| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.977640545144804, | |
| "global_step": 93500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.98418867120954e-05, | |
| "loss": 4.6657, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.9682176320272572e-05, | |
| "loss": 4.167, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.9522465928449746e-05, | |
| "loss": 4.0464, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.9362755536626915e-05, | |
| "loss": 3.9743, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.9203045144804092e-05, | |
| "loss": 3.9457, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.904333475298126e-05, | |
| "loss": 3.961, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.8883624361158435e-05, | |
| "loss": 3.9259, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.8723913969335604e-05, | |
| "loss": 3.9035, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.8564522998296424e-05, | |
| "loss": 3.8129, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.8404812606473597e-05, | |
| "loss": 3.9113, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.8245102214650767e-05, | |
| "loss": 3.8406, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.808539182282794e-05, | |
| "loss": 3.8943, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.7926000851788756e-05, | |
| "loss": 3.8532, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.776629045996593e-05, | |
| "loss": 3.8599, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.76065800681431e-05, | |
| "loss": 3.8322, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.744718909710392e-05, | |
| "loss": 3.8144, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.728747870528109e-05, | |
| "loss": 3.786, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.712776831345826e-05, | |
| "loss": 3.7725, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.6968057921635438e-05, | |
| "loss": 3.8074, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6808347529812607e-05, | |
| "loss": 3.8163, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.664863713798978e-05, | |
| "loss": 3.8075, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.648892674616695e-05, | |
| "loss": 3.7558, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.6329216354344123e-05, | |
| "loss": 3.7775, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.616982538330494e-05, | |
| "loss": 3.7994, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.6010114991482113e-05, | |
| "loss": 3.7975, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.5850404599659286e-05, | |
| "loss": 3.7754, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.569069420783646e-05, | |
| "loss": 3.7374, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.5531303236797275e-05, | |
| "loss": 3.7527, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.5371592844974445e-05, | |
| "loss": 3.7226, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.521188245315162e-05, | |
| "loss": 3.8012, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.505217206132879e-05, | |
| "loss": 3.7938, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.4892781090289607e-05, | |
| "loss": 3.7989, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.4733070698466784e-05, | |
| "loss": 3.7258, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.4573360306643953e-05, | |
| "loss": 3.7496, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.4413649914821123e-05, | |
| "loss": 3.7927, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.4254258943781943e-05, | |
| "loss": 3.7524, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.4094548551959116e-05, | |
| "loss": 3.7317, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_gen_len": 5.9904, | |
| "eval_loss": 4.179744243621826, | |
| "eval_rouge1": 12.674, | |
| "eval_rouge2": 2.9346, | |
| "eval_rougeL": 12.6863, | |
| "eval_rougeLsum": 12.6925, | |
| "eval_runtime": 104.7366, | |
| "eval_samples_per_second": 72.563, | |
| "eval_steps_per_second": 9.07, | |
| "step": 18784 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 2.3934838160136285e-05, | |
| "loss": 3.7256, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.377512776831346e-05, | |
| "loss": 3.7268, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 2.3615736797274278e-05, | |
| "loss": 3.7327, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 2.3456026405451448e-05, | |
| "loss": 3.7136, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 2.329631601362862e-05, | |
| "loss": 3.708, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 2.313660562180579e-05, | |
| "loss": 3.6911, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 2.297721465076661e-05, | |
| "loss": 3.6855, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.2817504258943783e-05, | |
| "loss": 3.7239, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.2657793867120953e-05, | |
| "loss": 3.6898, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.249808347529813e-05, | |
| "loss": 3.6916, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.2338692504258946e-05, | |
| "loss": 3.7224, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.2178982112436115e-05, | |
| "loss": 3.7074, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.201927172061329e-05, | |
| "loss": 3.6832, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.185956132879046e-05, | |
| "loss": 3.6984, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.1700170357751278e-05, | |
| "loss": 3.617, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.154045996592845e-05, | |
| "loss": 3.6604, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.1380749574105624e-05, | |
| "loss": 3.666, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.1221039182282794e-05, | |
| "loss": 3.6795, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.1061648211243613e-05, | |
| "loss": 3.7, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.0901937819420783e-05, | |
| "loss": 3.7005, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.0742227427597956e-05, | |
| "loss": 3.6941, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.058251703577513e-05, | |
| "loss": 3.6645, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.0423126064735945e-05, | |
| "loss": 3.6256, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.026341567291312e-05, | |
| "loss": 3.6873, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.010370528109029e-05, | |
| "loss": 3.6912, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.994399488926746e-05, | |
| "loss": 3.6764, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.9784603918228277e-05, | |
| "loss": 3.6543, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.9624893526405454e-05, | |
| "loss": 3.6974, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.9465183134582624e-05, | |
| "loss": 3.6267, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.9305472742759797e-05, | |
| "loss": 3.7287, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9146081771720616e-05, | |
| "loss": 3.6518, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.8986371379897786e-05, | |
| "loss": 3.659, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.8826660988074956e-05, | |
| "loss": 3.6865, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.866695059625213e-05, | |
| "loss": 3.649, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.850755962521295e-05, | |
| "loss": 3.69, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.8347849233390118e-05, | |
| "loss": 3.6672, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.818813884156729e-05, | |
| "loss": 3.6634, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.8028428449744464e-05, | |
| "loss": 3.6919, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_gen_len": 6.0964, | |
| "eval_loss": 4.154943943023682, | |
| "eval_rouge1": 12.7072, | |
| "eval_rouge2": 2.9431, | |
| "eval_rougeL": 12.7199, | |
| "eval_rougeLsum": 12.7279, | |
| "eval_runtime": 105.4406, | |
| "eval_samples_per_second": 72.079, | |
| "eval_steps_per_second": 9.01, | |
| "step": 37568 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.786903747870528e-05, | |
| "loss": 3.6437, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.7709327086882454e-05, | |
| "loss": 3.6416, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.7549616695059623e-05, | |
| "loss": 3.6792, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.73899063032368e-05, | |
| "loss": 3.6606, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.7230515332197616e-05, | |
| "loss": 3.6146, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.7070804940374786e-05, | |
| "loss": 3.6238, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.6911094548551962e-05, | |
| "loss": 3.6344, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.6751384156729132e-05, | |
| "loss": 3.6348, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.6591993185689948e-05, | |
| "loss": 3.6357, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.643228279386712e-05, | |
| "loss": 3.6303, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.6272572402044294e-05, | |
| "loss": 3.6216, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.6112862010221464e-05, | |
| "loss": 3.6247, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.5953471039182283e-05, | |
| "loss": 3.6372, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.5793760647359457e-05, | |
| "loss": 3.6249, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.5634050255536626e-05, | |
| "loss": 3.6202, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.54743398637138e-05, | |
| "loss": 3.6435, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1.531494889267462e-05, | |
| "loss": 3.6407, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1.5155238500851789e-05, | |
| "loss": 3.6552, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 1.4995528109028962e-05, | |
| "loss": 3.6302, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1.4835817717206133e-05, | |
| "loss": 3.5919, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1.4676426746166951e-05, | |
| "loss": 3.5796, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1.4516716354344122e-05, | |
| "loss": 3.6223, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 1.4357005962521294e-05, | |
| "loss": 3.5963, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.4197295570698467e-05, | |
| "loss": 3.6373, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 1.4037904599659285e-05, | |
| "loss": 3.5909, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1.3878194207836456e-05, | |
| "loss": 3.6421, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.371848381601363e-05, | |
| "loss": 3.5948, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.35587734241908e-05, | |
| "loss": 3.6234, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.3399382453151619e-05, | |
| "loss": 3.6113, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.323967206132879e-05, | |
| "loss": 3.5963, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.3079961669505963e-05, | |
| "loss": 3.628, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1.2920251277683136e-05, | |
| "loss": 3.6178, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 1.2760860306643952e-05, | |
| "loss": 3.6027, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.2601149914821126e-05, | |
| "loss": 3.6703, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.2441439522998297e-05, | |
| "loss": 3.6103, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 1.2281729131175468e-05, | |
| "loss": 3.596, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.2122338160136286e-05, | |
| "loss": 3.6205, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_gen_len": 6.1543, | |
| "eval_loss": 4.141897678375244, | |
| "eval_rouge1": 12.7145, | |
| "eval_rouge2": 2.93, | |
| "eval_rougeL": 12.7274, | |
| "eval_rougeLsum": 12.7376, | |
| "eval_runtime": 108.5251, | |
| "eval_samples_per_second": 70.03, | |
| "eval_steps_per_second": 8.754, | |
| "step": 56352 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 1.196262776831346e-05, | |
| "loss": 3.6473, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 1.180291737649063e-05, | |
| "loss": 3.5711, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.1643206984667802e-05, | |
| "loss": 3.6248, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 1.148381601362862e-05, | |
| "loss": 3.6218, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.1324105621805791e-05, | |
| "loss": 3.6035, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 1.1164395229982965e-05, | |
| "loss": 3.6007, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 1.1004684838160136e-05, | |
| "loss": 3.6097, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.0845293867120954e-05, | |
| "loss": 3.5995, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.0685583475298127e-05, | |
| "loss": 3.5552, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.0525873083475298e-05, | |
| "loss": 3.5995, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.0366162691652471e-05, | |
| "loss": 3.6177, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.0206771720613288e-05, | |
| "loss": 3.6039, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.004706132879046e-05, | |
| "loss": 3.6237, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 9.887350936967632e-06, | |
| "loss": 3.5536, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 9.727640545144804e-06, | |
| "loss": 3.6287, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 9.568249574105623e-06, | |
| "loss": 3.6013, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 9.408539182282794e-06, | |
| "loss": 3.6343, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 9.248828790459966e-06, | |
| "loss": 3.6494, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 9.089118398637137e-06, | |
| "loss": 3.6276, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 8.929727427597955e-06, | |
| "loss": 3.606, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 8.770017035775128e-06, | |
| "loss": 3.5757, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 8.6103066439523e-06, | |
| "loss": 3.5841, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 8.450596252129473e-06, | |
| "loss": 3.5987, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 8.291205281090289e-06, | |
| "loss": 3.5761, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 8.131494889267462e-06, | |
| "loss": 3.5931, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 7.971784497444633e-06, | |
| "loss": 3.5937, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 7.812074105621807e-06, | |
| "loss": 3.5696, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 7.652683134582624e-06, | |
| "loss": 3.6207, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 7.492972742759796e-06, | |
| "loss": 3.5782, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 7.333262350936967e-06, | |
| "loss": 3.5555, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 7.1735519591141395e-06, | |
| "loss": 3.5753, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 7.014160988074958e-06, | |
| "loss": 3.5215, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 6.85445059625213e-06, | |
| "loss": 3.5889, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 6.694740204429302e-06, | |
| "loss": 3.5766, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 6.535029812606473e-06, | |
| "loss": 3.5778, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 6.375638841567291e-06, | |
| "loss": 3.6086, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 6.2159284497444634e-06, | |
| "loss": 3.5562, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 6.056218057921636e-06, | |
| "loss": 3.5603, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_gen_len": 6.1403, | |
| "eval_loss": 4.137481212615967, | |
| "eval_rouge1": 12.6976, | |
| "eval_rouge2": 2.9239, | |
| "eval_rougeL": 12.7112, | |
| "eval_rougeLsum": 12.7168, | |
| "eval_runtime": 106.3438, | |
| "eval_samples_per_second": 71.466, | |
| "eval_steps_per_second": 8.933, | |
| "step": 75136 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.896507666098808e-06, | |
| "loss": 3.5449, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 5.737116695059626e-06, | |
| "loss": 3.5331, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 5.577406303236797e-06, | |
| "loss": 3.5443, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 5.4176959114139695e-06, | |
| "loss": 3.5557, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 5.257985519591142e-06, | |
| "loss": 3.6073, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 5.098594548551959e-06, | |
| "loss": 3.5791, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 4.938884156729132e-06, | |
| "loss": 3.6755, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 4.779173764906303e-06, | |
| "loss": 3.5966, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 4.6194633730834755e-06, | |
| "loss": 3.5897, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 4.459752981260648e-06, | |
| "loss": 3.5891, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 4.300362010221465e-06, | |
| "loss": 3.5978, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 4.140651618398637e-06, | |
| "loss": 3.5899, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 3.980941226575809e-06, | |
| "loss": 3.5805, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.8212308347529816e-06, | |
| "loss": 3.5577, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 3.6618398637137994e-06, | |
| "loss": 3.5753, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 3.5021294718909713e-06, | |
| "loss": 3.576, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 3.342419080068143e-06, | |
| "loss": 3.5951, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 3.182708688245315e-06, | |
| "loss": 3.5792, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 3.0233177172061332e-06, | |
| "loss": 3.5824, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 2.863607325383305e-06, | |
| "loss": 3.5487, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 2.703896933560477e-06, | |
| "loss": 3.5632, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 2.544186541737649e-06, | |
| "loss": 3.6142, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 2.384795570698467e-06, | |
| "loss": 3.4888, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 2.225085178875639e-06, | |
| "loss": 3.5823, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 2.0653747870528107e-06, | |
| "loss": 3.5525, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 1.9056643952299831e-06, | |
| "loss": 3.592, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 1.7462734241908006e-06, | |
| "loss": 3.5657, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 1.5865630323679728e-06, | |
| "loss": 3.5557, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 1.4268526405451449e-06, | |
| "loss": 3.5917, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 1.267142248722317e-06, | |
| "loss": 3.6155, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 1.1077512776831346e-06, | |
| "loss": 3.5609, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 9.480408858603066e-07, | |
| "loss": 3.5539, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 7.883304940374787e-07, | |
| "loss": 3.5676, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 6.286201022146508e-07, | |
| "loss": 3.6031, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 4.692291311754685e-07, | |
| "loss": 3.5581, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 3.095187393526406e-07, | |
| "loss": 3.5873, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 1.4980834752981263e-07, | |
| "loss": 3.5911, | |
| "step": 93500 | |
| } | |
| ], | |
| "max_steps": 93920, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.7064702150967296e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |