4.8276188998638195e-05, |
|
"loss": 1.475, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 4.784523624829774e-05, |
|
"loss": 1.4358, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 4.7414283497957284e-05, |
|
"loss": 1.4158, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 4.6983330747616836e-05, |
|
"loss": 1.3844, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.655237799727638e-05, |
|
"loss": 1.3696, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 4.6121425246935925e-05, |
|
"loss": 1.33, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 4.569047249659548e-05, |
|
"loss": 1.2824, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.525951974625502e-05, |
|
"loss": 1.2842, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 4.482856699591457e-05, |
|
"loss": 1.282, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 4.439761424557412e-05, |
|
"loss": 1.2593, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.396666149523366e-05, |
|
"loss": 1.238, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 4.3535708744893214e-05, |
|
"loss": 1.2322, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 4.310475599455276e-05, |
|
"loss": 1.2107, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.2673803244212304e-05, |
|
"loss": 1.2007, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.2242850493871855e-05, |
|
"loss": 1.1947, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.18118977435314e-05, |
|
"loss": 1.1792, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 4.138094499319095e-05, |
|
"loss": 1.1771, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.0949992242850496e-05, |
|
"loss": 1.1634, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 4.051903949251004e-05, |
|
"loss": 1.1487, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 4.008808674216959e-05, |
|
"loss": 1.1607, |
|
"step": 11500 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 3.965713399182914e-05, |
|
"loss": 1.1333, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 3.922618124148869e-05, |
|
"loss": 1.1344, |
|
"step": 12500 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 3.8795228491148233e-05, |
|
"loss": 1.1319, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 3.836427574080778e-05, |
|
"loss": 1.1236, |
|
"step": 13500 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 3.793332299046733e-05, |
|
"loss": 1.1088, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.7502370240126874e-05, |
|
"loss": 1.1135, |
|
"step": 14500 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 3.707141748978642e-05, |
|
"loss": 1.0985, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 3.664046473944597e-05, |
|
"loss": 1.0886, |
|
"step": 15500 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 3.6209511989105515e-05, |
|
"loss": 1.0961, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 3.577855923876507e-05, |
|
"loss": 1.0846, |
|
"step": 16500 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 3.534760648842461e-05, |
|
"loss": 1.0638, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 3.4916653738084156e-05, |
|
"loss": 1.0757, |
|
"step": 17500 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 3.448570098774371e-05, |
|
"loss": 1.0676, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 3.405474823740325e-05, |
|
"loss": 1.0547, |
|
"step": 18500 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 3.36237954870628e-05, |
|
"loss": 1.0657, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 3.319284273672235e-05, |
|
"loss": 1.0545, |
|
"step": 19500 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 3.2761889986381894e-05, |
|
"loss": 1.0325, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 3.2330937236041445e-05, |
|
"loss": 1.0263, |
|
"step": 20500 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 3.189998448570099e-05, |
|
"loss": 1.0273, |
|
"step": 21000 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 3.1469031735360535e-05, |
|
"loss": 1.0176, |
|
"step": 21500 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 3.1038078985020086e-05, |
|
"loss": 1.0119, |
|
"step": 22000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 3.060712623467963e-05, |
|
"loss": 1.0184, |
|
"step": 22500 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 3.0176173484339176e-05, |
|
"loss": 1.0226, |
|
"step": 23000 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 2.9745220733998724e-05, |
|
"loss": 0.9991, |
|
"step": 23500 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 2.9314267983658272e-05, |
|
"loss": 0.9955, |
|
"step": 24000 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 2.888331523331782e-05, |
|
"loss": 0.9834, |
|
"step": 24500 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 2.8452362482977368e-05, |
|
"loss": 0.9906, |
|
"step": 25000 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 2.8021409732636916e-05, |
|
"loss": 0.988, |
|
"step": 25500 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 2.7590456982296464e-05, |
|
"loss": 0.9742, |
|
"step": 26000 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 2.7159504231956013e-05, |
|
"loss": 0.9712, |
|
"step": 26500 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 2.6728551481615554e-05, |
|
"loss": 0.9755, |
|
"step": 27000 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 2.6297598731275102e-05, |
|
"loss": 0.9807, |
|
"step": 27500 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 2.586664598093465e-05, |
|
"loss": 0.9752, |
|
"step": 28000 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 2.54356932305942e-05, |
|
"loss": 0.9714, |
|
"step": 28500 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 2.5004740480253746e-05, |
|
"loss": 0.9579, |
|
"step": 29000 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 2.4573787729913295e-05, |
|
"loss": 0.9462, |
|
"step": 29500 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 2.414283497957284e-05, |
|
"loss": 0.9435, |
|
"step": 30000 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 2.3711882229232387e-05, |
|
"loss": 0.9429, |
|
"step": 30500 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 2.3280929478891936e-05, |
|
"loss": 0.9521, |
|
"step": 31000 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 2.2849976728551484e-05, |
|
"loss": 0.9409, |
|
"step": 31500 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 2.241902397821103e-05, |
|
"loss": 0.9434, |
|
"step": 32000 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"learning_rate": 2.1988071227870577e-05, |
|
"loss": 0.9357, |
|
"step": 32500 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 2.1557118477530125e-05, |
|
"loss": 0.9256, |
|
"step": 33000 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 2.1126165727189673e-05, |
|
"loss": 0.9379, |
|
"step": 33500 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"learning_rate": 2.0695212976849218e-05, |
|
"loss": 0.9346, |
|
"step": 34000 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 2.0264260226508766e-05, |
|
"loss": 0.9238, |
|
"step": 34500 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 1.9833307476168314e-05, |
|
"loss": 0.9266, |
|
"step": 35000 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 1.9402354725827862e-05, |
|
"loss": 0.924, |
|
"step": 35500 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 1.8971401975487407e-05, |
|
"loss": 0.9082, |
|
"step": 36000 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 1.8540449225146955e-05, |
|
"loss": 0.9124, |
|
"step": 36500 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"learning_rate": 1.8109496474806503e-05, |
|
"loss": 0.9265, |
|
"step": 37000 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 1.767854372446605e-05, |
|
"loss": 0.9153, |
|
"step": 37500 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 1.7247590974125596e-05, |
|
"loss": 0.9056, |
|
"step": 38000 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 1.6816638223785144e-05, |
|
"loss": 0.8929, |
|
"step": 38500 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 1.6385685473444692e-05, |
|
"loss": 0.8964, |
|
"step": 39000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 1.595473272310424e-05, |
|
"loss": 0.8852, |
|
"step": 39500 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 1.5523779972763785e-05, |
|
"loss": 0.8845, |
|
"step": 40000 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 1.5092827222423333e-05, |
|
"loss": 0.9005, |
|
"step": 40500 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 1.4661874472082881e-05, |
|
"loss": 0.8803, |
|
"step": 41000 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 1.423092172174243e-05, |
|
"loss": 0.8864, |
|
"step": 41500 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 1.3799968971401977e-05, |
|
"loss": 0.8821, |
|
"step": 42000 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.3369016221061522e-05, |
|
"loss": 0.887, |
|
"step": 42500 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 1.293806347072107e-05, |
|
"loss": 0.8738, |
|
"step": 43000 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.2507110720380618e-05, |
|
"loss": 0.8776, |
|
"step": 43500 |
|
}, |
|
{ |
|
"epoch": 2.28, |
|
"learning_rate": 1.2076157970040165e-05, |
|
"loss": 0.8589, |
|
"step": 44000 |
|
}, |
|
{ |
|
"epoch": 2.3, |
|
"learning_rate": 1.1645205219699713e-05, |
|
"loss": 0.8543, |
|
"step": 44500 |
|
}, |
|
{ |
|
"epoch": 2.33, |
|
"learning_rate": 1.121425246935926e-05, |
|
"loss": 0.8584, |
|
"step": 45000 |
|
}, |
|
{ |
|
"epoch": 2.35, |
|
"learning_rate": 1.0783299719018808e-05, |
|
"loss": 0.8614, |
|
"step": 45500 |
|
}, |
|
{ |
|
"epoch": 2.38, |
|
"learning_rate": 1.0352346968678354e-05, |
|
"loss": 0.8537, |
|
"step": 46000 |
|
}, |
|
{ |
|
"epoch": 2.4, |
|
"learning_rate": 9.921394218337902e-06, |
|
"loss": 0.8625, |
|
"step": 46500 |
|
}, |
|
{ |
|
"epoch": 2.43, |
|
"learning_rate": 9.490441467997449e-06, |
|
"loss": 0.8532, |
|
"step": 47000 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 9.059488717656997e-06, |
|
"loss": 0.8559, |
|
"step": 47500 |
|
}, |
|
{ |
|
"epoch": 2.48, |
|
"learning_rate": 8.628535967316543e-06, |
|
"loss": 0.8424, |
|
"step": 48000 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 8.197583216976091e-06, |
|
"loss": 0.8475, |
|
"step": 48500 |
|
}, |
|
{ |
|
"epoch": 2.53, |
|
"learning_rate": 7.766630466635638e-06, |
|
"loss": 0.8525, |
|
"step": 49000 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 7.335677716295186e-06, |
|
"loss": 0.8457, |
|
"step": 49500 |
|
}, |
|
{ |
|
"epoch": 2.59, |
|
"learning_rate": 6.904724965954733e-06, |
|
"loss": 0.8503, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 2.61, |
|
"learning_rate": 6.4737722156142804e-06, |
|
"loss": 0.8469, |
|
"step": 50500 |
|
}, |
|
{ |
|
"epoch": 2.64, |
|
"learning_rate": 6.042819465273828e-06, |
|
"loss": 0.8416, |
|
"step": 51000 |
|
}, |
|
{ |
|
"epoch": 2.66, |
|
"learning_rate": 5.611866714933375e-06, |
|
"loss": 0.8463, |
|
"step": 51500 |
|
}, |
|
{ |
|
"epoch": 2.69, |
|
"learning_rate": 5.180913964592922e-06, |
|
"loss": 0.843, |
|
"step": 52000 |
|
}, |
|
{ |
|
"epoch": 2.72, |
|
"learning_rate": 4.7499612142524696e-06, |
|
"loss": 0.8298, |
|
"step": 52500 |
|
}, |
|
{ |
|
"epoch": 2.74, |
|
"learning_rate": 4.319008463912017e-06, |
|
"loss": 0.8335, |
|
"step": 53000 |
|
}, |
|
{ |
|
"epoch": 2.77, |
|
"learning_rate": 3.888055713571564e-06, |
|
"loss": 0.8363, |
|
"step": 53500 |
|
}, |
|
{ |
|
"epoch": 2.79, |
|
"learning_rate": 3.457102963231112e-06, |
|
"loss": 0.834, |
|
"step": 54000 |
|
}, |
|
{ |
|
"epoch": 2.82, |
|
"learning_rate": 3.0261502128906587e-06, |
|
"loss": 0.8289, |
|
"step": 54500 |
|
}, |
|
{ |
|
"epoch": 2.84, |
|
"learning_rate": 2.595197462550206e-06, |
|
"loss": 0.8384, |
|
"step": 55000 |
|
}, |
|
{ |
|
"epoch": 2.87, |
|
"learning_rate": 2.1642447122097533e-06, |
|
"loss": 0.8255, |
|
"step": 55500 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"learning_rate": 1.7332919618693005e-06, |
|
"loss": 0.825, |
|
"step": 56000 |
|
}, |
|
{ |
|
"epoch": 2.92, |
|
"learning_rate": 1.302339211528848e-06, |
|
"loss": 0.8398, |
|
"step": 56500 |
|
}, |
|
{ |
|
"epoch": 2.95, |
|
"learning_rate": 8.713864611883953e-07, |
|
"loss": 0.8264, |
|
"step": 57000 |
|
}, |
|
{ |
|
"epoch": 2.97, |
|
"learning_rate": 4.4043371084794265e-07, |
|
"loss": 0.8196, |
|
"step": 57500 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 9.48096050748996e-09, |
|
"loss": 0.8307, |
|
"step": 58000 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 58011, |
|
"total_flos": 5.4106824484952474e+17, |
|
"train_loss": 1.0139114455800622, |
|
"train_runtime": 59211.1926, |
|
"train_samples_per_second": 9.797, |
|
"train_steps_per_second": 0.98 |
|
} |
|
], |
|
"max_steps": 58011, |
|
"num_train_epochs": 3, |
|
"total_flos": 5.4106824484952474e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|