diff --git "a/checkpoint-6000/trainer_state.json" "b/checkpoint-6000/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-6000/trainer_state.json" @@ -0,0 +1,4233 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.11413570735604633, + "eval_steps": 500, + "global_step": 6000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0001902261789267439, + "grad_norm": 7.031089782714844, + "learning_rate": 1.9022256039566294e-08, + "loss": 2.4444, + "step": 10 + }, + { + "epoch": 0.0003804523578534878, + "grad_norm": 13.060206413269043, + "learning_rate": 3.804451207913259e-08, + "loss": 2.614, + "step": 20 + }, + { + "epoch": 0.0005706785367802317, + "grad_norm": 15.187254905700684, + "learning_rate": 5.7066768118698886e-08, + "loss": 3.1213, + "step": 30 + }, + { + "epoch": 0.0007609047157069756, + "grad_norm": 19.703323364257812, + "learning_rate": 7.608902415826518e-08, + "loss": 4.1664, + "step": 40 + }, + { + "epoch": 0.0009511308946337195, + "grad_norm": 32.09828186035156, + "learning_rate": 9.511128019783147e-08, + "loss": 5.9187, + "step": 50 + }, + { + "epoch": 0.0011413570735604635, + "grad_norm": 6.527578830718994, + "learning_rate": 1.1413353623739777e-07, + "loss": 2.5169, + "step": 60 + }, + { + "epoch": 0.0013315832524872073, + "grad_norm": 8.701460838317871, + "learning_rate": 1.3315579227696404e-07, + "loss": 2.6031, + "step": 70 + }, + { + "epoch": 0.0015218094314139513, + "grad_norm": 19.97081184387207, + "learning_rate": 1.5217804831653035e-07, + "loss": 3.1127, + "step": 80 + }, + { + "epoch": 0.001712035610340695, + "grad_norm": 14.284954071044922, + "learning_rate": 1.7120030435609666e-07, + "loss": 3.9951, + "step": 90 + }, + { + "epoch": 0.001902261789267439, + "grad_norm": 25.341217041015625, + "learning_rate": 1.9022256039566295e-07, + "loss": 5.4591, + "step": 100 + }, + { + "epoch": 0.002092487968194183, + "grad_norm": 5.924285411834717, + "learning_rate": 2.0924481643522923e-07, + "loss": 2.4404, + "step": 110 + }, + { + "epoch": 0.002282714147120927, + "grad_norm": 11.63301944732666, + "learning_rate": 2.2826707247479554e-07, + "loss": 2.5059, + "step": 120 + }, + { + "epoch": 0.0024729403260476705, + "grad_norm": 11.728666305541992, + "learning_rate": 2.472893285143618e-07, + "loss": 3.1191, + "step": 130 + }, + { + "epoch": 0.0026631665049744145, + "grad_norm": 9.490508079528809, + "learning_rate": 2.663115845539281e-07, + "loss": 3.8237, + "step": 140 + }, + { + "epoch": 0.0028533926839011585, + "grad_norm": 24.791996002197266, + "learning_rate": 2.853338405934944e-07, + "loss": 5.5867, + "step": 150 + }, + { + "epoch": 0.0030436188628279025, + "grad_norm": 4.845604419708252, + "learning_rate": 3.043560966330607e-07, + "loss": 2.3633, + "step": 160 + }, + { + "epoch": 0.003233845041754646, + "grad_norm": 6.938657283782959, + "learning_rate": 3.2337835267262704e-07, + "loss": 2.3907, + "step": 170 + }, + { + "epoch": 0.00342407122068139, + "grad_norm": 8.972224235534668, + "learning_rate": 3.424006087121933e-07, + "loss": 2.6722, + "step": 180 + }, + { + "epoch": 0.003614297399608134, + "grad_norm": 9.294358253479004, + "learning_rate": 3.614228647517596e-07, + "loss": 3.4523, + "step": 190 + }, + { + "epoch": 0.003804523578534878, + "grad_norm": 19.425222396850586, + "learning_rate": 3.804451207913259e-07, + "loss": 5.2828, + "step": 200 + }, + { + "epoch": 0.003994749757461622, + "grad_norm": 3.1245577335357666, + "learning_rate": 3.994673768308922e-07, + "loss": 2.191, + "step": 210 + }, + { + "epoch": 0.004184975936388366, + "grad_norm": 4.416078567504883, + "learning_rate": 4.1848963287045846e-07, + "loss": 2.4138, + "step": 220 + }, + { + "epoch": 0.00437520211531511, + "grad_norm": 8.128155708312988, + "learning_rate": 4.3751188891002475e-07, + "loss": 2.4941, + "step": 230 + }, + { + "epoch": 0.004565428294241854, + "grad_norm": 5.850177764892578, + "learning_rate": 4.565341449495911e-07, + "loss": 3.153, + "step": 240 + }, + { + "epoch": 0.004755654473168598, + "grad_norm": 19.345319747924805, + "learning_rate": 4.7555640098915737e-07, + "loss": 5.0348, + "step": 250 + }, + { + "epoch": 0.004945880652095341, + "grad_norm": 2.327831983566284, + "learning_rate": 4.945786570287236e-07, + "loss": 2.1702, + "step": 260 + }, + { + "epoch": 0.005136106831022085, + "grad_norm": 2.7916853427886963, + "learning_rate": 5.136009130682899e-07, + "loss": 2.1161, + "step": 270 + }, + { + "epoch": 0.005326333009948829, + "grad_norm": 3.9358408451080322, + "learning_rate": 5.326231691078562e-07, + "loss": 2.2815, + "step": 280 + }, + { + "epoch": 0.005516559188875573, + "grad_norm": 4.891003131866455, + "learning_rate": 5.516454251474225e-07, + "loss": 2.8836, + "step": 290 + }, + { + "epoch": 0.005706785367802317, + "grad_norm": 22.499956130981445, + "learning_rate": 5.706676811869888e-07, + "loss": 4.7026, + "step": 300 + }, + { + "epoch": 0.005897011546729061, + "grad_norm": 1.9951050281524658, + "learning_rate": 5.896899372265551e-07, + "loss": 2.0446, + "step": 310 + }, + { + "epoch": 0.006087237725655805, + "grad_norm": 3.147122859954834, + "learning_rate": 6.087121932661214e-07, + "loss": 2.138, + "step": 320 + }, + { + "epoch": 0.006277463904582548, + "grad_norm": 3.4201011657714844, + "learning_rate": 6.277344493056876e-07, + "loss": 2.1024, + "step": 330 + }, + { + "epoch": 0.006467690083509292, + "grad_norm": 4.823904991149902, + "learning_rate": 6.467567053452541e-07, + "loss": 2.7573, + "step": 340 + }, + { + "epoch": 0.006657916262436036, + "grad_norm": 21.016569137573242, + "learning_rate": 6.657789613848204e-07, + "loss": 4.1976, + "step": 350 + }, + { + "epoch": 0.00684814244136278, + "grad_norm": 1.872706413269043, + "learning_rate": 6.848012174243867e-07, + "loss": 1.9986, + "step": 360 + }, + { + "epoch": 0.007038368620289524, + "grad_norm": 2.8600945472717285, + "learning_rate": 7.038234734639529e-07, + "loss": 1.9909, + "step": 370 + }, + { + "epoch": 0.007228594799216268, + "grad_norm": 3.7441070079803467, + "learning_rate": 7.228457295035192e-07, + "loss": 2.0065, + "step": 380 + }, + { + "epoch": 0.007418820978143012, + "grad_norm": 5.4876627922058105, + "learning_rate": 7.418679855430855e-07, + "loss": 2.7059, + "step": 390 + }, + { + "epoch": 0.007609047157069756, + "grad_norm": 24.044830322265625, + "learning_rate": 7.608902415826518e-07, + "loss": 4.0807, + "step": 400 + }, + { + "epoch": 0.0077992733359964995, + "grad_norm": 1.6859079599380493, + "learning_rate": 7.799124976222181e-07, + "loss": 1.974, + "step": 410 + }, + { + "epoch": 0.007989499514923244, + "grad_norm": 2.8750579357147217, + "learning_rate": 7.989347536617844e-07, + "loss": 1.835, + "step": 420 + }, + { + "epoch": 0.008179725693849988, + "grad_norm": 4.05927038192749, + "learning_rate": 8.179570097013506e-07, + "loss": 1.9796, + "step": 430 + }, + { + "epoch": 0.008369951872776732, + "grad_norm": 6.277343273162842, + "learning_rate": 8.369792657409169e-07, + "loss": 2.474, + "step": 440 + }, + { + "epoch": 0.008560178051703475, + "grad_norm": 32.71267318725586, + "learning_rate": 8.560015217804832e-07, + "loss": 3.7252, + "step": 450 + }, + { + "epoch": 0.00875040423063022, + "grad_norm": 1.6604670286178589, + "learning_rate": 8.750237778200495e-07, + "loss": 1.851, + "step": 460 + }, + { + "epoch": 0.008940630409556963, + "grad_norm": 2.756211280822754, + "learning_rate": 8.940460338596159e-07, + "loss": 1.6797, + "step": 470 + }, + { + "epoch": 0.009130856588483708, + "grad_norm": 4.476970195770264, + "learning_rate": 9.130682898991822e-07, + "loss": 1.8404, + "step": 480 + }, + { + "epoch": 0.00932108276741045, + "grad_norm": 6.676766395568848, + "learning_rate": 9.320905459387485e-07, + "loss": 2.2254, + "step": 490 + }, + { + "epoch": 0.009511308946337196, + "grad_norm": 33.66838836669922, + "learning_rate": 9.511128019783147e-07, + "loss": 3.3597, + "step": 500 + }, + { + "epoch": 0.009701535125263939, + "grad_norm": 1.771323561668396, + "learning_rate": 9.70135058017881e-07, + "loss": 1.7942, + "step": 510 + }, + { + "epoch": 0.009891761304190682, + "grad_norm": 3.0909101963043213, + "learning_rate": 9.891573140574472e-07, + "loss": 1.7758, + "step": 520 + }, + { + "epoch": 0.010081987483117427, + "grad_norm": 4.743190765380859, + "learning_rate": 1.0081795700970136e-06, + "loss": 1.7564, + "step": 530 + }, + { + "epoch": 0.01027221366204417, + "grad_norm": 7.06041145324707, + "learning_rate": 1.0272018261365798e-06, + "loss": 2.1511, + "step": 540 + }, + { + "epoch": 0.010462439840970915, + "grad_norm": 26.422157287597656, + "learning_rate": 1.0462240821761462e-06, + "loss": 2.8421, + "step": 550 + }, + { + "epoch": 0.010652666019897658, + "grad_norm": 1.8234245777130127, + "learning_rate": 1.0652463382157123e-06, + "loss": 1.7149, + "step": 560 + }, + { + "epoch": 0.010842892198824403, + "grad_norm": 3.059401273727417, + "learning_rate": 1.0842685942552787e-06, + "loss": 1.647, + "step": 570 + }, + { + "epoch": 0.011033118377751146, + "grad_norm": 4.818949222564697, + "learning_rate": 1.103290850294845e-06, + "loss": 1.5044, + "step": 580 + }, + { + "epoch": 0.01122334455667789, + "grad_norm": 6.994847774505615, + "learning_rate": 1.1223131063344113e-06, + "loss": 1.8919, + "step": 590 + }, + { + "epoch": 0.011413570735604634, + "grad_norm": 38.19530487060547, + "learning_rate": 1.1413353623739777e-06, + "loss": 2.4963, + "step": 600 + }, + { + "epoch": 0.011603796914531377, + "grad_norm": 2.0052971839904785, + "learning_rate": 1.160357618413544e-06, + "loss": 1.6709, + "step": 610 + }, + { + "epoch": 0.011794023093458122, + "grad_norm": 1.6200284957885742, + "learning_rate": 1.1793798744531103e-06, + "loss": 1.5749, + "step": 620 + }, + { + "epoch": 0.011984249272384865, + "grad_norm": 2.2375597953796387, + "learning_rate": 1.1984021304926766e-06, + "loss": 1.4411, + "step": 630 + }, + { + "epoch": 0.01217447545131161, + "grad_norm": 5.535067558288574, + "learning_rate": 1.2174243865322428e-06, + "loss": 1.845, + "step": 640 + }, + { + "epoch": 0.012364701630238353, + "grad_norm": 9.7356595993042, + "learning_rate": 1.2364466425718092e-06, + "loss": 2.524, + "step": 650 + }, + { + "epoch": 0.012554927809165096, + "grad_norm": 1.4515656232833862, + "learning_rate": 1.2554688986113752e-06, + "loss": 1.6263, + "step": 660 + }, + { + "epoch": 0.012745153988091841, + "grad_norm": 1.6473021507263184, + "learning_rate": 1.2744911546509418e-06, + "loss": 1.5024, + "step": 670 + }, + { + "epoch": 0.012935380167018585, + "grad_norm": 2.3517134189605713, + "learning_rate": 1.2935134106905082e-06, + "loss": 1.4407, + "step": 680 + }, + { + "epoch": 0.01312560634594533, + "grad_norm": 2.8996622562408447, + "learning_rate": 1.3125356667300743e-06, + "loss": 1.6358, + "step": 690 + }, + { + "epoch": 0.013315832524872073, + "grad_norm": 11.34484577178955, + "learning_rate": 1.3315579227696407e-06, + "loss": 2.2334, + "step": 700 + }, + { + "epoch": 0.013506058703798817, + "grad_norm": 1.1996574401855469, + "learning_rate": 1.350580178809207e-06, + "loss": 1.4753, + "step": 710 + }, + { + "epoch": 0.01369628488272556, + "grad_norm": 1.6772912740707397, + "learning_rate": 1.3696024348487733e-06, + "loss": 1.5256, + "step": 720 + }, + { + "epoch": 0.013886511061652304, + "grad_norm": 2.156625986099243, + "learning_rate": 1.3886246908883395e-06, + "loss": 1.4419, + "step": 730 + }, + { + "epoch": 0.014076737240579049, + "grad_norm": 3.710660219192505, + "learning_rate": 1.4076469469279059e-06, + "loss": 1.6054, + "step": 740 + }, + { + "epoch": 0.014266963419505792, + "grad_norm": 7.250610828399658, + "learning_rate": 1.426669202967472e-06, + "loss": 2.1039, + "step": 750 + }, + { + "epoch": 0.014457189598432537, + "grad_norm": 1.4629435539245605, + "learning_rate": 1.4456914590070384e-06, + "loss": 1.5061, + "step": 760 + }, + { + "epoch": 0.01464741577735928, + "grad_norm": 1.7025142908096313, + "learning_rate": 1.4647137150466046e-06, + "loss": 1.4471, + "step": 770 + }, + { + "epoch": 0.014837641956286025, + "grad_norm": 2.418815851211548, + "learning_rate": 1.483735971086171e-06, + "loss": 1.3384, + "step": 780 + }, + { + "epoch": 0.015027868135212768, + "grad_norm": 4.666192531585693, + "learning_rate": 1.5027582271257372e-06, + "loss": 1.599, + "step": 790 + }, + { + "epoch": 0.015218094314139513, + "grad_norm": 8.951377868652344, + "learning_rate": 1.5217804831653036e-06, + "loss": 1.9391, + "step": 800 + }, + { + "epoch": 0.015408320493066256, + "grad_norm": 1.383745551109314, + "learning_rate": 1.54080273920487e-06, + "loss": 1.5076, + "step": 810 + }, + { + "epoch": 0.015598546671992999, + "grad_norm": 1.4020535945892334, + "learning_rate": 1.5598249952444361e-06, + "loss": 1.3977, + "step": 820 + }, + { + "epoch": 0.015788772850919744, + "grad_norm": 2.2491354942321777, + "learning_rate": 1.5788472512840025e-06, + "loss": 1.3356, + "step": 830 + }, + { + "epoch": 0.015978999029846487, + "grad_norm": 3.5490710735321045, + "learning_rate": 1.5978695073235687e-06, + "loss": 1.5511, + "step": 840 + }, + { + "epoch": 0.01616922520877323, + "grad_norm": 11.946330070495605, + "learning_rate": 1.616891763363135e-06, + "loss": 1.8743, + "step": 850 + }, + { + "epoch": 0.016359451387699977, + "grad_norm": 1.4720561504364014, + "learning_rate": 1.6359140194027013e-06, + "loss": 1.4478, + "step": 860 + }, + { + "epoch": 0.01654967756662672, + "grad_norm": 1.6156224012374878, + "learning_rate": 1.6549362754422677e-06, + "loss": 1.4235, + "step": 870 + }, + { + "epoch": 0.016739903745553463, + "grad_norm": 2.14105224609375, + "learning_rate": 1.6739585314818339e-06, + "loss": 1.3183, + "step": 880 + }, + { + "epoch": 0.016930129924480206, + "grad_norm": 4.404923915863037, + "learning_rate": 1.6929807875214002e-06, + "loss": 1.3946, + "step": 890 + }, + { + "epoch": 0.01712035610340695, + "grad_norm": 7.182434558868408, + "learning_rate": 1.7120030435609664e-06, + "loss": 1.8036, + "step": 900 + }, + { + "epoch": 0.017310582282333696, + "grad_norm": 1.494486689567566, + "learning_rate": 1.7310252996005328e-06, + "loss": 1.4478, + "step": 910 + }, + { + "epoch": 0.01750080846126044, + "grad_norm": 1.6941497325897217, + "learning_rate": 1.750047555640099e-06, + "loss": 1.3241, + "step": 920 + }, + { + "epoch": 0.017691034640187182, + "grad_norm": 2.1194875240325928, + "learning_rate": 1.7690698116796654e-06, + "loss": 1.189, + "step": 930 + }, + { + "epoch": 0.017881260819113926, + "grad_norm": 2.8642826080322266, + "learning_rate": 1.7880920677192318e-06, + "loss": 1.3617, + "step": 940 + }, + { + "epoch": 0.01807148699804067, + "grad_norm": 9.907108306884766, + "learning_rate": 1.807114323758798e-06, + "loss": 1.7392, + "step": 950 + }, + { + "epoch": 0.018261713176967415, + "grad_norm": 1.627024531364441, + "learning_rate": 1.8261365797983643e-06, + "loss": 1.4536, + "step": 960 + }, + { + "epoch": 0.01845193935589416, + "grad_norm": 1.7732610702514648, + "learning_rate": 1.8451588358379305e-06, + "loss": 1.3351, + "step": 970 + }, + { + "epoch": 0.0186421655348209, + "grad_norm": 2.412036180496216, + "learning_rate": 1.864181091877497e-06, + "loss": 1.2028, + "step": 980 + }, + { + "epoch": 0.018832391713747645, + "grad_norm": 4.754890441894531, + "learning_rate": 1.883203347917063e-06, + "loss": 1.3517, + "step": 990 + }, + { + "epoch": 0.01902261789267439, + "grad_norm": 8.927350997924805, + "learning_rate": 1.9022256039566295e-06, + "loss": 1.6531, + "step": 1000 + }, + { + "epoch": 0.019212844071601134, + "grad_norm": 1.6348631381988525, + "learning_rate": 1.9212478599961954e-06, + "loss": 1.4934, + "step": 1010 + }, + { + "epoch": 0.019403070250527878, + "grad_norm": 1.9770840406417847, + "learning_rate": 1.940270116035762e-06, + "loss": 1.3571, + "step": 1020 + }, + { + "epoch": 0.01959329642945462, + "grad_norm": 2.404418468475342, + "learning_rate": 1.9592923720753282e-06, + "loss": 1.1291, + "step": 1030 + }, + { + "epoch": 0.019783522608381364, + "grad_norm": 3.4404196739196777, + "learning_rate": 1.9783146281148944e-06, + "loss": 1.2498, + "step": 1040 + }, + { + "epoch": 0.01997374878730811, + "grad_norm": 9.02733325958252, + "learning_rate": 1.9973368841544606e-06, + "loss": 1.5646, + "step": 1050 + }, + { + "epoch": 0.020163974966234854, + "grad_norm": 1.9403765201568604, + "learning_rate": 2.016359140194027e-06, + "loss": 1.3972, + "step": 1060 + }, + { + "epoch": 0.020354201145161597, + "grad_norm": 2.009138584136963, + "learning_rate": 2.0353813962335938e-06, + "loss": 1.303, + "step": 1070 + }, + { + "epoch": 0.02054442732408834, + "grad_norm": 2.1750669479370117, + "learning_rate": 2.0544036522731595e-06, + "loss": 1.1875, + "step": 1080 + }, + { + "epoch": 0.020734653503015087, + "grad_norm": 3.3346753120422363, + "learning_rate": 2.073425908312726e-06, + "loss": 1.262, + "step": 1090 + }, + { + "epoch": 0.02092487968194183, + "grad_norm": 8.939334869384766, + "learning_rate": 2.0924481643522923e-06, + "loss": 1.547, + "step": 1100 + }, + { + "epoch": 0.021115105860868573, + "grad_norm": 1.614304780960083, + "learning_rate": 2.111470420391859e-06, + "loss": 1.3818, + "step": 1110 + }, + { + "epoch": 0.021305332039795316, + "grad_norm": 1.98068106174469, + "learning_rate": 2.1304926764314247e-06, + "loss": 1.2253, + "step": 1120 + }, + { + "epoch": 0.02149555821872206, + "grad_norm": 2.3504743576049805, + "learning_rate": 2.1495149324709913e-06, + "loss": 1.1185, + "step": 1130 + }, + { + "epoch": 0.021685784397648806, + "grad_norm": 3.188448667526245, + "learning_rate": 2.1685371885105575e-06, + "loss": 1.2133, + "step": 1140 + }, + { + "epoch": 0.02187601057657555, + "grad_norm": 11.752523422241211, + "learning_rate": 2.187559444550124e-06, + "loss": 1.5604, + "step": 1150 + }, + { + "epoch": 0.022066236755502292, + "grad_norm": 1.4586597681045532, + "learning_rate": 2.20658170058969e-06, + "loss": 1.2922, + "step": 1160 + }, + { + "epoch": 0.022256462934429035, + "grad_norm": 1.83343505859375, + "learning_rate": 2.2256039566292564e-06, + "loss": 1.2738, + "step": 1170 + }, + { + "epoch": 0.02244668911335578, + "grad_norm": 2.553067684173584, + "learning_rate": 2.2446262126688226e-06, + "loss": 1.096, + "step": 1180 + }, + { + "epoch": 0.022636915292282525, + "grad_norm": 4.024831771850586, + "learning_rate": 2.263648468708389e-06, + "loss": 1.1661, + "step": 1190 + }, + { + "epoch": 0.022827141471209268, + "grad_norm": 11.252798080444336, + "learning_rate": 2.2826707247479554e-06, + "loss": 1.4848, + "step": 1200 + }, + { + "epoch": 0.02301736765013601, + "grad_norm": 2.0187082290649414, + "learning_rate": 2.3016929807875215e-06, + "loss": 1.2914, + "step": 1210 + }, + { + "epoch": 0.023207593829062755, + "grad_norm": 2.2791354656219482, + "learning_rate": 2.320715236827088e-06, + "loss": 1.2912, + "step": 1220 + }, + { + "epoch": 0.0233978200079895, + "grad_norm": 2.755721092224121, + "learning_rate": 2.3397374928666543e-06, + "loss": 1.1446, + "step": 1230 + }, + { + "epoch": 0.023588046186916244, + "grad_norm": 4.635268688201904, + "learning_rate": 2.3587597489062205e-06, + "loss": 1.156, + "step": 1240 + }, + { + "epoch": 0.023778272365842987, + "grad_norm": 11.180182456970215, + "learning_rate": 2.3777820049457867e-06, + "loss": 1.5128, + "step": 1250 + }, + { + "epoch": 0.02396849854476973, + "grad_norm": 2.6687514781951904, + "learning_rate": 2.3968042609853533e-06, + "loss": 1.3078, + "step": 1260 + }, + { + "epoch": 0.024158724723696474, + "grad_norm": 2.1617326736450195, + "learning_rate": 2.4158265170249195e-06, + "loss": 1.187, + "step": 1270 + }, + { + "epoch": 0.02434895090262322, + "grad_norm": 2.7917706966400146, + "learning_rate": 2.4348487730644856e-06, + "loss": 1.1008, + "step": 1280 + }, + { + "epoch": 0.024539177081549964, + "grad_norm": 3.9902455806732178, + "learning_rate": 2.453871029104052e-06, + "loss": 1.1694, + "step": 1290 + }, + { + "epoch": 0.024729403260476707, + "grad_norm": 10.19775104522705, + "learning_rate": 2.4728932851436184e-06, + "loss": 1.4882, + "step": 1300 + }, + { + "epoch": 0.02491962943940345, + "grad_norm": 1.815943956375122, + "learning_rate": 2.4919155411831846e-06, + "loss": 1.3296, + "step": 1310 + }, + { + "epoch": 0.025109855618330193, + "grad_norm": 1.7802958488464355, + "learning_rate": 2.5109377972227504e-06, + "loss": 1.1588, + "step": 1320 + }, + { + "epoch": 0.02530008179725694, + "grad_norm": 2.8108973503112793, + "learning_rate": 2.529960053262317e-06, + "loss": 1.0912, + "step": 1330 + }, + { + "epoch": 0.025490307976183683, + "grad_norm": 3.822803497314453, + "learning_rate": 2.5489823093018836e-06, + "loss": 1.1713, + "step": 1340 + }, + { + "epoch": 0.025680534155110426, + "grad_norm": 10.885619163513184, + "learning_rate": 2.5680045653414497e-06, + "loss": 1.4048, + "step": 1350 + }, + { + "epoch": 0.02587076033403717, + "grad_norm": 1.5321264266967773, + "learning_rate": 2.5870268213810163e-06, + "loss": 1.2555, + "step": 1360 + }, + { + "epoch": 0.026060986512963916, + "grad_norm": 2.115370512008667, + "learning_rate": 2.606049077420582e-06, + "loss": 1.1987, + "step": 1370 + }, + { + "epoch": 0.02625121269189066, + "grad_norm": 2.6233675479888916, + "learning_rate": 2.6250713334601487e-06, + "loss": 1.0758, + "step": 1380 + }, + { + "epoch": 0.026441438870817402, + "grad_norm": 3.8407323360443115, + "learning_rate": 2.644093589499715e-06, + "loss": 1.1887, + "step": 1390 + }, + { + "epoch": 0.026631665049744145, + "grad_norm": 12.603764533996582, + "learning_rate": 2.6631158455392815e-06, + "loss": 1.4082, + "step": 1400 + }, + { + "epoch": 0.026821891228670888, + "grad_norm": 1.6481189727783203, + "learning_rate": 2.6821381015788472e-06, + "loss": 1.2527, + "step": 1410 + }, + { + "epoch": 0.027012117407597635, + "grad_norm": 2.1282284259796143, + "learning_rate": 2.701160357618414e-06, + "loss": 1.2139, + "step": 1420 + }, + { + "epoch": 0.027202343586524378, + "grad_norm": 2.9524528980255127, + "learning_rate": 2.72018261365798e-06, + "loss": 1.09, + "step": 1430 + }, + { + "epoch": 0.02739256976545112, + "grad_norm": 4.275539398193359, + "learning_rate": 2.7392048696975466e-06, + "loss": 1.1149, + "step": 1440 + }, + { + "epoch": 0.027582795944377864, + "grad_norm": 12.224852561950684, + "learning_rate": 2.7582271257371124e-06, + "loss": 1.3974, + "step": 1450 + }, + { + "epoch": 0.027773022123304607, + "grad_norm": 2.3314363956451416, + "learning_rate": 2.777249381776679e-06, + "loss": 1.3061, + "step": 1460 + }, + { + "epoch": 0.027963248302231354, + "grad_norm": 2.170639991760254, + "learning_rate": 2.796271637816245e-06, + "loss": 1.2254, + "step": 1470 + }, + { + "epoch": 0.028153474481158097, + "grad_norm": 2.8608880043029785, + "learning_rate": 2.8152938938558117e-06, + "loss": 1.0807, + "step": 1480 + }, + { + "epoch": 0.02834370066008484, + "grad_norm": 3.440014600753784, + "learning_rate": 2.834316149895378e-06, + "loss": 1.1255, + "step": 1490 + }, + { + "epoch": 0.028533926839011584, + "grad_norm": 10.175577163696289, + "learning_rate": 2.853338405934944e-06, + "loss": 1.3387, + "step": 1500 + }, + { + "epoch": 0.02872415301793833, + "grad_norm": 2.22709321975708, + "learning_rate": 2.8723606619745103e-06, + "loss": 1.2654, + "step": 1510 + }, + { + "epoch": 0.028914379196865073, + "grad_norm": 1.8235417604446411, + "learning_rate": 2.891382918014077e-06, + "loss": 1.2287, + "step": 1520 + }, + { + "epoch": 0.029104605375791816, + "grad_norm": 2.5544168949127197, + "learning_rate": 2.910405174053643e-06, + "loss": 0.9776, + "step": 1530 + }, + { + "epoch": 0.02929483155471856, + "grad_norm": 4.163671016693115, + "learning_rate": 2.9294274300932092e-06, + "loss": 1.0837, + "step": 1540 + }, + { + "epoch": 0.029485057733645303, + "grad_norm": 10.698822021484375, + "learning_rate": 2.9484496861327754e-06, + "loss": 1.3228, + "step": 1550 + }, + { + "epoch": 0.02967528391257205, + "grad_norm": 1.6194937229156494, + "learning_rate": 2.967471942172342e-06, + "loss": 1.2534, + "step": 1560 + }, + { + "epoch": 0.029865510091498793, + "grad_norm": 1.9834601879119873, + "learning_rate": 2.986494198211908e-06, + "loss": 1.201, + "step": 1570 + }, + { + "epoch": 0.030055736270425536, + "grad_norm": 2.591292142868042, + "learning_rate": 3.0055164542514744e-06, + "loss": 1.0589, + "step": 1580 + }, + { + "epoch": 0.03024596244935228, + "grad_norm": 4.2067670822143555, + "learning_rate": 3.0245387102910406e-06, + "loss": 1.0603, + "step": 1590 + }, + { + "epoch": 0.030436188628279025, + "grad_norm": 9.938916206359863, + "learning_rate": 3.043560966330607e-06, + "loss": 1.2619, + "step": 1600 + }, + { + "epoch": 0.03062641480720577, + "grad_norm": 2.7832484245300293, + "learning_rate": 3.0625832223701733e-06, + "loss": 1.2533, + "step": 1610 + }, + { + "epoch": 0.030816640986132512, + "grad_norm": 2.117129325866699, + "learning_rate": 3.08160547840974e-06, + "loss": 1.1825, + "step": 1620 + }, + { + "epoch": 0.031006867165059255, + "grad_norm": 3.0919687747955322, + "learning_rate": 3.1006277344493057e-06, + "loss": 1.095, + "step": 1630 + }, + { + "epoch": 0.031197093343985998, + "grad_norm": 4.166520595550537, + "learning_rate": 3.1196499904888723e-06, + "loss": 1.047, + "step": 1640 + }, + { + "epoch": 0.031387319522912745, + "grad_norm": 12.0921630859375, + "learning_rate": 3.1386722465284385e-06, + "loss": 1.3526, + "step": 1650 + }, + { + "epoch": 0.03157754570183949, + "grad_norm": 1.9751317501068115, + "learning_rate": 3.157694502568005e-06, + "loss": 1.2864, + "step": 1660 + }, + { + "epoch": 0.03176777188076623, + "grad_norm": 2.2880234718322754, + "learning_rate": 3.176716758607571e-06, + "loss": 1.1079, + "step": 1670 + }, + { + "epoch": 0.031957998059692974, + "grad_norm": 2.804194688796997, + "learning_rate": 3.1957390146471374e-06, + "loss": 0.9666, + "step": 1680 + }, + { + "epoch": 0.03214822423861972, + "grad_norm": 4.810030937194824, + "learning_rate": 3.2147612706867036e-06, + "loss": 1.0979, + "step": 1690 + }, + { + "epoch": 0.03233845041754646, + "grad_norm": 9.718046188354492, + "learning_rate": 3.23378352672627e-06, + "loss": 1.2383, + "step": 1700 + }, + { + "epoch": 0.032528676596473204, + "grad_norm": 2.0624125003814697, + "learning_rate": 3.252805782765836e-06, + "loss": 1.2589, + "step": 1710 + }, + { + "epoch": 0.032718902775399954, + "grad_norm": 2.8751795291900635, + "learning_rate": 3.2718280388054026e-06, + "loss": 1.1319, + "step": 1720 + }, + { + "epoch": 0.0329091289543267, + "grad_norm": 2.805138349533081, + "learning_rate": 3.2908502948449687e-06, + "loss": 0.9922, + "step": 1730 + }, + { + "epoch": 0.03309935513325344, + "grad_norm": 4.693927764892578, + "learning_rate": 3.3098725508845353e-06, + "loss": 1.0518, + "step": 1740 + }, + { + "epoch": 0.03328958131218018, + "grad_norm": 9.922603607177734, + "learning_rate": 3.3288948069241015e-06, + "loss": 1.2417, + "step": 1750 + }, + { + "epoch": 0.033479807491106926, + "grad_norm": 1.9441560506820679, + "learning_rate": 3.3479170629636677e-06, + "loss": 1.2156, + "step": 1760 + }, + { + "epoch": 0.03367003367003367, + "grad_norm": 2.1576952934265137, + "learning_rate": 3.366939319003234e-06, + "loss": 1.1009, + "step": 1770 + }, + { + "epoch": 0.03386025984896041, + "grad_norm": 2.93638014793396, + "learning_rate": 3.3859615750428005e-06, + "loss": 0.9806, + "step": 1780 + }, + { + "epoch": 0.034050486027887156, + "grad_norm": 4.772079944610596, + "learning_rate": 3.4049838310823667e-06, + "loss": 1.0505, + "step": 1790 + }, + { + "epoch": 0.0342407122068139, + "grad_norm": 13.75841236114502, + "learning_rate": 3.424006087121933e-06, + "loss": 1.2653, + "step": 1800 + }, + { + "epoch": 0.03443093838574065, + "grad_norm": 2.011861801147461, + "learning_rate": 3.443028343161499e-06, + "loss": 1.1962, + "step": 1810 + }, + { + "epoch": 0.03462116456466739, + "grad_norm": 2.2049922943115234, + "learning_rate": 3.4620505992010656e-06, + "loss": 1.1636, + "step": 1820 + }, + { + "epoch": 0.034811390743594135, + "grad_norm": 3.468641757965088, + "learning_rate": 3.481072855240632e-06, + "loss": 1.006, + "step": 1830 + }, + { + "epoch": 0.03500161692252088, + "grad_norm": 5.194962024688721, + "learning_rate": 3.500095111280198e-06, + "loss": 1.035, + "step": 1840 + }, + { + "epoch": 0.03519184310144762, + "grad_norm": 9.455286979675293, + "learning_rate": 3.519117367319764e-06, + "loss": 1.1639, + "step": 1850 + }, + { + "epoch": 0.035382069280374365, + "grad_norm": 2.2892448902130127, + "learning_rate": 3.5381396233593308e-06, + "loss": 1.2648, + "step": 1860 + }, + { + "epoch": 0.03557229545930111, + "grad_norm": 2.668325185775757, + "learning_rate": 3.557161879398897e-06, + "loss": 1.1385, + "step": 1870 + }, + { + "epoch": 0.03576252163822785, + "grad_norm": 2.897562265396118, + "learning_rate": 3.5761841354384635e-06, + "loss": 0.9518, + "step": 1880 + }, + { + "epoch": 0.035952747817154594, + "grad_norm": 5.355794906616211, + "learning_rate": 3.5952063914780293e-06, + "loss": 1.059, + "step": 1890 + }, + { + "epoch": 0.03614297399608134, + "grad_norm": 11.731189727783203, + "learning_rate": 3.614228647517596e-06, + "loss": 1.2406, + "step": 1900 + }, + { + "epoch": 0.03633320017500809, + "grad_norm": 2.4913065433502197, + "learning_rate": 3.633250903557162e-06, + "loss": 1.1554, + "step": 1910 + }, + { + "epoch": 0.03652342635393483, + "grad_norm": 2.4660778045654297, + "learning_rate": 3.6522731595967287e-06, + "loss": 1.1604, + "step": 1920 + }, + { + "epoch": 0.036713652532861574, + "grad_norm": 2.8727564811706543, + "learning_rate": 3.6712954156362944e-06, + "loss": 0.986, + "step": 1930 + }, + { + "epoch": 0.03690387871178832, + "grad_norm": 5.380633354187012, + "learning_rate": 3.690317671675861e-06, + "loss": 0.9519, + "step": 1940 + }, + { + "epoch": 0.03709410489071506, + "grad_norm": 9.196549415588379, + "learning_rate": 3.709339927715427e-06, + "loss": 1.253, + "step": 1950 + }, + { + "epoch": 0.0372843310696418, + "grad_norm": 1.7812504768371582, + "learning_rate": 3.728362183754994e-06, + "loss": 1.2104, + "step": 1960 + }, + { + "epoch": 0.037474557248568546, + "grad_norm": 2.1316113471984863, + "learning_rate": 3.7473844397945596e-06, + "loss": 1.1207, + "step": 1970 + }, + { + "epoch": 0.03766478342749529, + "grad_norm": 2.813387155532837, + "learning_rate": 3.766406695834126e-06, + "loss": 0.9173, + "step": 1980 + }, + { + "epoch": 0.03785500960642203, + "grad_norm": 4.0899128913879395, + "learning_rate": 3.7854289518736923e-06, + "loss": 0.9493, + "step": 1990 + }, + { + "epoch": 0.03804523578534878, + "grad_norm": 10.877523422241211, + "learning_rate": 3.804451207913259e-06, + "loss": 1.2015, + "step": 2000 + }, + { + "epoch": 0.038235461964275526, + "grad_norm": 2.7726292610168457, + "learning_rate": 3.8234734639528255e-06, + "loss": 1.2061, + "step": 2010 + }, + { + "epoch": 0.03842568814320227, + "grad_norm": 2.1731584072113037, + "learning_rate": 3.842495719992391e-06, + "loss": 1.106, + "step": 2020 + }, + { + "epoch": 0.03861591432212901, + "grad_norm": 3.7517712116241455, + "learning_rate": 3.861517976031958e-06, + "loss": 0.9798, + "step": 2030 + }, + { + "epoch": 0.038806140501055755, + "grad_norm": 5.391069412231445, + "learning_rate": 3.880540232071524e-06, + "loss": 0.9718, + "step": 2040 + }, + { + "epoch": 0.0389963666799825, + "grad_norm": 9.744902610778809, + "learning_rate": 3.89956248811109e-06, + "loss": 1.2262, + "step": 2050 + }, + { + "epoch": 0.03918659285890924, + "grad_norm": 2.6106269359588623, + "learning_rate": 3.9185847441506564e-06, + "loss": 1.2071, + "step": 2060 + }, + { + "epoch": 0.039376819037835985, + "grad_norm": 3.034111738204956, + "learning_rate": 3.937607000190223e-06, + "loss": 1.1291, + "step": 2070 + }, + { + "epoch": 0.03956704521676273, + "grad_norm": 3.5086514949798584, + "learning_rate": 3.956629256229789e-06, + "loss": 0.9248, + "step": 2080 + }, + { + "epoch": 0.03975727139568948, + "grad_norm": 4.488050937652588, + "learning_rate": 3.975651512269356e-06, + "loss": 1.0084, + "step": 2090 + }, + { + "epoch": 0.03994749757461622, + "grad_norm": 10.10869312286377, + "learning_rate": 3.994673768308921e-06, + "loss": 1.188, + "step": 2100 + }, + { + "epoch": 0.040137723753542964, + "grad_norm": 2.787724256515503, + "learning_rate": 4.013696024348488e-06, + "loss": 1.153, + "step": 2110 + }, + { + "epoch": 0.04032794993246971, + "grad_norm": 2.309357166290283, + "learning_rate": 4.032718280388054e-06, + "loss": 1.0842, + "step": 2120 + }, + { + "epoch": 0.04051817611139645, + "grad_norm": 2.9359734058380127, + "learning_rate": 4.0517405364276205e-06, + "loss": 0.9549, + "step": 2130 + }, + { + "epoch": 0.040708402290323194, + "grad_norm": 4.532121658325195, + "learning_rate": 4.0707627924671876e-06, + "loss": 0.9548, + "step": 2140 + }, + { + "epoch": 0.04089862846924994, + "grad_norm": 9.86694049835205, + "learning_rate": 4.089785048506753e-06, + "loss": 1.1928, + "step": 2150 + }, + { + "epoch": 0.04108885464817668, + "grad_norm": 2.4668540954589844, + "learning_rate": 4.108807304546319e-06, + "loss": 1.2318, + "step": 2160 + }, + { + "epoch": 0.04127908082710342, + "grad_norm": 2.501939296722412, + "learning_rate": 4.127829560585886e-06, + "loss": 1.11, + "step": 2170 + }, + { + "epoch": 0.04146930700603017, + "grad_norm": 2.8509225845336914, + "learning_rate": 4.146851816625452e-06, + "loss": 0.8994, + "step": 2180 + }, + { + "epoch": 0.041659533184956916, + "grad_norm": 4.271004676818848, + "learning_rate": 4.1658740726650185e-06, + "loss": 0.9308, + "step": 2190 + }, + { + "epoch": 0.04184975936388366, + "grad_norm": 8.929543495178223, + "learning_rate": 4.184896328704585e-06, + "loss": 1.1605, + "step": 2200 + }, + { + "epoch": 0.0420399855428104, + "grad_norm": 2.176539421081543, + "learning_rate": 4.203918584744151e-06, + "loss": 1.1261, + "step": 2210 + }, + { + "epoch": 0.042230211721737146, + "grad_norm": 3.0560004711151123, + "learning_rate": 4.222940840783718e-06, + "loss": 1.1041, + "step": 2220 + }, + { + "epoch": 0.04242043790066389, + "grad_norm": 2.956686496734619, + "learning_rate": 4.241963096823283e-06, + "loss": 0.9188, + "step": 2230 + }, + { + "epoch": 0.04261066407959063, + "grad_norm": 4.251719951629639, + "learning_rate": 4.260985352862849e-06, + "loss": 1.0, + "step": 2240 + }, + { + "epoch": 0.042800890258517375, + "grad_norm": 11.712019920349121, + "learning_rate": 4.280007608902416e-06, + "loss": 1.1977, + "step": 2250 + }, + { + "epoch": 0.04299111643744412, + "grad_norm": 3.014216899871826, + "learning_rate": 4.2990298649419825e-06, + "loss": 1.175, + "step": 2260 + }, + { + "epoch": 0.04318134261637086, + "grad_norm": 2.524491786956787, + "learning_rate": 4.318052120981549e-06, + "loss": 1.0855, + "step": 2270 + }, + { + "epoch": 0.04337156879529761, + "grad_norm": 3.5122768878936768, + "learning_rate": 4.337074377021115e-06, + "loss": 0.8957, + "step": 2280 + }, + { + "epoch": 0.043561794974224355, + "grad_norm": 5.294569492340088, + "learning_rate": 4.356096633060681e-06, + "loss": 0.9801, + "step": 2290 + }, + { + "epoch": 0.0437520211531511, + "grad_norm": 12.318851470947266, + "learning_rate": 4.375118889100248e-06, + "loss": 1.2072, + "step": 2300 + }, + { + "epoch": 0.04394224733207784, + "grad_norm": 3.4695818424224854, + "learning_rate": 4.394141145139814e-06, + "loss": 1.1855, + "step": 2310 + }, + { + "epoch": 0.044132473511004584, + "grad_norm": 2.193481683731079, + "learning_rate": 4.41316340117938e-06, + "loss": 1.0368, + "step": 2320 + }, + { + "epoch": 0.04432269968993133, + "grad_norm": 3.3650193214416504, + "learning_rate": 4.432185657218947e-06, + "loss": 0.9339, + "step": 2330 + }, + { + "epoch": 0.04451292586885807, + "grad_norm": 4.817994594573975, + "learning_rate": 4.451207913258513e-06, + "loss": 0.9878, + "step": 2340 + }, + { + "epoch": 0.044703152047784814, + "grad_norm": 11.410785675048828, + "learning_rate": 4.470230169298079e-06, + "loss": 1.1553, + "step": 2350 + }, + { + "epoch": 0.04489337822671156, + "grad_norm": 2.329460620880127, + "learning_rate": 4.489252425337645e-06, + "loss": 1.1137, + "step": 2360 + }, + { + "epoch": 0.04508360440563831, + "grad_norm": 4.1374287605285645, + "learning_rate": 4.508274681377211e-06, + "loss": 1.0883, + "step": 2370 + }, + { + "epoch": 0.04527383058456505, + "grad_norm": 3.188446521759033, + "learning_rate": 4.527296937416778e-06, + "loss": 0.9459, + "step": 2380 + }, + { + "epoch": 0.04546405676349179, + "grad_norm": 6.629786014556885, + "learning_rate": 4.5463191934563446e-06, + "loss": 0.9519, + "step": 2390 + }, + { + "epoch": 0.045654282942418536, + "grad_norm": 12.22085189819336, + "learning_rate": 4.565341449495911e-06, + "loss": 1.1074, + "step": 2400 + }, + { + "epoch": 0.04584450912134528, + "grad_norm": 3.0764682292938232, + "learning_rate": 4.584363705535477e-06, + "loss": 1.1458, + "step": 2410 + }, + { + "epoch": 0.04603473530027202, + "grad_norm": 2.767094135284424, + "learning_rate": 4.603385961575043e-06, + "loss": 1.0658, + "step": 2420 + }, + { + "epoch": 0.046224961479198766, + "grad_norm": 3.757636308670044, + "learning_rate": 4.622408217614609e-06, + "loss": 0.9337, + "step": 2430 + }, + { + "epoch": 0.04641518765812551, + "grad_norm": 5.4339447021484375, + "learning_rate": 4.641430473654176e-06, + "loss": 0.9154, + "step": 2440 + }, + { + "epoch": 0.04660541383705225, + "grad_norm": 11.860993385314941, + "learning_rate": 4.660452729693742e-06, + "loss": 1.1127, + "step": 2450 + }, + { + "epoch": 0.046795640015979, + "grad_norm": 2.824115037918091, + "learning_rate": 4.679474985733309e-06, + "loss": 1.1366, + "step": 2460 + }, + { + "epoch": 0.046985866194905745, + "grad_norm": 2.538586139678955, + "learning_rate": 4.698497241772875e-06, + "loss": 1.0307, + "step": 2470 + }, + { + "epoch": 0.04717609237383249, + "grad_norm": 3.817042589187622, + "learning_rate": 4.717519497812441e-06, + "loss": 0.9, + "step": 2480 + }, + { + "epoch": 0.04736631855275923, + "grad_norm": 5.339418888092041, + "learning_rate": 4.736541753852007e-06, + "loss": 0.9951, + "step": 2490 + }, + { + "epoch": 0.047556544731685975, + "grad_norm": 12.489927291870117, + "learning_rate": 4.755564009891573e-06, + "loss": 1.1541, + "step": 2500 + }, + { + "epoch": 0.04774677091061272, + "grad_norm": 2.1420531272888184, + "learning_rate": 4.7745862659311395e-06, + "loss": 1.1063, + "step": 2510 + }, + { + "epoch": 0.04793699708953946, + "grad_norm": 2.2218401432037354, + "learning_rate": 4.7936085219707066e-06, + "loss": 1.0871, + "step": 2520 + }, + { + "epoch": 0.048127223268466204, + "grad_norm": 3.700441360473633, + "learning_rate": 4.812630778010273e-06, + "loss": 0.873, + "step": 2530 + }, + { + "epoch": 0.04831744944739295, + "grad_norm": 5.405223369598389, + "learning_rate": 4.831653034049839e-06, + "loss": 0.985, + "step": 2540 + }, + { + "epoch": 0.04850767562631969, + "grad_norm": 11.029736518859863, + "learning_rate": 4.850675290089405e-06, + "loss": 1.2141, + "step": 2550 + }, + { + "epoch": 0.04869790180524644, + "grad_norm": 2.7142674922943115, + "learning_rate": 4.869697546128971e-06, + "loss": 1.1559, + "step": 2560 + }, + { + "epoch": 0.048888127984173184, + "grad_norm": 2.459038734436035, + "learning_rate": 4.8887198021685375e-06, + "loss": 1.1262, + "step": 2570 + }, + { + "epoch": 0.04907835416309993, + "grad_norm": 3.2301881313323975, + "learning_rate": 4.907742058208104e-06, + "loss": 0.9321, + "step": 2580 + }, + { + "epoch": 0.04926858034202667, + "grad_norm": 5.523041248321533, + "learning_rate": 4.92676431424767e-06, + "loss": 0.9226, + "step": 2590 + }, + { + "epoch": 0.04945880652095341, + "grad_norm": 10.131287574768066, + "learning_rate": 4.945786570287237e-06, + "loss": 1.1409, + "step": 2600 + }, + { + "epoch": 0.049649032699880156, + "grad_norm": 2.8806002140045166, + "learning_rate": 4.964808826326803e-06, + "loss": 1.1364, + "step": 2610 + }, + { + "epoch": 0.0498392588788069, + "grad_norm": 2.400009870529175, + "learning_rate": 4.983831082366369e-06, + "loss": 1.0847, + "step": 2620 + }, + { + "epoch": 0.05002948505773364, + "grad_norm": 4.137524604797363, + "learning_rate": 5.002853338405935e-06, + "loss": 0.9026, + "step": 2630 + }, + { + "epoch": 0.050219711236660386, + "grad_norm": 4.834102153778076, + "learning_rate": 5.021875594445501e-06, + "loss": 0.936, + "step": 2640 + }, + { + "epoch": 0.050409937415587136, + "grad_norm": 10.480986595153809, + "learning_rate": 5.040897850485068e-06, + "loss": 1.1391, + "step": 2650 + }, + { + "epoch": 0.05060016359451388, + "grad_norm": 2.876462697982788, + "learning_rate": 5.059920106524634e-06, + "loss": 1.1186, + "step": 2660 + }, + { + "epoch": 0.05079038977344062, + "grad_norm": 3.254173755645752, + "learning_rate": 5.078942362564201e-06, + "loss": 1.0468, + "step": 2670 + }, + { + "epoch": 0.050980615952367365, + "grad_norm": 3.5466344356536865, + "learning_rate": 5.097964618603767e-06, + "loss": 0.8944, + "step": 2680 + }, + { + "epoch": 0.05117084213129411, + "grad_norm": 4.0647382736206055, + "learning_rate": 5.1169868746433324e-06, + "loss": 0.8855, + "step": 2690 + }, + { + "epoch": 0.05136106831022085, + "grad_norm": 12.590734481811523, + "learning_rate": 5.1360091306828995e-06, + "loss": 1.1354, + "step": 2700 + }, + { + "epoch": 0.051551294489147595, + "grad_norm": 2.858893632888794, + "learning_rate": 5.155031386722466e-06, + "loss": 1.1345, + "step": 2710 + }, + { + "epoch": 0.05174152066807434, + "grad_norm": 2.8196043968200684, + "learning_rate": 5.174053642762033e-06, + "loss": 1.0826, + "step": 2720 + }, + { + "epoch": 0.05193174684700108, + "grad_norm": 3.493431806564331, + "learning_rate": 5.193075898801598e-06, + "loss": 0.8708, + "step": 2730 + }, + { + "epoch": 0.05212197302592783, + "grad_norm": 5.68046236038208, + "learning_rate": 5.212098154841164e-06, + "loss": 0.937, + "step": 2740 + }, + { + "epoch": 0.052312199204854574, + "grad_norm": 13.049890518188477, + "learning_rate": 5.231120410880731e-06, + "loss": 1.1392, + "step": 2750 + }, + { + "epoch": 0.05250242538378132, + "grad_norm": 4.613514423370361, + "learning_rate": 5.250142666920297e-06, + "loss": 1.1284, + "step": 2760 + }, + { + "epoch": 0.05269265156270806, + "grad_norm": 2.3258795738220215, + "learning_rate": 5.269164922959863e-06, + "loss": 1.0144, + "step": 2770 + }, + { + "epoch": 0.052882877741634804, + "grad_norm": 3.8380930423736572, + "learning_rate": 5.28818717899943e-06, + "loss": 0.9067, + "step": 2780 + }, + { + "epoch": 0.05307310392056155, + "grad_norm": 5.028414249420166, + "learning_rate": 5.307209435038996e-06, + "loss": 0.9445, + "step": 2790 + }, + { + "epoch": 0.05326333009948829, + "grad_norm": 9.695432662963867, + "learning_rate": 5.326231691078563e-06, + "loss": 1.0988, + "step": 2800 + }, + { + "epoch": 0.05345355627841503, + "grad_norm": 3.135622262954712, + "learning_rate": 5.345253947118128e-06, + "loss": 1.1195, + "step": 2810 + }, + { + "epoch": 0.053643782457341777, + "grad_norm": 3.463083505630493, + "learning_rate": 5.3642762031576945e-06, + "loss": 1.0298, + "step": 2820 + }, + { + "epoch": 0.05383400863626853, + "grad_norm": 2.976191997528076, + "learning_rate": 5.3832984591972615e-06, + "loss": 0.8963, + "step": 2830 + }, + { + "epoch": 0.05402423481519527, + "grad_norm": 5.485313892364502, + "learning_rate": 5.402320715236828e-06, + "loss": 0.9498, + "step": 2840 + }, + { + "epoch": 0.05421446099412201, + "grad_norm": 12.000988006591797, + "learning_rate": 5.421342971276395e-06, + "loss": 1.1117, + "step": 2850 + }, + { + "epoch": 0.054404687173048756, + "grad_norm": 3.293025493621826, + "learning_rate": 5.44036522731596e-06, + "loss": 1.1391, + "step": 2860 + }, + { + "epoch": 0.0545949133519755, + "grad_norm": 3.216097354888916, + "learning_rate": 5.459387483355526e-06, + "loss": 1.0316, + "step": 2870 + }, + { + "epoch": 0.05478513953090224, + "grad_norm": 3.772177219390869, + "learning_rate": 5.478409739395093e-06, + "loss": 0.8565, + "step": 2880 + }, + { + "epoch": 0.054975365709828986, + "grad_norm": 4.893718242645264, + "learning_rate": 5.4974319954346586e-06, + "loss": 0.8826, + "step": 2890 + }, + { + "epoch": 0.05516559188875573, + "grad_norm": 12.316627502441406, + "learning_rate": 5.516454251474225e-06, + "loss": 1.1345, + "step": 2900 + }, + { + "epoch": 0.05535581806768247, + "grad_norm": 2.428457021713257, + "learning_rate": 5.535476507513792e-06, + "loss": 1.1332, + "step": 2910 + }, + { + "epoch": 0.055546044246609215, + "grad_norm": 2.9872822761535645, + "learning_rate": 5.554498763553358e-06, + "loss": 1.0649, + "step": 2920 + }, + { + "epoch": 0.055736270425535965, + "grad_norm": 3.698500156402588, + "learning_rate": 5.573521019592925e-06, + "loss": 0.9102, + "step": 2930 + }, + { + "epoch": 0.05592649660446271, + "grad_norm": 5.477416515350342, + "learning_rate": 5.59254327563249e-06, + "loss": 0.9307, + "step": 2940 + }, + { + "epoch": 0.05611672278338945, + "grad_norm": 13.570015907287598, + "learning_rate": 5.6115655316720565e-06, + "loss": 1.1574, + "step": 2950 + }, + { + "epoch": 0.056306948962316195, + "grad_norm": 4.1514434814453125, + "learning_rate": 5.6305877877116235e-06, + "loss": 1.0952, + "step": 2960 + }, + { + "epoch": 0.05649717514124294, + "grad_norm": 2.7112059593200684, + "learning_rate": 5.649610043751189e-06, + "loss": 1.1087, + "step": 2970 + }, + { + "epoch": 0.05668740132016968, + "grad_norm": 3.348348617553711, + "learning_rate": 5.668632299790756e-06, + "loss": 0.893, + "step": 2980 + }, + { + "epoch": 0.056877627499096424, + "grad_norm": 5.3222126960754395, + "learning_rate": 5.687654555830322e-06, + "loss": 0.9064, + "step": 2990 + }, + { + "epoch": 0.05706785367802317, + "grad_norm": 11.816397666931152, + "learning_rate": 5.706676811869888e-06, + "loss": 1.0827, + "step": 3000 + }, + { + "epoch": 0.05725807985694991, + "grad_norm": 2.0702426433563232, + "learning_rate": 5.725699067909455e-06, + "loss": 1.144, + "step": 3010 + }, + { + "epoch": 0.05744830603587666, + "grad_norm": 3.0293846130371094, + "learning_rate": 5.7447213239490206e-06, + "loss": 1.0186, + "step": 3020 + }, + { + "epoch": 0.057638532214803403, + "grad_norm": 3.4096548557281494, + "learning_rate": 5.763743579988587e-06, + "loss": 0.8879, + "step": 3030 + }, + { + "epoch": 0.05782875839373015, + "grad_norm": 6.768439769744873, + "learning_rate": 5.782765836028154e-06, + "loss": 0.8991, + "step": 3040 + }, + { + "epoch": 0.05801898457265689, + "grad_norm": 14.614982604980469, + "learning_rate": 5.801788092067719e-06, + "loss": 1.0877, + "step": 3050 + }, + { + "epoch": 0.05820921075158363, + "grad_norm": 2.3279948234558105, + "learning_rate": 5.820810348107286e-06, + "loss": 1.081, + "step": 3060 + }, + { + "epoch": 0.058399436930510376, + "grad_norm": 3.4593429565429688, + "learning_rate": 5.839832604146852e-06, + "loss": 1.0231, + "step": 3070 + }, + { + "epoch": 0.05858966310943712, + "grad_norm": 3.165771484375, + "learning_rate": 5.8588548601864185e-06, + "loss": 0.8792, + "step": 3080 + }, + { + "epoch": 0.05877988928836386, + "grad_norm": 5.759438991546631, + "learning_rate": 5.8778771162259855e-06, + "loss": 0.9131, + "step": 3090 + }, + { + "epoch": 0.058970115467290606, + "grad_norm": 10.794733047485352, + "learning_rate": 5.896899372265551e-06, + "loss": 1.0556, + "step": 3100 + }, + { + "epoch": 0.059160341646217356, + "grad_norm": 2.0568151473999023, + "learning_rate": 5.915921628305118e-06, + "loss": 1.0646, + "step": 3110 + }, + { + "epoch": 0.0593505678251441, + "grad_norm": 3.4895856380462646, + "learning_rate": 5.934943884344684e-06, + "loss": 1.0352, + "step": 3120 + }, + { + "epoch": 0.05954079400407084, + "grad_norm": 3.709806442260742, + "learning_rate": 5.953966140384249e-06, + "loss": 0.8928, + "step": 3130 + }, + { + "epoch": 0.059731020182997585, + "grad_norm": 4.89426326751709, + "learning_rate": 5.972988396423816e-06, + "loss": 0.881, + "step": 3140 + }, + { + "epoch": 0.05992124636192433, + "grad_norm": 13.595807075500488, + "learning_rate": 5.992010652463383e-06, + "loss": 1.0825, + "step": 3150 + }, + { + "epoch": 0.06011147254085107, + "grad_norm": 3.318986177444458, + "learning_rate": 6.011032908502949e-06, + "loss": 1.1474, + "step": 3160 + }, + { + "epoch": 0.060301698719777815, + "grad_norm": 3.450166940689087, + "learning_rate": 6.030055164542516e-06, + "loss": 1.0437, + "step": 3170 + }, + { + "epoch": 0.06049192489870456, + "grad_norm": 3.8872604370117188, + "learning_rate": 6.049077420582081e-06, + "loss": 0.8966, + "step": 3180 + }, + { + "epoch": 0.0606821510776313, + "grad_norm": 4.919650554656982, + "learning_rate": 6.068099676621648e-06, + "loss": 0.9155, + "step": 3190 + }, + { + "epoch": 0.06087237725655805, + "grad_norm": 12.513046264648438, + "learning_rate": 6.087121932661214e-06, + "loss": 1.0846, + "step": 3200 + }, + { + "epoch": 0.061062603435484794, + "grad_norm": 3.5747430324554443, + "learning_rate": 6.10614418870078e-06, + "loss": 1.1239, + "step": 3210 + }, + { + "epoch": 0.06125282961441154, + "grad_norm": 2.480370044708252, + "learning_rate": 6.125166444740347e-06, + "loss": 1.0142, + "step": 3220 + }, + { + "epoch": 0.06144305579333828, + "grad_norm": 3.4333581924438477, + "learning_rate": 6.144188700779913e-06, + "loss": 0.9134, + "step": 3230 + }, + { + "epoch": 0.061633281972265024, + "grad_norm": 5.920953750610352, + "learning_rate": 6.16321095681948e-06, + "loss": 0.9426, + "step": 3240 + }, + { + "epoch": 0.06182350815119177, + "grad_norm": 11.068296432495117, + "learning_rate": 6.182233212859046e-06, + "loss": 0.9753, + "step": 3250 + }, + { + "epoch": 0.06201373433011851, + "grad_norm": 2.670910120010376, + "learning_rate": 6.201255468898611e-06, + "loss": 1.0863, + "step": 3260 + }, + { + "epoch": 0.06220396050904525, + "grad_norm": 2.3734078407287598, + "learning_rate": 6.220277724938178e-06, + "loss": 1.0623, + "step": 3270 + }, + { + "epoch": 0.062394186687971996, + "grad_norm": 5.874599933624268, + "learning_rate": 6.239299980977745e-06, + "loss": 0.8873, + "step": 3280 + }, + { + "epoch": 0.06258441286689874, + "grad_norm": 6.023142337799072, + "learning_rate": 6.25832223701731e-06, + "loss": 0.8765, + "step": 3290 + }, + { + "epoch": 0.06277463904582549, + "grad_norm": 12.106161117553711, + "learning_rate": 6.277344493056877e-06, + "loss": 1.0472, + "step": 3300 + }, + { + "epoch": 0.06296486522475223, + "grad_norm": 2.7873780727386475, + "learning_rate": 6.296366749096443e-06, + "loss": 1.0871, + "step": 3310 + }, + { + "epoch": 0.06315509140367898, + "grad_norm": 2.410355806350708, + "learning_rate": 6.31538900513601e-06, + "loss": 1.0264, + "step": 3320 + }, + { + "epoch": 0.06334531758260571, + "grad_norm": 3.43695068359375, + "learning_rate": 6.334411261175576e-06, + "loss": 0.8591, + "step": 3330 + }, + { + "epoch": 0.06353554376153246, + "grad_norm": 5.210920810699463, + "learning_rate": 6.353433517215142e-06, + "loss": 0.9006, + "step": 3340 + }, + { + "epoch": 0.06372576994045921, + "grad_norm": 13.646373748779297, + "learning_rate": 6.372455773254709e-06, + "loss": 1.039, + "step": 3350 + }, + { + "epoch": 0.06391599611938595, + "grad_norm": 3.9488167762756348, + "learning_rate": 6.391478029294275e-06, + "loss": 1.1264, + "step": 3360 + }, + { + "epoch": 0.0641062222983127, + "grad_norm": 2.9603095054626465, + "learning_rate": 6.410500285333842e-06, + "loss": 1.0785, + "step": 3370 + }, + { + "epoch": 0.06429644847723943, + "grad_norm": 6.125018119812012, + "learning_rate": 6.429522541373407e-06, + "loss": 0.8835, + "step": 3380 + }, + { + "epoch": 0.06448667465616618, + "grad_norm": 6.037325382232666, + "learning_rate": 6.448544797412973e-06, + "loss": 0.9131, + "step": 3390 + }, + { + "epoch": 0.06467690083509292, + "grad_norm": 16.08112144470215, + "learning_rate": 6.46756705345254e-06, + "loss": 1.111, + "step": 3400 + }, + { + "epoch": 0.06486712701401967, + "grad_norm": 2.1979708671569824, + "learning_rate": 6.486589309492107e-06, + "loss": 1.0869, + "step": 3410 + }, + { + "epoch": 0.06505735319294641, + "grad_norm": 3.191075086593628, + "learning_rate": 6.505611565531672e-06, + "loss": 1.0359, + "step": 3420 + }, + { + "epoch": 0.06524757937187316, + "grad_norm": 4.361050128936768, + "learning_rate": 6.524633821571239e-06, + "loss": 0.8945, + "step": 3430 + }, + { + "epoch": 0.06543780555079991, + "grad_norm": 5.246515274047852, + "learning_rate": 6.543656077610805e-06, + "loss": 0.898, + "step": 3440 + }, + { + "epoch": 0.06562803172972664, + "grad_norm": 11.343035697937012, + "learning_rate": 6.562678333650372e-06, + "loss": 1.0329, + "step": 3450 + }, + { + "epoch": 0.0658182579086534, + "grad_norm": 3.319098711013794, + "learning_rate": 6.5817005896899375e-06, + "loss": 1.1156, + "step": 3460 + }, + { + "epoch": 0.06600848408758013, + "grad_norm": 2.946639060974121, + "learning_rate": 6.600722845729504e-06, + "loss": 1.018, + "step": 3470 + }, + { + "epoch": 0.06619871026650688, + "grad_norm": 2.6161160469055176, + "learning_rate": 6.619745101769071e-06, + "loss": 0.8507, + "step": 3480 + }, + { + "epoch": 0.06638893644543362, + "grad_norm": 5.574872970581055, + "learning_rate": 6.638767357808637e-06, + "loss": 0.9083, + "step": 3490 + }, + { + "epoch": 0.06657916262436037, + "grad_norm": 13.42422866821289, + "learning_rate": 6.657789613848203e-06, + "loss": 1.0902, + "step": 3500 + }, + { + "epoch": 0.0667693888032871, + "grad_norm": 2.8455140590667725, + "learning_rate": 6.676811869887769e-06, + "loss": 1.06, + "step": 3510 + }, + { + "epoch": 0.06695961498221385, + "grad_norm": 2.6613621711730957, + "learning_rate": 6.695834125927335e-06, + "loss": 0.989, + "step": 3520 + }, + { + "epoch": 0.0671498411611406, + "grad_norm": 3.7422332763671875, + "learning_rate": 6.7148563819669024e-06, + "loss": 0.8448, + "step": 3530 + }, + { + "epoch": 0.06734006734006734, + "grad_norm": 5.147526264190674, + "learning_rate": 6.733878638006468e-06, + "loss": 0.8717, + "step": 3540 + }, + { + "epoch": 0.06753029351899409, + "grad_norm": 12.03034782409668, + "learning_rate": 6.752900894046034e-06, + "loss": 1.048, + "step": 3550 + }, + { + "epoch": 0.06772051969792083, + "grad_norm": 3.214045763015747, + "learning_rate": 6.771923150085601e-06, + "loss": 1.0937, + "step": 3560 + }, + { + "epoch": 0.06791074587684758, + "grad_norm": 2.7989227771759033, + "learning_rate": 6.790945406125167e-06, + "loss": 1.0337, + "step": 3570 + }, + { + "epoch": 0.06810097205577431, + "grad_norm": 4.597136497497559, + "learning_rate": 6.809967662164733e-06, + "loss": 0.8806, + "step": 3580 + }, + { + "epoch": 0.06829119823470106, + "grad_norm": 4.616443634033203, + "learning_rate": 6.8289899182042995e-06, + "loss": 0.8547, + "step": 3590 + }, + { + "epoch": 0.0684814244136278, + "grad_norm": 11.670774459838867, + "learning_rate": 6.848012174243866e-06, + "loss": 1.0259, + "step": 3600 + }, + { + "epoch": 0.06867165059255455, + "grad_norm": 2.3249611854553223, + "learning_rate": 6.867034430283433e-06, + "loss": 1.045, + "step": 3610 + }, + { + "epoch": 0.0688618767714813, + "grad_norm": 4.721532344818115, + "learning_rate": 6.886056686322998e-06, + "loss": 1.0629, + "step": 3620 + }, + { + "epoch": 0.06905210295040803, + "grad_norm": 3.195577621459961, + "learning_rate": 6.905078942362565e-06, + "loss": 0.8311, + "step": 3630 + }, + { + "epoch": 0.06924232912933478, + "grad_norm": 5.499499320983887, + "learning_rate": 6.924101198402131e-06, + "loss": 0.8811, + "step": 3640 + }, + { + "epoch": 0.06943255530826152, + "grad_norm": 12.689728736877441, + "learning_rate": 6.943123454441697e-06, + "loss": 1.0668, + "step": 3650 + }, + { + "epoch": 0.06962278148718827, + "grad_norm": 2.8369383811950684, + "learning_rate": 6.962145710481264e-06, + "loss": 1.066, + "step": 3660 + }, + { + "epoch": 0.069813007666115, + "grad_norm": 5.4221415519714355, + "learning_rate": 6.98116796652083e-06, + "loss": 1.0518, + "step": 3670 + }, + { + "epoch": 0.07000323384504176, + "grad_norm": 2.920156717300415, + "learning_rate": 7.000190222560396e-06, + "loss": 0.8456, + "step": 3680 + }, + { + "epoch": 0.07019346002396849, + "grad_norm": 6.14293909072876, + "learning_rate": 7.019212478599963e-06, + "loss": 0.8598, + "step": 3690 + }, + { + "epoch": 0.07038368620289524, + "grad_norm": 14.368626594543457, + "learning_rate": 7.038234734639528e-06, + "loss": 1.0146, + "step": 3700 + }, + { + "epoch": 0.070573912381822, + "grad_norm": 3.4393813610076904, + "learning_rate": 7.057256990679095e-06, + "loss": 1.1011, + "step": 3710 + }, + { + "epoch": 0.07076413856074873, + "grad_norm": 2.811824083328247, + "learning_rate": 7.0762792467186615e-06, + "loss": 1.0224, + "step": 3720 + }, + { + "epoch": 0.07095436473967548, + "grad_norm": 3.577205181121826, + "learning_rate": 7.095301502758227e-06, + "loss": 0.8828, + "step": 3730 + }, + { + "epoch": 0.07114459091860222, + "grad_norm": 5.109273910522461, + "learning_rate": 7.114323758797794e-06, + "loss": 0.8378, + "step": 3740 + }, + { + "epoch": 0.07133481709752897, + "grad_norm": 12.113494873046875, + "learning_rate": 7.13334601483736e-06, + "loss": 1.0819, + "step": 3750 + }, + { + "epoch": 0.0715250432764557, + "grad_norm": 2.632756471633911, + "learning_rate": 7.152368270876927e-06, + "loss": 1.0574, + "step": 3760 + }, + { + "epoch": 0.07171526945538245, + "grad_norm": 2.616201162338257, + "learning_rate": 7.171390526916493e-06, + "loss": 1.0205, + "step": 3770 + }, + { + "epoch": 0.07190549563430919, + "grad_norm": 4.030309200286865, + "learning_rate": 7.190412782956059e-06, + "loss": 0.8747, + "step": 3780 + }, + { + "epoch": 0.07209572181323594, + "grad_norm": 5.87321138381958, + "learning_rate": 7.209435038995626e-06, + "loss": 0.9093, + "step": 3790 + }, + { + "epoch": 0.07228594799216267, + "grad_norm": 13.199514389038086, + "learning_rate": 7.228457295035192e-06, + "loss": 1.0841, + "step": 3800 + }, + { + "epoch": 0.07247617417108942, + "grad_norm": 3.2827208042144775, + "learning_rate": 7.247479551074757e-06, + "loss": 1.0602, + "step": 3810 + }, + { + "epoch": 0.07266640035001617, + "grad_norm": 2.744004726409912, + "learning_rate": 7.266501807114324e-06, + "loss": 0.9903, + "step": 3820 + }, + { + "epoch": 0.07285662652894291, + "grad_norm": 4.373607635498047, + "learning_rate": 7.28552406315389e-06, + "loss": 0.8325, + "step": 3830 + }, + { + "epoch": 0.07304685270786966, + "grad_norm": 5.0198140144348145, + "learning_rate": 7.304546319193457e-06, + "loss": 0.9334, + "step": 3840 + }, + { + "epoch": 0.0732370788867964, + "grad_norm": 13.514430046081543, + "learning_rate": 7.3235685752330235e-06, + "loss": 1.1131, + "step": 3850 + }, + { + "epoch": 0.07342730506572315, + "grad_norm": 2.1664793491363525, + "learning_rate": 7.342590831272589e-06, + "loss": 1.0527, + "step": 3860 + }, + { + "epoch": 0.07361753124464988, + "grad_norm": 6.562690258026123, + "learning_rate": 7.361613087312156e-06, + "loss": 0.9892, + "step": 3870 + }, + { + "epoch": 0.07380775742357663, + "grad_norm": 3.725823163986206, + "learning_rate": 7.380635343351722e-06, + "loss": 0.8403, + "step": 3880 + }, + { + "epoch": 0.07399798360250337, + "grad_norm": 5.953464031219482, + "learning_rate": 7.399657599391289e-06, + "loss": 0.8569, + "step": 3890 + }, + { + "epoch": 0.07418820978143012, + "grad_norm": 10.803670883178711, + "learning_rate": 7.418679855430854e-06, + "loss": 1.0628, + "step": 3900 + }, + { + "epoch": 0.07437843596035687, + "grad_norm": 2.3450770378112793, + "learning_rate": 7.437702111470421e-06, + "loss": 1.0782, + "step": 3910 + }, + { + "epoch": 0.0745686621392836, + "grad_norm": 2.5900728702545166, + "learning_rate": 7.456724367509988e-06, + "loss": 0.9397, + "step": 3920 + }, + { + "epoch": 0.07475888831821036, + "grad_norm": 3.811370849609375, + "learning_rate": 7.475746623549554e-06, + "loss": 0.8472, + "step": 3930 + }, + { + "epoch": 0.07494911449713709, + "grad_norm": 8.507999420166016, + "learning_rate": 7.494768879589119e-06, + "loss": 0.8761, + "step": 3940 + }, + { + "epoch": 0.07513934067606384, + "grad_norm": 11.520349502563477, + "learning_rate": 7.513791135628686e-06, + "loss": 1.0814, + "step": 3950 + }, + { + "epoch": 0.07532956685499058, + "grad_norm": 2.7183773517608643, + "learning_rate": 7.532813391668252e-06, + "loss": 1.0746, + "step": 3960 + }, + { + "epoch": 0.07551979303391733, + "grad_norm": 2.7066807746887207, + "learning_rate": 7.551835647707819e-06, + "loss": 0.9913, + "step": 3970 + }, + { + "epoch": 0.07571001921284407, + "grad_norm": 2.972797393798828, + "learning_rate": 7.570857903747385e-06, + "loss": 0.8899, + "step": 3980 + }, + { + "epoch": 0.07590024539177082, + "grad_norm": 5.18887996673584, + "learning_rate": 7.589880159786951e-06, + "loss": 0.8659, + "step": 3990 + }, + { + "epoch": 0.07609047157069757, + "grad_norm": 13.543824195861816, + "learning_rate": 7.608902415826518e-06, + "loss": 1.003, + "step": 4000 + }, + { + "epoch": 0.0762806977496243, + "grad_norm": 2.2497000694274902, + "learning_rate": 7.627924671866084e-06, + "loss": 1.0616, + "step": 4010 + }, + { + "epoch": 0.07647092392855105, + "grad_norm": 2.5572075843811035, + "learning_rate": 7.646946927905651e-06, + "loss": 0.9807, + "step": 4020 + }, + { + "epoch": 0.07666115010747779, + "grad_norm": 4.120774745941162, + "learning_rate": 7.665969183945216e-06, + "loss": 0.8635, + "step": 4030 + }, + { + "epoch": 0.07685137628640454, + "grad_norm": 5.530590534210205, + "learning_rate": 7.684991439984782e-06, + "loss": 0.8859, + "step": 4040 + }, + { + "epoch": 0.07704160246533127, + "grad_norm": 12.142224311828613, + "learning_rate": 7.704013696024349e-06, + "loss": 1.0475, + "step": 4050 + }, + { + "epoch": 0.07723182864425802, + "grad_norm": 3.392179250717163, + "learning_rate": 7.723035952063916e-06, + "loss": 1.0639, + "step": 4060 + }, + { + "epoch": 0.07742205482318476, + "grad_norm": 2.641266345977783, + "learning_rate": 7.742058208103481e-06, + "loss": 0.9801, + "step": 4070 + }, + { + "epoch": 0.07761228100211151, + "grad_norm": 4.2979350090026855, + "learning_rate": 7.761080464143048e-06, + "loss": 0.8867, + "step": 4080 + }, + { + "epoch": 0.07780250718103826, + "grad_norm": 5.953768730163574, + "learning_rate": 7.780102720182613e-06, + "loss": 0.8286, + "step": 4090 + }, + { + "epoch": 0.077992733359965, + "grad_norm": 9.22087574005127, + "learning_rate": 7.79912497622218e-06, + "loss": 1.0132, + "step": 4100 + }, + { + "epoch": 0.07818295953889175, + "grad_norm": 2.587709665298462, + "learning_rate": 7.818147232261748e-06, + "loss": 1.064, + "step": 4110 + }, + { + "epoch": 0.07837318571781848, + "grad_norm": 2.669003486633301, + "learning_rate": 7.837169488301313e-06, + "loss": 0.9844, + "step": 4120 + }, + { + "epoch": 0.07856341189674523, + "grad_norm": 3.0199615955352783, + "learning_rate": 7.85619174434088e-06, + "loss": 0.8422, + "step": 4130 + }, + { + "epoch": 0.07875363807567197, + "grad_norm": 4.246236324310303, + "learning_rate": 7.875214000380445e-06, + "loss": 0.8418, + "step": 4140 + }, + { + "epoch": 0.07894386425459872, + "grad_norm": 10.317477226257324, + "learning_rate": 7.894236256420012e-06, + "loss": 1.0039, + "step": 4150 + }, + { + "epoch": 0.07913409043352546, + "grad_norm": 2.923060655593872, + "learning_rate": 7.913258512459578e-06, + "loss": 1.0689, + "step": 4160 + }, + { + "epoch": 0.0793243166124522, + "grad_norm": 1.7650673389434814, + "learning_rate": 7.932280768499145e-06, + "loss": 0.9898, + "step": 4170 + }, + { + "epoch": 0.07951454279137896, + "grad_norm": 4.0497822761535645, + "learning_rate": 7.951303024538712e-06, + "loss": 0.8677, + "step": 4180 + }, + { + "epoch": 0.07970476897030569, + "grad_norm": 4.521123886108398, + "learning_rate": 7.970325280578277e-06, + "loss": 0.8269, + "step": 4190 + }, + { + "epoch": 0.07989499514923244, + "grad_norm": 12.454859733581543, + "learning_rate": 7.989347536617842e-06, + "loss": 0.9994, + "step": 4200 + }, + { + "epoch": 0.08008522132815918, + "grad_norm": 2.702120542526245, + "learning_rate": 8.00836979265741e-06, + "loss": 1.0442, + "step": 4210 + }, + { + "epoch": 0.08027544750708593, + "grad_norm": 2.9055328369140625, + "learning_rate": 8.027392048696976e-06, + "loss": 1.0382, + "step": 4220 + }, + { + "epoch": 0.08046567368601266, + "grad_norm": 3.210998773574829, + "learning_rate": 8.046414304736543e-06, + "loss": 0.8021, + "step": 4230 + }, + { + "epoch": 0.08065589986493941, + "grad_norm": 4.988993167877197, + "learning_rate": 8.065436560776109e-06, + "loss": 0.8846, + "step": 4240 + }, + { + "epoch": 0.08084612604386615, + "grad_norm": 13.53798770904541, + "learning_rate": 8.084458816815674e-06, + "loss": 1.0493, + "step": 4250 + }, + { + "epoch": 0.0810363522227929, + "grad_norm": 4.448582172393799, + "learning_rate": 8.103481072855241e-06, + "loss": 1.0174, + "step": 4260 + }, + { + "epoch": 0.08122657840171965, + "grad_norm": 3.8949131965637207, + "learning_rate": 8.122503328894808e-06, + "loss": 1.024, + "step": 4270 + }, + { + "epoch": 0.08141680458064639, + "grad_norm": 3.751685619354248, + "learning_rate": 8.141525584934375e-06, + "loss": 0.797, + "step": 4280 + }, + { + "epoch": 0.08160703075957314, + "grad_norm": 4.3078436851501465, + "learning_rate": 8.16054784097394e-06, + "loss": 0.8257, + "step": 4290 + }, + { + "epoch": 0.08179725693849987, + "grad_norm": 10.666089057922363, + "learning_rate": 8.179570097013506e-06, + "loss": 1.0389, + "step": 4300 + }, + { + "epoch": 0.08198748311742662, + "grad_norm": 2.153872489929199, + "learning_rate": 8.198592353053073e-06, + "loss": 1.0465, + "step": 4310 + }, + { + "epoch": 0.08217770929635336, + "grad_norm": 2.911468744277954, + "learning_rate": 8.217614609092638e-06, + "loss": 0.9862, + "step": 4320 + }, + { + "epoch": 0.08236793547528011, + "grad_norm": 3.601792573928833, + "learning_rate": 8.236636865132205e-06, + "loss": 0.8786, + "step": 4330 + }, + { + "epoch": 0.08255816165420685, + "grad_norm": 7.536445140838623, + "learning_rate": 8.255659121171772e-06, + "loss": 0.8042, + "step": 4340 + }, + { + "epoch": 0.0827483878331336, + "grad_norm": 12.88973617553711, + "learning_rate": 8.274681377211338e-06, + "loss": 1.0235, + "step": 4350 + }, + { + "epoch": 0.08293861401206035, + "grad_norm": 3.1542675495147705, + "learning_rate": 8.293703633250905e-06, + "loss": 1.0693, + "step": 4360 + }, + { + "epoch": 0.08312884019098708, + "grad_norm": 3.760368824005127, + "learning_rate": 8.31272588929047e-06, + "loss": 1.0306, + "step": 4370 + }, + { + "epoch": 0.08331906636991383, + "grad_norm": 3.9413628578186035, + "learning_rate": 8.331748145330037e-06, + "loss": 0.8654, + "step": 4380 + }, + { + "epoch": 0.08350929254884057, + "grad_norm": 4.6039862632751465, + "learning_rate": 8.350770401369604e-06, + "loss": 0.8311, + "step": 4390 + }, + { + "epoch": 0.08369951872776732, + "grad_norm": 11.079093933105469, + "learning_rate": 8.36979265740917e-06, + "loss": 0.9909, + "step": 4400 + }, + { + "epoch": 0.08388974490669406, + "grad_norm": 2.3281502723693848, + "learning_rate": 8.388814913448736e-06, + "loss": 1.0569, + "step": 4410 + }, + { + "epoch": 0.0840799710856208, + "grad_norm": 3.260143518447876, + "learning_rate": 8.407837169488302e-06, + "loss": 1.0086, + "step": 4420 + }, + { + "epoch": 0.08427019726454754, + "grad_norm": 2.812483310699463, + "learning_rate": 8.426859425527869e-06, + "loss": 0.8608, + "step": 4430 + }, + { + "epoch": 0.08446042344347429, + "grad_norm": 4.584824085235596, + "learning_rate": 8.445881681567436e-06, + "loss": 0.8322, + "step": 4440 + }, + { + "epoch": 0.08465064962240103, + "grad_norm": 9.15593147277832, + "learning_rate": 8.464903937607001e-06, + "loss": 1.0237, + "step": 4450 + }, + { + "epoch": 0.08484087580132778, + "grad_norm": 2.053774118423462, + "learning_rate": 8.483926193646566e-06, + "loss": 1.0547, + "step": 4460 + }, + { + "epoch": 0.08503110198025453, + "grad_norm": 2.387960910797119, + "learning_rate": 8.502948449686133e-06, + "loss": 0.9456, + "step": 4470 + }, + { + "epoch": 0.08522132815918126, + "grad_norm": 3.73870849609375, + "learning_rate": 8.521970705725699e-06, + "loss": 0.8484, + "step": 4480 + }, + { + "epoch": 0.08541155433810801, + "grad_norm": 5.459985256195068, + "learning_rate": 8.540992961765266e-06, + "loss": 0.8449, + "step": 4490 + }, + { + "epoch": 0.08560178051703475, + "grad_norm": 14.273500442504883, + "learning_rate": 8.560015217804833e-06, + "loss": 1.0177, + "step": 4500 + }, + { + "epoch": 0.0857920066959615, + "grad_norm": 3.626765012741089, + "learning_rate": 8.579037473844398e-06, + "loss": 1.0674, + "step": 4510 + }, + { + "epoch": 0.08598223287488824, + "grad_norm": 5.053386688232422, + "learning_rate": 8.598059729883965e-06, + "loss": 1.0287, + "step": 4520 + }, + { + "epoch": 0.08617245905381499, + "grad_norm": 3.2023839950561523, + "learning_rate": 8.61708198592353e-06, + "loss": 0.7877, + "step": 4530 + }, + { + "epoch": 0.08636268523274172, + "grad_norm": 9.036783218383789, + "learning_rate": 8.636104241963097e-06, + "loss": 0.8622, + "step": 4540 + }, + { + "epoch": 0.08655291141166847, + "grad_norm": 12.49517822265625, + "learning_rate": 8.655126498002664e-06, + "loss": 1.0252, + "step": 4550 + }, + { + "epoch": 0.08674313759059522, + "grad_norm": 2.4512569904327393, + "learning_rate": 8.67414875404223e-06, + "loss": 1.0266, + "step": 4560 + }, + { + "epoch": 0.08693336376952196, + "grad_norm": 2.0704665184020996, + "learning_rate": 8.693171010081797e-06, + "loss": 0.9304, + "step": 4570 + }, + { + "epoch": 0.08712358994844871, + "grad_norm": 2.670184850692749, + "learning_rate": 8.712193266121362e-06, + "loss": 0.8504, + "step": 4580 + }, + { + "epoch": 0.08731381612737545, + "grad_norm": 7.81472635269165, + "learning_rate": 8.73121552216093e-06, + "loss": 0.8383, + "step": 4590 + }, + { + "epoch": 0.0875040423063022, + "grad_norm": 11.205781936645508, + "learning_rate": 8.750237778200496e-06, + "loss": 1.0539, + "step": 4600 + }, + { + "epoch": 0.08769426848522893, + "grad_norm": 2.879502773284912, + "learning_rate": 8.769260034240062e-06, + "loss": 1.0363, + "step": 4610 + }, + { + "epoch": 0.08788449466415568, + "grad_norm": 4.640857696533203, + "learning_rate": 8.788282290279629e-06, + "loss": 1.0272, + "step": 4620 + }, + { + "epoch": 0.08807472084308242, + "grad_norm": 3.4126992225646973, + "learning_rate": 8.807304546319194e-06, + "loss": 0.8409, + "step": 4630 + }, + { + "epoch": 0.08826494702200917, + "grad_norm": 3.9957668781280518, + "learning_rate": 8.82632680235876e-06, + "loss": 0.876, + "step": 4640 + }, + { + "epoch": 0.08845517320093592, + "grad_norm": 11.795957565307617, + "learning_rate": 8.845349058398326e-06, + "loss": 1.0023, + "step": 4650 + }, + { + "epoch": 0.08864539937986265, + "grad_norm": 2.539607524871826, + "learning_rate": 8.864371314437893e-06, + "loss": 1.0589, + "step": 4660 + }, + { + "epoch": 0.0888356255587894, + "grad_norm": 2.778062343597412, + "learning_rate": 8.88339357047746e-06, + "loss": 0.9831, + "step": 4670 + }, + { + "epoch": 0.08902585173771614, + "grad_norm": 2.902723789215088, + "learning_rate": 8.902415826517026e-06, + "loss": 0.8115, + "step": 4680 + }, + { + "epoch": 0.08921607791664289, + "grad_norm": 6.224371910095215, + "learning_rate": 8.921438082556591e-06, + "loss": 0.8285, + "step": 4690 + }, + { + "epoch": 0.08940630409556963, + "grad_norm": 13.158405303955078, + "learning_rate": 8.940460338596158e-06, + "loss": 1.056, + "step": 4700 + }, + { + "epoch": 0.08959653027449638, + "grad_norm": 2.7186522483825684, + "learning_rate": 8.959482594635725e-06, + "loss": 1.0455, + "step": 4710 + }, + { + "epoch": 0.08978675645342311, + "grad_norm": 2.863405227661133, + "learning_rate": 8.97850485067529e-06, + "loss": 1.0269, + "step": 4720 + }, + { + "epoch": 0.08997698263234986, + "grad_norm": 2.861708641052246, + "learning_rate": 8.997527106714857e-06, + "loss": 0.7955, + "step": 4730 + }, + { + "epoch": 0.09016720881127661, + "grad_norm": 4.701228618621826, + "learning_rate": 9.016549362754423e-06, + "loss": 0.8354, + "step": 4740 + }, + { + "epoch": 0.09035743499020335, + "grad_norm": 12.41553020477295, + "learning_rate": 9.03557161879399e-06, + "loss": 1.131, + "step": 4750 + }, + { + "epoch": 0.0905476611691301, + "grad_norm": 2.774181842803955, + "learning_rate": 9.054593874833557e-06, + "loss": 1.045, + "step": 4760 + }, + { + "epoch": 0.09073788734805684, + "grad_norm": 2.170881748199463, + "learning_rate": 9.073616130873122e-06, + "loss": 0.974, + "step": 4770 + }, + { + "epoch": 0.09092811352698359, + "grad_norm": 3.6522154808044434, + "learning_rate": 9.092638386912689e-06, + "loss": 0.8233, + "step": 4780 + }, + { + "epoch": 0.09111833970591032, + "grad_norm": 4.515476226806641, + "learning_rate": 9.111660642952254e-06, + "loss": 0.8054, + "step": 4790 + }, + { + "epoch": 0.09130856588483707, + "grad_norm": 10.058849334716797, + "learning_rate": 9.130682898991821e-06, + "loss": 1.0239, + "step": 4800 + }, + { + "epoch": 0.09149879206376381, + "grad_norm": 4.930131912231445, + "learning_rate": 9.149705155031387e-06, + "loss": 1.0748, + "step": 4810 + }, + { + "epoch": 0.09168901824269056, + "grad_norm": 3.753122091293335, + "learning_rate": 9.168727411070954e-06, + "loss": 0.9262, + "step": 4820 + }, + { + "epoch": 0.09187924442161731, + "grad_norm": 3.604468822479248, + "learning_rate": 9.187749667110521e-06, + "loss": 0.8119, + "step": 4830 + }, + { + "epoch": 0.09206947060054405, + "grad_norm": 4.331033706665039, + "learning_rate": 9.206771923150086e-06, + "loss": 0.8029, + "step": 4840 + }, + { + "epoch": 0.0922596967794708, + "grad_norm": 11.139960289001465, + "learning_rate": 9.225794179189652e-06, + "loss": 0.9785, + "step": 4850 + }, + { + "epoch": 0.09244992295839753, + "grad_norm": 3.0269057750701904, + "learning_rate": 9.244816435229219e-06, + "loss": 1.0565, + "step": 4860 + }, + { + "epoch": 0.09264014913732428, + "grad_norm": 2.4621336460113525, + "learning_rate": 9.263838691268786e-06, + "loss": 0.9695, + "step": 4870 + }, + { + "epoch": 0.09283037531625102, + "grad_norm": 3.160784959793091, + "learning_rate": 9.282860947308353e-06, + "loss": 0.7952, + "step": 4880 + }, + { + "epoch": 0.09302060149517777, + "grad_norm": 6.189260005950928, + "learning_rate": 9.301883203347918e-06, + "loss": 0.8586, + "step": 4890 + }, + { + "epoch": 0.0932108276741045, + "grad_norm": 13.33193588256836, + "learning_rate": 9.320905459387483e-06, + "loss": 1.0753, + "step": 4900 + }, + { + "epoch": 0.09340105385303125, + "grad_norm": 2.4601781368255615, + "learning_rate": 9.33992771542705e-06, + "loss": 1.0345, + "step": 4910 + }, + { + "epoch": 0.093591280031958, + "grad_norm": 3.0945911407470703, + "learning_rate": 9.358949971466617e-06, + "loss": 0.9604, + "step": 4920 + }, + { + "epoch": 0.09378150621088474, + "grad_norm": 3.2006800174713135, + "learning_rate": 9.377972227506184e-06, + "loss": 0.8122, + "step": 4930 + }, + { + "epoch": 0.09397173238981149, + "grad_norm": 6.865549564361572, + "learning_rate": 9.39699448354575e-06, + "loss": 0.8086, + "step": 4940 + }, + { + "epoch": 0.09416195856873823, + "grad_norm": 11.536060333251953, + "learning_rate": 9.416016739585315e-06, + "loss": 0.9879, + "step": 4950 + }, + { + "epoch": 0.09435218474766498, + "grad_norm": 4.649012088775635, + "learning_rate": 9.435038995624882e-06, + "loss": 1.0615, + "step": 4960 + }, + { + "epoch": 0.09454241092659171, + "grad_norm": 2.240013599395752, + "learning_rate": 9.454061251664447e-06, + "loss": 0.9851, + "step": 4970 + }, + { + "epoch": 0.09473263710551846, + "grad_norm": 3.457646131515503, + "learning_rate": 9.473083507704014e-06, + "loss": 0.7976, + "step": 4980 + }, + { + "epoch": 0.0949228632844452, + "grad_norm": 4.621734142303467, + "learning_rate": 9.492105763743581e-06, + "loss": 0.7787, + "step": 4990 + }, + { + "epoch": 0.09511308946337195, + "grad_norm": 12.112364768981934, + "learning_rate": 9.511128019783147e-06, + "loss": 1.0011, + "step": 5000 + }, + { + "epoch": 0.0953033156422987, + "grad_norm": 2.524951696395874, + "learning_rate": 9.530150275822714e-06, + "loss": 1.0024, + "step": 5010 + }, + { + "epoch": 0.09549354182122544, + "grad_norm": 2.1611406803131104, + "learning_rate": 9.549172531862279e-06, + "loss": 0.9895, + "step": 5020 + }, + { + "epoch": 0.09568376800015219, + "grad_norm": 3.647883415222168, + "learning_rate": 9.568194787901846e-06, + "loss": 0.8134, + "step": 5030 + }, + { + "epoch": 0.09587399417907892, + "grad_norm": 5.40878963470459, + "learning_rate": 9.587217043941413e-06, + "loss": 0.7944, + "step": 5040 + }, + { + "epoch": 0.09606422035800567, + "grad_norm": 11.021536827087402, + "learning_rate": 9.606239299980978e-06, + "loss": 1.0177, + "step": 5050 + }, + { + "epoch": 0.09625444653693241, + "grad_norm": 2.3612380027770996, + "learning_rate": 9.625261556020545e-06, + "loss": 1.0267, + "step": 5060 + }, + { + "epoch": 0.09644467271585916, + "grad_norm": 2.3841519355773926, + "learning_rate": 9.64428381206011e-06, + "loss": 0.9979, + "step": 5070 + }, + { + "epoch": 0.0966348988947859, + "grad_norm": 2.819408416748047, + "learning_rate": 9.663306068099678e-06, + "loss": 0.7921, + "step": 5080 + }, + { + "epoch": 0.09682512507371265, + "grad_norm": 4.353105068206787, + "learning_rate": 9.682328324139245e-06, + "loss": 0.8087, + "step": 5090 + }, + { + "epoch": 0.09701535125263938, + "grad_norm": 9.611862182617188, + "learning_rate": 9.70135058017881e-06, + "loss": 0.9903, + "step": 5100 + }, + { + "epoch": 0.09720557743156613, + "grad_norm": 3.051971197128296, + "learning_rate": 9.720372836218376e-06, + "loss": 1.0251, + "step": 5110 + }, + { + "epoch": 0.09739580361049288, + "grad_norm": 3.2879152297973633, + "learning_rate": 9.739395092257943e-06, + "loss": 0.9787, + "step": 5120 + }, + { + "epoch": 0.09758602978941962, + "grad_norm": 3.3315176963806152, + "learning_rate": 9.758417348297508e-06, + "loss": 0.8526, + "step": 5130 + }, + { + "epoch": 0.09777625596834637, + "grad_norm": 4.358604907989502, + "learning_rate": 9.777439604337075e-06, + "loss": 0.841, + "step": 5140 + }, + { + "epoch": 0.0979664821472731, + "grad_norm": 10.932659149169922, + "learning_rate": 9.796461860376642e-06, + "loss": 1.0112, + "step": 5150 + }, + { + "epoch": 0.09815670832619985, + "grad_norm": 3.0365395545959473, + "learning_rate": 9.815484116416207e-06, + "loss": 1.0484, + "step": 5160 + }, + { + "epoch": 0.09834693450512659, + "grad_norm": 2.4415764808654785, + "learning_rate": 9.834506372455774e-06, + "loss": 0.9546, + "step": 5170 + }, + { + "epoch": 0.09853716068405334, + "grad_norm": 3.0862984657287598, + "learning_rate": 9.85352862849534e-06, + "loss": 0.8239, + "step": 5180 + }, + { + "epoch": 0.09872738686298008, + "grad_norm": 4.357607841491699, + "learning_rate": 9.872550884534907e-06, + "loss": 0.8166, + "step": 5190 + }, + { + "epoch": 0.09891761304190683, + "grad_norm": 10.748388290405273, + "learning_rate": 9.891573140574474e-06, + "loss": 1.0355, + "step": 5200 + }, + { + "epoch": 0.09910783922083358, + "grad_norm": 2.779499053955078, + "learning_rate": 9.910595396614039e-06, + "loss": 1.0779, + "step": 5210 + }, + { + "epoch": 0.09929806539976031, + "grad_norm": 3.65140700340271, + "learning_rate": 9.929617652653606e-06, + "loss": 0.9539, + "step": 5220 + }, + { + "epoch": 0.09948829157868706, + "grad_norm": 3.2343590259552, + "learning_rate": 9.948639908693171e-06, + "loss": 0.7727, + "step": 5230 + }, + { + "epoch": 0.0996785177576138, + "grad_norm": 5.205893516540527, + "learning_rate": 9.967662164732738e-06, + "loss": 0.8448, + "step": 5240 + }, + { + "epoch": 0.09986874393654055, + "grad_norm": 11.548497200012207, + "learning_rate": 9.986684420772305e-06, + "loss": 1.0132, + "step": 5250 + }, + { + "epoch": 0.10005897011546729, + "grad_norm": 2.170140504837036, + "learning_rate": 1e-05, + "loss": 0.9852, + "step": 5260 + }, + { + "epoch": 0.10024919629439404, + "grad_norm": 2.476912260055542, + "learning_rate": 1e-05, + "loss": 0.9804, + "step": 5270 + }, + { + "epoch": 0.10043942247332077, + "grad_norm": 3.0934178829193115, + "learning_rate": 1e-05, + "loss": 0.8043, + "step": 5280 + }, + { + "epoch": 0.10062964865224752, + "grad_norm": 4.3917999267578125, + "learning_rate": 1e-05, + "loss": 0.8387, + "step": 5290 + }, + { + "epoch": 0.10081987483117427, + "grad_norm": 14.04360580444336, + "learning_rate": 1e-05, + "loss": 1.0638, + "step": 5300 + }, + { + "epoch": 0.10101010101010101, + "grad_norm": 1.9007540941238403, + "learning_rate": 1e-05, + "loss": 1.0097, + "step": 5310 + }, + { + "epoch": 0.10120032718902776, + "grad_norm": 2.8841350078582764, + "learning_rate": 1e-05, + "loss": 0.9581, + "step": 5320 + }, + { + "epoch": 0.1013905533679545, + "grad_norm": 4.027973175048828, + "learning_rate": 1e-05, + "loss": 0.7922, + "step": 5330 + }, + { + "epoch": 0.10158077954688124, + "grad_norm": 4.386593818664551, + "learning_rate": 1e-05, + "loss": 0.8356, + "step": 5340 + }, + { + "epoch": 0.10177100572580798, + "grad_norm": 10.778451919555664, + "learning_rate": 1e-05, + "loss": 0.987, + "step": 5350 + }, + { + "epoch": 0.10196123190473473, + "grad_norm": 2.4819626808166504, + "learning_rate": 1e-05, + "loss": 1.0063, + "step": 5360 + }, + { + "epoch": 0.10215145808366147, + "grad_norm": 2.8411858081817627, + "learning_rate": 1e-05, + "loss": 0.9477, + "step": 5370 + }, + { + "epoch": 0.10234168426258822, + "grad_norm": 3.0888965129852295, + "learning_rate": 1e-05, + "loss": 0.7993, + "step": 5380 + }, + { + "epoch": 0.10253191044151497, + "grad_norm": 4.141193866729736, + "learning_rate": 1e-05, + "loss": 0.8134, + "step": 5390 + }, + { + "epoch": 0.1027221366204417, + "grad_norm": 9.52237319946289, + "learning_rate": 1e-05, + "loss": 0.9566, + "step": 5400 + }, + { + "epoch": 0.10291236279936845, + "grad_norm": 3.151825428009033, + "learning_rate": 1e-05, + "loss": 1.0167, + "step": 5410 + }, + { + "epoch": 0.10310258897829519, + "grad_norm": 2.040436267852783, + "learning_rate": 1e-05, + "loss": 0.9696, + "step": 5420 + }, + { + "epoch": 0.10329281515722194, + "grad_norm": 3.3553965091705322, + "learning_rate": 1e-05, + "loss": 0.8425, + "step": 5430 + }, + { + "epoch": 0.10348304133614868, + "grad_norm": 5.130828857421875, + "learning_rate": 1e-05, + "loss": 0.8082, + "step": 5440 + }, + { + "epoch": 0.10367326751507543, + "grad_norm": 9.019664764404297, + "learning_rate": 1e-05, + "loss": 0.9742, + "step": 5450 + }, + { + "epoch": 0.10386349369400216, + "grad_norm": 2.3726866245269775, + "learning_rate": 1e-05, + "loss": 0.9999, + "step": 5460 + }, + { + "epoch": 0.10405371987292891, + "grad_norm": 2.137408494949341, + "learning_rate": 1e-05, + "loss": 0.9862, + "step": 5470 + }, + { + "epoch": 0.10424394605185566, + "grad_norm": 3.0669353008270264, + "learning_rate": 1e-05, + "loss": 0.7806, + "step": 5480 + }, + { + "epoch": 0.1044341722307824, + "grad_norm": 3.7706305980682373, + "learning_rate": 1e-05, + "loss": 0.8121, + "step": 5490 + }, + { + "epoch": 0.10462439840970915, + "grad_norm": 10.3373441696167, + "learning_rate": 1e-05, + "loss": 0.9813, + "step": 5500 + }, + { + "epoch": 0.10481462458863589, + "grad_norm": 2.218104124069214, + "learning_rate": 1e-05, + "loss": 1.0445, + "step": 5510 + }, + { + "epoch": 0.10500485076756264, + "grad_norm": 2.3944590091705322, + "learning_rate": 1e-05, + "loss": 0.9092, + "step": 5520 + }, + { + "epoch": 0.10519507694648937, + "grad_norm": 3.1064295768737793, + "learning_rate": 1e-05, + "loss": 0.7949, + "step": 5530 + }, + { + "epoch": 0.10538530312541612, + "grad_norm": 5.8249030113220215, + "learning_rate": 1e-05, + "loss": 0.8293, + "step": 5540 + }, + { + "epoch": 0.10557552930434286, + "grad_norm": 9.393155097961426, + "learning_rate": 1e-05, + "loss": 0.9622, + "step": 5550 + }, + { + "epoch": 0.10576575548326961, + "grad_norm": 4.191644668579102, + "learning_rate": 1e-05, + "loss": 1.0034, + "step": 5560 + }, + { + "epoch": 0.10595598166219636, + "grad_norm": 2.126408576965332, + "learning_rate": 1e-05, + "loss": 0.9413, + "step": 5570 + }, + { + "epoch": 0.1061462078411231, + "grad_norm": 2.6248719692230225, + "learning_rate": 1e-05, + "loss": 0.7711, + "step": 5580 + }, + { + "epoch": 0.10633643402004984, + "grad_norm": 4.614845275878906, + "learning_rate": 1e-05, + "loss": 0.8127, + "step": 5590 + }, + { + "epoch": 0.10652666019897658, + "grad_norm": 10.200318336486816, + "learning_rate": 1e-05, + "loss": 0.9967, + "step": 5600 + }, + { + "epoch": 0.10671688637790333, + "grad_norm": 2.541682481765747, + "learning_rate": 1e-05, + "loss": 1.0298, + "step": 5610 + }, + { + "epoch": 0.10690711255683007, + "grad_norm": 2.1859960556030273, + "learning_rate": 1e-05, + "loss": 0.9212, + "step": 5620 + }, + { + "epoch": 0.10709733873575682, + "grad_norm": 2.971053123474121, + "learning_rate": 1e-05, + "loss": 0.7827, + "step": 5630 + }, + { + "epoch": 0.10728756491468355, + "grad_norm": 5.017271518707275, + "learning_rate": 1e-05, + "loss": 0.7759, + "step": 5640 + }, + { + "epoch": 0.1074777910936103, + "grad_norm": 9.602482795715332, + "learning_rate": 1e-05, + "loss": 0.9141, + "step": 5650 + }, + { + "epoch": 0.10766801727253705, + "grad_norm": 2.660590887069702, + "learning_rate": 1e-05, + "loss": 1.0346, + "step": 5660 + }, + { + "epoch": 0.10785824345146379, + "grad_norm": 2.3705356121063232, + "learning_rate": 1e-05, + "loss": 0.9302, + "step": 5670 + }, + { + "epoch": 0.10804846963039054, + "grad_norm": 2.7425448894500732, + "learning_rate": 1e-05, + "loss": 0.7813, + "step": 5680 + }, + { + "epoch": 0.10823869580931728, + "grad_norm": 4.8073601722717285, + "learning_rate": 1e-05, + "loss": 0.7984, + "step": 5690 + }, + { + "epoch": 0.10842892198824403, + "grad_norm": 11.548068046569824, + "learning_rate": 1e-05, + "loss": 0.9622, + "step": 5700 + }, + { + "epoch": 0.10861914816717076, + "grad_norm": 2.775019407272339, + "learning_rate": 1e-05, + "loss": 0.984, + "step": 5710 + }, + { + "epoch": 0.10880937434609751, + "grad_norm": 3.1400108337402344, + "learning_rate": 1e-05, + "loss": 0.9565, + "step": 5720 + }, + { + "epoch": 0.10899960052502425, + "grad_norm": 2.912726879119873, + "learning_rate": 1e-05, + "loss": 0.7693, + "step": 5730 + }, + { + "epoch": 0.109189826703951, + "grad_norm": 4.758162021636963, + "learning_rate": 1e-05, + "loss": 0.8397, + "step": 5740 + }, + { + "epoch": 0.10938005288287775, + "grad_norm": 11.496520042419434, + "learning_rate": 1e-05, + "loss": 0.9572, + "step": 5750 + }, + { + "epoch": 0.10957027906180448, + "grad_norm": 2.6489996910095215, + "learning_rate": 1e-05, + "loss": 1.0163, + "step": 5760 + }, + { + "epoch": 0.10976050524073123, + "grad_norm": 2.03450345993042, + "learning_rate": 1e-05, + "loss": 0.9425, + "step": 5770 + }, + { + "epoch": 0.10995073141965797, + "grad_norm": 2.9618072509765625, + "learning_rate": 1e-05, + "loss": 0.7925, + "step": 5780 + }, + { + "epoch": 0.11014095759858472, + "grad_norm": 3.8805148601531982, + "learning_rate": 1e-05, + "loss": 0.8022, + "step": 5790 + }, + { + "epoch": 0.11033118377751146, + "grad_norm": 10.308512687683105, + "learning_rate": 1e-05, + "loss": 0.9776, + "step": 5800 + }, + { + "epoch": 0.11052140995643821, + "grad_norm": 2.324984312057495, + "learning_rate": 1e-05, + "loss": 1.0327, + "step": 5810 + }, + { + "epoch": 0.11071163613536494, + "grad_norm": 2.211977005004883, + "learning_rate": 1e-05, + "loss": 0.9078, + "step": 5820 + }, + { + "epoch": 0.1109018623142917, + "grad_norm": 3.970721483230591, + "learning_rate": 1e-05, + "loss": 0.7686, + "step": 5830 + }, + { + "epoch": 0.11109208849321843, + "grad_norm": 4.543284893035889, + "learning_rate": 1e-05, + "loss": 0.7989, + "step": 5840 + }, + { + "epoch": 0.11128231467214518, + "grad_norm": 10.333953857421875, + "learning_rate": 1e-05, + "loss": 0.965, + "step": 5850 + }, + { + "epoch": 0.11147254085107193, + "grad_norm": 2.5103368759155273, + "learning_rate": 1e-05, + "loss": 1.0208, + "step": 5860 + }, + { + "epoch": 0.11166276702999867, + "grad_norm": 2.713374614715576, + "learning_rate": 1e-05, + "loss": 0.9197, + "step": 5870 + }, + { + "epoch": 0.11185299320892542, + "grad_norm": 2.925102710723877, + "learning_rate": 1e-05, + "loss": 0.763, + "step": 5880 + }, + { + "epoch": 0.11204321938785215, + "grad_norm": 4.616085052490234, + "learning_rate": 1e-05, + "loss": 0.7851, + "step": 5890 + }, + { + "epoch": 0.1122334455667789, + "grad_norm": 11.197551727294922, + "learning_rate": 1e-05, + "loss": 0.941, + "step": 5900 + }, + { + "epoch": 0.11242367174570564, + "grad_norm": 5.09884786605835, + "learning_rate": 1e-05, + "loss": 0.9611, + "step": 5910 + }, + { + "epoch": 0.11261389792463239, + "grad_norm": 2.7460286617279053, + "learning_rate": 1e-05, + "loss": 0.9698, + "step": 5920 + }, + { + "epoch": 0.11280412410355913, + "grad_norm": 3.23636794090271, + "learning_rate": 1e-05, + "loss": 0.7744, + "step": 5930 + }, + { + "epoch": 0.11299435028248588, + "grad_norm": 3.779413938522339, + "learning_rate": 1e-05, + "loss": 0.7861, + "step": 5940 + }, + { + "epoch": 0.11318457646141263, + "grad_norm": 12.377265930175781, + "learning_rate": 1e-05, + "loss": 0.9692, + "step": 5950 + }, + { + "epoch": 0.11337480264033936, + "grad_norm": 2.6297824382781982, + "learning_rate": 1e-05, + "loss": 0.9445, + "step": 5960 + }, + { + "epoch": 0.11356502881926611, + "grad_norm": 3.0455784797668457, + "learning_rate": 1e-05, + "loss": 0.9369, + "step": 5970 + }, + { + "epoch": 0.11375525499819285, + "grad_norm": 3.3172903060913086, + "learning_rate": 1e-05, + "loss": 0.7328, + "step": 5980 + }, + { + "epoch": 0.1139454811771196, + "grad_norm": 6.561709880828857, + "learning_rate": 1e-05, + "loss": 0.7805, + "step": 5990 + }, + { + "epoch": 0.11413570735604633, + "grad_norm": 8.712008476257324, + "learning_rate": 1e-05, + "loss": 0.9122, + "step": 6000 + } + ], + "logging_steps": 10, + "max_steps": 105138, + "num_input_tokens_seen": 0, + "num_train_epochs": 2, + "save_steps": 6000, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 8.096540477112515e+18, + "train_batch_size": 7, + "trial_name": null, + "trial_params": null +}