| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 26.0, | |
| "global_step": 98228, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5e-09, | |
| "loss": 10.4629, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.5e-06, | |
| "loss": 9.562, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5e-06, | |
| "loss": 7.9982, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.5e-06, | |
| "loss": 6.9798, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1e-05, | |
| "loss": 6.6752, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.25e-05, | |
| "loss": 6.5045, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.5e-05, | |
| "loss": 6.3892, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.75e-05, | |
| "loss": 6.2943, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 2e-05, | |
| "loss": 6.2221, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 2.25e-05, | |
| "loss": 6.154, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.5e-05, | |
| "loss": 6.0982, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.7500000000000004e-05, | |
| "loss": 6.0464, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3e-05, | |
| "loss": 6.0017, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.2500000000000004e-05, | |
| "loss": 5.9683, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.5e-05, | |
| "loss": 5.936, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 5.9005, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4e-05, | |
| "loss": 5.8763, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.2495e-05, | |
| "loss": 5.8522, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.4995000000000005e-05, | |
| "loss": 5.8356, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.7495e-05, | |
| "loss": 5.8125, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.9995000000000005e-05, | |
| "loss": 5.7984, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.998341326938449e-05, | |
| "loss": 5.7788, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.996675992539302e-05, | |
| "loss": 5.7609, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.995010658140154e-05, | |
| "loss": 5.7509, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 4.993345323741007e-05, | |
| "loss": 5.7398, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 4.991683320010658e-05, | |
| "loss": 5.7264, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 4.990017985611511e-05, | |
| "loss": 5.7196, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 4.988352651212364e-05, | |
| "loss": 5.7094, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 4.986687316813216e-05, | |
| "loss": 5.6992, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 4.9850253130828675e-05, | |
| "loss": 5.6917, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 4.9833599786837196e-05, | |
| "loss": 5.6858, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.9816946442845725e-05, | |
| "loss": 5.6734, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 4.980029309885425e-05, | |
| "loss": 5.6687, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 4.978367306155076e-05, | |
| "loss": 5.6662, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 4.976701971755929e-05, | |
| "loss": 5.6598, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 4.975036637356781e-05, | |
| "loss": 5.655, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 4.973374633626433e-05, | |
| "loss": 5.6471, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 4.971709299227285e-05, | |
| "loss": 5.6424, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 4.970043964828138e-05, | |
| "loss": 5.6371, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 4.96837863042899e-05, | |
| "loss": 5.6289, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 4.966713296029843e-05, | |
| "loss": 5.6248, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 4.965047961630696e-05, | |
| "loss": 5.622, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 4.963382627231548e-05, | |
| "loss": 5.6202, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 4.9617206235011995e-05, | |
| "loss": 5.6134, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 4.9600552891020516e-05, | |
| "loss": 5.6098, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 4.9583899547029045e-05, | |
| "loss": 5.6084, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 4.956724620303757e-05, | |
| "loss": 5.6018, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 4.9550592859046095e-05, | |
| "loss": 5.5955, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 4.953397282174261e-05, | |
| "loss": 5.5957, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 4.951731947775113e-05, | |
| "loss": 5.5875, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 4.950066613375966e-05, | |
| "loss": 5.5857, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 4.948401278976819e-05, | |
| "loss": 5.5856, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 4.94673927524647e-05, | |
| "loss": 5.5855, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 4.945073940847322e-05, | |
| "loss": 5.579, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 4.943408606448175e-05, | |
| "loss": 5.5772, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 4.941743272049028e-05, | |
| "loss": 5.5725, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 4.94007793764988e-05, | |
| "loss": 5.57, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 4.9384159339195315e-05, | |
| "loss": 5.5676, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 4.9367505995203836e-05, | |
| "loss": 5.5665, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 4.9350852651212365e-05, | |
| "loss": 5.5615, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 4.933419930722089e-05, | |
| "loss": 5.5169, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 4.9317545963229415e-05, | |
| "loss": 5.2484, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 4.930089261923794e-05, | |
| "loss": 5.043, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 4.928423927524647e-05, | |
| "loss": 4.8676, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 4.9267585931255e-05, | |
| "loss": 4.6987, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 8.6, | |
| "learning_rate": 4.925096589395151e-05, | |
| "loss": 4.5359, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 4.923434585664802e-05, | |
| "loss": 4.3603, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 4.921769251265655e-05, | |
| "loss": 4.1923, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 4.920103916866507e-05, | |
| "loss": 4.0239, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 4.91843858246736e-05, | |
| "loss": 3.7255, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 4.916773248068212e-05, | |
| "loss": 3.2457, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 4.9151112443378634e-05, | |
| "loss": 2.5753, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 4.9134459099387156e-05, | |
| "loss": 2.2821, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 4.9117805755395685e-05, | |
| "loss": 2.1267, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 4.910115241140421e-05, | |
| "loss": 2.0179, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 4.908453237410072e-05, | |
| "loss": 1.9368, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 4.906787903010925e-05, | |
| "loss": 1.8691, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 10.19, | |
| "learning_rate": 4.905125899280576e-05, | |
| "loss": 1.816, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 10.32, | |
| "learning_rate": 4.903460564881429e-05, | |
| "loss": 1.7673, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 10.46, | |
| "learning_rate": 4.901795230482281e-05, | |
| "loss": 1.7288, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 10.59, | |
| "learning_rate": 4.900129896083134e-05, | |
| "loss": 1.6907, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "learning_rate": 4.898464561683987e-05, | |
| "loss": 1.6572, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 4.896799227284839e-05, | |
| "loss": 1.6261, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 4.895133892885692e-05, | |
| "loss": 1.6001, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 4.893468558486544e-05, | |
| "loss": 1.5719, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 11.25, | |
| "learning_rate": 4.891803224087397e-05, | |
| "loss": 1.5495, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 11.38, | |
| "learning_rate": 4.8901412203570476e-05, | |
| "loss": 1.5277, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 11.51, | |
| "learning_rate": 4.8884758859579004e-05, | |
| "loss": 1.5069, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 11.65, | |
| "learning_rate": 4.886810551558753e-05, | |
| "loss": 1.4871, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 11.78, | |
| "learning_rate": 4.8851452171596055e-05, | |
| "loss": 1.4707, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 11.91, | |
| "learning_rate": 4.883479882760458e-05, | |
| "loss": 1.4522, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 4.881817879030109e-05, | |
| "loss": 1.4362, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 12.18, | |
| "learning_rate": 4.880152544630962e-05, | |
| "loss": 1.4219, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 12.31, | |
| "learning_rate": 4.878487210231815e-05, | |
| "loss": 1.4066, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "learning_rate": 4.876821875832667e-05, | |
| "loss": 1.3935, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 12.57, | |
| "learning_rate": 4.875159872102319e-05, | |
| "loss": 1.3809, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "learning_rate": 4.873494537703171e-05, | |
| "loss": 1.3663, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 12.84, | |
| "learning_rate": 4.871829203304024e-05, | |
| "loss": 1.3571, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 4.8701671995736746e-05, | |
| "loss": 1.3454, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 13.1, | |
| "learning_rate": 4.8685018651745274e-05, | |
| "loss": 1.3336, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 13.23, | |
| "learning_rate": 4.8668365307753796e-05, | |
| "loss": 1.3236, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 13.37, | |
| "learning_rate": 4.8651711963762324e-05, | |
| "loss": 1.3131, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 13.5, | |
| "learning_rate": 4.863505861977085e-05, | |
| "loss": 1.3029, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 4.8618405275779374e-05, | |
| "loss": 1.2961, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 13.76, | |
| "learning_rate": 4.86017519317879e-05, | |
| "loss": 1.2889, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 13.9, | |
| "learning_rate": 4.858513189448441e-05, | |
| "loss": 1.2814, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 4.856847855049294e-05, | |
| "loss": 1.2704, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 14.16, | |
| "learning_rate": 4.855182520650147e-05, | |
| "loss": 1.2605, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 14.29, | |
| "learning_rate": 4.853517186250999e-05, | |
| "loss": 1.2545, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 14.43, | |
| "learning_rate": 4.851851851851852e-05, | |
| "loss": 1.2487, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 14.56, | |
| "learning_rate": 4.850186517452705e-05, | |
| "loss": 1.2381, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 14.69, | |
| "learning_rate": 4.8485211830535576e-05, | |
| "loss": 1.2326, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 14.82, | |
| "learning_rate": 4.8468558486544104e-05, | |
| "loss": 1.224, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 14.96, | |
| "learning_rate": 4.845193844924061e-05, | |
| "loss": 1.2213, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 15.09, | |
| "learning_rate": 4.8435285105249135e-05, | |
| "loss": 1.2126, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 15.22, | |
| "learning_rate": 4.841863176125766e-05, | |
| "loss": 1.2047, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 15.35, | |
| "learning_rate": 4.840197841726619e-05, | |
| "loss": 1.1999, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 15.48, | |
| "learning_rate": 4.8385358379962694e-05, | |
| "loss": 1.1967, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 15.62, | |
| "learning_rate": 4.836870503597122e-05, | |
| "loss": 1.1901, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 15.75, | |
| "learning_rate": 4.835205169197975e-05, | |
| "loss": 1.1843, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 15.88, | |
| "learning_rate": 4.833539834798828e-05, | |
| "loss": 1.1792, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 4.831877831068479e-05, | |
| "loss": 1.1752, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 16.15, | |
| "learning_rate": 4.83021582733813e-05, | |
| "loss": 1.1657, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 16.28, | |
| "learning_rate": 4.8285504929389827e-05, | |
| "loss": 1.1628, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 16.41, | |
| "learning_rate": 4.826885158539835e-05, | |
| "loss": 1.1586, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 16.54, | |
| "learning_rate": 4.825219824140688e-05, | |
| "loss": 1.1547, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 16.68, | |
| "learning_rate": 4.8235578204103386e-05, | |
| "loss": 1.1486, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "learning_rate": 4.8218924860111914e-05, | |
| "loss": 1.1439, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 16.94, | |
| "learning_rate": 4.820227151612044e-05, | |
| "loss": 1.1409, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 17.07, | |
| "learning_rate": 4.8185618172128964e-05, | |
| "loss": 1.1349, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 17.2, | |
| "learning_rate": 4.816896482813749e-05, | |
| "loss": 1.1306, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 17.34, | |
| "learning_rate": 4.8152311484146014e-05, | |
| "loss": 1.1287, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 17.47, | |
| "learning_rate": 4.813569144684253e-05, | |
| "loss": 1.1224, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 17.6, | |
| "learning_rate": 4.811903810285105e-05, | |
| "loss": 1.1196, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 17.73, | |
| "learning_rate": 4.810238475885958e-05, | |
| "loss": 1.1162, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 17.87, | |
| "learning_rate": 4.808576472155609e-05, | |
| "loss": 1.1141, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 4.806911137756462e-05, | |
| "loss": 1.1098, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 18.13, | |
| "learning_rate": 4.8052458033573147e-05, | |
| "loss": 1.1043, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 18.26, | |
| "learning_rate": 4.803580468958167e-05, | |
| "loss": 1.1008, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 18.4, | |
| "learning_rate": 4.8019151345590197e-05, | |
| "loss": 1.0991, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 18.53, | |
| "learning_rate": 4.8002531308286706e-05, | |
| "loss": 1.0941, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 18.66, | |
| "learning_rate": 4.7985877964295234e-05, | |
| "loss": 1.0892, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 18.79, | |
| "learning_rate": 4.796922462030376e-05, | |
| "loss": 1.0847, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 18.93, | |
| "learning_rate": 4.7952571276312284e-05, | |
| "loss": 1.083, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 19.06, | |
| "learning_rate": 4.793591793232081e-05, | |
| "loss": 1.0802, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 19.19, | |
| "learning_rate": 4.791926458832934e-05, | |
| "loss": 1.0761, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 19.32, | |
| "learning_rate": 4.790261124433786e-05, | |
| "loss": 1.0726, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 19.45, | |
| "learning_rate": 4.788595790034639e-05, | |
| "loss": 1.0683, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 19.59, | |
| "learning_rate": 4.78693378630429e-05, | |
| "loss": 1.0667, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 19.72, | |
| "learning_rate": 4.785268451905143e-05, | |
| "loss": 1.0655, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 19.85, | |
| "learning_rate": 4.783603117505995e-05, | |
| "loss": 1.0617, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 19.98, | |
| "learning_rate": 4.781937783106848e-05, | |
| "loss": 1.0599, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 20.12, | |
| "learning_rate": 4.780272448707701e-05, | |
| "loss": 1.0554, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 20.25, | |
| "learning_rate": 4.778607114308553e-05, | |
| "loss": 1.0541, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 20.38, | |
| "learning_rate": 4.7769417799094064e-05, | |
| "loss": 1.0482, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 20.51, | |
| "learning_rate": 4.7752764455102586e-05, | |
| "loss": 1.0486, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 20.65, | |
| "learning_rate": 4.7736144417799095e-05, | |
| "loss": 1.0453, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 20.78, | |
| "learning_rate": 4.7719491073807623e-05, | |
| "loss": 1.0427, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 20.91, | |
| "learning_rate": 4.770287103650413e-05, | |
| "loss": 1.0399, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 21.04, | |
| "learning_rate": 4.768621769251266e-05, | |
| "loss": 1.0363, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 21.18, | |
| "learning_rate": 4.766956434852118e-05, | |
| "loss": 1.0321, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 21.31, | |
| "learning_rate": 4.765291100452971e-05, | |
| "loss": 1.0311, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 21.44, | |
| "learning_rate": 4.763629096722622e-05, | |
| "loss": 1.0287, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 21.57, | |
| "learning_rate": 4.761963762323475e-05, | |
| "loss": 1.0263, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 21.7, | |
| "learning_rate": 4.760298427924327e-05, | |
| "loss": 1.0241, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 21.84, | |
| "learning_rate": 4.75863309352518e-05, | |
| "loss": 1.0231, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 21.97, | |
| "learning_rate": 4.756967759126033e-05, | |
| "loss": 1.0223, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 22.1, | |
| "learning_rate": 4.7553057553956836e-05, | |
| "loss": 1.019, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 22.23, | |
| "learning_rate": 4.7536404209965365e-05, | |
| "loss": 1.0148, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 22.37, | |
| "learning_rate": 4.7519750865973887e-05, | |
| "loss": 1.012, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 22.5, | |
| "learning_rate": 4.7503097521982415e-05, | |
| "loss": 1.0117, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 22.63, | |
| "learning_rate": 4.7486477484678924e-05, | |
| "loss": 1.0089, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 22.76, | |
| "learning_rate": 4.746982414068745e-05, | |
| "loss": 1.0068, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 22.9, | |
| "learning_rate": 4.745317079669598e-05, | |
| "loss": 1.0042, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 23.03, | |
| "learning_rate": 4.74365174527045e-05, | |
| "loss": 1.002, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 23.16, | |
| "learning_rate": 4.741989741540102e-05, | |
| "loss": 1.0018, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 23.29, | |
| "learning_rate": 4.740324407140954e-05, | |
| "loss": 0.996, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 23.43, | |
| "learning_rate": 4.738659072741807e-05, | |
| "loss": 0.9964, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 23.56, | |
| "learning_rate": 4.736993738342659e-05, | |
| "loss": 0.9921, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 23.69, | |
| "learning_rate": 4.735328403943512e-05, | |
| "loss": 0.9919, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 23.82, | |
| "learning_rate": 4.733666400213163e-05, | |
| "loss": 0.9894, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 23.95, | |
| "learning_rate": 4.7320010658140156e-05, | |
| "loss": 0.9867, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 24.09, | |
| "learning_rate": 4.7303357314148685e-05, | |
| "loss": 0.9856, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 24.22, | |
| "learning_rate": 4.7286737276845194e-05, | |
| "loss": 0.9821, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 24.35, | |
| "learning_rate": 4.727008393285372e-05, | |
| "loss": 0.983, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 24.48, | |
| "learning_rate": 4.7253430588862244e-05, | |
| "loss": 0.9782, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 24.62, | |
| "learning_rate": 4.723677724487077e-05, | |
| "loss": 0.9789, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 24.75, | |
| "learning_rate": 4.72201239008793e-05, | |
| "loss": 0.9767, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 24.88, | |
| "learning_rate": 4.720347055688782e-05, | |
| "loss": 0.9741, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 25.01, | |
| "learning_rate": 4.718681721289635e-05, | |
| "loss": 0.9723, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 25.15, | |
| "learning_rate": 4.717016386890488e-05, | |
| "loss": 0.9688, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 25.28, | |
| "learning_rate": 4.71535105249134e-05, | |
| "loss": 0.9683, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 25.41, | |
| "learning_rate": 4.713689048760992e-05, | |
| "loss": 0.9683, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 25.54, | |
| "learning_rate": 4.712023714361844e-05, | |
| "loss": 0.9663, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 25.67, | |
| "learning_rate": 4.710358379962697e-05, | |
| "loss": 0.9623, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 25.81, | |
| "learning_rate": 4.708693045563549e-05, | |
| "loss": 0.9614, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 25.94, | |
| "learning_rate": 4.7070310418332005e-05, | |
| "loss": 0.9612, | |
| "step": 98000 | |
| } | |
| ], | |
| "max_steps": 1511200, | |
| "num_train_epochs": 400, | |
| "total_flos": 2.647131420139664e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |