| { | |
| "best_metric": 0.6568061113357544, | |
| "best_model_checkpoint": "./finetuned_model/checkpoint-129072", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 129072, | |
| "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 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 645360, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 15, | |
| "save_steps": 500, | |
| "total_flos": 4.4754523745125663e+18, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |