| { |
| "best_global_step": 1800, |
| "best_metric": 0.25532727229341157, |
| "best_model_checkpoint": "./SALAMA_NEW99/checkpoint-1800", |
| "epoch": 2.936378466557912, |
| "eval_steps": 600, |
| "global_step": 1800, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.01631321370309951, |
| "grad_norm": 0.8075992465019226, |
| "learning_rate": 1.8e-07, |
| "loss": 0.0109, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.03262642740619902, |
| "grad_norm": 1.4262839555740356, |
| "learning_rate": 3.8e-07, |
| "loss": 0.0094, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.048939641109298535, |
| "grad_norm": 2.3042991161346436, |
| "learning_rate": 5.800000000000001e-07, |
| "loss": 0.0093, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.06525285481239804, |
| "grad_norm": 0.546521782875061, |
| "learning_rate": 7.8e-07, |
| "loss": 0.0218, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.08156606851549755, |
| "grad_norm": 2.1409292221069336, |
| "learning_rate": 9.800000000000001e-07, |
| "loss": 0.0077, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.09787928221859707, |
| "grad_norm": 1.9533841609954834, |
| "learning_rate": 1.1800000000000001e-06, |
| "loss": 0.0061, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.11419249592169657, |
| "grad_norm": 1.0335050821304321, |
| "learning_rate": 1.3800000000000001e-06, |
| "loss": 0.0245, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.13050570962479607, |
| "grad_norm": 1.158021330833435, |
| "learning_rate": 1.5800000000000001e-06, |
| "loss": 0.009, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.1468189233278956, |
| "grad_norm": 2.7232327461242676, |
| "learning_rate": 1.7800000000000001e-06, |
| "loss": 0.0197, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.1631321370309951, |
| "grad_norm": 1.2299067974090576, |
| "learning_rate": 1.98e-06, |
| "loss": 0.007, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.17944535073409462, |
| "grad_norm": 2.173428535461426, |
| "learning_rate": 2.1800000000000003e-06, |
| "loss": 0.0098, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.19575856443719414, |
| "grad_norm": 1.4795992374420166, |
| "learning_rate": 2.38e-06, |
| "loss": 0.0098, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.21207177814029363, |
| "grad_norm": 0.48007962107658386, |
| "learning_rate": 2.5800000000000003e-06, |
| "loss": 0.0258, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.22838499184339314, |
| "grad_norm": 0.5283637642860413, |
| "learning_rate": 2.7800000000000005e-06, |
| "loss": 0.0078, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.24469820554649266, |
| "grad_norm": 1.748868465423584, |
| "learning_rate": 2.9800000000000003e-06, |
| "loss": 0.0083, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.26101141924959215, |
| "grad_norm": 2.2862608432769775, |
| "learning_rate": 3.1800000000000005e-06, |
| "loss": 0.012, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.27732463295269166, |
| "grad_norm": 1.9614684581756592, |
| "learning_rate": 3.3800000000000007e-06, |
| "loss": 0.0093, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.2936378466557912, |
| "grad_norm": 0.2269970178604126, |
| "learning_rate": 3.58e-06, |
| "loss": 0.0079, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.3099510603588907, |
| "grad_norm": 3.722224712371826, |
| "learning_rate": 3.7800000000000002e-06, |
| "loss": 0.0123, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.3262642740619902, |
| "grad_norm": 1.0197768211364746, |
| "learning_rate": 3.980000000000001e-06, |
| "loss": 0.0126, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.3425774877650897, |
| "grad_norm": 2.238046884536743, |
| "learning_rate": 4.18e-06, |
| "loss": 0.0155, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.35889070146818924, |
| "grad_norm": 2.2540082931518555, |
| "learning_rate": 4.38e-06, |
| "loss": 0.0064, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.37520391517128876, |
| "grad_norm": 4.272034168243408, |
| "learning_rate": 4.58e-06, |
| "loss": 0.0144, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.3915171288743883, |
| "grad_norm": 1.297784447669983, |
| "learning_rate": 4.78e-06, |
| "loss": 0.0183, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.4078303425774878, |
| "grad_norm": 1.1175105571746826, |
| "learning_rate": 4.980000000000001e-06, |
| "loss": 0.01, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.42414355628058725, |
| "grad_norm": 0.989632248878479, |
| "learning_rate": 5.18e-06, |
| "loss": 0.0185, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.44045676998368677, |
| "grad_norm": 2.2630860805511475, |
| "learning_rate": 5.380000000000001e-06, |
| "loss": 0.004, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.4567699836867863, |
| "grad_norm": 3.4459376335144043, |
| "learning_rate": 5.580000000000001e-06, |
| "loss": 0.0338, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.4730831973898858, |
| "grad_norm": 2.0962679386138916, |
| "learning_rate": 5.78e-06, |
| "loss": 0.0078, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.4893964110929853, |
| "grad_norm": 1.109755277633667, |
| "learning_rate": 5.98e-06, |
| "loss": 0.0133, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.5057096247960848, |
| "grad_norm": 1.8200187683105469, |
| "learning_rate": 6.18e-06, |
| "loss": 0.0128, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.5220228384991843, |
| "grad_norm": 0.8147674798965454, |
| "learning_rate": 6.380000000000001e-06, |
| "loss": 0.0139, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.5383360522022839, |
| "grad_norm": 2.6240575313568115, |
| "learning_rate": 6.5800000000000005e-06, |
| "loss": 0.0153, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.5546492659053833, |
| "grad_norm": 0.49103260040283203, |
| "learning_rate": 6.780000000000001e-06, |
| "loss": 0.0119, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.5709624796084829, |
| "grad_norm": 1.4433389902114868, |
| "learning_rate": 6.98e-06, |
| "loss": 0.0169, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.5872756933115824, |
| "grad_norm": 1.6671804189682007, |
| "learning_rate": 7.180000000000001e-06, |
| "loss": 0.0103, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.6035889070146819, |
| "grad_norm": 1.7344307899475098, |
| "learning_rate": 7.3800000000000005e-06, |
| "loss": 0.0234, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.6199021207177814, |
| "grad_norm": 1.4996588230133057, |
| "learning_rate": 7.58e-06, |
| "loss": 0.0138, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.636215334420881, |
| "grad_norm": 1.5783976316452026, |
| "learning_rate": 7.78e-06, |
| "loss": 0.0219, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.6525285481239804, |
| "grad_norm": 0.5455562472343445, |
| "learning_rate": 7.980000000000002e-06, |
| "loss": 0.0113, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.6688417618270799, |
| "grad_norm": 1.6277507543563843, |
| "learning_rate": 8.18e-06, |
| "loss": 0.018, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.6851549755301795, |
| "grad_norm": 2.473850965499878, |
| "learning_rate": 8.380000000000001e-06, |
| "loss": 0.0224, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.7014681892332789, |
| "grad_norm": 1.4439480304718018, |
| "learning_rate": 8.580000000000001e-06, |
| "loss": 0.0156, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.7177814029363785, |
| "grad_norm": 0.7241881489753723, |
| "learning_rate": 8.78e-06, |
| "loss": 0.0131, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.734094616639478, |
| "grad_norm": 2.1885428428649902, |
| "learning_rate": 8.98e-06, |
| "loss": 0.019, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.7504078303425775, |
| "grad_norm": 2.702345371246338, |
| "learning_rate": 9.180000000000002e-06, |
| "loss": 0.0253, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.766721044045677, |
| "grad_norm": 3.975011110305786, |
| "learning_rate": 9.38e-06, |
| "loss": 0.0107, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.7830342577487766, |
| "grad_norm": 2.3299198150634766, |
| "learning_rate": 9.58e-06, |
| "loss": 0.0212, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.799347471451876, |
| "grad_norm": 1.317580223083496, |
| "learning_rate": 9.780000000000001e-06, |
| "loss": 0.0227, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.8156606851549756, |
| "grad_norm": 2.4023892879486084, |
| "learning_rate": 9.980000000000001e-06, |
| "loss": 0.0365, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.831973898858075, |
| "grad_norm": 2.641399383544922, |
| "learning_rate": 9.932785660941001e-06, |
| "loss": 0.0138, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.8482871125611745, |
| "grad_norm": 2.7177205085754395, |
| "learning_rate": 9.858103061986558e-06, |
| "loss": 0.0145, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.8646003262642741, |
| "grad_norm": 1.9215168952941895, |
| "learning_rate": 9.783420463032114e-06, |
| "loss": 0.0305, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.8809135399673735, |
| "grad_norm": 2.8683998584747314, |
| "learning_rate": 9.708737864077671e-06, |
| "loss": 0.0183, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.8972267536704731, |
| "grad_norm": 1.9078762531280518, |
| "learning_rate": 9.634055265123227e-06, |
| "loss": 0.0235, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.9135399673735726, |
| "grad_norm": 3.7367427349090576, |
| "learning_rate": 9.559372666168784e-06, |
| "loss": 0.0175, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.9298531810766721, |
| "grad_norm": 1.7846318483352661, |
| "learning_rate": 9.48469006721434e-06, |
| "loss": 0.0138, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.9461663947797716, |
| "grad_norm": 3.428025245666504, |
| "learning_rate": 9.410007468259897e-06, |
| "loss": 0.0168, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.9624796084828712, |
| "grad_norm": 2.601677417755127, |
| "learning_rate": 9.335324869305452e-06, |
| "loss": 0.0192, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.9787928221859706, |
| "grad_norm": 3.509207248687744, |
| "learning_rate": 9.26064227035101e-06, |
| "loss": 0.0177, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.9787928221859706, |
| "eval_loss": 0.011165497824549675, |
| "eval_runtime": 1766.1438, |
| "eval_samples_per_second": 2.775, |
| "eval_steps_per_second": 0.347, |
| "eval_wer": 1.0189228529839884, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.9951060358890701, |
| "grad_norm": 2.4465436935424805, |
| "learning_rate": 9.185959671396565e-06, |
| "loss": 0.0168, |
| "step": 610 |
| }, |
| { |
| "epoch": 1.0114192495921697, |
| "grad_norm": 3.492405414581299, |
| "learning_rate": 9.111277072442123e-06, |
| "loss": 0.0096, |
| "step": 620 |
| }, |
| { |
| "epoch": 1.0277324632952691, |
| "grad_norm": 0.9142910242080688, |
| "learning_rate": 9.036594473487678e-06, |
| "loss": 0.0106, |
| "step": 630 |
| }, |
| { |
| "epoch": 1.0440456769983686, |
| "grad_norm": 0.7352218627929688, |
| "learning_rate": 8.961911874533234e-06, |
| "loss": 0.0085, |
| "step": 640 |
| }, |
| { |
| "epoch": 1.0603588907014683, |
| "grad_norm": 1.1586534976959229, |
| "learning_rate": 8.887229275578791e-06, |
| "loss": 0.0068, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.0766721044045677, |
| "grad_norm": 2.758241891860962, |
| "learning_rate": 8.812546676624347e-06, |
| "loss": 0.0084, |
| "step": 660 |
| }, |
| { |
| "epoch": 1.0929853181076672, |
| "grad_norm": 0.44431474804878235, |
| "learning_rate": 8.737864077669904e-06, |
| "loss": 0.0087, |
| "step": 670 |
| }, |
| { |
| "epoch": 1.1092985318107667, |
| "grad_norm": 1.2769174575805664, |
| "learning_rate": 8.66318147871546e-06, |
| "loss": 0.0096, |
| "step": 680 |
| }, |
| { |
| "epoch": 1.1256117455138663, |
| "grad_norm": 2.656780242919922, |
| "learning_rate": 8.588498879761017e-06, |
| "loss": 0.0127, |
| "step": 690 |
| }, |
| { |
| "epoch": 1.1419249592169658, |
| "grad_norm": 1.0895354747772217, |
| "learning_rate": 8.513816280806572e-06, |
| "loss": 0.0112, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.1582381729200653, |
| "grad_norm": 1.0903260707855225, |
| "learning_rate": 8.43913368185213e-06, |
| "loss": 0.0091, |
| "step": 710 |
| }, |
| { |
| "epoch": 1.1745513866231647, |
| "grad_norm": 2.6359965801239014, |
| "learning_rate": 8.364451082897685e-06, |
| "loss": 0.0193, |
| "step": 720 |
| }, |
| { |
| "epoch": 1.1908646003262642, |
| "grad_norm": 0.9313237071037292, |
| "learning_rate": 8.289768483943242e-06, |
| "loss": 0.0073, |
| "step": 730 |
| }, |
| { |
| "epoch": 1.2071778140293639, |
| "grad_norm": 0.7475559115409851, |
| "learning_rate": 8.215085884988798e-06, |
| "loss": 0.0075, |
| "step": 740 |
| }, |
| { |
| "epoch": 1.2234910277324633, |
| "grad_norm": 0.24283719062805176, |
| "learning_rate": 8.140403286034355e-06, |
| "loss": 0.0071, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.2398042414355628, |
| "grad_norm": 0.6300885677337646, |
| "learning_rate": 8.06572068707991e-06, |
| "loss": 0.006, |
| "step": 760 |
| }, |
| { |
| "epoch": 1.2561174551386622, |
| "grad_norm": 1.7239911556243896, |
| "learning_rate": 7.991038088125468e-06, |
| "loss": 0.0137, |
| "step": 770 |
| }, |
| { |
| "epoch": 1.272430668841762, |
| "grad_norm": 1.5679051876068115, |
| "learning_rate": 7.916355489171023e-06, |
| "loss": 0.0047, |
| "step": 780 |
| }, |
| { |
| "epoch": 1.2887438825448614, |
| "grad_norm": 1.0264155864715576, |
| "learning_rate": 7.84167289021658e-06, |
| "loss": 0.0074, |
| "step": 790 |
| }, |
| { |
| "epoch": 1.3050570962479608, |
| "grad_norm": 1.0627104043960571, |
| "learning_rate": 7.766990291262136e-06, |
| "loss": 0.0106, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.3213703099510603, |
| "grad_norm": 1.262370228767395, |
| "learning_rate": 7.692307692307694e-06, |
| "loss": 0.0072, |
| "step": 810 |
| }, |
| { |
| "epoch": 1.3376835236541598, |
| "grad_norm": 1.5836150646209717, |
| "learning_rate": 7.617625093353249e-06, |
| "loss": 0.0092, |
| "step": 820 |
| }, |
| { |
| "epoch": 1.3539967373572595, |
| "grad_norm": 1.4232890605926514, |
| "learning_rate": 7.542942494398806e-06, |
| "loss": 0.0089, |
| "step": 830 |
| }, |
| { |
| "epoch": 1.370309951060359, |
| "grad_norm": 0.49621379375457764, |
| "learning_rate": 7.468259895444362e-06, |
| "loss": 0.0066, |
| "step": 840 |
| }, |
| { |
| "epoch": 1.3866231647634584, |
| "grad_norm": 1.060057520866394, |
| "learning_rate": 7.393577296489919e-06, |
| "loss": 0.0065, |
| "step": 850 |
| }, |
| { |
| "epoch": 1.4029363784665578, |
| "grad_norm": 1.5507824420928955, |
| "learning_rate": 7.318894697535475e-06, |
| "loss": 0.0086, |
| "step": 860 |
| }, |
| { |
| "epoch": 1.4192495921696575, |
| "grad_norm": 0.6949241757392883, |
| "learning_rate": 7.244212098581031e-06, |
| "loss": 0.0066, |
| "step": 870 |
| }, |
| { |
| "epoch": 1.435562805872757, |
| "grad_norm": 3.595614194869995, |
| "learning_rate": 7.169529499626588e-06, |
| "loss": 0.0084, |
| "step": 880 |
| }, |
| { |
| "epoch": 1.4518760195758564, |
| "grad_norm": 0.38814422488212585, |
| "learning_rate": 7.094846900672144e-06, |
| "loss": 0.0086, |
| "step": 890 |
| }, |
| { |
| "epoch": 1.468189233278956, |
| "grad_norm": 0.612360417842865, |
| "learning_rate": 7.0201643017177005e-06, |
| "loss": 0.0087, |
| "step": 900 |
| }, |
| { |
| "epoch": 1.4845024469820554, |
| "grad_norm": 1.359798789024353, |
| "learning_rate": 6.945481702763257e-06, |
| "loss": 0.0052, |
| "step": 910 |
| }, |
| { |
| "epoch": 1.5008156606851548, |
| "grad_norm": 2.678591251373291, |
| "learning_rate": 6.870799103808813e-06, |
| "loss": 0.0079, |
| "step": 920 |
| }, |
| { |
| "epoch": 1.5171288743882545, |
| "grad_norm": 1.1047519445419312, |
| "learning_rate": 6.79611650485437e-06, |
| "loss": 0.0047, |
| "step": 930 |
| }, |
| { |
| "epoch": 1.533442088091354, |
| "grad_norm": 1.798194169998169, |
| "learning_rate": 6.721433905899925e-06, |
| "loss": 0.0125, |
| "step": 940 |
| }, |
| { |
| "epoch": 1.5497553017944536, |
| "grad_norm": 0.44330868124961853, |
| "learning_rate": 6.6467513069454825e-06, |
| "loss": 0.0074, |
| "step": 950 |
| }, |
| { |
| "epoch": 1.566068515497553, |
| "grad_norm": 1.6538296937942505, |
| "learning_rate": 6.572068707991038e-06, |
| "loss": 0.0118, |
| "step": 960 |
| }, |
| { |
| "epoch": 1.5823817292006526, |
| "grad_norm": 1.2385334968566895, |
| "learning_rate": 6.497386109036595e-06, |
| "loss": 0.0091, |
| "step": 970 |
| }, |
| { |
| "epoch": 1.598694942903752, |
| "grad_norm": 1.1827936172485352, |
| "learning_rate": 6.422703510082151e-06, |
| "loss": 0.008, |
| "step": 980 |
| }, |
| { |
| "epoch": 1.6150081566068515, |
| "grad_norm": 1.1670842170715332, |
| "learning_rate": 6.348020911127708e-06, |
| "loss": 0.004, |
| "step": 990 |
| }, |
| { |
| "epoch": 1.631321370309951, |
| "grad_norm": 1.9073982238769531, |
| "learning_rate": 6.273338312173264e-06, |
| "loss": 0.0105, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.6476345840130504, |
| "grad_norm": 0.6385743021965027, |
| "learning_rate": 6.198655713218821e-06, |
| "loss": 0.0047, |
| "step": 1010 |
| }, |
| { |
| "epoch": 1.66394779771615, |
| "grad_norm": 0.717432975769043, |
| "learning_rate": 6.123973114264377e-06, |
| "loss": 0.0104, |
| "step": 1020 |
| }, |
| { |
| "epoch": 1.6802610114192496, |
| "grad_norm": 1.6229373216629028, |
| "learning_rate": 6.049290515309934e-06, |
| "loss": 0.014, |
| "step": 1030 |
| }, |
| { |
| "epoch": 1.6965742251223492, |
| "grad_norm": 1.5651880502700806, |
| "learning_rate": 5.974607916355489e-06, |
| "loss": 0.0081, |
| "step": 1040 |
| }, |
| { |
| "epoch": 1.7128874388254487, |
| "grad_norm": 1.444212794303894, |
| "learning_rate": 5.899925317401046e-06, |
| "loss": 0.0099, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.7292006525285482, |
| "grad_norm": 1.1551159620285034, |
| "learning_rate": 5.825242718446602e-06, |
| "loss": 0.0087, |
| "step": 1060 |
| }, |
| { |
| "epoch": 1.7455138662316476, |
| "grad_norm": 0.7630689740180969, |
| "learning_rate": 5.750560119492159e-06, |
| "loss": 0.0056, |
| "step": 1070 |
| }, |
| { |
| "epoch": 1.761827079934747, |
| "grad_norm": 0.6756051778793335, |
| "learning_rate": 5.675877520537715e-06, |
| "loss": 0.0057, |
| "step": 1080 |
| }, |
| { |
| "epoch": 1.7781402936378465, |
| "grad_norm": 0.47443750500679016, |
| "learning_rate": 5.6011949215832715e-06, |
| "loss": 0.0068, |
| "step": 1090 |
| }, |
| { |
| "epoch": 1.7944535073409462, |
| "grad_norm": 0.8193678855895996, |
| "learning_rate": 5.526512322628828e-06, |
| "loss": 0.0079, |
| "step": 1100 |
| }, |
| { |
| "epoch": 1.8107667210440457, |
| "grad_norm": 0.7489306330680847, |
| "learning_rate": 5.451829723674384e-06, |
| "loss": 0.0072, |
| "step": 1110 |
| }, |
| { |
| "epoch": 1.8270799347471451, |
| "grad_norm": 0.9458585977554321, |
| "learning_rate": 5.37714712471994e-06, |
| "loss": 0.0047, |
| "step": 1120 |
| }, |
| { |
| "epoch": 1.8433931484502448, |
| "grad_norm": 0.8624395132064819, |
| "learning_rate": 5.302464525765497e-06, |
| "loss": 0.0053, |
| "step": 1130 |
| }, |
| { |
| "epoch": 1.8597063621533443, |
| "grad_norm": 0.5177104473114014, |
| "learning_rate": 5.227781926811053e-06, |
| "loss": 0.0074, |
| "step": 1140 |
| }, |
| { |
| "epoch": 1.8760195758564437, |
| "grad_norm": 1.2944625616073608, |
| "learning_rate": 5.15309932785661e-06, |
| "loss": 0.0151, |
| "step": 1150 |
| }, |
| { |
| "epoch": 1.8923327895595432, |
| "grad_norm": 1.9204165935516357, |
| "learning_rate": 5.0784167289021655e-06, |
| "loss": 0.0082, |
| "step": 1160 |
| }, |
| { |
| "epoch": 1.9086460032626427, |
| "grad_norm": 2.2888307571411133, |
| "learning_rate": 5.003734129947723e-06, |
| "loss": 0.0057, |
| "step": 1170 |
| }, |
| { |
| "epoch": 1.9249592169657421, |
| "grad_norm": 1.0358986854553223, |
| "learning_rate": 4.929051530993279e-06, |
| "loss": 0.0087, |
| "step": 1180 |
| }, |
| { |
| "epoch": 1.9412724306688418, |
| "grad_norm": 0.8882740139961243, |
| "learning_rate": 4.854368932038836e-06, |
| "loss": 0.0111, |
| "step": 1190 |
| }, |
| { |
| "epoch": 1.9575856443719413, |
| "grad_norm": 1.2315634489059448, |
| "learning_rate": 4.779686333084392e-06, |
| "loss": 0.0055, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.9575856443719413, |
| "eval_loss": 0.005837247706949711, |
| "eval_runtime": 1757.0337, |
| "eval_samples_per_second": 2.789, |
| "eval_steps_per_second": 0.349, |
| "eval_wer": 0.4533848760350299, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.9738988580750407, |
| "grad_norm": 1.227731466293335, |
| "learning_rate": 4.7050037341299485e-06, |
| "loss": 0.0039, |
| "step": 1210 |
| }, |
| { |
| "epoch": 1.9902120717781404, |
| "grad_norm": 2.077256917953491, |
| "learning_rate": 4.630321135175505e-06, |
| "loss": 0.0054, |
| "step": 1220 |
| }, |
| { |
| "epoch": 2.00652528548124, |
| "grad_norm": 0.7639275789260864, |
| "learning_rate": 4.555638536221061e-06, |
| "loss": 0.0055, |
| "step": 1230 |
| }, |
| { |
| "epoch": 2.0228384991843393, |
| "grad_norm": 0.3696356415748596, |
| "learning_rate": 4.480955937266617e-06, |
| "loss": 0.0044, |
| "step": 1240 |
| }, |
| { |
| "epoch": 2.039151712887439, |
| "grad_norm": 0.4084968566894531, |
| "learning_rate": 4.406273338312173e-06, |
| "loss": 0.0037, |
| "step": 1250 |
| }, |
| { |
| "epoch": 2.0554649265905383, |
| "grad_norm": 0.19460849463939667, |
| "learning_rate": 4.33159073935773e-06, |
| "loss": 0.0023, |
| "step": 1260 |
| }, |
| { |
| "epoch": 2.0717781402936377, |
| "grad_norm": 0.293778657913208, |
| "learning_rate": 4.256908140403286e-06, |
| "loss": 0.0034, |
| "step": 1270 |
| }, |
| { |
| "epoch": 2.088091353996737, |
| "grad_norm": 0.0963997095823288, |
| "learning_rate": 4.1822255414488425e-06, |
| "loss": 0.0011, |
| "step": 1280 |
| }, |
| { |
| "epoch": 2.104404567699837, |
| "grad_norm": 0.1960829347372055, |
| "learning_rate": 4.107542942494399e-06, |
| "loss": 0.002, |
| "step": 1290 |
| }, |
| { |
| "epoch": 2.1207177814029365, |
| "grad_norm": 0.22539754211902618, |
| "learning_rate": 4.032860343539955e-06, |
| "loss": 0.0033, |
| "step": 1300 |
| }, |
| { |
| "epoch": 2.137030995106036, |
| "grad_norm": 0.43771710991859436, |
| "learning_rate": 3.958177744585512e-06, |
| "loss": 0.0023, |
| "step": 1310 |
| }, |
| { |
| "epoch": 2.1533442088091355, |
| "grad_norm": 0.39533042907714844, |
| "learning_rate": 3.883495145631068e-06, |
| "loss": 0.0024, |
| "step": 1320 |
| }, |
| { |
| "epoch": 2.169657422512235, |
| "grad_norm": 0.1653490662574768, |
| "learning_rate": 3.8088125466766246e-06, |
| "loss": 0.0022, |
| "step": 1330 |
| }, |
| { |
| "epoch": 2.1859706362153344, |
| "grad_norm": 0.07081114500761032, |
| "learning_rate": 3.734129947722181e-06, |
| "loss": 0.0019, |
| "step": 1340 |
| }, |
| { |
| "epoch": 2.202283849918434, |
| "grad_norm": 0.4014896750450134, |
| "learning_rate": 3.6594473487677374e-06, |
| "loss": 0.0051, |
| "step": 1350 |
| }, |
| { |
| "epoch": 2.2185970636215333, |
| "grad_norm": 0.23356999456882477, |
| "learning_rate": 3.584764749813294e-06, |
| "loss": 0.003, |
| "step": 1360 |
| }, |
| { |
| "epoch": 2.2349102773246328, |
| "grad_norm": 0.10898713022470474, |
| "learning_rate": 3.5100821508588502e-06, |
| "loss": 0.0024, |
| "step": 1370 |
| }, |
| { |
| "epoch": 2.2512234910277327, |
| "grad_norm": 0.08161570131778717, |
| "learning_rate": 3.4353995519044066e-06, |
| "loss": 0.005, |
| "step": 1380 |
| }, |
| { |
| "epoch": 2.267536704730832, |
| "grad_norm": 0.5079241991043091, |
| "learning_rate": 3.3607169529499626e-06, |
| "loss": 0.0047, |
| "step": 1390 |
| }, |
| { |
| "epoch": 2.2838499184339316, |
| "grad_norm": 0.9440802335739136, |
| "learning_rate": 3.286034353995519e-06, |
| "loss": 0.005, |
| "step": 1400 |
| }, |
| { |
| "epoch": 2.300163132137031, |
| "grad_norm": 0.3201453685760498, |
| "learning_rate": 3.2113517550410755e-06, |
| "loss": 0.0025, |
| "step": 1410 |
| }, |
| { |
| "epoch": 2.3164763458401305, |
| "grad_norm": 0.4434724748134613, |
| "learning_rate": 3.136669156086632e-06, |
| "loss": 0.0046, |
| "step": 1420 |
| }, |
| { |
| "epoch": 2.33278955954323, |
| "grad_norm": 0.30961957573890686, |
| "learning_rate": 3.0619865571321883e-06, |
| "loss": 0.0017, |
| "step": 1430 |
| }, |
| { |
| "epoch": 2.3491027732463294, |
| "grad_norm": 0.16197660565376282, |
| "learning_rate": 2.9873039581777447e-06, |
| "loss": 0.0028, |
| "step": 1440 |
| }, |
| { |
| "epoch": 2.365415986949429, |
| "grad_norm": 0.6526560187339783, |
| "learning_rate": 2.912621359223301e-06, |
| "loss": 0.0048, |
| "step": 1450 |
| }, |
| { |
| "epoch": 2.3817292006525284, |
| "grad_norm": 0.07773681730031967, |
| "learning_rate": 2.8379387602688575e-06, |
| "loss": 0.0016, |
| "step": 1460 |
| }, |
| { |
| "epoch": 2.3980424143556283, |
| "grad_norm": 0.9661878943443298, |
| "learning_rate": 2.763256161314414e-06, |
| "loss": 0.0061, |
| "step": 1470 |
| }, |
| { |
| "epoch": 2.4143556280587277, |
| "grad_norm": 1.0361403226852417, |
| "learning_rate": 2.68857356235997e-06, |
| "loss": 0.0079, |
| "step": 1480 |
| }, |
| { |
| "epoch": 2.430668841761827, |
| "grad_norm": 0.09448342025279999, |
| "learning_rate": 2.6138909634055264e-06, |
| "loss": 0.0019, |
| "step": 1490 |
| }, |
| { |
| "epoch": 2.4469820554649266, |
| "grad_norm": 0.06374044716358185, |
| "learning_rate": 2.5392083644510828e-06, |
| "loss": 0.0016, |
| "step": 1500 |
| }, |
| { |
| "epoch": 2.463295269168026, |
| "grad_norm": 0.5809449553489685, |
| "learning_rate": 2.4645257654966396e-06, |
| "loss": 0.0023, |
| "step": 1510 |
| }, |
| { |
| "epoch": 2.4796084828711256, |
| "grad_norm": 0.1248660460114479, |
| "learning_rate": 2.389843166542196e-06, |
| "loss": 0.0022, |
| "step": 1520 |
| }, |
| { |
| "epoch": 2.495921696574225, |
| "grad_norm": 0.06973249465227127, |
| "learning_rate": 2.3151605675877524e-06, |
| "loss": 0.003, |
| "step": 1530 |
| }, |
| { |
| "epoch": 2.5122349102773245, |
| "grad_norm": 4.4841694831848145, |
| "learning_rate": 2.2404779686333084e-06, |
| "loss": 0.0028, |
| "step": 1540 |
| }, |
| { |
| "epoch": 2.528548123980424, |
| "grad_norm": 0.12578125298023224, |
| "learning_rate": 2.165795369678865e-06, |
| "loss": 0.0018, |
| "step": 1550 |
| }, |
| { |
| "epoch": 2.544861337683524, |
| "grad_norm": 0.4544123113155365, |
| "learning_rate": 2.0911127707244213e-06, |
| "loss": 0.0019, |
| "step": 1560 |
| }, |
| { |
| "epoch": 2.5611745513866233, |
| "grad_norm": 0.19402466714382172, |
| "learning_rate": 2.0164301717699777e-06, |
| "loss": 0.0022, |
| "step": 1570 |
| }, |
| { |
| "epoch": 2.5774877650897228, |
| "grad_norm": 0.790777325630188, |
| "learning_rate": 1.941747572815534e-06, |
| "loss": 0.0014, |
| "step": 1580 |
| }, |
| { |
| "epoch": 2.5938009787928222, |
| "grad_norm": 0.07585720717906952, |
| "learning_rate": 1.8670649738610905e-06, |
| "loss": 0.0016, |
| "step": 1590 |
| }, |
| { |
| "epoch": 2.6101141924959217, |
| "grad_norm": 1.340345025062561, |
| "learning_rate": 1.792382374906647e-06, |
| "loss": 0.0037, |
| "step": 1600 |
| }, |
| { |
| "epoch": 2.626427406199021, |
| "grad_norm": 1.7285263538360596, |
| "learning_rate": 1.7176997759522033e-06, |
| "loss": 0.0033, |
| "step": 1610 |
| }, |
| { |
| "epoch": 2.6427406199021206, |
| "grad_norm": 0.09814145416021347, |
| "learning_rate": 1.6430171769977595e-06, |
| "loss": 0.0041, |
| "step": 1620 |
| }, |
| { |
| "epoch": 2.65905383360522, |
| "grad_norm": 0.13084454834461212, |
| "learning_rate": 1.568334578043316e-06, |
| "loss": 0.0019, |
| "step": 1630 |
| }, |
| { |
| "epoch": 2.6753670473083195, |
| "grad_norm": 0.07720234245061874, |
| "learning_rate": 1.4936519790888724e-06, |
| "loss": 0.0032, |
| "step": 1640 |
| }, |
| { |
| "epoch": 2.6916802610114194, |
| "grad_norm": 0.8523921966552734, |
| "learning_rate": 1.4189693801344288e-06, |
| "loss": 0.0018, |
| "step": 1650 |
| }, |
| { |
| "epoch": 2.707993474714519, |
| "grad_norm": 0.09405750036239624, |
| "learning_rate": 1.344286781179985e-06, |
| "loss": 0.0021, |
| "step": 1660 |
| }, |
| { |
| "epoch": 2.7243066884176184, |
| "grad_norm": 0.11995333433151245, |
| "learning_rate": 1.2696041822255414e-06, |
| "loss": 0.0032, |
| "step": 1670 |
| }, |
| { |
| "epoch": 2.740619902120718, |
| "grad_norm": 0.6443132162094116, |
| "learning_rate": 1.194921583271098e-06, |
| "loss": 0.002, |
| "step": 1680 |
| }, |
| { |
| "epoch": 2.7569331158238173, |
| "grad_norm": 0.29005166888237, |
| "learning_rate": 1.1202389843166542e-06, |
| "loss": 0.0035, |
| "step": 1690 |
| }, |
| { |
| "epoch": 2.7732463295269167, |
| "grad_norm": 0.6452990770339966, |
| "learning_rate": 1.0455563853622106e-06, |
| "loss": 0.0018, |
| "step": 1700 |
| }, |
| { |
| "epoch": 2.789559543230016, |
| "grad_norm": 0.10402486473321915, |
| "learning_rate": 9.70873786407767e-07, |
| "loss": 0.0041, |
| "step": 1710 |
| }, |
| { |
| "epoch": 2.8058727569331157, |
| "grad_norm": 0.11018693447113037, |
| "learning_rate": 8.961911874533235e-07, |
| "loss": 0.0016, |
| "step": 1720 |
| }, |
| { |
| "epoch": 2.822185970636215, |
| "grad_norm": 0.1631583273410797, |
| "learning_rate": 8.215085884988798e-07, |
| "loss": 0.0018, |
| "step": 1730 |
| }, |
| { |
| "epoch": 2.838499184339315, |
| "grad_norm": 1.892919898033142, |
| "learning_rate": 7.468259895444362e-07, |
| "loss": 0.0097, |
| "step": 1740 |
| }, |
| { |
| "epoch": 2.8548123980424145, |
| "grad_norm": 1.0992926359176636, |
| "learning_rate": 6.721433905899925e-07, |
| "loss": 0.003, |
| "step": 1750 |
| }, |
| { |
| "epoch": 2.871125611745514, |
| "grad_norm": 0.08139196038246155, |
| "learning_rate": 5.97460791635549e-07, |
| "loss": 0.0017, |
| "step": 1760 |
| }, |
| { |
| "epoch": 2.8874388254486134, |
| "grad_norm": 2.4018540382385254, |
| "learning_rate": 5.227781926811053e-07, |
| "loss": 0.002, |
| "step": 1770 |
| }, |
| { |
| "epoch": 2.903752039151713, |
| "grad_norm": 1.0262824296951294, |
| "learning_rate": 4.4809559372666173e-07, |
| "loss": 0.0024, |
| "step": 1780 |
| }, |
| { |
| "epoch": 2.9200652528548123, |
| "grad_norm": 1.249171257019043, |
| "learning_rate": 3.734129947722181e-07, |
| "loss": 0.0029, |
| "step": 1790 |
| }, |
| { |
| "epoch": 2.936378466557912, |
| "grad_norm": 0.15619555115699768, |
| "learning_rate": 2.987303958177745e-07, |
| "loss": 0.0019, |
| "step": 1800 |
| }, |
| { |
| "epoch": 2.936378466557912, |
| "eval_loss": 0.003976329229772091, |
| "eval_runtime": 1760.2683, |
| "eval_samples_per_second": 2.784, |
| "eval_steps_per_second": 0.348, |
| "eval_wer": 0.25532727229341157, |
| "step": 1800 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 1839, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 600, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.661501583802368e+19, |
| "train_batch_size": 16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
