| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.3, |
| "global_step": 60000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9875000000000006e-05, |
| "loss": 6.535, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.975e-05, |
| "loss": 5.1906, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.962500000000001e-05, |
| "loss": 4.6323, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9500000000000004e-05, |
| "loss": 3.96, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.937500000000001e-05, |
| "loss": 3.6713, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.9250000000000004e-05, |
| "loss": 3.4882, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9125e-05, |
| "loss": 3.3843, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9e-05, |
| "loss": 3.3546, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8875e-05, |
| "loss": 3.223, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.875e-05, |
| "loss": 3.1837, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.8625e-05, |
| "loss": 3.063, |
| "step": 5500 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.85e-05, |
| "loss": 3.092, |
| "step": 6000 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.8375000000000004e-05, |
| "loss": 3.0056, |
| "step": 6500 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.825e-05, |
| "loss": 2.9768, |
| "step": 7000 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.8125000000000004e-05, |
| "loss": 2.9547, |
| "step": 7500 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.8e-05, |
| "loss": 2.8948, |
| "step": 8000 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.7875000000000005e-05, |
| "loss": 2.9324, |
| "step": 8500 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.775e-05, |
| "loss": 2.8575, |
| "step": 9000 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.7625000000000006e-05, |
| "loss": 2.8774, |
| "step": 9500 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.75e-05, |
| "loss": 2.8511, |
| "step": 10000 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.7375e-05, |
| "loss": 2.8635, |
| "step": 10500 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.7249999999999997e-05, |
| "loss": 2.7233, |
| "step": 11000 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.7125e-05, |
| "loss": 2.7953, |
| "step": 11500 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.7e-05, |
| "loss": 2.7092, |
| "step": 12000 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.6875e-05, |
| "loss": 2.7372, |
| "step": 12500 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.6750000000000005e-05, |
| "loss": 2.7834, |
| "step": 13000 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.6625e-05, |
| "loss": 2.7742, |
| "step": 13500 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.6500000000000005e-05, |
| "loss": 2.7567, |
| "step": 14000 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.6375e-05, |
| "loss": 2.7159, |
| "step": 14500 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.6250000000000006e-05, |
| "loss": 2.6708, |
| "step": 15000 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 4.6125e-05, |
| "loss": 2.7847, |
| "step": 15500 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 4.600000000000001e-05, |
| "loss": 2.7013, |
| "step": 16000 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 4.5875000000000004e-05, |
| "loss": 2.6901, |
| "step": 16500 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.575e-05, |
| "loss": 2.6959, |
| "step": 17000 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.5625e-05, |
| "loss": 2.6235, |
| "step": 17500 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.55e-05, |
| "loss": 2.698, |
| "step": 18000 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.5375e-05, |
| "loss": 2.635, |
| "step": 18500 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.525e-05, |
| "loss": 2.5864, |
| "step": 19000 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.5125e-05, |
| "loss": 2.7273, |
| "step": 19500 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.5e-05, |
| "loss": 2.628, |
| "step": 20000 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.4875e-05, |
| "loss": 2.6253, |
| "step": 20500 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.4750000000000004e-05, |
| "loss": 2.5907, |
| "step": 21000 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.4625e-05, |
| "loss": 2.6367, |
| "step": 21500 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.4500000000000004e-05, |
| "loss": 2.6509, |
| "step": 22000 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.4375e-05, |
| "loss": 2.5594, |
| "step": 22500 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.4250000000000005e-05, |
| "loss": 2.6131, |
| "step": 23000 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.4125e-05, |
| "loss": 2.5891, |
| "step": 23500 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.4000000000000006e-05, |
| "loss": 2.5567, |
| "step": 24000 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.3875e-05, |
| "loss": 2.5828, |
| "step": 24500 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.375e-05, |
| "loss": 2.5276, |
| "step": 25000 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 4.3625e-05, |
| "loss": 2.6037, |
| "step": 25500 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 4.35e-05, |
| "loss": 2.5522, |
| "step": 26000 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 4.3375000000000004e-05, |
| "loss": 2.612, |
| "step": 26500 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.325e-05, |
| "loss": 2.5103, |
| "step": 27000 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.3125000000000005e-05, |
| "loss": 2.5565, |
| "step": 27500 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.3e-05, |
| "loss": 2.5448, |
| "step": 28000 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.2875000000000005e-05, |
| "loss": 2.5231, |
| "step": 28500 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.275e-05, |
| "loss": 2.5115, |
| "step": 29000 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.2625000000000006e-05, |
| "loss": 2.5543, |
| "step": 29500 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.25e-05, |
| "loss": 2.5185, |
| "step": 30000 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.237500000000001e-05, |
| "loss": 2.5097, |
| "step": 30500 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.2250000000000004e-05, |
| "loss": 2.4648, |
| "step": 31000 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.2125e-05, |
| "loss": 2.5455, |
| "step": 31500 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.2e-05, |
| "loss": 2.4929, |
| "step": 32000 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.1875e-05, |
| "loss": 2.5141, |
| "step": 32500 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.175e-05, |
| "loss": 2.5016, |
| "step": 33000 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.1625e-05, |
| "loss": 2.515, |
| "step": 33500 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.15e-05, |
| "loss": 2.4849, |
| "step": 34000 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.1375e-05, |
| "loss": 2.4865, |
| "step": 34500 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.125e-05, |
| "loss": 2.4996, |
| "step": 35000 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.1125000000000004e-05, |
| "loss": 2.5202, |
| "step": 35500 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.1e-05, |
| "loss": 2.491, |
| "step": 36000 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.0875000000000004e-05, |
| "loss": 2.4586, |
| "step": 36500 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.075e-05, |
| "loss": 2.4067, |
| "step": 37000 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.0625000000000005e-05, |
| "loss": 2.4727, |
| "step": 37500 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.05e-05, |
| "loss": 2.473, |
| "step": 38000 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.0375e-05, |
| "loss": 2.5031, |
| "step": 38500 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 4.025e-05, |
| "loss": 2.4446, |
| "step": 39000 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 4.0125e-05, |
| "loss": 2.4294, |
| "step": 39500 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 4e-05, |
| "loss": 2.4211, |
| "step": 40000 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 3.9875e-05, |
| "loss": 2.4458, |
| "step": 40500 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 3.9750000000000004e-05, |
| "loss": 2.4372, |
| "step": 41000 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 3.9625e-05, |
| "loss": 2.4291, |
| "step": 41500 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 3.9500000000000005e-05, |
| "loss": 2.4459, |
| "step": 42000 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 3.9375e-05, |
| "loss": 2.4384, |
| "step": 42500 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 3.9250000000000005e-05, |
| "loss": 2.3212, |
| "step": 43000 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 3.9125e-05, |
| "loss": 2.4252, |
| "step": 43500 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 3.9000000000000006e-05, |
| "loss": 2.447, |
| "step": 44000 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 3.8875e-05, |
| "loss": 2.3701, |
| "step": 44500 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 3.875e-05, |
| "loss": 2.4841, |
| "step": 45000 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 3.8625e-05, |
| "loss": 2.4248, |
| "step": 45500 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 3.85e-05, |
| "loss": 2.4825, |
| "step": 46000 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 3.8375e-05, |
| "loss": 2.4443, |
| "step": 46500 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 3.825e-05, |
| "loss": 2.4231, |
| "step": 47000 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 3.8125e-05, |
| "loss": 2.4251, |
| "step": 47500 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 3.8e-05, |
| "loss": 2.3801, |
| "step": 48000 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 3.7875e-05, |
| "loss": 2.3641, |
| "step": 48500 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 3.775e-05, |
| "loss": 2.391, |
| "step": 49000 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 3.7625e-05, |
| "loss": 2.3917, |
| "step": 49500 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 3.7500000000000003e-05, |
| "loss": 2.4489, |
| "step": 50000 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 3.737500000000001e-05, |
| "loss": 2.413, |
| "step": 50500 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 3.7250000000000004e-05, |
| "loss": 2.424, |
| "step": 51000 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 3.7125e-05, |
| "loss": 2.3752, |
| "step": 51500 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 3.7e-05, |
| "loss": 2.418, |
| "step": 52000 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 3.6875e-05, |
| "loss": 2.3733, |
| "step": 52500 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 3.675e-05, |
| "loss": 2.4311, |
| "step": 53000 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 3.6625e-05, |
| "loss": 2.4121, |
| "step": 53500 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 3.65e-05, |
| "loss": 2.3514, |
| "step": 54000 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 3.6375e-05, |
| "loss": 2.3905, |
| "step": 54500 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 3.625e-05, |
| "loss": 2.288, |
| "step": 55000 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 3.6125000000000004e-05, |
| "loss": 2.3938, |
| "step": 55500 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 3.6e-05, |
| "loss": 2.3689, |
| "step": 56000 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 3.5875000000000005e-05, |
| "loss": 2.417, |
| "step": 56500 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 3.575e-05, |
| "loss": 2.3249, |
| "step": 57000 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 3.5625000000000005e-05, |
| "loss": 2.2635, |
| "step": 57500 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 3.55e-05, |
| "loss": 2.3325, |
| "step": 58000 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 3.5375e-05, |
| "loss": 2.3574, |
| "step": 58500 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 3.525e-05, |
| "loss": 2.3943, |
| "step": 59000 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 3.5125e-05, |
| "loss": 2.3558, |
| "step": 59500 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 3.5e-05, |
| "loss": 2.3984, |
| "step": 60000 |
| } |
| ], |
| "max_steps": 200000, |
| "num_train_epochs": 9223372036854775807, |
| "total_flos": 8.31675319713792e+18, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|