diff --git "a/checkpoint-30600/trainer_state.json" "b/checkpoint-30600/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-30600/trainer_state.json" @@ -0,0 +1,21912 @@ +{ + "best_metric": 0.17622125002269137, + "best_model_checkpoint": "./checkpoints/w2v-pa-v2/checkpoint-30600", + "epoch": 1.1119994185623956, + "eval_steps": 600, + "global_step": 30600, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.00036339850279816846, + "grad_norm": 12.712770462036133, + "learning_rate": 3.633985027981685e-08, + "loss": 8.1015, + "step": 10 + }, + { + "epoch": 0.0007267970055963369, + "grad_norm": 14.929631233215332, + "learning_rate": 7.26797005596337e-08, + "loss": 8.5111, + "step": 20 + }, + { + "epoch": 0.0010901955083945055, + "grad_norm": 12.774781227111816, + "learning_rate": 1.0901955083945056e-07, + "loss": 8.71, + "step": 30 + }, + { + "epoch": 0.0014535940111926739, + "grad_norm": 15.27083969116211, + "learning_rate": 1.417254160912857e-07, + "loss": 9.2894, + "step": 40 + }, + { + "epoch": 0.0018169925139908424, + "grad_norm": 31.209775924682617, + "learning_rate": 1.7806526637110256e-07, + "loss": 8.6779, + "step": 50 + }, + { + "epoch": 0.002180391016789011, + "grad_norm": 12.960335731506348, + "learning_rate": 2.1440511665091943e-07, + "loss": 8.0399, + "step": 60 + }, + { + "epoch": 0.002543789519587179, + "grad_norm": 14.452157020568848, + "learning_rate": 2.5074496693073626e-07, + "loss": 7.9357, + "step": 70 + }, + { + "epoch": 0.0029071880223853477, + "grad_norm": 12.74867057800293, + "learning_rate": 2.8708481721055307e-07, + "loss": 8.0764, + "step": 80 + }, + { + "epoch": 0.0032705865251835163, + "grad_norm": 19.905397415161133, + "learning_rate": 3.2342466749036993e-07, + "loss": 8.2522, + "step": 90 + }, + { + "epoch": 0.003633985027981685, + "grad_norm": Infinity, + "learning_rate": 3.5613053274220513e-07, + "loss": 7.8161, + "step": 100 + }, + { + "epoch": 0.003997383530779853, + "grad_norm": 15.877684593200684, + "learning_rate": 3.92470383022022e-07, + "loss": 7.4866, + "step": 110 + }, + { + "epoch": 0.004360782033578022, + "grad_norm": 19.216800689697266, + "learning_rate": 4.2881023330183885e-07, + "loss": 7.0324, + "step": 120 + }, + { + "epoch": 0.0047241805363761906, + "grad_norm": 16.937118530273438, + "learning_rate": 4.651500835816557e-07, + "loss": 6.7873, + "step": 130 + }, + { + "epoch": 0.005087579039174358, + "grad_norm": 27.858692169189453, + "learning_rate": 5.014899338614725e-07, + "loss": 7.2063, + "step": 140 + }, + { + "epoch": 0.005450977541972527, + "grad_norm": Infinity, + "learning_rate": 5.341957991133076e-07, + "loss": 6.5827, + "step": 150 + }, + { + "epoch": 0.005814376044770695, + "grad_norm": 21.252164840698242, + "learning_rate": 5.669016643651428e-07, + "loss": 7.819, + "step": 160 + }, + { + "epoch": 0.006177774547568864, + "grad_norm": 20.977886199951172, + "learning_rate": 6.032415146449597e-07, + "loss": 5.4741, + "step": 170 + }, + { + "epoch": 0.0065411730503670325, + "grad_norm": 35.25390625, + "learning_rate": 6.395813649247765e-07, + "loss": 5.7225, + "step": 180 + }, + { + "epoch": 0.006904571553165201, + "grad_norm": 7.165033340454102, + "learning_rate": 6.759212152045934e-07, + "loss": 4.7475, + "step": 190 + }, + { + "epoch": 0.00726797005596337, + "grad_norm": 14.877301216125488, + "learning_rate": 7.122610654844103e-07, + "loss": 4.6375, + "step": 200 + }, + { + "epoch": 0.007631368558761538, + "grad_norm": 5.826667785644531, + "learning_rate": 7.486009157642272e-07, + "loss": 4.356, + "step": 210 + }, + { + "epoch": 0.007994767061559707, + "grad_norm": 6.022212982177734, + "learning_rate": 7.84940766044044e-07, + "loss": 4.2138, + "step": 220 + }, + { + "epoch": 0.008358165564357875, + "grad_norm": 4.790489196777344, + "learning_rate": 8.212806163238608e-07, + "loss": 4.0662, + "step": 230 + }, + { + "epoch": 0.008721564067156044, + "grad_norm": 4.448057174682617, + "learning_rate": 8.576204666036777e-07, + "loss": 3.9507, + "step": 240 + }, + { + "epoch": 0.009084962569954213, + "grad_norm": 47.487003326416016, + "learning_rate": 8.939603168834945e-07, + "loss": 3.9576, + "step": 250 + }, + { + "epoch": 0.009448361072752381, + "grad_norm": 8.64856243133545, + "learning_rate": 9.303001671633114e-07, + "loss": 3.7102, + "step": 260 + }, + { + "epoch": 0.009811759575550548, + "grad_norm": 8.821709632873535, + "learning_rate": 9.66640017443128e-07, + "loss": 3.6644, + "step": 270 + }, + { + "epoch": 0.010175158078348717, + "grad_norm": 14.071539878845215, + "learning_rate": 1.002979867722945e-06, + "loss": 3.6909, + "step": 280 + }, + { + "epoch": 0.010538556581146885, + "grad_norm": 6.68039083480835, + "learning_rate": 1.0393197180027619e-06, + "loss": 3.6458, + "step": 290 + }, + { + "epoch": 0.010901955083945054, + "grad_norm": 20.664649963378906, + "learning_rate": 1.0756595682825787e-06, + "loss": 3.6332, + "step": 300 + }, + { + "epoch": 0.011265353586743222, + "grad_norm": 2.5272624492645264, + "learning_rate": 1.1119994185623955e-06, + "loss": 3.4835, + "step": 310 + }, + { + "epoch": 0.01162875208954139, + "grad_norm": 8.353235244750977, + "learning_rate": 1.1483392688422123e-06, + "loss": 3.4892, + "step": 320 + }, + { + "epoch": 0.01199215059233956, + "grad_norm": 7.0964531898498535, + "learning_rate": 1.1846791191220293e-06, + "loss": 3.5586, + "step": 330 + }, + { + "epoch": 0.012355549095137728, + "grad_norm": 4.734161376953125, + "learning_rate": 1.2210189694018461e-06, + "loss": 3.4996, + "step": 340 + }, + { + "epoch": 0.012718947597935897, + "grad_norm": 47.409996032714844, + "learning_rate": 1.257358819681663e-06, + "loss": 3.5565, + "step": 350 + }, + { + "epoch": 0.013082346100734065, + "grad_norm": 2.880244016647339, + "learning_rate": 1.2936986699614797e-06, + "loss": 3.4154, + "step": 360 + }, + { + "epoch": 0.013445744603532234, + "grad_norm": 6.637233734130859, + "learning_rate": 1.3300385202412968e-06, + "loss": 3.4119, + "step": 370 + }, + { + "epoch": 0.013809143106330402, + "grad_norm": 11.791736602783203, + "learning_rate": 1.3663783705211136e-06, + "loss": 3.4347, + "step": 380 + }, + { + "epoch": 0.01417254160912857, + "grad_norm": 8.274836540222168, + "learning_rate": 1.4027182208009304e-06, + "loss": 3.4253, + "step": 390 + }, + { + "epoch": 0.01453594011192674, + "grad_norm": 10.09929084777832, + "learning_rate": 1.4390580710807472e-06, + "loss": 3.4702, + "step": 400 + }, + { + "epoch": 0.014899338614724908, + "grad_norm": 6.32951545715332, + "learning_rate": 1.4753979213605642e-06, + "loss": 3.3513, + "step": 410 + }, + { + "epoch": 0.015262737117523077, + "grad_norm": 2.4888486862182617, + "learning_rate": 1.511737771640381e-06, + "loss": 3.3421, + "step": 420 + }, + { + "epoch": 0.015626135620321245, + "grad_norm": 3.02103328704834, + "learning_rate": 1.5480776219201978e-06, + "loss": 3.397, + "step": 430 + }, + { + "epoch": 0.015989534123119414, + "grad_norm": 7.464268207550049, + "learning_rate": 1.5844174722000146e-06, + "loss": 3.3582, + "step": 440 + }, + { + "epoch": 0.016352932625917582, + "grad_norm": 18.908123016357422, + "learning_rate": 1.6207573224798317e-06, + "loss": 3.4034, + "step": 450 + }, + { + "epoch": 0.01671633112871575, + "grad_norm": 2.487326145172119, + "learning_rate": 1.6570971727596485e-06, + "loss": 3.2229, + "step": 460 + }, + { + "epoch": 0.01707972963151392, + "grad_norm": 2.3999946117401123, + "learning_rate": 1.6934370230394653e-06, + "loss": 3.2185, + "step": 470 + }, + { + "epoch": 0.017443128134312088, + "grad_norm": 5.007234573364258, + "learning_rate": 1.729776873319282e-06, + "loss": 3.2069, + "step": 480 + }, + { + "epoch": 0.017806526637110257, + "grad_norm": 6.393301963806152, + "learning_rate": 1.766116723599099e-06, + "loss": 3.0687, + "step": 490 + }, + { + "epoch": 0.018169925139908425, + "grad_norm": 45.44938278198242, + "learning_rate": 1.802456573878916e-06, + "loss": 3.112, + "step": 500 + }, + { + "epoch": 0.018533323642706594, + "grad_norm": 7.32182502746582, + "learning_rate": 1.8387964241587327e-06, + "loss": 2.951, + "step": 510 + }, + { + "epoch": 0.018896722145504762, + "grad_norm": 3.3864173889160156, + "learning_rate": 1.8751362744385495e-06, + "loss": 2.8879, + "step": 520 + }, + { + "epoch": 0.019260120648302927, + "grad_norm": 5.429958343505859, + "learning_rate": 1.911476124718366e-06, + "loss": 2.7393, + "step": 530 + }, + { + "epoch": 0.019623519151101096, + "grad_norm": 5.3577985763549805, + "learning_rate": 1.947815974998183e-06, + "loss": 2.4813, + "step": 540 + }, + { + "epoch": 0.019986917653899264, + "grad_norm": 13.970659255981445, + "learning_rate": 1.9841558252779998e-06, + "loss": 2.3787, + "step": 550 + }, + { + "epoch": 0.020350316156697433, + "grad_norm": 5.2666754722595215, + "learning_rate": 2.0204956755578166e-06, + "loss": 2.207, + "step": 560 + }, + { + "epoch": 0.0207137146594956, + "grad_norm": 4.184991359710693, + "learning_rate": 2.0568355258376334e-06, + "loss": 2.0383, + "step": 570 + }, + { + "epoch": 0.02107711316229377, + "grad_norm": 6.312343597412109, + "learning_rate": 2.09317537611745e-06, + "loss": 1.8416, + "step": 580 + }, + { + "epoch": 0.02144051166509194, + "grad_norm": 4.754147529602051, + "learning_rate": 2.1295152263972674e-06, + "loss": 1.6002, + "step": 590 + }, + { + "epoch": 0.021803910167890107, + "grad_norm": 21.47913360595703, + "learning_rate": 2.1658550766770842e-06, + "loss": 1.6015, + "step": 600 + }, + { + "epoch": 0.021803910167890107, + "eval_loss": 1.5154471397399902, + "eval_runtime": 180.9184, + "eval_samples_per_second": 40.98, + "eval_steps_per_second": 5.124, + "eval_wer": 0.7997531177954872, + "step": 600 + }, + { + "epoch": 0.022167308670688276, + "grad_norm": 4.2374348640441895, + "learning_rate": 2.202194926956901e-06, + "loss": 1.4842, + "step": 610 + }, + { + "epoch": 0.022530707173486444, + "grad_norm": 4.392132759094238, + "learning_rate": 2.238534777236718e-06, + "loss": 1.3776, + "step": 620 + }, + { + "epoch": 0.022894105676284613, + "grad_norm": 4.682064533233643, + "learning_rate": 2.2748746275165347e-06, + "loss": 1.3177, + "step": 630 + }, + { + "epoch": 0.02325750417908278, + "grad_norm": 4.8396077156066895, + "learning_rate": 2.3112144777963515e-06, + "loss": 1.0737, + "step": 640 + }, + { + "epoch": 0.02362090268188095, + "grad_norm": 33.27382278442383, + "learning_rate": 2.3475543280761683e-06, + "loss": 1.3046, + "step": 650 + }, + { + "epoch": 0.02398430118467912, + "grad_norm": 5.410325050354004, + "learning_rate": 2.383894178355985e-06, + "loss": 1.1021, + "step": 660 + }, + { + "epoch": 0.024347699687477287, + "grad_norm": 3.9523680210113525, + "learning_rate": 2.420234028635802e-06, + "loss": 1.0602, + "step": 670 + }, + { + "epoch": 0.024711098190275456, + "grad_norm": 9.141073226928711, + "learning_rate": 2.456573878915619e-06, + "loss": 1.0631, + "step": 680 + }, + { + "epoch": 0.025074496693073624, + "grad_norm": 5.3534626960754395, + "learning_rate": 2.492913729195436e-06, + "loss": 0.8968, + "step": 690 + }, + { + "epoch": 0.025437895195871793, + "grad_norm": 32.30677795410156, + "learning_rate": 2.5292535794752527e-06, + "loss": 1.0439, + "step": 700 + }, + { + "epoch": 0.02580129369866996, + "grad_norm": 4.310474872589111, + "learning_rate": 2.5655934297550696e-06, + "loss": 0.954, + "step": 710 + }, + { + "epoch": 0.02616469220146813, + "grad_norm": 5.586440563201904, + "learning_rate": 2.6019332800348864e-06, + "loss": 1.0031, + "step": 720 + }, + { + "epoch": 0.0265280907042663, + "grad_norm": 3.6927313804626465, + "learning_rate": 2.638273130314703e-06, + "loss": 0.7956, + "step": 730 + }, + { + "epoch": 0.026891489207064467, + "grad_norm": 4.270529747009277, + "learning_rate": 2.67461298059452e-06, + "loss": 0.8874, + "step": 740 + }, + { + "epoch": 0.027254887709862636, + "grad_norm": 23.553489685058594, + "learning_rate": 2.710952830874337e-06, + "loss": 0.8523, + "step": 750 + }, + { + "epoch": 0.027618286212660804, + "grad_norm": 5.342041492462158, + "learning_rate": 2.747292681154154e-06, + "loss": 0.9029, + "step": 760 + }, + { + "epoch": 0.027981684715458973, + "grad_norm": 3.3802621364593506, + "learning_rate": 2.783632531433971e-06, + "loss": 0.8378, + "step": 770 + }, + { + "epoch": 0.02834508321825714, + "grad_norm": 6.378807067871094, + "learning_rate": 2.8199723817137876e-06, + "loss": 0.8085, + "step": 780 + }, + { + "epoch": 0.02870848172105531, + "grad_norm": 4.007000923156738, + "learning_rate": 2.8563122319936045e-06, + "loss": 0.8218, + "step": 790 + }, + { + "epoch": 0.02907188022385348, + "grad_norm": 68.16226196289062, + "learning_rate": 2.8926520822734213e-06, + "loss": 1.2055, + "step": 800 + }, + { + "epoch": 0.029435278726651647, + "grad_norm": 6.70043420791626, + "learning_rate": 2.928991932553238e-06, + "loss": 0.7641, + "step": 810 + }, + { + "epoch": 0.029798677229449816, + "grad_norm": 5.498161315917969, + "learning_rate": 2.965331782833055e-06, + "loss": 0.7739, + "step": 820 + }, + { + "epoch": 0.030162075732247984, + "grad_norm": 9.515852928161621, + "learning_rate": 3.0016716331128717e-06, + "loss": 0.8293, + "step": 830 + }, + { + "epoch": 0.030525474235046153, + "grad_norm": 13.3881196975708, + "learning_rate": 3.0380114833926885e-06, + "loss": 0.5597, + "step": 840 + }, + { + "epoch": 0.03088887273784432, + "grad_norm": 13.670549392700195, + "learning_rate": 3.0743513336725057e-06, + "loss": 0.7658, + "step": 850 + }, + { + "epoch": 0.03125227124064249, + "grad_norm": 3.58305287361145, + "learning_rate": 3.1106911839523226e-06, + "loss": 0.7036, + "step": 860 + }, + { + "epoch": 0.031615669743440655, + "grad_norm": 4.119450569152832, + "learning_rate": 3.147031034232139e-06, + "loss": 0.6842, + "step": 870 + }, + { + "epoch": 0.03197906824623883, + "grad_norm": 6.412299156188965, + "learning_rate": 3.183370884511956e-06, + "loss": 0.7148, + "step": 880 + }, + { + "epoch": 0.03234246674903699, + "grad_norm": 8.700023651123047, + "learning_rate": 3.2197107347917726e-06, + "loss": 1.4861, + "step": 890 + }, + { + "epoch": 0.032705865251835164, + "grad_norm": 18.78075408935547, + "learning_rate": 3.25605058507159e-06, + "loss": 0.7162, + "step": 900 + }, + { + "epoch": 0.03306926375463333, + "grad_norm": 4.078335762023926, + "learning_rate": 3.292390435351406e-06, + "loss": 0.643, + "step": 910 + }, + { + "epoch": 0.0334326622574315, + "grad_norm": 6.603452682495117, + "learning_rate": 3.3287302856312234e-06, + "loss": 0.6623, + "step": 920 + }, + { + "epoch": 0.03379606076022967, + "grad_norm": 5.817732334136963, + "learning_rate": 3.3650701359110402e-06, + "loss": 0.6265, + "step": 930 + }, + { + "epoch": 0.03415945926302784, + "grad_norm": 8.310086250305176, + "learning_rate": 3.4014099861908575e-06, + "loss": 0.5343, + "step": 940 + }, + { + "epoch": 0.034522857765826004, + "grad_norm": NaN, + "learning_rate": 3.4341158514426923e-06, + "loss": 0.6999, + "step": 950 + }, + { + "epoch": 0.034886256268624176, + "grad_norm": 4.416926860809326, + "learning_rate": 3.4704557017225087e-06, + "loss": 0.6688, + "step": 960 + }, + { + "epoch": 0.03524965477142234, + "grad_norm": 3.2407495975494385, + "learning_rate": 3.506795552002326e-06, + "loss": 0.5808, + "step": 970 + }, + { + "epoch": 0.03561305327422051, + "grad_norm": NaN, + "learning_rate": 3.539501417254161e-06, + "loss": 3.0266, + "step": 980 + }, + { + "epoch": 0.03597645177701868, + "grad_norm": 8.086112022399902, + "learning_rate": 3.575841267533978e-06, + "loss": 0.5829, + "step": 990 + }, + { + "epoch": 0.03633985027981685, + "grad_norm": 252.45077514648438, + "learning_rate": 3.612181117813795e-06, + "loss": 0.7371, + "step": 1000 + }, + { + "epoch": 0.036703248782615015, + "grad_norm": 3.5969936847686768, + "learning_rate": 3.648520968093612e-06, + "loss": 0.6632, + "step": 1010 + }, + { + "epoch": 0.03706664728541319, + "grad_norm": 3.0116841793060303, + "learning_rate": 3.6848608183734285e-06, + "loss": 0.537, + "step": 1020 + }, + { + "epoch": 0.03743004578821135, + "grad_norm": 5.494657039642334, + "learning_rate": 3.7212006686532457e-06, + "loss": 0.5422, + "step": 1030 + }, + { + "epoch": 0.037793444291009524, + "grad_norm": 21.526798248291016, + "learning_rate": 3.757540518933062e-06, + "loss": 0.5003, + "step": 1040 + }, + { + "epoch": 0.03815684279380769, + "grad_norm": 80.90055084228516, + "learning_rate": 3.7938803692128793e-06, + "loss": 0.6566, + "step": 1050 + }, + { + "epoch": 0.038520241296605855, + "grad_norm": 3.7678096294403076, + "learning_rate": 3.830220219492696e-06, + "loss": 0.5758, + "step": 1060 + }, + { + "epoch": 0.03888363979940403, + "grad_norm": 4.526616096496582, + "learning_rate": 3.866560069772512e-06, + "loss": 0.5648, + "step": 1070 + }, + { + "epoch": 0.03924703830220219, + "grad_norm": 4.571674346923828, + "learning_rate": 3.90289992005233e-06, + "loss": 0.5864, + "step": 1080 + }, + { + "epoch": 0.039610436805000364, + "grad_norm": 5.295219421386719, + "learning_rate": 3.939239770332146e-06, + "loss": 0.4476, + "step": 1090 + }, + { + "epoch": 0.03997383530779853, + "grad_norm": 16.631162643432617, + "learning_rate": 3.975579620611963e-06, + "loss": 0.6198, + "step": 1100 + }, + { + "epoch": 0.0403372338105967, + "grad_norm": 4.685397624969482, + "learning_rate": 4.01191947089178e-06, + "loss": 0.7512, + "step": 1110 + }, + { + "epoch": 0.040700632313394866, + "grad_norm": 3.333232879638672, + "learning_rate": 4.048259321171597e-06, + "loss": 0.5087, + "step": 1120 + }, + { + "epoch": 0.04106403081619304, + "grad_norm": 5.501911640167236, + "learning_rate": 4.084599171451414e-06, + "loss": 0.5772, + "step": 1130 + }, + { + "epoch": 0.0414274293189912, + "grad_norm": 8.066693305969238, + "learning_rate": 4.120939021731231e-06, + "loss": 0.4641, + "step": 1140 + }, + { + "epoch": 0.041790827821789375, + "grad_norm": 13.463829040527344, + "learning_rate": 4.1572788720110474e-06, + "loss": 0.5192, + "step": 1150 + }, + { + "epoch": 0.04215422632458754, + "grad_norm": 4.132773399353027, + "learning_rate": 4.193618722290864e-06, + "loss": 0.4696, + "step": 1160 + }, + { + "epoch": 0.04251762482738571, + "grad_norm": 6.176777362823486, + "learning_rate": 4.229958572570681e-06, + "loss": 0.4851, + "step": 1170 + }, + { + "epoch": 0.04288102333018388, + "grad_norm": 8.26610279083252, + "learning_rate": 4.266298422850498e-06, + "loss": 0.4967, + "step": 1180 + }, + { + "epoch": 0.04324442183298205, + "grad_norm": 3.9725544452667236, + "learning_rate": 4.302638273130315e-06, + "loss": 0.431, + "step": 1190 + }, + { + "epoch": 0.043607820335780215, + "grad_norm": 22.353294372558594, + "learning_rate": 4.338978123410132e-06, + "loss": 0.6523, + "step": 1200 + }, + { + "epoch": 0.043607820335780215, + "eval_loss": 0.5945897102355957, + "eval_runtime": 180.5168, + "eval_samples_per_second": 41.071, + "eval_steps_per_second": 5.135, + "eval_wer": 0.3718390908925881, + "step": 1200 + }, + { + "epoch": 0.04397121883857839, + "grad_norm": 3.7954189777374268, + "learning_rate": 4.375317973689948e-06, + "loss": 0.4511, + "step": 1210 + }, + { + "epoch": 0.04433461734137655, + "grad_norm": 5.583435535430908, + "learning_rate": 4.411657823969766e-06, + "loss": 1.4019, + "step": 1220 + }, + { + "epoch": 0.044698015844174724, + "grad_norm": 8.544243812561035, + "learning_rate": 4.447997674249582e-06, + "loss": 0.4467, + "step": 1230 + }, + { + "epoch": 0.04506141434697289, + "grad_norm": 3.8716418743133545, + "learning_rate": 4.4843375245293996e-06, + "loss": 0.4347, + "step": 1240 + }, + { + "epoch": 0.04542481284977106, + "grad_norm": 19.459606170654297, + "learning_rate": 4.5206773748092155e-06, + "loss": 0.616, + "step": 1250 + }, + { + "epoch": 0.045788211352569226, + "grad_norm": 5.474793434143066, + "learning_rate": 4.557017225089033e-06, + "loss": 0.4689, + "step": 1260 + }, + { + "epoch": 0.0461516098553674, + "grad_norm": 4.705495834350586, + "learning_rate": 4.593357075368849e-06, + "loss": 0.4623, + "step": 1270 + }, + { + "epoch": 0.04651500835816556, + "grad_norm": 6.779942035675049, + "learning_rate": 4.629696925648667e-06, + "loss": 0.4418, + "step": 1280 + }, + { + "epoch": 0.046878406860963735, + "grad_norm": 6.802936553955078, + "learning_rate": 4.666036775928484e-06, + "loss": 0.4429, + "step": 1290 + }, + { + "epoch": 0.0472418053637619, + "grad_norm": 17.47754669189453, + "learning_rate": 4.7023766262083004e-06, + "loss": 0.596, + "step": 1300 + }, + { + "epoch": 0.04760520386656007, + "grad_norm": 4.036036968231201, + "learning_rate": 4.738716476488117e-06, + "loss": 0.4362, + "step": 1310 + }, + { + "epoch": 0.04796860236935824, + "grad_norm": 6.022701740264893, + "learning_rate": 4.775056326767934e-06, + "loss": 0.5092, + "step": 1320 + }, + { + "epoch": 0.04833200087215641, + "grad_norm": 5.533923625946045, + "learning_rate": 4.811396177047751e-06, + "loss": 0.4358, + "step": 1330 + }, + { + "epoch": 0.048695399374954575, + "grad_norm": 3.4037017822265625, + "learning_rate": 4.847736027327568e-06, + "loss": 0.3684, + "step": 1340 + }, + { + "epoch": 0.04905879787775275, + "grad_norm": 13.625974655151367, + "learning_rate": 4.8840758776073845e-06, + "loss": 0.583, + "step": 1350 + }, + { + "epoch": 0.04942219638055091, + "grad_norm": 3.597294330596924, + "learning_rate": 4.920415727887201e-06, + "loss": 0.4561, + "step": 1360 + }, + { + "epoch": 0.049785594883349084, + "grad_norm": 2.8846936225891113, + "learning_rate": 4.956755578167018e-06, + "loss": 0.409, + "step": 1370 + }, + { + "epoch": 0.05014899338614725, + "grad_norm": 5.500187397003174, + "learning_rate": 4.993095428446836e-06, + "loss": 0.4531, + "step": 1380 + }, + { + "epoch": 0.05051239188894542, + "grad_norm": 3.1203413009643555, + "learning_rate": 5.029435278726652e-06, + "loss": 0.4004, + "step": 1390 + }, + { + "epoch": 0.050875790391743586, + "grad_norm": 98.18115234375, + "learning_rate": 5.0657751290064685e-06, + "loss": 0.522, + "step": 1400 + }, + { + "epoch": 0.05123918889454175, + "grad_norm": 3.901418924331665, + "learning_rate": 5.102114979286285e-06, + "loss": 0.4041, + "step": 1410 + }, + { + "epoch": 0.05160258739733992, + "grad_norm": 4.045637130737305, + "learning_rate": 5.138454829566102e-06, + "loss": 0.4051, + "step": 1420 + }, + { + "epoch": 0.05196598590013809, + "grad_norm": 6.835183143615723, + "learning_rate": 5.174794679845919e-06, + "loss": 0.4937, + "step": 1430 + }, + { + "epoch": 0.05232938440293626, + "grad_norm": 7.708272457122803, + "learning_rate": 5.211134530125736e-06, + "loss": 0.3818, + "step": 1440 + }, + { + "epoch": 0.052692782905734425, + "grad_norm": 24.2607364654541, + "learning_rate": 5.247474380405553e-06, + "loss": 0.5445, + "step": 1450 + }, + { + "epoch": 0.0530561814085326, + "grad_norm": 3.3517005443573, + "learning_rate": 5.283814230685369e-06, + "loss": 0.4079, + "step": 1460 + }, + { + "epoch": 0.05341957991133076, + "grad_norm": 12.727778434753418, + "learning_rate": 5.320154080965187e-06, + "loss": 0.4285, + "step": 1470 + }, + { + "epoch": 0.053782978414128935, + "grad_norm": 4.984294891357422, + "learning_rate": 5.356493931245003e-06, + "loss": 0.5006, + "step": 1480 + }, + { + "epoch": 0.0541463769169271, + "grad_norm": 3.3041558265686035, + "learning_rate": 5.392833781524821e-06, + "loss": 0.3729, + "step": 1490 + }, + { + "epoch": 0.05450977541972527, + "grad_norm": 38.074546813964844, + "learning_rate": 5.429173631804637e-06, + "loss": 0.5401, + "step": 1500 + }, + { + "epoch": 0.05487317392252344, + "grad_norm": 5.649720668792725, + "learning_rate": 5.465513482084454e-06, + "loss": 0.3879, + "step": 1510 + }, + { + "epoch": 0.05523657242532161, + "grad_norm": 3.107583522796631, + "learning_rate": 5.501853332364271e-06, + "loss": 0.4144, + "step": 1520 + }, + { + "epoch": 0.055599970928119774, + "grad_norm": 19.246564865112305, + "learning_rate": 5.538193182644088e-06, + "loss": 0.4314, + "step": 1530 + }, + { + "epoch": 0.055963369430917946, + "grad_norm": 4.72367525100708, + "learning_rate": 5.574533032923905e-06, + "loss": 0.3576, + "step": 1540 + }, + { + "epoch": 0.05632676793371611, + "grad_norm": 25.88886260986328, + "learning_rate": 5.6108728832037215e-06, + "loss": 0.5385, + "step": 1550 + }, + { + "epoch": 0.05669016643651428, + "grad_norm": 3.1524956226348877, + "learning_rate": 5.647212733483538e-06, + "loss": 0.4075, + "step": 1560 + }, + { + "epoch": 0.05705356493931245, + "grad_norm": 3.883281707763672, + "learning_rate": 5.683552583763355e-06, + "loss": 0.4242, + "step": 1570 + }, + { + "epoch": 0.05741696344211062, + "grad_norm": 16.935935974121094, + "learning_rate": 5.719892434043172e-06, + "loss": 0.6194, + "step": 1580 + }, + { + "epoch": 0.057780361944908785, + "grad_norm": 4.23909330368042, + "learning_rate": 5.756232284322989e-06, + "loss": 0.4206, + "step": 1590 + }, + { + "epoch": 0.05814376044770696, + "grad_norm": 16.6039981842041, + "learning_rate": 5.7925721346028056e-06, + "loss": 0.4854, + "step": 1600 + }, + { + "epoch": 0.05850715895050512, + "grad_norm": 2.5220890045166016, + "learning_rate": 5.828911984882622e-06, + "loss": 0.4186, + "step": 1610 + }, + { + "epoch": 0.058870557453303295, + "grad_norm": 3.075101613998413, + "learning_rate": 5.865251835162439e-06, + "loss": 0.8877, + "step": 1620 + }, + { + "epoch": 0.05923395595610146, + "grad_norm": 5.511383056640625, + "learning_rate": 5.901591685442257e-06, + "loss": 0.4219, + "step": 1630 + }, + { + "epoch": 0.05959735445889963, + "grad_norm": 2.9449989795684814, + "learning_rate": 5.937931535722073e-06, + "loss": 0.2992, + "step": 1640 + }, + { + "epoch": 0.0599607529616978, + "grad_norm": 31.823612213134766, + "learning_rate": 5.9742713860018905e-06, + "loss": 0.5939, + "step": 1650 + }, + { + "epoch": 0.06032415146449597, + "grad_norm": 4.240995407104492, + "learning_rate": 6.010611236281706e-06, + "loss": 0.4176, + "step": 1660 + }, + { + "epoch": 0.060687549967294134, + "grad_norm": 2.6084980964660645, + "learning_rate": 6.046951086561524e-06, + "loss": 0.3542, + "step": 1670 + }, + { + "epoch": 0.061050948470092306, + "grad_norm": 8.318774223327637, + "learning_rate": 6.08329093684134e-06, + "loss": 0.3968, + "step": 1680 + }, + { + "epoch": 0.06141434697289047, + "grad_norm": 5.18604850769043, + "learning_rate": 6.119630787121158e-06, + "loss": 0.3879, + "step": 1690 + }, + { + "epoch": 0.06177774547568864, + "grad_norm": 51.732086181640625, + "learning_rate": 6.1559706374009745e-06, + "loss": 0.5025, + "step": 1700 + }, + { + "epoch": 0.06214114397848681, + "grad_norm": 2.5876500606536865, + "learning_rate": 6.192310487680791e-06, + "loss": 0.3558, + "step": 1710 + }, + { + "epoch": 0.06250454248128498, + "grad_norm": 5.071794033050537, + "learning_rate": 6.228650337960608e-06, + "loss": 0.3534, + "step": 1720 + }, + { + "epoch": 0.06286794098408315, + "grad_norm": 11.539891242980957, + "learning_rate": 6.264990188240424e-06, + "loss": 0.4628, + "step": 1730 + }, + { + "epoch": 0.06323133948688131, + "grad_norm": 3.275383710861206, + "learning_rate": 6.301330038520243e-06, + "loss": 0.3368, + "step": 1740 + }, + { + "epoch": 0.06359473798967948, + "grad_norm": 41.4942741394043, + "learning_rate": 6.3376698888000586e-06, + "loss": 0.53, + "step": 1750 + }, + { + "epoch": 0.06395813649247765, + "grad_norm": 3.0071399211883545, + "learning_rate": 6.374009739079875e-06, + "loss": 0.3623, + "step": 1760 + }, + { + "epoch": 0.06432153499527582, + "grad_norm": 3.385955333709717, + "learning_rate": 6.410349589359692e-06, + "loss": 0.3476, + "step": 1770 + }, + { + "epoch": 0.06468493349807398, + "grad_norm": 3.872527599334717, + "learning_rate": 6.446689439639508e-06, + "loss": 0.3168, + "step": 1780 + }, + { + "epoch": 0.06504833200087215, + "grad_norm": 4.668768882751465, + "learning_rate": 6.483029289919327e-06, + "loss": 0.3813, + "step": 1790 + }, + { + "epoch": 0.06541173050367033, + "grad_norm": 69.33656311035156, + "learning_rate": 6.519369140199143e-06, + "loss": 0.4557, + "step": 1800 + }, + { + "epoch": 0.06541173050367033, + "eval_loss": 0.5579342246055603, + "eval_runtime": 180.31, + "eval_samples_per_second": 41.118, + "eval_steps_per_second": 5.141, + "eval_wer": 0.34055221740156477, + "step": 1800 + }, + { + "epoch": 0.0657751290064685, + "grad_norm": 4.2848381996154785, + "learning_rate": 6.555708990478959e-06, + "loss": 0.5481, + "step": 1810 + }, + { + "epoch": 0.06613852750926666, + "grad_norm": 11.31700325012207, + "learning_rate": 6.592048840758775e-06, + "loss": 0.3479, + "step": 1820 + }, + { + "epoch": 0.06650192601206482, + "grad_norm": 6.088991165161133, + "learning_rate": 6.628388691038594e-06, + "loss": 0.3994, + "step": 1830 + }, + { + "epoch": 0.066865324514863, + "grad_norm": 4.342681407928467, + "learning_rate": 6.66472854131841e-06, + "loss": 0.2953, + "step": 1840 + }, + { + "epoch": 0.06722872301766117, + "grad_norm": 75.97467041015625, + "learning_rate": 6.701068391598227e-06, + "loss": 0.4807, + "step": 1850 + }, + { + "epoch": 0.06759212152045933, + "grad_norm": 3.8739049434661865, + "learning_rate": 6.7374082418780435e-06, + "loss": 0.3963, + "step": 1860 + }, + { + "epoch": 0.0679555200232575, + "grad_norm": 2.6209168434143066, + "learning_rate": 6.773748092157861e-06, + "loss": 0.422, + "step": 1870 + }, + { + "epoch": 0.06831891852605568, + "grad_norm": 17.530773162841797, + "learning_rate": 6.810087942437678e-06, + "loss": 0.3939, + "step": 1880 + }, + { + "epoch": 0.06868231702885384, + "grad_norm": 3.475748300552368, + "learning_rate": 6.846427792717494e-06, + "loss": 0.2996, + "step": 1890 + }, + { + "epoch": 0.06904571553165201, + "grad_norm": 20.979995727539062, + "learning_rate": 6.882767642997311e-06, + "loss": 0.4528, + "step": 1900 + }, + { + "epoch": 0.06940911403445017, + "grad_norm": 3.8432774543762207, + "learning_rate": 6.919107493277128e-06, + "loss": 0.2959, + "step": 1910 + }, + { + "epoch": 0.06977251253724835, + "grad_norm": 7.830467700958252, + "learning_rate": 6.955447343556945e-06, + "loss": 0.3378, + "step": 1920 + }, + { + "epoch": 0.07013591104004652, + "grad_norm": 15.633039474487305, + "learning_rate": 6.991787193836762e-06, + "loss": 0.4, + "step": 1930 + }, + { + "epoch": 0.07049930954284468, + "grad_norm": 13.628314971923828, + "learning_rate": 7.028127044116578e-06, + "loss": 0.3255, + "step": 1940 + }, + { + "epoch": 0.07086270804564285, + "grad_norm": 33.001773834228516, + "learning_rate": 7.064466894396396e-06, + "loss": 0.4367, + "step": 1950 + }, + { + "epoch": 0.07122610654844103, + "grad_norm": 3.5115041732788086, + "learning_rate": 7.100806744676212e-06, + "loss": 0.3279, + "step": 1960 + }, + { + "epoch": 0.07158950505123919, + "grad_norm": 3.0497541427612305, + "learning_rate": 7.137146594956029e-06, + "loss": 2.8797, + "step": 1970 + }, + { + "epoch": 0.07195290355403736, + "grad_norm": 6.17769718170166, + "learning_rate": 7.173486445235845e-06, + "loss": 0.3534, + "step": 1980 + }, + { + "epoch": 0.07231630205683552, + "grad_norm": 5.4114789962768555, + "learning_rate": 7.209826295515664e-06, + "loss": 0.3309, + "step": 1990 + }, + { + "epoch": 0.0726797005596337, + "grad_norm": 11.600439071655273, + "learning_rate": 7.24616614579548e-06, + "loss": 0.4382, + "step": 2000 + }, + { + "epoch": 0.07304309906243187, + "grad_norm": 3.4476027488708496, + "learning_rate": 7.2825059960752965e-06, + "loss": 0.3487, + "step": 2010 + }, + { + "epoch": 0.07340649756523003, + "grad_norm": 5.642564296722412, + "learning_rate": 7.318845846355113e-06, + "loss": 0.3513, + "step": 2020 + }, + { + "epoch": 0.0737698960680282, + "grad_norm": 7.132052898406982, + "learning_rate": 7.355185696634931e-06, + "loss": 0.4564, + "step": 2030 + }, + { + "epoch": 0.07413329457082637, + "grad_norm": 6.583246231079102, + "learning_rate": 7.391525546914748e-06, + "loss": 0.3376, + "step": 2040 + }, + { + "epoch": 0.07449669307362454, + "grad_norm": 23.98805809020996, + "learning_rate": 7.427865397194564e-06, + "loss": 0.4217, + "step": 2050 + }, + { + "epoch": 0.0748600915764227, + "grad_norm": 3.9135584831237793, + "learning_rate": 7.4642052474743805e-06, + "loss": 0.3324, + "step": 2060 + }, + { + "epoch": 0.07522349007922087, + "grad_norm": 3.4022698402404785, + "learning_rate": 7.500545097754198e-06, + "loss": 0.3391, + "step": 2070 + }, + { + "epoch": 0.07558688858201905, + "grad_norm": 8.37547779083252, + "learning_rate": 7.536884948034015e-06, + "loss": 0.3119, + "step": 2080 + }, + { + "epoch": 0.07595028708481721, + "grad_norm": 6.2167558670043945, + "learning_rate": 7.573224798313831e-06, + "loss": 0.3247, + "step": 2090 + }, + { + "epoch": 0.07631368558761538, + "grad_norm": 81.76036834716797, + "learning_rate": 7.609564648593648e-06, + "loss": 0.4281, + "step": 2100 + }, + { + "epoch": 0.07667708409041354, + "grad_norm": 2.8961973190307617, + "learning_rate": 7.645904498873465e-06, + "loss": 0.3368, + "step": 2110 + }, + { + "epoch": 0.07704048259321171, + "grad_norm": 4.699477195739746, + "learning_rate": 7.682244349153282e-06, + "loss": 0.3403, + "step": 2120 + }, + { + "epoch": 0.07740388109600989, + "grad_norm": 4.429138660430908, + "learning_rate": 7.718584199433098e-06, + "loss": 0.3182, + "step": 2130 + }, + { + "epoch": 0.07776727959880805, + "grad_norm": 2.7269580364227295, + "learning_rate": 7.754924049712916e-06, + "loss": 0.2828, + "step": 2140 + }, + { + "epoch": 0.07813067810160622, + "grad_norm": 15.126232147216797, + "learning_rate": 7.791263899992732e-06, + "loss": 0.4606, + "step": 2150 + }, + { + "epoch": 0.07849407660440438, + "grad_norm": 10.14072322845459, + "learning_rate": 7.82760375027255e-06, + "loss": 0.3451, + "step": 2160 + }, + { + "epoch": 0.07885747510720256, + "grad_norm": 4.95914363861084, + "learning_rate": 7.863943600552365e-06, + "loss": 0.3612, + "step": 2170 + }, + { + "epoch": 0.07922087361000073, + "grad_norm": 4.115192413330078, + "learning_rate": 7.900283450832183e-06, + "loss": 0.3222, + "step": 2180 + }, + { + "epoch": 0.07958427211279889, + "grad_norm": 5.405594825744629, + "learning_rate": 7.936623301111999e-06, + "loss": 0.3474, + "step": 2190 + }, + { + "epoch": 0.07994767061559706, + "grad_norm": 23.328718185424805, + "learning_rate": 7.972963151391817e-06, + "loss": 0.4797, + "step": 2200 + }, + { + "epoch": 0.08031106911839524, + "grad_norm": 3.5595099925994873, + "learning_rate": 8.009303001671634e-06, + "loss": 0.3305, + "step": 2210 + }, + { + "epoch": 0.0806744676211934, + "grad_norm": 3.048445463180542, + "learning_rate": 8.04564285195145e-06, + "loss": 0.318, + "step": 2220 + }, + { + "epoch": 0.08103786612399157, + "grad_norm": 5.857702732086182, + "learning_rate": 8.081982702231266e-06, + "loss": 0.3497, + "step": 2230 + }, + { + "epoch": 0.08140126462678973, + "grad_norm": 3.0092968940734863, + "learning_rate": 8.118322552511084e-06, + "loss": 0.2995, + "step": 2240 + }, + { + "epoch": 0.08176466312958791, + "grad_norm": 9.337843894958496, + "learning_rate": 8.154662402790902e-06, + "loss": 0.4517, + "step": 2250 + }, + { + "epoch": 0.08212806163238608, + "grad_norm": 3.136950969696045, + "learning_rate": 8.191002253070718e-06, + "loss": 0.2927, + "step": 2260 + }, + { + "epoch": 0.08249146013518424, + "grad_norm": 4.228198051452637, + "learning_rate": 8.227342103350534e-06, + "loss": 1.2185, + "step": 2270 + }, + { + "epoch": 0.0828548586379824, + "grad_norm": 7.404679298400879, + "learning_rate": 8.263681953630351e-06, + "loss": 0.3448, + "step": 2280 + }, + { + "epoch": 0.08321825714078059, + "grad_norm": 7.873497009277344, + "learning_rate": 8.300021803910169e-06, + "loss": 0.2965, + "step": 2290 + }, + { + "epoch": 0.08358165564357875, + "grad_norm": 12.266081809997559, + "learning_rate": 8.336361654189985e-06, + "loss": 0.4631, + "step": 2300 + }, + { + "epoch": 0.08394505414637692, + "grad_norm": 3.3576557636260986, + "learning_rate": 8.3727015044698e-06, + "loss": 0.3339, + "step": 2310 + }, + { + "epoch": 0.08430845264917508, + "grad_norm": 3.0854902267456055, + "learning_rate": 8.40904135474962e-06, + "loss": 0.3448, + "step": 2320 + }, + { + "epoch": 0.08467185115197326, + "grad_norm": 6.1308746337890625, + "learning_rate": 8.445381205029436e-06, + "loss": 0.386, + "step": 2330 + }, + { + "epoch": 0.08503524965477142, + "grad_norm": 4.458275318145752, + "learning_rate": 8.481721055309252e-06, + "loss": 0.2916, + "step": 2340 + }, + { + "epoch": 0.08539864815756959, + "grad_norm": 25.443647384643555, + "learning_rate": 8.51806090558907e-06, + "loss": 0.4232, + "step": 2350 + }, + { + "epoch": 0.08576204666036776, + "grad_norm": 324.4353332519531, + "learning_rate": 8.554400755868887e-06, + "loss": 2.4995, + "step": 2360 + }, + { + "epoch": 0.08612544516316593, + "grad_norm": 17.593692779541016, + "learning_rate": 8.590740606148703e-06, + "loss": 0.2952, + "step": 2370 + }, + { + "epoch": 0.0864888436659641, + "grad_norm": 3.4646732807159424, + "learning_rate": 8.62708045642852e-06, + "loss": 0.2961, + "step": 2380 + }, + { + "epoch": 0.08685224216876226, + "grad_norm": 2.9895999431610107, + "learning_rate": 8.663420306708337e-06, + "loss": 0.2852, + "step": 2390 + }, + { + "epoch": 0.08721564067156043, + "grad_norm": 24.221176147460938, + "learning_rate": 8.699760156988155e-06, + "loss": 0.4343, + "step": 2400 + }, + { + "epoch": 0.08721564067156043, + "eval_loss": 0.47036415338516235, + "eval_runtime": 180.1154, + "eval_samples_per_second": 41.162, + "eval_steps_per_second": 5.147, + "eval_wer": 0.28054713453264835, + "step": 2400 + }, + { + "epoch": 0.08757903917435861, + "grad_norm": 161.69967651367188, + "learning_rate": 8.73610000726797e-06, + "loss": 1.4598, + "step": 2410 + }, + { + "epoch": 0.08794243767715677, + "grad_norm": 10.37559700012207, + "learning_rate": 8.772439857547786e-06, + "loss": 0.3042, + "step": 2420 + }, + { + "epoch": 0.08830583617995494, + "grad_norm": 5.90106725692749, + "learning_rate": 8.808779707827604e-06, + "loss": 0.3385, + "step": 2430 + }, + { + "epoch": 0.0886692346827531, + "grad_norm": 9.207955360412598, + "learning_rate": 8.845119558107422e-06, + "loss": 0.2963, + "step": 2440 + }, + { + "epoch": 0.08903263318555127, + "grad_norm": 22.280956268310547, + "learning_rate": 8.881459408387238e-06, + "loss": 0.4505, + "step": 2450 + }, + { + "epoch": 0.08939603168834945, + "grad_norm": 3.090710401535034, + "learning_rate": 8.917799258667055e-06, + "loss": 0.3114, + "step": 2460 + }, + { + "epoch": 0.08975943019114761, + "grad_norm": 4.144134044647217, + "learning_rate": 8.954139108946871e-06, + "loss": 0.2855, + "step": 2470 + }, + { + "epoch": 0.09012282869394578, + "grad_norm": 4.343112468719482, + "learning_rate": 8.990478959226687e-06, + "loss": 0.2906, + "step": 2480 + }, + { + "epoch": 0.09048622719674394, + "grad_norm": 2.6925292015075684, + "learning_rate": 9.026818809506505e-06, + "loss": 0.284, + "step": 2490 + }, + { + "epoch": 0.09084962569954212, + "grad_norm": 29.639341354370117, + "learning_rate": 9.063158659786323e-06, + "loss": 0.3411, + "step": 2500 + }, + { + "epoch": 0.09121302420234029, + "grad_norm": 4.425374984741211, + "learning_rate": 9.099498510066139e-06, + "loss": 0.3041, + "step": 2510 + }, + { + "epoch": 0.09157642270513845, + "grad_norm": 5.6643195152282715, + "learning_rate": 9.135838360345955e-06, + "loss": 0.3123, + "step": 2520 + }, + { + "epoch": 0.09193982120793662, + "grad_norm": 3.9098479747772217, + "learning_rate": 9.172178210625772e-06, + "loss": 0.3664, + "step": 2530 + }, + { + "epoch": 0.0923032197107348, + "grad_norm": 3.133389949798584, + "learning_rate": 9.20851806090559e-06, + "loss": 0.2708, + "step": 2540 + }, + { + "epoch": 0.09266661821353296, + "grad_norm": 43.00468063354492, + "learning_rate": 9.244857911185406e-06, + "loss": 0.4215, + "step": 2550 + }, + { + "epoch": 0.09303001671633113, + "grad_norm": 3.1411876678466797, + "learning_rate": 9.281197761465222e-06, + "loss": 0.2983, + "step": 2560 + }, + { + "epoch": 0.09339341521912929, + "grad_norm": 3.263828754425049, + "learning_rate": 9.317537611745041e-06, + "loss": 0.3137, + "step": 2570 + }, + { + "epoch": 0.09375681372192747, + "grad_norm": 3.618751049041748, + "learning_rate": 9.353877462024857e-06, + "loss": 0.3279, + "step": 2580 + }, + { + "epoch": 0.09412021222472564, + "grad_norm": 3.6551568508148193, + "learning_rate": 9.390217312304673e-06, + "loss": 0.2409, + "step": 2590 + }, + { + "epoch": 0.0944836107275238, + "grad_norm": 8.680901527404785, + "learning_rate": 9.42655716258449e-06, + "loss": 0.373, + "step": 2600 + }, + { + "epoch": 0.09484700923032197, + "grad_norm": 4.761026382446289, + "learning_rate": 9.462897012864308e-06, + "loss": 0.2777, + "step": 2610 + }, + { + "epoch": 0.09521040773312014, + "grad_norm": 3.142723321914673, + "learning_rate": 9.499236863144124e-06, + "loss": 0.2882, + "step": 2620 + }, + { + "epoch": 0.09557380623591831, + "grad_norm": 2.969968795776367, + "learning_rate": 9.53557671342394e-06, + "loss": 0.3086, + "step": 2630 + }, + { + "epoch": 0.09593720473871648, + "grad_norm": 3.754549264907837, + "learning_rate": 9.571916563703758e-06, + "loss": 0.259, + "step": 2640 + }, + { + "epoch": 0.09630060324151464, + "grad_norm": 23.7288761138916, + "learning_rate": 9.608256413983576e-06, + "loss": 0.4284, + "step": 2650 + }, + { + "epoch": 0.09666400174431282, + "grad_norm": 2.7727372646331787, + "learning_rate": 9.644596264263392e-06, + "loss": 0.2602, + "step": 2660 + }, + { + "epoch": 0.09702740024711098, + "grad_norm": 14.707064628601074, + "learning_rate": 9.680936114543208e-06, + "loss": 0.3059, + "step": 2670 + }, + { + "epoch": 0.09739079874990915, + "grad_norm": 3.8396642208099365, + "learning_rate": 9.717275964823025e-06, + "loss": 0.2811, + "step": 2680 + }, + { + "epoch": 0.09775419725270731, + "grad_norm": 2.9460713863372803, + "learning_rate": 9.753615815102843e-06, + "loss": 0.2686, + "step": 2690 + }, + { + "epoch": 0.0981175957555055, + "grad_norm": 20.107336044311523, + "learning_rate": 9.789955665382659e-06, + "loss": 0.4306, + "step": 2700 + }, + { + "epoch": 0.09848099425830366, + "grad_norm": 3.1286280155181885, + "learning_rate": 9.826295515662477e-06, + "loss": 0.3059, + "step": 2710 + }, + { + "epoch": 0.09884439276110182, + "grad_norm": 6.160215854644775, + "learning_rate": 9.862635365942292e-06, + "loss": 0.3046, + "step": 2720 + }, + { + "epoch": 0.09920779126389999, + "grad_norm": 6.1921186447143555, + "learning_rate": 9.89897521622211e-06, + "loss": 0.285, + "step": 2730 + }, + { + "epoch": 0.09957118976669817, + "grad_norm": 13.759759902954102, + "learning_rate": 9.935315066501926e-06, + "loss": 0.2888, + "step": 2740 + }, + { + "epoch": 0.09993458826949633, + "grad_norm": 13.92764949798584, + "learning_rate": 9.971654916781744e-06, + "loss": 0.4266, + "step": 2750 + }, + { + "epoch": 0.1002979867722945, + "grad_norm": 3.3999857902526855, + "learning_rate": 1.000799476706156e-05, + "loss": 0.2858, + "step": 2760 + }, + { + "epoch": 0.10066138527509266, + "grad_norm": 4.103928089141846, + "learning_rate": 1.0044334617341377e-05, + "loss": 0.262, + "step": 2770 + }, + { + "epoch": 0.10102478377789084, + "grad_norm": 6.15985107421875, + "learning_rate": 1.0080674467621195e-05, + "loss": 0.2866, + "step": 2780 + }, + { + "epoch": 0.10138818228068901, + "grad_norm": 4.904097557067871, + "learning_rate": 1.0117014317901011e-05, + "loss": 0.5057, + "step": 2790 + }, + { + "epoch": 0.10175158078348717, + "grad_norm": 15.2875337600708, + "learning_rate": 1.0153354168180827e-05, + "loss": 0.4345, + "step": 2800 + }, + { + "epoch": 0.10211497928628534, + "grad_norm": 2.4697763919830322, + "learning_rate": 1.0189694018460643e-05, + "loss": 0.2693, + "step": 2810 + }, + { + "epoch": 0.1024783777890835, + "grad_norm": 5.04618501663208, + "learning_rate": 1.0226033868740462e-05, + "loss": 0.2868, + "step": 2820 + }, + { + "epoch": 0.10284177629188168, + "grad_norm": 5.851120948791504, + "learning_rate": 1.0262373719020278e-05, + "loss": 0.3425, + "step": 2830 + }, + { + "epoch": 0.10320517479467985, + "grad_norm": 2.1007258892059326, + "learning_rate": 1.0298713569300094e-05, + "loss": 0.2394, + "step": 2840 + }, + { + "epoch": 0.10356857329747801, + "grad_norm": 23.411701202392578, + "learning_rate": 1.0335053419579912e-05, + "loss": 0.4125, + "step": 2850 + }, + { + "epoch": 0.10393197180027618, + "grad_norm": 4.178852558135986, + "learning_rate": 1.037139326985973e-05, + "loss": 0.2951, + "step": 2860 + }, + { + "epoch": 0.10429537030307436, + "grad_norm": 1.7873708009719849, + "learning_rate": 1.0407733120139545e-05, + "loss": 0.3272, + "step": 2870 + }, + { + "epoch": 0.10465876880587252, + "grad_norm": 7.603367328643799, + "learning_rate": 1.0444072970419361e-05, + "loss": 0.2779, + "step": 2880 + }, + { + "epoch": 0.10502216730867069, + "grad_norm": 3.468761444091797, + "learning_rate": 1.0480412820699179e-05, + "loss": 0.3007, + "step": 2890 + }, + { + "epoch": 0.10538556581146885, + "grad_norm": 16.35407829284668, + "learning_rate": 1.0516752670978997e-05, + "loss": 0.3918, + "step": 2900 + }, + { + "epoch": 0.10574896431426703, + "grad_norm": 3.4226725101470947, + "learning_rate": 1.0553092521258813e-05, + "loss": 3.7156, + "step": 2910 + }, + { + "epoch": 0.1061123628170652, + "grad_norm": 9.006295204162598, + "learning_rate": 1.058943237153863e-05, + "loss": 0.4075, + "step": 2920 + }, + { + "epoch": 0.10647576131986336, + "grad_norm": 4.993385314941406, + "learning_rate": 1.0625772221818446e-05, + "loss": 0.3588, + "step": 2930 + }, + { + "epoch": 0.10683915982266153, + "grad_norm": 3.7684736251831055, + "learning_rate": 1.0662112072098264e-05, + "loss": 0.2429, + "step": 2940 + }, + { + "epoch": 0.1072025583254597, + "grad_norm": 40.301170349121094, + "learning_rate": 1.069845192237808e-05, + "loss": 0.4739, + "step": 2950 + }, + { + "epoch": 0.10756595682825787, + "grad_norm": 3.772693157196045, + "learning_rate": 1.0734791772657898e-05, + "loss": 0.3284, + "step": 2960 + }, + { + "epoch": 0.10792935533105603, + "grad_norm": 3.0183212757110596, + "learning_rate": 1.0771131622937714e-05, + "loss": 0.38, + "step": 2970 + }, + { + "epoch": 0.1082927538338542, + "grad_norm": 6.61776876449585, + "learning_rate": 1.0807471473217531e-05, + "loss": 0.2793, + "step": 2980 + }, + { + "epoch": 0.10865615233665238, + "grad_norm": 6.112472057342529, + "learning_rate": 1.0843811323497347e-05, + "loss": 0.2447, + "step": 2990 + }, + { + "epoch": 0.10901955083945054, + "grad_norm": 10.800559997558594, + "learning_rate": 1.0880151173777165e-05, + "loss": 0.373, + "step": 3000 + }, + { + "epoch": 0.10901955083945054, + "eval_loss": 0.4652940630912781, + "eval_runtime": 180.0765, + "eval_samples_per_second": 41.171, + "eval_steps_per_second": 5.148, + "eval_wer": 0.27681667181004593, + "step": 3000 + }, + { + "epoch": 0.10938294934224871, + "grad_norm": 7.778831958770752, + "learning_rate": 1.091649102405698e-05, + "loss": 0.29, + "step": 3010 + }, + { + "epoch": 0.10974634784504687, + "grad_norm": 2.855592966079712, + "learning_rate": 1.0952830874336798e-05, + "loss": 0.2411, + "step": 3020 + }, + { + "epoch": 0.11010974634784505, + "grad_norm": 4.229335784912109, + "learning_rate": 1.0989170724616616e-05, + "loss": 0.3247, + "step": 3030 + }, + { + "epoch": 0.11047314485064322, + "grad_norm": 3.8145949840545654, + "learning_rate": 1.1025510574896432e-05, + "loss": 0.2242, + "step": 3040 + }, + { + "epoch": 0.11083654335344138, + "grad_norm": 22.571304321289062, + "learning_rate": 1.1061850425176248e-05, + "loss": 0.3959, + "step": 3050 + }, + { + "epoch": 0.11119994185623955, + "grad_norm": 2.4706461429595947, + "learning_rate": 1.1098190275456066e-05, + "loss": 0.2466, + "step": 3060 + }, + { + "epoch": 0.11156334035903773, + "grad_norm": 4.497069358825684, + "learning_rate": 1.1134530125735883e-05, + "loss": 2.1968, + "step": 3070 + }, + { + "epoch": 0.11192673886183589, + "grad_norm": 5.060062885284424, + "learning_rate": 1.11708699760157e-05, + "loss": 0.2921, + "step": 3080 + }, + { + "epoch": 0.11229013736463406, + "grad_norm": 2.7882325649261475, + "learning_rate": 1.1207209826295515e-05, + "loss": 0.2534, + "step": 3090 + }, + { + "epoch": 0.11265353586743222, + "grad_norm": 9.96241569519043, + "learning_rate": 1.1243549676575333e-05, + "loss": 0.421, + "step": 3100 + }, + { + "epoch": 0.1130169343702304, + "grad_norm": 31.262916564941406, + "learning_rate": 1.127988952685515e-05, + "loss": 0.4048, + "step": 3110 + }, + { + "epoch": 0.11338033287302857, + "grad_norm": 3.472343921661377, + "learning_rate": 1.1316229377134967e-05, + "loss": 0.2798, + "step": 3120 + }, + { + "epoch": 0.11374373137582673, + "grad_norm": 4.074085235595703, + "learning_rate": 1.1352569227414783e-05, + "loss": 0.299, + "step": 3130 + }, + { + "epoch": 0.1141071298786249, + "grad_norm": 2.879512310028076, + "learning_rate": 1.1388909077694602e-05, + "loss": 0.2137, + "step": 3140 + }, + { + "epoch": 0.11447052838142306, + "grad_norm": 125.17889404296875, + "learning_rate": 1.1425248927974418e-05, + "loss": 0.5418, + "step": 3150 + }, + { + "epoch": 0.11483392688422124, + "grad_norm": 4.171487808227539, + "learning_rate": 1.1461588778254234e-05, + "loss": 0.2685, + "step": 3160 + }, + { + "epoch": 0.1151973253870194, + "grad_norm": 2.1496529579162598, + "learning_rate": 1.1497928628534051e-05, + "loss": 0.2421, + "step": 3170 + }, + { + "epoch": 0.11556072388981757, + "grad_norm": 2.6266047954559326, + "learning_rate": 1.1534268478813867e-05, + "loss": 0.3288, + "step": 3180 + }, + { + "epoch": 0.11592412239261574, + "grad_norm": 3.7677230834960938, + "learning_rate": 1.1570608329093685e-05, + "loss": 0.3093, + "step": 3190 + }, + { + "epoch": 0.11628752089541392, + "grad_norm": 9.4945707321167, + "learning_rate": 1.1606948179373501e-05, + "loss": 0.3066, + "step": 3200 + }, + { + "epoch": 0.11665091939821208, + "grad_norm": 2.5509915351867676, + "learning_rate": 1.1643288029653319e-05, + "loss": 0.2615, + "step": 3210 + }, + { + "epoch": 0.11701431790101025, + "grad_norm": 3.066624641418457, + "learning_rate": 1.1679627879933135e-05, + "loss": 0.3224, + "step": 3220 + }, + { + "epoch": 0.11737771640380841, + "grad_norm": 6.494440078735352, + "learning_rate": 1.1715967730212952e-05, + "loss": 0.3017, + "step": 3230 + }, + { + "epoch": 0.11774111490660659, + "grad_norm": 3.4675605297088623, + "learning_rate": 1.1752307580492768e-05, + "loss": 0.2152, + "step": 3240 + }, + { + "epoch": 0.11810451340940475, + "grad_norm": 15.5110445022583, + "learning_rate": 1.1788647430772586e-05, + "loss": 0.349, + "step": 3250 + }, + { + "epoch": 0.11846791191220292, + "grad_norm": 1.972530484199524, + "learning_rate": 1.1824987281052402e-05, + "loss": 0.2728, + "step": 3260 + }, + { + "epoch": 0.11883131041500108, + "grad_norm": 4.018677711486816, + "learning_rate": 1.186132713133222e-05, + "loss": 0.254, + "step": 3270 + }, + { + "epoch": 0.11919470891779926, + "grad_norm": 4.95416784286499, + "learning_rate": 1.1897666981612037e-05, + "loss": 0.2465, + "step": 3280 + }, + { + "epoch": 0.11955810742059743, + "grad_norm": 3.165599822998047, + "learning_rate": 1.1934006831891853e-05, + "loss": 0.2537, + "step": 3290 + }, + { + "epoch": 0.1199215059233956, + "grad_norm": 8.508636474609375, + "learning_rate": 1.1970346682171669e-05, + "loss": 0.3655, + "step": 3300 + }, + { + "epoch": 0.12028490442619376, + "grad_norm": 2.3892879486083984, + "learning_rate": 1.2006686532451487e-05, + "loss": 0.252, + "step": 3310 + }, + { + "epoch": 0.12064830292899194, + "grad_norm": 3.591564178466797, + "learning_rate": 1.2043026382731304e-05, + "loss": 0.2401, + "step": 3320 + }, + { + "epoch": 0.1210117014317901, + "grad_norm": 3.891261577606201, + "learning_rate": 1.207936623301112e-05, + "loss": 0.2909, + "step": 3330 + }, + { + "epoch": 0.12137509993458827, + "grad_norm": 4.691511154174805, + "learning_rate": 1.2115706083290936e-05, + "loss": 0.2304, + "step": 3340 + }, + { + "epoch": 0.12173849843738643, + "grad_norm": 18.415170669555664, + "learning_rate": 1.2152045933570754e-05, + "loss": 0.35, + "step": 3350 + }, + { + "epoch": 0.12210189694018461, + "grad_norm": 3.9105615615844727, + "learning_rate": 1.2188385783850572e-05, + "loss": 0.3112, + "step": 3360 + }, + { + "epoch": 0.12246529544298278, + "grad_norm": 3.215313196182251, + "learning_rate": 1.2224725634130388e-05, + "loss": 0.2492, + "step": 3370 + }, + { + "epoch": 0.12282869394578094, + "grad_norm": 9.30749225616455, + "learning_rate": 1.2261065484410204e-05, + "loss": 0.2696, + "step": 3380 + }, + { + "epoch": 0.12319209244857911, + "grad_norm": 4.9797682762146, + "learning_rate": 1.2297405334690023e-05, + "loss": 0.2197, + "step": 3390 + }, + { + "epoch": 0.12355549095137729, + "grad_norm": 19.632797241210938, + "learning_rate": 1.2333745184969839e-05, + "loss": 0.3411, + "step": 3400 + }, + { + "epoch": 0.12391888945417545, + "grad_norm": 4.509830474853516, + "learning_rate": 1.2370085035249655e-05, + "loss": 0.2394, + "step": 3410 + }, + { + "epoch": 0.12428228795697362, + "grad_norm": 2.253514051437378, + "learning_rate": 1.2406424885529473e-05, + "loss": 0.266, + "step": 3420 + }, + { + "epoch": 0.12464568645977178, + "grad_norm": 3.123828172683716, + "learning_rate": 1.244276473580929e-05, + "loss": 0.2675, + "step": 3430 + }, + { + "epoch": 0.12500908496256996, + "grad_norm": 34.37680435180664, + "learning_rate": 1.2479104586089106e-05, + "loss": 0.279, + "step": 3440 + }, + { + "epoch": 0.1253724834653681, + "grad_norm": 10.051690101623535, + "learning_rate": 1.2515444436368922e-05, + "loss": 0.3364, + "step": 3450 + }, + { + "epoch": 0.1257358819681663, + "grad_norm": 2.1765711307525635, + "learning_rate": 1.255178428664874e-05, + "loss": 0.2288, + "step": 3460 + }, + { + "epoch": 0.12609928047096447, + "grad_norm": 2.4910778999328613, + "learning_rate": 1.2588124136928556e-05, + "loss": 0.2866, + "step": 3470 + }, + { + "epoch": 0.12646267897376262, + "grad_norm": 7.379613876342773, + "learning_rate": 1.2624463987208373e-05, + "loss": 0.2618, + "step": 3480 + }, + { + "epoch": 0.1268260774765608, + "grad_norm": 2.681814432144165, + "learning_rate": 1.266080383748819e-05, + "loss": 0.2405, + "step": 3490 + }, + { + "epoch": 0.12718947597935895, + "grad_norm": 83.93474578857422, + "learning_rate": 1.2697143687768007e-05, + "loss": 0.338, + "step": 3500 + }, + { + "epoch": 0.12755287448215713, + "grad_norm": 1.5564826726913452, + "learning_rate": 1.2733483538047825e-05, + "loss": 0.2305, + "step": 3510 + }, + { + "epoch": 0.1279162729849553, + "grad_norm": 2.6026437282562256, + "learning_rate": 1.2769823388327639e-05, + "loss": 0.2618, + "step": 3520 + }, + { + "epoch": 0.12827967148775346, + "grad_norm": 8.228372573852539, + "learning_rate": 1.2806163238607458e-05, + "loss": 0.2586, + "step": 3530 + }, + { + "epoch": 0.12864306999055164, + "grad_norm": 2.643139362335205, + "learning_rate": 1.2842503088887276e-05, + "loss": 0.2197, + "step": 3540 + }, + { + "epoch": 0.12900646849334982, + "grad_norm": 400.0296325683594, + "learning_rate": 1.287884293916709e-05, + "loss": 0.3586, + "step": 3550 + }, + { + "epoch": 0.12936986699614797, + "grad_norm": 1.6349281072616577, + "learning_rate": 1.2915182789446908e-05, + "loss": 0.2364, + "step": 3560 + }, + { + "epoch": 0.12973326549894615, + "grad_norm": 2.6573753356933594, + "learning_rate": 1.2951522639726724e-05, + "loss": 0.2195, + "step": 3570 + }, + { + "epoch": 0.1300966640017443, + "grad_norm": 4.2721686363220215, + "learning_rate": 1.2987862490006542e-05, + "loss": 0.3092, + "step": 3580 + }, + { + "epoch": 0.13046006250454248, + "grad_norm": 2.9982502460479736, + "learning_rate": 1.302420234028636e-05, + "loss": 0.2826, + "step": 3590 + }, + { + "epoch": 0.13082346100734066, + "grad_norm": 8.903009414672852, + "learning_rate": 1.3060542190566175e-05, + "loss": 0.3367, + "step": 3600 + }, + { + "epoch": 0.13082346100734066, + "eval_loss": 0.4490436017513275, + "eval_runtime": 179.8743, + "eval_samples_per_second": 41.218, + "eval_steps_per_second": 5.154, + "eval_wer": 0.2664058670829778, + "step": 3600 + }, + { + "epoch": 0.1311868595101388, + "grad_norm": 2.9746363162994385, + "learning_rate": 1.3096882040845993e-05, + "loss": 0.2418, + "step": 3610 + }, + { + "epoch": 0.131550258012937, + "grad_norm": 2.274872303009033, + "learning_rate": 1.313322189112581e-05, + "loss": 0.3052, + "step": 3620 + }, + { + "epoch": 0.13191365651573517, + "grad_norm": 7.114847660064697, + "learning_rate": 1.3169561741405625e-05, + "loss": 0.2821, + "step": 3630 + }, + { + "epoch": 0.13227705501853332, + "grad_norm": 3.2101128101348877, + "learning_rate": 1.3205901591685444e-05, + "loss": 0.2223, + "step": 3640 + }, + { + "epoch": 0.1326404535213315, + "grad_norm": 18.914968490600586, + "learning_rate": 1.3242241441965258e-05, + "loss": 0.3809, + "step": 3650 + }, + { + "epoch": 0.13300385202412965, + "grad_norm": 2.399569272994995, + "learning_rate": 1.3278581292245076e-05, + "loss": 0.2221, + "step": 3660 + }, + { + "epoch": 0.13336725052692783, + "grad_norm": 5.76792573928833, + "learning_rate": 1.3314921142524894e-05, + "loss": 0.2487, + "step": 3670 + }, + { + "epoch": 0.133730649029726, + "grad_norm": 3.6859967708587646, + "learning_rate": 1.335126099280471e-05, + "loss": 0.2781, + "step": 3680 + }, + { + "epoch": 0.13409404753252416, + "grad_norm": 2.9653141498565674, + "learning_rate": 1.3387600843084527e-05, + "loss": 0.2258, + "step": 3690 + }, + { + "epoch": 0.13445744603532234, + "grad_norm": 19.170753479003906, + "learning_rate": 1.3423940693364345e-05, + "loss": 0.3902, + "step": 3700 + }, + { + "epoch": 0.13482084453812052, + "grad_norm": 2.2880115509033203, + "learning_rate": 1.3460280543644161e-05, + "loss": 0.2745, + "step": 3710 + }, + { + "epoch": 0.13518424304091867, + "grad_norm": 2.5196125507354736, + "learning_rate": 1.3496620393923979e-05, + "loss": 0.2293, + "step": 3720 + }, + { + "epoch": 0.13554764154371685, + "grad_norm": 3.827986001968384, + "learning_rate": 1.3532960244203793e-05, + "loss": 0.259, + "step": 3730 + }, + { + "epoch": 0.135911040046515, + "grad_norm": 3.4211530685424805, + "learning_rate": 1.356930009448361e-05, + "loss": 0.3256, + "step": 3740 + }, + { + "epoch": 0.13627443854931318, + "grad_norm": 26.879398345947266, + "learning_rate": 1.360563994476343e-05, + "loss": 0.3208, + "step": 3750 + }, + { + "epoch": 0.13663783705211135, + "grad_norm": 2.316091775894165, + "learning_rate": 1.3641979795043244e-05, + "loss": 0.2316, + "step": 3760 + }, + { + "epoch": 0.1370012355549095, + "grad_norm": 4.098924160003662, + "learning_rate": 1.3678319645323062e-05, + "loss": 0.2399, + "step": 3770 + }, + { + "epoch": 0.13736463405770769, + "grad_norm": 6.9372687339782715, + "learning_rate": 1.371465949560288e-05, + "loss": 0.2858, + "step": 3780 + }, + { + "epoch": 0.13772803256050586, + "grad_norm": 2.509535789489746, + "learning_rate": 1.3750999345882695e-05, + "loss": 0.2113, + "step": 3790 + }, + { + "epoch": 0.13809143106330402, + "grad_norm": 7.7181077003479, + "learning_rate": 1.3787339196162513e-05, + "loss": 0.3279, + "step": 3800 + }, + { + "epoch": 0.1384548295661022, + "grad_norm": 2.6843245029449463, + "learning_rate": 1.3823679046442329e-05, + "loss": 0.2267, + "step": 3810 + }, + { + "epoch": 0.13881822806890035, + "grad_norm": 3.05159068107605, + "learning_rate": 1.3860018896722147e-05, + "loss": 0.229, + "step": 3820 + }, + { + "epoch": 0.13918162657169852, + "grad_norm": 5.029635429382324, + "learning_rate": 1.3896358747001964e-05, + "loss": 0.2627, + "step": 3830 + }, + { + "epoch": 0.1395450250744967, + "grad_norm": 2.8287103176116943, + "learning_rate": 1.3932698597281779e-05, + "loss": 0.2294, + "step": 3840 + }, + { + "epoch": 0.13990842357729485, + "grad_norm": 24.862224578857422, + "learning_rate": 1.3969038447561598e-05, + "loss": 0.3198, + "step": 3850 + }, + { + "epoch": 0.14027182208009303, + "grad_norm": 5.624647617340088, + "learning_rate": 1.4005378297841412e-05, + "loss": 0.2641, + "step": 3860 + }, + { + "epoch": 0.14063522058289118, + "grad_norm": 1.6199389696121216, + "learning_rate": 1.404171814812123e-05, + "loss": 0.2279, + "step": 3870 + }, + { + "epoch": 0.14099861908568936, + "grad_norm": 2.864058017730713, + "learning_rate": 1.4078057998401047e-05, + "loss": 0.2448, + "step": 3880 + }, + { + "epoch": 0.14136201758848754, + "grad_norm": 3.897899627685547, + "learning_rate": 1.4114397848680863e-05, + "loss": 0.2438, + "step": 3890 + }, + { + "epoch": 0.1417254160912857, + "grad_norm": 42.4840087890625, + "learning_rate": 1.4150737698960681e-05, + "loss": 0.3604, + "step": 3900 + }, + { + "epoch": 0.14208881459408387, + "grad_norm": 1.6532913446426392, + "learning_rate": 1.4187077549240499e-05, + "loss": 0.2469, + "step": 3910 + }, + { + "epoch": 0.14245221309688205, + "grad_norm": 2.3755931854248047, + "learning_rate": 1.4223417399520315e-05, + "loss": 0.2233, + "step": 3920 + }, + { + "epoch": 0.1428156115996802, + "grad_norm": 5.866461277008057, + "learning_rate": 1.4259757249800132e-05, + "loss": 0.2952, + "step": 3930 + }, + { + "epoch": 0.14317901010247838, + "grad_norm": 3.171570301055908, + "learning_rate": 1.4296097100079947e-05, + "loss": 0.2329, + "step": 3940 + }, + { + "epoch": 0.14354240860527653, + "grad_norm": 23.302635192871094, + "learning_rate": 1.4332436950359764e-05, + "loss": 0.3592, + "step": 3950 + }, + { + "epoch": 0.1439058071080747, + "grad_norm": 2.3609213829040527, + "learning_rate": 1.4368776800639584e-05, + "loss": 1.1975, + "step": 3960 + }, + { + "epoch": 0.1442692056108729, + "grad_norm": 2.857872486114502, + "learning_rate": 1.4405116650919398e-05, + "loss": 0.265, + "step": 3970 + }, + { + "epoch": 0.14463260411367104, + "grad_norm": 6.918335914611816, + "learning_rate": 1.4441456501199216e-05, + "loss": 0.4057, + "step": 3980 + }, + { + "epoch": 0.14499600261646922, + "grad_norm": 3.8019461631774902, + "learning_rate": 1.4477796351479033e-05, + "loss": 0.299, + "step": 3990 + }, + { + "epoch": 0.1453594011192674, + "grad_norm": 30.963428497314453, + "learning_rate": 1.451413620175885e-05, + "loss": 0.335, + "step": 4000 + }, + { + "epoch": 0.14572279962206555, + "grad_norm": 2.3968963623046875, + "learning_rate": 1.4550476052038667e-05, + "loss": 0.224, + "step": 4010 + }, + { + "epoch": 0.14608619812486373, + "grad_norm": 6.7229485511779785, + "learning_rate": 1.4586815902318481e-05, + "loss": 0.2657, + "step": 4020 + }, + { + "epoch": 0.14644959662766188, + "grad_norm": 17.447879791259766, + "learning_rate": 1.46231557525983e-05, + "loss": 0.2199, + "step": 4030 + }, + { + "epoch": 0.14681299513046006, + "grad_norm": 2.020756721496582, + "learning_rate": 1.4659495602878118e-05, + "loss": 0.31, + "step": 4040 + }, + { + "epoch": 0.14717639363325824, + "grad_norm": 38.28268814086914, + "learning_rate": 1.4695835453157932e-05, + "loss": 0.3861, + "step": 4050 + }, + { + "epoch": 0.1475397921360564, + "grad_norm": 2.085073232650757, + "learning_rate": 1.473217530343775e-05, + "loss": 0.2078, + "step": 4060 + }, + { + "epoch": 0.14790319063885457, + "grad_norm": 3.453597068786621, + "learning_rate": 1.476851515371757e-05, + "loss": 0.296, + "step": 4070 + }, + { + "epoch": 0.14826658914165275, + "grad_norm": 2.3039424419403076, + "learning_rate": 1.4804855003997384e-05, + "loss": 0.2346, + "step": 4080 + }, + { + "epoch": 0.1486299876444509, + "grad_norm": 3.217890977859497, + "learning_rate": 1.4841194854277201e-05, + "loss": 0.2243, + "step": 4090 + }, + { + "epoch": 0.14899338614724908, + "grad_norm": 12.48748779296875, + "learning_rate": 1.4877534704557017e-05, + "loss": 0.3378, + "step": 4100 + }, + { + "epoch": 0.14935678465004723, + "grad_norm": 2.781388282775879, + "learning_rate": 1.4913874554836835e-05, + "loss": 0.2167, + "step": 4110 + }, + { + "epoch": 0.1497201831528454, + "grad_norm": 2.564457893371582, + "learning_rate": 1.4950214405116653e-05, + "loss": 0.2187, + "step": 4120 + }, + { + "epoch": 0.1500835816556436, + "grad_norm": 9.590895652770996, + "learning_rate": 1.4986554255396469e-05, + "loss": 0.2444, + "step": 4130 + }, + { + "epoch": 0.15044698015844174, + "grad_norm": 2.8055028915405273, + "learning_rate": 1.5022894105676286e-05, + "loss": 0.2499, + "step": 4140 + }, + { + "epoch": 0.15081037866123992, + "grad_norm": 7.157045364379883, + "learning_rate": 1.5059233955956104e-05, + "loss": 0.361, + "step": 4150 + }, + { + "epoch": 0.1511737771640381, + "grad_norm": 3.369006633758545, + "learning_rate": 1.5095573806235918e-05, + "loss": 0.259, + "step": 4160 + }, + { + "epoch": 0.15153717566683625, + "grad_norm": 5.334355354309082, + "learning_rate": 1.5131913656515736e-05, + "loss": 0.2797, + "step": 4170 + }, + { + "epoch": 0.15190057416963443, + "grad_norm": 6.667120456695557, + "learning_rate": 1.5168253506795552e-05, + "loss": 0.2678, + "step": 4180 + }, + { + "epoch": 0.15226397267243258, + "grad_norm": 1.7419887781143188, + "learning_rate": 1.520459335707537e-05, + "loss": 0.2432, + "step": 4190 + }, + { + "epoch": 0.15262737117523076, + "grad_norm": 7.022573947906494, + "learning_rate": 1.5240933207355187e-05, + "loss": 0.2955, + "step": 4200 + }, + { + "epoch": 0.15262737117523076, + "eval_loss": 0.4361402690410614, + "eval_runtime": 180.5933, + "eval_samples_per_second": 41.054, + "eval_steps_per_second": 5.133, + "eval_wer": 0.2589540181894095, + "step": 4200 + }, + { + "epoch": 0.15299076967802894, + "grad_norm": 3.108078718185425, + "learning_rate": 1.5277273057635e-05, + "loss": 3.6147, + "step": 4210 + }, + { + "epoch": 0.1533541681808271, + "grad_norm": 2.6063787937164307, + "learning_rate": 1.531361290791482e-05, + "loss": 0.232, + "step": 4220 + }, + { + "epoch": 0.15371756668362527, + "grad_norm": 3.581697463989258, + "learning_rate": 1.5349952758194637e-05, + "loss": 0.2451, + "step": 4230 + }, + { + "epoch": 0.15408096518642342, + "grad_norm": 2.5910837650299072, + "learning_rate": 1.5386292608474453e-05, + "loss": 0.2283, + "step": 4240 + }, + { + "epoch": 0.1544443636892216, + "grad_norm": 70.38739013671875, + "learning_rate": 1.5422632458754272e-05, + "loss": 0.396, + "step": 4250 + }, + { + "epoch": 0.15480776219201978, + "grad_norm": 3.5658187866210938, + "learning_rate": 1.5458972309034088e-05, + "loss": 0.2116, + "step": 4260 + }, + { + "epoch": 0.15517116069481793, + "grad_norm": 5.393126487731934, + "learning_rate": 1.5495312159313904e-05, + "loss": 0.2382, + "step": 4270 + }, + { + "epoch": 0.1555345591976161, + "grad_norm": 10.135586738586426, + "learning_rate": 1.5531652009593723e-05, + "loss": 0.2485, + "step": 4280 + }, + { + "epoch": 0.15589795770041429, + "grad_norm": 2.1143031120300293, + "learning_rate": 1.5567991859873536e-05, + "loss": 0.1936, + "step": 4290 + }, + { + "epoch": 0.15626135620321244, + "grad_norm": 20.077383041381836, + "learning_rate": 1.5604331710153355e-05, + "loss": 0.3818, + "step": 4300 + }, + { + "epoch": 0.15662475470601062, + "grad_norm": 3.793126344680786, + "learning_rate": 1.564067156043317e-05, + "loss": 0.2245, + "step": 4310 + }, + { + "epoch": 0.15698815320880877, + "grad_norm": 3.2057955265045166, + "learning_rate": 1.5677011410712987e-05, + "loss": 0.2551, + "step": 4320 + }, + { + "epoch": 0.15735155171160695, + "grad_norm": 5.002716064453125, + "learning_rate": 1.5713351260992806e-05, + "loss": 0.2951, + "step": 4330 + }, + { + "epoch": 0.15771495021440513, + "grad_norm": 2.2240726947784424, + "learning_rate": 1.5749691111272622e-05, + "loss": 0.1993, + "step": 4340 + }, + { + "epoch": 0.15807834871720328, + "grad_norm": 55.30891036987305, + "learning_rate": 1.578603096155244e-05, + "loss": 0.2803, + "step": 4350 + }, + { + "epoch": 0.15844174722000146, + "grad_norm": 1.9186596870422363, + "learning_rate": 1.5822370811832258e-05, + "loss": 0.2234, + "step": 4360 + }, + { + "epoch": 0.15880514572279963, + "grad_norm": 1.7817661762237549, + "learning_rate": 1.5858710662112074e-05, + "loss": 0.2038, + "step": 4370 + }, + { + "epoch": 0.15916854422559779, + "grad_norm": 3.046330690383911, + "learning_rate": 1.589505051239189e-05, + "loss": 0.2809, + "step": 4380 + }, + { + "epoch": 0.15953194272839596, + "grad_norm": 5.43302583694458, + "learning_rate": 1.5931390362671706e-05, + "loss": 0.1896, + "step": 4390 + }, + { + "epoch": 0.15989534123119412, + "grad_norm": 12.185855865478516, + "learning_rate": 1.596773021295152e-05, + "loss": 0.2984, + "step": 4400 + }, + { + "epoch": 0.1602587397339923, + "grad_norm": 1.9507842063903809, + "learning_rate": 1.600407006323134e-05, + "loss": 0.2064, + "step": 4410 + }, + { + "epoch": 0.16062213823679047, + "grad_norm": 4.536543846130371, + "learning_rate": 1.6040409913511157e-05, + "loss": 0.2433, + "step": 4420 + }, + { + "epoch": 0.16098553673958862, + "grad_norm": 3.101174831390381, + "learning_rate": 1.6076749763790973e-05, + "loss": 0.2746, + "step": 4430 + }, + { + "epoch": 0.1613489352423868, + "grad_norm": 2.2098021507263184, + "learning_rate": 1.6113089614070792e-05, + "loss": 0.17, + "step": 4440 + }, + { + "epoch": 0.16171233374518498, + "grad_norm": 59.360809326171875, + "learning_rate": 1.6149429464350608e-05, + "loss": 0.379, + "step": 4450 + }, + { + "epoch": 0.16207573224798313, + "grad_norm": 6.364736557006836, + "learning_rate": 1.6185769314630424e-05, + "loss": 0.2224, + "step": 4460 + }, + { + "epoch": 0.1624391307507813, + "grad_norm": 3.2455356121063232, + "learning_rate": 1.622210916491024e-05, + "loss": 0.2195, + "step": 4470 + }, + { + "epoch": 0.16280252925357946, + "grad_norm": 6.399629592895508, + "learning_rate": 1.625844901519006e-05, + "loss": 0.266, + "step": 4480 + }, + { + "epoch": 0.16316592775637764, + "grad_norm": 16.19785499572754, + "learning_rate": 1.6294788865469875e-05, + "loss": 0.1836, + "step": 4490 + }, + { + "epoch": 0.16352932625917582, + "grad_norm": 7.909778594970703, + "learning_rate": 1.633112871574969e-05, + "loss": 0.6016, + "step": 4500 + }, + { + "epoch": 0.16389272476197397, + "grad_norm": 2.8134663105010986, + "learning_rate": 1.636746856602951e-05, + "loss": 0.2148, + "step": 4510 + }, + { + "epoch": 0.16425612326477215, + "grad_norm": 2.667999505996704, + "learning_rate": 1.6403808416309327e-05, + "loss": 0.2294, + "step": 4520 + }, + { + "epoch": 0.1646195217675703, + "grad_norm": 3.355242967605591, + "learning_rate": 1.6440148266589143e-05, + "loss": 0.2097, + "step": 4530 + }, + { + "epoch": 0.16498292027036848, + "grad_norm": 2.6241908073425293, + "learning_rate": 1.647648811686896e-05, + "loss": 0.2337, + "step": 4540 + }, + { + "epoch": 0.16534631877316666, + "grad_norm": 16.759428024291992, + "learning_rate": 1.6512827967148775e-05, + "loss": 0.2944, + "step": 4550 + }, + { + "epoch": 0.1657097172759648, + "grad_norm": 3.098898410797119, + "learning_rate": 1.6549167817428594e-05, + "loss": 0.1895, + "step": 4560 + }, + { + "epoch": 0.166073115778763, + "grad_norm": 4.042644023895264, + "learning_rate": 1.658550766770841e-05, + "loss": 0.2369, + "step": 4570 + }, + { + "epoch": 0.16643651428156117, + "grad_norm": 7.174807548522949, + "learning_rate": 1.6621847517988226e-05, + "loss": 0.2331, + "step": 4580 + }, + { + "epoch": 0.16679991278435932, + "grad_norm": 2.1805012226104736, + "learning_rate": 1.6658187368268045e-05, + "loss": 0.2422, + "step": 4590 + }, + { + "epoch": 0.1671633112871575, + "grad_norm": 18.097871780395508, + "learning_rate": 1.6694527218547858e-05, + "loss": 0.347, + "step": 4600 + }, + { + "epoch": 0.16752670978995565, + "grad_norm": 3.48561429977417, + "learning_rate": 1.6730867068827677e-05, + "loss": 0.2985, + "step": 4610 + }, + { + "epoch": 0.16789010829275383, + "grad_norm": 1.7519229650497437, + "learning_rate": 1.6767206919107496e-05, + "loss": 0.2204, + "step": 4620 + }, + { + "epoch": 0.168253506795552, + "grad_norm": 3.7641661167144775, + "learning_rate": 1.680354676938731e-05, + "loss": 0.2348, + "step": 4630 + }, + { + "epoch": 0.16861690529835016, + "grad_norm": 3.0688085556030273, + "learning_rate": 1.683988661966713e-05, + "loss": 0.2147, + "step": 4640 + }, + { + "epoch": 0.16898030380114834, + "grad_norm": 25.845094680786133, + "learning_rate": 1.6876226469946944e-05, + "loss": 0.3671, + "step": 4650 + }, + { + "epoch": 0.16934370230394652, + "grad_norm": 2.841994524002075, + "learning_rate": 1.691256632022676e-05, + "loss": 0.2182, + "step": 4660 + }, + { + "epoch": 0.16970710080674467, + "grad_norm": 1.0501997470855713, + "learning_rate": 1.694890617050658e-05, + "loss": 0.1791, + "step": 4670 + }, + { + "epoch": 0.17007049930954285, + "grad_norm": 3.3973441123962402, + "learning_rate": 1.6985246020786392e-05, + "loss": 0.3338, + "step": 4680 + }, + { + "epoch": 0.170433897812341, + "grad_norm": 1.8442267179489136, + "learning_rate": 1.702158587106621e-05, + "loss": 0.2528, + "step": 4690 + }, + { + "epoch": 0.17079729631513918, + "grad_norm": 42.373409271240234, + "learning_rate": 1.705792572134603e-05, + "loss": 0.2892, + "step": 4700 + }, + { + "epoch": 0.17116069481793736, + "grad_norm": 6.344671726226807, + "learning_rate": 1.7094265571625844e-05, + "loss": 0.2474, + "step": 4710 + }, + { + "epoch": 0.1715240933207355, + "grad_norm": 1.6177664995193481, + "learning_rate": 1.7130605421905663e-05, + "loss": 0.2364, + "step": 4720 + }, + { + "epoch": 0.1718874918235337, + "grad_norm": 4.98591423034668, + "learning_rate": 1.7166945272185482e-05, + "loss": 0.2046, + "step": 4730 + }, + { + "epoch": 0.17225089032633187, + "grad_norm": 7.943169116973877, + "learning_rate": 1.7203285122465295e-05, + "loss": 0.293, + "step": 4740 + }, + { + "epoch": 0.17261428882913002, + "grad_norm": 7.402034759521484, + "learning_rate": 1.7239624972745114e-05, + "loss": 0.2722, + "step": 4750 + }, + { + "epoch": 0.1729776873319282, + "grad_norm": 13.290019035339355, + "learning_rate": 1.727596482302493e-05, + "loss": 0.347, + "step": 4760 + }, + { + "epoch": 0.17334108583472635, + "grad_norm": 1.8591586351394653, + "learning_rate": 1.7312304673304746e-05, + "loss": 0.2291, + "step": 4770 + }, + { + "epoch": 0.17370448433752453, + "grad_norm": 2.5220861434936523, + "learning_rate": 1.7348644523584565e-05, + "loss": 0.2436, + "step": 4780 + }, + { + "epoch": 0.1740678828403227, + "grad_norm": 1.8692690134048462, + "learning_rate": 1.738498437386438e-05, + "loss": 0.1782, + "step": 4790 + }, + { + "epoch": 0.17443128134312086, + "grad_norm": 12.558557510375977, + "learning_rate": 1.7421324224144197e-05, + "loss": 0.3347, + "step": 4800 + }, + { + "epoch": 0.17443128134312086, + "eval_loss": 0.4148472547531128, + "eval_runtime": 180.0999, + "eval_samples_per_second": 41.166, + "eval_steps_per_second": 5.147, + "eval_wer": 0.23564543358687168, + "step": 4800 + }, + { + "epoch": 0.17479467984591904, + "grad_norm": 6.168694972991943, + "learning_rate": 1.7457664074424017e-05, + "loss": 0.2183, + "step": 4810 + }, + { + "epoch": 0.17515807834871722, + "grad_norm": 5.153416633605957, + "learning_rate": 1.749400392470383e-05, + "loss": 0.2689, + "step": 4820 + }, + { + "epoch": 0.17552147685151537, + "grad_norm": 2.8500893115997314, + "learning_rate": 1.753034377498365e-05, + "loss": 0.2848, + "step": 4830 + }, + { + "epoch": 0.17588487535431355, + "grad_norm": 17.89117431640625, + "learning_rate": 1.7566683625263465e-05, + "loss": 0.2539, + "step": 4840 + }, + { + "epoch": 0.1762482738571117, + "grad_norm": 19.455005645751953, + "learning_rate": 1.760302347554328e-05, + "loss": 0.3166, + "step": 4850 + }, + { + "epoch": 0.17661167235990988, + "grad_norm": 1.7975777387619019, + "learning_rate": 1.76393633258231e-05, + "loss": 0.1927, + "step": 4860 + }, + { + "epoch": 0.17697507086270806, + "grad_norm": 4.6790690422058105, + "learning_rate": 1.7675703176102916e-05, + "loss": 0.2248, + "step": 4870 + }, + { + "epoch": 0.1773384693655062, + "grad_norm": 3.2644243240356445, + "learning_rate": 1.7712043026382732e-05, + "loss": 0.2239, + "step": 4880 + }, + { + "epoch": 0.1777018678683044, + "grad_norm": 1.9375410079956055, + "learning_rate": 1.7748382876662548e-05, + "loss": 0.2053, + "step": 4890 + }, + { + "epoch": 0.17806526637110254, + "grad_norm": 15.435178756713867, + "learning_rate": 1.7784722726942367e-05, + "loss": 0.2903, + "step": 4900 + }, + { + "epoch": 0.17842866487390072, + "grad_norm": 2.486330270767212, + "learning_rate": 1.7821062577222183e-05, + "loss": 0.2598, + "step": 4910 + }, + { + "epoch": 0.1787920633766989, + "grad_norm": 2.5542314052581787, + "learning_rate": 1.7857402427502e-05, + "loss": 0.2305, + "step": 4920 + }, + { + "epoch": 0.17915546187949705, + "grad_norm": 3.6416103839874268, + "learning_rate": 1.7893742277781815e-05, + "loss": 1.046, + "step": 4930 + }, + { + "epoch": 0.17951886038229523, + "grad_norm": 1.9395058155059814, + "learning_rate": 1.7930082128061634e-05, + "loss": 0.2466, + "step": 4940 + }, + { + "epoch": 0.1798822588850934, + "grad_norm": 7.664824962615967, + "learning_rate": 1.796642197834145e-05, + "loss": 0.2871, + "step": 4950 + }, + { + "epoch": 0.18024565738789156, + "grad_norm": 2.0301320552825928, + "learning_rate": 1.8002761828621266e-05, + "loss": 0.1996, + "step": 4960 + }, + { + "epoch": 0.18060905589068973, + "grad_norm": 8.371182441711426, + "learning_rate": 1.8039101678901082e-05, + "loss": 0.1947, + "step": 4970 + }, + { + "epoch": 0.18097245439348789, + "grad_norm": 2.6746129989624023, + "learning_rate": 1.80754415291809e-05, + "loss": 0.2679, + "step": 4980 + }, + { + "epoch": 0.18133585289628606, + "grad_norm": 3.448202133178711, + "learning_rate": 1.8111781379460718e-05, + "loss": 0.1859, + "step": 4990 + }, + { + "epoch": 0.18169925139908424, + "grad_norm": 28.57021141052246, + "learning_rate": 1.8148121229740534e-05, + "loss": 0.3318, + "step": 5000 + }, + { + "epoch": 0.1820626499018824, + "grad_norm": 4.731750965118408, + "learning_rate": 1.8184461080020353e-05, + "loss": 0.2354, + "step": 5010 + }, + { + "epoch": 0.18242604840468057, + "grad_norm": 1.6815394163131714, + "learning_rate": 1.822080093030017e-05, + "loss": 0.2075, + "step": 5020 + }, + { + "epoch": 0.18278944690747875, + "grad_norm": 3.868263006210327, + "learning_rate": 1.8257140780579985e-05, + "loss": 0.2345, + "step": 5030 + }, + { + "epoch": 0.1831528454102769, + "grad_norm": 1.964240550994873, + "learning_rate": 1.82934806308598e-05, + "loss": 0.222, + "step": 5040 + }, + { + "epoch": 0.18351624391307508, + "grad_norm": 11.881858825683594, + "learning_rate": 1.8329820481139617e-05, + "loss": 0.3251, + "step": 5050 + }, + { + "epoch": 0.18387964241587323, + "grad_norm": 1.8463056087493896, + "learning_rate": 1.8366160331419436e-05, + "loss": 0.2255, + "step": 5060 + }, + { + "epoch": 0.1842430409186714, + "grad_norm": 2.592672348022461, + "learning_rate": 1.8402500181699252e-05, + "loss": 0.1904, + "step": 5070 + }, + { + "epoch": 0.1846064394214696, + "grad_norm": 4.0694074630737305, + "learning_rate": 1.8438840031979068e-05, + "loss": 0.2, + "step": 5080 + }, + { + "epoch": 0.18496983792426774, + "grad_norm": 2.101837396621704, + "learning_rate": 1.8475179882258887e-05, + "loss": 0.1927, + "step": 5090 + }, + { + "epoch": 0.18533323642706592, + "grad_norm": 22.162702560424805, + "learning_rate": 1.8511519732538703e-05, + "loss": 0.3481, + "step": 5100 + }, + { + "epoch": 0.1856966349298641, + "grad_norm": 2.7928340435028076, + "learning_rate": 1.854785958281852e-05, + "loss": 0.2344, + "step": 5110 + }, + { + "epoch": 0.18606003343266225, + "grad_norm": 1.8618485927581787, + "learning_rate": 1.858419943309834e-05, + "loss": 0.2139, + "step": 5120 + }, + { + "epoch": 0.18642343193546043, + "grad_norm": 2.9611120223999023, + "learning_rate": 1.862053928337815e-05, + "loss": 0.2194, + "step": 5130 + }, + { + "epoch": 0.18678683043825858, + "grad_norm": 5.181276321411133, + "learning_rate": 1.865687913365797e-05, + "loss": 0.2596, + "step": 5140 + }, + { + "epoch": 0.18715022894105676, + "grad_norm": 10.01041030883789, + "learning_rate": 1.8693218983937787e-05, + "loss": 0.3122, + "step": 5150 + }, + { + "epoch": 0.18751362744385494, + "grad_norm": 4.952126979827881, + "learning_rate": 1.8729558834217603e-05, + "loss": 0.2183, + "step": 5160 + }, + { + "epoch": 0.1878770259466531, + "grad_norm": 2.19279146194458, + "learning_rate": 1.8765898684497422e-05, + "loss": 0.2439, + "step": 5170 + }, + { + "epoch": 0.18824042444945127, + "grad_norm": 3.5189321041107178, + "learning_rate": 1.8802238534777238e-05, + "loss": 0.2343, + "step": 5180 + }, + { + "epoch": 0.18860382295224945, + "grad_norm": 2.0936787128448486, + "learning_rate": 1.8838578385057054e-05, + "loss": 0.1831, + "step": 5190 + }, + { + "epoch": 0.1889672214550476, + "grad_norm": 12.835061073303223, + "learning_rate": 1.8874918235336873e-05, + "loss": 0.2561, + "step": 5200 + }, + { + "epoch": 0.18933061995784578, + "grad_norm": 1.6738308668136597, + "learning_rate": 1.8911258085616686e-05, + "loss": 1.0257, + "step": 5210 + }, + { + "epoch": 0.18969401846064393, + "grad_norm": 2.7661142349243164, + "learning_rate": 1.8947597935896505e-05, + "loss": 0.2398, + "step": 5220 + }, + { + "epoch": 0.1900574169634421, + "grad_norm": 4.173921585083008, + "learning_rate": 1.8983937786176324e-05, + "loss": 0.2157, + "step": 5230 + }, + { + "epoch": 0.1904208154662403, + "grad_norm": 3.7037158012390137, + "learning_rate": 1.9020277636456137e-05, + "loss": 0.2182, + "step": 5240 + }, + { + "epoch": 0.19078421396903844, + "grad_norm": 16.288227081298828, + "learning_rate": 1.9056617486735956e-05, + "loss": 0.2829, + "step": 5250 + }, + { + "epoch": 0.19114761247183662, + "grad_norm": 2.0504090785980225, + "learning_rate": 1.9092957337015772e-05, + "loss": 0.201, + "step": 5260 + }, + { + "epoch": 0.19151101097463477, + "grad_norm": 1.2266415357589722, + "learning_rate": 1.9129297187295588e-05, + "loss": 0.2072, + "step": 5270 + }, + { + "epoch": 0.19187440947743295, + "grad_norm": 4.910546779632568, + "learning_rate": 1.9165637037575408e-05, + "loss": 0.1824, + "step": 5280 + }, + { + "epoch": 0.19223780798023113, + "grad_norm": 3.093318223953247, + "learning_rate": 1.9201976887855224e-05, + "loss": 0.2471, + "step": 5290 + }, + { + "epoch": 0.19260120648302928, + "grad_norm": 6.74167013168335, + "learning_rate": 1.923831673813504e-05, + "loss": 0.2912, + "step": 5300 + }, + { + "epoch": 0.19296460498582746, + "grad_norm": 2.0540058612823486, + "learning_rate": 1.927465658841486e-05, + "loss": 0.2599, + "step": 5310 + }, + { + "epoch": 0.19332800348862564, + "grad_norm": 2.407750129699707, + "learning_rate": 1.931099643869467e-05, + "loss": 0.2478, + "step": 5320 + }, + { + "epoch": 0.1936914019914238, + "grad_norm": 5.479567527770996, + "learning_rate": 1.934733628897449e-05, + "loss": 0.5936, + "step": 5330 + }, + { + "epoch": 0.19405480049422197, + "grad_norm": 1.912705659866333, + "learning_rate": 1.9383676139254307e-05, + "loss": 0.215, + "step": 5340 + }, + { + "epoch": 0.19441819899702012, + "grad_norm": 38.24689865112305, + "learning_rate": 1.9420015989534123e-05, + "loss": 0.286, + "step": 5350 + }, + { + "epoch": 0.1947815974998183, + "grad_norm": 3.4196550846099854, + "learning_rate": 1.9456355839813942e-05, + "loss": 0.4764, + "step": 5360 + }, + { + "epoch": 0.19514499600261648, + "grad_norm": 1.705702781677246, + "learning_rate": 1.9492695690093758e-05, + "loss": 0.195, + "step": 5370 + }, + { + "epoch": 0.19550839450541463, + "grad_norm": 2.7188572883605957, + "learning_rate": 1.9529035540373574e-05, + "loss": 0.2318, + "step": 5380 + }, + { + "epoch": 0.1958717930082128, + "grad_norm": 5.217918872833252, + "learning_rate": 1.9565375390653393e-05, + "loss": 0.2288, + "step": 5390 + }, + { + "epoch": 0.196235191511011, + "grad_norm": 7.094780921936035, + "learning_rate": 1.960171524093321e-05, + "loss": 0.3607, + "step": 5400 + }, + { + "epoch": 0.196235191511011, + "eval_loss": 0.3953820765018463, + "eval_runtime": 180.5214, + "eval_samples_per_second": 41.07, + "eval_steps_per_second": 5.135, + "eval_wer": 0.23525514186650207, + "step": 5400 + }, + { + "epoch": 0.19659859001380914, + "grad_norm": 2.379298448562622, + "learning_rate": 1.9638055091213025e-05, + "loss": 0.1981, + "step": 5410 + }, + { + "epoch": 0.19696198851660732, + "grad_norm": 1.2755372524261475, + "learning_rate": 1.967439494149284e-05, + "loss": 0.3185, + "step": 5420 + }, + { + "epoch": 0.19732538701940547, + "grad_norm": 2.6385338306427, + "learning_rate": 1.9710734791772657e-05, + "loss": 0.2231, + "step": 5430 + }, + { + "epoch": 0.19768878552220365, + "grad_norm": 4.030337810516357, + "learning_rate": 1.9747074642052477e-05, + "loss": 0.2417, + "step": 5440 + }, + { + "epoch": 0.19805218402500183, + "grad_norm": 10.988908767700195, + "learning_rate": 1.9783414492332293e-05, + "loss": 0.3163, + "step": 5450 + }, + { + "epoch": 0.19841558252779998, + "grad_norm": 2.8273231983184814, + "learning_rate": 1.981975434261211e-05, + "loss": 0.2062, + "step": 5460 + }, + { + "epoch": 0.19877898103059816, + "grad_norm": 1.880952000617981, + "learning_rate": 1.9856094192891928e-05, + "loss": 0.2103, + "step": 5470 + }, + { + "epoch": 0.19914237953339634, + "grad_norm": 12.882647514343262, + "learning_rate": 1.9892434043171744e-05, + "loss": 0.2513, + "step": 5480 + }, + { + "epoch": 0.1995057780361945, + "grad_norm": 2.8202428817749023, + "learning_rate": 1.992877389345156e-05, + "loss": 0.2002, + "step": 5490 + }, + { + "epoch": 0.19986917653899267, + "grad_norm": 11.30123519897461, + "learning_rate": 1.9965113743731376e-05, + "loss": 0.3399, + "step": 5500 + }, + { + "epoch": 0.20023257504179082, + "grad_norm": 3.016954183578491, + "learning_rate": 2.0001453594011195e-05, + "loss": 0.2016, + "step": 5510 + }, + { + "epoch": 0.200595973544589, + "grad_norm": 1.3506131172180176, + "learning_rate": 2.003779344429101e-05, + "loss": 0.6008, + "step": 5520 + }, + { + "epoch": 0.20095937204738717, + "grad_norm": 3.711284637451172, + "learning_rate": 2.0074133294570827e-05, + "loss": 0.2297, + "step": 5530 + }, + { + "epoch": 0.20132277055018533, + "grad_norm": 2.8310322761535645, + "learning_rate": 2.0110473144850643e-05, + "loss": 0.19, + "step": 5540 + }, + { + "epoch": 0.2016861690529835, + "grad_norm": 14.37038516998291, + "learning_rate": 2.0146812995130462e-05, + "loss": 0.3418, + "step": 5550 + }, + { + "epoch": 0.20204956755578168, + "grad_norm": 2.037245988845825, + "learning_rate": 2.0183152845410278e-05, + "loss": 0.2054, + "step": 5560 + }, + { + "epoch": 0.20241296605857984, + "grad_norm": 2.47495698928833, + "learning_rate": 2.0219492695690094e-05, + "loss": 0.2102, + "step": 5570 + }, + { + "epoch": 0.20277636456137801, + "grad_norm": 5.948564529418945, + "learning_rate": 2.025583254596991e-05, + "loss": 0.2299, + "step": 5580 + }, + { + "epoch": 0.20313976306417617, + "grad_norm": 2.010765552520752, + "learning_rate": 2.029217239624973e-05, + "loss": 0.2214, + "step": 5590 + }, + { + "epoch": 0.20350316156697434, + "grad_norm": 109.07927703857422, + "learning_rate": 2.0328512246529546e-05, + "loss": 0.327, + "step": 5600 + }, + { + "epoch": 0.20386656006977252, + "grad_norm": 2.708141565322876, + "learning_rate": 2.036485209680936e-05, + "loss": 0.2128, + "step": 5610 + }, + { + "epoch": 0.20422995857257067, + "grad_norm": 4.145051002502441, + "learning_rate": 2.040119194708918e-05, + "loss": 1.5499, + "step": 5620 + }, + { + "epoch": 0.20459335707536885, + "grad_norm": 5.204433917999268, + "learning_rate": 2.0437531797368993e-05, + "loss": 0.2238, + "step": 5630 + }, + { + "epoch": 0.204956755578167, + "grad_norm": 3.625671625137329, + "learning_rate": 2.0473871647648813e-05, + "loss": 0.2009, + "step": 5640 + }, + { + "epoch": 0.20532015408096518, + "grad_norm": 7.134413719177246, + "learning_rate": 2.051021149792863e-05, + "loss": 0.3236, + "step": 5650 + }, + { + "epoch": 0.20568355258376336, + "grad_norm": 3.090585708618164, + "learning_rate": 2.0546551348208445e-05, + "loss": 0.2245, + "step": 5660 + }, + { + "epoch": 0.20604695108656151, + "grad_norm": 1.5290725231170654, + "learning_rate": 2.0582891198488264e-05, + "loss": 0.9725, + "step": 5670 + }, + { + "epoch": 0.2064103495893597, + "grad_norm": 12.433088302612305, + "learning_rate": 2.061923104876808e-05, + "loss": 0.2755, + "step": 5680 + }, + { + "epoch": 0.20677374809215787, + "grad_norm": 4.399518013000488, + "learning_rate": 2.0655570899047896e-05, + "loss": 0.2136, + "step": 5690 + }, + { + "epoch": 0.20713714659495602, + "grad_norm": 12.662751197814941, + "learning_rate": 2.0691910749327715e-05, + "loss": 0.3022, + "step": 5700 + }, + { + "epoch": 0.2075005450977542, + "grad_norm": 1.8056265115737915, + "learning_rate": 2.0728250599607528e-05, + "loss": 0.3538, + "step": 5710 + }, + { + "epoch": 0.20786394360055235, + "grad_norm": 1.3133045434951782, + "learning_rate": 2.0764590449887347e-05, + "loss": 0.1829, + "step": 5720 + }, + { + "epoch": 0.20822734210335053, + "grad_norm": 6.10534143447876, + "learning_rate": 2.0800930300167167e-05, + "loss": 0.2819, + "step": 5730 + }, + { + "epoch": 0.2085907406061487, + "grad_norm": 4.327618598937988, + "learning_rate": 2.083727015044698e-05, + "loss": 0.2029, + "step": 5740 + }, + { + "epoch": 0.20895413910894686, + "grad_norm": 6.878536224365234, + "learning_rate": 2.08736100007268e-05, + "loss": 0.3301, + "step": 5750 + }, + { + "epoch": 0.20931753761174504, + "grad_norm": 2.8301913738250732, + "learning_rate": 2.0909949851006614e-05, + "loss": 0.2144, + "step": 5760 + }, + { + "epoch": 0.20968093611454322, + "grad_norm": 2.248054265975952, + "learning_rate": 2.094628970128643e-05, + "loss": 0.2046, + "step": 5770 + }, + { + "epoch": 0.21004433461734137, + "grad_norm": 4.619300842285156, + "learning_rate": 2.098262955156625e-05, + "loss": 0.2487, + "step": 5780 + }, + { + "epoch": 0.21040773312013955, + "grad_norm": 2.6446404457092285, + "learning_rate": 2.1018969401846066e-05, + "loss": 0.2222, + "step": 5790 + }, + { + "epoch": 0.2107711316229377, + "grad_norm": 7.827177047729492, + "learning_rate": 2.1055309252125882e-05, + "loss": 0.2684, + "step": 5800 + }, + { + "epoch": 0.21113453012573588, + "grad_norm": 5.37054967880249, + "learning_rate": 2.10916491024057e-05, + "loss": 0.216, + "step": 5810 + }, + { + "epoch": 0.21149792862853406, + "grad_norm": 1.5430680513381958, + "learning_rate": 2.1127988952685514e-05, + "loss": 0.1723, + "step": 5820 + }, + { + "epoch": 0.2118613271313322, + "grad_norm": 4.355040550231934, + "learning_rate": 2.1164328802965333e-05, + "loss": 0.3078, + "step": 5830 + }, + { + "epoch": 0.2122247256341304, + "grad_norm": 2.70613169670105, + "learning_rate": 2.1200668653245152e-05, + "loss": 0.1857, + "step": 5840 + }, + { + "epoch": 0.21258812413692857, + "grad_norm": 17.876861572265625, + "learning_rate": 2.1237008503524965e-05, + "loss": 0.335, + "step": 5850 + }, + { + "epoch": 0.21295152263972672, + "grad_norm": 2.048499822616577, + "learning_rate": 2.1273348353804784e-05, + "loss": 0.2588, + "step": 5860 + }, + { + "epoch": 0.2133149211425249, + "grad_norm": 2.2033607959747314, + "learning_rate": 2.13096882040846e-05, + "loss": 0.1973, + "step": 5870 + }, + { + "epoch": 0.21367831964532305, + "grad_norm": 5.563814640045166, + "learning_rate": 2.1346028054364416e-05, + "loss": 0.2632, + "step": 5880 + }, + { + "epoch": 0.21404171814812123, + "grad_norm": 1.4629203081130981, + "learning_rate": 2.1382367904644236e-05, + "loss": 0.1714, + "step": 5890 + }, + { + "epoch": 0.2144051166509194, + "grad_norm": 9.641836166381836, + "learning_rate": 2.141870775492405e-05, + "loss": 0.3329, + "step": 5900 + }, + { + "epoch": 0.21476851515371756, + "grad_norm": 3.0128610134124756, + "learning_rate": 2.1455047605203867e-05, + "loss": 0.205, + "step": 5910 + }, + { + "epoch": 0.21513191365651574, + "grad_norm": 6.38659143447876, + "learning_rate": 2.1491387455483687e-05, + "loss": 0.2066, + "step": 5920 + }, + { + "epoch": 0.2154953121593139, + "grad_norm": 3.397566080093384, + "learning_rate": 2.15277273057635e-05, + "loss": 0.2682, + "step": 5930 + }, + { + "epoch": 0.21585871066211207, + "grad_norm": 1.8110759258270264, + "learning_rate": 2.156406715604332e-05, + "loss": 0.2357, + "step": 5940 + }, + { + "epoch": 0.21622210916491025, + "grad_norm": 12.391556739807129, + "learning_rate": 2.1600407006323135e-05, + "loss": 0.3043, + "step": 5950 + }, + { + "epoch": 0.2165855076677084, + "grad_norm": 1.8203914165496826, + "learning_rate": 2.163674685660295e-05, + "loss": 0.2979, + "step": 5960 + }, + { + "epoch": 0.21694890617050658, + "grad_norm": 3.362252950668335, + "learning_rate": 2.167308670688277e-05, + "loss": 0.1667, + "step": 5970 + }, + { + "epoch": 0.21731230467330476, + "grad_norm": 4.1468000411987305, + "learning_rate": 2.1709426557162586e-05, + "loss": 0.3419, + "step": 5980 + }, + { + "epoch": 0.2176757031761029, + "grad_norm": 2.479288339614868, + "learning_rate": 2.1745766407442402e-05, + "loss": 0.1938, + "step": 5990 + }, + { + "epoch": 0.2180391016789011, + "grad_norm": 26.185468673706055, + "learning_rate": 2.1782106257722218e-05, + "loss": 0.2818, + "step": 6000 + }, + { + "epoch": 0.2180391016789011, + "eval_loss": 0.4106527864933014, + "eval_runtime": 179.9044, + "eval_samples_per_second": 41.211, + "eval_steps_per_second": 5.153, + "eval_wer": 0.2305625646704304, + "step": 6000 + }, + { + "epoch": 0.21840250018169924, + "grad_norm": 2.2452592849731445, + "learning_rate": 2.1818446108002037e-05, + "loss": 0.2208, + "step": 6010 + }, + { + "epoch": 0.21876589868449742, + "grad_norm": 2.273920774459839, + "learning_rate": 2.1854785958281853e-05, + "loss": 0.2268, + "step": 6020 + }, + { + "epoch": 0.2191292971872956, + "grad_norm": 1.9621226787567139, + "learning_rate": 2.189112580856167e-05, + "loss": 0.1965, + "step": 6030 + }, + { + "epoch": 0.21949269569009375, + "grad_norm": 2.866110324859619, + "learning_rate": 2.1927465658841485e-05, + "loss": 0.223, + "step": 6040 + }, + { + "epoch": 0.21985609419289193, + "grad_norm": 15.169930458068848, + "learning_rate": 2.1963805509121305e-05, + "loss": 0.254, + "step": 6050 + }, + { + "epoch": 0.2202194926956901, + "grad_norm": 2.174626350402832, + "learning_rate": 2.200014535940112e-05, + "loss": 0.2056, + "step": 6060 + }, + { + "epoch": 0.22058289119848826, + "grad_norm": 1.9627354145050049, + "learning_rate": 2.2036485209680936e-05, + "loss": 0.2211, + "step": 6070 + }, + { + "epoch": 0.22094628970128644, + "grad_norm": 5.444493770599365, + "learning_rate": 2.2072825059960752e-05, + "loss": 0.2819, + "step": 6080 + }, + { + "epoch": 0.2213096882040846, + "grad_norm": 2.5131990909576416, + "learning_rate": 2.2109164910240572e-05, + "loss": 0.2262, + "step": 6090 + }, + { + "epoch": 0.22167308670688277, + "grad_norm": 15.716779708862305, + "learning_rate": 2.2145504760520388e-05, + "loss": 0.2833, + "step": 6100 + }, + { + "epoch": 0.22203648520968094, + "grad_norm": 1.7514111995697021, + "learning_rate": 2.2181844610800204e-05, + "loss": 0.2238, + "step": 6110 + }, + { + "epoch": 0.2223998837124791, + "grad_norm": 1.8236886262893677, + "learning_rate": 2.2218184461080023e-05, + "loss": 0.1872, + "step": 6120 + }, + { + "epoch": 0.22276328221527728, + "grad_norm": 4.081092834472656, + "learning_rate": 2.225452431135984e-05, + "loss": 0.4926, + "step": 6130 + }, + { + "epoch": 0.22312668071807545, + "grad_norm": 3.3254685401916504, + "learning_rate": 2.2290864161639655e-05, + "loss": 0.2179, + "step": 6140 + }, + { + "epoch": 0.2234900792208736, + "grad_norm": 9.953665733337402, + "learning_rate": 2.232720401191947e-05, + "loss": 0.3221, + "step": 6150 + }, + { + "epoch": 0.22385347772367178, + "grad_norm": 3.531538724899292, + "learning_rate": 2.2363543862199287e-05, + "loss": 0.3639, + "step": 6160 + }, + { + "epoch": 0.22421687622646994, + "grad_norm": 1.6166915893554688, + "learning_rate": 2.2399883712479106e-05, + "loss": 0.1906, + "step": 6170 + }, + { + "epoch": 0.22458027472926811, + "grad_norm": 3.0561792850494385, + "learning_rate": 2.2436223562758922e-05, + "loss": 0.222, + "step": 6180 + }, + { + "epoch": 0.2249436732320663, + "grad_norm": 7.607283115386963, + "learning_rate": 2.2472563413038738e-05, + "loss": 0.1769, + "step": 6190 + }, + { + "epoch": 0.22530707173486444, + "grad_norm": 38.86745834350586, + "learning_rate": 2.2508903263318557e-05, + "loss": 0.3523, + "step": 6200 + }, + { + "epoch": 0.22567047023766262, + "grad_norm": 1.2490432262420654, + "learning_rate": 2.2545243113598373e-05, + "loss": 0.2241, + "step": 6210 + }, + { + "epoch": 0.2260338687404608, + "grad_norm": 3.8632936477661133, + "learning_rate": 2.258158296387819e-05, + "loss": 0.2761, + "step": 6220 + }, + { + "epoch": 0.22639726724325895, + "grad_norm": 6.057976722717285, + "learning_rate": 2.261792281415801e-05, + "loss": 0.2534, + "step": 6230 + }, + { + "epoch": 0.22676066574605713, + "grad_norm": 5.2983551025390625, + "learning_rate": 2.265426266443782e-05, + "loss": 0.1972, + "step": 6240 + }, + { + "epoch": 0.22712406424885528, + "grad_norm": 7.395950794219971, + "learning_rate": 2.269060251471764e-05, + "loss": 0.3446, + "step": 6250 + }, + { + "epoch": 0.22748746275165346, + "grad_norm": 2.7409260272979736, + "learning_rate": 2.2726942364997457e-05, + "loss": 0.1894, + "step": 6260 + }, + { + "epoch": 0.22785086125445164, + "grad_norm": 1.7545270919799805, + "learning_rate": 2.2763282215277273e-05, + "loss": 0.2376, + "step": 6270 + }, + { + "epoch": 0.2282142597572498, + "grad_norm": 112.10614013671875, + "learning_rate": 2.2799622065557092e-05, + "loss": 2.0322, + "step": 6280 + }, + { + "epoch": 0.22857765826004797, + "grad_norm": 3.6547396183013916, + "learning_rate": 2.2835961915836908e-05, + "loss": 0.2942, + "step": 6290 + }, + { + "epoch": 0.22894105676284612, + "grad_norm": 50.726261138916016, + "learning_rate": 2.2872301766116724e-05, + "loss": 0.3279, + "step": 6300 + }, + { + "epoch": 0.2293044552656443, + "grad_norm": 1.2374241352081299, + "learning_rate": 2.2908641616396543e-05, + "loss": 0.1912, + "step": 6310 + }, + { + "epoch": 0.22966785376844248, + "grad_norm": 1.6278152465820312, + "learning_rate": 2.2944981466676356e-05, + "loss": 0.1913, + "step": 6320 + }, + { + "epoch": 0.23003125227124063, + "grad_norm": 7.58544397354126, + "learning_rate": 2.2981321316956175e-05, + "loss": 0.2393, + "step": 6330 + }, + { + "epoch": 0.2303946507740388, + "grad_norm": 1.7094483375549316, + "learning_rate": 2.3017661167235995e-05, + "loss": 0.2333, + "step": 6340 + }, + { + "epoch": 0.230758049276837, + "grad_norm": 24.214885711669922, + "learning_rate": 2.3054001017515807e-05, + "loss": 0.3019, + "step": 6350 + }, + { + "epoch": 0.23112144777963514, + "grad_norm": 1.962106704711914, + "learning_rate": 2.3090340867795626e-05, + "loss": 0.8948, + "step": 6360 + }, + { + "epoch": 0.23148484628243332, + "grad_norm": 1.3703123331069946, + "learning_rate": 2.3126680718075442e-05, + "loss": 0.1936, + "step": 6370 + }, + { + "epoch": 0.23184824478523147, + "grad_norm": 7.507201194763184, + "learning_rate": 2.316302056835526e-05, + "loss": 0.2185, + "step": 6380 + }, + { + "epoch": 0.23221164328802965, + "grad_norm": 2.6310977935791016, + "learning_rate": 2.3199360418635078e-05, + "loss": 0.1961, + "step": 6390 + }, + { + "epoch": 0.23257504179082783, + "grad_norm": 4.186092376708984, + "learning_rate": 2.3235700268914894e-05, + "loss": 0.2734, + "step": 6400 + }, + { + "epoch": 0.23293844029362598, + "grad_norm": 1.817269206047058, + "learning_rate": 2.327204011919471e-05, + "loss": 0.1966, + "step": 6410 + }, + { + "epoch": 0.23330183879642416, + "grad_norm": 1.9503989219665527, + "learning_rate": 2.330837996947453e-05, + "loss": 2.7438, + "step": 6420 + }, + { + "epoch": 0.23366523729922234, + "grad_norm": 3.1107656955718994, + "learning_rate": 2.334471981975434e-05, + "loss": 0.2534, + "step": 6430 + }, + { + "epoch": 0.2340286358020205, + "grad_norm": 5.268273830413818, + "learning_rate": 2.338105967003416e-05, + "loss": 0.1963, + "step": 6440 + }, + { + "epoch": 0.23439203430481867, + "grad_norm": 9.586852073669434, + "learning_rate": 2.3417399520313977e-05, + "loss": 0.2342, + "step": 6450 + }, + { + "epoch": 0.23475543280761682, + "grad_norm": 3.0218632221221924, + "learning_rate": 2.3453739370593793e-05, + "loss": 0.231, + "step": 6460 + }, + { + "epoch": 0.235118831310415, + "grad_norm": 1.9708057641983032, + "learning_rate": 2.3490079220873612e-05, + "loss": 0.2156, + "step": 6470 + }, + { + "epoch": 0.23548222981321318, + "grad_norm": 3.6212944984436035, + "learning_rate": 2.3526419071153428e-05, + "loss": 0.2172, + "step": 6480 + }, + { + "epoch": 0.23584562831601133, + "grad_norm": 2.5205702781677246, + "learning_rate": 2.3562758921433244e-05, + "loss": 0.4643, + "step": 6490 + }, + { + "epoch": 0.2362090268188095, + "grad_norm": 4.1570305824279785, + "learning_rate": 2.3599098771713063e-05, + "loss": 0.2722, + "step": 6500 + }, + { + "epoch": 0.2365724253216077, + "grad_norm": 1.8376798629760742, + "learning_rate": 2.363543862199288e-05, + "loss": 0.2027, + "step": 6510 + }, + { + "epoch": 0.23693582382440584, + "grad_norm": 2.0464930534362793, + "learning_rate": 2.3671778472272695e-05, + "loss": 0.183, + "step": 6520 + }, + { + "epoch": 0.23729922232720402, + "grad_norm": 4.8776469230651855, + "learning_rate": 2.370811832255251e-05, + "loss": 0.2169, + "step": 6530 + }, + { + "epoch": 0.23766262083000217, + "grad_norm": 1.5764952898025513, + "learning_rate": 2.3744458172832327e-05, + "loss": 0.1917, + "step": 6540 + }, + { + "epoch": 0.23802601933280035, + "grad_norm": 16.132232666015625, + "learning_rate": 2.3780798023112147e-05, + "loss": 0.2732, + "step": 6550 + }, + { + "epoch": 0.23838941783559853, + "grad_norm": 8.105748176574707, + "learning_rate": 2.3817137873391963e-05, + "loss": 0.2055, + "step": 6560 + }, + { + "epoch": 0.23875281633839668, + "grad_norm": 2.087362051010132, + "learning_rate": 2.385347772367178e-05, + "loss": 0.186, + "step": 6570 + }, + { + "epoch": 0.23911621484119486, + "grad_norm": 2.8280205726623535, + "learning_rate": 2.3889817573951598e-05, + "loss": 0.201, + "step": 6580 + }, + { + "epoch": 0.23947961334399304, + "grad_norm": 1.2525794506072998, + "learning_rate": 2.3926157424231414e-05, + "loss": 0.1893, + "step": 6590 + }, + { + "epoch": 0.2398430118467912, + "grad_norm": 23.419832229614258, + "learning_rate": 2.396249727451123e-05, + "loss": 0.2554, + "step": 6600 + }, + { + "epoch": 0.2398430118467912, + "eval_loss": 0.4065987765789032, + "eval_runtime": 179.638, + "eval_samples_per_second": 41.272, + "eval_steps_per_second": 5.16, + "eval_wer": 0.24529380797647357, + "step": 6600 + }, + { + "epoch": 0.24020641034958937, + "grad_norm": 1.3757339715957642, + "learning_rate": 2.3998837124791046e-05, + "loss": 0.1962, + "step": 6610 + }, + { + "epoch": 0.24056980885238752, + "grad_norm": 4.00860071182251, + "learning_rate": 2.4035176975070865e-05, + "loss": 0.1848, + "step": 6620 + }, + { + "epoch": 0.2409332073551857, + "grad_norm": 5.544015407562256, + "learning_rate": 2.407151682535068e-05, + "loss": 0.245, + "step": 6630 + }, + { + "epoch": 0.24129660585798388, + "grad_norm": 1.0618844032287598, + "learning_rate": 2.4107856675630497e-05, + "loss": 0.191, + "step": 6640 + }, + { + "epoch": 0.24166000436078203, + "grad_norm": 125.15505981445312, + "learning_rate": 2.4144196525910313e-05, + "loss": 0.3055, + "step": 6650 + }, + { + "epoch": 0.2420234028635802, + "grad_norm": 5.015167713165283, + "learning_rate": 2.418053637619013e-05, + "loss": 0.2701, + "step": 6660 + }, + { + "epoch": 0.24238680136637836, + "grad_norm": 3.944514274597168, + "learning_rate": 2.421687622646995e-05, + "loss": 0.2107, + "step": 6670 + }, + { + "epoch": 0.24275019986917654, + "grad_norm": 3.1539418697357178, + "learning_rate": 2.4253216076749764e-05, + "loss": 0.232, + "step": 6680 + }, + { + "epoch": 0.24311359837197472, + "grad_norm": 2.980459213256836, + "learning_rate": 2.428955592702958e-05, + "loss": 0.2391, + "step": 6690 + }, + { + "epoch": 0.24347699687477287, + "grad_norm": 35.02157211303711, + "learning_rate": 2.43258957773094e-05, + "loss": 0.3172, + "step": 6700 + }, + { + "epoch": 0.24384039537757105, + "grad_norm": 1.606570839881897, + "learning_rate": 2.4362235627589216e-05, + "loss": 1.5707, + "step": 6710 + }, + { + "epoch": 0.24420379388036922, + "grad_norm": 3.940394401550293, + "learning_rate": 2.439857547786903e-05, + "loss": 0.1969, + "step": 6720 + }, + { + "epoch": 0.24456719238316738, + "grad_norm": 3.8990156650543213, + "learning_rate": 2.443491532814885e-05, + "loss": 0.2475, + "step": 6730 + }, + { + "epoch": 0.24493059088596555, + "grad_norm": 2.523500442504883, + "learning_rate": 2.4471255178428664e-05, + "loss": 0.194, + "step": 6740 + }, + { + "epoch": 0.2452939893887637, + "grad_norm": 4.920846939086914, + "learning_rate": 2.4507595028708483e-05, + "loss": 0.2417, + "step": 6750 + }, + { + "epoch": 0.24565738789156188, + "grad_norm": 2.2269723415374756, + "learning_rate": 2.4543934878988302e-05, + "loss": 0.2148, + "step": 6760 + }, + { + "epoch": 0.24602078639436006, + "grad_norm": 1.669722557067871, + "learning_rate": 2.4580274729268115e-05, + "loss": 0.1979, + "step": 6770 + }, + { + "epoch": 0.24638418489715821, + "grad_norm": 4.581501007080078, + "learning_rate": 2.4616614579547934e-05, + "loss": 0.2412, + "step": 6780 + }, + { + "epoch": 0.2467475833999564, + "grad_norm": 2.6605944633483887, + "learning_rate": 2.465295442982775e-05, + "loss": 0.1992, + "step": 6790 + }, + { + "epoch": 0.24711098190275457, + "grad_norm": 7.089646816253662, + "learning_rate": 2.4689294280107566e-05, + "loss": 0.2789, + "step": 6800 + }, + { + "epoch": 0.24747438040555272, + "grad_norm": 1.9901385307312012, + "learning_rate": 2.4725634130387385e-05, + "loss": 1.907, + "step": 6810 + }, + { + "epoch": 0.2478377789083509, + "grad_norm": 2.5120224952697754, + "learning_rate": 2.4761973980667198e-05, + "loss": 0.1908, + "step": 6820 + }, + { + "epoch": 0.24820117741114905, + "grad_norm": 1.553806185722351, + "learning_rate": 2.4794679845919035e-05, + "loss": 1.6707, + "step": 6830 + }, + { + "epoch": 0.24856457591394723, + "grad_norm": 2.130095958709717, + "learning_rate": 2.4831019696198855e-05, + "loss": 0.222, + "step": 6840 + }, + { + "epoch": 0.2489279744167454, + "grad_norm": 15.832701683044434, + "learning_rate": 2.486735954647867e-05, + "loss": 0.4634, + "step": 6850 + }, + { + "epoch": 0.24929137291954356, + "grad_norm": 1.87086820602417, + "learning_rate": 2.4903699396758487e-05, + "loss": 0.1887, + "step": 6860 + }, + { + "epoch": 0.24965477142234174, + "grad_norm": 2.32084584236145, + "learning_rate": 2.4940039247038303e-05, + "loss": 0.1881, + "step": 6870 + }, + { + "epoch": 0.2500181699251399, + "grad_norm": 3.3228461742401123, + "learning_rate": 2.497637909731812e-05, + "loss": 0.264, + "step": 6880 + }, + { + "epoch": 0.2503815684279381, + "grad_norm": 1.8676607608795166, + "learning_rate": 2.5012718947597935e-05, + "loss": 0.2102, + "step": 6890 + }, + { + "epoch": 0.2507449669307362, + "grad_norm": 17.540319442749023, + "learning_rate": 2.5049058797877757e-05, + "loss": 0.2567, + "step": 6900 + }, + { + "epoch": 0.25110836543353443, + "grad_norm": 1.6276856660842896, + "learning_rate": 2.508539864815757e-05, + "loss": 0.1917, + "step": 6910 + }, + { + "epoch": 0.2514717639363326, + "grad_norm": 2.347691059112549, + "learning_rate": 2.5121738498437386e-05, + "loss": 0.1998, + "step": 6920 + }, + { + "epoch": 0.25183516243913073, + "grad_norm": 3.5337650775909424, + "learning_rate": 2.5158078348717205e-05, + "loss": 0.2418, + "step": 6930 + }, + { + "epoch": 0.25219856094192894, + "grad_norm": 3.7415404319763184, + "learning_rate": 2.519441819899702e-05, + "loss": 0.2074, + "step": 6940 + }, + { + "epoch": 0.2525619594447271, + "grad_norm": 16.603042602539062, + "learning_rate": 2.5230758049276837e-05, + "loss": 0.3104, + "step": 6950 + }, + { + "epoch": 0.25292535794752524, + "grad_norm": 1.4864579439163208, + "learning_rate": 2.5267097899556656e-05, + "loss": 0.1771, + "step": 6960 + }, + { + "epoch": 0.25328875645032345, + "grad_norm": 1.7935876846313477, + "learning_rate": 2.5303437749836472e-05, + "loss": 0.1984, + "step": 6970 + }, + { + "epoch": 0.2536521549531216, + "grad_norm": 3.187351942062378, + "learning_rate": 2.533977760011629e-05, + "loss": 0.1828, + "step": 6980 + }, + { + "epoch": 0.25401555345591975, + "grad_norm": 1.7930549383163452, + "learning_rate": 2.5376117450396104e-05, + "loss": 0.2132, + "step": 6990 + }, + { + "epoch": 0.2543789519587179, + "grad_norm": 4.86196231842041, + "learning_rate": 2.5412457300675924e-05, + "loss": 0.2426, + "step": 7000 + }, + { + "epoch": 0.2547423504615161, + "grad_norm": 2.784335136413574, + "learning_rate": 2.544879715095574e-05, + "loss": 1.6557, + "step": 7010 + }, + { + "epoch": 0.25510574896431426, + "grad_norm": 1.460509181022644, + "learning_rate": 2.5485137001235552e-05, + "loss": 0.1812, + "step": 7020 + }, + { + "epoch": 0.2554691474671124, + "grad_norm": 2.5204946994781494, + "learning_rate": 2.5521476851515375e-05, + "loss": 0.3731, + "step": 7030 + }, + { + "epoch": 0.2558325459699106, + "grad_norm": 1.6122281551361084, + "learning_rate": 2.555781670179519e-05, + "loss": 0.2256, + "step": 7040 + }, + { + "epoch": 0.25619594447270877, + "grad_norm": 8.13974666595459, + "learning_rate": 2.5594156552075004e-05, + "loss": 0.2756, + "step": 7050 + }, + { + "epoch": 0.2565593429755069, + "grad_norm": 2.1560494899749756, + "learning_rate": 2.5630496402354826e-05, + "loss": 0.1869, + "step": 7060 + }, + { + "epoch": 0.25692274147830513, + "grad_norm": 2.938570737838745, + "learning_rate": 2.5666836252634642e-05, + "loss": 0.187, + "step": 7070 + }, + { + "epoch": 0.2572861399811033, + "grad_norm": 1.6697754859924316, + "learning_rate": 2.5703176102914455e-05, + "loss": 0.1841, + "step": 7080 + }, + { + "epoch": 0.25764953848390143, + "grad_norm": 2.500377655029297, + "learning_rate": 2.5739515953194278e-05, + "loss": 0.4097, + "step": 7090 + }, + { + "epoch": 0.25801293698669964, + "grad_norm": 6.614553928375244, + "learning_rate": 2.577585580347409e-05, + "loss": 0.2779, + "step": 7100 + }, + { + "epoch": 0.2583763354894978, + "grad_norm": 2.1538803577423096, + "learning_rate": 2.5812195653753906e-05, + "loss": 0.2035, + "step": 7110 + }, + { + "epoch": 0.25873973399229594, + "grad_norm": 2.64719820022583, + "learning_rate": 2.584853550403373e-05, + "loss": 0.1815, + "step": 7120 + }, + { + "epoch": 0.25910313249509415, + "grad_norm": 4.064308166503906, + "learning_rate": 2.588487535431354e-05, + "loss": 0.2115, + "step": 7130 + }, + { + "epoch": 0.2594665309978923, + "grad_norm": 4.535513877868652, + "learning_rate": 2.5921215204593357e-05, + "loss": 0.1733, + "step": 7140 + }, + { + "epoch": 0.25982992950069045, + "grad_norm": 14.761083602905273, + "learning_rate": 2.5957555054873173e-05, + "loss": 0.3061, + "step": 7150 + }, + { + "epoch": 0.2601933280034886, + "grad_norm": 2.902010202407837, + "learning_rate": 2.5993894905152993e-05, + "loss": 0.2539, + "step": 7160 + }, + { + "epoch": 0.2605567265062868, + "grad_norm": 2.6499462127685547, + "learning_rate": 2.603023475543281e-05, + "loss": 0.209, + "step": 7170 + }, + { + "epoch": 0.26092012500908496, + "grad_norm": 2.0298879146575928, + "learning_rate": 2.6066574605712625e-05, + "loss": 0.1966, + "step": 7180 + }, + { + "epoch": 0.2612835235118831, + "grad_norm": 5.285839080810547, + "learning_rate": 2.6102914455992444e-05, + "loss": 0.2416, + "step": 7190 + }, + { + "epoch": 0.2616469220146813, + "grad_norm": 14.89932918548584, + "learning_rate": 2.613925430627226e-05, + "loss": 0.2649, + "step": 7200 + }, + { + "epoch": 0.2616469220146813, + "eval_loss": 0.43822312355041504, + "eval_runtime": 180.398, + "eval_samples_per_second": 41.098, + "eval_steps_per_second": 5.139, + "eval_wer": 0.23023580881151634, + "step": 7200 + }, + { + "epoch": 0.26201032051747947, + "grad_norm": 2.9772818088531494, + "learning_rate": 2.6175594156552076e-05, + "loss": 0.2158, + "step": 7210 + }, + { + "epoch": 0.2623737190202776, + "grad_norm": 1.4703949689865112, + "learning_rate": 2.6211934006831895e-05, + "loss": 0.1925, + "step": 7220 + }, + { + "epoch": 0.2627371175230758, + "grad_norm": 2.6034176349639893, + "learning_rate": 2.624827385711171e-05, + "loss": 0.2065, + "step": 7230 + }, + { + "epoch": 0.263100516025874, + "grad_norm": 2.8392562866210938, + "learning_rate": 2.6284613707391527e-05, + "loss": 0.2097, + "step": 7240 + }, + { + "epoch": 0.2634639145286721, + "grad_norm": 8.892645835876465, + "learning_rate": 2.6320953557671347e-05, + "loss": 0.2835, + "step": 7250 + }, + { + "epoch": 0.26382731303147033, + "grad_norm": 1.616268277168274, + "learning_rate": 2.6357293407951162e-05, + "loss": 0.1875, + "step": 7260 + }, + { + "epoch": 0.2641907115342685, + "grad_norm": 2.1791138648986816, + "learning_rate": 2.6393633258230975e-05, + "loss": 0.1722, + "step": 7270 + }, + { + "epoch": 0.26455411003706664, + "grad_norm": 2.8691608905792236, + "learning_rate": 2.642997310851079e-05, + "loss": 0.2377, + "step": 7280 + }, + { + "epoch": 0.26491750853986484, + "grad_norm": 1.5673551559448242, + "learning_rate": 2.6466312958790614e-05, + "loss": 0.4404, + "step": 7290 + }, + { + "epoch": 0.265280907042663, + "grad_norm": 7.296738147735596, + "learning_rate": 2.6502652809070426e-05, + "loss": 0.3198, + "step": 7300 + }, + { + "epoch": 0.26564430554546115, + "grad_norm": 6.389322757720947, + "learning_rate": 2.6538992659350242e-05, + "loss": 0.2041, + "step": 7310 + }, + { + "epoch": 0.2660077040482593, + "grad_norm": 11.64201831817627, + "learning_rate": 2.657533250963006e-05, + "loss": 0.2014, + "step": 7320 + }, + { + "epoch": 0.2663711025510575, + "grad_norm": 4.454049587249756, + "learning_rate": 2.6611672359909878e-05, + "loss": 0.2295, + "step": 7330 + }, + { + "epoch": 0.26673450105385565, + "grad_norm": 2.091968297958374, + "learning_rate": 2.6648012210189694e-05, + "loss": 0.1784, + "step": 7340 + }, + { + "epoch": 0.2670978995566538, + "grad_norm": 6.904966354370117, + "learning_rate": 2.6684352060469513e-05, + "loss": 0.3303, + "step": 7350 + }, + { + "epoch": 0.267461298059452, + "grad_norm": 1.6893994808197021, + "learning_rate": 2.672069191074933e-05, + "loss": 0.2534, + "step": 7360 + }, + { + "epoch": 0.26782469656225016, + "grad_norm": 1.3456122875213623, + "learning_rate": 2.6757031761029145e-05, + "loss": 0.1829, + "step": 7370 + }, + { + "epoch": 0.2681880950650483, + "grad_norm": 7.959611892700195, + "learning_rate": 2.6793371611308964e-05, + "loss": 0.2425, + "step": 7380 + }, + { + "epoch": 0.2685514935678465, + "grad_norm": 1.5833840370178223, + "learning_rate": 2.682971146158878e-05, + "loss": 0.1988, + "step": 7390 + }, + { + "epoch": 0.2689148920706447, + "grad_norm": 19.886600494384766, + "learning_rate": 2.6866051311868596e-05, + "loss": 0.3563, + "step": 7400 + }, + { + "epoch": 0.2692782905734428, + "grad_norm": 2.55553936958313, + "learning_rate": 2.6902391162148415e-05, + "loss": 0.1857, + "step": 7410 + }, + { + "epoch": 0.26964168907624103, + "grad_norm": 2.125661849975586, + "learning_rate": 2.693873101242823e-05, + "loss": 0.7398, + "step": 7420 + }, + { + "epoch": 0.2700050875790392, + "grad_norm": 2.577770233154297, + "learning_rate": 2.6975070862708047e-05, + "loss": 0.5703, + "step": 7430 + }, + { + "epoch": 0.27036848608183733, + "grad_norm": 2.3848683834075928, + "learning_rate": 2.701141071298786e-05, + "loss": 0.173, + "step": 7440 + }, + { + "epoch": 0.2707318845846355, + "grad_norm": 22.96078109741211, + "learning_rate": 2.7047750563267683e-05, + "loss": 0.293, + "step": 7450 + }, + { + "epoch": 0.2710952830874337, + "grad_norm": 3.206329822540283, + "learning_rate": 2.70840904135475e-05, + "loss": 0.4585, + "step": 7460 + }, + { + "epoch": 0.27145868159023184, + "grad_norm": 2.251904010772705, + "learning_rate": 2.712043026382731e-05, + "loss": 0.2196, + "step": 7470 + }, + { + "epoch": 0.27182208009303, + "grad_norm": 3.7445387840270996, + "learning_rate": 2.7156770114107134e-05, + "loss": 0.2195, + "step": 7480 + }, + { + "epoch": 0.2721854785958282, + "grad_norm": 1.5370314121246338, + "learning_rate": 2.7193109964386947e-05, + "loss": 0.2007, + "step": 7490 + }, + { + "epoch": 0.27254887709862635, + "grad_norm": 18.44324493408203, + "learning_rate": 2.7229449814666763e-05, + "loss": 0.3091, + "step": 7500 + }, + { + "epoch": 0.2729122756014245, + "grad_norm": 1.5792795419692993, + "learning_rate": 2.7265789664946585e-05, + "loss": 0.1601, + "step": 7510 + }, + { + "epoch": 0.2732756741042227, + "grad_norm": 9.128384590148926, + "learning_rate": 2.7302129515226398e-05, + "loss": 0.178, + "step": 7520 + }, + { + "epoch": 0.27363907260702086, + "grad_norm": 2.2285592555999756, + "learning_rate": 2.7338469365506214e-05, + "loss": 2.4074, + "step": 7530 + }, + { + "epoch": 0.274002471109819, + "grad_norm": 2.2741541862487793, + "learning_rate": 2.7374809215786033e-05, + "loss": 0.246, + "step": 7540 + }, + { + "epoch": 0.2743658696126172, + "grad_norm": 17.185470581054688, + "learning_rate": 2.741114906606585e-05, + "loss": 0.2577, + "step": 7550 + }, + { + "epoch": 0.27472926811541537, + "grad_norm": 1.1907752752304077, + "learning_rate": 2.7447488916345665e-05, + "loss": 0.2073, + "step": 7560 + }, + { + "epoch": 0.2750926666182135, + "grad_norm": 3.535682201385498, + "learning_rate": 2.748382876662548e-05, + "loss": 0.2012, + "step": 7570 + }, + { + "epoch": 0.27545606512101173, + "grad_norm": 3.585460662841797, + "learning_rate": 2.75201686169053e-05, + "loss": 0.2147, + "step": 7580 + }, + { + "epoch": 0.2758194636238099, + "grad_norm": 1.9034504890441895, + "learning_rate": 2.7556508467185116e-05, + "loss": 0.1626, + "step": 7590 + }, + { + "epoch": 0.27618286212660803, + "grad_norm": 39.66155242919922, + "learning_rate": 2.7592848317464932e-05, + "loss": 0.2617, + "step": 7600 + }, + { + "epoch": 0.2765462606294062, + "grad_norm": 1.5698285102844238, + "learning_rate": 2.762918816774475e-05, + "loss": 0.3136, + "step": 7610 + }, + { + "epoch": 0.2769096591322044, + "grad_norm": 2.4866106510162354, + "learning_rate": 2.7665528018024568e-05, + "loss": 0.1971, + "step": 7620 + }, + { + "epoch": 0.27727305763500254, + "grad_norm": 9.244050025939941, + "learning_rate": 2.7701867868304384e-05, + "loss": 0.2025, + "step": 7630 + }, + { + "epoch": 0.2776364561378007, + "grad_norm": 2.1344380378723145, + "learning_rate": 2.7738207718584203e-05, + "loss": 0.2055, + "step": 7640 + }, + { + "epoch": 0.2779998546405989, + "grad_norm": 13.503227233886719, + "learning_rate": 2.777454756886402e-05, + "loss": 0.2671, + "step": 7650 + }, + { + "epoch": 0.27836325314339705, + "grad_norm": 2.238834857940674, + "learning_rate": 2.781088741914383e-05, + "loss": 0.1714, + "step": 7660 + }, + { + "epoch": 0.2787266516461952, + "grad_norm": 0.897280216217041, + "learning_rate": 2.7847227269423654e-05, + "loss": 0.1615, + "step": 7670 + }, + { + "epoch": 0.2790900501489934, + "grad_norm": 5.808285713195801, + "learning_rate": 2.788356711970347e-05, + "loss": 0.2052, + "step": 7680 + }, + { + "epoch": 0.27945344865179156, + "grad_norm": 1.8924663066864014, + "learning_rate": 2.7919906969983283e-05, + "loss": 0.1769, + "step": 7690 + }, + { + "epoch": 0.2798168471545897, + "grad_norm": 11.939653396606445, + "learning_rate": 2.7956246820263105e-05, + "loss": 0.2859, + "step": 7700 + }, + { + "epoch": 0.2801802456573879, + "grad_norm": 2.5077621936798096, + "learning_rate": 2.7992586670542918e-05, + "loss": 0.1767, + "step": 7710 + }, + { + "epoch": 0.28054364416018607, + "grad_norm": 2.0336718559265137, + "learning_rate": 2.8028926520822734e-05, + "loss": 0.6757, + "step": 7720 + }, + { + "epoch": 0.2809070426629842, + "grad_norm": 3.9547739028930664, + "learning_rate": 2.806526637110255e-05, + "loss": 0.2322, + "step": 7730 + }, + { + "epoch": 0.28127044116578237, + "grad_norm": 1.8082466125488281, + "learning_rate": 2.810160622138237e-05, + "loss": 0.1758, + "step": 7740 + }, + { + "epoch": 0.2816338396685806, + "grad_norm": 16.173986434936523, + "learning_rate": 2.8137946071662185e-05, + "loss": 0.2642, + "step": 7750 + }, + { + "epoch": 0.28199723817137873, + "grad_norm": 3.341475486755371, + "learning_rate": 2.8174285921942e-05, + "loss": 3.4407, + "step": 7760 + }, + { + "epoch": 0.2823606366741769, + "grad_norm": 1.7220288515090942, + "learning_rate": 2.821062577222182e-05, + "loss": 0.1965, + "step": 7770 + }, + { + "epoch": 0.2827240351769751, + "grad_norm": 3.8534610271453857, + "learning_rate": 2.8246965622501637e-05, + "loss": 0.1966, + "step": 7780 + }, + { + "epoch": 0.28308743367977324, + "grad_norm": 1.962780475616455, + "learning_rate": 2.8283305472781453e-05, + "loss": 0.1859, + "step": 7790 + }, + { + "epoch": 0.2834508321825714, + "grad_norm": 40.28166961669922, + "learning_rate": 2.8319645323061272e-05, + "loss": 0.6588, + "step": 7800 + }, + { + "epoch": 0.2834508321825714, + "eval_loss": 0.42970865964889526, + "eval_runtime": 180.6321, + "eval_samples_per_second": 41.045, + "eval_steps_per_second": 5.132, + "eval_wer": 0.2413455080145951, + "step": 7800 + }, + { + "epoch": 0.2838142306853696, + "grad_norm": 1.748349666595459, + "learning_rate": 2.8355985173341088e-05, + "loss": 0.1786, + "step": 7810 + }, + { + "epoch": 0.28417762918816775, + "grad_norm": 2.1137237548828125, + "learning_rate": 2.8392325023620904e-05, + "loss": 0.1803, + "step": 7820 + }, + { + "epoch": 0.2845410276909659, + "grad_norm": 1.59931218624115, + "learning_rate": 2.8428664873900723e-05, + "loss": 0.2107, + "step": 7830 + }, + { + "epoch": 0.2849044261937641, + "grad_norm": 2.263493061065674, + "learning_rate": 2.846500472418054e-05, + "loss": 0.1967, + "step": 7840 + }, + { + "epoch": 0.28526782469656226, + "grad_norm": 20.798656463623047, + "learning_rate": 2.8501344574460355e-05, + "loss": 0.268, + "step": 7850 + }, + { + "epoch": 0.2856312231993604, + "grad_norm": 3.0182480812072754, + "learning_rate": 2.8537684424740168e-05, + "loss": 0.1901, + "step": 7860 + }, + { + "epoch": 0.2859946217021586, + "grad_norm": 6.6378493309021, + "learning_rate": 2.857402427501999e-05, + "loss": 0.1804, + "step": 7870 + }, + { + "epoch": 0.28635802020495676, + "grad_norm": 2.5524067878723145, + "learning_rate": 2.8610364125299803e-05, + "loss": 0.233, + "step": 7880 + }, + { + "epoch": 0.2867214187077549, + "grad_norm": 2.6409335136413574, + "learning_rate": 2.864670397557962e-05, + "loss": 0.1717, + "step": 7890 + }, + { + "epoch": 0.28708481721055307, + "grad_norm": 6.834221363067627, + "learning_rate": 2.868304382585944e-05, + "loss": 0.2956, + "step": 7900 + }, + { + "epoch": 0.2874482157133513, + "grad_norm": 2.760669708251953, + "learning_rate": 2.8719383676139254e-05, + "loss": 0.1789, + "step": 7910 + }, + { + "epoch": 0.2878116142161494, + "grad_norm": 1.7543925046920776, + "learning_rate": 2.875572352641907e-05, + "loss": 0.2041, + "step": 7920 + }, + { + "epoch": 0.2881750127189476, + "grad_norm": 4.784151077270508, + "learning_rate": 2.879206337669889e-05, + "loss": 0.2259, + "step": 7930 + }, + { + "epoch": 0.2885384112217458, + "grad_norm": 2.1769356727600098, + "learning_rate": 2.8828403226978706e-05, + "loss": 0.2023, + "step": 7940 + }, + { + "epoch": 0.28890180972454393, + "grad_norm": 9.373051643371582, + "learning_rate": 2.886474307725852e-05, + "loss": 0.3511, + "step": 7950 + }, + { + "epoch": 0.2892652082273421, + "grad_norm": 1.895190715789795, + "learning_rate": 2.890108292753834e-05, + "loss": 0.1976, + "step": 7960 + }, + { + "epoch": 0.2896286067301403, + "grad_norm": 3.4400076866149902, + "learning_rate": 2.8937422777818157e-05, + "loss": 0.1902, + "step": 7970 + }, + { + "epoch": 0.28999200523293844, + "grad_norm": 9.663911819458008, + "learning_rate": 2.8973762628097973e-05, + "loss": 0.2551, + "step": 7980 + }, + { + "epoch": 0.2903554037357366, + "grad_norm": 5.1054463386535645, + "learning_rate": 2.9010102478377792e-05, + "loss": 0.2001, + "step": 7990 + }, + { + "epoch": 0.2907188022385348, + "grad_norm": 9.06143569946289, + "learning_rate": 2.9046442328657608e-05, + "loss": 0.2266, + "step": 8000 + }, + { + "epoch": 0.29108220074133295, + "grad_norm": 1.604077696800232, + "learning_rate": 2.9082782178937424e-05, + "loss": 0.1883, + "step": 8010 + }, + { + "epoch": 0.2914455992441311, + "grad_norm": 2.245687246322632, + "learning_rate": 2.911912202921724e-05, + "loss": 0.2093, + "step": 8020 + }, + { + "epoch": 0.29180899774692926, + "grad_norm": 3.8099372386932373, + "learning_rate": 2.915546187949706e-05, + "loss": 0.2283, + "step": 8030 + }, + { + "epoch": 0.29217239624972746, + "grad_norm": 2.135115623474121, + "learning_rate": 2.9191801729776875e-05, + "loss": 0.2369, + "step": 8040 + }, + { + "epoch": 0.2925357947525256, + "grad_norm": 5.596993446350098, + "learning_rate": 2.9228141580056688e-05, + "loss": 0.2709, + "step": 8050 + }, + { + "epoch": 0.29289919325532376, + "grad_norm": 1.3212496042251587, + "learning_rate": 2.926448143033651e-05, + "loss": 0.1968, + "step": 8060 + }, + { + "epoch": 0.29326259175812197, + "grad_norm": 1.9241231679916382, + "learning_rate": 2.9300821280616327e-05, + "loss": 0.3883, + "step": 8070 + }, + { + "epoch": 0.2936259902609201, + "grad_norm": 4.008016109466553, + "learning_rate": 2.933716113089614e-05, + "loss": 0.2074, + "step": 8080 + }, + { + "epoch": 0.2939893887637183, + "grad_norm": 1.5871399641036987, + "learning_rate": 2.9373500981175962e-05, + "loss": 0.1698, + "step": 8090 + }, + { + "epoch": 0.2943527872665165, + "grad_norm": 19.480670928955078, + "learning_rate": 2.9409840831455774e-05, + "loss": 0.4023, + "step": 8100 + }, + { + "epoch": 0.29471618576931463, + "grad_norm": 3.8420443534851074, + "learning_rate": 2.944618068173559e-05, + "loss": 0.181, + "step": 8110 + }, + { + "epoch": 0.2950795842721128, + "grad_norm": 1.9951499700546265, + "learning_rate": 2.9482520532015413e-05, + "loss": 0.2872, + "step": 8120 + }, + { + "epoch": 0.295442982774911, + "grad_norm": 4.958978176116943, + "learning_rate": 2.9518860382295226e-05, + "loss": 0.2359, + "step": 8130 + }, + { + "epoch": 0.29580638127770914, + "grad_norm": 1.5531708002090454, + "learning_rate": 2.9555200232575042e-05, + "loss": 0.2138, + "step": 8140 + }, + { + "epoch": 0.2961697797805073, + "grad_norm": 5.297884941101074, + "learning_rate": 2.9591540082854864e-05, + "loss": 0.2694, + "step": 8150 + }, + { + "epoch": 0.2965331782833055, + "grad_norm": 1.5989892482757568, + "learning_rate": 2.9627879933134677e-05, + "loss": 0.1686, + "step": 8160 + }, + { + "epoch": 0.29689657678610365, + "grad_norm": 3.347722291946411, + "learning_rate": 2.9664219783414493e-05, + "loss": 0.2206, + "step": 8170 + }, + { + "epoch": 0.2972599752889018, + "grad_norm": 2.9551491737365723, + "learning_rate": 2.970055963369431e-05, + "loss": 0.2274, + "step": 8180 + }, + { + "epoch": 0.29762337379169995, + "grad_norm": 2.527963638305664, + "learning_rate": 2.973689948397413e-05, + "loss": 0.1731, + "step": 8190 + }, + { + "epoch": 0.29798677229449816, + "grad_norm": 5.818012714385986, + "learning_rate": 2.9773239334253944e-05, + "loss": 0.265, + "step": 8200 + }, + { + "epoch": 0.2983501707972963, + "grad_norm": 1.5580624341964722, + "learning_rate": 2.980594519950578e-05, + "loss": 2.627, + "step": 8210 + }, + { + "epoch": 0.29871356930009446, + "grad_norm": 1.6011282205581665, + "learning_rate": 2.9842285049785594e-05, + "loss": 0.1811, + "step": 8220 + }, + { + "epoch": 0.29907696780289267, + "grad_norm": 44.825157165527344, + "learning_rate": 2.987862490006541e-05, + "loss": 0.4799, + "step": 8230 + }, + { + "epoch": 0.2994403663056908, + "grad_norm": 1.520982027053833, + "learning_rate": 2.991496475034523e-05, + "loss": 0.1935, + "step": 8240 + }, + { + "epoch": 0.29980376480848897, + "grad_norm": 6.3379058837890625, + "learning_rate": 2.9951304600625046e-05, + "loss": 0.2435, + "step": 8250 + }, + { + "epoch": 0.3001671633112872, + "grad_norm": 2.2493958473205566, + "learning_rate": 2.998764445090486e-05, + "loss": 0.1984, + "step": 8260 + }, + { + "epoch": 0.30053056181408533, + "grad_norm": 3.234196186065674, + "learning_rate": 3.002398430118468e-05, + "loss": 0.1785, + "step": 8270 + }, + { + "epoch": 0.3008939603168835, + "grad_norm": 4.99449348449707, + "learning_rate": 3.0060324151464497e-05, + "loss": 0.1888, + "step": 8280 + }, + { + "epoch": 0.3012573588196817, + "grad_norm": 1.8624048233032227, + "learning_rate": 3.0096664001744313e-05, + "loss": 1.6561, + "step": 8290 + }, + { + "epoch": 0.30162075732247984, + "grad_norm": 7.615640640258789, + "learning_rate": 3.0133003852024132e-05, + "loss": 0.2918, + "step": 8300 + }, + { + "epoch": 0.301984155825278, + "grad_norm": 1.6900697946548462, + "learning_rate": 3.0169343702303948e-05, + "loss": 0.2255, + "step": 8310 + }, + { + "epoch": 0.3023475543280762, + "grad_norm": 2.2034566402435303, + "learning_rate": 3.0205683552583764e-05, + "loss": 0.198, + "step": 8320 + }, + { + "epoch": 0.30271095283087435, + "grad_norm": 2.044597625732422, + "learning_rate": 3.0242023402863583e-05, + "loss": 0.1946, + "step": 8330 + }, + { + "epoch": 0.3030743513336725, + "grad_norm": 1.6171079874038696, + "learning_rate": 3.02783632531434e-05, + "loss": 0.1935, + "step": 8340 + }, + { + "epoch": 0.30343774983647065, + "grad_norm": 2.8435897827148438, + "learning_rate": 3.0314703103423215e-05, + "loss": 0.3876, + "step": 8350 + }, + { + "epoch": 0.30380114833926886, + "grad_norm": 2.023019552230835, + "learning_rate": 3.0351042953703035e-05, + "loss": 0.1879, + "step": 8360 + }, + { + "epoch": 0.304164546842067, + "grad_norm": 1.7610963582992554, + "learning_rate": 3.038738280398285e-05, + "loss": 0.1901, + "step": 8370 + }, + { + "epoch": 0.30452794534486516, + "grad_norm": 1.9482131004333496, + "learning_rate": 3.0423722654262667e-05, + "loss": 0.2119, + "step": 8380 + }, + { + "epoch": 0.30489134384766337, + "grad_norm": 1.6463958024978638, + "learning_rate": 3.046006250454248e-05, + "loss": 0.2067, + "step": 8390 + }, + { + "epoch": 0.3052547423504615, + "grad_norm": 10.607688903808594, + "learning_rate": 3.0496402354822302e-05, + "loss": 0.2709, + "step": 8400 + }, + { + "epoch": 0.3052547423504615, + "eval_loss": 0.3912598192691803, + "eval_runtime": 179.9461, + "eval_samples_per_second": 41.201, + "eval_steps_per_second": 5.152, + "eval_wer": 0.22865648882676493, + "step": 8400 + }, + { + "epoch": 0.30561814085325967, + "grad_norm": 5.675121307373047, + "learning_rate": 3.053274220510212e-05, + "loss": 0.1937, + "step": 8410 + }, + { + "epoch": 0.3059815393560579, + "grad_norm": 1.9001195430755615, + "learning_rate": 3.056908205538193e-05, + "loss": 0.1668, + "step": 8420 + }, + { + "epoch": 0.306344937858856, + "grad_norm": 6.807525157928467, + "learning_rate": 3.060542190566175e-05, + "loss": 0.2077, + "step": 8430 + }, + { + "epoch": 0.3067083363616542, + "grad_norm": 2.067265272140503, + "learning_rate": 3.064176175594157e-05, + "loss": 0.1596, + "step": 8440 + }, + { + "epoch": 0.3070717348644524, + "grad_norm": 15.267791748046875, + "learning_rate": 3.067810160622138e-05, + "loss": 0.2667, + "step": 8450 + }, + { + "epoch": 0.30743513336725053, + "grad_norm": 1.367903709411621, + "learning_rate": 3.07144414565012e-05, + "loss": 0.1819, + "step": 8460 + }, + { + "epoch": 0.3077985318700487, + "grad_norm": 1.531816840171814, + "learning_rate": 3.075078130678102e-05, + "loss": 0.1681, + "step": 8470 + }, + { + "epoch": 0.30816193037284684, + "grad_norm": 3.668304204940796, + "learning_rate": 3.078712115706083e-05, + "loss": 0.2488, + "step": 8480 + }, + { + "epoch": 0.30852532887564504, + "grad_norm": 2.2622220516204834, + "learning_rate": 3.082346100734065e-05, + "loss": 0.1866, + "step": 8490 + }, + { + "epoch": 0.3088887273784432, + "grad_norm": 6.450117111206055, + "learning_rate": 3.085980085762047e-05, + "loss": 0.2676, + "step": 8500 + }, + { + "epoch": 0.30925212588124135, + "grad_norm": 2.096731424331665, + "learning_rate": 3.0896140707900284e-05, + "loss": 0.1952, + "step": 8510 + }, + { + "epoch": 0.30961552438403955, + "grad_norm": 1.3809120655059814, + "learning_rate": 3.09324805581801e-05, + "loss": 0.3478, + "step": 8520 + }, + { + "epoch": 0.3099789228868377, + "grad_norm": 4.2257585525512695, + "learning_rate": 3.096882040845992e-05, + "loss": 0.2126, + "step": 8530 + }, + { + "epoch": 0.31034232138963586, + "grad_norm": 2.8543758392333984, + "learning_rate": 3.1005160258739736e-05, + "loss": 0.8169, + "step": 8540 + }, + { + "epoch": 0.31070571989243406, + "grad_norm": 5.897162437438965, + "learning_rate": 3.104150010901955e-05, + "loss": 0.2421, + "step": 8550 + }, + { + "epoch": 0.3110691183952322, + "grad_norm": 1.8980865478515625, + "learning_rate": 3.107783995929937e-05, + "loss": 0.193, + "step": 8560 + }, + { + "epoch": 0.31143251689803036, + "grad_norm": 2.113833427429199, + "learning_rate": 3.111417980957919e-05, + "loss": 0.1553, + "step": 8570 + }, + { + "epoch": 0.31179591540082857, + "grad_norm": 2.7569572925567627, + "learning_rate": 3.1150519659859e-05, + "loss": 0.2003, + "step": 8580 + }, + { + "epoch": 0.3121593139036267, + "grad_norm": 2.480473756790161, + "learning_rate": 3.118685951013882e-05, + "loss": 0.2173, + "step": 8590 + }, + { + "epoch": 0.3125227124064249, + "grad_norm": 12.174234390258789, + "learning_rate": 3.122319936041864e-05, + "loss": 0.3081, + "step": 8600 + }, + { + "epoch": 0.3128861109092231, + "grad_norm": 2.8075544834136963, + "learning_rate": 3.125953921069845e-05, + "loss": 0.263, + "step": 8610 + }, + { + "epoch": 0.31324950941202123, + "grad_norm": 16.535009384155273, + "learning_rate": 3.129587906097827e-05, + "loss": 0.1968, + "step": 8620 + }, + { + "epoch": 0.3136129079148194, + "grad_norm": 6.4783711433410645, + "learning_rate": 3.133221891125809e-05, + "loss": 0.2396, + "step": 8630 + }, + { + "epoch": 0.31397630641761753, + "grad_norm": 0.945353090763092, + "learning_rate": 3.13685587615379e-05, + "loss": 0.1623, + "step": 8640 + }, + { + "epoch": 0.31433970492041574, + "grad_norm": 7.135663032531738, + "learning_rate": 3.140489861181772e-05, + "loss": 0.3006, + "step": 8650 + }, + { + "epoch": 0.3147031034232139, + "grad_norm": 1.275896430015564, + "learning_rate": 3.144123846209754e-05, + "loss": 0.1845, + "step": 8660 + }, + { + "epoch": 0.31506650192601204, + "grad_norm": 2.1660525798797607, + "learning_rate": 3.147757831237735e-05, + "loss": 0.1614, + "step": 8670 + }, + { + "epoch": 0.31542990042881025, + "grad_norm": 3.878882646560669, + "learning_rate": 3.1513918162657166e-05, + "loss": 0.2124, + "step": 8680 + }, + { + "epoch": 0.3157932989316084, + "grad_norm": 3.452864170074463, + "learning_rate": 3.155025801293699e-05, + "loss": 0.1659, + "step": 8690 + }, + { + "epoch": 0.31615669743440655, + "grad_norm": 4.0493292808532715, + "learning_rate": 3.1586597863216805e-05, + "loss": 0.2653, + "step": 8700 + }, + { + "epoch": 0.31652009593720476, + "grad_norm": 1.9184757471084595, + "learning_rate": 3.162293771349662e-05, + "loss": 0.2043, + "step": 8710 + }, + { + "epoch": 0.3168834944400029, + "grad_norm": 4.22302770614624, + "learning_rate": 3.165927756377644e-05, + "loss": 0.2005, + "step": 8720 + }, + { + "epoch": 0.31724689294280106, + "grad_norm": 8.557464599609375, + "learning_rate": 3.1695617414056256e-05, + "loss": 0.2135, + "step": 8730 + }, + { + "epoch": 0.31761029144559927, + "grad_norm": 1.6090949773788452, + "learning_rate": 3.173195726433607e-05, + "loss": 0.1565, + "step": 8740 + }, + { + "epoch": 0.3179736899483974, + "grad_norm": 35.859737396240234, + "learning_rate": 3.1768297114615894e-05, + "loss": 0.3239, + "step": 8750 + }, + { + "epoch": 0.31833708845119557, + "grad_norm": 2.837944507598877, + "learning_rate": 3.180463696489571e-05, + "loss": 0.1902, + "step": 8760 + }, + { + "epoch": 0.3187004869539937, + "grad_norm": 1.6548888683319092, + "learning_rate": 3.184097681517552e-05, + "loss": 0.1732, + "step": 8770 + }, + { + "epoch": 0.31906388545679193, + "grad_norm": 3.840034246444702, + "learning_rate": 3.187731666545534e-05, + "loss": 0.2318, + "step": 8780 + }, + { + "epoch": 0.3194272839595901, + "grad_norm": 3.3684277534484863, + "learning_rate": 3.191365651573516e-05, + "loss": 0.1794, + "step": 8790 + }, + { + "epoch": 0.31979068246238823, + "grad_norm": 8.668655395507812, + "learning_rate": 3.194999636601497e-05, + "loss": 0.2745, + "step": 8800 + }, + { + "epoch": 0.32015408096518644, + "grad_norm": 1.412441611289978, + "learning_rate": 3.198633621629479e-05, + "loss": 0.1913, + "step": 8810 + }, + { + "epoch": 0.3205174794679846, + "grad_norm": 1.6273925304412842, + "learning_rate": 3.202267606657461e-05, + "loss": 0.1905, + "step": 8820 + }, + { + "epoch": 0.32088087797078274, + "grad_norm": 5.704558372497559, + "learning_rate": 3.205901591685442e-05, + "loss": 0.2217, + "step": 8830 + }, + { + "epoch": 0.32124427647358095, + "grad_norm": 2.248072385787964, + "learning_rate": 3.209535576713424e-05, + "loss": 0.1752, + "step": 8840 + }, + { + "epoch": 0.3216076749763791, + "grad_norm": 8.330979347229004, + "learning_rate": 3.213169561741406e-05, + "loss": 0.2693, + "step": 8850 + }, + { + "epoch": 0.32197107347917725, + "grad_norm": 6.713444709777832, + "learning_rate": 3.2168035467693873e-05, + "loss": 0.1821, + "step": 8860 + }, + { + "epoch": 0.32233447198197546, + "grad_norm": 1.7717983722686768, + "learning_rate": 3.220437531797369e-05, + "loss": 0.1572, + "step": 8870 + }, + { + "epoch": 0.3226978704847736, + "grad_norm": 3.8419570922851562, + "learning_rate": 3.224071516825351e-05, + "loss": 0.2168, + "step": 8880 + }, + { + "epoch": 0.32306126898757176, + "grad_norm": 1.8515948057174683, + "learning_rate": 3.2277055018533325e-05, + "loss": 0.1474, + "step": 8890 + }, + { + "epoch": 0.32342466749036997, + "grad_norm": 12.963587760925293, + "learning_rate": 3.231339486881314e-05, + "loss": 0.2349, + "step": 8900 + }, + { + "epoch": 0.3237880659931681, + "grad_norm": 1.078845500946045, + "learning_rate": 3.2349734719092963e-05, + "loss": 0.1968, + "step": 8910 + }, + { + "epoch": 0.32415146449596627, + "grad_norm": 1.5369044542312622, + "learning_rate": 3.2386074569372776e-05, + "loss": 0.1681, + "step": 8920 + }, + { + "epoch": 0.3245148629987644, + "grad_norm": 3.8013484477996826, + "learning_rate": 3.242241441965259e-05, + "loss": 0.2214, + "step": 8930 + }, + { + "epoch": 0.3248782615015626, + "grad_norm": 2.0259406566619873, + "learning_rate": 3.2458754269932415e-05, + "loss": 0.4227, + "step": 8940 + }, + { + "epoch": 0.3252416600043608, + "grad_norm": 6.423609256744385, + "learning_rate": 3.249509412021223e-05, + "loss": 0.2835, + "step": 8950 + }, + { + "epoch": 0.32560505850715893, + "grad_norm": 2.363159656524658, + "learning_rate": 3.253143397049204e-05, + "loss": 0.2038, + "step": 8960 + }, + { + "epoch": 0.32596845700995714, + "grad_norm": 2.4034435749053955, + "learning_rate": 3.256777382077186e-05, + "loss": 0.1907, + "step": 8970 + }, + { + "epoch": 0.3263318555127553, + "grad_norm": 4.032980442047119, + "learning_rate": 3.260411367105168e-05, + "loss": 0.1973, + "step": 8980 + }, + { + "epoch": 0.32669525401555344, + "grad_norm": 6.102022647857666, + "learning_rate": 3.264045352133149e-05, + "loss": 0.197, + "step": 8990 + }, + { + "epoch": 0.32705865251835164, + "grad_norm": 35.67893981933594, + "learning_rate": 3.267679337161131e-05, + "loss": 0.2682, + "step": 9000 + }, + { + "epoch": 0.32705865251835164, + "eval_loss": 0.40712428092956543, + "eval_runtime": 179.2194, + "eval_samples_per_second": 41.368, + "eval_steps_per_second": 5.172, + "eval_wer": 0.226941020567466, + "step": 9000 + }, + { + "epoch": 0.3274220510211498, + "grad_norm": 1.8014717102050781, + "learning_rate": 3.271313322189113e-05, + "loss": 0.1591, + "step": 9010 + }, + { + "epoch": 0.32778544952394795, + "grad_norm": 1.7404965162277222, + "learning_rate": 3.274947307217094e-05, + "loss": 0.17, + "step": 9020 + }, + { + "epoch": 0.32814884802674615, + "grad_norm": 3.7020771503448486, + "learning_rate": 3.278581292245076e-05, + "loss": 0.2225, + "step": 9030 + }, + { + "epoch": 0.3285122465295443, + "grad_norm": 1.045998454093933, + "learning_rate": 3.282215277273058e-05, + "loss": 0.1681, + "step": 9040 + }, + { + "epoch": 0.32887564503234246, + "grad_norm": 5.282716751098633, + "learning_rate": 3.2858492623010394e-05, + "loss": 0.2856, + "step": 9050 + }, + { + "epoch": 0.3292390435351406, + "grad_norm": 3.3956387042999268, + "learning_rate": 3.289483247329021e-05, + "loss": 0.1782, + "step": 9060 + }, + { + "epoch": 0.3296024420379388, + "grad_norm": 1.855603575706482, + "learning_rate": 3.293117232357003e-05, + "loss": 0.1582, + "step": 9070 + }, + { + "epoch": 0.32996584054073697, + "grad_norm": 7.214013576507568, + "learning_rate": 3.2967512173849845e-05, + "loss": 0.1691, + "step": 9080 + }, + { + "epoch": 0.3303292390435351, + "grad_norm": 3.140125036239624, + "learning_rate": 3.3003852024129664e-05, + "loss": 0.1872, + "step": 9090 + }, + { + "epoch": 0.3306926375463333, + "grad_norm": 17.094255447387695, + "learning_rate": 3.304019187440948e-05, + "loss": 0.2848, + "step": 9100 + }, + { + "epoch": 0.3310560360491315, + "grad_norm": 1.9439010620117188, + "learning_rate": 3.3076531724689296e-05, + "loss": 0.1625, + "step": 9110 + }, + { + "epoch": 0.3314194345519296, + "grad_norm": 1.609747290611267, + "learning_rate": 3.311287157496911e-05, + "loss": 0.1915, + "step": 9120 + }, + { + "epoch": 0.33178283305472783, + "grad_norm": 4.03629207611084, + "learning_rate": 3.314921142524893e-05, + "loss": 0.2291, + "step": 9130 + }, + { + "epoch": 0.332146231557526, + "grad_norm": 1.9643129110336304, + "learning_rate": 3.318555127552875e-05, + "loss": 0.1747, + "step": 9140 + }, + { + "epoch": 0.33250963006032414, + "grad_norm": 9.304847717285156, + "learning_rate": 3.322189112580856e-05, + "loss": 0.2539, + "step": 9150 + }, + { + "epoch": 0.33287302856312234, + "grad_norm": 1.991467833518982, + "learning_rate": 3.325823097608838e-05, + "loss": 3.61, + "step": 9160 + }, + { + "epoch": 0.3332364270659205, + "grad_norm": 2.7127187252044678, + "learning_rate": 3.32945708263682e-05, + "loss": 0.1985, + "step": 9170 + }, + { + "epoch": 0.33359982556871864, + "grad_norm": 2.831299304962158, + "learning_rate": 3.333091067664801e-05, + "loss": 1.7334, + "step": 9180 + }, + { + "epoch": 0.33396322407151685, + "grad_norm": 1.5434614419937134, + "learning_rate": 3.336725052692783e-05, + "loss": 0.1718, + "step": 9190 + }, + { + "epoch": 0.334326622574315, + "grad_norm": 10.254124641418457, + "learning_rate": 3.340359037720765e-05, + "loss": 0.3246, + "step": 9200 + }, + { + "epoch": 0.33469002107711315, + "grad_norm": 1.169886589050293, + "learning_rate": 3.343993022748746e-05, + "loss": 0.1936, + "step": 9210 + }, + { + "epoch": 0.3350534195799113, + "grad_norm": 3.697627544403076, + "learning_rate": 3.347627007776728e-05, + "loss": 0.205, + "step": 9220 + }, + { + "epoch": 0.3354168180827095, + "grad_norm": 3.15781307220459, + "learning_rate": 3.35126099280471e-05, + "loss": 0.2222, + "step": 9230 + }, + { + "epoch": 0.33578021658550766, + "grad_norm": 1.903701663017273, + "learning_rate": 3.3548949778326914e-05, + "loss": 0.1611, + "step": 9240 + }, + { + "epoch": 0.3361436150883058, + "grad_norm": 26.77275848388672, + "learning_rate": 3.358528962860673e-05, + "loss": 0.2872, + "step": 9250 + }, + { + "epoch": 0.336507013591104, + "grad_norm": 1.588224172592163, + "learning_rate": 3.3621629478886546e-05, + "loss": 3.404, + "step": 9260 + }, + { + "epoch": 0.33687041209390217, + "grad_norm": 1.8802090883255005, + "learning_rate": 3.3657969329166365e-05, + "loss": 0.1715, + "step": 9270 + }, + { + "epoch": 0.3372338105967003, + "grad_norm": 5.38352632522583, + "learning_rate": 3.3694309179446185e-05, + "loss": 0.1906, + "step": 9280 + }, + { + "epoch": 0.33759720909949853, + "grad_norm": 1.736177921295166, + "learning_rate": 3.3730649029726e-05, + "loss": 0.1881, + "step": 9290 + }, + { + "epoch": 0.3379606076022967, + "grad_norm": 17.865558624267578, + "learning_rate": 3.3766988880005816e-05, + "loss": 0.3003, + "step": 9300 + }, + { + "epoch": 0.33832400610509483, + "grad_norm": 1.532173991203308, + "learning_rate": 3.3803328730285636e-05, + "loss": 0.188, + "step": 9310 + }, + { + "epoch": 0.33868740460789304, + "grad_norm": 3.8595352172851562, + "learning_rate": 3.383966858056545e-05, + "loss": 0.1869, + "step": 9320 + }, + { + "epoch": 0.3390508031106912, + "grad_norm": 2.5906641483306885, + "learning_rate": 3.387600843084527e-05, + "loss": 0.1993, + "step": 9330 + }, + { + "epoch": 0.33941420161348934, + "grad_norm": 2.5224273204803467, + "learning_rate": 3.391234828112508e-05, + "loss": 0.1935, + "step": 9340 + }, + { + "epoch": 0.33977760011628755, + "grad_norm": 11.555095672607422, + "learning_rate": 3.39486881314049e-05, + "loss": 0.2891, + "step": 9350 + }, + { + "epoch": 0.3401409986190857, + "grad_norm": 1.3724703788757324, + "learning_rate": 3.398502798168472e-05, + "loss": 0.1656, + "step": 9360 + }, + { + "epoch": 0.34050439712188385, + "grad_norm": 2.1549072265625, + "learning_rate": 3.402136783196453e-05, + "loss": 0.1769, + "step": 9370 + }, + { + "epoch": 0.340867795624682, + "grad_norm": 1.793492317199707, + "learning_rate": 3.405770768224435e-05, + "loss": 0.2661, + "step": 9380 + }, + { + "epoch": 0.3412311941274802, + "grad_norm": 4.038620948791504, + "learning_rate": 3.409404753252417e-05, + "loss": 0.1871, + "step": 9390 + }, + { + "epoch": 0.34159459263027836, + "grad_norm": 31.7847900390625, + "learning_rate": 3.413038738280398e-05, + "loss": 0.2967, + "step": 9400 + }, + { + "epoch": 0.3419579911330765, + "grad_norm": 2.398646354675293, + "learning_rate": 3.41667272330838e-05, + "loss": 0.2086, + "step": 9410 + }, + { + "epoch": 0.3423213896358747, + "grad_norm": 2.2226221561431885, + "learning_rate": 3.4203067083363615e-05, + "loss": 0.1665, + "step": 9420 + }, + { + "epoch": 0.34268478813867287, + "grad_norm": 39.96380615234375, + "learning_rate": 3.4239406933643434e-05, + "loss": 0.9468, + "step": 9430 + }, + { + "epoch": 0.343048186641471, + "grad_norm": 1.5465339422225952, + "learning_rate": 3.4275746783923254e-05, + "loss": 0.1827, + "step": 9440 + }, + { + "epoch": 0.3434115851442692, + "grad_norm": 7.941345691680908, + "learning_rate": 3.4312086634203066e-05, + "loss": 0.2786, + "step": 9450 + }, + { + "epoch": 0.3437749836470674, + "grad_norm": 1.2575476169586182, + "learning_rate": 3.4348426484482885e-05, + "loss": 0.1764, + "step": 9460 + }, + { + "epoch": 0.34413838214986553, + "grad_norm": 1.3529596328735352, + "learning_rate": 3.4384766334762705e-05, + "loss": 0.207, + "step": 9470 + }, + { + "epoch": 0.34450178065266374, + "grad_norm": 3.2839174270629883, + "learning_rate": 3.442110618504252e-05, + "loss": 0.2672, + "step": 9480 + }, + { + "epoch": 0.3448651791554619, + "grad_norm": 3.246384859085083, + "learning_rate": 3.445744603532234e-05, + "loss": 0.1906, + "step": 9490 + }, + { + "epoch": 0.34522857765826004, + "grad_norm": 2.595038652420044, + "learning_rate": 3.4493785885602156e-05, + "loss": 0.2441, + "step": 9500 + }, + { + "epoch": 0.3455919761610582, + "grad_norm": 1.3803220987319946, + "learning_rate": 3.453012573588197e-05, + "loss": 0.1745, + "step": 9510 + }, + { + "epoch": 0.3459553746638564, + "grad_norm": 1.2091724872589111, + "learning_rate": 3.456646558616179e-05, + "loss": 0.1441, + "step": 9520 + }, + { + "epoch": 0.34631877316665455, + "grad_norm": 6.582603931427002, + "learning_rate": 3.460280543644161e-05, + "loss": 0.1835, + "step": 9530 + }, + { + "epoch": 0.3466821716694527, + "grad_norm": 2.6845383644104004, + "learning_rate": 3.463914528672142e-05, + "loss": 0.2048, + "step": 9540 + }, + { + "epoch": 0.3470455701722509, + "grad_norm": 11.775678634643555, + "learning_rate": 3.467548513700123e-05, + "loss": 0.2841, + "step": 9550 + }, + { + "epoch": 0.34740896867504906, + "grad_norm": 2.256279706954956, + "learning_rate": 3.471182498728106e-05, + "loss": 0.6472, + "step": 9560 + }, + { + "epoch": 0.3477723671778472, + "grad_norm": 1.4487576484680176, + "learning_rate": 3.474816483756087e-05, + "loss": 0.2722, + "step": 9570 + }, + { + "epoch": 0.3481357656806454, + "grad_norm": 3.843964099884033, + "learning_rate": 3.4784504687840684e-05, + "loss": 0.1855, + "step": 9580 + }, + { + "epoch": 0.34849916418344357, + "grad_norm": 1.5561772584915161, + "learning_rate": 3.48208445381205e-05, + "loss": 0.1908, + "step": 9590 + }, + { + "epoch": 0.3488625626862417, + "grad_norm": 3.757232666015625, + "learning_rate": 3.485718438840032e-05, + "loss": 0.2198, + "step": 9600 + }, + { + "epoch": 0.3488625626862417, + "eval_loss": 0.3895765244960785, + "eval_runtime": 179.7435, + "eval_samples_per_second": 41.248, + "eval_steps_per_second": 5.157, + "eval_wer": 0.21512335033674007, + "step": 9600 + }, + { + "epoch": 0.3492259611890399, + "grad_norm": 1.3912307024002075, + "learning_rate": 3.4893524238680135e-05, + "loss": 0.1616, + "step": 9610 + }, + { + "epoch": 0.3495893596918381, + "grad_norm": 2.4036080837249756, + "learning_rate": 3.4929864088959954e-05, + "loss": 0.1579, + "step": 9620 + }, + { + "epoch": 0.3499527581946362, + "grad_norm": 2.611175537109375, + "learning_rate": 3.4966203939239774e-05, + "loss": 0.1746, + "step": 9630 + }, + { + "epoch": 0.35031615669743443, + "grad_norm": 1.4045140743255615, + "learning_rate": 3.5002543789519586e-05, + "loss": 0.1594, + "step": 9640 + }, + { + "epoch": 0.3506795552002326, + "grad_norm": 12.708057403564453, + "learning_rate": 3.5038883639799406e-05, + "loss": 0.3118, + "step": 9650 + }, + { + "epoch": 0.35104295370303074, + "grad_norm": 3.0364696979522705, + "learning_rate": 3.5075223490079225e-05, + "loss": 0.3062, + "step": 9660 + }, + { + "epoch": 0.3514063522058289, + "grad_norm": 1.4527848958969116, + "learning_rate": 3.511156334035904e-05, + "loss": 0.1603, + "step": 9670 + }, + { + "epoch": 0.3517697507086271, + "grad_norm": 5.697939395904541, + "learning_rate": 3.514790319063886e-05, + "loss": 0.2069, + "step": 9680 + }, + { + "epoch": 0.35213314921142524, + "grad_norm": 2.1645712852478027, + "learning_rate": 3.5184243040918676e-05, + "loss": 0.162, + "step": 9690 + }, + { + "epoch": 0.3524965477142234, + "grad_norm": 8.024601936340332, + "learning_rate": 3.522058289119849e-05, + "loss": 0.898, + "step": 9700 + }, + { + "epoch": 0.3528599462170216, + "grad_norm": 1.4516103267669678, + "learning_rate": 3.52569227414783e-05, + "loss": 0.189, + "step": 9710 + }, + { + "epoch": 0.35322334471981975, + "grad_norm": 1.0467925071716309, + "learning_rate": 3.529326259175813e-05, + "loss": 0.1547, + "step": 9720 + }, + { + "epoch": 0.3535867432226179, + "grad_norm": 3.9237303733825684, + "learning_rate": 3.532960244203794e-05, + "loss": 0.1968, + "step": 9730 + }, + { + "epoch": 0.3539501417254161, + "grad_norm": 2.502257823944092, + "learning_rate": 3.536594229231775e-05, + "loss": 0.1645, + "step": 9740 + }, + { + "epoch": 0.35431354022821426, + "grad_norm": 30.662227630615234, + "learning_rate": 3.540228214259758e-05, + "loss": 0.2847, + "step": 9750 + }, + { + "epoch": 0.3546769387310124, + "grad_norm": 1.7106624841690063, + "learning_rate": 3.543862199287739e-05, + "loss": 0.1951, + "step": 9760 + }, + { + "epoch": 0.3550403372338106, + "grad_norm": 2.169036865234375, + "learning_rate": 3.5474961843157204e-05, + "loss": 0.172, + "step": 9770 + }, + { + "epoch": 0.3554037357366088, + "grad_norm": 6.116454124450684, + "learning_rate": 3.551130169343703e-05, + "loss": 0.1934, + "step": 9780 + }, + { + "epoch": 0.3557671342394069, + "grad_norm": 1.8530545234680176, + "learning_rate": 3.554764154371684e-05, + "loss": 0.217, + "step": 9790 + }, + { + "epoch": 0.3561305327422051, + "grad_norm": 11.060449600219727, + "learning_rate": 3.5583981393996655e-05, + "loss": 0.2145, + "step": 9800 + }, + { + "epoch": 0.3564939312450033, + "grad_norm": 7.748067378997803, + "learning_rate": 3.5620321244276475e-05, + "loss": 0.2114, + "step": 9810 + }, + { + "epoch": 0.35685732974780143, + "grad_norm": 3.562528610229492, + "learning_rate": 3.5656661094556294e-05, + "loss": 0.221, + "step": 9820 + }, + { + "epoch": 0.3572207282505996, + "grad_norm": 2.798417091369629, + "learning_rate": 3.5693000944836107e-05, + "loss": 0.2071, + "step": 9830 + }, + { + "epoch": 0.3575841267533978, + "grad_norm": 2.3908724784851074, + "learning_rate": 3.5729340795115926e-05, + "loss": 0.1678, + "step": 9840 + }, + { + "epoch": 0.35794752525619594, + "grad_norm": 7.205004692077637, + "learning_rate": 3.5765680645395745e-05, + "loss": 0.2953, + "step": 9850 + }, + { + "epoch": 0.3583109237589941, + "grad_norm": 2.5064749717712402, + "learning_rate": 3.580202049567556e-05, + "loss": 0.197, + "step": 9860 + }, + { + "epoch": 0.3586743222617923, + "grad_norm": 2.0985934734344482, + "learning_rate": 3.583836034595538e-05, + "loss": 0.1441, + "step": 9870 + }, + { + "epoch": 0.35903772076459045, + "grad_norm": 5.256442070007324, + "learning_rate": 3.5874700196235197e-05, + "loss": 0.203, + "step": 9880 + }, + { + "epoch": 0.3594011192673886, + "grad_norm": 2.3590219020843506, + "learning_rate": 3.591104004651501e-05, + "loss": 0.1811, + "step": 9890 + }, + { + "epoch": 0.3597645177701868, + "grad_norm": 24.96747398376465, + "learning_rate": 3.594737989679482e-05, + "loss": 0.293, + "step": 9900 + }, + { + "epoch": 0.36012791627298496, + "grad_norm": 1.727751612663269, + "learning_rate": 3.598371974707465e-05, + "loss": 0.1896, + "step": 9910 + }, + { + "epoch": 0.3604913147757831, + "grad_norm": 2.349269151687622, + "learning_rate": 3.602005959735446e-05, + "loss": 0.1649, + "step": 9920 + }, + { + "epoch": 0.3608547132785813, + "grad_norm": 3.139385223388672, + "learning_rate": 3.605639944763427e-05, + "loss": 0.2181, + "step": 9930 + }, + { + "epoch": 0.36121811178137947, + "grad_norm": 2.1249756813049316, + "learning_rate": 3.60927392979141e-05, + "loss": 0.1751, + "step": 9940 + }, + { + "epoch": 0.3615815102841776, + "grad_norm": 3.6616756916046143, + "learning_rate": 3.612907914819391e-05, + "loss": 0.2729, + "step": 9950 + }, + { + "epoch": 0.36194490878697577, + "grad_norm": 1.367600440979004, + "learning_rate": 3.6165418998473724e-05, + "loss": 0.1592, + "step": 9960 + }, + { + "epoch": 0.362308307289774, + "grad_norm": 1.8141239881515503, + "learning_rate": 3.620175884875355e-05, + "loss": 0.2867, + "step": 9970 + }, + { + "epoch": 0.36267170579257213, + "grad_norm": 7.0058794021606445, + "learning_rate": 3.623809869903336e-05, + "loss": 0.207, + "step": 9980 + }, + { + "epoch": 0.3630351042953703, + "grad_norm": 1.923048734664917, + "learning_rate": 3.6274438549313176e-05, + "loss": 0.346, + "step": 9990 + }, + { + "epoch": 0.3633985027981685, + "grad_norm": 16.30779457092285, + "learning_rate": 3.6310778399592995e-05, + "loss": 0.3107, + "step": 10000 + }, + { + "epoch": 0.36376190130096664, + "grad_norm": 1.979866862297058, + "learning_rate": 3.6347118249872814e-05, + "loss": 0.7999, + "step": 10010 + }, + { + "epoch": 0.3641252998037648, + "grad_norm": 2.7377023696899414, + "learning_rate": 3.638345810015263e-05, + "loss": 0.2005, + "step": 10020 + }, + { + "epoch": 0.364488698306563, + "grad_norm": 5.546159744262695, + "learning_rate": 3.6419797950432446e-05, + "loss": 0.1964, + "step": 10030 + }, + { + "epoch": 0.36485209680936115, + "grad_norm": 2.2417142391204834, + "learning_rate": 3.6456137800712265e-05, + "loss": 0.2078, + "step": 10040 + }, + { + "epoch": 0.3652154953121593, + "grad_norm": 7.2175092697143555, + "learning_rate": 3.649247765099208e-05, + "loss": 0.291, + "step": 10050 + }, + { + "epoch": 0.3655788938149575, + "grad_norm": 2.6172754764556885, + "learning_rate": 3.65288175012719e-05, + "loss": 0.2037, + "step": 10060 + }, + { + "epoch": 0.36594229231775566, + "grad_norm": 2.0634214878082275, + "learning_rate": 3.656515735155172e-05, + "loss": 0.1668, + "step": 10070 + }, + { + "epoch": 0.3663056908205538, + "grad_norm": 3.5431976318359375, + "learning_rate": 3.660149720183153e-05, + "loss": 0.475, + "step": 10080 + }, + { + "epoch": 0.366669089323352, + "grad_norm": 2.147472381591797, + "learning_rate": 3.663783705211135e-05, + "loss": 0.1869, + "step": 10090 + }, + { + "epoch": 0.36703248782615017, + "grad_norm": 18.726482391357422, + "learning_rate": 3.667417690239117e-05, + "loss": 0.2773, + "step": 10100 + }, + { + "epoch": 0.3673958863289483, + "grad_norm": 1.6554090976715088, + "learning_rate": 3.671051675267098e-05, + "loss": 0.1707, + "step": 10110 + }, + { + "epoch": 0.36775928483174647, + "grad_norm": 1.8967760801315308, + "learning_rate": 3.674685660295079e-05, + "loss": 0.2159, + "step": 10120 + }, + { + "epoch": 0.3681226833345447, + "grad_norm": 2.3765788078308105, + "learning_rate": 3.678319645323061e-05, + "loss": 0.2229, + "step": 10130 + }, + { + "epoch": 0.3684860818373428, + "grad_norm": 5.890452861785889, + "learning_rate": 3.681953630351043e-05, + "loss": 0.195, + "step": 10140 + }, + { + "epoch": 0.368849480340141, + "grad_norm": 5.045167446136475, + "learning_rate": 3.6855876153790244e-05, + "loss": 0.3111, + "step": 10150 + }, + { + "epoch": 0.3692128788429392, + "grad_norm": 2.37107253074646, + "learning_rate": 3.6892216004070064e-05, + "loss": 0.1942, + "step": 10160 + }, + { + "epoch": 0.36957627734573734, + "grad_norm": 1.9943170547485352, + "learning_rate": 3.692855585434988e-05, + "loss": 0.1906, + "step": 10170 + }, + { + "epoch": 0.3699396758485355, + "grad_norm": 3.16873836517334, + "learning_rate": 3.6964895704629696e-05, + "loss": 0.1791, + "step": 10180 + }, + { + "epoch": 0.3703030743513337, + "grad_norm": 15.252134323120117, + "learning_rate": 3.7001235554909515e-05, + "loss": 0.3702, + "step": 10190 + }, + { + "epoch": 0.37066647285413185, + "grad_norm": 8.845834732055664, + "learning_rate": 3.7037575405189334e-05, + "loss": 0.2765, + "step": 10200 + }, + { + "epoch": 0.37066647285413185, + "eval_loss": 0.4178149104118347, + "eval_runtime": 179.6523, + "eval_samples_per_second": 41.269, + "eval_steps_per_second": 5.16, + "eval_wer": 0.2237551509430537, + "step": 10200 + }, + { + "epoch": 0.37102987135693, + "grad_norm": 6.2689313888549805, + "learning_rate": 3.707391525546915e-05, + "loss": 0.1922, + "step": 10210 + }, + { + "epoch": 0.3713932698597282, + "grad_norm": 1.00067138671875, + "learning_rate": 3.7110255105748966e-05, + "loss": 0.1535, + "step": 10220 + }, + { + "epoch": 0.37175666836252635, + "grad_norm": 2.6602060794830322, + "learning_rate": 3.7146594956028786e-05, + "loss": 0.1959, + "step": 10230 + }, + { + "epoch": 0.3721200668653245, + "grad_norm": 4.743015766143799, + "learning_rate": 3.71829348063086e-05, + "loss": 0.2058, + "step": 10240 + }, + { + "epoch": 0.37248346536812266, + "grad_norm": 8.304347038269043, + "learning_rate": 3.721927465658842e-05, + "loss": 0.3027, + "step": 10250 + }, + { + "epoch": 0.37284686387092086, + "grad_norm": 1.8180521726608276, + "learning_rate": 3.725561450686824e-05, + "loss": 0.1708, + "step": 10260 + }, + { + "epoch": 0.373210262373719, + "grad_norm": 2.05625057220459, + "learning_rate": 3.729195435714805e-05, + "loss": 0.1824, + "step": 10270 + }, + { + "epoch": 0.37357366087651717, + "grad_norm": 2.426814317703247, + "learning_rate": 3.732829420742787e-05, + "loss": 0.197, + "step": 10280 + }, + { + "epoch": 0.3739370593793154, + "grad_norm": 1.658158540725708, + "learning_rate": 3.736463405770768e-05, + "loss": 0.1578, + "step": 10290 + }, + { + "epoch": 0.3743004578821135, + "grad_norm": 10.913407325744629, + "learning_rate": 3.74009739079875e-05, + "loss": 0.2728, + "step": 10300 + }, + { + "epoch": 0.3746638563849117, + "grad_norm": 1.6443781852722168, + "learning_rate": 3.743731375826732e-05, + "loss": 0.1656, + "step": 10310 + }, + { + "epoch": 0.3750272548877099, + "grad_norm": 1.0702744722366333, + "learning_rate": 3.747365360854713e-05, + "loss": 0.7132, + "step": 10320 + }, + { + "epoch": 0.37539065339050803, + "grad_norm": 5.8824052810668945, + "learning_rate": 3.750999345882695e-05, + "loss": 0.2701, + "step": 10330 + }, + { + "epoch": 0.3757540518933062, + "grad_norm": 4.373916149139404, + "learning_rate": 3.754633330910677e-05, + "loss": 0.2053, + "step": 10340 + }, + { + "epoch": 0.3761174503961044, + "grad_norm": 22.25397300720215, + "learning_rate": 3.7582673159386584e-05, + "loss": 0.2781, + "step": 10350 + }, + { + "epoch": 0.37648084889890254, + "grad_norm": 1.8272254467010498, + "learning_rate": 3.7619013009666403e-05, + "loss": 0.1833, + "step": 10360 + }, + { + "epoch": 0.3768442474017007, + "grad_norm": 3.286931037902832, + "learning_rate": 3.7655352859946216e-05, + "loss": 0.1576, + "step": 10370 + }, + { + "epoch": 0.3772076459044989, + "grad_norm": 5.283690929412842, + "learning_rate": 3.7691692710226035e-05, + "loss": 0.21, + "step": 10380 + }, + { + "epoch": 0.37757104440729705, + "grad_norm": 1.184476375579834, + "learning_rate": 3.7728032560505855e-05, + "loss": 0.2597, + "step": 10390 + }, + { + "epoch": 0.3779344429100952, + "grad_norm": 5.685116767883301, + "learning_rate": 3.776437241078567e-05, + "loss": 0.2476, + "step": 10400 + }, + { + "epoch": 0.37829784141289335, + "grad_norm": 1.1873399019241333, + "learning_rate": 3.7800712261065487e-05, + "loss": 0.1597, + "step": 10410 + }, + { + "epoch": 0.37866123991569156, + "grad_norm": 1.6136255264282227, + "learning_rate": 3.7837052111345306e-05, + "loss": 0.188, + "step": 10420 + }, + { + "epoch": 0.3790246384184897, + "grad_norm": 4.743179798126221, + "learning_rate": 3.787339196162512e-05, + "loss": 0.1962, + "step": 10430 + }, + { + "epoch": 0.37938803692128786, + "grad_norm": 2.603379011154175, + "learning_rate": 3.790973181190494e-05, + "loss": 0.1854, + "step": 10440 + }, + { + "epoch": 0.37975143542408607, + "grad_norm": 6.267378807067871, + "learning_rate": 3.794607166218475e-05, + "loss": 0.2569, + "step": 10450 + }, + { + "epoch": 0.3801148339268842, + "grad_norm": 5.370235919952393, + "learning_rate": 3.798241151246457e-05, + "loss": 0.1796, + "step": 10460 + }, + { + "epoch": 0.3804782324296824, + "grad_norm": 2.170964002609253, + "learning_rate": 3.801875136274439e-05, + "loss": 0.1713, + "step": 10470 + }, + { + "epoch": 0.3808416309324806, + "grad_norm": 4.134753704071045, + "learning_rate": 3.80550912130242e-05, + "loss": 0.2269, + "step": 10480 + }, + { + "epoch": 0.38120502943527873, + "grad_norm": 2.7026259899139404, + "learning_rate": 3.809143106330402e-05, + "loss": 0.1938, + "step": 10490 + }, + { + "epoch": 0.3815684279380769, + "grad_norm": 7.368224143981934, + "learning_rate": 3.812777091358384e-05, + "loss": 0.2617, + "step": 10500 + }, + { + "epoch": 0.3819318264408751, + "grad_norm": 1.3194938898086548, + "learning_rate": 3.816411076386365e-05, + "loss": 0.2066, + "step": 10510 + }, + { + "epoch": 0.38229522494367324, + "grad_norm": 1.901505470275879, + "learning_rate": 3.820045061414347e-05, + "loss": 0.1716, + "step": 10520 + }, + { + "epoch": 0.3826586234464714, + "grad_norm": 3.4045536518096924, + "learning_rate": 3.823679046442329e-05, + "loss": 0.1625, + "step": 10530 + }, + { + "epoch": 0.38302202194926954, + "grad_norm": 2.1540184020996094, + "learning_rate": 3.8273130314703104e-05, + "loss": 0.1829, + "step": 10540 + }, + { + "epoch": 0.38338542045206775, + "grad_norm": 14.377511024475098, + "learning_rate": 3.8309470164982924e-05, + "loss": 0.2747, + "step": 10550 + }, + { + "epoch": 0.3837488189548659, + "grad_norm": 1.9092762470245361, + "learning_rate": 3.834581001526274e-05, + "loss": 0.1728, + "step": 10560 + }, + { + "epoch": 0.38411221745766405, + "grad_norm": 1.867458462715149, + "learning_rate": 3.8382149865542556e-05, + "loss": 0.1752, + "step": 10570 + }, + { + "epoch": 0.38447561596046226, + "grad_norm": 5.246692657470703, + "learning_rate": 3.841848971582237e-05, + "loss": 0.1823, + "step": 10580 + }, + { + "epoch": 0.3848390144632604, + "grad_norm": 2.9294533729553223, + "learning_rate": 3.845482956610219e-05, + "loss": 0.2052, + "step": 10590 + }, + { + "epoch": 0.38520241296605856, + "grad_norm": 11.946113586425781, + "learning_rate": 3.849116941638201e-05, + "loss": 0.309, + "step": 10600 + }, + { + "epoch": 0.38556581146885677, + "grad_norm": 1.7155182361602783, + "learning_rate": 3.852750926666182e-05, + "loss": 0.175, + "step": 10610 + }, + { + "epoch": 0.3859292099716549, + "grad_norm": 1.1520076990127563, + "learning_rate": 3.856384911694164e-05, + "loss": 0.2129, + "step": 10620 + }, + { + "epoch": 0.38629260847445307, + "grad_norm": 1.9750351905822754, + "learning_rate": 3.860018896722146e-05, + "loss": 0.1725, + "step": 10630 + }, + { + "epoch": 0.3866560069772513, + "grad_norm": 4.309560298919678, + "learning_rate": 3.863652881750127e-05, + "loss": 0.1516, + "step": 10640 + }, + { + "epoch": 0.3870194054800494, + "grad_norm": 7.554156303405762, + "learning_rate": 3.867286866778109e-05, + "loss": 0.3069, + "step": 10650 + }, + { + "epoch": 0.3873828039828476, + "grad_norm": 3.7965683937072754, + "learning_rate": 3.870920851806091e-05, + "loss": 0.2014, + "step": 10660 + }, + { + "epoch": 0.3877462024856458, + "grad_norm": 3.8691935539245605, + "learning_rate": 3.874554836834072e-05, + "loss": 0.1678, + "step": 10670 + }, + { + "epoch": 0.38810960098844394, + "grad_norm": 4.144315719604492, + "learning_rate": 3.878188821862054e-05, + "loss": 0.2936, + "step": 10680 + }, + { + "epoch": 0.3884729994912421, + "grad_norm": 1.5667825937271118, + "learning_rate": 3.881822806890036e-05, + "loss": 0.1871, + "step": 10690 + }, + { + "epoch": 0.38883639799404024, + "grad_norm": 7.6076788902282715, + "learning_rate": 3.885456791918017e-05, + "loss": 0.2661, + "step": 10700 + }, + { + "epoch": 0.38919979649683845, + "grad_norm": 1.7828059196472168, + "learning_rate": 3.889090776945999e-05, + "loss": 0.1808, + "step": 10710 + }, + { + "epoch": 0.3895631949996366, + "grad_norm": 7.039370059967041, + "learning_rate": 3.892724761973981e-05, + "loss": 0.2484, + "step": 10720 + }, + { + "epoch": 0.38992659350243475, + "grad_norm": 2.1001148223876953, + "learning_rate": 3.8963587470019625e-05, + "loss": 0.1644, + "step": 10730 + }, + { + "epoch": 0.39028999200523296, + "grad_norm": 0.9235002398490906, + "learning_rate": 3.899992732029944e-05, + "loss": 0.172, + "step": 10740 + }, + { + "epoch": 0.3906533905080311, + "grad_norm": 10.066643714904785, + "learning_rate": 3.903626717057926e-05, + "loss": 0.2999, + "step": 10750 + }, + { + "epoch": 0.39101678901082926, + "grad_norm": 2.256965160369873, + "learning_rate": 3.9072607020859076e-05, + "loss": 0.2116, + "step": 10760 + }, + { + "epoch": 0.39138018751362746, + "grad_norm": 1.742125153541565, + "learning_rate": 3.910894687113889e-05, + "loss": 0.1838, + "step": 10770 + }, + { + "epoch": 0.3917435860164256, + "grad_norm": 5.397392749786377, + "learning_rate": 3.9145286721418714e-05, + "loss": 0.2213, + "step": 10780 + }, + { + "epoch": 0.39210698451922377, + "grad_norm": 2.439197540283203, + "learning_rate": 3.918162657169853e-05, + "loss": 0.1984, + "step": 10790 + }, + { + "epoch": 0.392470383022022, + "grad_norm": 6.7387895584106445, + "learning_rate": 3.921796642197834e-05, + "loss": 0.2842, + "step": 10800 + }, + { + "epoch": 0.392470383022022, + "eval_loss": 0.39516785740852356, + "eval_runtime": 180.1522, + "eval_samples_per_second": 41.154, + "eval_steps_per_second": 5.146, + "eval_wer": 0.21758309583023216, + "step": 10800 + }, + { + "epoch": 0.3928337815248201, + "grad_norm": 1.5229130983352661, + "learning_rate": 3.925430627225816e-05, + "loss": 0.1809, + "step": 10810 + }, + { + "epoch": 0.3931971800276183, + "grad_norm": 1.6385318040847778, + "learning_rate": 3.929064612253798e-05, + "loss": 0.155, + "step": 10820 + }, + { + "epoch": 0.3935605785304164, + "grad_norm": 2.403878927230835, + "learning_rate": 3.932698597281779e-05, + "loss": 0.2837, + "step": 10830 + }, + { + "epoch": 0.39392397703321463, + "grad_norm": 2.818368434906006, + "learning_rate": 3.936332582309761e-05, + "loss": 0.2298, + "step": 10840 + }, + { + "epoch": 0.3942873755360128, + "grad_norm": 6.08942174911499, + "learning_rate": 3.939966567337743e-05, + "loss": 0.2262, + "step": 10850 + }, + { + "epoch": 0.39465077403881094, + "grad_norm": 1.2632570266723633, + "learning_rate": 3.943600552365724e-05, + "loss": 0.2087, + "step": 10860 + }, + { + "epoch": 0.39501417254160914, + "grad_norm": 2.2119662761688232, + "learning_rate": 3.947234537393706e-05, + "loss": 0.1974, + "step": 10870 + }, + { + "epoch": 0.3953775710444073, + "grad_norm": 2.936021089553833, + "learning_rate": 3.950868522421688e-05, + "loss": 0.1909, + "step": 10880 + }, + { + "epoch": 0.39574096954720545, + "grad_norm": 1.3898749351501465, + "learning_rate": 3.9545025074496693e-05, + "loss": 0.184, + "step": 10890 + }, + { + "epoch": 0.39610436805000365, + "grad_norm": 9.063791275024414, + "learning_rate": 3.958136492477651e-05, + "loss": 0.338, + "step": 10900 + }, + { + "epoch": 0.3964677665528018, + "grad_norm": 1.3791584968566895, + "learning_rate": 3.961770477505633e-05, + "loss": 0.2256, + "step": 10910 + }, + { + "epoch": 0.39683116505559995, + "grad_norm": 0.9377845525741577, + "learning_rate": 3.9654044625336145e-05, + "loss": 0.9822, + "step": 10920 + }, + { + "epoch": 0.39719456355839816, + "grad_norm": 3.9755465984344482, + "learning_rate": 3.969038447561596e-05, + "loss": 0.2257, + "step": 10930 + }, + { + "epoch": 0.3975579620611963, + "grad_norm": 1.559699535369873, + "learning_rate": 3.9726724325895783e-05, + "loss": 0.2116, + "step": 10940 + }, + { + "epoch": 0.39792136056399446, + "grad_norm": 7.545668601989746, + "learning_rate": 3.9763064176175596e-05, + "loss": 0.2515, + "step": 10950 + }, + { + "epoch": 0.39828475906679267, + "grad_norm": 1.980197548866272, + "learning_rate": 3.979940402645541e-05, + "loss": 0.1721, + "step": 10960 + }, + { + "epoch": 0.3986481575695908, + "grad_norm": 2.5450973510742188, + "learning_rate": 3.9835743876735235e-05, + "loss": 1.7152, + "step": 10970 + }, + { + "epoch": 0.399011556072389, + "grad_norm": 3.518233060836792, + "learning_rate": 3.987208372701505e-05, + "loss": 0.2521, + "step": 10980 + }, + { + "epoch": 0.3993749545751871, + "grad_norm": 2.678774356842041, + "learning_rate": 3.990842357729486e-05, + "loss": 0.2025, + "step": 10990 + }, + { + "epoch": 0.39973835307798533, + "grad_norm": 11.46552848815918, + "learning_rate": 3.9944763427574686e-05, + "loss": 0.2683, + "step": 11000 + }, + { + "epoch": 0.4001017515807835, + "grad_norm": 2.3148844242095947, + "learning_rate": 3.99811032778545e-05, + "loss": 1.5331, + "step": 11010 + }, + { + "epoch": 0.40046515008358163, + "grad_norm": 1.2145686149597168, + "learning_rate": 4.001744312813431e-05, + "loss": 0.1931, + "step": 11020 + }, + { + "epoch": 0.40082854858637984, + "grad_norm": 3.581883192062378, + "learning_rate": 4.005378297841413e-05, + "loss": 0.18, + "step": 11030 + }, + { + "epoch": 0.401191947089178, + "grad_norm": 2.4645683765411377, + "learning_rate": 4.009012282869395e-05, + "loss": 0.2303, + "step": 11040 + }, + { + "epoch": 0.40155534559197614, + "grad_norm": 13.845566749572754, + "learning_rate": 4.012646267897376e-05, + "loss": 0.2515, + "step": 11050 + }, + { + "epoch": 0.40191874409477435, + "grad_norm": 1.6929864883422852, + "learning_rate": 4.016280252925358e-05, + "loss": 2.9232, + "step": 11060 + }, + { + "epoch": 0.4022821425975725, + "grad_norm": 1.5453213453292847, + "learning_rate": 4.01991423795334e-05, + "loss": 0.1703, + "step": 11070 + }, + { + "epoch": 0.40264554110037065, + "grad_norm": 1.5723987817764282, + "learning_rate": 4.0235482229813214e-05, + "loss": 0.1694, + "step": 11080 + }, + { + "epoch": 0.40300893960316886, + "grad_norm": 1.4501444101333618, + "learning_rate": 4.027182208009303e-05, + "loss": 0.2477, + "step": 11090 + }, + { + "epoch": 0.403372338105967, + "grad_norm": 20.50950813293457, + "learning_rate": 4.030816193037285e-05, + "loss": 0.2641, + "step": 11100 + }, + { + "epoch": 0.40373573660876516, + "grad_norm": 1.9846757650375366, + "learning_rate": 4.0344501780652665e-05, + "loss": 0.1807, + "step": 11110 + }, + { + "epoch": 0.40409913511156337, + "grad_norm": 1.3933240175247192, + "learning_rate": 4.0380841630932484e-05, + "loss": 0.1683, + "step": 11120 + }, + { + "epoch": 0.4044625336143615, + "grad_norm": 2.370534658432007, + "learning_rate": 4.0417181481212304e-05, + "loss": 0.2476, + "step": 11130 + }, + { + "epoch": 0.40482593211715967, + "grad_norm": 2.6382100582122803, + "learning_rate": 4.0453521331492116e-05, + "loss": 0.1723, + "step": 11140 + }, + { + "epoch": 0.4051893306199578, + "grad_norm": 27.381826400756836, + "learning_rate": 4.048986118177193e-05, + "loss": 0.2058, + "step": 11150 + }, + { + "epoch": 0.40555272912275603, + "grad_norm": 1.3622616529464722, + "learning_rate": 4.052620103205175e-05, + "loss": 0.1744, + "step": 11160 + }, + { + "epoch": 0.4059161276255542, + "grad_norm": 1.4734828472137451, + "learning_rate": 4.056254088233157e-05, + "loss": 0.1685, + "step": 11170 + }, + { + "epoch": 0.40627952612835233, + "grad_norm": 5.694312572479248, + "learning_rate": 4.059888073261138e-05, + "loss": 0.3549, + "step": 11180 + }, + { + "epoch": 0.40664292463115054, + "grad_norm": 1.9976438283920288, + "learning_rate": 4.06352205828912e-05, + "loss": 0.1525, + "step": 11190 + }, + { + "epoch": 0.4070063231339487, + "grad_norm": 5.735686779022217, + "learning_rate": 4.067156043317102e-05, + "loss": 0.2642, + "step": 11200 + }, + { + "epoch": 0.40736972163674684, + "grad_norm": 5.192315101623535, + "learning_rate": 4.070790028345083e-05, + "loss": 0.1636, + "step": 11210 + }, + { + "epoch": 0.40773312013954505, + "grad_norm": 2.6324477195739746, + "learning_rate": 4.074424013373065e-05, + "loss": 0.3451, + "step": 11220 + }, + { + "epoch": 0.4080965186423432, + "grad_norm": 2.496997356414795, + "learning_rate": 4.078057998401047e-05, + "loss": 0.1792, + "step": 11230 + }, + { + "epoch": 0.40845991714514135, + "grad_norm": 3.928255558013916, + "learning_rate": 4.081691983429028e-05, + "loss": 0.2203, + "step": 11240 + }, + { + "epoch": 0.40882331564793956, + "grad_norm": 14.433273315429688, + "learning_rate": 4.08532596845701e-05, + "loss": 0.3283, + "step": 11250 + }, + { + "epoch": 0.4091867141507377, + "grad_norm": 1.9282217025756836, + "learning_rate": 4.088959953484992e-05, + "loss": 0.2191, + "step": 11260 + }, + { + "epoch": 0.40955011265353586, + "grad_norm": 1.8360569477081299, + "learning_rate": 4.0925939385129734e-05, + "loss": 0.1623, + "step": 11270 + }, + { + "epoch": 0.409913511156334, + "grad_norm": 4.518060207366943, + "learning_rate": 4.096227923540955e-05, + "loss": 0.2036, + "step": 11280 + }, + { + "epoch": 0.4102769096591322, + "grad_norm": 1.4292632341384888, + "learning_rate": 4.099861908568937e-05, + "loss": 0.1515, + "step": 11290 + }, + { + "epoch": 0.41064030816193037, + "grad_norm": 23.795089721679688, + "learning_rate": 4.1034958935969185e-05, + "loss": 0.3228, + "step": 11300 + }, + { + "epoch": 0.4110037066647285, + "grad_norm": 1.7721456289291382, + "learning_rate": 4.1071298786249005e-05, + "loss": 0.15, + "step": 11310 + }, + { + "epoch": 0.4113671051675267, + "grad_norm": 3.544579029083252, + "learning_rate": 4.110763863652882e-05, + "loss": 0.2349, + "step": 11320 + }, + { + "epoch": 0.4117305036703249, + "grad_norm": 4.25554895401001, + "learning_rate": 4.1143978486808636e-05, + "loss": 0.5458, + "step": 11330 + }, + { + "epoch": 0.41209390217312303, + "grad_norm": 3.069894313812256, + "learning_rate": 4.1180318337088456e-05, + "loss": 0.2131, + "step": 11340 + }, + { + "epoch": 0.41245730067592123, + "grad_norm": 5.389547348022461, + "learning_rate": 4.121665818736827e-05, + "loss": 0.2895, + "step": 11350 + }, + { + "epoch": 0.4128206991787194, + "grad_norm": 2.308717727661133, + "learning_rate": 4.125299803764809e-05, + "loss": 0.209, + "step": 11360 + }, + { + "epoch": 0.41318409768151754, + "grad_norm": 2.071504831314087, + "learning_rate": 4.12893378879279e-05, + "loss": 0.225, + "step": 11370 + }, + { + "epoch": 0.41354749618431574, + "grad_norm": 10.397724151611328, + "learning_rate": 4.132567773820772e-05, + "loss": 0.5041, + "step": 11380 + }, + { + "epoch": 0.4139108946871139, + "grad_norm": 3.3916842937469482, + "learning_rate": 4.136201758848754e-05, + "loss": 0.2055, + "step": 11390 + }, + { + "epoch": 0.41427429318991205, + "grad_norm": 27.703519821166992, + "learning_rate": 4.139835743876735e-05, + "loss": 0.3002, + "step": 11400 + }, + { + "epoch": 0.41427429318991205, + "eval_loss": 0.40216270089149475, + "eval_runtime": 180.385, + "eval_samples_per_second": 41.101, + "eval_steps_per_second": 5.139, + "eval_wer": 0.22132263510447112, + "step": 11400 + }, + { + "epoch": 0.41463769169271025, + "grad_norm": 3.024658203125, + "learning_rate": 4.143469728904717e-05, + "loss": 0.168, + "step": 11410 + }, + { + "epoch": 0.4150010901955084, + "grad_norm": 2.899369478225708, + "learning_rate": 4.147103713932699e-05, + "loss": 3.0252, + "step": 11420 + }, + { + "epoch": 0.41536448869830656, + "grad_norm": 3.960700511932373, + "learning_rate": 4.15073769896068e-05, + "loss": 0.1972, + "step": 11430 + }, + { + "epoch": 0.4157278872011047, + "grad_norm": 1.542468786239624, + "learning_rate": 4.154371683988662e-05, + "loss": 0.1971, + "step": 11440 + }, + { + "epoch": 0.4160912857039029, + "grad_norm": 16.871423721313477, + "learning_rate": 4.158005669016644e-05, + "loss": 0.2768, + "step": 11450 + }, + { + "epoch": 0.41645468420670106, + "grad_norm": 3.142385721206665, + "learning_rate": 4.1616396540446254e-05, + "loss": 0.2173, + "step": 11460 + }, + { + "epoch": 0.4168180827094992, + "grad_norm": 0.9852932095527649, + "learning_rate": 4.1652736390726074e-05, + "loss": 0.2529, + "step": 11470 + }, + { + "epoch": 0.4171814812122974, + "grad_norm": 2.4834413528442383, + "learning_rate": 4.1689076241005886e-05, + "loss": 0.9175, + "step": 11480 + }, + { + "epoch": 0.4175448797150956, + "grad_norm": 2.7286272048950195, + "learning_rate": 4.1725416091285705e-05, + "loss": 0.1864, + "step": 11490 + }, + { + "epoch": 0.4179082782178937, + "grad_norm": 5.711360454559326, + "learning_rate": 4.1761755941565525e-05, + "loss": 0.2285, + "step": 11500 + }, + { + "epoch": 0.41827167672069193, + "grad_norm": 1.160866379737854, + "learning_rate": 4.179809579184534e-05, + "loss": 0.1959, + "step": 11510 + }, + { + "epoch": 0.4186350752234901, + "grad_norm": 2.5051305294036865, + "learning_rate": 4.183443564212516e-05, + "loss": 0.185, + "step": 11520 + }, + { + "epoch": 0.41899847372628823, + "grad_norm": 3.641874313354492, + "learning_rate": 4.1870775492404976e-05, + "loss": 0.199, + "step": 11530 + }, + { + "epoch": 0.41936187222908644, + "grad_norm": 1.676038146018982, + "learning_rate": 4.190711534268479e-05, + "loss": 0.1895, + "step": 11540 + }, + { + "epoch": 0.4197252707318846, + "grad_norm": 11.47658634185791, + "learning_rate": 4.194345519296461e-05, + "loss": 0.246, + "step": 11550 + }, + { + "epoch": 0.42008866923468274, + "grad_norm": 1.7632570266723633, + "learning_rate": 4.197979504324443e-05, + "loss": 0.1761, + "step": 11560 + }, + { + "epoch": 0.4204520677374809, + "grad_norm": 2.2994728088378906, + "learning_rate": 4.201613489352424e-05, + "loss": 0.1799, + "step": 11570 + }, + { + "epoch": 0.4208154662402791, + "grad_norm": 3.964228391647339, + "learning_rate": 4.205247474380406e-05, + "loss": 0.7376, + "step": 11580 + }, + { + "epoch": 0.42117886474307725, + "grad_norm": 1.866466760635376, + "learning_rate": 4.208881459408387e-05, + "loss": 0.2293, + "step": 11590 + }, + { + "epoch": 0.4215422632458754, + "grad_norm": 4.722428798675537, + "learning_rate": 4.212515444436369e-05, + "loss": 0.2303, + "step": 11600 + }, + { + "epoch": 0.4219056617486736, + "grad_norm": 2.8812968730926514, + "learning_rate": 4.2161494294643504e-05, + "loss": 0.1628, + "step": 11610 + }, + { + "epoch": 0.42226906025147176, + "grad_norm": 8.05451488494873, + "learning_rate": 4.219783414492332e-05, + "loss": 0.1978, + "step": 11620 + }, + { + "epoch": 0.4226324587542699, + "grad_norm": 3.4176700115203857, + "learning_rate": 4.223417399520314e-05, + "loss": 0.1986, + "step": 11630 + }, + { + "epoch": 0.4229958572570681, + "grad_norm": 5.204764366149902, + "learning_rate": 4.2270513845482955e-05, + "loss": 0.1959, + "step": 11640 + }, + { + "epoch": 0.42335925575986627, + "grad_norm": 6.184700965881348, + "learning_rate": 4.2306853695762774e-05, + "loss": 0.2822, + "step": 11650 + }, + { + "epoch": 0.4237226542626644, + "grad_norm": 2.288935422897339, + "learning_rate": 4.2343193546042594e-05, + "loss": 0.2073, + "step": 11660 + }, + { + "epoch": 0.42408605276546263, + "grad_norm": 3.8856844902038574, + "learning_rate": 4.2379533396322406e-05, + "loss": 0.2134, + "step": 11670 + }, + { + "epoch": 0.4244494512682608, + "grad_norm": 4.048069953918457, + "learning_rate": 4.2415873246602226e-05, + "loss": 0.1922, + "step": 11680 + }, + { + "epoch": 0.42481284977105893, + "grad_norm": 1.466927409172058, + "learning_rate": 4.2452213096882045e-05, + "loss": 0.1653, + "step": 11690 + }, + { + "epoch": 0.42517624827385714, + "grad_norm": 35.94015121459961, + "learning_rate": 4.248855294716186e-05, + "loss": 0.2398, + "step": 11700 + }, + { + "epoch": 0.4255396467766553, + "grad_norm": 2.575195789337158, + "learning_rate": 4.252489279744168e-05, + "loss": 0.2241, + "step": 11710 + }, + { + "epoch": 0.42590304527945344, + "grad_norm": 1.4232568740844727, + "learning_rate": 4.2561232647721496e-05, + "loss": 0.1817, + "step": 11720 + }, + { + "epoch": 0.4262664437822516, + "grad_norm": 2.8543412685394287, + "learning_rate": 4.259757249800131e-05, + "loss": 0.2094, + "step": 11730 + }, + { + "epoch": 0.4266298422850498, + "grad_norm": 0.85033118724823, + "learning_rate": 4.263391234828113e-05, + "loss": 0.1578, + "step": 11740 + }, + { + "epoch": 0.42699324078784795, + "grad_norm": 7.382369041442871, + "learning_rate": 4.267025219856095e-05, + "loss": 0.2763, + "step": 11750 + }, + { + "epoch": 0.4273566392906461, + "grad_norm": 1.3994635343551636, + "learning_rate": 4.270659204884076e-05, + "loss": 0.199, + "step": 11760 + }, + { + "epoch": 0.4277200377934443, + "grad_norm": 1.4978888034820557, + "learning_rate": 4.274293189912057e-05, + "loss": 0.1804, + "step": 11770 + }, + { + "epoch": 0.42808343629624246, + "grad_norm": 5.206210136413574, + "learning_rate": 4.27792717494004e-05, + "loss": 0.2483, + "step": 11780 + }, + { + "epoch": 0.4284468347990406, + "grad_norm": 1.4130820035934448, + "learning_rate": 4.281561159968021e-05, + "loss": 0.1792, + "step": 11790 + }, + { + "epoch": 0.4288102333018388, + "grad_norm": 2.60227370262146, + "learning_rate": 4.2851951449960024e-05, + "loss": 0.214, + "step": 11800 + }, + { + "epoch": 0.42917363180463697, + "grad_norm": 1.8874465227127075, + "learning_rate": 4.288829130023985e-05, + "loss": 0.1894, + "step": 11810 + }, + { + "epoch": 0.4295370303074351, + "grad_norm": 2.921766519546509, + "learning_rate": 4.292463115051966e-05, + "loss": 0.1608, + "step": 11820 + }, + { + "epoch": 0.4299004288102333, + "grad_norm": 2.812821626663208, + "learning_rate": 4.2960971000799475e-05, + "loss": 0.2381, + "step": 11830 + }, + { + "epoch": 0.4302638273130315, + "grad_norm": 1.8063637018203735, + "learning_rate": 4.2997310851079295e-05, + "loss": 0.198, + "step": 11840 + }, + { + "epoch": 0.43062722581582963, + "grad_norm": 16.433927536010742, + "learning_rate": 4.3033650701359114e-05, + "loss": 0.3015, + "step": 11850 + }, + { + "epoch": 0.4309906243186278, + "grad_norm": 1.295142650604248, + "learning_rate": 4.3069990551638927e-05, + "loss": 0.1678, + "step": 11860 + }, + { + "epoch": 0.431354022821426, + "grad_norm": 135.4871063232422, + "learning_rate": 4.3106330401918746e-05, + "loss": 1.8542, + "step": 11870 + }, + { + "epoch": 0.43171742132422414, + "grad_norm": 2.3314764499664307, + "learning_rate": 4.3142670252198565e-05, + "loss": 0.1983, + "step": 11880 + }, + { + "epoch": 0.4320808198270223, + "grad_norm": 1.6635117530822754, + "learning_rate": 4.317901010247838e-05, + "loss": 0.1737, + "step": 11890 + }, + { + "epoch": 0.4324442183298205, + "grad_norm": 32.102664947509766, + "learning_rate": 4.32153499527582e-05, + "loss": 0.3092, + "step": 11900 + }, + { + "epoch": 0.43280761683261865, + "grad_norm": 2.3491451740264893, + "learning_rate": 4.3251689803038017e-05, + "loss": 0.1849, + "step": 11910 + }, + { + "epoch": 0.4331710153354168, + "grad_norm": 3.8088629245758057, + "learning_rate": 4.328802965331783e-05, + "loss": 0.2023, + "step": 11920 + }, + { + "epoch": 0.433534413838215, + "grad_norm": 2.7132246494293213, + "learning_rate": 4.332436950359764e-05, + "loss": 0.1935, + "step": 11930 + }, + { + "epoch": 0.43389781234101316, + "grad_norm": 1.2917368412017822, + "learning_rate": 4.336070935387747e-05, + "loss": 0.1918, + "step": 11940 + }, + { + "epoch": 0.4342612108438113, + "grad_norm": 9.690601348876953, + "learning_rate": 4.339704920415728e-05, + "loss": 0.3059, + "step": 11950 + }, + { + "epoch": 0.4346246093466095, + "grad_norm": 1.2652380466461182, + "learning_rate": 4.343338905443709e-05, + "loss": 0.1587, + "step": 11960 + }, + { + "epoch": 0.43498800784940767, + "grad_norm": 0.9622058272361755, + "learning_rate": 4.346972890471692e-05, + "loss": 0.1755, + "step": 11970 + }, + { + "epoch": 0.4353514063522058, + "grad_norm": 5.316989898681641, + "learning_rate": 4.350606875499673e-05, + "loss": 0.1794, + "step": 11980 + }, + { + "epoch": 0.435714804855004, + "grad_norm": 3.428891181945801, + "learning_rate": 4.3542408605276544e-05, + "loss": 0.2105, + "step": 11990 + }, + { + "epoch": 0.4360782033578022, + "grad_norm": 12.879768371582031, + "learning_rate": 4.357874845555637e-05, + "loss": 0.2904, + "step": 12000 + }, + { + "epoch": 0.4360782033578022, + "eval_loss": 0.3918191194534302, + "eval_runtime": 180.0676, + "eval_samples_per_second": 41.173, + "eval_steps_per_second": 5.148, + "eval_wer": 0.22659611160527893, + "step": 12000 + }, + { + "epoch": 0.4364416018606003, + "grad_norm": 2.0471973419189453, + "learning_rate": 4.361508830583618e-05, + "loss": 0.2544, + "step": 12010 + }, + { + "epoch": 0.4368050003633985, + "grad_norm": 1.3883107900619507, + "learning_rate": 4.3651428156115995e-05, + "loss": 0.1957, + "step": 12020 + }, + { + "epoch": 0.4371683988661967, + "grad_norm": 1.786475419998169, + "learning_rate": 4.368776800639582e-05, + "loss": 0.1732, + "step": 12030 + }, + { + "epoch": 0.43753179736899483, + "grad_norm": 3.3099594116210938, + "learning_rate": 4.3724107856675634e-05, + "loss": 0.1871, + "step": 12040 + }, + { + "epoch": 0.437895195871793, + "grad_norm": 9.09699535369873, + "learning_rate": 4.376044770695545e-05, + "loss": 0.2745, + "step": 12050 + }, + { + "epoch": 0.4382585943745912, + "grad_norm": 2.0993807315826416, + "learning_rate": 4.3796787557235266e-05, + "loss": 0.2076, + "step": 12060 + }, + { + "epoch": 0.43862199287738934, + "grad_norm": 27.799428939819336, + "learning_rate": 4.3833127407515085e-05, + "loss": 0.548, + "step": 12070 + }, + { + "epoch": 0.4389853913801875, + "grad_norm": 3.8897557258605957, + "learning_rate": 4.38694672577949e-05, + "loss": 0.1799, + "step": 12080 + }, + { + "epoch": 0.4393487898829857, + "grad_norm": 3.4620189666748047, + "learning_rate": 4.390580710807472e-05, + "loss": 0.1735, + "step": 12090 + }, + { + "epoch": 0.43971218838578385, + "grad_norm": 9.587783813476562, + "learning_rate": 4.394214695835454e-05, + "loss": 0.3344, + "step": 12100 + }, + { + "epoch": 0.440075586888582, + "grad_norm": 1.2581641674041748, + "learning_rate": 4.397848680863435e-05, + "loss": 0.1863, + "step": 12110 + }, + { + "epoch": 0.4404389853913802, + "grad_norm": 1.3624401092529297, + "learning_rate": 4.401482665891417e-05, + "loss": 3.7692, + "step": 12120 + }, + { + "epoch": 0.44080238389417836, + "grad_norm": 2.0099213123321533, + "learning_rate": 4.405116650919399e-05, + "loss": 0.1999, + "step": 12130 + }, + { + "epoch": 0.4411657823969765, + "grad_norm": 2.7499871253967285, + "learning_rate": 4.40875063594738e-05, + "loss": 0.1854, + "step": 12140 + }, + { + "epoch": 0.4415291808997747, + "grad_norm": 6.473042964935303, + "learning_rate": 4.412384620975361e-05, + "loss": 0.2843, + "step": 12150 + }, + { + "epoch": 0.44189257940257287, + "grad_norm": 3.845900535583496, + "learning_rate": 4.416018606003344e-05, + "loss": 0.1747, + "step": 12160 + }, + { + "epoch": 0.442255977905371, + "grad_norm": 1.4052759408950806, + "learning_rate": 4.419652591031325e-05, + "loss": 0.16, + "step": 12170 + }, + { + "epoch": 0.4426193764081692, + "grad_norm": 3.5824673175811768, + "learning_rate": 4.4232865760593064e-05, + "loss": 0.7205, + "step": 12180 + }, + { + "epoch": 0.4429827749109674, + "grad_norm": 1.237358570098877, + "learning_rate": 4.426920561087289e-05, + "loss": 0.2043, + "step": 12190 + }, + { + "epoch": 0.44334617341376553, + "grad_norm": 11.106649398803711, + "learning_rate": 4.43055454611527e-05, + "loss": 0.2537, + "step": 12200 + }, + { + "epoch": 0.4437095719165637, + "grad_norm": 1.4566165208816528, + "learning_rate": 4.4341885311432516e-05, + "loss": 0.174, + "step": 12210 + }, + { + "epoch": 0.4440729704193619, + "grad_norm": 1.4067914485931396, + "learning_rate": 4.4378225161712335e-05, + "loss": 0.1672, + "step": 12220 + }, + { + "epoch": 0.44443636892216004, + "grad_norm": 3.1289005279541016, + "learning_rate": 4.4414565011992154e-05, + "loss": 0.2459, + "step": 12230 + }, + { + "epoch": 0.4447997674249582, + "grad_norm": 1.2487775087356567, + "learning_rate": 4.445090486227197e-05, + "loss": 0.1911, + "step": 12240 + }, + { + "epoch": 0.4451631659277564, + "grad_norm": 4.373108863830566, + "learning_rate": 4.4487244712551786e-05, + "loss": 0.265, + "step": 12250 + }, + { + "epoch": 0.44552656443055455, + "grad_norm": 3.0927655696868896, + "learning_rate": 4.4523584562831606e-05, + "loss": 0.166, + "step": 12260 + }, + { + "epoch": 0.4458899629333527, + "grad_norm": 1.4012075662612915, + "learning_rate": 4.455992441311142e-05, + "loss": 0.1631, + "step": 12270 + }, + { + "epoch": 0.4462533614361509, + "grad_norm": 3.9944920539855957, + "learning_rate": 4.459626426339124e-05, + "loss": 0.2616, + "step": 12280 + }, + { + "epoch": 0.44661675993894906, + "grad_norm": 2.412261962890625, + "learning_rate": 4.463260411367106e-05, + "loss": 0.1963, + "step": 12290 + }, + { + "epoch": 0.4469801584417472, + "grad_norm": 8.601739883422852, + "learning_rate": 4.466894396395087e-05, + "loss": 0.3057, + "step": 12300 + }, + { + "epoch": 0.44734355694454536, + "grad_norm": 2.1279587745666504, + "learning_rate": 4.470528381423069e-05, + "loss": 0.1931, + "step": 12310 + }, + { + "epoch": 0.44770695544734357, + "grad_norm": 2.465534210205078, + "learning_rate": 4.474162366451051e-05, + "loss": 0.1701, + "step": 12320 + }, + { + "epoch": 0.4480703539501417, + "grad_norm": 6.147269248962402, + "learning_rate": 4.477796351479032e-05, + "loss": 0.7176, + "step": 12330 + }, + { + "epoch": 0.44843375245293987, + "grad_norm": 1.6242046356201172, + "learning_rate": 4.481430336507014e-05, + "loss": 0.1769, + "step": 12340 + }, + { + "epoch": 0.4487971509557381, + "grad_norm": 7.065566539764404, + "learning_rate": 4.485064321534995e-05, + "loss": 0.2967, + "step": 12350 + }, + { + "epoch": 0.44916054945853623, + "grad_norm": 1.9389359951019287, + "learning_rate": 4.488698306562977e-05, + "loss": 0.1853, + "step": 12360 + }, + { + "epoch": 0.4495239479613344, + "grad_norm": 1.011250376701355, + "learning_rate": 4.492332291590959e-05, + "loss": 0.2036, + "step": 12370 + }, + { + "epoch": 0.4498873464641326, + "grad_norm": 2.459062099456787, + "learning_rate": 4.4959662766189404e-05, + "loss": 0.1865, + "step": 12380 + }, + { + "epoch": 0.45025074496693074, + "grad_norm": 1.8472875356674194, + "learning_rate": 4.499600261646922e-05, + "loss": 0.2178, + "step": 12390 + }, + { + "epoch": 0.4506141434697289, + "grad_norm": 40.6389045715332, + "learning_rate": 4.5032342466749036e-05, + "loss": 0.2506, + "step": 12400 + }, + { + "epoch": 0.4509775419725271, + "grad_norm": 3.9729344844818115, + "learning_rate": 4.5068682317028855e-05, + "loss": 0.1917, + "step": 12410 + }, + { + "epoch": 0.45134094047532525, + "grad_norm": 1.0262936353683472, + "learning_rate": 4.5105022167308675e-05, + "loss": 0.2115, + "step": 12420 + }, + { + "epoch": 0.4517043389781234, + "grad_norm": 1.5356003046035767, + "learning_rate": 4.514136201758849e-05, + "loss": 0.1907, + "step": 12430 + }, + { + "epoch": 0.4520677374809216, + "grad_norm": 1.3107296228408813, + "learning_rate": 4.5177701867868307e-05, + "loss": 0.195, + "step": 12440 + }, + { + "epoch": 0.45243113598371976, + "grad_norm": 11.025674819946289, + "learning_rate": 4.5214041718148126e-05, + "loss": 0.2794, + "step": 12450 + }, + { + "epoch": 0.4527945344865179, + "grad_norm": 1.8793771266937256, + "learning_rate": 4.525038156842794e-05, + "loss": 0.2143, + "step": 12460 + }, + { + "epoch": 0.45315793298931606, + "grad_norm": 1.6508142948150635, + "learning_rate": 4.528672141870776e-05, + "loss": 0.1863, + "step": 12470 + }, + { + "epoch": 0.45352133149211427, + "grad_norm": 4.942420959472656, + "learning_rate": 4.532306126898758e-05, + "loss": 0.1997, + "step": 12480 + }, + { + "epoch": 0.4538847299949124, + "grad_norm": 3.1977925300598145, + "learning_rate": 4.535940111926739e-05, + "loss": 1.9163, + "step": 12490 + }, + { + "epoch": 0.45424812849771057, + "grad_norm": 8.74572467803955, + "learning_rate": 4.539574096954721e-05, + "loss": 0.3186, + "step": 12500 + }, + { + "epoch": 0.4546115270005088, + "grad_norm": 1.5346311330795288, + "learning_rate": 4.543208081982702e-05, + "loss": 0.1958, + "step": 12510 + }, + { + "epoch": 0.4549749255033069, + "grad_norm": 1.622859239578247, + "learning_rate": 4.546842067010684e-05, + "loss": 0.1828, + "step": 12520 + }, + { + "epoch": 0.4553383240061051, + "grad_norm": 1.9394720792770386, + "learning_rate": 4.550476052038666e-05, + "loss": 0.2198, + "step": 12530 + }, + { + "epoch": 0.4557017225089033, + "grad_norm": 1.8405578136444092, + "learning_rate": 4.554110037066647e-05, + "loss": 0.1789, + "step": 12540 + }, + { + "epoch": 0.45606512101170144, + "grad_norm": 6.24867582321167, + "learning_rate": 4.557744022094629e-05, + "loss": 0.2593, + "step": 12550 + }, + { + "epoch": 0.4564285195144996, + "grad_norm": 1.6062959432601929, + "learning_rate": 4.561378007122611e-05, + "loss": 0.1665, + "step": 12560 + }, + { + "epoch": 0.4567919180172978, + "grad_norm": 1.1478540897369385, + "learning_rate": 4.5650119921505924e-05, + "loss": 0.1942, + "step": 12570 + }, + { + "epoch": 0.45715531652009594, + "grad_norm": 2.0299808979034424, + "learning_rate": 4.5686459771785744e-05, + "loss": 0.2092, + "step": 12580 + }, + { + "epoch": 0.4575187150228941, + "grad_norm": 1.6643180847167969, + "learning_rate": 4.572279962206556e-05, + "loss": 0.1714, + "step": 12590 + }, + { + "epoch": 0.45788211352569225, + "grad_norm": 10.169012069702148, + "learning_rate": 4.5759139472345376e-05, + "loss": 0.3101, + "step": 12600 + }, + { + "epoch": 0.45788211352569225, + "eval_loss": 0.408176064491272, + "eval_runtime": 179.7843, + "eval_samples_per_second": 41.238, + "eval_steps_per_second": 5.156, + "eval_wer": 0.24004756113057527, + "step": 12600 + }, + { + "epoch": 0.45824551202849045, + "grad_norm": 1.8151092529296875, + "learning_rate": 4.5795479322625195e-05, + "loss": 0.1739, + "step": 12610 + }, + { + "epoch": 0.4586089105312886, + "grad_norm": 1.1606543064117432, + "learning_rate": 4.583181917290501e-05, + "loss": 0.1781, + "step": 12620 + }, + { + "epoch": 0.45897230903408676, + "grad_norm": 2.5139431953430176, + "learning_rate": 4.586815902318483e-05, + "loss": 0.2101, + "step": 12630 + }, + { + "epoch": 0.45933570753688496, + "grad_norm": 3.1557183265686035, + "learning_rate": 4.590449887346464e-05, + "loss": 0.1925, + "step": 12640 + }, + { + "epoch": 0.4596991060396831, + "grad_norm": 13.978137016296387, + "learning_rate": 4.594083872374446e-05, + "loss": 0.3085, + "step": 12650 + }, + { + "epoch": 0.46006250454248127, + "grad_norm": 1.5187938213348389, + "learning_rate": 4.597717857402428e-05, + "loss": 0.1909, + "step": 12660 + }, + { + "epoch": 0.46042590304527947, + "grad_norm": 1.661890983581543, + "learning_rate": 4.601351842430409e-05, + "loss": 0.1729, + "step": 12670 + }, + { + "epoch": 0.4607893015480776, + "grad_norm": 5.693175792694092, + "learning_rate": 4.604985827458391e-05, + "loss": 0.2069, + "step": 12680 + }, + { + "epoch": 0.4611527000508758, + "grad_norm": 2.5228755474090576, + "learning_rate": 4.608619812486373e-05, + "loss": 0.1899, + "step": 12690 + }, + { + "epoch": 0.461516098553674, + "grad_norm": 12.629317283630371, + "learning_rate": 4.612253797514354e-05, + "loss": 0.2441, + "step": 12700 + }, + { + "epoch": 0.46187949705647213, + "grad_norm": 1.5003726482391357, + "learning_rate": 4.615887782542336e-05, + "loss": 0.1845, + "step": 12710 + }, + { + "epoch": 0.4622428955592703, + "grad_norm": 1.596705675125122, + "learning_rate": 4.619521767570318e-05, + "loss": 0.1942, + "step": 12720 + }, + { + "epoch": 0.4626062940620685, + "grad_norm": 4.299325466156006, + "learning_rate": 4.623155752598299e-05, + "loss": 0.1881, + "step": 12730 + }, + { + "epoch": 0.46296969256486664, + "grad_norm": 2.242932081222534, + "learning_rate": 4.626789737626281e-05, + "loss": 0.1655, + "step": 12740 + }, + { + "epoch": 0.4633330910676648, + "grad_norm": 17.353313446044922, + "learning_rate": 4.630423722654263e-05, + "loss": 0.3002, + "step": 12750 + }, + { + "epoch": 0.46369648957046294, + "grad_norm": 1.8967528343200684, + "learning_rate": 4.6340577076822444e-05, + "loss": 0.1967, + "step": 12760 + }, + { + "epoch": 0.46405988807326115, + "grad_norm": 1.9839125871658325, + "learning_rate": 4.6376916927102264e-05, + "loss": 0.1582, + "step": 12770 + }, + { + "epoch": 0.4644232865760593, + "grad_norm": 1.8139293193817139, + "learning_rate": 4.641325677738208e-05, + "loss": 0.2527, + "step": 12780 + }, + { + "epoch": 0.46478668507885745, + "grad_norm": 1.6944659948349, + "learning_rate": 4.6449596627661896e-05, + "loss": 0.1656, + "step": 12790 + }, + { + "epoch": 0.46515008358165566, + "grad_norm": 3.7842020988464355, + "learning_rate": 4.648593647794171e-05, + "loss": 0.2375, + "step": 12800 + }, + { + "epoch": 0.4655134820844538, + "grad_norm": 1.8103773593902588, + "learning_rate": 4.6522276328221534e-05, + "loss": 0.1932, + "step": 12810 + }, + { + "epoch": 0.46587688058725196, + "grad_norm": 1.4419440031051636, + "learning_rate": 4.655861617850135e-05, + "loss": 0.1808, + "step": 12820 + }, + { + "epoch": 0.46624027909005017, + "grad_norm": 6.361825466156006, + "learning_rate": 4.659495602878116e-05, + "loss": 0.2105, + "step": 12830 + }, + { + "epoch": 0.4666036775928483, + "grad_norm": 1.4687098264694214, + "learning_rate": 4.663129587906098e-05, + "loss": 0.18, + "step": 12840 + }, + { + "epoch": 0.46696707609564647, + "grad_norm": 14.758776664733887, + "learning_rate": 4.66676357293408e-05, + "loss": 0.3001, + "step": 12850 + }, + { + "epoch": 0.4673304745984447, + "grad_norm": 1.4836699962615967, + "learning_rate": 4.670397557962061e-05, + "loss": 0.1713, + "step": 12860 + }, + { + "epoch": 0.46769387310124283, + "grad_norm": 4.860133171081543, + "learning_rate": 4.674031542990043e-05, + "loss": 0.1791, + "step": 12870 + }, + { + "epoch": 0.468057271604041, + "grad_norm": 1.9861228466033936, + "learning_rate": 4.677665528018025e-05, + "loss": 0.2029, + "step": 12880 + }, + { + "epoch": 0.46842067010683913, + "grad_norm": 1.9190025329589844, + "learning_rate": 4.681299513046006e-05, + "loss": 0.1611, + "step": 12890 + }, + { + "epoch": 0.46878406860963734, + "grad_norm": 4.6381516456604, + "learning_rate": 4.684933498073988e-05, + "loss": 0.2646, + "step": 12900 + }, + { + "epoch": 0.4691474671124355, + "grad_norm": 1.2092620134353638, + "learning_rate": 4.68856748310197e-05, + "loss": 0.1865, + "step": 12910 + }, + { + "epoch": 0.46951086561523364, + "grad_norm": 2.7816121578216553, + "learning_rate": 4.6922014681299513e-05, + "loss": 0.2047, + "step": 12920 + }, + { + "epoch": 0.46987426411803185, + "grad_norm": 0.629324734210968, + "learning_rate": 4.6958354531579326e-05, + "loss": 0.2404, + "step": 12930 + }, + { + "epoch": 0.47023766262083, + "grad_norm": 4.156667232513428, + "learning_rate": 4.699469438185915e-05, + "loss": 0.1604, + "step": 12940 + }, + { + "epoch": 0.47060106112362815, + "grad_norm": 1.8534492254257202, + "learning_rate": 4.7031034232138965e-05, + "loss": 0.2364, + "step": 12950 + }, + { + "epoch": 0.47096445962642636, + "grad_norm": 1.382408857345581, + "learning_rate": 4.706737408241878e-05, + "loss": 0.2078, + "step": 12960 + }, + { + "epoch": 0.4713278581292245, + "grad_norm": 2.499023914337158, + "learning_rate": 4.7103713932698603e-05, + "loss": 0.1935, + "step": 12970 + }, + { + "epoch": 0.47169125663202266, + "grad_norm": 2.726032257080078, + "learning_rate": 4.7140053782978416e-05, + "loss": 0.2143, + "step": 12980 + }, + { + "epoch": 0.47205465513482087, + "grad_norm": 2.1388118267059326, + "learning_rate": 4.717639363325823e-05, + "loss": 0.1704, + "step": 12990 + }, + { + "epoch": 0.472418053637619, + "grad_norm": 5.408501148223877, + "learning_rate": 4.7212733483538055e-05, + "loss": 0.2492, + "step": 13000 + }, + { + "epoch": 0.47278145214041717, + "grad_norm": 1.8640841245651245, + "learning_rate": 4.724907333381787e-05, + "loss": 0.1958, + "step": 13010 + }, + { + "epoch": 0.4731448506432154, + "grad_norm": 1.4251651763916016, + "learning_rate": 4.728541318409768e-05, + "loss": 0.1969, + "step": 13020 + }, + { + "epoch": 0.4735082491460135, + "grad_norm": 2.2603137493133545, + "learning_rate": 4.7321753034377506e-05, + "loss": 0.1879, + "step": 13030 + }, + { + "epoch": 0.4738716476488117, + "grad_norm": 1.7813081741333008, + "learning_rate": 4.735809288465732e-05, + "loss": 0.1627, + "step": 13040 + }, + { + "epoch": 0.47423504615160983, + "grad_norm": 16.746126174926758, + "learning_rate": 4.739443273493713e-05, + "loss": 0.3058, + "step": 13050 + }, + { + "epoch": 0.47459844465440804, + "grad_norm": 2.56193470954895, + "learning_rate": 4.743077258521695e-05, + "loss": 0.1729, + "step": 13060 + }, + { + "epoch": 0.4749618431572062, + "grad_norm": 2.1787185668945312, + "learning_rate": 4.746711243549677e-05, + "loss": 0.1804, + "step": 13070 + }, + { + "epoch": 0.47532524166000434, + "grad_norm": 3.385338544845581, + "learning_rate": 4.750345228577658e-05, + "loss": 0.1884, + "step": 13080 + }, + { + "epoch": 0.47568864016280255, + "grad_norm": 2.48083233833313, + "learning_rate": 4.75397921360564e-05, + "loss": 0.1728, + "step": 13090 + }, + { + "epoch": 0.4760520386656007, + "grad_norm": 47.18072509765625, + "learning_rate": 4.757613198633622e-05, + "loss": 0.2427, + "step": 13100 + }, + { + "epoch": 0.47641543716839885, + "grad_norm": 1.3267533779144287, + "learning_rate": 4.7612471836616034e-05, + "loss": 0.1847, + "step": 13110 + }, + { + "epoch": 0.47677883567119705, + "grad_norm": 2.098389148712158, + "learning_rate": 4.764881168689585e-05, + "loss": 0.1682, + "step": 13120 + }, + { + "epoch": 0.4771422341739952, + "grad_norm": 1.1197071075439453, + "learning_rate": 4.768515153717567e-05, + "loss": 0.166, + "step": 13130 + }, + { + "epoch": 0.47750563267679336, + "grad_norm": 1.431281328201294, + "learning_rate": 4.7721491387455485e-05, + "loss": 0.3262, + "step": 13140 + }, + { + "epoch": 0.47786903117959156, + "grad_norm": 15.357772827148438, + "learning_rate": 4.7757831237735304e-05, + "loss": 0.2906, + "step": 13150 + }, + { + "epoch": 0.4782324296823897, + "grad_norm": 3.03275465965271, + "learning_rate": 4.7794171088015124e-05, + "loss": 0.207, + "step": 13160 + }, + { + "epoch": 0.47859582818518787, + "grad_norm": 1.0988962650299072, + "learning_rate": 4.7830510938294936e-05, + "loss": 0.1788, + "step": 13170 + }, + { + "epoch": 0.4789592266879861, + "grad_norm": 1.9456548690795898, + "learning_rate": 4.786685078857475e-05, + "loss": 0.2397, + "step": 13180 + }, + { + "epoch": 0.4793226251907842, + "grad_norm": 1.7383311986923218, + "learning_rate": 4.7903190638854575e-05, + "loss": 0.1841, + "step": 13190 + }, + { + "epoch": 0.4796860236935824, + "grad_norm": 5.512730121612549, + "learning_rate": 4.793953048913439e-05, + "loss": 0.2708, + "step": 13200 + }, + { + "epoch": 0.4796860236935824, + "eval_loss": 0.3998795747756958, + "eval_runtime": 180.9114, + "eval_samples_per_second": 40.981, + "eval_steps_per_second": 5.124, + "eval_wer": 0.2369433804708915, + "step": 13200 + }, + { + "epoch": 0.4800494221963805, + "grad_norm": 1.5843122005462646, + "learning_rate": 4.79758703394142e-05, + "loss": 1.0933, + "step": 13210 + }, + { + "epoch": 0.48041282069917873, + "grad_norm": 1.4696934223175049, + "learning_rate": 4.8012210189694026e-05, + "loss": 0.1771, + "step": 13220 + }, + { + "epoch": 0.4807762192019769, + "grad_norm": 2.5620357990264893, + "learning_rate": 4.804855003997384e-05, + "loss": 0.2202, + "step": 13230 + }, + { + "epoch": 0.48113961770477504, + "grad_norm": 73.08427429199219, + "learning_rate": 4.808488989025365e-05, + "loss": 0.2471, + "step": 13240 + }, + { + "epoch": 0.48150301620757324, + "grad_norm": 7.291989803314209, + "learning_rate": 4.812122974053347e-05, + "loss": 0.2542, + "step": 13250 + }, + { + "epoch": 0.4818664147103714, + "grad_norm": 1.7582112550735474, + "learning_rate": 4.815756959081329e-05, + "loss": 0.1884, + "step": 13260 + }, + { + "epoch": 0.48222981321316954, + "grad_norm": 0.9253680109977722, + "learning_rate": 4.81939094410931e-05, + "loss": 0.1797, + "step": 13270 + }, + { + "epoch": 0.48259321171596775, + "grad_norm": 8.042390823364258, + "learning_rate": 4.823024929137292e-05, + "loss": 0.192, + "step": 13280 + }, + { + "epoch": 0.4829566102187659, + "grad_norm": 3.2288219928741455, + "learning_rate": 4.826658914165274e-05, + "loss": 0.2041, + "step": 13290 + }, + { + "epoch": 0.48332000872156405, + "grad_norm": 7.657989978790283, + "learning_rate": 4.8302928991932554e-05, + "loss": 0.3034, + "step": 13300 + }, + { + "epoch": 0.48368340722436226, + "grad_norm": 2.9273271560668945, + "learning_rate": 4.833926884221237e-05, + "loss": 0.2028, + "step": 13310 + }, + { + "epoch": 0.4840468057271604, + "grad_norm": 4.2344865798950195, + "learning_rate": 4.837560869249219e-05, + "loss": 0.1817, + "step": 13320 + }, + { + "epoch": 0.48441020422995856, + "grad_norm": 4.074464797973633, + "learning_rate": 4.8411948542772005e-05, + "loss": 0.2197, + "step": 13330 + }, + { + "epoch": 0.4847736027327567, + "grad_norm": 1.7070029973983765, + "learning_rate": 4.8448288393051825e-05, + "loss": 0.2374, + "step": 13340 + }, + { + "epoch": 0.4851370012355549, + "grad_norm": 2.5278494358062744, + "learning_rate": 4.8484628243331644e-05, + "loss": 0.265, + "step": 13350 + }, + { + "epoch": 0.4855003997383531, + "grad_norm": 1.4800697565078735, + "learning_rate": 4.8520968093611456e-05, + "loss": 0.1597, + "step": 13360 + }, + { + "epoch": 0.4858637982411512, + "grad_norm": 1.238171935081482, + "learning_rate": 4.8557307943891276e-05, + "loss": 0.1862, + "step": 13370 + }, + { + "epoch": 0.48622719674394943, + "grad_norm": 2.7711944580078125, + "learning_rate": 4.859364779417109e-05, + "loss": 0.1572, + "step": 13380 + }, + { + "epoch": 0.4865905952467476, + "grad_norm": 2.386011838912964, + "learning_rate": 4.862998764445091e-05, + "loss": 0.1624, + "step": 13390 + }, + { + "epoch": 0.48695399374954573, + "grad_norm": 10.38249397277832, + "learning_rate": 4.866632749473072e-05, + "loss": 0.2182, + "step": 13400 + }, + { + "epoch": 0.48731739225234394, + "grad_norm": 1.1541043519973755, + "learning_rate": 4.870266734501054e-05, + "loss": 0.1867, + "step": 13410 + }, + { + "epoch": 0.4876807907551421, + "grad_norm": 0.7680534720420837, + "learning_rate": 4.873900719529036e-05, + "loss": 0.1619, + "step": 13420 + }, + { + "epoch": 0.48804418925794024, + "grad_norm": 2.6120142936706543, + "learning_rate": 4.877534704557017e-05, + "loss": 1.0657, + "step": 13430 + }, + { + "epoch": 0.48840758776073845, + "grad_norm": 2.1559348106384277, + "learning_rate": 4.881168689584999e-05, + "loss": 0.1576, + "step": 13440 + }, + { + "epoch": 0.4887709862635366, + "grad_norm": 8.222488403320312, + "learning_rate": 4.884802674612981e-05, + "loss": 0.2596, + "step": 13450 + }, + { + "epoch": 0.48913438476633475, + "grad_norm": 1.7630010843276978, + "learning_rate": 4.888436659640962e-05, + "loss": 0.1755, + "step": 13460 + }, + { + "epoch": 0.48949778326913296, + "grad_norm": 1.489050269126892, + "learning_rate": 4.892070644668944e-05, + "loss": 0.1844, + "step": 13470 + }, + { + "epoch": 0.4898611817719311, + "grad_norm": 4.412111759185791, + "learning_rate": 4.895704629696926e-05, + "loss": 0.2114, + "step": 13480 + }, + { + "epoch": 0.49022458027472926, + "grad_norm": 2.060366630554199, + "learning_rate": 4.8993386147249074e-05, + "loss": 0.1932, + "step": 13490 + }, + { + "epoch": 0.4905879787775274, + "grad_norm": 9.488603591918945, + "learning_rate": 4.9029725997528893e-05, + "loss": 0.303, + "step": 13500 + }, + { + "epoch": 0.4909513772803256, + "grad_norm": 2.295671224594116, + "learning_rate": 4.906606584780871e-05, + "loss": 0.1583, + "step": 13510 + }, + { + "epoch": 0.49131477578312377, + "grad_norm": 4.13812255859375, + "learning_rate": 4.9102405698088525e-05, + "loss": 1.9041, + "step": 13520 + }, + { + "epoch": 0.4916781742859219, + "grad_norm": 3.7411348819732666, + "learning_rate": 4.9138745548368345e-05, + "loss": 0.1927, + "step": 13530 + }, + { + "epoch": 0.4920415727887201, + "grad_norm": 1.523505449295044, + "learning_rate": 4.917508539864816e-05, + "loss": 0.1721, + "step": 13540 + }, + { + "epoch": 0.4924049712915183, + "grad_norm": 8.239662170410156, + "learning_rate": 4.921142524892798e-05, + "loss": 0.3205, + "step": 13550 + }, + { + "epoch": 0.49276836979431643, + "grad_norm": 1.8316904306411743, + "learning_rate": 4.9247765099207796e-05, + "loss": 0.172, + "step": 13560 + }, + { + "epoch": 0.49313176829711464, + "grad_norm": 4.627805233001709, + "learning_rate": 4.928410494948761e-05, + "loss": 0.1731, + "step": 13570 + }, + { + "epoch": 0.4934951667999128, + "grad_norm": 4.277485370635986, + "learning_rate": 4.932044479976743e-05, + "loss": 0.2522, + "step": 13580 + }, + { + "epoch": 0.49385856530271094, + "grad_norm": 2.131641149520874, + "learning_rate": 4.935678465004725e-05, + "loss": 0.1766, + "step": 13590 + }, + { + "epoch": 0.49422196380550915, + "grad_norm": 2.9195988178253174, + "learning_rate": 4.939312450032706e-05, + "loss": 0.3745, + "step": 13600 + }, + { + "epoch": 0.4945853623083073, + "grad_norm": 1.5876374244689941, + "learning_rate": 4.942946435060688e-05, + "loss": 0.1716, + "step": 13610 + }, + { + "epoch": 0.49494876081110545, + "grad_norm": 4.506389617919922, + "learning_rate": 4.946580420088669e-05, + "loss": 0.5847, + "step": 13620 + }, + { + "epoch": 0.4953121593139036, + "grad_norm": 3.497152090072632, + "learning_rate": 4.950214405116651e-05, + "loss": 0.2179, + "step": 13630 + }, + { + "epoch": 0.4956755578167018, + "grad_norm": 1.7728289365768433, + "learning_rate": 4.953848390144633e-05, + "loss": 0.165, + "step": 13640 + }, + { + "epoch": 0.49603895631949996, + "grad_norm": 12.01921558380127, + "learning_rate": 4.957482375172614e-05, + "loss": 0.2447, + "step": 13650 + }, + { + "epoch": 0.4964023548222981, + "grad_norm": 2.5448553562164307, + "learning_rate": 4.961116360200596e-05, + "loss": 0.2089, + "step": 13660 + }, + { + "epoch": 0.4967657533250963, + "grad_norm": 2.3643887042999268, + "learning_rate": 4.9647503452285775e-05, + "loss": 0.1724, + "step": 13670 + }, + { + "epoch": 0.49712915182789447, + "grad_norm": 2.096191644668579, + "learning_rate": 4.9683843302565594e-05, + "loss": 0.1759, + "step": 13680 + }, + { + "epoch": 0.4974925503306926, + "grad_norm": 0.9760168790817261, + "learning_rate": 4.9720183152845414e-05, + "loss": 0.1817, + "step": 13690 + }, + { + "epoch": 0.4978559488334908, + "grad_norm": 3.019702434539795, + "learning_rate": 4.9756523003125226e-05, + "loss": 0.2275, + "step": 13700 + }, + { + "epoch": 0.498219347336289, + "grad_norm": 1.0820231437683105, + "learning_rate": 4.9789228868377064e-05, + "loss": 2.5822, + "step": 13710 + }, + { + "epoch": 0.4985827458390871, + "grad_norm": 3.2908883094787598, + "learning_rate": 4.982556871865688e-05, + "loss": 0.1898, + "step": 13720 + }, + { + "epoch": 0.49894614434188533, + "grad_norm": 3.4303886890411377, + "learning_rate": 4.98619085689367e-05, + "loss": 0.2295, + "step": 13730 + }, + { + "epoch": 0.4993095428446835, + "grad_norm": 1.8785525560379028, + "learning_rate": 4.9898248419216515e-05, + "loss": 0.1699, + "step": 13740 + }, + { + "epoch": 0.49967294134748164, + "grad_norm": 7.539544105529785, + "learning_rate": 4.993458826949633e-05, + "loss": 0.2955, + "step": 13750 + }, + { + "epoch": 0.5000363398502798, + "grad_norm": 1.6091630458831787, + "learning_rate": 4.997092811977615e-05, + "loss": 0.1696, + "step": 13760 + }, + { + "epoch": 0.5003997383530779, + "grad_norm": 1.023695945739746, + "learning_rate": 4.9999999978456776e-05, + "loss": 0.1872, + "step": 13770 + }, + { + "epoch": 0.5007631368558761, + "grad_norm": 8.364274978637695, + "learning_rate": 4.999999922444405e-05, + "loss": 0.1844, + "step": 13780 + }, + { + "epoch": 0.5011265353586744, + "grad_norm": 1.7257829904556274, + "learning_rate": 4.99999973932703e-05, + "loss": 0.241, + "step": 13790 + }, + { + "epoch": 0.5014899338614724, + "grad_norm": 7.256163597106934, + "learning_rate": 4.999999448493561e-05, + "loss": 0.2714, + "step": 13800 + }, + { + "epoch": 0.5014899338614724, + "eval_loss": 0.4298999607563019, + "eval_runtime": 179.7223, + "eval_samples_per_second": 41.253, + "eval_steps_per_second": 5.158, + "eval_wer": 0.22517109299834806, + "step": 13800 + }, + { + "epoch": 0.5018533323642707, + "grad_norm": 3.087979316711426, + "learning_rate": 4.999999049944011e-05, + "loss": 0.3094, + "step": 13810 + }, + { + "epoch": 0.5022167308670689, + "grad_norm": 1.7626384496688843, + "learning_rate": 4.999998543678397e-05, + "loss": 0.2521, + "step": 13820 + }, + { + "epoch": 0.502580129369867, + "grad_norm": 2.257432699203491, + "learning_rate": 4.999997929696741e-05, + "loss": 0.1913, + "step": 13830 + }, + { + "epoch": 0.5029435278726652, + "grad_norm": 1.7763293981552124, + "learning_rate": 4.999997207999069e-05, + "loss": 0.1812, + "step": 13840 + }, + { + "epoch": 0.5033069263754634, + "grad_norm": 8.228759765625, + "learning_rate": 4.9999963785854124e-05, + "loss": 0.2953, + "step": 13850 + }, + { + "epoch": 0.5036703248782615, + "grad_norm": 1.200305461883545, + "learning_rate": 4.999995441455807e-05, + "loss": 0.3246, + "step": 13860 + }, + { + "epoch": 0.5040337233810597, + "grad_norm": 1.9264732599258423, + "learning_rate": 4.999994396610292e-05, + "loss": 0.1749, + "step": 13870 + }, + { + "epoch": 0.5043971218838579, + "grad_norm": 2.547212839126587, + "learning_rate": 4.999993244048915e-05, + "loss": 0.2714, + "step": 13880 + }, + { + "epoch": 0.504760520386656, + "grad_norm": 2.7918379306793213, + "learning_rate": 4.999991983771723e-05, + "loss": 0.1984, + "step": 13890 + }, + { + "epoch": 0.5051239188894542, + "grad_norm": 16.789764404296875, + "learning_rate": 4.999990615778772e-05, + "loss": 0.239, + "step": 13900 + }, + { + "epoch": 0.5054873173922524, + "grad_norm": 1.1825790405273438, + "learning_rate": 4.9999891400701205e-05, + "loss": 0.1774, + "step": 13910 + }, + { + "epoch": 0.5058507158950505, + "grad_norm": 2.1524746417999268, + "learning_rate": 4.999987556645832e-05, + "loss": 0.1956, + "step": 13920 + }, + { + "epoch": 0.5062141143978487, + "grad_norm": 2.8159048557281494, + "learning_rate": 4.999985865505974e-05, + "loss": 0.2315, + "step": 13930 + }, + { + "epoch": 0.5065775129006469, + "grad_norm": 1.7412035465240479, + "learning_rate": 4.99998406665062e-05, + "loss": 0.2106, + "step": 13940 + }, + { + "epoch": 0.506940911403445, + "grad_norm": 61.967708587646484, + "learning_rate": 4.999982160079848e-05, + "loss": 0.3067, + "step": 13950 + }, + { + "epoch": 0.5073043099062432, + "grad_norm": 2.378682851791382, + "learning_rate": 4.9999801457937404e-05, + "loss": 0.21, + "step": 13960 + }, + { + "epoch": 0.5076677084090414, + "grad_norm": 1.3668854236602783, + "learning_rate": 4.9999780237923824e-05, + "loss": 0.1529, + "step": 13970 + }, + { + "epoch": 0.5080311069118395, + "grad_norm": 2.2655959129333496, + "learning_rate": 4.9999757940758665e-05, + "loss": 0.1747, + "step": 13980 + }, + { + "epoch": 0.5083945054146377, + "grad_norm": 1.5975615978240967, + "learning_rate": 4.9999734566442877e-05, + "loss": 0.1728, + "step": 13990 + }, + { + "epoch": 0.5087579039174358, + "grad_norm": 9.869553565979004, + "learning_rate": 4.999971011497748e-05, + "loss": 0.3207, + "step": 14000 + }, + { + "epoch": 0.509121302420234, + "grad_norm": 2.3095829486846924, + "learning_rate": 4.999968458636353e-05, + "loss": 0.1872, + "step": 14010 + }, + { + "epoch": 0.5094847009230322, + "grad_norm": 2.059575080871582, + "learning_rate": 4.999965798060212e-05, + "loss": 0.3059, + "step": 14020 + }, + { + "epoch": 0.5098480994258303, + "grad_norm": 1.7138803005218506, + "learning_rate": 4.9999630297694395e-05, + "loss": 0.2025, + "step": 14030 + }, + { + "epoch": 0.5102114979286285, + "grad_norm": 2.831191062927246, + "learning_rate": 4.999960153764155e-05, + "loss": 0.1685, + "step": 14040 + }, + { + "epoch": 0.5105748964314267, + "grad_norm": 15.457362174987793, + "learning_rate": 4.999957170044482e-05, + "loss": 0.2165, + "step": 14050 + }, + { + "epoch": 0.5109382949342248, + "grad_norm": 3.923633337020874, + "learning_rate": 4.999954078610549e-05, + "loss": 0.1888, + "step": 14060 + }, + { + "epoch": 0.511301693437023, + "grad_norm": 0.8243936896324158, + "learning_rate": 4.999950879462491e-05, + "loss": 0.295, + "step": 14070 + }, + { + "epoch": 0.5116650919398212, + "grad_norm": 2.921447277069092, + "learning_rate": 4.9999475726004434e-05, + "loss": 0.3208, + "step": 14080 + }, + { + "epoch": 0.5120284904426193, + "grad_norm": 0.9395463466644287, + "learning_rate": 4.99994415802455e-05, + "loss": 0.1936, + "step": 14090 + }, + { + "epoch": 0.5123918889454175, + "grad_norm": 11.025691986083984, + "learning_rate": 4.999940635734958e-05, + "loss": 0.2581, + "step": 14100 + }, + { + "epoch": 0.5127552874482157, + "grad_norm": 2.2102460861206055, + "learning_rate": 4.999937005731818e-05, + "loss": 0.1888, + "step": 14110 + }, + { + "epoch": 0.5131186859510138, + "grad_norm": 1.6075447797775269, + "learning_rate": 4.9999332680152876e-05, + "loss": 0.1557, + "step": 14120 + }, + { + "epoch": 0.513482084453812, + "grad_norm": 3.0174403190612793, + "learning_rate": 4.999929422585528e-05, + "loss": 0.2137, + "step": 14130 + }, + { + "epoch": 0.5138454829566103, + "grad_norm": 3.2911272048950195, + "learning_rate": 4.999925469442705e-05, + "loss": 0.2249, + "step": 14140 + }, + { + "epoch": 0.5142088814594084, + "grad_norm": 4.0001444816589355, + "learning_rate": 4.999921408586986e-05, + "loss": 0.2548, + "step": 14150 + }, + { + "epoch": 0.5145722799622066, + "grad_norm": 2.7695538997650146, + "learning_rate": 4.9999172400185504e-05, + "loss": 0.2107, + "step": 14160 + }, + { + "epoch": 0.5149356784650048, + "grad_norm": 1.420189380645752, + "learning_rate": 4.999912963737574e-05, + "loss": 0.1887, + "step": 14170 + }, + { + "epoch": 0.5152990769678029, + "grad_norm": 1.4330711364746094, + "learning_rate": 4.9999085797442434e-05, + "loss": 0.2295, + "step": 14180 + }, + { + "epoch": 0.5156624754706011, + "grad_norm": 1.9518648386001587, + "learning_rate": 4.999904088038747e-05, + "loss": 0.181, + "step": 14190 + }, + { + "epoch": 0.5160258739733993, + "grad_norm": 9.763446807861328, + "learning_rate": 4.999899488621278e-05, + "loss": 0.2163, + "step": 14200 + }, + { + "epoch": 0.5163892724761974, + "grad_norm": 1.63487708568573, + "learning_rate": 4.999894781492035e-05, + "loss": 0.1675, + "step": 14210 + }, + { + "epoch": 0.5167526709789956, + "grad_norm": 1.3337619304656982, + "learning_rate": 4.99988996665122e-05, + "loss": 1.8258, + "step": 14220 + }, + { + "epoch": 0.5171160694817938, + "grad_norm": 4.741299152374268, + "learning_rate": 4.9998850440990414e-05, + "loss": 0.199, + "step": 14230 + }, + { + "epoch": 0.5174794679845919, + "grad_norm": 2.203994035720825, + "learning_rate": 4.9998800138357106e-05, + "loss": 0.1666, + "step": 14240 + }, + { + "epoch": 0.5178428664873901, + "grad_norm": 9.144301414489746, + "learning_rate": 4.999874875861444e-05, + "loss": 0.2567, + "step": 14250 + }, + { + "epoch": 0.5182062649901883, + "grad_norm": 1.432627558708191, + "learning_rate": 4.9998696301764644e-05, + "loss": 0.1842, + "step": 14260 + }, + { + "epoch": 0.5185696634929864, + "grad_norm": 1.5303106307983398, + "learning_rate": 4.999864276780998e-05, + "loss": 0.1726, + "step": 14270 + }, + { + "epoch": 0.5189330619957846, + "grad_norm": 13.468036651611328, + "learning_rate": 4.999858815675273e-05, + "loss": 0.1927, + "step": 14280 + }, + { + "epoch": 0.5192964604985827, + "grad_norm": 3.7133965492248535, + "learning_rate": 4.999853246859526e-05, + "loss": 0.1822, + "step": 14290 + }, + { + "epoch": 0.5196598590013809, + "grad_norm": 10.077652931213379, + "learning_rate": 4.999847570333998e-05, + "loss": 0.2847, + "step": 14300 + }, + { + "epoch": 0.5200232575041791, + "grad_norm": 2.3906922340393066, + "learning_rate": 4.9998417860989325e-05, + "loss": 0.1962, + "step": 14310 + }, + { + "epoch": 0.5203866560069772, + "grad_norm": 0.8041434288024902, + "learning_rate": 4.999835894154579e-05, + "loss": 0.1661, + "step": 14320 + }, + { + "epoch": 0.5207500545097754, + "grad_norm": 4.1071953773498535, + "learning_rate": 4.99982989450119e-05, + "loss": 0.2012, + "step": 14330 + }, + { + "epoch": 0.5211134530125736, + "grad_norm": 0.9645094871520996, + "learning_rate": 4.999823787139026e-05, + "loss": 0.209, + "step": 14340 + }, + { + "epoch": 0.5214768515153717, + "grad_norm": 19.18789291381836, + "learning_rate": 4.9998175720683506e-05, + "loss": 0.3019, + "step": 14350 + }, + { + "epoch": 0.5218402500181699, + "grad_norm": 1.6560392379760742, + "learning_rate": 4.999811249289429e-05, + "loss": 0.1696, + "step": 14360 + }, + { + "epoch": 0.5222036485209681, + "grad_norm": 1.993741512298584, + "learning_rate": 4.999804818802535e-05, + "loss": 0.1895, + "step": 14370 + }, + { + "epoch": 0.5225670470237662, + "grad_norm": 3.4508492946624756, + "learning_rate": 4.999798280607947e-05, + "loss": 0.2111, + "step": 14380 + }, + { + "epoch": 0.5229304455265644, + "grad_norm": 8.431037902832031, + "learning_rate": 4.999791634705944e-05, + "loss": 0.1898, + "step": 14390 + }, + { + "epoch": 0.5232938440293626, + "grad_norm": 10.659805297851562, + "learning_rate": 4.9997848810968137e-05, + "loss": 0.4744, + "step": 14400 + }, + { + "epoch": 0.5232938440293626, + "eval_loss": 0.40915772318840027, + "eval_runtime": 179.6286, + "eval_samples_per_second": 41.274, + "eval_steps_per_second": 5.161, + "eval_wer": 0.2273222357361991, + "step": 14400 + }, + { + "epoch": 0.5236572425321607, + "grad_norm": 1.676483392715454, + "learning_rate": 4.999778019780849e-05, + "loss": 0.1856, + "step": 14410 + }, + { + "epoch": 0.5240206410349589, + "grad_norm": 3.4859771728515625, + "learning_rate": 4.9997710507583414e-05, + "loss": 0.1641, + "step": 14420 + }, + { + "epoch": 0.5243840395377571, + "grad_norm": 2.583261251449585, + "learning_rate": 4.999763974029595e-05, + "loss": 0.2545, + "step": 14430 + }, + { + "epoch": 0.5247474380405552, + "grad_norm": 2.0467324256896973, + "learning_rate": 4.999756789594913e-05, + "loss": 0.1974, + "step": 14440 + }, + { + "epoch": 0.5251108365433534, + "grad_norm": 4.777310848236084, + "learning_rate": 4.999749497454605e-05, + "loss": 0.2653, + "step": 14450 + }, + { + "epoch": 0.5254742350461516, + "grad_norm": 1.6312458515167236, + "learning_rate": 4.999742097608984e-05, + "loss": 0.1503, + "step": 14460 + }, + { + "epoch": 0.5258376335489497, + "grad_norm": 1.1725629568099976, + "learning_rate": 4.999734590058371e-05, + "loss": 0.1636, + "step": 14470 + }, + { + "epoch": 0.526201032051748, + "grad_norm": 3.2061386108398438, + "learning_rate": 4.999726974803089e-05, + "loss": 0.1988, + "step": 14480 + }, + { + "epoch": 0.5265644305545462, + "grad_norm": 1.7078185081481934, + "learning_rate": 4.9997192518434655e-05, + "loss": 0.1763, + "step": 14490 + }, + { + "epoch": 0.5269278290573443, + "grad_norm": 3.5756313800811768, + "learning_rate": 4.999711421179833e-05, + "loss": 0.2651, + "step": 14500 + }, + { + "epoch": 0.5272912275601425, + "grad_norm": 1.8054040670394897, + "learning_rate": 4.99970348281253e-05, + "loss": 0.1923, + "step": 14510 + }, + { + "epoch": 0.5276546260629407, + "grad_norm": 2.8949921131134033, + "learning_rate": 4.9996954367418976e-05, + "loss": 0.1897, + "step": 14520 + }, + { + "epoch": 0.5280180245657388, + "grad_norm": 2.0020744800567627, + "learning_rate": 4.9996872829682825e-05, + "loss": 0.2469, + "step": 14530 + }, + { + "epoch": 0.528381423068537, + "grad_norm": 1.1650570631027222, + "learning_rate": 4.999679021492037e-05, + "loss": 0.2088, + "step": 14540 + }, + { + "epoch": 0.5287448215713352, + "grad_norm": 14.624237060546875, + "learning_rate": 4.999670652313516e-05, + "loss": 0.2918, + "step": 14550 + }, + { + "epoch": 0.5291082200741333, + "grad_norm": 1.6658445596694946, + "learning_rate": 4.99966217543308e-05, + "loss": 0.1936, + "step": 14560 + }, + { + "epoch": 0.5294716185769315, + "grad_norm": 2.0761842727661133, + "learning_rate": 4.9996535908510955e-05, + "loss": 0.2318, + "step": 14570 + }, + { + "epoch": 0.5298350170797297, + "grad_norm": 2.475193977355957, + "learning_rate": 4.999644898567931e-05, + "loss": 0.1682, + "step": 14580 + }, + { + "epoch": 0.5301984155825278, + "grad_norm": 1.7537975311279297, + "learning_rate": 4.9996360985839616e-05, + "loss": 0.1528, + "step": 14590 + }, + { + "epoch": 0.530561814085326, + "grad_norm": 6.486474990844727, + "learning_rate": 4.9996271908995666e-05, + "loss": 0.2571, + "step": 14600 + }, + { + "epoch": 0.5309252125881241, + "grad_norm": 2.308250665664673, + "learning_rate": 4.9996181755151294e-05, + "loss": 0.1764, + "step": 14610 + }, + { + "epoch": 0.5312886110909223, + "grad_norm": 4.871829032897949, + "learning_rate": 4.999609052431039e-05, + "loss": 0.3045, + "step": 14620 + }, + { + "epoch": 0.5316520095937205, + "grad_norm": 2.796844959259033, + "learning_rate": 4.999599821647688e-05, + "loss": 0.2102, + "step": 14630 + }, + { + "epoch": 0.5320154080965186, + "grad_norm": 1.9681658744812012, + "learning_rate": 4.999590483165475e-05, + "loss": 0.1882, + "step": 14640 + }, + { + "epoch": 0.5323788065993168, + "grad_norm": 5.858233451843262, + "learning_rate": 4.9995810369848006e-05, + "loss": 0.314, + "step": 14650 + }, + { + "epoch": 0.532742205102115, + "grad_norm": 6.469663143157959, + "learning_rate": 4.9995714831060736e-05, + "loss": 0.2103, + "step": 14660 + }, + { + "epoch": 0.5331056036049131, + "grad_norm": 1.8543453216552734, + "learning_rate": 4.999561821529705e-05, + "loss": 0.219, + "step": 14670 + }, + { + "epoch": 0.5334690021077113, + "grad_norm": 2.222320318222046, + "learning_rate": 4.99955205225611e-05, + "loss": 0.1879, + "step": 14680 + }, + { + "epoch": 0.5338324006105095, + "grad_norm": 5.018227577209473, + "learning_rate": 4.999542175285711e-05, + "loss": 0.1437, + "step": 14690 + }, + { + "epoch": 0.5341957991133076, + "grad_norm": 6.225541114807129, + "learning_rate": 4.999532190618933e-05, + "loss": 0.268, + "step": 14700 + }, + { + "epoch": 0.5345591976161058, + "grad_norm": 1.8122676610946655, + "learning_rate": 4.999522098256206e-05, + "loss": 0.1644, + "step": 14710 + }, + { + "epoch": 0.534922596118904, + "grad_norm": 2.4057557582855225, + "learning_rate": 4.999511898197966e-05, + "loss": 0.1663, + "step": 14720 + }, + { + "epoch": 0.5352859946217021, + "grad_norm": 1.756697416305542, + "learning_rate": 4.9995015904446513e-05, + "loss": 0.1771, + "step": 14730 + }, + { + "epoch": 0.5356493931245003, + "grad_norm": 1.5457457304000854, + "learning_rate": 4.999491174996706e-05, + "loss": 0.1889, + "step": 14740 + }, + { + "epoch": 0.5360127916272985, + "grad_norm": 3.108682155609131, + "learning_rate": 4.999480651854579e-05, + "loss": 0.2063, + "step": 14750 + }, + { + "epoch": 0.5363761901300966, + "grad_norm": 2.2037875652313232, + "learning_rate": 4.9994700210187246e-05, + "loss": 0.1579, + "step": 14760 + }, + { + "epoch": 0.5367395886328948, + "grad_norm": 1.2102454900741577, + "learning_rate": 4.9994592824895994e-05, + "loss": 0.2361, + "step": 14770 + }, + { + "epoch": 0.537102987135693, + "grad_norm": 6.5722455978393555, + "learning_rate": 4.999448436267667e-05, + "loss": 0.2165, + "step": 14780 + }, + { + "epoch": 0.5374663856384911, + "grad_norm": 1.606378197669983, + "learning_rate": 4.999437482353395e-05, + "loss": 0.1642, + "step": 14790 + }, + { + "epoch": 0.5378297841412893, + "grad_norm": 24.709177017211914, + "learning_rate": 4.999426420747255e-05, + "loss": 0.2628, + "step": 14800 + }, + { + "epoch": 0.5381931826440876, + "grad_norm": 2.543760299682617, + "learning_rate": 4.999415251449723e-05, + "loss": 0.1883, + "step": 14810 + }, + { + "epoch": 0.5385565811468856, + "grad_norm": 2.0813279151916504, + "learning_rate": 4.999403974461281e-05, + "loss": 0.1842, + "step": 14820 + }, + { + "epoch": 0.5389199796496839, + "grad_norm": 4.744104385375977, + "learning_rate": 4.9993925897824144e-05, + "loss": 0.1981, + "step": 14830 + }, + { + "epoch": 0.5392833781524821, + "grad_norm": 3.2407493591308594, + "learning_rate": 4.9993810974136146e-05, + "loss": 0.2169, + "step": 14840 + }, + { + "epoch": 0.5396467766552802, + "grad_norm": 13.33681869506836, + "learning_rate": 4.999369497355375e-05, + "loss": 0.2775, + "step": 14850 + }, + { + "epoch": 0.5400101751580784, + "grad_norm": 2.3192784786224365, + "learning_rate": 4.9993577896081975e-05, + "loss": 0.1987, + "step": 14860 + }, + { + "epoch": 0.5403735736608766, + "grad_norm": 1.6611911058425903, + "learning_rate": 4.999345974172586e-05, + "loss": 0.188, + "step": 14870 + }, + { + "epoch": 0.5407369721636747, + "grad_norm": 4.368532180786133, + "learning_rate": 4.9993340510490485e-05, + "loss": 0.2201, + "step": 14880 + }, + { + "epoch": 0.5411003706664729, + "grad_norm": 1.4825586080551147, + "learning_rate": 4.999322020238099e-05, + "loss": 0.185, + "step": 14890 + }, + { + "epoch": 0.541463769169271, + "grad_norm": 4.346343994140625, + "learning_rate": 4.9993098817402564e-05, + "loss": 0.2415, + "step": 14900 + }, + { + "epoch": 0.5418271676720692, + "grad_norm": 1.0175251960754395, + "learning_rate": 4.999297635556044e-05, + "loss": 0.1991, + "step": 14910 + }, + { + "epoch": 0.5421905661748674, + "grad_norm": 1.600205421447754, + "learning_rate": 4.999285281685989e-05, + "loss": 0.1706, + "step": 14920 + }, + { + "epoch": 0.5425539646776655, + "grad_norm": 4.332497596740723, + "learning_rate": 4.999272820130623e-05, + "loss": 0.1964, + "step": 14930 + }, + { + "epoch": 0.5429173631804637, + "grad_norm": 2.0384531021118164, + "learning_rate": 4.999260250890484e-05, + "loss": 0.1571, + "step": 14940 + }, + { + "epoch": 0.5432807616832619, + "grad_norm": 11.780756950378418, + "learning_rate": 4.999247573966114e-05, + "loss": 0.319, + "step": 14950 + }, + { + "epoch": 0.54364416018606, + "grad_norm": 2.7058663368225098, + "learning_rate": 4.999234789358057e-05, + "loss": 0.2009, + "step": 14960 + }, + { + "epoch": 0.5440075586888582, + "grad_norm": 1.966780662536621, + "learning_rate": 4.999221897066866e-05, + "loss": 0.177, + "step": 14970 + }, + { + "epoch": 0.5443709571916564, + "grad_norm": 2.2129642963409424, + "learning_rate": 4.999208897093096e-05, + "loss": 0.2472, + "step": 14980 + }, + { + "epoch": 0.5447343556944545, + "grad_norm": 2.726358652114868, + "learning_rate": 4.9991957894373064e-05, + "loss": 0.2239, + "step": 14990 + }, + { + "epoch": 0.5450977541972527, + "grad_norm": 28.577600479125977, + "learning_rate": 4.999182574100063e-05, + "loss": 0.2524, + "step": 15000 + }, + { + "epoch": 0.5450977541972527, + "eval_loss": 0.3972287178039551, + "eval_runtime": 180.8086, + "eval_samples_per_second": 41.005, + "eval_steps_per_second": 5.127, + "eval_wer": 0.2289560150307695, + "step": 15000 + }, + { + "epoch": 0.5454611527000509, + "grad_norm": 1.9243866205215454, + "learning_rate": 4.9991692510819335e-05, + "loss": 0.1679, + "step": 15010 + }, + { + "epoch": 0.545824551202849, + "grad_norm": 1.3926585912704468, + "learning_rate": 4.9991558203834944e-05, + "loss": 0.1933, + "step": 15020 + }, + { + "epoch": 0.5461879497056472, + "grad_norm": 5.275027751922607, + "learning_rate": 4.999142282005322e-05, + "loss": 0.2838, + "step": 15030 + }, + { + "epoch": 0.5465513482084454, + "grad_norm": 2.142784357070923, + "learning_rate": 4.999128635948e-05, + "loss": 0.1754, + "step": 15040 + }, + { + "epoch": 0.5469147467112435, + "grad_norm": 40.32966995239258, + "learning_rate": 4.999114882212119e-05, + "loss": 0.323, + "step": 15050 + }, + { + "epoch": 0.5472781452140417, + "grad_norm": 1.056662678718567, + "learning_rate": 4.999101020798268e-05, + "loss": 0.1462, + "step": 15060 + }, + { + "epoch": 0.5476415437168399, + "grad_norm": 3.7527568340301514, + "learning_rate": 4.9990870517070464e-05, + "loss": 0.2106, + "step": 15070 + }, + { + "epoch": 0.548004942219638, + "grad_norm": 3.396487236022949, + "learning_rate": 4.9990729749390555e-05, + "loss": 0.1995, + "step": 15080 + }, + { + "epoch": 0.5483683407224362, + "grad_norm": 1.650519609451294, + "learning_rate": 4.999058790494902e-05, + "loss": 0.195, + "step": 15090 + }, + { + "epoch": 0.5487317392252344, + "grad_norm": 16.096418380737305, + "learning_rate": 4.9990444983751975e-05, + "loss": 0.2705, + "step": 15100 + }, + { + "epoch": 0.5490951377280325, + "grad_norm": 1.273149847984314, + "learning_rate": 4.999030098580556e-05, + "loss": 0.2216, + "step": 15110 + }, + { + "epoch": 0.5494585362308307, + "grad_norm": 1.5414496660232544, + "learning_rate": 4.9990155911115995e-05, + "loss": 0.1876, + "step": 15120 + }, + { + "epoch": 0.549821934733629, + "grad_norm": 4.707805633544922, + "learning_rate": 4.9990009759689524e-05, + "loss": 0.1895, + "step": 15130 + }, + { + "epoch": 0.550185333236427, + "grad_norm": 2.033162832260132, + "learning_rate": 4.9989862531532456e-05, + "loss": 0.1705, + "step": 15140 + }, + { + "epoch": 0.5505487317392252, + "grad_norm": 7.349232196807861, + "learning_rate": 4.998971422665112e-05, + "loss": 0.2815, + "step": 15150 + }, + { + "epoch": 0.5509121302420235, + "grad_norm": 1.293078064918518, + "learning_rate": 4.9989564845051915e-05, + "loss": 0.1789, + "step": 15160 + }, + { + "epoch": 0.5512755287448216, + "grad_norm": 1.7343147993087769, + "learning_rate": 4.998941438674127e-05, + "loss": 0.1781, + "step": 15170 + }, + { + "epoch": 0.5516389272476198, + "grad_norm": 2.440030574798584, + "learning_rate": 4.9989262851725674e-05, + "loss": 0.1927, + "step": 15180 + }, + { + "epoch": 0.5520023257504179, + "grad_norm": 2.276111364364624, + "learning_rate": 4.998911024001165e-05, + "loss": 0.1774, + "step": 15190 + }, + { + "epoch": 0.5523657242532161, + "grad_norm": 9.360533714294434, + "learning_rate": 4.9988956551605783e-05, + "loss": 0.2761, + "step": 15200 + }, + { + "epoch": 0.5527291227560143, + "grad_norm": 3.8025522232055664, + "learning_rate": 4.998880178651468e-05, + "loss": 0.2855, + "step": 15210 + }, + { + "epoch": 0.5530925212588124, + "grad_norm": 3.816631555557251, + "learning_rate": 4.998864594474503e-05, + "loss": 0.1559, + "step": 15220 + }, + { + "epoch": 0.5534559197616106, + "grad_norm": 3.2255067825317383, + "learning_rate": 4.998848902630353e-05, + "loss": 0.1632, + "step": 15230 + }, + { + "epoch": 0.5538193182644088, + "grad_norm": 1.077268123626709, + "learning_rate": 4.9988331031196944e-05, + "loss": 0.1969, + "step": 15240 + }, + { + "epoch": 0.5541827167672069, + "grad_norm": 5.657801151275635, + "learning_rate": 4.998817195943209e-05, + "loss": 0.2361, + "step": 15250 + }, + { + "epoch": 0.5545461152700051, + "grad_norm": 1.180039882659912, + "learning_rate": 4.998801181101581e-05, + "loss": 0.1779, + "step": 15260 + }, + { + "epoch": 0.5549095137728033, + "grad_norm": 2.12725830078125, + "learning_rate": 4.998785058595501e-05, + "loss": 0.1505, + "step": 15270 + }, + { + "epoch": 0.5552729122756014, + "grad_norm": 2.0784361362457275, + "learning_rate": 4.998768828425664e-05, + "loss": 0.2221, + "step": 15280 + }, + { + "epoch": 0.5556363107783996, + "grad_norm": 2.0133538246154785, + "learning_rate": 4.998752490592768e-05, + "loss": 0.1759, + "step": 15290 + }, + { + "epoch": 0.5559997092811978, + "grad_norm": 3.3181140422821045, + "learning_rate": 4.998736045097518e-05, + "loss": 0.229, + "step": 15300 + }, + { + "epoch": 0.5563631077839959, + "grad_norm": 1.2881536483764648, + "learning_rate": 4.998719491940622e-05, + "loss": 0.1928, + "step": 15310 + }, + { + "epoch": 0.5567265062867941, + "grad_norm": 1.0155376195907593, + "learning_rate": 4.998702831122794e-05, + "loss": 0.1986, + "step": 15320 + }, + { + "epoch": 0.5570899047895923, + "grad_norm": 7.5557661056518555, + "learning_rate": 4.998686062644752e-05, + "loss": 0.2317, + "step": 15330 + }, + { + "epoch": 0.5574533032923904, + "grad_norm": 2.3196377754211426, + "learning_rate": 4.9986691865072176e-05, + "loss": 0.1827, + "step": 15340 + }, + { + "epoch": 0.5578167017951886, + "grad_norm": 25.910188674926758, + "learning_rate": 4.998652202710918e-05, + "loss": 0.2824, + "step": 15350 + }, + { + "epoch": 0.5581801002979868, + "grad_norm": 1.0091907978057861, + "learning_rate": 4.9986351112565846e-05, + "loss": 0.1946, + "step": 15360 + }, + { + "epoch": 0.5585434988007849, + "grad_norm": 3.0022408962249756, + "learning_rate": 4.998617912144956e-05, + "loss": 0.2028, + "step": 15370 + }, + { + "epoch": 0.5589068973035831, + "grad_norm": 2.9837419986724854, + "learning_rate": 4.99860060537677e-05, + "loss": 0.203, + "step": 15380 + }, + { + "epoch": 0.5592702958063813, + "grad_norm": 2.238867998123169, + "learning_rate": 4.9985831909527746e-05, + "loss": 0.1392, + "step": 15390 + }, + { + "epoch": 0.5596336943091794, + "grad_norm": 3.8585119247436523, + "learning_rate": 4.9985656688737205e-05, + "loss": 0.2289, + "step": 15400 + }, + { + "epoch": 0.5599970928119776, + "grad_norm": 2.4951331615448, + "learning_rate": 4.998548039140361e-05, + "loss": 0.1852, + "step": 15410 + }, + { + "epoch": 0.5603604913147758, + "grad_norm": 1.8404667377471924, + "learning_rate": 4.998530301753455e-05, + "loss": 0.1813, + "step": 15420 + }, + { + "epoch": 0.5607238898175739, + "grad_norm": 2.615247964859009, + "learning_rate": 4.9985124567137695e-05, + "loss": 0.228, + "step": 15430 + }, + { + "epoch": 0.5610872883203721, + "grad_norm": 1.2074272632598877, + "learning_rate": 4.9984945040220715e-05, + "loss": 0.1879, + "step": 15440 + }, + { + "epoch": 0.5614506868231703, + "grad_norm": 38.466712951660156, + "learning_rate": 4.9984764436791355e-05, + "loss": 0.3965, + "step": 15450 + }, + { + "epoch": 0.5618140853259684, + "grad_norm": 1.4196547269821167, + "learning_rate": 4.998458275685739e-05, + "loss": 0.2061, + "step": 15460 + }, + { + "epoch": 0.5621774838287666, + "grad_norm": 1.2451281547546387, + "learning_rate": 4.998440000042664e-05, + "loss": 0.2118, + "step": 15470 + }, + { + "epoch": 0.5625408823315647, + "grad_norm": 3.7021896839141846, + "learning_rate": 4.9984216167507005e-05, + "loss": 0.2294, + "step": 15480 + }, + { + "epoch": 0.562904280834363, + "grad_norm": 2.8826780319213867, + "learning_rate": 4.998403125810638e-05, + "loss": 0.1654, + "step": 15490 + }, + { + "epoch": 0.5632676793371612, + "grad_norm": 8.366926193237305, + "learning_rate": 4.998384527223274e-05, + "loss": 0.2467, + "step": 15500 + }, + { + "epoch": 0.5636310778399592, + "grad_norm": 2.2532148361206055, + "learning_rate": 4.99836582098941e-05, + "loss": 0.2569, + "step": 15510 + }, + { + "epoch": 0.5639944763427575, + "grad_norm": 2.164987325668335, + "learning_rate": 4.998347007109853e-05, + "loss": 0.2167, + "step": 15520 + }, + { + "epoch": 0.5643578748455557, + "grad_norm": 4.651108264923096, + "learning_rate": 4.998328085585411e-05, + "loss": 0.2138, + "step": 15530 + }, + { + "epoch": 0.5647212733483538, + "grad_norm": 1.5128902196884155, + "learning_rate": 4.9983090564169024e-05, + "loss": 0.1821, + "step": 15540 + }, + { + "epoch": 0.565084671851152, + "grad_norm": 8.516124725341797, + "learning_rate": 4.998289919605145e-05, + "loss": 0.2546, + "step": 15550 + }, + { + "epoch": 0.5654480703539502, + "grad_norm": 1.6480666399002075, + "learning_rate": 4.9982706751509635e-05, + "loss": 0.2069, + "step": 15560 + }, + { + "epoch": 0.5658114688567483, + "grad_norm": 1.3768938779830933, + "learning_rate": 4.998251323055187e-05, + "loss": 0.1775, + "step": 15570 + }, + { + "epoch": 0.5661748673595465, + "grad_norm": 1.8793795108795166, + "learning_rate": 4.998231863318651e-05, + "loss": 0.14, + "step": 15580 + }, + { + "epoch": 0.5665382658623447, + "grad_norm": 1.2361701726913452, + "learning_rate": 4.9982122959421924e-05, + "loss": 0.1797, + "step": 15590 + }, + { + "epoch": 0.5669016643651428, + "grad_norm": 14.16727352142334, + "learning_rate": 4.998192620926655e-05, + "loss": 0.3523, + "step": 15600 + }, + { + "epoch": 0.5669016643651428, + "eval_loss": 0.40661031007766724, + "eval_runtime": 180.2598, + "eval_samples_per_second": 41.13, + "eval_steps_per_second": 5.143, + "eval_wer": 0.21753771307204967, + "step": 15600 + }, + { + "epoch": 0.567265062867941, + "grad_norm": 2.460245370864868, + "learning_rate": 4.9981728382728855e-05, + "loss": 0.1824, + "step": 15610 + }, + { + "epoch": 0.5676284613707392, + "grad_norm": 1.603381633758545, + "learning_rate": 4.9981529479817366e-05, + "loss": 0.1506, + "step": 15620 + }, + { + "epoch": 0.5679918598735373, + "grad_norm": 3.650087356567383, + "learning_rate": 4.9981329500540664e-05, + "loss": 0.2351, + "step": 15630 + }, + { + "epoch": 0.5683552583763355, + "grad_norm": 2.2338075637817383, + "learning_rate": 4.9981128444907354e-05, + "loss": 0.1785, + "step": 15640 + }, + { + "epoch": 0.5687186568791337, + "grad_norm": 7.641642093658447, + "learning_rate": 4.998092631292611e-05, + "loss": 0.2816, + "step": 15650 + }, + { + "epoch": 0.5690820553819318, + "grad_norm": 1.5877048969268799, + "learning_rate": 4.998072310460562e-05, + "loss": 0.1784, + "step": 15660 + }, + { + "epoch": 0.56944545388473, + "grad_norm": 3.5917787551879883, + "learning_rate": 4.998051881995466e-05, + "loss": 0.1685, + "step": 15670 + }, + { + "epoch": 0.5698088523875282, + "grad_norm": 6.459184169769287, + "learning_rate": 4.998031345898203e-05, + "loss": 0.2031, + "step": 15680 + }, + { + "epoch": 0.5701722508903263, + "grad_norm": 2.7518184185028076, + "learning_rate": 4.9980107021696565e-05, + "loss": 0.168, + "step": 15690 + }, + { + "epoch": 0.5705356493931245, + "grad_norm": 9.814598083496094, + "learning_rate": 4.997989950810718e-05, + "loss": 0.2778, + "step": 15700 + }, + { + "epoch": 0.5708990478959227, + "grad_norm": 2.0985398292541504, + "learning_rate": 4.9979690918222785e-05, + "loss": 0.1864, + "step": 15710 + }, + { + "epoch": 0.5712624463987208, + "grad_norm": 1.9264591932296753, + "learning_rate": 4.997948125205241e-05, + "loss": 0.1682, + "step": 15720 + }, + { + "epoch": 0.571625844901519, + "grad_norm": 4.2961955070495605, + "learning_rate": 4.997927050960505e-05, + "loss": 0.198, + "step": 15730 + }, + { + "epoch": 0.5719892434043172, + "grad_norm": 4.524483680725098, + "learning_rate": 4.99790586908898e-05, + "loss": 0.3235, + "step": 15740 + }, + { + "epoch": 0.5723526419071153, + "grad_norm": 5.259559154510498, + "learning_rate": 4.997884579591578e-05, + "loss": 0.335, + "step": 15750 + }, + { + "epoch": 0.5727160404099135, + "grad_norm": 1.7875639200210571, + "learning_rate": 4.997863182469219e-05, + "loss": 0.1674, + "step": 15760 + }, + { + "epoch": 0.5730794389127116, + "grad_norm": 1.1852960586547852, + "learning_rate": 4.9978416777228216e-05, + "loss": 0.1968, + "step": 15770 + }, + { + "epoch": 0.5734428374155098, + "grad_norm": 1.253061294555664, + "learning_rate": 4.997820065353314e-05, + "loss": 0.2177, + "step": 15780 + }, + { + "epoch": 0.573806235918308, + "grad_norm": 2.0577871799468994, + "learning_rate": 4.9977983453616266e-05, + "loss": 0.1498, + "step": 15790 + }, + { + "epoch": 0.5741696344211061, + "grad_norm": 7.4168901443481445, + "learning_rate": 4.997776517748696e-05, + "loss": 0.3137, + "step": 15800 + }, + { + "epoch": 0.5745330329239043, + "grad_norm": 2.9957845211029053, + "learning_rate": 4.9977545825154625e-05, + "loss": 0.1819, + "step": 15810 + }, + { + "epoch": 0.5748964314267025, + "grad_norm": 1.251610517501831, + "learning_rate": 4.997732539662871e-05, + "loss": 0.1633, + "step": 15820 + }, + { + "epoch": 0.5752598299295006, + "grad_norm": 3.229581594467163, + "learning_rate": 4.997710389191871e-05, + "loss": 0.1888, + "step": 15830 + }, + { + "epoch": 0.5756232284322989, + "grad_norm": 1.2718089818954468, + "learning_rate": 4.997688131103417e-05, + "loss": 0.1938, + "step": 15840 + }, + { + "epoch": 0.5759866269350971, + "grad_norm": 4.77078104019165, + "learning_rate": 4.9976657653984694e-05, + "loss": 0.2311, + "step": 15850 + }, + { + "epoch": 0.5763500254378952, + "grad_norm": 1.9487907886505127, + "learning_rate": 4.9976432920779904e-05, + "loss": 0.7679, + "step": 15860 + }, + { + "epoch": 0.5767134239406934, + "grad_norm": 2.1322100162506104, + "learning_rate": 4.997620711142948e-05, + "loss": 0.204, + "step": 15870 + }, + { + "epoch": 0.5770768224434916, + "grad_norm": 3.0756008625030518, + "learning_rate": 4.997598022594316e-05, + "loss": 0.205, + "step": 15880 + }, + { + "epoch": 0.5774402209462897, + "grad_norm": 2.2399511337280273, + "learning_rate": 4.997575226433071e-05, + "loss": 0.183, + "step": 15890 + }, + { + "epoch": 0.5778036194490879, + "grad_norm": 4.17095947265625, + "learning_rate": 4.997552322660197e-05, + "loss": 0.2209, + "step": 15900 + }, + { + "epoch": 0.5781670179518861, + "grad_norm": 1.7085528373718262, + "learning_rate": 4.9975293112766794e-05, + "loss": 0.4196, + "step": 15910 + }, + { + "epoch": 0.5785304164546842, + "grad_norm": 1.6818984746932983, + "learning_rate": 4.99750619228351e-05, + "loss": 0.1592, + "step": 15920 + }, + { + "epoch": 0.5788938149574824, + "grad_norm": 2.526503324508667, + "learning_rate": 4.9974829656816846e-05, + "loss": 0.7523, + "step": 15930 + }, + { + "epoch": 0.5792572134602806, + "grad_norm": 4.456855297088623, + "learning_rate": 4.997459631472205e-05, + "loss": 0.1664, + "step": 15940 + }, + { + "epoch": 0.5796206119630787, + "grad_norm": 28.427839279174805, + "learning_rate": 4.9974361896560746e-05, + "loss": 0.2891, + "step": 15950 + }, + { + "epoch": 0.5799840104658769, + "grad_norm": 1.6720882654190063, + "learning_rate": 4.997412640234306e-05, + "loss": 0.1522, + "step": 15960 + }, + { + "epoch": 0.5803474089686751, + "grad_norm": 1.6327390670776367, + "learning_rate": 4.997388983207911e-05, + "loss": 0.1957, + "step": 15970 + }, + { + "epoch": 0.5807108074714732, + "grad_norm": 1.5792416334152222, + "learning_rate": 4.997365218577912e-05, + "loss": 0.2325, + "step": 15980 + }, + { + "epoch": 0.5810742059742714, + "grad_norm": 1.7585738897323608, + "learning_rate": 4.9973413463453305e-05, + "loss": 0.2023, + "step": 15990 + }, + { + "epoch": 0.5814376044770696, + "grad_norm": 8.14810562133789, + "learning_rate": 4.997317366511196e-05, + "loss": 0.2093, + "step": 16000 + }, + { + "epoch": 0.5818010029798677, + "grad_norm": 3.1430416107177734, + "learning_rate": 4.997293279076543e-05, + "loss": 0.1742, + "step": 16010 + }, + { + "epoch": 0.5821644014826659, + "grad_norm": 1.9447312355041504, + "learning_rate": 4.997269084042406e-05, + "loss": 0.1852, + "step": 16020 + }, + { + "epoch": 0.5825277999854641, + "grad_norm": 2.1479732990264893, + "learning_rate": 4.997244781409831e-05, + "loss": 0.2197, + "step": 16030 + }, + { + "epoch": 0.5828911984882622, + "grad_norm": 3.7066800594329834, + "learning_rate": 4.9972203711798625e-05, + "loss": 0.1899, + "step": 16040 + }, + { + "epoch": 0.5832545969910604, + "grad_norm": 4.3598432540893555, + "learning_rate": 4.9971958533535544e-05, + "loss": 0.237, + "step": 16050 + }, + { + "epoch": 0.5836179954938585, + "grad_norm": 3.410356283187866, + "learning_rate": 4.997171227931962e-05, + "loss": 0.1985, + "step": 16060 + }, + { + "epoch": 0.5839813939966567, + "grad_norm": 1.6299129724502563, + "learning_rate": 4.9971464949161454e-05, + "loss": 0.1758, + "step": 16070 + }, + { + "epoch": 0.5843447924994549, + "grad_norm": 1.993067979812622, + "learning_rate": 4.9971216543071716e-05, + "loss": 0.1822, + "step": 16080 + }, + { + "epoch": 0.584708191002253, + "grad_norm": 1.2057979106903076, + "learning_rate": 4.9970967061061104e-05, + "loss": 0.1953, + "step": 16090 + }, + { + "epoch": 0.5850715895050512, + "grad_norm": 34.54500961303711, + "learning_rate": 4.997071650314037e-05, + "loss": 0.277, + "step": 16100 + }, + { + "epoch": 0.5854349880078494, + "grad_norm": 1.243656039237976, + "learning_rate": 4.997046486932031e-05, + "loss": 0.172, + "step": 16110 + }, + { + "epoch": 0.5857983865106475, + "grad_norm": 0.6155187487602234, + "learning_rate": 4.997021215961176e-05, + "loss": 0.1523, + "step": 16120 + }, + { + "epoch": 0.5861617850134457, + "grad_norm": 2.0203208923339844, + "learning_rate": 4.9969958374025615e-05, + "loss": 0.1857, + "step": 16130 + }, + { + "epoch": 0.5865251835162439, + "grad_norm": 1.1912654638290405, + "learning_rate": 4.9969703512572805e-05, + "loss": 0.2169, + "step": 16140 + }, + { + "epoch": 0.586888582019042, + "grad_norm": 3.528538227081299, + "learning_rate": 4.9969447575264315e-05, + "loss": 0.2428, + "step": 16150 + }, + { + "epoch": 0.5872519805218402, + "grad_norm": 0.9166990518569946, + "learning_rate": 4.996919056211117e-05, + "loss": 0.2206, + "step": 16160 + }, + { + "epoch": 0.5876153790246385, + "grad_norm": 1.4956426620483398, + "learning_rate": 4.996893247312444e-05, + "loss": 0.168, + "step": 16170 + }, + { + "epoch": 0.5879787775274365, + "grad_norm": 1.4502993822097778, + "learning_rate": 4.996867330831526e-05, + "loss": 0.1767, + "step": 16180 + }, + { + "epoch": 0.5883421760302348, + "grad_norm": 0.9337482452392578, + "learning_rate": 4.9968413067694775e-05, + "loss": 0.2035, + "step": 16190 + }, + { + "epoch": 0.588705574533033, + "grad_norm": 13.500269889831543, + "learning_rate": 4.996815175127422e-05, + "loss": 0.2335, + "step": 16200 + }, + { + "epoch": 0.588705574533033, + "eval_loss": 0.4428017735481262, + "eval_runtime": 180.2159, + "eval_samples_per_second": 41.14, + "eval_steps_per_second": 5.144, + "eval_wer": 0.22281118957285748, + "step": 16200 + }, + { + "epoch": 0.5890689730358311, + "grad_norm": 5.091770648956299, + "learning_rate": 4.996788935906483e-05, + "loss": 0.1884, + "step": 16210 + }, + { + "epoch": 0.5894323715386293, + "grad_norm": 2.379033327102661, + "learning_rate": 4.996762589107793e-05, + "loss": 0.196, + "step": 16220 + }, + { + "epoch": 0.5897957700414275, + "grad_norm": 2.576484203338623, + "learning_rate": 4.996736134732487e-05, + "loss": 0.2069, + "step": 16230 + }, + { + "epoch": 0.5901591685442256, + "grad_norm": 1.4235923290252686, + "learning_rate": 4.9967095727817035e-05, + "loss": 0.1905, + "step": 16240 + }, + { + "epoch": 0.5905225670470238, + "grad_norm": 7.119918346405029, + "learning_rate": 4.9966829032565886e-05, + "loss": 0.2803, + "step": 16250 + }, + { + "epoch": 0.590885965549822, + "grad_norm": 1.1050286293029785, + "learning_rate": 4.99665612615829e-05, + "loss": 0.1914, + "step": 16260 + }, + { + "epoch": 0.5912493640526201, + "grad_norm": 1.403601884841919, + "learning_rate": 4.9966292414879625e-05, + "loss": 0.1649, + "step": 16270 + }, + { + "epoch": 0.5916127625554183, + "grad_norm": 5.439052104949951, + "learning_rate": 4.9966022492467635e-05, + "loss": 0.1897, + "step": 16280 + }, + { + "epoch": 0.5919761610582165, + "grad_norm": 1.0014379024505615, + "learning_rate": 4.996575149435857e-05, + "loss": 0.1472, + "step": 16290 + }, + { + "epoch": 0.5923395595610146, + "grad_norm": 9.480517387390137, + "learning_rate": 4.99654794205641e-05, + "loss": 0.2351, + "step": 16300 + }, + { + "epoch": 0.5927029580638128, + "grad_norm": 0.8406987190246582, + "learning_rate": 4.9965206271095955e-05, + "loss": 0.1795, + "step": 16310 + }, + { + "epoch": 0.593066356566611, + "grad_norm": 1.378169298171997, + "learning_rate": 4.996493204596589e-05, + "loss": 0.1597, + "step": 16320 + }, + { + "epoch": 0.5934297550694091, + "grad_norm": 3.9748549461364746, + "learning_rate": 4.996465674518573e-05, + "loss": 0.2264, + "step": 16330 + }, + { + "epoch": 0.5937931535722073, + "grad_norm": 2.2626171112060547, + "learning_rate": 4.996438036876734e-05, + "loss": 0.1647, + "step": 16340 + }, + { + "epoch": 0.5941565520750055, + "grad_norm": 3.8039205074310303, + "learning_rate": 4.996410291672262e-05, + "loss": 0.2204, + "step": 16350 + }, + { + "epoch": 0.5945199505778036, + "grad_norm": 1.5219416618347168, + "learning_rate": 4.996382438906353e-05, + "loss": 0.1518, + "step": 16360 + }, + { + "epoch": 0.5948833490806018, + "grad_norm": 1.4811570644378662, + "learning_rate": 4.9963544785802064e-05, + "loss": 0.2006, + "step": 16370 + }, + { + "epoch": 0.5952467475833999, + "grad_norm": 4.7030558586120605, + "learning_rate": 4.996326410695028e-05, + "loss": 0.2524, + "step": 16380 + }, + { + "epoch": 0.5956101460861981, + "grad_norm": 1.103624939918518, + "learning_rate": 4.996298235252026e-05, + "loss": 0.1558, + "step": 16390 + }, + { + "epoch": 0.5959735445889963, + "grad_norm": 4.654818534851074, + "learning_rate": 4.996269952252415e-05, + "loss": 0.2746, + "step": 16400 + }, + { + "epoch": 0.5963369430917944, + "grad_norm": 1.6746747493743896, + "learning_rate": 4.996241561697413e-05, + "loss": 0.1838, + "step": 16410 + }, + { + "epoch": 0.5967003415945926, + "grad_norm": 3.1955924034118652, + "learning_rate": 4.996213063588245e-05, + "loss": 0.1773, + "step": 16420 + }, + { + "epoch": 0.5970637400973908, + "grad_norm": 1.782669186592102, + "learning_rate": 4.996184457926137e-05, + "loss": 0.1939, + "step": 16430 + }, + { + "epoch": 0.5974271386001889, + "grad_norm": 1.2277849912643433, + "learning_rate": 4.996155744712322e-05, + "loss": 0.1724, + "step": 16440 + }, + { + "epoch": 0.5977905371029871, + "grad_norm": 25.578798294067383, + "learning_rate": 4.996126923948038e-05, + "loss": 0.2612, + "step": 16450 + }, + { + "epoch": 0.5981539356057853, + "grad_norm": 0.984426736831665, + "learning_rate": 4.9960979956345254e-05, + "loss": 0.1621, + "step": 16460 + }, + { + "epoch": 0.5985173341085834, + "grad_norm": 2.1299145221710205, + "learning_rate": 4.9960689597730315e-05, + "loss": 0.161, + "step": 16470 + }, + { + "epoch": 0.5988807326113816, + "grad_norm": 2.6153085231781006, + "learning_rate": 4.996039816364807e-05, + "loss": 0.2122, + "step": 16480 + }, + { + "epoch": 0.5992441311141798, + "grad_norm": 4.464552879333496, + "learning_rate": 4.996010565411108e-05, + "loss": 0.2417, + "step": 16490 + }, + { + "epoch": 0.5996075296169779, + "grad_norm": 26.441349029541016, + "learning_rate": 4.995981206913194e-05, + "loss": 0.3103, + "step": 16500 + }, + { + "epoch": 0.5999709281197761, + "grad_norm": 2.353302478790283, + "learning_rate": 4.995951740872331e-05, + "loss": 1.0256, + "step": 16510 + }, + { + "epoch": 0.6003343266225744, + "grad_norm": 0.8436356782913208, + "learning_rate": 4.995922167289788e-05, + "loss": 0.1563, + "step": 16520 + }, + { + "epoch": 0.6006977251253725, + "grad_norm": 3.3516342639923096, + "learning_rate": 4.99589248616684e-05, + "loss": 0.2441, + "step": 16530 + }, + { + "epoch": 0.6010611236281707, + "grad_norm": 2.0286059379577637, + "learning_rate": 4.995862697504764e-05, + "loss": 0.1767, + "step": 16540 + }, + { + "epoch": 0.6014245221309689, + "grad_norm": 18.248151779174805, + "learning_rate": 4.9958328013048464e-05, + "loss": 0.3522, + "step": 16550 + }, + { + "epoch": 0.601787920633767, + "grad_norm": 2.1514463424682617, + "learning_rate": 4.995802797568372e-05, + "loss": 0.1771, + "step": 16560 + }, + { + "epoch": 0.6021513191365652, + "grad_norm": 5.868020534515381, + "learning_rate": 4.995772686296635e-05, + "loss": 0.1776, + "step": 16570 + }, + { + "epoch": 0.6025147176393634, + "grad_norm": 4.539637565612793, + "learning_rate": 4.9957424674909336e-05, + "loss": 0.2002, + "step": 16580 + }, + { + "epoch": 0.6028781161421615, + "grad_norm": 1.7226190567016602, + "learning_rate": 4.99571214115257e-05, + "loss": 0.1927, + "step": 16590 + }, + { + "epoch": 0.6032415146449597, + "grad_norm": 22.087247848510742, + "learning_rate": 4.9956817072828485e-05, + "loss": 0.249, + "step": 16600 + }, + { + "epoch": 0.6036049131477579, + "grad_norm": 2.4267120361328125, + "learning_rate": 4.995651165883083e-05, + "loss": 0.1935, + "step": 16610 + }, + { + "epoch": 0.603968311650556, + "grad_norm": 2.5284249782562256, + "learning_rate": 4.995620516954588e-05, + "loss": 0.1495, + "step": 16620 + }, + { + "epoch": 0.6043317101533542, + "grad_norm": 1.5988596677780151, + "learning_rate": 4.995589760498684e-05, + "loss": 0.2329, + "step": 16630 + }, + { + "epoch": 0.6046951086561524, + "grad_norm": 1.0771689414978027, + "learning_rate": 4.9955588965166966e-05, + "loss": 0.1634, + "step": 16640 + }, + { + "epoch": 0.6050585071589505, + "grad_norm": 8.72423267364502, + "learning_rate": 4.995527925009956e-05, + "loss": 0.27, + "step": 16650 + }, + { + "epoch": 0.6054219056617487, + "grad_norm": 1.3176789283752441, + "learning_rate": 4.9954968459797955e-05, + "loss": 1.1913, + "step": 16660 + }, + { + "epoch": 0.6057853041645468, + "grad_norm": 1.8307547569274902, + "learning_rate": 4.9954656594275555e-05, + "loss": 0.188, + "step": 16670 + }, + { + "epoch": 0.606148702667345, + "grad_norm": 2.783604621887207, + "learning_rate": 4.9954343653545795e-05, + "loss": 0.1791, + "step": 16680 + }, + { + "epoch": 0.6065121011701432, + "grad_norm": 1.6639970541000366, + "learning_rate": 4.9954029637622146e-05, + "loss": 0.1829, + "step": 16690 + }, + { + "epoch": 0.6068754996729413, + "grad_norm": 11.055110931396484, + "learning_rate": 4.995371454651815e-05, + "loss": 0.2229, + "step": 16700 + }, + { + "epoch": 0.6072388981757395, + "grad_norm": 1.8166972398757935, + "learning_rate": 4.9953398380247384e-05, + "loss": 0.1734, + "step": 16710 + }, + { + "epoch": 0.6076022966785377, + "grad_norm": 4.851889610290527, + "learning_rate": 4.995308113882346e-05, + "loss": 0.1716, + "step": 16720 + }, + { + "epoch": 0.6079656951813358, + "grad_norm": 3.0047857761383057, + "learning_rate": 4.9952762822260056e-05, + "loss": 0.2125, + "step": 16730 + }, + { + "epoch": 0.608329093684134, + "grad_norm": 1.1506407260894775, + "learning_rate": 4.9952443430570887e-05, + "loss": 0.171, + "step": 16740 + }, + { + "epoch": 0.6086924921869322, + "grad_norm": 4.324979782104492, + "learning_rate": 4.995212296376971e-05, + "loss": 0.2365, + "step": 16750 + }, + { + "epoch": 0.6090558906897303, + "grad_norm": 1.2295490503311157, + "learning_rate": 4.995180142187033e-05, + "loss": 0.2002, + "step": 16760 + }, + { + "epoch": 0.6094192891925285, + "grad_norm": 1.454434871673584, + "learning_rate": 4.995147880488661e-05, + "loss": 0.1602, + "step": 16770 + }, + { + "epoch": 0.6097826876953267, + "grad_norm": 2.6185641288757324, + "learning_rate": 4.995115511283244e-05, + "loss": 0.1904, + "step": 16780 + }, + { + "epoch": 0.6101460861981248, + "grad_norm": 1.2603826522827148, + "learning_rate": 4.9950830345721774e-05, + "loss": 0.1892, + "step": 16790 + }, + { + "epoch": 0.610509484700923, + "grad_norm": 14.189190864562988, + "learning_rate": 4.9950504503568615e-05, + "loss": 0.281, + "step": 16800 + }, + { + "epoch": 0.610509484700923, + "eval_loss": 0.4123116433620453, + "eval_runtime": 179.5081, + "eval_samples_per_second": 41.302, + "eval_steps_per_second": 5.164, + "eval_wer": 0.22986367019441972, + "step": 16800 + }, + { + "epoch": 0.6108728832037212, + "grad_norm": 1.7670204639434814, + "learning_rate": 4.995017758638698e-05, + "loss": 0.2581, + "step": 16810 + }, + { + "epoch": 0.6112362817065193, + "grad_norm": 1.2099360227584839, + "learning_rate": 4.9949849594190964e-05, + "loss": 0.1762, + "step": 16820 + }, + { + "epoch": 0.6115996802093175, + "grad_norm": 2.7719335556030273, + "learning_rate": 4.9949520526994716e-05, + "loss": 0.1867, + "step": 16830 + }, + { + "epoch": 0.6119630787121157, + "grad_norm": 1.5935924053192139, + "learning_rate": 4.9949190384812386e-05, + "loss": 0.191, + "step": 16840 + }, + { + "epoch": 0.6123264772149138, + "grad_norm": 3.511439085006714, + "learning_rate": 4.994885916765821e-05, + "loss": 0.2132, + "step": 16850 + }, + { + "epoch": 0.612689875717712, + "grad_norm": 1.693789005279541, + "learning_rate": 4.994852687554647e-05, + "loss": 0.167, + "step": 16860 + }, + { + "epoch": 0.6130532742205103, + "grad_norm": 2.1199066638946533, + "learning_rate": 4.994819350849147e-05, + "loss": 0.172, + "step": 16870 + }, + { + "epoch": 0.6134166727233084, + "grad_norm": 2.724487543106079, + "learning_rate": 4.9947859066507575e-05, + "loss": 0.2083, + "step": 16880 + }, + { + "epoch": 0.6137800712261066, + "grad_norm": 0.926547110080719, + "learning_rate": 4.99475235496092e-05, + "loss": 0.1517, + "step": 16890 + }, + { + "epoch": 0.6141434697289048, + "grad_norm": 14.503059387207031, + "learning_rate": 4.99471869578108e-05, + "loss": 0.1945, + "step": 16900 + }, + { + "epoch": 0.6145068682317029, + "grad_norm": 3.2206919193267822, + "learning_rate": 4.994684929112687e-05, + "loss": 0.1882, + "step": 16910 + }, + { + "epoch": 0.6148702667345011, + "grad_norm": 2.004995107650757, + "learning_rate": 4.994651054957198e-05, + "loss": 0.1876, + "step": 16920 + }, + { + "epoch": 0.6152336652372993, + "grad_norm": 2.0580127239227295, + "learning_rate": 4.99461707331607e-05, + "loss": 0.2104, + "step": 16930 + }, + { + "epoch": 0.6155970637400974, + "grad_norm": 3.3028602600097656, + "learning_rate": 4.9945829841907684e-05, + "loss": 0.1494, + "step": 16940 + }, + { + "epoch": 0.6159604622428956, + "grad_norm": 7.572249412536621, + "learning_rate": 4.994548787582761e-05, + "loss": 0.2381, + "step": 16950 + }, + { + "epoch": 0.6163238607456937, + "grad_norm": 1.4220709800720215, + "learning_rate": 4.9945144834935234e-05, + "loss": 0.1916, + "step": 16960 + }, + { + "epoch": 0.6166872592484919, + "grad_norm": 1.2397724390029907, + "learning_rate": 4.994480071924531e-05, + "loss": 0.1593, + "step": 16970 + }, + { + "epoch": 0.6170506577512901, + "grad_norm": 2.2569403648376465, + "learning_rate": 4.9944455528772684e-05, + "loss": 0.1984, + "step": 16980 + }, + { + "epoch": 0.6174140562540882, + "grad_norm": 1.811727523803711, + "learning_rate": 4.994410926353221e-05, + "loss": 0.1838, + "step": 16990 + }, + { + "epoch": 0.6177774547568864, + "grad_norm": 2.783061981201172, + "learning_rate": 4.9943761923538834e-05, + "loss": 0.2217, + "step": 17000 + }, + { + "epoch": 0.6181408532596846, + "grad_norm": 2.816331148147583, + "learning_rate": 4.99434135088075e-05, + "loss": 0.1911, + "step": 17010 + }, + { + "epoch": 0.6185042517624827, + "grad_norm": 1.238916039466858, + "learning_rate": 4.9943064019353234e-05, + "loss": 0.1854, + "step": 17020 + }, + { + "epoch": 0.6188676502652809, + "grad_norm": 5.16685152053833, + "learning_rate": 4.9942713455191075e-05, + "loss": 0.1797, + "step": 17030 + }, + { + "epoch": 0.6192310487680791, + "grad_norm": 2.486461639404297, + "learning_rate": 4.9942361816336146e-05, + "loss": 0.1926, + "step": 17040 + }, + { + "epoch": 0.6195944472708772, + "grad_norm": 9.018515586853027, + "learning_rate": 4.994200910280359e-05, + "loss": 0.2193, + "step": 17050 + }, + { + "epoch": 0.6199578457736754, + "grad_norm": 1.804166555404663, + "learning_rate": 4.994165531460861e-05, + "loss": 0.1977, + "step": 17060 + }, + { + "epoch": 0.6203212442764736, + "grad_norm": 1.2862845659255981, + "learning_rate": 4.994130045176644e-05, + "loss": 0.1493, + "step": 17070 + }, + { + "epoch": 0.6206846427792717, + "grad_norm": 4.164750576019287, + "learning_rate": 4.994094451429237e-05, + "loss": 0.2548, + "step": 17080 + }, + { + "epoch": 0.6210480412820699, + "grad_norm": 1.577255368232727, + "learning_rate": 4.994058750220176e-05, + "loss": 0.1703, + "step": 17090 + }, + { + "epoch": 0.6214114397848681, + "grad_norm": 5.805021286010742, + "learning_rate": 4.994022941550996e-05, + "loss": 0.2976, + "step": 17100 + }, + { + "epoch": 0.6217748382876662, + "grad_norm": 0.9706230163574219, + "learning_rate": 4.993987025423241e-05, + "loss": 0.1454, + "step": 17110 + }, + { + "epoch": 0.6221382367904644, + "grad_norm": 1.4393014907836914, + "learning_rate": 4.993951001838459e-05, + "loss": 0.1496, + "step": 17120 + }, + { + "epoch": 0.6225016352932626, + "grad_norm": 1.839086651802063, + "learning_rate": 4.993914870798202e-05, + "loss": 0.2256, + "step": 17130 + }, + { + "epoch": 0.6228650337960607, + "grad_norm": 1.8924603462219238, + "learning_rate": 4.993878632304027e-05, + "loss": 0.1415, + "step": 17140 + }, + { + "epoch": 0.6232284322988589, + "grad_norm": 12.03149700164795, + "learning_rate": 4.993842286357494e-05, + "loss": 0.7236, + "step": 17150 + }, + { + "epoch": 0.6235918308016571, + "grad_norm": 2.0251877307891846, + "learning_rate": 4.993805832960171e-05, + "loss": 0.1913, + "step": 17160 + }, + { + "epoch": 0.6239552293044552, + "grad_norm": 2.341251850128174, + "learning_rate": 4.993769272113628e-05, + "loss": 0.1734, + "step": 17170 + }, + { + "epoch": 0.6243186278072534, + "grad_norm": 2.517820358276367, + "learning_rate": 4.993732603819438e-05, + "loss": 0.18, + "step": 17180 + }, + { + "epoch": 0.6246820263100517, + "grad_norm": 1.6384356021881104, + "learning_rate": 4.993695828079184e-05, + "loss": 0.1513, + "step": 17190 + }, + { + "epoch": 0.6250454248128497, + "grad_norm": 10.794693946838379, + "learning_rate": 4.993658944894449e-05, + "loss": 0.2282, + "step": 17200 + }, + { + "epoch": 0.625408823315648, + "grad_norm": 1.2552087306976318, + "learning_rate": 4.9936219542668236e-05, + "loss": 0.1938, + "step": 17210 + }, + { + "epoch": 0.6257722218184462, + "grad_norm": 2.423431634902954, + "learning_rate": 4.993584856197899e-05, + "loss": 0.1487, + "step": 17220 + }, + { + "epoch": 0.6261356203212443, + "grad_norm": 1.7924834489822388, + "learning_rate": 4.9935476506892763e-05, + "loss": 0.195, + "step": 17230 + }, + { + "epoch": 0.6264990188240425, + "grad_norm": 1.6521999835968018, + "learning_rate": 4.9935103377425566e-05, + "loss": 0.1652, + "step": 17240 + }, + { + "epoch": 0.6268624173268406, + "grad_norm": 6.472127437591553, + "learning_rate": 4.9934729173593494e-05, + "loss": 0.2481, + "step": 17250 + }, + { + "epoch": 0.6272258158296388, + "grad_norm": 1.8962410688400269, + "learning_rate": 4.993435389541265e-05, + "loss": 0.1487, + "step": 17260 + }, + { + "epoch": 0.627589214332437, + "grad_norm": 1.2054486274719238, + "learning_rate": 4.993397754289922e-05, + "loss": 0.1496, + "step": 17270 + }, + { + "epoch": 0.6279526128352351, + "grad_norm": 3.9840786457061768, + "learning_rate": 4.993360011606941e-05, + "loss": 0.1776, + "step": 17280 + }, + { + "epoch": 0.6283160113380333, + "grad_norm": 0.9625970125198364, + "learning_rate": 4.9933221614939485e-05, + "loss": 0.1652, + "step": 17290 + }, + { + "epoch": 0.6286794098408315, + "grad_norm": 11.166252136230469, + "learning_rate": 4.993284203952575e-05, + "loss": 0.233, + "step": 17300 + }, + { + "epoch": 0.6290428083436296, + "grad_norm": 2.356268882751465, + "learning_rate": 4.9932461389844566e-05, + "loss": 0.1498, + "step": 17310 + }, + { + "epoch": 0.6294062068464278, + "grad_norm": 0.9366337656974792, + "learning_rate": 4.993207966591234e-05, + "loss": 0.1483, + "step": 17320 + }, + { + "epoch": 0.629769605349226, + "grad_norm": 5.854847431182861, + "learning_rate": 4.9931696867745495e-05, + "loss": 0.1603, + "step": 17330 + }, + { + "epoch": 0.6301330038520241, + "grad_norm": 1.0090773105621338, + "learning_rate": 4.9931312995360546e-05, + "loss": 0.1475, + "step": 17340 + }, + { + "epoch": 0.6304964023548223, + "grad_norm": 3.896676540374756, + "learning_rate": 4.9930928048774024e-05, + "loss": 0.244, + "step": 17350 + }, + { + "epoch": 0.6308598008576205, + "grad_norm": 1.1872800588607788, + "learning_rate": 4.993054202800252e-05, + "loss": 0.1618, + "step": 17360 + }, + { + "epoch": 0.6312231993604186, + "grad_norm": 1.8078994750976562, + "learning_rate": 4.9930154933062654e-05, + "loss": 0.1554, + "step": 17370 + }, + { + "epoch": 0.6315865978632168, + "grad_norm": 1.8264563083648682, + "learning_rate": 4.9929766763971126e-05, + "loss": 0.162, + "step": 17380 + }, + { + "epoch": 0.631949996366015, + "grad_norm": 0.6304519176483154, + "learning_rate": 4.992937752074465e-05, + "loss": 0.209, + "step": 17390 + }, + { + "epoch": 0.6323133948688131, + "grad_norm": 4.7621917724609375, + "learning_rate": 4.992898720339998e-05, + "loss": 0.2393, + "step": 17400 + }, + { + "epoch": 0.6323133948688131, + "eval_loss": 0.3943130671977997, + "eval_runtime": 180.0553, + "eval_samples_per_second": 41.176, + "eval_steps_per_second": 5.148, + "eval_wer": 0.21144734692395664, + "step": 17400 + }, + { + "epoch": 0.6326767933716113, + "grad_norm": 1.1110138893127441, + "learning_rate": 4.992859581195396e-05, + "loss": 0.143, + "step": 17410 + }, + { + "epoch": 0.6330401918744095, + "grad_norm": 1.2453794479370117, + "learning_rate": 4.992820334642344e-05, + "loss": 0.1454, + "step": 17420 + }, + { + "epoch": 0.6334035903772076, + "grad_norm": 3.669144630432129, + "learning_rate": 4.9927809806825335e-05, + "loss": 0.2496, + "step": 17430 + }, + { + "epoch": 0.6337669888800058, + "grad_norm": 2.7898483276367188, + "learning_rate": 4.99274151931766e-05, + "loss": 0.1614, + "step": 17440 + }, + { + "epoch": 0.634130387382804, + "grad_norm": 6.725431442260742, + "learning_rate": 4.992701950549423e-05, + "loss": 0.2622, + "step": 17450 + }, + { + "epoch": 0.6344937858856021, + "grad_norm": 1.6481575965881348, + "learning_rate": 4.992662274379528e-05, + "loss": 0.1713, + "step": 17460 + }, + { + "epoch": 0.6348571843884003, + "grad_norm": 1.3567384481430054, + "learning_rate": 4.9926224908096856e-05, + "loss": 0.1725, + "step": 17470 + }, + { + "epoch": 0.6352205828911985, + "grad_norm": 1.8207722902297974, + "learning_rate": 4.9925825998416076e-05, + "loss": 0.1973, + "step": 17480 + }, + { + "epoch": 0.6355839813939966, + "grad_norm": 2.2345893383026123, + "learning_rate": 4.9925426014770146e-05, + "loss": 0.1847, + "step": 17490 + }, + { + "epoch": 0.6359473798967948, + "grad_norm": 7.193591594696045, + "learning_rate": 4.992502495717629e-05, + "loss": 0.2605, + "step": 17500 + }, + { + "epoch": 0.636310778399593, + "grad_norm": 1.346073865890503, + "learning_rate": 4.99246228256518e-05, + "loss": 0.1518, + "step": 17510 + }, + { + "epoch": 0.6366741769023911, + "grad_norm": 1.5637879371643066, + "learning_rate": 4.9924219620213995e-05, + "loss": 0.1648, + "step": 17520 + }, + { + "epoch": 0.6370375754051893, + "grad_norm": 3.2450170516967773, + "learning_rate": 4.9923815340880236e-05, + "loss": 0.1974, + "step": 17530 + }, + { + "epoch": 0.6374009739079874, + "grad_norm": 0.9553948640823364, + "learning_rate": 4.992340998766796e-05, + "loss": 0.1694, + "step": 17540 + }, + { + "epoch": 0.6377643724107857, + "grad_norm": 8.901055335998535, + "learning_rate": 4.9923003560594625e-05, + "loss": 0.2625, + "step": 17550 + }, + { + "epoch": 0.6381277709135839, + "grad_norm": 1.7500522136688232, + "learning_rate": 4.992259605967774e-05, + "loss": 0.1799, + "step": 17560 + }, + { + "epoch": 0.638491169416382, + "grad_norm": 1.4673160314559937, + "learning_rate": 4.9922187484934865e-05, + "loss": 0.1698, + "step": 17570 + }, + { + "epoch": 0.6388545679191802, + "grad_norm": 2.5377135276794434, + "learning_rate": 4.992177783638361e-05, + "loss": 0.1822, + "step": 17580 + }, + { + "epoch": 0.6392179664219784, + "grad_norm": 1.660311222076416, + "learning_rate": 4.9921367114041625e-05, + "loss": 0.1659, + "step": 17590 + }, + { + "epoch": 0.6395813649247765, + "grad_norm": 8.248649597167969, + "learning_rate": 4.9920955317926595e-05, + "loss": 0.2384, + "step": 17600 + }, + { + "epoch": 0.6399447634275747, + "grad_norm": 1.5581409931182861, + "learning_rate": 4.992054244805627e-05, + "loss": 0.1665, + "step": 17610 + }, + { + "epoch": 0.6403081619303729, + "grad_norm": 0.9654737710952759, + "learning_rate": 4.992012850444844e-05, + "loss": 0.3493, + "step": 17620 + }, + { + "epoch": 0.640671560433171, + "grad_norm": 3.4477317333221436, + "learning_rate": 4.9919713487120935e-05, + "loss": 0.2097, + "step": 17630 + }, + { + "epoch": 0.6410349589359692, + "grad_norm": 1.3745356798171997, + "learning_rate": 4.9919297396091634e-05, + "loss": 0.1459, + "step": 17640 + }, + { + "epoch": 0.6413983574387674, + "grad_norm": 4.813534259796143, + "learning_rate": 4.991888023137849e-05, + "loss": 0.1905, + "step": 17650 + }, + { + "epoch": 0.6417617559415655, + "grad_norm": 3.118452310562134, + "learning_rate": 4.9918461992999445e-05, + "loss": 0.1527, + "step": 17660 + }, + { + "epoch": 0.6421251544443637, + "grad_norm": 1.8424941301345825, + "learning_rate": 4.991804268097253e-05, + "loss": 0.1759, + "step": 17670 + }, + { + "epoch": 0.6424885529471619, + "grad_norm": 7.301458835601807, + "learning_rate": 4.9917622295315826e-05, + "loss": 0.1662, + "step": 17680 + }, + { + "epoch": 0.64285195144996, + "grad_norm": 3.133114814758301, + "learning_rate": 4.991720083604743e-05, + "loss": 0.1692, + "step": 17690 + }, + { + "epoch": 0.6432153499527582, + "grad_norm": 11.538620948791504, + "learning_rate": 4.99167783031855e-05, + "loss": 0.2443, + "step": 17700 + }, + { + "epoch": 0.6435787484555564, + "grad_norm": 1.3739595413208008, + "learning_rate": 4.991635469674825e-05, + "loss": 0.1465, + "step": 17710 + }, + { + "epoch": 0.6439421469583545, + "grad_norm": 1.6855549812316895, + "learning_rate": 4.991593001675393e-05, + "loss": 0.1819, + "step": 17720 + }, + { + "epoch": 0.6443055454611527, + "grad_norm": 1.692335844039917, + "learning_rate": 4.991550426322083e-05, + "loss": 0.1654, + "step": 17730 + }, + { + "epoch": 0.6446689439639509, + "grad_norm": 1.1132971048355103, + "learning_rate": 4.9915077436167313e-05, + "loss": 0.1688, + "step": 17740 + }, + { + "epoch": 0.645032342466749, + "grad_norm": 5.6813201904296875, + "learning_rate": 4.9914649535611756e-05, + "loss": 0.2235, + "step": 17750 + }, + { + "epoch": 0.6453957409695472, + "grad_norm": 1.5107471942901611, + "learning_rate": 4.99142205615726e-05, + "loss": 0.1747, + "step": 17760 + }, + { + "epoch": 0.6457591394723454, + "grad_norm": 2.4552764892578125, + "learning_rate": 4.9913790514068316e-05, + "loss": 0.1739, + "step": 17770 + }, + { + "epoch": 0.6461225379751435, + "grad_norm": 1.5664808750152588, + "learning_rate": 4.991335939311744e-05, + "loss": 0.1766, + "step": 17780 + }, + { + "epoch": 0.6464859364779417, + "grad_norm": 2.935850143432617, + "learning_rate": 4.9912927198738556e-05, + "loss": 0.2148, + "step": 17790 + }, + { + "epoch": 0.6468493349807399, + "grad_norm": 10.267364501953125, + "learning_rate": 4.991249393095028e-05, + "loss": 0.2521, + "step": 17800 + }, + { + "epoch": 0.647212733483538, + "grad_norm": 1.3392564058303833, + "learning_rate": 4.9912059589771274e-05, + "loss": 0.172, + "step": 17810 + }, + { + "epoch": 0.6475761319863362, + "grad_norm": 0.895491361618042, + "learning_rate": 4.991162417522026e-05, + "loss": 0.1379, + "step": 17820 + }, + { + "epoch": 0.6479395304891343, + "grad_norm": 2.536397695541382, + "learning_rate": 4.9911187687315997e-05, + "loss": 0.1477, + "step": 17830 + }, + { + "epoch": 0.6483029289919325, + "grad_norm": 1.7795464992523193, + "learning_rate": 4.9910750126077296e-05, + "loss": 0.1786, + "step": 17840 + }, + { + "epoch": 0.6486663274947307, + "grad_norm": 62.683929443359375, + "learning_rate": 4.9910311491523e-05, + "loss": 0.266, + "step": 17850 + }, + { + "epoch": 0.6490297259975288, + "grad_norm": 2.0866358280181885, + "learning_rate": 4.990987178367201e-05, + "loss": 0.1428, + "step": 17860 + }, + { + "epoch": 0.649393124500327, + "grad_norm": 1.5636661052703857, + "learning_rate": 4.990943100254328e-05, + "loss": 0.1845, + "step": 17870 + }, + { + "epoch": 0.6497565230031253, + "grad_norm": 3.540689468383789, + "learning_rate": 4.9908989148155796e-05, + "loss": 0.2348, + "step": 17880 + }, + { + "epoch": 0.6501199215059233, + "grad_norm": 1.720421314239502, + "learning_rate": 4.990854622052859e-05, + "loss": 0.1742, + "step": 17890 + }, + { + "epoch": 0.6504833200087216, + "grad_norm": 7.7201056480407715, + "learning_rate": 4.9908102219680756e-05, + "loss": 0.2573, + "step": 17900 + }, + { + "epoch": 0.6508467185115198, + "grad_norm": 3.826190948486328, + "learning_rate": 4.9907701701329876e-05, + "loss": 3.6024, + "step": 17910 + }, + { + "epoch": 0.6512101170143179, + "grad_norm": 2.047307252883911, + "learning_rate": 4.990725566141558e-05, + "loss": 0.1551, + "step": 17920 + }, + { + "epoch": 0.6515735155171161, + "grad_norm": 6.462743282318115, + "learning_rate": 4.990680854833626e-05, + "loss": 0.2109, + "step": 17930 + }, + { + "epoch": 0.6519369140199143, + "grad_norm": 1.7611109018325806, + "learning_rate": 4.9906360362111184e-05, + "loss": 0.1959, + "step": 17940 + }, + { + "epoch": 0.6523003125227124, + "grad_norm": 5.253514766693115, + "learning_rate": 4.9905911102759655e-05, + "loss": 0.2436, + "step": 17950 + }, + { + "epoch": 0.6526637110255106, + "grad_norm": 0.9357771873474121, + "learning_rate": 4.9905460770301035e-05, + "loss": 0.1664, + "step": 17960 + }, + { + "epoch": 0.6530271095283088, + "grad_norm": 1.219488263130188, + "learning_rate": 4.990500936475472e-05, + "loss": 0.2286, + "step": 17970 + }, + { + "epoch": 0.6533905080311069, + "grad_norm": 2.8499608039855957, + "learning_rate": 4.990455688614016e-05, + "loss": 0.2664, + "step": 17980 + }, + { + "epoch": 0.6537539065339051, + "grad_norm": 1.5652077198028564, + "learning_rate": 4.990410333447686e-05, + "loss": 0.1341, + "step": 17990 + }, + { + "epoch": 0.6541173050367033, + "grad_norm": 5.98219633102417, + "learning_rate": 4.9903648709784356e-05, + "loss": 0.2338, + "step": 18000 + }, + { + "epoch": 0.6541173050367033, + "eval_loss": 0.37892404198646545, + "eval_runtime": 180.0524, + "eval_samples_per_second": 41.177, + "eval_steps_per_second": 5.149, + "eval_wer": 0.200864087715795, + "step": 18000 + }, + { + "epoch": 0.6544807035395014, + "grad_norm": 2.105100154876709, + "learning_rate": 4.990319301208223e-05, + "loss": 0.1764, + "step": 18010 + }, + { + "epoch": 0.6548441020422996, + "grad_norm": 1.0867921113967896, + "learning_rate": 4.990273624139013e-05, + "loss": 0.1507, + "step": 18020 + }, + { + "epoch": 0.6552075005450978, + "grad_norm": 2.9895503520965576, + "learning_rate": 4.9902278397727734e-05, + "loss": 0.1479, + "step": 18030 + }, + { + "epoch": 0.6555708990478959, + "grad_norm": 0.9947407841682434, + "learning_rate": 4.990181948111475e-05, + "loss": 0.1558, + "step": 18040 + }, + { + "epoch": 0.6559342975506941, + "grad_norm": 7.774895191192627, + "learning_rate": 4.9901359491570974e-05, + "loss": 0.2202, + "step": 18050 + }, + { + "epoch": 0.6562976960534923, + "grad_norm": 1.8466017246246338, + "learning_rate": 4.990089842911622e-05, + "loss": 0.1929, + "step": 18060 + }, + { + "epoch": 0.6566610945562904, + "grad_norm": 0.8435410261154175, + "learning_rate": 4.9900436293770345e-05, + "loss": 0.1377, + "step": 18070 + }, + { + "epoch": 0.6570244930590886, + "grad_norm": 3.10648512840271, + "learning_rate": 4.989997308555326e-05, + "loss": 0.202, + "step": 18080 + }, + { + "epoch": 0.6573878915618868, + "grad_norm": 1.112806797027588, + "learning_rate": 4.989950880448494e-05, + "loss": 0.1486, + "step": 18090 + }, + { + "epoch": 0.6577512900646849, + "grad_norm": 18.821117401123047, + "learning_rate": 4.989904345058538e-05, + "loss": 0.2677, + "step": 18100 + }, + { + "epoch": 0.6581146885674831, + "grad_norm": 1.254798412322998, + "learning_rate": 4.989857702387463e-05, + "loss": 3.5769, + "step": 18110 + }, + { + "epoch": 0.6584780870702812, + "grad_norm": 0.9956761002540588, + "learning_rate": 4.989810952437277e-05, + "loss": 0.1958, + "step": 18120 + }, + { + "epoch": 0.6588414855730794, + "grad_norm": 2.9471828937530518, + "learning_rate": 4.9897640952099975e-05, + "loss": 0.1988, + "step": 18130 + }, + { + "epoch": 0.6592048840758776, + "grad_norm": 1.3806344270706177, + "learning_rate": 4.989717130707641e-05, + "loss": 0.1552, + "step": 18140 + }, + { + "epoch": 0.6595682825786757, + "grad_norm": 3.0857722759246826, + "learning_rate": 4.989670058932231e-05, + "loss": 0.2168, + "step": 18150 + }, + { + "epoch": 0.6599316810814739, + "grad_norm": 1.8781664371490479, + "learning_rate": 4.989622879885798e-05, + "loss": 0.1571, + "step": 18160 + }, + { + "epoch": 0.6602950795842721, + "grad_norm": 1.1139156818389893, + "learning_rate": 4.9895755935703725e-05, + "loss": 0.1365, + "step": 18170 + }, + { + "epoch": 0.6606584780870702, + "grad_norm": 2.3965742588043213, + "learning_rate": 4.9895281999879925e-05, + "loss": 0.1879, + "step": 18180 + }, + { + "epoch": 0.6610218765898684, + "grad_norm": 1.2575726509094238, + "learning_rate": 4.9894806991407e-05, + "loss": 0.2197, + "step": 18190 + }, + { + "epoch": 0.6613852750926666, + "grad_norm": 10.392169952392578, + "learning_rate": 4.989433091030542e-05, + "loss": 0.2318, + "step": 18200 + }, + { + "epoch": 0.6617486735954647, + "grad_norm": 0.8268498182296753, + "learning_rate": 4.98938537565957e-05, + "loss": 0.1416, + "step": 18210 + }, + { + "epoch": 0.662112072098263, + "grad_norm": 0.9257369637489319, + "learning_rate": 4.9893375530298384e-05, + "loss": 0.1855, + "step": 18220 + }, + { + "epoch": 0.6624754706010612, + "grad_norm": 1.7720370292663574, + "learning_rate": 4.9892896231434094e-05, + "loss": 1.0276, + "step": 18230 + }, + { + "epoch": 0.6628388691038593, + "grad_norm": 2.2012548446655273, + "learning_rate": 4.9892415860023476e-05, + "loss": 1.1909, + "step": 18240 + }, + { + "epoch": 0.6632022676066575, + "grad_norm": 9.690247535705566, + "learning_rate": 4.9891934416087224e-05, + "loss": 0.2603, + "step": 18250 + }, + { + "epoch": 0.6635656661094557, + "grad_norm": 2.528682231903076, + "learning_rate": 4.989145189964608e-05, + "loss": 0.1912, + "step": 18260 + }, + { + "epoch": 0.6639290646122538, + "grad_norm": 1.4666227102279663, + "learning_rate": 4.989096831072084e-05, + "loss": 0.2316, + "step": 18270 + }, + { + "epoch": 0.664292463115052, + "grad_norm": 1.463526725769043, + "learning_rate": 4.989048364933234e-05, + "loss": 0.1388, + "step": 18280 + }, + { + "epoch": 0.6646558616178502, + "grad_norm": 1.2156569957733154, + "learning_rate": 4.988999791550146e-05, + "loss": 0.4086, + "step": 18290 + }, + { + "epoch": 0.6650192601206483, + "grad_norm": 4.909139156341553, + "learning_rate": 4.988951110924913e-05, + "loss": 0.2631, + "step": 18300 + }, + { + "epoch": 0.6653826586234465, + "grad_norm": 1.3692512512207031, + "learning_rate": 4.988902323059632e-05, + "loss": 0.1525, + "step": 18310 + }, + { + "epoch": 0.6657460571262447, + "grad_norm": 1.153344988822937, + "learning_rate": 4.988853427956406e-05, + "loss": 0.1904, + "step": 18320 + }, + { + "epoch": 0.6661094556290428, + "grad_norm": 2.052828073501587, + "learning_rate": 4.988804425617341e-05, + "loss": 0.1979, + "step": 18330 + }, + { + "epoch": 0.666472854131841, + "grad_norm": 1.373213768005371, + "learning_rate": 4.988755316044548e-05, + "loss": 0.1836, + "step": 18340 + }, + { + "epoch": 0.6668362526346392, + "grad_norm": 24.185970306396484, + "learning_rate": 4.9887060992401436e-05, + "loss": 0.2546, + "step": 18350 + }, + { + "epoch": 0.6671996511374373, + "grad_norm": 1.702205777168274, + "learning_rate": 4.988656775206248e-05, + "loss": 0.1433, + "step": 18360 + }, + { + "epoch": 0.6675630496402355, + "grad_norm": 2.279100179672241, + "learning_rate": 4.9886073439449864e-05, + "loss": 0.1671, + "step": 18370 + }, + { + "epoch": 0.6679264481430337, + "grad_norm": 3.928740978240967, + "learning_rate": 4.98855780545849e-05, + "loss": 0.1506, + "step": 18380 + }, + { + "epoch": 0.6682898466458318, + "grad_norm": 2.2895402908325195, + "learning_rate": 4.988508159748891e-05, + "loss": 0.1523, + "step": 18390 + }, + { + "epoch": 0.66865324514863, + "grad_norm": 10.151689529418945, + "learning_rate": 4.98845840681833e-05, + "loss": 0.2284, + "step": 18400 + }, + { + "epoch": 0.6690166436514282, + "grad_norm": 1.268561840057373, + "learning_rate": 4.9884085466689504e-05, + "loss": 0.171, + "step": 18410 + }, + { + "epoch": 0.6693800421542263, + "grad_norm": 1.0731265544891357, + "learning_rate": 4.9883585793029e-05, + "loss": 0.7778, + "step": 18420 + }, + { + "epoch": 0.6697434406570245, + "grad_norm": 1.0762509107589722, + "learning_rate": 4.988308504722332e-05, + "loss": 0.1966, + "step": 18430 + }, + { + "epoch": 0.6701068391598226, + "grad_norm": 6.763409614562988, + "learning_rate": 4.9882583229294044e-05, + "loss": 0.156, + "step": 18440 + }, + { + "epoch": 0.6704702376626208, + "grad_norm": 8.312501907348633, + "learning_rate": 4.988208033926279e-05, + "loss": 0.2573, + "step": 18450 + }, + { + "epoch": 0.670833636165419, + "grad_norm": 1.7566003799438477, + "learning_rate": 4.988157637715122e-05, + "loss": 0.1639, + "step": 18460 + }, + { + "epoch": 0.6711970346682171, + "grad_norm": 2.336911916732788, + "learning_rate": 4.988107134298105e-05, + "loss": 0.1536, + "step": 18470 + }, + { + "epoch": 0.6715604331710153, + "grad_norm": 2.2477078437805176, + "learning_rate": 4.988056523677405e-05, + "loss": 0.2734, + "step": 18480 + }, + { + "epoch": 0.6719238316738135, + "grad_norm": 1.62912917137146, + "learning_rate": 4.9880058058552015e-05, + "loss": 0.1501, + "step": 18490 + }, + { + "epoch": 0.6722872301766116, + "grad_norm": 8.896906852722168, + "learning_rate": 4.98795498083368e-05, + "loss": 0.213, + "step": 18500 + }, + { + "epoch": 0.6726506286794098, + "grad_norm": 1.804291009902954, + "learning_rate": 4.987904048615031e-05, + "loss": 0.2175, + "step": 18510 + }, + { + "epoch": 0.673014027182208, + "grad_norm": 0.9261330366134644, + "learning_rate": 4.9878530092014486e-05, + "loss": 0.1553, + "step": 18520 + }, + { + "epoch": 0.6733774256850061, + "grad_norm": 4.854642868041992, + "learning_rate": 4.987801862595132e-05, + "loss": 0.2065, + "step": 18530 + }, + { + "epoch": 0.6737408241878043, + "grad_norm": 0.9362125992774963, + "learning_rate": 4.987750608798284e-05, + "loss": 0.1611, + "step": 18540 + }, + { + "epoch": 0.6741042226906026, + "grad_norm": 13.348092079162598, + "learning_rate": 4.987699247813114e-05, + "loss": 0.2834, + "step": 18550 + }, + { + "epoch": 0.6744676211934006, + "grad_norm": 1.3235937356948853, + "learning_rate": 4.987647779641835e-05, + "loss": 0.166, + "step": 18560 + }, + { + "epoch": 0.6748310196961989, + "grad_norm": 1.7941697835922241, + "learning_rate": 4.987596204286664e-05, + "loss": 0.186, + "step": 18570 + }, + { + "epoch": 0.6751944181989971, + "grad_norm": 6.945876121520996, + "learning_rate": 4.987544521749824e-05, + "loss": 0.1859, + "step": 18580 + }, + { + "epoch": 0.6755578167017952, + "grad_norm": 1.1671024560928345, + "learning_rate": 4.98749273203354e-05, + "loss": 0.2007, + "step": 18590 + }, + { + "epoch": 0.6759212152045934, + "grad_norm": 46.817718505859375, + "learning_rate": 4.987440835140046e-05, + "loss": 0.275, + "step": 18600 + }, + { + "epoch": 0.6759212152045934, + "eval_loss": 0.4186328053474426, + "eval_runtime": 180.6066, + "eval_samples_per_second": 41.051, + "eval_steps_per_second": 5.133, + "eval_wer": 0.21444260896400238, + "step": 18600 + }, + { + "epoch": 0.6762846137073916, + "grad_norm": 0.9619908928871155, + "learning_rate": 4.987388831071575e-05, + "loss": 0.2147, + "step": 18610 + }, + { + "epoch": 0.6766480122101897, + "grad_norm": 1.139666199684143, + "learning_rate": 4.9873367198303714e-05, + "loss": 0.2591, + "step": 18620 + }, + { + "epoch": 0.6770114107129879, + "grad_norm": 2.6673026084899902, + "learning_rate": 4.9872845014186776e-05, + "loss": 0.2013, + "step": 18630 + }, + { + "epoch": 0.6773748092157861, + "grad_norm": 1.0486637353897095, + "learning_rate": 4.987232175838745e-05, + "loss": 0.2326, + "step": 18640 + }, + { + "epoch": 0.6777382077185842, + "grad_norm": 6.457462787628174, + "learning_rate": 4.987179743092827e-05, + "loss": 0.2395, + "step": 18650 + }, + { + "epoch": 0.6781016062213824, + "grad_norm": 3.296480178833008, + "learning_rate": 4.987127203183183e-05, + "loss": 0.1857, + "step": 18660 + }, + { + "epoch": 0.6784650047241806, + "grad_norm": 2.828460454940796, + "learning_rate": 4.987074556112078e-05, + "loss": 0.1391, + "step": 18670 + }, + { + "epoch": 0.6788284032269787, + "grad_norm": 10.424219131469727, + "learning_rate": 4.987021801881779e-05, + "loss": 0.1583, + "step": 18680 + }, + { + "epoch": 0.6791918017297769, + "grad_norm": 5.248502254486084, + "learning_rate": 4.986968940494559e-05, + "loss": 0.1676, + "step": 18690 + }, + { + "epoch": 0.6795552002325751, + "grad_norm": 8.20375919342041, + "learning_rate": 4.986915971952696e-05, + "loss": 0.2844, + "step": 18700 + }, + { + "epoch": 0.6799185987353732, + "grad_norm": 2.415562152862549, + "learning_rate": 4.986862896258473e-05, + "loss": 0.1634, + "step": 18710 + }, + { + "epoch": 0.6802819972381714, + "grad_norm": 1.635680079460144, + "learning_rate": 4.986809713414176e-05, + "loss": 0.509, + "step": 18720 + }, + { + "epoch": 0.6806453957409695, + "grad_norm": 10.641048431396484, + "learning_rate": 4.986756423422095e-05, + "loss": 0.2015, + "step": 18730 + }, + { + "epoch": 0.6810087942437677, + "grad_norm": 1.3304156064987183, + "learning_rate": 4.986703026284529e-05, + "loss": 0.1598, + "step": 18740 + }, + { + "epoch": 0.6813721927465659, + "grad_norm": 4.707154750823975, + "learning_rate": 4.986649522003778e-05, + "loss": 0.2486, + "step": 18750 + }, + { + "epoch": 0.681735591249364, + "grad_norm": 1.671863317489624, + "learning_rate": 4.9865959105821454e-05, + "loss": 0.1628, + "step": 18760 + }, + { + "epoch": 0.6820989897521622, + "grad_norm": 2.4183709621429443, + "learning_rate": 4.986542192021942e-05, + "loss": 0.1636, + "step": 18770 + }, + { + "epoch": 0.6824623882549604, + "grad_norm": 120.8931884765625, + "learning_rate": 4.9864883663254836e-05, + "loss": 2.0172, + "step": 18780 + }, + { + "epoch": 0.6828257867577585, + "grad_norm": 2.785879135131836, + "learning_rate": 4.986434433495089e-05, + "loss": 0.1669, + "step": 18790 + }, + { + "epoch": 0.6831891852605567, + "grad_norm": 3.662753105163574, + "learning_rate": 4.98638039353308e-05, + "loss": 0.3095, + "step": 18800 + }, + { + "epoch": 0.6835525837633549, + "grad_norm": 1.1632777452468872, + "learning_rate": 4.986326246441787e-05, + "loss": 0.1632, + "step": 18810 + }, + { + "epoch": 0.683915982266153, + "grad_norm": 0.9660913348197937, + "learning_rate": 4.986271992223543e-05, + "loss": 0.1509, + "step": 18820 + }, + { + "epoch": 0.6842793807689512, + "grad_norm": 2.810391426086426, + "learning_rate": 4.986217630880684e-05, + "loss": 0.1507, + "step": 18830 + }, + { + "epoch": 0.6846427792717494, + "grad_norm": 2.008641242980957, + "learning_rate": 4.986163162415554e-05, + "loss": 0.1858, + "step": 18840 + }, + { + "epoch": 0.6850061777745475, + "grad_norm": 3.4007887840270996, + "learning_rate": 4.986108586830499e-05, + "loss": 0.2389, + "step": 18850 + }, + { + "epoch": 0.6853695762773457, + "grad_norm": 0.8250002861022949, + "learning_rate": 4.986053904127871e-05, + "loss": 0.1618, + "step": 18860 + }, + { + "epoch": 0.685732974780144, + "grad_norm": 0.792607307434082, + "learning_rate": 4.986004598111927e-05, + "loss": 1.5835, + "step": 18870 + }, + { + "epoch": 0.686096373282942, + "grad_norm": 2.740478038787842, + "learning_rate": 4.985949711892404e-05, + "loss": 0.2021, + "step": 18880 + }, + { + "epoch": 0.6864597717857402, + "grad_norm": 1.1361775398254395, + "learning_rate": 4.985894718562153e-05, + "loss": 0.2244, + "step": 18890 + }, + { + "epoch": 0.6868231702885385, + "grad_norm": 2.692542314529419, + "learning_rate": 4.985839618123543e-05, + "loss": 0.2095, + "step": 18900 + }, + { + "epoch": 0.6871865687913365, + "grad_norm": 1.2691428661346436, + "learning_rate": 4.9857844105789485e-05, + "loss": 0.1533, + "step": 18910 + }, + { + "epoch": 0.6875499672941348, + "grad_norm": 2.087209939956665, + "learning_rate": 4.9857290959307483e-05, + "loss": 0.1469, + "step": 18920 + }, + { + "epoch": 0.687913365796933, + "grad_norm": 1.5252209901809692, + "learning_rate": 4.985673674181326e-05, + "loss": 0.2099, + "step": 18930 + }, + { + "epoch": 0.6882767642997311, + "grad_norm": 1.81588876247406, + "learning_rate": 4.9856181453330685e-05, + "loss": 0.174, + "step": 18940 + }, + { + "epoch": 0.6886401628025293, + "grad_norm": 21.244775772094727, + "learning_rate": 4.9855625093883695e-05, + "loss": 0.2455, + "step": 18950 + }, + { + "epoch": 0.6890035613053275, + "grad_norm": 1.53201425075531, + "learning_rate": 4.9855067663496255e-05, + "loss": 0.1731, + "step": 18960 + }, + { + "epoch": 0.6893669598081256, + "grad_norm": 0.9922922849655151, + "learning_rate": 4.985450916219239e-05, + "loss": 0.1569, + "step": 18970 + }, + { + "epoch": 0.6897303583109238, + "grad_norm": 1.6983296871185303, + "learning_rate": 4.985394958999615e-05, + "loss": 0.1784, + "step": 18980 + }, + { + "epoch": 0.690093756813722, + "grad_norm": 2.5069353580474854, + "learning_rate": 4.9853388946931654e-05, + "loss": 0.1484, + "step": 18990 + }, + { + "epoch": 0.6904571553165201, + "grad_norm": 52.345367431640625, + "learning_rate": 4.985282723302306e-05, + "loss": 0.2431, + "step": 19000 + }, + { + "epoch": 0.6908205538193183, + "grad_norm": 1.5318138599395752, + "learning_rate": 4.9852264448294564e-05, + "loss": 0.1662, + "step": 19010 + }, + { + "epoch": 0.6911839523221164, + "grad_norm": 1.5980876684188843, + "learning_rate": 4.985170059277041e-05, + "loss": 1.3532, + "step": 19020 + }, + { + "epoch": 0.6915473508249146, + "grad_norm": 2.355023145675659, + "learning_rate": 4.9851135666474915e-05, + "loss": 0.1688, + "step": 19030 + }, + { + "epoch": 0.6919107493277128, + "grad_norm": 3.2141480445861816, + "learning_rate": 4.98505696694324e-05, + "loss": 0.1303, + "step": 19040 + }, + { + "epoch": 0.6922741478305109, + "grad_norm": 19.482290267944336, + "learning_rate": 4.985000260166725e-05, + "loss": 0.2337, + "step": 19050 + }, + { + "epoch": 0.6926375463333091, + "grad_norm": 0.8456101417541504, + "learning_rate": 4.9849434463203915e-05, + "loss": 0.1732, + "step": 19060 + }, + { + "epoch": 0.6930009448361073, + "grad_norm": 2.2158889770507812, + "learning_rate": 4.9848865254066856e-05, + "loss": 0.1524, + "step": 19070 + }, + { + "epoch": 0.6933643433389054, + "grad_norm": 2.0843331813812256, + "learning_rate": 4.9848294974280605e-05, + "loss": 0.1943, + "step": 19080 + }, + { + "epoch": 0.6937277418417036, + "grad_norm": 2.6970462799072266, + "learning_rate": 4.9847723623869734e-05, + "loss": 0.1697, + "step": 19090 + }, + { + "epoch": 0.6940911403445018, + "grad_norm": 9.394730567932129, + "learning_rate": 4.984715120285887e-05, + "loss": 0.2151, + "step": 19100 + }, + { + "epoch": 0.6944545388472999, + "grad_norm": 1.922090768814087, + "learning_rate": 4.9846577711272656e-05, + "loss": 0.1737, + "step": 19110 + }, + { + "epoch": 0.6948179373500981, + "grad_norm": 1.3870245218276978, + "learning_rate": 4.9846003149135815e-05, + "loss": 0.1694, + "step": 19120 + }, + { + "epoch": 0.6951813358528963, + "grad_norm": 1.6474970579147339, + "learning_rate": 4.9845427516473104e-05, + "loss": 0.219, + "step": 19130 + }, + { + "epoch": 0.6955447343556944, + "grad_norm": 1.4302411079406738, + "learning_rate": 4.984485081330932e-05, + "loss": 0.1489, + "step": 19140 + }, + { + "epoch": 0.6959081328584926, + "grad_norm": 3.888967990875244, + "learning_rate": 4.984427303966932e-05, + "loss": 0.2425, + "step": 19150 + }, + { + "epoch": 0.6962715313612908, + "grad_norm": 1.2002874612808228, + "learning_rate": 4.984369419557798e-05, + "loss": 0.1575, + "step": 19160 + }, + { + "epoch": 0.6966349298640889, + "grad_norm": 1.9064863920211792, + "learning_rate": 4.984311428106025e-05, + "loss": 0.1526, + "step": 19170 + }, + { + "epoch": 0.6969983283668871, + "grad_norm": 1.3838772773742676, + "learning_rate": 4.984253329614112e-05, + "loss": 0.1601, + "step": 19180 + }, + { + "epoch": 0.6973617268696853, + "grad_norm": 3.6261801719665527, + "learning_rate": 4.984195124084563e-05, + "loss": 0.1668, + "step": 19190 + }, + { + "epoch": 0.6977251253724834, + "grad_norm": 7.647263526916504, + "learning_rate": 4.984136811519884e-05, + "loss": 0.1879, + "step": 19200 + }, + { + "epoch": 0.6977251253724834, + "eval_loss": 0.3865276575088501, + "eval_runtime": 179.6651, + "eval_samples_per_second": 41.266, + "eval_steps_per_second": 5.16, + "eval_wer": 0.20815255867990634, + "step": 19200 + }, + { + "epoch": 0.6980885238752816, + "grad_norm": 1.7563225030899048, + "learning_rate": 4.984078391922589e-05, + "loss": 0.1481, + "step": 19210 + }, + { + "epoch": 0.6984519223780798, + "grad_norm": 1.8016029596328735, + "learning_rate": 4.984019865295194e-05, + "loss": 0.1713, + "step": 19220 + }, + { + "epoch": 0.6988153208808779, + "grad_norm": 2.0969181060791016, + "learning_rate": 4.983961231640221e-05, + "loss": 0.1959, + "step": 19230 + }, + { + "epoch": 0.6991787193836762, + "grad_norm": 1.6823608875274658, + "learning_rate": 4.9839024909601964e-05, + "loss": 0.1729, + "step": 19240 + }, + { + "epoch": 0.6995421178864744, + "grad_norm": 11.533753395080566, + "learning_rate": 4.983843643257652e-05, + "loss": 0.2264, + "step": 19250 + }, + { + "epoch": 0.6999055163892725, + "grad_norm": 4.1039204597473145, + "learning_rate": 4.983784688535122e-05, + "loss": 0.1738, + "step": 19260 + }, + { + "epoch": 0.7002689148920707, + "grad_norm": 1.1051629781723022, + "learning_rate": 4.983725626795147e-05, + "loss": 0.155, + "step": 19270 + }, + { + "epoch": 0.7006323133948689, + "grad_norm": 4.303994178771973, + "learning_rate": 4.983666458040273e-05, + "loss": 0.1593, + "step": 19280 + }, + { + "epoch": 0.700995711897667, + "grad_norm": 1.2324292659759521, + "learning_rate": 4.983607182273047e-05, + "loss": 0.1642, + "step": 19290 + }, + { + "epoch": 0.7013591104004652, + "grad_norm": 6.101926326751709, + "learning_rate": 4.983547799496024e-05, + "loss": 0.2338, + "step": 19300 + }, + { + "epoch": 0.7017225089032633, + "grad_norm": 1.1532049179077148, + "learning_rate": 4.983488309711763e-05, + "loss": 0.1591, + "step": 19310 + }, + { + "epoch": 0.7020859074060615, + "grad_norm": 0.8216233253479004, + "learning_rate": 4.983428712922828e-05, + "loss": 0.1489, + "step": 19320 + }, + { + "epoch": 0.7024493059088597, + "grad_norm": 1.489461064338684, + "learning_rate": 4.983369009131785e-05, + "loss": 0.2048, + "step": 19330 + }, + { + "epoch": 0.7028127044116578, + "grad_norm": 1.0493615865707397, + "learning_rate": 4.983309198341207e-05, + "loss": 0.1525, + "step": 19340 + }, + { + "epoch": 0.703176102914456, + "grad_norm": 10.2578706741333, + "learning_rate": 4.983249280553672e-05, + "loss": 0.2297, + "step": 19350 + }, + { + "epoch": 0.7035395014172542, + "grad_norm": 1.5366660356521606, + "learning_rate": 4.983189255771761e-05, + "loss": 0.1644, + "step": 19360 + }, + { + "epoch": 0.7039028999200523, + "grad_norm": 1.4915844202041626, + "learning_rate": 4.9831291239980596e-05, + "loss": 0.1599, + "step": 19370 + }, + { + "epoch": 0.7042662984228505, + "grad_norm": 1.3012590408325195, + "learning_rate": 4.98306888523516e-05, + "loss": 0.1907, + "step": 19380 + }, + { + "epoch": 0.7046296969256487, + "grad_norm": 1.5029476881027222, + "learning_rate": 4.983008539485656e-05, + "loss": 0.1391, + "step": 19390 + }, + { + "epoch": 0.7049930954284468, + "grad_norm": 3.0202033519744873, + "learning_rate": 4.9829480867521495e-05, + "loss": 0.2218, + "step": 19400 + }, + { + "epoch": 0.705356493931245, + "grad_norm": 1.7761317491531372, + "learning_rate": 4.9828875270372434e-05, + "loss": 0.1605, + "step": 19410 + }, + { + "epoch": 0.7057198924340432, + "grad_norm": 1.420793890953064, + "learning_rate": 4.9828268603435485e-05, + "loss": 1.5838, + "step": 19420 + }, + { + "epoch": 0.7060832909368413, + "grad_norm": 2.079665422439575, + "learning_rate": 4.982766086673678e-05, + "loss": 0.2146, + "step": 19430 + }, + { + "epoch": 0.7064466894396395, + "grad_norm": 2.440471887588501, + "learning_rate": 4.98270520603025e-05, + "loss": 0.1733, + "step": 19440 + }, + { + "epoch": 0.7068100879424377, + "grad_norm": 7.773731708526611, + "learning_rate": 4.982644218415889e-05, + "loss": 0.2126, + "step": 19450 + }, + { + "epoch": 0.7071734864452358, + "grad_norm": 0.9480405449867249, + "learning_rate": 4.982583123833221e-05, + "loss": 0.1575, + "step": 19460 + }, + { + "epoch": 0.707536884948034, + "grad_norm": 12.79196548461914, + "learning_rate": 4.982521922284881e-05, + "loss": 0.2745, + "step": 19470 + }, + { + "epoch": 0.7079002834508322, + "grad_norm": 4.492150783538818, + "learning_rate": 4.982460613773502e-05, + "loss": 0.1663, + "step": 19480 + }, + { + "epoch": 0.7082636819536303, + "grad_norm": 1.2373683452606201, + "learning_rate": 4.9823991983017295e-05, + "loss": 0.1699, + "step": 19490 + }, + { + "epoch": 0.7086270804564285, + "grad_norm": 5.8804402351379395, + "learning_rate": 4.982337675872207e-05, + "loss": 0.242, + "step": 19500 + }, + { + "epoch": 0.7089904789592267, + "grad_norm": 0.9465837478637695, + "learning_rate": 4.982276046487586e-05, + "loss": 0.1471, + "step": 19510 + }, + { + "epoch": 0.7093538774620248, + "grad_norm": 1.6178842782974243, + "learning_rate": 4.9822143101505226e-05, + "loss": 0.1619, + "step": 19520 + }, + { + "epoch": 0.709717275964823, + "grad_norm": 2.4963414669036865, + "learning_rate": 4.9821524668636766e-05, + "loss": 0.1426, + "step": 19530 + }, + { + "epoch": 0.7100806744676212, + "grad_norm": 1.1380610466003418, + "learning_rate": 4.982090516629712e-05, + "loss": 0.2364, + "step": 19540 + }, + { + "epoch": 0.7104440729704193, + "grad_norm": 5.2998046875, + "learning_rate": 4.982028459451298e-05, + "loss": 0.2661, + "step": 19550 + }, + { + "epoch": 0.7108074714732175, + "grad_norm": 1.1476637125015259, + "learning_rate": 4.9819662953311096e-05, + "loss": 0.1306, + "step": 19560 + }, + { + "epoch": 0.7111708699760158, + "grad_norm": 0.7960777878761292, + "learning_rate": 4.981904024271824e-05, + "loss": 0.1604, + "step": 19570 + }, + { + "epoch": 0.7115342684788138, + "grad_norm": 1.9035999774932861, + "learning_rate": 4.981841646276124e-05, + "loss": 0.1728, + "step": 19580 + }, + { + "epoch": 0.711897666981612, + "grad_norm": 0.9725393056869507, + "learning_rate": 4.981779161346699e-05, + "loss": 0.2529, + "step": 19590 + }, + { + "epoch": 0.7122610654844101, + "grad_norm": 5.759589672088623, + "learning_rate": 4.98171656948624e-05, + "loss": 0.25, + "step": 19600 + }, + { + "epoch": 0.7126244639872084, + "grad_norm": 1.3716357946395874, + "learning_rate": 4.9816538706974434e-05, + "loss": 0.1603, + "step": 19610 + }, + { + "epoch": 0.7129878624900066, + "grad_norm": 1.4253743886947632, + "learning_rate": 4.981591064983011e-05, + "loss": 0.1496, + "step": 19620 + }, + { + "epoch": 0.7133512609928047, + "grad_norm": 2.4253408908843994, + "learning_rate": 4.98152815234565e-05, + "loss": 0.1694, + "step": 19630 + }, + { + "epoch": 0.7137146594956029, + "grad_norm": 1.212689757347107, + "learning_rate": 4.9814651327880696e-05, + "loss": 0.1869, + "step": 19640 + }, + { + "epoch": 0.7140780579984011, + "grad_norm": 7.003270626068115, + "learning_rate": 4.981402006312986e-05, + "loss": 0.2709, + "step": 19650 + }, + { + "epoch": 0.7144414565011992, + "grad_norm": 1.6173512935638428, + "learning_rate": 4.981338772923119e-05, + "loss": 0.1651, + "step": 19660 + }, + { + "epoch": 0.7148048550039974, + "grad_norm": 2.2197723388671875, + "learning_rate": 4.981275432621192e-05, + "loss": 0.1657, + "step": 19670 + }, + { + "epoch": 0.7151682535067956, + "grad_norm": 1.8906898498535156, + "learning_rate": 4.981211985409936e-05, + "loss": 2.3111, + "step": 19680 + }, + { + "epoch": 0.7155316520095937, + "grad_norm": 3.50747013092041, + "learning_rate": 4.981148431292084e-05, + "loss": 0.1498, + "step": 19690 + }, + { + "epoch": 0.7158950505123919, + "grad_norm": 4.080805778503418, + "learning_rate": 4.981084770270373e-05, + "loss": 0.2094, + "step": 19700 + }, + { + "epoch": 0.7162584490151901, + "grad_norm": 2.1056652069091797, + "learning_rate": 4.981021002347547e-05, + "loss": 0.157, + "step": 19710 + }, + { + "epoch": 0.7166218475179882, + "grad_norm": 1.07776939868927, + "learning_rate": 4.980957127526354e-05, + "loss": 0.2049, + "step": 19720 + }, + { + "epoch": 0.7169852460207864, + "grad_norm": 3.5387072563171387, + "learning_rate": 4.980893145809546e-05, + "loss": 0.1706, + "step": 19730 + }, + { + "epoch": 0.7173486445235846, + "grad_norm": 1.5516027212142944, + "learning_rate": 4.980829057199879e-05, + "loss": 0.1371, + "step": 19740 + }, + { + "epoch": 0.7177120430263827, + "grad_norm": 6.618633270263672, + "learning_rate": 4.9807648617001145e-05, + "loss": 0.1833, + "step": 19750 + }, + { + "epoch": 0.7180754415291809, + "grad_norm": 1.7093079090118408, + "learning_rate": 4.980700559313019e-05, + "loss": 0.1592, + "step": 19760 + }, + { + "epoch": 0.7184388400319791, + "grad_norm": 1.1217936277389526, + "learning_rate": 4.9806361500413626e-05, + "loss": 0.145, + "step": 19770 + }, + { + "epoch": 0.7188022385347772, + "grad_norm": 1.869722604751587, + "learning_rate": 4.980571633887921e-05, + "loss": 0.1605, + "step": 19780 + }, + { + "epoch": 0.7191656370375754, + "grad_norm": 1.1555829048156738, + "learning_rate": 4.980507010855473e-05, + "loss": 0.1539, + "step": 19790 + }, + { + "epoch": 0.7195290355403736, + "grad_norm": 5.0145111083984375, + "learning_rate": 4.9804422809468046e-05, + "loss": 0.2334, + "step": 19800 + }, + { + "epoch": 0.7195290355403736, + "eval_loss": 0.394449919462204, + "eval_runtime": 180.0311, + "eval_samples_per_second": 41.182, + "eval_steps_per_second": 5.149, + "eval_wer": 0.2100677110752083, + "step": 19800 + }, + { + "epoch": 0.7198924340431717, + "grad_norm": 1.0865716934204102, + "learning_rate": 4.980377444164702e-05, + "loss": 0.1569, + "step": 19810 + }, + { + "epoch": 0.7202558325459699, + "grad_norm": 1.5475140810012817, + "learning_rate": 4.980312500511962e-05, + "loss": 0.1268, + "step": 19820 + }, + { + "epoch": 0.7206192310487681, + "grad_norm": 1.9507659673690796, + "learning_rate": 4.980247449991381e-05, + "loss": 0.2092, + "step": 19830 + }, + { + "epoch": 0.7209826295515662, + "grad_norm": 1.185339093208313, + "learning_rate": 4.980182292605762e-05, + "loss": 0.1432, + "step": 19840 + }, + { + "epoch": 0.7213460280543644, + "grad_norm": 5.294797420501709, + "learning_rate": 4.980117028357912e-05, + "loss": 0.2459, + "step": 19850 + }, + { + "epoch": 0.7217094265571626, + "grad_norm": 2.691941976547241, + "learning_rate": 4.980051657250645e-05, + "loss": 0.1747, + "step": 19860 + }, + { + "epoch": 0.7220728250599607, + "grad_norm": 1.3377537727355957, + "learning_rate": 4.9799861792867756e-05, + "loss": 0.1541, + "step": 19870 + }, + { + "epoch": 0.7224362235627589, + "grad_norm": 3.39907169342041, + "learning_rate": 4.979920594469124e-05, + "loss": 0.166, + "step": 19880 + }, + { + "epoch": 0.722799622065557, + "grad_norm": 1.738271951675415, + "learning_rate": 4.9798549028005195e-05, + "loss": 0.1591, + "step": 19890 + }, + { + "epoch": 0.7231630205683552, + "grad_norm": 4.062039852142334, + "learning_rate": 4.9797891042837893e-05, + "loss": 0.2372, + "step": 19900 + }, + { + "epoch": 0.7235264190711534, + "grad_norm": 2.46109676361084, + "learning_rate": 4.979723198921771e-05, + "loss": 0.1606, + "step": 19910 + }, + { + "epoch": 0.7238898175739515, + "grad_norm": 1.3511689901351929, + "learning_rate": 4.9796571867173017e-05, + "loss": 0.148, + "step": 19920 + }, + { + "epoch": 0.7242532160767498, + "grad_norm": 4.831977844238281, + "learning_rate": 4.979591067673227e-05, + "loss": 0.1832, + "step": 19930 + }, + { + "epoch": 0.724616614579548, + "grad_norm": 0.9530340433120728, + "learning_rate": 4.979524841792397e-05, + "loss": 0.1776, + "step": 19940 + }, + { + "epoch": 0.724980013082346, + "grad_norm": 2.886121988296509, + "learning_rate": 4.979458509077663e-05, + "loss": 0.217, + "step": 19950 + }, + { + "epoch": 0.7253434115851443, + "grad_norm": 2.6050822734832764, + "learning_rate": 4.979392069531883e-05, + "loss": 0.1709, + "step": 19960 + }, + { + "epoch": 0.7257068100879425, + "grad_norm": 1.1615772247314453, + "learning_rate": 4.979325523157921e-05, + "loss": 0.1891, + "step": 19970 + }, + { + "epoch": 0.7260702085907406, + "grad_norm": 5.947473526000977, + "learning_rate": 4.979258869958643e-05, + "loss": 0.1685, + "step": 19980 + }, + { + "epoch": 0.7264336070935388, + "grad_norm": 2.2721457481384277, + "learning_rate": 4.979192109936922e-05, + "loss": 0.1733, + "step": 19990 + }, + { + "epoch": 0.726797005596337, + "grad_norm": 2.83907413482666, + "learning_rate": 4.979125243095635e-05, + "loss": 0.2067, + "step": 20000 + }, + { + "epoch": 0.7271604040991351, + "grad_norm": 1.84774649143219, + "learning_rate": 4.9790582694376605e-05, + "loss": 0.1634, + "step": 20010 + }, + { + "epoch": 0.7275238026019333, + "grad_norm": 3.5162901878356934, + "learning_rate": 4.978991188965887e-05, + "loss": 0.1546, + "step": 20020 + }, + { + "epoch": 0.7278872011047315, + "grad_norm": 1.3396214246749878, + "learning_rate": 4.9789240016832026e-05, + "loss": 0.1549, + "step": 20030 + }, + { + "epoch": 0.7282505996075296, + "grad_norm": 0.8957159519195557, + "learning_rate": 4.978856707592503e-05, + "loss": 0.4856, + "step": 20040 + }, + { + "epoch": 0.7286139981103278, + "grad_norm": 3.291719913482666, + "learning_rate": 4.978789306696688e-05, + "loss": 0.1672, + "step": 20050 + }, + { + "epoch": 0.728977396613126, + "grad_norm": 1.2237446308135986, + "learning_rate": 4.978721798998661e-05, + "loss": 0.1547, + "step": 20060 + }, + { + "epoch": 0.7293407951159241, + "grad_norm": 1.5760120153427124, + "learning_rate": 4.978654184501331e-05, + "loss": 0.1491, + "step": 20070 + }, + { + "epoch": 0.7297041936187223, + "grad_norm": 2.661914587020874, + "learning_rate": 4.978586463207612e-05, + "loss": 0.2399, + "step": 20080 + }, + { + "epoch": 0.7300675921215205, + "grad_norm": 1.4015228748321533, + "learning_rate": 4.978518635120421e-05, + "loss": 0.1592, + "step": 20090 + }, + { + "epoch": 0.7304309906243186, + "grad_norm": 11.479881286621094, + "learning_rate": 4.9784507002426793e-05, + "loss": 0.2478, + "step": 20100 + }, + { + "epoch": 0.7307943891271168, + "grad_norm": 2.3282432556152344, + "learning_rate": 4.9783826585773164e-05, + "loss": 0.1565, + "step": 20110 + }, + { + "epoch": 0.731157787629915, + "grad_norm": 1.0281476974487305, + "learning_rate": 4.9783145101272625e-05, + "loss": 2.6872, + "step": 20120 + }, + { + "epoch": 0.7315211861327131, + "grad_norm": 1.4759191274642944, + "learning_rate": 4.978246254895455e-05, + "loss": 0.1755, + "step": 20130 + }, + { + "epoch": 0.7318845846355113, + "grad_norm": 1.1100878715515137, + "learning_rate": 4.978177892884833e-05, + "loss": 0.1519, + "step": 20140 + }, + { + "epoch": 0.7322479831383095, + "grad_norm": 5.326310157775879, + "learning_rate": 4.9781094240983435e-05, + "loss": 0.257, + "step": 20150 + }, + { + "epoch": 0.7326113816411076, + "grad_norm": 8.199230194091797, + "learning_rate": 4.978040848538936e-05, + "loss": 0.192, + "step": 20160 + }, + { + "epoch": 0.7329747801439058, + "grad_norm": 1.579663872718811, + "learning_rate": 4.9779721662095654e-05, + "loss": 0.1738, + "step": 20170 + }, + { + "epoch": 0.733338178646704, + "grad_norm": 3.319883346557617, + "learning_rate": 4.97790337711319e-05, + "loss": 0.1809, + "step": 20180 + }, + { + "epoch": 0.7337015771495021, + "grad_norm": 1.4813331365585327, + "learning_rate": 4.977834481252776e-05, + "loss": 0.1645, + "step": 20190 + }, + { + "epoch": 0.7340649756523003, + "grad_norm": 4.392731666564941, + "learning_rate": 4.9777654786312886e-05, + "loss": 0.1897, + "step": 20200 + }, + { + "epoch": 0.7344283741550984, + "grad_norm": 1.7336299419403076, + "learning_rate": 4.9776963692517034e-05, + "loss": 0.1751, + "step": 20210 + }, + { + "epoch": 0.7347917726578966, + "grad_norm": 1.6261765956878662, + "learning_rate": 4.977627153116998e-05, + "loss": 0.156, + "step": 20220 + }, + { + "epoch": 0.7351551711606948, + "grad_norm": 1.9801748991012573, + "learning_rate": 4.977557830230153e-05, + "loss": 0.2069, + "step": 20230 + }, + { + "epoch": 0.7355185696634929, + "grad_norm": 1.4615390300750732, + "learning_rate": 4.977488400594157e-05, + "loss": 0.1458, + "step": 20240 + }, + { + "epoch": 0.7358819681662911, + "grad_norm": 3.78981876373291, + "learning_rate": 4.977418864212e-05, + "loss": 0.1765, + "step": 20250 + }, + { + "epoch": 0.7362453666690894, + "grad_norm": 0.813947319984436, + "learning_rate": 4.97734922108668e-05, + "loss": 0.1482, + "step": 20260 + }, + { + "epoch": 0.7366087651718874, + "grad_norm": 1.1082271337509155, + "learning_rate": 4.977279471221195e-05, + "loss": 0.149, + "step": 20270 + }, + { + "epoch": 0.7369721636746857, + "grad_norm": 4.023866176605225, + "learning_rate": 4.9772096146185527e-05, + "loss": 0.1797, + "step": 20280 + }, + { + "epoch": 0.7373355621774839, + "grad_norm": 1.3649333715438843, + "learning_rate": 4.977139651281762e-05, + "loss": 0.182, + "step": 20290 + }, + { + "epoch": 0.737698960680282, + "grad_norm": 8.213293075561523, + "learning_rate": 4.977069581213837e-05, + "loss": 0.2117, + "step": 20300 + }, + { + "epoch": 0.7380623591830802, + "grad_norm": 1.0769990682601929, + "learning_rate": 4.9769994044177976e-05, + "loss": 0.1689, + "step": 20310 + }, + { + "epoch": 0.7384257576858784, + "grad_norm": 1.712949275970459, + "learning_rate": 4.9769291208966674e-05, + "loss": 0.1402, + "step": 20320 + }, + { + "epoch": 0.7387891561886765, + "grad_norm": 2.213164806365967, + "learning_rate": 4.976858730653473e-05, + "loss": 0.193, + "step": 20330 + }, + { + "epoch": 0.7391525546914747, + "grad_norm": 1.9228605031967163, + "learning_rate": 4.97678823369125e-05, + "loss": 0.1517, + "step": 20340 + }, + { + "epoch": 0.7395159531942729, + "grad_norm": 8.813825607299805, + "learning_rate": 4.976717630013034e-05, + "loss": 0.2682, + "step": 20350 + }, + { + "epoch": 0.739879351697071, + "grad_norm": 1.9778189659118652, + "learning_rate": 4.976646919621867e-05, + "loss": 0.1701, + "step": 20360 + }, + { + "epoch": 0.7402427501998692, + "grad_norm": 1.8553961515426636, + "learning_rate": 4.976576102520797e-05, + "loss": 0.1455, + "step": 20370 + }, + { + "epoch": 0.7406061487026674, + "grad_norm": 3.1159512996673584, + "learning_rate": 4.976505178712874e-05, + "loss": 0.2252, + "step": 20380 + }, + { + "epoch": 0.7409695472054655, + "grad_norm": 1.9035766124725342, + "learning_rate": 4.9764341482011545e-05, + "loss": 0.1815, + "step": 20390 + }, + { + "epoch": 0.7413329457082637, + "grad_norm": 2.228940725326538, + "learning_rate": 4.976363010988698e-05, + "loss": 0.1995, + "step": 20400 + }, + { + "epoch": 0.7413329457082637, + "eval_loss": 0.35944151878356934, + "eval_runtime": 179.8589, + "eval_samples_per_second": 41.221, + "eval_steps_per_second": 5.154, + "eval_wer": 0.200864087715795, + "step": 20400 + }, + { + "epoch": 0.7416963442110619, + "grad_norm": 1.5204256772994995, + "learning_rate": 4.976291767078571e-05, + "loss": 1.6497, + "step": 20410 + }, + { + "epoch": 0.74205974271386, + "grad_norm": 1.3520594835281372, + "learning_rate": 4.976220416473842e-05, + "loss": 0.1503, + "step": 20420 + }, + { + "epoch": 0.7424231412166582, + "grad_norm": 2.7322440147399902, + "learning_rate": 4.976148959177586e-05, + "loss": 0.1784, + "step": 20430 + }, + { + "epoch": 0.7427865397194564, + "grad_norm": 1.3193668127059937, + "learning_rate": 4.9760773951928815e-05, + "loss": 0.1685, + "step": 20440 + }, + { + "epoch": 0.7431499382222545, + "grad_norm": 11.000434875488281, + "learning_rate": 4.976005724522812e-05, + "loss": 0.2147, + "step": 20450 + }, + { + "epoch": 0.7435133367250527, + "grad_norm": 1.1825796365737915, + "learning_rate": 4.9759339471704656e-05, + "loss": 0.2116, + "step": 20460 + }, + { + "epoch": 0.7438767352278509, + "grad_norm": 1.1518877744674683, + "learning_rate": 4.975862063138934e-05, + "loss": 0.141, + "step": 20470 + }, + { + "epoch": 0.744240133730649, + "grad_norm": 6.054372310638428, + "learning_rate": 4.975790072431316e-05, + "loss": 0.1766, + "step": 20480 + }, + { + "epoch": 0.7446035322334472, + "grad_norm": 1.0629233121871948, + "learning_rate": 4.975717975050713e-05, + "loss": 0.1641, + "step": 20490 + }, + { + "epoch": 0.7449669307362453, + "grad_norm": 2.4782843589782715, + "learning_rate": 4.97564577100023e-05, + "loss": 0.2186, + "step": 20500 + }, + { + "epoch": 0.7453303292390435, + "grad_norm": 1.5713534355163574, + "learning_rate": 4.975573460282979e-05, + "loss": 0.1535, + "step": 20510 + }, + { + "epoch": 0.7456937277418417, + "grad_norm": 0.7279618382453918, + "learning_rate": 4.975501042902078e-05, + "loss": 0.1372, + "step": 20520 + }, + { + "epoch": 0.7460571262446398, + "grad_norm": 5.573297500610352, + "learning_rate": 4.975428518860643e-05, + "loss": 0.161, + "step": 20530 + }, + { + "epoch": 0.746420524747438, + "grad_norm": 1.022141695022583, + "learning_rate": 4.975355888161801e-05, + "loss": 0.1645, + "step": 20540 + }, + { + "epoch": 0.7467839232502362, + "grad_norm": 2.9584996700286865, + "learning_rate": 4.9752831508086805e-05, + "loss": 0.2085, + "step": 20550 + }, + { + "epoch": 0.7471473217530343, + "grad_norm": 2.2749557495117188, + "learning_rate": 4.975210306804418e-05, + "loss": 0.1531, + "step": 20560 + }, + { + "epoch": 0.7475107202558325, + "grad_norm": 1.877822995185852, + "learning_rate": 4.9751373561521484e-05, + "loss": 0.1654, + "step": 20570 + }, + { + "epoch": 0.7478741187586307, + "grad_norm": 7.727886199951172, + "learning_rate": 4.975064298855017e-05, + "loss": 0.2026, + "step": 20580 + }, + { + "epoch": 0.7482375172614288, + "grad_norm": 1.2424033880233765, + "learning_rate": 4.974991134916171e-05, + "loss": 0.1834, + "step": 20590 + }, + { + "epoch": 0.748600915764227, + "grad_norm": 7.272613525390625, + "learning_rate": 4.974917864338764e-05, + "loss": 0.2266, + "step": 20600 + }, + { + "epoch": 0.7489643142670253, + "grad_norm": 0.6424925327301025, + "learning_rate": 4.974844487125952e-05, + "loss": 0.1496, + "step": 20610 + }, + { + "epoch": 0.7493277127698234, + "grad_norm": 2.064819097518921, + "learning_rate": 4.974771003280896e-05, + "loss": 0.192, + "step": 20620 + }, + { + "epoch": 0.7496911112726216, + "grad_norm": 2.55157470703125, + "learning_rate": 4.974697412806763e-05, + "loss": 0.1863, + "step": 20630 + }, + { + "epoch": 0.7500545097754198, + "grad_norm": 1.10732901096344, + "learning_rate": 4.974623715706723e-05, + "loss": 0.1452, + "step": 20640 + }, + { + "epoch": 0.7504179082782179, + "grad_norm": 6.665337562561035, + "learning_rate": 4.9745499119839526e-05, + "loss": 0.2393, + "step": 20650 + }, + { + "epoch": 0.7507813067810161, + "grad_norm": 2.315764904022217, + "learning_rate": 4.974476001641631e-05, + "loss": 0.1724, + "step": 20660 + }, + { + "epoch": 0.7511447052838143, + "grad_norm": 1.7643327713012695, + "learning_rate": 4.974401984682942e-05, + "loss": 0.1676, + "step": 20670 + }, + { + "epoch": 0.7515081037866124, + "grad_norm": 2.556265115737915, + "learning_rate": 4.974327861111075e-05, + "loss": 0.1706, + "step": 20680 + }, + { + "epoch": 0.7518715022894106, + "grad_norm": 1.0939987897872925, + "learning_rate": 4.9742536309292257e-05, + "loss": 0.1514, + "step": 20690 + }, + { + "epoch": 0.7522349007922088, + "grad_norm": 2.3087685108184814, + "learning_rate": 4.97417929414059e-05, + "loss": 0.2064, + "step": 20700 + }, + { + "epoch": 0.7525982992950069, + "grad_norm": 1.6968719959259033, + "learning_rate": 4.974104850748372e-05, + "loss": 0.65, + "step": 20710 + }, + { + "epoch": 0.7529616977978051, + "grad_norm": 1.3144559860229492, + "learning_rate": 4.974030300755779e-05, + "loss": 3.2825, + "step": 20720 + }, + { + "epoch": 0.7533250963006033, + "grad_norm": 2.346266031265259, + "learning_rate": 4.973955644166022e-05, + "loss": 0.1621, + "step": 20730 + }, + { + "epoch": 0.7536884948034014, + "grad_norm": 0.8026605248451233, + "learning_rate": 4.973880880982319e-05, + "loss": 0.1566, + "step": 20740 + }, + { + "epoch": 0.7540518933061996, + "grad_norm": 8.70439624786377, + "learning_rate": 4.973806011207891e-05, + "loss": 0.2671, + "step": 20750 + }, + { + "epoch": 0.7544152918089978, + "grad_norm": 0.9762817025184631, + "learning_rate": 4.973731034845964e-05, + "loss": 0.1692, + "step": 20760 + }, + { + "epoch": 0.7547786903117959, + "grad_norm": 1.3316736221313477, + "learning_rate": 4.973655951899768e-05, + "loss": 0.1605, + "step": 20770 + }, + { + "epoch": 0.7551420888145941, + "grad_norm": 1.9772186279296875, + "learning_rate": 4.9735807623725394e-05, + "loss": 0.1551, + "step": 20780 + }, + { + "epoch": 0.7555054873173922, + "grad_norm": 1.4639058113098145, + "learning_rate": 4.9735054662675154e-05, + "loss": 0.2075, + "step": 20790 + }, + { + "epoch": 0.7558688858201904, + "grad_norm": 10.605428695678711, + "learning_rate": 4.973430063587943e-05, + "loss": 0.2542, + "step": 20800 + }, + { + "epoch": 0.7562322843229886, + "grad_norm": 1.9553091526031494, + "learning_rate": 4.9733545543370684e-05, + "loss": 0.1353, + "step": 20810 + }, + { + "epoch": 0.7565956828257867, + "grad_norm": 2.2855403423309326, + "learning_rate": 4.9732789385181466e-05, + "loss": 0.5004, + "step": 20820 + }, + { + "epoch": 0.7569590813285849, + "grad_norm": 1.7468841075897217, + "learning_rate": 4.973203216134435e-05, + "loss": 0.1433, + "step": 20830 + }, + { + "epoch": 0.7573224798313831, + "grad_norm": 0.9522268772125244, + "learning_rate": 4.973127387189197e-05, + "loss": 0.1488, + "step": 20840 + }, + { + "epoch": 0.7576858783341812, + "grad_norm": 13.445122718811035, + "learning_rate": 4.9730514516856996e-05, + "loss": 0.2154, + "step": 20850 + }, + { + "epoch": 0.7580492768369794, + "grad_norm": 1.0712549686431885, + "learning_rate": 4.972975409627214e-05, + "loss": 0.144, + "step": 20860 + }, + { + "epoch": 0.7584126753397776, + "grad_norm": 0.6894069314002991, + "learning_rate": 4.972899261017017e-05, + "loss": 0.1612, + "step": 20870 + }, + { + "epoch": 0.7587760738425757, + "grad_norm": 2.059844970703125, + "learning_rate": 4.9728230058583893e-05, + "loss": 0.1664, + "step": 20880 + }, + { + "epoch": 0.7591394723453739, + "grad_norm": 2.0392911434173584, + "learning_rate": 4.972746644154616e-05, + "loss": 0.1991, + "step": 20890 + }, + { + "epoch": 0.7595028708481721, + "grad_norm": 2.9800570011138916, + "learning_rate": 4.972670175908989e-05, + "loss": 0.2725, + "step": 20900 + }, + { + "epoch": 0.7598662693509702, + "grad_norm": 2.390784502029419, + "learning_rate": 4.972593601124801e-05, + "loss": 0.3158, + "step": 20910 + }, + { + "epoch": 0.7602296678537684, + "grad_norm": 6.595739364624023, + "learning_rate": 4.972516919805352e-05, + "loss": 0.1658, + "step": 20920 + }, + { + "epoch": 0.7605930663565666, + "grad_norm": 2.2043120861053467, + "learning_rate": 4.972440131953947e-05, + "loss": 0.163, + "step": 20930 + }, + { + "epoch": 0.7609564648593647, + "grad_norm": 0.9223461747169495, + "learning_rate": 4.972363237573894e-05, + "loss": 0.1276, + "step": 20940 + }, + { + "epoch": 0.761319863362163, + "grad_norm": 12.165254592895508, + "learning_rate": 4.972286236668505e-05, + "loss": 0.2105, + "step": 20950 + }, + { + "epoch": 0.7616832618649612, + "grad_norm": 1.2093875408172607, + "learning_rate": 4.9722091292410984e-05, + "loss": 0.1697, + "step": 20960 + }, + { + "epoch": 0.7620466603677593, + "grad_norm": 0.8847984075546265, + "learning_rate": 4.9721396414828535e-05, + "loss": 3.043, + "step": 20970 + }, + { + "epoch": 0.7624100588705575, + "grad_norm": 1.6682274341583252, + "learning_rate": 4.9720623316727705e-05, + "loss": 0.1841, + "step": 20980 + }, + { + "epoch": 0.7627734573733557, + "grad_norm": 1.2780869007110596, + "learning_rate": 4.971984915350317e-05, + "loss": 0.1412, + "step": 20990 + }, + { + "epoch": 0.7631368558761538, + "grad_norm": 36.68233108520508, + "learning_rate": 4.97190739251883e-05, + "loss": 0.2059, + "step": 21000 + }, + { + "epoch": 0.7631368558761538, + "eval_loss": 0.3906314969062805, + "eval_runtime": 180.3915, + "eval_samples_per_second": 41.1, + "eval_steps_per_second": 5.139, + "eval_wer": 0.21157441864686768, + "step": 21000 + }, + { + "epoch": 0.763500254378952, + "grad_norm": 24.036775588989258, + "learning_rate": 4.971829763181647e-05, + "loss": 0.3942, + "step": 21010 + }, + { + "epoch": 0.7638636528817502, + "grad_norm": 1.6546601057052612, + "learning_rate": 4.971752027342115e-05, + "loss": 0.1555, + "step": 21020 + }, + { + "epoch": 0.7642270513845483, + "grad_norm": 3.100032091140747, + "learning_rate": 4.971674185003583e-05, + "loss": 0.1917, + "step": 21030 + }, + { + "epoch": 0.7645904498873465, + "grad_norm": 3.2824084758758545, + "learning_rate": 4.9715962361694045e-05, + "loss": 0.1744, + "step": 21040 + }, + { + "epoch": 0.7649538483901447, + "grad_norm": 7.680720329284668, + "learning_rate": 4.9715181808429376e-05, + "loss": 0.2567, + "step": 21050 + }, + { + "epoch": 0.7653172468929428, + "grad_norm": 1.5478154420852661, + "learning_rate": 4.971440019027547e-05, + "loss": 0.1949, + "step": 21060 + }, + { + "epoch": 0.765680645395741, + "grad_norm": 1.1294565200805664, + "learning_rate": 4.971361750726598e-05, + "loss": 0.1546, + "step": 21070 + }, + { + "epoch": 0.7660440438985391, + "grad_norm": 3.339749813079834, + "learning_rate": 4.971283375943465e-05, + "loss": 0.1784, + "step": 21080 + }, + { + "epoch": 0.7664074424013373, + "grad_norm": 1.9784200191497803, + "learning_rate": 4.9712048946815244e-05, + "loss": 0.8969, + "step": 21090 + }, + { + "epoch": 0.7667708409041355, + "grad_norm": 13.550655364990234, + "learning_rate": 4.971126306944157e-05, + "loss": 0.2037, + "step": 21100 + }, + { + "epoch": 0.7671342394069336, + "grad_norm": 60.52021408081055, + "learning_rate": 4.971047612734749e-05, + "loss": 1.0649, + "step": 21110 + }, + { + "epoch": 0.7674976379097318, + "grad_norm": 1.7544801235198975, + "learning_rate": 4.970968812056693e-05, + "loss": 0.1619, + "step": 21120 + }, + { + "epoch": 0.76786103641253, + "grad_norm": 2.0749471187591553, + "learning_rate": 4.970889904913382e-05, + "loss": 0.1934, + "step": 21130 + }, + { + "epoch": 0.7682244349153281, + "grad_norm": 2.33097767829895, + "learning_rate": 4.970810891308215e-05, + "loss": 0.3121, + "step": 21140 + }, + { + "epoch": 0.7685878334181263, + "grad_norm": 3.5586440563201904, + "learning_rate": 4.9707317712445996e-05, + "loss": 0.2198, + "step": 21150 + }, + { + "epoch": 0.7689512319209245, + "grad_norm": 1.7430351972579956, + "learning_rate": 4.970652544725942e-05, + "loss": 0.1884, + "step": 21160 + }, + { + "epoch": 0.7693146304237226, + "grad_norm": 1.2475924491882324, + "learning_rate": 4.9705732117556574e-05, + "loss": 0.183, + "step": 21170 + }, + { + "epoch": 0.7696780289265208, + "grad_norm": 1.369491457939148, + "learning_rate": 4.970493772337164e-05, + "loss": 0.1854, + "step": 21180 + }, + { + "epoch": 0.770041427429319, + "grad_norm": 1.8093339204788208, + "learning_rate": 4.970414226473883e-05, + "loss": 0.1389, + "step": 21190 + }, + { + "epoch": 0.7704048259321171, + "grad_norm": 15.3746919631958, + "learning_rate": 4.9703345741692425e-05, + "loss": 0.2603, + "step": 21200 + }, + { + "epoch": 0.7707682244349153, + "grad_norm": 0.9604819416999817, + "learning_rate": 4.970254815426675e-05, + "loss": 0.1663, + "step": 21210 + }, + { + "epoch": 0.7711316229377135, + "grad_norm": 1.3457413911819458, + "learning_rate": 4.970174950249617e-05, + "loss": 0.1784, + "step": 21220 + }, + { + "epoch": 0.7714950214405116, + "grad_norm": 3.19975209236145, + "learning_rate": 4.970094978641509e-05, + "loss": 0.2369, + "step": 21230 + }, + { + "epoch": 0.7718584199433098, + "grad_norm": 1.4974329471588135, + "learning_rate": 4.970014900605797e-05, + "loss": 0.1553, + "step": 21240 + }, + { + "epoch": 0.772221818446108, + "grad_norm": 6.426448345184326, + "learning_rate": 4.969934716145932e-05, + "loss": 0.1848, + "step": 21250 + }, + { + "epoch": 0.7725852169489061, + "grad_norm": 4.081672668457031, + "learning_rate": 4.969854425265368e-05, + "loss": 0.2135, + "step": 21260 + }, + { + "epoch": 0.7729486154517043, + "grad_norm": 0.7796603441238403, + "learning_rate": 4.9697740279675635e-05, + "loss": 0.2853, + "step": 21270 + }, + { + "epoch": 0.7733120139545026, + "grad_norm": 1.2303035259246826, + "learning_rate": 4.969693524255984e-05, + "loss": 0.5319, + "step": 21280 + }, + { + "epoch": 0.7736754124573006, + "grad_norm": 0.9134958386421204, + "learning_rate": 4.9696129141340986e-05, + "loss": 0.1789, + "step": 21290 + }, + { + "epoch": 0.7740388109600989, + "grad_norm": 1.8099846839904785, + "learning_rate": 4.969532197605379e-05, + "loss": 0.1967, + "step": 21300 + }, + { + "epoch": 0.7744022094628971, + "grad_norm": 3.75593900680542, + "learning_rate": 4.969451374673304e-05, + "loss": 0.1908, + "step": 21310 + }, + { + "epoch": 0.7747656079656952, + "grad_norm": 2.851921319961548, + "learning_rate": 4.969370445341355e-05, + "loss": 0.1616, + "step": 21320 + }, + { + "epoch": 0.7751290064684934, + "grad_norm": 2.978349447250366, + "learning_rate": 4.96928940961302e-05, + "loss": 0.1682, + "step": 21330 + }, + { + "epoch": 0.7754924049712916, + "grad_norm": 2.945326089859009, + "learning_rate": 4.96920826749179e-05, + "loss": 0.1897, + "step": 21340 + }, + { + "epoch": 0.7758558034740897, + "grad_norm": 5.529159069061279, + "learning_rate": 4.9691270189811614e-05, + "loss": 0.2351, + "step": 21350 + }, + { + "epoch": 0.7762192019768879, + "grad_norm": 0.816582441329956, + "learning_rate": 4.969045664084634e-05, + "loss": 0.255, + "step": 21360 + }, + { + "epoch": 0.776582600479686, + "grad_norm": 3.373413324356079, + "learning_rate": 4.968964202805715e-05, + "loss": 0.165, + "step": 21370 + }, + { + "epoch": 0.7769459989824842, + "grad_norm": 1.4986653327941895, + "learning_rate": 4.968882635147912e-05, + "loss": 0.1803, + "step": 21380 + }, + { + "epoch": 0.7773093974852824, + "grad_norm": 4.049030303955078, + "learning_rate": 4.968800961114741e-05, + "loss": 0.2312, + "step": 21390 + }, + { + "epoch": 0.7776727959880805, + "grad_norm": 1.8616725206375122, + "learning_rate": 4.968719180709721e-05, + "loss": 0.2038, + "step": 21400 + }, + { + "epoch": 0.7780361944908787, + "grad_norm": 0.7410339117050171, + "learning_rate": 4.968637293936374e-05, + "loss": 0.1736, + "step": 21410 + }, + { + "epoch": 0.7783995929936769, + "grad_norm": 0.9004227519035339, + "learning_rate": 4.968555300798231e-05, + "loss": 0.6926, + "step": 21420 + }, + { + "epoch": 0.778762991496475, + "grad_norm": 1.9912917613983154, + "learning_rate": 4.968473201298822e-05, + "loss": 0.183, + "step": 21430 + }, + { + "epoch": 0.7791263899992732, + "grad_norm": 1.5098110437393188, + "learning_rate": 4.968390995441686e-05, + "loss": 0.1555, + "step": 21440 + }, + { + "epoch": 0.7794897885020714, + "grad_norm": 1.5687317848205566, + "learning_rate": 4.9683086832303655e-05, + "loss": 0.199, + "step": 21450 + }, + { + "epoch": 0.7798531870048695, + "grad_norm": 1.456758975982666, + "learning_rate": 4.9682262646684054e-05, + "loss": 0.1573, + "step": 21460 + }, + { + "epoch": 0.7802165855076677, + "grad_norm": 1.152894377708435, + "learning_rate": 4.9681437397593575e-05, + "loss": 0.136, + "step": 21470 + }, + { + "epoch": 0.7805799840104659, + "grad_norm": 6.458597183227539, + "learning_rate": 4.968061108506777e-05, + "loss": 0.2111, + "step": 21480 + }, + { + "epoch": 0.780943382513264, + "grad_norm": 1.3398655652999878, + "learning_rate": 4.967978370914226e-05, + "loss": 0.1785, + "step": 21490 + }, + { + "epoch": 0.7813067810160622, + "grad_norm": 12.363832473754883, + "learning_rate": 4.967895526985267e-05, + "loss": 0.217, + "step": 21500 + }, + { + "epoch": 0.7816701795188604, + "grad_norm": 3.800936698913574, + "learning_rate": 4.967812576723471e-05, + "loss": 0.1533, + "step": 21510 + }, + { + "epoch": 0.7820335780216585, + "grad_norm": 0.9531782865524292, + "learning_rate": 4.967729520132411e-05, + "loss": 1.353, + "step": 21520 + }, + { + "epoch": 0.7823969765244567, + "grad_norm": 1.3066377639770508, + "learning_rate": 4.967646357215667e-05, + "loss": 0.1338, + "step": 21530 + }, + { + "epoch": 0.7827603750272549, + "grad_norm": 1.1814554929733276, + "learning_rate": 4.967563087976821e-05, + "loss": 0.1735, + "step": 21540 + }, + { + "epoch": 0.783123773530053, + "grad_norm": 4.6233367919921875, + "learning_rate": 4.967479712419461e-05, + "loss": 0.2266, + "step": 21550 + }, + { + "epoch": 0.7834871720328512, + "grad_norm": 1.366377353668213, + "learning_rate": 4.96739623054718e-05, + "loss": 0.1595, + "step": 21560 + }, + { + "epoch": 0.7838505705356494, + "grad_norm": 2.0722217559814453, + "learning_rate": 4.967312642363574e-05, + "loss": 0.1721, + "step": 21570 + }, + { + "epoch": 0.7842139690384475, + "grad_norm": 2.186340570449829, + "learning_rate": 4.967228947872245e-05, + "loss": 0.1653, + "step": 21580 + }, + { + "epoch": 0.7845773675412457, + "grad_norm": 2.4222512245178223, + "learning_rate": 4.9671451470767996e-05, + "loss": 0.1446, + "step": 21590 + }, + { + "epoch": 0.784940766044044, + "grad_norm": 62.15577697753906, + "learning_rate": 4.9670612399808467e-05, + "loss": 0.2911, + "step": 21600 + }, + { + "epoch": 0.784940766044044, + "eval_loss": 0.3627218008041382, + "eval_runtime": 179.8971, + "eval_samples_per_second": 41.212, + "eval_steps_per_second": 5.153, + "eval_wer": 0.21580409170947773, + "step": 21600 + }, + { + "epoch": 0.785304164546842, + "grad_norm": 1.236609935760498, + "learning_rate": 4.9669772265880044e-05, + "loss": 0.1417, + "step": 21610 + }, + { + "epoch": 0.7856675630496402, + "grad_norm": 1.2447402477264404, + "learning_rate": 4.96689310690189e-05, + "loss": 0.1508, + "step": 21620 + }, + { + "epoch": 0.7860309615524385, + "grad_norm": 4.567975997924805, + "learning_rate": 4.966808880926129e-05, + "loss": 0.3503, + "step": 21630 + }, + { + "epoch": 0.7863943600552366, + "grad_norm": 0.9699403047561646, + "learning_rate": 4.96672454866435e-05, + "loss": 0.1615, + "step": 21640 + }, + { + "epoch": 0.7867577585580348, + "grad_norm": 11.004621505737305, + "learning_rate": 4.966640110120187e-05, + "loss": 0.2604, + "step": 21650 + }, + { + "epoch": 0.7871211570608329, + "grad_norm": 1.3322606086730957, + "learning_rate": 4.9665555652972784e-05, + "loss": 0.1958, + "step": 21660 + }, + { + "epoch": 0.7874845555636311, + "grad_norm": 1.0020729303359985, + "learning_rate": 4.966470914199266e-05, + "loss": 0.1207, + "step": 21670 + }, + { + "epoch": 0.7878479540664293, + "grad_norm": 3.457019567489624, + "learning_rate": 4.9663861568297976e-05, + "loss": 0.3774, + "step": 21680 + }, + { + "epoch": 0.7882113525692274, + "grad_norm": 2.4993362426757812, + "learning_rate": 4.9663012931925254e-05, + "loss": 0.1537, + "step": 21690 + }, + { + "epoch": 0.7885747510720256, + "grad_norm": 11.104598999023438, + "learning_rate": 4.966216323291106e-05, + "loss": 0.2472, + "step": 21700 + }, + { + "epoch": 0.7889381495748238, + "grad_norm": 1.5027676820755005, + "learning_rate": 4.9661312471291996e-05, + "loss": 0.154, + "step": 21710 + }, + { + "epoch": 0.7893015480776219, + "grad_norm": 1.1929068565368652, + "learning_rate": 4.9660460647104726e-05, + "loss": 0.1416, + "step": 21720 + }, + { + "epoch": 0.7896649465804201, + "grad_norm": 17.008617401123047, + "learning_rate": 4.965960776038594e-05, + "loss": 0.3858, + "step": 21730 + }, + { + "epoch": 0.7900283450832183, + "grad_norm": 1.6043013334274292, + "learning_rate": 4.96587538111724e-05, + "loss": 0.1624, + "step": 21740 + }, + { + "epoch": 0.7903917435860164, + "grad_norm": 10.960922241210938, + "learning_rate": 4.96578987995009e-05, + "loss": 0.2034, + "step": 21750 + }, + { + "epoch": 0.7907551420888146, + "grad_norm": 1.4807969331741333, + "learning_rate": 4.965704272540826e-05, + "loss": 0.1491, + "step": 21760 + }, + { + "epoch": 0.7911185405916128, + "grad_norm": 0.9724571108818054, + "learning_rate": 4.965618558893139e-05, + "loss": 0.1455, + "step": 21770 + }, + { + "epoch": 0.7914819390944109, + "grad_norm": 2.6035313606262207, + "learning_rate": 4.965532739010722e-05, + "loss": 0.1696, + "step": 21780 + }, + { + "epoch": 0.7918453375972091, + "grad_norm": 0.7998749017715454, + "learning_rate": 4.9654468128972695e-05, + "loss": 0.1549, + "step": 21790 + }, + { + "epoch": 0.7922087361000073, + "grad_norm": 14.13917350769043, + "learning_rate": 4.965360780556487e-05, + "loss": 0.2124, + "step": 21800 + }, + { + "epoch": 0.7925721346028054, + "grad_norm": 1.88883638381958, + "learning_rate": 4.9652746419920804e-05, + "loss": 0.1475, + "step": 21810 + }, + { + "epoch": 0.7929355331056036, + "grad_norm": 1.5585650205612183, + "learning_rate": 4.965188397207761e-05, + "loss": 0.1534, + "step": 21820 + }, + { + "epoch": 0.7932989316084018, + "grad_norm": 2.6418206691741943, + "learning_rate": 4.965102046207244e-05, + "loss": 0.1608, + "step": 21830 + }, + { + "epoch": 0.7936623301111999, + "grad_norm": 1.1672085523605347, + "learning_rate": 4.965015588994251e-05, + "loss": 0.1596, + "step": 21840 + }, + { + "epoch": 0.7940257286139981, + "grad_norm": 3.009610652923584, + "learning_rate": 4.964929025572507e-05, + "loss": 0.1805, + "step": 21850 + }, + { + "epoch": 0.7943891271167963, + "grad_norm": 1.8774985074996948, + "learning_rate": 4.964842355945742e-05, + "loss": 0.1583, + "step": 21860 + }, + { + "epoch": 0.7947525256195944, + "grad_norm": 1.1219382286071777, + "learning_rate": 4.964755580117689e-05, + "loss": 0.1524, + "step": 21870 + }, + { + "epoch": 0.7951159241223926, + "grad_norm": 6.0511627197265625, + "learning_rate": 4.964668698092088e-05, + "loss": 0.2349, + "step": 21880 + }, + { + "epoch": 0.7954793226251908, + "grad_norm": 3.4487464427948, + "learning_rate": 4.9645817098726824e-05, + "loss": 0.1915, + "step": 21890 + }, + { + "epoch": 0.7958427211279889, + "grad_norm": 4.096559524536133, + "learning_rate": 4.9644946154632196e-05, + "loss": 0.2067, + "step": 21900 + }, + { + "epoch": 0.7962061196307871, + "grad_norm": 4.144627571105957, + "learning_rate": 4.9644074148674526e-05, + "loss": 0.1564, + "step": 21910 + }, + { + "epoch": 0.7965695181335853, + "grad_norm": 1.3851386308670044, + "learning_rate": 4.9643201080891384e-05, + "loss": 0.1656, + "step": 21920 + }, + { + "epoch": 0.7969329166363834, + "grad_norm": 1.3050576448440552, + "learning_rate": 4.9642326951320384e-05, + "loss": 0.1555, + "step": 21930 + }, + { + "epoch": 0.7972963151391816, + "grad_norm": 1.578134298324585, + "learning_rate": 4.96414517599992e-05, + "loss": 0.1637, + "step": 21940 + }, + { + "epoch": 0.7976597136419797, + "grad_norm": 10.813237190246582, + "learning_rate": 4.9640575506965535e-05, + "loss": 0.3143, + "step": 21950 + }, + { + "epoch": 0.798023112144778, + "grad_norm": 0.7118828892707825, + "learning_rate": 4.963969819225713e-05, + "loss": 0.1581, + "step": 21960 + }, + { + "epoch": 0.7983865106475762, + "grad_norm": 1.389856219291687, + "learning_rate": 4.963881981591182e-05, + "loss": 0.1466, + "step": 21970 + }, + { + "epoch": 0.7987499091503742, + "grad_norm": 1.1921494007110596, + "learning_rate": 4.963794037796741e-05, + "loss": 0.1604, + "step": 21980 + }, + { + "epoch": 0.7991133076531725, + "grad_norm": 4.355441093444824, + "learning_rate": 4.963705987846182e-05, + "loss": 0.1792, + "step": 21990 + }, + { + "epoch": 0.7994767061559707, + "grad_norm": 8.20235824584961, + "learning_rate": 4.963617831743298e-05, + "loss": 0.2314, + "step": 22000 + }, + { + "epoch": 0.7998401046587688, + "grad_norm": 1.3720426559448242, + "learning_rate": 4.963529569491887e-05, + "loss": 0.1378, + "step": 22010 + }, + { + "epoch": 0.800203503161567, + "grad_norm": 1.490679383277893, + "learning_rate": 4.963441201095752e-05, + "loss": 0.1505, + "step": 22020 + }, + { + "epoch": 0.8005669016643652, + "grad_norm": 1.576416254043579, + "learning_rate": 4.963352726558701e-05, + "loss": 0.1379, + "step": 22030 + }, + { + "epoch": 0.8009303001671633, + "grad_norm": 1.547780156135559, + "learning_rate": 4.9632641458845454e-05, + "loss": 0.1584, + "step": 22040 + }, + { + "epoch": 0.8012936986699615, + "grad_norm": 41.95133972167969, + "learning_rate": 4.963175459077102e-05, + "loss": 0.6762, + "step": 22050 + }, + { + "epoch": 0.8016570971727597, + "grad_norm": 0.8984355330467224, + "learning_rate": 4.963086666140192e-05, + "loss": 0.1513, + "step": 22060 + }, + { + "epoch": 0.8020204956755578, + "grad_norm": 1.6865235567092896, + "learning_rate": 4.9629977670776404e-05, + "loss": 0.1659, + "step": 22070 + }, + { + "epoch": 0.802383894178356, + "grad_norm": 5.291965007781982, + "learning_rate": 4.96290876189328e-05, + "loss": 0.1735, + "step": 22080 + }, + { + "epoch": 0.8027472926811542, + "grad_norm": 0.9124179482460022, + "learning_rate": 4.962819650590943e-05, + "loss": 0.163, + "step": 22090 + }, + { + "epoch": 0.8031106911839523, + "grad_norm": 5.151334762573242, + "learning_rate": 4.9627304331744705e-05, + "loss": 0.2997, + "step": 22100 + }, + { + "epoch": 0.8034740896867505, + "grad_norm": 0.7093039155006409, + "learning_rate": 4.9626411096477066e-05, + "loss": 0.1297, + "step": 22110 + }, + { + "epoch": 0.8038374881895487, + "grad_norm": 0.7643496990203857, + "learning_rate": 4.962551680014499e-05, + "loss": 0.1568, + "step": 22120 + }, + { + "epoch": 0.8042008866923468, + "grad_norm": 2.0619888305664062, + "learning_rate": 4.9624621442787005e-05, + "loss": 0.1685, + "step": 22130 + }, + { + "epoch": 0.804564285195145, + "grad_norm": 1.3836963176727295, + "learning_rate": 4.9623725024441704e-05, + "loss": 0.1597, + "step": 22140 + }, + { + "epoch": 0.8049276836979432, + "grad_norm": 10.014172554016113, + "learning_rate": 4.96228275451477e-05, + "loss": 0.2371, + "step": 22150 + }, + { + "epoch": 0.8052910822007413, + "grad_norm": 0.8201650381088257, + "learning_rate": 4.962192900494367e-05, + "loss": 0.1457, + "step": 22160 + }, + { + "epoch": 0.8056544807035395, + "grad_norm": 2.9909164905548096, + "learning_rate": 4.962102940386832e-05, + "loss": 0.1584, + "step": 22170 + }, + { + "epoch": 0.8060178792063377, + "grad_norm": 1.8986990451812744, + "learning_rate": 4.9620128741960414e-05, + "loss": 0.1521, + "step": 22180 + }, + { + "epoch": 0.8063812777091358, + "grad_norm": 1.2521679401397705, + "learning_rate": 4.9619227019258766e-05, + "loss": 0.1398, + "step": 22190 + }, + { + "epoch": 0.806744676211934, + "grad_norm": 9.087230682373047, + "learning_rate": 4.9618324235802214e-05, + "loss": 0.2414, + "step": 22200 + }, + { + "epoch": 0.806744676211934, + "eval_loss": 0.3814217448234558, + "eval_runtime": 180.9296, + "eval_samples_per_second": 40.977, + "eval_steps_per_second": 5.124, + "eval_wer": 0.21859059306188394, + "step": 22200 + }, + { + "epoch": 0.8071080747147322, + "grad_norm": 1.3065155744552612, + "learning_rate": 4.9617420391629666e-05, + "loss": 0.1382, + "step": 22210 + }, + { + "epoch": 0.8074714732175303, + "grad_norm": 1.0691299438476562, + "learning_rate": 4.961651548678006e-05, + "loss": 0.1692, + "step": 22220 + }, + { + "epoch": 0.8078348717203285, + "grad_norm": 2.515131711959839, + "learning_rate": 4.961560952129239e-05, + "loss": 0.1719, + "step": 22230 + }, + { + "epoch": 0.8081982702231267, + "grad_norm": 1.3650884628295898, + "learning_rate": 4.9614702495205686e-05, + "loss": 0.1918, + "step": 22240 + }, + { + "epoch": 0.8085616687259248, + "grad_norm": 4.730445384979248, + "learning_rate": 4.961379440855903e-05, + "loss": 0.2002, + "step": 22250 + }, + { + "epoch": 0.808925067228723, + "grad_norm": 1.0421544313430786, + "learning_rate": 4.9612885261391555e-05, + "loss": 0.1544, + "step": 22260 + }, + { + "epoch": 0.8092884657315211, + "grad_norm": 1.1957643032073975, + "learning_rate": 4.961197505374242e-05, + "loss": 0.1471, + "step": 22270 + }, + { + "epoch": 0.8096518642343193, + "grad_norm": 2.936429977416992, + "learning_rate": 4.961106378565086e-05, + "loss": 0.2068, + "step": 22280 + }, + { + "epoch": 0.8100152627371175, + "grad_norm": 2.0803070068359375, + "learning_rate": 4.961015145715612e-05, + "loss": 0.1496, + "step": 22290 + }, + { + "epoch": 0.8103786612399156, + "grad_norm": 10.564451217651367, + "learning_rate": 4.960923806829752e-05, + "loss": 0.2549, + "step": 22300 + }, + { + "epoch": 0.8107420597427138, + "grad_norm": 1.0569120645523071, + "learning_rate": 4.9608323619114406e-05, + "loss": 0.1624, + "step": 22310 + }, + { + "epoch": 0.8111054582455121, + "grad_norm": 1.4505226612091064, + "learning_rate": 4.960740810964619e-05, + "loss": 0.1523, + "step": 22320 + }, + { + "epoch": 0.8114688567483102, + "grad_norm": 5.100767135620117, + "learning_rate": 4.960649153993231e-05, + "loss": 0.1562, + "step": 22330 + }, + { + "epoch": 0.8118322552511084, + "grad_norm": 2.2787342071533203, + "learning_rate": 4.960557391001226e-05, + "loss": 0.1691, + "step": 22340 + }, + { + "epoch": 0.8121956537539066, + "grad_norm": 15.405048370361328, + "learning_rate": 4.960465521992558e-05, + "loss": 0.2542, + "step": 22350 + }, + { + "epoch": 0.8125590522567047, + "grad_norm": 0.7388777732849121, + "learning_rate": 4.9603735469711845e-05, + "loss": 0.1522, + "step": 22360 + }, + { + "epoch": 0.8129224507595029, + "grad_norm": 0.9490914344787598, + "learning_rate": 4.960281465941069e-05, + "loss": 0.1317, + "step": 22370 + }, + { + "epoch": 0.8132858492623011, + "grad_norm": 2.281085252761841, + "learning_rate": 4.960189278906179e-05, + "loss": 0.1503, + "step": 22380 + }, + { + "epoch": 0.8136492477650992, + "grad_norm": 0.9328985810279846, + "learning_rate": 4.960096985870486e-05, + "loss": 0.1556, + "step": 22390 + }, + { + "epoch": 0.8140126462678974, + "grad_norm": 4.4524617195129395, + "learning_rate": 4.960004586837967e-05, + "loss": 0.2387, + "step": 22400 + }, + { + "epoch": 0.8143760447706956, + "grad_norm": 1.5577040910720825, + "learning_rate": 4.959912081812603e-05, + "loss": 0.1557, + "step": 22410 + }, + { + "epoch": 0.8147394432734937, + "grad_norm": 2.358896493911743, + "learning_rate": 4.95981947079838e-05, + "loss": 0.2016, + "step": 22420 + }, + { + "epoch": 0.8151028417762919, + "grad_norm": 2.1001386642456055, + "learning_rate": 4.9597267537992885e-05, + "loss": 0.1587, + "step": 22430 + }, + { + "epoch": 0.8154662402790901, + "grad_norm": 2.7561607360839844, + "learning_rate": 4.959633930819323e-05, + "loss": 0.1616, + "step": 22440 + }, + { + "epoch": 0.8158296387818882, + "grad_norm": 4.204514980316162, + "learning_rate": 4.959541001862482e-05, + "loss": 0.6089, + "step": 22450 + }, + { + "epoch": 0.8161930372846864, + "grad_norm": 1.3738398551940918, + "learning_rate": 4.959447966932771e-05, + "loss": 0.1756, + "step": 22460 + }, + { + "epoch": 0.8165564357874846, + "grad_norm": 0.705806314945221, + "learning_rate": 4.959354826034197e-05, + "loss": 0.1213, + "step": 22470 + }, + { + "epoch": 0.8169198342902827, + "grad_norm": 2.053788661956787, + "learning_rate": 4.9592615791707755e-05, + "loss": 0.1765, + "step": 22480 + }, + { + "epoch": 0.8172832327930809, + "grad_norm": 2.0120911598205566, + "learning_rate": 4.959168226346521e-05, + "loss": 0.1444, + "step": 22490 + }, + { + "epoch": 0.8176466312958791, + "grad_norm": 6.552361011505127, + "learning_rate": 4.959074767565458e-05, + "loss": 0.2201, + "step": 22500 + }, + { + "epoch": 0.8180100297986772, + "grad_norm": 1.3007264137268066, + "learning_rate": 4.958981202831613e-05, + "loss": 0.1488, + "step": 22510 + }, + { + "epoch": 0.8183734283014754, + "grad_norm": 1.7885551452636719, + "learning_rate": 4.958887532149016e-05, + "loss": 2.6491, + "step": 22520 + }, + { + "epoch": 0.8187368268042736, + "grad_norm": 1.7092806100845337, + "learning_rate": 4.9587937555217054e-05, + "loss": 0.1946, + "step": 22530 + }, + { + "epoch": 0.8191002253070717, + "grad_norm": 2.56215238571167, + "learning_rate": 4.958699872953719e-05, + "loss": 0.1676, + "step": 22540 + }, + { + "epoch": 0.8194636238098699, + "grad_norm": 2.085753917694092, + "learning_rate": 4.958605884449104e-05, + "loss": 0.2038, + "step": 22550 + }, + { + "epoch": 0.819827022312668, + "grad_norm": 0.8225610852241516, + "learning_rate": 4.958511790011909e-05, + "loss": 0.5185, + "step": 22560 + }, + { + "epoch": 0.8201904208154662, + "grad_norm": 1.6775872707366943, + "learning_rate": 4.9584175896461884e-05, + "loss": 0.17, + "step": 22570 + }, + { + "epoch": 0.8205538193182644, + "grad_norm": 3.4285826683044434, + "learning_rate": 4.958323283356001e-05, + "loss": 0.164, + "step": 22580 + }, + { + "epoch": 0.8209172178210625, + "grad_norm": 1.892842411994934, + "learning_rate": 4.95822887114541e-05, + "loss": 1.2783, + "step": 22590 + }, + { + "epoch": 0.8212806163238607, + "grad_norm": 4.959444522857666, + "learning_rate": 4.9581343530184834e-05, + "loss": 0.2062, + "step": 22600 + }, + { + "epoch": 0.8216440148266589, + "grad_norm": 2.4584267139434814, + "learning_rate": 4.958039728979293e-05, + "loss": 0.1443, + "step": 22610 + }, + { + "epoch": 0.822007413329457, + "grad_norm": 1.118804693222046, + "learning_rate": 4.957944999031917e-05, + "loss": 0.16, + "step": 22620 + }, + { + "epoch": 0.8223708118322552, + "grad_norm": 1.5434421300888062, + "learning_rate": 4.9578501631804365e-05, + "loss": 0.2104, + "step": 22630 + }, + { + "epoch": 0.8227342103350535, + "grad_norm": 1.3116744756698608, + "learning_rate": 4.9577552214289374e-05, + "loss": 0.1326, + "step": 22640 + }, + { + "epoch": 0.8230976088378515, + "grad_norm": 11.34653377532959, + "learning_rate": 4.95766017378151e-05, + "loss": 0.2231, + "step": 22650 + }, + { + "epoch": 0.8234610073406498, + "grad_norm": 1.0379194021224976, + "learning_rate": 4.957565020242251e-05, + "loss": 0.1805, + "step": 22660 + }, + { + "epoch": 0.823824405843448, + "grad_norm": 1.8218019008636475, + "learning_rate": 4.957469760815259e-05, + "loss": 0.1287, + "step": 22670 + }, + { + "epoch": 0.8241878043462461, + "grad_norm": 1.1962164640426636, + "learning_rate": 4.957374395504638e-05, + "loss": 0.4115, + "step": 22680 + }, + { + "epoch": 0.8245512028490443, + "grad_norm": 1.9947481155395508, + "learning_rate": 4.957278924314499e-05, + "loss": 0.1407, + "step": 22690 + }, + { + "epoch": 0.8249146013518425, + "grad_norm": 25.343172073364258, + "learning_rate": 4.957183347248953e-05, + "loss": 0.4247, + "step": 22700 + }, + { + "epoch": 0.8252779998546406, + "grad_norm": 1.4444775581359863, + "learning_rate": 4.95708766431212e-05, + "loss": 0.1641, + "step": 22710 + }, + { + "epoch": 0.8256413983574388, + "grad_norm": 1.621640920639038, + "learning_rate": 4.9569918755081216e-05, + "loss": 0.1289, + "step": 22720 + }, + { + "epoch": 0.826004796860237, + "grad_norm": 1.018471360206604, + "learning_rate": 4.9568959808410854e-05, + "loss": 0.1694, + "step": 22730 + }, + { + "epoch": 0.8263681953630351, + "grad_norm": 3.1913223266601562, + "learning_rate": 4.9567999803151424e-05, + "loss": 0.1898, + "step": 22740 + }, + { + "epoch": 0.8267315938658333, + "grad_norm": 8.095772743225098, + "learning_rate": 4.956703873934431e-05, + "loss": 0.2246, + "step": 22750 + }, + { + "epoch": 0.8270949923686315, + "grad_norm": 1.738887906074524, + "learning_rate": 4.956607661703089e-05, + "loss": 0.1678, + "step": 22760 + }, + { + "epoch": 0.8274583908714296, + "grad_norm": 0.9688615202903748, + "learning_rate": 4.9565113436252644e-05, + "loss": 0.1341, + "step": 22770 + }, + { + "epoch": 0.8278217893742278, + "grad_norm": 2.2478010654449463, + "learning_rate": 4.956414919705106e-05, + "loss": 0.1823, + "step": 22780 + }, + { + "epoch": 0.828185187877026, + "grad_norm": 1.6718928813934326, + "learning_rate": 4.956318389946769e-05, + "loss": 0.1543, + "step": 22790 + }, + { + "epoch": 0.8285485863798241, + "grad_norm": 5.168727874755859, + "learning_rate": 4.956221754354412e-05, + "loss": 0.1795, + "step": 22800 + }, + { + "epoch": 0.8285485863798241, + "eval_loss": 0.3908107876777649, + "eval_runtime": 180.5873, + "eval_samples_per_second": 41.055, + "eval_steps_per_second": 5.133, + "eval_wer": 0.20674569317624847, + "step": 22800 + }, + { + "epoch": 0.8289119848826223, + "grad_norm": 0.9549854397773743, + "learning_rate": 4.956125012932199e-05, + "loss": 0.1559, + "step": 22810 + }, + { + "epoch": 0.8292753833854205, + "grad_norm": 3.2057716846466064, + "learning_rate": 4.9560281656842977e-05, + "loss": 0.1675, + "step": 22820 + }, + { + "epoch": 0.8296387818882186, + "grad_norm": 1.7775851488113403, + "learning_rate": 4.955931212614882e-05, + "loss": 0.1997, + "step": 22830 + }, + { + "epoch": 0.8300021803910168, + "grad_norm": 1.7028132677078247, + "learning_rate": 4.9558341537281274e-05, + "loss": 0.1505, + "step": 22840 + }, + { + "epoch": 0.8303655788938149, + "grad_norm": 2.7027060985565186, + "learning_rate": 4.955736989028218e-05, + "loss": 0.2009, + "step": 22850 + }, + { + "epoch": 0.8307289773966131, + "grad_norm": 1.8419814109802246, + "learning_rate": 4.955639718519339e-05, + "loss": 0.1355, + "step": 22860 + }, + { + "epoch": 0.8310923758994113, + "grad_norm": 0.8633226156234741, + "learning_rate": 4.955542342205682e-05, + "loss": 0.178, + "step": 22870 + }, + { + "epoch": 0.8314557744022094, + "grad_norm": 6.966017723083496, + "learning_rate": 4.955444860091442e-05, + "loss": 0.1885, + "step": 22880 + }, + { + "epoch": 0.8318191729050076, + "grad_norm": 1.9565801620483398, + "learning_rate": 4.955347272180819e-05, + "loss": 0.1485, + "step": 22890 + }, + { + "epoch": 0.8321825714078058, + "grad_norm": 22.704593658447266, + "learning_rate": 4.9552495784780196e-05, + "loss": 0.2294, + "step": 22900 + }, + { + "epoch": 0.8325459699106039, + "grad_norm": 2.0515658855438232, + "learning_rate": 4.95515177898725e-05, + "loss": 0.166, + "step": 22910 + }, + { + "epoch": 0.8329093684134021, + "grad_norm": 2.9277150630950928, + "learning_rate": 4.9550538737127275e-05, + "loss": 0.8898, + "step": 22920 + }, + { + "epoch": 0.8332727669162003, + "grad_norm": 3.9280052185058594, + "learning_rate": 4.9549558626586676e-05, + "loss": 0.171, + "step": 22930 + }, + { + "epoch": 0.8336361654189984, + "grad_norm": 2.5431272983551025, + "learning_rate": 4.954857745829294e-05, + "loss": 0.1539, + "step": 22940 + }, + { + "epoch": 0.8339995639217966, + "grad_norm": 2.815434694290161, + "learning_rate": 4.954759523228835e-05, + "loss": 0.2126, + "step": 22950 + }, + { + "epoch": 0.8343629624245948, + "grad_norm": 0.6958141922950745, + "learning_rate": 4.9546611948615224e-05, + "loss": 0.2069, + "step": 22960 + }, + { + "epoch": 0.8347263609273929, + "grad_norm": 0.7068191766738892, + "learning_rate": 4.9545627607315924e-05, + "loss": 0.1287, + "step": 22970 + }, + { + "epoch": 0.8350897594301911, + "grad_norm": 1.8746801614761353, + "learning_rate": 4.954464220843287e-05, + "loss": 0.1488, + "step": 22980 + }, + { + "epoch": 0.8354531579329894, + "grad_norm": 1.5134693384170532, + "learning_rate": 4.95436557520085e-05, + "loss": 0.1337, + "step": 22990 + }, + { + "epoch": 0.8358165564357874, + "grad_norm": 4.778042316436768, + "learning_rate": 4.9542668238085344e-05, + "loss": 0.2172, + "step": 23000 + }, + { + "epoch": 0.8361799549385857, + "grad_norm": 1.074409008026123, + "learning_rate": 4.9541679666705924e-05, + "loss": 0.1696, + "step": 23010 + }, + { + "epoch": 0.8365433534413839, + "grad_norm": 1.6725049018859863, + "learning_rate": 4.954069003791286e-05, + "loss": 0.136, + "step": 23020 + }, + { + "epoch": 0.836906751944182, + "grad_norm": 3.194450616836548, + "learning_rate": 4.953969935174877e-05, + "loss": 0.2067, + "step": 23030 + }, + { + "epoch": 0.8372701504469802, + "grad_norm": 7.7923150062561035, + "learning_rate": 4.9538707608256345e-05, + "loss": 0.1938, + "step": 23040 + }, + { + "epoch": 0.8376335489497784, + "grad_norm": 8.767574310302734, + "learning_rate": 4.953771480747833e-05, + "loss": 0.2473, + "step": 23050 + }, + { + "epoch": 0.8379969474525765, + "grad_norm": 1.3911685943603516, + "learning_rate": 4.953672094945748e-05, + "loss": 0.1497, + "step": 23060 + }, + { + "epoch": 0.8383603459553747, + "grad_norm": 0.7775372266769409, + "learning_rate": 4.953572603423662e-05, + "loss": 0.7581, + "step": 23070 + }, + { + "epoch": 0.8387237444581729, + "grad_norm": 2.6937413215637207, + "learning_rate": 4.9534730061858634e-05, + "loss": 0.1849, + "step": 23080 + }, + { + "epoch": 0.839087142960971, + "grad_norm": 0.7375633716583252, + "learning_rate": 4.953373303236642e-05, + "loss": 0.1706, + "step": 23090 + }, + { + "epoch": 0.8394505414637692, + "grad_norm": 3.070746421813965, + "learning_rate": 4.953273494580295e-05, + "loss": 0.2114, + "step": 23100 + }, + { + "epoch": 0.8398139399665674, + "grad_norm": 0.7470118403434753, + "learning_rate": 4.953173580221121e-05, + "loss": 0.13, + "step": 23110 + }, + { + "epoch": 0.8401773384693655, + "grad_norm": 1.040595531463623, + "learning_rate": 4.953073560163426e-05, + "loss": 0.2088, + "step": 23120 + }, + { + "epoch": 0.8405407369721637, + "grad_norm": 3.9858949184417725, + "learning_rate": 4.95297343441152e-05, + "loss": 0.1528, + "step": 23130 + }, + { + "epoch": 0.8409041354749618, + "grad_norm": 1.4031178951263428, + "learning_rate": 4.952873202969716e-05, + "loss": 2.5826, + "step": 23140 + }, + { + "epoch": 0.84126753397776, + "grad_norm": 16.660646438598633, + "learning_rate": 4.952772865842332e-05, + "loss": 0.3101, + "step": 23150 + }, + { + "epoch": 0.8416309324805582, + "grad_norm": 1.21910560131073, + "learning_rate": 4.952672423033693e-05, + "loss": 0.1326, + "step": 23160 + }, + { + "epoch": 0.8419943309833563, + "grad_norm": 1.4494057893753052, + "learning_rate": 4.952571874548126e-05, + "loss": 0.1567, + "step": 23170 + }, + { + "epoch": 0.8423577294861545, + "grad_norm": 1.1903733015060425, + "learning_rate": 4.952471220389964e-05, + "loss": 0.1537, + "step": 23180 + }, + { + "epoch": 0.8427211279889527, + "grad_norm": 1.0293620824813843, + "learning_rate": 4.9523704605635414e-05, + "loss": 0.1695, + "step": 23190 + }, + { + "epoch": 0.8430845264917508, + "grad_norm": 9.536385536193848, + "learning_rate": 4.9522695950732025e-05, + "loss": 0.2702, + "step": 23200 + }, + { + "epoch": 0.843447924994549, + "grad_norm": 1.1565468311309814, + "learning_rate": 4.9521686239232915e-05, + "loss": 0.1452, + "step": 23210 + }, + { + "epoch": 0.8438113234973472, + "grad_norm": 1.0805953741073608, + "learning_rate": 4.9520675471181586e-05, + "loss": 0.1478, + "step": 23220 + }, + { + "epoch": 0.8441747220001453, + "grad_norm": 2.7216696739196777, + "learning_rate": 4.95196636466216e-05, + "loss": 0.1965, + "step": 23230 + }, + { + "epoch": 0.8445381205029435, + "grad_norm": 2.2064578533172607, + "learning_rate": 4.9518650765596564e-05, + "loss": 0.213, + "step": 23240 + }, + { + "epoch": 0.8449015190057417, + "grad_norm": 11.686285972595215, + "learning_rate": 4.951763682815009e-05, + "loss": 0.2929, + "step": 23250 + }, + { + "epoch": 0.8452649175085398, + "grad_norm": 1.6271568536758423, + "learning_rate": 4.9516621834325885e-05, + "loss": 0.1406, + "step": 23260 + }, + { + "epoch": 0.845628316011338, + "grad_norm": 2.791619300842285, + "learning_rate": 4.951560578416767e-05, + "loss": 0.1431, + "step": 23270 + }, + { + "epoch": 0.8459917145141362, + "grad_norm": 1.9396895170211792, + "learning_rate": 4.951458867771923e-05, + "loss": 0.1516, + "step": 23280 + }, + { + "epoch": 0.8463551130169343, + "grad_norm": 0.9364364147186279, + "learning_rate": 4.951357051502439e-05, + "loss": 0.1935, + "step": 23290 + }, + { + "epoch": 0.8467185115197325, + "grad_norm": 2.275146007537842, + "learning_rate": 4.9512551296127005e-05, + "loss": 0.1832, + "step": 23300 + }, + { + "epoch": 0.8470819100225307, + "grad_norm": 1.4089415073394775, + "learning_rate": 4.951153102107101e-05, + "loss": 0.1511, + "step": 23310 + }, + { + "epoch": 0.8474453085253288, + "grad_norm": 1.2446107864379883, + "learning_rate": 4.951050968990035e-05, + "loss": 0.282, + "step": 23320 + }, + { + "epoch": 0.847808707028127, + "grad_norm": 2.595438241958618, + "learning_rate": 4.950948730265905e-05, + "loss": 0.1643, + "step": 23330 + }, + { + "epoch": 0.8481721055309253, + "grad_norm": 1.1884585618972778, + "learning_rate": 4.950846385939114e-05, + "loss": 0.1445, + "step": 23340 + }, + { + "epoch": 0.8485355040337234, + "grad_norm": 33.609004974365234, + "learning_rate": 4.9507439360140716e-05, + "loss": 0.185, + "step": 23350 + }, + { + "epoch": 0.8488989025365216, + "grad_norm": 0.573637068271637, + "learning_rate": 4.950641380495194e-05, + "loss": 0.1417, + "step": 23360 + }, + { + "epoch": 0.8492623010393198, + "grad_norm": 1.1126424074172974, + "learning_rate": 4.9505387193868975e-05, + "loss": 0.1592, + "step": 23370 + }, + { + "epoch": 0.8496256995421179, + "grad_norm": 2.466045379638672, + "learning_rate": 4.9504359526936074e-05, + "loss": 0.1507, + "step": 23380 + }, + { + "epoch": 0.8499890980449161, + "grad_norm": 1.273472547531128, + "learning_rate": 4.95033308041975e-05, + "loss": 0.174, + "step": 23390 + }, + { + "epoch": 0.8503524965477143, + "grad_norm": 5.497190475463867, + "learning_rate": 4.9502301025697595e-05, + "loss": 0.2269, + "step": 23400 + }, + { + "epoch": 0.8503524965477143, + "eval_loss": 0.3661801218986511, + "eval_runtime": 181.0852, + "eval_samples_per_second": 40.942, + "eval_steps_per_second": 5.119, + "eval_wer": 0.198767404287763, + "step": 23400 + }, + { + "epoch": 0.8507158950505124, + "grad_norm": 0.740798830986023, + "learning_rate": 4.950127019148071e-05, + "loss": 0.148, + "step": 23410 + }, + { + "epoch": 0.8510792935533106, + "grad_norm": 1.7785030603408813, + "learning_rate": 4.950023830159127e-05, + "loss": 0.175, + "step": 23420 + }, + { + "epoch": 0.8514426920561087, + "grad_norm": 0.7675313949584961, + "learning_rate": 4.949920535607374e-05, + "loss": 0.1635, + "step": 23430 + }, + { + "epoch": 0.8518060905589069, + "grad_norm": 0.9880558252334595, + "learning_rate": 4.9498171354972617e-05, + "loss": 0.1732, + "step": 23440 + }, + { + "epoch": 0.8521694890617051, + "grad_norm": 5.804686069488525, + "learning_rate": 4.9497136298332454e-05, + "loss": 0.2142, + "step": 23450 + }, + { + "epoch": 0.8525328875645032, + "grad_norm": 1.063359022140503, + "learning_rate": 4.949610018619785e-05, + "loss": 0.1529, + "step": 23460 + }, + { + "epoch": 0.8528962860673014, + "grad_norm": 1.9043885469436646, + "learning_rate": 4.949506301861344e-05, + "loss": 0.1633, + "step": 23470 + }, + { + "epoch": 0.8532596845700996, + "grad_norm": 2.0380702018737793, + "learning_rate": 4.9494024795623926e-05, + "loss": 0.1595, + "step": 23480 + }, + { + "epoch": 0.8536230830728977, + "grad_norm": 1.65935218334198, + "learning_rate": 4.949298551727403e-05, + "loss": 0.1526, + "step": 23490 + }, + { + "epoch": 0.8539864815756959, + "grad_norm": 1.7575215101242065, + "learning_rate": 4.9491945183608536e-05, + "loss": 0.1924, + "step": 23500 + }, + { + "epoch": 0.8543498800784941, + "grad_norm": 2.332193374633789, + "learning_rate": 4.949090379467226e-05, + "loss": 0.1536, + "step": 23510 + }, + { + "epoch": 0.8547132785812922, + "grad_norm": 1.0475032329559326, + "learning_rate": 4.948986135051009e-05, + "loss": 0.1322, + "step": 23520 + }, + { + "epoch": 0.8550766770840904, + "grad_norm": 3.1753509044647217, + "learning_rate": 4.948881785116692e-05, + "loss": 0.1457, + "step": 23530 + }, + { + "epoch": 0.8554400755868886, + "grad_norm": 0.7468664646148682, + "learning_rate": 4.948777329668772e-05, + "loss": 0.1385, + "step": 23540 + }, + { + "epoch": 0.8558034740896867, + "grad_norm": 6.77406120300293, + "learning_rate": 4.9486727687117507e-05, + "loss": 0.19, + "step": 23550 + }, + { + "epoch": 0.8561668725924849, + "grad_norm": 1.6008226871490479, + "learning_rate": 4.9485681022501316e-05, + "loss": 0.1609, + "step": 23560 + }, + { + "epoch": 0.8565302710952831, + "grad_norm": 1.1062623262405396, + "learning_rate": 4.948463330288425e-05, + "loss": 0.1624, + "step": 23570 + }, + { + "epoch": 0.8568936695980812, + "grad_norm": 1.6599873304367065, + "learning_rate": 4.948358452831145e-05, + "loss": 0.1532, + "step": 23580 + }, + { + "epoch": 0.8572570681008794, + "grad_norm": 1.264592170715332, + "learning_rate": 4.9482534698828106e-05, + "loss": 0.1696, + "step": 23590 + }, + { + "epoch": 0.8576204666036776, + "grad_norm": 2.027796745300293, + "learning_rate": 4.948148381447945e-05, + "loss": 0.1913, + "step": 23600 + }, + { + "epoch": 0.8579838651064757, + "grad_norm": 1.3213417530059814, + "learning_rate": 4.948043187531076e-05, + "loss": 0.1517, + "step": 23610 + }, + { + "epoch": 0.8583472636092739, + "grad_norm": 1.6190669536590576, + "learning_rate": 4.9479378881367366e-05, + "loss": 0.1517, + "step": 23620 + }, + { + "epoch": 0.8587106621120721, + "grad_norm": 5.381803512573242, + "learning_rate": 4.947832483269464e-05, + "loss": 0.1504, + "step": 23630 + }, + { + "epoch": 0.8590740606148702, + "grad_norm": 3.4807474613189697, + "learning_rate": 4.947726972933798e-05, + "loss": 0.1887, + "step": 23640 + }, + { + "epoch": 0.8594374591176684, + "grad_norm": 4.890349864959717, + "learning_rate": 4.947621357134287e-05, + "loss": 0.219, + "step": 23650 + }, + { + "epoch": 0.8598008576204667, + "grad_norm": 1.1006419658660889, + "learning_rate": 4.947515635875479e-05, + "loss": 0.1743, + "step": 23660 + }, + { + "epoch": 0.8601642561232647, + "grad_norm": 0.9933237433433533, + "learning_rate": 4.9474098091619314e-05, + "loss": 0.1294, + "step": 23670 + }, + { + "epoch": 0.860527654626063, + "grad_norm": 3.392524480819702, + "learning_rate": 4.947303876998203e-05, + "loss": 0.1784, + "step": 23680 + }, + { + "epoch": 0.8608910531288612, + "grad_norm": 1.466454029083252, + "learning_rate": 4.947197839388857e-05, + "loss": 0.1828, + "step": 23690 + }, + { + "epoch": 0.8612544516316593, + "grad_norm": 3.670731544494629, + "learning_rate": 4.947091696338465e-05, + "loss": 0.1772, + "step": 23700 + }, + { + "epoch": 0.8616178501344575, + "grad_norm": 1.3586241006851196, + "learning_rate": 4.9469854478515976e-05, + "loss": 0.1512, + "step": 23710 + }, + { + "epoch": 0.8619812486372556, + "grad_norm": 0.8312864303588867, + "learning_rate": 4.9468790939328336e-05, + "loss": 0.1582, + "step": 23720 + }, + { + "epoch": 0.8623446471400538, + "grad_norm": 0.9825647473335266, + "learning_rate": 4.946772634586756e-05, + "loss": 0.1662, + "step": 23730 + }, + { + "epoch": 0.862708045642852, + "grad_norm": 2.7960050106048584, + "learning_rate": 4.94666606981795e-05, + "loss": 0.226, + "step": 23740 + }, + { + "epoch": 0.8630714441456501, + "grad_norm": 5.3017683029174805, + "learning_rate": 4.94655939963101e-05, + "loss": 0.2065, + "step": 23750 + }, + { + "epoch": 0.8634348426484483, + "grad_norm": 1.0958201885223389, + "learning_rate": 4.946452624030529e-05, + "loss": 0.2177, + "step": 23760 + }, + { + "epoch": 0.8637982411512465, + "grad_norm": 1.0320892333984375, + "learning_rate": 4.94634574302111e-05, + "loss": 0.1263, + "step": 23770 + }, + { + "epoch": 0.8641616396540446, + "grad_norm": 1.0401560068130493, + "learning_rate": 4.946238756607356e-05, + "loss": 0.6474, + "step": 23780 + }, + { + "epoch": 0.8645250381568428, + "grad_norm": 1.378184199333191, + "learning_rate": 4.9461316647938785e-05, + "loss": 0.1783, + "step": 23790 + }, + { + "epoch": 0.864888436659641, + "grad_norm": 7.429476261138916, + "learning_rate": 4.9460244675852906e-05, + "loss": 0.2744, + "step": 23800 + }, + { + "epoch": 0.8652518351624391, + "grad_norm": 2.2409234046936035, + "learning_rate": 4.945917164986211e-05, + "loss": 0.2088, + "step": 23810 + }, + { + "epoch": 0.8656152336652373, + "grad_norm": 1.1307353973388672, + "learning_rate": 4.945809757001264e-05, + "loss": 0.1311, + "step": 23820 + }, + { + "epoch": 0.8659786321680355, + "grad_norm": 1.6061898469924927, + "learning_rate": 4.945702243635077e-05, + "loss": 0.1683, + "step": 23830 + }, + { + "epoch": 0.8663420306708336, + "grad_norm": 1.0011060237884521, + "learning_rate": 4.945594624892281e-05, + "loss": 0.8323, + "step": 23840 + }, + { + "epoch": 0.8667054291736318, + "grad_norm": 6.631030082702637, + "learning_rate": 4.9454869007775154e-05, + "loss": 0.177, + "step": 23850 + }, + { + "epoch": 0.86706882767643, + "grad_norm": 2.8532910346984863, + "learning_rate": 4.9453790712954195e-05, + "loss": 0.145, + "step": 23860 + }, + { + "epoch": 0.8674322261792281, + "grad_norm": 2.6437554359436035, + "learning_rate": 4.945271136450641e-05, + "loss": 0.1496, + "step": 23870 + }, + { + "epoch": 0.8677956246820263, + "grad_norm": 3.0070180892944336, + "learning_rate": 4.945163096247829e-05, + "loss": 0.1582, + "step": 23880 + }, + { + "epoch": 0.8681590231848245, + "grad_norm": 0.8612903356552124, + "learning_rate": 4.9450549506916386e-05, + "loss": 0.157, + "step": 23890 + }, + { + "epoch": 0.8685224216876226, + "grad_norm": 9.475138664245605, + "learning_rate": 4.94494669978673e-05, + "loss": 0.312, + "step": 23900 + }, + { + "epoch": 0.8688858201904208, + "grad_norm": 0.789193868637085, + "learning_rate": 4.944838343537768e-05, + "loss": 0.1385, + "step": 23910 + }, + { + "epoch": 0.869249218693219, + "grad_norm": 0.9372280240058899, + "learning_rate": 4.94472988194942e-05, + "loss": 0.1581, + "step": 23920 + }, + { + "epoch": 0.8696126171960171, + "grad_norm": 4.738519191741943, + "learning_rate": 4.94462131502636e-05, + "loss": 0.1693, + "step": 23930 + }, + { + "epoch": 0.8699760156988153, + "grad_norm": 0.9660571217536926, + "learning_rate": 4.9445126427732654e-05, + "loss": 0.1578, + "step": 23940 + }, + { + "epoch": 0.8703394142016135, + "grad_norm": 8.137104034423828, + "learning_rate": 4.944403865194818e-05, + "loss": 0.1857, + "step": 23950 + }, + { + "epoch": 0.8707028127044116, + "grad_norm": 1.1240946054458618, + "learning_rate": 4.944294982295706e-05, + "loss": 0.2508, + "step": 23960 + }, + { + "epoch": 0.8710662112072098, + "grad_norm": 3.6192643642425537, + "learning_rate": 4.94418599408062e-05, + "loss": 0.1354, + "step": 23970 + }, + { + "epoch": 0.871429609710008, + "grad_norm": 2.76771879196167, + "learning_rate": 4.944076900554256e-05, + "loss": 0.1638, + "step": 23980 + }, + { + "epoch": 0.8717930082128061, + "grad_norm": 1.734529972076416, + "learning_rate": 4.9439677017213143e-05, + "loss": 0.1414, + "step": 23990 + }, + { + "epoch": 0.8721564067156043, + "grad_norm": 6.897458553314209, + "learning_rate": 4.9438583975864996e-05, + "loss": 0.2154, + "step": 24000 + }, + { + "epoch": 0.8721564067156043, + "eval_loss": 0.37997984886169434, + "eval_runtime": 180.3101, + "eval_samples_per_second": 41.118, + "eval_steps_per_second": 5.141, + "eval_wer": 0.20322399114128561, + "step": 24000 + }, + { + "epoch": 0.8725198052184026, + "grad_norm": 1.5639888048171997, + "learning_rate": 4.943748988154523e-05, + "loss": 0.1372, + "step": 24010 + }, + { + "epoch": 0.8728832037212007, + "grad_norm": 4.484424114227295, + "learning_rate": 4.943639473430096e-05, + "loss": 0.3205, + "step": 24020 + }, + { + "epoch": 0.8732466022239989, + "grad_norm": 1.9517849683761597, + "learning_rate": 4.9435298534179396e-05, + "loss": 0.2085, + "step": 24030 + }, + { + "epoch": 0.873610000726797, + "grad_norm": 1.3041925430297852, + "learning_rate": 4.943420128122776e-05, + "loss": 0.1446, + "step": 24040 + }, + { + "epoch": 0.8739733992295952, + "grad_norm": 29.67850685119629, + "learning_rate": 4.943310297549332e-05, + "loss": 0.2643, + "step": 24050 + }, + { + "epoch": 0.8743367977323934, + "grad_norm": 4.462527751922607, + "learning_rate": 4.9432003617023405e-05, + "loss": 0.2067, + "step": 24060 + }, + { + "epoch": 0.8747001962351915, + "grad_norm": 1.2176992893218994, + "learning_rate": 4.9430903205865384e-05, + "loss": 0.1353, + "step": 24070 + }, + { + "epoch": 0.8750635947379897, + "grad_norm": 2.044191360473633, + "learning_rate": 4.9429801742066675e-05, + "loss": 0.1632, + "step": 24080 + }, + { + "epoch": 0.8754269932407879, + "grad_norm": 3.0303845405578613, + "learning_rate": 4.942869922567473e-05, + "loss": 0.1533, + "step": 24090 + }, + { + "epoch": 0.875790391743586, + "grad_norm": 4.44179105758667, + "learning_rate": 4.942759565673705e-05, + "loss": 0.2054, + "step": 24100 + }, + { + "epoch": 0.8761537902463842, + "grad_norm": 2.158686637878418, + "learning_rate": 4.942649103530119e-05, + "loss": 0.1457, + "step": 24110 + }, + { + "epoch": 0.8765171887491824, + "grad_norm": 5.875476837158203, + "learning_rate": 4.942538536141473e-05, + "loss": 0.1941, + "step": 24120 + }, + { + "epoch": 0.8768805872519805, + "grad_norm": 1.7252172231674194, + "learning_rate": 4.9424278635125335e-05, + "loss": 0.155, + "step": 24130 + }, + { + "epoch": 0.8772439857547787, + "grad_norm": 1.6594487428665161, + "learning_rate": 4.9423170856480674e-05, + "loss": 0.1736, + "step": 24140 + }, + { + "epoch": 0.8776073842575769, + "grad_norm": 6.2919697761535645, + "learning_rate": 4.9422062025528474e-05, + "loss": 0.2313, + "step": 24150 + }, + { + "epoch": 0.877970782760375, + "grad_norm": 2.1133229732513428, + "learning_rate": 4.942095214231651e-05, + "loss": 0.1642, + "step": 24160 + }, + { + "epoch": 0.8783341812631732, + "grad_norm": 1.02867591381073, + "learning_rate": 4.941984120689262e-05, + "loss": 0.1554, + "step": 24170 + }, + { + "epoch": 0.8786975797659714, + "grad_norm": 1.7262704372406006, + "learning_rate": 4.941872921930465e-05, + "loss": 0.1428, + "step": 24180 + }, + { + "epoch": 0.8790609782687695, + "grad_norm": 1.095211386680603, + "learning_rate": 4.9417616179600526e-05, + "loss": 0.1683, + "step": 24190 + }, + { + "epoch": 0.8794243767715677, + "grad_norm": 9.772414207458496, + "learning_rate": 4.94165020878282e-05, + "loss": 0.2224, + "step": 24200 + }, + { + "epoch": 0.8797877752743659, + "grad_norm": 0.6741021871566772, + "learning_rate": 4.9415386944035665e-05, + "loss": 0.7216, + "step": 24210 + }, + { + "epoch": 0.880151173777164, + "grad_norm": 0.6714327335357666, + "learning_rate": 4.941427074827098e-05, + "loss": 0.1321, + "step": 24220 + }, + { + "epoch": 0.8805145722799622, + "grad_norm": 9.116118431091309, + "learning_rate": 4.941315350058223e-05, + "loss": 0.1738, + "step": 24230 + }, + { + "epoch": 0.8808779707827604, + "grad_norm": 1.119581937789917, + "learning_rate": 4.941203520101757e-05, + "loss": 0.1076, + "step": 24240 + }, + { + "epoch": 0.8812413692855585, + "grad_norm": 1.5630614757537842, + "learning_rate": 4.941091584962516e-05, + "loss": 0.1734, + "step": 24250 + }, + { + "epoch": 0.8816047677883567, + "grad_norm": 3.4376001358032227, + "learning_rate": 4.940979544645325e-05, + "loss": 0.1567, + "step": 24260 + }, + { + "epoch": 0.8819681662911549, + "grad_norm": 1.1688649654388428, + "learning_rate": 4.94086739915501e-05, + "loss": 0.137, + "step": 24270 + }, + { + "epoch": 0.882331564793953, + "grad_norm": 2.02235746383667, + "learning_rate": 4.9407551484964035e-05, + "loss": 0.1718, + "step": 24280 + }, + { + "epoch": 0.8826949632967512, + "grad_norm": 1.7484105825424194, + "learning_rate": 4.940642792674341e-05, + "loss": 0.1973, + "step": 24290 + }, + { + "epoch": 0.8830583617995494, + "grad_norm": 7.056839942932129, + "learning_rate": 4.940530331693666e-05, + "loss": 0.1916, + "step": 24300 + }, + { + "epoch": 0.8834217603023475, + "grad_norm": 1.4804614782333374, + "learning_rate": 4.940417765559221e-05, + "loss": 0.1418, + "step": 24310 + }, + { + "epoch": 0.8837851588051457, + "grad_norm": 1.3168327808380127, + "learning_rate": 4.940305094275859e-05, + "loss": 0.1466, + "step": 24320 + }, + { + "epoch": 0.8841485573079438, + "grad_norm": 2.4612350463867188, + "learning_rate": 4.9401923178484325e-05, + "loss": 0.1956, + "step": 24330 + }, + { + "epoch": 0.884511955810742, + "grad_norm": 0.8389832973480225, + "learning_rate": 4.9400794362818005e-05, + "loss": 0.1751, + "step": 24340 + }, + { + "epoch": 0.8848753543135403, + "grad_norm": 2.618521213531494, + "learning_rate": 4.939966449580828e-05, + "loss": 0.2133, + "step": 24350 + }, + { + "epoch": 0.8852387528163383, + "grad_norm": 0.767784833908081, + "learning_rate": 4.9398533577503826e-05, + "loss": 0.1256, + "step": 24360 + }, + { + "epoch": 0.8856021513191366, + "grad_norm": 1.7649836540222168, + "learning_rate": 4.939740160795336e-05, + "loss": 0.1925, + "step": 24370 + }, + { + "epoch": 0.8859655498219348, + "grad_norm": 2.182840347290039, + "learning_rate": 4.9396268587205685e-05, + "loss": 0.184, + "step": 24380 + }, + { + "epoch": 0.8863289483247329, + "grad_norm": 1.6524356603622437, + "learning_rate": 4.939513451530958e-05, + "loss": 0.1582, + "step": 24390 + }, + { + "epoch": 0.8866923468275311, + "grad_norm": 13.93655776977539, + "learning_rate": 4.939399939231394e-05, + "loss": 0.1813, + "step": 24400 + }, + { + "epoch": 0.8870557453303293, + "grad_norm": 1.9153752326965332, + "learning_rate": 4.939286321826766e-05, + "loss": 0.2093, + "step": 24410 + }, + { + "epoch": 0.8874191438331274, + "grad_norm": 1.9444178342819214, + "learning_rate": 4.9391725993219685e-05, + "loss": 0.1489, + "step": 24420 + }, + { + "epoch": 0.8877825423359256, + "grad_norm": 2.9371562004089355, + "learning_rate": 4.939058771721903e-05, + "loss": 0.1648, + "step": 24430 + }, + { + "epoch": 0.8881459408387238, + "grad_norm": 3.127439498901367, + "learning_rate": 4.938944839031473e-05, + "loss": 0.1756, + "step": 24440 + }, + { + "epoch": 0.8885093393415219, + "grad_norm": 11.735489845275879, + "learning_rate": 4.938830801255588e-05, + "loss": 0.2049, + "step": 24450 + }, + { + "epoch": 0.8888727378443201, + "grad_norm": 1.0685577392578125, + "learning_rate": 4.938716658399161e-05, + "loss": 0.147, + "step": 24460 + }, + { + "epoch": 0.8892361363471183, + "grad_norm": 3.6975417137145996, + "learning_rate": 4.93860241046711e-05, + "loss": 0.1402, + "step": 24470 + }, + { + "epoch": 0.8895995348499164, + "grad_norm": 1.703731894493103, + "learning_rate": 4.938488057464358e-05, + "loss": 0.1418, + "step": 24480 + }, + { + "epoch": 0.8899629333527146, + "grad_norm": 1.5911983251571655, + "learning_rate": 4.938373599395831e-05, + "loss": 0.1268, + "step": 24490 + }, + { + "epoch": 0.8903263318555128, + "grad_norm": 5.278975486755371, + "learning_rate": 4.9382590362664613e-05, + "loss": 0.2388, + "step": 24500 + }, + { + "epoch": 0.8906897303583109, + "grad_norm": 1.673403263092041, + "learning_rate": 4.9381443680811865e-05, + "loss": 0.1568, + "step": 24510 + }, + { + "epoch": 0.8910531288611091, + "grad_norm": 0.5384930968284607, + "learning_rate": 4.938029594844945e-05, + "loss": 0.1364, + "step": 24520 + }, + { + "epoch": 0.8914165273639073, + "grad_norm": 1.4231863021850586, + "learning_rate": 4.937914716562683e-05, + "loss": 0.1358, + "step": 24530 + }, + { + "epoch": 0.8917799258667054, + "grad_norm": 1.2151052951812744, + "learning_rate": 4.937799733239349e-05, + "loss": 0.1673, + "step": 24540 + }, + { + "epoch": 0.8921433243695036, + "grad_norm": 9.278292655944824, + "learning_rate": 4.937684644879899e-05, + "loss": 0.2505, + "step": 24550 + }, + { + "epoch": 0.8925067228723018, + "grad_norm": 2.3570127487182617, + "learning_rate": 4.937569451489291e-05, + "loss": 0.1447, + "step": 24560 + }, + { + "epoch": 0.8928701213750999, + "grad_norm": 0.44337037205696106, + "learning_rate": 4.937454153072488e-05, + "loss": 0.2015, + "step": 24570 + }, + { + "epoch": 0.8932335198778981, + "grad_norm": 2.4552314281463623, + "learning_rate": 4.937338749634458e-05, + "loss": 0.1838, + "step": 24580 + }, + { + "epoch": 0.8935969183806963, + "grad_norm": 0.9864338636398315, + "learning_rate": 4.937223241180174e-05, + "loss": 0.1356, + "step": 24590 + }, + { + "epoch": 0.8939603168834944, + "grad_norm": 8.218843460083008, + "learning_rate": 4.937107627714612e-05, + "loss": 0.2109, + "step": 24600 + }, + { + "epoch": 0.8939603168834944, + "eval_loss": 0.38069987297058105, + "eval_runtime": 180.4244, + "eval_samples_per_second": 41.092, + "eval_steps_per_second": 5.138, + "eval_wer": 0.20163559460489772, + "step": 24600 + }, + { + "epoch": 0.8943237153862926, + "grad_norm": 0.7269652485847473, + "learning_rate": 4.936991909242753e-05, + "loss": 0.1756, + "step": 24610 + }, + { + "epoch": 0.8946871138890907, + "grad_norm": 0.9835095405578613, + "learning_rate": 4.9368760857695836e-05, + "loss": 0.1297, + "step": 24620 + }, + { + "epoch": 0.8950505123918889, + "grad_norm": 3.5632708072662354, + "learning_rate": 4.9367601573000944e-05, + "loss": 0.146, + "step": 24630 + }, + { + "epoch": 0.8954139108946871, + "grad_norm": 0.7898311614990234, + "learning_rate": 4.93664412383928e-05, + "loss": 0.1693, + "step": 24640 + }, + { + "epoch": 0.8957773093974852, + "grad_norm": 3.8220248222351074, + "learning_rate": 4.93652798539214e-05, + "loss": 0.1739, + "step": 24650 + }, + { + "epoch": 0.8961407079002834, + "grad_norm": 0.7946699857711792, + "learning_rate": 4.936411741963678e-05, + "loss": 0.1271, + "step": 24660 + }, + { + "epoch": 0.8965041064030816, + "grad_norm": 1.5677101612091064, + "learning_rate": 4.936295393558903e-05, + "loss": 0.1365, + "step": 24670 + }, + { + "epoch": 0.8968675049058797, + "grad_norm": 18.39532470703125, + "learning_rate": 4.9361789401828285e-05, + "loss": 0.2035, + "step": 24680 + }, + { + "epoch": 0.897230903408678, + "grad_norm": 2.577984094619751, + "learning_rate": 4.93606238184047e-05, + "loss": 0.127, + "step": 24690 + }, + { + "epoch": 0.8975943019114762, + "grad_norm": 3.4822871685028076, + "learning_rate": 4.9359457185368515e-05, + "loss": 0.2335, + "step": 24700 + }, + { + "epoch": 0.8979577004142743, + "grad_norm": 1.6475412845611572, + "learning_rate": 4.935828950277e-05, + "loss": 0.1581, + "step": 24710 + }, + { + "epoch": 0.8983210989170725, + "grad_norm": 2.0972635746002197, + "learning_rate": 4.9357120770659446e-05, + "loss": 0.1608, + "step": 24720 + }, + { + "epoch": 0.8986844974198707, + "grad_norm": 3.194946050643921, + "learning_rate": 4.9355950989087226e-05, + "loss": 0.1911, + "step": 24730 + }, + { + "epoch": 0.8990478959226688, + "grad_norm": 1.1382654905319214, + "learning_rate": 4.9354780158103744e-05, + "loss": 0.1671, + "step": 24740 + }, + { + "epoch": 0.899411294425467, + "grad_norm": 7.309133052825928, + "learning_rate": 4.9353608277759433e-05, + "loss": 0.192, + "step": 24750 + }, + { + "epoch": 0.8997746929282652, + "grad_norm": 1.0215349197387695, + "learning_rate": 4.9352435348104786e-05, + "loss": 0.1713, + "step": 24760 + }, + { + "epoch": 0.9001380914310633, + "grad_norm": 2.319836378097534, + "learning_rate": 4.935126136919035e-05, + "loss": 0.1441, + "step": 24770 + }, + { + "epoch": 0.9005014899338615, + "grad_norm": 3.443413496017456, + "learning_rate": 4.9350086341066716e-05, + "loss": 0.2136, + "step": 24780 + }, + { + "epoch": 0.9008648884366597, + "grad_norm": 0.9862478971481323, + "learning_rate": 4.934891026378449e-05, + "loss": 0.134, + "step": 24790 + }, + { + "epoch": 0.9012282869394578, + "grad_norm": 10.3681640625, + "learning_rate": 4.934773313739435e-05, + "loss": 0.3034, + "step": 24800 + }, + { + "epoch": 0.901591685442256, + "grad_norm": 0.9848408102989197, + "learning_rate": 4.9346554961947014e-05, + "loss": 0.1503, + "step": 24810 + }, + { + "epoch": 0.9019550839450542, + "grad_norm": 1.3456752300262451, + "learning_rate": 4.934537573749326e-05, + "loss": 2.2881, + "step": 24820 + }, + { + "epoch": 0.9023184824478523, + "grad_norm": 0.8639931082725525, + "learning_rate": 4.9344195464083884e-05, + "loss": 0.1565, + "step": 24830 + }, + { + "epoch": 0.9026818809506505, + "grad_norm": 1.1297109127044678, + "learning_rate": 4.9343014141769744e-05, + "loss": 0.1338, + "step": 24840 + }, + { + "epoch": 0.9030452794534487, + "grad_norm": 20.8160343170166, + "learning_rate": 4.934183177060173e-05, + "loss": 0.2155, + "step": 24850 + }, + { + "epoch": 0.9034086779562468, + "grad_norm": 0.8113746643066406, + "learning_rate": 4.9340648350630804e-05, + "loss": 0.126, + "step": 24860 + }, + { + "epoch": 0.903772076459045, + "grad_norm": 1.7760541439056396, + "learning_rate": 4.9339463881907946e-05, + "loss": 0.119, + "step": 24870 + }, + { + "epoch": 0.9041354749618432, + "grad_norm": 1.8657050132751465, + "learning_rate": 4.933827836448418e-05, + "loss": 0.1772, + "step": 24880 + }, + { + "epoch": 0.9044988734646413, + "grad_norm": 1.2576991319656372, + "learning_rate": 4.9337091798410594e-05, + "loss": 0.1609, + "step": 24890 + }, + { + "epoch": 0.9048622719674395, + "grad_norm": 4.8249311447143555, + "learning_rate": 4.933590418373833e-05, + "loss": 1.7033, + "step": 24900 + }, + { + "epoch": 0.9052256704702376, + "grad_norm": 1.065819501876831, + "learning_rate": 4.9334715520518526e-05, + "loss": 0.1559, + "step": 24910 + }, + { + "epoch": 0.9055890689730358, + "grad_norm": 0.961330771446228, + "learning_rate": 4.933352580880242e-05, + "loss": 0.1459, + "step": 24920 + }, + { + "epoch": 0.905952467475834, + "grad_norm": 2.0911202430725098, + "learning_rate": 4.933233504864126e-05, + "loss": 0.2173, + "step": 24930 + }, + { + "epoch": 0.9063158659786321, + "grad_norm": 0.5074183940887451, + "learning_rate": 4.933114324008636e-05, + "loss": 0.1544, + "step": 24940 + }, + { + "epoch": 0.9066792644814303, + "grad_norm": 3.663172483444214, + "learning_rate": 4.932995038318907e-05, + "loss": 0.2042, + "step": 24950 + }, + { + "epoch": 0.9070426629842285, + "grad_norm": 1.691545844078064, + "learning_rate": 4.9328756478000784e-05, + "loss": 0.1616, + "step": 24960 + }, + { + "epoch": 0.9074060614870266, + "grad_norm": 1.6613342761993408, + "learning_rate": 4.9327561524572944e-05, + "loss": 0.1212, + "step": 24970 + }, + { + "epoch": 0.9077694599898248, + "grad_norm": 2.5737128257751465, + "learning_rate": 4.9326365522957044e-05, + "loss": 0.1753, + "step": 24980 + }, + { + "epoch": 0.908132858492623, + "grad_norm": 1.717429280281067, + "learning_rate": 4.932516847320459e-05, + "loss": 0.1436, + "step": 24990 + }, + { + "epoch": 0.9084962569954211, + "grad_norm": 13.324812889099121, + "learning_rate": 4.9323970375367186e-05, + "loss": 0.1983, + "step": 25000 + }, + { + "epoch": 0.9088596554982193, + "grad_norm": 1.374232530593872, + "learning_rate": 4.932277122949644e-05, + "loss": 0.1588, + "step": 25010 + }, + { + "epoch": 0.9092230540010175, + "grad_norm": 1.1790850162506104, + "learning_rate": 4.932157103564402e-05, + "loss": 0.1603, + "step": 25020 + }, + { + "epoch": 0.9095864525038156, + "grad_norm": 2.7326996326446533, + "learning_rate": 4.932036979386165e-05, + "loss": 0.1656, + "step": 25030 + }, + { + "epoch": 0.9099498510066139, + "grad_norm": 1.2364397048950195, + "learning_rate": 4.931916750420107e-05, + "loss": 0.2311, + "step": 25040 + }, + { + "epoch": 0.9103132495094121, + "grad_norm": 3.7070934772491455, + "learning_rate": 4.9317964166714095e-05, + "loss": 0.2286, + "step": 25050 + }, + { + "epoch": 0.9106766480122102, + "grad_norm": 2.05336594581604, + "learning_rate": 4.931675978145256e-05, + "loss": 0.1404, + "step": 25060 + }, + { + "epoch": 0.9110400465150084, + "grad_norm": 1.3064135313034058, + "learning_rate": 4.931555434846837e-05, + "loss": 0.1395, + "step": 25070 + }, + { + "epoch": 0.9114034450178066, + "grad_norm": 1.252254843711853, + "learning_rate": 4.931434786781346e-05, + "loss": 0.1595, + "step": 25080 + }, + { + "epoch": 0.9117668435206047, + "grad_norm": 1.399654507637024, + "learning_rate": 4.931314033953981e-05, + "loss": 0.1495, + "step": 25090 + }, + { + "epoch": 0.9121302420234029, + "grad_norm": 9.340110778808594, + "learning_rate": 4.931193176369945e-05, + "loss": 0.2489, + "step": 25100 + }, + { + "epoch": 0.9124936405262011, + "grad_norm": 1.4071942567825317, + "learning_rate": 4.931072214034445e-05, + "loss": 0.1409, + "step": 25110 + }, + { + "epoch": 0.9128570390289992, + "grad_norm": 1.5617743730545044, + "learning_rate": 4.9309511469526934e-05, + "loss": 0.2026, + "step": 25120 + }, + { + "epoch": 0.9132204375317974, + "grad_norm": 1.4382219314575195, + "learning_rate": 4.930829975129906e-05, + "loss": 0.1426, + "step": 25130 + }, + { + "epoch": 0.9135838360345956, + "grad_norm": 1.0388094186782837, + "learning_rate": 4.930708698571303e-05, + "loss": 0.132, + "step": 25140 + }, + { + "epoch": 0.9139472345373937, + "grad_norm": 3.9398436546325684, + "learning_rate": 4.9305873172821126e-05, + "loss": 0.2257, + "step": 25150 + }, + { + "epoch": 0.9143106330401919, + "grad_norm": 2.5586395263671875, + "learning_rate": 4.930465831267562e-05, + "loss": 0.1508, + "step": 25160 + }, + { + "epoch": 0.9146740315429901, + "grad_norm": 1.6908849477767944, + "learning_rate": 4.930344240532886e-05, + "loss": 0.1407, + "step": 25170 + }, + { + "epoch": 0.9150374300457882, + "grad_norm": 3.980564594268799, + "learning_rate": 4.930222545083324e-05, + "loss": 0.1749, + "step": 25180 + }, + { + "epoch": 0.9154008285485864, + "grad_norm": 1.7451142072677612, + "learning_rate": 4.930100744924119e-05, + "loss": 0.1415, + "step": 25190 + }, + { + "epoch": 0.9157642270513845, + "grad_norm": 11.09490966796875, + "learning_rate": 4.9299788400605194e-05, + "loss": 0.248, + "step": 25200 + }, + { + "epoch": 0.9157642270513845, + "eval_loss": 0.36305877566337585, + "eval_runtime": 180.0742, + "eval_samples_per_second": 41.172, + "eval_steps_per_second": 5.148, + "eval_wer": 0.19528200845934612, + "step": 25200 + }, + { + "epoch": 0.9161276255541827, + "grad_norm": 0.9552545547485352, + "learning_rate": 4.929856830497778e-05, + "loss": 0.1765, + "step": 25210 + }, + { + "epoch": 0.9164910240569809, + "grad_norm": 1.0652204751968384, + "learning_rate": 4.929734716241151e-05, + "loss": 0.1412, + "step": 25220 + }, + { + "epoch": 0.916854422559779, + "grad_norm": 2.473240375518799, + "learning_rate": 4.929612497295899e-05, + "loss": 0.1511, + "step": 25230 + }, + { + "epoch": 0.9172178210625772, + "grad_norm": 2.0563089847564697, + "learning_rate": 4.929490173667291e-05, + "loss": 0.1562, + "step": 25240 + }, + { + "epoch": 0.9175812195653754, + "grad_norm": 5.446952819824219, + "learning_rate": 4.929367745360593e-05, + "loss": 0.2416, + "step": 25250 + }, + { + "epoch": 0.9179446180681735, + "grad_norm": 0.724795937538147, + "learning_rate": 4.929245212381085e-05, + "loss": 0.1554, + "step": 25260 + }, + { + "epoch": 0.9183080165709717, + "grad_norm": 1.0962814092636108, + "learning_rate": 4.929122574734043e-05, + "loss": 0.1567, + "step": 25270 + }, + { + "epoch": 0.9186714150737699, + "grad_norm": 1.3689608573913574, + "learning_rate": 4.9289998324247524e-05, + "loss": 0.1498, + "step": 25280 + }, + { + "epoch": 0.919034813576568, + "grad_norm": 3.039569139480591, + "learning_rate": 4.9288769854585015e-05, + "loss": 0.1666, + "step": 25290 + }, + { + "epoch": 0.9193982120793662, + "grad_norm": 10.71928882598877, + "learning_rate": 4.928754033840583e-05, + "loss": 0.2487, + "step": 25300 + }, + { + "epoch": 0.9197616105821644, + "grad_norm": 0.47624918818473816, + "learning_rate": 4.928630977576295e-05, + "loss": 0.1457, + "step": 25310 + }, + { + "epoch": 0.9201250090849625, + "grad_norm": 1.2840664386749268, + "learning_rate": 4.9285078166709386e-05, + "loss": 0.1437, + "step": 25320 + }, + { + "epoch": 0.9204884075877607, + "grad_norm": 2.118415117263794, + "learning_rate": 4.928384551129822e-05, + "loss": 0.1861, + "step": 25330 + }, + { + "epoch": 0.9208518060905589, + "grad_norm": 0.8363248109817505, + "learning_rate": 4.928261180958255e-05, + "loss": 0.1494, + "step": 25340 + }, + { + "epoch": 0.921215204593357, + "grad_norm": 20.23488998413086, + "learning_rate": 4.928137706161553e-05, + "loss": 0.2414, + "step": 25350 + }, + { + "epoch": 0.9215786030961552, + "grad_norm": 1.1590826511383057, + "learning_rate": 4.928014126745037e-05, + "loss": 0.162, + "step": 25360 + }, + { + "epoch": 0.9219420015989535, + "grad_norm": 1.1986241340637207, + "learning_rate": 4.9278904427140315e-05, + "loss": 0.1323, + "step": 25370 + }, + { + "epoch": 0.9223054001017515, + "grad_norm": 5.075083255767822, + "learning_rate": 4.927766654073864e-05, + "loss": 0.2944, + "step": 25380 + }, + { + "epoch": 0.9226687986045498, + "grad_norm": 3.1853582859039307, + "learning_rate": 4.927642760829871e-05, + "loss": 0.1792, + "step": 25390 + }, + { + "epoch": 0.923032197107348, + "grad_norm": 5.919759273529053, + "learning_rate": 4.927518762987388e-05, + "loss": 0.2182, + "step": 25400 + }, + { + "epoch": 0.9233955956101461, + "grad_norm": 1.4639918804168701, + "learning_rate": 4.927394660551759e-05, + "loss": 0.1277, + "step": 25410 + }, + { + "epoch": 0.9237589941129443, + "grad_norm": 1.205178141593933, + "learning_rate": 4.927270453528331e-05, + "loss": 0.1197, + "step": 25420 + }, + { + "epoch": 0.9241223926157425, + "grad_norm": 1.4328303337097168, + "learning_rate": 4.927146141922455e-05, + "loss": 0.1522, + "step": 25430 + }, + { + "epoch": 0.9244857911185406, + "grad_norm": 0.6114678382873535, + "learning_rate": 4.927021725739488e-05, + "loss": 0.1661, + "step": 25440 + }, + { + "epoch": 0.9248491896213388, + "grad_norm": 34.13093566894531, + "learning_rate": 4.92689720498479e-05, + "loss": 0.2852, + "step": 25450 + }, + { + "epoch": 0.925212588124137, + "grad_norm": 0.9967424273490906, + "learning_rate": 4.9267725796637256e-05, + "loss": 0.1433, + "step": 25460 + }, + { + "epoch": 0.9255759866269351, + "grad_norm": 1.0493268966674805, + "learning_rate": 4.926647849781666e-05, + "loss": 0.1361, + "step": 25470 + }, + { + "epoch": 0.9259393851297333, + "grad_norm": 2.582016944885254, + "learning_rate": 4.926523015343985e-05, + "loss": 0.1829, + "step": 25480 + }, + { + "epoch": 0.9263027836325314, + "grad_norm": 5.122225284576416, + "learning_rate": 4.92639807635606e-05, + "loss": 0.1148, + "step": 25490 + }, + { + "epoch": 0.9266661821353296, + "grad_norm": 8.054966926574707, + "learning_rate": 4.9262730328232755e-05, + "loss": 0.2363, + "step": 25500 + }, + { + "epoch": 0.9270295806381278, + "grad_norm": 3.3668735027313232, + "learning_rate": 4.926147884751018e-05, + "loss": 0.1311, + "step": 25510 + }, + { + "epoch": 0.9273929791409259, + "grad_norm": 1.0643728971481323, + "learning_rate": 4.926022632144681e-05, + "loss": 0.1318, + "step": 25520 + }, + { + "epoch": 0.9277563776437241, + "grad_norm": 1.632354497909546, + "learning_rate": 4.9258972750096614e-05, + "loss": 0.1958, + "step": 25530 + }, + { + "epoch": 0.9281197761465223, + "grad_norm": 0.7638659477233887, + "learning_rate": 4.9257718133513586e-05, + "loss": 0.168, + "step": 25540 + }, + { + "epoch": 0.9284831746493204, + "grad_norm": 4.14115571975708, + "learning_rate": 4.9256462471751796e-05, + "loss": 0.1976, + "step": 25550 + }, + { + "epoch": 0.9288465731521186, + "grad_norm": 39.925689697265625, + "learning_rate": 4.925520576486534e-05, + "loss": 0.67, + "step": 25560 + }, + { + "epoch": 0.9292099716549168, + "grad_norm": 1.349623441696167, + "learning_rate": 4.9253948012908366e-05, + "loss": 0.1475, + "step": 25570 + }, + { + "epoch": 0.9295733701577149, + "grad_norm": 10.941556930541992, + "learning_rate": 4.925268921593508e-05, + "loss": 0.1696, + "step": 25580 + }, + { + "epoch": 0.9299367686605131, + "grad_norm": 1.5406817197799683, + "learning_rate": 4.925142937399969e-05, + "loss": 0.1444, + "step": 25590 + }, + { + "epoch": 0.9303001671633113, + "grad_norm": 3.9542319774627686, + "learning_rate": 4.925016848715651e-05, + "loss": 0.216, + "step": 25600 + }, + { + "epoch": 0.9306635656661094, + "grad_norm": 2.0055665969848633, + "learning_rate": 4.924890655545984e-05, + "loss": 0.1248, + "step": 25610 + }, + { + "epoch": 0.9310269641689076, + "grad_norm": 2.145512819290161, + "learning_rate": 4.924764357896408e-05, + "loss": 0.1278, + "step": 25620 + }, + { + "epoch": 0.9313903626717058, + "grad_norm": 6.076485633850098, + "learning_rate": 4.924637955772361e-05, + "loss": 0.1586, + "step": 25630 + }, + { + "epoch": 0.9317537611745039, + "grad_norm": 0.9482760429382324, + "learning_rate": 4.924511449179293e-05, + "loss": 0.1547, + "step": 25640 + }, + { + "epoch": 0.9321171596773021, + "grad_norm": 2.335090398788452, + "learning_rate": 4.924384838122653e-05, + "loss": 0.1709, + "step": 25650 + }, + { + "epoch": 0.9324805581801003, + "grad_norm": 2.1309449672698975, + "learning_rate": 4.924258122607895e-05, + "loss": 0.1425, + "step": 25660 + }, + { + "epoch": 0.9328439566828984, + "grad_norm": 1.092887282371521, + "learning_rate": 4.924131302640482e-05, + "loss": 0.1578, + "step": 25670 + }, + { + "epoch": 0.9332073551856966, + "grad_norm": 0.7325641512870789, + "learning_rate": 4.9240043782258746e-05, + "loss": 0.1473, + "step": 25680 + }, + { + "epoch": 0.9335707536884948, + "grad_norm": 1.296338677406311, + "learning_rate": 4.9238773493695443e-05, + "loss": 0.2279, + "step": 25690 + }, + { + "epoch": 0.9339341521912929, + "grad_norm": 1.196590542793274, + "learning_rate": 4.923750216076963e-05, + "loss": 0.1524, + "step": 25700 + }, + { + "epoch": 0.9342975506940912, + "grad_norm": 1.5417845249176025, + "learning_rate": 4.923622978353608e-05, + "loss": 0.1385, + "step": 25710 + }, + { + "epoch": 0.9346609491968894, + "grad_norm": 1.4865704774856567, + "learning_rate": 4.923495636204963e-05, + "loss": 0.1435, + "step": 25720 + }, + { + "epoch": 0.9350243476996875, + "grad_norm": 1.6445010900497437, + "learning_rate": 4.923368189636513e-05, + "loss": 0.223, + "step": 25730 + }, + { + "epoch": 0.9353877462024857, + "grad_norm": 0.6629343032836914, + "learning_rate": 4.9232406386537505e-05, + "loss": 0.1479, + "step": 25740 + }, + { + "epoch": 0.9357511447052839, + "grad_norm": 8.440834999084473, + "learning_rate": 4.923112983262171e-05, + "loss": 0.7624, + "step": 25750 + }, + { + "epoch": 0.936114543208082, + "grad_norm": 1.088809847831726, + "learning_rate": 4.922985223467274e-05, + "loss": 0.134, + "step": 25760 + }, + { + "epoch": 0.9364779417108802, + "grad_norm": 1.1839587688446045, + "learning_rate": 4.922857359274565e-05, + "loss": 0.1284, + "step": 25770 + }, + { + "epoch": 0.9368413402136783, + "grad_norm": 2.278588056564331, + "learning_rate": 4.922729390689553e-05, + "loss": 0.1873, + "step": 25780 + }, + { + "epoch": 0.9372047387164765, + "grad_norm": 1.6524765491485596, + "learning_rate": 4.9226013177177515e-05, + "loss": 0.1769, + "step": 25790 + }, + { + "epoch": 0.9375681372192747, + "grad_norm": 18.044713973999023, + "learning_rate": 4.922473140364679e-05, + "loss": 0.2122, + "step": 25800 + }, + { + "epoch": 0.9375681372192747, + "eval_loss": 0.3881298005580902, + "eval_runtime": 180.2979, + "eval_samples_per_second": 41.121, + "eval_steps_per_second": 5.141, + "eval_wer": 0.1963348884491804, + "step": 25800 + }, + { + "epoch": 0.9379315357220728, + "grad_norm": 1.1691884994506836, + "learning_rate": 4.9223448586358576e-05, + "loss": 0.1573, + "step": 25810 + }, + { + "epoch": 0.938294934224871, + "grad_norm": 1.1012376546859741, + "learning_rate": 4.9222164725368156e-05, + "loss": 0.1511, + "step": 25820 + }, + { + "epoch": 0.9386583327276692, + "grad_norm": 2.1937880516052246, + "learning_rate": 4.9220879820730844e-05, + "loss": 0.1684, + "step": 25830 + }, + { + "epoch": 0.9390217312304673, + "grad_norm": 1.5964059829711914, + "learning_rate": 4.921959387250199e-05, + "loss": 0.1897, + "step": 25840 + }, + { + "epoch": 0.9393851297332655, + "grad_norm": 6.693167209625244, + "learning_rate": 4.921830688073701e-05, + "loss": 0.2155, + "step": 25850 + }, + { + "epoch": 0.9397485282360637, + "grad_norm": 1.679046869277954, + "learning_rate": 4.921701884549136e-05, + "loss": 0.1566, + "step": 25860 + }, + { + "epoch": 0.9401119267388618, + "grad_norm": 0.648047924041748, + "learning_rate": 4.9215729766820536e-05, + "loss": 0.1398, + "step": 25870 + }, + { + "epoch": 0.94047532524166, + "grad_norm": 0.7286267876625061, + "learning_rate": 4.921443964478007e-05, + "loss": 0.1598, + "step": 25880 + }, + { + "epoch": 0.9408387237444582, + "grad_norm": 1.3676726818084717, + "learning_rate": 4.921314847942555e-05, + "loss": 0.1627, + "step": 25890 + }, + { + "epoch": 0.9412021222472563, + "grad_norm": 11.982099533081055, + "learning_rate": 4.921185627081263e-05, + "loss": 0.2181, + "step": 25900 + }, + { + "epoch": 0.9415655207500545, + "grad_norm": 0.8863544464111328, + "learning_rate": 4.9210563018996955e-05, + "loss": 0.1296, + "step": 25910 + }, + { + "epoch": 0.9419289192528527, + "grad_norm": 0.8388992547988892, + "learning_rate": 4.9209268724034265e-05, + "loss": 0.1406, + "step": 25920 + }, + { + "epoch": 0.9422923177556508, + "grad_norm": 2.4800333976745605, + "learning_rate": 4.9207973385980324e-05, + "loss": 0.1694, + "step": 25930 + }, + { + "epoch": 0.942655716258449, + "grad_norm": 4.2597174644470215, + "learning_rate": 4.920667700489093e-05, + "loss": 0.9439, + "step": 25940 + }, + { + "epoch": 0.9430191147612472, + "grad_norm": 5.32108736038208, + "learning_rate": 4.920537958082196e-05, + "loss": 0.1745, + "step": 25950 + }, + { + "epoch": 0.9433825132640453, + "grad_norm": 1.3563112020492554, + "learning_rate": 4.9204081113829316e-05, + "loss": 0.1554, + "step": 25960 + }, + { + "epoch": 0.9437459117668435, + "grad_norm": 8.575587272644043, + "learning_rate": 4.9202781603968926e-05, + "loss": 0.2015, + "step": 25970 + }, + { + "epoch": 0.9441093102696417, + "grad_norm": 6.85026216506958, + "learning_rate": 4.920148105129679e-05, + "loss": 0.1548, + "step": 25980 + }, + { + "epoch": 0.9444727087724398, + "grad_norm": 1.2886810302734375, + "learning_rate": 4.9200179455868944e-05, + "loss": 0.136, + "step": 25990 + }, + { + "epoch": 0.944836107275238, + "grad_norm": 2.0779457092285156, + "learning_rate": 4.919887681774148e-05, + "loss": 0.1744, + "step": 26000 + }, + { + "epoch": 0.9451995057780362, + "grad_norm": Infinity, + "learning_rate": 4.919770355196496e-05, + "loss": 2.7706, + "step": 26010 + }, + { + "epoch": 0.9455629042808343, + "grad_norm": 0.9514101147651672, + "learning_rate": 4.919639893286285e-05, + "loss": 0.1435, + "step": 26020 + }, + { + "epoch": 0.9459263027836325, + "grad_norm": 0.3761270046234131, + "learning_rate": 4.9195093271224016e-05, + "loss": 0.1525, + "step": 26030 + }, + { + "epoch": 0.9462897012864308, + "grad_norm": 1.2147834300994873, + "learning_rate": 4.919378656710469e-05, + "loss": 0.1922, + "step": 26040 + }, + { + "epoch": 0.9466530997892288, + "grad_norm": 15.408570289611816, + "learning_rate": 4.919247882056119e-05, + "loss": 0.2773, + "step": 26050 + }, + { + "epoch": 0.947016498292027, + "grad_norm": 2.2306370735168457, + "learning_rate": 4.919117003164985e-05, + "loss": 0.1446, + "step": 26060 + }, + { + "epoch": 0.9473798967948253, + "grad_norm": 1.3414242267608643, + "learning_rate": 4.918986020042706e-05, + "loss": 0.1484, + "step": 26070 + }, + { + "epoch": 0.9477432952976234, + "grad_norm": 1.9740337133407593, + "learning_rate": 4.9188549326949275e-05, + "loss": 0.1845, + "step": 26080 + }, + { + "epoch": 0.9481066938004216, + "grad_norm": 0.7002670764923096, + "learning_rate": 4.9187237411272955e-05, + "loss": 0.1559, + "step": 26090 + }, + { + "epoch": 0.9484700923032197, + "grad_norm": 8.308074951171875, + "learning_rate": 4.9185924453454635e-05, + "loss": 0.223, + "step": 26100 + }, + { + "epoch": 0.9488334908060179, + "grad_norm": 0.8129051327705383, + "learning_rate": 4.9184610453550884e-05, + "loss": 0.1459, + "step": 26110 + }, + { + "epoch": 0.9491968893088161, + "grad_norm": 1.5998592376708984, + "learning_rate": 4.918329541161831e-05, + "loss": 0.1394, + "step": 26120 + }, + { + "epoch": 0.9495602878116142, + "grad_norm": 1.8726842403411865, + "learning_rate": 4.918197932771359e-05, + "loss": 0.1859, + "step": 26130 + }, + { + "epoch": 0.9499236863144124, + "grad_norm": 1.1915557384490967, + "learning_rate": 4.9180662201893424e-05, + "loss": 0.1621, + "step": 26140 + }, + { + "epoch": 0.9502870848172106, + "grad_norm": 6.970126152038574, + "learning_rate": 4.917934403421455e-05, + "loss": 0.2613, + "step": 26150 + }, + { + "epoch": 0.9506504833200087, + "grad_norm": 1.0738050937652588, + "learning_rate": 4.9178024824733776e-05, + "loss": 0.1383, + "step": 26160 + }, + { + "epoch": 0.9510138818228069, + "grad_norm": 2.1130123138427734, + "learning_rate": 4.9176704573507933e-05, + "loss": 0.222, + "step": 26170 + }, + { + "epoch": 0.9513772803256051, + "grad_norm": 3.1722593307495117, + "learning_rate": 4.9175383280593925e-05, + "loss": 0.1624, + "step": 26180 + }, + { + "epoch": 0.9517406788284032, + "grad_norm": 0.9101456999778748, + "learning_rate": 4.917406094604865e-05, + "loss": 0.16, + "step": 26190 + }, + { + "epoch": 0.9521040773312014, + "grad_norm": 3.284672260284424, + "learning_rate": 4.917273756992911e-05, + "loss": 0.1817, + "step": 26200 + }, + { + "epoch": 0.9524674758339996, + "grad_norm": 2.2083284854888916, + "learning_rate": 4.917141315229232e-05, + "loss": 0.169, + "step": 26210 + }, + { + "epoch": 0.9528308743367977, + "grad_norm": 4.354351997375488, + "learning_rate": 4.9170087693195335e-05, + "loss": 0.2629, + "step": 26220 + }, + { + "epoch": 0.9531942728395959, + "grad_norm": 2.520522117614746, + "learning_rate": 4.916876119269526e-05, + "loss": 0.1852, + "step": 26230 + }, + { + "epoch": 0.9535576713423941, + "grad_norm": 0.8573399186134338, + "learning_rate": 4.9167433650849264e-05, + "loss": 0.1524, + "step": 26240 + }, + { + "epoch": 0.9539210698451922, + "grad_norm": 6.314918041229248, + "learning_rate": 4.916610506771454e-05, + "loss": 0.2685, + "step": 26250 + }, + { + "epoch": 0.9542844683479904, + "grad_norm": 3.0610973834991455, + "learning_rate": 4.916477544334833e-05, + "loss": 0.1374, + "step": 26260 + }, + { + "epoch": 0.9546478668507886, + "grad_norm": 0.9085964560508728, + "learning_rate": 4.916344477780793e-05, + "loss": 0.1754, + "step": 26270 + }, + { + "epoch": 0.9550112653535867, + "grad_norm": 3.8550243377685547, + "learning_rate": 4.916211307115067e-05, + "loss": 0.3546, + "step": 26280 + }, + { + "epoch": 0.9553746638563849, + "grad_norm": 5.278194904327393, + "learning_rate": 4.916078032343392e-05, + "loss": 0.1298, + "step": 26290 + }, + { + "epoch": 0.9557380623591831, + "grad_norm": 2.707965612411499, + "learning_rate": 4.9159446534715116e-05, + "loss": 0.1689, + "step": 26300 + }, + { + "epoch": 0.9561014608619812, + "grad_norm": 1.3821223974227905, + "learning_rate": 4.9158111705051716e-05, + "loss": 0.2117, + "step": 26310 + }, + { + "epoch": 0.9564648593647794, + "grad_norm": 1.0195057392120361, + "learning_rate": 4.915677583450123e-05, + "loss": 0.1151, + "step": 26320 + }, + { + "epoch": 0.9568282578675776, + "grad_norm": 2.078343629837036, + "learning_rate": 4.915543892312124e-05, + "loss": 0.1433, + "step": 26330 + }, + { + "epoch": 0.9571916563703757, + "grad_norm": 1.6972254514694214, + "learning_rate": 4.915410097096932e-05, + "loss": 0.1307, + "step": 26340 + }, + { + "epoch": 0.9575550548731739, + "grad_norm": 4.440702438354492, + "learning_rate": 4.915276197810313e-05, + "loss": 0.1806, + "step": 26350 + }, + { + "epoch": 0.9579184533759721, + "grad_norm": 0.778567373752594, + "learning_rate": 4.9151421944580374e-05, + "loss": 0.177, + "step": 26360 + }, + { + "epoch": 0.9582818518787702, + "grad_norm": 1.2955224514007568, + "learning_rate": 4.915008087045877e-05, + "loss": 0.1395, + "step": 26370 + }, + { + "epoch": 0.9586452503815684, + "grad_norm": 2.077195405960083, + "learning_rate": 4.9148738755796104e-05, + "loss": 0.1583, + "step": 26380 + }, + { + "epoch": 0.9590086488843665, + "grad_norm": 0.8736408352851868, + "learning_rate": 4.914739560065021e-05, + "loss": 0.1284, + "step": 26390 + }, + { + "epoch": 0.9593720473871648, + "grad_norm": 2.9465060234069824, + "learning_rate": 4.914605140507895e-05, + "loss": 0.1929, + "step": 26400 + }, + { + "epoch": 0.9593720473871648, + "eval_loss": 0.3894718587398529, + "eval_runtime": 180.7535, + "eval_samples_per_second": 41.017, + "eval_steps_per_second": 5.129, + "eval_wer": 0.18666836095630548, + "step": 26400 + }, + { + "epoch": 0.959735445889963, + "grad_norm": 0.7856747508049011, + "learning_rate": 4.9144706169140256e-05, + "loss": 0.1548, + "step": 26410 + }, + { + "epoch": 0.960098844392761, + "grad_norm": 3.3245174884796143, + "learning_rate": 4.914335989289208e-05, + "loss": 0.1328, + "step": 26420 + }, + { + "epoch": 0.9604622428955593, + "grad_norm": 4.848336219787598, + "learning_rate": 4.914201257639243e-05, + "loss": 0.145, + "step": 26430 + }, + { + "epoch": 0.9608256413983575, + "grad_norm": 2.2661678791046143, + "learning_rate": 4.9140664219699344e-05, + "loss": 0.1482, + "step": 26440 + }, + { + "epoch": 0.9611890399011556, + "grad_norm": 6.279752731323242, + "learning_rate": 4.913931482287094e-05, + "loss": 0.2087, + "step": 26450 + }, + { + "epoch": 0.9615524384039538, + "grad_norm": 0.9201165437698364, + "learning_rate": 4.913796438596534e-05, + "loss": 0.1641, + "step": 26460 + }, + { + "epoch": 0.961915836906752, + "grad_norm": 1.0935853719711304, + "learning_rate": 4.9136612909040746e-05, + "loss": 0.1678, + "step": 26470 + }, + { + "epoch": 0.9622792354095501, + "grad_norm": 5.511369705200195, + "learning_rate": 4.913526039215538e-05, + "loss": 0.2284, + "step": 26480 + }, + { + "epoch": 0.9626426339123483, + "grad_norm": 0.8109707832336426, + "learning_rate": 4.913390683536751e-05, + "loss": 0.1314, + "step": 26490 + }, + { + "epoch": 0.9630060324151465, + "grad_norm": 20.594274520874023, + "learning_rate": 4.9132552238735464e-05, + "loss": 0.2612, + "step": 26500 + }, + { + "epoch": 0.9633694309179446, + "grad_norm": 40.1435661315918, + "learning_rate": 4.913119660231761e-05, + "loss": 0.5943, + "step": 26510 + }, + { + "epoch": 0.9637328294207428, + "grad_norm": 0.6886749863624573, + "learning_rate": 4.912983992617235e-05, + "loss": 0.1445, + "step": 26520 + }, + { + "epoch": 0.964096227923541, + "grad_norm": 4.847496032714844, + "learning_rate": 4.912848221035815e-05, + "loss": 0.1645, + "step": 26530 + }, + { + "epoch": 0.9644596264263391, + "grad_norm": 1.3625943660736084, + "learning_rate": 4.912712345493349e-05, + "loss": 0.1403, + "step": 26540 + }, + { + "epoch": 0.9648230249291373, + "grad_norm": 6.022468090057373, + "learning_rate": 4.9125763659956934e-05, + "loss": 0.2215, + "step": 26550 + }, + { + "epoch": 0.9651864234319355, + "grad_norm": 1.5577186346054077, + "learning_rate": 4.912440282548706e-05, + "loss": 0.1401, + "step": 26560 + }, + { + "epoch": 0.9655498219347336, + "grad_norm": 1.5680512189865112, + "learning_rate": 4.91230409515825e-05, + "loss": 0.1373, + "step": 26570 + }, + { + "epoch": 0.9659132204375318, + "grad_norm": 2.3043782711029053, + "learning_rate": 4.912167803830193e-05, + "loss": 0.1501, + "step": 26580 + }, + { + "epoch": 0.96627661894033, + "grad_norm": 1.5168496370315552, + "learning_rate": 4.912031408570409e-05, + "loss": 0.1665, + "step": 26590 + }, + { + "epoch": 0.9666400174431281, + "grad_norm": 10.529095649719238, + "learning_rate": 4.911894909384773e-05, + "loss": 0.1642, + "step": 26600 + }, + { + "epoch": 0.9670034159459263, + "grad_norm": 2.458815097808838, + "learning_rate": 4.911758306279167e-05, + "loss": 3.3307, + "step": 26610 + }, + { + "epoch": 0.9673668144487245, + "grad_norm": 1.2745519876480103, + "learning_rate": 4.911621599259477e-05, + "loss": 0.1681, + "step": 26620 + }, + { + "epoch": 0.9677302129515226, + "grad_norm": 1.877960443496704, + "learning_rate": 4.911484788331593e-05, + "loss": 0.1445, + "step": 26630 + }, + { + "epoch": 0.9680936114543208, + "grad_norm": 1.3567255735397339, + "learning_rate": 4.911347873501408e-05, + "loss": 0.132, + "step": 26640 + }, + { + "epoch": 0.968457009957119, + "grad_norm": 3.44063138961792, + "learning_rate": 4.911210854774825e-05, + "loss": 0.205, + "step": 26650 + }, + { + "epoch": 0.9688204084599171, + "grad_norm": 5.335951805114746, + "learning_rate": 4.911073732157744e-05, + "loss": 0.1475, + "step": 26660 + }, + { + "epoch": 0.9691838069627153, + "grad_norm": 3.0675578117370605, + "learning_rate": 4.910936505656074e-05, + "loss": 0.1602, + "step": 26670 + }, + { + "epoch": 0.9695472054655134, + "grad_norm": 5.95693826675415, + "learning_rate": 4.910799175275729e-05, + "loss": 0.1888, + "step": 26680 + }, + { + "epoch": 0.9699106039683116, + "grad_norm": 1.7128913402557373, + "learning_rate": 4.910661741022625e-05, + "loss": 0.1402, + "step": 26690 + }, + { + "epoch": 0.9702740024711098, + "grad_norm": 11.855730056762695, + "learning_rate": 4.9105242029026844e-05, + "loss": 0.1939, + "step": 26700 + }, + { + "epoch": 0.9706374009739079, + "grad_norm": 2.21028208732605, + "learning_rate": 4.910386560921831e-05, + "loss": 0.1365, + "step": 26710 + }, + { + "epoch": 0.9710007994767061, + "grad_norm": 3.6761391162872314, + "learning_rate": 4.910248815085998e-05, + "loss": 0.1661, + "step": 26720 + }, + { + "epoch": 0.9713641979795044, + "grad_norm": 1.9474952220916748, + "learning_rate": 4.9101109654011196e-05, + "loss": 0.1176, + "step": 26730 + }, + { + "epoch": 0.9717275964823024, + "grad_norm": 4.190001010894775, + "learning_rate": 4.909973011873135e-05, + "loss": 0.1426, + "step": 26740 + }, + { + "epoch": 0.9720909949851007, + "grad_norm": 2.785562753677368, + "learning_rate": 4.909834954507987e-05, + "loss": 0.201, + "step": 26750 + }, + { + "epoch": 0.9724543934878989, + "grad_norm": 2.296952724456787, + "learning_rate": 4.909696793311625e-05, + "loss": 1.5478, + "step": 26760 + }, + { + "epoch": 0.972817791990697, + "grad_norm": 1.471690058708191, + "learning_rate": 4.909558528290002e-05, + "loss": 0.1254, + "step": 26770 + }, + { + "epoch": 0.9731811904934952, + "grad_norm": 5.213918685913086, + "learning_rate": 4.9094340010048675e-05, + "loss": 0.8881, + "step": 26780 + }, + { + "epoch": 0.9735445889962934, + "grad_norm": 1.5338894128799438, + "learning_rate": 4.909295538731665e-05, + "loss": 0.1621, + "step": 26790 + }, + { + "epoch": 0.9739079874990915, + "grad_norm": 4.493140697479248, + "learning_rate": 4.909156972650491e-05, + "loss": 0.1653, + "step": 26800 + }, + { + "epoch": 0.9742713860018897, + "grad_norm": 0.9602924585342407, + "learning_rate": 4.909018302767313e-05, + "loss": 0.1388, + "step": 26810 + }, + { + "epoch": 0.9746347845046879, + "grad_norm": 1.038445234298706, + "learning_rate": 4.9088795290881085e-05, + "loss": 0.145, + "step": 26820 + }, + { + "epoch": 0.974998183007486, + "grad_norm": 3.1368119716644287, + "learning_rate": 4.908740651618856e-05, + "loss": 0.1732, + "step": 26830 + }, + { + "epoch": 0.9753615815102842, + "grad_norm": 0.6875894069671631, + "learning_rate": 4.908601670365539e-05, + "loss": 0.1582, + "step": 26840 + }, + { + "epoch": 0.9757249800130824, + "grad_norm": 14.604360580444336, + "learning_rate": 4.908462585334146e-05, + "loss": 0.228, + "step": 26850 + }, + { + "epoch": 0.9760883785158805, + "grad_norm": 1.819300889968872, + "learning_rate": 4.9083233965306694e-05, + "loss": 0.5066, + "step": 26860 + }, + { + "epoch": 0.9764517770186787, + "grad_norm": 5.712610721588135, + "learning_rate": 4.908184103961106e-05, + "loss": 0.2109, + "step": 26870 + }, + { + "epoch": 0.9768151755214769, + "grad_norm": 2.8377017974853516, + "learning_rate": 4.908044707631459e-05, + "loss": 0.1417, + "step": 26880 + }, + { + "epoch": 0.977178574024275, + "grad_norm": 1.0483819246292114, + "learning_rate": 4.907905207547733e-05, + "loss": 0.1526, + "step": 26890 + }, + { + "epoch": 0.9775419725270732, + "grad_norm": 3.996112823486328, + "learning_rate": 4.907765603715938e-05, + "loss": 0.6109, + "step": 26900 + }, + { + "epoch": 0.9779053710298714, + "grad_norm": 1.336004614830017, + "learning_rate": 4.907625896142091e-05, + "loss": 0.8939, + "step": 26910 + }, + { + "epoch": 0.9782687695326695, + "grad_norm": 0.9394060373306274, + "learning_rate": 4.907486084832211e-05, + "loss": 0.1515, + "step": 26920 + }, + { + "epoch": 0.9786321680354677, + "grad_norm": 2.356201410293579, + "learning_rate": 4.907346169792321e-05, + "loss": 0.1567, + "step": 26930 + }, + { + "epoch": 0.9789955665382659, + "grad_norm": 0.926143229007721, + "learning_rate": 4.907206151028449e-05, + "loss": 0.1669, + "step": 26940 + }, + { + "epoch": 0.979358965041064, + "grad_norm": 4.815629482269287, + "learning_rate": 4.90706602854663e-05, + "loss": 0.2068, + "step": 26950 + }, + { + "epoch": 0.9797223635438622, + "grad_norm": 1.3679453134536743, + "learning_rate": 4.906925802352899e-05, + "loss": 0.15, + "step": 26960 + }, + { + "epoch": 0.9800857620466603, + "grad_norm": 1.1221717596054077, + "learning_rate": 4.9067854724533e-05, + "loss": 0.223, + "step": 26970 + }, + { + "epoch": 0.9804491605494585, + "grad_norm": 2.198657512664795, + "learning_rate": 4.906645038853878e-05, + "loss": 0.1662, + "step": 26980 + }, + { + "epoch": 0.9808125590522567, + "grad_norm": 1.7014293670654297, + "learning_rate": 4.906504501560684e-05, + "loss": 0.1601, + "step": 26990 + }, + { + "epoch": 0.9811759575550548, + "grad_norm": 8.204177856445312, + "learning_rate": 4.906363860579774e-05, + "loss": 0.2345, + "step": 27000 + }, + { + "epoch": 0.9811759575550548, + "eval_loss": 0.3534720242023468, + "eval_runtime": 180.0448, + "eval_samples_per_second": 41.179, + "eval_steps_per_second": 5.149, + "eval_wer": 0.19012652712981284, + "step": 27000 + }, + { + "epoch": 0.981539356057853, + "grad_norm": 1.1557930707931519, + "learning_rate": 4.906223115917207e-05, + "loss": 0.1357, + "step": 27010 + }, + { + "epoch": 0.9819027545606512, + "grad_norm": 0.7808053493499756, + "learning_rate": 4.906082267579047e-05, + "loss": 0.1366, + "step": 27020 + }, + { + "epoch": 0.9822661530634493, + "grad_norm": 1.4547855854034424, + "learning_rate": 4.9059413155713626e-05, + "loss": 0.1472, + "step": 27030 + }, + { + "epoch": 0.9826295515662475, + "grad_norm": 0.5997300148010254, + "learning_rate": 4.9058002599002275e-05, + "loss": 0.1455, + "step": 27040 + }, + { + "epoch": 0.9829929500690457, + "grad_norm": 12.663732528686523, + "learning_rate": 4.90565910057172e-05, + "loss": 0.2728, + "step": 27050 + }, + { + "epoch": 0.9833563485718438, + "grad_norm": 0.9098420739173889, + "learning_rate": 4.9055178375919196e-05, + "loss": 0.1467, + "step": 27060 + }, + { + "epoch": 0.983719747074642, + "grad_norm": 3.4135582447052, + "learning_rate": 4.9053764709669156e-05, + "loss": 0.2571, + "step": 27070 + }, + { + "epoch": 0.9840831455774403, + "grad_norm": 1.7984899282455444, + "learning_rate": 4.905235000702798e-05, + "loss": 0.2026, + "step": 27080 + }, + { + "epoch": 0.9844465440802384, + "grad_norm": 1.2022316455841064, + "learning_rate": 4.9050934268056615e-05, + "loss": 0.1378, + "step": 27090 + }, + { + "epoch": 0.9848099425830366, + "grad_norm": 1.499773621559143, + "learning_rate": 4.9049517492816066e-05, + "loss": 0.1765, + "step": 27100 + }, + { + "epoch": 0.9851733410858348, + "grad_norm": 0.8122308254241943, + "learning_rate": 4.9048099681367377e-05, + "loss": 0.1422, + "step": 27110 + }, + { + "epoch": 0.9855367395886329, + "grad_norm": 1.203873872756958, + "learning_rate": 4.904668083377164e-05, + "loss": 0.1436, + "step": 27120 + }, + { + "epoch": 0.9859001380914311, + "grad_norm": 1.6162346601486206, + "learning_rate": 4.9045260950089976e-05, + "loss": 0.1667, + "step": 27130 + }, + { + "epoch": 0.9862635365942293, + "grad_norm": 0.5100680589675903, + "learning_rate": 4.904384003038358e-05, + "loss": 0.1573, + "step": 27140 + }, + { + "epoch": 0.9866269350970274, + "grad_norm": 6.364781379699707, + "learning_rate": 4.904241807471366e-05, + "loss": 0.2245, + "step": 27150 + }, + { + "epoch": 0.9869903335998256, + "grad_norm": 0.6124529242515564, + "learning_rate": 4.9040995083141495e-05, + "loss": 0.2296, + "step": 27160 + }, + { + "epoch": 0.9873537321026238, + "grad_norm": 1.3477269411087036, + "learning_rate": 4.903957105572838e-05, + "loss": 0.1173, + "step": 27170 + }, + { + "epoch": 0.9877171306054219, + "grad_norm": 0.8505461812019348, + "learning_rate": 4.903814599253569e-05, + "loss": 0.1913, + "step": 27180 + }, + { + "epoch": 0.9880805291082201, + "grad_norm": 0.928269624710083, + "learning_rate": 4.903671989362481e-05, + "loss": 0.1449, + "step": 27190 + }, + { + "epoch": 0.9884439276110183, + "grad_norm": 9.115983963012695, + "learning_rate": 4.903529275905719e-05, + "loss": 0.2025, + "step": 27200 + }, + { + "epoch": 0.9888073261138164, + "grad_norm": 0.8631362318992615, + "learning_rate": 4.903386458889434e-05, + "loss": 0.1313, + "step": 27210 + }, + { + "epoch": 0.9891707246166146, + "grad_norm": 1.5814399719238281, + "learning_rate": 4.9032435383197764e-05, + "loss": 0.1547, + "step": 27220 + }, + { + "epoch": 0.9895341231194128, + "grad_norm": 2.2507669925689697, + "learning_rate": 4.9031005142029054e-05, + "loss": 2.0754, + "step": 27230 + }, + { + "epoch": 0.9898975216222109, + "grad_norm": 2.0611305236816406, + "learning_rate": 4.902957386544984e-05, + "loss": 0.1351, + "step": 27240 + }, + { + "epoch": 0.9902609201250091, + "grad_norm": 4.219666481018066, + "learning_rate": 4.9028141553521785e-05, + "loss": 0.194, + "step": 27250 + }, + { + "epoch": 0.9906243186278072, + "grad_norm": 2.4156904220581055, + "learning_rate": 4.90267082063066e-05, + "loss": 1.9594, + "step": 27260 + }, + { + "epoch": 0.9909877171306054, + "grad_norm": 4.805545806884766, + "learning_rate": 4.9025273823866046e-05, + "loss": 0.1608, + "step": 27270 + }, + { + "epoch": 0.9913511156334036, + "grad_norm": 3.431521415710449, + "learning_rate": 4.902383840626193e-05, + "loss": 0.1439, + "step": 27280 + }, + { + "epoch": 0.9917145141362017, + "grad_norm": 0.9847241640090942, + "learning_rate": 4.902240195355609e-05, + "loss": 0.1568, + "step": 27290 + }, + { + "epoch": 0.9920779126389999, + "grad_norm": 4.65169095993042, + "learning_rate": 4.9020964465810426e-05, + "loss": 0.2039, + "step": 27300 + }, + { + "epoch": 0.9924413111417981, + "grad_norm": 1.466956377029419, + "learning_rate": 4.9019525943086865e-05, + "loss": 0.1649, + "step": 27310 + }, + { + "epoch": 0.9928047096445962, + "grad_norm": 4.803518772125244, + "learning_rate": 4.901808638544739e-05, + "loss": 0.1645, + "step": 27320 + }, + { + "epoch": 0.9931681081473944, + "grad_norm": 3.4496331214904785, + "learning_rate": 4.901664579295404e-05, + "loss": 0.1751, + "step": 27330 + }, + { + "epoch": 0.9935315066501926, + "grad_norm": 0.9507334232330322, + "learning_rate": 4.9015204165668866e-05, + "loss": 0.1228, + "step": 27340 + }, + { + "epoch": 0.9938949051529907, + "grad_norm": 5.97396993637085, + "learning_rate": 4.901376150365399e-05, + "loss": 0.2178, + "step": 27350 + }, + { + "epoch": 0.9942583036557889, + "grad_norm": 1.7720214128494263, + "learning_rate": 4.9012317806971573e-05, + "loss": 0.1468, + "step": 27360 + }, + { + "epoch": 0.9946217021585871, + "grad_norm": 0.9194307923316956, + "learning_rate": 4.9010873075683825e-05, + "loss": 0.1481, + "step": 27370 + }, + { + "epoch": 0.9949851006613852, + "grad_norm": 2.8458971977233887, + "learning_rate": 4.9009427309852986e-05, + "loss": 0.1402, + "step": 27380 + }, + { + "epoch": 0.9953484991641834, + "grad_norm": 1.9232338666915894, + "learning_rate": 4.900798050954134e-05, + "loss": 0.155, + "step": 27390 + }, + { + "epoch": 0.9957118976669816, + "grad_norm": 4.017787456512451, + "learning_rate": 4.900653267481125e-05, + "loss": 0.2279, + "step": 27400 + }, + { + "epoch": 0.9960752961697797, + "grad_norm": 0.714726448059082, + "learning_rate": 4.9005083805725064e-05, + "loss": 0.1271, + "step": 27410 + }, + { + "epoch": 0.996438694672578, + "grad_norm": 0.8059016466140747, + "learning_rate": 4.900363390234524e-05, + "loss": 0.138, + "step": 27420 + }, + { + "epoch": 0.9968020931753762, + "grad_norm": 2.650024175643921, + "learning_rate": 4.9002182964734234e-05, + "loss": 0.1483, + "step": 27430 + }, + { + "epoch": 0.9971654916781743, + "grad_norm": 1.200749397277832, + "learning_rate": 4.900073099295456e-05, + "loss": 0.9832, + "step": 27440 + }, + { + "epoch": 0.9975288901809725, + "grad_norm": 3.3051798343658447, + "learning_rate": 4.8999277987068785e-05, + "loss": 0.2136, + "step": 27450 + }, + { + "epoch": 0.9978922886837707, + "grad_norm": 1.3630801439285278, + "learning_rate": 4.899782394713951e-05, + "loss": 0.2136, + "step": 27460 + }, + { + "epoch": 0.9982556871865688, + "grad_norm": 2.5952398777008057, + "learning_rate": 4.899636887322939e-05, + "loss": 0.1515, + "step": 27470 + }, + { + "epoch": 0.998619085689367, + "grad_norm": 5.025683879852295, + "learning_rate": 4.8994912765401116e-05, + "loss": 0.1862, + "step": 27480 + }, + { + "epoch": 0.9989824841921652, + "grad_norm": 1.1604958772659302, + "learning_rate": 4.8993455623717415e-05, + "loss": 0.1433, + "step": 27490 + }, + { + "epoch": 0.9993458826949633, + "grad_norm": 14.864492416381836, + "learning_rate": 4.899199744824109e-05, + "loss": 0.1752, + "step": 27500 + }, + { + "epoch": 0.9997092811977615, + "grad_norm": 1.072911024093628, + "learning_rate": 4.8990538239034956e-05, + "loss": 0.1524, + "step": 27510 + }, + { + "epoch": 1.0000726797005597, + "grad_norm": 0.41248244047164917, + "learning_rate": 4.898907799616188e-05, + "loss": 0.1457, + "step": 27520 + }, + { + "epoch": 1.000436078203358, + "grad_norm": 1.0402699708938599, + "learning_rate": 4.89876167196848e-05, + "loss": 0.1394, + "step": 27530 + }, + { + "epoch": 1.0007994767061559, + "grad_norm": 0.8177555203437805, + "learning_rate": 4.8986154409666654e-05, + "loss": 0.1134, + "step": 27540 + }, + { + "epoch": 1.001162875208954, + "grad_norm": 1.1209142208099365, + "learning_rate": 4.8984691066170465e-05, + "loss": 0.1574, + "step": 27550 + }, + { + "epoch": 1.0015262737117523, + "grad_norm": 1.4969863891601562, + "learning_rate": 4.8983226689259264e-05, + "loss": 0.1144, + "step": 27560 + }, + { + "epoch": 1.0018896722145505, + "grad_norm": 0.8014885783195496, + "learning_rate": 4.898176127899617e-05, + "loss": 0.1616, + "step": 27570 + }, + { + "epoch": 1.0022530707173487, + "grad_norm": 1.1477352380752563, + "learning_rate": 4.89802948354443e-05, + "loss": 0.1209, + "step": 27580 + }, + { + "epoch": 1.002616469220147, + "grad_norm": 1.0199166536331177, + "learning_rate": 4.897882735866686e-05, + "loss": 0.1422, + "step": 27590 + }, + { + "epoch": 1.002979867722945, + "grad_norm": 0.8987438678741455, + "learning_rate": 4.897735884872705e-05, + "loss": 0.1328, + "step": 27600 + }, + { + "epoch": 1.002979867722945, + "eval_loss": 0.38110727071762085, + "eval_runtime": 180.6687, + "eval_samples_per_second": 41.036, + "eval_steps_per_second": 5.131, + "eval_wer": 0.18068691342785048, + "step": 27600 + }, + { + "epoch": 1.003343266225743, + "grad_norm": 1.1313180923461914, + "learning_rate": 4.897588930568817e-05, + "loss": 0.1395, + "step": 27610 + }, + { + "epoch": 1.0037066647285413, + "grad_norm": 0.6739907264709473, + "learning_rate": 4.8974418729613526e-05, + "loss": 0.2011, + "step": 27620 + }, + { + "epoch": 1.0040700632313395, + "grad_norm": 0.986926257610321, + "learning_rate": 4.8972947120566475e-05, + "loss": 0.1212, + "step": 27630 + }, + { + "epoch": 1.0044334617341377, + "grad_norm": 0.795300304889679, + "learning_rate": 4.8971474478610437e-05, + "loss": 0.1404, + "step": 27640 + }, + { + "epoch": 1.004796860236936, + "grad_norm": 1.7036499977111816, + "learning_rate": 4.897000080380885e-05, + "loss": 0.1677, + "step": 27650 + }, + { + "epoch": 1.005160258739734, + "grad_norm": 1.4313631057739258, + "learning_rate": 4.896852609622521e-05, + "loss": 0.1188, + "step": 27660 + }, + { + "epoch": 1.0055236572425321, + "grad_norm": 0.5508180260658264, + "learning_rate": 4.896705035592306e-05, + "loss": 0.1627, + "step": 27670 + }, + { + "epoch": 1.0058870557453303, + "grad_norm": 2.3307416439056396, + "learning_rate": 4.896557358296599e-05, + "loss": 0.1173, + "step": 27680 + }, + { + "epoch": 1.0062504542481285, + "grad_norm": 3.0311474800109863, + "learning_rate": 4.896409577741762e-05, + "loss": 0.1176, + "step": 27690 + }, + { + "epoch": 1.0066138527509267, + "grad_norm": 1.8580576181411743, + "learning_rate": 4.896261693934163e-05, + "loss": 0.1647, + "step": 27700 + }, + { + "epoch": 1.0069772512537247, + "grad_norm": 1.094754934310913, + "learning_rate": 4.896113706880174e-05, + "loss": 0.1137, + "step": 27710 + }, + { + "epoch": 1.007340649756523, + "grad_norm": 0.8240002393722534, + "learning_rate": 4.89596561658617e-05, + "loss": 0.1835, + "step": 27720 + }, + { + "epoch": 1.0077040482593211, + "grad_norm": 1.4678568840026855, + "learning_rate": 4.895817423058533e-05, + "loss": 0.1612, + "step": 27730 + }, + { + "epoch": 1.0080674467621193, + "grad_norm": 0.8481863737106323, + "learning_rate": 4.8956691263036473e-05, + "loss": 0.1211, + "step": 27740 + }, + { + "epoch": 1.0084308452649176, + "grad_norm": 5.045682907104492, + "learning_rate": 4.895520726327903e-05, + "loss": 0.2747, + "step": 27750 + }, + { + "epoch": 1.0087942437677158, + "grad_norm": 2.3443167209625244, + "learning_rate": 4.895372223137694e-05, + "loss": 0.1437, + "step": 27760 + }, + { + "epoch": 1.0091576422705137, + "grad_norm": 1.6994588375091553, + "learning_rate": 4.895223616739418e-05, + "loss": 0.16, + "step": 27770 + }, + { + "epoch": 1.009521040773312, + "grad_norm": 2.073699712753296, + "learning_rate": 4.8950749071394794e-05, + "loss": 0.1341, + "step": 27780 + }, + { + "epoch": 1.0098844392761102, + "grad_norm": 1.1939536333084106, + "learning_rate": 4.894926094344284e-05, + "loss": 0.1284, + "step": 27790 + }, + { + "epoch": 1.0102478377789084, + "grad_norm": 1.4820387363433838, + "learning_rate": 4.8947771783602444e-05, + "loss": 0.1644, + "step": 27800 + }, + { + "epoch": 1.0106112362817066, + "grad_norm": 1.8140612840652466, + "learning_rate": 4.894628159193778e-05, + "loss": 0.1681, + "step": 27810 + }, + { + "epoch": 1.0109746347845048, + "grad_norm": 1.7120946645736694, + "learning_rate": 4.894479036851303e-05, + "loss": 0.2066, + "step": 27820 + }, + { + "epoch": 1.0113380332873028, + "grad_norm": 1.0871057510375977, + "learning_rate": 4.894329811339247e-05, + "loss": 0.1428, + "step": 27830 + }, + { + "epoch": 1.011701431790101, + "grad_norm": 0.897597074508667, + "learning_rate": 4.8941804826640375e-05, + "loss": 0.1202, + "step": 27840 + }, + { + "epoch": 1.0120648302928992, + "grad_norm": 1.2489410638809204, + "learning_rate": 4.89403105083211e-05, + "loss": 0.1435, + "step": 27850 + }, + { + "epoch": 1.0124282287956974, + "grad_norm": 1.043281078338623, + "learning_rate": 4.893881515849902e-05, + "loss": 0.1657, + "step": 27860 + }, + { + "epoch": 1.0127916272984956, + "grad_norm": 1.0345379114151, + "learning_rate": 4.893731877723857e-05, + "loss": 0.1669, + "step": 27870 + }, + { + "epoch": 1.0131550258012938, + "grad_norm": 3.5156590938568115, + "learning_rate": 4.893582136460423e-05, + "loss": 0.1356, + "step": 27880 + }, + { + "epoch": 1.0135184243040918, + "grad_norm": 1.0468858480453491, + "learning_rate": 4.893432292066051e-05, + "loss": 0.1201, + "step": 27890 + }, + { + "epoch": 1.01388182280689, + "grad_norm": 1.3872016668319702, + "learning_rate": 4.893282344547197e-05, + "loss": 0.1472, + "step": 27900 + }, + { + "epoch": 1.0142452213096882, + "grad_norm": 0.83976811170578, + "learning_rate": 4.893132293910322e-05, + "loss": 0.1467, + "step": 27910 + }, + { + "epoch": 1.0146086198124864, + "grad_norm": 0.625514566898346, + "learning_rate": 4.892982140161892e-05, + "loss": 0.1661, + "step": 27920 + }, + { + "epoch": 1.0149720183152846, + "grad_norm": 1.0802186727523804, + "learning_rate": 4.892831883308375e-05, + "loss": 0.1444, + "step": 27930 + }, + { + "epoch": 1.0153354168180828, + "grad_norm": 0.567722499370575, + "learning_rate": 4.892681523356246e-05, + "loss": 0.1003, + "step": 27940 + }, + { + "epoch": 1.0156988153208808, + "grad_norm": 1.1036186218261719, + "learning_rate": 4.892531060311985e-05, + "loss": 0.1438, + "step": 27950 + }, + { + "epoch": 1.016062213823679, + "grad_norm": 1.2610325813293457, + "learning_rate": 4.892380494182071e-05, + "loss": 0.1478, + "step": 27960 + }, + { + "epoch": 1.0164256123264772, + "grad_norm": 4.7541913986206055, + "learning_rate": 4.892229824972995e-05, + "loss": 0.2068, + "step": 27970 + }, + { + "epoch": 1.0167890108292754, + "grad_norm": 14.794916152954102, + "learning_rate": 4.8920790526912464e-05, + "loss": 0.4626, + "step": 27980 + }, + { + "epoch": 1.0171524093320736, + "grad_norm": 0.8083056807518005, + "learning_rate": 4.891928177343323e-05, + "loss": 0.4919, + "step": 27990 + }, + { + "epoch": 1.0175158078348718, + "grad_norm": 1.1072735786437988, + "learning_rate": 4.8917771989357246e-05, + "loss": 0.2863, + "step": 28000 + }, + { + "epoch": 1.0178792063376698, + "grad_norm": 0.8811991810798645, + "learning_rate": 4.891626117474957e-05, + "loss": 0.1361, + "step": 28010 + }, + { + "epoch": 1.018242604840468, + "grad_norm": 0.43256062269210815, + "learning_rate": 4.8914749329675294e-05, + "loss": 0.1668, + "step": 28020 + }, + { + "epoch": 1.0186060033432662, + "grad_norm": 1.7490280866622925, + "learning_rate": 4.891323645419956e-05, + "loss": 0.1328, + "step": 28030 + }, + { + "epoch": 1.0189694018460644, + "grad_norm": 1.5770010948181152, + "learning_rate": 4.891172254838755e-05, + "loss": 0.1429, + "step": 28040 + }, + { + "epoch": 1.0193328003488626, + "grad_norm": 0.5603241920471191, + "learning_rate": 4.8910207612304495e-05, + "loss": 0.1319, + "step": 28050 + }, + { + "epoch": 1.0196961988516606, + "grad_norm": 1.5490175485610962, + "learning_rate": 4.890869164601566e-05, + "loss": 0.1292, + "step": 28060 + }, + { + "epoch": 1.0200595973544588, + "grad_norm": 0.7562422752380371, + "learning_rate": 4.8907174649586376e-05, + "loss": 0.1978, + "step": 28070 + }, + { + "epoch": 1.020422995857257, + "grad_norm": 2.67669415473938, + "learning_rate": 4.8905656623082e-05, + "loss": 0.1367, + "step": 28080 + }, + { + "epoch": 1.0207863943600552, + "grad_norm": 1.4589964151382446, + "learning_rate": 4.890413756656793e-05, + "loss": 0.136, + "step": 28090 + }, + { + "epoch": 1.0211497928628535, + "grad_norm": 0.5042529702186584, + "learning_rate": 4.8902617480109626e-05, + "loss": 0.1768, + "step": 28100 + }, + { + "epoch": 1.0215131913656517, + "grad_norm": 3.3886609077453613, + "learning_rate": 4.890109636377258e-05, + "loss": 0.1827, + "step": 28110 + }, + { + "epoch": 1.0218765898684496, + "grad_norm": 0.8882365226745605, + "learning_rate": 4.889957421762234e-05, + "loss": 0.2176, + "step": 28120 + }, + { + "epoch": 1.0222399883712479, + "grad_norm": 1.5471583604812622, + "learning_rate": 4.889805104172447e-05, + "loss": 0.1934, + "step": 28130 + }, + { + "epoch": 1.022603386874046, + "grad_norm": 1.221699595451355, + "learning_rate": 4.889652683614461e-05, + "loss": 0.1217, + "step": 28140 + }, + { + "epoch": 1.0229667853768443, + "grad_norm": 1.1075172424316406, + "learning_rate": 4.8895001600948444e-05, + "loss": 2.944, + "step": 28150 + }, + { + "epoch": 1.0233301838796425, + "grad_norm": 0.9731149077415466, + "learning_rate": 4.889347533620167e-05, + "loss": 0.1123, + "step": 28160 + }, + { + "epoch": 1.0236935823824407, + "grad_norm": 0.8448407649993896, + "learning_rate": 4.889194804197006e-05, + "loss": 0.1755, + "step": 28170 + }, + { + "epoch": 1.0240569808852387, + "grad_norm": 0.8480188250541687, + "learning_rate": 4.8890419718319414e-05, + "loss": 0.1434, + "step": 28180 + }, + { + "epoch": 1.0244203793880369, + "grad_norm": 1.5608705282211304, + "learning_rate": 4.8888890365315584e-05, + "loss": 0.1309, + "step": 28190 + }, + { + "epoch": 1.024783777890835, + "grad_norm": 7.765607833862305, + "learning_rate": 4.888735998302447e-05, + "loss": 0.1584, + "step": 28200 + }, + { + "epoch": 1.024783777890835, + "eval_loss": 0.3628901541233063, + "eval_runtime": 180.7805, + "eval_samples_per_second": 41.011, + "eval_steps_per_second": 5.128, + "eval_wer": 0.1830921996115236, + "step": 28200 + }, + { + "epoch": 1.0251471763936333, + "grad_norm": 2.208989381790161, + "learning_rate": 4.8885828571512e-05, + "loss": 0.1206, + "step": 28210 + }, + { + "epoch": 1.0255105748964315, + "grad_norm": 0.4320629835128784, + "learning_rate": 4.8884296130844166e-05, + "loss": 0.2776, + "step": 28220 + }, + { + "epoch": 1.0258739733992297, + "grad_norm": 1.4430392980575562, + "learning_rate": 4.888276266108699e-05, + "loss": 0.1181, + "step": 28230 + }, + { + "epoch": 1.0262373719020277, + "grad_norm": 0.893260657787323, + "learning_rate": 4.888122816230655e-05, + "loss": 0.1141, + "step": 28240 + }, + { + "epoch": 1.026600770404826, + "grad_norm": 1.9237782955169678, + "learning_rate": 4.887969263456895e-05, + "loss": 0.1676, + "step": 28250 + }, + { + "epoch": 1.026964168907624, + "grad_norm": 1.0318949222564697, + "learning_rate": 4.8878156077940376e-05, + "loss": 0.1256, + "step": 28260 + }, + { + "epoch": 1.0273275674104223, + "grad_norm": 0.8919249773025513, + "learning_rate": 4.8876618492487e-05, + "loss": 0.2314, + "step": 28270 + }, + { + "epoch": 1.0276909659132205, + "grad_norm": 1.31845223903656, + "learning_rate": 4.8875079878275085e-05, + "loss": 0.1414, + "step": 28280 + }, + { + "epoch": 1.0280543644160187, + "grad_norm": 8.070326805114746, + "learning_rate": 4.887354023537094e-05, + "loss": 0.2391, + "step": 28290 + }, + { + "epoch": 1.0284177629188167, + "grad_norm": 0.7600485682487488, + "learning_rate": 4.887199956384088e-05, + "loss": 0.164, + "step": 28300 + }, + { + "epoch": 1.028781161421615, + "grad_norm": 1.0197162628173828, + "learning_rate": 4.88704578637513e-05, + "loss": 0.1324, + "step": 28310 + }, + { + "epoch": 1.0291445599244131, + "grad_norm": 0.5989790558815002, + "learning_rate": 4.886891513516861e-05, + "loss": 0.2162, + "step": 28320 + }, + { + "epoch": 1.0295079584272113, + "grad_norm": 1.2145419120788574, + "learning_rate": 4.88673713781593e-05, + "loss": 1.7629, + "step": 28330 + }, + { + "epoch": 1.0298713569300095, + "grad_norm": 0.7220103740692139, + "learning_rate": 4.8865826592789876e-05, + "loss": 0.105, + "step": 28340 + }, + { + "epoch": 1.0302347554328075, + "grad_norm": 1.2737821340560913, + "learning_rate": 4.88642807791269e-05, + "loss": 0.172, + "step": 28350 + }, + { + "epoch": 1.0305981539356057, + "grad_norm": 2.3391408920288086, + "learning_rate": 4.886273393723698e-05, + "loss": 0.1431, + "step": 28360 + }, + { + "epoch": 1.030961552438404, + "grad_norm": 1.1937615871429443, + "learning_rate": 4.8861186067186756e-05, + "loss": 0.1776, + "step": 28370 + }, + { + "epoch": 1.0313249509412021, + "grad_norm": 0.5789287090301514, + "learning_rate": 4.885963716904292e-05, + "loss": 0.1412, + "step": 28380 + }, + { + "epoch": 1.0316883494440003, + "grad_norm": 1.2566107511520386, + "learning_rate": 4.885808724287221e-05, + "loss": 0.1284, + "step": 28390 + }, + { + "epoch": 1.0320517479467985, + "grad_norm": 5.225760459899902, + "learning_rate": 4.885653628874141e-05, + "loss": 0.1411, + "step": 28400 + }, + { + "epoch": 1.0324151464495965, + "grad_norm": 1.2525557279586792, + "learning_rate": 4.885498430671735e-05, + "loss": 0.1372, + "step": 28410 + }, + { + "epoch": 1.0327785449523947, + "grad_norm": 0.5048568844795227, + "learning_rate": 4.885343129686688e-05, + "loss": 0.1595, + "step": 28420 + }, + { + "epoch": 1.033141943455193, + "grad_norm": 0.8768513202667236, + "learning_rate": 4.8851877259256933e-05, + "loss": 0.286, + "step": 28430 + }, + { + "epoch": 1.0335053419579912, + "grad_norm": 1.2799090147018433, + "learning_rate": 4.885032219395446e-05, + "loss": 0.1431, + "step": 28440 + }, + { + "epoch": 1.0338687404607894, + "grad_norm": 0.9944593906402588, + "learning_rate": 4.8848766101026466e-05, + "loss": 0.13, + "step": 28450 + }, + { + "epoch": 1.0342321389635876, + "grad_norm": 1.3601889610290527, + "learning_rate": 4.8847208980539994e-05, + "loss": 0.1379, + "step": 28460 + }, + { + "epoch": 1.0345955374663856, + "grad_norm": 0.6347102522850037, + "learning_rate": 4.884565083256213e-05, + "loss": 0.1833, + "step": 28470 + }, + { + "epoch": 1.0349589359691838, + "grad_norm": NaN, + "learning_rate": 4.884424762093241e-05, + "loss": 3.779, + "step": 28480 + }, + { + "epoch": 1.035322334471982, + "grad_norm": 1.6947808265686035, + "learning_rate": 4.8842687520905906e-05, + "loss": 0.1571, + "step": 28490 + }, + { + "epoch": 1.0356857329747802, + "grad_norm": 4.521624565124512, + "learning_rate": 4.884112639358283e-05, + "loss": 0.1429, + "step": 28500 + }, + { + "epoch": 1.0360491314775784, + "grad_norm": 1.9370489120483398, + "learning_rate": 4.883956423903044e-05, + "loss": 0.1375, + "step": 28510 + }, + { + "epoch": 1.0364125299803766, + "grad_norm": 2.3492047786712646, + "learning_rate": 4.883800105731606e-05, + "loss": 0.1496, + "step": 28520 + }, + { + "epoch": 1.0367759284831746, + "grad_norm": 1.1862452030181885, + "learning_rate": 4.8836436848507026e-05, + "loss": 0.1239, + "step": 28530 + }, + { + "epoch": 1.0371393269859728, + "grad_norm": 2.223708391189575, + "learning_rate": 4.883487161267074e-05, + "loss": 0.1159, + "step": 28540 + }, + { + "epoch": 1.037502725488771, + "grad_norm": 5.854187965393066, + "learning_rate": 4.8833305349874636e-05, + "loss": 0.1732, + "step": 28550 + }, + { + "epoch": 1.0378661239915692, + "grad_norm": 1.4000542163848877, + "learning_rate": 4.883173806018621e-05, + "loss": 0.1428, + "step": 28560 + }, + { + "epoch": 1.0382295224943674, + "grad_norm": 1.8862130641937256, + "learning_rate": 4.883016974367298e-05, + "loss": 0.2339, + "step": 28570 + }, + { + "epoch": 1.0385929209971656, + "grad_norm": 1.701545238494873, + "learning_rate": 4.8828600400402525e-05, + "loss": 0.2063, + "step": 28580 + }, + { + "epoch": 1.0389563194999636, + "grad_norm": 3.8795692920684814, + "learning_rate": 4.8827030030442466e-05, + "loss": 0.1317, + "step": 28590 + }, + { + "epoch": 1.0393197180027618, + "grad_norm": 1.0597456693649292, + "learning_rate": 4.882545863386046e-05, + "loss": 0.1783, + "step": 28600 + }, + { + "epoch": 1.03968311650556, + "grad_norm": 0.8949028849601746, + "learning_rate": 4.88238862107242e-05, + "loss": 0.1453, + "step": 28610 + }, + { + "epoch": 1.0400465150083582, + "grad_norm": 0.6270145773887634, + "learning_rate": 4.8822312761101456e-05, + "loss": 0.2118, + "step": 28620 + }, + { + "epoch": 1.0404099135111564, + "grad_norm": 0.8819754719734192, + "learning_rate": 4.8820738285060016e-05, + "loss": 0.1398, + "step": 28630 + }, + { + "epoch": 1.0407733120139544, + "grad_norm": 1.5963236093521118, + "learning_rate": 4.881916278266772e-05, + "loss": 0.1592, + "step": 28640 + }, + { + "epoch": 1.0411367105167526, + "grad_norm": 1.2960532903671265, + "learning_rate": 4.8817586253992445e-05, + "loss": 0.2044, + "step": 28650 + }, + { + "epoch": 1.0415001090195508, + "grad_norm": 1.6735124588012695, + "learning_rate": 4.881600869910212e-05, + "loss": 0.143, + "step": 28660 + }, + { + "epoch": 1.041863507522349, + "grad_norm": 1.2382493019104004, + "learning_rate": 4.8814430118064724e-05, + "loss": 0.182, + "step": 28670 + }, + { + "epoch": 1.0422269060251472, + "grad_norm": 1.614788293838501, + "learning_rate": 4.881285051094826e-05, + "loss": 0.149, + "step": 28680 + }, + { + "epoch": 1.0425903045279454, + "grad_norm": 1.549124002456665, + "learning_rate": 4.88112698778208e-05, + "loss": 0.1238, + "step": 28690 + }, + { + "epoch": 1.0429537030307434, + "grad_norm": 0.8877584338188171, + "learning_rate": 4.8809688218750435e-05, + "loss": 0.1541, + "step": 28700 + }, + { + "epoch": 1.0433171015335416, + "grad_norm": 1.1061103343963623, + "learning_rate": 4.8808105533805325e-05, + "loss": 0.1209, + "step": 28710 + }, + { + "epoch": 1.0436805000363398, + "grad_norm": 1.8957878351211548, + "learning_rate": 4.880652182305365e-05, + "loss": 0.1739, + "step": 28720 + }, + { + "epoch": 1.044043898539138, + "grad_norm": 0.9069591164588928, + "learning_rate": 4.880493708656366e-05, + "loss": 0.2014, + "step": 28730 + }, + { + "epoch": 1.0444072970419362, + "grad_norm": 0.7086552381515503, + "learning_rate": 4.880335132440364e-05, + "loss": 0.1149, + "step": 28740 + }, + { + "epoch": 1.0447706955447345, + "grad_norm": 0.5514993667602539, + "learning_rate": 4.8801764536641883e-05, + "loss": 0.163, + "step": 28750 + }, + { + "epoch": 1.0451340940475324, + "grad_norm": 0.5786269903182983, + "learning_rate": 4.880017672334679e-05, + "loss": 0.126, + "step": 28760 + }, + { + "epoch": 1.0454974925503306, + "grad_norm": 0.8554352521896362, + "learning_rate": 4.879858788458676e-05, + "loss": 0.2564, + "step": 28770 + }, + { + "epoch": 1.0458608910531288, + "grad_norm": 3.329148769378662, + "learning_rate": 4.8796998020430253e-05, + "loss": 0.1297, + "step": 28780 + }, + { + "epoch": 1.046224289555927, + "grad_norm": 1.1520358324050903, + "learning_rate": 4.879540713094578e-05, + "loss": 0.1156, + "step": 28790 + }, + { + "epoch": 1.0465876880587253, + "grad_norm": 1.6375194787979126, + "learning_rate": 4.879381521620187e-05, + "loss": 0.1418, + "step": 28800 + }, + { + "epoch": 1.0465876880587253, + "eval_loss": 0.35767313838005066, + "eval_runtime": 180.3335, + "eval_samples_per_second": 41.113, + "eval_steps_per_second": 5.14, + "eval_wer": 0.18009693757147785, + "step": 28800 + }, + { + "epoch": 1.0469510865615235, + "grad_norm": 2.6099300384521484, + "learning_rate": 4.879222227626712e-05, + "loss": 2.0354, + "step": 28810 + }, + { + "epoch": 1.0473144850643215, + "grad_norm": 0.9497049450874329, + "learning_rate": 4.879062831121017e-05, + "loss": 0.2014, + "step": 28820 + }, + { + "epoch": 1.0476778835671197, + "grad_norm": 1.100393533706665, + "learning_rate": 4.878903332109969e-05, + "loss": 0.1294, + "step": 28830 + }, + { + "epoch": 1.0480412820699179, + "grad_norm": 0.46238216757774353, + "learning_rate": 4.87874373060044e-05, + "loss": 0.1103, + "step": 28840 + }, + { + "epoch": 1.048404680572716, + "grad_norm": 1.111619234085083, + "learning_rate": 4.8785840265993085e-05, + "loss": 0.1635, + "step": 28850 + }, + { + "epoch": 1.0487680790755143, + "grad_norm": 1.8693902492523193, + "learning_rate": 4.8784242201134534e-05, + "loss": 0.1145, + "step": 28860 + }, + { + "epoch": 1.0491314775783125, + "grad_norm": 0.5382725596427917, + "learning_rate": 4.878264311149762e-05, + "loss": 0.1699, + "step": 28870 + }, + { + "epoch": 1.0494948760811105, + "grad_norm": 1.3384134769439697, + "learning_rate": 4.878104299715123e-05, + "loss": 0.1479, + "step": 28880 + }, + { + "epoch": 1.0498582745839087, + "grad_norm": 3.7112338542938232, + "learning_rate": 4.87794418581643e-05, + "loss": 2.828, + "step": 28890 + }, + { + "epoch": 1.0502216730867069, + "grad_norm": 0.8874093890190125, + "learning_rate": 4.8777839694605844e-05, + "loss": 0.1274, + "step": 28900 + }, + { + "epoch": 1.050585071589505, + "grad_norm": 2.278064489364624, + "learning_rate": 4.877623650654487e-05, + "loss": 0.1298, + "step": 28910 + }, + { + "epoch": 1.0509484700923033, + "grad_norm": 0.8750000596046448, + "learning_rate": 4.877463229405046e-05, + "loss": 0.2705, + "step": 28920 + }, + { + "epoch": 1.0513118685951013, + "grad_norm": 0.5634777545928955, + "learning_rate": 4.8773027057191735e-05, + "loss": 0.13, + "step": 28930 + }, + { + "epoch": 1.0516752670978995, + "grad_norm": 1.1990102529525757, + "learning_rate": 4.877142079603786e-05, + "loss": 0.1115, + "step": 28940 + }, + { + "epoch": 1.0520386656006977, + "grad_norm": 5.793541431427002, + "learning_rate": 4.8769813510658035e-05, + "loss": 0.1909, + "step": 28950 + }, + { + "epoch": 1.052402064103496, + "grad_norm": 1.0433887243270874, + "learning_rate": 4.876820520112153e-05, + "loss": 0.1225, + "step": 28960 + }, + { + "epoch": 1.0527654626062941, + "grad_norm": 0.8786159753799438, + "learning_rate": 4.8766595867497624e-05, + "loss": 0.1772, + "step": 28970 + }, + { + "epoch": 1.0531288611090923, + "grad_norm": 1.1270724534988403, + "learning_rate": 4.8764985509855664e-05, + "loss": 0.1683, + "step": 28980 + }, + { + "epoch": 1.0534922596118903, + "grad_norm": 0.9916827082633972, + "learning_rate": 4.876337412826504e-05, + "loss": 0.1834, + "step": 28990 + }, + { + "epoch": 1.0538556581146885, + "grad_norm": 1.1295456886291504, + "learning_rate": 4.876176172279517e-05, + "loss": 0.1677, + "step": 29000 + }, + { + "epoch": 1.0542190566174867, + "grad_norm": 1.625546932220459, + "learning_rate": 4.876014829351553e-05, + "loss": 0.1374, + "step": 29010 + }, + { + "epoch": 1.054582455120285, + "grad_norm": 0.4282989799976349, + "learning_rate": 4.875853384049564e-05, + "loss": 0.1826, + "step": 29020 + }, + { + "epoch": 1.0549458536230831, + "grad_norm": 0.8806937336921692, + "learning_rate": 4.875691836380507e-05, + "loss": 0.116, + "step": 29030 + }, + { + "epoch": 1.0553092521258813, + "grad_norm": 0.5082537531852722, + "learning_rate": 4.87553018635134e-05, + "loss": 0.1867, + "step": 29040 + }, + { + "epoch": 1.0556726506286793, + "grad_norm": 3.172614336013794, + "learning_rate": 4.875368433969031e-05, + "loss": 0.1872, + "step": 29050 + }, + { + "epoch": 1.0560360491314775, + "grad_norm": 4.570537090301514, + "learning_rate": 4.875206579240546e-05, + "loss": 0.1501, + "step": 29060 + }, + { + "epoch": 1.0563994476342757, + "grad_norm": 0.9751003980636597, + "learning_rate": 4.875044622172862e-05, + "loss": 0.1533, + "step": 29070 + }, + { + "epoch": 1.056762846137074, + "grad_norm": 0.9446988701820374, + "learning_rate": 4.874882562772955e-05, + "loss": 0.1462, + "step": 29080 + }, + { + "epoch": 1.0571262446398721, + "grad_norm": 5.769078254699707, + "learning_rate": 4.8747204010478086e-05, + "loss": 0.1796, + "step": 29090 + }, + { + "epoch": 1.0574896431426704, + "grad_norm": 6.486478328704834, + "learning_rate": 4.8745581370044094e-05, + "loss": 0.1674, + "step": 29100 + }, + { + "epoch": 1.0578530416454683, + "grad_norm": 0.622352123260498, + "learning_rate": 4.874395770649748e-05, + "loss": 0.137, + "step": 29110 + }, + { + "epoch": 1.0582164401482665, + "grad_norm": 0.5244133472442627, + "learning_rate": 4.8742333019908215e-05, + "loss": 0.2035, + "step": 29120 + }, + { + "epoch": 1.0585798386510648, + "grad_norm": 1.7058534622192383, + "learning_rate": 4.87407073103463e-05, + "loss": 0.1408, + "step": 29130 + }, + { + "epoch": 1.058943237153863, + "grad_norm": 0.9428019523620605, + "learning_rate": 4.873908057788177e-05, + "loss": 0.1128, + "step": 29140 + }, + { + "epoch": 1.0593066356566612, + "grad_norm": 0.4694746136665344, + "learning_rate": 4.8737452822584724e-05, + "loss": 0.142, + "step": 29150 + }, + { + "epoch": 1.0596700341594594, + "grad_norm": 1.3985977172851562, + "learning_rate": 4.873582404452529e-05, + "loss": 0.1169, + "step": 29160 + }, + { + "epoch": 1.0600334326622574, + "grad_norm": 0.8285462856292725, + "learning_rate": 4.873419424377366e-05, + "loss": 0.1945, + "step": 29170 + }, + { + "epoch": 1.0603968311650556, + "grad_norm": 1.657012939453125, + "learning_rate": 4.8732563420400037e-05, + "loss": 0.1904, + "step": 29180 + }, + { + "epoch": 1.0607602296678538, + "grad_norm": 1.4633735418319702, + "learning_rate": 4.87309315744747e-05, + "loss": 0.1183, + "step": 29190 + }, + { + "epoch": 1.061123628170652, + "grad_norm": 0.9722393155097961, + "learning_rate": 4.872929870606796e-05, + "loss": 0.1608, + "step": 29200 + }, + { + "epoch": 1.0614870266734502, + "grad_norm": 0.6080673933029175, + "learning_rate": 4.872766481525016e-05, + "loss": 0.1198, + "step": 29210 + }, + { + "epoch": 1.0618504251762482, + "grad_norm": 0.7502457499504089, + "learning_rate": 4.8726029902091715e-05, + "loss": 0.1893, + "step": 29220 + }, + { + "epoch": 1.0622138236790464, + "grad_norm": 1.7775638103485107, + "learning_rate": 4.8724393966663054e-05, + "loss": 1.4777, + "step": 29230 + }, + { + "epoch": 1.0625772221818446, + "grad_norm": 1.1095236539840698, + "learning_rate": 4.8722757009034666e-05, + "loss": 0.1443, + "step": 29240 + }, + { + "epoch": 1.0629406206846428, + "grad_norm": 0.6879424452781677, + "learning_rate": 4.872111902927709e-05, + "loss": 0.2048, + "step": 29250 + }, + { + "epoch": 1.063304019187441, + "grad_norm": 1.2532442808151245, + "learning_rate": 4.8719480027460895e-05, + "loss": 0.1343, + "step": 29260 + }, + { + "epoch": 1.0636674176902392, + "grad_norm": 1.0296350717544556, + "learning_rate": 4.87178400036567e-05, + "loss": 0.1656, + "step": 29270 + }, + { + "epoch": 1.0640308161930372, + "grad_norm": 1.0346356630325317, + "learning_rate": 4.871619895793517e-05, + "loss": 0.1466, + "step": 29280 + }, + { + "epoch": 1.0643942146958354, + "grad_norm": 1.9428579807281494, + "learning_rate": 4.8714556890367e-05, + "loss": 0.1496, + "step": 29290 + }, + { + "epoch": 1.0647576131986336, + "grad_norm": 2.6400890350341797, + "learning_rate": 4.871291380102295e-05, + "loss": 0.1857, + "step": 29300 + }, + { + "epoch": 1.0651210117014318, + "grad_norm": 7.17543888092041, + "learning_rate": 4.8711269689973826e-05, + "loss": 0.1242, + "step": 29310 + }, + { + "epoch": 1.06548441020423, + "grad_norm": 1.8619358539581299, + "learning_rate": 4.870962455729045e-05, + "loss": 0.2137, + "step": 29320 + }, + { + "epoch": 1.0658478087070282, + "grad_norm": 4.936455726623535, + "learning_rate": 4.8707978403043716e-05, + "loss": 0.1551, + "step": 29330 + }, + { + "epoch": 1.0662112072098262, + "grad_norm": 1.2196155786514282, + "learning_rate": 4.8706331227304533e-05, + "loss": 0.184, + "step": 29340 + }, + { + "epoch": 1.0665746057126244, + "grad_norm": 2.0982654094696045, + "learning_rate": 4.87046830301439e-05, + "loss": 0.1566, + "step": 29350 + }, + { + "epoch": 1.0669380042154226, + "grad_norm": 1.0265774726867676, + "learning_rate": 4.8703033811632806e-05, + "loss": 0.1263, + "step": 29360 + }, + { + "epoch": 1.0673014027182208, + "grad_norm": 2.413862705230713, + "learning_rate": 4.870138357184233e-05, + "loss": 0.135, + "step": 29370 + }, + { + "epoch": 1.067664801221019, + "grad_norm": 0.800736665725708, + "learning_rate": 4.869973231084356e-05, + "loss": 0.1169, + "step": 29380 + }, + { + "epoch": 1.0680281997238172, + "grad_norm": 4.07125186920166, + "learning_rate": 4.8698080028707647e-05, + "loss": 0.1745, + "step": 29390 + }, + { + "epoch": 1.0683915982266152, + "grad_norm": 1.1997871398925781, + "learning_rate": 4.8696426725505784e-05, + "loss": 0.1427, + "step": 29400 + }, + { + "epoch": 1.0683915982266152, + "eval_loss": 0.3479246199131012, + "eval_runtime": 180.3605, + "eval_samples_per_second": 41.107, + "eval_steps_per_second": 5.14, + "eval_wer": 0.1802693920525714, + "step": 29400 + }, + { + "epoch": 1.0687549967294134, + "grad_norm": 1.5684832334518433, + "learning_rate": 4.8694772401309205e-05, + "loss": 0.1491, + "step": 29410 + }, + { + "epoch": 1.0691183952322116, + "grad_norm": 1.3784462213516235, + "learning_rate": 4.8693117056189194e-05, + "loss": 0.1741, + "step": 29420 + }, + { + "epoch": 1.0694817937350098, + "grad_norm": 2.0766236782073975, + "learning_rate": 4.869146069021707e-05, + "loss": 0.1375, + "step": 29430 + }, + { + "epoch": 1.069845192237808, + "grad_norm": 0.6553940773010254, + "learning_rate": 4.86898033034642e-05, + "loss": 0.1371, + "step": 29440 + }, + { + "epoch": 1.0702085907406063, + "grad_norm": 0.9652252197265625, + "learning_rate": 4.868814489600199e-05, + "loss": 0.1446, + "step": 29450 + }, + { + "epoch": 1.0705719892434042, + "grad_norm": 1.123075008392334, + "learning_rate": 4.8686485467901896e-05, + "loss": 0.1628, + "step": 29460 + }, + { + "epoch": 1.0709353877462024, + "grad_norm": 1.3370702266693115, + "learning_rate": 4.868482501923543e-05, + "loss": 0.1822, + "step": 29470 + }, + { + "epoch": 1.0712987862490007, + "grad_norm": 1.1716543436050415, + "learning_rate": 4.868316355007412e-05, + "loss": 0.7928, + "step": 29480 + }, + { + "epoch": 1.0716621847517989, + "grad_norm": 2.255791187286377, + "learning_rate": 4.868150106048955e-05, + "loss": 0.3897, + "step": 29490 + }, + { + "epoch": 1.072025583254597, + "grad_norm": 2.2386605739593506, + "learning_rate": 4.8679837550553366e-05, + "loss": 0.174, + "step": 29500 + }, + { + "epoch": 1.072388981757395, + "grad_norm": 0.7938382625579834, + "learning_rate": 4.867817302033724e-05, + "loss": 0.1274, + "step": 29510 + }, + { + "epoch": 1.0727523802601933, + "grad_norm": 0.8619611263275146, + "learning_rate": 4.8676507469912866e-05, + "loss": 0.1756, + "step": 29520 + }, + { + "epoch": 1.0731157787629915, + "grad_norm": 5.2337727546691895, + "learning_rate": 4.867484089935205e-05, + "loss": 0.1131, + "step": 29530 + }, + { + "epoch": 1.0734791772657897, + "grad_norm": 1.339237093925476, + "learning_rate": 4.867317330872656e-05, + "loss": 0.2385, + "step": 29540 + }, + { + "epoch": 1.0738425757685879, + "grad_norm": 0.9693569540977478, + "learning_rate": 4.8671504698108266e-05, + "loss": 0.1455, + "step": 29550 + }, + { + "epoch": 1.074205974271386, + "grad_norm": 0.8684889674186707, + "learning_rate": 4.866983506756906e-05, + "loss": 0.1516, + "step": 29560 + }, + { + "epoch": 1.074569372774184, + "grad_norm": 0.5518342852592468, + "learning_rate": 4.866816441718088e-05, + "loss": 0.2068, + "step": 29570 + }, + { + "epoch": 1.0749327712769823, + "grad_norm": 1.049777865409851, + "learning_rate": 4.86664927470157e-05, + "loss": 0.1473, + "step": 29580 + }, + { + "epoch": 1.0752961697797805, + "grad_norm": 3.8799684047698975, + "learning_rate": 4.8664820057145556e-05, + "loss": 0.1398, + "step": 29590 + }, + { + "epoch": 1.0756595682825787, + "grad_norm": 0.9927829504013062, + "learning_rate": 4.866314634764252e-05, + "loss": 2.0386, + "step": 29600 + }, + { + "epoch": 1.076022966785377, + "grad_norm": 1.2022935152053833, + "learning_rate": 4.86614716185787e-05, + "loss": 0.1539, + "step": 29610 + }, + { + "epoch": 1.076386365288175, + "grad_norm": 0.7556710243225098, + "learning_rate": 4.865979587002625e-05, + "loss": 0.1718, + "step": 29620 + }, + { + "epoch": 1.076749763790973, + "grad_norm": 1.0953086614608765, + "learning_rate": 4.865811910205738e-05, + "loss": 0.1537, + "step": 29630 + }, + { + "epoch": 1.0771131622937713, + "grad_norm": 0.49788376688957214, + "learning_rate": 4.865644131474434e-05, + "loss": 0.1436, + "step": 29640 + }, + { + "epoch": 1.0774765607965695, + "grad_norm": 2.5504343509674072, + "learning_rate": 4.865476250815941e-05, + "loss": 0.1535, + "step": 29650 + }, + { + "epoch": 1.0778399592993677, + "grad_norm": 1.280085802078247, + "learning_rate": 4.865308268237492e-05, + "loss": 0.1389, + "step": 29660 + }, + { + "epoch": 1.078203357802166, + "grad_norm": 0.4341859519481659, + "learning_rate": 4.865140183746326e-05, + "loss": 0.2016, + "step": 29670 + }, + { + "epoch": 1.0785667563049641, + "grad_norm": 0.744679868221283, + "learning_rate": 4.864971997349685e-05, + "loss": 0.1216, + "step": 29680 + }, + { + "epoch": 1.078930154807762, + "grad_norm": 1.449559211730957, + "learning_rate": 4.8648037090548154e-05, + "loss": 0.1202, + "step": 29690 + }, + { + "epoch": 1.0792935533105603, + "grad_norm": 3.58284330368042, + "learning_rate": 4.8646353188689674e-05, + "loss": 0.145, + "step": 29700 + }, + { + "epoch": 1.0796569518133585, + "grad_norm": 1.7318589687347412, + "learning_rate": 4.864466826799398e-05, + "loss": 0.1361, + "step": 29710 + }, + { + "epoch": 1.0800203503161567, + "grad_norm": 0.44806694984436035, + "learning_rate": 4.864298232853364e-05, + "loss": 0.212, + "step": 29720 + }, + { + "epoch": 1.080383748818955, + "grad_norm": 0.8236504197120667, + "learning_rate": 4.864129537038132e-05, + "loss": 0.1446, + "step": 29730 + }, + { + "epoch": 1.0807471473217531, + "grad_norm": 0.920353889465332, + "learning_rate": 4.863960739360971e-05, + "loss": 0.1472, + "step": 29740 + }, + { + "epoch": 1.0811105458245511, + "grad_norm": 1.415685772895813, + "learning_rate": 4.8637918398291514e-05, + "loss": 1.9346, + "step": 29750 + }, + { + "epoch": 1.0814739443273493, + "grad_norm": 0.7517853379249573, + "learning_rate": 4.8636228384499524e-05, + "loss": 0.1276, + "step": 29760 + }, + { + "epoch": 1.0818373428301475, + "grad_norm": 0.5632757544517517, + "learning_rate": 4.8634537352306554e-05, + "loss": 0.1385, + "step": 29770 + }, + { + "epoch": 1.0822007413329457, + "grad_norm": 1.5767742395401, + "learning_rate": 4.8632845301785455e-05, + "loss": 0.1451, + "step": 29780 + }, + { + "epoch": 1.082564139835744, + "grad_norm": 0.7501896619796753, + "learning_rate": 4.8631152233009146e-05, + "loss": 0.1337, + "step": 29790 + }, + { + "epoch": 1.082927538338542, + "grad_norm": 0.7235280871391296, + "learning_rate": 4.862945814605056e-05, + "loss": 0.1727, + "step": 29800 + }, + { + "epoch": 1.0832909368413401, + "grad_norm": 0.9608789682388306, + "learning_rate": 4.86277630409827e-05, + "loss": 0.1346, + "step": 29810 + }, + { + "epoch": 1.0836543353441384, + "grad_norm": 0.5176007151603699, + "learning_rate": 4.862606691787859e-05, + "loss": 0.1477, + "step": 29820 + }, + { + "epoch": 1.0840177338469366, + "grad_norm": 1.1901780366897583, + "learning_rate": 4.862436977681133e-05, + "loss": 0.1478, + "step": 29830 + }, + { + "epoch": 1.0843811323497348, + "grad_norm": 4.09995698928833, + "learning_rate": 4.8622671617854026e-05, + "loss": 0.1369, + "step": 29840 + }, + { + "epoch": 1.084744530852533, + "grad_norm": 10.049054145812988, + "learning_rate": 4.8620972441079855e-05, + "loss": 0.8392, + "step": 29850 + }, + { + "epoch": 1.085107929355331, + "grad_norm": 1.131095051765442, + "learning_rate": 4.861927224656202e-05, + "loss": 0.1404, + "step": 29860 + }, + { + "epoch": 1.0854713278581292, + "grad_norm": 1.2740205526351929, + "learning_rate": 4.861757103437379e-05, + "loss": 0.1726, + "step": 29870 + }, + { + "epoch": 1.0858347263609274, + "grad_norm": 0.9203113317489624, + "learning_rate": 4.861586880458845e-05, + "loss": 0.1088, + "step": 29880 + }, + { + "epoch": 1.0861981248637256, + "grad_norm": 0.8646379113197327, + "learning_rate": 4.8614165557279345e-05, + "loss": 0.124, + "step": 29890 + }, + { + "epoch": 1.0865615233665238, + "grad_norm": 1.283758521080017, + "learning_rate": 4.861246129251987e-05, + "loss": 0.1616, + "step": 29900 + }, + { + "epoch": 1.086924921869322, + "grad_norm": 2.0251550674438477, + "learning_rate": 4.861075601038345e-05, + "loss": 0.1475, + "step": 29910 + }, + { + "epoch": 1.08728832037212, + "grad_norm": 0.7173452973365784, + "learning_rate": 4.860904971094356e-05, + "loss": 0.182, + "step": 29920 + }, + { + "epoch": 1.0876517188749182, + "grad_norm": 0.7154909372329712, + "learning_rate": 4.8607342394273725e-05, + "loss": 0.1263, + "step": 29930 + }, + { + "epoch": 1.0880151173777164, + "grad_norm": 2.5288286209106445, + "learning_rate": 4.860563406044749e-05, + "loss": 0.1289, + "step": 29940 + }, + { + "epoch": 1.0883785158805146, + "grad_norm": 1.4772063493728638, + "learning_rate": 4.860392470953848e-05, + "loss": 0.1556, + "step": 29950 + }, + { + "epoch": 1.0887419143833128, + "grad_norm": 1.623298168182373, + "learning_rate": 4.8602214341620346e-05, + "loss": 0.1493, + "step": 29960 + }, + { + "epoch": 1.089105312886111, + "grad_norm": 0.4302707016468048, + "learning_rate": 4.860050295676676e-05, + "loss": 0.2303, + "step": 29970 + }, + { + "epoch": 1.089468711388909, + "grad_norm": 1.4086140394210815, + "learning_rate": 4.8598790555051474e-05, + "loss": 0.1549, + "step": 29980 + }, + { + "epoch": 1.0898321098917072, + "grad_norm": 1.1924636363983154, + "learning_rate": 4.859707713654828e-05, + "loss": 0.1426, + "step": 29990 + }, + { + "epoch": 1.0901955083945054, + "grad_norm": 0.8468578457832336, + "learning_rate": 4.859536270133097e-05, + "loss": 0.1607, + "step": 30000 + }, + { + "epoch": 1.0901955083945054, + "eval_loss": 0.38150739669799805, + "eval_runtime": 180.8598, + "eval_samples_per_second": 40.993, + "eval_steps_per_second": 5.126, + "eval_wer": 0.18088659756385353, + "step": 30000 + }, + { + "epoch": 1.0905589068973036, + "grad_norm": 1.3293052911758423, + "learning_rate": 4.859364724947345e-05, + "loss": 0.1199, + "step": 30010 + }, + { + "epoch": 1.0909223054001018, + "grad_norm": 1.421976923942566, + "learning_rate": 4.859193078104961e-05, + "loss": 0.1932, + "step": 30020 + }, + { + "epoch": 1.0912857039029, + "grad_norm": 5.226151466369629, + "learning_rate": 4.8590213296133415e-05, + "loss": 0.1365, + "step": 30030 + }, + { + "epoch": 1.091649102405698, + "grad_norm": 1.6307711601257324, + "learning_rate": 4.8588494794798866e-05, + "loss": 0.1533, + "step": 30040 + }, + { + "epoch": 1.0920125009084962, + "grad_norm": 1.190746784210205, + "learning_rate": 4.858677527712e-05, + "loss": 0.1701, + "step": 30050 + }, + { + "epoch": 1.0923758994112944, + "grad_norm": 1.1558239459991455, + "learning_rate": 4.858505474317091e-05, + "loss": 0.1792, + "step": 30060 + }, + { + "epoch": 1.0927392979140926, + "grad_norm": 0.5284643769264221, + "learning_rate": 4.858333319302573e-05, + "loss": 0.1789, + "step": 30070 + }, + { + "epoch": 1.0931026964168908, + "grad_norm": 0.7858747243881226, + "learning_rate": 4.858161062675863e-05, + "loss": 0.1456, + "step": 30080 + }, + { + "epoch": 1.0934660949196888, + "grad_norm": 1.2685805559158325, + "learning_rate": 4.857988704444383e-05, + "loss": 0.1326, + "step": 30090 + }, + { + "epoch": 1.093829493422487, + "grad_norm": 0.9551296830177307, + "learning_rate": 4.8578162446155595e-05, + "loss": 0.327, + "step": 30100 + }, + { + "epoch": 1.0941928919252852, + "grad_norm": 3.6769495010375977, + "learning_rate": 4.857643683196823e-05, + "loss": 0.1111, + "step": 30110 + }, + { + "epoch": 1.0945562904280834, + "grad_norm": 0.757580041885376, + "learning_rate": 4.8574710201956095e-05, + "loss": 0.1622, + "step": 30120 + }, + { + "epoch": 1.0949196889308817, + "grad_norm": 0.762323796749115, + "learning_rate": 4.857298255619357e-05, + "loss": 0.1218, + "step": 30130 + }, + { + "epoch": 1.0952830874336799, + "grad_norm": 0.6065217852592468, + "learning_rate": 4.85712538947551e-05, + "loss": 0.1297, + "step": 30140 + }, + { + "epoch": 1.0956464859364778, + "grad_norm": 1.1257789134979248, + "learning_rate": 4.856952421771517e-05, + "loss": 0.1862, + "step": 30150 + }, + { + "epoch": 1.096009884439276, + "grad_norm": 1.128233790397644, + "learning_rate": 4.85677935251483e-05, + "loss": 0.1377, + "step": 30160 + }, + { + "epoch": 1.0963732829420743, + "grad_norm": 0.48844701051712036, + "learning_rate": 4.856606181712906e-05, + "loss": 0.1967, + "step": 30170 + }, + { + "epoch": 1.0967366814448725, + "grad_norm": 0.682921290397644, + "learning_rate": 4.856432909373206e-05, + "loss": 0.1268, + "step": 30180 + }, + { + "epoch": 1.0971000799476707, + "grad_norm": 0.8049948215484619, + "learning_rate": 4.856259535503197e-05, + "loss": 0.0971, + "step": 30190 + }, + { + "epoch": 1.0974634784504689, + "grad_norm": 0.6435711979866028, + "learning_rate": 4.8560860601103485e-05, + "loss": 1.0273, + "step": 30200 + }, + { + "epoch": 1.0978268769532669, + "grad_norm": 1.014172911643982, + "learning_rate": 4.855912483202134e-05, + "loss": 0.1137, + "step": 30210 + }, + { + "epoch": 1.098190275456065, + "grad_norm": 1.4760230779647827, + "learning_rate": 4.8557388047860334e-05, + "loss": 0.1585, + "step": 30220 + }, + { + "epoch": 1.0985536739588633, + "grad_norm": 1.4756141901016235, + "learning_rate": 4.855565024869529e-05, + "loss": 0.1261, + "step": 30230 + }, + { + "epoch": 1.0989170724616615, + "grad_norm": 2.1977133750915527, + "learning_rate": 4.8553911434601085e-05, + "loss": 0.1532, + "step": 30240 + }, + { + "epoch": 1.0992804709644597, + "grad_norm": 1.9084991216659546, + "learning_rate": 4.855217160565265e-05, + "loss": 0.1611, + "step": 30250 + }, + { + "epoch": 1.099643869467258, + "grad_norm": 1.3657923936843872, + "learning_rate": 4.855043076192494e-05, + "loss": 0.1569, + "step": 30260 + }, + { + "epoch": 1.1000072679700559, + "grad_norm": 0.6824470162391663, + "learning_rate": 4.8548688903492943e-05, + "loss": 0.1632, + "step": 30270 + }, + { + "epoch": 1.100370666472854, + "grad_norm": 0.595958948135376, + "learning_rate": 4.854694603043175e-05, + "loss": 0.1394, + "step": 30280 + }, + { + "epoch": 1.1007340649756523, + "grad_norm": 1.1626547574996948, + "learning_rate": 4.854520214281642e-05, + "loss": 0.1193, + "step": 30290 + }, + { + "epoch": 1.1010974634784505, + "grad_norm": 1.2703717947006226, + "learning_rate": 4.8543457240722104e-05, + "loss": 0.1949, + "step": 30300 + }, + { + "epoch": 1.1014608619812487, + "grad_norm": 1.7159488201141357, + "learning_rate": 4.854171132422399e-05, + "loss": 0.1524, + "step": 30310 + }, + { + "epoch": 1.101824260484047, + "grad_norm": 0.7651236057281494, + "learning_rate": 4.85399643933973e-05, + "loss": 0.1992, + "step": 30320 + }, + { + "epoch": 1.102187658986845, + "grad_norm": 0.7985833287239075, + "learning_rate": 4.8538216448317286e-05, + "loss": 0.1363, + "step": 30330 + }, + { + "epoch": 1.102551057489643, + "grad_norm": 1.2583733797073364, + "learning_rate": 4.853646748905928e-05, + "loss": 0.1215, + "step": 30340 + }, + { + "epoch": 1.1029144559924413, + "grad_norm": 1.1982141733169556, + "learning_rate": 4.853471751569864e-05, + "loss": 0.1656, + "step": 30350 + }, + { + "epoch": 1.1032778544952395, + "grad_norm": 2.399423599243164, + "learning_rate": 4.853296652831075e-05, + "loss": 0.1164, + "step": 30360 + }, + { + "epoch": 1.1036412529980377, + "grad_norm": 1.5785446166992188, + "learning_rate": 4.853121452697107e-05, + "loss": 0.1682, + "step": 30370 + }, + { + "epoch": 1.1040046515008357, + "grad_norm": 1.3818514347076416, + "learning_rate": 4.852946151175508e-05, + "loss": 0.5349, + "step": 30380 + }, + { + "epoch": 1.104368050003634, + "grad_norm": 1.6894676685333252, + "learning_rate": 4.8527707482738305e-05, + "loss": 0.1314, + "step": 30390 + }, + { + "epoch": 1.1047314485064321, + "grad_norm": 1.1517245769500732, + "learning_rate": 4.852595243999633e-05, + "loss": 0.1515, + "step": 30400 + }, + { + "epoch": 1.1050948470092303, + "grad_norm": 1.2149289846420288, + "learning_rate": 4.852419638360477e-05, + "loss": 0.2572, + "step": 30410 + }, + { + "epoch": 1.1054582455120285, + "grad_norm": 0.8241190314292908, + "learning_rate": 4.852243931363929e-05, + "loss": 0.2045, + "step": 30420 + }, + { + "epoch": 1.1058216440148267, + "grad_norm": 0.8909230828285217, + "learning_rate": 4.852068123017559e-05, + "loss": 0.1281, + "step": 30430 + }, + { + "epoch": 1.1061850425176247, + "grad_norm": 0.7718971967697144, + "learning_rate": 4.8518922133289424e-05, + "loss": 0.1207, + "step": 30440 + }, + { + "epoch": 1.106548441020423, + "grad_norm": 3.393324136734009, + "learning_rate": 4.8517162023056575e-05, + "loss": 0.4812, + "step": 30450 + }, + { + "epoch": 1.1069118395232211, + "grad_norm": 1.5000587701797485, + "learning_rate": 4.85154008995529e-05, + "loss": 0.1246, + "step": 30460 + }, + { + "epoch": 1.1072752380260193, + "grad_norm": 1.3177014589309692, + "learning_rate": 4.8513638762854264e-05, + "loss": 0.2352, + "step": 30470 + }, + { + "epoch": 1.1076386365288176, + "grad_norm": 0.8771611452102661, + "learning_rate": 4.8511875613036596e-05, + "loss": 1.4369, + "step": 30480 + }, + { + "epoch": 1.1080020350316158, + "grad_norm": 0.46630170941352844, + "learning_rate": 4.8510111450175865e-05, + "loss": 0.1184, + "step": 30490 + }, + { + "epoch": 1.1083654335344137, + "grad_norm": 1.582541823387146, + "learning_rate": 4.850834627434808e-05, + "loss": 0.154, + "step": 30500 + }, + { + "epoch": 1.108728832037212, + "grad_norm": 0.9425756335258484, + "learning_rate": 4.850658008562929e-05, + "loss": 0.1632, + "step": 30510 + }, + { + "epoch": 1.1090922305400102, + "grad_norm": 0.6453799605369568, + "learning_rate": 4.8504812884095616e-05, + "loss": 0.2024, + "step": 30520 + }, + { + "epoch": 1.1094556290428084, + "grad_norm": 0.8643505573272705, + "learning_rate": 4.850304466982317e-05, + "loss": 0.1373, + "step": 30530 + }, + { + "epoch": 1.1098190275456066, + "grad_norm": 0.6463938355445862, + "learning_rate": 4.850127544288816e-05, + "loss": 0.1308, + "step": 30540 + }, + { + "epoch": 1.1101824260484048, + "grad_norm": 1.2465693950653076, + "learning_rate": 4.8499505203366816e-05, + "loss": 0.1628, + "step": 30550 + }, + { + "epoch": 1.1105458245512028, + "grad_norm": 1.085317850112915, + "learning_rate": 4.84977339513354e-05, + "loss": 0.1533, + "step": 30560 + }, + { + "epoch": 1.110909223054001, + "grad_norm": 0.5834909081459045, + "learning_rate": 4.849596168687022e-05, + "loss": 0.1811, + "step": 30570 + }, + { + "epoch": 1.1112726215567992, + "grad_norm": 1.407309889793396, + "learning_rate": 4.849418841004766e-05, + "loss": 0.1621, + "step": 30580 + }, + { + "epoch": 1.1116360200595974, + "grad_norm": 1.0903669595718384, + "learning_rate": 4.8492414120944116e-05, + "loss": 0.271, + "step": 30590 + }, + { + "epoch": 1.1119994185623956, + "grad_norm": 1.6495404243469238, + "learning_rate": 4.8490638819636036e-05, + "loss": 0.1602, + "step": 30600 + }, + { + "epoch": 1.1119994185623956, + "eval_loss": 0.3786245882511139, + "eval_runtime": 180.2563, + "eval_samples_per_second": 41.13, + "eval_steps_per_second": 5.143, + "eval_wer": 0.17622125002269137, + "step": 30600 + } + ], + "logging_steps": 10, + "max_steps": 165108, + "num_input_tokens_seen": 0, + "num_train_epochs": 6, + "save_steps": 1800, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 1.0017977665702101e+20, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +}