{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.05239030779305828, "global_step": 200, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 6.9856793573174994e-09, "loss": 1.5302, "step": 1 }, { "epoch": 0.0, "learning_rate": 1.3971358714634999e-08, "loss": 1.3991, "step": 2 }, { "epoch": 0.0, "learning_rate": 2.09570380719525e-08, "loss": 1.5873, "step": 3 }, { "epoch": 0.0, "learning_rate": 2.7942717429269998e-08, "loss": 1.4977, "step": 4 }, { "epoch": 0.0, "learning_rate": 3.49283967865875e-08, "loss": 1.4561, "step": 5 }, { "epoch": 0.0, "learning_rate": 4.1914076143905e-08, "loss": 1.4481, "step": 6 }, { "epoch": 0.0, "learning_rate": 4.8899755501222494e-08, "loss": 1.5704, "step": 7 }, { "epoch": 0.0, "learning_rate": 5.5885434858539996e-08, "loss": 1.396, "step": 8 }, { "epoch": 0.0, "learning_rate": 6.28711142158575e-08, "loss": 1.4132, "step": 9 }, { "epoch": 0.0, "learning_rate": 6.9856793573175e-08, "loss": 1.3079, "step": 10 }, { "epoch": 0.0, "learning_rate": 7.684247293049249e-08, "loss": 1.381, "step": 11 }, { "epoch": 0.0, "learning_rate": 8.382815228781e-08, "loss": 1.5932, "step": 12 }, { "epoch": 0.0, "learning_rate": 9.08138316451275e-08, "loss": 1.4781, "step": 13 }, { "epoch": 0.0, "learning_rate": 9.779951100244499e-08, "loss": 1.4499, "step": 14 }, { "epoch": 0.0, "learning_rate": 1.047851903597625e-07, "loss": 1.3174, "step": 15 }, { "epoch": 0.0, "learning_rate": 1.1177086971707999e-07, "loss": 1.3322, "step": 16 }, { "epoch": 0.0, "learning_rate": 1.1875654907439749e-07, "loss": 1.3133, "step": 17 }, { "epoch": 0.0, "learning_rate": 1.25742228431715e-07, "loss": 1.4163, "step": 18 }, { "epoch": 0.0, "learning_rate": 1.327279077890325e-07, "loss": 1.3745, "step": 19 }, { "epoch": 0.01, "learning_rate": 1.3971358714635e-07, "loss": 1.5562, "step": 20 }, { "epoch": 0.01, "learning_rate": 1.466992665036675e-07, "loss": 1.5255, "step": 21 }, { "epoch": 0.01, "learning_rate": 1.5368494586098498e-07, "loss": 1.6313, "step": 22 }, { "epoch": 0.01, "learning_rate": 1.606706252183025e-07, "loss": 1.3289, "step": 23 }, { "epoch": 0.01, "learning_rate": 1.6765630457562e-07, "loss": 1.3399, "step": 24 }, { "epoch": 0.01, "learning_rate": 1.746419839329375e-07, "loss": 1.3479, "step": 25 }, { "epoch": 0.01, "learning_rate": 1.81627663290255e-07, "loss": 1.3508, "step": 26 }, { "epoch": 0.01, "learning_rate": 1.8861334264757248e-07, "loss": 1.5083, "step": 27 }, { "epoch": 0.01, "learning_rate": 1.9559902200488998e-07, "loss": 1.3118, "step": 28 }, { "epoch": 0.01, "learning_rate": 2.025847013622075e-07, "loss": 1.3354, "step": 29 }, { "epoch": 0.01, "learning_rate": 2.09570380719525e-07, "loss": 1.4193, "step": 30 }, { "epoch": 0.01, "learning_rate": 2.165560600768425e-07, "loss": 1.3794, "step": 31 }, { "epoch": 0.01, "learning_rate": 2.2354173943415998e-07, "loss": 1.4642, "step": 32 }, { "epoch": 0.01, "learning_rate": 2.3052741879147748e-07, "loss": 1.5252, "step": 33 }, { "epoch": 0.01, "learning_rate": 2.3751309814879497e-07, "loss": 1.2811, "step": 34 }, { "epoch": 0.01, "learning_rate": 2.444987775061125e-07, "loss": 1.5713, "step": 35 }, { "epoch": 0.01, "learning_rate": 2.5148445686343e-07, "loss": 1.3266, "step": 36 }, { "epoch": 0.01, "learning_rate": 2.584701362207475e-07, "loss": 1.2845, "step": 37 }, { "epoch": 0.01, "learning_rate": 2.65455815578065e-07, "loss": 1.3171, "step": 38 }, { "epoch": 0.01, "learning_rate": 2.724414949353825e-07, "loss": 1.569, "step": 39 }, { "epoch": 0.01, "learning_rate": 2.794271742927e-07, "loss": 1.3662, "step": 40 }, { "epoch": 0.01, "learning_rate": 2.864128536500175e-07, "loss": 1.3673, "step": 41 }, { "epoch": 0.01, "learning_rate": 2.93398533007335e-07, "loss": 1.5369, "step": 42 }, { "epoch": 0.01, "learning_rate": 3.003842123646525e-07, "loss": 1.2998, "step": 43 }, { "epoch": 0.01, "learning_rate": 3.0736989172196997e-07, "loss": 1.3358, "step": 44 }, { "epoch": 0.01, "learning_rate": 3.1435557107928746e-07, "loss": 1.2411, "step": 45 }, { "epoch": 0.01, "learning_rate": 3.21341250436605e-07, "loss": 1.3119, "step": 46 }, { "epoch": 0.01, "learning_rate": 3.2832692979392245e-07, "loss": 1.4592, "step": 47 }, { "epoch": 0.01, "learning_rate": 3.3531260915124e-07, "loss": 1.3613, "step": 48 }, { "epoch": 0.01, "learning_rate": 3.4229828850855744e-07, "loss": 1.3615, "step": 49 }, { "epoch": 0.01, "learning_rate": 3.49283967865875e-07, "loss": 1.3016, "step": 50 }, { "epoch": 0.01, "learning_rate": 3.562696472231925e-07, "loss": 1.4594, "step": 51 }, { "epoch": 0.01, "learning_rate": 3.6325532658051e-07, "loss": 1.4994, "step": 52 }, { "epoch": 0.01, "learning_rate": 3.7024100593782747e-07, "loss": 1.4328, "step": 53 }, { "epoch": 0.01, "learning_rate": 3.7722668529514497e-07, "loss": 1.4001, "step": 54 }, { "epoch": 0.01, "learning_rate": 3.8421236465246246e-07, "loss": 1.3456, "step": 55 }, { "epoch": 0.01, "learning_rate": 3.9119804400977996e-07, "loss": 1.3156, "step": 56 }, { "epoch": 0.01, "learning_rate": 3.981837233670975e-07, "loss": 1.4528, "step": 57 }, { "epoch": 0.02, "learning_rate": 4.05169402724415e-07, "loss": 1.2765, "step": 58 }, { "epoch": 0.02, "learning_rate": 4.121550820817325e-07, "loss": 1.3098, "step": 59 }, { "epoch": 0.02, "learning_rate": 4.1914076143905e-07, "loss": 1.5001, "step": 60 }, { "epoch": 0.02, "learning_rate": 4.261264407963675e-07, "loss": 1.5602, "step": 61 }, { "epoch": 0.02, "learning_rate": 4.33112120153685e-07, "loss": 1.3146, "step": 62 }, { "epoch": 0.02, "learning_rate": 4.4009779951100247e-07, "loss": 1.1927, "step": 63 }, { "epoch": 0.02, "learning_rate": 4.4708347886831996e-07, "loss": 1.3499, "step": 64 }, { "epoch": 0.02, "learning_rate": 4.540691582256375e-07, "loss": 1.4147, "step": 65 }, { "epoch": 0.02, "learning_rate": 4.6105483758295495e-07, "loss": 1.3828, "step": 66 }, { "epoch": 0.02, "learning_rate": 4.680405169402725e-07, "loss": 1.4336, "step": 67 }, { "epoch": 0.02, "learning_rate": 4.7502619629758994e-07, "loss": 1.3039, "step": 68 }, { "epoch": 0.02, "learning_rate": 4.820118756549074e-07, "loss": 1.2953, "step": 69 }, { "epoch": 0.02, "learning_rate": 4.88997555012225e-07, "loss": 1.2902, "step": 70 }, { "epoch": 0.02, "learning_rate": 4.959832343695424e-07, "loss": 1.3334, "step": 71 }, { "epoch": 0.02, "learning_rate": 5.0296891372686e-07, "loss": 1.5598, "step": 72 }, { "epoch": 0.02, "learning_rate": 5.099545930841775e-07, "loss": 1.3344, "step": 73 }, { "epoch": 0.02, "learning_rate": 5.16940272441495e-07, "loss": 1.2667, "step": 74 }, { "epoch": 0.02, "learning_rate": 5.239259517988125e-07, "loss": 1.2021, "step": 75 }, { "epoch": 0.02, "learning_rate": 5.3091163115613e-07, "loss": 1.4128, "step": 76 }, { "epoch": 0.02, "learning_rate": 5.378973105134475e-07, "loss": 1.5772, "step": 77 }, { "epoch": 0.02, "learning_rate": 5.44882989870765e-07, "loss": 1.3694, "step": 78 }, { "epoch": 0.02, "learning_rate": 5.518686692280825e-07, "loss": 1.4965, "step": 79 }, { "epoch": 0.02, "learning_rate": 5.588543485854e-07, "loss": 1.277, "step": 80 }, { "epoch": 0.02, "learning_rate": 5.658400279427175e-07, "loss": 1.3585, "step": 81 }, { "epoch": 0.02, "learning_rate": 5.72825707300035e-07, "loss": 1.3944, "step": 82 }, { "epoch": 0.02, "learning_rate": 5.798113866573525e-07, "loss": 1.4473, "step": 83 }, { "epoch": 0.02, "learning_rate": 5.8679706601467e-07, "loss": 1.1716, "step": 84 }, { "epoch": 0.02, "learning_rate": 5.937827453719875e-07, "loss": 1.3948, "step": 85 }, { "epoch": 0.02, "learning_rate": 6.00768424729305e-07, "loss": 1.3183, "step": 86 }, { "epoch": 0.02, "learning_rate": 6.077541040866225e-07, "loss": 1.4303, "step": 87 }, { "epoch": 0.02, "learning_rate": 6.147397834439399e-07, "loss": 1.2323, "step": 88 }, { "epoch": 0.02, "learning_rate": 6.217254628012575e-07, "loss": 1.4077, "step": 89 }, { "epoch": 0.02, "learning_rate": 6.287111421585749e-07, "loss": 1.3453, "step": 90 }, { "epoch": 0.02, "learning_rate": 6.356968215158925e-07, "loss": 1.4308, "step": 91 }, { "epoch": 0.02, "learning_rate": 6.4268250087321e-07, "loss": 1.3942, "step": 92 }, { "epoch": 0.02, "learning_rate": 6.496681802305274e-07, "loss": 1.7588, "step": 93 }, { "epoch": 0.02, "learning_rate": 6.566538595878449e-07, "loss": 1.3247, "step": 94 }, { "epoch": 0.02, "learning_rate": 6.636395389451625e-07, "loss": 1.2695, "step": 95 }, { "epoch": 0.03, "learning_rate": 6.7062521830248e-07, "loss": 1.2933, "step": 96 }, { "epoch": 0.03, "learning_rate": 6.776108976597975e-07, "loss": 1.3273, "step": 97 }, { "epoch": 0.03, "learning_rate": 6.845965770171149e-07, "loss": 1.3102, "step": 98 }, { "epoch": 0.03, "learning_rate": 6.915822563744325e-07, "loss": 1.3494, "step": 99 }, { "epoch": 0.03, "learning_rate": 6.9856793573175e-07, "loss": 1.3154, "step": 100 }, { "epoch": 0.03, "learning_rate": 7.055536150890676e-07, "loss": 1.2514, "step": 101 }, { "epoch": 0.03, "learning_rate": 7.12539294446385e-07, "loss": 1.2672, "step": 102 }, { "epoch": 0.03, "learning_rate": 7.195249738037025e-07, "loss": 1.7321, "step": 103 }, { "epoch": 0.03, "learning_rate": 7.2651065316102e-07, "loss": 1.2989, "step": 104 }, { "epoch": 0.03, "learning_rate": 7.334963325183376e-07, "loss": 1.5735, "step": 105 }, { "epoch": 0.03, "learning_rate": 7.404820118756549e-07, "loss": 1.2504, "step": 106 }, { "epoch": 0.03, "learning_rate": 7.474676912329724e-07, "loss": 1.2209, "step": 107 }, { "epoch": 0.03, "learning_rate": 7.544533705902899e-07, "loss": 1.31, "step": 108 }, { "epoch": 0.03, "learning_rate": 7.614390499476075e-07, "loss": 1.3797, "step": 109 }, { "epoch": 0.03, "learning_rate": 7.684247293049249e-07, "loss": 1.3813, "step": 110 }, { "epoch": 0.03, "learning_rate": 7.754104086622424e-07, "loss": 1.3116, "step": 111 }, { "epoch": 0.03, "learning_rate": 7.823960880195599e-07, "loss": 1.2662, "step": 112 }, { "epoch": 0.03, "learning_rate": 7.893817673768775e-07, "loss": 1.2918, "step": 113 }, { "epoch": 0.03, "learning_rate": 7.96367446734195e-07, "loss": 1.288, "step": 114 }, { "epoch": 0.03, "learning_rate": 8.033531260915124e-07, "loss": 1.2962, "step": 115 }, { "epoch": 0.03, "learning_rate": 8.1033880544883e-07, "loss": 1.4929, "step": 116 }, { "epoch": 0.03, "learning_rate": 8.173244848061475e-07, "loss": 1.4425, "step": 117 }, { "epoch": 0.03, "learning_rate": 8.24310164163465e-07, "loss": 1.4805, "step": 118 }, { "epoch": 0.03, "learning_rate": 8.312958435207824e-07, "loss": 1.6595, "step": 119 }, { "epoch": 0.03, "learning_rate": 8.382815228781e-07, "loss": 1.3928, "step": 120 }, { "epoch": 0.03, "learning_rate": 8.452672022354175e-07, "loss": 1.3824, "step": 121 }, { "epoch": 0.03, "learning_rate": 8.52252881592735e-07, "loss": 1.4096, "step": 122 }, { "epoch": 0.03, "learning_rate": 8.592385609500524e-07, "loss": 1.4056, "step": 123 }, { "epoch": 0.03, "learning_rate": 8.6622424030737e-07, "loss": 1.4939, "step": 124 }, { "epoch": 0.03, "learning_rate": 8.732099196646874e-07, "loss": 1.4946, "step": 125 }, { "epoch": 0.03, "learning_rate": 8.801955990220049e-07, "loss": 1.2152, "step": 126 }, { "epoch": 0.03, "learning_rate": 8.871812783793225e-07, "loss": 1.3495, "step": 127 }, { "epoch": 0.03, "learning_rate": 8.941669577366399e-07, "loss": 1.3502, "step": 128 }, { "epoch": 0.03, "learning_rate": 9.011526370939574e-07, "loss": 1.4499, "step": 129 }, { "epoch": 0.03, "learning_rate": 9.08138316451275e-07, "loss": 1.404, "step": 130 }, { "epoch": 0.03, "learning_rate": 9.151239958085925e-07, "loss": 1.2707, "step": 131 }, { "epoch": 0.03, "learning_rate": 9.221096751659099e-07, "loss": 1.3779, "step": 132 }, { "epoch": 0.03, "learning_rate": 9.290953545232274e-07, "loss": 1.4176, "step": 133 }, { "epoch": 0.04, "learning_rate": 9.36081033880545e-07, "loss": 1.3269, "step": 134 }, { "epoch": 0.04, "learning_rate": 9.430667132378625e-07, "loss": 1.4951, "step": 135 }, { "epoch": 0.04, "learning_rate": 9.500523925951799e-07, "loss": 1.3809, "step": 136 }, { "epoch": 0.04, "learning_rate": 9.570380719524975e-07, "loss": 1.4824, "step": 137 }, { "epoch": 0.04, "learning_rate": 9.640237513098149e-07, "loss": 1.4929, "step": 138 }, { "epoch": 0.04, "learning_rate": 9.710094306671325e-07, "loss": 1.3353, "step": 139 }, { "epoch": 0.04, "learning_rate": 9.7799511002445e-07, "loss": 1.6328, "step": 140 }, { "epoch": 0.04, "learning_rate": 9.849807893817675e-07, "loss": 1.4769, "step": 141 }, { "epoch": 0.04, "learning_rate": 9.919664687390849e-07, "loss": 1.5474, "step": 142 }, { "epoch": 0.04, "learning_rate": 9.989521480964025e-07, "loss": 1.36, "step": 143 }, { "epoch": 0.04, "learning_rate": 1.00593782745372e-06, "loss": 1.6957, "step": 144 }, { "epoch": 0.04, "learning_rate": 1.0129235068110374e-06, "loss": 1.1635, "step": 145 }, { "epoch": 0.04, "learning_rate": 1.019909186168355e-06, "loss": 1.4348, "step": 146 }, { "epoch": 0.04, "learning_rate": 1.0268948655256724e-06, "loss": 1.4559, "step": 147 }, { "epoch": 0.04, "learning_rate": 1.03388054488299e-06, "loss": 1.3772, "step": 148 }, { "epoch": 0.04, "learning_rate": 1.0408662242403074e-06, "loss": 1.3404, "step": 149 }, { "epoch": 0.04, "learning_rate": 1.047851903597625e-06, "loss": 1.3518, "step": 150 }, { "epoch": 0.04, "learning_rate": 1.0548375829549424e-06, "loss": 1.2911, "step": 151 }, { "epoch": 0.04, "learning_rate": 1.06182326231226e-06, "loss": 1.3891, "step": 152 }, { "epoch": 0.04, "learning_rate": 1.0688089416695776e-06, "loss": 1.3481, "step": 153 }, { "epoch": 0.04, "learning_rate": 1.075794621026895e-06, "loss": 1.2061, "step": 154 }, { "epoch": 0.04, "learning_rate": 1.0827803003842124e-06, "loss": 1.3228, "step": 155 }, { "epoch": 0.04, "learning_rate": 1.08976597974153e-06, "loss": 1.4932, "step": 156 }, { "epoch": 0.04, "learning_rate": 1.0967516590988476e-06, "loss": 1.0902, "step": 157 }, { "epoch": 0.04, "learning_rate": 1.103737338456165e-06, "loss": 1.4818, "step": 158 }, { "epoch": 0.04, "learning_rate": 1.1107230178134824e-06, "loss": 1.3129, "step": 159 }, { "epoch": 0.04, "learning_rate": 1.1177086971708e-06, "loss": 1.3357, "step": 160 }, { "epoch": 0.04, "learning_rate": 1.1246943765281176e-06, "loss": 1.609, "step": 161 }, { "epoch": 0.04, "learning_rate": 1.131680055885435e-06, "loss": 1.2163, "step": 162 }, { "epoch": 0.04, "learning_rate": 1.1386657352427523e-06, "loss": 1.3282, "step": 163 }, { "epoch": 0.04, "learning_rate": 1.14565141460007e-06, "loss": 1.3166, "step": 164 }, { "epoch": 0.04, "learning_rate": 1.1526370939573875e-06, "loss": 1.3252, "step": 165 }, { "epoch": 0.04, "learning_rate": 1.159622773314705e-06, "loss": 1.4827, "step": 166 }, { "epoch": 0.04, "learning_rate": 1.1666084526720223e-06, "loss": 1.4112, "step": 167 }, { "epoch": 0.04, "learning_rate": 1.17359413202934e-06, "loss": 1.3811, "step": 168 }, { "epoch": 0.04, "learning_rate": 1.1805798113866575e-06, "loss": 1.2234, "step": 169 }, { "epoch": 0.04, "learning_rate": 1.187565490743975e-06, "loss": 1.5784, "step": 170 }, { "epoch": 0.04, "learning_rate": 1.1945511701012923e-06, "loss": 1.4294, "step": 171 }, { "epoch": 0.05, "learning_rate": 1.20153684945861e-06, "loss": 1.2927, "step": 172 }, { "epoch": 0.05, "learning_rate": 1.2085225288159275e-06, "loss": 1.3468, "step": 173 }, { "epoch": 0.05, "learning_rate": 1.215508208173245e-06, "loss": 1.39, "step": 174 }, { "epoch": 0.05, "learning_rate": 1.2224938875305625e-06, "loss": 1.6262, "step": 175 }, { "epoch": 0.05, "learning_rate": 1.2294795668878799e-06, "loss": 1.273, "step": 176 }, { "epoch": 0.05, "learning_rate": 1.2364652462451975e-06, "loss": 1.3019, "step": 177 }, { "epoch": 0.05, "learning_rate": 1.243450925602515e-06, "loss": 1.4669, "step": 178 }, { "epoch": 0.05, "learning_rate": 1.2504366049598325e-06, "loss": 1.388, "step": 179 }, { "epoch": 0.05, "learning_rate": 1.2574222843171499e-06, "loss": 1.2819, "step": 180 }, { "epoch": 0.05, "learning_rate": 1.2644079636744675e-06, "loss": 1.4056, "step": 181 }, { "epoch": 0.05, "learning_rate": 1.271393643031785e-06, "loss": 1.4843, "step": 182 }, { "epoch": 0.05, "learning_rate": 1.2783793223891024e-06, "loss": 1.2905, "step": 183 }, { "epoch": 0.05, "learning_rate": 1.28536500174642e-06, "loss": 1.4571, "step": 184 }, { "epoch": 0.05, "learning_rate": 1.2923506811037376e-06, "loss": 1.3495, "step": 185 }, { "epoch": 0.05, "learning_rate": 1.2993363604610548e-06, "loss": 1.2697, "step": 186 }, { "epoch": 0.05, "learning_rate": 1.3063220398183724e-06, "loss": 1.3934, "step": 187 }, { "epoch": 0.05, "learning_rate": 1.3133077191756898e-06, "loss": 1.2987, "step": 188 }, { "epoch": 0.05, "learning_rate": 1.3202933985330074e-06, "loss": 1.2335, "step": 189 }, { "epoch": 0.05, "learning_rate": 1.327279077890325e-06, "loss": 1.3375, "step": 190 }, { "epoch": 0.05, "learning_rate": 1.3342647572476424e-06, "loss": 1.3014, "step": 191 }, { "epoch": 0.05, "learning_rate": 1.34125043660496e-06, "loss": 1.2776, "step": 192 }, { "epoch": 0.05, "learning_rate": 1.3482361159622776e-06, "loss": 1.1337, "step": 193 }, { "epoch": 0.05, "learning_rate": 1.355221795319595e-06, "loss": 1.3029, "step": 194 }, { "epoch": 0.05, "learning_rate": 1.3622074746769124e-06, "loss": 1.4668, "step": 195 }, { "epoch": 0.05, "learning_rate": 1.3691931540342298e-06, "loss": 1.3451, "step": 196 }, { "epoch": 0.05, "learning_rate": 1.3761788333915474e-06, "loss": 1.3971, "step": 197 }, { "epoch": 0.05, "learning_rate": 1.383164512748865e-06, "loss": 1.3155, "step": 198 }, { "epoch": 0.05, "learning_rate": 1.3901501921061826e-06, "loss": 1.2804, "step": 199 }, { "epoch": 0.05, "learning_rate": 1.3971358714635e-06, "loss": 1.3541, "step": 200 } ], "max_steps": 95425, "num_train_epochs": 25, "total_flos": 1535405170098176.0, "trial_name": null, "trial_params": null }