{ "best_global_step": 4000, "best_metric": 13.268986926675701, "best_model_checkpoint": "./SALAMA_NEWMED1/checkpoint-4000", "epoch": 2.3962848297213624, "eval_steps": 2000, "global_step": 6000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00399480675122341, "grad_norm": 29.325483322143555, "learning_rate": 1.8e-07, "loss": 3.9137, "step": 10 }, { "epoch": 0.00798961350244682, "grad_norm": 53.530357360839844, "learning_rate": 3.8e-07, "loss": 3.8712, "step": 20 }, { "epoch": 0.011984420253670228, "grad_norm": 36.358612060546875, "learning_rate": 5.800000000000001e-07, "loss": 3.0878, "step": 30 }, { "epoch": 0.01597922700489364, "grad_norm": 16.489139556884766, "learning_rate": 7.8e-07, "loss": 2.7678, "step": 40 }, { "epoch": 0.019974033756117046, "grad_norm": 12.16690444946289, "learning_rate": 9.800000000000001e-07, "loss": 2.3664, "step": 50 }, { "epoch": 0.023968840507340456, "grad_norm": 12.258421897888184, "learning_rate": 1.1800000000000001e-06, "loss": 2.1087, "step": 60 }, { "epoch": 0.027963647258563866, "grad_norm": 10.25462818145752, "learning_rate": 1.3800000000000001e-06, "loss": 1.8865, "step": 70 }, { "epoch": 0.03195845400978728, "grad_norm": 10.944440841674805, "learning_rate": 1.5800000000000001e-06, "loss": 1.7902, "step": 80 }, { "epoch": 0.035953260761010686, "grad_norm": 10.586995124816895, "learning_rate": 1.7800000000000001e-06, "loss": 1.6421, "step": 90 }, { "epoch": 0.03994806751223409, "grad_norm": 10.034771919250488, "learning_rate": 1.98e-06, "loss": 1.5449, "step": 100 }, { "epoch": 0.043942874263457506, "grad_norm": 10.443981170654297, "learning_rate": 2.1800000000000003e-06, "loss": 1.4428, "step": 110 }, { "epoch": 0.04793768101468091, "grad_norm": 9.427363395690918, "learning_rate": 2.38e-06, "loss": 1.2572, "step": 120 }, { "epoch": 0.051932487765904325, "grad_norm": 10.59061050415039, "learning_rate": 2.5800000000000003e-06, "loss": 1.072, "step": 130 }, { "epoch": 0.05592729451712773, "grad_norm": 7.301783084869385, "learning_rate": 2.7800000000000005e-06, "loss": 0.7508, "step": 140 }, { "epoch": 0.059922101268351145, "grad_norm": 8.623008728027344, "learning_rate": 2.9800000000000003e-06, "loss": 0.7196, "step": 150 }, { "epoch": 0.06391690801957456, "grad_norm": 8.511634826660156, "learning_rate": 3.1800000000000005e-06, "loss": 0.6404, "step": 160 }, { "epoch": 0.06791171477079797, "grad_norm": 8.052703857421875, "learning_rate": 3.3800000000000007e-06, "loss": 0.6112, "step": 170 }, { "epoch": 0.07190652152202137, "grad_norm": 7.108290672302246, "learning_rate": 3.58e-06, "loss": 0.6507, "step": 180 }, { "epoch": 0.07590132827324478, "grad_norm": 7.996196269989014, "learning_rate": 3.7800000000000002e-06, "loss": 0.6163, "step": 190 }, { "epoch": 0.07989613502446818, "grad_norm": 6.643161773681641, "learning_rate": 3.980000000000001e-06, "loss": 0.5806, "step": 200 }, { "epoch": 0.0838909417756916, "grad_norm": 7.640316009521484, "learning_rate": 4.18e-06, "loss": 0.5503, "step": 210 }, { "epoch": 0.08788574852691501, "grad_norm": 7.485447883605957, "learning_rate": 4.38e-06, "loss": 0.5672, "step": 220 }, { "epoch": 0.09188055527813842, "grad_norm": 7.693063735961914, "learning_rate": 4.58e-06, "loss": 0.5593, "step": 230 }, { "epoch": 0.09587536202936182, "grad_norm": 7.907495498657227, "learning_rate": 4.78e-06, "loss": 0.5122, "step": 240 }, { "epoch": 0.09987016878058524, "grad_norm": 6.673120021820068, "learning_rate": 4.980000000000001e-06, "loss": 0.5111, "step": 250 }, { "epoch": 0.10386497553180865, "grad_norm": 6.849791049957275, "learning_rate": 5.18e-06, "loss": 0.5122, "step": 260 }, { "epoch": 0.10785978228303206, "grad_norm": 6.444810390472412, "learning_rate": 5.380000000000001e-06, "loss": 0.4601, "step": 270 }, { "epoch": 0.11185458903425546, "grad_norm": 5.884876728057861, "learning_rate": 5.580000000000001e-06, "loss": 0.4452, "step": 280 }, { "epoch": 0.11584939578547888, "grad_norm": 5.974656105041504, "learning_rate": 5.78e-06, "loss": 0.495, "step": 290 }, { "epoch": 0.11984420253670229, "grad_norm": 5.884718418121338, "learning_rate": 5.98e-06, "loss": 0.4276, "step": 300 }, { "epoch": 0.1238390092879257, "grad_norm": 6.197109699249268, "learning_rate": 6.18e-06, "loss": 0.4378, "step": 310 }, { "epoch": 0.12783381603914912, "grad_norm": 6.656605243682861, "learning_rate": 6.380000000000001e-06, "loss": 0.4352, "step": 320 }, { "epoch": 0.13182862279037252, "grad_norm": 6.549432277679443, "learning_rate": 6.5800000000000005e-06, "loss": 0.4335, "step": 330 }, { "epoch": 0.13582342954159593, "grad_norm": 5.669299125671387, "learning_rate": 6.780000000000001e-06, "loss": 0.4776, "step": 340 }, { "epoch": 0.13981823629281934, "grad_norm": 5.265035152435303, "learning_rate": 6.98e-06, "loss": 0.4493, "step": 350 }, { "epoch": 0.14381304304404274, "grad_norm": 6.127915382385254, "learning_rate": 7.180000000000001e-06, "loss": 0.4505, "step": 360 }, { "epoch": 0.14780784979526615, "grad_norm": 6.895223617553711, "learning_rate": 7.3800000000000005e-06, "loss": 0.4369, "step": 370 }, { "epoch": 0.15180265654648956, "grad_norm": 5.936121940612793, "learning_rate": 7.58e-06, "loss": 0.4212, "step": 380 }, { "epoch": 0.15579746329771296, "grad_norm": 5.121963977813721, "learning_rate": 7.78e-06, "loss": 0.3993, "step": 390 }, { "epoch": 0.15979227004893637, "grad_norm": 6.143499851226807, "learning_rate": 7.980000000000002e-06, "loss": 0.4361, "step": 400 }, { "epoch": 0.1637870768001598, "grad_norm": 5.8908491134643555, "learning_rate": 8.18e-06, "loss": 0.4003, "step": 410 }, { "epoch": 0.1677818835513832, "grad_norm": 5.114202499389648, "learning_rate": 8.380000000000001e-06, "loss": 0.3558, "step": 420 }, { "epoch": 0.17177669030260662, "grad_norm": 5.836564540863037, "learning_rate": 8.580000000000001e-06, "loss": 0.3742, "step": 430 }, { "epoch": 0.17577149705383002, "grad_norm": 5.583286285400391, "learning_rate": 8.78e-06, "loss": 0.4112, "step": 440 }, { "epoch": 0.17976630380505343, "grad_norm": 6.04997444152832, "learning_rate": 8.98e-06, "loss": 0.3803, "step": 450 }, { "epoch": 0.18376111055627684, "grad_norm": 5.244260787963867, "learning_rate": 9.180000000000002e-06, "loss": 0.4266, "step": 460 }, { "epoch": 0.18775591730750024, "grad_norm": 5.963865756988525, "learning_rate": 9.38e-06, "loss": 0.3907, "step": 470 }, { "epoch": 0.19175072405872365, "grad_norm": 5.7191057205200195, "learning_rate": 9.58e-06, "loss": 0.3676, "step": 480 }, { "epoch": 0.19574553080994708, "grad_norm": 5.471884727478027, "learning_rate": 9.780000000000001e-06, "loss": 0.3764, "step": 490 }, { "epoch": 0.1997403375611705, "grad_norm": 6.182490825653076, "learning_rate": 9.980000000000001e-06, "loss": 0.3758, "step": 500 }, { "epoch": 0.2037351443123939, "grad_norm": 4.735085487365723, "learning_rate": 9.98716486023959e-06, "loss": 0.3518, "step": 510 }, { "epoch": 0.2077299510636173, "grad_norm": 5.506002426147461, "learning_rate": 9.972903593839133e-06, "loss": 0.3887, "step": 520 }, { "epoch": 0.2117247578148407, "grad_norm": 5.19524621963501, "learning_rate": 9.958642327438678e-06, "loss": 0.3979, "step": 530 }, { "epoch": 0.21571956456606411, "grad_norm": 6.315662860870361, "learning_rate": 9.944381061038221e-06, "loss": 0.3847, "step": 540 }, { "epoch": 0.21971437131728752, "grad_norm": 5.95819616317749, "learning_rate": 9.930119794637765e-06, "loss": 0.3516, "step": 550 }, { "epoch": 0.22370917806851093, "grad_norm": 5.994788646697998, "learning_rate": 9.91585852823731e-06, "loss": 0.346, "step": 560 }, { "epoch": 0.22770398481973433, "grad_norm": 4.4159369468688965, "learning_rate": 9.901597261836851e-06, "loss": 0.3584, "step": 570 }, { "epoch": 0.23169879157095777, "grad_norm": 4.532053470611572, "learning_rate": 9.887335995436396e-06, "loss": 0.4037, "step": 580 }, { "epoch": 0.23569359832218117, "grad_norm": 5.412591457366943, "learning_rate": 9.87307472903594e-06, "loss": 0.3547, "step": 590 }, { "epoch": 0.23968840507340458, "grad_norm": 5.319347381591797, "learning_rate": 9.858813462635483e-06, "loss": 0.3619, "step": 600 }, { "epoch": 0.243683211824628, "grad_norm": 4.947774887084961, "learning_rate": 9.844552196235026e-06, "loss": 0.3764, "step": 610 }, { "epoch": 0.2476780185758514, "grad_norm": 4.8416666984558105, "learning_rate": 9.83029092983457e-06, "loss": 0.3073, "step": 620 }, { "epoch": 0.2516728253270748, "grad_norm": 6.026658535003662, "learning_rate": 9.816029663434114e-06, "loss": 0.3709, "step": 630 }, { "epoch": 0.25566763207829823, "grad_norm": 6.480794429779053, "learning_rate": 9.801768397033657e-06, "loss": 0.3131, "step": 640 }, { "epoch": 0.2596624388295216, "grad_norm": 5.907824516296387, "learning_rate": 9.787507130633202e-06, "loss": 0.3447, "step": 650 }, { "epoch": 0.26365724558074505, "grad_norm": 4.916101932525635, "learning_rate": 9.773245864232744e-06, "loss": 0.3436, "step": 660 }, { "epoch": 0.2676520523319684, "grad_norm": 5.599504470825195, "learning_rate": 9.758984597832289e-06, "loss": 0.3236, "step": 670 }, { "epoch": 0.27164685908319186, "grad_norm": 6.6841654777526855, "learning_rate": 9.744723331431832e-06, "loss": 0.3491, "step": 680 }, { "epoch": 0.27564166583441524, "grad_norm": 6.430059909820557, "learning_rate": 9.730462065031375e-06, "loss": 0.3343, "step": 690 }, { "epoch": 0.2796364725856387, "grad_norm": 5.829887866973877, "learning_rate": 9.71620079863092e-06, "loss": 0.345, "step": 700 }, { "epoch": 0.28363127933686205, "grad_norm": 4.85678243637085, "learning_rate": 9.701939532230463e-06, "loss": 0.3756, "step": 710 }, { "epoch": 0.2876260860880855, "grad_norm": 4.557241439819336, "learning_rate": 9.687678265830007e-06, "loss": 0.2933, "step": 720 }, { "epoch": 0.2916208928393089, "grad_norm": 5.587825298309326, "learning_rate": 9.67341699942955e-06, "loss": 0.3247, "step": 730 }, { "epoch": 0.2956156995905323, "grad_norm": 5.785783290863037, "learning_rate": 9.659155733029095e-06, "loss": 0.3489, "step": 740 }, { "epoch": 0.29961050634175573, "grad_norm": 5.360633850097656, "learning_rate": 9.644894466628636e-06, "loss": 0.3727, "step": 750 }, { "epoch": 0.3036053130929791, "grad_norm": 5.479579448699951, "learning_rate": 9.630633200228181e-06, "loss": 0.3386, "step": 760 }, { "epoch": 0.30760011984420255, "grad_norm": 5.192019462585449, "learning_rate": 9.616371933827725e-06, "loss": 0.3268, "step": 770 }, { "epoch": 0.3115949265954259, "grad_norm": 5.177510738372803, "learning_rate": 9.602110667427268e-06, "loss": 0.3485, "step": 780 }, { "epoch": 0.31558973334664936, "grad_norm": 4.933345317840576, "learning_rate": 9.587849401026813e-06, "loss": 0.3028, "step": 790 }, { "epoch": 0.31958454009787274, "grad_norm": 4.921146869659424, "learning_rate": 9.573588134626356e-06, "loss": 0.3348, "step": 800 }, { "epoch": 0.32357934684909617, "grad_norm": 5.784856796264648, "learning_rate": 9.5593268682259e-06, "loss": 0.2961, "step": 810 }, { "epoch": 0.3275741536003196, "grad_norm": 5.327091693878174, "learning_rate": 9.545065601825442e-06, "loss": 0.2914, "step": 820 }, { "epoch": 0.331568960351543, "grad_norm": 4.885860443115234, "learning_rate": 9.530804335424987e-06, "loss": 0.3156, "step": 830 }, { "epoch": 0.3355637671027664, "grad_norm": 5.5808844566345215, "learning_rate": 9.516543069024529e-06, "loss": 0.3452, "step": 840 }, { "epoch": 0.3395585738539898, "grad_norm": 5.16964054107666, "learning_rate": 9.502281802624074e-06, "loss": 0.2737, "step": 850 }, { "epoch": 0.34355338060521323, "grad_norm": 4.464290142059326, "learning_rate": 9.488020536223617e-06, "loss": 0.2717, "step": 860 }, { "epoch": 0.3475481873564366, "grad_norm": 4.1654791831970215, "learning_rate": 9.47375926982316e-06, "loss": 0.2875, "step": 870 }, { "epoch": 0.35154299410766005, "grad_norm": 5.758086681365967, "learning_rate": 9.459498003422705e-06, "loss": 0.2947, "step": 880 }, { "epoch": 0.3555378008588834, "grad_norm": 4.330506801605225, "learning_rate": 9.445236737022249e-06, "loss": 0.2957, "step": 890 }, { "epoch": 0.35953260761010686, "grad_norm": 4.616450786590576, "learning_rate": 9.430975470621792e-06, "loss": 0.2752, "step": 900 }, { "epoch": 0.3635274143613303, "grad_norm": 4.364641189575195, "learning_rate": 9.416714204221335e-06, "loss": 0.33, "step": 910 }, { "epoch": 0.36752222111255367, "grad_norm": 4.513298034667969, "learning_rate": 9.40245293782088e-06, "loss": 0.3098, "step": 920 }, { "epoch": 0.3715170278637771, "grad_norm": 5.608431816101074, "learning_rate": 9.388191671420423e-06, "loss": 0.2962, "step": 930 }, { "epoch": 0.3755118346150005, "grad_norm": 3.523578405380249, "learning_rate": 9.373930405019966e-06, "loss": 0.2754, "step": 940 }, { "epoch": 0.3795066413662239, "grad_norm": 5.222875595092773, "learning_rate": 9.35966913861951e-06, "loss": 0.2875, "step": 950 }, { "epoch": 0.3835014481174473, "grad_norm": 5.467960357666016, "learning_rate": 9.345407872219053e-06, "loss": 0.3089, "step": 960 }, { "epoch": 0.38749625486867073, "grad_norm": 4.540034770965576, "learning_rate": 9.331146605818598e-06, "loss": 0.2808, "step": 970 }, { "epoch": 0.39149106161989417, "grad_norm": 4.5449628829956055, "learning_rate": 9.316885339418141e-06, "loss": 0.3063, "step": 980 }, { "epoch": 0.39548586837111754, "grad_norm": 5.014548301696777, "learning_rate": 9.302624073017684e-06, "loss": 0.2795, "step": 990 }, { "epoch": 0.399480675122341, "grad_norm": 4.645490646362305, "learning_rate": 9.288362806617228e-06, "loss": 0.2749, "step": 1000 }, { "epoch": 0.40347548187356436, "grad_norm": 4.577834606170654, "learning_rate": 9.274101540216773e-06, "loss": 0.2722, "step": 1010 }, { "epoch": 0.4074702886247878, "grad_norm": 3.580354928970337, "learning_rate": 9.259840273816316e-06, "loss": 0.2543, "step": 1020 }, { "epoch": 0.41146509537601117, "grad_norm": 4.645276069641113, "learning_rate": 9.245579007415859e-06, "loss": 0.2703, "step": 1030 }, { "epoch": 0.4154599021272346, "grad_norm": 4.813560485839844, "learning_rate": 9.231317741015402e-06, "loss": 0.2708, "step": 1040 }, { "epoch": 0.419454708878458, "grad_norm": 4.8925371170043945, "learning_rate": 9.217056474614946e-06, "loss": 0.2894, "step": 1050 }, { "epoch": 0.4234495156296814, "grad_norm": 4.675732135772705, "learning_rate": 9.20279520821449e-06, "loss": 0.2692, "step": 1060 }, { "epoch": 0.42744432238090485, "grad_norm": 4.893064975738525, "learning_rate": 9.188533941814034e-06, "loss": 0.3143, "step": 1070 }, { "epoch": 0.43143912913212823, "grad_norm": 4.4499592781066895, "learning_rate": 9.174272675413579e-06, "loss": 0.2846, "step": 1080 }, { "epoch": 0.43543393588335166, "grad_norm": 5.22846794128418, "learning_rate": 9.16001140901312e-06, "loss": 0.2384, "step": 1090 }, { "epoch": 0.43942874263457504, "grad_norm": 4.598273277282715, "learning_rate": 9.145750142612665e-06, "loss": 0.2798, "step": 1100 }, { "epoch": 0.4434235493857985, "grad_norm": 4.458007335662842, "learning_rate": 9.131488876212208e-06, "loss": 0.2677, "step": 1110 }, { "epoch": 0.44741835613702186, "grad_norm": 4.530386447906494, "learning_rate": 9.117227609811752e-06, "loss": 0.2985, "step": 1120 }, { "epoch": 0.4514131628882453, "grad_norm": 4.605809688568115, "learning_rate": 9.102966343411297e-06, "loss": 0.2862, "step": 1130 }, { "epoch": 0.45540796963946867, "grad_norm": 4.465124607086182, "learning_rate": 9.08870507701084e-06, "loss": 0.2603, "step": 1140 }, { "epoch": 0.4594027763906921, "grad_norm": 4.143098831176758, "learning_rate": 9.074443810610383e-06, "loss": 0.2777, "step": 1150 }, { "epoch": 0.46339758314191554, "grad_norm": 4.249698162078857, "learning_rate": 9.060182544209926e-06, "loss": 0.2545, "step": 1160 }, { "epoch": 0.4673923898931389, "grad_norm": 5.094964027404785, "learning_rate": 9.045921277809471e-06, "loss": 0.2644, "step": 1170 }, { "epoch": 0.47138719664436235, "grad_norm": 4.91431999206543, "learning_rate": 9.031660011409013e-06, "loss": 0.2718, "step": 1180 }, { "epoch": 0.47538200339558573, "grad_norm": 5.91095495223999, "learning_rate": 9.017398745008558e-06, "loss": 0.2519, "step": 1190 }, { "epoch": 0.47937681014680916, "grad_norm": 4.603298187255859, "learning_rate": 9.003137478608101e-06, "loss": 0.265, "step": 1200 }, { "epoch": 0.48337161689803254, "grad_norm": 4.797214031219482, "learning_rate": 8.988876212207644e-06, "loss": 0.2407, "step": 1210 }, { "epoch": 0.487366423649256, "grad_norm": 4.030250072479248, "learning_rate": 8.97461494580719e-06, "loss": 0.268, "step": 1220 }, { "epoch": 0.49136123040047935, "grad_norm": 5.035445213317871, "learning_rate": 8.960353679406733e-06, "loss": 0.2636, "step": 1230 }, { "epoch": 0.4953560371517028, "grad_norm": 4.499198913574219, "learning_rate": 8.946092413006276e-06, "loss": 0.2651, "step": 1240 }, { "epoch": 0.4993508439029262, "grad_norm": 3.754934310913086, "learning_rate": 8.931831146605819e-06, "loss": 0.2537, "step": 1250 }, { "epoch": 0.5033456506541496, "grad_norm": 4.640712738037109, "learning_rate": 8.917569880205364e-06, "loss": 0.2713, "step": 1260 }, { "epoch": 0.507340457405373, "grad_norm": 4.7820305824279785, "learning_rate": 8.903308613804906e-06, "loss": 0.2554, "step": 1270 }, { "epoch": 0.5113352641565965, "grad_norm": 5.130531311035156, "learning_rate": 8.88904734740445e-06, "loss": 0.2607, "step": 1280 }, { "epoch": 0.5153300709078198, "grad_norm": 4.392201900482178, "learning_rate": 8.874786081003994e-06, "loss": 0.2445, "step": 1290 }, { "epoch": 0.5193248776590432, "grad_norm": 3.3310599327087402, "learning_rate": 8.860524814603537e-06, "loss": 0.2355, "step": 1300 }, { "epoch": 0.5233196844102667, "grad_norm": 3.532226800918579, "learning_rate": 8.846263548203082e-06, "loss": 0.2255, "step": 1310 }, { "epoch": 0.5273144911614901, "grad_norm": 4.452306270599365, "learning_rate": 8.832002281802625e-06, "loss": 0.2671, "step": 1320 }, { "epoch": 0.5313092979127134, "grad_norm": 3.8220486640930176, "learning_rate": 8.817741015402168e-06, "loss": 0.2731, "step": 1330 }, { "epoch": 0.5353041046639369, "grad_norm": 4.6910319328308105, "learning_rate": 8.803479749001712e-06, "loss": 0.2617, "step": 1340 }, { "epoch": 0.5392989114151603, "grad_norm": 4.17935848236084, "learning_rate": 8.789218482601257e-06, "loss": 0.2433, "step": 1350 }, { "epoch": 0.5432937181663837, "grad_norm": 5.38765811920166, "learning_rate": 8.7749572162008e-06, "loss": 0.2466, "step": 1360 }, { "epoch": 0.5472885249176072, "grad_norm": 4.427072525024414, "learning_rate": 8.760695949800343e-06, "loss": 0.2465, "step": 1370 }, { "epoch": 0.5512833316688305, "grad_norm": 5.366933345794678, "learning_rate": 8.746434683399886e-06, "loss": 0.2568, "step": 1380 }, { "epoch": 0.5552781384200539, "grad_norm": 4.636244773864746, "learning_rate": 8.73217341699943e-06, "loss": 0.2652, "step": 1390 }, { "epoch": 0.5592729451712773, "grad_norm": 3.3288471698760986, "learning_rate": 8.717912150598975e-06, "loss": 0.2183, "step": 1400 }, { "epoch": 0.5632677519225008, "grad_norm": 5.008598327636719, "learning_rate": 8.703650884198518e-06, "loss": 0.2838, "step": 1410 }, { "epoch": 0.5672625586737241, "grad_norm": 5.04400110244751, "learning_rate": 8.689389617798061e-06, "loss": 0.2521, "step": 1420 }, { "epoch": 0.5712573654249475, "grad_norm": 3.676771640777588, "learning_rate": 8.675128351397604e-06, "loss": 0.2235, "step": 1430 }, { "epoch": 0.575252172176171, "grad_norm": 3.988940715789795, "learning_rate": 8.66086708499715e-06, "loss": 0.2561, "step": 1440 }, { "epoch": 0.5792469789273944, "grad_norm": 3.753837823867798, "learning_rate": 8.646605818596692e-06, "loss": 0.2349, "step": 1450 }, { "epoch": 0.5832417856786178, "grad_norm": 5.020956993103027, "learning_rate": 8.632344552196236e-06, "loss": 0.2605, "step": 1460 }, { "epoch": 0.5872365924298412, "grad_norm": 4.19816780090332, "learning_rate": 8.618083285795779e-06, "loss": 0.2229, "step": 1470 }, { "epoch": 0.5912313991810646, "grad_norm": 3.8098461627960205, "learning_rate": 8.603822019395322e-06, "loss": 0.2286, "step": 1480 }, { "epoch": 0.595226205932288, "grad_norm": 4.122703552246094, "learning_rate": 8.589560752994867e-06, "loss": 0.2257, "step": 1490 }, { "epoch": 0.5992210126835115, "grad_norm": 5.137128829956055, "learning_rate": 8.57529948659441e-06, "loss": 0.2531, "step": 1500 }, { "epoch": 0.6032158194347348, "grad_norm": 4.318984508514404, "learning_rate": 8.561038220193954e-06, "loss": 0.2384, "step": 1510 }, { "epoch": 0.6072106261859582, "grad_norm": 3.5681166648864746, "learning_rate": 8.546776953793497e-06, "loss": 0.282, "step": 1520 }, { "epoch": 0.6112054329371817, "grad_norm": 2.9932525157928467, "learning_rate": 8.532515687393042e-06, "loss": 0.2095, "step": 1530 }, { "epoch": 0.6152002396884051, "grad_norm": 3.375771999359131, "learning_rate": 8.518254420992585e-06, "loss": 0.2333, "step": 1540 }, { "epoch": 0.6191950464396285, "grad_norm": 4.841177940368652, "learning_rate": 8.503993154592128e-06, "loss": 0.2295, "step": 1550 }, { "epoch": 0.6231898531908519, "grad_norm": 4.5939717292785645, "learning_rate": 8.489731888191672e-06, "loss": 0.2494, "step": 1560 }, { "epoch": 0.6271846599420753, "grad_norm": 3.88786244392395, "learning_rate": 8.475470621791215e-06, "loss": 0.2324, "step": 1570 }, { "epoch": 0.6311794666932987, "grad_norm": 3.3802106380462646, "learning_rate": 8.46120935539076e-06, "loss": 0.2413, "step": 1580 }, { "epoch": 0.6351742734445222, "grad_norm": 4.619571685791016, "learning_rate": 8.446948088990303e-06, "loss": 0.2165, "step": 1590 }, { "epoch": 0.6391690801957455, "grad_norm": 4.9018874168396, "learning_rate": 8.432686822589846e-06, "loss": 0.2496, "step": 1600 }, { "epoch": 0.6431638869469689, "grad_norm": 4.832056522369385, "learning_rate": 8.41842555618939e-06, "loss": 0.1857, "step": 1610 }, { "epoch": 0.6471586936981923, "grad_norm": 4.44606351852417, "learning_rate": 8.404164289788934e-06, "loss": 0.2469, "step": 1620 }, { "epoch": 0.6511535004494158, "grad_norm": 4.6510138511657715, "learning_rate": 8.389903023388478e-06, "loss": 0.2419, "step": 1630 }, { "epoch": 0.6551483072006392, "grad_norm": 4.438849925994873, "learning_rate": 8.375641756988021e-06, "loss": 0.2488, "step": 1640 }, { "epoch": 0.6591431139518625, "grad_norm": 4.8355889320373535, "learning_rate": 8.361380490587566e-06, "loss": 0.2535, "step": 1650 }, { "epoch": 0.663137920703086, "grad_norm": 3.961348533630371, "learning_rate": 8.347119224187107e-06, "loss": 0.2338, "step": 1660 }, { "epoch": 0.6671327274543094, "grad_norm": 4.276237964630127, "learning_rate": 8.332857957786652e-06, "loss": 0.2417, "step": 1670 }, { "epoch": 0.6711275342055328, "grad_norm": 4.317446231842041, "learning_rate": 8.318596691386196e-06, "loss": 0.215, "step": 1680 }, { "epoch": 0.6751223409567562, "grad_norm": 4.8514018058776855, "learning_rate": 8.304335424985739e-06, "loss": 0.2462, "step": 1690 }, { "epoch": 0.6791171477079796, "grad_norm": 5.050139427185059, "learning_rate": 8.290074158585282e-06, "loss": 0.2228, "step": 1700 }, { "epoch": 0.683111954459203, "grad_norm": 3.6597845554351807, "learning_rate": 8.275812892184827e-06, "loss": 0.2337, "step": 1710 }, { "epoch": 0.6871067612104265, "grad_norm": 3.3534600734710693, "learning_rate": 8.26155162578437e-06, "loss": 0.2402, "step": 1720 }, { "epoch": 0.6911015679616499, "grad_norm": 3.987107276916504, "learning_rate": 8.247290359383914e-06, "loss": 0.2154, "step": 1730 }, { "epoch": 0.6950963747128732, "grad_norm": 3.4701499938964844, "learning_rate": 8.233029092983458e-06, "loss": 0.2446, "step": 1740 }, { "epoch": 0.6990911814640967, "grad_norm": 3.479576826095581, "learning_rate": 8.218767826583002e-06, "loss": 0.2138, "step": 1750 }, { "epoch": 0.7030859882153201, "grad_norm": 4.3463358879089355, "learning_rate": 8.204506560182545e-06, "loss": 0.2202, "step": 1760 }, { "epoch": 0.7070807949665435, "grad_norm": 3.920137882232666, "learning_rate": 8.190245293782088e-06, "loss": 0.2184, "step": 1770 }, { "epoch": 0.7110756017177668, "grad_norm": 4.3204121589660645, "learning_rate": 8.175984027381633e-06, "loss": 0.2467, "step": 1780 }, { "epoch": 0.7150704084689903, "grad_norm": 4.039252758026123, "learning_rate": 8.161722760981175e-06, "loss": 0.2238, "step": 1790 }, { "epoch": 0.7190652152202137, "grad_norm": 4.227252006530762, "learning_rate": 8.14746149458072e-06, "loss": 0.2023, "step": 1800 }, { "epoch": 0.7230600219714372, "grad_norm": 2.9367706775665283, "learning_rate": 8.133200228180263e-06, "loss": 0.1977, "step": 1810 }, { "epoch": 0.7270548287226606, "grad_norm": 4.3969831466674805, "learning_rate": 8.118938961779806e-06, "loss": 0.2494, "step": 1820 }, { "epoch": 0.7310496354738839, "grad_norm": 3.7488110065460205, "learning_rate": 8.104677695379351e-06, "loss": 0.2295, "step": 1830 }, { "epoch": 0.7350444422251073, "grad_norm": 5.048527717590332, "learning_rate": 8.090416428978894e-06, "loss": 0.2524, "step": 1840 }, { "epoch": 0.7390392489763308, "grad_norm": 4.360014915466309, "learning_rate": 8.076155162578438e-06, "loss": 0.21, "step": 1850 }, { "epoch": 0.7430340557275542, "grad_norm": 4.142049789428711, "learning_rate": 8.06189389617798e-06, "loss": 0.2198, "step": 1860 }, { "epoch": 0.7470288624787776, "grad_norm": 3.6296193599700928, "learning_rate": 8.047632629777526e-06, "loss": 0.2239, "step": 1870 }, { "epoch": 0.751023669230001, "grad_norm": 4.87466287612915, "learning_rate": 8.033371363377069e-06, "loss": 0.2253, "step": 1880 }, { "epoch": 0.7550184759812244, "grad_norm": 3.6398117542266846, "learning_rate": 8.019110096976612e-06, "loss": 0.2325, "step": 1890 }, { "epoch": 0.7590132827324478, "grad_norm": 3.6101794242858887, "learning_rate": 8.004848830576156e-06, "loss": 0.1967, "step": 1900 }, { "epoch": 0.7630080894836713, "grad_norm": 4.282483100891113, "learning_rate": 7.990587564175699e-06, "loss": 0.2183, "step": 1910 }, { "epoch": 0.7670028962348946, "grad_norm": 4.618693828582764, "learning_rate": 7.976326297775244e-06, "loss": 0.2356, "step": 1920 }, { "epoch": 0.770997702986118, "grad_norm": 3.517585277557373, "learning_rate": 7.962065031374787e-06, "loss": 0.2346, "step": 1930 }, { "epoch": 0.7749925097373415, "grad_norm": 3.8423843383789062, "learning_rate": 7.94780376497433e-06, "loss": 0.2436, "step": 1940 }, { "epoch": 0.7789873164885649, "grad_norm": 4.099793910980225, "learning_rate": 7.933542498573873e-06, "loss": 0.2091, "step": 1950 }, { "epoch": 0.7829821232397883, "grad_norm": 4.841848850250244, "learning_rate": 7.919281232173418e-06, "loss": 0.2328, "step": 1960 }, { "epoch": 0.7869769299910117, "grad_norm": 4.761816024780273, "learning_rate": 7.905019965772962e-06, "loss": 0.2159, "step": 1970 }, { "epoch": 0.7909717367422351, "grad_norm": 4.0117411613464355, "learning_rate": 7.890758699372505e-06, "loss": 0.2287, "step": 1980 }, { "epoch": 0.7949665434934585, "grad_norm": 4.176918029785156, "learning_rate": 7.876497432972048e-06, "loss": 0.2036, "step": 1990 }, { "epoch": 0.798961350244682, "grad_norm": 4.187548637390137, "learning_rate": 7.862236166571591e-06, "loss": 0.2175, "step": 2000 }, { "epoch": 0.798961350244682, "eval_loss": 0.17748136818408966, "eval_runtime": 8461.8051, "eval_samples_per_second": 2.367, "eval_steps_per_second": 0.296, "eval_wer": 15.578133881334441, "step": 2000 }, { "epoch": 0.8029561569959053, "grad_norm": 8.588675498962402, "learning_rate": 7.847974900171136e-06, "loss": 0.2548, "step": 2010 }, { "epoch": 0.8069509637471287, "grad_norm": 4.142924785614014, "learning_rate": 7.83371363377068e-06, "loss": 0.2145, "step": 2020 }, { "epoch": 0.8109457704983521, "grad_norm": 4.049276828765869, "learning_rate": 7.819452367370223e-06, "loss": 0.2227, "step": 2030 }, { "epoch": 0.8149405772495756, "grad_norm": 4.361568450927734, "learning_rate": 7.805191100969766e-06, "loss": 0.2352, "step": 2040 }, { "epoch": 0.818935384000799, "grad_norm": 4.02155876159668, "learning_rate": 7.790929834569311e-06, "loss": 0.1892, "step": 2050 }, { "epoch": 0.8229301907520223, "grad_norm": 4.061072826385498, "learning_rate": 7.776668568168854e-06, "loss": 0.2219, "step": 2060 }, { "epoch": 0.8269249975032458, "grad_norm": 3.2685811519622803, "learning_rate": 7.762407301768397e-06, "loss": 0.2057, "step": 2070 }, { "epoch": 0.8309198042544692, "grad_norm": 4.679708480834961, "learning_rate": 7.748146035367942e-06, "loss": 0.2364, "step": 2080 }, { "epoch": 0.8349146110056926, "grad_norm": 3.1732335090637207, "learning_rate": 7.733884768967484e-06, "loss": 0.2364, "step": 2090 }, { "epoch": 0.838909417756916, "grad_norm": 4.397407054901123, "learning_rate": 7.719623502567029e-06, "loss": 0.2251, "step": 2100 }, { "epoch": 0.8429042245081394, "grad_norm": 3.7484922409057617, "learning_rate": 7.705362236166572e-06, "loss": 0.2032, "step": 2110 }, { "epoch": 0.8468990312593628, "grad_norm": 4.67540168762207, "learning_rate": 7.691100969766115e-06, "loss": 0.2252, "step": 2120 }, { "epoch": 0.8508938380105863, "grad_norm": 3.038043975830078, "learning_rate": 7.676839703365659e-06, "loss": 0.2204, "step": 2130 }, { "epoch": 0.8548886447618097, "grad_norm": 4.430922985076904, "learning_rate": 7.662578436965204e-06, "loss": 0.2214, "step": 2140 }, { "epoch": 0.858883451513033, "grad_norm": 3.289482593536377, "learning_rate": 7.648317170564747e-06, "loss": 0.2301, "step": 2150 }, { "epoch": 0.8628782582642565, "grad_norm": 3.887826919555664, "learning_rate": 7.63405590416429e-06, "loss": 0.2448, "step": 2160 }, { "epoch": 0.8668730650154799, "grad_norm": 4.0688157081604, "learning_rate": 7.619794637763834e-06, "loss": 0.1966, "step": 2170 }, { "epoch": 0.8708678717667033, "grad_norm": 3.688675880432129, "learning_rate": 7.6055333713633774e-06, "loss": 0.197, "step": 2180 }, { "epoch": 0.8748626785179267, "grad_norm": 3.8685896396636963, "learning_rate": 7.5912721049629215e-06, "loss": 0.2144, "step": 2190 }, { "epoch": 0.8788574852691501, "grad_norm": 3.8697168827056885, "learning_rate": 7.577010838562466e-06, "loss": 0.21, "step": 2200 }, { "epoch": 0.8828522920203735, "grad_norm": 3.6243553161621094, "learning_rate": 7.562749572162008e-06, "loss": 0.2115, "step": 2210 }, { "epoch": 0.886847098771597, "grad_norm": 4.4122443199157715, "learning_rate": 7.548488305761552e-06, "loss": 0.2297, "step": 2220 }, { "epoch": 0.8908419055228204, "grad_norm": 3.5293548107147217, "learning_rate": 7.534227039361096e-06, "loss": 0.2012, "step": 2230 }, { "epoch": 0.8948367122740437, "grad_norm": 3.237689733505249, "learning_rate": 7.519965772960639e-06, "loss": 0.2104, "step": 2240 }, { "epoch": 0.8988315190252671, "grad_norm": 3.9817986488342285, "learning_rate": 7.505704506560183e-06, "loss": 0.2178, "step": 2250 }, { "epoch": 0.9028263257764906, "grad_norm": 4.46220064163208, "learning_rate": 7.491443240159727e-06, "loss": 0.2214, "step": 2260 }, { "epoch": 0.906821132527714, "grad_norm": 3.8697423934936523, "learning_rate": 7.47718197375927e-06, "loss": 0.1978, "step": 2270 }, { "epoch": 0.9108159392789373, "grad_norm": 3.2078206539154053, "learning_rate": 7.462920707358814e-06, "loss": 0.2084, "step": 2280 }, { "epoch": 0.9148107460301608, "grad_norm": 4.6600494384765625, "learning_rate": 7.448659440958358e-06, "loss": 0.2199, "step": 2290 }, { "epoch": 0.9188055527813842, "grad_norm": 4.262337684631348, "learning_rate": 7.434398174557901e-06, "loss": 0.2115, "step": 2300 }, { "epoch": 0.9228003595326076, "grad_norm": 4.174410820007324, "learning_rate": 7.420136908157445e-06, "loss": 0.1861, "step": 2310 }, { "epoch": 0.9267951662838311, "grad_norm": 3.7125275135040283, "learning_rate": 7.405875641756989e-06, "loss": 0.2176, "step": 2320 }, { "epoch": 0.9307899730350544, "grad_norm": 3.946467161178589, "learning_rate": 7.391614375356533e-06, "loss": 0.2119, "step": 2330 }, { "epoch": 0.9347847797862778, "grad_norm": 4.180386066436768, "learning_rate": 7.377353108956075e-06, "loss": 0.1986, "step": 2340 }, { "epoch": 0.9387795865375013, "grad_norm": 3.4374818801879883, "learning_rate": 7.3630918425556194e-06, "loss": 0.1921, "step": 2350 }, { "epoch": 0.9427743932887247, "grad_norm": 3.786980390548706, "learning_rate": 7.3488305761551635e-06, "loss": 0.2065, "step": 2360 }, { "epoch": 0.946769200039948, "grad_norm": 3.3135437965393066, "learning_rate": 7.334569309754707e-06, "loss": 0.2068, "step": 2370 }, { "epoch": 0.9507640067911715, "grad_norm": 3.5033023357391357, "learning_rate": 7.320308043354251e-06, "loss": 0.2276, "step": 2380 }, { "epoch": 0.9547588135423949, "grad_norm": 4.119509220123291, "learning_rate": 7.306046776953795e-06, "loss": 0.2009, "step": 2390 }, { "epoch": 0.9587536202936183, "grad_norm": 4.391267776489258, "learning_rate": 7.291785510553337e-06, "loss": 0.1966, "step": 2400 }, { "epoch": 0.9627484270448418, "grad_norm": 3.9926700592041016, "learning_rate": 7.2775242441528815e-06, "loss": 0.2245, "step": 2410 }, { "epoch": 0.9667432337960651, "grad_norm": 4.617944240570068, "learning_rate": 7.2632629777524256e-06, "loss": 0.1978, "step": 2420 }, { "epoch": 0.9707380405472885, "grad_norm": 3.33878231048584, "learning_rate": 7.249001711351969e-06, "loss": 0.1897, "step": 2430 }, { "epoch": 0.974732847298512, "grad_norm": 3.526387929916382, "learning_rate": 7.234740444951512e-06, "loss": 0.1706, "step": 2440 }, { "epoch": 0.9787276540497354, "grad_norm": 3.532831907272339, "learning_rate": 7.220479178551056e-06, "loss": 0.2061, "step": 2450 }, { "epoch": 0.9827224608009587, "grad_norm": 4.9069743156433105, "learning_rate": 7.206217912150599e-06, "loss": 0.207, "step": 2460 }, { "epoch": 0.9867172675521821, "grad_norm": 3.986372232437134, "learning_rate": 7.1919566457501435e-06, "loss": 0.2228, "step": 2470 }, { "epoch": 0.9907120743034056, "grad_norm": 3.50144100189209, "learning_rate": 7.1776953793496876e-06, "loss": 0.1928, "step": 2480 }, { "epoch": 0.994706881054629, "grad_norm": 3.42572283744812, "learning_rate": 7.16343411294923e-06, "loss": 0.2104, "step": 2490 }, { "epoch": 0.9987016878058524, "grad_norm": 3.9188385009765625, "learning_rate": 7.149172846548774e-06, "loss": 0.2023, "step": 2500 }, { "epoch": 1.002396884050734, "grad_norm": 2.447148084640503, "learning_rate": 7.134911580148318e-06, "loss": 0.1921, "step": 2510 }, { "epoch": 1.0063916908019575, "grad_norm": 2.6271724700927734, "learning_rate": 7.120650313747861e-06, "loss": 0.1441, "step": 2520 }, { "epoch": 1.0103864975531809, "grad_norm": 2.705094337463379, "learning_rate": 7.1063890473474055e-06, "loss": 0.1244, "step": 2530 }, { "epoch": 1.0143813043044043, "grad_norm": 2.762216091156006, "learning_rate": 7.092127780946949e-06, "loss": 0.1305, "step": 2540 }, { "epoch": 1.0183761110556278, "grad_norm": 3.086855888366699, "learning_rate": 7.077866514546492e-06, "loss": 0.1538, "step": 2550 }, { "epoch": 1.0223709178068512, "grad_norm": 2.8569979667663574, "learning_rate": 7.063605248146036e-06, "loss": 0.1351, "step": 2560 }, { "epoch": 1.0263657245580744, "grad_norm": 2.759294271469116, "learning_rate": 7.04934398174558e-06, "loss": 0.1457, "step": 2570 }, { "epoch": 1.0303605313092978, "grad_norm": 3.2542331218719482, "learning_rate": 7.035082715345123e-06, "loss": 0.1191, "step": 2580 }, { "epoch": 1.0343553380605213, "grad_norm": 2.826601028442383, "learning_rate": 7.020821448944667e-06, "loss": 0.1291, "step": 2590 }, { "epoch": 1.0383501448117447, "grad_norm": 3.4102251529693604, "learning_rate": 7.006560182544211e-06, "loss": 0.1574, "step": 2600 }, { "epoch": 1.0423449515629681, "grad_norm": 3.3488783836364746, "learning_rate": 6.992298916143754e-06, "loss": 0.1119, "step": 2610 }, { "epoch": 1.0463397583141916, "grad_norm": 2.898468494415283, "learning_rate": 6.978037649743298e-06, "loss": 0.1437, "step": 2620 }, { "epoch": 1.050334565065415, "grad_norm": 2.1492602825164795, "learning_rate": 6.963776383342841e-06, "loss": 0.1441, "step": 2630 }, { "epoch": 1.0543293718166384, "grad_norm": 3.253286361694336, "learning_rate": 6.949515116942385e-06, "loss": 0.141, "step": 2640 }, { "epoch": 1.0583241785678619, "grad_norm": 4.033700466156006, "learning_rate": 6.935253850541929e-06, "loss": 0.1474, "step": 2650 }, { "epoch": 1.062318985319085, "grad_norm": 3.222522258758545, "learning_rate": 6.920992584141473e-06, "loss": 0.126, "step": 2660 }, { "epoch": 1.0663137920703085, "grad_norm": 3.157158851623535, "learning_rate": 6.906731317741015e-06, "loss": 0.1451, "step": 2670 }, { "epoch": 1.070308598821532, "grad_norm": 3.311572551727295, "learning_rate": 6.892470051340559e-06, "loss": 0.1338, "step": 2680 }, { "epoch": 1.0743034055727554, "grad_norm": 2.2077033519744873, "learning_rate": 6.878208784940103e-06, "loss": 0.1155, "step": 2690 }, { "epoch": 1.0782982123239788, "grad_norm": 2.602149248123169, "learning_rate": 6.863947518539647e-06, "loss": 0.124, "step": 2700 }, { "epoch": 1.0822930190752023, "grad_norm": 3.072699785232544, "learning_rate": 6.849686252139191e-06, "loss": 0.1463, "step": 2710 }, { "epoch": 1.0862878258264257, "grad_norm": 2.7928380966186523, "learning_rate": 6.835424985738735e-06, "loss": 0.1233, "step": 2720 }, { "epoch": 1.0902826325776491, "grad_norm": 2.895634174346924, "learning_rate": 6.821163719338277e-06, "loss": 0.1333, "step": 2730 }, { "epoch": 1.0942774393288726, "grad_norm": 2.9929039478302, "learning_rate": 6.806902452937821e-06, "loss": 0.1244, "step": 2740 }, { "epoch": 1.098272246080096, "grad_norm": 2.920032501220703, "learning_rate": 6.792641186537365e-06, "loss": 0.1175, "step": 2750 }, { "epoch": 1.1022670528313192, "grad_norm": 3.432638168334961, "learning_rate": 6.778379920136909e-06, "loss": 0.1406, "step": 2760 }, { "epoch": 1.1062618595825426, "grad_norm": 2.5651278495788574, "learning_rate": 6.764118653736452e-06, "loss": 0.1303, "step": 2770 }, { "epoch": 1.110256666333766, "grad_norm": 3.10604190826416, "learning_rate": 6.749857387335996e-06, "loss": 0.1388, "step": 2780 }, { "epoch": 1.1142514730849895, "grad_norm": 3.3140065670013428, "learning_rate": 6.735596120935539e-06, "loss": 0.1368, "step": 2790 }, { "epoch": 1.118246279836213, "grad_norm": 2.830920696258545, "learning_rate": 6.721334854535083e-06, "loss": 0.1234, "step": 2800 }, { "epoch": 1.1222410865874364, "grad_norm": 3.0485241413116455, "learning_rate": 6.7070735881346274e-06, "loss": 0.1248, "step": 2810 }, { "epoch": 1.1262358933386598, "grad_norm": 2.8369975090026855, "learning_rate": 6.69281232173417e-06, "loss": 0.1465, "step": 2820 }, { "epoch": 1.1302307000898832, "grad_norm": 2.5955135822296143, "learning_rate": 6.678551055333714e-06, "loss": 0.1442, "step": 2830 }, { "epoch": 1.1342255068411067, "grad_norm": 3.3618500232696533, "learning_rate": 6.664289788933258e-06, "loss": 0.1343, "step": 2840 }, { "epoch": 1.13822031359233, "grad_norm": 2.4381425380706787, "learning_rate": 6.650028522532801e-06, "loss": 0.1302, "step": 2850 }, { "epoch": 1.1422151203435533, "grad_norm": 3.9529569149017334, "learning_rate": 6.635767256132345e-06, "loss": 0.1243, "step": 2860 }, { "epoch": 1.1462099270947768, "grad_norm": 2.211047410964966, "learning_rate": 6.621505989731889e-06, "loss": 0.1276, "step": 2870 }, { "epoch": 1.1502047338460002, "grad_norm": 3.246532678604126, "learning_rate": 6.607244723331432e-06, "loss": 0.1377, "step": 2880 }, { "epoch": 1.1541995405972236, "grad_norm": 2.9653186798095703, "learning_rate": 6.592983456930976e-06, "loss": 0.1431, "step": 2890 }, { "epoch": 1.158194347348447, "grad_norm": 3.3144278526306152, "learning_rate": 6.57872219053052e-06, "loss": 0.1608, "step": 2900 }, { "epoch": 1.1621891540996705, "grad_norm": 3.038832426071167, "learning_rate": 6.5644609241300625e-06, "loss": 0.1503, "step": 2910 }, { "epoch": 1.166183960850894, "grad_norm": 2.9031283855438232, "learning_rate": 6.5501996577296066e-06, "loss": 0.1403, "step": 2920 }, { "epoch": 1.1701787676021174, "grad_norm": 2.932734489440918, "learning_rate": 6.535938391329151e-06, "loss": 0.1162, "step": 2930 }, { "epoch": 1.1741735743533406, "grad_norm": 3.1002743244171143, "learning_rate": 6.521677124928694e-06, "loss": 0.1486, "step": 2940 }, { "epoch": 1.178168381104564, "grad_norm": 2.8298721313476562, "learning_rate": 6.507415858528238e-06, "loss": 0.1247, "step": 2950 }, { "epoch": 1.1821631878557874, "grad_norm": 2.830777406692505, "learning_rate": 6.493154592127781e-06, "loss": 0.1253, "step": 2960 }, { "epoch": 1.1861579946070109, "grad_norm": 2.3301286697387695, "learning_rate": 6.478893325727325e-06, "loss": 0.1302, "step": 2970 }, { "epoch": 1.1901528013582343, "grad_norm": 2.502960205078125, "learning_rate": 6.464632059326869e-06, "loss": 0.1185, "step": 2980 }, { "epoch": 1.1941476081094577, "grad_norm": 2.7609617710113525, "learning_rate": 6.450370792926413e-06, "loss": 0.126, "step": 2990 }, { "epoch": 1.1981424148606812, "grad_norm": 2.7776389122009277, "learning_rate": 6.436109526525957e-06, "loss": 0.1371, "step": 3000 }, { "epoch": 1.2021372216119046, "grad_norm": 2.744546413421631, "learning_rate": 6.421848260125499e-06, "loss": 0.1286, "step": 3010 }, { "epoch": 1.206132028363128, "grad_norm": 3.670236825942993, "learning_rate": 6.407586993725043e-06, "loss": 0.1434, "step": 3020 }, { "epoch": 1.2101268351143513, "grad_norm": 3.0598738193511963, "learning_rate": 6.393325727324587e-06, "loss": 0.1251, "step": 3030 }, { "epoch": 1.2141216418655747, "grad_norm": 3.536545991897583, "learning_rate": 6.379064460924131e-06, "loss": 0.1399, "step": 3040 }, { "epoch": 1.2181164486167981, "grad_norm": 4.090343952178955, "learning_rate": 6.364803194523675e-06, "loss": 0.1322, "step": 3050 }, { "epoch": 1.2221112553680216, "grad_norm": 2.5921967029571533, "learning_rate": 6.350541928123218e-06, "loss": 0.1233, "step": 3060 }, { "epoch": 1.226106062119245, "grad_norm": 2.831301689147949, "learning_rate": 6.336280661722761e-06, "loss": 0.1301, "step": 3070 }, { "epoch": 1.2301008688704684, "grad_norm": 2.6285064220428467, "learning_rate": 6.322019395322305e-06, "loss": 0.1357, "step": 3080 }, { "epoch": 1.2340956756216919, "grad_norm": 3.501486301422119, "learning_rate": 6.307758128921849e-06, "loss": 0.1387, "step": 3090 }, { "epoch": 1.2380904823729153, "grad_norm": 3.47078800201416, "learning_rate": 6.293496862521392e-06, "loss": 0.1312, "step": 3100 }, { "epoch": 1.2420852891241387, "grad_norm": 3.3166306018829346, "learning_rate": 6.279235596120936e-06, "loss": 0.1103, "step": 3110 }, { "epoch": 1.246080095875362, "grad_norm": 2.7375359535217285, "learning_rate": 6.26497432972048e-06, "loss": 0.1456, "step": 3120 }, { "epoch": 1.2500749026265854, "grad_norm": 2.8451974391937256, "learning_rate": 6.250713063320023e-06, "loss": 0.1323, "step": 3130 }, { "epoch": 1.2540697093778088, "grad_norm": 3.141065835952759, "learning_rate": 6.236451796919567e-06, "loss": 0.1307, "step": 3140 }, { "epoch": 1.2580645161290323, "grad_norm": 2.7583253383636475, "learning_rate": 6.222190530519111e-06, "loss": 0.1218, "step": 3150 }, { "epoch": 1.2620593228802557, "grad_norm": 2.8716115951538086, "learning_rate": 6.207929264118654e-06, "loss": 0.1211, "step": 3160 }, { "epoch": 1.2660541296314791, "grad_norm": 3.0804483890533447, "learning_rate": 6.193667997718198e-06, "loss": 0.1448, "step": 3170 }, { "epoch": 1.2700489363827026, "grad_norm": 2.831940174102783, "learning_rate": 6.179406731317742e-06, "loss": 0.1195, "step": 3180 }, { "epoch": 1.2740437431339258, "grad_norm": 3.2198126316070557, "learning_rate": 6.165145464917285e-06, "loss": 0.1337, "step": 3190 }, { "epoch": 1.2780385498851494, "grad_norm": 3.0133731365203857, "learning_rate": 6.1508841985168285e-06, "loss": 0.1368, "step": 3200 }, { "epoch": 1.2820333566363726, "grad_norm": 2.603764057159424, "learning_rate": 6.136622932116373e-06, "loss": 0.1169, "step": 3210 }, { "epoch": 1.286028163387596, "grad_norm": 2.296199083328247, "learning_rate": 6.122361665715916e-06, "loss": 0.1143, "step": 3220 }, { "epoch": 1.2900229701388195, "grad_norm": 2.61836838722229, "learning_rate": 6.10810039931546e-06, "loss": 0.119, "step": 3230 }, { "epoch": 1.294017776890043, "grad_norm": 3.154714584350586, "learning_rate": 6.093839132915004e-06, "loss": 0.1283, "step": 3240 }, { "epoch": 1.2980125836412664, "grad_norm": 3.181041717529297, "learning_rate": 6.0795778665145464e-06, "loss": 0.1149, "step": 3250 }, { "epoch": 1.3020073903924898, "grad_norm": 2.668581008911133, "learning_rate": 6.0653166001140905e-06, "loss": 0.1225, "step": 3260 }, { "epoch": 1.3060021971437132, "grad_norm": 3.9531939029693604, "learning_rate": 6.051055333713635e-06, "loss": 0.1317, "step": 3270 }, { "epoch": 1.3099970038949365, "grad_norm": 3.74332594871521, "learning_rate": 6.036794067313178e-06, "loss": 0.1332, "step": 3280 }, { "epoch": 1.31399181064616, "grad_norm": 2.8825340270996094, "learning_rate": 6.022532800912721e-06, "loss": 0.1334, "step": 3290 }, { "epoch": 1.3179866173973833, "grad_norm": 2.882206916809082, "learning_rate": 6.008271534512265e-06, "loss": 0.1192, "step": 3300 }, { "epoch": 1.3219814241486068, "grad_norm": 3.6865313053131104, "learning_rate": 5.9940102681118085e-06, "loss": 0.1203, "step": 3310 }, { "epoch": 1.3259762308998302, "grad_norm": 3.1911978721618652, "learning_rate": 5.9797490017113525e-06, "loss": 0.1218, "step": 3320 }, { "epoch": 1.3299710376510536, "grad_norm": 3.1022403240203857, "learning_rate": 5.965487735310897e-06, "loss": 0.1206, "step": 3330 }, { "epoch": 1.333965844402277, "grad_norm": 2.3271052837371826, "learning_rate": 5.951226468910439e-06, "loss": 0.1333, "step": 3340 }, { "epoch": 1.3379606511535005, "grad_norm": 2.547220230102539, "learning_rate": 5.936965202509983e-06, "loss": 0.1356, "step": 3350 }, { "epoch": 1.341955457904724, "grad_norm": 3.0954415798187256, "learning_rate": 5.922703936109527e-06, "loss": 0.1094, "step": 3360 }, { "epoch": 1.3459502646559471, "grad_norm": 2.3438808917999268, "learning_rate": 5.9084426697090705e-06, "loss": 0.1099, "step": 3370 }, { "epoch": 1.3499450714071708, "grad_norm": 3.047748565673828, "learning_rate": 5.8941814033086146e-06, "loss": 0.1308, "step": 3380 }, { "epoch": 1.353939878158394, "grad_norm": 2.9918041229248047, "learning_rate": 5.879920136908158e-06, "loss": 0.1325, "step": 3390 }, { "epoch": 1.3579346849096174, "grad_norm": 3.4993977546691895, "learning_rate": 5.865658870507701e-06, "loss": 0.1454, "step": 3400 }, { "epoch": 1.3619294916608409, "grad_norm": 4.395610809326172, "learning_rate": 5.851397604107245e-06, "loss": 0.1438, "step": 3410 }, { "epoch": 1.3659242984120643, "grad_norm": 2.523089647293091, "learning_rate": 5.837136337706789e-06, "loss": 0.1202, "step": 3420 }, { "epoch": 1.3699191051632877, "grad_norm": 2.9900269508361816, "learning_rate": 5.822875071306332e-06, "loss": 0.1404, "step": 3430 }, { "epoch": 1.3739139119145112, "grad_norm": 2.8332931995391846, "learning_rate": 5.808613804905876e-06, "loss": 0.1307, "step": 3440 }, { "epoch": 1.3779087186657346, "grad_norm": 3.6861443519592285, "learning_rate": 5.79435253850542e-06, "loss": 0.1308, "step": 3450 }, { "epoch": 1.3819035254169578, "grad_norm": 3.442070722579956, "learning_rate": 5.780091272104963e-06, "loss": 0.1174, "step": 3460 }, { "epoch": 1.3858983321681815, "grad_norm": 3.1186418533325195, "learning_rate": 5.765830005704507e-06, "loss": 0.1086, "step": 3470 }, { "epoch": 1.3898931389194047, "grad_norm": 3.1245079040527344, "learning_rate": 5.751568739304051e-06, "loss": 0.1168, "step": 3480 }, { "epoch": 1.3938879456706281, "grad_norm": 2.674656629562378, "learning_rate": 5.737307472903594e-06, "loss": 0.127, "step": 3490 }, { "epoch": 1.3978827524218516, "grad_norm": 2.8337886333465576, "learning_rate": 5.723046206503138e-06, "loss": 0.1585, "step": 3500 }, { "epoch": 1.401877559173075, "grad_norm": 3.2349750995635986, "learning_rate": 5.708784940102682e-06, "loss": 0.1329, "step": 3510 }, { "epoch": 1.4058723659242984, "grad_norm": 3.044790744781494, "learning_rate": 5.694523673702225e-06, "loss": 0.1277, "step": 3520 }, { "epoch": 1.4098671726755219, "grad_norm": 3.2932238578796387, "learning_rate": 5.680262407301768e-06, "loss": 0.1227, "step": 3530 }, { "epoch": 1.4138619794267453, "grad_norm": 3.259983777999878, "learning_rate": 5.6660011409013125e-06, "loss": 0.1201, "step": 3540 }, { "epoch": 1.4178567861779685, "grad_norm": 3.202609062194824, "learning_rate": 5.651739874500856e-06, "loss": 0.1436, "step": 3550 }, { "epoch": 1.4218515929291922, "grad_norm": 2.5954806804656982, "learning_rate": 5.6374786081004e-06, "loss": 0.1204, "step": 3560 }, { "epoch": 1.4258463996804154, "grad_norm": 2.437851667404175, "learning_rate": 5.623217341699944e-06, "loss": 0.1307, "step": 3570 }, { "epoch": 1.4298412064316388, "grad_norm": 2.9584476947784424, "learning_rate": 5.608956075299488e-06, "loss": 0.1232, "step": 3580 }, { "epoch": 1.4338360131828622, "grad_norm": 2.2511274814605713, "learning_rate": 5.59469480889903e-06, "loss": 0.1233, "step": 3590 }, { "epoch": 1.4378308199340857, "grad_norm": 3.3061892986297607, "learning_rate": 5.5804335424985745e-06, "loss": 0.1174, "step": 3600 }, { "epoch": 1.4418256266853091, "grad_norm": 2.1911284923553467, "learning_rate": 5.566172276098119e-06, "loss": 0.1263, "step": 3610 }, { "epoch": 1.4458204334365325, "grad_norm": 3.9662625789642334, "learning_rate": 5.551911009697661e-06, "loss": 0.1367, "step": 3620 }, { "epoch": 1.449815240187756, "grad_norm": 2.914937973022461, "learning_rate": 5.537649743297205e-06, "loss": 0.1349, "step": 3630 }, { "epoch": 1.4538100469389792, "grad_norm": 3.0488805770874023, "learning_rate": 5.523388476896749e-06, "loss": 0.1309, "step": 3640 }, { "epoch": 1.4578048536902029, "grad_norm": 3.5103049278259277, "learning_rate": 5.509127210496292e-06, "loss": 0.1419, "step": 3650 }, { "epoch": 1.461799660441426, "grad_norm": 2.7047817707061768, "learning_rate": 5.4948659440958365e-06, "loss": 0.1254, "step": 3660 }, { "epoch": 1.4657944671926495, "grad_norm": 2.7503600120544434, "learning_rate": 5.480604677695381e-06, "loss": 0.121, "step": 3670 }, { "epoch": 1.469789273943873, "grad_norm": 3.51440167427063, "learning_rate": 5.466343411294923e-06, "loss": 0.1132, "step": 3680 }, { "epoch": 1.4737840806950964, "grad_norm": 2.5173933506011963, "learning_rate": 5.452082144894467e-06, "loss": 0.1137, "step": 3690 }, { "epoch": 1.4777788874463198, "grad_norm": 2.483872175216675, "learning_rate": 5.437820878494011e-06, "loss": 0.1283, "step": 3700 }, { "epoch": 1.4817736941975432, "grad_norm": 2.768322229385376, "learning_rate": 5.4235596120935544e-06, "loss": 0.143, "step": 3710 }, { "epoch": 1.4857685009487667, "grad_norm": 3.048396587371826, "learning_rate": 5.409298345693098e-06, "loss": 0.1258, "step": 3720 }, { "epoch": 1.4897633076999899, "grad_norm": 2.805190324783325, "learning_rate": 5.395037079292642e-06, "loss": 0.1344, "step": 3730 }, { "epoch": 1.4937581144512135, "grad_norm": 3.2144975662231445, "learning_rate": 5.380775812892185e-06, "loss": 0.1216, "step": 3740 }, { "epoch": 1.4977529212024367, "grad_norm": 3.064366340637207, "learning_rate": 5.366514546491729e-06, "loss": 0.1255, "step": 3750 }, { "epoch": 1.5017477279536604, "grad_norm": 2.593935966491699, "learning_rate": 5.352253280091273e-06, "loss": 0.1128, "step": 3760 }, { "epoch": 1.5057425347048836, "grad_norm": 3.0325539112091064, "learning_rate": 5.337992013690816e-06, "loss": 0.128, "step": 3770 }, { "epoch": 1.509737341456107, "grad_norm": 2.8857476711273193, "learning_rate": 5.32373074729036e-06, "loss": 0.1269, "step": 3780 }, { "epoch": 1.5137321482073305, "grad_norm": 2.5328145027160645, "learning_rate": 5.309469480889904e-06, "loss": 0.1273, "step": 3790 }, { "epoch": 1.517726954958554, "grad_norm": 2.4243452548980713, "learning_rate": 5.295208214489447e-06, "loss": 0.1101, "step": 3800 }, { "epoch": 1.5217217617097774, "grad_norm": 2.239300012588501, "learning_rate": 5.280946948088991e-06, "loss": 0.1174, "step": 3810 }, { "epoch": 1.5257165684610006, "grad_norm": 2.756876230239868, "learning_rate": 5.266685681688534e-06, "loss": 0.1059, "step": 3820 }, { "epoch": 1.5297113752122242, "grad_norm": 1.9072672128677368, "learning_rate": 5.252424415288078e-06, "loss": 0.1061, "step": 3830 }, { "epoch": 1.5337061819634474, "grad_norm": 3.0514819622039795, "learning_rate": 5.238163148887622e-06, "loss": 0.1186, "step": 3840 }, { "epoch": 1.537700988714671, "grad_norm": 2.7836649417877197, "learning_rate": 5.223901882487166e-06, "loss": 0.1114, "step": 3850 }, { "epoch": 1.5416957954658943, "grad_norm": 2.7533199787139893, "learning_rate": 5.209640616086708e-06, "loss": 0.1168, "step": 3860 }, { "epoch": 1.5456906022171177, "grad_norm": 3.529172897338867, "learning_rate": 5.195379349686252e-06, "loss": 0.1409, "step": 3870 }, { "epoch": 1.5496854089683412, "grad_norm": 3.022292137145996, "learning_rate": 5.1811180832857964e-06, "loss": 0.107, "step": 3880 }, { "epoch": 1.5536802157195646, "grad_norm": 2.6646084785461426, "learning_rate": 5.16685681688534e-06, "loss": 0.1202, "step": 3890 }, { "epoch": 1.557675022470788, "grad_norm": 2.9116556644439697, "learning_rate": 5.152595550484884e-06, "loss": 0.1289, "step": 3900 }, { "epoch": 1.5616698292220113, "grad_norm": 2.6292238235473633, "learning_rate": 5.138334284084428e-06, "loss": 0.1191, "step": 3910 }, { "epoch": 1.565664635973235, "grad_norm": 3.42012619972229, "learning_rate": 5.12407301768397e-06, "loss": 0.1219, "step": 3920 }, { "epoch": 1.5696594427244581, "grad_norm": 3.27042555809021, "learning_rate": 5.109811751283514e-06, "loss": 0.1269, "step": 3930 }, { "epoch": 1.5736542494756818, "grad_norm": 3.3910858631134033, "learning_rate": 5.0955504848830584e-06, "loss": 0.1098, "step": 3940 }, { "epoch": 1.577649056226905, "grad_norm": 1.9884843826293945, "learning_rate": 5.081289218482601e-06, "loss": 0.1085, "step": 3950 }, { "epoch": 1.5816438629781284, "grad_norm": 2.9106059074401855, "learning_rate": 5.067027952082145e-06, "loss": 0.1173, "step": 3960 }, { "epoch": 1.5856386697293519, "grad_norm": 2.5543880462646484, "learning_rate": 5.052766685681689e-06, "loss": 0.1182, "step": 3970 }, { "epoch": 1.5896334764805753, "grad_norm": 3.7071778774261475, "learning_rate": 5.038505419281232e-06, "loss": 0.1245, "step": 3980 }, { "epoch": 1.5936282832317987, "grad_norm": 2.9774022102355957, "learning_rate": 5.024244152880776e-06, "loss": 0.1157, "step": 3990 }, { "epoch": 1.597623089983022, "grad_norm": 2.8045384883880615, "learning_rate": 5.0099828864803205e-06, "loss": 0.11, "step": 4000 }, { "epoch": 1.597623089983022, "eval_loss": 0.10785084217786789, "eval_runtime": 8524.6921, "eval_samples_per_second": 2.349, "eval_steps_per_second": 0.294, "eval_wer": 13.268986926675701, "step": 4000 }, { "epoch": 1.6016178967342456, "grad_norm": 2.8338892459869385, "learning_rate": 4.995721620079864e-06, "loss": 0.1284, "step": 4010 }, { "epoch": 1.6056127034854688, "grad_norm": 2.8798747062683105, "learning_rate": 4.981460353679407e-06, "loss": 0.1308, "step": 4020 }, { "epoch": 1.6096075102366925, "grad_norm": 2.7863898277282715, "learning_rate": 4.96719908727895e-06, "loss": 0.0964, "step": 4030 }, { "epoch": 1.6136023169879157, "grad_norm": 2.8436176776885986, "learning_rate": 4.952937820878494e-06, "loss": 0.1061, "step": 4040 }, { "epoch": 1.617597123739139, "grad_norm": 2.8740711212158203, "learning_rate": 4.9386765544780376e-06, "loss": 0.1259, "step": 4050 }, { "epoch": 1.6215919304903625, "grad_norm": 4.055492877960205, "learning_rate": 4.924415288077582e-06, "loss": 0.1316, "step": 4060 }, { "epoch": 1.625586737241586, "grad_norm": 2.7835795879364014, "learning_rate": 4.910154021677126e-06, "loss": 0.1003, "step": 4070 }, { "epoch": 1.6295815439928094, "grad_norm": 3.262213706970215, "learning_rate": 4.895892755276669e-06, "loss": 0.1372, "step": 4080 }, { "epoch": 1.6335763507440326, "grad_norm": 3.1588211059570312, "learning_rate": 4.881631488876213e-06, "loss": 0.1119, "step": 4090 }, { "epoch": 1.6375711574952563, "grad_norm": 2.552180528640747, "learning_rate": 4.867370222475756e-06, "loss": 0.1109, "step": 4100 }, { "epoch": 1.6415659642464795, "grad_norm": 3.10109543800354, "learning_rate": 4.8531089560753e-06, "loss": 0.1176, "step": 4110 }, { "epoch": 1.6455607709977031, "grad_norm": 3.3522074222564697, "learning_rate": 4.838847689674844e-06, "loss": 0.1171, "step": 4120 }, { "epoch": 1.6495555777489264, "grad_norm": 2.9772942066192627, "learning_rate": 4.824586423274387e-06, "loss": 0.1317, "step": 4130 }, { "epoch": 1.6535503845001498, "grad_norm": 2.6188857555389404, "learning_rate": 4.810325156873931e-06, "loss": 0.1134, "step": 4140 }, { "epoch": 1.6575451912513732, "grad_norm": 2.481497287750244, "learning_rate": 4.796063890473474e-06, "loss": 0.125, "step": 4150 }, { "epoch": 1.6615399980025967, "grad_norm": 2.825188636779785, "learning_rate": 4.781802624073018e-06, "loss": 0.1351, "step": 4160 }, { "epoch": 1.66553480475382, "grad_norm": 2.593604326248169, "learning_rate": 4.767541357672562e-06, "loss": 0.1134, "step": 4170 }, { "epoch": 1.6695296115050433, "grad_norm": 2.1480395793914795, "learning_rate": 4.753280091272106e-06, "loss": 0.1348, "step": 4180 }, { "epoch": 1.673524418256267, "grad_norm": 3.1574795246124268, "learning_rate": 4.739018824871649e-06, "loss": 0.1134, "step": 4190 }, { "epoch": 1.6775192250074902, "grad_norm": 2.4995102882385254, "learning_rate": 4.724757558471192e-06, "loss": 0.1205, "step": 4200 }, { "epoch": 1.6815140317587138, "grad_norm": 2.8229737281799316, "learning_rate": 4.710496292070736e-06, "loss": 0.1256, "step": 4210 }, { "epoch": 1.685508838509937, "grad_norm": 2.9614689350128174, "learning_rate": 4.6962350256702795e-06, "loss": 0.1121, "step": 4220 }, { "epoch": 1.6895036452611605, "grad_norm": 2.974914789199829, "learning_rate": 4.681973759269824e-06, "loss": 0.1254, "step": 4230 }, { "epoch": 1.693498452012384, "grad_norm": 2.6984448432922363, "learning_rate": 4.667712492869368e-06, "loss": 0.1242, "step": 4240 }, { "epoch": 1.6974932587636073, "grad_norm": 2.75011944770813, "learning_rate": 4.653451226468911e-06, "loss": 0.103, "step": 4250 }, { "epoch": 1.7014880655148308, "grad_norm": 3.0294549465179443, "learning_rate": 4.639189960068454e-06, "loss": 0.1076, "step": 4260 }, { "epoch": 1.705482872266054, "grad_norm": 2.964902639389038, "learning_rate": 4.624928693667998e-06, "loss": 0.1376, "step": 4270 }, { "epoch": 1.7094776790172777, "grad_norm": 3.420753240585327, "learning_rate": 4.6106674272675416e-06, "loss": 0.1281, "step": 4280 }, { "epoch": 1.7134724857685009, "grad_norm": 3.5108609199523926, "learning_rate": 4.596406160867085e-06, "loss": 0.1409, "step": 4290 }, { "epoch": 1.7174672925197245, "grad_norm": 3.1926028728485107, "learning_rate": 4.582144894466629e-06, "loss": 0.1341, "step": 4300 }, { "epoch": 1.7214620992709477, "grad_norm": 4.24528694152832, "learning_rate": 4.567883628066173e-06, "loss": 0.1215, "step": 4310 }, { "epoch": 1.7254569060221712, "grad_norm": 2.712660074234009, "learning_rate": 4.553622361665716e-06, "loss": 0.1034, "step": 4320 }, { "epoch": 1.7294517127733946, "grad_norm": 2.88460373878479, "learning_rate": 4.53936109526526e-06, "loss": 0.1112, "step": 4330 }, { "epoch": 1.733446519524618, "grad_norm": 3.443690538406372, "learning_rate": 4.525099828864804e-06, "loss": 0.1108, "step": 4340 }, { "epoch": 1.7374413262758415, "grad_norm": 2.7443666458129883, "learning_rate": 4.510838562464347e-06, "loss": 0.1305, "step": 4350 }, { "epoch": 1.7414361330270647, "grad_norm": 2.413006067276001, "learning_rate": 4.496577296063891e-06, "loss": 0.1156, "step": 4360 }, { "epoch": 1.7454309397782883, "grad_norm": 2.9180493354797363, "learning_rate": 4.482316029663434e-06, "loss": 0.1088, "step": 4370 }, { "epoch": 1.7494257465295115, "grad_norm": 2.8736770153045654, "learning_rate": 4.468054763262978e-06, "loss": 0.119, "step": 4380 }, { "epoch": 1.7534205532807352, "grad_norm": 3.3292946815490723, "learning_rate": 4.4537934968625215e-06, "loss": 0.1131, "step": 4390 }, { "epoch": 1.7574153600319584, "grad_norm": 2.740265369415283, "learning_rate": 4.439532230462066e-06, "loss": 0.1218, "step": 4400 }, { "epoch": 1.7614101667831819, "grad_norm": 2.3319180011749268, "learning_rate": 4.425270964061609e-06, "loss": 0.1086, "step": 4410 }, { "epoch": 1.7654049735344053, "grad_norm": 4.39361572265625, "learning_rate": 4.411009697661153e-06, "loss": 0.1226, "step": 4420 }, { "epoch": 1.7693997802856287, "grad_norm": 2.681518316268921, "learning_rate": 4.396748431260696e-06, "loss": 0.1185, "step": 4430 }, { "epoch": 1.7733945870368522, "grad_norm": 3.028081178665161, "learning_rate": 4.38248716486024e-06, "loss": 0.1227, "step": 4440 }, { "epoch": 1.7773893937880754, "grad_norm": 3.1598165035247803, "learning_rate": 4.3682258984597835e-06, "loss": 0.1143, "step": 4450 }, { "epoch": 1.781384200539299, "grad_norm": 2.9632198810577393, "learning_rate": 4.353964632059327e-06, "loss": 0.1401, "step": 4460 }, { "epoch": 1.7853790072905222, "grad_norm": 2.7030882835388184, "learning_rate": 4.339703365658871e-06, "loss": 0.1039, "step": 4470 }, { "epoch": 1.789373814041746, "grad_norm": 2.8958992958068848, "learning_rate": 4.325442099258414e-06, "loss": 0.1131, "step": 4480 }, { "epoch": 1.793368620792969, "grad_norm": 3.6256399154663086, "learning_rate": 4.311180832857958e-06, "loss": 0.1458, "step": 4490 }, { "epoch": 1.7973634275441925, "grad_norm": 2.9515862464904785, "learning_rate": 4.296919566457502e-06, "loss": 0.1163, "step": 4500 }, { "epoch": 1.801358234295416, "grad_norm": 2.9707515239715576, "learning_rate": 4.2826583000570456e-06, "loss": 0.1163, "step": 4510 }, { "epoch": 1.8053530410466394, "grad_norm": 2.949582099914551, "learning_rate": 4.268397033656589e-06, "loss": 0.1192, "step": 4520 }, { "epoch": 1.8093478477978628, "grad_norm": 2.750103712081909, "learning_rate": 4.254135767256133e-06, "loss": 0.1144, "step": 4530 }, { "epoch": 1.813342654549086, "grad_norm": 3.391284704208374, "learning_rate": 4.239874500855676e-06, "loss": 0.1024, "step": 4540 }, { "epoch": 1.8173374613003097, "grad_norm": 2.8504669666290283, "learning_rate": 4.225613234455219e-06, "loss": 0.1189, "step": 4550 }, { "epoch": 1.821332268051533, "grad_norm": 2.996779203414917, "learning_rate": 4.2113519680547635e-06, "loss": 0.1287, "step": 4560 }, { "epoch": 1.8253270748027566, "grad_norm": 3.1329922676086426, "learning_rate": 4.197090701654308e-06, "loss": 0.1305, "step": 4570 }, { "epoch": 1.8293218815539798, "grad_norm": 2.596558094024658, "learning_rate": 4.182829435253851e-06, "loss": 0.1128, "step": 4580 }, { "epoch": 1.8333166883052032, "grad_norm": 3.028841733932495, "learning_rate": 4.168568168853395e-06, "loss": 0.1213, "step": 4590 }, { "epoch": 1.8373114950564267, "grad_norm": 2.55002760887146, "learning_rate": 4.154306902452938e-06, "loss": 0.1246, "step": 4600 }, { "epoch": 1.84130630180765, "grad_norm": 2.3285882472991943, "learning_rate": 4.1400456360524814e-06, "loss": 0.1109, "step": 4610 }, { "epoch": 1.8453011085588735, "grad_norm": 3.2481889724731445, "learning_rate": 4.1257843696520255e-06, "loss": 0.1445, "step": 4620 }, { "epoch": 1.8492959153100967, "grad_norm": 4.245287895202637, "learning_rate": 4.111523103251569e-06, "loss": 0.1235, "step": 4630 }, { "epoch": 1.8532907220613204, "grad_norm": 2.241886854171753, "learning_rate": 4.097261836851113e-06, "loss": 0.1164, "step": 4640 }, { "epoch": 1.8572855288125436, "grad_norm": 2.427218437194824, "learning_rate": 4.083000570450656e-06, "loss": 0.1082, "step": 4650 }, { "epoch": 1.8612803355637673, "grad_norm": 2.9869871139526367, "learning_rate": 4.0687393040502e-06, "loss": 0.1176, "step": 4660 }, { "epoch": 1.8652751423149905, "grad_norm": 2.828096866607666, "learning_rate": 4.054478037649744e-06, "loss": 0.1102, "step": 4670 }, { "epoch": 1.869269949066214, "grad_norm": 2.7090885639190674, "learning_rate": 4.0402167712492876e-06, "loss": 0.1153, "step": 4680 }, { "epoch": 1.8732647558174373, "grad_norm": 2.018324375152588, "learning_rate": 4.025955504848831e-06, "loss": 0.1061, "step": 4690 }, { "epoch": 1.8772595625686608, "grad_norm": 3.0430712699890137, "learning_rate": 4.011694238448375e-06, "loss": 0.1257, "step": 4700 }, { "epoch": 1.8812543693198842, "grad_norm": 2.6465933322906494, "learning_rate": 3.997432972047918e-06, "loss": 0.1019, "step": 4710 }, { "epoch": 1.8852491760711074, "grad_norm": 2.4254000186920166, "learning_rate": 3.983171705647461e-06, "loss": 0.106, "step": 4720 }, { "epoch": 1.889243982822331, "grad_norm": 2.948087453842163, "learning_rate": 3.9689104392470055e-06, "loss": 0.1185, "step": 4730 }, { "epoch": 1.8932387895735543, "grad_norm": 2.6019253730773926, "learning_rate": 3.954649172846549e-06, "loss": 0.1233, "step": 4740 }, { "epoch": 1.897233596324778, "grad_norm": 2.469536066055298, "learning_rate": 3.940387906446093e-06, "loss": 0.1061, "step": 4750 }, { "epoch": 1.9012284030760012, "grad_norm": 2.6648082733154297, "learning_rate": 3.926126640045637e-06, "loss": 0.1219, "step": 4760 }, { "epoch": 1.9052232098272246, "grad_norm": 3.086829662322998, "learning_rate": 3.91186537364518e-06, "loss": 0.1185, "step": 4770 }, { "epoch": 1.909218016578448, "grad_norm": 2.9754223823547363, "learning_rate": 3.897604107244723e-06, "loss": 0.1197, "step": 4780 }, { "epoch": 1.9132128233296715, "grad_norm": 2.877070665359497, "learning_rate": 3.8833428408442675e-06, "loss": 0.1089, "step": 4790 }, { "epoch": 1.917207630080895, "grad_norm": 3.260694742202759, "learning_rate": 3.869081574443811e-06, "loss": 0.099, "step": 4800 }, { "epoch": 1.921202436832118, "grad_norm": 2.4807591438293457, "learning_rate": 3.854820308043354e-06, "loss": 0.119, "step": 4810 }, { "epoch": 1.9251972435833418, "grad_norm": 2.8273167610168457, "learning_rate": 3.840559041642898e-06, "loss": 0.1181, "step": 4820 }, { "epoch": 1.929192050334565, "grad_norm": 2.2303531169891357, "learning_rate": 3.826297775242442e-06, "loss": 0.1189, "step": 4830 }, { "epoch": 1.9331868570857886, "grad_norm": 3.1909899711608887, "learning_rate": 3.8120365088419854e-06, "loss": 0.1197, "step": 4840 }, { "epoch": 1.9371816638370118, "grad_norm": 2.892822742462158, "learning_rate": 3.797775242441529e-06, "loss": 0.1326, "step": 4850 }, { "epoch": 1.9411764705882353, "grad_norm": 2.9792587757110596, "learning_rate": 3.7835139760410728e-06, "loss": 0.129, "step": 4860 }, { "epoch": 1.9451712773394587, "grad_norm": 3.0600063800811768, "learning_rate": 3.769252709640616e-06, "loss": 0.0974, "step": 4870 }, { "epoch": 1.9491660840906821, "grad_norm": 3.256072521209717, "learning_rate": 3.75499144324016e-06, "loss": 0.1048, "step": 4880 }, { "epoch": 1.9531608908419056, "grad_norm": 2.7048609256744385, "learning_rate": 3.740730176839704e-06, "loss": 0.1084, "step": 4890 }, { "epoch": 1.9571556975931288, "grad_norm": 4.142454147338867, "learning_rate": 3.726468910439247e-06, "loss": 0.1157, "step": 4900 }, { "epoch": 1.9611505043443525, "grad_norm": 2.8769376277923584, "learning_rate": 3.712207644038791e-06, "loss": 0.1275, "step": 4910 }, { "epoch": 1.9651453110955757, "grad_norm": 2.7974257469177246, "learning_rate": 3.6979463776383344e-06, "loss": 0.1335, "step": 4920 }, { "epoch": 1.9691401178467993, "grad_norm": 4.0073065757751465, "learning_rate": 3.683685111237878e-06, "loss": 0.1164, "step": 4930 }, { "epoch": 1.9731349245980225, "grad_norm": 3.3923537731170654, "learning_rate": 3.669423844837422e-06, "loss": 0.1065, "step": 4940 }, { "epoch": 1.977129731349246, "grad_norm": 2.8357832431793213, "learning_rate": 3.6551625784369654e-06, "loss": 0.109, "step": 4950 }, { "epoch": 1.9811245381004694, "grad_norm": 2.486527442932129, "learning_rate": 3.640901312036509e-06, "loss": 0.102, "step": 4960 }, { "epoch": 1.9851193448516928, "grad_norm": 3.080864667892456, "learning_rate": 3.6266400456360527e-06, "loss": 0.1291, "step": 4970 }, { "epoch": 1.9891141516029163, "grad_norm": 2.906073570251465, "learning_rate": 3.6123787792355964e-06, "loss": 0.0989, "step": 4980 }, { "epoch": 1.9931089583541395, "grad_norm": 2.806203603744507, "learning_rate": 3.5981175128351405e-06, "loss": 0.0955, "step": 4990 }, { "epoch": 1.9971037651053631, "grad_norm": 2.900861978530884, "learning_rate": 3.5838562464346838e-06, "loss": 0.1138, "step": 5000 }, { "epoch": 2.000798961350245, "grad_norm": 1.7900011539459229, "learning_rate": 3.5695949800342274e-06, "loss": 0.1161, "step": 5010 }, { "epoch": 2.004793768101468, "grad_norm": 1.8336867094039917, "learning_rate": 3.555333713633771e-06, "loss": 0.0626, "step": 5020 }, { "epoch": 2.0087885748526917, "grad_norm": 1.8405265808105469, "learning_rate": 3.5410724472333148e-06, "loss": 0.0558, "step": 5030 }, { "epoch": 2.012783381603915, "grad_norm": 1.876847743988037, "learning_rate": 3.526811180832858e-06, "loss": 0.0596, "step": 5040 }, { "epoch": 2.016778188355138, "grad_norm": 1.7897323369979858, "learning_rate": 3.512549914432402e-06, "loss": 0.0642, "step": 5050 }, { "epoch": 2.0207729951063618, "grad_norm": 1.995587706565857, "learning_rate": 3.4982886480319454e-06, "loss": 0.0723, "step": 5060 }, { "epoch": 2.024767801857585, "grad_norm": 2.014132022857666, "learning_rate": 3.484027381631489e-06, "loss": 0.0615, "step": 5070 }, { "epoch": 2.0287626086088086, "grad_norm": 2.271591901779175, "learning_rate": 3.469766115231033e-06, "loss": 0.0627, "step": 5080 }, { "epoch": 2.032757415360032, "grad_norm": 1.8247534036636353, "learning_rate": 3.4555048488305764e-06, "loss": 0.0516, "step": 5090 }, { "epoch": 2.0367522221112555, "grad_norm": 1.6670212745666504, "learning_rate": 3.44124358243012e-06, "loss": 0.0608, "step": 5100 }, { "epoch": 2.0407470288624787, "grad_norm": 1.9770992994308472, "learning_rate": 3.4269823160296637e-06, "loss": 0.0597, "step": 5110 }, { "epoch": 2.0447418356137024, "grad_norm": 1.9557619094848633, "learning_rate": 3.4127210496292074e-06, "loss": 0.0594, "step": 5120 }, { "epoch": 2.0487366423649256, "grad_norm": 1.9880411624908447, "learning_rate": 3.3984597832287506e-06, "loss": 0.0571, "step": 5130 }, { "epoch": 2.052731449116149, "grad_norm": 2.5329978466033936, "learning_rate": 3.3841985168282947e-06, "loss": 0.0553, "step": 5140 }, { "epoch": 2.0567262558673725, "grad_norm": 1.9590691328048706, "learning_rate": 3.3699372504278384e-06, "loss": 0.0558, "step": 5150 }, { "epoch": 2.0607210626185957, "grad_norm": 1.8811684846878052, "learning_rate": 3.3556759840273816e-06, "loss": 0.0607, "step": 5160 }, { "epoch": 2.0647158693698193, "grad_norm": 1.5391110181808472, "learning_rate": 3.3414147176269257e-06, "loss": 0.0611, "step": 5170 }, { "epoch": 2.0687106761210425, "grad_norm": 2.170163631439209, "learning_rate": 3.327153451226469e-06, "loss": 0.0648, "step": 5180 }, { "epoch": 2.072705482872266, "grad_norm": 1.9479721784591675, "learning_rate": 3.3128921848260127e-06, "loss": 0.0537, "step": 5190 }, { "epoch": 2.0767002896234894, "grad_norm": 1.9854282140731812, "learning_rate": 3.2986309184255568e-06, "loss": 0.0616, "step": 5200 }, { "epoch": 2.080695096374713, "grad_norm": 1.7945525646209717, "learning_rate": 3.2843696520251e-06, "loss": 0.0549, "step": 5210 }, { "epoch": 2.0846899031259363, "grad_norm": 2.2716691493988037, "learning_rate": 3.2701083856246437e-06, "loss": 0.0618, "step": 5220 }, { "epoch": 2.0886847098771595, "grad_norm": 1.8477457761764526, "learning_rate": 3.2558471192241873e-06, "loss": 0.0519, "step": 5230 }, { "epoch": 2.092679516628383, "grad_norm": 2.674288749694824, "learning_rate": 3.241585852823731e-06, "loss": 0.067, "step": 5240 }, { "epoch": 2.0966743233796064, "grad_norm": 2.320908784866333, "learning_rate": 3.2273245864232743e-06, "loss": 0.0618, "step": 5250 }, { "epoch": 2.10066913013083, "grad_norm": 3.661378860473633, "learning_rate": 3.2130633200228184e-06, "loss": 0.0772, "step": 5260 }, { "epoch": 2.104663936882053, "grad_norm": 2.278613805770874, "learning_rate": 3.198802053622362e-06, "loss": 0.0689, "step": 5270 }, { "epoch": 2.108658743633277, "grad_norm": 2.430885076522827, "learning_rate": 3.1845407872219053e-06, "loss": 0.057, "step": 5280 }, { "epoch": 2.1126535503845, "grad_norm": 2.0284667015075684, "learning_rate": 3.1702795208214494e-06, "loss": 0.0592, "step": 5290 }, { "epoch": 2.1166483571357237, "grad_norm": 2.1612558364868164, "learning_rate": 3.1560182544209926e-06, "loss": 0.0594, "step": 5300 }, { "epoch": 2.120643163886947, "grad_norm": 2.5965471267700195, "learning_rate": 3.1417569880205367e-06, "loss": 0.0621, "step": 5310 }, { "epoch": 2.12463797063817, "grad_norm": 2.1125738620758057, "learning_rate": 3.1274957216200804e-06, "loss": 0.062, "step": 5320 }, { "epoch": 2.128632777389394, "grad_norm": 2.013819932937622, "learning_rate": 3.1132344552196236e-06, "loss": 0.0675, "step": 5330 }, { "epoch": 2.132627584140617, "grad_norm": 2.894634485244751, "learning_rate": 3.0989731888191677e-06, "loss": 0.0546, "step": 5340 }, { "epoch": 2.1366223908918407, "grad_norm": 1.700340747833252, "learning_rate": 3.084711922418711e-06, "loss": 0.0496, "step": 5350 }, { "epoch": 2.140617197643064, "grad_norm": 1.88705313205719, "learning_rate": 3.0704506560182546e-06, "loss": 0.0524, "step": 5360 }, { "epoch": 2.1446120043942876, "grad_norm": 2.7680206298828125, "learning_rate": 3.0561893896177987e-06, "loss": 0.0631, "step": 5370 }, { "epoch": 2.1486068111455108, "grad_norm": 1.6159018278121948, "learning_rate": 3.041928123217342e-06, "loss": 0.051, "step": 5380 }, { "epoch": 2.1526016178967344, "grad_norm": 2.0894980430603027, "learning_rate": 3.0276668568168852e-06, "loss": 0.0674, "step": 5390 }, { "epoch": 2.1565964246479576, "grad_norm": 2.091074228286743, "learning_rate": 3.0134055904164293e-06, "loss": 0.0619, "step": 5400 }, { "epoch": 2.160591231399181, "grad_norm": 1.9509477615356445, "learning_rate": 2.999144324015973e-06, "loss": 0.053, "step": 5410 }, { "epoch": 2.1645860381504045, "grad_norm": 2.5904581546783447, "learning_rate": 2.9848830576155162e-06, "loss": 0.0527, "step": 5420 }, { "epoch": 2.1685808449016277, "grad_norm": 2.7561097145080566, "learning_rate": 2.9706217912150603e-06, "loss": 0.0603, "step": 5430 }, { "epoch": 2.1725756516528514, "grad_norm": 2.5652804374694824, "learning_rate": 2.9563605248146036e-06, "loss": 0.0563, "step": 5440 }, { "epoch": 2.1765704584040746, "grad_norm": 2.3183577060699463, "learning_rate": 2.9420992584141473e-06, "loss": 0.0539, "step": 5450 }, { "epoch": 2.1805652651552982, "grad_norm": 1.536807656288147, "learning_rate": 2.9278379920136913e-06, "loss": 0.0621, "step": 5460 }, { "epoch": 2.1845600719065215, "grad_norm": 1.710230827331543, "learning_rate": 2.9135767256132346e-06, "loss": 0.0692, "step": 5470 }, { "epoch": 2.188554878657745, "grad_norm": 1.5010180473327637, "learning_rate": 2.8993154592127783e-06, "loss": 0.0576, "step": 5480 }, { "epoch": 2.1925496854089683, "grad_norm": 2.431806802749634, "learning_rate": 2.885054192812322e-06, "loss": 0.0561, "step": 5490 }, { "epoch": 2.196544492160192, "grad_norm": 2.0688700675964355, "learning_rate": 2.8707929264118656e-06, "loss": 0.0642, "step": 5500 }, { "epoch": 2.200539298911415, "grad_norm": 2.40775728225708, "learning_rate": 2.856531660011409e-06, "loss": 0.0607, "step": 5510 }, { "epoch": 2.2045341056626384, "grad_norm": 2.3045880794525146, "learning_rate": 2.842270393610953e-06, "loss": 0.0551, "step": 5520 }, { "epoch": 2.208528912413862, "grad_norm": 2.8836209774017334, "learning_rate": 2.8280091272104966e-06, "loss": 0.0677, "step": 5530 }, { "epoch": 2.2125237191650853, "grad_norm": 1.8057619333267212, "learning_rate": 2.81374786081004e-06, "loss": 0.061, "step": 5540 }, { "epoch": 2.216518525916309, "grad_norm": 3.2008817195892334, "learning_rate": 2.799486594409584e-06, "loss": 0.062, "step": 5550 }, { "epoch": 2.220513332667532, "grad_norm": 1.8195422887802124, "learning_rate": 2.785225328009127e-06, "loss": 0.0557, "step": 5560 }, { "epoch": 2.224508139418756, "grad_norm": 2.326125383377075, "learning_rate": 2.770964061608671e-06, "loss": 0.0607, "step": 5570 }, { "epoch": 2.228502946169979, "grad_norm": 2.0502982139587402, "learning_rate": 2.756702795208215e-06, "loss": 0.0641, "step": 5580 }, { "epoch": 2.2324977529212022, "grad_norm": 1.6687091588974, "learning_rate": 2.7424415288077582e-06, "loss": 0.0611, "step": 5590 }, { "epoch": 2.236492559672426, "grad_norm": 3.5430431365966797, "learning_rate": 2.728180262407302e-06, "loss": 0.0667, "step": 5600 }, { "epoch": 2.240487366423649, "grad_norm": 2.443739175796509, "learning_rate": 2.7139189960068456e-06, "loss": 0.0575, "step": 5610 }, { "epoch": 2.2444821731748728, "grad_norm": 2.1115715503692627, "learning_rate": 2.6996577296063892e-06, "loss": 0.0563, "step": 5620 }, { "epoch": 2.248476979926096, "grad_norm": 2.4915425777435303, "learning_rate": 2.6853964632059333e-06, "loss": 0.0689, "step": 5630 }, { "epoch": 2.2524717866773196, "grad_norm": 2.3705291748046875, "learning_rate": 2.6711351968054766e-06, "loss": 0.0699, "step": 5640 }, { "epoch": 2.256466593428543, "grad_norm": 2.797736406326294, "learning_rate": 2.6568739304050202e-06, "loss": 0.0585, "step": 5650 }, { "epoch": 2.2604614001797665, "grad_norm": 1.975651741027832, "learning_rate": 2.642612664004564e-06, "loss": 0.0584, "step": 5660 }, { "epoch": 2.2644562069309897, "grad_norm": 1.7696720361709595, "learning_rate": 2.6283513976041076e-06, "loss": 0.0567, "step": 5670 }, { "epoch": 2.2684510136822134, "grad_norm": 1.649061918258667, "learning_rate": 2.614090131203651e-06, "loss": 0.0623, "step": 5680 }, { "epoch": 2.2724458204334366, "grad_norm": 1.612545371055603, "learning_rate": 2.599828864803195e-06, "loss": 0.0585, "step": 5690 }, { "epoch": 2.27644062718466, "grad_norm": 2.1259803771972656, "learning_rate": 2.5855675984027386e-06, "loss": 0.0614, "step": 5700 }, { "epoch": 2.2804354339358834, "grad_norm": 2.3355531692504883, "learning_rate": 2.571306332002282e-06, "loss": 0.0592, "step": 5710 }, { "epoch": 2.2844302406871067, "grad_norm": 2.033362865447998, "learning_rate": 2.557045065601826e-06, "loss": 0.0675, "step": 5720 }, { "epoch": 2.2884250474383303, "grad_norm": 2.148087978363037, "learning_rate": 2.542783799201369e-06, "loss": 0.0566, "step": 5730 }, { "epoch": 2.2924198541895535, "grad_norm": 1.9608763456344604, "learning_rate": 2.528522532800913e-06, "loss": 0.0617, "step": 5740 }, { "epoch": 2.296414660940777, "grad_norm": 2.103911876678467, "learning_rate": 2.514261266400457e-06, "loss": 0.0459, "step": 5750 }, { "epoch": 2.3004094676920004, "grad_norm": 2.7615857124328613, "learning_rate": 2.5e-06, "loss": 0.0572, "step": 5760 }, { "epoch": 2.3044042744432236, "grad_norm": 2.5205304622650146, "learning_rate": 2.485738733599544e-06, "loss": 0.0572, "step": 5770 }, { "epoch": 2.3083990811944473, "grad_norm": 2.4511005878448486, "learning_rate": 2.4714774671990875e-06, "loss": 0.0609, "step": 5780 }, { "epoch": 2.3123938879456705, "grad_norm": 3.9531912803649902, "learning_rate": 2.4572162007986312e-06, "loss": 0.056, "step": 5790 }, { "epoch": 2.316388694696894, "grad_norm": 2.1331589221954346, "learning_rate": 2.442954934398175e-06, "loss": 0.0618, "step": 5800 }, { "epoch": 2.3203835014481173, "grad_norm": 1.9293180704116821, "learning_rate": 2.428693667997718e-06, "loss": 0.051, "step": 5810 }, { "epoch": 2.324378308199341, "grad_norm": 2.831864833831787, "learning_rate": 2.414432401597262e-06, "loss": 0.0575, "step": 5820 }, { "epoch": 2.328373114950564, "grad_norm": 2.2617709636688232, "learning_rate": 2.400171135196806e-06, "loss": 0.051, "step": 5830 }, { "epoch": 2.332367921701788, "grad_norm": 2.002591371536255, "learning_rate": 2.3859098687963496e-06, "loss": 0.0572, "step": 5840 }, { "epoch": 2.336362728453011, "grad_norm": 2.327680826187134, "learning_rate": 2.371648602395893e-06, "loss": 0.0539, "step": 5850 }, { "epoch": 2.3403575352042347, "grad_norm": 2.135169267654419, "learning_rate": 2.3573873359954365e-06, "loss": 0.0617, "step": 5860 }, { "epoch": 2.344352341955458, "grad_norm": 2.423444986343384, "learning_rate": 2.34312606959498e-06, "loss": 0.0535, "step": 5870 }, { "epoch": 2.348347148706681, "grad_norm": 2.7082910537719727, "learning_rate": 2.328864803194524e-06, "loss": 0.0533, "step": 5880 }, { "epoch": 2.352341955457905, "grad_norm": 1.4340418577194214, "learning_rate": 2.3146035367940675e-06, "loss": 0.0548, "step": 5890 }, { "epoch": 2.356336762209128, "grad_norm": 2.5168190002441406, "learning_rate": 2.300342270393611e-06, "loss": 0.0624, "step": 5900 }, { "epoch": 2.3603315689603517, "grad_norm": 1.581122636795044, "learning_rate": 2.286081003993155e-06, "loss": 0.044, "step": 5910 }, { "epoch": 2.364326375711575, "grad_norm": 1.8260554075241089, "learning_rate": 2.2718197375926985e-06, "loss": 0.0531, "step": 5920 }, { "epoch": 2.3683211824627985, "grad_norm": 2.107790946960449, "learning_rate": 2.257558471192242e-06, "loss": 0.048, "step": 5930 }, { "epoch": 2.3723159892140218, "grad_norm": 2.019073009490967, "learning_rate": 2.2432972047917854e-06, "loss": 0.0561, "step": 5940 }, { "epoch": 2.376310795965245, "grad_norm": 2.3659348487854004, "learning_rate": 2.229035938391329e-06, "loss": 0.0539, "step": 5950 }, { "epoch": 2.3803056027164686, "grad_norm": 1.6513670682907104, "learning_rate": 2.214774671990873e-06, "loss": 0.0502, "step": 5960 }, { "epoch": 2.384300409467692, "grad_norm": 2.1922340393066406, "learning_rate": 2.2005134055904164e-06, "loss": 0.049, "step": 5970 }, { "epoch": 2.3882952162189155, "grad_norm": 2.5353798866271973, "learning_rate": 2.18625213918996e-06, "loss": 0.0532, "step": 5980 }, { "epoch": 2.3922900229701387, "grad_norm": 2.41393780708313, "learning_rate": 2.171990872789504e-06, "loss": 0.0552, "step": 5990 }, { "epoch": 2.3962848297213624, "grad_norm": 1.9796687364578247, "learning_rate": 2.1577296063890475e-06, "loss": 0.0643, "step": 6000 }, { "epoch": 2.3962848297213624, "eval_loss": 0.07063105702400208, "eval_runtime": 8582.2825, "eval_samples_per_second": 2.333, "eval_steps_per_second": 0.292, "eval_wer": 15.794018626207862, "step": 6000 } ], "logging_steps": 10, "max_steps": 7512, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 2000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 1.9589898836901888e+20, "train_batch_size": 8, "trial_name": null, "trial_params": null }