{ "best_metric": null, "best_model_checkpoint": null, "epoch": 10.0, "global_step": 155200, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.03, "learning_rate": 4.98389175257732e-05, "loss": 2.2272, "step": 500 }, { "epoch": 0.06, "learning_rate": 4.9677835051546396e-05, "loss": 2.1881, "step": 1000 }, { "epoch": 0.1, "learning_rate": 4.951675257731959e-05, "loss": 2.1808, "step": 1500 }, { "epoch": 0.13, "learning_rate": 4.935567010309279e-05, "loss": 2.1891, "step": 2000 }, { "epoch": 0.16, "learning_rate": 4.919458762886598e-05, "loss": 2.2239, "step": 2500 }, { "epoch": 0.19, "learning_rate": 4.903350515463918e-05, "loss": 2.1983, "step": 3000 }, { "epoch": 0.23, "learning_rate": 4.8872422680412375e-05, "loss": 2.2063, "step": 3500 }, { "epoch": 0.26, "learning_rate": 4.871134020618557e-05, "loss": 2.2125, "step": 4000 }, { "epoch": 0.29, "learning_rate": 4.855025773195876e-05, "loss": 2.1918, "step": 4500 }, { "epoch": 0.32, "learning_rate": 4.838917525773196e-05, "loss": 2.2064, "step": 5000 }, { "epoch": 0.35, "learning_rate": 4.8228092783505156e-05, "loss": 2.1789, "step": 5500 }, { "epoch": 0.39, "learning_rate": 4.806701030927835e-05, "loss": 2.2123, "step": 6000 }, { "epoch": 0.42, "learning_rate": 4.790592783505155e-05, "loss": 2.1775, "step": 6500 }, { "epoch": 0.45, "learning_rate": 4.774484536082475e-05, "loss": 2.1837, "step": 7000 }, { "epoch": 0.48, "learning_rate": 4.758376288659794e-05, "loss": 2.1812, "step": 7500 }, { "epoch": 0.52, "learning_rate": 4.7422680412371134e-05, "loss": 2.1857, "step": 8000 }, { "epoch": 0.55, "learning_rate": 4.726159793814433e-05, "loss": 2.1788, "step": 8500 }, { "epoch": 0.58, "learning_rate": 4.710051546391753e-05, "loss": 2.2204, "step": 9000 }, { "epoch": 0.61, "learning_rate": 4.6939432989690725e-05, "loss": 2.1704, "step": 9500 }, { "epoch": 0.64, "learning_rate": 4.677835051546392e-05, "loss": 2.2155, "step": 10000 }, { "epoch": 0.68, "learning_rate": 4.661726804123712e-05, "loss": 2.2153, "step": 10500 }, { "epoch": 0.71, "learning_rate": 4.6456185567010316e-05, "loss": 2.2051, "step": 11000 }, { "epoch": 0.74, "learning_rate": 4.6295103092783506e-05, "loss": 2.2273, "step": 11500 }, { "epoch": 0.77, "learning_rate": 4.61340206185567e-05, "loss": 2.1818, "step": 12000 }, { "epoch": 0.81, "learning_rate": 4.59729381443299e-05, "loss": 2.1954, "step": 12500 }, { "epoch": 0.84, "learning_rate": 4.581185567010309e-05, "loss": 2.1635, "step": 13000 }, { "epoch": 0.87, "learning_rate": 4.565077319587629e-05, "loss": 2.1799, "step": 13500 }, { "epoch": 0.9, "learning_rate": 4.5489690721649484e-05, "loss": 2.1742, "step": 14000 }, { "epoch": 0.93, "learning_rate": 4.532860824742268e-05, "loss": 2.1402, "step": 14500 }, { "epoch": 0.97, "learning_rate": 4.516752577319588e-05, "loss": 2.1322, "step": 15000 }, { "epoch": 1.0, "learning_rate": 4.5006443298969075e-05, "loss": 2.1972, "step": 15500 }, { "epoch": 1.03, "learning_rate": 4.484536082474227e-05, "loss": 1.8657, "step": 16000 }, { "epoch": 1.06, "learning_rate": 4.468427835051547e-05, "loss": 1.8975, "step": 16500 }, { "epoch": 1.1, "learning_rate": 4.452319587628866e-05, "loss": 1.8662, "step": 17000 }, { "epoch": 1.13, "learning_rate": 4.4362113402061856e-05, "loss": 1.8394, "step": 17500 }, { "epoch": 1.16, "learning_rate": 4.4201030927835053e-05, "loss": 1.8698, "step": 18000 }, { "epoch": 1.19, "learning_rate": 4.403994845360825e-05, "loss": 1.8718, "step": 18500 }, { "epoch": 1.22, "learning_rate": 4.387886597938145e-05, "loss": 1.8674, "step": 19000 }, { "epoch": 1.26, "learning_rate": 4.3717783505154644e-05, "loss": 1.9007, "step": 19500 }, { "epoch": 1.29, "learning_rate": 4.3556701030927835e-05, "loss": 1.8961, "step": 20000 }, { "epoch": 1.32, "learning_rate": 4.339561855670103e-05, "loss": 1.8516, "step": 20500 }, { "epoch": 1.35, "learning_rate": 4.323453608247423e-05, "loss": 1.9022, "step": 21000 }, { "epoch": 1.39, "learning_rate": 4.3073453608247426e-05, "loss": 1.8913, "step": 21500 }, { "epoch": 1.42, "learning_rate": 4.2912371134020616e-05, "loss": 1.9092, "step": 22000 }, { "epoch": 1.45, "learning_rate": 4.275128865979381e-05, "loss": 1.8871, "step": 22500 }, { "epoch": 1.48, "learning_rate": 4.259020618556701e-05, "loss": 1.9027, "step": 23000 }, { "epoch": 1.51, "learning_rate": 4.242912371134021e-05, "loss": 1.9055, "step": 23500 }, { "epoch": 1.55, "learning_rate": 4.2268041237113404e-05, "loss": 1.8874, "step": 24000 }, { "epoch": 1.58, "learning_rate": 4.21069587628866e-05, "loss": 1.9049, "step": 24500 }, { "epoch": 1.61, "learning_rate": 4.19458762886598e-05, "loss": 1.9106, "step": 25000 }, { "epoch": 1.64, "learning_rate": 4.1784793814432995e-05, "loss": 1.9016, "step": 25500 }, { "epoch": 1.68, "learning_rate": 4.162371134020619e-05, "loss": 1.9368, "step": 26000 }, { "epoch": 1.71, "learning_rate": 4.146262886597939e-05, "loss": 1.9027, "step": 26500 }, { "epoch": 1.74, "learning_rate": 4.130154639175258e-05, "loss": 1.9143, "step": 27000 }, { "epoch": 1.77, "learning_rate": 4.1140463917525776e-05, "loss": 1.9305, "step": 27500 }, { "epoch": 1.8, "learning_rate": 4.097938144329897e-05, "loss": 1.9137, "step": 28000 }, { "epoch": 1.84, "learning_rate": 4.081829896907216e-05, "loss": 1.9027, "step": 28500 }, { "epoch": 1.87, "learning_rate": 4.065721649484536e-05, "loss": 1.9199, "step": 29000 }, { "epoch": 1.9, "learning_rate": 4.049613402061856e-05, "loss": 1.935, "step": 29500 }, { "epoch": 1.93, "learning_rate": 4.0335051546391754e-05, "loss": 1.9394, "step": 30000 }, { "epoch": 1.97, "learning_rate": 4.017396907216495e-05, "loss": 1.9376, "step": 30500 }, { "epoch": 2.0, "learning_rate": 4.001288659793815e-05, "loss": 1.8943, "step": 31000 }, { "epoch": 2.03, "learning_rate": 3.9851804123711345e-05, "loss": 1.6319, "step": 31500 }, { "epoch": 2.06, "learning_rate": 3.9690721649484535e-05, "loss": 1.5991, "step": 32000 }, { "epoch": 2.09, "learning_rate": 3.952963917525773e-05, "loss": 1.6505, "step": 32500 }, { "epoch": 2.13, "learning_rate": 3.936855670103093e-05, "loss": 1.6288, "step": 33000 }, { "epoch": 2.16, "learning_rate": 3.9207474226804126e-05, "loss": 1.6686, "step": 33500 }, { "epoch": 2.19, "learning_rate": 3.904639175257732e-05, "loss": 1.6538, "step": 34000 }, { "epoch": 2.22, "learning_rate": 3.888530927835052e-05, "loss": 1.6228, "step": 34500 }, { "epoch": 2.26, "learning_rate": 3.872422680412372e-05, "loss": 1.6465, "step": 35000 }, { "epoch": 2.29, "learning_rate": 3.8563144329896914e-05, "loss": 1.67, "step": 35500 }, { "epoch": 2.32, "learning_rate": 3.8402061855670104e-05, "loss": 1.6598, "step": 36000 }, { "epoch": 2.35, "learning_rate": 3.82409793814433e-05, "loss": 1.6909, "step": 36500 }, { "epoch": 2.38, "learning_rate": 3.807989690721649e-05, "loss": 1.6686, "step": 37000 }, { "epoch": 2.42, "learning_rate": 3.791881443298969e-05, "loss": 1.6587, "step": 37500 }, { "epoch": 2.45, "learning_rate": 3.7757731958762886e-05, "loss": 1.6826, "step": 38000 }, { "epoch": 2.48, "learning_rate": 3.759664948453608e-05, "loss": 1.6711, "step": 38500 }, { "epoch": 2.51, "learning_rate": 3.743556701030928e-05, "loss": 1.6633, "step": 39000 }, { "epoch": 2.55, "learning_rate": 3.7274484536082477e-05, "loss": 1.6806, "step": 39500 }, { "epoch": 2.58, "learning_rate": 3.7113402061855674e-05, "loss": 1.6888, "step": 40000 }, { "epoch": 2.61, "learning_rate": 3.695231958762887e-05, "loss": 1.6788, "step": 40500 }, { "epoch": 2.64, "learning_rate": 3.679123711340207e-05, "loss": 1.6988, "step": 41000 }, { "epoch": 2.67, "learning_rate": 3.663015463917526e-05, "loss": 1.6428, "step": 41500 }, { "epoch": 2.71, "learning_rate": 3.6469072164948455e-05, "loss": 1.6797, "step": 42000 }, { "epoch": 2.74, "learning_rate": 3.630798969072165e-05, "loss": 1.6623, "step": 42500 }, { "epoch": 2.77, "learning_rate": 3.614690721649485e-05, "loss": 1.6899, "step": 43000 }, { "epoch": 2.8, "learning_rate": 3.5985824742268046e-05, "loss": 1.706, "step": 43500 }, { "epoch": 2.84, "learning_rate": 3.5824742268041236e-05, "loss": 1.6955, "step": 44000 }, { "epoch": 2.87, "learning_rate": 3.566365979381443e-05, "loss": 1.7074, "step": 44500 }, { "epoch": 2.9, "learning_rate": 3.550257731958763e-05, "loss": 1.7156, "step": 45000 }, { "epoch": 2.93, "learning_rate": 3.534149484536083e-05, "loss": 1.6655, "step": 45500 }, { "epoch": 2.96, "learning_rate": 3.5180412371134024e-05, "loss": 1.6848, "step": 46000 }, { "epoch": 3.0, "learning_rate": 3.5019329896907214e-05, "loss": 1.6926, "step": 46500 }, { "epoch": 3.03, "learning_rate": 3.485824742268041e-05, "loss": 1.4406, "step": 47000 }, { "epoch": 3.06, "learning_rate": 3.469716494845361e-05, "loss": 1.4185, "step": 47500 }, { "epoch": 3.09, "learning_rate": 3.4536082474226805e-05, "loss": 1.4487, "step": 48000 }, { "epoch": 3.12, "learning_rate": 3.4375e-05, "loss": 1.4217, "step": 48500 }, { "epoch": 3.16, "learning_rate": 3.42139175257732e-05, "loss": 1.469, "step": 49000 }, { "epoch": 3.19, "learning_rate": 3.4052835051546396e-05, "loss": 1.4631, "step": 49500 }, { "epoch": 3.22, "learning_rate": 3.389175257731959e-05, "loss": 1.4367, "step": 50000 }, { "epoch": 3.25, "learning_rate": 3.373067010309279e-05, "loss": 1.4733, "step": 50500 }, { "epoch": 3.29, "learning_rate": 3.356958762886598e-05, "loss": 1.455, "step": 51000 }, { "epoch": 3.32, "learning_rate": 3.340850515463918e-05, "loss": 1.4458, "step": 51500 }, { "epoch": 3.35, "learning_rate": 3.3247422680412374e-05, "loss": 1.4681, "step": 52000 }, { "epoch": 3.38, "learning_rate": 3.3086340206185564e-05, "loss": 1.4796, "step": 52500 }, { "epoch": 3.41, "learning_rate": 3.292525773195876e-05, "loss": 1.4969, "step": 53000 }, { "epoch": 3.45, "learning_rate": 3.276417525773196e-05, "loss": 1.4861, "step": 53500 }, { "epoch": 3.48, "learning_rate": 3.2603092783505155e-05, "loss": 1.5032, "step": 54000 }, { "epoch": 3.51, "learning_rate": 3.244201030927835e-05, "loss": 1.463, "step": 54500 }, { "epoch": 3.54, "learning_rate": 3.228092783505155e-05, "loss": 1.4772, "step": 55000 }, { "epoch": 3.58, "learning_rate": 3.2119845360824746e-05, "loss": 1.4969, "step": 55500 }, { "epoch": 3.61, "learning_rate": 3.1958762886597937e-05, "loss": 1.5088, "step": 56000 }, { "epoch": 3.64, "learning_rate": 3.1797680412371134e-05, "loss": 1.487, "step": 56500 }, { "epoch": 3.67, "learning_rate": 3.163659793814433e-05, "loss": 1.477, "step": 57000 }, { "epoch": 3.7, "learning_rate": 3.147551546391753e-05, "loss": 1.486, "step": 57500 }, { "epoch": 3.74, "learning_rate": 3.1314432989690725e-05, "loss": 1.4838, "step": 58000 }, { "epoch": 3.77, "learning_rate": 3.115335051546392e-05, "loss": 1.4992, "step": 58500 }, { "epoch": 3.8, "learning_rate": 3.099226804123712e-05, "loss": 1.4949, "step": 59000 }, { "epoch": 3.83, "learning_rate": 3.0831185567010316e-05, "loss": 1.5133, "step": 59500 }, { "epoch": 3.87, "learning_rate": 3.0670103092783506e-05, "loss": 1.5086, "step": 60000 }, { "epoch": 3.9, "learning_rate": 3.0509020618556706e-05, "loss": 1.4887, "step": 60500 }, { "epoch": 3.93, "learning_rate": 3.0347938144329896e-05, "loss": 1.5304, "step": 61000 }, { "epoch": 3.96, "learning_rate": 3.0186855670103093e-05, "loss": 1.5248, "step": 61500 }, { "epoch": 3.99, "learning_rate": 3.002577319587629e-05, "loss": 1.4788, "step": 62000 }, { "epoch": 4.03, "learning_rate": 2.9864690721649484e-05, "loss": 1.3236, "step": 62500 }, { "epoch": 4.06, "learning_rate": 2.970360824742268e-05, "loss": 1.242, "step": 63000 }, { "epoch": 4.09, "learning_rate": 2.9542525773195878e-05, "loss": 1.2471, "step": 63500 }, { "epoch": 4.12, "learning_rate": 2.9381443298969075e-05, "loss": 1.2918, "step": 64000 }, { "epoch": 4.16, "learning_rate": 2.9220360824742272e-05, "loss": 1.2944, "step": 64500 }, { "epoch": 4.19, "learning_rate": 2.905927835051547e-05, "loss": 1.2813, "step": 65000 }, { "epoch": 4.22, "learning_rate": 2.889819587628866e-05, "loss": 1.2879, "step": 65500 }, { "epoch": 4.25, "learning_rate": 2.8737113402061856e-05, "loss": 1.28, "step": 66000 }, { "epoch": 4.28, "learning_rate": 2.857603092783505e-05, "loss": 1.2657, "step": 66500 }, { "epoch": 4.32, "learning_rate": 2.8414948453608247e-05, "loss": 1.3145, "step": 67000 }, { "epoch": 4.35, "learning_rate": 2.8253865979381444e-05, "loss": 1.3195, "step": 67500 }, { "epoch": 4.38, "learning_rate": 2.809278350515464e-05, "loss": 1.2844, "step": 68000 }, { "epoch": 4.41, "learning_rate": 2.7931701030927838e-05, "loss": 1.3339, "step": 68500 }, { "epoch": 4.45, "learning_rate": 2.7770618556701035e-05, "loss": 1.3075, "step": 69000 }, { "epoch": 4.48, "learning_rate": 2.760953608247423e-05, "loss": 1.314, "step": 69500 }, { "epoch": 4.51, "learning_rate": 2.7448453608247425e-05, "loss": 1.308, "step": 70000 }, { "epoch": 4.54, "learning_rate": 2.728737113402062e-05, "loss": 1.3234, "step": 70500 }, { "epoch": 4.57, "learning_rate": 2.7126288659793812e-05, "loss": 1.3307, "step": 71000 }, { "epoch": 4.61, "learning_rate": 2.696520618556701e-05, "loss": 1.3426, "step": 71500 }, { "epoch": 4.64, "learning_rate": 2.6804123711340206e-05, "loss": 1.3332, "step": 72000 }, { "epoch": 4.67, "learning_rate": 2.6643041237113403e-05, "loss": 1.3369, "step": 72500 }, { "epoch": 4.7, "learning_rate": 2.64819587628866e-05, "loss": 1.3446, "step": 73000 }, { "epoch": 4.74, "learning_rate": 2.6320876288659797e-05, "loss": 1.3266, "step": 73500 }, { "epoch": 4.77, "learning_rate": 2.615979381443299e-05, "loss": 1.3303, "step": 74000 }, { "epoch": 4.8, "learning_rate": 2.5998711340206188e-05, "loss": 1.332, "step": 74500 }, { "epoch": 4.83, "learning_rate": 2.5837628865979385e-05, "loss": 1.3235, "step": 75000 }, { "epoch": 4.86, "learning_rate": 2.5676546391752575e-05, "loss": 1.3405, "step": 75500 }, { "epoch": 4.9, "learning_rate": 2.5515463917525772e-05, "loss": 1.3568, "step": 76000 }, { "epoch": 4.93, "learning_rate": 2.535438144329897e-05, "loss": 1.3569, "step": 76500 }, { "epoch": 4.96, "learning_rate": 2.5193298969072166e-05, "loss": 1.3393, "step": 77000 }, { "epoch": 4.99, "learning_rate": 2.5032216494845363e-05, "loss": 1.3391, "step": 77500 }, { "epoch": 5.03, "learning_rate": 2.4871134020618557e-05, "loss": 1.1727, "step": 78000 }, { "epoch": 5.06, "learning_rate": 2.4710051546391754e-05, "loss": 1.1299, "step": 78500 }, { "epoch": 5.09, "learning_rate": 2.4548969072164947e-05, "loss": 1.1557, "step": 79000 }, { "epoch": 5.12, "learning_rate": 2.4387886597938144e-05, "loss": 1.1172, "step": 79500 }, { "epoch": 5.15, "learning_rate": 2.422680412371134e-05, "loss": 1.1455, "step": 80000 }, { "epoch": 5.19, "learning_rate": 2.406572164948454e-05, "loss": 1.1802, "step": 80500 }, { "epoch": 5.22, "learning_rate": 2.3904639175257735e-05, "loss": 1.156, "step": 81000 }, { "epoch": 5.25, "learning_rate": 2.374355670103093e-05, "loss": 1.1724, "step": 81500 }, { "epoch": 5.28, "learning_rate": 2.3582474226804126e-05, "loss": 1.1663, "step": 82000 }, { "epoch": 5.32, "learning_rate": 2.342139175257732e-05, "loss": 1.1727, "step": 82500 }, { "epoch": 5.35, "learning_rate": 2.3260309278350517e-05, "loss": 1.1767, "step": 83000 }, { "epoch": 5.38, "learning_rate": 2.3099226804123713e-05, "loss": 1.1817, "step": 83500 }, { "epoch": 5.41, "learning_rate": 2.2938144329896907e-05, "loss": 1.1878, "step": 84000 }, { "epoch": 5.44, "learning_rate": 2.2777061855670104e-05, "loss": 1.1664, "step": 84500 }, { "epoch": 5.48, "learning_rate": 2.26159793814433e-05, "loss": 1.1513, "step": 85000 }, { "epoch": 5.51, "learning_rate": 2.2454896907216498e-05, "loss": 1.1762, "step": 85500 }, { "epoch": 5.54, "learning_rate": 2.229381443298969e-05, "loss": 1.1857, "step": 86000 }, { "epoch": 5.57, "learning_rate": 2.2132731958762885e-05, "loss": 1.1986, "step": 86500 }, { "epoch": 5.61, "learning_rate": 2.1971649484536082e-05, "loss": 1.1738, "step": 87000 }, { "epoch": 5.64, "learning_rate": 2.181056701030928e-05, "loss": 1.2002, "step": 87500 }, { "epoch": 5.67, "learning_rate": 2.1649484536082476e-05, "loss": 1.1515, "step": 88000 }, { "epoch": 5.7, "learning_rate": 2.1488402061855673e-05, "loss": 1.2072, "step": 88500 }, { "epoch": 5.73, "learning_rate": 2.1327319587628867e-05, "loss": 1.1773, "step": 89000 }, { "epoch": 5.77, "learning_rate": 2.1166237113402064e-05, "loss": 1.1906, "step": 89500 }, { "epoch": 5.8, "learning_rate": 2.1005154639175257e-05, "loss": 1.2048, "step": 90000 }, { "epoch": 5.83, "learning_rate": 2.0844072164948454e-05, "loss": 1.2176, "step": 90500 }, { "epoch": 5.86, "learning_rate": 2.0682989690721648e-05, "loss": 1.1869, "step": 91000 }, { "epoch": 5.9, "learning_rate": 2.0521907216494845e-05, "loss": 1.1964, "step": 91500 }, { "epoch": 5.93, "learning_rate": 2.0360824742268042e-05, "loss": 1.176, "step": 92000 }, { "epoch": 5.96, "learning_rate": 2.019974226804124e-05, "loss": 1.2437, "step": 92500 }, { "epoch": 5.99, "learning_rate": 2.0038659793814436e-05, "loss": 1.2032, "step": 93000 }, { "epoch": 6.02, "learning_rate": 1.987757731958763e-05, "loss": 1.0818, "step": 93500 }, { "epoch": 6.06, "learning_rate": 1.9716494845360827e-05, "loss": 1.0236, "step": 94000 }, { "epoch": 6.09, "learning_rate": 1.955541237113402e-05, "loss": 1.0517, "step": 94500 }, { "epoch": 6.12, "learning_rate": 1.9394329896907217e-05, "loss": 1.0446, "step": 95000 }, { "epoch": 6.15, "learning_rate": 1.9233247422680414e-05, "loss": 1.0508, "step": 95500 }, { "epoch": 6.19, "learning_rate": 1.9072164948453608e-05, "loss": 1.0296, "step": 96000 }, { "epoch": 6.22, "learning_rate": 1.8911082474226805e-05, "loss": 1.0353, "step": 96500 }, { "epoch": 6.25, "learning_rate": 1.8750000000000002e-05, "loss": 1.0498, "step": 97000 }, { "epoch": 6.28, "learning_rate": 1.85889175257732e-05, "loss": 1.0701, "step": 97500 }, { "epoch": 6.31, "learning_rate": 1.8427835051546392e-05, "loss": 1.0654, "step": 98000 }, { "epoch": 6.35, "learning_rate": 1.8266752577319586e-05, "loss": 1.0504, "step": 98500 }, { "epoch": 6.38, "learning_rate": 1.8105670103092783e-05, "loss": 1.0527, "step": 99000 }, { "epoch": 6.41, "learning_rate": 1.794458762886598e-05, "loss": 1.0575, "step": 99500 }, { "epoch": 6.44, "learning_rate": 1.7783505154639177e-05, "loss": 1.0683, "step": 100000 }, { "epoch": 6.48, "learning_rate": 1.7622422680412374e-05, "loss": 1.0589, "step": 100500 }, { "epoch": 6.51, "learning_rate": 1.7461340206185568e-05, "loss": 1.0853, "step": 101000 }, { "epoch": 6.54, "learning_rate": 1.7300257731958764e-05, "loss": 1.0632, "step": 101500 }, { "epoch": 6.57, "learning_rate": 1.7139175257731958e-05, "loss": 1.0379, "step": 102000 }, { "epoch": 6.6, "learning_rate": 1.6978092783505155e-05, "loss": 1.052, "step": 102500 }, { "epoch": 6.64, "learning_rate": 1.6817010309278352e-05, "loss": 1.0583, "step": 103000 }, { "epoch": 6.67, "learning_rate": 1.6655927835051546e-05, "loss": 1.1123, "step": 103500 }, { "epoch": 6.7, "learning_rate": 1.6494845360824743e-05, "loss": 1.0807, "step": 104000 }, { "epoch": 6.73, "learning_rate": 1.633376288659794e-05, "loss": 1.0447, "step": 104500 }, { "epoch": 6.77, "learning_rate": 1.6172680412371137e-05, "loss": 1.0392, "step": 105000 }, { "epoch": 6.8, "learning_rate": 1.601159793814433e-05, "loss": 1.0944, "step": 105500 }, { "epoch": 6.83, "learning_rate": 1.5850515463917527e-05, "loss": 1.0617, "step": 106000 }, { "epoch": 6.86, "learning_rate": 1.568943298969072e-05, "loss": 1.0552, "step": 106500 }, { "epoch": 6.89, "learning_rate": 1.5528350515463918e-05, "loss": 1.1154, "step": 107000 }, { "epoch": 6.93, "learning_rate": 1.5367268041237115e-05, "loss": 1.0879, "step": 107500 }, { "epoch": 6.96, "learning_rate": 1.5206185567010308e-05, "loss": 1.0709, "step": 108000 }, { "epoch": 6.99, "learning_rate": 1.5045103092783505e-05, "loss": 1.0835, "step": 108500 }, { "epoch": 7.02, "learning_rate": 1.4884020618556702e-05, "loss": 0.974, "step": 109000 }, { "epoch": 7.06, "learning_rate": 1.4722938144329898e-05, "loss": 0.9417, "step": 109500 }, { "epoch": 7.09, "learning_rate": 1.4561855670103095e-05, "loss": 0.9393, "step": 110000 }, { "epoch": 7.12, "learning_rate": 1.4400773195876288e-05, "loss": 0.9574, "step": 110500 }, { "epoch": 7.15, "learning_rate": 1.4239690721649485e-05, "loss": 0.9629, "step": 111000 }, { "epoch": 7.18, "learning_rate": 1.407860824742268e-05, "loss": 0.97, "step": 111500 }, { "epoch": 7.22, "learning_rate": 1.3917525773195878e-05, "loss": 0.9266, "step": 112000 }, { "epoch": 7.25, "learning_rate": 1.3756443298969075e-05, "loss": 0.9694, "step": 112500 }, { "epoch": 7.28, "learning_rate": 1.3595360824742268e-05, "loss": 0.9429, "step": 113000 }, { "epoch": 7.31, "learning_rate": 1.3434278350515463e-05, "loss": 0.9373, "step": 113500 }, { "epoch": 7.35, "learning_rate": 1.327319587628866e-05, "loss": 0.9613, "step": 114000 }, { "epoch": 7.38, "learning_rate": 1.3112113402061857e-05, "loss": 0.9507, "step": 114500 }, { "epoch": 7.41, "learning_rate": 1.2951030927835053e-05, "loss": 0.9546, "step": 115000 }, { "epoch": 7.44, "learning_rate": 1.2789948453608246e-05, "loss": 0.9737, "step": 115500 }, { "epoch": 7.47, "learning_rate": 1.2628865979381443e-05, "loss": 0.9765, "step": 116000 }, { "epoch": 7.51, "learning_rate": 1.246778350515464e-05, "loss": 0.9772, "step": 116500 }, { "epoch": 7.54, "learning_rate": 1.2306701030927836e-05, "loss": 0.9745, "step": 117000 }, { "epoch": 7.57, "learning_rate": 1.2145618556701031e-05, "loss": 0.9693, "step": 117500 }, { "epoch": 7.6, "learning_rate": 1.1984536082474228e-05, "loss": 0.9641, "step": 118000 }, { "epoch": 7.64, "learning_rate": 1.1823453608247423e-05, "loss": 0.9696, "step": 118500 }, { "epoch": 7.67, "learning_rate": 1.1662371134020619e-05, "loss": 0.9849, "step": 119000 }, { "epoch": 7.7, "learning_rate": 1.1501288659793814e-05, "loss": 1.0011, "step": 119500 }, { "epoch": 7.73, "learning_rate": 1.134020618556701e-05, "loss": 0.9824, "step": 120000 }, { "epoch": 7.76, "learning_rate": 1.1179123711340208e-05, "loss": 0.9685, "step": 120500 }, { "epoch": 7.8, "learning_rate": 1.1018041237113403e-05, "loss": 0.9807, "step": 121000 }, { "epoch": 7.83, "learning_rate": 1.0856958762886598e-05, "loss": 0.9776, "step": 121500 }, { "epoch": 7.86, "learning_rate": 1.0695876288659794e-05, "loss": 0.9665, "step": 122000 }, { "epoch": 7.89, "learning_rate": 1.053479381443299e-05, "loss": 0.9841, "step": 122500 }, { "epoch": 7.93, "learning_rate": 1.0373711340206186e-05, "loss": 0.978, "step": 123000 }, { "epoch": 7.96, "learning_rate": 1.0212628865979381e-05, "loss": 0.9936, "step": 123500 }, { "epoch": 7.99, "learning_rate": 1.0051546391752578e-05, "loss": 0.979, "step": 124000 }, { "epoch": 8.02, "learning_rate": 9.890463917525774e-06, "loss": 0.9301, "step": 124500 }, { "epoch": 8.05, "learning_rate": 9.729381443298969e-06, "loss": 0.8839, "step": 125000 }, { "epoch": 8.09, "learning_rate": 9.568298969072164e-06, "loss": 0.8927, "step": 125500 }, { "epoch": 8.12, "learning_rate": 9.407216494845361e-06, "loss": 0.8764, "step": 126000 }, { "epoch": 8.15, "learning_rate": 9.246134020618558e-06, "loss": 0.8763, "step": 126500 }, { "epoch": 8.18, "learning_rate": 9.085051546391753e-06, "loss": 0.8848, "step": 127000 }, { "epoch": 8.22, "learning_rate": 8.923969072164949e-06, "loss": 0.8908, "step": 127500 }, { "epoch": 8.25, "learning_rate": 8.762886597938144e-06, "loss": 0.9014, "step": 128000 }, { "epoch": 8.28, "learning_rate": 8.601804123711341e-06, "loss": 0.882, "step": 128500 }, { "epoch": 8.31, "learning_rate": 8.440721649484536e-06, "loss": 0.9012, "step": 129000 }, { "epoch": 8.34, "learning_rate": 8.279639175257732e-06, "loss": 0.859, "step": 129500 }, { "epoch": 8.38, "learning_rate": 8.118556701030929e-06, "loss": 0.8917, "step": 130000 }, { "epoch": 8.41, "learning_rate": 7.957474226804124e-06, "loss": 0.9153, "step": 130500 }, { "epoch": 8.44, "learning_rate": 7.79639175257732e-06, "loss": 0.8929, "step": 131000 }, { "epoch": 8.47, "learning_rate": 7.635309278350516e-06, "loss": 0.8982, "step": 131500 }, { "epoch": 8.51, "learning_rate": 7.4742268041237115e-06, "loss": 0.8929, "step": 132000 }, { "epoch": 8.54, "learning_rate": 7.313144329896908e-06, "loss": 0.9249, "step": 132500 }, { "epoch": 8.57, "learning_rate": 7.152061855670103e-06, "loss": 0.8743, "step": 133000 }, { "epoch": 8.6, "learning_rate": 6.990979381443299e-06, "loss": 0.8956, "step": 133500 }, { "epoch": 8.63, "learning_rate": 6.829896907216494e-06, "loss": 0.8994, "step": 134000 }, { "epoch": 8.67, "learning_rate": 6.668814432989691e-06, "loss": 0.8952, "step": 134500 }, { "epoch": 8.7, "learning_rate": 6.5077319587628875e-06, "loss": 0.9033, "step": 135000 }, { "epoch": 8.73, "learning_rate": 6.346649484536083e-06, "loss": 0.9032, "step": 135500 }, { "epoch": 8.76, "learning_rate": 6.185567010309279e-06, "loss": 0.8964, "step": 136000 }, { "epoch": 8.8, "learning_rate": 6.024484536082474e-06, "loss": 0.8843, "step": 136500 }, { "epoch": 8.83, "learning_rate": 5.86340206185567e-06, "loss": 0.9027, "step": 137000 }, { "epoch": 8.86, "learning_rate": 5.7023195876288665e-06, "loss": 0.9043, "step": 137500 }, { "epoch": 8.89, "learning_rate": 5.541237113402062e-06, "loss": 0.8974, "step": 138000 }, { "epoch": 8.92, "learning_rate": 5.380154639175258e-06, "loss": 0.8937, "step": 138500 }, { "epoch": 8.96, "learning_rate": 5.219072164948454e-06, "loss": 0.8971, "step": 139000 }, { "epoch": 8.99, "learning_rate": 5.057989690721649e-06, "loss": 0.9233, "step": 139500 }, { "epoch": 9.02, "learning_rate": 4.8969072164948455e-06, "loss": 0.852, "step": 140000 }, { "epoch": 9.05, "learning_rate": 4.735824742268042e-06, "loss": 0.8142, "step": 140500 }, { "epoch": 9.09, "learning_rate": 4.574742268041237e-06, "loss": 0.8369, "step": 141000 }, { "epoch": 9.12, "learning_rate": 4.413659793814433e-06, "loss": 0.8492, "step": 141500 }, { "epoch": 9.15, "learning_rate": 4.252577319587629e-06, "loss": 0.8308, "step": 142000 }, { "epoch": 9.18, "learning_rate": 4.0914948453608246e-06, "loss": 0.8706, "step": 142500 }, { "epoch": 9.21, "learning_rate": 3.930412371134021e-06, "loss": 0.8536, "step": 143000 }, { "epoch": 9.25, "learning_rate": 3.7693298969072164e-06, "loss": 0.8231, "step": 143500 }, { "epoch": 9.28, "learning_rate": 3.608247422680412e-06, "loss": 0.8492, "step": 144000 }, { "epoch": 9.31, "learning_rate": 3.4471649484536087e-06, "loss": 0.8233, "step": 144500 }, { "epoch": 9.34, "learning_rate": 3.2860824742268044e-06, "loss": 0.8217, "step": 145000 }, { "epoch": 9.38, "learning_rate": 3.125e-06, "loss": 0.8694, "step": 145500 }, { "epoch": 9.41, "learning_rate": 2.963917525773196e-06, "loss": 0.866, "step": 146000 }, { "epoch": 9.44, "learning_rate": 2.802835051546392e-06, "loss": 0.8402, "step": 146500 }, { "epoch": 9.47, "learning_rate": 2.6417525773195877e-06, "loss": 0.84, "step": 147000 }, { "epoch": 9.5, "learning_rate": 2.4806701030927835e-06, "loss": 0.8501, "step": 147500 }, { "epoch": 9.54, "learning_rate": 2.3195876288659796e-06, "loss": 0.8374, "step": 148000 }, { "epoch": 9.57, "learning_rate": 2.1585051546391753e-06, "loss": 0.8645, "step": 148500 }, { "epoch": 9.6, "learning_rate": 1.997422680412371e-06, "loss": 0.8513, "step": 149000 }, { "epoch": 9.63, "learning_rate": 1.8363402061855672e-06, "loss": 0.8475, "step": 149500 }, { "epoch": 9.66, "learning_rate": 1.675257731958763e-06, "loss": 0.8474, "step": 150000 }, { "epoch": 9.7, "learning_rate": 1.5141752577319588e-06, "loss": 0.8414, "step": 150500 }, { "epoch": 9.73, "learning_rate": 1.3530927835051548e-06, "loss": 0.8441, "step": 151000 }, { "epoch": 9.76, "learning_rate": 1.1920103092783505e-06, "loss": 0.8762, "step": 151500 }, { "epoch": 9.79, "learning_rate": 1.0309278350515464e-06, "loss": 0.842, "step": 152000 }, { "epoch": 9.83, "learning_rate": 8.698453608247423e-07, "loss": 0.8571, "step": 152500 }, { "epoch": 9.86, "learning_rate": 7.087628865979382e-07, "loss": 0.8309, "step": 153000 }, { "epoch": 9.89, "learning_rate": 5.47680412371134e-07, "loss": 0.8334, "step": 153500 }, { "epoch": 9.92, "learning_rate": 3.8659793814432993e-07, "loss": 0.8501, "step": 154000 }, { "epoch": 9.95, "learning_rate": 2.2551546391752576e-07, "loss": 0.8423, "step": 154500 }, { "epoch": 9.99, "learning_rate": 6.443298969072165e-08, "loss": 0.8466, "step": 155000 }, { "epoch": 10.0, "step": 155200, "total_flos": 1.441343477907456e+17, "train_loss": 1.3494602605485424, "train_runtime": 31681.5151, "train_samples_per_second": 4.899, "train_steps_per_second": 4.899 } ], "max_steps": 155200, "num_train_epochs": 10, "total_flos": 1.441343477907456e+17, "trial_name": null, "trial_params": null }