diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,6422 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.9998824589585864, + "eval_steps": 500, + "global_step": 6380, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "ETA": 0.0, + "epoch": 0.0015672138855150257, + "fp16_scale": 512.0, + "global_step": 10, + "grad_norm": 44.605, + "learning_rate": 2.0833333333333333e-07, + "loss": 2.2146, + "step": 10 + }, + { + "ETA": 29.58, + "epoch": 0.0031344277710300514, + "fp16_scale": 512.0, + "global_step": 20, + "grad_norm": 18.403, + "learning_rate": 1.25e-06, + "loss": 1.7049, + "step": 20 + }, + { + "ETA": 29.6, + "epoch": 0.004701641656545077, + "fp16_scale": 512.0, + "global_step": 30, + "grad_norm": 7.563, + "learning_rate": 2.2916666666666666e-06, + "loss": 1.0428, + "step": 30 + }, + { + "ETA": 29.56, + "epoch": 0.006268855542060103, + "fp16_scale": 512.0, + "global_step": 40, + "grad_norm": 6.731, + "learning_rate": 3.3333333333333333e-06, + "loss": 1.0004, + "step": 40 + }, + { + "ETA": 29.5, + "epoch": 0.00783606942757513, + "fp16_scale": 512.0, + "global_step": 50, + "grad_norm": 8.091, + "learning_rate": 4.3750000000000005e-06, + "loss": 0.93, + "step": 50 + }, + { + "ETA": 29.47, + "epoch": 0.009403283313090155, + "fp16_scale": 512.0, + "global_step": 60, + "grad_norm": 6.857, + "learning_rate": 5.416666666666667e-06, + "loss": 0.9099, + "step": 60 + }, + { + "ETA": 29.41, + "epoch": 0.01097049719860518, + "fp16_scale": 512.0, + "global_step": 70, + "grad_norm": 4.815, + "learning_rate": 6.458333333333334e-06, + "loss": 0.8915, + "step": 70 + }, + { + "ETA": 29.36, + "epoch": 0.012537711084120206, + "fp16_scale": 512.0, + "global_step": 80, + "grad_norm": 4.539, + "learning_rate": 7.500000000000001e-06, + "loss": 0.8344, + "step": 80 + }, + { + "ETA": 29.3, + "epoch": 0.014104924969635231, + "fp16_scale": 512.0, + "global_step": 90, + "grad_norm": 5.123, + "learning_rate": 8.541666666666666e-06, + "loss": 0.8418, + "step": 90 + }, + { + "ETA": 29.25, + "epoch": 0.01567213885515026, + "fp16_scale": 512.0, + "global_step": 100, + "grad_norm": 5.271, + "learning_rate": 9.583333333333335e-06, + "loss": 0.7887, + "step": 100 + }, + { + "ETA": 29.2, + "epoch": 0.017239352740665282, + "fp16_scale": 512.0, + "global_step": 110, + "grad_norm": 5.762, + "learning_rate": 1.0625e-05, + "loss": 0.8702, + "step": 110 + }, + { + "ETA": 29.15, + "epoch": 0.01880656662618031, + "fp16_scale": 512.0, + "global_step": 120, + "grad_norm": 5.592, + "learning_rate": 1.1666666666666668e-05, + "loss": 0.8226, + "step": 120 + }, + { + "ETA": 29.1, + "epoch": 0.020373780511695333, + "fp16_scale": 512.0, + "global_step": 130, + "grad_norm": 4.187, + "learning_rate": 1.2708333333333333e-05, + "loss": 0.7926, + "step": 130 + }, + { + "ETA": 29.06, + "epoch": 0.02194099439721036, + "fp16_scale": 512.0, + "global_step": 140, + "grad_norm": 3.79, + "learning_rate": 1.375e-05, + "loss": 0.8084, + "step": 140 + }, + { + "ETA": 29.02, + "epoch": 0.023508208282725384, + "fp16_scale": 512.0, + "global_step": 150, + "grad_norm": 6.431, + "learning_rate": 1.479166666666667e-05, + "loss": 0.8406, + "step": 150 + }, + { + "ETA": 28.96, + "epoch": 0.02507542216824041, + "fp16_scale": 512.0, + "global_step": 160, + "grad_norm": 5.383, + "learning_rate": 1.5833333333333333e-05, + "loss": 0.8033, + "step": 160 + }, + { + "ETA": 28.93, + "epoch": 0.026642636053755435, + "fp16_scale": 512.0, + "global_step": 170, + "grad_norm": 3.825, + "learning_rate": 1.6875e-05, + "loss": 0.7992, + "step": 170 + }, + { + "ETA": 28.88, + "epoch": 0.028209849939270462, + "fp16_scale": 512.0, + "global_step": 180, + "grad_norm": 6.132, + "learning_rate": 1.7916666666666667e-05, + "loss": 0.831, + "step": 180 + }, + { + "ETA": 28.83, + "epoch": 0.029777063824785486, + "fp16_scale": 512.0, + "global_step": 190, + "grad_norm": 4.835, + "learning_rate": 1.8958333333333334e-05, + "loss": 0.8184, + "step": 190 + }, + { + "ETA": 28.79, + "epoch": 0.03134427771030052, + "fp16_scale": 512.0, + "global_step": 200, + "grad_norm": 6.233, + "learning_rate": 2e-05, + "loss": 0.8741, + "step": 200 + }, + { + "ETA": 28.74, + "epoch": 0.03291149159581554, + "fp16_scale": 512.0, + "global_step": 210, + "grad_norm": 3.753, + "learning_rate": 1.99998711251399e-05, + "loss": 0.8059, + "step": 210 + }, + { + "ETA": 28.69, + "epoch": 0.034478705481330564, + "fp16_scale": 512.0, + "global_step": 220, + "grad_norm": 3.802, + "learning_rate": 1.9999484503881354e-05, + "loss": 0.7927, + "step": 220 + }, + { + "ETA": 28.68, + "epoch": 0.03604591936684559, + "fp16_scale": 512.0, + "global_step": 230, + "grad_norm": 4.375, + "learning_rate": 1.99988401461895e-05, + "loss": 0.8292, + "step": 230 + }, + { + "ETA": 29.04, + "epoch": 0.03761313325236062, + "fp16_scale": 512.0, + "global_step": 240, + "grad_norm": 3.616, + "learning_rate": 1.9997938068672652e-05, + "loss": 0.822, + "step": 240 + }, + { + "ETA": 28.97, + "epoch": 0.03918034713787564, + "fp16_scale": 512.0, + "global_step": 250, + "grad_norm": 5.363, + "learning_rate": 1.9996778294581828e-05, + "loss": 0.8432, + "step": 250 + }, + { + "ETA": 28.91, + "epoch": 0.040747561023390666, + "fp16_scale": 512.0, + "global_step": 260, + "grad_norm": 4.838, + "learning_rate": 1.9995360853810172e-05, + "loss": 0.8458, + "step": 260 + }, + { + "ETA": 28.84, + "epoch": 0.04231477490890569, + "fp16_scale": 512.0, + "global_step": 270, + "grad_norm": 4.622, + "learning_rate": 1.9993685782892184e-05, + "loss": 0.8552, + "step": 270 + }, + { + "ETA": 28.78, + "epoch": 0.04388198879442072, + "fp16_scale": 512.0, + "global_step": 280, + "grad_norm": 4.467, + "learning_rate": 1.9991753125002766e-05, + "loss": 0.8134, + "step": 280 + }, + { + "ETA": 28.72, + "epoch": 0.04544920267993575, + "fp16_scale": 512.0, + "global_step": 290, + "grad_norm": 4.902, + "learning_rate": 1.998956292995612e-05, + "loss": 0.888, + "step": 290 + }, + { + "ETA": 28.66, + "epoch": 0.04701641656545077, + "fp16_scale": 512.0, + "global_step": 300, + "grad_norm": 8.023, + "learning_rate": 1.998711525420447e-05, + "loss": 0.8485, + "step": 300 + }, + { + "ETA": 28.6, + "epoch": 0.048583630450965795, + "fp16_scale": 512.0, + "global_step": 310, + "grad_norm": 6.401, + "learning_rate": 1.998441016083658e-05, + "loss": 0.8511, + "step": 310 + }, + { + "ETA": 28.54, + "epoch": 0.05015084433648082, + "fp16_scale": 512.0, + "global_step": 320, + "grad_norm": 4.692, + "learning_rate": 1.9981447719576163e-05, + "loss": 0.8344, + "step": 320 + }, + { + "ETA": 28.48, + "epoch": 0.05171805822199585, + "fp16_scale": 512.0, + "global_step": 330, + "grad_norm": 3.886, + "learning_rate": 1.9978228006780056e-05, + "loss": 0.7998, + "step": 330 + }, + { + "ETA": 28.42, + "epoch": 0.05328527210751087, + "fp16_scale": 512.0, + "global_step": 340, + "grad_norm": 3.093, + "learning_rate": 1.9974751105436266e-05, + "loss": 0.8319, + "step": 340 + }, + { + "ETA": 28.37, + "epoch": 0.0548524859930259, + "fp16_scale": 512.0, + "global_step": 350, + "grad_norm": 5.02, + "learning_rate": 1.9971017105161833e-05, + "loss": 0.8256, + "step": 350 + }, + { + "ETA": 28.31, + "epoch": 0.056419699878540924, + "fp16_scale": 512.0, + "global_step": 360, + "grad_norm": 3.542, + "learning_rate": 1.9967026102200503e-05, + "loss": 0.7992, + "step": 360 + }, + { + "ETA": 28.26, + "epoch": 0.05798691376405595, + "fp16_scale": 512.0, + "global_step": 370, + "grad_norm": 3.29, + "learning_rate": 1.9962778199420265e-05, + "loss": 0.8275, + "step": 370 + }, + { + "ETA": 28.2, + "epoch": 0.05955412764957097, + "fp16_scale": 512.0, + "global_step": 380, + "grad_norm": 4.119, + "learning_rate": 1.9958273506310703e-05, + "loss": 0.7866, + "step": 380 + }, + { + "ETA": 28.15, + "epoch": 0.061121341535086, + "fp16_scale": 512.0, + "global_step": 390, + "grad_norm": 3.126, + "learning_rate": 1.995351213898015e-05, + "loss": 0.7829, + "step": 390 + }, + { + "ETA": 28.09, + "epoch": 0.06268855542060103, + "fp16_scale": 512.0, + "global_step": 400, + "grad_norm": 4.012, + "learning_rate": 1.9948494220152714e-05, + "loss": 0.8086, + "step": 400 + }, + { + "ETA": 28.04, + "epoch": 0.06425576930611605, + "fp16_scale": 512.0, + "global_step": 410, + "grad_norm": 5.155, + "learning_rate": 1.9943219879165113e-05, + "loss": 0.791, + "step": 410 + }, + { + "ETA": 27.99, + "epoch": 0.06582298319163107, + "fp16_scale": 512.0, + "global_step": 420, + "grad_norm": 5.031, + "learning_rate": 1.9937689251963347e-05, + "loss": 0.7987, + "step": 420 + }, + { + "ETA": 27.94, + "epoch": 0.06739019707714611, + "fp16_scale": 512.0, + "global_step": 430, + "grad_norm": 5.236, + "learning_rate": 1.9931902481099163e-05, + "loss": 0.8278, + "step": 430 + }, + { + "ETA": 27.88, + "epoch": 0.06895741096266113, + "fp16_scale": 512.0, + "global_step": 440, + "grad_norm": 4.538, + "learning_rate": 1.992585971572643e-05, + "loss": 0.8123, + "step": 440 + }, + { + "ETA": 27.83, + "epoch": 0.07052462484817615, + "fp16_scale": 512.0, + "global_step": 450, + "grad_norm": 3.488, + "learning_rate": 1.9919561111597246e-05, + "loss": 0.7873, + "step": 450 + }, + { + "ETA": 27.78, + "epoch": 0.07209183873369118, + "fp16_scale": 512.0, + "global_step": 460, + "grad_norm": 4.643, + "learning_rate": 1.9913006831057967e-05, + "loss": 0.8107, + "step": 460 + }, + { + "ETA": 27.72, + "epoch": 0.0736590526192062, + "fp16_scale": 512.0, + "global_step": 470, + "grad_norm": 3.725, + "learning_rate": 1.990619704304499e-05, + "loss": 0.7758, + "step": 470 + }, + { + "ETA": 27.67, + "epoch": 0.07522626650472124, + "fp16_scale": 512.0, + "global_step": 480, + "grad_norm": 5.386, + "learning_rate": 1.98991319230804e-05, + "loss": 0.8376, + "step": 480 + }, + { + "ETA": 27.62, + "epoch": 0.07679348039023626, + "fp16_scale": 512.0, + "global_step": 490, + "grad_norm": 3.911, + "learning_rate": 1.989181165326747e-05, + "loss": 0.8483, + "step": 490 + }, + { + "ETA": 27.57, + "epoch": 0.07836069427575128, + "fp16_scale": 512.0, + "global_step": 500, + "grad_norm": 4.428, + "learning_rate": 1.988423642228596e-05, + "loss": 0.8042, + "step": 500 + }, + { + "ETA": 27.88, + "epoch": 0.07992790816126631, + "fp16_scale": 512.0, + "global_step": 510, + "grad_norm": 3.18, + "learning_rate": 1.9876406425387222e-05, + "loss": 0.7861, + "step": 510 + }, + { + "ETA": 27.82, + "epoch": 0.08149512204678133, + "fp16_scale": 512.0, + "global_step": 520, + "grad_norm": 4.472, + "learning_rate": 1.9868321864389216e-05, + "loss": 0.8244, + "step": 520 + }, + { + "ETA": 27.76, + "epoch": 0.08306233593229637, + "fp16_scale": 512.0, + "global_step": 530, + "grad_norm": 4.459, + "learning_rate": 1.9859982947671273e-05, + "loss": 0.8077, + "step": 530 + }, + { + "ETA": 27.7, + "epoch": 0.08462954981781139, + "fp16_scale": 512.0, + "global_step": 540, + "grad_norm": 3.43, + "learning_rate": 1.9851389890168738e-05, + "loss": 0.8235, + "step": 540 + }, + { + "ETA": 27.64, + "epoch": 0.0861967637033264, + "fp16_scale": 512.0, + "global_step": 550, + "grad_norm": 4.116, + "learning_rate": 1.984254291336743e-05, + "loss": 0.8453, + "step": 550 + }, + { + "ETA": 27.59, + "epoch": 0.08776397758884144, + "fp16_scale": 512.0, + "global_step": 560, + "grad_norm": 3.874, + "learning_rate": 1.9833442245297923e-05, + "loss": 0.8315, + "step": 560 + }, + { + "ETA": 27.53, + "epoch": 0.08933119147435646, + "fp16_scale": 512.0, + "global_step": 570, + "grad_norm": 4.0, + "learning_rate": 1.982408812052969e-05, + "loss": 0.8089, + "step": 570 + }, + { + "ETA": 27.47, + "epoch": 0.0908984053598715, + "fp16_scale": 512.0, + "global_step": 580, + "grad_norm": 3.54, + "learning_rate": 1.9814480780165026e-05, + "loss": 0.8185, + "step": 580 + }, + { + "ETA": 27.42, + "epoch": 0.09246561924538652, + "fp16_scale": 512.0, + "global_step": 590, + "grad_norm": 4.836, + "learning_rate": 1.980462047183287e-05, + "loss": 0.8105, + "step": 590 + }, + { + "ETA": 27.36, + "epoch": 0.09403283313090154, + "fp16_scale": 512.0, + "global_step": 600, + "grad_norm": 4.281, + "learning_rate": 1.9794507449682383e-05, + "loss": 0.778, + "step": 600 + }, + { + "ETA": 27.31, + "epoch": 0.09560004701641657, + "fp16_scale": 512.0, + "global_step": 610, + "grad_norm": 3.683, + "learning_rate": 1.9784141974376434e-05, + "loss": 0.7742, + "step": 610 + }, + { + "ETA": 27.25, + "epoch": 0.09716726090193159, + "fp16_scale": 512.0, + "global_step": 620, + "grad_norm": 4.075, + "learning_rate": 1.9773524313084857e-05, + "loss": 0.8509, + "step": 620 + }, + { + "ETA": 27.2, + "epoch": 0.09873447478744661, + "fp16_scale": 512.0, + "global_step": 630, + "grad_norm": 3.46, + "learning_rate": 1.9762654739477578e-05, + "loss": 0.7818, + "step": 630 + }, + { + "ETA": 27.14, + "epoch": 0.10030168867296164, + "fp16_scale": 512.0, + "global_step": 640, + "grad_norm": 3.695, + "learning_rate": 1.975153353371755e-05, + "loss": 0.8249, + "step": 640 + }, + { + "ETA": 27.09, + "epoch": 0.10186890255847666, + "fp16_scale": 512.0, + "global_step": 650, + "grad_norm": 4.314, + "learning_rate": 1.974016098245354e-05, + "loss": 0.7981, + "step": 650 + }, + { + "ETA": 27.04, + "epoch": 0.1034361164439917, + "fp16_scale": 512.0, + "global_step": 660, + "grad_norm": 3.079, + "learning_rate": 1.9728537378812738e-05, + "loss": 0.8327, + "step": 660 + }, + { + "ETA": 26.98, + "epoch": 0.10500333032950672, + "fp16_scale": 512.0, + "global_step": 670, + "grad_norm": 3.434, + "learning_rate": 1.9716663022393202e-05, + "loss": 0.7954, + "step": 670 + }, + { + "ETA": 26.93, + "epoch": 0.10657054421502174, + "fp16_scale": 512.0, + "global_step": 680, + "grad_norm": 5.473, + "learning_rate": 1.9704538219256143e-05, + "loss": 0.7967, + "step": 680 + }, + { + "ETA": 26.87, + "epoch": 0.10813775810053677, + "fp16_scale": 512.0, + "global_step": 690, + "grad_norm": 3.194, + "learning_rate": 1.9692163281918016e-05, + "loss": 0.8253, + "step": 690 + }, + { + "ETA": 26.82, + "epoch": 0.1097049719860518, + "fp16_scale": 512.0, + "global_step": 700, + "grad_norm": 3.881, + "learning_rate": 1.9679538529342487e-05, + "loss": 0.8237, + "step": 700 + }, + { + "ETA": 26.77, + "epoch": 0.11127218587156683, + "fp16_scale": 512.0, + "global_step": 710, + "grad_norm": 4.069, + "learning_rate": 1.9666664286932198e-05, + "loss": 0.8276, + "step": 710 + }, + { + "ETA": 26.72, + "epoch": 0.11283939975708185, + "fp16_scale": 512.0, + "global_step": 720, + "grad_norm": 3.983, + "learning_rate": 1.9653540886520387e-05, + "loss": 0.8102, + "step": 720 + }, + { + "ETA": 26.66, + "epoch": 0.11440661364259687, + "fp16_scale": 512.0, + "global_step": 730, + "grad_norm": 3.559, + "learning_rate": 1.964016866636234e-05, + "loss": 0.7812, + "step": 730 + }, + { + "ETA": 26.61, + "epoch": 0.1159738275281119, + "fp16_scale": 512.0, + "global_step": 740, + "grad_norm": 3.652, + "learning_rate": 1.9626547971126646e-05, + "loss": 0.7735, + "step": 740 + }, + { + "ETA": 26.56, + "epoch": 0.11754104141362692, + "fp16_scale": 512.0, + "global_step": 750, + "grad_norm": 4.565, + "learning_rate": 1.9612679151886352e-05, + "loss": 0.7897, + "step": 750 + }, + { + "ETA": 26.51, + "epoch": 0.11910825529914194, + "fp16_scale": 512.0, + "global_step": 760, + "grad_norm": 3.597, + "learning_rate": 1.959856256610988e-05, + "loss": 0.813, + "step": 760 + }, + { + "ETA": 26.46, + "epoch": 0.12067546918465698, + "fp16_scale": 512.0, + "global_step": 770, + "grad_norm": 3.112, + "learning_rate": 1.958419857765184e-05, + "loss": 0.8063, + "step": 770 + }, + { + "ETA": 26.43, + "epoch": 0.122242683070172, + "fp16_scale": 512.0, + "global_step": 780, + "grad_norm": 3.769, + "learning_rate": 1.9569587556743627e-05, + "loss": 0.7885, + "step": 780 + }, + { + "ETA": 26.51, + "epoch": 0.12380989695568703, + "fp16_scale": 512.0, + "global_step": 790, + "grad_norm": 4.049, + "learning_rate": 1.95547298799839e-05, + "loss": 0.7815, + "step": 790 + }, + { + "ETA": 26.46, + "epoch": 0.12537711084120207, + "fp16_scale": 512.0, + "global_step": 800, + "grad_norm": 3.712, + "learning_rate": 1.953962593032886e-05, + "loss": 0.7915, + "step": 800 + }, + { + "ETA": 26.41, + "epoch": 0.12694432472671707, + "fp16_scale": 512.0, + "global_step": 810, + "grad_norm": 3.061, + "learning_rate": 1.9524276097082383e-05, + "loss": 0.7629, + "step": 810 + }, + { + "ETA": 26.36, + "epoch": 0.1285115386122321, + "fp16_scale": 512.0, + "global_step": 820, + "grad_norm": 3.61, + "learning_rate": 1.9508680775886e-05, + "loss": 0.824, + "step": 820 + }, + { + "ETA": 26.3, + "epoch": 0.13007875249774714, + "fp16_scale": 512.0, + "global_step": 830, + "grad_norm": 3.23, + "learning_rate": 1.9492840368708668e-05, + "loss": 0.8129, + "step": 830 + }, + { + "ETA": 26.25, + "epoch": 0.13164596638326215, + "fp16_scale": 512.0, + "global_step": 840, + "grad_norm": 3.774, + "learning_rate": 1.9476755283836448e-05, + "loss": 0.7996, + "step": 840 + }, + { + "ETA": 26.2, + "epoch": 0.13321318026877718, + "fp16_scale": 512.0, + "global_step": 850, + "grad_norm": 3.48, + "learning_rate": 1.946042593586195e-05, + "loss": 0.8317, + "step": 850 + }, + { + "ETA": 26.14, + "epoch": 0.13478039415429222, + "fp16_scale": 512.0, + "global_step": 860, + "grad_norm": 4.229, + "learning_rate": 1.944385274567366e-05, + "loss": 0.8057, + "step": 860 + }, + { + "ETA": 26.09, + "epoch": 0.13634760803980722, + "fp16_scale": 512.0, + "global_step": 870, + "grad_norm": 3.373, + "learning_rate": 1.9427036140445087e-05, + "loss": 0.7926, + "step": 870 + }, + { + "ETA": 26.04, + "epoch": 0.13791482192532226, + "fp16_scale": 512.0, + "global_step": 880, + "grad_norm": 2.887, + "learning_rate": 1.9409976553623767e-05, + "loss": 0.7527, + "step": 880 + }, + { + "ETA": 25.99, + "epoch": 0.1394820358108373, + "fp16_scale": 512.0, + "global_step": 890, + "grad_norm": 3.633, + "learning_rate": 1.939267442492007e-05, + "loss": 0.8175, + "step": 890 + }, + { + "ETA": 25.93, + "epoch": 0.1410492496963523, + "fp16_scale": 512.0, + "global_step": 900, + "grad_norm": 3.6, + "learning_rate": 1.937513020029588e-05, + "loss": 0.7589, + "step": 900 + }, + { + "ETA": 25.88, + "epoch": 0.14261646358186733, + "fp16_scale": 512.0, + "global_step": 910, + "grad_norm": 9.442, + "learning_rate": 1.9357344331953095e-05, + "loss": 0.8359, + "step": 910 + }, + { + "ETA": 25.83, + "epoch": 0.14418367746738237, + "fp16_scale": 512.0, + "global_step": 920, + "grad_norm": 3.371, + "learning_rate": 1.9339317278321975e-05, + "loss": 0.7368, + "step": 920 + }, + { + "ETA": 25.78, + "epoch": 0.1457508913528974, + "fp16_scale": 512.0, + "global_step": 930, + "grad_norm": 3.693, + "learning_rate": 1.9321049504049325e-05, + "loss": 0.7861, + "step": 930 + }, + { + "ETA": 25.73, + "epoch": 0.1473181052384124, + "fp16_scale": 512.0, + "global_step": 940, + "grad_norm": 5.654, + "learning_rate": 1.930254147998651e-05, + "loss": 0.7698, + "step": 940 + }, + { + "ETA": 25.68, + "epoch": 0.14888531912392744, + "fp16_scale": 512.0, + "global_step": 950, + "grad_norm": 3.002, + "learning_rate": 1.9283793683177335e-05, + "loss": 0.802, + "step": 950 + }, + { + "ETA": 25.62, + "epoch": 0.15045253300944247, + "fp16_scale": 512.0, + "global_step": 960, + "grad_norm": 3.35, + "learning_rate": 1.9264806596845742e-05, + "loss": 0.7779, + "step": 960 + }, + { + "ETA": 25.57, + "epoch": 0.15201974689495748, + "fp16_scale": 512.0, + "global_step": 970, + "grad_norm": 2.751, + "learning_rate": 1.9245580710383344e-05, + "loss": 0.7804, + "step": 970 + }, + { + "ETA": 25.52, + "epoch": 0.15358696078047251, + "fp16_scale": 512.0, + "global_step": 980, + "grad_norm": 3.66, + "learning_rate": 1.922611651933683e-05, + "loss": 0.8006, + "step": 980 + }, + { + "ETA": 25.47, + "epoch": 0.15515417466598755, + "fp16_scale": 512.0, + "global_step": 990, + "grad_norm": 3.725, + "learning_rate": 1.920641452539518e-05, + "loss": 0.7695, + "step": 990 + }, + { + "ETA": 25.42, + "epoch": 0.15672138855150256, + "fp16_scale": 512.0, + "global_step": 1000, + "grad_norm": 3.159, + "learning_rate": 1.9186475236376733e-05, + "loss": 0.7952, + "step": 1000 + }, + { + "ETA": 25.53, + "epoch": 0.1582886024370176, + "fp16_scale": 1024.0, + "global_step": 1010, + "grad_norm": 1.366, + "learning_rate": 1.916629916621611e-05, + "loss": 0.7893, + "step": 1010 + }, + { + "ETA": 25.48, + "epoch": 0.15985581632253262, + "fp16_scale": 1024.0, + "global_step": 1020, + "grad_norm": 3.977, + "learning_rate": 1.914588683495095e-05, + "loss": 0.8, + "step": 1020 + }, + { + "ETA": 25.42, + "epoch": 0.16142303020804763, + "fp16_scale": 1024.0, + "global_step": 1030, + "grad_norm": 2.984, + "learning_rate": 1.9125238768708527e-05, + "loss": 0.7915, + "step": 1030 + }, + { + "ETA": 25.37, + "epoch": 0.16299024409356266, + "fp16_scale": 1024.0, + "global_step": 1040, + "grad_norm": 3.751, + "learning_rate": 1.9104355499692166e-05, + "loss": 0.7921, + "step": 1040 + }, + { + "ETA": 25.32, + "epoch": 0.1645574579790777, + "fp16_scale": 1024.0, + "global_step": 1050, + "grad_norm": 4.355, + "learning_rate": 1.908323756616754e-05, + "loss": 0.8083, + "step": 1050 + }, + { + "ETA": 25.27, + "epoch": 0.16612467186459273, + "fp16_scale": 1024.0, + "global_step": 1060, + "grad_norm": 3.678, + "learning_rate": 1.9061885512448797e-05, + "loss": 0.8066, + "step": 1060 + }, + { + "ETA": 25.21, + "epoch": 0.16769188575010774, + "fp16_scale": 1024.0, + "global_step": 1070, + "grad_norm": 3.258, + "learning_rate": 1.904029988888453e-05, + "loss": 0.7434, + "step": 1070 + }, + { + "ETA": 25.16, + "epoch": 0.16925909963562277, + "fp16_scale": 1024.0, + "global_step": 1080, + "grad_norm": 3.67, + "learning_rate": 1.901848125184357e-05, + "loss": 0.7847, + "step": 1080 + }, + { + "ETA": 25.11, + "epoch": 0.1708263135211378, + "fp16_scale": 1024.0, + "global_step": 1090, + "grad_norm": 3.202, + "learning_rate": 1.8996430163700686e-05, + "loss": 0.7697, + "step": 1090 + }, + { + "ETA": 25.06, + "epoch": 0.1723935274066528, + "fp16_scale": 1024.0, + "global_step": 1100, + "grad_norm": 3.825, + "learning_rate": 1.8974147192822053e-05, + "loss": 0.7835, + "step": 1100 + }, + { + "ETA": 25.01, + "epoch": 0.17396074129216785, + "fp16_scale": 1024.0, + "global_step": 1110, + "grad_norm": 3.314, + "learning_rate": 1.8951632913550625e-05, + "loss": 0.7981, + "step": 1110 + }, + { + "ETA": 24.96, + "epoch": 0.17552795517768288, + "fp16_scale": 1024.0, + "global_step": 1120, + "grad_norm": 5.387, + "learning_rate": 1.892888790619132e-05, + "loss": 0.7963, + "step": 1120 + }, + { + "ETA": 24.9, + "epoch": 0.1770951690631979, + "fp16_scale": 1024.0, + "global_step": 1130, + "grad_norm": 3.251, + "learning_rate": 1.890591275699606e-05, + "loss": 0.8005, + "step": 1130 + }, + { + "ETA": 24.85, + "epoch": 0.17866238294871292, + "fp16_scale": 1024.0, + "global_step": 1140, + "grad_norm": 4.961, + "learning_rate": 1.8882708058148683e-05, + "loss": 0.7651, + "step": 1140 + }, + { + "ETA": 24.8, + "epoch": 0.18022959683422796, + "fp16_scale": 1024.0, + "global_step": 1150, + "grad_norm": 3.24, + "learning_rate": 1.8859274407749646e-05, + "loss": 0.8029, + "step": 1150 + }, + { + "ETA": 24.75, + "epoch": 0.181796810719743, + "fp16_scale": 1024.0, + "global_step": 1160, + "grad_norm": 4.009, + "learning_rate": 1.8835612409800634e-05, + "loss": 0.7887, + "step": 1160 + }, + { + "ETA": 24.7, + "epoch": 0.183364024605258, + "fp16_scale": 1024.0, + "global_step": 1170, + "grad_norm": 3.594, + "learning_rate": 1.881172267418898e-05, + "loss": 0.7834, + "step": 1170 + }, + { + "ETA": 24.65, + "epoch": 0.18493123849077303, + "fp16_scale": 1024.0, + "global_step": 1180, + "grad_norm": 2.744, + "learning_rate": 1.8787605816671956e-05, + "loss": 0.7694, + "step": 1180 + }, + { + "ETA": 24.59, + "epoch": 0.18649845237628807, + "fp16_scale": 1024.0, + "global_step": 1190, + "grad_norm": 3.546, + "learning_rate": 1.876326245886088e-05, + "loss": 0.791, + "step": 1190 + }, + { + "ETA": 24.54, + "epoch": 0.18806566626180307, + "fp16_scale": 1024.0, + "global_step": 1200, + "grad_norm": 3.565, + "learning_rate": 1.873869322820513e-05, + "loss": 0.7762, + "step": 1200 + }, + { + "ETA": 24.49, + "epoch": 0.1896328801473181, + "fp16_scale": 1024.0, + "global_step": 1210, + "grad_norm": 2.926, + "learning_rate": 1.8713898757975935e-05, + "loss": 0.7669, + "step": 1210 + }, + { + "ETA": 24.44, + "epoch": 0.19120009403283314, + "fp16_scale": 1024.0, + "global_step": 1220, + "grad_norm": 4.165, + "learning_rate": 1.8688879687250067e-05, + "loss": 0.7813, + "step": 1220 + }, + { + "ETA": 24.39, + "epoch": 0.19276730791834815, + "fp16_scale": 1024.0, + "global_step": 1230, + "grad_norm": 2.299, + "learning_rate": 1.8663636660893378e-05, + "loss": 0.7547, + "step": 1230 + }, + { + "ETA": 24.34, + "epoch": 0.19433452180386318, + "fp16_scale": 1024.0, + "global_step": 1240, + "grad_norm": 3.092, + "learning_rate": 1.8638170329544164e-05, + "loss": 0.7791, + "step": 1240 + }, + { + "ETA": 24.29, + "epoch": 0.19590173568937821, + "fp16_scale": 1024.0, + "global_step": 1250, + "grad_norm": 3.193, + "learning_rate": 1.8612481349596406e-05, + "loss": 0.7505, + "step": 1250 + }, + { + "ETA": 24.24, + "epoch": 0.19746894957489322, + "fp16_scale": 1024.0, + "global_step": 1260, + "grad_norm": 3.13, + "learning_rate": 1.858657038318284e-05, + "loss": 0.7813, + "step": 1260 + }, + { + "ETA": 24.19, + "epoch": 0.19903616346040826, + "fp16_scale": 1024.0, + "global_step": 1270, + "grad_norm": 3.3, + "learning_rate": 1.85604380981579e-05, + "loss": 0.793, + "step": 1270 + }, + { + "ETA": 24.14, + "epoch": 0.2006033773459233, + "fp16_scale": 1024.0, + "global_step": 1280, + "grad_norm": 2.819, + "learning_rate": 1.8534085168080503e-05, + "loss": 0.7282, + "step": 1280 + }, + { + "ETA": 24.09, + "epoch": 0.20217059123143832, + "fp16_scale": 1024.0, + "global_step": 1290, + "grad_norm": 3.958, + "learning_rate": 1.850751227219669e-05, + "loss": 0.7718, + "step": 1290 + }, + { + "ETA": 24.04, + "epoch": 0.20373780511695333, + "fp16_scale": 1024.0, + "global_step": 1300, + "grad_norm": 3.133, + "learning_rate": 1.8480720095422096e-05, + "loss": 0.7958, + "step": 1300 + }, + { + "ETA": 23.99, + "epoch": 0.20530501900246836, + "fp16_scale": 1024.0, + "global_step": 1310, + "grad_norm": 3.184, + "learning_rate": 1.8453709328324337e-05, + "loss": 0.7885, + "step": 1310 + }, + { + "ETA": 23.94, + "epoch": 0.2068722328879834, + "fp16_scale": 1024.0, + "global_step": 1320, + "grad_norm": 2.849, + "learning_rate": 1.8426480667105178e-05, + "loss": 0.7554, + "step": 1320 + }, + { + "ETA": 23.9, + "epoch": 0.2084394467734984, + "fp16_scale": 1024.0, + "global_step": 1330, + "grad_norm": 3.621, + "learning_rate": 1.83990348135826e-05, + "loss": 0.7513, + "step": 1330 + }, + { + "ETA": 23.91, + "epoch": 0.21000666065901344, + "fp16_scale": 1024.0, + "global_step": 1340, + "grad_norm": 2.766, + "learning_rate": 1.8371372475172705e-05, + "loss": 0.7806, + "step": 1340 + }, + { + "ETA": 23.87, + "epoch": 0.21157387454452847, + "fp16_scale": 1024.0, + "global_step": 1350, + "grad_norm": 2.369, + "learning_rate": 1.8343494364871502e-05, + "loss": 0.7946, + "step": 1350 + }, + { + "ETA": 23.82, + "epoch": 0.21314108843004348, + "fp16_scale": 1024.0, + "global_step": 1360, + "grad_norm": 3.318, + "learning_rate": 1.8315401201236492e-05, + "loss": 0.7845, + "step": 1360 + }, + { + "ETA": 23.77, + "epoch": 0.2147083023155585, + "fp16_scale": 1024.0, + "global_step": 1370, + "grad_norm": 3.702, + "learning_rate": 1.828709370836819e-05, + "loss": 0.7898, + "step": 1370 + }, + { + "ETA": 23.72, + "epoch": 0.21627551620107355, + "fp16_scale": 1024.0, + "global_step": 1380, + "grad_norm": 2.894, + "learning_rate": 1.8258572615891427e-05, + "loss": 0.7749, + "step": 1380 + }, + { + "ETA": 23.67, + "epoch": 0.21784273008658855, + "fp16_scale": 1024.0, + "global_step": 1390, + "grad_norm": 3.881, + "learning_rate": 1.8229838658936566e-05, + "loss": 0.7623, + "step": 1390 + }, + { + "ETA": 23.62, + "epoch": 0.2194099439721036, + "fp16_scale": 1024.0, + "global_step": 1400, + "grad_norm": 3.222, + "learning_rate": 1.8200892578120544e-05, + "loss": 0.7537, + "step": 1400 + }, + { + "ETA": 23.57, + "epoch": 0.22097715785761862, + "fp16_scale": 1024.0, + "global_step": 1410, + "grad_norm": 2.809, + "learning_rate": 1.8171735119527784e-05, + "loss": 0.7634, + "step": 1410 + }, + { + "ETA": 23.52, + "epoch": 0.22254437174313366, + "fp16_scale": 1024.0, + "global_step": 1420, + "grad_norm": 2.92, + "learning_rate": 1.8142367034690967e-05, + "loss": 0.7852, + "step": 1420 + }, + { + "ETA": 23.47, + "epoch": 0.22411158562864866, + "fp16_scale": 1024.0, + "global_step": 1430, + "grad_norm": 2.293, + "learning_rate": 1.8112789080571655e-05, + "loss": 0.7808, + "step": 1430 + }, + { + "ETA": 23.42, + "epoch": 0.2256787995141637, + "fp16_scale": 1024.0, + "global_step": 1440, + "grad_norm": 3.209, + "learning_rate": 1.8083002019540784e-05, + "loss": 0.7904, + "step": 1440 + }, + { + "ETA": 23.37, + "epoch": 0.22724601339967873, + "fp16_scale": 1024.0, + "global_step": 1450, + "grad_norm": 2.592, + "learning_rate": 1.805300661935903e-05, + "loss": 0.7706, + "step": 1450 + }, + { + "ETA": 23.32, + "epoch": 0.22881322728519374, + "fp16_scale": 1024.0, + "global_step": 1460, + "grad_norm": 2.165, + "learning_rate": 1.8022803653156983e-05, + "loss": 0.7777, + "step": 1460 + }, + { + "ETA": 23.27, + "epoch": 0.23038044117070877, + "fp16_scale": 1024.0, + "global_step": 1470, + "grad_norm": 3.106, + "learning_rate": 1.799239389941526e-05, + "loss": 0.7475, + "step": 1470 + }, + { + "ETA": 23.22, + "epoch": 0.2319476550562238, + "fp16_scale": 1024.0, + "global_step": 1480, + "grad_norm": 3.043, + "learning_rate": 1.7961778141944407e-05, + "loss": 0.7672, + "step": 1480 + }, + { + "ETA": 23.17, + "epoch": 0.2335148689417388, + "fp16_scale": 1024.0, + "global_step": 1490, + "grad_norm": 2.615, + "learning_rate": 1.7930957169864713e-05, + "loss": 0.753, + "step": 1490 + }, + { + "ETA": 23.12, + "epoch": 0.23508208282725385, + "fp16_scale": 1024.0, + "global_step": 1500, + "grad_norm": 3.227, + "learning_rate": 1.789993177758588e-05, + "loss": 0.7833, + "step": 1500 + }, + { + "ETA": 23.17, + "epoch": 0.23664929671276888, + "fp16_scale": 1024.0, + "global_step": 1510, + "grad_norm": 3.618, + "learning_rate": 1.7868702764786522e-05, + "loss": 0.7465, + "step": 1510 + }, + { + "ETA": 23.12, + "epoch": 0.2382165105982839, + "fp16_scale": 1024.0, + "global_step": 1520, + "grad_norm": 3.637, + "learning_rate": 1.783727093639357e-05, + "loss": 0.7727, + "step": 1520 + }, + { + "ETA": 23.07, + "epoch": 0.23978372448379892, + "fp16_scale": 1024.0, + "global_step": 1530, + "grad_norm": 2.791, + "learning_rate": 1.7805637102561516e-05, + "loss": 0.7736, + "step": 1530 + }, + { + "ETA": 23.02, + "epoch": 0.24135093836931396, + "fp16_scale": 1024.0, + "global_step": 1540, + "grad_norm": 4.011, + "learning_rate": 1.777380207865155e-05, + "loss": 0.7794, + "step": 1540 + }, + { + "ETA": 22.97, + "epoch": 0.242918152254829, + "fp16_scale": 1024.0, + "global_step": 1550, + "grad_norm": 3.216, + "learning_rate": 1.7741766685210522e-05, + "loss": 0.7829, + "step": 1550 + }, + { + "ETA": 22.92, + "epoch": 0.244485366140344, + "fp16_scale": 1024.0, + "global_step": 1560, + "grad_norm": 3.06, + "learning_rate": 1.7709531747949796e-05, + "loss": 0.7472, + "step": 1560 + }, + { + "ETA": 22.87, + "epoch": 0.24605258002585903, + "fp16_scale": 1024.0, + "global_step": 1570, + "grad_norm": 2.834, + "learning_rate": 1.7677098097723985e-05, + "loss": 0.7557, + "step": 1570 + }, + { + "ETA": 22.82, + "epoch": 0.24761979391137406, + "fp16_scale": 1024.0, + "global_step": 1580, + "grad_norm": 2.833, + "learning_rate": 1.7644466570509508e-05, + "loss": 0.7649, + "step": 1580 + }, + { + "ETA": 22.76, + "epoch": 0.24918700779688907, + "fp16_scale": 1024.0, + "global_step": 1590, + "grad_norm": 2.454, + "learning_rate": 1.761163800738307e-05, + "loss": 0.7683, + "step": 1590 + }, + { + "ETA": 22.71, + "epoch": 0.25075422168240413, + "fp16_scale": 1024.0, + "global_step": 1600, + "grad_norm": 2.78, + "learning_rate": 1.757861325449997e-05, + "loss": 0.7333, + "step": 1600 + }, + { + "ETA": 22.66, + "epoch": 0.25232143556791914, + "fp16_scale": 1024.0, + "global_step": 1610, + "grad_norm": 2.601, + "learning_rate": 1.7545393163072285e-05, + "loss": 0.7917, + "step": 1610 + }, + { + "ETA": 22.61, + "epoch": 0.25388864945343415, + "fp16_scale": 1024.0, + "global_step": 1620, + "grad_norm": 3.483, + "learning_rate": 1.751197858934694e-05, + "loss": 0.77, + "step": 1620 + }, + { + "ETA": 22.56, + "epoch": 0.2554558633389492, + "fp16_scale": 1024.0, + "global_step": 1630, + "grad_norm": 2.619, + "learning_rate": 1.7478370394583647e-05, + "loss": 0.7957, + "step": 1630 + }, + { + "ETA": 22.51, + "epoch": 0.2570230772244642, + "fp16_scale": 1024.0, + "global_step": 1640, + "grad_norm": 2.546, + "learning_rate": 1.7444569445032677e-05, + "loss": 0.7811, + "step": 1640 + }, + { + "ETA": 22.46, + "epoch": 0.2585902911099792, + "fp16_scale": 1024.0, + "global_step": 1650, + "grad_norm": 4.994, + "learning_rate": 1.7410576611912563e-05, + "loss": 0.7777, + "step": 1650 + }, + { + "ETA": 22.41, + "epoch": 0.2601575049954943, + "fp16_scale": 1024.0, + "global_step": 1660, + "grad_norm": 2.53, + "learning_rate": 1.7376392771387623e-05, + "loss": 0.7764, + "step": 1660 + }, + { + "ETA": 22.36, + "epoch": 0.2617247188810093, + "fp16_scale": 1024.0, + "global_step": 1670, + "grad_norm": 3.218, + "learning_rate": 1.73420188045454e-05, + "loss": 0.7549, + "step": 1670 + }, + { + "ETA": 22.31, + "epoch": 0.2632919327665243, + "fp16_scale": 1024.0, + "global_step": 1680, + "grad_norm": 2.76, + "learning_rate": 1.7307455597373916e-05, + "loss": 0.7576, + "step": 1680 + }, + { + "ETA": 22.26, + "epoch": 0.26485914665203936, + "fp16_scale": 1024.0, + "global_step": 1690, + "grad_norm": 3.132, + "learning_rate": 1.7272704040738875e-05, + "loss": 0.7537, + "step": 1690 + }, + { + "ETA": 22.21, + "epoch": 0.26642636053755436, + "fp16_scale": 1024.0, + "global_step": 1700, + "grad_norm": 3.045, + "learning_rate": 1.723776503036068e-05, + "loss": 0.7826, + "step": 1700 + }, + { + "ETA": 22.16, + "epoch": 0.26799357442306937, + "fp16_scale": 1024.0, + "global_step": 1710, + "grad_norm": 4.358, + "learning_rate": 1.7202639466791336e-05, + "loss": 0.7511, + "step": 1710 + }, + { + "ETA": 22.11, + "epoch": 0.26956078830858443, + "fp16_scale": 1024.0, + "global_step": 1720, + "grad_norm": 2.452, + "learning_rate": 1.716732825539127e-05, + "loss": 0.7512, + "step": 1720 + }, + { + "ETA": 22.06, + "epoch": 0.27112800219409944, + "fp16_scale": 1024.0, + "global_step": 1730, + "grad_norm": 3.543, + "learning_rate": 1.7131832306305964e-05, + "loss": 0.76, + "step": 1730 + }, + { + "ETA": 22.01, + "epoch": 0.27269521607961444, + "fp16_scale": 1024.0, + "global_step": 1740, + "grad_norm": 3.003, + "learning_rate": 1.7096152534442515e-05, + "loss": 0.7576, + "step": 1740 + }, + { + "ETA": 21.96, + "epoch": 0.2742624299651295, + "fp16_scale": 1024.0, + "global_step": 1750, + "grad_norm": 3.431, + "learning_rate": 1.706028985944604e-05, + "loss": 0.7494, + "step": 1750 + }, + { + "ETA": 21.91, + "epoch": 0.2758296438506445, + "fp16_scale": 1024.0, + "global_step": 1760, + "grad_norm": 5.632, + "learning_rate": 1.7024245205675986e-05, + "loss": 0.7741, + "step": 1760 + }, + { + "ETA": 21.86, + "epoch": 0.2773968577361595, + "fp16_scale": 1024.0, + "global_step": 1770, + "grad_norm": 3.417, + "learning_rate": 1.6988019502182296e-05, + "loss": 0.7829, + "step": 1770 + }, + { + "ETA": 21.81, + "epoch": 0.2789640716216746, + "fp16_scale": 1024.0, + "global_step": 1780, + "grad_norm": 2.723, + "learning_rate": 1.6951613682681465e-05, + "loss": 0.7317, + "step": 1780 + }, + { + "ETA": 21.77, + "epoch": 0.2805312855071896, + "fp16_scale": 1024.0, + "global_step": 1790, + "grad_norm": 2.638, + "learning_rate": 1.691502868553247e-05, + "loss": 0.7567, + "step": 1790 + }, + { + "ETA": 21.72, + "epoch": 0.2820984993927046, + "fp16_scale": 1024.0, + "global_step": 1800, + "grad_norm": 4.379, + "learning_rate": 1.6878265453712587e-05, + "loss": 0.7854, + "step": 1800 + }, + { + "ETA": 21.67, + "epoch": 0.28366571327821966, + "fp16_scale": 1024.0, + "global_step": 1810, + "grad_norm": 2.736, + "learning_rate": 1.6841324934793096e-05, + "loss": 0.766, + "step": 1810 + }, + { + "ETA": 21.62, + "epoch": 0.28523292716373466, + "fp16_scale": 1024.0, + "global_step": 1820, + "grad_norm": 3.732, + "learning_rate": 1.6804208080914824e-05, + "loss": 0.7589, + "step": 1820 + }, + { + "ETA": 21.57, + "epoch": 0.2868001410492497, + "fp16_scale": 1024.0, + "global_step": 1830, + "grad_norm": 2.209, + "learning_rate": 1.6766915848763657e-05, + "loss": 0.7454, + "step": 1830 + }, + { + "ETA": 21.52, + "epoch": 0.28836735493476473, + "fp16_scale": 1024.0, + "global_step": 1840, + "grad_norm": 2.98, + "learning_rate": 1.6729449199545828e-05, + "loss": 0.7828, + "step": 1840 + }, + { + "ETA": 21.47, + "epoch": 0.28993456882027974, + "fp16_scale": 1024.0, + "global_step": 1850, + "grad_norm": 3.104, + "learning_rate": 1.669180909896317e-05, + "loss": 0.7484, + "step": 1850 + }, + { + "ETA": 21.42, + "epoch": 0.2915017827057948, + "fp16_scale": 1024.0, + "global_step": 1860, + "grad_norm": 2.728, + "learning_rate": 1.6653996517188224e-05, + "loss": 0.7652, + "step": 1860 + }, + { + "ETA": 21.37, + "epoch": 0.2930689965913098, + "fp16_scale": 1024.0, + "global_step": 1870, + "grad_norm": 3.01, + "learning_rate": 1.6616012428839226e-05, + "loss": 0.7734, + "step": 1870 + }, + { + "ETA": 21.34, + "epoch": 0.2946362104768248, + "fp16_scale": 1024.0, + "global_step": 1880, + "grad_norm": 3.439, + "learning_rate": 1.6577857812954994e-05, + "loss": 0.7893, + "step": 1880 + }, + { + "ETA": 21.32, + "epoch": 0.2962034243623399, + "fp16_scale": 1024.0, + "global_step": 1890, + "grad_norm": 2.866, + "learning_rate": 1.6539533652969683e-05, + "loss": 0.7249, + "step": 1890 + }, + { + "ETA": 21.27, + "epoch": 0.2977706382478549, + "fp16_scale": 1024.0, + "global_step": 1900, + "grad_norm": 2.747, + "learning_rate": 1.6501040936687444e-05, + "loss": 0.7319, + "step": 1900 + }, + { + "ETA": 21.22, + "epoch": 0.2993378521333699, + "fp16_scale": 1024.0, + "global_step": 1910, + "grad_norm": 2.916, + "learning_rate": 1.6462380656256962e-05, + "loss": 0.7661, + "step": 1910 + }, + { + "ETA": 21.17, + "epoch": 0.30090506601888495, + "fp16_scale": 1024.0, + "global_step": 1920, + "grad_norm": 3.348, + "learning_rate": 1.6423553808145886e-05, + "loss": 0.7494, + "step": 1920 + }, + { + "ETA": 21.12, + "epoch": 0.30247227990439995, + "fp16_scale": 1024.0, + "global_step": 1930, + "grad_norm": 2.77, + "learning_rate": 1.6384561393115135e-05, + "loss": 0.7675, + "step": 1930 + }, + { + "ETA": 21.07, + "epoch": 0.30403949378991496, + "fp16_scale": 1024.0, + "global_step": 1940, + "grad_norm": 3.482, + "learning_rate": 1.6345404416193117e-05, + "loss": 0.7661, + "step": 1940 + }, + { + "ETA": 21.02, + "epoch": 0.30560670767543, + "fp16_scale": 1024.0, + "global_step": 1950, + "grad_norm": 3.32, + "learning_rate": 1.6306083886649823e-05, + "loss": 0.7464, + "step": 1950 + }, + { + "ETA": 20.98, + "epoch": 0.30717392156094503, + "fp16_scale": 1024.0, + "global_step": 1960, + "grad_norm": 2.803, + "learning_rate": 1.6266600817970794e-05, + "loss": 0.7393, + "step": 1960 + }, + { + "ETA": 20.93, + "epoch": 0.30874113544646004, + "fp16_scale": 1024.0, + "global_step": 1970, + "grad_norm": 3.484, + "learning_rate": 1.6226956227831018e-05, + "loss": 0.7559, + "step": 1970 + }, + { + "ETA": 20.88, + "epoch": 0.3103083493319751, + "fp16_scale": 1024.0, + "global_step": 1980, + "grad_norm": 3.771, + "learning_rate": 1.6187151138068707e-05, + "loss": 0.7579, + "step": 1980 + }, + { + "ETA": 20.83, + "epoch": 0.3118755632174901, + "fp16_scale": 1024.0, + "global_step": 1990, + "grad_norm": 3.84, + "learning_rate": 1.6147186574658924e-05, + "loss": 0.7631, + "step": 1990 + }, + { + "ETA": 20.78, + "epoch": 0.3134427771030051, + "fp16_scale": 1024.0, + "global_step": 2000, + "grad_norm": 2.884, + "learning_rate": 1.6107063567687183e-05, + "loss": 0.7502, + "step": 2000 + }, + { + "ETA": 20.79, + "epoch": 0.31500999098852017, + "fp16_scale": 2048.0, + "global_step": 2010, + "grad_norm": 1.481, + "learning_rate": 1.6066783151322863e-05, + "loss": 0.7295, + "step": 2010 + }, + { + "ETA": 20.74, + "epoch": 0.3165772048740352, + "fp16_scale": 2048.0, + "global_step": 2020, + "grad_norm": 2.545, + "learning_rate": 1.6026346363792565e-05, + "loss": 0.7341, + "step": 2020 + }, + { + "ETA": 20.7, + "epoch": 0.3181444187595502, + "fp16_scale": 2048.0, + "global_step": 2030, + "grad_norm": 2.371, + "learning_rate": 1.598575424735336e-05, + "loss": 0.7362, + "step": 2030 + }, + { + "ETA": 20.65, + "epoch": 0.31971163264506525, + "fp16_scale": 2048.0, + "global_step": 2040, + "grad_norm": 2.798, + "learning_rate": 1.5945007848265912e-05, + "loss": 0.7554, + "step": 2040 + }, + { + "ETA": 20.6, + "epoch": 0.32127884653058025, + "fp16_scale": 2048.0, + "global_step": 2050, + "grad_norm": 2.313, + "learning_rate": 1.5904108216767516e-05, + "loss": 0.741, + "step": 2050 + }, + { + "ETA": 20.55, + "epoch": 0.32284606041609526, + "fp16_scale": 2048.0, + "global_step": 2060, + "grad_norm": 3.077, + "learning_rate": 1.5863056407045034e-05, + "loss": 0.7597, + "step": 2060 + }, + { + "ETA": 20.5, + "epoch": 0.3244132743016103, + "fp16_scale": 2048.0, + "global_step": 2070, + "grad_norm": 3.112, + "learning_rate": 1.582185347720771e-05, + "loss": 0.7317, + "step": 2070 + }, + { + "ETA": 20.45, + "epoch": 0.32598048818712533, + "fp16_scale": 2048.0, + "global_step": 2080, + "grad_norm": 2.469, + "learning_rate": 1.5780500489259907e-05, + "loss": 0.7093, + "step": 2080 + }, + { + "ETA": 20.4, + "epoch": 0.3275477020726404, + "fp16_scale": 2048.0, + "global_step": 2090, + "grad_norm": 2.714, + "learning_rate": 1.573899850907373e-05, + "loss": 0.7303, + "step": 2090 + }, + { + "ETA": 20.35, + "epoch": 0.3291149159581554, + "fp16_scale": 2048.0, + "global_step": 2100, + "grad_norm": 3.24, + "learning_rate": 1.5697348606361564e-05, + "loss": 0.7306, + "step": 2100 + }, + { + "ETA": 20.3, + "epoch": 0.3306821298436704, + "fp16_scale": 2048.0, + "global_step": 2110, + "grad_norm": 3.271, + "learning_rate": 1.5655551854648477e-05, + "loss": 0.7364, + "step": 2110 + }, + { + "ETA": 20.25, + "epoch": 0.33224934372918546, + "fp16_scale": 2048.0, + "global_step": 2120, + "grad_norm": 2.683, + "learning_rate": 1.5613609331244584e-05, + "loss": 0.7405, + "step": 2120 + }, + { + "ETA": 20.2, + "epoch": 0.33381655761470047, + "fp16_scale": 2048.0, + "global_step": 2130, + "grad_norm": 3.218, + "learning_rate": 1.557152211721725e-05, + "loss": 0.7515, + "step": 2130 + }, + { + "ETA": 20.15, + "epoch": 0.3353837715002155, + "fp16_scale": 2048.0, + "global_step": 2140, + "grad_norm": 3.1, + "learning_rate": 1.5529291297363235e-05, + "loss": 0.7298, + "step": 2140 + }, + { + "ETA": 20.1, + "epoch": 0.33695098538573054, + "fp16_scale": 2048.0, + "global_step": 2150, + "grad_norm": 3.992, + "learning_rate": 1.5486917960180742e-05, + "loss": 0.7228, + "step": 2150 + }, + { + "ETA": 20.05, + "epoch": 0.33851819927124555, + "fp16_scale": 2048.0, + "global_step": 2160, + "grad_norm": 2.571, + "learning_rate": 1.5444403197841345e-05, + "loss": 0.7533, + "step": 2160 + }, + { + "ETA": 20.0, + "epoch": 0.34008541315676055, + "fp16_scale": 2048.0, + "global_step": 2170, + "grad_norm": 3.083, + "learning_rate": 1.5401748106161868e-05, + "loss": 0.7298, + "step": 2170 + }, + { + "ETA": 19.95, + "epoch": 0.3416526270422756, + "fp16_scale": 2048.0, + "global_step": 2180, + "grad_norm": 2.534, + "learning_rate": 1.5358953784576093e-05, + "loss": 0.7323, + "step": 2180 + }, + { + "ETA": 19.9, + "epoch": 0.3432198409277906, + "fp16_scale": 2048.0, + "global_step": 2190, + "grad_norm": 2.848, + "learning_rate": 1.5316021336106463e-05, + "loss": 0.7388, + "step": 2190 + }, + { + "ETA": 19.85, + "epoch": 0.3447870548133056, + "fp16_scale": 2048.0, + "global_step": 2200, + "grad_norm": 2.76, + "learning_rate": 1.527295186733564e-05, + "loss": 0.723, + "step": 2200 + }, + { + "ETA": 19.81, + "epoch": 0.3463542686988207, + "fp16_scale": 2048.0, + "global_step": 2210, + "grad_norm": 3.625, + "learning_rate": 1.5229746488377974e-05, + "loss": 0.7092, + "step": 2210 + }, + { + "ETA": 19.76, + "epoch": 0.3479214825843357, + "fp16_scale": 2048.0, + "global_step": 2220, + "grad_norm": 3.283, + "learning_rate": 1.5186406312850901e-05, + "loss": 0.7439, + "step": 2220 + }, + { + "ETA": 19.71, + "epoch": 0.3494886964698507, + "fp16_scale": 2048.0, + "global_step": 2230, + "grad_norm": 2.736, + "learning_rate": 1.514293245784623e-05, + "loss": 0.7088, + "step": 2230 + }, + { + "ETA": 19.66, + "epoch": 0.35105591035536576, + "fp16_scale": 2048.0, + "global_step": 2240, + "grad_norm": 2.318, + "learning_rate": 1.5099326043901361e-05, + "loss": 0.7293, + "step": 2240 + }, + { + "ETA": 19.61, + "epoch": 0.35262312424088077, + "fp16_scale": 2048.0, + "global_step": 2250, + "grad_norm": 3.371, + "learning_rate": 1.505558819497039e-05, + "loss": 0.7234, + "step": 2250 + }, + { + "ETA": 19.56, + "epoch": 0.3541903381263958, + "fp16_scale": 2048.0, + "global_step": 2260, + "grad_norm": 2.881, + "learning_rate": 1.5011720038395145e-05, + "loss": 0.7456, + "step": 2260 + }, + { + "ETA": 19.51, + "epoch": 0.35575755201191084, + "fp16_scale": 2048.0, + "global_step": 2270, + "grad_norm": 2.577, + "learning_rate": 1.4967722704876147e-05, + "loss": 0.7459, + "step": 2270 + }, + { + "ETA": 19.46, + "epoch": 0.35732476589742584, + "fp16_scale": 2048.0, + "global_step": 2280, + "grad_norm": 2.892, + "learning_rate": 1.4923597328443423e-05, + "loss": 0.7415, + "step": 2280 + }, + { + "ETA": 19.41, + "epoch": 0.35889197978294085, + "fp16_scale": 2048.0, + "global_step": 2290, + "grad_norm": 2.461, + "learning_rate": 1.4879345046427322e-05, + "loss": 0.7487, + "step": 2290 + }, + { + "ETA": 19.36, + "epoch": 0.3604591936684559, + "fp16_scale": 2048.0, + "global_step": 2300, + "grad_norm": 2.269, + "learning_rate": 1.4834966999429179e-05, + "loss": 0.7339, + "step": 2300 + }, + { + "ETA": 19.31, + "epoch": 0.3620264075539709, + "fp16_scale": 2048.0, + "global_step": 2310, + "grad_norm": 2.551, + "learning_rate": 1.4790464331291906e-05, + "loss": 0.7049, + "step": 2310 + }, + { + "ETA": 19.26, + "epoch": 0.363593621439486, + "fp16_scale": 2048.0, + "global_step": 2320, + "grad_norm": 3.06, + "learning_rate": 1.4745838189070531e-05, + "loss": 0.7555, + "step": 2320 + }, + { + "ETA": 19.22, + "epoch": 0.365160835325001, + "fp16_scale": 2048.0, + "global_step": 2330, + "grad_norm": 3.508, + "learning_rate": 1.4701089723002623e-05, + "loss": 0.7164, + "step": 2330 + }, + { + "ETA": 19.17, + "epoch": 0.366728049210516, + "fp16_scale": 2048.0, + "global_step": 2340, + "grad_norm": 2.925, + "learning_rate": 1.4656220086478645e-05, + "loss": 0.7491, + "step": 2340 + }, + { + "ETA": 19.12, + "epoch": 0.36829526309603106, + "fp16_scale": 2048.0, + "global_step": 2350, + "grad_norm": 2.25, + "learning_rate": 1.4611230436012217e-05, + "loss": 0.7173, + "step": 2350 + }, + { + "ETA": 19.07, + "epoch": 0.36986247698154606, + "fp16_scale": 2048.0, + "global_step": 2360, + "grad_norm": 3.265, + "learning_rate": 1.4566121931210326e-05, + "loss": 0.7516, + "step": 2360 + }, + { + "ETA": 19.02, + "epoch": 0.37142969086706107, + "fp16_scale": 2048.0, + "global_step": 2370, + "grad_norm": 2.661, + "learning_rate": 1.4520895734743419e-05, + "loss": 0.7159, + "step": 2370 + }, + { + "ETA": 18.97, + "epoch": 0.37299690475257613, + "fp16_scale": 2048.0, + "global_step": 2380, + "grad_norm": 3.14, + "learning_rate": 1.4475553012315441e-05, + "loss": 0.7037, + "step": 2380 + }, + { + "ETA": 18.92, + "epoch": 0.37456411863809114, + "fp16_scale": 2048.0, + "global_step": 2390, + "grad_norm": 2.859, + "learning_rate": 1.44300949326338e-05, + "loss": 0.7347, + "step": 2390 + }, + { + "ETA": 18.87, + "epoch": 0.37613133252360614, + "fp16_scale": 2048.0, + "global_step": 2400, + "grad_norm": 4.006, + "learning_rate": 1.4384522667379229e-05, + "loss": 0.729, + "step": 2400 + }, + { + "ETA": 18.83, + "epoch": 0.3776985464091212, + "fp16_scale": 2048.0, + "global_step": 2410, + "grad_norm": 3.044, + "learning_rate": 1.4338837391175582e-05, + "loss": 0.741, + "step": 2410 + }, + { + "ETA": 18.78, + "epoch": 0.3792657602946362, + "fp16_scale": 2048.0, + "global_step": 2420, + "grad_norm": 2.674, + "learning_rate": 1.4297624991405243e-05, + "loss": 0.7268, + "step": 2420 + }, + { + "ETA": 18.74, + "epoch": 0.3808329741801512, + "fp16_scale": 2048.0, + "global_step": 2430, + "grad_norm": 2.866, + "learning_rate": 1.4251728240895031e-05, + "loss": 0.7816, + "step": 2430 + }, + { + "ETA": 18.71, + "epoch": 0.3824001880656663, + "fp16_scale": 2048.0, + "global_step": 2440, + "grad_norm": 2.794, + "learning_rate": 1.4205721902208371e-05, + "loss": 0.7469, + "step": 2440 + }, + { + "ETA": 18.66, + "epoch": 0.3839674019511813, + "fp16_scale": 2048.0, + "global_step": 2450, + "grad_norm": 2.944, + "learning_rate": 1.4159607161157363e-05, + "loss": 0.7123, + "step": 2450 + }, + { + "ETA": 18.62, + "epoch": 0.3855346158366963, + "fp16_scale": 2048.0, + "global_step": 2460, + "grad_norm": 3.239, + "learning_rate": 1.411338520634816e-05, + "loss": 0.7221, + "step": 2460 + }, + { + "ETA": 18.57, + "epoch": 0.38710182972221135, + "fp16_scale": 2048.0, + "global_step": 2470, + "grad_norm": 2.86, + "learning_rate": 1.4067057229150358e-05, + "loss": 0.7189, + "step": 2470 + }, + { + "ETA": 18.52, + "epoch": 0.38866904360772636, + "fp16_scale": 2048.0, + "global_step": 2480, + "grad_norm": 2.896, + "learning_rate": 1.402062442366627e-05, + "loss": 0.7206, + "step": 2480 + }, + { + "ETA": 18.47, + "epoch": 0.39023625749324137, + "fp16_scale": 2048.0, + "global_step": 2490, + "grad_norm": 2.715, + "learning_rate": 1.3974087986700163e-05, + "loss": 0.7439, + "step": 2490 + }, + { + "ETA": 18.42, + "epoch": 0.39180347137875643, + "fp16_scale": 2048.0, + "global_step": 2500, + "grad_norm": 2.595, + "learning_rate": 1.3927449117727392e-05, + "loss": 0.7346, + "step": 2500 + }, + { + "ETA": 18.42, + "epoch": 0.39337068526427144, + "fp16_scale": 2048.0, + "global_step": 2510, + "grad_norm": 2.875, + "learning_rate": 1.3880709018863504e-05, + "loss": 0.7652, + "step": 2510 + }, + { + "ETA": 18.37, + "epoch": 0.39493789914978644, + "fp16_scale": 2048.0, + "global_step": 2520, + "grad_norm": 3.692, + "learning_rate": 1.3833868894833238e-05, + "loss": 0.7012, + "step": 2520 + }, + { + "ETA": 18.32, + "epoch": 0.3965051130353015, + "fp16_scale": 2048.0, + "global_step": 2530, + "grad_norm": 3.356, + "learning_rate": 1.3786929952939478e-05, + "loss": 0.7451, + "step": 2530 + }, + { + "ETA": 18.27, + "epoch": 0.3980723269208165, + "fp16_scale": 2048.0, + "global_step": 2540, + "grad_norm": 2.35, + "learning_rate": 1.373989340303214e-05, + "loss": 0.7325, + "step": 2540 + }, + { + "ETA": 18.22, + "epoch": 0.3996395408063315, + "fp16_scale": 2048.0, + "global_step": 2550, + "grad_norm": 3.077, + "learning_rate": 1.3692760457476985e-05, + "loss": 0.7446, + "step": 2550 + }, + { + "ETA": 18.17, + "epoch": 0.4012067546918466, + "fp16_scale": 2048.0, + "global_step": 2560, + "grad_norm": 3.121, + "learning_rate": 1.3645532331124362e-05, + "loss": 0.7164, + "step": 2560 + }, + { + "ETA": 18.12, + "epoch": 0.4027739685773616, + "fp16_scale": 2048.0, + "global_step": 2570, + "grad_norm": 2.211, + "learning_rate": 1.3598210241277905e-05, + "loss": 0.7411, + "step": 2570 + }, + { + "ETA": 18.08, + "epoch": 0.40434118246287665, + "fp16_scale": 2048.0, + "global_step": 2580, + "grad_norm": 2.489, + "learning_rate": 1.3550795407663158e-05, + "loss": 0.7431, + "step": 2580 + }, + { + "ETA": 18.03, + "epoch": 0.40590839634839165, + "fp16_scale": 2048.0, + "global_step": 2590, + "grad_norm": 2.241, + "learning_rate": 1.3503289052396133e-05, + "loss": 0.7127, + "step": 2590 + }, + { + "ETA": 17.98, + "epoch": 0.40747561023390666, + "fp16_scale": 1024.0, + "global_step": 2600, + "grad_norm": 3.6, + "learning_rate": 1.3460456093623443e-05, + "loss": 0.7245, + "step": 2600 + }, + { + "ETA": 17.93, + "epoch": 0.4090428241194217, + "fp16_scale": 1024.0, + "global_step": 2610, + "grad_norm": 2.75, + "learning_rate": 1.3412779222569907e-05, + "loss": 0.7512, + "step": 2610 + }, + { + "ETA": 17.88, + "epoch": 0.41061003800493673, + "fp16_scale": 1024.0, + "global_step": 2620, + "grad_norm": 2.475, + "learning_rate": 1.3365014387227393e-05, + "loss": 0.7082, + "step": 2620 + }, + { + "ETA": 17.83, + "epoch": 0.41217725189045173, + "fp16_scale": 1024.0, + "global_step": 2630, + "grad_norm": 3.246, + "learning_rate": 1.3317162818733205e-05, + "loss": 0.7197, + "step": 2630 + }, + { + "ETA": 17.78, + "epoch": 0.4137444657759668, + "fp16_scale": 1024.0, + "global_step": 2640, + "grad_norm": 3.182, + "learning_rate": 1.326922575046018e-05, + "loss": 0.7903, + "step": 2640 + }, + { + "ETA": 17.73, + "epoch": 0.4153116796614818, + "fp16_scale": 512.0, + "global_step": 2650, + "grad_norm": 35.311, + "learning_rate": 1.3226010307862582e-05, + "loss": 0.7686, + "step": 2650 + }, + { + "ETA": 17.68, + "epoch": 0.4168788935469968, + "fp16_scale": 512.0, + "global_step": 2660, + "grad_norm": 2.301, + "learning_rate": 1.3177914195819018e-05, + "loss": 0.7842, + "step": 2660 + }, + { + "ETA": 17.63, + "epoch": 0.41844610743251187, + "fp16_scale": 512.0, + "global_step": 2670, + "grad_norm": 2.258, + "learning_rate": 1.3129736173125975e-05, + "loss": 0.7565, + "step": 2670 + }, + { + "ETA": 17.59, + "epoch": 0.4200133213180269, + "fp16_scale": 512.0, + "global_step": 2680, + "grad_norm": 2.581, + "learning_rate": 1.3081477481570642e-05, + "loss": 0.7429, + "step": 2680 + }, + { + "ETA": 17.54, + "epoch": 0.4215805352035419, + "fp16_scale": 512.0, + "global_step": 2690, + "grad_norm": 2.537, + "learning_rate": 1.303313936501944e-05, + "loss": 0.7517, + "step": 2690 + }, + { + "ETA": 17.49, + "epoch": 0.42314774908905695, + "fp16_scale": 512.0, + "global_step": 2700, + "grad_norm": 5.254, + "learning_rate": 1.2984723069385975e-05, + "loss": 0.7414, + "step": 2700 + }, + { + "ETA": 17.44, + "epoch": 0.42471496297457195, + "fp16_scale": 512.0, + "global_step": 2710, + "grad_norm": 2.486, + "learning_rate": 1.293622984259891e-05, + "loss": 0.7547, + "step": 2710 + }, + { + "ETA": 17.39, + "epoch": 0.42628217686008696, + "fp16_scale": 512.0, + "global_step": 2720, + "grad_norm": 2.129, + "learning_rate": 1.2887660934569809e-05, + "loss": 0.7298, + "step": 2720 + }, + { + "ETA": 17.34, + "epoch": 0.427849390745602, + "fp16_scale": 512.0, + "global_step": 2730, + "grad_norm": 3.215, + "learning_rate": 1.2839017597160916e-05, + "loss": 0.7346, + "step": 2730 + }, + { + "ETA": 17.29, + "epoch": 0.429416604631117, + "fp16_scale": 512.0, + "global_step": 2740, + "grad_norm": 2.886, + "learning_rate": 1.2790301084152889e-05, + "loss": 0.7197, + "step": 2740 + }, + { + "ETA": 17.24, + "epoch": 0.43098381851663203, + "fp16_scale": 512.0, + "global_step": 2750, + "grad_norm": 2.857, + "learning_rate": 1.2741512651212495e-05, + "loss": 0.7417, + "step": 2750 + }, + { + "ETA": 17.2, + "epoch": 0.4325510324021471, + "fp16_scale": 512.0, + "global_step": 2760, + "grad_norm": 2.604, + "learning_rate": 1.2692653555860224e-05, + "loss": 0.7411, + "step": 2760 + }, + { + "ETA": 17.15, + "epoch": 0.4341182462876621, + "fp16_scale": 512.0, + "global_step": 2770, + "grad_norm": 2.792, + "learning_rate": 1.264372505743789e-05, + "loss": 0.7263, + "step": 2770 + }, + { + "ETA": 17.1, + "epoch": 0.4356854601731771, + "fp16_scale": 512.0, + "global_step": 2780, + "grad_norm": 3.684, + "learning_rate": 1.2594728417076172e-05, + "loss": 0.7712, + "step": 2780 + }, + { + "ETA": 17.05, + "epoch": 0.43725267405869217, + "fp16_scale": 512.0, + "global_step": 2790, + "grad_norm": 2.179, + "learning_rate": 1.2545664897662108e-05, + "loss": 0.7155, + "step": 2790 + }, + { + "ETA": 17.0, + "epoch": 0.4388198879442072, + "fp16_scale": 512.0, + "global_step": 2800, + "grad_norm": 2.914, + "learning_rate": 1.2496535763806535e-05, + "loss": 0.728, + "step": 2800 + }, + { + "ETA": 16.95, + "epoch": 0.44038710182972224, + "fp16_scale": 512.0, + "global_step": 2810, + "grad_norm": 3.342, + "learning_rate": 1.2447342281811499e-05, + "loss": 0.7383, + "step": 2810 + }, + { + "ETA": 16.9, + "epoch": 0.44195431571523724, + "fp16_scale": 512.0, + "global_step": 2820, + "grad_norm": 2.831, + "learning_rate": 1.239808571963763e-05, + "loss": 0.7351, + "step": 2820 + }, + { + "ETA": 16.86, + "epoch": 0.44352152960075225, + "fp16_scale": 512.0, + "global_step": 2830, + "grad_norm": 2.444, + "learning_rate": 1.2348767346871438e-05, + "loss": 0.7054, + "step": 2830 + }, + { + "ETA": 16.81, + "epoch": 0.4450887434862673, + "fp16_scale": 512.0, + "global_step": 2840, + "grad_norm": 2.822, + "learning_rate": 1.2299388434692596e-05, + "loss": 0.7212, + "step": 2840 + }, + { + "ETA": 16.76, + "epoch": 0.4466559573717823, + "fp16_scale": 512.0, + "global_step": 2850, + "grad_norm": 2.994, + "learning_rate": 1.2249950255841187e-05, + "loss": 0.7232, + "step": 2850 + }, + { + "ETA": 16.71, + "epoch": 0.4482231712572973, + "fp16_scale": 512.0, + "global_step": 2860, + "grad_norm": 2.159, + "learning_rate": 1.220045408458489e-05, + "loss": 0.721, + "step": 2860 + }, + { + "ETA": 16.66, + "epoch": 0.4497903851428124, + "fp16_scale": 512.0, + "global_step": 2870, + "grad_norm": 2.573, + "learning_rate": 1.215090119668613e-05, + "loss": 0.7249, + "step": 2870 + }, + { + "ETA": 16.61, + "epoch": 0.4513575990283274, + "fp16_scale": 512.0, + "global_step": 2880, + "grad_norm": 2.2, + "learning_rate": 1.2101292869369208e-05, + "loss": 0.6849, + "step": 2880 + }, + { + "ETA": 16.56, + "epoch": 0.4529248129138424, + "fp16_scale": 512.0, + "global_step": 2890, + "grad_norm": 2.541, + "learning_rate": 1.2051630381287375e-05, + "loss": 0.7097, + "step": 2890 + }, + { + "ETA": 16.52, + "epoch": 0.45449202679935746, + "fp16_scale": 512.0, + "global_step": 2900, + "grad_norm": 2.185, + "learning_rate": 1.2001915012489869e-05, + "loss": 0.7336, + "step": 2900 + }, + { + "ETA": 16.47, + "epoch": 0.45605924068487247, + "fp16_scale": 512.0, + "global_step": 2910, + "grad_norm": 3.047, + "learning_rate": 1.195214804438893e-05, + "loss": 0.7295, + "step": 2910 + }, + { + "ETA": 16.42, + "epoch": 0.4576264545703875, + "fp16_scale": 512.0, + "global_step": 2920, + "grad_norm": 2.518, + "learning_rate": 1.1902330759726766e-05, + "loss": 0.7207, + "step": 2920 + }, + { + "ETA": 16.37, + "epoch": 0.45919366845590254, + "fp16_scale": 512.0, + "global_step": 2930, + "grad_norm": 2.452, + "learning_rate": 1.1852464442542501e-05, + "loss": 0.7031, + "step": 2930 + }, + { + "ETA": 16.32, + "epoch": 0.46076088234141754, + "fp16_scale": 512.0, + "global_step": 2940, + "grad_norm": 2.328, + "learning_rate": 1.180255037813906e-05, + "loss": 0.7034, + "step": 2940 + }, + { + "ETA": 16.27, + "epoch": 0.46232809622693255, + "fp16_scale": 512.0, + "global_step": 2950, + "grad_norm": 2.481, + "learning_rate": 1.1752589853050057e-05, + "loss": 0.712, + "step": 2950 + }, + { + "ETA": 16.23, + "epoch": 0.4638953101124476, + "fp16_scale": 512.0, + "global_step": 2960, + "grad_norm": 2.881, + "learning_rate": 1.1702584155006631e-05, + "loss": 0.7308, + "step": 2960 + }, + { + "ETA": 16.18, + "epoch": 0.4654625239979626, + "fp16_scale": 512.0, + "global_step": 2970, + "grad_norm": 2.929, + "learning_rate": 1.1652534572904248e-05, + "loss": 0.7251, + "step": 2970 + }, + { + "ETA": 16.14, + "epoch": 0.4670297378834776, + "fp16_scale": 512.0, + "global_step": 2980, + "grad_norm": 2.239, + "learning_rate": 1.1602442396769486e-05, + "loss": 0.6962, + "step": 2980 + }, + { + "ETA": 16.1, + "epoch": 0.4685969517689927, + "fp16_scale": 512.0, + "global_step": 2990, + "grad_norm": 2.487, + "learning_rate": 1.1552308917726786e-05, + "loss": 0.7391, + "step": 2990 + }, + { + "ETA": 16.06, + "epoch": 0.4701641656545077, + "fp16_scale": 512.0, + "global_step": 3000, + "grad_norm": 3.831, + "learning_rate": 1.1502135427965167e-05, + "loss": 0.715, + "step": 3000 + }, + { + "ETA": 16.04, + "epoch": 0.4717313795400227, + "fp16_scale": 512.0, + "global_step": 3010, + "grad_norm": 2.421, + "learning_rate": 1.1451923220704916e-05, + "loss": 0.7511, + "step": 3010 + }, + { + "ETA": 15.99, + "epoch": 0.47329859342553776, + "fp16_scale": 512.0, + "global_step": 3020, + "grad_norm": 2.23, + "learning_rate": 1.1401673590164281e-05, + "loss": 0.6914, + "step": 3020 + }, + { + "ETA": 15.94, + "epoch": 0.47486580731105277, + "fp16_scale": 512.0, + "global_step": 3030, + "grad_norm": 2.211, + "learning_rate": 1.135138783152608e-05, + "loss": 0.7333, + "step": 3030 + }, + { + "ETA": 15.9, + "epoch": 0.4764330211965678, + "fp16_scale": 512.0, + "global_step": 3040, + "grad_norm": 4.443, + "learning_rate": 1.1301067240904331e-05, + "loss": 0.7227, + "step": 3040 + }, + { + "ETA": 15.85, + "epoch": 0.47800023508208284, + "fp16_scale": 512.0, + "global_step": 3050, + "grad_norm": 3.145, + "learning_rate": 1.1250713115310852e-05, + "loss": 0.7343, + "step": 3050 + }, + { + "ETA": 15.8, + "epoch": 0.47956744896759784, + "fp16_scale": 512.0, + "global_step": 3060, + "grad_norm": 2.237, + "learning_rate": 1.1200326752621821e-05, + "loss": 0.7027, + "step": 3060 + }, + { + "ETA": 15.75, + "epoch": 0.4811346628531129, + "fp16_scale": 512.0, + "global_step": 3070, + "grad_norm": 3.649, + "learning_rate": 1.1149909451544327e-05, + "loss": 0.711, + "step": 3070 + }, + { + "ETA": 15.7, + "epoch": 0.4827018767386279, + "fp16_scale": 512.0, + "global_step": 3080, + "grad_norm": 6.593, + "learning_rate": 1.1099462511582893e-05, + "loss": 0.6981, + "step": 3080 + }, + { + "ETA": 15.65, + "epoch": 0.4842690906241429, + "fp16_scale": 512.0, + "global_step": 3090, + "grad_norm": 2.193, + "learning_rate": 1.104898723300599e-05, + "loss": 0.6855, + "step": 3090 + }, + { + "ETA": 15.6, + "epoch": 0.485836304509658, + "fp16_scale": 512.0, + "global_step": 3100, + "grad_norm": 2.081, + "learning_rate": 1.0998484916812503e-05, + "loss": 0.7143, + "step": 3100 + }, + { + "ETA": 15.55, + "epoch": 0.487403518395173, + "fp16_scale": 512.0, + "global_step": 3110, + "grad_norm": 3.085, + "learning_rate": 1.0947956864698223e-05, + "loss": 0.7145, + "step": 3110 + }, + { + "ETA": 15.51, + "epoch": 0.488970732280688, + "fp16_scale": 512.0, + "global_step": 3120, + "grad_norm": 2.29, + "learning_rate": 1.0897404379022281e-05, + "loss": 0.6886, + "step": 3120 + }, + { + "ETA": 15.46, + "epoch": 0.49053794616620305, + "fp16_scale": 512.0, + "global_step": 3130, + "grad_norm": 2.721, + "learning_rate": 1.0846828762773582e-05, + "loss": 0.6601, + "step": 3130 + }, + { + "ETA": 15.41, + "epoch": 0.49210516005171806, + "fp16_scale": 512.0, + "global_step": 3140, + "grad_norm": 3.726, + "learning_rate": 1.0796231319537214e-05, + "loss": 0.694, + "step": 3140 + }, + { + "ETA": 15.36, + "epoch": 0.49367237393723307, + "fp16_scale": 512.0, + "global_step": 3150, + "grad_norm": 2.391, + "learning_rate": 1.0745613353460862e-05, + "loss": 0.7002, + "step": 3150 + }, + { + "ETA": 15.31, + "epoch": 0.49523958782274813, + "fp16_scale": 512.0, + "global_step": 3160, + "grad_norm": 2.541, + "learning_rate": 1.069497616922119e-05, + "loss": 0.7195, + "step": 3160 + }, + { + "ETA": 15.26, + "epoch": 0.49680680170826313, + "fp16_scale": 512.0, + "global_step": 3170, + "grad_norm": 2.212, + "learning_rate": 1.0644321071990197e-05, + "loss": 0.7227, + "step": 3170 + }, + { + "ETA": 15.21, + "epoch": 0.49837401559377814, + "fp16_scale": 512.0, + "global_step": 3180, + "grad_norm": 2.843, + "learning_rate": 1.0593649367401607e-05, + "loss": 0.7215, + "step": 3180 + }, + { + "ETA": 15.17, + "epoch": 0.4999412294792932, + "fp16_scale": 512.0, + "global_step": 3190, + "grad_norm": 2.622, + "learning_rate": 1.0542962361517182e-05, + "loss": 0.6857, + "step": 3190 + }, + { + "ETA": 15.12, + "epoch": 0.5015084433648083, + "fp16_scale": 512.0, + "global_step": 3200, + "grad_norm": 4.081, + "learning_rate": 1.0492261360793082e-05, + "loss": 0.6919, + "step": 3200 + }, + { + "ETA": 15.07, + "epoch": 0.5030756572503232, + "fp16_scale": 512.0, + "global_step": 3210, + "grad_norm": 1.986, + "learning_rate": 1.0441547672046178e-05, + "loss": 0.7229, + "step": 3210 + }, + { + "ETA": 15.02, + "epoch": 0.5046428711358383, + "fp16_scale": 512.0, + "global_step": 3220, + "grad_norm": 2.561, + "learning_rate": 1.039082260242038e-05, + "loss": 0.692, + "step": 3220 + }, + { + "ETA": 14.97, + "epoch": 0.5062100850213533, + "fp16_scale": 512.0, + "global_step": 3230, + "grad_norm": 2.619, + "learning_rate": 1.0340087459352946e-05, + "loss": 0.72, + "step": 3230 + }, + { + "ETA": 14.92, + "epoch": 0.5077772989068683, + "fp16_scale": 512.0, + "global_step": 3240, + "grad_norm": 2.845, + "learning_rate": 1.028934355054076e-05, + "loss": 0.7005, + "step": 3240 + }, + { + "ETA": 14.87, + "epoch": 0.5093445127923834, + "fp16_scale": 512.0, + "global_step": 3250, + "grad_norm": 2.274, + "learning_rate": 1.023859218390665e-05, + "loss": 0.6779, + "step": 3250 + }, + { + "ETA": 14.83, + "epoch": 0.5109117266778984, + "fp16_scale": 512.0, + "global_step": 3260, + "grad_norm": 2.636, + "learning_rate": 1.0187834667565684e-05, + "loss": 0.7082, + "step": 3260 + }, + { + "ETA": 14.78, + "epoch": 0.5124789405634134, + "fp16_scale": 512.0, + "global_step": 3270, + "grad_norm": 2.666, + "learning_rate": 1.0137072309791411e-05, + "loss": 0.699, + "step": 3270 + }, + { + "ETA": 14.73, + "epoch": 0.5140461544489284, + "fp16_scale": 512.0, + "global_step": 3280, + "grad_norm": 2.36, + "learning_rate": 1.008630641898219e-05, + "loss": 0.7035, + "step": 3280 + }, + { + "ETA": 14.68, + "epoch": 0.5156133683344435, + "fp16_scale": 512.0, + "global_step": 3290, + "grad_norm": 2.33, + "learning_rate": 1.0035538303627437e-05, + "loss": 0.6767, + "step": 3290 + }, + { + "ETA": 14.63, + "epoch": 0.5171805822199584, + "fp16_scale": 512.0, + "global_step": 3300, + "grad_norm": 2.257, + "learning_rate": 9.9847692722739e-06, + "loss": 0.6767, + "step": 3300 + }, + { + "ETA": 14.58, + "epoch": 0.5187477961054735, + "fp16_scale": 512.0, + "global_step": 3310, + "grad_norm": 2.678, + "learning_rate": 9.934000633491945e-06, + "loss": 0.7028, + "step": 3310 + }, + { + "ETA": 14.54, + "epoch": 0.5203150099909886, + "fp16_scale": 512.0, + "global_step": 3320, + "grad_norm": 3.457, + "learning_rate": 9.883233695841814e-06, + "loss": 0.6642, + "step": 3320 + }, + { + "ETA": 14.49, + "epoch": 0.5218822238765035, + "fp16_scale": 512.0, + "global_step": 3330, + "grad_norm": 3.567, + "learning_rate": 9.832469767839908e-06, + "loss": 0.7094, + "step": 3330 + }, + { + "ETA": 14.44, + "epoch": 0.5234494377620186, + "fp16_scale": 512.0, + "global_step": 3340, + "grad_norm": 2.891, + "learning_rate": 9.781710157925047e-06, + "loss": 0.6836, + "step": 3340 + }, + { + "ETA": 14.39, + "epoch": 0.5250166516475336, + "fp16_scale": 512.0, + "global_step": 3350, + "grad_norm": 2.194, + "learning_rate": 9.730956174424757e-06, + "loss": 0.6913, + "step": 3350 + }, + { + "ETA": 14.34, + "epoch": 0.5265838655330486, + "fp16_scale": 512.0, + "global_step": 3360, + "grad_norm": 2.086, + "learning_rate": 9.680209125521545e-06, + "loss": 0.6788, + "step": 3360 + }, + { + "ETA": 14.29, + "epoch": 0.5281510794185637, + "fp16_scale": 512.0, + "global_step": 3370, + "grad_norm": 2.488, + "learning_rate": 9.629470319219173e-06, + "loss": 0.7221, + "step": 3370 + }, + { + "ETA": 14.25, + "epoch": 0.5297182933040787, + "fp16_scale": 512.0, + "global_step": 3380, + "grad_norm": 2.049, + "learning_rate": 9.578741063308954e-06, + "loss": 0.6611, + "step": 3380 + }, + { + "ETA": 14.2, + "epoch": 0.5312855071895937, + "fp16_scale": 512.0, + "global_step": 3390, + "grad_norm": 2.858, + "learning_rate": 9.528022665336045e-06, + "loss": 0.6552, + "step": 3390 + }, + { + "ETA": 14.15, + "epoch": 0.5328527210751087, + "fp16_scale": 512.0, + "global_step": 3400, + "grad_norm": 2.535, + "learning_rate": 9.477316432565727e-06, + "loss": 0.6672, + "step": 3400 + }, + { + "ETA": 14.1, + "epoch": 0.5344199349606238, + "fp16_scale": 512.0, + "global_step": 3410, + "grad_norm": 3.078, + "learning_rate": 9.426623671949735e-06, + "loss": 0.6831, + "step": 3410 + }, + { + "ETA": 14.05, + "epoch": 0.5359871488461387, + "fp16_scale": 512.0, + "global_step": 3420, + "grad_norm": 3.191, + "learning_rate": 9.37594569009256e-06, + "loss": 0.7036, + "step": 3420 + }, + { + "ETA": 14.0, + "epoch": 0.5375543627316538, + "fp16_scale": 512.0, + "global_step": 3430, + "grad_norm": 2.237, + "learning_rate": 9.325283793217758e-06, + "loss": 0.6793, + "step": 3430 + }, + { + "ETA": 13.96, + "epoch": 0.5391215766171689, + "fp16_scale": 512.0, + "global_step": 3440, + "grad_norm": 3.739, + "learning_rate": 9.274639287134309e-06, + "loss": 0.6925, + "step": 3440 + }, + { + "ETA": 13.91, + "epoch": 0.5406887905026838, + "fp16_scale": 512.0, + "global_step": 3450, + "grad_norm": 3.498, + "learning_rate": 9.224013477202939e-06, + "loss": 0.7115, + "step": 3450 + }, + { + "ETA": 13.86, + "epoch": 0.5422560043881989, + "fp16_scale": 512.0, + "global_step": 3460, + "grad_norm": 2.282, + "learning_rate": 9.17340766830248e-06, + "loss": 0.6965, + "step": 3460 + }, + { + "ETA": 13.81, + "epoch": 0.5438232182737139, + "fp16_scale": 512.0, + "global_step": 3470, + "grad_norm": 2.453, + "learning_rate": 9.122823164796241e-06, + "loss": 0.6978, + "step": 3470 + }, + { + "ETA": 13.76, + "epoch": 0.5453904321592289, + "fp16_scale": 512.0, + "global_step": 3480, + "grad_norm": 2.101, + "learning_rate": 9.072261270498389e-06, + "loss": 0.6903, + "step": 3480 + }, + { + "ETA": 13.71, + "epoch": 0.546957646044744, + "fp16_scale": 512.0, + "global_step": 3490, + "grad_norm": 2.317, + "learning_rate": 9.02172328864033e-06, + "loss": 0.6787, + "step": 3490 + }, + { + "ETA": 13.67, + "epoch": 0.548524859930259, + "fp16_scale": 512.0, + "global_step": 3500, + "grad_norm": 3.384, + "learning_rate": 8.971210521837133e-06, + "loss": 0.694, + "step": 3500 + }, + { + "ETA": 13.64, + "epoch": 0.550092073815774, + "fp16_scale": 512.0, + "global_step": 3510, + "grad_norm": 2.945, + "learning_rate": 8.920724272053952e-06, + "loss": 0.7047, + "step": 3510 + }, + { + "ETA": 13.6, + "epoch": 0.551659287701289, + "fp16_scale": 512.0, + "global_step": 3520, + "grad_norm": 2.045, + "learning_rate": 8.870265840572459e-06, + "loss": 0.6963, + "step": 3520 + }, + { + "ETA": 13.56, + "epoch": 0.5532265015868041, + "fp16_scale": 512.0, + "global_step": 3530, + "grad_norm": 3.728, + "learning_rate": 8.819836527957317e-06, + "loss": 0.7034, + "step": 3530 + }, + { + "ETA": 13.51, + "epoch": 0.554793715472319, + "fp16_scale": 512.0, + "global_step": 3540, + "grad_norm": 2.796, + "learning_rate": 8.769437634022642e-06, + "loss": 0.6766, + "step": 3540 + }, + { + "ETA": 13.47, + "epoch": 0.5563609293578341, + "fp16_scale": 512.0, + "global_step": 3550, + "grad_norm": 2.562, + "learning_rate": 8.71907045779852e-06, + "loss": 0.6392, + "step": 3550 + }, + { + "ETA": 13.42, + "epoch": 0.5579281432433492, + "fp16_scale": 512.0, + "global_step": 3560, + "grad_norm": 2.453, + "learning_rate": 8.668736297497511e-06, + "loss": 0.6982, + "step": 3560 + }, + { + "ETA": 13.37, + "epoch": 0.5594953571288641, + "fp16_scale": 512.0, + "global_step": 3570, + "grad_norm": 2.169, + "learning_rate": 8.618436450481182e-06, + "loss": 0.6625, + "step": 3570 + }, + { + "ETA": 13.32, + "epoch": 0.5610625710143792, + "fp16_scale": 512.0, + "global_step": 3580, + "grad_norm": 1.961, + "learning_rate": 8.568172213226684e-06, + "loss": 0.6839, + "step": 3580 + }, + { + "ETA": 13.27, + "epoch": 0.5626297848998942, + "fp16_scale": 512.0, + "global_step": 3590, + "grad_norm": 2.06, + "learning_rate": 8.517944881293327e-06, + "loss": 0.6616, + "step": 3590 + }, + { + "ETA": 13.23, + "epoch": 0.5641969987854092, + "fp16_scale": 512.0, + "global_step": 3600, + "grad_norm": 2.749, + "learning_rate": 8.467755749289186e-06, + "loss": 0.7188, + "step": 3600 + }, + { + "ETA": 13.18, + "epoch": 0.5657642126709242, + "fp16_scale": 512.0, + "global_step": 3610, + "grad_norm": 2.239, + "learning_rate": 8.417606110837738e-06, + "loss": 0.6778, + "step": 3610 + }, + { + "ETA": 13.13, + "epoch": 0.5673314265564393, + "fp16_scale": 512.0, + "global_step": 3620, + "grad_norm": 2.227, + "learning_rate": 8.367497258544507e-06, + "loss": 0.6858, + "step": 3620 + }, + { + "ETA": 13.08, + "epoch": 0.5688986404419543, + "fp16_scale": 512.0, + "global_step": 3630, + "grad_norm": 2.627, + "learning_rate": 8.317430483963758e-06, + "loss": 0.6977, + "step": 3630 + }, + { + "ETA": 13.03, + "epoch": 0.5704658543274693, + "fp16_scale": 512.0, + "global_step": 3640, + "grad_norm": 2.336, + "learning_rate": 8.267407077565206e-06, + "loss": 0.7021, + "step": 3640 + }, + { + "ETA": 12.98, + "epoch": 0.5720330682129844, + "fp16_scale": 1024.0, + "global_step": 3650, + "grad_norm": 2.043, + "learning_rate": 8.217428328700754e-06, + "loss": 0.6843, + "step": 3650 + }, + { + "ETA": 12.94, + "epoch": 0.5736002820984994, + "fp16_scale": 1024.0, + "global_step": 3660, + "grad_norm": 2.503, + "learning_rate": 8.167495525571249e-06, + "loss": 0.6849, + "step": 3660 + }, + { + "ETA": 12.89, + "epoch": 0.5751674959840144, + "fp16_scale": 1024.0, + "global_step": 3670, + "grad_norm": 8.048, + "learning_rate": 8.117609955193301e-06, + "loss": 0.7003, + "step": 3670 + }, + { + "ETA": 12.84, + "epoch": 0.5767347098695295, + "fp16_scale": 1024.0, + "global_step": 3680, + "grad_norm": 2.675, + "learning_rate": 8.067772903366087e-06, + "loss": 0.6441, + "step": 3680 + }, + { + "ETA": 12.79, + "epoch": 0.5783019237550445, + "fp16_scale": 1024.0, + "global_step": 3690, + "grad_norm": 2.125, + "learning_rate": 8.017985654638227e-06, + "loss": 0.6881, + "step": 3690 + }, + { + "ETA": 12.74, + "epoch": 0.5798691376405595, + "fp16_scale": 1024.0, + "global_step": 3700, + "grad_norm": 2.286, + "learning_rate": 7.96824949227466e-06, + "loss": 0.6574, + "step": 3700 + }, + { + "ETA": 12.69, + "epoch": 0.5814363515260745, + "fp16_scale": 1024.0, + "global_step": 3710, + "grad_norm": 2.344, + "learning_rate": 7.91856569822358e-06, + "loss": 0.6805, + "step": 3710 + }, + { + "ETA": 12.65, + "epoch": 0.5830035654115896, + "fp16_scale": 1024.0, + "global_step": 3720, + "grad_norm": 2.325, + "learning_rate": 7.868935553083391e-06, + "loss": 0.6865, + "step": 3720 + }, + { + "ETA": 12.6, + "epoch": 0.5845707792971045, + "fp16_scale": 1024.0, + "global_step": 3730, + "grad_norm": 3.179, + "learning_rate": 7.819360336069692e-06, + "loss": 0.6936, + "step": 3730 + }, + { + "ETA": 12.55, + "epoch": 0.5861379931826196, + "fp16_scale": 1024.0, + "global_step": 3740, + "grad_norm": 2.321, + "learning_rate": 7.769841324982315e-06, + "loss": 0.6764, + "step": 3740 + }, + { + "ETA": 12.5, + "epoch": 0.5877052070681347, + "fp16_scale": 1024.0, + "global_step": 3750, + "grad_norm": 2.531, + "learning_rate": 7.720379796172389e-06, + "loss": 0.6855, + "step": 3750 + }, + { + "ETA": 12.45, + "epoch": 0.5892724209536496, + "fp16_scale": 1024.0, + "global_step": 3760, + "grad_norm": 2.386, + "learning_rate": 7.670977024509432e-06, + "loss": 0.6872, + "step": 3760 + }, + { + "ETA": 12.41, + "epoch": 0.5908396348391647, + "fp16_scale": 1024.0, + "global_step": 3770, + "grad_norm": 2.79, + "learning_rate": 7.6216342833485e-06, + "loss": 0.6927, + "step": 3770 + }, + { + "ETA": 12.36, + "epoch": 0.5924068487246797, + "fp16_scale": 1024.0, + "global_step": 3780, + "grad_norm": 2.2, + "learning_rate": 7.572352844497369e-06, + "loss": 0.6648, + "step": 3780 + }, + { + "ETA": 12.31, + "epoch": 0.5939740626101947, + "fp16_scale": 1024.0, + "global_step": 3790, + "grad_norm": 2.125, + "learning_rate": 7.523133978183741e-06, + "loss": 0.6809, + "step": 3790 + }, + { + "ETA": 12.26, + "epoch": 0.5955412764957098, + "fp16_scale": 1024.0, + "global_step": 3800, + "grad_norm": 2.236, + "learning_rate": 7.4739789530225244e-06, + "loss": 0.6812, + "step": 3800 + }, + { + "ETA": 12.21, + "epoch": 0.5971084903812248, + "fp16_scale": 1024.0, + "global_step": 3810, + "grad_norm": 1.985, + "learning_rate": 7.424889035983116e-06, + "loss": 0.6879, + "step": 3810 + }, + { + "ETA": 12.16, + "epoch": 0.5986757042667398, + "fp16_scale": 1024.0, + "global_step": 3820, + "grad_norm": 3.346, + "learning_rate": 7.375865492356756e-06, + "loss": 0.6665, + "step": 3820 + }, + { + "ETA": 12.12, + "epoch": 0.6002429181522548, + "fp16_scale": 1024.0, + "global_step": 3830, + "grad_norm": 2.453, + "learning_rate": 7.326909585723901e-06, + "loss": 0.6788, + "step": 3830 + }, + { + "ETA": 12.07, + "epoch": 0.6018101320377699, + "fp16_scale": 1024.0, + "global_step": 3840, + "grad_norm": 2.547, + "learning_rate": 7.2780225779216825e-06, + "loss": 0.664, + "step": 3840 + }, + { + "ETA": 12.02, + "epoch": 0.6033773459232848, + "fp16_scale": 1024.0, + "global_step": 3850, + "grad_norm": 3.057, + "learning_rate": 7.22920572901136e-06, + "loss": 0.6709, + "step": 3850 + }, + { + "ETA": 11.97, + "epoch": 0.6049445598087999, + "fp16_scale": 1024.0, + "global_step": 3860, + "grad_norm": 2.438, + "learning_rate": 7.180460297245841e-06, + "loss": 0.668, + "step": 3860 + }, + { + "ETA": 11.92, + "epoch": 0.606511773694315, + "fp16_scale": 1024.0, + "global_step": 3870, + "grad_norm": 2.914, + "learning_rate": 7.131787539037269e-06, + "loss": 0.6619, + "step": 3870 + }, + { + "ETA": 11.88, + "epoch": 0.6080789875798299, + "fp16_scale": 1024.0, + "global_step": 3880, + "grad_norm": 3.034, + "learning_rate": 7.083188708924624e-06, + "loss": 0.6824, + "step": 3880 + }, + { + "ETA": 11.83, + "epoch": 0.609646201465345, + "fp16_scale": 1024.0, + "global_step": 3890, + "grad_norm": 2.359, + "learning_rate": 7.034665059541398e-06, + "loss": 0.665, + "step": 3890 + }, + { + "ETA": 11.78, + "epoch": 0.61121341535086, + "fp16_scale": 1024.0, + "global_step": 3900, + "grad_norm": 2.126, + "learning_rate": 6.98621784158329e-06, + "loss": 0.6999, + "step": 3900 + }, + { + "ETA": 11.73, + "epoch": 0.612780629236375, + "fp16_scale": 1024.0, + "global_step": 3910, + "grad_norm": 2.697, + "learning_rate": 6.937848303775986e-06, + "loss": 0.664, + "step": 3910 + }, + { + "ETA": 11.68, + "epoch": 0.6143478431218901, + "fp16_scale": 1024.0, + "global_step": 3920, + "grad_norm": 3.055, + "learning_rate": 6.889557692842976e-06, + "loss": 0.6968, + "step": 3920 + }, + { + "ETA": 11.64, + "epoch": 0.6159150570074051, + "fp16_scale": 1024.0, + "global_step": 3930, + "grad_norm": 2.679, + "learning_rate": 6.8413472534734e-06, + "loss": 0.6734, + "step": 3930 + }, + { + "ETA": 11.59, + "epoch": 0.6174822708929201, + "fp16_scale": 1024.0, + "global_step": 3940, + "grad_norm": 2.791, + "learning_rate": 6.793218228289986e-06, + "loss": 0.6812, + "step": 3940 + }, + { + "ETA": 11.54, + "epoch": 0.6190494847784351, + "fp16_scale": 1024.0, + "global_step": 3950, + "grad_norm": 2.579, + "learning_rate": 6.745171857817011e-06, + "loss": 0.663, + "step": 3950 + }, + { + "ETA": 11.49, + "epoch": 0.6206166986639502, + "fp16_scale": 1024.0, + "global_step": 3960, + "grad_norm": 2.878, + "learning_rate": 6.697209380448333e-06, + "loss": 0.6792, + "step": 3960 + }, + { + "ETA": 11.44, + "epoch": 0.6221839125494651, + "fp16_scale": 1024.0, + "global_step": 3970, + "grad_norm": 2.977, + "learning_rate": 6.649332032415459e-06, + "loss": 0.6796, + "step": 3970 + }, + { + "ETA": 11.39, + "epoch": 0.6237511264349802, + "fp16_scale": 1024.0, + "global_step": 3980, + "grad_norm": 2.891, + "learning_rate": 6.601541047755693e-06, + "loss": 0.6659, + "step": 3980 + }, + { + "ETA": 11.35, + "epoch": 0.6253183403204953, + "fp16_scale": 1024.0, + "global_step": 3990, + "grad_norm": 3.476, + "learning_rate": 6.5538376582803356e-06, + "loss": 0.694, + "step": 3990 + }, + { + "ETA": 11.3, + "epoch": 0.6268855542060102, + "fp16_scale": 1024.0, + "global_step": 4000, + "grad_norm": 2.474, + "learning_rate": 6.50622309354291e-06, + "loss": 0.6885, + "step": 4000 + }, + { + "ETA": 11.27, + "epoch": 0.6284527680915253, + "fp16_scale": 1024.0, + "global_step": 4010, + "grad_norm": 1.907, + "learning_rate": 6.458698580807495e-06, + "loss": 0.6842, + "step": 4010 + }, + { + "ETA": 11.22, + "epoch": 0.6300199819770403, + "fp16_scale": 1024.0, + "global_step": 4020, + "grad_norm": 3.952, + "learning_rate": 6.41126534501707e-06, + "loss": 0.694, + "step": 4020 + }, + { + "ETA": 11.17, + "epoch": 0.6315871958625553, + "fp16_scale": 1024.0, + "global_step": 4030, + "grad_norm": 1.945, + "learning_rate": 6.363924608761965e-06, + "loss": 0.6571, + "step": 4030 + }, + { + "ETA": 11.12, + "epoch": 0.6331544097480704, + "fp16_scale": 1024.0, + "global_step": 4040, + "grad_norm": 3.677, + "learning_rate": 6.316677592248331e-06, + "loss": 0.6579, + "step": 4040 + }, + { + "ETA": 11.08, + "epoch": 0.6347216236335854, + "fp16_scale": 1024.0, + "global_step": 4050, + "grad_norm": 2.402, + "learning_rate": 6.269525513266699e-06, + "loss": 0.6634, + "step": 4050 + }, + { + "ETA": 11.03, + "epoch": 0.6362888375191004, + "fp16_scale": 1024.0, + "global_step": 4060, + "grad_norm": 2.67, + "learning_rate": 6.22246958716058e-06, + "loss": 0.6879, + "step": 4060 + }, + { + "ETA": 10.98, + "epoch": 0.6378560514046154, + "fp16_scale": 1024.0, + "global_step": 4070, + "grad_norm": 2.808, + "learning_rate": 6.175511026795156e-06, + "loss": 0.6498, + "step": 4070 + }, + { + "ETA": 10.94, + "epoch": 0.6394232652901305, + "fp16_scale": 1024.0, + "global_step": 4080, + "grad_norm": 2.579, + "learning_rate": 6.128651042526006e-06, + "loss": 0.6925, + "step": 4080 + }, + { + "ETA": 10.89, + "epoch": 0.6409904791756454, + "fp16_scale": 1024.0, + "global_step": 4090, + "grad_norm": 2.52, + "learning_rate": 6.081890842167916e-06, + "loss": 0.6746, + "step": 4090 + }, + { + "ETA": 10.85, + "epoch": 0.6425576930611605, + "fp16_scale": 1024.0, + "global_step": 4100, + "grad_norm": 1.858, + "learning_rate": 6.035231630963737e-06, + "loss": 0.6818, + "step": 4100 + }, + { + "ETA": 10.8, + "epoch": 0.6441249069466756, + "fp16_scale": 1024.0, + "global_step": 4110, + "grad_norm": 2.397, + "learning_rate": 5.9886746115533356e-06, + "loss": 0.6525, + "step": 4110 + }, + { + "ETA": 10.75, + "epoch": 0.6456921208321905, + "fp16_scale": 1024.0, + "global_step": 4120, + "grad_norm": 2.411, + "learning_rate": 5.942220983942586e-06, + "loss": 0.6718, + "step": 4120 + }, + { + "ETA": 10.71, + "epoch": 0.6472593347177056, + "fp16_scale": 1024.0, + "global_step": 4130, + "grad_norm": 2.88, + "learning_rate": 5.895871945472434e-06, + "loss": 0.6455, + "step": 4130 + }, + { + "ETA": 10.66, + "epoch": 0.6488265486032206, + "fp16_scale": 1024.0, + "global_step": 4140, + "grad_norm": 2.438, + "learning_rate": 5.849628690788054e-06, + "loss": 0.6786, + "step": 4140 + }, + { + "ETA": 10.61, + "epoch": 0.6503937624887357, + "fp16_scale": 1024.0, + "global_step": 4150, + "grad_norm": 1.985, + "learning_rate": 5.803492411808042e-06, + "loss": 0.6437, + "step": 4150 + }, + { + "ETA": 10.56, + "epoch": 0.6519609763742507, + "fp16_scale": 1024.0, + "global_step": 4160, + "grad_norm": 2.398, + "learning_rate": 5.7574642976936945e-06, + "loss": 0.6379, + "step": 4160 + }, + { + "ETA": 10.51, + "epoch": 0.6535281902597657, + "fp16_scale": 1024.0, + "global_step": 4170, + "grad_norm": 2.936, + "learning_rate": 5.711545534818368e-06, + "loss": 0.6781, + "step": 4170 + }, + { + "ETA": 10.46, + "epoch": 0.6550954041452808, + "fp16_scale": 1024.0, + "global_step": 4180, + "grad_norm": 2.284, + "learning_rate": 5.665737306736889e-06, + "loss": 0.6458, + "step": 4180 + }, + { + "ETA": 10.42, + "epoch": 0.6566626180307957, + "fp16_scale": 1024.0, + "global_step": 4190, + "grad_norm": 2.256, + "learning_rate": 5.620040794155055e-06, + "loss": 0.6894, + "step": 4190 + }, + { + "ETA": 10.37, + "epoch": 0.6582298319163108, + "fp16_scale": 1024.0, + "global_step": 4200, + "grad_norm": 1.972, + "learning_rate": 5.5744571748992005e-06, + "loss": 0.6468, + "step": 4200 + }, + { + "ETA": 10.32, + "epoch": 0.6597970458018259, + "fp16_scale": 1024.0, + "global_step": 4210, + "grad_norm": 2.206, + "learning_rate": 5.528987623885834e-06, + "loss": 0.6404, + "step": 4210 + }, + { + "ETA": 10.27, + "epoch": 0.6613642596873408, + "fp16_scale": 1024.0, + "global_step": 4220, + "grad_norm": 1.952, + "learning_rate": 5.483633313091363e-06, + "loss": 0.6462, + "step": 4220 + }, + { + "ETA": 10.22, + "epoch": 0.6629314735728559, + "fp16_scale": 1024.0, + "global_step": 4230, + "grad_norm": 2.58, + "learning_rate": 5.4383954115218814e-06, + "loss": 0.6463, + "step": 4230 + }, + { + "ETA": 10.18, + "epoch": 0.6644986874583709, + "fp16_scale": 1024.0, + "global_step": 4240, + "grad_norm": 1.728, + "learning_rate": 5.393275085183029e-06, + "loss": 0.6645, + "step": 4240 + }, + { + "ETA": 10.13, + "epoch": 0.6660659013438859, + "fp16_scale": 1024.0, + "global_step": 4250, + "grad_norm": 2.223, + "learning_rate": 5.3482734970499605e-06, + "loss": 0.6531, + "step": 4250 + }, + { + "ETA": 10.08, + "epoch": 0.6676331152294009, + "fp16_scale": 1024.0, + "global_step": 4260, + "grad_norm": 2.054, + "learning_rate": 5.303391807037348e-06, + "loss": 0.6768, + "step": 4260 + }, + { + "ETA": 10.03, + "epoch": 0.669200329114916, + "fp16_scale": 1024.0, + "global_step": 4270, + "grad_norm": 2.35, + "learning_rate": 5.2586311719694974e-06, + "loss": 0.6359, + "step": 4270 + }, + { + "ETA": 9.98, + "epoch": 0.670767543000431, + "fp16_scale": 1024.0, + "global_step": 4280, + "grad_norm": 3.027, + "learning_rate": 5.213992745550524e-06, + "loss": 0.6412, + "step": 4280 + }, + { + "ETA": 9.94, + "epoch": 0.672334756885946, + "fp16_scale": 1024.0, + "global_step": 4290, + "grad_norm": 2.138, + "learning_rate": 5.16947767833462e-06, + "loss": 0.6668, + "step": 4290 + }, + { + "ETA": 9.89, + "epoch": 0.6739019707714611, + "fp16_scale": 1024.0, + "global_step": 4300, + "grad_norm": 2.281, + "learning_rate": 5.125087117696403e-06, + "loss": 0.6533, + "step": 4300 + }, + { + "ETA": 9.84, + "epoch": 0.675469184656976, + "fp16_scale": 1024.0, + "global_step": 4310, + "grad_norm": 2.512, + "learning_rate": 5.0808222078013214e-06, + "loss": 0.6543, + "step": 4310 + }, + { + "ETA": 9.79, + "epoch": 0.6770363985424911, + "fp16_scale": 1024.0, + "global_step": 4320, + "grad_norm": 2.208, + "learning_rate": 5.036684089576195e-06, + "loss": 0.6574, + "step": 4320 + }, + { + "ETA": 9.74, + "epoch": 0.6786036124280062, + "fp16_scale": 1024.0, + "global_step": 4330, + "grad_norm": 2.38, + "learning_rate": 4.992673900679784e-06, + "loss": 0.6942, + "step": 4330 + }, + { + "ETA": 9.7, + "epoch": 0.6801708263135211, + "fp16_scale": 1024.0, + "global_step": 4340, + "grad_norm": 1.937, + "learning_rate": 4.9487927754734775e-06, + "loss": 0.67, + "step": 4340 + }, + { + "ETA": 9.65, + "epoch": 0.6817380401990362, + "fp16_scale": 1024.0, + "global_step": 4350, + "grad_norm": 2.332, + "learning_rate": 4.90504184499205e-06, + "loss": 0.6347, + "step": 4350 + }, + { + "ETA": 9.6, + "epoch": 0.6833052540845512, + "fp16_scale": 1024.0, + "global_step": 4360, + "grad_norm": 3.316, + "learning_rate": 4.861422236914509e-06, + "loss": 0.6759, + "step": 4360 + }, + { + "ETA": 9.55, + "epoch": 0.6848724679700662, + "fp16_scale": 1024.0, + "global_step": 4370, + "grad_norm": 1.95, + "learning_rate": 4.817935075535033e-06, + "loss": 0.6925, + "step": 4370 + }, + { + "ETA": 9.5, + "epoch": 0.6864396818555812, + "fp16_scale": 1024.0, + "global_step": 4380, + "grad_norm": 2.019, + "learning_rate": 4.77458148173399e-06, + "loss": 0.6656, + "step": 4380 + }, + { + "ETA": 9.46, + "epoch": 0.6880068957410963, + "fp16_scale": 1024.0, + "global_step": 4390, + "grad_norm": 2.305, + "learning_rate": 4.7313625729490465e-06, + "loss": 0.6552, + "step": 4390 + }, + { + "ETA": 9.41, + "epoch": 0.6895741096266113, + "fp16_scale": 1024.0, + "global_step": 4400, + "grad_norm": 2.275, + "learning_rate": 4.688279463146368e-06, + "loss": 0.6474, + "step": 4400 + }, + { + "ETA": 9.36, + "epoch": 0.6911413235121263, + "fp16_scale": 1024.0, + "global_step": 4410, + "grad_norm": 2.844, + "learning_rate": 4.645333262791904e-06, + "loss": 0.6487, + "step": 4410 + }, + { + "ETA": 9.31, + "epoch": 0.6927085373976414, + "fp16_scale": 1024.0, + "global_step": 4420, + "grad_norm": 2.353, + "learning_rate": 4.602525078822766e-06, + "loss": 0.6221, + "step": 4420 + }, + { + "ETA": 9.26, + "epoch": 0.6942757512831563, + "fp16_scale": 1024.0, + "global_step": 4430, + "grad_norm": 2.665, + "learning_rate": 4.559856014618703e-06, + "loss": 0.6526, + "step": 4430 + }, + { + "ETA": 9.22, + "epoch": 0.6958429651686714, + "fp16_scale": 1024.0, + "global_step": 4440, + "grad_norm": 2.108, + "learning_rate": 4.517327169973643e-06, + "loss": 0.6532, + "step": 4440 + }, + { + "ETA": 9.17, + "epoch": 0.6974101790541865, + "fp16_scale": 1024.0, + "global_step": 4450, + "grad_norm": 2.701, + "learning_rate": 4.474939641067371e-06, + "loss": 0.6335, + "step": 4450 + }, + { + "ETA": 9.12, + "epoch": 0.6989773929397014, + "fp16_scale": 1024.0, + "global_step": 4460, + "grad_norm": 2.679, + "learning_rate": 4.432694520437257e-06, + "loss": 0.6579, + "step": 4460 + }, + { + "ETA": 9.07, + "epoch": 0.7005446068252165, + "fp16_scale": 1024.0, + "global_step": 4470, + "grad_norm": 1.97, + "learning_rate": 4.3905928969501054e-06, + "loss": 0.6379, + "step": 4470 + }, + { + "ETA": 9.02, + "epoch": 0.7021118207107315, + "fp16_scale": 1024.0, + "global_step": 4480, + "grad_norm": 3.684, + "learning_rate": 4.348635855774082e-06, + "loss": 0.6785, + "step": 4480 + }, + { + "ETA": 8.98, + "epoch": 0.7036790345962465, + "fp16_scale": 1024.0, + "global_step": 4490, + "grad_norm": 2.484, + "learning_rate": 4.30682447835075e-06, + "loss": 0.6561, + "step": 4490 + }, + { + "ETA": 8.93, + "epoch": 0.7052462484817615, + "fp16_scale": 1024.0, + "global_step": 4500, + "grad_norm": 3.453, + "learning_rate": 4.265159842367195e-06, + "loss": 0.6366, + "step": 4500 + }, + { + "ETA": 8.89, + "epoch": 0.7068134623672766, + "fp16_scale": 1024.0, + "global_step": 4510, + "grad_norm": 2.174, + "learning_rate": 4.223643021728237e-06, + "loss": 0.6589, + "step": 4510 + }, + { + "ETA": 8.84, + "epoch": 0.7083806762527916, + "fp16_scale": 1024.0, + "global_step": 4520, + "grad_norm": 2.416, + "learning_rate": 4.182275086528771e-06, + "loss": 0.6656, + "step": 4520 + }, + { + "ETA": 8.8, + "epoch": 0.7099478901383066, + "fp16_scale": 1024.0, + "global_step": 4530, + "grad_norm": 1.731, + "learning_rate": 4.141057103026168e-06, + "loss": 0.6437, + "step": 4530 + }, + { + "ETA": 8.75, + "epoch": 0.7115151040238217, + "fp16_scale": 1024.0, + "global_step": 4540, + "grad_norm": 2.158, + "learning_rate": 4.0999901336128015e-06, + "loss": 0.6332, + "step": 4540 + }, + { + "ETA": 8.7, + "epoch": 0.7130823179093366, + "fp16_scale": 1024.0, + "global_step": 4550, + "grad_norm": 2.984, + "learning_rate": 4.059075236788656e-06, + "loss": 0.6338, + "step": 4550 + }, + { + "ETA": 8.65, + "epoch": 0.7146495317948517, + "fp16_scale": 1024.0, + "global_step": 4560, + "grad_norm": 2.391, + "learning_rate": 4.018313467134054e-06, + "loss": 0.643, + "step": 4560 + }, + { + "ETA": 8.61, + "epoch": 0.7162167456803668, + "fp16_scale": 1024.0, + "global_step": 4570, + "grad_norm": 2.143, + "learning_rate": 3.977705875282468e-06, + "loss": 0.6483, + "step": 4570 + }, + { + "ETA": 8.56, + "epoch": 0.7177839595658817, + "fp16_scale": 1024.0, + "global_step": 4580, + "grad_norm": 2.292, + "learning_rate": 3.9372535078934415e-06, + "loss": 0.6624, + "step": 4580 + }, + { + "ETA": 8.51, + "epoch": 0.7193511734513968, + "fp16_scale": 1024.0, + "global_step": 4590, + "grad_norm": 2.035, + "learning_rate": 3.896957407625612e-06, + "loss": 0.6628, + "step": 4590 + }, + { + "ETA": 8.46, + "epoch": 0.7209183873369118, + "fp16_scale": 1024.0, + "global_step": 4600, + "grad_norm": 2.281, + "learning_rate": 3.856818613109836e-06, + "loss": 0.6493, + "step": 4600 + }, + { + "ETA": 8.41, + "epoch": 0.7224856012224268, + "fp16_scale": 1024.0, + "global_step": 4610, + "grad_norm": 2.059, + "learning_rate": 3.81683815892242e-06, + "loss": 0.6674, + "step": 4610 + }, + { + "ETA": 8.37, + "epoch": 0.7240528151079418, + "fp16_scale": 1024.0, + "global_step": 4620, + "grad_norm": 2.233, + "learning_rate": 3.7770170755584524e-06, + "loss": 0.6551, + "step": 4620 + }, + { + "ETA": 8.32, + "epoch": 0.7256200289934569, + "fp16_scale": 1024.0, + "global_step": 4630, + "grad_norm": 2.489, + "learning_rate": 3.7373563894052444e-06, + "loss": 0.6574, + "step": 4630 + }, + { + "ETA": 8.28, + "epoch": 0.727187242878972, + "fp16_scale": 1024.0, + "global_step": 4640, + "grad_norm": 2.365, + "learning_rate": 3.6978571227158655e-06, + "loss": 0.6672, + "step": 4640 + }, + { + "ETA": 8.23, + "epoch": 0.7287544567644869, + "fp16_scale": 2048.0, + "global_step": 4650, + "grad_norm": 2.451, + "learning_rate": 3.6585202935828133e-06, + "loss": 0.6598, + "step": 4650 + }, + { + "ETA": 8.18, + "epoch": 0.730321670650002, + "fp16_scale": 2048.0, + "global_step": 4660, + "grad_norm": 2.212, + "learning_rate": 3.619346915911759e-06, + "loss": 0.6387, + "step": 4660 + }, + { + "ETA": 8.13, + "epoch": 0.731888884535517, + "fp16_scale": 2048.0, + "global_step": 4670, + "grad_norm": 2.04, + "learning_rate": 3.5803379993954146e-06, + "loss": 0.6797, + "step": 4670 + }, + { + "ETA": 8.09, + "epoch": 0.733456098421032, + "fp16_scale": 2048.0, + "global_step": 4680, + "grad_norm": 2.082, + "learning_rate": 3.541494549487512e-06, + "loss": 0.6459, + "step": 4680 + }, + { + "ETA": 8.04, + "epoch": 0.735023312306547, + "fp16_scale": 2048.0, + "global_step": 4690, + "grad_norm": 2.181, + "learning_rate": 3.502817567376886e-06, + "loss": 0.6517, + "step": 4690 + }, + { + "ETA": 7.99, + "epoch": 0.7365905261920621, + "fp16_scale": 2048.0, + "global_step": 4700, + "grad_norm": 1.914, + "learning_rate": 3.464308049961671e-06, + "loss": 0.6311, + "step": 4700 + }, + { + "ETA": 7.94, + "epoch": 0.7381577400775771, + "fp16_scale": 2048.0, + "global_step": 4710, + "grad_norm": 2.169, + "learning_rate": 3.425966989823596e-06, + "loss": 0.6503, + "step": 4710 + }, + { + "ETA": 7.89, + "epoch": 0.7397249539630921, + "fp16_scale": 2048.0, + "global_step": 4720, + "grad_norm": 1.835, + "learning_rate": 3.387795375202414e-06, + "loss": 0.6289, + "step": 4720 + }, + { + "ETA": 7.85, + "epoch": 0.7412921678486072, + "fp16_scale": 2048.0, + "global_step": 4730, + "grad_norm": 2.463, + "learning_rate": 3.3497941899704267e-06, + "loss": 0.6407, + "step": 4730 + }, + { + "ETA": 7.8, + "epoch": 0.7428593817341221, + "fp16_scale": 2048.0, + "global_step": 4740, + "grad_norm": 3.038, + "learning_rate": 3.311964413607117e-06, + "loss": 0.6588, + "step": 4740 + }, + { + "ETA": 7.75, + "epoch": 0.7444265956196372, + "fp16_scale": 2048.0, + "global_step": 4750, + "grad_norm": 2.629, + "learning_rate": 3.274307021173911e-06, + "loss": 0.624, + "step": 4750 + }, + { + "ETA": 7.7, + "epoch": 0.7459938095051523, + "fp16_scale": 2048.0, + "global_step": 4760, + "grad_norm": 2.266, + "learning_rate": 3.2368229832890565e-06, + "loss": 0.6704, + "step": 4760 + }, + { + "ETA": 7.66, + "epoch": 0.7475610233906672, + "fp16_scale": 2048.0, + "global_step": 4770, + "grad_norm": 2.528, + "learning_rate": 3.1995132661025673e-06, + "loss": 0.6352, + "step": 4770 + }, + { + "ETA": 7.61, + "epoch": 0.7491282372761823, + "fp16_scale": 2048.0, + "global_step": 4780, + "grad_norm": 2.439, + "learning_rate": 3.1623788312713656e-06, + "loss": 0.6619, + "step": 4780 + }, + { + "ETA": 7.56, + "epoch": 0.7506954511616973, + "fp16_scale": 2048.0, + "global_step": 4790, + "grad_norm": 2.292, + "learning_rate": 3.12542063593447e-06, + "loss": 0.6274, + "step": 4790 + }, + { + "ETA": 7.51, + "epoch": 0.7522626650472123, + "fp16_scale": 2048.0, + "global_step": 4800, + "grad_norm": 3.361, + "learning_rate": 3.0886396326883305e-06, + "loss": 0.6196, + "step": 4800 + }, + { + "ETA": 7.46, + "epoch": 0.7538298789327273, + "fp16_scale": 2048.0, + "global_step": 4810, + "grad_norm": 10.937, + "learning_rate": 3.0520367695622766e-06, + "loss": 0.6547, + "step": 4810 + }, + { + "ETA": 7.42, + "epoch": 0.7553970928182424, + "fp16_scale": 2048.0, + "global_step": 4820, + "grad_norm": 3.288, + "learning_rate": 3.015612989994082e-06, + "loss": 0.6345, + "step": 4820 + }, + { + "ETA": 7.37, + "epoch": 0.7569643067037574, + "fp16_scale": 2048.0, + "global_step": 4830, + "grad_norm": 2.648, + "learning_rate": 2.97936923280565e-06, + "loss": 0.6281, + "step": 4830 + }, + { + "ETA": 7.32, + "epoch": 0.7585315205892724, + "fp16_scale": 2048.0, + "global_step": 4840, + "grad_norm": 2.144, + "learning_rate": 2.9433064321787996e-06, + "loss": 0.6253, + "step": 4840 + }, + { + "ETA": 7.27, + "epoch": 0.7600987344747875, + "fp16_scale": 2048.0, + "global_step": 4850, + "grad_norm": 3.614, + "learning_rate": 2.9074255176312115e-06, + "loss": 0.6505, + "step": 4850 + }, + { + "ETA": 7.22, + "epoch": 0.7616659483603024, + "fp16_scale": 2048.0, + "global_step": 4860, + "grad_norm": 2.036, + "learning_rate": 2.8717274139924556e-06, + "loss": 0.6299, + "step": 4860 + }, + { + "ETA": 7.18, + "epoch": 0.7632331622458175, + "fp16_scale": 2048.0, + "global_step": 4870, + "grad_norm": 3.092, + "learning_rate": 2.8362130413801524e-06, + "loss": 0.6549, + "step": 4870 + }, + { + "ETA": 7.13, + "epoch": 0.7648003761313326, + "fp16_scale": 2048.0, + "global_step": 4880, + "grad_norm": 2.695, + "learning_rate": 2.8008833151762636e-06, + "loss": 0.6202, + "step": 4880 + }, + { + "ETA": 7.08, + "epoch": 0.7663675900168475, + "fp16_scale": 2048.0, + "global_step": 4890, + "grad_norm": 2.407, + "learning_rate": 2.765739146003493e-06, + "loss": 0.6084, + "step": 4890 + }, + { + "ETA": 7.03, + "epoch": 0.7679348039023626, + "fp16_scale": 2048.0, + "global_step": 4900, + "grad_norm": 2.132, + "learning_rate": 2.7307814397018217e-06, + "loss": 0.6688, + "step": 4900 + }, + { + "ETA": 6.99, + "epoch": 0.7695020177878776, + "fp16_scale": 2048.0, + "global_step": 4910, + "grad_norm": 2.43, + "learning_rate": 2.6960110973051445e-06, + "loss": 0.6316, + "step": 4910 + }, + { + "ETA": 6.94, + "epoch": 0.7710692316733926, + "fp16_scale": 2048.0, + "global_step": 4920, + "grad_norm": 2.563, + "learning_rate": 2.6614290150180667e-06, + "loss": 0.6349, + "step": 4920 + }, + { + "ETA": 6.89, + "epoch": 0.7726364455589076, + "fp16_scale": 2048.0, + "global_step": 4930, + "grad_norm": 2.146, + "learning_rate": 2.6270360841927924e-06, + "loss": 0.6226, + "step": 4930 + }, + { + "ETA": 6.84, + "epoch": 0.7742036594444227, + "fp16_scale": 2048.0, + "global_step": 4940, + "grad_norm": 2.0, + "learning_rate": 2.592833191306151e-06, + "loss": 0.6323, + "step": 4940 + }, + { + "ETA": 6.79, + "epoch": 0.7757708733299377, + "fp16_scale": 2048.0, + "global_step": 4950, + "grad_norm": 4.297, + "learning_rate": 2.5588212179367487e-06, + "loss": 0.6356, + "step": 4950 + }, + { + "ETA": 6.75, + "epoch": 0.7773380872154527, + "fp16_scale": 2048.0, + "global_step": 4960, + "grad_norm": 2.479, + "learning_rate": 2.5250010407422485e-06, + "loss": 0.6219, + "step": 4960 + }, + { + "ETA": 6.7, + "epoch": 0.7789053011009678, + "fp16_scale": 2048.0, + "global_step": 4970, + "grad_norm": 2.719, + "learning_rate": 2.4913735314367715e-06, + "loss": 0.6485, + "step": 4970 + }, + { + "ETA": 6.65, + "epoch": 0.7804725149864827, + "fp16_scale": 2048.0, + "global_step": 4980, + "grad_norm": 2.164, + "learning_rate": 2.4579395567684284e-06, + "loss": 0.646, + "step": 4980 + }, + { + "ETA": 6.6, + "epoch": 0.7820397288719978, + "fp16_scale": 2048.0, + "global_step": 4990, + "grad_norm": 2.672, + "learning_rate": 2.4246999784969817e-06, + "loss": 0.6229, + "step": 4990 + }, + { + "ETA": 6.56, + "epoch": 0.7836069427575129, + "fp16_scale": 2048.0, + "global_step": 5000, + "grad_norm": 2.134, + "learning_rate": 2.3916556533716296e-06, + "loss": 0.62, + "step": 5000 + }, + { + "ETA": 6.52, + "epoch": 0.7851741566430278, + "fp16_scale": 2048.0, + "global_step": 5010, + "grad_norm": 2.442, + "learning_rate": 2.3588074331089296e-06, + "loss": 0.6711, + "step": 5010 + }, + { + "ETA": 6.47, + "epoch": 0.7867413705285429, + "fp16_scale": 2048.0, + "global_step": 5020, + "grad_norm": 2.502, + "learning_rate": 2.3261561643708387e-06, + "loss": 0.635, + "step": 5020 + }, + { + "ETA": 6.42, + "epoch": 0.7883085844140579, + "fp16_scale": 2048.0, + "global_step": 5030, + "grad_norm": 2.263, + "learning_rate": 2.293702688742898e-06, + "loss": 0.619, + "step": 5030 + }, + { + "ETA": 6.37, + "epoch": 0.7898757982995729, + "fp16_scale": 2048.0, + "global_step": 5040, + "grad_norm": 2.295, + "learning_rate": 2.2614478427125285e-06, + "loss": 0.6496, + "step": 5040 + }, + { + "ETA": 6.32, + "epoch": 0.791443012185088, + "fp16_scale": 2048.0, + "global_step": 5050, + "grad_norm": 2.363, + "learning_rate": 2.229392457647486e-06, + "loss": 0.6396, + "step": 5050 + }, + { + "ETA": 6.28, + "epoch": 0.793010226070603, + "fp16_scale": 2048.0, + "global_step": 5060, + "grad_norm": 2.289, + "learning_rate": 2.1975373597744242e-06, + "loss": 0.6366, + "step": 5060 + }, + { + "ETA": 6.23, + "epoch": 0.794577439956118, + "fp16_scale": 2048.0, + "global_step": 5070, + "grad_norm": 2.387, + "learning_rate": 2.165883370157601e-06, + "loss": 0.6459, + "step": 5070 + }, + { + "ETA": 6.18, + "epoch": 0.796144653841633, + "fp16_scale": 2048.0, + "global_step": 5080, + "grad_norm": 2.12, + "learning_rate": 2.1344313046777097e-06, + "loss": 0.6552, + "step": 5080 + }, + { + "ETA": 6.13, + "epoch": 0.7977118677271481, + "fp16_scale": 2048.0, + "global_step": 5090, + "grad_norm": 1.951, + "learning_rate": 2.1031819740108616e-06, + "loss": 0.6027, + "step": 5090 + }, + { + "ETA": 6.09, + "epoch": 0.799279081612663, + "fp16_scale": 2048.0, + "global_step": 5100, + "grad_norm": 1.878, + "learning_rate": 2.0721361836076814e-06, + "loss": 0.644, + "step": 5100 + }, + { + "ETA": 6.04, + "epoch": 0.8008462954981781, + "fp16_scale": 2048.0, + "global_step": 5110, + "grad_norm": 2.002, + "learning_rate": 2.041294733672542e-06, + "loss": 0.6535, + "step": 5110 + }, + { + "ETA": 5.99, + "epoch": 0.8024135093836932, + "fp16_scale": 2048.0, + "global_step": 5120, + "grad_norm": 2.605, + "learning_rate": 2.0106584191429556e-06, + "loss": 0.658, + "step": 5120 + }, + { + "ETA": 5.94, + "epoch": 0.8039807232692082, + "fp16_scale": 2048.0, + "global_step": 5130, + "grad_norm": 3.932, + "learning_rate": 1.9802280296690722e-06, + "loss": 0.6403, + "step": 5130 + }, + { + "ETA": 5.89, + "epoch": 0.8055479371547232, + "fp16_scale": 2048.0, + "global_step": 5140, + "grad_norm": 1.911, + "learning_rate": 1.9500043495933275e-06, + "loss": 0.6264, + "step": 5140 + }, + { + "ETA": 5.85, + "epoch": 0.8071151510402382, + "fp16_scale": 2048.0, + "global_step": 5150, + "grad_norm": 2.622, + "learning_rate": 1.919988157930236e-06, + "loss": 0.6415, + "step": 5150 + }, + { + "ETA": 5.8, + "epoch": 0.8086823649257533, + "fp16_scale": 2048.0, + "global_step": 5160, + "grad_norm": 3.352, + "learning_rate": 1.89018022834629e-06, + "loss": 0.6679, + "step": 5160 + }, + { + "ETA": 5.75, + "epoch": 0.8102495788112682, + "fp16_scale": 2048.0, + "global_step": 5170, + "grad_norm": 2.792, + "learning_rate": 1.8605813291400444e-06, + "loss": 0.6317, + "step": 5170 + }, + { + "ETA": 5.71, + "epoch": 0.8118167926967833, + "fp16_scale": 2048.0, + "global_step": 5180, + "grad_norm": 2.251, + "learning_rate": 1.8311922232222979e-06, + "loss": 0.6364, + "step": 5180 + }, + { + "ETA": 5.66, + "epoch": 0.8133840065822984, + "fp16_scale": 2048.0, + "global_step": 5190, + "grad_norm": 2.3, + "learning_rate": 1.8020136680964329e-06, + "loss": 0.6312, + "step": 5190 + }, + { + "ETA": 5.61, + "epoch": 0.8149512204678133, + "fp16_scale": 2048.0, + "global_step": 5200, + "grad_norm": 2.282, + "learning_rate": 1.7730464158388906e-06, + "loss": 0.6105, + "step": 5200 + }, + { + "ETA": 5.56, + "epoch": 0.8165184343533284, + "fp16_scale": 2048.0, + "global_step": 5210, + "grad_norm": 2.271, + "learning_rate": 1.7442912130797883e-06, + "loss": 0.6232, + "step": 5210 + }, + { + "ETA": 5.52, + "epoch": 0.8180856482388434, + "fp16_scale": 2048.0, + "global_step": 5220, + "grad_norm": 2.368, + "learning_rate": 1.715748800983672e-06, + "loss": 0.6293, + "step": 5220 + }, + { + "ETA": 5.47, + "epoch": 0.8196528621243584, + "fp16_scale": 2048.0, + "global_step": 5230, + "grad_norm": 2.348, + "learning_rate": 1.6874199152304182e-06, + "loss": 0.6553, + "step": 5230 + }, + { + "ETA": 5.42, + "epoch": 0.8212200760098735, + "fp16_scale": 2048.0, + "global_step": 5240, + "grad_norm": 1.922, + "learning_rate": 1.659305285996259e-06, + "loss": 0.6178, + "step": 5240 + }, + { + "ETA": 5.37, + "epoch": 0.8227872898953885, + "fp16_scale": 2048.0, + "global_step": 5250, + "grad_norm": 2.16, + "learning_rate": 1.6314056379349784e-06, + "loss": 0.641, + "step": 5250 + }, + { + "ETA": 5.33, + "epoch": 0.8243545037809035, + "fp16_scale": 2048.0, + "global_step": 5260, + "grad_norm": 1.997, + "learning_rate": 1.6037216901592245e-06, + "loss": 0.6361, + "step": 5260 + }, + { + "ETA": 5.28, + "epoch": 0.8259217176664185, + "fp16_scale": 2048.0, + "global_step": 5270, + "grad_norm": 2.167, + "learning_rate": 1.5762541562219779e-06, + "loss": 0.6505, + "step": 5270 + }, + { + "ETA": 5.23, + "epoch": 0.8274889315519336, + "fp16_scale": 2048.0, + "global_step": 5280, + "grad_norm": 2.206, + "learning_rate": 1.5490037440981564e-06, + "loss": 0.624, + "step": 5280 + }, + { + "ETA": 5.18, + "epoch": 0.8290561454374485, + "fp16_scale": 2048.0, + "global_step": 5290, + "grad_norm": 2.148, + "learning_rate": 1.5219711561663697e-06, + "loss": 0.6093, + "step": 5290 + }, + { + "ETA": 5.13, + "epoch": 0.8306233593229636, + "fp16_scale": 2048.0, + "global_step": 5300, + "grad_norm": 2.247, + "learning_rate": 1.4951570891908174e-06, + "loss": 0.6228, + "step": 5300 + }, + { + "ETA": 5.09, + "epoch": 0.8321905732084787, + "fp16_scale": 2048.0, + "global_step": 5310, + "grad_norm": 3.03, + "learning_rate": 1.4685622343033223e-06, + "loss": 0.6093, + "step": 5310 + }, + { + "ETA": 5.04, + "epoch": 0.8337577870939936, + "fp16_scale": 2048.0, + "global_step": 5320, + "grad_norm": 2.311, + "learning_rate": 1.4421872769855262e-06, + "loss": 0.6365, + "step": 5320 + }, + { + "ETA": 4.99, + "epoch": 0.8353250009795087, + "fp16_scale": 2048.0, + "global_step": 5330, + "grad_norm": 1.918, + "learning_rate": 1.4160328970512149e-06, + "loss": 0.6482, + "step": 5330 + }, + { + "ETA": 4.94, + "epoch": 0.8368922148650237, + "fp16_scale": 2048.0, + "global_step": 5340, + "grad_norm": 2.109, + "learning_rate": 1.390099768628802e-06, + "loss": 0.6297, + "step": 5340 + }, + { + "ETA": 4.9, + "epoch": 0.8384594287505387, + "fp16_scale": 2048.0, + "global_step": 5350, + "grad_norm": 2.198, + "learning_rate": 1.3643885601439488e-06, + "loss": 0.6396, + "step": 5350 + }, + { + "ETA": 4.85, + "epoch": 0.8400266426360538, + "fp16_scale": 2048.0, + "global_step": 5360, + "grad_norm": 2.682, + "learning_rate": 1.3388999343023278e-06, + "loss": 0.6397, + "step": 5360 + }, + { + "ETA": 4.8, + "epoch": 0.8415938565215688, + "fp16_scale": 2048.0, + "global_step": 5370, + "grad_norm": 2.607, + "learning_rate": 1.3136345480725621e-06, + "loss": 0.6441, + "step": 5370 + }, + { + "ETA": 4.75, + "epoch": 0.8431610704070838, + "fp16_scale": 2048.0, + "global_step": 5380, + "grad_norm": 2.553, + "learning_rate": 1.2885930526692736e-06, + "loss": 0.659, + "step": 5380 + }, + { + "ETA": 4.71, + "epoch": 0.8447282842925988, + "fp16_scale": 2048.0, + "global_step": 5390, + "grad_norm": 2.285, + "learning_rate": 1.2637760935363053e-06, + "loss": 0.6154, + "step": 5390 + }, + { + "ETA": 4.66, + "epoch": 0.8462954981781139, + "fp16_scale": 2048.0, + "global_step": 5400, + "grad_norm": 2.19, + "learning_rate": 1.2391843103300838e-06, + "loss": 0.6418, + "step": 5400 + }, + { + "ETA": 4.61, + "epoch": 0.8478627120636288, + "fp16_scale": 2048.0, + "global_step": 5410, + "grad_norm": 2.141, + "learning_rate": 1.214818336903134e-06, + "loss": 0.6329, + "step": 5410 + }, + { + "ETA": 4.56, + "epoch": 0.8494299259491439, + "fp16_scale": 2048.0, + "global_step": 5420, + "grad_norm": 2.279, + "learning_rate": 1.1906788012877423e-06, + "loss": 0.6217, + "step": 5420 + }, + { + "ETA": 4.51, + "epoch": 0.850997139834659, + "fp16_scale": 2048.0, + "global_step": 5430, + "grad_norm": 2.33, + "learning_rate": 1.1667663256797578e-06, + "loss": 0.6205, + "step": 5430 + }, + { + "ETA": 4.47, + "epoch": 0.8525643537201739, + "fp16_scale": 2048.0, + "global_step": 5440, + "grad_norm": 2.216, + "learning_rate": 1.1430815264225737e-06, + "loss": 0.6635, + "step": 5440 + }, + { + "ETA": 4.42, + "epoch": 0.854131567605689, + "fp16_scale": 2048.0, + "global_step": 5450, + "grad_norm": 2.259, + "learning_rate": 1.1196250139912268e-06, + "loss": 0.6828, + "step": 5450 + }, + { + "ETA": 4.37, + "epoch": 0.855698781491204, + "fp16_scale": 2048.0, + "global_step": 5460, + "grad_norm": 2.969, + "learning_rate": 1.096397392976669e-06, + "loss": 0.6305, + "step": 5460 + }, + { + "ETA": 4.32, + "epoch": 0.857265995376719, + "fp16_scale": 2048.0, + "global_step": 5470, + "grad_norm": 2.021, + "learning_rate": 1.073399262070184e-06, + "loss": 0.6284, + "step": 5470 + }, + { + "ETA": 4.28, + "epoch": 0.858833209262234, + "fp16_scale": 2048.0, + "global_step": 5480, + "grad_norm": 2.051, + "learning_rate": 1.0506312140479502e-06, + "loss": 0.6396, + "step": 5480 + }, + { + "ETA": 4.23, + "epoch": 0.8604004231477491, + "fp16_scale": 2048.0, + "global_step": 5490, + "grad_norm": 2.332, + "learning_rate": 1.028093835755769e-06, + "loss": 0.6275, + "step": 5490 + }, + { + "ETA": 4.18, + "epoch": 0.8619676370332641, + "fp16_scale": 2048.0, + "global_step": 5500, + "grad_norm": 3.388, + "learning_rate": 1.005787708093937e-06, + "loss": 0.6016, + "step": 5500 + }, + { + "ETA": 4.14, + "epoch": 0.8635348509187791, + "fp16_scale": 2048.0, + "global_step": 5510, + "grad_norm": 2.186, + "learning_rate": 9.837134060022668e-07, + "loss": 0.6251, + "step": 5510 + }, + { + "ETA": 4.09, + "epoch": 0.8651020648042942, + "fp16_scale": 2048.0, + "global_step": 5520, + "grad_norm": 2.527, + "learning_rate": 9.618714984452793e-07, + "loss": 0.6447, + "step": 5520 + }, + { + "ETA": 4.04, + "epoch": 0.8666692786898091, + "fp16_scale": 2048.0, + "global_step": 5530, + "grad_norm": 3.872, + "learning_rate": 9.40262548397528e-07, + "loss": 0.6198, + "step": 5530 + }, + { + "ETA": 3.99, + "epoch": 0.8682364925753242, + "fp16_scale": 2048.0, + "global_step": 5540, + "grad_norm": 1.967, + "learning_rate": 9.18887112829101e-07, + "loss": 0.6496, + "step": 5540 + }, + { + "ETA": 3.95, + "epoch": 0.8698037064608393, + "fp16_scale": 2048.0, + "global_step": 5550, + "grad_norm": 3.834, + "learning_rate": 8.977457426912517e-07, + "loss": 0.6364, + "step": 5550 + }, + { + "ETA": 3.9, + "epoch": 0.8713709203463542, + "fp16_scale": 2048.0, + "global_step": 5560, + "grad_norm": 2.293, + "learning_rate": 8.768389829021984e-07, + "loss": 0.6268, + "step": 5560 + }, + { + "ETA": 3.85, + "epoch": 0.8729381342318693, + "fp16_scale": 2048.0, + "global_step": 5570, + "grad_norm": 2.842, + "learning_rate": 8.561673723330932e-07, + "loss": 0.6419, + "step": 5570 + }, + { + "ETA": 3.8, + "epoch": 0.8745053481173843, + "fp16_scale": 2048.0, + "global_step": 5580, + "grad_norm": 2.526, + "learning_rate": 8.3573144379412e-07, + "loss": 0.6225, + "step": 5580 + }, + { + "ETA": 3.76, + "epoch": 0.8760725620028993, + "fp16_scale": 2048.0, + "global_step": 5590, + "grad_norm": 2.434, + "learning_rate": 8.15531724020765e-07, + "loss": 0.6463, + "step": 5590 + }, + { + "ETA": 3.71, + "epoch": 0.8776397758884144, + "fp16_scale": 2048.0, + "global_step": 5600, + "grad_norm": 2.044, + "learning_rate": 7.955687336602391e-07, + "loss": 0.64, + "step": 5600 + }, + { + "ETA": 3.66, + "epoch": 0.8792069897739294, + "fp16_scale": 2048.0, + "global_step": 5610, + "grad_norm": 4.355, + "learning_rate": 7.758429872580608e-07, + "loss": 0.6305, + "step": 5610 + }, + { + "ETA": 3.61, + "epoch": 0.8807742036594445, + "fp16_scale": 2048.0, + "global_step": 5620, + "grad_norm": 2.274, + "learning_rate": 7.563549932447944e-07, + "loss": 0.6042, + "step": 5620 + }, + { + "ETA": 3.57, + "epoch": 0.8823414175449594, + "fp16_scale": 2048.0, + "global_step": 5630, + "grad_norm": 2.577, + "learning_rate": 7.371052539229362e-07, + "loss": 0.6229, + "step": 5630 + }, + { + "ETA": 3.52, + "epoch": 0.8839086314304745, + "fp16_scale": 2048.0, + "global_step": 5640, + "grad_norm": 2.202, + "learning_rate": 7.180942654539802e-07, + "loss": 0.6616, + "step": 5640 + }, + { + "ETA": 3.47, + "epoch": 0.8854758453159896, + "fp16_scale": 4096.0, + "global_step": 5650, + "grad_norm": 3.963, + "learning_rate": 6.99322517845622e-07, + "loss": 0.6466, + "step": 5650 + }, + { + "ETA": 3.42, + "epoch": 0.8870430592015045, + "fp16_scale": 4096.0, + "global_step": 5660, + "grad_norm": 2.242, + "learning_rate": 6.807904949391319e-07, + "loss": 0.6464, + "step": 5660 + }, + { + "ETA": 3.37, + "epoch": 0.8886102730870196, + "fp16_scale": 4096.0, + "global_step": 5670, + "grad_norm": 2.595, + "learning_rate": 6.624986743968809e-07, + "loss": 0.6401, + "step": 5670 + }, + { + "ETA": 3.33, + "epoch": 0.8901774869725346, + "fp16_scale": 4096.0, + "global_step": 5680, + "grad_norm": 2.086, + "learning_rate": 6.44447527690032e-07, + "loss": 0.6126, + "step": 5680 + }, + { + "ETA": 3.28, + "epoch": 0.8917447008580496, + "fp16_scale": 4096.0, + "global_step": 5690, + "grad_norm": 2.494, + "learning_rate": 6.266375200863884e-07, + "loss": 0.613, + "step": 5690 + }, + { + "ETA": 3.23, + "epoch": 0.8933119147435646, + "fp16_scale": 4096.0, + "global_step": 5700, + "grad_norm": 2.06, + "learning_rate": 6.090691106383939e-07, + "loss": 0.6201, + "step": 5700 + }, + { + "ETA": 3.18, + "epoch": 0.8948791286290797, + "fp16_scale": 4096.0, + "global_step": 5710, + "grad_norm": 2.088, + "learning_rate": 5.917427521713115e-07, + "loss": 0.6254, + "step": 5710 + }, + { + "ETA": 3.14, + "epoch": 0.8964463425145947, + "fp16_scale": 4096.0, + "global_step": 5720, + "grad_norm": 2.035, + "learning_rate": 5.746588912715467e-07, + "loss": 0.6323, + "step": 5720 + }, + { + "ETA": 3.09, + "epoch": 0.8980135564001097, + "fp16_scale": 4096.0, + "global_step": 5730, + "grad_norm": 2.48, + "learning_rate": 5.578179682751372e-07, + "loss": 0.6391, + "step": 5730 + }, + { + "ETA": 3.04, + "epoch": 0.8995807702856248, + "fp16_scale": 4096.0, + "global_step": 5740, + "grad_norm": 2.13, + "learning_rate": 5.412204172564006e-07, + "loss": 0.6491, + "step": 5740 + }, + { + "ETA": 3.0, + "epoch": 0.9011479841711397, + "fp16_scale": 4096.0, + "global_step": 5750, + "grad_norm": 2.237, + "learning_rate": 5.248666660167523e-07, + "loss": 0.623, + "step": 5750 + }, + { + "ETA": 2.95, + "epoch": 0.9027151980566548, + "fp16_scale": 4096.0, + "global_step": 5760, + "grad_norm": 2.283, + "learning_rate": 5.087571360736698e-07, + "loss": 0.6169, + "step": 5760 + }, + { + "ETA": 2.9, + "epoch": 0.9042824119421699, + "fp16_scale": 4096.0, + "global_step": 5770, + "grad_norm": 1.802, + "learning_rate": 4.928922426498384e-07, + "loss": 0.6286, + "step": 5770 + }, + { + "ETA": 2.85, + "epoch": 0.9058496258276848, + "fp16_scale": 4096.0, + "global_step": 5780, + "grad_norm": 2.055, + "learning_rate": 4.772723946624414e-07, + "loss": 0.6375, + "step": 5780 + }, + { + "ETA": 2.81, + "epoch": 0.9074168397131999, + "fp16_scale": 4096.0, + "global_step": 5790, + "grad_norm": 2.604, + "learning_rate": 4.6189799471262343e-07, + "loss": 0.6312, + "step": 5790 + }, + { + "ETA": 2.76, + "epoch": 0.9089840535987149, + "fp16_scale": 4096.0, + "global_step": 5800, + "grad_norm": 2.046, + "learning_rate": 4.467694390751132e-07, + "loss": 0.619, + "step": 5800 + }, + { + "ETA": 2.71, + "epoch": 0.9105512674842299, + "fp16_scale": 4096.0, + "global_step": 5810, + "grad_norm": 2.427, + "learning_rate": 4.318871176880102e-07, + "loss": 0.6094, + "step": 5810 + }, + { + "ETA": 2.66, + "epoch": 0.9121184813697449, + "fp16_scale": 4096.0, + "global_step": 5820, + "grad_norm": 2.188, + "learning_rate": 4.1725141414273194e-07, + "loss": 0.6301, + "step": 5820 + }, + { + "ETA": 2.62, + "epoch": 0.91368569525526, + "fp16_scale": 4096.0, + "global_step": 5830, + "grad_norm": 2.529, + "learning_rate": 4.028627056741252e-07, + "loss": 0.6323, + "step": 5830 + }, + { + "ETA": 2.57, + "epoch": 0.915252909140775, + "fp16_scale": 4096.0, + "global_step": 5840, + "grad_norm": 2.326, + "learning_rate": 3.887213631507503e-07, + "loss": 0.6415, + "step": 5840 + }, + { + "ETA": 2.52, + "epoch": 0.91682012302629, + "fp16_scale": 4096.0, + "global_step": 5850, + "grad_norm": 2.328, + "learning_rate": 3.748277510653142e-07, + "loss": 0.6457, + "step": 5850 + }, + { + "ETA": 2.47, + "epoch": 0.9183873369118051, + "fp16_scale": 4096.0, + "global_step": 5860, + "grad_norm": 4.232, + "learning_rate": 3.6118222752527963e-07, + "loss": 0.625, + "step": 5860 + }, + { + "ETA": 2.42, + "epoch": 0.91995455079732, + "fp16_scale": 4096.0, + "global_step": 5870, + "grad_norm": 2.001, + "learning_rate": 3.477851442436342e-07, + "loss": 0.6162, + "step": 5870 + }, + { + "ETA": 2.38, + "epoch": 0.9215217646828351, + "fp16_scale": 4096.0, + "global_step": 5880, + "grad_norm": 1.914, + "learning_rate": 3.3463684652982555e-07, + "loss": 0.6074, + "step": 5880 + }, + { + "ETA": 2.33, + "epoch": 0.9230889785683501, + "fp16_scale": 4096.0, + "global_step": 5890, + "grad_norm": 1.723, + "learning_rate": 3.217376732808586e-07, + "loss": 0.6085, + "step": 5890 + }, + { + "ETA": 2.28, + "epoch": 0.9246561924538651, + "fp16_scale": 4096.0, + "global_step": 5900, + "grad_norm": 2.135, + "learning_rate": 3.0908795697256246e-07, + "loss": 0.6303, + "step": 5900 + }, + { + "ETA": 2.23, + "epoch": 0.9262234063393802, + "fp16_scale": 4096.0, + "global_step": 5910, + "grad_norm": 2.142, + "learning_rate": 2.9668802365102054e-07, + "loss": 0.6107, + "step": 5910 + }, + { + "ETA": 2.19, + "epoch": 0.9277906202248952, + "fp16_scale": 4096.0, + "global_step": 5920, + "grad_norm": 2.271, + "learning_rate": 2.8453819292416973e-07, + "loss": 0.6003, + "step": 5920 + }, + { + "ETA": 2.14, + "epoch": 0.9293578341104102, + "fp16_scale": 4096.0, + "global_step": 5930, + "grad_norm": 2.507, + "learning_rate": 2.7263877795355443e-07, + "loss": 0.6426, + "step": 5930 + }, + { + "ETA": 2.09, + "epoch": 0.9309250479959252, + "fp16_scale": 4096.0, + "global_step": 5940, + "grad_norm": 2.27, + "learning_rate": 2.6099008544626434e-07, + "loss": 0.6432, + "step": 5940 + }, + { + "ETA": 2.04, + "epoch": 0.9324922618814403, + "fp16_scale": 4096.0, + "global_step": 5950, + "grad_norm": 1.876, + "learning_rate": 2.495924156470231e-07, + "loss": 0.5796, + "step": 5950 + }, + { + "ETA": 2.0, + "epoch": 0.9340594757669552, + "fp16_scale": 4096.0, + "global_step": 5960, + "grad_norm": 2.29, + "learning_rate": 2.3844606233044855e-07, + "loss": 0.6381, + "step": 5960 + }, + { + "ETA": 1.95, + "epoch": 0.9356266896524703, + "fp16_scale": 4096.0, + "global_step": 5970, + "grad_norm": 1.93, + "learning_rate": 2.2755131279348807e-07, + "loss": 0.6265, + "step": 5970 + }, + { + "ETA": 1.9, + "epoch": 0.9371939035379854, + "fp16_scale": 4096.0, + "global_step": 5980, + "grad_norm": 2.107, + "learning_rate": 2.1690844784800437e-07, + "loss": 0.6197, + "step": 5980 + }, + { + "ETA": 1.85, + "epoch": 0.9387611174235003, + "fp16_scale": 4096.0, + "global_step": 5990, + "grad_norm": 2.435, + "learning_rate": 2.0651774181354445e-07, + "loss": 0.6203, + "step": 5990 + }, + { + "ETA": 1.81, + "epoch": 0.9403283313090154, + "fp16_scale": 4096.0, + "global_step": 6000, + "grad_norm": 2.107, + "learning_rate": 1.963794625102655e-07, + "loss": 0.6072, + "step": 6000 + }, + { + "ETA": 1.76, + "epoch": 0.9418955451945304, + "fp16_scale": 4096.0, + "global_step": 6010, + "grad_norm": 2.7, + "learning_rate": 1.8649387125203255e-07, + "loss": 0.6501, + "step": 6010 + }, + { + "ETA": 1.71, + "epoch": 0.9434627590800454, + "fp16_scale": 4096.0, + "global_step": 6020, + "grad_norm": 2.167, + "learning_rate": 1.7686122283968488e-07, + "loss": 0.6264, + "step": 6020 + }, + { + "ETA": 1.66, + "epoch": 0.9450299729655605, + "fp16_scale": 4096.0, + "global_step": 6030, + "grad_norm": 2.122, + "learning_rate": 1.674817655544647e-07, + "loss": 0.6247, + "step": 6030 + }, + { + "ETA": 1.62, + "epoch": 0.9465971868510755, + "fp16_scale": 4096.0, + "global_step": 6040, + "grad_norm": 2.076, + "learning_rate": 1.5835574115162123e-07, + "loss": 0.6271, + "step": 6040 + }, + { + "ETA": 1.57, + "epoch": 0.9481644007365905, + "fp16_scale": 4096.0, + "global_step": 6050, + "grad_norm": 2.815, + "learning_rate": 1.4948338485417768e-07, + "loss": 0.6186, + "step": 6050 + }, + { + "ETA": 1.52, + "epoch": 0.9497316146221055, + "fp16_scale": 4096.0, + "global_step": 6060, + "grad_norm": 2.341, + "learning_rate": 1.4086492534686968e-07, + "loss": 0.6258, + "step": 6060 + }, + { + "ETA": 1.47, + "epoch": 0.9512988285076206, + "fp16_scale": 4096.0, + "global_step": 6070, + "grad_norm": 2.646, + "learning_rate": 1.3250058477025095e-07, + "loss": 0.6234, + "step": 6070 + }, + { + "ETA": 1.43, + "epoch": 0.9528660423931355, + "fp16_scale": 4096.0, + "global_step": 6080, + "grad_norm": 2.8, + "learning_rate": 1.2439057871496463e-07, + "loss": 0.6148, + "step": 6080 + }, + { + "ETA": 1.38, + "epoch": 0.9544332562786506, + "fp16_scale": 4096.0, + "global_step": 6090, + "grad_norm": 1.999, + "learning_rate": 1.1653511621618985e-07, + "loss": 0.5933, + "step": 6090 + }, + { + "ETA": 1.33, + "epoch": 0.9560004701641657, + "fp16_scale": 4096.0, + "global_step": 6100, + "grad_norm": 2.628, + "learning_rate": 1.0893439974825393e-07, + "loss": 0.6104, + "step": 6100 + }, + { + "ETA": 1.28, + "epoch": 0.9575676840496807, + "fp16_scale": 4096.0, + "global_step": 6110, + "grad_norm": 2.091, + "learning_rate": 1.015886252194087e-07, + "loss": 0.6291, + "step": 6110 + }, + { + "ETA": 1.24, + "epoch": 0.9591348979351957, + "fp16_scale": 4096.0, + "global_step": 6120, + "grad_norm": 2.404, + "learning_rate": 9.449798196679016e-08, + "loss": 0.6221, + "step": 6120 + }, + { + "ETA": 1.19, + "epoch": 0.9607021118207107, + "fp16_scale": 4096.0, + "global_step": 6130, + "grad_norm": 2.17, + "learning_rate": 8.766265275152786e-08, + "loss": 0.6213, + "step": 6130 + }, + { + "ETA": 1.14, + "epoch": 0.9622693257062258, + "fp16_scale": 4096.0, + "global_step": 6140, + "grad_norm": 1.804, + "learning_rate": 8.108281375404093e-08, + "loss": 0.6415, + "step": 6140 + }, + { + "ETA": 1.09, + "epoch": 0.9638365395917408, + "fp16_scale": 4096.0, + "global_step": 6150, + "grad_norm": 1.722, + "learning_rate": 7.475863456949728e-08, + "loss": 0.6276, + "step": 6150 + }, + { + "ETA": 1.05, + "epoch": 0.9654037534772558, + "fp16_scale": 4096.0, + "global_step": 6160, + "grad_norm": 1.697, + "learning_rate": 6.869027820343598e-08, + "loss": 0.6077, + "step": 6160 + }, + { + "ETA": 1.0, + "epoch": 0.9669709673627709, + "fp16_scale": 4096.0, + "global_step": 6170, + "grad_norm": 2.37, + "learning_rate": 6.287790106757396e-08, + "loss": 0.6046, + "step": 6170 + }, + { + "ETA": 0.95, + "epoch": 0.9685381812482858, + "fp16_scale": 4096.0, + "global_step": 6180, + "grad_norm": 2.879, + "learning_rate": 5.7321652975769194e-08, + "loss": 0.6358, + "step": 6180 + }, + { + "ETA": 0.9, + "epoch": 0.9701053951338009, + "fp16_scale": 4096.0, + "global_step": 6190, + "grad_norm": 2.486, + "learning_rate": 5.20216771401616e-08, + "loss": 0.6194, + "step": 6190 + }, + { + "ETA": 0.86, + "epoch": 0.971672609019316, + "fp16_scale": 4096.0, + "global_step": 6200, + "grad_norm": 2.44, + "learning_rate": 4.697811016747711e-08, + "loss": 0.6041, + "step": 6200 + }, + { + "ETA": 0.81, + "epoch": 0.9732398229048309, + "fp16_scale": 4096.0, + "global_step": 6210, + "grad_norm": 2.378, + "learning_rate": 4.2191082055517143e-08, + "loss": 0.6194, + "step": 6210 + }, + { + "ETA": 0.76, + "epoch": 0.974807036790346, + "fp16_scale": 4096.0, + "global_step": 6220, + "grad_norm": 2.339, + "learning_rate": 3.766071618979461e-08, + "loss": 0.6172, + "step": 6220 + }, + { + "ETA": 0.71, + "epoch": 0.976374250675861, + "fp16_scale": 4096.0, + "global_step": 6230, + "grad_norm": 2.287, + "learning_rate": 3.338712934036425e-08, + "loss": 0.6208, + "step": 6230 + }, + { + "ETA": 0.67, + "epoch": 0.977941464561376, + "fp16_scale": 4096.0, + "global_step": 6240, + "grad_norm": 2.324, + "learning_rate": 2.9370431658806152e-08, + "loss": 0.6339, + "step": 6240 + }, + { + "ETA": 0.62, + "epoch": 0.979508678446891, + "fp16_scale": 4096.0, + "global_step": 6250, + "grad_norm": 2.071, + "learning_rate": 2.561072667539244e-08, + "loss": 0.6279, + "step": 6250 + }, + { + "ETA": 0.57, + "epoch": 0.9810758923324061, + "fp16_scale": 4096.0, + "global_step": 6260, + "grad_norm": 2.58, + "learning_rate": 2.2108111296412772e-08, + "loss": 0.6292, + "step": 6260 + }, + { + "ETA": 0.52, + "epoch": 0.9826431062179211, + "fp16_scale": 4096.0, + "global_step": 6270, + "grad_norm": 2.091, + "learning_rate": 1.8862675801681884e-08, + "loss": 0.6033, + "step": 6270 + }, + { + "ETA": 0.48, + "epoch": 0.9842103201034361, + "fp16_scale": 4096.0, + "global_step": 6280, + "grad_norm": 2.559, + "learning_rate": 1.58745038422059e-08, + "loss": 0.635, + "step": 6280 + }, + { + "ETA": 0.43, + "epoch": 0.9857775339889512, + "fp16_scale": 4096.0, + "global_step": 6290, + "grad_norm": 2.044, + "learning_rate": 1.3143672438037381e-08, + "loss": 0.6041, + "step": 6290 + }, + { + "ETA": 0.38, + "epoch": 0.9873447478744661, + "fp16_scale": 4096.0, + "global_step": 6300, + "grad_norm": 2.064, + "learning_rate": 1.0670251976275803e-08, + "loss": 0.6467, + "step": 6300 + }, + { + "ETA": 0.33, + "epoch": 0.9889119617599812, + "fp16_scale": 4096.0, + "global_step": 6310, + "grad_norm": 2.35, + "learning_rate": 8.454306209265684e-09, + "loss": 0.6252, + "step": 6310 + }, + { + "ETA": 0.29, + "epoch": 0.9904791756454963, + "fp16_scale": 4096.0, + "global_step": 6320, + "grad_norm": 2.157, + "learning_rate": 6.495892252947888e-09, + "loss": 0.639, + "step": 6320 + }, + { + "ETA": 0.24, + "epoch": 0.9920463895310112, + "fp16_scale": 4096.0, + "global_step": 6330, + "grad_norm": 2.505, + "learning_rate": 4.7950605853874785e-09, + "loss": 0.6403, + "step": 6330 + }, + { + "ETA": 0.19, + "epoch": 0.9936136034165263, + "fp16_scale": 4096.0, + "global_step": 6340, + "grad_norm": 2.051, + "learning_rate": 3.3518550454714195e-09, + "loss": 0.6109, + "step": 6340 + }, + { + "ETA": 0.14, + "epoch": 0.9951808173020413, + "fp16_scale": 4096.0, + "global_step": 6350, + "grad_norm": 1.954, + "learning_rate": 2.166312831783923e-09, + "loss": 0.602, + "step": 6350 + }, + { + "ETA": 0.1, + "epoch": 0.9967480311875563, + "fp16_scale": 4096.0, + "global_step": 6360, + "grad_norm": 1.921, + "learning_rate": 1.2384645016416674e-09, + "loss": 0.6085, + "step": 6360 + }, + { + "ETA": 0.05, + "epoch": 0.9983152450730713, + "fp16_scale": 4096.0, + "global_step": 6370, + "grad_norm": 3.47, + "learning_rate": 5.683339703088653e-10, + "loss": 0.6294, + "step": 6370 + }, + { + "ETA": 0.0, + "epoch": 0.9998824589585864, + "fp16_scale": 4096.0, + "global_step": 6380, + "grad_norm": 2.934, + "learning_rate": 1.5593851038109287e-10, + "loss": 0.6272, + "step": 6380 + }, + { + "epoch": 0.9998824589585864, + "step": 6380, + "total_flos": 6434699981881344.0, + "train_loss": 0.7134642638382866, + "train_runtime": 109318.7053, + "train_samples_per_second": 7.471, + "train_steps_per_second": 0.058 + } + ], + "logging_steps": 10, + "max_steps": 6380, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 6434699981881344.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}