| { | |
| "best_metric": 0.44585558772087097, | |
| "best_model_checkpoint": "english_xlm_xnli/checkpoint-24544", | |
| "epoch": 3.0, | |
| "global_step": 36816, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.932094741416776e-05, | |
| "loss": 0.8997, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.864189482833551e-05, | |
| "loss": 0.7341, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7962842242503265e-05, | |
| "loss": 0.6824, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.728378965667101e-05, | |
| "loss": 0.6543, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.660473707083877e-05, | |
| "loss": 0.6454, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.592568448500652e-05, | |
| "loss": 0.6381, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5246631899174276e-05, | |
| "loss": 0.6162, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.456757931334203e-05, | |
| "loss": 0.6062, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.3888526727509784e-05, | |
| "loss": 0.5852, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.320947414167754e-05, | |
| "loss": 0.587, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.2530421555845286e-05, | |
| "loss": 0.5819, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.185136897001304e-05, | |
| "loss": 0.5686, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1172316384180795e-05, | |
| "loss": 0.5706, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.049326379834855e-05, | |
| "loss": 0.5697, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.98142112125163e-05, | |
| "loss": 0.5562, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.913515862668405e-05, | |
| "loss": 0.5462, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.8456106040851805e-05, | |
| "loss": 0.551, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.777705345501956e-05, | |
| "loss": 0.5402, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.7098000869187314e-05, | |
| "loss": 0.539, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.641894828335506e-05, | |
| "loss": 0.5315, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.5739895697522816e-05, | |
| "loss": 0.5344, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.506084311169057e-05, | |
| "loss": 0.5359, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.4381790525858324e-05, | |
| "loss": 0.5276, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.370273794002607e-05, | |
| "loss": 0.5263, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8072289156626506, | |
| "eval_loss": 0.508637011051178, | |
| "eval_runtime": 5.1209, | |
| "eval_samples_per_second": 486.244, | |
| "eval_steps_per_second": 60.927, | |
| "step": 12272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.3023685354193826e-05, | |
| "loss": 0.4795, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.234463276836158e-05, | |
| "loss": 0.4424, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.1665580182529335e-05, | |
| "loss": 0.442, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.098652759669709e-05, | |
| "loss": 0.4355, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0307475010864843e-05, | |
| "loss": 0.4249, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9628422425032598e-05, | |
| "loss": 0.44, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.8949369839200345e-05, | |
| "loss": 0.4181, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.82703172533681e-05, | |
| "loss": 0.4275, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.7591264667535854e-05, | |
| "loss": 0.4289, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.6912212081703608e-05, | |
| "loss": 0.4324, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6233159495871362e-05, | |
| "loss": 0.4218, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5554106910039117e-05, | |
| "loss": 0.4227, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.4875054324206867e-05, | |
| "loss": 0.4232, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4196001738374622e-05, | |
| "loss": 0.4199, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.3516949152542376e-05, | |
| "loss": 0.4205, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2837896566710127e-05, | |
| "loss": 0.429, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.215884398087788e-05, | |
| "loss": 0.4188, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1479791395045636e-05, | |
| "loss": 0.4178, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0800738809213386e-05, | |
| "loss": 0.4232, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.012168622338114e-05, | |
| "loss": 0.4136, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.944263363754889e-05, | |
| "loss": 0.4164, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8763581051716646e-05, | |
| "loss": 0.4223, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8084528465884397e-05, | |
| "loss": 0.4036, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.740547588005215e-05, | |
| "loss": 0.4116, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6726423294219905e-05, | |
| "loss": 0.4115, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8176706827309237, | |
| "eval_loss": 0.44585558772087097, | |
| "eval_runtime": 5.1063, | |
| "eval_samples_per_second": 487.632, | |
| "eval_steps_per_second": 61.101, | |
| "step": 24544 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.6047370708387656e-05, | |
| "loss": 0.3125, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.536831812255541e-05, | |
| "loss": 0.318, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4689265536723165e-05, | |
| "loss": 0.3112, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.4010212950890916e-05, | |
| "loss": 0.3067, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.333116036505867e-05, | |
| "loss": 0.3087, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2652107779226425e-05, | |
| "loss": 0.2984, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.1973055193394177e-05, | |
| "loss": 0.3089, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1294002607561931e-05, | |
| "loss": 0.299, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0614950021729684e-05, | |
| "loss": 0.3042, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.935897435897435e-06, | |
| "loss": 0.3076, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.25684485006519e-06, | |
| "loss": 0.3016, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.577792264232942e-06, | |
| "loss": 0.3122, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.898739678400696e-06, | |
| "loss": 0.2986, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.219687092568449e-06, | |
| "loss": 0.3017, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.540634506736203e-06, | |
| "loss": 0.2947, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.861581920903955e-06, | |
| "loss": 0.2893, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.182529335071708e-06, | |
| "loss": 0.3045, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.503476749239461e-06, | |
| "loss": 0.2893, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.824424163407214e-06, | |
| "loss": 0.2916, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.1453715775749674e-06, | |
| "loss": 0.3041, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.466318991742721e-06, | |
| "loss": 0.283, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.7872664059104738e-06, | |
| "loss": 0.2954, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.108213820078227e-06, | |
| "loss": 0.2918, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.2916123424598005e-07, | |
| "loss": 0.2926, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8253012048192772, | |
| "eval_loss": 0.4829569160938263, | |
| "eval_runtime": 4.9822, | |
| "eval_samples_per_second": 499.779, | |
| "eval_steps_per_second": 62.623, | |
| "step": 36816 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 36816, | |
| "total_flos": 7.74938740264658e+16, | |
| "train_loss": 0.4398244035497845, | |
| "train_runtime": 7509.3386, | |
| "train_samples_per_second": 156.885, | |
| "train_steps_per_second": 4.903 | |
| } | |
| ], | |
| "max_steps": 36816, | |
| "num_train_epochs": 3, | |
| "total_flos": 7.74938740264658e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |