| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 0, | |
| "global_step": 325, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003076923076923077, | |
| "grad_norm": 0.06884765625, | |
| "learning_rate": 9.96923076923077e-06, | |
| "loss": 1.3291, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.006153846153846154, | |
| "grad_norm": 0.07568359375, | |
| "learning_rate": 9.93846153846154e-06, | |
| "loss": 1.4234, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.009230769230769232, | |
| "grad_norm": 0.07080078125, | |
| "learning_rate": 9.907692307692309e-06, | |
| "loss": 1.3606, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.012307692307692308, | |
| "grad_norm": 0.07080078125, | |
| "learning_rate": 9.876923076923077e-06, | |
| "loss": 1.3638, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.015384615384615385, | |
| "grad_norm": 0.07373046875, | |
| "learning_rate": 9.846153846153848e-06, | |
| "loss": 1.3257, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.018461538461538463, | |
| "grad_norm": 0.078125, | |
| "learning_rate": 9.815384615384617e-06, | |
| "loss": 1.3549, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.021538461538461538, | |
| "grad_norm": 0.07763671875, | |
| "learning_rate": 9.784615384615387e-06, | |
| "loss": 1.3114, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.024615384615384615, | |
| "grad_norm": 0.08447265625, | |
| "learning_rate": 9.753846153846154e-06, | |
| "loss": 1.3653, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.027692307692307693, | |
| "grad_norm": 0.0859375, | |
| "learning_rate": 9.723076923076924e-06, | |
| "loss": 1.3901, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03076923076923077, | |
| "grad_norm": 0.08740234375, | |
| "learning_rate": 9.692307692307693e-06, | |
| "loss": 1.3658, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.033846153846153845, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 9.661538461538462e-06, | |
| "loss": 1.3952, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.036923076923076927, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 9.630769230769232e-06, | |
| "loss": 1.4576, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 1.4002, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.043076923076923075, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 9.569230769230769e-06, | |
| "loss": 1.3855, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.046153846153846156, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 9.53846153846154e-06, | |
| "loss": 1.4135, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04923076923076923, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 9.50769230769231e-06, | |
| "loss": 1.3229, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.052307692307692305, | |
| "grad_norm": 0.119140625, | |
| "learning_rate": 9.476923076923079e-06, | |
| "loss": 1.4684, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.055384615384615386, | |
| "grad_norm": 0.11767578125, | |
| "learning_rate": 9.446153846153847e-06, | |
| "loss": 1.3685, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05846153846153846, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 9.415384615384616e-06, | |
| "loss": 1.36, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.06153846153846154, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 9.384615384615385e-06, | |
| "loss": 1.3234, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06461538461538462, | |
| "grad_norm": 0.12451171875, | |
| "learning_rate": 9.353846153846155e-06, | |
| "loss": 1.3713, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06769230769230769, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 9.323076923076924e-06, | |
| "loss": 1.3827, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.07076923076923076, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 9.292307692307694e-06, | |
| "loss": 1.251, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.07384615384615385, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 9.261538461538461e-06, | |
| "loss": 1.3138, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07692307692307693, | |
| "grad_norm": 0.1337890625, | |
| "learning_rate": 9.230769230769232e-06, | |
| "loss": 1.3053, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.146484375, | |
| "learning_rate": 9.200000000000002e-06, | |
| "loss": 1.3592, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.08307692307692308, | |
| "grad_norm": 0.138671875, | |
| "learning_rate": 9.169230769230771e-06, | |
| "loss": 1.3216, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.08615384615384615, | |
| "grad_norm": 0.1416015625, | |
| "learning_rate": 9.138461538461539e-06, | |
| "loss": 1.3195, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08923076923076922, | |
| "grad_norm": 0.158203125, | |
| "learning_rate": 9.107692307692308e-06, | |
| "loss": 1.3961, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.09230769230769231, | |
| "grad_norm": 0.310546875, | |
| "learning_rate": 9.076923076923078e-06, | |
| "loss": 1.4095, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09538461538461539, | |
| "grad_norm": 0.1640625, | |
| "learning_rate": 9.046153846153847e-06, | |
| "loss": 1.3865, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.09846153846153846, | |
| "grad_norm": 0.1552734375, | |
| "learning_rate": 9.015384615384616e-06, | |
| "loss": 1.3679, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.10153846153846154, | |
| "grad_norm": 0.1572265625, | |
| "learning_rate": 8.984615384615386e-06, | |
| "loss": 1.3037, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.10461538461538461, | |
| "grad_norm": 0.169921875, | |
| "learning_rate": 8.953846153846153e-06, | |
| "loss": 1.3945, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1076923076923077, | |
| "grad_norm": 0.171875, | |
| "learning_rate": 8.923076923076925e-06, | |
| "loss": 1.3785, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.11076923076923077, | |
| "grad_norm": 0.173828125, | |
| "learning_rate": 8.892307692307694e-06, | |
| "loss": 1.3318, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.11384615384615385, | |
| "grad_norm": 0.16796875, | |
| "learning_rate": 8.861538461538463e-06, | |
| "loss": 1.299, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.11692307692307692, | |
| "grad_norm": 0.189453125, | |
| "learning_rate": 8.830769230769231e-06, | |
| "loss": 1.4298, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.169921875, | |
| "learning_rate": 8.8e-06, | |
| "loss": 1.2916, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.12307692307692308, | |
| "grad_norm": 0.17578125, | |
| "learning_rate": 8.76923076923077e-06, | |
| "loss": 1.3395, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12615384615384614, | |
| "grad_norm": 0.1708984375, | |
| "learning_rate": 8.73846153846154e-06, | |
| "loss": 1.2629, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.12923076923076923, | |
| "grad_norm": 0.1953125, | |
| "learning_rate": 8.707692307692309e-06, | |
| "loss": 1.4473, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.13230769230769232, | |
| "grad_norm": 0.185546875, | |
| "learning_rate": 8.676923076923078e-06, | |
| "loss": 1.3714, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.13538461538461538, | |
| "grad_norm": 0.201171875, | |
| "learning_rate": 8.646153846153846e-06, | |
| "loss": 1.4174, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.13846153846153847, | |
| "grad_norm": 0.181640625, | |
| "learning_rate": 8.615384615384617e-06, | |
| "loss": 1.2992, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.14153846153846153, | |
| "grad_norm": 0.1767578125, | |
| "learning_rate": 8.584615384615386e-06, | |
| "loss": 1.2352, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.14461538461538462, | |
| "grad_norm": 0.1904296875, | |
| "learning_rate": 8.553846153846156e-06, | |
| "loss": 1.3411, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.1476923076923077, | |
| "grad_norm": 0.1806640625, | |
| "learning_rate": 8.523076923076923e-06, | |
| "loss": 1.263, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.15076923076923077, | |
| "grad_norm": 0.1943359375, | |
| "learning_rate": 8.492307692307693e-06, | |
| "loss": 1.3598, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.15384615384615385, | |
| "grad_norm": 0.189453125, | |
| "learning_rate": 8.461538461538462e-06, | |
| "loss": 1.2865, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15692307692307692, | |
| "grad_norm": 0.189453125, | |
| "learning_rate": 8.430769230769231e-06, | |
| "loss": 1.2995, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.189453125, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.3397, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.16307692307692306, | |
| "grad_norm": 0.1826171875, | |
| "learning_rate": 8.36923076923077e-06, | |
| "loss": 1.2509, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.16615384615384615, | |
| "grad_norm": 0.1845703125, | |
| "learning_rate": 8.338461538461538e-06, | |
| "loss": 1.2763, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.16923076923076924, | |
| "grad_norm": 0.177734375, | |
| "learning_rate": 8.307692307692309e-06, | |
| "loss": 1.2342, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1723076923076923, | |
| "grad_norm": 0.1845703125, | |
| "learning_rate": 8.276923076923078e-06, | |
| "loss": 1.2823, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.1753846153846154, | |
| "grad_norm": 0.1748046875, | |
| "learning_rate": 8.246153846153848e-06, | |
| "loss": 1.2181, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.17846153846153845, | |
| "grad_norm": 0.18359375, | |
| "learning_rate": 8.215384615384616e-06, | |
| "loss": 1.2414, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.18153846153846154, | |
| "grad_norm": 0.1845703125, | |
| "learning_rate": 8.184615384615385e-06, | |
| "loss": 1.2893, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.18461538461538463, | |
| "grad_norm": 0.193359375, | |
| "learning_rate": 8.153846153846154e-06, | |
| "loss": 1.2908, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.18769230769230769, | |
| "grad_norm": 0.197265625, | |
| "learning_rate": 8.123076923076924e-06, | |
| "loss": 1.3198, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.19076923076923077, | |
| "grad_norm": 0.1640625, | |
| "learning_rate": 8.092307692307693e-06, | |
| "loss": 1.1349, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.19384615384615383, | |
| "grad_norm": 0.1904296875, | |
| "learning_rate": 8.061538461538463e-06, | |
| "loss": 1.2279, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.19692307692307692, | |
| "grad_norm": 0.1787109375, | |
| "learning_rate": 8.03076923076923e-06, | |
| "loss": 1.2149, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.1796875, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.2458, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.20307692307692307, | |
| "grad_norm": 0.1904296875, | |
| "learning_rate": 7.96923076923077e-06, | |
| "loss": 1.2909, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.20615384615384616, | |
| "grad_norm": 0.1826171875, | |
| "learning_rate": 7.93846153846154e-06, | |
| "loss": 1.2465, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.20923076923076922, | |
| "grad_norm": 0.1923828125, | |
| "learning_rate": 7.907692307692308e-06, | |
| "loss": 1.3355, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2123076923076923, | |
| "grad_norm": 0.169921875, | |
| "learning_rate": 7.876923076923077e-06, | |
| "loss": 1.1948, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2153846153846154, | |
| "grad_norm": 0.1748046875, | |
| "learning_rate": 7.846153846153847e-06, | |
| "loss": 1.1997, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.21846153846153846, | |
| "grad_norm": 0.193359375, | |
| "learning_rate": 7.815384615384616e-06, | |
| "loss": 1.2977, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.22153846153846155, | |
| "grad_norm": 0.1826171875, | |
| "learning_rate": 7.784615384615385e-06, | |
| "loss": 1.232, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2246153846153846, | |
| "grad_norm": 0.1865234375, | |
| "learning_rate": 7.753846153846155e-06, | |
| "loss": 1.2615, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2276923076923077, | |
| "grad_norm": 0.1689453125, | |
| "learning_rate": 7.723076923076924e-06, | |
| "loss": 1.1747, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.23076923076923078, | |
| "grad_norm": 0.1787109375, | |
| "learning_rate": 7.692307692307694e-06, | |
| "loss": 1.2431, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.23384615384615384, | |
| "grad_norm": 0.16015625, | |
| "learning_rate": 7.661538461538463e-06, | |
| "loss": 1.1652, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.23692307692307693, | |
| "grad_norm": 0.1728515625, | |
| "learning_rate": 7.630769230769232e-06, | |
| "loss": 1.2095, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.173828125, | |
| "learning_rate": 7.600000000000001e-06, | |
| "loss": 1.2248, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.24307692307692308, | |
| "grad_norm": 0.162109375, | |
| "learning_rate": 7.5692307692307695e-06, | |
| "loss": 1.161, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.24615384615384617, | |
| "grad_norm": 0.169921875, | |
| "learning_rate": 7.538461538461539e-06, | |
| "loss": 1.2049, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.24923076923076923, | |
| "grad_norm": 0.171875, | |
| "learning_rate": 7.507692307692308e-06, | |
| "loss": 1.1334, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.2523076923076923, | |
| "grad_norm": 0.416015625, | |
| "learning_rate": 7.476923076923077e-06, | |
| "loss": 1.2324, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.2553846153846154, | |
| "grad_norm": 0.166015625, | |
| "learning_rate": 7.446153846153846e-06, | |
| "loss": 1.1635, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.25846153846153846, | |
| "grad_norm": 0.166015625, | |
| "learning_rate": 7.4153846153846164e-06, | |
| "loss": 1.165, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.26153846153846155, | |
| "grad_norm": 0.177734375, | |
| "learning_rate": 7.384615384615386e-06, | |
| "loss": 1.2228, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.26461538461538464, | |
| "grad_norm": 0.1494140625, | |
| "learning_rate": 7.353846153846154e-06, | |
| "loss": 1.1623, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.2676923076923077, | |
| "grad_norm": 0.177734375, | |
| "learning_rate": 7.323076923076924e-06, | |
| "loss": 1.2558, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.27076923076923076, | |
| "grad_norm": 0.1796875, | |
| "learning_rate": 7.292307692307693e-06, | |
| "loss": 1.2074, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.27384615384615385, | |
| "grad_norm": 0.16015625, | |
| "learning_rate": 7.261538461538462e-06, | |
| "loss": 1.1246, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.27692307692307694, | |
| "grad_norm": 0.169921875, | |
| "learning_rate": 7.230769230769231e-06, | |
| "loss": 1.1639, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.1806640625, | |
| "learning_rate": 7.2000000000000005e-06, | |
| "loss": 1.2529, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.28307692307692306, | |
| "grad_norm": 0.162109375, | |
| "learning_rate": 7.169230769230769e-06, | |
| "loss": 1.1334, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.28615384615384615, | |
| "grad_norm": 0.1787109375, | |
| "learning_rate": 7.1384615384615385e-06, | |
| "loss": 1.2247, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.28923076923076924, | |
| "grad_norm": 0.16796875, | |
| "learning_rate": 7.107692307692309e-06, | |
| "loss": 1.2117, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2923076923076923, | |
| "grad_norm": 0.189453125, | |
| "learning_rate": 7.076923076923078e-06, | |
| "loss": 1.2659, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2953846153846154, | |
| "grad_norm": 0.146484375, | |
| "learning_rate": 7.046153846153847e-06, | |
| "loss": 1.0678, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.29846153846153844, | |
| "grad_norm": 0.1630859375, | |
| "learning_rate": 7.015384615384616e-06, | |
| "loss": 1.1595, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.30153846153846153, | |
| "grad_norm": 0.1435546875, | |
| "learning_rate": 6.9846153846153854e-06, | |
| "loss": 1.0941, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.3046153846153846, | |
| "grad_norm": 0.142578125, | |
| "learning_rate": 6.953846153846154e-06, | |
| "loss": 1.0786, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.3076923076923077, | |
| "grad_norm": 0.16015625, | |
| "learning_rate": 6.923076923076923e-06, | |
| "loss": 1.158, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.31076923076923074, | |
| "grad_norm": 0.140625, | |
| "learning_rate": 6.892307692307693e-06, | |
| "loss": 1.1376, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.31384615384615383, | |
| "grad_norm": 0.158203125, | |
| "learning_rate": 6.861538461538461e-06, | |
| "loss": 1.1953, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.3169230769230769, | |
| "grad_norm": 0.1376953125, | |
| "learning_rate": 6.830769230769231e-06, | |
| "loss": 1.0991, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 1.0287, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.3230769230769231, | |
| "grad_norm": 0.1591796875, | |
| "learning_rate": 6.76923076923077e-06, | |
| "loss": 1.1518, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3261538461538461, | |
| "grad_norm": 0.1455078125, | |
| "learning_rate": 6.738461538461539e-06, | |
| "loss": 1.167, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3292307692307692, | |
| "grad_norm": 0.1962890625, | |
| "learning_rate": 6.707692307692308e-06, | |
| "loss": 1.2052, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.3323076923076923, | |
| "grad_norm": 0.1640625, | |
| "learning_rate": 6.676923076923078e-06, | |
| "loss": 1.1857, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.3353846153846154, | |
| "grad_norm": 0.1572265625, | |
| "learning_rate": 6.646153846153846e-06, | |
| "loss": 1.1613, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.3384615384615385, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 6.615384615384616e-06, | |
| "loss": 1.0379, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3415384615384615, | |
| "grad_norm": 0.2158203125, | |
| "learning_rate": 6.584615384615385e-06, | |
| "loss": 1.2315, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.3446153846153846, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 6.553846153846154e-06, | |
| "loss": 1.1071, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.3476923076923077, | |
| "grad_norm": 0.1435546875, | |
| "learning_rate": 6.523076923076923e-06, | |
| "loss": 1.1522, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.3507692307692308, | |
| "grad_norm": 0.12255859375, | |
| "learning_rate": 6.492307692307693e-06, | |
| "loss": 1.0503, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.35384615384615387, | |
| "grad_norm": 0.1435546875, | |
| "learning_rate": 6.461538461538463e-06, | |
| "loss": 1.1644, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.3569230769230769, | |
| "grad_norm": 0.140625, | |
| "learning_rate": 6.430769230769231e-06, | |
| "loss": 1.1099, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.12255859375, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 1.0527, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.3630769230769231, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 6.36923076923077e-06, | |
| "loss": 1.0771, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.36615384615384616, | |
| "grad_norm": 0.158203125, | |
| "learning_rate": 6.3384615384615385e-06, | |
| "loss": 1.1872, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.36923076923076925, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 6.307692307692308e-06, | |
| "loss": 1.0479, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3723076923076923, | |
| "grad_norm": 0.146484375, | |
| "learning_rate": 6.276923076923077e-06, | |
| "loss": 1.1463, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.37538461538461537, | |
| "grad_norm": 0.12451171875, | |
| "learning_rate": 6.246153846153846e-06, | |
| "loss": 1.0811, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.37846153846153846, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 6.215384615384615e-06, | |
| "loss": 1.0748, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.38153846153846155, | |
| "grad_norm": 0.142578125, | |
| "learning_rate": 6.1846153846153855e-06, | |
| "loss": 1.1335, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 0.1357421875, | |
| "learning_rate": 6.153846153846155e-06, | |
| "loss": 1.0854, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.38769230769230767, | |
| "grad_norm": 0.1376953125, | |
| "learning_rate": 6.123076923076923e-06, | |
| "loss": 1.085, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.39076923076923076, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 6.092307692307693e-06, | |
| "loss": 1.0989, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.39384615384615385, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 6.061538461538462e-06, | |
| "loss": 1.0747, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.39692307692307693, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 6.030769230769231e-06, | |
| "loss": 1.1073, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 6e-06, | |
| "loss": 1.0372, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.40307692307692305, | |
| "grad_norm": 0.134765625, | |
| "learning_rate": 5.9692307692307695e-06, | |
| "loss": 1.1499, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.40615384615384614, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 5.938461538461538e-06, | |
| "loss": 1.0192, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.40923076923076923, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 5.907692307692308e-06, | |
| "loss": 1.0566, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.4123076923076923, | |
| "grad_norm": 0.1357421875, | |
| "learning_rate": 5.876923076923078e-06, | |
| "loss": 1.1429, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.4153846153846154, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 5.846153846153847e-06, | |
| "loss": 1.1077, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.41846153846153844, | |
| "grad_norm": 0.138671875, | |
| "learning_rate": 5.815384615384616e-06, | |
| "loss": 1.1425, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.42153846153846153, | |
| "grad_norm": 0.1533203125, | |
| "learning_rate": 5.784615384615385e-06, | |
| "loss": 1.2069, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.4246153846153846, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 5.7538461538461545e-06, | |
| "loss": 1.1044, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.4276923076923077, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 5.723076923076923e-06, | |
| "loss": 1.0747, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.4307692307692308, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 5.692307692307692e-06, | |
| "loss": 1.0655, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.4338461538461538, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 5.661538461538462e-06, | |
| "loss": 1.0625, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.4369230769230769, | |
| "grad_norm": 0.12255859375, | |
| "learning_rate": 5.63076923076923e-06, | |
| "loss": 1.081, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 1.0705, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.4430769230769231, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 5.56923076923077e-06, | |
| "loss": 1.0234, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.4461538461538462, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 5.538461538461539e-06, | |
| "loss": 1.0902, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.4492307692307692, | |
| "grad_norm": 0.123046875, | |
| "learning_rate": 5.507692307692308e-06, | |
| "loss": 1.0407, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.4523076923076923, | |
| "grad_norm": 0.1484375, | |
| "learning_rate": 5.476923076923077e-06, | |
| "loss": 1.0833, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.4553846153846154, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 5.446153846153847e-06, | |
| "loss": 1.0868, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.4584615384615385, | |
| "grad_norm": 0.1162109375, | |
| "learning_rate": 5.415384615384615e-06, | |
| "loss": 1.0271, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.46153846153846156, | |
| "grad_norm": 0.1162109375, | |
| "learning_rate": 5.384615384615385e-06, | |
| "loss": 1.0906, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4646153846153846, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 5.353846153846154e-06, | |
| "loss": 1.0597, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.4676923076923077, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 5.323076923076923e-06, | |
| "loss": 1.0745, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.4707692307692308, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 5.292307692307693e-06, | |
| "loss": 1.1018, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.47384615384615386, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 5.261538461538462e-06, | |
| "loss": 1.0792, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.47692307692307695, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 5.230769230769232e-06, | |
| "loss": 1.0688, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 5.2e-06, | |
| "loss": 1.0883, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.48307692307692307, | |
| "grad_norm": 0.1357421875, | |
| "learning_rate": 5.16923076923077e-06, | |
| "loss": 1.0957, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.48615384615384616, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 5.138461538461539e-06, | |
| "loss": 1.0953, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.48923076923076925, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 5.1076923076923075e-06, | |
| "loss": 1.0632, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.49230769230769234, | |
| "grad_norm": 0.12109375, | |
| "learning_rate": 5.076923076923077e-06, | |
| "loss": 1.0575, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.49538461538461537, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 5.046153846153846e-06, | |
| "loss": 1.0383, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.49846153846153846, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 5.015384615384616e-06, | |
| "loss": 1.0679, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.5015384615384615, | |
| "grad_norm": 0.134765625, | |
| "learning_rate": 4.984615384615385e-06, | |
| "loss": 1.0539, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.5046153846153846, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 4.9538461538461545e-06, | |
| "loss": 1.106, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.5076923076923077, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 4.923076923076924e-06, | |
| "loss": 1.0879, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.5107692307692308, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 4.892307692307693e-06, | |
| "loss": 1.0629, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.5138461538461538, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 4.861538461538462e-06, | |
| "loss": 1.012, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.5169230769230769, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 4.830769230769231e-06, | |
| "loss": 1.0965, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 1.0352, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.5230769230769231, | |
| "grad_norm": 0.1201171875, | |
| "learning_rate": 4.76923076923077e-06, | |
| "loss": 1.0569, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5261538461538462, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 4.738461538461539e-06, | |
| "loss": 1.0371, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.5292307692307693, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 4.707692307692308e-06, | |
| "loss": 0.9938, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.5323076923076923, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 4.676923076923077e-06, | |
| "loss": 1.067, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.5353846153846153, | |
| "grad_norm": 0.140625, | |
| "learning_rate": 4.646153846153847e-06, | |
| "loss": 1.0799, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.5384615384615384, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 4.615384615384616e-06, | |
| "loss": 1.0737, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.5415384615384615, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 4.5846153846153855e-06, | |
| "loss": 1.039, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.5446153846153846, | |
| "grad_norm": 0.12060546875, | |
| "learning_rate": 4.553846153846154e-06, | |
| "loss": 0.9962, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.5476923076923077, | |
| "grad_norm": 0.1328125, | |
| "learning_rate": 4.5230769230769235e-06, | |
| "loss": 1.0563, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.5507692307692308, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 4.492307692307693e-06, | |
| "loss": 1.04, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.5538461538461539, | |
| "grad_norm": 0.11767578125, | |
| "learning_rate": 4.461538461538462e-06, | |
| "loss": 1.0855, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.556923076923077, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 4.430769230769232e-06, | |
| "loss": 1.0169, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 4.4e-06, | |
| "loss": 1.0089, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.563076923076923, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 4.36923076923077e-06, | |
| "loss": 1.079, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.5661538461538461, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 4.338461538461539e-06, | |
| "loss": 1.0521, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5692307692307692, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 4.307692307692308e-06, | |
| "loss": 1.0283, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5723076923076923, | |
| "grad_norm": 0.11328125, | |
| "learning_rate": 4.276923076923078e-06, | |
| "loss": 1.0415, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.5753846153846154, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 4.246153846153846e-06, | |
| "loss": 1.08, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.5784615384615385, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 4.215384615384616e-06, | |
| "loss": 1.0083, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5815384615384616, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 4.184615384615385e-06, | |
| "loss": 1.0441, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.5846153846153846, | |
| "grad_norm": 0.1328125, | |
| "learning_rate": 4.1538461538461545e-06, | |
| "loss": 1.0127, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5876923076923077, | |
| "grad_norm": 0.1201171875, | |
| "learning_rate": 4.123076923076924e-06, | |
| "loss": 1.0215, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.5907692307692308, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 4.0923076923076925e-06, | |
| "loss": 1.0584, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.5938461538461538, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 4.061538461538462e-06, | |
| "loss": 1.0506, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.5969230769230769, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 4.030769230769231e-06, | |
| "loss": 1.0189, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.9369, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.6030769230769231, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 3.96923076923077e-06, | |
| "loss": 1.1064, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.6061538461538462, | |
| "grad_norm": 0.11328125, | |
| "learning_rate": 3.938461538461539e-06, | |
| "loss": 0.9687, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.6092307692307692, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 3.907692307692308e-06, | |
| "loss": 1.0229, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.6123076923076923, | |
| "grad_norm": 0.119140625, | |
| "learning_rate": 3.876923076923077e-06, | |
| "loss": 1.0919, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.6153846153846154, | |
| "grad_norm": 0.12060546875, | |
| "learning_rate": 3.846153846153847e-06, | |
| "loss": 1.0418, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.6184615384615385, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 3.815384615384616e-06, | |
| "loss": 1.0216, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.6215384615384615, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 3.7846153846153847e-06, | |
| "loss": 1.0254, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.6246153846153846, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 3.753846153846154e-06, | |
| "loss": 1.03, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.6276923076923077, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 3.723076923076923e-06, | |
| "loss": 1.0292, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.6307692307692307, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 3.692307692307693e-06, | |
| "loss": 1.0453, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.6338461538461538, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 3.661538461538462e-06, | |
| "loss": 1.0143, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.6369230769230769, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 3.630769230769231e-06, | |
| "loss": 1.0696, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 0.9804, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.6430769230769231, | |
| "grad_norm": 0.212890625, | |
| "learning_rate": 3.5692307692307692e-06, | |
| "loss": 0.9922, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.6461538461538462, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 3.538461538461539e-06, | |
| "loss": 1.0517, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.6492307692307693, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 3.507692307692308e-06, | |
| "loss": 1.018, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.6523076923076923, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 3.476923076923077e-06, | |
| "loss": 1.0607, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.6553846153846153, | |
| "grad_norm": 0.11767578125, | |
| "learning_rate": 3.4461538461538464e-06, | |
| "loss": 0.991, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.6584615384615384, | |
| "grad_norm": 0.1435546875, | |
| "learning_rate": 3.4153846153846154e-06, | |
| "loss": 1.1046, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.6615384615384615, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 3.384615384615385e-06, | |
| "loss": 1.018, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.6646153846153846, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 3.353846153846154e-06, | |
| "loss": 1.0372, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.6676923076923077, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 3.323076923076923e-06, | |
| "loss": 1.03, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.6707692307692308, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 3.2923076923076925e-06, | |
| "loss": 1.0037, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.6738461538461539, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 3.2615384615384615e-06, | |
| "loss": 0.9792, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.676923076923077, | |
| "grad_norm": 0.123046875, | |
| "learning_rate": 3.2307692307692313e-06, | |
| "loss": 1.0411, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 1.0648, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.683076923076923, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 3.1692307692307693e-06, | |
| "loss": 1.0251, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.6861538461538461, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 3.1384615384615386e-06, | |
| "loss": 0.9741, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.6892307692307692, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 3.1076923076923076e-06, | |
| "loss": 1.0144, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.6923076923076923, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 3.0769230769230774e-06, | |
| "loss": 0.978, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6953846153846154, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 3.0461538461538464e-06, | |
| "loss": 1.009, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.6984615384615385, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 3.0153846153846154e-06, | |
| "loss": 0.9263, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.7015384615384616, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 2.9846153846153848e-06, | |
| "loss": 1.0537, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.7046153846153846, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 2.953846153846154e-06, | |
| "loss": 1.0333, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.7076923076923077, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 2.9230769230769236e-06, | |
| "loss": 0.9739, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.7107692307692308, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 2.8923076923076925e-06, | |
| "loss": 1.0017, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.7138461538461538, | |
| "grad_norm": 0.1337890625, | |
| "learning_rate": 2.8615384615384615e-06, | |
| "loss": 1.0938, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.7169230769230769, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 2.830769230769231e-06, | |
| "loss": 1.0264, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.11767578125, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 1.0092, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.7230769230769231, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 2.7692307692307697e-06, | |
| "loss": 0.9835, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.7261538461538461, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 2.7384615384615387e-06, | |
| "loss": 0.989, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.7292307692307692, | |
| "grad_norm": 0.134765625, | |
| "learning_rate": 2.7076923076923076e-06, | |
| "loss": 0.9958, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.7323076923076923, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 2.676923076923077e-06, | |
| "loss": 0.9686, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.7353846153846154, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 2.6461538461538464e-06, | |
| "loss": 1.0506, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.7384615384615385, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 2.615384615384616e-06, | |
| "loss": 1.0255, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7415384615384616, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 2.584615384615385e-06, | |
| "loss": 1.0417, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.7446153846153846, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 2.5538461538461538e-06, | |
| "loss": 1.0581, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.7476923076923077, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 2.523076923076923e-06, | |
| "loss": 1.06, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.7507692307692307, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 2.4923076923076926e-06, | |
| "loss": 0.9942, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.7538461538461538, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 2.461538461538462e-06, | |
| "loss": 1.0632, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.7569230769230769, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 2.430769230769231e-06, | |
| "loss": 1.0558, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.1162109375, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.9939, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.7630769230769231, | |
| "grad_norm": 0.1494140625, | |
| "learning_rate": 2.3692307692307697e-06, | |
| "loss": 1.047, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.7661538461538462, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 2.3384615384615387e-06, | |
| "loss": 1.0724, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 2.307692307692308e-06, | |
| "loss": 0.9348, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.7723076923076924, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 2.276923076923077e-06, | |
| "loss": 0.9966, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.7753846153846153, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 2.2461538461538464e-06, | |
| "loss": 0.9458, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.7784615384615384, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 2.215384615384616e-06, | |
| "loss": 1.0455, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.7815384615384615, | |
| "grad_norm": 0.12060546875, | |
| "learning_rate": 2.184615384615385e-06, | |
| "loss": 1.0082, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.7846153846153846, | |
| "grad_norm": 0.1435546875, | |
| "learning_rate": 2.153846153846154e-06, | |
| "loss": 1.0946, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.7876923076923077, | |
| "grad_norm": 0.119140625, | |
| "learning_rate": 2.123076923076923e-06, | |
| "loss": 1.0232, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.7907692307692308, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 2.0923076923076926e-06, | |
| "loss": 0.9456, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.7938461538461539, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 2.061538461538462e-06, | |
| "loss": 0.9781, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.796923076923077, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 2.030769230769231e-06, | |
| "loss": 1.0856, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.0517, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.803076923076923, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 1.9692307692307693e-06, | |
| "loss": 1.0173, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.8061538461538461, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 1.9384615384615387e-06, | |
| "loss": 1.0389, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.8092307692307692, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 1.907692307692308e-06, | |
| "loss": 0.9772, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.8123076923076923, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 1.876923076923077e-06, | |
| "loss": 1.0962, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.8153846153846154, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 1.8461538461538465e-06, | |
| "loss": 1.01, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.8184615384615385, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 1.8153846153846154e-06, | |
| "loss": 1.0083, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.8215384615384616, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 1.7846153846153846e-06, | |
| "loss": 1.0316, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.8246153846153846, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 1.753846153846154e-06, | |
| "loss": 1.0572, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.8276923076923077, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 1.7230769230769232e-06, | |
| "loss": 1.0571, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.8307692307692308, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 1.6923076923076926e-06, | |
| "loss": 1.0114, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.8338461538461538, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 1.6615384615384616e-06, | |
| "loss": 1.0088, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.8369230769230769, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 1.6307692307692307e-06, | |
| "loss": 1.0166, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 0.9437, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.8430769230769231, | |
| "grad_norm": 0.154296875, | |
| "learning_rate": 1.5692307692307693e-06, | |
| "loss": 1.0557, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.8461538461538461, | |
| "grad_norm": 0.1630859375, | |
| "learning_rate": 1.5384615384615387e-06, | |
| "loss": 1.0855, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.8492307692307692, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 1.5076923076923077e-06, | |
| "loss": 0.98, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.8523076923076923, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 1.476923076923077e-06, | |
| "loss": 1.0534, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.8553846153846154, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 1.4461538461538463e-06, | |
| "loss": 0.9787, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.8584615384615385, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 1.4153846153846155e-06, | |
| "loss": 1.0394, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.8615384615384616, | |
| "grad_norm": 0.11328125, | |
| "learning_rate": 1.3846153846153848e-06, | |
| "loss": 0.9793, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.8646153846153846, | |
| "grad_norm": 0.12060546875, | |
| "learning_rate": 1.3538461538461538e-06, | |
| "loss": 1.0406, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.8676923076923077, | |
| "grad_norm": 0.125, | |
| "learning_rate": 1.3230769230769232e-06, | |
| "loss": 1.0027, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.8707692307692307, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 1.2923076923076924e-06, | |
| "loss": 0.9823, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.8738461538461538, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 1.2615384615384616e-06, | |
| "loss": 0.9552, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.8769230769230769, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 1.230769230769231e-06, | |
| "loss": 0.9968, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 0.9435, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.8830769230769231, | |
| "grad_norm": 0.123046875, | |
| "learning_rate": 1.1692307692307693e-06, | |
| "loss": 1.0399, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.8861538461538462, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 1.1384615384615385e-06, | |
| "loss": 0.9963, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.8892307692307693, | |
| "grad_norm": 0.12451171875, | |
| "learning_rate": 1.107692307692308e-06, | |
| "loss": 1.0521, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.8923076923076924, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 1.076923076923077e-06, | |
| "loss": 1.0567, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.8953846153846153, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 1.0461538461538463e-06, | |
| "loss": 0.997, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.8984615384615384, | |
| "grad_norm": 0.1494140625, | |
| "learning_rate": 1.0153846153846155e-06, | |
| "loss": 1.0735, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.9015384615384615, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 9.846153846153847e-07, | |
| "loss": 1.0334, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.9046153846153846, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 9.53846153846154e-07, | |
| "loss": 0.9653, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.9076923076923077, | |
| "grad_norm": 0.1328125, | |
| "learning_rate": 9.230769230769232e-07, | |
| "loss": 1.0564, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.9107692307692308, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 8.923076923076923e-07, | |
| "loss": 1.0533, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.9138461538461539, | |
| "grad_norm": 0.12109375, | |
| "learning_rate": 8.615384615384616e-07, | |
| "loss": 0.9847, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.916923076923077, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 8.307692307692308e-07, | |
| "loss": 0.9929, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 1.0549, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.9230769230769231, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 7.692307692307694e-07, | |
| "loss": 1.0345, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.9261538461538461, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 7.384615384615385e-07, | |
| "loss": 1.0308, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.9292307692307692, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 7.076923076923077e-07, | |
| "loss": 1.0138, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.9323076923076923, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 6.769230769230769e-07, | |
| "loss": 0.9941, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.9353846153846154, | |
| "grad_norm": 0.1162109375, | |
| "learning_rate": 6.461538461538462e-07, | |
| "loss": 1.0061, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.9384615384615385, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 6.153846153846155e-07, | |
| "loss": 0.966, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.9415384615384615, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 5.846153846153847e-07, | |
| "loss": 0.9981, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.9446153846153846, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 5.53846153846154e-07, | |
| "loss": 0.954, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.9476923076923077, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 5.230769230769231e-07, | |
| "loss": 0.9617, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.9507692307692308, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 4.923076923076923e-07, | |
| "loss": 1.0047, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.9538461538461539, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 4.615384615384616e-07, | |
| "loss": 1.0354, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.9569230769230769, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 4.307692307692308e-07, | |
| "loss": 1.0476, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.1376953125, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 1.0779, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.963076923076923, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 3.6923076923076927e-07, | |
| "loss": 1.0165, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.9661538461538461, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 3.3846153846153845e-07, | |
| "loss": 0.9314, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.9692307692307692, | |
| "grad_norm": 0.11328125, | |
| "learning_rate": 3.0769230769230774e-07, | |
| "loss": 0.9821, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.9723076923076923, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 2.76923076923077e-07, | |
| "loss": 0.9925, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.9753846153846154, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 2.4615384615384616e-07, | |
| "loss": 1.0104, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.9784615384615385, | |
| "grad_norm": 0.1396484375, | |
| "learning_rate": 2.153846153846154e-07, | |
| "loss": 1.0548, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.9815384615384616, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 1.8461538461538464e-07, | |
| "loss": 0.9428, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.9846153846153847, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 1.5384615384615387e-07, | |
| "loss": 1.0099, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.9876923076923076, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 1.2307692307692308e-07, | |
| "loss": 1.0038, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.9907692307692307, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 9.230769230769232e-08, | |
| "loss": 1.0144, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.9938461538461538, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 6.153846153846154e-08, | |
| "loss": 0.9642, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.9969230769230769, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 3.076923076923077e-08, | |
| "loss": 1.0002, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 0.0, | |
| "loss": 0.9957, | |
| "step": 325 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 325, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 0, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.0200780709811978e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |