| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 690, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.468208092485549e-07, | |
| "loss": 6.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.936416184971098e-07, | |
| "loss": 6.0195, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.0404624277456647e-06, | |
| "loss": 6.0156, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.3872832369942195e-06, | |
| "loss": 6.0352, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.7341040462427746e-06, | |
| "loss": 6.0273, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.0809248554913294e-06, | |
| "loss": 5.9805, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.4277456647398847e-06, | |
| "loss": 6.0, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.774566473988439e-06, | |
| "loss": 5.9375, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.1213872832369943e-06, | |
| "loss": 5.9023, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.468208092485549e-06, | |
| "loss": 5.9219, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.8150289017341036e-06, | |
| "loss": 5.9531, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.161849710982659e-06, | |
| "loss": 5.9766, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.508670520231214e-06, | |
| "loss": 5.9141, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.855491329479769e-06, | |
| "loss": 5.8828, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.202312138728324e-06, | |
| "loss": 5.9219, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.549132947976878e-06, | |
| "loss": 5.8633, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.895953757225434e-06, | |
| "loss": 5.9062, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 6.242774566473989e-06, | |
| "loss": 5.8672, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.589595375722544e-06, | |
| "loss": 5.7695, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.936416184971098e-06, | |
| "loss": 5.8008, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 7.283236994219653e-06, | |
| "loss": 5.75, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 7.630057803468207e-06, | |
| "loss": 5.6953, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.976878612716762e-06, | |
| "loss": 5.6758, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 8.323699421965318e-06, | |
| "loss": 5.6055, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 8.670520231213873e-06, | |
| "loss": 5.5859, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.017341040462428e-06, | |
| "loss": 5.4531, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.364161849710983e-06, | |
| "loss": 5.5078, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.710982658959539e-06, | |
| "loss": 5.3711, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.0057803468208092e-05, | |
| "loss": 5.1953, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.0404624277456647e-05, | |
| "loss": 5.2578, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.0751445086705203e-05, | |
| "loss": 5.0273, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.1098265895953756e-05, | |
| "loss": 4.9766, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.1445086705202312e-05, | |
| "loss": 4.7852, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.1791907514450869e-05, | |
| "loss": 4.8984, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.2138728323699422e-05, | |
| "loss": 4.625, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.2485549132947977e-05, | |
| "loss": 4.5156, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.2832369942196533e-05, | |
| "loss": 4.5508, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.3179190751445088e-05, | |
| "loss": 4.3281, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.3526011560693641e-05, | |
| "loss": 4.5938, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.3872832369942197e-05, | |
| "loss": 4.1172, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.4219653179190752e-05, | |
| "loss": 4.2188, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.4566473988439305e-05, | |
| "loss": 4.0156, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.491329479768786e-05, | |
| "loss": 4.0234, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.5260115606936414e-05, | |
| "loss": 3.9141, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.560693641618497e-05, | |
| "loss": 3.6562, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.5953757225433525e-05, | |
| "loss": 3.6133, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.630057803468208e-05, | |
| "loss": 3.1582, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.6647398843930635e-05, | |
| "loss": 3.248, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.6994219653179194e-05, | |
| "loss": 3.0996, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.7341040462427746e-05, | |
| "loss": 3.1875, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.76878612716763e-05, | |
| "loss": 2.9941, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8034682080924856e-05, | |
| "loss": 2.9297, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.838150289017341e-05, | |
| "loss": 3.1094, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8728323699421967e-05, | |
| "loss": 2.7812, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9075144508670522e-05, | |
| "loss": 1.9785, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9421965317919077e-05, | |
| "loss": 3.4531, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.976878612716763e-05, | |
| "loss": 2.1875, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.0115606936416184e-05, | |
| "loss": 1.9697, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.046242774566474e-05, | |
| "loss": 2.1777, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.0809248554913295e-05, | |
| "loss": 1.8594, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.115606936416185e-05, | |
| "loss": 2.2441, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.1502890173410405e-05, | |
| "loss": 2.3242, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.184971098265896e-05, | |
| "loss": 1.9541, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.2196531791907513e-05, | |
| "loss": 1.7412, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.2543352601156068e-05, | |
| "loss": 2.2969, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.2890173410404623e-05, | |
| "loss": 1.3818, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.323699421965318e-05, | |
| "loss": 1.9668, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.3583815028901737e-05, | |
| "loss": 2.4727, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.3930635838150292e-05, | |
| "loss": 2.0645, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.4277456647398844e-05, | |
| "loss": 2.0312, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.46242774566474e-05, | |
| "loss": 2.0547, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.4971098265895955e-05, | |
| "loss": 1.7764, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.531791907514451e-05, | |
| "loss": 1.6074, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.5664739884393065e-05, | |
| "loss": 1.2148, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.601156069364162e-05, | |
| "loss": 1.3525, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.6358381502890176e-05, | |
| "loss": 1.6963, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.6705202312138728e-05, | |
| "loss": 1.4746, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.7052023121387283e-05, | |
| "loss": 1.8105, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.7398843930635838e-05, | |
| "loss": 1.5195, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.7745664739884393e-05, | |
| "loss": 1.1855, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.809248554913295e-05, | |
| "loss": 1.3115, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.8439306358381504e-05, | |
| "loss": 1.6318, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.878612716763006e-05, | |
| "loss": 2.2656, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.913294797687861e-05, | |
| "loss": 1.8789, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.9479768786127166e-05, | |
| "loss": 1.3555, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.982658959537572e-05, | |
| "loss": 1.1211, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.017341040462428e-05, | |
| "loss": 1.3418, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.052023121387283e-05, | |
| "loss": 1.5234, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.086705202312139e-05, | |
| "loss": 0.8867, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.121387283236994e-05, | |
| "loss": 1.2432, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.15606936416185e-05, | |
| "loss": 1.5752, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.190751445086705e-05, | |
| "loss": 1.541, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.225433526011561e-05, | |
| "loss": 1.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.260115606936416e-05, | |
| "loss": 1.4531, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.294797687861272e-05, | |
| "loss": 1.6367, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.329479768786127e-05, | |
| "loss": 1.0898, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.364161849710983e-05, | |
| "loss": 1.8867, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.398843930635839e-05, | |
| "loss": 1.4658, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.433526011560694e-05, | |
| "loss": 1.2217, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.468208092485549e-05, | |
| "loss": 1.3906, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.5028901734104043e-05, | |
| "loss": 1.4932, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.53757225433526e-05, | |
| "loss": 1.0254, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.5722543352601154e-05, | |
| "loss": 1.1143, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.606936416184971e-05, | |
| "loss": 1.6777, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.6416184971098265e-05, | |
| "loss": 0.6465, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.676300578034682e-05, | |
| "loss": 1.375, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.7109826589595375e-05, | |
| "loss": 1.1191, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.7456647398843934e-05, | |
| "loss": 1.8301, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.7803468208092486e-05, | |
| "loss": 1.6553, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.8150289017341044e-05, | |
| "loss": 1.5566, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.8497109826589596e-05, | |
| "loss": 1.6084, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.8843930635838155e-05, | |
| "loss": 1.3359, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.9190751445086707e-05, | |
| "loss": 1.1104, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.953757225433526e-05, | |
| "loss": 0.9771, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.988439306358381e-05, | |
| "loss": 1.4971, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.023121387283237e-05, | |
| "loss": 1.627, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.057803468208093e-05, | |
| "loss": 0.8262, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.092485549132948e-05, | |
| "loss": 1.5273, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.127167630057804e-05, | |
| "loss": 0.6133, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.161849710982659e-05, | |
| "loss": 1.1729, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.196531791907515e-05, | |
| "loss": 0.5942, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.23121387283237e-05, | |
| "loss": 1.0908, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.265895953757226e-05, | |
| "loss": 1.5547, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.300578034682081e-05, | |
| "loss": 0.8452, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.335260115606937e-05, | |
| "loss": 0.7603, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.369942196531792e-05, | |
| "loss": 0.7627, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.404624277456647e-05, | |
| "loss": 0.6387, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.4393063583815025e-05, | |
| "loss": 1.0439, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.4739884393063584e-05, | |
| "loss": 0.8623, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.5086705202312136e-05, | |
| "loss": 1.9482, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.5433526011560694e-05, | |
| "loss": 1.2148, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.5780346820809246e-05, | |
| "loss": 1.249, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.6127167630057805e-05, | |
| "loss": 1.2393, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.647398843930636e-05, | |
| "loss": 1.6406, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.6820809248554915e-05, | |
| "loss": 1.1143, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.7167630057803474e-05, | |
| "loss": 1.543, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.7514450867052026e-05, | |
| "loss": 1.9912, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.7861271676300585e-05, | |
| "loss": 0.9761, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.8208092485549136e-05, | |
| "loss": 1.4238, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.855491329479769e-05, | |
| "loss": 1.04, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.890173410404624e-05, | |
| "loss": 1.2793, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.92485549132948e-05, | |
| "loss": 1.8594, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.959537572254335e-05, | |
| "loss": 0.8799, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.994219653179191e-05, | |
| "loss": 0.2998, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.028901734104046e-05, | |
| "loss": 0.9536, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.063583815028902e-05, | |
| "loss": 1.4346, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.098265895953757e-05, | |
| "loss": 0.7554, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.132947976878613e-05, | |
| "loss": 0.7715, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.167630057803468e-05, | |
| "loss": 1.6406, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.202312138728324e-05, | |
| "loss": 0.6548, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.236994219653179e-05, | |
| "loss": 1.9697, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.271676300578035e-05, | |
| "loss": 1.2061, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.30635838150289e-05, | |
| "loss": 0.8584, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.3410404624277455e-05, | |
| "loss": 0.8721, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.3757225433526014e-05, | |
| "loss": 0.8486, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.4104046242774566e-05, | |
| "loss": 1.0762, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.4450867052023124e-05, | |
| "loss": 0.5454, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.4797687861271676e-05, | |
| "loss": 0.8379, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.5144508670520235e-05, | |
| "loss": 0.5874, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.5491329479768787e-05, | |
| "loss": 0.8701, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.5838150289017345e-05, | |
| "loss": 0.9492, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.61849710982659e-05, | |
| "loss": 1.2998, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.6531791907514456e-05, | |
| "loss": 1.7168, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.687861271676301e-05, | |
| "loss": 0.6938, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.7225433526011566e-05, | |
| "loss": 1.0918, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.757225433526012e-05, | |
| "loss": 1.1143, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.791907514450867e-05, | |
| "loss": 1.3672, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.826589595375722e-05, | |
| "loss": 1.0215, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.861271676300578e-05, | |
| "loss": 1.4531, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.895953757225433e-05, | |
| "loss": 1.3926, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.930635838150289e-05, | |
| "loss": 1.624, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.965317919075144e-05, | |
| "loss": 1.2578, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 6e-05, | |
| "loss": 1.1318, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.988394584139265e-05, | |
| "loss": 0.7456, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.97678916827853e-05, | |
| "loss": 1.1855, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.965183752417795e-05, | |
| "loss": 0.6221, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.9535783365570604e-05, | |
| "loss": 1.5713, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.9419729206963253e-05, | |
| "loss": 1.3613, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.9303675048355896e-05, | |
| "loss": 0.7422, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.918762088974855e-05, | |
| "loss": 0.6445, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.90715667311412e-05, | |
| "loss": 1.0869, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.895551257253385e-05, | |
| "loss": 2.543, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.88394584139265e-05, | |
| "loss": 1.0225, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.872340425531915e-05, | |
| "loss": 1.7207, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.86073500967118e-05, | |
| "loss": 1.0156, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.849129593810445e-05, | |
| "loss": 0.917, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.8375241779497095e-05, | |
| "loss": 1.4346, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.825918762088975e-05, | |
| "loss": 1.4922, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.81431334622824e-05, | |
| "loss": 1.3213, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.8027079303675055e-05, | |
| "loss": 1.0684, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.79110251450677e-05, | |
| "loss": 0.8037, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.7794970986460346e-05, | |
| "loss": 1.9531, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.7678916827853e-05, | |
| "loss": 1.0098, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.756286266924565e-05, | |
| "loss": 1.0127, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.74468085106383e-05, | |
| "loss": 0.4705, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.733075435203095e-05, | |
| "loss": 1.1465, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.72147001934236e-05, | |
| "loss": 1.5605, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.7098646034816254e-05, | |
| "loss": 1.6436, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.6982591876208896e-05, | |
| "loss": 0.7319, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.6866537717601545e-05, | |
| "loss": 1.4277, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.67504835589942e-05, | |
| "loss": 1.0391, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.6634429400386843e-05, | |
| "loss": 0.9307, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.65183752417795e-05, | |
| "loss": 1.6367, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.640232108317215e-05, | |
| "loss": 1.2861, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.62862669245648e-05, | |
| "loss": 1.4727, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.6170212765957446e-05, | |
| "loss": 0.9707, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.6054158607350095e-05, | |
| "loss": 0.937, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.593810444874275e-05, | |
| "loss": 1.1699, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.58220502901354e-05, | |
| "loss": 0.8452, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.570599613152804e-05, | |
| "loss": 0.999, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.55899419729207e-05, | |
| "loss": 1.3018, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.547388781431335e-05, | |
| "loss": 0.9292, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.5357833655706e-05, | |
| "loss": 0.998, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.5241779497098645e-05, | |
| "loss": 2.3984, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.5125725338491294e-05, | |
| "loss": 0.9868, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.500967117988395e-05, | |
| "loss": 0.8711, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.48936170212766e-05, | |
| "loss": 1.2734, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.477756286266924e-05, | |
| "loss": 1.6826, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.46615087040619e-05, | |
| "loss": 1.2676, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.4545454545454546e-05, | |
| "loss": 1.0039, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.44294003868472e-05, | |
| "loss": 1.1943, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.4313346228239844e-05, | |
| "loss": 0.9438, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.419729206963249e-05, | |
| "loss": 0.915, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.408123791102515e-05, | |
| "loss": 0.9854, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.39651837524178e-05, | |
| "loss": 0.7554, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.384912959381045e-05, | |
| "loss": 1.1641, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.3733075435203096e-05, | |
| "loss": 1.1104, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.3617021276595745e-05, | |
| "loss": 1.0586, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.3500967117988394e-05, | |
| "loss": 1.2461, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.338491295938104e-05, | |
| "loss": 0.5435, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.32688588007737e-05, | |
| "loss": 0.9458, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.315280464216635e-05, | |
| "loss": 1.1934, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.303675048355899e-05, | |
| "loss": 0.3909, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.2920696324951646e-05, | |
| "loss": 1.0605, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.2804642166344295e-05, | |
| "loss": 1.084, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.2688588007736944e-05, | |
| "loss": 0.4778, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.257253384912959e-05, | |
| "loss": 0.9976, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.245647969052224e-05, | |
| "loss": 0.4556, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.23404255319149e-05, | |
| "loss": 0.8711, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5.222437137330755e-05, | |
| "loss": 0.7954, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5.210831721470019e-05, | |
| "loss": 0.2117, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5.1992263056092845e-05, | |
| "loss": 1.2148, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5.1876208897485494e-05, | |
| "loss": 1.1367, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.176015473887815e-05, | |
| "loss": 1.1689, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.164410058027079e-05, | |
| "loss": 0.7588, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.152804642166344e-05, | |
| "loss": 1.0527, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5.14119922630561e-05, | |
| "loss": 1.3223, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5.1295938104448746e-05, | |
| "loss": 1.0264, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5.1179883945841395e-05, | |
| "loss": 0.9502, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5.1063829787234044e-05, | |
| "loss": 0.6611, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5.094777562862669e-05, | |
| "loss": 0.8711, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5.083172147001935e-05, | |
| "loss": 0.7632, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5.071566731141199e-05, | |
| "loss": 0.6128, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5.059961315280464e-05, | |
| "loss": 1.7568, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5.0483558994197296e-05, | |
| "loss": 0.6343, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5.036750483558994e-05, | |
| "loss": 1.0986, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5.0251450676982594e-05, | |
| "loss": 1.0752, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5.013539651837524e-05, | |
| "loss": 0.9414, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5.001934235976789e-05, | |
| "loss": 1.0146, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.990328820116054e-05, | |
| "loss": 0.5898, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.978723404255319e-05, | |
| "loss": 0.8062, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.9671179883945846e-05, | |
| "loss": 1.1055, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.9555125725338495e-05, | |
| "loss": 0.8584, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.943907156673114e-05, | |
| "loss": 0.8101, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.932301740812379e-05, | |
| "loss": 0.8618, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.920696324951644e-05, | |
| "loss": 0.5723, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.90909090909091e-05, | |
| "loss": 1.5088, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.897485493230174e-05, | |
| "loss": 0.407, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.885880077369439e-05, | |
| "loss": 0.8809, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.8742746615087045e-05, | |
| "loss": 1.0938, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.8626692456479694e-05, | |
| "loss": 0.7402, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.8510638297872336e-05, | |
| "loss": 1.0674, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.839458413926499e-05, | |
| "loss": 0.4993, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.827852998065764e-05, | |
| "loss": 0.998, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.81624758220503e-05, | |
| "loss": 1.2129, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.804642166344294e-05, | |
| "loss": 1.208, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.793036750483559e-05, | |
| "loss": 1.3545, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.7814313346228244e-05, | |
| "loss": 0.8594, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.769825918762089e-05, | |
| "loss": 0.7935, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.758220502901354e-05, | |
| "loss": 0.8599, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.746615087040619e-05, | |
| "loss": 0.9775, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.735009671179884e-05, | |
| "loss": 0.7212, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.723404255319149e-05, | |
| "loss": 0.9355, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.711798839458414e-05, | |
| "loss": 0.8828, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.7001934235976794e-05, | |
| "loss": 0.7603, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.688588007736944e-05, | |
| "loss": 0.417, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.6769825918762085e-05, | |
| "loss": 0.7583, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.665377176015474e-05, | |
| "loss": 0.5708, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.653771760154739e-05, | |
| "loss": 1.377, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.642166344294004e-05, | |
| "loss": 1.5684, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.630560928433269e-05, | |
| "loss": 0.9072, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.618955512572534e-05, | |
| "loss": 0.8252, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.607350096711799e-05, | |
| "loss": 1.085, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.595744680851064e-05, | |
| "loss": 0.8286, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.5841392649903284e-05, | |
| "loss": 1.1875, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.572533849129594e-05, | |
| "loss": 0.7661, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.560928433268859e-05, | |
| "loss": 0.9819, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.5493230174081244e-05, | |
| "loss": 1.1055, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.537717601547389e-05, | |
| "loss": 0.7915, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.5261121856866536e-05, | |
| "loss": 0.6772, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.514506769825919e-05, | |
| "loss": 0.9434, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.502901353965184e-05, | |
| "loss": 0.6226, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.491295938104449e-05, | |
| "loss": 0.5186, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.479690522243714e-05, | |
| "loss": 0.5127, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.468085106382979e-05, | |
| "loss": 0.998, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.4564796905222443e-05, | |
| "loss": 1.292, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.4448742746615086e-05, | |
| "loss": 1.3428, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.4332688588007735e-05, | |
| "loss": 1.5078, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.421663442940039e-05, | |
| "loss": 0.5425, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.410058027079303e-05, | |
| "loss": 1.1709, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.398452611218569e-05, | |
| "loss": 0.5474, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.386847195357834e-05, | |
| "loss": 0.7163, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.3752417794970987e-05, | |
| "loss": 0.9209, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.3636363636363636e-05, | |
| "loss": 0.4958, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.3520309477756285e-05, | |
| "loss": 0.5449, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.340425531914894e-05, | |
| "loss": 0.416, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.328820116054159e-05, | |
| "loss": 0.7725, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.317214700193423e-05, | |
| "loss": 0.9961, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.305609284332689e-05, | |
| "loss": 1.0986, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.2940038684719537e-05, | |
| "loss": 0.7969, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.282398452611219e-05, | |
| "loss": 0.6665, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.2707930367504835e-05, | |
| "loss": 1.0449, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.2591876208897484e-05, | |
| "loss": 0.5278, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.247582205029014e-05, | |
| "loss": 1.2695, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.235976789168279e-05, | |
| "loss": 1.2998, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.224371373307543e-05, | |
| "loss": 0.5835, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.2127659574468086e-05, | |
| "loss": 0.8564, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.2011605415860735e-05, | |
| "loss": 0.8145, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.189555125725339e-05, | |
| "loss": 0.8462, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.1779497098646034e-05, | |
| "loss": 0.5107, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.166344294003868e-05, | |
| "loss": 0.7842, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.154738878143134e-05, | |
| "loss": 1.0312, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.143133462282399e-05, | |
| "loss": 0.7061, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.1315280464216636e-05, | |
| "loss": 1.1055, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.1199226305609285e-05, | |
| "loss": 0.6479, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.1083172147001934e-05, | |
| "loss": 0.6284, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.0967117988394583e-05, | |
| "loss": 0.7573, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.085106382978723e-05, | |
| "loss": 1.6924, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.073500967117989e-05, | |
| "loss": 1.2656, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.061895551257254e-05, | |
| "loss": 1.209, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.050290135396518e-05, | |
| "loss": 0.6792, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.0386847195357835e-05, | |
| "loss": 0.5415, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.0270793036750484e-05, | |
| "loss": 0.6519, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.0154738878143133e-05, | |
| "loss": 0.5938, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.003868471953578e-05, | |
| "loss": 1.6006, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.992263056092843e-05, | |
| "loss": 1.0391, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.980657640232109e-05, | |
| "loss": 0.5776, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9690522243713736e-05, | |
| "loss": 0.7881, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.957446808510638e-05, | |
| "loss": 1.1357, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9458413926499034e-05, | |
| "loss": 0.6836, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.934235976789168e-05, | |
| "loss": 0.5996, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.922630560928434e-05, | |
| "loss": 0.25, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.911025145067698e-05, | |
| "loss": 1.0361, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.899419729206963e-05, | |
| "loss": 0.2573, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.8878143133462286e-05, | |
| "loss": 0.9463, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.8762088974854935e-05, | |
| "loss": 1.1104, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.8646034816247584e-05, | |
| "loss": 0.7646, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.852998065764023e-05, | |
| "loss": 0.5918, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.841392649903288e-05, | |
| "loss": 1.4531, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.829787234042554e-05, | |
| "loss": 0.9487, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.818181818181818e-05, | |
| "loss": 0.9604, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.806576402321083e-05, | |
| "loss": 0.6528, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.7949709864603485e-05, | |
| "loss": 1.0029, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.783365570599613e-05, | |
| "loss": 0.7422, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.771760154738878e-05, | |
| "loss": 0.7471, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.760154738878143e-05, | |
| "loss": 0.938, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.748549323017408e-05, | |
| "loss": 0.6191, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.736943907156673e-05, | |
| "loss": 0.46, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.725338491295938e-05, | |
| "loss": 1.1885, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.7137330754352035e-05, | |
| "loss": 1.3447, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.7021276595744684e-05, | |
| "loss": 0.749, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.6905222437137326e-05, | |
| "loss": 0.4883, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.678916827852998e-05, | |
| "loss": 0.5981, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.667311411992263e-05, | |
| "loss": 0.9893, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.655705996131529e-05, | |
| "loss": 0.7104, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.644100580270793e-05, | |
| "loss": 0.5493, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.632495164410058e-05, | |
| "loss": 0.5688, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.6208897485493234e-05, | |
| "loss": 0.4919, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.609284332688588e-05, | |
| "loss": 0.8325, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.5976789168278525e-05, | |
| "loss": 0.4885, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.586073500967118e-05, | |
| "loss": 1.084, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.574468085106383e-05, | |
| "loss": 0.6152, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.5628626692456486e-05, | |
| "loss": 0.489, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.551257253384913e-05, | |
| "loss": 0.667, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.539651837524178e-05, | |
| "loss": 0.606, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.528046421663443e-05, | |
| "loss": 0.4199, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.516441005802708e-05, | |
| "loss": 0.9307, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.504835589941973e-05, | |
| "loss": 0.8564, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.493230174081238e-05, | |
| "loss": 0.4531, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.481624758220503e-05, | |
| "loss": 0.875, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.470019342359768e-05, | |
| "loss": 1.0645, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.458413926499033e-05, | |
| "loss": 0.7993, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.446808510638298e-05, | |
| "loss": 0.8906, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.435203094777563e-05, | |
| "loss": 1.0947, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.4235976789168274e-05, | |
| "loss": 1.2812, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.411992263056093e-05, | |
| "loss": 0.4214, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.400386847195358e-05, | |
| "loss": 1.1426, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.388781431334623e-05, | |
| "loss": 0.5415, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.377176015473888e-05, | |
| "loss": 1.0723, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.3655705996131526e-05, | |
| "loss": 0.7412, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.353965183752418e-05, | |
| "loss": 0.9688, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.342359767891683e-05, | |
| "loss": 0.5547, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.330754352030947e-05, | |
| "loss": 0.5977, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.319148936170213e-05, | |
| "loss": 0.9434, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.307543520309478e-05, | |
| "loss": 0.9668, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.2959381044487434e-05, | |
| "loss": 0.8994, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.2843326885880076e-05, | |
| "loss": 0.5439, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.2727272727272725e-05, | |
| "loss": 1.168, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.261121856866538e-05, | |
| "loss": 1.292, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.249516441005803e-05, | |
| "loss": 0.6426, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.237911025145068e-05, | |
| "loss": 0.3635, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.226305609284333e-05, | |
| "loss": 0.9268, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.214700193423598e-05, | |
| "loss": 0.4749, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.203094777562863e-05, | |
| "loss": 0.8774, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.1914893617021275e-05, | |
| "loss": 1.0264, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.1798839458413924e-05, | |
| "loss": 0.8276, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.168278529980658e-05, | |
| "loss": 1.0762, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.156673114119923e-05, | |
| "loss": 1.0459, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.145067698259188e-05, | |
| "loss": 0.6641, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.133462282398453e-05, | |
| "loss": 1.9824, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.1218568665377176e-05, | |
| "loss": 1.2617, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.1102514506769825e-05, | |
| "loss": 0.5913, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.0986460348162474e-05, | |
| "loss": 0.6343, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.087040618955513e-05, | |
| "loss": 0.5215, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.075435203094778e-05, | |
| "loss": 0.7549, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.063829787234042e-05, | |
| "loss": 0.3704, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.052224371373308e-05, | |
| "loss": 0.6704, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.0406189555125726e-05, | |
| "loss": 0.9639, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.0290135396518378e-05, | |
| "loss": 0.8818, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.0174081237911027e-05, | |
| "loss": 0.6758, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.0058027079303673e-05, | |
| "loss": 0.728, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9941972920696325e-05, | |
| "loss": 0.6206, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.9825918762088974e-05, | |
| "loss": 0.7578, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.9709864603481627e-05, | |
| "loss": 0.6523, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.9593810444874276e-05, | |
| "loss": 0.4155, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.9477756286266925e-05, | |
| "loss": 0.5464, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.9361702127659574e-05, | |
| "loss": 0.4067, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.9245647969052226e-05, | |
| "loss": 1.3379, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.9129593810444875e-05, | |
| "loss": 1.0723, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.9013539651837528e-05, | |
| "loss": 0.6567, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8897485493230173e-05, | |
| "loss": 0.4177, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8781431334622826e-05, | |
| "loss": 0.4412, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8665377176015475e-05, | |
| "loss": 0.3015, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.8549323017408127e-05, | |
| "loss": 0.7876, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.8433268858800773e-05, | |
| "loss": 0.7197, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.8317214700193422e-05, | |
| "loss": 1.1719, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8201160541586074e-05, | |
| "loss": 1.2109, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8085106382978723e-05, | |
| "loss": 0.301, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.7969052224371376e-05, | |
| "loss": 0.7065, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.785299806576402e-05, | |
| "loss": 1.5684, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.7736943907156674e-05, | |
| "loss": 0.6211, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.7620889748549323e-05, | |
| "loss": 0.7876, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.7504835589941975e-05, | |
| "loss": 0.7261, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.738878143133462e-05, | |
| "loss": 0.6401, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.7272727272727273e-05, | |
| "loss": 0.4714, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.7156673114119922e-05, | |
| "loss": 0.6025, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.7040618955512575e-05, | |
| "loss": 0.708, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.6924564796905224e-05, | |
| "loss": 0.7397, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.6808510638297873e-05, | |
| "loss": 0.8315, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.669245647969052e-05, | |
| "loss": 0.7134, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.6576402321083174e-05, | |
| "loss": 0.4951, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.6460348162475823e-05, | |
| "loss": 0.8564, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.6344294003868472e-05, | |
| "loss": 0.6763, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.622823984526112e-05, | |
| "loss": 0.5425, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.6112185686653773e-05, | |
| "loss": 1.7285, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.5996131528046423e-05, | |
| "loss": 1.1133, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.5880077369439075e-05, | |
| "loss": 0.8857, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.576402321083172e-05, | |
| "loss": 0.4885, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.5647969052224373e-05, | |
| "loss": 0.5234, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.5531914893617022e-05, | |
| "loss": 1.1143, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.5415860735009674e-05, | |
| "loss": 1.0, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.529980657640232e-05, | |
| "loss": 0.4348, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.518375241779497e-05, | |
| "loss": 1.0801, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.506769825918762e-05, | |
| "loss": 0.8135, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.495164410058027e-05, | |
| "loss": 0.4106, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.4835589941972923e-05, | |
| "loss": 0.7329, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.471953578336557e-05, | |
| "loss": 1.46, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.460348162475822e-05, | |
| "loss": 0.2554, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.448742746615087e-05, | |
| "loss": 0.6641, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.4371373307543522e-05, | |
| "loss": 0.3413, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.4255319148936168e-05, | |
| "loss": 0.5776, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.413926499032882e-05, | |
| "loss": 0.5396, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.402321083172147e-05, | |
| "loss": 0.6875, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.3907156673114122e-05, | |
| "loss": 0.7529, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.379110251450677e-05, | |
| "loss": 0.5298, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.367504835589942e-05, | |
| "loss": 1.1924, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.355899419729207e-05, | |
| "loss": 0.479, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.344294003868472e-05, | |
| "loss": 0.73, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.332688588007737e-05, | |
| "loss": 0.7144, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.321083172147002e-05, | |
| "loss": 0.205, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.309477756286267e-05, | |
| "loss": 1.4053, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.297872340425532e-05, | |
| "loss": 0.5771, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.286266924564797e-05, | |
| "loss": 1.2969, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.2746615087040622e-05, | |
| "loss": 0.7012, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.2630560928433268e-05, | |
| "loss": 0.4915, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.251450676982592e-05, | |
| "loss": 0.7959, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.239845261121857e-05, | |
| "loss": 1.3574, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.2282398452611222e-05, | |
| "loss": 0.9609, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.2166344294003867e-05, | |
| "loss": 1.0781, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.2050290135396516e-05, | |
| "loss": 0.8276, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.193423597678917e-05, | |
| "loss": 0.6729, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.1818181818181818e-05, | |
| "loss": 0.6982, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.170212765957447e-05, | |
| "loss": 0.2954, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.1586073500967116e-05, | |
| "loss": 0.9126, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.1470019342359768e-05, | |
| "loss": 0.7505, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.1353965183752417e-05, | |
| "loss": 0.792, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.123791102514507e-05, | |
| "loss": 0.406, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.1121856866537715e-05, | |
| "loss": 1.3525, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.1005802707930368e-05, | |
| "loss": 0.4583, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.0889748549323017e-05, | |
| "loss": 0.5293, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.077369439071567e-05, | |
| "loss": 0.7393, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.0657640232108318e-05, | |
| "loss": 0.877, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.0541586073500967e-05, | |
| "loss": 0.4531, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.0425531914893616e-05, | |
| "loss": 1.2695, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.030947775628627e-05, | |
| "loss": 0.8027, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.0193423597678918e-05, | |
| "loss": 0.9111, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.0077369439071567e-05, | |
| "loss": 0.6533, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.9961315280464216e-05, | |
| "loss": 0.4072, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.9845261121856868e-05, | |
| "loss": 0.9634, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.9729206963249517e-05, | |
| "loss": 0.8545, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.961315280464217e-05, | |
| "loss": 0.7363, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.9497098646034815e-05, | |
| "loss": 0.6431, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.9381044487427468e-05, | |
| "loss": 0.5728, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.9264990328820117e-05, | |
| "loss": 0.3672, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.914893617021277e-05, | |
| "loss": 0.9966, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.9032882011605415e-05, | |
| "loss": 0.6738, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.8916827852998064e-05, | |
| "loss": 0.5571, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.8800773694390716e-05, | |
| "loss": 0.5581, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.8684719535783365e-05, | |
| "loss": 0.8774, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.8568665377176018e-05, | |
| "loss": 0.8042, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.8452611218568663e-05, | |
| "loss": 1.0947, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.8336557059961316e-05, | |
| "loss": 0.7212, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.8220502901353965e-05, | |
| "loss": 0.8325, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.8104448742746617e-05, | |
| "loss": 0.4131, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.7988394584139263e-05, | |
| "loss": 0.5898, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.7872340425531915e-05, | |
| "loss": 0.9541, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.7756286266924564e-05, | |
| "loss": 0.3879, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.7640232108317216e-05, | |
| "loss": 0.6372, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.7524177949709866e-05, | |
| "loss": 0.4324, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.7408123791102515e-05, | |
| "loss": 0.6602, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.7292069632495164e-05, | |
| "loss": 0.3093, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.7176015473887816e-05, | |
| "loss": 0.9644, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.7059961315280465e-05, | |
| "loss": 0.5732, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.6943907156673114e-05, | |
| "loss": 0.7998, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.6827852998065763e-05, | |
| "loss": 1.0518, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.6711798839458415e-05, | |
| "loss": 0.8706, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.6595744680851064e-05, | |
| "loss": 0.6118, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.6479690522243717e-05, | |
| "loss": 1.1592, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.6363636363636363e-05, | |
| "loss": 0.8555, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.6247582205029015e-05, | |
| "loss": 0.6455, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.6131528046421664e-05, | |
| "loss": 0.272, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.6015473887814316e-05, | |
| "loss": 0.9355, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.5899419729206962e-05, | |
| "loss": 0.6943, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.5783365570599614e-05, | |
| "loss": 1.0205, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.5667311411992263e-05, | |
| "loss": 0.8364, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.5551257253384912e-05, | |
| "loss": 1.0449, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.5435203094777565e-05, | |
| "loss": 0.4841, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.531914893617021e-05, | |
| "loss": 0.9956, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.5203094777562863e-05, | |
| "loss": 0.6279, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.5087040618955514e-05, | |
| "loss": 0.3101, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.4970986460348163e-05, | |
| "loss": 0.6255, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.4854932301740813e-05, | |
| "loss": 1.3223, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.4738878143133462e-05, | |
| "loss": 0.22, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.4622823984526113e-05, | |
| "loss": 0.4963, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.4506769825918764e-05, | |
| "loss": 1.002, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.4390715667311413e-05, | |
| "loss": 0.9609, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.4274661508704064e-05, | |
| "loss": 0.7744, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.4158607350096711e-05, | |
| "loss": 0.542, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.4042553191489362e-05, | |
| "loss": 0.668, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.392649903288201e-05, | |
| "loss": 0.8008, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3810444874274661e-05, | |
| "loss": 0.5571, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.369439071566731e-05, | |
| "loss": 0.8682, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3578336557059961e-05, | |
| "loss": 0.5957, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3462282398452612e-05, | |
| "loss": 0.5708, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.334622823984526e-05, | |
| "loss": 0.7559, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3230174081237912e-05, | |
| "loss": 0.4141, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.311411992263056e-05, | |
| "loss": 1.0518, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.2998065764023211e-05, | |
| "loss": 1.3086, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.288201160541586e-05, | |
| "loss": 0.8892, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.2765957446808511e-05, | |
| "loss": 0.447, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.264990328820116e-05, | |
| "loss": 0.6396, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.253384912959381e-05, | |
| "loss": 1.0215, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2417794970986461e-05, | |
| "loss": 0.5806, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.230174081237911e-05, | |
| "loss": 1.3672, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.2185686653771761e-05, | |
| "loss": 0.5278, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.206963249516441e-05, | |
| "loss": 0.8687, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.1953578336557061e-05, | |
| "loss": 0.8115, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.183752417794971e-05, | |
| "loss": 0.292, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.172147001934236e-05, | |
| "loss": 1.0811, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.160541586073501e-05, | |
| "loss": 0.6865, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.148936170212766e-05, | |
| "loss": 1.0137, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.1373307543520311e-05, | |
| "loss": 1.6211, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.125725338491296e-05, | |
| "loss": 0.8965, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.1141199226305611e-05, | |
| "loss": 0.2993, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.1025145067698258e-05, | |
| "loss": 0.5991, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.0909090909090909e-05, | |
| "loss": 0.6768, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.0793036750483558e-05, | |
| "loss": 0.6045, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.0676982591876209e-05, | |
| "loss": 0.3665, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.0560928433268858e-05, | |
| "loss": 1.4131, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.0444874274661508e-05, | |
| "loss": 1.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.0328820116054159e-05, | |
| "loss": 0.6123, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.0212765957446808e-05, | |
| "loss": 0.541, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.0096711798839459e-05, | |
| "loss": 0.7446, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 9.980657640232108e-06, | |
| "loss": 0.4622, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 9.864603481624759e-06, | |
| "loss": 0.4734, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 9.748549323017408e-06, | |
| "loss": 0.5908, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 9.632495164410058e-06, | |
| "loss": 0.8159, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 9.516441005802707e-06, | |
| "loss": 0.2949, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 9.400386847195358e-06, | |
| "loss": 0.4207, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 9.284332688588009e-06, | |
| "loss": 0.8066, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 9.168278529980658e-06, | |
| "loss": 0.5264, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 9.052224371373309e-06, | |
| "loss": 0.2637, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.936170212765958e-06, | |
| "loss": 0.8501, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.820116054158608e-06, | |
| "loss": 0.7407, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.704061895551257e-06, | |
| "loss": 1.3408, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.588007736943908e-06, | |
| "loss": 1.96, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.471953578336557e-06, | |
| "loss": 0.6353, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.355899419729208e-06, | |
| "loss": 1.1045, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.239845261121858e-06, | |
| "loss": 1.373, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.123791102514507e-06, | |
| "loss": 0.9888, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.007736943907158e-06, | |
| "loss": 0.4907, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.891682785299807e-06, | |
| "loss": 0.4163, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.775628626692456e-06, | |
| "loss": 0.7407, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.659574468085105e-06, | |
| "loss": 0.6738, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.543520309477757e-06, | |
| "loss": 0.7217, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.427466150870407e-06, | |
| "loss": 0.3872, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.3114119922630565e-06, | |
| "loss": 0.7197, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.195357833655706e-06, | |
| "loss": 1.125, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.0793036750483554e-06, | |
| "loss": 1.0352, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.963249516441005e-06, | |
| "loss": 0.6802, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.847195357833655e-06, | |
| "loss": 0.6938, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.731141199226306e-06, | |
| "loss": 1.0049, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.615087040618956e-06, | |
| "loss": 0.5635, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.499032882011606e-06, | |
| "loss": 0.7109, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.3829787234042555e-06, | |
| "loss": 1.1309, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.266924564796905e-06, | |
| "loss": 0.5781, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.150870406189555e-06, | |
| "loss": 0.7534, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.034816247582205e-06, | |
| "loss": 0.6694, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 5.918762088974855e-06, | |
| "loss": 0.9985, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.802707930367505e-06, | |
| "loss": 0.7793, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.6866537717601556e-06, | |
| "loss": 0.5532, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.5705996131528054e-06, | |
| "loss": 0.4526, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 1.2031, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 5.338491295938104e-06, | |
| "loss": 0.9697, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 5.222437137330754e-06, | |
| "loss": 1.1504, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 5.106382978723404e-06, | |
| "loss": 0.3545, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.990328820116054e-06, | |
| "loss": 0.7798, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.874274661508704e-06, | |
| "loss": 1.0352, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.758220502901354e-06, | |
| "loss": 0.9839, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.642166344294004e-06, | |
| "loss": 1.3174, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.526112185686654e-06, | |
| "loss": 0.7246, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.410058027079304e-06, | |
| "loss": 1.2266, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.294003868471954e-06, | |
| "loss": 0.5737, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.177949709864604e-06, | |
| "loss": 0.6113, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.061895551257254e-06, | |
| "loss": 0.8057, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.945841392649904e-06, | |
| "loss": 0.3179, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.829787234042553e-06, | |
| "loss": 0.3594, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.7137330754352033e-06, | |
| "loss": 0.3843, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.597678916827853e-06, | |
| "loss": 0.7217, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.4816247582205027e-06, | |
| "loss": 0.772, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.365570599613153e-06, | |
| "loss": 0.7642, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.249516441005803e-06, | |
| "loss": 0.4045, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.1334622823984527e-06, | |
| "loss": 0.8843, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.0174081237911026e-06, | |
| "loss": 0.48, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 2.9013539651837524e-06, | |
| "loss": 0.5723, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 2.7852998065764027e-06, | |
| "loss": 0.5728, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 2.669245647969052e-06, | |
| "loss": 0.8955, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.553191489361702e-06, | |
| "loss": 0.5557, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.437137330754352e-06, | |
| "loss": 0.9785, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.321083172147002e-06, | |
| "loss": 0.397, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.205029013539652e-06, | |
| "loss": 0.5137, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.088974854932302e-06, | |
| "loss": 0.4404, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.972920696324952e-06, | |
| "loss": 0.6499, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.8568665377176017e-06, | |
| "loss": 0.4558, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7408123791102513e-06, | |
| "loss": 1.3662, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.6247582205029014e-06, | |
| "loss": 1.3398, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.5087040618955513e-06, | |
| "loss": 1.0703, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.3926499032882014e-06, | |
| "loss": 0.208, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.276595744680851e-06, | |
| "loss": 0.5503, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.160541586073501e-06, | |
| "loss": 0.5762, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.044487427466151e-06, | |
| "loss": 0.627, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 9.284332688588008e-07, | |
| "loss": 0.5713, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.123791102514507e-07, | |
| "loss": 0.6621, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.963249516441007e-07, | |
| "loss": 0.6255, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.802707930367505e-07, | |
| "loss": 1.2549, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.642166344294004e-07, | |
| "loss": 1.0869, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 3.4816247582205034e-07, | |
| "loss": 0.5698, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 2.321083172147002e-07, | |
| "loss": 0.52, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.160541586073501e-07, | |
| "loss": 0.4097, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.9941, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 690, | |
| "total_flos": 2.215462037272658e+18, | |
| "train_loss": 1.2527947025022645, | |
| "train_runtime": 134.5045, | |
| "train_samples_per_second": 1316.952, | |
| "train_steps_per_second": 5.13 | |
| } | |
| ], | |
| "max_steps": 690, | |
| "num_train_epochs": 2, | |
| "total_flos": 2.215462037272658e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |