| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 6620, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 6.042296072507553e-06, | |
| "loss": 6.2033, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.2084592145015106e-05, | |
| "loss": 2.9531, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.812688821752266e-05, | |
| "loss": 2.6921, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.4169184290030212e-05, | |
| "loss": 2.5516, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.0211480362537768e-05, | |
| "loss": 2.4571, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.625377643504532e-05, | |
| "loss": 2.4143, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.974488083249413e-05, | |
| "loss": 2.3907, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.907351460221551e-05, | |
| "loss": 2.3511, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.840214837193689e-05, | |
| "loss": 2.318, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.773078214165828e-05, | |
| "loss": 2.3014, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.705941591137966e-05, | |
| "loss": 2.2913, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.638804968110104e-05, | |
| "loss": 2.2729, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.571668345082243e-05, | |
| "loss": 2.2123, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.504531722054381e-05, | |
| "loss": 2.1213, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.437395099026519e-05, | |
| "loss": 2.0869, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.370258475998657e-05, | |
| "loss": 2.0782, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.303121852970796e-05, | |
| "loss": 2.0852, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.235985229942934e-05, | |
| "loss": 2.0695, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.168848606915072e-05, | |
| "loss": 2.0743, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 3.101711983887211e-05, | |
| "loss": 2.0884, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.0345753608593493e-05, | |
| "loss": 2.059, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.967438737831487e-05, | |
| "loss": 2.0654, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.9003021148036254e-05, | |
| "loss": 2.0302, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.833165491775764e-05, | |
| "loss": 2.062, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.7660288687479022e-05, | |
| "loss": 2.0167, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.6988922457200404e-05, | |
| "loss": 2.0536, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.631755622692179e-05, | |
| "loss": 1.9644, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.5646189996643172e-05, | |
| "loss": 1.8799, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.4974823766364555e-05, | |
| "loss": 1.8726, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.4303457536085934e-05, | |
| "loss": 1.8819, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.363209130580732e-05, | |
| "loss": 1.8726, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 2.2960725075528702e-05, | |
| "loss": 1.8584, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 2.2289358845250084e-05, | |
| "loss": 1.8824, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.161799261497147e-05, | |
| "loss": 1.8721, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.0946626384692852e-05, | |
| "loss": 1.8858, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 2.0275260154414234e-05, | |
| "loss": 1.8946, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.9603893924135617e-05, | |
| "loss": 1.8755, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 1.8932527693857003e-05, | |
| "loss": 1.8765, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 1.826116146357838e-05, | |
| "loss": 1.8682, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.7589795233299767e-05, | |
| "loss": 1.8225, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 1.691842900302115e-05, | |
| "loss": 1.7437, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 1.6247062772742532e-05, | |
| "loss": 1.7122, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.5575696542463918e-05, | |
| "loss": 1.739, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.4904330312185298e-05, | |
| "loss": 1.7283, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.4232964081906682e-05, | |
| "loss": 1.7479, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 1.3561597851628063e-05, | |
| "loss": 1.7429, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.2890231621349447e-05, | |
| "loss": 1.7446, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 1.2218865391070831e-05, | |
| "loss": 1.7533, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 1.1547499160792213e-05, | |
| "loss": 1.7455, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.0876132930513597e-05, | |
| "loss": 1.7368, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "eval_bleu1": 29.9409, | |
| "eval_bleu2": 17.5869, | |
| "eval_bleu3": 10.4195, | |
| "eval_bleu4": 6.1345, | |
| "eval_gen_len": 50.5967, | |
| "eval_loss": 2.613112688064575, | |
| "eval_rouge1": 35.7499, | |
| "eval_rouge2": 13.0188, | |
| "eval_rougeL": 23.5089, | |
| "eval_runtime": 67.2771, | |
| "eval_samples_per_second": 6.377, | |
| "eval_steps_per_second": 0.059, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 1.0204766700234978e-05, | |
| "loss": 1.7469, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 9.533400469956362e-06, | |
| "loss": 1.7526, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 8.862034239677744e-06, | |
| "loss": 1.7479, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 8.190668009399128e-06, | |
| "loss": 1.6487, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 7.5193017791205116e-06, | |
| "loss": 1.6374, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 6.847935548841894e-06, | |
| "loss": 1.652, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 6.176569318563277e-06, | |
| "loss": 1.6542, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 5.505203088284659e-06, | |
| "loss": 1.6443, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 4.833836858006043e-06, | |
| "loss": 1.6469, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 4.162470627727426e-06, | |
| "loss": 1.6569, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 3.4911043974488085e-06, | |
| "loss": 1.6499, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.8197381671701917e-06, | |
| "loss": 1.6694, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.1483719368915745e-06, | |
| "loss": 1.6437, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 1.4770057066129575e-06, | |
| "loss": 1.6704, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 8.056394763343404e-07, | |
| "loss": 1.6467, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 1.342732460557234e-07, | |
| "loss": 1.6622, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 6620, | |
| "total_flos": 4.347230642786304e+16, | |
| "train_loss": 2.009976102938465, | |
| "train_runtime": 3338.0012, | |
| "train_samples_per_second": 63.462, | |
| "train_steps_per_second": 1.983 | |
| } | |
| ], | |
| "max_steps": 6620, | |
| "num_train_epochs": 5, | |
| "total_flos": 4.347230642786304e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |