| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 168750, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9851851851851855e-05, | |
| "loss": 0.0172, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.970370370370371e-05, | |
| "loss": 0.0128, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.955555555555556e-05, | |
| "loss": 0.0124, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.940740740740741e-05, | |
| "loss": 0.0125, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.925925925925926e-05, | |
| "loss": 0.0116, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9111111111111114e-05, | |
| "loss": 0.0117, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.896296296296297e-05, | |
| "loss": 0.0119, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.881481481481482e-05, | |
| "loss": 0.0129, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.866666666666667e-05, | |
| "loss": 0.0139, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.851851851851852e-05, | |
| "loss": 0.0131, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.837037037037037e-05, | |
| "loss": 0.013, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8222222222222225e-05, | |
| "loss": 0.0133, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.807407407407408e-05, | |
| "loss": 0.0132, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.792592592592593e-05, | |
| "loss": 0.0136, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7777777777777784e-05, | |
| "loss": 0.0132, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.762962962962963e-05, | |
| "loss": 0.0133, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7481481481481483e-05, | |
| "loss": 0.013, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7333333333333336e-05, | |
| "loss": 0.0131, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.718518518518519e-05, | |
| "loss": 0.0128, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.703703703703704e-05, | |
| "loss": 0.013, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6888888888888895e-05, | |
| "loss": 0.0132, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.674074074074074e-05, | |
| "loss": 0.0134, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6592592592592595e-05, | |
| "loss": 0.0131, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.644444444444445e-05, | |
| "loss": 0.0131, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.62962962962963e-05, | |
| "loss": 0.0131, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6148148148148154e-05, | |
| "loss": 0.0128, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.0131, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.585185185185185e-05, | |
| "loss": 0.0135, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5703703703703706e-05, | |
| "loss": 0.0131, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.555555555555556e-05, | |
| "loss": 0.0134, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.540740740740741e-05, | |
| "loss": 0.0131, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5259259259259265e-05, | |
| "loss": 0.013, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.511111111111112e-05, | |
| "loss": 0.0128, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.496296296296297e-05, | |
| "loss": 0.013, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.481481481481482e-05, | |
| "loss": 0.0127, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.466666666666667e-05, | |
| "loss": 0.013, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.4518518518518523e-05, | |
| "loss": 0.0127, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.4370370370370376e-05, | |
| "loss": 0.0125, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.422222222222222e-05, | |
| "loss": 0.0131, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.4074074074074076e-05, | |
| "loss": 0.0129, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.392592592592593e-05, | |
| "loss": 0.0131, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.377777777777778e-05, | |
| "loss": 0.0127, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.3629629629629635e-05, | |
| "loss": 0.0126, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.348148148148148e-05, | |
| "loss": 0.013, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.3333333333333334e-05, | |
| "loss": 0.0131, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.318518518518519e-05, | |
| "loss": 0.0131, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.303703703703704e-05, | |
| "loss": 0.0129, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.2888888888888886e-05, | |
| "loss": 0.0129, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.274074074074074e-05, | |
| "loss": 0.0128, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.259259259259259e-05, | |
| "loss": 0.0132, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.2444444444444445e-05, | |
| "loss": 0.0128, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.22962962962963e-05, | |
| "loss": 0.0131, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.2148148148148145e-05, | |
| "loss": 0.0131, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.0133, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.185185185185185e-05, | |
| "loss": 0.013, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.1703703703703704e-05, | |
| "loss": 0.0132, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.155555555555556e-05, | |
| "loss": 0.0132, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.140740740740741e-05, | |
| "loss": 0.0125, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.1259259259259256e-05, | |
| "loss": 0.0128, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.111111111111111e-05, | |
| "loss": 0.0126, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.096296296296296e-05, | |
| "loss": 0.0127, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.0814814814814815e-05, | |
| "loss": 0.0129, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.066666666666667e-05, | |
| "loss": 0.0129, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.051851851851852e-05, | |
| "loss": 0.0129, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.0370370370370374e-05, | |
| "loss": 0.0127, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.022222222222222e-05, | |
| "loss": 0.0127, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.007407407407407e-05, | |
| "loss": 0.0133, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.9925925925925926e-05, | |
| "loss": 0.0131, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.977777777777778e-05, | |
| "loss": 0.0126, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.962962962962963e-05, | |
| "loss": 0.013, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.9481481481481485e-05, | |
| "loss": 0.0128, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.933333333333333e-05, | |
| "loss": 0.0127, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.9185185185185185e-05, | |
| "loss": 0.0131, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.903703703703704e-05, | |
| "loss": 0.0135, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.888888888888889e-05, | |
| "loss": 0.0125, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.8740740740740744e-05, | |
| "loss": 0.013, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.85925925925926e-05, | |
| "loss": 0.0132, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.844444444444444e-05, | |
| "loss": 0.013, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.8296296296296296e-05, | |
| "loss": 0.0135, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.814814814814815e-05, | |
| "loss": 0.0128, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.013, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.7851851851851855e-05, | |
| "loss": 0.0128, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.770370370370371e-05, | |
| "loss": 0.0129, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.7555555555555554e-05, | |
| "loss": 0.0126, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.740740740740741e-05, | |
| "loss": 0.0129, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.725925925925926e-05, | |
| "loss": 0.0132, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.7111111111111113e-05, | |
| "loss": 0.0129, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.6962962962962966e-05, | |
| "loss": 0.0131, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.681481481481482e-05, | |
| "loss": 0.0126, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.6666666666666666e-05, | |
| "loss": 0.0132, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.651851851851852e-05, | |
| "loss": 0.0131, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.637037037037037e-05, | |
| "loss": 0.0129, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.6222222222222225e-05, | |
| "loss": 0.0126, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.607407407407408e-05, | |
| "loss": 0.0128, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.592592592592593e-05, | |
| "loss": 0.0129, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.577777777777778e-05, | |
| "loss": 0.0128, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.562962962962963e-05, | |
| "loss": 0.0127, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.548148148148148e-05, | |
| "loss": 0.0131, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.5333333333333336e-05, | |
| "loss": 0.0133, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.518518518518519e-05, | |
| "loss": 0.0128, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.503703703703704e-05, | |
| "loss": 0.0133, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.4888888888888895e-05, | |
| "loss": 0.0129, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.474074074074074e-05, | |
| "loss": 0.0128, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.4592592592592594e-05, | |
| "loss": 0.0133, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.444444444444445e-05, | |
| "loss": 0.013, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.42962962962963e-05, | |
| "loss": 0.0132, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4148148148148153e-05, | |
| "loss": 0.0129, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 0.0129, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.385185185185185e-05, | |
| "loss": 0.0127, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.3703703703703706e-05, | |
| "loss": 0.0125, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.355555555555556e-05, | |
| "loss": 0.0132, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.340740740740741e-05, | |
| "loss": 0.0129, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.013578644022345543, | |
| "eval_mse": 0.013578643091022968, | |
| "eval_runtime": 99.5064, | |
| "eval_samples_per_second": 140.694, | |
| "step": 56250 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.3259259259259265e-05, | |
| "loss": 0.0126, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.311111111111112e-05, | |
| "loss": 0.013, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.2962962962962964e-05, | |
| "loss": 0.0127, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.281481481481482e-05, | |
| "loss": 0.0129, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.266666666666667e-05, | |
| "loss": 0.0128, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.251851851851852e-05, | |
| "loss": 0.013, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.2370370370370376e-05, | |
| "loss": 0.0128, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.222222222222223e-05, | |
| "loss": 0.0132, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.2074074074074075e-05, | |
| "loss": 0.0129, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.192592592592593e-05, | |
| "loss": 0.0128, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.177777777777778e-05, | |
| "loss": 0.0129, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.1629629629629634e-05, | |
| "loss": 0.0131, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.148148148148148e-05, | |
| "loss": 0.0128, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.1333333333333334e-05, | |
| "loss": 0.0131, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.118518518518519e-05, | |
| "loss": 0.0134, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.103703703703704e-05, | |
| "loss": 0.0129, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.088888888888889e-05, | |
| "loss": 0.0129, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.074074074074074e-05, | |
| "loss": 0.0128, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.059259259259259e-05, | |
| "loss": 0.0132, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.044444444444445e-05, | |
| "loss": 0.0128, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.02962962962963e-05, | |
| "loss": 0.0131, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0148148148148148e-05, | |
| "loss": 0.0128, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3e-05, | |
| "loss": 0.0129, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9851851851851854e-05, | |
| "loss": 0.0128, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9703703703703707e-05, | |
| "loss": 0.0131, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.955555555555556e-05, | |
| "loss": 0.0131, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9407407407407413e-05, | |
| "loss": 0.0126, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.925925925925926e-05, | |
| "loss": 0.0133, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9111111111111112e-05, | |
| "loss": 0.0131, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.8962962962962965e-05, | |
| "loss": 0.0129, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8814814814814818e-05, | |
| "loss": 0.0131, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8666666666666668e-05, | |
| "loss": 0.0126, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.851851851851852e-05, | |
| "loss": 0.0125, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.837037037037037e-05, | |
| "loss": 0.0129, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.8222222222222223e-05, | |
| "loss": 0.013, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.8074074074074076e-05, | |
| "loss": 0.013, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.7925925925925926e-05, | |
| "loss": 0.0128, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.777777777777778e-05, | |
| "loss": 0.0132, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.7629629629629632e-05, | |
| "loss": 0.0128, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7481481481481482e-05, | |
| "loss": 0.0126, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.733333333333333e-05, | |
| "loss": 0.0127, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7185185185185184e-05, | |
| "loss": 0.0127, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.7037037037037037e-05, | |
| "loss": 0.0126, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.688888888888889e-05, | |
| "loss": 0.013, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6740740740740743e-05, | |
| "loss": 0.0129, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.659259259259259e-05, | |
| "loss": 0.013, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6444444444444443e-05, | |
| "loss": 0.0123, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6296296296296296e-05, | |
| "loss": 0.0132, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.614814814814815e-05, | |
| "loss": 0.0135, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 0.0128, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5851851851851855e-05, | |
| "loss": 0.013, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5703703703703708e-05, | |
| "loss": 0.013, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5555555555555554e-05, | |
| "loss": 0.0129, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5407407407407407e-05, | |
| "loss": 0.0128, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.525925925925926e-05, | |
| "loss": 0.0124, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5111111111111113e-05, | |
| "loss": 0.0132, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.4962962962962963e-05, | |
| "loss": 0.0129, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.4814814814814816e-05, | |
| "loss": 0.0126, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.466666666666667e-05, | |
| "loss": 0.0129, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.451851851851852e-05, | |
| "loss": 0.013, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.437037037037037e-05, | |
| "loss": 0.013, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4222222222222224e-05, | |
| "loss": 0.0131, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4074074074074074e-05, | |
| "loss": 0.0127, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.3925925925925927e-05, | |
| "loss": 0.013, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.377777777777778e-05, | |
| "loss": 0.013, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3629629629629633e-05, | |
| "loss": 0.0124, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.3481481481481483e-05, | |
| "loss": 0.0131, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.3333333333333336e-05, | |
| "loss": 0.0132, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.318518518518519e-05, | |
| "loss": 0.0126, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.303703703703704e-05, | |
| "loss": 0.0129, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.288888888888889e-05, | |
| "loss": 0.0131, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.2740740740740744e-05, | |
| "loss": 0.0127, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.2592592592592594e-05, | |
| "loss": 0.0131, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2444444444444447e-05, | |
| "loss": 0.013, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2296296296296297e-05, | |
| "loss": 0.0129, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.214814814814815e-05, | |
| "loss": 0.0127, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 0.0123, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1851851851851852e-05, | |
| "loss": 0.0127, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1703703703703705e-05, | |
| "loss": 0.0128, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1555555555555555e-05, | |
| "loss": 0.0129, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1407407407407408e-05, | |
| "loss": 0.0126, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1259259259259258e-05, | |
| "loss": 0.0126, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.111111111111111e-05, | |
| "loss": 0.013, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.0962962962962964e-05, | |
| "loss": 0.0132, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0814814814814813e-05, | |
| "loss": 0.0127, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.0666666666666666e-05, | |
| "loss": 0.0126, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.051851851851852e-05, | |
| "loss": 0.0127, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.037037037037037e-05, | |
| "loss": 0.0128, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0222222222222222e-05, | |
| "loss": 0.0129, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.0074074074074075e-05, | |
| "loss": 0.0125, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.9925925925925925e-05, | |
| "loss": 0.0126, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9777777777777778e-05, | |
| "loss": 0.0126, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.962962962962963e-05, | |
| "loss": 0.013, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.948148148148148e-05, | |
| "loss": 0.013, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9333333333333333e-05, | |
| "loss": 0.0128, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9185185185185186e-05, | |
| "loss": 0.0129, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.903703703703704e-05, | |
| "loss": 0.0127, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.888888888888889e-05, | |
| "loss": 0.0129, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8740740740740742e-05, | |
| "loss": 0.0128, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8592592592592595e-05, | |
| "loss": 0.0127, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8444444444444445e-05, | |
| "loss": 0.0128, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.8296296296296298e-05, | |
| "loss": 0.0124, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.814814814814815e-05, | |
| "loss": 0.0126, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.0128, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.7851851851851853e-05, | |
| "loss": 0.0128, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7703703703703706e-05, | |
| "loss": 0.0131, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.7555555555555556e-05, | |
| "loss": 0.0131, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.740740740740741e-05, | |
| "loss": 0.0127, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7259259259259262e-05, | |
| "loss": 0.0129, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7111111111111112e-05, | |
| "loss": 0.0128, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.6962962962962965e-05, | |
| "loss": 0.0127, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6814814814814818e-05, | |
| "loss": 0.0125, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 0.0126, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.013725973665714264, | |
| "eval_mse": 0.013725974597036839, | |
| "eval_runtime": 99.3457, | |
| "eval_samples_per_second": 140.922, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.651851851851852e-05, | |
| "loss": 0.0128, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.6370370370370374e-05, | |
| "loss": 0.0127, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6222222222222223e-05, | |
| "loss": 0.0129, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.6074074074074076e-05, | |
| "loss": 0.0128, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.5925925925925926e-05, | |
| "loss": 0.0126, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.577777777777778e-05, | |
| "loss": 0.0131, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.5629629629629632e-05, | |
| "loss": 0.0129, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.548148148148148e-05, | |
| "loss": 0.0129, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5333333333333334e-05, | |
| "loss": 0.0127, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5185185185185186e-05, | |
| "loss": 0.013, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.5037037037037039e-05, | |
| "loss": 0.0129, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4888888888888888e-05, | |
| "loss": 0.0126, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4740740740740741e-05, | |
| "loss": 0.0129, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4592592592592594e-05, | |
| "loss": 0.0125, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4444444444444444e-05, | |
| "loss": 0.013, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.4296296296296297e-05, | |
| "loss": 0.0128, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4148148148148148e-05, | |
| "loss": 0.0125, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 0.0129, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3851851851851853e-05, | |
| "loss": 0.0132, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3703703703703704e-05, | |
| "loss": 0.0126, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3555555555555557e-05, | |
| "loss": 0.0127, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3407407407407407e-05, | |
| "loss": 0.013, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.325925925925926e-05, | |
| "loss": 0.0134, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3111111111111113e-05, | |
| "loss": 0.0129, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.2962962962962962e-05, | |
| "loss": 0.0125, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.2814814814814815e-05, | |
| "loss": 0.0128, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2666666666666668e-05, | |
| "loss": 0.0125, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2518518518518518e-05, | |
| "loss": 0.0127, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2370370370370371e-05, | |
| "loss": 0.0127, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2222222222222222e-05, | |
| "loss": 0.0127, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.2074074074074075e-05, | |
| "loss": 0.013, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.1925925925925927e-05, | |
| "loss": 0.0125, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1777777777777778e-05, | |
| "loss": 0.0128, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.1629629629629631e-05, | |
| "loss": 0.0128, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.1481481481481482e-05, | |
| "loss": 0.013, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1333333333333334e-05, | |
| "loss": 0.0128, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1185185185185187e-05, | |
| "loss": 0.0131, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.1037037037037038e-05, | |
| "loss": 0.013, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.088888888888889e-05, | |
| "loss": 0.0126, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.074074074074074e-05, | |
| "loss": 0.0127, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0592592592592592e-05, | |
| "loss": 0.0127, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0444444444444445e-05, | |
| "loss": 0.0135, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0296296296296296e-05, | |
| "loss": 0.0127, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0148148148148148e-05, | |
| "loss": 0.0128, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1e-05, | |
| "loss": 0.0127, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.851851851851852e-06, | |
| "loss": 0.0126, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.703703703703703e-06, | |
| "loss": 0.0129, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.0128, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.407407407407408e-06, | |
| "loss": 0.0129, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.259259259259259e-06, | |
| "loss": 0.0127, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.013, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 8.962962962962963e-06, | |
| "loss": 0.0129, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.814814814814815e-06, | |
| "loss": 0.0126, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.0125, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.518518518518519e-06, | |
| "loss": 0.0129, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.37037037037037e-06, | |
| "loss": 0.013, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.222222222222223e-06, | |
| "loss": 0.0131, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.074074074074075e-06, | |
| "loss": 0.0127, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.925925925925926e-06, | |
| "loss": 0.0129, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.777777777777777e-06, | |
| "loss": 0.0129, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.629629629629629e-06, | |
| "loss": 0.0124, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.481481481481483e-06, | |
| "loss": 0.0128, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.333333333333334e-06, | |
| "loss": 0.0126, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.185185185185185e-06, | |
| "loss": 0.0127, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.0370370370370375e-06, | |
| "loss": 0.0124, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.888888888888889e-06, | |
| "loss": 0.0126, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.74074074074074e-06, | |
| "loss": 0.0126, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.592592592592593e-06, | |
| "loss": 0.0126, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.4444444444444445e-06, | |
| "loss": 0.0125, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.296296296296296e-06, | |
| "loss": 0.0125, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.148148148148149e-06, | |
| "loss": 0.0128, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 6e-06, | |
| "loss": 0.0127, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.851851851851852e-06, | |
| "loss": 0.0129, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.7037037037037045e-06, | |
| "loss": 0.0127, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.555555555555556e-06, | |
| "loss": 0.0126, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.407407407407407e-06, | |
| "loss": 0.0127, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.259259259259259e-06, | |
| "loss": 0.0127, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.0127, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.962962962962963e-06, | |
| "loss": 0.0124, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.814814814814815e-06, | |
| "loss": 0.0125, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.0126, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.5185185185185185e-06, | |
| "loss": 0.0126, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.370370370370371e-06, | |
| "loss": 0.0123, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.0126, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.074074074074075e-06, | |
| "loss": 0.0129, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.925925925925926e-06, | |
| "loss": 0.0124, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.0125, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.6296296296296302e-06, | |
| "loss": 0.0126, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.4814814814814816e-06, | |
| "loss": 0.0127, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.0131, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.1851851851851855e-06, | |
| "loss": 0.0124, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.0370370370370372e-06, | |
| "loss": 0.0124, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0125, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.7407407407407407e-06, | |
| "loss": 0.0129, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.5925925925925925e-06, | |
| "loss": 0.0129, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.0126, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.2962962962962964e-06, | |
| "loss": 0.0127, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.148148148148148e-06, | |
| "loss": 0.0125, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0128, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.8518518518518519e-06, | |
| "loss": 0.0125, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.7037037037037038e-06, | |
| "loss": 0.0127, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.5555555555555556e-06, | |
| "loss": 0.0124, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.4074074074074075e-06, | |
| "loss": 0.0129, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.2592592592592593e-06, | |
| "loss": 0.0128, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.1111111111111112e-06, | |
| "loss": 0.0124, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 9.62962962962963e-07, | |
| "loss": 0.0123, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.148148148148147e-07, | |
| "loss": 0.0128, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0129, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.185185185185186e-07, | |
| "loss": 0.0122, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.703703703703704e-07, | |
| "loss": 0.0128, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.0127, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.407407407407407e-08, | |
| "loss": 0.0127, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.013624305836856365, | |
| "eval_mse": 0.013624305836856365, | |
| "eval_runtime": 98.4295, | |
| "eval_samples_per_second": 142.234, | |
| "step": 168750 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 168750, | |
| "total_flos": 5045118796800000.0, | |
| "train_runtime": 51961.6794, | |
| "train_samples_per_second": 3.248 | |
| } | |
| ], | |
| "max_steps": 168750, | |
| "num_train_epochs": 3, | |
| "total_flos": 5045118796800000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |