| { | |
| "best_metric": 0.6417234539985657, | |
| "best_model_checkpoint": "./enko_mbartLarge_100p_sup2/checkpoint-129072", | |
| "epoch": 6.0, | |
| "eval_steps": 500, | |
| "global_step": 258144, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 2.2901, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.992e-05, | |
| "loss": 1.6084, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9920000000000005e-05, | |
| "loss": 1.449, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.9920000000000004e-05, | |
| "loss": 1.3874, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.992e-05, | |
| "loss": 1.3485, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.996142239367825e-05, | |
| "loss": 1.2926, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.9922533677628105e-05, | |
| "loss": 1.248, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.988364496157795e-05, | |
| "loss": 1.1931, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.98447562455278e-05, | |
| "loss": 1.1677, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.980594530690975e-05, | |
| "loss": 1.142, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.97670565908596e-05, | |
| "loss": 1.122, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.972816787480945e-05, | |
| "loss": 1.0855, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.968927915875929e-05, | |
| "loss": 1.0719, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.965039044270914e-05, | |
| "loss": 1.053, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9611501726658995e-05, | |
| "loss": 1.0359, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.957261301060884e-05, | |
| "loss": 1.0232, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.953380207199079e-05, | |
| "loss": 1.0169, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.949491335594064e-05, | |
| "loss": 1.0141, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.945602463989049e-05, | |
| "loss": 0.9855, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.941721370127244e-05, | |
| "loss": 0.9932, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9378324985222293e-05, | |
| "loss": 0.9656, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.933943626917214e-05, | |
| "loss": 0.9694, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.930054755312199e-05, | |
| "loss": 0.9609, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.926165883707184e-05, | |
| "loss": 0.9526, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9222770121021686e-05, | |
| "loss": 0.9359, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.9183959182403636e-05, | |
| "loss": 0.9324, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.914507046635349e-05, | |
| "loss": 0.9251, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.910625952773543e-05, | |
| "loss": 0.9216, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.906737081168528e-05, | |
| "loss": 0.9181, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.902855987306723e-05, | |
| "loss": 0.9098, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.898967115701708e-05, | |
| "loss": 0.9067, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.895078244096693e-05, | |
| "loss": 0.8931, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.891189372491678e-05, | |
| "loss": 0.8923, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.887300500886663e-05, | |
| "loss": 0.8703, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.8834116292816475e-05, | |
| "loss": 0.8861, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.879522757676633e-05, | |
| "loss": 0.8864, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.875633886071618e-05, | |
| "loss": 0.886, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.871752792209813e-05, | |
| "loss": 0.8737, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.867863920604798e-05, | |
| "loss": 0.8708, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.8639750489997824e-05, | |
| "loss": 0.865, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.8600861773947676e-05, | |
| "loss": 0.8571, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.856197305789753e-05, | |
| "loss": 0.8621, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.852308434184737e-05, | |
| "loss": 0.8607, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.848419562579722e-05, | |
| "loss": 0.8519, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.844530690974707e-05, | |
| "loss": 0.8472, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.840641819369692e-05, | |
| "loss": 0.8381, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.8367529477646765e-05, | |
| "loss": 0.8329, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.8328640761596616e-05, | |
| "loss": 0.8425, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.828975204554647e-05, | |
| "loss": 0.8227, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.825086332949631e-05, | |
| "loss": 0.8349, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.8211974613446164e-05, | |
| "loss": 0.8235, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.8173085897396015e-05, | |
| "loss": 0.8116, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.813419718134587e-05, | |
| "loss": 0.8367, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.809530846529571e-05, | |
| "loss": 0.8219, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.805641974924556e-05, | |
| "loss": 0.8206, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.8017531033195415e-05, | |
| "loss": 0.8199, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.797864231714526e-05, | |
| "loss": 0.816, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.793983137852721e-05, | |
| "loss": 0.8144, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.790102043990916e-05, | |
| "loss": 0.806, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.786220950129111e-05, | |
| "loss": 0.8134, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.782332078524095e-05, | |
| "loss": 0.8109, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.7784432069190805e-05, | |
| "loss": 0.792, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.7745543353140656e-05, | |
| "loss": 0.8046, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.77066546370905e-05, | |
| "loss": 0.7983, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.766776592104035e-05, | |
| "loss": 0.7937, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.7628877204990204e-05, | |
| "loss": 0.7978, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.758998848894005e-05, | |
| "loss": 0.7961, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.75510997728899e-05, | |
| "loss": 0.7854, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.751221105683975e-05, | |
| "loss": 0.8006, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.7473322340789597e-05, | |
| "loss": 0.7857, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.743443362473945e-05, | |
| "loss": 0.7857, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.73956226861214e-05, | |
| "loss": 0.7874, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.735673397007125e-05, | |
| "loss": 0.7826, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.73178452540211e-05, | |
| "loss": 0.7902, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.727895653797094e-05, | |
| "loss": 0.7695, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.724006782192079e-05, | |
| "loss": 0.7789, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.720117910587064e-05, | |
| "loss": 0.7739, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.716229038982049e-05, | |
| "loss": 0.7726, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.712340167377034e-05, | |
| "loss": 0.7722, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.708451295772019e-05, | |
| "loss": 0.7656, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.704562424167004e-05, | |
| "loss": 0.77, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.700689108048409e-05, | |
| "loss": 0.7814, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.696800236443394e-05, | |
| "loss": 0.7704, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.692919142581589e-05, | |
| "loss": 0.7611, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.689038048719784e-05, | |
| "loss": 0.7629, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6851491771147685e-05, | |
| "loss": 0.7676, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_bleu": 55.2526, | |
| "eval_gen_len": 16.382, | |
| "eval_loss": 0.7125306129455566, | |
| "eval_runtime": 7352.67, | |
| "eval_samples_per_second": 11.685, | |
| "eval_steps_per_second": 1.461, | |
| "step": 43024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6812603055097536e-05, | |
| "loss": 0.6434, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6773792116479486e-05, | |
| "loss": 0.6436, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.673490340042934e-05, | |
| "loss": 0.651, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.669601468437918e-05, | |
| "loss": 0.6517, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.665712596832903e-05, | |
| "loss": 0.6416, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.661823725227888e-05, | |
| "loss": 0.6517, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.657934853622873e-05, | |
| "loss": 0.6539, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.654045982017858e-05, | |
| "loss": 0.6587, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.6501571104128426e-05, | |
| "loss": 0.6518, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.646268238807828e-05, | |
| "loss": 0.6465, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.642379367202813e-05, | |
| "loss": 0.6437, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.6384904955977974e-05, | |
| "loss": 0.6458, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.6346016239927825e-05, | |
| "loss": 0.6476, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.630712752387768e-05, | |
| "loss": 0.6587, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.626823880782752e-05, | |
| "loss": 0.6392, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.622935009177737e-05, | |
| "loss": 0.6547, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.6190461375727225e-05, | |
| "loss": 0.6511, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.615157265967707e-05, | |
| "loss": 0.6522, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.611268394362692e-05, | |
| "loss": 0.6555, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.607395078244097e-05, | |
| "loss": 0.6461, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.6035062066390814e-05, | |
| "loss": 0.6471, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.5996173350340665e-05, | |
| "loss": 0.6563, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.5957284634290517e-05, | |
| "loss": 0.6447, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.591839591824036e-05, | |
| "loss": 0.6483, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.587958497962231e-05, | |
| "loss": 0.6475, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.584069626357216e-05, | |
| "loss": 0.6546, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.5801807547522014e-05, | |
| "loss": 0.6528, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.5762918831471866e-05, | |
| "loss": 0.6538, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.572403011542171e-05, | |
| "loss": 0.6511, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.568521917680366e-05, | |
| "loss": 0.6457, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.564633046075351e-05, | |
| "loss": 0.659, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.560744174470336e-05, | |
| "loss": 0.6358, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.556855302865321e-05, | |
| "loss": 0.6425, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.552966431260306e-05, | |
| "loss": 0.644, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.5490775596552904e-05, | |
| "loss": 0.6537, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.5451886880502756e-05, | |
| "loss": 0.6529, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.5413075941884705e-05, | |
| "loss": 0.6547, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.537418722583455e-05, | |
| "loss": 0.6443, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.53352985097844e-05, | |
| "loss": 0.6516, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.529640979373425e-05, | |
| "loss": 0.6439, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.52575210776841e-05, | |
| "loss": 0.6401, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.521863236163395e-05, | |
| "loss": 0.6545, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.51797436455838e-05, | |
| "loss": 0.6433, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.514085492953365e-05, | |
| "loss": 0.6538, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.51019662134835e-05, | |
| "loss": 0.6548, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.506307749743335e-05, | |
| "loss": 0.6458, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.50241887813832e-05, | |
| "loss": 0.649, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.4985300065333045e-05, | |
| "loss": 0.6471, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.4946411349282896e-05, | |
| "loss": 0.6526, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.4907600410664846e-05, | |
| "loss": 0.646, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.486878947204679e-05, | |
| "loss": 0.6491, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.482990075599664e-05, | |
| "loss": 0.6484, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.479101203994649e-05, | |
| "loss": 0.6421, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.475220110132844e-05, | |
| "loss": 0.6498, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.4713312385278286e-05, | |
| "loss": 0.6497, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.467442366922814e-05, | |
| "loss": 0.6503, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.463553495317799e-05, | |
| "loss": 0.6314, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.459672401455994e-05, | |
| "loss": 0.6495, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.4557835298509784e-05, | |
| "loss": 0.643, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.4518946582459635e-05, | |
| "loss": 0.6542, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.448005786640949e-05, | |
| "loss": 0.6522, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.444116915035934e-05, | |
| "loss": 0.6476, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.440228043430918e-05, | |
| "loss": 0.6337, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.4363391718259035e-05, | |
| "loss": 0.6546, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.4324503002208886e-05, | |
| "loss": 0.647, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.4285614286158724e-05, | |
| "loss": 0.6467, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.4246725570108576e-05, | |
| "loss": 0.6432, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.420783685405843e-05, | |
| "loss": 0.6399, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.416894813800828e-05, | |
| "loss": 0.6382, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.4130059421958123e-05, | |
| "loss": 0.649, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.409124848334007e-05, | |
| "loss": 0.6385, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.4052359767289925e-05, | |
| "loss": 0.6448, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.4013471051239776e-05, | |
| "loss": 0.638, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.397458233518962e-05, | |
| "loss": 0.6317, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.393569361913947e-05, | |
| "loss": 0.6338, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.3896804903089324e-05, | |
| "loss": 0.6406, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.385791618703917e-05, | |
| "loss": 0.6363, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.381902747098902e-05, | |
| "loss": 0.6381, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.378021653237097e-05, | |
| "loss": 0.6383, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.374140559375292e-05, | |
| "loss": 0.6351, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.370251687770277e-05, | |
| "loss": 0.642, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.3663705939084714e-05, | |
| "loss": 0.6399, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.3624817223034566e-05, | |
| "loss": 0.6351, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.358592850698442e-05, | |
| "loss": 0.631, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.354703979093426e-05, | |
| "loss": 0.6298, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.350822885231621e-05, | |
| "loss": 0.6349, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bleu": 58.202, | |
| "eval_gen_len": 15.9466, | |
| "eval_loss": 0.6546894311904907, | |
| "eval_runtime": 6844.8665, | |
| "eval_samples_per_second": 12.552, | |
| "eval_steps_per_second": 1.569, | |
| "step": 86048 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.346934013626606e-05, | |
| "loss": 0.5093, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.3430451420215914e-05, | |
| "loss": 0.5066, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.339156270416576e-05, | |
| "loss": 0.4956, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.335267398811561e-05, | |
| "loss": 0.5, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.331378527206546e-05, | |
| "loss": 0.5034, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.327489655601531e-05, | |
| "loss": 0.505, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.323600783996516e-05, | |
| "loss": 0.4995, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.319719690134711e-05, | |
| "loss": 0.5064, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.315830818529696e-05, | |
| "loss": 0.5065, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.311949724667891e-05, | |
| "loss": 0.5031, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.3080608530628754e-05, | |
| "loss": 0.4967, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.30417198145786e-05, | |
| "loss": 0.5112, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.300290887596055e-05, | |
| "loss": 0.5069, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.29640201599104e-05, | |
| "loss": 0.5106, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.292513144386025e-05, | |
| "loss": 0.5077, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.28862427278101e-05, | |
| "loss": 0.5179, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.284735401175995e-05, | |
| "loss": 0.5192, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.28084652957098e-05, | |
| "loss": 0.5103, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.276957657965965e-05, | |
| "loss": 0.5203, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.2730687863609496e-05, | |
| "loss": 0.5179, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.269179914755935e-05, | |
| "loss": 0.515, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.26529104315092e-05, | |
| "loss": 0.5133, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.2614021715459043e-05, | |
| "loss": 0.5147, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.2575132999408895e-05, | |
| "loss": 0.5308, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.2536322060790845e-05, | |
| "loss": 0.5207, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.2497511122172794e-05, | |
| "loss": 0.5228, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.245862240612264e-05, | |
| "loss": 0.5207, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.241973369007249e-05, | |
| "loss": 0.5211, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.2380844974022335e-05, | |
| "loss": 0.5169, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.234195625797219e-05, | |
| "loss": 0.5302, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.230306754192204e-05, | |
| "loss": 0.5155, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.226417882587188e-05, | |
| "loss": 0.5193, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.2225290109821735e-05, | |
| "loss": 0.518, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.2186401393771586e-05, | |
| "loss": 0.5171, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 4.214751267772144e-05, | |
| "loss": 0.5208, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.210862396167128e-05, | |
| "loss": 0.5231, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.2069735245621134e-05, | |
| "loss": 0.5267, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.2030846529570985e-05, | |
| "loss": 0.5196, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 4.199195781352083e-05, | |
| "loss": 0.5234, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.1953224652334885e-05, | |
| "loss": 0.526, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.191449149114893e-05, | |
| "loss": 0.5312, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 4.187560277509878e-05, | |
| "loss": 0.5286, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.183671405904863e-05, | |
| "loss": 0.5225, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.1797825342998474e-05, | |
| "loss": 0.5207, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.1758936626948325e-05, | |
| "loss": 0.5263, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.172004791089818e-05, | |
| "loss": 0.5243, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.1681236972280126e-05, | |
| "loss": 0.5332, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.164234825622997e-05, | |
| "loss": 0.5297, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.160345954017982e-05, | |
| "loss": 0.5274, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.1564570824129674e-05, | |
| "loss": 0.5241, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.152568210807952e-05, | |
| "loss": 0.528, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.148679339202937e-05, | |
| "loss": 0.53, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.144790467597922e-05, | |
| "loss": 0.5291, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.1409015959929073e-05, | |
| "loss": 0.5431, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.137012724387892e-05, | |
| "loss": 0.5336, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.133123852782877e-05, | |
| "loss": 0.531, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.129234981177862e-05, | |
| "loss": 0.5278, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.1253461095728466e-05, | |
| "loss": 0.5214, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.121457237967831e-05, | |
| "loss": 0.5249, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.117568366362816e-05, | |
| "loss": 0.5351, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.1136794947578014e-05, | |
| "loss": 0.5275, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.109790623152786e-05, | |
| "loss": 0.5309, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.105901751547771e-05, | |
| "loss": 0.5291, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.102012879942756e-05, | |
| "loss": 0.5274, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.098131786080951e-05, | |
| "loss": 0.532, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.0942429144759356e-05, | |
| "loss": 0.5278, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.090354042870921e-05, | |
| "loss": 0.5338, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.086465171265906e-05, | |
| "loss": 0.5289, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.082576299660891e-05, | |
| "loss": 0.5388, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.0786874280558755e-05, | |
| "loss": 0.5337, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.074798556450861e-05, | |
| "loss": 0.5308, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.0709174625890556e-05, | |
| "loss": 0.5294, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.067028590984041e-05, | |
| "loss": 0.5385, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.063139719379025e-05, | |
| "loss": 0.5379, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.05925084777401e-05, | |
| "loss": 0.5485, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.055361976168995e-05, | |
| "loss": 0.5391, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.05147310456398e-05, | |
| "loss": 0.5414, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.0475842329589645e-05, | |
| "loss": 0.5237, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.04369536135395e-05, | |
| "loss": 0.5363, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.039806489748935e-05, | |
| "loss": 0.543, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.0359331736303396e-05, | |
| "loss": 0.5265, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.032044302025325e-05, | |
| "loss": 0.5379, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.028155430420309e-05, | |
| "loss": 0.5311, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.0242665588152944e-05, | |
| "loss": 0.5361, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.0203776872102795e-05, | |
| "loss": 0.5405, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.016488815605265e-05, | |
| "loss": 0.537, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bleu": 59.1835, | |
| "eval_gen_len": 15.7226, | |
| "eval_loss": 0.6417234539985657, | |
| "eval_runtime": 6645.2011, | |
| "eval_samples_per_second": 12.929, | |
| "eval_steps_per_second": 1.616, | |
| "step": 129072 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.0126077217434597e-05, | |
| "loss": 0.4217, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.008718850138444e-05, | |
| "loss": 0.3942, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.0048299785334286e-05, | |
| "loss": 0.3948, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.0009488846716236e-05, | |
| "loss": 0.3895, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.9970677908098185e-05, | |
| "loss": 0.3912, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.993178919204804e-05, | |
| "loss": 0.3989, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.989290047599789e-05, | |
| "loss": 0.3975, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.985401175994773e-05, | |
| "loss": 0.4036, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.9815123043897585e-05, | |
| "loss": 0.4028, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.9776234327847436e-05, | |
| "loss": 0.3985, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.973734561179728e-05, | |
| "loss": 0.4002, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.969845689574713e-05, | |
| "loss": 0.3975, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.9659568179696984e-05, | |
| "loss": 0.4024, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.962067946364683e-05, | |
| "loss": 0.4016, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.958179074759668e-05, | |
| "loss": 0.4084, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.954290203154653e-05, | |
| "loss": 0.4054, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.9504013315496377e-05, | |
| "loss": 0.4061, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.946512459944623e-05, | |
| "loss": 0.4098, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.942623588339607e-05, | |
| "loss": 0.4115, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.9387347167345924e-05, | |
| "loss": 0.4068, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.9348536228727874e-05, | |
| "loss": 0.4058, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.930964751267772e-05, | |
| "loss": 0.4059, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.927075879662757e-05, | |
| "loss": 0.4145, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.9232025635441625e-05, | |
| "loss": 0.4104, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.919313691939147e-05, | |
| "loss": 0.4141, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.915424820334132e-05, | |
| "loss": 0.4158, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.911535948729117e-05, | |
| "loss": 0.4115, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.907647077124102e-05, | |
| "loss": 0.4197, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.903758205519087e-05, | |
| "loss": 0.4082, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.899869333914072e-05, | |
| "loss": 0.4231, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.8959804623090565e-05, | |
| "loss": 0.4237, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.892091590704042e-05, | |
| "loss": 0.4162, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.888202719099027e-05, | |
| "loss": 0.4189, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.884321625237221e-05, | |
| "loss": 0.4154, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.880432753632206e-05, | |
| "loss": 0.4193, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.876543882027191e-05, | |
| "loss": 0.4238, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.872655010422176e-05, | |
| "loss": 0.4274, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.868766138817161e-05, | |
| "loss": 0.4165, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.8648772672121455e-05, | |
| "loss": 0.4223, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.860988395607131e-05, | |
| "loss": 0.4213, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.8571073017453256e-05, | |
| "loss": 0.422, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.853218430140311e-05, | |
| "loss": 0.4245, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.849329558535296e-05, | |
| "loss": 0.4247, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.8454406869302804e-05, | |
| "loss": 0.4267, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.8415518153252656e-05, | |
| "loss": 0.4253, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.837662943720251e-05, | |
| "loss": 0.4311, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.833774072115235e-05, | |
| "loss": 0.4204, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.8298852005102203e-05, | |
| "loss": 0.4246, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.8259963289052055e-05, | |
| "loss": 0.4424, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.82210745730019e-05, | |
| "loss": 0.426, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.8182185856951744e-05, | |
| "loss": 0.4304, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.8143374918333694e-05, | |
| "loss": 0.4292, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.8104486202283546e-05, | |
| "loss": 0.426, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 3.80655974862334e-05, | |
| "loss": 0.4247, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.802670877018324e-05, | |
| "loss": 0.4271, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.798789783156519e-05, | |
| "loss": 0.4279, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.794900911551504e-05, | |
| "loss": 0.4357, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.791019817689699e-05, | |
| "loss": 0.435, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.7871309460846844e-05, | |
| "loss": 0.4336, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.7832420744796696e-05, | |
| "loss": 0.4303, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.779353202874654e-05, | |
| "loss": 0.4231, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.775464331269639e-05, | |
| "loss": 0.4303, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.7715754596646244e-05, | |
| "loss": 0.4301, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.767686588059609e-05, | |
| "loss": 0.4317, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.763797716454593e-05, | |
| "loss": 0.4421, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.7599088448495785e-05, | |
| "loss": 0.4338, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.7560199732445636e-05, | |
| "loss": 0.4315, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.752131101639548e-05, | |
| "loss": 0.4318, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.748242230034533e-05, | |
| "loss": 0.4378, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.7443533584295184e-05, | |
| "loss": 0.438, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.740464486824503e-05, | |
| "loss": 0.4385, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.736575615219488e-05, | |
| "loss": 0.4365, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.732686743614473e-05, | |
| "loss": 0.4286, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.7287978720094576e-05, | |
| "loss": 0.4369, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.7249167781476526e-05, | |
| "loss": 0.4361, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.721027906542638e-05, | |
| "loss": 0.4371, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.717139034937623e-05, | |
| "loss": 0.4351, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.713257941075818e-05, | |
| "loss": 0.4361, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.709376847214013e-05, | |
| "loss": 0.434, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.705487975608998e-05, | |
| "loss": 0.4408, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.701606881747192e-05, | |
| "loss": 0.4398, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.6977180101421774e-05, | |
| "loss": 0.4396, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.693829138537162e-05, | |
| "loss": 0.438, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.689940266932147e-05, | |
| "loss": 0.4374, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.686051395327132e-05, | |
| "loss": 0.4407, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.682162523722117e-05, | |
| "loss": 0.434, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bleu": 59.6194, | |
| "eval_gen_len": 15.702, | |
| "eval_loss": 0.6589247584342957, | |
| "eval_runtime": 6589.9687, | |
| "eval_samples_per_second": 13.037, | |
| "eval_steps_per_second": 1.63, | |
| "step": 172096 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.678273652117102e-05, | |
| "loss": 0.3325, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.674384780512087e-05, | |
| "loss": 0.3007, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.6704959089070715e-05, | |
| "loss": 0.305, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.6666070373020566e-05, | |
| "loss": 0.3016, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.662718165697042e-05, | |
| "loss": 0.2961, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.6588526273216564e-05, | |
| "loss": 0.3007, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.6549637557166415e-05, | |
| "loss": 0.3012, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.651074884111627e-05, | |
| "loss": 0.3094, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.647186012506612e-05, | |
| "loss": 0.3069, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.6432971409015956e-05, | |
| "loss": 0.3091, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.639408269296581e-05, | |
| "loss": 0.3147, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.635519397691566e-05, | |
| "loss": 0.3112, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.6316305260865504e-05, | |
| "loss": 0.3129, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.6277416544815356e-05, | |
| "loss": 0.309, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.623852782876521e-05, | |
| "loss": 0.3195, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.619963911271506e-05, | |
| "loss": 0.3219, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.6160750396664903e-05, | |
| "loss": 0.3134, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 3.6121861680614755e-05, | |
| "loss": 0.3182, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.6083050741996705e-05, | |
| "loss": 0.3131, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 3.6044162025946556e-05, | |
| "loss": 0.3193, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.6005351087328506e-05, | |
| "loss": 0.3153, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.596646237127835e-05, | |
| "loss": 0.3264, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.59276514326603e-05, | |
| "loss": 0.3195, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.588876271661015e-05, | |
| "loss": 0.3197, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 3.5849874000559996e-05, | |
| "loss": 0.3235, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.581098528450985e-05, | |
| "loss": 0.3188, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 3.577209656845969e-05, | |
| "loss": 0.3183, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 3.5733207852409544e-05, | |
| "loss": 0.3271, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 3.5694319136359396e-05, | |
| "loss": 0.3333, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.565543042030924e-05, | |
| "loss": 0.3216, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 3.561654170425909e-05, | |
| "loss": 0.3296, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.5577652988208944e-05, | |
| "loss": 0.3282, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.5538764272158795e-05, | |
| "loss": 0.3297, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 3.549987555610864e-05, | |
| "loss": 0.3283, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.546106461749059e-05, | |
| "loss": 0.3294, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 3.542217590144044e-05, | |
| "loss": 0.3275, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 3.538328718539029e-05, | |
| "loss": 0.324, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 3.534439846934014e-05, | |
| "loss": 0.3285, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 3.530558753072209e-05, | |
| "loss": 0.3338, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 3.526669881467194e-05, | |
| "loss": 0.3347, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 3.522781009862178e-05, | |
| "loss": 0.338, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 3.5188921382571635e-05, | |
| "loss": 0.3312, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 3.515003266652148e-05, | |
| "loss": 0.3364, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 3.511114395047133e-05, | |
| "loss": 0.3347, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 3.507225523442118e-05, | |
| "loss": 0.3334, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 3.503336651837103e-05, | |
| "loss": 0.3319, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 3.499455557975298e-05, | |
| "loss": 0.3332, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 3.495566686370283e-05, | |
| "loss": 0.3418, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 3.491677814765268e-05, | |
| "loss": 0.3402, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 3.487796720903463e-05, | |
| "loss": 0.3342, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 3.483907849298448e-05, | |
| "loss": 0.3294, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 3.4800189776934326e-05, | |
| "loss": 0.3393, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 3.476130106088418e-05, | |
| "loss": 0.3446, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 3.472241234483403e-05, | |
| "loss": 0.3375, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 3.4683523628783874e-05, | |
| "loss": 0.3399, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.4644634912733725e-05, | |
| "loss": 0.3405, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 3.460574619668357e-05, | |
| "loss": 0.3378, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 3.456693525806552e-05, | |
| "loss": 0.346, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.452804654201537e-05, | |
| "loss": 0.3455, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 3.4489157825965216e-05, | |
| "loss": 0.3432, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 3.4450346887347166e-05, | |
| "loss": 0.3505, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 3.441145817129702e-05, | |
| "loss": 0.3432, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 3.437256945524687e-05, | |
| "loss": 0.3405, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 3.433368073919671e-05, | |
| "loss": 0.3394, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 3.4294792023146565e-05, | |
| "loss": 0.3477, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 3.4255903307096416e-05, | |
| "loss": 0.352, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 3.421701459104627e-05, | |
| "loss": 0.3442, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 3.417812587499611e-05, | |
| "loss": 0.3487, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 3.4139237158945964e-05, | |
| "loss": 0.346, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 3.4100348442895816e-05, | |
| "loss": 0.3533, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 3.406145972684566e-05, | |
| "loss": 0.3501, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 3.4022571010795505e-05, | |
| "loss": 0.3472, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 3.398368229474536e-05, | |
| "loss": 0.351, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 3.3944871356127306e-05, | |
| "loss": 0.3415, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 3.390598264007716e-05, | |
| "loss": 0.3478, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 3.386717170145911e-05, | |
| "loss": 0.3475, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 3.382828298540895e-05, | |
| "loss": 0.3486, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 3.3789394269358804e-05, | |
| "loss": 0.3564, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 3.3750505553308655e-05, | |
| "loss": 0.3484, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 3.37116168372585e-05, | |
| "loss": 0.3541, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 3.367280589864045e-05, | |
| "loss": 0.3464, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 3.36339949600224e-05, | |
| "loss": 0.35, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 3.359510624397225e-05, | |
| "loss": 0.3475, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.35562175279221e-05, | |
| "loss": 0.3479, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 3.3517328811871954e-05, | |
| "loss": 0.3556, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.34784400958218e-05, | |
| "loss": 0.3504, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_bleu": 59.352, | |
| "eval_gen_len": 15.7454, | |
| "eval_loss": 0.7117229700088501, | |
| "eval_runtime": 6592.0433, | |
| "eval_samples_per_second": 13.033, | |
| "eval_steps_per_second": 1.629, | |
| "step": 215120 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.343955137977165e-05, | |
| "loss": 0.2566, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.3400662663721495e-05, | |
| "loss": 0.2267, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.336177394767134e-05, | |
| "loss": 0.2283, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.332288523162119e-05, | |
| "loss": 0.226, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.328399651557104e-05, | |
| "loss": 0.2285, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.3245107799520894e-05, | |
| "loss": 0.2307, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.320621908347074e-05, | |
| "loss": 0.2282, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.316733036742059e-05, | |
| "loss": 0.2302, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.312851942880254e-05, | |
| "loss": 0.2312, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 3.308970849018449e-05, | |
| "loss": 0.2318, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 3.305089755156644e-05, | |
| "loss": 0.2389, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 3.301200883551629e-05, | |
| "loss": 0.2371, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.2973120119466136e-05, | |
| "loss": 0.2396, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 3.293423140341599e-05, | |
| "loss": 0.2356, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 3.289534268736584e-05, | |
| "loss": 0.236, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 3.285645397131569e-05, | |
| "loss": 0.2381, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 3.281756525526553e-05, | |
| "loss": 0.2396, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 3.277875431664748e-05, | |
| "loss": 0.2361, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 3.273986560059733e-05, | |
| "loss": 0.2423, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 3.270097688454718e-05, | |
| "loss": 0.2407, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 3.266216594592913e-05, | |
| "loss": 0.2427, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 3.262327722987898e-05, | |
| "loss": 0.2454, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 3.258438851382883e-05, | |
| "loss": 0.2484, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 3.254549979777868e-05, | |
| "loss": 0.2467, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 3.250661108172853e-05, | |
| "loss": 0.2439, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 3.2467722365678375e-05, | |
| "loss": 0.2496, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 3.2428833649628226e-05, | |
| "loss": 0.2486, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 3.238994493357808e-05, | |
| "loss": 0.2466, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 3.235105621752792e-05, | |
| "loss": 0.2497, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 3.2312167501477774e-05, | |
| "loss": 0.2455, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 3.2273278785427626e-05, | |
| "loss": 0.2532, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 3.223439006937747e-05, | |
| "loss": 0.2454, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 3.219557913075942e-05, | |
| "loss": 0.2491, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 3.2156690414709265e-05, | |
| "loss": 0.2501, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 3.2117801698659116e-05, | |
| "loss": 0.2528, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 3.207891298260897e-05, | |
| "loss": 0.2482, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 3.204002426655881e-05, | |
| "loss": 0.2548, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 3.200121332794076e-05, | |
| "loss": 0.2565, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 3.1962324611890614e-05, | |
| "loss": 0.253, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 3.1923513673272564e-05, | |
| "loss": 0.251, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 3.1884624957222415e-05, | |
| "loss": 0.2542, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 3.1845736241172267e-05, | |
| "loss": 0.2611, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 3.180684752512211e-05, | |
| "loss": 0.2583, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 3.176795880907196e-05, | |
| "loss": 0.2577, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 3.1729070093021814e-05, | |
| "loss": 0.2555, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 3.169018137697166e-05, | |
| "loss": 0.2581, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 3.165129266092151e-05, | |
| "loss": 0.2525, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 3.1612403944871355e-05, | |
| "loss": 0.2607, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 3.157351522882121e-05, | |
| "loss": 0.2623, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 3.153462651277105e-05, | |
| "loss": 0.2644, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 3.14957377967209e-05, | |
| "loss": 0.2635, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 3.1456849080670755e-05, | |
| "loss": 0.2644, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 3.14179603646206e-05, | |
| "loss": 0.2586, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 3.137907164857045e-05, | |
| "loss": 0.2688, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 3.13401829325203e-05, | |
| "loss": 0.2592, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 3.1301294216470154e-05, | |
| "loss": 0.2627, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 3.12626388327163e-05, | |
| "loss": 0.2659, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 3.122375011666615e-05, | |
| "loss": 0.2631, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 3.1184861400616e-05, | |
| "loss": 0.2647, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 3.114597268456585e-05, | |
| "loss": 0.2662, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 3.11070839685157e-05, | |
| "loss": 0.2647, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 3.106819525246555e-05, | |
| "loss": 0.2735, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 3.10293843138475e-05, | |
| "loss": 0.27, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 3.0990495597797345e-05, | |
| "loss": 0.2696, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 3.095160688174719e-05, | |
| "loss": 0.2652, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 3.091279594312914e-05, | |
| "loss": 0.2704, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 3.087390722707899e-05, | |
| "loss": 0.2633, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 3.083501851102884e-05, | |
| "loss": 0.2696, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 3.079612979497869e-05, | |
| "loss": 0.2657, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 3.075724107892854e-05, | |
| "loss": 0.2705, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 3.071835236287839e-05, | |
| "loss": 0.2727, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 3.0679463646828235e-05, | |
| "loss": 0.2636, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 3.064057493077809e-05, | |
| "loss": 0.2772, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 3.060168621472794e-05, | |
| "loss": 0.2658, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 3.056279749867779e-05, | |
| "loss": 0.2744, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 3.0523908782627634e-05, | |
| "loss": 0.2711, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 3.0485020066577486e-05, | |
| "loss": 0.2668, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 3.0446131350527334e-05, | |
| "loss": 0.2706, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 3.040724263447718e-05, | |
| "loss": 0.2698, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 3.0368353918427027e-05, | |
| "loss": 0.2714, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 3.0329542979808977e-05, | |
| "loss": 0.2734, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 3.0290654263758828e-05, | |
| "loss": 0.2696, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 3.0251765547708676e-05, | |
| "loss": 0.2752, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 3.0212876831658528e-05, | |
| "loss": 0.2732, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 3.0173988115608376e-05, | |
| "loss": 0.2778, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 3.0135177176990326e-05, | |
| "loss": 0.2799, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_bleu": 59.2034, | |
| "eval_gen_len": 15.6702, | |
| "eval_loss": 0.7783872485160828, | |
| "eval_runtime": 6542.9489, | |
| "eval_samples_per_second": 13.131, | |
| "eval_steps_per_second": 1.641, | |
| "step": 258144 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 645360, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 15, | |
| "save_steps": 500, | |
| "total_flos": 8.950904749025133e+18, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |