| { | |
| "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": 9.063444108761328e-06, | |
| "loss": 5.7175, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8126888217522657e-05, | |
| "loss": 2.8477, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.719033232628399e-05, | |
| "loss": 2.6157, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.625377643504531e-05, | |
| "loss": 2.494, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.531722054380665e-05, | |
| "loss": 2.4171, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.438066465256798e-05, | |
| "loss": 2.3849, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.961732124874119e-05, | |
| "loss": 2.3699, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.8610271903323265e-05, | |
| "loss": 2.3347, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.7603222557905336e-05, | |
| "loss": 2.3025, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5.6596173212487415e-05, | |
| "loss": 2.2882, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5.558912386706949e-05, | |
| "loss": 2.2799, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.458207452165156e-05, | |
| "loss": 2.2598, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.357502517623364e-05, | |
| "loss": 2.1986, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5.2567975830815716e-05, | |
| "loss": 2.0755, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.156092648539779e-05, | |
| "loss": 2.0341, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.055387713997986e-05, | |
| "loss": 2.0281, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.954682779456194e-05, | |
| "loss": 2.0348, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.853977844914401e-05, | |
| "loss": 2.0219, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.753272910372608e-05, | |
| "loss": 2.0266, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.6525679758308154e-05, | |
| "loss": 2.0429, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.551863041289023e-05, | |
| "loss": 2.0147, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.4511581067472304e-05, | |
| "loss": 2.0208, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.350453172205438e-05, | |
| "loss": 1.9841, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.249748237663646e-05, | |
| "loss": 1.9728, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.149043303121853e-05, | |
| "loss": 1.924, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.0483383685800605e-05, | |
| "loss": 1.9596, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 3.947633434038268e-05, | |
| "loss": 1.8471, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 3.8469284994964755e-05, | |
| "loss": 1.7412, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.746223564954683e-05, | |
| "loss": 1.7375, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 3.64551863041289e-05, | |
| "loss": 1.7486, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3.544813695871098e-05, | |
| "loss": 1.7393, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.444108761329305e-05, | |
| "loss": 1.7279, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 3.343403826787513e-05, | |
| "loss": 1.7482, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 3.2426988922457206e-05, | |
| "loss": 1.7409, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 3.141993957703928e-05, | |
| "loss": 1.756, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 3.041289023162135e-05, | |
| "loss": 1.765, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 2.9405840886203425e-05, | |
| "loss": 1.746, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.83987915407855e-05, | |
| "loss": 1.745, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 2.7391742195367572e-05, | |
| "loss": 1.7377, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 2.6384692849949647e-05, | |
| "loss": 1.6794, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.5377643504531723e-05, | |
| "loss": 1.5671, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.4370594159113798e-05, | |
| "loss": 1.5392, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 2.3363544813695873e-05, | |
| "loss": 1.5624, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 2.2356495468277945e-05, | |
| "loss": 1.5562, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 2.134944612286002e-05, | |
| "loss": 1.5765, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 2.0342396777442095e-05, | |
| "loss": 1.5726, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.933534743202417e-05, | |
| "loss": 1.5744, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 1.8328298086606246e-05, | |
| "loss": 1.5826, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 1.7321248741188317e-05, | |
| "loss": 1.5756, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.6314199395770393e-05, | |
| "loss": 1.5664, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "eval_bleu1": 30.0048, | |
| "eval_bleu2": 17.5297, | |
| "eval_bleu3": 10.3153, | |
| "eval_bleu4": 5.9092, | |
| "eval_gen_len": 50.8508, | |
| "eval_loss": 2.610959053039551, | |
| "eval_rouge1": 35.8879, | |
| "eval_rouge2": 12.9302, | |
| "eval_rougeL": 23.7819, | |
| "eval_runtime": 75.2321, | |
| "eval_samples_per_second": 5.702, | |
| "eval_steps_per_second": 0.053, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 1.5307150050352468e-05, | |
| "loss": 1.5826, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.4300100704934543e-05, | |
| "loss": 1.581, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.3293051359516617e-05, | |
| "loss": 1.578, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 1.228600201409869e-05, | |
| "loss": 1.4451, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 1.1278952668680767e-05, | |
| "loss": 1.4347, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 1.027190332326284e-05, | |
| "loss": 1.4428, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 9.264853977844916e-06, | |
| "loss": 1.4495, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 8.257804632426989e-06, | |
| "loss": 1.4419, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 7.250755287009064e-06, | |
| "loss": 1.4394, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 6.243705941591138e-06, | |
| "loss": 1.4542, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 5.236656596173212e-06, | |
| "loss": 1.4499, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 4.229607250755287e-06, | |
| "loss": 1.463, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 3.2225579053373617e-06, | |
| "loss": 1.442, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.215508559919436e-06, | |
| "loss": 1.4651, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 1.2084592145015106e-06, | |
| "loss": 1.4429, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 2.014098690835851e-07, | |
| "loss": 1.4553, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 6620, | |
| "total_flos": 4.347230642786304e+16, | |
| "train_loss": 1.8827527066371952, | |
| "train_runtime": 3344.9943, | |
| "train_samples_per_second": 63.329, | |
| "train_steps_per_second": 1.979 | |
| } | |
| ], | |
| "max_steps": 6620, | |
| "num_train_epochs": 5, | |
| "total_flos": 4.347230642786304e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |