{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.1370319545187217, "eval_steps": 500, "global_step": 4000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.008276432857438443, "grad_norm": 0.0, "learning_rate": 0.0, "loss": 2.0924, "step": 10 }, { "epoch": 0.016552865714876887, "grad_norm": 17.15061378479004, "learning_rate": 5e-06, "loss": 2.0976, "step": 20 }, { "epoch": 0.024829298572315334, "grad_norm": 5.300135135650635, "learning_rate": 3e-05, "loss": 1.4782, "step": 30 }, { "epoch": 0.033105731429753774, "grad_norm": 2.799682855606079, "learning_rate": 4.9999994694145566e-05, "loss": 1.3569, "step": 40 }, { "epoch": 0.04138216428719222, "grad_norm": 2.3753180503845215, "learning_rate": 4.9999808989476856e-05, "loss": 1.3787, "step": 50 }, { "epoch": 0.04965859714463067, "grad_norm": 2.270704984664917, "learning_rate": 4.99993579943386e-05, "loss": 1.3241, "step": 60 }, { "epoch": 0.05793503000206911, "grad_norm": 3.3539905548095703, "learning_rate": 4.999864171351664e-05, "loss": 1.3225, "step": 70 }, { "epoch": 0.06621146285950755, "grad_norm": 3.482184410095215, "learning_rate": 4.999766015461193e-05, "loss": 1.3114, "step": 80 }, { "epoch": 0.074487895716946, "grad_norm": 4.038243770599365, "learning_rate": 4.999641332804046e-05, "loss": 1.2537, "step": 90 }, { "epoch": 0.08276432857438444, "grad_norm": 3.1982836723327637, "learning_rate": 4.999490124703319e-05, "loss": 1.2443, "step": 100 }, { "epoch": 0.09104076143182288, "grad_norm": 2.335792064666748, "learning_rate": 4.9993123927635874e-05, "loss": 1.2643, "step": 110 }, { "epoch": 0.09931719428926133, "grad_norm": 2.2586803436279297, "learning_rate": 4.99910813887089e-05, "loss": 1.2241, "step": 120 }, { "epoch": 0.10759362714669977, "grad_norm": 2.543644428253174, "learning_rate": 4.9988773651927064e-05, "loss": 1.2263, "step": 130 }, { "epoch": 0.11587006000413821, "grad_norm": 2.1221492290496826, "learning_rate": 4.998620074177938e-05, "loss": 1.2634, "step": 140 }, { "epoch": 0.12414649286157665, "grad_norm": 1.6611372232437134, "learning_rate": 4.998336268556879e-05, "loss": 1.2401, "step": 150 }, { "epoch": 0.1324229257190151, "grad_norm": 1.7521588802337646, "learning_rate": 4.998025951341191e-05, "loss": 1.242, "step": 160 }, { "epoch": 0.14069935857645355, "grad_norm": 1.445685625076294, "learning_rate": 4.997689125823867e-05, "loss": 1.2534, "step": 170 }, { "epoch": 0.148975791433892, "grad_norm": 1.8729900121688843, "learning_rate": 4.997325795579197e-05, "loss": 1.2583, "step": 180 }, { "epoch": 0.15725222429133043, "grad_norm": 1.734971284866333, "learning_rate": 4.9969359644627334e-05, "loss": 1.2543, "step": 190 }, { "epoch": 0.16552865714876888, "grad_norm": 1.8854798078536987, "learning_rate": 4.9965196366112464e-05, "loss": 1.2386, "step": 200 }, { "epoch": 0.17380509000620734, "grad_norm": 1.6403270959854126, "learning_rate": 4.996076816442683e-05, "loss": 1.2014, "step": 210 }, { "epoch": 0.18208152286364576, "grad_norm": 1.6998963356018066, "learning_rate": 4.995607508656118e-05, "loss": 1.2327, "step": 220 }, { "epoch": 0.19035795572108422, "grad_norm": 2.0543465614318848, "learning_rate": 4.995111718231704e-05, "loss": 1.215, "step": 230 }, { "epoch": 0.19863438857852267, "grad_norm": 2.155479907989502, "learning_rate": 4.9945894504306214e-05, "loss": 1.1843, "step": 240 }, { "epoch": 0.2069108214359611, "grad_norm": 1.2477573156356812, "learning_rate": 4.9940407107950175e-05, "loss": 1.171, "step": 250 }, { "epoch": 0.21518725429339955, "grad_norm": 1.8000760078430176, "learning_rate": 4.993465505147954e-05, "loss": 1.2036, "step": 260 }, { "epoch": 0.22346368715083798, "grad_norm": 1.8096606731414795, "learning_rate": 4.99286383959334e-05, "loss": 1.1794, "step": 270 }, { "epoch": 0.23174012000827643, "grad_norm": 1.412726879119873, "learning_rate": 4.9922357205158695e-05, "loss": 1.1787, "step": 280 }, { "epoch": 0.24001655286571488, "grad_norm": 1.3341314792633057, "learning_rate": 4.9915811545809557e-05, "loss": 1.1993, "step": 290 }, { "epoch": 0.2482929857231533, "grad_norm": 2.0190353393554688, "learning_rate": 4.990900148734653e-05, "loss": 1.1576, "step": 300 }, { "epoch": 0.2565694185805918, "grad_norm": 1.9342364072799683, "learning_rate": 4.9901927102035925e-05, "loss": 1.2081, "step": 310 }, { "epoch": 0.2648458514380302, "grad_norm": 2.2545788288116455, "learning_rate": 4.989458846494899e-05, "loss": 1.1935, "step": 320 }, { "epoch": 0.27312228429546864, "grad_norm": 1.927686095237732, "learning_rate": 4.988698565396115e-05, "loss": 1.1508, "step": 330 }, { "epoch": 0.2813987171529071, "grad_norm": 1.3246594667434692, "learning_rate": 4.987911874975114e-05, "loss": 1.138, "step": 340 }, { "epoch": 0.28967515001034555, "grad_norm": 1.9527952671051025, "learning_rate": 4.9870987835800194e-05, "loss": 1.2036, "step": 350 }, { "epoch": 0.297951582867784, "grad_norm": 1.1734226942062378, "learning_rate": 4.9862592998391125e-05, "loss": 1.1692, "step": 360 }, { "epoch": 0.3062280157252224, "grad_norm": 2.067718982696533, "learning_rate": 4.9853934326607434e-05, "loss": 1.1401, "step": 370 }, { "epoch": 0.31450444858266086, "grad_norm": 1.2361276149749756, "learning_rate": 4.984501191233235e-05, "loss": 1.1868, "step": 380 }, { "epoch": 0.3227808814400993, "grad_norm": 1.379475712776184, "learning_rate": 4.983582585024784e-05, "loss": 1.1458, "step": 390 }, { "epoch": 0.33105731429753776, "grad_norm": 1.45447838306427, "learning_rate": 4.982637623783366e-05, "loss": 1.1424, "step": 400 }, { "epoch": 0.3393337471549762, "grad_norm": 1.4039075374603271, "learning_rate": 4.981666317536624e-05, "loss": 1.1405, "step": 410 }, { "epoch": 0.34761018001241467, "grad_norm": 1.7842013835906982, "learning_rate": 4.980668676591769e-05, "loss": 1.1444, "step": 420 }, { "epoch": 0.35588661286985307, "grad_norm": 1.3413935899734497, "learning_rate": 4.9796447115354685e-05, "loss": 1.1256, "step": 430 }, { "epoch": 0.3641630457272915, "grad_norm": 1.3572813272476196, "learning_rate": 4.97859443323373e-05, "loss": 1.152, "step": 440 }, { "epoch": 0.37243947858473, "grad_norm": 1.4994474649429321, "learning_rate": 4.977517852831794e-05, "loss": 1.1264, "step": 450 }, { "epoch": 0.38071591144216843, "grad_norm": 1.2120469808578491, "learning_rate": 4.9764149817540074e-05, "loss": 1.1035, "step": 460 }, { "epoch": 0.3889923442996069, "grad_norm": 1.2979780435562134, "learning_rate": 4.975285831703707e-05, "loss": 1.1299, "step": 470 }, { "epoch": 0.39726877715704534, "grad_norm": 1.1150094270706177, "learning_rate": 4.974130414663094e-05, "loss": 1.1232, "step": 480 }, { "epoch": 0.40554521001448374, "grad_norm": 1.1811398267745972, "learning_rate": 4.9729487428931084e-05, "loss": 1.1106, "step": 490 }, { "epoch": 0.4138216428719222, "grad_norm": 1.285284399986267, "learning_rate": 4.971740828933295e-05, "loss": 1.1226, "step": 500 }, { "epoch": 0.4138216428719222, "eval_loss": 1.124096393585205, "eval_runtime": 494.2301, "eval_samples_per_second": 39.109, "eval_steps_per_second": 6.519, "step": 500 }, { "epoch": 0.42209807572936064, "grad_norm": 1.6036468744277954, "learning_rate": 4.970631279885502e-05, "loss": 1.1088, "step": 510 }, { "epoch": 0.4303745085867991, "grad_norm": 1.0415977239608765, "learning_rate": 4.969373541308714e-05, "loss": 1.1154, "step": 520 }, { "epoch": 0.43865094144423755, "grad_norm": 1.207888126373291, "learning_rate": 4.9680895984810636e-05, "loss": 1.1014, "step": 530 }, { "epoch": 0.44692737430167595, "grad_norm": 1.2686097621917725, "learning_rate": 4.966779465027368e-05, "loss": 1.1097, "step": 540 }, { "epoch": 0.4552038071591144, "grad_norm": 1.795076847076416, "learning_rate": 4.965443154850371e-05, "loss": 1.091, "step": 550 }, { "epoch": 0.46348024001655286, "grad_norm": 1.7811661958694458, "learning_rate": 4.964080682130594e-05, "loss": 1.0852, "step": 560 }, { "epoch": 0.4717566728739913, "grad_norm": 1.7507492303848267, "learning_rate": 4.962692061326189e-05, "loss": 1.0961, "step": 570 }, { "epoch": 0.48003310573142977, "grad_norm": 1.575803279876709, "learning_rate": 4.961277307172783e-05, "loss": 1.1022, "step": 580 }, { "epoch": 0.4883095385888682, "grad_norm": 1.2174866199493408, "learning_rate": 4.959836434683322e-05, "loss": 1.0551, "step": 590 }, { "epoch": 0.4965859714463066, "grad_norm": 1.272923231124878, "learning_rate": 4.9583694591479144e-05, "loss": 1.113, "step": 600 }, { "epoch": 0.5048624043037451, "grad_norm": 1.100860357284546, "learning_rate": 4.9568763961336626e-05, "loss": 1.0731, "step": 610 }, { "epoch": 0.5131388371611836, "grad_norm": 1.1055638790130615, "learning_rate": 4.955357261484503e-05, "loss": 1.0928, "step": 620 }, { "epoch": 0.521415270018622, "grad_norm": 1.1676515340805054, "learning_rate": 4.953812071321037e-05, "loss": 1.1089, "step": 630 }, { "epoch": 0.5296917028760604, "grad_norm": 1.578491449356079, "learning_rate": 4.952240842040359e-05, "loss": 1.0972, "step": 640 }, { "epoch": 0.5379681357334989, "grad_norm": 1.2124993801116943, "learning_rate": 4.950643590315882e-05, "loss": 1.0518, "step": 650 }, { "epoch": 0.5462445685909373, "grad_norm": 1.1784847974777222, "learning_rate": 4.949020333097163e-05, "loss": 1.0882, "step": 660 }, { "epoch": 0.5545210014483758, "grad_norm": 1.2548474073410034, "learning_rate": 4.9473710876097197e-05, "loss": 1.0578, "step": 670 }, { "epoch": 0.5627974343058142, "grad_norm": 1.05464506149292, "learning_rate": 4.945695871354849e-05, "loss": 1.0833, "step": 680 }, { "epoch": 0.5710738671632526, "grad_norm": 1.4869188070297241, "learning_rate": 4.943994702109446e-05, "loss": 1.0485, "step": 690 }, { "epoch": 0.5793503000206911, "grad_norm": 1.298762559890747, "learning_rate": 4.942267597925805e-05, "loss": 1.0704, "step": 700 }, { "epoch": 0.5876267328781295, "grad_norm": 1.2249423265457153, "learning_rate": 4.940514577131439e-05, "loss": 1.0614, "step": 710 }, { "epoch": 0.595903165735568, "grad_norm": 1.1190897226333618, "learning_rate": 4.938735658328878e-05, "loss": 1.0466, "step": 720 }, { "epoch": 0.6041795985930064, "grad_norm": 1.0651410818099976, "learning_rate": 4.936930860395474e-05, "loss": 1.0352, "step": 730 }, { "epoch": 0.6124560314504448, "grad_norm": 1.2162879705429077, "learning_rate": 4.9351002024832026e-05, "loss": 1.0527, "step": 740 }, { "epoch": 0.6207324643078833, "grad_norm": 1.3244678974151611, "learning_rate": 4.933243704018454e-05, "loss": 1.0415, "step": 750 }, { "epoch": 0.6290088971653217, "grad_norm": 1.3377256393432617, "learning_rate": 4.931361384701835e-05, "loss": 1.0753, "step": 760 }, { "epoch": 0.6372853300227602, "grad_norm": 1.3565047979354858, "learning_rate": 4.9294532645079504e-05, "loss": 1.064, "step": 770 }, { "epoch": 0.6455617628801986, "grad_norm": 1.2356196641921997, "learning_rate": 4.9275193636852014e-05, "loss": 1.0539, "step": 780 }, { "epoch": 0.6538381957376371, "grad_norm": 1.0150561332702637, "learning_rate": 4.9255597027555617e-05, "loss": 1.0389, "step": 790 }, { "epoch": 0.6621146285950755, "grad_norm": 1.0787653923034668, "learning_rate": 4.9235743025143656e-05, "loss": 1.0415, "step": 800 }, { "epoch": 0.6703910614525139, "grad_norm": 1.1886955499649048, "learning_rate": 4.921563184030084e-05, "loss": 1.027, "step": 810 }, { "epoch": 0.6786674943099524, "grad_norm": 1.1936701536178589, "learning_rate": 4.919526368644103e-05, "loss": 1.022, "step": 820 }, { "epoch": 0.6869439271673908, "grad_norm": 1.2435652017593384, "learning_rate": 4.917463877970496e-05, "loss": 1.0269, "step": 830 }, { "epoch": 0.6952203600248293, "grad_norm": 1.1247020959854126, "learning_rate": 4.915375733895796e-05, "loss": 1.0319, "step": 840 }, { "epoch": 0.7034967928822677, "grad_norm": 1.8782734870910645, "learning_rate": 4.91326195857876e-05, "loss": 1.0327, "step": 850 }, { "epoch": 0.7117732257397061, "grad_norm": 1.1094200611114502, "learning_rate": 4.911122574450137e-05, "loss": 1.0139, "step": 860 }, { "epoch": 0.7200496585971446, "grad_norm": 1.0171282291412354, "learning_rate": 4.90895760421243e-05, "loss": 1.0061, "step": 870 }, { "epoch": 0.728326091454583, "grad_norm": 1.22124445438385, "learning_rate": 4.906767070839653e-05, "loss": 1.0087, "step": 880 }, { "epoch": 0.7366025243120216, "grad_norm": 1.43118155002594, "learning_rate": 4.90455099757709e-05, "loss": 1.0265, "step": 890 }, { "epoch": 0.74487895716946, "grad_norm": 1.6183593273162842, "learning_rate": 4.902309407941043e-05, "loss": 0.998, "step": 900 }, { "epoch": 0.7531553900268984, "grad_norm": 1.1645337343215942, "learning_rate": 4.9000423257185904e-05, "loss": 1.0293, "step": 910 }, { "epoch": 0.7614318228843369, "grad_norm": 1.1429606676101685, "learning_rate": 4.897749774967326e-05, "loss": 1.0308, "step": 920 }, { "epoch": 0.7697082557417753, "grad_norm": 1.0337284803390503, "learning_rate": 4.895431780015114e-05, "loss": 1.0067, "step": 930 }, { "epoch": 0.7779846885992138, "grad_norm": 1.1541541814804077, "learning_rate": 4.893088365459817e-05, "loss": 1.0004, "step": 940 }, { "epoch": 0.7862611214566522, "grad_norm": 1.2892276048660278, "learning_rate": 4.8907195561690496e-05, "loss": 0.9858, "step": 950 }, { "epoch": 0.7945375543140907, "grad_norm": 1.098703145980835, "learning_rate": 4.8883253772799046e-05, "loss": 0.9817, "step": 960 }, { "epoch": 0.8028139871715291, "grad_norm": 1.1893219947814941, "learning_rate": 4.885905854198688e-05, "loss": 0.983, "step": 970 }, { "epoch": 0.8110904200289675, "grad_norm": 1.0653300285339355, "learning_rate": 4.8834610126006555e-05, "loss": 1.0175, "step": 980 }, { "epoch": 0.819366852886406, "grad_norm": 1.2211158275604248, "learning_rate": 4.88099087842973e-05, "loss": 0.9644, "step": 990 }, { "epoch": 0.8276432857438444, "grad_norm": 1.2194496393203735, "learning_rate": 4.878495477898235e-05, "loss": 0.9766, "step": 1000 }, { "epoch": 0.8276432857438444, "eval_loss": 1.0402740240097046, "eval_runtime": 580.0141, "eval_samples_per_second": 33.325, "eval_steps_per_second": 5.555, "step": 1000 }, { "epoch": 0.7920015683199373, "grad_norm": 1.1055457592010498, "learning_rate": 4.888596070759588e-05, "loss": 0.7878, "step": 1010 }, { "epoch": 0.7998431680062733, "grad_norm": 1.0589972734451294, "learning_rate": 4.8863072734908147e-05, "loss": 0.774, "step": 1020 }, { "epoch": 0.8076847676926093, "grad_norm": 0.9427498579025269, "learning_rate": 4.8839957498448887e-05, "loss": 0.7741, "step": 1030 }, { "epoch": 0.8155263673789453, "grad_norm": 1.094714879989624, "learning_rate": 4.881661521835972e-05, "loss": 0.7552, "step": 1040 }, { "epoch": 0.8233679670652814, "grad_norm": 1.029213786125183, "learning_rate": 4.879304611694457e-05, "loss": 0.7632, "step": 1050 }, { "epoch": 0.8312095667516173, "grad_norm": 1.1656430959701538, "learning_rate": 4.876925041866754e-05, "loss": 0.7533, "step": 1060 }, { "epoch": 0.8390511664379533, "grad_norm": 1.0112115144729614, "learning_rate": 4.8745228350150734e-05, "loss": 0.7536, "step": 1070 }, { "epoch": 0.8468927661242893, "grad_norm": 1.1366935968399048, "learning_rate": 4.8720980140172144e-05, "loss": 0.7892, "step": 1080 }, { "epoch": 0.8547343658106253, "grad_norm": 1.4567644596099854, "learning_rate": 4.869650601966348e-05, "loss": 0.7585, "step": 1090 }, { "epoch": 0.8625759654969614, "grad_norm": 1.1882156133651733, "learning_rate": 4.8671806221707886e-05, "loss": 0.7565, "step": 1100 }, { "epoch": 0.8704175651832974, "grad_norm": 1.055330753326416, "learning_rate": 4.864688098153786e-05, "loss": 0.775, "step": 1110 }, { "epoch": 0.8782591648696334, "grad_norm": 1.073933720588684, "learning_rate": 4.862173053653286e-05, "loss": 0.7524, "step": 1120 }, { "epoch": 0.8861007645559694, "grad_norm": 1.1609364748001099, "learning_rate": 4.859635512621718e-05, "loss": 0.7492, "step": 1130 }, { "epoch": 0.8939423642423054, "grad_norm": 0.9947859048843384, "learning_rate": 4.857075499225756e-05, "loss": 0.7526, "step": 1140 }, { "epoch": 0.9017839639286415, "grad_norm": 6.8848652839660645, "learning_rate": 4.8544930378460954e-05, "loss": 0.7702, "step": 1150 }, { "epoch": 0.9096255636149775, "grad_norm": 1.043473243713379, "learning_rate": 4.851888153077219e-05, "loss": 0.7756, "step": 1160 }, { "epoch": 0.9174671633013135, "grad_norm": 1.2918435335159302, "learning_rate": 4.849260869727161e-05, "loss": 0.7663, "step": 1170 }, { "epoch": 0.9253087629876495, "grad_norm": 1.2701385021209717, "learning_rate": 4.846611212817273e-05, "loss": 0.7656, "step": 1180 }, { "epoch": 0.9331503626739855, "grad_norm": 1.0570497512817383, "learning_rate": 4.843939207581983e-05, "loss": 0.757, "step": 1190 }, { "epoch": 0.9409919623603215, "grad_norm": 1.145662784576416, "learning_rate": 4.841244879468561e-05, "loss": 0.7464, "step": 1200 }, { "epoch": 0.9488335620466575, "grad_norm": 1.1173900365829468, "learning_rate": 4.838528254136866e-05, "loss": 0.7375, "step": 1210 }, { "epoch": 0.9566751617329935, "grad_norm": 1.2657852172851562, "learning_rate": 4.835789357459114e-05, "loss": 0.7682, "step": 1220 }, { "epoch": 0.9645167614193295, "grad_norm": 1.141322135925293, "learning_rate": 4.833028215519623e-05, "loss": 0.7425, "step": 1230 }, { "epoch": 0.9723583611056655, "grad_norm": 1.0891972780227661, "learning_rate": 4.830244854614568e-05, "loss": 0.7735, "step": 1240 }, { "epoch": 0.9801999607920016, "grad_norm": 1.1772578954696655, "learning_rate": 4.827439301251729e-05, "loss": 0.7591, "step": 1250 }, { "epoch": 0.9880415604783376, "grad_norm": 1.1350594758987427, "learning_rate": 4.824611582150241e-05, "loss": 0.7522, "step": 1260 }, { "epoch": 0.9958831601646736, "grad_norm": 1.03608238697052, "learning_rate": 4.821761724240336e-05, "loss": 0.7608, "step": 1270 }, { "epoch": 1.0037247598510095, "grad_norm": 1.1283738613128662, "learning_rate": 4.81888975466309e-05, "loss": 0.7299, "step": 1280 }, { "epoch": 1.0115663595373456, "grad_norm": 1.297022819519043, "learning_rate": 4.815995700770162e-05, "loss": 0.6897, "step": 1290 }, { "epoch": 1.0194079592236815, "grad_norm": 1.198197603225708, "learning_rate": 4.813079590123535e-05, "loss": 0.7174, "step": 1300 }, { "epoch": 1.0272495589100177, "grad_norm": 1.2182512283325195, "learning_rate": 4.810141450495254e-05, "loss": 0.6865, "step": 1310 }, { "epoch": 1.0350911585963536, "grad_norm": 1.0651652812957764, "learning_rate": 4.807181309867157e-05, "loss": 0.6786, "step": 1320 }, { "epoch": 1.0429327582826897, "grad_norm": 1.1763614416122437, "learning_rate": 4.8041991964306164e-05, "loss": 0.6762, "step": 1330 }, { "epoch": 1.0507743579690256, "grad_norm": 1.0765550136566162, "learning_rate": 4.801195138586263e-05, "loss": 0.6623, "step": 1340 }, { "epoch": 1.0586159576553618, "grad_norm": 1.19382905960083, "learning_rate": 4.7981691649437214e-05, "loss": 0.6796, "step": 1350 }, { "epoch": 1.0664575573416977, "grad_norm": 1.323279619216919, "learning_rate": 4.795121304321332e-05, "loss": 0.6904, "step": 1360 }, { "epoch": 1.0742991570280338, "grad_norm": 1.0708096027374268, "learning_rate": 4.7920515857458816e-05, "loss": 0.6986, "step": 1370 }, { "epoch": 1.0821407567143697, "grad_norm": 0.9783353209495544, "learning_rate": 4.788960038452322e-05, "loss": 0.6743, "step": 1380 }, { "epoch": 1.0899823564007058, "grad_norm": 1.1598165035247803, "learning_rate": 4.785846691883498e-05, "loss": 0.6881, "step": 1390 }, { "epoch": 1.0978239560870418, "grad_norm": 0.9942623376846313, "learning_rate": 4.782711575689859e-05, "loss": 0.6658, "step": 1400 }, { "epoch": 1.1056655557733777, "grad_norm": 1.0791629552841187, "learning_rate": 4.7795547197291844e-05, "loss": 0.6896, "step": 1410 }, { "epoch": 1.1135071554597138, "grad_norm": 1.3438973426818848, "learning_rate": 4.7763761540662945e-05, "loss": 0.6592, "step": 1420 }, { "epoch": 1.1213487551460497, "grad_norm": 1.0666946172714233, "learning_rate": 4.773175908972765e-05, "loss": 0.6656, "step": 1430 }, { "epoch": 1.1291903548323858, "grad_norm": 0.9979410767555237, "learning_rate": 4.76995401492664e-05, "loss": 0.6668, "step": 1440 }, { "epoch": 1.1370319545187217, "grad_norm": 1.0160318613052368, "learning_rate": 4.766710502612138e-05, "loss": 0.6719, "step": 1450 }, { "epoch": 1.1448735542050579, "grad_norm": 1.1360750198364258, "learning_rate": 4.763445402919369e-05, "loss": 0.6651, "step": 1460 }, { "epoch": 1.1527151538913938, "grad_norm": 1.293106198310852, "learning_rate": 4.7601587469440255e-05, "loss": 0.6736, "step": 1470 }, { "epoch": 1.16055675357773, "grad_norm": 1.0604827404022217, "learning_rate": 4.756850565987101e-05, "loss": 0.6627, "step": 1480 }, { "epoch": 1.1683983532640658, "grad_norm": 1.276926875114441, "learning_rate": 4.753520891554584e-05, "loss": 0.6714, "step": 1490 }, { "epoch": 1.176239952950402, "grad_norm": 0.9686921238899231, "learning_rate": 4.7501697553571564e-05, "loss": 0.6781, "step": 1500 }, { "epoch": 1.176239952950402, "eval_loss": 1.0382379293441772, "eval_runtime": 250.4218, "eval_samples_per_second": 38.595, "eval_steps_per_second": 6.433, "step": 1500 }, { "epoch": 1.1840815526367379, "grad_norm": 1.0833640098571777, "learning_rate": 4.7467971893099e-05, "loss": 0.6692, "step": 1510 }, { "epoch": 1.191923152323074, "grad_norm": 1.011971116065979, "learning_rate": 4.743403225531983e-05, "loss": 0.6576, "step": 1520 }, { "epoch": 1.19976475200941, "grad_norm": 1.0864920616149902, "learning_rate": 4.740330389782858e-05, "loss": 0.6742, "step": 1530 }, { "epoch": 1.207606351695746, "grad_norm": 1.0627334117889404, "learning_rate": 4.736895859534605e-05, "loss": 0.6878, "step": 1540 }, { "epoch": 1.215447951382082, "grad_norm": 1.0697227716445923, "learning_rate": 4.7334400258525804e-05, "loss": 0.6952, "step": 1550 }, { "epoch": 1.2232895510684179, "grad_norm": 1.106345534324646, "learning_rate": 4.7299629216489685e-05, "loss": 0.6788, "step": 1560 }, { "epoch": 1.231131150754754, "grad_norm": 1.0687874555587769, "learning_rate": 4.7264645800385235e-05, "loss": 0.6828, "step": 1570 }, { "epoch": 1.23897275044109, "grad_norm": 1.0802325010299683, "learning_rate": 4.7229450343382585e-05, "loss": 0.6767, "step": 1580 }, { "epoch": 1.246814350127426, "grad_norm": 1.0141079425811768, "learning_rate": 4.7194043180671274e-05, "loss": 0.6676, "step": 1590 }, { "epoch": 1.254655949813762, "grad_norm": 1.02689790725708, "learning_rate": 4.715842464945706e-05, "loss": 0.6627, "step": 1600 }, { "epoch": 1.262497549500098, "grad_norm": 0.9644928574562073, "learning_rate": 4.712259508895868e-05, "loss": 0.6752, "step": 1610 }, { "epoch": 1.270339149186434, "grad_norm": 1.024220585823059, "learning_rate": 4.7086554840404676e-05, "loss": 0.6773, "step": 1620 }, { "epoch": 1.2781807488727701, "grad_norm": 1.1297568082809448, "learning_rate": 4.705030424703006e-05, "loss": 0.6787, "step": 1630 }, { "epoch": 1.286022348559106, "grad_norm": 1.1851943731307983, "learning_rate": 4.701384365407316e-05, "loss": 0.6784, "step": 1640 }, { "epoch": 1.2938639482454422, "grad_norm": 0.9990951418876648, "learning_rate": 4.6977173408772226e-05, "loss": 0.6575, "step": 1650 }, { "epoch": 1.301705547931778, "grad_norm": 0.9888284206390381, "learning_rate": 4.694029386036216e-05, "loss": 0.6774, "step": 1660 }, { "epoch": 1.309547147618114, "grad_norm": 1.090669870376587, "learning_rate": 4.6903205360071235e-05, "loss": 0.6822, "step": 1670 }, { "epoch": 1.31738874730445, "grad_norm": 1.0397796630859375, "learning_rate": 4.686590826111768e-05, "loss": 0.6787, "step": 1680 }, { "epoch": 1.3252303469907862, "grad_norm": 1.1437417268753052, "learning_rate": 4.682840291870635e-05, "loss": 0.6845, "step": 1690 }, { "epoch": 1.3330719466771221, "grad_norm": 1.1360801458358765, "learning_rate": 4.679068969002534e-05, "loss": 0.6715, "step": 1700 }, { "epoch": 1.340913546363458, "grad_norm": 0.9892112612724304, "learning_rate": 4.675276893424259e-05, "loss": 0.6758, "step": 1710 }, { "epoch": 1.3487551460497942, "grad_norm": 1.129503607749939, "learning_rate": 4.671464101250246e-05, "loss": 0.6822, "step": 1720 }, { "epoch": 1.35659674573613, "grad_norm": 1.2884234189987183, "learning_rate": 4.667630628792227e-05, "loss": 0.6657, "step": 1730 }, { "epoch": 1.3644383454224662, "grad_norm": 1.1253407001495361, "learning_rate": 4.663776512558887e-05, "loss": 0.6644, "step": 1740 }, { "epoch": 1.3722799451088021, "grad_norm": 1.035405158996582, "learning_rate": 4.659901789255517e-05, "loss": 0.692, "step": 1750 }, { "epoch": 1.3801215447951383, "grad_norm": 0.9973240494728088, "learning_rate": 4.656006495783661e-05, "loss": 0.6493, "step": 1760 }, { "epoch": 1.3879631444814742, "grad_norm": 1.12758469581604, "learning_rate": 4.6520906692407645e-05, "loss": 0.6827, "step": 1770 }, { "epoch": 1.3958047441678103, "grad_norm": 0.9674776196479797, "learning_rate": 4.6481543469198265e-05, "loss": 0.6778, "step": 1780 }, { "epoch": 1.4036463438541462, "grad_norm": 0.9787268042564392, "learning_rate": 4.644197566309039e-05, "loss": 0.6653, "step": 1790 }, { "epoch": 1.4114879435404823, "grad_norm": 1.0090129375457764, "learning_rate": 4.640220365091431e-05, "loss": 0.6685, "step": 1800 }, { "epoch": 1.4193295432268183, "grad_norm": 1.0591384172439575, "learning_rate": 4.6362227811445124e-05, "loss": 0.6795, "step": 1810 }, { "epoch": 1.4271711429131542, "grad_norm": 1.0188828706741333, "learning_rate": 4.63220485253991e-05, "loss": 0.652, "step": 1820 }, { "epoch": 1.4350127425994903, "grad_norm": 0.908219039440155, "learning_rate": 4.628166617543005e-05, "loss": 0.6611, "step": 1830 }, { "epoch": 1.4428543422858264, "grad_norm": 1.6317641735076904, "learning_rate": 4.6241081146125735e-05, "loss": 0.6893, "step": 1840 }, { "epoch": 1.4506959419721623, "grad_norm": 1.042119026184082, "learning_rate": 4.620029382400414e-05, "loss": 0.6805, "step": 1850 }, { "epoch": 1.4585375416584982, "grad_norm": 1.0308270454406738, "learning_rate": 4.615930459750979e-05, "loss": 0.6641, "step": 1860 }, { "epoch": 1.4663791413448344, "grad_norm": 1.0121833086013794, "learning_rate": 4.611811385701014e-05, "loss": 0.6506, "step": 1870 }, { "epoch": 1.4742207410311703, "grad_norm": 0.9907941222190857, "learning_rate": 4.607672199479175e-05, "loss": 0.6742, "step": 1880 }, { "epoch": 1.4820623407175064, "grad_norm": 1.0068496465682983, "learning_rate": 4.603512940505661e-05, "loss": 0.6496, "step": 1890 }, { "epoch": 1.4899039404038423, "grad_norm": 0.9669331908226013, "learning_rate": 4.599333648391837e-05, "loss": 0.6683, "step": 1900 }, { "epoch": 1.4977455400901785, "grad_norm": 0.9877241849899292, "learning_rate": 4.595134362939856e-05, "loss": 0.6746, "step": 1910 }, { "epoch": 1.5055871397765144, "grad_norm": 0.9874888062477112, "learning_rate": 4.590915124142282e-05, "loss": 0.6741, "step": 1920 }, { "epoch": 1.5134287394628503, "grad_norm": 0.9484206438064575, "learning_rate": 4.586675972181708e-05, "loss": 0.6678, "step": 1930 }, { "epoch": 1.5212703391491864, "grad_norm": 0.9802061915397644, "learning_rate": 4.582416947430371e-05, "loss": 0.6304, "step": 1940 }, { "epoch": 1.5291119388355225, "grad_norm": 0.9652764797210693, "learning_rate": 4.578138090449772e-05, "loss": 0.6663, "step": 1950 }, { "epoch": 1.5369535385218585, "grad_norm": 1.1324354410171509, "learning_rate": 4.573839441990286e-05, "loss": 0.6613, "step": 1960 }, { "epoch": 1.5447951382081944, "grad_norm": 0.989467978477478, "learning_rate": 4.5695210429907774e-05, "loss": 0.6856, "step": 1970 }, { "epoch": 1.5526367378945305, "grad_norm": 0.9154842495918274, "learning_rate": 4.5651829345782046e-05, "loss": 0.6706, "step": 1980 }, { "epoch": 1.5604783375808666, "grad_norm": 0.9855336546897888, "learning_rate": 4.560825158067233e-05, "loss": 0.6608, "step": 1990 }, { "epoch": 1.5683199372672025, "grad_norm": 0.9735700488090515, "learning_rate": 4.556447754959844e-05, "loss": 0.6524, "step": 2000 }, { "epoch": 1.5683199372672025, "eval_loss": 1.007663607597351, "eval_runtime": 244.3994, "eval_samples_per_second": 39.546, "eval_steps_per_second": 6.592, "step": 2000 }, { "epoch": 1.5761615369535384, "grad_norm": 0.8914473056793213, "learning_rate": 4.5520507669449296e-05, "loss": 0.6428, "step": 2010 }, { "epoch": 1.5840031366398746, "grad_norm": 1.0634649991989136, "learning_rate": 4.547634235897906e-05, "loss": 0.6756, "step": 2020 }, { "epoch": 1.5918447363262107, "grad_norm": 0.9682461619377136, "learning_rate": 4.5431982038803113e-05, "loss": 0.6728, "step": 2030 }, { "epoch": 1.5996863360125464, "grad_norm": 0.9319469928741455, "learning_rate": 4.5387427131394006e-05, "loss": 0.6549, "step": 2040 }, { "epoch": 1.6075279356988825, "grad_norm": 1.000883936882019, "learning_rate": 4.5342678061077495e-05, "loss": 0.6605, "step": 2050 }, { "epoch": 1.6153695353852187, "grad_norm": 0.9240787029266357, "learning_rate": 4.5297735254028475e-05, "loss": 0.6627, "step": 2060 }, { "epoch": 1.6232111350715546, "grad_norm": 0.935151994228363, "learning_rate": 4.5252599138266926e-05, "loss": 0.6577, "step": 2070 }, { "epoch": 1.6310527347578905, "grad_norm": 0.8398969769477844, "learning_rate": 4.520727014365383e-05, "loss": 0.6807, "step": 2080 }, { "epoch": 1.6388943344442266, "grad_norm": 1.0277074575424194, "learning_rate": 4.516174870188707e-05, "loss": 0.6708, "step": 2090 }, { "epoch": 1.6467359341305627, "grad_norm": 1.0168735980987549, "learning_rate": 4.511603524649737e-05, "loss": 0.6306, "step": 2100 }, { "epoch": 1.6545775338168986, "grad_norm": 0.9712296724319458, "learning_rate": 4.5070130212844075e-05, "loss": 0.6664, "step": 2110 }, { "epoch": 1.6624191335032346, "grad_norm": 0.8896380662918091, "learning_rate": 4.502403403811109e-05, "loss": 0.6614, "step": 2120 }, { "epoch": 1.6702607331895707, "grad_norm": 0.876569926738739, "learning_rate": 4.497774716130266e-05, "loss": 0.644, "step": 2130 }, { "epoch": 1.6781023328759068, "grad_norm": 0.8911204934120178, "learning_rate": 4.493127002323925e-05, "loss": 0.6564, "step": 2140 }, { "epoch": 1.6859439325622427, "grad_norm": 1.0690345764160156, "learning_rate": 4.488460306655325e-05, "loss": 0.655, "step": 2150 }, { "epoch": 1.6937855322485786, "grad_norm": 0.9396145343780518, "learning_rate": 4.4837746735684874e-05, "loss": 0.6873, "step": 2160 }, { "epoch": 1.7016271319349148, "grad_norm": 0.9179625511169434, "learning_rate": 4.479070147687783e-05, "loss": 0.6711, "step": 2170 }, { "epoch": 1.7094687316212507, "grad_norm": 0.8691576719284058, "learning_rate": 4.4743467738175145e-05, "loss": 0.6568, "step": 2180 }, { "epoch": 1.7173103313075866, "grad_norm": 0.8838081955909729, "learning_rate": 4.469604596941483e-05, "loss": 0.6531, "step": 2190 }, { "epoch": 1.7251519309939227, "grad_norm": 1.0800485610961914, "learning_rate": 4.4648436622225676e-05, "loss": 0.6915, "step": 2200 }, { "epoch": 1.7329935306802589, "grad_norm": 1.1134532690048218, "learning_rate": 4.460064015002284e-05, "loss": 0.6667, "step": 2210 }, { "epoch": 1.7408351303665948, "grad_norm": 0.8759159445762634, "learning_rate": 4.4552657008003676e-05, "loss": 0.6578, "step": 2220 }, { "epoch": 1.7486767300529307, "grad_norm": 0.9356112480163574, "learning_rate": 4.4504487653143235e-05, "loss": 0.6706, "step": 2230 }, { "epoch": 1.7565183297392668, "grad_norm": 1.0078978538513184, "learning_rate": 4.445613254419006e-05, "loss": 0.6486, "step": 2240 }, { "epoch": 1.764359929425603, "grad_norm": 0.984367311000824, "learning_rate": 4.440759214166172e-05, "loss": 0.6643, "step": 2250 }, { "epoch": 1.7722015291119388, "grad_norm": 1.0800858736038208, "learning_rate": 4.435886690784047e-05, "loss": 0.6621, "step": 2260 }, { "epoch": 1.7800431287982748, "grad_norm": 0.9246965050697327, "learning_rate": 4.430995730676882e-05, "loss": 0.6671, "step": 2270 }, { "epoch": 1.7878847284846109, "grad_norm": 1.0341116189956665, "learning_rate": 4.4260863804245146e-05, "loss": 0.6736, "step": 2280 }, { "epoch": 1.795726328170947, "grad_norm": 0.9403420686721802, "learning_rate": 4.421652280262614e-05, "loss": 0.6223, "step": 2290 }, { "epoch": 1.803567927857283, "grad_norm": 0.9806979894638062, "learning_rate": 4.4167081176887246e-05, "loss": 0.6622, "step": 2300 }, { "epoch": 1.8114095275436188, "grad_norm": 1.016072392463684, "learning_rate": 4.4117457010399885e-05, "loss": 0.6526, "step": 2310 }, { "epoch": 1.819251127229955, "grad_norm": 0.9248052835464478, "learning_rate": 4.4067650775767717e-05, "loss": 0.6459, "step": 2320 }, { "epoch": 1.8270927269162909, "grad_norm": 0.9529238343238831, "learning_rate": 4.401766294732834e-05, "loss": 0.6584, "step": 2330 }, { "epoch": 1.8349343266026268, "grad_norm": 0.9078419208526611, "learning_rate": 4.396749400114879e-05, "loss": 0.6832, "step": 2340 }, { "epoch": 1.842775926288963, "grad_norm": 0.9941418766975403, "learning_rate": 4.391714441502103e-05, "loss": 0.6664, "step": 2350 }, { "epoch": 1.850617525975299, "grad_norm": 0.944197952747345, "learning_rate": 4.387167573663176e-05, "loss": 0.6578, "step": 2360 }, { "epoch": 1.858459125661635, "grad_norm": 0.9415684938430786, "learning_rate": 4.382098425707695e-05, "loss": 0.6409, "step": 2370 }, { "epoch": 1.8663007253479709, "grad_norm": 0.9796119928359985, "learning_rate": 4.377011353288277e-05, "loss": 0.6593, "step": 2380 }, { "epoch": 1.874142325034307, "grad_norm": 0.9953365325927734, "learning_rate": 4.3719064048524696e-05, "loss": 0.6624, "step": 2390 }, { "epoch": 1.8819839247206431, "grad_norm": 0.8975547552108765, "learning_rate": 4.3667836290180586e-05, "loss": 0.6541, "step": 2400 }, { "epoch": 1.889825524406979, "grad_norm": 1.0561144351959229, "learning_rate": 4.361643074572617e-05, "loss": 0.6307, "step": 2410 }, { "epoch": 1.897667124093315, "grad_norm": 0.9564677476882935, "learning_rate": 4.356484790473032e-05, "loss": 0.6293, "step": 2420 }, { "epoch": 1.905508723779651, "grad_norm": 10.329718589782715, "learning_rate": 4.351308825845044e-05, "loss": 0.6628, "step": 2430 }, { "epoch": 1.9133503234659872, "grad_norm": 0.9198611378669739, "learning_rate": 4.3461152299827754e-05, "loss": 0.6617, "step": 2440 }, { "epoch": 1.9211919231523231, "grad_norm": 0.893352210521698, "learning_rate": 4.340904052348264e-05, "loss": 0.6266, "step": 2450 }, { "epoch": 1.929033522838659, "grad_norm": 0.9610190391540527, "learning_rate": 4.3356753425709885e-05, "loss": 0.6603, "step": 2460 }, { "epoch": 1.9368751225249952, "grad_norm": 0.9027713537216187, "learning_rate": 4.330429150447399e-05, "loss": 0.6759, "step": 2470 }, { "epoch": 1.944716722211331, "grad_norm": 0.8384199142456055, "learning_rate": 4.325165525940442e-05, "loss": 0.6429, "step": 2480 }, { "epoch": 1.952558321897667, "grad_norm": 0.8894764184951782, "learning_rate": 4.319884519179082e-05, "loss": 0.6529, "step": 2490 }, { "epoch": 1.9603999215840031, "grad_norm": 0.8820430040359497, "learning_rate": 4.3145861804578283e-05, "loss": 0.6591, "step": 2500 }, { "epoch": 1.9603999215840031, "eval_loss": 0.9828659892082214, "eval_runtime": 242.2012, "eval_samples_per_second": 39.905, "eval_steps_per_second": 6.651, "step": 2500 }, { "epoch": 1.9684375612624976, "grad_norm": 0.9603639841079712, "learning_rate": 4.3092705602362536e-05, "loss": 0.6308, "step": 2510 }, { "epoch": 1.9762791609488337, "grad_norm": 0.9399013519287109, "learning_rate": 4.303937709138512e-05, "loss": 0.6523, "step": 2520 }, { "epoch": 1.9841207606351696, "grad_norm": 0.8963415026664734, "learning_rate": 4.298587677952861e-05, "loss": 0.6764, "step": 2530 }, { "epoch": 1.9919623603215055, "grad_norm": 0.8300694227218628, "learning_rate": 4.2932205176311726e-05, "loss": 0.6513, "step": 2540 }, { "epoch": 1.9998039600078417, "grad_norm": 0.8571862578392029, "learning_rate": 4.287836279288453e-05, "loss": 0.6521, "step": 2550 }, { "epoch": 2.007645559694178, "grad_norm": 1.0097789764404297, "learning_rate": 4.282435014202354e-05, "loss": 0.3983, "step": 2560 }, { "epoch": 2.0154871593805135, "grad_norm": 0.9734821319580078, "learning_rate": 4.2770167738126834e-05, "loss": 0.3739, "step": 2570 }, { "epoch": 2.0233287590668496, "grad_norm": 1.0220587253570557, "learning_rate": 4.271581609720915e-05, "loss": 0.3803, "step": 2580 }, { "epoch": 2.0311703587531857, "grad_norm": 1.0062997341156006, "learning_rate": 4.266129573689699e-05, "loss": 0.3539, "step": 2590 }, { "epoch": 2.039011958439522, "grad_norm": 1.014708161354065, "learning_rate": 4.260660717642369e-05, "loss": 0.3699, "step": 2600 }, { "epoch": 2.0468535581258576, "grad_norm": 1.0316507816314697, "learning_rate": 4.255175093662446e-05, "loss": 0.3725, "step": 2610 }, { "epoch": 2.0546951578121937, "grad_norm": 1.0374858379364014, "learning_rate": 4.249672753993143e-05, "loss": 0.3732, "step": 2620 }, { "epoch": 2.06253675749853, "grad_norm": 0.9619321823120117, "learning_rate": 4.2441537510368675e-05, "loss": 0.369, "step": 2630 }, { "epoch": 2.0703783571848655, "grad_norm": 0.8847309350967407, "learning_rate": 4.238618137354723e-05, "loss": 0.3582, "step": 2640 }, { "epoch": 2.0782199568712016, "grad_norm": 0.8922093510627747, "learning_rate": 4.2330659656660075e-05, "loss": 0.3553, "step": 2650 }, { "epoch": 2.0860615565575378, "grad_norm": 0.9724310040473938, "learning_rate": 4.227497288847712e-05, "loss": 0.3771, "step": 2660 }, { "epoch": 2.093903156243874, "grad_norm": 1.001301646232605, "learning_rate": 4.221912159934016e-05, "loss": 0.3773, "step": 2670 }, { "epoch": 2.1017447559302096, "grad_norm": 0.971805214881897, "learning_rate": 4.216310632115784e-05, "loss": 0.3748, "step": 2680 }, { "epoch": 2.1095863556165457, "grad_norm": 0.9344645142555237, "learning_rate": 4.2106927587400566e-05, "loss": 0.3748, "step": 2690 }, { "epoch": 2.117427955302882, "grad_norm": 0.9709860682487488, "learning_rate": 4.205058593309548e-05, "loss": 0.3792, "step": 2700 }, { "epoch": 2.125269554989218, "grad_norm": 1.0208593606948853, "learning_rate": 4.199408189482127e-05, "loss": 0.3724, "step": 2710 }, { "epoch": 2.1331111546755537, "grad_norm": 0.995505154132843, "learning_rate": 4.193741601070315e-05, "loss": 0.3843, "step": 2720 }, { "epoch": 2.14095275436189, "grad_norm": 1.059967279434204, "learning_rate": 4.1880588820407673e-05, "loss": 0.377, "step": 2730 }, { "epoch": 2.148794354048226, "grad_norm": 1.062596082687378, "learning_rate": 4.1823600865137643e-05, "loss": 0.3779, "step": 2740 }, { "epoch": 2.156635953734562, "grad_norm": 0.9699451327323914, "learning_rate": 4.1766452687626926e-05, "loss": 0.3889, "step": 2750 }, { "epoch": 2.1644775534208978, "grad_norm": 1.0266555547714233, "learning_rate": 4.170914483213528e-05, "loss": 0.3711, "step": 2760 }, { "epoch": 2.172319153107234, "grad_norm": 0.9311341047286987, "learning_rate": 4.165167784444319e-05, "loss": 0.3577, "step": 2770 }, { "epoch": 2.18016075279357, "grad_norm": 0.9153704047203064, "learning_rate": 4.1594052271846663e-05, "loss": 0.3778, "step": 2780 }, { "epoch": 2.1880023524799057, "grad_norm": 0.9365727305412292, "learning_rate": 4.1536268663152005e-05, "loss": 0.3924, "step": 2790 }, { "epoch": 2.195843952166242, "grad_norm": 1.0049713850021362, "learning_rate": 4.147832756867062e-05, "loss": 0.4038, "step": 2800 }, { "epoch": 2.203685551852578, "grad_norm": 1.0050500631332397, "learning_rate": 4.142022954021374e-05, "loss": 0.4056, "step": 2810 }, { "epoch": 2.211527151538914, "grad_norm": 0.9921131730079651, "learning_rate": 4.1361975131087184e-05, "loss": 0.3959, "step": 2820 }, { "epoch": 2.21936875122525, "grad_norm": 0.9911596179008484, "learning_rate": 4.1303564896086086e-05, "loss": 0.4093, "step": 2830 }, { "epoch": 2.227210350911586, "grad_norm": 0.9569091796875, "learning_rate": 4.1244999391489636e-05, "loss": 0.3944, "step": 2840 }, { "epoch": 2.235051950597922, "grad_norm": 1.010050654411316, "learning_rate": 4.1186279175055715e-05, "loss": 0.393, "step": 2850 }, { "epoch": 2.242893550284258, "grad_norm": 0.9854335784912109, "learning_rate": 4.112740480601566e-05, "loss": 0.3923, "step": 2860 }, { "epoch": 2.250735149970594, "grad_norm": 0.9931651949882507, "learning_rate": 4.1068376845068904e-05, "loss": 0.4148, "step": 2870 }, { "epoch": 2.25857674965693, "grad_norm": 0.9877751469612122, "learning_rate": 4.100919585437762e-05, "loss": 0.3918, "step": 2880 }, { "epoch": 2.266418349343266, "grad_norm": 1.1037381887435913, "learning_rate": 4.094986239756139e-05, "loss": 0.386, "step": 2890 }, { "epoch": 2.2742599490296023, "grad_norm": 0.9635111093521118, "learning_rate": 4.089037703969184e-05, "loss": 0.404, "step": 2900 }, { "epoch": 2.282101548715938, "grad_norm": 1.041157603263855, "learning_rate": 4.083074034728725e-05, "loss": 0.4102, "step": 2910 }, { "epoch": 2.289943148402274, "grad_norm": 0.9768211245536804, "learning_rate": 4.0770952888307143e-05, "loss": 0.3885, "step": 2920 }, { "epoch": 2.29778474808861, "grad_norm": 1.0234442949295044, "learning_rate": 4.07110152321469e-05, "loss": 0.409, "step": 2930 }, { "epoch": 2.305626347774946, "grad_norm": 0.9377536773681641, "learning_rate": 4.0650927949632335e-05, "loss": 0.4072, "step": 2940 }, { "epoch": 2.313467947461282, "grad_norm": 0.8978007435798645, "learning_rate": 4.0590691613014244e-05, "loss": 0.4, "step": 2950 }, { "epoch": 2.321309547147618, "grad_norm": 46.86246109008789, "learning_rate": 4.053030679596297e-05, "loss": 0.4276, "step": 2960 }, { "epoch": 2.3291511468339543, "grad_norm": 0.9630893468856812, "learning_rate": 4.0469774073562927e-05, "loss": 0.4096, "step": 2970 }, { "epoch": 2.33699274652029, "grad_norm": 0.9448952674865723, "learning_rate": 4.040909402230714e-05, "loss": 0.4159, "step": 2980 }, { "epoch": 2.344834346206626, "grad_norm": 0.9361767768859863, "learning_rate": 4.034826722009172e-05, "loss": 0.4089, "step": 2990 }, { "epoch": 2.3526759458929623, "grad_norm": 0.9969122409820557, "learning_rate": 4.0287294246210414e-05, "loss": 0.422, "step": 3000 }, { "epoch": 2.3526759458929623, "eval_loss": 1.1007354259490967, "eval_runtime": 250.2946, "eval_samples_per_second": 38.615, "eval_steps_per_second": 6.436, "step": 3000 }, { "epoch": 2.3607135855714567, "grad_norm": 1.070020318031311, "learning_rate": 4.022617568134903e-05, "loss": 0.4181, "step": 3010 }, { "epoch": 2.3685551852577924, "grad_norm": 0.9444131255149841, "learning_rate": 4.01710449737401e-05, "loss": 0.4108, "step": 3020 }, { "epoch": 2.3763967849441285, "grad_norm": 0.9726797342300415, "learning_rate": 4.010965139076752e-05, "loss": 0.4116, "step": 3030 }, { "epoch": 2.3842383846304647, "grad_norm": 0.9778685569763184, "learning_rate": 4.004811390862488e-05, "loss": 0.4285, "step": 3040 }, { "epoch": 2.392079984316801, "grad_norm": 0.9821404814720154, "learning_rate": 3.998643311337419e-05, "loss": 0.4278, "step": 3050 }, { "epoch": 2.3999215840031365, "grad_norm": 0.979962944984436, "learning_rate": 3.9924609592442326e-05, "loss": 0.415, "step": 3060 }, { "epoch": 2.4077631836894726, "grad_norm": 1.1424775123596191, "learning_rate": 3.986264393461542e-05, "loss": 0.4253, "step": 3070 }, { "epoch": 2.4156047833758088, "grad_norm": 0.9540061354637146, "learning_rate": 3.9800536730033274e-05, "loss": 0.3971, "step": 3080 }, { "epoch": 2.423446383062145, "grad_norm": 0.914680004119873, "learning_rate": 3.973828857018374e-05, "loss": 0.4206, "step": 3090 }, { "epoch": 2.4312879827484806, "grad_norm": 1.0090687274932861, "learning_rate": 3.967590004789709e-05, "loss": 0.4158, "step": 3100 }, { "epoch": 2.4391295824348167, "grad_norm": 1.0544170141220093, "learning_rate": 3.961337175734032e-05, "loss": 0.4218, "step": 3110 }, { "epoch": 2.446971182121153, "grad_norm": 1.0221105813980103, "learning_rate": 3.9550704294011585e-05, "loss": 0.4092, "step": 3120 }, { "epoch": 2.4548127818074885, "grad_norm": 1.0691900253295898, "learning_rate": 3.948789825473443e-05, "loss": 0.4105, "step": 3130 }, { "epoch": 2.4626543814938247, "grad_norm": 1.0215855836868286, "learning_rate": 3.9424954237652166e-05, "loss": 0.4139, "step": 3140 }, { "epoch": 2.470495981180161, "grad_norm": 1.0506454706192017, "learning_rate": 3.936187284222218e-05, "loss": 0.4075, "step": 3150 }, { "epoch": 2.478337580866497, "grad_norm": 1.0375871658325195, "learning_rate": 3.929865466921016e-05, "loss": 0.423, "step": 3160 }, { "epoch": 2.4861791805528326, "grad_norm": 0.9551310539245605, "learning_rate": 3.923530032068445e-05, "loss": 0.4285, "step": 3170 }, { "epoch": 2.4940207802391687, "grad_norm": 0.9317300319671631, "learning_rate": 3.917181040001027e-05, "loss": 0.4125, "step": 3180 }, { "epoch": 2.501862379925505, "grad_norm": 0.9303199648857117, "learning_rate": 3.910818551184398e-05, "loss": 0.4047, "step": 3190 }, { "epoch": 2.5097039796118406, "grad_norm": 1.0755215883255005, "learning_rate": 3.904442626212733e-05, "loss": 0.4225, "step": 3200 }, { "epoch": 2.5175455792981767, "grad_norm": 1.0713340044021606, "learning_rate": 3.8980533258081684e-05, "loss": 0.4115, "step": 3210 }, { "epoch": 2.525387178984513, "grad_norm": 0.9300441145896912, "learning_rate": 3.891650710820225e-05, "loss": 0.4398, "step": 3220 }, { "epoch": 2.533228778670849, "grad_norm": 0.9100314378738403, "learning_rate": 3.885234842225225e-05, "loss": 0.4414, "step": 3230 }, { "epoch": 2.541070378357185, "grad_norm": 1.015506386756897, "learning_rate": 3.8788057811257126e-05, "loss": 0.4109, "step": 3240 }, { "epoch": 2.5489119780435208, "grad_norm": 1.0658026933670044, "learning_rate": 3.8723635887498766e-05, "loss": 0.4209, "step": 3250 }, { "epoch": 2.556753577729857, "grad_norm": 0.9921278953552246, "learning_rate": 3.865908326450961e-05, "loss": 0.4244, "step": 3260 }, { "epoch": 2.564595177416193, "grad_norm": 0.968429684638977, "learning_rate": 3.8594400557066835e-05, "loss": 0.4278, "step": 3270 }, { "epoch": 2.5724367771025287, "grad_norm": 1.021282434463501, "learning_rate": 3.852958838118649e-05, "loss": 0.4253, "step": 3280 }, { "epoch": 2.580278376788865, "grad_norm": 1.0124115943908691, "learning_rate": 3.846464735411767e-05, "loss": 0.4198, "step": 3290 }, { "epoch": 2.588119976475201, "grad_norm": 1.1336549520492554, "learning_rate": 3.839957809433657e-05, "loss": 0.4206, "step": 3300 }, { "epoch": 2.595961576161537, "grad_norm": 0.9238972663879395, "learning_rate": 3.833438122154065e-05, "loss": 0.4468, "step": 3310 }, { "epoch": 2.6038031758478732, "grad_norm": 0.9685710072517395, "learning_rate": 3.826905735664269e-05, "loss": 0.4393, "step": 3320 }, { "epoch": 2.611644775534209, "grad_norm": 1.1224950551986694, "learning_rate": 3.820360712176494e-05, "loss": 0.4407, "step": 3330 }, { "epoch": 2.619486375220545, "grad_norm": 0.9622184634208679, "learning_rate": 3.81380311402331e-05, "loss": 0.4289, "step": 3340 }, { "epoch": 2.6273279749068807, "grad_norm": 0.9730106592178345, "learning_rate": 3.807233003657049e-05, "loss": 0.4153, "step": 3350 }, { "epoch": 2.635169574593217, "grad_norm": 1.0938421487808228, "learning_rate": 3.800650443649203e-05, "loss": 0.4161, "step": 3360 }, { "epoch": 2.643011174279553, "grad_norm": 0.9540886282920837, "learning_rate": 3.794055496689829e-05, "loss": 0.4333, "step": 3370 }, { "epoch": 2.650852773965889, "grad_norm": 0.986900806427002, "learning_rate": 3.787448225586954e-05, "loss": 0.4217, "step": 3380 }, { "epoch": 2.6586943736522253, "grad_norm": 0.9545695781707764, "learning_rate": 3.780828693265976e-05, "loss": 0.4456, "step": 3390 }, { "epoch": 2.666535973338561, "grad_norm": 0.9835594892501831, "learning_rate": 3.7741969627690656e-05, "loss": 0.4397, "step": 3400 }, { "epoch": 2.674377573024897, "grad_norm": 0.9485548734664917, "learning_rate": 3.7675530972545636e-05, "loss": 0.4246, "step": 3410 }, { "epoch": 2.6822191727112332, "grad_norm": 0.9652374386787415, "learning_rate": 3.76089715999638e-05, "loss": 0.442, "step": 3420 }, { "epoch": 2.690060772397569, "grad_norm": 1.1110929250717163, "learning_rate": 3.7542292143833926e-05, "loss": 0.4426, "step": 3430 }, { "epoch": 2.697902372083905, "grad_norm": 1.0148861408233643, "learning_rate": 3.747549323918844e-05, "loss": 0.4619, "step": 3440 }, { "epoch": 2.705743971770241, "grad_norm": 1.0639394521713257, "learning_rate": 3.7408575522197334e-05, "loss": 0.4284, "step": 3450 }, { "epoch": 2.7135855714565773, "grad_norm": 1.1294188499450684, "learning_rate": 3.734153963016215e-05, "loss": 0.422, "step": 3460 }, { "epoch": 2.721427171142913, "grad_norm": 1.133299708366394, "learning_rate": 3.727438620150985e-05, "loss": 0.4356, "step": 3470 }, { "epoch": 2.729268770829249, "grad_norm": 1.0103888511657715, "learning_rate": 3.720711587578682e-05, "loss": 0.4293, "step": 3480 }, { "epoch": 2.7371103705155853, "grad_norm": 0.9841229319572449, "learning_rate": 3.7139729293652715e-05, "loss": 0.4314, "step": 3490 }, { "epoch": 2.744951970201921, "grad_norm": 1.0465657711029053, "learning_rate": 3.7072227096874366e-05, "loss": 0.4357, "step": 3500 }, { "epoch": 2.744951970201921, "eval_loss": 1.100440263748169, "eval_runtime": 251.9109, "eval_samples_per_second": 38.367, "eval_steps_per_second": 6.395, "step": 3500 }, { "epoch": 2.752793569888257, "grad_norm": 1.0336558818817139, "learning_rate": 3.7004609928319703e-05, "loss": 0.4326, "step": 3510 }, { "epoch": 2.760635169574593, "grad_norm": 1.018001675605774, "learning_rate": 3.693687843195158e-05, "loss": 0.4249, "step": 3520 }, { "epoch": 2.7684767692609293, "grad_norm": 0.970750093460083, "learning_rate": 3.686903325282169e-05, "loss": 0.4395, "step": 3530 }, { "epoch": 2.7763183689472655, "grad_norm": 0.9996511936187744, "learning_rate": 3.6801075037064394e-05, "loss": 0.4237, "step": 3540 }, { "epoch": 2.784159968633601, "grad_norm": 0.9123607873916626, "learning_rate": 3.673300443189057e-05, "loss": 0.4393, "step": 3550 }, { "epoch": 2.7920015683199373, "grad_norm": 0.979560375213623, "learning_rate": 3.666482208558147e-05, "loss": 0.4446, "step": 3560 }, { "epoch": 2.7998431680062734, "grad_norm": 1.0506279468536377, "learning_rate": 3.659652864748253e-05, "loss": 0.4172, "step": 3570 }, { "epoch": 2.807684767692609, "grad_norm": 1.0336925983428955, "learning_rate": 3.652812476799716e-05, "loss": 0.4408, "step": 3580 }, { "epoch": 2.8155263673789452, "grad_norm": 1.0965049266815186, "learning_rate": 3.645961109858062e-05, "loss": 0.4439, "step": 3590 }, { "epoch": 2.8233679670652814, "grad_norm": 0.9744886159896851, "learning_rate": 3.6390988291733704e-05, "loss": 0.4397, "step": 3600 }, { "epoch": 2.8312095667516175, "grad_norm": 0.925773561000824, "learning_rate": 3.6322257000996665e-05, "loss": 0.4444, "step": 3610 }, { "epoch": 2.839051166437953, "grad_norm": 1.040789008140564, "learning_rate": 3.6253417880942896e-05, "loss": 0.442, "step": 3620 }, { "epoch": 2.8468927661242893, "grad_norm": 1.1857627630233765, "learning_rate": 3.61844715871727e-05, "loss": 0.4325, "step": 3630 }, { "epoch": 2.8547343658106255, "grad_norm": 0.9822872281074524, "learning_rate": 3.6115418776307084e-05, "loss": 0.4277, "step": 3640 }, { "epoch": 2.862575965496961, "grad_norm": 0.9385848045349121, "learning_rate": 3.6046260105981485e-05, "loss": 0.421, "step": 3650 }, { "epoch": 2.8704175651832973, "grad_norm": 1.0122392177581787, "learning_rate": 3.5976996234839505e-05, "loss": 0.4475, "step": 3660 }, { "epoch": 2.8782591648696334, "grad_norm": 1.0014116764068604, "learning_rate": 3.590762782252665e-05, "loss": 0.4499, "step": 3670 }, { "epoch": 2.8861007645559695, "grad_norm": 0.9408068656921387, "learning_rate": 3.583815552968403e-05, "loss": 0.4349, "step": 3680 }, { "epoch": 2.8939423642423057, "grad_norm": 1.007498025894165, "learning_rate": 3.5768580017942096e-05, "loss": 0.4479, "step": 3690 }, { "epoch": 2.9017839639286414, "grad_norm": 1.084996223449707, "learning_rate": 3.569890194991429e-05, "loss": 0.4439, "step": 3700 }, { "epoch": 2.9096255636149775, "grad_norm": 1.1018431186676025, "learning_rate": 3.56291219891908e-05, "loss": 0.454, "step": 3710 }, { "epoch": 2.9174671633013136, "grad_norm": 1.0079305171966553, "learning_rate": 3.555924080033218e-05, "loss": 0.4449, "step": 3720 }, { "epoch": 2.9253087629876493, "grad_norm": 1.0269763469696045, "learning_rate": 3.5489259048863054e-05, "loss": 0.4339, "step": 3730 }, { "epoch": 2.9331503626739854, "grad_norm": 0.9952816963195801, "learning_rate": 3.5419177401265784e-05, "loss": 0.4551, "step": 3740 }, { "epoch": 2.9409919623603216, "grad_norm": 0.8597270846366882, "learning_rate": 3.5348996524974086e-05, "loss": 0.4441, "step": 3750 }, { "epoch": 2.9488335620466577, "grad_norm": 0.9447600245475769, "learning_rate": 3.5278717088366703e-05, "loss": 0.4505, "step": 3760 }, { "epoch": 2.9566751617329934, "grad_norm": 0.9347754120826721, "learning_rate": 3.520833976076104e-05, "loss": 0.4189, "step": 3770 }, { "epoch": 2.9645167614193295, "grad_norm": 1.0394740104675293, "learning_rate": 3.5137865212406785e-05, "loss": 0.4533, "step": 3780 }, { "epoch": 2.9723583611056656, "grad_norm": 0.9572747945785522, "learning_rate": 3.506729411447951e-05, "loss": 0.4338, "step": 3790 }, { "epoch": 2.9801999607920013, "grad_norm": 1.0237895250320435, "learning_rate": 3.49966271390743e-05, "loss": 0.4345, "step": 3800 }, { "epoch": 2.9880415604783375, "grad_norm": 1.1056692600250244, "learning_rate": 3.492586495919934e-05, "loss": 0.4607, "step": 3810 }, { "epoch": 2.9958831601646736, "grad_norm": 1.1012498140335083, "learning_rate": 3.485500824876953e-05, "loss": 0.4491, "step": 3820 }, { "epoch": 3.0037247598510097, "grad_norm": 1.2760201692581177, "learning_rate": 3.47840576826e-05, "loss": 0.3312, "step": 3830 }, { "epoch": 3.0115663595373454, "grad_norm": 0.9876531958580017, "learning_rate": 3.471301393639979e-05, "loss": 0.1973, "step": 3840 }, { "epoch": 3.0194079592236815, "grad_norm": 0.9299415349960327, "learning_rate": 3.46418776867653e-05, "loss": 0.1983, "step": 3850 }, { "epoch": 3.0272495589100177, "grad_norm": 0.9327115416526794, "learning_rate": 3.457064961117396e-05, "loss": 0.1952, "step": 3860 }, { "epoch": 3.035091158596354, "grad_norm": 0.956949770450592, "learning_rate": 3.449933038797765e-05, "loss": 0.1963, "step": 3870 }, { "epoch": 3.0429327582826895, "grad_norm": 0.9961047768592834, "learning_rate": 3.4427920696396355e-05, "loss": 0.196, "step": 3880 }, { "epoch": 3.0507743579690256, "grad_norm": 0.9145923852920532, "learning_rate": 3.435642121651165e-05, "loss": 0.1983, "step": 3890 }, { "epoch": 3.0586159576553618, "grad_norm": 0.8782720565795898, "learning_rate": 3.42848326292602e-05, "loss": 0.1929, "step": 3900 }, { "epoch": 3.066457557341698, "grad_norm": 0.9555357694625854, "learning_rate": 3.421315561642732e-05, "loss": 0.2011, "step": 3910 }, { "epoch": 3.0742991570280336, "grad_norm": 0.8813580274581909, "learning_rate": 3.4141390860640454e-05, "loss": 0.1986, "step": 3920 }, { "epoch": 3.0821407567143697, "grad_norm": 0.9723597168922424, "learning_rate": 3.406953904536266e-05, "loss": 0.2079, "step": 3930 }, { "epoch": 3.089982356400706, "grad_norm": 0.9829353094100952, "learning_rate": 3.399760085488615e-05, "loss": 0.2018, "step": 3940 }, { "epoch": 3.0978239560870415, "grad_norm": 0.9612635970115662, "learning_rate": 3.392557697432573e-05, "loss": 0.1948, "step": 3950 }, { "epoch": 3.1056655557733777, "grad_norm": 1.003859043121338, "learning_rate": 3.385346808961229e-05, "loss": 0.2051, "step": 3960 }, { "epoch": 3.113507155459714, "grad_norm": 0.881211519241333, "learning_rate": 3.378127488748627e-05, "loss": 0.1965, "step": 3970 }, { "epoch": 3.12134875514605, "grad_norm": 1.045480728149414, "learning_rate": 3.3708998055491114e-05, "loss": 0.2073, "step": 3980 }, { "epoch": 3.1291903548323856, "grad_norm": 0.9101377725601196, "learning_rate": 3.363663828196675e-05, "loss": 0.2, "step": 3990 }, { "epoch": 3.1370319545187217, "grad_norm": 0.9816673398017883, "learning_rate": 3.3564196256043e-05, "loss": 0.2055, "step": 4000 }, { "epoch": 3.1370319545187217, "eval_loss": 1.250553846359253, "eval_runtime": 246.4616, "eval_samples_per_second": 39.215, "eval_steps_per_second": 6.537, "step": 4000 } ], "logging_steps": 10, "max_steps": 10200, "num_input_tokens_seen": 0, "num_train_epochs": 8, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 3.911561497694909e+19, "train_batch_size": 6, "trial_name": null, "trial_params": null }