| { |
| "best_metric": 1.691104769706726, |
| "best_model_checkpoint": "./test-mlm/checkpoint-8000", |
| "epoch": 10.0, |
| "global_step": 8420, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.59, |
| "learning_rate": 0.00047030878859857485, |
| "loss": 2.5064, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.59, |
| "eval_accuracy": 0.5923440629322982, |
| "eval_loss": 2.000737190246582, |
| "eval_runtime": 40.8995, |
| "eval_samples_per_second": 15.99, |
| "eval_steps_per_second": 2.005, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 0.0004406175771971497, |
| "loss": 2.3261, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.19, |
| "eval_accuracy": 0.6004155844155844, |
| "eval_loss": 1.9369549751281738, |
| "eval_runtime": 40.9751, |
| "eval_samples_per_second": 15.961, |
| "eval_steps_per_second": 2.001, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 0.00041092636579572443, |
| "loss": 2.2689, |
| "step": 1500 |
| }, |
| { |
| "epoch": 1.78, |
| "eval_accuracy": 0.6049467180826803, |
| "eval_loss": 1.9199129343032837, |
| "eval_runtime": 40.8894, |
| "eval_samples_per_second": 15.994, |
| "eval_steps_per_second": 2.005, |
| "step": 1500 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 0.00038123515439429927, |
| "loss": 2.2101, |
| "step": 2000 |
| }, |
| { |
| "epoch": 2.38, |
| "eval_accuracy": 0.6167493796526055, |
| "eval_loss": 1.8502798080444336, |
| "eval_runtime": 41.0954, |
| "eval_samples_per_second": 15.914, |
| "eval_steps_per_second": 1.995, |
| "step": 2000 |
| }, |
| { |
| "epoch": 2.97, |
| "learning_rate": 0.0003515439429928741, |
| "loss": 2.2015, |
| "step": 2500 |
| }, |
| { |
| "epoch": 2.97, |
| "eval_accuracy": 0.6148145072557039, |
| "eval_loss": 1.858150601387024, |
| "eval_runtime": 40.9285, |
| "eval_samples_per_second": 15.979, |
| "eval_steps_per_second": 2.003, |
| "step": 2500 |
| }, |
| { |
| "epoch": 3.56, |
| "learning_rate": 0.00032185273159144896, |
| "loss": 2.1468, |
| "step": 3000 |
| }, |
| { |
| "epoch": 3.56, |
| "eval_accuracy": 0.6186465352300141, |
| "eval_loss": 1.8364105224609375, |
| "eval_runtime": 40.9793, |
| "eval_samples_per_second": 15.959, |
| "eval_steps_per_second": 2.001, |
| "step": 3000 |
| }, |
| { |
| "epoch": 4.16, |
| "learning_rate": 0.00029216152019002374, |
| "loss": 2.1213, |
| "step": 3500 |
| }, |
| { |
| "epoch": 4.16, |
| "eval_accuracy": 0.6203441651000888, |
| "eval_loss": 1.8127774000167847, |
| "eval_runtime": 40.94, |
| "eval_samples_per_second": 15.975, |
| "eval_steps_per_second": 2.003, |
| "step": 3500 |
| }, |
| { |
| "epoch": 4.75, |
| "learning_rate": 0.0002624703087885986, |
| "loss": 2.0938, |
| "step": 4000 |
| }, |
| { |
| "epoch": 4.75, |
| "eval_accuracy": 0.6253609489581818, |
| "eval_loss": 1.788784384727478, |
| "eval_runtime": 41.0869, |
| "eval_samples_per_second": 15.917, |
| "eval_steps_per_second": 1.996, |
| "step": 4000 |
| }, |
| { |
| "epoch": 5.34, |
| "learning_rate": 0.0002327790973871734, |
| "loss": 2.0888, |
| "step": 4500 |
| }, |
| { |
| "epoch": 5.34, |
| "eval_accuracy": 0.628534333525678, |
| "eval_loss": 1.7835372686386108, |
| "eval_runtime": 41.0712, |
| "eval_samples_per_second": 15.924, |
| "eval_steps_per_second": 1.997, |
| "step": 4500 |
| }, |
| { |
| "epoch": 5.94, |
| "learning_rate": 0.00020308788598574824, |
| "loss": 2.0619, |
| "step": 5000 |
| }, |
| { |
| "epoch": 5.94, |
| "eval_accuracy": 0.6320160722422434, |
| "eval_loss": 1.7468773126602173, |
| "eval_runtime": 41.0197, |
| "eval_samples_per_second": 15.944, |
| "eval_steps_per_second": 1.999, |
| "step": 5000 |
| }, |
| { |
| "epoch": 6.53, |
| "learning_rate": 0.00017339667458432303, |
| "loss": 2.045, |
| "step": 5500 |
| }, |
| { |
| "epoch": 6.53, |
| "eval_accuracy": 0.6350336264873254, |
| "eval_loss": 1.7321710586547852, |
| "eval_runtime": 41.1615, |
| "eval_samples_per_second": 15.889, |
| "eval_steps_per_second": 1.992, |
| "step": 5500 |
| }, |
| { |
| "epoch": 7.13, |
| "learning_rate": 0.00014370546318289787, |
| "loss": 2.032, |
| "step": 6000 |
| }, |
| { |
| "epoch": 7.13, |
| "eval_accuracy": 0.6363391096814361, |
| "eval_loss": 1.7278571128845215, |
| "eval_runtime": 41.1852, |
| "eval_samples_per_second": 15.88, |
| "eval_steps_per_second": 1.991, |
| "step": 6000 |
| }, |
| { |
| "epoch": 7.72, |
| "learning_rate": 0.00011401425178147269, |
| "loss": 2.012, |
| "step": 6500 |
| }, |
| { |
| "epoch": 7.72, |
| "eval_accuracy": 0.6403126628452319, |
| "eval_loss": 1.7044711112976074, |
| "eval_runtime": 41.1592, |
| "eval_samples_per_second": 15.89, |
| "eval_steps_per_second": 1.992, |
| "step": 6500 |
| }, |
| { |
| "epoch": 8.31, |
| "learning_rate": 8.432304038004752e-05, |
| "loss": 1.9992, |
| "step": 7000 |
| }, |
| { |
| "epoch": 8.31, |
| "eval_accuracy": 0.6397965988662888, |
| "eval_loss": 1.7062609195709229, |
| "eval_runtime": 40.9227, |
| "eval_samples_per_second": 15.981, |
| "eval_steps_per_second": 2.004, |
| "step": 7000 |
| }, |
| { |
| "epoch": 8.91, |
| "learning_rate": 5.463182897862233e-05, |
| "loss": 1.993, |
| "step": 7500 |
| }, |
| { |
| "epoch": 8.91, |
| "eval_accuracy": 0.640505019145193, |
| "eval_loss": 1.7070534229278564, |
| "eval_runtime": 41.0391, |
| "eval_samples_per_second": 15.936, |
| "eval_steps_per_second": 1.998, |
| "step": 7500 |
| }, |
| { |
| "epoch": 9.5, |
| "learning_rate": 2.494061757719715e-05, |
| "loss": 1.9774, |
| "step": 8000 |
| }, |
| { |
| "epoch": 9.5, |
| "eval_accuracy": 0.6419018087855297, |
| "eval_loss": 1.691104769706726, |
| "eval_runtime": 41.0068, |
| "eval_samples_per_second": 15.949, |
| "eval_steps_per_second": 2.0, |
| "step": 8000 |
| }, |
| { |
| "epoch": 10.0, |
| "step": 8420, |
| "total_flos": 1.798091044383744e+16, |
| "train_loss": 2.122876556468973, |
| "train_runtime": 10299.32, |
| "train_samples_per_second": 6.535, |
| "train_steps_per_second": 0.818 |
| } |
| ], |
| "max_steps": 8420, |
| "num_train_epochs": 10, |
| "total_flos": 1.798091044383744e+16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
