| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 293, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 1.0441, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 1.0424, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.0489, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 1.0717, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.0396, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.0776, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 1.0178, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.0666666666666667e-05, | |
| "loss": 1.0144, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.9903, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.9258, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.4666666666666666e-05, | |
| "loss": 0.9325, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.9532, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.7333333333333336e-05, | |
| "loss": 0.9473, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.866666666666667e-05, | |
| "loss": 0.9213, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.866666666666667e-05, | |
| "loss": 0.9497, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2e-05, | |
| "loss": 0.9721, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9999361478484043e-05, | |
| "loss": 0.9294, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.999744599547812e-05, | |
| "loss": 0.9948, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.999425379559765e-05, | |
| "loss": 0.9043, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9989785286500294e-05, | |
| "loss": 0.9031, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.99840410388339e-05, | |
| "loss": 0.8928, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.99770217861636e-05, | |
| "loss": 0.9324, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9968728424878178e-05, | |
| "loss": 0.9068, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9959162014075553e-05, | |
| "loss": 0.9057, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.994832377542755e-05, | |
| "loss": 0.8217, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9936215093023884e-05, | |
| "loss": 0.8921, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9922837513195406e-05, | |
| "loss": 0.9066, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.990819274431662e-05, | |
| "loss": 0.8279, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.989228265658754e-05, | |
| "loss": 0.8751, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9875109281794828e-05, | |
| "loss": 0.8937, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9856674813052345e-05, | |
| "loss": 0.8608, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9836981604521077e-05, | |
| "loss": 0.8593, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.98160321711085e-05, | |
| "loss": 0.9067, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9793829188147406e-05, | |
| "loss": 0.8843, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9770375491054264e-05, | |
| "loss": 0.815, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.974567407496712e-05, | |
| "loss": 0.8244, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9719728094363103e-05, | |
| "loss": 0.8525, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9692540862655587e-05, | |
| "loss": 0.7885, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.966411585177105e-05, | |
| "loss": 0.8094, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9634456691705705e-05, | |
| "loss": 0.8224, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9603567170061918e-05, | |
| "loss": 0.8554, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9571451231564523e-05, | |
| "loss": 0.8125, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9571451231564523e-05, | |
| "loss": 0.7916, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9538112977557077e-05, | |
| "loss": 0.8607, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9503556665478066e-05, | |
| "loss": 0.8381, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9467786708317257e-05, | |
| "loss": 0.8083, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9430807674052092e-05, | |
| "loss": 0.8169, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.939262428506438e-05, | |
| "loss": 0.7989, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9353241417537216e-05, | |
| "loss": 0.8313, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9312664100832236e-05, | |
| "loss": 0.8218, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9270897516847406e-05, | |
| "loss": 0.8098, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9227946999355226e-05, | |
| "loss": 0.806, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9183818033321612e-05, | |
| "loss": 0.8, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9138516254205416e-05, | |
| "loss": 0.8299, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9092047447238775e-05, | |
| "loss": 0.841, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9044417546688295e-05, | |
| "loss": 0.8153, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.899563263509725e-05, | |
| "loss": 0.8235, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.894569894250877e-05, | |
| "loss": 0.8133, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8894622845670282e-05, | |
| "loss": 0.827, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8842410867219137e-05, | |
| "loss": 0.8848, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.878906967484966e-05, | |
| "loss": 0.8284, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8734606080461657e-05, | |
| "loss": 0.8398, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.86790270392905e-05, | |
| "loss": 0.8236, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8622339649018907e-05, | |
| "loss": 0.8271, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.856455114887056e-05, | |
| "loss": 0.8735, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8505668918685603e-05, | |
| "loss": 0.7842, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8445700477978207e-05, | |
| "loss": 0.7925, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8384653484976305e-05, | |
| "loss": 0.7709, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.8322535735643604e-05, | |
| "loss": 0.7795, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.8259355162684e-05, | |
| "loss": 0.8165, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.8195119834528535e-05, | |
| "loss": 0.7843, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.8129837954305033e-05, | |
| "loss": 0.7773, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.8063517858790517e-05, | |
| "loss": 0.813, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.799616801734657e-05, | |
| "loss": 0.7894, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.792779703083777e-05, | |
| "loss": 0.795, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7858413630533305e-05, | |
| "loss": 0.8139, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.778802667699196e-05, | |
| "loss": 0.814, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.77166451589306e-05, | |
| "loss": 0.7781, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.764427819207624e-05, | |
| "loss": 0.7791, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.757093501800196e-05, | |
| "loss": 0.7927, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.7496625002946702e-05, | |
| "loss": 0.764, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.7421357636619153e-05, | |
| "loss": 0.8185, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.734514253098589e-05, | |
| "loss": 0.7295, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.726798941904386e-05, | |
| "loss": 0.7698, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.7189908153577473e-05, | |
| "loss": 0.7951, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.7110908705900322e-05, | |
| "loss": 0.894, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.7031001164581828e-05, | |
| "loss": 0.8151, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6950195734158874e-05, | |
| "loss": 0.7179, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6868502733832647e-05, | |
| "loss": 0.7924, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6785932596150827e-05, | |
| "loss": 0.7925, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.670249586567531e-05, | |
| "loss": 0.7784, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6618203197635624e-05, | |
| "loss": 0.7663, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6533065356568206e-05, | |
| "loss": 0.874, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6447093214941727e-05, | |
| "loss": 0.8626, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.636029775176862e-05, | |
| "loss": 0.7891, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.627269005120304e-05, | |
| "loss": 0.8171, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.618428130112533e-05, | |
| "loss": 0.7662, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.6095082791713322e-05, | |
| "loss": 0.8031, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.6005105914000508e-05, | |
| "loss": 0.881, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5914362158421352e-05, | |
| "loss": 0.7381, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5822863113343934e-05, | |
| "loss": 0.749, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.5730620463590052e-05, | |
| "loss": 0.774, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.5637645988943008e-05, | |
| "loss": 0.8499, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.554395156264331e-05, | |
| "loss": 0.7523, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.544954914987238e-05, | |
| "loss": 0.8426, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.5354450806224553e-05, | |
| "loss": 0.8087, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.5258668676167548e-05, | |
| "loss": 0.7308, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.516221499149154e-05, | |
| "loss": 0.8656, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.5065102069747117e-05, | |
| "loss": 0.8439, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4967342312672283e-05, | |
| "loss": 0.8121, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.48689482046087e-05, | |
| "loss": 0.7898, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4769932310907372e-05, | |
| "loss": 0.8082, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.467030727632401e-05, | |
| "loss": 0.7991, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.4570085823404232e-05, | |
| "loss": 0.9332, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.4469280750858854e-05, | |
| "loss": 0.7493, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.4367904931929422e-05, | |
| "loss": 0.818, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.4265971312744252e-05, | |
| "loss": 0.817, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.4163492910665153e-05, | |
| "loss": 0.8618, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.4060482812625055e-05, | |
| "loss": 0.8142, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.395695417345675e-05, | |
| "loss": 0.7781, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.3852920214212966e-05, | |
| "loss": 0.7698, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.3748394220477972e-05, | |
| "loss": 0.7905, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.3643389540670963e-05, | |
| "loss": 0.7785, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.3537919584341413e-05, | |
| "loss": 0.775, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.3431997820456592e-05, | |
| "loss": 0.8126, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.3325637775681561e-05, | |
| "loss": 0.7391, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.3218853032651719e-05, | |
| "loss": 0.7936, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.3111657228238263e-05, | |
| "loss": 0.8208, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.3004064051806712e-05, | |
| "loss": 0.845, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2896087243468673e-05, | |
| "loss": 0.8119, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2787740592327232e-05, | |
| "loss": 0.8512, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.267903793471597e-05, | |
| "loss": 0.8217, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2569993152432028e-05, | |
| "loss": 0.7735, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.2460620170963353e-05, | |
| "loss": 0.845, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.2350932957710322e-05, | |
| "loss": 0.7313, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.2240945520202079e-05, | |
| "loss": 0.7404, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.2130671904307692e-05, | |
| "loss": 0.7769, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.202012619244243e-05, | |
| "loss": 0.7828, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1909322501769407e-05, | |
| "loss": 0.7855, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1798274982396728e-05, | |
| "loss": 0.7496, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1686997815570473e-05, | |
| "loss": 0.7661, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.15755052118637e-05, | |
| "loss": 0.8014, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.1463811409361667e-05, | |
| "loss": 0.7606, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.13519306718436e-05, | |
| "loss": 0.7683, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.1239877286961123e-05, | |
| "loss": 0.7931, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.112766556441367e-05, | |
| "loss": 0.783, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.1015309834121083e-05, | |
| "loss": 0.7784, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.0902824444393602e-05, | |
| "loss": 0.8107, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.079022376009955e-05, | |
| "loss": 0.7373, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.067752216083085e-05, | |
| "loss": 0.777, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.05647340390667e-05, | |
| "loss": 0.8342, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.0451873798335605e-05, | |
| "loss": 0.7869, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.0338955851375962e-05, | |
| "loss": 0.7407, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.0225994618295507e-05, | |
| "loss": 0.744, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.01130045247298e-05, | |
| "loss": 0.795, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8239, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.886995475270205e-06, | |
| "loss": 0.9011, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.774005381704498e-06, | |
| "loss": 0.7928, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.661044148624038e-06, | |
| "loss": 0.8258, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.548126201664398e-06, | |
| "loss": 0.7536, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.435265960933304e-06, | |
| "loss": 0.739, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.322477839169156e-06, | |
| "loss": 0.7535, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.209776239900453e-06, | |
| "loss": 0.8141, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.097175555606396e-06, | |
| "loss": 0.7229, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.98469016587892e-06, | |
| "loss": 0.8472, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.872334435586333e-06, | |
| "loss": 0.7363, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.76012271303888e-06, | |
| "loss": 0.7183, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.648069328156403e-06, | |
| "loss": 0.741, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.536188590638334e-06, | |
| "loss": 0.7531, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.424494788136303e-06, | |
| "loss": 0.8059, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.313002184429529e-06, | |
| "loss": 0.8084, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.201725017603277e-06, | |
| "loss": 0.8038, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.090677498230598e-06, | |
| "loss": 0.7679, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.97987380755757e-06, | |
| "loss": 0.7534, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.869328095692313e-06, | |
| "loss": 0.6547, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.759054479797924e-06, | |
| "loss": 0.752, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.649067042289681e-06, | |
| "loss": 0.7703, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.539379829036652e-06, | |
| "loss": 0.7418, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.430006847567972e-06, | |
| "loss": 0.7699, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.320962065284032e-06, | |
| "loss": 0.7668, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.2122594076727705e-06, | |
| "loss": 0.6996, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.1039127565313285e-06, | |
| "loss": 0.8008, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.995935948193294e-06, | |
| "loss": 0.8206, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.888342771761737e-06, | |
| "loss": 0.789, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.781146967348283e-06, | |
| "loss": 0.7528, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.6743622243184405e-06, | |
| "loss": 0.7225, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.568002179543409e-06, | |
| "loss": 0.8076, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.462080415658591e-06, | |
| "loss": 0.77, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 6.356610459329038e-06, | |
| "loss": 0.7526, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 6.251605779522032e-06, | |
| "loss": 0.8311, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 6.147079785787038e-06, | |
| "loss": 0.8706, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.043045826543254e-06, | |
| "loss": 0.7817, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.93951718737495e-06, | |
| "loss": 0.8314, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.836507089334849e-06, | |
| "loss": 0.7774, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.7340286872557515e-06, | |
| "loss": 0.7791, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.6320950680705826e-06, | |
| "loss": 0.7239, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.530719249141148e-06, | |
| "loss": 0.8481, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.429914176595772e-06, | |
| "loss": 0.7794, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.329692723675994e-06, | |
| "loss": 0.7363, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.230067689092629e-06, | |
| "loss": 0.8312, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.131051795391302e-06, | |
| "loss": 0.7527, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.03265768732772e-06, | |
| "loss": 0.7228, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.934897930252887e-06, | |
| "loss": 0.8204, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.837785008508462e-06, | |
| "loss": 0.7834, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.7413313238324556e-06, | |
| "loss": 0.7664, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.645549193775452e-06, | |
| "loss": 0.7509, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.550450850127626e-06, | |
| "loss": 0.7836, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.4560484373566945e-06, | |
| "loss": 0.75, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.3623540110569935e-06, | |
| "loss": 0.7374, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.26937953640995e-06, | |
| "loss": 0.7208, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.177136886656067e-06, | |
| "loss": 0.7699, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.085637841578652e-06, | |
| "loss": 0.7434, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.9948940859994964e-06, | |
| "loss": 0.7305, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.9049172082866786e-06, | |
| "loss": 0.7795, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.815718698874672e-06, | |
| "loss": 0.7743, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.727309948796963e-06, | |
| "loss": 0.706, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.6397022482313804e-06, | |
| "loss": 0.7155, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.552906785058278e-06, | |
| "loss": 0.7618, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.466934643431795e-06, | |
| "loss": 0.776, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.3817968023643766e-06, | |
| "loss": 0.7556, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.2975041343246937e-06, | |
| "loss": 0.7645, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.214067403849179e-06, | |
| "loss": 0.7757, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.1314972661673572e-06, | |
| "loss": 0.778, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.0498042658411276e-06, | |
| "loss": 0.7783, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.9689988354181742e-06, | |
| "loss": 0.7602, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.8890912940996784e-06, | |
| "loss": 0.7106, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.8100918464225304e-06, | |
| "loss": 0.6951, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.7320105809561415e-06, | |
| "loss": 0.7384, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.654857469014113e-06, | |
| "loss": 0.7446, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.5786423633808487e-06, | |
| "loss": 0.7311, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.5033749970533015e-06, | |
| "loss": 0.734, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.4290649819980404e-06, | |
| "loss": 0.7467, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.3557218079237608e-06, | |
| "loss": 0.7501, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.283354841069403e-06, | |
| "loss": 0.746, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.211973323008041e-06, | |
| "loss": 0.7515, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.1415863694666973e-06, | |
| "loss": 0.7672, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.072202969162234e-06, | |
| "loss": 0.7621, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0038319826534312e-06, | |
| "loss": 0.7432, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.936482141209486e-06, | |
| "loss": 0.8442, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.870162045694971e-06, | |
| "loss": 0.7738, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8048801654714687e-06, | |
| "loss": 0.7054, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7406448373160024e-06, | |
| "loss": 0.6819, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.6774642643563955e-06, | |
| "loss": 0.7322, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.615346515023698e-06, | |
| "loss": 0.7504, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.5542995220217961e-06, | |
| "loss": 0.7721, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.4943310813144006e-06, | |
| "loss": 0.6799, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.4354488511294418e-06, | |
| "loss": 0.7524, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.3776603509810938e-06, | |
| "loss": 0.7943, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.3209729607095022e-06, | |
| "loss": 0.772, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.2653939195383448e-06, | |
| "loss": 0.7735, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.2109303251503434e-06, | |
| "loss": 0.8155, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.1575891327808664e-06, | |
| "loss": 0.7834, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.1053771543297198e-06, | |
| "loss": 0.7563, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.0543010574912305e-06, | |
| "loss": 0.7114, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.0043673649027519e-06, | |
| "loss": 0.8082, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.555824533117064e-07, | |
| "loss": 0.7351, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.079525527612321e-07, | |
| "loss": 0.6883, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.614837457945868e-07, | |
| "loss": 0.7643, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.161819666783888e-07, | |
| "loss": 0.8028, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.720530006447735e-07, | |
| "loss": 0.7222, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.291024831525961e-07, | |
| "loss": 0.7265, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.87335899167767e-07, | |
| "loss": 0.7656, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.467585824627886e-07, | |
| "loss": 0.7666, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.073757149356185e-07, | |
| "loss": 0.7724, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.691923259479093e-07, | |
| "loss": 0.7669, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.322132916827483e-07, | |
| "loss": 0.6936, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.964433345219354e-07, | |
| "loss": 0.7554, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.6188702244292614e-07, | |
| "loss": 0.7711, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.285487684354772e-07, | |
| "loss": 0.7357, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.96432829938086e-07, | |
| "loss": 0.7777, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.6554330829429716e-07, | |
| "loss": 0.805, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.3588414822895097e-07, | |
| "loss": 0.7552, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.0745913734441357e-07, | |
| "loss": 0.7379, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.8027190563689745e-07, | |
| "loss": 0.7398, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.5432592503288e-07, | |
| "loss": 0.7244, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.2962450894573606e-07, | |
| "loss": 0.7314, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.0617081185259512e-07, | |
| "loss": 0.7479, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.8396782889150144e-07, | |
| "loss": 0.7381, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.630183954789233e-07, | |
| "loss": 0.7126, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.4332518694765708e-07, | |
| "loss": 0.7229, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.2489071820517394e-07, | |
| "loss": 0.6955, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.0771734341246121e-07, | |
| "loss": 0.7215, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.180725568338045e-08, | |
| "loss": 0.7687, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.716248680459726e-08, | |
| "loss": 0.7748, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.378490697611761e-08, | |
| "loss": 0.761, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.1676224572452246e-08, | |
| "loss": 0.7746, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.083798592444899e-08, | |
| "loss": 0.7059, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.127157512182288e-08, | |
| "loss": 0.7545, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.2978213836400974e-08, | |
| "loss": 0.7726, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.5958961166104847e-08, | |
| "loss": 0.6931, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0214713499706596e-08, | |
| "loss": 0.7208, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.7462044023515186e-09, | |
| "loss": 0.7379, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.5540045218819256e-09, | |
| "loss": 0.7908, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 293, | |
| "total_flos": 133458268946432.0, | |
| "train_loss": 0.798602851381074, | |
| "train_runtime": 7530.3249, | |
| "train_samples_per_second": 1.866, | |
| "train_steps_per_second": 0.039 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 293, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 133458268946432.0, | |
| "train_batch_size": 12, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |