{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "global_step": 59035, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.9915304480392985e-05, "loss": 3.6733, "step": 100 }, { "epoch": 0.0, "learning_rate": 4.983060896078598e-05, "loss": 3.625, "step": 200 }, { "epoch": 0.01, "learning_rate": 4.9745913441178964e-05, "loss": 3.5944, "step": 300 }, { "epoch": 0.01, "learning_rate": 4.966121792157195e-05, "loss": 3.5842, "step": 400 }, { "epoch": 0.01, "learning_rate": 4.9576522401964936e-05, "loss": 3.5584, "step": 500 }, { "epoch": 0.01, "learning_rate": 4.9491826882357925e-05, "loss": 3.5428, "step": 600 }, { "epoch": 0.01, "learning_rate": 4.9407131362750914e-05, "loss": 3.5307, "step": 700 }, { "epoch": 0.01, "learning_rate": 4.93224358431439e-05, "loss": 3.5306, "step": 800 }, { "epoch": 0.02, "learning_rate": 4.9237740323536886e-05, "loss": 3.527, "step": 900 }, { "epoch": 0.02, "learning_rate": 4.9153044803929875e-05, "loss": 3.5105, "step": 1000 }, { "epoch": 0.02, "learning_rate": 4.9068349284322865e-05, "loss": 3.5141, "step": 1100 }, { "epoch": 0.02, "learning_rate": 4.898365376471585e-05, "loss": 3.4989, "step": 1200 }, { "epoch": 0.02, "learning_rate": 4.889895824510884e-05, "loss": 3.4976, "step": 1300 }, { "epoch": 0.02, "learning_rate": 4.8814262725501826e-05, "loss": 3.5103, "step": 1400 }, { "epoch": 0.03, "learning_rate": 4.872956720589481e-05, "loss": 3.5097, "step": 1500 }, { "epoch": 0.03, "learning_rate": 4.86448716862878e-05, "loss": 3.5, "step": 1600 }, { "epoch": 0.03, "learning_rate": 4.856017616668079e-05, "loss": 3.4815, "step": 1700 }, { "epoch": 0.03, "learning_rate": 4.847548064707377e-05, "loss": 3.4799, "step": 1800 }, { "epoch": 0.03, "learning_rate": 4.839078512746676e-05, "loss": 3.4787, "step": 1900 }, { "epoch": 0.03, "learning_rate": 4.830608960785974e-05, "loss": 3.462, "step": 2000 }, { "epoch": 0.04, "learning_rate": 4.822139408825274e-05, "loss": 3.467, "step": 2100 }, { "epoch": 0.04, "learning_rate": 4.813669856864572e-05, "loss": 3.4687, "step": 2200 }, { "epoch": 0.04, "learning_rate": 4.805200304903871e-05, "loss": 3.4582, "step": 2300 }, { "epoch": 0.04, "learning_rate": 4.796730752943169e-05, "loss": 3.4537, "step": 2400 }, { "epoch": 0.04, "learning_rate": 4.788261200982468e-05, "loss": 3.4518, "step": 2500 }, { "epoch": 0.04, "learning_rate": 4.779791649021767e-05, "loss": 3.4474, "step": 2600 }, { "epoch": 0.05, "learning_rate": 4.7713220970610654e-05, "loss": 3.4428, "step": 2700 }, { "epoch": 0.05, "learning_rate": 4.762852545100364e-05, "loss": 3.4255, "step": 2800 }, { "epoch": 0.05, "learning_rate": 4.754382993139663e-05, "loss": 3.4417, "step": 2900 }, { "epoch": 0.05, "learning_rate": 4.745913441178962e-05, "loss": 3.4478, "step": 3000 }, { "epoch": 0.05, "learning_rate": 4.7374438892182604e-05, "loss": 3.4525, "step": 3100 }, { "epoch": 0.05, "learning_rate": 4.7289743372575594e-05, "loss": 3.4398, "step": 3200 }, { "epoch": 0.06, "learning_rate": 4.720504785296858e-05, "loss": 3.4446, "step": 3300 }, { "epoch": 0.06, "learning_rate": 4.7120352333361565e-05, "loss": 3.4213, "step": 3400 }, { "epoch": 0.06, "learning_rate": 4.7035656813754555e-05, "loss": 3.4228, "step": 3500 }, { "epoch": 0.06, "learning_rate": 4.6950961294147544e-05, "loss": 3.4218, "step": 3600 }, { "epoch": 0.06, "learning_rate": 4.686626577454053e-05, "loss": 3.4119, "step": 3700 }, { "epoch": 0.06, "learning_rate": 4.6781570254933516e-05, "loss": 3.4226, "step": 3800 }, { "epoch": 0.07, "learning_rate": 4.66968747353265e-05, "loss": 3.4109, "step": 3900 }, { "epoch": 0.07, "learning_rate": 4.6612179215719495e-05, "loss": 3.4229, "step": 4000 }, { "epoch": 0.07, "learning_rate": 4.652748369611248e-05, "loss": 3.4124, "step": 4100 }, { "epoch": 0.07, "learning_rate": 4.644278817650547e-05, "loss": 3.4037, "step": 4200 }, { "epoch": 0.07, "learning_rate": 4.635809265689845e-05, "loss": 3.418, "step": 4300 }, { "epoch": 0.07, "learning_rate": 4.627339713729144e-05, "loss": 3.4158, "step": 4400 }, { "epoch": 0.08, "learning_rate": 4.618870161768443e-05, "loss": 3.4151, "step": 4500 }, { "epoch": 0.08, "learning_rate": 4.610400609807741e-05, "loss": 3.3993, "step": 4600 }, { "epoch": 0.08, "learning_rate": 4.6019310578470407e-05, "loss": 3.4016, "step": 4700 }, { "epoch": 0.08, "learning_rate": 4.593461505886339e-05, "loss": 3.4012, "step": 4800 }, { "epoch": 0.08, "learning_rate": 4.584991953925638e-05, "loss": 3.4052, "step": 4900 }, { "epoch": 0.08, "learning_rate": 4.576522401964936e-05, "loss": 3.3892, "step": 5000 }, { "epoch": 0.09, "learning_rate": 4.568052850004235e-05, "loss": 3.3988, "step": 5100 }, { "epoch": 0.09, "learning_rate": 4.559583298043534e-05, "loss": 3.3904, "step": 5200 }, { "epoch": 0.09, "learning_rate": 4.551113746082832e-05, "loss": 3.3914, "step": 5300 }, { "epoch": 0.09, "learning_rate": 4.542644194122131e-05, "loss": 3.3866, "step": 5400 }, { "epoch": 0.09, "learning_rate": 4.53417464216143e-05, "loss": 3.4047, "step": 5500 }, { "epoch": 0.09, "learning_rate": 4.5257050902007284e-05, "loss": 3.3882, "step": 5600 }, { "epoch": 0.1, "learning_rate": 4.517235538240027e-05, "loss": 3.386, "step": 5700 }, { "epoch": 0.1, "learning_rate": 4.5087659862793255e-05, "loss": 3.3753, "step": 5800 }, { "epoch": 0.1, "learning_rate": 4.500296434318625e-05, "loss": 3.3788, "step": 5900 }, { "epoch": 0.1, "learning_rate": 4.4918268823579234e-05, "loss": 3.3694, "step": 6000 }, { "epoch": 0.1, "learning_rate": 4.4833573303972223e-05, "loss": 3.3764, "step": 6100 }, { "epoch": 0.11, "learning_rate": 4.474887778436521e-05, "loss": 3.3728, "step": 6200 }, { "epoch": 0.11, "learning_rate": 4.4664182264758195e-05, "loss": 3.3889, "step": 6300 }, { "epoch": 0.11, "learning_rate": 4.4579486745151185e-05, "loss": 3.3798, "step": 6400 }, { "epoch": 0.11, "learning_rate": 4.449479122554417e-05, "loss": 3.3877, "step": 6500 }, { "epoch": 0.11, "learning_rate": 4.4410095705937163e-05, "loss": 3.3784, "step": 6600 }, { "epoch": 0.11, "learning_rate": 4.4325400186330146e-05, "loss": 3.3791, "step": 6700 }, { "epoch": 0.12, "learning_rate": 4.424070466672313e-05, "loss": 3.3634, "step": 6800 }, { "epoch": 0.12, "learning_rate": 4.415600914711612e-05, "loss": 3.3735, "step": 6900 }, { "epoch": 0.12, "learning_rate": 4.407131362750911e-05, "loss": 3.3628, "step": 7000 }, { "epoch": 0.12, "learning_rate": 4.3986618107902097e-05, "loss": 3.3631, "step": 7100 }, { "epoch": 0.12, "learning_rate": 4.390192258829508e-05, "loss": 3.3781, "step": 7200 }, { "epoch": 0.12, "learning_rate": 4.381722706868807e-05, "loss": 3.3661, "step": 7300 }, { "epoch": 0.13, "learning_rate": 4.373253154908106e-05, "loss": 3.357, "step": 7400 }, { "epoch": 0.13, "learning_rate": 4.364783602947404e-05, "loss": 3.3617, "step": 7500 }, { "epoch": 0.13, "learning_rate": 4.356314050986703e-05, "loss": 3.3608, "step": 7600 }, { "epoch": 0.13, "learning_rate": 4.347844499026002e-05, "loss": 3.3688, "step": 7700 }, { "epoch": 0.13, "learning_rate": 4.339374947065301e-05, "loss": 3.3568, "step": 7800 }, { "epoch": 0.13, "learning_rate": 4.330905395104599e-05, "loss": 3.3558, "step": 7900 }, { "epoch": 0.14, "learning_rate": 4.322435843143898e-05, "loss": 3.3519, "step": 8000 }, { "epoch": 0.14, "learning_rate": 4.313966291183197e-05, "loss": 3.3446, "step": 8100 }, { "epoch": 0.14, "learning_rate": 4.305496739222495e-05, "loss": 3.3601, "step": 8200 }, { "epoch": 0.14, "learning_rate": 4.297027187261794e-05, "loss": 3.3536, "step": 8300 }, { "epoch": 0.14, "learning_rate": 4.2885576353010924e-05, "loss": 3.3516, "step": 8400 }, { "epoch": 0.14, "learning_rate": 4.280088083340392e-05, "loss": 3.3506, "step": 8500 }, { "epoch": 0.15, "learning_rate": 4.27161853137969e-05, "loss": 3.3596, "step": 8600 }, { "epoch": 0.15, "learning_rate": 4.2631489794189885e-05, "loss": 3.3403, "step": 8700 }, { "epoch": 0.15, "learning_rate": 4.2546794274582875e-05, "loss": 3.3462, "step": 8800 }, { "epoch": 0.15, "learning_rate": 4.2462098754975864e-05, "loss": 3.3527, "step": 8900 }, { "epoch": 0.15, "learning_rate": 4.237740323536885e-05, "loss": 3.3366, "step": 9000 }, { "epoch": 0.15, "learning_rate": 4.2292707715761836e-05, "loss": 3.3477, "step": 9100 }, { "epoch": 0.16, "learning_rate": 4.2208012196154825e-05, "loss": 3.3442, "step": 9200 }, { "epoch": 0.16, "learning_rate": 4.2123316676547815e-05, "loss": 3.3409, "step": 9300 }, { "epoch": 0.16, "learning_rate": 4.20386211569408e-05, "loss": 3.3388, "step": 9400 }, { "epoch": 0.16, "learning_rate": 4.1953925637333787e-05, "loss": 3.3477, "step": 9500 }, { "epoch": 0.16, "learning_rate": 4.1869230117726776e-05, "loss": 3.3408, "step": 9600 }, { "epoch": 0.16, "learning_rate": 4.1784534598119765e-05, "loss": 3.3139, "step": 9700 }, { "epoch": 0.17, "learning_rate": 4.169983907851275e-05, "loss": 3.3397, "step": 9800 }, { "epoch": 0.17, "learning_rate": 4.161514355890573e-05, "loss": 3.3409, "step": 9900 }, { "epoch": 0.17, "learning_rate": 4.1530448039298726e-05, "loss": 3.3583, "step": 10000 }, { "epoch": 0.17, "learning_rate": 4.144575251969171e-05, "loss": 3.327, "step": 10100 }, { "epoch": 0.17, "learning_rate": 4.13610570000847e-05, "loss": 3.3296, "step": 10200 }, { "epoch": 0.17, "learning_rate": 4.127636148047768e-05, "loss": 3.3406, "step": 10300 }, { "epoch": 0.18, "learning_rate": 4.119166596087067e-05, "loss": 3.3341, "step": 10400 }, { "epoch": 0.18, "learning_rate": 4.110697044126366e-05, "loss": 3.3396, "step": 10500 }, { "epoch": 0.18, "learning_rate": 4.102227492165664e-05, "loss": 3.333, "step": 10600 }, { "epoch": 0.18, "learning_rate": 4.093757940204964e-05, "loss": 3.3402, "step": 10700 }, { "epoch": 0.18, "learning_rate": 4.085288388244262e-05, "loss": 3.3197, "step": 10800 }, { "epoch": 0.18, "learning_rate": 4.076818836283561e-05, "loss": 3.3265, "step": 10900 }, { "epoch": 0.19, "learning_rate": 4.068349284322859e-05, "loss": 3.322, "step": 11000 }, { "epoch": 0.19, "learning_rate": 4.059879732362158e-05, "loss": 3.3263, "step": 11100 }, { "epoch": 0.19, "learning_rate": 4.051410180401457e-05, "loss": 3.3376, "step": 11200 }, { "epoch": 0.19, "learning_rate": 4.0429406284407554e-05, "loss": 3.3221, "step": 11300 }, { "epoch": 0.19, "learning_rate": 4.034471076480054e-05, "loss": 3.3336, "step": 11400 }, { "epoch": 0.19, "learning_rate": 4.026001524519353e-05, "loss": 3.3214, "step": 11500 }, { "epoch": 0.2, "learning_rate": 4.017531972558652e-05, "loss": 3.3162, "step": 11600 }, { "epoch": 0.2, "learning_rate": 4.0090624205979505e-05, "loss": 3.3146, "step": 11700 }, { "epoch": 0.2, "learning_rate": 4.000592868637249e-05, "loss": 3.313, "step": 11800 }, { "epoch": 0.2, "learning_rate": 3.992123316676548e-05, "loss": 3.3124, "step": 11900 }, { "epoch": 0.2, "learning_rate": 3.9836537647158466e-05, "loss": 3.3081, "step": 12000 }, { "epoch": 0.2, "learning_rate": 3.9751842127551455e-05, "loss": 3.3001, "step": 12100 }, { "epoch": 0.21, "learning_rate": 3.9667146607944445e-05, "loss": 3.3163, "step": 12200 }, { "epoch": 0.21, "learning_rate": 3.958245108833743e-05, "loss": 3.3065, "step": 12300 }, { "epoch": 0.21, "learning_rate": 3.9497755568730416e-05, "loss": 3.3106, "step": 12400 }, { "epoch": 0.21, "learning_rate": 3.94130600491234e-05, "loss": 3.3182, "step": 12500 }, { "epoch": 0.21, "learning_rate": 3.9328364529516395e-05, "loss": 3.3078, "step": 12600 }, { "epoch": 0.22, "learning_rate": 3.924366900990938e-05, "loss": 3.3173, "step": 12700 }, { "epoch": 0.22, "learning_rate": 3.915897349030237e-05, "loss": 3.3037, "step": 12800 }, { "epoch": 0.22, "learning_rate": 3.907427797069535e-05, "loss": 3.3128, "step": 12900 }, { "epoch": 0.22, "learning_rate": 3.898958245108834e-05, "loss": 3.3012, "step": 13000 }, { "epoch": 0.22, "learning_rate": 3.890488693148133e-05, "loss": 3.3003, "step": 13100 }, { "epoch": 0.22, "learning_rate": 3.882019141187431e-05, "loss": 3.3, "step": 13200 }, { "epoch": 0.23, "learning_rate": 3.87354958922673e-05, "loss": 3.3083, "step": 13300 }, { "epoch": 0.23, "learning_rate": 3.865080037266029e-05, "loss": 3.3119, "step": 13400 }, { "epoch": 0.23, "learning_rate": 3.856610485305327e-05, "loss": 3.3086, "step": 13500 }, { "epoch": 0.23, "learning_rate": 3.848140933344626e-05, "loss": 3.315, "step": 13600 }, { "epoch": 0.23, "learning_rate": 3.839671381383925e-05, "loss": 3.3085, "step": 13700 }, { "epoch": 0.23, "learning_rate": 3.831201829423224e-05, "loss": 3.302, "step": 13800 }, { "epoch": 0.24, "learning_rate": 3.822732277462522e-05, "loss": 3.3055, "step": 13900 }, { "epoch": 0.24, "learning_rate": 3.814262725501821e-05, "loss": 3.2976, "step": 14000 }, { "epoch": 0.24, "learning_rate": 3.80579317354112e-05, "loss": 3.3026, "step": 14100 }, { "epoch": 0.24, "learning_rate": 3.7973236215804184e-05, "loss": 3.2877, "step": 14200 }, { "epoch": 0.24, "learning_rate": 3.788854069619717e-05, "loss": 3.2788, "step": 14300 }, { "epoch": 0.24, "learning_rate": 3.7803845176590156e-05, "loss": 3.302, "step": 14400 }, { "epoch": 0.25, "learning_rate": 3.771914965698315e-05, "loss": 3.2954, "step": 14500 }, { "epoch": 0.25, "learning_rate": 3.7634454137376135e-05, "loss": 3.2901, "step": 14600 }, { "epoch": 0.25, "learning_rate": 3.7549758617769124e-05, "loss": 3.2979, "step": 14700 }, { "epoch": 0.25, "learning_rate": 3.7465063098162106e-05, "loss": 3.2923, "step": 14800 }, { "epoch": 0.25, "learning_rate": 3.7380367578555096e-05, "loss": 3.2981, "step": 14900 }, { "epoch": 0.25, "learning_rate": 3.7295672058948085e-05, "loss": 3.2964, "step": 15000 }, { "epoch": 0.26, "learning_rate": 3.721097653934107e-05, "loss": 3.2995, "step": 15100 }, { "epoch": 0.26, "learning_rate": 3.7126281019734064e-05, "loss": 3.2988, "step": 15200 }, { "epoch": 0.26, "learning_rate": 3.7041585500127046e-05, "loss": 3.2948, "step": 15300 }, { "epoch": 0.26, "learning_rate": 3.695688998052003e-05, "loss": 3.288, "step": 15400 }, { "epoch": 0.26, "learning_rate": 3.687219446091302e-05, "loss": 3.2881, "step": 15500 }, { "epoch": 0.26, "learning_rate": 3.678749894130601e-05, "loss": 3.2768, "step": 15600 }, { "epoch": 0.27, "learning_rate": 3.6702803421699e-05, "loss": 3.2781, "step": 15700 }, { "epoch": 0.27, "learning_rate": 3.661810790209198e-05, "loss": 3.2889, "step": 15800 }, { "epoch": 0.27, "learning_rate": 3.653341238248497e-05, "loss": 3.2875, "step": 15900 }, { "epoch": 0.27, "learning_rate": 3.644871686287796e-05, "loss": 3.2949, "step": 16000 }, { "epoch": 0.27, "learning_rate": 3.636402134327094e-05, "loss": 3.281, "step": 16100 }, { "epoch": 0.27, "learning_rate": 3.627932582366393e-05, "loss": 3.2902, "step": 16200 }, { "epoch": 0.28, "learning_rate": 3.619463030405691e-05, "loss": 3.2926, "step": 16300 }, { "epoch": 0.28, "learning_rate": 3.610993478444991e-05, "loss": 3.2912, "step": 16400 }, { "epoch": 0.28, "learning_rate": 3.602523926484289e-05, "loss": 3.2835, "step": 16500 }, { "epoch": 0.28, "learning_rate": 3.594054374523588e-05, "loss": 3.2712, "step": 16600 }, { "epoch": 0.28, "learning_rate": 3.585584822562887e-05, "loss": 3.2824, "step": 16700 }, { "epoch": 0.28, "learning_rate": 3.577115270602185e-05, "loss": 3.2741, "step": 16800 }, { "epoch": 0.29, "learning_rate": 3.568645718641484e-05, "loss": 3.2763, "step": 16900 }, { "epoch": 0.29, "learning_rate": 3.5601761666807825e-05, "loss": 3.2611, "step": 17000 }, { "epoch": 0.29, "learning_rate": 3.5517066147200814e-05, "loss": 3.2923, "step": 17100 }, { "epoch": 0.29, "learning_rate": 3.54323706275938e-05, "loss": 3.2667, "step": 17200 }, { "epoch": 0.29, "learning_rate": 3.5347675107986786e-05, "loss": 3.2792, "step": 17300 }, { "epoch": 0.29, "learning_rate": 3.5262979588379775e-05, "loss": 3.2777, "step": 17400 }, { "epoch": 0.3, "learning_rate": 3.5178284068772764e-05, "loss": 3.275, "step": 17500 }, { "epoch": 0.3, "learning_rate": 3.5093588549165754e-05, "loss": 3.286, "step": 17600 }, { "epoch": 0.3, "learning_rate": 3.5008893029558736e-05, "loss": 3.2784, "step": 17700 }, { "epoch": 0.3, "learning_rate": 3.4924197509951726e-05, "loss": 3.2707, "step": 17800 }, { "epoch": 0.3, "learning_rate": 3.4839501990344715e-05, "loss": 3.2591, "step": 17900 }, { "epoch": 0.3, "learning_rate": 3.47548064707377e-05, "loss": 3.268, "step": 18000 }, { "epoch": 0.31, "learning_rate": 3.467011095113069e-05, "loss": 3.2802, "step": 18100 }, { "epoch": 0.31, "learning_rate": 3.4585415431523676e-05, "loss": 3.2788, "step": 18200 }, { "epoch": 0.31, "learning_rate": 3.4500719911916666e-05, "loss": 3.2516, "step": 18300 }, { "epoch": 0.31, "learning_rate": 3.441602439230965e-05, "loss": 3.2708, "step": 18400 }, { "epoch": 0.31, "learning_rate": 3.433132887270263e-05, "loss": 3.269, "step": 18500 }, { "epoch": 0.32, "learning_rate": 3.424663335309563e-05, "loss": 3.2583, "step": 18600 }, { "epoch": 0.32, "learning_rate": 3.416193783348861e-05, "loss": 3.2607, "step": 18700 }, { "epoch": 0.32, "learning_rate": 3.40772423138816e-05, "loss": 3.2738, "step": 18800 }, { "epoch": 0.32, "learning_rate": 3.399254679427458e-05, "loss": 3.2747, "step": 18900 }, { "epoch": 0.32, "learning_rate": 3.390785127466757e-05, "loss": 3.2722, "step": 19000 }, { "epoch": 0.32, "learning_rate": 3.382315575506056e-05, "loss": 3.28, "step": 19100 }, { "epoch": 0.33, "learning_rate": 3.373846023545354e-05, "loss": 3.2739, "step": 19200 }, { "epoch": 0.33, "learning_rate": 3.365376471584653e-05, "loss": 3.2595, "step": 19300 }, { "epoch": 0.33, "learning_rate": 3.356906919623952e-05, "loss": 3.2672, "step": 19400 }, { "epoch": 0.33, "learning_rate": 3.348437367663251e-05, "loss": 3.2611, "step": 19500 }, { "epoch": 0.33, "learning_rate": 3.339967815702549e-05, "loss": 3.2656, "step": 19600 }, { "epoch": 0.33, "learning_rate": 3.331498263741848e-05, "loss": 3.2765, "step": 19700 }, { "epoch": 0.34, "learning_rate": 3.323028711781147e-05, "loss": 3.2694, "step": 19800 }, { "epoch": 0.34, "learning_rate": 3.3145591598204454e-05, "loss": 3.2673, "step": 19900 }, { "epoch": 0.34, "learning_rate": 3.3060896078597444e-05, "loss": 3.2698, "step": 20000 }, { "epoch": 0.34, "learning_rate": 3.297620055899043e-05, "loss": 3.2659, "step": 20100 }, { "epoch": 0.34, "learning_rate": 3.2891505039383416e-05, "loss": 3.2683, "step": 20200 }, { "epoch": 0.34, "learning_rate": 3.2806809519776405e-05, "loss": 3.2527, "step": 20300 }, { "epoch": 0.35, "learning_rate": 3.272211400016939e-05, "loss": 3.2603, "step": 20400 }, { "epoch": 0.35, "learning_rate": 3.2637418480562384e-05, "loss": 3.2634, "step": 20500 }, { "epoch": 0.35, "learning_rate": 3.2552722960955366e-05, "loss": 3.2569, "step": 20600 }, { "epoch": 0.35, "learning_rate": 3.2468027441348356e-05, "loss": 3.2552, "step": 20700 }, { "epoch": 0.35, "learning_rate": 3.238333192174134e-05, "loss": 3.2618, "step": 20800 }, { "epoch": 0.35, "learning_rate": 3.229863640213433e-05, "loss": 3.2546, "step": 20900 }, { "epoch": 0.36, "learning_rate": 3.221394088252732e-05, "loss": 3.2551, "step": 21000 }, { "epoch": 0.36, "learning_rate": 3.21292453629203e-05, "loss": 3.2576, "step": 21100 }, { "epoch": 0.36, "learning_rate": 3.2044549843313296e-05, "loss": 3.2724, "step": 21200 }, { "epoch": 0.36, "learning_rate": 3.195985432370628e-05, "loss": 3.2538, "step": 21300 }, { "epoch": 0.36, "learning_rate": 3.187515880409927e-05, "loss": 3.2595, "step": 21400 }, { "epoch": 0.36, "learning_rate": 3.179046328449225e-05, "loss": 3.2421, "step": 21500 }, { "epoch": 0.37, "learning_rate": 3.170576776488524e-05, "loss": 3.2536, "step": 21600 }, { "epoch": 0.37, "learning_rate": 3.162107224527823e-05, "loss": 3.2726, "step": 21700 }, { "epoch": 0.37, "learning_rate": 3.153637672567121e-05, "loss": 3.2631, "step": 21800 }, { "epoch": 0.37, "learning_rate": 3.14516812060642e-05, "loss": 3.2627, "step": 21900 }, { "epoch": 0.37, "learning_rate": 3.136698568645719e-05, "loss": 3.2431, "step": 22000 }, { "epoch": 0.37, "learning_rate": 3.128229016685017e-05, "loss": 3.2649, "step": 22100 }, { "epoch": 0.38, "learning_rate": 3.119759464724316e-05, "loss": 3.2347, "step": 22200 }, { "epoch": 0.38, "learning_rate": 3.1112899127636144e-05, "loss": 3.2413, "step": 22300 }, { "epoch": 0.38, "learning_rate": 3.102820360802914e-05, "loss": 3.2626, "step": 22400 }, { "epoch": 0.38, "learning_rate": 3.094350808842212e-05, "loss": 3.2423, "step": 22500 }, { "epoch": 0.38, "learning_rate": 3.085881256881511e-05, "loss": 3.2457, "step": 22600 }, { "epoch": 0.38, "learning_rate": 3.07741170492081e-05, "loss": 3.239, "step": 22700 }, { "epoch": 0.39, "learning_rate": 3.0689421529601084e-05, "loss": 3.2517, "step": 22800 }, { "epoch": 0.39, "learning_rate": 3.0604726009994074e-05, "loss": 3.251, "step": 22900 }, { "epoch": 0.39, "learning_rate": 3.0520030490387056e-05, "loss": 3.2454, "step": 23000 }, { "epoch": 0.39, "learning_rate": 3.043533497078005e-05, "loss": 3.2455, "step": 23100 }, { "epoch": 0.39, "learning_rate": 3.0350639451173035e-05, "loss": 3.2572, "step": 23200 }, { "epoch": 0.39, "learning_rate": 3.026594393156602e-05, "loss": 3.2525, "step": 23300 }, { "epoch": 0.4, "learning_rate": 3.0181248411959007e-05, "loss": 3.2423, "step": 23400 }, { "epoch": 0.4, "learning_rate": 3.0096552892352e-05, "loss": 3.2602, "step": 23500 }, { "epoch": 0.4, "learning_rate": 3.0011857372744982e-05, "loss": 3.2442, "step": 23600 }, { "epoch": 0.4, "learning_rate": 2.9927161853137968e-05, "loss": 3.242, "step": 23700 }, { "epoch": 0.4, "learning_rate": 2.9842466333530954e-05, "loss": 3.2454, "step": 23800 }, { "epoch": 0.4, "learning_rate": 2.9757770813923947e-05, "loss": 3.2343, "step": 23900 }, { "epoch": 0.41, "learning_rate": 2.9673075294316933e-05, "loss": 3.2436, "step": 24000 }, { "epoch": 0.41, "learning_rate": 2.958837977470992e-05, "loss": 3.237, "step": 24100 }, { "epoch": 0.41, "learning_rate": 2.9503684255102908e-05, "loss": 3.2552, "step": 24200 }, { "epoch": 0.41, "learning_rate": 2.9418988735495894e-05, "loss": 3.2511, "step": 24300 }, { "epoch": 0.41, "learning_rate": 2.933429321588888e-05, "loss": 3.2352, "step": 24400 }, { "epoch": 0.42, "learning_rate": 2.9249597696281866e-05, "loss": 3.2356, "step": 24500 }, { "epoch": 0.42, "learning_rate": 2.916490217667486e-05, "loss": 3.2547, "step": 24600 }, { "epoch": 0.42, "learning_rate": 2.9080206657067845e-05, "loss": 3.2578, "step": 24700 }, { "epoch": 0.42, "learning_rate": 2.899551113746083e-05, "loss": 3.2299, "step": 24800 }, { "epoch": 0.42, "learning_rate": 2.8910815617853816e-05, "loss": 3.2308, "step": 24900 }, { "epoch": 0.42, "learning_rate": 2.8826120098246806e-05, "loss": 3.231, "step": 25000 }, { "epoch": 0.43, "learning_rate": 2.8741424578639792e-05, "loss": 3.244, "step": 25100 }, { "epoch": 0.43, "learning_rate": 2.8656729059032778e-05, "loss": 3.2523, "step": 25200 }, { "epoch": 0.43, "learning_rate": 2.8572033539425764e-05, "loss": 3.2394, "step": 25300 }, { "epoch": 0.43, "learning_rate": 2.8487338019818753e-05, "loss": 3.2434, "step": 25400 }, { "epoch": 0.43, "learning_rate": 2.840264250021174e-05, "loss": 3.2412, "step": 25500 }, { "epoch": 0.43, "learning_rate": 2.8317946980604725e-05, "loss": 3.2396, "step": 25600 }, { "epoch": 0.44, "learning_rate": 2.8233251460997718e-05, "loss": 3.2516, "step": 25700 }, { "epoch": 0.44, "learning_rate": 2.8148555941390704e-05, "loss": 3.221, "step": 25800 }, { "epoch": 0.44, "learning_rate": 2.806386042178369e-05, "loss": 3.2295, "step": 25900 }, { "epoch": 0.44, "learning_rate": 2.7979164902176675e-05, "loss": 3.236, "step": 26000 }, { "epoch": 0.44, "learning_rate": 2.7894469382569665e-05, "loss": 3.2444, "step": 26100 }, { "epoch": 0.44, "learning_rate": 2.780977386296265e-05, "loss": 3.2383, "step": 26200 }, { "epoch": 0.45, "learning_rate": 2.7725078343355637e-05, "loss": 3.2314, "step": 26300 }, { "epoch": 0.45, "learning_rate": 2.7640382823748623e-05, "loss": 3.2459, "step": 26400 }, { "epoch": 0.45, "learning_rate": 2.7555687304141615e-05, "loss": 3.2391, "step": 26500 }, { "epoch": 0.45, "learning_rate": 2.74709917845346e-05, "loss": 3.2394, "step": 26600 }, { "epoch": 0.45, "learning_rate": 2.7386296264927584e-05, "loss": 3.2304, "step": 26700 }, { "epoch": 0.45, "learning_rate": 2.730160074532057e-05, "loss": 3.2376, "step": 26800 }, { "epoch": 0.46, "learning_rate": 2.7216905225713563e-05, "loss": 3.2229, "step": 26900 }, { "epoch": 0.46, "learning_rate": 2.713220970610655e-05, "loss": 3.2386, "step": 27000 }, { "epoch": 0.46, "learning_rate": 2.7047514186499535e-05, "loss": 3.2269, "step": 27100 }, { "epoch": 0.46, "learning_rate": 2.6962818666892524e-05, "loss": 3.2357, "step": 27200 }, { "epoch": 0.46, "learning_rate": 2.687812314728551e-05, "loss": 3.2211, "step": 27300 }, { "epoch": 0.46, "learning_rate": 2.6793427627678496e-05, "loss": 3.2306, "step": 27400 }, { "epoch": 0.47, "learning_rate": 2.6708732108071482e-05, "loss": 3.246, "step": 27500 }, { "epoch": 0.47, "learning_rate": 2.6624036588464474e-05, "loss": 3.2277, "step": 27600 }, { "epoch": 0.47, "learning_rate": 2.653934106885746e-05, "loss": 3.2324, "step": 27700 }, { "epoch": 0.47, "learning_rate": 2.6454645549250446e-05, "loss": 3.2292, "step": 27800 }, { "epoch": 0.47, "learning_rate": 2.6369950029643432e-05, "loss": 3.2308, "step": 27900 }, { "epoch": 0.47, "learning_rate": 2.628525451003642e-05, "loss": 3.2243, "step": 28000 }, { "epoch": 0.48, "learning_rate": 2.6200558990429408e-05, "loss": 3.2343, "step": 28100 }, { "epoch": 0.48, "learning_rate": 2.6115863470822394e-05, "loss": 3.237, "step": 28200 }, { "epoch": 0.48, "learning_rate": 2.603116795121538e-05, "loss": 3.2222, "step": 28300 }, { "epoch": 0.48, "learning_rate": 2.5946472431608372e-05, "loss": 3.234, "step": 28400 }, { "epoch": 0.48, "learning_rate": 2.5861776912001355e-05, "loss": 3.2234, "step": 28500 }, { "epoch": 0.48, "learning_rate": 2.577708139239434e-05, "loss": 3.2333, "step": 28600 }, { "epoch": 0.49, "learning_rate": 2.5692385872787333e-05, "loss": 3.2221, "step": 28700 }, { "epoch": 0.49, "learning_rate": 2.560769035318032e-05, "loss": 3.2205, "step": 28800 }, { "epoch": 0.49, "learning_rate": 2.5522994833573305e-05, "loss": 3.2371, "step": 28900 }, { "epoch": 0.49, "learning_rate": 2.543829931396629e-05, "loss": 3.2169, "step": 29000 }, { "epoch": 0.49, "learning_rate": 2.535360379435928e-05, "loss": 3.2225, "step": 29100 }, { "epoch": 0.49, "learning_rate": 2.5268908274752267e-05, "loss": 3.2218, "step": 29200 }, { "epoch": 0.5, "learning_rate": 2.5184212755145253e-05, "loss": 3.2285, "step": 29300 }, { "epoch": 0.5, "learning_rate": 2.509951723553824e-05, "loss": 3.2374, "step": 29400 }, { "epoch": 0.5, "learning_rate": 2.501482171593123e-05, "loss": 3.2276, "step": 29500 }, { "epoch": 0.5, "learning_rate": 2.4930126196324217e-05, "loss": 3.2242, "step": 29600 }, { "epoch": 0.5, "learning_rate": 2.4845430676717203e-05, "loss": 3.2227, "step": 29700 }, { "epoch": 0.5, "learning_rate": 2.476073515711019e-05, "loss": 3.2058, "step": 29800 }, { "epoch": 0.51, "learning_rate": 2.4676039637503175e-05, "loss": 3.2233, "step": 29900 }, { "epoch": 0.51, "learning_rate": 2.4591344117896164e-05, "loss": 3.2197, "step": 30000 }, { "epoch": 0.51, "learning_rate": 2.4506648598289154e-05, "loss": 3.2137, "step": 30100 }, { "epoch": 0.51, "learning_rate": 2.442195307868214e-05, "loss": 3.2201, "step": 30200 }, { "epoch": 0.51, "learning_rate": 2.4337257559075126e-05, "loss": 3.2254, "step": 30300 }, { "epoch": 0.51, "learning_rate": 2.425256203946811e-05, "loss": 3.2302, "step": 30400 }, { "epoch": 0.52, "learning_rate": 2.41678665198611e-05, "loss": 3.2154, "step": 30500 }, { "epoch": 0.52, "learning_rate": 2.4083171000254087e-05, "loss": 3.2012, "step": 30600 }, { "epoch": 0.52, "learning_rate": 2.3998475480647076e-05, "loss": 3.2228, "step": 30700 }, { "epoch": 0.52, "learning_rate": 2.3913779961040062e-05, "loss": 3.2212, "step": 30800 }, { "epoch": 0.52, "learning_rate": 2.382908444143305e-05, "loss": 3.222, "step": 30900 }, { "epoch": 0.53, "learning_rate": 2.3744388921826034e-05, "loss": 3.215, "step": 31000 }, { "epoch": 0.53, "learning_rate": 2.3659693402219023e-05, "loss": 3.2235, "step": 31100 }, { "epoch": 0.53, "learning_rate": 2.357499788261201e-05, "loss": 3.2163, "step": 31200 }, { "epoch": 0.53, "learning_rate": 2.3490302363005e-05, "loss": 3.2061, "step": 31300 }, { "epoch": 0.53, "learning_rate": 2.3405606843397985e-05, "loss": 3.2124, "step": 31400 }, { "epoch": 0.53, "learning_rate": 2.3320911323790974e-05, "loss": 3.217, "step": 31500 }, { "epoch": 0.54, "learning_rate": 2.323621580418396e-05, "loss": 3.2221, "step": 31600 }, { "epoch": 0.54, "learning_rate": 2.3151520284576946e-05, "loss": 3.2334, "step": 31700 }, { "epoch": 0.54, "learning_rate": 2.3066824764969935e-05, "loss": 3.2209, "step": 31800 }, { "epoch": 0.54, "learning_rate": 2.298212924536292e-05, "loss": 3.2292, "step": 31900 }, { "epoch": 0.54, "learning_rate": 2.289743372575591e-05, "loss": 3.2192, "step": 32000 }, { "epoch": 0.54, "learning_rate": 2.2812738206148897e-05, "loss": 3.2103, "step": 32100 }, { "epoch": 0.55, "learning_rate": 2.2728042686541883e-05, "loss": 3.2196, "step": 32200 }, { "epoch": 0.55, "learning_rate": 2.264334716693487e-05, "loss": 3.2, "step": 32300 }, { "epoch": 0.55, "learning_rate": 2.2558651647327858e-05, "loss": 3.2053, "step": 32400 }, { "epoch": 0.55, "learning_rate": 2.2473956127720844e-05, "loss": 3.2075, "step": 32500 }, { "epoch": 0.55, "learning_rate": 2.2389260608113833e-05, "loss": 3.2158, "step": 32600 }, { "epoch": 0.55, "learning_rate": 2.230456508850682e-05, "loss": 3.2149, "step": 32700 }, { "epoch": 0.56, "learning_rate": 2.2219869568899805e-05, "loss": 3.2082, "step": 32800 }, { "epoch": 0.56, "learning_rate": 2.213517404929279e-05, "loss": 3.2068, "step": 32900 }, { "epoch": 0.56, "learning_rate": 2.205047852968578e-05, "loss": 3.2078, "step": 33000 }, { "epoch": 0.56, "learning_rate": 2.196578301007877e-05, "loss": 3.2165, "step": 33100 }, { "epoch": 0.56, "learning_rate": 2.1881087490471756e-05, "loss": 3.2233, "step": 33200 }, { "epoch": 0.56, "learning_rate": 2.1796391970864745e-05, "loss": 3.2029, "step": 33300 }, { "epoch": 0.57, "learning_rate": 2.171169645125773e-05, "loss": 3.2079, "step": 33400 }, { "epoch": 0.57, "learning_rate": 2.1627000931650717e-05, "loss": 3.2147, "step": 33500 }, { "epoch": 0.57, "learning_rate": 2.1542305412043703e-05, "loss": 3.2175, "step": 33600 }, { "epoch": 0.57, "learning_rate": 2.1457609892436692e-05, "loss": 3.2167, "step": 33700 }, { "epoch": 0.57, "learning_rate": 2.1372914372829678e-05, "loss": 3.1926, "step": 33800 }, { "epoch": 0.57, "learning_rate": 2.1288218853222667e-05, "loss": 3.198, "step": 33900 }, { "epoch": 0.58, "learning_rate": 2.1203523333615653e-05, "loss": 3.2076, "step": 34000 }, { "epoch": 0.58, "learning_rate": 2.111882781400864e-05, "loss": 3.2119, "step": 34100 }, { "epoch": 0.58, "learning_rate": 2.1034132294401625e-05, "loss": 3.2017, "step": 34200 }, { "epoch": 0.58, "learning_rate": 2.0949436774794615e-05, "loss": 3.2065, "step": 34300 }, { "epoch": 0.58, "learning_rate": 2.08647412551876e-05, "loss": 3.1921, "step": 34400 }, { "epoch": 0.58, "learning_rate": 2.078004573558059e-05, "loss": 3.2301, "step": 34500 }, { "epoch": 0.59, "learning_rate": 2.0695350215973576e-05, "loss": 3.2104, "step": 34600 }, { "epoch": 0.59, "learning_rate": 2.0610654696366562e-05, "loss": 3.1944, "step": 34700 }, { "epoch": 0.59, "learning_rate": 2.052595917675955e-05, "loss": 3.2097, "step": 34800 }, { "epoch": 0.59, "learning_rate": 2.0441263657152537e-05, "loss": 3.1961, "step": 34900 }, { "epoch": 0.59, "learning_rate": 2.0356568137545526e-05, "loss": 3.2053, "step": 35000 }, { "epoch": 0.59, "learning_rate": 2.0271872617938512e-05, "loss": 3.1948, "step": 35100 }, { "epoch": 0.6, "learning_rate": 2.01871770983315e-05, "loss": 3.2027, "step": 35200 }, { "epoch": 0.6, "learning_rate": 2.0102481578724484e-05, "loss": 3.2068, "step": 35300 }, { "epoch": 0.6, "learning_rate": 2.0017786059117474e-05, "loss": 3.2119, "step": 35400 }, { "epoch": 0.6, "learning_rate": 1.993309053951046e-05, "loss": 3.1933, "step": 35500 }, { "epoch": 0.6, "learning_rate": 1.984839501990345e-05, "loss": 3.1979, "step": 35600 }, { "epoch": 0.6, "learning_rate": 1.9763699500296435e-05, "loss": 3.2042, "step": 35700 }, { "epoch": 0.61, "learning_rate": 1.9679003980689424e-05, "loss": 3.1985, "step": 35800 }, { "epoch": 0.61, "learning_rate": 1.9594308461082407e-05, "loss": 3.2023, "step": 35900 }, { "epoch": 0.61, "learning_rate": 1.9509612941475396e-05, "loss": 3.2023, "step": 36000 }, { "epoch": 0.61, "learning_rate": 1.9424917421868386e-05, "loss": 3.2116, "step": 36100 }, { "epoch": 0.61, "learning_rate": 1.934022190226137e-05, "loss": 3.1966, "step": 36200 }, { "epoch": 0.61, "learning_rate": 1.925552638265436e-05, "loss": 3.1994, "step": 36300 }, { "epoch": 0.62, "learning_rate": 1.9170830863047347e-05, "loss": 3.1967, "step": 36400 }, { "epoch": 0.62, "learning_rate": 1.9086135343440333e-05, "loss": 3.2149, "step": 36500 }, { "epoch": 0.62, "learning_rate": 1.900143982383332e-05, "loss": 3.202, "step": 36600 }, { "epoch": 0.62, "learning_rate": 1.8916744304226308e-05, "loss": 3.2015, "step": 36700 }, { "epoch": 0.62, "learning_rate": 1.8832048784619294e-05, "loss": 3.2088, "step": 36800 }, { "epoch": 0.63, "learning_rate": 1.8747353265012283e-05, "loss": 3.1967, "step": 36900 }, { "epoch": 0.63, "learning_rate": 1.866265774540527e-05, "loss": 3.2063, "step": 37000 }, { "epoch": 0.63, "learning_rate": 1.8577962225798255e-05, "loss": 3.2164, "step": 37100 }, { "epoch": 0.63, "learning_rate": 1.849326670619124e-05, "loss": 3.177, "step": 37200 }, { "epoch": 0.63, "learning_rate": 1.840857118658423e-05, "loss": 3.1965, "step": 37300 }, { "epoch": 0.63, "learning_rate": 1.8323875666977216e-05, "loss": 3.1846, "step": 37400 }, { "epoch": 0.64, "learning_rate": 1.8239180147370206e-05, "loss": 3.1875, "step": 37500 }, { "epoch": 0.64, "learning_rate": 1.8154484627763192e-05, "loss": 3.2023, "step": 37600 }, { "epoch": 0.64, "learning_rate": 1.8069789108156178e-05, "loss": 3.203, "step": 37700 }, { "epoch": 0.64, "learning_rate": 1.7985093588549167e-05, "loss": 3.1991, "step": 37800 }, { "epoch": 0.64, "learning_rate": 1.7900398068942153e-05, "loss": 3.2008, "step": 37900 }, { "epoch": 0.64, "learning_rate": 1.7815702549335142e-05, "loss": 3.1955, "step": 38000 }, { "epoch": 0.65, "learning_rate": 1.7731007029728128e-05, "loss": 3.1963, "step": 38100 }, { "epoch": 0.65, "learning_rate": 1.7646311510121118e-05, "loss": 3.1964, "step": 38200 }, { "epoch": 0.65, "learning_rate": 1.7561615990514104e-05, "loss": 3.1881, "step": 38300 }, { "epoch": 0.65, "learning_rate": 1.747692047090709e-05, "loss": 3.1924, "step": 38400 }, { "epoch": 0.65, "learning_rate": 1.7392224951300075e-05, "loss": 3.1953, "step": 38500 }, { "epoch": 0.65, "learning_rate": 1.7307529431693065e-05, "loss": 3.1933, "step": 38600 }, { "epoch": 0.66, "learning_rate": 1.722283391208605e-05, "loss": 3.1976, "step": 38700 }, { "epoch": 0.66, "learning_rate": 1.713813839247904e-05, "loss": 3.1868, "step": 38800 }, { "epoch": 0.66, "learning_rate": 1.7053442872872026e-05, "loss": 3.191, "step": 38900 }, { "epoch": 0.66, "learning_rate": 1.6968747353265012e-05, "loss": 3.1914, "step": 39000 }, { "epoch": 0.66, "learning_rate": 1.6884051833657998e-05, "loss": 3.1831, "step": 39100 }, { "epoch": 0.66, "learning_rate": 1.6799356314050987e-05, "loss": 3.2032, "step": 39200 }, { "epoch": 0.67, "learning_rate": 1.6714660794443977e-05, "loss": 3.2008, "step": 39300 }, { "epoch": 0.67, "learning_rate": 1.6629965274836963e-05, "loss": 3.191, "step": 39400 }, { "epoch": 0.67, "learning_rate": 1.654526975522995e-05, "loss": 3.1821, "step": 39500 }, { "epoch": 0.67, "learning_rate": 1.6460574235622935e-05, "loss": 3.1753, "step": 39600 }, { "epoch": 0.67, "learning_rate": 1.6375878716015924e-05, "loss": 3.1947, "step": 39700 }, { "epoch": 0.67, "learning_rate": 1.629118319640891e-05, "loss": 3.1815, "step": 39800 }, { "epoch": 0.68, "learning_rate": 1.62064876768019e-05, "loss": 3.1991, "step": 39900 }, { "epoch": 0.68, "learning_rate": 1.6121792157194885e-05, "loss": 3.1912, "step": 40000 }, { "epoch": 0.68, "learning_rate": 1.6037096637587874e-05, "loss": 3.1912, "step": 40100 }, { "epoch": 0.68, "learning_rate": 1.5952401117980857e-05, "loss": 3.1865, "step": 40200 }, { "epoch": 0.68, "learning_rate": 1.5867705598373846e-05, "loss": 3.1933, "step": 40300 }, { "epoch": 0.68, "learning_rate": 1.5783010078766832e-05, "loss": 3.2054, "step": 40400 }, { "epoch": 0.69, "learning_rate": 1.569831455915982e-05, "loss": 3.187, "step": 40500 }, { "epoch": 0.69, "learning_rate": 1.5613619039552808e-05, "loss": 3.207, "step": 40600 }, { "epoch": 0.69, "learning_rate": 1.5528923519945797e-05, "loss": 3.1933, "step": 40700 }, { "epoch": 0.69, "learning_rate": 1.5444228000338783e-05, "loss": 3.1906, "step": 40800 }, { "epoch": 0.69, "learning_rate": 1.535953248073177e-05, "loss": 3.1899, "step": 40900 }, { "epoch": 0.69, "learning_rate": 1.5274836961124758e-05, "loss": 3.1834, "step": 41000 }, { "epoch": 0.7, "learning_rate": 1.5190141441517744e-05, "loss": 3.1793, "step": 41100 }, { "epoch": 0.7, "learning_rate": 1.5105445921910732e-05, "loss": 3.1903, "step": 41200 }, { "epoch": 0.7, "learning_rate": 1.5020750402303718e-05, "loss": 3.1749, "step": 41300 }, { "epoch": 0.7, "learning_rate": 1.4936054882696707e-05, "loss": 3.2057, "step": 41400 }, { "epoch": 0.7, "learning_rate": 1.4851359363089693e-05, "loss": 3.1976, "step": 41500 }, { "epoch": 0.7, "learning_rate": 1.476666384348268e-05, "loss": 3.1903, "step": 41600 }, { "epoch": 0.71, "learning_rate": 1.4681968323875667e-05, "loss": 3.1933, "step": 41700 }, { "epoch": 0.71, "learning_rate": 1.4597272804268656e-05, "loss": 3.1893, "step": 41800 }, { "epoch": 0.71, "learning_rate": 1.451257728466164e-05, "loss": 3.1758, "step": 41900 }, { "epoch": 0.71, "learning_rate": 1.442788176505463e-05, "loss": 3.1862, "step": 42000 }, { "epoch": 0.71, "learning_rate": 1.4343186245447616e-05, "loss": 3.1919, "step": 42100 }, { "epoch": 0.71, "learning_rate": 1.4258490725840603e-05, "loss": 3.1743, "step": 42200 }, { "epoch": 0.72, "learning_rate": 1.4173795206233593e-05, "loss": 3.1762, "step": 42300 }, { "epoch": 0.72, "learning_rate": 1.4089099686626578e-05, "loss": 3.1939, "step": 42400 }, { "epoch": 0.72, "learning_rate": 1.4004404167019566e-05, "loss": 3.1814, "step": 42500 }, { "epoch": 0.72, "learning_rate": 1.3919708647412552e-05, "loss": 3.1925, "step": 42600 }, { "epoch": 0.72, "learning_rate": 1.3835013127805541e-05, "loss": 3.1708, "step": 42700 }, { "epoch": 0.72, "learning_rate": 1.3750317608198526e-05, "loss": 3.1635, "step": 42800 }, { "epoch": 0.73, "learning_rate": 1.3665622088591515e-05, "loss": 3.1827, "step": 42900 }, { "epoch": 0.73, "learning_rate": 1.3580926568984501e-05, "loss": 3.1721, "step": 43000 }, { "epoch": 0.73, "learning_rate": 1.3496231049377489e-05, "loss": 3.1853, "step": 43100 }, { "epoch": 0.73, "learning_rate": 1.3411535529770475e-05, "loss": 3.1747, "step": 43200 }, { "epoch": 0.73, "learning_rate": 1.3326840010163464e-05, "loss": 3.1799, "step": 43300 }, { "epoch": 0.74, "learning_rate": 1.3242144490556448e-05, "loss": 3.1843, "step": 43400 }, { "epoch": 0.74, "learning_rate": 1.3157448970949438e-05, "loss": 3.1985, "step": 43500 }, { "epoch": 0.74, "learning_rate": 1.3072753451342423e-05, "loss": 3.1907, "step": 43600 }, { "epoch": 0.74, "learning_rate": 1.2988057931735411e-05, "loss": 3.1804, "step": 43700 }, { "epoch": 0.74, "learning_rate": 1.29033624121284e-05, "loss": 3.1787, "step": 43800 }, { "epoch": 0.74, "learning_rate": 1.2818666892521386e-05, "loss": 3.181, "step": 43900 }, { "epoch": 0.75, "learning_rate": 1.2733971372914374e-05, "loss": 3.1677, "step": 44000 }, { "epoch": 0.75, "learning_rate": 1.264927585330736e-05, "loss": 3.1876, "step": 44100 }, { "epoch": 0.75, "learning_rate": 1.256458033370035e-05, "loss": 3.1952, "step": 44200 }, { "epoch": 0.75, "learning_rate": 1.2479884814093334e-05, "loss": 3.1881, "step": 44300 }, { "epoch": 0.75, "learning_rate": 1.2395189294486323e-05, "loss": 3.1789, "step": 44400 }, { "epoch": 0.75, "learning_rate": 1.231049377487931e-05, "loss": 3.177, "step": 44500 }, { "epoch": 0.76, "learning_rate": 1.2225798255272297e-05, "loss": 3.1733, "step": 44600 }, { "epoch": 0.76, "learning_rate": 1.2141102735665284e-05, "loss": 3.17, "step": 44700 }, { "epoch": 0.76, "learning_rate": 1.2056407216058272e-05, "loss": 3.1849, "step": 44800 }, { "epoch": 0.76, "learning_rate": 1.1971711696451258e-05, "loss": 3.1875, "step": 44900 }, { "epoch": 0.76, "learning_rate": 1.1887016176844245e-05, "loss": 3.1659, "step": 45000 }, { "epoch": 0.76, "learning_rate": 1.1802320657237233e-05, "loss": 3.1877, "step": 45100 }, { "epoch": 0.77, "learning_rate": 1.1717625137630219e-05, "loss": 3.1808, "step": 45200 }, { "epoch": 0.77, "learning_rate": 1.1632929618023207e-05, "loss": 3.1831, "step": 45300 }, { "epoch": 0.77, "learning_rate": 1.1548234098416194e-05, "loss": 3.1852, "step": 45400 }, { "epoch": 0.77, "learning_rate": 1.1463538578809182e-05, "loss": 3.1855, "step": 45500 }, { "epoch": 0.77, "learning_rate": 1.1378843059202168e-05, "loss": 3.1771, "step": 45600 }, { "epoch": 0.77, "learning_rate": 1.1294147539595156e-05, "loss": 3.1826, "step": 45700 }, { "epoch": 0.78, "learning_rate": 1.1209452019988143e-05, "loss": 3.1634, "step": 45800 }, { "epoch": 0.78, "learning_rate": 1.1124756500381131e-05, "loss": 3.1812, "step": 45900 }, { "epoch": 0.78, "learning_rate": 1.1040060980774119e-05, "loss": 3.1836, "step": 46000 }, { "epoch": 0.78, "learning_rate": 1.0955365461167104e-05, "loss": 3.1879, "step": 46100 }, { "epoch": 0.78, "learning_rate": 1.0870669941560092e-05, "loss": 3.1714, "step": 46200 }, { "epoch": 0.78, "learning_rate": 1.078597442195308e-05, "loss": 3.1747, "step": 46300 }, { "epoch": 0.79, "learning_rate": 1.0701278902346067e-05, "loss": 3.1795, "step": 46400 }, { "epoch": 0.79, "learning_rate": 1.0616583382739053e-05, "loss": 3.1776, "step": 46500 }, { "epoch": 0.79, "learning_rate": 1.0531887863132041e-05, "loss": 3.1687, "step": 46600 }, { "epoch": 0.79, "learning_rate": 1.0447192343525029e-05, "loss": 3.1842, "step": 46700 }, { "epoch": 0.79, "learning_rate": 1.0362496823918015e-05, "loss": 3.1702, "step": 46800 }, { "epoch": 0.79, "learning_rate": 1.0277801304311002e-05, "loss": 3.1601, "step": 46900 }, { "epoch": 0.8, "learning_rate": 1.019310578470399e-05, "loss": 3.1822, "step": 47000 }, { "epoch": 0.8, "learning_rate": 1.0108410265096976e-05, "loss": 3.1776, "step": 47100 }, { "epoch": 0.8, "learning_rate": 1.0023714745489964e-05, "loss": 3.1768, "step": 47200 }, { "epoch": 0.8, "learning_rate": 9.939019225882951e-06, "loss": 3.1709, "step": 47300 }, { "epoch": 0.8, "learning_rate": 9.854323706275939e-06, "loss": 3.172, "step": 47400 }, { "epoch": 0.8, "learning_rate": 9.769628186668926e-06, "loss": 3.1755, "step": 47500 }, { "epoch": 0.81, "learning_rate": 9.684932667061914e-06, "loss": 3.1717, "step": 47600 }, { "epoch": 0.81, "learning_rate": 9.6002371474549e-06, "loss": 3.1659, "step": 47700 }, { "epoch": 0.81, "learning_rate": 9.515541627847888e-06, "loss": 3.1713, "step": 47800 }, { "epoch": 0.81, "learning_rate": 9.430846108240875e-06, "loss": 3.1833, "step": 47900 }, { "epoch": 0.81, "learning_rate": 9.346150588633861e-06, "loss": 3.1755, "step": 48000 }, { "epoch": 0.81, "learning_rate": 9.261455069026849e-06, "loss": 3.1693, "step": 48100 }, { "epoch": 0.82, "learning_rate": 9.176759549419837e-06, "loss": 3.1777, "step": 48200 }, { "epoch": 0.82, "learning_rate": 9.092064029812823e-06, "loss": 3.1638, "step": 48300 }, { "epoch": 0.82, "learning_rate": 9.00736851020581e-06, "loss": 3.1786, "step": 48400 }, { "epoch": 0.82, "learning_rate": 8.922672990598798e-06, "loss": 3.1759, "step": 48500 }, { "epoch": 0.82, "learning_rate": 8.837977470991784e-06, "loss": 3.1712, "step": 48600 }, { "epoch": 0.82, "learning_rate": 8.753281951384771e-06, "loss": 3.1689, "step": 48700 }, { "epoch": 0.83, "learning_rate": 8.668586431777759e-06, "loss": 3.1861, "step": 48800 }, { "epoch": 0.83, "learning_rate": 8.583890912170747e-06, "loss": 3.1732, "step": 48900 }, { "epoch": 0.83, "learning_rate": 8.499195392563734e-06, "loss": 3.1735, "step": 49000 }, { "epoch": 0.83, "learning_rate": 8.414499872956722e-06, "loss": 3.1827, "step": 49100 }, { "epoch": 0.83, "learning_rate": 8.329804353349708e-06, "loss": 3.1681, "step": 49200 }, { "epoch": 0.84, "learning_rate": 8.245108833742696e-06, "loss": 3.1506, "step": 49300 }, { "epoch": 0.84, "learning_rate": 8.160413314135683e-06, "loss": 3.1628, "step": 49400 }, { "epoch": 0.84, "learning_rate": 8.07571779452867e-06, "loss": 3.1808, "step": 49500 }, { "epoch": 0.84, "learning_rate": 7.991022274921657e-06, "loss": 3.1682, "step": 49600 }, { "epoch": 0.84, "learning_rate": 7.906326755314645e-06, "loss": 3.1668, "step": 49700 }, { "epoch": 0.84, "learning_rate": 7.82163123570763e-06, "loss": 3.1663, "step": 49800 }, { "epoch": 0.85, "learning_rate": 7.736935716100618e-06, "loss": 3.1656, "step": 49900 }, { "epoch": 0.85, "learning_rate": 7.652240196493606e-06, "loss": 3.1749, "step": 50000 }, { "epoch": 0.85, "learning_rate": 7.567544676886593e-06, "loss": 3.1667, "step": 50100 }, { "epoch": 0.85, "learning_rate": 7.482849157279579e-06, "loss": 3.1658, "step": 50200 }, { "epoch": 0.85, "learning_rate": 7.398153637672567e-06, "loss": 3.1689, "step": 50300 }, { "epoch": 0.85, "learning_rate": 7.3134581180655555e-06, "loss": 3.1652, "step": 50400 }, { "epoch": 0.86, "learning_rate": 7.228762598458542e-06, "loss": 3.1625, "step": 50500 }, { "epoch": 0.86, "learning_rate": 7.14406707885153e-06, "loss": 3.1584, "step": 50600 }, { "epoch": 0.86, "learning_rate": 7.059371559244517e-06, "loss": 3.1695, "step": 50700 }, { "epoch": 0.86, "learning_rate": 6.974676039637504e-06, "loss": 3.1612, "step": 50800 }, { "epoch": 0.86, "learning_rate": 6.889980520030491e-06, "loss": 3.1564, "step": 50900 }, { "epoch": 0.86, "learning_rate": 6.805285000423478e-06, "loss": 3.1747, "step": 51000 }, { "epoch": 0.87, "learning_rate": 6.720589480816465e-06, "loss": 3.165, "step": 51100 }, { "epoch": 0.87, "learning_rate": 6.6358939612094525e-06, "loss": 3.1672, "step": 51200 }, { "epoch": 0.87, "learning_rate": 6.551198441602439e-06, "loss": 3.1618, "step": 51300 }, { "epoch": 0.87, "learning_rate": 6.466502921995427e-06, "loss": 3.172, "step": 51400 }, { "epoch": 0.87, "learning_rate": 6.381807402388414e-06, "loss": 3.1683, "step": 51500 }, { "epoch": 0.87, "learning_rate": 6.2971118827814005e-06, "loss": 3.1752, "step": 51600 }, { "epoch": 0.88, "learning_rate": 6.212416363174388e-06, "loss": 3.1698, "step": 51700 }, { "epoch": 0.88, "learning_rate": 6.127720843567376e-06, "loss": 3.1615, "step": 51800 }, { "epoch": 0.88, "learning_rate": 6.043025323960363e-06, "loss": 3.1667, "step": 51900 }, { "epoch": 0.88, "learning_rate": 5.95832980435335e-06, "loss": 3.1699, "step": 52000 }, { "epoch": 0.88, "learning_rate": 5.873634284746337e-06, "loss": 3.1621, "step": 52100 }, { "epoch": 0.88, "learning_rate": 5.788938765139324e-06, "loss": 3.1789, "step": 52200 }, { "epoch": 0.89, "learning_rate": 5.7042432455323115e-06, "loss": 3.1679, "step": 52300 }, { "epoch": 0.89, "learning_rate": 5.619547725925299e-06, "loss": 3.1751, "step": 52400 }, { "epoch": 0.89, "learning_rate": 5.534852206318286e-06, "loss": 3.1589, "step": 52500 }, { "epoch": 0.89, "learning_rate": 5.450156686711274e-06, "loss": 3.1644, "step": 52600 }, { "epoch": 0.89, "learning_rate": 5.36546116710426e-06, "loss": 3.1777, "step": 52700 }, { "epoch": 0.89, "learning_rate": 5.280765647497247e-06, "loss": 3.1678, "step": 52800 }, { "epoch": 0.9, "learning_rate": 5.196070127890235e-06, "loss": 3.1648, "step": 52900 }, { "epoch": 0.9, "learning_rate": 5.111374608283222e-06, "loss": 3.1621, "step": 53000 }, { "epoch": 0.9, "learning_rate": 5.026679088676209e-06, "loss": 3.1644, "step": 53100 }, { "epoch": 0.9, "learning_rate": 4.941983569069197e-06, "loss": 3.1822, "step": 53200 }, { "epoch": 0.9, "learning_rate": 4.857288049462184e-06, "loss": 3.173, "step": 53300 }, { "epoch": 0.9, "learning_rate": 4.7725925298551706e-06, "loss": 3.1617, "step": 53400 }, { "epoch": 0.91, "learning_rate": 4.687897010248158e-06, "loss": 3.1615, "step": 53500 }, { "epoch": 0.91, "learning_rate": 4.603201490641145e-06, "loss": 3.1535, "step": 53600 }, { "epoch": 0.91, "learning_rate": 4.518505971034133e-06, "loss": 3.1688, "step": 53700 }, { "epoch": 0.91, "learning_rate": 4.43381045142712e-06, "loss": 3.1839, "step": 53800 }, { "epoch": 0.91, "learning_rate": 4.349114931820107e-06, "loss": 3.1769, "step": 53900 }, { "epoch": 0.91, "learning_rate": 4.264419412213094e-06, "loss": 3.1649, "step": 54000 }, { "epoch": 0.92, "learning_rate": 4.1797238926060815e-06, "loss": 3.1644, "step": 54100 }, { "epoch": 0.92, "learning_rate": 4.095028372999068e-06, "loss": 3.1616, "step": 54200 }, { "epoch": 0.92, "learning_rate": 4.010332853392056e-06, "loss": 3.1578, "step": 54300 }, { "epoch": 0.92, "learning_rate": 3.925637333785043e-06, "loss": 3.1621, "step": 54400 }, { "epoch": 0.92, "learning_rate": 3.84094181417803e-06, "loss": 3.1523, "step": 54500 }, { "epoch": 0.92, "learning_rate": 3.7562462945710177e-06, "loss": 3.1667, "step": 54600 }, { "epoch": 0.93, "learning_rate": 3.671550774964005e-06, "loss": 3.1748, "step": 54700 }, { "epoch": 0.93, "learning_rate": 3.5868552553569917e-06, "loss": 3.1773, "step": 54800 }, { "epoch": 0.93, "learning_rate": 3.502159735749979e-06, "loss": 3.1418, "step": 54900 }, { "epoch": 0.93, "learning_rate": 3.417464216142966e-06, "loss": 3.1613, "step": 55000 }, { "epoch": 0.93, "learning_rate": 3.3327686965359534e-06, "loss": 3.154, "step": 55100 }, { "epoch": 0.94, "learning_rate": 3.24807317692894e-06, "loss": 3.1776, "step": 55200 }, { "epoch": 0.94, "learning_rate": 3.1633776573219282e-06, "loss": 3.1513, "step": 55300 }, { "epoch": 0.94, "learning_rate": 3.078682137714915e-06, "loss": 3.1608, "step": 55400 }, { "epoch": 0.94, "learning_rate": 2.9939866181079023e-06, "loss": 3.1759, "step": 55500 }, { "epoch": 0.94, "learning_rate": 2.9092910985008895e-06, "loss": 3.1549, "step": 55600 }, { "epoch": 0.94, "learning_rate": 2.8245955788938767e-06, "loss": 3.1651, "step": 55700 }, { "epoch": 0.95, "learning_rate": 2.739900059286864e-06, "loss": 3.167, "step": 55800 }, { "epoch": 0.95, "learning_rate": 2.655204539679851e-06, "loss": 3.1577, "step": 55900 }, { "epoch": 0.95, "learning_rate": 2.5705090200728384e-06, "loss": 3.1514, "step": 56000 }, { "epoch": 0.95, "learning_rate": 2.4858135004658256e-06, "loss": 3.1693, "step": 56100 }, { "epoch": 0.95, "learning_rate": 2.4011179808588124e-06, "loss": 3.1473, "step": 56200 }, { "epoch": 0.95, "learning_rate": 2.3164224612518e-06, "loss": 3.1574, "step": 56300 }, { "epoch": 0.96, "learning_rate": 2.2317269416447873e-06, "loss": 3.1603, "step": 56400 }, { "epoch": 0.96, "learning_rate": 2.147031422037774e-06, "loss": 3.1687, "step": 56500 }, { "epoch": 0.96, "learning_rate": 2.0623359024307617e-06, "loss": 3.1608, "step": 56600 }, { "epoch": 0.96, "learning_rate": 1.977640382823749e-06, "loss": 3.1538, "step": 56700 }, { "epoch": 0.96, "learning_rate": 1.892944863216736e-06, "loss": 3.1622, "step": 56800 }, { "epoch": 0.96, "learning_rate": 1.808249343609723e-06, "loss": 3.1582, "step": 56900 }, { "epoch": 0.97, "learning_rate": 1.7235538240027104e-06, "loss": 3.1573, "step": 57000 }, { "epoch": 0.97, "learning_rate": 1.6388583043956976e-06, "loss": 3.1431, "step": 57100 }, { "epoch": 0.97, "learning_rate": 1.5541627847886846e-06, "loss": 3.1634, "step": 57200 }, { "epoch": 0.97, "learning_rate": 1.469467265181672e-06, "loss": 3.1581, "step": 57300 }, { "epoch": 0.97, "learning_rate": 1.384771745574659e-06, "loss": 3.1531, "step": 57400 }, { "epoch": 0.97, "learning_rate": 1.3000762259676463e-06, "loss": 3.1531, "step": 57500 }, { "epoch": 0.98, "learning_rate": 1.2153807063606335e-06, "loss": 3.1499, "step": 57600 }, { "epoch": 0.98, "learning_rate": 1.1306851867536208e-06, "loss": 3.1624, "step": 57700 }, { "epoch": 0.98, "learning_rate": 1.045989667146608e-06, "loss": 3.1609, "step": 57800 }, { "epoch": 0.98, "learning_rate": 9.612941475395952e-07, "loss": 3.1707, "step": 57900 }, { "epoch": 0.98, "learning_rate": 8.765986279325823e-07, "loss": 3.1626, "step": 58000 }, { "epoch": 0.98, "learning_rate": 7.919031083255696e-07, "loss": 3.1698, "step": 58100 }, { "epoch": 0.99, "learning_rate": 7.072075887185568e-07, "loss": 3.1589, "step": 58200 }, { "epoch": 0.99, "learning_rate": 6.22512069111544e-07, "loss": 3.1557, "step": 58300 }, { "epoch": 0.99, "learning_rate": 5.378165495045312e-07, "loss": 3.1603, "step": 58400 }, { "epoch": 0.99, "learning_rate": 4.5312102989751844e-07, "loss": 3.1588, "step": 58500 }, { "epoch": 0.99, "learning_rate": 3.6842551029050566e-07, "loss": 3.1622, "step": 58600 }, { "epoch": 0.99, "learning_rate": 2.837299906834929e-07, "loss": 3.1547, "step": 58700 }, { "epoch": 1.0, "learning_rate": 1.9903447107648006e-07, "loss": 3.1522, "step": 58800 }, { "epoch": 1.0, "learning_rate": 1.1433895146946727e-07, "loss": 3.1583, "step": 58900 }, { "epoch": 1.0, "learning_rate": 2.9643431862454477e-08, "loss": 3.1542, "step": 59000 }, { "epoch": 1.0, "step": 59035, "total_flos": 2.318685330743415e+18, "train_loss": 3.2513424668918267, "train_runtime": 267734.5376, "train_samples_per_second": 7.056, "train_steps_per_second": 0.22 } ], "max_steps": 59035, "num_train_epochs": 1, "total_flos": 2.318685330743415e+18, "trial_name": null, "trial_params": null }