{ "best_metric": null, "best_model_checkpoint": null, "epoch": 6.740196078431373, "global_step": 110000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.03, "learning_rate": 2.3999999999999997e-05, "loss": 0.8848, "step": 500 }, { "epoch": 0.06, "learning_rate": 4.7999999999999994e-05, "loss": 0.7411, "step": 1000 }, { "epoch": 0.06, "eval_loss": 0.9070383906364441, "eval_runtime": 1.3223, "eval_samples_per_second": 756.258, "eval_steps_per_second": 12.1, "step": 1000 }, { "epoch": 0.09, "learning_rate": 7.199999999999999e-05, "loss": 0.7398, "step": 1500 }, { "epoch": 0.12, "learning_rate": 9.599999999999999e-05, "loss": 0.7395, "step": 2000 }, { "epoch": 0.12, "eval_loss": 0.9064154028892517, "eval_runtime": 1.194, "eval_samples_per_second": 837.495, "eval_steps_per_second": 13.4, "step": 2000 }, { "epoch": 0.15, "learning_rate": 0.00011999999999999999, "loss": 0.7392, "step": 2500 }, { "epoch": 0.18, "learning_rate": 0.00014399999999999998, "loss": 0.7387, "step": 3000 }, { "epoch": 0.18, "eval_loss": 0.9047173857688904, "eval_runtime": 1.3077, "eval_samples_per_second": 764.675, "eval_steps_per_second": 12.235, "step": 3000 }, { "epoch": 0.21, "learning_rate": 0.000168, "loss": 0.7384, "step": 3500 }, { "epoch": 0.25, "learning_rate": 0.00019199999999999998, "loss": 0.7382, "step": 4000 }, { "epoch": 0.25, "eval_loss": 0.9014638662338257, "eval_runtime": 1.2645, "eval_samples_per_second": 790.849, "eval_steps_per_second": 12.654, "step": 4000 }, { "epoch": 0.28, "learning_rate": 0.00021599999999999996, "loss": 0.7382, "step": 4500 }, { "epoch": 0.31, "learning_rate": 0.00023999999999999998, "loss": 0.7381, "step": 5000 }, { "epoch": 0.31, "eval_loss": 0.9044438600540161, "eval_runtime": 1.2394, "eval_samples_per_second": 806.828, "eval_steps_per_second": 12.909, "step": 5000 }, { "epoch": 0.34, "learning_rate": 0.00026399999999999997, "loss": 0.7379, "step": 5500 }, { "epoch": 0.37, "learning_rate": 0.00028799999999999995, "loss": 0.7379, "step": 6000 }, { "epoch": 0.37, "eval_loss": 0.9041813015937805, "eval_runtime": 1.2368, "eval_samples_per_second": 808.566, "eval_steps_per_second": 12.937, "step": 6000 }, { "epoch": 0.4, "learning_rate": 0.000312, "loss": 0.7381, "step": 6500 }, { "epoch": 0.43, "learning_rate": 0.000336, "loss": 0.7379, "step": 7000 }, { "epoch": 0.43, "eval_loss": 0.9053561687469482, "eval_runtime": 1.2638, "eval_samples_per_second": 791.247, "eval_steps_per_second": 12.66, "step": 7000 }, { "epoch": 0.46, "learning_rate": 0.00035999999999999997, "loss": 0.7378, "step": 7500 }, { "epoch": 0.49, "learning_rate": 0.00038399999999999996, "loss": 0.7378, "step": 8000 }, { "epoch": 0.49, "eval_loss": 0.9035025835037231, "eval_runtime": 1.2079, "eval_samples_per_second": 827.911, "eval_steps_per_second": 13.247, "step": 8000 }, { "epoch": 0.52, "learning_rate": 0.000408, "loss": 0.7378, "step": 8500 }, { "epoch": 0.55, "learning_rate": 0.00043199999999999993, "loss": 0.7378, "step": 9000 }, { "epoch": 0.55, "eval_loss": 0.9025757312774658, "eval_runtime": 1.2745, "eval_samples_per_second": 784.598, "eval_steps_per_second": 12.554, "step": 9000 }, { "epoch": 0.58, "learning_rate": 0.00045599999999999997, "loss": 0.7375, "step": 9500 }, { "epoch": 0.61, "learning_rate": 0.00047999999999999996, "loss": 0.7371, "step": 10000 }, { "epoch": 0.61, "eval_loss": 0.903812050819397, "eval_runtime": 1.2376, "eval_samples_per_second": 807.983, "eval_steps_per_second": 12.928, "step": 10000 }, { "epoch": 0.64, "learning_rate": 0.0005039999999999999, "loss": 0.7369, "step": 10500 }, { "epoch": 0.67, "learning_rate": 0.0005279999999999999, "loss": 0.7369, "step": 11000 }, { "epoch": 0.67, "eval_loss": 0.9027426838874817, "eval_runtime": 1.2459, "eval_samples_per_second": 802.657, "eval_steps_per_second": 12.843, "step": 11000 }, { "epoch": 0.7, "learning_rate": 0.000552, "loss": 0.7369, "step": 11500 }, { "epoch": 0.74, "learning_rate": 0.0005759999999999999, "loss": 0.7368, "step": 12000 }, { "epoch": 0.74, "eval_loss": 0.9022310972213745, "eval_runtime": 1.2295, "eval_samples_per_second": 813.368, "eval_steps_per_second": 13.014, "step": 12000 }, { "epoch": 0.77, "learning_rate": 0.0006, "loss": 0.7367, "step": 12500 }, { "epoch": 0.8, "learning_rate": 0.0005999935478721662, "loss": 0.7368, "step": 13000 }, { "epoch": 0.8, "eval_loss": 0.8987133502960205, "eval_runtime": 1.2643, "eval_samples_per_second": 790.939, "eval_steps_per_second": 12.655, "step": 13000 }, { "epoch": 0.83, "learning_rate": 0.000599974191770902, "loss": 0.7375, "step": 13500 }, { "epoch": 0.86, "learning_rate": 0.0005999419325429058, "loss": 0.7374, "step": 14000 }, { "epoch": 0.86, "eval_loss": 0.9013972282409668, "eval_runtime": 1.236, "eval_samples_per_second": 809.042, "eval_steps_per_second": 12.945, "step": 14000 }, { "epoch": 0.89, "learning_rate": 0.0005998967715993009, "loss": 0.7369, "step": 14500 }, { "epoch": 0.92, "learning_rate": 0.0005998387109155732, "loss": 0.7369, "step": 15000 }, { "epoch": 0.92, "eval_loss": 0.9001522660255432, "eval_runtime": 1.235, "eval_samples_per_second": 809.697, "eval_steps_per_second": 12.955, "step": 15000 }, { "epoch": 0.95, "learning_rate": 0.000599767753031485, "loss": 0.737, "step": 15500 }, { "epoch": 0.98, "learning_rate": 0.0005996839010509641, "loss": 0.7369, "step": 16000 }, { "epoch": 0.98, "eval_loss": 0.9001864790916443, "eval_runtime": 1.1685, "eval_samples_per_second": 855.784, "eval_steps_per_second": 13.693, "step": 16000 }, { "epoch": 1.01, "learning_rate": 0.0005995871586419678, "loss": 0.7369, "step": 16500 }, { "epoch": 1.04, "learning_rate": 0.0005994775300363225, "loss": 0.7372, "step": 17000 }, { "epoch": 1.04, "eval_loss": 0.9018534421920776, "eval_runtime": 1.1865, "eval_samples_per_second": 842.819, "eval_steps_per_second": 13.485, "step": 17000 }, { "epoch": 1.07, "learning_rate": 0.0005993550200295384, "loss": 0.7371, "step": 17500 }, { "epoch": 1.1, "learning_rate": 0.0005992196339806002, "loss": 0.737, "step": 18000 }, { "epoch": 1.1, "eval_loss": 0.9001176953315735, "eval_runtime": 1.2219, "eval_samples_per_second": 818.375, "eval_steps_per_second": 13.094, "step": 18000 }, { "epoch": 1.13, "learning_rate": 0.0005990713778117324, "loss": 0.7369, "step": 18500 }, { "epoch": 1.16, "learning_rate": 0.0005989102580081398, "loss": 0.737, "step": 19000 }, { "epoch": 1.16, "eval_loss": 0.9006143808364868, "eval_runtime": 1.2544, "eval_samples_per_second": 797.163, "eval_steps_per_second": 12.755, "step": 19000 }, { "epoch": 1.19, "learning_rate": 0.0005987362816177249, "loss": 0.7369, "step": 19500 }, { "epoch": 1.23, "learning_rate": 0.0005985494562507783, "loss": 0.7369, "step": 20000 }, { "epoch": 1.23, "eval_loss": 0.9006676077842712, "eval_runtime": 1.2815, "eval_samples_per_second": 780.312, "eval_steps_per_second": 12.485, "step": 20000 }, { "epoch": 1.26, "learning_rate": 0.000598349790079647, "loss": 0.737, "step": 20500 }, { "epoch": 1.29, "learning_rate": 0.000598137291838376, "loss": 0.7365, "step": 21000 }, { "epoch": 1.29, "eval_loss": 0.8698467016220093, "eval_runtime": 1.2584, "eval_samples_per_second": 794.654, "eval_steps_per_second": 12.714, "step": 21000 }, { "epoch": 1.32, "learning_rate": 0.000597911970822327, "loss": 0.7364, "step": 21500 }, { "epoch": 1.35, "learning_rate": 0.000597673836887771, "loss": 0.7363, "step": 22000 }, { "epoch": 1.35, "eval_loss": 0.870046079158783, "eval_runtime": 1.2567, "eval_samples_per_second": 795.734, "eval_steps_per_second": 12.732, "step": 22000 }, { "epoch": 1.38, "learning_rate": 0.0005974229004514577, "loss": 0.7363, "step": 22500 }, { "epoch": 1.41, "learning_rate": 0.0005971591724901598, "loss": 0.7366, "step": 23000 }, { "epoch": 1.41, "eval_loss": 0.9021453857421875, "eval_runtime": 1.277, "eval_samples_per_second": 783.107, "eval_steps_per_second": 12.53, "step": 23000 }, { "epoch": 1.44, "learning_rate": 0.0005968826645401927, "loss": 0.7368, "step": 23500 }, { "epoch": 1.47, "learning_rate": 0.00059659338869691, "loss": 0.7362, "step": 24000 }, { "epoch": 1.47, "eval_loss": 0.8762597441673279, "eval_runtime": 1.2918, "eval_samples_per_second": 774.091, "eval_steps_per_second": 12.385, "step": 24000 }, { "epoch": 1.5, "learning_rate": 0.0005962913576141742, "loss": 0.7354, "step": 24500 }, { "epoch": 1.53, "learning_rate": 0.000595976584503803, "loss": 0.7082, "step": 25000 }, { "epoch": 1.53, "eval_loss": 0.8719052672386169, "eval_runtime": 1.2648, "eval_samples_per_second": 790.61, "eval_steps_per_second": 12.65, "step": 25000 }, { "epoch": 1.56, "learning_rate": 0.0005956490831349923, "loss": 0.6914, "step": 25500 }, { "epoch": 1.59, "learning_rate": 0.0005953088678337129, "loss": 0.6774, "step": 26000 }, { "epoch": 1.59, "eval_loss": 0.8876385688781738, "eval_runtime": 1.3046, "eval_samples_per_second": 766.511, "eval_steps_per_second": 12.264, "step": 26000 }, { "epoch": 1.62, "learning_rate": 0.0005949559534820841, "loss": 0.6675, "step": 26500 }, { "epoch": 1.65, "learning_rate": 0.0005945903555177229, "loss": 0.6525, "step": 27000 }, { "epoch": 1.65, "eval_loss": 0.8904880285263062, "eval_runtime": 1.319, "eval_samples_per_second": 758.176, "eval_steps_per_second": 12.131, "step": 27000 }, { "epoch": 1.69, "learning_rate": 0.0005942120899330687, "loss": 0.6186, "step": 27500 }, { "epoch": 1.72, "learning_rate": 0.0005938211732746836, "loss": 0.6022, "step": 28000 }, { "epoch": 1.72, "eval_loss": 0.8856265544891357, "eval_runtime": 1.2877, "eval_samples_per_second": 776.587, "eval_steps_per_second": 12.425, "step": 28000 }, { "epoch": 1.75, "learning_rate": 0.0005934176226425286, "loss": 0.593, "step": 28500 }, { "epoch": 1.78, "learning_rate": 0.0005930014556892158, "loss": 0.5874, "step": 29000 }, { "epoch": 1.78, "eval_loss": 0.8793675303459167, "eval_runtime": 1.2489, "eval_samples_per_second": 800.707, "eval_steps_per_second": 12.811, "step": 29000 }, { "epoch": 1.81, "learning_rate": 0.0005925726906192357, "loss": 0.5837, "step": 29500 }, { "epoch": 1.84, "learning_rate": 0.0005921313461881617, "loss": 0.5765, "step": 30000 }, { "epoch": 1.84, "eval_loss": 0.8805580735206604, "eval_runtime": 1.2758, "eval_samples_per_second": 783.816, "eval_steps_per_second": 12.541, "step": 30000 }, { "epoch": 1.87, "learning_rate": 0.0005916774417018287, "loss": 0.5714, "step": 30500 }, { "epoch": 1.9, "learning_rate": 0.0005912109970154897, "loss": 0.5685, "step": 31000 }, { "epoch": 1.9, "eval_loss": 0.8747313022613525, "eval_runtime": 1.3003, "eval_samples_per_second": 769.027, "eval_steps_per_second": 12.304, "step": 31000 }, { "epoch": 1.93, "learning_rate": 0.0005907320325329461, "loss": 0.566, "step": 31500 }, { "epoch": 1.96, "learning_rate": 0.0005902405692056561, "loss": 0.564, "step": 32000 }, { "epoch": 1.96, "eval_loss": 0.8779122233390808, "eval_runtime": 1.293, "eval_samples_per_second": 773.405, "eval_steps_per_second": 12.374, "step": 32000 }, { "epoch": 1.99, "learning_rate": 0.0005897366285318178, "loss": 0.5617, "step": 32500 }, { "epoch": 2.02, "learning_rate": 0.0005892202325554288, "loss": 0.5606, "step": 33000 }, { "epoch": 2.02, "eval_loss": 0.8761873245239258, "eval_runtime": 1.3816, "eval_samples_per_second": 723.824, "eval_steps_per_second": 11.581, "step": 33000 }, { "epoch": 2.05, "learning_rate": 0.0005886914038653217, "loss": 0.5583, "step": 33500 }, { "epoch": 2.08, "learning_rate": 0.0005881501655941771, "loss": 0.5574, "step": 34000 }, { "epoch": 2.08, "eval_loss": 0.8702684044837952, "eval_runtime": 1.3086, "eval_samples_per_second": 764.158, "eval_steps_per_second": 12.227, "step": 34000 }, { "epoch": 2.11, "learning_rate": 0.00058759654141751, "loss": 0.5548, "step": 34500 }, { "epoch": 2.14, "learning_rate": 0.0005870305555526355, "loss": 0.5528, "step": 35000 }, { "epoch": 2.14, "eval_loss": 0.8663867115974426, "eval_runtime": 1.2619, "eval_samples_per_second": 792.469, "eval_steps_per_second": 12.679, "step": 35000 }, { "epoch": 2.18, "learning_rate": 0.0005864522327576088, "loss": 0.5509, "step": 35500 }, { "epoch": 2.21, "learning_rate": 0.0005858615983301424, "loss": 0.5494, "step": 36000 }, { "epoch": 2.21, "eval_loss": 0.8717171549797058, "eval_runtime": 1.2691, "eval_samples_per_second": 787.953, "eval_steps_per_second": 12.607, "step": 36000 }, { "epoch": 2.24, "learning_rate": 0.0005852586781064997, "loss": 0.5471, "step": 36500 }, { "epoch": 2.27, "learning_rate": 0.0005846434984603645, "loss": 0.5448, "step": 37000 }, { "epoch": 2.27, "eval_loss": 0.8672583103179932, "eval_runtime": 1.268, "eval_samples_per_second": 788.629, "eval_steps_per_second": 12.618, "step": 37000 }, { "epoch": 2.3, "learning_rate": 0.0005840160863016872, "loss": 0.5433, "step": 37500 }, { "epoch": 2.33, "learning_rate": 0.0005833764690755083, "loss": 0.5419, "step": 38000 }, { "epoch": 2.33, "eval_loss": 0.8636866211891174, "eval_runtime": 1.3526, "eval_samples_per_second": 739.295, "eval_steps_per_second": 11.829, "step": 38000 }, { "epoch": 2.36, "learning_rate": 0.0005827246747607574, "loss": 0.5398, "step": 38500 }, { "epoch": 2.39, "learning_rate": 0.0005820607318690293, "loss": 0.5385, "step": 39000 }, { "epoch": 2.39, "eval_loss": 0.8634054064750671, "eval_runtime": 1.2705, "eval_samples_per_second": 787.122, "eval_steps_per_second": 12.594, "step": 39000 }, { "epoch": 2.42, "learning_rate": 0.0005813846694433368, "loss": 0.5374, "step": 39500 }, { "epoch": 2.45, "learning_rate": 0.0005806965170568409, "loss": 0.536, "step": 40000 }, { "epoch": 2.45, "eval_loss": 0.8661102652549744, "eval_runtime": 1.3401, "eval_samples_per_second": 746.204, "eval_steps_per_second": 11.939, "step": 40000 }, { "epoch": 2.48, "learning_rate": 0.0005799963048115559, "loss": 0.5353, "step": 40500 }, { "epoch": 2.51, "learning_rate": 0.0005792840633370341, "loss": 0.5336, "step": 41000 }, { "epoch": 2.51, "eval_loss": 0.8631040453910828, "eval_runtime": 1.3088, "eval_samples_per_second": 764.037, "eval_steps_per_second": 12.225, "step": 41000 }, { "epoch": 2.54, "learning_rate": 0.0005785598237890247, "loss": 0.5327, "step": 41500 }, { "epoch": 2.57, "learning_rate": 0.0005778236178481119, "loss": 0.5316, "step": 42000 }, { "epoch": 2.57, "eval_loss": 0.8605585098266602, "eval_runtime": 1.2836, "eval_samples_per_second": 779.077, "eval_steps_per_second": 12.465, "step": 42000 }, { "epoch": 2.6, "learning_rate": 0.0005770754777183285, "loss": 0.5306, "step": 42500 }, { "epoch": 2.63, "learning_rate": 0.0005763154361257473, "loss": 0.5297, "step": 43000 }, { "epoch": 2.63, "eval_loss": 0.8589205145835876, "eval_runtime": 1.2763, "eval_samples_per_second": 783.485, "eval_steps_per_second": 12.536, "step": 43000 }, { "epoch": 2.67, "learning_rate": 0.0005755435263170498, "loss": 0.5287, "step": 43500 }, { "epoch": 2.7, "learning_rate": 0.0005747597820580717, "loss": 0.5305, "step": 44000 }, { "epoch": 2.7, "eval_loss": 0.8569635152816772, "eval_runtime": 1.2713, "eval_samples_per_second": 786.624, "eval_steps_per_second": 12.586, "step": 44000 }, { "epoch": 2.73, "learning_rate": 0.000573964237632326, "loss": 0.527, "step": 44500 }, { "epoch": 2.76, "learning_rate": 0.0005731569278395029, "loss": 0.5262, "step": 45000 }, { "epoch": 2.76, "eval_loss": 0.8558768033981323, "eval_runtime": 1.2738, "eval_samples_per_second": 785.051, "eval_steps_per_second": 12.561, "step": 45000 }, { "epoch": 2.79, "learning_rate": 0.0005723378879939481, "loss": 0.5254, "step": 45500 }, { "epoch": 2.82, "learning_rate": 0.0005715071539231178, "loss": 0.5247, "step": 46000 }, { "epoch": 2.82, "eval_loss": 0.8633874654769897, "eval_runtime": 1.2747, "eval_samples_per_second": 784.504, "eval_steps_per_second": 12.552, "step": 46000 }, { "epoch": 2.85, "learning_rate": 0.0005706647619660116, "loss": 0.5243, "step": 46500 }, { "epoch": 2.88, "learning_rate": 0.0005698107489715823, "loss": 0.5235, "step": 47000 }, { "epoch": 2.88, "eval_loss": 0.8606237769126892, "eval_runtime": 1.3283, "eval_samples_per_second": 752.838, "eval_steps_per_second": 12.045, "step": 47000 }, { "epoch": 2.91, "learning_rate": 0.0005689451522971252, "loss": 0.5228, "step": 47500 }, { "epoch": 2.94, "learning_rate": 0.0005680680098066429, "loss": 0.5227, "step": 48000 }, { "epoch": 2.94, "eval_loss": 0.8610469698905945, "eval_runtime": 1.2783, "eval_samples_per_second": 782.274, "eval_steps_per_second": 12.516, "step": 48000 }, { "epoch": 2.97, "learning_rate": 0.0005671793598691895, "loss": 0.5215, "step": 48500 }, { "epoch": 3.0, "learning_rate": 0.0005662792413571921, "loss": 0.5206, "step": 49000 }, { "epoch": 3.0, "eval_loss": 0.8610268235206604, "eval_runtime": 1.3555, "eval_samples_per_second": 737.72, "eval_steps_per_second": 11.804, "step": 49000 }, { "epoch": 3.03, "learning_rate": 0.0005653676936447504, "loss": 0.5201, "step": 49500 }, { "epoch": 3.06, "learning_rate": 0.0005644447566059142, "loss": 0.5194, "step": 50000 }, { "epoch": 3.06, "eval_loss": 0.8611247539520264, "eval_runtime": 1.2902, "eval_samples_per_second": 775.099, "eval_steps_per_second": 12.402, "step": 50000 }, { "epoch": 3.09, "learning_rate": 0.0005635104706129397, "loss": 0.5189, "step": 50500 }, { "epoch": 3.12, "learning_rate": 0.0005625648765345228, "loss": 0.5183, "step": 51000 }, { "epoch": 3.12, "eval_loss": 0.8579334616661072, "eval_runtime": 1.3546, "eval_samples_per_second": 738.246, "eval_steps_per_second": 11.812, "step": 51000 }, { "epoch": 3.16, "learning_rate": 0.0005616080157340118, "loss": 0.5178, "step": 51500 }, { "epoch": 3.19, "learning_rate": 0.0005606399300675978, "loss": 0.5175, "step": 52000 }, { "epoch": 3.19, "eval_loss": 0.8597957491874695, "eval_runtime": 1.3415, "eval_samples_per_second": 745.449, "eval_steps_per_second": 11.927, "step": 52000 }, { "epoch": 3.22, "learning_rate": 0.0005596606618824843, "loss": 0.5186, "step": 52500 }, { "epoch": 3.25, "learning_rate": 0.0005586702540150338, "loss": 0.5163, "step": 53000 }, { "epoch": 3.25, "eval_loss": 0.8520950078964233, "eval_runtime": 1.3505, "eval_samples_per_second": 740.484, "eval_steps_per_second": 11.848, "step": 53000 }, { "epoch": 3.28, "learning_rate": 0.000557668749788895, "loss": 0.5159, "step": 53500 }, { "epoch": 3.31, "learning_rate": 0.0005566561930131072, "loss": 0.5156, "step": 54000 }, { "epoch": 3.31, "eval_loss": 0.8550420999526978, "eval_runtime": 1.3983, "eval_samples_per_second": 715.143, "eval_steps_per_second": 11.442, "step": 54000 }, { "epoch": 3.34, "learning_rate": 0.000555632627980184, "loss": 0.5153, "step": 54500 }, { "epoch": 3.37, "learning_rate": 0.0005545980994641758, "loss": 0.5148, "step": 55000 }, { "epoch": 3.37, "eval_loss": 0.8503537178039551, "eval_runtime": 1.2794, "eval_samples_per_second": 781.636, "eval_steps_per_second": 12.506, "step": 55000 }, { "epoch": 3.4, "learning_rate": 0.0005535526527187115, "loss": 0.5144, "step": 55500 }, { "epoch": 3.43, "learning_rate": 0.0005524963334750183, "loss": 0.5139, "step": 56000 }, { "epoch": 3.43, "eval_loss": 0.8530069589614868, "eval_runtime": 1.2989, "eval_samples_per_second": 769.886, "eval_steps_per_second": 12.318, "step": 56000 }, { "epoch": 3.46, "learning_rate": 0.0005514291879399219, "loss": 0.5136, "step": 56500 }, { "epoch": 3.49, "learning_rate": 0.000550351262793825, "loss": 0.5133, "step": 57000 }, { "epoch": 3.49, "eval_loss": 0.8588612675666809, "eval_runtime": 1.2931, "eval_samples_per_second": 773.351, "eval_steps_per_second": 12.374, "step": 57000 }, { "epoch": 3.52, "learning_rate": 0.0005492626051886659, "loss": 0.5129, "step": 57500 }, { "epoch": 3.55, "learning_rate": 0.0005481632627458546, "loss": 0.5126, "step": 58000 }, { "epoch": 3.55, "eval_loss": 0.8561201095581055, "eval_runtime": 1.244, "eval_samples_per_second": 803.841, "eval_steps_per_second": 12.861, "step": 58000 }, { "epoch": 3.58, "learning_rate": 0.0005470532835541911, "loss": 0.5122, "step": 58500 }, { "epoch": 3.62, "learning_rate": 0.000545932716167761, "loss": 0.5119, "step": 59000 }, { "epoch": 3.62, "eval_loss": 0.8574157953262329, "eval_runtime": 1.3446, "eval_samples_per_second": 743.721, "eval_steps_per_second": 11.9, "step": 59000 }, { "epoch": 3.65, "learning_rate": 0.000544801609603812, "loss": 0.5115, "step": 59500 }, { "epoch": 3.68, "learning_rate": 0.0005436600133406095, "loss": 0.5127, "step": 60000 }, { "epoch": 3.68, "eval_loss": 0.8624024987220764, "eval_runtime": 1.324, "eval_samples_per_second": 755.286, "eval_steps_per_second": 12.085, "step": 60000 }, { "epoch": 3.71, "learning_rate": 0.0005425079773152721, "loss": 0.5113, "step": 60500 }, { "epoch": 3.74, "learning_rate": 0.0005413455519215879, "loss": 0.5105, "step": 61000 }, { "epoch": 3.74, "eval_loss": 0.8522316813468933, "eval_runtime": 1.2934, "eval_samples_per_second": 773.185, "eval_steps_per_second": 12.371, "step": 61000 }, { "epoch": 3.77, "learning_rate": 0.0005401727880078093, "loss": 0.5102, "step": 61500 }, { "epoch": 3.8, "learning_rate": 0.0005389897368744289, "loss": 0.5099, "step": 62000 }, { "epoch": 3.8, "eval_loss": 0.8549780249595642, "eval_runtime": 1.3356, "eval_samples_per_second": 748.732, "eval_steps_per_second": 11.98, "step": 62000 }, { "epoch": 3.83, "learning_rate": 0.0005377964502719361, "loss": 0.5099, "step": 62500 }, { "epoch": 3.86, "learning_rate": 0.0005365929803985524, "loss": 0.5094, "step": 63000 }, { "epoch": 3.86, "eval_loss": 0.8536927700042725, "eval_runtime": 1.2799, "eval_samples_per_second": 781.291, "eval_steps_per_second": 12.501, "step": 63000 }, { "epoch": 3.89, "learning_rate": 0.0005353793798979489, "loss": 0.5093, "step": 63500 }, { "epoch": 3.92, "learning_rate": 0.000534155701856943, "loss": 0.509, "step": 64000 }, { "epoch": 3.92, "eval_loss": 0.8535122871398926, "eval_runtime": 1.2782, "eval_samples_per_second": 782.329, "eval_steps_per_second": 12.517, "step": 64000 }, { "epoch": 3.95, "learning_rate": 0.0005329219998031763, "loss": 0.5087, "step": 64500 }, { "epoch": 3.98, "learning_rate": 0.0005316783277027734, "loss": 0.5091, "step": 65000 }, { "epoch": 3.98, "eval_loss": 0.8591586351394653, "eval_runtime": 1.3143, "eval_samples_per_second": 760.882, "eval_steps_per_second": 12.174, "step": 65000 }, { "epoch": 4.01, "learning_rate": 0.0005304247399579808, "loss": 0.508, "step": 65500 }, { "epoch": 4.04, "learning_rate": 0.0005291612914047876, "loss": 0.5079, "step": 66000 }, { "epoch": 4.04, "eval_loss": 0.8554427027702332, "eval_runtime": 1.3483, "eval_samples_per_second": 741.678, "eval_steps_per_second": 11.867, "step": 66000 }, { "epoch": 4.07, "learning_rate": 0.0005278880373105263, "loss": 0.5077, "step": 66500 }, { "epoch": 4.11, "learning_rate": 0.0005266050333714561, "loss": 0.5074, "step": 67000 }, { "epoch": 4.11, "eval_loss": 0.8515585660934448, "eval_runtime": 1.2858, "eval_samples_per_second": 777.712, "eval_steps_per_second": 12.443, "step": 67000 }, { "epoch": 4.14, "learning_rate": 0.0005253123357103253, "loss": 0.5074, "step": 67500 }, { "epoch": 4.17, "learning_rate": 0.0005240100008739177, "loss": 0.5069, "step": 68000 }, { "epoch": 4.17, "eval_loss": 0.8490995168685913, "eval_runtime": 1.2974, "eval_samples_per_second": 770.751, "eval_steps_per_second": 12.332, "step": 68000 }, { "epoch": 4.2, "learning_rate": 0.0005226980858305778, "loss": 0.5067, "step": 68500 }, { "epoch": 4.23, "learning_rate": 0.0005213766479677197, "loss": 0.5066, "step": 69000 }, { "epoch": 4.23, "eval_loss": 0.8570588231086731, "eval_runtime": 1.2957, "eval_samples_per_second": 771.773, "eval_steps_per_second": 12.348, "step": 69000 }, { "epoch": 4.26, "learning_rate": 0.0005200457450893163, "loss": 0.5063, "step": 69500 }, { "epoch": 4.29, "learning_rate": 0.0005187054354133712, "loss": 0.5068, "step": 70000 }, { "epoch": 4.29, "eval_loss": 0.8535866141319275, "eval_runtime": 1.3217, "eval_samples_per_second": 756.603, "eval_steps_per_second": 12.106, "step": 70000 }, { "epoch": 4.32, "learning_rate": 0.0005173557775693715, "loss": 0.5058, "step": 70500 }, { "epoch": 4.35, "learning_rate": 0.0005159968305957235, "loss": 0.5066, "step": 71000 }, { "epoch": 4.35, "eval_loss": 0.9288005232810974, "eval_runtime": 1.2769, "eval_samples_per_second": 783.148, "eval_steps_per_second": 12.53, "step": 71000 }, { "epoch": 4.38, "learning_rate": 0.0005146286539371703, "loss": 0.5084, "step": 71500 }, { "epoch": 4.41, "learning_rate": 0.0005132513074421913, "loss": 0.5051, "step": 72000 }, { "epoch": 4.41, "eval_loss": 0.8597245812416077, "eval_runtime": 1.3157, "eval_samples_per_second": 760.048, "eval_steps_per_second": 12.161, "step": 72000 }, { "epoch": 4.44, "learning_rate": 0.0005118648513603841, "loss": 0.5051, "step": 72500 }, { "epoch": 4.47, "learning_rate": 0.0005104693463398293, "loss": 0.5045, "step": 73000 }, { "epoch": 4.47, "eval_loss": 0.8554754257202148, "eval_runtime": 1.3269, "eval_samples_per_second": 753.64, "eval_steps_per_second": 12.058, "step": 73000 }, { "epoch": 4.5, "learning_rate": 0.0005090648534244371, "loss": 0.5045, "step": 73500 }, { "epoch": 4.53, "learning_rate": 0.0005076514340512776, "loss": 0.5043, "step": 74000 }, { "epoch": 4.53, "eval_loss": 0.8547331094741821, "eval_runtime": 1.2736, "eval_samples_per_second": 785.148, "eval_steps_per_second": 12.562, "step": 74000 }, { "epoch": 4.56, "learning_rate": 0.0005062291500478931, "loss": 0.5039, "step": 74500 }, { "epoch": 4.6, "learning_rate": 0.0005047980636295937, "loss": 0.5039, "step": 75000 }, { "epoch": 4.6, "eval_loss": 0.8560617566108704, "eval_runtime": 1.2429, "eval_samples_per_second": 804.574, "eval_steps_per_second": 12.873, "step": 75000 }, { "epoch": 4.63, "learning_rate": 0.0005033582373967348, "loss": 0.5036, "step": 75500 }, { "epoch": 4.66, "learning_rate": 0.0005019097343319809, "loss": 0.504, "step": 76000 }, { "epoch": 4.66, "eval_loss": 0.854142427444458, "eval_runtime": 1.3399, "eval_samples_per_second": 746.336, "eval_steps_per_second": 11.941, "step": 76000 }, { "epoch": 4.69, "learning_rate": 0.0005004526177975481, "loss": 0.503, "step": 76500 }, { "epoch": 4.72, "learning_rate": 0.0004989869515324342, "loss": 0.5026, "step": 77000 }, { "epoch": 4.72, "eval_loss": 0.8489631414413452, "eval_runtime": 1.2942, "eval_samples_per_second": 772.659, "eval_steps_per_second": 12.363, "step": 77000 }, { "epoch": 4.75, "learning_rate": 0.0004975127996496297, "loss": 0.5024, "step": 77500 }, { "epoch": 4.78, "learning_rate": 0.0004960302266333135, "loss": 0.5024, "step": 78000 }, { "epoch": 4.78, "eval_loss": 0.8499003648757935, "eval_runtime": 1.3145, "eval_samples_per_second": 760.727, "eval_steps_per_second": 12.172, "step": 78000 }, { "epoch": 4.81, "learning_rate": 0.0004945392973360323, "loss": 0.502, "step": 78500 }, { "epoch": 4.84, "learning_rate": 0.0004930400769758634, "loss": 0.5019, "step": 79000 }, { "epoch": 4.84, "eval_loss": 0.8521845936775208, "eval_runtime": 1.3201, "eval_samples_per_second": 757.527, "eval_steps_per_second": 12.12, "step": 79000 }, { "epoch": 4.87, "learning_rate": 0.0004915326311335622, "loss": 0.5017, "step": 79500 }, { "epoch": 4.9, "learning_rate": 0.0004900170257496933, "loss": 0.5014, "step": 80000 }, { "epoch": 4.9, "eval_loss": 0.8508275747299194, "eval_runtime": 1.3001, "eval_samples_per_second": 769.194, "eval_steps_per_second": 12.307, "step": 80000 }, { "epoch": 4.93, "learning_rate": 0.0004884933271217461, "loss": 0.501, "step": 80500 }, { "epoch": 4.96, "learning_rate": 0.0004869616019012347, "loss": 0.5008, "step": 81000 }, { "epoch": 4.96, "eval_loss": 0.8512425422668457, "eval_runtime": 1.2521, "eval_samples_per_second": 798.677, "eval_steps_per_second": 12.779, "step": 81000 }, { "epoch": 4.99, "learning_rate": 0.00048542191709078226, "loss": 0.5004, "step": 81500 }, { "epoch": 5.02, "learning_rate": 0.00048387434004119027, "loss": 0.5002, "step": 82000 }, { "epoch": 5.02, "eval_loss": 0.8470181822776794, "eval_runtime": 1.2894, "eval_samples_per_second": 775.545, "eval_steps_per_second": 12.409, "step": 82000 }, { "epoch": 5.06, "learning_rate": 0.0004823189384484924, "loss": 0.5007, "step": 82500 }, { "epoch": 5.09, "learning_rate": 0.000480755780350993, "loss": 0.4995, "step": 83000 }, { "epoch": 5.09, "eval_loss": 0.846192479133606, "eval_runtime": 1.2808, "eval_samples_per_second": 780.79, "eval_steps_per_second": 12.493, "step": 83000 }, { "epoch": 5.12, "learning_rate": 0.0004791849341262914, "loss": 0.4993, "step": 83500 }, { "epoch": 5.15, "learning_rate": 0.0004776064684882901, "loss": 0.4991, "step": 84000 }, { "epoch": 5.15, "eval_loss": 0.8454980254173279, "eval_runtime": 1.3356, "eval_samples_per_second": 748.729, "eval_steps_per_second": 11.98, "step": 84000 }, { "epoch": 5.18, "learning_rate": 0.0004760204524841897, "loss": 0.4985, "step": 84500 }, { "epoch": 5.21, "learning_rate": 0.0004744269554914683, "loss": 0.4982, "step": 85000 }, { "epoch": 5.21, "eval_loss": 0.8464594483375549, "eval_runtime": 1.368, "eval_samples_per_second": 730.977, "eval_steps_per_second": 11.696, "step": 85000 }, { "epoch": 5.24, "learning_rate": 0.00047282604721484643, "loss": 0.4979, "step": 85500 }, { "epoch": 5.27, "learning_rate": 0.0004712177976832385, "loss": 0.4978, "step": 86000 }, { "epoch": 5.27, "eval_loss": 0.8433752655982971, "eval_runtime": 1.3279, "eval_samples_per_second": 753.055, "eval_steps_per_second": 12.049, "step": 86000 }, { "epoch": 5.3, "learning_rate": 0.0004696022772466888, "loss": 0.4974, "step": 86500 }, { "epoch": 5.33, "learning_rate": 0.00046797955657329487, "loss": 0.4969, "step": 87000 }, { "epoch": 5.33, "eval_loss": 0.843237042427063, "eval_runtime": 1.3349, "eval_samples_per_second": 749.108, "eval_steps_per_second": 11.986, "step": 87000 }, { "epoch": 5.36, "learning_rate": 0.000466349706646116, "loss": 0.4967, "step": 87500 }, { "epoch": 5.39, "learning_rate": 0.00046471279876006763, "loss": 0.4964, "step": 88000 }, { "epoch": 5.39, "eval_loss": 0.8417282104492188, "eval_runtime": 1.3291, "eval_samples_per_second": 752.394, "eval_steps_per_second": 12.038, "step": 88000 }, { "epoch": 5.42, "learning_rate": 0.00046306890451880395, "loss": 0.4964, "step": 88500 }, { "epoch": 5.45, "learning_rate": 0.0004614180958315844, "loss": 0.4957, "step": 89000 }, { "epoch": 5.45, "eval_loss": 0.8362923860549927, "eval_runtime": 1.2708, "eval_samples_per_second": 786.921, "eval_steps_per_second": 12.591, "step": 89000 }, { "epoch": 5.48, "learning_rate": 0.00045976044491012884, "loss": 0.4954, "step": 89500 }, { "epoch": 5.51, "learning_rate": 0.00045809602426545847, "loss": 0.495, "step": 90000 }, { "epoch": 5.51, "eval_loss": 0.8392152190208435, "eval_runtime": 1.26, "eval_samples_per_second": 793.671, "eval_steps_per_second": 12.699, "step": 90000 }, { "epoch": 5.55, "learning_rate": 0.00045642490670472436, "loss": 0.4947, "step": 90500 }, { "epoch": 5.58, "learning_rate": 0.0004547471653280225, "loss": 0.4946, "step": 91000 }, { "epoch": 5.58, "eval_loss": 0.8400516510009766, "eval_runtime": 1.2358, "eval_samples_per_second": 809.182, "eval_steps_per_second": 12.947, "step": 91000 }, { "epoch": 5.61, "learning_rate": 0.00045306287352519543, "loss": 0.4939, "step": 91500 }, { "epoch": 5.64, "learning_rate": 0.00045137210497262333, "loss": 0.4935, "step": 92000 }, { "epoch": 5.64, "eval_loss": 0.8373098969459534, "eval_runtime": 1.3142, "eval_samples_per_second": 760.929, "eval_steps_per_second": 12.175, "step": 92000 }, { "epoch": 5.67, "learning_rate": 0.0004496749336299999, "loss": 0.4931, "step": 92500 }, { "epoch": 5.7, "learning_rate": 0.0004479714337370977, "loss": 0.4929, "step": 93000 }, { "epoch": 5.7, "eval_loss": 0.840123176574707, "eval_runtime": 1.2896, "eval_samples_per_second": 775.451, "eval_steps_per_second": 12.407, "step": 93000 }, { "epoch": 5.73, "learning_rate": 0.00044626167981052036, "loss": 0.4924, "step": 93500 }, { "epoch": 5.76, "learning_rate": 0.00044454574664044404, "loss": 0.492, "step": 94000 }, { "epoch": 5.76, "eval_loss": 0.8355880379676819, "eval_runtime": 1.2671, "eval_samples_per_second": 789.19, "eval_steps_per_second": 12.627, "step": 94000 }, { "epoch": 5.79, "learning_rate": 0.000442823709287344, "loss": 0.4916, "step": 94500 }, { "epoch": 5.82, "learning_rate": 0.0004410956430787129, "loss": 0.4912, "step": 95000 }, { "epoch": 5.82, "eval_loss": 0.8333644866943359, "eval_runtime": 1.299, "eval_samples_per_second": 769.834, "eval_steps_per_second": 12.317, "step": 95000 }, { "epoch": 5.85, "learning_rate": 0.0004393616236057647, "loss": 0.4912, "step": 95500 }, { "epoch": 5.88, "learning_rate": 0.00043762172672012875, "loss": 0.4904, "step": 96000 }, { "epoch": 5.88, "eval_loss": 0.8280515074729919, "eval_runtime": 1.2803, "eval_samples_per_second": 781.038, "eval_steps_per_second": 12.497, "step": 96000 }, { "epoch": 5.91, "learning_rate": 0.0004358760285305312, "loss": 0.4901, "step": 96500 }, { "epoch": 5.94, "learning_rate": 0.0004341246053994663, "loss": 0.4898, "step": 97000 }, { "epoch": 5.94, "eval_loss": 0.8338386416435242, "eval_runtime": 1.3216, "eval_samples_per_second": 756.685, "eval_steps_per_second": 12.107, "step": 97000 }, { "epoch": 5.97, "learning_rate": 0.00043236753393985534, "loss": 0.4892, "step": 97500 }, { "epoch": 6.0, "learning_rate": 0.0004306048910116964, "loss": 0.4891, "step": 98000 }, { "epoch": 6.0, "eval_loss": 0.8300430774688721, "eval_runtime": 1.316, "eval_samples_per_second": 759.879, "eval_steps_per_second": 12.158, "step": 98000 }, { "epoch": 6.04, "learning_rate": 0.0004288367537187012, "loss": 0.4887, "step": 98500 }, { "epoch": 6.07, "learning_rate": 0.00042706319940492284, "loss": 0.4882, "step": 99000 }, { "epoch": 6.07, "eval_loss": 0.8262238502502441, "eval_runtime": 1.3745, "eval_samples_per_second": 727.548, "eval_steps_per_second": 11.641, "step": 99000 }, { "epoch": 6.1, "learning_rate": 0.00042528430565137254, "loss": 0.488, "step": 99500 }, { "epoch": 6.13, "learning_rate": 0.00042350015027262593, "loss": 0.4876, "step": 100000 }, { "epoch": 6.13, "eval_loss": 0.8171582221984863, "eval_runtime": 1.3024, "eval_samples_per_second": 767.786, "eval_steps_per_second": 12.285, "step": 100000 }, { "epoch": 6.16, "learning_rate": 0.00042171081131341917, "loss": 0.4877, "step": 100500 }, { "epoch": 6.19, "learning_rate": 0.00041991636704523497, "loss": 0.4868, "step": 101000 }, { "epoch": 6.19, "eval_loss": 0.8239555954933167, "eval_runtime": 1.2567, "eval_samples_per_second": 795.735, "eval_steps_per_second": 12.732, "step": 101000 }, { "epoch": 6.22, "learning_rate": 0.00041811689596287893, "loss": 0.4864, "step": 101500 }, { "epoch": 6.25, "learning_rate": 0.0004163124767810454, "loss": 0.4861, "step": 102000 }, { "epoch": 6.25, "eval_loss": 0.8212010264396667, "eval_runtime": 1.2247, "eval_samples_per_second": 816.535, "eval_steps_per_second": 13.065, "step": 102000 }, { "epoch": 6.28, "learning_rate": 0.00041450318843087506, "loss": 0.4858, "step": 102500 }, { "epoch": 6.31, "learning_rate": 0.00041268911005650166, "loss": 0.4854, "step": 103000 }, { "epoch": 6.31, "eval_loss": 0.8242572546005249, "eval_runtime": 1.357, "eval_samples_per_second": 736.942, "eval_steps_per_second": 11.791, "step": 103000 }, { "epoch": 6.34, "learning_rate": 0.00041087032101159006, "loss": 0.4851, "step": 103500 }, { "epoch": 6.37, "learning_rate": 0.00040904690085586515, "loss": 0.4847, "step": 104000 }, { "epoch": 6.37, "eval_loss": 0.8227641582489014, "eval_runtime": 1.3326, "eval_samples_per_second": 750.437, "eval_steps_per_second": 12.007, "step": 104000 }, { "epoch": 6.4, "learning_rate": 0.0004072189293516316, "loss": 0.4844, "step": 104500 }, { "epoch": 6.43, "learning_rate": 0.0004053864864602847, "loss": 0.4841, "step": 105000 }, { "epoch": 6.43, "eval_loss": 0.8184976577758789, "eval_runtime": 1.3395, "eval_samples_per_second": 746.538, "eval_steps_per_second": 11.945, "step": 105000 }, { "epoch": 6.46, "learning_rate": 0.00040354965233881297, "loss": 0.4846, "step": 105500 }, { "epoch": 6.5, "learning_rate": 0.0004017085073362913, "loss": 0.4837, "step": 106000 }, { "epoch": 6.5, "eval_loss": 0.8177208304405212, "eval_runtime": 1.2838, "eval_samples_per_second": 778.932, "eval_steps_per_second": 12.463, "step": 106000 }, { "epoch": 6.53, "learning_rate": 0.00039986313199036664, "loss": 0.4828, "step": 106500 }, { "epoch": 6.56, "learning_rate": 0.00039801360702373484, "loss": 0.4827, "step": 107000 }, { "epoch": 6.56, "eval_loss": 0.8140051364898682, "eval_runtime": 1.3554, "eval_samples_per_second": 737.763, "eval_steps_per_second": 11.804, "step": 107000 }, { "epoch": 6.59, "learning_rate": 0.00039616001334060954, "loss": 0.4824, "step": 107500 }, { "epoch": 6.62, "learning_rate": 0.00039430243202318314, "loss": 0.4819, "step": 108000 }, { "epoch": 6.62, "eval_loss": 0.8147432208061218, "eval_runtime": 1.3502, "eval_samples_per_second": 740.632, "eval_steps_per_second": 11.85, "step": 108000 }, { "epoch": 6.65, "learning_rate": 0.00039244094432808034, "loss": 0.4816, "step": 108500 }, { "epoch": 6.68, "learning_rate": 0.0003905756316828033, "loss": 0.4813, "step": 109000 }, { "epoch": 6.68, "eval_loss": 0.8172094225883484, "eval_runtime": 1.4168, "eval_samples_per_second": 705.811, "eval_steps_per_second": 11.293, "step": 109000 }, { "epoch": 6.71, "learning_rate": 0.00038870657568216963, "loss": 0.4808, "step": 109500 }, { "epoch": 6.74, "learning_rate": 0.00038683385808474416, "loss": 0.4807, "step": 110000 }, { "epoch": 6.74, "eval_loss": 0.8148666024208069, "eval_runtime": 1.2865, "eval_samples_per_second": 777.303, "eval_steps_per_second": 12.437, "step": 110000 } ], "max_steps": 250000, "num_train_epochs": 16, "total_flos": 1.7617879685021544e+21, "trial_name": null, "trial_params": null }