| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.999848249174855, | |
| "global_step": 105432, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9810311468568605e-08, | |
| "loss": 0.6518, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.962062293713722e-08, | |
| "loss": 0.6229, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9430934405705826e-08, | |
| "loss": 0.5878, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.924124587427444e-08, | |
| "loss": 0.5654, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9051557342843047e-08, | |
| "loss": 0.5387, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.886186881141166e-08, | |
| "loss": 0.512, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.867218027998027e-08, | |
| "loss": 0.4914, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.848249174854888e-08, | |
| "loss": 0.4605, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.829280321711749e-08, | |
| "loss": 0.445, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.81031146856861e-08, | |
| "loss": 0.4245, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.791342615425471e-08, | |
| "loss": 0.4061, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.772373762282332e-08, | |
| "loss": 0.3907, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.753404909139193e-08, | |
| "loss": 0.3671, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.7344360559960544e-08, | |
| "loss": 0.3672, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.715467202852915e-08, | |
| "loss": 0.3369, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.6964983497097764e-08, | |
| "loss": 0.33, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.677529496566637e-08, | |
| "loss": 0.3152, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.6585606434234985e-08, | |
| "loss": 0.2996, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.639591790280359e-08, | |
| "loss": 0.2835, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.6206229371372206e-08, | |
| "loss": 0.2747, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.601654083994081e-08, | |
| "loss": 0.2597, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.582685230850942e-08, | |
| "loss": 0.2509, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.5637163777078034e-08, | |
| "loss": 0.2321, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.544747524564664e-08, | |
| "loss": 0.2216, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.5257786714215255e-08, | |
| "loss": 0.2138, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.506809818278386e-08, | |
| "loss": 0.1964, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.4878409651352476e-08, | |
| "loss": 0.183, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.468872111992108e-08, | |
| "loss": 0.1738, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.4499032588489697e-08, | |
| "loss": 0.1628, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.4309344057058304e-08, | |
| "loss": 0.155, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.411965552562692e-08, | |
| "loss": 0.1433, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.3929966994195524e-08, | |
| "loss": 0.1336, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.374027846276414e-08, | |
| "loss": 0.125, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.3550589931332745e-08, | |
| "loss": 0.115, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.336090139990136e-08, | |
| "loss": 0.1095, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.317121286846997e-08, | |
| "loss": 0.1024, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.298152433703858e-08, | |
| "loss": 0.0947, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.2791835805607194e-08, | |
| "loss": 0.0908, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.26021472741758e-08, | |
| "loss": 0.0832, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.2412458742744414e-08, | |
| "loss": 0.0846, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.222277021131302e-08, | |
| "loss": 0.0722, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.2033081679881635e-08, | |
| "loss": 0.0673, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.184339314845024e-08, | |
| "loss": 0.0656, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1653704617018856e-08, | |
| "loss": 0.0585, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.146401608558746e-08, | |
| "loss": 0.0589, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.127432755415608e-08, | |
| "loss": 0.0543, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.1084639022724684e-08, | |
| "loss": 0.0548, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.08949504912933e-08, | |
| "loss": 0.0496, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.0705261959861905e-08, | |
| "loss": 0.055, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.051557342843052e-08, | |
| "loss": 0.0437, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.0325884896999126e-08, | |
| "loss": 0.0429, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.013619636556774e-08, | |
| "loss": 0.0451, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.03802574425935745, | |
| "eval_runtime": 313.8581, | |
| "eval_samples_per_second": 125.974, | |
| "eval_steps_per_second": 15.749, | |
| "step": 26359 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9946507834136347e-08, | |
| "loss": 0.0423, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.975681930270496e-08, | |
| "loss": 0.0437, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.956713077127357e-08, | |
| "loss": 0.0409, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.937744223984218e-08, | |
| "loss": 0.034, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.918775370841079e-08, | |
| "loss": 0.0421, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.89980651769794e-08, | |
| "loss": 0.038, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.880837664554801e-08, | |
| "loss": 0.0391, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.861868811411662e-08, | |
| "loss": 0.0341, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.842899958268523e-08, | |
| "loss": 0.0438, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.8239311051253844e-08, | |
| "loss": 0.0398, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.804962251982245e-08, | |
| "loss": 0.0402, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.7859933988391064e-08, | |
| "loss": 0.0424, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.767024545695967e-08, | |
| "loss": 0.0381, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.748055692552828e-08, | |
| "loss": 0.027, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.729086839409689e-08, | |
| "loss": 0.0339, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.71011798626655e-08, | |
| "loss": 0.0318, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.691149133123411e-08, | |
| "loss": 0.0415, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.672180279980272e-08, | |
| "loss": 0.0323, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.6532114268371334e-08, | |
| "loss": 0.0349, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 3.634242573693994e-08, | |
| "loss": 0.0278, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.6152737205508555e-08, | |
| "loss": 0.0367, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 3.596304867407716e-08, | |
| "loss": 0.0292, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 3.5773360142645776e-08, | |
| "loss": 0.0322, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.558367161121438e-08, | |
| "loss": 0.0326, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.5393983079782997e-08, | |
| "loss": 0.0288, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.5204294548351604e-08, | |
| "loss": 0.0398, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.501460601692022e-08, | |
| "loss": 0.0265, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 3.4824917485488824e-08, | |
| "loss": 0.0372, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 3.463522895405744e-08, | |
| "loss": 0.042, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.4445540422626045e-08, | |
| "loss": 0.0383, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 3.425585189119466e-08, | |
| "loss": 0.0397, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.4066163359763266e-08, | |
| "loss": 0.0319, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.387647482833188e-08, | |
| "loss": 0.0286, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.368678629690049e-08, | |
| "loss": 0.033, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.3497097765469094e-08, | |
| "loss": 0.0361, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.330740923403771e-08, | |
| "loss": 0.0329, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.3117720702606315e-08, | |
| "loss": 0.0343, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.292803217117493e-08, | |
| "loss": 0.0302, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 3.2738343639743536e-08, | |
| "loss": 0.0237, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.254865510831215e-08, | |
| "loss": 0.0321, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.2358966576880757e-08, | |
| "loss": 0.0227, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.216927804544937e-08, | |
| "loss": 0.0324, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.197958951401798e-08, | |
| "loss": 0.0262, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 3.178990098258659e-08, | |
| "loss": 0.0337, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.16002124511552e-08, | |
| "loss": 0.0314, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.141052391972381e-08, | |
| "loss": 0.0433, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.122083538829242e-08, | |
| "loss": 0.0351, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.103114685686103e-08, | |
| "loss": 0.02, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.084145832542964e-08, | |
| "loss": 0.023, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.0651769793998254e-08, | |
| "loss": 0.0311, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.046208126256686e-08, | |
| "loss": 0.0231, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.0272392731135474e-08, | |
| "loss": 0.0277, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 3.008270419970408e-08, | |
| "loss": 0.0233, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.029595419764518738, | |
| "eval_runtime": 313.3391, | |
| "eval_samples_per_second": 126.183, | |
| "eval_steps_per_second": 15.775, | |
| "step": 52718 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.9893015668272695e-08, | |
| "loss": 0.0279, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.9703327136841306e-08, | |
| "loss": 0.0324, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.9513638605409913e-08, | |
| "loss": 0.0245, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 2.9323950073978527e-08, | |
| "loss": 0.0397, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.9134261542547134e-08, | |
| "loss": 0.0306, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.8944573011115747e-08, | |
| "loss": 0.0215, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.8754884479684354e-08, | |
| "loss": 0.0227, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.8565195948252968e-08, | |
| "loss": 0.0246, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 2.8375507416821575e-08, | |
| "loss": 0.0318, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 2.818581888539019e-08, | |
| "loss": 0.0454, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 2.7996130353958796e-08, | |
| "loss": 0.0236, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 2.780644182252741e-08, | |
| "loss": 0.032, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 2.7616753291096017e-08, | |
| "loss": 0.0315, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 2.742706475966463e-08, | |
| "loss": 0.0367, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.7237376228233238e-08, | |
| "loss": 0.0244, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.704768769680185e-08, | |
| "loss": 0.0268, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 2.685799916537046e-08, | |
| "loss": 0.0197, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.6668310633939072e-08, | |
| "loss": 0.037, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.647862210250768e-08, | |
| "loss": 0.0239, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.6288933571076293e-08, | |
| "loss": 0.0392, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.60992450396449e-08, | |
| "loss": 0.0269, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 2.5909556508213514e-08, | |
| "loss": 0.016, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.571986797678212e-08, | |
| "loss": 0.0293, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 2.553017944535073e-08, | |
| "loss": 0.0368, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.5340490913919342e-08, | |
| "loss": 0.0173, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 2.5150802382487952e-08, | |
| "loss": 0.037, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 2.4961113851056563e-08, | |
| "loss": 0.0328, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 2.4771425319625177e-08, | |
| "loss": 0.0312, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 2.4581736788193787e-08, | |
| "loss": 0.0435, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.4392048256762397e-08, | |
| "loss": 0.0302, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 2.4202359725331008e-08, | |
| "loss": 0.0327, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 2.4012671193899618e-08, | |
| "loss": 0.0266, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 2.382298266246823e-08, | |
| "loss": 0.0242, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.363329413103684e-08, | |
| "loss": 0.0203, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.3443605599605446e-08, | |
| "loss": 0.025, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 2.3253917068174057e-08, | |
| "loss": 0.0373, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 2.3064228536742667e-08, | |
| "loss": 0.0357, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 2.2874540005311277e-08, | |
| "loss": 0.0294, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 2.2684851473879888e-08, | |
| "loss": 0.0361, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 2.2495162942448498e-08, | |
| "loss": 0.0271, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 2.230547441101711e-08, | |
| "loss": 0.0417, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 2.211578587958572e-08, | |
| "loss": 0.0348, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 2.192609734815433e-08, | |
| "loss": 0.0322, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.173640881672294e-08, | |
| "loss": 0.0245, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.154672028529155e-08, | |
| "loss": 0.0393, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.135703175386016e-08, | |
| "loss": 0.0275, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.116734322242877e-08, | |
| "loss": 0.0221, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 2.097765469099738e-08, | |
| "loss": 0.0401, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 2.0787966159565992e-08, | |
| "loss": 0.0392, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 2.0598277628134602e-08, | |
| "loss": 0.0271, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.0408589096703213e-08, | |
| "loss": 0.027, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 2.0218900565271823e-08, | |
| "loss": 0.0267, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 2.0029212033840434e-08, | |
| "loss": 0.0379, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.02849128097295761, | |
| "eval_runtime": 312.9013, | |
| "eval_samples_per_second": 126.359, | |
| "eval_steps_per_second": 15.797, | |
| "step": 79077 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.9839523502409044e-08, | |
| "loss": 0.0335, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.9649834970977654e-08, | |
| "loss": 0.0283, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1.946014643954626e-08, | |
| "loss": 0.0296, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 1.9270457908114872e-08, | |
| "loss": 0.0174, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 1.9080769376683482e-08, | |
| "loss": 0.0228, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.8891080845252093e-08, | |
| "loss": 0.0271, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 1.8701392313820703e-08, | |
| "loss": 0.0178, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 1.8511703782389314e-08, | |
| "loss": 0.0293, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 1.8322015250957924e-08, | |
| "loss": 0.0405, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.8132326719526538e-08, | |
| "loss": 0.0273, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.7942638188095148e-08, | |
| "loss": 0.0292, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.775294965666376e-08, | |
| "loss": 0.0233, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.756326112523237e-08, | |
| "loss": 0.0327, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.737357259380098e-08, | |
| "loss": 0.046, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.718388406236959e-08, | |
| "loss": 0.0195, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.69941955309382e-08, | |
| "loss": 0.0266, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.680450699950681e-08, | |
| "loss": 0.0241, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.661481846807542e-08, | |
| "loss": 0.0169, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.642512993664403e-08, | |
| "loss": 0.0283, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.6235441405212642e-08, | |
| "loss": 0.0369, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.6045752873781252e-08, | |
| "loss": 0.0295, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.5856064342349863e-08, | |
| "loss": 0.0308, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.5666375810918473e-08, | |
| "loss": 0.0299, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 1.5476687279487084e-08, | |
| "loss": 0.0218, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 1.528699874805569e-08, | |
| "loss": 0.0207, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 1.50973102166243e-08, | |
| "loss": 0.0214, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 1.490762168519291e-08, | |
| "loss": 0.0302, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 1.4717933153761522e-08, | |
| "loss": 0.0246, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.4528244622330132e-08, | |
| "loss": 0.0241, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 1.4338556090898743e-08, | |
| "loss": 0.0516, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 1.4148867559467353e-08, | |
| "loss": 0.0328, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 1.3959179028035964e-08, | |
| "loss": 0.0495, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 1.3769490496604574e-08, | |
| "loss": 0.0216, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 1.3579801965173184e-08, | |
| "loss": 0.0389, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 1.3390113433741795e-08, | |
| "loss": 0.0189, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 1.3200424902310405e-08, | |
| "loss": 0.0179, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 1.3010736370879016e-08, | |
| "loss": 0.032, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 1.2821047839447626e-08, | |
| "loss": 0.0329, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 1.2631359308016237e-08, | |
| "loss": 0.0292, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 1.2441670776584847e-08, | |
| "loss": 0.026, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.2251982245153457e-08, | |
| "loss": 0.0323, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.2062293713722068e-08, | |
| "loss": 0.0223, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.1872605182290678e-08, | |
| "loss": 0.0243, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 1.1682916650859289e-08, | |
| "loss": 0.0167, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 1.1493228119427899e-08, | |
| "loss": 0.0222, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 1.130353958799651e-08, | |
| "loss": 0.0218, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.111385105656512e-08, | |
| "loss": 0.0304, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 1.092416252513373e-08, | |
| "loss": 0.0357, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.073447399370234e-08, | |
| "loss": 0.0377, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 1.0544785462270951e-08, | |
| "loss": 0.0301, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 1.0355096930839562e-08, | |
| "loss": 0.0274, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.016540839940817e-08, | |
| "loss": 0.0276, | |
| "step": 105000 | |
| } | |
| ], | |
| "max_steps": 131795, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.6644037006596096e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |