| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.0, | |
| "eval_steps": 0, | |
| "global_step": 1300, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003076923076923077, | |
| "grad_norm": 0.07861328125, | |
| "learning_rate": 9.992307692307694e-06, | |
| "loss": 1.3766, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.006153846153846154, | |
| "grad_norm": 0.07080078125, | |
| "learning_rate": 9.984615384615386e-06, | |
| "loss": 1.3768, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.009230769230769232, | |
| "grad_norm": 0.0732421875, | |
| "learning_rate": 9.976923076923077e-06, | |
| "loss": 1.3894, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.012307692307692308, | |
| "grad_norm": 0.07373046875, | |
| "learning_rate": 9.96923076923077e-06, | |
| "loss": 1.4846, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.015384615384615385, | |
| "grad_norm": 0.07177734375, | |
| "learning_rate": 9.961538461538463e-06, | |
| "loss": 1.2595, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.018461538461538463, | |
| "grad_norm": 0.07861328125, | |
| "learning_rate": 9.953846153846156e-06, | |
| "loss": 1.4687, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.021538461538461538, | |
| "grad_norm": 0.080078125, | |
| "learning_rate": 9.946153846153847e-06, | |
| "loss": 1.38, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.024615384615384615, | |
| "grad_norm": 0.08447265625, | |
| "learning_rate": 9.93846153846154e-06, | |
| "loss": 1.4664, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.027692307692307693, | |
| "grad_norm": 0.07861328125, | |
| "learning_rate": 9.930769230769232e-06, | |
| "loss": 1.3434, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03076923076923077, | |
| "grad_norm": 0.08349609375, | |
| "learning_rate": 9.923076923076923e-06, | |
| "loss": 1.337, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.033846153846153845, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 9.915384615384616e-06, | |
| "loss": 1.4769, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.036923076923076927, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 9.907692307692309e-06, | |
| "loss": 1.389, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 9.9e-06, | |
| "loss": 1.383, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.043076923076923075, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 9.892307692307693e-06, | |
| "loss": 1.3875, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.046153846153846156, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 9.884615384615386e-06, | |
| "loss": 1.3463, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04923076923076923, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 9.876923076923077e-06, | |
| "loss": 1.5405, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.052307692307692305, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 9.86923076923077e-06, | |
| "loss": 1.3709, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.055384615384615386, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 9.861538461538462e-06, | |
| "loss": 1.2591, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05846153846153846, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 9.853846153846155e-06, | |
| "loss": 1.3693, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.06153846153846154, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 9.846153846153848e-06, | |
| "loss": 1.4028, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06461538461538462, | |
| "grad_norm": 0.1396484375, | |
| "learning_rate": 9.838461538461539e-06, | |
| "loss": 1.5402, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06769230769230769, | |
| "grad_norm": 0.125, | |
| "learning_rate": 9.830769230769232e-06, | |
| "loss": 1.3354, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.07076923076923076, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 9.823076923076924e-06, | |
| "loss": 1.3587, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.07384615384615385, | |
| "grad_norm": 0.146484375, | |
| "learning_rate": 9.815384615384617e-06, | |
| "loss": 1.4719, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07692307692307693, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 9.807692307692308e-06, | |
| "loss": 1.2647, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.1533203125, | |
| "learning_rate": 9.800000000000001e-06, | |
| "loss": 1.4166, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.08307692307692308, | |
| "grad_norm": 0.146484375, | |
| "learning_rate": 9.792307692307694e-06, | |
| "loss": 1.3832, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.08615384615384615, | |
| "grad_norm": 0.1455078125, | |
| "learning_rate": 9.784615384615387e-06, | |
| "loss": 1.297, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08923076923076922, | |
| "grad_norm": 0.1484375, | |
| "learning_rate": 9.776923076923078e-06, | |
| "loss": 1.3337, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.09230769230769231, | |
| "grad_norm": 0.1513671875, | |
| "learning_rate": 9.76923076923077e-06, | |
| "loss": 1.3031, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09538461538461539, | |
| "grad_norm": 0.1708984375, | |
| "learning_rate": 9.761538461538462e-06, | |
| "loss": 1.4593, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.09846153846153846, | |
| "grad_norm": 0.1728515625, | |
| "learning_rate": 9.753846153846154e-06, | |
| "loss": 1.4258, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.10153846153846154, | |
| "grad_norm": 0.1708984375, | |
| "learning_rate": 9.746153846153847e-06, | |
| "loss": 1.4554, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.10461538461538461, | |
| "grad_norm": 0.18359375, | |
| "learning_rate": 9.738461538461538e-06, | |
| "loss": 1.4829, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1076923076923077, | |
| "grad_norm": 0.177734375, | |
| "learning_rate": 9.730769230769231e-06, | |
| "loss": 1.3776, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.11076923076923077, | |
| "grad_norm": 0.1806640625, | |
| "learning_rate": 9.723076923076924e-06, | |
| "loss": 1.3531, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.11384615384615385, | |
| "grad_norm": 0.1787109375, | |
| "learning_rate": 9.715384615384616e-06, | |
| "loss": 1.3573, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.11692307692307692, | |
| "grad_norm": 0.189453125, | |
| "learning_rate": 9.707692307692308e-06, | |
| "loss": 1.3863, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.19140625, | |
| "learning_rate": 9.7e-06, | |
| "loss": 1.3953, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.12307692307692308, | |
| "grad_norm": 0.1787109375, | |
| "learning_rate": 9.692307692307693e-06, | |
| "loss": 1.2809, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12615384615384614, | |
| "grad_norm": 0.1728515625, | |
| "learning_rate": 9.684615384615386e-06, | |
| "loss": 1.2503, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.12923076923076923, | |
| "grad_norm": 0.1865234375, | |
| "learning_rate": 9.676923076923079e-06, | |
| "loss": 1.3506, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.13230769230769232, | |
| "grad_norm": 0.18359375, | |
| "learning_rate": 9.66923076923077e-06, | |
| "loss": 1.2746, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.13538461538461538, | |
| "grad_norm": 0.1865234375, | |
| "learning_rate": 9.661538461538462e-06, | |
| "loss": 1.2677, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.13846153846153847, | |
| "grad_norm": 0.1923828125, | |
| "learning_rate": 9.653846153846155e-06, | |
| "loss": 1.2677, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.14153846153846153, | |
| "grad_norm": 0.2021484375, | |
| "learning_rate": 9.646153846153848e-06, | |
| "loss": 1.387, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.14461538461538462, | |
| "grad_norm": 0.19921875, | |
| "learning_rate": 9.638461538461539e-06, | |
| "loss": 1.4374, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.1476923076923077, | |
| "grad_norm": 0.203125, | |
| "learning_rate": 9.630769230769232e-06, | |
| "loss": 1.3077, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.15076923076923077, | |
| "grad_norm": 0.1953125, | |
| "learning_rate": 9.623076923076925e-06, | |
| "loss": 1.3261, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.15384615384615385, | |
| "grad_norm": 0.19140625, | |
| "learning_rate": 9.615384615384616e-06, | |
| "loss": 1.2521, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15692307692307692, | |
| "grad_norm": 0.1953125, | |
| "learning_rate": 9.607692307692308e-06, | |
| "loss": 1.2529, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.1982421875, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 1.2926, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.16307692307692306, | |
| "grad_norm": 0.19140625, | |
| "learning_rate": 9.592307692307692e-06, | |
| "loss": 1.2683, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.16615384615384615, | |
| "grad_norm": 0.1962890625, | |
| "learning_rate": 9.584615384615385e-06, | |
| "loss": 1.3059, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.16923076923076924, | |
| "grad_norm": 0.1962890625, | |
| "learning_rate": 9.576923076923078e-06, | |
| "loss": 1.3085, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1723076923076923, | |
| "grad_norm": 0.1904296875, | |
| "learning_rate": 9.569230769230769e-06, | |
| "loss": 1.2002, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.1753846153846154, | |
| "grad_norm": 0.1962890625, | |
| "learning_rate": 9.561538461538462e-06, | |
| "loss": 1.2754, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.17846153846153845, | |
| "grad_norm": 0.1943359375, | |
| "learning_rate": 9.553846153846155e-06, | |
| "loss": 1.2916, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.18153846153846154, | |
| "grad_norm": 0.19140625, | |
| "learning_rate": 9.546153846153847e-06, | |
| "loss": 1.2647, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.18461538461538463, | |
| "grad_norm": 0.19921875, | |
| "learning_rate": 9.53846153846154e-06, | |
| "loss": 1.3039, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.18769230769230769, | |
| "grad_norm": 0.1748046875, | |
| "learning_rate": 9.530769230769231e-06, | |
| "loss": 1.2012, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.19076923076923077, | |
| "grad_norm": 0.1953125, | |
| "learning_rate": 9.523076923076924e-06, | |
| "loss": 1.3133, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.19384615384615383, | |
| "grad_norm": 0.197265625, | |
| "learning_rate": 9.515384615384617e-06, | |
| "loss": 1.2921, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.19692307692307692, | |
| "grad_norm": 0.203125, | |
| "learning_rate": 9.50769230769231e-06, | |
| "loss": 1.2733, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.177734375, | |
| "learning_rate": 9.5e-06, | |
| "loss": 1.2117, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.20307692307692307, | |
| "grad_norm": 0.158203125, | |
| "learning_rate": 9.492307692307693e-06, | |
| "loss": 1.0862, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.20615384615384616, | |
| "grad_norm": 0.1962890625, | |
| "learning_rate": 9.484615384615386e-06, | |
| "loss": 1.3095, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.20923076923076922, | |
| "grad_norm": 0.1962890625, | |
| "learning_rate": 9.476923076923079e-06, | |
| "loss": 1.3025, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2123076923076923, | |
| "grad_norm": 0.1796875, | |
| "learning_rate": 9.46923076923077e-06, | |
| "loss": 1.178, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2153846153846154, | |
| "grad_norm": 0.1787109375, | |
| "learning_rate": 9.461538461538463e-06, | |
| "loss": 1.1691, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.21846153846153846, | |
| "grad_norm": 0.173828125, | |
| "learning_rate": 9.453846153846154e-06, | |
| "loss": 1.1692, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.22153846153846155, | |
| "grad_norm": 0.17578125, | |
| "learning_rate": 9.446153846153847e-06, | |
| "loss": 1.2111, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2246153846153846, | |
| "grad_norm": 0.18359375, | |
| "learning_rate": 9.43846153846154e-06, | |
| "loss": 1.21, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2276923076923077, | |
| "grad_norm": 0.1767578125, | |
| "learning_rate": 9.43076923076923e-06, | |
| "loss": 1.1997, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.23076923076923078, | |
| "grad_norm": 0.1875, | |
| "learning_rate": 9.423076923076923e-06, | |
| "loss": 1.1826, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.23384615384615384, | |
| "grad_norm": 0.177734375, | |
| "learning_rate": 9.415384615384616e-06, | |
| "loss": 1.2182, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.23692307692307693, | |
| "grad_norm": 0.1494140625, | |
| "learning_rate": 9.407692307692309e-06, | |
| "loss": 1.0469, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.16015625, | |
| "learning_rate": 9.4e-06, | |
| "loss": 1.1651, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.24307692307692308, | |
| "grad_norm": 0.1708984375, | |
| "learning_rate": 9.392307692307693e-06, | |
| "loss": 1.1454, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.24615384615384617, | |
| "grad_norm": 0.1640625, | |
| "learning_rate": 9.384615384615385e-06, | |
| "loss": 1.1784, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.24923076923076923, | |
| "grad_norm": 0.1865234375, | |
| "learning_rate": 9.376923076923078e-06, | |
| "loss": 1.2164, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.2523076923076923, | |
| "grad_norm": 0.150390625, | |
| "learning_rate": 9.369230769230771e-06, | |
| "loss": 1.0507, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.2553846153846154, | |
| "grad_norm": 0.1611328125, | |
| "learning_rate": 9.361538461538462e-06, | |
| "loss": 1.2013, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.25846153846153846, | |
| "grad_norm": 0.1650390625, | |
| "learning_rate": 9.353846153846155e-06, | |
| "loss": 1.1708, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.26153846153846155, | |
| "grad_norm": 0.1630859375, | |
| "learning_rate": 9.346153846153847e-06, | |
| "loss": 1.1342, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.26461538461538464, | |
| "grad_norm": 0.146484375, | |
| "learning_rate": 9.33846153846154e-06, | |
| "loss": 1.0999, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.2676923076923077, | |
| "grad_norm": 0.1904296875, | |
| "learning_rate": 9.330769230769231e-06, | |
| "loss": 1.261, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.27076923076923076, | |
| "grad_norm": 0.142578125, | |
| "learning_rate": 9.323076923076924e-06, | |
| "loss": 0.9943, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.27384615384615385, | |
| "grad_norm": 0.1572265625, | |
| "learning_rate": 9.315384615384617e-06, | |
| "loss": 1.1976, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.27692307692307694, | |
| "grad_norm": 0.1513671875, | |
| "learning_rate": 9.307692307692308e-06, | |
| "loss": 1.1669, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.166015625, | |
| "learning_rate": 9.3e-06, | |
| "loss": 1.1616, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.28307692307692306, | |
| "grad_norm": 0.16015625, | |
| "learning_rate": 9.292307692307694e-06, | |
| "loss": 1.2202, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.28615384615384615, | |
| "grad_norm": 0.1533203125, | |
| "learning_rate": 9.284615384615385e-06, | |
| "loss": 1.1244, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.28923076923076924, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 9.276923076923077e-06, | |
| "loss": 1.0587, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2923076923076923, | |
| "grad_norm": 0.1494140625, | |
| "learning_rate": 9.26923076923077e-06, | |
| "loss": 1.139, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2953846153846154, | |
| "grad_norm": 0.1533203125, | |
| "learning_rate": 9.261538461538461e-06, | |
| "loss": 1.1293, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.29846153846153844, | |
| "grad_norm": 0.1455078125, | |
| "learning_rate": 9.253846153846154e-06, | |
| "loss": 1.1064, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.30153846153846153, | |
| "grad_norm": 0.1416015625, | |
| "learning_rate": 9.246153846153847e-06, | |
| "loss": 1.142, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.3046153846153846, | |
| "grad_norm": 0.150390625, | |
| "learning_rate": 9.23846153846154e-06, | |
| "loss": 1.1003, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.3076923076923077, | |
| "grad_norm": 0.142578125, | |
| "learning_rate": 9.230769230769232e-06, | |
| "loss": 1.1111, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.31076923076923074, | |
| "grad_norm": 0.1533203125, | |
| "learning_rate": 9.223076923076923e-06, | |
| "loss": 1.2121, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.31384615384615383, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 9.215384615384616e-06, | |
| "loss": 1.0288, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.3169230769230769, | |
| "grad_norm": 0.140625, | |
| "learning_rate": 9.207692307692309e-06, | |
| "loss": 1.1273, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 9.200000000000002e-06, | |
| "loss": 1.0176, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.3230769230769231, | |
| "grad_norm": 0.1376953125, | |
| "learning_rate": 9.192307692307693e-06, | |
| "loss": 1.1277, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3261538461538461, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 9.184615384615386e-06, | |
| "loss": 1.0563, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3292307692307692, | |
| "grad_norm": 0.150390625, | |
| "learning_rate": 9.176923076923078e-06, | |
| "loss": 1.1409, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.3323076923076923, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 9.169230769230771e-06, | |
| "loss": 1.0699, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.3353846153846154, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 9.161538461538462e-06, | |
| "loss": 1.111, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.3384615384615385, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 9.153846153846155e-06, | |
| "loss": 1.0962, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3415384615384615, | |
| "grad_norm": 0.1435546875, | |
| "learning_rate": 9.146153846153846e-06, | |
| "loss": 1.1233, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.3446153846153846, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 9.138461538461539e-06, | |
| "loss": 1.0922, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.3476923076923077, | |
| "grad_norm": 0.12255859375, | |
| "learning_rate": 9.130769230769232e-06, | |
| "loss": 1.0232, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.3507692307692308, | |
| "grad_norm": 0.125, | |
| "learning_rate": 9.123076923076923e-06, | |
| "loss": 1.0629, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.35384615384615387, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 9.115384615384615e-06, | |
| "loss": 1.0904, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.3569230769230769, | |
| "grad_norm": 0.12255859375, | |
| "learning_rate": 9.107692307692308e-06, | |
| "loss": 1.0123, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 9.100000000000001e-06, | |
| "loss": 1.0697, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.3630769230769231, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 9.092307692307692e-06, | |
| "loss": 1.042, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.36615384615384616, | |
| "grad_norm": 0.123046875, | |
| "learning_rate": 9.084615384615385e-06, | |
| "loss": 1.0818, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.36923076923076925, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 9.076923076923078e-06, | |
| "loss": 1.0598, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3723076923076923, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 9.06923076923077e-06, | |
| "loss": 1.0614, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.37538461538461537, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 9.061538461538463e-06, | |
| "loss": 1.004, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.37846153846153846, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 9.053846153846154e-06, | |
| "loss": 1.0717, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.38153846153846155, | |
| "grad_norm": 0.123046875, | |
| "learning_rate": 9.046153846153847e-06, | |
| "loss": 1.0261, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 9.03846153846154e-06, | |
| "loss": 1.0789, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.38769230769230767, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 9.030769230769233e-06, | |
| "loss": 1.083, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.39076923076923076, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 9.023076923076924e-06, | |
| "loss": 1.0129, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.39384615384615385, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 9.015384615384616e-06, | |
| "loss": 1.0588, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.39692307692307693, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 9.007692307692309e-06, | |
| "loss": 1.0988, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 9e-06, | |
| "loss": 1.0002, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.40307692307692305, | |
| "grad_norm": 0.12451171875, | |
| "learning_rate": 8.992307692307693e-06, | |
| "loss": 1.0977, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.40615384615384614, | |
| "grad_norm": 0.119140625, | |
| "learning_rate": 8.984615384615386e-06, | |
| "loss": 1.075, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.40923076923076923, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 8.976923076923077e-06, | |
| "loss": 1.0013, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.4123076923076923, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 8.96923076923077e-06, | |
| "loss": 1.0471, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.4153846153846154, | |
| "grad_norm": 0.119140625, | |
| "learning_rate": 8.961538461538462e-06, | |
| "loss": 1.0296, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.41846153846153844, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 8.953846153846153e-06, | |
| "loss": 0.9648, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.42153846153846153, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 8.946153846153846e-06, | |
| "loss": 1.004, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.4246153846153846, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 8.938461538461539e-06, | |
| "loss": 1.0301, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.4276923076923077, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 8.930769230769232e-06, | |
| "loss": 1.003, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.4307692307692308, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 8.923076923076925e-06, | |
| "loss": 0.9966, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.4338461538461538, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 8.915384615384616e-06, | |
| "loss": 1.0372, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.4369230769230769, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 8.907692307692308e-06, | |
| "loss": 1.0301, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.125, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 1.0993, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.4430769230769231, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 8.892307692307694e-06, | |
| "loss": 1.0152, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.4461538461538462, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 8.884615384615385e-06, | |
| "loss": 0.9818, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.4492307692307692, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 8.876923076923078e-06, | |
| "loss": 0.9707, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.4523076923076923, | |
| "grad_norm": 0.11328125, | |
| "learning_rate": 8.86923076923077e-06, | |
| "loss": 0.9819, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.4553846153846154, | |
| "grad_norm": 0.125, | |
| "learning_rate": 8.861538461538463e-06, | |
| "loss": 1.0015, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.4584615384615385, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 8.853846153846154e-06, | |
| "loss": 0.9939, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.46153846153846156, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 8.846153846153847e-06, | |
| "loss": 0.9875, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4646153846153846, | |
| "grad_norm": 0.12255859375, | |
| "learning_rate": 8.838461538461538e-06, | |
| "loss": 1.0281, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.4676923076923077, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 8.830769230769231e-06, | |
| "loss": 0.9997, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.4707692307692308, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 8.823076923076924e-06, | |
| "loss": 1.0186, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.47384615384615386, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 8.815384615384615e-06, | |
| "loss": 1.023, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.47692307692307695, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 8.807692307692308e-06, | |
| "loss": 0.9963, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 8.8e-06, | |
| "loss": 1.008, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.48307692307692307, | |
| "grad_norm": 0.1337890625, | |
| "learning_rate": 8.792307692307693e-06, | |
| "loss": 1.0809, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.48615384615384616, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 8.784615384615386e-06, | |
| "loss": 1.0142, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.48923076923076925, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 8.776923076923077e-06, | |
| "loss": 0.9912, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.49230769230769234, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 8.76923076923077e-06, | |
| "loss": 0.9856, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.49538461538461537, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 8.761538461538463e-06, | |
| "loss": 0.8901, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.49846153846153846, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 8.753846153846155e-06, | |
| "loss": 1.0399, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.5015384615384615, | |
| "grad_norm": 0.12109375, | |
| "learning_rate": 8.746153846153846e-06, | |
| "loss": 0.9985, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.5046153846153846, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 8.73846153846154e-06, | |
| "loss": 1.0299, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.5076923076923077, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 8.730769230769232e-06, | |
| "loss": 0.9807, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.5107692307692308, | |
| "grad_norm": 0.142578125, | |
| "learning_rate": 8.723076923076925e-06, | |
| "loss": 1.1017, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.5138461538461538, | |
| "grad_norm": 0.1337890625, | |
| "learning_rate": 8.715384615384616e-06, | |
| "loss": 1.047, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.5169230769230769, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 8.707692307692309e-06, | |
| "loss": 1.003, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 8.700000000000001e-06, | |
| "loss": 0.9674, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.5230769230769231, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 8.692307692307692e-06, | |
| "loss": 0.995, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5261538461538462, | |
| "grad_norm": 0.1484375, | |
| "learning_rate": 8.684615384615385e-06, | |
| "loss": 0.9515, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.5292307692307693, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 8.676923076923078e-06, | |
| "loss": 0.9964, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.5323076923076923, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 8.669230769230769e-06, | |
| "loss": 0.9572, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.5353846153846153, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 8.661538461538462e-06, | |
| "loss": 0.9454, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.5384615384615384, | |
| "grad_norm": 0.1982421875, | |
| "learning_rate": 8.653846153846155e-06, | |
| "loss": 0.9738, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.5415384615384615, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 8.646153846153846e-06, | |
| "loss": 0.9537, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.5446153846153846, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 8.638461538461539e-06, | |
| "loss": 1.0275, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.5476923076923077, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 8.630769230769231e-06, | |
| "loss": 0.9752, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.5507692307692308, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 8.623076923076924e-06, | |
| "loss": 0.9401, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.5538461538461539, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 8.615384615384617e-06, | |
| "loss": 0.9614, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.556923076923077, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 8.607692307692308e-06, | |
| "loss": 0.958, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 8.6e-06, | |
| "loss": 0.9176, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.563076923076923, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 8.592307692307693e-06, | |
| "loss": 0.9948, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.5661538461538461, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 8.584615384615386e-06, | |
| "loss": 0.9877, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5692307692307692, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 8.576923076923077e-06, | |
| "loss": 0.9587, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5723076923076923, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 8.56923076923077e-06, | |
| "loss": 0.9679, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.5753846153846154, | |
| "grad_norm": 0.1376953125, | |
| "learning_rate": 8.561538461538463e-06, | |
| "loss": 1.0528, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.5784615384615385, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 8.553846153846156e-06, | |
| "loss": 0.946, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5815384615384616, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 8.546153846153847e-06, | |
| "loss": 0.9383, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.5846153846153846, | |
| "grad_norm": 0.134765625, | |
| "learning_rate": 8.53846153846154e-06, | |
| "loss": 1.0041, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5876923076923077, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 8.53076923076923e-06, | |
| "loss": 1.0136, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.5907692307692308, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 8.523076923076923e-06, | |
| "loss": 0.9866, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.5938461538461538, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 8.515384615384616e-06, | |
| "loss": 0.9633, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.5969230769230769, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 8.507692307692307e-06, | |
| "loss": 0.9552, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.1201171875, | |
| "learning_rate": 8.5e-06, | |
| "loss": 1.0123, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.6030769230769231, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 8.492307692307693e-06, | |
| "loss": 0.9365, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.6061538461538462, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 8.484615384615385e-06, | |
| "loss": 0.95, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.6092307692307692, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 8.476923076923078e-06, | |
| "loss": 0.8824, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.6123076923076923, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 8.46923076923077e-06, | |
| "loss": 0.9532, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.6153846153846154, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 8.461538461538462e-06, | |
| "loss": 0.933, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.6184615384615385, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 8.453846153846155e-06, | |
| "loss": 0.9753, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.6215384615384615, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 8.446153846153848e-06, | |
| "loss": 0.9129, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.6246153846153846, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 8.438461538461539e-06, | |
| "loss": 0.9672, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.6276923076923077, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 8.430769230769231e-06, | |
| "loss": 0.9243, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.6307692307692307, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 8.423076923076924e-06, | |
| "loss": 0.9044, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.6338461538461538, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 8.415384615384617e-06, | |
| "loss": 0.9285, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.6369230769230769, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 8.407692307692308e-06, | |
| "loss": 0.9435, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.0324, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.6430769230769231, | |
| "grad_norm": 0.125, | |
| "learning_rate": 8.392307692307694e-06, | |
| "loss": 0.9932, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.6461538461538462, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 8.384615384615385e-06, | |
| "loss": 1.0314, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.6492307692307693, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 8.376923076923078e-06, | |
| "loss": 0.9245, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.6523076923076923, | |
| "grad_norm": 0.11767578125, | |
| "learning_rate": 8.36923076923077e-06, | |
| "loss": 0.9759, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.6553846153846153, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 8.361538461538461e-06, | |
| "loss": 0.9581, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.6584615384615384, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 8.353846153846154e-06, | |
| "loss": 1.0027, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.6615384615384615, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 8.346153846153847e-06, | |
| "loss": 0.9302, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.6646153846153846, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 8.338461538461538e-06, | |
| "loss": 0.9282, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.6676923076923077, | |
| "grad_norm": 0.162109375, | |
| "learning_rate": 8.33076923076923e-06, | |
| "loss": 1.0222, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.6707692307692308, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 8.323076923076924e-06, | |
| "loss": 0.9857, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.6738461538461539, | |
| "grad_norm": 0.1513671875, | |
| "learning_rate": 8.315384615384616e-06, | |
| "loss": 1.0328, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.676923076923077, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 8.307692307692309e-06, | |
| "loss": 0.9697, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 8.3e-06, | |
| "loss": 0.9329, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.683076923076923, | |
| "grad_norm": 0.119140625, | |
| "learning_rate": 8.292307692307693e-06, | |
| "loss": 0.9555, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.6861538461538461, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 8.284615384615386e-06, | |
| "loss": 0.9182, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.6892307692307692, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 8.276923076923078e-06, | |
| "loss": 0.9429, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.6923076923076923, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 8.26923076923077e-06, | |
| "loss": 0.8966, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6953846153846154, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 8.261538461538462e-06, | |
| "loss": 0.9731, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.6984615384615385, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 8.253846153846155e-06, | |
| "loss": 0.8996, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.7015384615384616, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 8.246153846153848e-06, | |
| "loss": 0.9365, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.7046153846153846, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 8.238461538461539e-06, | |
| "loss": 0.9287, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.7076923076923077, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 8.230769230769232e-06, | |
| "loss": 0.9866, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.7107692307692308, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 8.223076923076923e-06, | |
| "loss": 0.9364, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.7138461538461538, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 8.215384615384616e-06, | |
| "loss": 0.9377, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.7169230769230769, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 8.207692307692308e-06, | |
| "loss": 0.9102, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.1494140625, | |
| "learning_rate": 8.2e-06, | |
| "loss": 1.0207, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.7230769230769231, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 8.192307692307692e-06, | |
| "loss": 0.9463, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.7261538461538461, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 8.184615384615385e-06, | |
| "loss": 0.977, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.7292307692307692, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 8.176923076923078e-06, | |
| "loss": 0.9035, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.7323076923076923, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 8.16923076923077e-06, | |
| "loss": 0.9152, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.7353846153846154, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 8.161538461538462e-06, | |
| "loss": 0.964, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.7384615384615385, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 8.153846153846154e-06, | |
| "loss": 0.9655, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7415384615384616, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 8.146153846153847e-06, | |
| "loss": 0.9406, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.7446153846153846, | |
| "grad_norm": 0.1201171875, | |
| "learning_rate": 8.13846153846154e-06, | |
| "loss": 0.955, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.7476923076923077, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 8.130769230769231e-06, | |
| "loss": 0.8998, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.7507692307692307, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 8.123076923076924e-06, | |
| "loss": 0.9753, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.7538461538461538, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 8.115384615384617e-06, | |
| "loss": 0.9659, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.7569230769230769, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 8.10769230769231e-06, | |
| "loss": 0.9104, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 8.1e-06, | |
| "loss": 0.8778, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.7630769230769231, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 8.092307692307693e-06, | |
| "loss": 0.8775, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.7661538461538462, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 8.084615384615386e-06, | |
| "loss": 0.9472, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 8.076923076923077e-06, | |
| "loss": 0.9444, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.7723076923076924, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 8.06923076923077e-06, | |
| "loss": 0.9107, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.7753846153846153, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 8.061538461538463e-06, | |
| "loss": 0.9417, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.7784615384615384, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 8.053846153846154e-06, | |
| "loss": 0.9135, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.7815384615384615, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 8.046153846153846e-06, | |
| "loss": 0.9175, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.7846153846153846, | |
| "grad_norm": 0.08056640625, | |
| "learning_rate": 8.03846153846154e-06, | |
| "loss": 0.8171, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.7876923076923077, | |
| "grad_norm": 0.123046875, | |
| "learning_rate": 8.03076923076923e-06, | |
| "loss": 0.9672, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.7907692307692308, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 8.023076923076923e-06, | |
| "loss": 0.8799, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.7938461538461539, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 8.015384615384616e-06, | |
| "loss": 0.913, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.796923076923077, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 8.007692307692309e-06, | |
| "loss": 0.8654, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.9201, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.803076923076923, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 7.992307692307692e-06, | |
| "loss": 0.9131, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.8061538461538461, | |
| "grad_norm": 0.08447265625, | |
| "learning_rate": 7.984615384615385e-06, | |
| "loss": 0.9123, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.8092307692307692, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 7.976923076923078e-06, | |
| "loss": 0.9436, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.8123076923076923, | |
| "grad_norm": 0.08056640625, | |
| "learning_rate": 7.96923076923077e-06, | |
| "loss": 0.8846, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.8153846153846154, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 7.961538461538462e-06, | |
| "loss": 0.9456, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.8184615384615385, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 7.953846153846155e-06, | |
| "loss": 0.8611, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.8215384615384616, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 7.946153846153847e-06, | |
| "loss": 0.9421, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.8246153846153846, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 7.93846153846154e-06, | |
| "loss": 0.9192, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.8276923076923077, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 7.930769230769231e-06, | |
| "loss": 0.8858, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.8307692307692308, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 7.923076923076924e-06, | |
| "loss": 0.9422, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.8338461538461538, | |
| "grad_norm": 0.177734375, | |
| "learning_rate": 7.915384615384615e-06, | |
| "loss": 0.9481, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.8369230769230769, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 7.907692307692308e-06, | |
| "loss": 0.869, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.9813, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.8430769230769231, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 7.892307692307692e-06, | |
| "loss": 0.8997, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.8461538461538461, | |
| "grad_norm": 0.125, | |
| "learning_rate": 7.884615384615384e-06, | |
| "loss": 0.9491, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.8492307692307692, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 7.876923076923077e-06, | |
| "loss": 0.9195, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.8523076923076923, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 7.86923076923077e-06, | |
| "loss": 0.9129, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.8553846153846154, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 7.861538461538463e-06, | |
| "loss": 0.9908, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.8584615384615385, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 7.853846153846154e-06, | |
| "loss": 0.8949, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.8615384615384616, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 7.846153846153847e-06, | |
| "loss": 0.9472, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.8646153846153846, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 7.83846153846154e-06, | |
| "loss": 0.8985, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.8676923076923077, | |
| "grad_norm": 0.08642578125, | |
| "learning_rate": 7.830769230769232e-06, | |
| "loss": 0.8989, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.8707692307692307, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 7.823076923076923e-06, | |
| "loss": 0.875, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.8738461538461538, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 7.815384615384616e-06, | |
| "loss": 0.9091, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.8769230769230769, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 7.807692307692309e-06, | |
| "loss": 0.936, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 7.800000000000002e-06, | |
| "loss": 0.9323, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.8830769230769231, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 7.792307692307693e-06, | |
| "loss": 0.8884, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.8861538461538462, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 7.784615384615385e-06, | |
| "loss": 0.9148, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.8892307692307693, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 7.776923076923078e-06, | |
| "loss": 0.9018, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.8923076923076924, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 7.76923076923077e-06, | |
| "loss": 0.9304, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.8953846153846153, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 7.761538461538462e-06, | |
| "loss": 0.8613, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.8984615384615384, | |
| "grad_norm": 0.08935546875, | |
| "learning_rate": 7.753846153846155e-06, | |
| "loss": 0.8766, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.9015384615384615, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 7.746153846153846e-06, | |
| "loss": 0.9412, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.9046153846153846, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 7.738461538461539e-06, | |
| "loss": 0.9288, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.9076923076923077, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 7.730769230769231e-06, | |
| "loss": 0.9423, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.9107692307692308, | |
| "grad_norm": 0.0888671875, | |
| "learning_rate": 7.723076923076924e-06, | |
| "loss": 0.872, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.9138461538461539, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 7.715384615384615e-06, | |
| "loss": 0.8896, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.916923076923077, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 7.707692307692308e-06, | |
| "loss": 0.877, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 7.7e-06, | |
| "loss": 0.8395, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.9230769230769231, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 7.692307692307694e-06, | |
| "loss": 0.8947, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.9261538461538461, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 7.684615384615385e-06, | |
| "loss": 0.9335, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.9292307692307692, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 7.676923076923077e-06, | |
| "loss": 0.9047, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.9323076923076923, | |
| "grad_norm": 0.0859375, | |
| "learning_rate": 7.66923076923077e-06, | |
| "loss": 0.9168, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.9353846153846154, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 7.661538461538463e-06, | |
| "loss": 0.8469, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.9384615384615385, | |
| "grad_norm": 0.0859375, | |
| "learning_rate": 7.653846153846154e-06, | |
| "loss": 0.8624, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.9415384615384615, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 7.646153846153847e-06, | |
| "loss": 0.9436, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.9446153846153846, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 7.63846153846154e-06, | |
| "loss": 0.9133, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.9476923076923077, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 7.630769230769232e-06, | |
| "loss": 0.836, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.9507692307692308, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 7.623076923076924e-06, | |
| "loss": 0.9272, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.9538461538461539, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 7.615384615384615e-06, | |
| "loss": 0.8932, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.9569230769230769, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 7.607692307692308e-06, | |
| "loss": 0.8389, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.0888671875, | |
| "learning_rate": 7.600000000000001e-06, | |
| "loss": 0.831, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.963076923076923, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 7.592307692307693e-06, | |
| "loss": 0.8732, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.9661538461538461, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 7.584615384615385e-06, | |
| "loss": 0.9093, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.9692307692307692, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 7.5769230769230775e-06, | |
| "loss": 0.962, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.9723076923076923, | |
| "grad_norm": 0.0869140625, | |
| "learning_rate": 7.5692307692307695e-06, | |
| "loss": 0.8741, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.9753846153846154, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 7.561538461538462e-06, | |
| "loss": 0.8955, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.9784615384615385, | |
| "grad_norm": 0.08837890625, | |
| "learning_rate": 7.553846153846155e-06, | |
| "loss": 0.8394, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.9815384615384616, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 7.546153846153846e-06, | |
| "loss": 0.8809, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.9846153846153847, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 7.538461538461539e-06, | |
| "loss": 0.9349, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.9876923076923076, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 7.530769230769232e-06, | |
| "loss": 0.9177, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.9907692307692307, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 7.523076923076924e-06, | |
| "loss": 0.9004, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.9938461538461538, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 7.5153846153846155e-06, | |
| "loss": 0.8698, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.9969230769230769, | |
| "grad_norm": 0.322265625, | |
| "learning_rate": 7.507692307692308e-06, | |
| "loss": 0.889, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.9119, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.003076923076923, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 7.492307692307693e-06, | |
| "loss": 0.8353, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.0061538461538462, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 7.484615384615386e-06, | |
| "loss": 0.9124, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.0092307692307692, | |
| "grad_norm": 0.08349609375, | |
| "learning_rate": 7.476923076923077e-06, | |
| "loss": 0.8289, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.0123076923076924, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 7.46923076923077e-06, | |
| "loss": 0.8407, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.0153846153846153, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 7.461538461538462e-06, | |
| "loss": 0.8651, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.0184615384615385, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 7.453846153846155e-06, | |
| "loss": 0.8838, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.0215384615384615, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 7.446153846153846e-06, | |
| "loss": 0.9404, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.0246153846153847, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 7.438461538461539e-06, | |
| "loss": 0.8734, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.0276923076923077, | |
| "grad_norm": 0.083984375, | |
| "learning_rate": 7.430769230769232e-06, | |
| "loss": 0.8729, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.0307692307692307, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 7.423076923076924e-06, | |
| "loss": 0.9136, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.0338461538461539, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 7.4153846153846164e-06, | |
| "loss": 0.9294, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.0369230769230768, | |
| "grad_norm": 0.12109375, | |
| "learning_rate": 7.407692307692308e-06, | |
| "loss": 0.9243, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.9192, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.043076923076923, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 7.392307692307693e-06, | |
| "loss": 0.8743, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.0461538461538462, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 7.384615384615386e-06, | |
| "loss": 0.9251, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.0492307692307692, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 7.376923076923077e-06, | |
| "loss": 0.9009, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.0523076923076924, | |
| "grad_norm": 0.1328125, | |
| "learning_rate": 7.36923076923077e-06, | |
| "loss": 0.9855, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.0553846153846154, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 7.3615384615384625e-06, | |
| "loss": 0.8874, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.0584615384615386, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 7.353846153846154e-06, | |
| "loss": 0.8889, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.0615384615384615, | |
| "grad_norm": 0.08203125, | |
| "learning_rate": 7.346153846153847e-06, | |
| "loss": 0.8963, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.0646153846153845, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 7.338461538461539e-06, | |
| "loss": 0.8842, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.0676923076923077, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 7.330769230769231e-06, | |
| "loss": 0.8583, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.0707692307692307, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 7.323076923076924e-06, | |
| "loss": 0.9322, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.073846153846154, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 7.3153846153846166e-06, | |
| "loss": 0.8769, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.0769230769230769, | |
| "grad_norm": 0.08935546875, | |
| "learning_rate": 7.307692307692308e-06, | |
| "loss": 0.8697, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 0.08935546875, | |
| "learning_rate": 7.3e-06, | |
| "loss": 0.8632, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.083076923076923, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 7.292307692307693e-06, | |
| "loss": 0.813, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.0861538461538462, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 7.284615384615386e-06, | |
| "loss": 0.8745, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.0892307692307692, | |
| "grad_norm": 0.0869140625, | |
| "learning_rate": 7.276923076923077e-06, | |
| "loss": 0.8735, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.0923076923076924, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 7.26923076923077e-06, | |
| "loss": 0.8606, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.0953846153846154, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 7.261538461538462e-06, | |
| "loss": 0.89, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.0984615384615384, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 7.2538461538461545e-06, | |
| "loss": 0.8824, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.1015384615384616, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 7.246153846153847e-06, | |
| "loss": 0.8998, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.1046153846153846, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 7.238461538461538e-06, | |
| "loss": 0.8922, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.1076923076923078, | |
| "grad_norm": 0.0888671875, | |
| "learning_rate": 7.230769230769231e-06, | |
| "loss": 0.8264, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.1107692307692307, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 7.223076923076924e-06, | |
| "loss": 0.8757, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.113846153846154, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 7.215384615384617e-06, | |
| "loss": 0.9294, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.116923076923077, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 7.207692307692308e-06, | |
| "loss": 0.877, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 7.2000000000000005e-06, | |
| "loss": 0.8758, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.123076923076923, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 7.192307692307693e-06, | |
| "loss": 0.8682, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.126153846153846, | |
| "grad_norm": 0.08837890625, | |
| "learning_rate": 7.184615384615385e-06, | |
| "loss": 0.8567, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.1292307692307693, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 7.176923076923078e-06, | |
| "loss": 0.9063, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.1323076923076922, | |
| "grad_norm": 0.08447265625, | |
| "learning_rate": 7.169230769230769e-06, | |
| "loss": 0.8677, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.1353846153846154, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 7.161538461538462e-06, | |
| "loss": 0.9018, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.1384615384615384, | |
| "grad_norm": 0.0888671875, | |
| "learning_rate": 7.153846153846155e-06, | |
| "loss": 0.8491, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.1415384615384616, | |
| "grad_norm": 0.1328125, | |
| "learning_rate": 7.146153846153847e-06, | |
| "loss": 0.9698, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.1446153846153846, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 7.1384615384615385e-06, | |
| "loss": 0.8739, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.1476923076923078, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 7.130769230769231e-06, | |
| "loss": 0.8469, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.1507692307692308, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 7.123076923076924e-06, | |
| "loss": 0.8319, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.1538461538461537, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 7.115384615384616e-06, | |
| "loss": 0.9033, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.156923076923077, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 7.107692307692309e-06, | |
| "loss": 0.846, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 7.100000000000001e-06, | |
| "loss": 0.8865, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.1630769230769231, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 7.0923076923076926e-06, | |
| "loss": 0.8769, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.166153846153846, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 7.084615384615385e-06, | |
| "loss": 0.8891, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.1692307692307693, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 7.076923076923078e-06, | |
| "loss": 0.9003, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.1723076923076923, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 7.069230769230769e-06, | |
| "loss": 0.8442, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.1753846153846155, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 7.061538461538462e-06, | |
| "loss": 0.8752, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.1784615384615384, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 7.053846153846155e-06, | |
| "loss": 0.8524, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.1815384615384614, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 7.046153846153847e-06, | |
| "loss": 0.8719, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.1846153846153846, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 7.038461538461539e-06, | |
| "loss": 0.8479, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.1876923076923076, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 7.030769230769231e-06, | |
| "loss": 0.8653, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.1907692307692308, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 7.023076923076923e-06, | |
| "loss": 0.8723, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.1938461538461538, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 7.015384615384616e-06, | |
| "loss": 0.8887, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.196923076923077, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 7.007692307692309e-06, | |
| "loss": 0.9512, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 7e-06, | |
| "loss": 0.882, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.2030769230769232, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 6.992307692307693e-06, | |
| "loss": 0.8859, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.2061538461538461, | |
| "grad_norm": 0.142578125, | |
| "learning_rate": 6.9846153846153854e-06, | |
| "loss": 0.9472, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.209230769230769, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 6.976923076923078e-06, | |
| "loss": 0.8808, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.2123076923076923, | |
| "grad_norm": 0.0859375, | |
| "learning_rate": 6.96923076923077e-06, | |
| "loss": 0.8486, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.2153846153846155, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 6.961538461538462e-06, | |
| "loss": 0.9193, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.2184615384615385, | |
| "grad_norm": 0.08447265625, | |
| "learning_rate": 6.953846153846154e-06, | |
| "loss": 0.8518, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.2215384615384615, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 6.946153846153847e-06, | |
| "loss": 0.9028, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.2246153846153847, | |
| "grad_norm": 0.08984375, | |
| "learning_rate": 6.9384615384615395e-06, | |
| "loss": 0.8646, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.2276923076923076, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 6.930769230769231e-06, | |
| "loss": 0.8605, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.2307692307692308, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 6.923076923076923e-06, | |
| "loss": 0.8978, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.2338461538461538, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 6.915384615384616e-06, | |
| "loss": 0.8515, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.236923076923077, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 6.907692307692309e-06, | |
| "loss": 0.8307, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 0.1650390625, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.8954, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.2430769230769232, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 6.892307692307693e-06, | |
| "loss": 0.8327, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.2461538461538462, | |
| "grad_norm": 0.08935546875, | |
| "learning_rate": 6.8846153846153855e-06, | |
| "loss": 0.8104, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.2492307692307691, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 6.8769230769230775e-06, | |
| "loss": 0.9422, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.2523076923076923, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 6.86923076923077e-06, | |
| "loss": 0.8732, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.2553846153846153, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 6.861538461538461e-06, | |
| "loss": 0.9171, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.2584615384615385, | |
| "grad_norm": 0.08642578125, | |
| "learning_rate": 6.853846153846154e-06, | |
| "loss": 0.8369, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.2615384615384615, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 6.846153846153847e-06, | |
| "loss": 0.847, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.2646153846153847, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 6.83846153846154e-06, | |
| "loss": 0.867, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.2676923076923077, | |
| "grad_norm": 0.11328125, | |
| "learning_rate": 6.830769230769231e-06, | |
| "loss": 0.9069, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.2707692307692309, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 6.8230769230769235e-06, | |
| "loss": 0.9174, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.2738461538461539, | |
| "grad_norm": 0.08349609375, | |
| "learning_rate": 6.815384615384616e-06, | |
| "loss": 0.8343, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.2769230769230768, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 6.807692307692308e-06, | |
| "loss": 0.8455, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 0.8232, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.283076923076923, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 6.792307692307693e-06, | |
| "loss": 0.8524, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.2861538461538462, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 6.784615384615385e-06, | |
| "loss": 0.8099, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.2892307692307692, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 6.776923076923078e-06, | |
| "loss": 0.8905, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.2923076923076924, | |
| "grad_norm": 0.0859375, | |
| "learning_rate": 6.76923076923077e-06, | |
| "loss": 0.8504, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.2953846153846154, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 6.7615384615384614e-06, | |
| "loss": 0.8929, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.2984615384615386, | |
| "grad_norm": 0.08935546875, | |
| "learning_rate": 6.753846153846154e-06, | |
| "loss": 0.8709, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.3015384615384615, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 6.746153846153847e-06, | |
| "loss": 0.8906, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.3046153846153845, | |
| "grad_norm": 0.08056640625, | |
| "learning_rate": 6.738461538461539e-06, | |
| "loss": 0.8365, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.3076923076923077, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 6.730769230769232e-06, | |
| "loss": 0.8303, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.3107692307692307, | |
| "grad_norm": 0.0869140625, | |
| "learning_rate": 6.723076923076924e-06, | |
| "loss": 0.8528, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.3138461538461539, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 6.7153846153846155e-06, | |
| "loss": 0.8426, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.3169230769230769, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 6.707692307692308e-06, | |
| "loss": 0.8677, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 6.700000000000001e-06, | |
| "loss": 0.9237, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.323076923076923, | |
| "grad_norm": 0.1640625, | |
| "learning_rate": 6.692307692307692e-06, | |
| "loss": 0.9263, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.3261538461538462, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 6.684615384615385e-06, | |
| "loss": 0.8738, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.3292307692307692, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 6.676923076923078e-06, | |
| "loss": 0.8671, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.3323076923076922, | |
| "grad_norm": 0.08447265625, | |
| "learning_rate": 6.6692307692307705e-06, | |
| "loss": 0.8577, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.3353846153846154, | |
| "grad_norm": 0.083984375, | |
| "learning_rate": 6.661538461538462e-06, | |
| "loss": 0.827, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.3384615384615386, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 6.653846153846154e-06, | |
| "loss": 0.8578, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.3415384615384616, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 6.646153846153846e-06, | |
| "loss": 0.8166, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.3446153846153845, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 6.638461538461539e-06, | |
| "loss": 0.8947, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.3476923076923077, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 6.630769230769232e-06, | |
| "loss": 0.932, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.3507692307692307, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 6.623076923076923e-06, | |
| "loss": 0.8804, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.353846153846154, | |
| "grad_norm": 0.1376953125, | |
| "learning_rate": 6.615384615384616e-06, | |
| "loss": 0.9784, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.356923076923077, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 6.607692307692308e-06, | |
| "loss": 0.896, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.3599999999999999, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 6.600000000000001e-06, | |
| "loss": 0.8738, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.363076923076923, | |
| "grad_norm": 0.140625, | |
| "learning_rate": 6.592307692307692e-06, | |
| "loss": 0.8838, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.3661538461538463, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 6.584615384615385e-06, | |
| "loss": 0.8805, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.3692307692307693, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 6.576923076923078e-06, | |
| "loss": 0.8042, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.3723076923076922, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 6.56923076923077e-06, | |
| "loss": 0.8788, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.3753846153846154, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 6.5615384615384625e-06, | |
| "loss": 0.8689, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.3784615384615384, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 6.553846153846154e-06, | |
| "loss": 0.884, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.3815384615384616, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 6.546153846153846e-06, | |
| "loss": 0.8406, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.3846153846153846, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 6.538461538461539e-06, | |
| "loss": 0.8729, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.3876923076923076, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 6.530769230769232e-06, | |
| "loss": 0.9129, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.3907692307692308, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 6.523076923076923e-06, | |
| "loss": 0.8604, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.393846153846154, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 6.515384615384616e-06, | |
| "loss": 0.8803, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.396923076923077, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 6.5076923076923085e-06, | |
| "loss": 0.8238, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.8554, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.403076923076923, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 6.492307692307693e-06, | |
| "loss": 0.8371, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.406153846153846, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 6.484615384615385e-06, | |
| "loss": 0.8189, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.4092307692307693, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 6.476923076923077e-06, | |
| "loss": 0.8848, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.4123076923076923, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 6.46923076923077e-06, | |
| "loss": 0.8912, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.4153846153846155, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 6.461538461538463e-06, | |
| "loss": 0.8049, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.4184615384615384, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 6.453846153846154e-06, | |
| "loss": 0.8725, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.4215384615384616, | |
| "grad_norm": 0.123046875, | |
| "learning_rate": 6.4461538461538465e-06, | |
| "loss": 0.8227, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.4246153846153846, | |
| "grad_norm": 0.08349609375, | |
| "learning_rate": 6.438461538461539e-06, | |
| "loss": 0.8258, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.4276923076923076, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 6.430769230769231e-06, | |
| "loss": 0.8589, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.4307692307692308, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 6.423076923076924e-06, | |
| "loss": 0.8708, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.4338461538461538, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 6.415384615384616e-06, | |
| "loss": 0.8335, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.436923076923077, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 6.407692307692308e-06, | |
| "loss": 0.8539, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.9145, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.4430769230769231, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 6.392307692307693e-06, | |
| "loss": 0.8831, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.4461538461538461, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 6.384615384615384e-06, | |
| "loss": 0.8228, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.4492307692307693, | |
| "grad_norm": 0.0888671875, | |
| "learning_rate": 6.376923076923077e-06, | |
| "loss": 0.8476, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.4523076923076923, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 6.36923076923077e-06, | |
| "loss": 0.8544, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.4553846153846153, | |
| "grad_norm": 0.12109375, | |
| "learning_rate": 6.361538461538463e-06, | |
| "loss": 0.8947, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.4584615384615385, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 6.353846153846155e-06, | |
| "loss": 0.8674, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.4615384615384617, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 6.3461538461538466e-06, | |
| "loss": 0.8743, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.4646153846153847, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 6.3384615384615385e-06, | |
| "loss": 0.8286, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.4676923076923076, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 6.330769230769231e-06, | |
| "loss": 0.8129, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.4707692307692308, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 6.323076923076924e-06, | |
| "loss": 0.831, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.4738461538461538, | |
| "grad_norm": 0.14453125, | |
| "learning_rate": 6.315384615384615e-06, | |
| "loss": 0.8477, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.476923076923077, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 6.307692307692308e-06, | |
| "loss": 0.8861, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 6.300000000000001e-06, | |
| "loss": 0.9379, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.483076923076923, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 6.2923076923076934e-06, | |
| "loss": 0.8746, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.4861538461538462, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 6.284615384615385e-06, | |
| "loss": 0.8321, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.4892307692307694, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 6.276923076923077e-06, | |
| "loss": 0.8557, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.4923076923076923, | |
| "grad_norm": 0.08837890625, | |
| "learning_rate": 6.26923076923077e-06, | |
| "loss": 0.8314, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.4953846153846153, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 6.261538461538462e-06, | |
| "loss": 0.828, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.4984615384615385, | |
| "grad_norm": 0.1474609375, | |
| "learning_rate": 6.253846153846155e-06, | |
| "loss": 0.8561, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.5015384615384615, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 6.246153846153846e-06, | |
| "loss": 0.8693, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.5046153846153847, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 6.238461538461539e-06, | |
| "loss": 0.8596, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.5076923076923077, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 6.230769230769231e-06, | |
| "loss": 0.8595, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.5107692307692306, | |
| "grad_norm": 0.0869140625, | |
| "learning_rate": 6.223076923076924e-06, | |
| "loss": 0.8311, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.5138461538461538, | |
| "grad_norm": 0.08251953125, | |
| "learning_rate": 6.215384615384615e-06, | |
| "loss": 0.7852, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.516923076923077, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 6.207692307692308e-06, | |
| "loss": 0.8492, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 6.200000000000001e-06, | |
| "loss": 0.8243, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.523076923076923, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 6.192307692307693e-06, | |
| "loss": 0.8339, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.5261538461538462, | |
| "grad_norm": 0.083984375, | |
| "learning_rate": 6.1846153846153855e-06, | |
| "loss": 0.7964, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.5292307692307694, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 6.176923076923077e-06, | |
| "loss": 0.8816, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.5323076923076924, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 6.169230769230769e-06, | |
| "loss": 0.8943, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.5353846153846153, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 6.161538461538462e-06, | |
| "loss": 0.8551, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 6.153846153846155e-06, | |
| "loss": 0.8639, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.5415384615384615, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 6.146153846153846e-06, | |
| "loss": 0.8517, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.5446153846153847, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 6.138461538461539e-06, | |
| "loss": 0.8829, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.5476923076923077, | |
| "grad_norm": 0.125, | |
| "learning_rate": 6.1307692307692315e-06, | |
| "loss": 0.8361, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.5507692307692307, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 6.123076923076923e-06, | |
| "loss": 0.86, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.5538461538461539, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 6.115384615384616e-06, | |
| "loss": 0.8502, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.556923076923077, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 6.107692307692308e-06, | |
| "loss": 0.8355, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "grad_norm": 0.08251953125, | |
| "learning_rate": 6.1e-06, | |
| "loss": 0.7808, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.563076923076923, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 6.092307692307693e-06, | |
| "loss": 0.8956, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.566153846153846, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 6.084615384615386e-06, | |
| "loss": 0.8874, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.5692307692307692, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 6.076923076923077e-06, | |
| "loss": 0.8608, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.5723076923076924, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 6.0692307692307694e-06, | |
| "loss": 0.8447, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.5753846153846154, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 6.061538461538462e-06, | |
| "loss": 0.8339, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.5784615384615384, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 6.053846153846155e-06, | |
| "loss": 0.889, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.5815384615384616, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 6.046153846153847e-06, | |
| "loss": 0.8211, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.5846153846153848, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 6.038461538461539e-06, | |
| "loss": 0.8564, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.5876923076923077, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 6.030769230769231e-06, | |
| "loss": 0.9038, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.5907692307692307, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 6.0230769230769235e-06, | |
| "loss": 0.9108, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.5938461538461537, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 6.015384615384616e-06, | |
| "loss": 0.8443, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.596923076923077, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 6.007692307692307e-06, | |
| "loss": 0.868, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 0.083984375, | |
| "learning_rate": 6e-06, | |
| "loss": 0.8111, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.603076923076923, | |
| "grad_norm": 0.07958984375, | |
| "learning_rate": 5.992307692307693e-06, | |
| "loss": 0.8221, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.606153846153846, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 5.984615384615386e-06, | |
| "loss": 0.8409, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.6092307692307692, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 5.976923076923078e-06, | |
| "loss": 0.8427, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.6123076923076924, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 5.9692307692307695e-06, | |
| "loss": 0.7984, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.6153846153846154, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 5.961538461538462e-06, | |
| "loss": 0.9146, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.6184615384615384, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 5.953846153846154e-06, | |
| "loss": 0.8561, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.6215384615384614, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 5.946153846153847e-06, | |
| "loss": 0.8429, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.6246153846153846, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 5.938461538461538e-06, | |
| "loss": 0.8613, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.6276923076923078, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 5.930769230769231e-06, | |
| "loss": 0.8389, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.6307692307692307, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 5.923076923076924e-06, | |
| "loss": 0.8731, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.6338461538461537, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 5.915384615384616e-06, | |
| "loss": 0.885, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.636923076923077, | |
| "grad_norm": 0.125, | |
| "learning_rate": 5.907692307692308e-06, | |
| "loss": 0.886, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.6400000000000001, | |
| "grad_norm": 0.08447265625, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.7995, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.643076923076923, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 5.892307692307693e-06, | |
| "loss": 0.8518, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.646153846153846, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 5.884615384615385e-06, | |
| "loss": 0.8754, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.6492307692307693, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 5.876923076923078e-06, | |
| "loss": 0.8217, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.6523076923076923, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 5.86923076923077e-06, | |
| "loss": 0.9, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.6553846153846155, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 5.861538461538462e-06, | |
| "loss": 0.8791, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.6584615384615384, | |
| "grad_norm": 0.0888671875, | |
| "learning_rate": 5.853846153846154e-06, | |
| "loss": 0.8708, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.6615384615384614, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 5.846153846153847e-06, | |
| "loss": 0.8456, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.6646153846153846, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 5.838461538461538e-06, | |
| "loss": 0.8326, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.6676923076923078, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 5.830769230769231e-06, | |
| "loss": 0.8674, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.6707692307692308, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 5.823076923076924e-06, | |
| "loss": 0.8614, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.6738461538461538, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 5.815384615384616e-06, | |
| "loss": 0.8943, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.676923076923077, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 5.8076923076923084e-06, | |
| "loss": 0.7728, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.6800000000000002, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 5.8e-06, | |
| "loss": 0.8528, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.6830769230769231, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 5.792307692307692e-06, | |
| "loss": 0.837, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.6861538461538461, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 5.784615384615385e-06, | |
| "loss": 0.8721, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.689230769230769, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 5.776923076923078e-06, | |
| "loss": 0.8491, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.6923076923076923, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 5.769230769230769e-06, | |
| "loss": 0.8654, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.6953846153846155, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 5.761538461538462e-06, | |
| "loss": 0.8744, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.6984615384615385, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 5.7538461538461545e-06, | |
| "loss": 0.8588, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.7015384615384614, | |
| "grad_norm": 0.1357421875, | |
| "learning_rate": 5.746153846153847e-06, | |
| "loss": 0.8006, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.7046153846153846, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 5.738461538461539e-06, | |
| "loss": 0.8487, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.7076923076923078, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 5.730769230769231e-06, | |
| "loss": 0.832, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.7107692307692308, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 5.723076923076923e-06, | |
| "loss": 0.8665, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.7138461538461538, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 5.715384615384616e-06, | |
| "loss": 0.8505, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.7169230769230768, | |
| "grad_norm": 0.08740234375, | |
| "learning_rate": 5.7076923076923086e-06, | |
| "loss": 0.8525, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 5.7e-06, | |
| "loss": 0.8791, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.7230769230769232, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 5.692307692307692e-06, | |
| "loss": 0.8009, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.7261538461538461, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 5.684615384615385e-06, | |
| "loss": 0.8542, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.7292307692307691, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 5.676923076923078e-06, | |
| "loss": 0.8773, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.7323076923076923, | |
| "grad_norm": 0.08642578125, | |
| "learning_rate": 5.66923076923077e-06, | |
| "loss": 0.8037, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.7353846153846155, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 5.661538461538462e-06, | |
| "loss": 0.8224, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.7384615384615385, | |
| "grad_norm": 0.08447265625, | |
| "learning_rate": 5.6538461538461546e-06, | |
| "loss": 0.794, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.7415384615384615, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 5.6461538461538465e-06, | |
| "loss": 0.8755, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.7446153846153845, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 5.638461538461539e-06, | |
| "loss": 0.8717, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.7476923076923077, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 5.63076923076923e-06, | |
| "loss": 0.8368, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.7507692307692309, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 5.623076923076923e-06, | |
| "loss": 0.8683, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.7538461538461538, | |
| "grad_norm": 0.1474609375, | |
| "learning_rate": 5.615384615384616e-06, | |
| "loss": 0.8568, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.7569230769230768, | |
| "grad_norm": 0.119140625, | |
| "learning_rate": 5.607692307692309e-06, | |
| "loss": 0.8209, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 0.831, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.7630769230769232, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 5.5923076923076925e-06, | |
| "loss": 0.8715, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.7661538461538462, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 5.584615384615385e-06, | |
| "loss": 0.8058, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.7692307692307692, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 5.576923076923077e-06, | |
| "loss": 0.8551, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.7723076923076924, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 5.56923076923077e-06, | |
| "loss": 0.8302, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.7753846153846153, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 5.561538461538462e-06, | |
| "loss": 0.8815, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.7784615384615385, | |
| "grad_norm": 0.1328125, | |
| "learning_rate": 5.553846153846154e-06, | |
| "loss": 0.8345, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.7815384615384615, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 5.546153846153847e-06, | |
| "loss": 0.8221, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.7846153846153845, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 5.538461538461539e-06, | |
| "loss": 0.8232, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.7876923076923077, | |
| "grad_norm": 0.08740234375, | |
| "learning_rate": 5.5307692307692305e-06, | |
| "loss": 0.8116, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.790769230769231, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 5.523076923076923e-06, | |
| "loss": 0.8241, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.7938461538461539, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 5.515384615384616e-06, | |
| "loss": 0.8391, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.7969230769230768, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 5.507692307692308e-06, | |
| "loss": 0.8248, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.8478, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.803076923076923, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 5.492307692307693e-06, | |
| "loss": 0.8783, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.8061538461538462, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 5.4846153846153846e-06, | |
| "loss": 0.8399, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.8092307692307692, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 5.476923076923077e-06, | |
| "loss": 0.8846, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.8123076923076922, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 5.46923076923077e-06, | |
| "loss": 0.8854, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.8153846153846154, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 5.461538461538461e-06, | |
| "loss": 0.8517, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.8184615384615386, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 5.453846153846154e-06, | |
| "loss": 0.8367, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.8215384615384616, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 5.446153846153847e-06, | |
| "loss": 0.8222, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.8246153846153845, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 5.4384615384615395e-06, | |
| "loss": 0.8283, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.8276923076923077, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 5.430769230769231e-06, | |
| "loss": 0.8428, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.830769230769231, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 5.423076923076923e-06, | |
| "loss": 0.8184, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.833846153846154, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 5.415384615384615e-06, | |
| "loss": 0.8506, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.8369230769230769, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 5.407692307692308e-06, | |
| "loss": 0.8197, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.8399999999999999, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.8162, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.843076923076923, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 5.392307692307692e-06, | |
| "loss": 0.8506, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.8461538461538463, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 5.384615384615385e-06, | |
| "loss": 0.8537, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.8492307692307692, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 5.3769230769230774e-06, | |
| "loss": 0.8144, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.8523076923076922, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 5.36923076923077e-06, | |
| "loss": 0.8869, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.8553846153846154, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 5.361538461538462e-06, | |
| "loss": 0.8415, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.8584615384615386, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 5.353846153846154e-06, | |
| "loss": 0.8345, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.8615384615384616, | |
| "grad_norm": 0.08740234375, | |
| "learning_rate": 5.346153846153847e-06, | |
| "loss": 0.7687, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.8646153846153846, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 5.338461538461539e-06, | |
| "loss": 0.8033, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.8676923076923075, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 5.3307692307692315e-06, | |
| "loss": 0.858, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.8707692307692307, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 5.323076923076923e-06, | |
| "loss": 0.8792, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.873846153846154, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 5.315384615384615e-06, | |
| "loss": 0.8612, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.876923076923077, | |
| "grad_norm": 0.08837890625, | |
| "learning_rate": 5.307692307692308e-06, | |
| "loss": 0.864, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 5.300000000000001e-06, | |
| "loss": 0.8218, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.883076923076923, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 5.292307692307693e-06, | |
| "loss": 0.8678, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.8861538461538463, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 5.284615384615385e-06, | |
| "loss": 0.8332, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.8892307692307693, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 5.2769230769230775e-06, | |
| "loss": 0.8268, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.8923076923076922, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 5.2692307692307695e-06, | |
| "loss": 0.8476, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.8953846153846152, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 5.261538461538462e-06, | |
| "loss": 0.8443, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.8984615384615384, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 5.253846153846154e-06, | |
| "loss": 0.8532, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.9015384615384616, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 5.246153846153846e-06, | |
| "loss": 0.8679, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.9046153846153846, | |
| "grad_norm": 0.1630859375, | |
| "learning_rate": 5.238461538461539e-06, | |
| "loss": 0.9119, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.9076923076923076, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 5.230769230769232e-06, | |
| "loss": 0.8649, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.9107692307692308, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 5.223076923076924e-06, | |
| "loss": 0.7897, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.913846153846154, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 5.2153846153846155e-06, | |
| "loss": 0.8454, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.916923076923077, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 5.207692307692308e-06, | |
| "loss": 0.8672, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 5.2e-06, | |
| "loss": 0.9015, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.9230769230769231, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 5.192307692307693e-06, | |
| "loss": 0.8276, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.926153846153846, | |
| "grad_norm": 0.08740234375, | |
| "learning_rate": 5.184615384615385e-06, | |
| "loss": 0.8351, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.9292307692307693, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 5.176923076923077e-06, | |
| "loss": 0.8533, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.9323076923076923, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 5.16923076923077e-06, | |
| "loss": 0.9086, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.9353846153846153, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 5.161538461538462e-06, | |
| "loss": 0.842, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.9384615384615385, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 5.1538461538461534e-06, | |
| "loss": 0.8796, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.9415384615384617, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 5.146153846153846e-06, | |
| "loss": 0.8477, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.9446153846153846, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 5.138461538461539e-06, | |
| "loss": 0.9049, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.9476923076923076, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 5.130769230769232e-06, | |
| "loss": 0.8354, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.9507692307692308, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 5.123076923076924e-06, | |
| "loss": 0.849, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.953846153846154, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 5.115384615384616e-06, | |
| "loss": 0.8896, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.956923076923077, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 5.1076923076923075e-06, | |
| "loss": 0.779, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 5.1e-06, | |
| "loss": 0.8864, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.963076923076923, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 5.092307692307693e-06, | |
| "loss": 0.8349, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.9661538461538461, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 5.084615384615384e-06, | |
| "loss": 0.8751, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.9692307692307693, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 5.076923076923077e-06, | |
| "loss": 0.8253, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.9723076923076923, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 5.06923076923077e-06, | |
| "loss": 0.8697, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.9753846153846153, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 5.0615384615384625e-06, | |
| "loss": 0.8202, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.9784615384615385, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 5.053846153846154e-06, | |
| "loss": 0.8122, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.9815384615384617, | |
| "grad_norm": 0.0849609375, | |
| "learning_rate": 5.046153846153846e-06, | |
| "loss": 0.8222, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.9846153846153847, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 5.038461538461539e-06, | |
| "loss": 0.8236, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.9876923076923076, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 5.030769230769231e-06, | |
| "loss": 0.8374, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.9907692307692306, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 5.023076923076924e-06, | |
| "loss": 0.8623, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.9938461538461538, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 5.015384615384616e-06, | |
| "loss": 0.7917, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.996923076923077, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 5.007692307692308e-06, | |
| "loss": 0.814, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 5e-06, | |
| "loss": 0.7785, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.003076923076923, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 4.992307692307693e-06, | |
| "loss": 0.7981, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 2.006153846153846, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 4.984615384615385e-06, | |
| "loss": 0.8724, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 2.0092307692307694, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 4.976923076923078e-06, | |
| "loss": 0.8657, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 2.0123076923076924, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 4.96923076923077e-06, | |
| "loss": 0.8614, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 2.0153846153846153, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 4.961538461538462e-06, | |
| "loss": 0.8729, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.0184615384615383, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 4.9538461538461545e-06, | |
| "loss": 0.8664, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 2.0215384615384617, | |
| "grad_norm": 0.0849609375, | |
| "learning_rate": 4.9461538461538464e-06, | |
| "loss": 0.8065, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 2.0246153846153847, | |
| "grad_norm": 0.1201171875, | |
| "learning_rate": 4.938461538461538e-06, | |
| "loss": 0.8713, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 2.0276923076923077, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 4.930769230769231e-06, | |
| "loss": 0.8107, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 2.0307692307692307, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 4.923076923076924e-06, | |
| "loss": 0.906, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.0338461538461536, | |
| "grad_norm": 1.140625, | |
| "learning_rate": 4.915384615384616e-06, | |
| "loss": 0.8177, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 2.036923076923077, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 4.907692307692309e-06, | |
| "loss": 0.8576, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.8598, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 2.043076923076923, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 4.892307692307693e-06, | |
| "loss": 0.8523, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 2.046153846153846, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 4.884615384615385e-06, | |
| "loss": 0.8192, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.0492307692307694, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 4.876923076923077e-06, | |
| "loss": 0.8609, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 2.0523076923076924, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 4.869230769230769e-06, | |
| "loss": 0.8368, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 2.0553846153846154, | |
| "grad_norm": 0.08837890625, | |
| "learning_rate": 4.861538461538462e-06, | |
| "loss": 0.8177, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 2.0584615384615383, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 4.853846153846154e-06, | |
| "loss": 0.7955, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 2.0615384615384613, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 4.8461538461538465e-06, | |
| "loss": 0.8117, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.0646153846153847, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 4.838461538461539e-06, | |
| "loss": 0.7957, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 2.0676923076923077, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 4.830769230769231e-06, | |
| "loss": 0.8208, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 2.0707692307692307, | |
| "grad_norm": 0.119140625, | |
| "learning_rate": 4.823076923076924e-06, | |
| "loss": 0.8736, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 2.0738461538461537, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 4.815384615384616e-06, | |
| "loss": 0.8157, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 2.076923076923077, | |
| "grad_norm": 0.0859375, | |
| "learning_rate": 4.807692307692308e-06, | |
| "loss": 0.8487, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 0.8638, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 2.083076923076923, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 4.7923076923076926e-06, | |
| "loss": 0.8407, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 2.086153846153846, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 4.7846153846153845e-06, | |
| "loss": 0.8237, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 2.0892307692307694, | |
| "grad_norm": 0.08740234375, | |
| "learning_rate": 4.776923076923077e-06, | |
| "loss": 0.8204, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 2.0923076923076924, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 4.76923076923077e-06, | |
| "loss": 0.8663, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.0953846153846154, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 4.761538461538462e-06, | |
| "loss": 0.8336, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 2.0984615384615384, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 4.753846153846155e-06, | |
| "loss": 0.8479, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 2.1015384615384614, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 4.746153846153847e-06, | |
| "loss": 0.7951, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 2.1046153846153848, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 4.738461538461539e-06, | |
| "loss": 0.8763, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 2.1076923076923078, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 4.730769230769231e-06, | |
| "loss": 0.838, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.1107692307692307, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 4.723076923076923e-06, | |
| "loss": 0.8611, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 2.1138461538461537, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 4.715384615384615e-06, | |
| "loss": 0.7858, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 2.116923076923077, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 4.707692307692308e-06, | |
| "loss": 0.871, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "grad_norm": 0.1640625, | |
| "learning_rate": 4.7e-06, | |
| "loss": 0.8521, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 2.123076923076923, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 4.692307692307693e-06, | |
| "loss": 0.8546, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.126153846153846, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 4.6846153846153854e-06, | |
| "loss": 0.8083, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 2.129230769230769, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 4.676923076923077e-06, | |
| "loss": 0.8394, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 2.1323076923076925, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 4.66923076923077e-06, | |
| "loss": 0.8492, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 2.1353846153846154, | |
| "grad_norm": 0.12109375, | |
| "learning_rate": 4.661538461538462e-06, | |
| "loss": 0.8377, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 2.1384615384615384, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 4.653846153846154e-06, | |
| "loss": 0.8229, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.1415384615384614, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 4.646153846153847e-06, | |
| "loss": 0.8503, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 2.144615384615385, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 4.638461538461539e-06, | |
| "loss": 0.8387, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 2.147692307692308, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 4.630769230769231e-06, | |
| "loss": 0.8573, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 2.1507692307692308, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 4.623076923076923e-06, | |
| "loss": 0.8396, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 2.1538461538461537, | |
| "grad_norm": 0.08740234375, | |
| "learning_rate": 4.615384615384616e-06, | |
| "loss": 0.8074, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.1569230769230767, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 4.607692307692308e-06, | |
| "loss": 0.8798, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 4.600000000000001e-06, | |
| "loss": 0.8265, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 2.163076923076923, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 4.592307692307693e-06, | |
| "loss": 0.8472, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 2.166153846153846, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 4.5846153846153855e-06, | |
| "loss": 0.7898, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 2.169230769230769, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 4.5769230769230775e-06, | |
| "loss": 0.8883, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.1723076923076925, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 4.569230769230769e-06, | |
| "loss": 0.8465, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 2.1753846153846155, | |
| "grad_norm": 0.158203125, | |
| "learning_rate": 4.561538461538461e-06, | |
| "loss": 0.8762, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 2.1784615384615384, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 4.553846153846154e-06, | |
| "loss": 0.848, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 2.1815384615384614, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 4.546153846153846e-06, | |
| "loss": 0.8394, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 2.184615384615385, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 4.538461538461539e-06, | |
| "loss": 0.844, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.187692307692308, | |
| "grad_norm": 0.1162109375, | |
| "learning_rate": 4.5307692307692316e-06, | |
| "loss": 0.9077, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 2.190769230769231, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 4.5230769230769235e-06, | |
| "loss": 0.8238, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 2.193846153846154, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 4.515384615384616e-06, | |
| "loss": 0.8957, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 2.1969230769230768, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 4.507692307692308e-06, | |
| "loss": 0.8179, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.7926, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.203076923076923, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 4.492307692307693e-06, | |
| "loss": 0.8131, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 2.206153846153846, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 4.484615384615385e-06, | |
| "loss": 0.8277, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 2.209230769230769, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 4.476923076923077e-06, | |
| "loss": 0.8576, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 2.212307692307692, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 4.4692307692307695e-06, | |
| "loss": 0.8988, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 2.2153846153846155, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 4.461538461538462e-06, | |
| "loss": 0.8054, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.2184615384615385, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 4.453846153846154e-06, | |
| "loss": 0.8193, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 2.2215384615384615, | |
| "grad_norm": 0.138671875, | |
| "learning_rate": 4.446153846153847e-06, | |
| "loss": 0.842, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 2.2246153846153844, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 4.438461538461539e-06, | |
| "loss": 0.7995, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 2.227692307692308, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 4.430769230769232e-06, | |
| "loss": 0.8814, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 2.230769230769231, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 4.423076923076924e-06, | |
| "loss": 0.8994, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.233846153846154, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 4.4153846153846155e-06, | |
| "loss": 0.8067, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 2.236923076923077, | |
| "grad_norm": 0.138671875, | |
| "learning_rate": 4.4076923076923075e-06, | |
| "loss": 0.8696, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 0.08251953125, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.7729, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 2.243076923076923, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 4.392307692307693e-06, | |
| "loss": 0.7773, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 2.246153846153846, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 4.384615384615385e-06, | |
| "loss": 0.8462, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.249230769230769, | |
| "grad_norm": 0.0810546875, | |
| "learning_rate": 4.376923076923078e-06, | |
| "loss": 0.7747, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 2.252307692307692, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 4.36923076923077e-06, | |
| "loss": 0.8327, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 2.2553846153846155, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 4.361538461538462e-06, | |
| "loss": 0.843, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 2.2584615384615385, | |
| "grad_norm": 0.14453125, | |
| "learning_rate": 4.353846153846154e-06, | |
| "loss": 0.8971, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 2.2615384615384615, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 4.346153846153846e-06, | |
| "loss": 0.8006, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.2646153846153845, | |
| "grad_norm": 0.1357421875, | |
| "learning_rate": 4.338461538461539e-06, | |
| "loss": 0.8402, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 2.2676923076923075, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 4.330769230769231e-06, | |
| "loss": 0.8706, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 2.270769230769231, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 4.323076923076923e-06, | |
| "loss": 0.8393, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 2.273846153846154, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 4.315384615384616e-06, | |
| "loss": 0.8291, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 2.276923076923077, | |
| "grad_norm": 0.08447265625, | |
| "learning_rate": 4.307692307692308e-06, | |
| "loss": 0.7642, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.2800000000000002, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 4.3e-06, | |
| "loss": 0.8262, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 2.2830769230769232, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 4.292307692307693e-06, | |
| "loss": 0.7732, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 2.286153846153846, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 4.284615384615385e-06, | |
| "loss": 0.8437, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 2.289230769230769, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 4.276923076923078e-06, | |
| "loss": 0.8348, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 2.292307692307692, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 4.26923076923077e-06, | |
| "loss": 0.8891, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 2.2953846153846156, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 4.261538461538462e-06, | |
| "loss": 0.875, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 2.2984615384615386, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 4.253846153846154e-06, | |
| "loss": 0.8852, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 2.3015384615384615, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 4.246153846153846e-06, | |
| "loss": 0.8323, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 2.3046153846153845, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 4.238461538461539e-06, | |
| "loss": 0.8213, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 2.3076923076923075, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 4.230769230769231e-06, | |
| "loss": 0.8336, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.310769230769231, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 4.223076923076924e-06, | |
| "loss": 0.8863, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 2.313846153846154, | |
| "grad_norm": 0.08740234375, | |
| "learning_rate": 4.215384615384616e-06, | |
| "loss": 0.7972, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 2.316923076923077, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 4.2076923076923085e-06, | |
| "loss": 0.8558, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 0.8157, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 2.3230769230769233, | |
| "grad_norm": 0.11767578125, | |
| "learning_rate": 4.192307692307692e-06, | |
| "loss": 0.9043, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.3261538461538462, | |
| "grad_norm": 0.1357421875, | |
| "learning_rate": 4.184615384615385e-06, | |
| "loss": 0.8562, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 2.329230769230769, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 4.176923076923077e-06, | |
| "loss": 0.8145, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 2.332307692307692, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 4.169230769230769e-06, | |
| "loss": 0.842, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 2.3353846153846156, | |
| "grad_norm": 0.123046875, | |
| "learning_rate": 4.161538461538462e-06, | |
| "loss": 0.9075, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 2.3384615384615386, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 4.1538461538461545e-06, | |
| "loss": 0.844, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.3415384615384616, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 4.1461538461538465e-06, | |
| "loss": 0.8607, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 2.3446153846153845, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 4.138461538461539e-06, | |
| "loss": 0.8982, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 2.3476923076923075, | |
| "grad_norm": 0.08837890625, | |
| "learning_rate": 4.130769230769231e-06, | |
| "loss": 0.8, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 2.350769230769231, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 4.123076923076924e-06, | |
| "loss": 0.8328, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 2.353846153846154, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 4.115384615384616e-06, | |
| "loss": 0.8801, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.356923076923077, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 4.107692307692308e-06, | |
| "loss": 0.8381, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 4.1e-06, | |
| "loss": 0.8489, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 2.363076923076923, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 4.0923076923076925e-06, | |
| "loss": 0.8314, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 2.3661538461538463, | |
| "grad_norm": 0.193359375, | |
| "learning_rate": 4.084615384615385e-06, | |
| "loss": 0.8445, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 2.3692307692307693, | |
| "grad_norm": 0.1728515625, | |
| "learning_rate": 4.076923076923077e-06, | |
| "loss": 0.8301, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.3723076923076922, | |
| "grad_norm": 0.0888671875, | |
| "learning_rate": 4.06923076923077e-06, | |
| "loss": 0.8352, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 2.375384615384615, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 4.061538461538462e-06, | |
| "loss": 0.8491, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 2.3784615384615386, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 4.053846153846155e-06, | |
| "loss": 0.8352, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 2.3815384615384616, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 4.0461538461538466e-06, | |
| "loss": 0.8162, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 2.3846153846153846, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 4.0384615384615385e-06, | |
| "loss": 0.8674, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.3876923076923076, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 4.030769230769231e-06, | |
| "loss": 0.8667, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 2.390769230769231, | |
| "grad_norm": 0.177734375, | |
| "learning_rate": 4.023076923076923e-06, | |
| "loss": 0.8449, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 2.393846153846154, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 4.015384615384615e-06, | |
| "loss": 0.8385, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 2.396923076923077, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 4.007692307692308e-06, | |
| "loss": 0.8664, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.8424, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.403076923076923, | |
| "grad_norm": 0.0869140625, | |
| "learning_rate": 3.992307692307693e-06, | |
| "loss": 0.8286, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 2.4061538461538463, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 3.984615384615385e-06, | |
| "loss": 0.8398, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 2.4092307692307693, | |
| "grad_norm": 5.875, | |
| "learning_rate": 3.976923076923077e-06, | |
| "loss": 0.8831, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 2.4123076923076923, | |
| "grad_norm": 0.08642578125, | |
| "learning_rate": 3.96923076923077e-06, | |
| "loss": 0.7996, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 2.4153846153846152, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 3.961538461538462e-06, | |
| "loss": 0.8037, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.418461538461538, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 3.953846153846154e-06, | |
| "loss": 0.8088, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 2.4215384615384616, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 3.946153846153846e-06, | |
| "loss": 0.842, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 2.4246153846153846, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 3.938461538461539e-06, | |
| "loss": 0.839, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 2.4276923076923076, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 3.930769230769231e-06, | |
| "loss": 0.815, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 2.430769230769231, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 3.923076923076923e-06, | |
| "loss": 0.7884, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.433846153846154, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 3.915384615384616e-06, | |
| "loss": 0.8539, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 2.436923076923077, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 3.907692307692308e-06, | |
| "loss": 0.8395, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.875, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 2.443076923076923, | |
| "grad_norm": 0.173828125, | |
| "learning_rate": 3.892307692307693e-06, | |
| "loss": 0.8621, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 2.4461538461538463, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 3.884615384615385e-06, | |
| "loss": 0.782, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.4492307692307693, | |
| "grad_norm": 0.134765625, | |
| "learning_rate": 3.876923076923077e-06, | |
| "loss": 0.8388, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 2.4523076923076923, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 3.869230769230769e-06, | |
| "loss": 0.839, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 2.4553846153846153, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 3.861538461538462e-06, | |
| "loss": 0.821, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 2.4584615384615383, | |
| "grad_norm": 0.11328125, | |
| "learning_rate": 3.853846153846154e-06, | |
| "loss": 0.8577, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 2.4615384615384617, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 3.846153846153847e-06, | |
| "loss": 0.8354, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.4646153846153847, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 3.838461538461539e-06, | |
| "loss": 0.8276, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 2.4676923076923076, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 3.8307692307692315e-06, | |
| "loss": 0.8319, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 2.4707692307692306, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 3.823076923076923e-06, | |
| "loss": 0.8379, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 2.473846153846154, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 3.815384615384616e-06, | |
| "loss": 0.8468, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 2.476923076923077, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 3.8076923076923077e-06, | |
| "loss": 0.8328, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 3.8000000000000005e-06, | |
| "loss": 0.8269, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 2.483076923076923, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 3.7923076923076924e-06, | |
| "loss": 0.8509, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 2.4861538461538464, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 3.7846153846153847e-06, | |
| "loss": 0.8148, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 2.4892307692307694, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 3.7769230769230775e-06, | |
| "loss": 0.8323, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 2.4923076923076923, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 3.7692307692307694e-06, | |
| "loss": 0.8523, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.4953846153846153, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 3.761538461538462e-06, | |
| "loss": 0.8342, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 2.4984615384615383, | |
| "grad_norm": 0.150390625, | |
| "learning_rate": 3.753846153846154e-06, | |
| "loss": 0.8209, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 2.5015384615384617, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 3.7461538461538465e-06, | |
| "loss": 0.7825, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 2.5046153846153847, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 3.7384615384615384e-06, | |
| "loss": 0.8882, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 2.5076923076923077, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 3.730769230769231e-06, | |
| "loss": 0.8659, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.5107692307692306, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 3.723076923076923e-06, | |
| "loss": 0.8428, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 2.5138461538461536, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 3.715384615384616e-06, | |
| "loss": 0.8087, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 2.516923076923077, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 3.7076923076923082e-06, | |
| "loss": 0.8225, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 3.7e-06, | |
| "loss": 0.8777, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 2.523076923076923, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 3.692307692307693e-06, | |
| "loss": 0.8345, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.5261538461538464, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 3.684615384615385e-06, | |
| "loss": 0.8693, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 2.5292307692307694, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 3.676923076923077e-06, | |
| "loss": 0.8334, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 2.5323076923076924, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 3.6692307692307695e-06, | |
| "loss": 0.8881, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 2.5353846153846153, | |
| "grad_norm": 0.11572265625, | |
| "learning_rate": 3.661538461538462e-06, | |
| "loss": 0.8611, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 2.5384615384615383, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 3.653846153846154e-06, | |
| "loss": 0.8416, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.5415384615384617, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 3.6461538461538466e-06, | |
| "loss": 0.8201, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 2.5446153846153847, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 3.6384615384615385e-06, | |
| "loss": 0.8278, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 2.5476923076923077, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 3.630769230769231e-06, | |
| "loss": 0.8115, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 2.5507692307692307, | |
| "grad_norm": 0.1220703125, | |
| "learning_rate": 3.6230769230769236e-06, | |
| "loss": 0.827, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 2.5538461538461537, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 3.6153846153846156e-06, | |
| "loss": 0.8178, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.556923076923077, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 3.6076923076923083e-06, | |
| "loss": 0.864, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 0.815, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 2.563076923076923, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 3.5923076923076926e-06, | |
| "loss": 0.8371, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 2.566153846153846, | |
| "grad_norm": 0.0849609375, | |
| "learning_rate": 3.5846153846153845e-06, | |
| "loss": 0.8077, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 2.569230769230769, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 3.5769230769230773e-06, | |
| "loss": 0.8622, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 2.5723076923076924, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 3.5692307692307692e-06, | |
| "loss": 0.809, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 2.5753846153846154, | |
| "grad_norm": 0.0888671875, | |
| "learning_rate": 3.561538461538462e-06, | |
| "loss": 0.8176, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 2.5784615384615384, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 3.5538461538461544e-06, | |
| "loss": 0.8104, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 2.581538461538462, | |
| "grad_norm": 0.1162109375, | |
| "learning_rate": 3.5461538461538463e-06, | |
| "loss": 0.8187, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 2.5846153846153848, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 3.538461538461539e-06, | |
| "loss": 0.8103, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.5876923076923077, | |
| "grad_norm": 0.1201171875, | |
| "learning_rate": 3.530769230769231e-06, | |
| "loss": 0.8413, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 2.5907692307692307, | |
| "grad_norm": 0.1162109375, | |
| "learning_rate": 3.5230769230769233e-06, | |
| "loss": 0.7912, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 2.5938461538461537, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 3.5153846153846157e-06, | |
| "loss": 0.815, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 2.596923076923077, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 3.507692307692308e-06, | |
| "loss": 0.861, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.8215, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 2.603076923076923, | |
| "grad_norm": 0.08837890625, | |
| "learning_rate": 3.4923076923076927e-06, | |
| "loss": 0.8155, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 2.606153846153846, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 3.484615384615385e-06, | |
| "loss": 0.865, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 2.609230769230769, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 3.476923076923077e-06, | |
| "loss": 0.7879, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 2.6123076923076924, | |
| "grad_norm": 0.1201171875, | |
| "learning_rate": 3.4692307692307698e-06, | |
| "loss": 0.8391, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 2.6153846153846154, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 3.4615384615384617e-06, | |
| "loss": 0.8525, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.6184615384615384, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 3.4538461538461545e-06, | |
| "loss": 0.8282, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 2.6215384615384614, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 3.4461538461538464e-06, | |
| "loss": 0.8206, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 2.6246153846153844, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 3.4384615384615387e-06, | |
| "loss": 0.8105, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 2.6276923076923078, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 3.4307692307692307e-06, | |
| "loss": 0.8337, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 2.6307692307692307, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 3.4230769230769234e-06, | |
| "loss": 0.8593, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 2.6338461538461537, | |
| "grad_norm": 0.08349609375, | |
| "learning_rate": 3.4153846153846154e-06, | |
| "loss": 0.7969, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 2.636923076923077, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 3.407692307692308e-06, | |
| "loss": 0.9084, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.8435, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 2.643076923076923, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 3.3923076923076924e-06, | |
| "loss": 0.8719, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 2.646153846153846, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 3.384615384615385e-06, | |
| "loss": 0.8013, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.649230769230769, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 3.376923076923077e-06, | |
| "loss": 0.7983, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 2.6523076923076925, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 3.3692307692307695e-06, | |
| "loss": 0.8435, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 2.6553846153846155, | |
| "grad_norm": 0.125, | |
| "learning_rate": 3.361538461538462e-06, | |
| "loss": 0.8516, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 2.6584615384615384, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 3.353846153846154e-06, | |
| "loss": 0.8525, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 2.6615384615384614, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 3.346153846153846e-06, | |
| "loss": 0.8108, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 2.6646153846153844, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 3.338461538461539e-06, | |
| "loss": 0.8115, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 2.667692307692308, | |
| "grad_norm": 0.138671875, | |
| "learning_rate": 3.330769230769231e-06, | |
| "loss": 0.8325, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 2.670769230769231, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 3.323076923076923e-06, | |
| "loss": 0.8666, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 2.6738461538461538, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 3.315384615384616e-06, | |
| "loss": 0.821, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 2.676923076923077, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 3.307692307692308e-06, | |
| "loss": 0.9241, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 3.3000000000000006e-06, | |
| "loss": 0.8109, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 2.683076923076923, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 3.2923076923076925e-06, | |
| "loss": 0.8678, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 2.686153846153846, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 3.284615384615385e-06, | |
| "loss": 0.8489, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 2.689230769230769, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 3.276923076923077e-06, | |
| "loss": 0.834, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 2.6923076923076925, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 3.2692307692307696e-06, | |
| "loss": 0.8562, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.6953846153846155, | |
| "grad_norm": 0.1416015625, | |
| "learning_rate": 3.2615384615384615e-06, | |
| "loss": 0.796, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 2.6984615384615385, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 3.2538461538461543e-06, | |
| "loss": 0.8541, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 2.7015384615384614, | |
| "grad_norm": 0.1552734375, | |
| "learning_rate": 3.2461538461538466e-06, | |
| "loss": 0.9153, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 2.7046153846153844, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 3.2384615384615385e-06, | |
| "loss": 0.8241, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 2.707692307692308, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 3.2307692307692313e-06, | |
| "loss": 0.8111, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.710769230769231, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 3.2230769230769232e-06, | |
| "loss": 0.862, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 2.713846153846154, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 3.2153846153846156e-06, | |
| "loss": 0.8111, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 2.7169230769230768, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 3.207692307692308e-06, | |
| "loss": 0.8156, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 2.7199999999999998, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 0.8807, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 2.723076923076923, | |
| "grad_norm": 0.0830078125, | |
| "learning_rate": 3.192307692307692e-06, | |
| "loss": 0.7752, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 2.726153846153846, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 3.184615384615385e-06, | |
| "loss": 0.8353, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 2.729230769230769, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 3.1769230769230773e-06, | |
| "loss": 0.8478, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 2.7323076923076925, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 3.1692307692307693e-06, | |
| "loss": 0.8805, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 2.7353846153846155, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 3.161538461538462e-06, | |
| "loss": 0.8351, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 2.7384615384615385, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 3.153846153846154e-06, | |
| "loss": 0.8231, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.7415384615384615, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 3.1461538461538467e-06, | |
| "loss": 0.8297, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 2.7446153846153845, | |
| "grad_norm": 0.12060546875, | |
| "learning_rate": 3.1384615384615386e-06, | |
| "loss": 0.8146, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 2.747692307692308, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 3.130769230769231e-06, | |
| "loss": 0.8378, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 2.750769230769231, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 3.123076923076923e-06, | |
| "loss": 0.8502, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 2.753846153846154, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 3.1153846153846157e-06, | |
| "loss": 0.8299, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 2.756923076923077, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 3.1076923076923076e-06, | |
| "loss": 0.8527, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 3.1000000000000004e-06, | |
| "loss": 0.8136, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 2.763076923076923, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 3.0923076923076927e-06, | |
| "loss": 0.8182, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 2.766153846153846, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 3.0846153846153847e-06, | |
| "loss": 0.8575, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 2.769230769230769, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 3.0769230769230774e-06, | |
| "loss": 0.8378, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.7723076923076926, | |
| "grad_norm": 0.08984375, | |
| "learning_rate": 3.0692307692307694e-06, | |
| "loss": 0.8273, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 2.775384615384615, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 3.0615384615384617e-06, | |
| "loss": 0.8271, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 2.7784615384615385, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 3.053846153846154e-06, | |
| "loss": 0.8232, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 2.7815384615384615, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 3.0461538461538464e-06, | |
| "loss": 0.8078, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.7846153846153845, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 3.0384615384615383e-06, | |
| "loss": 0.8171, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.787692307692308, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 3.030769230769231e-06, | |
| "loss": 0.8276, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 2.790769230769231, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 3.0230769230769235e-06, | |
| "loss": 0.8269, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 2.793846153846154, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 3.0153846153846154e-06, | |
| "loss": 0.8309, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.796923076923077, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 3.007692307692308e-06, | |
| "loss": 0.8517, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 3e-06, | |
| "loss": 0.802, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.8030769230769232, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 2.992307692307693e-06, | |
| "loss": 0.8316, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 2.806153846153846, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 2.9846153846153848e-06, | |
| "loss": 0.9295, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.809230769230769, | |
| "grad_norm": 0.12451171875, | |
| "learning_rate": 2.976923076923077e-06, | |
| "loss": 0.922, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 2.812307692307692, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 2.969230769230769e-06, | |
| "loss": 0.8601, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.815384615384615, | |
| "grad_norm": 0.1396484375, | |
| "learning_rate": 2.961538461538462e-06, | |
| "loss": 0.844, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.8184615384615386, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 2.953846153846154e-06, | |
| "loss": 0.8618, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.8215384615384616, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 2.9461538461538465e-06, | |
| "loss": 0.8424, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 2.8246153846153845, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 2.938461538461539e-06, | |
| "loss": 0.8492, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.827692307692308, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 2.930769230769231e-06, | |
| "loss": 0.8379, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 2.830769230769231, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 2.9230769230769236e-06, | |
| "loss": 0.751, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.833846153846154, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 2.9153846153846155e-06, | |
| "loss": 0.7931, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 2.836923076923077, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 2.907692307692308e-06, | |
| "loss": 0.8097, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.8504, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 2.8430769230769233, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 2.8923076923076925e-06, | |
| "loss": 0.8251, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.8461538461538463, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 2.8846153846153845e-06, | |
| "loss": 0.8366, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.8492307692307692, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 2.8769230769230772e-06, | |
| "loss": 0.8211, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 2.852307692307692, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 2.8692307692307696e-06, | |
| "loss": 0.8284, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 2.855384615384615, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 2.8615384615384615e-06, | |
| "loss": 0.8488, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 2.8584615384615386, | |
| "grad_norm": 0.1611328125, | |
| "learning_rate": 2.8538461538461543e-06, | |
| "loss": 0.8934, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 2.8615384615384616, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 2.846153846153846e-06, | |
| "loss": 0.8847, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.8646153846153846, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 2.838461538461539e-06, | |
| "loss": 0.8357, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 2.8676923076923075, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 2.830769230769231e-06, | |
| "loss": 0.79, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 2.8707692307692305, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 2.8230769230769233e-06, | |
| "loss": 0.7898, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 2.873846153846154, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 2.815384615384615e-06, | |
| "loss": 0.8311, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 2.876923076923077, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 2.807692307692308e-06, | |
| "loss": 0.8601, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 0.1455078125, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 0.8447, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 2.8830769230769233, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 2.7923076923076926e-06, | |
| "loss": 0.879, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 2.8861538461538463, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 2.784615384615385e-06, | |
| "loss": 0.8843, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 2.8892307692307693, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 2.776923076923077e-06, | |
| "loss": 0.767, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 2.8923076923076922, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 2.7692307692307697e-06, | |
| "loss": 0.8124, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.8953846153846152, | |
| "grad_norm": 0.15625, | |
| "learning_rate": 2.7615384615384616e-06, | |
| "loss": 0.8878, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 2.8984615384615386, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 2.753846153846154e-06, | |
| "loss": 0.854, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 2.9015384615384616, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 2.7461538461538463e-06, | |
| "loss": 0.8826, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 2.9046153846153846, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 2.7384615384615387e-06, | |
| "loss": 0.8155, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 2.9076923076923076, | |
| "grad_norm": 0.12158203125, | |
| "learning_rate": 2.7307692307692306e-06, | |
| "loss": 0.8203, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.9107692307692306, | |
| "grad_norm": 0.125, | |
| "learning_rate": 2.7230769230769234e-06, | |
| "loss": 0.8637, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 2.913846153846154, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 2.7153846153846157e-06, | |
| "loss": 0.8762, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 2.916923076923077, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 2.7076923076923076e-06, | |
| "loss": 0.7989, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 2.7000000000000004e-06, | |
| "loss": 0.8157, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 2.9230769230769234, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 2.6923076923076923e-06, | |
| "loss": 0.8408, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.926153846153846, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 2.684615384615385e-06, | |
| "loss": 0.8257, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 2.9292307692307693, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 2.676923076923077e-06, | |
| "loss": 0.7587, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 2.9323076923076923, | |
| "grad_norm": 0.08642578125, | |
| "learning_rate": 2.6692307692307694e-06, | |
| "loss": 0.81, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 2.9353846153846153, | |
| "grad_norm": 0.1337890625, | |
| "learning_rate": 2.6615384615384613e-06, | |
| "loss": 0.8491, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 2.9384615384615387, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 2.653846153846154e-06, | |
| "loss": 0.8225, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.9415384615384617, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 2.6461538461538464e-06, | |
| "loss": 0.7967, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 2.9446153846153846, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 2.6384615384615388e-06, | |
| "loss": 0.874, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 2.9476923076923076, | |
| "grad_norm": 0.08642578125, | |
| "learning_rate": 2.630769230769231e-06, | |
| "loss": 0.7837, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 2.9507692307692306, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 2.623076923076923e-06, | |
| "loss": 0.8478, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 2.953846153846154, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 2.615384615384616e-06, | |
| "loss": 0.8429, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.956923076923077, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 2.6076923076923077e-06, | |
| "loss": 0.8162, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 0.1484375, | |
| "learning_rate": 2.6e-06, | |
| "loss": 0.8765, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 2.963076923076923, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 2.5923076923076924e-06, | |
| "loss": 0.8266, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 2.966153846153846, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 2.584615384615385e-06, | |
| "loss": 0.8262, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 2.9692307692307693, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 2.5769230769230767e-06, | |
| "loss": 0.8319, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.9723076923076923, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 2.5692307692307695e-06, | |
| "loss": 0.8662, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 2.9753846153846153, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 2.561538461538462e-06, | |
| "loss": 0.8603, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 2.9784615384615387, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 2.5538461538461538e-06, | |
| "loss": 0.8483, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 2.9815384615384617, | |
| "grad_norm": 0.140625, | |
| "learning_rate": 2.5461538461538465e-06, | |
| "loss": 0.8661, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 2.9846153846153847, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 2.5384615384615385e-06, | |
| "loss": 0.8083, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.9876923076923076, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 2.5307692307692312e-06, | |
| "loss": 0.776, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 2.9907692307692306, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 2.523076923076923e-06, | |
| "loss": 0.8315, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.993846153846154, | |
| "grad_norm": 0.12060546875, | |
| "learning_rate": 2.5153846153846155e-06, | |
| "loss": 0.8195, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 2.996923076923077, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 2.507692307692308e-06, | |
| "loss": 0.7853, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.8047, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 3.003076923076923, | |
| "grad_norm": 0.1923828125, | |
| "learning_rate": 2.4923076923076926e-06, | |
| "loss": 0.8049, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 3.006153846153846, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 2.484615384615385e-06, | |
| "loss": 0.87, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 3.0092307692307694, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 2.4769230769230773e-06, | |
| "loss": 0.7973, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 3.0123076923076924, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 2.469230769230769e-06, | |
| "loss": 0.795, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 3.0153846153846153, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 2.461538461538462e-06, | |
| "loss": 0.836, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.0184615384615383, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 2.4538461538461543e-06, | |
| "loss": 0.8617, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 3.0215384615384617, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 2.4461538461538466e-06, | |
| "loss": 0.8123, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 3.0246153846153847, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 2.4384615384615386e-06, | |
| "loss": 0.8631, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 3.0276923076923077, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 2.430769230769231e-06, | |
| "loss": 0.8203, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 3.0307692307692307, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 2.4230769230769233e-06, | |
| "loss": 0.843, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 3.0338461538461536, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 2.4153846153846156e-06, | |
| "loss": 0.8456, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 3.036923076923077, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 2.407692307692308e-06, | |
| "loss": 0.8643, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.8085, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 3.043076923076923, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 2.3923076923076922e-06, | |
| "loss": 0.7958, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 3.046153846153846, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 2.384615384615385e-06, | |
| "loss": 0.8405, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 3.0492307692307694, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 2.3769230769230774e-06, | |
| "loss": 0.8325, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 3.0523076923076924, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 2.3692307692307697e-06, | |
| "loss": 0.8314, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 3.0553846153846154, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 2.3615384615384616e-06, | |
| "loss": 0.8583, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 3.0584615384615383, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 2.353846153846154e-06, | |
| "loss": 0.8523, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 3.0615384615384613, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 2.3461538461538463e-06, | |
| "loss": 0.8121, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 3.0646153846153847, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 2.3384615384615387e-06, | |
| "loss": 0.8164, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 3.0676923076923077, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 2.330769230769231e-06, | |
| "loss": 0.8583, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 3.0707692307692307, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 2.3230769230769234e-06, | |
| "loss": 0.8755, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 3.0738461538461537, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 2.3153846153846153e-06, | |
| "loss": 0.8861, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 3.076923076923077, | |
| "grad_norm": 0.125, | |
| "learning_rate": 2.307692307692308e-06, | |
| "loss": 0.8472, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 2.3000000000000004e-06, | |
| "loss": 0.778, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 3.083076923076923, | |
| "grad_norm": 0.11767578125, | |
| "learning_rate": 2.2923076923076928e-06, | |
| "loss": 0.785, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 3.086153846153846, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 2.2846153846153847e-06, | |
| "loss": 0.8447, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 3.0892307692307694, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 2.276923076923077e-06, | |
| "loss": 0.8487, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 3.0923076923076924, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 2.2692307692307694e-06, | |
| "loss": 0.8539, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 3.0953846153846154, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 2.2615384615384617e-06, | |
| "loss": 0.8309, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 3.0984615384615384, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 2.253846153846154e-06, | |
| "loss": 0.8168, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 3.1015384615384614, | |
| "grad_norm": 0.1201171875, | |
| "learning_rate": 2.2461538461538464e-06, | |
| "loss": 0.8052, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 3.1046153846153848, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 2.2384615384615384e-06, | |
| "loss": 0.8163, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 3.1076923076923078, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 2.230769230769231e-06, | |
| "loss": 0.8507, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 3.1107692307692307, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 2.2230769230769235e-06, | |
| "loss": 0.803, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 3.1138461538461537, | |
| "grad_norm": 0.15234375, | |
| "learning_rate": 2.215384615384616e-06, | |
| "loss": 0.8384, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 3.116923076923077, | |
| "grad_norm": 0.11328125, | |
| "learning_rate": 2.2076923076923078e-06, | |
| "loss": 0.8313, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 2.2e-06, | |
| "loss": 0.8227, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 3.123076923076923, | |
| "grad_norm": 0.125, | |
| "learning_rate": 2.1923076923076925e-06, | |
| "loss": 0.901, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 3.126153846153846, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 2.184615384615385e-06, | |
| "loss": 0.7983, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 3.129230769230769, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 2.176923076923077e-06, | |
| "loss": 0.8535, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 3.1323076923076925, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 2.1692307692307695e-06, | |
| "loss": 0.8274, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 3.1353846153846154, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 2.1615384615384614e-06, | |
| "loss": 0.8226, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 3.1384615384615384, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 2.153846153846154e-06, | |
| "loss": 0.8483, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 3.1415384615384614, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 2.1461538461538466e-06, | |
| "loss": 0.8762, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 3.144615384615385, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 2.138461538461539e-06, | |
| "loss": 0.8003, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 3.147692307692308, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 2.130769230769231e-06, | |
| "loss": 0.8454, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 3.1507692307692308, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 2.123076923076923e-06, | |
| "loss": 0.8665, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 3.1538461538461537, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 2.1153846153846155e-06, | |
| "loss": 0.8421, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 3.1569230769230767, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 2.107692307692308e-06, | |
| "loss": 0.8482, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 2.1000000000000002e-06, | |
| "loss": 0.8336, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 3.163076923076923, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 2.0923076923076926e-06, | |
| "loss": 0.8339, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 3.166153846153846, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 2.0846153846153845e-06, | |
| "loss": 0.8346, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 3.169230769230769, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 2.0769230769230773e-06, | |
| "loss": 0.7939, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 3.1723076923076925, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 2.0692307692307696e-06, | |
| "loss": 0.8439, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 3.1753846153846155, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 2.061538461538462e-06, | |
| "loss": 0.8165, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 3.1784615384615384, | |
| "grad_norm": 0.08740234375, | |
| "learning_rate": 2.053846153846154e-06, | |
| "loss": 0.8137, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 3.1815384615384614, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 2.0461538461538462e-06, | |
| "loss": 0.9064, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 3.184615384615385, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 2.0384615384615386e-06, | |
| "loss": 0.8686, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 3.187692307692308, | |
| "grad_norm": 0.11328125, | |
| "learning_rate": 2.030769230769231e-06, | |
| "loss": 0.8569, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 3.190769230769231, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 2.0230769230769233e-06, | |
| "loss": 0.8218, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 3.193846153846154, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 2.0153846153846156e-06, | |
| "loss": 0.8085, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 3.1969230769230768, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 2.0076923076923076e-06, | |
| "loss": 0.8159, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.7709, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 3.203076923076923, | |
| "grad_norm": 0.0859375, | |
| "learning_rate": 1.9923076923076927e-06, | |
| "loss": 0.8077, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 3.206153846153846, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 1.984615384615385e-06, | |
| "loss": 0.8088, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 3.209230769230769, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 1.976923076923077e-06, | |
| "loss": 0.827, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 3.212307692307692, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 1.9692307692307693e-06, | |
| "loss": 0.8193, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 3.2153846153846155, | |
| "grad_norm": 0.1171875, | |
| "learning_rate": 1.9615384615384617e-06, | |
| "loss": 0.8797, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 3.2184615384615385, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 1.953846153846154e-06, | |
| "loss": 0.7951, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 3.2215384615384615, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 1.9461538461538464e-06, | |
| "loss": 0.8373, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 3.2246153846153844, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 1.9384615384615387e-06, | |
| "loss": 0.7948, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 3.227692307692308, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 1.930769230769231e-06, | |
| "loss": 0.8643, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 3.230769230769231, | |
| "grad_norm": 0.173828125, | |
| "learning_rate": 1.9230769230769234e-06, | |
| "loss": 0.8228, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 3.233846153846154, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 1.9153846153846157e-06, | |
| "loss": 0.8109, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 3.236923076923077, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 1.907692307692308e-06, | |
| "loss": 0.8279, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 0.1162109375, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.8123, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 3.243076923076923, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 1.8923076923076924e-06, | |
| "loss": 0.8399, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 3.246153846153846, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 1.8846153846153847e-06, | |
| "loss": 0.8361, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 3.249230769230769, | |
| "grad_norm": 0.08984375, | |
| "learning_rate": 1.876923076923077e-06, | |
| "loss": 0.7757, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 3.252307692307692, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 1.8692307692307692e-06, | |
| "loss": 0.7979, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 3.2553846153846155, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 1.8615384615384616e-06, | |
| "loss": 0.8985, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 3.2584615384615385, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 1.8538461538461541e-06, | |
| "loss": 0.7853, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 3.2615384615384615, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 1.8461538461538465e-06, | |
| "loss": 0.8481, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 3.2646153846153845, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 1.8384615384615386e-06, | |
| "loss": 0.8863, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 3.2676923076923075, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 1.830769230769231e-06, | |
| "loss": 0.8518, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 3.270769230769231, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 1.8230769230769233e-06, | |
| "loss": 0.833, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 3.273846153846154, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 1.8153846153846154e-06, | |
| "loss": 0.8268, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 3.276923076923077, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 1.8076923076923078e-06, | |
| "loss": 0.8645, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 3.2800000000000002, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 1.8000000000000001e-06, | |
| "loss": 0.8502, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 3.2830769230769232, | |
| "grad_norm": 0.12109375, | |
| "learning_rate": 1.7923076923076923e-06, | |
| "loss": 0.841, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 3.286153846153846, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 1.7846153846153846e-06, | |
| "loss": 0.8629, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 3.289230769230769, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 1.7769230769230772e-06, | |
| "loss": 0.8157, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 3.292307692307692, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 1.7692307692307695e-06, | |
| "loss": 0.7898, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 3.2953846153846156, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 1.7615384615384617e-06, | |
| "loss": 0.8861, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 3.2984615384615386, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 1.753846153846154e-06, | |
| "loss": 0.8543, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 3.3015384615384615, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 1.7461538461538464e-06, | |
| "loss": 0.8631, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 3.3046153846153845, | |
| "grad_norm": 0.0849609375, | |
| "learning_rate": 1.7384615384615385e-06, | |
| "loss": 0.7981, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 3.3076923076923075, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 1.7307692307692308e-06, | |
| "loss": 0.8093, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 3.310769230769231, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 1.7230769230769232e-06, | |
| "loss": 0.8114, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 3.313846153846154, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 1.7153846153846153e-06, | |
| "loss": 0.8292, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 3.316923076923077, | |
| "grad_norm": 0.087890625, | |
| "learning_rate": 1.7076923076923077e-06, | |
| "loss": 0.7871, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 1.7000000000000002e-06, | |
| "loss": 0.8283, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 3.3230769230769233, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 1.6923076923076926e-06, | |
| "loss": 0.8013, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 3.3261538461538462, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 1.6846153846153847e-06, | |
| "loss": 0.8398, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 3.329230769230769, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 1.676923076923077e-06, | |
| "loss": 0.8165, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 3.332307692307692, | |
| "grad_norm": 0.15234375, | |
| "learning_rate": 1.6692307692307694e-06, | |
| "loss": 0.8179, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 3.3353846153846156, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 1.6615384615384616e-06, | |
| "loss": 0.8267, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 3.3384615384615386, | |
| "grad_norm": 0.10888671875, | |
| "learning_rate": 1.653846153846154e-06, | |
| "loss": 0.8222, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 3.3415384615384616, | |
| "grad_norm": 0.11767578125, | |
| "learning_rate": 1.6461538461538463e-06, | |
| "loss": 0.844, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 3.3446153846153845, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 1.6384615384615384e-06, | |
| "loss": 0.8368, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 3.3476923076923075, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 1.6307692307692307e-06, | |
| "loss": 0.8426, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 3.350769230769231, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 1.6230769230769233e-06, | |
| "loss": 0.836, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 3.353846153846154, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 1.6153846153846157e-06, | |
| "loss": 0.8111, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 3.356923076923077, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 1.6076923076923078e-06, | |
| "loss": 0.8183, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 0.7812, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 3.363076923076923, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 1.5923076923076925e-06, | |
| "loss": 0.8596, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 3.3661538461538463, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 1.5846153846153846e-06, | |
| "loss": 0.8522, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 3.3692307692307693, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 1.576923076923077e-06, | |
| "loss": 0.8545, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 3.3723076923076922, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 1.5692307692307693e-06, | |
| "loss": 0.8151, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 3.375384615384615, | |
| "grad_norm": 0.09130859375, | |
| "learning_rate": 1.5615384615384615e-06, | |
| "loss": 0.7809, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 3.3784615384615386, | |
| "grad_norm": 0.11474609375, | |
| "learning_rate": 1.5538461538461538e-06, | |
| "loss": 0.8873, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 3.3815384615384616, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 1.5461538461538464e-06, | |
| "loss": 0.8103, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 3.3846153846153846, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 1.5384615384615387e-06, | |
| "loss": 0.8441, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 3.3876923076923076, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 1.5307692307692309e-06, | |
| "loss": 0.8216, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 3.390769230769231, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 1.5230769230769232e-06, | |
| "loss": 0.8705, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 3.393846153846154, | |
| "grad_norm": 0.1064453125, | |
| "learning_rate": 1.5153846153846156e-06, | |
| "loss": 0.868, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 3.396923076923077, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 1.5076923076923077e-06, | |
| "loss": 0.8434, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.7793, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 3.403076923076923, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 1.4923076923076924e-06, | |
| "loss": 0.8467, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 3.4061538461538463, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 1.4846153846153845e-06, | |
| "loss": 0.8539, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 3.4092307692307693, | |
| "grad_norm": 0.1279296875, | |
| "learning_rate": 1.476923076923077e-06, | |
| "loss": 0.8569, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 3.4123076923076923, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 1.4692307692307694e-06, | |
| "loss": 0.8191, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 3.4153846153846152, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 1.4615384615384618e-06, | |
| "loss": 0.8282, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 3.418461538461538, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 1.453846153846154e-06, | |
| "loss": 0.8447, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 3.4215384615384616, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 1.4461538461538463e-06, | |
| "loss": 0.8446, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 3.4246153846153846, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 1.4384615384615386e-06, | |
| "loss": 0.8197, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 3.4276923076923076, | |
| "grad_norm": 0.119140625, | |
| "learning_rate": 1.4307692307692308e-06, | |
| "loss": 0.8327, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 3.430769230769231, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 1.423076923076923e-06, | |
| "loss": 0.8227, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 3.433846153846154, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 1.4153846153846155e-06, | |
| "loss": 0.8294, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 3.436923076923077, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 1.4076923076923076e-06, | |
| "loss": 0.8556, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "grad_norm": 0.09228515625, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.8163, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 3.443076923076923, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 1.3923076923076925e-06, | |
| "loss": 0.863, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 3.4461538461538463, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 1.3846153846153848e-06, | |
| "loss": 0.8732, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 3.4492307692307693, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 1.376923076923077e-06, | |
| "loss": 0.868, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 3.4523076923076923, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 1.3692307692307693e-06, | |
| "loss": 0.8602, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 3.4553846153846153, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 1.3615384615384617e-06, | |
| "loss": 0.8319, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 3.4584615384615383, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 1.3538461538461538e-06, | |
| "loss": 0.8406, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 3.4615384615384617, | |
| "grad_norm": 0.146484375, | |
| "learning_rate": 1.3461538461538462e-06, | |
| "loss": 0.8992, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 3.4646153846153847, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 1.3384615384615385e-06, | |
| "loss": 0.8461, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 3.4676923076923076, | |
| "grad_norm": 0.10400390625, | |
| "learning_rate": 1.3307692307692307e-06, | |
| "loss": 0.8658, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 3.4707692307692306, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 1.3230769230769232e-06, | |
| "loss": 0.8373, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 3.473846153846154, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 1.3153846153846156e-06, | |
| "loss": 0.833, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 3.476923076923077, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 1.307692307692308e-06, | |
| "loss": 0.8334, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 1.3e-06, | |
| "loss": 0.8678, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 3.483076923076923, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 1.2923076923076924e-06, | |
| "loss": 0.8138, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 3.4861538461538464, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 1.2846153846153847e-06, | |
| "loss": 0.8388, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 3.4892307692307694, | |
| "grad_norm": 0.103515625, | |
| "learning_rate": 1.2769230769230769e-06, | |
| "loss": 0.8652, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 3.4923076923076923, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 1.2692307692307692e-06, | |
| "loss": 0.8244, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 3.4953846153846153, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 1.2615384615384616e-06, | |
| "loss": 0.8061, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 3.4984615384615383, | |
| "grad_norm": 0.12109375, | |
| "learning_rate": 1.253846153846154e-06, | |
| "loss": 0.8481, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 3.5015384615384617, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 1.2461538461538463e-06, | |
| "loss": 0.8455, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 3.5046153846153847, | |
| "grad_norm": 0.0908203125, | |
| "learning_rate": 1.2384615384615386e-06, | |
| "loss": 0.8128, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 3.5076923076923077, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 1.230769230769231e-06, | |
| "loss": 0.8121, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 3.5107692307692306, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 1.2230769230769233e-06, | |
| "loss": 0.8232, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 3.5138461538461536, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 1.2153846153846155e-06, | |
| "loss": 0.8167, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 3.516923076923077, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 1.2076923076923078e-06, | |
| "loss": 0.8449, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 0.853, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 3.523076923076923, | |
| "grad_norm": 0.0986328125, | |
| "learning_rate": 1.1923076923076925e-06, | |
| "loss": 0.8073, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 3.5261538461538464, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 1.1846153846153849e-06, | |
| "loss": 0.845, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 3.5292307692307694, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 1.176923076923077e-06, | |
| "loss": 0.8308, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 3.5323076923076924, | |
| "grad_norm": 0.130859375, | |
| "learning_rate": 1.1692307692307693e-06, | |
| "loss": 0.8282, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 3.5353846153846153, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 1.1615384615384617e-06, | |
| "loss": 0.8193, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 3.5384615384615383, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 1.153846153846154e-06, | |
| "loss": 0.8435, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 3.5415384615384617, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 1.1461538461538464e-06, | |
| "loss": 0.8574, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 3.5446153846153847, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 1.1384615384615385e-06, | |
| "loss": 0.8292, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 3.5476923076923077, | |
| "grad_norm": 0.08642578125, | |
| "learning_rate": 1.1307692307692309e-06, | |
| "loss": 0.7969, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 3.5507692307692307, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 1.1230769230769232e-06, | |
| "loss": 0.8751, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 3.5538461538461537, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 1.1153846153846156e-06, | |
| "loss": 0.8251, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 3.556923076923077, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 1.107692307692308e-06, | |
| "loss": 0.8217, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 1.1e-06, | |
| "loss": 0.8384, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 3.563076923076923, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 1.0923076923076924e-06, | |
| "loss": 0.8301, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 3.566153846153846, | |
| "grad_norm": 0.11279296875, | |
| "learning_rate": 1.0846153846153848e-06, | |
| "loss": 0.8404, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 3.569230769230769, | |
| "grad_norm": 0.09033203125, | |
| "learning_rate": 1.076923076923077e-06, | |
| "loss": 0.795, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 3.5723076923076924, | |
| "grad_norm": 0.08544921875, | |
| "learning_rate": 1.0692307692307695e-06, | |
| "loss": 0.8084, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 3.5753846153846154, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 1.0615384615384616e-06, | |
| "loss": 0.838, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 3.5784615384615384, | |
| "grad_norm": 0.1484375, | |
| "learning_rate": 1.053846153846154e-06, | |
| "loss": 0.8278, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 3.581538461538462, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 1.0461538461538463e-06, | |
| "loss": 0.8401, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 3.5846153846153848, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 1.0384615384615386e-06, | |
| "loss": 0.8157, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 3.5876923076923077, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 1.030769230769231e-06, | |
| "loss": 0.8324, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 3.5907692307692307, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 1.0230769230769231e-06, | |
| "loss": 0.7968, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 3.5938461538461537, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 1.0153846153846155e-06, | |
| "loss": 0.8276, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 3.596923076923077, | |
| "grad_norm": 0.1142578125, | |
| "learning_rate": 1.0076923076923078e-06, | |
| "loss": 0.8435, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.7808, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 3.603076923076923, | |
| "grad_norm": 0.12353515625, | |
| "learning_rate": 9.923076923076925e-07, | |
| "loss": 0.8327, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 3.606153846153846, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 9.846153846153847e-07, | |
| "loss": 0.8335, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 3.609230769230769, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 9.76923076923077e-07, | |
| "loss": 0.8104, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 3.6123076923076924, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 9.692307692307693e-07, | |
| "loss": 0.7962, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 3.6153846153846154, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 9.615384615384617e-07, | |
| "loss": 0.8524, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 3.6184615384615384, | |
| "grad_norm": 0.12060546875, | |
| "learning_rate": 9.53846153846154e-07, | |
| "loss": 0.8479, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 3.6215384615384614, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 9.461538461538462e-07, | |
| "loss": 0.8133, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 3.6246153846153844, | |
| "grad_norm": 0.125, | |
| "learning_rate": 9.384615384615385e-07, | |
| "loss": 0.8317, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 3.6276923076923078, | |
| "grad_norm": 0.091796875, | |
| "learning_rate": 9.307692307692308e-07, | |
| "loss": 0.8014, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 3.6307692307692307, | |
| "grad_norm": 0.10791015625, | |
| "learning_rate": 9.230769230769232e-07, | |
| "loss": 0.8556, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 3.6338461538461537, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 9.153846153846155e-07, | |
| "loss": 0.8507, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 3.636923076923077, | |
| "grad_norm": 0.1328125, | |
| "learning_rate": 9.076923076923077e-07, | |
| "loss": 0.8051, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "grad_norm": 0.1162109375, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.8373, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 3.643076923076923, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 8.923076923076923e-07, | |
| "loss": 0.8564, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 3.646153846153846, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 8.846153846153848e-07, | |
| "loss": 0.8807, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 3.649230769230769, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 8.76923076923077e-07, | |
| "loss": 0.7891, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 3.6523076923076925, | |
| "grad_norm": 0.134765625, | |
| "learning_rate": 8.692307692307692e-07, | |
| "loss": 0.8818, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 3.6553846153846155, | |
| "grad_norm": 0.09326171875, | |
| "learning_rate": 8.615384615384616e-07, | |
| "loss": 0.8475, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 3.6584615384615384, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 8.538461538461538e-07, | |
| "loss": 0.8738, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 3.6615384615384614, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 8.461538461538463e-07, | |
| "loss": 0.8134, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 3.6646153846153844, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 8.384615384615385e-07, | |
| "loss": 0.827, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 3.667692307692308, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 8.307692307692308e-07, | |
| "loss": 0.8348, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 3.670769230769231, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 8.230769230769231e-07, | |
| "loss": 0.8127, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 3.6738461538461538, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 8.153846153846154e-07, | |
| "loss": 0.8176, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 3.676923076923077, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 8.076923076923078e-07, | |
| "loss": 0.8511, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 0.8424, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 3.683076923076923, | |
| "grad_norm": 0.140625, | |
| "learning_rate": 7.923076923076923e-07, | |
| "loss": 0.8447, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 3.686153846153846, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 7.846153846153847e-07, | |
| "loss": 0.7968, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 3.689230769230769, | |
| "grad_norm": 0.1357421875, | |
| "learning_rate": 7.769230769230769e-07, | |
| "loss": 0.8404, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 3.6923076923076925, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 7.692307692307694e-07, | |
| "loss": 0.8546, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.6953846153846155, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 7.615384615384616e-07, | |
| "loss": 0.8703, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 3.6984615384615385, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 7.538461538461538e-07, | |
| "loss": 0.8375, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 3.7015384615384614, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 7.461538461538462e-07, | |
| "loss": 0.8488, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 3.7046153846153844, | |
| "grad_norm": 0.11669921875, | |
| "learning_rate": 7.384615384615385e-07, | |
| "loss": 0.8808, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 3.707692307692308, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 7.307692307692309e-07, | |
| "loss": 0.8681, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 3.710769230769231, | |
| "grad_norm": 0.1005859375, | |
| "learning_rate": 7.230769230769231e-07, | |
| "loss": 0.8673, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 3.713846153846154, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 7.153846153846154e-07, | |
| "loss": 0.8337, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 3.7169230769230768, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 7.076923076923077e-07, | |
| "loss": 0.8621, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 3.7199999999999998, | |
| "grad_norm": 0.10986328125, | |
| "learning_rate": 7.000000000000001e-07, | |
| "loss": 0.8694, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 3.723076923076923, | |
| "grad_norm": 0.111328125, | |
| "learning_rate": 6.923076923076924e-07, | |
| "loss": 0.8435, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 3.726153846153846, | |
| "grad_norm": 0.0888671875, | |
| "learning_rate": 6.846153846153847e-07, | |
| "loss": 0.8212, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 3.729230769230769, | |
| "grad_norm": 0.08837890625, | |
| "learning_rate": 6.769230769230769e-07, | |
| "loss": 0.7855, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 3.7323076923076925, | |
| "grad_norm": 0.158203125, | |
| "learning_rate": 6.692307692307693e-07, | |
| "loss": 0.828, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 3.7353846153846155, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 6.615384615384616e-07, | |
| "loss": 0.8708, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 3.7384615384615385, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 6.53846153846154e-07, | |
| "loss": 0.8409, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 3.7415384615384615, | |
| "grad_norm": 0.10302734375, | |
| "learning_rate": 6.461538461538462e-07, | |
| "loss": 0.8498, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 3.7446153846153845, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 6.384615384615384e-07, | |
| "loss": 0.8559, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 3.747692307692308, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 6.307692307692308e-07, | |
| "loss": 0.8116, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 3.750769230769231, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 6.230769230769231e-07, | |
| "loss": 0.8675, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 3.753846153846154, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 6.153846153846155e-07, | |
| "loss": 0.8459, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 3.756923076923077, | |
| "grad_norm": 0.08642578125, | |
| "learning_rate": 6.076923076923077e-07, | |
| "loss": 0.7934, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.08251953125, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 0.8066, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 3.763076923076923, | |
| "grad_norm": 0.0869140625, | |
| "learning_rate": 5.923076923076924e-07, | |
| "loss": 0.7788, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 3.766153846153846, | |
| "grad_norm": 0.1318359375, | |
| "learning_rate": 5.846153846153847e-07, | |
| "loss": 0.7919, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 3.769230769230769, | |
| "grad_norm": 0.08544921875, | |
| "learning_rate": 5.76923076923077e-07, | |
| "loss": 0.7924, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 3.7723076923076926, | |
| "grad_norm": 0.140625, | |
| "learning_rate": 5.692307692307693e-07, | |
| "loss": 0.861, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 3.775384615384615, | |
| "grad_norm": 0.08935546875, | |
| "learning_rate": 5.615384615384616e-07, | |
| "loss": 0.8055, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 3.7784615384615385, | |
| "grad_norm": 0.11865234375, | |
| "learning_rate": 5.53846153846154e-07, | |
| "loss": 0.8312, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 3.7815384615384615, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 5.461538461538462e-07, | |
| "loss": 0.7921, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 3.7846153846153845, | |
| "grad_norm": 0.17578125, | |
| "learning_rate": 5.384615384615386e-07, | |
| "loss": 0.8188, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 3.787692307692308, | |
| "grad_norm": 0.12451171875, | |
| "learning_rate": 5.307692307692308e-07, | |
| "loss": 0.9232, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 3.790769230769231, | |
| "grad_norm": 0.09814453125, | |
| "learning_rate": 5.230769230769231e-07, | |
| "loss": 0.8224, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 3.793846153846154, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 5.153846153846155e-07, | |
| "loss": 0.9331, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 3.796923076923077, | |
| "grad_norm": 0.1513671875, | |
| "learning_rate": 5.076923076923077e-07, | |
| "loss": 0.8822, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "grad_norm": 0.0947265625, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.8223, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 3.8030769230769232, | |
| "grad_norm": 0.1259765625, | |
| "learning_rate": 4.923076923076923e-07, | |
| "loss": 0.8982, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 3.806153846153846, | |
| "grad_norm": 0.10205078125, | |
| "learning_rate": 4.846153846153847e-07, | |
| "loss": 0.8546, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 3.809230769230769, | |
| "grad_norm": 0.10595703125, | |
| "learning_rate": 4.76923076923077e-07, | |
| "loss": 0.8254, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 3.812307692307692, | |
| "grad_norm": 0.08984375, | |
| "learning_rate": 4.6923076923076927e-07, | |
| "loss": 0.773, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 3.815384615384615, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 4.615384615384616e-07, | |
| "loss": 0.8391, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 3.8184615384615386, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 4.5384615384615386e-07, | |
| "loss": 0.8583, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 3.8215384615384616, | |
| "grad_norm": 0.1025390625, | |
| "learning_rate": 4.4615384615384615e-07, | |
| "loss": 0.8279, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 3.8246153846153845, | |
| "grad_norm": 0.1103515625, | |
| "learning_rate": 4.384615384615385e-07, | |
| "loss": 0.8759, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 3.827692307692308, | |
| "grad_norm": 0.09521484375, | |
| "learning_rate": 4.307692307692308e-07, | |
| "loss": 0.7975, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 3.830769230769231, | |
| "grad_norm": 0.0927734375, | |
| "learning_rate": 4.2307692307692315e-07, | |
| "loss": 0.8052, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 3.833846153846154, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 4.153846153846154e-07, | |
| "loss": 0.8466, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 3.836923076923077, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 4.076923076923077e-07, | |
| "loss": 0.9025, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.8181, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 3.8430769230769233, | |
| "grad_norm": 0.107421875, | |
| "learning_rate": 3.9230769230769233e-07, | |
| "loss": 0.8039, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 3.8461538461538463, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 3.846153846153847e-07, | |
| "loss": 0.8404, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.8492307692307692, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 3.769230769230769e-07, | |
| "loss": 0.7762, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 3.852307692307692, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 3.6923076923076927e-07, | |
| "loss": 0.863, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 3.855384615384615, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 3.6153846153846157e-07, | |
| "loss": 0.8174, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 3.8584615384615386, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 3.5384615384615386e-07, | |
| "loss": 0.8448, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 3.8615384615384616, | |
| "grad_norm": 0.099609375, | |
| "learning_rate": 3.461538461538462e-07, | |
| "loss": 0.7968, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 3.8646153846153846, | |
| "grad_norm": 0.083984375, | |
| "learning_rate": 3.3846153846153845e-07, | |
| "loss": 0.7687, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 3.8676923076923075, | |
| "grad_norm": 0.0869140625, | |
| "learning_rate": 3.307692307692308e-07, | |
| "loss": 0.792, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 3.8707692307692305, | |
| "grad_norm": 0.09619140625, | |
| "learning_rate": 3.230769230769231e-07, | |
| "loss": 0.8406, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 3.873846153846154, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 3.153846153846154e-07, | |
| "loss": 0.7823, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 3.876923076923077, | |
| "grad_norm": 0.09375, | |
| "learning_rate": 3.0769230769230774e-07, | |
| "loss": 0.8473, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "grad_norm": 0.1201171875, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 0.8691, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 3.8830769230769233, | |
| "grad_norm": 0.1474609375, | |
| "learning_rate": 2.9230769230769234e-07, | |
| "loss": 0.95, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 3.8861538461538463, | |
| "grad_norm": 0.1083984375, | |
| "learning_rate": 2.8461538461538463e-07, | |
| "loss": 0.8264, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 3.8892307692307693, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 2.76923076923077e-07, | |
| "loss": 0.8361, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 3.8923076923076922, | |
| "grad_norm": 0.115234375, | |
| "learning_rate": 2.692307692307693e-07, | |
| "loss": 0.8266, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 3.8953846153846152, | |
| "grad_norm": 0.09423828125, | |
| "learning_rate": 2.6153846153846157e-07, | |
| "loss": 0.7736, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 3.8984615384615386, | |
| "grad_norm": 0.1015625, | |
| "learning_rate": 2.5384615384615387e-07, | |
| "loss": 0.8011, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 3.9015384615384616, | |
| "grad_norm": 0.10498046875, | |
| "learning_rate": 2.4615384615384616e-07, | |
| "loss": 0.8297, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 3.9046153846153846, | |
| "grad_norm": 0.1796875, | |
| "learning_rate": 2.384615384615385e-07, | |
| "loss": 0.8527, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 3.9076923076923076, | |
| "grad_norm": 0.095703125, | |
| "learning_rate": 2.307692307692308e-07, | |
| "loss": 0.803, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 3.9107692307692306, | |
| "grad_norm": 0.1123046875, | |
| "learning_rate": 2.2307692307692308e-07, | |
| "loss": 0.8298, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 3.913846153846154, | |
| "grad_norm": 0.09765625, | |
| "learning_rate": 2.153846153846154e-07, | |
| "loss": 0.8245, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 3.916923076923077, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 2.076923076923077e-07, | |
| "loss": 0.8274, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 2.0000000000000002e-07, | |
| "loss": 0.7781, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 3.9230769230769234, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 1.9230769230769234e-07, | |
| "loss": 0.8573, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 3.926153846153846, | |
| "grad_norm": 0.11181640625, | |
| "learning_rate": 1.8461538461538464e-07, | |
| "loss": 0.8652, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 3.9292307692307693, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 1.7692307692307693e-07, | |
| "loss": 0.9406, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 3.9323076923076923, | |
| "grad_norm": 0.0859375, | |
| "learning_rate": 1.6923076923076923e-07, | |
| "loss": 0.8042, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 3.9353846153846153, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 1.6153846153846155e-07, | |
| "loss": 0.8742, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 3.9384615384615387, | |
| "grad_norm": 0.10546875, | |
| "learning_rate": 1.5384615384615387e-07, | |
| "loss": 0.8719, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 3.9415384615384617, | |
| "grad_norm": 0.1181640625, | |
| "learning_rate": 1.4615384615384617e-07, | |
| "loss": 0.8524, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 3.9446153846153846, | |
| "grad_norm": 0.10107421875, | |
| "learning_rate": 1.384615384615385e-07, | |
| "loss": 0.8243, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 3.9476923076923076, | |
| "grad_norm": 0.154296875, | |
| "learning_rate": 1.3076923076923079e-07, | |
| "loss": 0.8309, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 3.9507692307692306, | |
| "grad_norm": 0.10009765625, | |
| "learning_rate": 1.2307692307692308e-07, | |
| "loss": 0.8549, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 3.953846153846154, | |
| "grad_norm": 0.09716796875, | |
| "learning_rate": 1.153846153846154e-07, | |
| "loss": 0.8192, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 3.956923076923077, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 1.076923076923077e-07, | |
| "loss": 0.8136, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "grad_norm": 0.0966796875, | |
| "learning_rate": 1.0000000000000001e-07, | |
| "loss": 0.8045, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 3.963076923076923, | |
| "grad_norm": 0.1455078125, | |
| "learning_rate": 9.230769230769232e-08, | |
| "loss": 0.9089, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 3.966153846153846, | |
| "grad_norm": 0.1044921875, | |
| "learning_rate": 8.461538461538461e-08, | |
| "loss": 0.8479, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 3.9692307692307693, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 7.692307692307694e-08, | |
| "loss": 0.8336, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 3.9723076923076923, | |
| "grad_norm": 0.109375, | |
| "learning_rate": 6.923076923076924e-08, | |
| "loss": 0.8912, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 3.9753846153846153, | |
| "grad_norm": 0.12109375, | |
| "learning_rate": 6.153846153846154e-08, | |
| "loss": 0.8512, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 3.9784615384615387, | |
| "grad_norm": 0.138671875, | |
| "learning_rate": 5.384615384615385e-08, | |
| "loss": 0.8587, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 3.9815384615384617, | |
| "grad_norm": 0.11083984375, | |
| "learning_rate": 4.615384615384616e-08, | |
| "loss": 0.8593, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 3.9846153846153847, | |
| "grad_norm": 0.134765625, | |
| "learning_rate": 3.846153846153847e-08, | |
| "loss": 0.9444, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 3.9876923076923076, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 3.076923076923077e-08, | |
| "loss": 0.8725, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 3.9907692307692306, | |
| "grad_norm": 0.1240234375, | |
| "learning_rate": 2.307692307692308e-08, | |
| "loss": 0.8362, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 3.993846153846154, | |
| "grad_norm": 0.10693359375, | |
| "learning_rate": 1.5384615384615385e-08, | |
| "loss": 0.8544, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 3.996923076923077, | |
| "grad_norm": 0.11376953125, | |
| "learning_rate": 7.692307692307693e-09, | |
| "loss": 0.8339, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.09912109375, | |
| "learning_rate": 0.0, | |
| "loss": 0.8022, | |
| "step": 1300 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 1300, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "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": 4.0797918831386296e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |