{ "best_metric": 0.3763063217406202, "best_model_checkpoint": "d:\\DataTicon\\Whisper-Khmer-Small\\whisper-khmer\\outputs\\whisper-small-khmer\\checkpoint-5000", "epoch": 2.995805871779509, "eval_steps": 1000, "global_step": 5000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.005991611743559017, "grad_norm": 22.960840225219727, "learning_rate": 7.000000000000001e-06, "loss": 3.0494, "step": 10 }, { "epoch": 0.011983223487118035, "grad_norm": 15.023651123046875, "learning_rate": 1.7000000000000003e-05, "loss": 2.1354, "step": 20 }, { "epoch": 0.017974835230677052, "grad_norm": 20.588504791259766, "learning_rate": 2.7000000000000002e-05, "loss": 1.7606, "step": 30 }, { "epoch": 0.02396644697423607, "grad_norm": 12.844827651977539, "learning_rate": 3.7e-05, "loss": 1.6323, "step": 40 }, { "epoch": 0.029958058717795086, "grad_norm": 21.74739646911621, "learning_rate": 4.7e-05, "loss": 1.5285, "step": 50 }, { "epoch": 0.035949670461354104, "grad_norm": 14.430631637573242, "learning_rate": 4.992939277788985e-05, "loss": 1.4681, "step": 60 }, { "epoch": 0.041941282204913125, "grad_norm": 16.150493621826172, "learning_rate": 4.98285253177325e-05, "loss": 1.4109, "step": 70 }, { "epoch": 0.04793289394847214, "grad_norm": 14.220664024353027, "learning_rate": 4.972765785757515e-05, "loss": 1.3271, "step": 80 }, { "epoch": 0.05392450569203116, "grad_norm": 10.489856719970703, "learning_rate": 4.962679039741779e-05, "loss": 1.2089, "step": 90 }, { "epoch": 0.05991611743559017, "grad_norm": 15.242402076721191, "learning_rate": 4.952592293726044e-05, "loss": 1.2212, "step": 100 }, { "epoch": 0.0659077291791492, "grad_norm": 8.778143882751465, "learning_rate": 4.942505547710309e-05, "loss": 1.0803, "step": 110 }, { "epoch": 0.07189934092270821, "grad_norm": 8.006437301635742, "learning_rate": 4.932418801694574e-05, "loss": 1.0374, "step": 120 }, { "epoch": 0.07789095266626722, "grad_norm": 10.208110809326172, "learning_rate": 4.922332055678838e-05, "loss": 0.9244, "step": 130 }, { "epoch": 0.08388256440982625, "grad_norm": 11.258834838867188, "learning_rate": 4.9122453096631024e-05, "loss": 0.7758, "step": 140 }, { "epoch": 0.08987417615338526, "grad_norm": 11.1365327835083, "learning_rate": 4.902158563647368e-05, "loss": 0.7568, "step": 150 }, { "epoch": 0.09586578789694428, "grad_norm": 9.645950317382812, "learning_rate": 4.892071817631632e-05, "loss": 0.666, "step": 160 }, { "epoch": 0.10185739964050329, "grad_norm": 7.856279373168945, "learning_rate": 4.881985071615897e-05, "loss": 0.6093, "step": 170 }, { "epoch": 0.10784901138406232, "grad_norm": 8.624720573425293, "learning_rate": 4.871898325600161e-05, "loss": 0.611, "step": 180 }, { "epoch": 0.11384062312762133, "grad_norm": 7.039414882659912, "learning_rate": 4.861811579584427e-05, "loss": 0.5455, "step": 190 }, { "epoch": 0.11983223487118035, "grad_norm": 8.36270523071289, "learning_rate": 4.851724833568691e-05, "loss": 0.5192, "step": 200 }, { "epoch": 0.12582384661473936, "grad_norm": 6.983541488647461, "learning_rate": 4.841638087552955e-05, "loss": 0.5046, "step": 210 }, { "epoch": 0.1318154583582984, "grad_norm": 6.642160415649414, "learning_rate": 4.83155134153722e-05, "loss": 0.5473, "step": 220 }, { "epoch": 0.1378070701018574, "grad_norm": 8.442938804626465, "learning_rate": 4.821464595521485e-05, "loss": 0.5307, "step": 230 }, { "epoch": 0.14379868184541642, "grad_norm": 5.670754432678223, "learning_rate": 4.81137784950575e-05, "loss": 0.4801, "step": 240 }, { "epoch": 0.14979029358897544, "grad_norm": 7.541809558868408, "learning_rate": 4.801291103490014e-05, "loss": 0.4827, "step": 250 }, { "epoch": 0.15578190533253444, "grad_norm": 4.786922454833984, "learning_rate": 4.791204357474279e-05, "loss": 0.4322, "step": 260 }, { "epoch": 0.16177351707609347, "grad_norm": 6.316529750823975, "learning_rate": 4.781117611458544e-05, "loss": 0.4823, "step": 270 }, { "epoch": 0.1677651288196525, "grad_norm": 6.494561195373535, "learning_rate": 4.771030865442808e-05, "loss": 0.4177, "step": 280 }, { "epoch": 0.1737567405632115, "grad_norm": 6.1072998046875, "learning_rate": 4.760944119427073e-05, "loss": 0.4238, "step": 290 }, { "epoch": 0.17974835230677053, "grad_norm": 7.656487464904785, "learning_rate": 4.750857373411338e-05, "loss": 0.3656, "step": 300 }, { "epoch": 0.18573996405032953, "grad_norm": 4.906504154205322, "learning_rate": 4.740770627395603e-05, "loss": 0.4198, "step": 310 }, { "epoch": 0.19173157579388855, "grad_norm": 5.329479694366455, "learning_rate": 4.730683881379867e-05, "loss": 0.3753, "step": 320 }, { "epoch": 0.19772318753744758, "grad_norm": 4.617819309234619, "learning_rate": 4.720597135364131e-05, "loss": 0.3791, "step": 330 }, { "epoch": 0.20371479928100658, "grad_norm": 6.305775165557861, "learning_rate": 4.710510389348396e-05, "loss": 0.3902, "step": 340 }, { "epoch": 0.2097064110245656, "grad_norm": 3.827413320541382, "learning_rate": 4.700423643332661e-05, "loss": 0.3879, "step": 350 }, { "epoch": 0.21569802276812464, "grad_norm": 5.891330242156982, "learning_rate": 4.690336897316926e-05, "loss": 0.3855, "step": 360 }, { "epoch": 0.22168963451168364, "grad_norm": 4.497730731964111, "learning_rate": 4.68025015130119e-05, "loss": 0.3275, "step": 370 }, { "epoch": 0.22768124625524266, "grad_norm": 3.898587703704834, "learning_rate": 4.670163405285455e-05, "loss": 0.3645, "step": 380 }, { "epoch": 0.23367285799880166, "grad_norm": 4.55585241317749, "learning_rate": 4.66007665926972e-05, "loss": 0.3683, "step": 390 }, { "epoch": 0.2396644697423607, "grad_norm": 6.788579940795898, "learning_rate": 4.649989913253984e-05, "loss": 0.3637, "step": 400 }, { "epoch": 0.24565608148591972, "grad_norm": 5.430655002593994, "learning_rate": 4.639903167238249e-05, "loss": 0.3766, "step": 410 }, { "epoch": 0.2516476932294787, "grad_norm": 6.993104934692383, "learning_rate": 4.629816421222514e-05, "loss": 0.3482, "step": 420 }, { "epoch": 0.2576393049730378, "grad_norm": 4.983759880065918, "learning_rate": 4.619729675206779e-05, "loss": 0.3489, "step": 430 }, { "epoch": 0.2636309167165968, "grad_norm": 4.883998394012451, "learning_rate": 4.609642929191043e-05, "loss": 0.3228, "step": 440 }, { "epoch": 0.2696225284601558, "grad_norm": 4.701789855957031, "learning_rate": 4.599556183175308e-05, "loss": 0.3075, "step": 450 }, { "epoch": 0.2756141402037148, "grad_norm": 6.912998676300049, "learning_rate": 4.589469437159573e-05, "loss": 0.3267, "step": 460 }, { "epoch": 0.28160575194727383, "grad_norm": 5.111138343811035, "learning_rate": 4.579382691143837e-05, "loss": 0.3471, "step": 470 }, { "epoch": 0.28759736369083283, "grad_norm": 4.039860725402832, "learning_rate": 4.569295945128102e-05, "loss": 0.2972, "step": 480 }, { "epoch": 0.29358897543439183, "grad_norm": 4.554994106292725, "learning_rate": 4.559209199112366e-05, "loss": 0.3344, "step": 490 }, { "epoch": 0.2995805871779509, "grad_norm": 5.8619184494018555, "learning_rate": 4.549122453096631e-05, "loss": 0.3375, "step": 500 }, { "epoch": 0.3055721989215099, "grad_norm": 4.201863765716553, "learning_rate": 4.539035707080896e-05, "loss": 0.2924, "step": 510 }, { "epoch": 0.3115638106650689, "grad_norm": 4.574302673339844, "learning_rate": 4.528948961065161e-05, "loss": 0.3193, "step": 520 }, { "epoch": 0.31755542240862794, "grad_norm": 5.951114654541016, "learning_rate": 4.518862215049425e-05, "loss": 0.3522, "step": 530 }, { "epoch": 0.32354703415218694, "grad_norm": 5.023397922515869, "learning_rate": 4.50877546903369e-05, "loss": 0.3144, "step": 540 }, { "epoch": 0.32953864589574594, "grad_norm": 4.05376672744751, "learning_rate": 4.498688723017955e-05, "loss": 0.3183, "step": 550 }, { "epoch": 0.335530257639305, "grad_norm": 3.973388910293579, "learning_rate": 4.488601977002219e-05, "loss": 0.289, "step": 560 }, { "epoch": 0.341521869382864, "grad_norm": 4.413971424102783, "learning_rate": 4.478515230986484e-05, "loss": 0.3045, "step": 570 }, { "epoch": 0.347513481126423, "grad_norm": 4.1166791915893555, "learning_rate": 4.468428484970749e-05, "loss": 0.2692, "step": 580 }, { "epoch": 0.35350509286998205, "grad_norm": 4.370257377624512, "learning_rate": 4.458341738955013e-05, "loss": 0.3231, "step": 590 }, { "epoch": 0.35949670461354105, "grad_norm": 2.8509249687194824, "learning_rate": 4.448254992939278e-05, "loss": 0.2658, "step": 600 }, { "epoch": 0.36548831635710005, "grad_norm": 4.245826244354248, "learning_rate": 4.438168246923542e-05, "loss": 0.2796, "step": 610 }, { "epoch": 0.37147992810065905, "grad_norm": 2.2622575759887695, "learning_rate": 4.428081500907808e-05, "loss": 0.2793, "step": 620 }, { "epoch": 0.3774715398442181, "grad_norm": 5.839344024658203, "learning_rate": 4.417994754892072e-05, "loss": 0.2948, "step": 630 }, { "epoch": 0.3834631515877771, "grad_norm": 4.588991165161133, "learning_rate": 4.407908008876337e-05, "loss": 0.2899, "step": 640 }, { "epoch": 0.3894547633313361, "grad_norm": 4.319995880126953, "learning_rate": 4.397821262860601e-05, "loss": 0.2889, "step": 650 }, { "epoch": 0.39544637507489516, "grad_norm": 4.287087440490723, "learning_rate": 4.387734516844866e-05, "loss": 0.2961, "step": 660 }, { "epoch": 0.40143798681845416, "grad_norm": 2.974928855895996, "learning_rate": 4.377647770829131e-05, "loss": 0.287, "step": 670 }, { "epoch": 0.40742959856201316, "grad_norm": 3.6927688121795654, "learning_rate": 4.367561024813395e-05, "loss": 0.3006, "step": 680 }, { "epoch": 0.4134212103055722, "grad_norm": 2.942836284637451, "learning_rate": 4.35747427879766e-05, "loss": 0.2678, "step": 690 }, { "epoch": 0.4194128220491312, "grad_norm": 3.953315019607544, "learning_rate": 4.347387532781925e-05, "loss": 0.3063, "step": 700 }, { "epoch": 0.4254044337926902, "grad_norm": 3.223515510559082, "learning_rate": 4.33730078676619e-05, "loss": 0.2575, "step": 710 }, { "epoch": 0.4313960455362493, "grad_norm": 4.100767612457275, "learning_rate": 4.327214040750454e-05, "loss": 0.2581, "step": 720 }, { "epoch": 0.4373876572798083, "grad_norm": 3.5469157695770264, "learning_rate": 4.317127294734718e-05, "loss": 0.2511, "step": 730 }, { "epoch": 0.4433792690233673, "grad_norm": 3.4238710403442383, "learning_rate": 4.307040548718984e-05, "loss": 0.2835, "step": 740 }, { "epoch": 0.44937088076692633, "grad_norm": 3.2563302516937256, "learning_rate": 4.296953802703248e-05, "loss": 0.2559, "step": 750 }, { "epoch": 0.45536249251048533, "grad_norm": 3.1551856994628906, "learning_rate": 4.286867056687513e-05, "loss": 0.2367, "step": 760 }, { "epoch": 0.46135410425404433, "grad_norm": 3.8980541229248047, "learning_rate": 4.276780310671777e-05, "loss": 0.2376, "step": 770 }, { "epoch": 0.46734571599760333, "grad_norm": 5.483028888702393, "learning_rate": 4.266693564656043e-05, "loss": 0.2515, "step": 780 }, { "epoch": 0.4733373277411624, "grad_norm": 5.038168907165527, "learning_rate": 4.256606818640307e-05, "loss": 0.258, "step": 790 }, { "epoch": 0.4793289394847214, "grad_norm": 4.069836616516113, "learning_rate": 4.246520072624571e-05, "loss": 0.2717, "step": 800 }, { "epoch": 0.4853205512282804, "grad_norm": 3.8578696250915527, "learning_rate": 4.236433326608836e-05, "loss": 0.2353, "step": 810 }, { "epoch": 0.49131216297183944, "grad_norm": 3.4313695430755615, "learning_rate": 4.226346580593101e-05, "loss": 0.25, "step": 820 }, { "epoch": 0.49730377471539844, "grad_norm": 4.114861488342285, "learning_rate": 4.216259834577366e-05, "loss": 0.2516, "step": 830 }, { "epoch": 0.5032953864589574, "grad_norm": 3.718224048614502, "learning_rate": 4.20617308856163e-05, "loss": 0.2438, "step": 840 }, { "epoch": 0.5092869982025164, "grad_norm": 4.2352070808410645, "learning_rate": 4.196086342545895e-05, "loss": 0.2427, "step": 850 }, { "epoch": 0.5152786099460755, "grad_norm": 3.1277523040771484, "learning_rate": 4.18599959653016e-05, "loss": 0.2577, "step": 860 }, { "epoch": 0.5212702216896345, "grad_norm": 3.432342767715454, "learning_rate": 4.175912850514424e-05, "loss": 0.2632, "step": 870 }, { "epoch": 0.5272618334331935, "grad_norm": 2.9332733154296875, "learning_rate": 4.165826104498689e-05, "loss": 0.2466, "step": 880 }, { "epoch": 0.5332534451767525, "grad_norm": 3.802907705307007, "learning_rate": 4.155739358482953e-05, "loss": 0.2335, "step": 890 }, { "epoch": 0.5392450569203115, "grad_norm": 2.909348726272583, "learning_rate": 4.145652612467219e-05, "loss": 0.249, "step": 900 }, { "epoch": 0.5452366686638705, "grad_norm": 3.496734142303467, "learning_rate": 4.135565866451483e-05, "loss": 0.2347, "step": 910 }, { "epoch": 0.5512282804074295, "grad_norm": 3.264517307281494, "learning_rate": 4.125479120435747e-05, "loss": 0.2367, "step": 920 }, { "epoch": 0.5572198921509887, "grad_norm": 3.3015294075012207, "learning_rate": 4.115392374420012e-05, "loss": 0.2285, "step": 930 }, { "epoch": 0.5632115038945477, "grad_norm": 3.386230945587158, "learning_rate": 4.105305628404277e-05, "loss": 0.2373, "step": 940 }, { "epoch": 0.5692031156381067, "grad_norm": 4.057402610778809, "learning_rate": 4.095218882388542e-05, "loss": 0.2336, "step": 950 }, { "epoch": 0.5751947273816657, "grad_norm": 3.4949593544006348, "learning_rate": 4.085132136372806e-05, "loss": 0.214, "step": 960 }, { "epoch": 0.5811863391252247, "grad_norm": 2.452944278717041, "learning_rate": 4.075045390357071e-05, "loss": 0.2142, "step": 970 }, { "epoch": 0.5871779508687837, "grad_norm": 4.477965831756592, "learning_rate": 4.064958644341336e-05, "loss": 0.2443, "step": 980 }, { "epoch": 0.5931695626123428, "grad_norm": 3.8095345497131348, "learning_rate": 4.0548718983256e-05, "loss": 0.2074, "step": 990 }, { "epoch": 0.5991611743559018, "grad_norm": 3.586312770843506, "learning_rate": 4.044785152309865e-05, "loss": 0.2352, "step": 1000 }, { "epoch": 0.5991611743559018, "eval_loss": 0.3941349983215332, "eval_runtime": 1145.4733, "eval_samples_per_second": 1.238, "eval_steps_per_second": 0.078, "eval_wer": 0.47164639369539146, "step": 1000 }, { "epoch": 0.6051527860994608, "grad_norm": 3.0073084831237793, "learning_rate": 4.03469840629413e-05, "loss": 0.1915, "step": 1010 }, { "epoch": 0.6111443978430198, "grad_norm": 3.476414680480957, "learning_rate": 4.024611660278395e-05, "loss": 0.2131, "step": 1020 }, { "epoch": 0.6171360095865788, "grad_norm": 3.182587146759033, "learning_rate": 4.015533588864232e-05, "loss": 0.2186, "step": 1030 }, { "epoch": 0.6231276213301378, "grad_norm": 2.873277187347412, "learning_rate": 4.005446842848497e-05, "loss": 0.2182, "step": 1040 }, { "epoch": 0.6291192330736968, "grad_norm": 3.587416172027588, "learning_rate": 3.995360096832762e-05, "loss": 0.2358, "step": 1050 }, { "epoch": 0.6351108448172559, "grad_norm": 2.8519115447998047, "learning_rate": 3.9852733508170263e-05, "loss": 0.202, "step": 1060 }, { "epoch": 0.6411024565608149, "grad_norm": 4.239020824432373, "learning_rate": 3.975186604801291e-05, "loss": 0.252, "step": 1070 }, { "epoch": 0.6470940683043739, "grad_norm": 3.040431499481201, "learning_rate": 3.965099858785556e-05, "loss": 0.2191, "step": 1080 }, { "epoch": 0.6530856800479329, "grad_norm": 4.314202785491943, "learning_rate": 3.955013112769821e-05, "loss": 0.2341, "step": 1090 }, { "epoch": 0.6590772917914919, "grad_norm": 4.510042190551758, "learning_rate": 3.944926366754085e-05, "loss": 0.2149, "step": 1100 }, { "epoch": 0.6650689035350509, "grad_norm": 3.784174919128418, "learning_rate": 3.9348396207383495e-05, "loss": 0.2413, "step": 1110 }, { "epoch": 0.67106051527861, "grad_norm": 3.5287516117095947, "learning_rate": 3.924752874722615e-05, "loss": 0.2268, "step": 1120 }, { "epoch": 0.677052127022169, "grad_norm": 3.4882707595825195, "learning_rate": 3.914666128706879e-05, "loss": 0.2451, "step": 1130 }, { "epoch": 0.683043738765728, "grad_norm": 2.9120218753814697, "learning_rate": 3.904579382691144e-05, "loss": 0.2256, "step": 1140 }, { "epoch": 0.689035350509287, "grad_norm": 4.245069980621338, "learning_rate": 3.8944926366754084e-05, "loss": 0.2168, "step": 1150 }, { "epoch": 0.695026962252846, "grad_norm": 2.571373701095581, "learning_rate": 3.884405890659674e-05, "loss": 0.2323, "step": 1160 }, { "epoch": 0.701018573996405, "grad_norm": 3.3030152320861816, "learning_rate": 3.874319144643938e-05, "loss": 0.2123, "step": 1170 }, { "epoch": 0.7070101857399641, "grad_norm": 3.8424742221832275, "learning_rate": 3.8642323986282024e-05, "loss": 0.2239, "step": 1180 }, { "epoch": 0.7130017974835231, "grad_norm": 2.93353271484375, "learning_rate": 3.854145652612467e-05, "loss": 0.2153, "step": 1190 }, { "epoch": 0.7189934092270821, "grad_norm": 5.56205940246582, "learning_rate": 3.844058906596732e-05, "loss": 0.2362, "step": 1200 }, { "epoch": 0.7249850209706411, "grad_norm": 3.792433023452759, "learning_rate": 3.833972160580997e-05, "loss": 0.2338, "step": 1210 }, { "epoch": 0.7309766327142001, "grad_norm": 3.8626973628997803, "learning_rate": 3.823885414565261e-05, "loss": 0.2245, "step": 1220 }, { "epoch": 0.7369682444577591, "grad_norm": 4.392445087432861, "learning_rate": 3.813798668549526e-05, "loss": 0.2469, "step": 1230 }, { "epoch": 0.7429598562013181, "grad_norm": 2.464857578277588, "learning_rate": 3.803711922533791e-05, "loss": 0.2249, "step": 1240 }, { "epoch": 0.7489514679448772, "grad_norm": 4.464147567749023, "learning_rate": 3.793625176518055e-05, "loss": 0.239, "step": 1250 }, { "epoch": 0.7549430796884362, "grad_norm": 3.0969793796539307, "learning_rate": 3.78353843050232e-05, "loss": 0.2422, "step": 1260 }, { "epoch": 0.7609346914319952, "grad_norm": 2.4737939834594727, "learning_rate": 3.7734516844865844e-05, "loss": 0.2013, "step": 1270 }, { "epoch": 0.7669263031755542, "grad_norm": 3.358137845993042, "learning_rate": 3.76336493847085e-05, "loss": 0.1985, "step": 1280 }, { "epoch": 0.7729179149191132, "grad_norm": 2.7187469005584717, "learning_rate": 3.753278192455114e-05, "loss": 0.2036, "step": 1290 }, { "epoch": 0.7789095266626722, "grad_norm": 2.8770973682403564, "learning_rate": 3.7431914464393784e-05, "loss": 0.1973, "step": 1300 }, { "epoch": 0.7849011384062313, "grad_norm": 3.322476387023926, "learning_rate": 3.733104700423643e-05, "loss": 0.2165, "step": 1310 }, { "epoch": 0.7908927501497903, "grad_norm": 3.7058589458465576, "learning_rate": 3.723017954407908e-05, "loss": 0.2101, "step": 1320 }, { "epoch": 0.7968843618933493, "grad_norm": 3.665559768676758, "learning_rate": 3.712931208392173e-05, "loss": 0.2365, "step": 1330 }, { "epoch": 0.8028759736369083, "grad_norm": 2.0677387714385986, "learning_rate": 3.702844462376437e-05, "loss": 0.2019, "step": 1340 }, { "epoch": 0.8088675853804673, "grad_norm": 3.7458584308624268, "learning_rate": 3.692757716360702e-05, "loss": 0.1937, "step": 1350 }, { "epoch": 0.8148591971240263, "grad_norm": 3.561161994934082, "learning_rate": 3.682670970344967e-05, "loss": 0.2034, "step": 1360 }, { "epoch": 0.8208508088675854, "grad_norm": 3.4886398315429688, "learning_rate": 3.672584224329231e-05, "loss": 0.2074, "step": 1370 }, { "epoch": 0.8268424206111444, "grad_norm": 2.348358631134033, "learning_rate": 3.662497478313496e-05, "loss": 0.2011, "step": 1380 }, { "epoch": 0.8328340323547034, "grad_norm": 4.720375061035156, "learning_rate": 3.652410732297761e-05, "loss": 0.2075, "step": 1390 }, { "epoch": 0.8388256440982624, "grad_norm": 2.795454740524292, "learning_rate": 3.642323986282026e-05, "loss": 0.2136, "step": 1400 }, { "epoch": 0.8448172558418214, "grad_norm": 1.7271804809570312, "learning_rate": 3.63223724026629e-05, "loss": 0.2004, "step": 1410 }, { "epoch": 0.8508088675853804, "grad_norm": 3.5545125007629395, "learning_rate": 3.622150494250555e-05, "loss": 0.1925, "step": 1420 }, { "epoch": 0.8568004793289394, "grad_norm": 2.8007822036743164, "learning_rate": 3.61206374823482e-05, "loss": 0.2042, "step": 1430 }, { "epoch": 0.8627920910724985, "grad_norm": 2.51857852935791, "learning_rate": 3.601977002219084e-05, "loss": 0.1843, "step": 1440 }, { "epoch": 0.8687837028160575, "grad_norm": 2.5232300758361816, "learning_rate": 3.591890256203349e-05, "loss": 0.1921, "step": 1450 }, { "epoch": 0.8747753145596165, "grad_norm": 3.585514783859253, "learning_rate": 3.581803510187613e-05, "loss": 0.2168, "step": 1460 }, { "epoch": 0.8807669263031755, "grad_norm": 3.1116299629211426, "learning_rate": 3.571716764171878e-05, "loss": 0.2088, "step": 1470 }, { "epoch": 0.8867585380467345, "grad_norm": 2.461395263671875, "learning_rate": 3.561630018156143e-05, "loss": 0.1727, "step": 1480 }, { "epoch": 0.8927501497902935, "grad_norm": 3.6343369483947754, "learning_rate": 3.551543272140408e-05, "loss": 0.2092, "step": 1490 }, { "epoch": 0.8987417615338527, "grad_norm": 3.2808003425598145, "learning_rate": 3.541456526124672e-05, "loss": 0.2121, "step": 1500 }, { "epoch": 0.9047333732774117, "grad_norm": 2.9965360164642334, "learning_rate": 3.531369780108937e-05, "loss": 0.1769, "step": 1510 }, { "epoch": 0.9107249850209707, "grad_norm": 3.1718783378601074, "learning_rate": 3.521283034093202e-05, "loss": 0.2018, "step": 1520 }, { "epoch": 0.9167165967645297, "grad_norm": 3.313483238220215, "learning_rate": 3.511196288077466e-05, "loss": 0.2089, "step": 1530 }, { "epoch": 0.9227082085080887, "grad_norm": 4.256625652313232, "learning_rate": 3.501109542061731e-05, "loss": 0.2107, "step": 1540 }, { "epoch": 0.9286998202516477, "grad_norm": 4.17952299118042, "learning_rate": 3.491022796045996e-05, "loss": 0.1871, "step": 1550 }, { "epoch": 0.9346914319952067, "grad_norm": 2.3444721698760986, "learning_rate": 3.48093605003026e-05, "loss": 0.1679, "step": 1560 }, { "epoch": 0.9406830437387658, "grad_norm": 3.664823293685913, "learning_rate": 3.470849304014525e-05, "loss": 0.1886, "step": 1570 }, { "epoch": 0.9466746554823248, "grad_norm": 1.821067214012146, "learning_rate": 3.460762557998789e-05, "loss": 0.1583, "step": 1580 }, { "epoch": 0.9526662672258838, "grad_norm": 4.162514686584473, "learning_rate": 3.450675811983055e-05, "loss": 0.1845, "step": 1590 }, { "epoch": 0.9586578789694428, "grad_norm": 3.205587387084961, "learning_rate": 3.440589065967319e-05, "loss": 0.2166, "step": 1600 }, { "epoch": 0.9646494907130018, "grad_norm": 2.805971622467041, "learning_rate": 3.430502319951584e-05, "loss": 0.1621, "step": 1610 }, { "epoch": 0.9706411024565608, "grad_norm": 2.5269038677215576, "learning_rate": 3.420415573935848e-05, "loss": 0.1698, "step": 1620 }, { "epoch": 0.9766327142001199, "grad_norm": 2.486302375793457, "learning_rate": 3.410328827920113e-05, "loss": 0.2071, "step": 1630 }, { "epoch": 0.9826243259436789, "grad_norm": 3.5762979984283447, "learning_rate": 3.400242081904378e-05, "loss": 0.182, "step": 1640 }, { "epoch": 0.9886159376872379, "grad_norm": 2.278040885925293, "learning_rate": 3.390155335888642e-05, "loss": 0.1826, "step": 1650 }, { "epoch": 0.9946075494307969, "grad_norm": 2.813933849334717, "learning_rate": 3.380068589872907e-05, "loss": 0.1842, "step": 1660 }, { "epoch": 1.0005991611743559, "grad_norm": 2.1713950634002686, "learning_rate": 3.369981843857172e-05, "loss": 0.1858, "step": 1670 }, { "epoch": 1.0065907729179149, "grad_norm": 2.670079469680786, "learning_rate": 3.359895097841437e-05, "loss": 0.1146, "step": 1680 }, { "epoch": 1.0125823846614739, "grad_norm": 1.8872461318969727, "learning_rate": 3.349808351825701e-05, "loss": 0.1229, "step": 1690 }, { "epoch": 1.0185739964050329, "grad_norm": 2.9682705402374268, "learning_rate": 3.339721605809965e-05, "loss": 0.1212, "step": 1700 }, { "epoch": 1.0245656081485919, "grad_norm": 2.689821720123291, "learning_rate": 3.329634859794231e-05, "loss": 0.1404, "step": 1710 }, { "epoch": 1.030557219892151, "grad_norm": 2.6816999912261963, "learning_rate": 3.319548113778495e-05, "loss": 0.144, "step": 1720 }, { "epoch": 1.03654883163571, "grad_norm": 2.6932919025421143, "learning_rate": 3.30946136776276e-05, "loss": 0.135, "step": 1730 }, { "epoch": 1.042540443379269, "grad_norm": 2.344484567642212, "learning_rate": 3.299374621747024e-05, "loss": 0.1101, "step": 1740 }, { "epoch": 1.048532055122828, "grad_norm": 3.399644613265991, "learning_rate": 3.28928787573129e-05, "loss": 0.1328, "step": 1750 }, { "epoch": 1.054523666866387, "grad_norm": 2.168858289718628, "learning_rate": 3.279201129715554e-05, "loss": 0.1184, "step": 1760 }, { "epoch": 1.060515278609946, "grad_norm": 2.3398423194885254, "learning_rate": 3.269114383699818e-05, "loss": 0.1464, "step": 1770 }, { "epoch": 1.066506890353505, "grad_norm": 2.287614107131958, "learning_rate": 3.259027637684083e-05, "loss": 0.1323, "step": 1780 }, { "epoch": 1.072498502097064, "grad_norm": 3.1191556453704834, "learning_rate": 3.248940891668348e-05, "loss": 0.1314, "step": 1790 }, { "epoch": 1.078490113840623, "grad_norm": 2.123906135559082, "learning_rate": 3.238854145652613e-05, "loss": 0.1254, "step": 1800 }, { "epoch": 1.084481725584182, "grad_norm": 2.0637898445129395, "learning_rate": 3.228767399636877e-05, "loss": 0.1296, "step": 1810 }, { "epoch": 1.090473337327741, "grad_norm": 3.5648674964904785, "learning_rate": 3.218680653621142e-05, "loss": 0.1445, "step": 1820 }, { "epoch": 1.0964649490713, "grad_norm": 3.08638334274292, "learning_rate": 3.208593907605407e-05, "loss": 0.122, "step": 1830 }, { "epoch": 1.102456560814859, "grad_norm": 2.7976272106170654, "learning_rate": 3.198507161589671e-05, "loss": 0.1176, "step": 1840 }, { "epoch": 1.1084481725584183, "grad_norm": 2.4950814247131348, "learning_rate": 3.188420415573936e-05, "loss": 0.1516, "step": 1850 }, { "epoch": 1.1144397843019773, "grad_norm": 3.1080756187438965, "learning_rate": 3.1783336695582e-05, "loss": 0.1309, "step": 1860 }, { "epoch": 1.1204313960455363, "grad_norm": 1.801734209060669, "learning_rate": 3.168246923542466e-05, "loss": 0.1276, "step": 1870 }, { "epoch": 1.1264230077890953, "grad_norm": 1.831494927406311, "learning_rate": 3.15816017752673e-05, "loss": 0.1353, "step": 1880 }, { "epoch": 1.1324146195326543, "grad_norm": 2.9599852561950684, "learning_rate": 3.148073431510994e-05, "loss": 0.1241, "step": 1890 }, { "epoch": 1.1384062312762133, "grad_norm": 1.5874921083450317, "learning_rate": 3.137986685495259e-05, "loss": 0.1259, "step": 1900 }, { "epoch": 1.1443978430197723, "grad_norm": 2.19362211227417, "learning_rate": 3.127899939479524e-05, "loss": 0.1177, "step": 1910 }, { "epoch": 1.1503894547633313, "grad_norm": 1.6142268180847168, "learning_rate": 3.117813193463789e-05, "loss": 0.1196, "step": 1920 }, { "epoch": 1.1563810665068903, "grad_norm": 3.0195200443267822, "learning_rate": 3.107726447448053e-05, "loss": 0.1235, "step": 1930 }, { "epoch": 1.1623726782504493, "grad_norm": 2.401179075241089, "learning_rate": 3.097639701432318e-05, "loss": 0.1349, "step": 1940 }, { "epoch": 1.1683642899940083, "grad_norm": 3.15936279296875, "learning_rate": 3.087552955416583e-05, "loss": 0.1273, "step": 1950 }, { "epoch": 1.1743559017375673, "grad_norm": 2.6735379695892334, "learning_rate": 3.077466209400847e-05, "loss": 0.1003, "step": 1960 }, { "epoch": 1.1803475134811263, "grad_norm": 2.687624216079712, "learning_rate": 3.067379463385112e-05, "loss": 0.1053, "step": 1970 }, { "epoch": 1.1863391252246855, "grad_norm": 1.7122023105621338, "learning_rate": 3.057292717369377e-05, "loss": 0.1237, "step": 1980 }, { "epoch": 1.1923307369682445, "grad_norm": 2.6045424938201904, "learning_rate": 3.0472059713536415e-05, "loss": 0.135, "step": 1990 }, { "epoch": 1.1983223487118035, "grad_norm": 2.4615840911865234, "learning_rate": 3.037119225337906e-05, "loss": 0.1086, "step": 2000 }, { "epoch": 1.1983223487118035, "eval_loss": 0.3957676291465759, "eval_runtime": 1529.837, "eval_samples_per_second": 0.927, "eval_steps_per_second": 0.058, "eval_wer": 0.42359088572897036, "step": 2000 }, { "epoch": 1.2043139604553625, "grad_norm": 2.78117036819458, "learning_rate": 3.027032479322171e-05, "loss": 0.1399, "step": 2010 }, { "epoch": 1.2103055721989215, "grad_norm": 3.1339714527130127, "learning_rate": 3.0169457333064355e-05, "loss": 0.1489, "step": 2020 }, { "epoch": 1.2162971839424805, "grad_norm": 2.7637593746185303, "learning_rate": 3.0068589872907e-05, "loss": 0.1129, "step": 2030 }, { "epoch": 1.2222887956860395, "grad_norm": 2.9713125228881836, "learning_rate": 2.996772241274965e-05, "loss": 0.1118, "step": 2040 }, { "epoch": 1.2282804074295985, "grad_norm": 3.4637749195098877, "learning_rate": 2.9866854952592295e-05, "loss": 0.1173, "step": 2050 }, { "epoch": 1.2342720191731575, "grad_norm": 3.3388004302978516, "learning_rate": 2.9765987492434944e-05, "loss": 0.1347, "step": 2060 }, { "epoch": 1.2402636309167165, "grad_norm": 1.4394806623458862, "learning_rate": 2.966512003227759e-05, "loss": 0.1301, "step": 2070 }, { "epoch": 1.2462552426602755, "grad_norm": 2.6756908893585205, "learning_rate": 2.956425257212024e-05, "loss": 0.1267, "step": 2080 }, { "epoch": 1.2522468544038348, "grad_norm": 3.740405559539795, "learning_rate": 2.9463385111962884e-05, "loss": 0.1339, "step": 2090 }, { "epoch": 1.2582384661473935, "grad_norm": 2.376574993133545, "learning_rate": 2.9362517651805526e-05, "loss": 0.1471, "step": 2100 }, { "epoch": 1.2642300778909528, "grad_norm": 2.841569662094116, "learning_rate": 2.9261650191648175e-05, "loss": 0.107, "step": 2110 }, { "epoch": 1.2702216896345118, "grad_norm": 2.9141831398010254, "learning_rate": 2.916078273149082e-05, "loss": 0.1216, "step": 2120 }, { "epoch": 1.2762133013780708, "grad_norm": 2.759247303009033, "learning_rate": 2.905991527133347e-05, "loss": 0.1326, "step": 2130 }, { "epoch": 1.2822049131216298, "grad_norm": 2.4907028675079346, "learning_rate": 2.8959047811176115e-05, "loss": 0.1193, "step": 2140 }, { "epoch": 1.2881965248651888, "grad_norm": 3.2932450771331787, "learning_rate": 2.8858180351018764e-05, "loss": 0.1218, "step": 2150 }, { "epoch": 1.2941881366087478, "grad_norm": 2.4969570636749268, "learning_rate": 2.875731289086141e-05, "loss": 0.1111, "step": 2160 }, { "epoch": 1.3001797483523068, "grad_norm": 2.1168346405029297, "learning_rate": 2.8656445430704055e-05, "loss": 0.1267, "step": 2170 }, { "epoch": 1.3061713600958658, "grad_norm": 2.1791086196899414, "learning_rate": 2.8555577970546704e-05, "loss": 0.1269, "step": 2180 }, { "epoch": 1.3121629718394248, "grad_norm": 2.5179903507232666, "learning_rate": 2.845471051038935e-05, "loss": 0.1088, "step": 2190 }, { "epoch": 1.3181545835829838, "grad_norm": 2.5732688903808594, "learning_rate": 2.8353843050232e-05, "loss": 0.1181, "step": 2200 }, { "epoch": 1.3241461953265428, "grad_norm": 2.7886123657226562, "learning_rate": 2.8252975590074644e-05, "loss": 0.1242, "step": 2210 }, { "epoch": 1.330137807070102, "grad_norm": 3.418480396270752, "learning_rate": 2.8152108129917286e-05, "loss": 0.1219, "step": 2220 }, { "epoch": 1.3361294188136608, "grad_norm": 1.7312183380126953, "learning_rate": 2.805124066975994e-05, "loss": 0.1096, "step": 2230 }, { "epoch": 1.34212103055722, "grad_norm": 1.4882137775421143, "learning_rate": 2.795037320960258e-05, "loss": 0.1034, "step": 2240 }, { "epoch": 1.348112642300779, "grad_norm": 3.208034038543701, "learning_rate": 2.7849505749445233e-05, "loss": 0.1288, "step": 2250 }, { "epoch": 1.354104254044338, "grad_norm": 2.543548583984375, "learning_rate": 2.7748638289287875e-05, "loss": 0.0991, "step": 2260 }, { "epoch": 1.360095865787897, "grad_norm": 2.1029109954833984, "learning_rate": 2.7647770829130528e-05, "loss": 0.12, "step": 2270 }, { "epoch": 1.366087477531456, "grad_norm": 1.8528416156768799, "learning_rate": 2.754690336897317e-05, "loss": 0.1116, "step": 2280 }, { "epoch": 1.372079089275015, "grad_norm": 2.3777570724487305, "learning_rate": 2.7446035908815815e-05, "loss": 0.143, "step": 2290 }, { "epoch": 1.378070701018574, "grad_norm": 2.3329262733459473, "learning_rate": 2.7345168448658464e-05, "loss": 0.1067, "step": 2300 }, { "epoch": 1.384062312762133, "grad_norm": 1.7974488735198975, "learning_rate": 2.724430098850111e-05, "loss": 0.1287, "step": 2310 }, { "epoch": 1.390053924505692, "grad_norm": 2.3591809272766113, "learning_rate": 2.714343352834376e-05, "loss": 0.1142, "step": 2320 }, { "epoch": 1.396045536249251, "grad_norm": 1.9029614925384521, "learning_rate": 2.7042566068186404e-05, "loss": 0.1204, "step": 2330 }, { "epoch": 1.40203714799281, "grad_norm": 2.7638425827026367, "learning_rate": 2.6941698608029053e-05, "loss": 0.1188, "step": 2340 }, { "epoch": 1.4080287597363692, "grad_norm": 2.830389976501465, "learning_rate": 2.68408311478717e-05, "loss": 0.1167, "step": 2350 }, { "epoch": 1.414020371479928, "grad_norm": 2.2718679904937744, "learning_rate": 2.673996368771434e-05, "loss": 0.1286, "step": 2360 }, { "epoch": 1.4200119832234872, "grad_norm": 2.5862531661987305, "learning_rate": 2.6639096227556993e-05, "loss": 0.122, "step": 2370 }, { "epoch": 1.4260035949670462, "grad_norm": 2.345581531524658, "learning_rate": 2.6538228767399635e-05, "loss": 0.1102, "step": 2380 }, { "epoch": 1.4319952067106052, "grad_norm": 1.8267598152160645, "learning_rate": 2.6437361307242288e-05, "loss": 0.1139, "step": 2390 }, { "epoch": 1.4379868184541642, "grad_norm": 2.5517992973327637, "learning_rate": 2.633649384708493e-05, "loss": 0.1173, "step": 2400 }, { "epoch": 1.4439784301977232, "grad_norm": 2.2720143795013428, "learning_rate": 2.6235626386927582e-05, "loss": 0.1129, "step": 2410 }, { "epoch": 1.4499700419412822, "grad_norm": 2.7546370029449463, "learning_rate": 2.6134758926770224e-05, "loss": 0.1394, "step": 2420 }, { "epoch": 1.4559616536848412, "grad_norm": 3.8197317123413086, "learning_rate": 2.603389146661287e-05, "loss": 0.1276, "step": 2430 }, { "epoch": 1.4619532654284002, "grad_norm": 1.800384283065796, "learning_rate": 2.593302400645552e-05, "loss": 0.0952, "step": 2440 }, { "epoch": 1.4679448771719592, "grad_norm": 1.8722529411315918, "learning_rate": 2.5832156546298164e-05, "loss": 0.1015, "step": 2450 }, { "epoch": 1.4739364889155182, "grad_norm": 2.2280287742614746, "learning_rate": 2.5731289086140813e-05, "loss": 0.109, "step": 2460 }, { "epoch": 1.4799281006590772, "grad_norm": 2.547365665435791, "learning_rate": 2.563042162598346e-05, "loss": 0.1164, "step": 2470 }, { "epoch": 1.4859197124026364, "grad_norm": 2.6080679893493652, "learning_rate": 2.5529554165826108e-05, "loss": 0.1213, "step": 2480 }, { "epoch": 1.4919113241461952, "grad_norm": 2.250347852706909, "learning_rate": 2.5428686705668753e-05, "loss": 0.1107, "step": 2490 }, { "epoch": 1.4979029358897544, "grad_norm": 2.15417742729187, "learning_rate": 2.5327819245511396e-05, "loss": 0.132, "step": 2500 }, { "epoch": 1.5038945476333132, "grad_norm": 2.5671651363372803, "learning_rate": 2.5226951785354048e-05, "loss": 0.1218, "step": 2510 }, { "epoch": 1.5098861593768724, "grad_norm": 2.304866313934326, "learning_rate": 2.512608432519669e-05, "loss": 0.1236, "step": 2520 }, { "epoch": 1.5158777711204314, "grad_norm": 2.581188201904297, "learning_rate": 2.5025216865039342e-05, "loss": 0.1155, "step": 2530 }, { "epoch": 1.5218693828639904, "grad_norm": 2.099909782409668, "learning_rate": 2.4924349404881985e-05, "loss": 0.1271, "step": 2540 }, { "epoch": 1.5278609946075494, "grad_norm": 2.0890612602233887, "learning_rate": 2.4823481944724633e-05, "loss": 0.1107, "step": 2550 }, { "epoch": 1.5338526063511084, "grad_norm": 2.7176012992858887, "learning_rate": 2.472261448456728e-05, "loss": 0.1244, "step": 2560 }, { "epoch": 1.5398442180946674, "grad_norm": 2.306342840194702, "learning_rate": 2.4621747024409928e-05, "loss": 0.1197, "step": 2570 }, { "epoch": 1.5458358298382264, "grad_norm": 1.9637123346328735, "learning_rate": 2.4520879564252573e-05, "loss": 0.121, "step": 2580 }, { "epoch": 1.5518274415817856, "grad_norm": 2.171283006668091, "learning_rate": 2.442001210409522e-05, "loss": 0.1272, "step": 2590 }, { "epoch": 1.5578190533253444, "grad_norm": 2.2798995971679688, "learning_rate": 2.4319144643937865e-05, "loss": 0.1181, "step": 2600 }, { "epoch": 1.5638106650689036, "grad_norm": 2.048116683959961, "learning_rate": 2.4218277183780514e-05, "loss": 0.1296, "step": 2610 }, { "epoch": 1.5698022768124624, "grad_norm": 2.062462091445923, "learning_rate": 2.411740972362316e-05, "loss": 0.1019, "step": 2620 }, { "epoch": 1.5757938885560216, "grad_norm": 3.318964958190918, "learning_rate": 2.4016542263465808e-05, "loss": 0.1156, "step": 2630 }, { "epoch": 1.5817855002995804, "grad_norm": 1.8189563751220703, "learning_rate": 2.3915674803308454e-05, "loss": 0.1131, "step": 2640 }, { "epoch": 1.5877771120431396, "grad_norm": 2.640641689300537, "learning_rate": 2.38148073431511e-05, "loss": 0.1153, "step": 2650 }, { "epoch": 1.5937687237866986, "grad_norm": 2.267590045928955, "learning_rate": 2.3713939882993748e-05, "loss": 0.1276, "step": 2660 }, { "epoch": 1.5997603355302576, "grad_norm": 2.572810649871826, "learning_rate": 2.3613072422836394e-05, "loss": 0.1042, "step": 2670 }, { "epoch": 1.6057519472738166, "grad_norm": 2.604325771331787, "learning_rate": 2.351220496267904e-05, "loss": 0.1387, "step": 2680 }, { "epoch": 1.6117435590173756, "grad_norm": 1.9912638664245605, "learning_rate": 2.3411337502521688e-05, "loss": 0.1066, "step": 2690 }, { "epoch": 1.6177351707609346, "grad_norm": 2.712186336517334, "learning_rate": 2.3310470042364334e-05, "loss": 0.105, "step": 2700 }, { "epoch": 1.6237267825044936, "grad_norm": 3.0537188053131104, "learning_rate": 2.3209602582206983e-05, "loss": 0.1257, "step": 2710 }, { "epoch": 1.6297183942480529, "grad_norm": 2.5793514251708984, "learning_rate": 2.3108735122049628e-05, "loss": 0.13, "step": 2720 }, { "epoch": 1.6357100059916116, "grad_norm": 2.015709400177002, "learning_rate": 2.3007867661892274e-05, "loss": 0.1185, "step": 2730 }, { "epoch": 1.6417016177351709, "grad_norm": 2.600104331970215, "learning_rate": 2.2907000201734923e-05, "loss": 0.1208, "step": 2740 }, { "epoch": 1.6476932294787296, "grad_norm": 1.7748316526412964, "learning_rate": 2.2806132741577568e-05, "loss": 0.0968, "step": 2750 }, { "epoch": 1.6536848412222889, "grad_norm": 3.2551629543304443, "learning_rate": 2.2705265281420217e-05, "loss": 0.105, "step": 2760 }, { "epoch": 1.6596764529658476, "grad_norm": 2.5319693088531494, "learning_rate": 2.2604397821262863e-05, "loss": 0.1035, "step": 2770 }, { "epoch": 1.6656680647094069, "grad_norm": 2.2642369270324707, "learning_rate": 2.2503530361105508e-05, "loss": 0.1082, "step": 2780 }, { "epoch": 1.6716596764529659, "grad_norm": 2.3224713802337646, "learning_rate": 2.2402662900948154e-05, "loss": 0.1113, "step": 2790 }, { "epoch": 1.6776512881965249, "grad_norm": 1.8430181741714478, "learning_rate": 2.2301795440790803e-05, "loss": 0.0963, "step": 2800 }, { "epoch": 1.6836428999400839, "grad_norm": 2.5304131507873535, "learning_rate": 2.2200927980633448e-05, "loss": 0.1057, "step": 2810 }, { "epoch": 1.6896345116836429, "grad_norm": 1.7761938571929932, "learning_rate": 2.2100060520476097e-05, "loss": 0.1158, "step": 2820 }, { "epoch": 1.6956261234272019, "grad_norm": 2.3258302211761475, "learning_rate": 2.1999193060318743e-05, "loss": 0.1345, "step": 2830 }, { "epoch": 1.7016177351707609, "grad_norm": 1.5975005626678467, "learning_rate": 2.1898325600161392e-05, "loss": 0.0934, "step": 2840 }, { "epoch": 1.70760934691432, "grad_norm": 2.0967276096343994, "learning_rate": 2.1797458140004034e-05, "loss": 0.1222, "step": 2850 }, { "epoch": 1.7136009586578789, "grad_norm": 2.689154863357544, "learning_rate": 2.1696590679846683e-05, "loss": 0.1021, "step": 2860 }, { "epoch": 1.719592570401438, "grad_norm": 2.287780523300171, "learning_rate": 2.159572321968933e-05, "loss": 0.1163, "step": 2870 }, { "epoch": 1.7255841821449969, "grad_norm": 2.54907488822937, "learning_rate": 2.1494855759531977e-05, "loss": 0.1093, "step": 2880 }, { "epoch": 1.731575793888556, "grad_norm": 2.6872503757476807, "learning_rate": 2.1393988299374623e-05, "loss": 0.1195, "step": 2890 }, { "epoch": 1.737567405632115, "grad_norm": 2.690882682800293, "learning_rate": 2.1293120839217272e-05, "loss": 0.1036, "step": 2900 }, { "epoch": 1.743559017375674, "grad_norm": 1.8093252182006836, "learning_rate": 2.1192253379059917e-05, "loss": 0.0956, "step": 2910 }, { "epoch": 1.749550629119233, "grad_norm": 2.9592061042785645, "learning_rate": 2.1091385918902563e-05, "loss": 0.1186, "step": 2920 }, { "epoch": 1.755542240862792, "grad_norm": 2.5291833877563477, "learning_rate": 2.099051845874521e-05, "loss": 0.1039, "step": 2930 }, { "epoch": 1.761533852606351, "grad_norm": 1.820418119430542, "learning_rate": 2.0889650998587857e-05, "loss": 0.0998, "step": 2940 }, { "epoch": 1.76752546434991, "grad_norm": 2.9867782592773438, "learning_rate": 2.0788783538430503e-05, "loss": 0.1226, "step": 2950 }, { "epoch": 1.773517076093469, "grad_norm": 3.5829522609710693, "learning_rate": 2.0687916078273152e-05, "loss": 0.1139, "step": 2960 }, { "epoch": 1.779508687837028, "grad_norm": 1.5877478122711182, "learning_rate": 2.0587048618115797e-05, "loss": 0.0974, "step": 2970 }, { "epoch": 1.7855002995805873, "grad_norm": 2.036705255508423, "learning_rate": 2.0486181157958443e-05, "loss": 0.1058, "step": 2980 }, { "epoch": 1.791491911324146, "grad_norm": 2.361468553543091, "learning_rate": 2.038531369780109e-05, "loss": 0.1031, "step": 2990 }, { "epoch": 1.7974835230677053, "grad_norm": 3.4837324619293213, "learning_rate": 2.0284446237643737e-05, "loss": 0.1181, "step": 3000 }, { "epoch": 1.7974835230677053, "eval_loss": 0.39589497447013855, "eval_runtime": 1534.4493, "eval_samples_per_second": 0.924, "eval_steps_per_second": 0.058, "eval_wer": 0.39158243389869224, "step": 3000 }, { "epoch": 1.803475134811264, "grad_norm": 2.0314831733703613, "learning_rate": 2.0183578777486383e-05, "loss": 0.0952, "step": 3010 }, { "epoch": 1.8094667465548233, "grad_norm": 2.6841511726379395, "learning_rate": 2.0082711317329032e-05, "loss": 0.1172, "step": 3020 }, { "epoch": 1.8154583582983823, "grad_norm": 2.687436103820801, "learning_rate": 1.9981843857171677e-05, "loss": 0.0986, "step": 3030 }, { "epoch": 1.8214499700419413, "grad_norm": 2.0085389614105225, "learning_rate": 1.9880976397014326e-05, "loss": 0.1105, "step": 3040 }, { "epoch": 1.8274415817855003, "grad_norm": 1.7776066064834595, "learning_rate": 1.9790195682872706e-05, "loss": 0.1068, "step": 3050 }, { "epoch": 1.8334331935290593, "grad_norm": 2.3934295177459717, "learning_rate": 1.9689328222715355e-05, "loss": 0.1042, "step": 3060 }, { "epoch": 1.8394248052726183, "grad_norm": 1.7487627267837524, "learning_rate": 1.9588460762558e-05, "loss": 0.0973, "step": 3070 }, { "epoch": 1.8454164170161773, "grad_norm": 1.7244535684585571, "learning_rate": 1.9487593302400646e-05, "loss": 0.1118, "step": 3080 }, { "epoch": 1.8514080287597365, "grad_norm": 3.1214139461517334, "learning_rate": 1.9386725842243295e-05, "loss": 0.1077, "step": 3090 }, { "epoch": 1.8573996405032953, "grad_norm": 2.187126398086548, "learning_rate": 1.928585838208594e-05, "loss": 0.1208, "step": 3100 }, { "epoch": 1.8633912522468545, "grad_norm": 1.8465856313705444, "learning_rate": 1.9184990921928586e-05, "loss": 0.098, "step": 3110 }, { "epoch": 1.8693828639904133, "grad_norm": 3.0469441413879395, "learning_rate": 1.9084123461771235e-05, "loss": 0.0905, "step": 3120 }, { "epoch": 1.8753744757339725, "grad_norm": 1.5280126333236694, "learning_rate": 1.898325600161388e-05, "loss": 0.0998, "step": 3130 }, { "epoch": 1.8813660874775313, "grad_norm": 1.8420058488845825, "learning_rate": 1.888238854145653e-05, "loss": 0.1108, "step": 3140 }, { "epoch": 1.8873576992210905, "grad_norm": 1.4494051933288574, "learning_rate": 1.8781521081299175e-05, "loss": 0.095, "step": 3150 }, { "epoch": 1.8933493109646495, "grad_norm": 2.12457013130188, "learning_rate": 1.868065362114182e-05, "loss": 0.1015, "step": 3160 }, { "epoch": 1.8993409227082085, "grad_norm": 1.672364592552185, "learning_rate": 1.8579786160984466e-05, "loss": 0.1107, "step": 3170 }, { "epoch": 1.9053325344517675, "grad_norm": 1.6993036270141602, "learning_rate": 1.8478918700827115e-05, "loss": 0.0929, "step": 3180 }, { "epoch": 1.9113241461953265, "grad_norm": 2.326103448867798, "learning_rate": 1.837805124066976e-05, "loss": 0.1141, "step": 3190 }, { "epoch": 1.9173157579388855, "grad_norm": 2.1768336296081543, "learning_rate": 1.827718378051241e-05, "loss": 0.0883, "step": 3200 }, { "epoch": 1.9233073696824445, "grad_norm": 1.8029934167861938, "learning_rate": 1.8176316320355055e-05, "loss": 0.0974, "step": 3210 }, { "epoch": 1.9292989814260038, "grad_norm": 1.255569338798523, "learning_rate": 1.8075448860197704e-05, "loss": 0.1032, "step": 3220 }, { "epoch": 1.9352905931695625, "grad_norm": 1.9688622951507568, "learning_rate": 1.7974581400040346e-05, "loss": 0.1133, "step": 3230 }, { "epoch": 1.9412822049131218, "grad_norm": 1.9603787660598755, "learning_rate": 1.7873713939882995e-05, "loss": 0.0965, "step": 3240 }, { "epoch": 1.9472738166566805, "grad_norm": 2.000797748565674, "learning_rate": 1.777284647972564e-05, "loss": 0.0958, "step": 3250 }, { "epoch": 1.9532654284002398, "grad_norm": 2.369920015335083, "learning_rate": 1.767197901956829e-05, "loss": 0.0887, "step": 3260 }, { "epoch": 1.9592570401437985, "grad_norm": 2.7224764823913574, "learning_rate": 1.7571111559410935e-05, "loss": 0.0886, "step": 3270 }, { "epoch": 1.9652486518873578, "grad_norm": 2.3583805561065674, "learning_rate": 1.7470244099253584e-05, "loss": 0.1129, "step": 3280 }, { "epoch": 1.9712402636309168, "grad_norm": 1.8538249731063843, "learning_rate": 1.7369376639096226e-05, "loss": 0.1154, "step": 3290 }, { "epoch": 1.9772318753744758, "grad_norm": 3.2503836154937744, "learning_rate": 1.7268509178938875e-05, "loss": 0.0976, "step": 3300 }, { "epoch": 1.9832234871180348, "grad_norm": 2.481663703918457, "learning_rate": 1.716764171878152e-05, "loss": 0.1121, "step": 3310 }, { "epoch": 1.9892150988615938, "grad_norm": 1.6555776596069336, "learning_rate": 1.706677425862417e-05, "loss": 0.0901, "step": 3320 }, { "epoch": 1.9952067106051528, "grad_norm": 2.6977438926696777, "learning_rate": 1.6965906798466815e-05, "loss": 0.1068, "step": 3330 }, { "epoch": 2.0011983223487118, "grad_norm": 1.913468599319458, "learning_rate": 1.6865039338309464e-05, "loss": 0.086, "step": 3340 }, { "epoch": 2.007189934092271, "grad_norm": 1.3262066841125488, "learning_rate": 1.676417187815211e-05, "loss": 0.0658, "step": 3350 }, { "epoch": 2.0131815458358298, "grad_norm": 1.445493459701538, "learning_rate": 1.6663304417994755e-05, "loss": 0.0605, "step": 3360 }, { "epoch": 2.019173157579389, "grad_norm": 1.2603121995925903, "learning_rate": 1.65624369578374e-05, "loss": 0.0566, "step": 3370 }, { "epoch": 2.0251647693229478, "grad_norm": 1.7415363788604736, "learning_rate": 1.646156949768005e-05, "loss": 0.0543, "step": 3380 }, { "epoch": 2.031156381066507, "grad_norm": 1.135882019996643, "learning_rate": 1.6360702037522695e-05, "loss": 0.0547, "step": 3390 }, { "epoch": 2.0371479928100658, "grad_norm": 1.364913821220398, "learning_rate": 1.6259834577365344e-05, "loss": 0.0494, "step": 3400 }, { "epoch": 2.043139604553625, "grad_norm": 2.1429061889648438, "learning_rate": 1.615896711720799e-05, "loss": 0.0599, "step": 3410 }, { "epoch": 2.0491312162971838, "grad_norm": 2.314098596572876, "learning_rate": 1.605809965705064e-05, "loss": 0.0562, "step": 3420 }, { "epoch": 2.055122828040743, "grad_norm": 1.4434194564819336, "learning_rate": 1.595723219689328e-05, "loss": 0.0568, "step": 3430 }, { "epoch": 2.061114439784302, "grad_norm": 2.2602288722991943, "learning_rate": 1.585636473673593e-05, "loss": 0.0539, "step": 3440 }, { "epoch": 2.067106051527861, "grad_norm": 1.6304532289505005, "learning_rate": 1.5755497276578575e-05, "loss": 0.0532, "step": 3450 }, { "epoch": 2.07309766327142, "grad_norm": 2.1285336017608643, "learning_rate": 1.5654629816421224e-05, "loss": 0.0608, "step": 3460 }, { "epoch": 2.079089275014979, "grad_norm": 1.5324342250823975, "learning_rate": 1.555376235626387e-05, "loss": 0.056, "step": 3470 }, { "epoch": 2.085080886758538, "grad_norm": 2.044532060623169, "learning_rate": 1.545289489610652e-05, "loss": 0.0544, "step": 3480 }, { "epoch": 2.091072498502097, "grad_norm": 1.922879934310913, "learning_rate": 1.535202743594916e-05, "loss": 0.0624, "step": 3490 }, { "epoch": 2.097064110245656, "grad_norm": 2.054161310195923, "learning_rate": 1.525115997579181e-05, "loss": 0.0573, "step": 3500 }, { "epoch": 2.103055721989215, "grad_norm": 1.7563127279281616, "learning_rate": 1.5150292515634457e-05, "loss": 0.0689, "step": 3510 }, { "epoch": 2.109047333732774, "grad_norm": 1.2950539588928223, "learning_rate": 1.5049425055477104e-05, "loss": 0.0572, "step": 3520 }, { "epoch": 2.115038945476333, "grad_norm": 2.48503041267395, "learning_rate": 1.4948557595319751e-05, "loss": 0.0584, "step": 3530 }, { "epoch": 2.121030557219892, "grad_norm": 3.5458927154541016, "learning_rate": 1.4847690135162399e-05, "loss": 0.0591, "step": 3540 }, { "epoch": 2.127022168963451, "grad_norm": 1.8503930568695068, "learning_rate": 1.4746822675005046e-05, "loss": 0.0525, "step": 3550 }, { "epoch": 2.13301378070701, "grad_norm": 1.5099945068359375, "learning_rate": 1.464595521484769e-05, "loss": 0.059, "step": 3560 }, { "epoch": 2.139005392450569, "grad_norm": 1.6359002590179443, "learning_rate": 1.4545087754690337e-05, "loss": 0.0482, "step": 3570 }, { "epoch": 2.144997004194128, "grad_norm": 1.9500666856765747, "learning_rate": 1.4444220294532984e-05, "loss": 0.0596, "step": 3580 }, { "epoch": 2.1509886159376874, "grad_norm": 1.7838146686553955, "learning_rate": 1.4343352834375632e-05, "loss": 0.0574, "step": 3590 }, { "epoch": 2.156980227681246, "grad_norm": 1.5071059465408325, "learning_rate": 1.4242485374218279e-05, "loss": 0.0424, "step": 3600 }, { "epoch": 2.1629718394248054, "grad_norm": 2.2160496711730957, "learning_rate": 1.4141617914060926e-05, "loss": 0.0608, "step": 3610 }, { "epoch": 2.168963451168364, "grad_norm": 1.6579121351242065, "learning_rate": 1.404075045390357e-05, "loss": 0.0547, "step": 3620 }, { "epoch": 2.1749550629119234, "grad_norm": 2.632821798324585, "learning_rate": 1.3939882993746217e-05, "loss": 0.0779, "step": 3630 }, { "epoch": 2.180946674655482, "grad_norm": 1.9470138549804688, "learning_rate": 1.3839015533588864e-05, "loss": 0.0596, "step": 3640 }, { "epoch": 2.1869382863990414, "grad_norm": 1.6838595867156982, "learning_rate": 1.3738148073431512e-05, "loss": 0.051, "step": 3650 }, { "epoch": 2.1929298981426, "grad_norm": 1.5005348920822144, "learning_rate": 1.3637280613274159e-05, "loss": 0.061, "step": 3660 }, { "epoch": 2.1989215098861594, "grad_norm": 1.6386892795562744, "learning_rate": 1.3536413153116806e-05, "loss": 0.0482, "step": 3670 }, { "epoch": 2.204913121629718, "grad_norm": 1.9598783254623413, "learning_rate": 1.3435545692959453e-05, "loss": 0.0551, "step": 3680 }, { "epoch": 2.2109047333732774, "grad_norm": 2.5483479499816895, "learning_rate": 1.3334678232802097e-05, "loss": 0.0664, "step": 3690 }, { "epoch": 2.2168963451168366, "grad_norm": 1.7083418369293213, "learning_rate": 1.3233810772644744e-05, "loss": 0.08, "step": 3700 }, { "epoch": 2.2228879568603954, "grad_norm": 1.9843679666519165, "learning_rate": 1.3132943312487392e-05, "loss": 0.0531, "step": 3710 }, { "epoch": 2.2288795686039546, "grad_norm": 1.2589327096939087, "learning_rate": 1.3032075852330039e-05, "loss": 0.0569, "step": 3720 }, { "epoch": 2.2348711803475134, "grad_norm": 2.206603527069092, "learning_rate": 1.2931208392172686e-05, "loss": 0.052, "step": 3730 }, { "epoch": 2.2408627920910726, "grad_norm": 1.1368589401245117, "learning_rate": 1.2830340932015333e-05, "loss": 0.0607, "step": 3740 }, { "epoch": 2.2468544038346314, "grad_norm": 1.757006049156189, "learning_rate": 1.272947347185798e-05, "loss": 0.0376, "step": 3750 }, { "epoch": 2.2528460155781906, "grad_norm": 1.6406621932983398, "learning_rate": 1.2628606011700625e-05, "loss": 0.0696, "step": 3760 }, { "epoch": 2.2588376273217494, "grad_norm": 1.9685550928115845, "learning_rate": 1.2527738551543272e-05, "loss": 0.0531, "step": 3770 }, { "epoch": 2.2648292390653086, "grad_norm": 1.26189386844635, "learning_rate": 1.2426871091385919e-05, "loss": 0.0542, "step": 3780 }, { "epoch": 2.2708208508088674, "grad_norm": 2.2154734134674072, "learning_rate": 1.2326003631228566e-05, "loss": 0.0647, "step": 3790 }, { "epoch": 2.2768124625524266, "grad_norm": 1.3167383670806885, "learning_rate": 1.2225136171071214e-05, "loss": 0.0536, "step": 3800 }, { "epoch": 2.282804074295986, "grad_norm": 3.2605197429656982, "learning_rate": 1.2124268710913859e-05, "loss": 0.0607, "step": 3810 }, { "epoch": 2.2887956860395446, "grad_norm": 1.6796486377716064, "learning_rate": 1.2023401250756506e-05, "loss": 0.0556, "step": 3820 }, { "epoch": 2.2947872977831034, "grad_norm": 1.8123384714126587, "learning_rate": 1.1922533790599154e-05, "loss": 0.0492, "step": 3830 }, { "epoch": 2.3007789095266626, "grad_norm": 2.183927059173584, "learning_rate": 1.1821666330441799e-05, "loss": 0.0555, "step": 3840 }, { "epoch": 2.306770521270222, "grad_norm": 1.6841498613357544, "learning_rate": 1.1720798870284446e-05, "loss": 0.0521, "step": 3850 }, { "epoch": 2.3127621330137806, "grad_norm": 1.5228015184402466, "learning_rate": 1.1619931410127094e-05, "loss": 0.0518, "step": 3860 }, { "epoch": 2.31875374475734, "grad_norm": 1.1061369180679321, "learning_rate": 1.151906394996974e-05, "loss": 0.0467, "step": 3870 }, { "epoch": 2.3247453565008986, "grad_norm": 1.0257222652435303, "learning_rate": 1.1418196489812386e-05, "loss": 0.0602, "step": 3880 }, { "epoch": 2.330736968244458, "grad_norm": 1.333539366722107, "learning_rate": 1.1317329029655034e-05, "loss": 0.065, "step": 3890 }, { "epoch": 2.3367285799880166, "grad_norm": 1.477871298789978, "learning_rate": 1.121646156949768e-05, "loss": 0.0578, "step": 3900 }, { "epoch": 2.342720191731576, "grad_norm": 1.7921220064163208, "learning_rate": 1.1115594109340326e-05, "loss": 0.0475, "step": 3910 }, { "epoch": 2.3487118034751346, "grad_norm": 1.7656147480010986, "learning_rate": 1.1014726649182974e-05, "loss": 0.0602, "step": 3920 }, { "epoch": 2.354703415218694, "grad_norm": 1.8412047624588013, "learning_rate": 1.0913859189025621e-05, "loss": 0.0641, "step": 3930 }, { "epoch": 2.3606950269622526, "grad_norm": 1.4820233583450317, "learning_rate": 1.0812991728868266e-05, "loss": 0.0677, "step": 3940 }, { "epoch": 2.366686638705812, "grad_norm": 1.366188883781433, "learning_rate": 1.0712124268710914e-05, "loss": 0.0587, "step": 3950 }, { "epoch": 2.372678250449371, "grad_norm": 1.3289388418197632, "learning_rate": 1.0611256808553561e-05, "loss": 0.0573, "step": 3960 }, { "epoch": 2.37866986219293, "grad_norm": 1.1990536451339722, "learning_rate": 1.0510389348396207e-05, "loss": 0.0515, "step": 3970 }, { "epoch": 2.384661473936489, "grad_norm": 2.262007236480713, "learning_rate": 1.0409521888238854e-05, "loss": 0.0573, "step": 3980 }, { "epoch": 2.390653085680048, "grad_norm": 1.7814624309539795, "learning_rate": 1.0308654428081501e-05, "loss": 0.0511, "step": 3990 }, { "epoch": 2.396644697423607, "grad_norm": 1.9160945415496826, "learning_rate": 1.0207786967924148e-05, "loss": 0.054, "step": 4000 }, { "epoch": 2.396644697423607, "eval_loss": 0.38849014043807983, "eval_runtime": 1731.1558, "eval_samples_per_second": 0.819, "eval_steps_per_second": 0.051, "eval_wer": 0.38532922163211697, "step": 4000 }, { "epoch": 2.402636309167166, "grad_norm": 1.8122146129608154, "learning_rate": 1.0106919507766796e-05, "loss": 0.0453, "step": 4010 }, { "epoch": 2.408627920910725, "grad_norm": 1.0477360486984253, "learning_rate": 1.0006052047609441e-05, "loss": 0.053, "step": 4020 }, { "epoch": 2.414619532654284, "grad_norm": 1.326602578163147, "learning_rate": 9.905184587452088e-06, "loss": 0.0499, "step": 4030 }, { "epoch": 2.420611144397843, "grad_norm": 1.8951810598373413, "learning_rate": 9.804317127294736e-06, "loss": 0.052, "step": 4040 }, { "epoch": 2.426602756141402, "grad_norm": 3.2080581188201904, "learning_rate": 9.703449667137383e-06, "loss": 0.058, "step": 4050 }, { "epoch": 2.432594367884961, "grad_norm": 2.5068373680114746, "learning_rate": 9.602582206980028e-06, "loss": 0.0499, "step": 4060 }, { "epoch": 2.4385859796285203, "grad_norm": 1.7612242698669434, "learning_rate": 9.501714746822676e-06, "loss": 0.0533, "step": 4070 }, { "epoch": 2.444577591372079, "grad_norm": 1.7996723651885986, "learning_rate": 9.400847286665323e-06, "loss": 0.0564, "step": 4080 }, { "epoch": 2.450569203115638, "grad_norm": 1.415356993675232, "learning_rate": 9.29997982650797e-06, "loss": 0.0508, "step": 4090 }, { "epoch": 2.456560814859197, "grad_norm": 1.7406309843063354, "learning_rate": 9.199112366350616e-06, "loss": 0.0509, "step": 4100 }, { "epoch": 2.4625524266027563, "grad_norm": 1.6858899593353271, "learning_rate": 9.098244906193263e-06, "loss": 0.0597, "step": 4110 }, { "epoch": 2.468544038346315, "grad_norm": 1.3853732347488403, "learning_rate": 8.99737744603591e-06, "loss": 0.0642, "step": 4120 }, { "epoch": 2.4745356500898743, "grad_norm": 1.2157524824142456, "learning_rate": 8.896509985878557e-06, "loss": 0.0428, "step": 4130 }, { "epoch": 2.480527261833433, "grad_norm": 1.7540271282196045, "learning_rate": 8.795642525721203e-06, "loss": 0.0466, "step": 4140 }, { "epoch": 2.4865188735769923, "grad_norm": 2.2730844020843506, "learning_rate": 8.69477506556385e-06, "loss": 0.0554, "step": 4150 }, { "epoch": 2.492510485320551, "grad_norm": 1.7424499988555908, "learning_rate": 8.593907605406497e-06, "loss": 0.054, "step": 4160 }, { "epoch": 2.4985020970641103, "grad_norm": 2.139974594116211, "learning_rate": 8.493040145249143e-06, "loss": 0.0567, "step": 4170 }, { "epoch": 2.5044937088076695, "grad_norm": 1.306476354598999, "learning_rate": 8.39217268509179e-06, "loss": 0.0482, "step": 4180 }, { "epoch": 2.5104853205512283, "grad_norm": 1.921081304550171, "learning_rate": 8.291305224934437e-06, "loss": 0.0417, "step": 4190 }, { "epoch": 2.516476932294787, "grad_norm": 2.030853271484375, "learning_rate": 8.190437764777083e-06, "loss": 0.0489, "step": 4200 }, { "epoch": 2.5224685440383463, "grad_norm": 1.5279465913772583, "learning_rate": 8.08957030461973e-06, "loss": 0.0455, "step": 4210 }, { "epoch": 2.5284601557819055, "grad_norm": 1.7720401287078857, "learning_rate": 7.988702844462377e-06, "loss": 0.0684, "step": 4220 }, { "epoch": 2.5344517675254643, "grad_norm": 1.1482131481170654, "learning_rate": 7.887835384305023e-06, "loss": 0.0544, "step": 4230 }, { "epoch": 2.5404433792690235, "grad_norm": 2.373537302017212, "learning_rate": 7.78696792414767e-06, "loss": 0.0709, "step": 4240 }, { "epoch": 2.5464349910125823, "grad_norm": 1.1651583909988403, "learning_rate": 7.686100463990318e-06, "loss": 0.0465, "step": 4250 }, { "epoch": 2.5524266027561415, "grad_norm": 2.0321195125579834, "learning_rate": 7.585233003832965e-06, "loss": 0.0465, "step": 4260 }, { "epoch": 2.5584182144997003, "grad_norm": 1.7529507875442505, "learning_rate": 7.48436554367561e-06, "loss": 0.0487, "step": 4270 }, { "epoch": 2.5644098262432595, "grad_norm": 1.5366095304489136, "learning_rate": 7.383498083518258e-06, "loss": 0.0592, "step": 4280 }, { "epoch": 2.5704014379868183, "grad_norm": 1.4841970205307007, "learning_rate": 7.282630623360905e-06, "loss": 0.0469, "step": 4290 }, { "epoch": 2.5763930497303775, "grad_norm": 1.1961781978607178, "learning_rate": 7.18176316320355e-06, "loss": 0.036, "step": 4300 }, { "epoch": 2.5823846614739363, "grad_norm": 0.8614912629127502, "learning_rate": 7.080895703046198e-06, "loss": 0.0426, "step": 4310 }, { "epoch": 2.5883762732174955, "grad_norm": 2.814953327178955, "learning_rate": 6.980028242888845e-06, "loss": 0.0566, "step": 4320 }, { "epoch": 2.5943678849610547, "grad_norm": 2.4379405975341797, "learning_rate": 6.87916078273149e-06, "loss": 0.0594, "step": 4330 }, { "epoch": 2.6003594967046135, "grad_norm": 1.8499358892440796, "learning_rate": 6.778293322574138e-06, "loss": 0.0512, "step": 4340 }, { "epoch": 2.6063511084481723, "grad_norm": 1.0436636209487915, "learning_rate": 6.677425862416785e-06, "loss": 0.0597, "step": 4350 }, { "epoch": 2.6123427201917315, "grad_norm": 1.7677953243255615, "learning_rate": 6.576558402259432e-06, "loss": 0.0546, "step": 4360 }, { "epoch": 2.6183343319352907, "grad_norm": 2.6549458503723145, "learning_rate": 6.475690942102078e-06, "loss": 0.056, "step": 4370 }, { "epoch": 2.6243259436788495, "grad_norm": 0.9973770976066589, "learning_rate": 6.374823481944725e-06, "loss": 0.0497, "step": 4380 }, { "epoch": 2.6303175554224087, "grad_norm": 2.5903773307800293, "learning_rate": 6.273956021787372e-06, "loss": 0.0522, "step": 4390 }, { "epoch": 2.6363091671659675, "grad_norm": 1.0804693698883057, "learning_rate": 6.173088561630019e-06, "loss": 0.0356, "step": 4400 }, { "epoch": 2.6423007789095267, "grad_norm": 1.8099851608276367, "learning_rate": 6.072221101472665e-06, "loss": 0.051, "step": 4410 }, { "epoch": 2.6482923906530855, "grad_norm": 1.1087106466293335, "learning_rate": 5.971353641315311e-06, "loss": 0.0587, "step": 4420 }, { "epoch": 2.6542840023966447, "grad_norm": 1.498772144317627, "learning_rate": 5.870486181157959e-06, "loss": 0.0497, "step": 4430 }, { "epoch": 2.660275614140204, "grad_norm": 1.4975790977478027, "learning_rate": 5.769618721000605e-06, "loss": 0.0544, "step": 4440 }, { "epoch": 2.6662672258837627, "grad_norm": 1.380449652671814, "learning_rate": 5.668751260843252e-06, "loss": 0.0511, "step": 4450 }, { "epoch": 2.6722588376273215, "grad_norm": 1.132257103919983, "learning_rate": 5.567883800685899e-06, "loss": 0.0397, "step": 4460 }, { "epoch": 2.6782504493708807, "grad_norm": 1.5480509996414185, "learning_rate": 5.467016340528545e-06, "loss": 0.0461, "step": 4470 }, { "epoch": 2.68424206111444, "grad_norm": 1.469515085220337, "learning_rate": 5.366148880371192e-06, "loss": 0.0418, "step": 4480 }, { "epoch": 2.6902336728579987, "grad_norm": 1.5454354286193848, "learning_rate": 5.265281420213839e-06, "loss": 0.0492, "step": 4490 }, { "epoch": 2.696225284601558, "grad_norm": 1.7315654754638672, "learning_rate": 5.164413960056486e-06, "loss": 0.0575, "step": 4500 }, { "epoch": 2.7022168963451167, "grad_norm": 1.5216388702392578, "learning_rate": 5.063546499899132e-06, "loss": 0.0491, "step": 4510 }, { "epoch": 2.708208508088676, "grad_norm": 1.5422757863998413, "learning_rate": 4.96267903974178e-06, "loss": 0.0541, "step": 4520 }, { "epoch": 2.7142001198322347, "grad_norm": 2.1533663272857666, "learning_rate": 4.861811579584427e-06, "loss": 0.0492, "step": 4530 }, { "epoch": 2.720191731575794, "grad_norm": 1.8315476179122925, "learning_rate": 4.760944119427073e-06, "loss": 0.0519, "step": 4540 }, { "epoch": 2.7261833433193527, "grad_norm": 1.1926990747451782, "learning_rate": 4.6600766592697205e-06, "loss": 0.0417, "step": 4550 }, { "epoch": 2.732174955062912, "grad_norm": 1.0338667631149292, "learning_rate": 4.559209199112367e-06, "loss": 0.0544, "step": 4560 }, { "epoch": 2.7381665668064707, "grad_norm": 1.5781855583190918, "learning_rate": 4.458341738955013e-06, "loss": 0.043, "step": 4570 }, { "epoch": 2.74415817855003, "grad_norm": 1.4570903778076172, "learning_rate": 4.3574742787976605e-06, "loss": 0.0471, "step": 4580 }, { "epoch": 2.750149790293589, "grad_norm": 1.8981298208236694, "learning_rate": 4.256606818640307e-06, "loss": 0.0445, "step": 4590 }, { "epoch": 2.756141402037148, "grad_norm": 1.8429211378097534, "learning_rate": 4.155739358482954e-06, "loss": 0.0247, "step": 4600 }, { "epoch": 2.7621330137807067, "grad_norm": 1.007271409034729, "learning_rate": 4.0548718983256006e-06, "loss": 0.0408, "step": 4610 }, { "epoch": 2.768124625524266, "grad_norm": 1.6639927625656128, "learning_rate": 3.954004438168247e-06, "loss": 0.0572, "step": 4620 }, { "epoch": 2.774116237267825, "grad_norm": 2.048002243041992, "learning_rate": 3.853136978010894e-06, "loss": 0.0535, "step": 4630 }, { "epoch": 2.780107849011384, "grad_norm": 1.2363957166671753, "learning_rate": 3.7522695178535406e-06, "loss": 0.0381, "step": 4640 }, { "epoch": 2.786099460754943, "grad_norm": 1.5503149032592773, "learning_rate": 3.651402057696188e-06, "loss": 0.0606, "step": 4650 }, { "epoch": 2.792091072498502, "grad_norm": 0.9487207531929016, "learning_rate": 3.5505345975388343e-06, "loss": 0.0441, "step": 4660 }, { "epoch": 2.798082684242061, "grad_norm": 1.1120935678482056, "learning_rate": 3.4496671373814807e-06, "loss": 0.0504, "step": 4670 }, { "epoch": 2.80407429598562, "grad_norm": 1.4742374420166016, "learning_rate": 3.348799677224128e-06, "loss": 0.0482, "step": 4680 }, { "epoch": 2.810065907729179, "grad_norm": 1.9465519189834595, "learning_rate": 3.2479322170667743e-06, "loss": 0.0376, "step": 4690 }, { "epoch": 2.8160575194727384, "grad_norm": 1.6899336576461792, "learning_rate": 3.1470647569094215e-06, "loss": 0.0459, "step": 4700 }, { "epoch": 2.822049131216297, "grad_norm": 1.3006277084350586, "learning_rate": 3.046197296752068e-06, "loss": 0.0426, "step": 4710 }, { "epoch": 2.828040742959856, "grad_norm": 1.4417216777801514, "learning_rate": 2.9453298365947148e-06, "loss": 0.047, "step": 4720 }, { "epoch": 2.834032354703415, "grad_norm": 1.2588374614715576, "learning_rate": 2.844462376437361e-06, "loss": 0.0483, "step": 4730 }, { "epoch": 2.8400239664469744, "grad_norm": 1.2371790409088135, "learning_rate": 2.743594916280008e-06, "loss": 0.0506, "step": 4740 }, { "epoch": 2.846015578190533, "grad_norm": 1.0114707946777344, "learning_rate": 2.642727456122655e-06, "loss": 0.0378, "step": 4750 }, { "epoch": 2.8520071899340924, "grad_norm": 2.299523115158081, "learning_rate": 2.5418599959653016e-06, "loss": 0.0602, "step": 4760 }, { "epoch": 2.857998801677651, "grad_norm": 1.0991910696029663, "learning_rate": 2.4409925358079484e-06, "loss": 0.0536, "step": 4770 }, { "epoch": 2.8639904134212104, "grad_norm": 1.8296887874603271, "learning_rate": 2.3401250756505953e-06, "loss": 0.0381, "step": 4780 }, { "epoch": 2.869982025164769, "grad_norm": 1.669256329536438, "learning_rate": 2.239257615493242e-06, "loss": 0.039, "step": 4790 }, { "epoch": 2.8759736369083284, "grad_norm": 1.4977699518203735, "learning_rate": 2.138390155335889e-06, "loss": 0.0461, "step": 4800 }, { "epoch": 2.8819652486518876, "grad_norm": 1.5286149978637695, "learning_rate": 2.0375226951785357e-06, "loss": 0.0541, "step": 4810 }, { "epoch": 2.8879568603954464, "grad_norm": 1.9209306240081787, "learning_rate": 1.936655235021182e-06, "loss": 0.0584, "step": 4820 }, { "epoch": 2.893948472139005, "grad_norm": 2.085172653198242, "learning_rate": 1.835787774863829e-06, "loss": 0.0417, "step": 4830 }, { "epoch": 2.8999400838825644, "grad_norm": 0.9194634556770325, "learning_rate": 1.7349203147064758e-06, "loss": 0.0412, "step": 4840 }, { "epoch": 2.9059316956261236, "grad_norm": 0.8978023529052734, "learning_rate": 1.6340528545491226e-06, "loss": 0.0475, "step": 4850 }, { "epoch": 2.9119233073696824, "grad_norm": 1.5397992134094238, "learning_rate": 1.5331853943917692e-06, "loss": 0.0411, "step": 4860 }, { "epoch": 2.917914919113241, "grad_norm": 2.272191047668457, "learning_rate": 1.432317934234416e-06, "loss": 0.0442, "step": 4870 }, { "epoch": 2.9239065308568004, "grad_norm": 1.2417062520980835, "learning_rate": 1.3314504740770626e-06, "loss": 0.048, "step": 4880 }, { "epoch": 2.9298981426003596, "grad_norm": 0.9558309316635132, "learning_rate": 1.2305830139197097e-06, "loss": 0.0411, "step": 4890 }, { "epoch": 2.9358897543439184, "grad_norm": 1.373119592666626, "learning_rate": 1.1297155537623565e-06, "loss": 0.055, "step": 4900 }, { "epoch": 2.9418813660874776, "grad_norm": 1.019445538520813, "learning_rate": 1.028848093605003e-06, "loss": 0.0464, "step": 4910 }, { "epoch": 2.9478729778310364, "grad_norm": 2.2195796966552734, "learning_rate": 9.279806334476499e-07, "loss": 0.0584, "step": 4920 }, { "epoch": 2.9538645895745956, "grad_norm": 1.278170108795166, "learning_rate": 8.271131732902965e-07, "loss": 0.0334, "step": 4930 }, { "epoch": 2.9598562013181544, "grad_norm": 2.264543294906616, "learning_rate": 7.262457131329434e-07, "loss": 0.0549, "step": 4940 }, { "epoch": 2.9658478130617136, "grad_norm": 1.332661509513855, "learning_rate": 6.253782529755901e-07, "loss": 0.0483, "step": 4950 }, { "epoch": 2.971839424805273, "grad_norm": 0.6639212965965271, "learning_rate": 5.245107928182369e-07, "loss": 0.0399, "step": 4960 }, { "epoch": 2.9778310365488316, "grad_norm": 1.3559752702713013, "learning_rate": 4.236433326608836e-07, "loss": 0.0456, "step": 4970 }, { "epoch": 2.9838226482923904, "grad_norm": 2.040262222290039, "learning_rate": 3.2277587250353037e-07, "loss": 0.0604, "step": 4980 }, { "epoch": 2.9898142600359496, "grad_norm": 1.7835615873336792, "learning_rate": 2.2190841234617714e-07, "loss": 0.048, "step": 4990 }, { "epoch": 2.995805871779509, "grad_norm": 1.4808944463729858, "learning_rate": 1.2104095218882388e-07, "loss": 0.05, "step": 5000 }, { "epoch": 2.995805871779509, "eval_loss": 0.3848415017127991, "eval_runtime": 2452.1783, "eval_samples_per_second": 0.578, "eval_steps_per_second": 0.036, "eval_wer": 0.3763063217406202, "step": 5000 } ], "logging_steps": 10, "max_steps": 5007, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 2.30868320256e+19, "train_batch_size": 16, "trial_name": null, "trial_params": null }