| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.5968691500040696, | |
| "global_step": 44000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.886136363636363e-08, | |
| "loss": 0.0254, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.7725e-08, | |
| "loss": 0.0239, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.658863636363636e-08, | |
| "loss": 0.0247, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.545227272727273e-08, | |
| "loss": 0.0258, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.431590909090909e-08, | |
| "loss": 0.0233, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.318181818181817e-08, | |
| "loss": 0.0256, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.204545454545454e-08, | |
| "loss": 0.0253, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.09090909090909e-08, | |
| "loss": 0.0278, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 8.977272727272727e-08, | |
| "loss": 0.0255, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 8.863636363636363e-08, | |
| "loss": 0.0275, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 8.75e-08, | |
| "loss": 0.0251, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.636363636363636e-08, | |
| "loss": 0.027, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.522727272727273e-08, | |
| "loss": 0.0241, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.409090909090909e-08, | |
| "loss": 0.0247, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.295454545454546e-08, | |
| "loss": 0.0245, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.181818181818182e-08, | |
| "loss": 0.0236, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.068181818181817e-08, | |
| "loss": 0.0242, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.954545454545454e-08, | |
| "loss": 0.0257, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.841136363636363e-08, | |
| "loss": 0.0269, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.727727272727272e-08, | |
| "loss": 0.0271, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.614090909090909e-08, | |
| "loss": 0.0256, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.500454545454545e-08, | |
| "loss": 0.0269, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.386818181818182e-08, | |
| "loss": 0.0259, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.273181818181818e-08, | |
| "loss": 0.0246, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 7.159545454545454e-08, | |
| "loss": 0.0254, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.046136363636363e-08, | |
| "loss": 0.0272, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.9325e-08, | |
| "loss": 0.0269, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.818863636363636e-08, | |
| "loss": 0.0251, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.705227272727272e-08, | |
| "loss": 0.0262, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.591590909090909e-08, | |
| "loss": 0.023, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 6.477954545454545e-08, | |
| "loss": 0.0239, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 6.364318181818182e-08, | |
| "loss": 0.0257, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 6.250681818181818e-08, | |
| "loss": 0.024, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 6.137045454545455e-08, | |
| "loss": 0.0287, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 6.023409090909091e-08, | |
| "loss": 0.0253, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.910227272727273e-08, | |
| "loss": 0.0265, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5.7965909090909093e-08, | |
| "loss": 0.0255, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5.6829545454545445e-08, | |
| "loss": 0.024, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5.569318181818181e-08, | |
| "loss": 0.0258, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 5.4556818181818175e-08, | |
| "loss": 0.0255, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5.342045454545454e-08, | |
| "loss": 0.0249, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5.2284090909090905e-08, | |
| "loss": 0.0264, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.114772727272727e-08, | |
| "loss": 0.0251, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5.0011363636363634e-08, | |
| "loss": 0.0258, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.8875e-08, | |
| "loss": 0.0252, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.7740909090909086e-08, | |
| "loss": 0.0253, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.660454545454545e-08, | |
| "loss": 0.0254, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.5468181818181816e-08, | |
| "loss": 0.0259, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.433181818181818e-08, | |
| "loss": 0.0245, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.3195454545454546e-08, | |
| "loss": 0.0246, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.206136363636363e-08, | |
| "loss": 0.0262, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0925e-08, | |
| "loss": 0.0252, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.978863636363636e-08, | |
| "loss": 0.0248, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.865227272727273e-08, | |
| "loss": 0.025, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.7515909090909086e-08, | |
| "loss": 0.024, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.637954545454545e-08, | |
| "loss": 0.0277, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.5243181818181815e-08, | |
| "loss": 0.0235, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.410681818181818e-08, | |
| "loss": 0.0249, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.2974999999999996e-08, | |
| "loss": 0.0267, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.183863636363636e-08, | |
| "loss": 0.0247, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.0702272727272726e-08, | |
| "loss": 0.0251, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.956590909090909e-08, | |
| "loss": 0.0274, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.8429545454545452e-08, | |
| "loss": 0.0257, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.7293181818181814e-08, | |
| "loss": 0.0253, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.615681818181818e-08, | |
| "loss": 0.0258, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.5020454545454543e-08, | |
| "loss": 0.0249, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.3886363636363634e-08, | |
| "loss": 0.0249, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.275e-08, | |
| "loss": 0.0233, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.161590909090909e-08, | |
| "loss": 0.024, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.0479545454545454e-08, | |
| "loss": 0.0262, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.9343181818181815e-08, | |
| "loss": 0.025, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.820681818181818e-08, | |
| "loss": 0.0253, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.7070454545454545e-08, | |
| "loss": 0.0258, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.5938636363636364e-08, | |
| "loss": 0.0275, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.4802272727272727e-08, | |
| "loss": 0.025, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.3665909090909089e-08, | |
| "loss": 0.0267, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.2531818181818182e-08, | |
| "loss": 0.0277, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.1395454545454546e-08, | |
| "loss": 0.0255, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.0259090909090909e-08, | |
| "loss": 0.0255, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.122727272727272e-09, | |
| "loss": 0.0255, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 7.986363636363635e-09, | |
| "loss": 0.0246, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 6.852272727272727e-09, | |
| "loss": 0.0276, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.71590909090909e-09, | |
| "loss": 0.024, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.5795454545454544e-09, | |
| "loss": 0.0254, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.443181818181818e-09, | |
| "loss": 0.0232, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.3068181818181817e-09, | |
| "loss": 0.0263, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.1727272727272726e-09, | |
| "loss": 0.0254, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.636363636363636e-11, | |
| "loss": 0.0247, | |
| "step": 44000 | |
| } | |
| ], | |
| "max_steps": 44000, | |
| "num_train_epochs": 1, | |
| "total_flos": 4.597453878932275e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |