| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.5968691500040696, | |
| "global_step": 44000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.943068181818182e-05, | |
| "loss": 0.0316, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.88625e-05, | |
| "loss": 0.0341, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.829431818181818e-05, | |
| "loss": 0.0346, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.772613636363636e-05, | |
| "loss": 0.0352, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.715795454545455e-05, | |
| "loss": 0.0369, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.658977272727273e-05, | |
| "loss": 0.0377, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.602159090909091e-05, | |
| "loss": 0.0355, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.545340909090909e-05, | |
| "loss": 0.0359, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.488522727272728e-05, | |
| "loss": 0.0366, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.431704545454546e-05, | |
| "loss": 0.0364, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.374886363636364e-05, | |
| "loss": 0.0362, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.318181818181819e-05, | |
| "loss": 0.0382, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.261363636363637e-05, | |
| "loss": 0.036, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.204545454545455e-05, | |
| "loss": 0.0338, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.1477272727272734e-05, | |
| "loss": 0.0358, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.0909090909090915e-05, | |
| "loss": 0.0355, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.0342045454545456e-05, | |
| "loss": 0.0382, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.9773863636363637e-05, | |
| "loss": 0.0368, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.9205681818181824e-05, | |
| "loss": 0.0352, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.8638636363636365e-05, | |
| "loss": 0.0344, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.8070454545454546e-05, | |
| "loss": 0.0351, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.750227272727273e-05, | |
| "loss": 0.0358, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.693409090909091e-05, | |
| "loss": 0.0346, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.636590909090909e-05, | |
| "loss": 0.0352, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 3.5797727272727274e-05, | |
| "loss": 0.0368, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.5229545454545454e-05, | |
| "loss": 0.0367, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.4661363636363634e-05, | |
| "loss": 0.036, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.409318181818182e-05, | |
| "loss": 0.0343, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.3525e-05, | |
| "loss": 0.0353, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.295681818181818e-05, | |
| "loss": 0.034, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.238863636363636e-05, | |
| "loss": 0.0341, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.182159090909091e-05, | |
| "loss": 0.0348, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.125340909090909e-05, | |
| "loss": 0.0312, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.068522727272727e-05, | |
| "loss": 0.034, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.0118181818181816e-05, | |
| "loss": 0.0341, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.955e-05, | |
| "loss": 0.0333, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.8981818181818184e-05, | |
| "loss": 0.0351, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.8413636363636364e-05, | |
| "loss": 0.0341, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.7845454545454548e-05, | |
| "loss": 0.0337, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.727840909090909e-05, | |
| "loss": 0.0336, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.6710227272727273e-05, | |
| "loss": 0.0335, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.6142045454545457e-05, | |
| "loss": 0.0335, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.5573863636363637e-05, | |
| "loss": 0.0364, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.500568181818182e-05, | |
| "loss": 0.036, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.4438636363636362e-05, | |
| "loss": 0.0347, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.3870454545454546e-05, | |
| "loss": 0.032, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.330227272727273e-05, | |
| "loss": 0.0331, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.273409090909091e-05, | |
| "loss": 0.0324, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.2167045454545455e-05, | |
| "loss": 0.0334, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.1598863636363635e-05, | |
| "loss": 0.0327, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.103068181818182e-05, | |
| "loss": 0.0303, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.0463636363636364e-05, | |
| "loss": 0.03, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9895454545454548e-05, | |
| "loss": 0.0316, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.9327272727272728e-05, | |
| "loss": 0.0327, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8759090909090908e-05, | |
| "loss": 0.0326, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.8192045454545457e-05, | |
| "loss": 0.038, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.7623863636363637e-05, | |
| "loss": 0.0298, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.7055681818181817e-05, | |
| "loss": 0.0335, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.64875e-05, | |
| "loss": 0.0304, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.591931818181818e-05, | |
| "loss": 0.0322, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.5351136363636365e-05, | |
| "loss": 0.033, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.4782954545454547e-05, | |
| "loss": 0.0307, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.4214772727272727e-05, | |
| "loss": 0.0324, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.364659090909091e-05, | |
| "loss": 0.0312, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.3078409090909091e-05, | |
| "loss": 0.0306, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2510227272727273e-05, | |
| "loss": 0.0323, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1942045454545455e-05, | |
| "loss": 0.0303, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1373863636363637e-05, | |
| "loss": 0.0295, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.080568181818182e-05, | |
| "loss": 0.0318, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.02375e-05, | |
| "loss": 0.0317, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.670454545454546e-06, | |
| "loss": 0.0309, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.102272727272727e-06, | |
| "loss": 0.0287, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.53409090909091e-06, | |
| "loss": 0.0309, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 7.965909090909092e-06, | |
| "loss": 0.0295, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 7.397727272727273e-06, | |
| "loss": 0.0302, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.829545454545455e-06, | |
| "loss": 0.0286, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.261363636363636e-06, | |
| "loss": 0.0291, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.693181818181818e-06, | |
| "loss": 0.0303, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.125e-06, | |
| "loss": 0.0284, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.557954545454546e-06, | |
| "loss": 0.0308, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.989772727272727e-06, | |
| "loss": 0.0305, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.421590909090909e-06, | |
| "loss": 0.0302, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.853409090909091e-06, | |
| "loss": 0.0289, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.2863636363636364e-06, | |
| "loss": 0.0289, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.7181818181818184e-06, | |
| "loss": 0.0302, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.15e-06, | |
| "loss": 0.0308, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.818181818181818e-07, | |
| "loss": 0.0291, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.3636363636363638e-08, | |
| "loss": 0.0284, | |
| "step": 44000 | |
| } | |
| ], | |
| "max_steps": 44000, | |
| "num_train_epochs": 1, | |
| "total_flos": 4.597453878932275e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |