| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.03, | |
| "eval_steps": 1000, | |
| "global_step": 75, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 2.0489, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00013333333333333334, | |
| "loss": 1.8868, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 1.6718, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019722222222222225, | |
| "loss": 1.9564, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019444444444444446, | |
| "loss": 1.6946, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019166666666666667, | |
| "loss": 2.2475, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018888888888888888, | |
| "loss": 2.4487, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018611111111111112, | |
| "loss": 1.4369, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018333333333333334, | |
| "loss": 1.9525, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018055555555555557, | |
| "loss": 1.9572, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017777777777777779, | |
| "loss": 1.727, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000175, | |
| "loss": 1.9873, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017222222222222224, | |
| "loss": 1.7861, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00016944444444444445, | |
| "loss": 1.7475, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001666666666666667, | |
| "loss": 1.7548, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001638888888888889, | |
| "loss": 1.5408, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001611111111111111, | |
| "loss": 1.7227, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00015833333333333332, | |
| "loss": 2.6196, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00015555555555555556, | |
| "loss": 1.4832, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00015277777777777777, | |
| "loss": 1.9515, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 1.8943, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00014722222222222223, | |
| "loss": 1.7673, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00014444444444444444, | |
| "loss": 1.5981, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00014166666666666668, | |
| "loss": 1.9063, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001388888888888889, | |
| "loss": 1.5963, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00013611111111111113, | |
| "loss": 1.7514, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00013333333333333334, | |
| "loss": 1.4565, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00013055555555555555, | |
| "loss": 1.5058, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00012777777777777776, | |
| "loss": 1.6539, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000125, | |
| "loss": 1.6019, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00012222222222222224, | |
| "loss": 1.8105, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00011944444444444445, | |
| "loss": 1.4344, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00011666666666666668, | |
| "loss": 2.1983, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00011388888888888889, | |
| "loss": 1.6288, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00011111111111111112, | |
| "loss": 1.7846, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00010833333333333333, | |
| "loss": 1.6324, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00010555555555555557, | |
| "loss": 1.534, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00010277777777777778, | |
| "loss": 1.5642, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001, | |
| "loss": 2.209, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.722222222222223e-05, | |
| "loss": 1.4805, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.444444444444444e-05, | |
| "loss": 1.4624, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.166666666666667e-05, | |
| "loss": 1.7335, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.888888888888889e-05, | |
| "loss": 1.976, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.611111111111112e-05, | |
| "loss": 1.7132, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 1.8266, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.055555555555556e-05, | |
| "loss": 1.4724, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.777777777777778e-05, | |
| "loss": 1.9093, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 1.6004, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.222222222222222e-05, | |
| "loss": 1.8015, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.944444444444444e-05, | |
| "loss": 1.5109, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 1.8057, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.388888888888888e-05, | |
| "loss": 1.6599, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.111111111111112e-05, | |
| "loss": 1.6122, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.833333333333334e-05, | |
| "loss": 1.4713, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.555555555555556e-05, | |
| "loss": 1.5983, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.2777777777777784e-05, | |
| "loss": 1.7994, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7283, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.722222222222222e-05, | |
| "loss": 1.7313, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "loss": 1.9441, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.166666666666667e-05, | |
| "loss": 1.6845, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.888888888888889e-05, | |
| "loss": 1.753, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.611111111111111e-05, | |
| "loss": 1.6686, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 1.6792, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.055555555555556e-05, | |
| "loss": 1.4095, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.777777777777778e-05, | |
| "loss": 1.511, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.9625, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.2222222222222223e-05, | |
| "loss": 1.6922, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9444444444444445e-05, | |
| "loss": 1.6134, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 1.7452, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.388888888888889e-05, | |
| "loss": 1.7016, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.1111111111111112e-05, | |
| "loss": 2.2515, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 2.047, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.555555555555556e-06, | |
| "loss": 1.3849, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.777777777777778e-06, | |
| "loss": 1.4436, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0, | |
| "loss": 1.6971, | |
| "step": 75 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 75, | |
| "num_train_epochs": 1, | |
| "save_steps": 5, | |
| "total_flos": 4668793953730560.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |