| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 145, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00013679786386078908, | |
| "loss": 3.2135, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00013559617012171195, | |
| "loss": 3.0835, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00013360900754314024, | |
| "loss": 3.0207, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00013085967378942767, | |
| "loss": 2.9809, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0001273804022850966, | |
| "loss": 3.0379, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00012321198430740717, | |
| "loss": 2.7644, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00011840329074592678, | |
| "loss": 2.8189, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00011301069913603339, | |
| "loss": 2.6556, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00010709743268385944, | |
| "loss": 2.9363, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00010073281903200561, | |
| "loss": 2.5794, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.399147745631814e-05, | |
| "loss": 2.4779, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.695244402310457e-05, | |
| "loss": 2.9374, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.969824496351968e-05, | |
| "loss": 2.6178, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.231392912895965e-05, | |
| "loss": 2.7956, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.488607087104036e-05, | |
| "loss": 2.7369, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.750175503648034e-05, | |
| "loss": 2.8983, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.0247555976895465e-05, | |
| "loss": 2.6613, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.320852254368187e-05, | |
| "loss": 2.8101, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.64671809679944e-05, | |
| "loss": 2.7433, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.0102567316140575e-05, | |
| "loss": 2.6317, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.418930086396663e-05, | |
| "loss": 2.8128, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.879670925407324e-05, | |
| "loss": 2.4389, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.3988015692592839e-05, | |
| "loss": 2.6243, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.819597714903392e-06, | |
| "loss": 2.6688, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.340326210572357e-06, | |
| "loss": 2.5304, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.590992456859759e-06, | |
| "loss": 2.6001, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.6038298782880553e-06, | |
| "loss": 2.5575, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.0213613921093164e-07, | |
| "loss": 2.6308, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 2.8741, | |
| "step": 145 | |
| } | |
| ], | |
| "max_steps": 145, | |
| "num_train_epochs": 1, | |
| "total_flos": 150765502464000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |