| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 1.3711, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 1.334, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5e-05, | |
| "loss": 1.3864, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 2.248, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 1.9428, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5513, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.999344418328162e-05, | |
| "loss": 1.3746, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.997377845227576e-05, | |
| "loss": 1.2993, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.994100796397954e-05, | |
| "loss": 2.3052, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.989514131188559e-05, | |
| "loss": 1.2252, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.983619052372848e-05, | |
| "loss": 1.1473, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.97641710583307e-05, | |
| "loss": 1.474, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.967910180154889e-05, | |
| "loss": 1.5391, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.958100506132127e-05, | |
| "loss": 1.0238, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.946990656181781e-05, | |
| "loss": 0.6343, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.934583543669453e-05, | |
| "loss": 1.375, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.920882422145372e-05, | |
| "loss": 0.9674, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.905890884491195e-05, | |
| "loss": 0.6452, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.889612861977853e-05, | |
| "loss": 0.5237, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.872052623234632e-05, | |
| "loss": 1.2983, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.853214773129796e-05, | |
| "loss": 0.8931, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.833104251563056e-05, | |
| "loss": 1.1171, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.811726332170153e-05, | |
| "loss": 1.3356, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.789086620939936e-05, | |
| "loss": 0.9511, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.765191054744305e-05, | |
| "loss": 1.0608, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.740045899781352e-05, | |
| "loss": 1.0137, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.713657749932172e-05, | |
| "loss": 1.0402, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.686033525031719e-05, | |
| "loss": 0.5888, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.657180469054213e-05, | |
| "loss": 1.0759, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.627106148213522e-05, | |
| "loss": 1.1284, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.595818448979061e-05, | |
| "loss": 0.5581, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.563325576007701e-05, | |
| "loss": 0.1291, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.529636049992234e-05, | |
| "loss": 0.5968, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.494758705426978e-05, | |
| "loss": 0.0704, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.458702688291073e-05, | |
| "loss": 0.567, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.421477453650118e-05, | |
| "loss": 0.4475, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.38309276317674e-05, | |
| "loss": 1.4415, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.343558682590756e-05, | |
| "loss": 0.7743, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.302885579019627e-05, | |
| "loss": 0.6328, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.261084118279847e-05, | |
| "loss": 0.2475, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.218165262080023e-05, | |
| "loss": 1.2443, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.174140265146356e-05, | |
| "loss": 0.8085, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.129020672271283e-05, | |
| "loss": 0.7764, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.082818315286055e-05, | |
| "loss": 0.5008, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.035545309958046e-05, | |
| "loss": 0.5582, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.987214052813604e-05, | |
| "loss": 1.2532, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.937837217887273e-05, | |
| "loss": 0.5482, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.887427753398248e-05, | |
| "loss": 1.1794, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.835998878354931e-05, | |
| "loss": 1.0944, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.783564079088477e-05, | |
| "loss": 0.9639, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.73013710571623e-05, | |
| "loss": 0.6401, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.675731968536002e-05, | |
| "loss": 1.1377, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.620362934352109e-05, | |
| "loss": 0.5624, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.564044522734147e-05, | |
| "loss": 1.1549, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.506791502209496e-05, | |
| "loss": 0.7098, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.448618886390522e-05, | |
| "loss": 0.6059, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.389541930037516e-05, | |
| "loss": 0.6728, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.329576125058406e-05, | |
| "loss": 1.0251, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.268737196446264e-05, | |
| "loss": 0.7898, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.2070410981557e-05, | |
| "loss": 0.8504, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.144504008919222e-05, | |
| "loss": 0.5914, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.081142328004637e-05, | |
| "loss": 1.1653, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.016972670914624e-05, | |
| "loss": 1.1441, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.952011865029614e-05, | |
| "loss": 1.1371, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.886276945195099e-05, | |
| "loss": 0.7263, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.819785149254532e-05, | |
| "loss": 0.6626, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.752553913529018e-05, | |
| "loss": 1.0817, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.68460086824492e-05, | |
| "loss": 0.6464, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.61594383291065e-05, | |
| "loss": 0.6011, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.546600811643816e-05, | |
| "loss": 1.0604, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.476589988449939e-05, | |
| "loss": 0.5161, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.405929722454026e-05, | |
| "loss": 0.8083, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.334638543086203e-05, | |
| "loss": 1.0716, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.262735145222696e-05, | |
| "loss": 0.5927, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.190238384283412e-05, | |
| "loss": 1.2094, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.117167271287453e-05, | |
| "loss": 0.5113, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.043540967867782e-05, | |
| "loss": 0.9976, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.969378781246436e-05, | |
| "loss": 1.0694, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.894700159171534e-05, | |
| "loss": 0.0773, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.819524684817438e-05, | |
| "loss": 1.2459, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 6.743872071649411e-05, | |
| "loss": 1.2277, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 6.667762158254104e-05, | |
| "loss": 0.7503, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 6.59121490313722e-05, | |
| "loss": 0.8949, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.514250379489753e-05, | |
| "loss": 1.0714, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.436888769924142e-05, | |
| "loss": 0.764, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.359150361181715e-05, | |
| "loss": 1.0263, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.281055538812861e-05, | |
| "loss": 1.0915, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.202624781831268e-05, | |
| "loss": 1.1416, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.123878657343648e-05, | |
| "loss": 1.0471, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.044837815156377e-05, | |
| "loss": 0.7535, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.9655229823604406e-05, | |
| "loss": 1.1383, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.885954957896115e-05, | |
| "loss": 1.1546, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.8061546070987994e-05, | |
| "loss": 0.6833, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.726142856227452e-05, | |
| "loss": 0.7396, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.645940686977033e-05, | |
| "loss": 1.0028, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.565569130976422e-05, | |
| "loss": 0.7586, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.4850492642732406e-05, | |
| "loss": 1.0252, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.4044022018070214e-05, | |
| "loss": 0.8318, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.3236490918721794e-05, | |
| "loss": 1.4927, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.242811110572242e-05, | |
| "loss": 1.0549, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.1619094562667804e-05, | |
| "loss": 1.1034, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 5.080965344012508e-05, | |
| "loss": 1.1269, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 5e-05, | |
| "loss": 0.5282, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.919034655987493e-05, | |
| "loss": 0.5633, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.838090543733222e-05, | |
| "loss": 0.508, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.7571888894277604e-05, | |
| "loss": 1.0521, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.676350908127822e-05, | |
| "loss": 0.6679, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.59559779819298e-05, | |
| "loss": 0.5803, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.51495073572676e-05, | |
| "loss": 0.6257, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.434430869023579e-05, | |
| "loss": 1.1256, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.35405931302297e-05, | |
| "loss": 1.0836, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.27385714377255e-05, | |
| "loss": 1.0069, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.193845392901201e-05, | |
| "loss": 0.0824, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.114045042103887e-05, | |
| "loss": 0.887, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.0344770176395606e-05, | |
| "loss": 0.9928, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.955162184843625e-05, | |
| "loss": 0.6192, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.876121342656355e-05, | |
| "loss": 0.9845, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.7973752181687335e-05, | |
| "loss": 1.0146, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.718944461187138e-05, | |
| "loss": 0.6376, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.640849638818286e-05, | |
| "loss": 1.1646, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.5631112300758595e-05, | |
| "loss": 0.6622, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.4857496205102474e-05, | |
| "loss": 0.1604, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.408785096862782e-05, | |
| "loss": 0.5158, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.332237841745898e-05, | |
| "loss": 0.6361, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.2561279283505883e-05, | |
| "loss": 1.1773, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.180475315182563e-05, | |
| "loss": 1.1003, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.105299840828466e-05, | |
| "loss": 0.5131, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.0306212187535653e-05, | |
| "loss": 0.9371, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.9564590321322207e-05, | |
| "loss": 0.4125, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.882832728712551e-05, | |
| "loss": 1.034, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.8097616157165883e-05, | |
| "loss": 0.5951, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.737264854777306e-05, | |
| "loss": 0.3212, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.6653614569137968e-05, | |
| "loss": 0.6433, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.5940702775459747e-05, | |
| "loss": 0.7006, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.5234100115500643e-05, | |
| "loss": 1.091, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.4533991883561868e-05, | |
| "loss": 0.4494, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.3840561670893496e-05, | |
| "loss": 0.4309, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.315399131755081e-05, | |
| "loss": 0.9503, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.2474460864709824e-05, | |
| "loss": 1.1667, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.180214850745467e-05, | |
| "loss": 1.1528, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.1137230548049043e-05, | |
| "loss": 0.5514, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.0479881349703883e-05, | |
| "loss": 1.0104, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.983027329085377e-05, | |
| "loss": 1.2217, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.9188576719953633e-05, | |
| "loss": 1.1916, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.8554959910807775e-05, | |
| "loss": 0.5268, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.7929589018443016e-05, | |
| "loss": 0.6524, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.7312628035537387e-05, | |
| "loss": 0.5125, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.6704238749415957e-05, | |
| "loss": 0.577, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.6104580699624837e-05, | |
| "loss": 1.098, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.5513811136094787e-05, | |
| "loss": 0.9055, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.4932084977905042e-05, | |
| "loss": 0.8727, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.4359554772658552e-05, | |
| "loss": 0.9075, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.3796370656478935e-05, | |
| "loss": 1.1641, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.3242680314639993e-05, | |
| "loss": 0.5507, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.2698628942837699e-05, | |
| "loss": 0.5801, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.2164359209115234e-05, | |
| "loss": 1.1929, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.1640011216450691e-05, | |
| "loss": 1.2133, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.1125722466017547e-05, | |
| "loss": 0.6737, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.0621627821127289e-05, | |
| "loss": 1.0575, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.012785947186397e-05, | |
| "loss": 0.5367, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.644546900419533e-06, | |
| "loss": 0.876, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.171816847139448e-06, | |
| "loss": 0.8905, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 8.70979327728718e-06, | |
| "loss": 0.6048, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 8.25859734853645e-06, | |
| "loss": 0.4433, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 7.81834737919978e-06, | |
| "loss": 1.1322, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 7.389158817201542e-06, | |
| "loss": 1.1292, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 6.9711442098037375e-06, | |
| "loss": 0.5969, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 6.564413174092443e-06, | |
| "loss": 0.815, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 6.16907236823262e-06, | |
| "loss": 1.1361, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 5.785225463498828e-06, | |
| "loss": 0.3949, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 5.412973117089287e-06, | |
| "loss": 1.0599, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 5.05241294573024e-06, | |
| "loss": 0.177, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.703639500077656e-06, | |
| "loss": 1.0362, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.366744239922998e-06, | |
| "loss": 0.3497, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.041815510209396e-06, | |
| "loss": 0.9584, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.728938517864794e-06, | |
| "loss": 0.9382, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.4281953094578877e-06, | |
| "loss": 0.9351, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.1396647496828247e-06, | |
| "loss": 1.3268, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.8634225006782865e-06, | |
| "loss": 0.4768, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.5995410021864787e-06, | |
| "loss": 0.0659, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.3480894525569562e-06, | |
| "loss": 0.0854, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.1091337906006482e-06, | |
| "loss": 1.0339, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.8827366782984913e-06, | |
| "loss": 0.7292, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.6689574843694433e-06, | |
| "loss": 1.082, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.4678522687020413e-06, | |
| "loss": 1.2546, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.2794737676536994e-06, | |
| "loss": 1.0967, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.1038713802214717e-06, | |
| "loss": 0.5755, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 9.410911550880475e-07, | |
| "loss": 0.6927, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 7.911757785462881e-07, | |
| "loss": 1.0575, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.54164563305465e-07, | |
| "loss": 0.9652, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 5.300934381821998e-07, | |
| "loss": 0.7263, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.189949386787462e-07, | |
| "loss": 0.4148, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.208981984511195e-07, | |
| "loss": 0.9471, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.3582894166930268e-07, | |
| "loss": 1.0083, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.6380947627153143e-07, | |
| "loss": 0.1093, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.0485868811441757e-07, | |
| "loss": 0.8541, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.899203602046655e-08, | |
| "loss": 1.032, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.6221547724253337e-08, | |
| "loss": 1.1813, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 6.5558167183898955e-09, | |
| "loss": 0.0601, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.1166, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 200, | |
| "total_flos": 3.406531187714294e+17, | |
| "train_loss": 0.8767247783765196, | |
| "train_runtime": 2128.2481, | |
| "train_samples_per_second": 0.376, | |
| "train_steps_per_second": 0.094 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 10000, | |
| "total_flos": 3.406531187714294e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |