| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 50582, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001, | |
| "loss": 3.3233, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.99998927054457e-05, | |
| "loss": 2.8783, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.999957082224324e-05, | |
| "loss": 2.6917, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.999903435177409e-05, | |
| "loss": 2.7154, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.999828329634069e-05, | |
| "loss": 2.7366, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.999731765916636e-05, | |
| "loss": 2.7151, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.999613744439543e-05, | |
| "loss": 2.6935, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.999474265709312e-05, | |
| "loss": 2.6431, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.999313330324557e-05, | |
| "loss": 2.6171, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.999130938975975e-05, | |
| "loss": 2.6553, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.998927092446351e-05, | |
| "loss": 2.5905, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.99870179161055e-05, | |
| "loss": 2.5851, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.998455037435515e-05, | |
| "loss": 2.7004, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.998186830980259e-05, | |
| "loss": 2.5175, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.997897173395868e-05, | |
| "loss": 2.5745, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.997586065925489e-05, | |
| "loss": 2.515, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.997253509904324e-05, | |
| "loss": 2.5784, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.996899506759634e-05, | |
| "loss": 2.5426, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.996524058010725e-05, | |
| "loss": 2.5677, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.996127165268938e-05, | |
| "loss": 2.5871, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.995708830237652e-05, | |
| "loss": 2.6239, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.995269054712269e-05, | |
| "loss": 2.5366, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.994807840580211e-05, | |
| "loss": 2.5484, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.994325189820907e-05, | |
| "loss": 2.5512, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.99382110450579e-05, | |
| "loss": 2.5573, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.993295586798282e-05, | |
| "loss": 2.5584, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.992748638953795e-05, | |
| "loss": 2.5176, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.992180263319706e-05, | |
| "loss": 2.5322, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.991590462335362e-05, | |
| "loss": 2.4881, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.990979238532059e-05, | |
| "loss": 2.5357, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.990346594533036e-05, | |
| "loss": 2.4982, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.989692533053464e-05, | |
| "loss": 2.5111, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.989017056900434e-05, | |
| "loss": 2.5439, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.988320168972938e-05, | |
| "loss": 2.4806, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.987601872261871e-05, | |
| "loss": 2.4715, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 9.986862169850007e-05, | |
| "loss": 2.4796, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.986101064911984e-05, | |
| "loss": 2.5061, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.985318560714301e-05, | |
| "loss": 2.4945, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.984514660615293e-05, | |
| "loss": 2.4611, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.983689368065128e-05, | |
| "loss": 2.5054, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.98284268660578e-05, | |
| "loss": 2.5272, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.981974619871019e-05, | |
| "loss": 2.4692, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.981085171586402e-05, | |
| "loss": 2.4757, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.980174345569246e-05, | |
| "loss": 2.5493, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.979242145728618e-05, | |
| "loss": 2.4543, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.978288576065315e-05, | |
| "loss": 2.4532, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 9.977313640671853e-05, | |
| "loss": 2.4311, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.97631734373244e-05, | |
| "loss": 2.4512, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.975299689522967e-05, | |
| "loss": 2.46, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.974260682410984e-05, | |
| "loss": 2.5014, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_bleu": 1.0, | |
| "eval_brevity_penalty": 1.0, | |
| "eval_length_ratio": 1.0, | |
| "eval_loss": 2.3814520835876465, | |
| "eval_precisions": [ | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0 | |
| ], | |
| "eval_reference_length": 1439232, | |
| "eval_runtime": 2544.152, | |
| "eval_samples_per_second": 1.105, | |
| "eval_steps_per_second": 1.105, | |
| "eval_translation_length": 1439232, | |
| "step": 25291 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.973200326855681e-05, | |
| "loss": 2.3745, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.972118627407877e-05, | |
| "loss": 2.2548, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.971015588709987e-05, | |
| "loss": 2.1978, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.969891215496015e-05, | |
| "loss": 2.1931, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.968745512591524e-05, | |
| "loss": 2.1611, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.967578484913623e-05, | |
| "loss": 2.2166, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.96639013747094e-05, | |
| "loss": 2.1401, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 9.965180475363603e-05, | |
| "loss": 2.1456, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.963949503783216e-05, | |
| "loss": 2.2447, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.962697228012848e-05, | |
| "loss": 2.2138, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 9.961423653426987e-05, | |
| "loss": 2.1736, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.96012878549154e-05, | |
| "loss": 2.2721, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 9.958812629763798e-05, | |
| "loss": 2.1877, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.957475191892413e-05, | |
| "loss": 2.1888, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 9.956116477617382e-05, | |
| "loss": 2.1881, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 9.954736492770007e-05, | |
| "loss": 2.22, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 9.95333524327288e-05, | |
| "loss": 2.1357, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.951912735139864e-05, | |
| "loss": 2.2336, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 9.950468974476051e-05, | |
| "loss": 2.1933, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.949003967477749e-05, | |
| "loss": 2.2787, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 9.947517720432445e-05, | |
| "loss": 2.1992, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 9.946010239718791e-05, | |
| "loss": 2.2127, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.944481531806567e-05, | |
| "loss": 2.1832, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 9.942931603256652e-05, | |
| "loss": 2.226, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 9.941360460721e-05, | |
| "loss": 2.2111, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.939768110942617e-05, | |
| "loss": 2.233, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 9.938154560755516e-05, | |
| "loss": 2.2399, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.93651981708471e-05, | |
| "loss": 2.2297, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.934863886946158e-05, | |
| "loss": 2.2551, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 9.93318677744675e-05, | |
| "loss": 2.1899, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.931488495784278e-05, | |
| "loss": 2.118, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.929769049247397e-05, | |
| "loss": 2.2226, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.928028445215595e-05, | |
| "loss": 2.2213, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 9.926266691159165e-05, | |
| "loss": 2.1881, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 9.924483794639175e-05, | |
| "loss": 2.238, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 9.922679763307426e-05, | |
| "loss": 2.2178, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.920854604906427e-05, | |
| "loss": 2.1311, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 9.919008327269361e-05, | |
| "loss": 2.2308, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 9.917140938320049e-05, | |
| "loss": 2.2121, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 9.91525244607292e-05, | |
| "loss": 2.1756, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 9.913342858632968e-05, | |
| "loss": 2.2231, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 9.91141218419573e-05, | |
| "loss": 2.197, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.909460431047235e-05, | |
| "loss": 2.2141, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 9.907487607563988e-05, | |
| "loss": 2.2276, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 9.905493722212913e-05, | |
| "loss": 2.278, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 9.903478783551334e-05, | |
| "loss": 2.2759, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 9.90144280022693e-05, | |
| "loss": 2.2556, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 9.899385780977695e-05, | |
| "loss": 2.1704, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 9.89730773463191e-05, | |
| "loss": 2.2612, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 9.895208670108096e-05, | |
| "loss": 2.238, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9.893088596414981e-05, | |
| "loss": 2.2049, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bleu": 1.0, | |
| "eval_brevity_penalty": 1.0, | |
| "eval_length_ratio": 1.0, | |
| "eval_loss": 2.3368308544158936, | |
| "eval_precisions": [ | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0 | |
| ], | |
| "eval_reference_length": 1439232, | |
| "eval_runtime": 2375.7581, | |
| "eval_samples_per_second": 1.183, | |
| "eval_steps_per_second": 1.183, | |
| "eval_translation_length": 1439232, | |
| "step": 50582 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 758730, | |
| "num_train_epochs": 30, | |
| "save_steps": 500, | |
| "total_flos": 1.1657977248979354e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |