| { | |
| "best_metric": 0.40209120512008667, | |
| "best_model_checkpoint": "./disaster-tweet-distilbert-classification/checkpoint-378", | |
| "epoch": 1.0, | |
| "global_step": 378, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.7503, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.7789, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3e-06, | |
| "loss": 0.7344, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.7709, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5e-06, | |
| "loss": 0.6884, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6e-06, | |
| "loss": 0.7087, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 0.6655, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.6978, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9e-06, | |
| "loss": 0.7435, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1e-05, | |
| "loss": 0.719, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 0.7129, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.7249, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 0.7436, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 0.6886, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.5e-05, | |
| "loss": 0.702, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.7105, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 0.6709, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.6767, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9e-05, | |
| "loss": 0.6784, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7016, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.1e-05, | |
| "loss": 0.6308, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 0.645, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 0.6845, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.4e-05, | |
| "loss": 0.6891, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.5914, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 0.6916, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 0.5884, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 0.6587, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.9e-05, | |
| "loss": 0.6008, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3e-05, | |
| "loss": 0.6717, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.1e-05, | |
| "loss": 0.5795, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.6358, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.3e-05, | |
| "loss": 0.7508, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 0.6601, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.5e-05, | |
| "loss": 0.6573, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.5695, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.7e-05, | |
| "loss": 0.5535, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.5813, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 0.5224, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4e-05, | |
| "loss": 0.4757, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.1e-05, | |
| "loss": 0.5529, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.4964, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.4565, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.5848, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.7333, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.5224, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7e-05, | |
| "loss": 0.4826, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.4328, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9e-05, | |
| "loss": 0.2546, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5e-05, | |
| "loss": 0.2636, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.998659517426274e-05, | |
| "loss": 0.4116, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.997319034852547e-05, | |
| "loss": 0.3191, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.995978552278821e-05, | |
| "loss": 0.2822, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.994638069705094e-05, | |
| "loss": 0.8821, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.993297587131368e-05, | |
| "loss": 0.483, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9919571045576406e-05, | |
| "loss": 0.7164, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.990616621983915e-05, | |
| "loss": 0.4161, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.989276139410188e-05, | |
| "loss": 0.2668, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.987935656836462e-05, | |
| "loss": 0.5255, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.986595174262735e-05, | |
| "loss": 0.3784, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.985254691689008e-05, | |
| "loss": 0.5065, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.983914209115282e-05, | |
| "loss": 0.1988, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.982573726541555e-05, | |
| "loss": 0.4362, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.981233243967829e-05, | |
| "loss": 0.6619, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.979892761394102e-05, | |
| "loss": 0.3217, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.978552278820375e-05, | |
| "loss": 0.2967, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.977211796246649e-05, | |
| "loss": 0.2429, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.975871313672922e-05, | |
| "loss": 0.6642, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.974530831099196e-05, | |
| "loss": 0.56, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.973190348525469e-05, | |
| "loss": 1.2979, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.9718498659517427e-05, | |
| "loss": 0.5287, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.970509383378016e-05, | |
| "loss": 0.4684, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.96916890080429e-05, | |
| "loss": 0.283, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.967828418230563e-05, | |
| "loss": 0.6818, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.966487935656837e-05, | |
| "loss": 0.6141, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.96514745308311e-05, | |
| "loss": 0.5046, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.963806970509384e-05, | |
| "loss": 0.5266, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.962466487935657e-05, | |
| "loss": 0.5944, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.961126005361931e-05, | |
| "loss": 0.5631, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.959785522788204e-05, | |
| "loss": 0.4791, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.958445040214477e-05, | |
| "loss": 0.5645, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.957104557640751e-05, | |
| "loss": 0.4349, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.955764075067024e-05, | |
| "loss": 0.3865, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.954423592493298e-05, | |
| "loss": 0.486, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.953083109919571e-05, | |
| "loss": 0.2179, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.951742627345845e-05, | |
| "loss": 0.3896, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.950402144772118e-05, | |
| "loss": 0.4247, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.949061662198392e-05, | |
| "loss": 0.4906, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.947721179624665e-05, | |
| "loss": 0.4483, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.946380697050939e-05, | |
| "loss": 0.557, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9450402144772116e-05, | |
| "loss": 0.7521, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.943699731903486e-05, | |
| "loss": 0.3103, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.9423592493297586e-05, | |
| "loss": 0.757, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.941018766756033e-05, | |
| "loss": 0.8248, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.9396782841823056e-05, | |
| "loss": 0.4591, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.938337801608579e-05, | |
| "loss": 0.3912, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.9369973190348526e-05, | |
| "loss": 0.5289, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.935656836461126e-05, | |
| "loss": 0.3264, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.9343163538874e-05, | |
| "loss": 0.2947, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.932975871313673e-05, | |
| "loss": 0.2647, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.931635388739946e-05, | |
| "loss": 0.3691, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.93029490616622e-05, | |
| "loss": 0.4796, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.928954423592493e-05, | |
| "loss": 0.4827, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.927613941018767e-05, | |
| "loss": 0.2672, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.92627345844504e-05, | |
| "loss": 0.7456, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.9249329758713136e-05, | |
| "loss": 0.5206, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.923592493297587e-05, | |
| "loss": 0.3576, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9222520107238606e-05, | |
| "loss": 0.2596, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.920911528150134e-05, | |
| "loss": 0.4115, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9195710455764076e-05, | |
| "loss": 0.3481, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.918230563002681e-05, | |
| "loss": 0.4387, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.916890080428955e-05, | |
| "loss": 0.5023, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.915549597855228e-05, | |
| "loss": 0.5916, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.914209115281502e-05, | |
| "loss": 0.5467, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.912868632707775e-05, | |
| "loss": 0.5631, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.911528150134049e-05, | |
| "loss": 0.5512, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.910187667560322e-05, | |
| "loss": 0.5546, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.908847184986595e-05, | |
| "loss": 0.4209, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.907506702412869e-05, | |
| "loss": 0.6064, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.906166219839142e-05, | |
| "loss": 0.5301, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.904825737265416e-05, | |
| "loss": 0.436, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.903485254691689e-05, | |
| "loss": 0.726, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.9021447721179626e-05, | |
| "loss": 0.5288, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.900804289544236e-05, | |
| "loss": 0.4887, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.8994638069705097e-05, | |
| "loss": 0.3752, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.898123324396783e-05, | |
| "loss": 0.4895, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.896782841823057e-05, | |
| "loss": 0.5046, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.8954423592493295e-05, | |
| "loss": 0.3953, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.894101876675604e-05, | |
| "loss": 0.2015, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.8927613941018765e-05, | |
| "loss": 0.5165, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.891420911528151e-05, | |
| "loss": 0.4237, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.8900804289544236e-05, | |
| "loss": 0.239, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.888739946380697e-05, | |
| "loss": 0.5515, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.8873994638069706e-05, | |
| "loss": 0.303, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.886058981233244e-05, | |
| "loss": 0.2867, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.8847184986595176e-05, | |
| "loss": 0.6756, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.883378016085791e-05, | |
| "loss": 0.4996, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.8820375335120646e-05, | |
| "loss": 0.2798, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.880697050938338e-05, | |
| "loss": 0.8877, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.879356568364612e-05, | |
| "loss": 0.5022, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.878016085790885e-05, | |
| "loss": 0.2177, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.876675603217159e-05, | |
| "loss": 0.8708, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.8753351206434315e-05, | |
| "loss": 0.719, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.873994638069706e-05, | |
| "loss": 0.562, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.8726541554959786e-05, | |
| "loss": 0.492, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.871313672922253e-05, | |
| "loss": 0.4637, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.8699731903485256e-05, | |
| "loss": 0.4132, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.868632707774799e-05, | |
| "loss": 0.2889, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.8672922252010726e-05, | |
| "loss": 0.2213, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.865951742627346e-05, | |
| "loss": 0.9268, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.8646112600536196e-05, | |
| "loss": 0.2852, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.863270777479893e-05, | |
| "loss": 0.4599, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.861930294906166e-05, | |
| "loss": 0.1913, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.86058981233244e-05, | |
| "loss": 0.4488, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.859249329758713e-05, | |
| "loss": 0.9022, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.857908847184987e-05, | |
| "loss": 0.5221, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.85656836461126e-05, | |
| "loss": 0.2394, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.8552278820375336e-05, | |
| "loss": 0.3332, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.853887399463807e-05, | |
| "loss": 0.4015, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.8525469168900806e-05, | |
| "loss": 0.4461, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.851206434316354e-05, | |
| "loss": 0.337, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.8498659517426276e-05, | |
| "loss": 0.4908, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.848525469168901e-05, | |
| "loss": 0.526, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.8471849865951746e-05, | |
| "loss": 0.5262, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.845844504021448e-05, | |
| "loss": 0.6818, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.8445040214477217e-05, | |
| "loss": 0.3154, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.843163538873995e-05, | |
| "loss": 0.5963, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.841823056300268e-05, | |
| "loss": 0.4451, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.840482573726542e-05, | |
| "loss": 0.5969, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.839142091152815e-05, | |
| "loss": 0.438, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.837801608579089e-05, | |
| "loss": 0.4827, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.836461126005362e-05, | |
| "loss": 0.2029, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.8351206434316356e-05, | |
| "loss": 0.5195, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.833780160857909e-05, | |
| "loss": 0.517, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.8324396782841826e-05, | |
| "loss": 0.5532, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.831099195710456e-05, | |
| "loss": 0.4198, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.8297587131367296e-05, | |
| "loss": 0.8386, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.8284182305630025e-05, | |
| "loss": 0.575, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.8270777479892766e-05, | |
| "loss": 0.6156, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.8257372654155495e-05, | |
| "loss": 0.7044, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.824396782841824e-05, | |
| "loss": 0.5712, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.8230563002680965e-05, | |
| "loss": 0.34, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.82171581769437e-05, | |
| "loss": 0.5773, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.8203753351206435e-05, | |
| "loss": 0.546, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.819034852546917e-05, | |
| "loss": 0.3955, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.8176943699731906e-05, | |
| "loss": 0.5921, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.816353887399464e-05, | |
| "loss": 0.3108, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.8150134048257376e-05, | |
| "loss": 0.5469, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.813672922252011e-05, | |
| "loss": 0.64, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.8123324396782846e-05, | |
| "loss": 0.5153, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.810991957104558e-05, | |
| "loss": 0.4719, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.8096514745308316e-05, | |
| "loss": 0.52, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.8083109919571045e-05, | |
| "loss": 0.5114, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.806970509383379e-05, | |
| "loss": 0.5469, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.8056300268096515e-05, | |
| "loss": 0.3435, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.804289544235926e-05, | |
| "loss": 0.6469, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.8029490616621985e-05, | |
| "loss": 0.6595, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.801608579088472e-05, | |
| "loss": 0.5503, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.8002680965147456e-05, | |
| "loss": 0.3799, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.798927613941019e-05, | |
| "loss": 0.417, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.7975871313672926e-05, | |
| "loss": 0.5281, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.796246648793566e-05, | |
| "loss": 0.3439, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.794906166219839e-05, | |
| "loss": 0.5777, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.793565683646113e-05, | |
| "loss": 0.5286, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.792225201072386e-05, | |
| "loss": 0.4302, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.79088471849866e-05, | |
| "loss": 0.5413, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.789544235924933e-05, | |
| "loss": 0.3087, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.7882037533512065e-05, | |
| "loss": 0.4385, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.78686327077748e-05, | |
| "loss": 0.7137, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.7855227882037535e-05, | |
| "loss": 0.6278, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.784182305630027e-05, | |
| "loss": 0.2832, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.7828418230563005e-05, | |
| "loss": 0.5899, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.7815013404825734e-05, | |
| "loss": 0.4421, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.7801608579088476e-05, | |
| "loss": 0.3411, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.7788203753351204e-05, | |
| "loss": 0.4236, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.7774798927613946e-05, | |
| "loss": 0.4444, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.7761394101876674e-05, | |
| "loss": 0.4765, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.774798927613941e-05, | |
| "loss": 0.2453, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.7734584450402145e-05, | |
| "loss": 0.2765, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.772117962466488e-05, | |
| "loss": 0.2075, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.7707774798927615e-05, | |
| "loss": 0.5905, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.769436997319035e-05, | |
| "loss": 0.3457, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.7680965147453085e-05, | |
| "loss": 0.5986, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.766756032171582e-05, | |
| "loss": 0.3881, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.7654155495978555e-05, | |
| "loss": 0.5655, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.764075067024129e-05, | |
| "loss": 0.6156, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.7627345844504026e-05, | |
| "loss": 0.4784, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.7613941018766754e-05, | |
| "loss": 0.5195, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.7600536193029496e-05, | |
| "loss": 0.66, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.7587131367292224e-05, | |
| "loss": 0.4104, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.7573726541554966e-05, | |
| "loss": 0.3478, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.7560321715817695e-05, | |
| "loss": 0.2037, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.754691689008043e-05, | |
| "loss": 0.75, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.7533512064343165e-05, | |
| "loss": 0.4237, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.75201072386059e-05, | |
| "loss": 0.2372, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.7506702412868635e-05, | |
| "loss": 0.7874, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.749329758713137e-05, | |
| "loss": 0.5751, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.7479892761394105e-05, | |
| "loss": 0.5801, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.746648793565684e-05, | |
| "loss": 0.4983, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.745308310991957e-05, | |
| "loss": 0.4215, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.743967828418231e-05, | |
| "loss": 0.3655, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.742627345844504e-05, | |
| "loss": 0.523, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.741286863270778e-05, | |
| "loss": 0.5952, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.739946380697051e-05, | |
| "loss": 0.4226, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.7386058981233244e-05, | |
| "loss": 0.316, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.737265415549598e-05, | |
| "loss": 0.5096, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.7359249329758715e-05, | |
| "loss": 0.435, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.734584450402145e-05, | |
| "loss": 0.7516, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.7332439678284185e-05, | |
| "loss": 0.531, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.731903485254692e-05, | |
| "loss": 0.4096, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.7305630026809655e-05, | |
| "loss": 0.387, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.729222520107239e-05, | |
| "loss": 0.5468, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.7278820375335125e-05, | |
| "loss": 0.4613, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.726541554959786e-05, | |
| "loss": 0.8437, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.725201072386059e-05, | |
| "loss": 0.522, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.723860589812333e-05, | |
| "loss": 0.3922, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.722520107238606e-05, | |
| "loss": 0.5114, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.72117962466488e-05, | |
| "loss": 0.6148, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.719839142091153e-05, | |
| "loss": 0.4578, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.7184986595174265e-05, | |
| "loss": 0.6286, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.7171581769437e-05, | |
| "loss": 0.5883, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.7158176943699735e-05, | |
| "loss": 0.5634, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.714477211796247e-05, | |
| "loss": 0.4085, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.7131367292225205e-05, | |
| "loss": 0.2988, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.7117962466487934e-05, | |
| "loss": 0.6353, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.7104557640750675e-05, | |
| "loss": 0.4598, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.7091152815013404e-05, | |
| "loss": 0.5072, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.7077747989276146e-05, | |
| "loss": 0.49, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.7064343163538874e-05, | |
| "loss": 0.7225, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.705093833780161e-05, | |
| "loss": 0.5332, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.7037533512064344e-05, | |
| "loss": 0.6064, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.702412868632708e-05, | |
| "loss": 0.3518, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.7010723860589815e-05, | |
| "loss": 0.3673, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.699731903485255e-05, | |
| "loss": 0.4688, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.6983914209115285e-05, | |
| "loss": 0.5389, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.697050938337802e-05, | |
| "loss": 0.374, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.6957104557640755e-05, | |
| "loss": 0.475, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.694369973190349e-05, | |
| "loss": 0.5397, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.6930294906166225e-05, | |
| "loss": 0.3821, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.6916890080428954e-05, | |
| "loss": 0.3372, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.6903485254691696e-05, | |
| "loss": 0.6652, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.6890080428954424e-05, | |
| "loss": 0.2894, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.6876675603217166e-05, | |
| "loss": 0.5639, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.6863270777479894e-05, | |
| "loss": 0.353, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.684986595174263e-05, | |
| "loss": 0.2932, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.6836461126005364e-05, | |
| "loss": 0.467, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.68230563002681e-05, | |
| "loss": 0.4732, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.6809651474530835e-05, | |
| "loss": 0.1808, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.679624664879357e-05, | |
| "loss": 0.6031, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.67828418230563e-05, | |
| "loss": 0.2555, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.676943699731904e-05, | |
| "loss": 0.4041, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.675603217158177e-05, | |
| "loss": 0.7822, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.674262734584451e-05, | |
| "loss": 0.138, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.672922252010724e-05, | |
| "loss": 0.2746, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.6715817694369974e-05, | |
| "loss": 0.5835, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.670241286863271e-05, | |
| "loss": 0.2367, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.6689008042895444e-05, | |
| "loss": 0.3247, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.667560321715818e-05, | |
| "loss": 0.306, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.6662198391420914e-05, | |
| "loss": 0.2825, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.664879356568364e-05, | |
| "loss": 0.6102, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.6635388739946385e-05, | |
| "loss": 0.3613, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.662198391420911e-05, | |
| "loss": 0.6327, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.6608579088471855e-05, | |
| "loss": 0.4059, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.659517426273458e-05, | |
| "loss": 0.4027, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.658176943699732e-05, | |
| "loss": 0.9133, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.6568364611260054e-05, | |
| "loss": 0.1869, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.655495978552279e-05, | |
| "loss": 0.3987, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.6541554959785524e-05, | |
| "loss": 0.6114, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.652815013404826e-05, | |
| "loss": 0.4406, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6514745308310994e-05, | |
| "loss": 0.2954, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.650134048257373e-05, | |
| "loss": 0.5067, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.6487935656836464e-05, | |
| "loss": 0.3985, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.64745308310992e-05, | |
| "loss": 0.3756, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.6461126005361935e-05, | |
| "loss": 0.2618, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.644772117962466e-05, | |
| "loss": 0.4992, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.6434316353887405e-05, | |
| "loss": 0.8224, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.642091152815013e-05, | |
| "loss": 0.3425, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.6407506702412875e-05, | |
| "loss": 0.4062, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.6394101876675603e-05, | |
| "loss": 0.4748, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.638069705093834e-05, | |
| "loss": 0.6857, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.6367292225201074e-05, | |
| "loss": 0.5368, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.635388739946381e-05, | |
| "loss": 0.5571, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.6340482573726544e-05, | |
| "loss": 0.3045, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.632707774798928e-05, | |
| "loss": 0.3189, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.631367292225201e-05, | |
| "loss": 0.3033, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.630026809651475e-05, | |
| "loss": 0.4659, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.628686327077748e-05, | |
| "loss": 0.4058, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.627345844504022e-05, | |
| "loss": 0.6745, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.626005361930295e-05, | |
| "loss": 0.3259, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.624664879356568e-05, | |
| "loss": 0.5126, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.623324396782842e-05, | |
| "loss": 0.2759, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.621983914209115e-05, | |
| "loss": 0.2512, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.620643431635389e-05, | |
| "loss": 0.3046, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.6193029490616624e-05, | |
| "loss": 0.3931, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.617962466487936e-05, | |
| "loss": 0.4838, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.6166219839142094e-05, | |
| "loss": 0.2925, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.615281501340483e-05, | |
| "loss": 0.4481, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.6139410187667564e-05, | |
| "loss": 0.4528, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.61260053619303e-05, | |
| "loss": 0.2934, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.611260053619303e-05, | |
| "loss": 0.609, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.609919571045577e-05, | |
| "loss": 0.8988, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.60857908847185e-05, | |
| "loss": 1.1222, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.607238605898124e-05, | |
| "loss": 0.3265, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.605898123324397e-05, | |
| "loss": 0.4722, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.60455764075067e-05, | |
| "loss": 0.2791, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.603217158176944e-05, | |
| "loss": 0.4183, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.6018766756032174e-05, | |
| "loss": 0.5323, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.600536193029491e-05, | |
| "loss": 0.6108, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.5991957104557644e-05, | |
| "loss": 0.2875, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.597855227882037e-05, | |
| "loss": 0.6642, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.5965147453083114e-05, | |
| "loss": 0.5244, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.595174262734584e-05, | |
| "loss": 0.5562, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.5938337801608584e-05, | |
| "loss": 0.3972, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.592493297587131e-05, | |
| "loss": 0.4532, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.591152815013405e-05, | |
| "loss": 0.3368, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.589812332439678e-05, | |
| "loss": 0.171, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.588471849865952e-05, | |
| "loss": 0.4036, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.587131367292225e-05, | |
| "loss": 0.4305, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.585790884718499e-05, | |
| "loss": 0.2643, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.5844504021447723e-05, | |
| "loss": 0.3782, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.583109919571046e-05, | |
| "loss": 0.1673, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.5817694369973194e-05, | |
| "loss": 0.3969, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.580428954423593e-05, | |
| "loss": 0.3249, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.5790884718498664e-05, | |
| "loss": 0.1656, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.57774798927614e-05, | |
| "loss": 0.4551, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.5764075067024134e-05, | |
| "loss": 0.6075, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.575067024128686e-05, | |
| "loss": 0.4699, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.5737265415549605e-05, | |
| "loss": 0.5752, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.572386058981233e-05, | |
| "loss": 0.3114, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.5710455764075075e-05, | |
| "loss": 0.7407, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.56970509383378e-05, | |
| "loss": 0.6427, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.568364611260054e-05, | |
| "loss": 0.5021, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.5670241286863273e-05, | |
| "loss": 0.4209, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.565683646112601e-05, | |
| "loss": 0.5957, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.5643431635388744e-05, | |
| "loss": 0.495, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.563002680965148e-05, | |
| "loss": 0.3101, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.561662198391421e-05, | |
| "loss": 0.472, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.560321715817695e-05, | |
| "loss": 0.4607, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_f1": 0.7914963205233032, | |
| "eval_loss": 0.40209120512008667, | |
| "eval_runtime": 2.1585, | |
| "eval_samples_per_second": 700.949, | |
| "eval_steps_per_second": 44.012, | |
| "step": 378 | |
| } | |
| ], | |
| "max_steps": 3780, | |
| "num_train_epochs": 10, | |
| "total_flos": 96750876293376.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |