| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 28.960817717206133, | |
| "global_step": 34000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.369363916964232e-05, | |
| "loss": 0.5598, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.218362282878412e-05, | |
| "loss": 0.5495, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.067058040307452e-05, | |
| "loss": 0.5008, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.915753797736489e-05, | |
| "loss": 0.4756, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 8.764449555165527e-05, | |
| "loss": 0.4689, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 8.613447921079707e-05, | |
| "loss": 0.4223, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 8.462143678508746e-05, | |
| "loss": 0.4444, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 8.310839435937784e-05, | |
| "loss": 0.3734, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 8.159535193366822e-05, | |
| "loss": 0.4001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 8.00823095079586e-05, | |
| "loss": 0.3875, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 7.856926708224899e-05, | |
| "loss": 0.3557, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 7.705622465653937e-05, | |
| "loss": 0.34, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 7.554318223082976e-05, | |
| "loss": 0.3244, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 7.403013980512014e-05, | |
| "loss": 0.3372, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 7.251709737941052e-05, | |
| "loss": 0.3099, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 7.10040549537009e-05, | |
| "loss": 0.3108, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 6.949101252799129e-05, | |
| "loss": 0.2705, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 6.797797010228167e-05, | |
| "loss": 0.2856, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 6.646795376142348e-05, | |
| "loss": 0.2885, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 6.495491133571386e-05, | |
| "loss": 0.2729, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 8.94, | |
| "learning_rate": 6.344186891000424e-05, | |
| "loss": 0.274, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 6.192882648429462e-05, | |
| "loss": 0.2465, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 6.041881014343642e-05, | |
| "loss": 0.2517, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 10.22, | |
| "learning_rate": 5.890576771772681e-05, | |
| "loss": 0.2512, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 10.65, | |
| "learning_rate": 5.739272529201719e-05, | |
| "loss": 0.239, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 11.07, | |
| "learning_rate": 5.587968286630757e-05, | |
| "loss": 0.2302, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 11.5, | |
| "learning_rate": 5.436966652544938e-05, | |
| "loss": 0.2256, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 5.285662409973976e-05, | |
| "loss": 0.2259, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 12.35, | |
| "learning_rate": 5.1343581674030136e-05, | |
| "loss": 0.206, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 12.78, | |
| "learning_rate": 4.9830539248320526e-05, | |
| "loss": 0.2197, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "learning_rate": 4.831749682261091e-05, | |
| "loss": 0.2008, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 4.680445439690129e-05, | |
| "loss": 0.1843, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 14.05, | |
| "learning_rate": 4.5291411971191675e-05, | |
| "loss": 0.1896, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 14.48, | |
| "learning_rate": 4.377836954548206e-05, | |
| "loss": 0.184, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 14.91, | |
| "learning_rate": 4.226532711977244e-05, | |
| "loss": 0.1788, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 15.33, | |
| "learning_rate": 4.0752284694062824e-05, | |
| "loss": 0.174, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 15.76, | |
| "learning_rate": 3.923924226835321e-05, | |
| "loss": 0.177, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 16.18, | |
| "learning_rate": 3.772619984264359e-05, | |
| "loss": 0.161, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 16.61, | |
| "learning_rate": 3.6213157416933974e-05, | |
| "loss": 0.1548, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 3.470616716092719e-05, | |
| "loss": 0.1719, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 17.46, | |
| "learning_rate": 3.3193124735217576e-05, | |
| "loss": 0.159, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 17.89, | |
| "learning_rate": 3.168008230950796e-05, | |
| "loss": 0.1466, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 18.31, | |
| "learning_rate": 3.0167039883798342e-05, | |
| "loss": 0.153, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 18.74, | |
| "learning_rate": 2.865399745808873e-05, | |
| "loss": 0.1399, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 19.17, | |
| "learning_rate": 2.7140955032379112e-05, | |
| "loss": 0.1369, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 19.59, | |
| "learning_rate": 2.5630938691520913e-05, | |
| "loss": 0.1397, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 2.4117896265811293e-05, | |
| "loss": 0.1351, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 20.44, | |
| "learning_rate": 2.260485384010168e-05, | |
| "loss": 0.1256, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 20.87, | |
| "learning_rate": 2.109181141439206e-05, | |
| "loss": 0.128, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 21.29, | |
| "learning_rate": 1.9581795073533863e-05, | |
| "loss": 0.1184, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 21.72, | |
| "learning_rate": 1.8068752647824246e-05, | |
| "loss": 0.1312, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 22.15, | |
| "learning_rate": 1.655571022211463e-05, | |
| "loss": 0.1138, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 22.57, | |
| "learning_rate": 1.5042667796405011e-05, | |
| "loss": 0.1123, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 1.3529625370695396e-05, | |
| "loss": 0.1144, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 23.42, | |
| "learning_rate": 1.2019609029837197e-05, | |
| "loss": 0.1076, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 23.85, | |
| "learning_rate": 1.0509592688979e-05, | |
| "loss": 0.1054, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 24.28, | |
| "learning_rate": 8.996550263269383e-06, | |
| "loss": 0.1033, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 24.7, | |
| "learning_rate": 7.483507837559765e-06, | |
| "loss": 0.0993, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 25.13, | |
| "learning_rate": 5.970465411850148e-06, | |
| "loss": 0.1, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 25.55, | |
| "learning_rate": 4.457422986140532e-06, | |
| "loss": 0.0985, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 25.98, | |
| "learning_rate": 2.9443805604309144e-06, | |
| "loss": 0.0928, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 26.41, | |
| "learning_rate": 1.4313381347212977e-06, | |
| "loss": 0.0913, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 26.83, | |
| "learning_rate": 0.0, | |
| "loss": 0.0947, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 27.26, | |
| "learning_rate": 0.0, | |
| "loss": 0.1028, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 27.68, | |
| "learning_rate": 0.0, | |
| "loss": 0.0895, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 28.11, | |
| "learning_rate": 0.0, | |
| "loss": 0.0894, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 28.53, | |
| "learning_rate": 0.0, | |
| "loss": 0.0894, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 28.96, | |
| "learning_rate": 0.0, | |
| "loss": 0.0924, | |
| "step": 34000 | |
| } | |
| ], | |
| "max_steps": 34046, | |
| "num_train_epochs": 29, | |
| "total_flos": 9.909309665241907e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |