{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9409887541615739, "eval_steps": 500, "global_step": 46000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0002045627726438204, "grad_norm": 1322.4091796875, "learning_rate": 1.226993865030675e-07, "loss": 2.497515106201172, "step": 10 }, { "epoch": 0.0004091255452876408, "grad_norm": 48.41965103149414, "learning_rate": 2.5903203817314247e-07, "loss": 1.0519317626953124, "step": 20 }, { "epoch": 0.0006136883179314613, "grad_norm": 59.999263763427734, "learning_rate": 3.953646898432175e-07, "loss": 0.33386669158935545, "step": 30 }, { "epoch": 0.0008182510905752816, "grad_norm": 20.514692306518555, "learning_rate": 5.316973415132925e-07, "loss": 0.3147749662399292, "step": 40 }, { "epoch": 0.0010228138632191021, "grad_norm": 18.04642105102539, "learning_rate": 6.680299931833676e-07, "loss": 0.28726825714111326, "step": 50 }, { "epoch": 0.0012273766358629225, "grad_norm": 11.525564193725586, "learning_rate": 8.043626448534425e-07, "loss": 0.270180869102478, "step": 60 }, { "epoch": 0.0014319394085067429, "grad_norm": 9.384416580200195, "learning_rate": 9.406952965235175e-07, "loss": 0.27215096950531004, "step": 70 }, { "epoch": 0.0016365021811505633, "grad_norm": 10.75300407409668, "learning_rate": 1.0770279481935925e-06, "loss": 0.28212790489196776, "step": 80 }, { "epoch": 0.0018410649537943836, "grad_norm": 11.66346549987793, "learning_rate": 1.2133605998636674e-06, "loss": 0.26888844966888426, "step": 90 }, { "epoch": 0.0020456277264382042, "grad_norm": 9.192279815673828, "learning_rate": 1.3496932515337425e-06, "loss": 0.26565074920654297, "step": 100 }, { "epoch": 0.0022501904990820244, "grad_norm": 10.468182563781738, "learning_rate": 1.4860259032038174e-06, "loss": 0.2623821020126343, "step": 110 }, { "epoch": 0.002454753271725845, "grad_norm": 19.531068801879883, "learning_rate": 1.6223585548738923e-06, "loss": 0.2785211086273193, "step": 120 }, { "epoch": 0.002659316044369665, "grad_norm": 12.359622955322266, "learning_rate": 1.7586912065439674e-06, "loss": 0.2741536140441895, "step": 130 }, { "epoch": 0.0028638788170134858, "grad_norm": 16.013368606567383, "learning_rate": 1.8950238582140423e-06, "loss": 0.2747157573699951, "step": 140 }, { "epoch": 0.0030684415896573064, "grad_norm": 22.088977813720703, "learning_rate": 2.031356509884117e-06, "loss": 0.28859868049621584, "step": 150 }, { "epoch": 0.0032730043623011265, "grad_norm": 29.792007446289062, "learning_rate": 2.1676891615541925e-06, "loss": 0.2553713798522949, "step": 160 }, { "epoch": 0.003477567134944947, "grad_norm": 9.39339828491211, "learning_rate": 2.3040218132242673e-06, "loss": 0.25037808418273927, "step": 170 }, { "epoch": 0.0036821299075887673, "grad_norm": 13.860047340393066, "learning_rate": 2.4403544648943427e-06, "loss": 0.2612645626068115, "step": 180 }, { "epoch": 0.003886692680232588, "grad_norm": 22.42649269104004, "learning_rate": 2.5766871165644175e-06, "loss": 0.25780482292175294, "step": 190 }, { "epoch": 0.0040912554528764085, "grad_norm": 14.081650733947754, "learning_rate": 2.713019768234492e-06, "loss": 0.24367117881774902, "step": 200 }, { "epoch": 0.004295818225520229, "grad_norm": 16.992481231689453, "learning_rate": 2.8493524199045673e-06, "loss": 0.2279716968536377, "step": 210 }, { "epoch": 0.004500380998164049, "grad_norm": 8.51474666595459, "learning_rate": 2.985685071574642e-06, "loss": 0.26891887187957764, "step": 220 }, { "epoch": 0.004704943770807869, "grad_norm": 6.3475341796875, "learning_rate": 3.1220177232447175e-06, "loss": 0.2477013111114502, "step": 230 }, { "epoch": 0.00490950654345169, "grad_norm": 11.599653244018555, "learning_rate": 3.258350374914792e-06, "loss": 0.25707106590270995, "step": 240 }, { "epoch": 0.005114069316095511, "grad_norm": 13.862350463867188, "learning_rate": 3.3946830265848673e-06, "loss": 0.25145750045776366, "step": 250 }, { "epoch": 0.00531863208873933, "grad_norm": 20.093097686767578, "learning_rate": 3.5310156782549426e-06, "loss": 0.2787784576416016, "step": 260 }, { "epoch": 0.005523194861383151, "grad_norm": 19.383190155029297, "learning_rate": 3.6673483299250175e-06, "loss": 0.26395373344421386, "step": 270 }, { "epoch": 0.0057277576340269715, "grad_norm": 11.810567855834961, "learning_rate": 3.8036809815950928e-06, "loss": 0.2555896520614624, "step": 280 }, { "epoch": 0.005932320406670792, "grad_norm": 15.424582481384277, "learning_rate": 3.940013633265167e-06, "loss": 0.26258389949798583, "step": 290 }, { "epoch": 0.006136883179314613, "grad_norm": 9.57666301727295, "learning_rate": 4.076346284935242e-06, "loss": 0.25736584663391116, "step": 300 }, { "epoch": 0.0063414459519584325, "grad_norm": 16.089977264404297, "learning_rate": 4.212678936605317e-06, "loss": 0.25005536079406737, "step": 310 }, { "epoch": 0.006546008724602253, "grad_norm": 16.00985336303711, "learning_rate": 4.349011588275393e-06, "loss": 0.24516136646270753, "step": 320 }, { "epoch": 0.006750571497246074, "grad_norm": 15.225984573364258, "learning_rate": 4.485344239945467e-06, "loss": 0.24549686908721924, "step": 330 }, { "epoch": 0.006955134269889894, "grad_norm": 4.800993919372559, "learning_rate": 4.6216768916155425e-06, "loss": 0.24107637405395507, "step": 340 }, { "epoch": 0.007159697042533715, "grad_norm": 7.910202980041504, "learning_rate": 4.758009543285617e-06, "loss": 0.2549219846725464, "step": 350 }, { "epoch": 0.007364259815177535, "grad_norm": 15.663596153259277, "learning_rate": 4.894342194955692e-06, "loss": 0.25311932563781736, "step": 360 }, { "epoch": 0.007568822587821355, "grad_norm": 11.66611099243164, "learning_rate": 5.030674846625767e-06, "loss": 0.25794410705566406, "step": 370 }, { "epoch": 0.007773385360465176, "grad_norm": 10.642110824584961, "learning_rate": 5.167007498295842e-06, "loss": 0.2601644992828369, "step": 380 }, { "epoch": 0.007977948133108996, "grad_norm": 8.631386756896973, "learning_rate": 5.303340149965918e-06, "loss": 0.2338024854660034, "step": 390 }, { "epoch": 0.008182510905752817, "grad_norm": 6.575963973999023, "learning_rate": 5.439672801635993e-06, "loss": 0.25810861587524414, "step": 400 }, { "epoch": 0.008387073678396638, "grad_norm": 5.1855669021606445, "learning_rate": 5.576005453306067e-06, "loss": 0.24390020370483398, "step": 410 }, { "epoch": 0.008591636451040458, "grad_norm": 16.527450561523438, "learning_rate": 5.712338104976142e-06, "loss": 0.2471834182739258, "step": 420 }, { "epoch": 0.008796199223684277, "grad_norm": 10.236855506896973, "learning_rate": 5.848670756646217e-06, "loss": 0.24280643463134766, "step": 430 }, { "epoch": 0.009000761996328098, "grad_norm": 11.396493911743164, "learning_rate": 5.985003408316292e-06, "loss": 0.24722771644592284, "step": 440 }, { "epoch": 0.009205324768971918, "grad_norm": 12.100617408752441, "learning_rate": 6.121336059986367e-06, "loss": 0.24162616729736328, "step": 450 }, { "epoch": 0.009409887541615739, "grad_norm": 5.7156548500061035, "learning_rate": 6.257668711656443e-06, "loss": 0.24008772373199463, "step": 460 }, { "epoch": 0.00961445031425956, "grad_norm": 7.945430755615234, "learning_rate": 6.394001363326518e-06, "loss": 0.24152028560638428, "step": 470 }, { "epoch": 0.00981901308690338, "grad_norm": 10.174844741821289, "learning_rate": 6.530334014996592e-06, "loss": 0.24351162910461427, "step": 480 }, { "epoch": 0.0100235758595472, "grad_norm": 10.981884956359863, "learning_rate": 6.666666666666667e-06, "loss": 0.2503940582275391, "step": 490 }, { "epoch": 0.010228138632191021, "grad_norm": 9.064326286315918, "learning_rate": 6.802999318336742e-06, "loss": 0.25287790298461915, "step": 500 }, { "epoch": 0.010432701404834842, "grad_norm": 11.30699634552002, "learning_rate": 6.939331970006817e-06, "loss": 0.24177236557006837, "step": 510 }, { "epoch": 0.01063726417747866, "grad_norm": 15.712932586669922, "learning_rate": 7.075664621676892e-06, "loss": 0.2306075096130371, "step": 520 }, { "epoch": 0.010841826950122481, "grad_norm": 4.898688793182373, "learning_rate": 7.211997273346967e-06, "loss": 0.24057083129882811, "step": 530 }, { "epoch": 0.011046389722766302, "grad_norm": 5.052396297454834, "learning_rate": 7.348329925017043e-06, "loss": 0.2333120346069336, "step": 540 }, { "epoch": 0.011250952495410122, "grad_norm": 7.262669086456299, "learning_rate": 7.484662576687118e-06, "loss": 0.256821346282959, "step": 550 }, { "epoch": 0.011455515268053943, "grad_norm": 3.87188982963562, "learning_rate": 7.620995228357192e-06, "loss": 0.23729300498962402, "step": 560 }, { "epoch": 0.011660078040697764, "grad_norm": 6.580966472625732, "learning_rate": 7.757327880027267e-06, "loss": 0.24562397003173828, "step": 570 }, { "epoch": 0.011864640813341584, "grad_norm": 8.317594528198242, "learning_rate": 7.893660531697342e-06, "loss": 0.24440286159515381, "step": 580 }, { "epoch": 0.012069203585985405, "grad_norm": 9.459776878356934, "learning_rate": 8.029993183367417e-06, "loss": 0.2650471210479736, "step": 590 }, { "epoch": 0.012273766358629225, "grad_norm": 4.107361316680908, "learning_rate": 8.166325835037492e-06, "loss": 0.25248258113861083, "step": 600 }, { "epoch": 0.012478329131273046, "grad_norm": 10.486988067626953, "learning_rate": 8.302658486707567e-06, "loss": 0.23548486232757568, "step": 610 }, { "epoch": 0.012682891903916865, "grad_norm": 5.970145225524902, "learning_rate": 8.438991138377642e-06, "loss": 0.23853745460510253, "step": 620 }, { "epoch": 0.012887454676560686, "grad_norm": 8.760467529296875, "learning_rate": 8.575323790047717e-06, "loss": 0.24659357070922852, "step": 630 }, { "epoch": 0.013092017449204506, "grad_norm": 8.476232528686523, "learning_rate": 8.711656441717792e-06, "loss": 0.22962665557861328, "step": 640 }, { "epoch": 0.013296580221848327, "grad_norm": 9.617321968078613, "learning_rate": 8.847989093387868e-06, "loss": 0.24794669151306153, "step": 650 }, { "epoch": 0.013501142994492147, "grad_norm": 5.2855987548828125, "learning_rate": 8.984321745057943e-06, "loss": 0.23107781410217285, "step": 660 }, { "epoch": 0.013705705767135968, "grad_norm": 5.700791835784912, "learning_rate": 9.120654396728016e-06, "loss": 0.2604004144668579, "step": 670 }, { "epoch": 0.013910268539779789, "grad_norm": 6.054930210113525, "learning_rate": 9.256987048398091e-06, "loss": 0.23733630180358886, "step": 680 }, { "epoch": 0.014114831312423609, "grad_norm": 8.92485237121582, "learning_rate": 9.393319700068168e-06, "loss": 0.2559273481369019, "step": 690 }, { "epoch": 0.01431939408506743, "grad_norm": 13.244160652160645, "learning_rate": 9.529652351738243e-06, "loss": 0.2447065830230713, "step": 700 }, { "epoch": 0.014523956857711249, "grad_norm": 6.798666477203369, "learning_rate": 9.665985003408318e-06, "loss": 0.25516366958618164, "step": 710 }, { "epoch": 0.01472851963035507, "grad_norm": 12.444178581237793, "learning_rate": 9.802317655078392e-06, "loss": 0.2647365093231201, "step": 720 }, { "epoch": 0.01493308240299889, "grad_norm": 8.227578163146973, "learning_rate": 9.938650306748467e-06, "loss": 0.24309248924255372, "step": 730 }, { "epoch": 0.01513764517564271, "grad_norm": 10.260855674743652, "learning_rate": 1.0074982958418542e-05, "loss": 0.24093937873840332, "step": 740 }, { "epoch": 0.015342207948286531, "grad_norm": 8.81863784790039, "learning_rate": 1.0211315610088617e-05, "loss": 0.2509352207183838, "step": 750 }, { "epoch": 0.015546770720930352, "grad_norm": 10.271194458007812, "learning_rate": 1.0347648261758692e-05, "loss": 0.2532621383666992, "step": 760 }, { "epoch": 0.01575133349357417, "grad_norm": 7.823755741119385, "learning_rate": 1.0483980913428767e-05, "loss": 0.2455000400543213, "step": 770 }, { "epoch": 0.015955896266217993, "grad_norm": 8.706862449645996, "learning_rate": 1.0620313565098843e-05, "loss": 0.24046711921691893, "step": 780 }, { "epoch": 0.01616045903886181, "grad_norm": 5.840294361114502, "learning_rate": 1.0756646216768918e-05, "loss": 0.26148841381072996, "step": 790 }, { "epoch": 0.016365021811505634, "grad_norm": 7.217397689819336, "learning_rate": 1.0892978868438993e-05, "loss": 0.2459665298461914, "step": 800 }, { "epoch": 0.016569584584149453, "grad_norm": 7.5736799240112305, "learning_rate": 1.1029311520109068e-05, "loss": 0.25462751388549804, "step": 810 }, { "epoch": 0.016774147356793275, "grad_norm": 8.155405044555664, "learning_rate": 1.1165644171779141e-05, "loss": 0.24461240768432618, "step": 820 }, { "epoch": 0.016978710129437094, "grad_norm": 12.247174263000488, "learning_rate": 1.1301976823449216e-05, "loss": 0.23934953212738036, "step": 830 }, { "epoch": 0.017183272902080916, "grad_norm": 7.307239532470703, "learning_rate": 1.1438309475119291e-05, "loss": 0.2308333158493042, "step": 840 }, { "epoch": 0.017387835674724735, "grad_norm": 6.284276485443115, "learning_rate": 1.1574642126789366e-05, "loss": 0.241119384765625, "step": 850 }, { "epoch": 0.017592398447368554, "grad_norm": 7.363729000091553, "learning_rate": 1.1710974778459442e-05, "loss": 0.24161911010742188, "step": 860 }, { "epoch": 0.017796961220012376, "grad_norm": 4.721088409423828, "learning_rate": 1.1847307430129517e-05, "loss": 0.2243041515350342, "step": 870 }, { "epoch": 0.018001523992656195, "grad_norm": 3.2835986614227295, "learning_rate": 1.1983640081799592e-05, "loss": 0.23877978324890137, "step": 880 }, { "epoch": 0.018206086765300018, "grad_norm": 3.2119874954223633, "learning_rate": 1.2119972733469667e-05, "loss": 0.2254627227783203, "step": 890 }, { "epoch": 0.018410649537943836, "grad_norm": 11.27684497833252, "learning_rate": 1.2256305385139742e-05, "loss": 0.22723329067230225, "step": 900 }, { "epoch": 0.01861521231058766, "grad_norm": 5.16454553604126, "learning_rate": 1.2392638036809817e-05, "loss": 0.24502649307250976, "step": 910 }, { "epoch": 0.018819775083231478, "grad_norm": 6.6756181716918945, "learning_rate": 1.2528970688479892e-05, "loss": 0.25124826431274416, "step": 920 }, { "epoch": 0.0190243378558753, "grad_norm": 1.9527130126953125, "learning_rate": 1.2665303340149967e-05, "loss": 0.23971734046936036, "step": 930 }, { "epoch": 0.01922890062851912, "grad_norm": 3.1325552463531494, "learning_rate": 1.2801635991820043e-05, "loss": 0.23688170909881592, "step": 940 }, { "epoch": 0.019433463401162938, "grad_norm": 3.955563545227051, "learning_rate": 1.2937968643490118e-05, "loss": 0.2558871030807495, "step": 950 }, { "epoch": 0.01963802617380676, "grad_norm": 3.817612409591675, "learning_rate": 1.3074301295160191e-05, "loss": 0.23308591842651366, "step": 960 }, { "epoch": 0.01984258894645058, "grad_norm": 4.015598773956299, "learning_rate": 1.3210633946830266e-05, "loss": 0.242836332321167, "step": 970 }, { "epoch": 0.0200471517190944, "grad_norm": 5.876497268676758, "learning_rate": 1.3346966598500341e-05, "loss": 0.2353851318359375, "step": 980 }, { "epoch": 0.02025171449173822, "grad_norm": 3.2260093688964844, "learning_rate": 1.3483299250170416e-05, "loss": 0.2296689748764038, "step": 990 }, { "epoch": 0.020456277264382042, "grad_norm": 6.555721759796143, "learning_rate": 1.3619631901840491e-05, "loss": 0.23085966110229492, "step": 1000 }, { "epoch": 0.02066084003702586, "grad_norm": 6.5157856941223145, "learning_rate": 1.3755964553510567e-05, "loss": 0.23325190544128419, "step": 1010 }, { "epoch": 0.020865402809669684, "grad_norm": 4.8217854499816895, "learning_rate": 1.3892297205180642e-05, "loss": 0.24951257705688476, "step": 1020 }, { "epoch": 0.021069965582313503, "grad_norm": 6.93559455871582, "learning_rate": 1.4028629856850717e-05, "loss": 0.24221696853637695, "step": 1030 }, { "epoch": 0.02127452835495732, "grad_norm": 5.210307598114014, "learning_rate": 1.4164962508520792e-05, "loss": 0.23380188941955565, "step": 1040 }, { "epoch": 0.021479091127601144, "grad_norm": 9.44946002960205, "learning_rate": 1.4301295160190867e-05, "loss": 0.2348261594772339, "step": 1050 }, { "epoch": 0.021683653900244963, "grad_norm": 4.345777988433838, "learning_rate": 1.4437627811860942e-05, "loss": 0.2554929733276367, "step": 1060 }, { "epoch": 0.021888216672888785, "grad_norm": 6.244737148284912, "learning_rate": 1.4573960463531017e-05, "loss": 0.2546764135360718, "step": 1070 }, { "epoch": 0.022092779445532604, "grad_norm": 3.7343943119049072, "learning_rate": 1.471029311520109e-05, "loss": 0.2443603515625, "step": 1080 }, { "epoch": 0.022297342218176426, "grad_norm": 12.457460403442383, "learning_rate": 1.4846625766871168e-05, "loss": 0.23022689819335937, "step": 1090 }, { "epoch": 0.022501904990820245, "grad_norm": 3.6894562244415283, "learning_rate": 1.4982958418541243e-05, "loss": 0.23204188346862792, "step": 1100 }, { "epoch": 0.022706467763464067, "grad_norm": 4.578464031219482, "learning_rate": 1.5119291070211316e-05, "loss": 0.23379945755004883, "step": 1110 }, { "epoch": 0.022911030536107886, "grad_norm": 7.050217151641846, "learning_rate": 1.5255623721881391e-05, "loss": 0.2370091199874878, "step": 1120 }, { "epoch": 0.02311559330875171, "grad_norm": 9.051339149475098, "learning_rate": 1.5391956373551468e-05, "loss": 0.23648247718811036, "step": 1130 }, { "epoch": 0.023320156081395527, "grad_norm": 5.062636375427246, "learning_rate": 1.5528289025221543e-05, "loss": 0.23285813331604005, "step": 1140 }, { "epoch": 0.023524718854039346, "grad_norm": 3.9475834369659424, "learning_rate": 1.5664621676891618e-05, "loss": 0.2275324583053589, "step": 1150 }, { "epoch": 0.02372928162668317, "grad_norm": 4.990899085998535, "learning_rate": 1.5800954328561692e-05, "loss": 0.24655742645263673, "step": 1160 }, { "epoch": 0.023933844399326987, "grad_norm": 2.9217071533203125, "learning_rate": 1.5937286980231767e-05, "loss": 0.24099085330963135, "step": 1170 }, { "epoch": 0.02413840717197081, "grad_norm": 9.939743041992188, "learning_rate": 1.6073619631901842e-05, "loss": 0.2466895341873169, "step": 1180 }, { "epoch": 0.02434296994461463, "grad_norm": 5.999149322509766, "learning_rate": 1.6209952283571917e-05, "loss": 0.24030523300170897, "step": 1190 }, { "epoch": 0.02454753271725845, "grad_norm": 2.884174108505249, "learning_rate": 1.6346284935241992e-05, "loss": 0.24676980972290039, "step": 1200 }, { "epoch": 0.02475209548990227, "grad_norm": 3.1035685539245605, "learning_rate": 1.6482617586912067e-05, "loss": 0.22700700759887696, "step": 1210 }, { "epoch": 0.024956658262546092, "grad_norm": 3.7177746295928955, "learning_rate": 1.6618950238582142e-05, "loss": 0.24163634777069093, "step": 1220 }, { "epoch": 0.02516122103518991, "grad_norm": 11.649608612060547, "learning_rate": 1.6755282890252217e-05, "loss": 0.22782137393951415, "step": 1230 }, { "epoch": 0.02536578380783373, "grad_norm": 5.20341682434082, "learning_rate": 1.689161554192229e-05, "loss": 0.24848592281341553, "step": 1240 }, { "epoch": 0.025570346580477552, "grad_norm": 8.562382698059082, "learning_rate": 1.7027948193592366e-05, "loss": 0.24461679458618163, "step": 1250 }, { "epoch": 0.02577490935312137, "grad_norm": 4.436246871948242, "learning_rate": 1.716428084526244e-05, "loss": 0.25495865345001223, "step": 1260 }, { "epoch": 0.025979472125765193, "grad_norm": 6.774163246154785, "learning_rate": 1.7300613496932516e-05, "loss": 0.24368605613708497, "step": 1270 }, { "epoch": 0.026184034898409012, "grad_norm": 3.9284379482269287, "learning_rate": 1.743694614860259e-05, "loss": 0.23642210960388182, "step": 1280 }, { "epoch": 0.026388597671052835, "grad_norm": 2.610867738723755, "learning_rate": 1.7573278800272666e-05, "loss": 0.22842812538146973, "step": 1290 }, { "epoch": 0.026593160443696653, "grad_norm": 9.507801055908203, "learning_rate": 1.770961145194274e-05, "loss": 0.2367171049118042, "step": 1300 }, { "epoch": 0.026797723216340476, "grad_norm": 5.404077529907227, "learning_rate": 1.784594410361282e-05, "loss": 0.22744860649108886, "step": 1310 }, { "epoch": 0.027002285988984295, "grad_norm": 2.946303367614746, "learning_rate": 1.7982276755282894e-05, "loss": 0.2508368015289307, "step": 1320 }, { "epoch": 0.027206848761628113, "grad_norm": 6.781836986541748, "learning_rate": 1.811860940695297e-05, "loss": 0.2442564010620117, "step": 1330 }, { "epoch": 0.027411411534271936, "grad_norm": 4.447707176208496, "learning_rate": 1.825494205862304e-05, "loss": 0.22991580963134767, "step": 1340 }, { "epoch": 0.027615974306915755, "grad_norm": 4.919931888580322, "learning_rate": 1.8391274710293115e-05, "loss": 0.23134150505065917, "step": 1350 }, { "epoch": 0.027820537079559577, "grad_norm": 2.5548417568206787, "learning_rate": 1.852760736196319e-05, "loss": 0.24369499683380128, "step": 1360 }, { "epoch": 0.028025099852203396, "grad_norm": 2.7064099311828613, "learning_rate": 1.8663940013633265e-05, "loss": 0.23427906036376953, "step": 1370 }, { "epoch": 0.028229662624847218, "grad_norm": 9.153219223022461, "learning_rate": 1.880027266530334e-05, "loss": 0.2305999994277954, "step": 1380 }, { "epoch": 0.028434225397491037, "grad_norm": 8.615388870239258, "learning_rate": 1.8936605316973418e-05, "loss": 0.23182849884033202, "step": 1390 }, { "epoch": 0.02863878817013486, "grad_norm": 3.289309501647949, "learning_rate": 1.9072937968643493e-05, "loss": 0.24958360195159912, "step": 1400 }, { "epoch": 0.02884335094277868, "grad_norm": 4.615565776824951, "learning_rate": 1.9209270620313568e-05, "loss": 0.22406892776489257, "step": 1410 }, { "epoch": 0.029047913715422497, "grad_norm": 2.141252040863037, "learning_rate": 1.9345603271983643e-05, "loss": 0.22823643684387207, "step": 1420 }, { "epoch": 0.02925247648806632, "grad_norm": 2.5040884017944336, "learning_rate": 1.9481935923653718e-05, "loss": 0.2513998746871948, "step": 1430 }, { "epoch": 0.02945703926071014, "grad_norm": 5.795921325683594, "learning_rate": 1.9618268575323793e-05, "loss": 0.2422318458557129, "step": 1440 }, { "epoch": 0.02966160203335396, "grad_norm": 5.276379585266113, "learning_rate": 1.9754601226993868e-05, "loss": 0.22441024780273439, "step": 1450 }, { "epoch": 0.02986616480599778, "grad_norm": 5.921865463256836, "learning_rate": 1.989093387866394e-05, "loss": 0.23780419826507568, "step": 1460 }, { "epoch": 0.030070727578641602, "grad_norm": 5.715939521789551, "learning_rate": 1.999999991221036e-05, "loss": 0.22156381607055664, "step": 1470 }, { "epoch": 0.03027529035128542, "grad_norm": 6.823939323425293, "learning_rate": 1.9999996839573068e-05, "loss": 0.2474888324737549, "step": 1480 }, { "epoch": 0.030479853123929243, "grad_norm": 8.417153358459473, "learning_rate": 1.9999989377455237e-05, "loss": 0.22731573581695558, "step": 1490 }, { "epoch": 0.030684415896573062, "grad_norm": 2.8374781608581543, "learning_rate": 1.999997752586015e-05, "loss": 0.2284107208251953, "step": 1500 }, { "epoch": 0.030888978669216884, "grad_norm": 5.383684158325195, "learning_rate": 1.9999961284793004e-05, "loss": 0.25558278560638426, "step": 1510 }, { "epoch": 0.031093541441860703, "grad_norm": 3.3066437244415283, "learning_rate": 1.9999940654260928e-05, "loss": 0.22794761657714843, "step": 1520 }, { "epoch": 0.031298104214504525, "grad_norm": 4.142764568328857, "learning_rate": 1.9999915634272978e-05, "loss": 0.22768206596374513, "step": 1530 }, { "epoch": 0.03150266698714834, "grad_norm": 8.678051948547363, "learning_rate": 1.9999886224840135e-05, "loss": 0.2392820119857788, "step": 1540 }, { "epoch": 0.03170722975979216, "grad_norm": 1.4955272674560547, "learning_rate": 1.9999852425975312e-05, "loss": 0.2528279542922974, "step": 1550 }, { "epoch": 0.031911792532435986, "grad_norm": 4.196108341217041, "learning_rate": 1.9999814237693345e-05, "loss": 0.24402837753295897, "step": 1560 }, { "epoch": 0.03211635530507981, "grad_norm": 5.532372951507568, "learning_rate": 1.999977166001099e-05, "loss": 0.22701892852783204, "step": 1570 }, { "epoch": 0.03232091807772362, "grad_norm": 4.543706893920898, "learning_rate": 1.9999724692946944e-05, "loss": 0.24147312641143798, "step": 1580 }, { "epoch": 0.032525480850367446, "grad_norm": 3.561648368835449, "learning_rate": 1.9999673336521823e-05, "loss": 0.23969759941101074, "step": 1590 }, { "epoch": 0.03273004362301127, "grad_norm": 3.5796561241149902, "learning_rate": 1.9999617590758166e-05, "loss": 0.22214760780334472, "step": 1600 }, { "epoch": 0.03293460639565508, "grad_norm": 2.6527392864227295, "learning_rate": 1.999955745568044e-05, "loss": 0.2306440830230713, "step": 1610 }, { "epoch": 0.033139169168298906, "grad_norm": 2.5623395442962646, "learning_rate": 1.9999492931315045e-05, "loss": 0.23867559432983398, "step": 1620 }, { "epoch": 0.03334373194094273, "grad_norm": 6.984724998474121, "learning_rate": 1.9999424017690306e-05, "loss": 0.23410024642944335, "step": 1630 }, { "epoch": 0.03354829471358655, "grad_norm": 3.238554000854492, "learning_rate": 1.9999350714836468e-05, "loss": 0.23847858905792235, "step": 1640 }, { "epoch": 0.033752857486230366, "grad_norm": 7.299864292144775, "learning_rate": 1.999927302278571e-05, "loss": 0.23235070705413818, "step": 1650 }, { "epoch": 0.03395742025887419, "grad_norm": 4.142194747924805, "learning_rate": 1.9999190941572137e-05, "loss": 0.2489847183227539, "step": 1660 }, { "epoch": 0.03416198303151801, "grad_norm": 6.413217544555664, "learning_rate": 1.9999104471231775e-05, "loss": 0.22639312744140624, "step": 1670 }, { "epoch": 0.03436654580416183, "grad_norm": 9.625746726989746, "learning_rate": 1.999901361180258e-05, "loss": 0.2562007188796997, "step": 1680 }, { "epoch": 0.03457110857680565, "grad_norm": 6.126235485076904, "learning_rate": 1.999891836332443e-05, "loss": 0.23477575778961182, "step": 1690 }, { "epoch": 0.03477567134944947, "grad_norm": 3.9975342750549316, "learning_rate": 1.9998818725839143e-05, "loss": 0.23544483184814452, "step": 1700 }, { "epoch": 0.03498023412209329, "grad_norm": 5.323237419128418, "learning_rate": 1.9998714699390453e-05, "loss": 0.22694244384765624, "step": 1710 }, { "epoch": 0.03518479689473711, "grad_norm": 8.781648635864258, "learning_rate": 1.9998606284024017e-05, "loss": 0.22633099555969238, "step": 1720 }, { "epoch": 0.03538935966738093, "grad_norm": 4.632314682006836, "learning_rate": 1.9998493479787426e-05, "loss": 0.22183871269226074, "step": 1730 }, { "epoch": 0.03559392244002475, "grad_norm": 5.12542200088501, "learning_rate": 1.99983762867302e-05, "loss": 0.2292017698287964, "step": 1740 }, { "epoch": 0.035798485212668575, "grad_norm": 3.480116844177246, "learning_rate": 1.9998254704903775e-05, "loss": 0.22254834175109864, "step": 1750 }, { "epoch": 0.03600304798531239, "grad_norm": 2.4217300415039062, "learning_rate": 1.999812873436152e-05, "loss": 0.22176332473754884, "step": 1760 }, { "epoch": 0.03620761075795621, "grad_norm": 3.350013256072998, "learning_rate": 1.999799837515873e-05, "loss": 0.2308824062347412, "step": 1770 }, { "epoch": 0.036412173530600035, "grad_norm": 2.7582473754882812, "learning_rate": 1.9997863627352627e-05, "loss": 0.2328019142150879, "step": 1780 }, { "epoch": 0.03661673630324386, "grad_norm": 2.296842098236084, "learning_rate": 1.999772449100236e-05, "loss": 0.23819632530212403, "step": 1790 }, { "epoch": 0.03682129907588767, "grad_norm": 6.964822292327881, "learning_rate": 1.9997580966168997e-05, "loss": 0.21697392463684081, "step": 1800 }, { "epoch": 0.037025861848531495, "grad_norm": 8.177725791931152, "learning_rate": 1.9997433052915544e-05, "loss": 0.23704142570495607, "step": 1810 }, { "epoch": 0.03723042462117532, "grad_norm": 7.24596643447876, "learning_rate": 1.9997280751306922e-05, "loss": 0.23726887702941896, "step": 1820 }, { "epoch": 0.03743498739381913, "grad_norm": 6.032847881317139, "learning_rate": 1.9997124061409985e-05, "loss": 0.2284160852432251, "step": 1830 }, { "epoch": 0.037639550166462955, "grad_norm": 3.3339858055114746, "learning_rate": 1.9996962983293516e-05, "loss": 0.22085819244384766, "step": 1840 }, { "epoch": 0.03784411293910678, "grad_norm": 2.6622371673583984, "learning_rate": 1.9996797517028216e-05, "loss": 0.23529601097106934, "step": 1850 }, { "epoch": 0.0380486757117506, "grad_norm": 3.6977927684783936, "learning_rate": 1.9996627662686717e-05, "loss": 0.23434889316558838, "step": 1860 }, { "epoch": 0.038253238484394415, "grad_norm": 4.789235591888428, "learning_rate": 1.9996453420343572e-05, "loss": 0.22377028465270996, "step": 1870 }, { "epoch": 0.03845780125703824, "grad_norm": 7.804076671600342, "learning_rate": 1.9996274790075273e-05, "loss": 0.2337170124053955, "step": 1880 }, { "epoch": 0.03866236402968206, "grad_norm": 3.680337429046631, "learning_rate": 1.9996091771960226e-05, "loss": 0.2505925178527832, "step": 1890 }, { "epoch": 0.038866926802325875, "grad_norm": 4.174388408660889, "learning_rate": 1.9995904366078763e-05, "loss": 0.2556272506713867, "step": 1900 }, { "epoch": 0.0390714895749697, "grad_norm": 6.916759014129639, "learning_rate": 1.9995712572513145e-05, "loss": 0.2329261302947998, "step": 1910 }, { "epoch": 0.03927605234761352, "grad_norm": 5.425793647766113, "learning_rate": 1.9995516391347567e-05, "loss": 0.23824799060821533, "step": 1920 }, { "epoch": 0.03948061512025734, "grad_norm": 3.5766313076019287, "learning_rate": 1.9995315822668135e-05, "loss": 0.2322160243988037, "step": 1930 }, { "epoch": 0.03968517789290116, "grad_norm": 8.428255081176758, "learning_rate": 1.9995110866562893e-05, "loss": 0.24530091285705566, "step": 1940 }, { "epoch": 0.03988974066554498, "grad_norm": 2.1243624687194824, "learning_rate": 1.9994901523121802e-05, "loss": 0.23711862564086914, "step": 1950 }, { "epoch": 0.0400943034381888, "grad_norm": 3.699021816253662, "learning_rate": 1.9994687792436756e-05, "loss": 0.2368396043777466, "step": 1960 }, { "epoch": 0.040298866210832625, "grad_norm": 5.656952857971191, "learning_rate": 1.9994469674601568e-05, "loss": 0.2398841381072998, "step": 1970 }, { "epoch": 0.04050342898347644, "grad_norm": 4.257737636566162, "learning_rate": 1.999424716971199e-05, "loss": 0.22450828552246094, "step": 1980 }, { "epoch": 0.04070799175612026, "grad_norm": 4.136907577514648, "learning_rate": 1.999402027786568e-05, "loss": 0.22495903968811035, "step": 1990 }, { "epoch": 0.040912554528764085, "grad_norm": 5.321060657501221, "learning_rate": 1.9993788999162232e-05, "loss": 0.23318157196044922, "step": 2000 }, { "epoch": 0.0411171173014079, "grad_norm": 2.727281093597412, "learning_rate": 1.9993553333703167e-05, "loss": 0.23663928508758544, "step": 2010 }, { "epoch": 0.04132168007405172, "grad_norm": 2.2334887981414795, "learning_rate": 1.9993313281591935e-05, "loss": 0.23388416767120362, "step": 2020 }, { "epoch": 0.041526242846695545, "grad_norm": 3.891329050064087, "learning_rate": 1.99930688429339e-05, "loss": 0.21784915924072265, "step": 2030 }, { "epoch": 0.04173080561933937, "grad_norm": 3.7121307849884033, "learning_rate": 1.9992820017836367e-05, "loss": 0.21789174079895018, "step": 2040 }, { "epoch": 0.04193536839198318, "grad_norm": 3.4435536861419678, "learning_rate": 1.9992566806408545e-05, "loss": 0.2281360149383545, "step": 2050 }, { "epoch": 0.042139931164627005, "grad_norm": 7.706423759460449, "learning_rate": 1.9992309208761586e-05, "loss": 0.23873677253723144, "step": 2060 }, { "epoch": 0.04234449393727083, "grad_norm": 5.429510593414307, "learning_rate": 1.9992047225008563e-05, "loss": 0.2143171787261963, "step": 2070 }, { "epoch": 0.04254905670991464, "grad_norm": 5.18685245513916, "learning_rate": 1.9991780855264477e-05, "loss": 0.2274139404296875, "step": 2080 }, { "epoch": 0.042753619482558465, "grad_norm": 4.11610221862793, "learning_rate": 1.9991510099646244e-05, "loss": 0.22634682655334473, "step": 2090 }, { "epoch": 0.04295818225520229, "grad_norm": 7.453866481781006, "learning_rate": 1.9991234958272713e-05, "loss": 0.22540969848632814, "step": 2100 }, { "epoch": 0.04316274502784611, "grad_norm": 4.376588344573975, "learning_rate": 1.999095543126466e-05, "loss": 0.22890939712524414, "step": 2110 }, { "epoch": 0.043367307800489925, "grad_norm": 5.869319915771484, "learning_rate": 1.999067151874478e-05, "loss": 0.2333230495452881, "step": 2120 }, { "epoch": 0.04357187057313375, "grad_norm": 2.881117343902588, "learning_rate": 1.9990383220837696e-05, "loss": 0.2314704418182373, "step": 2130 }, { "epoch": 0.04377643334577757, "grad_norm": 3.2926015853881836, "learning_rate": 1.9990090537669958e-05, "loss": 0.23309178352355958, "step": 2140 }, { "epoch": 0.04398099611842139, "grad_norm": 8.442134857177734, "learning_rate": 1.9989793469370037e-05, "loss": 0.26082017421722414, "step": 2150 }, { "epoch": 0.04418555889106521, "grad_norm": 4.935953140258789, "learning_rate": 1.9989492016068327e-05, "loss": 0.254431939125061, "step": 2160 }, { "epoch": 0.04439012166370903, "grad_norm": 4.99823522567749, "learning_rate": 1.998918617789716e-05, "loss": 0.21173014640808105, "step": 2170 }, { "epoch": 0.04459468443635285, "grad_norm": 2.6556849479675293, "learning_rate": 1.9988875954990775e-05, "loss": 0.24159138202667235, "step": 2180 }, { "epoch": 0.04479924720899667, "grad_norm": 3.6498281955718994, "learning_rate": 1.9988561347485346e-05, "loss": 0.25468103885650634, "step": 2190 }, { "epoch": 0.04500380998164049, "grad_norm": 2.3046369552612305, "learning_rate": 1.998824235551897e-05, "loss": 0.25986692905426023, "step": 2200 }, { "epoch": 0.04520837275428431, "grad_norm": 5.01034688949585, "learning_rate": 1.9987918979231668e-05, "loss": 0.24984979629516602, "step": 2210 }, { "epoch": 0.045412935526928135, "grad_norm": 7.0939226150512695, "learning_rate": 1.9987591218765385e-05, "loss": 0.22473559379577637, "step": 2220 }, { "epoch": 0.04561749829957195, "grad_norm": 5.0322675704956055, "learning_rate": 1.998725907426399e-05, "loss": 0.23248586654663086, "step": 2230 }, { "epoch": 0.04582206107221577, "grad_norm": 2.8150432109832764, "learning_rate": 1.9986922545873277e-05, "loss": 0.22469544410705566, "step": 2240 }, { "epoch": 0.046026623844859595, "grad_norm": 4.815343379974365, "learning_rate": 1.9986581633740968e-05, "loss": 0.233856463432312, "step": 2250 }, { "epoch": 0.04623118661750342, "grad_norm": 4.954966068267822, "learning_rate": 1.9986236338016704e-05, "loss": 0.226265287399292, "step": 2260 }, { "epoch": 0.04643574939014723, "grad_norm": 3.7112975120544434, "learning_rate": 1.9985886658852048e-05, "loss": 0.22617614269256592, "step": 2270 }, { "epoch": 0.046640312162791055, "grad_norm": 6.671778202056885, "learning_rate": 1.9985532596400494e-05, "loss": 0.217997145652771, "step": 2280 }, { "epoch": 0.04684487493543488, "grad_norm": 3.9801599979400635, "learning_rate": 1.998517415081746e-05, "loss": 0.23063278198242188, "step": 2290 }, { "epoch": 0.04704943770807869, "grad_norm": 5.790965557098389, "learning_rate": 1.998481132226028e-05, "loss": 0.2457277774810791, "step": 2300 }, { "epoch": 0.047254000480722515, "grad_norm": 5.142910480499268, "learning_rate": 1.9984444110888222e-05, "loss": 0.23697175979614257, "step": 2310 }, { "epoch": 0.04745856325336634, "grad_norm": 2.7815282344818115, "learning_rate": 1.998407251686247e-05, "loss": 0.23086061477661132, "step": 2320 }, { "epoch": 0.04766312602601016, "grad_norm": 7.20952033996582, "learning_rate": 1.998369654034613e-05, "loss": 0.21584722995758057, "step": 2330 }, { "epoch": 0.047867688798653975, "grad_norm": 7.083583354949951, "learning_rate": 1.9983316181504245e-05, "loss": 0.226311993598938, "step": 2340 }, { "epoch": 0.0480722515712978, "grad_norm": 5.458742618560791, "learning_rate": 1.998293144050377e-05, "loss": 0.23033745288848878, "step": 2350 }, { "epoch": 0.04827681434394162, "grad_norm": 6.400622367858887, "learning_rate": 1.9982542317513585e-05, "loss": 0.23051657676696777, "step": 2360 }, { "epoch": 0.048481377116585435, "grad_norm": 4.909924507141113, "learning_rate": 1.9982148812704492e-05, "loss": 0.22982873916625976, "step": 2370 }, { "epoch": 0.04868593988922926, "grad_norm": 4.237866401672363, "learning_rate": 1.9981750926249222e-05, "loss": 0.23156361579895018, "step": 2380 }, { "epoch": 0.04889050266187308, "grad_norm": 4.822798728942871, "learning_rate": 1.998134865832243e-05, "loss": 0.2381134510040283, "step": 2390 }, { "epoch": 0.0490950654345169, "grad_norm": 4.664863109588623, "learning_rate": 1.9980942009100687e-05, "loss": 0.216498064994812, "step": 2400 }, { "epoch": 0.04929962820716072, "grad_norm": 5.378891944885254, "learning_rate": 1.9980530978762486e-05, "loss": 0.2357170343399048, "step": 2410 }, { "epoch": 0.04950419097980454, "grad_norm": 6.585239410400391, "learning_rate": 1.998011556748826e-05, "loss": 0.23174066543579103, "step": 2420 }, { "epoch": 0.04970875375244836, "grad_norm": 5.2492756843566895, "learning_rate": 1.997969577546034e-05, "loss": 0.23436312675476073, "step": 2430 }, { "epoch": 0.049913316525092184, "grad_norm": 6.168591499328613, "learning_rate": 1.9979271602863006e-05, "loss": 0.2283571720123291, "step": 2440 }, { "epoch": 0.050117879297736, "grad_norm": 11.391996383666992, "learning_rate": 1.9978843049882438e-05, "loss": 0.23030741214752198, "step": 2450 }, { "epoch": 0.05032244207037982, "grad_norm": 7.351150989532471, "learning_rate": 1.9978410116706753e-05, "loss": 0.2244765043258667, "step": 2460 }, { "epoch": 0.050527004843023644, "grad_norm": 3.622337818145752, "learning_rate": 1.9977972803525985e-05, "loss": 0.23735227584838867, "step": 2470 }, { "epoch": 0.05073156761566746, "grad_norm": 2.3177034854888916, "learning_rate": 1.997753111053209e-05, "loss": 0.22052240371704102, "step": 2480 }, { "epoch": 0.05093613038831128, "grad_norm": 2.170651912689209, "learning_rate": 1.9977085037918952e-05, "loss": 0.21583142280578613, "step": 2490 }, { "epoch": 0.051140693160955104, "grad_norm": 3.4149904251098633, "learning_rate": 1.997663458588237e-05, "loss": 0.2204876184463501, "step": 2500 }, { "epoch": 0.05134525593359893, "grad_norm": 5.3642377853393555, "learning_rate": 1.997617975462007e-05, "loss": 0.21982192993164062, "step": 2510 }, { "epoch": 0.05154981870624274, "grad_norm": 2.2598729133605957, "learning_rate": 1.9975720544331704e-05, "loss": 0.20249342918395996, "step": 2520 }, { "epoch": 0.051754381478886564, "grad_norm": 5.380239009857178, "learning_rate": 1.9975256955218836e-05, "loss": 0.22621347904205322, "step": 2530 }, { "epoch": 0.05195894425153039, "grad_norm": 4.141851902008057, "learning_rate": 1.997478898748496e-05, "loss": 0.22701678276062012, "step": 2540 }, { "epoch": 0.05216350702417421, "grad_norm": 3.027956485748291, "learning_rate": 1.997431664133549e-05, "loss": 0.23860950469970704, "step": 2550 }, { "epoch": 0.052368069796818024, "grad_norm": 5.199517250061035, "learning_rate": 1.997383991697776e-05, "loss": 0.22420401573181153, "step": 2560 }, { "epoch": 0.05257263256946185, "grad_norm": 2.3898239135742188, "learning_rate": 1.9973358814621024e-05, "loss": 0.22157559394836426, "step": 2570 }, { "epoch": 0.05277719534210567, "grad_norm": 2.4073550701141357, "learning_rate": 1.997287333447647e-05, "loss": 0.20753512382507325, "step": 2580 }, { "epoch": 0.052981758114749485, "grad_norm": 1.7918883562088013, "learning_rate": 1.9972383476757193e-05, "loss": 0.21786985397338868, "step": 2590 }, { "epoch": 0.05318632088739331, "grad_norm": 7.485311508178711, "learning_rate": 1.997188924167821e-05, "loss": 0.21655683517456054, "step": 2600 }, { "epoch": 0.05339088366003713, "grad_norm": 3.730452060699463, "learning_rate": 1.9971390629456478e-05, "loss": 0.21680479049682616, "step": 2610 }, { "epoch": 0.05359544643268095, "grad_norm": 3.3867616653442383, "learning_rate": 1.997088764031085e-05, "loss": 0.2303628921508789, "step": 2620 }, { "epoch": 0.05380000920532477, "grad_norm": 2.8324835300445557, "learning_rate": 1.9970380274462117e-05, "loss": 0.218426513671875, "step": 2630 }, { "epoch": 0.05400457197796859, "grad_norm": 5.816864490509033, "learning_rate": 1.9969868532132985e-05, "loss": 0.24464144706726074, "step": 2640 }, { "epoch": 0.05420913475061241, "grad_norm": 1.5930027961730957, "learning_rate": 1.9969352413548083e-05, "loss": 0.22381606101989746, "step": 2650 }, { "epoch": 0.05441369752325623, "grad_norm": 3.1099541187286377, "learning_rate": 1.9968831918933965e-05, "loss": 0.21513850688934327, "step": 2660 }, { "epoch": 0.05461826029590005, "grad_norm": 2.6531238555908203, "learning_rate": 1.996830704851909e-05, "loss": 0.22004990577697753, "step": 2670 }, { "epoch": 0.05482282306854387, "grad_norm": 7.358564853668213, "learning_rate": 1.9967777802533857e-05, "loss": 0.2111391544342041, "step": 2680 }, { "epoch": 0.055027385841187694, "grad_norm": 3.2013542652130127, "learning_rate": 1.996724418121058e-05, "loss": 0.22046942710876466, "step": 2690 }, { "epoch": 0.05523194861383151, "grad_norm": 5.250951766967773, "learning_rate": 1.9966706184783486e-05, "loss": 0.2325218439102173, "step": 2700 }, { "epoch": 0.05543651138647533, "grad_norm": 3.8651089668273926, "learning_rate": 1.9966163813488726e-05, "loss": 0.22282052040100098, "step": 2710 }, { "epoch": 0.055641074159119154, "grad_norm": 2.9634315967559814, "learning_rate": 1.9965617067564376e-05, "loss": 0.24078893661499023, "step": 2720 }, { "epoch": 0.055845636931762976, "grad_norm": 1.9246870279312134, "learning_rate": 1.9965065947250434e-05, "loss": 0.20954108238220215, "step": 2730 }, { "epoch": 0.05605019970440679, "grad_norm": 5.52718448638916, "learning_rate": 1.9964510452788803e-05, "loss": 0.2094501733779907, "step": 2740 }, { "epoch": 0.056254762477050614, "grad_norm": 4.314077854156494, "learning_rate": 1.9963950584423322e-05, "loss": 0.22867903709411622, "step": 2750 }, { "epoch": 0.056459325249694436, "grad_norm": 3.9680280685424805, "learning_rate": 1.996338634239974e-05, "loss": 0.22112517356872557, "step": 2760 }, { "epoch": 0.05666388802233825, "grad_norm": 1.8138667345046997, "learning_rate": 1.9962817726965744e-05, "loss": 0.2288844108581543, "step": 2770 }, { "epoch": 0.056868450794982074, "grad_norm": 6.2132158279418945, "learning_rate": 1.9962244738370906e-05, "loss": 0.20632386207580566, "step": 2780 }, { "epoch": 0.057073013567625897, "grad_norm": 10.264894485473633, "learning_rate": 1.9961667376866753e-05, "loss": 0.22453222274780274, "step": 2790 }, { "epoch": 0.05727757634026972, "grad_norm": 3.534597158432007, "learning_rate": 1.9961085642706713e-05, "loss": 0.2189408540725708, "step": 2800 }, { "epoch": 0.057482139112913534, "grad_norm": 1.7509410381317139, "learning_rate": 1.9960499536146133e-05, "loss": 0.22011346817016603, "step": 2810 }, { "epoch": 0.05768670188555736, "grad_norm": 3.878335475921631, "learning_rate": 1.995990905744229e-05, "loss": 0.2013526439666748, "step": 2820 }, { "epoch": 0.05789126465820118, "grad_norm": 4.220477104187012, "learning_rate": 1.995931420685437e-05, "loss": 0.22043094635009766, "step": 2830 }, { "epoch": 0.058095827430844994, "grad_norm": 3.1176280975341797, "learning_rate": 1.995871498464348e-05, "loss": 0.21216676235198975, "step": 2840 }, { "epoch": 0.05830039020348882, "grad_norm": 6.409346103668213, "learning_rate": 1.9958111391072653e-05, "loss": 0.21947455406188965, "step": 2850 }, { "epoch": 0.05850495297613264, "grad_norm": 2.127732515335083, "learning_rate": 1.995750342640683e-05, "loss": 0.209859299659729, "step": 2860 }, { "epoch": 0.05870951574877646, "grad_norm": 2.731273651123047, "learning_rate": 1.9956891090912877e-05, "loss": 0.21423735618591308, "step": 2870 }, { "epoch": 0.05891407852142028, "grad_norm": 1.6941038370132446, "learning_rate": 1.9956274384859577e-05, "loss": 0.2220925807952881, "step": 2880 }, { "epoch": 0.0591186412940641, "grad_norm": 5.057610034942627, "learning_rate": 1.9955653308517637e-05, "loss": 0.2214745283126831, "step": 2890 }, { "epoch": 0.05932320406670792, "grad_norm": 8.144679069519043, "learning_rate": 1.9955027862159673e-05, "loss": 0.22471818923950196, "step": 2900 }, { "epoch": 0.059527766839351744, "grad_norm": 3.9700963497161865, "learning_rate": 1.9954398046060226e-05, "loss": 0.23228297233581544, "step": 2910 }, { "epoch": 0.05973232961199556, "grad_norm": 3.674983024597168, "learning_rate": 1.9953763860495747e-05, "loss": 0.21056554317474366, "step": 2920 }, { "epoch": 0.05993689238463938, "grad_norm": 3.2998857498168945, "learning_rate": 1.9953125305744615e-05, "loss": 0.225657057762146, "step": 2930 }, { "epoch": 0.060141455157283204, "grad_norm": 5.187435626983643, "learning_rate": 1.9952482382087126e-05, "loss": 0.21955699920654298, "step": 2940 }, { "epoch": 0.06034601792992702, "grad_norm": 5.251687049865723, "learning_rate": 1.9951835089805486e-05, "loss": 0.22294752597808837, "step": 2950 }, { "epoch": 0.06055058070257084, "grad_norm": 5.826038360595703, "learning_rate": 1.9951183429183822e-05, "loss": 0.24071662425994872, "step": 2960 }, { "epoch": 0.060755143475214664, "grad_norm": 1.6238867044448853, "learning_rate": 1.9950527400508174e-05, "loss": 0.22152292728424072, "step": 2970 }, { "epoch": 0.060959706247858486, "grad_norm": 3.2169294357299805, "learning_rate": 1.994986700406652e-05, "loss": 0.219663667678833, "step": 2980 }, { "epoch": 0.0611642690205023, "grad_norm": 1.925646185874939, "learning_rate": 1.9949202240148725e-05, "loss": 0.23248934745788574, "step": 2990 }, { "epoch": 0.061368831793146124, "grad_norm": 3.0944604873657227, "learning_rate": 1.9948533109046592e-05, "loss": 0.22403507232666015, "step": 3000 }, { "epoch": 0.061573394565789946, "grad_norm": 3.4826793670654297, "learning_rate": 1.9947859611053838e-05, "loss": 0.22067732810974122, "step": 3010 }, { "epoch": 0.06177795733843377, "grad_norm": 3.178823471069336, "learning_rate": 1.9947181746466087e-05, "loss": 0.2167065143585205, "step": 3020 }, { "epoch": 0.061982520111077584, "grad_norm": 6.40047025680542, "learning_rate": 1.9946499515580888e-05, "loss": 0.2081939697265625, "step": 3030 }, { "epoch": 0.062187082883721406, "grad_norm": 6.335292339324951, "learning_rate": 1.9945812918697712e-05, "loss": 0.2470637798309326, "step": 3040 }, { "epoch": 0.06239164565636523, "grad_norm": 3.2206838130950928, "learning_rate": 1.994512195611793e-05, "loss": 0.2184401273727417, "step": 3050 }, { "epoch": 0.06259620842900905, "grad_norm": 2.272294521331787, "learning_rate": 1.9944426628144844e-05, "loss": 0.22498927116394044, "step": 3060 }, { "epoch": 0.06280077120165287, "grad_norm": 4.60621452331543, "learning_rate": 1.9943726935083667e-05, "loss": 0.21953530311584474, "step": 3070 }, { "epoch": 0.06300533397429668, "grad_norm": 3.0964508056640625, "learning_rate": 1.9943022877241526e-05, "loss": 0.21847856044769287, "step": 3080 }, { "epoch": 0.06320989674694051, "grad_norm": 2.1208527088165283, "learning_rate": 1.9942314454927466e-05, "loss": 0.22152557373046874, "step": 3090 }, { "epoch": 0.06341445951958433, "grad_norm": 6.5011186599731445, "learning_rate": 1.9941601668452447e-05, "loss": 0.21972060203552246, "step": 3100 }, { "epoch": 0.06361902229222816, "grad_norm": 3.367846727371216, "learning_rate": 1.994088451812935e-05, "loss": 0.19217391014099122, "step": 3110 }, { "epoch": 0.06382358506487197, "grad_norm": 4.2246198654174805, "learning_rate": 1.9940163004272964e-05, "loss": 0.2132814884185791, "step": 3120 }, { "epoch": 0.06402814783751579, "grad_norm": 2.764387845993042, "learning_rate": 1.9939437127199994e-05, "loss": 0.22181398868560792, "step": 3130 }, { "epoch": 0.06423271061015962, "grad_norm": 3.0552256107330322, "learning_rate": 1.9938706887229064e-05, "loss": 0.21867012977600098, "step": 3140 }, { "epoch": 0.06443727338280343, "grad_norm": 2.635167360305786, "learning_rate": 1.9937972284680714e-05, "loss": 0.2183786392211914, "step": 3150 }, { "epoch": 0.06464183615544725, "grad_norm": 1.6867821216583252, "learning_rate": 1.993723331987739e-05, "loss": 0.204626202583313, "step": 3160 }, { "epoch": 0.06484639892809108, "grad_norm": 3.6625547409057617, "learning_rate": 1.993648999314347e-05, "loss": 0.219002628326416, "step": 3170 }, { "epoch": 0.06505096170073489, "grad_norm": 5.903679370880127, "learning_rate": 1.9935742304805224e-05, "loss": 0.2241436004638672, "step": 3180 }, { "epoch": 0.0652555244733787, "grad_norm": 3.84879994392395, "learning_rate": 1.9934990255190855e-05, "loss": 0.21058015823364257, "step": 3190 }, { "epoch": 0.06546008724602254, "grad_norm": 1.6802250146865845, "learning_rate": 1.9934233844630474e-05, "loss": 0.21831612586975097, "step": 3200 }, { "epoch": 0.06566465001866635, "grad_norm": 4.9870991706848145, "learning_rate": 1.9933473073456104e-05, "loss": 0.2205021858215332, "step": 3210 }, { "epoch": 0.06586921279131017, "grad_norm": 1.5696327686309814, "learning_rate": 1.9932707942001684e-05, "loss": 0.20469362735748292, "step": 3220 }, { "epoch": 0.066073775563954, "grad_norm": 4.932891845703125, "learning_rate": 1.993193845060307e-05, "loss": 0.2142056941986084, "step": 3230 }, { "epoch": 0.06627833833659781, "grad_norm": 4.247636318206787, "learning_rate": 1.9931164599598026e-05, "loss": 0.21333723068237304, "step": 3240 }, { "epoch": 0.06648290110924164, "grad_norm": 2.7460989952087402, "learning_rate": 1.9930386389326232e-05, "loss": 0.20758528709411622, "step": 3250 }, { "epoch": 0.06668746388188546, "grad_norm": 3.3783581256866455, "learning_rate": 1.9929603820129285e-05, "loss": 0.21734399795532228, "step": 3260 }, { "epoch": 0.06689202665452927, "grad_norm": 5.406643867492676, "learning_rate": 1.992881689235069e-05, "loss": 0.21778416633605957, "step": 3270 }, { "epoch": 0.0670965894271731, "grad_norm": 2.5971150398254395, "learning_rate": 1.9928025606335865e-05, "loss": 0.21945557594299317, "step": 3280 }, { "epoch": 0.06730115219981692, "grad_norm": 1.9850115776062012, "learning_rate": 1.9927229962432154e-05, "loss": 0.2076267719268799, "step": 3290 }, { "epoch": 0.06750571497246073, "grad_norm": 5.1384358406066895, "learning_rate": 1.992642996098879e-05, "loss": 0.22581658363342286, "step": 3300 }, { "epoch": 0.06771027774510456, "grad_norm": 2.72475266456604, "learning_rate": 1.992562560235694e-05, "loss": 0.20792698860168457, "step": 3310 }, { "epoch": 0.06791484051774838, "grad_norm": 3.258375406265259, "learning_rate": 1.9924816886889676e-05, "loss": 0.2247303009033203, "step": 3320 }, { "epoch": 0.06811940329039219, "grad_norm": 2.0706238746643066, "learning_rate": 1.992400381494198e-05, "loss": 0.2058290958404541, "step": 3330 }, { "epoch": 0.06832396606303602, "grad_norm": 5.244489669799805, "learning_rate": 1.9923186386870746e-05, "loss": 0.22904341220855712, "step": 3340 }, { "epoch": 0.06852852883567984, "grad_norm": 4.545787811279297, "learning_rate": 1.9922364603034786e-05, "loss": 0.21855378150939941, "step": 3350 }, { "epoch": 0.06873309160832367, "grad_norm": 4.749356269836426, "learning_rate": 1.992153846379482e-05, "loss": 0.21837882995605468, "step": 3360 }, { "epoch": 0.06893765438096748, "grad_norm": 2.0649945735931396, "learning_rate": 1.992070796951348e-05, "loss": 0.20008227825164795, "step": 3370 }, { "epoch": 0.0691422171536113, "grad_norm": 1.9501770734786987, "learning_rate": 1.9919873120555314e-05, "loss": 0.20721228122711183, "step": 3380 }, { "epoch": 0.06934677992625513, "grad_norm": 4.354963779449463, "learning_rate": 1.991903391728677e-05, "loss": 0.21923341751098632, "step": 3390 }, { "epoch": 0.06955134269889894, "grad_norm": 1.6531859636306763, "learning_rate": 1.991819036007622e-05, "loss": 0.2055530548095703, "step": 3400 }, { "epoch": 0.06975590547154276, "grad_norm": 2.6039650440216064, "learning_rate": 1.991734244929394e-05, "loss": 0.20354537963867186, "step": 3410 }, { "epoch": 0.06996046824418659, "grad_norm": 1.7356629371643066, "learning_rate": 1.9916490185312117e-05, "loss": 0.2060065507888794, "step": 3420 }, { "epoch": 0.0701650310168304, "grad_norm": 3.8267149925231934, "learning_rate": 1.9915633568504855e-05, "loss": 0.2124016523361206, "step": 3430 }, { "epoch": 0.07036959378947422, "grad_norm": 2.545083999633789, "learning_rate": 1.991477259924816e-05, "loss": 0.2192770481109619, "step": 3440 }, { "epoch": 0.07057415656211805, "grad_norm": 2.809861898422241, "learning_rate": 1.991390727791996e-05, "loss": 0.22858133316040039, "step": 3450 }, { "epoch": 0.07077871933476186, "grad_norm": 1.9883414506912231, "learning_rate": 1.9913037604900077e-05, "loss": 0.21977543830871582, "step": 3460 }, { "epoch": 0.07098328210740569, "grad_norm": 2.0472004413604736, "learning_rate": 1.9912163580570257e-05, "loss": 0.2209764003753662, "step": 3470 }, { "epoch": 0.0711878448800495, "grad_norm": 1.6384565830230713, "learning_rate": 1.991128520531415e-05, "loss": 0.21211299896240235, "step": 3480 }, { "epoch": 0.07139240765269332, "grad_norm": 4.367480754852295, "learning_rate": 1.9910402479517326e-05, "loss": 0.2164222478866577, "step": 3490 }, { "epoch": 0.07159697042533715, "grad_norm": 2.6124322414398193, "learning_rate": 1.9909515403567243e-05, "loss": 0.1984264373779297, "step": 3500 }, { "epoch": 0.07180153319798097, "grad_norm": 2.3840274810791016, "learning_rate": 1.9908623977853287e-05, "loss": 0.21925783157348633, "step": 3510 }, { "epoch": 0.07200609597062478, "grad_norm": 5.228074550628662, "learning_rate": 1.990772820276675e-05, "loss": 0.220589017868042, "step": 3520 }, { "epoch": 0.07221065874326861, "grad_norm": 4.289828777313232, "learning_rate": 1.9906828078700823e-05, "loss": 0.20656197071075438, "step": 3530 }, { "epoch": 0.07241522151591243, "grad_norm": 3.0354902744293213, "learning_rate": 1.9905923606050622e-05, "loss": 0.20514974594116211, "step": 3540 }, { "epoch": 0.07261978428855624, "grad_norm": 4.882221221923828, "learning_rate": 1.990501478521316e-05, "loss": 0.21801490783691407, "step": 3550 }, { "epoch": 0.07282434706120007, "grad_norm": 2.320871353149414, "learning_rate": 1.9904101616587365e-05, "loss": 0.2153409719467163, "step": 3560 }, { "epoch": 0.07302890983384389, "grad_norm": 2.8027141094207764, "learning_rate": 1.9903184100574072e-05, "loss": 0.22388508319854736, "step": 3570 }, { "epoch": 0.07323347260648772, "grad_norm": 2.6599373817443848, "learning_rate": 1.9902262237576015e-05, "loss": 0.21730031967163085, "step": 3580 }, { "epoch": 0.07343803537913153, "grad_norm": 9.284223556518555, "learning_rate": 1.9901336027997848e-05, "loss": 0.22580137252807617, "step": 3590 }, { "epoch": 0.07364259815177535, "grad_norm": 1.9037426710128784, "learning_rate": 1.990040547224613e-05, "loss": 0.23079209327697753, "step": 3600 }, { "epoch": 0.07384716092441918, "grad_norm": 3.7070741653442383, "learning_rate": 1.9899470570729332e-05, "loss": 0.21867711544036866, "step": 3610 }, { "epoch": 0.07405172369706299, "grad_norm": 8.329113006591797, "learning_rate": 1.9898531323857817e-05, "loss": 0.2298736333847046, "step": 3620 }, { "epoch": 0.0742562864697068, "grad_norm": 2.479327440261841, "learning_rate": 1.989758773204387e-05, "loss": 0.23611030578613282, "step": 3630 }, { "epoch": 0.07446084924235064, "grad_norm": 1.7591407299041748, "learning_rate": 1.9896639795701682e-05, "loss": 0.2105861186981201, "step": 3640 }, { "epoch": 0.07466541201499445, "grad_norm": 3.2169151306152344, "learning_rate": 1.9895687515247348e-05, "loss": 0.21970832347869873, "step": 3650 }, { "epoch": 0.07486997478763827, "grad_norm": 3.842913866043091, "learning_rate": 1.989473089109886e-05, "loss": 0.2060927152633667, "step": 3660 }, { "epoch": 0.0750745375602821, "grad_norm": 3.420506477355957, "learning_rate": 1.9893769923676142e-05, "loss": 0.20785984992980958, "step": 3670 }, { "epoch": 0.07527910033292591, "grad_norm": 4.122020244598389, "learning_rate": 1.9892804613400994e-05, "loss": 0.20759119987487792, "step": 3680 }, { "epoch": 0.07548366310556973, "grad_norm": 2.437120199203491, "learning_rate": 1.989183496069715e-05, "loss": 0.2116391897201538, "step": 3690 }, { "epoch": 0.07568822587821356, "grad_norm": 2.458986282348633, "learning_rate": 1.9890860965990227e-05, "loss": 0.21951518058776856, "step": 3700 }, { "epoch": 0.07589278865085737, "grad_norm": 3.2828969955444336, "learning_rate": 1.988988262970776e-05, "loss": 0.2200260877609253, "step": 3710 }, { "epoch": 0.0760973514235012, "grad_norm": 4.531951427459717, "learning_rate": 1.9888899952279196e-05, "loss": 0.2174576997756958, "step": 3720 }, { "epoch": 0.07630191419614502, "grad_norm": 1.729113221168518, "learning_rate": 1.9887912934135872e-05, "loss": 0.2112560749053955, "step": 3730 }, { "epoch": 0.07650647696878883, "grad_norm": 3.9216346740722656, "learning_rate": 1.9886921575711037e-05, "loss": 0.21511080265045165, "step": 3740 }, { "epoch": 0.07671103974143266, "grad_norm": 8.38648796081543, "learning_rate": 1.9885925877439852e-05, "loss": 0.21266918182373046, "step": 3750 }, { "epoch": 0.07691560251407648, "grad_norm": 3.828251600265503, "learning_rate": 1.9884925839759375e-05, "loss": 0.22929680347442627, "step": 3760 }, { "epoch": 0.07712016528672029, "grad_norm": 3.172247886657715, "learning_rate": 1.9883921463108568e-05, "loss": 0.21023499965667725, "step": 3770 }, { "epoch": 0.07732472805936412, "grad_norm": 1.929677963256836, "learning_rate": 1.98829127479283e-05, "loss": 0.22052254676818847, "step": 3780 }, { "epoch": 0.07752929083200794, "grad_norm": 2.8186819553375244, "learning_rate": 1.9881899694661343e-05, "loss": 0.21946685314178466, "step": 3790 }, { "epoch": 0.07773385360465175, "grad_norm": 5.191129207611084, "learning_rate": 1.9880882303752383e-05, "loss": 0.206931734085083, "step": 3800 }, { "epoch": 0.07793841637729558, "grad_norm": 6.963578701019287, "learning_rate": 1.9879860575648e-05, "loss": 0.2242159366607666, "step": 3810 }, { "epoch": 0.0781429791499394, "grad_norm": 2.846674680709839, "learning_rate": 1.9878834510796673e-05, "loss": 0.22210607528686524, "step": 3820 }, { "epoch": 0.07834754192258322, "grad_norm": 3.391895294189453, "learning_rate": 1.9877804109648793e-05, "loss": 0.21150834560394288, "step": 3830 }, { "epoch": 0.07855210469522704, "grad_norm": 6.789275169372559, "learning_rate": 1.9876769372656656e-05, "loss": 0.2154078006744385, "step": 3840 }, { "epoch": 0.07875666746787086, "grad_norm": 2.8720250129699707, "learning_rate": 1.987573030027446e-05, "loss": 0.21336650848388672, "step": 3850 }, { "epoch": 0.07896123024051468, "grad_norm": 3.087949275970459, "learning_rate": 1.98746868929583e-05, "loss": 0.22108817100524902, "step": 3860 }, { "epoch": 0.0791657930131585, "grad_norm": 6.568836688995361, "learning_rate": 1.9873639151166174e-05, "loss": 0.22451262474060057, "step": 3870 }, { "epoch": 0.07937035578580232, "grad_norm": 2.5872724056243896, "learning_rate": 1.9872587075357995e-05, "loss": 0.22275402545928955, "step": 3880 }, { "epoch": 0.07957491855844614, "grad_norm": 4.651001453399658, "learning_rate": 1.987153066599557e-05, "loss": 0.2176722764968872, "step": 3890 }, { "epoch": 0.07977948133108996, "grad_norm": 2.8224332332611084, "learning_rate": 1.9870469923542598e-05, "loss": 0.20844850540161133, "step": 3900 }, { "epoch": 0.07998404410373378, "grad_norm": 2.1334571838378906, "learning_rate": 1.9869404848464695e-05, "loss": 0.20222914218902588, "step": 3910 }, { "epoch": 0.0801886068763776, "grad_norm": 9.605443000793457, "learning_rate": 1.9868335441229376e-05, "loss": 0.2083744764328003, "step": 3920 }, { "epoch": 0.08039316964902142, "grad_norm": 1.5568506717681885, "learning_rate": 1.9867261702306054e-05, "loss": 0.19006719589233398, "step": 3930 }, { "epoch": 0.08059773242166525, "grad_norm": 3.4468159675598145, "learning_rate": 1.9866183632166045e-05, "loss": 0.22170896530151368, "step": 3940 }, { "epoch": 0.08080229519430907, "grad_norm": 2.736926317214966, "learning_rate": 1.9865101231282563e-05, "loss": 0.21456520557403563, "step": 3950 }, { "epoch": 0.08100685796695288, "grad_norm": 2.9144704341888428, "learning_rate": 1.986401450013073e-05, "loss": 0.22095632553100586, "step": 3960 }, { "epoch": 0.08121142073959671, "grad_norm": 4.44222354888916, "learning_rate": 1.9862923439187564e-05, "loss": 0.21698975563049316, "step": 3970 }, { "epoch": 0.08141598351224053, "grad_norm": 6.009205341339111, "learning_rate": 1.9861828048931982e-05, "loss": 0.23071713447570802, "step": 3980 }, { "epoch": 0.08162054628488434, "grad_norm": 1.7207918167114258, "learning_rate": 1.986072832984481e-05, "loss": 0.2193828344345093, "step": 3990 }, { "epoch": 0.08182510905752817, "grad_norm": 1.9913793802261353, "learning_rate": 1.9859624282408755e-05, "loss": 0.20769238471984863, "step": 4000 }, { "epoch": 0.08202967183017199, "grad_norm": 6.752035140991211, "learning_rate": 1.9858515907108448e-05, "loss": 0.21514654159545898, "step": 4010 }, { "epoch": 0.0822342346028158, "grad_norm": 2.745028257369995, "learning_rate": 1.98574032044304e-05, "loss": 0.22077980041503906, "step": 4020 }, { "epoch": 0.08243879737545963, "grad_norm": 1.9957289695739746, "learning_rate": 1.985628617486304e-05, "loss": 0.21864030361175538, "step": 4030 }, { "epoch": 0.08264336014810345, "grad_norm": 1.6869831085205078, "learning_rate": 1.985516481889668e-05, "loss": 0.21733396053314208, "step": 4040 }, { "epoch": 0.08284792292074727, "grad_norm": 2.732828378677368, "learning_rate": 1.9854039137023534e-05, "loss": 0.21394579410552977, "step": 4050 }, { "epoch": 0.08305248569339109, "grad_norm": 5.195559024810791, "learning_rate": 1.9852909129737725e-05, "loss": 0.22119054794311524, "step": 4060 }, { "epoch": 0.0832570484660349, "grad_norm": 4.204400062561035, "learning_rate": 1.9851774797535263e-05, "loss": 0.21811771392822266, "step": 4070 }, { "epoch": 0.08346161123867873, "grad_norm": 3.093108892440796, "learning_rate": 1.9850636140914062e-05, "loss": 0.2039113998413086, "step": 4080 }, { "epoch": 0.08366617401132255, "grad_norm": 4.973918437957764, "learning_rate": 1.9849493160373936e-05, "loss": 0.21868622303009033, "step": 4090 }, { "epoch": 0.08387073678396637, "grad_norm": 3.7061727046966553, "learning_rate": 1.9848345856416592e-05, "loss": 0.20774917602539061, "step": 4100 }, { "epoch": 0.0840752995566102, "grad_norm": 2.793447971343994, "learning_rate": 1.9847194229545633e-05, "loss": 0.22514011859893798, "step": 4110 }, { "epoch": 0.08427986232925401, "grad_norm": 4.061790466308594, "learning_rate": 1.984603828026657e-05, "loss": 0.2085574150085449, "step": 4120 }, { "epoch": 0.08448442510189783, "grad_norm": 1.1082043647766113, "learning_rate": 1.98448780090868e-05, "loss": 0.21018643379211427, "step": 4130 }, { "epoch": 0.08468898787454165, "grad_norm": 1.2160954475402832, "learning_rate": 1.9843713416515628e-05, "loss": 0.20883545875549317, "step": 4140 }, { "epoch": 0.08489355064718547, "grad_norm": 1.2808533906936646, "learning_rate": 1.9842544503064244e-05, "loss": 0.2119051456451416, "step": 4150 }, { "epoch": 0.08509811341982929, "grad_norm": 3.038712978363037, "learning_rate": 1.9841371269245743e-05, "loss": 0.2022111177444458, "step": 4160 }, { "epoch": 0.08530267619247311, "grad_norm": 4.799592971801758, "learning_rate": 1.9840193715575112e-05, "loss": 0.20100085735321044, "step": 4170 }, { "epoch": 0.08550723896511693, "grad_norm": 2.322873592376709, "learning_rate": 1.983901184256924e-05, "loss": 0.217940092086792, "step": 4180 }, { "epoch": 0.08571180173776076, "grad_norm": 1.3996471166610718, "learning_rate": 1.9837825650746906e-05, "loss": 0.22198922634124757, "step": 4190 }, { "epoch": 0.08591636451040457, "grad_norm": 3.070347309112549, "learning_rate": 1.9836635140628782e-05, "loss": 0.21477127075195312, "step": 4200 }, { "epoch": 0.08612092728304839, "grad_norm": 1.7430442571640015, "learning_rate": 1.9835440312737445e-05, "loss": 0.20833101272583007, "step": 4210 }, { "epoch": 0.08632549005569222, "grad_norm": 2.230907917022705, "learning_rate": 1.9834241167597367e-05, "loss": 0.21876320838928223, "step": 4220 }, { "epoch": 0.08653005282833603, "grad_norm": 2.301772117614746, "learning_rate": 1.9833037705734898e-05, "loss": 0.20289735794067382, "step": 4230 }, { "epoch": 0.08673461560097985, "grad_norm": 1.9660015106201172, "learning_rate": 1.983182992767831e-05, "loss": 0.19728105068206786, "step": 4240 }, { "epoch": 0.08693917837362368, "grad_norm": 2.930114507675171, "learning_rate": 1.9830617833957743e-05, "loss": 0.20877699851989745, "step": 4250 }, { "epoch": 0.0871437411462675, "grad_norm": 3.1202123165130615, "learning_rate": 1.982940142510525e-05, "loss": 0.2068026065826416, "step": 4260 }, { "epoch": 0.08734830391891131, "grad_norm": 2.0632288455963135, "learning_rate": 1.982818070165477e-05, "loss": 0.20422470569610596, "step": 4270 }, { "epoch": 0.08755286669155514, "grad_norm": 2.8836512565612793, "learning_rate": 1.982695566414214e-05, "loss": 0.20105409622192383, "step": 4280 }, { "epoch": 0.08775742946419895, "grad_norm": 3.017942428588867, "learning_rate": 1.9825726313105082e-05, "loss": 0.22649993896484374, "step": 4290 }, { "epoch": 0.08796199223684278, "grad_norm": 6.286077499389648, "learning_rate": 1.982449264908322e-05, "loss": 0.21065983772277833, "step": 4300 }, { "epoch": 0.0881665550094866, "grad_norm": 1.8946268558502197, "learning_rate": 1.982325467261807e-05, "loss": 0.22223892211914062, "step": 4310 }, { "epoch": 0.08837111778213042, "grad_norm": 2.3912806510925293, "learning_rate": 1.982201238425304e-05, "loss": 0.211181640625, "step": 4320 }, { "epoch": 0.08857568055477424, "grad_norm": 3.308166027069092, "learning_rate": 1.9820765784533428e-05, "loss": 0.19122258424758912, "step": 4330 }, { "epoch": 0.08878024332741806, "grad_norm": 4.467013835906982, "learning_rate": 1.9819514874006427e-05, "loss": 0.20047144889831542, "step": 4340 }, { "epoch": 0.08898480610006188, "grad_norm": 4.231367111206055, "learning_rate": 1.9818259653221124e-05, "loss": 0.21748776435852052, "step": 4350 }, { "epoch": 0.0891893688727057, "grad_norm": 1.8153002262115479, "learning_rate": 1.981700012272849e-05, "loss": 0.20622341632843016, "step": 4360 }, { "epoch": 0.08939393164534952, "grad_norm": 2.5196681022644043, "learning_rate": 1.9815736283081405e-05, "loss": 0.20402040481567382, "step": 4370 }, { "epoch": 0.08959849441799334, "grad_norm": 2.1073715686798096, "learning_rate": 1.981446813483462e-05, "loss": 0.224548864364624, "step": 4380 }, { "epoch": 0.08980305719063716, "grad_norm": 1.4441951513290405, "learning_rate": 1.9813195678544784e-05, "loss": 0.22961878776550293, "step": 4390 }, { "epoch": 0.09000761996328098, "grad_norm": 1.8481369018554688, "learning_rate": 1.9811918914770448e-05, "loss": 0.2158677577972412, "step": 4400 }, { "epoch": 0.09021218273592481, "grad_norm": 2.871293544769287, "learning_rate": 1.981063784407204e-05, "loss": 0.20779380798339844, "step": 4410 }, { "epoch": 0.09041674550856862, "grad_norm": 4.360907077789307, "learning_rate": 1.9809352467011886e-05, "loss": 0.21477084159851073, "step": 4420 }, { "epoch": 0.09062130828121244, "grad_norm": 3.1711513996124268, "learning_rate": 1.9808062784154195e-05, "loss": 0.2129807710647583, "step": 4430 }, { "epoch": 0.09082587105385627, "grad_norm": 4.900604724884033, "learning_rate": 1.9806768796065078e-05, "loss": 0.21121664047241212, "step": 4440 }, { "epoch": 0.09103043382650008, "grad_norm": 2.2539541721343994, "learning_rate": 1.9805470503312527e-05, "loss": 0.20538501739501952, "step": 4450 }, { "epoch": 0.0912349965991439, "grad_norm": 2.696607828140259, "learning_rate": 1.980416790646642e-05, "loss": 0.20343823432922364, "step": 4460 }, { "epoch": 0.09143955937178773, "grad_norm": 2.1665709018707275, "learning_rate": 1.9802861006098535e-05, "loss": 0.1952465295791626, "step": 4470 }, { "epoch": 0.09164412214443154, "grad_norm": 9.676446914672852, "learning_rate": 1.980154980278253e-05, "loss": 0.2154768705368042, "step": 4480 }, { "epoch": 0.09184868491707536, "grad_norm": 2.6115357875823975, "learning_rate": 1.9800234297093957e-05, "loss": 0.22924551963806153, "step": 4490 }, { "epoch": 0.09205324768971919, "grad_norm": 2.204408884048462, "learning_rate": 1.979891448961026e-05, "loss": 0.20650367736816405, "step": 4500 }, { "epoch": 0.092257810462363, "grad_norm": 4.136912822723389, "learning_rate": 1.9797590380910758e-05, "loss": 0.2205209732055664, "step": 4510 }, { "epoch": 0.09246237323500683, "grad_norm": 5.069903373718262, "learning_rate": 1.9796261971576666e-05, "loss": 0.2200350284576416, "step": 4520 }, { "epoch": 0.09266693600765065, "grad_norm": 3.2915306091308594, "learning_rate": 1.9794929262191095e-05, "loss": 0.21113643646240235, "step": 4530 }, { "epoch": 0.09287149878029446, "grad_norm": 5.914007186889648, "learning_rate": 1.9793592253339028e-05, "loss": 0.21345167160034179, "step": 4540 }, { "epoch": 0.0930760615529383, "grad_norm": 3.8568949699401855, "learning_rate": 1.9792250945607345e-05, "loss": 0.21831860542297363, "step": 4550 }, { "epoch": 0.09328062432558211, "grad_norm": 2.360004186630249, "learning_rate": 1.979090533958481e-05, "loss": 0.20819497108459473, "step": 4560 }, { "epoch": 0.09348518709822592, "grad_norm": 2.6647911071777344, "learning_rate": 1.9789555435862075e-05, "loss": 0.2030043601989746, "step": 4570 }, { "epoch": 0.09368974987086975, "grad_norm": 3.087996006011963, "learning_rate": 1.9788201235031677e-05, "loss": 0.2009918212890625, "step": 4580 }, { "epoch": 0.09389431264351357, "grad_norm": 1.7625164985656738, "learning_rate": 1.9786842737688044e-05, "loss": 0.21056685447692872, "step": 4590 }, { "epoch": 0.09409887541615738, "grad_norm": 3.5456607341766357, "learning_rate": 1.978547994442748e-05, "loss": 0.20545511245727538, "step": 4600 }, { "epoch": 0.09430343818880121, "grad_norm": 3.9885683059692383, "learning_rate": 1.9784112855848184e-05, "loss": 0.1974404573440552, "step": 4610 }, { "epoch": 0.09450800096144503, "grad_norm": 1.7743279933929443, "learning_rate": 1.9782741472550237e-05, "loss": 0.21213080883026122, "step": 4620 }, { "epoch": 0.09471256373408886, "grad_norm": 3.0555238723754883, "learning_rate": 1.9781365795135602e-05, "loss": 0.21288645267486572, "step": 4630 }, { "epoch": 0.09491712650673267, "grad_norm": 2.089367628097534, "learning_rate": 1.9779985824208136e-05, "loss": 0.21135931015014647, "step": 4640 }, { "epoch": 0.09512168927937649, "grad_norm": 2.1434357166290283, "learning_rate": 1.9778601560373572e-05, "loss": 0.20566079616546631, "step": 4650 }, { "epoch": 0.09532625205202032, "grad_norm": 1.965035319328308, "learning_rate": 1.9777213004239525e-05, "loss": 0.2001131534576416, "step": 4660 }, { "epoch": 0.09553081482466413, "grad_norm": 4.746573448181152, "learning_rate": 1.977582015641551e-05, "loss": 0.22283148765563965, "step": 4670 }, { "epoch": 0.09573537759730795, "grad_norm": 2.0454230308532715, "learning_rate": 1.9774423017512902e-05, "loss": 0.2184225559234619, "step": 4680 }, { "epoch": 0.09593994036995178, "grad_norm": 2.8777382373809814, "learning_rate": 1.9773021588144983e-05, "loss": 0.20322706699371337, "step": 4690 }, { "epoch": 0.0961445031425956, "grad_norm": 2.4922518730163574, "learning_rate": 1.9771615868926906e-05, "loss": 0.21377451419830323, "step": 4700 }, { "epoch": 0.09634906591523941, "grad_norm": 3.474825859069824, "learning_rate": 1.9770205860475705e-05, "loss": 0.21592049598693847, "step": 4710 }, { "epoch": 0.09655362868788324, "grad_norm": 7.151583671569824, "learning_rate": 1.9768791563410302e-05, "loss": 0.21598949432373046, "step": 4720 }, { "epoch": 0.09675819146052705, "grad_norm": 8.013775825500488, "learning_rate": 1.9767372978351504e-05, "loss": 0.20511474609375, "step": 4730 }, { "epoch": 0.09696275423317087, "grad_norm": 1.841440200805664, "learning_rate": 1.9765950105921992e-05, "loss": 0.22217803001403807, "step": 4740 }, { "epoch": 0.0971673170058147, "grad_norm": 2.7362771034240723, "learning_rate": 1.9764522946746334e-05, "loss": 0.22945046424865723, "step": 4750 }, { "epoch": 0.09737187977845851, "grad_norm": 2.7783045768737793, "learning_rate": 1.9763091501450978e-05, "loss": 0.20293116569519043, "step": 4760 }, { "epoch": 0.09757644255110234, "grad_norm": 2.0085458755493164, "learning_rate": 1.9761655770664258e-05, "loss": 0.19502609968185425, "step": 4770 }, { "epoch": 0.09778100532374616, "grad_norm": 4.707954406738281, "learning_rate": 1.9760215755016386e-05, "loss": 0.19629571437835694, "step": 4780 }, { "epoch": 0.09798556809638997, "grad_norm": 2.612626314163208, "learning_rate": 1.9758771455139448e-05, "loss": 0.20421924591064453, "step": 4790 }, { "epoch": 0.0981901308690338, "grad_norm": 2.416125535964966, "learning_rate": 1.975732287166742e-05, "loss": 0.22086892127990723, "step": 4800 }, { "epoch": 0.09839469364167762, "grad_norm": 2.984590768814087, "learning_rate": 1.975587000523616e-05, "loss": 0.20924825668334962, "step": 4810 }, { "epoch": 0.09859925641432143, "grad_norm": 4.043100357055664, "learning_rate": 1.975441285648339e-05, "loss": 0.23313865661621094, "step": 4820 }, { "epoch": 0.09880381918696526, "grad_norm": 1.466413140296936, "learning_rate": 1.9752951426048734e-05, "loss": 0.2075420379638672, "step": 4830 }, { "epoch": 0.09900838195960908, "grad_norm": 3.6269047260284424, "learning_rate": 1.975148571457368e-05, "loss": 0.2051316261291504, "step": 4840 }, { "epoch": 0.0992129447322529, "grad_norm": 2.0994908809661865, "learning_rate": 1.9750015722701596e-05, "loss": 0.18989657163619994, "step": 4850 }, { "epoch": 0.09941750750489672, "grad_norm": 2.8478260040283203, "learning_rate": 1.9748541451077738e-05, "loss": 0.20132358074188234, "step": 4860 }, { "epoch": 0.09962207027754054, "grad_norm": 3.920109510421753, "learning_rate": 1.974706290034923e-05, "loss": 0.19346601963043214, "step": 4870 }, { "epoch": 0.09982663305018437, "grad_norm": 2.346301555633545, "learning_rate": 1.9745580071165084e-05, "loss": 0.21529505252838135, "step": 4880 }, { "epoch": 0.10003119582282818, "grad_norm": 2.390984296798706, "learning_rate": 1.9744092964176182e-05, "loss": 0.19779043197631835, "step": 4890 }, { "epoch": 0.100235758595472, "grad_norm": 3.4184601306915283, "learning_rate": 1.9742601580035286e-05, "loss": 0.20527997016906738, "step": 4900 }, { "epoch": 0.10044032136811583, "grad_norm": 2.1983349323272705, "learning_rate": 1.9741105919397035e-05, "loss": 0.19292960166931153, "step": 4910 }, { "epoch": 0.10064488414075964, "grad_norm": 2.6796209812164307, "learning_rate": 1.9739605982917957e-05, "loss": 0.21770477294921875, "step": 4920 }, { "epoch": 0.10084944691340346, "grad_norm": 4.158708572387695, "learning_rate": 1.9738101771256434e-05, "loss": 0.21301641464233398, "step": 4930 }, { "epoch": 0.10105400968604729, "grad_norm": 2.09440541267395, "learning_rate": 1.9736593285072744e-05, "loss": 0.22133054733276367, "step": 4940 }, { "epoch": 0.1012585724586911, "grad_norm": 4.496572017669678, "learning_rate": 1.9735080525029028e-05, "loss": 0.21240551471710206, "step": 4950 }, { "epoch": 0.10146313523133492, "grad_norm": 3.4465720653533936, "learning_rate": 1.9733563491789317e-05, "loss": 0.22789816856384276, "step": 4960 }, { "epoch": 0.10166769800397875, "grad_norm": 3.032696008682251, "learning_rate": 1.9732042186019506e-05, "loss": 0.2094024419784546, "step": 4970 }, { "epoch": 0.10187226077662256, "grad_norm": 2.887159585952759, "learning_rate": 1.973051660838737e-05, "loss": 0.20635910034179689, "step": 4980 }, { "epoch": 0.1020768235492664, "grad_norm": 6.8926591873168945, "learning_rate": 1.9728986759562557e-05, "loss": 0.20425598621368407, "step": 4990 }, { "epoch": 0.10228138632191021, "grad_norm": 2.2751758098602295, "learning_rate": 1.9727452640216595e-05, "loss": 0.2050168991088867, "step": 5000 }, { "epoch": 0.10248594909455402, "grad_norm": 2.1304290294647217, "learning_rate": 1.9725914251022878e-05, "loss": 0.1940804362297058, "step": 5010 }, { "epoch": 0.10269051186719785, "grad_norm": 4.872063636779785, "learning_rate": 1.9724371592656682e-05, "loss": 0.21028733253479004, "step": 5020 }, { "epoch": 0.10289507463984167, "grad_norm": 2.9817464351654053, "learning_rate": 1.9722824665795153e-05, "loss": 0.20840034484863282, "step": 5030 }, { "epoch": 0.10309963741248548, "grad_norm": 3.2252678871154785, "learning_rate": 1.9721273471117316e-05, "loss": 0.20271039009094238, "step": 5040 }, { "epoch": 0.10330420018512931, "grad_norm": 2.499755859375, "learning_rate": 1.971971800930406e-05, "loss": 0.21355714797973632, "step": 5050 }, { "epoch": 0.10350876295777313, "grad_norm": 1.800626516342163, "learning_rate": 1.9718158281038155e-05, "loss": 0.20759527683258056, "step": 5060 }, { "epoch": 0.10371332573041694, "grad_norm": 2.0142195224761963, "learning_rate": 1.971659428700424e-05, "loss": 0.20920779705047607, "step": 5070 }, { "epoch": 0.10391788850306077, "grad_norm": 4.044199466705322, "learning_rate": 1.9715026027888825e-05, "loss": 0.21640148162841796, "step": 5080 }, { "epoch": 0.10412245127570459, "grad_norm": 2.713390588760376, "learning_rate": 1.97134535043803e-05, "loss": 0.19274301528930665, "step": 5090 }, { "epoch": 0.10432701404834842, "grad_norm": 1.7610270977020264, "learning_rate": 1.9711876717168913e-05, "loss": 0.19623059034347534, "step": 5100 }, { "epoch": 0.10453157682099223, "grad_norm": 4.917872428894043, "learning_rate": 1.97102956669468e-05, "loss": 0.1932083249092102, "step": 5110 }, { "epoch": 0.10473613959363605, "grad_norm": 1.687882661819458, "learning_rate": 1.9708710354407957e-05, "loss": 0.1859729766845703, "step": 5120 }, { "epoch": 0.10494070236627988, "grad_norm": 4.907532215118408, "learning_rate": 1.9707120780248253e-05, "loss": 0.212684965133667, "step": 5130 }, { "epoch": 0.1051452651389237, "grad_norm": 3.3256683349609375, "learning_rate": 1.970552694516543e-05, "loss": 0.20309035778045653, "step": 5140 }, { "epoch": 0.10534982791156751, "grad_norm": 1.5229039192199707, "learning_rate": 1.97039288498591e-05, "loss": 0.2068228006362915, "step": 5150 }, { "epoch": 0.10555439068421134, "grad_norm": 1.2316884994506836, "learning_rate": 1.9702326495030737e-05, "loss": 0.20912208557128906, "step": 5160 }, { "epoch": 0.10575895345685515, "grad_norm": 2.738706111907959, "learning_rate": 1.970071988138371e-05, "loss": 0.18875842094421386, "step": 5170 }, { "epoch": 0.10596351622949897, "grad_norm": 2.0723884105682373, "learning_rate": 1.9699109009623216e-05, "loss": 0.20781230926513672, "step": 5180 }, { "epoch": 0.1061680790021428, "grad_norm": 4.855881214141846, "learning_rate": 1.969749388045636e-05, "loss": 0.2050405263900757, "step": 5190 }, { "epoch": 0.10637264177478661, "grad_norm": 1.7537163496017456, "learning_rate": 1.9695874494592094e-05, "loss": 0.20662717819213866, "step": 5200 }, { "epoch": 0.10657720454743043, "grad_norm": 2.991532564163208, "learning_rate": 1.9694250852741243e-05, "loss": 0.22052111625671386, "step": 5210 }, { "epoch": 0.10678176732007426, "grad_norm": 3.276876211166382, "learning_rate": 1.969262295561651e-05, "loss": 0.19912232160568238, "step": 5220 }, { "epoch": 0.10698633009271807, "grad_norm": 5.158731460571289, "learning_rate": 1.9690990803932446e-05, "loss": 0.21907930374145507, "step": 5230 }, { "epoch": 0.1071908928653619, "grad_norm": 2.323293924331665, "learning_rate": 1.9689354398405486e-05, "loss": 0.21451787948608397, "step": 5240 }, { "epoch": 0.10739545563800572, "grad_norm": 1.7724477052688599, "learning_rate": 1.9687713739753934e-05, "loss": 0.20086841583251952, "step": 5250 }, { "epoch": 0.10760001841064953, "grad_norm": 2.8315927982330322, "learning_rate": 1.9686068828697944e-05, "loss": 0.21100871562957763, "step": 5260 }, { "epoch": 0.10780458118329336, "grad_norm": 3.908470630645752, "learning_rate": 1.9684419665959552e-05, "loss": 0.19466829299926758, "step": 5270 }, { "epoch": 0.10800914395593718, "grad_norm": 2.20355224609375, "learning_rate": 1.9682766252262655e-05, "loss": 0.21872832775115966, "step": 5280 }, { "epoch": 0.108213706728581, "grad_norm": 2.324165105819702, "learning_rate": 1.9681108588333014e-05, "loss": 0.2159559726715088, "step": 5290 }, { "epoch": 0.10841826950122482, "grad_norm": 4.381373882293701, "learning_rate": 1.9679446674898257e-05, "loss": 0.21519718170166016, "step": 5300 }, { "epoch": 0.10862283227386864, "grad_norm": 2.655592441558838, "learning_rate": 1.967778051268788e-05, "loss": 0.20879881381988524, "step": 5310 }, { "epoch": 0.10882739504651245, "grad_norm": 4.93942928314209, "learning_rate": 1.9676110102433243e-05, "loss": 0.19308314323425294, "step": 5320 }, { "epoch": 0.10903195781915628, "grad_norm": 2.387967586517334, "learning_rate": 1.9674435444867567e-05, "loss": 0.20240561962127684, "step": 5330 }, { "epoch": 0.1092365205918001, "grad_norm": 3.6662700176239014, "learning_rate": 1.967275654072594e-05, "loss": 0.20921897888183594, "step": 5340 }, { "epoch": 0.10944108336444393, "grad_norm": 7.475968360900879, "learning_rate": 1.9671073390745316e-05, "loss": 0.21029014587402345, "step": 5350 }, { "epoch": 0.10964564613708774, "grad_norm": 2.357041120529175, "learning_rate": 1.9669385995664506e-05, "loss": 0.19659903049468994, "step": 5360 }, { "epoch": 0.10985020890973156, "grad_norm": 3.0072226524353027, "learning_rate": 1.9667694356224194e-05, "loss": 0.20448248386383056, "step": 5370 }, { "epoch": 0.11005477168237539, "grad_norm": 3.1756067276000977, "learning_rate": 1.966599847316692e-05, "loss": 0.2098087787628174, "step": 5380 }, { "epoch": 0.1102593344550192, "grad_norm": 3.7780730724334717, "learning_rate": 1.9664298347237093e-05, "loss": 0.20176343917846679, "step": 5390 }, { "epoch": 0.11046389722766302, "grad_norm": 5.776988506317139, "learning_rate": 1.9662593979180974e-05, "loss": 0.21308083534240724, "step": 5400 }, { "epoch": 0.11066846000030685, "grad_norm": 1.956768274307251, "learning_rate": 1.9660885369746692e-05, "loss": 0.20399031639099122, "step": 5410 }, { "epoch": 0.11087302277295066, "grad_norm": 3.4470932483673096, "learning_rate": 1.965917251968424e-05, "loss": 0.19718527793884277, "step": 5420 }, { "epoch": 0.11107758554559448, "grad_norm": 2.044259548187256, "learning_rate": 1.9657455429745476e-05, "loss": 0.21142785549163817, "step": 5430 }, { "epoch": 0.11128214831823831, "grad_norm": 2.5563976764678955, "learning_rate": 1.9655734100684105e-05, "loss": 0.2079789161682129, "step": 5440 }, { "epoch": 0.11148671109088212, "grad_norm": 2.8347318172454834, "learning_rate": 1.9654008533255705e-05, "loss": 0.20593936443328859, "step": 5450 }, { "epoch": 0.11169127386352595, "grad_norm": 2.404296398162842, "learning_rate": 1.9652278728217716e-05, "loss": 0.19688718318939208, "step": 5460 }, { "epoch": 0.11189583663616977, "grad_norm": 2.176346778869629, "learning_rate": 1.9650544686329423e-05, "loss": 0.21096925735473632, "step": 5470 }, { "epoch": 0.11210039940881358, "grad_norm": 3.8437914848327637, "learning_rate": 1.9648806408351982e-05, "loss": 0.2083665370941162, "step": 5480 }, { "epoch": 0.11230496218145741, "grad_norm": 2.0247814655303955, "learning_rate": 1.9647063895048415e-05, "loss": 0.2068653106689453, "step": 5490 }, { "epoch": 0.11250952495410123, "grad_norm": 2.552976369857788, "learning_rate": 1.964531714718359e-05, "loss": 0.19748187065124512, "step": 5500 }, { "epoch": 0.11271408772674504, "grad_norm": 2.6884491443634033, "learning_rate": 1.9643566165524234e-05, "loss": 0.19506224393844604, "step": 5510 }, { "epoch": 0.11291865049938887, "grad_norm": 1.4507445096969604, "learning_rate": 1.9641810950838945e-05, "loss": 0.20797595977783204, "step": 5520 }, { "epoch": 0.11312321327203269, "grad_norm": 2.8974545001983643, "learning_rate": 1.964005150389817e-05, "loss": 0.21526846885681153, "step": 5530 }, { "epoch": 0.1133277760446765, "grad_norm": 5.785240173339844, "learning_rate": 1.9638287825474213e-05, "loss": 0.22717103958129883, "step": 5540 }, { "epoch": 0.11353233881732033, "grad_norm": 2.5795204639434814, "learning_rate": 1.9636519916341235e-05, "loss": 0.22100262641906737, "step": 5550 }, { "epoch": 0.11373690158996415, "grad_norm": 6.124509334564209, "learning_rate": 1.9634747777275263e-05, "loss": 0.21935615539550782, "step": 5560 }, { "epoch": 0.11394146436260798, "grad_norm": 2.959225654602051, "learning_rate": 1.9632971409054166e-05, "loss": 0.22073986530303955, "step": 5570 }, { "epoch": 0.11414602713525179, "grad_norm": 2.2329704761505127, "learning_rate": 1.9631190812457688e-05, "loss": 0.20728468894958496, "step": 5580 }, { "epoch": 0.11435058990789561, "grad_norm": 2.3505406379699707, "learning_rate": 1.962940598826741e-05, "loss": 0.20477471351623536, "step": 5590 }, { "epoch": 0.11455515268053944, "grad_norm": 1.3277642726898193, "learning_rate": 1.962761693726678e-05, "loss": 0.2116774797439575, "step": 5600 }, { "epoch": 0.11475971545318325, "grad_norm": 1.6227459907531738, "learning_rate": 1.96258236602411e-05, "loss": 0.20244112014770507, "step": 5610 }, { "epoch": 0.11496427822582707, "grad_norm": 2.5453102588653564, "learning_rate": 1.962402615797752e-05, "loss": 0.191015887260437, "step": 5620 }, { "epoch": 0.1151688409984709, "grad_norm": 2.0798683166503906, "learning_rate": 1.9622224431265062e-05, "loss": 0.1989881157875061, "step": 5630 }, { "epoch": 0.11537340377111471, "grad_norm": 5.322508335113525, "learning_rate": 1.9620418480894582e-05, "loss": 0.19377758502960205, "step": 5640 }, { "epoch": 0.11557796654375853, "grad_norm": 5.288447856903076, "learning_rate": 1.9618608307658796e-05, "loss": 0.2064598798751831, "step": 5650 }, { "epoch": 0.11578252931640236, "grad_norm": 6.083659648895264, "learning_rate": 1.9616793912352285e-05, "loss": 0.20019731521606446, "step": 5660 }, { "epoch": 0.11598709208904617, "grad_norm": 4.292911052703857, "learning_rate": 1.9614975295771467e-05, "loss": 0.21078922748565673, "step": 5670 }, { "epoch": 0.11619165486168999, "grad_norm": 5.209534168243408, "learning_rate": 1.961315245871463e-05, "loss": 0.19796881675720215, "step": 5680 }, { "epoch": 0.11639621763433382, "grad_norm": 2.0684564113616943, "learning_rate": 1.9611325401981892e-05, "loss": 0.20869555473327636, "step": 5690 }, { "epoch": 0.11660078040697763, "grad_norm": 4.821783065795898, "learning_rate": 1.9609494126375245e-05, "loss": 0.20314631462097169, "step": 5700 }, { "epoch": 0.11680534317962146, "grad_norm": 5.762752056121826, "learning_rate": 1.9607658632698525e-05, "loss": 0.20609130859375, "step": 5710 }, { "epoch": 0.11700990595226528, "grad_norm": 2.7702629566192627, "learning_rate": 1.960581892175741e-05, "loss": 0.2259828567504883, "step": 5720 }, { "epoch": 0.1172144687249091, "grad_norm": 3.4293324947357178, "learning_rate": 1.9603974994359447e-05, "loss": 0.21138415336608887, "step": 5730 }, { "epoch": 0.11741903149755292, "grad_norm": 3.2839715480804443, "learning_rate": 1.9602126851314022e-05, "loss": 0.20781950950622557, "step": 5740 }, { "epoch": 0.11762359427019674, "grad_norm": 3.156954526901245, "learning_rate": 1.960027449343237e-05, "loss": 0.20679769515991211, "step": 5750 }, { "epoch": 0.11782815704284055, "grad_norm": 3.727285385131836, "learning_rate": 1.9598417921527586e-05, "loss": 0.2193986415863037, "step": 5760 }, { "epoch": 0.11803271981548438, "grad_norm": 3.406757116317749, "learning_rate": 1.9596557136414606e-05, "loss": 0.1910797357559204, "step": 5770 }, { "epoch": 0.1182372825881282, "grad_norm": 4.96753454208374, "learning_rate": 1.9594692138910217e-05, "loss": 0.20905613899230957, "step": 5780 }, { "epoch": 0.11844184536077201, "grad_norm": 1.5120480060577393, "learning_rate": 1.9592822929833057e-05, "loss": 0.2127589464187622, "step": 5790 }, { "epoch": 0.11864640813341584, "grad_norm": 4.597858428955078, "learning_rate": 1.9590949510003613e-05, "loss": 0.2130443572998047, "step": 5800 }, { "epoch": 0.11885097090605966, "grad_norm": 2.2210824489593506, "learning_rate": 1.958907188024422e-05, "loss": 0.20098752975463868, "step": 5810 }, { "epoch": 0.11905553367870349, "grad_norm": 3.4446232318878174, "learning_rate": 1.958719004137906e-05, "loss": 0.20822455883026122, "step": 5820 }, { "epoch": 0.1192600964513473, "grad_norm": 2.9204516410827637, "learning_rate": 1.958530399423416e-05, "loss": 0.20161962509155273, "step": 5830 }, { "epoch": 0.11946465922399112, "grad_norm": 1.7185012102127075, "learning_rate": 1.95834137396374e-05, "loss": 0.20545239448547364, "step": 5840 }, { "epoch": 0.11966922199663495, "grad_norm": 4.138160228729248, "learning_rate": 1.95815192784185e-05, "loss": 0.18912433385848998, "step": 5850 }, { "epoch": 0.11987378476927876, "grad_norm": 4.067440032958984, "learning_rate": 1.9579620611409035e-05, "loss": 0.18035645484924318, "step": 5860 }, { "epoch": 0.12007834754192258, "grad_norm": 1.5969173908233643, "learning_rate": 1.957771773944242e-05, "loss": 0.21958906650543214, "step": 5870 }, { "epoch": 0.12028291031456641, "grad_norm": 1.2343945503234863, "learning_rate": 1.9575810663353918e-05, "loss": 0.20454397201538085, "step": 5880 }, { "epoch": 0.12048747308721022, "grad_norm": 3.657676935195923, "learning_rate": 1.957389938398063e-05, "loss": 0.20840010643005372, "step": 5890 }, { "epoch": 0.12069203585985404, "grad_norm": 3.2154765129089355, "learning_rate": 1.957198390216152e-05, "loss": 0.19131126403808593, "step": 5900 }, { "epoch": 0.12089659863249787, "grad_norm": 6.477786540985107, "learning_rate": 1.9570064218737377e-05, "loss": 0.20407943725585936, "step": 5910 }, { "epoch": 0.12110116140514168, "grad_norm": 3.948146104812622, "learning_rate": 1.9568140334550844e-05, "loss": 0.21661200523376464, "step": 5920 }, { "epoch": 0.12130572417778551, "grad_norm": 3.875690460205078, "learning_rate": 1.9566212250446404e-05, "loss": 0.1987699031829834, "step": 5930 }, { "epoch": 0.12151028695042933, "grad_norm": 2.8424124717712402, "learning_rate": 1.9564279967270393e-05, "loss": 0.20139241218566895, "step": 5940 }, { "epoch": 0.12171484972307314, "grad_norm": 5.166370391845703, "learning_rate": 1.9562343485870982e-05, "loss": 0.2188418388366699, "step": 5950 }, { "epoch": 0.12191941249571697, "grad_norm": 4.076542854309082, "learning_rate": 1.956040280709818e-05, "loss": 0.20252671241760253, "step": 5960 }, { "epoch": 0.12212397526836079, "grad_norm": 3.377389669418335, "learning_rate": 1.955845793180385e-05, "loss": 0.19956473112106324, "step": 5970 }, { "epoch": 0.1223285380410046, "grad_norm": 1.570038080215454, "learning_rate": 1.955650886084169e-05, "loss": 0.20520095825195311, "step": 5980 }, { "epoch": 0.12253310081364843, "grad_norm": 3.042245864868164, "learning_rate": 1.9554555595067237e-05, "loss": 0.18694989681243895, "step": 5990 }, { "epoch": 0.12273766358629225, "grad_norm": 5.3008575439453125, "learning_rate": 1.955259813533788e-05, "loss": 0.1874991774559021, "step": 6000 }, { "epoch": 0.12294222635893606, "grad_norm": 3.429866313934326, "learning_rate": 1.9550636482512834e-05, "loss": 0.2082390308380127, "step": 6010 }, { "epoch": 0.12314678913157989, "grad_norm": 3.034698486328125, "learning_rate": 1.9548670637453173e-05, "loss": 0.19780371189117432, "step": 6020 }, { "epoch": 0.12335135190422371, "grad_norm": 2.109842300415039, "learning_rate": 1.9546700601021796e-05, "loss": 0.20364010334014893, "step": 6030 }, { "epoch": 0.12355591467686754, "grad_norm": 4.057469367980957, "learning_rate": 1.954472637408345e-05, "loss": 0.19937132596969603, "step": 6040 }, { "epoch": 0.12376047744951135, "grad_norm": 2.6811256408691406, "learning_rate": 1.9542747957504713e-05, "loss": 0.20711896419525147, "step": 6050 }, { "epoch": 0.12396504022215517, "grad_norm": 2.4318995475769043, "learning_rate": 1.9540765352154007e-05, "loss": 0.20909478664398193, "step": 6060 }, { "epoch": 0.124169602994799, "grad_norm": 3.7821755409240723, "learning_rate": 1.9538778558901602e-05, "loss": 0.21094040870666503, "step": 6070 }, { "epoch": 0.12437416576744281, "grad_norm": 1.722357988357544, "learning_rate": 1.953678757861959e-05, "loss": 0.20092060565948486, "step": 6080 }, { "epoch": 0.12457872854008663, "grad_norm": 2.532917022705078, "learning_rate": 1.953479241218191e-05, "loss": 0.19564690589904785, "step": 6090 }, { "epoch": 0.12478329131273046, "grad_norm": 3.200904369354248, "learning_rate": 1.9532793060464334e-05, "loss": 0.19932323694229126, "step": 6100 }, { "epoch": 0.12498785408537427, "grad_norm": 2.818772077560425, "learning_rate": 1.9530789524344476e-05, "loss": 0.20391666889190674, "step": 6110 }, { "epoch": 0.1251924168580181, "grad_norm": 1.6635379791259766, "learning_rate": 1.9528781804701788e-05, "loss": 0.2106684684753418, "step": 6120 }, { "epoch": 0.1253969796306619, "grad_norm": 2.578709363937378, "learning_rate": 1.952676990241755e-05, "loss": 0.19879130125045777, "step": 6130 }, { "epoch": 0.12560154240330573, "grad_norm": 4.763222694396973, "learning_rate": 1.9524753818374886e-05, "loss": 0.1992211103439331, "step": 6140 }, { "epoch": 0.12580610517594956, "grad_norm": 2.0874130725860596, "learning_rate": 1.952273355345875e-05, "loss": 0.20967369079589843, "step": 6150 }, { "epoch": 0.12601066794859336, "grad_norm": 3.2336039543151855, "learning_rate": 1.9520709108555935e-05, "loss": 0.20968477725982665, "step": 6160 }, { "epoch": 0.1262152307212372, "grad_norm": 2.7383294105529785, "learning_rate": 1.9518680484555066e-05, "loss": 0.20421042442321777, "step": 6170 }, { "epoch": 0.12641979349388102, "grad_norm": 3.9240965843200684, "learning_rate": 1.951664768234661e-05, "loss": 0.2090524196624756, "step": 6180 }, { "epoch": 0.12662435626652482, "grad_norm": 2.3636116981506348, "learning_rate": 1.9514610702822852e-05, "loss": 0.21276700496673584, "step": 6190 }, { "epoch": 0.12682891903916865, "grad_norm": 3.157752513885498, "learning_rate": 1.9512569546877924e-05, "loss": 0.20339016914367675, "step": 6200 }, { "epoch": 0.12703348181181248, "grad_norm": 3.3983373641967773, "learning_rate": 1.951052421540779e-05, "loss": 0.2076042890548706, "step": 6210 }, { "epoch": 0.1272380445844563, "grad_norm": 4.156719207763672, "learning_rate": 1.9508474709310245e-05, "loss": 0.19986438751220703, "step": 6220 }, { "epoch": 0.1274426073571001, "grad_norm": 1.3414092063903809, "learning_rate": 1.9506421029484913e-05, "loss": 0.1973353862762451, "step": 6230 }, { "epoch": 0.12764717012974394, "grad_norm": 3.5671234130859375, "learning_rate": 1.9504363176833257e-05, "loss": 0.20624680519104005, "step": 6240 }, { "epoch": 0.12785173290238777, "grad_norm": 3.2377076148986816, "learning_rate": 1.9502301152258564e-05, "loss": 0.20398783683776855, "step": 6250 }, { "epoch": 0.12805629567503157, "grad_norm": 2.573010206222534, "learning_rate": 1.9500234956665956e-05, "loss": 0.20875988006591797, "step": 6260 }, { "epoch": 0.1282608584476754, "grad_norm": 1.9022302627563477, "learning_rate": 1.9498164590962393e-05, "loss": 0.19852855205535888, "step": 6270 }, { "epoch": 0.12846542122031923, "grad_norm": 2.4834158420562744, "learning_rate": 1.9496090056056648e-05, "loss": 0.19844080209732057, "step": 6280 }, { "epoch": 0.12866998399296303, "grad_norm": 7.103064060211182, "learning_rate": 1.9494011352859337e-05, "loss": 0.20241732597351075, "step": 6290 }, { "epoch": 0.12887454676560686, "grad_norm": 3.402212619781494, "learning_rate": 1.9491928482282907e-05, "loss": 0.20084042549133302, "step": 6300 }, { "epoch": 0.1290791095382507, "grad_norm": 1.928949236869812, "learning_rate": 1.9489841445241625e-05, "loss": 0.22112905979156494, "step": 6310 }, { "epoch": 0.1292836723108945, "grad_norm": 4.039140224456787, "learning_rate": 1.94877502426516e-05, "loss": 0.18904756307601928, "step": 6320 }, { "epoch": 0.12948823508353832, "grad_norm": 2.4201042652130127, "learning_rate": 1.9485654875430753e-05, "loss": 0.1989264726638794, "step": 6330 }, { "epoch": 0.12969279785618215, "grad_norm": 3.221057653427124, "learning_rate": 1.9483555344498848e-05, "loss": 0.20985004901885987, "step": 6340 }, { "epoch": 0.12989736062882595, "grad_norm": 1.633543848991394, "learning_rate": 1.9481451650777466e-05, "loss": 0.20043861865997314, "step": 6350 }, { "epoch": 0.13010192340146978, "grad_norm": 2.0345029830932617, "learning_rate": 1.947934379519002e-05, "loss": 0.2019228458404541, "step": 6360 }, { "epoch": 0.1303064861741136, "grad_norm": 2.247330665588379, "learning_rate": 1.9477231778661754e-05, "loss": 0.20375189781188965, "step": 6370 }, { "epoch": 0.1305110489467574, "grad_norm": 2.475524425506592, "learning_rate": 1.947511560211973e-05, "loss": 0.20152835845947265, "step": 6380 }, { "epoch": 0.13071561171940124, "grad_norm": 1.870516300201416, "learning_rate": 1.9472995266492836e-05, "loss": 0.20888123512268067, "step": 6390 }, { "epoch": 0.13092017449204507, "grad_norm": 2.4840426445007324, "learning_rate": 1.9470870772711798e-05, "loss": 0.18873603343963624, "step": 6400 }, { "epoch": 0.13112473726468887, "grad_norm": 2.1789557933807373, "learning_rate": 1.9468742121709155e-05, "loss": 0.20105478763580323, "step": 6410 }, { "epoch": 0.1313293000373327, "grad_norm": 4.587846755981445, "learning_rate": 1.9466609314419273e-05, "loss": 0.21024694442749023, "step": 6420 }, { "epoch": 0.13153386280997653, "grad_norm": 2.7457315921783447, "learning_rate": 1.9464472351778343e-05, "loss": 0.2042691707611084, "step": 6430 }, { "epoch": 0.13173842558262033, "grad_norm": 2.952850580215454, "learning_rate": 1.9462331234724386e-05, "loss": 0.2062549591064453, "step": 6440 }, { "epoch": 0.13194298835526416, "grad_norm": 1.2966537475585938, "learning_rate": 1.9460185964197234e-05, "loss": 0.21331892013549805, "step": 6450 }, { "epoch": 0.132147551127908, "grad_norm": 2.926558017730713, "learning_rate": 1.945803654113856e-05, "loss": 0.20357508659362794, "step": 6460 }, { "epoch": 0.13235211390055182, "grad_norm": 1.997822642326355, "learning_rate": 1.945588296649184e-05, "loss": 0.20648009777069093, "step": 6470 }, { "epoch": 0.13255667667319562, "grad_norm": 5.226727485656738, "learning_rate": 1.9453725241202384e-05, "loss": 0.19657326936721803, "step": 6480 }, { "epoch": 0.13276123944583945, "grad_norm": 4.423237323760986, "learning_rate": 1.9451563366217325e-05, "loss": 0.20528812408447267, "step": 6490 }, { "epoch": 0.13296580221848328, "grad_norm": 4.056731700897217, "learning_rate": 1.9449397342485608e-05, "loss": 0.2105506420135498, "step": 6500 }, { "epoch": 0.13317036499112708, "grad_norm": 2.310882091522217, "learning_rate": 1.944722717095801e-05, "loss": 0.20199937820434571, "step": 6510 }, { "epoch": 0.1333749277637709, "grad_norm": 1.8181730508804321, "learning_rate": 1.9445052852587125e-05, "loss": 0.1911325693130493, "step": 6520 }, { "epoch": 0.13357949053641474, "grad_norm": 3.982248306274414, "learning_rate": 1.9442874388327363e-05, "loss": 0.22015013694763183, "step": 6530 }, { "epoch": 0.13378405330905854, "grad_norm": 3.102811813354492, "learning_rate": 1.944069177913496e-05, "loss": 0.19735740423202514, "step": 6540 }, { "epoch": 0.13398861608170237, "grad_norm": 3.7971677780151367, "learning_rate": 1.943850502596796e-05, "loss": 0.1980788826942444, "step": 6550 }, { "epoch": 0.1341931788543462, "grad_norm": 3.4741814136505127, "learning_rate": 1.9436314129786248e-05, "loss": 0.20477714538574218, "step": 6560 }, { "epoch": 0.13439774162699, "grad_norm": 2.393639326095581, "learning_rate": 1.94341190915515e-05, "loss": 0.21086082458496094, "step": 6570 }, { "epoch": 0.13460230439963383, "grad_norm": 3.7378506660461426, "learning_rate": 1.9431919912227238e-05, "loss": 0.19599554538726807, "step": 6580 }, { "epoch": 0.13480686717227766, "grad_norm": 2.3515589237213135, "learning_rate": 1.9429716592778775e-05, "loss": 0.2048488140106201, "step": 6590 }, { "epoch": 0.13501142994492146, "grad_norm": 3.5860166549682617, "learning_rate": 1.942750913417326e-05, "loss": 0.19626845121383668, "step": 6600 }, { "epoch": 0.1352159927175653, "grad_norm": 2.9817943572998047, "learning_rate": 1.9425297537379654e-05, "loss": 0.20950582027435302, "step": 6610 }, { "epoch": 0.13542055549020912, "grad_norm": 2.0552449226379395, "learning_rate": 1.942308180336873e-05, "loss": 0.2093959093093872, "step": 6620 }, { "epoch": 0.13562511826285292, "grad_norm": 5.055057048797607, "learning_rate": 1.9420861933113083e-05, "loss": 0.20399293899536133, "step": 6630 }, { "epoch": 0.13582968103549675, "grad_norm": 2.4517009258270264, "learning_rate": 1.9418637927587117e-05, "loss": 0.20755982398986816, "step": 6640 }, { "epoch": 0.13603424380814058, "grad_norm": 2.3688604831695557, "learning_rate": 1.9416409787767066e-05, "loss": 0.18947794437408447, "step": 6650 }, { "epoch": 0.13623880658078438, "grad_norm": 1.3900024890899658, "learning_rate": 1.9414177514630956e-05, "loss": 0.19028708934783936, "step": 6660 }, { "epoch": 0.1364433693534282, "grad_norm": 4.924600601196289, "learning_rate": 1.9411941109158642e-05, "loss": 0.1896311402320862, "step": 6670 }, { "epoch": 0.13664793212607204, "grad_norm": 1.3108241558074951, "learning_rate": 1.9409700572331792e-05, "loss": 0.20342302322387695, "step": 6680 }, { "epoch": 0.13685249489871587, "grad_norm": 4.340987205505371, "learning_rate": 1.9407455905133892e-05, "loss": 0.21173784732818604, "step": 6690 }, { "epoch": 0.13705705767135967, "grad_norm": 5.295881271362305, "learning_rate": 1.940520710855022e-05, "loss": 0.2046372652053833, "step": 6700 }, { "epoch": 0.1372616204440035, "grad_norm": 2.803255081176758, "learning_rate": 1.9402954183567894e-05, "loss": 0.2041844367980957, "step": 6710 }, { "epoch": 0.13746618321664733, "grad_norm": 5.397721767425537, "learning_rate": 1.9400697131175826e-05, "loss": 0.1862511157989502, "step": 6720 }, { "epoch": 0.13767074598929113, "grad_norm": 4.811631679534912, "learning_rate": 1.9398435952364743e-05, "loss": 0.209557843208313, "step": 6730 }, { "epoch": 0.13787530876193496, "grad_norm": 3.270479917526245, "learning_rate": 1.939617064812719e-05, "loss": 0.18961185216903687, "step": 6740 }, { "epoch": 0.1380798715345788, "grad_norm": 3.3926568031311035, "learning_rate": 1.939390121945752e-05, "loss": 0.22242674827575684, "step": 6750 }, { "epoch": 0.1382844343072226, "grad_norm": 2.900690793991089, "learning_rate": 1.9391627667351885e-05, "loss": 0.19621800184249877, "step": 6760 }, { "epoch": 0.13848899707986642, "grad_norm": 5.089657306671143, "learning_rate": 1.9389349992808264e-05, "loss": 0.19418163299560548, "step": 6770 }, { "epoch": 0.13869355985251025, "grad_norm": 1.4914050102233887, "learning_rate": 1.9387068196826434e-05, "loss": 0.21317996978759765, "step": 6780 }, { "epoch": 0.13889812262515405, "grad_norm": 3.0177550315856934, "learning_rate": 1.9384782280407992e-05, "loss": 0.19646302461624146, "step": 6790 }, { "epoch": 0.13910268539779788, "grad_norm": 2.6852915287017822, "learning_rate": 1.9382492244556333e-05, "loss": 0.18625689744949342, "step": 6800 }, { "epoch": 0.1393072481704417, "grad_norm": 2.9123032093048096, "learning_rate": 1.9380198090276662e-05, "loss": 0.2051666259765625, "step": 6810 }, { "epoch": 0.1395118109430855, "grad_norm": 1.8890711069107056, "learning_rate": 1.9377899818575995e-05, "loss": 0.20902442932128906, "step": 6820 }, { "epoch": 0.13971637371572934, "grad_norm": 1.9752880334854126, "learning_rate": 1.9375597430463157e-05, "loss": 0.19661890268325805, "step": 6830 }, { "epoch": 0.13992093648837317, "grad_norm": 3.791746139526367, "learning_rate": 1.937329092694877e-05, "loss": 0.2013789653778076, "step": 6840 }, { "epoch": 0.14012549926101697, "grad_norm": 1.787878394126892, "learning_rate": 1.9370980309045277e-05, "loss": 0.2092000961303711, "step": 6850 }, { "epoch": 0.1403300620336608, "grad_norm": 1.7937291860580444, "learning_rate": 1.936866557776692e-05, "loss": 0.21595613956451415, "step": 6860 }, { "epoch": 0.14053462480630463, "grad_norm": 1.2309014797210693, "learning_rate": 1.936634673412974e-05, "loss": 0.2048292875289917, "step": 6870 }, { "epoch": 0.14073918757894843, "grad_norm": 3.576612949371338, "learning_rate": 1.9364023779151594e-05, "loss": 0.1935584545135498, "step": 6880 }, { "epoch": 0.14094375035159226, "grad_norm": 1.7655757665634155, "learning_rate": 1.9361696713852135e-05, "loss": 0.19613423347473144, "step": 6890 }, { "epoch": 0.1411483131242361, "grad_norm": 1.4287970066070557, "learning_rate": 1.935936553925283e-05, "loss": 0.19047908782958983, "step": 6900 }, { "epoch": 0.1413528758968799, "grad_norm": 1.2125201225280762, "learning_rate": 1.9357030256376937e-05, "loss": 0.19918121099472047, "step": 6910 }, { "epoch": 0.14155743866952372, "grad_norm": 2.586432933807373, "learning_rate": 1.9354690866249532e-05, "loss": 0.18899195194244384, "step": 6920 }, { "epoch": 0.14176200144216755, "grad_norm": 2.2496907711029053, "learning_rate": 1.9352347369897475e-05, "loss": 0.1889805793762207, "step": 6930 }, { "epoch": 0.14196656421481138, "grad_norm": 1.7276713848114014, "learning_rate": 1.934999976834945e-05, "loss": 0.19936851263046265, "step": 6940 }, { "epoch": 0.14217112698745518, "grad_norm": 5.151482582092285, "learning_rate": 1.9347648062635927e-05, "loss": 0.1954187870025635, "step": 6950 }, { "epoch": 0.142375689760099, "grad_norm": 2.3329355716705322, "learning_rate": 1.934529225378918e-05, "loss": 0.2104471206665039, "step": 6960 }, { "epoch": 0.14258025253274284, "grad_norm": 1.4809755086898804, "learning_rate": 1.9342932342843302e-05, "loss": 0.18869824409484864, "step": 6970 }, { "epoch": 0.14278481530538664, "grad_norm": 2.137760639190674, "learning_rate": 1.9340568330834154e-05, "loss": 0.2081890106201172, "step": 6980 }, { "epoch": 0.14298937807803047, "grad_norm": 4.57649564743042, "learning_rate": 1.9338200218799422e-05, "loss": 0.20476093292236328, "step": 6990 }, { "epoch": 0.1431939408506743, "grad_norm": 4.493127822875977, "learning_rate": 1.9335828007778587e-05, "loss": 0.1932983160018921, "step": 7000 }, { "epoch": 0.1433985036233181, "grad_norm": 2.10902738571167, "learning_rate": 1.9333451698812918e-05, "loss": 0.20787622928619384, "step": 7010 }, { "epoch": 0.14360306639596193, "grad_norm": 2.297637701034546, "learning_rate": 1.9331071292945504e-05, "loss": 0.1978832960128784, "step": 7020 }, { "epoch": 0.14380762916860576, "grad_norm": 2.813577175140381, "learning_rate": 1.9328686791221213e-05, "loss": 0.201112699508667, "step": 7030 }, { "epoch": 0.14401219194124956, "grad_norm": 2.3630332946777344, "learning_rate": 1.9326298194686718e-05, "loss": 0.1903133988380432, "step": 7040 }, { "epoch": 0.1442167547138934, "grad_norm": 1.519186019897461, "learning_rate": 1.932390550439049e-05, "loss": 0.18624114990234375, "step": 7050 }, { "epoch": 0.14442131748653722, "grad_norm": 3.841773271560669, "learning_rate": 1.9321508721382792e-05, "loss": 0.19995572566986083, "step": 7060 }, { "epoch": 0.14462588025918102, "grad_norm": 1.8641512393951416, "learning_rate": 1.931910784671569e-05, "loss": 0.18977271318435668, "step": 7070 }, { "epoch": 0.14483044303182485, "grad_norm": 2.800644636154175, "learning_rate": 1.931670288144305e-05, "loss": 0.22018826007843018, "step": 7080 }, { "epoch": 0.14503500580446868, "grad_norm": 2.7263715267181396, "learning_rate": 1.9314293826620517e-05, "loss": 0.21303842067718506, "step": 7090 }, { "epoch": 0.14523956857711248, "grad_norm": 3.0434744358062744, "learning_rate": 1.9311880683305547e-05, "loss": 0.19830844402313233, "step": 7100 }, { "epoch": 0.1454441313497563, "grad_norm": 3.042694568634033, "learning_rate": 1.9309463452557385e-05, "loss": 0.1901804566383362, "step": 7110 }, { "epoch": 0.14564869412240014, "grad_norm": 1.8786485195159912, "learning_rate": 1.930704213543707e-05, "loss": 0.1955502986907959, "step": 7120 }, { "epoch": 0.14585325689504394, "grad_norm": 2.1625359058380127, "learning_rate": 1.9304616733007434e-05, "loss": 0.2020191431045532, "step": 7130 }, { "epoch": 0.14605781966768777, "grad_norm": 1.4912737607955933, "learning_rate": 1.93021872463331e-05, "loss": 0.19476643800735474, "step": 7140 }, { "epoch": 0.1462623824403316, "grad_norm": 1.855069875717163, "learning_rate": 1.9299753676480494e-05, "loss": 0.21094846725463867, "step": 7150 }, { "epoch": 0.14646694521297543, "grad_norm": 3.67063307762146, "learning_rate": 1.929731602451782e-05, "loss": 0.19373540878295897, "step": 7160 }, { "epoch": 0.14667150798561923, "grad_norm": 2.202709913253784, "learning_rate": 1.9294874291515082e-05, "loss": 0.20761613845825194, "step": 7170 }, { "epoch": 0.14687607075826306, "grad_norm": 3.201435089111328, "learning_rate": 1.9292428478544076e-05, "loss": 0.18923375606536866, "step": 7180 }, { "epoch": 0.1470806335309069, "grad_norm": 1.5726420879364014, "learning_rate": 1.9289978586678387e-05, "loss": 0.20611767768859862, "step": 7190 }, { "epoch": 0.1472851963035507, "grad_norm": 2.435359001159668, "learning_rate": 1.928752461699339e-05, "loss": 0.18946633338928223, "step": 7200 }, { "epoch": 0.14748975907619452, "grad_norm": 3.9266092777252197, "learning_rate": 1.928506657056625e-05, "loss": 0.19452003240585328, "step": 7210 }, { "epoch": 0.14769432184883835, "grad_norm": 2.0061519145965576, "learning_rate": 1.928260444847593e-05, "loss": 0.2060300350189209, "step": 7220 }, { "epoch": 0.14789888462148215, "grad_norm": 6.403658390045166, "learning_rate": 1.9280138251803165e-05, "loss": 0.2057652473449707, "step": 7230 }, { "epoch": 0.14810344739412598, "grad_norm": 4.425238609313965, "learning_rate": 1.9277667981630487e-05, "loss": 0.20090391635894775, "step": 7240 }, { "epoch": 0.1483080101667698, "grad_norm": 2.314406394958496, "learning_rate": 1.927519363904222e-05, "loss": 0.20668067932128906, "step": 7250 }, { "epoch": 0.1485125729394136, "grad_norm": 3.908456325531006, "learning_rate": 1.927271522512447e-05, "loss": 0.2155350923538208, "step": 7260 }, { "epoch": 0.14871713571205744, "grad_norm": 2.5477454662323, "learning_rate": 1.927023274096514e-05, "loss": 0.19797061681747435, "step": 7270 }, { "epoch": 0.14892169848470127, "grad_norm": 3.2877519130706787, "learning_rate": 1.9267746187653903e-05, "loss": 0.2111807346343994, "step": 7280 }, { "epoch": 0.14912626125734507, "grad_norm": 3.537240743637085, "learning_rate": 1.9265255566282227e-05, "loss": 0.19609345197677613, "step": 7290 }, { "epoch": 0.1493308240299889, "grad_norm": 23.016305923461914, "learning_rate": 1.926276087794337e-05, "loss": 2.550894546508789, "step": 7300 }, { "epoch": 0.14953538680263273, "grad_norm": 3.15549373626709, "learning_rate": 1.9260262123732373e-05, "loss": 0.6483545780181885, "step": 7310 }, { "epoch": 0.14973994957527653, "grad_norm": 4.8404364585876465, "learning_rate": 1.9257759304746054e-05, "loss": 0.3754291534423828, "step": 7320 }, { "epoch": 0.14994451234792036, "grad_norm": 2.8259408473968506, "learning_rate": 1.925525242208302e-05, "loss": 0.2789908885955811, "step": 7330 }, { "epoch": 0.1501490751205642, "grad_norm": 2.3227756023406982, "learning_rate": 1.925274147684367e-05, "loss": 0.2816964626312256, "step": 7340 }, { "epoch": 0.150353637893208, "grad_norm": 2.5063464641571045, "learning_rate": 1.9250226470130172e-05, "loss": 0.2504149913787842, "step": 7350 }, { "epoch": 0.15055820066585182, "grad_norm": 2.715757131576538, "learning_rate": 1.924770740304649e-05, "loss": 0.2692726135253906, "step": 7360 }, { "epoch": 0.15076276343849565, "grad_norm": 2.17575740814209, "learning_rate": 1.9245184276698355e-05, "loss": 0.2609139919281006, "step": 7370 }, { "epoch": 0.15096732621113945, "grad_norm": 4.237732410430908, "learning_rate": 1.92426570921933e-05, "loss": 0.25251379013061526, "step": 7380 }, { "epoch": 0.15117188898378328, "grad_norm": 4.021399021148682, "learning_rate": 1.9240125850640617e-05, "loss": 0.2539872169494629, "step": 7390 }, { "epoch": 0.1513764517564271, "grad_norm": 2.089712619781494, "learning_rate": 1.9237590553151396e-05, "loss": 0.24187374114990234, "step": 7400 }, { "epoch": 0.15158101452907094, "grad_norm": 2.6592392921447754, "learning_rate": 1.9235051200838502e-05, "loss": 0.23484227657318116, "step": 7410 }, { "epoch": 0.15178557730171474, "grad_norm": 3.434704065322876, "learning_rate": 1.9232507794816576e-05, "loss": 0.23966851234436035, "step": 7420 }, { "epoch": 0.15199014007435857, "grad_norm": 2.231567144393921, "learning_rate": 1.9229960336202045e-05, "loss": 0.22122640609741212, "step": 7430 }, { "epoch": 0.1521947028470024, "grad_norm": 3.4217305183410645, "learning_rate": 1.9227408826113107e-05, "loss": 0.24497480392456056, "step": 7440 }, { "epoch": 0.1523992656196462, "grad_norm": 1.7165706157684326, "learning_rate": 1.9224853265669744e-05, "loss": 0.22288126945495607, "step": 7450 }, { "epoch": 0.15260382839229003, "grad_norm": 2.047271251678467, "learning_rate": 1.922229365599372e-05, "loss": 0.22021198272705078, "step": 7460 }, { "epoch": 0.15280839116493386, "grad_norm": 3.1419625282287598, "learning_rate": 1.9219729998208563e-05, "loss": 0.2097097635269165, "step": 7470 }, { "epoch": 0.15301295393757766, "grad_norm": 2.8029394149780273, "learning_rate": 1.921716229343959e-05, "loss": 0.19685648679733275, "step": 7480 }, { "epoch": 0.1532175167102215, "grad_norm": 5.459460258483887, "learning_rate": 1.9214590542813888e-05, "loss": 0.23476753234863282, "step": 7490 }, { "epoch": 0.15342207948286532, "grad_norm": 2.1736562252044678, "learning_rate": 1.9212014747460328e-05, "loss": 0.20065727233886718, "step": 7500 }, { "epoch": 0.15362664225550912, "grad_norm": 2.144090175628662, "learning_rate": 1.9209434908509544e-05, "loss": 0.19514750242233275, "step": 7510 }, { "epoch": 0.15383120502815295, "grad_norm": 3.446798324584961, "learning_rate": 1.9206851027093958e-05, "loss": 0.21259984970092774, "step": 7520 }, { "epoch": 0.15403576780079678, "grad_norm": 4.164159774780273, "learning_rate": 1.9204263104347755e-05, "loss": 0.2261418342590332, "step": 7530 }, { "epoch": 0.15424033057344058, "grad_norm": 1.6668827533721924, "learning_rate": 1.92016711414069e-05, "loss": 0.21841959953308104, "step": 7540 }, { "epoch": 0.1544448933460844, "grad_norm": 2.3023316860198975, "learning_rate": 1.919907513940913e-05, "loss": 0.20594754219055175, "step": 7550 }, { "epoch": 0.15464945611872824, "grad_norm": 2.4465863704681396, "learning_rate": 1.9196475099493957e-05, "loss": 0.20071992874145508, "step": 7560 }, { "epoch": 0.15485401889137204, "grad_norm": 3.1194581985473633, "learning_rate": 1.9193871022802666e-05, "loss": 0.21410057544708253, "step": 7570 }, { "epoch": 0.15505858166401587, "grad_norm": 3.9930450916290283, "learning_rate": 1.9191262910478307e-05, "loss": 0.2122809648513794, "step": 7580 }, { "epoch": 0.1552631444366597, "grad_norm": 3.181513786315918, "learning_rate": 1.918865076366571e-05, "loss": 0.2120812177658081, "step": 7590 }, { "epoch": 0.1554677072093035, "grad_norm": 0.9615801572799683, "learning_rate": 1.9186034583511465e-05, "loss": 0.21173839569091796, "step": 7600 }, { "epoch": 0.15567226998194733, "grad_norm": 1.8856161832809448, "learning_rate": 1.918341437116395e-05, "loss": 0.20046639442443848, "step": 7610 }, { "epoch": 0.15587683275459116, "grad_norm": 1.9387249946594238, "learning_rate": 1.9180790127773298e-05, "loss": 0.2044283628463745, "step": 7620 }, { "epoch": 0.156081395527235, "grad_norm": 2.781381845474243, "learning_rate": 1.917816185449141e-05, "loss": 0.20567002296447753, "step": 7630 }, { "epoch": 0.1562859582998788, "grad_norm": 2.591874599456787, "learning_rate": 1.9175529552471974e-05, "loss": 0.18772988319396972, "step": 7640 }, { "epoch": 0.15649052107252262, "grad_norm": 4.575911998748779, "learning_rate": 1.917289322287042e-05, "loss": 0.21329948902130128, "step": 7650 }, { "epoch": 0.15669508384516645, "grad_norm": 2.7959702014923096, "learning_rate": 1.917025286684397e-05, "loss": 0.19244089126586914, "step": 7660 }, { "epoch": 0.15689964661781025, "grad_norm": 2.2618682384490967, "learning_rate": 1.9167608485551606e-05, "loss": 0.20245704650878907, "step": 7670 }, { "epoch": 0.15710420939045408, "grad_norm": 3.4579787254333496, "learning_rate": 1.916496008015406e-05, "loss": 0.20928993225097656, "step": 7680 }, { "epoch": 0.1573087721630979, "grad_norm": 4.360982418060303, "learning_rate": 1.9162307651813863e-05, "loss": 0.21614644527435303, "step": 7690 }, { "epoch": 0.1575133349357417, "grad_norm": 3.75226092338562, "learning_rate": 1.915965120169528e-05, "loss": 0.21344218254089356, "step": 7700 }, { "epoch": 0.15771789770838554, "grad_norm": 1.385000228881836, "learning_rate": 1.9156990730964364e-05, "loss": 0.22002410888671875, "step": 7710 }, { "epoch": 0.15792246048102937, "grad_norm": 1.961225152015686, "learning_rate": 1.9154326240788914e-05, "loss": 0.1980452060699463, "step": 7720 }, { "epoch": 0.15812702325367317, "grad_norm": 2.774507999420166, "learning_rate": 1.915165773233851e-05, "loss": 0.22439026832580566, "step": 7730 }, { "epoch": 0.158331586026317, "grad_norm": 3.8355889320373535, "learning_rate": 1.9148985206784488e-05, "loss": 0.1878036379814148, "step": 7740 }, { "epoch": 0.15853614879896083, "grad_norm": 3.4371538162231445, "learning_rate": 1.914630866529995e-05, "loss": 0.20822253227233886, "step": 7750 }, { "epoch": 0.15874071157160463, "grad_norm": 4.682356834411621, "learning_rate": 1.9143628109059755e-05, "loss": 0.19313712120056153, "step": 7760 }, { "epoch": 0.15894527434424846, "grad_norm": 2.6162798404693604, "learning_rate": 1.914094353924053e-05, "loss": 0.20734767913818358, "step": 7770 }, { "epoch": 0.1591498371168923, "grad_norm": 1.9381966590881348, "learning_rate": 1.9138254957020664e-05, "loss": 0.20519630908966063, "step": 7780 }, { "epoch": 0.1593543998895361, "grad_norm": 5.331486225128174, "learning_rate": 1.91355623635803e-05, "loss": 0.19855448007583618, "step": 7790 }, { "epoch": 0.15955896266217992, "grad_norm": 3.0999951362609863, "learning_rate": 1.9132865760101355e-05, "loss": 0.1967594265937805, "step": 7800 }, { "epoch": 0.15976352543482375, "grad_norm": 1.780242681503296, "learning_rate": 1.913016514776749e-05, "loss": 0.18975123167037963, "step": 7810 }, { "epoch": 0.15996808820746755, "grad_norm": 3.307103395462036, "learning_rate": 1.912746052776414e-05, "loss": 0.1997169852256775, "step": 7820 }, { "epoch": 0.16017265098011138, "grad_norm": 1.8082481622695923, "learning_rate": 1.912475190127849e-05, "loss": 0.19931379556655884, "step": 7830 }, { "epoch": 0.1603772137527552, "grad_norm": 1.78659188747406, "learning_rate": 1.9122039269499488e-05, "loss": 0.19686729907989503, "step": 7840 }, { "epoch": 0.160581776525399, "grad_norm": 2.2578952312469482, "learning_rate": 1.9119322633617838e-05, "loss": 0.18392138481140136, "step": 7850 }, { "epoch": 0.16078633929804284, "grad_norm": 2.3891372680664062, "learning_rate": 1.9116601994826002e-05, "loss": 0.20663919448852539, "step": 7860 }, { "epoch": 0.16099090207068667, "grad_norm": 2.670315980911255, "learning_rate": 1.9113877354318202e-05, "loss": 0.18594683408737184, "step": 7870 }, { "epoch": 0.1611954648433305, "grad_norm": 7.225728988647461, "learning_rate": 1.911114871329041e-05, "loss": 0.20041134357452392, "step": 7880 }, { "epoch": 0.1614000276159743, "grad_norm": 4.274582862854004, "learning_rate": 1.910841607294036e-05, "loss": 0.2143160581588745, "step": 7890 }, { "epoch": 0.16160459038861813, "grad_norm": 4.071975231170654, "learning_rate": 1.9105679434467544e-05, "loss": 0.2118499755859375, "step": 7900 }, { "epoch": 0.16180915316126196, "grad_norm": 2.405636787414551, "learning_rate": 1.91029387990732e-05, "loss": 0.19898041486740112, "step": 7910 }, { "epoch": 0.16201371593390576, "grad_norm": 1.5764822959899902, "learning_rate": 1.9100194167960322e-05, "loss": 0.19597249031066893, "step": 7920 }, { "epoch": 0.1622182787065496, "grad_norm": 3.4361166954040527, "learning_rate": 1.9097445542333662e-05, "loss": 0.1861456036567688, "step": 7930 }, { "epoch": 0.16242284147919342, "grad_norm": 3.6971356868743896, "learning_rate": 1.909469292339973e-05, "loss": 0.1998357057571411, "step": 7940 }, { "epoch": 0.16262740425183722, "grad_norm": 1.1150234937667847, "learning_rate": 1.909193631236678e-05, "loss": 0.20401487350463868, "step": 7950 }, { "epoch": 0.16283196702448105, "grad_norm": 4.943863391876221, "learning_rate": 1.908917571044482e-05, "loss": 0.17840291261672975, "step": 7960 }, { "epoch": 0.16303652979712488, "grad_norm": 6.2282490730285645, "learning_rate": 1.908641111884561e-05, "loss": 0.2083420753479004, "step": 7970 }, { "epoch": 0.16324109256976868, "grad_norm": 3.2929463386535645, "learning_rate": 1.9083642538782668e-05, "loss": 0.19327929019927978, "step": 7980 }, { "epoch": 0.1634456553424125, "grad_norm": 2.783440589904785, "learning_rate": 1.9080869971471253e-05, "loss": 0.19245270490646363, "step": 7990 }, { "epoch": 0.16365021811505634, "grad_norm": 1.2005380392074585, "learning_rate": 1.9078093418128382e-05, "loss": 0.19534523487091066, "step": 8000 }, { "epoch": 0.16385478088770014, "grad_norm": 3.8213894367218018, "learning_rate": 1.9075312879972808e-05, "loss": 0.19612051248550416, "step": 8010 }, { "epoch": 0.16405934366034397, "grad_norm": 3.4468538761138916, "learning_rate": 1.9072528358225057e-05, "loss": 0.20270512104034424, "step": 8020 }, { "epoch": 0.1642639064329878, "grad_norm": 1.5454055070877075, "learning_rate": 1.906973985410738e-05, "loss": 0.19566009044647217, "step": 8030 }, { "epoch": 0.1644684692056316, "grad_norm": 6.07951545715332, "learning_rate": 1.906694736884379e-05, "loss": 0.20123233795166015, "step": 8040 }, { "epoch": 0.16467303197827543, "grad_norm": 3.031359910964966, "learning_rate": 1.9064150903660043e-05, "loss": 0.20627710819244385, "step": 8050 }, { "epoch": 0.16487759475091926, "grad_norm": 5.057559013366699, "learning_rate": 1.9061350459783643e-05, "loss": 0.2040529727935791, "step": 8060 }, { "epoch": 0.16508215752356306, "grad_norm": 3.608736991882324, "learning_rate": 1.9058546038443837e-05, "loss": 0.19901137351989745, "step": 8070 }, { "epoch": 0.1652867202962069, "grad_norm": 5.81221342086792, "learning_rate": 1.9055737640871622e-05, "loss": 0.192382550239563, "step": 8080 }, { "epoch": 0.16549128306885072, "grad_norm": 2.668978691101074, "learning_rate": 1.905292526829974e-05, "loss": 0.20438816547393798, "step": 8090 }, { "epoch": 0.16569584584149455, "grad_norm": 2.834829568862915, "learning_rate": 1.9050108921962675e-05, "loss": 0.19665929079055786, "step": 8100 }, { "epoch": 0.16590040861413835, "grad_norm": 4.737696170806885, "learning_rate": 1.904728860309666e-05, "loss": 0.2058798313140869, "step": 8110 }, { "epoch": 0.16610497138678218, "grad_norm": 1.4150917530059814, "learning_rate": 1.904446431293967e-05, "loss": 0.1911359906196594, "step": 8120 }, { "epoch": 0.166309534159426, "grad_norm": 3.7209410667419434, "learning_rate": 1.9041636052731415e-05, "loss": 0.19076687097549438, "step": 8130 }, { "epoch": 0.1665140969320698, "grad_norm": 2.4184682369232178, "learning_rate": 1.9038803823713357e-05, "loss": 0.20573780536651612, "step": 8140 }, { "epoch": 0.16671865970471364, "grad_norm": 2.4659228324890137, "learning_rate": 1.9035967627128705e-05, "loss": 0.19187896251678466, "step": 8150 }, { "epoch": 0.16692322247735747, "grad_norm": 5.546408176422119, "learning_rate": 1.9033127464222393e-05, "loss": 0.20835909843444825, "step": 8160 }, { "epoch": 0.16712778525000127, "grad_norm": 1.7434555292129517, "learning_rate": 1.9030283336241112e-05, "loss": 0.19770486354827882, "step": 8170 }, { "epoch": 0.1673323480226451, "grad_norm": 2.9782073497772217, "learning_rate": 1.9027435244433284e-05, "loss": 0.19235713481903077, "step": 8180 }, { "epoch": 0.16753691079528893, "grad_norm": 2.8840157985687256, "learning_rate": 1.902458319004907e-05, "loss": 0.19673073291778564, "step": 8190 }, { "epoch": 0.16774147356793273, "grad_norm": 2.1326615810394287, "learning_rate": 1.902172717434039e-05, "loss": 0.20575737953186035, "step": 8200 }, { "epoch": 0.16794603634057656, "grad_norm": 2.993520975112915, "learning_rate": 1.9018867198560865e-05, "loss": 0.18845127820968627, "step": 8210 }, { "epoch": 0.1681505991132204, "grad_norm": 3.270054578781128, "learning_rate": 1.901600326396589e-05, "loss": 0.18662030696868898, "step": 8220 }, { "epoch": 0.1683551618858642, "grad_norm": 2.649249792098999, "learning_rate": 1.9013135371812578e-05, "loss": 0.19193129539489745, "step": 8230 }, { "epoch": 0.16855972465850802, "grad_norm": 6.131075859069824, "learning_rate": 1.9010263523359786e-05, "loss": 0.19039456844329833, "step": 8240 }, { "epoch": 0.16876428743115185, "grad_norm": 2.839677095413208, "learning_rate": 1.9007387719868113e-05, "loss": 0.1994660258293152, "step": 8250 }, { "epoch": 0.16896885020379565, "grad_norm": 1.6985896825790405, "learning_rate": 1.900450796259988e-05, "loss": 0.192476487159729, "step": 8260 }, { "epoch": 0.16917341297643948, "grad_norm": 0.8741666674613953, "learning_rate": 1.9001624252819152e-05, "loss": 0.20351431369781495, "step": 8270 }, { "epoch": 0.1693779757490833, "grad_norm": 2.740201473236084, "learning_rate": 1.8998736591791732e-05, "loss": 0.19171590805053712, "step": 8280 }, { "epoch": 0.1695825385217271, "grad_norm": 2.9323058128356934, "learning_rate": 1.8995844980785154e-05, "loss": 0.1999297857284546, "step": 8290 }, { "epoch": 0.16978710129437094, "grad_norm": 1.7256803512573242, "learning_rate": 1.899294942106868e-05, "loss": 0.207281231880188, "step": 8300 }, { "epoch": 0.16999166406701477, "grad_norm": 2.65315580368042, "learning_rate": 1.8990049913913317e-05, "loss": 0.18801995515823364, "step": 8310 }, { "epoch": 0.17019622683965857, "grad_norm": 3.611525297164917, "learning_rate": 1.8987146460591793e-05, "loss": 0.19129812717437744, "step": 8320 }, { "epoch": 0.1704007896123024, "grad_norm": 1.447031021118164, "learning_rate": 1.8984239062378576e-05, "loss": 0.19522907733917236, "step": 8330 }, { "epoch": 0.17060535238494623, "grad_norm": 1.7795121669769287, "learning_rate": 1.8981327720549862e-05, "loss": 0.1900084376335144, "step": 8340 }, { "epoch": 0.17080991515759006, "grad_norm": 1.8126002550125122, "learning_rate": 1.8978412436383582e-05, "loss": 0.19626412391662598, "step": 8350 }, { "epoch": 0.17101447793023386, "grad_norm": 1.058048963546753, "learning_rate": 1.8975493211159394e-05, "loss": 0.19064246416091918, "step": 8360 }, { "epoch": 0.1712190407028777, "grad_norm": 3.113712787628174, "learning_rate": 1.8972570046158684e-05, "loss": 0.19929959774017333, "step": 8370 }, { "epoch": 0.17142360347552152, "grad_norm": 2.954845666885376, "learning_rate": 1.8969642942664573e-05, "loss": 0.19482636451721191, "step": 8380 }, { "epoch": 0.17162816624816532, "grad_norm": 1.5402919054031372, "learning_rate": 1.8966711901961908e-05, "loss": 0.19156895875930785, "step": 8390 }, { "epoch": 0.17183272902080915, "grad_norm": 4.714947700500488, "learning_rate": 1.8963776925337258e-05, "loss": 0.1960767388343811, "step": 8400 }, { "epoch": 0.17203729179345298, "grad_norm": 2.1750409603118896, "learning_rate": 1.8960838014078935e-05, "loss": 0.18521888256073, "step": 8410 }, { "epoch": 0.17224185456609678, "grad_norm": 6.5535569190979, "learning_rate": 1.895789516947696e-05, "loss": 0.1981738567352295, "step": 8420 }, { "epoch": 0.1724464173387406, "grad_norm": 1.2423266172409058, "learning_rate": 1.8954948392823095e-05, "loss": 0.1850841999053955, "step": 8430 }, { "epoch": 0.17265098011138444, "grad_norm": 2.7180230617523193, "learning_rate": 1.8951997685410824e-05, "loss": 0.21523137092590333, "step": 8440 }, { "epoch": 0.17285554288402824, "grad_norm": 1.3749593496322632, "learning_rate": 1.8949043048535346e-05, "loss": 0.18556530475616456, "step": 8450 }, { "epoch": 0.17306010565667207, "grad_norm": 4.508784294128418, "learning_rate": 1.89460844834936e-05, "loss": 0.18209328651428222, "step": 8460 }, { "epoch": 0.1732646684293159, "grad_norm": 2.283027172088623, "learning_rate": 1.8943121991584245e-05, "loss": 0.1953737258911133, "step": 8470 }, { "epoch": 0.1734692312019597, "grad_norm": 4.609721660614014, "learning_rate": 1.894015557410765e-05, "loss": 0.21650419235229493, "step": 8480 }, { "epoch": 0.17367379397460353, "grad_norm": 1.587167501449585, "learning_rate": 1.8937185232365936e-05, "loss": 0.19665076732635497, "step": 8490 }, { "epoch": 0.17387835674724736, "grad_norm": 2.295788049697876, "learning_rate": 1.8934210967662917e-05, "loss": 0.19242478609085084, "step": 8500 }, { "epoch": 0.17408291951989116, "grad_norm": 6.266669750213623, "learning_rate": 1.893123278130414e-05, "loss": 0.19846930503845214, "step": 8510 }, { "epoch": 0.174287482292535, "grad_norm": 2.1787381172180176, "learning_rate": 1.8928250674596883e-05, "loss": 0.1957713007926941, "step": 8520 }, { "epoch": 0.17449204506517882, "grad_norm": 1.5971405506134033, "learning_rate": 1.892526464885013e-05, "loss": 0.19293899536132814, "step": 8530 }, { "epoch": 0.17469660783782262, "grad_norm": 3.084695339202881, "learning_rate": 1.892227470537459e-05, "loss": 0.20295238494873047, "step": 8540 }, { "epoch": 0.17490117061046645, "grad_norm": 0.8799822330474854, "learning_rate": 1.8919280845482703e-05, "loss": 0.18811478614807128, "step": 8550 }, { "epoch": 0.17510573338311028, "grad_norm": 1.4151545763015747, "learning_rate": 1.8916283070488605e-05, "loss": 0.19163367748260499, "step": 8560 }, { "epoch": 0.1753102961557541, "grad_norm": 2.8617489337921143, "learning_rate": 1.8913281381708176e-05, "loss": 0.18946282863616942, "step": 8570 }, { "epoch": 0.1755148589283979, "grad_norm": 2.5072484016418457, "learning_rate": 1.8910275780458996e-05, "loss": 0.19681282043457032, "step": 8580 }, { "epoch": 0.17571942170104174, "grad_norm": 3.2715513706207275, "learning_rate": 1.890726626806037e-05, "loss": 0.18595559597015382, "step": 8590 }, { "epoch": 0.17592398447368557, "grad_norm": 1.6092523336410522, "learning_rate": 1.890425284583331e-05, "loss": 0.20074105262756348, "step": 8600 }, { "epoch": 0.17612854724632937, "grad_norm": 2.074946880340576, "learning_rate": 1.890123551510057e-05, "loss": 0.18922221660614014, "step": 8610 }, { "epoch": 0.1763331100189732, "grad_norm": 0.7999511361122131, "learning_rate": 1.8898214277186585e-05, "loss": 0.18926154375076293, "step": 8620 }, { "epoch": 0.17653767279161703, "grad_norm": 2.239790916442871, "learning_rate": 1.889518913341753e-05, "loss": 0.196597158908844, "step": 8630 }, { "epoch": 0.17674223556426083, "grad_norm": 2.502469778060913, "learning_rate": 1.8892160085121284e-05, "loss": 0.19700063467025758, "step": 8640 }, { "epoch": 0.17694679833690466, "grad_norm": 3.010788679122925, "learning_rate": 1.888912713362744e-05, "loss": 0.1999910831451416, "step": 8650 }, { "epoch": 0.1771513611095485, "grad_norm": 2.1828043460845947, "learning_rate": 1.8886090280267317e-05, "loss": 0.19313855171203614, "step": 8660 }, { "epoch": 0.1773559238821923, "grad_norm": 2.3415093421936035, "learning_rate": 1.888304952637392e-05, "loss": 0.1935926556587219, "step": 8670 }, { "epoch": 0.17756048665483612, "grad_norm": 1.4111946821212769, "learning_rate": 1.8880004873281993e-05, "loss": 0.18870381116867066, "step": 8680 }, { "epoch": 0.17776504942747995, "grad_norm": 1.39119291305542, "learning_rate": 1.8876956322327982e-05, "loss": 0.18612868785858155, "step": 8690 }, { "epoch": 0.17796961220012375, "grad_norm": 3.587512731552124, "learning_rate": 1.887390387485004e-05, "loss": 0.19368467330932618, "step": 8700 }, { "epoch": 0.17817417497276758, "grad_norm": 2.2868354320526123, "learning_rate": 1.8870847532188027e-05, "loss": 0.1996023178100586, "step": 8710 }, { "epoch": 0.1783787377454114, "grad_norm": 2.0066237449645996, "learning_rate": 1.8867787295683532e-05, "loss": 0.19883270263671876, "step": 8720 }, { "epoch": 0.1785833005180552, "grad_norm": 1.9442800283432007, "learning_rate": 1.8864723166679833e-05, "loss": 0.19443185329437257, "step": 8730 }, { "epoch": 0.17878786329069904, "grad_norm": 3.8042027950286865, "learning_rate": 1.8861655146521924e-05, "loss": 0.1850648283958435, "step": 8740 }, { "epoch": 0.17899242606334287, "grad_norm": 3.3110055923461914, "learning_rate": 1.8858583236556505e-05, "loss": 0.1927794933319092, "step": 8750 }, { "epoch": 0.17919698883598667, "grad_norm": 3.0865426063537598, "learning_rate": 1.885550743813199e-05, "loss": 0.18805651664733886, "step": 8760 }, { "epoch": 0.1794015516086305, "grad_norm": 1.4249659776687622, "learning_rate": 1.885242775259849e-05, "loss": 0.18402069807052612, "step": 8770 }, { "epoch": 0.17960611438127433, "grad_norm": 1.7101948261260986, "learning_rate": 1.8849344181307832e-05, "loss": 0.18040956258773805, "step": 8780 }, { "epoch": 0.17981067715391813, "grad_norm": 2.2418270111083984, "learning_rate": 1.8846256725613545e-05, "loss": 0.18936655521392823, "step": 8790 }, { "epoch": 0.18001523992656196, "grad_norm": 3.259767532348633, "learning_rate": 1.8843165386870857e-05, "loss": 0.18663867712020873, "step": 8800 }, { "epoch": 0.1802198026992058, "grad_norm": 1.939625859260559, "learning_rate": 1.884007016643671e-05, "loss": 0.17881470918655396, "step": 8810 }, { "epoch": 0.18042436547184962, "grad_norm": 5.014800548553467, "learning_rate": 1.883697106566974e-05, "loss": 0.18116600513458253, "step": 8820 }, { "epoch": 0.18062892824449342, "grad_norm": 4.965804576873779, "learning_rate": 1.8833868085930295e-05, "loss": 0.19197078943252563, "step": 8830 }, { "epoch": 0.18083349101713725, "grad_norm": 1.7046258449554443, "learning_rate": 1.8830761228580426e-05, "loss": 0.17146960496902466, "step": 8840 }, { "epoch": 0.18103805378978108, "grad_norm": 2.7605576515197754, "learning_rate": 1.8827650494983875e-05, "loss": 0.1935096263885498, "step": 8850 }, { "epoch": 0.18124261656242488, "grad_norm": 3.2380783557891846, "learning_rate": 1.88245358865061e-05, "loss": 0.1806708812713623, "step": 8860 }, { "epoch": 0.1814471793350687, "grad_norm": 1.3233723640441895, "learning_rate": 1.8821417404514247e-05, "loss": 0.19607110023498536, "step": 8870 }, { "epoch": 0.18165174210771254, "grad_norm": 4.230253219604492, "learning_rate": 1.8818295050377175e-05, "loss": 0.19721651077270508, "step": 8880 }, { "epoch": 0.18185630488035634, "grad_norm": 5.854376316070557, "learning_rate": 1.8815168825465427e-05, "loss": 0.19974133968353272, "step": 8890 }, { "epoch": 0.18206086765300017, "grad_norm": 2.8179593086242676, "learning_rate": 1.881203873115126e-05, "loss": 0.19623435735702516, "step": 8900 }, { "epoch": 0.182265430425644, "grad_norm": 2.764845609664917, "learning_rate": 1.8808904768808617e-05, "loss": 0.21367104053497316, "step": 8910 }, { "epoch": 0.1824699931982878, "grad_norm": 2.125593900680542, "learning_rate": 1.8805766939813154e-05, "loss": 0.20418910980224608, "step": 8920 }, { "epoch": 0.18267455597093163, "grad_norm": 1.4295629262924194, "learning_rate": 1.8802625245542207e-05, "loss": 0.19116146564483644, "step": 8930 }, { "epoch": 0.18287911874357546, "grad_norm": 4.543377876281738, "learning_rate": 1.8799479687374824e-05, "loss": 0.19346776008605956, "step": 8940 }, { "epoch": 0.18308368151621926, "grad_norm": 1.8602231740951538, "learning_rate": 1.879633026669174e-05, "loss": 0.2000744104385376, "step": 8950 }, { "epoch": 0.1832882442888631, "grad_norm": 2.136807680130005, "learning_rate": 1.8793176984875382e-05, "loss": 0.19294683933258056, "step": 8960 }, { "epoch": 0.18349280706150692, "grad_norm": 1.8407955169677734, "learning_rate": 1.8790019843309885e-05, "loss": 0.19648200273513794, "step": 8970 }, { "epoch": 0.18369736983415072, "grad_norm": 3.5111992359161377, "learning_rate": 1.8786858843381063e-05, "loss": 0.20078730583190918, "step": 8980 }, { "epoch": 0.18390193260679455, "grad_norm": 2.4160220623016357, "learning_rate": 1.8783693986476443e-05, "loss": 0.1880112648010254, "step": 8990 }, { "epoch": 0.18410649537943838, "grad_norm": 2.1852831840515137, "learning_rate": 1.8780525273985217e-05, "loss": 0.2064128637313843, "step": 9000 }, { "epoch": 0.18431105815208218, "grad_norm": 1.5893770456314087, "learning_rate": 1.87773527072983e-05, "loss": 0.189544939994812, "step": 9010 }, { "epoch": 0.184515620924726, "grad_norm": 2.1557140350341797, "learning_rate": 1.877417628780828e-05, "loss": 0.1993694543838501, "step": 9020 }, { "epoch": 0.18472018369736984, "grad_norm": 3.282424211502075, "learning_rate": 1.8770996016909433e-05, "loss": 0.19275002479553222, "step": 9030 }, { "epoch": 0.18492474647001367, "grad_norm": 2.720569610595703, "learning_rate": 1.8767811895997744e-05, "loss": 0.20602164268493653, "step": 9040 }, { "epoch": 0.18512930924265747, "grad_norm": 0.9802066683769226, "learning_rate": 1.876462392647087e-05, "loss": 0.20562334060668946, "step": 9050 }, { "epoch": 0.1853338720153013, "grad_norm": 1.4663554430007935, "learning_rate": 1.876143210972817e-05, "loss": 0.1937481164932251, "step": 9060 }, { "epoch": 0.18553843478794513, "grad_norm": 1.7657731771469116, "learning_rate": 1.8758236447170683e-05, "loss": 0.208791184425354, "step": 9070 }, { "epoch": 0.18574299756058893, "grad_norm": 2.3464550971984863, "learning_rate": 1.8755036940201138e-05, "loss": 0.18272714614868163, "step": 9080 }, { "epoch": 0.18594756033323276, "grad_norm": 4.703680992126465, "learning_rate": 1.8751833590223955e-05, "loss": 0.20458791255950928, "step": 9090 }, { "epoch": 0.1861521231058766, "grad_norm": 1.057194709777832, "learning_rate": 1.8748626398645238e-05, "loss": 0.18802168369293212, "step": 9100 }, { "epoch": 0.1863566858785204, "grad_norm": 2.0974388122558594, "learning_rate": 1.8745415366872778e-05, "loss": 0.18518414497375488, "step": 9110 }, { "epoch": 0.18656124865116422, "grad_norm": 3.7831480503082275, "learning_rate": 1.874220049631605e-05, "loss": 0.1879005789756775, "step": 9120 }, { "epoch": 0.18676581142380805, "grad_norm": 2.7066164016723633, "learning_rate": 1.8738981788386218e-05, "loss": 0.1960893154144287, "step": 9130 }, { "epoch": 0.18697037419645185, "grad_norm": 2.266883611679077, "learning_rate": 1.8735759244496127e-05, "loss": 0.19070122241973878, "step": 9140 }, { "epoch": 0.18717493696909568, "grad_norm": 1.4991413354873657, "learning_rate": 1.8732532866060307e-05, "loss": 0.19623020887374878, "step": 9150 }, { "epoch": 0.1873794997417395, "grad_norm": 1.4893726110458374, "learning_rate": 1.8729302654494975e-05, "loss": 0.2040769100189209, "step": 9160 }, { "epoch": 0.1875840625143833, "grad_norm": 3.829759120941162, "learning_rate": 1.872606861121802e-05, "loss": 0.20055136680603028, "step": 9170 }, { "epoch": 0.18778862528702714, "grad_norm": 3.8646841049194336, "learning_rate": 1.8722830737649022e-05, "loss": 0.1999640703201294, "step": 9180 }, { "epoch": 0.18799318805967097, "grad_norm": 1.6870585680007935, "learning_rate": 1.8719589035209238e-05, "loss": 0.20028867721557617, "step": 9190 }, { "epoch": 0.18819775083231477, "grad_norm": 2.6140992641448975, "learning_rate": 1.871634350532161e-05, "loss": 0.21966269016265869, "step": 9200 }, { "epoch": 0.1884023136049586, "grad_norm": 5.111854076385498, "learning_rate": 1.8713094149410756e-05, "loss": 0.1925598382949829, "step": 9210 }, { "epoch": 0.18860687637760243, "grad_norm": 2.9745490550994873, "learning_rate": 1.8709840968902978e-05, "loss": 0.1949130892753601, "step": 9220 }, { "epoch": 0.18881143915024623, "grad_norm": 3.8389699459075928, "learning_rate": 1.870658396522625e-05, "loss": 0.1970098614692688, "step": 9230 }, { "epoch": 0.18901600192289006, "grad_norm": 2.892639398574829, "learning_rate": 1.8703323139810225e-05, "loss": 0.1843941926956177, "step": 9240 }, { "epoch": 0.1892205646955339, "grad_norm": 1.8193765878677368, "learning_rate": 1.8700058494086247e-05, "loss": 0.19748635292053224, "step": 9250 }, { "epoch": 0.18942512746817772, "grad_norm": 1.266087532043457, "learning_rate": 1.8696790029487314e-05, "loss": 0.19795918464660645, "step": 9260 }, { "epoch": 0.18962969024082152, "grad_norm": 2.665250301361084, "learning_rate": 1.8693517747448124e-05, "loss": 0.1837458372116089, "step": 9270 }, { "epoch": 0.18983425301346535, "grad_norm": 2.5477795600891113, "learning_rate": 1.869024164940503e-05, "loss": 0.17776002883911132, "step": 9280 }, { "epoch": 0.19003881578610918, "grad_norm": 3.521106004714966, "learning_rate": 1.8686961736796077e-05, "loss": 0.17941243648529054, "step": 9290 }, { "epoch": 0.19024337855875298, "grad_norm": 4.769209861755371, "learning_rate": 1.8683678011060968e-05, "loss": 0.18230600357055665, "step": 9300 }, { "epoch": 0.1904479413313968, "grad_norm": 3.763676643371582, "learning_rate": 1.8680390473641096e-05, "loss": 0.19033300876617432, "step": 9310 }, { "epoch": 0.19065250410404064, "grad_norm": 1.9343332052230835, "learning_rate": 1.8677099125979517e-05, "loss": 0.18896957635879516, "step": 9320 }, { "epoch": 0.19085706687668444, "grad_norm": 1.4857178926467896, "learning_rate": 1.867380396952096e-05, "loss": 0.18542726039886476, "step": 9330 }, { "epoch": 0.19106162964932827, "grad_norm": 4.349238872528076, "learning_rate": 1.8670505005711835e-05, "loss": 0.2020730495452881, "step": 9340 }, { "epoch": 0.1912661924219721, "grad_norm": 2.838798761367798, "learning_rate": 1.8667202236000207e-05, "loss": 0.1916113018989563, "step": 9350 }, { "epoch": 0.1914707551946159, "grad_norm": 1.1979948282241821, "learning_rate": 1.8663895661835826e-05, "loss": 0.1710929036140442, "step": 9360 }, { "epoch": 0.19167531796725973, "grad_norm": 2.6972594261169434, "learning_rate": 1.8660585284670106e-05, "loss": 0.19603843688964845, "step": 9370 }, { "epoch": 0.19187988073990356, "grad_norm": 1.3451707363128662, "learning_rate": 1.865727110595613e-05, "loss": 0.18747906684875487, "step": 9380 }, { "epoch": 0.19208444351254736, "grad_norm": 3.1925861835479736, "learning_rate": 1.865395312714865e-05, "loss": 0.21373488903045654, "step": 9390 }, { "epoch": 0.1922890062851912, "grad_norm": 6.357202053070068, "learning_rate": 1.865063134970409e-05, "loss": 0.18737925291061402, "step": 9400 }, { "epoch": 0.19249356905783502, "grad_norm": 2.038133144378662, "learning_rate": 1.8647305775080538e-05, "loss": 0.19225711822509767, "step": 9410 }, { "epoch": 0.19269813183047882, "grad_norm": 0.8306653499603271, "learning_rate": 1.8643976404737744e-05, "loss": 0.1997586965560913, "step": 9420 }, { "epoch": 0.19290269460312265, "grad_norm": 1.6636625528335571, "learning_rate": 1.8640643240137137e-05, "loss": 0.1821353793144226, "step": 9430 }, { "epoch": 0.19310725737576648, "grad_norm": 5.574807643890381, "learning_rate": 1.8637306282741795e-05, "loss": 0.19833941459655763, "step": 9440 }, { "epoch": 0.19331182014841028, "grad_norm": 1.338553786277771, "learning_rate": 1.8633965534016478e-05, "loss": 0.20194606781005858, "step": 9450 }, { "epoch": 0.1935163829210541, "grad_norm": 1.9814414978027344, "learning_rate": 1.8630620995427592e-05, "loss": 0.1906484603881836, "step": 9460 }, { "epoch": 0.19372094569369794, "grad_norm": 4.35590934753418, "learning_rate": 1.8627272668443227e-05, "loss": 0.20082628726959229, "step": 9470 }, { "epoch": 0.19392550846634174, "grad_norm": 2.0415186882019043, "learning_rate": 1.862392055453312e-05, "loss": 0.19197829961776733, "step": 9480 }, { "epoch": 0.19413007123898557, "grad_norm": 2.5416884422302246, "learning_rate": 1.862056465516867e-05, "loss": 0.19137226343154906, "step": 9490 }, { "epoch": 0.1943346340116294, "grad_norm": 2.114551544189453, "learning_rate": 1.8617204971822946e-05, "loss": 0.20526883602142335, "step": 9500 }, { "epoch": 0.19453919678427323, "grad_norm": 3.047677993774414, "learning_rate": 1.8613841505970682e-05, "loss": 0.19657540321350098, "step": 9510 }, { "epoch": 0.19474375955691703, "grad_norm": 3.3535678386688232, "learning_rate": 1.861047425908826e-05, "loss": 0.18196277618408202, "step": 9520 }, { "epoch": 0.19494832232956086, "grad_norm": 4.25262975692749, "learning_rate": 1.860710323265372e-05, "loss": 0.18630975484848022, "step": 9530 }, { "epoch": 0.1951528851022047, "grad_norm": 2.977511167526245, "learning_rate": 1.8603728428146782e-05, "loss": 0.1891119122505188, "step": 9540 }, { "epoch": 0.1953574478748485, "grad_norm": 1.7275809049606323, "learning_rate": 1.8600349847048803e-05, "loss": 0.20659363269805908, "step": 9550 }, { "epoch": 0.19556201064749232, "grad_norm": 2.8382441997528076, "learning_rate": 1.85969674908428e-05, "loss": 0.1932645320892334, "step": 9560 }, { "epoch": 0.19576657342013615, "grad_norm": 2.2791314125061035, "learning_rate": 1.859358136101346e-05, "loss": 0.20137932300567626, "step": 9570 }, { "epoch": 0.19597113619277995, "grad_norm": 1.3569124937057495, "learning_rate": 1.859019145904712e-05, "loss": 0.20966196060180664, "step": 9580 }, { "epoch": 0.19617569896542378, "grad_norm": 1.036468744277954, "learning_rate": 1.8586797786431763e-05, "loss": 0.18605144023895265, "step": 9590 }, { "epoch": 0.1963802617380676, "grad_norm": 2.678760290145874, "learning_rate": 1.8583400344657045e-05, "loss": 0.18995270729064942, "step": 9600 }, { "epoch": 0.1965848245107114, "grad_norm": 2.928781032562256, "learning_rate": 1.8579999135214254e-05, "loss": 0.18340697288513183, "step": 9610 }, { "epoch": 0.19678938728335524, "grad_norm": 2.877279043197632, "learning_rate": 1.8576594159596353e-05, "loss": 0.19269182682037353, "step": 9620 }, { "epoch": 0.19699395005599907, "grad_norm": 3.8734757900238037, "learning_rate": 1.8573185419297957e-05, "loss": 0.20141401290893554, "step": 9630 }, { "epoch": 0.19719851282864287, "grad_norm": 1.7699073553085327, "learning_rate": 1.8569772915815312e-05, "loss": 0.19717307090759278, "step": 9640 }, { "epoch": 0.1974030756012867, "grad_norm": 2.2300851345062256, "learning_rate": 1.856635665064634e-05, "loss": 0.184845232963562, "step": 9650 }, { "epoch": 0.19760763837393053, "grad_norm": 1.6765954494476318, "learning_rate": 1.8562936625290597e-05, "loss": 0.18191399574279785, "step": 9660 }, { "epoch": 0.19781220114657433, "grad_norm": 2.5465383529663086, "learning_rate": 1.8559512841249305e-05, "loss": 0.1921546220779419, "step": 9670 }, { "epoch": 0.19801676391921816, "grad_norm": 2.788834810256958, "learning_rate": 1.8556085300025323e-05, "loss": 0.18241989612579346, "step": 9680 }, { "epoch": 0.198221326691862, "grad_norm": 4.487758636474609, "learning_rate": 1.855265400312317e-05, "loss": 0.20050957202911376, "step": 9690 }, { "epoch": 0.1984258894645058, "grad_norm": 5.972116470336914, "learning_rate": 1.8549218952048998e-05, "loss": 0.20719516277313232, "step": 9700 }, { "epoch": 0.19863045223714962, "grad_norm": 5.243858337402344, "learning_rate": 1.854578014831062e-05, "loss": 0.1856932282447815, "step": 9710 }, { "epoch": 0.19883501500979345, "grad_norm": 2.0731210708618164, "learning_rate": 1.8542337593417498e-05, "loss": 0.20925827026367189, "step": 9720 }, { "epoch": 0.19903957778243728, "grad_norm": 2.029282331466675, "learning_rate": 1.8538891288880727e-05, "loss": 0.20261826515197753, "step": 9730 }, { "epoch": 0.19924414055508108, "grad_norm": 1.983244776725769, "learning_rate": 1.853544123621306e-05, "loss": 0.19844751358032225, "step": 9740 }, { "epoch": 0.1994487033277249, "grad_norm": 3.181241989135742, "learning_rate": 1.8531987436928894e-05, "loss": 0.17676538228988647, "step": 9750 }, { "epoch": 0.19965326610036874, "grad_norm": 1.7804092168807983, "learning_rate": 1.8528529892544268e-05, "loss": 0.19972290992736816, "step": 9760 }, { "epoch": 0.19985782887301254, "grad_norm": 2.5562260150909424, "learning_rate": 1.852506860457686e-05, "loss": 0.19490333795547485, "step": 9770 }, { "epoch": 0.20006239164565637, "grad_norm": 1.3444923162460327, "learning_rate": 1.8521603574546e-05, "loss": 0.17786693572998047, "step": 9780 }, { "epoch": 0.2002669544183002, "grad_norm": 3.309335231781006, "learning_rate": 1.8518134803972653e-05, "loss": 0.20584390163421631, "step": 9790 }, { "epoch": 0.200471517190944, "grad_norm": 3.085564374923706, "learning_rate": 1.851466229437943e-05, "loss": 0.1901384949684143, "step": 9800 }, { "epoch": 0.20067607996358783, "grad_norm": 1.6935843229293823, "learning_rate": 1.8511186047290586e-05, "loss": 0.17772897481918334, "step": 9810 }, { "epoch": 0.20088064273623166, "grad_norm": 3.499213933944702, "learning_rate": 1.8507706064232015e-05, "loss": 0.19131844043731688, "step": 9820 }, { "epoch": 0.20108520550887546, "grad_norm": 1.8902896642684937, "learning_rate": 1.850422234673124e-05, "loss": 0.1947648286819458, "step": 9830 }, { "epoch": 0.2012897682815193, "grad_norm": 2.06543231010437, "learning_rate": 1.8500734896317448e-05, "loss": 0.20338129997253418, "step": 9840 }, { "epoch": 0.20149433105416312, "grad_norm": 2.1353397369384766, "learning_rate": 1.849724371452143e-05, "loss": 0.19417471885681153, "step": 9850 }, { "epoch": 0.20169889382680692, "grad_norm": 1.9629493951797485, "learning_rate": 1.8493748802875653e-05, "loss": 0.18792904615402223, "step": 9860 }, { "epoch": 0.20190345659945075, "grad_norm": 3.3382174968719482, "learning_rate": 1.849025016291419e-05, "loss": 0.18922295570373535, "step": 9870 }, { "epoch": 0.20210801937209458, "grad_norm": 2.481440544128418, "learning_rate": 1.848674779617276e-05, "loss": 0.18560416698455812, "step": 9880 }, { "epoch": 0.20231258214473838, "grad_norm": 3.489211320877075, "learning_rate": 1.8483241704188733e-05, "loss": 0.1980239748954773, "step": 9890 }, { "epoch": 0.2025171449173822, "grad_norm": 2.8080453872680664, "learning_rate": 1.8479731888501092e-05, "loss": 0.1895769238471985, "step": 9900 }, { "epoch": 0.20272170769002604, "grad_norm": 4.5923967361450195, "learning_rate": 1.8476218350650465e-05, "loss": 0.18495333194732666, "step": 9910 }, { "epoch": 0.20292627046266984, "grad_norm": 2.2449746131896973, "learning_rate": 1.8472701092179116e-05, "loss": 0.19248886108398439, "step": 9920 }, { "epoch": 0.20313083323531367, "grad_norm": 3.8116848468780518, "learning_rate": 1.8469180114630942e-05, "loss": 0.18859212398529052, "step": 9930 }, { "epoch": 0.2033353960079575, "grad_norm": 4.672685146331787, "learning_rate": 1.8465655419551464e-05, "loss": 0.19373561143875123, "step": 9940 }, { "epoch": 0.2035399587806013, "grad_norm": 0.9698997139930725, "learning_rate": 1.8462127008487844e-05, "loss": 0.20182740688323975, "step": 9950 }, { "epoch": 0.20374452155324513, "grad_norm": 3.723280429840088, "learning_rate": 1.8458594882988866e-05, "loss": 0.20402674674987792, "step": 9960 }, { "epoch": 0.20394908432588896, "grad_norm": 6.9297075271606445, "learning_rate": 1.845505904460496e-05, "loss": 0.19281859397888185, "step": 9970 }, { "epoch": 0.2041536470985328, "grad_norm": 5.537939548492432, "learning_rate": 1.8451519494888167e-05, "loss": 0.19748482704162598, "step": 9980 }, { "epoch": 0.2043582098711766, "grad_norm": 2.517792224884033, "learning_rate": 1.8447976235392172e-05, "loss": 0.1924793004989624, "step": 9990 }, { "epoch": 0.20456277264382042, "grad_norm": 1.8314493894577026, "learning_rate": 1.8444429267672276e-05, "loss": 0.17099299430847167, "step": 10000 }, { "epoch": 0.20476733541646425, "grad_norm": 2.865375280380249, "learning_rate": 1.844087859328542e-05, "loss": 0.18529839515686036, "step": 10010 }, { "epoch": 0.20497189818910805, "grad_norm": 1.2321321964263916, "learning_rate": 1.8437324213790163e-05, "loss": 0.19397523403167724, "step": 10020 }, { "epoch": 0.20517646096175188, "grad_norm": 2.6419661045074463, "learning_rate": 1.8433766130746694e-05, "loss": 0.20508289337158203, "step": 10030 }, { "epoch": 0.2053810237343957, "grad_norm": 2.1352412700653076, "learning_rate": 1.843020434571683e-05, "loss": 0.1883484125137329, "step": 10040 }, { "epoch": 0.2055855865070395, "grad_norm": 1.8877846002578735, "learning_rate": 1.8426638860264003e-05, "loss": 0.2024524688720703, "step": 10050 }, { "epoch": 0.20579014927968334, "grad_norm": 3.027242660522461, "learning_rate": 1.8423069675953287e-05, "loss": 0.18888435363769532, "step": 10060 }, { "epoch": 0.20599471205232717, "grad_norm": 5.64384126663208, "learning_rate": 1.8419496794351358e-05, "loss": 0.20614964962005616, "step": 10070 }, { "epoch": 0.20619927482497097, "grad_norm": 1.6474030017852783, "learning_rate": 1.841592021702653e-05, "loss": 0.19070278406143187, "step": 10080 }, { "epoch": 0.2064038375976148, "grad_norm": 3.01785945892334, "learning_rate": 1.841233994554874e-05, "loss": 0.18806781768798828, "step": 10090 }, { "epoch": 0.20660840037025863, "grad_norm": 1.4154314994812012, "learning_rate": 1.8408755981489537e-05, "loss": 0.2018526554107666, "step": 10100 }, { "epoch": 0.20681296314290243, "grad_norm": 3.2179148197174072, "learning_rate": 1.8405168326422094e-05, "loss": 0.1963186740875244, "step": 10110 }, { "epoch": 0.20701752591554626, "grad_norm": 1.6937501430511475, "learning_rate": 1.840157698192121e-05, "loss": 0.19184411764144899, "step": 10120 }, { "epoch": 0.2072220886881901, "grad_norm": 1.2564424276351929, "learning_rate": 1.8397981949563294e-05, "loss": 0.19184789657592774, "step": 10130 }, { "epoch": 0.2074266514608339, "grad_norm": 6.225679397583008, "learning_rate": 1.8394383230926384e-05, "loss": 0.2028555154800415, "step": 10140 }, { "epoch": 0.20763121423347772, "grad_norm": 2.8041205406188965, "learning_rate": 1.839078082759013e-05, "loss": 0.1907306432723999, "step": 10150 }, { "epoch": 0.20783577700612155, "grad_norm": 2.113346576690674, "learning_rate": 1.8387174741135802e-05, "loss": 0.21746540069580078, "step": 10160 }, { "epoch": 0.20804033977876535, "grad_norm": 1.8229120969772339, "learning_rate": 1.8383564973146276e-05, "loss": 0.20107529163360596, "step": 10170 }, { "epoch": 0.20824490255140918, "grad_norm": 2.643320322036743, "learning_rate": 1.8379951525206065e-05, "loss": 0.18950116634368896, "step": 10180 }, { "epoch": 0.208449465324053, "grad_norm": 5.441706657409668, "learning_rate": 1.8376334398901278e-05, "loss": 0.18928639888763427, "step": 10190 }, { "epoch": 0.20865402809669684, "grad_norm": 1.3613706827163696, "learning_rate": 1.8372713595819648e-05, "loss": 0.1889427423477173, "step": 10200 }, { "epoch": 0.20885859086934064, "grad_norm": 1.879714012145996, "learning_rate": 1.836908911755052e-05, "loss": 0.18251656293869017, "step": 10210 }, { "epoch": 0.20906315364198447, "grad_norm": 3.9316494464874268, "learning_rate": 1.8365460965684854e-05, "loss": 0.19027557373046874, "step": 10220 }, { "epoch": 0.2092677164146283, "grad_norm": 1.6127643585205078, "learning_rate": 1.8361829141815217e-05, "loss": 0.20176498889923095, "step": 10230 }, { "epoch": 0.2094722791872721, "grad_norm": 2.0217161178588867, "learning_rate": 1.8358193647535798e-05, "loss": 0.21330065727233888, "step": 10240 }, { "epoch": 0.20967684195991593, "grad_norm": 3.704240322113037, "learning_rate": 1.8354554484442384e-05, "loss": 0.18418030738830565, "step": 10250 }, { "epoch": 0.20988140473255976, "grad_norm": 1.5551867485046387, "learning_rate": 1.835091165413238e-05, "loss": 0.18865596055984496, "step": 10260 }, { "epoch": 0.21008596750520356, "grad_norm": 1.1695066690444946, "learning_rate": 1.83472651582048e-05, "loss": 0.19784400463104249, "step": 10270 }, { "epoch": 0.2102905302778474, "grad_norm": 2.397261619567871, "learning_rate": 1.834361499826027e-05, "loss": 0.19010844230651855, "step": 10280 }, { "epoch": 0.21049509305049122, "grad_norm": 1.2267801761627197, "learning_rate": 1.833996117590102e-05, "loss": 0.19613800048828126, "step": 10290 }, { "epoch": 0.21069965582313502, "grad_norm": 1.1795318126678467, "learning_rate": 1.8336303692730887e-05, "loss": 0.18159809112548828, "step": 10300 }, { "epoch": 0.21090421859577885, "grad_norm": 2.412719488143921, "learning_rate": 1.8332642550355317e-05, "loss": 0.17535216808319093, "step": 10310 }, { "epoch": 0.21110878136842268, "grad_norm": 2.0423195362091064, "learning_rate": 1.8328977750381362e-05, "loss": 0.19014967679977418, "step": 10320 }, { "epoch": 0.21131334414106648, "grad_norm": 1.7807270288467407, "learning_rate": 1.832530929441768e-05, "loss": 0.19716932773590087, "step": 10330 }, { "epoch": 0.2115179069137103, "grad_norm": 1.389997959136963, "learning_rate": 1.8321637184074532e-05, "loss": 0.182747483253479, "step": 10340 }, { "epoch": 0.21172246968635414, "grad_norm": 3.08141827583313, "learning_rate": 1.8317961420963787e-05, "loss": 0.19124071598052977, "step": 10350 }, { "epoch": 0.21192703245899794, "grad_norm": 5.043427467346191, "learning_rate": 1.831428200669891e-05, "loss": 0.18404932022094728, "step": 10360 }, { "epoch": 0.21213159523164177, "grad_norm": 1.0577071905136108, "learning_rate": 1.8310598942894974e-05, "loss": 0.18076212406158448, "step": 10370 }, { "epoch": 0.2123361580042856, "grad_norm": 2.0872371196746826, "learning_rate": 1.830691223116866e-05, "loss": 0.17147160768508912, "step": 10380 }, { "epoch": 0.2125407207769294, "grad_norm": 2.557239294052124, "learning_rate": 1.8303221873138234e-05, "loss": 0.20959024429321288, "step": 10390 }, { "epoch": 0.21274528354957323, "grad_norm": 1.0346312522888184, "learning_rate": 1.8299527870423576e-05, "loss": 0.187096107006073, "step": 10400 }, { "epoch": 0.21294984632221706, "grad_norm": 3.2105138301849365, "learning_rate": 1.8295830224646166e-05, "loss": 0.18541886806488037, "step": 10410 }, { "epoch": 0.21315440909486086, "grad_norm": 2.6790947914123535, "learning_rate": 1.8292128937429073e-05, "loss": 0.2169800281524658, "step": 10420 }, { "epoch": 0.2133589718675047, "grad_norm": 2.7473504543304443, "learning_rate": 1.8288424010396975e-05, "loss": 0.19269158840179443, "step": 10430 }, { "epoch": 0.21356353464014852, "grad_norm": 1.6983048915863037, "learning_rate": 1.8284715445176135e-05, "loss": 0.1916511058807373, "step": 10440 }, { "epoch": 0.21376809741279235, "grad_norm": 2.4066216945648193, "learning_rate": 1.828100324339443e-05, "loss": 0.20334572792053224, "step": 10450 }, { "epoch": 0.21397266018543615, "grad_norm": 2.955571413040161, "learning_rate": 1.8277287406681317e-05, "loss": 0.181571364402771, "step": 10460 }, { "epoch": 0.21417722295807998, "grad_norm": 2.461711883544922, "learning_rate": 1.8273567936667868e-05, "loss": 0.1954846978187561, "step": 10470 }, { "epoch": 0.2143817857307238, "grad_norm": 3.3258402347564697, "learning_rate": 1.8269844834986722e-05, "loss": 0.1871412754058838, "step": 10480 }, { "epoch": 0.2145863485033676, "grad_norm": 4.3028998374938965, "learning_rate": 1.8266118103272134e-05, "loss": 0.18752957582473756, "step": 10490 }, { "epoch": 0.21479091127601144, "grad_norm": 1.044303059577942, "learning_rate": 1.826238774315995e-05, "loss": 0.18222061395645142, "step": 10500 }, { "epoch": 0.21499547404865527, "grad_norm": 1.5647931098937988, "learning_rate": 1.82586537562876e-05, "loss": 0.18981654644012452, "step": 10510 }, { "epoch": 0.21520003682129907, "grad_norm": 1.9634814262390137, "learning_rate": 1.8254916144294113e-05, "loss": 0.19643584489822388, "step": 10520 }, { "epoch": 0.2154045995939429, "grad_norm": 2.6694045066833496, "learning_rate": 1.825117490882011e-05, "loss": 0.18899978399276735, "step": 10530 }, { "epoch": 0.21560916236658673, "grad_norm": 2.693565607070923, "learning_rate": 1.8247430051507792e-05, "loss": 0.1940521240234375, "step": 10540 }, { "epoch": 0.21581372513923053, "grad_norm": 2.207050085067749, "learning_rate": 1.8243681574000965e-05, "loss": 0.20494499206542968, "step": 10550 }, { "epoch": 0.21601828791187436, "grad_norm": 2.894209146499634, "learning_rate": 1.8239929477945012e-05, "loss": 0.19802141189575195, "step": 10560 }, { "epoch": 0.2162228506845182, "grad_norm": 1.6167683601379395, "learning_rate": 1.8236173764986906e-05, "loss": 0.21204845905303954, "step": 10570 }, { "epoch": 0.216427413457162, "grad_norm": 2.421713352203369, "learning_rate": 1.823241443677522e-05, "loss": 0.18270301818847656, "step": 10580 }, { "epoch": 0.21663197622980582, "grad_norm": 2.2367639541625977, "learning_rate": 1.8228651494960094e-05, "loss": 0.18503210544586182, "step": 10590 }, { "epoch": 0.21683653900244965, "grad_norm": 3.289639949798584, "learning_rate": 1.8224884941193272e-05, "loss": 0.17914299964904784, "step": 10600 }, { "epoch": 0.21704110177509345, "grad_norm": 1.4342178106307983, "learning_rate": 1.822111477712807e-05, "loss": 0.20616655349731444, "step": 10610 }, { "epoch": 0.21724566454773728, "grad_norm": 1.2095361948013306, "learning_rate": 1.82173410044194e-05, "loss": 0.19374276399612428, "step": 10620 }, { "epoch": 0.2174502273203811, "grad_norm": 3.137984275817871, "learning_rate": 1.821356362472375e-05, "loss": 0.1804553508758545, "step": 10630 }, { "epoch": 0.2176547900930249, "grad_norm": 2.8637123107910156, "learning_rate": 1.820978263969919e-05, "loss": 0.19098777770996095, "step": 10640 }, { "epoch": 0.21785935286566874, "grad_norm": 4.30389404296875, "learning_rate": 1.820599805100539e-05, "loss": 0.17470879554748536, "step": 10650 }, { "epoch": 0.21806391563831257, "grad_norm": 2.0229299068450928, "learning_rate": 1.820220986030357e-05, "loss": 0.19976465702056884, "step": 10660 }, { "epoch": 0.2182684784109564, "grad_norm": 1.082769751548767, "learning_rate": 1.8198418069256566e-05, "loss": 0.18347456455230712, "step": 10670 }, { "epoch": 0.2184730411836002, "grad_norm": 2.238208532333374, "learning_rate": 1.8194622679528767e-05, "loss": 0.19044184684753418, "step": 10680 }, { "epoch": 0.21867760395624403, "grad_norm": 3.812363624572754, "learning_rate": 1.8190823692786155e-05, "loss": 0.206382417678833, "step": 10690 }, { "epoch": 0.21888216672888786, "grad_norm": 2.7688136100769043, "learning_rate": 1.818702111069629e-05, "loss": 0.18607620000839234, "step": 10700 }, { "epoch": 0.21908672950153166, "grad_norm": 2.4708986282348633, "learning_rate": 1.8183214934928307e-05, "loss": 0.19450714588165283, "step": 10710 }, { "epoch": 0.2192912922741755, "grad_norm": 1.5513678789138794, "learning_rate": 1.8179405167152927e-05, "loss": 0.19346189498901367, "step": 10720 }, { "epoch": 0.21949585504681932, "grad_norm": 2.383089303970337, "learning_rate": 1.817559180904243e-05, "loss": 0.1877532958984375, "step": 10730 }, { "epoch": 0.21970041781946312, "grad_norm": 2.951349973678589, "learning_rate": 1.8171774862270687e-05, "loss": 0.17482550144195558, "step": 10740 }, { "epoch": 0.21990498059210695, "grad_norm": 3.718825340270996, "learning_rate": 1.816795432851314e-05, "loss": 0.18111784458160402, "step": 10750 }, { "epoch": 0.22010954336475078, "grad_norm": 2.22825288772583, "learning_rate": 1.8164130209446806e-05, "loss": 0.18638978004455567, "step": 10760 }, { "epoch": 0.22031410613739458, "grad_norm": 1.9429043531417847, "learning_rate": 1.8160302506750276e-05, "loss": 0.18389655351638795, "step": 10770 }, { "epoch": 0.2205186689100384, "grad_norm": 1.6354457139968872, "learning_rate": 1.815647122210371e-05, "loss": 0.18331332206726075, "step": 10780 }, { "epoch": 0.22072323168268224, "grad_norm": 1.8793126344680786, "learning_rate": 1.8152636357188845e-05, "loss": 0.18056156635284423, "step": 10790 }, { "epoch": 0.22092779445532604, "grad_norm": 2.6486408710479736, "learning_rate": 1.814879791368899e-05, "loss": 0.1834555149078369, "step": 10800 }, { "epoch": 0.22113235722796987, "grad_norm": 3.7628026008605957, "learning_rate": 1.8144955893289017e-05, "loss": 0.18602330684661866, "step": 10810 }, { "epoch": 0.2213369200006137, "grad_norm": 3.181959629058838, "learning_rate": 1.8141110297675383e-05, "loss": 0.18018430471420288, "step": 10820 }, { "epoch": 0.2215414827732575, "grad_norm": 3.5983657836914062, "learning_rate": 1.8137261128536098e-05, "loss": 0.21112947463989257, "step": 10830 }, { "epoch": 0.22174604554590133, "grad_norm": 1.8087515830993652, "learning_rate": 1.8133408387560752e-05, "loss": 0.19344708919525147, "step": 10840 }, { "epoch": 0.22195060831854516, "grad_norm": 3.1362550258636475, "learning_rate": 1.8129552076440494e-05, "loss": 0.18813393115997315, "step": 10850 }, { "epoch": 0.22215517109118896, "grad_norm": 1.2146503925323486, "learning_rate": 1.812569219686805e-05, "loss": 0.18283343315124512, "step": 10860 }, { "epoch": 0.2223597338638328, "grad_norm": 1.8582091331481934, "learning_rate": 1.8121828750537703e-05, "loss": 0.1809253692626953, "step": 10870 }, { "epoch": 0.22256429663647662, "grad_norm": 2.3379318714141846, "learning_rate": 1.8117961739145312e-05, "loss": 0.20963082313537598, "step": 10880 }, { "epoch": 0.22276885940912042, "grad_norm": 3.7752363681793213, "learning_rate": 1.8114091164388284e-05, "loss": 0.17845230102539061, "step": 10890 }, { "epoch": 0.22297342218176425, "grad_norm": 3.485220193862915, "learning_rate": 1.811021702796561e-05, "loss": 0.1948938250541687, "step": 10900 }, { "epoch": 0.22317798495440808, "grad_norm": 3.598393201828003, "learning_rate": 1.8106339331577833e-05, "loss": 0.2044370174407959, "step": 10910 }, { "epoch": 0.2233825477270519, "grad_norm": 3.4304633140563965, "learning_rate": 1.8102458076927058e-05, "loss": 0.17962379455566407, "step": 10920 }, { "epoch": 0.2235871104996957, "grad_norm": 1.5019276142120361, "learning_rate": 1.8098573265716956e-05, "loss": 0.1836946725845337, "step": 10930 }, { "epoch": 0.22379167327233954, "grad_norm": 1.433642029762268, "learning_rate": 1.8094684899652763e-05, "loss": 0.1750802993774414, "step": 10940 }, { "epoch": 0.22399623604498337, "grad_norm": 1.7453207969665527, "learning_rate": 1.809079298044126e-05, "loss": 0.1916882038116455, "step": 10950 }, { "epoch": 0.22420079881762717, "grad_norm": 2.672844409942627, "learning_rate": 1.808689750979081e-05, "loss": 0.19008498191833495, "step": 10960 }, { "epoch": 0.224405361590271, "grad_norm": 2.4677035808563232, "learning_rate": 1.808299848941131e-05, "loss": 0.18059310913085938, "step": 10970 }, { "epoch": 0.22460992436291483, "grad_norm": 1.1246260404586792, "learning_rate": 1.807909592101424e-05, "loss": 0.21183929443359376, "step": 10980 }, { "epoch": 0.22481448713555863, "grad_norm": 2.0646860599517822, "learning_rate": 1.807518980631262e-05, "loss": 0.17664663791656493, "step": 10990 }, { "epoch": 0.22501904990820246, "grad_norm": 5.032137393951416, "learning_rate": 1.807128014702103e-05, "loss": 0.1962032437324524, "step": 11000 }, { "epoch": 0.22522361268084629, "grad_norm": 3.5176942348480225, "learning_rate": 1.8067366944855606e-05, "loss": 0.19665056467056274, "step": 11010 }, { "epoch": 0.2254281754534901, "grad_norm": 1.4115886688232422, "learning_rate": 1.806345020153405e-05, "loss": 0.17624703645706177, "step": 11020 }, { "epoch": 0.22563273822613392, "grad_norm": 1.226189136505127, "learning_rate": 1.8059529918775603e-05, "loss": 0.17702031135559082, "step": 11030 }, { "epoch": 0.22583730099877775, "grad_norm": 2.383345365524292, "learning_rate": 1.8055606098301064e-05, "loss": 0.19088425636291503, "step": 11040 }, { "epoch": 0.22604186377142155, "grad_norm": 1.9393593072891235, "learning_rate": 1.8051678741832792e-05, "loss": 0.19584392309188842, "step": 11050 }, { "epoch": 0.22624642654406538, "grad_norm": 7.322976112365723, "learning_rate": 1.8047747851094692e-05, "loss": 0.19018410444259642, "step": 11060 }, { "epoch": 0.2264509893167092, "grad_norm": 3.8167600631713867, "learning_rate": 1.804381342781222e-05, "loss": 0.18421504497528077, "step": 11070 }, { "epoch": 0.226655552089353, "grad_norm": 2.23028564453125, "learning_rate": 1.8039875473712386e-05, "loss": 0.17822396755218506, "step": 11080 }, { "epoch": 0.22686011486199684, "grad_norm": 3.960665225982666, "learning_rate": 1.8035933990523742e-05, "loss": 0.19422502517700196, "step": 11090 }, { "epoch": 0.22706467763464067, "grad_norm": 4.505288600921631, "learning_rate": 1.80319889799764e-05, "loss": 0.1892271876335144, "step": 11100 }, { "epoch": 0.22726924040728447, "grad_norm": 4.338014125823975, "learning_rate": 1.802804044380202e-05, "loss": 0.20216012001037598, "step": 11110 }, { "epoch": 0.2274738031799283, "grad_norm": 0.9395129084587097, "learning_rate": 1.8024088383733795e-05, "loss": 0.19167696237564086, "step": 11120 }, { "epoch": 0.22767836595257213, "grad_norm": 5.211683750152588, "learning_rate": 1.802013280150648e-05, "loss": 0.1980413556098938, "step": 11130 }, { "epoch": 0.22788292872521596, "grad_norm": 4.494865894317627, "learning_rate": 1.8016173698856366e-05, "loss": 0.1716765880584717, "step": 11140 }, { "epoch": 0.22808749149785976, "grad_norm": 3.2410717010498047, "learning_rate": 1.8012211077521303e-05, "loss": 0.18467459678649903, "step": 11150 }, { "epoch": 0.22829205427050359, "grad_norm": 2.1182003021240234, "learning_rate": 1.8008244939240668e-05, "loss": 0.18205387592315675, "step": 11160 }, { "epoch": 0.22849661704314742, "grad_norm": 1.3041967153549194, "learning_rate": 1.8004275285755392e-05, "loss": 0.17290464639663697, "step": 11170 }, { "epoch": 0.22870117981579122, "grad_norm": 4.247189521789551, "learning_rate": 1.800030211880795e-05, "loss": 0.16670161485671997, "step": 11180 }, { "epoch": 0.22890574258843505, "grad_norm": 2.667555332183838, "learning_rate": 1.7996325440142356e-05, "loss": 0.17727088928222656, "step": 11190 }, { "epoch": 0.22911030536107888, "grad_norm": 1.1562058925628662, "learning_rate": 1.7992345251504164e-05, "loss": 0.1988302707672119, "step": 11200 }, { "epoch": 0.22931486813372268, "grad_norm": 2.3273661136627197, "learning_rate": 1.7988361554640473e-05, "loss": 0.1945033550262451, "step": 11210 }, { "epoch": 0.2295194309063665, "grad_norm": 4.216678619384766, "learning_rate": 1.7984374351299917e-05, "loss": 0.205853271484375, "step": 11220 }, { "epoch": 0.22972399367901034, "grad_norm": 2.908479928970337, "learning_rate": 1.7980383643232668e-05, "loss": 0.1845534324645996, "step": 11230 }, { "epoch": 0.22992855645165414, "grad_norm": 3.645430564880371, "learning_rate": 1.797638943219045e-05, "loss": 0.19838953018188477, "step": 11240 }, { "epoch": 0.23013311922429797, "grad_norm": 1.0307420492172241, "learning_rate": 1.797239171992651e-05, "loss": 0.18479799032211303, "step": 11250 }, { "epoch": 0.2303376819969418, "grad_norm": 2.010467767715454, "learning_rate": 1.7968390508195633e-05, "loss": 0.18709821701049806, "step": 11260 }, { "epoch": 0.2305422447695856, "grad_norm": 0.9815008640289307, "learning_rate": 1.7964385798754145e-05, "loss": 0.18827234506607055, "step": 11270 }, { "epoch": 0.23074680754222943, "grad_norm": 3.7047300338745117, "learning_rate": 1.796037759335991e-05, "loss": 0.19489774703979493, "step": 11280 }, { "epoch": 0.23095137031487326, "grad_norm": 3.2681829929351807, "learning_rate": 1.795636589377232e-05, "loss": 0.18300037384033202, "step": 11290 }, { "epoch": 0.23115593308751706, "grad_norm": 1.2012252807617188, "learning_rate": 1.7952350701752307e-05, "loss": 0.18594007492065429, "step": 11300 }, { "epoch": 0.2313604958601609, "grad_norm": 2.21368408203125, "learning_rate": 1.7948332019062324e-05, "loss": 0.1826460599899292, "step": 11310 }, { "epoch": 0.23156505863280472, "grad_norm": 2.7402501106262207, "learning_rate": 1.794430984746637e-05, "loss": 0.19798033237457274, "step": 11320 }, { "epoch": 0.23176962140544852, "grad_norm": 2.4484245777130127, "learning_rate": 1.794028418872997e-05, "loss": 0.18695602416992188, "step": 11330 }, { "epoch": 0.23197418417809235, "grad_norm": 1.6962209939956665, "learning_rate": 1.7936255044620178e-05, "loss": 0.1817249059677124, "step": 11340 }, { "epoch": 0.23217874695073618, "grad_norm": 2.2422525882720947, "learning_rate": 1.793222241690558e-05, "loss": 0.19061558246612548, "step": 11350 }, { "epoch": 0.23238330972337998, "grad_norm": 2.119173526763916, "learning_rate": 1.7928186307356294e-05, "loss": 0.17430962324142457, "step": 11360 }, { "epoch": 0.2325878724960238, "grad_norm": 1.4819562435150146, "learning_rate": 1.792414671774396e-05, "loss": 0.19523978233337402, "step": 11370 }, { "epoch": 0.23279243526866764, "grad_norm": 3.6659886837005615, "learning_rate": 1.7920103649841746e-05, "loss": 0.2006072759628296, "step": 11380 }, { "epoch": 0.23299699804131146, "grad_norm": 2.410548210144043, "learning_rate": 1.7916057105424348e-05, "loss": 0.1825239658355713, "step": 11390 }, { "epoch": 0.23320156081395527, "grad_norm": 2.917274236679077, "learning_rate": 1.7912007086267998e-05, "loss": 0.18718457221984863, "step": 11400 }, { "epoch": 0.2334061235865991, "grad_norm": 2.5113744735717773, "learning_rate": 1.790795359415044e-05, "loss": 0.17611750364303588, "step": 11410 }, { "epoch": 0.23361068635924293, "grad_norm": 1.3754655122756958, "learning_rate": 1.7903896630850948e-05, "loss": 0.19267162084579467, "step": 11420 }, { "epoch": 0.23381524913188673, "grad_norm": 2.845876693725586, "learning_rate": 1.7899836198150317e-05, "loss": 0.18578054904937744, "step": 11430 }, { "epoch": 0.23401981190453056, "grad_norm": 2.1456196308135986, "learning_rate": 1.7895772297830864e-05, "loss": 0.18046659231185913, "step": 11440 }, { "epoch": 0.23422437467717439, "grad_norm": 1.7781281471252441, "learning_rate": 1.7891704931676436e-05, "loss": 0.16980259418487548, "step": 11450 }, { "epoch": 0.2344289374498182, "grad_norm": 3.127838134765625, "learning_rate": 1.7887634101472393e-05, "loss": 0.20432772636413574, "step": 11460 }, { "epoch": 0.23463350022246202, "grad_norm": 2.791696548461914, "learning_rate": 1.788355980900562e-05, "loss": 0.18309980630874634, "step": 11470 }, { "epoch": 0.23483806299510585, "grad_norm": 2.5616068840026855, "learning_rate": 1.7879482056064523e-05, "loss": 0.18176686763763428, "step": 11480 }, { "epoch": 0.23504262576774965, "grad_norm": 4.770889759063721, "learning_rate": 1.7875400844439016e-05, "loss": 0.19521422386169435, "step": 11490 }, { "epoch": 0.23524718854039348, "grad_norm": 2.5445210933685303, "learning_rate": 1.7871316175920547e-05, "loss": 0.19727714061737062, "step": 11500 }, { "epoch": 0.2354517513130373, "grad_norm": 2.2193405628204346, "learning_rate": 1.786722805230207e-05, "loss": 0.18066587448120117, "step": 11510 }, { "epoch": 0.2356563140856811, "grad_norm": 2.2315101623535156, "learning_rate": 1.7863136475378062e-05, "loss": 0.17498064041137695, "step": 11520 }, { "epoch": 0.23586087685832494, "grad_norm": 1.9320162534713745, "learning_rate": 1.7859041446944512e-05, "loss": 0.1889073967933655, "step": 11530 }, { "epoch": 0.23606543963096877, "grad_norm": 7.519484043121338, "learning_rate": 1.7854942968798923e-05, "loss": 0.19370874166488647, "step": 11540 }, { "epoch": 0.23627000240361257, "grad_norm": 1.6337300539016724, "learning_rate": 1.7850841042740318e-05, "loss": 0.17867354154586793, "step": 11550 }, { "epoch": 0.2364745651762564, "grad_norm": 3.772616386413574, "learning_rate": 1.784673567056923e-05, "loss": 0.169921875, "step": 11560 }, { "epoch": 0.23667912794890023, "grad_norm": 2.2315611839294434, "learning_rate": 1.7842626854087704e-05, "loss": 0.21243104934692383, "step": 11570 }, { "epoch": 0.23688369072154403, "grad_norm": 1.7009682655334473, "learning_rate": 1.7838514595099298e-05, "loss": 0.19846091270446778, "step": 11580 }, { "epoch": 0.23708825349418786, "grad_norm": 4.897308349609375, "learning_rate": 1.7834398895409077e-05, "loss": 0.18650920391082765, "step": 11590 }, { "epoch": 0.23729281626683169, "grad_norm": 0.8987330794334412, "learning_rate": 1.7830279756823627e-05, "loss": 0.19174652099609374, "step": 11600 }, { "epoch": 0.23749737903947551, "grad_norm": 1.577282190322876, "learning_rate": 1.7826157181151028e-05, "loss": 0.1904728889465332, "step": 11610 }, { "epoch": 0.23770194181211932, "grad_norm": 1.266573190689087, "learning_rate": 1.7822031170200884e-05, "loss": 0.19980225563049317, "step": 11620 }, { "epoch": 0.23790650458476315, "grad_norm": 2.3325326442718506, "learning_rate": 1.7817901725784298e-05, "loss": 0.1822401762008667, "step": 11630 }, { "epoch": 0.23811106735740697, "grad_norm": 1.561966896057129, "learning_rate": 1.7813768849713882e-05, "loss": 0.18788591623306275, "step": 11640 }, { "epoch": 0.23831563013005078, "grad_norm": 5.440273284912109, "learning_rate": 1.7809632543803752e-05, "loss": 0.18511335849761962, "step": 11650 }, { "epoch": 0.2385201929026946, "grad_norm": 2.167936086654663, "learning_rate": 1.7805492809869535e-05, "loss": 0.18840633630752562, "step": 11660 }, { "epoch": 0.23872475567533843, "grad_norm": 2.3930604457855225, "learning_rate": 1.7801349649728358e-05, "loss": 0.19567159414291382, "step": 11670 }, { "epoch": 0.23892931844798224, "grad_norm": 3.3971478939056396, "learning_rate": 1.7797203065198858e-05, "loss": 0.18265390396118164, "step": 11680 }, { "epoch": 0.23913388122062607, "grad_norm": 5.422078609466553, "learning_rate": 1.7793053058101163e-05, "loss": 0.19961915016174317, "step": 11690 }, { "epoch": 0.2393384439932699, "grad_norm": 2.6623728275299072, "learning_rate": 1.778889963025692e-05, "loss": 0.18590893745422363, "step": 11700 }, { "epoch": 0.2395430067659137, "grad_norm": 5.354931831359863, "learning_rate": 1.7784742783489258e-05, "loss": 0.19175437688827515, "step": 11710 }, { "epoch": 0.23974756953855753, "grad_norm": 1.3926974534988403, "learning_rate": 1.778058251962283e-05, "loss": 0.1899067521095276, "step": 11720 }, { "epoch": 0.23995213231120135, "grad_norm": 1.6321500539779663, "learning_rate": 1.7776418840483764e-05, "loss": 0.19214868545532227, "step": 11730 }, { "epoch": 0.24015669508384516, "grad_norm": 2.1378040313720703, "learning_rate": 1.7772251747899713e-05, "loss": 0.17702922821044922, "step": 11740 }, { "epoch": 0.24036125785648899, "grad_norm": 1.9284634590148926, "learning_rate": 1.77680812436998e-05, "loss": 0.18223285675048828, "step": 11750 }, { "epoch": 0.24056582062913281, "grad_norm": 2.037046194076538, "learning_rate": 1.776390732971467e-05, "loss": 0.17731103897094727, "step": 11760 }, { "epoch": 0.24077038340177662, "grad_norm": 2.154947519302368, "learning_rate": 1.775973000777645e-05, "loss": 0.17967212200164795, "step": 11770 }, { "epoch": 0.24097494617442045, "grad_norm": 3.8639886379241943, "learning_rate": 1.775554927971877e-05, "loss": 0.18176681995391847, "step": 11780 }, { "epoch": 0.24117950894706428, "grad_norm": 1.6827601194381714, "learning_rate": 1.7751365147376755e-05, "loss": 0.18827242851257325, "step": 11790 }, { "epoch": 0.24138407171970808, "grad_norm": 2.7476601600646973, "learning_rate": 1.774717761258702e-05, "loss": 0.18480929136276245, "step": 11800 }, { "epoch": 0.2415886344923519, "grad_norm": 1.8989062309265137, "learning_rate": 1.7742986677187677e-05, "loss": 0.19193825721740723, "step": 11810 }, { "epoch": 0.24179319726499574, "grad_norm": 2.0697872638702393, "learning_rate": 1.7738792343018327e-05, "loss": 0.1893148183822632, "step": 11820 }, { "epoch": 0.24199776003763954, "grad_norm": 3.7236995697021484, "learning_rate": 1.773459461192007e-05, "loss": 0.18756095170974732, "step": 11830 }, { "epoch": 0.24220232281028337, "grad_norm": 4.436959743499756, "learning_rate": 1.773039348573549e-05, "loss": 0.18864028453826903, "step": 11840 }, { "epoch": 0.2424068855829272, "grad_norm": 1.4180974960327148, "learning_rate": 1.772618896630866e-05, "loss": 0.17821481227874755, "step": 11850 }, { "epoch": 0.24261144835557102, "grad_norm": 2.0774173736572266, "learning_rate": 1.7721981055485153e-05, "loss": 0.19402639865875243, "step": 11860 }, { "epoch": 0.24281601112821483, "grad_norm": 1.537695288658142, "learning_rate": 1.7717769755112017e-05, "loss": 0.1763428807258606, "step": 11870 }, { "epoch": 0.24302057390085866, "grad_norm": 1.2795071601867676, "learning_rate": 1.77135550670378e-05, "loss": 0.2021373987197876, "step": 11880 }, { "epoch": 0.24322513667350248, "grad_norm": 2.287200927734375, "learning_rate": 1.770933699311253e-05, "loss": 0.18632931709289552, "step": 11890 }, { "epoch": 0.24342969944614629, "grad_norm": 2.1794190406799316, "learning_rate": 1.7705115535187724e-05, "loss": 0.17749031782150268, "step": 11900 }, { "epoch": 0.24363426221879012, "grad_norm": 1.7447954416275024, "learning_rate": 1.770089069511638e-05, "loss": 0.19834479093551635, "step": 11910 }, { "epoch": 0.24383882499143394, "grad_norm": 4.572577476501465, "learning_rate": 1.7696662474752987e-05, "loss": 0.19863346815109253, "step": 11920 }, { "epoch": 0.24404338776407775, "grad_norm": 2.1133556365966797, "learning_rate": 1.7692430875953514e-05, "loss": 0.190339994430542, "step": 11930 }, { "epoch": 0.24424795053672158, "grad_norm": 3.0801749229431152, "learning_rate": 1.7688195900575412e-05, "loss": 0.17928047180175782, "step": 11940 }, { "epoch": 0.2444525133093654, "grad_norm": 2.7844159603118896, "learning_rate": 1.768395755047762e-05, "loss": 0.17366033792495728, "step": 11950 }, { "epoch": 0.2446570760820092, "grad_norm": 1.7852895259857178, "learning_rate": 1.7679715827520548e-05, "loss": 0.1918109178543091, "step": 11960 }, { "epoch": 0.24486163885465304, "grad_norm": 2.2925167083740234, "learning_rate": 1.76754707335661e-05, "loss": 0.19511737823486328, "step": 11970 }, { "epoch": 0.24506620162729686, "grad_norm": 2.432133197784424, "learning_rate": 1.7671222270477645e-05, "loss": 0.184904146194458, "step": 11980 }, { "epoch": 0.24527076439994067, "grad_norm": 1.9327373504638672, "learning_rate": 1.766697044012004e-05, "loss": 0.17506673336029052, "step": 11990 }, { "epoch": 0.2454753271725845, "grad_norm": 3.7196784019470215, "learning_rate": 1.7662715244359624e-05, "loss": 0.19474217891693116, "step": 12000 }, { "epoch": 0.24567988994522832, "grad_norm": 1.115447759628296, "learning_rate": 1.76584566850642e-05, "loss": 0.18891139030456544, "step": 12010 }, { "epoch": 0.24588445271787213, "grad_norm": 1.9484039545059204, "learning_rate": 1.765419476410306e-05, "loss": 0.19154973030090333, "step": 12020 }, { "epoch": 0.24608901549051596, "grad_norm": 3.1960954666137695, "learning_rate": 1.7649929483346962e-05, "loss": 0.17075666189193725, "step": 12030 }, { "epoch": 0.24629357826315978, "grad_norm": 4.2917680740356445, "learning_rate": 1.764566084466815e-05, "loss": 0.18596622943878174, "step": 12040 }, { "epoch": 0.2464981410358036, "grad_norm": 1.8461551666259766, "learning_rate": 1.7641388849940328e-05, "loss": 0.19257954359054566, "step": 12050 }, { "epoch": 0.24670270380844742, "grad_norm": 2.1019883155822754, "learning_rate": 1.7637113501038684e-05, "loss": 0.18602352142333983, "step": 12060 }, { "epoch": 0.24690726658109124, "grad_norm": 1.8283239603042603, "learning_rate": 1.7632834799839874e-05, "loss": 0.18201298713684083, "step": 12070 }, { "epoch": 0.24711182935373507, "grad_norm": 2.913616895675659, "learning_rate": 1.762855274822203e-05, "loss": 0.1625762939453125, "step": 12080 }, { "epoch": 0.24731639212637888, "grad_norm": 1.9201730489730835, "learning_rate": 1.762426734806474e-05, "loss": 0.1849072217941284, "step": 12090 }, { "epoch": 0.2475209548990227, "grad_norm": 3.1741886138916016, "learning_rate": 1.7619978601249082e-05, "loss": 0.19790363311767578, "step": 12100 }, { "epoch": 0.24772551767166653, "grad_norm": 2.731626272201538, "learning_rate": 1.7615686509657596e-05, "loss": 0.18829834461212158, "step": 12110 }, { "epoch": 0.24793008044431034, "grad_norm": 5.0207672119140625, "learning_rate": 1.761139107517428e-05, "loss": 0.18401345014572143, "step": 12120 }, { "epoch": 0.24813464321695416, "grad_norm": 4.095311164855957, "learning_rate": 1.760709229968461e-05, "loss": 0.19772374629974365, "step": 12130 }, { "epoch": 0.248339205989598, "grad_norm": 5.392574787139893, "learning_rate": 1.7602790185075526e-05, "loss": 0.18505679368972777, "step": 12140 }, { "epoch": 0.2485437687622418, "grad_norm": 1.8108729124069214, "learning_rate": 1.7598484733235433e-05, "loss": 0.19113492965698242, "step": 12150 }, { "epoch": 0.24874833153488562, "grad_norm": 1.5840141773223877, "learning_rate": 1.7594175946054206e-05, "loss": 0.17510490417480468, "step": 12160 }, { "epoch": 0.24895289430752945, "grad_norm": 2.0661988258361816, "learning_rate": 1.7589863825423173e-05, "loss": 0.19465802907943724, "step": 12170 }, { "epoch": 0.24915745708017326, "grad_norm": 1.4229148626327515, "learning_rate": 1.7585548373235132e-05, "loss": 0.1799420475959778, "step": 12180 }, { "epoch": 0.24936201985281709, "grad_norm": 0.9316917657852173, "learning_rate": 1.758122959138435e-05, "loss": 0.19801807403564453, "step": 12190 }, { "epoch": 0.24956658262546091, "grad_norm": 7.101391315460205, "learning_rate": 1.7576907481766537e-05, "loss": 0.19760508537292482, "step": 12200 }, { "epoch": 0.24977114539810472, "grad_norm": 2.958735466003418, "learning_rate": 1.7572582046278884e-05, "loss": 0.20355758666992188, "step": 12210 }, { "epoch": 0.24997570817074855, "grad_norm": 2.3886830806732178, "learning_rate": 1.756825328682003e-05, "loss": 0.19146143198013305, "step": 12220 }, { "epoch": 0.25018027094339235, "grad_norm": 1.7569304704666138, "learning_rate": 1.7563921205290077e-05, "loss": 0.18167494535446166, "step": 12230 }, { "epoch": 0.2503848337160362, "grad_norm": 1.0563368797302246, "learning_rate": 1.7559585803590585e-05, "loss": 0.18557435274124146, "step": 12240 }, { "epoch": 0.25058939648868, "grad_norm": 2.9053266048431396, "learning_rate": 1.7555247083624567e-05, "loss": 0.18181605339050294, "step": 12250 }, { "epoch": 0.2507939592613238, "grad_norm": 2.8300137519836426, "learning_rate": 1.7550905047296502e-05, "loss": 0.18966150283813477, "step": 12260 }, { "epoch": 0.25099852203396766, "grad_norm": 1.8838859796524048, "learning_rate": 1.7546559696512316e-05, "loss": 0.18904638290405273, "step": 12270 }, { "epoch": 0.25120308480661147, "grad_norm": 2.330176591873169, "learning_rate": 1.754221103317939e-05, "loss": 0.2008307933807373, "step": 12280 }, { "epoch": 0.25140764757925527, "grad_norm": 1.1679787635803223, "learning_rate": 1.7537859059206568e-05, "loss": 0.18064969778060913, "step": 12290 }, { "epoch": 0.2516122103518991, "grad_norm": 1.1558595895767212, "learning_rate": 1.7533503776504135e-05, "loss": 0.1808405637741089, "step": 12300 }, { "epoch": 0.2518167731245429, "grad_norm": 1.754555106163025, "learning_rate": 1.752914518698384e-05, "loss": 0.17375390529632567, "step": 12310 }, { "epoch": 0.2520213358971867, "grad_norm": 3.1663601398468018, "learning_rate": 1.7524783292558877e-05, "loss": 0.21277241706848143, "step": 12320 }, { "epoch": 0.2522258986698306, "grad_norm": 2.4418296813964844, "learning_rate": 1.752041809514389e-05, "loss": 0.1887232780456543, "step": 12330 }, { "epoch": 0.2524304614424744, "grad_norm": 2.7133185863494873, "learning_rate": 1.751604959665497e-05, "loss": 0.17567453384399415, "step": 12340 }, { "epoch": 0.2526350242151182, "grad_norm": 3.7925283908843994, "learning_rate": 1.7511677799009666e-05, "loss": 0.17696197032928468, "step": 12350 }, { "epoch": 0.25283958698776204, "grad_norm": 2.0280678272247314, "learning_rate": 1.7507302704126972e-05, "loss": 0.1885829210281372, "step": 12360 }, { "epoch": 0.25304414976040585, "grad_norm": 1.6530683040618896, "learning_rate": 1.7502924313927328e-05, "loss": 0.18273024559020995, "step": 12370 }, { "epoch": 0.25324871253304965, "grad_norm": 1.0529818534851074, "learning_rate": 1.749854263033262e-05, "loss": 0.17460720539093016, "step": 12380 }, { "epoch": 0.2534532753056935, "grad_norm": 3.0367441177368164, "learning_rate": 1.7494157655266177e-05, "loss": 0.18605210781097412, "step": 12390 }, { "epoch": 0.2536578380783373, "grad_norm": 0.7505576014518738, "learning_rate": 1.7489769390652783e-05, "loss": 0.1933127760887146, "step": 12400 }, { "epoch": 0.2538624008509811, "grad_norm": 2.280388116836548, "learning_rate": 1.748537783841865e-05, "loss": 0.1819485306739807, "step": 12410 }, { "epoch": 0.25406696362362496, "grad_norm": 2.3147661685943604, "learning_rate": 1.7480983000491446e-05, "loss": 0.19717226028442383, "step": 12420 }, { "epoch": 0.25427152639626877, "grad_norm": 2.296907901763916, "learning_rate": 1.747658487880028e-05, "loss": 0.19565274715423583, "step": 12430 }, { "epoch": 0.2544760891689126, "grad_norm": 2.482694149017334, "learning_rate": 1.747218347527569e-05, "loss": 0.190592360496521, "step": 12440 }, { "epoch": 0.2546806519415564, "grad_norm": 1.1888107061386108, "learning_rate": 1.7467778791849678e-05, "loss": 0.16547249555587767, "step": 12450 }, { "epoch": 0.2548852147142002, "grad_norm": 1.4413257837295532, "learning_rate": 1.7463370830455663e-05, "loss": 0.18706204891204833, "step": 12460 }, { "epoch": 0.2550897774868441, "grad_norm": 3.9549598693847656, "learning_rate": 1.745895959302851e-05, "loss": 0.188092839717865, "step": 12470 }, { "epoch": 0.2552943402594879, "grad_norm": 1.6081697940826416, "learning_rate": 1.745454508150453e-05, "loss": 0.1832775354385376, "step": 12480 }, { "epoch": 0.2554989030321317, "grad_norm": 1.8182296752929688, "learning_rate": 1.745012729782146e-05, "loss": 0.17186684608459474, "step": 12490 }, { "epoch": 0.25570346580477554, "grad_norm": 1.935123324394226, "learning_rate": 1.744570624391848e-05, "loss": 0.18740264177322388, "step": 12500 }, { "epoch": 0.25590802857741934, "grad_norm": 1.099234700202942, "learning_rate": 1.7441281921736205e-05, "loss": 0.18795287609100342, "step": 12510 }, { "epoch": 0.25611259135006315, "grad_norm": 1.7251936197280884, "learning_rate": 1.7436854333216683e-05, "loss": 0.1841917634010315, "step": 12520 }, { "epoch": 0.256317154122707, "grad_norm": 1.8655884265899658, "learning_rate": 1.7432423480303394e-05, "loss": 0.18297221660614013, "step": 12530 }, { "epoch": 0.2565217168953508, "grad_norm": 2.7421069145202637, "learning_rate": 1.742798936494125e-05, "loss": 0.19577317237854003, "step": 12540 }, { "epoch": 0.2567262796679946, "grad_norm": 1.902388095855713, "learning_rate": 1.7423551989076605e-05, "loss": 0.17884200811386108, "step": 12550 }, { "epoch": 0.25693084244063846, "grad_norm": 3.7331912517547607, "learning_rate": 1.7419111354657236e-05, "loss": 0.19040045738220215, "step": 12560 }, { "epoch": 0.25713540521328226, "grad_norm": 3.1251869201660156, "learning_rate": 1.7414667463632345e-05, "loss": 0.1623647093772888, "step": 12570 }, { "epoch": 0.25733996798592607, "grad_norm": 3.4886481761932373, "learning_rate": 1.7410220317952576e-05, "loss": 0.1893907904624939, "step": 12580 }, { "epoch": 0.2575445307585699, "grad_norm": 3.914154052734375, "learning_rate": 1.7405769919569993e-05, "loss": 0.18360505104064942, "step": 12590 }, { "epoch": 0.2577490935312137, "grad_norm": 1.4318269491195679, "learning_rate": 1.7401316270438094e-05, "loss": 0.18412272930145263, "step": 12600 }, { "epoch": 0.2579536563038575, "grad_norm": 3.6004252433776855, "learning_rate": 1.7396859372511792e-05, "loss": 0.186953341960907, "step": 12610 }, { "epoch": 0.2581582190765014, "grad_norm": 3.65057373046875, "learning_rate": 1.7392399227747444e-05, "loss": 0.19709854125976561, "step": 12620 }, { "epoch": 0.2583627818491452, "grad_norm": 2.125488758087158, "learning_rate": 1.7387935838102817e-05, "loss": 0.19672220945358276, "step": 12630 }, { "epoch": 0.258567344621789, "grad_norm": 2.228093147277832, "learning_rate": 1.7383469205537107e-05, "loss": 0.1751378893852234, "step": 12640 }, { "epoch": 0.25877190739443284, "grad_norm": 1.4957780838012695, "learning_rate": 1.7378999332010937e-05, "loss": 0.20010945796966553, "step": 12650 }, { "epoch": 0.25897647016707664, "grad_norm": 1.755332589149475, "learning_rate": 1.737452621948635e-05, "loss": 0.19292802810668946, "step": 12660 }, { "epoch": 0.25918103293972045, "grad_norm": 4.380967140197754, "learning_rate": 1.737004986992681e-05, "loss": 0.18193590641021729, "step": 12670 }, { "epoch": 0.2593855957123643, "grad_norm": 1.8270378112792969, "learning_rate": 1.73655702852972e-05, "loss": 0.18007047176361085, "step": 12680 }, { "epoch": 0.2595901584850081, "grad_norm": 1.5605697631835938, "learning_rate": 1.736108746756383e-05, "loss": 0.19835602045059203, "step": 12690 }, { "epoch": 0.2597947212576519, "grad_norm": 2.0378212928771973, "learning_rate": 1.7356601418694422e-05, "loss": 0.1908400297164917, "step": 12700 }, { "epoch": 0.25999928403029576, "grad_norm": 2.1641921997070312, "learning_rate": 1.735211214065812e-05, "loss": 0.19231779575347902, "step": 12710 }, { "epoch": 0.26020384680293956, "grad_norm": 1.352246880531311, "learning_rate": 1.734761963542548e-05, "loss": 0.18568756580352783, "step": 12720 }, { "epoch": 0.26040840957558337, "grad_norm": 4.8931474685668945, "learning_rate": 1.734312390496849e-05, "loss": 0.19405252933502198, "step": 12730 }, { "epoch": 0.2606129723482272, "grad_norm": 2.734344482421875, "learning_rate": 1.733862495126053e-05, "loss": 0.19840434789657593, "step": 12740 }, { "epoch": 0.260817535120871, "grad_norm": 2.153346061706543, "learning_rate": 1.7334122776276413e-05, "loss": 0.1949142575263977, "step": 12750 }, { "epoch": 0.2610220978935148, "grad_norm": 2.1957693099975586, "learning_rate": 1.7329617381992365e-05, "loss": 0.18804094791412354, "step": 12760 }, { "epoch": 0.2612266606661587, "grad_norm": 1.947432279586792, "learning_rate": 1.7325108770386016e-05, "loss": 0.18863682746887206, "step": 12770 }, { "epoch": 0.2614312234388025, "grad_norm": 3.3944756984710693, "learning_rate": 1.7320596943436414e-05, "loss": 0.18319553136825562, "step": 12780 }, { "epoch": 0.2616357862114463, "grad_norm": 1.6647859811782837, "learning_rate": 1.7316081903124014e-05, "loss": 0.19301618337631227, "step": 12790 }, { "epoch": 0.26184034898409014, "grad_norm": 3.0470383167266846, "learning_rate": 1.7311563651430692e-05, "loss": 0.19326651096343994, "step": 12800 }, { "epoch": 0.26204491175673394, "grad_norm": 1.7925070524215698, "learning_rate": 1.730704219033972e-05, "loss": 0.1767452359199524, "step": 12810 }, { "epoch": 0.26224947452937775, "grad_norm": 4.262508869171143, "learning_rate": 1.7302517521835788e-05, "loss": 0.19560936689376832, "step": 12820 }, { "epoch": 0.2624540373020216, "grad_norm": 3.6782517433166504, "learning_rate": 1.729798964790499e-05, "loss": 0.19189242124557496, "step": 12830 }, { "epoch": 0.2626586000746654, "grad_norm": 2.6887898445129395, "learning_rate": 1.729345857053483e-05, "loss": 0.19187920093536376, "step": 12840 }, { "epoch": 0.2628631628473092, "grad_norm": 2.4207727909088135, "learning_rate": 1.728892429171421e-05, "loss": 0.18788809776306153, "step": 12850 }, { "epoch": 0.26306772561995306, "grad_norm": 1.7739465236663818, "learning_rate": 1.7284386813433453e-05, "loss": 0.1813512086868286, "step": 12860 }, { "epoch": 0.26327228839259686, "grad_norm": 1.7319978475570679, "learning_rate": 1.7279846137684266e-05, "loss": 0.19549086093902587, "step": 12870 }, { "epoch": 0.26347685116524067, "grad_norm": 1.3025435209274292, "learning_rate": 1.7275302266459783e-05, "loss": 0.1798069715499878, "step": 12880 }, { "epoch": 0.2636814139378845, "grad_norm": 2.7282230854034424, "learning_rate": 1.7270755201754518e-05, "loss": 0.19341530799865722, "step": 12890 }, { "epoch": 0.2638859767105283, "grad_norm": 2.8360016345977783, "learning_rate": 1.72662049455644e-05, "loss": 0.18998868465423585, "step": 12900 }, { "epoch": 0.2640905394831722, "grad_norm": 2.255140781402588, "learning_rate": 1.7261651499886757e-05, "loss": 0.17957022190093994, "step": 12910 }, { "epoch": 0.264295102255816, "grad_norm": 3.506589889526367, "learning_rate": 1.7257094866720315e-05, "loss": 0.19567240476608277, "step": 12920 }, { "epoch": 0.2644996650284598, "grad_norm": 2.3658435344696045, "learning_rate": 1.72525350480652e-05, "loss": 0.18438918590545655, "step": 12930 }, { "epoch": 0.26470422780110364, "grad_norm": 1.9933770895004272, "learning_rate": 1.7247972045922936e-05, "loss": 0.1872640371322632, "step": 12940 }, { "epoch": 0.26490879057374744, "grad_norm": 2.4051575660705566, "learning_rate": 1.7243405862296442e-05, "loss": 0.19611091613769532, "step": 12950 }, { "epoch": 0.26511335334639125, "grad_norm": 1.3284071683883667, "learning_rate": 1.723883649919004e-05, "loss": 0.18915555477142335, "step": 12960 }, { "epoch": 0.2653179161190351, "grad_norm": 2.7841920852661133, "learning_rate": 1.7234263958609444e-05, "loss": 0.18118611574172974, "step": 12970 }, { "epoch": 0.2655224788916789, "grad_norm": 2.582265853881836, "learning_rate": 1.7229688242561758e-05, "loss": 0.1785653352737427, "step": 12980 }, { "epoch": 0.2657270416643227, "grad_norm": 3.372117757797241, "learning_rate": 1.722510935305549e-05, "loss": 0.17627557516098022, "step": 12990 }, { "epoch": 0.26593160443696656, "grad_norm": 2.619372606277466, "learning_rate": 1.722052729210053e-05, "loss": 0.17374662160873414, "step": 13000 }, { "epoch": 0.26613616720961036, "grad_norm": 2.038968086242676, "learning_rate": 1.7215942061708167e-05, "loss": 0.17828433513641356, "step": 13010 }, { "epoch": 0.26634072998225417, "grad_norm": 1.951658010482788, "learning_rate": 1.7211353663891082e-05, "loss": 0.18975962400436402, "step": 13020 }, { "epoch": 0.266545292754898, "grad_norm": 1.3590141534805298, "learning_rate": 1.720676210066334e-05, "loss": 0.18696633577346802, "step": 13030 }, { "epoch": 0.2667498555275418, "grad_norm": 2.8755533695220947, "learning_rate": 1.7202167374040404e-05, "loss": 0.18514879941940307, "step": 13040 }, { "epoch": 0.2669544183001856, "grad_norm": 3.5562736988067627, "learning_rate": 1.7197569486039114e-05, "loss": 0.1876865267753601, "step": 13050 }, { "epoch": 0.2671589810728295, "grad_norm": 2.925126552581787, "learning_rate": 1.719296843867771e-05, "loss": 0.16763519048690795, "step": 13060 }, { "epoch": 0.2673635438454733, "grad_norm": 4.002114295959473, "learning_rate": 1.7188364233975817e-05, "loss": 0.20204238891601561, "step": 13070 }, { "epoch": 0.2675681066181171, "grad_norm": 1.0482903718948364, "learning_rate": 1.7183756873954432e-05, "loss": 0.19611029624938964, "step": 13080 }, { "epoch": 0.26777266939076094, "grad_norm": 5.52549409866333, "learning_rate": 1.7179146360635955e-05, "loss": 0.19101667404174805, "step": 13090 }, { "epoch": 0.26797723216340474, "grad_norm": 1.0657540559768677, "learning_rate": 1.717453269604416e-05, "loss": 0.16647180318832397, "step": 13100 }, { "epoch": 0.26818179493604855, "grad_norm": 4.127771377563477, "learning_rate": 1.7169915882204212e-05, "loss": 0.20387134552001954, "step": 13110 }, { "epoch": 0.2683863577086924, "grad_norm": 1.5688806772232056, "learning_rate": 1.7165295921142644e-05, "loss": 0.19195311069488524, "step": 13120 }, { "epoch": 0.2685909204813362, "grad_norm": 2.45520281791687, "learning_rate": 1.7160672814887386e-05, "loss": 0.19279519319534302, "step": 13130 }, { "epoch": 0.26879548325398, "grad_norm": 5.674112319946289, "learning_rate": 1.7156046565467736e-05, "loss": 0.2008615255355835, "step": 13140 }, { "epoch": 0.26900004602662386, "grad_norm": 1.9232244491577148, "learning_rate": 1.7151417174914384e-05, "loss": 0.19706538915634156, "step": 13150 }, { "epoch": 0.26920460879926766, "grad_norm": 2.6596481800079346, "learning_rate": 1.7146784645259392e-05, "loss": 0.19457417726516724, "step": 13160 }, { "epoch": 0.26940917157191147, "grad_norm": 2.84069561958313, "learning_rate": 1.7142148978536196e-05, "loss": 0.19176867008209228, "step": 13170 }, { "epoch": 0.2696137343445553, "grad_norm": 2.4939792156219482, "learning_rate": 1.7137510176779622e-05, "loss": 0.17828773260116576, "step": 13180 }, { "epoch": 0.2698182971171991, "grad_norm": 1.9698364734649658, "learning_rate": 1.7132868242025854e-05, "loss": 0.19221029281616211, "step": 13190 }, { "epoch": 0.2700228598898429, "grad_norm": 1.3507376909255981, "learning_rate": 1.7128223176312465e-05, "loss": 0.17683533430099488, "step": 13200 }, { "epoch": 0.2702274226624868, "grad_norm": 3.924225330352783, "learning_rate": 1.7123574981678398e-05, "loss": 0.17226841449737548, "step": 13210 }, { "epoch": 0.2704319854351306, "grad_norm": 1.673667073249817, "learning_rate": 1.711892366016397e-05, "loss": 0.1840964674949646, "step": 13220 }, { "epoch": 0.2706365482077744, "grad_norm": 2.1483876705169678, "learning_rate": 1.7114269213810874e-05, "loss": 0.18305397033691406, "step": 13230 }, { "epoch": 0.27084111098041824, "grad_norm": 1.1932281255722046, "learning_rate": 1.7109611644662163e-05, "loss": 0.1872950792312622, "step": 13240 }, { "epoch": 0.27104567375306204, "grad_norm": 2.904383897781372, "learning_rate": 1.7104950954762273e-05, "loss": 0.18604724407196044, "step": 13250 }, { "epoch": 0.27125023652570585, "grad_norm": 3.973797559738159, "learning_rate": 1.7100287146157006e-05, "loss": 0.179701566696167, "step": 13260 }, { "epoch": 0.2714547992983497, "grad_norm": 2.633539915084839, "learning_rate": 1.7095620220893532e-05, "loss": 0.1948474168777466, "step": 13270 }, { "epoch": 0.2716593620709935, "grad_norm": 2.952022075653076, "learning_rate": 1.7090950181020386e-05, "loss": 0.19061172008514404, "step": 13280 }, { "epoch": 0.2718639248436373, "grad_norm": 5.057570934295654, "learning_rate": 1.7086277028587477e-05, "loss": 0.18785949945449829, "step": 13290 }, { "epoch": 0.27206848761628116, "grad_norm": 4.87427282333374, "learning_rate": 1.7081600765646077e-05, "loss": 0.20406813621520997, "step": 13300 }, { "epoch": 0.27227305038892496, "grad_norm": 2.444648027420044, "learning_rate": 1.707692139424882e-05, "loss": 0.18560242652893066, "step": 13310 }, { "epoch": 0.27247761316156877, "grad_norm": 4.142312526702881, "learning_rate": 1.707223891644971e-05, "loss": 0.1788311243057251, "step": 13320 }, { "epoch": 0.2726821759342126, "grad_norm": 1.977915644645691, "learning_rate": 1.706755333430411e-05, "loss": 0.1889405608177185, "step": 13330 }, { "epoch": 0.2728867387068564, "grad_norm": 2.9605867862701416, "learning_rate": 1.7062864649868757e-05, "loss": 0.19345053434371948, "step": 13340 }, { "epoch": 0.2730913014795002, "grad_norm": 1.2046802043914795, "learning_rate": 1.7058172865201728e-05, "loss": 0.18053547143936158, "step": 13350 }, { "epoch": 0.2732958642521441, "grad_norm": 2.6854488849639893, "learning_rate": 1.7053477982362482e-05, "loss": 0.19556775093078613, "step": 13360 }, { "epoch": 0.2735004270247879, "grad_norm": 1.3108793497085571, "learning_rate": 1.7048780003411822e-05, "loss": 0.18305550813674926, "step": 13370 }, { "epoch": 0.27370498979743174, "grad_norm": 5.16260290145874, "learning_rate": 1.704407893041192e-05, "loss": 0.19241206645965575, "step": 13380 }, { "epoch": 0.27390955257007554, "grad_norm": 1.2837984561920166, "learning_rate": 1.703937476542631e-05, "loss": 0.1802686333656311, "step": 13390 }, { "epoch": 0.27411411534271934, "grad_norm": 1.2187341451644897, "learning_rate": 1.703466751051987e-05, "loss": 0.19413495063781738, "step": 13400 }, { "epoch": 0.2743186781153632, "grad_norm": 2.823453903198242, "learning_rate": 1.7029957167758843e-05, "loss": 0.19479031562805177, "step": 13410 }, { "epoch": 0.274523240888007, "grad_norm": 1.0689475536346436, "learning_rate": 1.7025243739210825e-05, "loss": 0.19619263410568238, "step": 13420 }, { "epoch": 0.2747278036606508, "grad_norm": 2.6662368774414062, "learning_rate": 1.7020527226944766e-05, "loss": 0.17076423168182372, "step": 13430 }, { "epoch": 0.27493236643329466, "grad_norm": 2.2935233116149902, "learning_rate": 1.701580763303097e-05, "loss": 0.17992310523986815, "step": 13440 }, { "epoch": 0.27513692920593846, "grad_norm": 1.151465654373169, "learning_rate": 1.7011084959541098e-05, "loss": 0.17683844566345214, "step": 13450 }, { "epoch": 0.27534149197858226, "grad_norm": 1.1503018140792847, "learning_rate": 1.7006359208548156e-05, "loss": 0.1825258255004883, "step": 13460 }, { "epoch": 0.2755460547512261, "grad_norm": 1.4811893701553345, "learning_rate": 1.7001630382126506e-05, "loss": 0.18383704423904418, "step": 13470 }, { "epoch": 0.2757506175238699, "grad_norm": 2.1848959922790527, "learning_rate": 1.6996898482351854e-05, "loss": 0.18682395219802855, "step": 13480 }, { "epoch": 0.2759551802965137, "grad_norm": 3.365086555480957, "learning_rate": 1.6992163511301263e-05, "loss": 0.17150847911834716, "step": 13490 }, { "epoch": 0.2761597430691576, "grad_norm": 1.4307379722595215, "learning_rate": 1.698742547105314e-05, "loss": 0.1851707935333252, "step": 13500 }, { "epoch": 0.2763643058418014, "grad_norm": 2.7614402770996094, "learning_rate": 1.6982684363687232e-05, "loss": 0.1918013095855713, "step": 13510 }, { "epoch": 0.2765688686144452, "grad_norm": 2.4478673934936523, "learning_rate": 1.6977940191284653e-05, "loss": 0.18337298631668092, "step": 13520 }, { "epoch": 0.27677343138708904, "grad_norm": 1.6050304174423218, "learning_rate": 1.6973192955927832e-05, "loss": 0.18209713697433472, "step": 13530 }, { "epoch": 0.27697799415973284, "grad_norm": 3.018066167831421, "learning_rate": 1.6968442659700576e-05, "loss": 0.18447089195251465, "step": 13540 }, { "epoch": 0.27718255693237664, "grad_norm": 1.8974385261535645, "learning_rate": 1.6963689304688005e-05, "loss": 0.16669023036956787, "step": 13550 }, { "epoch": 0.2773871197050205, "grad_norm": 2.0737950801849365, "learning_rate": 1.6958932892976606e-05, "loss": 0.1896853804588318, "step": 13560 }, { "epoch": 0.2775916824776643, "grad_norm": 3.3021011352539062, "learning_rate": 1.695417342665419e-05, "loss": 0.2010371208190918, "step": 13570 }, { "epoch": 0.2777962452503081, "grad_norm": 4.007541656494141, "learning_rate": 1.694941090780992e-05, "loss": 0.1802680730819702, "step": 13580 }, { "epoch": 0.27800080802295196, "grad_norm": 1.8748719692230225, "learning_rate": 1.6944645338534295e-05, "loss": 0.19286071062088012, "step": 13590 }, { "epoch": 0.27820537079559576, "grad_norm": 2.1632888317108154, "learning_rate": 1.693987672091915e-05, "loss": 0.17127513885498047, "step": 13600 }, { "epoch": 0.27840993356823956, "grad_norm": 1.800730586051941, "learning_rate": 1.693510505705767e-05, "loss": 0.17233757972717284, "step": 13610 }, { "epoch": 0.2786144963408834, "grad_norm": 5.737215042114258, "learning_rate": 1.6930330349044356e-05, "loss": 0.18205748796463012, "step": 13620 }, { "epoch": 0.2788190591135272, "grad_norm": 1.8651351928710938, "learning_rate": 1.6925552598975064e-05, "loss": 0.17149900197982787, "step": 13630 }, { "epoch": 0.279023621886171, "grad_norm": 2.119588613510132, "learning_rate": 1.6920771808946978e-05, "loss": 0.18354798555374147, "step": 13640 }, { "epoch": 0.2792281846588149, "grad_norm": 3.05942964553833, "learning_rate": 1.6915987981058618e-05, "loss": 0.1854243516921997, "step": 13650 }, { "epoch": 0.2794327474314587, "grad_norm": 2.465205192565918, "learning_rate": 1.6911201117409833e-05, "loss": 0.18557884693145751, "step": 13660 }, { "epoch": 0.2796373102041025, "grad_norm": 2.2668845653533936, "learning_rate": 1.6906411220101814e-05, "loss": 0.18975284099578857, "step": 13670 }, { "epoch": 0.27984187297674634, "grad_norm": 5.116244792938232, "learning_rate": 1.6901618291237073e-05, "loss": 0.19075658321380615, "step": 13680 }, { "epoch": 0.28004643574939014, "grad_norm": 2.3340771198272705, "learning_rate": 1.689682233291946e-05, "loss": 0.16986427307128907, "step": 13690 }, { "epoch": 0.28025099852203395, "grad_norm": 2.6908252239227295, "learning_rate": 1.689202334725415e-05, "loss": 0.1869903564453125, "step": 13700 }, { "epoch": 0.2804555612946778, "grad_norm": 2.541250705718994, "learning_rate": 1.688722133634765e-05, "loss": 0.1841808795928955, "step": 13710 }, { "epoch": 0.2806601240673216, "grad_norm": 1.5392893552780151, "learning_rate": 1.6882416302307794e-05, "loss": 0.18024892807006837, "step": 13720 }, { "epoch": 0.2808646868399654, "grad_norm": 0.9936197400093079, "learning_rate": 1.6877608247243745e-05, "loss": 0.16119377613067626, "step": 13730 }, { "epoch": 0.28106924961260926, "grad_norm": 3.6228041648864746, "learning_rate": 1.6872797173265987e-05, "loss": 0.18578908443450928, "step": 13740 }, { "epoch": 0.28127381238525306, "grad_norm": 3.499932050704956, "learning_rate": 1.6867983082486334e-05, "loss": 0.18237770795822145, "step": 13750 }, { "epoch": 0.28147837515789687, "grad_norm": 1.9052306413650513, "learning_rate": 1.686316597701792e-05, "loss": 0.1842581510543823, "step": 13760 }, { "epoch": 0.2816829379305407, "grad_norm": 1.2345625162124634, "learning_rate": 1.6858345858975208e-05, "loss": 0.17327096462249755, "step": 13770 }, { "epoch": 0.2818875007031845, "grad_norm": 2.7474067211151123, "learning_rate": 1.685352273047398e-05, "loss": 0.16985139846801758, "step": 13780 }, { "epoch": 0.2820920634758283, "grad_norm": 2.161217451095581, "learning_rate": 1.684869659363134e-05, "loss": 0.18712332248687744, "step": 13790 }, { "epoch": 0.2822966262484722, "grad_norm": 1.216394066810608, "learning_rate": 1.6843867450565708e-05, "loss": 0.18020708560943605, "step": 13800 }, { "epoch": 0.282501189021116, "grad_norm": 1.6335827112197876, "learning_rate": 1.683903530339683e-05, "loss": 0.18192871809005737, "step": 13810 }, { "epoch": 0.2827057517937598, "grad_norm": 5.318080902099609, "learning_rate": 1.683420015424577e-05, "loss": 0.20345258712768555, "step": 13820 }, { "epoch": 0.28291031456640364, "grad_norm": 1.695279598236084, "learning_rate": 1.6829362005234902e-05, "loss": 0.18318966627120972, "step": 13830 }, { "epoch": 0.28311487733904744, "grad_norm": 2.0835864543914795, "learning_rate": 1.6824520858487932e-05, "loss": 0.17513484954833985, "step": 13840 }, { "epoch": 0.2833194401116913, "grad_norm": 1.8314003944396973, "learning_rate": 1.6819676716129865e-05, "loss": 0.16967917680740358, "step": 13850 }, { "epoch": 0.2835240028843351, "grad_norm": 3.109528064727783, "learning_rate": 1.6814829580287032e-05, "loss": 0.18129130601882934, "step": 13860 }, { "epoch": 0.2837285656569789, "grad_norm": 2.5597941875457764, "learning_rate": 1.680997945308707e-05, "loss": 0.1942615270614624, "step": 13870 }, { "epoch": 0.28393312842962276, "grad_norm": 2.4175939559936523, "learning_rate": 1.6805126336658944e-05, "loss": 0.18240203857421874, "step": 13880 }, { "epoch": 0.28413769120226656, "grad_norm": 1.3763625621795654, "learning_rate": 1.680027023313291e-05, "loss": 0.18090847730636597, "step": 13890 }, { "epoch": 0.28434225397491036, "grad_norm": 2.2648260593414307, "learning_rate": 1.679541114464055e-05, "loss": 0.17562729120254517, "step": 13900 }, { "epoch": 0.2845468167475542, "grad_norm": 1.2645450830459595, "learning_rate": 1.6790549073314752e-05, "loss": 0.19328887462615968, "step": 13910 }, { "epoch": 0.284751379520198, "grad_norm": 2.7189433574676514, "learning_rate": 1.6785684021289713e-05, "loss": 0.1855617046356201, "step": 13920 }, { "epoch": 0.2849559422928418, "grad_norm": 3.805114269256592, "learning_rate": 1.678081599070094e-05, "loss": 0.190021288394928, "step": 13930 }, { "epoch": 0.2851605050654857, "grad_norm": 1.5494416952133179, "learning_rate": 1.6775944983685247e-05, "loss": 0.19763808250427245, "step": 13940 }, { "epoch": 0.2853650678381295, "grad_norm": 3.5277674198150635, "learning_rate": 1.677107100238075e-05, "loss": 0.1856139898300171, "step": 13950 }, { "epoch": 0.2855696306107733, "grad_norm": 2.5180132389068604, "learning_rate": 1.6766194048926875e-05, "loss": 0.17851006984710693, "step": 13960 }, { "epoch": 0.28577419338341714, "grad_norm": 1.8164949417114258, "learning_rate": 1.6761314125464356e-05, "loss": 0.18033318519592284, "step": 13970 }, { "epoch": 0.28597875615606094, "grad_norm": 1.0165736675262451, "learning_rate": 1.6756431234135224e-05, "loss": 0.17120585441589356, "step": 13980 }, { "epoch": 0.28618331892870474, "grad_norm": 3.0944952964782715, "learning_rate": 1.6751545377082815e-05, "loss": 0.18562085628509523, "step": 13990 }, { "epoch": 0.2863878817013486, "grad_norm": 2.8359875679016113, "learning_rate": 1.6746656556451766e-05, "loss": 0.18291885852813722, "step": 14000 }, { "epoch": 0.2865924444739924, "grad_norm": 5.558093547821045, "learning_rate": 1.6741764774388015e-05, "loss": 0.19069011211395265, "step": 14010 }, { "epoch": 0.2867970072466362, "grad_norm": 1.5875073671340942, "learning_rate": 1.673687003303881e-05, "loss": 0.18035731315612794, "step": 14020 }, { "epoch": 0.28700157001928006, "grad_norm": 2.451277732849121, "learning_rate": 1.6731972334552676e-05, "loss": 0.18929438591003417, "step": 14030 }, { "epoch": 0.28720613279192386, "grad_norm": 2.020711660385132, "learning_rate": 1.672707168107946e-05, "loss": 0.18105652332305908, "step": 14040 }, { "epoch": 0.28741069556456766, "grad_norm": 2.8180294036865234, "learning_rate": 1.6722168074770285e-05, "loss": 0.19240365028381348, "step": 14050 }, { "epoch": 0.2876152583372115, "grad_norm": 1.2034715414047241, "learning_rate": 1.6717261517777586e-05, "loss": 0.1750725507736206, "step": 14060 }, { "epoch": 0.2878198211098553, "grad_norm": 4.473069667816162, "learning_rate": 1.6712352012255088e-05, "loss": 0.1799041748046875, "step": 14070 }, { "epoch": 0.2880243838824991, "grad_norm": 1.9020237922668457, "learning_rate": 1.6707439560357806e-05, "loss": 0.19398813247680663, "step": 14080 }, { "epoch": 0.288228946655143, "grad_norm": 2.14123797416687, "learning_rate": 1.6702524164242054e-05, "loss": 0.18113714456558228, "step": 14090 }, { "epoch": 0.2884335094277868, "grad_norm": 2.210048198699951, "learning_rate": 1.669760582606544e-05, "loss": 0.17958219051361085, "step": 14100 }, { "epoch": 0.2886380722004306, "grad_norm": 1.3550251722335815, "learning_rate": 1.6692684547986854e-05, "loss": 0.1773930311203003, "step": 14110 }, { "epoch": 0.28884263497307444, "grad_norm": 1.4169702529907227, "learning_rate": 1.668776033216648e-05, "loss": 0.1789454221725464, "step": 14120 }, { "epoch": 0.28904719774571824, "grad_norm": 2.0579330921173096, "learning_rate": 1.6682833180765802e-05, "loss": 0.18061180114746095, "step": 14130 }, { "epoch": 0.28925176051836204, "grad_norm": 3.380537509918213, "learning_rate": 1.6677903095947576e-05, "loss": 0.18838318586349487, "step": 14140 }, { "epoch": 0.2894563232910059, "grad_norm": 1.6327323913574219, "learning_rate": 1.667297007987586e-05, "loss": 0.1802741050720215, "step": 14150 }, { "epoch": 0.2896608860636497, "grad_norm": 3.641195058822632, "learning_rate": 1.6668034134715993e-05, "loss": 0.1778885841369629, "step": 14160 }, { "epoch": 0.2898654488362935, "grad_norm": 4.966531276702881, "learning_rate": 1.666309526263459e-05, "loss": 0.18891530036926268, "step": 14170 }, { "epoch": 0.29007001160893736, "grad_norm": 2.1618900299072266, "learning_rate": 1.665815346579957e-05, "loss": 0.19656363725662232, "step": 14180 }, { "epoch": 0.29027457438158116, "grad_norm": 2.419365167617798, "learning_rate": 1.6653208746380117e-05, "loss": 0.17395318746566774, "step": 14190 }, { "epoch": 0.29047913715422496, "grad_norm": 0.8159370422363281, "learning_rate": 1.664826110654672e-05, "loss": 0.173176372051239, "step": 14200 }, { "epoch": 0.2906836999268688, "grad_norm": 1.7462066411972046, "learning_rate": 1.664331054847112e-05, "loss": 0.17132006883621215, "step": 14210 }, { "epoch": 0.2908882626995126, "grad_norm": 1.4035680294036865, "learning_rate": 1.663835707432637e-05, "loss": 0.1749478816986084, "step": 14220 }, { "epoch": 0.2910928254721564, "grad_norm": 1.1267619132995605, "learning_rate": 1.6633400686286777e-05, "loss": 0.18387298583984374, "step": 14230 }, { "epoch": 0.2912973882448003, "grad_norm": 2.64019775390625, "learning_rate": 1.6628441386527948e-05, "loss": 0.2010817050933838, "step": 14240 }, { "epoch": 0.2915019510174441, "grad_norm": 2.367182493209839, "learning_rate": 1.6623479177226754e-05, "loss": 0.1900556802749634, "step": 14250 }, { "epoch": 0.2917065137900879, "grad_norm": 2.774932384490967, "learning_rate": 1.6618514060561346e-05, "loss": 0.1891147494316101, "step": 14260 }, { "epoch": 0.29191107656273174, "grad_norm": 2.097684383392334, "learning_rate": 1.6613546038711156e-05, "loss": 0.1825414776802063, "step": 14270 }, { "epoch": 0.29211563933537554, "grad_norm": 2.8894054889678955, "learning_rate": 1.660857511385689e-05, "loss": 0.16807618141174316, "step": 14280 }, { "epoch": 0.29232020210801934, "grad_norm": 3.7788712978363037, "learning_rate": 1.6603601288180518e-05, "loss": 0.17964912652969361, "step": 14290 }, { "epoch": 0.2925247648806632, "grad_norm": 1.4760582447052002, "learning_rate": 1.6598624563865303e-05, "loss": 0.1758614420890808, "step": 14300 }, { "epoch": 0.292729327653307, "grad_norm": 3.197882652282715, "learning_rate": 1.659364494309576e-05, "loss": 0.19582148790359497, "step": 14310 }, { "epoch": 0.29293389042595086, "grad_norm": 2.054865837097168, "learning_rate": 1.658866242805769e-05, "loss": 0.1849205493927002, "step": 14320 }, { "epoch": 0.29313845319859466, "grad_norm": 2.1254589557647705, "learning_rate": 1.658367702093816e-05, "loss": 0.18920769691467285, "step": 14330 }, { "epoch": 0.29334301597123846, "grad_norm": 4.735405921936035, "learning_rate": 1.65786887239255e-05, "loss": 0.18776636123657225, "step": 14340 }, { "epoch": 0.2935475787438823, "grad_norm": 2.243389844894409, "learning_rate": 1.6573697539209318e-05, "loss": 0.17539854049682618, "step": 14350 }, { "epoch": 0.2937521415165261, "grad_norm": 2.7384378910064697, "learning_rate": 1.6568703468980484e-05, "loss": 0.18359289169311524, "step": 14360 }, { "epoch": 0.2939567042891699, "grad_norm": 3.3772614002227783, "learning_rate": 1.6563706515431138e-05, "loss": 0.18129732608795165, "step": 14370 }, { "epoch": 0.2941612670618138, "grad_norm": 2.6461896896362305, "learning_rate": 1.655870668075468e-05, "loss": 0.20216588973999022, "step": 14380 }, { "epoch": 0.2943658298344576, "grad_norm": 1.93759024143219, "learning_rate": 1.655370396714578e-05, "loss": 0.17256534099578857, "step": 14390 }, { "epoch": 0.2945703926071014, "grad_norm": 6.083837509155273, "learning_rate": 1.6548698376800368e-05, "loss": 0.17856069803237914, "step": 14400 }, { "epoch": 0.29477495537974524, "grad_norm": 4.030029296875, "learning_rate": 1.6543689911915646e-05, "loss": 0.19832346439361573, "step": 14410 }, { "epoch": 0.29497951815238904, "grad_norm": 2.07696270942688, "learning_rate": 1.6538678574690067e-05, "loss": 0.19705092906951904, "step": 14420 }, { "epoch": 0.29518408092503284, "grad_norm": 4.552485942840576, "learning_rate": 1.653366436732334e-05, "loss": 0.17666771411895751, "step": 14430 }, { "epoch": 0.2953886436976767, "grad_norm": 2.3936524391174316, "learning_rate": 1.6528647292016457e-05, "loss": 0.19011210203170775, "step": 14440 }, { "epoch": 0.2955932064703205, "grad_norm": 2.8573360443115234, "learning_rate": 1.6523627350971645e-05, "loss": 0.18483688831329345, "step": 14450 }, { "epoch": 0.2957977692429643, "grad_norm": 1.1878163814544678, "learning_rate": 1.6518604546392398e-05, "loss": 0.1684914708137512, "step": 14460 }, { "epoch": 0.29600233201560816, "grad_norm": 2.699352979660034, "learning_rate": 1.651357888048347e-05, "loss": 0.18455549478530883, "step": 14470 }, { "epoch": 0.29620689478825196, "grad_norm": 2.2322261333465576, "learning_rate": 1.650855035545087e-05, "loss": 0.18938839435577393, "step": 14480 }, { "epoch": 0.29641145756089576, "grad_norm": 2.630073070526123, "learning_rate": 1.6503518973501852e-05, "loss": 0.17366011142730714, "step": 14490 }, { "epoch": 0.2966160203335396, "grad_norm": 3.7383196353912354, "learning_rate": 1.649848473684494e-05, "loss": 0.18873023986816406, "step": 14500 }, { "epoch": 0.2968205831061834, "grad_norm": 1.4055439233779907, "learning_rate": 1.6493447647689898e-05, "loss": 0.18776957988739013, "step": 14510 }, { "epoch": 0.2970251458788272, "grad_norm": 1.917275071144104, "learning_rate": 1.648840770824775e-05, "loss": 0.18863667249679567, "step": 14520 }, { "epoch": 0.2972297086514711, "grad_norm": 1.7316594123840332, "learning_rate": 1.6483364920730767e-05, "loss": 0.19126291275024415, "step": 14530 }, { "epoch": 0.2974342714241149, "grad_norm": 3.7964892387390137, "learning_rate": 1.647831928735247e-05, "loss": 0.19695039987564086, "step": 14540 }, { "epoch": 0.2976388341967587, "grad_norm": 1.4169528484344482, "learning_rate": 1.6473270810327637e-05, "loss": 0.19583945274353026, "step": 14550 }, { "epoch": 0.29784339696940254, "grad_norm": 1.6031855344772339, "learning_rate": 1.6468219491872278e-05, "loss": 0.20007381439208985, "step": 14560 }, { "epoch": 0.29804795974204634, "grad_norm": 1.7805312871932983, "learning_rate": 1.646316533420367e-05, "loss": 0.17743219137191774, "step": 14570 }, { "epoch": 0.29825252251469014, "grad_norm": 1.840511679649353, "learning_rate": 1.645810833954032e-05, "loss": 0.18461519479751587, "step": 14580 }, { "epoch": 0.298457085287334, "grad_norm": 2.2962324619293213, "learning_rate": 1.645304851010199e-05, "loss": 0.1866754412651062, "step": 14590 }, { "epoch": 0.2986616480599778, "grad_norm": 1.7575962543487549, "learning_rate": 1.644798584810968e-05, "loss": 0.17844048738479615, "step": 14600 }, { "epoch": 0.2988662108326216, "grad_norm": 1.8069833517074585, "learning_rate": 1.6442920355785635e-05, "loss": 0.17971153259277345, "step": 14610 }, { "epoch": 0.29907077360526546, "grad_norm": 2.6869449615478516, "learning_rate": 1.6437852035353352e-05, "loss": 0.195261549949646, "step": 14620 }, { "epoch": 0.29927533637790926, "grad_norm": 1.6382756233215332, "learning_rate": 1.6432780889037553e-05, "loss": 0.17132503986358644, "step": 14630 }, { "epoch": 0.29947989915055306, "grad_norm": 2.3585152626037598, "learning_rate": 1.642770691906421e-05, "loss": 0.19093382358551025, "step": 14640 }, { "epoch": 0.2996844619231969, "grad_norm": 2.3877410888671875, "learning_rate": 1.6422630127660536e-05, "loss": 0.18927621841430664, "step": 14650 }, { "epoch": 0.2998890246958407, "grad_norm": 2.47477388381958, "learning_rate": 1.6417550517054976e-05, "loss": 0.18373926877975463, "step": 14660 }, { "epoch": 0.3000935874684845, "grad_norm": 3.3840785026550293, "learning_rate": 1.6412468089477215e-05, "loss": 0.18923399448394776, "step": 14670 }, { "epoch": 0.3002981502411284, "grad_norm": 1.047803521156311, "learning_rate": 1.6407382847158185e-05, "loss": 0.18519375324249268, "step": 14680 }, { "epoch": 0.3005027130137722, "grad_norm": 1.9172086715698242, "learning_rate": 1.6402294792330032e-05, "loss": 0.17587929964065552, "step": 14690 }, { "epoch": 0.300707275786416, "grad_norm": 1.2076033353805542, "learning_rate": 1.6397203927226152e-05, "loss": 0.17745195627212523, "step": 14700 }, { "epoch": 0.30091183855905984, "grad_norm": 5.478806972503662, "learning_rate": 1.6392110254081176e-05, "loss": 0.1804277181625366, "step": 14710 }, { "epoch": 0.30111640133170364, "grad_norm": 2.9847843647003174, "learning_rate": 1.6387013775130953e-05, "loss": 0.18869316577911377, "step": 14720 }, { "epoch": 0.30132096410434744, "grad_norm": 2.0210464000701904, "learning_rate": 1.6381914492612585e-05, "loss": 0.17323193550109864, "step": 14730 }, { "epoch": 0.3015255268769913, "grad_norm": 3.1675610542297363, "learning_rate": 1.6376812408764383e-05, "loss": 0.1740504264831543, "step": 14740 }, { "epoch": 0.3017300896496351, "grad_norm": 5.122091293334961, "learning_rate": 1.6371707525825904e-05, "loss": 0.17833032608032226, "step": 14750 }, { "epoch": 0.3019346524222789, "grad_norm": 1.433624505996704, "learning_rate": 1.636659984603792e-05, "loss": 0.20210096836090088, "step": 14760 }, { "epoch": 0.30213921519492276, "grad_norm": 2.7304770946502686, "learning_rate": 1.636148937164245e-05, "loss": 0.17552518844604492, "step": 14770 }, { "epoch": 0.30234377796756656, "grad_norm": 2.2285733222961426, "learning_rate": 1.6356376104882715e-05, "loss": 0.1994379758834839, "step": 14780 }, { "epoch": 0.3025483407402104, "grad_norm": 1.4382553100585938, "learning_rate": 1.635126004800318e-05, "loss": 0.1732187509536743, "step": 14790 }, { "epoch": 0.3027529035128542, "grad_norm": 1.9991097450256348, "learning_rate": 1.6346141203249528e-05, "loss": 0.18865840435028075, "step": 14800 }, { "epoch": 0.302957466285498, "grad_norm": 1.5612506866455078, "learning_rate": 1.6341019572868667e-05, "loss": 0.18577929735183715, "step": 14810 }, { "epoch": 0.3031620290581419, "grad_norm": 2.541827917098999, "learning_rate": 1.6335895159108726e-05, "loss": 0.17614424228668213, "step": 14820 }, { "epoch": 0.3033665918307857, "grad_norm": 1.2341564893722534, "learning_rate": 1.6330767964219057e-05, "loss": 0.190514874458313, "step": 14830 }, { "epoch": 0.3035711546034295, "grad_norm": 1.3853538036346436, "learning_rate": 1.6325637990450234e-05, "loss": 0.20880434513092042, "step": 14840 }, { "epoch": 0.30377571737607334, "grad_norm": 2.1743009090423584, "learning_rate": 1.632050524005405e-05, "loss": 0.17650748491287233, "step": 14850 }, { "epoch": 0.30398028014871714, "grad_norm": 3.093930721282959, "learning_rate": 1.631536971528352e-05, "loss": 0.17536140680313111, "step": 14860 }, { "epoch": 0.30418484292136094, "grad_norm": 2.883366823196411, "learning_rate": 1.6310231418392866e-05, "loss": 0.17027565240859985, "step": 14870 }, { "epoch": 0.3043894056940048, "grad_norm": 3.1404125690460205, "learning_rate": 1.6305090351637542e-05, "loss": 0.18922187089920045, "step": 14880 }, { "epoch": 0.3045939684666486, "grad_norm": 1.5840624570846558, "learning_rate": 1.62999465172742e-05, "loss": 0.20086252689361572, "step": 14890 }, { "epoch": 0.3047985312392924, "grad_norm": 1.4504209756851196, "learning_rate": 1.6294799917560725e-05, "loss": 0.18048287630081178, "step": 14900 }, { "epoch": 0.30500309401193626, "grad_norm": 2.8961942195892334, "learning_rate": 1.6289650554756207e-05, "loss": 0.16825265884399415, "step": 14910 }, { "epoch": 0.30520765678458006, "grad_norm": 2.681407928466797, "learning_rate": 1.6284498431120947e-05, "loss": 0.17862071990966796, "step": 14920 }, { "epoch": 0.30541221955722386, "grad_norm": 3.6562891006469727, "learning_rate": 1.627934354891646e-05, "loss": 0.17877776622772218, "step": 14930 }, { "epoch": 0.3056167823298677, "grad_norm": 2.554445743560791, "learning_rate": 1.627418591040547e-05, "loss": 0.1776371955871582, "step": 14940 }, { "epoch": 0.3058213451025115, "grad_norm": 3.107114315032959, "learning_rate": 1.6269025517851917e-05, "loss": 0.1874755620956421, "step": 14950 }, { "epoch": 0.3060259078751553, "grad_norm": 5.085423469543457, "learning_rate": 1.626386237352095e-05, "loss": 0.1687152624130249, "step": 14960 }, { "epoch": 0.3062304706477992, "grad_norm": 5.180761814117432, "learning_rate": 1.6258696479678915e-05, "loss": 0.192275071144104, "step": 14970 }, { "epoch": 0.306435033420443, "grad_norm": 1.40413236618042, "learning_rate": 1.6253527838593375e-05, "loss": 0.18191064596176149, "step": 14980 }, { "epoch": 0.3066395961930868, "grad_norm": 1.1539645195007324, "learning_rate": 1.6248356452533093e-05, "loss": 0.17772905826568602, "step": 14990 }, { "epoch": 0.30684415896573064, "grad_norm": 2.3098645210266113, "learning_rate": 1.624318232376804e-05, "loss": 0.18674752712249756, "step": 15000 }, { "epoch": 0.30704872173837444, "grad_norm": 1.0344550609588623, "learning_rate": 1.6238005454569392e-05, "loss": 0.16566803455352783, "step": 15010 }, { "epoch": 0.30725328451101824, "grad_norm": 3.670915365219116, "learning_rate": 1.6232825847209526e-05, "loss": 0.1859275221824646, "step": 15020 }, { "epoch": 0.3074578472836621, "grad_norm": 1.9891263246536255, "learning_rate": 1.6227643503962023e-05, "loss": 0.17741796970367432, "step": 15030 }, { "epoch": 0.3076624100563059, "grad_norm": 3.789841890335083, "learning_rate": 1.622245842710166e-05, "loss": 0.17935464382171631, "step": 15040 }, { "epoch": 0.3078669728289497, "grad_norm": 1.800336241722107, "learning_rate": 1.6217270618904422e-05, "loss": 0.1840044379234314, "step": 15050 }, { "epoch": 0.30807153560159356, "grad_norm": 2.7612881660461426, "learning_rate": 1.621208008164748e-05, "loss": 0.17498139142990113, "step": 15060 }, { "epoch": 0.30827609837423736, "grad_norm": 1.530293583869934, "learning_rate": 1.620688681760922e-05, "loss": 0.18677887916564942, "step": 15070 }, { "epoch": 0.30848066114688116, "grad_norm": 5.089416980743408, "learning_rate": 1.6201690829069205e-05, "loss": 0.2055762529373169, "step": 15080 }, { "epoch": 0.308685223919525, "grad_norm": 2.309250593185425, "learning_rate": 1.6196492118308217e-05, "loss": 0.1846071481704712, "step": 15090 }, { "epoch": 0.3088897866921688, "grad_norm": 0.9047659635543823, "learning_rate": 1.6191290687608212e-05, "loss": 0.18727318048477173, "step": 15100 }, { "epoch": 0.3090943494648126, "grad_norm": 1.9241015911102295, "learning_rate": 1.618608653925235e-05, "loss": 0.19246299266815187, "step": 15110 }, { "epoch": 0.3092989122374565, "grad_norm": 1.401828646659851, "learning_rate": 1.6180879675524986e-05, "loss": 0.19010531902313232, "step": 15120 }, { "epoch": 0.3095034750101003, "grad_norm": 3.8008599281311035, "learning_rate": 1.6175670098711664e-05, "loss": 0.17412703037261962, "step": 15130 }, { "epoch": 0.3097080377827441, "grad_norm": 3.0314362049102783, "learning_rate": 1.6170457811099116e-05, "loss": 0.20034019947052, "step": 15140 }, { "epoch": 0.30991260055538794, "grad_norm": 1.637444019317627, "learning_rate": 1.6165242814975265e-05, "loss": 0.1810365080833435, "step": 15150 }, { "epoch": 0.31011716332803174, "grad_norm": 1.7204713821411133, "learning_rate": 1.6160025112629222e-05, "loss": 0.1894751787185669, "step": 15160 }, { "epoch": 0.31032172610067554, "grad_norm": 2.232956647872925, "learning_rate": 1.6154804706351292e-05, "loss": 0.1810903310775757, "step": 15170 }, { "epoch": 0.3105262888733194, "grad_norm": 1.213097095489502, "learning_rate": 1.614958159843296e-05, "loss": 0.1691259503364563, "step": 15180 }, { "epoch": 0.3107308516459632, "grad_norm": 2.127955436706543, "learning_rate": 1.6144355791166907e-05, "loss": 0.19069867134094237, "step": 15190 }, { "epoch": 0.310935414418607, "grad_norm": 1.1546725034713745, "learning_rate": 1.6139127286846986e-05, "loss": 0.1729802370071411, "step": 15200 }, { "epoch": 0.31113997719125086, "grad_norm": 1.1220051050186157, "learning_rate": 1.6133896087768237e-05, "loss": 0.1698810338973999, "step": 15210 }, { "epoch": 0.31134453996389466, "grad_norm": 1.3436158895492554, "learning_rate": 1.612866219622689e-05, "loss": 0.1857912063598633, "step": 15220 }, { "epoch": 0.31154910273653846, "grad_norm": 3.178337335586548, "learning_rate": 1.612342561452035e-05, "loss": 0.16781601905822754, "step": 15230 }, { "epoch": 0.3117536655091823, "grad_norm": 3.884321689605713, "learning_rate": 1.6118186344947204e-05, "loss": 0.16381486654281616, "step": 15240 }, { "epoch": 0.3119582282818261, "grad_norm": 2.199737310409546, "learning_rate": 1.6112944389807225e-05, "loss": 0.18148163557052613, "step": 15250 }, { "epoch": 0.31216279105447, "grad_norm": 1.478926181793213, "learning_rate": 1.6107699751401354e-05, "loss": 0.18474533557891845, "step": 15260 }, { "epoch": 0.3123673538271138, "grad_norm": 3.405669927597046, "learning_rate": 1.610245243203172e-05, "loss": 0.18347983360290526, "step": 15270 }, { "epoch": 0.3125719165997576, "grad_norm": 4.750792026519775, "learning_rate": 1.609720243400162e-05, "loss": 0.19081616401672363, "step": 15280 }, { "epoch": 0.31277647937240144, "grad_norm": 3.1389548778533936, "learning_rate": 1.6091949759615534e-05, "loss": 0.17739518880844116, "step": 15290 }, { "epoch": 0.31298104214504524, "grad_norm": 2.3710265159606934, "learning_rate": 1.608669441117911e-05, "loss": 0.16871309280395508, "step": 15300 }, { "epoch": 0.31318560491768904, "grad_norm": 0.7519932389259338, "learning_rate": 1.608143639099918e-05, "loss": 0.18796753883361816, "step": 15310 }, { "epoch": 0.3133901676903329, "grad_norm": 0.9698041081428528, "learning_rate": 1.6076175701383743e-05, "loss": 0.17995494604110718, "step": 15320 }, { "epoch": 0.3135947304629767, "grad_norm": 1.720611572265625, "learning_rate": 1.6070912344641963e-05, "loss": 0.17759113311767577, "step": 15330 }, { "epoch": 0.3137992932356205, "grad_norm": 0.5770703554153442, "learning_rate": 1.6065646323084184e-05, "loss": 0.1737258553504944, "step": 15340 }, { "epoch": 0.31400385600826436, "grad_norm": 3.16677188873291, "learning_rate": 1.606037763902191e-05, "loss": 0.1963411331176758, "step": 15350 }, { "epoch": 0.31420841878090816, "grad_norm": 1.3175365924835205, "learning_rate": 1.6055106294767836e-05, "loss": 0.17091139554977416, "step": 15360 }, { "epoch": 0.31441298155355196, "grad_norm": 2.1247787475585938, "learning_rate": 1.6049832292635793e-05, "loss": 0.16795403957366944, "step": 15370 }, { "epoch": 0.3146175443261958, "grad_norm": 1.0741311311721802, "learning_rate": 1.6044555634940806e-05, "loss": 0.17301157712936402, "step": 15380 }, { "epoch": 0.3148221070988396, "grad_norm": 2.1767048835754395, "learning_rate": 1.6039276323999045e-05, "loss": 0.17756520509719848, "step": 15390 }, { "epoch": 0.3150266698714834, "grad_norm": 0.843258798122406, "learning_rate": 1.6033994362127862e-05, "loss": 0.1688625454902649, "step": 15400 }, { "epoch": 0.3152312326441273, "grad_norm": 3.0079190731048584, "learning_rate": 1.6028709751645757e-05, "loss": 0.18361462354660035, "step": 15410 }, { "epoch": 0.3154357954167711, "grad_norm": 0.9052479863166809, "learning_rate": 1.602342249487241e-05, "loss": 0.1900031566619873, "step": 15420 }, { "epoch": 0.3156403581894149, "grad_norm": 3.113614559173584, "learning_rate": 1.6018132594128644e-05, "loss": 0.17707982063293456, "step": 15430 }, { "epoch": 0.31584492096205874, "grad_norm": 5.104113578796387, "learning_rate": 1.6012840051736454e-05, "loss": 0.1874373435974121, "step": 15440 }, { "epoch": 0.31604948373470254, "grad_norm": 1.5970427989959717, "learning_rate": 1.6007544870018995e-05, "loss": 0.18345751762390136, "step": 15450 }, { "epoch": 0.31625404650734634, "grad_norm": 1.172570824623108, "learning_rate": 1.6002247051300573e-05, "loss": 0.1637399673461914, "step": 15460 }, { "epoch": 0.3164586092799902, "grad_norm": 6.900490760803223, "learning_rate": 1.599694659790666e-05, "loss": 0.18508548736572267, "step": 15470 }, { "epoch": 0.316663172052634, "grad_norm": 3.0206944942474365, "learning_rate": 1.5991643512163874e-05, "loss": 0.17810291051864624, "step": 15480 }, { "epoch": 0.3168677348252778, "grad_norm": 3.610217571258545, "learning_rate": 1.5986337796400005e-05, "loss": 0.1803203344345093, "step": 15490 }, { "epoch": 0.31707229759792166, "grad_norm": 2.321093797683716, "learning_rate": 1.5981029452943978e-05, "loss": 0.18188765048980712, "step": 15500 }, { "epoch": 0.31727686037056546, "grad_norm": 3.376317262649536, "learning_rate": 1.5975718484125885e-05, "loss": 0.17709959745407106, "step": 15510 }, { "epoch": 0.31748142314320926, "grad_norm": 3.243656635284424, "learning_rate": 1.5970404892276965e-05, "loss": 0.18679046630859375, "step": 15520 }, { "epoch": 0.3176859859158531, "grad_norm": 0.8233914375305176, "learning_rate": 1.5965088679729613e-05, "loss": 0.18545083999633788, "step": 15530 }, { "epoch": 0.3178905486884969, "grad_norm": 2.6119604110717773, "learning_rate": 1.5959769848817362e-05, "loss": 0.17565460205078126, "step": 15540 }, { "epoch": 0.3180951114611407, "grad_norm": 3.5265371799468994, "learning_rate": 1.595444840187492e-05, "loss": 0.19443683624267577, "step": 15550 }, { "epoch": 0.3182996742337846, "grad_norm": 3.0054359436035156, "learning_rate": 1.5949124341238104e-05, "loss": 0.17173552513122559, "step": 15560 }, { "epoch": 0.3185042370064284, "grad_norm": 3.3166277408599854, "learning_rate": 1.594379766924392e-05, "loss": 0.1826813817024231, "step": 15570 }, { "epoch": 0.3187087997790722, "grad_norm": 1.732690691947937, "learning_rate": 1.593846838823049e-05, "loss": 0.1752653956413269, "step": 15580 }, { "epoch": 0.31891336255171604, "grad_norm": 1.7622116804122925, "learning_rate": 1.5933136500537095e-05, "loss": 0.1876833200454712, "step": 15590 }, { "epoch": 0.31911792532435984, "grad_norm": 1.102705955505371, "learning_rate": 1.592780200850416e-05, "loss": 0.18470335006713867, "step": 15600 }, { "epoch": 0.31932248809700364, "grad_norm": 2.805177688598633, "learning_rate": 1.5922464914473247e-05, "loss": 0.1800963044166565, "step": 15610 }, { "epoch": 0.3195270508696475, "grad_norm": 1.9803484678268433, "learning_rate": 1.5917125220787068e-05, "loss": 0.17798082828521727, "step": 15620 }, { "epoch": 0.3197316136422913, "grad_norm": 2.683511972427368, "learning_rate": 1.591178292978947e-05, "loss": 0.1842937469482422, "step": 15630 }, { "epoch": 0.3199361764149351, "grad_norm": 2.0444388389587402, "learning_rate": 1.5906438043825442e-05, "loss": 0.18161096572875976, "step": 15640 }, { "epoch": 0.32014073918757896, "grad_norm": 2.0647077560424805, "learning_rate": 1.590109056524111e-05, "loss": 0.18847963809967042, "step": 15650 }, { "epoch": 0.32034530196022276, "grad_norm": 1.775665283203125, "learning_rate": 1.589574049638374e-05, "loss": 0.18471683263778688, "step": 15660 }, { "epoch": 0.32054986473286656, "grad_norm": 1.803008794784546, "learning_rate": 1.5890387839601738e-05, "loss": 0.17647545337677, "step": 15670 }, { "epoch": 0.3207544275055104, "grad_norm": 2.758277416229248, "learning_rate": 1.5885032597244644e-05, "loss": 0.19588106870651245, "step": 15680 }, { "epoch": 0.3209589902781542, "grad_norm": 1.2817565202713013, "learning_rate": 1.5879674771663128e-05, "loss": 0.18818655014038085, "step": 15690 }, { "epoch": 0.321163553050798, "grad_norm": 1.6375970840454102, "learning_rate": 1.5874314365209e-05, "loss": 0.17913875579833985, "step": 15700 }, { "epoch": 0.3213681158234419, "grad_norm": 3.5379652976989746, "learning_rate": 1.5868951380235197e-05, "loss": 0.17095584869384767, "step": 15710 }, { "epoch": 0.3215726785960857, "grad_norm": 1.5097100734710693, "learning_rate": 1.5863585819095796e-05, "loss": 0.17255886793136596, "step": 15720 }, { "epoch": 0.32177724136872954, "grad_norm": 1.531005859375, "learning_rate": 1.5858217684146e-05, "loss": 0.1754847288131714, "step": 15730 }, { "epoch": 0.32198180414137334, "grad_norm": 1.683474063873291, "learning_rate": 1.5852846977742143e-05, "loss": 0.1764955163002014, "step": 15740 }, { "epoch": 0.32218636691401714, "grad_norm": 2.5080857276916504, "learning_rate": 1.5847473702241684e-05, "loss": 0.17346510887145997, "step": 15750 }, { "epoch": 0.322390929686661, "grad_norm": 1.8422327041625977, "learning_rate": 1.5842097860003214e-05, "loss": 0.17513010501861573, "step": 15760 }, { "epoch": 0.3225954924593048, "grad_norm": 2.3501832485198975, "learning_rate": 1.5836719453386446e-05, "loss": 0.17799968719482423, "step": 15770 }, { "epoch": 0.3228000552319486, "grad_norm": 1.37575364112854, "learning_rate": 1.583133848475223e-05, "loss": 0.1691657304763794, "step": 15780 }, { "epoch": 0.32300461800459246, "grad_norm": 1.8383575677871704, "learning_rate": 1.5825954956462524e-05, "loss": 0.17874534130096437, "step": 15790 }, { "epoch": 0.32320918077723626, "grad_norm": 2.2314844131469727, "learning_rate": 1.5820568870880423e-05, "loss": 0.1889496088027954, "step": 15800 }, { "epoch": 0.32341374354988006, "grad_norm": 3.1731948852539062, "learning_rate": 1.5815180230370135e-05, "loss": 0.19242745637893677, "step": 15810 }, { "epoch": 0.3236183063225239, "grad_norm": 1.1694272756576538, "learning_rate": 1.5809789037296997e-05, "loss": 0.16518291234970092, "step": 15820 }, { "epoch": 0.3238228690951677, "grad_norm": 2.566990613937378, "learning_rate": 1.5804395294027467e-05, "loss": 0.182535719871521, "step": 15830 }, { "epoch": 0.3240274318678115, "grad_norm": 2.4980309009552, "learning_rate": 1.579899900292911e-05, "loss": 0.1758872389793396, "step": 15840 }, { "epoch": 0.3242319946404554, "grad_norm": 1.8079370260238647, "learning_rate": 1.5793600166370628e-05, "loss": 0.1790791153907776, "step": 15850 }, { "epoch": 0.3244365574130992, "grad_norm": 1.209368348121643, "learning_rate": 1.5788198786721823e-05, "loss": 0.17509884834289552, "step": 15860 }, { "epoch": 0.324641120185743, "grad_norm": 0.945625364780426, "learning_rate": 1.5782794866353627e-05, "loss": 0.18237887620925902, "step": 15870 }, { "epoch": 0.32484568295838684, "grad_norm": 1.2784258127212524, "learning_rate": 1.5777388407638078e-05, "loss": 0.17824372053146362, "step": 15880 }, { "epoch": 0.32505024573103064, "grad_norm": 2.161280870437622, "learning_rate": 1.5771979412948328e-05, "loss": 0.16149556636810303, "step": 15890 }, { "epoch": 0.32525480850367444, "grad_norm": 1.7745016813278198, "learning_rate": 1.5766567884658648e-05, "loss": 0.1847790837287903, "step": 15900 }, { "epoch": 0.3254593712763183, "grad_norm": 2.036806344985962, "learning_rate": 1.5761153825144422e-05, "loss": 0.17309949398040772, "step": 15910 }, { "epoch": 0.3256639340489621, "grad_norm": 2.0046980381011963, "learning_rate": 1.5755737236782135e-05, "loss": 0.16515512466430665, "step": 15920 }, { "epoch": 0.3258684968216059, "grad_norm": 2.5690388679504395, "learning_rate": 1.5750318121949397e-05, "loss": 0.17674334049224855, "step": 15930 }, { "epoch": 0.32607305959424976, "grad_norm": 3.6370749473571777, "learning_rate": 1.574489648302491e-05, "loss": 0.17974607944488524, "step": 15940 }, { "epoch": 0.32627762236689356, "grad_norm": 1.1523374319076538, "learning_rate": 1.57394723223885e-05, "loss": 0.18137340545654296, "step": 15950 }, { "epoch": 0.32648218513953736, "grad_norm": 2.610947847366333, "learning_rate": 1.573404564242109e-05, "loss": 0.1819605827331543, "step": 15960 }, { "epoch": 0.3266867479121812, "grad_norm": 1.3087149858474731, "learning_rate": 1.5728616445504704e-05, "loss": 0.17825981378555297, "step": 15970 }, { "epoch": 0.326891310684825, "grad_norm": 5.53115701675415, "learning_rate": 1.5723184734022487e-05, "loss": 0.1826465606689453, "step": 15980 }, { "epoch": 0.3270958734574688, "grad_norm": 2.6233677864074707, "learning_rate": 1.571775051035868e-05, "loss": 0.17375471591949462, "step": 15990 }, { "epoch": 0.3273004362301127, "grad_norm": 2.9503562450408936, "learning_rate": 1.5712313776898623e-05, "loss": 0.1793125629425049, "step": 16000 }, { "epoch": 0.3275049990027565, "grad_norm": 2.34934663772583, "learning_rate": 1.5706874536028757e-05, "loss": 0.18514729738235475, "step": 16010 }, { "epoch": 0.3277095617754003, "grad_norm": 3.253110885620117, "learning_rate": 1.570143279013663e-05, "loss": 0.18347939252853393, "step": 16020 }, { "epoch": 0.32791412454804414, "grad_norm": 1.2918996810913086, "learning_rate": 1.5695988541610882e-05, "loss": 0.17667639255523682, "step": 16030 }, { "epoch": 0.32811868732068794, "grad_norm": 2.646404266357422, "learning_rate": 1.5690541792841267e-05, "loss": 0.19093260765075684, "step": 16040 }, { "epoch": 0.32832325009333174, "grad_norm": 1.8765095472335815, "learning_rate": 1.5685092546218616e-05, "loss": 0.186770498752594, "step": 16050 }, { "epoch": 0.3285278128659756, "grad_norm": 2.3215763568878174, "learning_rate": 1.567964080413487e-05, "loss": 0.17376444339752198, "step": 16060 }, { "epoch": 0.3287323756386194, "grad_norm": 2.0288312435150146, "learning_rate": 1.5674186568983057e-05, "loss": 0.16347343921661378, "step": 16070 }, { "epoch": 0.3289369384112632, "grad_norm": 3.680819511413574, "learning_rate": 1.5668729843157308e-05, "loss": 0.18826106786727906, "step": 16080 }, { "epoch": 0.32914150118390706, "grad_norm": 1.847021222114563, "learning_rate": 1.566327062905284e-05, "loss": 0.19004373550415038, "step": 16090 }, { "epoch": 0.32934606395655086, "grad_norm": 2.8239974975585938, "learning_rate": 1.565780892906597e-05, "loss": 0.18378221988677979, "step": 16100 }, { "epoch": 0.32955062672919466, "grad_norm": 2.517467975616455, "learning_rate": 1.5652344745594094e-05, "loss": 0.17909170389175416, "step": 16110 }, { "epoch": 0.3297551895018385, "grad_norm": 2.780397891998291, "learning_rate": 1.5646878081035714e-05, "loss": 0.18660367727279664, "step": 16120 }, { "epoch": 0.3299597522744823, "grad_norm": 1.3225092887878418, "learning_rate": 1.5641408937790404e-05, "loss": 0.18204686641693116, "step": 16130 }, { "epoch": 0.3301643150471261, "grad_norm": 2.3219714164733887, "learning_rate": 1.5635937318258842e-05, "loss": 0.1775282382965088, "step": 16140 }, { "epoch": 0.33036887781977, "grad_norm": 2.695068597793579, "learning_rate": 1.563046322484278e-05, "loss": 0.18774921894073487, "step": 16150 }, { "epoch": 0.3305734405924138, "grad_norm": 1.804656982421875, "learning_rate": 1.5624986659945065e-05, "loss": 0.17289717197418214, "step": 16160 }, { "epoch": 0.3307780033650576, "grad_norm": 1.571671724319458, "learning_rate": 1.5619507625969624e-05, "loss": 0.17190194129943848, "step": 16170 }, { "epoch": 0.33098256613770144, "grad_norm": 1.2224199771881104, "learning_rate": 1.5614026125321467e-05, "loss": 0.17824386358261107, "step": 16180 }, { "epoch": 0.33118712891034524, "grad_norm": 2.824817419052124, "learning_rate": 1.560854216040669e-05, "loss": 0.16621147394180297, "step": 16190 }, { "epoch": 0.3313916916829891, "grad_norm": 2.5005247592926025, "learning_rate": 1.5603055733632474e-05, "loss": 0.182555890083313, "step": 16200 }, { "epoch": 0.3315962544556329, "grad_norm": 2.266810417175293, "learning_rate": 1.559756684740707e-05, "loss": 0.19154422283172606, "step": 16210 }, { "epoch": 0.3318008172282767, "grad_norm": 2.57190203666687, "learning_rate": 1.5592075504139815e-05, "loss": 0.1807039499282837, "step": 16220 }, { "epoch": 0.33200538000092056, "grad_norm": 2.523033618927002, "learning_rate": 1.558658170624113e-05, "loss": 0.17434933185577392, "step": 16230 }, { "epoch": 0.33220994277356436, "grad_norm": 3.699246406555176, "learning_rate": 1.55810854561225e-05, "loss": 0.18088810443878173, "step": 16240 }, { "epoch": 0.33241450554620816, "grad_norm": 1.7253594398498535, "learning_rate": 1.55755867561965e-05, "loss": 0.18372809886932373, "step": 16250 }, { "epoch": 0.332619068318852, "grad_norm": 2.459291696548462, "learning_rate": 1.557008560887677e-05, "loss": 0.1847238540649414, "step": 16260 }, { "epoch": 0.3328236310914958, "grad_norm": 2.1571476459503174, "learning_rate": 1.556458201657803e-05, "loss": 0.1869566798210144, "step": 16270 }, { "epoch": 0.3330281938641396, "grad_norm": 1.8166630268096924, "learning_rate": 1.5559075981716075e-05, "loss": 0.18355480432510377, "step": 16280 }, { "epoch": 0.3332327566367835, "grad_norm": 3.956813097000122, "learning_rate": 1.555356750670777e-05, "loss": 0.1812732219696045, "step": 16290 }, { "epoch": 0.3334373194094273, "grad_norm": 2.1986308097839355, "learning_rate": 1.554805659397104e-05, "loss": 0.1742470979690552, "step": 16300 }, { "epoch": 0.3336418821820711, "grad_norm": 1.811492919921875, "learning_rate": 1.5542543245924903e-05, "loss": 0.18203688859939576, "step": 16310 }, { "epoch": 0.33384644495471494, "grad_norm": 1.2659555673599243, "learning_rate": 1.5537027464989427e-05, "loss": 0.17581963539123535, "step": 16320 }, { "epoch": 0.33405100772735874, "grad_norm": 1.5565755367279053, "learning_rate": 1.5531509253585754e-05, "loss": 0.16859607696533202, "step": 16330 }, { "epoch": 0.33425557050000254, "grad_norm": 1.210793137550354, "learning_rate": 1.552598861413609e-05, "loss": 0.17724258899688722, "step": 16340 }, { "epoch": 0.3344601332726464, "grad_norm": 1.6068360805511475, "learning_rate": 1.5520465549063713e-05, "loss": 0.18708134889602662, "step": 16350 }, { "epoch": 0.3346646960452902, "grad_norm": 1.041003942489624, "learning_rate": 1.5514940060792966e-05, "loss": 0.17310343980789183, "step": 16360 }, { "epoch": 0.334869258817934, "grad_norm": 2.9381978511810303, "learning_rate": 1.5509412151749246e-05, "loss": 0.19196161031723022, "step": 16370 }, { "epoch": 0.33507382159057786, "grad_norm": 2.420393943786621, "learning_rate": 1.5503881824359024e-05, "loss": 0.1779308795928955, "step": 16380 }, { "epoch": 0.33527838436322166, "grad_norm": 4.570771217346191, "learning_rate": 1.5498349081049822e-05, "loss": 0.18143553733825685, "step": 16390 }, { "epoch": 0.33548294713586546, "grad_norm": 3.2330055236816406, "learning_rate": 1.549281392425023e-05, "loss": 0.17932312488555907, "step": 16400 }, { "epoch": 0.3356875099085093, "grad_norm": 1.0821253061294556, "learning_rate": 1.5487276356389896e-05, "loss": 0.17659244537353516, "step": 16410 }, { "epoch": 0.3358920726811531, "grad_norm": 1.458929419517517, "learning_rate": 1.5481736379899525e-05, "loss": 0.18128379583358764, "step": 16420 }, { "epoch": 0.3360966354537969, "grad_norm": 1.9696145057678223, "learning_rate": 1.547619399721088e-05, "loss": 0.1882353663444519, "step": 16430 }, { "epoch": 0.3363011982264408, "grad_norm": 3.3930485248565674, "learning_rate": 1.5470649210756774e-05, "loss": 0.18931903839111328, "step": 16440 }, { "epoch": 0.3365057609990846, "grad_norm": 1.3475186824798584, "learning_rate": 1.546510202297109e-05, "loss": 0.17904651165008545, "step": 16450 }, { "epoch": 0.3367103237717284, "grad_norm": 2.076892137527466, "learning_rate": 1.5459552436288754e-05, "loss": 0.17729365825653076, "step": 16460 }, { "epoch": 0.33691488654437224, "grad_norm": 1.1680634021759033, "learning_rate": 1.5454000453145742e-05, "loss": 0.17534881830215454, "step": 16470 }, { "epoch": 0.33711944931701604, "grad_norm": 1.4488316774368286, "learning_rate": 1.5448446075979094e-05, "loss": 0.16196322441101074, "step": 16480 }, { "epoch": 0.33732401208965984, "grad_norm": 1.8636071681976318, "learning_rate": 1.5442889307226884e-05, "loss": 0.17245786190032958, "step": 16490 }, { "epoch": 0.3375285748623037, "grad_norm": 1.2661771774291992, "learning_rate": 1.5437330149328254e-05, "loss": 0.18172097206115723, "step": 16500 }, { "epoch": 0.3377331376349475, "grad_norm": 1.1547704935073853, "learning_rate": 1.5431768604723387e-05, "loss": 0.18398902416229249, "step": 16510 }, { "epoch": 0.3379377004075913, "grad_norm": 1.2263226509094238, "learning_rate": 1.5426204675853503e-05, "loss": 0.1809561014175415, "step": 16520 }, { "epoch": 0.33814226318023516, "grad_norm": 1.8792144060134888, "learning_rate": 1.542063836516089e-05, "loss": 0.17512348890304566, "step": 16530 }, { "epoch": 0.33834682595287896, "grad_norm": 1.6577421426773071, "learning_rate": 1.5415069675088867e-05, "loss": 0.16864922046661376, "step": 16540 }, { "epoch": 0.33855138872552276, "grad_norm": 2.4657914638519287, "learning_rate": 1.54094986080818e-05, "loss": 0.18002243041992189, "step": 16550 }, { "epoch": 0.3387559514981666, "grad_norm": 1.6724166870117188, "learning_rate": 1.540392516658509e-05, "loss": 0.19217681884765625, "step": 16560 }, { "epoch": 0.3389605142708104, "grad_norm": 2.625864267349243, "learning_rate": 1.5398349353045205e-05, "loss": 0.18341751098632814, "step": 16570 }, { "epoch": 0.3391650770434542, "grad_norm": 2.8264200687408447, "learning_rate": 1.5392771169909625e-05, "loss": 0.19124771356582643, "step": 16580 }, { "epoch": 0.3393696398160981, "grad_norm": 1.280354619026184, "learning_rate": 1.5387190619626888e-05, "loss": 0.17405290603637696, "step": 16590 }, { "epoch": 0.3395742025887419, "grad_norm": 1.992074966430664, "learning_rate": 1.538160770464657e-05, "loss": 0.18953933715820312, "step": 16600 }, { "epoch": 0.3397787653613857, "grad_norm": 2.127117395401001, "learning_rate": 1.5376022427419276e-05, "loss": 0.18506537675857543, "step": 16610 }, { "epoch": 0.33998332813402954, "grad_norm": 4.24228048324585, "learning_rate": 1.5370434790396656e-05, "loss": 0.17151886224746704, "step": 16620 }, { "epoch": 0.34018789090667334, "grad_norm": 2.1921486854553223, "learning_rate": 1.5364844796031392e-05, "loss": 0.1723082184791565, "step": 16630 }, { "epoch": 0.34039245367931714, "grad_norm": 5.056528091430664, "learning_rate": 1.5359252446777207e-05, "loss": 0.18922873735427856, "step": 16640 }, { "epoch": 0.340597016451961, "grad_norm": 2.4246323108673096, "learning_rate": 1.5353657745088843e-05, "loss": 0.18129844665527345, "step": 16650 }, { "epoch": 0.3408015792246048, "grad_norm": 3.5564398765563965, "learning_rate": 1.5348060693422093e-05, "loss": 0.15946346521377563, "step": 16660 }, { "epoch": 0.34100614199724866, "grad_norm": 1.1411381959915161, "learning_rate": 1.5342461294233765e-05, "loss": 0.1837529420852661, "step": 16670 }, { "epoch": 0.34121070476989246, "grad_norm": 3.7002687454223633, "learning_rate": 1.533685954998171e-05, "loss": 0.19757431745529175, "step": 16680 }, { "epoch": 0.34141526754253626, "grad_norm": 2.4714198112487793, "learning_rate": 1.5331255463124802e-05, "loss": 0.17487359046936035, "step": 16690 }, { "epoch": 0.3416198303151801, "grad_norm": 2.146658182144165, "learning_rate": 1.5325649036122952e-05, "loss": 0.1746264934539795, "step": 16700 }, { "epoch": 0.3418243930878239, "grad_norm": 3.6259474754333496, "learning_rate": 1.532004027143708e-05, "loss": 0.17651790380477905, "step": 16710 }, { "epoch": 0.3420289558604677, "grad_norm": 1.8567205667495728, "learning_rate": 1.531442917152915e-05, "loss": 0.18043916225433348, "step": 16720 }, { "epoch": 0.3422335186331116, "grad_norm": 5.322687149047852, "learning_rate": 1.5308815738862136e-05, "loss": 0.18952224254608155, "step": 16730 }, { "epoch": 0.3424380814057554, "grad_norm": 1.958450436592102, "learning_rate": 1.530319997590006e-05, "loss": 0.17002781629562377, "step": 16740 }, { "epoch": 0.3426426441783992, "grad_norm": 3.1940133571624756, "learning_rate": 1.5297581885107934e-05, "loss": 0.17073380947113037, "step": 16750 }, { "epoch": 0.34284720695104304, "grad_norm": 3.6893110275268555, "learning_rate": 1.5291961468951818e-05, "loss": 0.17174139022827148, "step": 16760 }, { "epoch": 0.34305176972368684, "grad_norm": 3.5552709102630615, "learning_rate": 1.528633872989878e-05, "loss": 0.17708151340484618, "step": 16770 }, { "epoch": 0.34325633249633064, "grad_norm": 1.8827202320098877, "learning_rate": 1.528071367041691e-05, "loss": 0.2003525972366333, "step": 16780 }, { "epoch": 0.3434608952689745, "grad_norm": 4.953907012939453, "learning_rate": 1.5275086292975324e-05, "loss": 0.17287023067474366, "step": 16790 }, { "epoch": 0.3436654580416183, "grad_norm": 1.4979020357131958, "learning_rate": 1.5269456600044147e-05, "loss": 0.18837230205535888, "step": 16800 }, { "epoch": 0.3438700208142621, "grad_norm": 3.233369827270508, "learning_rate": 1.526382459409452e-05, "loss": 0.16950418949127197, "step": 16810 }, { "epoch": 0.34407458358690596, "grad_norm": 2.2933859825134277, "learning_rate": 1.5258190277598599e-05, "loss": 0.17604224681854247, "step": 16820 }, { "epoch": 0.34427914635954976, "grad_norm": 1.17534601688385, "learning_rate": 1.5252553653029565e-05, "loss": 0.18739335536956786, "step": 16830 }, { "epoch": 0.34448370913219356, "grad_norm": 3.9665844440460205, "learning_rate": 1.5246914722861596e-05, "loss": 0.18096542358398438, "step": 16840 }, { "epoch": 0.3446882719048374, "grad_norm": 4.206308364868164, "learning_rate": 1.5241273489569897e-05, "loss": 0.1823729991912842, "step": 16850 }, { "epoch": 0.3448928346774812, "grad_norm": 1.3262863159179688, "learning_rate": 1.5235629955630673e-05, "loss": 0.19328267574310304, "step": 16860 }, { "epoch": 0.345097397450125, "grad_norm": 1.0777509212493896, "learning_rate": 1.5229984123521142e-05, "loss": 0.18178954124450683, "step": 16870 }, { "epoch": 0.3453019602227689, "grad_norm": 2.5643694400787354, "learning_rate": 1.5224335995719537e-05, "loss": 0.1791430950164795, "step": 16880 }, { "epoch": 0.3455065229954127, "grad_norm": 1.100229263305664, "learning_rate": 1.521868557470509e-05, "loss": 0.1689671277999878, "step": 16890 }, { "epoch": 0.3457110857680565, "grad_norm": 2.8275675773620605, "learning_rate": 1.5213032862958047e-05, "loss": 0.17248647212982177, "step": 16900 }, { "epoch": 0.34591564854070034, "grad_norm": 3.7699522972106934, "learning_rate": 1.5207377862959647e-05, "loss": 0.18687610626220702, "step": 16910 }, { "epoch": 0.34612021131334414, "grad_norm": 3.6203677654266357, "learning_rate": 1.5201720577192149e-05, "loss": 0.1729116916656494, "step": 16920 }, { "epoch": 0.34632477408598794, "grad_norm": 2.800001382827759, "learning_rate": 1.5196061008138808e-05, "loss": 0.18984984159469603, "step": 16930 }, { "epoch": 0.3465293368586318, "grad_norm": 5.217186450958252, "learning_rate": 1.5190399158283878e-05, "loss": 0.17789702415466307, "step": 16940 }, { "epoch": 0.3467338996312756, "grad_norm": 1.808676838874817, "learning_rate": 1.5184735030112622e-05, "loss": 0.17491230964660645, "step": 16950 }, { "epoch": 0.3469384624039194, "grad_norm": 4.7941789627075195, "learning_rate": 1.5179068626111295e-05, "loss": 0.17143133878707886, "step": 16960 }, { "epoch": 0.34714302517656326, "grad_norm": 2.2438971996307373, "learning_rate": 1.5173399948767157e-05, "loss": 0.17945777177810668, "step": 16970 }, { "epoch": 0.34734758794920706, "grad_norm": 5.157105445861816, "learning_rate": 1.5167729000568466e-05, "loss": 0.16949670314788817, "step": 16980 }, { "epoch": 0.34755215072185086, "grad_norm": 2.1599762439727783, "learning_rate": 1.516205578400447e-05, "loss": 0.18228172063827514, "step": 16990 }, { "epoch": 0.3477567134944947, "grad_norm": 1.9320621490478516, "learning_rate": 1.5156380301565418e-05, "loss": 0.1836859703063965, "step": 17000 }, { "epoch": 0.3479612762671385, "grad_norm": 1.9096988439559937, "learning_rate": 1.5150702555742556e-05, "loss": 0.17097439765930175, "step": 17010 }, { "epoch": 0.3481658390397823, "grad_norm": 1.0666979551315308, "learning_rate": 1.5145022549028117e-05, "loss": 0.1650954008102417, "step": 17020 }, { "epoch": 0.3483704018124262, "grad_norm": 2.203930377960205, "learning_rate": 1.5139340283915327e-05, "loss": 0.19042131900787354, "step": 17030 }, { "epoch": 0.34857496458507, "grad_norm": 2.2821083068847656, "learning_rate": 1.5133655762898412e-05, "loss": 0.18865818977355958, "step": 17040 }, { "epoch": 0.3487795273577138, "grad_norm": 2.0024352073669434, "learning_rate": 1.5127968988472583e-05, "loss": 0.17103580236434937, "step": 17050 }, { "epoch": 0.34898409013035764, "grad_norm": 3.782804012298584, "learning_rate": 1.5122279963134032e-05, "loss": 0.18183367252349852, "step": 17060 }, { "epoch": 0.34918865290300144, "grad_norm": 2.3718113899230957, "learning_rate": 1.5116588689379952e-05, "loss": 0.17513854503631593, "step": 17070 }, { "epoch": 0.34939321567564524, "grad_norm": 3.5365331172943115, "learning_rate": 1.5110895169708517e-05, "loss": 0.17538025379180908, "step": 17080 }, { "epoch": 0.3495977784482891, "grad_norm": 1.384932279586792, "learning_rate": 1.5105199406618884e-05, "loss": 0.1661006450653076, "step": 17090 }, { "epoch": 0.3498023412209329, "grad_norm": 1.3015817403793335, "learning_rate": 1.5099501402611204e-05, "loss": 0.1781975507736206, "step": 17100 }, { "epoch": 0.3500069039935767, "grad_norm": 0.9076962471008301, "learning_rate": 1.5093801160186596e-05, "loss": 0.20163393020629883, "step": 17110 }, { "epoch": 0.35021146676622056, "grad_norm": 1.648490309715271, "learning_rate": 1.5088098681847182e-05, "loss": 0.16324384212493898, "step": 17120 }, { "epoch": 0.35041602953886436, "grad_norm": 2.461819648742676, "learning_rate": 1.5082393970096046e-05, "loss": 0.1729576826095581, "step": 17130 }, { "epoch": 0.3506205923115082, "grad_norm": 2.7237343788146973, "learning_rate": 1.5076687027437264e-05, "loss": 0.17065430879592897, "step": 17140 }, { "epoch": 0.350825155084152, "grad_norm": 1.1136269569396973, "learning_rate": 1.5070977856375887e-05, "loss": 0.1694238305091858, "step": 17150 }, { "epoch": 0.3510297178567958, "grad_norm": 0.9695799946784973, "learning_rate": 1.5065266459417953e-05, "loss": 0.18246922492980958, "step": 17160 }, { "epoch": 0.3512342806294397, "grad_norm": 1.771206259727478, "learning_rate": 1.5059552839070457e-05, "loss": 0.17400786876678467, "step": 17170 }, { "epoch": 0.3514388434020835, "grad_norm": 1.798903465270996, "learning_rate": 1.5053836997841388e-05, "loss": 0.17442235946655274, "step": 17180 }, { "epoch": 0.3516434061747273, "grad_norm": 3.3299574851989746, "learning_rate": 1.5048118938239708e-05, "loss": 0.19016013145446778, "step": 17190 }, { "epoch": 0.35184796894737114, "grad_norm": 3.3487868309020996, "learning_rate": 1.5042398662775346e-05, "loss": 0.18197771310806274, "step": 17200 }, { "epoch": 0.35205253172001494, "grad_norm": 2.023590087890625, "learning_rate": 1.5036676173959202e-05, "loss": 0.1666611909866333, "step": 17210 }, { "epoch": 0.35225709449265874, "grad_norm": 1.330245018005371, "learning_rate": 1.5030951474303158e-05, "loss": 0.17529282569885254, "step": 17220 }, { "epoch": 0.3524616572653026, "grad_norm": 2.7755491733551025, "learning_rate": 1.502522456632006e-05, "loss": 0.1903551459312439, "step": 17230 }, { "epoch": 0.3526662200379464, "grad_norm": 1.8206876516342163, "learning_rate": 1.501949545252372e-05, "loss": 0.18976675271987914, "step": 17240 }, { "epoch": 0.3528707828105902, "grad_norm": 1.1245828866958618, "learning_rate": 1.5013764135428925e-05, "loss": 0.1764678478240967, "step": 17250 }, { "epoch": 0.35307534558323406, "grad_norm": 2.175429582595825, "learning_rate": 1.5008030617551428e-05, "loss": 0.1835167407989502, "step": 17260 }, { "epoch": 0.35327990835587786, "grad_norm": 1.4376507997512817, "learning_rate": 1.5002294901407943e-05, "loss": 0.17238123416900636, "step": 17270 }, { "epoch": 0.35348447112852166, "grad_norm": 4.507654190063477, "learning_rate": 1.4996556989516153e-05, "loss": 0.16917877197265624, "step": 17280 }, { "epoch": 0.3536890339011655, "grad_norm": 1.5361945629119873, "learning_rate": 1.4990816884394708e-05, "loss": 0.18215142488479613, "step": 17290 }, { "epoch": 0.3538935966738093, "grad_norm": 1.502471923828125, "learning_rate": 1.498507458856321e-05, "loss": 0.17728912830352783, "step": 17300 }, { "epoch": 0.3540981594464531, "grad_norm": 2.2976605892181396, "learning_rate": 1.4979330104542234e-05, "loss": 0.17793352603912355, "step": 17310 }, { "epoch": 0.354302722219097, "grad_norm": 2.3934521675109863, "learning_rate": 1.4973583434853307e-05, "loss": 0.18035993576049805, "step": 17320 }, { "epoch": 0.3545072849917408, "grad_norm": 3.636798143386841, "learning_rate": 1.4967834582018926e-05, "loss": 0.16977022886276244, "step": 17330 }, { "epoch": 0.3547118477643846, "grad_norm": 1.4174673557281494, "learning_rate": 1.4962083548562532e-05, "loss": 0.1855005741119385, "step": 17340 }, { "epoch": 0.35491641053702844, "grad_norm": 1.70539391040802, "learning_rate": 1.4956330337008538e-05, "loss": 0.16292965412139893, "step": 17350 }, { "epoch": 0.35512097330967224, "grad_norm": 0.9944117069244385, "learning_rate": 1.4950574949882298e-05, "loss": 0.17919728755950928, "step": 17360 }, { "epoch": 0.35532553608231604, "grad_norm": 1.2217155694961548, "learning_rate": 1.4944817389710133e-05, "loss": 0.17576899528503417, "step": 17370 }, { "epoch": 0.3555300988549599, "grad_norm": 4.069941520690918, "learning_rate": 1.4939057659019315e-05, "loss": 0.17903535366058348, "step": 17380 }, { "epoch": 0.3557346616276037, "grad_norm": 2.668233633041382, "learning_rate": 1.4933295760338067e-05, "loss": 0.18717606067657472, "step": 17390 }, { "epoch": 0.3559392244002475, "grad_norm": 1.4516352415084839, "learning_rate": 1.492753169619556e-05, "loss": 0.18468445539474487, "step": 17400 }, { "epoch": 0.35614378717289136, "grad_norm": 1.5021946430206299, "learning_rate": 1.4921765469121924e-05, "loss": 0.18844319581985475, "step": 17410 }, { "epoch": 0.35634834994553516, "grad_norm": 2.2566537857055664, "learning_rate": 1.4915997081648233e-05, "loss": 0.18350725173950194, "step": 17420 }, { "epoch": 0.35655291271817896, "grad_norm": 1.3527357578277588, "learning_rate": 1.4910226536306506e-05, "loss": 0.1800433874130249, "step": 17430 }, { "epoch": 0.3567574754908228, "grad_norm": 1.7555783987045288, "learning_rate": 1.490445383562972e-05, "loss": 0.17183024883270265, "step": 17440 }, { "epoch": 0.3569620382634666, "grad_norm": 3.412698268890381, "learning_rate": 1.489867898215179e-05, "loss": 0.1877611517906189, "step": 17450 }, { "epoch": 0.3571666010361104, "grad_norm": 4.285518169403076, "learning_rate": 1.4892901978407575e-05, "loss": 0.17817444801330568, "step": 17460 }, { "epoch": 0.3573711638087543, "grad_norm": 2.090785503387451, "learning_rate": 1.4887122826932878e-05, "loss": 0.17809969186782837, "step": 17470 }, { "epoch": 0.3575757265813981, "grad_norm": 0.9530273079872131, "learning_rate": 1.4881341530264453e-05, "loss": 0.1750824570655823, "step": 17480 }, { "epoch": 0.3577802893540419, "grad_norm": 0.7797870635986328, "learning_rate": 1.4875558090939987e-05, "loss": 0.1716103434562683, "step": 17490 }, { "epoch": 0.35798485212668574, "grad_norm": 2.540658950805664, "learning_rate": 1.486977251149811e-05, "loss": 0.16901350021362305, "step": 17500 }, { "epoch": 0.35818941489932954, "grad_norm": 0.9062615036964417, "learning_rate": 1.4863984794478392e-05, "loss": 0.16933612823486327, "step": 17510 }, { "epoch": 0.35839397767197334, "grad_norm": 3.104527473449707, "learning_rate": 1.4858194942421342e-05, "loss": 0.18683335781097413, "step": 17520 }, { "epoch": 0.3585985404446172, "grad_norm": 1.255643367767334, "learning_rate": 1.4852402957868404e-05, "loss": 0.1705415964126587, "step": 17530 }, { "epoch": 0.358803103217261, "grad_norm": 1.624967336654663, "learning_rate": 1.4846608843361959e-05, "loss": 0.17730445861816407, "step": 17540 }, { "epoch": 0.3590076659899048, "grad_norm": 2.065911293029785, "learning_rate": 1.484081260144532e-05, "loss": 0.17402877807617187, "step": 17550 }, { "epoch": 0.35921222876254866, "grad_norm": 2.0087339878082275, "learning_rate": 1.4835014234662746e-05, "loss": 0.18862415552139283, "step": 17560 }, { "epoch": 0.35941679153519246, "grad_norm": 2.3166213035583496, "learning_rate": 1.482921374555941e-05, "loss": 0.1754591941833496, "step": 17570 }, { "epoch": 0.35962135430783626, "grad_norm": 5.04824161529541, "learning_rate": 1.4823411136681432e-05, "loss": 0.17891427278518676, "step": 17580 }, { "epoch": 0.3598259170804801, "grad_norm": 1.2800947427749634, "learning_rate": 1.4817606410575852e-05, "loss": 0.17980291843414306, "step": 17590 }, { "epoch": 0.3600304798531239, "grad_norm": 1.8555268049240112, "learning_rate": 1.4811799569790647e-05, "loss": 0.1758336305618286, "step": 17600 }, { "epoch": 0.3602350426257678, "grad_norm": 1.8556932210922241, "learning_rate": 1.4805990616874722e-05, "loss": 0.18026257753372193, "step": 17610 }, { "epoch": 0.3604396053984116, "grad_norm": 1.984663963317871, "learning_rate": 1.48001795543779e-05, "loss": 0.1589486837387085, "step": 17620 }, { "epoch": 0.3606441681710554, "grad_norm": 1.4673222303390503, "learning_rate": 1.4794366384850943e-05, "loss": 0.1752488613128662, "step": 17630 }, { "epoch": 0.36084873094369924, "grad_norm": 1.5678138732910156, "learning_rate": 1.4788551110845525e-05, "loss": 0.18127706050872802, "step": 17640 }, { "epoch": 0.36105329371634304, "grad_norm": 1.4870737791061401, "learning_rate": 1.4782733734914256e-05, "loss": 0.16526129245758056, "step": 17650 }, { "epoch": 0.36125785648898684, "grad_norm": 1.2144293785095215, "learning_rate": 1.4776914259610657e-05, "loss": 0.1705762267112732, "step": 17660 }, { "epoch": 0.3614624192616307, "grad_norm": 1.5280362367630005, "learning_rate": 1.4771092687489184e-05, "loss": 0.18374541997909546, "step": 17670 }, { "epoch": 0.3616669820342745, "grad_norm": 1.8993827104568481, "learning_rate": 1.4765269021105192e-05, "loss": 0.1721095085144043, "step": 17680 }, { "epoch": 0.3618715448069183, "grad_norm": 1.9441465139389038, "learning_rate": 1.4759443263014983e-05, "loss": 0.18488914966583253, "step": 17690 }, { "epoch": 0.36207610757956216, "grad_norm": 2.4148354530334473, "learning_rate": 1.4753615415775754e-05, "loss": 0.17613381147384644, "step": 17700 }, { "epoch": 0.36228067035220596, "grad_norm": 1.5812311172485352, "learning_rate": 1.4747785481945638e-05, "loss": 0.17724502086639404, "step": 17710 }, { "epoch": 0.36248523312484976, "grad_norm": 1.4942600727081299, "learning_rate": 1.474195346408366e-05, "loss": 0.1598062753677368, "step": 17720 }, { "epoch": 0.3626897958974936, "grad_norm": 2.186331272125244, "learning_rate": 1.4736119364749783e-05, "loss": 0.17434836626052858, "step": 17730 }, { "epoch": 0.3628943586701374, "grad_norm": 1.444250226020813, "learning_rate": 1.473028318650487e-05, "loss": 0.16779645681381225, "step": 17740 }, { "epoch": 0.3630989214427812, "grad_norm": 2.496086359024048, "learning_rate": 1.4724444931910708e-05, "loss": 0.17529900074005128, "step": 17750 }, { "epoch": 0.3633034842154251, "grad_norm": 2.550921678543091, "learning_rate": 1.471860460352998e-05, "loss": 0.17463536262512208, "step": 17760 }, { "epoch": 0.3635080469880689, "grad_norm": 1.9803906679153442, "learning_rate": 1.4712762203926293e-05, "loss": 0.18379056453704834, "step": 17770 }, { "epoch": 0.3637126097607127, "grad_norm": 3.394538164138794, "learning_rate": 1.4706917735664155e-05, "loss": 0.18373844623565674, "step": 17780 }, { "epoch": 0.36391717253335654, "grad_norm": 4.167691707611084, "learning_rate": 1.4701071201308982e-05, "loss": 0.17777087688446044, "step": 17790 }, { "epoch": 0.36412173530600034, "grad_norm": 2.1083760261535645, "learning_rate": 1.4695222603427106e-05, "loss": 0.1479318618774414, "step": 17800 }, { "epoch": 0.36432629807864414, "grad_norm": 2.3372747898101807, "learning_rate": 1.4689371944585755e-05, "loss": 0.18811172246932983, "step": 17810 }, { "epoch": 0.364530860851288, "grad_norm": 1.9442590475082397, "learning_rate": 1.4683519227353068e-05, "loss": 0.18263237476348876, "step": 17820 }, { "epoch": 0.3647354236239318, "grad_norm": 1.0086705684661865, "learning_rate": 1.4677664454298081e-05, "loss": 0.1693894386291504, "step": 17830 }, { "epoch": 0.3649399863965756, "grad_norm": 2.2051141262054443, "learning_rate": 1.4671807627990737e-05, "loss": 0.167620849609375, "step": 17840 }, { "epoch": 0.36514454916921946, "grad_norm": 1.685771107673645, "learning_rate": 1.4665948751001882e-05, "loss": 0.17331440448760987, "step": 17850 }, { "epoch": 0.36534911194186326, "grad_norm": 1.874157190322876, "learning_rate": 1.4660087825903255e-05, "loss": 0.16943325996398925, "step": 17860 }, { "epoch": 0.36555367471450706, "grad_norm": 2.9292633533477783, "learning_rate": 1.46542248552675e-05, "loss": 0.17189481258392333, "step": 17870 }, { "epoch": 0.3657582374871509, "grad_norm": 1.751441478729248, "learning_rate": 1.464835984166816e-05, "loss": 0.18028948307037354, "step": 17880 }, { "epoch": 0.3659628002597947, "grad_norm": 3.0336132049560547, "learning_rate": 1.4642492787679672e-05, "loss": 0.17462589740753173, "step": 17890 }, { "epoch": 0.3661673630324385, "grad_norm": 1.351780652999878, "learning_rate": 1.4636623695877363e-05, "loss": 0.16520540714263915, "step": 17900 }, { "epoch": 0.3663719258050824, "grad_norm": 1.9320975542068481, "learning_rate": 1.4630752568837465e-05, "loss": 0.1778683543205261, "step": 17910 }, { "epoch": 0.3665764885777262, "grad_norm": 3.411566734313965, "learning_rate": 1.46248794091371e-05, "loss": 0.17399526834487916, "step": 17920 }, { "epoch": 0.36678105135037, "grad_norm": 4.251964569091797, "learning_rate": 1.4619004219354275e-05, "loss": 0.16818408966064452, "step": 17930 }, { "epoch": 0.36698561412301384, "grad_norm": 1.8415169715881348, "learning_rate": 1.4613127002067902e-05, "loss": 0.16956040859222413, "step": 17940 }, { "epoch": 0.36719017689565764, "grad_norm": 3.7910051345825195, "learning_rate": 1.4607247759857766e-05, "loss": 0.17510766983032228, "step": 17950 }, { "epoch": 0.36739473966830144, "grad_norm": 3.123573064804077, "learning_rate": 1.4601366495304558e-05, "loss": 0.17042529582977295, "step": 17960 }, { "epoch": 0.3675993024409453, "grad_norm": 3.1084578037261963, "learning_rate": 1.4595483210989842e-05, "loss": 0.1917470335960388, "step": 17970 }, { "epoch": 0.3678038652135891, "grad_norm": 0.8238036036491394, "learning_rate": 1.4589597909496078e-05, "loss": 0.17731770277023315, "step": 17980 }, { "epoch": 0.3680084279862329, "grad_norm": 1.7936991453170776, "learning_rate": 1.458371059340661e-05, "loss": 0.1625174641609192, "step": 17990 }, { "epoch": 0.36821299075887676, "grad_norm": 1.6407138109207153, "learning_rate": 1.4577821265305663e-05, "loss": 0.17316849231719972, "step": 18000 }, { "epoch": 0.36841755353152056, "grad_norm": 1.406689167022705, "learning_rate": 1.4571929927778346e-05, "loss": 0.18254311084747316, "step": 18010 }, { "epoch": 0.36862211630416436, "grad_norm": 2.1043272018432617, "learning_rate": 1.456603658341065e-05, "loss": 0.17160028219223022, "step": 18020 }, { "epoch": 0.3688266790768082, "grad_norm": 1.4400930404663086, "learning_rate": 1.456014123478945e-05, "loss": 0.1626047372817993, "step": 18030 }, { "epoch": 0.369031241849452, "grad_norm": 1.8056172132492065, "learning_rate": 1.4554243884502501e-05, "loss": 0.1805182933807373, "step": 18040 }, { "epoch": 0.3692358046220959, "grad_norm": 0.9717869162559509, "learning_rate": 1.454834453513843e-05, "loss": 0.18034257888793945, "step": 18050 }, { "epoch": 0.3694403673947397, "grad_norm": 1.0743391513824463, "learning_rate": 1.4542443189286743e-05, "loss": 0.18988187313079835, "step": 18060 }, { "epoch": 0.3696449301673835, "grad_norm": 0.7944437861442566, "learning_rate": 1.4536539849537836e-05, "loss": 0.17247719764709474, "step": 18070 }, { "epoch": 0.36984949294002734, "grad_norm": 3.2354094982147217, "learning_rate": 1.453063451848296e-05, "loss": 0.17776409387588502, "step": 18080 }, { "epoch": 0.37005405571267114, "grad_norm": 1.3647960424423218, "learning_rate": 1.4524727198714252e-05, "loss": 0.15969376564025878, "step": 18090 }, { "epoch": 0.37025861848531494, "grad_norm": 1.5882233381271362, "learning_rate": 1.4518817892824718e-05, "loss": 0.1579761505126953, "step": 18100 }, { "epoch": 0.3704631812579588, "grad_norm": 1.289442777633667, "learning_rate": 1.451290660340824e-05, "loss": 0.17666089534759521, "step": 18110 }, { "epoch": 0.3706677440306026, "grad_norm": 2.1477396488189697, "learning_rate": 1.4506993333059565e-05, "loss": 0.1741098165512085, "step": 18120 }, { "epoch": 0.3708723068032464, "grad_norm": 1.5364314317703247, "learning_rate": 1.4501078084374316e-05, "loss": 0.17452455759048463, "step": 18130 }, { "epoch": 0.37107686957589026, "grad_norm": 2.690429925918579, "learning_rate": 1.4495160859948975e-05, "loss": 0.1734003782272339, "step": 18140 }, { "epoch": 0.37128143234853406, "grad_norm": 2.303323745727539, "learning_rate": 1.4489241662380904e-05, "loss": 0.17092127799987794, "step": 18150 }, { "epoch": 0.37148599512117786, "grad_norm": 1.1932915449142456, "learning_rate": 1.4483320494268318e-05, "loss": 0.1698566198348999, "step": 18160 }, { "epoch": 0.3716905578938217, "grad_norm": 3.0367681980133057, "learning_rate": 1.4477397358210306e-05, "loss": 0.1584364652633667, "step": 18170 }, { "epoch": 0.3718951206664655, "grad_norm": 1.5594199895858765, "learning_rate": 1.4471472256806815e-05, "loss": 0.1629232406616211, "step": 18180 }, { "epoch": 0.3720996834391093, "grad_norm": 3.4741640090942383, "learning_rate": 1.4465545192658662e-05, "loss": 0.1676548480987549, "step": 18190 }, { "epoch": 0.3723042462117532, "grad_norm": 3.340463399887085, "learning_rate": 1.445961616836752e-05, "loss": 0.18497815132141113, "step": 18200 }, { "epoch": 0.372508808984397, "grad_norm": 2.129289388656616, "learning_rate": 1.4453685186535922e-05, "loss": 0.1624639391899109, "step": 18210 }, { "epoch": 0.3727133717570408, "grad_norm": 1.6400822401046753, "learning_rate": 1.4447752249767261e-05, "loss": 0.18459267616271974, "step": 18220 }, { "epoch": 0.37291793452968464, "grad_norm": 2.504424810409546, "learning_rate": 1.4441817360665793e-05, "loss": 0.17114092111587526, "step": 18230 }, { "epoch": 0.37312249730232844, "grad_norm": 1.530019998550415, "learning_rate": 1.4435880521836621e-05, "loss": 0.1802661895751953, "step": 18240 }, { "epoch": 0.37332706007497224, "grad_norm": 2.0332133769989014, "learning_rate": 1.4429941735885714e-05, "loss": 0.18192977905273439, "step": 18250 }, { "epoch": 0.3735316228476161, "grad_norm": 2.187422037124634, "learning_rate": 1.442400100541989e-05, "loss": 0.17275114059448243, "step": 18260 }, { "epoch": 0.3737361856202599, "grad_norm": 2.2351572513580322, "learning_rate": 1.4418058333046818e-05, "loss": 0.15973042249679564, "step": 18270 }, { "epoch": 0.3739407483929037, "grad_norm": 3.2247657775878906, "learning_rate": 1.4412113721375031e-05, "loss": 0.16664443016052247, "step": 18280 }, { "epoch": 0.37414531116554756, "grad_norm": 1.1901322603225708, "learning_rate": 1.4406167173013899e-05, "loss": 0.17505725622177123, "step": 18290 }, { "epoch": 0.37434987393819136, "grad_norm": 0.8570743203163147, "learning_rate": 1.4400218690573655e-05, "loss": 0.18142186403274535, "step": 18300 }, { "epoch": 0.37455443671083516, "grad_norm": 5.566702842712402, "learning_rate": 1.4394268276665368e-05, "loss": 0.17422821521759033, "step": 18310 }, { "epoch": 0.374758999483479, "grad_norm": 1.6620440483093262, "learning_rate": 1.4388315933900967e-05, "loss": 0.17670631408691406, "step": 18320 }, { "epoch": 0.3749635622561228, "grad_norm": 1.9589062929153442, "learning_rate": 1.4382361664893214e-05, "loss": 0.164029860496521, "step": 18330 }, { "epoch": 0.3751681250287666, "grad_norm": 1.295175552368164, "learning_rate": 1.4376405472255734e-05, "loss": 0.17500903606414794, "step": 18340 }, { "epoch": 0.3753726878014105, "grad_norm": 2.4734978675842285, "learning_rate": 1.4370447358602986e-05, "loss": 0.1715761661529541, "step": 18350 }, { "epoch": 0.3755772505740543, "grad_norm": 1.4860824346542358, "learning_rate": 1.4364487326550265e-05, "loss": 0.17762467861175538, "step": 18360 }, { "epoch": 0.3757818133466981, "grad_norm": 1.7690300941467285, "learning_rate": 1.4358525378713726e-05, "loss": 0.17672560214996338, "step": 18370 }, { "epoch": 0.37598637611934194, "grad_norm": 2.2470109462738037, "learning_rate": 1.4352561517710348e-05, "loss": 0.1747344970703125, "step": 18380 }, { "epoch": 0.37619093889198574, "grad_norm": 2.194199800491333, "learning_rate": 1.434659574615796e-05, "loss": 0.1784507155418396, "step": 18390 }, { "epoch": 0.37639550166462954, "grad_norm": 1.6204451322555542, "learning_rate": 1.4340628066675226e-05, "loss": 0.1690070390701294, "step": 18400 }, { "epoch": 0.3766000644372734, "grad_norm": 2.025117874145508, "learning_rate": 1.4334658481881652e-05, "loss": 0.18640669584274291, "step": 18410 }, { "epoch": 0.3768046272099172, "grad_norm": 2.219336986541748, "learning_rate": 1.4328686994397567e-05, "loss": 0.16574862003326415, "step": 18420 }, { "epoch": 0.377009189982561, "grad_norm": 2.5014288425445557, "learning_rate": 1.4322713606844154e-05, "loss": 0.17505578994750975, "step": 18430 }, { "epoch": 0.37721375275520486, "grad_norm": 1.7735222578048706, "learning_rate": 1.431673832184342e-05, "loss": 0.16299951076507568, "step": 18440 }, { "epoch": 0.37741831552784866, "grad_norm": 3.507171154022217, "learning_rate": 1.4310761142018201e-05, "loss": 0.17959051132202147, "step": 18450 }, { "epoch": 0.37762287830049246, "grad_norm": 1.565077304840088, "learning_rate": 1.4304782069992169e-05, "loss": 0.17090387344360353, "step": 18460 }, { "epoch": 0.3778274410731363, "grad_norm": 1.5147217512130737, "learning_rate": 1.429880110838983e-05, "loss": 0.16057121753692627, "step": 18470 }, { "epoch": 0.3780320038457801, "grad_norm": 1.2937675714492798, "learning_rate": 1.4292818259836515e-05, "loss": 0.16463303565979004, "step": 18480 }, { "epoch": 0.3782365666184239, "grad_norm": 2.123530387878418, "learning_rate": 1.4286833526958387e-05, "loss": 0.17106192111968993, "step": 18490 }, { "epoch": 0.3784411293910678, "grad_norm": 1.4837032556533813, "learning_rate": 1.4280846912382428e-05, "loss": 0.16939065456390381, "step": 18500 }, { "epoch": 0.3786456921637116, "grad_norm": 1.0977190732955933, "learning_rate": 1.427485841873646e-05, "loss": 0.1811454176902771, "step": 18510 }, { "epoch": 0.37885025493635544, "grad_norm": 1.2331700325012207, "learning_rate": 1.4268868048649114e-05, "loss": 0.16859209537506104, "step": 18520 }, { "epoch": 0.37905481770899924, "grad_norm": 2.226064682006836, "learning_rate": 1.4262875804749852e-05, "loss": 0.17638168334960938, "step": 18530 }, { "epoch": 0.37925938048164304, "grad_norm": 2.61299204826355, "learning_rate": 1.4256881689668964e-05, "loss": 0.16620750427246095, "step": 18540 }, { "epoch": 0.3794639432542869, "grad_norm": 1.9569380283355713, "learning_rate": 1.4250885706037556e-05, "loss": 0.1738607406616211, "step": 18550 }, { "epoch": 0.3796685060269307, "grad_norm": 2.3040425777435303, "learning_rate": 1.4244887856487549e-05, "loss": 0.17647358179092407, "step": 18560 }, { "epoch": 0.3798730687995745, "grad_norm": 1.294469952583313, "learning_rate": 1.4238888143651688e-05, "loss": 0.1621049165725708, "step": 18570 }, { "epoch": 0.38007763157221836, "grad_norm": 2.1563596725463867, "learning_rate": 1.4232886570163542e-05, "loss": 0.17970597743988037, "step": 18580 }, { "epoch": 0.38028219434486216, "grad_norm": 4.818857192993164, "learning_rate": 1.4226883138657489e-05, "loss": 0.16289138793945312, "step": 18590 }, { "epoch": 0.38048675711750596, "grad_norm": 4.538371562957764, "learning_rate": 1.4220877851768722e-05, "loss": 0.17791160345077514, "step": 18600 }, { "epoch": 0.3806913198901498, "grad_norm": 1.286694049835205, "learning_rate": 1.4214870712133248e-05, "loss": 0.1709395170211792, "step": 18610 }, { "epoch": 0.3808958826627936, "grad_norm": 2.8194665908813477, "learning_rate": 1.4208861722387899e-05, "loss": 0.1742150068283081, "step": 18620 }, { "epoch": 0.3811004454354374, "grad_norm": 1.475781798362732, "learning_rate": 1.4202850885170305e-05, "loss": 0.1856175422668457, "step": 18630 }, { "epoch": 0.3813050082080813, "grad_norm": 1.7942453622817993, "learning_rate": 1.419683820311891e-05, "loss": 0.1673583507537842, "step": 18640 }, { "epoch": 0.3815095709807251, "grad_norm": 2.98897123336792, "learning_rate": 1.4190823678872971e-05, "loss": 0.16703670024871825, "step": 18650 }, { "epoch": 0.3817141337533689, "grad_norm": 1.4324926137924194, "learning_rate": 1.4184807315072557e-05, "loss": 0.18050065040588378, "step": 18660 }, { "epoch": 0.38191869652601274, "grad_norm": 3.288813352584839, "learning_rate": 1.4178789114358537e-05, "loss": 0.16796295642852782, "step": 18670 }, { "epoch": 0.38212325929865654, "grad_norm": 2.7898406982421875, "learning_rate": 1.4172769079372589e-05, "loss": 0.17066867351531984, "step": 18680 }, { "epoch": 0.38232782207130034, "grad_norm": 4.0410847663879395, "learning_rate": 1.4166747212757193e-05, "loss": 0.16910786628723146, "step": 18690 }, { "epoch": 0.3825323848439442, "grad_norm": 1.460898518562317, "learning_rate": 1.4160723517155643e-05, "loss": 0.1819972038269043, "step": 18700 }, { "epoch": 0.382736947616588, "grad_norm": 2.500519037246704, "learning_rate": 1.4154697995212027e-05, "loss": 0.16976075172424315, "step": 18710 }, { "epoch": 0.3829415103892318, "grad_norm": 1.5660074949264526, "learning_rate": 1.4148670649571234e-05, "loss": 0.1694591999053955, "step": 18720 }, { "epoch": 0.38314607316187566, "grad_norm": 2.2071375846862793, "learning_rate": 1.4142641482878961e-05, "loss": 0.16457256078720092, "step": 18730 }, { "epoch": 0.38335063593451946, "grad_norm": 3.1832447052001953, "learning_rate": 1.4136610497781697e-05, "loss": 0.17619725465774536, "step": 18740 }, { "epoch": 0.38355519870716326, "grad_norm": 2.446082353591919, "learning_rate": 1.4130577696926732e-05, "loss": 0.16821545362472534, "step": 18750 }, { "epoch": 0.3837597614798071, "grad_norm": 2.385406255722046, "learning_rate": 1.4124543082962155e-05, "loss": 0.16032307147979735, "step": 18760 }, { "epoch": 0.3839643242524509, "grad_norm": 2.135983467102051, "learning_rate": 1.4118506658536846e-05, "loss": 0.17031543254852294, "step": 18770 }, { "epoch": 0.3841688870250947, "grad_norm": 3.5508434772491455, "learning_rate": 1.4112468426300485e-05, "loss": 0.1806311011314392, "step": 18780 }, { "epoch": 0.3843734497977386, "grad_norm": 1.6983802318572998, "learning_rate": 1.4106428388903544e-05, "loss": 0.14821429252624513, "step": 18790 }, { "epoch": 0.3845780125703824, "grad_norm": 3.1783549785614014, "learning_rate": 1.4100386548997281e-05, "loss": 0.18367080688476561, "step": 18800 }, { "epoch": 0.3847825753430262, "grad_norm": 4.604497909545898, "learning_rate": 1.4094342909233756e-05, "loss": 0.1627212643623352, "step": 18810 }, { "epoch": 0.38498713811567004, "grad_norm": 1.2977852821350098, "learning_rate": 1.4088297472265812e-05, "loss": 0.17519264221191405, "step": 18820 }, { "epoch": 0.38519170088831384, "grad_norm": 2.6955862045288086, "learning_rate": 1.4082250240747084e-05, "loss": 0.16620321273803712, "step": 18830 }, { "epoch": 0.38539626366095764, "grad_norm": 2.1139895915985107, "learning_rate": 1.407620121733199e-05, "loss": 0.16475927829742432, "step": 18840 }, { "epoch": 0.3856008264336015, "grad_norm": 6.860621452331543, "learning_rate": 1.4070150404675746e-05, "loss": 0.17833240032196046, "step": 18850 }, { "epoch": 0.3858053892062453, "grad_norm": 1.1892517805099487, "learning_rate": 1.406409780543433e-05, "loss": 0.17594873905181885, "step": 18860 }, { "epoch": 0.3860099519788891, "grad_norm": 1.8429553508758545, "learning_rate": 1.4058043422264538e-05, "loss": 0.1907630205154419, "step": 18870 }, { "epoch": 0.38621451475153296, "grad_norm": 2.45159912109375, "learning_rate": 1.4051987257823915e-05, "loss": 0.1699777364730835, "step": 18880 }, { "epoch": 0.38641907752417676, "grad_norm": 4.13884162902832, "learning_rate": 1.4045929314770813e-05, "loss": 0.17674405574798585, "step": 18890 }, { "epoch": 0.38662364029682056, "grad_norm": 2.244272232055664, "learning_rate": 1.403986959576435e-05, "loss": 0.17576609849929808, "step": 18900 }, { "epoch": 0.3868282030694644, "grad_norm": 4.929913520812988, "learning_rate": 1.4033808103464432e-05, "loss": 0.17309898138046265, "step": 18910 }, { "epoch": 0.3870327658421082, "grad_norm": 1.672104001045227, "learning_rate": 1.4027744840531739e-05, "loss": 0.18729504346847534, "step": 18920 }, { "epoch": 0.387237328614752, "grad_norm": 1.2562761306762695, "learning_rate": 1.402167980962773e-05, "loss": 0.17513368129730225, "step": 18930 }, { "epoch": 0.3874418913873959, "grad_norm": 1.0230847597122192, "learning_rate": 1.4015613013414638e-05, "loss": 0.16837217807769775, "step": 18940 }, { "epoch": 0.3876464541600397, "grad_norm": 1.562605619430542, "learning_rate": 1.4009544454555468e-05, "loss": 0.1617227554321289, "step": 18950 }, { "epoch": 0.3878510169326835, "grad_norm": 2.2180604934692383, "learning_rate": 1.4003474135714015e-05, "loss": 0.17749178409576416, "step": 18960 }, { "epoch": 0.38805557970532734, "grad_norm": 2.413090467453003, "learning_rate": 1.3997402059554823e-05, "loss": 0.1673878788948059, "step": 18970 }, { "epoch": 0.38826014247797114, "grad_norm": 1.5147838592529297, "learning_rate": 1.3991328228743224e-05, "loss": 0.1616837501525879, "step": 18980 }, { "epoch": 0.388464705250615, "grad_norm": 1.3911709785461426, "learning_rate": 1.3985252645945311e-05, "loss": 0.17092444896697997, "step": 18990 }, { "epoch": 0.3886692680232588, "grad_norm": 1.6074141263961792, "learning_rate": 1.397917531382796e-05, "loss": 0.16694045066833496, "step": 19000 }, { "epoch": 0.3888738307959026, "grad_norm": 16.554750442504883, "learning_rate": 1.3973096235058791e-05, "loss": 0.54737548828125, "step": 19010 }, { "epoch": 0.38907839356854645, "grad_norm": 2.130061388015747, "learning_rate": 1.3967015412306216e-05, "loss": 0.27931962013244627, "step": 19020 }, { "epoch": 0.38928295634119026, "grad_norm": 4.904281139373779, "learning_rate": 1.3960932848239395e-05, "loss": 0.19678382873535155, "step": 19030 }, { "epoch": 0.38948751911383406, "grad_norm": 1.4381903409957886, "learning_rate": 1.3954848545528261e-05, "loss": 0.2033824920654297, "step": 19040 }, { "epoch": 0.3896920818864779, "grad_norm": 1.9050090312957764, "learning_rate": 1.3948762506843505e-05, "loss": 0.1919124722480774, "step": 19050 }, { "epoch": 0.3898966446591217, "grad_norm": 3.916109323501587, "learning_rate": 1.3942674734856588e-05, "loss": 0.1683865189552307, "step": 19060 }, { "epoch": 0.3901012074317655, "grad_norm": 2.164775848388672, "learning_rate": 1.393658523223972e-05, "loss": 0.19087980985641478, "step": 19070 }, { "epoch": 0.3903057702044094, "grad_norm": 3.7924115657806396, "learning_rate": 1.393049400166588e-05, "loss": 0.18551368713378907, "step": 19080 }, { "epoch": 0.3905103329770532, "grad_norm": 1.7565501928329468, "learning_rate": 1.3924401045808807e-05, "loss": 0.1792805314064026, "step": 19090 }, { "epoch": 0.390714895749697, "grad_norm": 2.492790460586548, "learning_rate": 1.3918306367342985e-05, "loss": 0.18102235794067384, "step": 19100 }, { "epoch": 0.39091945852234083, "grad_norm": 4.2847771644592285, "learning_rate": 1.391220996894367e-05, "loss": 0.18146963119506837, "step": 19110 }, { "epoch": 0.39112402129498464, "grad_norm": 2.0022900104522705, "learning_rate": 1.3906111853286856e-05, "loss": 0.17349159717559814, "step": 19120 }, { "epoch": 0.39132858406762844, "grad_norm": 1.1436779499053955, "learning_rate": 1.390001202304931e-05, "loss": 0.1698472261428833, "step": 19130 }, { "epoch": 0.3915331468402723, "grad_norm": 1.4344422817230225, "learning_rate": 1.3893910480908534e-05, "loss": 0.17295589447021484, "step": 19140 }, { "epoch": 0.3917377096129161, "grad_norm": 4.172408103942871, "learning_rate": 1.3887807229542793e-05, "loss": 0.17159250974655152, "step": 19150 }, { "epoch": 0.3919422723855599, "grad_norm": 1.7289539575576782, "learning_rate": 1.3881702271631094e-05, "loss": 0.1656977653503418, "step": 19160 }, { "epoch": 0.39214683515820375, "grad_norm": 1.3298696279525757, "learning_rate": 1.3875595609853201e-05, "loss": 0.1698831796646118, "step": 19170 }, { "epoch": 0.39235139793084756, "grad_norm": 2.7541425228118896, "learning_rate": 1.386948724688962e-05, "loss": 0.16068639755249023, "step": 19180 }, { "epoch": 0.39255596070349136, "grad_norm": 2.1675069332122803, "learning_rate": 1.386337718542161e-05, "loss": 0.16799209117889405, "step": 19190 }, { "epoch": 0.3927605234761352, "grad_norm": 2.3784046173095703, "learning_rate": 1.3857265428131163e-05, "loss": 0.17992007732391357, "step": 19200 }, { "epoch": 0.392965086248779, "grad_norm": 2.0638787746429443, "learning_rate": 1.3851151977701033e-05, "loss": 0.1749492406845093, "step": 19210 }, { "epoch": 0.3931696490214228, "grad_norm": 4.28890323638916, "learning_rate": 1.3845036836814703e-05, "loss": 0.16361360549926757, "step": 19220 }, { "epoch": 0.3933742117940667, "grad_norm": 1.9938527345657349, "learning_rate": 1.3838920008156404e-05, "loss": 0.1575866460800171, "step": 19230 }, { "epoch": 0.3935787745667105, "grad_norm": 1.4261621236801147, "learning_rate": 1.3832801494411107e-05, "loss": 0.19359078407287597, "step": 19240 }, { "epoch": 0.3937833373393543, "grad_norm": 0.9745628833770752, "learning_rate": 1.3826681298264521e-05, "loss": 0.17273160219192504, "step": 19250 }, { "epoch": 0.39398790011199813, "grad_norm": 2.6194067001342773, "learning_rate": 1.3820559422403099e-05, "loss": 0.1555500864982605, "step": 19260 }, { "epoch": 0.39419246288464194, "grad_norm": 4.1294026374816895, "learning_rate": 1.3814435869514022e-05, "loss": 0.15089370012283326, "step": 19270 }, { "epoch": 0.39439702565728574, "grad_norm": 3.896780490875244, "learning_rate": 1.3808310642285218e-05, "loss": 0.1860931396484375, "step": 19280 }, { "epoch": 0.3946015884299296, "grad_norm": 0.8517377376556396, "learning_rate": 1.3802183743405342e-05, "loss": 0.1634660243988037, "step": 19290 }, { "epoch": 0.3948061512025734, "grad_norm": 1.7377727031707764, "learning_rate": 1.3796055175563782e-05, "loss": 0.17002584934234619, "step": 19300 }, { "epoch": 0.3950107139752172, "grad_norm": 1.1440916061401367, "learning_rate": 1.3789924941450665e-05, "loss": 0.15864164829254152, "step": 19310 }, { "epoch": 0.39521527674786106, "grad_norm": 4.356170654296875, "learning_rate": 1.378379304375685e-05, "loss": 0.1703243374824524, "step": 19320 }, { "epoch": 0.39541983952050486, "grad_norm": 1.688089370727539, "learning_rate": 1.3777659485173916e-05, "loss": 0.15996141433715821, "step": 19330 }, { "epoch": 0.39562440229314866, "grad_norm": 1.172156810760498, "learning_rate": 1.377152426839418e-05, "loss": 0.16092350482940673, "step": 19340 }, { "epoch": 0.3958289650657925, "grad_norm": 0.9783849716186523, "learning_rate": 1.3765387396110682e-05, "loss": 0.1497013211250305, "step": 19350 }, { "epoch": 0.3960335278384363, "grad_norm": 0.8063821792602539, "learning_rate": 1.3759248871017192e-05, "loss": 0.15781840085983276, "step": 19360 }, { "epoch": 0.3962380906110801, "grad_norm": 1.9794878959655762, "learning_rate": 1.375310869580821e-05, "loss": 0.1700427770614624, "step": 19370 }, { "epoch": 0.396442653383724, "grad_norm": 1.0535688400268555, "learning_rate": 1.3746966873178948e-05, "loss": 0.16781210899353027, "step": 19380 }, { "epoch": 0.3966472161563678, "grad_norm": 4.402553558349609, "learning_rate": 1.374082340582535e-05, "loss": 0.17199262380599975, "step": 19390 }, { "epoch": 0.3968517789290116, "grad_norm": 0.8645378351211548, "learning_rate": 1.3734678296444082e-05, "loss": 0.16543503999710082, "step": 19400 }, { "epoch": 0.39705634170165544, "grad_norm": 2.686694860458374, "learning_rate": 1.3728531547732526e-05, "loss": 0.16354167461395264, "step": 19410 }, { "epoch": 0.39726090447429924, "grad_norm": 4.611863613128662, "learning_rate": 1.3722383162388785e-05, "loss": 0.1643027663230896, "step": 19420 }, { "epoch": 0.39746546724694304, "grad_norm": 1.6757291555404663, "learning_rate": 1.3716233143111686e-05, "loss": 0.1679433226585388, "step": 19430 }, { "epoch": 0.3976700300195869, "grad_norm": 2.60752272605896, "learning_rate": 1.3710081492600766e-05, "loss": 0.16190557479858397, "step": 19440 }, { "epoch": 0.3978745927922307, "grad_norm": 2.1790246963500977, "learning_rate": 1.3703928213556284e-05, "loss": 0.15416685342788697, "step": 19450 }, { "epoch": 0.39807915556487455, "grad_norm": 1.567936658859253, "learning_rate": 1.3697773308679203e-05, "loss": 0.16877565383911133, "step": 19460 }, { "epoch": 0.39828371833751836, "grad_norm": 2.1676018238067627, "learning_rate": 1.3691616780671216e-05, "loss": 0.17276580333709718, "step": 19470 }, { "epoch": 0.39848828111016216, "grad_norm": 2.0197465419769287, "learning_rate": 1.3685458632234716e-05, "loss": 0.16304469108581543, "step": 19480 }, { "epoch": 0.398692843882806, "grad_norm": 3.1295652389526367, "learning_rate": 1.3679298866072812e-05, "loss": 0.15975346565246581, "step": 19490 }, { "epoch": 0.3988974066554498, "grad_norm": 2.482905864715576, "learning_rate": 1.3673137484889319e-05, "loss": 0.1608902096748352, "step": 19500 }, { "epoch": 0.3991019694280936, "grad_norm": 1.5870181322097778, "learning_rate": 1.366697449138877e-05, "loss": 0.16182432174682618, "step": 19510 }, { "epoch": 0.3993065322007375, "grad_norm": 3.367802143096924, "learning_rate": 1.3660809888276394e-05, "loss": 0.16933585405349733, "step": 19520 }, { "epoch": 0.3995110949733813, "grad_norm": 2.5711381435394287, "learning_rate": 1.3654643678258137e-05, "loss": 0.17068209648132324, "step": 19530 }, { "epoch": 0.3997156577460251, "grad_norm": 2.0820271968841553, "learning_rate": 1.3648475864040642e-05, "loss": 0.16281903982162477, "step": 19540 }, { "epoch": 0.39992022051866893, "grad_norm": 1.9431060552597046, "learning_rate": 1.3642306448331266e-05, "loss": 0.17131199836730956, "step": 19550 }, { "epoch": 0.40012478329131274, "grad_norm": 2.7426064014434814, "learning_rate": 1.3636135433838053e-05, "loss": 0.16550731658935547, "step": 19560 }, { "epoch": 0.40032934606395654, "grad_norm": 1.568113923072815, "learning_rate": 1.362996282326977e-05, "loss": 0.16900469064712526, "step": 19570 }, { "epoch": 0.4005339088366004, "grad_norm": 2.4076085090637207, "learning_rate": 1.362378861933586e-05, "loss": 0.16235095262527466, "step": 19580 }, { "epoch": 0.4007384716092442, "grad_norm": 1.6353840827941895, "learning_rate": 1.3617612824746493e-05, "loss": 0.1719159245491028, "step": 19590 }, { "epoch": 0.400943034381888, "grad_norm": 4.544003963470459, "learning_rate": 1.3611435442212512e-05, "loss": 0.15931738615036012, "step": 19600 }, { "epoch": 0.40114759715453185, "grad_norm": 1.227373480796814, "learning_rate": 1.3605256474445474e-05, "loss": 0.17688945531845093, "step": 19610 }, { "epoch": 0.40135215992717566, "grad_norm": 1.9189575910568237, "learning_rate": 1.3599075924157621e-05, "loss": 0.15823280811309814, "step": 19620 }, { "epoch": 0.40155672269981946, "grad_norm": 3.3339083194732666, "learning_rate": 1.3592893794061901e-05, "loss": 0.17848891019821167, "step": 19630 }, { "epoch": 0.4017612854724633, "grad_norm": 2.0192923545837402, "learning_rate": 1.3586710086871944e-05, "loss": 0.16777234077453612, "step": 19640 }, { "epoch": 0.4019658482451071, "grad_norm": 2.211348056793213, "learning_rate": 1.3580524805302076e-05, "loss": 0.1762405276298523, "step": 19650 }, { "epoch": 0.4021704110177509, "grad_norm": 2.41453218460083, "learning_rate": 1.357433795206732e-05, "loss": 0.17689192295074463, "step": 19660 }, { "epoch": 0.4023749737903948, "grad_norm": 1.6622400283813477, "learning_rate": 1.3568149529883377e-05, "loss": 0.15154082775115968, "step": 19670 }, { "epoch": 0.4025795365630386, "grad_norm": 1.0737769603729248, "learning_rate": 1.3561959541466651e-05, "loss": 0.16609463691711426, "step": 19680 }, { "epoch": 0.4027840993356824, "grad_norm": 2.576298952102661, "learning_rate": 1.355576798953422e-05, "loss": 0.1587965726852417, "step": 19690 }, { "epoch": 0.40298866210832623, "grad_norm": 2.574516773223877, "learning_rate": 1.3549574876803864e-05, "loss": 0.15906786918640137, "step": 19700 }, { "epoch": 0.40319322488097004, "grad_norm": 2.6246845722198486, "learning_rate": 1.3543380205994025e-05, "loss": 0.17712409496307374, "step": 19710 }, { "epoch": 0.40339778765361384, "grad_norm": 1.2764484882354736, "learning_rate": 1.3537183979823856e-05, "loss": 0.15801727771759033, "step": 19720 }, { "epoch": 0.4036023504262577, "grad_norm": 1.101374626159668, "learning_rate": 1.353098620101317e-05, "loss": 0.1694082260131836, "step": 19730 }, { "epoch": 0.4038069131989015, "grad_norm": 3.087960958480835, "learning_rate": 1.3524786872282474e-05, "loss": 0.1658942937850952, "step": 19740 }, { "epoch": 0.4040114759715453, "grad_norm": 2.2574052810668945, "learning_rate": 1.351858599635295e-05, "loss": 0.1706781268119812, "step": 19750 }, { "epoch": 0.40421603874418915, "grad_norm": 4.793893337249756, "learning_rate": 1.3512383575946468e-05, "loss": 0.15653555393218993, "step": 19760 }, { "epoch": 0.40442060151683296, "grad_norm": 1.197415828704834, "learning_rate": 1.3506179613785561e-05, "loss": 0.16576032638549804, "step": 19770 }, { "epoch": 0.40462516428947676, "grad_norm": 1.9602288007736206, "learning_rate": 1.3499974112593455e-05, "loss": 0.16481542587280273, "step": 19780 }, { "epoch": 0.4048297270621206, "grad_norm": 1.730912685394287, "learning_rate": 1.3493767075094034e-05, "loss": 0.17183563709259034, "step": 19790 }, { "epoch": 0.4050342898347644, "grad_norm": 2.3069260120391846, "learning_rate": 1.348755850401187e-05, "loss": 0.17000975608825683, "step": 19800 }, { "epoch": 0.4052388526074082, "grad_norm": 2.797097682952881, "learning_rate": 1.3481348402072212e-05, "loss": 0.16683008670806884, "step": 19810 }, { "epoch": 0.4054434153800521, "grad_norm": 2.0542283058166504, "learning_rate": 1.3475136772000959e-05, "loss": 0.16581114530563354, "step": 19820 }, { "epoch": 0.4056479781526959, "grad_norm": 2.79704213142395, "learning_rate": 1.3468923616524705e-05, "loss": 0.15973172187805176, "step": 19830 }, { "epoch": 0.4058525409253397, "grad_norm": 1.1830729246139526, "learning_rate": 1.3462708938370701e-05, "loss": 0.17225452661514282, "step": 19840 }, { "epoch": 0.40605710369798353, "grad_norm": 0.9309203624725342, "learning_rate": 1.3456492740266867e-05, "loss": 0.15443507432937623, "step": 19850 }, { "epoch": 0.40626166647062734, "grad_norm": 1.7387902736663818, "learning_rate": 1.3450275024941792e-05, "loss": 0.1607633113861084, "step": 19860 }, { "epoch": 0.40646622924327114, "grad_norm": 2.438342809677124, "learning_rate": 1.3444055795124733e-05, "loss": 0.18181066513061522, "step": 19870 }, { "epoch": 0.406670792015915, "grad_norm": 3.0286343097686768, "learning_rate": 1.3437835053545607e-05, "loss": 0.15720784664154053, "step": 19880 }, { "epoch": 0.4068753547885588, "grad_norm": 1.5024635791778564, "learning_rate": 1.3431612802935e-05, "loss": 0.17262532711029052, "step": 19890 }, { "epoch": 0.4070799175612026, "grad_norm": 1.1018705368041992, "learning_rate": 1.3425389046024155e-05, "loss": 0.16323602199554443, "step": 19900 }, { "epoch": 0.40728448033384645, "grad_norm": 1.7781858444213867, "learning_rate": 1.3419163785544985e-05, "loss": 0.16642351150512696, "step": 19910 }, { "epoch": 0.40748904310649026, "grad_norm": 1.6705667972564697, "learning_rate": 1.3412937024230046e-05, "loss": 0.17800371646881102, "step": 19920 }, { "epoch": 0.4076936058791341, "grad_norm": 2.5949854850769043, "learning_rate": 1.3406708764812572e-05, "loss": 0.1675642490386963, "step": 19930 }, { "epoch": 0.4078981686517779, "grad_norm": 3.2430968284606934, "learning_rate": 1.3400479010026443e-05, "loss": 0.186439049243927, "step": 19940 }, { "epoch": 0.4081027314244217, "grad_norm": 1.6586329936981201, "learning_rate": 1.3394247762606204e-05, "loss": 0.166973876953125, "step": 19950 }, { "epoch": 0.4083072941970656, "grad_norm": 1.5126913785934448, "learning_rate": 1.338801502528704e-05, "loss": 0.1636947512626648, "step": 19960 }, { "epoch": 0.4085118569697094, "grad_norm": 3.4123847484588623, "learning_rate": 1.3381780800804806e-05, "loss": 0.1703795909881592, "step": 19970 }, { "epoch": 0.4087164197423532, "grad_norm": 2.440290689468384, "learning_rate": 1.3375545091896003e-05, "loss": 0.16590590476989747, "step": 19980 }, { "epoch": 0.40892098251499703, "grad_norm": 3.4815609455108643, "learning_rate": 1.3369307901297786e-05, "loss": 0.17058860063552855, "step": 19990 }, { "epoch": 0.40912554528764083, "grad_norm": 1.7869229316711426, "learning_rate": 1.3363069231747955e-05, "loss": 0.15970284938812257, "step": 20000 }, { "epoch": 0.40933010806028464, "grad_norm": 0.9551007747650146, "learning_rate": 1.3356829085984962e-05, "loss": 0.15269501209259034, "step": 20010 }, { "epoch": 0.4095346708329285, "grad_norm": 1.508941411972046, "learning_rate": 1.3350587466747914e-05, "loss": 0.15925732851028443, "step": 20020 }, { "epoch": 0.4097392336055723, "grad_norm": 1.1582881212234497, "learning_rate": 1.334434437677655e-05, "loss": 0.1701311707496643, "step": 20030 }, { "epoch": 0.4099437963782161, "grad_norm": 2.139085054397583, "learning_rate": 1.333809981881127e-05, "loss": 0.16311416625976563, "step": 20040 }, { "epoch": 0.41014835915085995, "grad_norm": 3.7610528469085693, "learning_rate": 1.3331853795593106e-05, "loss": 0.15302605628967286, "step": 20050 }, { "epoch": 0.41035292192350376, "grad_norm": 2.584172487258911, "learning_rate": 1.3325606309863744e-05, "loss": 0.16879658699035643, "step": 20060 }, { "epoch": 0.41055748469614756, "grad_norm": 1.1245172023773193, "learning_rate": 1.3319357364365504e-05, "loss": 0.16511194705963134, "step": 20070 }, { "epoch": 0.4107620474687914, "grad_norm": 0.8485332727432251, "learning_rate": 1.331310696184135e-05, "loss": 0.15399937629699706, "step": 20080 }, { "epoch": 0.4109666102414352, "grad_norm": 0.9198453426361084, "learning_rate": 1.3306855105034879e-05, "loss": 0.1572748064994812, "step": 20090 }, { "epoch": 0.411171173014079, "grad_norm": 4.034480571746826, "learning_rate": 1.330060179669034e-05, "loss": 0.149431049823761, "step": 20100 }, { "epoch": 0.4113757357867229, "grad_norm": 1.6122839450836182, "learning_rate": 1.3294347039552611e-05, "loss": 0.17334483861923217, "step": 20110 }, { "epoch": 0.4115802985593667, "grad_norm": 1.5983432531356812, "learning_rate": 1.3288090836367202e-05, "loss": 0.16062872409820556, "step": 20120 }, { "epoch": 0.4117848613320105, "grad_norm": 1.095428228378296, "learning_rate": 1.3281833189880264e-05, "loss": 0.1602558374404907, "step": 20130 }, { "epoch": 0.41198942410465433, "grad_norm": 2.6945321559906006, "learning_rate": 1.3275574102838583e-05, "loss": 0.1483849048614502, "step": 20140 }, { "epoch": 0.41219398687729814, "grad_norm": 2.184734582901001, "learning_rate": 1.3269313577989566e-05, "loss": 0.15984578132629396, "step": 20150 }, { "epoch": 0.41239854964994194, "grad_norm": 1.8106111288070679, "learning_rate": 1.3263051618081266e-05, "loss": 0.15003955364227295, "step": 20160 }, { "epoch": 0.4126031124225858, "grad_norm": 1.2055795192718506, "learning_rate": 1.3256788225862356e-05, "loss": 0.16043710708618164, "step": 20170 }, { "epoch": 0.4128076751952296, "grad_norm": 1.9679830074310303, "learning_rate": 1.3250523404082142e-05, "loss": 0.17651901245117188, "step": 20180 }, { "epoch": 0.4130122379678734, "grad_norm": 1.58089017868042, "learning_rate": 1.3244257155490558e-05, "loss": 0.16462587118148803, "step": 20190 }, { "epoch": 0.41321680074051725, "grad_norm": 2.070117950439453, "learning_rate": 1.3237989482838156e-05, "loss": 0.15690953731536866, "step": 20200 }, { "epoch": 0.41342136351316106, "grad_norm": 5.249818325042725, "learning_rate": 1.3231720388876127e-05, "loss": 0.16346430778503418, "step": 20210 }, { "epoch": 0.41362592628580486, "grad_norm": 2.4406614303588867, "learning_rate": 1.3225449876356274e-05, "loss": 0.17051661014556885, "step": 20220 }, { "epoch": 0.4138304890584487, "grad_norm": 1.242598056793213, "learning_rate": 1.321917794803103e-05, "loss": 0.1624152898788452, "step": 20230 }, { "epoch": 0.4140350518310925, "grad_norm": 0.7872424125671387, "learning_rate": 1.3212904606653441e-05, "loss": 0.1587973117828369, "step": 20240 }, { "epoch": 0.4142396146037363, "grad_norm": 2.794581890106201, "learning_rate": 1.3206629854977187e-05, "loss": 0.15869786739349365, "step": 20250 }, { "epoch": 0.4144441773763802, "grad_norm": 2.3744921684265137, "learning_rate": 1.3200353695756552e-05, "loss": 0.1641894221305847, "step": 20260 }, { "epoch": 0.414648740149024, "grad_norm": 1.0033103227615356, "learning_rate": 1.3194076131746446e-05, "loss": 0.1567571759223938, "step": 20270 }, { "epoch": 0.4148533029216678, "grad_norm": 2.8326730728149414, "learning_rate": 1.3187797165702399e-05, "loss": 0.17565326690673827, "step": 20280 }, { "epoch": 0.41505786569431163, "grad_norm": 2.45084285736084, "learning_rate": 1.3181516800380549e-05, "loss": 0.15435101985931396, "step": 20290 }, { "epoch": 0.41526242846695544, "grad_norm": 1.9088891744613647, "learning_rate": 1.3175235038537649e-05, "loss": 0.16909717321395873, "step": 20300 }, { "epoch": 0.41546699123959924, "grad_norm": 1.6679376363754272, "learning_rate": 1.3168951882931069e-05, "loss": 0.1760468602180481, "step": 20310 }, { "epoch": 0.4156715540122431, "grad_norm": 2.358635902404785, "learning_rate": 1.3162667336318789e-05, "loss": 0.16582534313201905, "step": 20320 }, { "epoch": 0.4158761167848869, "grad_norm": 3.797717571258545, "learning_rate": 1.3156381401459396e-05, "loss": 0.17056070566177367, "step": 20330 }, { "epoch": 0.4160806795575307, "grad_norm": 1.5394076108932495, "learning_rate": 1.3150094081112095e-05, "loss": 0.15907175540924073, "step": 20340 }, { "epoch": 0.41628524233017455, "grad_norm": 2.6025357246398926, "learning_rate": 1.314380537803669e-05, "loss": 0.1560142755508423, "step": 20350 }, { "epoch": 0.41648980510281836, "grad_norm": 2.451016902923584, "learning_rate": 1.31375152949936e-05, "loss": 0.1585231065750122, "step": 20360 }, { "epoch": 0.41669436787546216, "grad_norm": 3.859377861022949, "learning_rate": 1.3131223834743841e-05, "loss": 0.17051804065704346, "step": 20370 }, { "epoch": 0.416898930648106, "grad_norm": 2.6210031509399414, "learning_rate": 1.312493100004904e-05, "loss": 0.16874318122863768, "step": 20380 }, { "epoch": 0.4171034934207498, "grad_norm": 1.742218017578125, "learning_rate": 1.3118636793671424e-05, "loss": 0.18177850246429444, "step": 20390 }, { "epoch": 0.4173080561933937, "grad_norm": 2.3554813861846924, "learning_rate": 1.3112341218373827e-05, "loss": 0.17051939964294432, "step": 20400 }, { "epoch": 0.4175126189660375, "grad_norm": 1.414450764656067, "learning_rate": 1.3106044276919678e-05, "loss": 0.1596900701522827, "step": 20410 }, { "epoch": 0.4177171817386813, "grad_norm": 5.511402606964111, "learning_rate": 1.3099745972073009e-05, "loss": 0.1657518744468689, "step": 20420 }, { "epoch": 0.41792174451132513, "grad_norm": 1.1284469366073608, "learning_rate": 1.3093446306598447e-05, "loss": 0.16683177947998046, "step": 20430 }, { "epoch": 0.41812630728396893, "grad_norm": 1.5868672132492065, "learning_rate": 1.3087145283261223e-05, "loss": 0.16665395498275756, "step": 20440 }, { "epoch": 0.41833087005661274, "grad_norm": 1.2874783277511597, "learning_rate": 1.3080842904827154e-05, "loss": 0.15998523235321044, "step": 20450 }, { "epoch": 0.4185354328292566, "grad_norm": 1.7015557289123535, "learning_rate": 1.3074539174062663e-05, "loss": 0.14708025455474855, "step": 20460 }, { "epoch": 0.4187399956019004, "grad_norm": 2.9710729122161865, "learning_rate": 1.3068234093734763e-05, "loss": 0.15198099613189697, "step": 20470 }, { "epoch": 0.4189445583745442, "grad_norm": 1.9075146913528442, "learning_rate": 1.306192766661105e-05, "loss": 0.1724087119102478, "step": 20480 }, { "epoch": 0.41914912114718805, "grad_norm": 2.0187876224517822, "learning_rate": 1.305561989545972e-05, "loss": 0.15380704402923584, "step": 20490 }, { "epoch": 0.41935368391983185, "grad_norm": 2.622821569442749, "learning_rate": 1.3049310783049565e-05, "loss": 0.1601744294166565, "step": 20500 }, { "epoch": 0.41955824669247566, "grad_norm": 1.4687033891677856, "learning_rate": 1.3043000332149954e-05, "loss": 0.15933122634887695, "step": 20510 }, { "epoch": 0.4197628094651195, "grad_norm": 2.50335431098938, "learning_rate": 1.3036688545530845e-05, "loss": 0.15987117290496827, "step": 20520 }, { "epoch": 0.4199673722377633, "grad_norm": 3.042267322540283, "learning_rate": 1.303037542596279e-05, "loss": 0.16323635578155518, "step": 20530 }, { "epoch": 0.4201719350104071, "grad_norm": 1.5393168926239014, "learning_rate": 1.3024060976216919e-05, "loss": 0.14786936044692994, "step": 20540 }, { "epoch": 0.420376497783051, "grad_norm": 2.2464115619659424, "learning_rate": 1.3017745199064949e-05, "loss": 0.15303798913955688, "step": 20550 }, { "epoch": 0.4205810605556948, "grad_norm": 1.5008997917175293, "learning_rate": 1.3011428097279178e-05, "loss": 0.1862849235534668, "step": 20560 }, { "epoch": 0.4207856233283386, "grad_norm": 1.5942156314849854, "learning_rate": 1.3005109673632487e-05, "loss": 0.1585829734802246, "step": 20570 }, { "epoch": 0.42099018610098243, "grad_norm": 1.702408790588379, "learning_rate": 1.2998789930898342e-05, "loss": 0.1563529849052429, "step": 20580 }, { "epoch": 0.42119474887362623, "grad_norm": 2.3256006240844727, "learning_rate": 1.2992468871850774e-05, "loss": 0.17653450965881348, "step": 20590 }, { "epoch": 0.42139931164627004, "grad_norm": 1.6751513481140137, "learning_rate": 1.2986146499264402e-05, "loss": 0.15505070686340333, "step": 20600 }, { "epoch": 0.4216038744189139, "grad_norm": 1.0792083740234375, "learning_rate": 1.2979822815914426e-05, "loss": 0.16315932273864747, "step": 20610 }, { "epoch": 0.4218084371915577, "grad_norm": 1.1973263025283813, "learning_rate": 1.297349782457661e-05, "loss": 0.15560693740844728, "step": 20620 }, { "epoch": 0.4220129999642015, "grad_norm": 1.4946712255477905, "learning_rate": 1.2967171528027301e-05, "loss": 0.162648606300354, "step": 20630 }, { "epoch": 0.42221756273684535, "grad_norm": 3.8293004035949707, "learning_rate": 1.296084392904341e-05, "loss": 0.15264465808868408, "step": 20640 }, { "epoch": 0.42242212550948915, "grad_norm": 1.449447751045227, "learning_rate": 1.295451503040243e-05, "loss": 0.16430835723876952, "step": 20650 }, { "epoch": 0.42262668828213296, "grad_norm": 3.6814658641815186, "learning_rate": 1.2948184834882421e-05, "loss": 0.15731410980224608, "step": 20660 }, { "epoch": 0.4228312510547768, "grad_norm": 4.394618034362793, "learning_rate": 1.2941853345262005e-05, "loss": 0.15338753461837767, "step": 20670 }, { "epoch": 0.4230358138274206, "grad_norm": 4.41146183013916, "learning_rate": 1.2935520564320378e-05, "loss": 0.16013531684875487, "step": 20680 }, { "epoch": 0.4232403766000644, "grad_norm": 2.238292694091797, "learning_rate": 1.2929186494837312e-05, "loss": 0.16191774606704712, "step": 20690 }, { "epoch": 0.4234449393727083, "grad_norm": 2.6844098567962646, "learning_rate": 1.2922851139593127e-05, "loss": 0.1585986614227295, "step": 20700 }, { "epoch": 0.4236495021453521, "grad_norm": 2.0999104976654053, "learning_rate": 1.2916514501368718e-05, "loss": 0.1652313470840454, "step": 20710 }, { "epoch": 0.4238540649179959, "grad_norm": 1.708060383796692, "learning_rate": 1.2910176582945537e-05, "loss": 0.16416754722595214, "step": 20720 }, { "epoch": 0.42405862769063973, "grad_norm": 0.9740471243858337, "learning_rate": 1.2903837387105608e-05, "loss": 0.1596771478652954, "step": 20730 }, { "epoch": 0.42426319046328353, "grad_norm": 3.1823747158050537, "learning_rate": 1.2897496916631509e-05, "loss": 0.1432097315788269, "step": 20740 }, { "epoch": 0.42446775323592734, "grad_norm": 3.0126700401306152, "learning_rate": 1.2891155174306373e-05, "loss": 0.15922050476074218, "step": 20750 }, { "epoch": 0.4246723160085712, "grad_norm": 3.2672321796417236, "learning_rate": 1.2884812162913903e-05, "loss": 0.1596675157546997, "step": 20760 }, { "epoch": 0.424876878781215, "grad_norm": 2.0901849269866943, "learning_rate": 1.2878467885238348e-05, "loss": 0.1467190146446228, "step": 20770 }, { "epoch": 0.4250814415538588, "grad_norm": 1.5064059495925903, "learning_rate": 1.2872122344064517e-05, "loss": 0.16508169174194337, "step": 20780 }, { "epoch": 0.42528600432650265, "grad_norm": 2.3832693099975586, "learning_rate": 1.2865775542177775e-05, "loss": 0.17498011589050294, "step": 20790 }, { "epoch": 0.42549056709914646, "grad_norm": 1.2956453561782837, "learning_rate": 1.2859427482364038e-05, "loss": 0.15446679592132567, "step": 20800 }, { "epoch": 0.42569512987179026, "grad_norm": 3.397007942199707, "learning_rate": 1.2853078167409778e-05, "loss": 0.16671926975250245, "step": 20810 }, { "epoch": 0.4258996926444341, "grad_norm": 1.6163417100906372, "learning_rate": 1.2846727600102015e-05, "loss": 0.16378564834594728, "step": 20820 }, { "epoch": 0.4261042554170779, "grad_norm": 2.192237615585327, "learning_rate": 1.2840375783228317e-05, "loss": 0.15558031797409058, "step": 20830 }, { "epoch": 0.4263088181897217, "grad_norm": 2.538668632507324, "learning_rate": 1.2834022719576804e-05, "loss": 0.16193320751190185, "step": 20840 }, { "epoch": 0.4265133809623656, "grad_norm": 2.928345203399658, "learning_rate": 1.2827668411936139e-05, "loss": 0.16543543338775635, "step": 20850 }, { "epoch": 0.4267179437350094, "grad_norm": 1.880028247833252, "learning_rate": 1.2821312863095536e-05, "loss": 0.1471657156944275, "step": 20860 }, { "epoch": 0.42692250650765323, "grad_norm": 3.125154495239258, "learning_rate": 1.2814956075844753e-05, "loss": 0.15558620691299438, "step": 20870 }, { "epoch": 0.42712706928029703, "grad_norm": 3.2227494716644287, "learning_rate": 1.2808598052974094e-05, "loss": 0.14524440765380858, "step": 20880 }, { "epoch": 0.42733163205294084, "grad_norm": 1.4823678731918335, "learning_rate": 1.280223879727439e-05, "loss": 0.15727087259292602, "step": 20890 }, { "epoch": 0.4275361948255847, "grad_norm": 1.390446662902832, "learning_rate": 1.279587831153703e-05, "loss": 0.15186002254486083, "step": 20900 }, { "epoch": 0.4277407575982285, "grad_norm": 1.7268157005310059, "learning_rate": 1.2789516598553944e-05, "loss": 0.15162545442581177, "step": 20910 }, { "epoch": 0.4279453203708723, "grad_norm": 2.0298428535461426, "learning_rate": 1.2783153661117587e-05, "loss": 0.15150331258773803, "step": 20920 }, { "epoch": 0.42814988314351615, "grad_norm": 2.227720022201538, "learning_rate": 1.2776789502020964e-05, "loss": 0.1580786943435669, "step": 20930 }, { "epoch": 0.42835444591615995, "grad_norm": 1.6134629249572754, "learning_rate": 1.2770424124057605e-05, "loss": 0.1607067584991455, "step": 20940 }, { "epoch": 0.42855900868880376, "grad_norm": 2.7161641120910645, "learning_rate": 1.276405753002159e-05, "loss": 0.15639653205871581, "step": 20950 }, { "epoch": 0.4287635714614476, "grad_norm": 1.8740904331207275, "learning_rate": 1.2757689722707514e-05, "loss": 0.1529465913772583, "step": 20960 }, { "epoch": 0.4289681342340914, "grad_norm": 2.409268856048584, "learning_rate": 1.275132070491052e-05, "loss": 0.1509261131286621, "step": 20970 }, { "epoch": 0.4291726970067352, "grad_norm": 1.166475772857666, "learning_rate": 1.2744950479426277e-05, "loss": 0.16758313179016113, "step": 20980 }, { "epoch": 0.4293772597793791, "grad_norm": 1.839056372642517, "learning_rate": 1.2738579049050985e-05, "loss": 0.16898388862609864, "step": 20990 }, { "epoch": 0.4295818225520229, "grad_norm": 1.674228549003601, "learning_rate": 1.2732206416581366e-05, "loss": 0.1529913067817688, "step": 21000 }, { "epoch": 0.4297863853246667, "grad_norm": 2.3885886669158936, "learning_rate": 1.2725832584814681e-05, "loss": 0.16556222438812257, "step": 21010 }, { "epoch": 0.42999094809731053, "grad_norm": 1.6197919845581055, "learning_rate": 1.2719457556548712e-05, "loss": 0.1508520245552063, "step": 21020 }, { "epoch": 0.43019551086995433, "grad_norm": 3.481506109237671, "learning_rate": 1.2713081334581763e-05, "loss": 0.168154776096344, "step": 21030 }, { "epoch": 0.43040007364259814, "grad_norm": 1.6359323263168335, "learning_rate": 1.2706703921712664e-05, "loss": 0.16263206005096437, "step": 21040 }, { "epoch": 0.430604636415242, "grad_norm": 3.0420589447021484, "learning_rate": 1.2700325320740773e-05, "loss": 0.16945796012878417, "step": 21050 }, { "epoch": 0.4308091991878858, "grad_norm": 2.0701370239257812, "learning_rate": 1.2693945534465967e-05, "loss": 0.1515645742416382, "step": 21060 }, { "epoch": 0.4310137619605296, "grad_norm": 1.2966365814208984, "learning_rate": 1.2687564565688636e-05, "loss": 0.15575947761535644, "step": 21070 }, { "epoch": 0.43121832473317345, "grad_norm": 2.1871144771575928, "learning_rate": 1.2681182417209697e-05, "loss": 0.16587557792663574, "step": 21080 }, { "epoch": 0.43142288750581725, "grad_norm": 1.490976333618164, "learning_rate": 1.2674799091830582e-05, "loss": 0.16132650375366211, "step": 21090 }, { "epoch": 0.43162745027846106, "grad_norm": 1.2108783721923828, "learning_rate": 1.2668414592353244e-05, "loss": 0.1602320671081543, "step": 21100 }, { "epoch": 0.4318320130511049, "grad_norm": 1.6839730739593506, "learning_rate": 1.266202892158014e-05, "loss": 0.1567065715789795, "step": 21110 }, { "epoch": 0.4320365758237487, "grad_norm": 1.8356071710586548, "learning_rate": 1.2655642082314255e-05, "loss": 0.15154778957366943, "step": 21120 }, { "epoch": 0.4322411385963925, "grad_norm": 1.8164796829223633, "learning_rate": 1.2649254077359079e-05, "loss": 0.156777286529541, "step": 21130 }, { "epoch": 0.4324457013690364, "grad_norm": 2.279370069503784, "learning_rate": 1.2642864909518616e-05, "loss": 0.16347355842590333, "step": 21140 }, { "epoch": 0.4326502641416802, "grad_norm": 2.317613363265991, "learning_rate": 1.2636474581597373e-05, "loss": 0.1467137336730957, "step": 21150 }, { "epoch": 0.432854826914324, "grad_norm": 1.8831928968429565, "learning_rate": 1.2630083096400382e-05, "loss": 0.15288190841674804, "step": 21160 }, { "epoch": 0.43305938968696783, "grad_norm": 1.0856702327728271, "learning_rate": 1.262369045673317e-05, "loss": 0.15336275100708008, "step": 21170 }, { "epoch": 0.43326395245961163, "grad_norm": 1.2626376152038574, "learning_rate": 1.2617296665401778e-05, "loss": 0.1687880277633667, "step": 21180 }, { "epoch": 0.43346851523225544, "grad_norm": 1.6946929693222046, "learning_rate": 1.261090172521274e-05, "loss": 0.1561334490776062, "step": 21190 }, { "epoch": 0.4336730780048993, "grad_norm": 2.882568836212158, "learning_rate": 1.2604505638973112e-05, "loss": 0.15361528396606444, "step": 21200 }, { "epoch": 0.4338776407775431, "grad_norm": 4.898351192474365, "learning_rate": 1.2598108409490444e-05, "loss": 0.15800752639770507, "step": 21210 }, { "epoch": 0.4340822035501869, "grad_norm": 1.3758996725082397, "learning_rate": 1.2591710039572784e-05, "loss": 0.16058659553527832, "step": 21220 }, { "epoch": 0.43428676632283075, "grad_norm": 1.305069923400879, "learning_rate": 1.2585310532028686e-05, "loss": 0.14982964992523193, "step": 21230 }, { "epoch": 0.43449132909547455, "grad_norm": 1.428295373916626, "learning_rate": 1.2578909889667204e-05, "loss": 0.17465684413909913, "step": 21240 }, { "epoch": 0.43469589186811836, "grad_norm": 1.504897952079773, "learning_rate": 1.257250811529789e-05, "loss": 0.15050845146179198, "step": 21250 }, { "epoch": 0.4349004546407622, "grad_norm": 2.6199846267700195, "learning_rate": 1.2566105211730788e-05, "loss": 0.152894127368927, "step": 21260 }, { "epoch": 0.435105017413406, "grad_norm": 1.062104344367981, "learning_rate": 1.255970118177644e-05, "loss": 0.15626683235168456, "step": 21270 }, { "epoch": 0.4353095801860498, "grad_norm": 3.7402939796447754, "learning_rate": 1.2553296028245887e-05, "loss": 0.1522066593170166, "step": 21280 }, { "epoch": 0.4355141429586937, "grad_norm": 3.4564595222473145, "learning_rate": 1.2546889753950657e-05, "loss": 0.15725724697113036, "step": 21290 }, { "epoch": 0.4357187057313375, "grad_norm": 1.50242018699646, "learning_rate": 1.2540482361702773e-05, "loss": 0.15876729488372804, "step": 21300 }, { "epoch": 0.4359232685039813, "grad_norm": 1.9530973434448242, "learning_rate": 1.253407385431475e-05, "loss": 0.15943821668624877, "step": 21310 }, { "epoch": 0.43612783127662513, "grad_norm": 2.02726674079895, "learning_rate": 1.252766423459959e-05, "loss": 0.15006752014160157, "step": 21320 }, { "epoch": 0.43633239404926893, "grad_norm": 1.1992466449737549, "learning_rate": 1.2521253505370785e-05, "loss": 0.1546098470687866, "step": 21330 }, { "epoch": 0.4365369568219128, "grad_norm": 1.1153978109359741, "learning_rate": 1.2514841669442306e-05, "loss": 0.16013824939727783, "step": 21340 }, { "epoch": 0.4367415195945566, "grad_norm": 2.7946741580963135, "learning_rate": 1.2508428729628626e-05, "loss": 0.15149264335632323, "step": 21350 }, { "epoch": 0.4369460823672004, "grad_norm": 4.382969856262207, "learning_rate": 1.250201468874469e-05, "loss": 0.16991639137268066, "step": 21360 }, { "epoch": 0.43715064513984425, "grad_norm": 1.7807716131210327, "learning_rate": 1.2495599549605929e-05, "loss": 0.16727595329284667, "step": 21370 }, { "epoch": 0.43735520791248805, "grad_norm": 1.7390514612197876, "learning_rate": 1.2489183315028256e-05, "loss": 0.16628572940826417, "step": 21380 }, { "epoch": 0.43755977068513185, "grad_norm": 2.036824941635132, "learning_rate": 1.2482765987828064e-05, "loss": 0.16345876455307007, "step": 21390 }, { "epoch": 0.4377643334577757, "grad_norm": 1.7545980215072632, "learning_rate": 1.2476347570822233e-05, "loss": 0.1420891523361206, "step": 21400 }, { "epoch": 0.4379688962304195, "grad_norm": 2.0027823448181152, "learning_rate": 1.2469928066828112e-05, "loss": 0.17321239709854125, "step": 21410 }, { "epoch": 0.4381734590030633, "grad_norm": 1.4402368068695068, "learning_rate": 1.246350747866353e-05, "loss": 0.14795750379562378, "step": 21420 }, { "epoch": 0.43837802177570717, "grad_norm": 2.152351140975952, "learning_rate": 1.2457085809146795e-05, "loss": 0.16025173664093018, "step": 21430 }, { "epoch": 0.438582584548351, "grad_norm": 2.692755937576294, "learning_rate": 1.2450663061096684e-05, "loss": 0.1638244867324829, "step": 21440 }, { "epoch": 0.4387871473209948, "grad_norm": 3.391995668411255, "learning_rate": 1.2444239237332449e-05, "loss": 0.14944689273834227, "step": 21450 }, { "epoch": 0.43899171009363863, "grad_norm": 1.24317467212677, "learning_rate": 1.2437814340673824e-05, "loss": 0.1587376832962036, "step": 21460 }, { "epoch": 0.43919627286628243, "grad_norm": 1.2365412712097168, "learning_rate": 1.2431388373941002e-05, "loss": 0.149245548248291, "step": 21470 }, { "epoch": 0.43940083563892623, "grad_norm": 2.221557140350342, "learning_rate": 1.2424961339954646e-05, "loss": 0.16525721549987793, "step": 21480 }, { "epoch": 0.4396053984115701, "grad_norm": 1.6652549505233765, "learning_rate": 1.2418533241535894e-05, "loss": 0.15999459028244017, "step": 21490 }, { "epoch": 0.4398099611842139, "grad_norm": 2.027979612350464, "learning_rate": 1.2412104081506349e-05, "loss": 0.14611848592758178, "step": 21500 }, { "epoch": 0.4400145239568577, "grad_norm": 2.089385509490967, "learning_rate": 1.2405673862688073e-05, "loss": 0.16007888317108154, "step": 21510 }, { "epoch": 0.44021908672950155, "grad_norm": 1.9373059272766113, "learning_rate": 1.2399242587903608e-05, "loss": 0.16686182022094725, "step": 21520 }, { "epoch": 0.44042364950214535, "grad_norm": 3.0409159660339355, "learning_rate": 1.2392810259975943e-05, "loss": 0.15962225198745728, "step": 21530 }, { "epoch": 0.44062821227478915, "grad_norm": 0.8499221205711365, "learning_rate": 1.2386376881728543e-05, "loss": 0.16086626052856445, "step": 21540 }, { "epoch": 0.440832775047433, "grad_norm": 1.9519306421279907, "learning_rate": 1.2379942455985321e-05, "loss": 0.16157320737838746, "step": 21550 }, { "epoch": 0.4410373378200768, "grad_norm": 2.392235040664673, "learning_rate": 1.2373506985570662e-05, "loss": 0.16334433555603028, "step": 21560 }, { "epoch": 0.4412419005927206, "grad_norm": 1.3229011297225952, "learning_rate": 1.2367070473309402e-05, "loss": 0.1655963182449341, "step": 21570 }, { "epoch": 0.44144646336536447, "grad_norm": 1.7772696018218994, "learning_rate": 1.2360632922026841e-05, "loss": 0.16717169284820557, "step": 21580 }, { "epoch": 0.4416510261380083, "grad_norm": 1.2895638942718506, "learning_rate": 1.235419433454872e-05, "loss": 0.1570512294769287, "step": 21590 }, { "epoch": 0.4418555889106521, "grad_norm": 1.5036901235580444, "learning_rate": 1.2347754713701252e-05, "loss": 0.15906670093536376, "step": 21600 }, { "epoch": 0.44206015168329593, "grad_norm": 1.3374576568603516, "learning_rate": 1.2341314062311101e-05, "loss": 0.15555533170700073, "step": 21610 }, { "epoch": 0.44226471445593973, "grad_norm": 1.943380355834961, "learning_rate": 1.2334872383205369e-05, "loss": 0.15806578397750853, "step": 21620 }, { "epoch": 0.44246927722858354, "grad_norm": 5.910507678985596, "learning_rate": 1.2328429679211629e-05, "loss": 0.15254130363464355, "step": 21630 }, { "epoch": 0.4426738400012274, "grad_norm": 3.2374472618103027, "learning_rate": 1.2321985953157887e-05, "loss": 0.1565652847290039, "step": 21640 }, { "epoch": 0.4428784027738712, "grad_norm": 3.218411684036255, "learning_rate": 1.2315541207872612e-05, "loss": 0.16636474132537843, "step": 21650 }, { "epoch": 0.443082965546515, "grad_norm": 0.8829190135002136, "learning_rate": 1.2309095446184707e-05, "loss": 0.15448641777038574, "step": 21660 }, { "epoch": 0.44328752831915885, "grad_norm": 2.2233498096466064, "learning_rate": 1.2302648670923531e-05, "loss": 0.15856430530548096, "step": 21670 }, { "epoch": 0.44349209109180265, "grad_norm": 2.0779147148132324, "learning_rate": 1.2296200884918884e-05, "loss": 0.1550661325454712, "step": 21680 }, { "epoch": 0.44369665386444646, "grad_norm": 0.7588064074516296, "learning_rate": 1.2289752091001008e-05, "loss": 0.14981664419174195, "step": 21690 }, { "epoch": 0.4439012166370903, "grad_norm": 1.2153713703155518, "learning_rate": 1.2283302292000589e-05, "loss": 0.1573217511177063, "step": 21700 }, { "epoch": 0.4441057794097341, "grad_norm": 4.048986911773682, "learning_rate": 1.227685149074876e-05, "loss": 0.16430239677429198, "step": 21710 }, { "epoch": 0.4443103421823779, "grad_norm": 2.219569683074951, "learning_rate": 1.2270399690077084e-05, "loss": 0.17331233024597167, "step": 21720 }, { "epoch": 0.4445149049550218, "grad_norm": 2.334214448928833, "learning_rate": 1.2263946892817569e-05, "loss": 0.16784048080444336, "step": 21730 }, { "epoch": 0.4447194677276656, "grad_norm": 2.637510061264038, "learning_rate": 1.2257493101802655e-05, "loss": 0.15383338928222656, "step": 21740 }, { "epoch": 0.4449240305003094, "grad_norm": 1.1429145336151123, "learning_rate": 1.2251038319865227e-05, "loss": 0.14674835205078124, "step": 21750 }, { "epoch": 0.44512859327295323, "grad_norm": 1.9647897481918335, "learning_rate": 1.2244582549838597e-05, "loss": 0.15579451322555543, "step": 21760 }, { "epoch": 0.44533315604559703, "grad_norm": 1.507128119468689, "learning_rate": 1.2238125794556514e-05, "loss": 0.15521761178970336, "step": 21770 }, { "epoch": 0.44553771881824084, "grad_norm": 2.1103591918945312, "learning_rate": 1.2231668056853159e-05, "loss": 0.16680731773376464, "step": 21780 }, { "epoch": 0.4457422815908847, "grad_norm": 1.5995588302612305, "learning_rate": 1.2225209339563144e-05, "loss": 0.15906494855880737, "step": 21790 }, { "epoch": 0.4459468443635285, "grad_norm": 1.0063380002975464, "learning_rate": 1.2218749645521515e-05, "loss": 0.15358588695526124, "step": 21800 }, { "epoch": 0.44615140713617235, "grad_norm": 1.5571209192276, "learning_rate": 1.2212288977563737e-05, "loss": 0.16106076240539552, "step": 21810 }, { "epoch": 0.44635596990881615, "grad_norm": 3.9514622688293457, "learning_rate": 1.2205827338525711e-05, "loss": 0.15914334058761598, "step": 21820 }, { "epoch": 0.44656053268145995, "grad_norm": 1.5913498401641846, "learning_rate": 1.2199364731243766e-05, "loss": 0.1575561285018921, "step": 21830 }, { "epoch": 0.4467650954541038, "grad_norm": 2.507077693939209, "learning_rate": 1.2192901158554645e-05, "loss": 0.1579175591468811, "step": 21840 }, { "epoch": 0.4469696582267476, "grad_norm": 4.0983734130859375, "learning_rate": 1.2186436623295524e-05, "loss": 0.164908766746521, "step": 21850 }, { "epoch": 0.4471742209993914, "grad_norm": 2.744823932647705, "learning_rate": 1.2179971128304e-05, "loss": 0.16531939506530763, "step": 21860 }, { "epoch": 0.44737878377203527, "grad_norm": 1.6977040767669678, "learning_rate": 1.2173504676418091e-05, "loss": 0.14444139003753662, "step": 21870 }, { "epoch": 0.4475833465446791, "grad_norm": 1.7269864082336426, "learning_rate": 1.2167037270476235e-05, "loss": 0.15216885805130004, "step": 21880 }, { "epoch": 0.4477879093173229, "grad_norm": 2.220299243927002, "learning_rate": 1.2160568913317281e-05, "loss": 0.15671249628067016, "step": 21890 }, { "epoch": 0.44799247208996673, "grad_norm": 3.2569127082824707, "learning_rate": 1.2154099607780512e-05, "loss": 0.16189265251159668, "step": 21900 }, { "epoch": 0.44819703486261053, "grad_norm": 1.1394513845443726, "learning_rate": 1.2147629356705613e-05, "loss": 0.142666494846344, "step": 21910 }, { "epoch": 0.44840159763525433, "grad_norm": 1.979775309562683, "learning_rate": 1.2141158162932691e-05, "loss": 0.15858204364776612, "step": 21920 }, { "epoch": 0.4486061604078982, "grad_norm": 1.1478694677352905, "learning_rate": 1.213468602930226e-05, "loss": 0.14754788875579833, "step": 21930 }, { "epoch": 0.448810723180542, "grad_norm": 3.0783474445343018, "learning_rate": 1.212821295865526e-05, "loss": 0.15589461326599122, "step": 21940 }, { "epoch": 0.4490152859531858, "grad_norm": 1.492993712425232, "learning_rate": 1.212173895383303e-05, "loss": 0.14567986726760865, "step": 21950 }, { "epoch": 0.44921984872582965, "grad_norm": 1.4801222085952759, "learning_rate": 1.2115264017677317e-05, "loss": 0.15224568843841552, "step": 21960 }, { "epoch": 0.44942441149847345, "grad_norm": 3.605713367462158, "learning_rate": 1.2108788153030289e-05, "loss": 0.16190462112426757, "step": 21970 }, { "epoch": 0.44962897427111725, "grad_norm": 1.2695536613464355, "learning_rate": 1.2102311362734514e-05, "loss": 0.15279898643493653, "step": 21980 }, { "epoch": 0.4498335370437611, "grad_norm": 1.0890552997589111, "learning_rate": 1.2095833649632965e-05, "loss": 0.1460193157196045, "step": 21990 }, { "epoch": 0.4500380998164049, "grad_norm": 2.0749547481536865, "learning_rate": 1.2089355016569025e-05, "loss": 0.1564727544784546, "step": 22000 }, { "epoch": 0.4502426625890487, "grad_norm": 2.604055643081665, "learning_rate": 1.2082875466386477e-05, "loss": 0.15275295972824096, "step": 22010 }, { "epoch": 0.45044722536169257, "grad_norm": 1.7419785261154175, "learning_rate": 1.2076395001929509e-05, "loss": 0.16750116348266603, "step": 22020 }, { "epoch": 0.4506517881343364, "grad_norm": 1.4391591548919678, "learning_rate": 1.2069913626042709e-05, "loss": 0.15493035316467285, "step": 22030 }, { "epoch": 0.4508563509069802, "grad_norm": 2.349548101425171, "learning_rate": 1.206343134157106e-05, "loss": 0.1519089460372925, "step": 22040 }, { "epoch": 0.45106091367962403, "grad_norm": 2.2280426025390625, "learning_rate": 1.2056948151359959e-05, "loss": 0.16254889965057373, "step": 22050 }, { "epoch": 0.45126547645226783, "grad_norm": 1.7343060970306396, "learning_rate": 1.2050464058255185e-05, "loss": 0.16543641090393066, "step": 22060 }, { "epoch": 0.45147003922491163, "grad_norm": 1.6561012268066406, "learning_rate": 1.2043979065102918e-05, "loss": 0.1541245698928833, "step": 22070 }, { "epoch": 0.4516746019975555, "grad_norm": 0.9380149245262146, "learning_rate": 1.2037493174749733e-05, "loss": 0.16186835765838622, "step": 22080 }, { "epoch": 0.4518791647701993, "grad_norm": 2.6780221462249756, "learning_rate": 1.2031006390042606e-05, "loss": 0.1713528275489807, "step": 22090 }, { "epoch": 0.4520837275428431, "grad_norm": 1.1201090812683105, "learning_rate": 1.2024518713828892e-05, "loss": 0.15614862442016603, "step": 22100 }, { "epoch": 0.45228829031548695, "grad_norm": 1.6351228952407837, "learning_rate": 1.201803014895635e-05, "loss": 0.17161649465560913, "step": 22110 }, { "epoch": 0.45249285308813075, "grad_norm": 1.259497880935669, "learning_rate": 1.2011540698273119e-05, "loss": 0.1474716305732727, "step": 22120 }, { "epoch": 0.45269741586077455, "grad_norm": 2.0687143802642822, "learning_rate": 1.2005050364627739e-05, "loss": 0.15896377563476563, "step": 22130 }, { "epoch": 0.4529019786334184, "grad_norm": 3.3965795040130615, "learning_rate": 1.1998559150869124e-05, "loss": 0.1689465284347534, "step": 22140 }, { "epoch": 0.4531065414060622, "grad_norm": 2.070222854614258, "learning_rate": 1.1992067059846584e-05, "loss": 0.1764465093612671, "step": 22150 }, { "epoch": 0.453311104178706, "grad_norm": 1.9777206182479858, "learning_rate": 1.1985574094409806e-05, "loss": 0.15799906253814697, "step": 22160 }, { "epoch": 0.45351566695134987, "grad_norm": 2.1671459674835205, "learning_rate": 1.1979080257408873e-05, "loss": 0.1508244514465332, "step": 22170 }, { "epoch": 0.4537202297239937, "grad_norm": 1.3096648454666138, "learning_rate": 1.1972585551694234e-05, "loss": 0.1645442485809326, "step": 22180 }, { "epoch": 0.4539247924966375, "grad_norm": 2.5277395248413086, "learning_rate": 1.1966089980116733e-05, "loss": 0.15815092325210572, "step": 22190 }, { "epoch": 0.45412935526928133, "grad_norm": 2.0879673957824707, "learning_rate": 1.1959593545527594e-05, "loss": 0.16788227558135987, "step": 22200 }, { "epoch": 0.45433391804192513, "grad_norm": 1.8253706693649292, "learning_rate": 1.1953096250778407e-05, "loss": 0.16059943437576293, "step": 22210 }, { "epoch": 0.45453848081456893, "grad_norm": 1.8903239965438843, "learning_rate": 1.1946598098721151e-05, "loss": 0.15217987298965455, "step": 22220 }, { "epoch": 0.4547430435872128, "grad_norm": 0.8650144934654236, "learning_rate": 1.1940099092208178e-05, "loss": 0.15176353454589844, "step": 22230 }, { "epoch": 0.4549476063598566, "grad_norm": 3.6059188842773438, "learning_rate": 1.193359923409222e-05, "loss": 0.14006081819534302, "step": 22240 }, { "epoch": 0.4551521691325004, "grad_norm": 2.8173372745513916, "learning_rate": 1.1927098527226369e-05, "loss": 0.161092209815979, "step": 22250 }, { "epoch": 0.45535673190514425, "grad_norm": 0.8243618607521057, "learning_rate": 1.1920596974464102e-05, "loss": 0.1582557201385498, "step": 22260 }, { "epoch": 0.45556129467778805, "grad_norm": 1.7008723020553589, "learning_rate": 1.1914094578659269e-05, "loss": 0.1472676396369934, "step": 22270 }, { "epoch": 0.4557658574504319, "grad_norm": 1.5317524671554565, "learning_rate": 1.190759134266608e-05, "loss": 0.15879368782043457, "step": 22280 }, { "epoch": 0.4559704202230757, "grad_norm": 1.4012775421142578, "learning_rate": 1.1901087269339116e-05, "loss": 0.1587069272994995, "step": 22290 }, { "epoch": 0.4561749829957195, "grad_norm": 1.5066555738449097, "learning_rate": 1.1894582361533335e-05, "loss": 0.1489718198776245, "step": 22300 }, { "epoch": 0.45637954576836337, "grad_norm": 1.446406602859497, "learning_rate": 1.188807662210405e-05, "loss": 0.16022148132324218, "step": 22310 }, { "epoch": 0.45658410854100717, "grad_norm": 3.081547975540161, "learning_rate": 1.1881570053906945e-05, "loss": 0.171738600730896, "step": 22320 }, { "epoch": 0.456788671313651, "grad_norm": 2.896972179412842, "learning_rate": 1.1875062659798064e-05, "loss": 0.1613783836364746, "step": 22330 }, { "epoch": 0.45699323408629483, "grad_norm": 1.6568291187286377, "learning_rate": 1.1868554442633822e-05, "loss": 0.16316578388214112, "step": 22340 }, { "epoch": 0.45719779685893863, "grad_norm": 2.571964740753174, "learning_rate": 1.1862045405270983e-05, "loss": 0.16715784072875978, "step": 22350 }, { "epoch": 0.45740235963158243, "grad_norm": 1.4767165184020996, "learning_rate": 1.1855535550566682e-05, "loss": 0.15821110010147094, "step": 22360 }, { "epoch": 0.4576069224042263, "grad_norm": 1.9308390617370605, "learning_rate": 1.1849024881378399e-05, "loss": 0.14460995197296142, "step": 22370 }, { "epoch": 0.4578114851768701, "grad_norm": 2.581252336502075, "learning_rate": 1.1842513400563993e-05, "loss": 0.14620498418807984, "step": 22380 }, { "epoch": 0.4580160479495139, "grad_norm": 2.5888595581054688, "learning_rate": 1.1836001110981656e-05, "loss": 0.16529961824417114, "step": 22390 }, { "epoch": 0.45822061072215775, "grad_norm": 1.4748364686965942, "learning_rate": 1.1829488015489952e-05, "loss": 0.13424501419067383, "step": 22400 }, { "epoch": 0.45842517349480155, "grad_norm": 2.5284993648529053, "learning_rate": 1.1822974116947792e-05, "loss": 0.14626526832580566, "step": 22410 }, { "epoch": 0.45862973626744535, "grad_norm": 1.6719118356704712, "learning_rate": 1.1816459418214436e-05, "loss": 0.15661509037017823, "step": 22420 }, { "epoch": 0.4588342990400892, "grad_norm": 3.354477882385254, "learning_rate": 1.1809943922149506e-05, "loss": 0.15669184923171997, "step": 22430 }, { "epoch": 0.459038861812733, "grad_norm": 2.3012397289276123, "learning_rate": 1.1803427631612958e-05, "loss": 0.15138092041015624, "step": 22440 }, { "epoch": 0.4592434245853768, "grad_norm": 2.3005199432373047, "learning_rate": 1.1796910549465116e-05, "loss": 0.14694904088973998, "step": 22450 }, { "epoch": 0.45944798735802067, "grad_norm": 1.6742346286773682, "learning_rate": 1.1790392678566633e-05, "loss": 0.1480247974395752, "step": 22460 }, { "epoch": 0.4596525501306645, "grad_norm": 2.0144989490509033, "learning_rate": 1.1783874021778523e-05, "loss": 0.1527927875518799, "step": 22470 }, { "epoch": 0.4598571129033083, "grad_norm": 3.5663936138153076, "learning_rate": 1.1777354581962132e-05, "loss": 0.15931034088134766, "step": 22480 }, { "epoch": 0.46006167567595213, "grad_norm": 1.4541399478912354, "learning_rate": 1.1770834361979165e-05, "loss": 0.16903836727142335, "step": 22490 }, { "epoch": 0.46026623844859593, "grad_norm": 2.0443308353424072, "learning_rate": 1.1764313364691652e-05, "loss": 0.161848247051239, "step": 22500 }, { "epoch": 0.46047080122123973, "grad_norm": 1.000137448310852, "learning_rate": 1.175779159296198e-05, "loss": 0.15344598293304443, "step": 22510 }, { "epoch": 0.4606753639938836, "grad_norm": 1.7603141069412231, "learning_rate": 1.1751269049652862e-05, "loss": 0.15610101222991943, "step": 22520 }, { "epoch": 0.4608799267665274, "grad_norm": 1.3088374137878418, "learning_rate": 1.1744745737627364e-05, "loss": 0.14977505207061767, "step": 22530 }, { "epoch": 0.4610844895391712, "grad_norm": 2.497988224029541, "learning_rate": 1.1738221659748877e-05, "loss": 0.16373027563095094, "step": 22540 }, { "epoch": 0.46128905231181505, "grad_norm": 2.2721126079559326, "learning_rate": 1.1731696818881135e-05, "loss": 0.15726070404052733, "step": 22550 }, { "epoch": 0.46149361508445885, "grad_norm": 0.9507908821105957, "learning_rate": 1.1725171217888202e-05, "loss": 0.14816415309906006, "step": 22560 }, { "epoch": 0.46169817785710265, "grad_norm": 1.287941575050354, "learning_rate": 1.1718644859634485e-05, "loss": 0.16014159917831422, "step": 22570 }, { "epoch": 0.4619027406297465, "grad_norm": 0.9765119552612305, "learning_rate": 1.1712117746984713e-05, "loss": 0.15216314792633057, "step": 22580 }, { "epoch": 0.4621073034023903, "grad_norm": 5.780767440795898, "learning_rate": 1.1705589882803947e-05, "loss": 0.15215781927108765, "step": 22590 }, { "epoch": 0.4623118661750341, "grad_norm": 3.5879364013671875, "learning_rate": 1.169906126995759e-05, "loss": 0.1593770742416382, "step": 22600 }, { "epoch": 0.46251642894767797, "grad_norm": 1.3473283052444458, "learning_rate": 1.1692531911311357e-05, "loss": 0.15114763975143433, "step": 22610 }, { "epoch": 0.4627209917203218, "grad_norm": 2.3521831035614014, "learning_rate": 1.1686001809731304e-05, "loss": 0.1456700086593628, "step": 22620 }, { "epoch": 0.4629255544929656, "grad_norm": 0.7956385612487793, "learning_rate": 1.1679470968083802e-05, "loss": 0.14943430423736573, "step": 22630 }, { "epoch": 0.46313011726560943, "grad_norm": 2.2714011669158936, "learning_rate": 1.1672939389235554e-05, "loss": 0.15783374309539794, "step": 22640 }, { "epoch": 0.46333468003825323, "grad_norm": 3.9229037761688232, "learning_rate": 1.1666407076053587e-05, "loss": 0.16381263732910156, "step": 22650 }, { "epoch": 0.46353924281089703, "grad_norm": 2.061089277267456, "learning_rate": 1.1659874031405245e-05, "loss": 0.14663008451461793, "step": 22660 }, { "epoch": 0.4637438055835409, "grad_norm": 1.122887134552002, "learning_rate": 1.1653340258158197e-05, "loss": 0.16017353534698486, "step": 22670 }, { "epoch": 0.4639483683561847, "grad_norm": 1.6591325998306274, "learning_rate": 1.1646805759180435e-05, "loss": 0.16163620948791504, "step": 22680 }, { "epoch": 0.4641529311288285, "grad_norm": 1.344881296157837, "learning_rate": 1.1640270537340257e-05, "loss": 0.1415133595466614, "step": 22690 }, { "epoch": 0.46435749390147235, "grad_norm": 2.0075788497924805, "learning_rate": 1.1633734595506296e-05, "loss": 0.1546395778656006, "step": 22700 }, { "epoch": 0.46456205667411615, "grad_norm": 2.4228365421295166, "learning_rate": 1.1627197936547484e-05, "loss": 0.1584046006202698, "step": 22710 }, { "epoch": 0.46476661944675995, "grad_norm": 2.6006462574005127, "learning_rate": 1.1620660563333084e-05, "loss": 0.15386741161346434, "step": 22720 }, { "epoch": 0.4649711822194038, "grad_norm": 1.4525196552276611, "learning_rate": 1.1614122478732654e-05, "loss": 0.1590623140335083, "step": 22730 }, { "epoch": 0.4651757449920476, "grad_norm": 1.1004104614257812, "learning_rate": 1.160758368561608e-05, "loss": 0.15238893032073975, "step": 22740 }, { "epoch": 0.46538030776469147, "grad_norm": 1.3258740901947021, "learning_rate": 1.1601044186853554e-05, "loss": 0.15280331373214723, "step": 22750 }, { "epoch": 0.46558487053733527, "grad_norm": 1.5870336294174194, "learning_rate": 1.1594503985315579e-05, "loss": 0.15176445245742798, "step": 22760 }, { "epoch": 0.4657894333099791, "grad_norm": 0.7328482866287231, "learning_rate": 1.1587963083872959e-05, "loss": 0.15790609121322632, "step": 22770 }, { "epoch": 0.46599399608262293, "grad_norm": 1.399656057357788, "learning_rate": 1.1581421485396816e-05, "loss": 0.16787405014038087, "step": 22780 }, { "epoch": 0.46619855885526673, "grad_norm": 2.015800714492798, "learning_rate": 1.1574879192758572e-05, "loss": 0.1473463535308838, "step": 22790 }, { "epoch": 0.46640312162791053, "grad_norm": 1.9941242933273315, "learning_rate": 1.1568336208829948e-05, "loss": 0.1463654398918152, "step": 22800 }, { "epoch": 0.4666076844005544, "grad_norm": 1.6988775730133057, "learning_rate": 1.1561792536482986e-05, "loss": 0.15301687717437745, "step": 22810 }, { "epoch": 0.4668122471731982, "grad_norm": 2.2877488136291504, "learning_rate": 1.1555248178590008e-05, "loss": 0.14531898498535156, "step": 22820 }, { "epoch": 0.467016809945842, "grad_norm": 0.9909458756446838, "learning_rate": 1.1548703138023658e-05, "loss": 0.15101081132888794, "step": 22830 }, { "epoch": 0.46722137271848585, "grad_norm": 2.3711233139038086, "learning_rate": 1.1542157417656862e-05, "loss": 0.14375174045562744, "step": 22840 }, { "epoch": 0.46742593549112965, "grad_norm": 1.0611847639083862, "learning_rate": 1.1535611020362855e-05, "loss": 0.15457142591476442, "step": 22850 }, { "epoch": 0.46763049826377345, "grad_norm": 2.501171588897705, "learning_rate": 1.1529063949015172e-05, "loss": 0.1376083731651306, "step": 22860 }, { "epoch": 0.4678350610364173, "grad_norm": 1.171398639678955, "learning_rate": 1.152251620648763e-05, "loss": 0.14793174266815184, "step": 22870 }, { "epoch": 0.4680396238090611, "grad_norm": 4.17897367477417, "learning_rate": 1.1515967795654345e-05, "loss": 0.15629634857177735, "step": 22880 }, { "epoch": 0.4682441865817049, "grad_norm": 2.4553380012512207, "learning_rate": 1.1509418719389742e-05, "loss": 0.15158519744873047, "step": 22890 }, { "epoch": 0.46844874935434877, "grad_norm": 2.122321844100952, "learning_rate": 1.150286898056852e-05, "loss": 0.1566143035888672, "step": 22900 }, { "epoch": 0.46865331212699257, "grad_norm": 2.169347047805786, "learning_rate": 1.1496318582065677e-05, "loss": 0.14467045068740844, "step": 22910 }, { "epoch": 0.4688578748996364, "grad_norm": 1.857832670211792, "learning_rate": 1.1489767526756494e-05, "loss": 0.1581888437271118, "step": 22920 }, { "epoch": 0.46906243767228023, "grad_norm": 1.4407988786697388, "learning_rate": 1.148321581751655e-05, "loss": 0.14798028469085694, "step": 22930 }, { "epoch": 0.46926700044492403, "grad_norm": 2.5253031253814697, "learning_rate": 1.1476663457221705e-05, "loss": 0.17058870792388917, "step": 22940 }, { "epoch": 0.46947156321756783, "grad_norm": 1.2285168170928955, "learning_rate": 1.1470110448748102e-05, "loss": 0.147206711769104, "step": 22950 }, { "epoch": 0.4696761259902117, "grad_norm": 1.6796151399612427, "learning_rate": 1.1463556794972177e-05, "loss": 0.13362655639648438, "step": 22960 }, { "epoch": 0.4698806887628555, "grad_norm": 2.053062915802002, "learning_rate": 1.1457002498770644e-05, "loss": 0.14114096164703369, "step": 22970 }, { "epoch": 0.4700852515354993, "grad_norm": 1.6498734951019287, "learning_rate": 1.1450447563020498e-05, "loss": 0.15296504497528077, "step": 22980 }, { "epoch": 0.47028981430814315, "grad_norm": 1.0075814723968506, "learning_rate": 1.1443891990599012e-05, "loss": 0.15875459909439088, "step": 22990 }, { "epoch": 0.47049437708078695, "grad_norm": 3.5721869468688965, "learning_rate": 1.143733578438375e-05, "loss": 0.14186465740203857, "step": 23000 }, { "epoch": 0.47069893985343075, "grad_norm": 0.947544515132904, "learning_rate": 1.1430778947252543e-05, "loss": 0.14955708980560303, "step": 23010 }, { "epoch": 0.4709035026260746, "grad_norm": 2.9685075283050537, "learning_rate": 1.1424221482083504e-05, "loss": 0.15126945972442626, "step": 23020 }, { "epoch": 0.4711080653987184, "grad_norm": 1.3852752447128296, "learning_rate": 1.1417663391755016e-05, "loss": 0.16300132274627685, "step": 23030 }, { "epoch": 0.4713126281713622, "grad_norm": 1.908621907234192, "learning_rate": 1.1411104679145751e-05, "loss": 0.1542879104614258, "step": 23040 }, { "epoch": 0.47151719094400607, "grad_norm": 1.2470240592956543, "learning_rate": 1.1404545347134634e-05, "loss": 0.1440783977508545, "step": 23050 }, { "epoch": 0.47172175371664987, "grad_norm": 1.2339116334915161, "learning_rate": 1.1397985398600876e-05, "loss": 0.15530444383621217, "step": 23060 }, { "epoch": 0.4719263164892937, "grad_norm": 2.959122657775879, "learning_rate": 1.1391424836423956e-05, "loss": 0.15689677000045776, "step": 23070 }, { "epoch": 0.47213087926193753, "grad_norm": 1.8914650678634644, "learning_rate": 1.1384863663483619e-05, "loss": 0.1389837384223938, "step": 23080 }, { "epoch": 0.47233544203458133, "grad_norm": 4.400355339050293, "learning_rate": 1.1378301882659882e-05, "loss": 0.15097726583480836, "step": 23090 }, { "epoch": 0.47254000480722513, "grad_norm": 4.005594253540039, "learning_rate": 1.1371739496833023e-05, "loss": 0.15617747306823732, "step": 23100 }, { "epoch": 0.472744567579869, "grad_norm": 1.939961314201355, "learning_rate": 1.1365176508883588e-05, "loss": 0.16677327156066896, "step": 23110 }, { "epoch": 0.4729491303525128, "grad_norm": 4.681675434112549, "learning_rate": 1.1358612921692397e-05, "loss": 0.15454933643341065, "step": 23120 }, { "epoch": 0.4731536931251566, "grad_norm": 1.3616794347763062, "learning_rate": 1.135204873814052e-05, "loss": 0.1461155891418457, "step": 23130 }, { "epoch": 0.47335825589780045, "grad_norm": 2.0062289237976074, "learning_rate": 1.1345483961109292e-05, "loss": 0.18671102523803712, "step": 23140 }, { "epoch": 0.47356281867044425, "grad_norm": 1.413435935974121, "learning_rate": 1.1338918593480311e-05, "loss": 0.1592781662940979, "step": 23150 }, { "epoch": 0.47376738144308805, "grad_norm": 1.4708549976348877, "learning_rate": 1.1332352638135436e-05, "loss": 0.16542272567749022, "step": 23160 }, { "epoch": 0.4739719442157319, "grad_norm": 2.40427565574646, "learning_rate": 1.1325786097956776e-05, "loss": 0.1550566077232361, "step": 23170 }, { "epoch": 0.4741765069883757, "grad_norm": 1.882614016532898, "learning_rate": 1.1319218975826705e-05, "loss": 0.14604365825653076, "step": 23180 }, { "epoch": 0.4743810697610195, "grad_norm": 2.205280303955078, "learning_rate": 1.131265127462785e-05, "loss": 0.15553089380264282, "step": 23190 }, { "epoch": 0.47458563253366337, "grad_norm": 1.7434614896774292, "learning_rate": 1.1306082997243093e-05, "loss": 0.151763916015625, "step": 23200 }, { "epoch": 0.4747901953063072, "grad_norm": 2.79470157623291, "learning_rate": 1.1299514146555564e-05, "loss": 0.14931633472442626, "step": 23210 }, { "epoch": 0.47499475807895103, "grad_norm": 1.16139554977417, "learning_rate": 1.1292944725448647e-05, "loss": 0.156988263130188, "step": 23220 }, { "epoch": 0.47519932085159483, "grad_norm": 2.160153865814209, "learning_rate": 1.1286374736805983e-05, "loss": 0.14246680736541747, "step": 23230 }, { "epoch": 0.47540388362423863, "grad_norm": 2.6390674114227295, "learning_rate": 1.1279804183511454e-05, "loss": 0.13240689039230347, "step": 23240 }, { "epoch": 0.4756084463968825, "grad_norm": 3.923506259918213, "learning_rate": 1.1273233068449193e-05, "loss": 0.15105217695236206, "step": 23250 }, { "epoch": 0.4758130091695263, "grad_norm": 3.324162483215332, "learning_rate": 1.1266661394503573e-05, "loss": 0.1471993088722229, "step": 23260 }, { "epoch": 0.4760175719421701, "grad_norm": 1.6952316761016846, "learning_rate": 1.1260089164559229e-05, "loss": 0.1523300290107727, "step": 23270 }, { "epoch": 0.47622213471481395, "grad_norm": 1.6874322891235352, "learning_rate": 1.1253516381501019e-05, "loss": 0.14086391925811767, "step": 23280 }, { "epoch": 0.47642669748745775, "grad_norm": 4.782733917236328, "learning_rate": 1.124694304821406e-05, "loss": 0.16366202831268312, "step": 23290 }, { "epoch": 0.47663126026010155, "grad_norm": 2.22645902633667, "learning_rate": 1.1240369167583709e-05, "loss": 0.14845306873321534, "step": 23300 }, { "epoch": 0.4768358230327454, "grad_norm": 1.689258098602295, "learning_rate": 1.123379474249555e-05, "loss": 0.1472862958908081, "step": 23310 }, { "epoch": 0.4770403858053892, "grad_norm": 2.233253002166748, "learning_rate": 1.122721977583542e-05, "loss": 0.1513150930404663, "step": 23320 }, { "epoch": 0.477244948578033, "grad_norm": 2.0875678062438965, "learning_rate": 1.1220644270489386e-05, "loss": 0.1533881425857544, "step": 23330 }, { "epoch": 0.47744951135067687, "grad_norm": 1.9253661632537842, "learning_rate": 1.121406822934376e-05, "loss": 0.15004798173904418, "step": 23340 }, { "epoch": 0.47765407412332067, "grad_norm": 1.433323621749878, "learning_rate": 1.1207491655285079e-05, "loss": 0.15889480113983154, "step": 23350 }, { "epoch": 0.4778586368959645, "grad_norm": 1.766023874282837, "learning_rate": 1.1200914551200117e-05, "loss": 0.15757811069488525, "step": 23360 }, { "epoch": 0.47806319966860833, "grad_norm": 4.024032115936279, "learning_rate": 1.1194336919975881e-05, "loss": 0.15182281732559205, "step": 23370 }, { "epoch": 0.47826776244125213, "grad_norm": 1.108782172203064, "learning_rate": 1.1187758764499619e-05, "loss": 0.15767766237258912, "step": 23380 }, { "epoch": 0.47847232521389593, "grad_norm": 1.5615742206573486, "learning_rate": 1.118118008765879e-05, "loss": 0.14668002128601074, "step": 23390 }, { "epoch": 0.4786768879865398, "grad_norm": 2.4231269359588623, "learning_rate": 1.1174600892341098e-05, "loss": 0.15133187770843506, "step": 23400 }, { "epoch": 0.4788814507591836, "grad_norm": 2.152953624725342, "learning_rate": 1.1168021181434465e-05, "loss": 0.1475584864616394, "step": 23410 }, { "epoch": 0.4790860135318274, "grad_norm": 1.491510033607483, "learning_rate": 1.1161440957827053e-05, "loss": 0.15281176567077637, "step": 23420 }, { "epoch": 0.47929057630447125, "grad_norm": 3.1853175163269043, "learning_rate": 1.1154860224407225e-05, "loss": 0.1530160427093506, "step": 23430 }, { "epoch": 0.47949513907711505, "grad_norm": 2.5417211055755615, "learning_rate": 1.1148278984063589e-05, "loss": 0.16163251399993897, "step": 23440 }, { "epoch": 0.47969970184975885, "grad_norm": 1.1828701496124268, "learning_rate": 1.1141697239684973e-05, "loss": 0.15232641696929933, "step": 23450 }, { "epoch": 0.4799042646224027, "grad_norm": 1.1346876621246338, "learning_rate": 1.1135114994160416e-05, "loss": 0.15487043857574462, "step": 23460 }, { "epoch": 0.4801088273950465, "grad_norm": 2.8632450103759766, "learning_rate": 1.1128532250379183e-05, "loss": 0.14645401239395142, "step": 23470 }, { "epoch": 0.4803133901676903, "grad_norm": 1.3642600774765015, "learning_rate": 1.1121949011230757e-05, "loss": 0.16413424015045167, "step": 23480 }, { "epoch": 0.48051795294033417, "grad_norm": 1.537406325340271, "learning_rate": 1.1115365279604845e-05, "loss": 0.1552674412727356, "step": 23490 }, { "epoch": 0.48072251571297797, "grad_norm": 2.2251036167144775, "learning_rate": 1.1108781058391355e-05, "loss": 0.149982225894928, "step": 23500 }, { "epoch": 0.4809270784856218, "grad_norm": 1.7505650520324707, "learning_rate": 1.110219635048043e-05, "loss": 0.16070730686187745, "step": 23510 }, { "epoch": 0.48113164125826563, "grad_norm": 1.7418594360351562, "learning_rate": 1.1095611158762403e-05, "loss": 0.1511581063270569, "step": 23520 }, { "epoch": 0.48133620403090943, "grad_norm": 1.4329968690872192, "learning_rate": 1.1089025486127845e-05, "loss": 0.1675554871559143, "step": 23530 }, { "epoch": 0.48154076680355323, "grad_norm": 1.1913548707962036, "learning_rate": 1.1082439335467511e-05, "loss": 0.15028122663497925, "step": 23540 }, { "epoch": 0.4817453295761971, "grad_norm": 2.8947393894195557, "learning_rate": 1.1075852709672393e-05, "loss": 0.13918052911758422, "step": 23550 }, { "epoch": 0.4819498923488409, "grad_norm": 3.563530683517456, "learning_rate": 1.106926561163367e-05, "loss": 0.15739891529083253, "step": 23560 }, { "epoch": 0.4821544551214847, "grad_norm": 1.0061067342758179, "learning_rate": 1.1062678044242743e-05, "loss": 0.13870385885238648, "step": 23570 }, { "epoch": 0.48235901789412855, "grad_norm": 1.3848521709442139, "learning_rate": 1.1056090010391205e-05, "loss": 0.15491420030593872, "step": 23580 }, { "epoch": 0.48256358066677235, "grad_norm": 2.15932035446167, "learning_rate": 1.1049501512970869e-05, "loss": 0.16666197776794434, "step": 23590 }, { "epoch": 0.48276814343941615, "grad_norm": 2.021423101425171, "learning_rate": 1.104291255487374e-05, "loss": 0.160233473777771, "step": 23600 }, { "epoch": 0.48297270621206, "grad_norm": 2.2020044326782227, "learning_rate": 1.1036323138992031e-05, "loss": 0.1448265790939331, "step": 23610 }, { "epoch": 0.4831772689847038, "grad_norm": 1.4126173257827759, "learning_rate": 1.102973326821815e-05, "loss": 0.15309994220733641, "step": 23620 }, { "epoch": 0.4833818317573476, "grad_norm": 1.5755608081817627, "learning_rate": 1.1023142945444716e-05, "loss": 0.15258713960647582, "step": 23630 }, { "epoch": 0.48358639452999147, "grad_norm": 2.0080320835113525, "learning_rate": 1.1016552173564536e-05, "loss": 0.162902569770813, "step": 23640 }, { "epoch": 0.48379095730263527, "grad_norm": 1.5263265371322632, "learning_rate": 1.1009960955470615e-05, "loss": 0.15717817544937135, "step": 23650 }, { "epoch": 0.4839955200752791, "grad_norm": 4.082408428192139, "learning_rate": 1.1003369294056158e-05, "loss": 0.15276663303375243, "step": 23660 }, { "epoch": 0.48420008284792293, "grad_norm": 0.9501427412033081, "learning_rate": 1.0996777192214564e-05, "loss": 0.15382397174835205, "step": 23670 }, { "epoch": 0.48440464562056673, "grad_norm": 1.5568097829818726, "learning_rate": 1.0990184652839426e-05, "loss": 0.1495049476623535, "step": 23680 }, { "epoch": 0.4846092083932106, "grad_norm": 2.1445720195770264, "learning_rate": 1.098359167882452e-05, "loss": 0.15942578315734862, "step": 23690 }, { "epoch": 0.4848137711658544, "grad_norm": 2.180278778076172, "learning_rate": 1.0976998273063827e-05, "loss": 0.15590347051620485, "step": 23700 }, { "epoch": 0.4850183339384982, "grad_norm": 1.0552561283111572, "learning_rate": 1.0970404438451508e-05, "loss": 0.16111687421798707, "step": 23710 }, { "epoch": 0.48522289671114205, "grad_norm": 1.2456964254379272, "learning_rate": 1.0963810177881915e-05, "loss": 0.1542827606201172, "step": 23720 }, { "epoch": 0.48542745948378585, "grad_norm": 1.451084852218628, "learning_rate": 1.0957215494249585e-05, "loss": 0.15960079431533813, "step": 23730 }, { "epoch": 0.48563202225642965, "grad_norm": 3.59124755859375, "learning_rate": 1.0950620390449247e-05, "loss": 0.14581960439682007, "step": 23740 }, { "epoch": 0.4858365850290735, "grad_norm": 1.5614055395126343, "learning_rate": 1.0944024869375807e-05, "loss": 0.14483287334442138, "step": 23750 }, { "epoch": 0.4860411478017173, "grad_norm": 0.9371262192726135, "learning_rate": 1.0937428933924355e-05, "loss": 0.16086874008178711, "step": 23760 }, { "epoch": 0.4862457105743611, "grad_norm": 1.4623452425003052, "learning_rate": 1.0930832586990166e-05, "loss": 0.1650084972381592, "step": 23770 }, { "epoch": 0.48645027334700497, "grad_norm": 3.135225534439087, "learning_rate": 1.0924235831468697e-05, "loss": 0.14808785915374756, "step": 23780 }, { "epoch": 0.48665483611964877, "grad_norm": 0.9653758406639099, "learning_rate": 1.0917638670255581e-05, "loss": 0.15503541231155396, "step": 23790 }, { "epoch": 0.48685939889229257, "grad_norm": 1.0632565021514893, "learning_rate": 1.0911041106246632e-05, "loss": 0.14007892608642578, "step": 23800 }, { "epoch": 0.48706396166493643, "grad_norm": 2.6223185062408447, "learning_rate": 1.090444314233783e-05, "loss": 0.1639781951904297, "step": 23810 }, { "epoch": 0.48726852443758023, "grad_norm": 1.273026943206787, "learning_rate": 1.0897844781425351e-05, "loss": 0.15431749820709229, "step": 23820 }, { "epoch": 0.48747308721022403, "grad_norm": 3.2363553047180176, "learning_rate": 1.0891246026405526e-05, "loss": 0.14959239959716797, "step": 23830 }, { "epoch": 0.4876776499828679, "grad_norm": 1.0329649448394775, "learning_rate": 1.0884646880174868e-05, "loss": 0.15170977115631104, "step": 23840 }, { "epoch": 0.4878822127555117, "grad_norm": 1.563338279724121, "learning_rate": 1.0878047345630063e-05, "loss": 0.1468080997467041, "step": 23850 }, { "epoch": 0.4880867755281555, "grad_norm": 2.43656063079834, "learning_rate": 1.0871447425667962e-05, "loss": 0.1604165554046631, "step": 23860 }, { "epoch": 0.48829133830079935, "grad_norm": 5.750553607940674, "learning_rate": 1.0864847123185589e-05, "loss": 0.16180906295776368, "step": 23870 }, { "epoch": 0.48849590107344315, "grad_norm": 1.4912803173065186, "learning_rate": 1.0858246441080133e-05, "loss": 0.14535462856292725, "step": 23880 }, { "epoch": 0.48870046384608695, "grad_norm": 0.9325462579727173, "learning_rate": 1.0851645382248955e-05, "loss": 0.14906073808670045, "step": 23890 }, { "epoch": 0.4889050266187308, "grad_norm": 2.485401153564453, "learning_rate": 1.0845043949589575e-05, "loss": 0.15750203132629395, "step": 23900 }, { "epoch": 0.4891095893913746, "grad_norm": 1.7112332582473755, "learning_rate": 1.0838442145999684e-05, "loss": 0.16025631427764891, "step": 23910 }, { "epoch": 0.4893141521640184, "grad_norm": 1.9214507341384888, "learning_rate": 1.0831839974377123e-05, "loss": 0.15047091245651245, "step": 23920 }, { "epoch": 0.48951871493666227, "grad_norm": 2.341247320175171, "learning_rate": 1.0825237437619913e-05, "loss": 0.16430622339248657, "step": 23930 }, { "epoch": 0.48972327770930607, "grad_norm": 3.896634101867676, "learning_rate": 1.081863453862622e-05, "loss": 0.1387711524963379, "step": 23940 }, { "epoch": 0.48992784048194987, "grad_norm": 1.5774900913238525, "learning_rate": 1.0812031280294379e-05, "loss": 0.14713773727416993, "step": 23950 }, { "epoch": 0.49013240325459373, "grad_norm": 1.6167712211608887, "learning_rate": 1.0805427665522872e-05, "loss": 0.1757320284843445, "step": 23960 }, { "epoch": 0.49033696602723753, "grad_norm": 2.650892496109009, "learning_rate": 1.079882369721035e-05, "loss": 0.15725650787353515, "step": 23970 }, { "epoch": 0.49054152879988133, "grad_norm": 1.9368517398834229, "learning_rate": 1.0792219378255607e-05, "loss": 0.16318833827972412, "step": 23980 }, { "epoch": 0.4907460915725252, "grad_norm": 1.374526858329773, "learning_rate": 1.0785614711557602e-05, "loss": 0.14618990421295167, "step": 23990 }, { "epoch": 0.490950654345169, "grad_norm": 3.155329465866089, "learning_rate": 1.0779009700015437e-05, "loss": 0.15776455402374268, "step": 24000 }, { "epoch": 0.4911552171178128, "grad_norm": 1.0036568641662598, "learning_rate": 1.0772404346528376e-05, "loss": 0.15718982219696045, "step": 24010 }, { "epoch": 0.49135977989045665, "grad_norm": 2.5941786766052246, "learning_rate": 1.076579865399582e-05, "loss": 0.15791606903076172, "step": 24020 }, { "epoch": 0.49156434266310045, "grad_norm": 1.5596450567245483, "learning_rate": 1.075919262531733e-05, "loss": 0.15093146562576293, "step": 24030 }, { "epoch": 0.49176890543574425, "grad_norm": 2.5227134227752686, "learning_rate": 1.0752586263392612e-05, "loss": 0.14687862396240234, "step": 24040 }, { "epoch": 0.4919734682083881, "grad_norm": 3.619389057159424, "learning_rate": 1.0745979571121512e-05, "loss": 0.15535573959350585, "step": 24050 }, { "epoch": 0.4921780309810319, "grad_norm": 1.6418277025222778, "learning_rate": 1.0739372551404028e-05, "loss": 0.14397557973861694, "step": 24060 }, { "epoch": 0.4923825937536757, "grad_norm": 2.4425790309906006, "learning_rate": 1.0732765207140297e-05, "loss": 0.15041587352752686, "step": 24070 }, { "epoch": 0.49258715652631957, "grad_norm": 1.432869553565979, "learning_rate": 1.0726157541230609e-05, "loss": 0.16189243793487548, "step": 24080 }, { "epoch": 0.49279171929896337, "grad_norm": 2.888408660888672, "learning_rate": 1.0719549556575375e-05, "loss": 0.15184555053710938, "step": 24090 }, { "epoch": 0.4929962820716072, "grad_norm": 1.4904330968856812, "learning_rate": 1.0712941256075164e-05, "loss": 0.14574570655822755, "step": 24100 }, { "epoch": 0.49320084484425103, "grad_norm": 1.7921135425567627, "learning_rate": 1.0706332642630679e-05, "loss": 0.1502575159072876, "step": 24110 }, { "epoch": 0.49340540761689483, "grad_norm": 2.2281651496887207, "learning_rate": 1.0699723719142761e-05, "loss": 0.15252552032470704, "step": 24120 }, { "epoch": 0.49360997038953863, "grad_norm": 1.707601547241211, "learning_rate": 1.0693114488512376e-05, "loss": 0.15043267011642455, "step": 24130 }, { "epoch": 0.4938145331621825, "grad_norm": 1.8503894805908203, "learning_rate": 1.0686504953640642e-05, "loss": 0.14508955478668212, "step": 24140 }, { "epoch": 0.4940190959348263, "grad_norm": 2.7447590827941895, "learning_rate": 1.0679895117428798e-05, "loss": 0.1482865571975708, "step": 24150 }, { "epoch": 0.49422365870747015, "grad_norm": 1.344062328338623, "learning_rate": 1.0673284982778223e-05, "loss": 0.1388784408569336, "step": 24160 }, { "epoch": 0.49442822148011395, "grad_norm": 2.7471606731414795, "learning_rate": 1.066667455259042e-05, "loss": 0.14263211488723754, "step": 24170 }, { "epoch": 0.49463278425275775, "grad_norm": 4.50209379196167, "learning_rate": 1.066006382976703e-05, "loss": 0.14123013019561767, "step": 24180 }, { "epoch": 0.4948373470254016, "grad_norm": 1.9235825538635254, "learning_rate": 1.0653452817209811e-05, "loss": 0.14821609258651733, "step": 24190 }, { "epoch": 0.4950419097980454, "grad_norm": 1.6379320621490479, "learning_rate": 1.0646841517820663e-05, "loss": 0.14786989688873292, "step": 24200 }, { "epoch": 0.4952464725706892, "grad_norm": 1.7109426259994507, "learning_rate": 1.0640229934501596e-05, "loss": 0.14515237808227538, "step": 24210 }, { "epoch": 0.49545103534333307, "grad_norm": 1.7046195268630981, "learning_rate": 1.0633618070154759e-05, "loss": 0.1496081233024597, "step": 24220 }, { "epoch": 0.49565559811597687, "grad_norm": 1.5046120882034302, "learning_rate": 1.0627005927682417e-05, "loss": 0.1535443663597107, "step": 24230 }, { "epoch": 0.49586016088862067, "grad_norm": 2.092991590499878, "learning_rate": 1.0620393509986952e-05, "loss": 0.1617838501930237, "step": 24240 }, { "epoch": 0.49606472366126453, "grad_norm": 1.5946635007858276, "learning_rate": 1.0613780819970882e-05, "loss": 0.16017227172851561, "step": 24250 }, { "epoch": 0.49626928643390833, "grad_norm": 2.0969929695129395, "learning_rate": 1.060716786053683e-05, "loss": 0.13789160251617433, "step": 24260 }, { "epoch": 0.49647384920655213, "grad_norm": 3.0359957218170166, "learning_rate": 1.060055463458754e-05, "loss": 0.16987918615341185, "step": 24270 }, { "epoch": 0.496678411979196, "grad_norm": 1.8756686449050903, "learning_rate": 1.0593941145025882e-05, "loss": 0.1585860252380371, "step": 24280 }, { "epoch": 0.4968829747518398, "grad_norm": 1.4772788286209106, "learning_rate": 1.058732739475483e-05, "loss": 0.1469278335571289, "step": 24290 }, { "epoch": 0.4970875375244836, "grad_norm": 1.4477145671844482, "learning_rate": 1.0580713386677482e-05, "loss": 0.15283496379852296, "step": 24300 }, { "epoch": 0.49729210029712745, "grad_norm": 1.452884316444397, "learning_rate": 1.057409912369704e-05, "loss": 0.1403254747390747, "step": 24310 }, { "epoch": 0.49749666306977125, "grad_norm": 1.3025572299957275, "learning_rate": 1.0567484608716825e-05, "loss": 0.15323259830474853, "step": 24320 }, { "epoch": 0.49770122584241505, "grad_norm": 1.7845410108566284, "learning_rate": 1.056086984464027e-05, "loss": 0.15812839269638063, "step": 24330 }, { "epoch": 0.4979057886150589, "grad_norm": 1.5413849353790283, "learning_rate": 1.055425483437091e-05, "loss": 0.15163506269454957, "step": 24340 }, { "epoch": 0.4981103513877027, "grad_norm": 2.6919822692871094, "learning_rate": 1.054763958081239e-05, "loss": 0.14954395294189454, "step": 24350 }, { "epoch": 0.4983149141603465, "grad_norm": 2.6268484592437744, "learning_rate": 1.0541024086868465e-05, "loss": 0.1591383695602417, "step": 24360 }, { "epoch": 0.49851947693299037, "grad_norm": 2.470045804977417, "learning_rate": 1.0534408355443e-05, "loss": 0.16765115261077881, "step": 24370 }, { "epoch": 0.49872403970563417, "grad_norm": 2.2790045738220215, "learning_rate": 1.0527792389439951e-05, "loss": 0.13927677869796753, "step": 24380 }, { "epoch": 0.49892860247827797, "grad_norm": 1.4169044494628906, "learning_rate": 1.0521176191763385e-05, "loss": 0.15489563941955567, "step": 24390 }, { "epoch": 0.49913316525092183, "grad_norm": 1.2902538776397705, "learning_rate": 1.0514559765317474e-05, "loss": 0.15809811353683473, "step": 24400 }, { "epoch": 0.49933772802356563, "grad_norm": 2.594231128692627, "learning_rate": 1.0507943113006483e-05, "loss": 0.14121031761169434, "step": 24410 }, { "epoch": 0.49954229079620943, "grad_norm": 1.5759632587432861, "learning_rate": 1.050132623773478e-05, "loss": 0.14437971115112305, "step": 24420 }, { "epoch": 0.4997468535688533, "grad_norm": 2.4272005558013916, "learning_rate": 1.049470914240683e-05, "loss": 0.15091187953948976, "step": 24430 }, { "epoch": 0.4999514163414971, "grad_norm": 3.1904754638671875, "learning_rate": 1.04880918299272e-05, "loss": 0.15450352430343628, "step": 24440 }, { "epoch": 0.5001559791141409, "grad_norm": 1.4919952154159546, "learning_rate": 1.0481474303200544e-05, "loss": 0.13858959674835206, "step": 24450 }, { "epoch": 0.5003605418867847, "grad_norm": 3.614833354949951, "learning_rate": 1.0474856565131612e-05, "loss": 0.15695966482162477, "step": 24460 }, { "epoch": 0.5005651046594286, "grad_norm": 0.9744639992713928, "learning_rate": 1.0468238618625246e-05, "loss": 0.15462583303451538, "step": 24470 }, { "epoch": 0.5007696674320724, "grad_norm": 3.058028221130371, "learning_rate": 1.046162046658639e-05, "loss": 0.14308313131332398, "step": 24480 }, { "epoch": 0.5009742302047162, "grad_norm": 3.5375733375549316, "learning_rate": 1.045500211192006e-05, "loss": 0.1682592511177063, "step": 24490 }, { "epoch": 0.50117879297736, "grad_norm": 2.1513562202453613, "learning_rate": 1.0448383557531377e-05, "loss": 0.1515866756439209, "step": 24500 }, { "epoch": 0.5013833557500038, "grad_norm": 2.433215856552124, "learning_rate": 1.044176480632554e-05, "loss": 0.14356226921081544, "step": 24510 }, { "epoch": 0.5015879185226476, "grad_norm": 1.3518141508102417, "learning_rate": 1.0435145861207843e-05, "loss": 0.14168164730072022, "step": 24520 }, { "epoch": 0.5017924812952915, "grad_norm": 2.1470963954925537, "learning_rate": 1.0428526725083657e-05, "loss": 0.13812425136566162, "step": 24530 }, { "epoch": 0.5019970440679353, "grad_norm": 2.7136332988739014, "learning_rate": 1.0421907400858437e-05, "loss": 0.15608373880386353, "step": 24540 }, { "epoch": 0.5022016068405791, "grad_norm": 2.554481267929077, "learning_rate": 1.0415287891437726e-05, "loss": 0.15686495304107667, "step": 24550 }, { "epoch": 0.5024061696132229, "grad_norm": 2.1357309818267822, "learning_rate": 1.040866819972715e-05, "loss": 0.14707186222076415, "step": 24560 }, { "epoch": 0.5026107323858667, "grad_norm": 1.559985637664795, "learning_rate": 1.0402048328632403e-05, "loss": 0.1484889030456543, "step": 24570 }, { "epoch": 0.5028152951585105, "grad_norm": 2.7930314540863037, "learning_rate": 1.0395428281059268e-05, "loss": 0.1627239465713501, "step": 24580 }, { "epoch": 0.5030198579311544, "grad_norm": 3.1060423851013184, "learning_rate": 1.0388808059913606e-05, "loss": 0.15684888362884522, "step": 24590 }, { "epoch": 0.5032244207037982, "grad_norm": 2.8530004024505615, "learning_rate": 1.0382187668101348e-05, "loss": 0.15431748628616332, "step": 24600 }, { "epoch": 0.503428983476442, "grad_norm": 2.069997549057007, "learning_rate": 1.0375567108528506e-05, "loss": 0.1502068042755127, "step": 24610 }, { "epoch": 0.5036335462490859, "grad_norm": 0.9331933259963989, "learning_rate": 1.0368946384101157e-05, "loss": 0.14290299415588378, "step": 24620 }, { "epoch": 0.5038381090217297, "grad_norm": 2.810030698776245, "learning_rate": 1.0362325497725463e-05, "loss": 0.16200729608535766, "step": 24630 }, { "epoch": 0.5040426717943735, "grad_norm": 2.192915439605713, "learning_rate": 1.0355704452307644e-05, "loss": 0.1521608352661133, "step": 24640 }, { "epoch": 0.5042472345670174, "grad_norm": 1.3225152492523193, "learning_rate": 1.0349083250754001e-05, "loss": 0.14510881900787354, "step": 24650 }, { "epoch": 0.5044517973396612, "grad_norm": 1.7073116302490234, "learning_rate": 1.0342461895970894e-05, "loss": 0.16522070169448852, "step": 24660 }, { "epoch": 0.504656360112305, "grad_norm": 2.068251848220825, "learning_rate": 1.033584039086476e-05, "loss": 0.14302101135253906, "step": 24670 }, { "epoch": 0.5048609228849488, "grad_norm": 1.7340149879455566, "learning_rate": 1.0329218738342091e-05, "loss": 0.15028750896453857, "step": 24680 }, { "epoch": 0.5050654856575926, "grad_norm": 3.2702314853668213, "learning_rate": 1.0322596941309452e-05, "loss": 0.16486697196960448, "step": 24690 }, { "epoch": 0.5052700484302364, "grad_norm": 2.130218982696533, "learning_rate": 1.0315975002673465e-05, "loss": 0.16327651739120483, "step": 24700 }, { "epoch": 0.5054746112028803, "grad_norm": 0.9404845833778381, "learning_rate": 1.030935292534083e-05, "loss": 0.1497323751449585, "step": 24710 }, { "epoch": 0.5056791739755241, "grad_norm": 2.194937229156494, "learning_rate": 1.0302730712218283e-05, "loss": 0.15532805919647216, "step": 24720 }, { "epoch": 0.5058837367481679, "grad_norm": 1.6421886682510376, "learning_rate": 1.0296108366212638e-05, "loss": 0.15167320966720582, "step": 24730 }, { "epoch": 0.5060882995208117, "grad_norm": 2.0843095779418945, "learning_rate": 1.0289485890230761e-05, "loss": 0.13658907413482665, "step": 24740 }, { "epoch": 0.5062928622934555, "grad_norm": 2.396087408065796, "learning_rate": 1.0282863287179575e-05, "loss": 0.13886580467224122, "step": 24750 }, { "epoch": 0.5064974250660993, "grad_norm": 2.315354585647583, "learning_rate": 1.0276240559966059e-05, "loss": 0.1549270272254944, "step": 24760 }, { "epoch": 0.5067019878387432, "grad_norm": 1.6213504076004028, "learning_rate": 1.0269617711497248e-05, "loss": 0.15424106121063233, "step": 24770 }, { "epoch": 0.506906550611387, "grad_norm": 3.454822301864624, "learning_rate": 1.0262994744680235e-05, "loss": 0.1461993932723999, "step": 24780 }, { "epoch": 0.5071111133840308, "grad_norm": 1.5869792699813843, "learning_rate": 1.0256371662422152e-05, "loss": 0.14698395729064942, "step": 24790 }, { "epoch": 0.5073156761566746, "grad_norm": 1.7624175548553467, "learning_rate": 1.024974846763019e-05, "loss": 0.1648624300956726, "step": 24800 }, { "epoch": 0.5075202389293184, "grad_norm": 1.8952114582061768, "learning_rate": 1.024312516321159e-05, "loss": 0.13766515254974365, "step": 24810 }, { "epoch": 0.5077248017019622, "grad_norm": 1.511256217956543, "learning_rate": 1.0236501752073639e-05, "loss": 0.1533735513687134, "step": 24820 }, { "epoch": 0.5079293644746061, "grad_norm": 4.744894027709961, "learning_rate": 1.0229878237123672e-05, "loss": 0.151222562789917, "step": 24830 }, { "epoch": 0.5081339272472499, "grad_norm": 2.6836817264556885, "learning_rate": 1.0223254621269068e-05, "loss": 0.15621368885040282, "step": 24840 }, { "epoch": 0.5083384900198937, "grad_norm": 1.0246586799621582, "learning_rate": 1.0216630907417254e-05, "loss": 0.14813525676727296, "step": 24850 }, { "epoch": 0.5085430527925375, "grad_norm": 3.078878402709961, "learning_rate": 1.0210007098475691e-05, "loss": 0.14957504272460936, "step": 24860 }, { "epoch": 0.5087476155651813, "grad_norm": 1.2597236633300781, "learning_rate": 1.0203383197351893e-05, "loss": 0.15111931562423705, "step": 24870 }, { "epoch": 0.5089521783378252, "grad_norm": 2.914991855621338, "learning_rate": 1.0196759206953408e-05, "loss": 0.1584030270576477, "step": 24880 }, { "epoch": 0.509156741110469, "grad_norm": 1.7836757898330688, "learning_rate": 1.0190135130187826e-05, "loss": 0.14978058338165284, "step": 24890 }, { "epoch": 0.5093613038831128, "grad_norm": 1.4034514427185059, "learning_rate": 1.0183510969962772e-05, "loss": 0.1560114026069641, "step": 24900 }, { "epoch": 0.5095658666557566, "grad_norm": 1.5200003385543823, "learning_rate": 1.0176886729185906e-05, "loss": 0.1629801034927368, "step": 24910 }, { "epoch": 0.5097704294284005, "grad_norm": 3.295088052749634, "learning_rate": 1.0170262410764936e-05, "loss": 0.14483660459518433, "step": 24920 }, { "epoch": 0.5099749922010443, "grad_norm": 1.621128797531128, "learning_rate": 1.0163638017607585e-05, "loss": 0.1396289587020874, "step": 24930 }, { "epoch": 0.5101795549736882, "grad_norm": 1.0091108083724976, "learning_rate": 1.015701355262162e-05, "loss": 0.13103368282318115, "step": 24940 }, { "epoch": 0.510384117746332, "grad_norm": 3.3367202281951904, "learning_rate": 1.0150389018714844e-05, "loss": 0.15665783882141113, "step": 24950 }, { "epoch": 0.5105886805189758, "grad_norm": 3.381240129470825, "learning_rate": 1.0143764418795077e-05, "loss": 0.1326984643936157, "step": 24960 }, { "epoch": 0.5107932432916196, "grad_norm": 2.1771299839019775, "learning_rate": 1.013713975577018e-05, "loss": 0.1480647087097168, "step": 24970 }, { "epoch": 0.5109978060642634, "grad_norm": 1.980443000793457, "learning_rate": 1.0130515032548034e-05, "loss": 0.14932756423950194, "step": 24980 }, { "epoch": 0.5112023688369072, "grad_norm": 2.8916072845458984, "learning_rate": 1.0123890252036552e-05, "loss": 0.13671553134918213, "step": 24990 }, { "epoch": 0.5114069316095511, "grad_norm": 2.9219295978546143, "learning_rate": 1.0117265417143667e-05, "loss": 0.14206783771514891, "step": 25000 }, { "epoch": 0.5116114943821949, "grad_norm": 1.2385562658309937, "learning_rate": 1.011064053077734e-05, "loss": 0.15508687496185303, "step": 25010 }, { "epoch": 0.5118160571548387, "grad_norm": 2.541264057159424, "learning_rate": 1.0104015595845548e-05, "loss": 0.15234694480895997, "step": 25020 }, { "epoch": 0.5120206199274825, "grad_norm": 1.4572417736053467, "learning_rate": 1.0097390615256303e-05, "loss": 0.1670420289039612, "step": 25030 }, { "epoch": 0.5122251827001263, "grad_norm": 2.2110278606414795, "learning_rate": 1.0090765591917624e-05, "loss": 0.13766582012176515, "step": 25040 }, { "epoch": 0.5124297454727701, "grad_norm": 2.2832469940185547, "learning_rate": 1.0084140528737549e-05, "loss": 0.14827542304992675, "step": 25050 }, { "epoch": 0.512634308245414, "grad_norm": 2.0080630779266357, "learning_rate": 1.007751542862414e-05, "loss": 0.1532963991165161, "step": 25060 }, { "epoch": 0.5128388710180578, "grad_norm": 1.7737191915512085, "learning_rate": 1.0070890294485478e-05, "loss": 0.14237525463104247, "step": 25070 }, { "epoch": 0.5130434337907016, "grad_norm": 1.4714702367782593, "learning_rate": 1.0064265129229649e-05, "loss": 0.16793310642242432, "step": 25080 }, { "epoch": 0.5132479965633454, "grad_norm": 0.8678575158119202, "learning_rate": 1.0057639935764756e-05, "loss": 0.14706445932388307, "step": 25090 }, { "epoch": 0.5134525593359892, "grad_norm": 1.36239492893219, "learning_rate": 1.0051014716998915e-05, "loss": 0.14466177225112914, "step": 25100 }, { "epoch": 0.513657122108633, "grad_norm": 1.5015443563461304, "learning_rate": 1.004438947584026e-05, "loss": 0.15233192443847657, "step": 25110 }, { "epoch": 0.5138616848812769, "grad_norm": 1.2144553661346436, "learning_rate": 1.0037764215196923e-05, "loss": 0.1596190094947815, "step": 25120 }, { "epoch": 0.5140662476539207, "grad_norm": 2.372497320175171, "learning_rate": 1.0031138937977052e-05, "loss": 0.15887476205825807, "step": 25130 }, { "epoch": 0.5142708104265645, "grad_norm": 2.3293492794036865, "learning_rate": 1.0024513647088803e-05, "loss": 0.1605835199356079, "step": 25140 }, { "epoch": 0.5144753731992083, "grad_norm": 3.6987392902374268, "learning_rate": 1.0017888345440329e-05, "loss": 0.15182743072509766, "step": 25150 }, { "epoch": 0.5146799359718521, "grad_norm": 2.854356288909912, "learning_rate": 1.0011263035939802e-05, "loss": 0.15214970111846923, "step": 25160 }, { "epoch": 0.5148844987444959, "grad_norm": 3.0447440147399902, "learning_rate": 1.0004637721495382e-05, "loss": 0.15232977867126465, "step": 25170 }, { "epoch": 0.5150890615171398, "grad_norm": 0.826844334602356, "learning_rate": 9.998012405015246e-06, "loss": 0.13899580240249634, "step": 25180 }, { "epoch": 0.5152936242897836, "grad_norm": 2.948395252227783, "learning_rate": 9.991387089407562e-06, "loss": 0.16484766006469725, "step": 25190 }, { "epoch": 0.5154981870624274, "grad_norm": 1.4972420930862427, "learning_rate": 9.984761777580492e-06, "loss": 0.1429931640625, "step": 25200 }, { "epoch": 0.5157027498350713, "grad_norm": 2.3384361267089844, "learning_rate": 9.978136472442219e-06, "loss": 0.16276514530181885, "step": 25210 }, { "epoch": 0.515907312607715, "grad_norm": 1.8763206005096436, "learning_rate": 9.971511176900903e-06, "loss": 0.14000951051712035, "step": 25220 }, { "epoch": 0.5161118753803589, "grad_norm": 2.1846811771392822, "learning_rate": 9.964885893864702e-06, "loss": 0.14818212985992432, "step": 25230 }, { "epoch": 0.5163164381530028, "grad_norm": 1.687934160232544, "learning_rate": 9.95826062624177e-06, "loss": 0.13870376348495483, "step": 25240 }, { "epoch": 0.5165210009256466, "grad_norm": 1.6190356016159058, "learning_rate": 9.951635376940265e-06, "loss": 0.15940775871276855, "step": 25250 }, { "epoch": 0.5167255636982904, "grad_norm": 2.324876308441162, "learning_rate": 9.945010148868325e-06, "loss": 0.15278656482696534, "step": 25260 }, { "epoch": 0.5169301264709342, "grad_norm": 1.9810988903045654, "learning_rate": 9.93838494493408e-06, "loss": 0.1599063515663147, "step": 25270 }, { "epoch": 0.517134689243578, "grad_norm": 2.2147974967956543, "learning_rate": 9.93175976804565e-06, "loss": 0.14602117538452147, "step": 25280 }, { "epoch": 0.5173392520162218, "grad_norm": 1.6385263204574585, "learning_rate": 9.925134621111146e-06, "loss": 0.14868963956832887, "step": 25290 }, { "epoch": 0.5175438147888657, "grad_norm": 1.9422268867492676, "learning_rate": 9.91850950703867e-06, "loss": 0.13220471143722534, "step": 25300 }, { "epoch": 0.5177483775615095, "grad_norm": 3.2540621757507324, "learning_rate": 9.91188442873629e-06, "loss": 0.14894521236419678, "step": 25310 }, { "epoch": 0.5179529403341533, "grad_norm": 1.6824440956115723, "learning_rate": 9.905259389112087e-06, "loss": 0.14604580402374268, "step": 25320 }, { "epoch": 0.5181575031067971, "grad_norm": 1.1514886617660522, "learning_rate": 9.898634391074106e-06, "loss": 0.1566234350204468, "step": 25330 }, { "epoch": 0.5183620658794409, "grad_norm": 3.630298614501953, "learning_rate": 9.892009437530374e-06, "loss": 0.1394833207130432, "step": 25340 }, { "epoch": 0.5185666286520848, "grad_norm": 2.732518434524536, "learning_rate": 9.8853845313889e-06, "loss": 0.15807676315307617, "step": 25350 }, { "epoch": 0.5187711914247286, "grad_norm": 1.5870295763015747, "learning_rate": 9.87875967555768e-06, "loss": 0.1464979648590088, "step": 25360 }, { "epoch": 0.5189757541973724, "grad_norm": 1.059028148651123, "learning_rate": 9.872134872944686e-06, "loss": 0.14716806411743164, "step": 25370 }, { "epoch": 0.5191803169700162, "grad_norm": 2.033562183380127, "learning_rate": 9.865510126457856e-06, "loss": 0.1470654249191284, "step": 25380 }, { "epoch": 0.51938487974266, "grad_norm": 4.599057674407959, "learning_rate": 9.858885439005107e-06, "loss": 0.1462847352027893, "step": 25390 }, { "epoch": 0.5195894425153038, "grad_norm": 1.4720921516418457, "learning_rate": 9.852260813494347e-06, "loss": 0.14426298141479493, "step": 25400 }, { "epoch": 0.5197940052879477, "grad_norm": 1.2552800178527832, "learning_rate": 9.845636252833434e-06, "loss": 0.15787575244903565, "step": 25410 }, { "epoch": 0.5199985680605915, "grad_norm": 1.910134196281433, "learning_rate": 9.83901175993021e-06, "loss": 0.1619476079940796, "step": 25420 }, { "epoch": 0.5202031308332353, "grad_norm": 1.811896562576294, "learning_rate": 9.832387337692477e-06, "loss": 0.14627792835235595, "step": 25430 }, { "epoch": 0.5204076936058791, "grad_norm": 1.4860931634902954, "learning_rate": 9.825762989028025e-06, "loss": 0.14778532981872558, "step": 25440 }, { "epoch": 0.5206122563785229, "grad_norm": 2.137648582458496, "learning_rate": 9.819138716844593e-06, "loss": 0.15522074699401855, "step": 25450 }, { "epoch": 0.5208168191511667, "grad_norm": 2.830615997314453, "learning_rate": 9.812514524049894e-06, "loss": 0.1621718168258667, "step": 25460 }, { "epoch": 0.5210213819238106, "grad_norm": 3.5578277111053467, "learning_rate": 9.805890413551602e-06, "loss": 0.14575212001800536, "step": 25470 }, { "epoch": 0.5212259446964544, "grad_norm": 2.474271297454834, "learning_rate": 9.799266388257367e-06, "loss": 0.14190843105316162, "step": 25480 }, { "epoch": 0.5214305074690982, "grad_norm": 2.7939059734344482, "learning_rate": 9.792642451074787e-06, "loss": 0.14457858800888063, "step": 25490 }, { "epoch": 0.521635070241742, "grad_norm": 3.077867269515991, "learning_rate": 9.786018604911424e-06, "loss": 0.14344639778137208, "step": 25500 }, { "epoch": 0.5218396330143859, "grad_norm": 1.160791039466858, "learning_rate": 9.779394852674814e-06, "loss": 0.15483486652374268, "step": 25510 }, { "epoch": 0.5220441957870297, "grad_norm": 2.871281147003174, "learning_rate": 9.772771197272431e-06, "loss": 0.15614402294158936, "step": 25520 }, { "epoch": 0.5222487585596736, "grad_norm": 2.1550328731536865, "learning_rate": 9.766147641611721e-06, "loss": 0.14959588050842285, "step": 25530 }, { "epoch": 0.5224533213323174, "grad_norm": 1.4014708995819092, "learning_rate": 9.759524188600078e-06, "loss": 0.1455210566520691, "step": 25540 }, { "epoch": 0.5226578841049612, "grad_norm": 1.6717891693115234, "learning_rate": 9.752900841144863e-06, "loss": 0.15776684284210205, "step": 25550 }, { "epoch": 0.522862446877605, "grad_norm": 1.5867509841918945, "learning_rate": 9.746277602153375e-06, "loss": 0.15535433292388917, "step": 25560 }, { "epoch": 0.5230670096502488, "grad_norm": 1.091355800628662, "learning_rate": 9.739654474532875e-06, "loss": 0.1583601117134094, "step": 25570 }, { "epoch": 0.5232715724228926, "grad_norm": 2.9981741905212402, "learning_rate": 9.733031461190573e-06, "loss": 0.14258410930633544, "step": 25580 }, { "epoch": 0.5234761351955365, "grad_norm": 2.7577571868896484, "learning_rate": 9.726408565033628e-06, "loss": 0.15319302082061767, "step": 25590 }, { "epoch": 0.5236806979681803, "grad_norm": 3.496299982070923, "learning_rate": 9.719785788969149e-06, "loss": 0.15447765588760376, "step": 25600 }, { "epoch": 0.5238852607408241, "grad_norm": 1.1319304704666138, "learning_rate": 9.713163135904193e-06, "loss": 0.14618655443191528, "step": 25610 }, { "epoch": 0.5240898235134679, "grad_norm": 2.4092857837677, "learning_rate": 9.706540608745754e-06, "loss": 0.14016162157058715, "step": 25620 }, { "epoch": 0.5242943862861117, "grad_norm": 4.228646278381348, "learning_rate": 9.699918210400791e-06, "loss": 0.14142128229141235, "step": 25630 }, { "epoch": 0.5244989490587555, "grad_norm": 3.5814766883850098, "learning_rate": 9.693295943776184e-06, "loss": 0.14435701370239257, "step": 25640 }, { "epoch": 0.5247035118313994, "grad_norm": 0.9558322429656982, "learning_rate": 9.68667381177877e-06, "loss": 0.15285089015960693, "step": 25650 }, { "epoch": 0.5249080746040432, "grad_norm": 1.8438537120819092, "learning_rate": 9.680051817315316e-06, "loss": 0.14397051334381103, "step": 25660 }, { "epoch": 0.525112637376687, "grad_norm": 3.1677260398864746, "learning_rate": 9.673429963292541e-06, "loss": 0.1450897216796875, "step": 25670 }, { "epoch": 0.5253172001493308, "grad_norm": 2.5189006328582764, "learning_rate": 9.666808252617094e-06, "loss": 0.15114023685455322, "step": 25680 }, { "epoch": 0.5255217629219746, "grad_norm": 1.9116544723510742, "learning_rate": 9.660186688195562e-06, "loss": 0.14349857568740845, "step": 25690 }, { "epoch": 0.5257263256946184, "grad_norm": 2.0905070304870605, "learning_rate": 9.653565272934468e-06, "loss": 0.1390989303588867, "step": 25700 }, { "epoch": 0.5259308884672623, "grad_norm": 1.7185906171798706, "learning_rate": 9.646944009740272e-06, "loss": 0.14730790853500367, "step": 25710 }, { "epoch": 0.5261354512399061, "grad_norm": 2.265794515609741, "learning_rate": 9.640322901519365e-06, "loss": 0.15413321256637574, "step": 25720 }, { "epoch": 0.5263400140125499, "grad_norm": 1.3661683797836304, "learning_rate": 9.633701951178067e-06, "loss": 0.15581216812133789, "step": 25730 }, { "epoch": 0.5265445767851937, "grad_norm": 3.87410831451416, "learning_rate": 9.627081161622639e-06, "loss": 0.14794645309448243, "step": 25740 }, { "epoch": 0.5267491395578375, "grad_norm": 3.7900283336639404, "learning_rate": 9.62046053575926e-06, "loss": 0.14469220638275146, "step": 25750 }, { "epoch": 0.5269537023304813, "grad_norm": 1.7336715459823608, "learning_rate": 9.613840076494045e-06, "loss": 0.1438666820526123, "step": 25760 }, { "epoch": 0.5271582651031252, "grad_norm": 1.8922957181930542, "learning_rate": 9.607219786733025e-06, "loss": 0.13985056877136232, "step": 25770 }, { "epoch": 0.527362827875769, "grad_norm": 2.787465810775757, "learning_rate": 9.600599669382174e-06, "loss": 0.1412588357925415, "step": 25780 }, { "epoch": 0.5275673906484128, "grad_norm": 2.0925095081329346, "learning_rate": 9.593979727347374e-06, "loss": 0.14610350131988525, "step": 25790 }, { "epoch": 0.5277719534210566, "grad_norm": 1.2452112436294556, "learning_rate": 9.58735996353444e-06, "loss": 0.1516667604446411, "step": 25800 }, { "epoch": 0.5279765161937005, "grad_norm": 1.509514331817627, "learning_rate": 9.580740380849102e-06, "loss": 0.1436445951461792, "step": 25810 }, { "epoch": 0.5281810789663444, "grad_norm": 1.1632887125015259, "learning_rate": 9.574120982197019e-06, "loss": 0.14690672159194945, "step": 25820 }, { "epoch": 0.5283856417389882, "grad_norm": 1.6910810470581055, "learning_rate": 9.56750177048376e-06, "loss": 0.15254902839660645, "step": 25830 }, { "epoch": 0.528590204511632, "grad_norm": 2.1774685382843018, "learning_rate": 9.560882748614816e-06, "loss": 0.15854227542877197, "step": 25840 }, { "epoch": 0.5287947672842758, "grad_norm": 1.9063167572021484, "learning_rate": 9.554263919495596e-06, "loss": 0.1555425524711609, "step": 25850 }, { "epoch": 0.5289993300569196, "grad_norm": 1.440644383430481, "learning_rate": 9.547645286031423e-06, "loss": 0.14792338609695435, "step": 25860 }, { "epoch": 0.5292038928295634, "grad_norm": 1.6211647987365723, "learning_rate": 9.541026851127535e-06, "loss": 0.15630002021789552, "step": 25870 }, { "epoch": 0.5294084556022073, "grad_norm": 0.9897461533546448, "learning_rate": 9.534408617689078e-06, "loss": 0.15635173320770263, "step": 25880 }, { "epoch": 0.5296130183748511, "grad_norm": 1.2520085573196411, "learning_rate": 9.52779058862112e-06, "loss": 0.1574692726135254, "step": 25890 }, { "epoch": 0.5298175811474949, "grad_norm": 3.634322166442871, "learning_rate": 9.521172766828629e-06, "loss": 0.15113513469696044, "step": 25900 }, { "epoch": 0.5300221439201387, "grad_norm": 2.163731575012207, "learning_rate": 9.514555155216486e-06, "loss": 0.14602310657501222, "step": 25910 }, { "epoch": 0.5302267066927825, "grad_norm": 1.094274878501892, "learning_rate": 9.507937756689477e-06, "loss": 0.14659180641174316, "step": 25920 }, { "epoch": 0.5304312694654263, "grad_norm": 1.8971641063690186, "learning_rate": 9.501320574152304e-06, "loss": 0.15335285663604736, "step": 25930 }, { "epoch": 0.5306358322380702, "grad_norm": 0.7969937920570374, "learning_rate": 9.494703610509563e-06, "loss": 0.1503931164741516, "step": 25940 }, { "epoch": 0.530840395010714, "grad_norm": 1.2850377559661865, "learning_rate": 9.488086868665759e-06, "loss": 0.14706661701202392, "step": 25950 }, { "epoch": 0.5310449577833578, "grad_norm": 1.8169384002685547, "learning_rate": 9.481470351525297e-06, "loss": 0.14282081127166749, "step": 25960 }, { "epoch": 0.5312495205560016, "grad_norm": 3.0622050762176514, "learning_rate": 9.474854061992489e-06, "loss": 0.1381825566291809, "step": 25970 }, { "epoch": 0.5314540833286454, "grad_norm": 2.2263219356536865, "learning_rate": 9.46823800297154e-06, "loss": 0.14853882789611816, "step": 25980 }, { "epoch": 0.5316586461012892, "grad_norm": 1.5200244188308716, "learning_rate": 9.461622177366559e-06, "loss": 0.15647302865982055, "step": 25990 }, { "epoch": 0.5318632088739331, "grad_norm": 2.100013494491577, "learning_rate": 9.455006588081546e-06, "loss": 0.1396464705467224, "step": 26000 }, { "epoch": 0.5320677716465769, "grad_norm": 2.414034605026245, "learning_rate": 9.448391238020412e-06, "loss": 0.13984999656677247, "step": 26010 }, { "epoch": 0.5322723344192207, "grad_norm": 1.8173755407333374, "learning_rate": 9.441776130086943e-06, "loss": 0.15358933210372924, "step": 26020 }, { "epoch": 0.5324768971918645, "grad_norm": 2.1173479557037354, "learning_rate": 9.435161267184835e-06, "loss": 0.143957257270813, "step": 26030 }, { "epoch": 0.5326814599645083, "grad_norm": 1.4571723937988281, "learning_rate": 9.428546652217665e-06, "loss": 0.140824031829834, "step": 26040 }, { "epoch": 0.5328860227371521, "grad_norm": 2.2074179649353027, "learning_rate": 9.42193228808891e-06, "loss": 0.14825046062469482, "step": 26050 }, { "epoch": 0.533090585509796, "grad_norm": 1.3041566610336304, "learning_rate": 9.415318177701934e-06, "loss": 0.1541332483291626, "step": 26060 }, { "epoch": 0.5332951482824398, "grad_norm": 3.5291836261749268, "learning_rate": 9.408704323959985e-06, "loss": 0.1540430784225464, "step": 26070 }, { "epoch": 0.5334997110550836, "grad_norm": 3.355560064315796, "learning_rate": 9.402090729766206e-06, "loss": 0.13947741985321044, "step": 26080 }, { "epoch": 0.5337042738277274, "grad_norm": 2.552543878555298, "learning_rate": 9.395477398023618e-06, "loss": 0.14538434743881226, "step": 26090 }, { "epoch": 0.5339088366003713, "grad_norm": 2.8495428562164307, "learning_rate": 9.388864331635136e-06, "loss": 0.1497891902923584, "step": 26100 }, { "epoch": 0.534113399373015, "grad_norm": 3.0006508827209473, "learning_rate": 9.382251533503549e-06, "loss": 0.15693511962890624, "step": 26110 }, { "epoch": 0.534317962145659, "grad_norm": 3.472243547439575, "learning_rate": 9.375639006531534e-06, "loss": 0.14650589227676392, "step": 26120 }, { "epoch": 0.5345225249183028, "grad_norm": 2.6815099716186523, "learning_rate": 9.369026753621651e-06, "loss": 0.14962377548217773, "step": 26130 }, { "epoch": 0.5347270876909466, "grad_norm": 2.8945724964141846, "learning_rate": 9.362414777676333e-06, "loss": 0.1445850610733032, "step": 26140 }, { "epoch": 0.5349316504635904, "grad_norm": 2.6527578830718994, "learning_rate": 9.355803081597893e-06, "loss": 0.15336222648620607, "step": 26150 }, { "epoch": 0.5351362132362342, "grad_norm": 2.5123465061187744, "learning_rate": 9.34919166828853e-06, "loss": 0.15116209983825685, "step": 26160 }, { "epoch": 0.535340776008878, "grad_norm": 2.1579906940460205, "learning_rate": 9.342580540650306e-06, "loss": 0.14810500144958497, "step": 26170 }, { "epoch": 0.5355453387815219, "grad_norm": 3.2854623794555664, "learning_rate": 9.335969701585166e-06, "loss": 0.14117753505706787, "step": 26180 }, { "epoch": 0.5357499015541657, "grad_norm": 1.5961483716964722, "learning_rate": 9.329359153994922e-06, "loss": 0.13832032680511475, "step": 26190 }, { "epoch": 0.5359544643268095, "grad_norm": 1.7943816184997559, "learning_rate": 9.322748900781268e-06, "loss": 0.1553572654724121, "step": 26200 }, { "epoch": 0.5361590270994533, "grad_norm": 2.3498573303222656, "learning_rate": 9.31613894484576e-06, "loss": 0.14173638820648193, "step": 26210 }, { "epoch": 0.5363635898720971, "grad_norm": 1.3598304986953735, "learning_rate": 9.309529289089828e-06, "loss": 0.14193376302719116, "step": 26220 }, { "epoch": 0.5365681526447409, "grad_norm": 2.924802780151367, "learning_rate": 9.30291993641476e-06, "loss": 0.13427674770355225, "step": 26230 }, { "epoch": 0.5367727154173848, "grad_norm": 1.8299363851547241, "learning_rate": 9.296310889721729e-06, "loss": 0.13830604553222656, "step": 26240 }, { "epoch": 0.5369772781900286, "grad_norm": 1.6774296760559082, "learning_rate": 9.289702151911763e-06, "loss": 0.14546236991882325, "step": 26250 }, { "epoch": 0.5371818409626724, "grad_norm": 1.4337273836135864, "learning_rate": 9.283093725885748e-06, "loss": 0.15178132057189941, "step": 26260 }, { "epoch": 0.5373864037353162, "grad_norm": 1.8690083026885986, "learning_rate": 9.276485614544452e-06, "loss": 0.15555435419082642, "step": 26270 }, { "epoch": 0.53759096650796, "grad_norm": 1.0262233018875122, "learning_rate": 9.269877820788485e-06, "loss": 0.13745648860931398, "step": 26280 }, { "epoch": 0.5377955292806039, "grad_norm": 1.6446095705032349, "learning_rate": 9.26327034751833e-06, "loss": 0.1643434762954712, "step": 26290 }, { "epoch": 0.5380000920532477, "grad_norm": 2.8850131034851074, "learning_rate": 9.256663197634323e-06, "loss": 0.16050243377685547, "step": 26300 }, { "epoch": 0.5382046548258915, "grad_norm": 2.5824732780456543, "learning_rate": 9.250056374036662e-06, "loss": 0.1450740098953247, "step": 26310 }, { "epoch": 0.5384092175985353, "grad_norm": 1.1819854974746704, "learning_rate": 9.2434498796254e-06, "loss": 0.13515238761901854, "step": 26320 }, { "epoch": 0.5386137803711791, "grad_norm": 1.2001771926879883, "learning_rate": 9.236843717300448e-06, "loss": 0.14258272647857667, "step": 26330 }, { "epoch": 0.5388183431438229, "grad_norm": 4.429309368133545, "learning_rate": 9.230237889961563e-06, "loss": 0.15462729930877686, "step": 26340 }, { "epoch": 0.5390229059164668, "grad_norm": 1.640128493309021, "learning_rate": 9.223632400508365e-06, "loss": 0.15554593801498412, "step": 26350 }, { "epoch": 0.5392274686891106, "grad_norm": 1.139879584312439, "learning_rate": 9.217027251840324e-06, "loss": 0.1408859372138977, "step": 26360 }, { "epoch": 0.5394320314617544, "grad_norm": 1.7325992584228516, "learning_rate": 9.210422446856755e-06, "loss": 0.13616102933883667, "step": 26370 }, { "epoch": 0.5396365942343982, "grad_norm": 1.909394383430481, "learning_rate": 9.203817988456822e-06, "loss": 0.15803084373474122, "step": 26380 }, { "epoch": 0.539841157007042, "grad_norm": 1.763077735900879, "learning_rate": 9.197213879539547e-06, "loss": 0.1521664023399353, "step": 26390 }, { "epoch": 0.5400457197796859, "grad_norm": 1.9226620197296143, "learning_rate": 9.19061012300379e-06, "loss": 0.1414309024810791, "step": 26400 }, { "epoch": 0.5402502825523298, "grad_norm": 2.292762041091919, "learning_rate": 9.184006721748252e-06, "loss": 0.1421845316886902, "step": 26410 }, { "epoch": 0.5404548453249736, "grad_norm": 1.3838574886322021, "learning_rate": 9.177403678671491e-06, "loss": 0.14019293785095216, "step": 26420 }, { "epoch": 0.5406594080976174, "grad_norm": 1.6159336566925049, "learning_rate": 9.170800996671897e-06, "loss": 0.14231972694396972, "step": 26430 }, { "epoch": 0.5408639708702612, "grad_norm": 1.6191049814224243, "learning_rate": 9.16419867864771e-06, "loss": 0.14634296894073487, "step": 26440 }, { "epoch": 0.541068533642905, "grad_norm": 2.4243221282958984, "learning_rate": 9.157596727496997e-06, "loss": 0.13663430213928224, "step": 26450 }, { "epoch": 0.5412730964155488, "grad_norm": 1.0594935417175293, "learning_rate": 9.15099514611768e-06, "loss": 0.1445634961128235, "step": 26460 }, { "epoch": 0.5414776591881927, "grad_norm": 1.7005687952041626, "learning_rate": 9.14439393740751e-06, "loss": 0.14396491050720214, "step": 26470 }, { "epoch": 0.5416822219608365, "grad_norm": 1.5341980457305908, "learning_rate": 9.137793104264077e-06, "loss": 0.1444873809814453, "step": 26480 }, { "epoch": 0.5418867847334803, "grad_norm": 2.1261472702026367, "learning_rate": 9.131192649584795e-06, "loss": 0.1355496644973755, "step": 26490 }, { "epoch": 0.5420913475061241, "grad_norm": 2.2470474243164062, "learning_rate": 9.124592576266933e-06, "loss": 0.1356404185295105, "step": 26500 }, { "epoch": 0.5422959102787679, "grad_norm": 1.0284444093704224, "learning_rate": 9.11799288720758e-06, "loss": 0.14177210330963136, "step": 26510 }, { "epoch": 0.5425004730514117, "grad_norm": 5.007339954376221, "learning_rate": 9.111393585303652e-06, "loss": 0.1346570372581482, "step": 26520 }, { "epoch": 0.5427050358240556, "grad_norm": 1.6940791606903076, "learning_rate": 9.1047946734519e-06, "loss": 0.153302001953125, "step": 26530 }, { "epoch": 0.5429095985966994, "grad_norm": 1.0166832208633423, "learning_rate": 9.098196154548911e-06, "loss": 0.1441063404083252, "step": 26540 }, { "epoch": 0.5431141613693432, "grad_norm": 4.046903133392334, "learning_rate": 9.091598031491093e-06, "loss": 0.15074532032012938, "step": 26550 }, { "epoch": 0.543318724141987, "grad_norm": 3.087740182876587, "learning_rate": 9.085000307174676e-06, "loss": 0.15656840801239014, "step": 26560 }, { "epoch": 0.5435232869146308, "grad_norm": 1.2816277742385864, "learning_rate": 9.078402984495713e-06, "loss": 0.13049596548080444, "step": 26570 }, { "epoch": 0.5437278496872746, "grad_norm": 2.801152467727661, "learning_rate": 9.071806066350103e-06, "loss": 0.1329960346221924, "step": 26580 }, { "epoch": 0.5439324124599185, "grad_norm": 0.9909166693687439, "learning_rate": 9.06520955563354e-06, "loss": 0.13125417232513428, "step": 26590 }, { "epoch": 0.5441369752325623, "grad_norm": 3.1421167850494385, "learning_rate": 9.058613455241546e-06, "loss": 0.15548514127731322, "step": 26600 }, { "epoch": 0.5443415380052061, "grad_norm": 3.6385974884033203, "learning_rate": 9.05201776806948e-06, "loss": 0.12920129299163818, "step": 26610 }, { "epoch": 0.5445461007778499, "grad_norm": 2.0799527168273926, "learning_rate": 9.045422497012503e-06, "loss": 0.13799972534179689, "step": 26620 }, { "epoch": 0.5447506635504937, "grad_norm": 2.615668296813965, "learning_rate": 9.038827644965594e-06, "loss": 0.14706294536590575, "step": 26630 }, { "epoch": 0.5449552263231375, "grad_norm": 3.215198278427124, "learning_rate": 9.032233214823547e-06, "loss": 0.14860947132110597, "step": 26640 }, { "epoch": 0.5451597890957814, "grad_norm": 1.573262095451355, "learning_rate": 9.025639209480986e-06, "loss": 0.14592605829238892, "step": 26650 }, { "epoch": 0.5453643518684252, "grad_norm": 1.5701700448989868, "learning_rate": 9.019045631832336e-06, "loss": 0.14927952289581298, "step": 26660 }, { "epoch": 0.545568914641069, "grad_norm": 1.5106228590011597, "learning_rate": 9.012452484771829e-06, "loss": 0.14256925582885743, "step": 26670 }, { "epoch": 0.5457734774137128, "grad_norm": 1.3244032859802246, "learning_rate": 9.005859771193514e-06, "loss": 0.12963927984237672, "step": 26680 }, { "epoch": 0.5459780401863567, "grad_norm": 4.142077922821045, "learning_rate": 8.999267493991263e-06, "loss": 0.15913569927215576, "step": 26690 }, { "epoch": 0.5461826029590005, "grad_norm": 1.036234974861145, "learning_rate": 8.992675656058733e-06, "loss": 0.1417352080345154, "step": 26700 }, { "epoch": 0.5463871657316444, "grad_norm": 1.560770034790039, "learning_rate": 8.986084260289404e-06, "loss": 0.13345156908035277, "step": 26710 }, { "epoch": 0.5465917285042882, "grad_norm": 1.2556002140045166, "learning_rate": 8.979493309576551e-06, "loss": 0.14552011489868164, "step": 26720 }, { "epoch": 0.546796291276932, "grad_norm": 2.233091354370117, "learning_rate": 8.972902806813272e-06, "loss": 0.13730562925338746, "step": 26730 }, { "epoch": 0.5470008540495758, "grad_norm": 3.906418800354004, "learning_rate": 8.966312754892448e-06, "loss": 0.144132661819458, "step": 26740 }, { "epoch": 0.5472054168222196, "grad_norm": 2.1705031394958496, "learning_rate": 8.95972315670677e-06, "loss": 0.14680149555206298, "step": 26750 }, { "epoch": 0.5474099795948635, "grad_norm": 1.3124885559082031, "learning_rate": 8.95313401514873e-06, "loss": 0.1553375244140625, "step": 26760 }, { "epoch": 0.5476145423675073, "grad_norm": 2.479705572128296, "learning_rate": 8.946545333110626e-06, "loss": 0.14177865982055665, "step": 26770 }, { "epoch": 0.5478191051401511, "grad_norm": 1.2468239068984985, "learning_rate": 8.93995711348454e-06, "loss": 0.1343381404876709, "step": 26780 }, { "epoch": 0.5480236679127949, "grad_norm": 2.7841460704803467, "learning_rate": 8.933369359162361e-06, "loss": 0.13055081367492677, "step": 26790 }, { "epoch": 0.5482282306854387, "grad_norm": 1.045581579208374, "learning_rate": 8.926782073035782e-06, "loss": 0.12859828472137452, "step": 26800 }, { "epoch": 0.5484327934580825, "grad_norm": 0.9199334979057312, "learning_rate": 8.920195257996267e-06, "loss": 0.13291633129119873, "step": 26810 }, { "epoch": 0.5486373562307264, "grad_norm": 4.724035263061523, "learning_rate": 8.91360891693509e-06, "loss": 0.14839136600494385, "step": 26820 }, { "epoch": 0.5488419190033702, "grad_norm": 1.4291012287139893, "learning_rate": 8.907023052743314e-06, "loss": 0.1542038917541504, "step": 26830 }, { "epoch": 0.549046481776014, "grad_norm": 1.071242332458496, "learning_rate": 8.900437668311799e-06, "loss": 0.1504976511001587, "step": 26840 }, { "epoch": 0.5492510445486578, "grad_norm": 1.0257326364517212, "learning_rate": 8.89385276653118e-06, "loss": 0.14130032062530518, "step": 26850 }, { "epoch": 0.5494556073213016, "grad_norm": 1.628620982170105, "learning_rate": 8.88726835029189e-06, "loss": 0.14336576461791992, "step": 26860 }, { "epoch": 0.5496601700939454, "grad_norm": 1.8397676944732666, "learning_rate": 8.880684422484141e-06, "loss": 0.1550780177116394, "step": 26870 }, { "epoch": 0.5498647328665893, "grad_norm": 2.1933231353759766, "learning_rate": 8.874100985997945e-06, "loss": 0.1583791494369507, "step": 26880 }, { "epoch": 0.5500692956392331, "grad_norm": 2.5708200931549072, "learning_rate": 8.867518043723087e-06, "loss": 0.13979251384735109, "step": 26890 }, { "epoch": 0.5502738584118769, "grad_norm": 0.9252556562423706, "learning_rate": 8.860935598549136e-06, "loss": 0.14528336524963378, "step": 26900 }, { "epoch": 0.5504784211845207, "grad_norm": 1.1075152158737183, "learning_rate": 8.854353653365443e-06, "loss": 0.1386043667793274, "step": 26910 }, { "epoch": 0.5506829839571645, "grad_norm": 1.7123295068740845, "learning_rate": 8.847772211061146e-06, "loss": 0.16106693744659423, "step": 26920 }, { "epoch": 0.5508875467298083, "grad_norm": 1.2496294975280762, "learning_rate": 8.841191274525152e-06, "loss": 0.1452459216117859, "step": 26930 }, { "epoch": 0.5510921095024522, "grad_norm": 2.6190547943115234, "learning_rate": 8.834610846646154e-06, "loss": 0.14083982706069947, "step": 26940 }, { "epoch": 0.551296672275096, "grad_norm": 2.551565170288086, "learning_rate": 8.828030930312616e-06, "loss": 0.13828504085540771, "step": 26950 }, { "epoch": 0.5515012350477398, "grad_norm": 2.3341524600982666, "learning_rate": 8.821451528412786e-06, "loss": 0.15063272714614867, "step": 26960 }, { "epoch": 0.5517057978203836, "grad_norm": 2.3598458766937256, "learning_rate": 8.814872643834676e-06, "loss": 0.1395152688026428, "step": 26970 }, { "epoch": 0.5519103605930274, "grad_norm": 1.2633455991744995, "learning_rate": 8.808294279466075e-06, "loss": 0.14176172018051147, "step": 26980 }, { "epoch": 0.5521149233656713, "grad_norm": 2.817859172821045, "learning_rate": 8.801716438194548e-06, "loss": 0.1433631658554077, "step": 26990 }, { "epoch": 0.5523194861383152, "grad_norm": 1.0453029870986938, "learning_rate": 8.795139122907423e-06, "loss": 0.141844642162323, "step": 27000 }, { "epoch": 0.552524048910959, "grad_norm": 1.9895884990692139, "learning_rate": 8.788562336491804e-06, "loss": 0.13572468757629394, "step": 27010 }, { "epoch": 0.5527286116836028, "grad_norm": 1.7295989990234375, "learning_rate": 8.781986081834553e-06, "loss": 0.1441339373588562, "step": 27020 }, { "epoch": 0.5529331744562466, "grad_norm": 2.591198682785034, "learning_rate": 8.775410361822314e-06, "loss": 0.16288392543792723, "step": 27030 }, { "epoch": 0.5531377372288904, "grad_norm": 3.094557523727417, "learning_rate": 8.768835179341483e-06, "loss": 0.13827221393585204, "step": 27040 }, { "epoch": 0.5533423000015342, "grad_norm": 1.8355046510696411, "learning_rate": 8.762260537278223e-06, "loss": 0.1541795015335083, "step": 27050 }, { "epoch": 0.5535468627741781, "grad_norm": 3.2747316360473633, "learning_rate": 8.755686438518462e-06, "loss": 0.1372392177581787, "step": 27060 }, { "epoch": 0.5537514255468219, "grad_norm": 1.3758375644683838, "learning_rate": 8.74911288594789e-06, "loss": 0.12616467475891113, "step": 27070 }, { "epoch": 0.5539559883194657, "grad_norm": 1.6312592029571533, "learning_rate": 8.742539882451957e-06, "loss": 0.1495019793510437, "step": 27080 }, { "epoch": 0.5541605510921095, "grad_norm": 1.5256547927856445, "learning_rate": 8.735967430915868e-06, "loss": 0.14494088888168336, "step": 27090 }, { "epoch": 0.5543651138647533, "grad_norm": 0.9909873008728027, "learning_rate": 8.729395534224589e-06, "loss": 0.1347351551055908, "step": 27100 }, { "epoch": 0.5545696766373971, "grad_norm": 1.45493483543396, "learning_rate": 8.722824195262847e-06, "loss": 0.15245454311370848, "step": 27110 }, { "epoch": 0.554774239410041, "grad_norm": 1.1073857545852661, "learning_rate": 8.716253416915114e-06, "loss": 0.1482866883277893, "step": 27120 }, { "epoch": 0.5549788021826848, "grad_norm": 1.495032548904419, "learning_rate": 8.709683202065623e-06, "loss": 0.13552048206329345, "step": 27130 }, { "epoch": 0.5551833649553286, "grad_norm": 2.36287522315979, "learning_rate": 8.703113553598356e-06, "loss": 0.13303403854370116, "step": 27140 }, { "epoch": 0.5553879277279724, "grad_norm": 2.092292308807373, "learning_rate": 8.696544474397052e-06, "loss": 0.14969909191131592, "step": 27150 }, { "epoch": 0.5555924905006162, "grad_norm": 2.2035858631134033, "learning_rate": 8.689975967345197e-06, "loss": 0.15338413715362548, "step": 27160 }, { "epoch": 0.55579705327326, "grad_norm": 2.640188694000244, "learning_rate": 8.68340803532602e-06, "loss": 0.15862537622451783, "step": 27170 }, { "epoch": 0.5560016160459039, "grad_norm": 3.2934114933013916, "learning_rate": 8.676840681222506e-06, "loss": 0.15190526247024536, "step": 27180 }, { "epoch": 0.5562061788185477, "grad_norm": 3.269451856613159, "learning_rate": 8.670273907917386e-06, "loss": 0.1590815305709839, "step": 27190 }, { "epoch": 0.5564107415911915, "grad_norm": 1.040513277053833, "learning_rate": 8.663707718293128e-06, "loss": 0.1294699192047119, "step": 27200 }, { "epoch": 0.5566153043638353, "grad_norm": 1.527409315109253, "learning_rate": 8.65714211523195e-06, "loss": 0.1555829405784607, "step": 27210 }, { "epoch": 0.5568198671364791, "grad_norm": 1.8376716375350952, "learning_rate": 8.650577101615815e-06, "loss": 0.14142136573791503, "step": 27220 }, { "epoch": 0.557024429909123, "grad_norm": 1.6688014268875122, "learning_rate": 8.644012680326423e-06, "loss": 0.14626712799072267, "step": 27230 }, { "epoch": 0.5572289926817668, "grad_norm": 1.8963282108306885, "learning_rate": 8.637448854245214e-06, "loss": 0.14169785976409913, "step": 27240 }, { "epoch": 0.5574335554544106, "grad_norm": 2.4516396522521973, "learning_rate": 8.630885626253365e-06, "loss": 0.14638185501098633, "step": 27250 }, { "epoch": 0.5576381182270544, "grad_norm": 2.3825299739837646, "learning_rate": 8.624322999231796e-06, "loss": 0.14769102334976197, "step": 27260 }, { "epoch": 0.5578426809996982, "grad_norm": 2.2920455932617188, "learning_rate": 8.61776097606116e-06, "loss": 0.16153476238250733, "step": 27270 }, { "epoch": 0.558047243772342, "grad_norm": 0.8199582695960999, "learning_rate": 8.611199559621846e-06, "loss": 0.1454608917236328, "step": 27280 }, { "epoch": 0.558251806544986, "grad_norm": 1.0610754489898682, "learning_rate": 8.604638752793974e-06, "loss": 0.1451387882232666, "step": 27290 }, { "epoch": 0.5584563693176298, "grad_norm": 3.5576610565185547, "learning_rate": 8.598078558457399e-06, "loss": 0.14639201164245605, "step": 27300 }, { "epoch": 0.5586609320902736, "grad_norm": 2.579381227493286, "learning_rate": 8.591518979491707e-06, "loss": 0.13592488765716554, "step": 27310 }, { "epoch": 0.5588654948629174, "grad_norm": 2.7936646938323975, "learning_rate": 8.584960018776214e-06, "loss": 0.1439821481704712, "step": 27320 }, { "epoch": 0.5590700576355612, "grad_norm": 1.6835155487060547, "learning_rate": 8.578401679189959e-06, "loss": 0.14584112167358398, "step": 27330 }, { "epoch": 0.559274620408205, "grad_norm": 2.2649760246276855, "learning_rate": 8.57184396361172e-06, "loss": 0.15130093097686767, "step": 27340 }, { "epoch": 0.5594791831808489, "grad_norm": 0.8855767250061035, "learning_rate": 8.565286874919992e-06, "loss": 0.13667545318603516, "step": 27350 }, { "epoch": 0.5596837459534927, "grad_norm": 1.2117124795913696, "learning_rate": 8.558730415992993e-06, "loss": 0.14133294820785522, "step": 27360 }, { "epoch": 0.5598883087261365, "grad_norm": 1.3014973402023315, "learning_rate": 8.552174589708678e-06, "loss": 0.15654062032699584, "step": 27370 }, { "epoch": 0.5600928714987803, "grad_norm": 1.6077297925949097, "learning_rate": 8.545619398944707e-06, "loss": 0.13231464624404907, "step": 27380 }, { "epoch": 0.5602974342714241, "grad_norm": 1.3931525945663452, "learning_rate": 8.539064846578471e-06, "loss": 0.14110898971557617, "step": 27390 }, { "epoch": 0.5605019970440679, "grad_norm": 2.7827789783477783, "learning_rate": 8.532510935487079e-06, "loss": 0.138644802570343, "step": 27400 }, { "epoch": 0.5607065598167118, "grad_norm": 2.520470380783081, "learning_rate": 8.52595766854736e-06, "loss": 0.14032679796218872, "step": 27410 }, { "epoch": 0.5609111225893556, "grad_norm": 1.174377679824829, "learning_rate": 8.51940504863586e-06, "loss": 0.1397937536239624, "step": 27420 }, { "epoch": 0.5611156853619994, "grad_norm": 2.5165176391601562, "learning_rate": 8.512853078628836e-06, "loss": 0.15607714653015137, "step": 27430 }, { "epoch": 0.5613202481346432, "grad_norm": 1.4462144374847412, "learning_rate": 8.506301761402263e-06, "loss": 0.16055208444595337, "step": 27440 }, { "epoch": 0.561524810907287, "grad_norm": 1.472350835800171, "learning_rate": 8.499751099831834e-06, "loss": 0.13978997468948365, "step": 27450 }, { "epoch": 0.5617293736799308, "grad_norm": 2.002704620361328, "learning_rate": 8.493201096792948e-06, "loss": 0.1368165969848633, "step": 27460 }, { "epoch": 0.5619339364525747, "grad_norm": 2.674475908279419, "learning_rate": 8.486651755160717e-06, "loss": 0.13803924322128297, "step": 27470 }, { "epoch": 0.5621384992252185, "grad_norm": 1.87522292137146, "learning_rate": 8.48010307780996e-06, "loss": 0.15195765495300292, "step": 27480 }, { "epoch": 0.5623430619978623, "grad_norm": 1.9092339277267456, "learning_rate": 8.473555067615211e-06, "loss": 0.1397395610809326, "step": 27490 }, { "epoch": 0.5625476247705061, "grad_norm": 1.8611197471618652, "learning_rate": 8.467007727450707e-06, "loss": 0.14227166175842285, "step": 27500 }, { "epoch": 0.5627521875431499, "grad_norm": 1.9555526971817017, "learning_rate": 8.460461060190386e-06, "loss": 0.14803937673568726, "step": 27510 }, { "epoch": 0.5629567503157937, "grad_norm": 1.866774082183838, "learning_rate": 8.453915068707902e-06, "loss": 0.14333770275115967, "step": 27520 }, { "epoch": 0.5631613130884376, "grad_norm": 1.7105339765548706, "learning_rate": 8.447369755876607e-06, "loss": 0.14547946453094482, "step": 27530 }, { "epoch": 0.5633658758610814, "grad_norm": 4.473988056182861, "learning_rate": 8.440825124569546e-06, "loss": 0.15063745975494386, "step": 27540 }, { "epoch": 0.5635704386337252, "grad_norm": 2.4788739681243896, "learning_rate": 8.43428117765948e-06, "loss": 0.15319100618362427, "step": 27550 }, { "epoch": 0.563775001406369, "grad_norm": 1.5518476963043213, "learning_rate": 8.42773791801886e-06, "loss": 0.14978222846984862, "step": 27560 }, { "epoch": 0.5639795641790128, "grad_norm": 2.059217929840088, "learning_rate": 8.421195348519842e-06, "loss": 0.14222640991210939, "step": 27570 }, { "epoch": 0.5641841269516567, "grad_norm": 1.406249761581421, "learning_rate": 8.414653472034273e-06, "loss": 0.15562785863876344, "step": 27580 }, { "epoch": 0.5643886897243006, "grad_norm": 0.9494521021842957, "learning_rate": 8.408112291433687e-06, "loss": 0.1489521861076355, "step": 27590 }, { "epoch": 0.5645932524969444, "grad_norm": 2.12808895111084, "learning_rate": 8.40157180958934e-06, "loss": 0.13333969116210936, "step": 27600 }, { "epoch": 0.5647978152695882, "grad_norm": 1.9228136539459229, "learning_rate": 8.39503202937216e-06, "loss": 0.13379679918289183, "step": 27610 }, { "epoch": 0.565002378042232, "grad_norm": 1.7356725931167603, "learning_rate": 8.388492953652766e-06, "loss": 0.12490259408950806, "step": 27620 }, { "epoch": 0.5652069408148758, "grad_norm": 2.1537649631500244, "learning_rate": 8.381954585301475e-06, "loss": 0.1574362635612488, "step": 27630 }, { "epoch": 0.5654115035875196, "grad_norm": 2.3144407272338867, "learning_rate": 8.375416927188297e-06, "loss": 0.1421681523323059, "step": 27640 }, { "epoch": 0.5656160663601635, "grad_norm": 1.2837461233139038, "learning_rate": 8.368879982182922e-06, "loss": 0.15322203636169435, "step": 27650 }, { "epoch": 0.5658206291328073, "grad_norm": 1.9572871923446655, "learning_rate": 8.362343753154731e-06, "loss": 0.15038504600524902, "step": 27660 }, { "epoch": 0.5660251919054511, "grad_norm": 2.9734528064727783, "learning_rate": 8.355808242972786e-06, "loss": 0.14243035316467284, "step": 27670 }, { "epoch": 0.5662297546780949, "grad_norm": 0.9408449530601501, "learning_rate": 8.349273454505842e-06, "loss": 0.13891241550445557, "step": 27680 }, { "epoch": 0.5664343174507387, "grad_norm": 3.004117727279663, "learning_rate": 8.342739390622333e-06, "loss": 0.15035046339035035, "step": 27690 }, { "epoch": 0.5666388802233826, "grad_norm": 2.022489547729492, "learning_rate": 8.336206054190367e-06, "loss": 0.13795628547668456, "step": 27700 }, { "epoch": 0.5668434429960264, "grad_norm": 1.4928399324417114, "learning_rate": 8.329673448077751e-06, "loss": 0.13686649799346923, "step": 27710 }, { "epoch": 0.5670480057686702, "grad_norm": 1.0523717403411865, "learning_rate": 8.323141575151957e-06, "loss": 0.1416727662086487, "step": 27720 }, { "epoch": 0.567252568541314, "grad_norm": 0.7523421049118042, "learning_rate": 8.316610438280136e-06, "loss": 0.14042131900787352, "step": 27730 }, { "epoch": 0.5674571313139578, "grad_norm": 2.7183496952056885, "learning_rate": 8.310080040329116e-06, "loss": 0.1429669141769409, "step": 27740 }, { "epoch": 0.5676616940866016, "grad_norm": 1.0157064199447632, "learning_rate": 8.303550384165411e-06, "loss": 0.13190155029296874, "step": 27750 }, { "epoch": 0.5678662568592455, "grad_norm": 2.428396701812744, "learning_rate": 8.2970214726552e-06, "loss": 0.14873602390289306, "step": 27760 }, { "epoch": 0.5680708196318893, "grad_norm": 1.8867446184158325, "learning_rate": 8.290493308664334e-06, "loss": 0.15122690200805664, "step": 27770 }, { "epoch": 0.5682753824045331, "grad_norm": 1.34743332862854, "learning_rate": 8.283965895058335e-06, "loss": 0.12874925136566162, "step": 27780 }, { "epoch": 0.5684799451771769, "grad_norm": 1.7079190015792847, "learning_rate": 8.27743923470241e-06, "loss": 0.13416361808776855, "step": 27790 }, { "epoch": 0.5686845079498207, "grad_norm": 1.6522154808044434, "learning_rate": 8.270913330461418e-06, "loss": 0.13115452527999877, "step": 27800 }, { "epoch": 0.5688890707224645, "grad_norm": 2.037475109100342, "learning_rate": 8.264388185199895e-06, "loss": 0.17198560237884522, "step": 27810 }, { "epoch": 0.5690936334951084, "grad_norm": 1.5533654689788818, "learning_rate": 8.257863801782036e-06, "loss": 0.15641295909881592, "step": 27820 }, { "epoch": 0.5692981962677522, "grad_norm": 2.184884786605835, "learning_rate": 8.251340183071714e-06, "loss": 0.1570178747177124, "step": 27830 }, { "epoch": 0.569502759040396, "grad_norm": 1.938044786453247, "learning_rate": 8.244817331932462e-06, "loss": 0.14585145711898803, "step": 27840 }, { "epoch": 0.5697073218130398, "grad_norm": 2.49876070022583, "learning_rate": 8.238295251227467e-06, "loss": 0.14377617835998535, "step": 27850 }, { "epoch": 0.5699118845856836, "grad_norm": 2.6036174297332764, "learning_rate": 8.231773943819583e-06, "loss": 0.14813539981842042, "step": 27860 }, { "epoch": 0.5701164473583274, "grad_norm": 2.334580898284912, "learning_rate": 8.225253412571335e-06, "loss": 0.14739387035369872, "step": 27870 }, { "epoch": 0.5703210101309714, "grad_norm": 2.276038408279419, "learning_rate": 8.21873366034489e-06, "loss": 0.14402647018432618, "step": 27880 }, { "epoch": 0.5705255729036152, "grad_norm": 2.2511041164398193, "learning_rate": 8.212214690002084e-06, "loss": 0.1446582317352295, "step": 27890 }, { "epoch": 0.570730135676259, "grad_norm": 2.7825729846954346, "learning_rate": 8.205696504404411e-06, "loss": 0.1422065019607544, "step": 27900 }, { "epoch": 0.5709346984489028, "grad_norm": 1.9982891082763672, "learning_rate": 8.199179106413015e-06, "loss": 0.13860094547271729, "step": 27910 }, { "epoch": 0.5711392612215466, "grad_norm": 2.3969709873199463, "learning_rate": 8.192662498888694e-06, "loss": 0.1479475498199463, "step": 27920 }, { "epoch": 0.5713438239941904, "grad_norm": 2.1123874187469482, "learning_rate": 8.186146684691897e-06, "loss": 0.14169418811798096, "step": 27930 }, { "epoch": 0.5715483867668343, "grad_norm": 6.472072601318359, "learning_rate": 8.17963166668274e-06, "loss": 0.14870284795761107, "step": 27940 }, { "epoch": 0.5717529495394781, "grad_norm": 1.1653112173080444, "learning_rate": 8.173117447720972e-06, "loss": 0.13530662059783935, "step": 27950 }, { "epoch": 0.5719575123121219, "grad_norm": 1.8055839538574219, "learning_rate": 8.166604030665996e-06, "loss": 0.13460137844085693, "step": 27960 }, { "epoch": 0.5721620750847657, "grad_norm": 1.7301676273345947, "learning_rate": 8.160091418376863e-06, "loss": 0.15464556217193604, "step": 27970 }, { "epoch": 0.5723666378574095, "grad_norm": 1.7591346502304077, "learning_rate": 8.153579613712283e-06, "loss": 0.15223138332366942, "step": 27980 }, { "epoch": 0.5725712006300533, "grad_norm": 1.5369523763656616, "learning_rate": 8.14706861953059e-06, "loss": 0.1387478828430176, "step": 27990 }, { "epoch": 0.5727757634026972, "grad_norm": 3.0635571479797363, "learning_rate": 8.140558438689778e-06, "loss": 0.14459710121154784, "step": 28000 }, { "epoch": 0.572980326175341, "grad_norm": 2.5502870082855225, "learning_rate": 8.134049074047474e-06, "loss": 0.13874266147613526, "step": 28010 }, { "epoch": 0.5731848889479848, "grad_norm": 1.2121398448944092, "learning_rate": 8.127540528460961e-06, "loss": 0.13920135498046876, "step": 28020 }, { "epoch": 0.5733894517206286, "grad_norm": 1.655808687210083, "learning_rate": 8.121032804787148e-06, "loss": 0.1470586895942688, "step": 28030 }, { "epoch": 0.5735940144932724, "grad_norm": 0.9034110307693481, "learning_rate": 8.114525905882589e-06, "loss": 0.12666553258895874, "step": 28040 }, { "epoch": 0.5737985772659162, "grad_norm": 0.9482545852661133, "learning_rate": 8.108019834603472e-06, "loss": 0.13681132793426515, "step": 28050 }, { "epoch": 0.5740031400385601, "grad_norm": 0.9968993663787842, "learning_rate": 8.101514593805631e-06, "loss": 0.12596960067749025, "step": 28060 }, { "epoch": 0.5742077028112039, "grad_norm": 1.1802520751953125, "learning_rate": 8.095010186344528e-06, "loss": 0.1493183970451355, "step": 28070 }, { "epoch": 0.5744122655838477, "grad_norm": 1.5086534023284912, "learning_rate": 8.088506615075253e-06, "loss": 0.1367443323135376, "step": 28080 }, { "epoch": 0.5746168283564915, "grad_norm": 3.4541962146759033, "learning_rate": 8.082003882852552e-06, "loss": 0.15686845779418945, "step": 28090 }, { "epoch": 0.5748213911291353, "grad_norm": 3.65504789352417, "learning_rate": 8.075501992530778e-06, "loss": 0.13144453763961791, "step": 28100 }, { "epoch": 0.5750259539017791, "grad_norm": 1.5494920015335083, "learning_rate": 8.069000946963926e-06, "loss": 0.1424545168876648, "step": 28110 }, { "epoch": 0.575230516674423, "grad_norm": 3.1846327781677246, "learning_rate": 8.062500749005614e-06, "loss": 0.1577472686767578, "step": 28120 }, { "epoch": 0.5754350794470668, "grad_norm": 3.0065133571624756, "learning_rate": 8.056001401509099e-06, "loss": 0.1517184853553772, "step": 28130 }, { "epoch": 0.5756396422197106, "grad_norm": 1.312317967414856, "learning_rate": 8.049502907327255e-06, "loss": 0.15029211044311525, "step": 28140 }, { "epoch": 0.5758442049923544, "grad_norm": 1.4042867422103882, "learning_rate": 8.043005269312583e-06, "loss": 0.13657238483428955, "step": 28150 }, { "epoch": 0.5760487677649982, "grad_norm": 1.8824357986450195, "learning_rate": 8.036508490317209e-06, "loss": 0.15829505920410156, "step": 28160 }, { "epoch": 0.5762533305376422, "grad_norm": 2.57979416847229, "learning_rate": 8.030012573192885e-06, "loss": 0.13840174674987793, "step": 28170 }, { "epoch": 0.576457893310286, "grad_norm": 0.9533601403236389, "learning_rate": 8.023517520790983e-06, "loss": 0.14252955913543702, "step": 28180 }, { "epoch": 0.5766624560829298, "grad_norm": 1.9143904447555542, "learning_rate": 8.017023335962489e-06, "loss": 0.14497584104537964, "step": 28190 }, { "epoch": 0.5768670188555736, "grad_norm": 1.410996913909912, "learning_rate": 8.010530021558018e-06, "loss": 0.14176018238067628, "step": 28200 }, { "epoch": 0.5770715816282174, "grad_norm": 1.8560534715652466, "learning_rate": 8.004037580427799e-06, "loss": 0.14659957885742186, "step": 28210 }, { "epoch": 0.5772761444008612, "grad_norm": 3.153102159500122, "learning_rate": 7.997546015421675e-06, "loss": 0.13960549831390381, "step": 28220 }, { "epoch": 0.5774807071735051, "grad_norm": 3.313138723373413, "learning_rate": 7.99105532938911e-06, "loss": 0.13546860218048096, "step": 28230 }, { "epoch": 0.5776852699461489, "grad_norm": 1.698984980583191, "learning_rate": 7.984565525179173e-06, "loss": 0.14541139602661132, "step": 28240 }, { "epoch": 0.5778898327187927, "grad_norm": 4.195036888122559, "learning_rate": 7.978076605640556e-06, "loss": 0.15436990261077882, "step": 28250 }, { "epoch": 0.5780943954914365, "grad_norm": 1.1701676845550537, "learning_rate": 7.97158857362156e-06, "loss": 0.14463977813720702, "step": 28260 }, { "epoch": 0.5782989582640803, "grad_norm": 1.9131593704223633, "learning_rate": 7.96510143197009e-06, "loss": 0.13883355855941773, "step": 28270 }, { "epoch": 0.5785035210367241, "grad_norm": 2.7632811069488525, "learning_rate": 7.95861518353367e-06, "loss": 0.13424917459487914, "step": 28280 }, { "epoch": 0.578708083809368, "grad_norm": 1.4650226831436157, "learning_rate": 7.952129831159427e-06, "loss": 0.13170931339263917, "step": 28290 }, { "epoch": 0.5789126465820118, "grad_norm": 1.4002851247787476, "learning_rate": 7.94564537769409e-06, "loss": 0.13591716289520264, "step": 28300 }, { "epoch": 0.5791172093546556, "grad_norm": 1.8241708278656006, "learning_rate": 7.939161825984e-06, "loss": 0.14896926879882813, "step": 28310 }, { "epoch": 0.5793217721272994, "grad_norm": 2.7170729637145996, "learning_rate": 7.932679178875105e-06, "loss": 0.13750452995300294, "step": 28320 }, { "epoch": 0.5795263348999432, "grad_norm": 1.045839548110962, "learning_rate": 7.926197439212945e-06, "loss": 0.12511518001556396, "step": 28330 }, { "epoch": 0.579730897672587, "grad_norm": 2.015458106994629, "learning_rate": 7.919716609842672e-06, "loss": 0.1422576427459717, "step": 28340 }, { "epoch": 0.5799354604452309, "grad_norm": 2.209105968475342, "learning_rate": 7.913236693609028e-06, "loss": 0.14847447872161865, "step": 28350 }, { "epoch": 0.5801400232178747, "grad_norm": 2.4103565216064453, "learning_rate": 7.906757693356369e-06, "loss": 0.15247184038162231, "step": 28360 }, { "epoch": 0.5803445859905185, "grad_norm": 1.3170263767242432, "learning_rate": 7.900279611928636e-06, "loss": 0.13426473140716552, "step": 28370 }, { "epoch": 0.5805491487631623, "grad_norm": 1.4175275564193726, "learning_rate": 7.893802452169372e-06, "loss": 0.13579927682876586, "step": 28380 }, { "epoch": 0.5807537115358061, "grad_norm": 1.4078543186187744, "learning_rate": 7.887326216921713e-06, "loss": 0.15313210487365722, "step": 28390 }, { "epoch": 0.5809582743084499, "grad_norm": 1.859035611152649, "learning_rate": 7.880850909028394e-06, "loss": 0.14550631046295165, "step": 28400 }, { "epoch": 0.5811628370810938, "grad_norm": 2.106496810913086, "learning_rate": 7.874376531331738e-06, "loss": 0.14496079683303834, "step": 28410 }, { "epoch": 0.5813673998537376, "grad_norm": 2.0130198001861572, "learning_rate": 7.867903086673659e-06, "loss": 0.14262948036193848, "step": 28420 }, { "epoch": 0.5815719626263814, "grad_norm": 2.0813939571380615, "learning_rate": 7.861430577895666e-06, "loss": 0.14098787307739258, "step": 28430 }, { "epoch": 0.5817765253990252, "grad_norm": 1.923445463180542, "learning_rate": 7.854959007838856e-06, "loss": 0.13894453048706054, "step": 28440 }, { "epoch": 0.581981088171669, "grad_norm": 1.3964192867279053, "learning_rate": 7.848488379343912e-06, "loss": 0.13599939346313478, "step": 28450 }, { "epoch": 0.5821856509443128, "grad_norm": 2.6838371753692627, "learning_rate": 7.842018695251104e-06, "loss": 0.1332763910293579, "step": 28460 }, { "epoch": 0.5823902137169568, "grad_norm": 3.3925576210021973, "learning_rate": 7.835549958400289e-06, "loss": 0.14465686082839965, "step": 28470 }, { "epoch": 0.5825947764896006, "grad_norm": 2.4203591346740723, "learning_rate": 7.829082171630908e-06, "loss": 0.14071030616760255, "step": 28480 }, { "epoch": 0.5827993392622444, "grad_norm": 1.9108465909957886, "learning_rate": 7.822615337781984e-06, "loss": 0.13888964653015137, "step": 28490 }, { "epoch": 0.5830039020348882, "grad_norm": 1.338627815246582, "learning_rate": 7.816149459692118e-06, "loss": 0.13399899005889893, "step": 28500 }, { "epoch": 0.583208464807532, "grad_norm": 1.0532715320587158, "learning_rate": 7.809684540199503e-06, "loss": 0.13802053928375244, "step": 28510 }, { "epoch": 0.5834130275801758, "grad_norm": 1.9644051790237427, "learning_rate": 7.803220582141897e-06, "loss": 0.1507067561149597, "step": 28520 }, { "epoch": 0.5836175903528197, "grad_norm": 1.5035717487335205, "learning_rate": 7.796757588356648e-06, "loss": 0.12874436378479004, "step": 28530 }, { "epoch": 0.5838221531254635, "grad_norm": 2.259497880935669, "learning_rate": 7.79029556168067e-06, "loss": 0.14933228492736816, "step": 28540 }, { "epoch": 0.5840267158981073, "grad_norm": 2.5033254623413086, "learning_rate": 7.783834504950463e-06, "loss": 0.13129184246063233, "step": 28550 }, { "epoch": 0.5842312786707511, "grad_norm": 4.747262477874756, "learning_rate": 7.777374421002094e-06, "loss": 0.14504966735839844, "step": 28560 }, { "epoch": 0.5844358414433949, "grad_norm": 1.628740668296814, "learning_rate": 7.770915312671203e-06, "loss": 0.13909471035003662, "step": 28570 }, { "epoch": 0.5846404042160387, "grad_norm": 2.815523386001587, "learning_rate": 7.764457182793006e-06, "loss": 0.13559603691101074, "step": 28580 }, { "epoch": 0.5848449669886826, "grad_norm": 1.8603935241699219, "learning_rate": 7.758000034202288e-06, "loss": 0.13664214611053466, "step": 28590 }, { "epoch": 0.5850495297613264, "grad_norm": 2.533705949783325, "learning_rate": 7.751543869733404e-06, "loss": 0.13485044240951538, "step": 28600 }, { "epoch": 0.5852540925339702, "grad_norm": 1.4390820264816284, "learning_rate": 7.745088692220273e-06, "loss": 0.14815077781677247, "step": 28610 }, { "epoch": 0.585458655306614, "grad_norm": 2.027740001678467, "learning_rate": 7.738634504496383e-06, "loss": 0.14844988584518432, "step": 28620 }, { "epoch": 0.5856632180792578, "grad_norm": 1.278440237045288, "learning_rate": 7.732181309394791e-06, "loss": 0.13520663976669312, "step": 28630 }, { "epoch": 0.5858677808519017, "grad_norm": 2.1790366172790527, "learning_rate": 7.725729109748115e-06, "loss": 0.14459360837936402, "step": 28640 }, { "epoch": 0.5860723436245455, "grad_norm": 1.6060452461242676, "learning_rate": 7.719277908388532e-06, "loss": 0.14319493770599365, "step": 28650 }, { "epoch": 0.5862769063971893, "grad_norm": 1.485805869102478, "learning_rate": 7.71282770814779e-06, "loss": 0.14266408681869508, "step": 28660 }, { "epoch": 0.5864814691698331, "grad_norm": 1.595706582069397, "learning_rate": 7.706378511857192e-06, "loss": 0.14122402667999268, "step": 28670 }, { "epoch": 0.5866860319424769, "grad_norm": 1.7851969003677368, "learning_rate": 7.699930322347598e-06, "loss": 0.14801112413406373, "step": 28680 }, { "epoch": 0.5868905947151207, "grad_norm": 0.9494929909706116, "learning_rate": 7.693483142449432e-06, "loss": 0.1342146873474121, "step": 28690 }, { "epoch": 0.5870951574877646, "grad_norm": 1.5433175563812256, "learning_rate": 7.687036974992671e-06, "loss": 0.1533634066581726, "step": 28700 }, { "epoch": 0.5872997202604084, "grad_norm": 2.9558918476104736, "learning_rate": 7.68059182280685e-06, "loss": 0.13103599548339845, "step": 28710 }, { "epoch": 0.5875042830330522, "grad_norm": 2.740300178527832, "learning_rate": 7.674147688721055e-06, "loss": 0.14728775024414062, "step": 28720 }, { "epoch": 0.587708845805696, "grad_norm": 1.2289938926696777, "learning_rate": 7.667704575563926e-06, "loss": 0.14961885213851928, "step": 28730 }, { "epoch": 0.5879134085783398, "grad_norm": 2.343311309814453, "learning_rate": 7.661262486163659e-06, "loss": 0.13180029392242432, "step": 28740 }, { "epoch": 0.5881179713509836, "grad_norm": 0.9331393837928772, "learning_rate": 7.654821423347996e-06, "loss": 0.13385589122772218, "step": 28750 }, { "epoch": 0.5883225341236276, "grad_norm": 1.4349231719970703, "learning_rate": 7.648381389944232e-06, "loss": 0.14243202209472655, "step": 28760 }, { "epoch": 0.5885270968962714, "grad_norm": 1.8669425249099731, "learning_rate": 7.641942388779202e-06, "loss": 0.13904926776885987, "step": 28770 }, { "epoch": 0.5887316596689152, "grad_norm": 1.448439121246338, "learning_rate": 7.6355044226793e-06, "loss": 0.14005712270736695, "step": 28780 }, { "epoch": 0.588936222441559, "grad_norm": 2.506898880004883, "learning_rate": 7.629067494470458e-06, "loss": 0.1400708317756653, "step": 28790 }, { "epoch": 0.5891407852142028, "grad_norm": 2.0326569080352783, "learning_rate": 7.6226316069781504e-06, "loss": 0.14454212188720703, "step": 28800 }, { "epoch": 0.5893453479868466, "grad_norm": 1.2435529232025146, "learning_rate": 7.6161967630274035e-06, "loss": 0.1382544755935669, "step": 28810 }, { "epoch": 0.5895499107594905, "grad_norm": 2.4997024536132812, "learning_rate": 7.6097629654427795e-06, "loss": 0.14267294406890868, "step": 28820 }, { "epoch": 0.5897544735321343, "grad_norm": 3.641263961791992, "learning_rate": 7.603330217048381e-06, "loss": 0.14101805686950683, "step": 28830 }, { "epoch": 0.5899590363047781, "grad_norm": 3.1178576946258545, "learning_rate": 7.596898520667848e-06, "loss": 0.14776194095611572, "step": 28840 }, { "epoch": 0.5901635990774219, "grad_norm": 1.1636226177215576, "learning_rate": 7.590467879124368e-06, "loss": 0.1336905002593994, "step": 28850 }, { "epoch": 0.5903681618500657, "grad_norm": 1.6588433980941772, "learning_rate": 7.5840382952406576e-06, "loss": 0.14595384597778321, "step": 28860 }, { "epoch": 0.5905727246227095, "grad_norm": 1.658588171005249, "learning_rate": 7.577609771838972e-06, "loss": 0.13341604471206664, "step": 28870 }, { "epoch": 0.5907772873953534, "grad_norm": 1.6076215505599976, "learning_rate": 7.571182311741091e-06, "loss": 0.1361333131790161, "step": 28880 }, { "epoch": 0.5909818501679972, "grad_norm": 2.6706979274749756, "learning_rate": 7.564755917768348e-06, "loss": 0.13847770690917968, "step": 28890 }, { "epoch": 0.591186412940641, "grad_norm": 2.3564634323120117, "learning_rate": 7.5583305927415936e-06, "loss": 0.14393410682678223, "step": 28900 }, { "epoch": 0.5913909757132848, "grad_norm": 1.7174146175384521, "learning_rate": 7.5519063394812145e-06, "loss": 0.1380467653274536, "step": 28910 }, { "epoch": 0.5915955384859286, "grad_norm": 3.158790349960327, "learning_rate": 7.545483160807115e-06, "loss": 0.14915146827697753, "step": 28920 }, { "epoch": 0.5918001012585724, "grad_norm": 2.269022226333618, "learning_rate": 7.539061059538748e-06, "loss": 0.13765428066253663, "step": 28930 }, { "epoch": 0.5920046640312163, "grad_norm": 1.611506462097168, "learning_rate": 7.5326400384950826e-06, "loss": 0.1360754370689392, "step": 28940 }, { "epoch": 0.5922092268038601, "grad_norm": 1.1981569528579712, "learning_rate": 7.5262201004946136e-06, "loss": 0.13251791000366211, "step": 28950 }, { "epoch": 0.5924137895765039, "grad_norm": 2.125182628631592, "learning_rate": 7.519801248355353e-06, "loss": 0.14053404331207275, "step": 28960 }, { "epoch": 0.5926183523491477, "grad_norm": 3.9386115074157715, "learning_rate": 7.513383484894855e-06, "loss": 0.1542930006980896, "step": 28970 }, { "epoch": 0.5928229151217915, "grad_norm": 1.4686803817749023, "learning_rate": 7.506966812930184e-06, "loss": 0.13682348728179933, "step": 28980 }, { "epoch": 0.5930274778944353, "grad_norm": 1.8383498191833496, "learning_rate": 7.5005512352779175e-06, "loss": 0.13696391582489015, "step": 28990 }, { "epoch": 0.5932320406670792, "grad_norm": 1.7708237171173096, "learning_rate": 7.494136754754172e-06, "loss": 0.1402537703514099, "step": 29000 }, { "epoch": 0.593436603439723, "grad_norm": 1.5733743906021118, "learning_rate": 7.487723374174568e-06, "loss": 0.1366109609603882, "step": 29010 }, { "epoch": 0.5936411662123668, "grad_norm": 1.1882712841033936, "learning_rate": 7.48131109635425e-06, "loss": 0.1552217960357666, "step": 29020 }, { "epoch": 0.5938457289850106, "grad_norm": 1.403876543045044, "learning_rate": 7.474899924107868e-06, "loss": 0.137265408039093, "step": 29030 }, { "epoch": 0.5940502917576544, "grad_norm": 1.344299077987671, "learning_rate": 7.468489860249602e-06, "loss": 0.14494402408599855, "step": 29040 }, { "epoch": 0.5942548545302982, "grad_norm": 2.643433094024658, "learning_rate": 7.462080907593139e-06, "loss": 0.14775946140289306, "step": 29050 }, { "epoch": 0.5944594173029422, "grad_norm": 1.4804375171661377, "learning_rate": 7.455673068951672e-06, "loss": 0.13373143672943116, "step": 29060 }, { "epoch": 0.594663980075586, "grad_norm": 2.6887729167938232, "learning_rate": 7.449266347137907e-06, "loss": 0.1338234066963196, "step": 29070 }, { "epoch": 0.5948685428482298, "grad_norm": 1.4009284973144531, "learning_rate": 7.442860744964074e-06, "loss": 0.14242856502532958, "step": 29080 }, { "epoch": 0.5950731056208736, "grad_norm": 3.114438772201538, "learning_rate": 7.436456265241894e-06, "loss": 0.15131609439849852, "step": 29090 }, { "epoch": 0.5952776683935174, "grad_norm": 2.378265619277954, "learning_rate": 7.4300529107826016e-06, "loss": 0.13407516479492188, "step": 29100 }, { "epoch": 0.5954822311661613, "grad_norm": 3.475909471511841, "learning_rate": 7.423650684396933e-06, "loss": 0.14563405513763428, "step": 29110 }, { "epoch": 0.5956867939388051, "grad_norm": 3.9038233757019043, "learning_rate": 7.417249588895146e-06, "loss": 0.13430302143096923, "step": 29120 }, { "epoch": 0.5958913567114489, "grad_norm": 2.5543744564056396, "learning_rate": 7.4108496270869835e-06, "loss": 0.14090741872787477, "step": 29130 }, { "epoch": 0.5960959194840927, "grad_norm": 2.1200127601623535, "learning_rate": 7.404450801781695e-06, "loss": 0.13792858123779297, "step": 29140 }, { "epoch": 0.5963004822567365, "grad_norm": 2.500429153442383, "learning_rate": 7.398053115788031e-06, "loss": 0.12962477207183837, "step": 29150 }, { "epoch": 0.5965050450293803, "grad_norm": 2.264514207839966, "learning_rate": 7.391656571914254e-06, "loss": 0.13130404949188232, "step": 29160 }, { "epoch": 0.5967096078020242, "grad_norm": 1.2341018915176392, "learning_rate": 7.385261172968109e-06, "loss": 0.14670510292053224, "step": 29170 }, { "epoch": 0.596914170574668, "grad_norm": 2.713508129119873, "learning_rate": 7.3788669217568396e-06, "loss": 0.1429043412208557, "step": 29180 }, { "epoch": 0.5971187333473118, "grad_norm": 1.5532511472702026, "learning_rate": 7.372473821087203e-06, "loss": 0.14372150897979735, "step": 29190 }, { "epoch": 0.5973232961199556, "grad_norm": 3.9072656631469727, "learning_rate": 7.3660818737654335e-06, "loss": 0.13980569839477539, "step": 29200 }, { "epoch": 0.5975278588925994, "grad_norm": 2.0896291732788086, "learning_rate": 7.359691082597264e-06, "loss": 0.14398422241210937, "step": 29210 }, { "epoch": 0.5977324216652432, "grad_norm": 1.3015165328979492, "learning_rate": 7.353301450387915e-06, "loss": 0.14061187505722045, "step": 29220 }, { "epoch": 0.5979369844378871, "grad_norm": 3.2431252002716064, "learning_rate": 7.3469129799421154e-06, "loss": 0.13112026453018188, "step": 29230 }, { "epoch": 0.5981415472105309, "grad_norm": 1.7675554752349854, "learning_rate": 7.340525674064067e-06, "loss": 0.15073401927948, "step": 29240 }, { "epoch": 0.5983461099831747, "grad_norm": 0.68858802318573, "learning_rate": 7.334139535557468e-06, "loss": 0.1262032389640808, "step": 29250 }, { "epoch": 0.5985506727558185, "grad_norm": 1.7017072439193726, "learning_rate": 7.327754567225493e-06, "loss": 0.1370156526565552, "step": 29260 }, { "epoch": 0.5987552355284623, "grad_norm": 1.8361809253692627, "learning_rate": 7.3213707718708285e-06, "loss": 0.134023118019104, "step": 29270 }, { "epoch": 0.5989597983011061, "grad_norm": 2.5524487495422363, "learning_rate": 7.314988152295619e-06, "loss": 0.14465868473052979, "step": 29280 }, { "epoch": 0.59916436107375, "grad_norm": 1.0057487487792969, "learning_rate": 7.3086067113015066e-06, "loss": 0.14428560733795165, "step": 29290 }, { "epoch": 0.5993689238463938, "grad_norm": 1.42833411693573, "learning_rate": 7.302226451689609e-06, "loss": 0.14654967784881592, "step": 29300 }, { "epoch": 0.5995734866190376, "grad_norm": 1.590480923652649, "learning_rate": 7.295847376260541e-06, "loss": 0.13989081382751464, "step": 29310 }, { "epoch": 0.5997780493916814, "grad_norm": 1.4842617511749268, "learning_rate": 7.289469487814376e-06, "loss": 0.1165959358215332, "step": 29320 }, { "epoch": 0.5999826121643252, "grad_norm": 2.7549071311950684, "learning_rate": 7.2830927891506785e-06, "loss": 0.15563502311706542, "step": 29330 }, { "epoch": 0.600187174936969, "grad_norm": 4.895695209503174, "learning_rate": 7.276717283068489e-06, "loss": 0.14173519611358643, "step": 29340 }, { "epoch": 0.600391737709613, "grad_norm": 3.3709444999694824, "learning_rate": 7.270342972366327e-06, "loss": 0.15199582576751708, "step": 29350 }, { "epoch": 0.6005963004822568, "grad_norm": 2.924534320831299, "learning_rate": 7.263969859842184e-06, "loss": 0.1282575845718384, "step": 29360 }, { "epoch": 0.6008008632549006, "grad_norm": 1.7926656007766724, "learning_rate": 7.25759794829352e-06, "loss": 0.13908770084381103, "step": 29370 }, { "epoch": 0.6010054260275444, "grad_norm": 2.2846901416778564, "learning_rate": 7.2512272405172846e-06, "loss": 0.12875360250473022, "step": 29380 }, { "epoch": 0.6012099888001882, "grad_norm": 2.1360297203063965, "learning_rate": 7.244857739309883e-06, "loss": 0.1465645432472229, "step": 29390 }, { "epoch": 0.601414551572832, "grad_norm": 1.8355995416641235, "learning_rate": 7.238489447467195e-06, "loss": 0.1378185272216797, "step": 29400 }, { "epoch": 0.6016191143454759, "grad_norm": 1.6747092008590698, "learning_rate": 7.232122367784571e-06, "loss": 0.1409808874130249, "step": 29410 }, { "epoch": 0.6018236771181197, "grad_norm": 2.49489688873291, "learning_rate": 7.2257565030568315e-06, "loss": 0.15105001926422118, "step": 29420 }, { "epoch": 0.6020282398907635, "grad_norm": 3.1110570430755615, "learning_rate": 7.219391856078259e-06, "loss": 0.1458060026168823, "step": 29430 }, { "epoch": 0.6022328026634073, "grad_norm": 1.3619942665100098, "learning_rate": 7.2130284296426065e-06, "loss": 0.13487677574157714, "step": 29440 }, { "epoch": 0.6024373654360511, "grad_norm": 5.798102378845215, "learning_rate": 7.2066662265430846e-06, "loss": 0.14859952926635742, "step": 29450 }, { "epoch": 0.6026419282086949, "grad_norm": 1.4683606624603271, "learning_rate": 7.200305249572374e-06, "loss": 0.14329349994659424, "step": 29460 }, { "epoch": 0.6028464909813388, "grad_norm": 1.6503267288208008, "learning_rate": 7.1939455015226156e-06, "loss": 0.13940865993499757, "step": 29470 }, { "epoch": 0.6030510537539826, "grad_norm": 0.8636177182197571, "learning_rate": 7.1875869851854055e-06, "loss": 0.12787452936172486, "step": 29480 }, { "epoch": 0.6032556165266264, "grad_norm": 1.3839367628097534, "learning_rate": 7.181229703351802e-06, "loss": 0.1363150119781494, "step": 29490 }, { "epoch": 0.6034601792992702, "grad_norm": 1.5483132600784302, "learning_rate": 7.174873658812327e-06, "loss": 0.15465247631072998, "step": 29500 }, { "epoch": 0.603664742071914, "grad_norm": 0.998254656791687, "learning_rate": 7.1685188543569524e-06, "loss": 0.13565939664840698, "step": 29510 }, { "epoch": 0.6038693048445578, "grad_norm": 1.3169981241226196, "learning_rate": 7.162165292775109e-06, "loss": 0.14127997159957886, "step": 29520 }, { "epoch": 0.6040738676172017, "grad_norm": 1.3985885381698608, "learning_rate": 7.155812976855678e-06, "loss": 0.13796231746673585, "step": 29530 }, { "epoch": 0.6042784303898455, "grad_norm": 1.5802260637283325, "learning_rate": 7.149461909387001e-06, "loss": 0.1428377866744995, "step": 29540 }, { "epoch": 0.6044829931624893, "grad_norm": 1.2737421989440918, "learning_rate": 7.143112093156866e-06, "loss": 0.1434590458869934, "step": 29550 }, { "epoch": 0.6046875559351331, "grad_norm": 1.252449631690979, "learning_rate": 7.136763530952513e-06, "loss": 0.13061599731445311, "step": 29560 }, { "epoch": 0.6048921187077769, "grad_norm": 2.7644357681274414, "learning_rate": 7.130416225560633e-06, "loss": 0.1454838752746582, "step": 29570 }, { "epoch": 0.6050966814804208, "grad_norm": 1.3550857305526733, "learning_rate": 7.124070179767364e-06, "loss": 0.14021235704421997, "step": 29580 }, { "epoch": 0.6053012442530646, "grad_norm": 1.9846247434616089, "learning_rate": 7.1177253963582906e-06, "loss": 0.1299112319946289, "step": 29590 }, { "epoch": 0.6055058070257084, "grad_norm": 2.1818172931671143, "learning_rate": 7.111381878118442e-06, "loss": 0.13514584302902222, "step": 29600 }, { "epoch": 0.6057103697983522, "grad_norm": 2.8258216381073, "learning_rate": 7.1050396278322976e-06, "loss": 0.1399569630622864, "step": 29610 }, { "epoch": 0.605914932570996, "grad_norm": 1.476476788520813, "learning_rate": 7.098698648283777e-06, "loss": 0.12809059619903565, "step": 29620 }, { "epoch": 0.6061194953436398, "grad_norm": 1.7603150606155396, "learning_rate": 7.092358942256239e-06, "loss": 0.13185166120529174, "step": 29630 }, { "epoch": 0.6063240581162838, "grad_norm": 2.286550521850586, "learning_rate": 7.086020512532486e-06, "loss": 0.13123273849487305, "step": 29640 }, { "epoch": 0.6065286208889276, "grad_norm": 3.881781578063965, "learning_rate": 7.079683361894763e-06, "loss": 0.14767374992370605, "step": 29650 }, { "epoch": 0.6067331836615714, "grad_norm": 2.22881817817688, "learning_rate": 7.073347493124749e-06, "loss": 0.1365530252456665, "step": 29660 }, { "epoch": 0.6069377464342152, "grad_norm": 1.4885371923446655, "learning_rate": 7.067012909003563e-06, "loss": 0.13511521816253663, "step": 29670 }, { "epoch": 0.607142309206859, "grad_norm": 2.7741525173187256, "learning_rate": 7.060679612311757e-06, "loss": 0.1501840591430664, "step": 29680 }, { "epoch": 0.6073468719795028, "grad_norm": 3.953176975250244, "learning_rate": 7.054347605829323e-06, "loss": 0.1444241762161255, "step": 29690 }, { "epoch": 0.6075514347521467, "grad_norm": 3.9250614643096924, "learning_rate": 7.0480168923356825e-06, "loss": 0.1469182014465332, "step": 29700 }, { "epoch": 0.6077559975247905, "grad_norm": 0.9229586124420166, "learning_rate": 7.041687474609693e-06, "loss": 0.13517916202545166, "step": 29710 }, { "epoch": 0.6079605602974343, "grad_norm": 2.8434829711914062, "learning_rate": 7.035359355429635e-06, "loss": 0.14248961210250854, "step": 29720 }, { "epoch": 0.6081651230700781, "grad_norm": 1.7157000303268433, "learning_rate": 7.02903253757323e-06, "loss": 0.14512078762054442, "step": 29730 }, { "epoch": 0.6083696858427219, "grad_norm": 2.5767340660095215, "learning_rate": 7.022707023817625e-06, "loss": 0.13190302848815919, "step": 29740 }, { "epoch": 0.6085742486153657, "grad_norm": 2.264890670776367, "learning_rate": 7.0163828169393856e-06, "loss": 0.14332367181777955, "step": 29750 }, { "epoch": 0.6087788113880096, "grad_norm": 2.7151663303375244, "learning_rate": 7.010059919714518e-06, "loss": 0.13032834529876708, "step": 29760 }, { "epoch": 0.6089833741606534, "grad_norm": 1.0676329135894775, "learning_rate": 7.003738334918444e-06, "loss": 0.14262343645095826, "step": 29770 }, { "epoch": 0.6091879369332972, "grad_norm": 2.118722438812256, "learning_rate": 6.997418065326012e-06, "loss": 0.12407764196395873, "step": 29780 }, { "epoch": 0.609392499705941, "grad_norm": 1.2835743427276611, "learning_rate": 6.991099113711489e-06, "loss": 0.12485229969024658, "step": 29790 }, { "epoch": 0.6095970624785848, "grad_norm": 1.6159322261810303, "learning_rate": 6.984781482848574e-06, "loss": 0.14748225212097169, "step": 29800 }, { "epoch": 0.6098016252512286, "grad_norm": 2.0000386238098145, "learning_rate": 6.978465175510376e-06, "loss": 0.1529846429824829, "step": 29810 }, { "epoch": 0.6100061880238725, "grad_norm": 1.456248164176941, "learning_rate": 6.9721501944694285e-06, "loss": 0.1372464656829834, "step": 29820 }, { "epoch": 0.6102107507965163, "grad_norm": 1.9129807949066162, "learning_rate": 6.965836542497677e-06, "loss": 0.13442676067352294, "step": 29830 }, { "epoch": 0.6104153135691601, "grad_norm": 3.234279155731201, "learning_rate": 6.959524222366492e-06, "loss": 0.1348554253578186, "step": 29840 }, { "epoch": 0.6106198763418039, "grad_norm": 1.4059429168701172, "learning_rate": 6.953213236846654e-06, "loss": 0.12694908380508424, "step": 29850 }, { "epoch": 0.6108244391144477, "grad_norm": 1.6420843601226807, "learning_rate": 6.946903588708358e-06, "loss": 0.15261359214782716, "step": 29860 }, { "epoch": 0.6110290018870915, "grad_norm": 1.708228588104248, "learning_rate": 6.9405952807212095e-06, "loss": 0.13923170566558837, "step": 29870 }, { "epoch": 0.6112335646597354, "grad_norm": 4.186265468597412, "learning_rate": 6.934288315654236e-06, "loss": 0.16218115091323854, "step": 29880 }, { "epoch": 0.6114381274323792, "grad_norm": 2.3738300800323486, "learning_rate": 6.927982696275864e-06, "loss": 0.137612247467041, "step": 29890 }, { "epoch": 0.611642690205023, "grad_norm": 1.5665931701660156, "learning_rate": 6.921678425353931e-06, "loss": 0.14013874530792236, "step": 29900 }, { "epoch": 0.6118472529776668, "grad_norm": 3.1104540824890137, "learning_rate": 6.915375505655693e-06, "loss": 0.1250202775001526, "step": 29910 }, { "epoch": 0.6120518157503106, "grad_norm": 1.8742629289627075, "learning_rate": 6.9090739399477994e-06, "loss": 0.14098103046417237, "step": 29920 }, { "epoch": 0.6122563785229544, "grad_norm": 1.5166465044021606, "learning_rate": 6.902773730996312e-06, "loss": 0.14241912364959716, "step": 29930 }, { "epoch": 0.6124609412955984, "grad_norm": 3.6345276832580566, "learning_rate": 6.8964748815666945e-06, "loss": 0.14776005744934081, "step": 29940 }, { "epoch": 0.6126655040682422, "grad_norm": 1.4359360933303833, "learning_rate": 6.89017739442382e-06, "loss": 0.13837053775787353, "step": 29950 }, { "epoch": 0.612870066840886, "grad_norm": 1.6858576536178589, "learning_rate": 6.883881272331958e-06, "loss": 0.13532021045684814, "step": 29960 }, { "epoch": 0.6130746296135298, "grad_norm": 0.9002386331558228, "learning_rate": 6.877586518054777e-06, "loss": 0.12110804319381714, "step": 29970 }, { "epoch": 0.6132791923861736, "grad_norm": 2.2975473403930664, "learning_rate": 6.871293134355348e-06, "loss": 0.1468024730682373, "step": 29980 }, { "epoch": 0.6134837551588174, "grad_norm": 1.1775044202804565, "learning_rate": 6.865001123996145e-06, "loss": 0.12738144397735596, "step": 29990 }, { "epoch": 0.6136883179314613, "grad_norm": 2.29819655418396, "learning_rate": 6.858710489739029e-06, "loss": 0.14129685163497924, "step": 30000 }, { "epoch": 0.6138928807041051, "grad_norm": 2.3073365688323975, "learning_rate": 6.852421234345268e-06, "loss": 0.1389009714126587, "step": 30010 }, { "epoch": 0.6140974434767489, "grad_norm": 2.4316163063049316, "learning_rate": 6.846133360575512e-06, "loss": 0.1332458257675171, "step": 30020 }, { "epoch": 0.6143020062493927, "grad_norm": 3.7758970260620117, "learning_rate": 6.839846871189819e-06, "loss": 0.13388899564743043, "step": 30030 }, { "epoch": 0.6145065690220365, "grad_norm": 3.7012033462524414, "learning_rate": 6.8335617689476294e-06, "loss": 0.1294163465499878, "step": 30040 }, { "epoch": 0.6147111317946804, "grad_norm": 2.3212571144104004, "learning_rate": 6.82727805660778e-06, "loss": 0.1341669201850891, "step": 30050 }, { "epoch": 0.6149156945673242, "grad_norm": 1.0708112716674805, "learning_rate": 6.820995736928486e-06, "loss": 0.1365211009979248, "step": 30060 }, { "epoch": 0.615120257339968, "grad_norm": 1.7434744834899902, "learning_rate": 6.8147148126673714e-06, "loss": 0.13405282497406007, "step": 30070 }, { "epoch": 0.6153248201126118, "grad_norm": 3.6070663928985596, "learning_rate": 6.8084352865814316e-06, "loss": 0.14456486701965332, "step": 30080 }, { "epoch": 0.6155293828852556, "grad_norm": 1.0177334547042847, "learning_rate": 6.80215716142705e-06, "loss": 0.146124005317688, "step": 30090 }, { "epoch": 0.6157339456578994, "grad_norm": 1.319043517112732, "learning_rate": 6.7958804399600054e-06, "loss": 0.13612304925918578, "step": 30100 }, { "epoch": 0.6159385084305433, "grad_norm": 2.7478322982788086, "learning_rate": 6.7896051249354485e-06, "loss": 0.13942774534225463, "step": 30110 }, { "epoch": 0.6161430712031871, "grad_norm": 1.076524257659912, "learning_rate": 6.78333121910792e-06, "loss": 0.14507436752319336, "step": 30120 }, { "epoch": 0.6163476339758309, "grad_norm": 1.450455904006958, "learning_rate": 6.777058725231336e-06, "loss": 0.13659865856170655, "step": 30130 }, { "epoch": 0.6165521967484747, "grad_norm": 2.371373176574707, "learning_rate": 6.770787646058998e-06, "loss": 0.13997769355773926, "step": 30140 }, { "epoch": 0.6167567595211185, "grad_norm": 0.925098717212677, "learning_rate": 6.764517984343588e-06, "loss": 0.13537228107452393, "step": 30150 }, { "epoch": 0.6169613222937623, "grad_norm": 1.521271824836731, "learning_rate": 6.758249742837162e-06, "loss": 0.14187970161437988, "step": 30160 }, { "epoch": 0.6171658850664062, "grad_norm": 1.6309962272644043, "learning_rate": 6.751982924291145e-06, "loss": 0.14701462984085084, "step": 30170 }, { "epoch": 0.61737044783905, "grad_norm": 3.0372323989868164, "learning_rate": 6.745717531456355e-06, "loss": 0.12093936204910279, "step": 30180 }, { "epoch": 0.6175750106116938, "grad_norm": 4.822818279266357, "learning_rate": 6.739453567082973e-06, "loss": 0.13881187438964843, "step": 30190 }, { "epoch": 0.6177795733843376, "grad_norm": 1.3157246112823486, "learning_rate": 6.733191033920556e-06, "loss": 0.13458003997802734, "step": 30200 }, { "epoch": 0.6179841361569814, "grad_norm": 3.6327755451202393, "learning_rate": 6.7269299347180225e-06, "loss": 0.1463434100151062, "step": 30210 }, { "epoch": 0.6181886989296252, "grad_norm": 1.7211936712265015, "learning_rate": 6.7206702722236815e-06, "loss": 0.12483272552490235, "step": 30220 }, { "epoch": 0.6183932617022692, "grad_norm": 2.4277443885803223, "learning_rate": 6.714412049185199e-06, "loss": 0.15421760082244873, "step": 30230 }, { "epoch": 0.618597824474913, "grad_norm": 3.6509287357330322, "learning_rate": 6.7081552683496064e-06, "loss": 0.1302397608757019, "step": 30240 }, { "epoch": 0.6188023872475568, "grad_norm": 1.2922015190124512, "learning_rate": 6.701899932463304e-06, "loss": 0.13170077800750732, "step": 30250 }, { "epoch": 0.6190069500202006, "grad_norm": 2.6113152503967285, "learning_rate": 6.695646044272068e-06, "loss": 0.15288435220718383, "step": 30260 }, { "epoch": 0.6192115127928444, "grad_norm": 2.40045166015625, "learning_rate": 6.68939360652103e-06, "loss": 0.12694023847579955, "step": 30270 }, { "epoch": 0.6194160755654882, "grad_norm": 2.2326583862304688, "learning_rate": 6.683142621954677e-06, "loss": 0.1433804750442505, "step": 30280 }, { "epoch": 0.6196206383381321, "grad_norm": 1.211056113243103, "learning_rate": 6.67689309331688e-06, "loss": 0.13180575370788575, "step": 30290 }, { "epoch": 0.6198252011107759, "grad_norm": 1.8953564167022705, "learning_rate": 6.670645023350851e-06, "loss": 0.12974188327789307, "step": 30300 }, { "epoch": 0.6200297638834197, "grad_norm": 1.3848096132278442, "learning_rate": 6.6643984147991735e-06, "loss": 0.1274498701095581, "step": 30310 }, { "epoch": 0.6202343266560635, "grad_norm": 1.1240062713623047, "learning_rate": 6.658153270403776e-06, "loss": 0.13669147491455078, "step": 30320 }, { "epoch": 0.6204388894287073, "grad_norm": 1.2976996898651123, "learning_rate": 6.651909592905963e-06, "loss": 0.14307966232299804, "step": 30330 }, { "epoch": 0.6206434522013511, "grad_norm": 1.79576575756073, "learning_rate": 6.6456673850463856e-06, "loss": 0.1378461241722107, "step": 30340 }, { "epoch": 0.620848014973995, "grad_norm": 3.6707537174224854, "learning_rate": 6.639426649565042e-06, "loss": 0.14272806644439698, "step": 30350 }, { "epoch": 0.6210525777466388, "grad_norm": 2.4572863578796387, "learning_rate": 6.6331873892012925e-06, "loss": 0.13370590209960936, "step": 30360 }, { "epoch": 0.6212571405192826, "grad_norm": 1.9289289712905884, "learning_rate": 6.626949606693858e-06, "loss": 0.12952330112457275, "step": 30370 }, { "epoch": 0.6214617032919264, "grad_norm": 2.1006975173950195, "learning_rate": 6.620713304780796e-06, "loss": 0.1360421895980835, "step": 30380 }, { "epoch": 0.6216662660645702, "grad_norm": 1.1368988752365112, "learning_rate": 6.614478486199518e-06, "loss": 0.11644564867019654, "step": 30390 }, { "epoch": 0.621870828837214, "grad_norm": 1.8519951105117798, "learning_rate": 6.608245153686786e-06, "loss": 0.14534249305725097, "step": 30400 }, { "epoch": 0.6220753916098579, "grad_norm": 1.9223647117614746, "learning_rate": 6.602013309978719e-06, "loss": 0.12804746627807617, "step": 30410 }, { "epoch": 0.6222799543825017, "grad_norm": 2.0408530235290527, "learning_rate": 6.595782957810763e-06, "loss": 0.1430150032043457, "step": 30420 }, { "epoch": 0.6224845171551455, "grad_norm": 0.7382440567016602, "learning_rate": 6.5895540999177255e-06, "loss": 0.12849960327148438, "step": 30430 }, { "epoch": 0.6226890799277893, "grad_norm": 1.5994818210601807, "learning_rate": 6.5833267390337465e-06, "loss": 0.13273961544036866, "step": 30440 }, { "epoch": 0.6228936427004331, "grad_norm": 1.6523798704147339, "learning_rate": 6.577100877892326e-06, "loss": 0.14510514736175537, "step": 30450 }, { "epoch": 0.6230982054730769, "grad_norm": 2.0781826972961426, "learning_rate": 6.570876519226285e-06, "loss": 0.13150004148483277, "step": 30460 }, { "epoch": 0.6233027682457208, "grad_norm": 2.7389450073242188, "learning_rate": 6.564653665767793e-06, "loss": 0.13708747625350953, "step": 30470 }, { "epoch": 0.6235073310183646, "grad_norm": 2.857909679412842, "learning_rate": 6.558432320248368e-06, "loss": 0.1389464855194092, "step": 30480 }, { "epoch": 0.6237118937910084, "grad_norm": 3.1358418464660645, "learning_rate": 6.552212485398858e-06, "loss": 0.136018705368042, "step": 30490 }, { "epoch": 0.6239164565636522, "grad_norm": 1.683140516281128, "learning_rate": 6.5459941639494415e-06, "loss": 0.1467163920402527, "step": 30500 }, { "epoch": 0.624121019336296, "grad_norm": 1.8960450887680054, "learning_rate": 6.53977735862964e-06, "loss": 0.13656480312347413, "step": 30510 }, { "epoch": 0.62432558210894, "grad_norm": 4.093841552734375, "learning_rate": 6.533562072168318e-06, "loss": 0.1454852819442749, "step": 30520 }, { "epoch": 0.6245301448815838, "grad_norm": 1.3703030347824097, "learning_rate": 6.527348307293656e-06, "loss": 0.13672471046447754, "step": 30530 }, { "epoch": 0.6247347076542276, "grad_norm": 3.234839916229248, "learning_rate": 6.521136066733178e-06, "loss": 0.13070181608200074, "step": 30540 }, { "epoch": 0.6249392704268714, "grad_norm": 3.715797185897827, "learning_rate": 6.51492535321373e-06, "loss": 0.12691206932067872, "step": 30550 }, { "epoch": 0.6251438331995152, "grad_norm": 1.8274312019348145, "learning_rate": 6.508716169461504e-06, "loss": 0.14274346828460693, "step": 30560 }, { "epoch": 0.625348395972159, "grad_norm": 2.0140209197998047, "learning_rate": 6.502508518202002e-06, "loss": 0.1407869577407837, "step": 30570 }, { "epoch": 0.6255529587448029, "grad_norm": 1.4175843000411987, "learning_rate": 6.4963024021600655e-06, "loss": 0.14660441875457764, "step": 30580 }, { "epoch": 0.6257575215174467, "grad_norm": 1.3614912033081055, "learning_rate": 6.490097824059852e-06, "loss": 0.1256251811981201, "step": 30590 }, { "epoch": 0.6259620842900905, "grad_norm": 1.7429490089416504, "learning_rate": 6.483894786624857e-06, "loss": 0.12996516227722169, "step": 30600 }, { "epoch": 0.6261666470627343, "grad_norm": 2.8870694637298584, "learning_rate": 6.477693292577889e-06, "loss": 0.1259645104408264, "step": 30610 }, { "epoch": 0.6263712098353781, "grad_norm": 2.119896173477173, "learning_rate": 6.471493344641083e-06, "loss": 0.14768340587615966, "step": 30620 }, { "epoch": 0.6265757726080219, "grad_norm": 1.7679911851882935, "learning_rate": 6.4652949455358935e-06, "loss": 0.13028247356414796, "step": 30630 }, { "epoch": 0.6267803353806658, "grad_norm": 1.2777440547943115, "learning_rate": 6.459098097983099e-06, "loss": 0.14338381290435792, "step": 30640 }, { "epoch": 0.6269848981533096, "grad_norm": 2.1265487670898438, "learning_rate": 6.452902804702795e-06, "loss": 0.1274484157562256, "step": 30650 }, { "epoch": 0.6271894609259534, "grad_norm": 1.3831578493118286, "learning_rate": 6.446709068414389e-06, "loss": 0.12061508893966674, "step": 30660 }, { "epoch": 0.6273940236985972, "grad_norm": 2.517914295196533, "learning_rate": 6.440516891836619e-06, "loss": 0.13787670135498048, "step": 30670 }, { "epoch": 0.627598586471241, "grad_norm": 1.5852943658828735, "learning_rate": 6.434326277687524e-06, "loss": 0.13979381322860718, "step": 30680 }, { "epoch": 0.6278031492438848, "grad_norm": 1.3762459754943848, "learning_rate": 6.428137228684463e-06, "loss": 0.1324792504310608, "step": 30690 }, { "epoch": 0.6280077120165287, "grad_norm": 1.2066023349761963, "learning_rate": 6.421949747544107e-06, "loss": 0.142132568359375, "step": 30700 }, { "epoch": 0.6282122747891725, "grad_norm": 2.6758787631988525, "learning_rate": 6.415763836982442e-06, "loss": 0.14072216749191285, "step": 30710 }, { "epoch": 0.6284168375618163, "grad_norm": 1.7637284994125366, "learning_rate": 6.409579499714764e-06, "loss": 0.11544580459594726, "step": 30720 }, { "epoch": 0.6286214003344601, "grad_norm": 1.3847811222076416, "learning_rate": 6.403396738455674e-06, "loss": 0.13692654371261598, "step": 30730 }, { "epoch": 0.6288259631071039, "grad_norm": 1.3687011003494263, "learning_rate": 6.397215555919081e-06, "loss": 0.12751317024230957, "step": 30740 }, { "epoch": 0.6290305258797477, "grad_norm": 1.4142926931381226, "learning_rate": 6.3910359548182085e-06, "loss": 0.13040468692779542, "step": 30750 }, { "epoch": 0.6292350886523916, "grad_norm": 1.8281641006469727, "learning_rate": 6.38485793786558e-06, "loss": 0.11955113410949707, "step": 30760 }, { "epoch": 0.6294396514250354, "grad_norm": 2.961440324783325, "learning_rate": 6.378681507773027e-06, "loss": 0.12838213443756102, "step": 30770 }, { "epoch": 0.6296442141976792, "grad_norm": 3.129422187805176, "learning_rate": 6.3725066672516765e-06, "loss": 0.13278058767318726, "step": 30780 }, { "epoch": 0.629848776970323, "grad_norm": 3.357698440551758, "learning_rate": 6.36633341901197e-06, "loss": 0.13024178743362427, "step": 30790 }, { "epoch": 0.6300533397429668, "grad_norm": 1.2304209470748901, "learning_rate": 6.36016176576364e-06, "loss": 0.13173506259918213, "step": 30800 }, { "epoch": 0.6302579025156106, "grad_norm": 2.792025327682495, "learning_rate": 6.353991710215721e-06, "loss": 0.1218438982963562, "step": 30810 }, { "epoch": 0.6304624652882546, "grad_norm": 1.8324501514434814, "learning_rate": 6.34782325507655e-06, "loss": 0.13297669887542723, "step": 30820 }, { "epoch": 0.6306670280608984, "grad_norm": 1.7467906475067139, "learning_rate": 6.341656403053762e-06, "loss": 0.1313770055770874, "step": 30830 }, { "epoch": 0.6308715908335422, "grad_norm": 1.4067243337631226, "learning_rate": 6.335491156854283e-06, "loss": 0.12904590368270874, "step": 30840 }, { "epoch": 0.631076153606186, "grad_norm": 1.2668333053588867, "learning_rate": 6.329327519184333e-06, "loss": 0.1318972587585449, "step": 30850 }, { "epoch": 0.6312807163788298, "grad_norm": 2.849165678024292, "learning_rate": 6.323165492749436e-06, "loss": 0.1473197340965271, "step": 30860 }, { "epoch": 0.6314852791514736, "grad_norm": 1.234747290611267, "learning_rate": 6.317005080254399e-06, "loss": 0.12743443250656128, "step": 30870 }, { "epoch": 0.6316898419241175, "grad_norm": 1.119268774986267, "learning_rate": 6.3108462844033244e-06, "loss": 0.1271827459335327, "step": 30880 }, { "epoch": 0.6318944046967613, "grad_norm": 1.278789758682251, "learning_rate": 6.304689107899602e-06, "loss": 0.13804444074630737, "step": 30890 }, { "epoch": 0.6320989674694051, "grad_norm": 2.5473597049713135, "learning_rate": 6.298533553445919e-06, "loss": 0.12878067493438722, "step": 30900 }, { "epoch": 0.6323035302420489, "grad_norm": 2.5792832374572754, "learning_rate": 6.29237962374424e-06, "loss": 0.14299501180648805, "step": 30910 }, { "epoch": 0.6325080930146927, "grad_norm": 1.7068064212799072, "learning_rate": 6.286227321495825e-06, "loss": 0.1288081407546997, "step": 30920 }, { "epoch": 0.6327126557873365, "grad_norm": 1.5229934453964233, "learning_rate": 6.280076649401211e-06, "loss": 0.12930049896240234, "step": 30930 }, { "epoch": 0.6329172185599804, "grad_norm": 1.2730321884155273, "learning_rate": 6.273927610160228e-06, "loss": 0.13068290948867797, "step": 30940 }, { "epoch": 0.6331217813326242, "grad_norm": 2.009401798248291, "learning_rate": 6.267780206471987e-06, "loss": 0.1326945185661316, "step": 30950 }, { "epoch": 0.633326344105268, "grad_norm": 2.914762496948242, "learning_rate": 6.261634441034878e-06, "loss": 0.13526903390884398, "step": 30960 }, { "epoch": 0.6335309068779118, "grad_norm": 2.5992090702056885, "learning_rate": 6.25549031654657e-06, "loss": 0.1340259790420532, "step": 30970 }, { "epoch": 0.6337354696505556, "grad_norm": 1.9673912525177002, "learning_rate": 6.24934783570402e-06, "loss": 0.1364763855934143, "step": 30980 }, { "epoch": 0.6339400324231995, "grad_norm": 1.6883090734481812, "learning_rate": 6.24320700120346e-06, "loss": 0.14686866998672485, "step": 30990 }, { "epoch": 0.6341445951958433, "grad_norm": 1.6990578174591064, "learning_rate": 6.237067815740392e-06, "loss": 0.14219369888305664, "step": 31000 }, { "epoch": 0.6343491579684871, "grad_norm": 3.459831714630127, "learning_rate": 6.2309302820096065e-06, "loss": 0.13577210903167725, "step": 31010 }, { "epoch": 0.6345537207411309, "grad_norm": 1.8959897756576538, "learning_rate": 6.224794402705161e-06, "loss": 0.14193733930587768, "step": 31020 }, { "epoch": 0.6347582835137747, "grad_norm": 1.9897644519805908, "learning_rate": 6.218660180520388e-06, "loss": 0.13904550075531005, "step": 31030 }, { "epoch": 0.6349628462864185, "grad_norm": 2.0530691146850586, "learning_rate": 6.21252761814789e-06, "loss": 0.13155457973480225, "step": 31040 }, { "epoch": 0.6351674090590624, "grad_norm": 1.8225996494293213, "learning_rate": 6.20639671827955e-06, "loss": 0.14637061357498168, "step": 31050 }, { "epoch": 0.6353719718317062, "grad_norm": 1.5239019393920898, "learning_rate": 6.200267483606515e-06, "loss": 0.1309475302696228, "step": 31060 }, { "epoch": 0.63557653460435, "grad_norm": 1.9934290647506714, "learning_rate": 6.194139916819197e-06, "loss": 0.13008508682250977, "step": 31070 }, { "epoch": 0.6357810973769938, "grad_norm": 2.127655267715454, "learning_rate": 6.1880140206072815e-06, "loss": 0.1223823070526123, "step": 31080 }, { "epoch": 0.6359856601496376, "grad_norm": 1.3914884328842163, "learning_rate": 6.181889797659723e-06, "loss": 0.12531838417053223, "step": 31090 }, { "epoch": 0.6361902229222814, "grad_norm": 3.372171640396118, "learning_rate": 6.175767250664735e-06, "loss": 0.13408282995224, "step": 31100 }, { "epoch": 0.6363947856949254, "grad_norm": 2.2082557678222656, "learning_rate": 6.169646382309801e-06, "loss": 0.1271184802055359, "step": 31110 }, { "epoch": 0.6365993484675692, "grad_norm": 1.9729970693588257, "learning_rate": 6.16352719528166e-06, "loss": 0.12482857704162598, "step": 31120 }, { "epoch": 0.636803911240213, "grad_norm": 1.2222285270690918, "learning_rate": 6.1574096922663255e-06, "loss": 0.14020102024078368, "step": 31130 }, { "epoch": 0.6370084740128568, "grad_norm": 1.2497903108596802, "learning_rate": 6.151293875949058e-06, "loss": 0.13120505809783936, "step": 31140 }, { "epoch": 0.6372130367855006, "grad_norm": 1.4874796867370605, "learning_rate": 6.145179749014387e-06, "loss": 0.1454329252243042, "step": 31150 }, { "epoch": 0.6374175995581444, "grad_norm": 1.701647162437439, "learning_rate": 6.139067314146096e-06, "loss": 0.12874190807342528, "step": 31160 }, { "epoch": 0.6376221623307883, "grad_norm": 1.6196404695510864, "learning_rate": 6.132956574027229e-06, "loss": 0.133049476146698, "step": 31170 }, { "epoch": 0.6378267251034321, "grad_norm": 1.4434468746185303, "learning_rate": 6.126847531340085e-06, "loss": 0.11939560174942017, "step": 31180 }, { "epoch": 0.6380312878760759, "grad_norm": 1.2914727926254272, "learning_rate": 6.120740188766212e-06, "loss": 0.13306522369384766, "step": 31190 }, { "epoch": 0.6382358506487197, "grad_norm": 1.0882508754730225, "learning_rate": 6.114634548986424e-06, "loss": 0.14066942930221557, "step": 31200 }, { "epoch": 0.6384404134213635, "grad_norm": 1.774300217628479, "learning_rate": 6.108530614680777e-06, "loss": 0.13746435642242433, "step": 31210 }, { "epoch": 0.6386449761940073, "grad_norm": 4.287378787994385, "learning_rate": 6.102428388528582e-06, "loss": 0.14537569284439086, "step": 31220 }, { "epoch": 0.6388495389666512, "grad_norm": 2.824911117553711, "learning_rate": 6.096327873208397e-06, "loss": 0.1381535768508911, "step": 31230 }, { "epoch": 0.639054101739295, "grad_norm": 2.01005482673645, "learning_rate": 6.090229071398039e-06, "loss": 0.13217651844024658, "step": 31240 }, { "epoch": 0.6392586645119388, "grad_norm": 4.134884834289551, "learning_rate": 6.084131985774563e-06, "loss": 0.14592533111572265, "step": 31250 }, { "epoch": 0.6394632272845826, "grad_norm": 0.8188876509666443, "learning_rate": 6.078036619014272e-06, "loss": 0.13837380409240724, "step": 31260 }, { "epoch": 0.6396677900572264, "grad_norm": 2.4421961307525635, "learning_rate": 6.071942973792716e-06, "loss": 0.11689014434814453, "step": 31270 }, { "epoch": 0.6398723528298702, "grad_norm": 2.3404016494750977, "learning_rate": 6.065851052784693e-06, "loss": 0.1371586799621582, "step": 31280 }, { "epoch": 0.6400769156025141, "grad_norm": 1.5226045846939087, "learning_rate": 6.059760858664239e-06, "loss": 0.1348918318748474, "step": 31290 }, { "epoch": 0.6402814783751579, "grad_norm": 2.5327653884887695, "learning_rate": 6.053672394104633e-06, "loss": 0.15111217498779297, "step": 31300 }, { "epoch": 0.6404860411478017, "grad_norm": 1.3438048362731934, "learning_rate": 6.047585661778394e-06, "loss": 0.1403345823287964, "step": 31310 }, { "epoch": 0.6406906039204455, "grad_norm": 1.5226895809173584, "learning_rate": 6.041500664357286e-06, "loss": 0.12502126693725585, "step": 31320 }, { "epoch": 0.6408951666930893, "grad_norm": 1.807092547416687, "learning_rate": 6.035417404512306e-06, "loss": 0.13917362689971924, "step": 31330 }, { "epoch": 0.6410997294657331, "grad_norm": 1.8215538263320923, "learning_rate": 6.029335884913692e-06, "loss": 0.13359572887420654, "step": 31340 }, { "epoch": 0.641304292238377, "grad_norm": 1.7851842641830444, "learning_rate": 6.023256108230906e-06, "loss": 0.1415521264076233, "step": 31350 }, { "epoch": 0.6415088550110208, "grad_norm": 2.2285096645355225, "learning_rate": 6.017178077132668e-06, "loss": 0.12833337783813475, "step": 31360 }, { "epoch": 0.6417134177836646, "grad_norm": 1.9447193145751953, "learning_rate": 6.011101794286913e-06, "loss": 0.14280588626861573, "step": 31370 }, { "epoch": 0.6419179805563084, "grad_norm": 2.6568126678466797, "learning_rate": 6.005027262360814e-06, "loss": 0.1450817108154297, "step": 31380 }, { "epoch": 0.6421225433289522, "grad_norm": 1.773498296737671, "learning_rate": 5.998954484020777e-06, "loss": 0.1315446376800537, "step": 31390 }, { "epoch": 0.642327106101596, "grad_norm": 1.7414170503616333, "learning_rate": 5.9928834619324375e-06, "loss": 0.128196120262146, "step": 31400 }, { "epoch": 0.64253166887424, "grad_norm": 1.1657987833023071, "learning_rate": 5.98681419876066e-06, "loss": 0.13135778903961182, "step": 31410 }, { "epoch": 0.6427362316468838, "grad_norm": 1.1655184030532837, "learning_rate": 5.980746697169531e-06, "loss": 0.1186147689819336, "step": 31420 }, { "epoch": 0.6429407944195276, "grad_norm": 2.887585163116455, "learning_rate": 5.9746809598223765e-06, "loss": 0.13351560831069947, "step": 31430 }, { "epoch": 0.6431453571921714, "grad_norm": 1.4057390689849854, "learning_rate": 5.968616989381737e-06, "loss": 0.15391175746917723, "step": 31440 }, { "epoch": 0.6433499199648152, "grad_norm": 3.4474825859069824, "learning_rate": 5.962554788509387e-06, "loss": 0.13289237022399902, "step": 31450 }, { "epoch": 0.6435544827374591, "grad_norm": 1.8194791078567505, "learning_rate": 5.956494359866307e-06, "loss": 0.1380021572113037, "step": 31460 }, { "epoch": 0.6437590455101029, "grad_norm": 2.89391827583313, "learning_rate": 5.950435706112723e-06, "loss": 0.13811490535736085, "step": 31470 }, { "epoch": 0.6439636082827467, "grad_norm": 1.785996675491333, "learning_rate": 5.944378829908064e-06, "loss": 0.12907860279083253, "step": 31480 }, { "epoch": 0.6441681710553905, "grad_norm": 2.209742784500122, "learning_rate": 5.9383237339109915e-06, "loss": 0.1150315523147583, "step": 31490 }, { "epoch": 0.6443727338280343, "grad_norm": 1.2970058917999268, "learning_rate": 5.932270420779365e-06, "loss": 0.150899338722229, "step": 31500 }, { "epoch": 0.6445772966006781, "grad_norm": 1.8134080171585083, "learning_rate": 5.926218893170288e-06, "loss": 0.13288838863372804, "step": 31510 }, { "epoch": 0.644781859373322, "grad_norm": 1.5173914432525635, "learning_rate": 5.920169153740067e-06, "loss": 0.14717631340026854, "step": 31520 }, { "epoch": 0.6449864221459658, "grad_norm": 1.5694376230239868, "learning_rate": 5.914121205144219e-06, "loss": 0.1256725311279297, "step": 31530 }, { "epoch": 0.6451909849186096, "grad_norm": 2.1887166500091553, "learning_rate": 5.908075050037478e-06, "loss": 0.14584611654281615, "step": 31540 }, { "epoch": 0.6453955476912534, "grad_norm": 1.9366282224655151, "learning_rate": 5.902030691073801e-06, "loss": 0.12826282978057862, "step": 31550 }, { "epoch": 0.6456001104638972, "grad_norm": 1.900657296180725, "learning_rate": 5.895988130906346e-06, "loss": 0.12827836275100707, "step": 31560 }, { "epoch": 0.645804673236541, "grad_norm": 1.547014832496643, "learning_rate": 5.889947372187477e-06, "loss": 0.12686529159545898, "step": 31570 }, { "epoch": 0.6460092360091849, "grad_norm": 2.3388686180114746, "learning_rate": 5.883908417568782e-06, "loss": 0.13085763454437255, "step": 31580 }, { "epoch": 0.6462137987818287, "grad_norm": 1.1806443929672241, "learning_rate": 5.87787126970105e-06, "loss": 0.13291730880737304, "step": 31590 }, { "epoch": 0.6464183615544725, "grad_norm": 1.5335214138031006, "learning_rate": 5.871835931234268e-06, "loss": 0.14031665325164794, "step": 31600 }, { "epoch": 0.6466229243271163, "grad_norm": 3.1097657680511475, "learning_rate": 5.86580240481764e-06, "loss": 0.13297611474990845, "step": 31610 }, { "epoch": 0.6468274870997601, "grad_norm": 2.281114101409912, "learning_rate": 5.859770693099578e-06, "loss": 0.14602952003479003, "step": 31620 }, { "epoch": 0.6470320498724039, "grad_norm": 1.5451555252075195, "learning_rate": 5.853740798727686e-06, "loss": 0.13632433414459227, "step": 31630 }, { "epoch": 0.6472366126450478, "grad_norm": 1.305038332939148, "learning_rate": 5.847712724348774e-06, "loss": 0.1445746660232544, "step": 31640 }, { "epoch": 0.6474411754176916, "grad_norm": 2.8565022945404053, "learning_rate": 5.841686472608852e-06, "loss": 0.12294031381607055, "step": 31650 }, { "epoch": 0.6476457381903354, "grad_norm": 1.9282054901123047, "learning_rate": 5.835662046153141e-06, "loss": 0.12853329181671141, "step": 31660 }, { "epoch": 0.6478503009629792, "grad_norm": 1.6069916486740112, "learning_rate": 5.829639447626047e-06, "loss": 0.1335081934928894, "step": 31670 }, { "epoch": 0.648054863735623, "grad_norm": 3.2304141521453857, "learning_rate": 5.823618679671177e-06, "loss": 0.1437103867530823, "step": 31680 }, { "epoch": 0.6482594265082668, "grad_norm": 4.355749607086182, "learning_rate": 5.817599744931335e-06, "loss": 0.13412730693817138, "step": 31690 }, { "epoch": 0.6484639892809108, "grad_norm": 1.0030843019485474, "learning_rate": 5.811582646048528e-06, "loss": 0.12560662031173705, "step": 31700 }, { "epoch": 0.6486685520535546, "grad_norm": 1.0422309637069702, "learning_rate": 5.805567385663947e-06, "loss": 0.1398850440979004, "step": 31710 }, { "epoch": 0.6488731148261984, "grad_norm": 1.1719530820846558, "learning_rate": 5.7995539664179815e-06, "loss": 0.12875916957855224, "step": 31720 }, { "epoch": 0.6490776775988422, "grad_norm": 1.4751464128494263, "learning_rate": 5.793542390950208e-06, "loss": 0.1335953950881958, "step": 31730 }, { "epoch": 0.649282240371486, "grad_norm": 4.06113862991333, "learning_rate": 5.787532661899398e-06, "loss": 0.12804958820343018, "step": 31740 }, { "epoch": 0.6494868031441298, "grad_norm": 2.003573179244995, "learning_rate": 5.781524781903511e-06, "loss": 0.12783122062683105, "step": 31750 }, { "epoch": 0.6496913659167737, "grad_norm": 1.6411911249160767, "learning_rate": 5.7755187535996915e-06, "loss": 0.13449894189834594, "step": 31760 }, { "epoch": 0.6498959286894175, "grad_norm": 3.009418249130249, "learning_rate": 5.7695145796242845e-06, "loss": 0.1406106948852539, "step": 31770 }, { "epoch": 0.6501004914620613, "grad_norm": 1.6342520713806152, "learning_rate": 5.763512262612803e-06, "loss": 0.1376536726951599, "step": 31780 }, { "epoch": 0.6503050542347051, "grad_norm": 4.064123153686523, "learning_rate": 5.757511805199958e-06, "loss": 0.1372677803039551, "step": 31790 }, { "epoch": 0.6505096170073489, "grad_norm": 1.789623498916626, "learning_rate": 5.751513210019634e-06, "loss": 0.12445542812347413, "step": 31800 }, { "epoch": 0.6507141797799927, "grad_norm": 1.895176887512207, "learning_rate": 5.745516479704909e-06, "loss": 0.13657550811767577, "step": 31810 }, { "epoch": 0.6509187425526366, "grad_norm": 1.5958733558654785, "learning_rate": 5.739521616888034e-06, "loss": 0.130646276473999, "step": 31820 }, { "epoch": 0.6511233053252804, "grad_norm": 3.209839105606079, "learning_rate": 5.733528624200444e-06, "loss": 0.1308511257171631, "step": 31830 }, { "epoch": 0.6513278680979242, "grad_norm": 3.4294657707214355, "learning_rate": 5.727537504272747e-06, "loss": 0.12838494777679443, "step": 31840 }, { "epoch": 0.651532430870568, "grad_norm": 1.280937910079956, "learning_rate": 5.7215482597347435e-06, "loss": 0.1251215934753418, "step": 31850 }, { "epoch": 0.6517369936432118, "grad_norm": 2.896454095840454, "learning_rate": 5.715560893215398e-06, "loss": 0.14187419414520264, "step": 31860 }, { "epoch": 0.6519415564158556, "grad_norm": 1.4025285243988037, "learning_rate": 5.709575407342853e-06, "loss": 0.14019675254821778, "step": 31870 }, { "epoch": 0.6521461191884995, "grad_norm": 1.715736746788025, "learning_rate": 5.703591804744429e-06, "loss": 0.1180113434791565, "step": 31880 }, { "epoch": 0.6523506819611433, "grad_norm": 2.319931745529175, "learning_rate": 5.6976100880466135e-06, "loss": 0.11831502914428711, "step": 31890 }, { "epoch": 0.6525552447337871, "grad_norm": 2.565547227859497, "learning_rate": 5.691630259875074e-06, "loss": 0.12067873477935791, "step": 31900 }, { "epoch": 0.6527598075064309, "grad_norm": 3.0370657444000244, "learning_rate": 5.685652322854642e-06, "loss": 0.13206474781036376, "step": 31910 }, { "epoch": 0.6529643702790747, "grad_norm": 1.3418469429016113, "learning_rate": 5.679676279609322e-06, "loss": 0.12995190620422364, "step": 31920 }, { "epoch": 0.6531689330517186, "grad_norm": 1.5904730558395386, "learning_rate": 5.6737021327622925e-06, "loss": 0.1232572078704834, "step": 31930 }, { "epoch": 0.6533734958243624, "grad_norm": 2.094399929046631, "learning_rate": 5.6677298849358905e-06, "loss": 0.1378987669944763, "step": 31940 }, { "epoch": 0.6535780585970062, "grad_norm": 1.0094918012619019, "learning_rate": 5.661759538751626e-06, "loss": 0.13141411542892456, "step": 31950 }, { "epoch": 0.65378262136965, "grad_norm": 1.66864013671875, "learning_rate": 5.655791096830168e-06, "loss": 0.12555395364761351, "step": 31960 }, { "epoch": 0.6539871841422938, "grad_norm": 2.087423801422119, "learning_rate": 5.649824561791356e-06, "loss": 0.141765296459198, "step": 31970 }, { "epoch": 0.6541917469149376, "grad_norm": 2.1499814987182617, "learning_rate": 5.643859936254189e-06, "loss": 0.12668607234954835, "step": 31980 }, { "epoch": 0.6543963096875816, "grad_norm": 1.7213654518127441, "learning_rate": 5.637897222836825e-06, "loss": 0.13636338710784912, "step": 31990 }, { "epoch": 0.6546008724602254, "grad_norm": 2.3785252571105957, "learning_rate": 5.631936424156592e-06, "loss": 0.12782049179077148, "step": 32000 }, { "epoch": 0.6548054352328692, "grad_norm": 1.2031382322311401, "learning_rate": 5.625977542829971e-06, "loss": 0.14117856025695802, "step": 32010 }, { "epoch": 0.655009998005513, "grad_norm": 2.376636028289795, "learning_rate": 5.620020581472605e-06, "loss": 0.12554435729980468, "step": 32020 }, { "epoch": 0.6552145607781568, "grad_norm": 2.4983162879943848, "learning_rate": 5.614065542699278e-06, "loss": 0.13413374423980712, "step": 32030 }, { "epoch": 0.6554191235508006, "grad_norm": 1.1644130945205688, "learning_rate": 5.608112429123957e-06, "loss": 0.1266770362854004, "step": 32040 }, { "epoch": 0.6556236863234445, "grad_norm": 1.9189209938049316, "learning_rate": 5.602161243359747e-06, "loss": 0.12789992094039918, "step": 32050 }, { "epoch": 0.6558282490960883, "grad_norm": 0.7979558110237122, "learning_rate": 5.596211988018909e-06, "loss": 0.13234398365020753, "step": 32060 }, { "epoch": 0.6560328118687321, "grad_norm": 1.5770999193191528, "learning_rate": 5.590264665712854e-06, "loss": 0.1355266571044922, "step": 32070 }, { "epoch": 0.6562373746413759, "grad_norm": 1.2869828939437866, "learning_rate": 5.584319279052157e-06, "loss": 0.1281929850578308, "step": 32080 }, { "epoch": 0.6564419374140197, "grad_norm": 3.341726303100586, "learning_rate": 5.578375830646534e-06, "loss": 0.1434725046157837, "step": 32090 }, { "epoch": 0.6566465001866635, "grad_norm": 1.9049110412597656, "learning_rate": 5.572434323104843e-06, "loss": 0.13023746013641357, "step": 32100 }, { "epoch": 0.6568510629593074, "grad_norm": 1.5866318941116333, "learning_rate": 5.5664947590350995e-06, "loss": 0.11860960721969604, "step": 32110 }, { "epoch": 0.6570556257319512, "grad_norm": 2.756096124649048, "learning_rate": 5.56055714104447e-06, "loss": 0.13864036798477172, "step": 32120 }, { "epoch": 0.657260188504595, "grad_norm": 1.4404157400131226, "learning_rate": 5.554621471739259e-06, "loss": 0.12468864917755126, "step": 32130 }, { "epoch": 0.6574647512772388, "grad_norm": 2.7244205474853516, "learning_rate": 5.548687753724913e-06, "loss": 0.1343362808227539, "step": 32140 }, { "epoch": 0.6576693140498826, "grad_norm": 1.5040111541748047, "learning_rate": 5.542755989606034e-06, "loss": 0.1393275260925293, "step": 32150 }, { "epoch": 0.6578738768225264, "grad_norm": 3.7378125190734863, "learning_rate": 5.536826181986358e-06, "loss": 0.12202599048614501, "step": 32160 }, { "epoch": 0.6580784395951703, "grad_norm": 1.6178430318832397, "learning_rate": 5.530898333468764e-06, "loss": 0.11479804515838624, "step": 32170 }, { "epoch": 0.6582830023678141, "grad_norm": 2.7080371379852295, "learning_rate": 5.524972446655261e-06, "loss": 0.13901102542877197, "step": 32180 }, { "epoch": 0.6584875651404579, "grad_norm": 3.0469744205474854, "learning_rate": 5.519048524147016e-06, "loss": 0.12114301919937134, "step": 32190 }, { "epoch": 0.6586921279131017, "grad_norm": 1.3385320901870728, "learning_rate": 5.5131265685443224e-06, "loss": 0.12947943210601806, "step": 32200 }, { "epoch": 0.6588966906857455, "grad_norm": 2.0321192741394043, "learning_rate": 5.507206582446609e-06, "loss": 0.11816548109054566, "step": 32210 }, { "epoch": 0.6591012534583893, "grad_norm": 2.095531940460205, "learning_rate": 5.501288568452443e-06, "loss": 0.12806324958801268, "step": 32220 }, { "epoch": 0.6593058162310332, "grad_norm": 1.1698507070541382, "learning_rate": 5.495372529159531e-06, "loss": 0.1312075138092041, "step": 32230 }, { "epoch": 0.659510379003677, "grad_norm": 3.0865211486816406, "learning_rate": 5.489458467164706e-06, "loss": 0.12056277990341187, "step": 32240 }, { "epoch": 0.6597149417763208, "grad_norm": 1.6649938821792603, "learning_rate": 5.483546385063932e-06, "loss": 0.12112473249435425, "step": 32250 }, { "epoch": 0.6599195045489646, "grad_norm": 2.3280701637268066, "learning_rate": 5.477636285452304e-06, "loss": 0.12690353393554688, "step": 32260 }, { "epoch": 0.6601240673216084, "grad_norm": 3.8041059970855713, "learning_rate": 5.471728170924055e-06, "loss": 0.13789395093917847, "step": 32270 }, { "epoch": 0.6603286300942522, "grad_norm": 2.368626594543457, "learning_rate": 5.465822044072542e-06, "loss": 0.129323148727417, "step": 32280 }, { "epoch": 0.6605331928668962, "grad_norm": 2.372290849685669, "learning_rate": 5.459917907490242e-06, "loss": 0.12560330629348754, "step": 32290 }, { "epoch": 0.66073775563954, "grad_norm": 2.262410879135132, "learning_rate": 5.454015763768774e-06, "loss": 0.1397786855697632, "step": 32300 }, { "epoch": 0.6609423184121838, "grad_norm": 3.0484750270843506, "learning_rate": 5.448115615498873e-06, "loss": 0.12854163646697997, "step": 32310 }, { "epoch": 0.6611468811848276, "grad_norm": 2.3798694610595703, "learning_rate": 5.442217465270388e-06, "loss": 0.13144400119781494, "step": 32320 }, { "epoch": 0.6613514439574714, "grad_norm": 2.3375678062438965, "learning_rate": 5.4363213156723064e-06, "loss": 0.13141417503356934, "step": 32330 }, { "epoch": 0.6615560067301152, "grad_norm": 2.1864631175994873, "learning_rate": 5.430427169292736e-06, "loss": 0.14141230583190917, "step": 32340 }, { "epoch": 0.6617605695027591, "grad_norm": 2.3339104652404785, "learning_rate": 5.424535028718899e-06, "loss": 0.13144874572753906, "step": 32350 }, { "epoch": 0.6619651322754029, "grad_norm": 2.2369322776794434, "learning_rate": 5.418644896537142e-06, "loss": 0.11857906579971314, "step": 32360 }, { "epoch": 0.6621696950480467, "grad_norm": 2.2882425785064697, "learning_rate": 5.41275677533292e-06, "loss": 0.13851362466812134, "step": 32370 }, { "epoch": 0.6623742578206905, "grad_norm": 2.0252325534820557, "learning_rate": 5.4068706676908285e-06, "loss": 0.12557191848754884, "step": 32380 }, { "epoch": 0.6625788205933343, "grad_norm": 3.0157086849212646, "learning_rate": 5.40098657619455e-06, "loss": 0.13829689025878905, "step": 32390 }, { "epoch": 0.6627833833659782, "grad_norm": 1.9456373453140259, "learning_rate": 5.3951045034269e-06, "loss": 0.12755277156829833, "step": 32400 }, { "epoch": 0.662987946138622, "grad_norm": 1.8654470443725586, "learning_rate": 5.389224451969801e-06, "loss": 0.13832099437713624, "step": 32410 }, { "epoch": 0.6631925089112658, "grad_norm": 2.1227669715881348, "learning_rate": 5.383346424404299e-06, "loss": 0.12640275955200195, "step": 32420 }, { "epoch": 0.6633970716839096, "grad_norm": 2.708590507507324, "learning_rate": 5.377470423310536e-06, "loss": 0.12330260276794433, "step": 32430 }, { "epoch": 0.6636016344565534, "grad_norm": 2.2497403621673584, "learning_rate": 5.371596451267776e-06, "loss": 0.1296964168548584, "step": 32440 }, { "epoch": 0.6638061972291972, "grad_norm": 1.8399890661239624, "learning_rate": 5.365724510854387e-06, "loss": 0.13504469394683838, "step": 32450 }, { "epoch": 0.6640107600018411, "grad_norm": 2.107741594314575, "learning_rate": 5.3598546046478474e-06, "loss": 0.12432637214660644, "step": 32460 }, { "epoch": 0.6642153227744849, "grad_norm": 2.9153518676757812, "learning_rate": 5.353986735224741e-06, "loss": 0.14031410217285156, "step": 32470 }, { "epoch": 0.6644198855471287, "grad_norm": 1.0106456279754639, "learning_rate": 5.348120905160754e-06, "loss": 0.12622426748275756, "step": 32480 }, { "epoch": 0.6646244483197725, "grad_norm": 2.451568603515625, "learning_rate": 5.342257117030691e-06, "loss": 0.12851994037628173, "step": 32490 }, { "epoch": 0.6648290110924163, "grad_norm": 0.9793629050254822, "learning_rate": 5.336395373408447e-06, "loss": 0.12709789276123046, "step": 32500 }, { "epoch": 0.6650335738650601, "grad_norm": 2.2305281162261963, "learning_rate": 5.330535676867023e-06, "loss": 0.13977313041687012, "step": 32510 }, { "epoch": 0.665238136637704, "grad_norm": 1.3065013885498047, "learning_rate": 5.324678029978519e-06, "loss": 0.12956385612487792, "step": 32520 }, { "epoch": 0.6654426994103478, "grad_norm": 1.2456899881362915, "learning_rate": 5.31882243531415e-06, "loss": 0.13110694885253907, "step": 32530 }, { "epoch": 0.6656472621829916, "grad_norm": 3.9394495487213135, "learning_rate": 5.312968895444208e-06, "loss": 0.1306365966796875, "step": 32540 }, { "epoch": 0.6658518249556354, "grad_norm": 0.9506731033325195, "learning_rate": 5.307117412938096e-06, "loss": 0.13254184722900392, "step": 32550 }, { "epoch": 0.6660563877282792, "grad_norm": 4.359739303588867, "learning_rate": 5.301267990364309e-06, "loss": 0.14371011257171631, "step": 32560 }, { "epoch": 0.666260950500923, "grad_norm": 2.4184749126434326, "learning_rate": 5.2954206302904466e-06, "loss": 0.1179648756980896, "step": 32570 }, { "epoch": 0.666465513273567, "grad_norm": 2.129343271255493, "learning_rate": 5.289575335283195e-06, "loss": 0.12494727373123168, "step": 32580 }, { "epoch": 0.6666700760462108, "grad_norm": 1.2713935375213623, "learning_rate": 5.283732107908335e-06, "loss": 0.13306471109390258, "step": 32590 }, { "epoch": 0.6668746388188546, "grad_norm": 2.2111897468566895, "learning_rate": 5.277890950730741e-06, "loss": 0.1209554672241211, "step": 32600 }, { "epoch": 0.6670792015914984, "grad_norm": 2.5273120403289795, "learning_rate": 5.272051866314378e-06, "loss": 0.1295168876647949, "step": 32610 }, { "epoch": 0.6672837643641422, "grad_norm": 2.4940919876098633, "learning_rate": 5.266214857222302e-06, "loss": 0.11965727806091309, "step": 32620 }, { "epoch": 0.667488327136786, "grad_norm": 1.0952070951461792, "learning_rate": 5.260379926016657e-06, "loss": 0.12312567234039307, "step": 32630 }, { "epoch": 0.6676928899094299, "grad_norm": 1.0753049850463867, "learning_rate": 5.254547075258671e-06, "loss": 0.1165823221206665, "step": 32640 }, { "epoch": 0.6678974526820737, "grad_norm": 1.2835569381713867, "learning_rate": 5.248716307508672e-06, "loss": 0.12359788417816162, "step": 32650 }, { "epoch": 0.6681020154547175, "grad_norm": 0.9371799826622009, "learning_rate": 5.242887625326063e-06, "loss": 0.12774498462677003, "step": 32660 }, { "epoch": 0.6683065782273613, "grad_norm": 2.754762887954712, "learning_rate": 5.23706103126933e-06, "loss": 0.13042829036712647, "step": 32670 }, { "epoch": 0.6685111410000051, "grad_norm": 1.162441611289978, "learning_rate": 5.231236527896047e-06, "loss": 0.1337862491607666, "step": 32680 }, { "epoch": 0.6687157037726489, "grad_norm": 1.8831509351730347, "learning_rate": 5.225414117762871e-06, "loss": 0.12194844484329223, "step": 32690 }, { "epoch": 0.6689202665452928, "grad_norm": 2.1384122371673584, "learning_rate": 5.219593803425538e-06, "loss": 0.12959213256835939, "step": 32700 }, { "epoch": 0.6691248293179366, "grad_norm": 2.3262979984283447, "learning_rate": 5.2137755874388575e-06, "loss": 0.12129992246627808, "step": 32710 }, { "epoch": 0.6693293920905804, "grad_norm": 1.7273792028427124, "learning_rate": 5.207959472356736e-06, "loss": 0.12710123062133788, "step": 32720 }, { "epoch": 0.6695339548632242, "grad_norm": 1.649460792541504, "learning_rate": 5.202145460732141e-06, "loss": 0.13450514078140258, "step": 32730 }, { "epoch": 0.669738517635868, "grad_norm": 1.489539623260498, "learning_rate": 5.196333555117123e-06, "loss": 0.12560912370681762, "step": 32740 }, { "epoch": 0.6699430804085118, "grad_norm": 1.8511313199996948, "learning_rate": 5.190523758062809e-06, "loss": 0.11934508085250854, "step": 32750 }, { "epoch": 0.6701476431811557, "grad_norm": 1.829804539680481, "learning_rate": 5.184716072119396e-06, "loss": 0.12200891971588135, "step": 32760 }, { "epoch": 0.6703522059537995, "grad_norm": 0.8478111028671265, "learning_rate": 5.1789104998361585e-06, "loss": 0.11918476819992066, "step": 32770 }, { "epoch": 0.6705567687264433, "grad_norm": 1.7699662446975708, "learning_rate": 5.173107043761443e-06, "loss": 0.13099058866500854, "step": 32780 }, { "epoch": 0.6707613314990871, "grad_norm": 3.465122938156128, "learning_rate": 5.167305706442661e-06, "loss": 0.13514872789382934, "step": 32790 }, { "epoch": 0.6709658942717309, "grad_norm": 2.3468706607818604, "learning_rate": 5.161506490426307e-06, "loss": 0.11230201721191406, "step": 32800 }, { "epoch": 0.6711704570443747, "grad_norm": 1.6508289575576782, "learning_rate": 5.1557093982579334e-06, "loss": 0.1277984619140625, "step": 32810 }, { "epoch": 0.6713750198170186, "grad_norm": 3.69405460357666, "learning_rate": 5.149914432482161e-06, "loss": 0.14384729862213136, "step": 32820 }, { "epoch": 0.6715795825896624, "grad_norm": 3.4100279808044434, "learning_rate": 5.144121595642683e-06, "loss": 0.13599320650100707, "step": 32830 }, { "epoch": 0.6717841453623062, "grad_norm": 2.028301954269409, "learning_rate": 5.1383308902822505e-06, "loss": 0.12807087898254393, "step": 32840 }, { "epoch": 0.67198870813495, "grad_norm": 1.2484341859817505, "learning_rate": 5.132542318942688e-06, "loss": 0.1251958966255188, "step": 32850 }, { "epoch": 0.6721932709075938, "grad_norm": 1.2652798891067505, "learning_rate": 5.12675588416487e-06, "loss": 0.14028518199920653, "step": 32860 }, { "epoch": 0.6723978336802378, "grad_norm": 1.6647930145263672, "learning_rate": 5.120971588488752e-06, "loss": 0.13631702661514283, "step": 32870 }, { "epoch": 0.6726023964528816, "grad_norm": 1.8982353210449219, "learning_rate": 5.115189434453336e-06, "loss": 0.14059951305389404, "step": 32880 }, { "epoch": 0.6728069592255254, "grad_norm": 3.712538957595825, "learning_rate": 5.1094094245966865e-06, "loss": 0.1256887674331665, "step": 32890 }, { "epoch": 0.6730115219981692, "grad_norm": 3.1089560985565186, "learning_rate": 5.103631561455931e-06, "loss": 0.12031866312026977, "step": 32900 }, { "epoch": 0.673216084770813, "grad_norm": 1.349224328994751, "learning_rate": 5.097855847567248e-06, "loss": 0.1316269040107727, "step": 32910 }, { "epoch": 0.6734206475434568, "grad_norm": 1.4967671632766724, "learning_rate": 5.092082285465882e-06, "loss": 0.12039973735809326, "step": 32920 }, { "epoch": 0.6736252103161007, "grad_norm": 2.191127061843872, "learning_rate": 5.086310877686122e-06, "loss": 0.12381563186645508, "step": 32930 }, { "epoch": 0.6738297730887445, "grad_norm": 1.5549670457839966, "learning_rate": 5.080541626761319e-06, "loss": 0.12596747875213624, "step": 32940 }, { "epoch": 0.6740343358613883, "grad_norm": 1.9836074113845825, "learning_rate": 5.074774535223879e-06, "loss": 0.12919855117797852, "step": 32950 }, { "epoch": 0.6742388986340321, "grad_norm": 1.5540814399719238, "learning_rate": 5.0690096056052526e-06, "loss": 0.13200616836547852, "step": 32960 }, { "epoch": 0.6744434614066759, "grad_norm": 2.953725814819336, "learning_rate": 5.063246840435947e-06, "loss": 0.1325872302055359, "step": 32970 }, { "epoch": 0.6746480241793197, "grad_norm": 2.32991623878479, "learning_rate": 5.057486242245518e-06, "loss": 0.12815337181091307, "step": 32980 }, { "epoch": 0.6748525869519636, "grad_norm": 1.4296715259552002, "learning_rate": 5.051727813562566e-06, "loss": 0.13081969022750856, "step": 32990 }, { "epoch": 0.6750571497246074, "grad_norm": 1.5592888593673706, "learning_rate": 5.045971556914746e-06, "loss": 0.12293422222137451, "step": 33000 }, { "epoch": 0.6752617124972512, "grad_norm": 2.067215919494629, "learning_rate": 5.040217474828757e-06, "loss": 0.11455776691436767, "step": 33010 }, { "epoch": 0.675466275269895, "grad_norm": 1.0354689359664917, "learning_rate": 5.034465569830338e-06, "loss": 0.1270461082458496, "step": 33020 }, { "epoch": 0.6756708380425388, "grad_norm": 1.6308289766311646, "learning_rate": 5.028715844444283e-06, "loss": 0.1406090259552002, "step": 33030 }, { "epoch": 0.6758754008151826, "grad_norm": 1.3732808828353882, "learning_rate": 5.022968301194424e-06, "loss": 0.13080118894577025, "step": 33040 }, { "epoch": 0.6760799635878265, "grad_norm": 1.7542541027069092, "learning_rate": 5.01722294260363e-06, "loss": 0.14136085510253907, "step": 33050 }, { "epoch": 0.6762845263604703, "grad_norm": 2.055833101272583, "learning_rate": 5.011479771193819e-06, "loss": 0.13772246837615967, "step": 33060 }, { "epoch": 0.6764890891331141, "grad_norm": 1.0088231563568115, "learning_rate": 5.0057387894859455e-06, "loss": 0.13005418777465821, "step": 33070 }, { "epoch": 0.6766936519057579, "grad_norm": 1.2249860763549805, "learning_rate": 5.000000000000003e-06, "loss": 0.12247557640075683, "step": 33080 }, { "epoch": 0.6768982146784017, "grad_norm": 1.9240621328353882, "learning_rate": 4.994263405255016e-06, "loss": 0.11551960706710815, "step": 33090 }, { "epoch": 0.6771027774510455, "grad_norm": 2.5059964656829834, "learning_rate": 4.988529007769064e-06, "loss": 0.12766356468200685, "step": 33100 }, { "epoch": 0.6773073402236894, "grad_norm": 2.4565036296844482, "learning_rate": 4.982796810059245e-06, "loss": 0.13025944232940673, "step": 33110 }, { "epoch": 0.6775119029963332, "grad_norm": 1.8625359535217285, "learning_rate": 4.977066814641696e-06, "loss": 0.11494181156158448, "step": 33120 }, { "epoch": 0.677716465768977, "grad_norm": 2.868666172027588, "learning_rate": 4.9713390240315895e-06, "loss": 0.11779847145080566, "step": 33130 }, { "epoch": 0.6779210285416208, "grad_norm": 1.1382967233657837, "learning_rate": 4.965613440743127e-06, "loss": 0.14037753343582154, "step": 33140 }, { "epoch": 0.6781255913142646, "grad_norm": 3.8277008533477783, "learning_rate": 4.959890067289546e-06, "loss": 0.12122528553009033, "step": 33150 }, { "epoch": 0.6783301540869084, "grad_norm": 1.3484957218170166, "learning_rate": 4.954168906183108e-06, "loss": 0.11750997304916382, "step": 33160 }, { "epoch": 0.6785347168595524, "grad_norm": 1.6154025793075562, "learning_rate": 4.9484499599351045e-06, "loss": 0.11992912292480469, "step": 33170 }, { "epoch": 0.6787392796321962, "grad_norm": 1.0102299451828003, "learning_rate": 4.942733231055862e-06, "loss": 0.12017390727996827, "step": 33180 }, { "epoch": 0.67894384240484, "grad_norm": 2.2324705123901367, "learning_rate": 4.937018722054727e-06, "loss": 0.11991791725158692, "step": 33190 }, { "epoch": 0.6791484051774838, "grad_norm": 4.480473518371582, "learning_rate": 4.931306435440075e-06, "loss": 0.13932406902313232, "step": 33200 }, { "epoch": 0.6793529679501276, "grad_norm": 1.6194053888320923, "learning_rate": 4.925596373719292e-06, "loss": 0.13730744123458863, "step": 33210 }, { "epoch": 0.6795575307227714, "grad_norm": 0.8771636486053467, "learning_rate": 4.919888539398814e-06, "loss": 0.14244258403778076, "step": 33220 }, { "epoch": 0.6797620934954153, "grad_norm": 2.486022710800171, "learning_rate": 4.9141829349840766e-06, "loss": 0.1217614769935608, "step": 33230 }, { "epoch": 0.6799666562680591, "grad_norm": 2.6482810974121094, "learning_rate": 4.9084795629795435e-06, "loss": 0.13394467830657958, "step": 33240 }, { "epoch": 0.6801712190407029, "grad_norm": 2.269176959991455, "learning_rate": 4.902778425888706e-06, "loss": 0.1263999819755554, "step": 33250 }, { "epoch": 0.6803757818133467, "grad_norm": 3.566709041595459, "learning_rate": 4.897079526214064e-06, "loss": 0.11283787488937377, "step": 33260 }, { "epoch": 0.6805803445859905, "grad_norm": 3.495098352432251, "learning_rate": 4.891382866457144e-06, "loss": 0.12594738006591796, "step": 33270 }, { "epoch": 0.6807849073586343, "grad_norm": 2.0026345252990723, "learning_rate": 4.8856884491184715e-06, "loss": 0.1215167760848999, "step": 33280 }, { "epoch": 0.6809894701312782, "grad_norm": 1.4974360466003418, "learning_rate": 4.879996276697614e-06, "loss": 0.13433837890625, "step": 33290 }, { "epoch": 0.681194032903922, "grad_norm": 1.483069896697998, "learning_rate": 4.8743063516931355e-06, "loss": 0.13605536222457887, "step": 33300 }, { "epoch": 0.6813985956765658, "grad_norm": 1.4575058221817017, "learning_rate": 4.868618676602618e-06, "loss": 0.12657902240753174, "step": 33310 }, { "epoch": 0.6816031584492096, "grad_norm": 1.2367172241210938, "learning_rate": 4.8629332539226525e-06, "loss": 0.1236048698425293, "step": 33320 }, { "epoch": 0.6818077212218534, "grad_norm": 2.2732133865356445, "learning_rate": 4.857250086148854e-06, "loss": 0.11929316520690918, "step": 33330 }, { "epoch": 0.6820122839944973, "grad_norm": 2.411620616912842, "learning_rate": 4.851569175775832e-06, "loss": 0.12851362228393554, "step": 33340 }, { "epoch": 0.6822168467671411, "grad_norm": 1.2317806482315063, "learning_rate": 4.845890525297219e-06, "loss": 0.10316973924636841, "step": 33350 }, { "epoch": 0.6824214095397849, "grad_norm": 2.672203302383423, "learning_rate": 4.840214137205634e-06, "loss": 0.13378589153289794, "step": 33360 }, { "epoch": 0.6826259723124287, "grad_norm": 1.7733699083328247, "learning_rate": 4.834540013992731e-06, "loss": 0.13083479404449463, "step": 33370 }, { "epoch": 0.6828305350850725, "grad_norm": 2.8911707401275635, "learning_rate": 4.828868158149152e-06, "loss": 0.12669246196746825, "step": 33380 }, { "epoch": 0.6830350978577163, "grad_norm": 1.1203526258468628, "learning_rate": 4.823198572164541e-06, "loss": 0.11484830379486084, "step": 33390 }, { "epoch": 0.6832396606303602, "grad_norm": 3.653177499771118, "learning_rate": 4.817531258527565e-06, "loss": 0.13801791667938232, "step": 33400 }, { "epoch": 0.683444223403004, "grad_norm": 2.68183970451355, "learning_rate": 4.811866219725873e-06, "loss": 0.11722137928009033, "step": 33410 }, { "epoch": 0.6836487861756478, "grad_norm": 3.119208574295044, "learning_rate": 4.806203458246131e-06, "loss": 0.12637102603912354, "step": 33420 }, { "epoch": 0.6838533489482916, "grad_norm": 1.796674370765686, "learning_rate": 4.800542976573984e-06, "loss": 0.12847977876663208, "step": 33430 }, { "epoch": 0.6840579117209354, "grad_norm": 2.9902775287628174, "learning_rate": 4.7948847771941024e-06, "loss": 0.12612411975860596, "step": 33440 }, { "epoch": 0.6842624744935792, "grad_norm": 2.019129991531372, "learning_rate": 4.7892288625901386e-06, "loss": 0.1327721357345581, "step": 33450 }, { "epoch": 0.6844670372662232, "grad_norm": 2.432800054550171, "learning_rate": 4.783575235244746e-06, "loss": 0.13057732582092285, "step": 33460 }, { "epoch": 0.684671600038867, "grad_norm": 1.9346282482147217, "learning_rate": 4.777923897639571e-06, "loss": 0.1282055616378784, "step": 33470 }, { "epoch": 0.6848761628115108, "grad_norm": 1.8794633150100708, "learning_rate": 4.772274852255265e-06, "loss": 0.12624828815460204, "step": 33480 }, { "epoch": 0.6850807255841546, "grad_norm": 3.5647385120391846, "learning_rate": 4.766628101571466e-06, "loss": 0.12268140316009521, "step": 33490 }, { "epoch": 0.6852852883567984, "grad_norm": 1.3259268999099731, "learning_rate": 4.7609836480667985e-06, "loss": 0.12551485300064086, "step": 33500 }, { "epoch": 0.6854898511294422, "grad_norm": 1.606615424156189, "learning_rate": 4.7553414942188835e-06, "loss": 0.12143890857696533, "step": 33510 }, { "epoch": 0.6856944139020861, "grad_norm": 2.454752206802368, "learning_rate": 4.749701642504343e-06, "loss": 0.1208943247795105, "step": 33520 }, { "epoch": 0.6858989766747299, "grad_norm": 2.221681594848633, "learning_rate": 4.744064095398776e-06, "loss": 0.13147323131561278, "step": 33530 }, { "epoch": 0.6861035394473737, "grad_norm": 1.6281449794769287, "learning_rate": 4.738428855376773e-06, "loss": 0.11877031326293945, "step": 33540 }, { "epoch": 0.6863081022200175, "grad_norm": 1.500366449356079, "learning_rate": 4.73279592491191e-06, "loss": 0.13484058380126954, "step": 33550 }, { "epoch": 0.6865126649926613, "grad_norm": 2.164041042327881, "learning_rate": 4.7271653064767605e-06, "loss": 0.1296682357788086, "step": 33560 }, { "epoch": 0.6867172277653051, "grad_norm": 2.2904744148254395, "learning_rate": 4.721537002542865e-06, "loss": 0.12313714027404785, "step": 33570 }, { "epoch": 0.686921790537949, "grad_norm": 1.0727165937423706, "learning_rate": 4.715911015580756e-06, "loss": 0.11841121912002564, "step": 33580 }, { "epoch": 0.6871263533105928, "grad_norm": 1.9244304895401, "learning_rate": 4.71028734805996e-06, "loss": 0.12831946611404418, "step": 33590 }, { "epoch": 0.6873309160832366, "grad_norm": 1.6351051330566406, "learning_rate": 4.70466600244897e-06, "loss": 0.13114821910858154, "step": 33600 }, { "epoch": 0.6875354788558804, "grad_norm": 1.9728341102600098, "learning_rate": 4.699046981215266e-06, "loss": 0.12930535078048705, "step": 33610 }, { "epoch": 0.6877400416285242, "grad_norm": 3.9103353023529053, "learning_rate": 4.693430286825304e-06, "loss": 0.134604811668396, "step": 33620 }, { "epoch": 0.687944604401168, "grad_norm": 2.2640721797943115, "learning_rate": 4.687815921744531e-06, "loss": 0.12806787490844726, "step": 33630 }, { "epoch": 0.6881491671738119, "grad_norm": 2.313824415206909, "learning_rate": 4.682203888437353e-06, "loss": 0.12592382431030275, "step": 33640 }, { "epoch": 0.6883537299464557, "grad_norm": 1.789633870124817, "learning_rate": 4.676594189367166e-06, "loss": 0.11930578947067261, "step": 33650 }, { "epoch": 0.6885582927190995, "grad_norm": 1.3022372722625732, "learning_rate": 4.670986826996334e-06, "loss": 0.12218005657196045, "step": 33660 }, { "epoch": 0.6887628554917433, "grad_norm": 2.974609136581421, "learning_rate": 4.665381803786203e-06, "loss": 0.11394139528274536, "step": 33670 }, { "epoch": 0.6889674182643871, "grad_norm": 2.0726418495178223, "learning_rate": 4.659779122197085e-06, "loss": 0.12601513862609864, "step": 33680 }, { "epoch": 0.6891719810370309, "grad_norm": 3.454249858856201, "learning_rate": 4.6541787846882706e-06, "loss": 0.11225204467773438, "step": 33690 }, { "epoch": 0.6893765438096748, "grad_norm": 2.922086477279663, "learning_rate": 4.64858079371801e-06, "loss": 0.13846893310546876, "step": 33700 }, { "epoch": 0.6895811065823186, "grad_norm": 1.9963281154632568, "learning_rate": 4.6429851517435445e-06, "loss": 0.12765104770660402, "step": 33710 }, { "epoch": 0.6897856693549624, "grad_norm": 3.0150859355926514, "learning_rate": 4.6373918612210585e-06, "loss": 0.12277973890304565, "step": 33720 }, { "epoch": 0.6899902321276062, "grad_norm": 1.6481759548187256, "learning_rate": 4.6318009246057215e-06, "loss": 0.1297064781188965, "step": 33730 }, { "epoch": 0.69019479490025, "grad_norm": 3.32610821723938, "learning_rate": 4.6262123443516616e-06, "loss": 0.13509680032730104, "step": 33740 }, { "epoch": 0.6903993576728938, "grad_norm": 3.9811551570892334, "learning_rate": 4.620626122911982e-06, "loss": 0.1252094864845276, "step": 33750 }, { "epoch": 0.6906039204455378, "grad_norm": 1.5995815992355347, "learning_rate": 4.615042262738742e-06, "loss": 0.1329840064048767, "step": 33760 }, { "epoch": 0.6908084832181816, "grad_norm": 1.281201720237732, "learning_rate": 4.609460766282963e-06, "loss": 0.12896459102630614, "step": 33770 }, { "epoch": 0.6910130459908254, "grad_norm": 3.2350375652313232, "learning_rate": 4.603881635994643e-06, "loss": 0.13048231601715088, "step": 33780 }, { "epoch": 0.6912176087634692, "grad_norm": 1.8316994905471802, "learning_rate": 4.598304874322721e-06, "loss": 0.11684116125106811, "step": 33790 }, { "epoch": 0.691422171536113, "grad_norm": 1.788419246673584, "learning_rate": 4.592730483715111e-06, "loss": 0.11657769680023193, "step": 33800 }, { "epoch": 0.6916267343087569, "grad_norm": 2.663471221923828, "learning_rate": 4.587158466618675e-06, "loss": 0.1170589804649353, "step": 33810 }, { "epoch": 0.6918312970814007, "grad_norm": 2.1367340087890625, "learning_rate": 4.58158882547925e-06, "loss": 0.12882139682769775, "step": 33820 }, { "epoch": 0.6920358598540445, "grad_norm": 1.941686749458313, "learning_rate": 4.576021562741616e-06, "loss": 0.12801871299743653, "step": 33830 }, { "epoch": 0.6922404226266883, "grad_norm": 1.2375198602676392, "learning_rate": 4.570456680849512e-06, "loss": 0.14394423961639405, "step": 33840 }, { "epoch": 0.6924449853993321, "grad_norm": 1.3468290567398071, "learning_rate": 4.564894182245632e-06, "loss": 0.12876118421554567, "step": 33850 }, { "epoch": 0.6926495481719759, "grad_norm": 2.5664222240448, "learning_rate": 4.559334069371627e-06, "loss": 0.1285875678062439, "step": 33860 }, { "epoch": 0.6928541109446198, "grad_norm": 1.2101329565048218, "learning_rate": 4.553776344668097e-06, "loss": 0.13215374946594238, "step": 33870 }, { "epoch": 0.6930586737172636, "grad_norm": 2.1499619483947754, "learning_rate": 4.548221010574595e-06, "loss": 0.1219041109085083, "step": 33880 }, { "epoch": 0.6932632364899074, "grad_norm": 1.8590030670166016, "learning_rate": 4.5426680695296224e-06, "loss": 0.11935346126556397, "step": 33890 }, { "epoch": 0.6934677992625512, "grad_norm": 2.169517755508423, "learning_rate": 4.537117523970639e-06, "loss": 0.1277535080909729, "step": 33900 }, { "epoch": 0.693672362035195, "grad_norm": 2.7211008071899414, "learning_rate": 4.5315693763340425e-06, "loss": 0.12355630397796631, "step": 33910 }, { "epoch": 0.6938769248078388, "grad_norm": 1.2333908081054688, "learning_rate": 4.5260236290551855e-06, "loss": 0.12845909595489502, "step": 33920 }, { "epoch": 0.6940814875804827, "grad_norm": 1.1503349542617798, "learning_rate": 4.520480284568359e-06, "loss": 0.11818031072616578, "step": 33930 }, { "epoch": 0.6942860503531265, "grad_norm": 2.311488151550293, "learning_rate": 4.514939345306808e-06, "loss": 0.13457082509994506, "step": 33940 }, { "epoch": 0.6944906131257703, "grad_norm": 2.2554221153259277, "learning_rate": 4.5094008137027145e-06, "loss": 0.1215505838394165, "step": 33950 }, { "epoch": 0.6946951758984141, "grad_norm": 2.8627402782440186, "learning_rate": 4.503864692187207e-06, "loss": 0.1371755838394165, "step": 33960 }, { "epoch": 0.6948997386710579, "grad_norm": 3.1150217056274414, "learning_rate": 4.49833098319036e-06, "loss": 0.1144447684288025, "step": 33970 }, { "epoch": 0.6951043014437017, "grad_norm": 2.771576166152954, "learning_rate": 4.492799689141182e-06, "loss": 0.1152238130569458, "step": 33980 }, { "epoch": 0.6953088642163456, "grad_norm": 1.4456641674041748, "learning_rate": 4.487270812467626e-06, "loss": 0.12236654758453369, "step": 33990 }, { "epoch": 0.6955134269889894, "grad_norm": 2.087125301361084, "learning_rate": 4.481744355596582e-06, "loss": 0.11998085975646973, "step": 34000 }, { "epoch": 0.6957179897616332, "grad_norm": 1.5243691205978394, "learning_rate": 4.476220320953878e-06, "loss": 0.11438980102539062, "step": 34010 }, { "epoch": 0.695922552534277, "grad_norm": 2.1532089710235596, "learning_rate": 4.470698710964278e-06, "loss": 0.13692272901535035, "step": 34020 }, { "epoch": 0.6961271153069208, "grad_norm": 2.5812647342681885, "learning_rate": 4.4651795280514855e-06, "loss": 0.11925599575042725, "step": 34030 }, { "epoch": 0.6963316780795646, "grad_norm": 1.4060642719268799, "learning_rate": 4.459662774638128e-06, "loss": 0.12544645071029664, "step": 34040 }, { "epoch": 0.6965362408522086, "grad_norm": 3.409611940383911, "learning_rate": 4.454148453145786e-06, "loss": 0.1287415146827698, "step": 34050 }, { "epoch": 0.6967408036248524, "grad_norm": 1.719347596168518, "learning_rate": 4.448636565994955e-06, "loss": 0.1266083002090454, "step": 34060 }, { "epoch": 0.6969453663974962, "grad_norm": 1.7309709787368774, "learning_rate": 4.443127115605067e-06, "loss": 0.12161567211151122, "step": 34070 }, { "epoch": 0.69714992917014, "grad_norm": 1.8339436054229736, "learning_rate": 4.4376201043944876e-06, "loss": 0.11969969272613526, "step": 34080 }, { "epoch": 0.6973544919427838, "grad_norm": 1.1029446125030518, "learning_rate": 4.432115534780509e-06, "loss": 0.11418726444244384, "step": 34090 }, { "epoch": 0.6975590547154276, "grad_norm": 1.6706430912017822, "learning_rate": 4.42661340917935e-06, "loss": 0.12543152570724486, "step": 34100 }, { "epoch": 0.6977636174880715, "grad_norm": 2.8702919483184814, "learning_rate": 4.421113730006162e-06, "loss": 0.12601010799407958, "step": 34110 }, { "epoch": 0.6979681802607153, "grad_norm": 1.7761482000350952, "learning_rate": 4.415616499675012e-06, "loss": 0.1372743010520935, "step": 34120 }, { "epoch": 0.6981727430333591, "grad_norm": 2.8028764724731445, "learning_rate": 4.4101217205989086e-06, "loss": 0.11292445659637451, "step": 34130 }, { "epoch": 0.6983773058060029, "grad_norm": 2.4250593185424805, "learning_rate": 4.404629395189772e-06, "loss": 0.1349808692932129, "step": 34140 }, { "epoch": 0.6985818685786467, "grad_norm": 2.63177227973938, "learning_rate": 4.399139525858448e-06, "loss": 0.12403502464294433, "step": 34150 }, { "epoch": 0.6987864313512905, "grad_norm": 1.698150396347046, "learning_rate": 4.3936521150147045e-06, "loss": 0.12677117586135864, "step": 34160 }, { "epoch": 0.6989909941239344, "grad_norm": 0.985196053981781, "learning_rate": 4.388167165067231e-06, "loss": 0.13509150743484497, "step": 34170 }, { "epoch": 0.6991955568965782, "grad_norm": 2.719296455383301, "learning_rate": 4.382684678423634e-06, "loss": 0.12748459577560425, "step": 34180 }, { "epoch": 0.699400119669222, "grad_norm": 1.6338303089141846, "learning_rate": 4.377204657490441e-06, "loss": 0.11703474521636963, "step": 34190 }, { "epoch": 0.6996046824418658, "grad_norm": 1.4958014488220215, "learning_rate": 4.371727104673103e-06, "loss": 0.12586774826049804, "step": 34200 }, { "epoch": 0.6998092452145096, "grad_norm": 2.408937454223633, "learning_rate": 4.366252022375978e-06, "loss": 0.11503974199295045, "step": 34210 }, { "epoch": 0.7000138079871534, "grad_norm": 2.2029428482055664, "learning_rate": 4.360779413002345e-06, "loss": 0.12366080284118652, "step": 34220 }, { "epoch": 0.7002183707597973, "grad_norm": 1.426977515220642, "learning_rate": 4.355309278954394e-06, "loss": 0.12692978382110595, "step": 34230 }, { "epoch": 0.7004229335324411, "grad_norm": 2.49001145362854, "learning_rate": 4.349841622633231e-06, "loss": 0.1273016095161438, "step": 34240 }, { "epoch": 0.7006274963050849, "grad_norm": 1.6883597373962402, "learning_rate": 4.344376446438875e-06, "loss": 0.1356796383857727, "step": 34250 }, { "epoch": 0.7008320590777287, "grad_norm": 2.5103070735931396, "learning_rate": 4.338913752770253e-06, "loss": 0.1355372190475464, "step": 34260 }, { "epoch": 0.7010366218503725, "grad_norm": 1.6822837591171265, "learning_rate": 4.333453544025203e-06, "loss": 0.12104942798614501, "step": 34270 }, { "epoch": 0.7012411846230164, "grad_norm": 1.4494236707687378, "learning_rate": 4.32799582260048e-06, "loss": 0.14456896781921386, "step": 34280 }, { "epoch": 0.7014457473956602, "grad_norm": 1.8154969215393066, "learning_rate": 4.322540590891737e-06, "loss": 0.13227304220199584, "step": 34290 }, { "epoch": 0.701650310168304, "grad_norm": 1.8385032415390015, "learning_rate": 4.31708785129354e-06, "loss": 0.12675817012786866, "step": 34300 }, { "epoch": 0.7018548729409478, "grad_norm": 1.5340851545333862, "learning_rate": 4.311637606199356e-06, "loss": 0.1358325481414795, "step": 34310 }, { "epoch": 0.7020594357135916, "grad_norm": 2.4910166263580322, "learning_rate": 4.306189858001563e-06, "loss": 0.11960372924804688, "step": 34320 }, { "epoch": 0.7022639984862354, "grad_norm": 1.1658416986465454, "learning_rate": 4.3007446090914394e-06, "loss": 0.12316267490386963, "step": 34330 }, { "epoch": 0.7024685612588794, "grad_norm": 2.9993371963500977, "learning_rate": 4.295301861859163e-06, "loss": 0.12332721948623657, "step": 34340 }, { "epoch": 0.7026731240315232, "grad_norm": 2.17899227142334, "learning_rate": 4.289861618693827e-06, "loss": 0.11697278022766114, "step": 34350 }, { "epoch": 0.702877686804167, "grad_norm": 1.5448161363601685, "learning_rate": 4.284423881983411e-06, "loss": 0.13227349519729614, "step": 34360 }, { "epoch": 0.7030822495768108, "grad_norm": 1.7468149662017822, "learning_rate": 4.2789886541148e-06, "loss": 0.13105059862136842, "step": 34370 }, { "epoch": 0.7032868123494546, "grad_norm": 1.380553960800171, "learning_rate": 4.273555937473777e-06, "loss": 0.12455558776855469, "step": 34380 }, { "epoch": 0.7034913751220984, "grad_norm": 2.364638566970825, "learning_rate": 4.2681257344450246e-06, "loss": 0.11899487972259522, "step": 34390 }, { "epoch": 0.7036959378947423, "grad_norm": 1.1689321994781494, "learning_rate": 4.262698047412118e-06, "loss": 0.11376044750213624, "step": 34400 }, { "epoch": 0.7039005006673861, "grad_norm": 2.529144048690796, "learning_rate": 4.2572728787575346e-06, "loss": 0.12225215435028076, "step": 34410 }, { "epoch": 0.7041050634400299, "grad_norm": 1.4957417249679565, "learning_rate": 4.251850230862635e-06, "loss": 0.12306311130523681, "step": 34420 }, { "epoch": 0.7043096262126737, "grad_norm": 3.124459743499756, "learning_rate": 4.2464301061076895e-06, "loss": 0.12285745143890381, "step": 34430 }, { "epoch": 0.7045141889853175, "grad_norm": 2.4367685317993164, "learning_rate": 4.2410125068718484e-06, "loss": 0.13067203760147095, "step": 34440 }, { "epoch": 0.7047187517579613, "grad_norm": 2.021334409713745, "learning_rate": 4.2355974355331605e-06, "loss": 0.12431695461273193, "step": 34450 }, { "epoch": 0.7049233145306052, "grad_norm": 1.266413927078247, "learning_rate": 4.230184894468552e-06, "loss": 0.1195615291595459, "step": 34460 }, { "epoch": 0.705127877303249, "grad_norm": 3.468745708465576, "learning_rate": 4.224774886053857e-06, "loss": 0.13229185342788696, "step": 34470 }, { "epoch": 0.7053324400758928, "grad_norm": 3.327815532684326, "learning_rate": 4.219367412663786e-06, "loss": 0.11022217273712158, "step": 34480 }, { "epoch": 0.7055370028485366, "grad_norm": 1.9318064451217651, "learning_rate": 4.2139624766719365e-06, "loss": 0.14050592184066774, "step": 34490 }, { "epoch": 0.7057415656211804, "grad_norm": 2.0268046855926514, "learning_rate": 4.208560080450802e-06, "loss": 0.12684400081634523, "step": 34500 }, { "epoch": 0.7059461283938242, "grad_norm": 2.2087104320526123, "learning_rate": 4.203160226371751e-06, "loss": 0.1322023630142212, "step": 34510 }, { "epoch": 0.7061506911664681, "grad_norm": 2.616074800491333, "learning_rate": 4.197762916805039e-06, "loss": 0.12135759592056275, "step": 34520 }, { "epoch": 0.7063552539391119, "grad_norm": 3.0655932426452637, "learning_rate": 4.192368154119808e-06, "loss": 0.11177107095718383, "step": 34530 }, { "epoch": 0.7065598167117557, "grad_norm": 1.2057281732559204, "learning_rate": 4.186975940684076e-06, "loss": 0.11975986957550049, "step": 34540 }, { "epoch": 0.7067643794843995, "grad_norm": 1.5556504726409912, "learning_rate": 4.181586278864747e-06, "loss": 0.12773653268814086, "step": 34550 }, { "epoch": 0.7069689422570433, "grad_norm": 1.0667058229446411, "learning_rate": 4.176199171027604e-06, "loss": 0.1170009970664978, "step": 34560 }, { "epoch": 0.7071735050296871, "grad_norm": 1.5907223224639893, "learning_rate": 4.170814619537302e-06, "loss": 0.12165144681930543, "step": 34570 }, { "epoch": 0.707378067802331, "grad_norm": 3.3954992294311523, "learning_rate": 4.165432626757391e-06, "loss": 0.14153480529785156, "step": 34580 }, { "epoch": 0.7075826305749748, "grad_norm": 1.2224962711334229, "learning_rate": 4.160053195050281e-06, "loss": 0.10952842235565186, "step": 34590 }, { "epoch": 0.7077871933476186, "grad_norm": 2.975778341293335, "learning_rate": 4.154676326777267e-06, "loss": 0.11970064640045167, "step": 34600 }, { "epoch": 0.7079917561202624, "grad_norm": 1.3410905599594116, "learning_rate": 4.149302024298507e-06, "loss": 0.1257818341255188, "step": 34610 }, { "epoch": 0.7081963188929062, "grad_norm": 1.1147263050079346, "learning_rate": 4.14393028997305e-06, "loss": 0.12629303932189942, "step": 34620 }, { "epoch": 0.70840088166555, "grad_norm": 1.6933684349060059, "learning_rate": 4.138561126158806e-06, "loss": 0.11892161369323731, "step": 34630 }, { "epoch": 0.708605444438194, "grad_norm": 2.1651577949523926, "learning_rate": 4.133194535212561e-06, "loss": 0.12011301517486572, "step": 34640 }, { "epoch": 0.7088100072108378, "grad_norm": 3.1252663135528564, "learning_rate": 4.127830519489966e-06, "loss": 0.13169143199920655, "step": 34650 }, { "epoch": 0.7090145699834816, "grad_norm": 2.0112650394439697, "learning_rate": 4.122469081345553e-06, "loss": 0.12347431182861328, "step": 34660 }, { "epoch": 0.7092191327561254, "grad_norm": 2.1636648178100586, "learning_rate": 4.117110223132714e-06, "loss": 0.13153295516967772, "step": 34670 }, { "epoch": 0.7094236955287692, "grad_norm": 1.4031505584716797, "learning_rate": 4.111753947203703e-06, "loss": 0.12592735290527343, "step": 34680 }, { "epoch": 0.709628258301413, "grad_norm": 2.5547866821289062, "learning_rate": 4.106400255909656e-06, "loss": 0.12384788990020752, "step": 34690 }, { "epoch": 0.7098328210740569, "grad_norm": 2.5011560916900635, "learning_rate": 4.101049151600563e-06, "loss": 0.12688030004501344, "step": 34700 }, { "epoch": 0.7100373838467007, "grad_norm": 1.389089822769165, "learning_rate": 4.09570063662528e-06, "loss": 0.11660153865814209, "step": 34710 }, { "epoch": 0.7102419466193445, "grad_norm": 1.0928080081939697, "learning_rate": 4.090354713331528e-06, "loss": 0.12187976837158203, "step": 34720 }, { "epoch": 0.7104465093919883, "grad_norm": 2.4767844676971436, "learning_rate": 4.0850113840658946e-06, "loss": 0.1363250732421875, "step": 34730 }, { "epoch": 0.7106510721646321, "grad_norm": 3.048572540283203, "learning_rate": 4.079670651173826e-06, "loss": 0.12903201580047607, "step": 34740 }, { "epoch": 0.710855634937276, "grad_norm": 1.9216656684875488, "learning_rate": 4.074332516999618e-06, "loss": 0.12228317260742187, "step": 34750 }, { "epoch": 0.7110601977099198, "grad_norm": 2.6832354068756104, "learning_rate": 4.068996983886439e-06, "loss": 0.12387733459472657, "step": 34760 }, { "epoch": 0.7112647604825636, "grad_norm": 1.7121853828430176, "learning_rate": 4.063664054176314e-06, "loss": 0.11432368755340576, "step": 34770 }, { "epoch": 0.7114693232552074, "grad_norm": 2.7575666904449463, "learning_rate": 4.0583337302101225e-06, "loss": 0.12741291522979736, "step": 34780 }, { "epoch": 0.7116738860278512, "grad_norm": 2.323166847229004, "learning_rate": 4.0530060143275985e-06, "loss": 0.11372959613800049, "step": 34790 }, { "epoch": 0.711878448800495, "grad_norm": 2.412249803543091, "learning_rate": 4.04768090886733e-06, "loss": 0.12630133628845214, "step": 34800 }, { "epoch": 0.7120830115731389, "grad_norm": 1.8455746173858643, "learning_rate": 4.042358416166776e-06, "loss": 0.11911115646362305, "step": 34810 }, { "epoch": 0.7122875743457827, "grad_norm": 1.763832688331604, "learning_rate": 4.037038538562221e-06, "loss": 0.12173267602920532, "step": 34820 }, { "epoch": 0.7124921371184265, "grad_norm": 1.2301124334335327, "learning_rate": 4.031721278388821e-06, "loss": 0.13914988040924073, "step": 34830 }, { "epoch": 0.7126966998910703, "grad_norm": 1.9546626806259155, "learning_rate": 4.026406637980573e-06, "loss": 0.13280144929885865, "step": 34840 }, { "epoch": 0.7129012626637141, "grad_norm": 4.0491743087768555, "learning_rate": 4.021094619670337e-06, "loss": 0.1264018416404724, "step": 34850 }, { "epoch": 0.7131058254363579, "grad_norm": 3.7818498611450195, "learning_rate": 4.01578522578981e-06, "loss": 0.12006893157958984, "step": 34860 }, { "epoch": 0.7133103882090018, "grad_norm": 1.5488423109054565, "learning_rate": 4.010478458669537e-06, "loss": 0.11511778831481934, "step": 34870 }, { "epoch": 0.7135149509816456, "grad_norm": 2.0957276821136475, "learning_rate": 4.0051743206389206e-06, "loss": 0.10815321207046509, "step": 34880 }, { "epoch": 0.7137195137542894, "grad_norm": 1.6619906425476074, "learning_rate": 3.999872814026203e-06, "loss": 0.112542724609375, "step": 34890 }, { "epoch": 0.7139240765269332, "grad_norm": 2.6288371086120605, "learning_rate": 3.994573941158464e-06, "loss": 0.12501755952835084, "step": 34900 }, { "epoch": 0.714128639299577, "grad_norm": 1.830257534980774, "learning_rate": 3.989277704361634e-06, "loss": 0.11244940757751465, "step": 34910 }, { "epoch": 0.7143332020722208, "grad_norm": 3.3864691257476807, "learning_rate": 3.9839841059604944e-06, "loss": 0.11573271751403809, "step": 34920 }, { "epoch": 0.7145377648448648, "grad_norm": 1.6902153491973877, "learning_rate": 3.978693148278656e-06, "loss": 0.12932748794555665, "step": 34930 }, { "epoch": 0.7147423276175086, "grad_norm": 1.606546401977539, "learning_rate": 3.973404833638576e-06, "loss": 0.11963129043579102, "step": 34940 }, { "epoch": 0.7149468903901524, "grad_norm": 1.720569133758545, "learning_rate": 3.968119164361546e-06, "loss": 0.12384968996047974, "step": 34950 }, { "epoch": 0.7151514531627962, "grad_norm": 1.4583171606063843, "learning_rate": 3.962836142767712e-06, "loss": 0.1351083040237427, "step": 34960 }, { "epoch": 0.71535601593544, "grad_norm": 2.801194906234741, "learning_rate": 3.957555771176037e-06, "loss": 0.1122926950454712, "step": 34970 }, { "epoch": 0.7155605787080838, "grad_norm": 2.084244728088379, "learning_rate": 3.952278051904333e-06, "loss": 0.1122143030166626, "step": 34980 }, { "epoch": 0.7157651414807277, "grad_norm": 2.7452855110168457, "learning_rate": 3.9470029872692426e-06, "loss": 0.1273850679397583, "step": 34990 }, { "epoch": 0.7159697042533715, "grad_norm": 1.786505937576294, "learning_rate": 3.941730579586253e-06, "loss": 0.10346906185150147, "step": 35000 }, { "epoch": 0.7161742670260153, "grad_norm": 2.098776340484619, "learning_rate": 3.936460831169674e-06, "loss": 0.1182438850402832, "step": 35010 }, { "epoch": 0.7163788297986591, "grad_norm": 1.3575143814086914, "learning_rate": 3.931193744332652e-06, "loss": 0.11730644702911378, "step": 35020 }, { "epoch": 0.7165833925713029, "grad_norm": 2.241920232772827, "learning_rate": 3.925929321387165e-06, "loss": 0.12142086029052734, "step": 35030 }, { "epoch": 0.7167879553439467, "grad_norm": 1.8531467914581299, "learning_rate": 3.920667564644023e-06, "loss": 0.12379825115203857, "step": 35040 }, { "epoch": 0.7169925181165906, "grad_norm": 1.4374780654907227, "learning_rate": 3.915408476412864e-06, "loss": 0.12653234004974365, "step": 35050 }, { "epoch": 0.7171970808892344, "grad_norm": 1.4702534675598145, "learning_rate": 3.9101520590021524e-06, "loss": 0.12128746509552002, "step": 35060 }, { "epoch": 0.7174016436618782, "grad_norm": 0.915717601776123, "learning_rate": 3.904898314719189e-06, "loss": 0.11769139766693115, "step": 35070 }, { "epoch": 0.717606206434522, "grad_norm": 1.927059292793274, "learning_rate": 3.899647245870092e-06, "loss": 0.12701785564422607, "step": 35080 }, { "epoch": 0.7178107692071658, "grad_norm": 2.679797410964966, "learning_rate": 3.894398854759809e-06, "loss": 0.12693599462509156, "step": 35090 }, { "epoch": 0.7180153319798096, "grad_norm": 1.8673739433288574, "learning_rate": 3.889153143692112e-06, "loss": 0.11927423477172852, "step": 35100 }, { "epoch": 0.7182198947524535, "grad_norm": 3.8137571811676025, "learning_rate": 3.883910114969597e-06, "loss": 0.1263887405395508, "step": 35110 }, { "epoch": 0.7184244575250973, "grad_norm": 2.2836718559265137, "learning_rate": 3.878669770893681e-06, "loss": 0.13458707332611083, "step": 35120 }, { "epoch": 0.7186290202977411, "grad_norm": 2.1442134380340576, "learning_rate": 3.873432113764602e-06, "loss": 0.1285272717475891, "step": 35130 }, { "epoch": 0.7188335830703849, "grad_norm": 2.3323702812194824, "learning_rate": 3.86819714588142e-06, "loss": 0.11729657649993896, "step": 35140 }, { "epoch": 0.7190381458430287, "grad_norm": 1.6565531492233276, "learning_rate": 3.862964869542019e-06, "loss": 0.10860774517059327, "step": 35150 }, { "epoch": 0.7192427086156725, "grad_norm": 2.603283405303955, "learning_rate": 3.8577352870430955e-06, "loss": 0.11232376098632812, "step": 35160 }, { "epoch": 0.7194472713883164, "grad_norm": 2.4575603008270264, "learning_rate": 3.852508400680164e-06, "loss": 0.11531182527542114, "step": 35170 }, { "epoch": 0.7196518341609602, "grad_norm": 2.876697301864624, "learning_rate": 3.8472842127475586e-06, "loss": 0.11376359462738037, "step": 35180 }, { "epoch": 0.719856396933604, "grad_norm": 2.4831652641296387, "learning_rate": 3.842062725538424e-06, "loss": 0.12218214273452759, "step": 35190 }, { "epoch": 0.7200609597062478, "grad_norm": 4.322883605957031, "learning_rate": 3.836843941344726e-06, "loss": 0.11786596775054932, "step": 35200 }, { "epoch": 0.7202655224788916, "grad_norm": 2.369879722595215, "learning_rate": 3.831627862457238e-06, "loss": 0.12191234827041626, "step": 35210 }, { "epoch": 0.7204700852515356, "grad_norm": 2.4881341457366943, "learning_rate": 3.826414491165546e-06, "loss": 0.1085160255432129, "step": 35220 }, { "epoch": 0.7206746480241794, "grad_norm": 3.3995134830474854, "learning_rate": 3.821203829758058e-06, "loss": 0.12871729135513305, "step": 35230 }, { "epoch": 0.7208792107968232, "grad_norm": 1.7368732690811157, "learning_rate": 3.815995880521979e-06, "loss": 0.11736738681793213, "step": 35240 }, { "epoch": 0.721083773569467, "grad_norm": 1.9782094955444336, "learning_rate": 3.8107906457433284e-06, "loss": 0.12455010414123535, "step": 35250 }, { "epoch": 0.7212883363421108, "grad_norm": 1.4751276969909668, "learning_rate": 3.8055881277069374e-06, "loss": 0.13472722768783568, "step": 35260 }, { "epoch": 0.7214928991147546, "grad_norm": 2.8002336025238037, "learning_rate": 3.8003883286964395e-06, "loss": 0.12300026416778564, "step": 35270 }, { "epoch": 0.7216974618873985, "grad_norm": 2.138040781021118, "learning_rate": 3.7951912509942777e-06, "loss": 0.12646008729934693, "step": 35280 }, { "epoch": 0.7219020246600423, "grad_norm": 1.2374231815338135, "learning_rate": 3.789996896881697e-06, "loss": 0.11177700757980347, "step": 35290 }, { "epoch": 0.7221065874326861, "grad_norm": 1.3857158422470093, "learning_rate": 3.784805268638756e-06, "loss": 0.120978844165802, "step": 35300 }, { "epoch": 0.7223111502053299, "grad_norm": 2.313321113586426, "learning_rate": 3.779616368544308e-06, "loss": 0.12820019721984863, "step": 35310 }, { "epoch": 0.7225157129779737, "grad_norm": 1.6629526615142822, "learning_rate": 3.774430198876009e-06, "loss": 0.11801908016204835, "step": 35320 }, { "epoch": 0.7227202757506175, "grad_norm": 1.6712790727615356, "learning_rate": 3.769246761910321e-06, "loss": 0.12226498126983643, "step": 35330 }, { "epoch": 0.7229248385232614, "grad_norm": 1.6703463792800903, "learning_rate": 3.7640660599225043e-06, "loss": 0.12370051145553589, "step": 35340 }, { "epoch": 0.7231294012959052, "grad_norm": 1.8073246479034424, "learning_rate": 3.758888095186618e-06, "loss": 0.1124636173248291, "step": 35350 }, { "epoch": 0.723333964068549, "grad_norm": 2.5975112915039062, "learning_rate": 3.753712869975521e-06, "loss": 0.11664439439773559, "step": 35360 }, { "epoch": 0.7235385268411928, "grad_norm": 2.2920727729797363, "learning_rate": 3.7485403865608637e-06, "loss": 0.12149423360824585, "step": 35370 }, { "epoch": 0.7237430896138366, "grad_norm": 1.621379017829895, "learning_rate": 3.743370647213107e-06, "loss": 0.11116753816604615, "step": 35380 }, { "epoch": 0.7239476523864804, "grad_norm": 1.2018598318099976, "learning_rate": 3.7382036542014956e-06, "loss": 0.1066655158996582, "step": 35390 }, { "epoch": 0.7241522151591243, "grad_norm": 1.442348599433899, "learning_rate": 3.733039409794069e-06, "loss": 0.12936482429504395, "step": 35400 }, { "epoch": 0.7243567779317681, "grad_norm": 2.912463665008545, "learning_rate": 3.7278779162576673e-06, "loss": 0.13800244331359862, "step": 35410 }, { "epoch": 0.7245613407044119, "grad_norm": 4.063827991485596, "learning_rate": 3.722719175857915e-06, "loss": 0.11520228385925294, "step": 35420 }, { "epoch": 0.7247659034770557, "grad_norm": 1.5545985698699951, "learning_rate": 3.7175631908592326e-06, "loss": 0.11440271139144897, "step": 35430 }, { "epoch": 0.7249704662496995, "grad_norm": 3.1880199909210205, "learning_rate": 3.712409963524828e-06, "loss": 0.12164275646209717, "step": 35440 }, { "epoch": 0.7251750290223433, "grad_norm": 0.9801573753356934, "learning_rate": 3.7072594961167085e-06, "loss": 0.11123331785202026, "step": 35450 }, { "epoch": 0.7253795917949872, "grad_norm": 1.5699238777160645, "learning_rate": 3.702111790895657e-06, "loss": 0.11107139587402344, "step": 35460 }, { "epoch": 0.725584154567631, "grad_norm": 2.7458574771881104, "learning_rate": 3.6969668501212495e-06, "loss": 0.13210248947143555, "step": 35470 }, { "epoch": 0.7257887173402748, "grad_norm": 2.050502061843872, "learning_rate": 3.6918246760518504e-06, "loss": 0.12473039627075196, "step": 35480 }, { "epoch": 0.7259932801129186, "grad_norm": 1.9934293031692505, "learning_rate": 3.6866852709446064e-06, "loss": 0.11792312860488892, "step": 35490 }, { "epoch": 0.7261978428855624, "grad_norm": 1.3004939556121826, "learning_rate": 3.681548637055451e-06, "loss": 0.1147811770439148, "step": 35500 }, { "epoch": 0.7264024056582062, "grad_norm": 2.4098663330078125, "learning_rate": 3.676414776639099e-06, "loss": 0.1245067834854126, "step": 35510 }, { "epoch": 0.7266069684308502, "grad_norm": 2.315943956375122, "learning_rate": 3.6712836919490468e-06, "loss": 0.1256554126739502, "step": 35520 }, { "epoch": 0.726811531203494, "grad_norm": 1.9395877122879028, "learning_rate": 3.666155385237581e-06, "loss": 0.12232948541641235, "step": 35530 }, { "epoch": 0.7270160939761378, "grad_norm": 1.115745186805725, "learning_rate": 3.661029858755759e-06, "loss": 0.12585694789886476, "step": 35540 }, { "epoch": 0.7272206567487816, "grad_norm": 1.6488757133483887, "learning_rate": 3.655907114753422e-06, "loss": 0.1203235149383545, "step": 35550 }, { "epoch": 0.7274252195214254, "grad_norm": 2.3090779781341553, "learning_rate": 3.6507871554791885e-06, "loss": 0.11356645822525024, "step": 35560 }, { "epoch": 0.7276297822940692, "grad_norm": 1.4342212677001953, "learning_rate": 3.6456699831804567e-06, "loss": 0.1071843147277832, "step": 35570 }, { "epoch": 0.7278343450667131, "grad_norm": 1.9384334087371826, "learning_rate": 3.640555600103398e-06, "loss": 0.11857860088348389, "step": 35580 }, { "epoch": 0.7280389078393569, "grad_norm": 2.112189769744873, "learning_rate": 3.63544400849296e-06, "loss": 0.12635765075683594, "step": 35590 }, { "epoch": 0.7282434706120007, "grad_norm": 1.2992579936981201, "learning_rate": 3.6303352105928733e-06, "loss": 0.10459678173065186, "step": 35600 }, { "epoch": 0.7284480333846445, "grad_norm": 1.8427317142486572, "learning_rate": 3.625229208645631e-06, "loss": 0.12205730676651001, "step": 35610 }, { "epoch": 0.7286525961572883, "grad_norm": 1.7918146848678589, "learning_rate": 3.620126004892505e-06, "loss": 0.1240992784500122, "step": 35620 }, { "epoch": 0.7288571589299322, "grad_norm": 1.8929136991500854, "learning_rate": 3.615025601573535e-06, "loss": 0.11718647480010987, "step": 35630 }, { "epoch": 0.729061721702576, "grad_norm": 1.487849473953247, "learning_rate": 3.609928000927536e-06, "loss": 0.1123380184173584, "step": 35640 }, { "epoch": 0.7292662844752198, "grad_norm": 1.1836504936218262, "learning_rate": 3.60483320519209e-06, "loss": 0.11156517267227173, "step": 35650 }, { "epoch": 0.7294708472478636, "grad_norm": 1.3112519979476929, "learning_rate": 3.5997412166035483e-06, "loss": 0.10878185033798218, "step": 35660 }, { "epoch": 0.7296754100205074, "grad_norm": 1.6516841650009155, "learning_rate": 3.594652037397026e-06, "loss": 0.12293515205383301, "step": 35670 }, { "epoch": 0.7298799727931512, "grad_norm": 2.127901077270508, "learning_rate": 3.5895656698064175e-06, "loss": 0.12946550846099852, "step": 35680 }, { "epoch": 0.7300845355657951, "grad_norm": 1.8631808757781982, "learning_rate": 3.5844821160643695e-06, "loss": 0.117686927318573, "step": 35690 }, { "epoch": 0.7302890983384389, "grad_norm": 2.365816116333008, "learning_rate": 3.5794013784022997e-06, "loss": 0.12999603748321534, "step": 35700 }, { "epoch": 0.7304936611110827, "grad_norm": 1.5871573686599731, "learning_rate": 3.574323459050388e-06, "loss": 0.1220768928527832, "step": 35710 }, { "epoch": 0.7306982238837265, "grad_norm": 3.571991205215454, "learning_rate": 3.5692483602375793e-06, "loss": 0.11838352680206299, "step": 35720 }, { "epoch": 0.7309027866563703, "grad_norm": 2.159677267074585, "learning_rate": 3.5641760841915774e-06, "loss": 0.12125611305236816, "step": 35730 }, { "epoch": 0.7311073494290141, "grad_norm": 2.605245590209961, "learning_rate": 3.55910663313885e-06, "loss": 0.12809081077575685, "step": 35740 }, { "epoch": 0.731311912201658, "grad_norm": 2.0756771564483643, "learning_rate": 3.554040009304619e-06, "loss": 0.11896662712097168, "step": 35750 }, { "epoch": 0.7315164749743018, "grad_norm": 1.2818580865859985, "learning_rate": 3.548976214912878e-06, "loss": 0.13416769504547119, "step": 35760 }, { "epoch": 0.7317210377469456, "grad_norm": 1.8479281663894653, "learning_rate": 3.543915252186365e-06, "loss": 0.11253933906555176, "step": 35770 }, { "epoch": 0.7319256005195894, "grad_norm": 1.6088777780532837, "learning_rate": 3.538857123346583e-06, "loss": 0.1284428596496582, "step": 35780 }, { "epoch": 0.7321301632922332, "grad_norm": 2.2755489349365234, "learning_rate": 3.5338018306137855e-06, "loss": 0.11005910634994506, "step": 35790 }, { "epoch": 0.732334726064877, "grad_norm": 2.371213436126709, "learning_rate": 3.528749376206987e-06, "loss": 0.11545331478118896, "step": 35800 }, { "epoch": 0.732539288837521, "grad_norm": 2.3691999912261963, "learning_rate": 3.5236997623439505e-06, "loss": 0.12492235898971557, "step": 35810 }, { "epoch": 0.7327438516101648, "grad_norm": 2.223280668258667, "learning_rate": 3.518652991241195e-06, "loss": 0.11515583992004394, "step": 35820 }, { "epoch": 0.7329484143828086, "grad_norm": 2.004854202270508, "learning_rate": 3.5136090651139944e-06, "loss": 0.12681167125701903, "step": 35830 }, { "epoch": 0.7331529771554524, "grad_norm": 2.029726982116699, "learning_rate": 3.50856798617637e-06, "loss": 0.11363626718521118, "step": 35840 }, { "epoch": 0.7333575399280962, "grad_norm": 2.147834300994873, "learning_rate": 3.5035297566410963e-06, "loss": 0.1206203579902649, "step": 35850 }, { "epoch": 0.73356210270074, "grad_norm": 2.2393836975097656, "learning_rate": 3.498494378719687e-06, "loss": 0.13029371500015258, "step": 35860 }, { "epoch": 0.7337666654733839, "grad_norm": 2.2492971420288086, "learning_rate": 3.493461854622421e-06, "loss": 0.12152979373931885, "step": 35870 }, { "epoch": 0.7339712282460277, "grad_norm": 2.081326484680176, "learning_rate": 3.4884321865583103e-06, "loss": 0.11856601238250733, "step": 35880 }, { "epoch": 0.7341757910186715, "grad_norm": 1.0615943670272827, "learning_rate": 3.483405376735123e-06, "loss": 0.11837002038955688, "step": 35890 }, { "epoch": 0.7343803537913153, "grad_norm": 1.4571964740753174, "learning_rate": 3.4783814273593606e-06, "loss": 0.10073102712631225, "step": 35900 }, { "epoch": 0.7345849165639591, "grad_norm": 2.763873815536499, "learning_rate": 3.473360340636286e-06, "loss": 0.1132930040359497, "step": 35910 }, { "epoch": 0.7347894793366029, "grad_norm": 3.273087978363037, "learning_rate": 3.468342118769894e-06, "loss": 0.12559597492218016, "step": 35920 }, { "epoch": 0.7349940421092468, "grad_norm": 2.279306411743164, "learning_rate": 3.46332676396292e-06, "loss": 0.12629886865615844, "step": 35930 }, { "epoch": 0.7351986048818906, "grad_norm": 3.8849194049835205, "learning_rate": 3.458314278416842e-06, "loss": 0.12288186550140381, "step": 35940 }, { "epoch": 0.7354031676545344, "grad_norm": 3.49189829826355, "learning_rate": 3.45330466433189e-06, "loss": 0.12742855548858642, "step": 35950 }, { "epoch": 0.7356077304271782, "grad_norm": 2.6775951385498047, "learning_rate": 3.4482979239070226e-06, "loss": 0.12350754737854004, "step": 35960 }, { "epoch": 0.735812293199822, "grad_norm": 1.5151385068893433, "learning_rate": 3.4432940593399332e-06, "loss": 0.11035189628601075, "step": 35970 }, { "epoch": 0.7360168559724658, "grad_norm": 1.3762644529342651, "learning_rate": 3.438293072827068e-06, "loss": 0.11661758422851562, "step": 35980 }, { "epoch": 0.7362214187451097, "grad_norm": 2.19059681892395, "learning_rate": 3.4332949665635975e-06, "loss": 0.11547369956970215, "step": 35990 }, { "epoch": 0.7364259815177535, "grad_norm": 2.181316375732422, "learning_rate": 3.428299742743434e-06, "loss": 0.12068355083465576, "step": 36000 }, { "epoch": 0.7366305442903973, "grad_norm": 1.4152367115020752, "learning_rate": 3.4233074035592118e-06, "loss": 0.11451888084411621, "step": 36010 }, { "epoch": 0.7368351070630411, "grad_norm": 2.1437106132507324, "learning_rate": 3.41831795120232e-06, "loss": 0.11903369426727295, "step": 36020 }, { "epoch": 0.7370396698356849, "grad_norm": 2.0970747470855713, "learning_rate": 3.413331387862866e-06, "loss": 0.11239324808120728, "step": 36030 }, { "epoch": 0.7372442326083287, "grad_norm": 1.722307562828064, "learning_rate": 3.408347715729693e-06, "loss": 0.1091805100440979, "step": 36040 }, { "epoch": 0.7374487953809726, "grad_norm": 2.08353853225708, "learning_rate": 3.403366936990371e-06, "loss": 0.12819671630859375, "step": 36050 }, { "epoch": 0.7376533581536164, "grad_norm": 2.566129684448242, "learning_rate": 3.3983890538312116e-06, "loss": 0.12216682434082031, "step": 36060 }, { "epoch": 0.7378579209262602, "grad_norm": 1.4964097738265991, "learning_rate": 3.3934140684372454e-06, "loss": 0.1237375020980835, "step": 36070 }, { "epoch": 0.738062483698904, "grad_norm": 1.7396669387817383, "learning_rate": 3.388441982992229e-06, "loss": 0.11541461944580078, "step": 36080 }, { "epoch": 0.7382670464715478, "grad_norm": 1.6042191982269287, "learning_rate": 3.3834727996786487e-06, "loss": 0.1127699375152588, "step": 36090 }, { "epoch": 0.7384716092441918, "grad_norm": 2.299248695373535, "learning_rate": 3.3785065206777257e-06, "loss": 0.1174281358718872, "step": 36100 }, { "epoch": 0.7386761720168356, "grad_norm": 2.1685733795166016, "learning_rate": 3.373543148169397e-06, "loss": 0.12406003475189209, "step": 36110 }, { "epoch": 0.7388807347894794, "grad_norm": 1.7537325620651245, "learning_rate": 3.3685826843323246e-06, "loss": 0.1346908450126648, "step": 36120 }, { "epoch": 0.7390852975621232, "grad_norm": 1.5875651836395264, "learning_rate": 3.3636251313438904e-06, "loss": 0.11820716857910156, "step": 36130 }, { "epoch": 0.739289860334767, "grad_norm": 1.5566469430923462, "learning_rate": 3.358670491380217e-06, "loss": 0.11636309623718262, "step": 36140 }, { "epoch": 0.7394944231074108, "grad_norm": 1.859330177307129, "learning_rate": 3.353718766616122e-06, "loss": 0.12260429859161377, "step": 36150 }, { "epoch": 0.7396989858800547, "grad_norm": 1.5596803426742554, "learning_rate": 3.348769959225158e-06, "loss": 0.11447587013244628, "step": 36160 }, { "epoch": 0.7399035486526985, "grad_norm": 1.5941722393035889, "learning_rate": 3.343824071379599e-06, "loss": 0.1131603479385376, "step": 36170 }, { "epoch": 0.7401081114253423, "grad_norm": 1.8871006965637207, "learning_rate": 3.3388811052504324e-06, "loss": 0.11564323902130128, "step": 36180 }, { "epoch": 0.7403126741979861, "grad_norm": 2.820683717727661, "learning_rate": 3.333941063007362e-06, "loss": 0.11888689994812011, "step": 36190 }, { "epoch": 0.7405172369706299, "grad_norm": 2.4065399169921875, "learning_rate": 3.3290039468188106e-06, "loss": 0.13053679466247559, "step": 36200 }, { "epoch": 0.7407217997432737, "grad_norm": 1.3964186906814575, "learning_rate": 3.324069758851923e-06, "loss": 0.12982919216156005, "step": 36210 }, { "epoch": 0.7409263625159176, "grad_norm": 1.51077401638031, "learning_rate": 3.319138501272544e-06, "loss": 0.12254202365875244, "step": 36220 }, { "epoch": 0.7411309252885614, "grad_norm": 4.047049522399902, "learning_rate": 3.314210176245244e-06, "loss": 0.11521098613739014, "step": 36230 }, { "epoch": 0.7413354880612052, "grad_norm": 2.8401856422424316, "learning_rate": 3.3092847859332965e-06, "loss": 0.11625771522521973, "step": 36240 }, { "epoch": 0.741540050833849, "grad_norm": 1.7414060831069946, "learning_rate": 3.304362332498702e-06, "loss": 0.11148158311843873, "step": 36250 }, { "epoch": 0.7417446136064928, "grad_norm": 1.5475789308547974, "learning_rate": 3.2994428181021575e-06, "loss": 0.11665995121002197, "step": 36260 }, { "epoch": 0.7419491763791366, "grad_norm": 2.1615588665008545, "learning_rate": 3.2945262449030768e-06, "loss": 0.10875562429428101, "step": 36270 }, { "epoch": 0.7421537391517805, "grad_norm": 1.5856786966323853, "learning_rate": 3.2896126150595785e-06, "loss": 0.10675514936447143, "step": 36280 }, { "epoch": 0.7423583019244243, "grad_norm": 1.8410050868988037, "learning_rate": 3.2847019307284933e-06, "loss": 0.11747121810913086, "step": 36290 }, { "epoch": 0.7425628646970681, "grad_norm": 1.3598679304122925, "learning_rate": 3.2797941940653556e-06, "loss": 0.1116415023803711, "step": 36300 }, { "epoch": 0.7427674274697119, "grad_norm": 2.1899449825286865, "learning_rate": 3.2748894072244087e-06, "loss": 0.1222966194152832, "step": 36310 }, { "epoch": 0.7429719902423557, "grad_norm": 2.0106308460235596, "learning_rate": 3.269987572358597e-06, "loss": 0.11503407955169678, "step": 36320 }, { "epoch": 0.7431765530149995, "grad_norm": 1.6321748495101929, "learning_rate": 3.265088691619578e-06, "loss": 0.1135839581489563, "step": 36330 }, { "epoch": 0.7433811157876434, "grad_norm": 1.525999665260315, "learning_rate": 3.2601927671577026e-06, "loss": 0.1173673391342163, "step": 36340 }, { "epoch": 0.7435856785602872, "grad_norm": 2.1645348072052, "learning_rate": 3.255299801122026e-06, "loss": 0.12219831943511963, "step": 36350 }, { "epoch": 0.743790241332931, "grad_norm": 3.3188352584838867, "learning_rate": 3.2504097956603154e-06, "loss": 0.11771972179412842, "step": 36360 }, { "epoch": 0.7439948041055748, "grad_norm": 1.8750399351119995, "learning_rate": 3.2455227529190215e-06, "loss": 0.12659986019134523, "step": 36370 }, { "epoch": 0.7441993668782186, "grad_norm": 1.7271907329559326, "learning_rate": 3.240638675043305e-06, "loss": 0.11236906051635742, "step": 36380 }, { "epoch": 0.7444039296508624, "grad_norm": 3.434962749481201, "learning_rate": 3.23575756417702e-06, "loss": 0.1133357048034668, "step": 36390 }, { "epoch": 0.7446084924235064, "grad_norm": 2.297898769378662, "learning_rate": 3.230879422462728e-06, "loss": 0.12128504514694213, "step": 36400 }, { "epoch": 0.7448130551961502, "grad_norm": 2.2754733562469482, "learning_rate": 3.226004252041677e-06, "loss": 0.1192152738571167, "step": 36410 }, { "epoch": 0.745017617968794, "grad_norm": 1.6304231882095337, "learning_rate": 3.221132055053814e-06, "loss": 0.11335189342498779, "step": 36420 }, { "epoch": 0.7452221807414378, "grad_norm": 2.1700427532196045, "learning_rate": 3.216262833637782e-06, "loss": 0.11608661413192749, "step": 36430 }, { "epoch": 0.7454267435140816, "grad_norm": 2.0382697582244873, "learning_rate": 3.2113965899309164e-06, "loss": 0.11981576681137085, "step": 36440 }, { "epoch": 0.7456313062867254, "grad_norm": 1.7857966423034668, "learning_rate": 3.2065333260692467e-06, "loss": 0.11310023069381714, "step": 36450 }, { "epoch": 0.7458358690593693, "grad_norm": 2.2672359943389893, "learning_rate": 3.2016730441874923e-06, "loss": 0.11972370147705078, "step": 36460 }, { "epoch": 0.7460404318320131, "grad_norm": 2.3473775386810303, "learning_rate": 3.196815746419063e-06, "loss": 0.12581377029418944, "step": 36470 }, { "epoch": 0.7462449946046569, "grad_norm": 2.1821534633636475, "learning_rate": 3.191961434896067e-06, "loss": 0.12203001976013184, "step": 36480 }, { "epoch": 0.7464495573773007, "grad_norm": 2.0921404361724854, "learning_rate": 3.187110111749292e-06, "loss": 0.11290847063064575, "step": 36490 }, { "epoch": 0.7466541201499445, "grad_norm": 2.7962138652801514, "learning_rate": 3.182261779108219e-06, "loss": 0.12946720123291017, "step": 36500 }, { "epoch": 0.7468586829225883, "grad_norm": 3.844698667526245, "learning_rate": 3.177416439101013e-06, "loss": 0.13684167861938476, "step": 36510 }, { "epoch": 0.7470632456952322, "grad_norm": 2.2799479961395264, "learning_rate": 3.172574093854529e-06, "loss": 0.11143262386322021, "step": 36520 }, { "epoch": 0.747267808467876, "grad_norm": 1.8613498210906982, "learning_rate": 3.167734745494304e-06, "loss": 0.10977815389633179, "step": 36530 }, { "epoch": 0.7474723712405198, "grad_norm": 2.513310432434082, "learning_rate": 3.162898396144558e-06, "loss": 0.11629858016967773, "step": 36540 }, { "epoch": 0.7476769340131636, "grad_norm": 2.0046563148498535, "learning_rate": 3.158065047928206e-06, "loss": 0.11948730945587158, "step": 36550 }, { "epoch": 0.7478814967858074, "grad_norm": 3.134539842605591, "learning_rate": 3.153234702966833e-06, "loss": 0.10846922397613526, "step": 36560 }, { "epoch": 0.7480860595584513, "grad_norm": 1.8478529453277588, "learning_rate": 3.148407363380711e-06, "loss": 0.10286040306091308, "step": 36570 }, { "epoch": 0.7482906223310951, "grad_norm": 1.8837872743606567, "learning_rate": 3.1435830312887906e-06, "loss": 0.10930511951446534, "step": 36580 }, { "epoch": 0.7484951851037389, "grad_norm": 2.8895840644836426, "learning_rate": 3.138761708808705e-06, "loss": 0.12376210689544678, "step": 36590 }, { "epoch": 0.7486997478763827, "grad_norm": 1.9932477474212646, "learning_rate": 3.1339433980567646e-06, "loss": 0.11729518175125123, "step": 36600 }, { "epoch": 0.7489043106490265, "grad_norm": 1.6345919370651245, "learning_rate": 3.1291281011479568e-06, "loss": 0.1202324628829956, "step": 36610 }, { "epoch": 0.7491088734216703, "grad_norm": 2.2650368213653564, "learning_rate": 3.1243158201959468e-06, "loss": 0.1141711950302124, "step": 36620 }, { "epoch": 0.7493134361943142, "grad_norm": 5.070582389831543, "learning_rate": 3.119506557313081e-06, "loss": 0.12175199985504151, "step": 36630 }, { "epoch": 0.749517998966958, "grad_norm": 1.9194847345352173, "learning_rate": 3.114700314610375e-06, "loss": 0.11585799455642701, "step": 36640 }, { "epoch": 0.7497225617396018, "grad_norm": 2.035280227661133, "learning_rate": 3.1098970941975183e-06, "loss": 0.11258153915405274, "step": 36650 }, { "epoch": 0.7499271245122456, "grad_norm": 1.5747873783111572, "learning_rate": 3.105096898182878e-06, "loss": 0.11286060810089112, "step": 36660 }, { "epoch": 0.7501316872848894, "grad_norm": 2.3806517124176025, "learning_rate": 3.1002997286734904e-06, "loss": 0.11106382608413697, "step": 36670 }, { "epoch": 0.7503362500575332, "grad_norm": 2.8953237533569336, "learning_rate": 3.095505587775064e-06, "loss": 0.10623807907104492, "step": 36680 }, { "epoch": 0.7505408128301772, "grad_norm": 1.8367462158203125, "learning_rate": 3.0907144775919797e-06, "loss": 0.11289877891540527, "step": 36690 }, { "epoch": 0.750745375602821, "grad_norm": 2.0378599166870117, "learning_rate": 3.0859264002272814e-06, "loss": 0.11016502380371093, "step": 36700 }, { "epoch": 0.7509499383754648, "grad_norm": 1.3886181116104126, "learning_rate": 3.0811413577826953e-06, "loss": 0.10918376445770264, "step": 36710 }, { "epoch": 0.7511545011481086, "grad_norm": 1.9512579441070557, "learning_rate": 3.076359352358603e-06, "loss": 0.10368432998657226, "step": 36720 }, { "epoch": 0.7513590639207524, "grad_norm": 2.0567052364349365, "learning_rate": 3.0715803860540574e-06, "loss": 0.10967028141021729, "step": 36730 }, { "epoch": 0.7515636266933962, "grad_norm": 1.8321845531463623, "learning_rate": 3.066804460966777e-06, "loss": 0.12331467866897583, "step": 36740 }, { "epoch": 0.7517681894660401, "grad_norm": 1.9302456378936768, "learning_rate": 3.0620315791931455e-06, "loss": 0.11053702831268311, "step": 36750 }, { "epoch": 0.7519727522386839, "grad_norm": 1.8144909143447876, "learning_rate": 3.0572617428282105e-06, "loss": 0.11294584274291992, "step": 36760 }, { "epoch": 0.7521773150113277, "grad_norm": 3.20654034614563, "learning_rate": 3.05249495396568e-06, "loss": 0.11183665990829468, "step": 36770 }, { "epoch": 0.7523818777839715, "grad_norm": 2.521188497543335, "learning_rate": 3.0477312146979344e-06, "loss": 0.11683967113494872, "step": 36780 }, { "epoch": 0.7525864405566153, "grad_norm": 1.4607806205749512, "learning_rate": 3.0429705271160035e-06, "loss": 0.11629369258880615, "step": 36790 }, { "epoch": 0.7527910033292591, "grad_norm": 1.6960718631744385, "learning_rate": 3.0382128933095847e-06, "loss": 0.1069232702255249, "step": 36800 }, { "epoch": 0.752995566101903, "grad_norm": 1.730867862701416, "learning_rate": 3.0334583153670316e-06, "loss": 0.10735393762588501, "step": 36810 }, { "epoch": 0.7532001288745468, "grad_norm": 1.8189083337783813, "learning_rate": 3.028706795375358e-06, "loss": 0.10441347360610961, "step": 36820 }, { "epoch": 0.7534046916471906, "grad_norm": 1.7759982347488403, "learning_rate": 3.023958335420234e-06, "loss": 0.11103798151016235, "step": 36830 }, { "epoch": 0.7536092544198344, "grad_norm": 2.52777099609375, "learning_rate": 3.0192129375859893e-06, "loss": 0.11625714302062988, "step": 36840 }, { "epoch": 0.7538138171924782, "grad_norm": 2.6110172271728516, "learning_rate": 3.0144706039556025e-06, "loss": 0.12161712646484375, "step": 36850 }, { "epoch": 0.754018379965122, "grad_norm": 1.674654245376587, "learning_rate": 3.0097313366107204e-06, "loss": 0.12713892459869386, "step": 36860 }, { "epoch": 0.7542229427377659, "grad_norm": 1.6526833772659302, "learning_rate": 3.0049951376316313e-06, "loss": 0.11984338760375976, "step": 36870 }, { "epoch": 0.7544275055104097, "grad_norm": 2.8238770961761475, "learning_rate": 3.000262009097282e-06, "loss": 0.11878776550292969, "step": 36880 }, { "epoch": 0.7546320682830535, "grad_norm": 2.5967111587524414, "learning_rate": 2.995531953085271e-06, "loss": 0.1165350079536438, "step": 36890 }, { "epoch": 0.7548366310556973, "grad_norm": 2.8336009979248047, "learning_rate": 2.9908049716718468e-06, "loss": 0.12249056100845337, "step": 36900 }, { "epoch": 0.7550411938283411, "grad_norm": 1.8056553602218628, "learning_rate": 2.9860810669319105e-06, "loss": 0.11750254631042481, "step": 36910 }, { "epoch": 0.7552457566009849, "grad_norm": 2.9175918102264404, "learning_rate": 2.9813602409390086e-06, "loss": 0.11909763813018799, "step": 36920 }, { "epoch": 0.7554503193736288, "grad_norm": 1.6229579448699951, "learning_rate": 2.9766424957653427e-06, "loss": 0.11647422313690185, "step": 36930 }, { "epoch": 0.7556548821462726, "grad_norm": 2.058022975921631, "learning_rate": 2.9719278334817592e-06, "loss": 0.11358931064605712, "step": 36940 }, { "epoch": 0.7558594449189164, "grad_norm": 1.9323077201843262, "learning_rate": 2.967216256157748e-06, "loss": 0.12315716743469238, "step": 36950 }, { "epoch": 0.7560640076915602, "grad_norm": 2.2447023391723633, "learning_rate": 2.9625077658614475e-06, "loss": 0.11151050329208374, "step": 36960 }, { "epoch": 0.756268570464204, "grad_norm": 2.1556637287139893, "learning_rate": 2.9578023646596434e-06, "loss": 0.12061929702758789, "step": 36970 }, { "epoch": 0.7564731332368478, "grad_norm": 1.2678357362747192, "learning_rate": 2.953100054617759e-06, "loss": 0.10629173517227172, "step": 36980 }, { "epoch": 0.7566776960094918, "grad_norm": 2.0225210189819336, "learning_rate": 2.9484008377998684e-06, "loss": 0.10620439052581787, "step": 36990 }, { "epoch": 0.7568822587821356, "grad_norm": 1.6452574729919434, "learning_rate": 2.9437047162686806e-06, "loss": 0.10576498508453369, "step": 37000 }, { "epoch": 0.7570868215547794, "grad_norm": 1.7231546640396118, "learning_rate": 2.9390116920855548e-06, "loss": 0.12337725162506104, "step": 37010 }, { "epoch": 0.7572913843274232, "grad_norm": 1.3803656101226807, "learning_rate": 2.9343217673104828e-06, "loss": 0.12000672817230225, "step": 37020 }, { "epoch": 0.757495947100067, "grad_norm": 1.911597728729248, "learning_rate": 2.929634944002101e-06, "loss": 0.09981738328933716, "step": 37030 }, { "epoch": 0.7577005098727109, "grad_norm": 3.1166441440582275, "learning_rate": 2.9249512242176735e-06, "loss": 0.11740479469299317, "step": 37040 }, { "epoch": 0.7579050726453547, "grad_norm": 2.201512336730957, "learning_rate": 2.9202706100131207e-06, "loss": 0.10246930122375489, "step": 37050 }, { "epoch": 0.7581096354179985, "grad_norm": 2.398397922515869, "learning_rate": 2.9155931034429853e-06, "loss": 0.11395845413208008, "step": 37060 }, { "epoch": 0.7583141981906423, "grad_norm": 2.9091410636901855, "learning_rate": 2.910918706560448e-06, "loss": 0.11276564598083497, "step": 37070 }, { "epoch": 0.7585187609632861, "grad_norm": 1.2721292972564697, "learning_rate": 2.906247421417333e-06, "loss": 0.10498238801956176, "step": 37080 }, { "epoch": 0.7587233237359299, "grad_norm": 1.558608055114746, "learning_rate": 2.9015792500640884e-06, "loss": 0.10093621015548707, "step": 37090 }, { "epoch": 0.7589278865085738, "grad_norm": 2.16060209274292, "learning_rate": 2.8969141945498038e-06, "loss": 0.10663008689880371, "step": 37100 }, { "epoch": 0.7591324492812176, "grad_norm": 3.46504282951355, "learning_rate": 2.892252256922188e-06, "loss": 0.10568307638168335, "step": 37110 }, { "epoch": 0.7593370120538614, "grad_norm": 1.9944051504135132, "learning_rate": 2.8875934392275975e-06, "loss": 0.1131861686706543, "step": 37120 }, { "epoch": 0.7595415748265052, "grad_norm": 1.6632497310638428, "learning_rate": 2.8829377435110105e-06, "loss": 0.10445597171783447, "step": 37130 }, { "epoch": 0.759746137599149, "grad_norm": 1.3884109258651733, "learning_rate": 2.8782851718160355e-06, "loss": 0.1086657166481018, "step": 37140 }, { "epoch": 0.7599507003717928, "grad_norm": 1.5380219221115112, "learning_rate": 2.873635726184907e-06, "loss": 0.10785382986068726, "step": 37150 }, { "epoch": 0.7601552631444367, "grad_norm": 1.5903549194335938, "learning_rate": 2.868989408658498e-06, "loss": 0.10916513204574585, "step": 37160 }, { "epoch": 0.7603598259170805, "grad_norm": 3.3049588203430176, "learning_rate": 2.8643462212762984e-06, "loss": 0.10751763582229615, "step": 37170 }, { "epoch": 0.7605643886897243, "grad_norm": 2.0356295108795166, "learning_rate": 2.859706166076428e-06, "loss": 0.10619181394577026, "step": 37180 }, { "epoch": 0.7607689514623681, "grad_norm": 1.9373596906661987, "learning_rate": 2.855069245095623e-06, "loss": 0.11134641170501709, "step": 37190 }, { "epoch": 0.7609735142350119, "grad_norm": 2.277549982070923, "learning_rate": 2.8504354603692595e-06, "loss": 0.1310736656188965, "step": 37200 }, { "epoch": 0.7611780770076557, "grad_norm": 2.1478097438812256, "learning_rate": 2.845804813931328e-06, "loss": 0.11542258262634278, "step": 37210 }, { "epoch": 0.7613826397802996, "grad_norm": 1.1459487676620483, "learning_rate": 2.841177307814439e-06, "loss": 0.1150560736656189, "step": 37220 }, { "epoch": 0.7615872025529434, "grad_norm": 2.8967058658599854, "learning_rate": 2.8365529440498285e-06, "loss": 0.11174634695053101, "step": 37230 }, { "epoch": 0.7617917653255872, "grad_norm": 2.0394392013549805, "learning_rate": 2.831931724667356e-06, "loss": 0.1106600284576416, "step": 37240 }, { "epoch": 0.761996328098231, "grad_norm": 1.900736689567566, "learning_rate": 2.8273136516954993e-06, "loss": 0.10632292032241822, "step": 37250 }, { "epoch": 0.7622008908708748, "grad_norm": 2.2949371337890625, "learning_rate": 2.822698727161344e-06, "loss": 0.12267322540283203, "step": 37260 }, { "epoch": 0.7624054536435186, "grad_norm": 1.7459162473678589, "learning_rate": 2.81808695309061e-06, "loss": 0.11745264530181884, "step": 37270 }, { "epoch": 0.7626100164161626, "grad_norm": 2.5070626735687256, "learning_rate": 2.8134783315076253e-06, "loss": 0.11220680475234986, "step": 37280 }, { "epoch": 0.7628145791888064, "grad_norm": 1.8811691999435425, "learning_rate": 2.8088728644353357e-06, "loss": 0.12201969623565674, "step": 37290 }, { "epoch": 0.7630191419614502, "grad_norm": 1.4142696857452393, "learning_rate": 2.804270553895301e-06, "loss": 0.1300602912902832, "step": 37300 }, { "epoch": 0.763223704734094, "grad_norm": 1.3549116849899292, "learning_rate": 2.799671401907702e-06, "loss": 0.11143121719360352, "step": 37310 }, { "epoch": 0.7634282675067378, "grad_norm": 2.6713504791259766, "learning_rate": 2.795075410491328e-06, "loss": 0.11925867795944214, "step": 37320 }, { "epoch": 0.7636328302793816, "grad_norm": 2.2766361236572266, "learning_rate": 2.7904825816635762e-06, "loss": 0.10647480487823487, "step": 37330 }, { "epoch": 0.7638373930520255, "grad_norm": 1.9133449792861938, "learning_rate": 2.785892917440459e-06, "loss": 0.11431999206542968, "step": 37340 }, { "epoch": 0.7640419558246693, "grad_norm": 3.731667995452881, "learning_rate": 2.781306419836608e-06, "loss": 0.1208958387374878, "step": 37350 }, { "epoch": 0.7642465185973131, "grad_norm": 2.181931257247925, "learning_rate": 2.7767230908652553e-06, "loss": 0.11180411577224732, "step": 37360 }, { "epoch": 0.7644510813699569, "grad_norm": 2.7029170989990234, "learning_rate": 2.772142932538244e-06, "loss": 0.11217080354690552, "step": 37370 }, { "epoch": 0.7646556441426007, "grad_norm": 2.2148377895355225, "learning_rate": 2.767565946866024e-06, "loss": 0.10856218338012695, "step": 37380 }, { "epoch": 0.7648602069152445, "grad_norm": 1.310896873474121, "learning_rate": 2.762992135857664e-06, "loss": 0.11391940116882324, "step": 37390 }, { "epoch": 0.7650647696878884, "grad_norm": 2.387132167816162, "learning_rate": 2.7584215015208203e-06, "loss": 0.11332616806030274, "step": 37400 }, { "epoch": 0.7652693324605322, "grad_norm": 1.7271270751953125, "learning_rate": 2.7538540458617693e-06, "loss": 0.11802979707717895, "step": 37410 }, { "epoch": 0.765473895233176, "grad_norm": 2.1518032550811768, "learning_rate": 2.7492897708853816e-06, "loss": 0.10841482877731323, "step": 37420 }, { "epoch": 0.7656784580058198, "grad_norm": 1.6183804273605347, "learning_rate": 2.7447286785951467e-06, "loss": 0.10916883945465088, "step": 37430 }, { "epoch": 0.7658830207784636, "grad_norm": 2.5959243774414062, "learning_rate": 2.7401707709931424e-06, "loss": 0.10818029642105102, "step": 37440 }, { "epoch": 0.7660875835511074, "grad_norm": 2.0860447883605957, "learning_rate": 2.735616050080051e-06, "loss": 0.11884876489639282, "step": 37450 }, { "epoch": 0.7662921463237513, "grad_norm": 2.2913401126861572, "learning_rate": 2.73106451785517e-06, "loss": 0.11351001262664795, "step": 37460 }, { "epoch": 0.7664967090963951, "grad_norm": 1.865866780281067, "learning_rate": 2.726516176316375e-06, "loss": 0.10556821823120117, "step": 37470 }, { "epoch": 0.7667012718690389, "grad_norm": 2.062530994415283, "learning_rate": 2.7219710274601574e-06, "loss": 0.11219527721405029, "step": 37480 }, { "epoch": 0.7669058346416827, "grad_norm": 2.068713903427124, "learning_rate": 2.717429073281597e-06, "loss": 0.10668483972549439, "step": 37490 }, { "epoch": 0.7671103974143265, "grad_norm": 3.8022313117980957, "learning_rate": 2.7128903157743834e-06, "loss": 0.09739171266555786, "step": 37500 }, { "epoch": 0.7673149601869704, "grad_norm": 1.483424186706543, "learning_rate": 2.7083547569307943e-06, "loss": 0.12156156301498414, "step": 37510 }, { "epoch": 0.7675195229596142, "grad_norm": 2.1341254711151123, "learning_rate": 2.7038223987417023e-06, "loss": 0.11233744621276856, "step": 37520 }, { "epoch": 0.767724085732258, "grad_norm": 1.4104015827178955, "learning_rate": 2.6992932431965767e-06, "loss": 0.12441830635070801, "step": 37530 }, { "epoch": 0.7679286485049018, "grad_norm": 1.587752342224121, "learning_rate": 2.6947672922834913e-06, "loss": 0.0981667160987854, "step": 37540 }, { "epoch": 0.7681332112775456, "grad_norm": 2.020847797393799, "learning_rate": 2.690244547989094e-06, "loss": 0.13049808740615845, "step": 37550 }, { "epoch": 0.7683377740501894, "grad_norm": 1.3852474689483643, "learning_rate": 2.685725012298639e-06, "loss": 0.11567125320434571, "step": 37560 }, { "epoch": 0.7685423368228333, "grad_norm": 2.0150415897369385, "learning_rate": 2.681208687195964e-06, "loss": 0.12437779903411865, "step": 37570 }, { "epoch": 0.7687468995954772, "grad_norm": 2.016599655151367, "learning_rate": 2.6766955746635105e-06, "loss": 0.10771752595901489, "step": 37580 }, { "epoch": 0.768951462368121, "grad_norm": 1.790790319442749, "learning_rate": 2.672185676682295e-06, "loss": 0.10119154453277587, "step": 37590 }, { "epoch": 0.7691560251407648, "grad_norm": 2.6433475017547607, "learning_rate": 2.6676789952319304e-06, "loss": 0.1117091178894043, "step": 37600 }, { "epoch": 0.7693605879134086, "grad_norm": 2.095615863800049, "learning_rate": 2.6631755322906174e-06, "loss": 0.11628191471099854, "step": 37610 }, { "epoch": 0.7695651506860524, "grad_norm": 3.364290952682495, "learning_rate": 2.6586752898351407e-06, "loss": 0.12718508243560792, "step": 37620 }, { "epoch": 0.7697697134586963, "grad_norm": 2.0871012210845947, "learning_rate": 2.6541782698408756e-06, "loss": 0.11775901317596435, "step": 37630 }, { "epoch": 0.7699742762313401, "grad_norm": 1.7001590728759766, "learning_rate": 2.6496844742817774e-06, "loss": 0.11497931480407715, "step": 37640 }, { "epoch": 0.7701788390039839, "grad_norm": 1.2571759223937988, "learning_rate": 2.6451939051303943e-06, "loss": 0.10923748016357422, "step": 37650 }, { "epoch": 0.7703834017766277, "grad_norm": 1.744255542755127, "learning_rate": 2.6407065643578524e-06, "loss": 0.10832127332687377, "step": 37660 }, { "epoch": 0.7705879645492715, "grad_norm": 2.4971625804901123, "learning_rate": 2.63622245393386e-06, "loss": 0.10108823776245117, "step": 37670 }, { "epoch": 0.7707925273219153, "grad_norm": 2.03993558883667, "learning_rate": 2.631741575826711e-06, "loss": 0.09677438735961914, "step": 37680 }, { "epoch": 0.7709970900945592, "grad_norm": 1.7370386123657227, "learning_rate": 2.6272639320032766e-06, "loss": 0.10561540126800537, "step": 37690 }, { "epoch": 0.771201652867203, "grad_norm": 3.0390145778656006, "learning_rate": 2.6227895244290135e-06, "loss": 0.11293554306030273, "step": 37700 }, { "epoch": 0.7714062156398468, "grad_norm": 2.261362314224243, "learning_rate": 2.6183183550679513e-06, "loss": 0.12150132656097412, "step": 37710 }, { "epoch": 0.7716107784124906, "grad_norm": 1.524483323097229, "learning_rate": 2.6138504258827025e-06, "loss": 0.12303328514099121, "step": 37720 }, { "epoch": 0.7718153411851344, "grad_norm": 1.8698630332946777, "learning_rate": 2.609385738834459e-06, "loss": 0.11493358612060547, "step": 37730 }, { "epoch": 0.7720199039577782, "grad_norm": 2.1830079555511475, "learning_rate": 2.6049242958829857e-06, "loss": 0.11586050987243653, "step": 37740 }, { "epoch": 0.7722244667304221, "grad_norm": 1.4621007442474365, "learning_rate": 2.600466098986625e-06, "loss": 0.10780487060546876, "step": 37750 }, { "epoch": 0.7724290295030659, "grad_norm": 3.1174988746643066, "learning_rate": 2.5960111501022943e-06, "loss": 0.1092875599861145, "step": 37760 }, { "epoch": 0.7726335922757097, "grad_norm": 1.9124690294265747, "learning_rate": 2.5915594511854857e-06, "loss": 0.10067262649536132, "step": 37770 }, { "epoch": 0.7728381550483535, "grad_norm": 1.5483888387680054, "learning_rate": 2.587111004190264e-06, "loss": 0.10127061605453491, "step": 37780 }, { "epoch": 0.7730427178209973, "grad_norm": 1.4569870233535767, "learning_rate": 2.582665811069266e-06, "loss": 0.11133799552917481, "step": 37790 }, { "epoch": 0.7732472805936411, "grad_norm": 1.6153029203414917, "learning_rate": 2.5782238737736996e-06, "loss": 0.10704324245452881, "step": 37800 }, { "epoch": 0.773451843366285, "grad_norm": 1.4150699377059937, "learning_rate": 2.573785194253351e-06, "loss": 0.1010897159576416, "step": 37810 }, { "epoch": 0.7736564061389288, "grad_norm": 1.9240509271621704, "learning_rate": 2.5693497744565665e-06, "loss": 0.10095585584640503, "step": 37820 }, { "epoch": 0.7738609689115726, "grad_norm": 4.1515092849731445, "learning_rate": 2.5649176163302668e-06, "loss": 0.11835229396820068, "step": 37830 }, { "epoch": 0.7740655316842164, "grad_norm": 2.21287202835083, "learning_rate": 2.5604887218199393e-06, "loss": 0.1196668028831482, "step": 37840 }, { "epoch": 0.7742700944568602, "grad_norm": 1.6699591875076294, "learning_rate": 2.5560630928696385e-06, "loss": 0.11956161260604858, "step": 37850 }, { "epoch": 0.774474657229504, "grad_norm": 1.755700707435608, "learning_rate": 2.5516407314219883e-06, "loss": 0.09349315166473389, "step": 37860 }, { "epoch": 0.774679220002148, "grad_norm": 2.458226203918457, "learning_rate": 2.54722163941817e-06, "loss": 0.11414179801940919, "step": 37870 }, { "epoch": 0.7748837827747918, "grad_norm": 1.5581953525543213, "learning_rate": 2.5428058187979454e-06, "loss": 0.10093090534210206, "step": 37880 }, { "epoch": 0.7750883455474356, "grad_norm": 1.9924731254577637, "learning_rate": 2.5383932714996253e-06, "loss": 0.12633649110794068, "step": 37890 }, { "epoch": 0.7752929083200794, "grad_norm": 3.0385220050811768, "learning_rate": 2.5339839994600912e-06, "loss": 0.10252686738967895, "step": 37900 }, { "epoch": 0.7754974710927232, "grad_norm": 2.75899338722229, "learning_rate": 2.5295780046147843e-06, "loss": 0.1079315185546875, "step": 37910 }, { "epoch": 0.775702033865367, "grad_norm": 1.7033958435058594, "learning_rate": 2.525175288897709e-06, "loss": 0.1064453363418579, "step": 37920 }, { "epoch": 0.7759065966380109, "grad_norm": 2.9731314182281494, "learning_rate": 2.520775854241427e-06, "loss": 0.10630736351013184, "step": 37930 }, { "epoch": 0.7761111594106547, "grad_norm": 1.7260313034057617, "learning_rate": 2.516379702577064e-06, "loss": 0.12034068107604981, "step": 37940 }, { "epoch": 0.7763157221832985, "grad_norm": 1.7160730361938477, "learning_rate": 2.5119868358343012e-06, "loss": 0.11474639177322388, "step": 37950 }, { "epoch": 0.7765202849559423, "grad_norm": 2.397632598876953, "learning_rate": 2.507597255941383e-06, "loss": 0.09924910068511963, "step": 37960 }, { "epoch": 0.7767248477285861, "grad_norm": 2.1823675632476807, "learning_rate": 2.5032109648251045e-06, "loss": 0.10761584043502807, "step": 37970 }, { "epoch": 0.77692941050123, "grad_norm": 3.9554364681243896, "learning_rate": 2.4988279644108227e-06, "loss": 0.1065606951713562, "step": 37980 }, { "epoch": 0.7771339732738738, "grad_norm": 2.9010472297668457, "learning_rate": 2.494448256622446e-06, "loss": 0.12465107440948486, "step": 37990 }, { "epoch": 0.7773385360465176, "grad_norm": 1.7796984910964966, "learning_rate": 2.490071843382439e-06, "loss": 0.10573267936706543, "step": 38000 }, { "epoch": 0.7775430988191614, "grad_norm": 3.2137720584869385, "learning_rate": 2.4856987266118217e-06, "loss": 0.11005188226699829, "step": 38010 }, { "epoch": 0.7777476615918052, "grad_norm": 1.9194767475128174, "learning_rate": 2.4813289082301606e-06, "loss": 0.11314823627471923, "step": 38020 }, { "epoch": 0.777952224364449, "grad_norm": 1.3511934280395508, "learning_rate": 2.476962390155587e-06, "loss": 0.10586860179901122, "step": 38030 }, { "epoch": 0.7781567871370929, "grad_norm": 1.5644294023513794, "learning_rate": 2.472599174304774e-06, "loss": 0.1019355058670044, "step": 38040 }, { "epoch": 0.7783613499097367, "grad_norm": 1.8834331035614014, "learning_rate": 2.468239262592946e-06, "loss": 0.11335376501083375, "step": 38050 }, { "epoch": 0.7785659126823805, "grad_norm": 2.179847002029419, "learning_rate": 2.4638826569338792e-06, "loss": 0.11159403324127197, "step": 38060 }, { "epoch": 0.7787704754550243, "grad_norm": 1.6992837190628052, "learning_rate": 2.4595293592398973e-06, "loss": 0.10976185798645019, "step": 38070 }, { "epoch": 0.7789750382276681, "grad_norm": 3.0669493675231934, "learning_rate": 2.455179371421872e-06, "loss": 0.10432316064834594, "step": 38080 }, { "epoch": 0.7791796010003119, "grad_norm": 1.6060482263565063, "learning_rate": 2.4508326953892227e-06, "loss": 0.11336609125137329, "step": 38090 }, { "epoch": 0.7793841637729558, "grad_norm": 2.6975185871124268, "learning_rate": 2.446489333049913e-06, "loss": 0.11774784326553345, "step": 38100 }, { "epoch": 0.7795887265455996, "grad_norm": 1.9109752178192139, "learning_rate": 2.442149286310459e-06, "loss": 0.10709831714630128, "step": 38110 }, { "epoch": 0.7797932893182434, "grad_norm": 3.3764021396636963, "learning_rate": 2.4378125570759135e-06, "loss": 0.10573340654373169, "step": 38120 }, { "epoch": 0.7799978520908872, "grad_norm": 2.5753397941589355, "learning_rate": 2.4334791472498763e-06, "loss": 0.12223125696182251, "step": 38130 }, { "epoch": 0.780202414863531, "grad_norm": 3.1776340007781982, "learning_rate": 2.429149058734489e-06, "loss": 0.12106530666351319, "step": 38140 }, { "epoch": 0.7804069776361748, "grad_norm": 1.63589346408844, "learning_rate": 2.424822293430438e-06, "loss": 0.11285240650177002, "step": 38150 }, { "epoch": 0.7806115404088187, "grad_norm": 1.7597217559814453, "learning_rate": 2.420498853236947e-06, "loss": 0.12631697654724122, "step": 38160 }, { "epoch": 0.7808161031814626, "grad_norm": 2.1608426570892334, "learning_rate": 2.41617874005178e-06, "loss": 0.10823792219161987, "step": 38170 }, { "epoch": 0.7810206659541064, "grad_norm": 1.7593015432357788, "learning_rate": 2.4118619557712496e-06, "loss": 0.10308661460876464, "step": 38180 }, { "epoch": 0.7812252287267502, "grad_norm": 1.989099144935608, "learning_rate": 2.407548502290196e-06, "loss": 0.10825893878936768, "step": 38190 }, { "epoch": 0.781429791499394, "grad_norm": 3.2497451305389404, "learning_rate": 2.4032383815020033e-06, "loss": 0.11774786710739135, "step": 38200 }, { "epoch": 0.7816343542720378, "grad_norm": 1.510526180267334, "learning_rate": 2.3989315952985905e-06, "loss": 0.1220935344696045, "step": 38210 }, { "epoch": 0.7818389170446817, "grad_norm": 2.05145263671875, "learning_rate": 2.3946281455704134e-06, "loss": 0.11289916038513184, "step": 38220 }, { "epoch": 0.7820434798173255, "grad_norm": 1.8972818851470947, "learning_rate": 2.390328034206464e-06, "loss": 0.10926778316497802, "step": 38230 }, { "epoch": 0.7822480425899693, "grad_norm": 3.162391424179077, "learning_rate": 2.3860312630942685e-06, "loss": 0.11822477579116822, "step": 38240 }, { "epoch": 0.7824526053626131, "grad_norm": 2.9875998497009277, "learning_rate": 2.3817378341198825e-06, "loss": 0.10414283275604248, "step": 38250 }, { "epoch": 0.7826571681352569, "grad_norm": 1.976986050605774, "learning_rate": 2.377447749167907e-06, "loss": 0.10298303365707398, "step": 38260 }, { "epoch": 0.7828617309079007, "grad_norm": 1.4238414764404297, "learning_rate": 2.373161010121463e-06, "loss": 0.10241665840148925, "step": 38270 }, { "epoch": 0.7830662936805446, "grad_norm": 1.7034944295883179, "learning_rate": 2.3688776188622086e-06, "loss": 0.11694638729095459, "step": 38280 }, { "epoch": 0.7832708564531884, "grad_norm": 3.1723480224609375, "learning_rate": 2.3645975772703236e-06, "loss": 0.11384539604187012, "step": 38290 }, { "epoch": 0.7834754192258322, "grad_norm": 2.2858777046203613, "learning_rate": 2.360320887224533e-06, "loss": 0.10639140605926514, "step": 38300 }, { "epoch": 0.783679981998476, "grad_norm": 1.4777079820632935, "learning_rate": 2.3560475506020785e-06, "loss": 0.11007845401763916, "step": 38310 }, { "epoch": 0.7838845447711198, "grad_norm": 2.5865180492401123, "learning_rate": 2.3517775692787335e-06, "loss": 0.11620573997497559, "step": 38320 }, { "epoch": 0.7840891075437636, "grad_norm": 1.4917521476745605, "learning_rate": 2.347510945128797e-06, "loss": 0.10781742334365844, "step": 38330 }, { "epoch": 0.7842936703164075, "grad_norm": 1.770315408706665, "learning_rate": 2.3432476800251e-06, "loss": 0.10997622013092041, "step": 38340 }, { "epoch": 0.7844982330890513, "grad_norm": 1.8332401514053345, "learning_rate": 2.3389877758389935e-06, "loss": 0.11304588317871093, "step": 38350 }, { "epoch": 0.7847027958616951, "grad_norm": 2.048692464828491, "learning_rate": 2.334731234440354e-06, "loss": 0.10177284479141235, "step": 38360 }, { "epoch": 0.7849073586343389, "grad_norm": 2.21632981300354, "learning_rate": 2.3304780576975828e-06, "loss": 0.11739214658737182, "step": 38370 }, { "epoch": 0.7851119214069827, "grad_norm": 2.135451316833496, "learning_rate": 2.3262282474776045e-06, "loss": 0.10686101913452148, "step": 38380 }, { "epoch": 0.7853164841796265, "grad_norm": 3.1245977878570557, "learning_rate": 2.3219818056458642e-06, "loss": 0.09978626370429992, "step": 38390 }, { "epoch": 0.7855210469522704, "grad_norm": 2.915512800216675, "learning_rate": 2.3177387340663294e-06, "loss": 0.11450613737106323, "step": 38400 }, { "epoch": 0.7857256097249142, "grad_norm": 2.1072492599487305, "learning_rate": 2.3134990346014917e-06, "loss": 0.11521704196929931, "step": 38410 }, { "epoch": 0.785930172497558, "grad_norm": 2.399761438369751, "learning_rate": 2.30926270911236e-06, "loss": 0.11541416645050048, "step": 38420 }, { "epoch": 0.7861347352702018, "grad_norm": 2.6730098724365234, "learning_rate": 2.3050297594584613e-06, "loss": 0.1097151517868042, "step": 38430 }, { "epoch": 0.7863392980428456, "grad_norm": 2.34251070022583, "learning_rate": 2.3008001874978347e-06, "loss": 0.10818893909454345, "step": 38440 }, { "epoch": 0.7865438608154895, "grad_norm": 1.6707743406295776, "learning_rate": 2.2965739950870504e-06, "loss": 0.10056992769241332, "step": 38450 }, { "epoch": 0.7867484235881333, "grad_norm": 2.2207858562469482, "learning_rate": 2.2923511840811862e-06, "loss": 0.11035416126251221, "step": 38460 }, { "epoch": 0.7869529863607772, "grad_norm": 2.285762310028076, "learning_rate": 2.2881317563338358e-06, "loss": 0.10599910020828247, "step": 38470 }, { "epoch": 0.787157549133421, "grad_norm": 1.5632643699645996, "learning_rate": 2.283915713697109e-06, "loss": 0.10898125171661377, "step": 38480 }, { "epoch": 0.7873621119060648, "grad_norm": 3.112654447555542, "learning_rate": 2.279703058021634e-06, "loss": 0.12809178829193116, "step": 38490 }, { "epoch": 0.7875666746787086, "grad_norm": 1.4679243564605713, "learning_rate": 2.275493791156548e-06, "loss": 0.10983977317810059, "step": 38500 }, { "epoch": 0.7877712374513525, "grad_norm": 2.457707166671753, "learning_rate": 2.271287914949496e-06, "loss": 0.10932893753051758, "step": 38510 }, { "epoch": 0.7879758002239963, "grad_norm": 2.5351321697235107, "learning_rate": 2.267085431246641e-06, "loss": 0.10163252353668213, "step": 38520 }, { "epoch": 0.7881803629966401, "grad_norm": 1.8343595266342163, "learning_rate": 2.2628863418926595e-06, "loss": 0.10310530662536621, "step": 38530 }, { "epoch": 0.7883849257692839, "grad_norm": 1.933158278465271, "learning_rate": 2.2586906487307324e-06, "loss": 0.11536186933517456, "step": 38540 }, { "epoch": 0.7885894885419277, "grad_norm": 2.1972262859344482, "learning_rate": 2.2544983536025488e-06, "loss": 0.1060139536857605, "step": 38550 }, { "epoch": 0.7887940513145715, "grad_norm": 2.0978009700775146, "learning_rate": 2.250309458348313e-06, "loss": 0.10516898632049561, "step": 38560 }, { "epoch": 0.7889986140872154, "grad_norm": 2.2229068279266357, "learning_rate": 2.246123964806736e-06, "loss": 0.11170964241027832, "step": 38570 }, { "epoch": 0.7892031768598592, "grad_norm": 1.4290353059768677, "learning_rate": 2.2419418748150246e-06, "loss": 0.12084226608276367, "step": 38580 }, { "epoch": 0.789407739632503, "grad_norm": 1.5221469402313232, "learning_rate": 2.2377631902089004e-06, "loss": 0.09919031858444213, "step": 38590 }, { "epoch": 0.7896123024051468, "grad_norm": 2.7323250770568848, "learning_rate": 2.2335879128225946e-06, "loss": 0.11178992986679077, "step": 38600 }, { "epoch": 0.7898168651777906, "grad_norm": 2.020939350128174, "learning_rate": 2.229416044488836e-06, "loss": 0.1066232442855835, "step": 38610 }, { "epoch": 0.7900214279504344, "grad_norm": 1.90806245803833, "learning_rate": 2.225247587038858e-06, "loss": 0.1006395697593689, "step": 38620 }, { "epoch": 0.7902259907230783, "grad_norm": 1.644002914428711, "learning_rate": 2.221082542302394e-06, "loss": 0.11108253002166749, "step": 38630 }, { "epoch": 0.7904305534957221, "grad_norm": 1.6938693523406982, "learning_rate": 2.216920912107693e-06, "loss": 0.10886939764022827, "step": 38640 }, { "epoch": 0.7906351162683659, "grad_norm": 1.6322404146194458, "learning_rate": 2.2127626982814854e-06, "loss": 0.13081464767456055, "step": 38650 }, { "epoch": 0.7908396790410097, "grad_norm": 2.83921217918396, "learning_rate": 2.2086079026490136e-06, "loss": 0.11487922668457032, "step": 38660 }, { "epoch": 0.7910442418136535, "grad_norm": 1.2258260250091553, "learning_rate": 2.2044565270340166e-06, "loss": 0.10270067453384399, "step": 38670 }, { "epoch": 0.7912488045862973, "grad_norm": 1.7727537155151367, "learning_rate": 2.200308573258738e-06, "loss": 0.1118201732635498, "step": 38680 }, { "epoch": 0.7914533673589412, "grad_norm": 1.9163087606430054, "learning_rate": 2.1961640431439113e-06, "loss": 0.10187307596206666, "step": 38690 }, { "epoch": 0.791657930131585, "grad_norm": 1.7912627458572388, "learning_rate": 2.192022938508772e-06, "loss": 0.11396812200546265, "step": 38700 }, { "epoch": 0.7918624929042288, "grad_norm": 1.2926572561264038, "learning_rate": 2.1878852611710465e-06, "loss": 0.10681917667388915, "step": 38710 }, { "epoch": 0.7920670556768726, "grad_norm": 1.808458924293518, "learning_rate": 2.1837510129469684e-06, "loss": 0.09578131437301636, "step": 38720 }, { "epoch": 0.7922716184495164, "grad_norm": 2.0680391788482666, "learning_rate": 2.179620195651253e-06, "loss": 0.11249276399612426, "step": 38730 }, { "epoch": 0.7924761812221602, "grad_norm": 1.2376376390457153, "learning_rate": 2.175492811097112e-06, "loss": 0.11884726285934448, "step": 38740 }, { "epoch": 0.7926807439948041, "grad_norm": 1.4903088808059692, "learning_rate": 2.1713688610962603e-06, "loss": 0.0991642951965332, "step": 38750 }, { "epoch": 0.792885306767448, "grad_norm": 1.5967028141021729, "learning_rate": 2.1672483474588957e-06, "loss": 0.10649325847625732, "step": 38760 }, { "epoch": 0.7930898695400918, "grad_norm": 2.2719388008117676, "learning_rate": 2.163131271993709e-06, "loss": 0.11546827554702759, "step": 38770 }, { "epoch": 0.7932944323127356, "grad_norm": 2.4517874717712402, "learning_rate": 2.159017636507883e-06, "loss": 0.10837267637252808, "step": 38780 }, { "epoch": 0.7934989950853794, "grad_norm": 1.8881374597549438, "learning_rate": 2.1549074428070958e-06, "loss": 0.10603690147399902, "step": 38790 }, { "epoch": 0.7937035578580232, "grad_norm": 2.3587677478790283, "learning_rate": 2.1508006926955028e-06, "loss": 0.10370814800262451, "step": 38800 }, { "epoch": 0.7939081206306671, "grad_norm": 1.9987338781356812, "learning_rate": 2.146697387975757e-06, "loss": 0.10984375476837158, "step": 38810 }, { "epoch": 0.7941126834033109, "grad_norm": 3.2694201469421387, "learning_rate": 2.1425975304489933e-06, "loss": 0.10816649198532105, "step": 38820 }, { "epoch": 0.7943172461759547, "grad_norm": 2.9725842475891113, "learning_rate": 2.138501121914843e-06, "loss": 0.1167292594909668, "step": 38830 }, { "epoch": 0.7945218089485985, "grad_norm": 1.8117685317993164, "learning_rate": 2.1344081641714143e-06, "loss": 0.11983928680419922, "step": 38840 }, { "epoch": 0.7947263717212423, "grad_norm": 2.3097519874572754, "learning_rate": 2.1303186590153035e-06, "loss": 0.10000686645507813, "step": 38850 }, { "epoch": 0.7949309344938861, "grad_norm": 1.607010841369629, "learning_rate": 2.1262326082415918e-06, "loss": 0.10750739574432373, "step": 38860 }, { "epoch": 0.79513549726653, "grad_norm": 3.1367664337158203, "learning_rate": 2.1221500136438433e-06, "loss": 0.10893675088882446, "step": 38870 }, { "epoch": 0.7953400600391738, "grad_norm": 1.7939095497131348, "learning_rate": 2.1180708770141054e-06, "loss": 0.1121891975402832, "step": 38880 }, { "epoch": 0.7955446228118176, "grad_norm": 1.6029633283615112, "learning_rate": 2.113995200142909e-06, "loss": 0.09836262464523315, "step": 38890 }, { "epoch": 0.7957491855844614, "grad_norm": 1.454397201538086, "learning_rate": 2.1099229848192603e-06, "loss": 0.10917530059814454, "step": 38900 }, { "epoch": 0.7959537483571052, "grad_norm": 2.0924172401428223, "learning_rate": 2.1058542328306586e-06, "loss": 0.11135506629943848, "step": 38910 }, { "epoch": 0.7961583111297491, "grad_norm": 1.7150490283966064, "learning_rate": 2.10178894596307e-06, "loss": 0.10614275932312012, "step": 38920 }, { "epoch": 0.7963628739023929, "grad_norm": 2.4876224994659424, "learning_rate": 2.0977271260009468e-06, "loss": 0.10031207799911498, "step": 38930 }, { "epoch": 0.7965674366750367, "grad_norm": 1.4339423179626465, "learning_rate": 2.0936687747272176e-06, "loss": 0.09336166977882385, "step": 38940 }, { "epoch": 0.7967719994476805, "grad_norm": 2.068338394165039, "learning_rate": 2.089613893923287e-06, "loss": 0.10977132320404052, "step": 38950 }, { "epoch": 0.7969765622203243, "grad_norm": 1.712157130241394, "learning_rate": 2.085562485369038e-06, "loss": 0.11769801378250122, "step": 38960 }, { "epoch": 0.7971811249929681, "grad_norm": 1.906382441520691, "learning_rate": 2.081514550842827e-06, "loss": 0.11915931701660157, "step": 38970 }, { "epoch": 0.797385687765612, "grad_norm": 3.7214043140411377, "learning_rate": 2.077470092121493e-06, "loss": 0.10700531005859375, "step": 38980 }, { "epoch": 0.7975902505382558, "grad_norm": 2.9033291339874268, "learning_rate": 2.0734291109803405e-06, "loss": 0.09871301054954529, "step": 38990 }, { "epoch": 0.7977948133108996, "grad_norm": 2.1287589073181152, "learning_rate": 2.069391609193151e-06, "loss": 0.11089701652526855, "step": 39000 }, { "epoch": 0.7979993760835434, "grad_norm": 2.0892417430877686, "learning_rate": 2.0653575885321786e-06, "loss": 0.11248335838317872, "step": 39010 }, { "epoch": 0.7982039388561872, "grad_norm": 3.3696165084838867, "learning_rate": 2.0613270507681493e-06, "loss": 0.10351876020431519, "step": 39020 }, { "epoch": 0.798408501628831, "grad_norm": 1.93574857711792, "learning_rate": 2.05729999767026e-06, "loss": 0.09991247653961181, "step": 39030 }, { "epoch": 0.798613064401475, "grad_norm": 1.7372055053710938, "learning_rate": 2.0532764310061804e-06, "loss": 0.10628306865692139, "step": 39040 }, { "epoch": 0.7988176271741187, "grad_norm": 2.3167362213134766, "learning_rate": 2.0492563525420427e-06, "loss": 0.10862182378768921, "step": 39050 }, { "epoch": 0.7990221899467626, "grad_norm": 2.19582462310791, "learning_rate": 2.0452397640424594e-06, "loss": 0.10960246324539184, "step": 39060 }, { "epoch": 0.7992267527194064, "grad_norm": 3.470255136489868, "learning_rate": 2.041226667270503e-06, "loss": 0.1230545997619629, "step": 39070 }, { "epoch": 0.7994313154920502, "grad_norm": 4.27147102355957, "learning_rate": 2.037217063987713e-06, "loss": 0.11796410083770752, "step": 39080 }, { "epoch": 0.799635878264694, "grad_norm": 1.9705582857131958, "learning_rate": 2.033210955954099e-06, "loss": 0.1005859136581421, "step": 39090 }, { "epoch": 0.7998404410373379, "grad_norm": 1.8417928218841553, "learning_rate": 2.029208344928134e-06, "loss": 0.10892616510391236, "step": 39100 }, { "epoch": 0.8000450038099817, "grad_norm": 2.2170209884643555, "learning_rate": 2.0252092326667582e-06, "loss": 0.11386938095092773, "step": 39110 }, { "epoch": 0.8002495665826255, "grad_norm": 2.2890501022338867, "learning_rate": 2.021213620925371e-06, "loss": 0.10240598917007446, "step": 39120 }, { "epoch": 0.8004541293552693, "grad_norm": 4.2798919677734375, "learning_rate": 2.0172215114578443e-06, "loss": 0.10601060390472412, "step": 39130 }, { "epoch": 0.8006586921279131, "grad_norm": 1.828971028327942, "learning_rate": 2.0132329060165055e-06, "loss": 0.10247856378555298, "step": 39140 }, { "epoch": 0.8008632549005569, "grad_norm": 1.9240636825561523, "learning_rate": 2.009247806352145e-06, "loss": 0.11477237939834595, "step": 39150 }, { "epoch": 0.8010678176732008, "grad_norm": 2.1824631690979004, "learning_rate": 2.0052662142140145e-06, "loss": 0.1056670069694519, "step": 39160 }, { "epoch": 0.8012723804458446, "grad_norm": 1.6657848358154297, "learning_rate": 2.0012881313498276e-06, "loss": 0.1072617530822754, "step": 39170 }, { "epoch": 0.8014769432184884, "grad_norm": 2.01271915435791, "learning_rate": 1.997313559505757e-06, "loss": 0.10831387042999267, "step": 39180 }, { "epoch": 0.8016815059911322, "grad_norm": 2.7166712284088135, "learning_rate": 1.993342500426433e-06, "loss": 0.1198662281036377, "step": 39190 }, { "epoch": 0.801886068763776, "grad_norm": 3.110973834991455, "learning_rate": 1.9893749558549424e-06, "loss": 0.1098186731338501, "step": 39200 }, { "epoch": 0.8020906315364198, "grad_norm": 1.4152488708496094, "learning_rate": 1.985410927532837e-06, "loss": 0.12448842525482177, "step": 39210 }, { "epoch": 0.8022951943090637, "grad_norm": 1.8436695337295532, "learning_rate": 1.981450417200117e-06, "loss": 0.10697729587554931, "step": 39220 }, { "epoch": 0.8024997570817075, "grad_norm": 3.6777665615081787, "learning_rate": 1.9774934265952416e-06, "loss": 0.1046146273612976, "step": 39230 }, { "epoch": 0.8027043198543513, "grad_norm": 2.5695626735687256, "learning_rate": 1.9735399574551238e-06, "loss": 0.09344412088394165, "step": 39240 }, { "epoch": 0.8029088826269951, "grad_norm": 1.6791857481002808, "learning_rate": 1.9695900115151333e-06, "loss": 0.12690966129302977, "step": 39250 }, { "epoch": 0.8031134453996389, "grad_norm": 2.197429895401001, "learning_rate": 1.9656435905090907e-06, "loss": 0.10056931972503662, "step": 39260 }, { "epoch": 0.8033180081722827, "grad_norm": 1.9683492183685303, "learning_rate": 1.9617006961692674e-06, "loss": 0.11320292949676514, "step": 39270 }, { "epoch": 0.8035225709449266, "grad_norm": 1.985602855682373, "learning_rate": 1.9577613302263965e-06, "loss": 0.11708188056945801, "step": 39280 }, { "epoch": 0.8037271337175704, "grad_norm": 2.0172276496887207, "learning_rate": 1.953825494409651e-06, "loss": 0.09499784708023071, "step": 39290 }, { "epoch": 0.8039316964902142, "grad_norm": 2.6515052318573, "learning_rate": 1.949893190446661e-06, "loss": 0.10426812171936035, "step": 39300 }, { "epoch": 0.804136259262858, "grad_norm": 1.6932506561279297, "learning_rate": 1.945964420063502e-06, "loss": 0.11200275421142578, "step": 39310 }, { "epoch": 0.8043408220355018, "grad_norm": 1.9518014192581177, "learning_rate": 1.942039184984702e-06, "loss": 0.11628603935241699, "step": 39320 }, { "epoch": 0.8045453848081456, "grad_norm": 2.0056071281433105, "learning_rate": 1.9381174869332366e-06, "loss": 0.10347049236297608, "step": 39330 }, { "epoch": 0.8047499475807895, "grad_norm": 3.0582847595214844, "learning_rate": 1.934199327630526e-06, "loss": 0.10284000635147095, "step": 39340 }, { "epoch": 0.8049545103534334, "grad_norm": 2.4364757537841797, "learning_rate": 1.9302847087964384e-06, "loss": 0.09600051641464233, "step": 39350 }, { "epoch": 0.8051590731260772, "grad_norm": 2.1473348140716553, "learning_rate": 1.926373632149292e-06, "loss": 0.11525084972381591, "step": 39360 }, { "epoch": 0.805363635898721, "grad_norm": 1.424973726272583, "learning_rate": 1.9224660994058454e-06, "loss": 0.10028610229492188, "step": 39370 }, { "epoch": 0.8055681986713648, "grad_norm": 3.693542003631592, "learning_rate": 1.918562112281304e-06, "loss": 0.11432865858078003, "step": 39380 }, { "epoch": 0.8057727614440087, "grad_norm": 2.8650102615356445, "learning_rate": 1.914661672489314e-06, "loss": 0.10694766044616699, "step": 39390 }, { "epoch": 0.8059773242166525, "grad_norm": 2.005457878112793, "learning_rate": 1.9107647817419674e-06, "loss": 0.11301925182342529, "step": 39400 }, { "epoch": 0.8061818869892963, "grad_norm": 1.6251450777053833, "learning_rate": 1.906871441749797e-06, "loss": 0.1026957631111145, "step": 39410 }, { "epoch": 0.8063864497619401, "grad_norm": 2.133244037628174, "learning_rate": 1.9029816542217771e-06, "loss": 0.10011479854583741, "step": 39420 }, { "epoch": 0.8065910125345839, "grad_norm": 1.539507508277893, "learning_rate": 1.8990954208653224e-06, "loss": 0.11881160736083984, "step": 39430 }, { "epoch": 0.8067955753072277, "grad_norm": 2.185555934906006, "learning_rate": 1.8952127433862898e-06, "loss": 0.10739722251892089, "step": 39440 }, { "epoch": 0.8070001380798716, "grad_norm": 2.7975001335144043, "learning_rate": 1.8913336234889734e-06, "loss": 0.09902814626693726, "step": 39450 }, { "epoch": 0.8072047008525154, "grad_norm": 1.6028571128845215, "learning_rate": 1.8874580628761053e-06, "loss": 0.10374658107757569, "step": 39460 }, { "epoch": 0.8074092636251592, "grad_norm": 2.1347270011901855, "learning_rate": 1.8835860632488545e-06, "loss": 0.11456267833709717, "step": 39470 }, { "epoch": 0.807613826397803, "grad_norm": 1.8647754192352295, "learning_rate": 1.8797176263068307e-06, "loss": 0.10766434669494629, "step": 39480 }, { "epoch": 0.8078183891704468, "grad_norm": 2.1175143718719482, "learning_rate": 1.8758527537480753e-06, "loss": 0.10789623260498046, "step": 39490 }, { "epoch": 0.8080229519430906, "grad_norm": 2.2685365676879883, "learning_rate": 1.871991447269066e-06, "loss": 0.11738160848617554, "step": 39500 }, { "epoch": 0.8082275147157345, "grad_norm": 2.3332161903381348, "learning_rate": 1.86813370856472e-06, "loss": 0.10193769931793213, "step": 39510 }, { "epoch": 0.8084320774883783, "grad_norm": 1.5174484252929688, "learning_rate": 1.8642795393283831e-06, "loss": 0.10452057123184204, "step": 39520 }, { "epoch": 0.8086366402610221, "grad_norm": 2.740415334701538, "learning_rate": 1.860428941251836e-06, "loss": 0.0970265507698059, "step": 39530 }, { "epoch": 0.8088412030336659, "grad_norm": 3.4629580974578857, "learning_rate": 1.8565819160252907e-06, "loss": 0.11924357414245605, "step": 39540 }, { "epoch": 0.8090457658063097, "grad_norm": 2.227118968963623, "learning_rate": 1.8527384653373937e-06, "loss": 0.11005699634552002, "step": 39550 }, { "epoch": 0.8092503285789535, "grad_norm": 2.0493602752685547, "learning_rate": 1.8488985908752188e-06, "loss": 0.1036527395248413, "step": 39560 }, { "epoch": 0.8094548913515974, "grad_norm": 1.875789999961853, "learning_rate": 1.8450622943242736e-06, "loss": 0.09792742133140564, "step": 39570 }, { "epoch": 0.8096594541242412, "grad_norm": 4.663150787353516, "learning_rate": 1.8412295773684908e-06, "loss": 0.11122641563415528, "step": 39580 }, { "epoch": 0.809864016896885, "grad_norm": 1.9640196561813354, "learning_rate": 1.837400441690238e-06, "loss": 0.1160825252532959, "step": 39590 }, { "epoch": 0.8100685796695288, "grad_norm": 1.3674118518829346, "learning_rate": 1.8335748889703064e-06, "loss": 0.09185457229614258, "step": 39600 }, { "epoch": 0.8102731424421726, "grad_norm": 2.2079946994781494, "learning_rate": 1.8297529208879183e-06, "loss": 0.10579426288604736, "step": 39610 }, { "epoch": 0.8104777052148164, "grad_norm": 2.091188907623291, "learning_rate": 1.8259345391207107e-06, "loss": 0.10228838920593261, "step": 39620 }, { "epoch": 0.8106822679874603, "grad_norm": 2.6659607887268066, "learning_rate": 1.8221197453447636e-06, "loss": 0.11926888227462769, "step": 39630 }, { "epoch": 0.8108868307601041, "grad_norm": 1.698639154434204, "learning_rate": 1.8183085412345714e-06, "loss": 0.1035414457321167, "step": 39640 }, { "epoch": 0.811091393532748, "grad_norm": 1.8345891237258911, "learning_rate": 1.814500928463052e-06, "loss": 0.10898096561431884, "step": 39650 }, { "epoch": 0.8112959563053918, "grad_norm": 2.789746046066284, "learning_rate": 1.8106969087015568e-06, "loss": 0.11488142013549804, "step": 39660 }, { "epoch": 0.8115005190780356, "grad_norm": 2.200723171234131, "learning_rate": 1.8068964836198499e-06, "loss": 0.10423245429992675, "step": 39670 }, { "epoch": 0.8117050818506794, "grad_norm": 3.508835792541504, "learning_rate": 1.8030996548861235e-06, "loss": 0.10643768310546875, "step": 39680 }, { "epoch": 0.8119096446233233, "grad_norm": 2.5113565921783447, "learning_rate": 1.799306424166981e-06, "loss": 0.10242135524749756, "step": 39690 }, { "epoch": 0.8121142073959671, "grad_norm": 3.363468647003174, "learning_rate": 1.795516793127462e-06, "loss": 0.09986741542816162, "step": 39700 }, { "epoch": 0.8123187701686109, "grad_norm": 2.6473710536956787, "learning_rate": 1.7917307634310155e-06, "loss": 0.09757233262062073, "step": 39710 }, { "epoch": 0.8125233329412547, "grad_norm": 2.063588857650757, "learning_rate": 1.7879483367395134e-06, "loss": 0.10789461135864258, "step": 39720 }, { "epoch": 0.8127278957138985, "grad_norm": 2.30661678314209, "learning_rate": 1.7841695147132409e-06, "loss": 0.10620677471160889, "step": 39730 }, { "epoch": 0.8129324584865423, "grad_norm": 2.1276373863220215, "learning_rate": 1.7803942990109103e-06, "loss": 0.11834533214569092, "step": 39740 }, { "epoch": 0.8131370212591862, "grad_norm": 2.595062732696533, "learning_rate": 1.776622691289648e-06, "loss": 0.11811152696609498, "step": 39750 }, { "epoch": 0.81334158403183, "grad_norm": 1.5137872695922852, "learning_rate": 1.772854693204986e-06, "loss": 0.09190692901611328, "step": 39760 }, { "epoch": 0.8135461468044738, "grad_norm": 1.7170382738113403, "learning_rate": 1.7690903064108823e-06, "loss": 0.09875326156616211, "step": 39770 }, { "epoch": 0.8137507095771176, "grad_norm": 3.2160778045654297, "learning_rate": 1.765329532559713e-06, "loss": 0.10164968967437744, "step": 39780 }, { "epoch": 0.8139552723497614, "grad_norm": 2.3867781162261963, "learning_rate": 1.7615723733022593e-06, "loss": 0.10492781400680543, "step": 39790 }, { "epoch": 0.8141598351224052, "grad_norm": 1.9092929363250732, "learning_rate": 1.7578188302877198e-06, "loss": 0.10727154016494751, "step": 39800 }, { "epoch": 0.8143643978950491, "grad_norm": 3.9493696689605713, "learning_rate": 1.754068905163704e-06, "loss": 0.10912730693817138, "step": 39810 }, { "epoch": 0.8145689606676929, "grad_norm": 2.06720232963562, "learning_rate": 1.7503225995762407e-06, "loss": 0.11312119960784912, "step": 39820 }, { "epoch": 0.8147735234403367, "grad_norm": 2.497100830078125, "learning_rate": 1.746579915169757e-06, "loss": 0.10337047576904297, "step": 39830 }, { "epoch": 0.8149780862129805, "grad_norm": 2.096078634262085, "learning_rate": 1.7428408535870967e-06, "loss": 0.10896420478820801, "step": 39840 }, { "epoch": 0.8151826489856243, "grad_norm": 1.9920443296432495, "learning_rate": 1.739105416469521e-06, "loss": 0.09804841876029968, "step": 39850 }, { "epoch": 0.8153872117582682, "grad_norm": 1.8152753114700317, "learning_rate": 1.7353736054566905e-06, "loss": 0.10022211074829102, "step": 39860 }, { "epoch": 0.815591774530912, "grad_norm": 1.8105381727218628, "learning_rate": 1.7316454221866751e-06, "loss": 0.11109751462936401, "step": 39870 }, { "epoch": 0.8157963373035558, "grad_norm": 3.1258842945098877, "learning_rate": 1.7279208682959526e-06, "loss": 0.11312873363494873, "step": 39880 }, { "epoch": 0.8160009000761996, "grad_norm": 3.579395294189453, "learning_rate": 1.7241999454194148e-06, "loss": 0.10901604890823365, "step": 39890 }, { "epoch": 0.8162054628488434, "grad_norm": 2.5867085456848145, "learning_rate": 1.720482655190353e-06, "loss": 0.0974218487739563, "step": 39900 }, { "epoch": 0.8164100256214872, "grad_norm": 1.305061936378479, "learning_rate": 1.71676899924046e-06, "loss": 0.10083293914794922, "step": 39910 }, { "epoch": 0.8166145883941311, "grad_norm": 2.1464977264404297, "learning_rate": 1.7130589791998398e-06, "loss": 0.12119243144989014, "step": 39920 }, { "epoch": 0.816819151166775, "grad_norm": 1.9380675554275513, "learning_rate": 1.7093525966970025e-06, "loss": 0.09547887444496155, "step": 39930 }, { "epoch": 0.8170237139394187, "grad_norm": 2.697464942932129, "learning_rate": 1.7056498533588573e-06, "loss": 0.10034617185592651, "step": 39940 }, { "epoch": 0.8172282767120626, "grad_norm": 1.8708711862564087, "learning_rate": 1.7019507508107148e-06, "loss": 0.10030632019042969, "step": 39950 }, { "epoch": 0.8174328394847064, "grad_norm": 2.0624091625213623, "learning_rate": 1.6982552906762873e-06, "loss": 0.1038435935974121, "step": 39960 }, { "epoch": 0.8176374022573502, "grad_norm": 2.2545385360717773, "learning_rate": 1.6945634745776984e-06, "loss": 0.10108296871185303, "step": 39970 }, { "epoch": 0.8178419650299941, "grad_norm": 2.3396248817443848, "learning_rate": 1.6908753041354564e-06, "loss": 0.1047990083694458, "step": 39980 }, { "epoch": 0.8180465278026379, "grad_norm": 3.0256664752960205, "learning_rate": 1.6871907809684783e-06, "loss": 0.1275913953781128, "step": 39990 }, { "epoch": 0.8182510905752817, "grad_norm": 2.026683807373047, "learning_rate": 1.6835099066940798e-06, "loss": 0.1060490608215332, "step": 40000 }, { "epoch": 0.8184556533479255, "grad_norm": 2.4816412925720215, "learning_rate": 1.6798326829279754e-06, "loss": 0.11146030426025391, "step": 40010 }, { "epoch": 0.8186602161205693, "grad_norm": 3.1778223514556885, "learning_rate": 1.6761591112842746e-06, "loss": 0.10491346120834351, "step": 40020 }, { "epoch": 0.8188647788932131, "grad_norm": 2.112105131149292, "learning_rate": 1.6724891933754827e-06, "loss": 0.10833417177200318, "step": 40030 }, { "epoch": 0.819069341665857, "grad_norm": 3.106194257736206, "learning_rate": 1.6688229308125104e-06, "loss": 0.08874717354774475, "step": 40040 }, { "epoch": 0.8192739044385008, "grad_norm": 2.3281853199005127, "learning_rate": 1.6651603252046511e-06, "loss": 0.10752100944519043, "step": 40050 }, { "epoch": 0.8194784672111446, "grad_norm": 2.024083375930786, "learning_rate": 1.6615013781596e-06, "loss": 0.10705904960632324, "step": 40060 }, { "epoch": 0.8196830299837884, "grad_norm": 2.0615601539611816, "learning_rate": 1.6578460912834437e-06, "loss": 0.10762667655944824, "step": 40070 }, { "epoch": 0.8198875927564322, "grad_norm": 2.8381941318511963, "learning_rate": 1.6541944661806675e-06, "loss": 0.11184424161911011, "step": 40080 }, { "epoch": 0.820092155529076, "grad_norm": 2.5311741828918457, "learning_rate": 1.6505465044541435e-06, "loss": 0.1032839298248291, "step": 40090 }, { "epoch": 0.8202967183017199, "grad_norm": 2.2115695476531982, "learning_rate": 1.646902207705139e-06, "loss": 0.11144717931747436, "step": 40100 }, { "epoch": 0.8205012810743637, "grad_norm": 1.8231028318405151, "learning_rate": 1.6432615775333104e-06, "loss": 0.10538294315338134, "step": 40110 }, { "epoch": 0.8207058438470075, "grad_norm": 2.5214920043945312, "learning_rate": 1.6396246155367067e-06, "loss": 0.11507070064544678, "step": 40120 }, { "epoch": 0.8209104066196513, "grad_norm": 2.3818159103393555, "learning_rate": 1.6359913233117651e-06, "loss": 0.11036512851715088, "step": 40130 }, { "epoch": 0.8211149693922951, "grad_norm": 2.8562920093536377, "learning_rate": 1.6323617024533124e-06, "loss": 0.09829012155532837, "step": 40140 }, { "epoch": 0.8213195321649389, "grad_norm": 1.830640196800232, "learning_rate": 1.6287357545545623e-06, "loss": 0.11135170459747315, "step": 40150 }, { "epoch": 0.8215240949375828, "grad_norm": 3.8979175090789795, "learning_rate": 1.6251134812071222e-06, "loss": 0.11236310005187988, "step": 40160 }, { "epoch": 0.8217286577102266, "grad_norm": 2.5558767318725586, "learning_rate": 1.6214948840009804e-06, "loss": 0.09262273311614991, "step": 40170 }, { "epoch": 0.8219332204828704, "grad_norm": 3.160308599472046, "learning_rate": 1.6178799645245147e-06, "loss": 0.10797713994979859, "step": 40180 }, { "epoch": 0.8221377832555142, "grad_norm": 1.7347691059112549, "learning_rate": 1.614268724364485e-06, "loss": 0.10472596883773803, "step": 40190 }, { "epoch": 0.822342346028158, "grad_norm": 2.22299861907959, "learning_rate": 1.610661165106041e-06, "loss": 0.11268017292022706, "step": 40200 }, { "epoch": 0.8225469088008018, "grad_norm": 2.666135549545288, "learning_rate": 1.607057288332713e-06, "loss": 0.11893411874771118, "step": 40210 }, { "epoch": 0.8227514715734457, "grad_norm": 2.295088291168213, "learning_rate": 1.603457095626415e-06, "loss": 0.10883266925811767, "step": 40220 }, { "epoch": 0.8229560343460895, "grad_norm": 3.4068596363067627, "learning_rate": 1.5998605885674478e-06, "loss": 0.09957383275032043, "step": 40230 }, { "epoch": 0.8231605971187334, "grad_norm": 1.8457828760147095, "learning_rate": 1.596267768734492e-06, "loss": 0.08592262268066406, "step": 40240 }, { "epoch": 0.8233651598913772, "grad_norm": 2.7981655597686768, "learning_rate": 1.5926786377046066e-06, "loss": 0.11233310699462891, "step": 40250 }, { "epoch": 0.823569722664021, "grad_norm": 2.0954229831695557, "learning_rate": 1.5890931970532364e-06, "loss": 0.09485563039779663, "step": 40260 }, { "epoch": 0.8237742854366648, "grad_norm": 2.236227512359619, "learning_rate": 1.5855114483542022e-06, "loss": 0.10783636569976807, "step": 40270 }, { "epoch": 0.8239788482093087, "grad_norm": 2.0524210929870605, "learning_rate": 1.5819333931797075e-06, "loss": 0.09782493710517884, "step": 40280 }, { "epoch": 0.8241834109819525, "grad_norm": 1.8869831562042236, "learning_rate": 1.578359033100334e-06, "loss": 0.09940866827964782, "step": 40290 }, { "epoch": 0.8243879737545963, "grad_norm": 1.923824429512024, "learning_rate": 1.5747883696850351e-06, "loss": 0.10142544507980347, "step": 40300 }, { "epoch": 0.8245925365272401, "grad_norm": 2.6170594692230225, "learning_rate": 1.5712214045011543e-06, "loss": 0.1082697868347168, "step": 40310 }, { "epoch": 0.8247970992998839, "grad_norm": 2.538022756576538, "learning_rate": 1.5676581391144019e-06, "loss": 0.10489311218261718, "step": 40320 }, { "epoch": 0.8250016620725278, "grad_norm": 2.0950427055358887, "learning_rate": 1.564098575088866e-06, "loss": 0.09807431697845459, "step": 40330 }, { "epoch": 0.8252062248451716, "grad_norm": 1.3592088222503662, "learning_rate": 1.560542713987012e-06, "loss": 0.10216237306594848, "step": 40340 }, { "epoch": 0.8254107876178154, "grad_norm": 2.575716495513916, "learning_rate": 1.5569905573696776e-06, "loss": 0.0926913559436798, "step": 40350 }, { "epoch": 0.8256153503904592, "grad_norm": 2.2143478393554688, "learning_rate": 1.5534421067960747e-06, "loss": 0.11048270463943481, "step": 40360 }, { "epoch": 0.825819913163103, "grad_norm": 2.406733274459839, "learning_rate": 1.5498973638237891e-06, "loss": 0.1078100323677063, "step": 40370 }, { "epoch": 0.8260244759357468, "grad_norm": 3.889514446258545, "learning_rate": 1.5463563300087836e-06, "loss": 0.09834232330322265, "step": 40380 }, { "epoch": 0.8262290387083907, "grad_norm": 1.5723484754562378, "learning_rate": 1.542819006905384e-06, "loss": 0.10427072048187255, "step": 40390 }, { "epoch": 0.8264336014810345, "grad_norm": 2.7481369972229004, "learning_rate": 1.5392853960662945e-06, "loss": 0.10578441619873047, "step": 40400 }, { "epoch": 0.8266381642536783, "grad_norm": 3.1885855197906494, "learning_rate": 1.535755499042586e-06, "loss": 0.10266652107238769, "step": 40410 }, { "epoch": 0.8268427270263221, "grad_norm": 2.702484369277954, "learning_rate": 1.5322293173837011e-06, "loss": 0.11027222871780396, "step": 40420 }, { "epoch": 0.8270472897989659, "grad_norm": 2.672466278076172, "learning_rate": 1.5287068526374494e-06, "loss": 0.09985281229019165, "step": 40430 }, { "epoch": 0.8272518525716097, "grad_norm": 1.9219701290130615, "learning_rate": 1.5251881063500119e-06, "loss": 0.09863163828849793, "step": 40440 }, { "epoch": 0.8274564153442536, "grad_norm": 2.950132131576538, "learning_rate": 1.521673080065933e-06, "loss": 0.1115225076675415, "step": 40450 }, { "epoch": 0.8276609781168974, "grad_norm": 1.7119600772857666, "learning_rate": 1.518161775328132e-06, "loss": 0.09795212745666504, "step": 40460 }, { "epoch": 0.8278655408895412, "grad_norm": 3.7593257427215576, "learning_rate": 1.514654193677888e-06, "loss": 0.09580729007720948, "step": 40470 }, { "epoch": 0.828070103662185, "grad_norm": 1.8859914541244507, "learning_rate": 1.5111503366548463e-06, "loss": 0.11680371761322021, "step": 40480 }, { "epoch": 0.8282746664348288, "grad_norm": 2.2248566150665283, "learning_rate": 1.507650205797019e-06, "loss": 0.09213466048240662, "step": 40490 }, { "epoch": 0.8284792292074726, "grad_norm": 2.4792275428771973, "learning_rate": 1.504153802640782e-06, "loss": 0.11219416856765747, "step": 40500 }, { "epoch": 0.8286837919801165, "grad_norm": 2.2595834732055664, "learning_rate": 1.5006611287208772e-06, "loss": 0.10633624792098999, "step": 40510 }, { "epoch": 0.8288883547527603, "grad_norm": 2.568371534347534, "learning_rate": 1.4971721855704047e-06, "loss": 0.09314001202583314, "step": 40520 }, { "epoch": 0.8290929175254041, "grad_norm": 2.2005739212036133, "learning_rate": 1.4936869747208294e-06, "loss": 0.10652203559875488, "step": 40530 }, { "epoch": 0.829297480298048, "grad_norm": 2.1662070751190186, "learning_rate": 1.490205497701981e-06, "loss": 0.09790992140769958, "step": 40540 }, { "epoch": 0.8295020430706918, "grad_norm": 2.8966853618621826, "learning_rate": 1.486727756042048e-06, "loss": 0.12253262996673583, "step": 40550 }, { "epoch": 0.8297066058433356, "grad_norm": 1.5708670616149902, "learning_rate": 1.483253751267576e-06, "loss": 0.09113614559173584, "step": 40560 }, { "epoch": 0.8299111686159795, "grad_norm": 1.7517144680023193, "learning_rate": 1.4797834849034754e-06, "loss": 0.10930161476135254, "step": 40570 }, { "epoch": 0.8301157313886233, "grad_norm": 1.8344714641571045, "learning_rate": 1.4763169584730118e-06, "loss": 0.09986476302146911, "step": 40580 }, { "epoch": 0.8303202941612671, "grad_norm": 2.091543674468994, "learning_rate": 1.4728541734978108e-06, "loss": 0.10525300502777099, "step": 40590 }, { "epoch": 0.8305248569339109, "grad_norm": 3.283360004425049, "learning_rate": 1.469395131497855e-06, "loss": 0.11811754703521729, "step": 40600 }, { "epoch": 0.8307294197065547, "grad_norm": 2.2868685722351074, "learning_rate": 1.4659398339914876e-06, "loss": 0.10768847465515137, "step": 40610 }, { "epoch": 0.8309339824791985, "grad_norm": 2.156522274017334, "learning_rate": 1.4624882824954023e-06, "loss": 0.10116767883300781, "step": 40620 }, { "epoch": 0.8311385452518424, "grad_norm": 2.080390691757202, "learning_rate": 1.4590404785246537e-06, "loss": 0.09529653787612916, "step": 40630 }, { "epoch": 0.8313431080244862, "grad_norm": 3.0976219177246094, "learning_rate": 1.4555964235926466e-06, "loss": 0.09512776136398315, "step": 40640 }, { "epoch": 0.83154767079713, "grad_norm": 2.1828737258911133, "learning_rate": 1.4521561192111444e-06, "loss": 0.09998639822006225, "step": 40650 }, { "epoch": 0.8317522335697738, "grad_norm": 2.0740737915039062, "learning_rate": 1.4487195668902622e-06, "loss": 0.10325440168380737, "step": 40660 }, { "epoch": 0.8319567963424176, "grad_norm": 1.9889795780181885, "learning_rate": 1.4452867681384675e-06, "loss": 0.1058894157409668, "step": 40670 }, { "epoch": 0.8321613591150614, "grad_norm": 2.8796916007995605, "learning_rate": 1.4418577244625798e-06, "loss": 0.118892240524292, "step": 40680 }, { "epoch": 0.8323659218877053, "grad_norm": 2.359455108642578, "learning_rate": 1.438432437367776e-06, "loss": 0.10034652948379516, "step": 40690 }, { "epoch": 0.8325704846603491, "grad_norm": 3.1059300899505615, "learning_rate": 1.4350109083575781e-06, "loss": 0.10123920440673828, "step": 40700 }, { "epoch": 0.8327750474329929, "grad_norm": 1.7796612977981567, "learning_rate": 1.4315931389338588e-06, "loss": 0.09231947660446167, "step": 40710 }, { "epoch": 0.8329796102056367, "grad_norm": 2.5481722354888916, "learning_rate": 1.4281791305968417e-06, "loss": 0.10711090564727783, "step": 40720 }, { "epoch": 0.8331841729782805, "grad_norm": 3.1366639137268066, "learning_rate": 1.4247688848451003e-06, "loss": 0.09498519897460937, "step": 40730 }, { "epoch": 0.8333887357509243, "grad_norm": 2.328035354614258, "learning_rate": 1.4213624031755558e-06, "loss": 0.1057063102722168, "step": 40740 }, { "epoch": 0.8335932985235682, "grad_norm": 1.9657732248306274, "learning_rate": 1.4179596870834755e-06, "loss": 0.108225417137146, "step": 40750 }, { "epoch": 0.833797861296212, "grad_norm": 2.1626358032226562, "learning_rate": 1.4145607380624783e-06, "loss": 0.08841596841812134, "step": 40760 }, { "epoch": 0.8340024240688558, "grad_norm": 3.1533825397491455, "learning_rate": 1.4111655576045258e-06, "loss": 0.09798285961151124, "step": 40770 }, { "epoch": 0.8342069868414996, "grad_norm": 2.601947069168091, "learning_rate": 1.4077741471999263e-06, "loss": 0.10276378393173217, "step": 40780 }, { "epoch": 0.8344115496141434, "grad_norm": 2.8881115913391113, "learning_rate": 1.404386508337332e-06, "loss": 0.11184134483337402, "step": 40790 }, { "epoch": 0.8346161123867873, "grad_norm": 2.441200017929077, "learning_rate": 1.4010026425037426e-06, "loss": 0.10493683815002441, "step": 40800 }, { "epoch": 0.8348206751594311, "grad_norm": 2.335071325302124, "learning_rate": 1.3976225511844987e-06, "loss": 0.0981215000152588, "step": 40810 }, { "epoch": 0.835025237932075, "grad_norm": 2.25405216217041, "learning_rate": 1.394246235863287e-06, "loss": 0.09644067287445068, "step": 40820 }, { "epoch": 0.8352298007047188, "grad_norm": 2.03503155708313, "learning_rate": 1.3908736980221304e-06, "loss": 0.10208585262298583, "step": 40830 }, { "epoch": 0.8354343634773626, "grad_norm": 2.601855516433716, "learning_rate": 1.3875049391414031e-06, "loss": 0.08808106184005737, "step": 40840 }, { "epoch": 0.8356389262500064, "grad_norm": 2.549710512161255, "learning_rate": 1.3841399606998151e-06, "loss": 0.12339544296264648, "step": 40850 }, { "epoch": 0.8358434890226503, "grad_norm": 2.8002893924713135, "learning_rate": 1.3807787641744185e-06, "loss": 0.10403268337249756, "step": 40860 }, { "epoch": 0.8360480517952941, "grad_norm": 2.1659841537475586, "learning_rate": 1.3774213510405987e-06, "loss": 0.10026853084564209, "step": 40870 }, { "epoch": 0.8362526145679379, "grad_norm": 3.0204694271087646, "learning_rate": 1.3740677227720923e-06, "loss": 0.11671148538589478, "step": 40880 }, { "epoch": 0.8364571773405817, "grad_norm": 2.25034236907959, "learning_rate": 1.370717880840965e-06, "loss": 0.12032697200775147, "step": 40890 }, { "epoch": 0.8366617401132255, "grad_norm": 2.4368703365325928, "learning_rate": 1.3673718267176261e-06, "loss": 0.10176293849945069, "step": 40900 }, { "epoch": 0.8368663028858693, "grad_norm": 1.8692975044250488, "learning_rate": 1.3640295618708165e-06, "loss": 0.10160398483276367, "step": 40910 }, { "epoch": 0.8370708656585132, "grad_norm": 3.4016127586364746, "learning_rate": 1.3606910877676216e-06, "loss": 0.09938461184501649, "step": 40920 }, { "epoch": 0.837275428431157, "grad_norm": 3.0462169647216797, "learning_rate": 1.357356405873459e-06, "loss": 0.1059085488319397, "step": 40930 }, { "epoch": 0.8374799912038008, "grad_norm": 2.0214710235595703, "learning_rate": 1.3540255176520745e-06, "loss": 0.104710054397583, "step": 40940 }, { "epoch": 0.8376845539764446, "grad_norm": 1.9761567115783691, "learning_rate": 1.3506984245655618e-06, "loss": 0.09551416635513306, "step": 40950 }, { "epoch": 0.8378891167490884, "grad_norm": 3.044142246246338, "learning_rate": 1.347375128074342e-06, "loss": 0.10412411689758301, "step": 40960 }, { "epoch": 0.8380936795217322, "grad_norm": 3.8956217765808105, "learning_rate": 1.3440556296371666e-06, "loss": 0.09150987267494201, "step": 40970 }, { "epoch": 0.8382982422943761, "grad_norm": 2.592001438140869, "learning_rate": 1.3407399307111245e-06, "loss": 0.11736365556716918, "step": 40980 }, { "epoch": 0.8385028050670199, "grad_norm": 2.861807346343994, "learning_rate": 1.3374280327516388e-06, "loss": 0.09418863654136658, "step": 40990 }, { "epoch": 0.8387073678396637, "grad_norm": 3.0962307453155518, "learning_rate": 1.3341199372124613e-06, "loss": 0.09595870971679688, "step": 41000 }, { "epoch": 0.8389119306123075, "grad_norm": 2.1032257080078125, "learning_rate": 1.33081564554567e-06, "loss": 0.10624728202819825, "step": 41010 }, { "epoch": 0.8391164933849513, "grad_norm": 1.8940842151641846, "learning_rate": 1.3275151592016777e-06, "loss": 0.10379941463470459, "step": 41020 }, { "epoch": 0.8393210561575951, "grad_norm": 2.087843179702759, "learning_rate": 1.3242184796292313e-06, "loss": 0.10755971670150757, "step": 41030 }, { "epoch": 0.839525618930239, "grad_norm": 2.135300636291504, "learning_rate": 1.3209256082753997e-06, "loss": 0.1035807490348816, "step": 41040 }, { "epoch": 0.8397301817028828, "grad_norm": 3.2947161197662354, "learning_rate": 1.3176365465855845e-06, "loss": 0.11387996673583985, "step": 41050 }, { "epoch": 0.8399347444755266, "grad_norm": 1.7458224296569824, "learning_rate": 1.31435129600351e-06, "loss": 0.09396942257881165, "step": 41060 }, { "epoch": 0.8401393072481704, "grad_norm": 2.1417653560638428, "learning_rate": 1.3110698579712356e-06, "loss": 0.10506352186203002, "step": 41070 }, { "epoch": 0.8403438700208142, "grad_norm": 4.045594692230225, "learning_rate": 1.3077922339291417e-06, "loss": 0.10202522277832031, "step": 41080 }, { "epoch": 0.840548432793458, "grad_norm": 2.760430097579956, "learning_rate": 1.3045184253159327e-06, "loss": 0.11228822469711304, "step": 41090 }, { "epoch": 0.840752995566102, "grad_norm": 1.8847142457962036, "learning_rate": 1.3012484335686392e-06, "loss": 0.09158333539962768, "step": 41100 }, { "epoch": 0.8409575583387457, "grad_norm": 2.4021072387695312, "learning_rate": 1.2979822601226245e-06, "loss": 0.09271177649497986, "step": 41110 }, { "epoch": 0.8411621211113895, "grad_norm": 1.5670725107192993, "learning_rate": 1.2947199064115655e-06, "loss": 0.10022714138031005, "step": 41120 }, { "epoch": 0.8413666838840334, "grad_norm": 2.07887601852417, "learning_rate": 1.2914613738674653e-06, "loss": 0.10042915344238282, "step": 41130 }, { "epoch": 0.8415712466566772, "grad_norm": 3.6783220767974854, "learning_rate": 1.2882066639206547e-06, "loss": 0.10347371101379395, "step": 41140 }, { "epoch": 0.841775809429321, "grad_norm": 2.265331745147705, "learning_rate": 1.2849557779997834e-06, "loss": 0.09328060746192932, "step": 41150 }, { "epoch": 0.8419803722019649, "grad_norm": 2.496710777282715, "learning_rate": 1.2817087175318176e-06, "loss": 0.1073456883430481, "step": 41160 }, { "epoch": 0.8421849349746087, "grad_norm": 2.6079227924346924, "learning_rate": 1.2784654839420484e-06, "loss": 0.09462481737136841, "step": 41170 }, { "epoch": 0.8423894977472525, "grad_norm": 2.2923617362976074, "learning_rate": 1.2752260786540905e-06, "loss": 0.0939246654510498, "step": 41180 }, { "epoch": 0.8425940605198963, "grad_norm": 3.4029006958007812, "learning_rate": 1.2719905030898749e-06, "loss": 0.099380624294281, "step": 41190 }, { "epoch": 0.8427986232925401, "grad_norm": 2.694415807723999, "learning_rate": 1.2687587586696494e-06, "loss": 0.09621770977973938, "step": 41200 }, { "epoch": 0.8430031860651839, "grad_norm": 1.8776108026504517, "learning_rate": 1.265530846811982e-06, "loss": 0.09856394529342652, "step": 41210 }, { "epoch": 0.8432077488378278, "grad_norm": 1.9665148258209229, "learning_rate": 1.2623067689337653e-06, "loss": 0.0996223509311676, "step": 41220 }, { "epoch": 0.8434123116104716, "grad_norm": 2.262085437774658, "learning_rate": 1.259086526450195e-06, "loss": 0.10418176651000977, "step": 41230 }, { "epoch": 0.8436168743831154, "grad_norm": 4.063558101654053, "learning_rate": 1.2558701207747926e-06, "loss": 0.08770965337753296, "step": 41240 }, { "epoch": 0.8438214371557592, "grad_norm": 2.470606803894043, "learning_rate": 1.252657553319392e-06, "loss": 0.10042374134063721, "step": 41250 }, { "epoch": 0.844025999928403, "grad_norm": 2.7585270404815674, "learning_rate": 1.2494488254941485e-06, "loss": 0.10442850589752198, "step": 41260 }, { "epoch": 0.8442305627010469, "grad_norm": 2.247995615005493, "learning_rate": 1.2462439387075253e-06, "loss": 0.09992085099220276, "step": 41270 }, { "epoch": 0.8444351254736907, "grad_norm": 1.5666146278381348, "learning_rate": 1.2430428943663008e-06, "loss": 0.10015608072280884, "step": 41280 }, { "epoch": 0.8446396882463345, "grad_norm": 1.577820062637329, "learning_rate": 1.2398456938755687e-06, "loss": 0.10080480575561523, "step": 41290 }, { "epoch": 0.8448442510189783, "grad_norm": 2.7658066749572754, "learning_rate": 1.236652338638733e-06, "loss": 0.09806890487670898, "step": 41300 }, { "epoch": 0.8450488137916221, "grad_norm": 2.5040910243988037, "learning_rate": 1.2334628300575125e-06, "loss": 0.09546063542366028, "step": 41310 }, { "epoch": 0.8452533765642659, "grad_norm": 1.9973654747009277, "learning_rate": 1.2302771695319326e-06, "loss": 0.1033826231956482, "step": 41320 }, { "epoch": 0.8454579393369098, "grad_norm": 2.2140567302703857, "learning_rate": 1.2270953584603395e-06, "loss": 0.09924153685569763, "step": 41330 }, { "epoch": 0.8456625021095536, "grad_norm": 2.743718385696411, "learning_rate": 1.2239173982393805e-06, "loss": 0.10284702777862549, "step": 41340 }, { "epoch": 0.8458670648821974, "grad_norm": 2.7520720958709717, "learning_rate": 1.2207432902640148e-06, "loss": 0.10966298580169678, "step": 41350 }, { "epoch": 0.8460716276548412, "grad_norm": 3.150507688522339, "learning_rate": 1.2175730359275118e-06, "loss": 0.11178388595581054, "step": 41360 }, { "epoch": 0.846276190427485, "grad_norm": 2.0701475143432617, "learning_rate": 1.2144066366214491e-06, "loss": 0.10754637718200684, "step": 41370 }, { "epoch": 0.8464807532001288, "grad_norm": 2.536555051803589, "learning_rate": 1.2112440937357107e-06, "loss": 0.10160068273544312, "step": 41380 }, { "epoch": 0.8466853159727727, "grad_norm": 2.6814682483673096, "learning_rate": 1.2080854086584914e-06, "loss": 0.10453946590423584, "step": 41390 }, { "epoch": 0.8468898787454165, "grad_norm": 2.8143820762634277, "learning_rate": 1.204930582776287e-06, "loss": 0.10209896564483642, "step": 41400 }, { "epoch": 0.8470944415180603, "grad_norm": 3.5805552005767822, "learning_rate": 1.201779617473906e-06, "loss": 0.09985234141349793, "step": 41410 }, { "epoch": 0.8472990042907041, "grad_norm": 2.056119203567505, "learning_rate": 1.1986325141344578e-06, "loss": 0.10146327018737793, "step": 41420 }, { "epoch": 0.847503567063348, "grad_norm": 2.570087194442749, "learning_rate": 1.1954892741393575e-06, "loss": 0.10032542943954467, "step": 41430 }, { "epoch": 0.8477081298359918, "grad_norm": 1.8122777938842773, "learning_rate": 1.1923498988683246e-06, "loss": 0.09122350215911865, "step": 41440 }, { "epoch": 0.8479126926086357, "grad_norm": 2.001364231109619, "learning_rate": 1.1892143896993834e-06, "loss": 0.09728973507881164, "step": 41450 }, { "epoch": 0.8481172553812795, "grad_norm": 1.9356110095977783, "learning_rate": 1.1860827480088577e-06, "loss": 0.09522459506988526, "step": 41460 }, { "epoch": 0.8483218181539233, "grad_norm": 2.234039068222046, "learning_rate": 1.1829549751713753e-06, "loss": 0.10804362297058105, "step": 41470 }, { "epoch": 0.8485263809265671, "grad_norm": 2.8061325550079346, "learning_rate": 1.1798310725598695e-06, "loss": 0.09671496748924255, "step": 41480 }, { "epoch": 0.8487309436992109, "grad_norm": 1.9552263021469116, "learning_rate": 1.1767110415455718e-06, "loss": 0.09861636161804199, "step": 41490 }, { "epoch": 0.8489355064718547, "grad_norm": 2.142625093460083, "learning_rate": 1.1735948834980116e-06, "loss": 0.10114817619323731, "step": 41500 }, { "epoch": 0.8491400692444986, "grad_norm": 3.7800204753875732, "learning_rate": 1.1704825997850222e-06, "loss": 0.09885722994804383, "step": 41510 }, { "epoch": 0.8493446320171424, "grad_norm": 2.4083800315856934, "learning_rate": 1.1673741917727356e-06, "loss": 0.0865324854850769, "step": 41520 }, { "epoch": 0.8495491947897862, "grad_norm": 2.504227876663208, "learning_rate": 1.16426966082558e-06, "loss": 0.09567999839782715, "step": 41530 }, { "epoch": 0.84975375756243, "grad_norm": 2.7570810317993164, "learning_rate": 1.1611690083062854e-06, "loss": 0.09272018671035767, "step": 41540 }, { "epoch": 0.8499583203350738, "grad_norm": 3.7296316623687744, "learning_rate": 1.1580722355758744e-06, "loss": 0.10717630386352539, "step": 41550 }, { "epoch": 0.8501628831077176, "grad_norm": 2.44802188873291, "learning_rate": 1.1549793439936752e-06, "loss": 0.11041728258132935, "step": 41560 }, { "epoch": 0.8503674458803615, "grad_norm": 2.3659071922302246, "learning_rate": 1.1518903349173028e-06, "loss": 0.09969156980514526, "step": 41570 }, { "epoch": 0.8505720086530053, "grad_norm": 2.0140769481658936, "learning_rate": 1.1488052097026747e-06, "loss": 0.10773887634277343, "step": 41580 }, { "epoch": 0.8507765714256491, "grad_norm": 2.1148362159729004, "learning_rate": 1.145723969703999e-06, "loss": 0.0921266496181488, "step": 41590 }, { "epoch": 0.8509811341982929, "grad_norm": 1.8653794527053833, "learning_rate": 1.1426466162737803e-06, "loss": 0.10459022521972657, "step": 41600 }, { "epoch": 0.8511856969709367, "grad_norm": 2.2097630500793457, "learning_rate": 1.1395731507628193e-06, "loss": 0.09471639394760131, "step": 41610 }, { "epoch": 0.8513902597435805, "grad_norm": 3.3762922286987305, "learning_rate": 1.1365035745202069e-06, "loss": 0.1142848014831543, "step": 41620 }, { "epoch": 0.8515948225162244, "grad_norm": 1.8816982507705688, "learning_rate": 1.133437888893326e-06, "loss": 0.09848214387893676, "step": 41630 }, { "epoch": 0.8517993852888682, "grad_norm": 2.2500157356262207, "learning_rate": 1.1303760952278576e-06, "loss": 0.10929549932479858, "step": 41640 }, { "epoch": 0.852003948061512, "grad_norm": 2.250623941421509, "learning_rate": 1.127318194867769e-06, "loss": 0.09826035499572754, "step": 41650 }, { "epoch": 0.8522085108341558, "grad_norm": 2.7509326934814453, "learning_rate": 1.1242641891553208e-06, "loss": 0.09792498350143433, "step": 41660 }, { "epoch": 0.8524130736067996, "grad_norm": 1.879056453704834, "learning_rate": 1.121214079431061e-06, "loss": 0.08152998685836792, "step": 41670 }, { "epoch": 0.8526176363794434, "grad_norm": 2.366943597793579, "learning_rate": 1.1181678670338313e-06, "loss": 0.10182819366455079, "step": 41680 }, { "epoch": 0.8528221991520873, "grad_norm": 2.8244402408599854, "learning_rate": 1.115125553300762e-06, "loss": 0.08579021096229553, "step": 41690 }, { "epoch": 0.8530267619247311, "grad_norm": 2.256617784500122, "learning_rate": 1.1120871395672672e-06, "loss": 0.09880518317222595, "step": 41700 }, { "epoch": 0.853231324697375, "grad_norm": 2.5197958946228027, "learning_rate": 1.1090526271670587e-06, "loss": 0.09954456090927125, "step": 41710 }, { "epoch": 0.8534358874700188, "grad_norm": 3.36533522605896, "learning_rate": 1.1060220174321268e-06, "loss": 0.10114195346832275, "step": 41720 }, { "epoch": 0.8536404502426626, "grad_norm": 2.7471165657043457, "learning_rate": 1.1029953116927528e-06, "loss": 0.11130259037017823, "step": 41730 }, { "epoch": 0.8538450130153065, "grad_norm": 3.1654906272888184, "learning_rate": 1.0999725112775051e-06, "loss": 0.09738863706588745, "step": 41740 }, { "epoch": 0.8540495757879503, "grad_norm": 3.0436182022094727, "learning_rate": 1.0969536175132344e-06, "loss": 0.0994236946105957, "step": 41750 }, { "epoch": 0.8542541385605941, "grad_norm": 1.9069856405258179, "learning_rate": 1.093938631725079e-06, "loss": 0.11678731441497803, "step": 41760 }, { "epoch": 0.8544587013332379, "grad_norm": 2.5217201709747314, "learning_rate": 1.090927555236463e-06, "loss": 0.08757926225662231, "step": 41770 }, { "epoch": 0.8546632641058817, "grad_norm": 3.1792147159576416, "learning_rate": 1.0879203893690893e-06, "loss": 0.10251922607421875, "step": 41780 }, { "epoch": 0.8548678268785255, "grad_norm": 2.46087384223938, "learning_rate": 1.0849171354429523e-06, "loss": 0.09572468996047974, "step": 41790 }, { "epoch": 0.8550723896511694, "grad_norm": 2.5984599590301514, "learning_rate": 1.0819177947763238e-06, "loss": 0.10594003200531006, "step": 41800 }, { "epoch": 0.8552769524238132, "grad_norm": 2.6225714683532715, "learning_rate": 1.0789223686857574e-06, "loss": 0.10230109691619874, "step": 41810 }, { "epoch": 0.855481515196457, "grad_norm": 4.262283802032471, "learning_rate": 1.0759308584860917e-06, "loss": 0.10138044357299805, "step": 41820 }, { "epoch": 0.8556860779691008, "grad_norm": 1.8892093896865845, "learning_rate": 1.0729432654904438e-06, "loss": 0.10659546852111816, "step": 41830 }, { "epoch": 0.8558906407417446, "grad_norm": 1.7323594093322754, "learning_rate": 1.0699595910102122e-06, "loss": 0.09895220398902893, "step": 41840 }, { "epoch": 0.8560952035143884, "grad_norm": 1.491500973701477, "learning_rate": 1.0669798363550743e-06, "loss": 0.10634334087371826, "step": 41850 }, { "epoch": 0.8562997662870323, "grad_norm": 1.893746256828308, "learning_rate": 1.06400400283299e-06, "loss": 0.10248857736587524, "step": 41860 }, { "epoch": 0.8565043290596761, "grad_norm": 3.1219122409820557, "learning_rate": 1.061032091750196e-06, "loss": 0.09133328199386596, "step": 41870 }, { "epoch": 0.8567088918323199, "grad_norm": 3.115088939666748, "learning_rate": 1.058064104411206e-06, "loss": 0.11114755868911744, "step": 41880 }, { "epoch": 0.8569134546049637, "grad_norm": 2.1482090950012207, "learning_rate": 1.055100042118814e-06, "loss": 0.09948391318321229, "step": 41890 }, { "epoch": 0.8571180173776075, "grad_norm": 2.7682249546051025, "learning_rate": 1.052139906174089e-06, "loss": 0.10278627872467042, "step": 41900 }, { "epoch": 0.8573225801502513, "grad_norm": 2.1927905082702637, "learning_rate": 1.049183697876377e-06, "loss": 0.10718188285827637, "step": 41910 }, { "epoch": 0.8575271429228952, "grad_norm": 1.9906980991363525, "learning_rate": 1.0462314185233002e-06, "loss": 0.09468124508857727, "step": 41920 }, { "epoch": 0.857731705695539, "grad_norm": 4.255490779876709, "learning_rate": 1.0432830694107554e-06, "loss": 0.09639756679534912, "step": 41930 }, { "epoch": 0.8579362684681828, "grad_norm": 2.6592049598693848, "learning_rate": 1.0403386518329183e-06, "loss": 0.09760783910751343, "step": 41940 }, { "epoch": 0.8581408312408266, "grad_norm": 2.2214646339416504, "learning_rate": 1.0373981670822342e-06, "loss": 0.10666570663452149, "step": 41950 }, { "epoch": 0.8583453940134704, "grad_norm": 1.741876482963562, "learning_rate": 1.0344616164494227e-06, "loss": 0.09770535826683044, "step": 41960 }, { "epoch": 0.8585499567861142, "grad_norm": 3.2162444591522217, "learning_rate": 1.031529001223478e-06, "loss": 0.09241122007369995, "step": 41970 }, { "epoch": 0.8587545195587581, "grad_norm": 2.4975287914276123, "learning_rate": 1.0286003226916662e-06, "loss": 0.09979990720748902, "step": 41980 }, { "epoch": 0.858959082331402, "grad_norm": 2.126906394958496, "learning_rate": 1.025675582139526e-06, "loss": 0.09222307205200195, "step": 41990 }, { "epoch": 0.8591636451040457, "grad_norm": 2.801483154296875, "learning_rate": 1.0227547808508654e-06, "loss": 0.1080385684967041, "step": 42000 }, { "epoch": 0.8593682078766895, "grad_norm": 2.5963919162750244, "learning_rate": 1.0198379201077647e-06, "loss": 0.10762262344360352, "step": 42010 }, { "epoch": 0.8595727706493334, "grad_norm": 1.7134605646133423, "learning_rate": 1.0169250011905762e-06, "loss": 0.09542151689529418, "step": 42020 }, { "epoch": 0.8597773334219772, "grad_norm": 3.03226375579834, "learning_rate": 1.014016025377922e-06, "loss": 0.09433495998382568, "step": 42030 }, { "epoch": 0.8599818961946211, "grad_norm": 3.066803455352783, "learning_rate": 1.011110993946688e-06, "loss": 0.10134690999984741, "step": 42040 }, { "epoch": 0.8601864589672649, "grad_norm": 2.0334911346435547, "learning_rate": 1.0082099081720354e-06, "loss": 0.10523964166641235, "step": 42050 }, { "epoch": 0.8603910217399087, "grad_norm": 2.4319498538970947, "learning_rate": 1.005312769327389e-06, "loss": 0.10065212249755859, "step": 42060 }, { "epoch": 0.8605955845125525, "grad_norm": 3.9581456184387207, "learning_rate": 1.0024195786844426e-06, "loss": 0.10276695489883422, "step": 42070 }, { "epoch": 0.8608001472851963, "grad_norm": 2.812511920928955, "learning_rate": 9.99530337513157e-07, "loss": 0.10033688545227051, "step": 42080 }, { "epoch": 0.8610047100578401, "grad_norm": 2.5057220458984375, "learning_rate": 9.966450470817602e-07, "loss": 0.09474114179611207, "step": 42090 }, { "epoch": 0.861209272830484, "grad_norm": 3.4452362060546875, "learning_rate": 9.937637086567453e-07, "loss": 0.09307560324668884, "step": 42100 }, { "epoch": 0.8614138356031278, "grad_norm": 2.017284393310547, "learning_rate": 9.90886323502872e-07, "loss": 0.0913865327835083, "step": 42110 }, { "epoch": 0.8616183983757716, "grad_norm": 2.7039623260498047, "learning_rate": 9.88012892883158e-07, "loss": 0.11187384128570557, "step": 42120 }, { "epoch": 0.8618229611484154, "grad_norm": 2.504464626312256, "learning_rate": 9.85143418058896e-07, "loss": 0.10150341987609864, "step": 42130 }, { "epoch": 0.8620275239210592, "grad_norm": 3.2271971702575684, "learning_rate": 9.822779002896342e-07, "loss": 0.09632359743118286, "step": 42140 }, { "epoch": 0.862232086693703, "grad_norm": 3.0323562622070312, "learning_rate": 9.79416340833187e-07, "loss": 0.10005238056182861, "step": 42150 }, { "epoch": 0.8624366494663469, "grad_norm": 1.8546184301376343, "learning_rate": 9.76558740945629e-07, "loss": 0.09842500686645508, "step": 42160 }, { "epoch": 0.8626412122389907, "grad_norm": 3.1417672634124756, "learning_rate": 9.737051018813027e-07, "loss": 0.10384441614151001, "step": 42170 }, { "epoch": 0.8628457750116345, "grad_norm": 2.2951183319091797, "learning_rate": 9.70855424892807e-07, "loss": 0.09152894616127014, "step": 42180 }, { "epoch": 0.8630503377842783, "grad_norm": 1.9812356233596802, "learning_rate": 9.680097112309982e-07, "loss": 0.09826791882514954, "step": 42190 }, { "epoch": 0.8632549005569221, "grad_norm": 2.4018962383270264, "learning_rate": 9.651679621449984e-07, "loss": 0.10558748245239258, "step": 42200 }, { "epoch": 0.863459463329566, "grad_norm": 1.5419062376022339, "learning_rate": 9.623301788821915e-07, "loss": 0.08439652919769287, "step": 42210 }, { "epoch": 0.8636640261022098, "grad_norm": 3.2515480518341064, "learning_rate": 9.594963626882147e-07, "loss": 0.10090103149414062, "step": 42220 }, { "epoch": 0.8638685888748536, "grad_norm": 1.8303240537643433, "learning_rate": 9.566665148069666e-07, "loss": 0.0913021445274353, "step": 42230 }, { "epoch": 0.8640731516474974, "grad_norm": 2.9535303115844727, "learning_rate": 9.538406364806051e-07, "loss": 0.0947589933872223, "step": 42240 }, { "epoch": 0.8642777144201412, "grad_norm": 2.3379571437835693, "learning_rate": 9.510187289495443e-07, "loss": 0.10658292770385742, "step": 42250 }, { "epoch": 0.864482277192785, "grad_norm": 3.0603277683258057, "learning_rate": 9.482007934524562e-07, "loss": 0.09259152412414551, "step": 42260 }, { "epoch": 0.8646868399654289, "grad_norm": 2.070141315460205, "learning_rate": 9.453868312262637e-07, "loss": 0.0905637264251709, "step": 42270 }, { "epoch": 0.8648914027380727, "grad_norm": 1.8030163049697876, "learning_rate": 9.425768435061567e-07, "loss": 0.11139533519744874, "step": 42280 }, { "epoch": 0.8650959655107165, "grad_norm": 2.135593891143799, "learning_rate": 9.397708315255704e-07, "loss": 0.10242164134979248, "step": 42290 }, { "epoch": 0.8653005282833603, "grad_norm": 1.8607219457626343, "learning_rate": 9.369687965162e-07, "loss": 0.10314139127731323, "step": 42300 }, { "epoch": 0.8655050910560041, "grad_norm": 1.7895866632461548, "learning_rate": 9.341707397079914e-07, "loss": 0.08528602719306946, "step": 42310 }, { "epoch": 0.865709653828648, "grad_norm": 2.058980941772461, "learning_rate": 9.313766623291498e-07, "loss": 0.09607845544815063, "step": 42320 }, { "epoch": 0.8659142166012919, "grad_norm": 1.4596108198165894, "learning_rate": 9.28586565606131e-07, "loss": 0.09427461624145508, "step": 42330 }, { "epoch": 0.8661187793739357, "grad_norm": 2.467801570892334, "learning_rate": 9.258004507636397e-07, "loss": 0.11007261276245117, "step": 42340 }, { "epoch": 0.8663233421465795, "grad_norm": 2.6229183673858643, "learning_rate": 9.230183190246366e-07, "loss": 0.08030056357383727, "step": 42350 }, { "epoch": 0.8665279049192233, "grad_norm": 2.320037364959717, "learning_rate": 9.202401716103348e-07, "loss": 0.11603994369506836, "step": 42360 }, { "epoch": 0.8667324676918671, "grad_norm": 1.7062532901763916, "learning_rate": 9.174660097401966e-07, "loss": 0.0965226948261261, "step": 42370 }, { "epoch": 0.8669370304645109, "grad_norm": 2.1868462562561035, "learning_rate": 9.146958346319346e-07, "loss": 0.09913714528083802, "step": 42380 }, { "epoch": 0.8671415932371548, "grad_norm": 2.25441312789917, "learning_rate": 9.119296475015116e-07, "loss": 0.09256514310836791, "step": 42390 }, { "epoch": 0.8673461560097986, "grad_norm": 1.9729464054107666, "learning_rate": 9.091674495631453e-07, "loss": 0.09189914464950562, "step": 42400 }, { "epoch": 0.8675507187824424, "grad_norm": 2.414226770401001, "learning_rate": 9.064092420292913e-07, "loss": 0.08998509645462036, "step": 42410 }, { "epoch": 0.8677552815550862, "grad_norm": 2.625713348388672, "learning_rate": 9.036550261106614e-07, "loss": 0.09368718266487122, "step": 42420 }, { "epoch": 0.86795984432773, "grad_norm": 1.8284556865692139, "learning_rate": 9.00904803016216e-07, "loss": 0.09423928260803223, "step": 42430 }, { "epoch": 0.8681644071003738, "grad_norm": 3.3825342655181885, "learning_rate": 8.981585739531584e-07, "loss": 0.08714605569839477, "step": 42440 }, { "epoch": 0.8683689698730177, "grad_norm": 2.8675858974456787, "learning_rate": 8.954163401269422e-07, "loss": 0.08935041427612304, "step": 42450 }, { "epoch": 0.8685735326456615, "grad_norm": 3.2805604934692383, "learning_rate": 8.926781027412634e-07, "loss": 0.09644453525543213, "step": 42460 }, { "epoch": 0.8687780954183053, "grad_norm": 2.926638603210449, "learning_rate": 8.899438629980728e-07, "loss": 0.09752184748649598, "step": 42470 }, { "epoch": 0.8689826581909491, "grad_norm": 2.412273406982422, "learning_rate": 8.872136220975525e-07, "loss": 0.10197889804840088, "step": 42480 }, { "epoch": 0.8691872209635929, "grad_norm": 1.982802152633667, "learning_rate": 8.844873812381405e-07, "loss": 0.09788657426834106, "step": 42490 }, { "epoch": 0.8693917837362367, "grad_norm": 2.6268861293792725, "learning_rate": 8.81765141616513e-07, "loss": 0.09955962896347045, "step": 42500 }, { "epoch": 0.8695963465088806, "grad_norm": 3.189220428466797, "learning_rate": 8.790469044275951e-07, "loss": 0.09440487623214722, "step": 42510 }, { "epoch": 0.8698009092815244, "grad_norm": 2.9939465522766113, "learning_rate": 8.763326708645514e-07, "loss": 0.09423521757125855, "step": 42520 }, { "epoch": 0.8700054720541682, "grad_norm": 3.690722703933716, "learning_rate": 8.736224421187889e-07, "loss": 0.09470462799072266, "step": 42530 }, { "epoch": 0.870210034826812, "grad_norm": 2.2473676204681396, "learning_rate": 8.709162193799569e-07, "loss": 0.0930273175239563, "step": 42540 }, { "epoch": 0.8704145975994558, "grad_norm": 2.8469908237457275, "learning_rate": 8.682140038359499e-07, "loss": 0.10738511085510254, "step": 42550 }, { "epoch": 0.8706191603720996, "grad_norm": 2.219310760498047, "learning_rate": 8.655157966728977e-07, "loss": 0.0903561294078827, "step": 42560 }, { "epoch": 0.8708237231447435, "grad_norm": 1.9279570579528809, "learning_rate": 8.628215990751721e-07, "loss": 0.10147497653961182, "step": 42570 }, { "epoch": 0.8710282859173873, "grad_norm": 3.133665084838867, "learning_rate": 8.601314122253912e-07, "loss": 0.10696033239364625, "step": 42580 }, { "epoch": 0.8712328486900311, "grad_norm": 1.8843393325805664, "learning_rate": 8.57445237304404e-07, "loss": 0.09811513423919678, "step": 42590 }, { "epoch": 0.871437411462675, "grad_norm": 2.297562599182129, "learning_rate": 8.547630754913039e-07, "loss": 0.10329445600509643, "step": 42600 }, { "epoch": 0.8716419742353188, "grad_norm": 2.610138177871704, "learning_rate": 8.520849279634191e-07, "loss": 0.09924301505088806, "step": 42610 }, { "epoch": 0.8718465370079626, "grad_norm": 2.087007761001587, "learning_rate": 8.494107958963215e-07, "loss": 0.10560858249664307, "step": 42620 }, { "epoch": 0.8720510997806065, "grad_norm": 2.459803819656372, "learning_rate": 8.467406804638112e-07, "loss": 0.0926332950592041, "step": 42630 }, { "epoch": 0.8722556625532503, "grad_norm": 2.294088840484619, "learning_rate": 8.440745828379337e-07, "loss": 0.0973060131072998, "step": 42640 }, { "epoch": 0.8724602253258941, "grad_norm": 3.8828773498535156, "learning_rate": 8.414125041889664e-07, "loss": 0.11776903867721558, "step": 42650 }, { "epoch": 0.8726647880985379, "grad_norm": 2.6773715019226074, "learning_rate": 8.387544456854258e-07, "loss": 0.09263467788696289, "step": 42660 }, { "epoch": 0.8728693508711817, "grad_norm": 3.5233731269836426, "learning_rate": 8.361004084940605e-07, "loss": 0.10078191757202148, "step": 42670 }, { "epoch": 0.8730739136438256, "grad_norm": 3.147778272628784, "learning_rate": 8.334503937798555e-07, "loss": 0.10914474725723267, "step": 42680 }, { "epoch": 0.8732784764164694, "grad_norm": 2.5674819946289062, "learning_rate": 8.308044027060303e-07, "loss": 0.09785717725753784, "step": 42690 }, { "epoch": 0.8734830391891132, "grad_norm": 2.430992603302002, "learning_rate": 8.281624364340368e-07, "loss": 0.10308805704116822, "step": 42700 }, { "epoch": 0.873687601961757, "grad_norm": 2.140009880065918, "learning_rate": 8.255244961235632e-07, "loss": 0.10936274528503417, "step": 42710 }, { "epoch": 0.8738921647344008, "grad_norm": 2.709892749786377, "learning_rate": 8.228905829325273e-07, "loss": 0.08993752002716064, "step": 42720 }, { "epoch": 0.8740967275070446, "grad_norm": 2.5497138500213623, "learning_rate": 8.202606980170779e-07, "loss": 0.0975091814994812, "step": 42730 }, { "epoch": 0.8743012902796885, "grad_norm": 1.438055157661438, "learning_rate": 8.176348425316039e-07, "loss": 0.09150252938270569, "step": 42740 }, { "epoch": 0.8745058530523323, "grad_norm": 2.7158989906311035, "learning_rate": 8.150130176287174e-07, "loss": 0.08986763954162598, "step": 42750 }, { "epoch": 0.8747104158249761, "grad_norm": 2.0167477130889893, "learning_rate": 8.123952244592637e-07, "loss": 0.09030214548110962, "step": 42760 }, { "epoch": 0.8749149785976199, "grad_norm": 2.5947024822235107, "learning_rate": 8.097814641723167e-07, "loss": 0.1049875259399414, "step": 42770 }, { "epoch": 0.8751195413702637, "grad_norm": 4.440253734588623, "learning_rate": 8.071717379151834e-07, "loss": 0.10836577415466309, "step": 42780 }, { "epoch": 0.8753241041429075, "grad_norm": 2.8808047771453857, "learning_rate": 8.045660468333994e-07, "loss": 0.09211840629577636, "step": 42790 }, { "epoch": 0.8755286669155514, "grad_norm": 3.149700880050659, "learning_rate": 8.019643920707231e-07, "loss": 0.09291287064552307, "step": 42800 }, { "epoch": 0.8757332296881952, "grad_norm": 2.1425957679748535, "learning_rate": 7.993667747691535e-07, "loss": 0.09683237075805665, "step": 42810 }, { "epoch": 0.875937792460839, "grad_norm": 1.7793656587600708, "learning_rate": 7.967731960689051e-07, "loss": 0.09731236100196838, "step": 42820 }, { "epoch": 0.8761423552334828, "grad_norm": 2.190244197845459, "learning_rate": 7.94183657108426e-07, "loss": 0.09307746291160583, "step": 42830 }, { "epoch": 0.8763469180061266, "grad_norm": 1.3796919584274292, "learning_rate": 7.915981590243904e-07, "loss": 0.09782171249389648, "step": 42840 }, { "epoch": 0.8765514807787704, "grad_norm": 3.723215103149414, "learning_rate": 7.890167029516959e-07, "loss": 0.10041557550430298, "step": 42850 }, { "epoch": 0.8767560435514143, "grad_norm": 1.6901307106018066, "learning_rate": 7.864392900234696e-07, "loss": 0.09454631805419922, "step": 42860 }, { "epoch": 0.8769606063240581, "grad_norm": 2.1535067558288574, "learning_rate": 7.83865921371062e-07, "loss": 0.09311556816101074, "step": 42870 }, { "epoch": 0.877165169096702, "grad_norm": 2.4177114963531494, "learning_rate": 7.812965981240472e-07, "loss": 0.08037184476852417, "step": 42880 }, { "epoch": 0.8773697318693457, "grad_norm": 3.322697877883911, "learning_rate": 7.787313214102277e-07, "loss": 0.1082659363746643, "step": 42890 }, { "epoch": 0.8775742946419895, "grad_norm": 3.309983015060425, "learning_rate": 7.761700923556248e-07, "loss": 0.10263789892196655, "step": 42900 }, { "epoch": 0.8777788574146334, "grad_norm": 1.9585587978363037, "learning_rate": 7.736129120844871e-07, "loss": 0.08256921768188477, "step": 42910 }, { "epoch": 0.8779834201872773, "grad_norm": 2.8189220428466797, "learning_rate": 7.710597817192834e-07, "loss": 0.10358405113220215, "step": 42920 }, { "epoch": 0.8781879829599211, "grad_norm": 1.957263708114624, "learning_rate": 7.685107023807048e-07, "loss": 0.0854286789894104, "step": 42930 }, { "epoch": 0.8783925457325649, "grad_norm": 2.4941792488098145, "learning_rate": 7.659656751876666e-07, "loss": 0.09514263272285461, "step": 42940 }, { "epoch": 0.8785971085052087, "grad_norm": 3.2300846576690674, "learning_rate": 7.634247012573015e-07, "loss": 0.08960132002830505, "step": 42950 }, { "epoch": 0.8788016712778525, "grad_norm": 2.0802855491638184, "learning_rate": 7.608877817049676e-07, "loss": 0.09444501399993896, "step": 42960 }, { "epoch": 0.8790062340504963, "grad_norm": 2.443725824356079, "learning_rate": 7.583549176442418e-07, "loss": 0.10950803756713867, "step": 42970 }, { "epoch": 0.8792107968231402, "grad_norm": 2.52451229095459, "learning_rate": 7.558261101869191e-07, "loss": 0.09782232642173767, "step": 42980 }, { "epoch": 0.879415359595784, "grad_norm": 2.3680872917175293, "learning_rate": 7.533013604430139e-07, "loss": 0.10573616027832031, "step": 42990 }, { "epoch": 0.8796199223684278, "grad_norm": 3.111039161682129, "learning_rate": 7.50780669520762e-07, "loss": 0.08340484499931336, "step": 43000 }, { "epoch": 0.8798244851410716, "grad_norm": 1.51006019115448, "learning_rate": 7.48264038526616e-07, "loss": 0.09332603216171265, "step": 43010 }, { "epoch": 0.8800290479137154, "grad_norm": 2.181248664855957, "learning_rate": 7.457514685652467e-07, "loss": 0.10479509830474854, "step": 43020 }, { "epoch": 0.8802336106863592, "grad_norm": 2.070763349533081, "learning_rate": 7.432429607395386e-07, "loss": 0.089536052942276, "step": 43030 }, { "epoch": 0.8804381734590031, "grad_norm": 2.5993783473968506, "learning_rate": 7.407385161506019e-07, "loss": 0.10307813882827759, "step": 43040 }, { "epoch": 0.8806427362316469, "grad_norm": 2.518876314163208, "learning_rate": 7.382381358977564e-07, "loss": 0.10284897089004516, "step": 43050 }, { "epoch": 0.8808472990042907, "grad_norm": 2.1310510635375977, "learning_rate": 7.357418210785384e-07, "loss": 0.10188069343566894, "step": 43060 }, { "epoch": 0.8810518617769345, "grad_norm": 2.8136088848114014, "learning_rate": 7.332495727887012e-07, "loss": 0.09901711940765381, "step": 43070 }, { "epoch": 0.8812564245495783, "grad_norm": 2.166034698486328, "learning_rate": 7.307613921222123e-07, "loss": 0.08862468600273132, "step": 43080 }, { "epoch": 0.8814609873222221, "grad_norm": 3.2628865242004395, "learning_rate": 7.282772801712557e-07, "loss": 0.08511799573898315, "step": 43090 }, { "epoch": 0.881665550094866, "grad_norm": 2.8041586875915527, "learning_rate": 7.257972380262257e-07, "loss": 0.10370081663131714, "step": 43100 }, { "epoch": 0.8818701128675098, "grad_norm": 2.283175468444824, "learning_rate": 7.233212667757328e-07, "loss": 0.0866556167602539, "step": 43110 }, { "epoch": 0.8820746756401536, "grad_norm": 2.1968936920166016, "learning_rate": 7.208493675066009e-07, "loss": 0.10301556587219238, "step": 43120 }, { "epoch": 0.8822792384127974, "grad_norm": 2.3202760219573975, "learning_rate": 7.183815413038664e-07, "loss": 0.09216111898422241, "step": 43130 }, { "epoch": 0.8824838011854412, "grad_norm": 3.318720579147339, "learning_rate": 7.159177892507762e-07, "loss": 0.09433913230895996, "step": 43140 }, { "epoch": 0.8826883639580851, "grad_norm": 3.943277597427368, "learning_rate": 7.134581124287898e-07, "loss": 0.09640771746635438, "step": 43150 }, { "epoch": 0.8828929267307289, "grad_norm": 2.642425775527954, "learning_rate": 7.110025119175778e-07, "loss": 0.08870747685432434, "step": 43160 }, { "epoch": 0.8830974895033727, "grad_norm": 3.016057252883911, "learning_rate": 7.085509887950215e-07, "loss": 0.0910951018333435, "step": 43170 }, { "epoch": 0.8833020522760165, "grad_norm": 2.071544885635376, "learning_rate": 7.06103544137211e-07, "loss": 0.08641647100448609, "step": 43180 }, { "epoch": 0.8835066150486603, "grad_norm": 2.3414926528930664, "learning_rate": 7.036601790184517e-07, "loss": 0.08918828368186951, "step": 43190 }, { "epoch": 0.8837111778213042, "grad_norm": 3.249558448791504, "learning_rate": 7.01220894511252e-07, "loss": 0.09277529716491699, "step": 43200 }, { "epoch": 0.8839157405939481, "grad_norm": 2.167829751968384, "learning_rate": 6.987856916863312e-07, "loss": 0.09463545083999633, "step": 43210 }, { "epoch": 0.8841203033665919, "grad_norm": 3.4121336936950684, "learning_rate": 6.963545716126174e-07, "loss": 0.10466055870056153, "step": 43220 }, { "epoch": 0.8843248661392357, "grad_norm": 2.3919312953948975, "learning_rate": 6.939275353572461e-07, "loss": 0.0943914771080017, "step": 43230 }, { "epoch": 0.8845294289118795, "grad_norm": 3.327089309692383, "learning_rate": 6.915045839855617e-07, "loss": 0.09509564638137817, "step": 43240 }, { "epoch": 0.8847339916845233, "grad_norm": 2.795529365539551, "learning_rate": 6.890857185611132e-07, "loss": 0.09464994668960572, "step": 43250 }, { "epoch": 0.8849385544571671, "grad_norm": 1.8786324262619019, "learning_rate": 6.866709401456562e-07, "loss": 0.09419904947280884, "step": 43260 }, { "epoch": 0.885143117229811, "grad_norm": 3.08400297164917, "learning_rate": 6.842602497991546e-07, "loss": 0.09676127433776856, "step": 43270 }, { "epoch": 0.8853476800024548, "grad_norm": 2.8885695934295654, "learning_rate": 6.818536485797777e-07, "loss": 0.09507544040679931, "step": 43280 }, { "epoch": 0.8855522427750986, "grad_norm": 3.4119980335235596, "learning_rate": 6.794511375438995e-07, "loss": 0.09059889316558838, "step": 43290 }, { "epoch": 0.8857568055477424, "grad_norm": 2.170494794845581, "learning_rate": 6.770527177460928e-07, "loss": 0.10255272388458252, "step": 43300 }, { "epoch": 0.8859613683203862, "grad_norm": 1.9391752481460571, "learning_rate": 6.746583902391435e-07, "loss": 0.09599505662918091, "step": 43310 }, { "epoch": 0.88616593109303, "grad_norm": 2.342733144760132, "learning_rate": 6.722681560740374e-07, "loss": 0.10567126274108887, "step": 43320 }, { "epoch": 0.8863704938656739, "grad_norm": 2.263944387435913, "learning_rate": 6.698820162999609e-07, "loss": 0.09872909784317016, "step": 43330 }, { "epoch": 0.8865750566383177, "grad_norm": 2.4934849739074707, "learning_rate": 6.674999719643094e-07, "loss": 0.10585286617279052, "step": 43340 }, { "epoch": 0.8867796194109615, "grad_norm": 2.326606512069702, "learning_rate": 6.651220241126744e-07, "loss": 0.08403881788253784, "step": 43350 }, { "epoch": 0.8869841821836053, "grad_norm": 1.9726029634475708, "learning_rate": 6.627481737888541e-07, "loss": 0.0916244387626648, "step": 43360 }, { "epoch": 0.8871887449562491, "grad_norm": 2.7197892665863037, "learning_rate": 6.603784220348419e-07, "loss": 0.1087526559829712, "step": 43370 }, { "epoch": 0.8873933077288929, "grad_norm": 1.8476738929748535, "learning_rate": 6.58012769890839e-07, "loss": 0.09432479739189148, "step": 43380 }, { "epoch": 0.8875978705015368, "grad_norm": 2.587588310241699, "learning_rate": 6.556512183952446e-07, "loss": 0.10366365909576417, "step": 43390 }, { "epoch": 0.8878024332741806, "grad_norm": 2.362419366836548, "learning_rate": 6.532937685846552e-07, "loss": 0.09770406484603882, "step": 43400 }, { "epoch": 0.8880069960468244, "grad_norm": 2.9779818058013916, "learning_rate": 6.509404214938697e-07, "loss": 0.11165394783020019, "step": 43410 }, { "epoch": 0.8882115588194682, "grad_norm": 2.847105026245117, "learning_rate": 6.485911781558862e-07, "loss": 0.0990295171737671, "step": 43420 }, { "epoch": 0.888416121592112, "grad_norm": 2.0577666759490967, "learning_rate": 6.462460396019021e-07, "loss": 0.09129841923713684, "step": 43430 }, { "epoch": 0.8886206843647558, "grad_norm": 2.2785227298736572, "learning_rate": 6.439050068613117e-07, "loss": 0.08687044382095337, "step": 43440 }, { "epoch": 0.8888252471373997, "grad_norm": 2.1882941722869873, "learning_rate": 6.415680809617031e-07, "loss": 0.10784759521484374, "step": 43450 }, { "epoch": 0.8890298099100435, "grad_norm": 2.2384161949157715, "learning_rate": 6.39235262928869e-07, "loss": 0.0847083568572998, "step": 43460 }, { "epoch": 0.8892343726826873, "grad_norm": 2.4498281478881836, "learning_rate": 6.369065537867958e-07, "loss": 0.08694566488265991, "step": 43470 }, { "epoch": 0.8894389354553311, "grad_norm": 2.5363829135894775, "learning_rate": 6.345819545576648e-07, "loss": 0.11205191612243652, "step": 43480 }, { "epoch": 0.889643498227975, "grad_norm": 2.1455657482147217, "learning_rate": 6.322614662618542e-07, "loss": 0.11207611560821533, "step": 43490 }, { "epoch": 0.8898480610006188, "grad_norm": 2.474905490875244, "learning_rate": 6.299450899179405e-07, "loss": 0.09871325492858887, "step": 43500 }, { "epoch": 0.8900526237732627, "grad_norm": 3.107602119445801, "learning_rate": 6.276328265426923e-07, "loss": 0.08856587409973145, "step": 43510 }, { "epoch": 0.8902571865459065, "grad_norm": 2.492253541946411, "learning_rate": 6.253246771510713e-07, "loss": 0.09142745733261108, "step": 43520 }, { "epoch": 0.8904617493185503, "grad_norm": 2.8726298809051514, "learning_rate": 6.230206427562369e-07, "loss": 0.09400997757911682, "step": 43530 }, { "epoch": 0.8906663120911941, "grad_norm": 3.6955530643463135, "learning_rate": 6.207207243695423e-07, "loss": 0.1025304913520813, "step": 43540 }, { "epoch": 0.8908708748638379, "grad_norm": 1.6476889848709106, "learning_rate": 6.184249230005301e-07, "loss": 0.08254374265670776, "step": 43550 }, { "epoch": 0.8910754376364817, "grad_norm": 2.073918581008911, "learning_rate": 6.161332396569375e-07, "loss": 0.08536555171012879, "step": 43560 }, { "epoch": 0.8912800004091256, "grad_norm": 2.979947090148926, "learning_rate": 6.138456753446986e-07, "loss": 0.08994355201721191, "step": 43570 }, { "epoch": 0.8914845631817694, "grad_norm": 2.4962995052337646, "learning_rate": 6.115622310679336e-07, "loss": 0.1016700267791748, "step": 43580 }, { "epoch": 0.8916891259544132, "grad_norm": 2.0486297607421875, "learning_rate": 6.092829078289564e-07, "loss": 0.08646578788757324, "step": 43590 }, { "epoch": 0.891893688727057, "grad_norm": 2.21398663520813, "learning_rate": 6.070077066282687e-07, "loss": 0.08466529846191406, "step": 43600 }, { "epoch": 0.8920982514997008, "grad_norm": 1.9971749782562256, "learning_rate": 6.047366284645706e-07, "loss": 0.0895429790019989, "step": 43610 }, { "epoch": 0.8923028142723447, "grad_norm": 3.1059670448303223, "learning_rate": 6.024696743347458e-07, "loss": 0.09902310967445374, "step": 43620 }, { "epoch": 0.8925073770449885, "grad_norm": 6.552926063537598, "learning_rate": 6.002068452338694e-07, "loss": 0.09921208620071412, "step": 43630 }, { "epoch": 0.8927119398176323, "grad_norm": 2.5200629234313965, "learning_rate": 5.979481421552058e-07, "loss": 0.08947327136993408, "step": 43640 }, { "epoch": 0.8929165025902761, "grad_norm": 2.0755531787872314, "learning_rate": 5.956935660902119e-07, "loss": 0.08905205726623536, "step": 43650 }, { "epoch": 0.8931210653629199, "grad_norm": 2.926283121109009, "learning_rate": 5.934431180285261e-07, "loss": 0.09932542443275452, "step": 43660 }, { "epoch": 0.8933256281355637, "grad_norm": 1.9908461570739746, "learning_rate": 5.911967989579792e-07, "loss": 0.10339906215667724, "step": 43670 }, { "epoch": 0.8935301909082076, "grad_norm": 2.1166858673095703, "learning_rate": 5.889546098645882e-07, "loss": 0.07973402738571167, "step": 43680 }, { "epoch": 0.8937347536808514, "grad_norm": 1.4859466552734375, "learning_rate": 5.867165517325602e-07, "loss": 0.09361168146133422, "step": 43690 }, { "epoch": 0.8939393164534952, "grad_norm": 2.893933057785034, "learning_rate": 5.84482625544286e-07, "loss": 0.09643532037734985, "step": 43700 }, { "epoch": 0.894143879226139, "grad_norm": 2.4281818866729736, "learning_rate": 5.822528322803411e-07, "loss": 0.09591342806816101, "step": 43710 }, { "epoch": 0.8943484419987828, "grad_norm": 1.9627232551574707, "learning_rate": 5.800271729194939e-07, "loss": 0.08359389305114746, "step": 43720 }, { "epoch": 0.8945530047714266, "grad_norm": 1.7499762773513794, "learning_rate": 5.778056484386918e-07, "loss": 0.09286352396011352, "step": 43730 }, { "epoch": 0.8947575675440705, "grad_norm": 2.9771194458007812, "learning_rate": 5.755882598130658e-07, "loss": 0.0989766538143158, "step": 43740 }, { "epoch": 0.8949621303167143, "grad_norm": 3.947394609451294, "learning_rate": 5.733750080159361e-07, "loss": 0.11068743467330933, "step": 43750 }, { "epoch": 0.8951666930893581, "grad_norm": 2.0900466442108154, "learning_rate": 5.71165894018808e-07, "loss": 0.10655286312103271, "step": 43760 }, { "epoch": 0.895371255862002, "grad_norm": 2.8817076683044434, "learning_rate": 5.68960918791367e-07, "loss": 0.0993448793888092, "step": 43770 }, { "epoch": 0.8955758186346457, "grad_norm": 3.726990222930908, "learning_rate": 5.66760083301482e-07, "loss": 0.09601417779922486, "step": 43780 }, { "epoch": 0.8957803814072895, "grad_norm": 2.907583713531494, "learning_rate": 5.645633885152046e-07, "loss": 0.0934243619441986, "step": 43790 }, { "epoch": 0.8959849441799335, "grad_norm": 2.7383389472961426, "learning_rate": 5.623708353967738e-07, "loss": 0.111226487159729, "step": 43800 }, { "epoch": 0.8961895069525773, "grad_norm": 1.8614428043365479, "learning_rate": 5.601824249086041e-07, "loss": 0.09640430212020874, "step": 43810 }, { "epoch": 0.8963940697252211, "grad_norm": 3.282627582550049, "learning_rate": 5.579981580112947e-07, "loss": 0.09795597195625305, "step": 43820 }, { "epoch": 0.8965986324978649, "grad_norm": 2.123814821243286, "learning_rate": 5.558180356636245e-07, "loss": 0.08741587400436401, "step": 43830 }, { "epoch": 0.8968031952705087, "grad_norm": 2.9851911067962646, "learning_rate": 5.536420588225566e-07, "loss": 0.09323046207427979, "step": 43840 }, { "epoch": 0.8970077580431525, "grad_norm": 2.6582961082458496, "learning_rate": 5.514702284432316e-07, "loss": 0.08174501657485962, "step": 43850 }, { "epoch": 0.8972123208157964, "grad_norm": 3.3474833965301514, "learning_rate": 5.49302545478968e-07, "loss": 0.09716386795043945, "step": 43860 }, { "epoch": 0.8974168835884402, "grad_norm": 3.055326461791992, "learning_rate": 5.471390108812713e-07, "loss": 0.08757844567298889, "step": 43870 }, { "epoch": 0.897621446361084, "grad_norm": 4.335091590881348, "learning_rate": 5.449796255998175e-07, "loss": 0.12648370265960693, "step": 43880 }, { "epoch": 0.8978260091337278, "grad_norm": 2.8681557178497314, "learning_rate": 5.428243905824648e-07, "loss": 0.10894155502319336, "step": 43890 }, { "epoch": 0.8980305719063716, "grad_norm": 3.1283011436462402, "learning_rate": 5.406733067752479e-07, "loss": 0.10253826379776002, "step": 43900 }, { "epoch": 0.8982351346790154, "grad_norm": 2.1265180110931396, "learning_rate": 5.385263751223868e-07, "loss": 0.0830946147441864, "step": 43910 }, { "epoch": 0.8984396974516593, "grad_norm": 3.0862972736358643, "learning_rate": 5.363835965662689e-07, "loss": 0.09221752882003784, "step": 43920 }, { "epoch": 0.8986442602243031, "grad_norm": 2.866511583328247, "learning_rate": 5.342449720474641e-07, "loss": 0.09418004751205444, "step": 43930 }, { "epoch": 0.8988488229969469, "grad_norm": 2.496271848678589, "learning_rate": 5.32110502504719e-07, "loss": 0.08617280721664429, "step": 43940 }, { "epoch": 0.8990533857695907, "grad_norm": 3.225020408630371, "learning_rate": 5.299801888749534e-07, "loss": 0.09835340976715087, "step": 43950 }, { "epoch": 0.8992579485422345, "grad_norm": 3.9674620628356934, "learning_rate": 5.278540320932645e-07, "loss": 0.09547508358955384, "step": 43960 }, { "epoch": 0.8994625113148783, "grad_norm": 3.5495615005493164, "learning_rate": 5.257320330929261e-07, "loss": 0.09023234844207764, "step": 43970 }, { "epoch": 0.8996670740875222, "grad_norm": 1.94346022605896, "learning_rate": 5.23614192805384e-07, "loss": 0.08902382850646973, "step": 43980 }, { "epoch": 0.899871636860166, "grad_norm": 2.1993026733398438, "learning_rate": 5.215005121602635e-07, "loss": 0.09300599694252014, "step": 43990 }, { "epoch": 0.9000761996328098, "grad_norm": 3.205775737762451, "learning_rate": 5.19390992085359e-07, "loss": 0.08960033059120179, "step": 44000 }, { "epoch": 0.9002807624054536, "grad_norm": 2.158900499343872, "learning_rate": 5.172856335066412e-07, "loss": 0.08911086320877075, "step": 44010 }, { "epoch": 0.9004853251780974, "grad_norm": 2.539487600326538, "learning_rate": 5.151844373482517e-07, "loss": 0.10808675289154053, "step": 44020 }, { "epoch": 0.9006898879507412, "grad_norm": 2.657205581665039, "learning_rate": 5.130874045325085e-07, "loss": 0.0949118673801422, "step": 44030 }, { "epoch": 0.9008944507233851, "grad_norm": 2.702410936355591, "learning_rate": 5.109945359799007e-07, "loss": 0.08897305727005005, "step": 44040 }, { "epoch": 0.9010990134960289, "grad_norm": 3.6894378662109375, "learning_rate": 5.089058326090856e-07, "loss": 0.09165235161781311, "step": 44050 }, { "epoch": 0.9013035762686727, "grad_norm": 2.424349308013916, "learning_rate": 5.068212953369001e-07, "loss": 0.10099287033081054, "step": 44060 }, { "epoch": 0.9015081390413165, "grad_norm": 2.2836413383483887, "learning_rate": 5.047409250783475e-07, "loss": 0.09301097393035888, "step": 44070 }, { "epoch": 0.9017127018139603, "grad_norm": 2.5848047733306885, "learning_rate": 5.026647227466008e-07, "loss": 0.09270010590553283, "step": 44080 }, { "epoch": 0.9019172645866043, "grad_norm": 3.767608642578125, "learning_rate": 5.005926892530055e-07, "loss": 0.0867903470993042, "step": 44090 }, { "epoch": 0.9021218273592481, "grad_norm": 2.1569018363952637, "learning_rate": 4.985248255070785e-07, "loss": 0.08944818377494812, "step": 44100 }, { "epoch": 0.9023263901318919, "grad_norm": 3.551694393157959, "learning_rate": 4.964611324165025e-07, "loss": 0.09632312059402466, "step": 44110 }, { "epoch": 0.9025309529045357, "grad_norm": 2.9304769039154053, "learning_rate": 4.944016108871341e-07, "loss": 0.10036134719848633, "step": 44120 }, { "epoch": 0.9027355156771795, "grad_norm": 2.6196837425231934, "learning_rate": 4.923462618229935e-07, "loss": 0.09516656994819642, "step": 44130 }, { "epoch": 0.9029400784498233, "grad_norm": 2.7835423946380615, "learning_rate": 4.902950861262768e-07, "loss": 0.08830618858337402, "step": 44140 }, { "epoch": 0.9031446412224672, "grad_norm": 3.1890554428100586, "learning_rate": 4.882480846973415e-07, "loss": 0.08620677590370178, "step": 44150 }, { "epoch": 0.903349203995111, "grad_norm": 2.355515480041504, "learning_rate": 4.862052584347144e-07, "loss": 0.09461070895195008, "step": 44160 }, { "epoch": 0.9035537667677548, "grad_norm": 2.008028030395508, "learning_rate": 4.841666082350915e-07, "loss": 0.08919670581817626, "step": 44170 }, { "epoch": 0.9037583295403986, "grad_norm": 2.347405195236206, "learning_rate": 4.82132134993335e-07, "loss": 0.10651764869689942, "step": 44180 }, { "epoch": 0.9039628923130424, "grad_norm": 1.6580125093460083, "learning_rate": 4.801018396024726e-07, "loss": 0.08421536087989807, "step": 44190 }, { "epoch": 0.9041674550856862, "grad_norm": 1.9383140802383423, "learning_rate": 4.780757229536981e-07, "loss": 0.09771100282669068, "step": 44200 }, { "epoch": 0.9043720178583301, "grad_norm": 2.3421294689178467, "learning_rate": 4.760537859363723e-07, "loss": 0.09540545344352722, "step": 44210 }, { "epoch": 0.9045765806309739, "grad_norm": 2.5857927799224854, "learning_rate": 4.7403602943802194e-07, "loss": 0.09562551975250244, "step": 44220 }, { "epoch": 0.9047811434036177, "grad_norm": 2.8809478282928467, "learning_rate": 4.7202245434433633e-07, "loss": 0.0977371096611023, "step": 44230 }, { "epoch": 0.9049857061762615, "grad_norm": 1.9760634899139404, "learning_rate": 4.700130615391707e-07, "loss": 0.09258942604064942, "step": 44240 }, { "epoch": 0.9051902689489053, "grad_norm": 2.1034302711486816, "learning_rate": 4.68007851904545e-07, "loss": 0.09784966707229614, "step": 44250 }, { "epoch": 0.9053948317215491, "grad_norm": 2.0181407928466797, "learning_rate": 4.660068263206419e-07, "loss": 0.09191895127296448, "step": 44260 }, { "epoch": 0.905599394494193, "grad_norm": 2.1525793075561523, "learning_rate": 4.640099856658076e-07, "loss": 0.09176376461982727, "step": 44270 }, { "epoch": 0.9058039572668368, "grad_norm": 2.4318759441375732, "learning_rate": 4.62017330816551e-07, "loss": 0.08996669054031373, "step": 44280 }, { "epoch": 0.9060085200394806, "grad_norm": 2.8934032917022705, "learning_rate": 4.6002886264754687e-07, "loss": 0.10478767156600952, "step": 44290 }, { "epoch": 0.9062130828121244, "grad_norm": 2.459003210067749, "learning_rate": 4.5804458203162815e-07, "loss": 0.10168789625167847, "step": 44300 }, { "epoch": 0.9064176455847682, "grad_norm": 2.279198408126831, "learning_rate": 4.5606448983979256e-07, "loss": 0.09920179843902588, "step": 44310 }, { "epoch": 0.906622208357412, "grad_norm": 3.941563844680786, "learning_rate": 4.54088586941196e-07, "loss": 0.08869248628616333, "step": 44320 }, { "epoch": 0.9068267711300559, "grad_norm": 2.4051706790924072, "learning_rate": 4.521168742031579e-07, "loss": 0.0936751663684845, "step": 44330 }, { "epoch": 0.9070313339026997, "grad_norm": 3.0084588527679443, "learning_rate": 4.5014935249115933e-07, "loss": 0.09836254119873047, "step": 44340 }, { "epoch": 0.9072358966753435, "grad_norm": 2.5936625003814697, "learning_rate": 4.4818602266883835e-07, "loss": 0.07532877922058105, "step": 44350 }, { "epoch": 0.9074404594479873, "grad_norm": 2.222386598587036, "learning_rate": 4.4622688559799565e-07, "loss": 0.10045346021652221, "step": 44360 }, { "epoch": 0.9076450222206311, "grad_norm": 2.885936737060547, "learning_rate": 4.4427194213859216e-07, "loss": 0.098591148853302, "step": 44370 }, { "epoch": 0.907849584993275, "grad_norm": 3.330799102783203, "learning_rate": 4.4232119314874697e-07, "loss": 0.0968855857849121, "step": 44380 }, { "epoch": 0.9080541477659189, "grad_norm": 2.39896821975708, "learning_rate": 4.403746394847364e-07, "loss": 0.08816602826118469, "step": 44390 }, { "epoch": 0.9082587105385627, "grad_norm": 2.409266233444214, "learning_rate": 4.3843228200099675e-07, "loss": 0.09660599231719971, "step": 44400 }, { "epoch": 0.9084632733112065, "grad_norm": 3.429788112640381, "learning_rate": 4.3649412155012284e-07, "loss": 0.1035139799118042, "step": 44410 }, { "epoch": 0.9086678360838503, "grad_norm": 2.6331560611724854, "learning_rate": 4.345601589828674e-07, "loss": 0.0933738112449646, "step": 44420 }, { "epoch": 0.9088723988564941, "grad_norm": 2.1611106395721436, "learning_rate": 4.32630395148137e-07, "loss": 0.08663521409034729, "step": 44430 }, { "epoch": 0.9090769616291379, "grad_norm": 2.0559427738189697, "learning_rate": 4.3070483089300175e-07, "loss": 0.10088460445404053, "step": 44440 }, { "epoch": 0.9092815244017818, "grad_norm": 3.1601929664611816, "learning_rate": 4.2878346706268227e-07, "loss": 0.08831902146339417, "step": 44450 }, { "epoch": 0.9094860871744256, "grad_norm": 2.421583652496338, "learning_rate": 4.2686630450055835e-07, "loss": 0.08835607171058654, "step": 44460 }, { "epoch": 0.9096906499470694, "grad_norm": 2.5732171535491943, "learning_rate": 4.249533440481646e-07, "loss": 0.10166344642639161, "step": 44470 }, { "epoch": 0.9098952127197132, "grad_norm": 2.091740846633911, "learning_rate": 4.2304458654519267e-07, "loss": 0.0920646071434021, "step": 44480 }, { "epoch": 0.910099775492357, "grad_norm": 2.0394680500030518, "learning_rate": 4.211400328294868e-07, "loss": 0.10067710876464844, "step": 44490 }, { "epoch": 0.9103043382650008, "grad_norm": 3.1078124046325684, "learning_rate": 4.192396837370494e-07, "loss": 0.08957197666168212, "step": 44500 }, { "epoch": 0.9105089010376447, "grad_norm": 2.5759830474853516, "learning_rate": 4.173435401020309e-07, "loss": 0.08839917778968812, "step": 44510 }, { "epoch": 0.9107134638102885, "grad_norm": 2.165893077850342, "learning_rate": 4.1545160275674566e-07, "loss": 0.0883021354675293, "step": 44520 }, { "epoch": 0.9109180265829323, "grad_norm": 1.9231550693511963, "learning_rate": 4.1356387253165375e-07, "loss": 0.08719404339790345, "step": 44530 }, { "epoch": 0.9111225893555761, "grad_norm": 3.4922473430633545, "learning_rate": 4.116803502553712e-07, "loss": 0.08586718440055847, "step": 44540 }, { "epoch": 0.9113271521282199, "grad_norm": 2.241591453552246, "learning_rate": 4.0980103675466456e-07, "loss": 0.09663833379745483, "step": 44550 }, { "epoch": 0.9115317149008638, "grad_norm": 3.41910457611084, "learning_rate": 4.079259328544571e-07, "loss": 0.09790133833885192, "step": 44560 }, { "epoch": 0.9117362776735076, "grad_norm": 2.666214942932129, "learning_rate": 4.060550393778229e-07, "loss": 0.0988755464553833, "step": 44570 }, { "epoch": 0.9119408404461514, "grad_norm": 2.1307175159454346, "learning_rate": 4.04188357145987e-07, "loss": 0.0858331561088562, "step": 44580 }, { "epoch": 0.9121454032187952, "grad_norm": 2.2588491439819336, "learning_rate": 4.023258869783231e-07, "loss": 0.09588480591773987, "step": 44590 }, { "epoch": 0.912349965991439, "grad_norm": 1.9901018142700195, "learning_rate": 4.0046762969236374e-07, "loss": 0.08460821509361267, "step": 44600 }, { "epoch": 0.9125545287640828, "grad_norm": 1.985503911972046, "learning_rate": 3.9861358610378544e-07, "loss": 0.10644452571868897, "step": 44610 }, { "epoch": 0.9127590915367267, "grad_norm": 2.7230231761932373, "learning_rate": 3.9676375702641714e-07, "loss": 0.0753018856048584, "step": 44620 }, { "epoch": 0.9129636543093705, "grad_norm": 2.560096502304077, "learning_rate": 3.949181432722382e-07, "loss": 0.10446081161499024, "step": 44630 }, { "epoch": 0.9131682170820143, "grad_norm": 2.86299467086792, "learning_rate": 3.930767456513784e-07, "loss": 0.08559950590133666, "step": 44640 }, { "epoch": 0.9133727798546581, "grad_norm": 3.614875555038452, "learning_rate": 3.9123956497211346e-07, "loss": 0.085133957862854, "step": 44650 }, { "epoch": 0.913577342627302, "grad_norm": 2.0381052494049072, "learning_rate": 3.8940660204087156e-07, "loss": 0.09466844201087951, "step": 44660 }, { "epoch": 0.9137819053999457, "grad_norm": 2.394942045211792, "learning_rate": 3.875778576622302e-07, "loss": 0.09604616761207581, "step": 44670 }, { "epoch": 0.9139864681725897, "grad_norm": 2.2925477027893066, "learning_rate": 3.857533326389118e-07, "loss": 0.09709187746047973, "step": 44680 }, { "epoch": 0.9141910309452335, "grad_norm": 1.7934221029281616, "learning_rate": 3.83933027771789e-07, "loss": 0.08450474739074706, "step": 44690 }, { "epoch": 0.9143955937178773, "grad_norm": 2.747802972793579, "learning_rate": 3.8211694385987927e-07, "loss": 0.10055309534072876, "step": 44700 }, { "epoch": 0.9146001564905211, "grad_norm": 4.357390880584717, "learning_rate": 3.8030508170035283e-07, "loss": 0.09296976327896118, "step": 44710 }, { "epoch": 0.9148047192631649, "grad_norm": 2.8114914894104004, "learning_rate": 3.7849744208852124e-07, "loss": 0.09468072652816772, "step": 44720 }, { "epoch": 0.9150092820358087, "grad_norm": 2.318570137023926, "learning_rate": 3.766940258178442e-07, "loss": 0.08972617387771606, "step": 44730 }, { "epoch": 0.9152138448084526, "grad_norm": 3.743408679962158, "learning_rate": 3.7489483367992854e-07, "loss": 0.09382185935974122, "step": 44740 }, { "epoch": 0.9154184075810964, "grad_norm": 2.375697374343872, "learning_rate": 3.730998664645269e-07, "loss": 0.0936055064201355, "step": 44750 }, { "epoch": 0.9156229703537402, "grad_norm": 2.6611130237579346, "learning_rate": 3.71309124959538e-07, "loss": 0.08822187185287475, "step": 44760 }, { "epoch": 0.915827533126384, "grad_norm": 2.9032578468322754, "learning_rate": 3.69522609951003e-07, "loss": 0.09476357698440552, "step": 44770 }, { "epoch": 0.9160320958990278, "grad_norm": 1.9842578172683716, "learning_rate": 3.677403222231068e-07, "loss": 0.09701523184776306, "step": 44780 }, { "epoch": 0.9162366586716716, "grad_norm": 2.524489402770996, "learning_rate": 3.65962262558186e-07, "loss": 0.09474073648452759, "step": 44790 }, { "epoch": 0.9164412214443155, "grad_norm": 3.8114097118377686, "learning_rate": 3.64188431736715e-07, "loss": 0.099583899974823, "step": 44800 }, { "epoch": 0.9166457842169593, "grad_norm": 2.7971670627593994, "learning_rate": 3.624188305373122e-07, "loss": 0.10230076313018799, "step": 44810 }, { "epoch": 0.9168503469896031, "grad_norm": 1.7568837404251099, "learning_rate": 3.606534597367439e-07, "loss": 0.09799272418022156, "step": 44820 }, { "epoch": 0.9170549097622469, "grad_norm": 3.839010238647461, "learning_rate": 3.588923201099148e-07, "loss": 0.1040876865386963, "step": 44830 }, { "epoch": 0.9172594725348907, "grad_norm": 1.9870631694793701, "learning_rate": 3.5713541242987426e-07, "loss": 0.09368621706962585, "step": 44840 }, { "epoch": 0.9174640353075345, "grad_norm": 2.030782461166382, "learning_rate": 3.553827374678109e-07, "loss": 0.08436415195465088, "step": 44850 }, { "epoch": 0.9176685980801784, "grad_norm": 3.3590660095214844, "learning_rate": 3.536342959930616e-07, "loss": 0.08877157568931579, "step": 44860 }, { "epoch": 0.9178731608528222, "grad_norm": 2.5618064403533936, "learning_rate": 3.518900887731014e-07, "loss": 0.09506987929344177, "step": 44870 }, { "epoch": 0.918077723625466, "grad_norm": 2.2263293266296387, "learning_rate": 3.501501165735455e-07, "loss": 0.09504585266113282, "step": 44880 }, { "epoch": 0.9182822863981098, "grad_norm": 3.0498414039611816, "learning_rate": 3.484143801581508e-07, "loss": 0.08869723081588746, "step": 44890 }, { "epoch": 0.9184868491707536, "grad_norm": 1.946256160736084, "learning_rate": 3.466828802888189e-07, "loss": 0.09803326725959778, "step": 44900 }, { "epoch": 0.9186914119433974, "grad_norm": 3.0589261054992676, "learning_rate": 3.449556177255875e-07, "loss": 0.08990461230278016, "step": 44910 }, { "epoch": 0.9188959747160413, "grad_norm": 1.6130075454711914, "learning_rate": 3.4323259322663226e-07, "loss": 0.09234728217124939, "step": 44920 }, { "epoch": 0.9191005374886851, "grad_norm": 2.2249622344970703, "learning_rate": 3.4151380754827377e-07, "loss": 0.08873885869979858, "step": 44930 }, { "epoch": 0.919305100261329, "grad_norm": 3.8013906478881836, "learning_rate": 3.397992614449719e-07, "loss": 0.09535343647003174, "step": 44940 }, { "epoch": 0.9195096630339727, "grad_norm": 2.354323387145996, "learning_rate": 3.380889556693212e-07, "loss": 0.09674569964408875, "step": 44950 }, { "epoch": 0.9197142258066165, "grad_norm": 2.473215341567993, "learning_rate": 3.363828909720568e-07, "loss": 0.11852185726165772, "step": 44960 }, { "epoch": 0.9199187885792603, "grad_norm": 2.3816585540771484, "learning_rate": 3.34681068102054e-07, "loss": 0.08742247819900513, "step": 44970 }, { "epoch": 0.9201233513519043, "grad_norm": 2.436234474182129, "learning_rate": 3.329834878063265e-07, "loss": 0.08990947604179382, "step": 44980 }, { "epoch": 0.9203279141245481, "grad_norm": 2.6652743816375732, "learning_rate": 3.3129015083002147e-07, "loss": 0.08298628330230713, "step": 44990 }, { "epoch": 0.9205324768971919, "grad_norm": 2.275831937789917, "learning_rate": 3.296010579164255e-07, "loss": 0.0933750331401825, "step": 45000 }, { "epoch": 0.9207370396698357, "grad_norm": 3.5867228507995605, "learning_rate": 3.2791620980696437e-07, "loss": 0.09443082213401795, "step": 45010 }, { "epoch": 0.9209416024424795, "grad_norm": 3.533769130706787, "learning_rate": 3.2623560724119987e-07, "loss": 0.09819875359535217, "step": 45020 }, { "epoch": 0.9211461652151234, "grad_norm": 2.9582035541534424, "learning_rate": 3.245592509568285e-07, "loss": 0.08736634254455566, "step": 45030 }, { "epoch": 0.9213507279877672, "grad_norm": 2.6238057613372803, "learning_rate": 3.228871416896828e-07, "loss": 0.07979904413223267, "step": 45040 }, { "epoch": 0.921555290760411, "grad_norm": 2.3571479320526123, "learning_rate": 3.212192801737357e-07, "loss": 0.09613209962844849, "step": 45050 }, { "epoch": 0.9217598535330548, "grad_norm": 2.446254253387451, "learning_rate": 3.195556671410882e-07, "loss": 0.0701751708984375, "step": 45060 }, { "epoch": 0.9219644163056986, "grad_norm": 3.0160114765167236, "learning_rate": 3.178963033219806e-07, "loss": 0.1046443223953247, "step": 45070 }, { "epoch": 0.9221689790783424, "grad_norm": 3.3507094383239746, "learning_rate": 3.1624118944478923e-07, "loss": 0.09455388188362121, "step": 45080 }, { "epoch": 0.9223735418509863, "grad_norm": 2.8879175186157227, "learning_rate": 3.145903262360228e-07, "loss": 0.100706946849823, "step": 45090 }, { "epoch": 0.9225781046236301, "grad_norm": 2.826775312423706, "learning_rate": 3.1294371442032403e-07, "loss": 0.1016754150390625, "step": 45100 }, { "epoch": 0.9227826673962739, "grad_norm": 3.6052722930908203, "learning_rate": 3.1130135472047127e-07, "loss": 0.09476290345191955, "step": 45110 }, { "epoch": 0.9229872301689177, "grad_norm": 3.51741361618042, "learning_rate": 3.096632478573747e-07, "loss": 0.10828909873962403, "step": 45120 }, { "epoch": 0.9231917929415615, "grad_norm": 3.4520864486694336, "learning_rate": 3.08029394550079e-07, "loss": 0.08551549315452575, "step": 45130 }, { "epoch": 0.9233963557142053, "grad_norm": 3.3283908367156982, "learning_rate": 3.0639979551576047e-07, "loss": 0.09631754159927368, "step": 45140 }, { "epoch": 0.9236009184868492, "grad_norm": 3.129732847213745, "learning_rate": 3.0477445146972815e-07, "loss": 0.1049575686454773, "step": 45150 }, { "epoch": 0.923805481259493, "grad_norm": 1.9000849723815918, "learning_rate": 3.0315336312542465e-07, "loss": 0.09336387515068054, "step": 45160 }, { "epoch": 0.9240100440321368, "grad_norm": 1.8216203451156616, "learning_rate": 3.015365311944241e-07, "loss": 0.09113326668739319, "step": 45170 }, { "epoch": 0.9242146068047806, "grad_norm": 5.174433708190918, "learning_rate": 2.99923956386432e-07, "loss": 0.09970353245735168, "step": 45180 }, { "epoch": 0.9244191695774244, "grad_norm": 2.4007601737976074, "learning_rate": 2.983156394092845e-07, "loss": 0.09181075096130371, "step": 45190 }, { "epoch": 0.9246237323500682, "grad_norm": 2.4699878692626953, "learning_rate": 2.9671158096894913e-07, "loss": 0.08734872937202454, "step": 45200 }, { "epoch": 0.9248282951227121, "grad_norm": 2.570117712020874, "learning_rate": 2.9511178176952594e-07, "loss": 0.08907262086868287, "step": 45210 }, { "epoch": 0.9250328578953559, "grad_norm": 3.0108911991119385, "learning_rate": 2.9351624251324206e-07, "loss": 0.09638167023658753, "step": 45220 }, { "epoch": 0.9252374206679997, "grad_norm": 2.8208563327789307, "learning_rate": 2.919249639004562e-07, "loss": 0.0869291067123413, "step": 45230 }, { "epoch": 0.9254419834406435, "grad_norm": 3.868539810180664, "learning_rate": 2.9033794662965963e-07, "loss": 0.10114855766296386, "step": 45240 }, { "epoch": 0.9256465462132873, "grad_norm": 2.078955888748169, "learning_rate": 2.8875519139746955e-07, "loss": 0.09462754130363464, "step": 45250 }, { "epoch": 0.9258511089859311, "grad_norm": 2.1542158126831055, "learning_rate": 2.8717669889863355e-07, "loss": 0.09763721227645875, "step": 45260 }, { "epoch": 0.9260556717585751, "grad_norm": 2.4588449001312256, "learning_rate": 2.8560246982602735e-07, "loss": 0.08588069677352905, "step": 45270 }, { "epoch": 0.9262602345312189, "grad_norm": 1.827466607093811, "learning_rate": 2.8403250487065713e-07, "loss": 0.09077121019363403, "step": 45280 }, { "epoch": 0.9264647973038627, "grad_norm": 2.4959774017333984, "learning_rate": 2.824668047216561e-07, "loss": 0.08934874534606933, "step": 45290 }, { "epoch": 0.9266693600765065, "grad_norm": 2.315190076828003, "learning_rate": 2.809053700662845e-07, "loss": 0.08569937348365783, "step": 45300 }, { "epoch": 0.9268739228491503, "grad_norm": 3.6766982078552246, "learning_rate": 2.793482015899307e-07, "loss": 0.09746639132499695, "step": 45310 }, { "epoch": 0.9270784856217941, "grad_norm": 2.8994429111480713, "learning_rate": 2.777952999761124e-07, "loss": 0.08036257028579712, "step": 45320 }, { "epoch": 0.927283048394438, "grad_norm": 3.4301586151123047, "learning_rate": 2.762466659064733e-07, "loss": 0.10223796367645263, "step": 45330 }, { "epoch": 0.9274876111670818, "grad_norm": 2.3661961555480957, "learning_rate": 2.747023000607829e-07, "loss": 0.10133639574050904, "step": 45340 }, { "epoch": 0.9276921739397256, "grad_norm": 3.4962875843048096, "learning_rate": 2.7316220311693786e-07, "loss": 0.10135078430175781, "step": 45350 }, { "epoch": 0.9278967367123694, "grad_norm": 2.3293659687042236, "learning_rate": 2.7162637575096076e-07, "loss": 0.10309020280838013, "step": 45360 }, { "epoch": 0.9281012994850132, "grad_norm": 2.8563404083251953, "learning_rate": 2.70094818637e-07, "loss": 0.08936557769775391, "step": 45370 }, { "epoch": 0.928305862257657, "grad_norm": 1.9051976203918457, "learning_rate": 2.6856753244732913e-07, "loss": 0.09674990177154541, "step": 45380 }, { "epoch": 0.9285104250303009, "grad_norm": 2.0146021842956543, "learning_rate": 2.670445178523495e-07, "loss": 0.08910205960273743, "step": 45390 }, { "epoch": 0.9287149878029447, "grad_norm": 3.4236834049224854, "learning_rate": 2.6552577552058533e-07, "loss": 0.09387630820274354, "step": 45400 }, { "epoch": 0.9289195505755885, "grad_norm": 2.5568971633911133, "learning_rate": 2.6401130611868553e-07, "loss": 0.08911013603210449, "step": 45410 }, { "epoch": 0.9291241133482323, "grad_norm": 3.4704737663269043, "learning_rate": 2.6250111031142276e-07, "loss": 0.0977121651172638, "step": 45420 }, { "epoch": 0.9293286761208761, "grad_norm": 3.178131103515625, "learning_rate": 2.609951887616946e-07, "loss": 0.0922701597213745, "step": 45430 }, { "epoch": 0.9295332388935199, "grad_norm": 2.467193841934204, "learning_rate": 2.594935421305245e-07, "loss": 0.08748127222061157, "step": 45440 }, { "epoch": 0.9297378016661638, "grad_norm": 2.7409417629241943, "learning_rate": 2.579961710770562e-07, "loss": 0.09152541756629944, "step": 45450 }, { "epoch": 0.9299423644388076, "grad_norm": 2.518768072128296, "learning_rate": 2.5650307625855743e-07, "loss": 0.1044304609298706, "step": 45460 }, { "epoch": 0.9301469272114514, "grad_norm": 3.7438881397247314, "learning_rate": 2.550142583304216e-07, "loss": 0.0925140619277954, "step": 45470 }, { "epoch": 0.9303514899840952, "grad_norm": 2.18361234664917, "learning_rate": 2.535297179461604e-07, "loss": 0.09567705392837525, "step": 45480 }, { "epoch": 0.930556052756739, "grad_norm": 2.5896694660186768, "learning_rate": 2.520494557574116e-07, "loss": 0.09975137114524842, "step": 45490 }, { "epoch": 0.9307606155293829, "grad_norm": 2.4925968647003174, "learning_rate": 2.5057347241393303e-07, "loss": 0.09096682071685791, "step": 45500 }, { "epoch": 0.9309651783020267, "grad_norm": 2.1762783527374268, "learning_rate": 2.491017685636044e-07, "loss": 0.09061336517333984, "step": 45510 }, { "epoch": 0.9311697410746705, "grad_norm": 1.9913448095321655, "learning_rate": 2.4763434485242874e-07, "loss": 0.09811891317367553, "step": 45520 }, { "epoch": 0.9313743038473143, "grad_norm": 2.960387706756592, "learning_rate": 2.4617120192452773e-07, "loss": 0.08399608731269836, "step": 45530 }, { "epoch": 0.9315788666199581, "grad_norm": 2.8752825260162354, "learning_rate": 2.4471234042214634e-07, "loss": 0.08656484484672547, "step": 45540 }, { "epoch": 0.931783429392602, "grad_norm": 1.9587346315383911, "learning_rate": 2.4325776098564793e-07, "loss": 0.08840080499649047, "step": 45550 }, { "epoch": 0.9319879921652459, "grad_norm": 2.392794370651245, "learning_rate": 2.4180746425351953e-07, "loss": 0.09378288984298706, "step": 45560 }, { "epoch": 0.9321925549378897, "grad_norm": 1.7859488725662231, "learning_rate": 2.4036145086236417e-07, "loss": 0.08304296731948853, "step": 45570 }, { "epoch": 0.9323971177105335, "grad_norm": 2.3812806606292725, "learning_rate": 2.389197214469086e-07, "loss": 0.10390235185623169, "step": 45580 }, { "epoch": 0.9326016804831773, "grad_norm": 3.6194489002227783, "learning_rate": 2.3748227663999645e-07, "loss": 0.09303859472274781, "step": 45590 }, { "epoch": 0.9328062432558211, "grad_norm": 2.466953754425049, "learning_rate": 2.3604911707259093e-07, "loss": 0.08551293611526489, "step": 45600 }, { "epoch": 0.9330108060284649, "grad_norm": 2.9501025676727295, "learning_rate": 2.346202433737743e-07, "loss": 0.10444436073303223, "step": 45610 }, { "epoch": 0.9332153688011088, "grad_norm": 2.1651690006256104, "learning_rate": 2.3319565617075045e-07, "loss": 0.08600758314132691, "step": 45620 }, { "epoch": 0.9334199315737526, "grad_norm": 2.368302822113037, "learning_rate": 2.3177535608883805e-07, "loss": 0.08543329238891602, "step": 45630 }, { "epoch": 0.9336244943463964, "grad_norm": 1.7219672203063965, "learning_rate": 2.3035934375147617e-07, "loss": 0.07854313850402832, "step": 45640 }, { "epoch": 0.9338290571190402, "grad_norm": 2.1427290439605713, "learning_rate": 2.2894761978021985e-07, "loss": 0.09179819822311401, "step": 45650 }, { "epoch": 0.934033619891684, "grad_norm": 2.4353911876678467, "learning_rate": 2.2754018479474226e-07, "loss": 0.09995261430740357, "step": 45660 }, { "epoch": 0.9342381826643278, "grad_norm": 2.509415864944458, "learning_rate": 2.2613703941283482e-07, "loss": 0.09079781770706177, "step": 45670 }, { "epoch": 0.9344427454369717, "grad_norm": 2.3756086826324463, "learning_rate": 2.2473818425040704e-07, "loss": 0.0885625958442688, "step": 45680 }, { "epoch": 0.9346473082096155, "grad_norm": 3.89725923538208, "learning_rate": 2.2334361992148112e-07, "loss": 0.09526216983795166, "step": 45690 }, { "epoch": 0.9348518709822593, "grad_norm": 1.846510887145996, "learning_rate": 2.219533470382018e-07, "loss": 0.08059484958648681, "step": 45700 }, { "epoch": 0.9350564337549031, "grad_norm": 1.8619775772094727, "learning_rate": 2.205673662108243e-07, "loss": 0.09139471650123596, "step": 45710 }, { "epoch": 0.9352609965275469, "grad_norm": 2.2335205078125, "learning_rate": 2.1918567804772418e-07, "loss": 0.09401378035545349, "step": 45720 }, { "epoch": 0.9354655593001907, "grad_norm": 2.123922109603882, "learning_rate": 2.178082831553896e-07, "loss": 0.08929270505905151, "step": 45730 }, { "epoch": 0.9356701220728346, "grad_norm": 2.2778098583221436, "learning_rate": 2.1643518213842696e-07, "loss": 0.09001833200454712, "step": 45740 }, { "epoch": 0.9358746848454784, "grad_norm": 2.1443727016448975, "learning_rate": 2.1506637559955524e-07, "loss": 0.0904222548007965, "step": 45750 }, { "epoch": 0.9360792476181222, "grad_norm": 2.7124996185302734, "learning_rate": 2.1370186413960937e-07, "loss": 0.09591434001922608, "step": 45760 }, { "epoch": 0.936283810390766, "grad_norm": 2.070481061935425, "learning_rate": 2.1234164835754023e-07, "loss": 0.07881026864051818, "step": 45770 }, { "epoch": 0.9364883731634098, "grad_norm": 2.5273079872131348, "learning_rate": 2.1098572885041136e-07, "loss": 0.0823244571685791, "step": 45780 }, { "epoch": 0.9366929359360536, "grad_norm": 3.2763423919677734, "learning_rate": 2.096341062134022e-07, "loss": 0.09175297021865844, "step": 45790 }, { "epoch": 0.9368974987086975, "grad_norm": 2.647588014602661, "learning_rate": 2.082867810398037e-07, "loss": 0.08072956800460815, "step": 45800 }, { "epoch": 0.9371020614813413, "grad_norm": 3.3868727684020996, "learning_rate": 2.0694375392102284e-07, "loss": 0.09411004781723023, "step": 45810 }, { "epoch": 0.9373066242539851, "grad_norm": 2.6502487659454346, "learning_rate": 2.0560502544657912e-07, "loss": 0.08334428071975708, "step": 45820 }, { "epoch": 0.937511187026629, "grad_norm": 2.689103603363037, "learning_rate": 2.042705962041036e-07, "loss": 0.09128422141075135, "step": 45830 }, { "epoch": 0.9377157497992727, "grad_norm": 2.4865119457244873, "learning_rate": 2.029404667793411e-07, "loss": 0.09736162424087524, "step": 45840 }, { "epoch": 0.9379203125719165, "grad_norm": 2.8247010707855225, "learning_rate": 2.016146377561523e-07, "loss": 0.09151309728622437, "step": 45850 }, { "epoch": 0.9381248753445605, "grad_norm": 3.5275161266326904, "learning_rate": 2.0029310971650618e-07, "loss": 0.09304993152618408, "step": 45860 }, { "epoch": 0.9383294381172043, "grad_norm": 2.6610517501831055, "learning_rate": 1.9897588324048533e-07, "loss": 0.08963409662246705, "step": 45870 }, { "epoch": 0.9385340008898481, "grad_norm": 4.3371710777282715, "learning_rate": 1.9766295890628284e-07, "loss": 0.09067069888114929, "step": 45880 }, { "epoch": 0.9387385636624919, "grad_norm": 2.7824788093566895, "learning_rate": 1.9635433729020436e-07, "loss": 0.09332885146141053, "step": 45890 }, { "epoch": 0.9389431264351357, "grad_norm": 2.9782044887542725, "learning_rate": 1.950500189666693e-07, "loss": 0.10236377716064453, "step": 45900 }, { "epoch": 0.9391476892077795, "grad_norm": 2.9458954334259033, "learning_rate": 1.9375000450820191e-07, "loss": 0.08787844181060792, "step": 45910 }, { "epoch": 0.9393522519804234, "grad_norm": 2.1760549545288086, "learning_rate": 1.9245429448544572e-07, "loss": 0.09175267815589905, "step": 45920 }, { "epoch": 0.9395568147530672, "grad_norm": 2.261941909790039, "learning_rate": 1.911628894671469e-07, "loss": 0.0918358325958252, "step": 45930 }, { "epoch": 0.939761377525711, "grad_norm": 2.255453109741211, "learning_rate": 1.8987579002016865e-07, "loss": 0.09332352876663208, "step": 45940 }, { "epoch": 0.9399659402983548, "grad_norm": 3.0926361083984375, "learning_rate": 1.8859299670947574e-07, "loss": 0.08291418552398681, "step": 45950 }, { "epoch": 0.9401705030709986, "grad_norm": 3.1108295917510986, "learning_rate": 1.8731451009815328e-07, "loss": 0.0882340669631958, "step": 45960 }, { "epoch": 0.9403750658436425, "grad_norm": 3.1909983158111572, "learning_rate": 1.860403307473879e-07, "loss": 0.09990223050117493, "step": 45970 }, { "epoch": 0.9405796286162863, "grad_norm": 2.5921666622161865, "learning_rate": 1.8477045921647897e-07, "loss": 0.0831292986869812, "step": 45980 }, { "epoch": 0.9407841913889301, "grad_norm": 3.164383888244629, "learning_rate": 1.8350489606283382e-07, "loss": 0.0882460594177246, "step": 45990 }, { "epoch": 0.9409887541615739, "grad_norm": 2.4415340423583984, "learning_rate": 1.822436418419693e-07, "loss": 0.08096357583999633, "step": 46000 } ], "logging_steps": 10, "max_steps": 48885, "num_input_tokens_seen": 0, "num_train_epochs": 1, "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": 2.5060899022271047e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }