| { | |
| "best_metric": 0.6568061113357544, | |
| "best_model_checkpoint": "./finetuned_model/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": 9.96e-06, | |
| "loss": 2.2631, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.992e-05, | |
| "loss": 1.5976, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9920000000000005e-05, | |
| "loss": 1.4641, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.9920000000000004e-05, | |
| "loss": 1.3868, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.992e-05, | |
| "loss": 1.344, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.996142239367825e-05, | |
| "loss": 1.3138, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.9922533677628105e-05, | |
| "loss": 1.244, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.988364496157795e-05, | |
| "loss": 1.2047, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.98447562455278e-05, | |
| "loss": 1.1707, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.980586752947765e-05, | |
| "loss": 1.1331, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.97670565908596e-05, | |
| "loss": 1.1021, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.972816787480945e-05, | |
| "loss": 1.0923, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.968927915875929e-05, | |
| "loss": 1.0713, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.965039044270914e-05, | |
| "loss": 1.0609, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9611501726658995e-05, | |
| "loss": 1.043, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.957261301060884e-05, | |
| "loss": 1.0345, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.953380207199079e-05, | |
| "loss": 1.02, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.949491335594064e-05, | |
| "loss": 1.0057, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.945602463989049e-05, | |
| "loss": 0.9978, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.9417135923840344e-05, | |
| "loss": 0.9801, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.937824720779019e-05, | |
| "loss": 0.9752, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.933935849174004e-05, | |
| "loss": 0.9736, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.930046977568989e-05, | |
| "loss": 0.9519, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.9261581059639736e-05, | |
| "loss": 0.9559, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9222770121021686e-05, | |
| "loss": 0.936, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.9183959182403636e-05, | |
| "loss": 0.9276, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.914507046635349e-05, | |
| "loss": 0.9384, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.910618175030333e-05, | |
| "loss": 0.9347, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.9067293034253183e-05, | |
| "loss": 0.9182, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.902840431820303e-05, | |
| "loss": 0.9136, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.898951560215288e-05, | |
| "loss": 0.9096, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.895062688610273e-05, | |
| "loss": 0.8926, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.891181594748468e-05, | |
| "loss": 0.8946, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.887300500886663e-05, | |
| "loss": 0.8895, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.8834116292816475e-05, | |
| "loss": 0.8897, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.879522757676633e-05, | |
| "loss": 0.8901, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.875633886071618e-05, | |
| "loss": 0.8839, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.871745014466603e-05, | |
| "loss": 0.8737, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.867863920604798e-05, | |
| "loss": 0.8613, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.8639750489997824e-05, | |
| "loss": 0.8547, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.8600861773947676e-05, | |
| "loss": 0.8623, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.856197305789753e-05, | |
| "loss": 0.8615, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.852308434184737e-05, | |
| "loss": 0.8614, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.848419562579722e-05, | |
| "loss": 0.846, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.8445384687179167e-05, | |
| "loss": 0.8487, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.840649597112902e-05, | |
| "loss": 0.8426, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.836760725507887e-05, | |
| "loss": 0.8398, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.8328718539028714e-05, | |
| "loss": 0.8407, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.828998537784277e-05, | |
| "loss": 0.8438, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.825117443922472e-05, | |
| "loss": 0.8257, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.821228572317456e-05, | |
| "loss": 0.8328, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.8173397007124415e-05, | |
| "loss": 0.8229, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.8134508291074266e-05, | |
| "loss": 0.8249, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.809561957502412e-05, | |
| "loss": 0.8232, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.805673085897396e-05, | |
| "loss": 0.83, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.801791992035591e-05, | |
| "loss": 0.8339, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.7979031204305764e-05, | |
| "loss": 0.817, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.7940142488255615e-05, | |
| "loss": 0.8223, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.790125377220545e-05, | |
| "loss": 0.8041, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.7862365056155305e-05, | |
| "loss": 0.8126, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.7823476340105156e-05, | |
| "loss": 0.8039, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.778458762405501e-05, | |
| "loss": 0.8071, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.774569890800485e-05, | |
| "loss": 0.8065, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.7706810191954704e-05, | |
| "loss": 0.7998, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.7667921475904556e-05, | |
| "loss": 0.8034, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.76290327598544e-05, | |
| "loss": 0.8031, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.759014404380425e-05, | |
| "loss": 0.8057, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.7551255327754103e-05, | |
| "loss": 0.7905, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.751236661170395e-05, | |
| "loss": 0.794, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.74734778956538e-05, | |
| "loss": 0.7906, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.743458917960365e-05, | |
| "loss": 0.7865, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.73957004635535e-05, | |
| "loss": 0.7644, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.735681174750335e-05, | |
| "loss": 0.7737, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.73179230314532e-05, | |
| "loss": 0.7746, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.7279034315403044e-05, | |
| "loss": 0.7832, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.724014559935289e-05, | |
| "loss": 0.7706, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.720141243816694e-05, | |
| "loss": 0.7778, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.7162523722116795e-05, | |
| "loss": 0.7713, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.712363500606664e-05, | |
| "loss": 0.772, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.708474629001649e-05, | |
| "loss": 0.775, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.704593535139844e-05, | |
| "loss": 0.7718, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.700704663534829e-05, | |
| "loss": 0.7724, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.696815791929814e-05, | |
| "loss": 0.7629, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.692926920324799e-05, | |
| "loss": 0.7731, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.689038048719784e-05, | |
| "loss": 0.7597, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6851491771147685e-05, | |
| "loss": 0.7629, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_bleu": 54.9608, | |
| "eval_gen_len": 16.1273, | |
| "eval_loss": 0.7230631709098816, | |
| "eval_runtime": 7044.4442, | |
| "eval_samples_per_second": 12.196, | |
| "eval_steps_per_second": 1.525, | |
| "step": 43024 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.6812680832529634e-05, | |
| "loss": 0.6436, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6773869893911584e-05, | |
| "loss": 0.6446, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.6734981177861435e-05, | |
| "loss": 0.6489, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.669617023924338e-05, | |
| "loss": 0.6512, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.665728152319323e-05, | |
| "loss": 0.648, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.661839280714308e-05, | |
| "loss": 0.6426, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.657958186852503e-05, | |
| "loss": 0.6489, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.6540693152474876e-05, | |
| "loss": 0.6417, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.650180443642473e-05, | |
| "loss": 0.645, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.646291572037458e-05, | |
| "loss": 0.6543, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.642402700432443e-05, | |
| "loss": 0.6555, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.6385138288274275e-05, | |
| "loss": 0.6501, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.634624957222413e-05, | |
| "loss": 0.6447, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.630736085617398e-05, | |
| "loss": 0.6432, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.626847214012382e-05, | |
| "loss": 0.6516, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.6229583424073674e-05, | |
| "loss": 0.6517, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.6190694708023526e-05, | |
| "loss": 0.6534, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.615180599197337e-05, | |
| "loss": 0.6482, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.611291727592322e-05, | |
| "loss": 0.6504, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.607402855987307e-05, | |
| "loss": 0.6481, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.603513984382292e-05, | |
| "loss": 0.6444, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.599625112777276e-05, | |
| "loss": 0.6509, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.5957362411722615e-05, | |
| "loss": 0.6485, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.5918551473104564e-05, | |
| "loss": 0.6474, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.5879662757054416e-05, | |
| "loss": 0.6467, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.584077404100427e-05, | |
| "loss": 0.6442, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.580188532495411e-05, | |
| "loss": 0.6537, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.576307438633606e-05, | |
| "loss": 0.6563, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.5724185670285913e-05, | |
| "loss": 0.6587, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.5685296954235765e-05, | |
| "loss": 0.6535, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.564640823818561e-05, | |
| "loss": 0.6623, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.560759729956756e-05, | |
| "loss": 0.6458, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.556870858351741e-05, | |
| "loss": 0.6557, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.552981986746726e-05, | |
| "loss": 0.6458, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.549093115141711e-05, | |
| "loss": 0.6565, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.545204243536695e-05, | |
| "loss": 0.6521, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.54131537193168e-05, | |
| "loss": 0.6485, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.5374265003266655e-05, | |
| "loss": 0.6503, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.53353762872165e-05, | |
| "loss": 0.6483, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.529656534859845e-05, | |
| "loss": 0.6425, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.52576766325483e-05, | |
| "loss": 0.6566, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.521878791649815e-05, | |
| "loss": 0.6537, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.5179899200448e-05, | |
| "loss": 0.653, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.514116603926205e-05, | |
| "loss": 0.6448, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.51022773232119e-05, | |
| "loss": 0.6594, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.506338860716175e-05, | |
| "loss": 0.6551, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.50245776685437e-05, | |
| "loss": 0.6559, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.498568895249355e-05, | |
| "loss": 0.6506, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.49468002364434e-05, | |
| "loss": 0.6504, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.4907911520393245e-05, | |
| "loss": 0.6643, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.486902280434309e-05, | |
| "loss": 0.6435, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.483013408829294e-05, | |
| "loss": 0.6389, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.479124537224279e-05, | |
| "loss": 0.6495, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.475235665619264e-05, | |
| "loss": 0.6395, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.471346794014249e-05, | |
| "loss": 0.6404, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.467457922409234e-05, | |
| "loss": 0.6438, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.4635690508042186e-05, | |
| "loss": 0.6361, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.459680179199204e-05, | |
| "loss": 0.6486, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.455791307594189e-05, | |
| "loss": 0.6489, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.4519024359891734e-05, | |
| "loss": 0.6532, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.4480135643841585e-05, | |
| "loss": 0.6529, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.4441246927791437e-05, | |
| "loss": 0.6451, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.440235821174129e-05, | |
| "loss": 0.6443, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.436346949569113e-05, | |
| "loss": 0.6456, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.4324580779640984e-05, | |
| "loss": 0.6515, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.428569206359083e-05, | |
| "loss": 0.6425, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.424680334754068e-05, | |
| "loss": 0.6394, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.4207914631490525e-05, | |
| "loss": 0.6433, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.416902591544038e-05, | |
| "loss": 0.6459, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.4130214976822327e-05, | |
| "loss": 0.6421, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.409132626077218e-05, | |
| "loss": 0.6428, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.405251532215413e-05, | |
| "loss": 0.6406, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.401362660610397e-05, | |
| "loss": 0.6357, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.3974737890053824e-05, | |
| "loss": 0.6355, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.3935849174003676e-05, | |
| "loss": 0.6459, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.389696045795352e-05, | |
| "loss": 0.6422, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.385807174190337e-05, | |
| "loss": 0.6413, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.381918302585322e-05, | |
| "loss": 0.6391, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.3780294309803075e-05, | |
| "loss": 0.6409, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.374140559375292e-05, | |
| "loss": 0.632, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.370251687770277e-05, | |
| "loss": 0.6462, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.3663628161652616e-05, | |
| "loss": 0.6388, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.3624817223034566e-05, | |
| "loss": 0.642, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.358592850698442e-05, | |
| "loss": 0.6409, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.354703979093426e-05, | |
| "loss": 0.6406, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.350815107488411e-05, | |
| "loss": 0.6355, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bleu": 57.5735, | |
| "eval_gen_len": 15.9938, | |
| "eval_loss": 0.6641375422477722, | |
| "eval_runtime": 6884.7279, | |
| "eval_samples_per_second": 12.479, | |
| "eval_steps_per_second": 1.56, | |
| "step": 86048 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.3469262358833965e-05, | |
| "loss": 0.5199, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.343037364278381e-05, | |
| "loss": 0.4964, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.339148492673366e-05, | |
| "loss": 0.5025, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.335259621068351e-05, | |
| "loss": 0.508, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.331378527206546e-05, | |
| "loss": 0.5096, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.327497433344741e-05, | |
| "loss": 0.5106, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.323616339482936e-05, | |
| "loss": 0.5108, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.3197274678779206e-05, | |
| "loss": 0.5057, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.3158463740161156e-05, | |
| "loss": 0.5108, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.311957502411101e-05, | |
| "loss": 0.5044, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.308068630806085e-05, | |
| "loss": 0.5077, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.3041797592010704e-05, | |
| "loss": 0.5007, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.300290887596055e-05, | |
| "loss": 0.5068, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.29640201599104e-05, | |
| "loss": 0.5038, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.292513144386025e-05, | |
| "loss": 0.5107, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.28862427278101e-05, | |
| "loss": 0.5117, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.284750956662415e-05, | |
| "loss": 0.5129, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.2808620850574e-05, | |
| "loss": 0.5119, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.276973213452385e-05, | |
| "loss": 0.5198, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.27308434184737e-05, | |
| "loss": 0.5138, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.269195470242355e-05, | |
| "loss": 0.516, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.2653065986373395e-05, | |
| "loss": 0.521, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.2614177270323247e-05, | |
| "loss": 0.5162, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.25752885542731e-05, | |
| "loss": 0.5175, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.253639983822294e-05, | |
| "loss": 0.518, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.249758889960489e-05, | |
| "loss": 0.5234, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.245870018355474e-05, | |
| "loss": 0.524, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.241981146750459e-05, | |
| "loss": 0.5263, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.238092275145444e-05, | |
| "loss": 0.5171, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.2342034035404285e-05, | |
| "loss": 0.5265, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.2303145319354137e-05, | |
| "loss": 0.5156, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.226425660330399e-05, | |
| "loss": 0.5255, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.222536788725383e-05, | |
| "loss": 0.5187, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.218655694863579e-05, | |
| "loss": 0.5298, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 4.2147668232585634e-05, | |
| "loss": 0.5258, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.2108779516535485e-05, | |
| "loss": 0.5206, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.206989080048534e-05, | |
| "loss": 0.5292, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.203100208443518e-05, | |
| "loss": 0.5156, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 4.199211336838503e-05, | |
| "loss": 0.5156, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.1953224652334885e-05, | |
| "loss": 0.5259, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.191433593628473e-05, | |
| "loss": 0.5236, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 4.187552499766668e-05, | |
| "loss": 0.5282, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.1836636281616524e-05, | |
| "loss": 0.5342, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.1797747565566375e-05, | |
| "loss": 0.5246, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.175885884951623e-05, | |
| "loss": 0.5238, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.171997013346607e-05, | |
| "loss": 0.5329, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.168108141741592e-05, | |
| "loss": 0.5223, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.164234825622997e-05, | |
| "loss": 0.5333, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.160345954017982e-05, | |
| "loss": 0.5353, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.1564570824129674e-05, | |
| "loss": 0.5285, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.152568210807952e-05, | |
| "loss": 0.5365, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.1486948946893573e-05, | |
| "loss": 0.5214, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.1448060230843425e-05, | |
| "loss": 0.534, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.140917151479327e-05, | |
| "loss": 0.5375, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.137028279874312e-05, | |
| "loss": 0.532, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.133139408269297e-05, | |
| "loss": 0.5232, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.129250536664282e-05, | |
| "loss": 0.5333, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.125361665059266e-05, | |
| "loss": 0.5269, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.1214727934542514e-05, | |
| "loss": 0.5186, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.1175839218492365e-05, | |
| "loss": 0.5226, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.1137028279874315e-05, | |
| "loss": 0.5278, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.109813956382416e-05, | |
| "loss": 0.5311, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.105925084777401e-05, | |
| "loss": 0.5274, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.102036213172386e-05, | |
| "loss": 0.5348, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.098147341567371e-05, | |
| "loss": 0.5359, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.094258469962356e-05, | |
| "loss": 0.5416, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.090369598357341e-05, | |
| "loss": 0.526, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.086488504495536e-05, | |
| "loss": 0.5404, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.0825996328905205e-05, | |
| "loss": 0.5321, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.0787107612855056e-05, | |
| "loss": 0.5378, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.074821889680491e-05, | |
| "loss": 0.5371, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.070933018075476e-05, | |
| "loss": 0.5303, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.06704414647046e-05, | |
| "loss": 0.5321, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.063163052608655e-05, | |
| "loss": 0.5291, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.05927418100364e-05, | |
| "loss": 0.5391, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.055385309398625e-05, | |
| "loss": 0.5344, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.05149643779361e-05, | |
| "loss": 0.5381, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.0476075661885946e-05, | |
| "loss": 0.5348, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.04371869458358e-05, | |
| "loss": 0.5367, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.039829822978565e-05, | |
| "loss": 0.5294, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.0359409513735494e-05, | |
| "loss": 0.5257, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.0320520797685346e-05, | |
| "loss": 0.5241, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.02816320816352e-05, | |
| "loss": 0.5375, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.024274336558504e-05, | |
| "loss": 0.5332, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.0203854649534894e-05, | |
| "loss": 0.532, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.0164965933484745e-05, | |
| "loss": 0.5464, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bleu": 58.837, | |
| "eval_gen_len": 15.7689, | |
| "eval_loss": 0.6568061113357544, | |
| "eval_runtime": 6740.2004, | |
| "eval_samples_per_second": 12.747, | |
| "eval_steps_per_second": 1.593, | |
| "step": 129072 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.0126077217434597e-05, | |
| "loss": 0.4193, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.0087344056248644e-05, | |
| "loss": 0.399, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.004853311763059e-05, | |
| "loss": 0.3996, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.0009799956444635e-05, | |
| "loss": 0.3937, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.997091124039449e-05, | |
| "loss": 0.3955, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.993202252434434e-05, | |
| "loss": 0.3998, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.989313380829419e-05, | |
| "loss": 0.3934, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.9854245092244034e-05, | |
| "loss": 0.3987, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.9815434153625984e-05, | |
| "loss": 0.4048, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.9776545437575836e-05, | |
| "loss": 0.406, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.973765672152569e-05, | |
| "loss": 0.3987, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.969876800547553e-05, | |
| "loss": 0.4085, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.9659879289425383e-05, | |
| "loss": 0.4047, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.9620990573375235e-05, | |
| "loss": 0.4021, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.958210185732508e-05, | |
| "loss": 0.4056, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.954321314127493e-05, | |
| "loss": 0.4019, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.950440220265688e-05, | |
| "loss": 0.4125, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.9465513486606726e-05, | |
| "loss": 0.4087, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.942662477055658e-05, | |
| "loss": 0.3971, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.938773605450642e-05, | |
| "loss": 0.4122, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.9348847338456273e-05, | |
| "loss": 0.4086, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.931003639983822e-05, | |
| "loss": 0.4173, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.9271147683788075e-05, | |
| "loss": 0.4158, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.923225896773792e-05, | |
| "loss": 0.4072, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.919337025168777e-05, | |
| "loss": 0.4156, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.915448153563762e-05, | |
| "loss": 0.4174, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.9115592819587474e-05, | |
| "loss": 0.416, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.9076781880969424e-05, | |
| "loss": 0.4151, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.903789316491927e-05, | |
| "loss": 0.4188, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.899900444886912e-05, | |
| "loss": 0.419, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.896011573281897e-05, | |
| "loss": 0.4206, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.8921227016768816e-05, | |
| "loss": 0.4243, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.888233830071866e-05, | |
| "loss": 0.4131, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.884344958466851e-05, | |
| "loss": 0.4192, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.8804560868618364e-05, | |
| "loss": 0.4179, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.876567215256821e-05, | |
| "loss": 0.4264, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.872678343651806e-05, | |
| "loss": 0.4231, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.868789472046791e-05, | |
| "loss": 0.4217, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.8649006004417756e-05, | |
| "loss": 0.4189, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.861011728836761e-05, | |
| "loss": 0.4198, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.857122857231746e-05, | |
| "loss": 0.4237, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.853233985626731e-05, | |
| "loss": 0.4147, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.849352891764926e-05, | |
| "loss": 0.4219, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.8454640201599105e-05, | |
| "loss": 0.4274, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.8415829262981055e-05, | |
| "loss": 0.4283, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.8376940546930907e-05, | |
| "loss": 0.4222, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.833805183088076e-05, | |
| "loss": 0.4269, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 3.82991631148306e-05, | |
| "loss": 0.4271, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 3.826035217621255e-05, | |
| "loss": 0.425, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.82214634601624e-05, | |
| "loss": 0.4222, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.818257474411225e-05, | |
| "loss": 0.4236, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.81436860280621e-05, | |
| "loss": 0.4327, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 3.8104797312011945e-05, | |
| "loss": 0.4268, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 3.8065908595961797e-05, | |
| "loss": 0.4319, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.802701987991165e-05, | |
| "loss": 0.4312, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.798813116386149e-05, | |
| "loss": 0.4328, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.7949242447811344e-05, | |
| "loss": 0.4292, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.7910353731761196e-05, | |
| "loss": 0.4317, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.787146501571104e-05, | |
| "loss": 0.4309, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.7832654077093e-05, | |
| "loss": 0.4425, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.779376536104284e-05, | |
| "loss": 0.4335, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.775487664499269e-05, | |
| "loss": 0.4295, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 3.7715987928942545e-05, | |
| "loss": 0.4305, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.767709921289239e-05, | |
| "loss": 0.4264, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 3.7638210496842234e-05, | |
| "loss": 0.43, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.7599321780792086e-05, | |
| "loss": 0.4329, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 3.756043306474194e-05, | |
| "loss": 0.4324, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.752154434869178e-05, | |
| "loss": 0.434, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.7482655632641634e-05, | |
| "loss": 0.4317, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 3.744384469402358e-05, | |
| "loss": 0.4379, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.7404955977973435e-05, | |
| "loss": 0.4427, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 3.736606726192328e-05, | |
| "loss": 0.433, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.732717854587313e-05, | |
| "loss": 0.4407, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.728828982982298e-05, | |
| "loss": 0.4416, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.724940111377283e-05, | |
| "loss": 0.4407, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.721051239772268e-05, | |
| "loss": 0.4379, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.717170145910463e-05, | |
| "loss": 0.4345, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.713289052048658e-05, | |
| "loss": 0.435, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 3.709400180443643e-05, | |
| "loss": 0.4366, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.7055113088386275e-05, | |
| "loss": 0.4336, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.701622437233612e-05, | |
| "loss": 0.4438, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.697741343371807e-05, | |
| "loss": 0.4389, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 3.693852471766792e-05, | |
| "loss": 0.4394, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.689963600161777e-05, | |
| "loss": 0.4378, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.686082506299972e-05, | |
| "loss": 0.4443, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.682193634694957e-05, | |
| "loss": 0.4444, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bleu": 59.3055, | |
| "eval_gen_len": 15.7545, | |
| "eval_loss": 0.6776524782180786, | |
| "eval_runtime": 6664.893, | |
| "eval_samples_per_second": 12.891, | |
| "eval_steps_per_second": 1.611, | |
| "step": 172096 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.678304763089942e-05, | |
| "loss": 0.3253, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.674415891484927e-05, | |
| "loss": 0.302, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.670534797623122e-05, | |
| "loss": 0.3034, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.666645926018107e-05, | |
| "loss": 0.3075, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.6627570544130915e-05, | |
| "loss": 0.3039, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.6588759605512865e-05, | |
| "loss": 0.3102, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.6549870889462717e-05, | |
| "loss": 0.3005, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.651098217341257e-05, | |
| "loss": 0.3052, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.647217123479451e-05, | |
| "loss": 0.313, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.643328251874436e-05, | |
| "loss": 0.3103, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.639439380269421e-05, | |
| "loss": 0.3167, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.635550508664406e-05, | |
| "loss": 0.3129, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.631669414802601e-05, | |
| "loss": 0.3067, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.627780543197586e-05, | |
| "loss": 0.3091, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.623891671592571e-05, | |
| "loss": 0.3131, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.6200027999875556e-05, | |
| "loss": 0.3135, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.616113928382541e-05, | |
| "loss": 0.3133, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 3.612225056777526e-05, | |
| "loss": 0.3167, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.6083361851725104e-05, | |
| "loss": 0.3129, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 3.6044473135674956e-05, | |
| "loss": 0.3232, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.600558441962481e-05, | |
| "loss": 0.3161, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.596669570357465e-05, | |
| "loss": 0.3211, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.59278847649566e-05, | |
| "loss": 0.3248, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.588899604890645e-05, | |
| "loss": 0.3238, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 3.58501073328563e-05, | |
| "loss": 0.3209, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.581121861680615e-05, | |
| "loss": 0.318, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 3.5772329900755994e-05, | |
| "loss": 0.3255, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 3.5733441184705846e-05, | |
| "loss": 0.3265, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 3.56945524686557e-05, | |
| "loss": 0.3235, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.565574153003765e-05, | |
| "loss": 0.3304, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 3.561685281398749e-05, | |
| "loss": 0.3331, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.557796409793734e-05, | |
| "loss": 0.3254, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.5539075381887194e-05, | |
| "loss": 0.3305, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 3.5500186665837046e-05, | |
| "loss": 0.3264, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.546129794978689e-05, | |
| "loss": 0.3286, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 3.542240923373674e-05, | |
| "loss": 0.3303, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 3.5383520517686594e-05, | |
| "loss": 0.3332, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 3.534463180163644e-05, | |
| "loss": 0.3284, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 3.530574308558629e-05, | |
| "loss": 0.3297, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 3.526685436953614e-05, | |
| "loss": 0.3267, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 3.5227965653485986e-05, | |
| "loss": 0.3396, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 3.518907693743583e-05, | |
| "loss": 0.3321, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 3.515018822138568e-05, | |
| "loss": 0.3411, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 3.5111299505335534e-05, | |
| "loss": 0.3356, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 3.507241078928538e-05, | |
| "loss": 0.3367, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 3.503359985066733e-05, | |
| "loss": 0.3372, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 3.499471113461718e-05, | |
| "loss": 0.3316, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 3.495582241856703e-05, | |
| "loss": 0.3388, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 3.4916933702516876e-05, | |
| "loss": 0.3348, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 3.487804498646673e-05, | |
| "loss": 0.3382, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 3.483915627041658e-05, | |
| "loss": 0.3376, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 3.480026755436643e-05, | |
| "loss": 0.3353, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 3.4761378838316276e-05, | |
| "loss": 0.3437, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 3.472249012226613e-05, | |
| "loss": 0.339, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 3.468360140621598e-05, | |
| "loss": 0.343, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.464479046759792e-05, | |
| "loss": 0.3402, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 3.460590175154777e-05, | |
| "loss": 0.3476, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 3.456701303549762e-05, | |
| "loss": 0.3432, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.452812431944747e-05, | |
| "loss": 0.3435, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 3.448923560339732e-05, | |
| "loss": 0.3396, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 3.445042466477927e-05, | |
| "loss": 0.3421, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 3.4411535948729115e-05, | |
| "loss": 0.3405, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 3.437264723267897e-05, | |
| "loss": 0.3434, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 3.433375851662882e-05, | |
| "loss": 0.3465, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 3.429486980057866e-05, | |
| "loss": 0.3459, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 3.4255981084528515e-05, | |
| "loss": 0.3458, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 3.4217092368478366e-05, | |
| "loss": 0.3486, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 3.4178281429860316e-05, | |
| "loss": 0.3491, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 3.4139470491242265e-05, | |
| "loss": 0.3509, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 3.410058177519212e-05, | |
| "loss": 0.3475, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 3.406169305914196e-05, | |
| "loss": 0.3537, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 3.4022804343091806e-05, | |
| "loss": 0.3485, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 3.398391562704166e-05, | |
| "loss": 0.3402, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 3.394502691099151e-05, | |
| "loss": 0.3582, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 3.390629374980556e-05, | |
| "loss": 0.3509, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 3.386740503375541e-05, | |
| "loss": 0.3431, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 3.3828516317705254e-05, | |
| "loss": 0.3494, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 3.37897053790872e-05, | |
| "loss": 0.3514, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 3.3750816663037055e-05, | |
| "loss": 0.3555, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 3.3711927946986906e-05, | |
| "loss": 0.3513, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 3.367303923093675e-05, | |
| "loss": 0.3466, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 3.36341505148866e-05, | |
| "loss": 0.3509, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 3.3595261798836454e-05, | |
| "loss": 0.347, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.35563730827863e-05, | |
| "loss": 0.3511, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 3.351748436673615e-05, | |
| "loss": 0.3524, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.3478595650686e-05, | |
| "loss": 0.3458, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_bleu": 59.1341, | |
| "eval_gen_len": 15.7016, | |
| "eval_loss": 0.7214592099189758, | |
| "eval_runtime": 6616.423, | |
| "eval_samples_per_second": 12.985, | |
| "eval_steps_per_second": 1.623, | |
| "step": 215120 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.343970693463585e-05, | |
| "loss": 0.2535, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.3400895996017796e-05, | |
| "loss": 0.2198, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.336200727996764e-05, | |
| "loss": 0.2281, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.332311856391749e-05, | |
| "loss": 0.2251, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.3284229847867344e-05, | |
| "loss": 0.2282, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.3245341131817196e-05, | |
| "loss": 0.2293, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.320645241576704e-05, | |
| "loss": 0.2246, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.316764147714899e-05, | |
| "loss": 0.2282, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.312875276109884e-05, | |
| "loss": 0.2311, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 3.308986404504869e-05, | |
| "loss": 0.239, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 3.305113088386274e-05, | |
| "loss": 0.2323, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 3.301224216781259e-05, | |
| "loss": 0.2317, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.297343122919454e-05, | |
| "loss": 0.2324, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 3.293454251314439e-05, | |
| "loss": 0.2376, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 3.289565379709424e-05, | |
| "loss": 0.2441, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 3.285676508104409e-05, | |
| "loss": 0.24, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 3.2817876364993935e-05, | |
| "loss": 0.2411, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 3.277898764894378e-05, | |
| "loss": 0.2454, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 3.274009893289363e-05, | |
| "loss": 0.2455, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 3.270121021684348e-05, | |
| "loss": 0.2429, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 3.266232150079333e-05, | |
| "loss": 0.2461, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 3.262343278474318e-05, | |
| "loss": 0.2414, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 3.258454406869303e-05, | |
| "loss": 0.253, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 3.254573313007498e-05, | |
| "loss": 0.2481, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 3.250684441402483e-05, | |
| "loss": 0.2506, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 3.2467955697974676e-05, | |
| "loss": 0.249, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 3.242906698192453e-05, | |
| "loss": 0.2448, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 3.239017826587438e-05, | |
| "loss": 0.2472, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 3.2351289549824224e-05, | |
| "loss": 0.2516, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 3.2312400833774075e-05, | |
| "loss": 0.2564, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 3.227351211772393e-05, | |
| "loss": 0.248, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 3.223462340167377e-05, | |
| "loss": 0.2533, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 3.2195734685623616e-05, | |
| "loss": 0.2537, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 3.215684596957347e-05, | |
| "loss": 0.2544, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 3.211795725352332e-05, | |
| "loss": 0.2543, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 3.2079068537473164e-05, | |
| "loss": 0.2559, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 3.2040179821423016e-05, | |
| "loss": 0.2496, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 3.2001368882804965e-05, | |
| "loss": 0.2496, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 3.196248016675482e-05, | |
| "loss": 0.2536, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 3.192359145070467e-05, | |
| "loss": 0.256, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 3.188470273465451e-05, | |
| "loss": 0.2588, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 3.1845814018604365e-05, | |
| "loss": 0.2521, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 3.1806925302554216e-05, | |
| "loss": 0.2596, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 3.176803658650406e-05, | |
| "loss": 0.2565, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 3.172914787045391e-05, | |
| "loss": 0.2632, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 3.169033693183586e-05, | |
| "loss": 0.2642, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 3.1651448215785714e-05, | |
| "loss": 0.261, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 3.1612637277167657e-05, | |
| "loss": 0.2607, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 3.157374856111751e-05, | |
| "loss": 0.2626, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 3.153485984506735e-05, | |
| "loss": 0.2561, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 3.1495971129017204e-05, | |
| "loss": 0.258, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 3.1457082412967056e-05, | |
| "loss": 0.2593, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 3.14181936969169e-05, | |
| "loss": 0.265, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 3.137930498086675e-05, | |
| "loss": 0.2566, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 3.1340416264816604e-05, | |
| "loss": 0.2615, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 3.130160532619855e-05, | |
| "loss": 0.2621, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 3.12627943875805e-05, | |
| "loss": 0.2634, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 3.1223905671530355e-05, | |
| "loss": 0.2635, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 3.11850169554802e-05, | |
| "loss": 0.2719, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 3.114620601686215e-05, | |
| "loss": 0.2657, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 3.11073950782441e-05, | |
| "loss": 0.2674, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 3.106850636219395e-05, | |
| "loss": 0.2658, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 3.1029617646143795e-05, | |
| "loss": 0.2711, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 3.0990728930093646e-05, | |
| "loss": 0.2726, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 3.095184021404349e-05, | |
| "loss": 0.2653, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 3.091295149799334e-05, | |
| "loss": 0.2695, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 3.0874062781943194e-05, | |
| "loss": 0.276, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 3.083517406589304e-05, | |
| "loss": 0.2718, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 3.079628534984289e-05, | |
| "loss": 0.2687, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 3.075739663379274e-05, | |
| "loss": 0.2674, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 3.071850791774259e-05, | |
| "loss": 0.2713, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 3.067961920169244e-05, | |
| "loss": 0.2671, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 3.064073048564229e-05, | |
| "loss": 0.2721, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 3.0601841769592135e-05, | |
| "loss": 0.2647, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 3.0562953053541986e-05, | |
| "loss": 0.2723, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 3.052406433749184e-05, | |
| "loss": 0.2722, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 3.0485175621441686e-05, | |
| "loss": 0.2744, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 3.0446286905391537e-05, | |
| "loss": 0.2708, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 3.040739818934138e-05, | |
| "loss": 0.271, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 3.036850947329123e-05, | |
| "loss": 0.2686, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 3.0329620757241078e-05, | |
| "loss": 0.2703, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 3.0290732041190926e-05, | |
| "loss": 0.2786, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 3.0251843325140778e-05, | |
| "loss": 0.2687, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 3.0212954609090626e-05, | |
| "loss": 0.2727, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 3.0174065893040478e-05, | |
| "loss": 0.2726, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 3.0135177176990326e-05, | |
| "loss": 0.2702, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_bleu": 59.0721, | |
| "eval_gen_len": 15.6936, | |
| "eval_loss": 0.8013012409210205, | |
| "eval_runtime": 6628.012, | |
| "eval_samples_per_second": 12.962, | |
| "eval_steps_per_second": 1.62, | |
| "step": 258144 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "step": 258144, | |
| "total_flos": 8.950904749025133e+18, | |
| "train_loss": 0.5181208364993382, | |
| "train_runtime": 241803.971, | |
| "train_samples_per_second": 42.703, | |
| "train_steps_per_second": 2.669 | |
| } | |
| ], | |
| "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 | |
| } | |