diff --git "a/LLaVA/text_augmention/2/trainer_state.json" "b/LLaVA/text_augmention/2/trainer_state.json" deleted file mode 100755--- "a/LLaVA/text_augmention/2/trainer_state.json" +++ /dev/null @@ -1,8880 +0,0 @@ -{ - "best_metric": null, - "best_model_checkpoint": null, - "epoch": 5.0, - "eval_steps": 500, - "global_step": 1475, - "is_hyper_param_search": false, - "is_local_process_zero": true, - "is_world_process_zero": true, - "log_history": [ - { - "epoch": 0.0, - "learning_rate": 4.444444444444445e-07, - "loss": 2.0062, - "step": 1 - }, - { - "epoch": 0.01, - "learning_rate": 8.88888888888889e-07, - "loss": 2.1163, - "step": 2 - }, - { - "epoch": 0.01, - "learning_rate": 1.3333333333333334e-06, - "loss": 2.0676, - "step": 3 - }, - { - "epoch": 0.01, - "learning_rate": 1.777777777777778e-06, - "loss": 2.0428, - "step": 4 - }, - { - "epoch": 0.02, - "learning_rate": 2.222222222222222e-06, - "loss": 2.0071, - "step": 5 - }, - { - "epoch": 0.02, - "learning_rate": 2.666666666666667e-06, - "loss": 1.9834, - "step": 6 - }, - { - "epoch": 0.02, - "learning_rate": 3.1111111111111116e-06, - "loss": 2.0107, - "step": 7 - }, - { - "epoch": 0.03, - "learning_rate": 3.555555555555556e-06, - "loss": 1.9564, - "step": 8 - }, - { - "epoch": 0.03, - "learning_rate": 4.000000000000001e-06, - "loss": 1.9384, - "step": 9 - }, - { - "epoch": 0.03, - "learning_rate": 4.444444444444444e-06, - "loss": 1.8986, - "step": 10 - }, - { - "epoch": 0.04, - "learning_rate": 4.888888888888889e-06, - "loss": 1.8231, - "step": 11 - }, - { - "epoch": 0.04, - "learning_rate": 5.333333333333334e-06, - "loss": 1.8277, - "step": 12 - }, - { - "epoch": 0.04, - "learning_rate": 5.777777777777778e-06, - "loss": 1.7736, - "step": 13 - }, - { - "epoch": 0.05, - "learning_rate": 6.222222222222223e-06, - "loss": 1.7968, - "step": 14 - }, - { - "epoch": 0.05, - "learning_rate": 6.666666666666667e-06, - "loss": 1.8345, - "step": 15 - }, - { - "epoch": 0.05, - "learning_rate": 7.111111111111112e-06, - "loss": 1.8006, - "step": 16 - }, - { - "epoch": 0.06, - "learning_rate": 7.555555555555556e-06, - "loss": 1.7911, - "step": 17 - }, - { - "epoch": 0.06, - "learning_rate": 8.000000000000001e-06, - "loss": 1.7357, - "step": 18 - }, - { - "epoch": 0.06, - "learning_rate": 8.444444444444446e-06, - "loss": 1.7564, - "step": 19 - }, - { - "epoch": 0.07, - "learning_rate": 8.888888888888888e-06, - "loss": 1.7716, - "step": 20 - }, - { - "epoch": 0.07, - "learning_rate": 9.333333333333334e-06, - "loss": 1.6546, - "step": 21 - }, - { - "epoch": 0.07, - "learning_rate": 9.777777777777779e-06, - "loss": 1.6842, - "step": 22 - }, - { - "epoch": 0.08, - "learning_rate": 1.0222222222222223e-05, - "loss": 1.7244, - "step": 23 - }, - { - "epoch": 0.08, - "learning_rate": 1.0666666666666667e-05, - "loss": 1.6798, - "step": 24 - }, - { - "epoch": 0.08, - "learning_rate": 1.1111111111111113e-05, - "loss": 1.6966, - "step": 25 - }, - { - "epoch": 0.09, - "learning_rate": 1.1555555555555556e-05, - "loss": 1.6571, - "step": 26 - }, - { - "epoch": 0.09, - "learning_rate": 1.2e-05, - "loss": 1.656, - "step": 27 - }, - { - "epoch": 0.09, - "learning_rate": 1.2444444444444446e-05, - "loss": 1.6646, - "step": 28 - }, - { - "epoch": 0.1, - "learning_rate": 1.288888888888889e-05, - "loss": 1.6003, - "step": 29 - }, - { - "epoch": 0.1, - "learning_rate": 1.3333333333333333e-05, - "loss": 1.6425, - "step": 30 - }, - { - "epoch": 0.11, - "learning_rate": 1.377777777777778e-05, - "loss": 1.6368, - "step": 31 - }, - { - "epoch": 0.11, - "learning_rate": 1.4222222222222224e-05, - "loss": 1.607, - "step": 32 - }, - { - "epoch": 0.11, - "learning_rate": 1.4666666666666666e-05, - "loss": 1.5804, - "step": 33 - }, - { - "epoch": 0.12, - "learning_rate": 1.5111111111111112e-05, - "loss": 1.5997, - "step": 34 - }, - { - "epoch": 0.12, - "learning_rate": 1.555555555555556e-05, - "loss": 1.5804, - "step": 35 - }, - { - "epoch": 0.12, - "learning_rate": 1.6000000000000003e-05, - "loss": 1.5807, - "step": 36 - }, - { - "epoch": 0.13, - "learning_rate": 1.6444444444444444e-05, - "loss": 1.6291, - "step": 37 - }, - { - "epoch": 0.13, - "learning_rate": 1.688888888888889e-05, - "loss": 1.5867, - "step": 38 - }, - { - "epoch": 0.13, - "learning_rate": 1.7333333333333336e-05, - "loss": 1.5732, - "step": 39 - }, - { - "epoch": 0.14, - "learning_rate": 1.7777777777777777e-05, - "loss": 1.5915, - "step": 40 - }, - { - "epoch": 0.14, - "learning_rate": 1.8222222222222224e-05, - "loss": 1.6182, - "step": 41 - }, - { - "epoch": 0.14, - "learning_rate": 1.866666666666667e-05, - "loss": 1.5594, - "step": 42 - }, - { - "epoch": 0.15, - "learning_rate": 1.9111111111111113e-05, - "loss": 1.6067, - "step": 43 - }, - { - "epoch": 0.15, - "learning_rate": 1.9555555555555557e-05, - "loss": 1.5946, - "step": 44 - }, - { - "epoch": 0.15, - "learning_rate": 2e-05, - "loss": 1.6166, - "step": 45 - }, - { - "epoch": 0.16, - "learning_rate": 1.999997586776754e-05, - "loss": 1.5883, - "step": 46 - }, - { - "epoch": 0.16, - "learning_rate": 1.9999903471186634e-05, - "loss": 1.5457, - "step": 47 - }, - { - "epoch": 0.16, - "learning_rate": 1.9999782810606697e-05, - "loss": 1.5317, - "step": 48 - }, - { - "epoch": 0.17, - "learning_rate": 1.9999613886610097e-05, - "loss": 1.5255, - "step": 49 - }, - { - "epoch": 0.17, - "learning_rate": 1.9999396700012127e-05, - "loss": 1.521, - "step": 50 - }, - { - "epoch": 0.17, - "learning_rate": 1.9999131251861037e-05, - "loss": 1.5829, - "step": 51 - }, - { - "epoch": 0.18, - "learning_rate": 1.999881754343799e-05, - "loss": 1.5769, - "step": 52 - }, - { - "epoch": 0.18, - "learning_rate": 1.999845557625709e-05, - "loss": 1.5923, - "step": 53 - }, - { - "epoch": 0.18, - "learning_rate": 1.9998045352065342e-05, - "loss": 1.534, - "step": 54 - }, - { - "epoch": 0.19, - "learning_rate": 1.9997586872842683e-05, - "loss": 1.5258, - "step": 55 - }, - { - "epoch": 0.19, - "learning_rate": 1.9997080140801932e-05, - "loss": 1.5375, - "step": 56 - }, - { - "epoch": 0.19, - "learning_rate": 1.9996525158388804e-05, - "loss": 1.5352, - "step": 57 - }, - { - "epoch": 0.2, - "learning_rate": 1.9995921928281893e-05, - "loss": 1.472, - "step": 58 - }, - { - "epoch": 0.2, - "learning_rate": 1.999527045339266e-05, - "loss": 1.495, - "step": 59 - }, - { - "epoch": 0.2, - "learning_rate": 1.9994570736865407e-05, - "loss": 1.4784, - "step": 60 - }, - { - "epoch": 0.21, - "learning_rate": 1.9993822782077282e-05, - "loss": 1.497, - "step": 61 - }, - { - "epoch": 0.21, - "learning_rate": 1.999302659263825e-05, - "loss": 1.4781, - "step": 62 - }, - { - "epoch": 0.21, - "learning_rate": 1.999218217239108e-05, - "loss": 1.5233, - "step": 63 - }, - { - "epoch": 0.22, - "learning_rate": 1.9991289525411308e-05, - "loss": 1.5066, - "step": 64 - }, - { - "epoch": 0.22, - "learning_rate": 1.999034865600726e-05, - "loss": 1.5117, - "step": 65 - }, - { - "epoch": 0.22, - "learning_rate": 1.9989359568719988e-05, - "loss": 1.488, - "step": 66 - }, - { - "epoch": 0.23, - "learning_rate": 1.998832226832327e-05, - "loss": 1.5254, - "step": 67 - }, - { - "epoch": 0.23, - "learning_rate": 1.9987236759823573e-05, - "loss": 1.5089, - "step": 68 - }, - { - "epoch": 0.23, - "learning_rate": 1.9986103048460056e-05, - "loss": 1.5109, - "step": 69 - }, - { - "epoch": 0.24, - "learning_rate": 1.998492113970451e-05, - "loss": 1.4985, - "step": 70 - }, - { - "epoch": 0.24, - "learning_rate": 1.9983691039261358e-05, - "loss": 1.4718, - "step": 71 - }, - { - "epoch": 0.24, - "learning_rate": 1.998241275306761e-05, - "loss": 1.4964, - "step": 72 - }, - { - "epoch": 0.25, - "learning_rate": 1.9981086287292853e-05, - "loss": 1.4554, - "step": 73 - }, - { - "epoch": 0.25, - "learning_rate": 1.9979711648339192e-05, - "loss": 1.4822, - "step": 74 - }, - { - "epoch": 0.25, - "learning_rate": 1.997828884284126e-05, - "loss": 1.4467, - "step": 75 - }, - { - "epoch": 0.26, - "learning_rate": 1.9976817877666143e-05, - "loss": 1.3985, - "step": 76 - }, - { - "epoch": 0.26, - "learning_rate": 1.9975298759913382e-05, - "loss": 1.4838, - "step": 77 - }, - { - "epoch": 0.26, - "learning_rate": 1.9973731496914914e-05, - "loss": 1.5225, - "step": 78 - }, - { - "epoch": 0.27, - "learning_rate": 1.9972116096235047e-05, - "loss": 1.4701, - "step": 79 - }, - { - "epoch": 0.27, - "learning_rate": 1.9970452565670432e-05, - "loss": 1.4671, - "step": 80 - }, - { - "epoch": 0.27, - "learning_rate": 1.996874091325001e-05, - "loss": 1.4761, - "step": 81 - }, - { - "epoch": 0.28, - "learning_rate": 1.9966981147234975e-05, - "loss": 1.4852, - "step": 82 - }, - { - "epoch": 0.28, - "learning_rate": 1.9965173276118747e-05, - "loss": 1.4404, - "step": 83 - }, - { - "epoch": 0.28, - "learning_rate": 1.9963317308626916e-05, - "loss": 1.518, - "step": 84 - }, - { - "epoch": 0.29, - "learning_rate": 1.9961413253717214e-05, - "loss": 1.4429, - "step": 85 - }, - { - "epoch": 0.29, - "learning_rate": 1.9959461120579457e-05, - "loss": 1.3691, - "step": 86 - }, - { - "epoch": 0.29, - "learning_rate": 1.9957460918635513e-05, - "loss": 1.411, - "step": 87 - }, - { - "epoch": 0.3, - "learning_rate": 1.995541265753925e-05, - "loss": 1.4051, - "step": 88 - }, - { - "epoch": 0.3, - "learning_rate": 1.995331634717649e-05, - "loss": 1.4474, - "step": 89 - }, - { - "epoch": 0.31, - "learning_rate": 1.995117199766496e-05, - "loss": 1.4626, - "step": 90 - }, - { - "epoch": 0.31, - "learning_rate": 1.9948979619354253e-05, - "loss": 1.3975, - "step": 91 - }, - { - "epoch": 0.31, - "learning_rate": 1.994673922282576e-05, - "loss": 1.4355, - "step": 92 - }, - { - "epoch": 0.32, - "learning_rate": 1.994445081889264e-05, - "loss": 1.3808, - "step": 93 - }, - { - "epoch": 0.32, - "learning_rate": 1.994211441859975e-05, - "loss": 1.4312, - "step": 94 - }, - { - "epoch": 0.32, - "learning_rate": 1.99397300332236e-05, - "loss": 1.3892, - "step": 95 - }, - { - "epoch": 0.33, - "learning_rate": 1.9937297674272302e-05, - "loss": 1.362, - "step": 96 - }, - { - "epoch": 0.33, - "learning_rate": 1.99348173534855e-05, - "loss": 1.3932, - "step": 97 - }, - { - "epoch": 0.33, - "learning_rate": 1.9932289082834342e-05, - "loss": 1.4252, - "step": 98 - }, - { - "epoch": 0.34, - "learning_rate": 1.9929712874521375e-05, - "loss": 1.4457, - "step": 99 - }, - { - "epoch": 0.34, - "learning_rate": 1.992708874098054e-05, - "loss": 1.4276, - "step": 100 - }, - { - "epoch": 0.34, - "learning_rate": 1.9924416694877077e-05, - "loss": 1.3817, - "step": 101 - }, - { - "epoch": 0.35, - "learning_rate": 1.992169674910747e-05, - "loss": 1.4091, - "step": 102 - }, - { - "epoch": 0.35, - "learning_rate": 1.9918928916799395e-05, - "loss": 1.3773, - "step": 103 - }, - { - "epoch": 0.35, - "learning_rate": 1.9916113211311647e-05, - "loss": 1.3813, - "step": 104 - }, - { - "epoch": 0.36, - "learning_rate": 1.9913249646234072e-05, - "loss": 1.4594, - "step": 105 - }, - { - "epoch": 0.36, - "learning_rate": 1.9910338235387517e-05, - "loss": 1.3849, - "step": 106 - }, - { - "epoch": 0.36, - "learning_rate": 1.9907378992823755e-05, - "loss": 1.3684, - "step": 107 - }, - { - "epoch": 0.37, - "learning_rate": 1.9904371932825407e-05, - "loss": 1.4027, - "step": 108 - }, - { - "epoch": 0.37, - "learning_rate": 1.990131706990589e-05, - "loss": 1.3902, - "step": 109 - }, - { - "epoch": 0.37, - "learning_rate": 1.989821441880933e-05, - "loss": 1.3502, - "step": 110 - }, - { - "epoch": 0.38, - "learning_rate": 1.9895063994510512e-05, - "loss": 1.3576, - "step": 111 - }, - { - "epoch": 0.38, - "learning_rate": 1.9891865812214793e-05, - "loss": 1.3827, - "step": 112 - }, - { - "epoch": 0.38, - "learning_rate": 1.9888619887358024e-05, - "loss": 1.3901, - "step": 113 - }, - { - "epoch": 0.39, - "learning_rate": 1.988532623560649e-05, - "loss": 1.3338, - "step": 114 - }, - { - "epoch": 0.39, - "learning_rate": 1.988198487285682e-05, - "loss": 1.3797, - "step": 115 - }, - { - "epoch": 0.39, - "learning_rate": 1.987859581523593e-05, - "loss": 1.3316, - "step": 116 - }, - { - "epoch": 0.4, - "learning_rate": 1.9875159079100917e-05, - "loss": 1.3499, - "step": 117 - }, - { - "epoch": 0.4, - "learning_rate": 1.9871674681039013e-05, - "loss": 1.3013, - "step": 118 - }, - { - "epoch": 0.4, - "learning_rate": 1.9868142637867474e-05, - "loss": 1.3837, - "step": 119 - }, - { - "epoch": 0.41, - "learning_rate": 1.9864562966633517e-05, - "loss": 1.363, - "step": 120 - }, - { - "epoch": 0.41, - "learning_rate": 1.9860935684614235e-05, - "loss": 1.3804, - "step": 121 - }, - { - "epoch": 0.41, - "learning_rate": 1.985726080931651e-05, - "loss": 1.3706, - "step": 122 - }, - { - "epoch": 0.42, - "learning_rate": 1.9853538358476933e-05, - "loss": 1.3095, - "step": 123 - }, - { - "epoch": 0.42, - "learning_rate": 1.984976835006171e-05, - "loss": 1.2888, - "step": 124 - }, - { - "epoch": 0.42, - "learning_rate": 1.9845950802266584e-05, - "loss": 1.3517, - "step": 125 - }, - { - "epoch": 0.43, - "learning_rate": 1.9842085733516753e-05, - "loss": 1.3224, - "step": 126 - }, - { - "epoch": 0.43, - "learning_rate": 1.983817316246676e-05, - "loss": 1.3442, - "step": 127 - }, - { - "epoch": 0.43, - "learning_rate": 1.9834213108000414e-05, - "loss": 1.3859, - "step": 128 - }, - { - "epoch": 0.44, - "learning_rate": 1.9830205589230713e-05, - "loss": 1.321, - "step": 129 - }, - { - "epoch": 0.44, - "learning_rate": 1.9826150625499733e-05, - "loss": 1.3494, - "step": 130 - }, - { - "epoch": 0.44, - "learning_rate": 1.9822048236378536e-05, - "loss": 1.287, - "step": 131 - }, - { - "epoch": 0.45, - "learning_rate": 1.9817898441667082e-05, - "loss": 1.3149, - "step": 132 - }, - { - "epoch": 0.45, - "learning_rate": 1.9813701261394136e-05, - "loss": 1.3033, - "step": 133 - }, - { - "epoch": 0.45, - "learning_rate": 1.9809456715817163e-05, - "loss": 1.3605, - "step": 134 - }, - { - "epoch": 0.46, - "learning_rate": 1.980516482542224e-05, - "loss": 1.2911, - "step": 135 - }, - { - "epoch": 0.46, - "learning_rate": 1.9800825610923937e-05, - "loss": 1.338, - "step": 136 - }, - { - "epoch": 0.46, - "learning_rate": 1.9796439093265245e-05, - "loss": 1.2991, - "step": 137 - }, - { - "epoch": 0.47, - "learning_rate": 1.979200529361746e-05, - "loss": 1.3225, - "step": 138 - }, - { - "epoch": 0.47, - "learning_rate": 1.9787524233380076e-05, - "loss": 1.3249, - "step": 139 - }, - { - "epoch": 0.47, - "learning_rate": 1.9782995934180687e-05, - "loss": 1.3159, - "step": 140 - }, - { - "epoch": 0.48, - "learning_rate": 1.9778420417874894e-05, - "loss": 1.2342, - "step": 141 - }, - { - "epoch": 0.48, - "learning_rate": 1.9773797706546176e-05, - "loss": 1.2517, - "step": 142 - }, - { - "epoch": 0.48, - "learning_rate": 1.9769127822505805e-05, - "loss": 1.2726, - "step": 143 - }, - { - "epoch": 0.49, - "learning_rate": 1.9764410788292724e-05, - "loss": 1.3744, - "step": 144 - }, - { - "epoch": 0.49, - "learning_rate": 1.9759646626673445e-05, - "loss": 1.2745, - "step": 145 - }, - { - "epoch": 0.49, - "learning_rate": 1.9754835360641945e-05, - "loss": 1.3011, - "step": 146 - }, - { - "epoch": 0.5, - "learning_rate": 1.9749977013419536e-05, - "loss": 1.2741, - "step": 147 - }, - { - "epoch": 0.5, - "learning_rate": 1.9745071608454777e-05, - "loss": 1.2579, - "step": 148 - }, - { - "epoch": 0.51, - "learning_rate": 1.9740119169423337e-05, - "loss": 1.3241, - "step": 149 - }, - { - "epoch": 0.51, - "learning_rate": 1.97351197202279e-05, - "loss": 1.3209, - "step": 150 - }, - { - "epoch": 0.51, - "learning_rate": 1.973007328499804e-05, - "loss": 1.2672, - "step": 151 - }, - { - "epoch": 0.52, - "learning_rate": 1.972497988809011e-05, - "loss": 1.2575, - "step": 152 - }, - { - "epoch": 0.52, - "learning_rate": 1.9719839554087108e-05, - "loss": 1.2332, - "step": 153 - }, - { - "epoch": 0.52, - "learning_rate": 1.971465230779859e-05, - "loss": 1.2199, - "step": 154 - }, - { - "epoch": 0.53, - "learning_rate": 1.9709418174260523e-05, - "loss": 1.2798, - "step": 155 - }, - { - "epoch": 0.53, - "learning_rate": 1.9704137178735164e-05, - "loss": 1.3037, - "step": 156 - }, - { - "epoch": 0.53, - "learning_rate": 1.9698809346710965e-05, - "loss": 1.2093, - "step": 157 - }, - { - "epoch": 0.54, - "learning_rate": 1.9693434703902417e-05, - "loss": 1.3282, - "step": 158 - }, - { - "epoch": 0.54, - "learning_rate": 1.9688013276249947e-05, - "loss": 1.2727, - "step": 159 - }, - { - "epoch": 0.54, - "learning_rate": 1.9682545089919784e-05, - "loss": 1.258, - "step": 160 - }, - { - "epoch": 0.55, - "learning_rate": 1.9677030171303842e-05, - "loss": 1.3024, - "step": 161 - }, - { - "epoch": 0.55, - "learning_rate": 1.9671468547019575e-05, - "loss": 1.2703, - "step": 162 - }, - { - "epoch": 0.55, - "learning_rate": 1.966586024390986e-05, - "loss": 1.2392, - "step": 163 - }, - { - "epoch": 0.56, - "learning_rate": 1.9660205289042887e-05, - "loss": 1.3063, - "step": 164 - }, - { - "epoch": 0.56, - "learning_rate": 1.9654503709711984e-05, - "loss": 1.1974, - "step": 165 - }, - { - "epoch": 0.56, - "learning_rate": 1.9648755533435517e-05, - "loss": 1.2569, - "step": 166 - }, - { - "epoch": 0.57, - "learning_rate": 1.964296078795675e-05, - "loss": 1.2074, - "step": 167 - }, - { - "epoch": 0.57, - "learning_rate": 1.963711950124372e-05, - "loss": 1.2476, - "step": 168 - }, - { - "epoch": 0.57, - "learning_rate": 1.9631231701489083e-05, - "loss": 1.2487, - "step": 169 - }, - { - "epoch": 0.58, - "learning_rate": 1.9625297417109982e-05, - "loss": 1.2503, - "step": 170 - }, - { - "epoch": 0.58, - "learning_rate": 1.9619316676747928e-05, - "loss": 1.2256, - "step": 171 - }, - { - "epoch": 0.58, - "learning_rate": 1.9613289509268647e-05, - "loss": 1.239, - "step": 172 - }, - { - "epoch": 0.59, - "learning_rate": 1.9607215943761933e-05, - "loss": 1.2047, - "step": 173 - }, - { - "epoch": 0.59, - "learning_rate": 1.960109600954153e-05, - "loss": 1.2187, - "step": 174 - }, - { - "epoch": 0.59, - "learning_rate": 1.9594929736144978e-05, - "loss": 1.2475, - "step": 175 - }, - { - "epoch": 0.6, - "learning_rate": 1.9588717153333456e-05, - "loss": 1.2239, - "step": 176 - }, - { - "epoch": 0.6, - "learning_rate": 1.9582458291091664e-05, - "loss": 1.2095, - "step": 177 - }, - { - "epoch": 0.6, - "learning_rate": 1.957615317962767e-05, - "loss": 1.2643, - "step": 178 - }, - { - "epoch": 0.61, - "learning_rate": 1.9569801849372757e-05, - "loss": 1.2429, - "step": 179 - }, - { - "epoch": 0.61, - "learning_rate": 1.9563404330981276e-05, - "loss": 1.2308, - "step": 180 - }, - { - "epoch": 0.61, - "learning_rate": 1.9556960655330512e-05, - "loss": 1.1902, - "step": 181 - }, - { - "epoch": 0.62, - "learning_rate": 1.955047085352052e-05, - "loss": 1.1985, - "step": 182 - }, - { - "epoch": 0.62, - "learning_rate": 1.954393495687398e-05, - "loss": 1.2143, - "step": 183 - }, - { - "epoch": 0.62, - "learning_rate": 1.9537352996936046e-05, - "loss": 1.1882, - "step": 184 - }, - { - "epoch": 0.63, - "learning_rate": 1.9530725005474195e-05, - "loss": 1.1892, - "step": 185 - }, - { - "epoch": 0.63, - "learning_rate": 1.9524051014478078e-05, - "loss": 1.2489, - "step": 186 - }, - { - "epoch": 0.63, - "learning_rate": 1.9517331056159353e-05, - "loss": 1.1247, - "step": 187 - }, - { - "epoch": 0.64, - "learning_rate": 1.9510565162951538e-05, - "loss": 1.1524, - "step": 188 - }, - { - "epoch": 0.64, - "learning_rate": 1.9503753367509855e-05, - "loss": 1.1487, - "step": 189 - }, - { - "epoch": 0.64, - "learning_rate": 1.949689570271107e-05, - "loss": 1.1452, - "step": 190 - }, - { - "epoch": 0.65, - "learning_rate": 1.9489992201653337e-05, - "loss": 1.1414, - "step": 191 - }, - { - "epoch": 0.65, - "learning_rate": 1.9483042897656032e-05, - "loss": 1.1512, - "step": 192 - }, - { - "epoch": 0.65, - "learning_rate": 1.9476047824259602e-05, - "loss": 1.1686, - "step": 193 - }, - { - "epoch": 0.66, - "learning_rate": 1.946900701522539e-05, - "loss": 1.1615, - "step": 194 - }, - { - "epoch": 0.66, - "learning_rate": 1.946192050453549e-05, - "loss": 1.1298, - "step": 195 - }, - { - "epoch": 0.66, - "learning_rate": 1.9454788326392558e-05, - "loss": 1.1324, - "step": 196 - }, - { - "epoch": 0.67, - "learning_rate": 1.944761051521968e-05, - "loss": 1.2683, - "step": 197 - }, - { - "epoch": 0.67, - "learning_rate": 1.944038710566017e-05, - "loss": 1.177, - "step": 198 - }, - { - "epoch": 0.67, - "learning_rate": 1.9433118132577432e-05, - "loss": 1.1371, - "step": 199 - }, - { - "epoch": 0.68, - "learning_rate": 1.9425803631054773e-05, - "loss": 1.1523, - "step": 200 - }, - { - "epoch": 0.68, - "learning_rate": 1.941844363639525e-05, - "loss": 1.1983, - "step": 201 - }, - { - "epoch": 0.68, - "learning_rate": 1.941103818412147e-05, - "loss": 1.1456, - "step": 202 - }, - { - "epoch": 0.69, - "learning_rate": 1.9403587309975467e-05, - "loss": 1.161, - "step": 203 - }, - { - "epoch": 0.69, - "learning_rate": 1.9396091049918478e-05, - "loss": 1.2068, - "step": 204 - }, - { - "epoch": 0.69, - "learning_rate": 1.93885494401308e-05, - "loss": 1.2629, - "step": 205 - }, - { - "epoch": 0.7, - "learning_rate": 1.9380962517011614e-05, - "loss": 1.1606, - "step": 206 - }, - { - "epoch": 0.7, - "learning_rate": 1.9373330317178797e-05, - "loss": 1.0766, - "step": 207 - }, - { - "epoch": 0.71, - "learning_rate": 1.9365652877468747e-05, - "loss": 1.1331, - "step": 208 - }, - { - "epoch": 0.71, - "learning_rate": 1.9357930234936228e-05, - "loss": 1.1454, - "step": 209 - }, - { - "epoch": 0.71, - "learning_rate": 1.9350162426854152e-05, - "loss": 1.1342, - "step": 210 - }, - { - "epoch": 0.72, - "learning_rate": 1.9342349490713427e-05, - "loss": 1.1042, - "step": 211 - }, - { - "epoch": 0.72, - "learning_rate": 1.933449146422278e-05, - "loss": 1.1235, - "step": 212 - }, - { - "epoch": 0.72, - "learning_rate": 1.932658838530855e-05, - "loss": 1.1172, - "step": 213 - }, - { - "epoch": 0.73, - "learning_rate": 1.9318640292114526e-05, - "loss": 1.1251, - "step": 214 - }, - { - "epoch": 0.73, - "learning_rate": 1.9310647223001752e-05, - "loss": 1.0935, - "step": 215 - }, - { - "epoch": 0.73, - "learning_rate": 1.9302609216548352e-05, - "loss": 1.0804, - "step": 216 - }, - { - "epoch": 0.74, - "learning_rate": 1.929452631154933e-05, - "loss": 1.1533, - "step": 217 - }, - { - "epoch": 0.74, - "learning_rate": 1.9286398547016398e-05, - "loss": 1.1265, - "step": 218 - }, - { - "epoch": 0.74, - "learning_rate": 1.9278225962177776e-05, - "loss": 1.0824, - "step": 219 - }, - { - "epoch": 0.75, - "learning_rate": 1.9270008596478008e-05, - "loss": 1.0924, - "step": 220 - }, - { - "epoch": 0.75, - "learning_rate": 1.9261746489577767e-05, - "loss": 1.1053, - "step": 221 - }, - { - "epoch": 0.75, - "learning_rate": 1.9253439681353673e-05, - "loss": 1.1316, - "step": 222 - }, - { - "epoch": 0.76, - "learning_rate": 1.9245088211898086e-05, - "loss": 1.0886, - "step": 223 - }, - { - "epoch": 0.76, - "learning_rate": 1.9236692121518934e-05, - "loss": 1.1089, - "step": 224 - }, - { - "epoch": 0.76, - "learning_rate": 1.9228251450739495e-05, - "loss": 1.0796, - "step": 225 - }, - { - "epoch": 0.77, - "learning_rate": 1.921976624029821e-05, - "loss": 1.1229, - "step": 226 - }, - { - "epoch": 0.77, - "learning_rate": 1.92112365311485e-05, - "loss": 1.1545, - "step": 227 - }, - { - "epoch": 0.77, - "learning_rate": 1.920266236445855e-05, - "loss": 1.1277, - "step": 228 - }, - { - "epoch": 0.78, - "learning_rate": 1.919404378161111e-05, - "loss": 1.0889, - "step": 229 - }, - { - "epoch": 0.78, - "learning_rate": 1.918538082420332e-05, - "loss": 1.1454, - "step": 230 - }, - { - "epoch": 0.78, - "learning_rate": 1.9176673534046465e-05, - "loss": 0.9981, - "step": 231 - }, - { - "epoch": 0.79, - "learning_rate": 1.9167921953165827e-05, - "loss": 1.0854, - "step": 232 - }, - { - "epoch": 0.79, - "learning_rate": 1.9159126123800437e-05, - "loss": 1.0537, - "step": 233 - }, - { - "epoch": 0.79, - "learning_rate": 1.9150286088402898e-05, - "loss": 1.0426, - "step": 234 - }, - { - "epoch": 0.8, - "learning_rate": 1.9141401889639167e-05, - "loss": 1.0671, - "step": 235 - }, - { - "epoch": 0.8, - "learning_rate": 1.9132473570388354e-05, - "loss": 1.1121, - "step": 236 - }, - { - "epoch": 0.8, - "learning_rate": 1.9123501173742514e-05, - "loss": 1.1043, - "step": 237 - }, - { - "epoch": 0.81, - "learning_rate": 1.911448474300644e-05, - "loss": 1.0639, - "step": 238 - }, - { - "epoch": 0.81, - "learning_rate": 1.910542432169745e-05, - "loss": 1.0932, - "step": 239 - }, - { - "epoch": 0.81, - "learning_rate": 1.9096319953545186e-05, - "loss": 1.1092, - "step": 240 - }, - { - "epoch": 0.82, - "learning_rate": 1.908717168249139e-05, - "loss": 1.1152, - "step": 241 - }, - { - "epoch": 0.82, - "learning_rate": 1.9077979552689708e-05, - "loss": 1.0238, - "step": 242 - }, - { - "epoch": 0.82, - "learning_rate": 1.9068743608505454e-05, - "loss": 1.1107, - "step": 243 - }, - { - "epoch": 0.83, - "learning_rate": 1.9059463894515427e-05, - "loss": 1.0607, - "step": 244 - }, - { - "epoch": 0.83, - "learning_rate": 1.905014045550767e-05, - "loss": 1.1252, - "step": 245 - }, - { - "epoch": 0.83, - "learning_rate": 1.904077333648126e-05, - "loss": 1.049, - "step": 246 - }, - { - "epoch": 0.84, - "learning_rate": 1.903136258264609e-05, - "loss": 1.0101, - "step": 247 - }, - { - "epoch": 0.84, - "learning_rate": 1.9021908239422665e-05, - "loss": 1.0361, - "step": 248 - }, - { - "epoch": 0.84, - "learning_rate": 1.9012410352441866e-05, - "loss": 1.1218, - "step": 249 - }, - { - "epoch": 0.85, - "learning_rate": 1.9002868967544743e-05, - "loss": 1.0832, - "step": 250 - }, - { - "epoch": 0.85, - "learning_rate": 1.899328413078227e-05, - "loss": 1.0628, - "step": 251 - }, - { - "epoch": 0.85, - "learning_rate": 1.898365588841515e-05, - "loss": 0.9965, - "step": 252 - }, - { - "epoch": 0.86, - "learning_rate": 1.8973984286913584e-05, - "loss": 1.0602, - "step": 253 - }, - { - "epoch": 0.86, - "learning_rate": 1.896426937295704e-05, - "loss": 1.0364, - "step": 254 - }, - { - "epoch": 0.86, - "learning_rate": 1.8954511193434024e-05, - "loss": 1.1074, - "step": 255 - }, - { - "epoch": 0.87, - "learning_rate": 1.8944709795441874e-05, - "loss": 1.0293, - "step": 256 - }, - { - "epoch": 0.87, - "learning_rate": 1.8934865226286507e-05, - "loss": 1.0202, - "step": 257 - }, - { - "epoch": 0.87, - "learning_rate": 1.892497753348221e-05, - "loss": 1.0194, - "step": 258 - }, - { - "epoch": 0.88, - "learning_rate": 1.891504676475141e-05, - "loss": 1.06, - "step": 259 - }, - { - "epoch": 0.88, - "learning_rate": 1.8905072968024424e-05, - "loss": 1.0867, - "step": 260 - }, - { - "epoch": 0.88, - "learning_rate": 1.8895056191439252e-05, - "loss": 1.004, - "step": 261 - }, - { - "epoch": 0.89, - "learning_rate": 1.888499648334133e-05, - "loss": 1.0414, - "step": 262 - }, - { - "epoch": 0.89, - "learning_rate": 1.8874893892283296e-05, - "loss": 1.0281, - "step": 263 - }, - { - "epoch": 0.89, - "learning_rate": 1.886474846702477e-05, - "loss": 1.022, - "step": 264 - }, - { - "epoch": 0.9, - "learning_rate": 1.8854560256532098e-05, - "loss": 0.9762, - "step": 265 - }, - { - "epoch": 0.9, - "learning_rate": 1.8844329309978146e-05, - "loss": 0.9825, - "step": 266 - }, - { - "epoch": 0.91, - "learning_rate": 1.8834055676742018e-05, - "loss": 1.0496, - "step": 267 - }, - { - "epoch": 0.91, - "learning_rate": 1.8823739406408855e-05, - "loss": 1.0346, - "step": 268 - }, - { - "epoch": 0.91, - "learning_rate": 1.8813380548769594e-05, - "loss": 1.0791, - "step": 269 - }, - { - "epoch": 0.92, - "learning_rate": 1.88029791538207e-05, - "loss": 1.0589, - "step": 270 - }, - { - "epoch": 0.92, - "learning_rate": 1.8792535271763944e-05, - "loss": 1.0257, - "step": 271 - }, - { - "epoch": 0.92, - "learning_rate": 1.8782048953006176e-05, - "loss": 0.9571, - "step": 272 - }, - { - "epoch": 0.93, - "learning_rate": 1.8771520248159044e-05, - "loss": 1.0126, - "step": 273 - }, - { - "epoch": 0.93, - "learning_rate": 1.8760949208038782e-05, - "loss": 1.0058, - "step": 274 - }, - { - "epoch": 0.93, - "learning_rate": 1.8750335883665948e-05, - "loss": 0.9504, - "step": 275 - }, - { - "epoch": 0.94, - "learning_rate": 1.873968032626518e-05, - "loss": 1.0416, - "step": 276 - }, - { - "epoch": 0.94, - "learning_rate": 1.8728982587264965e-05, - "loss": 0.9966, - "step": 277 - }, - { - "epoch": 0.94, - "learning_rate": 1.8718242718297358e-05, - "loss": 0.9941, - "step": 278 - }, - { - "epoch": 0.95, - "learning_rate": 1.8707460771197773e-05, - "loss": 0.9734, - "step": 279 - }, - { - "epoch": 0.95, - "learning_rate": 1.8696636798004693e-05, - "loss": 0.9301, - "step": 280 - }, - { - "epoch": 0.95, - "learning_rate": 1.8685770850959444e-05, - "loss": 0.9928, - "step": 281 - }, - { - "epoch": 0.96, - "learning_rate": 1.8674862982505946e-05, - "loss": 0.9124, - "step": 282 - }, - { - "epoch": 0.96, - "learning_rate": 1.8663913245290433e-05, - "loss": 1.0034, - "step": 283 - }, - { - "epoch": 0.96, - "learning_rate": 1.8652921692161233e-05, - "loss": 0.9096, - "step": 284 - }, - { - "epoch": 0.97, - "learning_rate": 1.8641888376168483e-05, - "loss": 0.9138, - "step": 285 - }, - { - "epoch": 0.97, - "learning_rate": 1.8630813350563898e-05, - "loss": 1.0206, - "step": 286 - }, - { - "epoch": 0.97, - "learning_rate": 1.8619696668800494e-05, - "loss": 0.9181, - "step": 287 - }, - { - "epoch": 0.98, - "learning_rate": 1.860853838453234e-05, - "loss": 0.9952, - "step": 288 - }, - { - "epoch": 0.98, - "learning_rate": 1.85973385516143e-05, - "loss": 0.9674, - "step": 289 - }, - { - "epoch": 0.98, - "learning_rate": 1.8586097224101767e-05, - "loss": 1.0001, - "step": 290 - }, - { - "epoch": 0.99, - "learning_rate": 1.8574814456250406e-05, - "loss": 0.9524, - "step": 291 - }, - { - "epoch": 0.99, - "learning_rate": 1.856349030251589e-05, - "loss": 0.9039, - "step": 292 - }, - { - "epoch": 0.99, - "learning_rate": 1.855212481755365e-05, - "loss": 0.9838, - "step": 293 - }, - { - "epoch": 1.0, - "learning_rate": 1.854071805621858e-05, - "loss": 0.9493, - "step": 294 - }, - { - "epoch": 1.0, - "learning_rate": 1.852927007356481e-05, - "loss": 0.9276, - "step": 295 - }, - { - "epoch": 1.0, - "learning_rate": 1.8517780924845415e-05, - "loss": 0.6039, - "step": 296 - }, - { - "epoch": 1.01, - "learning_rate": 1.8506250665512156e-05, - "loss": 0.6049, - "step": 297 - }, - { - "epoch": 1.01, - "learning_rate": 1.8494679351215212e-05, - "loss": 0.6009, - "step": 298 - }, - { - "epoch": 1.01, - "learning_rate": 1.848306703780291e-05, - "loss": 0.5639, - "step": 299 - }, - { - "epoch": 1.02, - "learning_rate": 1.847141378132147e-05, - "loss": 0.5836, - "step": 300 - }, - { - "epoch": 1.02, - "learning_rate": 1.8459719638014693e-05, - "loss": 0.5774, - "step": 301 - }, - { - "epoch": 1.02, - "learning_rate": 1.844798466432375e-05, - "loss": 0.5712, - "step": 302 - }, - { - "epoch": 1.03, - "learning_rate": 1.843620891688686e-05, - "loss": 0.6007, - "step": 303 - }, - { - "epoch": 1.03, - "learning_rate": 1.842439245253904e-05, - "loss": 0.6161, - "step": 304 - }, - { - "epoch": 1.03, - "learning_rate": 1.8412535328311813e-05, - "loss": 0.5793, - "step": 305 - }, - { - "epoch": 1.04, - "learning_rate": 1.840063760143297e-05, - "loss": 0.6075, - "step": 306 - }, - { - "epoch": 1.04, - "learning_rate": 1.8388699329326237e-05, - "loss": 0.5691, - "step": 307 - }, - { - "epoch": 1.04, - "learning_rate": 1.8376720569611057e-05, - "loss": 0.575, - "step": 308 - }, - { - "epoch": 1.05, - "learning_rate": 1.8364701380102267e-05, - "loss": 0.5826, - "step": 309 - }, - { - "epoch": 1.05, - "learning_rate": 1.835264181880985e-05, - "loss": 0.5927, - "step": 310 - }, - { - "epoch": 1.05, - "learning_rate": 1.8340541943938623e-05, - "loss": 0.5579, - "step": 311 - }, - { - "epoch": 1.06, - "learning_rate": 1.8328401813887994e-05, - "loss": 0.5605, - "step": 312 - }, - { - "epoch": 1.06, - "learning_rate": 1.831622148725164e-05, - "loss": 0.5859, - "step": 313 - }, - { - "epoch": 1.06, - "learning_rate": 1.8304001022817267e-05, - "loss": 0.5536, - "step": 314 - }, - { - "epoch": 1.07, - "learning_rate": 1.8291740479566286e-05, - "loss": 0.5639, - "step": 315 - }, - { - "epoch": 1.07, - "learning_rate": 1.8279439916673553e-05, - "loss": 0.6251, - "step": 316 - }, - { - "epoch": 1.07, - "learning_rate": 1.8267099393507083e-05, - "loss": 0.5586, - "step": 317 - }, - { - "epoch": 1.08, - "learning_rate": 1.825471896962774e-05, - "loss": 0.5879, - "step": 318 - }, - { - "epoch": 1.08, - "learning_rate": 1.8242298704788988e-05, - "loss": 0.6132, - "step": 319 - }, - { - "epoch": 1.08, - "learning_rate": 1.8229838658936566e-05, - "loss": 0.5929, - "step": 320 - }, - { - "epoch": 1.09, - "learning_rate": 1.821733889220822e-05, - "loss": 0.5901, - "step": 321 - }, - { - "epoch": 1.09, - "learning_rate": 1.82047994649334e-05, - "loss": 0.6531, - "step": 322 - }, - { - "epoch": 1.09, - "learning_rate": 1.819222043763299e-05, - "loss": 0.5478, - "step": 323 - }, - { - "epoch": 1.1, - "learning_rate": 1.8179601871018983e-05, - "loss": 0.5703, - "step": 324 - }, - { - "epoch": 1.1, - "learning_rate": 1.816694382599422e-05, - "loss": 0.5654, - "step": 325 - }, - { - "epoch": 1.11, - "learning_rate": 1.815424636365208e-05, - "loss": 0.5563, - "step": 326 - }, - { - "epoch": 1.11, - "learning_rate": 1.814150954527618e-05, - "loss": 0.5631, - "step": 327 - }, - { - "epoch": 1.11, - "learning_rate": 1.8128733432340095e-05, - "loss": 0.5728, - "step": 328 - }, - { - "epoch": 1.12, - "learning_rate": 1.8115918086507054e-05, - "loss": 0.6115, - "step": 329 - }, - { - "epoch": 1.12, - "learning_rate": 1.8103063569629635e-05, - "loss": 0.6014, - "step": 330 - }, - { - "epoch": 1.12, - "learning_rate": 1.8090169943749477e-05, - "loss": 0.5588, - "step": 331 - }, - { - "epoch": 1.13, - "learning_rate": 1.8077237271096972e-05, - "loss": 0.5805, - "step": 332 - }, - { - "epoch": 1.13, - "learning_rate": 1.8064265614090976e-05, - "loss": 0.5993, - "step": 333 - }, - { - "epoch": 1.13, - "learning_rate": 1.8051255035338494e-05, - "loss": 0.5282, - "step": 334 - }, - { - "epoch": 1.14, - "learning_rate": 1.8038205597634392e-05, - "loss": 0.5771, - "step": 335 - }, - { - "epoch": 1.14, - "learning_rate": 1.8025117363961083e-05, - "loss": 0.5962, - "step": 336 - }, - { - "epoch": 1.14, - "learning_rate": 1.801199039748822e-05, - "loss": 0.5572, - "step": 337 - }, - { - "epoch": 1.15, - "learning_rate": 1.7998824761572415e-05, - "loss": 0.5663, - "step": 338 - }, - { - "epoch": 1.15, - "learning_rate": 1.7985620519756897e-05, - "loss": 0.568, - "step": 339 - }, - { - "epoch": 1.15, - "learning_rate": 1.7972377735771234e-05, - "loss": 0.5895, - "step": 340 - }, - { - "epoch": 1.16, - "learning_rate": 1.7959096473531016e-05, - "loss": 0.5457, - "step": 341 - }, - { - "epoch": 1.16, - "learning_rate": 1.7945776797137544e-05, - "loss": 0.6028, - "step": 342 - }, - { - "epoch": 1.16, - "learning_rate": 1.7932418770877523e-05, - "loss": 0.6354, - "step": 343 - }, - { - "epoch": 1.17, - "learning_rate": 1.7919022459222754e-05, - "loss": 0.5837, - "step": 344 - }, - { - "epoch": 1.17, - "learning_rate": 1.7905587926829815e-05, - "loss": 0.5694, - "step": 345 - }, - { - "epoch": 1.17, - "learning_rate": 1.7892115238539757e-05, - "loss": 0.5789, - "step": 346 - }, - { - "epoch": 1.18, - "learning_rate": 1.7878604459377795e-05, - "loss": 0.6188, - "step": 347 - }, - { - "epoch": 1.18, - "learning_rate": 1.7865055654552977e-05, - "loss": 0.5655, - "step": 348 - }, - { - "epoch": 1.18, - "learning_rate": 1.7851468889457883e-05, - "loss": 0.5804, - "step": 349 - }, - { - "epoch": 1.19, - "learning_rate": 1.7837844229668312e-05, - "loss": 0.5997, - "step": 350 - }, - { - "epoch": 1.19, - "learning_rate": 1.7824181740942958e-05, - "loss": 0.5823, - "step": 351 - }, - { - "epoch": 1.19, - "learning_rate": 1.7810481489223082e-05, - "loss": 0.5819, - "step": 352 - }, - { - "epoch": 1.2, - "learning_rate": 1.7796743540632226e-05, - "loss": 0.6087, - "step": 353 - }, - { - "epoch": 1.2, - "learning_rate": 1.7782967961475855e-05, - "loss": 0.5732, - "step": 354 - }, - { - "epoch": 1.2, - "learning_rate": 1.776915481824107e-05, - "loss": 0.5494, - "step": 355 - }, - { - "epoch": 1.21, - "learning_rate": 1.775530417759627e-05, - "loss": 0.6098, - "step": 356 - }, - { - "epoch": 1.21, - "learning_rate": 1.7741416106390828e-05, - "loss": 0.5868, - "step": 357 - }, - { - "epoch": 1.21, - "learning_rate": 1.772749067165477e-05, - "loss": 0.5796, - "step": 358 - }, - { - "epoch": 1.22, - "learning_rate": 1.7713527940598473e-05, - "loss": 0.5503, - "step": 359 - }, - { - "epoch": 1.22, - "learning_rate": 1.7699527980612306e-05, - "loss": 0.5471, - "step": 360 - }, - { - "epoch": 1.22, - "learning_rate": 1.7685490859266324e-05, - "loss": 0.5942, - "step": 361 - }, - { - "epoch": 1.23, - "learning_rate": 1.7671416644309945e-05, - "loss": 0.5742, - "step": 362 - }, - { - "epoch": 1.23, - "learning_rate": 1.7657305403671618e-05, - "loss": 0.5866, - "step": 363 - }, - { - "epoch": 1.23, - "learning_rate": 1.7643157205458483e-05, - "loss": 0.6442, - "step": 364 - }, - { - "epoch": 1.24, - "learning_rate": 1.762897211795607e-05, - "loss": 0.5661, - "step": 365 - }, - { - "epoch": 1.24, - "learning_rate": 1.7614750209627938e-05, - "loss": 0.5688, - "step": 366 - }, - { - "epoch": 1.24, - "learning_rate": 1.760049154911537e-05, - "loss": 0.5874, - "step": 367 - }, - { - "epoch": 1.25, - "learning_rate": 1.758619620523703e-05, - "loss": 0.5656, - "step": 368 - }, - { - "epoch": 1.25, - "learning_rate": 1.7571864246988623e-05, - "loss": 0.5356, - "step": 369 - }, - { - "epoch": 1.25, - "learning_rate": 1.7557495743542586e-05, - "loss": 0.6017, - "step": 370 - }, - { - "epoch": 1.26, - "learning_rate": 1.7543090764247726e-05, - "loss": 0.5672, - "step": 371 - }, - { - "epoch": 1.26, - "learning_rate": 1.7528649378628912e-05, - "loss": 0.5879, - "step": 372 - }, - { - "epoch": 1.26, - "learning_rate": 1.751417165638671e-05, - "loss": 0.6135, - "step": 373 - }, - { - "epoch": 1.27, - "learning_rate": 1.7499657667397083e-05, - "loss": 0.5555, - "step": 374 - }, - { - "epoch": 1.27, - "learning_rate": 1.7485107481711014e-05, - "loss": 0.5694, - "step": 375 - }, - { - "epoch": 1.27, - "learning_rate": 1.7470521169554196e-05, - "loss": 0.5627, - "step": 376 - }, - { - "epoch": 1.28, - "learning_rate": 1.7455898801326685e-05, - "loss": 0.5864, - "step": 377 - }, - { - "epoch": 1.28, - "learning_rate": 1.7441240447602565e-05, - "loss": 0.5609, - "step": 378 - }, - { - "epoch": 1.28, - "learning_rate": 1.742654617912958e-05, - "loss": 0.5667, - "step": 379 - }, - { - "epoch": 1.29, - "learning_rate": 1.7411816066828852e-05, - "loss": 0.579, - "step": 380 - }, - { - "epoch": 1.29, - "learning_rate": 1.7397050181794463e-05, - "loss": 0.5748, - "step": 381 - }, - { - "epoch": 1.29, - "learning_rate": 1.7382248595293175e-05, - "loss": 0.5787, - "step": 382 - }, - { - "epoch": 1.3, - "learning_rate": 1.736741137876405e-05, - "loss": 0.5834, - "step": 383 - }, - { - "epoch": 1.3, - "learning_rate": 1.7352538603818124e-05, - "loss": 0.5654, - "step": 384 - }, - { - "epoch": 1.31, - "learning_rate": 1.733763034223804e-05, - "loss": 0.5427, - "step": 385 - }, - { - "epoch": 1.31, - "learning_rate": 1.7322686665977738e-05, - "loss": 0.561, - "step": 386 - }, - { - "epoch": 1.31, - "learning_rate": 1.730770764716206e-05, - "loss": 0.5425, - "step": 387 - }, - { - "epoch": 1.32, - "learning_rate": 1.7292693358086447e-05, - "loss": 0.5378, - "step": 388 - }, - { - "epoch": 1.32, - "learning_rate": 1.7277643871216558e-05, - "loss": 0.5599, - "step": 389 - }, - { - "epoch": 1.32, - "learning_rate": 1.7262559259187936e-05, - "loss": 0.5224, - "step": 390 - }, - { - "epoch": 1.33, - "learning_rate": 1.724743959480565e-05, - "loss": 0.5494, - "step": 391 - }, - { - "epoch": 1.33, - "learning_rate": 1.7232284951043962e-05, - "loss": 0.5812, - "step": 392 - }, - { - "epoch": 1.33, - "learning_rate": 1.721709540104594e-05, - "loss": 0.5803, - "step": 393 - }, - { - "epoch": 1.34, - "learning_rate": 1.720187101812314e-05, - "loss": 0.5658, - "step": 394 - }, - { - "epoch": 1.34, - "learning_rate": 1.7186611875755227e-05, - "loss": 0.5744, - "step": 395 - }, - { - "epoch": 1.34, - "learning_rate": 1.7171318047589637e-05, - "loss": 0.525, - "step": 396 - }, - { - "epoch": 1.35, - "learning_rate": 1.715598960744121e-05, - "loss": 0.5297, - "step": 397 - }, - { - "epoch": 1.35, - "learning_rate": 1.7140626629291853e-05, - "loss": 0.5845, - "step": 398 - }, - { - "epoch": 1.35, - "learning_rate": 1.712522918729014e-05, - "loss": 0.5429, - "step": 399 - }, - { - "epoch": 1.36, - "learning_rate": 1.7109797355751017e-05, - "loss": 0.5884, - "step": 400 - }, - { - "epoch": 1.36, - "learning_rate": 1.7094331209155394e-05, - "loss": 0.5819, - "step": 401 - }, - { - "epoch": 1.36, - "learning_rate": 1.7078830822149784e-05, - "loss": 0.5503, - "step": 402 - }, - { - "epoch": 1.37, - "learning_rate": 1.7063296269545988e-05, - "loss": 0.5726, - "step": 403 - }, - { - "epoch": 1.37, - "learning_rate": 1.7047727626320688e-05, - "loss": 0.5583, - "step": 404 - }, - { - "epoch": 1.37, - "learning_rate": 1.7032124967615112e-05, - "loss": 0.5696, - "step": 405 - }, - { - "epoch": 1.38, - "learning_rate": 1.7016488368734654e-05, - "loss": 0.5691, - "step": 406 - }, - { - "epoch": 1.38, - "learning_rate": 1.7000817905148523e-05, - "loss": 0.5183, - "step": 407 - }, - { - "epoch": 1.38, - "learning_rate": 1.6985113652489374e-05, - "loss": 0.5458, - "step": 408 - }, - { - "epoch": 1.39, - "learning_rate": 1.696937568655294e-05, - "loss": 0.5713, - "step": 409 - }, - { - "epoch": 1.39, - "learning_rate": 1.6953604083297665e-05, - "loss": 0.5261, - "step": 410 - }, - { - "epoch": 1.39, - "learning_rate": 1.6937798918844363e-05, - "loss": 0.5924, - "step": 411 - }, - { - "epoch": 1.4, - "learning_rate": 1.6921960269475806e-05, - "loss": 0.5371, - "step": 412 - }, - { - "epoch": 1.4, - "learning_rate": 1.6906088211636387e-05, - "loss": 0.5421, - "step": 413 - }, - { - "epoch": 1.4, - "learning_rate": 1.6890182821931746e-05, - "loss": 0.55, - "step": 414 - }, - { - "epoch": 1.41, - "learning_rate": 1.6874244177128395e-05, - "loss": 0.5721, - "step": 415 - }, - { - "epoch": 1.41, - "learning_rate": 1.685827235415335e-05, - "loss": 0.5511, - "step": 416 - }, - { - "epoch": 1.41, - "learning_rate": 1.6842267430093762e-05, - "loss": 0.5382, - "step": 417 - }, - { - "epoch": 1.42, - "learning_rate": 1.6826229482196535e-05, - "loss": 0.544, - "step": 418 - }, - { - "epoch": 1.42, - "learning_rate": 1.6810158587867973e-05, - "loss": 0.5523, - "step": 419 - }, - { - "epoch": 1.42, - "learning_rate": 1.679405482467338e-05, - "loss": 0.5371, - "step": 420 - }, - { - "epoch": 1.43, - "learning_rate": 1.6777918270336718e-05, - "loss": 0.5424, - "step": 421 - }, - { - "epoch": 1.43, - "learning_rate": 1.6761749002740195e-05, - "loss": 0.5645, - "step": 422 - }, - { - "epoch": 1.43, - "learning_rate": 1.6745547099923917e-05, - "loss": 0.5525, - "step": 423 - }, - { - "epoch": 1.44, - "learning_rate": 1.6729312640085504e-05, - "loss": 0.5397, - "step": 424 - }, - { - "epoch": 1.44, - "learning_rate": 1.6713045701579705e-05, - "loss": 0.5744, - "step": 425 - }, - { - "epoch": 1.44, - "learning_rate": 1.6696746362918027e-05, - "loss": 0.5761, - "step": 426 - }, - { - "epoch": 1.45, - "learning_rate": 1.6680414702768358e-05, - "loss": 0.552, - "step": 427 - }, - { - "epoch": 1.45, - "learning_rate": 1.6664050799954587e-05, - "loss": 0.4994, - "step": 428 - }, - { - "epoch": 1.45, - "learning_rate": 1.66476547334562e-05, - "loss": 0.5104, - "step": 429 - }, - { - "epoch": 1.46, - "learning_rate": 1.6631226582407954e-05, - "loss": 0.5306, - "step": 430 - }, - { - "epoch": 1.46, - "learning_rate": 1.661476642609943e-05, - "loss": 0.5542, - "step": 431 - }, - { - "epoch": 1.46, - "learning_rate": 1.6598274343974688e-05, - "loss": 0.5211, - "step": 432 - }, - { - "epoch": 1.47, - "learning_rate": 1.658175041563189e-05, - "loss": 0.5458, - "step": 433 - }, - { - "epoch": 1.47, - "learning_rate": 1.6565194720822885e-05, - "loss": 0.5635, - "step": 434 - }, - { - "epoch": 1.47, - "learning_rate": 1.6548607339452853e-05, - "loss": 0.5422, - "step": 435 - }, - { - "epoch": 1.48, - "learning_rate": 1.6531988351579897e-05, - "loss": 0.5728, - "step": 436 - }, - { - "epoch": 1.48, - "learning_rate": 1.6515337837414677e-05, - "loss": 0.5289, - "step": 437 - }, - { - "epoch": 1.48, - "learning_rate": 1.6498655877320008e-05, - "loss": 0.544, - "step": 438 - }, - { - "epoch": 1.49, - "learning_rate": 1.6481942551810476e-05, - "loss": 0.5285, - "step": 439 - }, - { - "epoch": 1.49, - "learning_rate": 1.6465197941552054e-05, - "loss": 0.5561, - "step": 440 - }, - { - "epoch": 1.49, - "learning_rate": 1.6448422127361707e-05, - "loss": 0.5698, - "step": 441 - }, - { - "epoch": 1.5, - "learning_rate": 1.6431615190207003e-05, - "loss": 0.5469, - "step": 442 - }, - { - "epoch": 1.5, - "learning_rate": 1.641477721120573e-05, - "loss": 0.5705, - "step": 443 - }, - { - "epoch": 1.51, - "learning_rate": 1.6397908271625488e-05, - "loss": 0.5406, - "step": 444 - }, - { - "epoch": 1.51, - "learning_rate": 1.638100845288331e-05, - "loss": 0.501, - "step": 445 - }, - { - "epoch": 1.51, - "learning_rate": 1.636407783654527e-05, - "loss": 0.5456, - "step": 446 - }, - { - "epoch": 1.52, - "learning_rate": 1.6347116504326082e-05, - "loss": 0.5441, - "step": 447 - }, - { - "epoch": 1.52, - "learning_rate": 1.6330124538088705e-05, - "loss": 0.5348, - "step": 448 - }, - { - "epoch": 1.52, - "learning_rate": 1.631310201984396e-05, - "loss": 0.5394, - "step": 449 - }, - { - "epoch": 1.53, - "learning_rate": 1.6296049031750114e-05, - "loss": 0.5392, - "step": 450 - }, - { - "epoch": 1.53, - "learning_rate": 1.627896565611251e-05, - "loss": 0.579, - "step": 451 - }, - { - "epoch": 1.53, - "learning_rate": 1.626185197538314e-05, - "loss": 0.5285, - "step": 452 - }, - { - "epoch": 1.54, - "learning_rate": 1.6244708072160267e-05, - "loss": 0.5486, - "step": 453 - }, - { - "epoch": 1.54, - "learning_rate": 1.6227534029188027e-05, - "loss": 0.544, - "step": 454 - }, - { - "epoch": 1.54, - "learning_rate": 1.6210329929356017e-05, - "loss": 0.5521, - "step": 455 - }, - { - "epoch": 1.55, - "learning_rate": 1.619309585569891e-05, - "loss": 0.5514, - "step": 456 - }, - { - "epoch": 1.55, - "learning_rate": 1.6175831891396034e-05, - "loss": 0.5325, - "step": 457 - }, - { - "epoch": 1.55, - "learning_rate": 1.615853811977099e-05, - "loss": 0.5125, - "step": 458 - }, - { - "epoch": 1.56, - "learning_rate": 1.614121462429124e-05, - "loss": 0.5711, - "step": 459 - }, - { - "epoch": 1.56, - "learning_rate": 1.612386148856771e-05, - "loss": 0.5071, - "step": 460 - }, - { - "epoch": 1.56, - "learning_rate": 1.6106478796354382e-05, - "loss": 0.5746, - "step": 461 - }, - { - "epoch": 1.57, - "learning_rate": 1.6089066631547893e-05, - "loss": 0.5424, - "step": 462 - }, - { - "epoch": 1.57, - "learning_rate": 1.6071625078187113e-05, - "loss": 0.5437, - "step": 463 - }, - { - "epoch": 1.57, - "learning_rate": 1.6054154220452776e-05, - "loss": 0.5771, - "step": 464 - }, - { - "epoch": 1.58, - "learning_rate": 1.6036654142667043e-05, - "loss": 0.5236, - "step": 465 - }, - { - "epoch": 1.58, - "learning_rate": 1.6019124929293097e-05, - "loss": 0.5263, - "step": 466 - }, - { - "epoch": 1.58, - "learning_rate": 1.600156666493475e-05, - "loss": 0.5398, - "step": 467 - }, - { - "epoch": 1.59, - "learning_rate": 1.598397943433603e-05, - "loss": 0.5534, - "step": 468 - }, - { - "epoch": 1.59, - "learning_rate": 1.596636332238076e-05, - "loss": 0.52, - "step": 469 - }, - { - "epoch": 1.59, - "learning_rate": 1.5948718414092163e-05, - "loss": 0.5447, - "step": 470 - }, - { - "epoch": 1.6, - "learning_rate": 1.593104479463244e-05, - "loss": 0.5277, - "step": 471 - }, - { - "epoch": 1.6, - "learning_rate": 1.5913342549302378e-05, - "loss": 0.5042, - "step": 472 - }, - { - "epoch": 1.6, - "learning_rate": 1.5895611763540914e-05, - "loss": 0.5387, - "step": 473 - }, - { - "epoch": 1.61, - "learning_rate": 1.5877852522924733e-05, - "loss": 0.5401, - "step": 474 - }, - { - "epoch": 1.61, - "learning_rate": 1.5860064913167863e-05, - "loss": 0.555, - "step": 475 - }, - { - "epoch": 1.61, - "learning_rate": 1.584224902012125e-05, - "loss": 0.535, - "step": 476 - }, - { - "epoch": 1.62, - "learning_rate": 1.5824404929772347e-05, - "loss": 0.5408, - "step": 477 - }, - { - "epoch": 1.62, - "learning_rate": 1.5806532728244707e-05, - "loss": 0.5068, - "step": 478 - }, - { - "epoch": 1.62, - "learning_rate": 1.5788632501797545e-05, - "loss": 0.5368, - "step": 479 - }, - { - "epoch": 1.63, - "learning_rate": 1.5770704336825354e-05, - "loss": 0.4959, - "step": 480 - }, - { - "epoch": 1.63, - "learning_rate": 1.575274831985746e-05, - "loss": 0.5182, - "step": 481 - }, - { - "epoch": 1.63, - "learning_rate": 1.5734764537557617e-05, - "loss": 0.5431, - "step": 482 - }, - { - "epoch": 1.64, - "learning_rate": 1.5716753076723594e-05, - "loss": 0.5475, - "step": 483 - }, - { - "epoch": 1.64, - "learning_rate": 1.5698714024286733e-05, - "loss": 0.5151, - "step": 484 - }, - { - "epoch": 1.64, - "learning_rate": 1.568064746731156e-05, - "loss": 0.5075, - "step": 485 - }, - { - "epoch": 1.65, - "learning_rate": 1.5662553492995347e-05, - "loss": 0.5275, - "step": 486 - }, - { - "epoch": 1.65, - "learning_rate": 1.5644432188667695e-05, - "loss": 0.5053, - "step": 487 - }, - { - "epoch": 1.65, - "learning_rate": 1.5626283641790105e-05, - "loss": 0.5317, - "step": 488 - }, - { - "epoch": 1.66, - "learning_rate": 1.5608107939955565e-05, - "loss": 0.4834, - "step": 489 - }, - { - "epoch": 1.66, - "learning_rate": 1.5589905170888136e-05, - "loss": 0.5277, - "step": 490 - }, - { - "epoch": 1.66, - "learning_rate": 1.5571675422442504e-05, - "loss": 0.4736, - "step": 491 - }, - { - "epoch": 1.67, - "learning_rate": 1.5553418782603574e-05, - "loss": 0.5216, - "step": 492 - }, - { - "epoch": 1.67, - "learning_rate": 1.5535135339486044e-05, - "loss": 0.5005, - "step": 493 - }, - { - "epoch": 1.67, - "learning_rate": 1.5516825181333967e-05, - "loss": 0.5245, - "step": 494 - }, - { - "epoch": 1.68, - "learning_rate": 1.549848839652035e-05, - "loss": 0.5113, - "step": 495 - }, - { - "epoch": 1.68, - "learning_rate": 1.5480125073546705e-05, - "loss": 0.4958, - "step": 496 - }, - { - "epoch": 1.68, - "learning_rate": 1.5461735301042615e-05, - "loss": 0.5181, - "step": 497 - }, - { - "epoch": 1.69, - "learning_rate": 1.5443319167765345e-05, - "loss": 0.516, - "step": 498 - }, - { - "epoch": 1.69, - "learning_rate": 1.542487676259937e-05, - "loss": 0.5231, - "step": 499 - }, - { - "epoch": 1.69, - "learning_rate": 1.5406408174555978e-05, - "loss": 0.5178, - "step": 500 - }, - { - "epoch": 1.7, - "learning_rate": 1.5387913492772816e-05, - "loss": 0.4902, - "step": 501 - }, - { - "epoch": 1.7, - "learning_rate": 1.536939280651348e-05, - "loss": 0.5652, - "step": 502 - }, - { - "epoch": 1.71, - "learning_rate": 1.5350846205167065e-05, - "loss": 0.5512, - "step": 503 - }, - { - "epoch": 1.71, - "learning_rate": 1.533227377824775e-05, - "loss": 0.5095, - "step": 504 - }, - { - "epoch": 1.71, - "learning_rate": 1.5313675615394373e-05, - "loss": 0.5275, - "step": 505 - }, - { - "epoch": 1.72, - "learning_rate": 1.529505180636996e-05, - "loss": 0.4856, - "step": 506 - }, - { - "epoch": 1.72, - "learning_rate": 1.527640244106133e-05, - "loss": 0.5308, - "step": 507 - }, - { - "epoch": 1.72, - "learning_rate": 1.5257727609478647e-05, - "loss": 0.5119, - "step": 508 - }, - { - "epoch": 1.73, - "learning_rate": 1.5239027401754987e-05, - "loss": 0.5395, - "step": 509 - }, - { - "epoch": 1.73, - "learning_rate": 1.5220301908145905e-05, - "loss": 0.4969, - "step": 510 - }, - { - "epoch": 1.73, - "learning_rate": 1.5201551219028988e-05, - "loss": 0.4827, - "step": 511 - }, - { - "epoch": 1.74, - "learning_rate": 1.5182775424903437e-05, - "loss": 0.5061, - "step": 512 - }, - { - "epoch": 1.74, - "learning_rate": 1.5163974616389621e-05, - "loss": 0.509, - "step": 513 - }, - { - "epoch": 1.74, - "learning_rate": 1.514514888422863e-05, - "loss": 0.5285, - "step": 514 - }, - { - "epoch": 1.75, - "learning_rate": 1.5126298319281859e-05, - "loss": 0.4944, - "step": 515 - }, - { - "epoch": 1.75, - "learning_rate": 1.5107423012530546e-05, - "loss": 0.5059, - "step": 516 - }, - { - "epoch": 1.75, - "learning_rate": 1.508852305507535e-05, - "loss": 0.4976, - "step": 517 - }, - { - "epoch": 1.76, - "learning_rate": 1.5069598538135905e-05, - "loss": 0.4888, - "step": 518 - }, - { - "epoch": 1.76, - "learning_rate": 1.5050649553050383e-05, - "loss": 0.4979, - "step": 519 - }, - { - "epoch": 1.76, - "learning_rate": 1.503167619127504e-05, - "loss": 0.4918, - "step": 520 - }, - { - "epoch": 1.77, - "learning_rate": 1.5012678544383797e-05, - "loss": 0.5289, - "step": 521 - }, - { - "epoch": 1.77, - "learning_rate": 1.4993656704067777e-05, - "loss": 0.5224, - "step": 522 - }, - { - "epoch": 1.77, - "learning_rate": 1.4974610762134875e-05, - "loss": 0.5227, - "step": 523 - }, - { - "epoch": 1.78, - "learning_rate": 1.495554081050931e-05, - "loss": 0.5171, - "step": 524 - }, - { - "epoch": 1.78, - "learning_rate": 1.4936446941231186e-05, - "loss": 0.5269, - "step": 525 - }, - { - "epoch": 1.78, - "learning_rate": 1.491732924645604e-05, - "loss": 0.5261, - "step": 526 - }, - { - "epoch": 1.79, - "learning_rate": 1.4898187818454401e-05, - "loss": 0.5047, - "step": 527 - }, - { - "epoch": 1.79, - "learning_rate": 1.4879022749611349e-05, - "loss": 0.4995, - "step": 528 - }, - { - "epoch": 1.79, - "learning_rate": 1.485983413242606e-05, - "loss": 0.529, - "step": 529 - }, - { - "epoch": 1.8, - "learning_rate": 1.4840622059511376e-05, - "loss": 0.5077, - "step": 530 - }, - { - "epoch": 1.8, - "learning_rate": 1.4821386623593332e-05, - "loss": 0.4926, - "step": 531 - }, - { - "epoch": 1.8, - "learning_rate": 1.4802127917510731e-05, - "loss": 0.524, - "step": 532 - }, - { - "epoch": 1.81, - "learning_rate": 1.4782846034214689e-05, - "loss": 0.4617, - "step": 533 - }, - { - "epoch": 1.81, - "learning_rate": 1.476354106676818e-05, - "loss": 0.5225, - "step": 534 - }, - { - "epoch": 1.81, - "learning_rate": 1.4744213108345605e-05, - "loss": 0.5161, - "step": 535 - }, - { - "epoch": 1.82, - "learning_rate": 1.4724862252232313e-05, - "loss": 0.5005, - "step": 536 - }, - { - "epoch": 1.82, - "learning_rate": 1.4705488591824182e-05, - "loss": 0.5428, - "step": 537 - }, - { - "epoch": 1.82, - "learning_rate": 1.4686092220627139e-05, - "loss": 0.5027, - "step": 538 - }, - { - "epoch": 1.83, - "learning_rate": 1.4666673232256738e-05, - "loss": 0.4999, - "step": 539 - }, - { - "epoch": 1.83, - "learning_rate": 1.4647231720437687e-05, - "loss": 0.4784, - "step": 540 - }, - { - "epoch": 1.83, - "learning_rate": 1.4627767779003402e-05, - "loss": 0.502, - "step": 541 - }, - { - "epoch": 1.84, - "learning_rate": 1.4608281501895551e-05, - "loss": 0.4816, - "step": 542 - }, - { - "epoch": 1.84, - "learning_rate": 1.4588772983163612e-05, - "loss": 0.5095, - "step": 543 - }, - { - "epoch": 1.84, - "learning_rate": 1.4569242316964409e-05, - "loss": 0.5141, - "step": 544 - }, - { - "epoch": 1.85, - "learning_rate": 1.4549689597561652e-05, - "loss": 0.5095, - "step": 545 - }, - { - "epoch": 1.85, - "learning_rate": 1.4530114919325498e-05, - "loss": 0.504, - "step": 546 - }, - { - "epoch": 1.85, - "learning_rate": 1.4510518376732081e-05, - "loss": 0.4933, - "step": 547 - }, - { - "epoch": 1.86, - "learning_rate": 1.449090006436307e-05, - "loss": 0.5067, - "step": 548 - }, - { - "epoch": 1.86, - "learning_rate": 1.4471260076905198e-05, - "loss": 0.4989, - "step": 549 - }, - { - "epoch": 1.86, - "learning_rate": 1.4451598509149809e-05, - "loss": 0.5219, - "step": 550 - }, - { - "epoch": 1.87, - "learning_rate": 1.4431915455992416e-05, - "loss": 0.4648, - "step": 551 - }, - { - "epoch": 1.87, - "learning_rate": 1.4412211012432213e-05, - "loss": 0.4914, - "step": 552 - }, - { - "epoch": 1.87, - "learning_rate": 1.4392485273571652e-05, - "loss": 0.4437, - "step": 553 - }, - { - "epoch": 1.88, - "learning_rate": 1.4372738334615947e-05, - "loss": 0.5462, - "step": 554 - }, - { - "epoch": 1.88, - "learning_rate": 1.435297029087265e-05, - "loss": 0.4987, - "step": 555 - }, - { - "epoch": 1.88, - "learning_rate": 1.4333181237751159e-05, - "loss": 0.504, - "step": 556 - }, - { - "epoch": 1.89, - "learning_rate": 1.431337127076229e-05, - "loss": 0.4892, - "step": 557 - }, - { - "epoch": 1.89, - "learning_rate": 1.4293540485517778e-05, - "loss": 0.5334, - "step": 558 - }, - { - "epoch": 1.89, - "learning_rate": 1.4273688977729852e-05, - "loss": 0.4754, - "step": 559 - }, - { - "epoch": 1.9, - "learning_rate": 1.4253816843210751e-05, - "loss": 0.5114, - "step": 560 - }, - { - "epoch": 1.9, - "learning_rate": 1.4233924177872269e-05, - "loss": 0.4848, - "step": 561 - }, - { - "epoch": 1.91, - "learning_rate": 1.4214011077725293e-05, - "loss": 0.4579, - "step": 562 - }, - { - "epoch": 1.91, - "learning_rate": 1.4194077638879333e-05, - "loss": 0.4888, - "step": 563 - }, - { - "epoch": 1.91, - "learning_rate": 1.4174123957542065e-05, - "loss": 0.5123, - "step": 564 - }, - { - "epoch": 1.92, - "learning_rate": 1.4154150130018867e-05, - "loss": 0.4603, - "step": 565 - }, - { - "epoch": 1.92, - "learning_rate": 1.4134156252712343e-05, - "loss": 0.4861, - "step": 566 - }, - { - "epoch": 1.92, - "learning_rate": 1.4114142422121879e-05, - "loss": 0.5048, - "step": 567 - }, - { - "epoch": 1.93, - "learning_rate": 1.4094108734843155e-05, - "loss": 0.4481, - "step": 568 - }, - { - "epoch": 1.93, - "learning_rate": 1.4074055287567685e-05, - "loss": 0.4898, - "step": 569 - }, - { - "epoch": 1.93, - "learning_rate": 1.4053982177082369e-05, - "loss": 0.5118, - "step": 570 - }, - { - "epoch": 1.94, - "learning_rate": 1.4033889500268991e-05, - "loss": 0.4851, - "step": 571 - }, - { - "epoch": 1.94, - "learning_rate": 1.401377735410379e-05, - "loss": 0.4634, - "step": 572 - }, - { - "epoch": 1.94, - "learning_rate": 1.3993645835656955e-05, - "loss": 0.4689, - "step": 573 - }, - { - "epoch": 1.95, - "learning_rate": 1.3973495042092192e-05, - "loss": 0.5102, - "step": 574 - }, - { - "epoch": 1.95, - "learning_rate": 1.3953325070666215e-05, - "loss": 0.4948, - "step": 575 - }, - { - "epoch": 1.95, - "learning_rate": 1.3933136018728324e-05, - "loss": 0.4979, - "step": 576 - }, - { - "epoch": 1.96, - "learning_rate": 1.3912927983719888e-05, - "loss": 0.4977, - "step": 577 - }, - { - "epoch": 1.96, - "learning_rate": 1.3892701063173917e-05, - "loss": 0.4832, - "step": 578 - }, - { - "epoch": 1.96, - "learning_rate": 1.3872455354714552e-05, - "loss": 0.4712, - "step": 579 - }, - { - "epoch": 1.97, - "learning_rate": 1.3852190956056623e-05, - "loss": 0.4994, - "step": 580 - }, - { - "epoch": 1.97, - "learning_rate": 1.3831907965005173e-05, - "loss": 0.4945, - "step": 581 - }, - { - "epoch": 1.97, - "learning_rate": 1.3811606479454961e-05, - "loss": 0.4728, - "step": 582 - }, - { - "epoch": 1.98, - "learning_rate": 1.3791286597390035e-05, - "loss": 0.4995, - "step": 583 - }, - { - "epoch": 1.98, - "learning_rate": 1.3770948416883205e-05, - "loss": 0.4929, - "step": 584 - }, - { - "epoch": 1.98, - "learning_rate": 1.375059203609562e-05, - "loss": 0.4713, - "step": 585 - }, - { - "epoch": 1.99, - "learning_rate": 1.3730217553276257e-05, - "loss": 0.4571, - "step": 586 - }, - { - "epoch": 1.99, - "learning_rate": 1.370982506676147e-05, - "loss": 0.4625, - "step": 587 - }, - { - "epoch": 1.99, - "learning_rate": 1.3689414674974506e-05, - "loss": 0.4983, - "step": 588 - }, - { - "epoch": 2.0, - "learning_rate": 1.3668986476425024e-05, - "loss": 0.5002, - "step": 589 - }, - { - "epoch": 2.0, - "learning_rate": 1.3648540569708637e-05, - "loss": 0.3674, - "step": 590 - }, - { - "epoch": 2.0, - "learning_rate": 1.362807705350641e-05, - "loss": 0.2295, - "step": 591 - }, - { - "epoch": 2.01, - "learning_rate": 1.3607596026584423e-05, - "loss": 0.2268, - "step": 592 - }, - { - "epoch": 2.01, - "learning_rate": 1.3587097587793243e-05, - "loss": 0.2299, - "step": 593 - }, - { - "epoch": 2.01, - "learning_rate": 1.3566581836067495e-05, - "loss": 0.2095, - "step": 594 - }, - { - "epoch": 2.02, - "learning_rate": 1.3546048870425356e-05, - "loss": 0.2089, - "step": 595 - }, - { - "epoch": 2.02, - "learning_rate": 1.3525498789968088e-05, - "loss": 0.208, - "step": 596 - }, - { - "epoch": 2.02, - "learning_rate": 1.3504931693879553e-05, - "loss": 0.1965, - "step": 597 - }, - { - "epoch": 2.03, - "learning_rate": 1.3484347681425739e-05, - "loss": 0.1979, - "step": 598 - }, - { - "epoch": 2.03, - "learning_rate": 1.3463746851954275e-05, - "loss": 0.2079, - "step": 599 - }, - { - "epoch": 2.03, - "learning_rate": 1.3443129304893974e-05, - "loss": 0.2018, - "step": 600 - }, - { - "epoch": 2.04, - "learning_rate": 1.342249513975432e-05, - "loss": 0.2409, - "step": 601 - }, - { - "epoch": 2.04, - "learning_rate": 1.3401844456125002e-05, - "loss": 0.204, - "step": 602 - }, - { - "epoch": 2.04, - "learning_rate": 1.3381177353675441e-05, - "loss": 0.1925, - "step": 603 - }, - { - "epoch": 2.05, - "learning_rate": 1.3360493932154301e-05, - "loss": 0.2057, - "step": 604 - }, - { - "epoch": 2.05, - "learning_rate": 1.3339794291389015e-05, - "loss": 0.2057, - "step": 605 - }, - { - "epoch": 2.05, - "learning_rate": 1.3319078531285286e-05, - "loss": 0.2213, - "step": 606 - }, - { - "epoch": 2.06, - "learning_rate": 1.3298346751826624e-05, - "loss": 0.1901, - "step": 607 - }, - { - "epoch": 2.06, - "learning_rate": 1.3277599053073848e-05, - "loss": 0.2079, - "step": 608 - }, - { - "epoch": 2.06, - "learning_rate": 1.3256835535164622e-05, - "loss": 0.2179, - "step": 609 - }, - { - "epoch": 2.07, - "learning_rate": 1.3236056298312957e-05, - "loss": 0.1957, - "step": 610 - }, - { - "epoch": 2.07, - "learning_rate": 1.3215261442808718e-05, - "loss": 0.2138, - "step": 611 - }, - { - "epoch": 2.07, - "learning_rate": 1.319445106901717e-05, - "loss": 0.2158, - "step": 612 - }, - { - "epoch": 2.08, - "learning_rate": 1.3173625277378464e-05, - "loss": 0.2158, - "step": 613 - }, - { - "epoch": 2.08, - "learning_rate": 1.315278416840717e-05, - "loss": 0.2261, - "step": 614 - }, - { - "epoch": 2.08, - "learning_rate": 1.3131927842691793e-05, - "loss": 0.2062, - "step": 615 - }, - { - "epoch": 2.09, - "learning_rate": 1.311105640089426e-05, - "loss": 0.211, - "step": 616 - }, - { - "epoch": 2.09, - "learning_rate": 1.3090169943749475e-05, - "loss": 0.2198, - "step": 617 - }, - { - "epoch": 2.09, - "learning_rate": 1.3069268572064808e-05, - "loss": 0.2225, - "step": 618 - }, - { - "epoch": 2.1, - "learning_rate": 1.3048352386719605e-05, - "loss": 0.2157, - "step": 619 - }, - { - "epoch": 2.1, - "learning_rate": 1.3027421488664723e-05, - "loss": 0.2235, - "step": 620 - }, - { - "epoch": 2.11, - "learning_rate": 1.3006475978922013e-05, - "loss": 0.2182, - "step": 621 - }, - { - "epoch": 2.11, - "learning_rate": 1.2985515958583865e-05, - "loss": 0.213, - "step": 622 - }, - { - "epoch": 2.11, - "learning_rate": 1.2964541528812689e-05, - "loss": 0.2183, - "step": 623 - }, - { - "epoch": 2.12, - "learning_rate": 1.2943552790840452e-05, - "loss": 0.2143, - "step": 624 - }, - { - "epoch": 2.12, - "learning_rate": 1.2922549845968174e-05, - "loss": 0.2114, - "step": 625 - }, - { - "epoch": 2.12, - "learning_rate": 1.2901532795565444e-05, - "loss": 0.2087, - "step": 626 - }, - { - "epoch": 2.13, - "learning_rate": 1.2880501741069931e-05, - "loss": 0.1909, - "step": 627 - }, - { - "epoch": 2.13, - "learning_rate": 1.2859456783986892e-05, - "loss": 0.2076, - "step": 628 - }, - { - "epoch": 2.13, - "learning_rate": 1.2838398025888695e-05, - "loss": 0.2152, - "step": 629 - }, - { - "epoch": 2.14, - "learning_rate": 1.2817325568414299e-05, - "loss": 0.2044, - "step": 630 - }, - { - "epoch": 2.14, - "learning_rate": 1.2796239513268796e-05, - "loss": 0.2093, - "step": 631 - }, - { - "epoch": 2.14, - "learning_rate": 1.2775139962222905e-05, - "loss": 0.2197, - "step": 632 - }, - { - "epoch": 2.15, - "learning_rate": 1.275402701711248e-05, - "loss": 0.2034, - "step": 633 - }, - { - "epoch": 2.15, - "learning_rate": 1.2732900779838016e-05, - "loss": 0.2107, - "step": 634 - }, - { - "epoch": 2.15, - "learning_rate": 1.2711761352364172e-05, - "loss": 0.2193, - "step": 635 - }, - { - "epoch": 2.16, - "learning_rate": 1.2690608836719261e-05, - "loss": 0.206, - "step": 636 - }, - { - "epoch": 2.16, - "learning_rate": 1.2669443334994768e-05, - "loss": 0.218, - "step": 637 - }, - { - "epoch": 2.16, - "learning_rate": 1.2648264949344858e-05, - "loss": 0.2173, - "step": 638 - }, - { - "epoch": 2.17, - "learning_rate": 1.262707378198587e-05, - "loss": 0.1951, - "step": 639 - }, - { - "epoch": 2.17, - "learning_rate": 1.2605869935195844e-05, - "loss": 0.222, - "step": 640 - }, - { - "epoch": 2.17, - "learning_rate": 1.2584653511314012e-05, - "loss": 0.2104, - "step": 641 - }, - { - "epoch": 2.18, - "learning_rate": 1.2563424612740307e-05, - "loss": 0.1999, - "step": 642 - }, - { - "epoch": 2.18, - "learning_rate": 1.2542183341934873e-05, - "loss": 0.1986, - "step": 643 - }, - { - "epoch": 2.18, - "learning_rate": 1.252092980141756e-05, - "loss": 0.2229, - "step": 644 - }, - { - "epoch": 2.19, - "learning_rate": 1.2499664093767458e-05, - "loss": 0.1983, - "step": 645 - }, - { - "epoch": 2.19, - "learning_rate": 1.2478386321622356e-05, - "loss": 0.2037, - "step": 646 - }, - { - "epoch": 2.19, - "learning_rate": 1.245709658767829e-05, - "loss": 0.2086, - "step": 647 - }, - { - "epoch": 2.2, - "learning_rate": 1.243579499468901e-05, - "loss": 0.2188, - "step": 648 - }, - { - "epoch": 2.2, - "learning_rate": 1.241448164546553e-05, - "loss": 0.2194, - "step": 649 - }, - { - "epoch": 2.2, - "learning_rate": 1.2393156642875579e-05, - "loss": 0.2226, - "step": 650 - }, - { - "epoch": 2.21, - "learning_rate": 1.2371820089843145e-05, - "loss": 0.2217, - "step": 651 - }, - { - "epoch": 2.21, - "learning_rate": 1.2350472089347957e-05, - "loss": 0.2152, - "step": 652 - }, - { - "epoch": 2.21, - "learning_rate": 1.2329112744425e-05, - "loss": 0.2066, - "step": 653 - }, - { - "epoch": 2.22, - "learning_rate": 1.2307742158164012e-05, - "loss": 0.2069, - "step": 654 - }, - { - "epoch": 2.22, - "learning_rate": 1.2286360433708976e-05, - "loss": 0.2104, - "step": 655 - }, - { - "epoch": 2.22, - "learning_rate": 1.2264967674257647e-05, - "loss": 0.2, - "step": 656 - }, - { - "epoch": 2.23, - "learning_rate": 1.2243563983061029e-05, - "loss": 0.2111, - "step": 657 - }, - { - "epoch": 2.23, - "learning_rate": 1.2222149463422898e-05, - "loss": 0.227, - "step": 658 - }, - { - "epoch": 2.23, - "learning_rate": 1.2200724218699284e-05, - "loss": 0.2078, - "step": 659 - }, - { - "epoch": 2.24, - "learning_rate": 1.2179288352297985e-05, - "loss": 0.2232, - "step": 660 - }, - { - "epoch": 2.24, - "learning_rate": 1.2157841967678064e-05, - "loss": 0.2145, - "step": 661 - }, - { - "epoch": 2.24, - "learning_rate": 1.2136385168349345e-05, - "loss": 0.2216, - "step": 662 - }, - { - "epoch": 2.25, - "learning_rate": 1.2114918057871928e-05, - "loss": 0.2032, - "step": 663 - }, - { - "epoch": 2.25, - "learning_rate": 1.2093440739855669e-05, - "loss": 0.215, - "step": 664 - }, - { - "epoch": 2.25, - "learning_rate": 1.2071953317959692e-05, - "loss": 0.1997, - "step": 665 - }, - { - "epoch": 2.26, - "learning_rate": 1.2050455895891893e-05, - "loss": 0.2244, - "step": 666 - }, - { - "epoch": 2.26, - "learning_rate": 1.202894857740843e-05, - "loss": 0.2284, - "step": 667 - }, - { - "epoch": 2.26, - "learning_rate": 1.2007431466313224e-05, - "loss": 0.2108, - "step": 668 - }, - { - "epoch": 2.27, - "learning_rate": 1.1985904666457455e-05, - "loss": 0.2046, - "step": 669 - }, - { - "epoch": 2.27, - "learning_rate": 1.1964368281739078e-05, - "loss": 0.2199, - "step": 670 - }, - { - "epoch": 2.27, - "learning_rate": 1.19428224161023e-05, - "loss": 0.1945, - "step": 671 - }, - { - "epoch": 2.28, - "learning_rate": 1.1921267173537085e-05, - "loss": 0.2173, - "step": 672 - }, - { - "epoch": 2.28, - "learning_rate": 1.1899702658078663e-05, - "loss": 0.2246, - "step": 673 - }, - { - "epoch": 2.28, - "learning_rate": 1.1878128973807005e-05, - "loss": 0.2205, - "step": 674 - }, - { - "epoch": 2.29, - "learning_rate": 1.1856546224846354e-05, - "loss": 0.2249, - "step": 675 - }, - { - "epoch": 2.29, - "learning_rate": 1.183495451536469e-05, - "loss": 0.2272, - "step": 676 - }, - { - "epoch": 2.29, - "learning_rate": 1.181335394957324e-05, - "loss": 0.228, - "step": 677 - }, - { - "epoch": 2.3, - "learning_rate": 1.1791744631725983e-05, - "loss": 0.2138, - "step": 678 - }, - { - "epoch": 2.3, - "learning_rate": 1.1770126666119133e-05, - "loss": 0.2205, - "step": 679 - }, - { - "epoch": 2.31, - "learning_rate": 1.1748500157090645e-05, - "loss": 0.2008, - "step": 680 - }, - { - "epoch": 2.31, - "learning_rate": 1.1726865209019709e-05, - "loss": 0.226, - "step": 681 - }, - { - "epoch": 2.31, - "learning_rate": 1.170522192632624e-05, - "loss": 0.2368, - "step": 682 - }, - { - "epoch": 2.32, - "learning_rate": 1.1683570413470384e-05, - "loss": 0.2143, - "step": 683 - }, - { - "epoch": 2.32, - "learning_rate": 1.1661910774952019e-05, - "loss": 0.236, - "step": 684 - }, - { - "epoch": 2.32, - "learning_rate": 1.1640243115310219e-05, - "loss": 0.2217, - "step": 685 - }, - { - "epoch": 2.33, - "learning_rate": 1.1618567539122794e-05, - "loss": 0.2321, - "step": 686 - }, - { - "epoch": 2.33, - "learning_rate": 1.1596884151005743e-05, - "loss": 0.2265, - "step": 687 - }, - { - "epoch": 2.33, - "learning_rate": 1.1575193055612785e-05, - "loss": 0.2132, - "step": 688 - }, - { - "epoch": 2.34, - "learning_rate": 1.155349435763483e-05, - "loss": 0.2185, - "step": 689 - }, - { - "epoch": 2.34, - "learning_rate": 1.153178816179948e-05, - "loss": 0.2348, - "step": 690 - }, - { - "epoch": 2.34, - "learning_rate": 1.1510074572870533e-05, - "loss": 0.2307, - "step": 691 - }, - { - "epoch": 2.35, - "learning_rate": 1.1488353695647456e-05, - "loss": 0.2066, - "step": 692 - }, - { - "epoch": 2.35, - "learning_rate": 1.1466625634964911e-05, - "loss": 0.2256, - "step": 693 - }, - { - "epoch": 2.35, - "learning_rate": 1.1444890495692214e-05, - "loss": 0.2137, - "step": 694 - }, - { - "epoch": 2.36, - "learning_rate": 1.1423148382732854e-05, - "loss": 0.2234, - "step": 695 - }, - { - "epoch": 2.36, - "learning_rate": 1.1401399401023974e-05, - "loss": 0.2198, - "step": 696 - }, - { - "epoch": 2.36, - "learning_rate": 1.1379643655535869e-05, - "loss": 0.2189, - "step": 697 - }, - { - "epoch": 2.37, - "learning_rate": 1.1357881251271482e-05, - "loss": 0.2165, - "step": 698 - }, - { - "epoch": 2.37, - "learning_rate": 1.1336112293265896e-05, - "loss": 0.2212, - "step": 699 - }, - { - "epoch": 2.37, - "learning_rate": 1.1314336886585818e-05, - "loss": 0.2131, - "step": 700 - }, - { - "epoch": 2.38, - "learning_rate": 1.1292555136329082e-05, - "loss": 0.2404, - "step": 701 - }, - { - "epoch": 2.38, - "learning_rate": 1.1270767147624146e-05, - "loss": 0.2149, - "step": 702 - }, - { - "epoch": 2.38, - "learning_rate": 1.1248973025629567e-05, - "loss": 0.226, - "step": 703 - }, - { - "epoch": 2.39, - "learning_rate": 1.122717287553351e-05, - "loss": 0.2159, - "step": 704 - }, - { - "epoch": 2.39, - "learning_rate": 1.1205366802553231e-05, - "loss": 0.2167, - "step": 705 - }, - { - "epoch": 2.39, - "learning_rate": 1.1183554911934578e-05, - "loss": 0.2221, - "step": 706 - }, - { - "epoch": 2.4, - "learning_rate": 1.1161737308951473e-05, - "loss": 0.2234, - "step": 707 - }, - { - "epoch": 2.4, - "learning_rate": 1.1139914098905406e-05, - "loss": 0.2169, - "step": 708 - }, - { - "epoch": 2.4, - "learning_rate": 1.111808538712494e-05, - "loss": 0.2197, - "step": 709 - }, - { - "epoch": 2.41, - "learning_rate": 1.1096251278965173e-05, - "loss": 0.2297, - "step": 710 - }, - { - "epoch": 2.41, - "learning_rate": 1.1074411879807271e-05, - "loss": 0.2165, - "step": 711 - }, - { - "epoch": 2.41, - "learning_rate": 1.1052567295057921e-05, - "loss": 0.2249, - "step": 712 - }, - { - "epoch": 2.42, - "learning_rate": 1.1030717630148839e-05, - "loss": 0.2146, - "step": 713 - }, - { - "epoch": 2.42, - "learning_rate": 1.1008862990536268e-05, - "loss": 0.2278, - "step": 714 - }, - { - "epoch": 2.42, - "learning_rate": 1.0987003481700456e-05, - "loss": 0.2445, - "step": 715 - }, - { - "epoch": 2.43, - "learning_rate": 1.0965139209145153e-05, - "loss": 0.235, - "step": 716 - }, - { - "epoch": 2.43, - "learning_rate": 1.0943270278397097e-05, - "loss": 0.2254, - "step": 717 - }, - { - "epoch": 2.43, - "learning_rate": 1.0921396795005518e-05, - "loss": 0.2354, - "step": 718 - }, - { - "epoch": 2.44, - "learning_rate": 1.0899518864541607e-05, - "loss": 0.2267, - "step": 719 - }, - { - "epoch": 2.44, - "learning_rate": 1.087763659259803e-05, - "loss": 0.2135, - "step": 720 - }, - { - "epoch": 2.44, - "learning_rate": 1.08557500847884e-05, - "loss": 0.2325, - "step": 721 - }, - { - "epoch": 2.45, - "learning_rate": 1.0833859446746773e-05, - "loss": 0.2261, - "step": 722 - }, - { - "epoch": 2.45, - "learning_rate": 1.0811964784127145e-05, - "loss": 0.2341, - "step": 723 - }, - { - "epoch": 2.45, - "learning_rate": 1.0790066202602931e-05, - "loss": 0.2165, - "step": 724 - }, - { - "epoch": 2.46, - "learning_rate": 1.076816380786647e-05, - "loss": 0.2416, - "step": 725 - }, - { - "epoch": 2.46, - "learning_rate": 1.0746257705628491e-05, - "loss": 0.223, - "step": 726 - }, - { - "epoch": 2.46, - "learning_rate": 1.0724348001617626e-05, - "loss": 0.2196, - "step": 727 - }, - { - "epoch": 2.47, - "learning_rate": 1.0702434801579888e-05, - "loss": 0.2211, - "step": 728 - }, - { - "epoch": 2.47, - "learning_rate": 1.0680518211278169e-05, - "loss": 0.2168, - "step": 729 - }, - { - "epoch": 2.47, - "learning_rate": 1.0658598336491715e-05, - "loss": 0.2318, - "step": 730 - }, - { - "epoch": 2.48, - "learning_rate": 1.063667528301563e-05, - "loss": 0.2269, - "step": 731 - }, - { - "epoch": 2.48, - "learning_rate": 1.0614749156660357e-05, - "loss": 0.2309, - "step": 732 - }, - { - "epoch": 2.48, - "learning_rate": 1.0592820063251177e-05, - "loss": 0.2381, - "step": 733 - }, - { - "epoch": 2.49, - "learning_rate": 1.0570888108627682e-05, - "loss": 0.221, - "step": 734 - }, - { - "epoch": 2.49, - "learning_rate": 1.0548953398643276e-05, - "loss": 0.2285, - "step": 735 - }, - { - "epoch": 2.49, - "learning_rate": 1.0527016039164665e-05, - "loss": 0.2225, - "step": 736 - }, - { - "epoch": 2.5, - "learning_rate": 1.0505076136071342e-05, - "loss": 0.2196, - "step": 737 - }, - { - "epoch": 2.5, - "learning_rate": 1.0483133795255072e-05, - "loss": 0.2324, - "step": 738 - }, - { - "epoch": 2.51, - "learning_rate": 1.0461189122619394e-05, - "loss": 0.2397, - "step": 739 - }, - { - "epoch": 2.51, - "learning_rate": 1.043924222407909e-05, - "loss": 0.2428, - "step": 740 - }, - { - "epoch": 2.51, - "learning_rate": 1.0417293205559694e-05, - "loss": 0.2372, - "step": 741 - }, - { - "epoch": 2.52, - "learning_rate": 1.0395342172996969e-05, - "loss": 0.2306, - "step": 742 - }, - { - "epoch": 2.52, - "learning_rate": 1.0373389232336404e-05, - "loss": 0.2154, - "step": 743 - }, - { - "epoch": 2.52, - "learning_rate": 1.0351434489532685e-05, - "loss": 0.2303, - "step": 744 - }, - { - "epoch": 2.53, - "learning_rate": 1.0329478050549208e-05, - "loss": 0.2182, - "step": 745 - }, - { - "epoch": 2.53, - "learning_rate": 1.0307520021357552e-05, - "loss": 0.2277, - "step": 746 - }, - { - "epoch": 2.53, - "learning_rate": 1.0285560507936962e-05, - "loss": 0.221, - "step": 747 - }, - { - "epoch": 2.54, - "learning_rate": 1.0263599616273868e-05, - "loss": 0.2261, - "step": 748 - }, - { - "epoch": 2.54, - "learning_rate": 1.0241637452361323e-05, - "loss": 0.2362, - "step": 749 - }, - { - "epoch": 2.54, - "learning_rate": 1.0219674122198548e-05, - "loss": 0.2294, - "step": 750 - }, - { - "epoch": 2.55, - "learning_rate": 1.0197709731790375e-05, - "loss": 0.2162, - "step": 751 - }, - { - "epoch": 2.55, - "learning_rate": 1.0175744387146763e-05, - "loss": 0.2376, - "step": 752 - }, - { - "epoch": 2.55, - "learning_rate": 1.0153778194282269e-05, - "loss": 0.2327, - "step": 753 - }, - { - "epoch": 2.56, - "learning_rate": 1.0131811259215545e-05, - "loss": 0.2273, - "step": 754 - }, - { - "epoch": 2.56, - "learning_rate": 1.0109843687968837e-05, - "loss": 0.2202, - "step": 755 - }, - { - "epoch": 2.56, - "learning_rate": 1.0087875586567443e-05, - "loss": 0.2126, - "step": 756 - }, - { - "epoch": 2.57, - "learning_rate": 1.0065907061039234e-05, - "loss": 0.2377, - "step": 757 - }, - { - "epoch": 2.57, - "learning_rate": 1.004393821741412e-05, - "loss": 0.2293, - "step": 758 - }, - { - "epoch": 2.57, - "learning_rate": 1.0021969161723555e-05, - "loss": 0.2272, - "step": 759 - }, - { - "epoch": 2.58, - "learning_rate": 1e-05, - "loss": 0.2208, - "step": 760 - }, - { - "epoch": 2.58, - "learning_rate": 9.97803083827645e-06, - "loss": 0.2246, - "step": 761 - }, - { - "epoch": 2.58, - "learning_rate": 9.956061782585882e-06, - "loss": 0.2167, - "step": 762 - }, - { - "epoch": 2.59, - "learning_rate": 9.934092938960769e-06, - "loss": 0.2521, - "step": 763 - }, - { - "epoch": 2.59, - "learning_rate": 9.912124413432562e-06, - "loss": 0.2367, - "step": 764 - }, - { - "epoch": 2.59, - "learning_rate": 9.890156312031165e-06, - "loss": 0.2401, - "step": 765 - }, - { - "epoch": 2.6, - "learning_rate": 9.868188740784455e-06, - "loss": 0.2266, - "step": 766 - }, - { - "epoch": 2.6, - "learning_rate": 9.846221805717734e-06, - "loss": 0.2462, - "step": 767 - }, - { - "epoch": 2.6, - "learning_rate": 9.82425561285324e-06, - "loss": 0.2233, - "step": 768 - }, - { - "epoch": 2.61, - "learning_rate": 9.802290268209627e-06, - "loss": 0.2162, - "step": 769 - }, - { - "epoch": 2.61, - "learning_rate": 9.780325877801455e-06, - "loss": 0.2258, - "step": 770 - }, - { - "epoch": 2.61, - "learning_rate": 9.75836254763868e-06, - "loss": 0.2373, - "step": 771 - }, - { - "epoch": 2.62, - "learning_rate": 9.736400383726138e-06, - "loss": 0.2255, - "step": 772 - }, - { - "epoch": 2.62, - "learning_rate": 9.71443949206304e-06, - "loss": 0.216, - "step": 773 - }, - { - "epoch": 2.62, - "learning_rate": 9.69247997864245e-06, - "loss": 0.2286, - "step": 774 - }, - { - "epoch": 2.63, - "learning_rate": 9.670521949450793e-06, - "loss": 0.2398, - "step": 775 - }, - { - "epoch": 2.63, - "learning_rate": 9.648565510467316e-06, - "loss": 0.2271, - "step": 776 - }, - { - "epoch": 2.63, - "learning_rate": 9.6266107676636e-06, - "loss": 0.2323, - "step": 777 - }, - { - "epoch": 2.64, - "learning_rate": 9.604657827003033e-06, - "loss": 0.2355, - "step": 778 - }, - { - "epoch": 2.64, - "learning_rate": 9.58270679444031e-06, - "loss": 0.2363, - "step": 779 - }, - { - "epoch": 2.64, - "learning_rate": 9.560757775920917e-06, - "loss": 0.2226, - "step": 780 - }, - { - "epoch": 2.65, - "learning_rate": 9.538810877380611e-06, - "loss": 0.2188, - "step": 781 - }, - { - "epoch": 2.65, - "learning_rate": 9.516866204744932e-06, - "loss": 0.2287, - "step": 782 - }, - { - "epoch": 2.65, - "learning_rate": 9.49492386392866e-06, - "loss": 0.2094, - "step": 783 - }, - { - "epoch": 2.66, - "learning_rate": 9.472983960835338e-06, - "loss": 0.2221, - "step": 784 - }, - { - "epoch": 2.66, - "learning_rate": 9.451046601356725e-06, - "loss": 0.2273, - "step": 785 - }, - { - "epoch": 2.66, - "learning_rate": 9.42911189137232e-06, - "loss": 0.2396, - "step": 786 - }, - { - "epoch": 2.67, - "learning_rate": 9.407179936748827e-06, - "loss": 0.2262, - "step": 787 - }, - { - "epoch": 2.67, - "learning_rate": 9.385250843339644e-06, - "loss": 0.2236, - "step": 788 - }, - { - "epoch": 2.67, - "learning_rate": 9.363324716984375e-06, - "loss": 0.2286, - "step": 789 - }, - { - "epoch": 2.68, - "learning_rate": 9.34140166350829e-06, - "loss": 0.225, - "step": 790 - }, - { - "epoch": 2.68, - "learning_rate": 9.319481788721833e-06, - "loss": 0.2304, - "step": 791 - }, - { - "epoch": 2.68, - "learning_rate": 9.297565198420112e-06, - "loss": 0.2239, - "step": 792 - }, - { - "epoch": 2.69, - "learning_rate": 9.275651998382377e-06, - "loss": 0.2263, - "step": 793 - }, - { - "epoch": 2.69, - "learning_rate": 9.253742294371512e-06, - "loss": 0.2237, - "step": 794 - }, - { - "epoch": 2.69, - "learning_rate": 9.231836192133532e-06, - "loss": 0.2269, - "step": 795 - }, - { - "epoch": 2.7, - "learning_rate": 9.20993379739707e-06, - "loss": 0.2253, - "step": 796 - }, - { - "epoch": 2.7, - "learning_rate": 9.188035215872858e-06, - "loss": 0.2281, - "step": 797 - }, - { - "epoch": 2.71, - "learning_rate": 9.16614055325323e-06, - "loss": 0.2258, - "step": 798 - }, - { - "epoch": 2.71, - "learning_rate": 9.144249915211605e-06, - "loss": 0.2283, - "step": 799 - }, - { - "epoch": 2.71, - "learning_rate": 9.122363407401971e-06, - "loss": 0.2436, - "step": 800 - }, - { - "epoch": 2.72, - "learning_rate": 9.100481135458393e-06, - "loss": 0.2354, - "step": 801 - }, - { - "epoch": 2.72, - "learning_rate": 9.078603204994484e-06, - "loss": 0.2353, - "step": 802 - }, - { - "epoch": 2.72, - "learning_rate": 9.056729721602904e-06, - "loss": 0.2346, - "step": 803 - }, - { - "epoch": 2.73, - "learning_rate": 9.034860790854848e-06, - "loss": 0.2146, - "step": 804 - }, - { - "epoch": 2.73, - "learning_rate": 9.012996518299547e-06, - "loss": 0.2299, - "step": 805 - }, - { - "epoch": 2.73, - "learning_rate": 8.991137009463735e-06, - "loss": 0.2252, - "step": 806 - }, - { - "epoch": 2.74, - "learning_rate": 8.969282369851163e-06, - "loss": 0.2305, - "step": 807 - }, - { - "epoch": 2.74, - "learning_rate": 8.947432704942085e-06, - "loss": 0.2243, - "step": 808 - }, - { - "epoch": 2.74, - "learning_rate": 8.92558812019273e-06, - "loss": 0.235, - "step": 809 - }, - { - "epoch": 2.75, - "learning_rate": 8.903748721034827e-06, - "loss": 0.236, - "step": 810 - }, - { - "epoch": 2.75, - "learning_rate": 8.881914612875062e-06, - "loss": 0.2457, - "step": 811 - }, - { - "epoch": 2.75, - "learning_rate": 8.860085901094595e-06, - "loss": 0.2303, - "step": 812 - }, - { - "epoch": 2.76, - "learning_rate": 8.838262691048529e-06, - "loss": 0.2315, - "step": 813 - }, - { - "epoch": 2.76, - "learning_rate": 8.816445088065425e-06, - "loss": 0.2217, - "step": 814 - }, - { - "epoch": 2.76, - "learning_rate": 8.79463319744677e-06, - "loss": 0.2278, - "step": 815 - }, - { - "epoch": 2.77, - "learning_rate": 8.772827124466495e-06, - "loss": 0.2194, - "step": 816 - }, - { - "epoch": 2.77, - "learning_rate": 8.751026974370438e-06, - "loss": 0.2294, - "step": 817 - }, - { - "epoch": 2.77, - "learning_rate": 8.729232852375855e-06, - "loss": 0.2366, - "step": 818 - }, - { - "epoch": 2.78, - "learning_rate": 8.70744486367092e-06, - "loss": 0.2233, - "step": 819 - }, - { - "epoch": 2.78, - "learning_rate": 8.685663113414186e-06, - "loss": 0.2462, - "step": 820 - }, - { - "epoch": 2.78, - "learning_rate": 8.66388770673411e-06, - "loss": 0.2234, - "step": 821 - }, - { - "epoch": 2.79, - "learning_rate": 8.64211874872852e-06, - "loss": 0.2214, - "step": 822 - }, - { - "epoch": 2.79, - "learning_rate": 8.620356344464135e-06, - "loss": 0.2166, - "step": 823 - }, - { - "epoch": 2.79, - "learning_rate": 8.598600598976033e-06, - "loss": 0.2144, - "step": 824 - }, - { - "epoch": 2.8, - "learning_rate": 8.576851617267151e-06, - "loss": 0.2276, - "step": 825 - }, - { - "epoch": 2.8, - "learning_rate": 8.55510950430779e-06, - "loss": 0.2269, - "step": 826 - }, - { - "epoch": 2.8, - "learning_rate": 8.533374365035089e-06, - "loss": 0.2209, - "step": 827 - }, - { - "epoch": 2.81, - "learning_rate": 8.511646304352544e-06, - "loss": 0.2335, - "step": 828 - }, - { - "epoch": 2.81, - "learning_rate": 8.489925427129469e-06, - "loss": 0.2226, - "step": 829 - }, - { - "epoch": 2.81, - "learning_rate": 8.468211838200523e-06, - "loss": 0.2278, - "step": 830 - }, - { - "epoch": 2.82, - "learning_rate": 8.446505642365174e-06, - "loss": 0.2308, - "step": 831 - }, - { - "epoch": 2.82, - "learning_rate": 8.424806944387219e-06, - "loss": 0.2322, - "step": 832 - }, - { - "epoch": 2.82, - "learning_rate": 8.40311584899426e-06, - "loss": 0.2338, - "step": 833 - }, - { - "epoch": 2.83, - "learning_rate": 8.381432460877213e-06, - "loss": 0.2114, - "step": 834 - }, - { - "epoch": 2.83, - "learning_rate": 8.359756884689785e-06, - "loss": 0.2296, - "step": 835 - }, - { - "epoch": 2.83, - "learning_rate": 8.338089225047983e-06, - "loss": 0.2242, - "step": 836 - }, - { - "epoch": 2.84, - "learning_rate": 8.316429586529616e-06, - "loss": 0.2297, - "step": 837 - }, - { - "epoch": 2.84, - "learning_rate": 8.294778073673762e-06, - "loss": 0.2265, - "step": 838 - }, - { - "epoch": 2.84, - "learning_rate": 8.273134790980295e-06, - "loss": 0.2227, - "step": 839 - }, - { - "epoch": 2.85, - "learning_rate": 8.251499842909358e-06, - "loss": 0.2239, - "step": 840 - }, - { - "epoch": 2.85, - "learning_rate": 8.22987333388087e-06, - "loss": 0.2237, - "step": 841 - }, - { - "epoch": 2.85, - "learning_rate": 8.20825536827402e-06, - "loss": 0.2265, - "step": 842 - }, - { - "epoch": 2.86, - "learning_rate": 8.186646050426763e-06, - "loss": 0.2314, - "step": 843 - }, - { - "epoch": 2.86, - "learning_rate": 8.16504548463531e-06, - "loss": 0.2357, - "step": 844 - }, - { - "epoch": 2.86, - "learning_rate": 8.143453775153646e-06, - "loss": 0.2231, - "step": 845 - }, - { - "epoch": 2.87, - "learning_rate": 8.121871026192996e-06, - "loss": 0.2117, - "step": 846 - }, - { - "epoch": 2.87, - "learning_rate": 8.100297341921342e-06, - "loss": 0.2315, - "step": 847 - }, - { - "epoch": 2.87, - "learning_rate": 8.078732826462917e-06, - "loss": 0.2198, - "step": 848 - }, - { - "epoch": 2.88, - "learning_rate": 8.057177583897704e-06, - "loss": 0.2201, - "step": 849 - }, - { - "epoch": 2.88, - "learning_rate": 8.035631718260923e-06, - "loss": 0.2259, - "step": 850 - }, - { - "epoch": 2.88, - "learning_rate": 8.014095333542548e-06, - "loss": 0.2272, - "step": 851 - }, - { - "epoch": 2.89, - "learning_rate": 7.992568533686782e-06, - "loss": 0.2181, - "step": 852 - }, - { - "epoch": 2.89, - "learning_rate": 7.971051422591571e-06, - "loss": 0.2215, - "step": 853 - }, - { - "epoch": 2.89, - "learning_rate": 7.949544104108107e-06, - "loss": 0.2303, - "step": 854 - }, - { - "epoch": 2.9, - "learning_rate": 7.928046682040311e-06, - "loss": 0.232, - "step": 855 - }, - { - "epoch": 2.9, - "learning_rate": 7.906559260144336e-06, - "loss": 0.2273, - "step": 856 - }, - { - "epoch": 2.91, - "learning_rate": 7.885081942128074e-06, - "loss": 0.2121, - "step": 857 - }, - { - "epoch": 2.91, - "learning_rate": 7.863614831650658e-06, - "loss": 0.2228, - "step": 858 - }, - { - "epoch": 2.91, - "learning_rate": 7.84215803232194e-06, - "loss": 0.2422, - "step": 859 - }, - { - "epoch": 2.92, - "learning_rate": 7.820711647702017e-06, - "loss": 0.2273, - "step": 860 - }, - { - "epoch": 2.92, - "learning_rate": 7.79927578130072e-06, - "loss": 0.2395, - "step": 861 - }, - { - "epoch": 2.92, - "learning_rate": 7.777850536577104e-06, - "loss": 0.2156, - "step": 862 - }, - { - "epoch": 2.93, - "learning_rate": 7.756436016938973e-06, - "loss": 0.2352, - "step": 863 - }, - { - "epoch": 2.93, - "learning_rate": 7.735032325742355e-06, - "loss": 0.2302, - "step": 864 - }, - { - "epoch": 2.93, - "learning_rate": 7.713639566291028e-06, - "loss": 0.2341, - "step": 865 - }, - { - "epoch": 2.94, - "learning_rate": 7.692257841835992e-06, - "loss": 0.2431, - "step": 866 - }, - { - "epoch": 2.94, - "learning_rate": 7.670887255575003e-06, - "loss": 0.2337, - "step": 867 - }, - { - "epoch": 2.94, - "learning_rate": 7.649527910652044e-06, - "loss": 0.218, - "step": 868 - }, - { - "epoch": 2.95, - "learning_rate": 7.628179910156859e-06, - "loss": 0.2299, - "step": 869 - }, - { - "epoch": 2.95, - "learning_rate": 7.606843357124426e-06, - "loss": 0.2308, - "step": 870 - }, - { - "epoch": 2.95, - "learning_rate": 7.585518354534473e-06, - "loss": 0.2269, - "step": 871 - }, - { - "epoch": 2.96, - "learning_rate": 7.56420500531099e-06, - "loss": 0.2269, - "step": 872 - }, - { - "epoch": 2.96, - "learning_rate": 7.542903412321714e-06, - "loss": 0.2264, - "step": 873 - }, - { - "epoch": 2.96, - "learning_rate": 7.521613678377646e-06, - "loss": 0.219, - "step": 874 - }, - { - "epoch": 2.97, - "learning_rate": 7.500335906232544e-06, - "loss": 0.2366, - "step": 875 - }, - { - "epoch": 2.97, - "learning_rate": 7.479070198582441e-06, - "loss": 0.2371, - "step": 876 - }, - { - "epoch": 2.97, - "learning_rate": 7.4578166580651335e-06, - "loss": 0.2369, - "step": 877 - }, - { - "epoch": 2.98, - "learning_rate": 7.436575387259697e-06, - "loss": 0.2171, - "step": 878 - }, - { - "epoch": 2.98, - "learning_rate": 7.4153464886859925e-06, - "loss": 0.218, - "step": 879 - }, - { - "epoch": 2.98, - "learning_rate": 7.394130064804157e-06, - "loss": 0.2341, - "step": 880 - }, - { - "epoch": 2.99, - "learning_rate": 7.372926218014131e-06, - "loss": 0.2267, - "step": 881 - }, - { - "epoch": 2.99, - "learning_rate": 7.3517350506551446e-06, - "loss": 0.2052, - "step": 882 - }, - { - "epoch": 2.99, - "learning_rate": 7.330556665005235e-06, - "loss": 0.2183, - "step": 883 - }, - { - "epoch": 3.0, - "learning_rate": 7.3093911632807415e-06, - "loss": 0.2208, - "step": 884 - }, - { - "epoch": 3.0, - "learning_rate": 7.2882386476358304e-06, - "loss": 0.1735, - "step": 885 - }, - { - "epoch": 3.0, - "learning_rate": 7.267099220161989e-06, - "loss": 0.0956, - "step": 886 - }, - { - "epoch": 3.01, - "learning_rate": 7.2459729828875256e-06, - "loss": 0.0949, - "step": 887 - }, - { - "epoch": 3.01, - "learning_rate": 7.224860037777095e-06, - "loss": 0.0964, - "step": 888 - }, - { - "epoch": 3.01, - "learning_rate": 7.203760486731204e-06, - "loss": 0.0887, - "step": 889 - }, - { - "epoch": 3.02, - "learning_rate": 7.182674431585703e-06, - "loss": 0.0871, - "step": 890 - }, - { - "epoch": 3.02, - "learning_rate": 7.161601974111308e-06, - "loss": 0.0896, - "step": 891 - }, - { - "epoch": 3.02, - "learning_rate": 7.140543216013109e-06, - "loss": 0.0807, - "step": 892 - }, - { - "epoch": 3.03, - "learning_rate": 7.119498258930073e-06, - "loss": 0.0819, - "step": 893 - }, - { - "epoch": 3.03, - "learning_rate": 7.098467204434559e-06, - "loss": 0.0841, - "step": 894 - }, - { - "epoch": 3.03, - "learning_rate": 7.0774501540318305e-06, - "loss": 0.0822, - "step": 895 - }, - { - "epoch": 3.04, - "learning_rate": 7.056447209159552e-06, - "loss": 0.0813, - "step": 896 - }, - { - "epoch": 3.04, - "learning_rate": 7.035458471187312e-06, - "loss": 0.0763, - "step": 897 - }, - { - "epoch": 3.04, - "learning_rate": 7.014484041416137e-06, - "loss": 0.0864, - "step": 898 - }, - { - "epoch": 3.05, - "learning_rate": 6.993524021077989e-06, - "loss": 0.0833, - "step": 899 - }, - { - "epoch": 3.05, - "learning_rate": 6.97257851133528e-06, - "loss": 0.0834, - "step": 900 - }, - { - "epoch": 3.05, - "learning_rate": 6.951647613280397e-06, - "loss": 0.0903, - "step": 901 - }, - { - "epoch": 3.06, - "learning_rate": 6.930731427935196e-06, - "loss": 0.0809, - "step": 902 - }, - { - "epoch": 3.06, - "learning_rate": 6.909830056250527e-06, - "loss": 0.0809, - "step": 903 - }, - { - "epoch": 3.06, - "learning_rate": 6.888943599105745e-06, - "loss": 0.0801, - "step": 904 - }, - { - "epoch": 3.07, - "learning_rate": 6.868072157308213e-06, - "loss": 0.0873, - "step": 905 - }, - { - "epoch": 3.07, - "learning_rate": 6.84721583159283e-06, - "loss": 0.0837, - "step": 906 - }, - { - "epoch": 3.07, - "learning_rate": 6.826374722621536e-06, - "loss": 0.0759, - "step": 907 - }, - { - "epoch": 3.08, - "learning_rate": 6.805548930982832e-06, - "loss": 0.0798, - "step": 908 - }, - { - "epoch": 3.08, - "learning_rate": 6.784738557191284e-06, - "loss": 0.0846, - "step": 909 - }, - { - "epoch": 3.08, - "learning_rate": 6.763943701687046e-06, - "loss": 0.0723, - "step": 910 - }, - { - "epoch": 3.09, - "learning_rate": 6.7431644648353785e-06, - "loss": 0.0872, - "step": 911 - }, - { - "epoch": 3.09, - "learning_rate": 6.7224009469261535e-06, - "loss": 0.0889, - "step": 912 - }, - { - "epoch": 3.09, - "learning_rate": 6.701653248173382e-06, - "loss": 0.0737, - "step": 913 - }, - { - "epoch": 3.1, - "learning_rate": 6.680921468714718e-06, - "loss": 0.0788, - "step": 914 - }, - { - "epoch": 3.1, - "learning_rate": 6.660205708610987e-06, - "loss": 0.089, - "step": 915 - }, - { - "epoch": 3.11, - "learning_rate": 6.639506067845698e-06, - "loss": 0.0821, - "step": 916 - }, - { - "epoch": 3.11, - "learning_rate": 6.618822646324563e-06, - "loss": 0.0838, - "step": 917 - }, - { - "epoch": 3.11, - "learning_rate": 6.598155543875002e-06, - "loss": 0.0841, - "step": 918 - }, - { - "epoch": 3.12, - "learning_rate": 6.577504860245684e-06, - "loss": 0.0873, - "step": 919 - }, - { - "epoch": 3.12, - "learning_rate": 6.556870695106028e-06, - "loss": 0.0861, - "step": 920 - }, - { - "epoch": 3.12, - "learning_rate": 6.536253148045726e-06, - "loss": 0.0761, - "step": 921 - }, - { - "epoch": 3.13, - "learning_rate": 6.515652318574268e-06, - "loss": 0.079, - "step": 922 - }, - { - "epoch": 3.13, - "learning_rate": 6.495068306120452e-06, - "loss": 0.089, - "step": 923 - }, - { - "epoch": 3.13, - "learning_rate": 6.474501210031914e-06, - "loss": 0.0736, - "step": 924 - }, - { - "epoch": 3.14, - "learning_rate": 6.453951129574644e-06, - "loss": 0.0827, - "step": 925 - }, - { - "epoch": 3.14, - "learning_rate": 6.433418163932508e-06, - "loss": 0.0844, - "step": 926 - }, - { - "epoch": 3.14, - "learning_rate": 6.41290241220676e-06, - "loss": 0.0903, - "step": 927 - }, - { - "epoch": 3.15, - "learning_rate": 6.392403973415582e-06, - "loss": 0.0875, - "step": 928 - }, - { - "epoch": 3.15, - "learning_rate": 6.3719229464935915e-06, - "loss": 0.0833, - "step": 929 - }, - { - "epoch": 3.15, - "learning_rate": 6.351459430291369e-06, - "loss": 0.084, - "step": 930 - }, - { - "epoch": 3.16, - "learning_rate": 6.331013523574978e-06, - "loss": 0.0879, - "step": 931 - }, - { - "epoch": 3.16, - "learning_rate": 6.310585325025499e-06, - "loss": 0.0871, - "step": 932 - }, - { - "epoch": 3.16, - "learning_rate": 6.290174933238531e-06, - "loss": 0.0744, - "step": 933 - }, - { - "epoch": 3.17, - "learning_rate": 6.2697824467237445e-06, - "loss": 0.0872, - "step": 934 - }, - { - "epoch": 3.17, - "learning_rate": 6.249407963904381e-06, - "loss": 0.0851, - "step": 935 - }, - { - "epoch": 3.17, - "learning_rate": 6.229051583116796e-06, - "loss": 0.0941, - "step": 936 - }, - { - "epoch": 3.18, - "learning_rate": 6.208713402609968e-06, - "loss": 0.0888, - "step": 937 - }, - { - "epoch": 3.18, - "learning_rate": 6.1883935205450396e-06, - "loss": 0.0902, - "step": 938 - }, - { - "epoch": 3.18, - "learning_rate": 6.168092034994832e-06, - "loss": 0.0857, - "step": 939 - }, - { - "epoch": 3.19, - "learning_rate": 6.1478090439433776e-06, - "loss": 0.0894, - "step": 940 - }, - { - "epoch": 3.19, - "learning_rate": 6.127544645285448e-06, - "loss": 0.0818, - "step": 941 - }, - { - "epoch": 3.19, - "learning_rate": 6.107298936826086e-06, - "loss": 0.0843, - "step": 942 - }, - { - "epoch": 3.2, - "learning_rate": 6.087072016280111e-06, - "loss": 0.0864, - "step": 943 - }, - { - "epoch": 3.2, - "learning_rate": 6.066863981271678e-06, - "loss": 0.0842, - "step": 944 - }, - { - "epoch": 3.2, - "learning_rate": 6.046674929333787e-06, - "loss": 0.0808, - "step": 945 - }, - { - "epoch": 3.21, - "learning_rate": 6.0265049579078125e-06, - "loss": 0.0859, - "step": 946 - }, - { - "epoch": 3.21, - "learning_rate": 6.006354164343047e-06, - "loss": 0.0904, - "step": 947 - }, - { - "epoch": 3.21, - "learning_rate": 5.986222645896214e-06, - "loss": 0.0838, - "step": 948 - }, - { - "epoch": 3.22, - "learning_rate": 5.96611049973101e-06, - "loss": 0.077, - "step": 949 - }, - { - "epoch": 3.22, - "learning_rate": 5.946017822917632e-06, - "loss": 0.0941, - "step": 950 - }, - { - "epoch": 3.22, - "learning_rate": 5.925944712432317e-06, - "loss": 0.085, - "step": 951 - }, - { - "epoch": 3.23, - "learning_rate": 5.905891265156849e-06, - "loss": 0.0819, - "step": 952 - }, - { - "epoch": 3.23, - "learning_rate": 5.885857577878122e-06, - "loss": 0.0861, - "step": 953 - }, - { - "epoch": 3.23, - "learning_rate": 5.865843747287659e-06, - "loss": 0.0805, - "step": 954 - }, - { - "epoch": 3.24, - "learning_rate": 5.845849869981137e-06, - "loss": 0.0895, - "step": 955 - }, - { - "epoch": 3.24, - "learning_rate": 5.825876042457939e-06, - "loss": 0.0817, - "step": 956 - }, - { - "epoch": 3.24, - "learning_rate": 5.8059223611206716e-06, - "loss": 0.0848, - "step": 957 - }, - { - "epoch": 3.25, - "learning_rate": 5.785988922274711e-06, - "loss": 0.0759, - "step": 958 - }, - { - "epoch": 3.25, - "learning_rate": 5.766075822127735e-06, - "loss": 0.0781, - "step": 959 - }, - { - "epoch": 3.25, - "learning_rate": 5.746183156789252e-06, - "loss": 0.0826, - "step": 960 - }, - { - "epoch": 3.26, - "learning_rate": 5.726311022270152e-06, - "loss": 0.0868, - "step": 961 - }, - { - "epoch": 3.26, - "learning_rate": 5.706459514482226e-06, - "loss": 0.093, - "step": 962 - }, - { - "epoch": 3.26, - "learning_rate": 5.686628729237713e-06, - "loss": 0.0862, - "step": 963 - }, - { - "epoch": 3.27, - "learning_rate": 5.66681876224884e-06, - "loss": 0.0895, - "step": 964 - }, - { - "epoch": 3.27, - "learning_rate": 5.647029709127355e-06, - "loss": 0.0881, - "step": 965 - }, - { - "epoch": 3.27, - "learning_rate": 5.627261665384056e-06, - "loss": 0.089, - "step": 966 - }, - { - "epoch": 3.28, - "learning_rate": 5.6075147264283526e-06, - "loss": 0.0807, - "step": 967 - }, - { - "epoch": 3.28, - "learning_rate": 5.587788987567785e-06, - "loss": 0.0896, - "step": 968 - }, - { - "epoch": 3.28, - "learning_rate": 5.5680845440075885e-06, - "loss": 0.0863, - "step": 969 - }, - { - "epoch": 3.29, - "learning_rate": 5.548401490850193e-06, - "loss": 0.0887, - "step": 970 - }, - { - "epoch": 3.29, - "learning_rate": 5.528739923094806e-06, - "loss": 0.0862, - "step": 971 - }, - { - "epoch": 3.29, - "learning_rate": 5.509099935636932e-06, - "loss": 0.0908, - "step": 972 - }, - { - "epoch": 3.3, - "learning_rate": 5.4894816232679195e-06, - "loss": 0.099, - "step": 973 - }, - { - "epoch": 3.3, - "learning_rate": 5.469885080674508e-06, - "loss": 0.0918, - "step": 974 - }, - { - "epoch": 3.31, - "learning_rate": 5.450310402438353e-06, - "loss": 0.0844, - "step": 975 - }, - { - "epoch": 3.31, - "learning_rate": 5.4307576830355945e-06, - "loss": 0.0939, - "step": 976 - }, - { - "epoch": 3.31, - "learning_rate": 5.4112270168363854e-06, - "loss": 0.0881, - "step": 977 - }, - { - "epoch": 3.32, - "learning_rate": 5.391718498104451e-06, - "loss": 0.0914, - "step": 978 - }, - { - "epoch": 3.32, - "learning_rate": 5.3722322209966024e-06, - "loss": 0.0901, - "step": 979 - }, - { - "epoch": 3.32, - "learning_rate": 5.352768279562315e-06, - "loss": 0.0823, - "step": 980 - }, - { - "epoch": 3.33, - "learning_rate": 5.333326767743263e-06, - "loss": 0.0753, - "step": 981 - }, - { - "epoch": 3.33, - "learning_rate": 5.313907779372862e-06, - "loss": 0.092, - "step": 982 - }, - { - "epoch": 3.33, - "learning_rate": 5.294511408175825e-06, - "loss": 0.0921, - "step": 983 - }, - { - "epoch": 3.34, - "learning_rate": 5.275137747767691e-06, - "loss": 0.0847, - "step": 984 - }, - { - "epoch": 3.34, - "learning_rate": 5.2557868916543996e-06, - "loss": 0.0879, - "step": 985 - }, - { - "epoch": 3.34, - "learning_rate": 5.236458933231818e-06, - "loss": 0.0986, - "step": 986 - }, - { - "epoch": 3.35, - "learning_rate": 5.217153965785315e-06, - "loss": 0.0929, - "step": 987 - }, - { - "epoch": 3.35, - "learning_rate": 5.1978720824892725e-06, - "loss": 0.0905, - "step": 988 - }, - { - "epoch": 3.35, - "learning_rate": 5.178613376406672e-06, - "loss": 0.0914, - "step": 989 - }, - { - "epoch": 3.36, - "learning_rate": 5.1593779404886255e-06, - "loss": 0.0925, - "step": 990 - }, - { - "epoch": 3.36, - "learning_rate": 5.14016586757394e-06, - "loss": 0.0881, - "step": 991 - }, - { - "epoch": 3.36, - "learning_rate": 5.120977250388657e-06, - "loss": 0.0892, - "step": 992 - }, - { - "epoch": 3.37, - "learning_rate": 5.1018121815456045e-06, - "loss": 0.0899, - "step": 993 - }, - { - "epoch": 3.37, - "learning_rate": 5.082670753543961e-06, - "loss": 0.0944, - "step": 994 - }, - { - "epoch": 3.37, - "learning_rate": 5.063553058768814e-06, - "loss": 0.0831, - "step": 995 - }, - { - "epoch": 3.38, - "learning_rate": 5.044459189490694e-06, - "loss": 0.0902, - "step": 996 - }, - { - "epoch": 3.38, - "learning_rate": 5.025389237865128e-06, - "loss": 0.0892, - "step": 997 - }, - { - "epoch": 3.38, - "learning_rate": 5.0063432959322265e-06, - "loss": 0.0917, - "step": 998 - }, - { - "epoch": 3.39, - "learning_rate": 4.987321455616206e-06, - "loss": 0.0841, - "step": 999 - }, - { - "epoch": 3.39, - "learning_rate": 4.968323808724962e-06, - "loss": 0.0911, - "step": 1000 - }, - { - "epoch": 3.39, - "learning_rate": 4.9493504469496235e-06, - "loss": 0.0944, - "step": 1001 - }, - { - "epoch": 3.4, - "learning_rate": 4.930401461864099e-06, - "loss": 0.092, - "step": 1002 - }, - { - "epoch": 3.4, - "learning_rate": 4.911476944924651e-06, - "loss": 0.0875, - "step": 1003 - }, - { - "epoch": 3.4, - "learning_rate": 4.892576987469456e-06, - "loss": 0.0951, - "step": 1004 - }, - { - "epoch": 3.41, - "learning_rate": 4.873701680718146e-06, - "loss": 0.0896, - "step": 1005 - }, - { - "epoch": 3.41, - "learning_rate": 4.854851115771373e-06, - "loss": 0.0851, - "step": 1006 - }, - { - "epoch": 3.41, - "learning_rate": 4.836025383610382e-06, - "loss": 0.0899, - "step": 1007 - }, - { - "epoch": 3.42, - "learning_rate": 4.817224575096564e-06, - "loss": 0.0964, - "step": 1008 - }, - { - "epoch": 3.42, - "learning_rate": 4.798448780971013e-06, - "loss": 0.0878, - "step": 1009 - }, - { - "epoch": 3.42, - "learning_rate": 4.779698091854098e-06, - "loss": 0.0817, - "step": 1010 - }, - { - "epoch": 3.43, - "learning_rate": 4.7609725982450176e-06, - "loss": 0.0879, - "step": 1011 - }, - { - "epoch": 3.43, - "learning_rate": 4.742272390521354e-06, - "loss": 0.084, - "step": 1012 - }, - { - "epoch": 3.43, - "learning_rate": 4.7235975589386715e-06, - "loss": 0.0958, - "step": 1013 - }, - { - "epoch": 3.44, - "learning_rate": 4.704948193630041e-06, - "loss": 0.0938, - "step": 1014 - }, - { - "epoch": 3.44, - "learning_rate": 4.686324384605629e-06, - "loss": 0.0946, - "step": 1015 - }, - { - "epoch": 3.44, - "learning_rate": 4.667726221752249e-06, - "loss": 0.0885, - "step": 1016 - }, - { - "epoch": 3.45, - "learning_rate": 4.649153794832939e-06, - "loss": 0.0878, - "step": 1017 - }, - { - "epoch": 3.45, - "learning_rate": 4.630607193486525e-06, - "loss": 0.0888, - "step": 1018 - }, - { - "epoch": 3.45, - "learning_rate": 4.612086507227186e-06, - "loss": 0.0978, - "step": 1019 - }, - { - "epoch": 3.46, - "learning_rate": 4.593591825444028e-06, - "loss": 0.0856, - "step": 1020 - }, - { - "epoch": 3.46, - "learning_rate": 4.5751232374006304e-06, - "loss": 0.0905, - "step": 1021 - }, - { - "epoch": 3.46, - "learning_rate": 4.556680832234657e-06, - "loss": 0.0847, - "step": 1022 - }, - { - "epoch": 3.47, - "learning_rate": 4.538264698957387e-06, - "loss": 0.0872, - "step": 1023 - }, - { - "epoch": 3.47, - "learning_rate": 4.519874926453303e-06, - "loss": 0.0845, - "step": 1024 - }, - { - "epoch": 3.47, - "learning_rate": 4.501511603479653e-06, - "loss": 0.0943, - "step": 1025 - }, - { - "epoch": 3.48, - "learning_rate": 4.483174818666034e-06, - "loss": 0.089, - "step": 1026 - }, - { - "epoch": 3.48, - "learning_rate": 4.4648646605139605e-06, - "loss": 0.0863, - "step": 1027 - }, - { - "epoch": 3.48, - "learning_rate": 4.446581217396428e-06, - "loss": 0.0837, - "step": 1028 - }, - { - "epoch": 3.49, - "learning_rate": 4.428324577557501e-06, - "loss": 0.0837, - "step": 1029 - }, - { - "epoch": 3.49, - "learning_rate": 4.410094829111865e-06, - "loss": 0.0863, - "step": 1030 - }, - { - "epoch": 3.49, - "learning_rate": 4.391892060044435e-06, - "loss": 0.0875, - "step": 1031 - }, - { - "epoch": 3.5, - "learning_rate": 4.373716358209898e-06, - "loss": 0.0975, - "step": 1032 - }, - { - "epoch": 3.5, - "learning_rate": 4.355567811332311e-06, - "loss": 0.0825, - "step": 1033 - }, - { - "epoch": 3.51, - "learning_rate": 4.337446507004656e-06, - "loss": 0.0823, - "step": 1034 - }, - { - "epoch": 3.51, - "learning_rate": 4.319352532688444e-06, - "loss": 0.0876, - "step": 1035 - }, - { - "epoch": 3.51, - "learning_rate": 4.3012859757132715e-06, - "loss": 0.0937, - "step": 1036 - }, - { - "epoch": 3.52, - "learning_rate": 4.283246923276411e-06, - "loss": 0.0909, - "step": 1037 - }, - { - "epoch": 3.52, - "learning_rate": 4.265235462442389e-06, - "loss": 0.0894, - "step": 1038 - }, - { - "epoch": 3.52, - "learning_rate": 4.247251680142542e-06, - "loss": 0.0856, - "step": 1039 - }, - { - "epoch": 3.53, - "learning_rate": 4.2292956631746475e-06, - "loss": 0.0946, - "step": 1040 - }, - { - "epoch": 3.53, - "learning_rate": 4.211367498202456e-06, - "loss": 0.0876, - "step": 1041 - }, - { - "epoch": 3.53, - "learning_rate": 4.1934672717552986e-06, - "loss": 0.0885, - "step": 1042 - }, - { - "epoch": 3.54, - "learning_rate": 4.175595070227655e-06, - "loss": 0.0835, - "step": 1043 - }, - { - "epoch": 3.54, - "learning_rate": 4.157750979878753e-06, - "loss": 0.092, - "step": 1044 - }, - { - "epoch": 3.54, - "learning_rate": 4.13993508683214e-06, - "loss": 0.0844, - "step": 1045 - }, - { - "epoch": 3.55, - "learning_rate": 4.12214747707527e-06, - "loss": 0.0857, - "step": 1046 - }, - { - "epoch": 3.55, - "learning_rate": 4.1043882364590895e-06, - "loss": 0.0967, - "step": 1047 - }, - { - "epoch": 3.55, - "learning_rate": 4.086657450697623e-06, - "loss": 0.0876, - "step": 1048 - }, - { - "epoch": 3.56, - "learning_rate": 4.068955205367559e-06, - "loss": 0.0922, - "step": 1049 - }, - { - "epoch": 3.56, - "learning_rate": 4.051281585907841e-06, - "loss": 0.0881, - "step": 1050 - }, - { - "epoch": 3.56, - "learning_rate": 4.033636677619242e-06, - "loss": 0.0953, - "step": 1051 - }, - { - "epoch": 3.57, - "learning_rate": 4.016020565663974e-06, - "loss": 0.0862, - "step": 1052 - }, - { - "epoch": 3.57, - "learning_rate": 3.998433335065251e-06, - "loss": 0.0883, - "step": 1053 - }, - { - "epoch": 3.57, - "learning_rate": 3.980875070706906e-06, - "loss": 0.0974, - "step": 1054 - }, - { - "epoch": 3.58, - "learning_rate": 3.96334585733296e-06, - "loss": 0.0912, - "step": 1055 - }, - { - "epoch": 3.58, - "learning_rate": 3.9458457795472245e-06, - "loss": 0.0928, - "step": 1056 - }, - { - "epoch": 3.58, - "learning_rate": 3.9283749218128885e-06, - "loss": 0.095, - "step": 1057 - }, - { - "epoch": 3.59, - "learning_rate": 3.910933368452112e-06, - "loss": 0.0916, - "step": 1058 - }, - { - "epoch": 3.59, - "learning_rate": 3.893521203645618e-06, - "loss": 0.079, - "step": 1059 - }, - { - "epoch": 3.59, - "learning_rate": 3.8761385114322905e-06, - "loss": 0.0918, - "step": 1060 - }, - { - "epoch": 3.6, - "learning_rate": 3.858785375708764e-06, - "loss": 0.0846, - "step": 1061 - }, - { - "epoch": 3.6, - "learning_rate": 3.841461880229016e-06, - "loss": 0.0878, - "step": 1062 - }, - { - "epoch": 3.6, - "learning_rate": 3.824168108603971e-06, - "loss": 0.0927, - "step": 1063 - }, - { - "epoch": 3.61, - "learning_rate": 3.8069041443010924e-06, - "loss": 0.0873, - "step": 1064 - }, - { - "epoch": 3.61, - "learning_rate": 3.7896700706439826e-06, - "loss": 0.085, - "step": 1065 - }, - { - "epoch": 3.61, - "learning_rate": 3.7724659708119737e-06, - "loss": 0.0863, - "step": 1066 - }, - { - "epoch": 3.62, - "learning_rate": 3.7552919278397335e-06, - "loss": 0.0842, - "step": 1067 - }, - { - "epoch": 3.62, - "learning_rate": 3.738148024616863e-06, - "loss": 0.0894, - "step": 1068 - }, - { - "epoch": 3.62, - "learning_rate": 3.7210343438874917e-06, - "loss": 0.0942, - "step": 1069 - }, - { - "epoch": 3.63, - "learning_rate": 3.7039509682498887e-06, - "loss": 0.0868, - "step": 1070 - }, - { - "epoch": 3.63, - "learning_rate": 3.6868979801560443e-06, - "loss": 0.0906, - "step": 1071 - }, - { - "epoch": 3.63, - "learning_rate": 3.6698754619112974e-06, - "loss": 0.088, - "step": 1072 - }, - { - "epoch": 3.64, - "learning_rate": 3.6528834956739224e-06, - "loss": 0.0883, - "step": 1073 - }, - { - "epoch": 3.64, - "learning_rate": 3.6359221634547324e-06, - "loss": 0.0904, - "step": 1074 - }, - { - "epoch": 3.64, - "learning_rate": 3.6189915471166927e-06, - "loss": 0.0911, - "step": 1075 - }, - { - "epoch": 3.65, - "learning_rate": 3.602091728374515e-06, - "loss": 0.0886, - "step": 1076 - }, - { - "epoch": 3.65, - "learning_rate": 3.5852227887942713e-06, - "loss": 0.0883, - "step": 1077 - }, - { - "epoch": 3.65, - "learning_rate": 3.5683848097929963e-06, - "loss": 0.0916, - "step": 1078 - }, - { - "epoch": 3.66, - "learning_rate": 3.5515778726382967e-06, - "loss": 0.09, - "step": 1079 - }, - { - "epoch": 3.66, - "learning_rate": 3.5348020584479492e-06, - "loss": 0.0875, - "step": 1080 - }, - { - "epoch": 3.66, - "learning_rate": 3.518057448189527e-06, - "loss": 0.0863, - "step": 1081 - }, - { - "epoch": 3.67, - "learning_rate": 3.501344122679995e-06, - "loss": 0.0889, - "step": 1082 - }, - { - "epoch": 3.67, - "learning_rate": 3.4846621625853248e-06, - "loss": 0.089, - "step": 1083 - }, - { - "epoch": 3.67, - "learning_rate": 3.4680116484201055e-06, - "loss": 0.0912, - "step": 1084 - }, - { - "epoch": 3.68, - "learning_rate": 3.4513926605471504e-06, - "loss": 0.0926, - "step": 1085 - }, - { - "epoch": 3.68, - "learning_rate": 3.4348052791771158e-06, - "loss": 0.0932, - "step": 1086 - }, - { - "epoch": 3.68, - "learning_rate": 3.4182495843681117e-06, - "loss": 0.0923, - "step": 1087 - }, - { - "epoch": 3.69, - "learning_rate": 3.401725656025315e-06, - "loss": 0.0833, - "step": 1088 - }, - { - "epoch": 3.69, - "learning_rate": 3.385233573900576e-06, - "loss": 0.0927, - "step": 1089 - }, - { - "epoch": 3.69, - "learning_rate": 3.3687734175920505e-06, - "loss": 0.0903, - "step": 1090 - }, - { - "epoch": 3.7, - "learning_rate": 3.3523452665438004e-06, - "loss": 0.0926, - "step": 1091 - }, - { - "epoch": 3.7, - "learning_rate": 3.3359492000454186e-06, - "loss": 0.091, - "step": 1092 - }, - { - "epoch": 3.71, - "learning_rate": 3.3195852972316435e-06, - "loss": 0.0829, - "step": 1093 - }, - { - "epoch": 3.71, - "learning_rate": 3.3032536370819746e-06, - "loss": 0.0884, - "step": 1094 - }, - { - "epoch": 3.71, - "learning_rate": 3.2869542984202974e-06, - "loss": 0.0889, - "step": 1095 - }, - { - "epoch": 3.72, - "learning_rate": 3.2706873599144973e-06, - "loss": 0.0927, - "step": 1096 - }, - { - "epoch": 3.72, - "learning_rate": 3.254452900076083e-06, - "loss": 0.0861, - "step": 1097 - }, - { - "epoch": 3.72, - "learning_rate": 3.2382509972598087e-06, - "loss": 0.0916, - "step": 1098 - }, - { - "epoch": 3.73, - "learning_rate": 3.2220817296632845e-06, - "loss": 0.0885, - "step": 1099 - }, - { - "epoch": 3.73, - "learning_rate": 3.205945175326617e-06, - "loss": 0.0908, - "step": 1100 - }, - { - "epoch": 3.73, - "learning_rate": 3.1898414121320277e-06, - "loss": 0.0966, - "step": 1101 - }, - { - "epoch": 3.74, - "learning_rate": 3.173770517803467e-06, - "loss": 0.0953, - "step": 1102 - }, - { - "epoch": 3.74, - "learning_rate": 3.1577325699062424e-06, - "loss": 0.0801, - "step": 1103 - }, - { - "epoch": 3.74, - "learning_rate": 3.1417276458466515e-06, - "loss": 0.0889, - "step": 1104 - }, - { - "epoch": 3.75, - "learning_rate": 3.125755822871607e-06, - "loss": 0.0906, - "step": 1105 - }, - { - "epoch": 3.75, - "learning_rate": 3.1098171780682553e-06, - "loss": 0.0818, - "step": 1106 - }, - { - "epoch": 3.75, - "learning_rate": 3.093911788363617e-06, - "loss": 0.0864, - "step": 1107 - }, - { - "epoch": 3.76, - "learning_rate": 3.078039730524198e-06, - "loss": 0.0891, - "step": 1108 - }, - { - "epoch": 3.76, - "learning_rate": 3.062201081155637e-06, - "loss": 0.0977, - "step": 1109 - }, - { - "epoch": 3.76, - "learning_rate": 3.0463959167023336e-06, - "loss": 0.0866, - "step": 1110 - }, - { - "epoch": 3.77, - "learning_rate": 3.0306243134470668e-06, - "loss": 0.0859, - "step": 1111 - }, - { - "epoch": 3.77, - "learning_rate": 3.0148863475106315e-06, - "loss": 0.0899, - "step": 1112 - }, - { - "epoch": 3.77, - "learning_rate": 2.9991820948514795e-06, - "loss": 0.0912, - "step": 1113 - }, - { - "epoch": 3.78, - "learning_rate": 2.9835116312653477e-06, - "loss": 0.0854, - "step": 1114 - }, - { - "epoch": 3.78, - "learning_rate": 2.9678750323848893e-06, - "loss": 0.0919, - "step": 1115 - }, - { - "epoch": 3.78, - "learning_rate": 2.952272373679316e-06, - "loss": 0.0874, - "step": 1116 - }, - { - "epoch": 3.79, - "learning_rate": 2.936703730454017e-06, - "loss": 0.0932, - "step": 1117 - }, - { - "epoch": 3.79, - "learning_rate": 2.9211691778502173e-06, - "loss": 0.0953, - "step": 1118 - }, - { - "epoch": 3.79, - "learning_rate": 2.90566879084461e-06, - "loss": 0.0986, - "step": 1119 - }, - { - "epoch": 3.8, - "learning_rate": 2.890202644248983e-06, - "loss": 0.0892, - "step": 1120 - }, - { - "epoch": 3.8, - "learning_rate": 2.8747708127098593e-06, - "loss": 0.0892, - "step": 1121 - }, - { - "epoch": 3.8, - "learning_rate": 2.8593733707081516e-06, - "loss": 0.0836, - "step": 1122 - }, - { - "epoch": 3.81, - "learning_rate": 2.8440103925587904e-06, - "loss": 0.0826, - "step": 1123 - }, - { - "epoch": 3.81, - "learning_rate": 2.8286819524103657e-06, - "loss": 0.0844, - "step": 1124 - }, - { - "epoch": 3.81, - "learning_rate": 2.813388124244778e-06, - "loss": 0.0905, - "step": 1125 - }, - { - "epoch": 3.82, - "learning_rate": 2.798128981876864e-06, - "loss": 0.0864, - "step": 1126 - }, - { - "epoch": 3.82, - "learning_rate": 2.7829045989540594e-06, - "loss": 0.0845, - "step": 1127 - }, - { - "epoch": 3.82, - "learning_rate": 2.7677150489560378e-06, - "loss": 0.0906, - "step": 1128 - }, - { - "epoch": 3.83, - "learning_rate": 2.7525604051943512e-06, - "loss": 0.0886, - "step": 1129 - }, - { - "epoch": 3.83, - "learning_rate": 2.7374407408120685e-06, - "loss": 0.1002, - "step": 1130 - }, - { - "epoch": 3.83, - "learning_rate": 2.7223561287834467e-06, - "loss": 0.087, - "step": 1131 - }, - { - "epoch": 3.84, - "learning_rate": 2.707306641913556e-06, - "loss": 0.0928, - "step": 1132 - }, - { - "epoch": 3.84, - "learning_rate": 2.692292352837942e-06, - "loss": 0.0926, - "step": 1133 - }, - { - "epoch": 3.84, - "learning_rate": 2.6773133340222677e-06, - "loss": 0.0892, - "step": 1134 - }, - { - "epoch": 3.85, - "learning_rate": 2.662369657761963e-06, - "loss": 0.0855, - "step": 1135 - }, - { - "epoch": 3.85, - "learning_rate": 2.6474613961818785e-06, - "loss": 0.0859, - "step": 1136 - }, - { - "epoch": 3.85, - "learning_rate": 2.6325886212359496e-06, - "loss": 0.0915, - "step": 1137 - }, - { - "epoch": 3.86, - "learning_rate": 2.6177514047068287e-06, - "loss": 0.0861, - "step": 1138 - }, - { - "epoch": 3.86, - "learning_rate": 2.602949818205539e-06, - "loss": 0.0825, - "step": 1139 - }, - { - "epoch": 3.86, - "learning_rate": 2.5881839331711524e-06, - "loss": 0.0855, - "step": 1140 - }, - { - "epoch": 3.87, - "learning_rate": 2.5734538208704197e-06, - "loss": 0.0873, - "step": 1141 - }, - { - "epoch": 3.87, - "learning_rate": 2.5587595523974408e-06, - "loss": 0.0882, - "step": 1142 - }, - { - "epoch": 3.87, - "learning_rate": 2.5441011986733165e-06, - "loss": 0.0888, - "step": 1143 - }, - { - "epoch": 3.88, - "learning_rate": 2.5294788304458063e-06, - "loss": 0.0886, - "step": 1144 - }, - { - "epoch": 3.88, - "learning_rate": 2.514892518288988e-06, - "loss": 0.0875, - "step": 1145 - }, - { - "epoch": 3.88, - "learning_rate": 2.5003423326029187e-06, - "loss": 0.0872, - "step": 1146 - }, - { - "epoch": 3.89, - "learning_rate": 2.485828343613288e-06, - "loss": 0.0823, - "step": 1147 - }, - { - "epoch": 3.89, - "learning_rate": 2.4713506213710924e-06, - "loss": 0.0926, - "step": 1148 - }, - { - "epoch": 3.89, - "learning_rate": 2.456909235752276e-06, - "loss": 0.0826, - "step": 1149 - }, - { - "epoch": 3.9, - "learning_rate": 2.4425042564574186e-06, - "loss": 0.0912, - "step": 1150 - }, - { - "epoch": 3.9, - "learning_rate": 2.4281357530113804e-06, - "loss": 0.0887, - "step": 1151 - }, - { - "epoch": 3.91, - "learning_rate": 2.4138037947629743e-06, - "loss": 0.0919, - "step": 1152 - }, - { - "epoch": 3.91, - "learning_rate": 2.399508450884631e-06, - "loss": 0.0882, - "step": 1153 - }, - { - "epoch": 3.91, - "learning_rate": 2.3852497903720626e-06, - "loss": 0.092, - "step": 1154 - }, - { - "epoch": 3.92, - "learning_rate": 2.3710278820439313e-06, - "loss": 0.0955, - "step": 1155 - }, - { - "epoch": 3.92, - "learning_rate": 2.3568427945415163e-06, - "loss": 0.0866, - "step": 1156 - }, - { - "epoch": 3.92, - "learning_rate": 2.3426945963283853e-06, - "loss": 0.0833, - "step": 1157 - }, - { - "epoch": 3.93, - "learning_rate": 2.328583355690056e-06, - "loss": 0.0872, - "step": 1158 - }, - { - "epoch": 3.93, - "learning_rate": 2.3145091407336785e-06, - "loss": 0.0939, - "step": 1159 - }, - { - "epoch": 3.93, - "learning_rate": 2.3004720193876972e-06, - "loss": 0.0837, - "step": 1160 - }, - { - "epoch": 3.94, - "learning_rate": 2.2864720594015288e-06, - "loss": 0.0881, - "step": 1161 - }, - { - "epoch": 3.94, - "learning_rate": 2.2725093283452305e-06, - "loss": 0.0909, - "step": 1162 - }, - { - "epoch": 3.94, - "learning_rate": 2.2585838936091753e-06, - "loss": 0.0907, - "step": 1163 - }, - { - "epoch": 3.95, - "learning_rate": 2.244695822403731e-06, - "loss": 0.0843, - "step": 1164 - }, - { - "epoch": 3.95, - "learning_rate": 2.230845181758928e-06, - "loss": 0.09, - "step": 1165 - }, - { - "epoch": 3.95, - "learning_rate": 2.2170320385241475e-06, - "loss": 0.0881, - "step": 1166 - }, - { - "epoch": 3.96, - "learning_rate": 2.2032564593677773e-06, - "loss": 0.0891, - "step": 1167 - }, - { - "epoch": 3.96, - "learning_rate": 2.189518510776919e-06, - "loss": 0.0853, - "step": 1168 - }, - { - "epoch": 3.96, - "learning_rate": 2.1758182590570454e-06, - "loss": 0.0857, - "step": 1169 - }, - { - "epoch": 3.97, - "learning_rate": 2.1621557703316876e-06, - "loss": 0.0921, - "step": 1170 - }, - { - "epoch": 3.97, - "learning_rate": 2.148531110542118e-06, - "loss": 0.093, - "step": 1171 - }, - { - "epoch": 3.97, - "learning_rate": 2.1349443454470254e-06, - "loss": 0.0859, - "step": 1172 - }, - { - "epoch": 3.98, - "learning_rate": 2.1213955406222076e-06, - "loss": 0.0867, - "step": 1173 - }, - { - "epoch": 3.98, - "learning_rate": 2.1078847614602437e-06, - "loss": 0.093, - "step": 1174 - }, - { - "epoch": 3.98, - "learning_rate": 2.09441207317019e-06, - "loss": 0.0919, - "step": 1175 - }, - { - "epoch": 3.99, - "learning_rate": 2.0809775407772505e-06, - "loss": 0.0912, - "step": 1176 - }, - { - "epoch": 3.99, - "learning_rate": 2.0675812291224796e-06, - "loss": 0.0898, - "step": 1177 - }, - { - "epoch": 3.99, - "learning_rate": 2.0542232028624585e-06, - "loss": 0.0868, - "step": 1178 - }, - { - "epoch": 4.0, - "learning_rate": 2.0409035264689857e-06, - "loss": 0.0991, - "step": 1179 - }, - { - "epoch": 4.0, - "learning_rate": 2.027622264228768e-06, - "loss": 0.0673, - "step": 1180 - }, - { - "epoch": 4.0, - "learning_rate": 2.014379480243105e-06, - "loss": 0.04, - "step": 1181 - }, - { - "epoch": 4.01, - "learning_rate": 2.0011752384275862e-06, - "loss": 0.0396, - "step": 1182 - }, - { - "epoch": 4.01, - "learning_rate": 1.988009602511779e-06, - "loss": 0.04, - "step": 1183 - }, - { - "epoch": 4.01, - "learning_rate": 1.9748826360389216e-06, - "loss": 0.0384, - "step": 1184 - }, - { - "epoch": 4.02, - "learning_rate": 1.961794402365611e-06, - "loss": 0.0378, - "step": 1185 - }, - { - "epoch": 4.02, - "learning_rate": 1.9487449646615087e-06, - "loss": 0.0407, - "step": 1186 - }, - { - "epoch": 4.02, - "learning_rate": 1.935734385909028e-06, - "loss": 0.0399, - "step": 1187 - }, - { - "epoch": 4.03, - "learning_rate": 1.9227627289030315e-06, - "loss": 0.0361, - "step": 1188 - }, - { - "epoch": 4.03, - "learning_rate": 1.9098300562505266e-06, - "loss": 0.0381, - "step": 1189 - }, - { - "epoch": 4.03, - "learning_rate": 1.8969364303703664e-06, - "loss": 0.0381, - "step": 1190 - }, - { - "epoch": 4.04, - "learning_rate": 1.8840819134929467e-06, - "loss": 0.0373, - "step": 1191 - }, - { - "epoch": 4.04, - "learning_rate": 1.871266567659905e-06, - "loss": 0.034, - "step": 1192 - }, - { - "epoch": 4.04, - "learning_rate": 1.8584904547238214e-06, - "loss": 0.0398, - "step": 1193 - }, - { - "epoch": 4.05, - "learning_rate": 1.8457536363479257e-06, - "loss": 0.0367, - "step": 1194 - }, - { - "epoch": 4.05, - "learning_rate": 1.8330561740057839e-06, - "loss": 0.0359, - "step": 1195 - }, - { - "epoch": 4.05, - "learning_rate": 1.8203981289810212e-06, - "loss": 0.034, - "step": 1196 - }, - { - "epoch": 4.06, - "learning_rate": 1.8077795623670135e-06, - "loss": 0.0364, - "step": 1197 - }, - { - "epoch": 4.06, - "learning_rate": 1.7952005350666023e-06, - "loss": 0.0332, - "step": 1198 - }, - { - "epoch": 4.06, - "learning_rate": 1.7826611077917843e-06, - "loss": 0.0364, - "step": 1199 - }, - { - "epoch": 4.07, - "learning_rate": 1.7701613410634367e-06, - "loss": 0.032, - "step": 1200 - }, - { - "epoch": 4.07, - "learning_rate": 1.757701295211014e-06, - "loss": 0.037, - "step": 1201 - }, - { - "epoch": 4.07, - "learning_rate": 1.74528103037226e-06, - "loss": 0.0325, - "step": 1202 - }, - { - "epoch": 4.08, - "learning_rate": 1.7329006064929232e-06, - "loss": 0.0339, - "step": 1203 - }, - { - "epoch": 4.08, - "learning_rate": 1.7205600833264501e-06, - "loss": 0.031, - "step": 1204 - }, - { - "epoch": 4.08, - "learning_rate": 1.7082595204337183e-06, - "loss": 0.0332, - "step": 1205 - }, - { - "epoch": 4.09, - "learning_rate": 1.6959989771827346e-06, - "loss": 0.0377, - "step": 1206 - }, - { - "epoch": 4.09, - "learning_rate": 1.683778512748362e-06, - "loss": 0.0354, - "step": 1207 - }, - { - "epoch": 4.09, - "learning_rate": 1.6715981861120112e-06, - "loss": 0.0363, - "step": 1208 - }, - { - "epoch": 4.1, - "learning_rate": 1.6594580560613782e-06, - "loss": 0.0367, - "step": 1209 - }, - { - "epoch": 4.1, - "learning_rate": 1.6473581811901529e-06, - "loss": 0.0314, - "step": 1210 - }, - { - "epoch": 4.11, - "learning_rate": 1.6352986198977327e-06, - "loss": 0.0337, - "step": 1211 - }, - { - "epoch": 4.11, - "learning_rate": 1.6232794303889466e-06, - "loss": 0.0371, - "step": 1212 - }, - { - "epoch": 4.11, - "learning_rate": 1.6113006706737667e-06, - "loss": 0.0343, - "step": 1213 - }, - { - "epoch": 4.12, - "learning_rate": 1.599362398567037e-06, - "loss": 0.0335, - "step": 1214 - }, - { - "epoch": 4.12, - "learning_rate": 1.587464671688187e-06, - "loss": 0.034, - "step": 1215 - }, - { - "epoch": 4.12, - "learning_rate": 1.5756075474609667e-06, - "loss": 0.0346, - "step": 1216 - }, - { - "epoch": 4.13, - "learning_rate": 1.563791083113142e-06, - "loss": 0.0326, - "step": 1217 - }, - { - "epoch": 4.13, - "learning_rate": 1.5520153356762514e-06, - "loss": 0.035, - "step": 1218 - }, - { - "epoch": 4.13, - "learning_rate": 1.540280361985308e-06, - "loss": 0.0341, - "step": 1219 - }, - { - "epoch": 4.14, - "learning_rate": 1.528586218678535e-06, - "loss": 0.0336, - "step": 1220 - }, - { - "epoch": 4.14, - "learning_rate": 1.5169329621970918e-06, - "loss": 0.0347, - "step": 1221 - }, - { - "epoch": 4.14, - "learning_rate": 1.5053206487847916e-06, - "loss": 0.0327, - "step": 1222 - }, - { - "epoch": 4.15, - "learning_rate": 1.4937493344878474e-06, - "loss": 0.0378, - "step": 1223 - }, - { - "epoch": 4.15, - "learning_rate": 1.482219075154585e-06, - "loss": 0.033, - "step": 1224 - }, - { - "epoch": 4.15, - "learning_rate": 1.4707299264351914e-06, - "loss": 0.0338, - "step": 1225 - }, - { - "epoch": 4.16, - "learning_rate": 1.459281943781422e-06, - "loss": 0.0353, - "step": 1226 - }, - { - "epoch": 4.16, - "learning_rate": 1.4478751824463543e-06, - "loss": 0.0338, - "step": 1227 - }, - { - "epoch": 4.16, - "learning_rate": 1.436509697484111e-06, - "loss": 0.0321, - "step": 1228 - }, - { - "epoch": 4.17, - "learning_rate": 1.4251855437495976e-06, - "loss": 0.0313, - "step": 1229 - }, - { - "epoch": 4.17, - "learning_rate": 1.413902775898236e-06, - "loss": 0.0374, - "step": 1230 - }, - { - "epoch": 4.17, - "learning_rate": 1.4026614483857037e-06, - "loss": 0.0379, - "step": 1231 - }, - { - "epoch": 4.18, - "learning_rate": 1.391461615467663e-06, - "loss": 0.0362, - "step": 1232 - }, - { - "epoch": 4.18, - "learning_rate": 1.3803033311995072e-06, - "loss": 0.0349, - "step": 1233 - }, - { - "epoch": 4.18, - "learning_rate": 1.3691866494361029e-06, - "loss": 0.0356, - "step": 1234 - }, - { - "epoch": 4.19, - "learning_rate": 1.3581116238315194e-06, - "loss": 0.0383, - "step": 1235 - }, - { - "epoch": 4.19, - "learning_rate": 1.3470783078387705e-06, - "loss": 0.0376, - "step": 1236 - }, - { - "epoch": 4.19, - "learning_rate": 1.336086754709569e-06, - "loss": 0.0397, - "step": 1237 - }, - { - "epoch": 4.2, - "learning_rate": 1.3251370174940582e-06, - "loss": 0.0341, - "step": 1238 - }, - { - "epoch": 4.2, - "learning_rate": 1.3142291490405568e-06, - "loss": 0.0356, - "step": 1239 - }, - { - "epoch": 4.2, - "learning_rate": 1.3033632019953113e-06, - "loss": 0.0326, - "step": 1240 - }, - { - "epoch": 4.21, - "learning_rate": 1.2925392288022299e-06, - "loss": 0.0363, - "step": 1241 - }, - { - "epoch": 4.21, - "learning_rate": 1.2817572817026402e-06, - "loss": 0.0326, - "step": 1242 - }, - { - "epoch": 4.21, - "learning_rate": 1.2710174127350362e-06, - "loss": 0.0336, - "step": 1243 - }, - { - "epoch": 4.22, - "learning_rate": 1.2603196737348211e-06, - "loss": 0.0344, - "step": 1244 - }, - { - "epoch": 4.22, - "learning_rate": 1.2496641163340562e-06, - "loss": 0.0337, - "step": 1245 - }, - { - "epoch": 4.22, - "learning_rate": 1.2390507919612215e-06, - "loss": 0.0344, - "step": 1246 - }, - { - "epoch": 4.23, - "learning_rate": 1.2284797518409575e-06, - "loss": 0.0328, - "step": 1247 - }, - { - "epoch": 4.23, - "learning_rate": 1.217951046993826e-06, - "loss": 0.0347, - "step": 1248 - }, - { - "epoch": 4.23, - "learning_rate": 1.2074647282360573e-06, - "loss": 0.0327, - "step": 1249 - }, - { - "epoch": 4.24, - "learning_rate": 1.197020846179303e-06, - "loss": 0.0356, - "step": 1250 - }, - { - "epoch": 4.24, - "learning_rate": 1.1866194512304075e-06, - "loss": 0.0348, - "step": 1251 - }, - { - "epoch": 4.24, - "learning_rate": 1.1762605935911432e-06, - "loss": 0.0358, - "step": 1252 - }, - { - "epoch": 4.25, - "learning_rate": 1.165944323257986e-06, - "loss": 0.0363, - "step": 1253 - }, - { - "epoch": 4.25, - "learning_rate": 1.1556706900218572e-06, - "loss": 0.0382, - "step": 1254 - }, - { - "epoch": 4.25, - "learning_rate": 1.1454397434679022e-06, - "loss": 0.0316, - "step": 1255 - }, - { - "epoch": 4.26, - "learning_rate": 1.1352515329752345e-06, - "loss": 0.0378, - "step": 1256 - }, - { - "epoch": 4.26, - "learning_rate": 1.125106107716708e-06, - "loss": 0.0354, - "step": 1257 - }, - { - "epoch": 4.26, - "learning_rate": 1.115003516658677e-06, - "loss": 0.0345, - "step": 1258 - }, - { - "epoch": 4.27, - "learning_rate": 1.10494380856075e-06, - "loss": 0.035, - "step": 1259 - }, - { - "epoch": 4.27, - "learning_rate": 1.0949270319755768e-06, - "loss": 0.0328, - "step": 1260 - }, - { - "epoch": 4.27, - "learning_rate": 1.0849532352485903e-06, - "loss": 0.0336, - "step": 1261 - }, - { - "epoch": 4.28, - "learning_rate": 1.075022466517791e-06, - "loss": 0.0348, - "step": 1262 - }, - { - "epoch": 4.28, - "learning_rate": 1.0651347737134965e-06, - "loss": 0.0394, - "step": 1263 - }, - { - "epoch": 4.28, - "learning_rate": 1.0552902045581305e-06, - "loss": 0.0361, - "step": 1264 - }, - { - "epoch": 4.29, - "learning_rate": 1.0454888065659775e-06, - "loss": 0.0368, - "step": 1265 - }, - { - "epoch": 4.29, - "learning_rate": 1.0357306270429623e-06, - "loss": 0.0377, - "step": 1266 - }, - { - "epoch": 4.29, - "learning_rate": 1.0260157130864178e-06, - "loss": 0.0353, - "step": 1267 - }, - { - "epoch": 4.3, - "learning_rate": 1.0163441115848506e-06, - "loss": 0.0333, - "step": 1268 - }, - { - "epoch": 4.3, - "learning_rate": 1.0067158692177325e-06, - "loss": 0.0361, - "step": 1269 - }, - { - "epoch": 4.31, - "learning_rate": 9.971310324552597e-07, - "loss": 0.0368, - "step": 1270 - }, - { - "epoch": 4.31, - "learning_rate": 9.87589647558135e-07, - "loss": 0.0378, - "step": 1271 - }, - { - "epoch": 4.31, - "learning_rate": 9.780917605773376e-07, - "loss": 0.0354, - "step": 1272 - }, - { - "epoch": 4.32, - "learning_rate": 9.686374173539147e-07, - "loss": 0.0348, - "step": 1273 - }, - { - "epoch": 4.32, - "learning_rate": 9.592266635187464e-07, - "loss": 0.0368, - "step": 1274 - }, - { - "epoch": 4.32, - "learning_rate": 9.49859544492332e-07, - "loss": 0.033, - "step": 1275 - }, - { - "epoch": 4.33, - "learning_rate": 9.40536105484573e-07, - "loss": 0.0377, - "step": 1276 - }, - { - "epoch": 4.33, - "learning_rate": 9.312563914945461e-07, - "loss": 0.0346, - "step": 1277 - }, - { - "epoch": 4.33, - "learning_rate": 9.220204473102945e-07, - "loss": 0.0351, - "step": 1278 - }, - { - "epoch": 4.34, - "learning_rate": 9.128283175086106e-07, - "loss": 0.0373, - "step": 1279 - }, - { - "epoch": 4.34, - "learning_rate": 9.036800464548157e-07, - "loss": 0.034, - "step": 1280 - }, - { - "epoch": 4.34, - "learning_rate": 8.945756783025528e-07, - "loss": 0.0329, - "step": 1281 - }, - { - "epoch": 4.35, - "learning_rate": 8.855152569935632e-07, - "loss": 0.0341, - "step": 1282 - }, - { - "epoch": 4.35, - "learning_rate": 8.76498826257488e-07, - "loss": 0.0352, - "step": 1283 - }, - { - "epoch": 4.35, - "learning_rate": 8.675264296116481e-07, - "loss": 0.0346, - "step": 1284 - }, - { - "epoch": 4.36, - "learning_rate": 8.585981103608343e-07, - "loss": 0.0359, - "step": 1285 - }, - { - "epoch": 4.36, - "learning_rate": 8.497139115971031e-07, - "loss": 0.0355, - "step": 1286 - }, - { - "epoch": 4.36, - "learning_rate": 8.40873876199565e-07, - "loss": 0.0356, - "step": 1287 - }, - { - "epoch": 4.37, - "learning_rate": 8.320780468341761e-07, - "loss": 0.034, - "step": 1288 - }, - { - "epoch": 4.37, - "learning_rate": 8.233264659535367e-07, - "loss": 0.0345, - "step": 1289 - }, - { - "epoch": 4.37, - "learning_rate": 8.146191757966859e-07, - "loss": 0.0353, - "step": 1290 - }, - { - "epoch": 4.38, - "learning_rate": 8.059562183888903e-07, - "loss": 0.0357, - "step": 1291 - }, - { - "epoch": 4.38, - "learning_rate": 7.97337635541452e-07, - "loss": 0.0354, - "step": 1292 - }, - { - "epoch": 4.38, - "learning_rate": 7.887634688515e-07, - "loss": 0.0313, - "step": 1293 - }, - { - "epoch": 4.39, - "learning_rate": 7.802337597017895e-07, - "loss": 0.0342, - "step": 1294 - }, - { - "epoch": 4.39, - "learning_rate": 7.71748549260507e-07, - "loss": 0.0351, - "step": 1295 - }, - { - "epoch": 4.39, - "learning_rate": 7.633078784810666e-07, - "loss": 0.0359, - "step": 1296 - }, - { - "epoch": 4.4, - "learning_rate": 7.549117881019141e-07, - "loss": 0.034, - "step": 1297 - }, - { - "epoch": 4.4, - "learning_rate": 7.465603186463299e-07, - "loss": 0.0342, - "step": 1298 - }, - { - "epoch": 4.4, - "learning_rate": 7.382535104222366e-07, - "loss": 0.0363, - "step": 1299 - }, - { - "epoch": 4.41, - "learning_rate": 7.299914035219957e-07, - "loss": 0.0378, - "step": 1300 - }, - { - "epoch": 4.41, - "learning_rate": 7.21774037822226e-07, - "loss": 0.0357, - "step": 1301 - }, - { - "epoch": 4.41, - "learning_rate": 7.136014529836033e-07, - "loss": 0.0377, - "step": 1302 - }, - { - "epoch": 4.42, - "learning_rate": 7.054736884506718e-07, - "loss": 0.0355, - "step": 1303 - }, - { - "epoch": 4.42, - "learning_rate": 6.973907834516513e-07, - "loss": 0.0377, - "step": 1304 - }, - { - "epoch": 4.42, - "learning_rate": 6.8935277699825e-07, - "loss": 0.0386, - "step": 1305 - }, - { - "epoch": 4.43, - "learning_rate": 6.813597078854772e-07, - "loss": 0.0353, - "step": 1306 - }, - { - "epoch": 4.43, - "learning_rate": 6.734116146914516e-07, - "loss": 0.0332, - "step": 1307 - }, - { - "epoch": 4.43, - "learning_rate": 6.655085357772229e-07, - "loss": 0.0355, - "step": 1308 - }, - { - "epoch": 4.44, - "learning_rate": 6.576505092865748e-07, - "loss": 0.0331, - "step": 1309 - }, - { - "epoch": 4.44, - "learning_rate": 6.498375731458529e-07, - "loss": 0.0335, - "step": 1310 - }, - { - "epoch": 4.44, - "learning_rate": 6.420697650637753e-07, - "loss": 0.0368, - "step": 1311 - }, - { - "epoch": 4.45, - "learning_rate": 6.343471225312536e-07, - "loss": 0.0399, - "step": 1312 - }, - { - "epoch": 4.45, - "learning_rate": 6.266696828212071e-07, - "loss": 0.0374, - "step": 1313 - }, - { - "epoch": 4.45, - "learning_rate": 6.190374829883883e-07, - "loss": 0.0369, - "step": 1314 - }, - { - "epoch": 4.46, - "learning_rate": 6.114505598692011e-07, - "loss": 0.0384, - "step": 1315 - }, - { - "epoch": 4.46, - "learning_rate": 6.039089500815243e-07, - "loss": 0.0355, - "step": 1316 - }, - { - "epoch": 4.46, - "learning_rate": 5.964126900245359e-07, - "loss": 0.0353, - "step": 1317 - }, - { - "epoch": 4.47, - "learning_rate": 5.889618158785304e-07, - "loss": 0.0346, - "step": 1318 - }, - { - "epoch": 4.47, - "learning_rate": 5.815563636047539e-07, - "loss": 0.0359, - "step": 1319 - }, - { - "epoch": 4.47, - "learning_rate": 5.741963689452268e-07, - "loss": 0.0371, - "step": 1320 - }, - { - "epoch": 4.48, - "learning_rate": 5.668818674225684e-07, - "loss": 0.0354, - "step": 1321 - }, - { - "epoch": 4.48, - "learning_rate": 5.596128943398316e-07, - "loss": 0.0321, - "step": 1322 - }, - { - "epoch": 4.48, - "learning_rate": 5.523894847803235e-07, - "loss": 0.0354, - "step": 1323 - }, - { - "epoch": 4.49, - "learning_rate": 5.452116736074431e-07, - "loss": 0.0317, - "step": 1324 - }, - { - "epoch": 4.49, - "learning_rate": 5.380794954645141e-07, - "loss": 0.0388, - "step": 1325 - }, - { - "epoch": 4.49, - "learning_rate": 5.30992984774612e-07, - "loss": 0.036, - "step": 1326 - }, - { - "epoch": 4.5, - "learning_rate": 5.23952175740402e-07, - "loss": 0.0355, - "step": 1327 - }, - { - "epoch": 4.5, - "learning_rate": 5.169571023439712e-07, - "loss": 0.0368, - "step": 1328 - }, - { - "epoch": 4.51, - "learning_rate": 5.100077983466667e-07, - "loss": 0.0352, - "step": 1329 - }, - { - "epoch": 4.51, - "learning_rate": 5.031042972889311e-07, - "loss": 0.0352, - "step": 1330 - }, - { - "epoch": 4.51, - "learning_rate": 4.962466324901483e-07, - "loss": 0.0345, - "step": 1331 - }, - { - "epoch": 4.52, - "learning_rate": 4.894348370484648e-07, - "loss": 0.0383, - "step": 1332 - }, - { - "epoch": 4.52, - "learning_rate": 4.826689438406495e-07, - "loss": 0.0349, - "step": 1333 - }, - { - "epoch": 4.52, - "learning_rate": 4.759489855219235e-07, - "loss": 0.0359, - "step": 1334 - }, - { - "epoch": 4.53, - "learning_rate": 4.6927499452580574e-07, - "loss": 0.0362, - "step": 1335 - }, - { - "epoch": 4.53, - "learning_rate": 4.626470030639574e-07, - "loss": 0.0345, - "step": 1336 - }, - { - "epoch": 4.53, - "learning_rate": 4.5606504312602384e-07, - "loss": 0.0351, - "step": 1337 - }, - { - "epoch": 4.54, - "learning_rate": 4.4952914647948264e-07, - "loss": 0.0351, - "step": 1338 - }, - { - "epoch": 4.54, - "learning_rate": 4.4303934466948804e-07, - "loss": 0.0367, - "step": 1339 - }, - { - "epoch": 4.54, - "learning_rate": 4.365956690187256e-07, - "loss": 0.0373, - "step": 1340 - }, - { - "epoch": 4.55, - "learning_rate": 4.3019815062724567e-07, - "loss": 0.037, - "step": 1341 - }, - { - "epoch": 4.55, - "learning_rate": 4.2384682037233115e-07, - "loss": 0.0345, - "step": 1342 - }, - { - "epoch": 4.55, - "learning_rate": 4.1754170890833777e-07, - "loss": 0.0366, - "step": 1343 - }, - { - "epoch": 4.56, - "learning_rate": 4.112828466665486e-07, - "loss": 0.0337, - "step": 1344 - }, - { - "epoch": 4.56, - "learning_rate": 4.0507026385502747e-07, - "loss": 0.0342, - "step": 1345 - }, - { - "epoch": 4.56, - "learning_rate": 3.9890399045847127e-07, - "loss": 0.0357, - "step": 1346 - }, - { - "epoch": 4.57, - "learning_rate": 3.9278405623806914e-07, - "loss": 0.0336, - "step": 1347 - }, - { - "epoch": 4.57, - "learning_rate": 3.867104907313557e-07, - "loss": 0.0333, - "step": 1348 - }, - { - "epoch": 4.57, - "learning_rate": 3.806833232520746e-07, - "loss": 0.0375, - "step": 1349 - }, - { - "epoch": 4.58, - "learning_rate": 3.747025828900208e-07, - "loss": 0.0322, - "step": 1350 - }, - { - "epoch": 4.58, - "learning_rate": 3.687682985109209e-07, - "loss": 0.0335, - "step": 1351 - }, - { - "epoch": 4.58, - "learning_rate": 3.628804987562795e-07, - "loss": 0.0332, - "step": 1352 - }, - { - "epoch": 4.59, - "learning_rate": 3.5703921204324863e-07, - "loss": 0.0369, - "step": 1353 - }, - { - "epoch": 4.59, - "learning_rate": 3.5124446656448654e-07, - "loss": 0.0364, - "step": 1354 - }, - { - "epoch": 4.59, - "learning_rate": 3.454962902880199e-07, - "loss": 0.0348, - "step": 1355 - }, - { - "epoch": 4.6, - "learning_rate": 3.397947109571131e-07, - "loss": 0.0324, - "step": 1356 - }, - { - "epoch": 4.6, - "learning_rate": 3.3413975609013713e-07, - "loss": 0.0353, - "step": 1357 - }, - { - "epoch": 4.6, - "learning_rate": 3.2853145298042954e-07, - "loss": 0.0327, - "step": 1358 - }, - { - "epoch": 4.61, - "learning_rate": 3.2296982869616134e-07, - "loss": 0.038, - "step": 1359 - }, - { - "epoch": 4.61, - "learning_rate": 3.1745491008021603e-07, - "loss": 0.034, - "step": 1360 - }, - { - "epoch": 4.61, - "learning_rate": 3.1198672375005403e-07, - "loss": 0.0354, - "step": 1361 - }, - { - "epoch": 4.62, - "learning_rate": 3.06565296097584e-07, - "loss": 0.0359, - "step": 1362 - }, - { - "epoch": 4.62, - "learning_rate": 3.0119065328903517e-07, - "loss": 0.0336, - "step": 1363 - }, - { - "epoch": 4.62, - "learning_rate": 2.9586282126483625e-07, - "loss": 0.0329, - "step": 1364 - }, - { - "epoch": 4.63, - "learning_rate": 2.905818257394799e-07, - "loss": 0.0343, - "step": 1365 - }, - { - "epoch": 4.63, - "learning_rate": 2.853476922014098e-07, - "loss": 0.0357, - "step": 1366 - }, - { - "epoch": 4.63, - "learning_rate": 2.801604459128926e-07, - "loss": 0.0359, - "step": 1367 - }, - { - "epoch": 4.64, - "learning_rate": 2.75020111909895e-07, - "loss": 0.037, - "step": 1368 - }, - { - "epoch": 4.64, - "learning_rate": 2.6992671500196134e-07, - "loss": 0.0352, - "step": 1369 - }, - { - "epoch": 4.64, - "learning_rate": 2.6488027977210175e-07, - "loss": 0.0354, - "step": 1370 - }, - { - "epoch": 4.65, - "learning_rate": 2.5988083057666534e-07, - "loss": 0.0326, - "step": 1371 - }, - { - "epoch": 4.65, - "learning_rate": 2.5492839154522495e-07, - "loss": 0.0375, - "step": 1372 - }, - { - "epoch": 4.65, - "learning_rate": 2.5002298658046484e-07, - "loss": 0.0348, - "step": 1373 - }, - { - "epoch": 4.66, - "learning_rate": 2.4516463935805644e-07, - "loss": 0.0349, - "step": 1374 - }, - { - "epoch": 4.66, - "learning_rate": 2.4035337332655504e-07, - "loss": 0.0354, - "step": 1375 - }, - { - "epoch": 4.66, - "learning_rate": 2.355892117072789e-07, - "loss": 0.0362, - "step": 1376 - }, - { - "epoch": 4.67, - "learning_rate": 2.308721774941991e-07, - "loss": 0.036, - "step": 1377 - }, - { - "epoch": 4.67, - "learning_rate": 2.262022934538266e-07, - "loss": 0.0342, - "step": 1378 - }, - { - "epoch": 4.67, - "learning_rate": 2.2157958212510877e-07, - "loss": 0.0348, - "step": 1379 - }, - { - "epoch": 4.68, - "learning_rate": 2.1700406581931398e-07, - "loss": 0.0356, - "step": 1380 - }, - { - "epoch": 4.68, - "learning_rate": 2.124757666199273e-07, - "loss": 0.0377, - "step": 1381 - }, - { - "epoch": 4.68, - "learning_rate": 2.079947063825427e-07, - "loss": 0.0316, - "step": 1382 - }, - { - "epoch": 4.69, - "learning_rate": 2.035609067347566e-07, - "loss": 0.0346, - "step": 1383 - }, - { - "epoch": 4.69, - "learning_rate": 1.9917438907606556e-07, - "loss": 0.0369, - "step": 1384 - }, - { - "epoch": 4.69, - "learning_rate": 1.9483517457776436e-07, - "loss": 0.0362, - "step": 1385 - }, - { - "epoch": 4.7, - "learning_rate": 1.9054328418283808e-07, - "loss": 0.0355, - "step": 1386 - }, - { - "epoch": 4.7, - "learning_rate": 1.8629873860586567e-07, - "loss": 0.0381, - "step": 1387 - }, - { - "epoch": 4.71, - "learning_rate": 1.8210155833291998e-07, - "loss": 0.034, - "step": 1388 - }, - { - "epoch": 4.71, - "learning_rate": 1.7795176362146783e-07, - "loss": 0.0312, - "step": 1389 - }, - { - "epoch": 4.71, - "learning_rate": 1.7384937450026895e-07, - "loss": 0.0338, - "step": 1390 - }, - { - "epoch": 4.72, - "learning_rate": 1.6979441076928837e-07, - "loss": 0.0335, - "step": 1391 - }, - { - "epoch": 4.72, - "learning_rate": 1.6578689199958753e-07, - "loss": 0.0358, - "step": 1392 - }, - { - "epoch": 4.72, - "learning_rate": 1.6182683753324435e-07, - "loss": 0.0376, - "step": 1393 - }, - { - "epoch": 4.73, - "learning_rate": 1.5791426648324893e-07, - "loss": 0.0342, - "step": 1394 - }, - { - "epoch": 4.73, - "learning_rate": 1.5404919773341576e-07, - "loss": 0.036, - "step": 1395 - }, - { - "epoch": 4.73, - "learning_rate": 1.5023164993829277e-07, - "loss": 0.0343, - "step": 1396 - }, - { - "epoch": 4.74, - "learning_rate": 1.464616415230702e-07, - "loss": 0.0365, - "step": 1397 - }, - { - "epoch": 4.74, - "learning_rate": 1.4273919068349184e-07, - "loss": 0.0423, - "step": 1398 - }, - { - "epoch": 4.74, - "learning_rate": 1.3906431538576626e-07, - "loss": 0.0365, - "step": 1399 - }, - { - "epoch": 4.75, - "learning_rate": 1.354370333664845e-07, - "loss": 0.0361, - "step": 1400 - }, - { - "epoch": 4.75, - "learning_rate": 1.3185736213252808e-07, - "loss": 0.0317, - "step": 1401 - }, - { - "epoch": 4.75, - "learning_rate": 1.2832531896098788e-07, - "loss": 0.0357, - "step": 1402 - }, - { - "epoch": 4.76, - "learning_rate": 1.2484092089908307e-07, - "loss": 0.0309, - "step": 1403 - }, - { - "epoch": 4.76, - "learning_rate": 1.2140418476407457e-07, - "loss": 0.0319, - "step": 1404 - }, - { - "epoch": 4.76, - "learning_rate": 1.1801512714318286e-07, - "loss": 0.0347, - "step": 1405 - }, - { - "epoch": 4.77, - "learning_rate": 1.1467376439351474e-07, - "loss": 0.0346, - "step": 1406 - }, - { - "epoch": 4.77, - "learning_rate": 1.113801126419789e-07, - "loss": 0.0355, - "step": 1407 - }, - { - "epoch": 4.77, - "learning_rate": 1.0813418778521046e-07, - "loss": 0.0355, - "step": 1408 - }, - { - "epoch": 4.78, - "learning_rate": 1.0493600548948879e-07, - "loss": 0.0383, - "step": 1409 - }, - { - "epoch": 4.78, - "learning_rate": 1.0178558119067316e-07, - "loss": 0.0371, - "step": 1410 - }, - { - "epoch": 4.78, - "learning_rate": 9.8682930094115e-08, - "loss": 0.0363, - "step": 1411 - }, - { - "epoch": 4.79, - "learning_rate": 9.562806717459572e-08, - "loss": 0.0356, - "step": 1412 - }, - { - "epoch": 4.79, - "learning_rate": 9.262100717624678e-08, - "loss": 0.0354, - "step": 1413 - }, - { - "epoch": 4.79, - "learning_rate": 8.966176461248422e-08, - "loss": 0.036, - "step": 1414 - }, - { - "epoch": 4.8, - "learning_rate": 8.675035376593088e-08, - "loss": 0.0364, - "step": 1415 - }, - { - "epoch": 4.8, - "learning_rate": 8.388678868835653e-08, - "loss": 0.0372, - "step": 1416 - }, - { - "epoch": 4.8, - "learning_rate": 8.107108320060675e-08, - "loss": 0.0353, - "step": 1417 - }, - { - "epoch": 4.81, - "learning_rate": 7.830325089253077e-08, - "loss": 0.037, - "step": 1418 - }, - { - "epoch": 4.81, - "learning_rate": 7.558330512292378e-08, - "loss": 0.0361, - "step": 1419 - }, - { - "epoch": 4.81, - "learning_rate": 7.291125901946027e-08, - "loss": 0.0343, - "step": 1420 - }, - { - "epoch": 4.82, - "learning_rate": 7.028712547862526e-08, - "loss": 0.0348, - "step": 1421 - }, - { - "epoch": 4.82, - "learning_rate": 6.771091716566091e-08, - "loss": 0.0369, - "step": 1422 - }, - { - "epoch": 4.82, - "learning_rate": 6.51826465144978e-08, - "loss": 0.0335, - "step": 1423 - }, - { - "epoch": 4.83, - "learning_rate": 6.27023257276993e-08, - "loss": 0.0375, - "step": 1424 - }, - { - "epoch": 4.83, - "learning_rate": 6.026996677640062e-08, - "loss": 0.0342, - "step": 1425 - }, - { - "epoch": 4.83, - "learning_rate": 5.788558140025213e-08, - "loss": 0.0356, - "step": 1426 - }, - { - "epoch": 4.84, - "learning_rate": 5.5549181107362734e-08, - "loss": 0.0371, - "step": 1427 - }, - { - "epoch": 4.84, - "learning_rate": 5.326077717424216e-08, - "loss": 0.0353, - "step": 1428 - }, - { - "epoch": 4.84, - "learning_rate": 5.102038064575099e-08, - "loss": 0.0403, - "step": 1429 - }, - { - "epoch": 4.85, - "learning_rate": 4.882800233504292e-08, - "loss": 0.0362, - "step": 1430 - }, - { - "epoch": 4.85, - "learning_rate": 4.6683652823513725e-08, - "loss": 0.0342, - "step": 1431 - }, - { - "epoch": 4.85, - "learning_rate": 4.458734246075236e-08, - "loss": 0.0349, - "step": 1432 - }, - { - "epoch": 4.86, - "learning_rate": 4.253908136448881e-08, - "loss": 0.0352, - "step": 1433 - }, - { - "epoch": 4.86, - "learning_rate": 4.053887942054524e-08, - "loss": 0.0357, - "step": 1434 - }, - { - "epoch": 4.86, - "learning_rate": 3.858674628278825e-08, - "loss": 0.0346, - "step": 1435 - }, - { - "epoch": 4.87, - "learning_rate": 3.668269137308666e-08, - "loss": 0.0362, - "step": 1436 - }, - { - "epoch": 4.87, - "learning_rate": 3.482672388125719e-08, - "loss": 0.0391, - "step": 1437 - }, - { - "epoch": 4.87, - "learning_rate": 3.3018852765027696e-08, - "loss": 0.0343, - "step": 1438 - }, - { - "epoch": 4.88, - "learning_rate": 3.125908674999289e-08, - "loss": 0.0331, - "step": 1439 - }, - { - "epoch": 4.88, - "learning_rate": 2.9547434329568747e-08, - "loss": 0.0353, - "step": 1440 - }, - { - "epoch": 4.88, - "learning_rate": 2.7883903764953647e-08, - "loss": 0.0386, - "step": 1441 - }, - { - "epoch": 4.89, - "learning_rate": 2.6268503085089547e-08, - "loss": 0.0357, - "step": 1442 - }, - { - "epoch": 4.89, - "learning_rate": 2.470124008661978e-08, - "loss": 0.0348, - "step": 1443 - }, - { - "epoch": 4.89, - "learning_rate": 2.318212233385686e-08, - "loss": 0.036, - "step": 1444 - }, - { - "epoch": 4.9, - "learning_rate": 2.171115715874139e-08, - "loss": 0.0363, - "step": 1445 - }, - { - "epoch": 4.9, - "learning_rate": 2.028835166080767e-08, - "loss": 0.033, - "step": 1446 - }, - { - "epoch": 4.91, - "learning_rate": 1.8913712707149255e-08, - "loss": 0.0358, - "step": 1447 - }, - { - "epoch": 4.91, - "learning_rate": 1.7587246932389003e-08, - "loss": 0.039, - "step": 1448 - }, - { - "epoch": 4.91, - "learning_rate": 1.630896073864352e-08, - "loss": 0.0348, - "step": 1449 - }, - { - "epoch": 4.92, - "learning_rate": 1.5078860295490995e-08, - "loss": 0.0356, - "step": 1450 - }, - { - "epoch": 4.92, - "learning_rate": 1.3896951539945635e-08, - "loss": 0.033, - "step": 1451 - }, - { - "epoch": 4.92, - "learning_rate": 1.2763240176427715e-08, - "loss": 0.0364, - "step": 1452 - }, - { - "epoch": 4.93, - "learning_rate": 1.1677731676733584e-08, - "loss": 0.0362, - "step": 1453 - }, - { - "epoch": 4.93, - "learning_rate": 1.0640431280013463e-08, - "loss": 0.0362, - "step": 1454 - }, - { - "epoch": 4.93, - "learning_rate": 9.651343992740369e-09, - "loss": 0.0358, - "step": 1455 - }, - { - "epoch": 4.94, - "learning_rate": 8.71047458869234e-09, - "loss": 0.0383, - "step": 1456 - }, - { - "epoch": 4.94, - "learning_rate": 7.817827608924689e-09, - "loss": 0.0363, - "step": 1457 - }, - { - "epoch": 4.94, - "learning_rate": 6.973407361750006e-09, - "loss": 0.0317, - "step": 1458 - }, - { - "epoch": 4.95, - "learning_rate": 6.1772179227181926e-09, - "loss": 0.0321, - "step": 1459 - }, - { - "epoch": 4.95, - "learning_rate": 5.429263134594242e-09, - "loss": 0.0348, - "step": 1460 - }, - { - "epoch": 4.95, - "learning_rate": 4.7295466073427055e-09, - "loss": 0.0344, - "step": 1461 - }, - { - "epoch": 4.96, - "learning_rate": 4.0780717181077015e-09, - "loss": 0.036, - "step": 1462 - }, - { - "epoch": 4.96, - "learning_rate": 3.474841611197377e-09, - "loss": 0.0351, - "step": 1463 - }, - { - "epoch": 4.96, - "learning_rate": 2.9198591980705847e-09, - "loss": 0.0334, - "step": 1464 - }, - { - "epoch": 4.97, - "learning_rate": 2.4131271573191172e-09, - "loss": 0.0362, - "step": 1465 - }, - { - "epoch": 4.97, - "learning_rate": 1.9546479346588265e-09, - "loss": 0.0367, - "step": 1466 - }, - { - "epoch": 4.97, - "learning_rate": 1.5444237429140806e-09, - "loss": 0.034, - "step": 1467 - }, - { - "epoch": 4.98, - "learning_rate": 1.182456562012213e-09, - "loss": 0.0328, - "step": 1468 - }, - { - "epoch": 4.98, - "learning_rate": 8.687481389657582e-10, - "loss": 0.0376, - "step": 1469 - }, - { - "epoch": 4.98, - "learning_rate": 6.032999878735624e-10, - "loss": 0.0335, - "step": 1470 - }, - { - "epoch": 4.99, - "learning_rate": 3.861133899063507e-10, - "loss": 0.0329, - "step": 1471 - }, - { - "epoch": 4.99, - "learning_rate": 2.171893933033964e-10, - "loss": 0.0385, - "step": 1472 - }, - { - "epoch": 4.99, - "learning_rate": 9.652881336696951e-11, - "loss": 0.034, - "step": 1473 - }, - { - "epoch": 5.0, - "learning_rate": 2.4132232460116754e-11, - "loss": 0.0342, - "step": 1474 - }, - { - "epoch": 5.0, - "learning_rate": 0.0, - "loss": 0.0257, - "step": 1475 - }, - { - "epoch": 5.0, - "step": 1475, - "total_flos": 164477023027200.0, - "train_loss": 0.44021065391852693, - "train_runtime": 10298.8214, - "train_samples_per_second": 13.723, - "train_steps_per_second": 0.143 - } - ], - "logging_steps": 1.0, - "max_steps": 1475, - "num_input_tokens_seen": 0, - "num_train_epochs": 5, - "save_steps": 50000, - "total_flos": 164477023027200.0, - "train_batch_size": 24, - "trial_name": null, - "trial_params": null -}