{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 5.04072, "eval_steps": 100.0, "global_step": 45000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.001, "grad_norm": 191.5121800601927, "learning_rate": 3.0000000000000004e-07, "loss": 20.9094, "step": 10 }, { "epoch": 0.002, "grad_norm": 166.89736589540473, "learning_rate": 6.333333333333334e-07, "loss": 20.4268, "step": 20 }, { "epoch": 0.003, "grad_norm": 137.82136442107483, "learning_rate": 9.666666666666668e-07, "loss": 18.7467, "step": 30 }, { "epoch": 0.004, "grad_norm": 63.49699978675954, "learning_rate": 1.3e-06, "loss": 16.2005, "step": 40 }, { "epoch": 0.005, "grad_norm": 43.006122977427026, "learning_rate": 1.6333333333333335e-06, "loss": 14.5155, "step": 50 }, { "epoch": 0.006, "grad_norm": 26.032050334864362, "learning_rate": 1.9666666666666668e-06, "loss": 13.2723, "step": 60 }, { "epoch": 0.007, "grad_norm": 18.233641308799314, "learning_rate": 2.3000000000000004e-06, "loss": 12.4314, "step": 70 }, { "epoch": 0.008, "grad_norm": 19.007782108673528, "learning_rate": 2.6333333333333332e-06, "loss": 12.0204, "step": 80 }, { "epoch": 0.009, "grad_norm": 14.861953193687505, "learning_rate": 2.9666666666666673e-06, "loss": 11.8464, "step": 90 }, { "epoch": 0.01, "grad_norm": 14.000733321156797, "learning_rate": 3.3000000000000006e-06, "loss": 11.7284, "step": 100 }, { "epoch": 0.011, "grad_norm": 15.262858168736027, "learning_rate": 3.633333333333334e-06, "loss": 11.6123, "step": 110 }, { "epoch": 0.012, "grad_norm": 16.925218591947747, "learning_rate": 3.966666666666667e-06, "loss": 11.4731, "step": 120 }, { "epoch": 0.013, "grad_norm": 20.46035073998815, "learning_rate": 4.3e-06, "loss": 11.249, "step": 130 }, { "epoch": 0.014, "grad_norm": 32.33844507809916, "learning_rate": 4.633333333333334e-06, "loss": 10.8626, "step": 140 }, { "epoch": 0.015, "grad_norm": 33.371956027773024, "learning_rate": 4.966666666666667e-06, "loss": 10.2549, "step": 150 }, { "epoch": 0.016, "grad_norm": 31.0786160176865, "learning_rate": 5.300000000000001e-06, "loss": 9.5112, "step": 160 }, { "epoch": 0.017, "grad_norm": 32.49956295615944, "learning_rate": 5.633333333333334e-06, "loss": 8.7382, "step": 170 }, { "epoch": 0.018, "grad_norm": 30.833996151003387, "learning_rate": 5.966666666666667e-06, "loss": 7.9581, "step": 180 }, { "epoch": 0.019, "grad_norm": 26.424923301418264, "learning_rate": 6.300000000000001e-06, "loss": 7.1813, "step": 190 }, { "epoch": 0.02, "grad_norm": 26.700061275281364, "learning_rate": 6.633333333333334e-06, "loss": 6.5096, "step": 200 }, { "epoch": 0.021, "grad_norm": 24.429714861277336, "learning_rate": 6.966666666666667e-06, "loss": 6.0824, "step": 210 }, { "epoch": 0.022, "grad_norm": 24.299573449882466, "learning_rate": 7.3e-06, "loss": 5.7781, "step": 220 }, { "epoch": 0.023, "grad_norm": 21.16470038801262, "learning_rate": 7.633333333333334e-06, "loss": 5.6361, "step": 230 }, { "epoch": 0.024, "grad_norm": 24.554107034310842, "learning_rate": 7.966666666666668e-06, "loss": 5.5408, "step": 240 }, { "epoch": 0.025, "grad_norm": 22.976985506621336, "learning_rate": 8.3e-06, "loss": 5.5048, "step": 250 }, { "epoch": 0.026, "grad_norm": 22.417798976409415, "learning_rate": 8.633333333333334e-06, "loss": 5.4858, "step": 260 }, { "epoch": 0.027, "grad_norm": 22.207587512124054, "learning_rate": 8.966666666666667e-06, "loss": 5.4191, "step": 270 }, { "epoch": 0.028, "grad_norm": 21.492675567727563, "learning_rate": 9.3e-06, "loss": 5.4313, "step": 280 }, { "epoch": 0.029, "grad_norm": 22.264802978199526, "learning_rate": 9.633333333333335e-06, "loss": 5.4145, "step": 290 }, { "epoch": 0.03, "grad_norm": 23.203034409431837, "learning_rate": 9.966666666666667e-06, "loss": 5.393, "step": 300 }, { "epoch": 0.031, "grad_norm": 21.117016788142152, "learning_rate": 9.999978758704688e-06, "loss": 5.4209, "step": 310 }, { "epoch": 0.032, "grad_norm": 22.431695764462315, "learning_rate": 9.99990533223654e-06, "loss": 5.3804, "step": 320 }, { "epoch": 0.033, "grad_norm": 22.87746413043485, "learning_rate": 9.999779459127373e-06, "loss": 5.4108, "step": 330 }, { "epoch": 0.034, "grad_norm": 21.393044087809855, "learning_rate": 9.999601140697532e-06, "loss": 5.3923, "step": 340 }, { "epoch": 0.035, "grad_norm": 22.08820561743038, "learning_rate": 9.999370378817499e-06, "loss": 5.3987, "step": 350 }, { "epoch": 0.036, "grad_norm": 22.340286091495667, "learning_rate": 9.999087175907852e-06, "loss": 5.3695, "step": 360 }, { "epoch": 0.037, "grad_norm": 22.99764648723962, "learning_rate": 9.998751534939258e-06, "loss": 5.3854, "step": 370 }, { "epoch": 0.038, "grad_norm": 20.892854670564656, "learning_rate": 9.998363459432428e-06, "loss": 5.3847, "step": 380 }, { "epoch": 0.039, "grad_norm": 22.074447678901205, "learning_rate": 9.997922953458095e-06, "loss": 5.3503, "step": 390 }, { "epoch": 0.04, "grad_norm": 25.693103177645956, "learning_rate": 9.997430021636957e-06, "loss": 5.3942, "step": 400 }, { "epoch": 0.041, "grad_norm": 23.870741078336724, "learning_rate": 9.996884669139635e-06, "loss": 5.3734, "step": 410 }, { "epoch": 0.042, "grad_norm": 22.455134619511508, "learning_rate": 9.996286901686621e-06, "loss": 5.3888, "step": 420 }, { "epoch": 0.043, "grad_norm": 22.682163788995346, "learning_rate": 9.995636725548207e-06, "loss": 5.362, "step": 430 }, { "epoch": 0.044, "grad_norm": 21.40275135110395, "learning_rate": 9.994934147544438e-06, "loss": 5.3522, "step": 440 }, { "epoch": 0.045, "grad_norm": 22.005191674826133, "learning_rate": 9.994179175045023e-06, "loss": 5.3486, "step": 450 }, { "epoch": 0.046, "grad_norm": 21.232139842257375, "learning_rate": 9.993371815969268e-06, "loss": 5.3837, "step": 460 }, { "epoch": 0.047, "grad_norm": 22.835047672122716, "learning_rate": 9.992512078785986e-06, "loss": 5.3582, "step": 470 }, { "epoch": 0.048, "grad_norm": 22.235351584497582, "learning_rate": 9.991599972513411e-06, "loss": 5.3506, "step": 480 }, { "epoch": 0.049, "grad_norm": 21.9837502204522, "learning_rate": 9.99063550671911e-06, "loss": 5.3183, "step": 490 }, { "epoch": 0.05, "grad_norm": 23.801021188451493, "learning_rate": 9.989618691519873e-06, "loss": 5.3503, "step": 500 }, { "epoch": 0.051, "grad_norm": 22.277736847848917, "learning_rate": 9.988549537581608e-06, "loss": 5.2892, "step": 510 }, { "epoch": 0.052, "grad_norm": 24.21904365726824, "learning_rate": 9.987428056119235e-06, "loss": 5.2847, "step": 520 }, { "epoch": 0.053, "grad_norm": 22.912532173312083, "learning_rate": 9.986254258896563e-06, "loss": 5.2505, "step": 530 }, { "epoch": 0.054, "grad_norm": 23.48174488866164, "learning_rate": 9.98502815822617e-06, "loss": 5.2641, "step": 540 }, { "epoch": 0.055, "grad_norm": 25.02061433819434, "learning_rate": 9.983749766969272e-06, "loss": 5.2376, "step": 550 }, { "epoch": 0.056, "grad_norm": 22.88902384245407, "learning_rate": 9.982419098535587e-06, "loss": 5.2408, "step": 560 }, { "epoch": 0.057, "grad_norm": 21.332962350481747, "learning_rate": 9.9810361668832e-06, "loss": 5.196, "step": 570 }, { "epoch": 0.058, "grad_norm": 22.341775552467663, "learning_rate": 9.979600986518408e-06, "loss": 5.1314, "step": 580 }, { "epoch": 0.059, "grad_norm": 22.052485197352027, "learning_rate": 9.978113572495575e-06, "loss": 5.1646, "step": 590 }, { "epoch": 0.06, "grad_norm": 21.325152547432815, "learning_rate": 9.976573940416967e-06, "loss": 5.1357, "step": 600 }, { "epoch": 0.061, "grad_norm": 22.947282303464075, "learning_rate": 9.9749821064326e-06, "loss": 5.1217, "step": 610 }, { "epoch": 0.062, "grad_norm": 21.902019018288282, "learning_rate": 9.973338087240057e-06, "loss": 5.134, "step": 620 }, { "epoch": 0.063, "grad_norm": 23.312254564262773, "learning_rate": 9.971641900084322e-06, "loss": 5.1222, "step": 630 }, { "epoch": 0.064, "grad_norm": 21.900908238963005, "learning_rate": 9.969893562757594e-06, "loss": 5.0231, "step": 640 }, { "epoch": 0.065, "grad_norm": 21.895466113157568, "learning_rate": 9.968093093599107e-06, "loss": 5.0238, "step": 650 }, { "epoch": 0.066, "grad_norm": 23.714144452230002, "learning_rate": 9.966240511494929e-06, "loss": 4.9657, "step": 660 }, { "epoch": 0.067, "grad_norm": 22.127527389166946, "learning_rate": 9.96433583587777e-06, "loss": 4.9612, "step": 670 }, { "epoch": 0.068, "grad_norm": 24.189739794174656, "learning_rate": 9.962379086726777e-06, "loss": 4.9, "step": 680 }, { "epoch": 0.069, "grad_norm": 21.208409630224036, "learning_rate": 9.96037028456732e-06, "loss": 4.8682, "step": 690 }, { "epoch": 0.07, "grad_norm": 24.131593207777115, "learning_rate": 9.958309450470784e-06, "loss": 4.8783, "step": 700 }, { "epoch": 0.071, "grad_norm": 27.095911251000587, "learning_rate": 9.956196606054346e-06, "loss": 4.8478, "step": 710 }, { "epoch": 0.072, "grad_norm": 25.67355495673078, "learning_rate": 9.954031773480741e-06, "loss": 4.8249, "step": 720 }, { "epoch": 0.073, "grad_norm": 24.163788830136188, "learning_rate": 9.951814975458041e-06, "loss": 4.8443, "step": 730 }, { "epoch": 0.074, "grad_norm": 23.78825959136402, "learning_rate": 9.949546235239406e-06, "loss": 4.8138, "step": 740 }, { "epoch": 0.075, "grad_norm": 23.14404478359282, "learning_rate": 9.947225576622848e-06, "loss": 4.7972, "step": 750 }, { "epoch": 0.076, "grad_norm": 22.535710363850576, "learning_rate": 9.944853023950978e-06, "loss": 4.7017, "step": 760 }, { "epoch": 0.077, "grad_norm": 22.661539693523853, "learning_rate": 9.942428602110751e-06, "loss": 4.6246, "step": 770 }, { "epoch": 0.078, "grad_norm": 23.339874017225192, "learning_rate": 9.939952336533205e-06, "loss": 4.6568, "step": 780 }, { "epoch": 0.079, "grad_norm": 22.62219086029143, "learning_rate": 9.937424253193196e-06, "loss": 4.717, "step": 790 }, { "epoch": 0.08, "grad_norm": 22.783638896150393, "learning_rate": 9.934844378609117e-06, "loss": 4.6578, "step": 800 }, { "epoch": 0.081, "grad_norm": 23.533500175522235, "learning_rate": 9.932212739842634e-06, "loss": 4.6231, "step": 810 }, { "epoch": 0.082, "grad_norm": 23.066256645098754, "learning_rate": 9.929529364498394e-06, "loss": 4.5839, "step": 820 }, { "epoch": 0.083, "grad_norm": 23.628289033516634, "learning_rate": 9.926794280723735e-06, "loss": 4.5309, "step": 830 }, { "epoch": 0.084, "grad_norm": 24.76861079838377, "learning_rate": 9.924007517208388e-06, "loss": 4.4814, "step": 840 }, { "epoch": 0.085, "grad_norm": 22.742218891850197, "learning_rate": 9.921169103184188e-06, "loss": 4.4505, "step": 850 }, { "epoch": 0.086, "grad_norm": 21.926466244318494, "learning_rate": 9.91827906842475e-06, "loss": 4.5008, "step": 860 }, { "epoch": 0.087, "grad_norm": 22.742384821861354, "learning_rate": 9.915337443245175e-06, "loss": 4.3937, "step": 870 }, { "epoch": 0.088, "grad_norm": 22.408404275815325, "learning_rate": 9.912344258501716e-06, "loss": 4.4625, "step": 880 }, { "epoch": 0.089, "grad_norm": 27.310500891991374, "learning_rate": 9.90929954559147e-06, "loss": 4.486, "step": 890 }, { "epoch": 0.09, "grad_norm": 23.645527024427818, "learning_rate": 9.906203336452029e-06, "loss": 4.4662, "step": 900 }, { "epoch": 0.091, "grad_norm": 24.26381380633954, "learning_rate": 9.903055663561166e-06, "loss": 4.4778, "step": 910 }, { "epoch": 0.092, "grad_norm": 23.298837696921417, "learning_rate": 9.899856559936476e-06, "loss": 4.2975, "step": 920 }, { "epoch": 0.093, "grad_norm": 22.603634620121774, "learning_rate": 9.896606059135045e-06, "loss": 4.3188, "step": 930 }, { "epoch": 0.094, "grad_norm": 22.481228710809305, "learning_rate": 9.893304195253088e-06, "loss": 4.2939, "step": 940 }, { "epoch": 0.095, "grad_norm": 24.551612391013666, "learning_rate": 9.889951002925594e-06, "loss": 4.308, "step": 950 }, { "epoch": 0.096, "grad_norm": 22.638329042508637, "learning_rate": 9.886546517325964e-06, "loss": 4.2845, "step": 960 }, { "epoch": 0.097, "grad_norm": 25.92641626399578, "learning_rate": 9.883090774165642e-06, "loss": 4.2674, "step": 970 }, { "epoch": 0.098, "grad_norm": 24.816515447496528, "learning_rate": 9.879583809693737e-06, "loss": 4.2915, "step": 980 }, { "epoch": 0.099, "grad_norm": 26.18107106720688, "learning_rate": 9.87602566069665e-06, "loss": 4.2607, "step": 990 }, { "epoch": 0.1, "grad_norm": 23.418651046462916, "learning_rate": 9.872416364497675e-06, "loss": 4.2707, "step": 1000 }, { "epoch": 0.101, "grad_norm": 22.829135748474357, "learning_rate": 9.868755958956625e-06, "loss": 4.234, "step": 1010 }, { "epoch": 0.102, "grad_norm": 21.522299198555253, "learning_rate": 9.865044482469421e-06, "loss": 4.2311, "step": 1020 }, { "epoch": 0.103, "grad_norm": 23.989884740211636, "learning_rate": 9.861281973967695e-06, "loss": 4.1197, "step": 1030 }, { "epoch": 0.104, "grad_norm": 25.54162814761193, "learning_rate": 9.857468472918382e-06, "loss": 4.1035, "step": 1040 }, { "epoch": 0.105, "grad_norm": 24.509224738118522, "learning_rate": 9.853604019323303e-06, "loss": 4.0812, "step": 1050 }, { "epoch": 0.106, "grad_norm": 24.6565009266652, "learning_rate": 9.849688653718745e-06, "loss": 4.1835, "step": 1060 }, { "epoch": 0.107, "grad_norm": 24.37373148595233, "learning_rate": 9.845722417175044e-06, "loss": 4.1729, "step": 1070 }, { "epoch": 0.108, "grad_norm": 22.328956119788682, "learning_rate": 9.841705351296142e-06, "loss": 4.0466, "step": 1080 }, { "epoch": 0.109, "grad_norm": 22.333756682355045, "learning_rate": 9.837637498219162e-06, "loss": 4.083, "step": 1090 }, { "epoch": 0.11, "grad_norm": 24.638185594781152, "learning_rate": 9.833518900613957e-06, "loss": 4.1452, "step": 1100 }, { "epoch": 0.111, "grad_norm": 25.90941567119859, "learning_rate": 9.829349601682664e-06, "loss": 4.0439, "step": 1110 }, { "epoch": 0.112, "grad_norm": 25.202965431802678, "learning_rate": 9.82512964515926e-06, "loss": 4.0828, "step": 1120 }, { "epoch": 0.113, "grad_norm": 22.576978277958307, "learning_rate": 9.820859075309088e-06, "loss": 3.9728, "step": 1130 }, { "epoch": 0.114, "grad_norm": 25.75780891671629, "learning_rate": 9.81653793692841e-06, "loss": 3.9884, "step": 1140 }, { "epoch": 0.115, "grad_norm": 20.930205581044806, "learning_rate": 9.812166275343917e-06, "loss": 3.9746, "step": 1150 }, { "epoch": 0.116, "grad_norm": 22.29066921669641, "learning_rate": 9.807744136412276e-06, "loss": 4.0443, "step": 1160 }, { "epoch": 0.117, "grad_norm": 22.323337818346655, "learning_rate": 9.803271566519627e-06, "loss": 3.9717, "step": 1170 }, { "epoch": 0.118, "grad_norm": 23.85436114025019, "learning_rate": 9.798748612581117e-06, "loss": 3.9414, "step": 1180 }, { "epoch": 0.119, "grad_norm": 22.036908494286084, "learning_rate": 9.794175322040388e-06, "loss": 3.9784, "step": 1190 }, { "epoch": 0.12, "grad_norm": 21.34683696754092, "learning_rate": 9.789551742869096e-06, "loss": 3.8458, "step": 1200 }, { "epoch": 0.121, "grad_norm": 22.813680211480126, "learning_rate": 9.784877923566398e-06, "loss": 3.9741, "step": 1210 }, { "epoch": 0.122, "grad_norm": 23.105729666312968, "learning_rate": 9.780153913158446e-06, "loss": 3.9021, "step": 1220 }, { "epoch": 0.123, "grad_norm": 23.027323320663502, "learning_rate": 9.775379761197876e-06, "loss": 3.8876, "step": 1230 }, { "epoch": 0.124, "grad_norm": 25.584002997501507, "learning_rate": 9.770555517763281e-06, "loss": 3.9344, "step": 1240 }, { "epoch": 0.125, "grad_norm": 24.812419781569233, "learning_rate": 9.765681233458694e-06, "loss": 3.9631, "step": 1250 }, { "epoch": 0.126, "grad_norm": 23.628646212594592, "learning_rate": 9.76075695941305e-06, "loss": 3.8281, "step": 1260 }, { "epoch": 0.127, "grad_norm": 22.05747214779236, "learning_rate": 9.755782747279649e-06, "loss": 3.8601, "step": 1270 }, { "epoch": 0.128, "grad_norm": 22.178176526507933, "learning_rate": 9.75075864923563e-06, "loss": 3.8407, "step": 1280 }, { "epoch": 0.129, "grad_norm": 23.0475684286927, "learning_rate": 9.745684717981401e-06, "loss": 3.8942, "step": 1290 }, { "epoch": 0.13, "grad_norm": 24.282574871679643, "learning_rate": 9.7405610067401e-06, "loss": 3.7417, "step": 1300 }, { "epoch": 0.131, "grad_norm": 22.449241778033034, "learning_rate": 9.735387569257031e-06, "loss": 3.8135, "step": 1310 }, { "epoch": 0.132, "grad_norm": 23.355287073687737, "learning_rate": 9.730164459799108e-06, "loss": 3.8165, "step": 1320 }, { "epoch": 0.133, "grad_norm": 23.73370539328191, "learning_rate": 9.724891733154276e-06, "loss": 3.7663, "step": 1330 }, { "epoch": 0.134, "grad_norm": 23.099427356078, "learning_rate": 9.719569444630944e-06, "loss": 3.6456, "step": 1340 }, { "epoch": 0.135, "grad_norm": 21.398306110086555, "learning_rate": 9.714197650057402e-06, "loss": 3.724, "step": 1350 }, { "epoch": 0.136, "grad_norm": 21.091672060788312, "learning_rate": 9.708776405781229e-06, "loss": 3.7181, "step": 1360 }, { "epoch": 0.137, "grad_norm": 22.95075179863374, "learning_rate": 9.703305768668713e-06, "loss": 3.7837, "step": 1370 }, { "epoch": 0.138, "grad_norm": 26.073301595865512, "learning_rate": 9.69778579610425e-06, "loss": 3.7436, "step": 1380 }, { "epoch": 0.139, "grad_norm": 24.202878327157745, "learning_rate": 9.69221654598974e-06, "loss": 3.7349, "step": 1390 }, { "epoch": 0.14, "grad_norm": 24.345618200665633, "learning_rate": 9.686598076743981e-06, "loss": 3.7148, "step": 1400 }, { "epoch": 0.141, "grad_norm": 22.491396233440565, "learning_rate": 9.680930447302058e-06, "loss": 3.7294, "step": 1410 }, { "epoch": 0.142, "grad_norm": 22.393360798961915, "learning_rate": 9.67521371711472e-06, "loss": 3.768, "step": 1420 }, { "epoch": 0.143, "grad_norm": 26.99990107456492, "learning_rate": 9.669447946147763e-06, "loss": 3.6831, "step": 1430 }, { "epoch": 0.144, "grad_norm": 24.751464864836837, "learning_rate": 9.663633194881394e-06, "loss": 3.7367, "step": 1440 }, { "epoch": 0.145, "grad_norm": 24.69603609228496, "learning_rate": 9.657769524309606e-06, "loss": 3.758, "step": 1450 }, { "epoch": 0.146, "grad_norm": 22.84201571095169, "learning_rate": 9.651856995939526e-06, "loss": 3.7002, "step": 1460 }, { "epoch": 0.147, "grad_norm": 24.626128545429513, "learning_rate": 9.645895671790778e-06, "loss": 3.6695, "step": 1470 }, { "epoch": 0.148, "grad_norm": 24.911567343494106, "learning_rate": 9.639885614394835e-06, "loss": 3.6367, "step": 1480 }, { "epoch": 0.149, "grad_norm": 25.615488910662624, "learning_rate": 9.633826886794351e-06, "loss": 3.6069, "step": 1490 }, { "epoch": 0.15, "grad_norm": 22.48458222482155, "learning_rate": 9.627719552542515e-06, "loss": 3.6477, "step": 1500 }, { "epoch": 0.151, "grad_norm": 22.61820302053004, "learning_rate": 9.621563675702372e-06, "loss": 3.6222, "step": 1510 }, { "epoch": 0.152, "grad_norm": 22.227065511210427, "learning_rate": 9.615359320846156e-06, "loss": 3.6263, "step": 1520 }, { "epoch": 0.153, "grad_norm": 23.134734806301392, "learning_rate": 9.609106553054617e-06, "loss": 3.5971, "step": 1530 }, { "epoch": 0.154, "grad_norm": 23.555174086868778, "learning_rate": 9.602805437916324e-06, "loss": 3.6728, "step": 1540 }, { "epoch": 0.155, "grad_norm": 24.660554446342175, "learning_rate": 9.596456041527001e-06, "loss": 3.6199, "step": 1550 }, { "epoch": 0.156, "grad_norm": 23.023056725545683, "learning_rate": 9.590058430488807e-06, "loss": 3.6098, "step": 1560 }, { "epoch": 0.157, "grad_norm": 22.834975585604365, "learning_rate": 9.583612671909655e-06, "loss": 3.6631, "step": 1570 }, { "epoch": 0.158, "grad_norm": 24.723584349187675, "learning_rate": 9.577118833402503e-06, "loss": 3.6218, "step": 1580 }, { "epoch": 0.159, "grad_norm": 24.314693287777896, "learning_rate": 9.570576983084644e-06, "loss": 3.5868, "step": 1590 }, { "epoch": 0.16, "grad_norm": 23.117296732843023, "learning_rate": 9.563987189576991e-06, "loss": 3.5894, "step": 1600 }, { "epoch": 0.161, "grad_norm": 22.41097627664212, "learning_rate": 9.557349522003363e-06, "loss": 3.554, "step": 1610 }, { "epoch": 0.162, "grad_norm": 22.79794306624016, "learning_rate": 9.550664049989748e-06, "loss": 3.5648, "step": 1620 }, { "epoch": 0.163, "grad_norm": 24.214326234284798, "learning_rate": 9.543930843663585e-06, "loss": 3.5563, "step": 1630 }, { "epoch": 0.164, "grad_norm": 23.50823079835679, "learning_rate": 9.537149973653026e-06, "loss": 3.5483, "step": 1640 }, { "epoch": 0.165, "grad_norm": 24.195064184437427, "learning_rate": 9.530321511086182e-06, "loss": 3.5255, "step": 1650 }, { "epoch": 0.166, "grad_norm": 24.48737467923783, "learning_rate": 9.5234455275904e-06, "loss": 3.6298, "step": 1660 }, { "epoch": 0.167, "grad_norm": 22.952096083715436, "learning_rate": 9.51652209529149e-06, "loss": 3.5452, "step": 1670 }, { "epoch": 0.168, "grad_norm": 24.332503011367194, "learning_rate": 9.509551286812983e-06, "loss": 3.5813, "step": 1680 }, { "epoch": 0.169, "grad_norm": 24.06297702933911, "learning_rate": 9.502533175275362e-06, "loss": 3.5345, "step": 1690 }, { "epoch": 0.17, "grad_norm": 22.927394095288108, "learning_rate": 9.495467834295292e-06, "loss": 3.4628, "step": 1700 }, { "epoch": 0.171, "grad_norm": 23.564468578137184, "learning_rate": 9.488355337984858e-06, "loss": 3.5022, "step": 1710 }, { "epoch": 0.172, "grad_norm": 23.376728452748708, "learning_rate": 9.481195760950782e-06, "loss": 3.4329, "step": 1720 }, { "epoch": 0.173, "grad_norm": 23.443321355874644, "learning_rate": 9.473989178293642e-06, "loss": 3.4551, "step": 1730 }, { "epoch": 0.174, "grad_norm": 25.188588920395837, "learning_rate": 9.466735665607075e-06, "loss": 3.4936, "step": 1740 }, { "epoch": 0.175, "grad_norm": 24.02989588062241, "learning_rate": 9.459435298976998e-06, "loss": 3.4129, "step": 1750 }, { "epoch": 0.176, "grad_norm": 25.25515813836085, "learning_rate": 9.452088154980803e-06, "loss": 3.4433, "step": 1760 }, { "epoch": 0.177, "grad_norm": 22.71910646985039, "learning_rate": 9.444694310686554e-06, "loss": 3.4517, "step": 1770 }, { "epoch": 0.178, "grad_norm": 22.177317572693116, "learning_rate": 9.43725384365218e-06, "loss": 3.4649, "step": 1780 }, { "epoch": 0.179, "grad_norm": 24.13024671922974, "learning_rate": 9.429766831924659e-06, "loss": 3.4621, "step": 1790 }, { "epoch": 0.18, "grad_norm": 23.911033745387464, "learning_rate": 9.422233354039199e-06, "loss": 3.4405, "step": 1800 }, { "epoch": 0.181, "grad_norm": 23.73595507565857, "learning_rate": 9.414653489018419e-06, "loss": 3.4221, "step": 1810 }, { "epoch": 0.182, "grad_norm": 24.6021919036614, "learning_rate": 9.407027316371519e-06, "loss": 3.4076, "step": 1820 }, { "epoch": 0.183, "grad_norm": 24.65212303310678, "learning_rate": 9.39935491609344e-06, "loss": 3.382, "step": 1830 }, { "epoch": 0.184, "grad_norm": 23.07392583195514, "learning_rate": 9.39163636866403e-06, "loss": 3.5137, "step": 1840 }, { "epoch": 0.185, "grad_norm": 25.490612484070066, "learning_rate": 9.383871755047198e-06, "loss": 3.3762, "step": 1850 }, { "epoch": 0.186, "grad_norm": 24.962536013499314, "learning_rate": 9.376061156690068e-06, "loss": 3.4027, "step": 1860 }, { "epoch": 0.187, "grad_norm": 21.927242813619642, "learning_rate": 9.368204655522122e-06, "loss": 3.3624, "step": 1870 }, { "epoch": 0.188, "grad_norm": 22.90346457974155, "learning_rate": 9.360302333954337e-06, "loss": 3.2714, "step": 1880 }, { "epoch": 0.189, "grad_norm": 21.42728890556583, "learning_rate": 9.352354274878329e-06, "loss": 3.4341, "step": 1890 }, { "epoch": 0.19, "grad_norm": 24.2954221362493, "learning_rate": 9.344360561665476e-06, "loss": 3.319, "step": 1900 }, { "epoch": 0.191, "grad_norm": 24.195541492062425, "learning_rate": 9.336321278166047e-06, "loss": 3.3551, "step": 1910 }, { "epoch": 0.192, "grad_norm": 27.124161917689207, "learning_rate": 9.328236508708322e-06, "loss": 3.359, "step": 1920 }, { "epoch": 0.193, "grad_norm": 23.28646313605679, "learning_rate": 9.320106338097708e-06, "loss": 3.3477, "step": 1930 }, { "epoch": 0.194, "grad_norm": 24.301536656866322, "learning_rate": 9.311930851615845e-06, "loss": 3.4302, "step": 1940 }, { "epoch": 0.195, "grad_norm": 25.53306385839391, "learning_rate": 9.30371013501972e-06, "loss": 3.2911, "step": 1950 }, { "epoch": 0.196, "grad_norm": 25.21626457511393, "learning_rate": 9.295444274540757e-06, "loss": 3.3898, "step": 1960 }, { "epoch": 0.197, "grad_norm": 23.25154130434933, "learning_rate": 9.287133356883925e-06, "loss": 3.3684, "step": 1970 }, { "epoch": 0.198, "grad_norm": 25.106432355414906, "learning_rate": 9.278777469226815e-06, "loss": 3.3229, "step": 1980 }, { "epoch": 0.199, "grad_norm": 21.394826627734666, "learning_rate": 9.270376699218737e-06, "loss": 3.3207, "step": 1990 }, { "epoch": 0.2, "grad_norm": 23.489372528771135, "learning_rate": 9.26193113497979e-06, "loss": 3.3743, "step": 2000 }, { "epoch": 0.201, "grad_norm": 25.75762830216029, "learning_rate": 9.25344086509995e-06, "loss": 3.2556, "step": 2010 }, { "epoch": 0.202, "grad_norm": 23.416865213254997, "learning_rate": 9.244905978638127e-06, "loss": 3.2713, "step": 2020 }, { "epoch": 0.203, "grad_norm": 28.189975318882162, "learning_rate": 9.236326565121242e-06, "loss": 3.2561, "step": 2030 }, { "epoch": 0.204, "grad_norm": 24.01178695754152, "learning_rate": 9.227702714543288e-06, "loss": 3.2741, "step": 2040 }, { "epoch": 0.205, "grad_norm": 22.302087692875403, "learning_rate": 9.219034517364369e-06, "loss": 3.2485, "step": 2050 }, { "epoch": 0.206, "grad_norm": 25.54566886899162, "learning_rate": 9.210322064509778e-06, "loss": 3.3286, "step": 2060 }, { "epoch": 0.207, "grad_norm": 23.26287762742553, "learning_rate": 9.201565447369019e-06, "loss": 3.3011, "step": 2070 }, { "epoch": 0.208, "grad_norm": 23.703640948336055, "learning_rate": 9.192764757794863e-06, "loss": 3.3323, "step": 2080 }, { "epoch": 0.209, "grad_norm": 23.357452612025767, "learning_rate": 9.18392008810238e-06, "loss": 3.2175, "step": 2090 }, { "epoch": 0.21, "grad_norm": 23.51313898370609, "learning_rate": 9.17503153106797e-06, "loss": 3.318, "step": 2100 }, { "epoch": 0.211, "grad_norm": 21.730315616740754, "learning_rate": 9.166099179928391e-06, "loss": 3.1358, "step": 2110 }, { "epoch": 0.212, "grad_norm": 23.798908528617197, "learning_rate": 9.157123128379778e-06, "loss": 3.2717, "step": 2120 }, { "epoch": 0.213, "grad_norm": 23.15411276423997, "learning_rate": 9.148103470576667e-06, "loss": 3.2984, "step": 2130 }, { "epoch": 0.214, "grad_norm": 25.6751311476953, "learning_rate": 9.139040301130995e-06, "loss": 3.3074, "step": 2140 }, { "epoch": 0.215, "grad_norm": 24.53621575638315, "learning_rate": 9.129933715111126e-06, "loss": 3.2676, "step": 2150 }, { "epoch": 0.216, "grad_norm": 23.633558372571866, "learning_rate": 9.120783808040837e-06, "loss": 3.2197, "step": 2160 }, { "epoch": 0.217, "grad_norm": 21.224259585548143, "learning_rate": 9.111590675898321e-06, "loss": 3.3398, "step": 2170 }, { "epoch": 0.218, "grad_norm": 21.845275197826645, "learning_rate": 9.102354415115186e-06, "loss": 3.3465, "step": 2180 }, { "epoch": 0.219, "grad_norm": 22.763700480310355, "learning_rate": 9.093075122575438e-06, "loss": 3.2089, "step": 2190 }, { "epoch": 0.22, "grad_norm": 26.440914868643688, "learning_rate": 9.083752895614466e-06, "loss": 3.3029, "step": 2200 }, { "epoch": 0.221, "grad_norm": 26.933703656298324, "learning_rate": 9.074387832018014e-06, "loss": 3.2648, "step": 2210 }, { "epoch": 0.222, "grad_norm": 24.363181677557844, "learning_rate": 9.064980030021173e-06, "loss": 3.2516, "step": 2220 }, { "epoch": 0.223, "grad_norm": 26.28683157023169, "learning_rate": 9.055529588307332e-06, "loss": 3.2713, "step": 2230 }, { "epoch": 0.224, "grad_norm": 25.52108125717531, "learning_rate": 9.046036606007154e-06, "loss": 3.3863, "step": 2240 }, { "epoch": 0.225, "grad_norm": 23.636426252948127, "learning_rate": 9.03650118269753e-06, "loss": 3.2403, "step": 2250 }, { "epoch": 0.226, "grad_norm": 21.973575058376877, "learning_rate": 9.026923418400538e-06, "loss": 3.2839, "step": 2260 }, { "epoch": 0.227, "grad_norm": 24.103564550272957, "learning_rate": 9.017303413582397e-06, "loss": 3.1788, "step": 2270 }, { "epoch": 0.228, "grad_norm": 24.302549287531424, "learning_rate": 9.0076412691524e-06, "loss": 3.284, "step": 2280 }, { "epoch": 0.229, "grad_norm": 24.128669441347036, "learning_rate": 8.997937086461874e-06, "loss": 3.1998, "step": 2290 }, { "epoch": 0.23, "grad_norm": 22.83300066091983, "learning_rate": 8.9881909673031e-06, "loss": 3.2838, "step": 2300 }, { "epoch": 0.231, "grad_norm": 22.833967131630054, "learning_rate": 8.97840301390826e-06, "loss": 3.2641, "step": 2310 }, { "epoch": 0.232, "grad_norm": 22.669664313225056, "learning_rate": 8.968573328948354e-06, "loss": 3.2564, "step": 2320 }, { "epoch": 0.233, "grad_norm": 23.91876883524881, "learning_rate": 8.958702015532123e-06, "loss": 3.2581, "step": 2330 }, { "epoch": 0.234, "grad_norm": 24.480554952881253, "learning_rate": 8.948789177204972e-06, "loss": 3.2595, "step": 2340 }, { "epoch": 0.235, "grad_norm": 21.59460421453609, "learning_rate": 8.93883491794789e-06, "loss": 3.1977, "step": 2350 }, { "epoch": 0.236, "grad_norm": 21.21134669814089, "learning_rate": 8.928839342176342e-06, "loss": 3.2995, "step": 2360 }, { "epoch": 0.237, "grad_norm": 27.75618242855786, "learning_rate": 8.91880255473919e-06, "loss": 3.1716, "step": 2370 }, { "epoch": 0.238, "grad_norm": 22.179107320115204, "learning_rate": 8.908724660917583e-06, "loss": 3.1607, "step": 2380 }, { "epoch": 0.239, "grad_norm": 22.610763887544532, "learning_rate": 8.89860576642386e-06, "loss": 3.2004, "step": 2390 }, { "epoch": 0.24, "grad_norm": 24.35570838505421, "learning_rate": 8.888445977400435e-06, "loss": 3.2074, "step": 2400 }, { "epoch": 0.241, "grad_norm": 25.35604530733846, "learning_rate": 8.878245400418689e-06, "loss": 3.1654, "step": 2410 }, { "epoch": 0.242, "grad_norm": 23.61985699970461, "learning_rate": 8.868004142477846e-06, "loss": 3.1347, "step": 2420 }, { "epoch": 0.243, "grad_norm": 25.934109942553864, "learning_rate": 8.857722311003857e-06, "loss": 3.2037, "step": 2430 }, { "epoch": 0.244, "grad_norm": 21.64044955602141, "learning_rate": 8.847400013848268e-06, "loss": 3.1032, "step": 2440 }, { "epoch": 0.245, "grad_norm": 21.616332033681037, "learning_rate": 8.837037359287093e-06, "loss": 3.0841, "step": 2450 }, { "epoch": 0.246, "grad_norm": 24.087994146507274, "learning_rate": 8.826634456019673e-06, "loss": 3.0812, "step": 2460 }, { "epoch": 0.247, "grad_norm": 22.62182748703039, "learning_rate": 8.816191413167541e-06, "loss": 3.1154, "step": 2470 }, { "epoch": 0.248, "grad_norm": 22.98613865587244, "learning_rate": 8.805708340273275e-06, "loss": 3.1732, "step": 2480 }, { "epoch": 0.249, "grad_norm": 22.41515107722678, "learning_rate": 8.79518534729935e-06, "loss": 3.1031, "step": 2490 }, { "epoch": 0.25, "grad_norm": 21.737785627599212, "learning_rate": 8.784622544626985e-06, "loss": 3.0525, "step": 2500 }, { "epoch": 0.251, "grad_norm": 22.862970889097603, "learning_rate": 8.774020043054979e-06, "loss": 3.0877, "step": 2510 }, { "epoch": 0.252, "grad_norm": 22.46171748551351, "learning_rate": 8.763377953798555e-06, "loss": 3.0955, "step": 2520 }, { "epoch": 0.253, "grad_norm": 23.60371454257512, "learning_rate": 8.752696388488196e-06, "loss": 3.0229, "step": 2530 }, { "epoch": 0.254, "grad_norm": 23.777335628783284, "learning_rate": 8.741975459168467e-06, "loss": 3.1196, "step": 2540 }, { "epoch": 0.255, "grad_norm": 24.602127079427465, "learning_rate": 8.731215278296844e-06, "loss": 3.1527, "step": 2550 }, { "epoch": 0.256, "grad_norm": 22.491230184088046, "learning_rate": 8.720415958742532e-06, "loss": 3.1121, "step": 2560 }, { "epoch": 0.257, "grad_norm": 21.77091915970421, "learning_rate": 8.709577613785282e-06, "loss": 3.0764, "step": 2570 }, { "epoch": 0.258, "grad_norm": 22.038704729276436, "learning_rate": 8.698700357114207e-06, "loss": 3.0718, "step": 2580 }, { "epoch": 0.259, "grad_norm": 22.465250049834957, "learning_rate": 8.68778430282658e-06, "loss": 3.056, "step": 2590 }, { "epoch": 0.26, "grad_norm": 23.118244488626626, "learning_rate": 8.676829565426645e-06, "loss": 3.1346, "step": 2600 }, { "epoch": 0.261, "grad_norm": 23.026358377092517, "learning_rate": 8.665836259824417e-06, "loss": 3.0114, "step": 2610 }, { "epoch": 0.262, "grad_norm": 24.585695588515986, "learning_rate": 8.654804501334466e-06, "loss": 3.0735, "step": 2620 }, { "epoch": 0.263, "grad_norm": 21.19937777499543, "learning_rate": 8.643734405674719e-06, "loss": 3.0856, "step": 2630 }, { "epoch": 0.264, "grad_norm": 22.668270691256566, "learning_rate": 8.632626088965241e-06, "loss": 3.087, "step": 2640 }, { "epoch": 0.265, "grad_norm": 22.199203403282162, "learning_rate": 8.62147966772702e-06, "loss": 3.0942, "step": 2650 }, { "epoch": 0.266, "grad_norm": 21.853682607561833, "learning_rate": 8.61029525888074e-06, "loss": 3.0371, "step": 2660 }, { "epoch": 0.267, "grad_norm": 22.991096956289013, "learning_rate": 8.599072979745557e-06, "loss": 2.9975, "step": 2670 }, { "epoch": 0.268, "grad_norm": 23.325970492158056, "learning_rate": 8.587812948037866e-06, "loss": 3.0528, "step": 2680 }, { "epoch": 0.269, "grad_norm": 22.167560935491636, "learning_rate": 8.576515281870075e-06, "loss": 2.9703, "step": 2690 }, { "epoch": 0.27, "grad_norm": 25.80311797544179, "learning_rate": 8.565180099749355e-06, "loss": 3.1497, "step": 2700 }, { "epoch": 0.271, "grad_norm": 23.149575497246605, "learning_rate": 8.553807520576404e-06, "loss": 3.0629, "step": 2710 }, { "epoch": 0.272, "grad_norm": 23.43806263925292, "learning_rate": 8.542397663644193e-06, "loss": 3.0056, "step": 2720 }, { "epoch": 0.273, "grad_norm": 20.598526802687527, "learning_rate": 8.530950648636725e-06, "loss": 2.9975, "step": 2730 }, { "epoch": 0.274, "grad_norm": 23.395232439765152, "learning_rate": 8.51946659562777e-06, "loss": 3.0091, "step": 2740 }, { "epoch": 0.275, "grad_norm": 27.521377354120137, "learning_rate": 8.50794562507961e-06, "loss": 2.9817, "step": 2750 }, { "epoch": 0.276, "grad_norm": 23.554507487051907, "learning_rate": 8.496387857841772e-06, "loss": 3.0144, "step": 2760 }, { "epoch": 0.277, "grad_norm": 22.248471802661705, "learning_rate": 8.48479341514977e-06, "loss": 2.9956, "step": 2770 }, { "epoch": 0.278, "grad_norm": 23.920360565286423, "learning_rate": 8.473162418623818e-06, "loss": 2.9666, "step": 2780 }, { "epoch": 0.279, "grad_norm": 24.498063373758548, "learning_rate": 8.461494990267568e-06, "loss": 3.0246, "step": 2790 }, { "epoch": 0.28, "grad_norm": 22.130838306134713, "learning_rate": 8.449791252466818e-06, "loss": 2.987, "step": 2800 }, { "epoch": 0.281, "grad_norm": 22.071686070004766, "learning_rate": 8.438051327988246e-06, "loss": 3.0414, "step": 2810 }, { "epoch": 0.282, "grad_norm": 22.885241899871122, "learning_rate": 8.4262753399781e-06, "loss": 3.0177, "step": 2820 }, { "epoch": 0.283, "grad_norm": 23.73524023792324, "learning_rate": 8.414463411960922e-06, "loss": 2.9935, "step": 2830 }, { "epoch": 0.284, "grad_norm": 20.534636994036997, "learning_rate": 8.402615667838253e-06, "loss": 3.0122, "step": 2840 }, { "epoch": 0.285, "grad_norm": 22.543252392191082, "learning_rate": 8.390732231887314e-06, "loss": 2.9858, "step": 2850 }, { "epoch": 0.286, "grad_norm": 23.753114005682967, "learning_rate": 8.37881322875973e-06, "loss": 3.0714, "step": 2860 }, { "epoch": 0.287, "grad_norm": 21.509731585882395, "learning_rate": 8.366858783480205e-06, "loss": 2.9722, "step": 2870 }, { "epoch": 0.288, "grad_norm": 22.75584298828593, "learning_rate": 8.35486902144521e-06, "loss": 2.9434, "step": 2880 }, { "epoch": 0.289, "grad_norm": 23.480554393750634, "learning_rate": 8.342844068421678e-06, "loss": 2.9462, "step": 2890 }, { "epoch": 0.29, "grad_norm": 22.857741176994672, "learning_rate": 8.330784050545672e-06, "loss": 2.9492, "step": 2900 }, { "epoch": 0.291, "grad_norm": 22.83755236126708, "learning_rate": 8.318689094321078e-06, "loss": 2.9077, "step": 2910 }, { "epoch": 0.292, "grad_norm": 22.273864654510238, "learning_rate": 8.30655932661826e-06, "loss": 2.9717, "step": 2920 }, { "epoch": 0.293, "grad_norm": 23.184809701967776, "learning_rate": 8.294394874672744e-06, "loss": 2.9619, "step": 2930 }, { "epoch": 0.294, "grad_norm": 22.032577490070956, "learning_rate": 8.282195866083872e-06, "loss": 2.9748, "step": 2940 }, { "epoch": 0.295, "grad_norm": 21.886379251272942, "learning_rate": 8.269962428813474e-06, "loss": 2.8442, "step": 2950 }, { "epoch": 0.296, "grad_norm": 22.811966988927562, "learning_rate": 8.257694691184517e-06, "loss": 2.9171, "step": 2960 }, { "epoch": 0.297, "grad_norm": 23.384747347892507, "learning_rate": 8.245392781879765e-06, "loss": 2.8893, "step": 2970 }, { "epoch": 0.298, "grad_norm": 23.841760221766826, "learning_rate": 8.233056829940423e-06, "loss": 2.9231, "step": 2980 }, { "epoch": 0.299, "grad_norm": 23.27198792113873, "learning_rate": 8.220686964764792e-06, "loss": 2.9246, "step": 2990 }, { "epoch": 0.3, "grad_norm": 22.292544418423027, "learning_rate": 8.208283316106901e-06, "loss": 2.8433, "step": 3000 }, { "epoch": 0.301, "grad_norm": 22.12189691936275, "learning_rate": 8.195846014075161e-06, "loss": 2.8666, "step": 3010 }, { "epoch": 0.302, "grad_norm": 22.751439509296812, "learning_rate": 8.18337518913098e-06, "loss": 2.9506, "step": 3020 }, { "epoch": 0.303, "grad_norm": 29.242292798748753, "learning_rate": 8.170870972087412e-06, "loss": 2.9248, "step": 3030 }, { "epoch": 0.304, "grad_norm": 21.77319818428595, "learning_rate": 8.158333494107779e-06, "loss": 2.9063, "step": 3040 }, { "epoch": 0.305, "grad_norm": 21.897072508229098, "learning_rate": 8.145762886704288e-06, "loss": 2.9129, "step": 3050 }, { "epoch": 0.306, "grad_norm": 24.65353392973142, "learning_rate": 8.133159281736665e-06, "loss": 2.8249, "step": 3060 }, { "epoch": 0.307, "grad_norm": 23.507339515007676, "learning_rate": 8.120522811410763e-06, "loss": 2.8132, "step": 3070 }, { "epoch": 0.308, "grad_norm": 20.5559811305243, "learning_rate": 8.107853608277176e-06, "loss": 2.9035, "step": 3080 }, { "epoch": 0.309, "grad_norm": 22.037713822234664, "learning_rate": 8.095151805229848e-06, "loss": 2.8378, "step": 3090 }, { "epoch": 0.31, "grad_norm": 22.519898191765222, "learning_rate": 8.082417535504682e-06, "loss": 2.7761, "step": 3100 }, { "epoch": 0.311, "grad_norm": 23.098033680131245, "learning_rate": 8.069650932678146e-06, "loss": 2.8211, "step": 3110 }, { "epoch": 0.312, "grad_norm": 24.92701947245253, "learning_rate": 8.05685213066586e-06, "loss": 2.8556, "step": 3120 }, { "epoch": 0.313, "grad_norm": 24.904407264952166, "learning_rate": 8.0440212637212e-06, "loss": 2.9596, "step": 3130 }, { "epoch": 0.314, "grad_norm": 24.020594042453077, "learning_rate": 8.031158466433889e-06, "loss": 2.9234, "step": 3140 }, { "epoch": 0.315, "grad_norm": 23.73994421946332, "learning_rate": 8.018263873728586e-06, "loss": 2.9004, "step": 3150 }, { "epoch": 0.316, "grad_norm": 21.30677964784449, "learning_rate": 8.005337620863465e-06, "loss": 2.9174, "step": 3160 }, { "epoch": 0.317, "grad_norm": 23.42753229463555, "learning_rate": 7.992379843428801e-06, "loss": 2.8446, "step": 3170 }, { "epoch": 0.318, "grad_norm": 24.148353049427204, "learning_rate": 7.97939067734555e-06, "loss": 2.8335, "step": 3180 }, { "epoch": 0.319, "grad_norm": 23.36406140539168, "learning_rate": 7.966370258863918e-06, "loss": 2.8962, "step": 3190 }, { "epoch": 0.32, "grad_norm": 22.312623082846226, "learning_rate": 7.953318724561932e-06, "loss": 2.8152, "step": 3200 }, { "epoch": 0.321, "grad_norm": 24.33003576719242, "learning_rate": 7.940236211344011e-06, "loss": 2.8617, "step": 3210 }, { "epoch": 0.322, "grad_norm": 22.531915397315693, "learning_rate": 7.92712285643953e-06, "loss": 2.8531, "step": 3220 }, { "epoch": 0.323, "grad_norm": 23.830897140373573, "learning_rate": 7.913978797401374e-06, "loss": 2.881, "step": 3230 }, { "epoch": 0.324, "grad_norm": 26.17870869062595, "learning_rate": 7.900804172104506e-06, "loss": 2.8144, "step": 3240 }, { "epoch": 0.325, "grad_norm": 22.342738532499705, "learning_rate": 7.887599118744508e-06, "loss": 2.8616, "step": 3250 }, { "epoch": 0.326, "grad_norm": 25.382437702862997, "learning_rate": 7.874363775836144e-06, "loss": 2.921, "step": 3260 }, { "epoch": 0.327, "grad_norm": 21.765457019570892, "learning_rate": 7.861098282211896e-06, "loss": 2.7945, "step": 3270 }, { "epoch": 0.328, "grad_norm": 23.164122849384523, "learning_rate": 7.847802777020518e-06, "loss": 2.779, "step": 3280 }, { "epoch": 0.329, "grad_norm": 23.085600865230653, "learning_rate": 7.834477399725564e-06, "loss": 2.8444, "step": 3290 }, { "epoch": 0.33, "grad_norm": 22.611759906246267, "learning_rate": 7.821122290103938e-06, "loss": 2.8238, "step": 3300 }, { "epoch": 0.331, "grad_norm": 23.96471666785546, "learning_rate": 7.807737588244422e-06, "loss": 2.852, "step": 3310 }, { "epoch": 0.332, "grad_norm": 29.45188942740737, "learning_rate": 7.794323434546205e-06, "loss": 2.8929, "step": 3320 }, { "epoch": 0.333, "grad_norm": 21.68240585365781, "learning_rate": 7.780879969717409e-06, "loss": 2.83, "step": 3330 }, { "epoch": 0.334, "grad_norm": 22.685349778281424, "learning_rate": 7.76740733477362e-06, "loss": 2.8292, "step": 3340 }, { "epoch": 0.335, "grad_norm": 23.456315869016926, "learning_rate": 7.753905671036403e-06, "loss": 2.8169, "step": 3350 }, { "epoch": 0.336, "grad_norm": 21.822513744622988, "learning_rate": 7.740375120131826e-06, "loss": 2.779, "step": 3360 }, { "epoch": 0.337, "grad_norm": 24.073227711631965, "learning_rate": 7.72681582398896e-06, "loss": 2.8207, "step": 3370 }, { "epoch": 0.338, "grad_norm": 22.74996503336556, "learning_rate": 7.713227924838407e-06, "loss": 2.7891, "step": 3380 }, { "epoch": 0.339, "grad_norm": 22.229780852752135, "learning_rate": 7.699611565210801e-06, "loss": 2.8154, "step": 3390 }, { "epoch": 0.34, "grad_norm": 24.443239990227884, "learning_rate": 7.685966887935309e-06, "loss": 2.7987, "step": 3400 }, { "epoch": 0.341, "grad_norm": 23.147260545331584, "learning_rate": 7.67229403613814e-06, "loss": 2.8111, "step": 3410 }, { "epoch": 0.342, "grad_norm": 22.55582400125668, "learning_rate": 7.658593153241041e-06, "loss": 2.865, "step": 3420 }, { "epoch": 0.343, "grad_norm": 22.85955171360443, "learning_rate": 7.644864382959787e-06, "loss": 2.7519, "step": 3430 }, { "epoch": 0.344, "grad_norm": 21.50029735100571, "learning_rate": 7.631107869302685e-06, "loss": 2.8122, "step": 3440 }, { "epoch": 0.345, "grad_norm": 22.571884756705916, "learning_rate": 7.617323756569053e-06, "loss": 2.7594, "step": 3450 }, { "epoch": 0.346, "grad_norm": 21.517950714110395, "learning_rate": 7.6035121893477115e-06, "loss": 2.7576, "step": 3460 }, { "epoch": 0.347, "grad_norm": 20.201621635236233, "learning_rate": 7.5896733125154646e-06, "loss": 2.7714, "step": 3470 }, { "epoch": 0.348, "grad_norm": 21.320057713979505, "learning_rate": 7.57580727123558e-06, "loss": 2.7672, "step": 3480 }, { "epoch": 0.349, "grad_norm": 21.77419275648284, "learning_rate": 7.561914210956272e-06, "loss": 2.7646, "step": 3490 }, { "epoch": 0.35, "grad_norm": 22.42807635608928, "learning_rate": 7.547994277409168e-06, "loss": 2.705, "step": 3500 }, { "epoch": 0.351, "grad_norm": 21.288135887785597, "learning_rate": 7.534047616607784e-06, "loss": 2.7893, "step": 3510 }, { "epoch": 0.352, "grad_norm": 20.2124745536486, "learning_rate": 7.520074374845992e-06, "loss": 2.8025, "step": 3520 }, { "epoch": 0.353, "grad_norm": 21.322943257388207, "learning_rate": 7.5060746986964884e-06, "loss": 2.7657, "step": 3530 }, { "epoch": 0.354, "grad_norm": 23.762576289110868, "learning_rate": 7.492048735009248e-06, "loss": 2.8, "step": 3540 }, { "epoch": 0.355, "grad_norm": 23.222446413227672, "learning_rate": 7.477996630909994e-06, "loss": 2.8626, "step": 3550 }, { "epoch": 0.356, "grad_norm": 24.046336162054114, "learning_rate": 7.4639185337986504e-06, "loss": 2.8695, "step": 3560 }, { "epoch": 0.357, "grad_norm": 23.702870091257722, "learning_rate": 7.449814591347794e-06, "loss": 2.8251, "step": 3570 }, { "epoch": 0.358, "grad_norm": 23.341227587431113, "learning_rate": 7.4356849515011055e-06, "loss": 2.828, "step": 3580 }, { "epoch": 0.359, "grad_norm": 21.517491524357975, "learning_rate": 7.421529762471822e-06, "loss": 2.7802, "step": 3590 }, { "epoch": 0.36, "grad_norm": 22.39000100685624, "learning_rate": 7.4073491727411795e-06, "loss": 2.7937, "step": 3600 }, { "epoch": 0.361, "grad_norm": 23.019647163380704, "learning_rate": 7.393143331056852e-06, "loss": 2.7362, "step": 3610 }, { "epoch": 0.362, "grad_norm": 22.973077334604596, "learning_rate": 7.3789123864313986e-06, "loss": 2.7692, "step": 3620 }, { "epoch": 0.363, "grad_norm": 22.00723881602885, "learning_rate": 7.364656488140693e-06, "loss": 2.7412, "step": 3630 }, { "epoch": 0.364, "grad_norm": 21.611581714239197, "learning_rate": 7.350375785722362e-06, "loss": 2.7337, "step": 3640 }, { "epoch": 0.365, "grad_norm": 22.218995380939777, "learning_rate": 7.336070428974218e-06, "loss": 2.7321, "step": 3650 }, { "epoch": 0.366, "grad_norm": 21.075551620705937, "learning_rate": 7.321740567952683e-06, "loss": 2.8026, "step": 3660 }, { "epoch": 0.367, "grad_norm": 22.28316890877753, "learning_rate": 7.307386352971218e-06, "loss": 2.7924, "step": 3670 }, { "epoch": 0.368, "grad_norm": 21.224316520911447, "learning_rate": 7.2930079345987495e-06, "loss": 2.6659, "step": 3680 }, { "epoch": 0.369, "grad_norm": 22.698719038421157, "learning_rate": 7.278605463658079e-06, "loss": 2.7666, "step": 3690 }, { "epoch": 0.37, "grad_norm": 22.868759324337105, "learning_rate": 7.264179091224311e-06, "loss": 2.764, "step": 3700 }, { "epoch": 0.371, "grad_norm": 22.604326901777995, "learning_rate": 7.249728968623268e-06, "loss": 2.7935, "step": 3710 }, { "epoch": 0.372, "grad_norm": 23.441679377771507, "learning_rate": 7.235255247429898e-06, "loss": 2.7883, "step": 3720 }, { "epoch": 0.373, "grad_norm": 24.749228577963667, "learning_rate": 7.220758079466687e-06, "loss": 2.7725, "step": 3730 }, { "epoch": 0.374, "grad_norm": 23.12073033528547, "learning_rate": 7.206237616802065e-06, "loss": 2.7888, "step": 3740 }, { "epoch": 0.375, "grad_norm": 23.4645449986288, "learning_rate": 7.1916940117488174e-06, "loss": 2.7483, "step": 3750 }, { "epoch": 0.376, "grad_norm": 23.553531891560045, "learning_rate": 7.177127416862479e-06, "loss": 2.7735, "step": 3760 }, { "epoch": 0.377, "grad_norm": 22.5705022507931, "learning_rate": 7.162537984939737e-06, "loss": 2.8026, "step": 3770 }, { "epoch": 0.378, "grad_norm": 21.866200264448686, "learning_rate": 7.1479258690168275e-06, "loss": 2.7079, "step": 3780 }, { "epoch": 0.379, "grad_norm": 22.567049791015194, "learning_rate": 7.133291222367933e-06, "loss": 2.7485, "step": 3790 }, { "epoch": 0.38, "grad_norm": 22.161790491713205, "learning_rate": 7.118634198503573e-06, "loss": 2.7353, "step": 3800 }, { "epoch": 0.381, "grad_norm": 22.159992715546473, "learning_rate": 7.103954951168989e-06, "loss": 2.7487, "step": 3810 }, { "epoch": 0.382, "grad_norm": 23.430323162407856, "learning_rate": 7.089253634342541e-06, "loss": 2.7594, "step": 3820 }, { "epoch": 0.383, "grad_norm": 21.732423534756798, "learning_rate": 7.074530402234089e-06, "loss": 2.8149, "step": 3830 }, { "epoch": 0.384, "grad_norm": 23.12599597095445, "learning_rate": 7.059785409283368e-06, "loss": 2.6969, "step": 3840 }, { "epoch": 0.385, "grad_norm": 20.812947965200824, "learning_rate": 7.045018810158376e-06, "loss": 2.77, "step": 3850 }, { "epoch": 0.386, "grad_norm": 22.16745164190667, "learning_rate": 7.030230759753753e-06, "loss": 2.7487, "step": 3860 }, { "epoch": 0.387, "grad_norm": 21.151010727688792, "learning_rate": 7.015421413189149e-06, "loss": 2.8205, "step": 3870 }, { "epoch": 0.388, "grad_norm": 23.765312330139245, "learning_rate": 7.000590925807601e-06, "loss": 2.8722, "step": 3880 }, { "epoch": 0.389, "grad_norm": 20.34754481359727, "learning_rate": 6.985739453173903e-06, "loss": 2.7729, "step": 3890 }, { "epoch": 0.39, "grad_norm": 22.21616802354248, "learning_rate": 6.97086715107298e-06, "loss": 2.8245, "step": 3900 }, { "epoch": 0.391, "grad_norm": 27.312233478380737, "learning_rate": 6.9559741755082376e-06, "loss": 2.8805, "step": 3910 }, { "epoch": 0.392, "grad_norm": 23.569152803244304, "learning_rate": 6.941060682699947e-06, "loss": 2.7894, "step": 3920 }, { "epoch": 0.393, "grad_norm": 21.73378810991857, "learning_rate": 6.926126829083589e-06, "loss": 2.7384, "step": 3930 }, { "epoch": 0.394, "grad_norm": 24.362282786966546, "learning_rate": 6.911172771308221e-06, "loss": 2.8074, "step": 3940 }, { "epoch": 0.395, "grad_norm": 23.226443853589167, "learning_rate": 6.896198666234833e-06, "loss": 2.838, "step": 3950 }, { "epoch": 0.396, "grad_norm": 25.173366483313533, "learning_rate": 6.881204670934702e-06, "loss": 2.7735, "step": 3960 }, { "epoch": 0.397, "grad_norm": 24.258599445619318, "learning_rate": 6.866190942687744e-06, "loss": 2.7651, "step": 3970 }, { "epoch": 0.398, "grad_norm": 23.856003550036373, "learning_rate": 6.851157638980862e-06, "loss": 2.8627, "step": 3980 }, { "epoch": 0.399, "grad_norm": 22.60502015921121, "learning_rate": 6.836104917506301e-06, "loss": 2.7454, "step": 3990 }, { "epoch": 0.4, "grad_norm": 23.880266300879466, "learning_rate": 6.8210329361599855e-06, "loss": 2.8108, "step": 4000 }, { "epoch": 0.401, "grad_norm": 22.170258973598216, "learning_rate": 6.805941853039868e-06, "loss": 2.7842, "step": 4010 }, { "epoch": 0.402, "grad_norm": 25.97072084505222, "learning_rate": 6.790831826444268e-06, "loss": 2.8059, "step": 4020 }, { "epoch": 0.403, "grad_norm": 23.645861761335333, "learning_rate": 6.775703014870217e-06, "loss": 2.747, "step": 4030 }, { "epoch": 0.404, "grad_norm": 25.142796697644982, "learning_rate": 6.760555577011787e-06, "loss": 2.8086, "step": 4040 }, { "epoch": 0.405, "grad_norm": 24.59679214803799, "learning_rate": 6.745389671758435e-06, "loss": 2.7327, "step": 4050 }, { "epoch": 0.406, "grad_norm": 22.604911335040605, "learning_rate": 6.730205458193331e-06, "loss": 2.6873, "step": 4060 }, { "epoch": 0.407, "grad_norm": 27.021526904460234, "learning_rate": 6.715003095591689e-06, "loss": 2.6904, "step": 4070 }, { "epoch": 0.408, "grad_norm": 23.563325367103346, "learning_rate": 6.6997827434191e-06, "loss": 2.7319, "step": 4080 }, { "epoch": 0.409, "grad_norm": 21.524319223644145, "learning_rate": 6.684544561329855e-06, "loss": 2.7127, "step": 4090 }, { "epoch": 0.41, "grad_norm": 22.76281027951707, "learning_rate": 6.669288709165276e-06, "loss": 2.7847, "step": 4100 }, { "epoch": 0.411, "grad_norm": 22.903464574845078, "learning_rate": 6.65401534695203e-06, "loss": 2.7309, "step": 4110 }, { "epoch": 0.412, "grad_norm": 22.62038664233793, "learning_rate": 6.638724634900462e-06, "loss": 2.7964, "step": 4120 }, { "epoch": 0.413, "grad_norm": 22.91425427112768, "learning_rate": 6.6234167334029035e-06, "loss": 2.7224, "step": 4130 }, { "epoch": 0.414, "grad_norm": 21.424216775890002, "learning_rate": 6.608091803032e-06, "loss": 2.7015, "step": 4140 }, { "epoch": 0.415, "grad_norm": 23.316858466161655, "learning_rate": 6.592750004539019e-06, "loss": 2.6925, "step": 4150 }, { "epoch": 0.416, "grad_norm": 24.053777556462254, "learning_rate": 6.577391498852168e-06, "loss": 2.6725, "step": 4160 }, { "epoch": 0.417, "grad_norm": 24.198966863080148, "learning_rate": 6.5620164470749015e-06, "loss": 2.7306, "step": 4170 }, { "epoch": 0.418, "grad_norm": 25.504201004678368, "learning_rate": 6.546625010484239e-06, "loss": 2.7101, "step": 4180 }, { "epoch": 0.419, "grad_norm": 23.221886590032188, "learning_rate": 6.531217350529067e-06, "loss": 2.7561, "step": 4190 }, { "epoch": 0.42, "grad_norm": 23.005146336040827, "learning_rate": 6.515793628828446e-06, "loss": 2.6854, "step": 4200 }, { "epoch": 0.421, "grad_norm": 22.55646260424012, "learning_rate": 6.500354007169921e-06, "loss": 2.6104, "step": 4210 }, { "epoch": 0.422, "grad_norm": 23.089753826069582, "learning_rate": 6.484898647507816e-06, "loss": 2.7905, "step": 4220 }, { "epoch": 0.423, "grad_norm": 22.060617249308248, "learning_rate": 6.4694277119615405e-06, "loss": 2.7317, "step": 4230 }, { "epoch": 0.424, "grad_norm": 22.449382027099922, "learning_rate": 6.453941362813888e-06, "loss": 2.7411, "step": 4240 }, { "epoch": 0.425, "grad_norm": 22.792073785709025, "learning_rate": 6.438439762509332e-06, "loss": 2.6544, "step": 4250 }, { "epoch": 0.426, "grad_norm": 20.983028397728273, "learning_rate": 6.422923073652325e-06, "loss": 2.7134, "step": 4260 }, { "epoch": 0.427, "grad_norm": 23.20418795677239, "learning_rate": 6.4073914590055916e-06, "loss": 2.6774, "step": 4270 }, { "epoch": 0.428, "grad_norm": 24.351898534600377, "learning_rate": 6.391845081488419e-06, "loss": 2.7382, "step": 4280 }, { "epoch": 0.429, "grad_norm": 22.231680384417274, "learning_rate": 6.376284104174952e-06, "loss": 2.7824, "step": 4290 }, { "epoch": 0.43, "grad_norm": 23.682280110148493, "learning_rate": 6.360708690292479e-06, "loss": 2.6709, "step": 4300 }, { "epoch": 0.431, "grad_norm": 22.460135300591457, "learning_rate": 6.34511900321972e-06, "loss": 2.7371, "step": 4310 }, { "epoch": 0.432, "grad_norm": 19.541508882195817, "learning_rate": 6.329515206485118e-06, "loss": 2.6906, "step": 4320 }, { "epoch": 0.433, "grad_norm": 21.30285746459095, "learning_rate": 6.313897463765115e-06, "loss": 2.7562, "step": 4330 }, { "epoch": 0.434, "grad_norm": 22.358449810513083, "learning_rate": 6.298265938882443e-06, "loss": 2.6672, "step": 4340 }, { "epoch": 0.435, "grad_norm": 24.017120859434723, "learning_rate": 6.282620795804399e-06, "loss": 2.7771, "step": 4350 }, { "epoch": 0.436, "grad_norm": 22.870374001102483, "learning_rate": 6.266962198641134e-06, "loss": 2.7262, "step": 4360 }, { "epoch": 0.437, "grad_norm": 22.163916420629548, "learning_rate": 6.251290311643923e-06, "loss": 2.645, "step": 4370 }, { "epoch": 0.438, "grad_norm": 22.669015981263485, "learning_rate": 6.235605299203441e-06, "loss": 2.7119, "step": 4380 }, { "epoch": 0.439, "grad_norm": 23.390258197492447, "learning_rate": 6.21990732584805e-06, "loss": 2.6374, "step": 4390 }, { "epoch": 0.44, "grad_norm": 23.895631435935133, "learning_rate": 6.204196556242061e-06, "loss": 2.5934, "step": 4400 }, { "epoch": 0.441, "grad_norm": 22.889469767682222, "learning_rate": 6.188473155184011e-06, "loss": 2.6668, "step": 4410 }, { "epoch": 0.442, "grad_norm": 22.410255690903092, "learning_rate": 6.172737287604942e-06, "loss": 2.7256, "step": 4420 }, { "epoch": 0.443, "grad_norm": 21.320466243105425, "learning_rate": 6.156989118566654e-06, "loss": 2.62, "step": 4430 }, { "epoch": 0.444, "grad_norm": 22.782556480105853, "learning_rate": 6.14122881325999e-06, "loss": 2.5637, "step": 4440 }, { "epoch": 0.445, "grad_norm": 22.49177584062071, "learning_rate": 6.125456537003095e-06, "loss": 2.6251, "step": 4450 }, { "epoch": 0.446, "grad_norm": 21.372244405828038, "learning_rate": 6.109672455239684e-06, "loss": 2.7102, "step": 4460 }, { "epoch": 0.447, "grad_norm": 27.450721787848217, "learning_rate": 6.0938767335373064e-06, "loss": 2.7011, "step": 4470 }, { "epoch": 0.448, "grad_norm": 24.133527593850953, "learning_rate": 6.078069537585606e-06, "loss": 2.7473, "step": 4480 }, { "epoch": 0.449, "grad_norm": 21.43754960843299, "learning_rate": 6.062251033194591e-06, "loss": 2.6581, "step": 4490 }, { "epoch": 0.45, "grad_norm": 23.064579033226895, "learning_rate": 6.046421386292887e-06, "loss": 2.696, "step": 4500 }, { "epoch": 0.451, "grad_norm": 24.7272336423561, "learning_rate": 6.030580762925996e-06, "loss": 2.5792, "step": 4510 }, { "epoch": 0.452, "grad_norm": 22.07287300706964, "learning_rate": 6.0147293292545675e-06, "loss": 2.663, "step": 4520 }, { "epoch": 0.453, "grad_norm": 23.750253369210135, "learning_rate": 5.998867251552635e-06, "loss": 2.6106, "step": 4530 }, { "epoch": 0.454, "grad_norm": 22.743885278765717, "learning_rate": 5.98299469620589e-06, "loss": 2.648, "step": 4540 }, { "epoch": 0.455, "grad_norm": 22.586552946089537, "learning_rate": 5.9671118297099276e-06, "loss": 2.6789, "step": 4550 }, { "epoch": 0.456, "grad_norm": 22.670162341519074, "learning_rate": 5.9512188186685e-06, "loss": 2.6389, "step": 4560 }, { "epoch": 0.457, "grad_norm": 23.835709991905297, "learning_rate": 5.935315829791774e-06, "loss": 2.6504, "step": 4570 }, { "epoch": 0.458, "grad_norm": 22.69994060294616, "learning_rate": 5.9194030298945755e-06, "loss": 2.647, "step": 4580 }, { "epoch": 0.459, "grad_norm": 24.250720504170367, "learning_rate": 5.9034805858946486e-06, "loss": 2.6325, "step": 4590 }, { "epoch": 0.46, "grad_norm": 22.937346415520437, "learning_rate": 5.887548664810896e-06, "loss": 2.687, "step": 4600 }, { "epoch": 0.461, "grad_norm": 21.095665547987682, "learning_rate": 5.87160743376163e-06, "loss": 2.6402, "step": 4610 }, { "epoch": 0.462, "grad_norm": 23.2617807209512, "learning_rate": 5.855657059962822e-06, "loss": 2.5852, "step": 4620 }, { "epoch": 0.463, "grad_norm": 23.999518189424556, "learning_rate": 5.839697710726344e-06, "loss": 2.6363, "step": 4630 }, { "epoch": 0.464, "grad_norm": 23.346858751488963, "learning_rate": 5.823729553458219e-06, "loss": 2.6574, "step": 4640 }, { "epoch": 0.465, "grad_norm": 25.52605053929966, "learning_rate": 5.807752755656862e-06, "loss": 2.6056, "step": 4650 }, { "epoch": 0.466, "grad_norm": 22.694045397767084, "learning_rate": 5.791767484911317e-06, "loss": 2.6164, "step": 4660 }, { "epoch": 0.467, "grad_norm": 23.563056237853544, "learning_rate": 5.775773908899515e-06, "loss": 2.6201, "step": 4670 }, { "epoch": 0.468, "grad_norm": 25.047596778721836, "learning_rate": 5.759772195386498e-06, "loss": 2.6398, "step": 4680 }, { "epoch": 0.469, "grad_norm": 23.081258846168645, "learning_rate": 5.74376251222267e-06, "loss": 2.5974, "step": 4690 }, { "epoch": 0.47, "grad_norm": 21.859782203352324, "learning_rate": 5.727745027342031e-06, "loss": 2.6427, "step": 4700 }, { "epoch": 0.471, "grad_norm": 23.320743968315114, "learning_rate": 5.711719908760418e-06, "loss": 2.6305, "step": 4710 }, { "epoch": 0.472, "grad_norm": 21.867623347325473, "learning_rate": 5.695687324573741e-06, "loss": 2.5915, "step": 4720 }, { "epoch": 0.473, "grad_norm": 24.942219065962178, "learning_rate": 5.6796474429562245e-06, "loss": 2.6812, "step": 4730 }, { "epoch": 0.474, "grad_norm": 24.064171165336997, "learning_rate": 5.663600432158635e-06, "loss": 2.5497, "step": 4740 }, { "epoch": 0.475, "grad_norm": 21.791099594213005, "learning_rate": 5.64754646050652e-06, "loss": 2.5604, "step": 4750 }, { "epoch": 0.476, "grad_norm": 23.00628257037668, "learning_rate": 5.63148569639845e-06, "loss": 2.7038, "step": 4760 }, { "epoch": 0.477, "grad_norm": 21.937157565097678, "learning_rate": 5.615418308304239e-06, "loss": 2.613, "step": 4770 }, { "epoch": 0.478, "grad_norm": 26.53009219035451, "learning_rate": 5.5993444647631846e-06, "loss": 2.5808, "step": 4780 }, { "epoch": 0.479, "grad_norm": 23.881103735777327, "learning_rate": 5.5832643343823e-06, "loss": 2.5633, "step": 4790 }, { "epoch": 0.48, "grad_norm": 23.575388360153507, "learning_rate": 5.5671780858345424e-06, "loss": 2.6206, "step": 4800 }, { "epoch": 0.481, "grad_norm": 28.737044163253415, "learning_rate": 5.5510858878570485e-06, "loss": 2.5753, "step": 4810 }, { "epoch": 0.482, "grad_norm": 24.21379156588361, "learning_rate": 5.534987909249362e-06, "loss": 2.6488, "step": 4820 }, { "epoch": 0.483, "grad_norm": 22.55782891596325, "learning_rate": 5.518884318871658e-06, "loss": 2.6269, "step": 4830 }, { "epoch": 0.484, "grad_norm": 23.71018818404809, "learning_rate": 5.502775285642982e-06, "loss": 2.535, "step": 4840 }, { "epoch": 0.485, "grad_norm": 24.025261694952377, "learning_rate": 5.486660978539468e-06, "loss": 2.6328, "step": 4850 }, { "epoch": 0.486, "grad_norm": 22.37649926429674, "learning_rate": 5.470541566592573e-06, "loss": 2.5307, "step": 4860 }, { "epoch": 0.487, "grad_norm": 25.10295490842964, "learning_rate": 5.4544172188873e-06, "loss": 2.5795, "step": 4870 }, { "epoch": 0.488, "grad_norm": 22.410849307805687, "learning_rate": 5.4382881045604285e-06, "loss": 2.5982, "step": 4880 }, { "epoch": 0.489, "grad_norm": 21.834141185442974, "learning_rate": 5.422154392798735e-06, "loss": 2.5662, "step": 4890 }, { "epoch": 0.49, "grad_norm": 23.83709193096916, "learning_rate": 5.40601625283722e-06, "loss": 2.5812, "step": 4900 }, { "epoch": 0.491, "grad_norm": 24.71986870161946, "learning_rate": 5.389873853957336e-06, "loss": 2.5802, "step": 4910 }, { "epoch": 0.492, "grad_norm": 23.409507237557516, "learning_rate": 5.373727365485211e-06, "loss": 2.6007, "step": 4920 }, { "epoch": 0.493, "grad_norm": 24.4222897630817, "learning_rate": 5.357576956789865e-06, "loss": 2.6069, "step": 4930 }, { "epoch": 0.494, "grad_norm": 22.937279375469757, "learning_rate": 5.341422797281446e-06, "loss": 2.5652, "step": 4940 }, { "epoch": 0.495, "grad_norm": 26.049573965216798, "learning_rate": 5.325265056409442e-06, "loss": 2.5718, "step": 4950 }, { "epoch": 0.496, "grad_norm": 23.69623725013249, "learning_rate": 5.3091039036609085e-06, "loss": 2.6196, "step": 4960 }, { "epoch": 0.497, "grad_norm": 23.91538584283533, "learning_rate": 5.29293950855869e-06, "loss": 2.68, "step": 4970 }, { "epoch": 0.498, "grad_norm": 24.452221891881866, "learning_rate": 5.276772040659645e-06, "loss": 2.5604, "step": 4980 }, { "epoch": 0.499, "grad_norm": 23.824868358077204, "learning_rate": 5.260601669552859e-06, "loss": 2.5685, "step": 4990 }, { "epoch": 0.5, "grad_norm": 23.824569884787365, "learning_rate": 5.244428564857874e-06, "loss": 2.5118, "step": 5000 }, { "epoch": 0.501, "grad_norm": 23.161386117275736, "learning_rate": 5.228252896222906e-06, "loss": 2.5627, "step": 5010 }, { "epoch": 0.502, "grad_norm": 24.15705704620734, "learning_rate": 5.212074833323062e-06, "loss": 2.5688, "step": 5020 }, { "epoch": 0.503, "grad_norm": 22.38183932188608, "learning_rate": 5.195894545858571e-06, "loss": 2.6117, "step": 5030 }, { "epoch": 0.504, "grad_norm": 25.94119507519535, "learning_rate": 5.179712203552988e-06, "loss": 2.6507, "step": 5040 }, { "epoch": 0.505, "grad_norm": 24.15538457203914, "learning_rate": 5.163527976151428e-06, "loss": 2.6148, "step": 5050 }, { "epoch": 0.506, "grad_norm": 22.499141980048613, "learning_rate": 5.147342033418778e-06, "loss": 2.6151, "step": 5060 }, { "epoch": 0.507, "grad_norm": 23.69450370611486, "learning_rate": 5.1311545451379205e-06, "loss": 2.6802, "step": 5070 }, { "epoch": 0.508, "grad_norm": 21.285344716324524, "learning_rate": 5.1149656811079455e-06, "loss": 2.6103, "step": 5080 }, { "epoch": 0.509, "grad_norm": 24.293138847419378, "learning_rate": 5.098775611142374e-06, "loss": 2.5966, "step": 5090 }, { "epoch": 0.51, "grad_norm": 23.612520580973456, "learning_rate": 5.082584505067384e-06, "loss": 2.5463, "step": 5100 }, { "epoch": 0.511, "grad_norm": 23.244139713394286, "learning_rate": 5.0663925327200135e-06, "loss": 2.6011, "step": 5110 }, { "epoch": 0.512, "grad_norm": 22.110339523144702, "learning_rate": 5.05019986394639e-06, "loss": 2.538, "step": 5120 }, { "epoch": 0.513, "grad_norm": 23.14195116452269, "learning_rate": 5.034006668599949e-06, "loss": 2.5723, "step": 5130 }, { "epoch": 0.514, "grad_norm": 24.193123143329505, "learning_rate": 5.017813116539646e-06, "loss": 2.515, "step": 5140 }, { "epoch": 0.515, "grad_norm": 27.26184416186975, "learning_rate": 5.00161937762818e-06, "loss": 2.6846, "step": 5150 }, { "epoch": 0.516, "grad_norm": 26.28061767814624, "learning_rate": 4.985425621730208e-06, "loss": 2.5746, "step": 5160 }, { "epoch": 0.517, "grad_norm": 22.791660570243288, "learning_rate": 4.969232018710569e-06, "loss": 2.557, "step": 5170 }, { "epoch": 0.518, "grad_norm": 24.194281363764066, "learning_rate": 4.953038738432493e-06, "loss": 2.5847, "step": 5180 }, { "epoch": 0.519, "grad_norm": 22.01820129887373, "learning_rate": 4.936845950755829e-06, "loss": 2.4581, "step": 5190 }, { "epoch": 0.52, "grad_norm": 26.161317207525276, "learning_rate": 4.920653825535258e-06, "loss": 2.4735, "step": 5200 }, { "epoch": 0.521, "grad_norm": 22.557490606188164, "learning_rate": 4.904462532618511e-06, "loss": 2.6211, "step": 5210 }, { "epoch": 0.522, "grad_norm": 22.20124606486481, "learning_rate": 4.888272241844591e-06, "loss": 2.5358, "step": 5220 }, { "epoch": 0.523, "grad_norm": 24.187520950083194, "learning_rate": 4.872083123041983e-06, "loss": 2.5981, "step": 5230 }, { "epoch": 0.524, "grad_norm": 23.0050117991877, "learning_rate": 4.855895346026887e-06, "loss": 2.5931, "step": 5240 }, { "epoch": 0.525, "grad_norm": 21.505669665336715, "learning_rate": 4.839709080601419e-06, "loss": 2.5626, "step": 5250 }, { "epoch": 0.526, "grad_norm": 25.077109737936013, "learning_rate": 4.82352449655185e-06, "loss": 2.5676, "step": 5260 }, { "epoch": 0.527, "grad_norm": 22.235029219202698, "learning_rate": 4.807341763646804e-06, "loss": 2.4843, "step": 5270 }, { "epoch": 0.528, "grad_norm": 22.998195217341973, "learning_rate": 4.791161051635494e-06, "loss": 2.5424, "step": 5280 }, { "epoch": 0.529, "grad_norm": 24.405172186461435, "learning_rate": 4.774982530245931e-06, "loss": 2.5669, "step": 5290 }, { "epoch": 0.53, "grad_norm": 24.11476321578105, "learning_rate": 4.7588063691831495e-06, "loss": 2.5367, "step": 5300 }, { "epoch": 0.531, "grad_norm": 22.75488756484615, "learning_rate": 4.742632738127424e-06, "loss": 2.4767, "step": 5310 }, { "epoch": 0.532, "grad_norm": 22.14081201035399, "learning_rate": 4.726461806732492e-06, "loss": 2.5896, "step": 5320 }, { "epoch": 0.533, "grad_norm": 23.078124584269265, "learning_rate": 4.710293744623772e-06, "loss": 2.5732, "step": 5330 }, { "epoch": 0.534, "grad_norm": 23.988187229620564, "learning_rate": 4.694128721396583e-06, "loss": 2.5839, "step": 5340 }, { "epoch": 0.535, "grad_norm": 22.69399474229365, "learning_rate": 4.677966906614372e-06, "loss": 2.5532, "step": 5350 }, { "epoch": 0.536, "grad_norm": 22.209447746400876, "learning_rate": 4.661808469806925e-06, "loss": 2.5533, "step": 5360 }, { "epoch": 0.537, "grad_norm": 24.219441088645986, "learning_rate": 4.645653580468602e-06, "loss": 2.5507, "step": 5370 }, { "epoch": 0.538, "grad_norm": 24.39450465555395, "learning_rate": 4.629502408056546e-06, "loss": 2.4993, "step": 5380 }, { "epoch": 0.539, "grad_norm": 22.39543987758126, "learning_rate": 4.6133551219889125e-06, "loss": 2.5656, "step": 5390 }, { "epoch": 0.54, "grad_norm": 23.808893657382605, "learning_rate": 4.597211891643093e-06, "loss": 2.5149, "step": 5400 }, { "epoch": 0.541, "grad_norm": 24.110909939686525, "learning_rate": 4.581072886353934e-06, "loss": 2.5428, "step": 5410 }, { "epoch": 0.542, "grad_norm": 24.062801137984263, "learning_rate": 4.564938275411963e-06, "loss": 2.5677, "step": 5420 }, { "epoch": 0.543, "grad_norm": 23.60119202442823, "learning_rate": 4.548808228061615e-06, "loss": 2.6112, "step": 5430 }, { "epoch": 0.544, "grad_norm": 22.274021868313095, "learning_rate": 4.532682913499456e-06, "loss": 2.5543, "step": 5440 }, { "epoch": 0.545, "grad_norm": 22.595867697797207, "learning_rate": 4.516562500872403e-06, "loss": 2.486, "step": 5450 }, { "epoch": 0.546, "grad_norm": 24.16682103796466, "learning_rate": 4.500447159275957e-06, "loss": 2.5214, "step": 5460 }, { "epoch": 0.547, "grad_norm": 26.88842137920154, "learning_rate": 4.484337057752427e-06, "loss": 2.538, "step": 5470 }, { "epoch": 0.548, "grad_norm": 24.656395682617887, "learning_rate": 4.468232365289153e-06, "loss": 2.5728, "step": 5480 }, { "epoch": 0.549, "grad_norm": 23.14117797302742, "learning_rate": 4.452133250816739e-06, "loss": 2.5608, "step": 5490 }, { "epoch": 0.55, "grad_norm": 24.67193536939508, "learning_rate": 4.436039883207277e-06, "loss": 2.5801, "step": 5500 }, { "epoch": 0.551, "grad_norm": 23.58592040589172, "learning_rate": 4.419952431272579e-06, "loss": 2.4841, "step": 5510 }, { "epoch": 0.552, "grad_norm": 25.78539357913926, "learning_rate": 4.403871063762404e-06, "loss": 2.4872, "step": 5520 }, { "epoch": 0.553, "grad_norm": 22.180451935723287, "learning_rate": 4.387795949362687e-06, "loss": 2.4889, "step": 5530 }, { "epoch": 0.554, "grad_norm": 24.09610341753377, "learning_rate": 4.371727256693771e-06, "loss": 2.5527, "step": 5540 }, { "epoch": 0.555, "grad_norm": 23.753547846081307, "learning_rate": 4.355665154308637e-06, "loss": 2.5091, "step": 5550 }, { "epoch": 0.556, "grad_norm": 24.943771860199742, "learning_rate": 4.339609810691141e-06, "loss": 2.5541, "step": 5560 }, { "epoch": 0.557, "grad_norm": 22.326972803344738, "learning_rate": 4.323561394254241e-06, "loss": 2.5974, "step": 5570 }, { "epoch": 0.558, "grad_norm": 25.310115099732784, "learning_rate": 4.30752007333823e-06, "loss": 2.6046, "step": 5580 }, { "epoch": 0.559, "grad_norm": 22.62402860073653, "learning_rate": 4.291486016208973e-06, "loss": 2.5099, "step": 5590 }, { "epoch": 0.56, "grad_norm": 24.67831072157042, "learning_rate": 4.275459391056142e-06, "loss": 2.6291, "step": 5600 }, { "epoch": 0.561, "grad_norm": 375.6993525592239, "learning_rate": 4.259440365991454e-06, "loss": 2.5269, "step": 5610 }, { "epoch": 0.562, "grad_norm": 24.28568264046934, "learning_rate": 4.243429109046899e-06, "loss": 2.6224, "step": 5620 }, { "epoch": 0.563, "grad_norm": 22.91072957810827, "learning_rate": 4.227425788172987e-06, "loss": 2.6187, "step": 5630 }, { "epoch": 0.564, "grad_norm": 24.05888329176886, "learning_rate": 4.211430571236982e-06, "loss": 2.6254, "step": 5640 }, { "epoch": 0.565, "grad_norm": 24.53267546357263, "learning_rate": 4.1954436260211396e-06, "loss": 2.542, "step": 5650 }, { "epoch": 0.566, "grad_norm": 23.289902006444073, "learning_rate": 4.179465120220951e-06, "loss": 2.5592, "step": 5660 }, { "epoch": 0.567, "grad_norm": 25.4553761281628, "learning_rate": 4.16349522144338e-06, "loss": 2.6529, "step": 5670 }, { "epoch": 0.568, "grad_norm": 22.856385173542364, "learning_rate": 4.147534097205109e-06, "loss": 2.5321, "step": 5680 }, { "epoch": 0.569, "grad_norm": 25.45543592141224, "learning_rate": 4.131581914930779e-06, "loss": 2.6453, "step": 5690 }, { "epoch": 0.57, "grad_norm": 23.555889605377104, "learning_rate": 4.115638841951233e-06, "loss": 2.536, "step": 5700 }, { "epoch": 0.571, "grad_norm": 24.452957095355384, "learning_rate": 4.0997050455017614e-06, "loss": 2.4935, "step": 5710 }, { "epoch": 0.572, "grad_norm": 24.689543320783432, "learning_rate": 4.083780692720352e-06, "loss": 2.6099, "step": 5720 }, { "epoch": 0.573, "grad_norm": 25.490951368379363, "learning_rate": 4.067865950645927e-06, "loss": 2.6015, "step": 5730 }, { "epoch": 0.574, "grad_norm": 22.14617692557331, "learning_rate": 4.0519609862166025e-06, "loss": 2.5188, "step": 5740 }, { "epoch": 0.575, "grad_norm": 23.06684049197533, "learning_rate": 4.036065966267927e-06, "loss": 2.5459, "step": 5750 }, { "epoch": 0.576, "grad_norm": 26.74202000742249, "learning_rate": 4.020181057531141e-06, "loss": 2.5441, "step": 5760 }, { "epoch": 0.577, "grad_norm": 23.846141616834306, "learning_rate": 4.004306426631418e-06, "loss": 2.5333, "step": 5770 }, { "epoch": 0.578, "grad_norm": 24.047080119079514, "learning_rate": 3.988442240086127e-06, "loss": 2.5821, "step": 5780 }, { "epoch": 0.579, "grad_norm": 25.14274613292411, "learning_rate": 3.972588664303076e-06, "loss": 2.6028, "step": 5790 }, { "epoch": 0.58, "grad_norm": 23.181592358517253, "learning_rate": 3.956745865578773e-06, "loss": 2.6589, "step": 5800 }, { "epoch": 0.581, "grad_norm": 23.370877629035377, "learning_rate": 3.940914010096681e-06, "loss": 2.5681, "step": 5810 }, { "epoch": 0.582, "grad_norm": 23.646912789846475, "learning_rate": 3.925093263925472e-06, "loss": 2.5682, "step": 5820 }, { "epoch": 0.583, "grad_norm": 23.651020584467055, "learning_rate": 3.909283793017289e-06, "loss": 2.5612, "step": 5830 }, { "epoch": 0.584, "grad_norm": 24.092916939569776, "learning_rate": 3.893485763205998e-06, "loss": 2.4995, "step": 5840 }, { "epoch": 0.585, "grad_norm": 25.607134976571647, "learning_rate": 3.877699340205455e-06, "loss": 2.4823, "step": 5850 }, { "epoch": 0.586, "grad_norm": 23.352953107957234, "learning_rate": 3.861924689607773e-06, "loss": 2.4791, "step": 5860 }, { "epoch": 0.587, "grad_norm": 25.41281282289825, "learning_rate": 3.846161976881568e-06, "loss": 2.5612, "step": 5870 }, { "epoch": 0.588, "grad_norm": 24.143360977028642, "learning_rate": 3.830411367370239e-06, "loss": 2.5415, "step": 5880 }, { "epoch": 0.589, "grad_norm": 23.076011392716836, "learning_rate": 3.8146730262902277e-06, "loss": 2.5226, "step": 5890 }, { "epoch": 0.59, "grad_norm": 24.738282780599096, "learning_rate": 3.798947118729282e-06, "loss": 2.4536, "step": 5900 }, { "epoch": 0.591, "grad_norm": 24.753991448856453, "learning_rate": 3.783233809644732e-06, "loss": 2.4868, "step": 5910 }, { "epoch": 0.592, "grad_norm": 20.876244609298247, "learning_rate": 3.767533263861756e-06, "loss": 2.4731, "step": 5920 }, { "epoch": 0.593, "grad_norm": 22.506849507929147, "learning_rate": 3.751845646071649e-06, "loss": 2.5043, "step": 5930 }, { "epoch": 0.594, "grad_norm": 23.703751096125526, "learning_rate": 3.7361711208300984e-06, "loss": 2.4899, "step": 5940 }, { "epoch": 0.595, "grad_norm": 22.47517721892105, "learning_rate": 3.720509852555456e-06, "loss": 2.4825, "step": 5950 }, { "epoch": 0.596, "grad_norm": 24.206163833569313, "learning_rate": 3.7048620055270167e-06, "loss": 2.4673, "step": 5960 }, { "epoch": 0.597, "grad_norm": 24.30922963514069, "learning_rate": 3.6892277438832903e-06, "loss": 2.5032, "step": 5970 }, { "epoch": 0.598, "grad_norm": 25.055558457418407, "learning_rate": 3.673607231620283e-06, "loss": 2.5146, "step": 5980 }, { "epoch": 0.599, "grad_norm": 23.62573569501835, "learning_rate": 3.6580006325897786e-06, "loss": 2.4632, "step": 5990 }, { "epoch": 0.6, "grad_norm": 25.393092326630224, "learning_rate": 3.6424081104976162e-06, "loss": 2.4686, "step": 6000 }, { "epoch": 0.601, "grad_norm": 26.31085054674852, "learning_rate": 3.6268298289019723e-06, "loss": 2.4759, "step": 6010 }, { "epoch": 0.602, "grad_norm": 25.043785809333475, "learning_rate": 3.6112659512116497e-06, "loss": 2.5168, "step": 6020 }, { "epoch": 0.603, "grad_norm": 24.83442212579718, "learning_rate": 3.5957166406843623e-06, "loss": 2.493, "step": 6030 }, { "epoch": 0.604, "grad_norm": 23.79579044549792, "learning_rate": 3.580182060425017e-06, "loss": 2.4691, "step": 6040 }, { "epoch": 0.605, "grad_norm": 22.81412577399071, "learning_rate": 3.5646623733840135e-06, "loss": 2.4453, "step": 6050 }, { "epoch": 0.606, "grad_norm": 23.940111036353677, "learning_rate": 3.5491577423555222e-06, "loss": 2.5141, "step": 6060 }, { "epoch": 0.607, "grad_norm": 23.278910766411826, "learning_rate": 3.533668329975788e-06, "loss": 2.4489, "step": 6070 }, { "epoch": 0.608, "grad_norm": 24.18680454985014, "learning_rate": 3.5181942987214145e-06, "loss": 2.4289, "step": 6080 }, { "epoch": 0.609, "grad_norm": 23.02575005126682, "learning_rate": 3.5027358109076704e-06, "loss": 2.5228, "step": 6090 }, { "epoch": 0.61, "grad_norm": 24.31643748980866, "learning_rate": 3.487293028686776e-06, "loss": 2.4567, "step": 6100 }, { "epoch": 0.611, "grad_norm": 23.426436332241842, "learning_rate": 3.4718661140462097e-06, "loss": 2.4609, "step": 6110 }, { "epoch": 0.612, "grad_norm": 24.087987491554916, "learning_rate": 3.4564552288070073e-06, "loss": 2.5466, "step": 6120 }, { "epoch": 0.613, "grad_norm": 24.40604651426191, "learning_rate": 3.4410605346220616e-06, "loss": 2.5099, "step": 6130 }, { "epoch": 0.614, "grad_norm": 24.0600808868819, "learning_rate": 3.4256821929744304e-06, "loss": 2.4797, "step": 6140 }, { "epoch": 0.615, "grad_norm": 23.57955835990851, "learning_rate": 3.4103203651756406e-06, "loss": 2.5222, "step": 6150 }, { "epoch": 0.616, "grad_norm": 25.08017764264328, "learning_rate": 3.3949752123639946e-06, "loss": 2.487, "step": 6160 }, { "epoch": 0.617, "grad_norm": 26.01587829061594, "learning_rate": 3.3796468955028883e-06, "loss": 2.4756, "step": 6170 }, { "epoch": 0.618, "grad_norm": 23.412523088744965, "learning_rate": 3.364335575379111e-06, "loss": 2.5229, "step": 6180 }, { "epoch": 0.619, "grad_norm": 23.81404834642887, "learning_rate": 3.3490414126011638e-06, "loss": 2.4768, "step": 6190 }, { "epoch": 0.62, "grad_norm": 24.947210845220088, "learning_rate": 3.333764567597579e-06, "loss": 2.4874, "step": 6200 }, { "epoch": 0.621, "grad_norm": 25.545329442783355, "learning_rate": 3.3185052006152296e-06, "loss": 2.645, "step": 6210 }, { "epoch": 0.622, "grad_norm": 23.908367133982093, "learning_rate": 3.303263471717657e-06, "loss": 2.5261, "step": 6220 }, { "epoch": 0.623, "grad_norm": 23.63178901892679, "learning_rate": 3.2880395407833856e-06, "loss": 2.5407, "step": 6230 }, { "epoch": 0.624, "grad_norm": 25.536655114146072, "learning_rate": 3.2728335675042444e-06, "loss": 2.556, "step": 6240 }, { "epoch": 0.625, "grad_norm": 25.919259022727154, "learning_rate": 3.257645711383699e-06, "loss": 2.5913, "step": 6250 }, { "epoch": 0.626, "grad_norm": 24.455472268236417, "learning_rate": 3.242476131735174e-06, "loss": 2.5115, "step": 6260 }, { "epoch": 0.627, "grad_norm": 23.40444449172011, "learning_rate": 3.22732498768038e-06, "loss": 2.555, "step": 6270 }, { "epoch": 0.628, "grad_norm": 22.30901739957592, "learning_rate": 3.21219243814765e-06, "loss": 2.4957, "step": 6280 }, { "epoch": 0.629, "grad_norm": 22.91903765098284, "learning_rate": 3.197078641870268e-06, "loss": 2.4647, "step": 6290 }, { "epoch": 0.63, "grad_norm": 22.449520840833635, "learning_rate": 3.1819837573848054e-06, "loss": 2.4711, "step": 6300 }, { "epoch": 0.631, "grad_norm": 23.029326601328876, "learning_rate": 3.166907943029458e-06, "loss": 2.4051, "step": 6310 }, { "epoch": 0.632, "grad_norm": 27.005343322029155, "learning_rate": 3.1518513569423834e-06, "loss": 2.5069, "step": 6320 }, { "epoch": 0.633, "grad_norm": 24.099137528741753, "learning_rate": 3.136814157060046e-06, "loss": 2.4731, "step": 6330 }, { "epoch": 0.634, "grad_norm": 25.523043364405, "learning_rate": 3.121796501115556e-06, "loss": 2.5227, "step": 6340 }, { "epoch": 0.635, "grad_norm": 21.66043931518163, "learning_rate": 3.106798546637019e-06, "loss": 2.431, "step": 6350 }, { "epoch": 0.636, "grad_norm": 24.531748932841516, "learning_rate": 3.09182045094588e-06, "loss": 2.3906, "step": 6360 }, { "epoch": 0.637, "grad_norm": 22.86113062181456, "learning_rate": 3.0768623711552757e-06, "loss": 2.479, "step": 6370 }, { "epoch": 0.638, "grad_norm": 24.829979937034118, "learning_rate": 3.0619244641683856e-06, "loss": 2.443, "step": 6380 }, { "epoch": 0.639, "grad_norm": 24.134760225374325, "learning_rate": 3.0470068866767844e-06, "loss": 2.4159, "step": 6390 }, { "epoch": 0.64, "grad_norm": 25.408584527839956, "learning_rate": 3.032109795158802e-06, "loss": 2.3555, "step": 6400 }, { "epoch": 0.641, "grad_norm": 25.837865061565417, "learning_rate": 3.0172333458778783e-06, "loss": 2.3964, "step": 6410 }, { "epoch": 0.642, "grad_norm": 25.772932058522212, "learning_rate": 3.002377694880927e-06, "loss": 2.4641, "step": 6420 }, { "epoch": 0.643, "grad_norm": 26.894924706153088, "learning_rate": 2.987542997996698e-06, "loss": 2.443, "step": 6430 }, { "epoch": 0.644, "grad_norm": 23.47713860818913, "learning_rate": 2.972729410834142e-06, "loss": 2.3693, "step": 6440 }, { "epoch": 0.645, "grad_norm": 25.22869223104333, "learning_rate": 2.957937088780779e-06, "loss": 2.3972, "step": 6450 }, { "epoch": 0.646, "grad_norm": 25.11105213723829, "learning_rate": 2.943166187001064e-06, "loss": 2.4288, "step": 6460 }, { "epoch": 0.647, "grad_norm": 25.833206194468033, "learning_rate": 2.9284168604347735e-06, "loss": 2.3966, "step": 6470 }, { "epoch": 0.648, "grad_norm": 25.031998369439282, "learning_rate": 2.913689263795361e-06, "loss": 2.4732, "step": 6480 }, { "epoch": 0.649, "grad_norm": 23.18307563556915, "learning_rate": 2.898983551568344e-06, "loss": 2.3751, "step": 6490 }, { "epoch": 0.65, "grad_norm": 24.282231305821284, "learning_rate": 2.88429987800969e-06, "loss": 2.4621, "step": 6500 }, { "epoch": 0.651, "grad_norm": 24.545237201673817, "learning_rate": 2.869638397144181e-06, "loss": 2.5018, "step": 6510 }, { "epoch": 0.652, "grad_norm": 25.296351311035608, "learning_rate": 2.8549992627638186e-06, "loss": 2.3955, "step": 6520 }, { "epoch": 0.653, "grad_norm": 25.57719033937348, "learning_rate": 2.8403826284261906e-06, "loss": 2.4357, "step": 6530 }, { "epoch": 0.654, "grad_norm": 25.630884177705095, "learning_rate": 2.8257886474528794e-06, "loss": 2.4297, "step": 6540 }, { "epoch": 0.655, "grad_norm": 26.0560417779101, "learning_rate": 2.8112174729278354e-06, "loss": 2.4473, "step": 6550 }, { "epoch": 0.656, "grad_norm": 25.31079668239842, "learning_rate": 2.7966692576957868e-06, "loss": 2.4355, "step": 6560 }, { "epoch": 0.657, "grad_norm": 24.036005123462697, "learning_rate": 2.7821441543606275e-06, "loss": 2.5058, "step": 6570 }, { "epoch": 0.658, "grad_norm": 23.192462777498655, "learning_rate": 2.767642315283817e-06, "loss": 2.4819, "step": 6580 }, { "epoch": 0.659, "grad_norm": 24.331936278596, "learning_rate": 2.753163892582787e-06, "loss": 2.4075, "step": 6590 }, { "epoch": 0.66, "grad_norm": 25.794720714952284, "learning_rate": 2.7387090381293374e-06, "loss": 2.486, "step": 6600 }, { "epoch": 0.661, "grad_norm": 25.864540182072883, "learning_rate": 2.7242779035480556e-06, "loss": 2.4766, "step": 6610 }, { "epoch": 0.662, "grad_norm": 24.413756964280687, "learning_rate": 2.70987064021471e-06, "loss": 2.4185, "step": 6620 }, { "epoch": 0.663, "grad_norm": 26.124836337917497, "learning_rate": 2.6954873992546782e-06, "loss": 2.4909, "step": 6630 }, { "epoch": 0.664, "grad_norm": 23.172839632514748, "learning_rate": 2.681128331541354e-06, "loss": 2.4572, "step": 6640 }, { "epoch": 0.665, "grad_norm": 24.468174552604577, "learning_rate": 2.6667935876945583e-06, "loss": 2.4475, "step": 6650 }, { "epoch": 0.666, "grad_norm": 25.37978848164465, "learning_rate": 2.652483318078976e-06, "loss": 2.5147, "step": 6660 }, { "epoch": 0.667, "grad_norm": 26.067862632371757, "learning_rate": 2.6381976728025595e-06, "loss": 2.4542, "step": 6670 }, { "epoch": 0.668, "grad_norm": 25.302964897083438, "learning_rate": 2.6239368017149707e-06, "loss": 2.3765, "step": 6680 }, { "epoch": 0.669, "grad_norm": 23.865772457424207, "learning_rate": 2.6097008544059966e-06, "loss": 2.3889, "step": 6690 }, { "epoch": 0.67, "grad_norm": 25.333840987003736, "learning_rate": 2.59548998020399e-06, "loss": 2.4245, "step": 6700 }, { "epoch": 0.671, "grad_norm": 24.94064529852266, "learning_rate": 2.5813043281742984e-06, "loss": 2.4764, "step": 6710 }, { "epoch": 0.672, "grad_norm": 24.025257907929984, "learning_rate": 2.567144047117695e-06, "loss": 2.3844, "step": 6720 }, { "epoch": 0.673, "grad_norm": 29.731483890364103, "learning_rate": 2.5530092855688314e-06, "loss": 2.4027, "step": 6730 }, { "epoch": 0.674, "grad_norm": 26.171570102354803, "learning_rate": 2.5389001917946655e-06, "loss": 2.4366, "step": 6740 }, { "epoch": 0.675, "grad_norm": 24.315784281917395, "learning_rate": 2.524816913792916e-06, "loss": 2.4022, "step": 6750 }, { "epoch": 0.676, "grad_norm": 24.476459416161937, "learning_rate": 2.510759599290501e-06, "loss": 2.3781, "step": 6760 }, { "epoch": 0.677, "grad_norm": 22.883793338553513, "learning_rate": 2.4967283957420046e-06, "loss": 2.4373, "step": 6770 }, { "epoch": 0.678, "grad_norm": 27.034132314809487, "learning_rate": 2.4827234503281076e-06, "loss": 2.3305, "step": 6780 }, { "epoch": 0.679, "grad_norm": 25.89172946337445, "learning_rate": 2.4687449099540623e-06, "loss": 2.405, "step": 6790 }, { "epoch": 0.68, "grad_norm": 25.73628471136655, "learning_rate": 2.4547929212481436e-06, "loss": 2.3535, "step": 6800 }, { "epoch": 0.681, "grad_norm": 24.092239336988715, "learning_rate": 2.440867630560108e-06, "loss": 2.4279, "step": 6810 }, { "epoch": 0.682, "grad_norm": 24.7644049756857, "learning_rate": 2.426969183959669e-06, "loss": 2.3513, "step": 6820 }, { "epoch": 0.683, "grad_norm": 26.979421268065682, "learning_rate": 2.4130977272349523e-06, "loss": 2.3772, "step": 6830 }, { "epoch": 0.684, "grad_norm": 26.12502983487333, "learning_rate": 2.3992534058909765e-06, "loss": 2.4459, "step": 6840 }, { "epoch": 0.685, "grad_norm": 24.617381699815382, "learning_rate": 2.3854363651481195e-06, "loss": 2.3867, "step": 6850 }, { "epoch": 0.686, "grad_norm": 23.848485840021343, "learning_rate": 2.3716467499406017e-06, "loss": 2.4004, "step": 6860 }, { "epoch": 0.687, "grad_norm": 24.090100001880774, "learning_rate": 2.357884704914963e-06, "loss": 2.3315, "step": 6870 }, { "epoch": 0.688, "grad_norm": 26.403583211291842, "learning_rate": 2.3441503744285416e-06, "loss": 2.34, "step": 6880 }, { "epoch": 0.689, "grad_norm": 25.906301562095862, "learning_rate": 2.3304439025479676e-06, "loss": 2.3543, "step": 6890 }, { "epoch": 0.69, "grad_norm": 25.629763972473107, "learning_rate": 2.3167654330476412e-06, "loss": 2.3219, "step": 6900 }, { "epoch": 0.691, "grad_norm": 24.757161222522654, "learning_rate": 2.30311510940824e-06, "loss": 2.3035, "step": 6910 }, { "epoch": 0.692, "grad_norm": 25.783255487419826, "learning_rate": 2.289493074815195e-06, "loss": 2.4057, "step": 6920 }, { "epoch": 0.693, "grad_norm": 25.846604590717583, "learning_rate": 2.275899472157206e-06, "loss": 2.2895, "step": 6930 }, { "epoch": 0.694, "grad_norm": 26.29588391677404, "learning_rate": 2.262334444024733e-06, "loss": 2.3872, "step": 6940 }, { "epoch": 0.695, "grad_norm": 26.61730141440346, "learning_rate": 2.2487981327084994e-06, "loss": 2.2877, "step": 6950 }, { "epoch": 0.696, "grad_norm": 26.291579252389788, "learning_rate": 2.2352906801980097e-06, "loss": 2.3399, "step": 6960 }, { "epoch": 0.697, "grad_norm": 25.919397474398465, "learning_rate": 2.221812228180045e-06, "loss": 2.3595, "step": 6970 }, { "epoch": 0.698, "grad_norm": 25.476082686818277, "learning_rate": 2.208362918037192e-06, "loss": 2.335, "step": 6980 }, { "epoch": 0.699, "grad_norm": 23.1763232741886, "learning_rate": 2.1949428908463476e-06, "loss": 2.3368, "step": 6990 }, { "epoch": 0.7, "grad_norm": 24.25705043766323, "learning_rate": 2.1815522873772475e-06, "loss": 2.3476, "step": 7000 }, { "epoch": 0.701, "grad_norm": 24.07514922261785, "learning_rate": 2.168191248090989e-06, "loss": 2.3446, "step": 7010 }, { "epoch": 0.702, "grad_norm": 28.072947567529344, "learning_rate": 2.154859913138547e-06, "loss": 2.4, "step": 7020 }, { "epoch": 0.703, "grad_norm": 27.404631735901916, "learning_rate": 2.1415584223593212e-06, "loss": 2.3723, "step": 7030 }, { "epoch": 0.704, "grad_norm": 26.372344389105972, "learning_rate": 2.1282869152796514e-06, "loss": 2.3225, "step": 7040 }, { "epoch": 0.705, "grad_norm": 26.92085078677194, "learning_rate": 2.115045531111371e-06, "loss": 2.3455, "step": 7050 }, { "epoch": 0.706, "grad_norm": 24.78375496657101, "learning_rate": 2.101834408750331e-06, "loss": 2.4154, "step": 7060 }, { "epoch": 0.707, "grad_norm": 24.576018995860384, "learning_rate": 2.088653686774954e-06, "loss": 2.3094, "step": 7070 }, { "epoch": 0.708, "grad_norm": 24.551864672843873, "learning_rate": 2.075503503444778e-06, "loss": 2.2821, "step": 7080 }, { "epoch": 0.709, "grad_norm": 26.587151654876983, "learning_rate": 2.0623839966990045e-06, "loss": 2.2956, "step": 7090 }, { "epoch": 0.71, "grad_norm": 27.055603274037892, "learning_rate": 2.0492953041550485e-06, "loss": 2.2905, "step": 7100 }, { "epoch": 0.711, "grad_norm": 24.230108970851546, "learning_rate": 2.036237563107102e-06, "loss": 2.2827, "step": 7110 }, { "epoch": 0.712, "grad_norm": 22.728350264172107, "learning_rate": 2.0232109105246924e-06, "loss": 2.2897, "step": 7120 }, { "epoch": 0.713, "grad_norm": 27.505449893080915, "learning_rate": 2.010215483051237e-06, "loss": 2.3555, "step": 7130 }, { "epoch": 0.714, "grad_norm": 26.107252424122066, "learning_rate": 1.9972514170026243e-06, "loss": 2.3104, "step": 7140 }, { "epoch": 0.715, "grad_norm": 27.816169163605057, "learning_rate": 1.98431884836577e-06, "loss": 2.324, "step": 7150 }, { "epoch": 0.716, "grad_norm": 23.181094540000753, "learning_rate": 1.9714179127972016e-06, "loss": 2.2577, "step": 7160 }, { "epoch": 0.717, "grad_norm": 24.483803771670253, "learning_rate": 1.958548745621626e-06, "loss": 2.2789, "step": 7170 }, { "epoch": 0.718, "grad_norm": 26.333254671117768, "learning_rate": 1.9457114818305195e-06, "loss": 2.3651, "step": 7180 }, { "epoch": 0.719, "grad_norm": 25.838410858711878, "learning_rate": 1.9329062560807053e-06, "loss": 2.2637, "step": 7190 }, { "epoch": 0.72, "grad_norm": 26.672575828780516, "learning_rate": 1.92013320269294e-06, "loss": 2.3527, "step": 7200 }, { "epoch": 0.721, "grad_norm": 24.339995634745673, "learning_rate": 1.907392455650512e-06, "loss": 2.2096, "step": 7210 }, { "epoch": 0.722, "grad_norm": 26.727316556104213, "learning_rate": 1.8946841485978256e-06, "loss": 2.2764, "step": 7220 }, { "epoch": 0.723, "grad_norm": 27.4845705912396, "learning_rate": 1.8820084148390121e-06, "loss": 2.3463, "step": 7230 }, { "epoch": 0.724, "grad_norm": 23.85364893103126, "learning_rate": 1.8693653873365175e-06, "loss": 2.2716, "step": 7240 }, { "epoch": 0.725, "grad_norm": 24.247932401412996, "learning_rate": 1.8567551987097188e-06, "loss": 2.32, "step": 7250 }, { "epoch": 0.726, "grad_norm": 26.73137078188418, "learning_rate": 1.8441779812335292e-06, "loss": 2.344, "step": 7260 }, { "epoch": 0.727, "grad_norm": 24.253572136206408, "learning_rate": 1.8316338668370053e-06, "loss": 2.2814, "step": 7270 }, { "epoch": 0.728, "grad_norm": 26.632443691378253, "learning_rate": 1.8191229871019755e-06, "loss": 2.3811, "step": 7280 }, { "epoch": 0.729, "grad_norm": 28.58114144588924, "learning_rate": 1.8066454732616445e-06, "loss": 2.2837, "step": 7290 }, { "epoch": 0.73, "grad_norm": 26.74150213834366, "learning_rate": 1.794201456199231e-06, "loss": 2.3375, "step": 7300 }, { "epoch": 0.731, "grad_norm": 26.00854178236627, "learning_rate": 1.781791066446582e-06, "loss": 2.3048, "step": 7310 }, { "epoch": 0.732, "grad_norm": 27.068503844916602, "learning_rate": 1.7694144341828168e-06, "loss": 2.3482, "step": 7320 }, { "epoch": 0.733, "grad_norm": 26.95650993876601, "learning_rate": 1.7570716892329514e-06, "loss": 2.3809, "step": 7330 }, { "epoch": 0.734, "grad_norm": 27.50208158457101, "learning_rate": 1.7447629610665368e-06, "loss": 2.3702, "step": 7340 }, { "epoch": 0.735, "grad_norm": 26.27880650546668, "learning_rate": 1.7324883787963108e-06, "loss": 2.379, "step": 7350 }, { "epoch": 0.736, "grad_norm": 23.99253134471804, "learning_rate": 1.7202480711768305e-06, "loss": 2.2477, "step": 7360 }, { "epoch": 0.737, "grad_norm": 24.92658676502121, "learning_rate": 1.7080421666031311e-06, "loss": 2.2969, "step": 7370 }, { "epoch": 0.738, "grad_norm": 24.144501986027635, "learning_rate": 1.695870793109376e-06, "loss": 2.2202, "step": 7380 }, { "epoch": 0.739, "grad_norm": 26.10598808146374, "learning_rate": 1.6837340783675166e-06, "loss": 2.2424, "step": 7390 }, { "epoch": 0.74, "grad_norm": 27.300820930013252, "learning_rate": 1.6716321496859428e-06, "loss": 2.2026, "step": 7400 }, { "epoch": 0.741, "grad_norm": 24.232467551224033, "learning_rate": 1.6595651340081625e-06, "loss": 2.3171, "step": 7410 }, { "epoch": 0.742, "grad_norm": 25.77800150009333, "learning_rate": 1.6475331579114617e-06, "loss": 2.273, "step": 7420 }, { "epoch": 0.743, "grad_norm": 27.747218867338635, "learning_rate": 1.635536347605573e-06, "loss": 2.3209, "step": 7430 }, { "epoch": 0.744, "grad_norm": 26.375029850374176, "learning_rate": 1.6235748289313625e-06, "loss": 2.2768, "step": 7440 }, { "epoch": 0.745, "grad_norm": 26.083555238629184, "learning_rate": 1.6116487273594984e-06, "loss": 2.1958, "step": 7450 }, { "epoch": 0.746, "grad_norm": 28.51279565291339, "learning_rate": 1.5997581679891455e-06, "loss": 2.3105, "step": 7460 }, { "epoch": 0.747, "grad_norm": 22.487747833733575, "learning_rate": 1.5879032755466416e-06, "loss": 2.2906, "step": 7470 }, { "epoch": 0.748, "grad_norm": 26.162905848088915, "learning_rate": 1.5760841743842015e-06, "loss": 2.2388, "step": 7480 }, { "epoch": 0.749, "grad_norm": 27.03061809411013, "learning_rate": 1.5643009884786047e-06, "loss": 2.1925, "step": 7490 }, { "epoch": 0.75, "grad_norm": 28.507260754854872, "learning_rate": 1.5525538414298924e-06, "loss": 2.2306, "step": 7500 }, { "epoch": 0.751, "grad_norm": 24.460066196565702, "learning_rate": 1.5408428564600814e-06, "loss": 2.1724, "step": 7510 }, { "epoch": 0.752, "grad_norm": 25.467051496612523, "learning_rate": 1.5291681564118594e-06, "loss": 2.219, "step": 7520 }, { "epoch": 0.753, "grad_norm": 25.735831538247922, "learning_rate": 1.5175298637473095e-06, "loss": 2.1385, "step": 7530 }, { "epoch": 0.754, "grad_norm": 27.0360497229217, "learning_rate": 1.5059281005466104e-06, "loss": 2.1873, "step": 7540 }, { "epoch": 0.755, "grad_norm": 25.694799853275956, "learning_rate": 1.4943629885067707e-06, "loss": 2.1987, "step": 7550 }, { "epoch": 0.756, "grad_norm": 28.272742345909823, "learning_rate": 1.482834648940344e-06, "loss": 2.1962, "step": 7560 }, { "epoch": 0.757, "grad_norm": 25.557074228341754, "learning_rate": 1.4713432027741542e-06, "loss": 2.2423, "step": 7570 }, { "epoch": 0.758, "grad_norm": 25.60263509771645, "learning_rate": 1.4598887705480374e-06, "loss": 2.2165, "step": 7580 }, { "epoch": 0.759, "grad_norm": 24.077746805841073, "learning_rate": 1.4484714724135645e-06, "loss": 2.1744, "step": 7590 }, { "epoch": 0.76, "grad_norm": 23.99506491939666, "learning_rate": 1.437091428132793e-06, "loss": 2.1723, "step": 7600 }, { "epoch": 0.761, "grad_norm": 26.35544513574848, "learning_rate": 1.4257487570769996e-06, "loss": 2.1868, "step": 7610 }, { "epoch": 0.762, "grad_norm": 25.77855196550203, "learning_rate": 1.4144435782254373e-06, "loss": 2.1938, "step": 7620 }, { "epoch": 0.763, "grad_norm": 23.807810707527555, "learning_rate": 1.4031760101640857e-06, "loss": 2.2489, "step": 7630 }, { "epoch": 0.764, "grad_norm": 25.729277905952316, "learning_rate": 1.391946171084398e-06, "loss": 2.1859, "step": 7640 }, { "epoch": 0.765, "grad_norm": 26.287185866809487, "learning_rate": 1.3807541787820755e-06, "loss": 2.2054, "step": 7650 }, { "epoch": 0.766, "grad_norm": 22.842704897976155, "learning_rate": 1.3696001506558181e-06, "loss": 2.1293, "step": 7660 }, { "epoch": 0.767, "grad_norm": 27.699164490163895, "learning_rate": 1.3584842037061053e-06, "loss": 2.1594, "step": 7670 }, { "epoch": 0.768, "grad_norm": 25.31103410822779, "learning_rate": 1.34740645453396e-06, "loss": 2.085, "step": 7680 }, { "epoch": 0.769, "grad_norm": 27.729696866634114, "learning_rate": 1.3363670193397304e-06, "loss": 2.1651, "step": 7690 }, { "epoch": 0.77, "grad_norm": 26.540278530318623, "learning_rate": 1.3253660139218648e-06, "loss": 2.2037, "step": 7700 }, { "epoch": 0.771, "grad_norm": 25.257065405712982, "learning_rate": 1.3144035536757088e-06, "loss": 2.2154, "step": 7710 }, { "epoch": 0.772, "grad_norm": 32.06343384404941, "learning_rate": 1.3034797535922783e-06, "loss": 2.1474, "step": 7720 }, { "epoch": 0.773, "grad_norm": 28.179289621156702, "learning_rate": 1.2925947282570695e-06, "loss": 2.124, "step": 7730 }, { "epoch": 1.0008, "grad_norm": 28.096443580541486, "learning_rate": 1.2817485918488493e-06, "loss": 2.8048, "step": 7740 }, { "epoch": 1.0018, "grad_norm": 33.12007199930344, "learning_rate": 1.2709414581384526e-06, "loss": 2.7396, "step": 7750 }, { "epoch": 1.0028, "grad_norm": 28.10444747358109, "learning_rate": 1.2601734404876015e-06, "loss": 2.6114, "step": 7760 }, { "epoch": 1.0038, "grad_norm": 28.9389777960749, "learning_rate": 1.2494446518477022e-06, "loss": 2.6638, "step": 7770 }, { "epoch": 1.0048, "grad_norm": 26.511320233424396, "learning_rate": 1.2387552047586747e-06, "loss": 2.69, "step": 7780 }, { "epoch": 1.0058, "grad_norm": 23.7599020464409, "learning_rate": 1.2281052113477575e-06, "loss": 2.6294, "step": 7790 }, { "epoch": 1.0068, "grad_norm": 24.48510219972502, "learning_rate": 1.2174947833283434e-06, "loss": 2.4996, "step": 7800 }, { "epoch": 1.0078, "grad_norm": 23.556204861121092, "learning_rate": 1.2069240319988035e-06, "loss": 2.4405, "step": 7810 }, { "epoch": 1.0088, "grad_norm": 26.016733382097975, "learning_rate": 1.1963930682413133e-06, "loss": 2.5715, "step": 7820 }, { "epoch": 1.0098, "grad_norm": 26.216010703422697, "learning_rate": 1.185902002520703e-06, "loss": 2.5144, "step": 7830 }, { "epoch": 1.0108, "grad_norm": 24.814722399815572, "learning_rate": 1.175450944883285e-06, "loss": 2.5445, "step": 7840 }, { "epoch": 1.0118, "grad_norm": 25.69654933840267, "learning_rate": 1.1650400049557093e-06, "loss": 2.4173, "step": 7850 }, { "epoch": 1.0128, "grad_norm": 24.91611451062141, "learning_rate": 1.1546692919438118e-06, "loss": 2.4522, "step": 7860 }, { "epoch": 1.0138, "grad_norm": 23.877781256311582, "learning_rate": 1.1443389146314609e-06, "loss": 2.4324, "step": 7870 }, { "epoch": 1.0148, "grad_norm": 27.56070320001817, "learning_rate": 1.1340489813794297e-06, "loss": 2.3838, "step": 7880 }, { "epoch": 1.0158, "grad_norm": 24.65559356968328, "learning_rate": 1.1237996001242457e-06, "loss": 2.4206, "step": 7890 }, { "epoch": 1.0168, "grad_norm": 22.922372251998187, "learning_rate": 1.1135908783770715e-06, "loss": 2.4669, "step": 7900 }, { "epoch": 1.0178, "grad_norm": 23.564915673072843, "learning_rate": 1.1034229232225646e-06, "loss": 2.39, "step": 7910 }, { "epoch": 1.0188, "grad_norm": 28.266737179460407, "learning_rate": 1.0932958413177647e-06, "loss": 2.4413, "step": 7920 }, { "epoch": 1.0198, "grad_norm": 25.494878024354723, "learning_rate": 1.0832097388909707e-06, "loss": 2.4092, "step": 7930 }, { "epoch": 1.0208, "grad_norm": 26.019352040791436, "learning_rate": 1.0731647217406204e-06, "loss": 2.3811, "step": 7940 }, { "epoch": 1.0218, "grad_norm": 26.822832639050564, "learning_rate": 1.0631608952341942e-06, "loss": 2.4522, "step": 7950 }, { "epoch": 1.0228, "grad_norm": 27.106340742211213, "learning_rate": 1.053198364307096e-06, "loss": 2.4628, "step": 7960 }, { "epoch": 1.0238, "grad_norm": 25.243913665211945, "learning_rate": 1.0432772334615626e-06, "loss": 2.4413, "step": 7970 }, { "epoch": 1.0248, "grad_norm": 25.70874850822169, "learning_rate": 1.0333976067655589e-06, "loss": 2.5097, "step": 7980 }, { "epoch": 1.0258, "grad_norm": 27.362461438988063, "learning_rate": 1.0235595878516985e-06, "loss": 2.3634, "step": 7990 }, { "epoch": 1.0268, "grad_norm": 23.356195526997034, "learning_rate": 1.0137632799161411e-06, "loss": 2.4136, "step": 8000 }, { "epoch": 1.0278, "grad_norm": 24.87623384423107, "learning_rate": 1.0040087857175246e-06, "loss": 2.3788, "step": 8010 }, { "epoch": 1.0288, "grad_norm": 32.81628039949997, "learning_rate": 9.942962075758738e-07, "loss": 2.4341, "step": 8020 }, { "epoch": 1.0298, "grad_norm": 27.87459574894831, "learning_rate": 9.846256473715394e-07, "loss": 2.4134, "step": 8030 }, { "epoch": 1.0308, "grad_norm": 26.281711293781964, "learning_rate": 9.749972065441228e-07, "loss": 2.4196, "step": 8040 }, { "epoch": 1.0318, "grad_norm": 25.476574997118313, "learning_rate": 9.654109860914096e-07, "loss": 2.3931, "step": 8050 }, { "epoch": 1.0328, "grad_norm": 24.458305970638136, "learning_rate": 9.55867086568319e-07, "loss": 2.3635, "step": 8060 }, { "epoch": 1.0338, "grad_norm": 24.563518460221605, "learning_rate": 9.463656080858386e-07, "loss": 2.3809, "step": 8070 }, { "epoch": 1.0348, "grad_norm": 26.191407686350722, "learning_rate": 9.369066503099838e-07, "loss": 2.4373, "step": 8080 }, { "epoch": 1.0358, "grad_norm": 30.140088434622264, "learning_rate": 9.274903124607432e-07, "loss": 2.4661, "step": 8090 }, { "epoch": 1.0368, "grad_norm": 24.514048117020785, "learning_rate": 9.18116693311047e-07, "loss": 2.3979, "step": 8100 }, { "epoch": 1.0378, "grad_norm": 26.629293576133037, "learning_rate": 9.087858911857245e-07, "loss": 2.3623, "step": 8110 }, { "epoch": 1.0388, "grad_norm": 29.005175520768276, "learning_rate": 8.994980039604739e-07, "loss": 2.3407, "step": 8120 }, { "epoch": 1.0398, "grad_norm": 27.859486947015935, "learning_rate": 8.902531290608384e-07, "loss": 2.3804, "step": 8130 }, { "epoch": 1.0408, "grad_norm": 25.594108020945733, "learning_rate": 8.810513634611795e-07, "loss": 2.3326, "step": 8140 }, { "epoch": 1.0418, "grad_norm": 28.326276304275005, "learning_rate": 8.71892803683666e-07, "loss": 2.3607, "step": 8150 }, { "epoch": 1.0428, "grad_norm": 25.19709681818264, "learning_rate": 8.627775457972537e-07, "loss": 2.3288, "step": 8160 }, { "epoch": 1.0438, "grad_norm": 27.9821645737335, "learning_rate": 8.537056854166864e-07, "loss": 2.36, "step": 8170 }, { "epoch": 1.0448, "grad_norm": 27.420215610945476, "learning_rate": 8.446773177014877e-07, "loss": 2.3033, "step": 8180 }, { "epoch": 1.0458, "grad_norm": 25.1432949673955, "learning_rate": 8.356925373549601e-07, "loss": 2.341, "step": 8190 }, { "epoch": 1.0468, "grad_norm": 25.60216843086142, "learning_rate": 8.267514386232007e-07, "loss": 2.3913, "step": 8200 }, { "epoch": 1.0478, "grad_norm": 28.588709408335454, "learning_rate": 8.178541152941016e-07, "loss": 2.3983, "step": 8210 }, { "epoch": 1.0488, "grad_norm": 26.58098016107466, "learning_rate": 8.090006606963769e-07, "loss": 2.3198, "step": 8220 }, { "epoch": 1.0498, "grad_norm": 26.18202991539712, "learning_rate": 8.001911676985735e-07, "loss": 2.3381, "step": 8230 }, { "epoch": 1.0508, "grad_norm": 24.96453982252671, "learning_rate": 7.914257287081062e-07, "loss": 2.3448, "step": 8240 }, { "epoch": 1.0518, "grad_norm": 27.864940645069236, "learning_rate": 7.827044356702823e-07, "loss": 2.2768, "step": 8250 }, { "epoch": 1.0528, "grad_norm": 26.93943293613992, "learning_rate": 7.740273800673376e-07, "loss": 2.4058, "step": 8260 }, { "epoch": 1.0538, "grad_norm": 24.559291167940508, "learning_rate": 7.653946529174816e-07, "loss": 2.3062, "step": 8270 }, { "epoch": 1.0548, "grad_norm": 25.895865937226787, "learning_rate": 7.568063447739327e-07, "loss": 2.2638, "step": 8280 }, { "epoch": 1.0558, "grad_norm": 27.611511617253388, "learning_rate": 7.482625457239856e-07, "loss": 2.3578, "step": 8290 }, { "epoch": 1.0568, "grad_norm": 26.776974072505606, "learning_rate": 7.397633453880449e-07, "loss": 2.4621, "step": 8300 }, { "epoch": 1.0578, "grad_norm": 25.908173277516216, "learning_rate": 7.313088329187029e-07, "loss": 2.2832, "step": 8310 }, { "epoch": 1.0588, "grad_norm": 25.507274649330185, "learning_rate": 7.228990969997924e-07, "loss": 2.2994, "step": 8320 }, { "epoch": 1.0598, "grad_norm": 29.05199811394889, "learning_rate": 7.145342258454657e-07, "loss": 2.3014, "step": 8330 }, { "epoch": 1.0608, "grad_norm": 27.44061385868196, "learning_rate": 7.062143071992634e-07, "loss": 2.2716, "step": 8340 }, { "epoch": 1.0618, "grad_norm": 26.16743292830829, "learning_rate": 6.979394283331925e-07, "loss": 2.2931, "step": 8350 }, { "epoch": 1.0628, "grad_norm": 27.08917098833595, "learning_rate": 6.897096760468203e-07, "loss": 2.4061, "step": 8360 }, { "epoch": 1.0638, "grad_norm": 24.43992363805331, "learning_rate": 6.815251366663505e-07, "loss": 2.3016, "step": 8370 }, { "epoch": 1.0648, "grad_norm": 28.24092734216542, "learning_rate": 6.733858960437312e-07, "loss": 2.3434, "step": 8380 }, { "epoch": 1.0658, "grad_norm": 26.93047495154958, "learning_rate": 6.652920395557427e-07, "loss": 2.2562, "step": 8390 }, { "epoch": 1.0668, "grad_norm": 28.063654292062576, "learning_rate": 6.572436521031111e-07, "loss": 2.3761, "step": 8400 }, { "epoch": 1.0678, "grad_norm": 26.209448670430756, "learning_rate": 6.492408181096133e-07, "loss": 2.3419, "step": 8410 }, { "epoch": 1.0688, "grad_norm": 25.93335280623505, "learning_rate": 6.412836215211881e-07, "loss": 2.3152, "step": 8420 }, { "epoch": 1.0698, "grad_norm": 25.21843409076808, "learning_rate": 6.333721458050657e-07, "loss": 2.3486, "step": 8430 }, { "epoch": 1.0708, "grad_norm": 25.557791039075923, "learning_rate": 6.255064739488797e-07, "loss": 2.3693, "step": 8440 }, { "epoch": 1.0718, "grad_norm": 28.222336751949182, "learning_rate": 6.176866884598087e-07, "loss": 2.3328, "step": 8450 }, { "epoch": 1.0728, "grad_norm": 26.67164314242069, "learning_rate": 6.099128713636998e-07, "loss": 2.3694, "step": 8460 }, { "epoch": 1.0738, "grad_norm": 23.067056854173817, "learning_rate": 6.021851042042165e-07, "loss": 2.3211, "step": 8470 }, { "epoch": 1.0748, "grad_norm": 27.00815764932028, "learning_rate": 5.945034680419815e-07, "loss": 2.3435, "step": 8480 }, { "epoch": 1.0758, "grad_norm": 28.771321339274607, "learning_rate": 5.868680434537205e-07, "loss": 2.3436, "step": 8490 }, { "epoch": 1.0768, "grad_norm": 24.10866778156386, "learning_rate": 5.792789105314262e-07, "loss": 2.3198, "step": 8500 }, { "epoch": 1.0778, "grad_norm": 25.25287563136629, "learning_rate": 5.71736148881511e-07, "loss": 2.2544, "step": 8510 }, { "epoch": 1.0788, "grad_norm": 24.517833035349707, "learning_rate": 5.64239837623976e-07, "loss": 2.314, "step": 8520 }, { "epoch": 1.0798, "grad_norm": 28.60357797619362, "learning_rate": 5.567900553915784e-07, "loss": 2.3069, "step": 8530 }, { "epoch": 1.0808, "grad_norm": 26.942889806341707, "learning_rate": 5.493868803290087e-07, "loss": 2.322, "step": 8540 }, { "epoch": 1.0818, "grad_norm": 26.483948694077913, "learning_rate": 5.420303900920715e-07, "loss": 2.2676, "step": 8550 }, { "epoch": 1.0828, "grad_norm": 24.746296849050733, "learning_rate": 5.347206618468664e-07, "loss": 2.3108, "step": 8560 }, { "epoch": 1.0838, "grad_norm": 28.42986398797777, "learning_rate": 5.274577722689855e-07, "loss": 2.3006, "step": 8570 }, { "epoch": 1.0848, "grad_norm": 27.553258379355135, "learning_rate": 5.202417975427015e-07, "loss": 2.3418, "step": 8580 }, { "epoch": 1.0858, "grad_norm": 27.043988156633453, "learning_rate": 5.130728133601781e-07, "loss": 2.3669, "step": 8590 }, { "epoch": 1.0868, "grad_norm": 25.815783614498034, "learning_rate": 5.059508949206643e-07, "loss": 2.3847, "step": 8600 }, { "epoch": 1.0878, "grad_norm": 27.797884275143595, "learning_rate": 4.988761169297163e-07, "loss": 2.3429, "step": 8610 }, { "epoch": 1.0888, "grad_norm": 25.639045580227002, "learning_rate": 4.918485535984052e-07, "loss": 2.3604, "step": 8620 }, { "epoch": 1.0898, "grad_norm": 29.390128387373004, "learning_rate": 4.848682786425479e-07, "loss": 2.3184, "step": 8630 }, { "epoch": 1.0908, "grad_norm": 29.328928772545446, "learning_rate": 4.779353652819229e-07, "loss": 2.3019, "step": 8640 }, { "epoch": 1.0918, "grad_norm": 26.72275601584289, "learning_rate": 4.710498862395119e-07, "loss": 2.3028, "step": 8650 }, { "epoch": 1.0928, "grad_norm": 27.195556721853382, "learning_rate": 4.642119137407336e-07, "loss": 2.2894, "step": 8660 }, { "epoch": 1.0937999999999999, "grad_norm": 24.84217045101147, "learning_rate": 4.5742151951268075e-07, "loss": 2.2962, "step": 8670 }, { "epoch": 1.0948, "grad_norm": 27.421769087652248, "learning_rate": 4.5067877478337795e-07, "loss": 2.3474, "step": 8680 }, { "epoch": 1.0958, "grad_norm": 26.09212800155693, "learning_rate": 4.4398375028102513e-07, "loss": 2.3787, "step": 8690 }, { "epoch": 1.0968, "grad_norm": 27.01612930173846, "learning_rate": 4.373365162332632e-07, "loss": 2.3201, "step": 8700 }, { "epoch": 1.0977999999999999, "grad_norm": 26.27456433596079, "learning_rate": 4.307371423664297e-07, "loss": 2.3494, "step": 8710 }, { "epoch": 1.0988, "grad_norm": 26.80387034428916, "learning_rate": 4.241856979048353e-07, "loss": 2.3873, "step": 8720 }, { "epoch": 1.0998, "grad_norm": 26.69892635627918, "learning_rate": 4.176822515700324e-07, "loss": 2.308, "step": 8730 }, { "epoch": 1.1008, "grad_norm": 27.712745551313446, "learning_rate": 4.112268715800943e-07, "loss": 2.2557, "step": 8740 }, { "epoch": 1.1018, "grad_norm": 27.377655752750112, "learning_rate": 4.0481962564890454e-07, "loss": 2.3439, "step": 8750 }, { "epoch": 1.1028, "grad_norm": 25.635683401118108, "learning_rate": 3.9846058098543863e-07, "loss": 2.2742, "step": 8760 }, { "epoch": 1.1038000000000001, "grad_norm": 26.634050911069927, "learning_rate": 3.9214980429306827e-07, "loss": 2.3637, "step": 8770 }, { "epoch": 1.1048, "grad_norm": 28.444270395195964, "learning_rate": 3.858873617688524e-07, "loss": 2.315, "step": 8780 }, { "epoch": 1.1058, "grad_norm": 27.333694580077466, "learning_rate": 3.7967331910285175e-07, "loss": 2.3616, "step": 8790 }, { "epoch": 1.1068, "grad_norm": 25.687427348023267, "learning_rate": 3.7350774147743353e-07, "loss": 2.4186, "step": 8800 }, { "epoch": 1.1078000000000001, "grad_norm": 26.631452203403473, "learning_rate": 3.6739069356658773e-07, "loss": 2.2656, "step": 8810 }, { "epoch": 1.1088, "grad_norm": 25.300817381097126, "learning_rate": 3.613222395352545e-07, "loss": 2.3755, "step": 8820 }, { "epoch": 1.1098, "grad_norm": 26.09137437276164, "learning_rate": 3.5530244303864316e-07, "loss": 2.3339, "step": 8830 }, { "epoch": 1.1108, "grad_norm": 24.406065715871588, "learning_rate": 3.49331367221572e-07, "loss": 2.25, "step": 8840 }, { "epoch": 1.1118, "grad_norm": 29.196050928620913, "learning_rate": 3.434090747177998e-07, "loss": 2.3672, "step": 8850 }, { "epoch": 1.1128, "grad_norm": 27.990494282797997, "learning_rate": 3.37535627649373e-07, "loss": 2.3087, "step": 8860 }, { "epoch": 1.1138, "grad_norm": 24.76767648327167, "learning_rate": 3.317110876259727e-07, "loss": 2.3111, "step": 8870 }, { "epoch": 1.1148, "grad_norm": 25.51110573488626, "learning_rate": 3.259355157442651e-07, "loss": 2.3466, "step": 8880 }, { "epoch": 1.1158, "grad_norm": 26.772769008204936, "learning_rate": 3.202089725872698e-07, "loss": 2.2406, "step": 8890 }, { "epoch": 1.1168, "grad_norm": 26.545604829247825, "learning_rate": 3.145315182237124e-07, "loss": 2.2638, "step": 8900 }, { "epoch": 1.1178, "grad_norm": 26.13495912905379, "learning_rate": 3.0890321220740425e-07, "loss": 2.2614, "step": 8910 }, { "epoch": 1.1188, "grad_norm": 24.89655264382916, "learning_rate": 3.0332411357661073e-07, "loss": 2.2991, "step": 8920 }, { "epoch": 1.1198, "grad_norm": 26.76510292542302, "learning_rate": 2.977942808534379e-07, "loss": 2.317, "step": 8930 }, { "epoch": 1.1208, "grad_norm": 23.70936647653368, "learning_rate": 2.923137720432123e-07, "loss": 2.3648, "step": 8940 }, { "epoch": 1.1218, "grad_norm": 27.901996205312333, "learning_rate": 2.868826446338796e-07, "loss": 2.2745, "step": 8950 }, { "epoch": 1.1228, "grad_norm": 25.153545235685183, "learning_rate": 2.8150095559539624e-07, "loss": 2.2573, "step": 8960 }, { "epoch": 1.1238, "grad_norm": 25.436315619991916, "learning_rate": 2.761687613791331e-07, "loss": 2.3115, "step": 8970 }, { "epoch": 1.1248, "grad_norm": 27.1581933743001, "learning_rate": 2.708861179172856e-07, "loss": 2.3748, "step": 8980 }, { "epoch": 1.1258, "grad_norm": 27.314489705507533, "learning_rate": 2.656530806222818e-07, "loss": 2.2885, "step": 8990 }, { "epoch": 1.1268, "grad_norm": 26.79264738934702, "learning_rate": 2.604697043862087e-07, "loss": 2.3333, "step": 9000 }, { "epoch": 1.1278, "grad_norm": 26.168238762677678, "learning_rate": 2.553360435802293e-07, "loss": 2.2624, "step": 9010 }, { "epoch": 1.1288, "grad_norm": 28.353926372931568, "learning_rate": 2.5025215205401633e-07, "loss": 2.3605, "step": 9020 }, { "epoch": 1.1298, "grad_norm": 27.525244207000945, "learning_rate": 2.4521808313518694e-07, "loss": 2.2292, "step": 9030 }, { "epoch": 1.1308, "grad_norm": 29.368436797551205, "learning_rate": 2.402338896287404e-07, "loss": 2.2584, "step": 9040 }, { "epoch": 1.1318, "grad_norm": 26.52980526587858, "learning_rate": 2.3529962381651007e-07, "loss": 2.3502, "step": 9050 }, { "epoch": 1.1328, "grad_norm": 27.823248991619334, "learning_rate": 2.3041533745660817e-07, "loss": 2.3517, "step": 9060 }, { "epoch": 1.1338, "grad_norm": 24.16331552540572, "learning_rate": 2.2558108178288795e-07, "loss": 2.319, "step": 9070 }, { "epoch": 1.1348, "grad_norm": 29.55031395532824, "learning_rate": 2.2079690750440364e-07, "loss": 2.2323, "step": 9080 }, { "epoch": 1.1358, "grad_norm": 25.415159343884735, "learning_rate": 2.160628648048796e-07, "loss": 2.3044, "step": 9090 }, { "epoch": 1.1368, "grad_norm": 25.471389785275303, "learning_rate": 2.113790033421842e-07, "loss": 2.2913, "step": 9100 }, { "epoch": 1.1378, "grad_norm": 26.645572563691893, "learning_rate": 2.0674537224780633e-07, "loss": 2.3593, "step": 9110 }, { "epoch": 1.1388, "grad_norm": 25.46999564556415, "learning_rate": 2.021620201263458e-07, "loss": 2.2681, "step": 9120 }, { "epoch": 1.1398, "grad_norm": 27.472772309176353, "learning_rate": 1.9762899505499532e-07, "loss": 2.2704, "step": 9130 }, { "epoch": 1.1408, "grad_norm": 27.02405224520672, "learning_rate": 1.9314634458304505e-07, "loss": 2.1873, "step": 9140 }, { "epoch": 1.1418, "grad_norm": 28.27054912434433, "learning_rate": 1.8871411573137488e-07, "loss": 2.4105, "step": 9150 }, { "epoch": 1.1428, "grad_norm": 27.308212110420918, "learning_rate": 1.843323549919701e-07, "loss": 2.2756, "step": 9160 }, { "epoch": 1.1438, "grad_norm": 24.62064607531292, "learning_rate": 1.8000110832742722e-07, "loss": 2.2932, "step": 9170 }, { "epoch": 1.1448, "grad_norm": 28.158504012134998, "learning_rate": 1.7572042117047272e-07, "loss": 2.3736, "step": 9180 }, { "epoch": 1.1458, "grad_norm": 25.690214619166227, "learning_rate": 1.7149033842349116e-07, "loss": 2.3062, "step": 9190 }, { "epoch": 1.1468, "grad_norm": 24.652939575243856, "learning_rate": 1.6731090445804787e-07, "loss": 2.3043, "step": 9200 }, { "epoch": 1.1478, "grad_norm": 24.536919819245142, "learning_rate": 1.6318216311442814e-07, "loss": 2.2841, "step": 9210 }, { "epoch": 1.1488, "grad_norm": 28.74133807734041, "learning_rate": 1.5910415770117428e-07, "loss": 2.2844, "step": 9220 }, { "epoch": 1.1498, "grad_norm": 24.79279090865744, "learning_rate": 1.5507693099463428e-07, "loss": 2.322, "step": 9230 }, { "epoch": 1.1508, "grad_norm": 26.691122594068645, "learning_rate": 1.5110052523850894e-07, "loss": 2.2836, "step": 9240 }, { "epoch": 1.1518, "grad_norm": 28.69056174254246, "learning_rate": 1.4717498214341486e-07, "loss": 2.3825, "step": 9250 }, { "epoch": 1.1528, "grad_norm": 25.75925190864323, "learning_rate": 1.43300342886441e-07, "loss": 2.3919, "step": 9260 }, { "epoch": 1.1538, "grad_norm": 28.38184177591823, "learning_rate": 1.3947664811071959e-07, "loss": 2.4032, "step": 9270 }, { "epoch": 1.1548, "grad_norm": 26.733748015267956, "learning_rate": 1.357039379249997e-07, "loss": 2.3352, "step": 9280 }, { "epoch": 1.1558, "grad_norm": 25.619253294505135, "learning_rate": 1.3198225190322488e-07, "loss": 2.3419, "step": 9290 }, { "epoch": 1.1568, "grad_norm": 24.60161658083279, "learning_rate": 1.2831162908412244e-07, "loss": 2.3361, "step": 9300 }, { "epoch": 1.1578, "grad_norm": 25.62154772753153, "learning_rate": 1.246921079707869e-07, "loss": 2.2865, "step": 9310 }, { "epoch": 1.1588, "grad_norm": 27.677289242505708, "learning_rate": 1.211237265302828e-07, "loss": 2.3919, "step": 9320 }, { "epoch": 1.1598, "grad_norm": 28.424833504286372, "learning_rate": 1.176065221932432e-07, "loss": 2.313, "step": 9330 }, { "epoch": 1.1608, "grad_norm": 25.754479273863822, "learning_rate": 1.1414053185347663e-07, "loss": 2.3427, "step": 9340 }, { "epoch": 1.1618, "grad_norm": 25.48874547338306, "learning_rate": 1.1072579186758303e-07, "loss": 2.3226, "step": 9350 }, { "epoch": 1.1628, "grad_norm": 28.80653772673039, "learning_rate": 1.0736233805456741e-07, "loss": 2.3387, "step": 9360 }, { "epoch": 1.1638, "grad_norm": 27.458728371294356, "learning_rate": 1.0405020569546953e-07, "loss": 2.3002, "step": 9370 }, { "epoch": 1.1648, "grad_norm": 25.85562685518478, "learning_rate": 1.0078942953298976e-07, "loss": 2.3121, "step": 9380 }, { "epoch": 1.1658, "grad_norm": 24.735707412339835, "learning_rate": 9.758004377112829e-08, "loss": 2.4236, "step": 9390 }, { "epoch": 1.1668, "grad_norm": 27.59806507889652, "learning_rate": 9.442208207482262e-08, "loss": 2.3063, "step": 9400 }, { "epoch": 1.1678, "grad_norm": 26.189586613172487, "learning_rate": 9.13155775695962e-08, "loss": 2.3735, "step": 9410 }, { "epoch": 1.1688, "grad_norm": 26.44024540588109, "learning_rate": 8.826056284121199e-08, "loss": 2.2855, "step": 9420 }, { "epoch": 1.1698, "grad_norm": 26.14403738198444, "learning_rate": 8.525706993532834e-08, "loss": 2.2897, "step": 9430 }, { "epoch": 1.1708, "grad_norm": 25.189474390071606, "learning_rate": 8.230513035716537e-08, "loss": 2.3502, "step": 9440 }, { "epoch": 1.1718, "grad_norm": 25.907173180185104, "learning_rate": 7.940477507117072e-08, "loss": 2.289, "step": 9450 }, { "epoch": 1.1728, "grad_norm": 25.64112920798971, "learning_rate": 7.655603450070048e-08, "loss": 2.2908, "step": 9460 }, { "epoch": 1.1738, "grad_norm": 28.398885404106423, "learning_rate": 7.375893852769544e-08, "loss": 2.3698, "step": 9470 }, { "epoch": 1.1748, "grad_norm": 24.832379685746886, "learning_rate": 7.101351649236754e-08, "loss": 2.2532, "step": 9480 }, { "epoch": 1.1758, "grad_norm": 28.477294704877952, "learning_rate": 6.83197971928956e-08, "loss": 2.3198, "step": 9490 }, { "epoch": 1.1768, "grad_norm": 27.0957467118128, "learning_rate": 6.56778088851201e-08, "loss": 2.3151, "step": 9500 }, { "epoch": 1.1778, "grad_norm": 23.946890567818333, "learning_rate": 6.308757928224885e-08, "loss": 2.2793, "step": 9510 }, { "epoch": 1.1788, "grad_norm": 26.91283749052814, "learning_rate": 6.054913555456344e-08, "loss": 2.36, "step": 9520 }, { "epoch": 1.1798, "grad_norm": 25.716116587948775, "learning_rate": 5.806250432913885e-08, "loss": 2.3364, "step": 9530 }, { "epoch": 1.1808, "grad_norm": 27.836500682148433, "learning_rate": 5.562771168955927e-08, "loss": 2.3363, "step": 9540 }, { "epoch": 1.1818, "grad_norm": 25.282203128369027, "learning_rate": 5.324478317564774e-08, "loss": 2.2719, "step": 9550 }, { "epoch": 1.1828, "grad_norm": 26.719339242846456, "learning_rate": 5.09137437831958e-08, "loss": 2.3416, "step": 9560 }, { "epoch": 1.1838, "grad_norm": 26.487823952227128, "learning_rate": 4.8634617963704277e-08, "loss": 2.3358, "step": 9570 }, { "epoch": 1.1848, "grad_norm": 25.312991823625193, "learning_rate": 4.6407429624125144e-08, "loss": 2.2529, "step": 9580 }, { "epoch": 1.1858, "grad_norm": 29.122749708881404, "learning_rate": 4.42322021266095e-08, "loss": 2.3331, "step": 9590 }, { "epoch": 1.1868, "grad_norm": 25.942939707895146, "learning_rate": 4.2108958288264446e-08, "loss": 2.2885, "step": 9600 }, { "epoch": 1.1878, "grad_norm": 24.898091767371202, "learning_rate": 4.00377203809138e-08, "loss": 2.2779, "step": 9610 }, { "epoch": 1.1888, "grad_norm": 26.452241128286474, "learning_rate": 3.801851013086222e-08, "loss": 2.3353, "step": 9620 }, { "epoch": 1.1898, "grad_norm": 25.722772077963615, "learning_rate": 3.6051348718670333e-08, "loss": 2.2603, "step": 9630 }, { "epoch": 1.1908, "grad_norm": 26.261274204477235, "learning_rate": 3.413625677893051e-08, "loss": 2.2891, "step": 9640 }, { "epoch": 1.1918, "grad_norm": 27.021967256306635, "learning_rate": 3.227325440005091e-08, "loss": 2.388, "step": 9650 }, { "epoch": 1.1928, "grad_norm": 26.939450380940638, "learning_rate": 3.0462361124044545e-08, "loss": 2.3235, "step": 9660 }, { "epoch": 1.1938, "grad_norm": 28.65360420429247, "learning_rate": 2.8703595946325547e-08, "loss": 2.3289, "step": 9670 }, { "epoch": 1.1948, "grad_norm": 27.805112621267664, "learning_rate": 2.6996977315507656e-08, "loss": 2.3184, "step": 9680 }, { "epoch": 1.1958, "grad_norm": 26.30230479393334, "learning_rate": 2.5342523133213835e-08, "loss": 2.4011, "step": 9690 }, { "epoch": 1.1968, "grad_norm": 25.020461728659416, "learning_rate": 2.3740250753884754e-08, "loss": 2.2706, "step": 9700 }, { "epoch": 1.1978, "grad_norm": 23.576883339896334, "learning_rate": 2.219017698460002e-08, "loss": 2.2469, "step": 9710 }, { "epoch": 1.1988, "grad_norm": 26.079668998342232, "learning_rate": 2.0692318084899465e-08, "loss": 2.3442, "step": 9720 }, { "epoch": 1.1998, "grad_norm": 26.467579093561607, "learning_rate": 1.924668976661437e-08, "loss": 2.2691, "step": 9730 }, { "epoch": 1.2008, "grad_norm": 27.35931792679388, "learning_rate": 1.7853307193702042e-08, "loss": 2.3084, "step": 9740 }, { "epoch": 1.2018, "grad_norm": 24.90413026591534, "learning_rate": 1.6512184982085953e-08, "loss": 2.3481, "step": 9750 }, { "epoch": 1.2028, "grad_norm": 25.958840263608604, "learning_rate": 1.522333719950364e-08, "loss": 2.2888, "step": 9760 }, { "epoch": 1.2038, "grad_norm": 23.86333087620246, "learning_rate": 1.3986777365358473e-08, "loss": 2.3272, "step": 9770 }, { "epoch": 1.2048, "grad_norm": 26.24604808246558, "learning_rate": 1.2802518450578117e-08, "loss": 2.2976, "step": 9780 }, { "epoch": 1.2058, "grad_norm": 25.9224707454181, "learning_rate": 1.1670572877477416e-08, "loss": 2.3473, "step": 9790 }, { "epoch": 1.2068, "grad_norm": 27.07824570292251, "learning_rate": 1.0590952519630715e-08, "loss": 2.3584, "step": 9800 }, { "epoch": 1.2078, "grad_norm": 25.55332869553411, "learning_rate": 9.563668701744743e-09, "loss": 2.278, "step": 9810 }, { "epoch": 1.2088, "grad_norm": 23.62365215608855, "learning_rate": 8.588732199540373e-09, "loss": 2.2512, "step": 9820 }, { "epoch": 1.2098, "grad_norm": 29.092888772465923, "learning_rate": 7.666153239639929e-09, "loss": 2.3522, "step": 9830 }, { "epoch": 1.2107999999999999, "grad_norm": 26.611552812245982, "learning_rate": 6.795941499460057e-09, "loss": 2.2586, "step": 9840 }, { "epoch": 1.2118, "grad_norm": 25.817026903330063, "learning_rate": 5.9781061071106924e-09, "loss": 2.3364, "step": 9850 }, { "epoch": 1.2128, "grad_norm": 26.826754192765534, "learning_rate": 5.212655641297915e-09, "loss": 2.3343, "step": 9860 }, { "epoch": 1.2138, "grad_norm": 27.60028434410644, "learning_rate": 4.499598131234573e-09, "loss": 2.402, "step": 9870 }, { "epoch": 1.2147999999999999, "grad_norm": 25.238494475772402, "learning_rate": 3.838941056557022e-09, "loss": 2.2858, "step": 9880 }, { "epoch": 1.2158, "grad_norm": 26.24382882674059, "learning_rate": 3.230691347245185e-09, "loss": 2.2625, "step": 9890 }, { "epoch": 1.2168, "grad_norm": 24.43899713679754, "learning_rate": 2.67485538355039e-09, "loss": 2.3847, "step": 9900 }, { "epoch": 1.2178, "grad_norm": 25.331421290768578, "learning_rate": 2.171438995929309e-09, "loss": 2.3224, "step": 9910 }, { "epoch": 1.2187999999999999, "grad_norm": 25.68620467949693, "learning_rate": 1.7204474649812341e-09, "loss": 2.3186, "step": 9920 }, { "epoch": 1.2198, "grad_norm": 26.93000800287073, "learning_rate": 1.3218855213947834e-09, "loss": 2.2933, "step": 9930 }, { "epoch": 1.2208, "grad_norm": 26.733108107133468, "learning_rate": 9.757573458946124e-10, "loss": 2.3584, "step": 9940 }, { "epoch": 1.2218, "grad_norm": 26.756446717772384, "learning_rate": 6.820665692031104e-10, "loss": 2.2663, "step": 9950 }, { "epoch": 1.2227999999999999, "grad_norm": 28.400435640943517, "learning_rate": 4.40816271996547e-10, "loss": 2.3739, "step": 9960 }, { "epoch": 1.2238, "grad_norm": 26.212982137586962, "learning_rate": 2.520089848762064e-10, "loss": 2.3999, "step": 9970 }, { "epoch": 1.2248, "grad_norm": 24.851989544436243, "learning_rate": 1.1564668834229686e-10, "loss": 2.3618, "step": 9980 }, { "epoch": 1.2258, "grad_norm": 27.43314026734001, "learning_rate": 3.173081277008105e-11, "loss": 2.3472, "step": 9990 }, { "epoch": 1.2268, "grad_norm": 24.99759608887615, "learning_rate": 2.6223839710848297e-13, "loss": 2.3423, "step": 10000 }, { "epoch": 1.0001, "grad_norm": 25.029163225108956, "learning_rate": 9.871724819144172e-06, "loss": 2.1332, "step": 10010 }, { "epoch": 1.0002, "grad_norm": 30.718769446104776, "learning_rate": 9.87136010690795e-06, "loss": 2.4332, "step": 10020 }, { "epoch": 1.0003, "grad_norm": 28.346234037928227, "learning_rate": 9.87099488368866e-06, "loss": 2.5722, "step": 10030 }, { "epoch": 1.0004, "grad_norm": 25.84316414552352, "learning_rate": 9.870629149524615e-06, "loss": 2.6877, "step": 10040 }, { "epoch": 1.0005, "grad_norm": 24.952913644899414, "learning_rate": 9.870262904454175e-06, "loss": 2.6958, "step": 10050 }, { "epoch": 1.0006, "grad_norm": 26.64928324296097, "learning_rate": 9.869896148515763e-06, "loss": 2.6294, "step": 10060 }, { "epoch": 1.0007, "grad_norm": 25.03291460989249, "learning_rate": 9.869528881747845e-06, "loss": 2.5662, "step": 10070 }, { "epoch": 1.0008, "grad_norm": 24.290311348010782, "learning_rate": 9.86916110418895e-06, "loss": 2.4842, "step": 10080 }, { "epoch": 1.0009, "grad_norm": 25.400491250758385, "learning_rate": 9.86879281587765e-06, "loss": 2.5536, "step": 10090 }, { "epoch": 1.001, "grad_norm": 26.186162416729623, "learning_rate": 9.868424016852582e-06, "loss": 2.5972, "step": 10100 }, { "epoch": 1.0011, "grad_norm": 24.685865397740706, "learning_rate": 9.868054707152431e-06, "loss": 2.6177, "step": 10110 }, { "epoch": 1.0012, "grad_norm": 25.291609504557588, "learning_rate": 9.867684886815932e-06, "loss": 2.5605, "step": 10120 }, { "epoch": 1.0013, "grad_norm": 21.514417492331386, "learning_rate": 9.867314555881881e-06, "loss": 2.6097, "step": 10130 }, { "epoch": 1.0014, "grad_norm": 24.43877894315048, "learning_rate": 9.866943714389124e-06, "loss": 2.5513, "step": 10140 }, { "epoch": 1.0015, "grad_norm": 22.667378114167633, "learning_rate": 9.866572362376558e-06, "loss": 2.472, "step": 10150 }, { "epoch": 1.0016, "grad_norm": 25.136046631463888, "learning_rate": 9.866200499883138e-06, "loss": 2.5714, "step": 10160 }, { "epoch": 1.0017, "grad_norm": 23.275764810136586, "learning_rate": 9.865828126947872e-06, "loss": 2.5207, "step": 10170 }, { "epoch": 1.0018, "grad_norm": 25.740810020286546, "learning_rate": 9.865455243609817e-06, "loss": 2.5515, "step": 10180 }, { "epoch": 1.0019, "grad_norm": 24.40265418312963, "learning_rate": 9.86508184990809e-06, "loss": 2.4685, "step": 10190 }, { "epoch": 1.002, "grad_norm": 25.97313908720894, "learning_rate": 9.864707945881853e-06, "loss": 2.4921, "step": 10200 }, { "epoch": 1.0021, "grad_norm": 26.489299817977987, "learning_rate": 9.864333531570332e-06, "loss": 2.6013, "step": 10210 }, { "epoch": 1.0022, "grad_norm": 23.581789252060922, "learning_rate": 9.8639586070128e-06, "loss": 2.6379, "step": 10220 }, { "epoch": 1.0023, "grad_norm": 22.71729341671351, "learning_rate": 9.863583172248585e-06, "loss": 2.5677, "step": 10230 }, { "epoch": 1.0024, "grad_norm": 25.68389216433485, "learning_rate": 9.863207227317066e-06, "loss": 2.5199, "step": 10240 }, { "epoch": 1.0025, "grad_norm": 25.663973642330994, "learning_rate": 9.862830772257682e-06, "loss": 2.6112, "step": 10250 }, { "epoch": 1.0026, "grad_norm": 26.4869117164885, "learning_rate": 9.862453807109917e-06, "loss": 2.5606, "step": 10260 }, { "epoch": 1.0027, "grad_norm": 22.748440242824177, "learning_rate": 9.862076331913317e-06, "loss": 2.5222, "step": 10270 }, { "epoch": 1.0028, "grad_norm": 23.560791459544614, "learning_rate": 9.861698346707474e-06, "loss": 2.567, "step": 10280 }, { "epoch": 1.0029, "grad_norm": 23.389707668315943, "learning_rate": 9.86131985153204e-06, "loss": 2.5617, "step": 10290 }, { "epoch": 1.003, "grad_norm": 23.45162920063435, "learning_rate": 9.860940846426712e-06, "loss": 2.5002, "step": 10300 }, { "epoch": 1.0031, "grad_norm": 22.60439967162541, "learning_rate": 9.860561331431253e-06, "loss": 2.5739, "step": 10310 }, { "epoch": 1.0032, "grad_norm": 23.273048980592385, "learning_rate": 9.860181306585468e-06, "loss": 2.4667, "step": 10320 }, { "epoch": 1.0033, "grad_norm": 24.72133845166431, "learning_rate": 9.85980077192922e-06, "loss": 2.4624, "step": 10330 }, { "epoch": 1.0034, "grad_norm": 25.170695326269165, "learning_rate": 9.859419727502427e-06, "loss": 2.5394, "step": 10340 }, { "epoch": 1.0035, "grad_norm": 22.95645190151738, "learning_rate": 9.859038173345056e-06, "loss": 2.565, "step": 10350 }, { "epoch": 1.0036, "grad_norm": 23.667901990028536, "learning_rate": 9.858656109497134e-06, "loss": 2.6625, "step": 10360 }, { "epoch": 1.0037, "grad_norm": 22.959570981525374, "learning_rate": 9.858273535998735e-06, "loss": 2.5798, "step": 10370 }, { "epoch": 1.0038, "grad_norm": 23.966877688662525, "learning_rate": 9.857890452889989e-06, "loss": 2.5456, "step": 10380 }, { "epoch": 1.0039, "grad_norm": 21.940217870533285, "learning_rate": 9.857506860211084e-06, "loss": 2.453, "step": 10390 }, { "epoch": 1.004, "grad_norm": 24.045198681707664, "learning_rate": 9.85712275800225e-06, "loss": 2.5093, "step": 10400 }, { "epoch": 1.0041, "grad_norm": 26.642279136535183, "learning_rate": 9.856738146303783e-06, "loss": 2.5848, "step": 10410 }, { "epoch": 1.0042, "grad_norm": 23.139492832273575, "learning_rate": 9.856353025156026e-06, "loss": 2.5367, "step": 10420 }, { "epoch": 1.0043, "grad_norm": 23.386689673055304, "learning_rate": 9.855967394599375e-06, "loss": 2.4797, "step": 10430 }, { "epoch": 1.0044, "grad_norm": 23.25256150439783, "learning_rate": 9.855581254674282e-06, "loss": 2.4931, "step": 10440 }, { "epoch": 1.0045, "grad_norm": 23.76833633511625, "learning_rate": 9.85519460542125e-06, "loss": 2.4304, "step": 10450 }, { "epoch": 1.0046, "grad_norm": 23.913019118419584, "learning_rate": 9.854807446880837e-06, "loss": 2.5871, "step": 10460 }, { "epoch": 1.0047, "grad_norm": 23.857124700606267, "learning_rate": 9.854419779093656e-06, "loss": 2.5893, "step": 10470 }, { "epoch": 1.0048, "grad_norm": 21.90192966692284, "learning_rate": 9.85403160210037e-06, "loss": 2.6484, "step": 10480 }, { "epoch": 1.0049, "grad_norm": 22.759068442836135, "learning_rate": 9.853642915941696e-06, "loss": 2.5049, "step": 10490 }, { "epoch": 1.005, "grad_norm": 23.20623315910518, "learning_rate": 9.853253720658407e-06, "loss": 2.5392, "step": 10500 }, { "epoch": 1.0051, "grad_norm": 23.045094306346748, "learning_rate": 9.852864016291329e-06, "loss": 2.5032, "step": 10510 }, { "epoch": 1.0052, "grad_norm": 25.079658136574093, "learning_rate": 9.852473802881337e-06, "loss": 2.5151, "step": 10520 }, { "epoch": 1.0053, "grad_norm": 24.584797257273454, "learning_rate": 9.852083080469365e-06, "loss": 2.5889, "step": 10530 }, { "epoch": 1.0054, "grad_norm": 21.24094324880013, "learning_rate": 9.851691849096397e-06, "loss": 2.4575, "step": 10540 }, { "epoch": 1.0055, "grad_norm": 22.73192520633209, "learning_rate": 9.85130010880347e-06, "loss": 2.4907, "step": 10550 }, { "epoch": 1.0056, "grad_norm": 24.927514492768374, "learning_rate": 9.850907859631678e-06, "loss": 2.5335, "step": 10560 }, { "epoch": 1.0057, "grad_norm": 21.11424601453074, "learning_rate": 9.850515101622166e-06, "loss": 2.5668, "step": 10570 }, { "epoch": 1.0058, "grad_norm": 21.936670271052876, "learning_rate": 9.850121834816132e-06, "loss": 2.4058, "step": 10580 }, { "epoch": 1.0059, "grad_norm": 23.588158696582457, "learning_rate": 9.849728059254827e-06, "loss": 2.5335, "step": 10590 }, { "epoch": 1.006, "grad_norm": 24.366770751107214, "learning_rate": 9.849333774979555e-06, "loss": 2.5042, "step": 10600 }, { "epoch": 1.0061, "grad_norm": 21.433318692852822, "learning_rate": 9.84893898203168e-06, "loss": 2.4618, "step": 10610 }, { "epoch": 1.0062, "grad_norm": 20.834505131776325, "learning_rate": 9.848543680452607e-06, "loss": 2.4881, "step": 10620 }, { "epoch": 1.0063, "grad_norm": 23.027566201793864, "learning_rate": 9.848147870283807e-06, "loss": 2.5101, "step": 10630 }, { "epoch": 1.0064, "grad_norm": 21.749016748869266, "learning_rate": 9.847751551566795e-06, "loss": 2.489, "step": 10640 }, { "epoch": 1.0065, "grad_norm": 23.09511868321867, "learning_rate": 9.847354724343146e-06, "loss": 2.4602, "step": 10650 }, { "epoch": 1.0066, "grad_norm": 25.420326740376588, "learning_rate": 9.846957388654483e-06, "loss": 2.4481, "step": 10660 }, { "epoch": 1.0067, "grad_norm": 24.34712528706796, "learning_rate": 9.846559544542484e-06, "loss": 2.5433, "step": 10670 }, { "epoch": 1.0068, "grad_norm": 24.20763053557463, "learning_rate": 9.846161192048884e-06, "loss": 2.4956, "step": 10680 }, { "epoch": 1.0069, "grad_norm": 23.977913280463586, "learning_rate": 9.845762331215465e-06, "loss": 2.4841, "step": 10690 }, { "epoch": 1.007, "grad_norm": 24.742698050741836, "learning_rate": 9.845362962084069e-06, "loss": 2.5892, "step": 10700 }, { "epoch": 1.0071, "grad_norm": 23.77416282193363, "learning_rate": 9.844963084696586e-06, "loss": 2.5373, "step": 10710 }, { "epoch": 1.0072, "grad_norm": 22.069999657193183, "learning_rate": 9.844562699094962e-06, "loss": 2.6089, "step": 10720 }, { "epoch": 1.0073, "grad_norm": 22.749570374366055, "learning_rate": 9.844161805321192e-06, "loss": 2.5395, "step": 10730 }, { "epoch": 1.0074, "grad_norm": 24.36971996622367, "learning_rate": 9.843760403417335e-06, "loss": 2.5775, "step": 10740 }, { "epoch": 1.0075, "grad_norm": 26.113706230465638, "learning_rate": 9.84335849342549e-06, "loss": 2.47, "step": 10750 }, { "epoch": 1.0076, "grad_norm": 22.28366988018183, "learning_rate": 9.84295607538782e-06, "loss": 2.5177, "step": 10760 }, { "epoch": 1.0077, "grad_norm": 22.551982437101962, "learning_rate": 9.842553149346533e-06, "loss": 2.5169, "step": 10770 }, { "epoch": 1.0078, "grad_norm": 22.006094202848256, "learning_rate": 9.842149715343896e-06, "loss": 2.4745, "step": 10780 }, { "epoch": 1.0079, "grad_norm": 22.168597898063314, "learning_rate": 9.841745773422227e-06, "loss": 2.4636, "step": 10790 }, { "epoch": 1.008, "grad_norm": 22.116618537652784, "learning_rate": 9.841341323623897e-06, "loss": 2.5233, "step": 10800 }, { "epoch": 1.0081, "grad_norm": 21.439797174520344, "learning_rate": 9.840936365991332e-06, "loss": 2.4679, "step": 10810 }, { "epoch": 1.0082, "grad_norm": 23.555498891221863, "learning_rate": 9.840530900567009e-06, "loss": 2.56, "step": 10820 }, { "epoch": 1.0083, "grad_norm": 24.40366103924378, "learning_rate": 9.84012492739346e-06, "loss": 2.5452, "step": 10830 }, { "epoch": 1.0084, "grad_norm": 23.250235382246867, "learning_rate": 9.839718446513272e-06, "loss": 2.5326, "step": 10840 }, { "epoch": 1.0085, "grad_norm": 21.891336210277547, "learning_rate": 9.839311457969079e-06, "loss": 2.5161, "step": 10850 }, { "epoch": 1.0086, "grad_norm": 25.020017455286336, "learning_rate": 9.838903961803574e-06, "loss": 2.5271, "step": 10860 }, { "epoch": 1.0087, "grad_norm": 21.940915803439076, "learning_rate": 9.8384959580595e-06, "loss": 2.5279, "step": 10870 }, { "epoch": 1.0088, "grad_norm": 22.337258547647394, "learning_rate": 9.838087446779659e-06, "loss": 2.5592, "step": 10880 }, { "epoch": 1.0089, "grad_norm": 20.060519289587152, "learning_rate": 9.837678428006899e-06, "loss": 2.5995, "step": 10890 }, { "epoch": 1.009, "grad_norm": 22.04523897159908, "learning_rate": 9.837268901784123e-06, "loss": 2.477, "step": 10900 }, { "epoch": 1.0091, "grad_norm": 26.174715383176366, "learning_rate": 9.836858868154289e-06, "loss": 2.5919, "step": 10910 }, { "epoch": 1.0092, "grad_norm": 23.264773640208958, "learning_rate": 9.836448327160409e-06, "loss": 2.4005, "step": 10920 }, { "epoch": 1.0093, "grad_norm": 22.09752622950856, "learning_rate": 9.836037278845547e-06, "loss": 2.54, "step": 10930 }, { "epoch": 1.0094, "grad_norm": 23.105448004386588, "learning_rate": 9.835625723252817e-06, "loss": 2.4615, "step": 10940 }, { "epoch": 1.0095, "grad_norm": 24.972289468767666, "learning_rate": 9.835213660425393e-06, "loss": 2.5077, "step": 10950 }, { "epoch": 1.0096, "grad_norm": 21.44106683365171, "learning_rate": 9.834801090406499e-06, "loss": 2.5615, "step": 10960 }, { "epoch": 1.0097, "grad_norm": 22.99964145179965, "learning_rate": 9.834388013239407e-06, "loss": 2.5555, "step": 10970 }, { "epoch": 1.0098, "grad_norm": 22.959538460701282, "learning_rate": 9.833974428967449e-06, "loss": 2.4989, "step": 10980 }, { "epoch": 1.0099, "grad_norm": 24.076881320413218, "learning_rate": 9.83356033763401e-06, "loss": 2.5446, "step": 10990 }, { "epoch": 1.01, "grad_norm": 21.663257129069386, "learning_rate": 9.833145739282524e-06, "loss": 2.5848, "step": 11000 }, { "epoch": 1.0101, "grad_norm": 21.500128440571377, "learning_rate": 9.832730633956482e-06, "loss": 2.3937, "step": 11010 }, { "epoch": 1.0102, "grad_norm": 20.634983765630437, "learning_rate": 9.832315021699426e-06, "loss": 2.4751, "step": 11020 }, { "epoch": 1.0103, "grad_norm": 22.9881263403536, "learning_rate": 9.83189890255495e-06, "loss": 2.4087, "step": 11030 }, { "epoch": 1.0104, "grad_norm": 24.428059277189348, "learning_rate": 9.831482276566706e-06, "loss": 2.4901, "step": 11040 }, { "epoch": 1.0105, "grad_norm": 22.598080241395163, "learning_rate": 9.831065143778395e-06, "loss": 2.5368, "step": 11050 }, { "epoch": 1.0106, "grad_norm": 21.98003062510308, "learning_rate": 9.830647504233773e-06, "loss": 2.5922, "step": 11060 }, { "epoch": 1.0107, "grad_norm": 21.86573359415885, "learning_rate": 9.830229357976646e-06, "loss": 2.5099, "step": 11070 }, { "epoch": 1.0108, "grad_norm": 24.501592615652516, "learning_rate": 9.829810705050877e-06, "loss": 2.4782, "step": 11080 }, { "epoch": 1.0109, "grad_norm": 21.494250222626764, "learning_rate": 9.829391545500382e-06, "loss": 2.4879, "step": 11090 }, { "epoch": 1.011, "grad_norm": 23.60943477888212, "learning_rate": 9.828971879369127e-06, "loss": 2.4803, "step": 11100 }, { "epoch": 1.0111, "grad_norm": 23.78495256277646, "learning_rate": 9.828551706701133e-06, "loss": 2.4565, "step": 11110 }, { "epoch": 1.0112, "grad_norm": 21.818078089044985, "learning_rate": 9.828131027540476e-06, "loss": 2.4358, "step": 11120 }, { "epoch": 1.0113, "grad_norm": 21.58924404517791, "learning_rate": 9.827709841931282e-06, "loss": 2.4082, "step": 11130 }, { "epoch": 1.0114, "grad_norm": 23.966973764187113, "learning_rate": 9.827288149917732e-06, "loss": 2.4361, "step": 11140 }, { "epoch": 1.0115, "grad_norm": 21.249401836719077, "learning_rate": 9.826865951544058e-06, "loss": 2.5517, "step": 11150 }, { "epoch": 1.0116, "grad_norm": 20.585164035682237, "learning_rate": 9.82644324685455e-06, "loss": 2.4635, "step": 11160 }, { "epoch": 1.0117, "grad_norm": 21.055675085723056, "learning_rate": 9.826020035893545e-06, "loss": 2.5217, "step": 11170 }, { "epoch": 1.0118, "grad_norm": 21.58581543709135, "learning_rate": 9.825596318705435e-06, "loss": 2.4044, "step": 11180 }, { "epoch": 1.0119, "grad_norm": 22.31234954975908, "learning_rate": 9.82517209533467e-06, "loss": 2.4603, "step": 11190 }, { "epoch": 1.012, "grad_norm": 19.660900780553927, "learning_rate": 9.824747365825744e-06, "loss": 2.4388, "step": 11200 }, { "epoch": 1.0121, "grad_norm": 20.927118414926483, "learning_rate": 9.824322130223213e-06, "loss": 2.5618, "step": 11210 }, { "epoch": 1.0122, "grad_norm": 20.649350120739737, "learning_rate": 9.82389638857168e-06, "loss": 2.4227, "step": 11220 }, { "epoch": 1.0123, "grad_norm": 21.147234249384066, "learning_rate": 9.823470140915806e-06, "loss": 2.462, "step": 11230 }, { "epoch": 1.0124, "grad_norm": 23.68637167912905, "learning_rate": 9.8230433873003e-06, "loss": 2.5055, "step": 11240 }, { "epoch": 1.0125, "grad_norm": 22.42997992085665, "learning_rate": 9.822616127769925e-06, "loss": 2.4873, "step": 11250 }, { "epoch": 1.0126, "grad_norm": 21.861540558887476, "learning_rate": 9.822188362369502e-06, "loss": 2.4236, "step": 11260 }, { "epoch": 1.0127, "grad_norm": 21.61733268243067, "learning_rate": 9.8217600911439e-06, "loss": 2.4798, "step": 11270 }, { "epoch": 1.0128, "grad_norm": 21.054824424521243, "learning_rate": 9.821331314138044e-06, "loss": 2.4319, "step": 11280 }, { "epoch": 1.0129, "grad_norm": 22.565559048406342, "learning_rate": 9.82090203139691e-06, "loss": 2.5247, "step": 11290 }, { "epoch": 1.013, "grad_norm": 22.41896375120415, "learning_rate": 9.820472242965525e-06, "loss": 2.5352, "step": 11300 }, { "epoch": 1.0131000000000001, "grad_norm": 20.679126235949283, "learning_rate": 9.820041948888976e-06, "loss": 2.4583, "step": 11310 }, { "epoch": 1.0132, "grad_norm": 22.785496145198714, "learning_rate": 9.819611149212397e-06, "loss": 2.3958, "step": 11320 }, { "epoch": 1.0133, "grad_norm": 22.51971644029265, "learning_rate": 9.819179843980976e-06, "loss": 2.4531, "step": 11330 }, { "epoch": 1.0134, "grad_norm": 20.466467103067448, "learning_rate": 9.818748033239957e-06, "loss": 2.3584, "step": 11340 }, { "epoch": 1.0135, "grad_norm": 20.990652697670438, "learning_rate": 9.818315717034633e-06, "loss": 2.3871, "step": 11350 }, { "epoch": 1.0136, "grad_norm": 22.275104773816476, "learning_rate": 9.817882895410354e-06, "loss": 2.4063, "step": 11360 }, { "epoch": 1.0137, "grad_norm": 23.725084967713965, "learning_rate": 9.817449568412517e-06, "loss": 2.4621, "step": 11370 }, { "epoch": 1.0138, "grad_norm": 22.488231686450405, "learning_rate": 9.817015736086582e-06, "loss": 2.5689, "step": 11380 }, { "epoch": 1.0139, "grad_norm": 19.71513024563764, "learning_rate": 9.81658139847805e-06, "loss": 2.4568, "step": 11390 }, { "epoch": 1.014, "grad_norm": 20.98383182119441, "learning_rate": 9.816146555632487e-06, "loss": 2.4493, "step": 11400 }, { "epoch": 1.0141, "grad_norm": 21.002054159678778, "learning_rate": 9.8157112075955e-06, "loss": 2.3662, "step": 11410 }, { "epoch": 1.0142, "grad_norm": 23.72650572427314, "learning_rate": 9.815275354412758e-06, "loss": 2.4807, "step": 11420 }, { "epoch": 1.0143, "grad_norm": 21.92325133098789, "learning_rate": 9.81483899612998e-06, "loss": 2.4296, "step": 11430 }, { "epoch": 1.0144, "grad_norm": 22.795784897099516, "learning_rate": 9.814402132792939e-06, "loss": 2.4555, "step": 11440 }, { "epoch": 1.0145, "grad_norm": 22.341454331961977, "learning_rate": 9.813964764447456e-06, "loss": 2.5082, "step": 11450 }, { "epoch": 1.0146, "grad_norm": 19.868791074782056, "learning_rate": 9.813526891139414e-06, "loss": 2.4669, "step": 11460 }, { "epoch": 1.0147, "grad_norm": 20.73294504779266, "learning_rate": 9.813088512914741e-06, "loss": 2.4284, "step": 11470 }, { "epoch": 1.0148, "grad_norm": 21.15866822814226, "learning_rate": 9.812649629819419e-06, "loss": 2.4723, "step": 11480 }, { "epoch": 1.0149, "grad_norm": 19.802833091086082, "learning_rate": 9.81221024189949e-06, "loss": 2.4724, "step": 11490 }, { "epoch": 1.015, "grad_norm": 20.742664358912528, "learning_rate": 9.81177034920104e-06, "loss": 2.4716, "step": 11500 }, { "epoch": 1.0151, "grad_norm": 21.952502874230607, "learning_rate": 9.811329951770212e-06, "loss": 2.4522, "step": 11510 }, { "epoch": 1.0152, "grad_norm": 22.904972479484595, "learning_rate": 9.810889049653201e-06, "loss": 2.4646, "step": 11520 }, { "epoch": 1.0153, "grad_norm": 21.567734422171878, "learning_rate": 9.810447642896259e-06, "loss": 2.4797, "step": 11530 }, { "epoch": 1.0154, "grad_norm": 23.225528730843273, "learning_rate": 9.810005731545685e-06, "loss": 2.5394, "step": 11540 }, { "epoch": 1.0155, "grad_norm": 22.061701429274343, "learning_rate": 9.809563315647831e-06, "loss": 2.4762, "step": 11550 }, { "epoch": 1.0156, "grad_norm": 21.20204996806157, "learning_rate": 9.80912039524911e-06, "loss": 2.4729, "step": 11560 }, { "epoch": 1.0157, "grad_norm": 23.011370255782605, "learning_rate": 9.808676970395977e-06, "loss": 2.4936, "step": 11570 }, { "epoch": 1.0158, "grad_norm": 24.76874675847236, "learning_rate": 9.80823304113495e-06, "loss": 2.4366, "step": 11580 }, { "epoch": 1.0159, "grad_norm": 24.405975700044607, "learning_rate": 9.807788607512589e-06, "loss": 2.5539, "step": 11590 }, { "epoch": 1.016, "grad_norm": 21.782791234963142, "learning_rate": 9.807343669575519e-06, "loss": 2.4353, "step": 11600 }, { "epoch": 1.0161, "grad_norm": 22.32850196485686, "learning_rate": 9.806898227370411e-06, "loss": 2.4462, "step": 11610 }, { "epoch": 1.0162, "grad_norm": 20.76166310230141, "learning_rate": 9.806452280943988e-06, "loss": 2.4558, "step": 11620 }, { "epoch": 1.0163, "grad_norm": 21.73077540289691, "learning_rate": 9.806005830343026e-06, "loss": 2.4604, "step": 11630 }, { "epoch": 1.0164, "grad_norm": 21.982597255461346, "learning_rate": 9.805558875614361e-06, "loss": 2.4569, "step": 11640 }, { "epoch": 1.0165, "grad_norm": 22.00012817424087, "learning_rate": 9.80511141680487e-06, "loss": 2.4709, "step": 11650 }, { "epoch": 1.0166, "grad_norm": 22.44581987787833, "learning_rate": 9.804663453961494e-06, "loss": 2.4737, "step": 11660 }, { "epoch": 1.0167, "grad_norm": 21.943005187314007, "learning_rate": 9.804214987131221e-06, "loss": 2.4535, "step": 11670 }, { "epoch": 1.0168, "grad_norm": 23.2939074090717, "learning_rate": 9.803766016361093e-06, "loss": 2.5774, "step": 11680 }, { "epoch": 1.0169, "grad_norm": 23.872416149371208, "learning_rate": 9.803316541698204e-06, "loss": 2.451, "step": 11690 }, { "epoch": 1.017, "grad_norm": 22.283568013514316, "learning_rate": 9.802866563189704e-06, "loss": 2.4289, "step": 11700 }, { "epoch": 1.0171, "grad_norm": 20.68970808235879, "learning_rate": 9.80241608088279e-06, "loss": 2.4652, "step": 11710 }, { "epoch": 1.0172, "grad_norm": 21.06930197673156, "learning_rate": 9.80196509482472e-06, "loss": 2.425, "step": 11720 }, { "epoch": 1.0173, "grad_norm": 22.59159691052905, "learning_rate": 9.801513605062798e-06, "loss": 2.4379, "step": 11730 }, { "epoch": 1.0174, "grad_norm": 22.718877108154032, "learning_rate": 9.801061611644382e-06, "loss": 2.5381, "step": 11740 }, { "epoch": 1.0175, "grad_norm": 22.25909803728629, "learning_rate": 9.800609114616885e-06, "loss": 2.4121, "step": 11750 }, { "epoch": 1.0176, "grad_norm": 22.011316207150173, "learning_rate": 9.800156114027775e-06, "loss": 2.4338, "step": 11760 }, { "epoch": 1.0177, "grad_norm": 21.683431974891356, "learning_rate": 9.799702609924564e-06, "loss": 2.4406, "step": 11770 }, { "epoch": 1.0178, "grad_norm": 28.73125735129282, "learning_rate": 9.799248602354826e-06, "loss": 2.4259, "step": 11780 }, { "epoch": 1.0179, "grad_norm": 21.806196059880207, "learning_rate": 9.798794091366187e-06, "loss": 2.4909, "step": 11790 }, { "epoch": 1.018, "grad_norm": 24.55819224446337, "learning_rate": 9.798339077006316e-06, "loss": 2.4892, "step": 11800 }, { "epoch": 1.0181, "grad_norm": 21.66145264201029, "learning_rate": 9.797883559322947e-06, "loss": 2.4864, "step": 11810 }, { "epoch": 1.0182, "grad_norm": 21.674247575106737, "learning_rate": 9.79742753836386e-06, "loss": 2.458, "step": 11820 }, { "epoch": 1.0183, "grad_norm": 20.320948780281654, "learning_rate": 9.796971014176889e-06, "loss": 2.4918, "step": 11830 }, { "epoch": 1.0184, "grad_norm": 20.305658507012414, "learning_rate": 9.796513986809922e-06, "loss": 2.5147, "step": 11840 }, { "epoch": 1.0185, "grad_norm": 22.139123899151876, "learning_rate": 9.7960564563109e-06, "loss": 2.5226, "step": 11850 }, { "epoch": 1.0186, "grad_norm": 21.692579930914384, "learning_rate": 9.795598422727814e-06, "loss": 2.453, "step": 11860 }, { "epoch": 1.0187, "grad_norm": 21.093386260521456, "learning_rate": 9.79513988610871e-06, "loss": 2.4625, "step": 11870 }, { "epoch": 1.0188, "grad_norm": 21.138551538631944, "learning_rate": 9.79468084650169e-06, "loss": 2.3337, "step": 11880 }, { "epoch": 1.0189, "grad_norm": 20.226250623191547, "learning_rate": 9.794221303954898e-06, "loss": 2.4776, "step": 11890 }, { "epoch": 1.019, "grad_norm": 21.593158566003144, "learning_rate": 9.793761258516545e-06, "loss": 2.4159, "step": 11900 }, { "epoch": 1.0191, "grad_norm": 22.94798734777286, "learning_rate": 9.793300710234883e-06, "loss": 2.4017, "step": 11910 }, { "epoch": 1.0192, "grad_norm": 22.69888990506391, "learning_rate": 9.792839659158223e-06, "loss": 2.4788, "step": 11920 }, { "epoch": 1.0193, "grad_norm": 22.710397207765492, "learning_rate": 9.792378105334928e-06, "loss": 2.434, "step": 11930 }, { "epoch": 1.0194, "grad_norm": 21.39378062733835, "learning_rate": 9.79191604881341e-06, "loss": 2.4534, "step": 11940 }, { "epoch": 1.0195, "grad_norm": 21.950375286260012, "learning_rate": 9.79145348964214e-06, "loss": 2.4099, "step": 11950 }, { "epoch": 1.0196, "grad_norm": 21.38878023139135, "learning_rate": 9.790990427869636e-06, "loss": 2.4854, "step": 11960 }, { "epoch": 1.0197, "grad_norm": 23.894030241434724, "learning_rate": 9.790526863544472e-06, "loss": 2.423, "step": 11970 }, { "epoch": 1.0198, "grad_norm": 21.389449368786888, "learning_rate": 9.790062796715274e-06, "loss": 2.3733, "step": 11980 }, { "epoch": 1.0199, "grad_norm": 21.03837415012722, "learning_rate": 9.78959822743072e-06, "loss": 2.4041, "step": 11990 }, { "epoch": 1.02, "grad_norm": 21.204385065061768, "learning_rate": 9.789133155739541e-06, "loss": 2.4501, "step": 12000 }, { "epoch": 1.0201, "grad_norm": 23.764260473167656, "learning_rate": 9.788667581690522e-06, "loss": 2.3892, "step": 12010 }, { "epoch": 1.0202, "grad_norm": 20.352413184282934, "learning_rate": 9.788201505332496e-06, "loss": 2.3536, "step": 12020 }, { "epoch": 1.0203, "grad_norm": 22.33288422976887, "learning_rate": 9.787734926714356e-06, "loss": 2.3866, "step": 12030 }, { "epoch": 1.0204, "grad_norm": 22.44077635233536, "learning_rate": 9.787267845885043e-06, "loss": 2.4239, "step": 12040 }, { "epoch": 1.0205, "grad_norm": 20.713502334364563, "learning_rate": 9.786800262893552e-06, "loss": 2.4498, "step": 12050 }, { "epoch": 1.0206, "grad_norm": 23.86383986302549, "learning_rate": 9.786332177788928e-06, "loss": 2.4475, "step": 12060 }, { "epoch": 1.0207, "grad_norm": 20.553140844981353, "learning_rate": 9.785863590620275e-06, "loss": 2.5149, "step": 12070 }, { "epoch": 1.0208, "grad_norm": 22.17252133403497, "learning_rate": 9.785394501436742e-06, "loss": 2.4182, "step": 12080 }, { "epoch": 1.0209, "grad_norm": 22.260875694763598, "learning_rate": 9.784924910287534e-06, "loss": 2.3814, "step": 12090 }, { "epoch": 1.021, "grad_norm": 24.061041649471587, "learning_rate": 9.784454817221912e-06, "loss": 2.4528, "step": 12100 }, { "epoch": 1.0211, "grad_norm": 20.887826357420835, "learning_rate": 9.783984222289185e-06, "loss": 2.3695, "step": 12110 }, { "epoch": 1.0212, "grad_norm": 21.411019486972876, "learning_rate": 9.783513125538717e-06, "loss": 2.4797, "step": 12120 }, { "epoch": 1.0213, "grad_norm": 22.57946764809226, "learning_rate": 9.78304152701992e-06, "loss": 2.4754, "step": 12130 }, { "epoch": 1.0214, "grad_norm": 22.337487270504287, "learning_rate": 9.782569426782269e-06, "loss": 2.4634, "step": 12140 }, { "epoch": 1.0215, "grad_norm": 21.86453471086327, "learning_rate": 9.78209682487528e-06, "loss": 2.4468, "step": 12150 }, { "epoch": 1.0216, "grad_norm": 23.103545024584513, "learning_rate": 9.781623721348529e-06, "loss": 2.3619, "step": 12160 }, { "epoch": 1.0217, "grad_norm": 20.085232149063977, "learning_rate": 9.78115011625164e-06, "loss": 2.5498, "step": 12170 }, { "epoch": 1.0218, "grad_norm": 20.11090911901453, "learning_rate": 9.780676009634297e-06, "loss": 2.5355, "step": 12180 }, { "epoch": 1.0219, "grad_norm": 21.26472908441453, "learning_rate": 9.780201401546227e-06, "loss": 2.404, "step": 12190 }, { "epoch": 1.022, "grad_norm": 21.997319343100934, "learning_rate": 9.779726292037214e-06, "loss": 2.4418, "step": 12200 }, { "epoch": 1.0221, "grad_norm": 20.996431148806508, "learning_rate": 9.779250681157099e-06, "loss": 2.4157, "step": 12210 }, { "epoch": 1.0222, "grad_norm": 21.752921208047145, "learning_rate": 9.778774568955769e-06, "loss": 2.424, "step": 12220 }, { "epoch": 1.0223, "grad_norm": 25.265084621503465, "learning_rate": 9.778297955483164e-06, "loss": 2.448, "step": 12230 }, { "epoch": 1.0224, "grad_norm": 21.401316001491548, "learning_rate": 9.77782084078928e-06, "loss": 2.5449, "step": 12240 }, { "epoch": 1.0225, "grad_norm": 20.13687431005082, "learning_rate": 9.777343224924164e-06, "loss": 2.411, "step": 12250 }, { "epoch": 1.0226, "grad_norm": 20.778913613870206, "learning_rate": 9.776865107937919e-06, "loss": 2.443, "step": 12260 }, { "epoch": 1.0227, "grad_norm": 22.05075262316107, "learning_rate": 9.776386489880692e-06, "loss": 2.3805, "step": 12270 }, { "epoch": 1.0228, "grad_norm": 21.838434692426404, "learning_rate": 9.77590737080269e-06, "loss": 2.4874, "step": 12280 }, { "epoch": 1.0229, "grad_norm": 22.056576038273107, "learning_rate": 9.77542775075417e-06, "loss": 2.424, "step": 12290 }, { "epoch": 1.023, "grad_norm": 21.59052487865488, "learning_rate": 9.774947629785443e-06, "loss": 2.5205, "step": 12300 }, { "epoch": 1.0231, "grad_norm": 19.606706358903907, "learning_rate": 9.77446700794687e-06, "loss": 2.5026, "step": 12310 }, { "epoch": 1.0232, "grad_norm": 19.75251841114292, "learning_rate": 9.773985885288869e-06, "loss": 2.5372, "step": 12320 }, { "epoch": 1.0233, "grad_norm": 20.383773801152355, "learning_rate": 9.773504261861904e-06, "loss": 2.5057, "step": 12330 }, { "epoch": 1.0234, "grad_norm": 22.35858594142115, "learning_rate": 9.773022137716496e-06, "loss": 2.5047, "step": 12340 }, { "epoch": 1.0235, "grad_norm": 19.375921199415107, "learning_rate": 9.77253951290322e-06, "loss": 2.4555, "step": 12350 }, { "epoch": 1.0236, "grad_norm": 20.998684685919287, "learning_rate": 9.772056387472698e-06, "loss": 2.5258, "step": 12360 }, { "epoch": 1.0237, "grad_norm": 20.758975460183336, "learning_rate": 9.771572761475608e-06, "loss": 2.4802, "step": 12370 }, { "epoch": 1.0238, "grad_norm": 23.023973088202602, "learning_rate": 9.771088634962681e-06, "loss": 2.469, "step": 12380 }, { "epoch": 1.0239, "grad_norm": 20.92525321200045, "learning_rate": 9.7706040079847e-06, "loss": 2.4116, "step": 12390 }, { "epoch": 1.024, "grad_norm": 21.94660371141383, "learning_rate": 9.7701188805925e-06, "loss": 2.4336, "step": 12400 }, { "epoch": 1.0241, "grad_norm": 21.13436243733965, "learning_rate": 9.769633252836969e-06, "loss": 2.3863, "step": 12410 }, { "epoch": 1.0242, "grad_norm": 23.094600053792462, "learning_rate": 9.769147124769045e-06, "loss": 2.3762, "step": 12420 }, { "epoch": 1.0243, "grad_norm": 23.683320693921736, "learning_rate": 9.768660496439721e-06, "loss": 2.4332, "step": 12430 }, { "epoch": 1.0244, "grad_norm": 20.38479244898512, "learning_rate": 9.768173367900045e-06, "loss": 2.3684, "step": 12440 }, { "epoch": 1.0245, "grad_norm": 20.644295300154376, "learning_rate": 9.767685739201112e-06, "loss": 2.3542, "step": 12450 }, { "epoch": 1.0246, "grad_norm": 21.659694389361057, "learning_rate": 9.767197610394072e-06, "loss": 2.4375, "step": 12460 }, { "epoch": 1.0247, "grad_norm": 20.580013741460213, "learning_rate": 9.766708981530127e-06, "loss": 2.4492, "step": 12470 }, { "epoch": 1.0248, "grad_norm": 22.442860885519046, "learning_rate": 9.766219852660534e-06, "loss": 2.4561, "step": 12480 }, { "epoch": 1.0249, "grad_norm": 20.195137801273116, "learning_rate": 9.765730223836599e-06, "loss": 2.3956, "step": 12490 }, { "epoch": 1.025, "grad_norm": 19.556715662216522, "learning_rate": 9.765240095109683e-06, "loss": 2.3649, "step": 12500 }, { "epoch": 1.0251, "grad_norm": 21.55679165794895, "learning_rate": 9.764749466531196e-06, "loss": 2.4281, "step": 12510 }, { "epoch": 1.0252, "grad_norm": 20.22508091667548, "learning_rate": 9.764258338152604e-06, "loss": 2.3673, "step": 12520 }, { "epoch": 1.0253, "grad_norm": 20.94988855944023, "learning_rate": 9.763766710025422e-06, "loss": 2.3344, "step": 12530 }, { "epoch": 1.0254, "grad_norm": 21.445544434748555, "learning_rate": 9.763274582201224e-06, "loss": 2.4594, "step": 12540 }, { "epoch": 1.0255, "grad_norm": 22.02940251984698, "learning_rate": 9.762781954731628e-06, "loss": 2.457, "step": 12550 }, { "epoch": 1.0256, "grad_norm": 19.312624209102918, "learning_rate": 9.76228882766831e-06, "loss": 2.376, "step": 12560 }, { "epoch": 1.0257, "grad_norm": 19.60513151336716, "learning_rate": 9.761795201062997e-06, "loss": 2.4008, "step": 12570 }, { "epoch": 1.0258, "grad_norm": 21.33412349605374, "learning_rate": 9.761301074967468e-06, "loss": 2.3663, "step": 12580 }, { "epoch": 1.0259, "grad_norm": 21.13365495721114, "learning_rate": 9.760806449433554e-06, "loss": 2.3589, "step": 12590 }, { "epoch": 1.026, "grad_norm": 21.248868628009113, "learning_rate": 9.760311324513139e-06, "loss": 2.3143, "step": 12600 }, { "epoch": 1.0261, "grad_norm": 21.27723402483877, "learning_rate": 9.759815700258159e-06, "loss": 2.3428, "step": 12610 }, { "epoch": 1.0262, "grad_norm": 21.6742774016493, "learning_rate": 9.759319576720602e-06, "loss": 2.3576, "step": 12620 }, { "epoch": 1.0263, "grad_norm": 20.345908649218707, "learning_rate": 9.75882295395251e-06, "loss": 2.3542, "step": 12630 }, { "epoch": 1.0264, "grad_norm": 22.21398833841648, "learning_rate": 9.758325832005979e-06, "loss": 2.3793, "step": 12640 }, { "epoch": 1.0265, "grad_norm": 19.59608106364157, "learning_rate": 9.75782821093315e-06, "loss": 2.4226, "step": 12650 }, { "epoch": 1.0266, "grad_norm": 19.671618458532595, "learning_rate": 9.757330090786226e-06, "loss": 2.4481, "step": 12660 }, { "epoch": 1.0267, "grad_norm": 20.879279499511785, "learning_rate": 9.756831471617451e-06, "loss": 2.2755, "step": 12670 }, { "epoch": 1.0268, "grad_norm": 20.97952883834971, "learning_rate": 9.756332353479134e-06, "loss": 2.3996, "step": 12680 }, { "epoch": 1.0269, "grad_norm": 21.494562137979017, "learning_rate": 9.755832736423629e-06, "loss": 2.3833, "step": 12690 }, { "epoch": 1.027, "grad_norm": 21.09773254322001, "learning_rate": 9.755332620503342e-06, "loss": 2.4246, "step": 12700 }, { "epoch": 1.0271, "grad_norm": 20.434309552791948, "learning_rate": 9.754832005770732e-06, "loss": 2.3379, "step": 12710 }, { "epoch": 1.0272, "grad_norm": 20.445182767199174, "learning_rate": 9.754330892278313e-06, "loss": 2.3314, "step": 12720 }, { "epoch": 1.0273, "grad_norm": 21.4094684380216, "learning_rate": 9.753829280078648e-06, "loss": 2.3465, "step": 12730 }, { "epoch": 1.0274, "grad_norm": 20.258411801657328, "learning_rate": 9.753327169224356e-06, "loss": 2.3508, "step": 12740 }, { "epoch": 1.0275, "grad_norm": 20.562121736995806, "learning_rate": 9.752824559768106e-06, "loss": 2.343, "step": 12750 }, { "epoch": 1.0276, "grad_norm": 21.195726786181826, "learning_rate": 9.752321451762616e-06, "loss": 2.3629, "step": 12760 }, { "epoch": 1.0277, "grad_norm": 22.16873754738101, "learning_rate": 9.751817845260663e-06, "loss": 2.3462, "step": 12770 }, { "epoch": 1.0278, "grad_norm": 20.323023194641546, "learning_rate": 9.751313740315071e-06, "loss": 2.3458, "step": 12780 }, { "epoch": 1.0279, "grad_norm": 22.092656020679293, "learning_rate": 9.75080913697872e-06, "loss": 2.2967, "step": 12790 }, { "epoch": 1.028, "grad_norm": 21.522660395918983, "learning_rate": 9.75030403530454e-06, "loss": 2.2762, "step": 12800 }, { "epoch": 1.0281, "grad_norm": 22.158811061156598, "learning_rate": 9.749798435345513e-06, "loss": 2.3583, "step": 12810 }, { "epoch": 1.0282, "grad_norm": 22.198990747437964, "learning_rate": 9.749292337154675e-06, "loss": 2.3909, "step": 12820 }, { "epoch": 1.0283, "grad_norm": 21.670654413716772, "learning_rate": 9.748785740785112e-06, "loss": 2.303, "step": 12830 }, { "epoch": 1.0284, "grad_norm": 19.516455461901128, "learning_rate": 9.748278646289965e-06, "loss": 2.2923, "step": 12840 }, { "epoch": 1.0285, "grad_norm": 19.85734293239578, "learning_rate": 9.747771053722425e-06, "loss": 2.3408, "step": 12850 }, { "epoch": 1.0286, "grad_norm": 23.38828940886913, "learning_rate": 9.747262963135738e-06, "loss": 2.3749, "step": 12860 }, { "epoch": 1.0287, "grad_norm": 23.816206132462217, "learning_rate": 9.746754374583198e-06, "loss": 2.3947, "step": 12870 }, { "epoch": 1.0288, "grad_norm": 29.602326123095168, "learning_rate": 9.746245288118155e-06, "loss": 2.3702, "step": 12880 }, { "epoch": 1.0289, "grad_norm": 21.93693851153825, "learning_rate": 9.745735703794009e-06, "loss": 2.3362, "step": 12890 }, { "epoch": 1.029, "grad_norm": 22.26507413510132, "learning_rate": 9.745225621664212e-06, "loss": 2.4435, "step": 12900 }, { "epoch": 1.0291, "grad_norm": 22.066323159665377, "learning_rate": 9.744715041782273e-06, "loss": 2.3589, "step": 12910 }, { "epoch": 1.0292, "grad_norm": 23.847927990414718, "learning_rate": 9.744203964201745e-06, "loss": 2.3878, "step": 12920 }, { "epoch": 1.0293, "grad_norm": 19.90612145839872, "learning_rate": 9.74369238897624e-06, "loss": 2.4172, "step": 12930 }, { "epoch": 1.0294, "grad_norm": 20.781344798827664, "learning_rate": 9.74318031615942e-06, "loss": 2.3901, "step": 12940 }, { "epoch": 1.0295, "grad_norm": 20.425147269547043, "learning_rate": 9.742667745804999e-06, "loss": 2.3039, "step": 12950 }, { "epoch": 1.0296, "grad_norm": 22.160468106563766, "learning_rate": 9.742154677966742e-06, "loss": 2.354, "step": 12960 }, { "epoch": 1.0297, "grad_norm": 21.404537899066504, "learning_rate": 9.74164111269847e-06, "loss": 2.3282, "step": 12970 }, { "epoch": 1.0298, "grad_norm": 22.150838534195216, "learning_rate": 9.74112705005405e-06, "loss": 2.3456, "step": 12980 }, { "epoch": 1.0299, "grad_norm": 20.37605581977292, "learning_rate": 9.740612490087406e-06, "loss": 2.3319, "step": 12990 }, { "epoch": 1.03, "grad_norm": 19.98682406562198, "learning_rate": 9.740097432852515e-06, "loss": 2.2477, "step": 13000 }, { "epoch": 1.0301, "grad_norm": 21.278309012328574, "learning_rate": 9.739581878403404e-06, "loss": 2.2587, "step": 13010 }, { "epoch": 1.0302, "grad_norm": 20.43130768872209, "learning_rate": 9.73906582679415e-06, "loss": 2.3161, "step": 13020 }, { "epoch": 1.0303, "grad_norm": 20.37736324004418, "learning_rate": 9.738549278078886e-06, "loss": 2.2954, "step": 13030 }, { "epoch": 1.0304, "grad_norm": 20.314703021314692, "learning_rate": 9.738032232311794e-06, "loss": 2.3031, "step": 13040 }, { "epoch": 1.0305, "grad_norm": 20.97669424925769, "learning_rate": 9.737514689547111e-06, "loss": 2.2774, "step": 13050 }, { "epoch": 1.0306, "grad_norm": 20.83906252862653, "learning_rate": 9.736996649839123e-06, "loss": 2.2813, "step": 13060 }, { "epoch": 1.0307, "grad_norm": 20.927861526142937, "learning_rate": 9.736478113242173e-06, "loss": 2.2558, "step": 13070 }, { "epoch": 1.0308, "grad_norm": 19.659550612801148, "learning_rate": 9.735959079810654e-06, "loss": 2.2926, "step": 13080 }, { "epoch": 1.0309, "grad_norm": 20.245556381349367, "learning_rate": 9.735439549599003e-06, "loss": 2.2593, "step": 13090 }, { "epoch": 1.031, "grad_norm": 21.666730104911508, "learning_rate": 9.734919522661725e-06, "loss": 2.2519, "step": 13100 }, { "epoch": 1.0311, "grad_norm": 20.308500330964147, "learning_rate": 9.734398999053363e-06, "loss": 2.2896, "step": 13110 }, { "epoch": 1.0312, "grad_norm": 19.445147001426797, "learning_rate": 9.733877978828518e-06, "loss": 2.3294, "step": 13120 }, { "epoch": 1.0313, "grad_norm": 23.20824018116495, "learning_rate": 9.733356462041845e-06, "loss": 2.2771, "step": 13130 }, { "epoch": 1.0314, "grad_norm": 25.286575791174695, "learning_rate": 9.732834448748048e-06, "loss": 2.3011, "step": 13140 }, { "epoch": 1.0315, "grad_norm": 21.573188707350344, "learning_rate": 9.732311939001882e-06, "loss": 2.3053, "step": 13150 }, { "epoch": 1.0316, "grad_norm": 19.987762847324454, "learning_rate": 9.731788932858157e-06, "loss": 2.2422, "step": 13160 }, { "epoch": 1.0317, "grad_norm": 22.025806368309258, "learning_rate": 9.731265430371735e-06, "loss": 2.2883, "step": 13170 }, { "epoch": 1.0318, "grad_norm": 20.238293979066064, "learning_rate": 9.730741431597526e-06, "loss": 2.2294, "step": 13180 }, { "epoch": 1.0319, "grad_norm": 21.576466081363645, "learning_rate": 9.730216936590497e-06, "loss": 2.3324, "step": 13190 }, { "epoch": 1.032, "grad_norm": 20.233109275817128, "learning_rate": 9.729691945405666e-06, "loss": 2.2702, "step": 13200 }, { "epoch": 1.0321, "grad_norm": 22.007260815019468, "learning_rate": 9.7291664580981e-06, "loss": 2.2147, "step": 13210 }, { "epoch": 1.0322, "grad_norm": 24.70344917921432, "learning_rate": 9.728640474722924e-06, "loss": 2.3237, "step": 13220 }, { "epoch": 1.0323, "grad_norm": 21.570558016430674, "learning_rate": 9.728113995335306e-06, "loss": 2.2477, "step": 13230 }, { "epoch": 1.0324, "grad_norm": 18.64020053687625, "learning_rate": 9.727587019990475e-06, "loss": 2.2846, "step": 13240 }, { "epoch": 1.0325, "grad_norm": 21.715663129881317, "learning_rate": 9.727059548743705e-06, "loss": 2.2614, "step": 13250 }, { "epoch": 1.0326, "grad_norm": 23.199738447890553, "learning_rate": 9.726531581650329e-06, "loss": 2.3445, "step": 13260 }, { "epoch": 1.0327, "grad_norm": 21.43485262870059, "learning_rate": 9.726003118765725e-06, "loss": 2.2666, "step": 13270 }, { "epoch": 1.0328, "grad_norm": 21.05234457324313, "learning_rate": 9.72547416014533e-06, "loss": 2.3352, "step": 13280 }, { "epoch": 1.0329, "grad_norm": 21.094198067906564, "learning_rate": 9.724944705844626e-06, "loss": 2.3556, "step": 13290 }, { "epoch": 1.033, "grad_norm": 22.897007905658846, "learning_rate": 9.724414755919151e-06, "loss": 2.3253, "step": 13300 }, { "epoch": 1.0331, "grad_norm": 22.965870227074323, "learning_rate": 9.723884310424497e-06, "loss": 2.2741, "step": 13310 }, { "epoch": 1.0332, "grad_norm": 21.213815687790028, "learning_rate": 9.7233533694163e-06, "loss": 2.3819, "step": 13320 }, { "epoch": 1.0333, "grad_norm": 20.700743653063906, "learning_rate": 9.722821932950259e-06, "loss": 2.2516, "step": 13330 }, { "epoch": 1.0334, "grad_norm": 20.4580310265341, "learning_rate": 9.722290001082114e-06, "loss": 2.3096, "step": 13340 }, { "epoch": 1.0335, "grad_norm": 22.589568310531305, "learning_rate": 9.721757573867667e-06, "loss": 2.2658, "step": 13350 }, { "epoch": 1.0336, "grad_norm": 20.163323648319682, "learning_rate": 9.721224651362765e-06, "loss": 2.2696, "step": 13360 }, { "epoch": 1.0337, "grad_norm": 22.17452445282732, "learning_rate": 9.720691233623307e-06, "loss": 2.3023, "step": 13370 }, { "epoch": 1.0338, "grad_norm": 21.081582176050215, "learning_rate": 9.72015732070525e-06, "loss": 2.3098, "step": 13380 }, { "epoch": 1.0339, "grad_norm": 20.523236316451133, "learning_rate": 9.719622912664595e-06, "loss": 2.2775, "step": 13390 }, { "epoch": 1.034, "grad_norm": 22.62131907008175, "learning_rate": 9.719088009557402e-06, "loss": 2.2615, "step": 13400 }, { "epoch": 1.0341, "grad_norm": 19.365529528686288, "learning_rate": 9.718552611439778e-06, "loss": 2.2653, "step": 13410 }, { "epoch": 1.0342, "grad_norm": 20.062919907055658, "learning_rate": 9.718016718367887e-06, "loss": 2.2786, "step": 13420 }, { "epoch": 1.0343, "grad_norm": 21.775443741555062, "learning_rate": 9.717480330397937e-06, "loss": 2.2343, "step": 13430 }, { "epoch": 1.0344, "grad_norm": 20.373001451617437, "learning_rate": 9.716943447586195e-06, "loss": 2.2382, "step": 13440 }, { "epoch": 1.0345, "grad_norm": 20.975545226746227, "learning_rate": 9.716406069988978e-06, "loss": 2.2533, "step": 13450 }, { "epoch": 1.0346, "grad_norm": 19.728893699292705, "learning_rate": 9.715868197662652e-06, "loss": 2.2438, "step": 13460 }, { "epoch": 1.0347, "grad_norm": 19.20511463189764, "learning_rate": 9.715329830663642e-06, "loss": 2.2652, "step": 13470 }, { "epoch": 1.0348, "grad_norm": 20.577730286205938, "learning_rate": 9.714790969048416e-06, "loss": 2.2365, "step": 13480 }, { "epoch": 1.0349, "grad_norm": 22.194019212649394, "learning_rate": 9.714251612873498e-06, "loss": 2.2187, "step": 13490 }, { "epoch": 1.035, "grad_norm": 22.56131217931325, "learning_rate": 9.713711762195466e-06, "loss": 2.2423, "step": 13500 }, { "epoch": 1.0351, "grad_norm": 21.381380866905225, "learning_rate": 9.713171417070949e-06, "loss": 2.2543, "step": 13510 }, { "epoch": 1.0352, "grad_norm": 21.340079323993685, "learning_rate": 9.712630577556623e-06, "loss": 2.377, "step": 13520 }, { "epoch": 1.0352999999999999, "grad_norm": 19.146739299772182, "learning_rate": 9.712089243709222e-06, "loss": 2.2877, "step": 13530 }, { "epoch": 1.0354, "grad_norm": 20.095115648815447, "learning_rate": 9.711547415585528e-06, "loss": 2.254, "step": 13540 }, { "epoch": 1.0355, "grad_norm": 22.12720118303292, "learning_rate": 9.711005093242379e-06, "loss": 2.3124, "step": 13550 }, { "epoch": 1.0356, "grad_norm": 21.94906606779667, "learning_rate": 9.710462276736659e-06, "loss": 2.2557, "step": 13560 }, { "epoch": 1.0357, "grad_norm": 21.537550515982584, "learning_rate": 9.709918966125306e-06, "loss": 2.2994, "step": 13570 }, { "epoch": 1.0358, "grad_norm": 20.7486633186625, "learning_rate": 9.709375161465316e-06, "loss": 2.3359, "step": 13580 }, { "epoch": 1.0359, "grad_norm": 19.954995729437083, "learning_rate": 9.708830862813728e-06, "loss": 2.1952, "step": 13590 }, { "epoch": 1.036, "grad_norm": 20.783700942461994, "learning_rate": 9.708286070227635e-06, "loss": 2.2227, "step": 13600 }, { "epoch": 1.0361, "grad_norm": 21.558644236634137, "learning_rate": 9.707740783764185e-06, "loss": 2.2293, "step": 13610 }, { "epoch": 1.0362, "grad_norm": 22.592543212860868, "learning_rate": 9.707195003480579e-06, "loss": 2.2539, "step": 13620 }, { "epoch": 1.0363, "grad_norm": 23.072120606784516, "learning_rate": 9.706648729434063e-06, "loss": 2.3499, "step": 13630 }, { "epoch": 1.0364, "grad_norm": 19.231337065706555, "learning_rate": 9.706101961681938e-06, "loss": 2.2363, "step": 13640 }, { "epoch": 1.0365, "grad_norm": 20.203534398469447, "learning_rate": 9.705554700281558e-06, "loss": 2.2656, "step": 13650 }, { "epoch": 1.0366, "grad_norm": 20.412576876730668, "learning_rate": 9.705006945290332e-06, "loss": 2.3357, "step": 13660 }, { "epoch": 1.0367, "grad_norm": 21.062551139985434, "learning_rate": 9.704458696765713e-06, "loss": 2.2976, "step": 13670 }, { "epoch": 1.0368, "grad_norm": 21.194362773503805, "learning_rate": 9.703909954765211e-06, "loss": 2.234, "step": 13680 }, { "epoch": 1.0369, "grad_norm": 20.976838154366867, "learning_rate": 9.703360719346384e-06, "loss": 2.2064, "step": 13690 }, { "epoch": 1.037, "grad_norm": 21.48171812040072, "learning_rate": 9.70281099056685e-06, "loss": 2.2604, "step": 13700 }, { "epoch": 1.0371, "grad_norm": 24.065541951267974, "learning_rate": 9.702260768484268e-06, "loss": 2.3713, "step": 13710 }, { "epoch": 1.0372, "grad_norm": 20.20243958299827, "learning_rate": 9.701710053156355e-06, "loss": 2.3058, "step": 13720 }, { "epoch": 1.0373, "grad_norm": 22.762155191939694, "learning_rate": 9.701158844640878e-06, "loss": 2.2964, "step": 13730 }, { "epoch": 1.0374, "grad_norm": 19.646967530360875, "learning_rate": 9.700607142995659e-06, "loss": 2.3115, "step": 13740 }, { "epoch": 1.0375, "grad_norm": 22.29415045751823, "learning_rate": 9.700054948278565e-06, "loss": 2.2664, "step": 13750 }, { "epoch": 1.0376, "grad_norm": 21.06432124135892, "learning_rate": 9.699502260547522e-06, "loss": 2.309, "step": 13760 }, { "epoch": 1.0377, "grad_norm": 20.608387992058383, "learning_rate": 9.698949079860503e-06, "loss": 2.3468, "step": 13770 }, { "epoch": 1.0378, "grad_norm": 19.593293886402268, "learning_rate": 9.698395406275531e-06, "loss": 2.254, "step": 13780 }, { "epoch": 1.0379, "grad_norm": 19.77419355071034, "learning_rate": 9.697841239850689e-06, "loss": 2.2766, "step": 13790 }, { "epoch": 1.038, "grad_norm": 21.864885533054668, "learning_rate": 9.697286580644105e-06, "loss": 2.2764, "step": 13800 }, { "epoch": 1.0381, "grad_norm": 20.385994368044745, "learning_rate": 9.696731428713959e-06, "loss": 2.236, "step": 13810 }, { "epoch": 1.0382, "grad_norm": 22.218040028981992, "learning_rate": 9.696175784118485e-06, "loss": 2.2742, "step": 13820 }, { "epoch": 1.0383, "grad_norm": 23.423837512257663, "learning_rate": 9.695619646915966e-06, "loss": 2.3326, "step": 13830 }, { "epoch": 1.0384, "grad_norm": 21.057279317968337, "learning_rate": 9.695063017164738e-06, "loss": 2.2572, "step": 13840 }, { "epoch": 1.0385, "grad_norm": 19.499958376571346, "learning_rate": 9.694505894923191e-06, "loss": 2.335, "step": 13850 }, { "epoch": 1.0386, "grad_norm": 21.10242343340171, "learning_rate": 9.693948280249763e-06, "loss": 2.2307, "step": 13860 }, { "epoch": 1.0387, "grad_norm": 22.18892314332344, "learning_rate": 9.693390173202947e-06, "loss": 2.3499, "step": 13870 }, { "epoch": 1.0388, "grad_norm": 21.832648952954624, "learning_rate": 9.692831573841284e-06, "loss": 2.3919, "step": 13880 }, { "epoch": 1.0389, "grad_norm": 18.81624788670358, "learning_rate": 9.692272482223369e-06, "loss": 2.33, "step": 13890 }, { "epoch": 1.039, "grad_norm": 21.693024281882273, "learning_rate": 9.691712898407847e-06, "loss": 2.3721, "step": 13900 }, { "epoch": 1.0391, "grad_norm": 24.1142651089109, "learning_rate": 9.691152822453418e-06, "loss": 2.3295, "step": 13910 }, { "epoch": 1.0392, "grad_norm": 21.862494188088508, "learning_rate": 9.69059225441883e-06, "loss": 2.3019, "step": 13920 }, { "epoch": 1.0393, "grad_norm": 20.055669961178637, "learning_rate": 9.690031194362884e-06, "loss": 2.2781, "step": 13930 }, { "epoch": 1.0394, "grad_norm": 21.105691760034787, "learning_rate": 9.689469642344433e-06, "loss": 2.2466, "step": 13940 }, { "epoch": 1.0395, "grad_norm": 21.371603588185007, "learning_rate": 9.688907598422381e-06, "loss": 2.3467, "step": 13950 }, { "epoch": 1.0396, "grad_norm": 22.493539992801544, "learning_rate": 9.688345062655684e-06, "loss": 2.2909, "step": 13960 }, { "epoch": 1.0397, "grad_norm": 24.89252327675876, "learning_rate": 9.687782035103349e-06, "loss": 2.2733, "step": 13970 }, { "epoch": 1.0398, "grad_norm": 20.083884514972354, "learning_rate": 9.687218515824435e-06, "loss": 2.3432, "step": 13980 }, { "epoch": 1.0399, "grad_norm": 21.179457865660076, "learning_rate": 9.686654504878053e-06, "loss": 2.3247, "step": 13990 }, { "epoch": 1.04, "grad_norm": 20.405047596953004, "learning_rate": 9.686090002323365e-06, "loss": 2.3211, "step": 14000 }, { "epoch": 1.0401, "grad_norm": 20.593765760282412, "learning_rate": 9.685525008219585e-06, "loss": 2.3236, "step": 14010 }, { "epoch": 1.0402, "grad_norm": 26.58065959717382, "learning_rate": 9.684959522625977e-06, "loss": 2.3704, "step": 14020 }, { "epoch": 1.0403, "grad_norm": 20.601819542076214, "learning_rate": 9.684393545601859e-06, "loss": 2.248, "step": 14030 }, { "epoch": 1.0404, "grad_norm": 21.669871062201025, "learning_rate": 9.683827077206598e-06, "loss": 2.3379, "step": 14040 }, { "epoch": 1.0405, "grad_norm": 23.581374674021003, "learning_rate": 9.683260117499616e-06, "loss": 2.2388, "step": 14050 }, { "epoch": 1.0406, "grad_norm": 20.386030673953332, "learning_rate": 9.682692666540385e-06, "loss": 2.2532, "step": 14060 }, { "epoch": 1.0407, "grad_norm": 22.86895884617281, "learning_rate": 9.682124724388423e-06, "loss": 2.2226, "step": 14070 }, { "epoch": 1.0408, "grad_norm": 20.505789809040806, "learning_rate": 9.68155629110331e-06, "loss": 2.3168, "step": 14080 }, { "epoch": 1.0409, "grad_norm": 20.61227293193883, "learning_rate": 9.68098736674467e-06, "loss": 2.3071, "step": 14090 }, { "epoch": 1.041, "grad_norm": 19.807701186487943, "learning_rate": 9.680417951372178e-06, "loss": 2.3179, "step": 14100 }, { "epoch": 1.0411, "grad_norm": 22.405889294424867, "learning_rate": 9.679848045045568e-06, "loss": 2.3444, "step": 14110 }, { "epoch": 1.0412, "grad_norm": 21.57759473077795, "learning_rate": 9.679277647824616e-06, "loss": 2.309, "step": 14120 }, { "epoch": 1.0413000000000001, "grad_norm": 21.685812437576054, "learning_rate": 9.678706759769157e-06, "loss": 2.3033, "step": 14130 }, { "epoch": 1.0414, "grad_norm": 19.69300267835324, "learning_rate": 9.678135380939074e-06, "loss": 2.1956, "step": 14140 }, { "epoch": 1.0415, "grad_norm": 22.640068741735035, "learning_rate": 9.6775635113943e-06, "loss": 2.2681, "step": 14150 }, { "epoch": 1.0416, "grad_norm": 21.11680686718869, "learning_rate": 9.676991151194823e-06, "loss": 2.2081, "step": 14160 }, { "epoch": 1.0417, "grad_norm": 23.28203992995155, "learning_rate": 9.676418300400682e-06, "loss": 2.2639, "step": 14170 }, { "epoch": 1.0418, "grad_norm": 23.289185515631992, "learning_rate": 9.675844959071966e-06, "loss": 2.2127, "step": 14180 }, { "epoch": 1.0419, "grad_norm": 21.250698924650912, "learning_rate": 9.675271127268812e-06, "loss": 2.2941, "step": 14190 }, { "epoch": 1.042, "grad_norm": 22.140209686530106, "learning_rate": 9.674696805051417e-06, "loss": 2.2724, "step": 14200 }, { "epoch": 1.0421, "grad_norm": 24.14475652483479, "learning_rate": 9.674121992480024e-06, "loss": 2.2029, "step": 14210 }, { "epoch": 1.0422, "grad_norm": 22.678000119027008, "learning_rate": 9.673546689614926e-06, "loss": 2.3782, "step": 14220 }, { "epoch": 1.0423, "grad_norm": 22.574909464675994, "learning_rate": 9.672970896516471e-06, "loss": 2.2506, "step": 14230 }, { "epoch": 1.0424, "grad_norm": 22.250982110746452, "learning_rate": 9.672394613245058e-06, "loss": 2.3143, "step": 14240 }, { "epoch": 1.0425, "grad_norm": 23.01901898379262, "learning_rate": 9.671817839861134e-06, "loss": 2.3307, "step": 14250 }, { "epoch": 1.0426, "grad_norm": 21.019275946374275, "learning_rate": 9.671240576425202e-06, "loss": 2.3063, "step": 14260 }, { "epoch": 1.0427, "grad_norm": 21.095951350938396, "learning_rate": 9.670662822997813e-06, "loss": 2.2765, "step": 14270 }, { "epoch": 1.0428, "grad_norm": 20.718639644846814, "learning_rate": 9.670084579639571e-06, "loss": 2.2875, "step": 14280 }, { "epoch": 1.0429, "grad_norm": 22.29520518818035, "learning_rate": 9.669505846411132e-06, "loss": 2.4031, "step": 14290 }, { "epoch": 1.043, "grad_norm": 20.61944777378674, "learning_rate": 9.6689266233732e-06, "loss": 2.2472, "step": 14300 }, { "epoch": 1.0431, "grad_norm": 21.10309411498565, "learning_rate": 9.668346910586535e-06, "loss": 2.2836, "step": 14310 }, { "epoch": 1.0432, "grad_norm": 18.045368832251064, "learning_rate": 9.667766708111946e-06, "loss": 2.2419, "step": 14320 }, { "epoch": 1.0433, "grad_norm": 20.363273369618657, "learning_rate": 9.667186016010293e-06, "loss": 2.2782, "step": 14330 }, { "epoch": 1.0434, "grad_norm": 21.334564327660903, "learning_rate": 9.666604834342487e-06, "loss": 2.2602, "step": 14340 }, { "epoch": 1.0435, "grad_norm": 24.795786652916956, "learning_rate": 9.666023163169493e-06, "loss": 2.3337, "step": 14350 }, { "epoch": 1.0436, "grad_norm": 22.518101938180422, "learning_rate": 9.665441002552325e-06, "loss": 2.2745, "step": 14360 }, { "epoch": 1.0437, "grad_norm": 19.994225389941803, "learning_rate": 9.66485835255205e-06, "loss": 2.2954, "step": 14370 }, { "epoch": 1.0438, "grad_norm": 22.675358432488913, "learning_rate": 9.664275213229782e-06, "loss": 2.2927, "step": 14380 }, { "epoch": 1.0439, "grad_norm": 20.540763890240203, "learning_rate": 9.663691584646693e-06, "loss": 2.3383, "step": 14390 }, { "epoch": 1.044, "grad_norm": 18.37897558402848, "learning_rate": 9.663107466864001e-06, "loss": 2.2327, "step": 14400 }, { "epoch": 1.0441, "grad_norm": 20.900310551828643, "learning_rate": 9.662522859942977e-06, "loss": 2.2853, "step": 14410 }, { "epoch": 1.0442, "grad_norm": 22.393905876331505, "learning_rate": 9.661937763944947e-06, "loss": 2.2555, "step": 14420 }, { "epoch": 1.0443, "grad_norm": 20.054172787053954, "learning_rate": 9.661352178931281e-06, "loss": 2.2543, "step": 14430 }, { "epoch": 1.0444, "grad_norm": 20.39271524222734, "learning_rate": 9.660766104963406e-06, "loss": 2.2242, "step": 14440 }, { "epoch": 1.0445, "grad_norm": 20.598837238138394, "learning_rate": 9.660179542102796e-06, "loss": 2.2927, "step": 14450 }, { "epoch": 1.0446, "grad_norm": 21.099032027894335, "learning_rate": 9.659592490410982e-06, "loss": 2.3095, "step": 14460 }, { "epoch": 1.0447, "grad_norm": 22.55652915640962, "learning_rate": 9.65900494994954e-06, "loss": 2.2943, "step": 14470 }, { "epoch": 1.0448, "grad_norm": 21.293678720755445, "learning_rate": 9.658416920780106e-06, "loss": 2.2639, "step": 14480 }, { "epoch": 1.0449, "grad_norm": 20.85645548026876, "learning_rate": 9.657828402964353e-06, "loss": 2.2557, "step": 14490 }, { "epoch": 1.045, "grad_norm": 22.038741422801685, "learning_rate": 9.65723939656402e-06, "loss": 2.3155, "step": 14500 }, { "epoch": 1.0451, "grad_norm": 22.62938251925604, "learning_rate": 9.65664990164089e-06, "loss": 2.183, "step": 14510 }, { "epoch": 1.0452, "grad_norm": 22.068846224409153, "learning_rate": 9.656059918256795e-06, "loss": 2.2504, "step": 14520 }, { "epoch": 1.0453, "grad_norm": 21.305857642274038, "learning_rate": 9.655469446473626e-06, "loss": 2.2554, "step": 14530 }, { "epoch": 1.0454, "grad_norm": 20.49485232076078, "learning_rate": 9.654878486353318e-06, "loss": 2.2233, "step": 14540 }, { "epoch": 1.0455, "grad_norm": 22.791049516151194, "learning_rate": 9.65428703795786e-06, "loss": 2.2533, "step": 14550 }, { "epoch": 1.0456, "grad_norm": 22.430252101323013, "learning_rate": 9.653695101349295e-06, "loss": 2.3113, "step": 14560 }, { "epoch": 1.0457, "grad_norm": 21.013815416776495, "learning_rate": 9.653102676589711e-06, "loss": 2.271, "step": 14570 }, { "epoch": 1.0458, "grad_norm": 20.704136455653078, "learning_rate": 9.65250976374125e-06, "loss": 2.2945, "step": 14580 }, { "epoch": 1.0459, "grad_norm": 20.6011355990914, "learning_rate": 9.651916362866111e-06, "loss": 2.2363, "step": 14590 }, { "epoch": 1.046, "grad_norm": 20.751728358062508, "learning_rate": 9.651322474026536e-06, "loss": 2.2822, "step": 14600 }, { "epoch": 1.0461, "grad_norm": 18.191455362697493, "learning_rate": 9.650728097284818e-06, "loss": 2.2503, "step": 14610 }, { "epoch": 1.0462, "grad_norm": 20.682962074266825, "learning_rate": 9.65013323270331e-06, "loss": 2.244, "step": 14620 }, { "epoch": 1.0463, "grad_norm": 20.964405865719115, "learning_rate": 9.649537880344406e-06, "loss": 2.2479, "step": 14630 }, { "epoch": 1.0464, "grad_norm": 21.719916355955455, "learning_rate": 9.648942040270559e-06, "loss": 2.1938, "step": 14640 }, { "epoch": 1.0465, "grad_norm": 22.9247738870213, "learning_rate": 9.648345712544266e-06, "loss": 2.1897, "step": 14650 }, { "epoch": 1.0466, "grad_norm": 20.817862819518087, "learning_rate": 9.647748897228085e-06, "loss": 2.2528, "step": 14660 }, { "epoch": 1.0467, "grad_norm": 21.417650316276088, "learning_rate": 9.647151594384612e-06, "loss": 2.2339, "step": 14670 }, { "epoch": 1.0468, "grad_norm": 22.17148990404459, "learning_rate": 9.646553804076506e-06, "loss": 2.2665, "step": 14680 }, { "epoch": 1.0469, "grad_norm": 20.44996447448358, "learning_rate": 9.645955526366472e-06, "loss": 2.1797, "step": 14690 }, { "epoch": 1.047, "grad_norm": 20.389518451972734, "learning_rate": 9.645356761317266e-06, "loss": 2.2603, "step": 14700 }, { "epoch": 1.0471, "grad_norm": 22.39986033595841, "learning_rate": 9.644757508991695e-06, "loss": 2.2775, "step": 14710 }, { "epoch": 1.0472, "grad_norm": 20.01533524444188, "learning_rate": 9.64415776945262e-06, "loss": 2.2637, "step": 14720 }, { "epoch": 1.0473, "grad_norm": 22.059579049366892, "learning_rate": 9.643557542762947e-06, "loss": 2.2723, "step": 14730 }, { "epoch": 1.0474, "grad_norm": 20.88296773692123, "learning_rate": 9.642956828985638e-06, "loss": 2.2255, "step": 14740 }, { "epoch": 1.0475, "grad_norm": 21.26148218637891, "learning_rate": 9.642355628183708e-06, "loss": 2.1896, "step": 14750 }, { "epoch": 1.0476, "grad_norm": 21.741873435648472, "learning_rate": 9.64175394042022e-06, "loss": 2.2619, "step": 14760 }, { "epoch": 1.0477, "grad_norm": 20.223789787857452, "learning_rate": 9.641151765758284e-06, "loss": 2.2739, "step": 14770 }, { "epoch": 1.0478, "grad_norm": 22.66199055398407, "learning_rate": 9.64054910426107e-06, "loss": 2.2246, "step": 14780 }, { "epoch": 1.0479, "grad_norm": 22.27086377862522, "learning_rate": 9.639945955991791e-06, "loss": 2.2089, "step": 14790 }, { "epoch": 1.048, "grad_norm": 23.727908008431687, "learning_rate": 9.639342321013716e-06, "loss": 2.2164, "step": 14800 }, { "epoch": 1.0481, "grad_norm": 19.77524203045611, "learning_rate": 9.638738199390164e-06, "loss": 2.1639, "step": 14810 }, { "epoch": 1.0482, "grad_norm": 21.815060595645775, "learning_rate": 9.638133591184505e-06, "loss": 2.2695, "step": 14820 }, { "epoch": 1.0483, "grad_norm": 21.026085436432293, "learning_rate": 9.637528496460156e-06, "loss": 2.3116, "step": 14830 }, { "epoch": 1.0484, "grad_norm": 18.64278416300453, "learning_rate": 9.636922915280593e-06, "loss": 2.1679, "step": 14840 }, { "epoch": 1.0485, "grad_norm": 24.664948584400836, "learning_rate": 9.636316847709335e-06, "loss": 2.2374, "step": 14850 }, { "epoch": 1.0486, "grad_norm": 20.380738839645087, "learning_rate": 9.635710293809958e-06, "loss": 2.1545, "step": 14860 }, { "epoch": 1.0487, "grad_norm": 20.12311143260926, "learning_rate": 9.635103253646088e-06, "loss": 2.2284, "step": 14870 }, { "epoch": 1.0488, "grad_norm": 19.691543358046644, "learning_rate": 9.634495727281398e-06, "loss": 2.2494, "step": 14880 }, { "epoch": 1.0489, "grad_norm": 21.950477086884227, "learning_rate": 9.633887714779615e-06, "loss": 2.2725, "step": 14890 }, { "epoch": 1.049, "grad_norm": 21.017865281921743, "learning_rate": 9.633279216204518e-06, "loss": 2.2514, "step": 14900 }, { "epoch": 1.0491, "grad_norm": 21.777802040494187, "learning_rate": 9.632670231619934e-06, "loss": 2.2066, "step": 14910 }, { "epoch": 1.0492, "grad_norm": 24.47857121443935, "learning_rate": 9.632060761089745e-06, "loss": 2.2799, "step": 14920 }, { "epoch": 1.0493, "grad_norm": 19.661073449317747, "learning_rate": 9.631450804677879e-06, "loss": 2.2579, "step": 14930 }, { "epoch": 1.0493999999999999, "grad_norm": 20.02199417201613, "learning_rate": 9.630840362448319e-06, "loss": 2.1859, "step": 14940 }, { "epoch": 1.0495, "grad_norm": 23.645670080588268, "learning_rate": 9.630229434465097e-06, "loss": 2.1929, "step": 14950 }, { "epoch": 1.0496, "grad_norm": 20.709886494837846, "learning_rate": 9.629618020792297e-06, "loss": 2.2423, "step": 14960 }, { "epoch": 1.0497, "grad_norm": 21.152416728905628, "learning_rate": 9.629006121494054e-06, "loss": 2.3427, "step": 14970 }, { "epoch": 1.0498, "grad_norm": 21.37840897339183, "learning_rate": 9.628393736634554e-06, "loss": 2.2323, "step": 14980 }, { "epoch": 1.0499, "grad_norm": 23.861627040066395, "learning_rate": 9.627780866278029e-06, "loss": 2.1899, "step": 14990 }, { "epoch": 1.05, "grad_norm": 20.19219083747057, "learning_rate": 9.627167510488771e-06, "loss": 2.2182, "step": 15000 }, { "epoch": 1.0501, "grad_norm": 20.492508376370925, "learning_rate": 9.626553669331117e-06, "loss": 2.1998, "step": 15010 }, { "epoch": 1.0502, "grad_norm": 22.157214485442605, "learning_rate": 9.625939342869454e-06, "loss": 2.2264, "step": 15020 }, { "epoch": 1.0503, "grad_norm": 21.912096186880667, "learning_rate": 9.625324531168225e-06, "loss": 2.2814, "step": 15030 }, { "epoch": 1.0504, "grad_norm": 21.04177533698656, "learning_rate": 9.62470923429192e-06, "loss": 2.2558, "step": 15040 }, { "epoch": 1.0505, "grad_norm": 20.959836786903068, "learning_rate": 9.62409345230508e-06, "loss": 2.2444, "step": 15050 }, { "epoch": 1.0506, "grad_norm": 20.241583446649354, "learning_rate": 9.623477185272297e-06, "loss": 2.2441, "step": 15060 }, { "epoch": 1.0507, "grad_norm": 22.63739278940165, "learning_rate": 9.622860433258214e-06, "loss": 2.4092, "step": 15070 }, { "epoch": 1.0508, "grad_norm": 22.212016172105223, "learning_rate": 9.62224319632753e-06, "loss": 2.2812, "step": 15080 }, { "epoch": 1.0509, "grad_norm": 22.375355686890735, "learning_rate": 9.621625474544984e-06, "loss": 2.2293, "step": 15090 }, { "epoch": 1.051, "grad_norm": 21.589151491974448, "learning_rate": 9.621007267975377e-06, "loss": 2.2417, "step": 15100 }, { "epoch": 1.0511, "grad_norm": 21.77826086734877, "learning_rate": 9.620388576683555e-06, "loss": 2.266, "step": 15110 }, { "epoch": 1.0512, "grad_norm": 20.22376082458117, "learning_rate": 9.619769400734415e-06, "loss": 2.2411, "step": 15120 }, { "epoch": 1.0513, "grad_norm": 21.396724096892406, "learning_rate": 9.619149740192905e-06, "loss": 2.2684, "step": 15130 }, { "epoch": 1.0514000000000001, "grad_norm": 21.904603432535833, "learning_rate": 9.618529595124026e-06, "loss": 2.165, "step": 15140 }, { "epoch": 1.0515, "grad_norm": 28.94935927780261, "learning_rate": 9.617908965592828e-06, "loss": 2.3441, "step": 15150 }, { "epoch": 1.0516, "grad_norm": 24.3682764424965, "learning_rate": 9.617287851664411e-06, "loss": 2.2536, "step": 15160 }, { "epoch": 1.0517, "grad_norm": 22.505245831494275, "learning_rate": 9.616666253403929e-06, "loss": 2.2232, "step": 15170 }, { "epoch": 1.0518, "grad_norm": 20.131960678055105, "learning_rate": 9.616044170876582e-06, "loss": 2.2508, "step": 15180 }, { "epoch": 1.0519, "grad_norm": 20.077436268715086, "learning_rate": 9.615421604147627e-06, "loss": 2.171, "step": 15190 }, { "epoch": 1.052, "grad_norm": 23.57143205228217, "learning_rate": 9.614798553282364e-06, "loss": 2.2077, "step": 15200 }, { "epoch": 1.0521, "grad_norm": 21.054664229298, "learning_rate": 9.614175018346154e-06, "loss": 2.2109, "step": 15210 }, { "epoch": 1.0522, "grad_norm": 24.07529101977983, "learning_rate": 9.613550999404398e-06, "loss": 2.2121, "step": 15220 }, { "epoch": 1.0523, "grad_norm": 23.076968994669894, "learning_rate": 9.612926496522554e-06, "loss": 2.291, "step": 15230 }, { "epoch": 1.0524, "grad_norm": 21.553694742493203, "learning_rate": 9.612301509766131e-06, "loss": 2.2935, "step": 15240 }, { "epoch": 1.0525, "grad_norm": 19.59292423286615, "learning_rate": 9.611676039200686e-06, "loss": 2.2644, "step": 15250 }, { "epoch": 1.0526, "grad_norm": 21.033179845943167, "learning_rate": 9.611050084891826e-06, "loss": 2.2656, "step": 15260 }, { "epoch": 1.0527, "grad_norm": 19.01660080503283, "learning_rate": 9.610423646905213e-06, "loss": 2.2498, "step": 15270 }, { "epoch": 1.0528, "grad_norm": 20.61782992019266, "learning_rate": 9.609796725306559e-06, "loss": 2.2144, "step": 15280 }, { "epoch": 1.0529, "grad_norm": 22.21966278596986, "learning_rate": 9.609169320161621e-06, "loss": 2.1881, "step": 15290 }, { "epoch": 1.053, "grad_norm": 23.299486202915176, "learning_rate": 9.608541431536216e-06, "loss": 2.2525, "step": 15300 }, { "epoch": 1.0531, "grad_norm": 21.650522600554265, "learning_rate": 9.607913059496201e-06, "loss": 2.1855, "step": 15310 }, { "epoch": 1.0532, "grad_norm": 21.783090632768648, "learning_rate": 9.607284204107493e-06, "loss": 2.2522, "step": 15320 }, { "epoch": 1.0533, "grad_norm": 21.36630757694711, "learning_rate": 9.606654865436056e-06, "loss": 2.3306, "step": 15330 }, { "epoch": 1.0534, "grad_norm": 21.442551284028244, "learning_rate": 9.606025043547901e-06, "loss": 2.261, "step": 15340 }, { "epoch": 1.0535, "grad_norm": 19.76441601968123, "learning_rate": 9.605394738509099e-06, "loss": 2.2667, "step": 15350 }, { "epoch": 1.0536, "grad_norm": 20.43707399811401, "learning_rate": 9.604763950385763e-06, "loss": 2.3401, "step": 15360 }, { "epoch": 1.0537, "grad_norm": 19.60420811501762, "learning_rate": 9.604132679244058e-06, "loss": 2.2963, "step": 15370 }, { "epoch": 1.0538, "grad_norm": 22.365697003741403, "learning_rate": 9.603500925150204e-06, "loss": 2.2371, "step": 15380 }, { "epoch": 1.0539, "grad_norm": 21.54358829984902, "learning_rate": 9.60286868817047e-06, "loss": 2.2831, "step": 15390 }, { "epoch": 1.054, "grad_norm": 21.53556692845969, "learning_rate": 9.602235968371173e-06, "loss": 2.2354, "step": 15400 }, { "epoch": 1.0541, "grad_norm": 24.071572800587944, "learning_rate": 9.601602765818681e-06, "loss": 2.2859, "step": 15410 }, { "epoch": 1.0542, "grad_norm": 20.152588949386086, "learning_rate": 9.600969080579417e-06, "loss": 2.309, "step": 15420 }, { "epoch": 1.0543, "grad_norm": 20.753305440995355, "learning_rate": 9.600334912719849e-06, "loss": 2.3367, "step": 15430 }, { "epoch": 1.0544, "grad_norm": 21.725069128116882, "learning_rate": 9.599700262306499e-06, "loss": 2.2234, "step": 15440 }, { "epoch": 1.0545, "grad_norm": 19.80261734449859, "learning_rate": 9.59906512940594e-06, "loss": 2.1669, "step": 15450 }, { "epoch": 1.0546, "grad_norm": 21.263864306352012, "learning_rate": 9.598429514084794e-06, "loss": 2.2476, "step": 15460 }, { "epoch": 1.0547, "grad_norm": 25.20878467574775, "learning_rate": 9.597793416409736e-06, "loss": 2.2553, "step": 15470 }, { "epoch": 1.0548, "grad_norm": 21.873090156649823, "learning_rate": 9.597156836447485e-06, "loss": 2.318, "step": 15480 }, { "epoch": 1.0549, "grad_norm": 21.468261965475413, "learning_rate": 9.596519774264818e-06, "loss": 2.2708, "step": 15490 }, { "epoch": 1.055, "grad_norm": 21.698234003340914, "learning_rate": 9.595882229928562e-06, "loss": 2.3021, "step": 15500 }, { "epoch": 1.0551, "grad_norm": 20.095815104164092, "learning_rate": 9.595244203505588e-06, "loss": 2.1486, "step": 15510 }, { "epoch": 1.0552, "grad_norm": 25.866892908819636, "learning_rate": 9.594605695062824e-06, "loss": 2.2016, "step": 15520 }, { "epoch": 1.0553, "grad_norm": 21.7092913869501, "learning_rate": 9.593966704667248e-06, "loss": 2.2255, "step": 15530 }, { "epoch": 1.0554, "grad_norm": 21.487442403096843, "learning_rate": 9.593327232385885e-06, "loss": 2.2985, "step": 15540 }, { "epoch": 1.0555, "grad_norm": 23.247792672608288, "learning_rate": 9.592687278285815e-06, "loss": 2.2844, "step": 15550 }, { "epoch": 1.0556, "grad_norm": 20.554028807703613, "learning_rate": 9.592046842434165e-06, "loss": 2.2986, "step": 15560 }, { "epoch": 1.0557, "grad_norm": 19.650095695930954, "learning_rate": 9.591405924898113e-06, "loss": 2.3123, "step": 15570 }, { "epoch": 1.0558, "grad_norm": 22.95629289435578, "learning_rate": 9.590764525744888e-06, "loss": 2.381, "step": 15580 }, { "epoch": 1.0559, "grad_norm": 20.735489687647192, "learning_rate": 9.590122645041771e-06, "loss": 2.2183, "step": 15590 }, { "epoch": 1.056, "grad_norm": 22.20938480998334, "learning_rate": 9.589480282856092e-06, "loss": 2.3369, "step": 15600 }, { "epoch": 1.0561, "grad_norm": 23.407908955170242, "learning_rate": 9.588837439255232e-06, "loss": 2.2212, "step": 15610 }, { "epoch": 1.0562, "grad_norm": 22.617251905953722, "learning_rate": 9.588194114306621e-06, "loss": 2.2814, "step": 15620 }, { "epoch": 1.0563, "grad_norm": 21.494349206916915, "learning_rate": 9.587550308077744e-06, "loss": 2.3335, "step": 15630 }, { "epoch": 1.0564, "grad_norm": 22.486254653080255, "learning_rate": 9.58690602063613e-06, "loss": 2.3771, "step": 15640 }, { "epoch": 1.0565, "grad_norm": 19.73308663656104, "learning_rate": 9.586261252049363e-06, "loss": 2.3257, "step": 15650 }, { "epoch": 1.0566, "grad_norm": 19.037945869154104, "learning_rate": 9.585616002385076e-06, "loss": 2.2742, "step": 15660 }, { "epoch": 1.0567, "grad_norm": 20.562443732573396, "learning_rate": 9.584970271710953e-06, "loss": 2.3393, "step": 15670 }, { "epoch": 1.0568, "grad_norm": 18.4069608438605, "learning_rate": 9.584324060094727e-06, "loss": 2.2718, "step": 15680 }, { "epoch": 1.0569, "grad_norm": 21.89056662541719, "learning_rate": 9.583677367604184e-06, "loss": 2.3497, "step": 15690 }, { "epoch": 1.057, "grad_norm": 20.911258229833066, "learning_rate": 9.583030194307161e-06, "loss": 2.2878, "step": 15700 }, { "epoch": 1.0571, "grad_norm": 25.152127286611858, "learning_rate": 9.582382540271538e-06, "loss": 2.2082, "step": 15710 }, { "epoch": 1.0572, "grad_norm": 21.207562746846932, "learning_rate": 9.581734405565257e-06, "loss": 2.3409, "step": 15720 }, { "epoch": 1.0573, "grad_norm": 21.12804680788987, "learning_rate": 9.5810857902563e-06, "loss": 2.3203, "step": 15730 }, { "epoch": 1.0574, "grad_norm": 20.15035846390543, "learning_rate": 9.580436694412703e-06, "loss": 2.2994, "step": 15740 }, { "epoch": 1.0575, "grad_norm": 21.125513771638072, "learning_rate": 9.579787118102556e-06, "loss": 2.2402, "step": 15750 }, { "epoch": 1.0576, "grad_norm": 24.291307454384206, "learning_rate": 9.579137061393999e-06, "loss": 2.3, "step": 15760 }, { "epoch": 1.0577, "grad_norm": 21.02451982190205, "learning_rate": 9.578486524355214e-06, "loss": 2.3141, "step": 15770 }, { "epoch": 1.0578, "grad_norm": 21.068821692552785, "learning_rate": 9.577835507054443e-06, "loss": 2.2852, "step": 15780 }, { "epoch": 1.0579, "grad_norm": 22.78820552884228, "learning_rate": 9.577184009559973e-06, "loss": 2.3739, "step": 15790 }, { "epoch": 1.058, "grad_norm": 20.686584848273252, "learning_rate": 9.576532031940143e-06, "loss": 2.4449, "step": 15800 }, { "epoch": 1.0581, "grad_norm": 20.66683226226077, "learning_rate": 9.575879574263344e-06, "loss": 2.3534, "step": 15810 }, { "epoch": 1.0582, "grad_norm": 22.126883467197192, "learning_rate": 9.575226636598015e-06, "loss": 2.3229, "step": 15820 }, { "epoch": 1.0583, "grad_norm": 22.483033340975282, "learning_rate": 9.574573219012646e-06, "loss": 2.3545, "step": 15830 }, { "epoch": 1.0584, "grad_norm": 20.131569999837424, "learning_rate": 9.573919321575777e-06, "loss": 2.267, "step": 15840 }, { "epoch": 1.0585, "grad_norm": 22.004725099378135, "learning_rate": 9.573264944356001e-06, "loss": 2.2296, "step": 15850 }, { "epoch": 1.0586, "grad_norm": 22.740239048374796, "learning_rate": 9.572610087421957e-06, "loss": 2.2352, "step": 15860 }, { "epoch": 1.0587, "grad_norm": 24.099946456985847, "learning_rate": 9.571954750842336e-06, "loss": 2.3093, "step": 15870 }, { "epoch": 1.0588, "grad_norm": 22.41887784027121, "learning_rate": 9.571298934685882e-06, "loss": 2.3731, "step": 15880 }, { "epoch": 1.0589, "grad_norm": 20.38311250629664, "learning_rate": 9.570642639021386e-06, "loss": 2.2816, "step": 15890 }, { "epoch": 1.059, "grad_norm": 19.491456726880315, "learning_rate": 9.569985863917687e-06, "loss": 2.2566, "step": 15900 }, { "epoch": 1.0591, "grad_norm": 22.185766038833233, "learning_rate": 9.569328609443685e-06, "loss": 2.2823, "step": 15910 }, { "epoch": 1.0592, "grad_norm": 19.328664715931566, "learning_rate": 9.568670875668317e-06, "loss": 2.2267, "step": 15920 }, { "epoch": 1.0593, "grad_norm": 20.619754702237643, "learning_rate": 9.568012662660577e-06, "loss": 2.3452, "step": 15930 }, { "epoch": 1.0594, "grad_norm": 21.729428145084324, "learning_rate": 9.56735397048951e-06, "loss": 2.2632, "step": 15940 }, { "epoch": 1.0594999999999999, "grad_norm": 20.59891805251018, "learning_rate": 9.56669479922421e-06, "loss": 2.2715, "step": 15950 }, { "epoch": 1.0596, "grad_norm": 20.221541643137613, "learning_rate": 9.56603514893382e-06, "loss": 2.2525, "step": 15960 }, { "epoch": 1.0597, "grad_norm": 21.362906015086885, "learning_rate": 9.565375019687536e-06, "loss": 2.2645, "step": 15970 }, { "epoch": 1.0598, "grad_norm": 20.687308981395468, "learning_rate": 9.564714411554598e-06, "loss": 2.2947, "step": 15980 }, { "epoch": 1.0599, "grad_norm": 20.645699786630107, "learning_rate": 9.564053324604305e-06, "loss": 2.3048, "step": 15990 }, { "epoch": 1.06, "grad_norm": 23.820527377120467, "learning_rate": 9.563391758906002e-06, "loss": 2.2838, "step": 16000 }, { "epoch": 1.0601, "grad_norm": 22.04155516304168, "learning_rate": 9.562729714529081e-06, "loss": 2.2773, "step": 16010 }, { "epoch": 1.0602, "grad_norm": 21.923900279683405, "learning_rate": 9.56206719154299e-06, "loss": 2.2726, "step": 16020 }, { "epoch": 1.0603, "grad_norm": 21.847162348796598, "learning_rate": 9.561404190017224e-06, "loss": 2.2769, "step": 16030 }, { "epoch": 1.0604, "grad_norm": 19.90387227227998, "learning_rate": 9.560740710021329e-06, "loss": 2.3249, "step": 16040 }, { "epoch": 1.0605, "grad_norm": 19.298272814113997, "learning_rate": 9.560076751624899e-06, "loss": 2.2179, "step": 16050 }, { "epoch": 1.0606, "grad_norm": 20.190066063707786, "learning_rate": 9.559412314897583e-06, "loss": 2.2955, "step": 16060 }, { "epoch": 1.0607, "grad_norm": 19.45444148360682, "learning_rate": 9.558747399909075e-06, "loss": 2.2788, "step": 16070 }, { "epoch": 1.0608, "grad_norm": 20.262029831622577, "learning_rate": 9.558082006729123e-06, "loss": 2.2952, "step": 16080 }, { "epoch": 1.0609, "grad_norm": 20.94899371772441, "learning_rate": 9.557416135427524e-06, "loss": 2.3078, "step": 16090 }, { "epoch": 1.061, "grad_norm": 20.093835666863573, "learning_rate": 9.556749786074122e-06, "loss": 2.2384, "step": 16100 }, { "epoch": 1.0611, "grad_norm": 20.08681029523278, "learning_rate": 9.556082958738818e-06, "loss": 2.2487, "step": 16110 }, { "epoch": 1.0612, "grad_norm": 21.077286135707574, "learning_rate": 9.555415653491558e-06, "loss": 2.3192, "step": 16120 }, { "epoch": 1.0613, "grad_norm": 20.61060580896263, "learning_rate": 9.554747870402336e-06, "loss": 2.2564, "step": 16130 }, { "epoch": 1.0614, "grad_norm": 21.62421053435006, "learning_rate": 9.554079609541202e-06, "loss": 2.197, "step": 16140 }, { "epoch": 1.0615, "grad_norm": 21.518410076662086, "learning_rate": 9.553410870978253e-06, "loss": 2.2427, "step": 16150 }, { "epoch": 1.0616, "grad_norm": 24.07847840354005, "learning_rate": 9.552741654783637e-06, "loss": 2.2854, "step": 16160 }, { "epoch": 1.0617, "grad_norm": 23.0517304841047, "learning_rate": 9.552071961027551e-06, "loss": 2.2581, "step": 16170 }, { "epoch": 1.0618, "grad_norm": 21.892833191430366, "learning_rate": 9.551401789780245e-06, "loss": 2.3046, "step": 16180 }, { "epoch": 1.0619, "grad_norm": 21.726909989300204, "learning_rate": 9.550731141112013e-06, "loss": 2.2873, "step": 16190 }, { "epoch": 1.062, "grad_norm": 22.319645714548688, "learning_rate": 9.550060015093205e-06, "loss": 2.334, "step": 16200 }, { "epoch": 1.0621, "grad_norm": 24.348611659321513, "learning_rate": 9.54938841179422e-06, "loss": 2.3381, "step": 16210 }, { "epoch": 1.0622, "grad_norm": 21.360731348943403, "learning_rate": 9.548716331285503e-06, "loss": 2.2696, "step": 16220 }, { "epoch": 1.0623, "grad_norm": 21.908440119102718, "learning_rate": 9.548043773637556e-06, "loss": 2.3044, "step": 16230 }, { "epoch": 1.0624, "grad_norm": 20.37152371432606, "learning_rate": 9.547370738920925e-06, "loss": 2.3154, "step": 16240 }, { "epoch": 1.0625, "grad_norm": 22.268046927613916, "learning_rate": 9.546697227206208e-06, "loss": 2.3275, "step": 16250 }, { "epoch": 1.0626, "grad_norm": 20.52456763701356, "learning_rate": 9.546023238564056e-06, "loss": 2.3156, "step": 16260 }, { "epoch": 1.0627, "grad_norm": 20.25179652156818, "learning_rate": 9.545348773065161e-06, "loss": 2.3125, "step": 16270 }, { "epoch": 1.0628, "grad_norm": 19.22598845041978, "learning_rate": 9.544673830780277e-06, "loss": 2.3584, "step": 16280 }, { "epoch": 1.0629, "grad_norm": 20.19043107470104, "learning_rate": 9.543998411780202e-06, "loss": 2.2768, "step": 16290 }, { "epoch": 1.063, "grad_norm": 20.323183751910317, "learning_rate": 9.543322516135782e-06, "loss": 2.2425, "step": 16300 }, { "epoch": 1.0631, "grad_norm": 20.746220654384942, "learning_rate": 9.542646143917914e-06, "loss": 2.1997, "step": 16310 }, { "epoch": 1.0632, "grad_norm": 25.50876669874953, "learning_rate": 9.541969295197551e-06, "loss": 2.2691, "step": 16320 }, { "epoch": 1.0633, "grad_norm": 20.305050193915072, "learning_rate": 9.541291970045689e-06, "loss": 2.2484, "step": 16330 }, { "epoch": 1.0634, "grad_norm": 22.613612791753916, "learning_rate": 9.540614168533376e-06, "loss": 2.3322, "step": 16340 }, { "epoch": 1.0635, "grad_norm": 19.620875404510063, "learning_rate": 9.539935890731711e-06, "loss": 2.234, "step": 16350 }, { "epoch": 1.0636, "grad_norm": 25.320763377793142, "learning_rate": 9.539257136711839e-06, "loss": 2.2355, "step": 16360 }, { "epoch": 1.0637, "grad_norm": 19.557598738144947, "learning_rate": 9.538577906544963e-06, "loss": 2.3904, "step": 16370 }, { "epoch": 1.0638, "grad_norm": 19.712927738568872, "learning_rate": 9.537898200302328e-06, "loss": 2.2365, "step": 16380 }, { "epoch": 1.0639, "grad_norm": 20.179517708730348, "learning_rate": 9.537218018055232e-06, "loss": 2.2991, "step": 16390 }, { "epoch": 1.064, "grad_norm": 20.725467007113267, "learning_rate": 9.536537359875026e-06, "loss": 2.2002, "step": 16400 }, { "epoch": 1.0641, "grad_norm": 22.322763388775524, "learning_rate": 9.535856225833105e-06, "loss": 2.2452, "step": 16410 }, { "epoch": 1.0642, "grad_norm": 21.55479266458448, "learning_rate": 9.535174616000915e-06, "loss": 2.2191, "step": 16420 }, { "epoch": 1.0643, "grad_norm": 20.529668769385825, "learning_rate": 9.53449253044996e-06, "loss": 2.3139, "step": 16430 }, { "epoch": 1.0644, "grad_norm": 20.24142714712095, "learning_rate": 9.53380996925178e-06, "loss": 2.2252, "step": 16440 }, { "epoch": 1.0645, "grad_norm": 20.57207534895006, "learning_rate": 9.533126932477978e-06, "loss": 2.194, "step": 16450 }, { "epoch": 1.0646, "grad_norm": 19.69397568903691, "learning_rate": 9.5324434202002e-06, "loss": 2.2444, "step": 16460 }, { "epoch": 1.0647, "grad_norm": 37.2886756692888, "learning_rate": 9.531759432490143e-06, "loss": 2.2113, "step": 16470 }, { "epoch": 1.0648, "grad_norm": 22.551889179580712, "learning_rate": 9.531074969419553e-06, "loss": 2.2615, "step": 16480 }, { "epoch": 1.0649, "grad_norm": 25.167741197521085, "learning_rate": 9.530390031060231e-06, "loss": 2.2605, "step": 16490 }, { "epoch": 1.065, "grad_norm": 20.050737129590352, "learning_rate": 9.529704617484017e-06, "loss": 2.2412, "step": 16500 }, { "epoch": 1.0651, "grad_norm": 22.994367239272048, "learning_rate": 9.529018728762813e-06, "loss": 2.2773, "step": 16510 }, { "epoch": 1.0652, "grad_norm": 21.3975206286401, "learning_rate": 9.528332364968566e-06, "loss": 2.259, "step": 16520 }, { "epoch": 1.0653, "grad_norm": 20.324316818713303, "learning_rate": 9.527645526173268e-06, "loss": 2.2708, "step": 16530 }, { "epoch": 1.0654, "grad_norm": 21.819692201668303, "learning_rate": 9.526958212448969e-06, "loss": 2.2957, "step": 16540 }, { "epoch": 1.0655000000000001, "grad_norm": 20.482402384927006, "learning_rate": 9.526270423867764e-06, "loss": 2.3265, "step": 16550 }, { "epoch": 1.0656, "grad_norm": 20.585569836983343, "learning_rate": 9.525582160501798e-06, "loss": 2.3084, "step": 16560 }, { "epoch": 1.0657, "grad_norm": 21.162080172530082, "learning_rate": 9.524893422423268e-06, "loss": 2.2849, "step": 16570 }, { "epoch": 1.0658, "grad_norm": 19.193623953263955, "learning_rate": 9.524204209704418e-06, "loss": 2.3296, "step": 16580 }, { "epoch": 1.0659, "grad_norm": 20.12763671237878, "learning_rate": 9.523514522417544e-06, "loss": 2.2522, "step": 16590 }, { "epoch": 1.066, "grad_norm": 21.117211900366915, "learning_rate": 9.522824360634991e-06, "loss": 2.3356, "step": 16600 }, { "epoch": 1.0661, "grad_norm": 20.412171987125728, "learning_rate": 9.522133724429155e-06, "loss": 2.3003, "step": 16610 }, { "epoch": 1.0662, "grad_norm": 21.257673607890347, "learning_rate": 9.521442613872476e-06, "loss": 2.2505, "step": 16620 }, { "epoch": 1.0663, "grad_norm": 21.421214850324848, "learning_rate": 9.520751029037455e-06, "loss": 2.3516, "step": 16630 }, { "epoch": 1.0664, "grad_norm": 20.13610207736905, "learning_rate": 9.52005896999663e-06, "loss": 2.3052, "step": 16640 }, { "epoch": 1.0665, "grad_norm": 22.205957481227305, "learning_rate": 9.519366436822597e-06, "loss": 2.3451, "step": 16650 }, { "epoch": 1.0666, "grad_norm": 21.70476201648675, "learning_rate": 9.518673429588001e-06, "loss": 2.2961, "step": 16660 }, { "epoch": 1.0667, "grad_norm": 23.065364475039694, "learning_rate": 9.517979948365535e-06, "loss": 2.2926, "step": 16670 }, { "epoch": 1.0668, "grad_norm": 19.36646435801435, "learning_rate": 9.517285993227939e-06, "loss": 2.3016, "step": 16680 }, { "epoch": 1.0669, "grad_norm": 19.074009327855205, "learning_rate": 9.51659156424801e-06, "loss": 2.2417, "step": 16690 }, { "epoch": 1.067, "grad_norm": 19.843934390853804, "learning_rate": 9.515896661498585e-06, "loss": 2.3054, "step": 16700 }, { "epoch": 1.0671, "grad_norm": 22.11052325728412, "learning_rate": 9.515201285052561e-06, "loss": 2.3151, "step": 16710 }, { "epoch": 1.0672, "grad_norm": 19.469452667179443, "learning_rate": 9.514505434982878e-06, "loss": 2.2056, "step": 16720 }, { "epoch": 1.0673, "grad_norm": 22.72922366408455, "learning_rate": 9.513809111362527e-06, "loss": 2.213, "step": 16730 }, { "epoch": 1.0674, "grad_norm": 21.63918056672638, "learning_rate": 9.51311231426455e-06, "loss": 2.3519, "step": 16740 }, { "epoch": 1.0675, "grad_norm": 22.254283285241943, "learning_rate": 9.512415043762035e-06, "loss": 2.2506, "step": 16750 }, { "epoch": 1.0676, "grad_norm": 22.875044660469378, "learning_rate": 9.511717299928126e-06, "loss": 2.2661, "step": 16760 }, { "epoch": 1.0677, "grad_norm": 19.592324205474075, "learning_rate": 9.511019082836014e-06, "loss": 2.3189, "step": 16770 }, { "epoch": 1.0678, "grad_norm": 21.178836586605343, "learning_rate": 9.510320392558933e-06, "loss": 2.2079, "step": 16780 }, { "epoch": 1.0679, "grad_norm": 21.355432211183906, "learning_rate": 9.509621229170178e-06, "loss": 2.285, "step": 16790 }, { "epoch": 1.068, "grad_norm": 22.42066656808085, "learning_rate": 9.508921592743086e-06, "loss": 2.309, "step": 16800 }, { "epoch": 1.0681, "grad_norm": 21.849675311407122, "learning_rate": 9.508221483351044e-06, "loss": 2.2349, "step": 16810 }, { "epoch": 1.0682, "grad_norm": 20.61201558068523, "learning_rate": 9.507520901067493e-06, "loss": 2.2099, "step": 16820 }, { "epoch": 1.0683, "grad_norm": 20.98751697785219, "learning_rate": 9.50681984596592e-06, "loss": 2.2727, "step": 16830 }, { "epoch": 1.0684, "grad_norm": 19.919833035123528, "learning_rate": 9.50611831811986e-06, "loss": 2.2581, "step": 16840 }, { "epoch": 1.0685, "grad_norm": 19.651124194945368, "learning_rate": 9.505416317602902e-06, "loss": 2.2248, "step": 16850 }, { "epoch": 1.0686, "grad_norm": 20.9445595730814, "learning_rate": 9.504713844488685e-06, "loss": 2.2536, "step": 16860 }, { "epoch": 1.0687, "grad_norm": 23.224031127282167, "learning_rate": 9.50401089885089e-06, "loss": 2.2881, "step": 16870 }, { "epoch": 1.0688, "grad_norm": 21.178597509894217, "learning_rate": 9.503307480763258e-06, "loss": 2.2327, "step": 16880 }, { "epoch": 1.0689, "grad_norm": 21.40572294172296, "learning_rate": 9.502603590299571e-06, "loss": 2.1938, "step": 16890 }, { "epoch": 1.069, "grad_norm": 20.5429849740656, "learning_rate": 9.501899227533665e-06, "loss": 2.1508, "step": 16900 }, { "epoch": 1.0691, "grad_norm": 21.89024316135195, "learning_rate": 9.501194392539422e-06, "loss": 2.2362, "step": 16910 }, { "epoch": 1.0692, "grad_norm": 19.919227334783344, "learning_rate": 9.500489085390779e-06, "loss": 2.2675, "step": 16920 }, { "epoch": 1.0693, "grad_norm": 21.048879617582134, "learning_rate": 9.49978330616172e-06, "loss": 2.2172, "step": 16930 }, { "epoch": 1.0694, "grad_norm": 20.30208308460738, "learning_rate": 9.499077054926274e-06, "loss": 2.3114, "step": 16940 }, { "epoch": 1.0695000000000001, "grad_norm": 19.57152435065954, "learning_rate": 9.498370331758524e-06, "loss": 2.1503, "step": 16950 }, { "epoch": 1.0695999999999999, "grad_norm": 21.102732175913367, "learning_rate": 9.497663136732608e-06, "loss": 2.2121, "step": 16960 }, { "epoch": 1.0697, "grad_norm": 20.585697435279332, "learning_rate": 9.496955469922701e-06, "loss": 2.1879, "step": 16970 }, { "epoch": 1.0698, "grad_norm": 20.625509820214827, "learning_rate": 9.496247331403037e-06, "loss": 2.2334, "step": 16980 }, { "epoch": 1.0699, "grad_norm": 21.128546639178726, "learning_rate": 9.495538721247894e-06, "loss": 2.2553, "step": 16990 }, { "epoch": 1.07, "grad_norm": 19.759217302415102, "learning_rate": 9.494829639531607e-06, "loss": 2.2227, "step": 17000 }, { "epoch": 1.0701, "grad_norm": 19.25962538499256, "learning_rate": 9.494120086328549e-06, "loss": 2.2367, "step": 17010 }, { "epoch": 1.0702, "grad_norm": 22.377709630811992, "learning_rate": 9.493410061713154e-06, "loss": 2.2617, "step": 17020 }, { "epoch": 1.0703, "grad_norm": 21.243133335484004, "learning_rate": 9.492699565759896e-06, "loss": 2.243, "step": 17030 }, { "epoch": 1.0704, "grad_norm": 21.589450637339237, "learning_rate": 9.491988598543305e-06, "loss": 2.2569, "step": 17040 }, { "epoch": 1.0705, "grad_norm": 21.583652502513274, "learning_rate": 9.491277160137959e-06, "loss": 2.2158, "step": 17050 }, { "epoch": 1.0706, "grad_norm": 18.934594058801483, "learning_rate": 9.490565250618485e-06, "loss": 2.2623, "step": 17060 }, { "epoch": 1.0707, "grad_norm": 19.19971696566494, "learning_rate": 9.489852870059556e-06, "loss": 2.2649, "step": 17070 }, { "epoch": 1.0708, "grad_norm": 19.81730362436409, "learning_rate": 9.489140018535898e-06, "loss": 2.2196, "step": 17080 }, { "epoch": 1.0709, "grad_norm": 20.688175192806273, "learning_rate": 9.488426696122286e-06, "loss": 2.1969, "step": 17090 }, { "epoch": 1.071, "grad_norm": 20.69435547631227, "learning_rate": 9.487712902893547e-06, "loss": 2.1209, "step": 17100 }, { "epoch": 1.0711, "grad_norm": 22.71790103966948, "learning_rate": 9.486998638924552e-06, "loss": 2.1755, "step": 17110 }, { "epoch": 1.0712, "grad_norm": 19.36830452720831, "learning_rate": 9.486283904290226e-06, "loss": 2.1686, "step": 17120 }, { "epoch": 1.0713, "grad_norm": 20.735696147635444, "learning_rate": 9.485568699065538e-06, "loss": 2.3096, "step": 17130 }, { "epoch": 1.0714, "grad_norm": 20.639188445328255, "learning_rate": 9.484853023325511e-06, "loss": 2.2128, "step": 17140 }, { "epoch": 1.0715, "grad_norm": 20.117755868630667, "learning_rate": 9.484136877145219e-06, "loss": 2.2362, "step": 17150 }, { "epoch": 1.0716, "grad_norm": 18.336705532844956, "learning_rate": 9.483420260599779e-06, "loss": 2.144, "step": 17160 }, { "epoch": 1.0717, "grad_norm": 20.051469127480665, "learning_rate": 9.482703173764362e-06, "loss": 2.187, "step": 17170 }, { "epoch": 1.0718, "grad_norm": 21.44119216722438, "learning_rate": 9.481985616714186e-06, "loss": 2.2657, "step": 17180 }, { "epoch": 1.0719, "grad_norm": 20.213437850471117, "learning_rate": 9.481267589524523e-06, "loss": 2.2131, "step": 17190 }, { "epoch": 1.072, "grad_norm": 21.076401782467528, "learning_rate": 9.480549092270686e-06, "loss": 2.2321, "step": 17200 }, { "epoch": 1.0721, "grad_norm": 20.305097781495864, "learning_rate": 9.479830125028044e-06, "loss": 2.1265, "step": 17210 }, { "epoch": 1.0722, "grad_norm": 20.208017956224552, "learning_rate": 9.479110687872014e-06, "loss": 2.2054, "step": 17220 }, { "epoch": 1.0723, "grad_norm": 22.33837294042243, "learning_rate": 9.478390780878061e-06, "loss": 2.2339, "step": 17230 }, { "epoch": 1.0724, "grad_norm": 19.288857746142362, "learning_rate": 9.4776704041217e-06, "loss": 2.1605, "step": 17240 }, { "epoch": 1.0725, "grad_norm": 19.73948522487547, "learning_rate": 9.476949557678495e-06, "loss": 2.2113, "step": 17250 }, { "epoch": 1.0726, "grad_norm": 21.51485004752134, "learning_rate": 9.476228241624059e-06, "loss": 2.2122, "step": 17260 }, { "epoch": 1.0727, "grad_norm": 20.22824511939977, "learning_rate": 9.475506456034056e-06, "loss": 2.1672, "step": 17270 }, { "epoch": 1.0728, "grad_norm": 20.742002070926016, "learning_rate": 9.474784200984197e-06, "loss": 2.2792, "step": 17280 }, { "epoch": 1.0729, "grad_norm": 20.79528006718728, "learning_rate": 9.474061476550243e-06, "loss": 2.1991, "step": 17290 }, { "epoch": 1.073, "grad_norm": 20.862656620421525, "learning_rate": 9.473338282808006e-06, "loss": 2.2079, "step": 17300 }, { "epoch": 1.0731, "grad_norm": 20.781253247747593, "learning_rate": 9.472614619833344e-06, "loss": 2.191, "step": 17310 }, { "epoch": 1.0732, "grad_norm": 19.876974422253312, "learning_rate": 9.471890487702166e-06, "loss": 2.3061, "step": 17320 }, { "epoch": 1.0733, "grad_norm": 20.083909089802184, "learning_rate": 9.47116588649043e-06, "loss": 2.2606, "step": 17330 }, { "epoch": 1.0734, "grad_norm": 20.600035322720398, "learning_rate": 9.470440816274145e-06, "loss": 2.2048, "step": 17340 }, { "epoch": 1.0735, "grad_norm": 21.884726344427495, "learning_rate": 9.469715277129366e-06, "loss": 2.2751, "step": 17350 }, { "epoch": 1.0735999999999999, "grad_norm": 19.893678028123436, "learning_rate": 9.4689892691322e-06, "loss": 2.1518, "step": 17360 }, { "epoch": 1.0737, "grad_norm": 19.363050888243222, "learning_rate": 9.468262792358798e-06, "loss": 2.2461, "step": 17370 }, { "epoch": 1.0738, "grad_norm": 20.40713722149602, "learning_rate": 9.46753584688537e-06, "loss": 2.1925, "step": 17380 }, { "epoch": 1.0739, "grad_norm": 19.437769054259743, "learning_rate": 9.466808432788165e-06, "loss": 2.235, "step": 17390 }, { "epoch": 1.074, "grad_norm": 19.403425029936457, "learning_rate": 9.466080550143486e-06, "loss": 2.1396, "step": 17400 }, { "epoch": 1.0741, "grad_norm": 18.551398961058098, "learning_rate": 9.465352199027687e-06, "loss": 2.202, "step": 17410 }, { "epoch": 1.0742, "grad_norm": 20.592625620656488, "learning_rate": 9.464623379517165e-06, "loss": 2.2186, "step": 17420 }, { "epoch": 1.0743, "grad_norm": 21.90412760393634, "learning_rate": 9.46389409168837e-06, "loss": 2.2599, "step": 17430 }, { "epoch": 1.0744, "grad_norm": 20.055484984053443, "learning_rate": 9.463164335617806e-06, "loss": 2.2071, "step": 17440 }, { "epoch": 1.0745, "grad_norm": 20.60793052563804, "learning_rate": 9.462434111382013e-06, "loss": 2.1357, "step": 17450 }, { "epoch": 1.0746, "grad_norm": 19.83045683991372, "learning_rate": 9.461703419057595e-06, "loss": 2.2487, "step": 17460 }, { "epoch": 1.0747, "grad_norm": 17.53628177239973, "learning_rate": 9.460972258721196e-06, "loss": 2.2252, "step": 17470 }, { "epoch": 1.0748, "grad_norm": 18.71054477810363, "learning_rate": 9.46024063044951e-06, "loss": 2.1639, "step": 17480 }, { "epoch": 1.0749, "grad_norm": 21.692589357146456, "learning_rate": 9.459508534319284e-06, "loss": 2.1213, "step": 17490 }, { "epoch": 1.075, "grad_norm": 21.004204022871985, "learning_rate": 9.458775970407307e-06, "loss": 2.2162, "step": 17500 }, { "epoch": 1.0751, "grad_norm": 17.267165763792082, "learning_rate": 9.458042938790429e-06, "loss": 2.0787, "step": 17510 }, { "epoch": 1.0752, "grad_norm": 20.50412102990186, "learning_rate": 9.457309439545535e-06, "loss": 2.1552, "step": 17520 }, { "epoch": 1.0753, "grad_norm": 21.75024729245215, "learning_rate": 9.456575472749566e-06, "loss": 2.0867, "step": 17530 }, { "epoch": 1.0754, "grad_norm": 20.515850265681593, "learning_rate": 9.455841038479515e-06, "loss": 2.196, "step": 17540 }, { "epoch": 1.0755, "grad_norm": 18.727187866823076, "learning_rate": 9.45510613681242e-06, "loss": 2.1349, "step": 17550 }, { "epoch": 1.0756000000000001, "grad_norm": 19.985649023359603, "learning_rate": 9.454370767825367e-06, "loss": 2.1043, "step": 17560 }, { "epoch": 1.0757, "grad_norm": 21.32924421442994, "learning_rate": 9.453634931595494e-06, "loss": 2.1915, "step": 17570 }, { "epoch": 1.0758, "grad_norm": 21.17478483456811, "learning_rate": 9.452898628199986e-06, "loss": 2.1567, "step": 17580 }, { "epoch": 1.0759, "grad_norm": 18.50111613685343, "learning_rate": 9.45216185771608e-06, "loss": 2.1338, "step": 17590 }, { "epoch": 1.076, "grad_norm": 19.184956080261976, "learning_rate": 9.451424620221058e-06, "loss": 2.1519, "step": 17600 }, { "epoch": 1.0761, "grad_norm": 20.238392290112206, "learning_rate": 9.450686915792252e-06, "loss": 2.1502, "step": 17610 }, { "epoch": 1.0762, "grad_norm": 20.558572055077075, "learning_rate": 9.449948744507045e-06, "loss": 2.1258, "step": 17620 }, { "epoch": 1.0763, "grad_norm": 19.942656547370895, "learning_rate": 9.449210106442868e-06, "loss": 2.2281, "step": 17630 }, { "epoch": 1.0764, "grad_norm": 19.44987815460152, "learning_rate": 9.448471001677202e-06, "loss": 2.1747, "step": 17640 }, { "epoch": 1.0765, "grad_norm": 20.597896435751032, "learning_rate": 9.44773143028757e-06, "loss": 2.1286, "step": 17650 }, { "epoch": 1.0766, "grad_norm": 19.87055261378047, "learning_rate": 9.446991392351556e-06, "loss": 2.0788, "step": 17660 }, { "epoch": 1.0767, "grad_norm": 22.024087946042957, "learning_rate": 9.446250887946785e-06, "loss": 2.13, "step": 17670 }, { "epoch": 1.0768, "grad_norm": 18.841813150498844, "learning_rate": 9.445509917150928e-06, "loss": 2.0392, "step": 17680 }, { "epoch": 1.0769, "grad_norm": 20.398447303444282, "learning_rate": 9.444768480041717e-06, "loss": 2.0825, "step": 17690 }, { "epoch": 1.077, "grad_norm": 21.116743414200386, "learning_rate": 9.444026576696918e-06, "loss": 2.19, "step": 17700 }, { "epoch": 1.0771, "grad_norm": 21.15821080520982, "learning_rate": 9.443284207194358e-06, "loss": 2.1591, "step": 17710 }, { "epoch": 1.0772, "grad_norm": 20.627879057808773, "learning_rate": 9.442541371611906e-06, "loss": 2.1306, "step": 17720 }, { "epoch": 1.0773, "grad_norm": 19.974605351412517, "learning_rate": 9.441798070027483e-06, "loss": 2.079, "step": 17730 }, { "epoch": 2.00008, "grad_norm": 25.516720363581317, "learning_rate": 9.441054302519058e-06, "loss": 2.4887, "step": 17740 }, { "epoch": 2.00018, "grad_norm": 20.293051651185788, "learning_rate": 9.440310069164649e-06, "loss": 2.3488, "step": 17750 }, { "epoch": 2.00028, "grad_norm": 19.268537071248307, "learning_rate": 9.439565370042321e-06, "loss": 2.3311, "step": 17760 }, { "epoch": 2.00038, "grad_norm": 22.517631871086927, "learning_rate": 9.438820205230188e-06, "loss": 2.2667, "step": 17770 }, { "epoch": 2.00048, "grad_norm": 21.222140906569834, "learning_rate": 9.43807457480642e-06, "loss": 2.3382, "step": 17780 }, { "epoch": 2.00058, "grad_norm": 20.27359615199726, "learning_rate": 9.437328478849225e-06, "loss": 2.2875, "step": 17790 }, { "epoch": 2.00068, "grad_norm": 19.639594881439464, "learning_rate": 9.436581917436867e-06, "loss": 2.226, "step": 17800 }, { "epoch": 2.00078, "grad_norm": 21.275623361040786, "learning_rate": 9.435834890647656e-06, "loss": 2.1541, "step": 17810 }, { "epoch": 2.00088, "grad_norm": 24.669960213608768, "learning_rate": 9.435087398559954e-06, "loss": 2.279, "step": 17820 }, { "epoch": 2.00098, "grad_norm": 20.119403772743937, "learning_rate": 9.434339441252167e-06, "loss": 2.1767, "step": 17830 }, { "epoch": 2.00108, "grad_norm": 20.560785405420184, "learning_rate": 9.433591018802753e-06, "loss": 2.2601, "step": 17840 }, { "epoch": 2.00118, "grad_norm": 20.515240417553187, "learning_rate": 9.432842131290219e-06, "loss": 2.1853, "step": 17850 }, { "epoch": 2.00128, "grad_norm": 20.471117249290412, "learning_rate": 9.432092778793119e-06, "loss": 2.1978, "step": 17860 }, { "epoch": 2.00138, "grad_norm": 19.71767415599374, "learning_rate": 9.431342961390055e-06, "loss": 2.1546, "step": 17870 }, { "epoch": 2.00148, "grad_norm": 21.658086376966608, "learning_rate": 9.43059267915968e-06, "loss": 2.1196, "step": 17880 }, { "epoch": 2.00158, "grad_norm": 20.927026258384466, "learning_rate": 9.429841932180698e-06, "loss": 2.1862, "step": 17890 }, { "epoch": 2.00168, "grad_norm": 20.485473965084367, "learning_rate": 9.429090720531857e-06, "loss": 2.1834, "step": 17900 }, { "epoch": 2.00178, "grad_norm": 19.65327293006979, "learning_rate": 9.428339044291955e-06, "loss": 2.1636, "step": 17910 }, { "epoch": 2.00188, "grad_norm": 22.328865560072764, "learning_rate": 9.427586903539838e-06, "loss": 2.2155, "step": 17920 }, { "epoch": 2.00198, "grad_norm": 21.510693240853485, "learning_rate": 9.426834298354405e-06, "loss": 2.1553, "step": 17930 }, { "epoch": 2.00208, "grad_norm": 18.528227124052687, "learning_rate": 9.426081228814601e-06, "loss": 2.1791, "step": 17940 }, { "epoch": 2.00218, "grad_norm": 21.515519142109834, "learning_rate": 9.425327694999417e-06, "loss": 2.1698, "step": 17950 }, { "epoch": 2.00228, "grad_norm": 21.502362330755776, "learning_rate": 9.424573696987897e-06, "loss": 2.1765, "step": 17960 }, { "epoch": 2.00238, "grad_norm": 21.04708794035362, "learning_rate": 9.42381923485913e-06, "loss": 2.2073, "step": 17970 }, { "epoch": 2.00248, "grad_norm": 20.194296928776257, "learning_rate": 9.423064308692258e-06, "loss": 2.2231, "step": 17980 }, { "epoch": 2.00258, "grad_norm": 20.76306538347405, "learning_rate": 9.422308918566467e-06, "loss": 2.2017, "step": 17990 }, { "epoch": 2.00268, "grad_norm": 17.584917052463695, "learning_rate": 9.421553064560996e-06, "loss": 2.1491, "step": 18000 }, { "epoch": 2.00278, "grad_norm": 20.55663265346591, "learning_rate": 9.42079674675513e-06, "loss": 2.1917, "step": 18010 }, { "epoch": 2.00288, "grad_norm": 20.71880181226246, "learning_rate": 9.420039965228202e-06, "loss": 2.1999, "step": 18020 }, { "epoch": 2.00298, "grad_norm": 21.96423461773572, "learning_rate": 9.419282720059595e-06, "loss": 2.1584, "step": 18030 }, { "epoch": 2.00308, "grad_norm": 20.257723736889737, "learning_rate": 9.418525011328744e-06, "loss": 2.1425, "step": 18040 }, { "epoch": 2.00318, "grad_norm": 21.569145887170656, "learning_rate": 9.417766839115124e-06, "loss": 2.1858, "step": 18050 }, { "epoch": 2.00328, "grad_norm": 18.798660667600664, "learning_rate": 9.417008203498266e-06, "loss": 2.0461, "step": 18060 }, { "epoch": 2.00338, "grad_norm": 19.459490047320223, "learning_rate": 9.416249104557747e-06, "loss": 2.1557, "step": 18070 }, { "epoch": 2.00348, "grad_norm": 21.37536041175924, "learning_rate": 9.415489542373195e-06, "loss": 2.1385, "step": 18080 }, { "epoch": 2.00358, "grad_norm": 23.129358973512872, "learning_rate": 9.414729517024282e-06, "loss": 2.2013, "step": 18090 }, { "epoch": 2.00368, "grad_norm": 19.419577404269752, "learning_rate": 9.413969028590733e-06, "loss": 2.1131, "step": 18100 }, { "epoch": 2.00378, "grad_norm": 21.463568951094075, "learning_rate": 9.413208077152316e-06, "loss": 2.1281, "step": 18110 }, { "epoch": 2.00388, "grad_norm": 20.991597482326675, "learning_rate": 9.412446662788857e-06, "loss": 2.1353, "step": 18120 }, { "epoch": 2.00398, "grad_norm": 21.396461411476263, "learning_rate": 9.411684785580222e-06, "loss": 2.1818, "step": 18130 }, { "epoch": 2.00408, "grad_norm": 21.301799300036357, "learning_rate": 9.410922445606326e-06, "loss": 2.1575, "step": 18140 }, { "epoch": 2.00418, "grad_norm": 20.743446105930865, "learning_rate": 9.410159642947138e-06, "loss": 2.1596, "step": 18150 }, { "epoch": 2.00428, "grad_norm": 19.47723355372513, "learning_rate": 9.409396377682673e-06, "loss": 2.1299, "step": 18160 }, { "epoch": 2.00438, "grad_norm": 19.450364073553494, "learning_rate": 9.40863264989299e-06, "loss": 2.1181, "step": 18170 }, { "epoch": 2.00448, "grad_norm": 21.594479416170127, "learning_rate": 9.407868459658207e-06, "loss": 2.1122, "step": 18180 }, { "epoch": 2.00458, "grad_norm": 19.616520430817904, "learning_rate": 9.407103807058476e-06, "loss": 2.1488, "step": 18190 }, { "epoch": 2.00468, "grad_norm": 21.336884105179603, "learning_rate": 9.406338692174012e-06, "loss": 2.1274, "step": 18200 }, { "epoch": 2.00478, "grad_norm": 22.157318606314846, "learning_rate": 9.405573115085069e-06, "loss": 2.1516, "step": 18210 }, { "epoch": 2.00488, "grad_norm": 20.27286524099686, "learning_rate": 9.404807075871954e-06, "loss": 2.1201, "step": 18220 }, { "epoch": 2.00498, "grad_norm": 20.241073803159576, "learning_rate": 9.404040574615018e-06, "loss": 2.1723, "step": 18230 }, { "epoch": 2.00508, "grad_norm": 18.904273109490045, "learning_rate": 9.403273611394668e-06, "loss": 2.0621, "step": 18240 }, { "epoch": 2.00518, "grad_norm": 23.519841532716764, "learning_rate": 9.402506186291351e-06, "loss": 2.0357, "step": 18250 }, { "epoch": 2.00528, "grad_norm": 20.63419113035106, "learning_rate": 9.401738299385568e-06, "loss": 2.2304, "step": 18260 }, { "epoch": 2.00538, "grad_norm": 20.608880777472518, "learning_rate": 9.400969950757865e-06, "loss": 2.1917, "step": 18270 }, { "epoch": 2.00548, "grad_norm": 19.453396642722918, "learning_rate": 9.400201140488842e-06, "loss": 2.1356, "step": 18280 }, { "epoch": 2.00558, "grad_norm": 21.42097297627571, "learning_rate": 9.399431868659142e-06, "loss": 2.1147, "step": 18290 }, { "epoch": 2.00568, "grad_norm": 20.035024153310815, "learning_rate": 9.398662135349454e-06, "loss": 2.2281, "step": 18300 }, { "epoch": 2.00578, "grad_norm": 20.397478366415456, "learning_rate": 9.397891940640526e-06, "loss": 2.0897, "step": 18310 }, { "epoch": 2.00588, "grad_norm": 20.182208059420056, "learning_rate": 9.397121284613142e-06, "loss": 2.0839, "step": 18320 }, { "epoch": 2.00598, "grad_norm": 21.446577740939897, "learning_rate": 9.396350167348145e-06, "loss": 2.1295, "step": 18330 }, { "epoch": 2.00608, "grad_norm": 22.119187488335736, "learning_rate": 9.39557858892642e-06, "loss": 2.0775, "step": 18340 }, { "epoch": 2.00618, "grad_norm": 18.737010729247075, "learning_rate": 9.394806549428899e-06, "loss": 2.0772, "step": 18350 }, { "epoch": 2.00628, "grad_norm": 20.347459328621564, "learning_rate": 9.39403404893657e-06, "loss": 2.176, "step": 18360 }, { "epoch": 2.00638, "grad_norm": 19.646051457600517, "learning_rate": 9.39326108753046e-06, "loss": 2.1014, "step": 18370 }, { "epoch": 2.00648, "grad_norm": 22.23484633930676, "learning_rate": 9.392487665291654e-06, "loss": 2.0558, "step": 18380 }, { "epoch": 2.00658, "grad_norm": 21.308504305555076, "learning_rate": 9.391713782301279e-06, "loss": 2.0643, "step": 18390 }, { "epoch": 2.00668, "grad_norm": 19.964768564787825, "learning_rate": 9.390939438640508e-06, "loss": 2.1448, "step": 18400 }, { "epoch": 2.00678, "grad_norm": 18.769123416266066, "learning_rate": 9.390164634390571e-06, "loss": 2.098, "step": 18410 }, { "epoch": 2.00688, "grad_norm": 21.657218880587386, "learning_rate": 9.389389369632737e-06, "loss": 2.1235, "step": 18420 }, { "epoch": 2.00698, "grad_norm": 21.083717620504412, "learning_rate": 9.388613644448332e-06, "loss": 2.2231, "step": 18430 }, { "epoch": 2.00708, "grad_norm": 19.29439980358174, "learning_rate": 9.387837458918725e-06, "loss": 2.1792, "step": 18440 }, { "epoch": 2.00718, "grad_norm": 20.90637464931983, "learning_rate": 9.38706081312533e-06, "loss": 2.1698, "step": 18450 }, { "epoch": 2.00728, "grad_norm": 20.43475887667099, "learning_rate": 9.386283707149619e-06, "loss": 2.16, "step": 18460 }, { "epoch": 2.00738, "grad_norm": 18.984266754138982, "learning_rate": 9.385506141073104e-06, "loss": 2.1561, "step": 18470 }, { "epoch": 2.00748, "grad_norm": 20.48741930959714, "learning_rate": 9.384728114977347e-06, "loss": 2.1079, "step": 18480 }, { "epoch": 2.00758, "grad_norm": 22.017905183851443, "learning_rate": 9.383949628943963e-06, "loss": 2.1174, "step": 18490 }, { "epoch": 2.00768, "grad_norm": 20.16033947007011, "learning_rate": 9.383170683054607e-06, "loss": 2.0825, "step": 18500 }, { "epoch": 2.00778, "grad_norm": 20.44767024344433, "learning_rate": 9.382391277390992e-06, "loss": 2.0357, "step": 18510 }, { "epoch": 2.00788, "grad_norm": 21.89010141707249, "learning_rate": 9.38161141203487e-06, "loss": 2.0871, "step": 18520 }, { "epoch": 2.00798, "grad_norm": 21.351812400106553, "learning_rate": 9.380831087068046e-06, "loss": 2.1496, "step": 18530 }, { "epoch": 2.00808, "grad_norm": 19.20074761855518, "learning_rate": 9.380050302572375e-06, "loss": 2.108, "step": 18540 }, { "epoch": 2.00818, "grad_norm": 22.910804000221045, "learning_rate": 9.379269058629752e-06, "loss": 2.1134, "step": 18550 }, { "epoch": 2.00828, "grad_norm": 20.784032438802296, "learning_rate": 9.378487355322133e-06, "loss": 2.1026, "step": 18560 }, { "epoch": 2.00838, "grad_norm": 21.148512033066872, "learning_rate": 9.37770519273151e-06, "loss": 2.077, "step": 18570 }, { "epoch": 2.00848, "grad_norm": 21.57854954375592, "learning_rate": 9.37692257093993e-06, "loss": 2.054, "step": 18580 }, { "epoch": 2.00858, "grad_norm": 21.52798511405371, "learning_rate": 9.376139490029488e-06, "loss": 2.2062, "step": 18590 }, { "epoch": 2.00868, "grad_norm": 23.50739438832569, "learning_rate": 9.375355950082322e-06, "loss": 2.2228, "step": 18600 }, { "epoch": 2.00878, "grad_norm": 19.885862664370794, "learning_rate": 9.374571951180624e-06, "loss": 2.1285, "step": 18610 }, { "epoch": 2.00888, "grad_norm": 19.983634607241978, "learning_rate": 9.373787493406632e-06, "loss": 2.1564, "step": 18620 }, { "epoch": 2.00898, "grad_norm": 19.954552223585935, "learning_rate": 9.37300257684263e-06, "loss": 2.094, "step": 18630 }, { "epoch": 2.00908, "grad_norm": 22.01272858750007, "learning_rate": 9.372217201570954e-06, "loss": 2.1426, "step": 18640 }, { "epoch": 2.00918, "grad_norm": 21.04473336814992, "learning_rate": 9.371431367673986e-06, "loss": 2.1137, "step": 18650 }, { "epoch": 2.00928, "grad_norm": 21.54230698759729, "learning_rate": 9.370645075234155e-06, "loss": 2.0914, "step": 18660 }, { "epoch": 2.00938, "grad_norm": 20.675681302471535, "learning_rate": 9.369858324333942e-06, "loss": 2.061, "step": 18670 }, { "epoch": 2.00948, "grad_norm": 21.516416220858215, "learning_rate": 9.36907111505587e-06, "loss": 2.0743, "step": 18680 }, { "epoch": 2.00958, "grad_norm": 21.11465481922855, "learning_rate": 9.368283447482517e-06, "loss": 2.1055, "step": 18690 }, { "epoch": 2.00968, "grad_norm": 19.43194710391026, "learning_rate": 9.367495321696504e-06, "loss": 2.0933, "step": 18700 }, { "epoch": 2.00978, "grad_norm": 21.10293516210816, "learning_rate": 9.366706737780501e-06, "loss": 2.1343, "step": 18710 }, { "epoch": 2.00988, "grad_norm": 21.331292773994676, "learning_rate": 9.365917695817228e-06, "loss": 2.1358, "step": 18720 }, { "epoch": 2.00998, "grad_norm": 19.611854545608015, "learning_rate": 9.365128195889452e-06, "loss": 2.0907, "step": 18730 }, { "epoch": 2.01008, "grad_norm": 22.129589224012832, "learning_rate": 9.364338238079987e-06, "loss": 2.015, "step": 18740 }, { "epoch": 2.01018, "grad_norm": 23.096650879217965, "learning_rate": 9.363547822471697e-06, "loss": 2.1356, "step": 18750 }, { "epoch": 2.01028, "grad_norm": 19.442857985947004, "learning_rate": 9.362756949147491e-06, "loss": 2.0484, "step": 18760 }, { "epoch": 2.01038, "grad_norm": 21.32237210446752, "learning_rate": 9.361965618190332e-06, "loss": 2.1097, "step": 18770 }, { "epoch": 2.01048, "grad_norm": 21.65062008606395, "learning_rate": 9.361173829683221e-06, "loss": 2.0771, "step": 18780 }, { "epoch": 2.01058, "grad_norm": 21.475386142419495, "learning_rate": 9.360381583709217e-06, "loss": 2.1182, "step": 18790 }, { "epoch": 2.01068, "grad_norm": 20.218448977706682, "learning_rate": 9.359588880351422e-06, "loss": 2.1562, "step": 18800 }, { "epoch": 2.01078, "grad_norm": 19.66779954374081, "learning_rate": 9.358795719692988e-06, "loss": 2.0647, "step": 18810 }, { "epoch": 2.0108800000000002, "grad_norm": 20.3054723745376, "learning_rate": 9.358002101817112e-06, "loss": 2.0958, "step": 18820 }, { "epoch": 2.01098, "grad_norm": 20.19599188224446, "learning_rate": 9.357208026807042e-06, "loss": 2.0533, "step": 18830 }, { "epoch": 2.01108, "grad_norm": 22.82396879310755, "learning_rate": 9.356413494746072e-06, "loss": 2.0386, "step": 18840 }, { "epoch": 2.01118, "grad_norm": 20.078654852146144, "learning_rate": 9.355618505717545e-06, "loss": 2.0755, "step": 18850 }, { "epoch": 2.01128, "grad_norm": 22.10618675911664, "learning_rate": 9.354823059804853e-06, "loss": 2.029, "step": 18860 }, { "epoch": 2.01138, "grad_norm": 20.35935046655356, "learning_rate": 9.354027157091433e-06, "loss": 2.1135, "step": 18870 }, { "epoch": 2.01148, "grad_norm": 19.94176355953663, "learning_rate": 9.35323079766077e-06, "loss": 2.0785, "step": 18880 }, { "epoch": 2.01158, "grad_norm": 20.98334987368305, "learning_rate": 9.352433981596403e-06, "loss": 2.0322, "step": 18890 }, { "epoch": 2.01168, "grad_norm": 20.50358756911303, "learning_rate": 9.351636708981909e-06, "loss": 2.0378, "step": 18900 }, { "epoch": 2.01178, "grad_norm": 22.009819731598743, "learning_rate": 9.350838979900923e-06, "loss": 2.0281, "step": 18910 }, { "epoch": 2.01188, "grad_norm": 19.58110885902401, "learning_rate": 9.35004079443712e-06, "loss": 2.0112, "step": 18920 }, { "epoch": 2.01198, "grad_norm": 23.153100891534354, "learning_rate": 9.349242152674228e-06, "loss": 2.1035, "step": 18930 }, { "epoch": 2.01208, "grad_norm": 21.32422008485755, "learning_rate": 9.348443054696017e-06, "loss": 2.1121, "step": 18940 }, { "epoch": 2.01218, "grad_norm": 22.085447091914563, "learning_rate": 9.347643500586312e-06, "loss": 2.0121, "step": 18950 }, { "epoch": 2.01228, "grad_norm": 20.987073154911666, "learning_rate": 9.346843490428982e-06, "loss": 2.0379, "step": 18960 }, { "epoch": 2.01238, "grad_norm": 22.52203597928885, "learning_rate": 9.346043024307946e-06, "loss": 2.0749, "step": 18970 }, { "epoch": 2.01248, "grad_norm": 21.0037113266792, "learning_rate": 9.345242102307167e-06, "loss": 2.0829, "step": 18980 }, { "epoch": 2.01258, "grad_norm": 21.741362566677765, "learning_rate": 9.344440724510656e-06, "loss": 2.1148, "step": 18990 }, { "epoch": 2.01268, "grad_norm": 19.18230414567883, "learning_rate": 9.343638891002477e-06, "loss": 2.0288, "step": 19000 }, { "epoch": 2.01278, "grad_norm": 21.349478552453743, "learning_rate": 9.34283660186674e-06, "loss": 2.0217, "step": 19010 }, { "epoch": 2.01288, "grad_norm": 18.70467013024421, "learning_rate": 9.342033857187597e-06, "loss": 2.0667, "step": 19020 }, { "epoch": 2.01298, "grad_norm": 20.18745255479525, "learning_rate": 9.341230657049254e-06, "loss": 2.0549, "step": 19030 }, { "epoch": 2.01308, "grad_norm": 21.000357006381666, "learning_rate": 9.340427001535966e-06, "loss": 2.0643, "step": 19040 }, { "epoch": 2.01318, "grad_norm": 20.470981477051655, "learning_rate": 9.339622890732027e-06, "loss": 2.0506, "step": 19050 }, { "epoch": 2.01328, "grad_norm": 22.7426607503996, "learning_rate": 9.33881832472179e-06, "loss": 1.9948, "step": 19060 }, { "epoch": 2.01338, "grad_norm": 20.994669818707607, "learning_rate": 9.338013303589646e-06, "loss": 2.0261, "step": 19070 }, { "epoch": 2.01348, "grad_norm": 19.992477028485663, "learning_rate": 9.33720782742004e-06, "loss": 1.9572, "step": 19080 }, { "epoch": 2.01358, "grad_norm": 19.676936900804986, "learning_rate": 9.336401896297465e-06, "loss": 2.069, "step": 19090 }, { "epoch": 2.01368, "grad_norm": 18.831013248718556, "learning_rate": 9.335595510306455e-06, "loss": 1.9787, "step": 19100 }, { "epoch": 2.01378, "grad_norm": 21.895562410729347, "learning_rate": 9.334788669531597e-06, "loss": 2.0819, "step": 19110 }, { "epoch": 2.01388, "grad_norm": 21.34295519850289, "learning_rate": 9.333981374057527e-06, "loss": 2.0211, "step": 19120 }, { "epoch": 2.01398, "grad_norm": 20.155914141808704, "learning_rate": 9.333173623968925e-06, "loss": 2.0997, "step": 19130 }, { "epoch": 2.01408, "grad_norm": 19.18260007378981, "learning_rate": 9.332365419350522e-06, "loss": 1.9629, "step": 19140 }, { "epoch": 2.01418, "grad_norm": 21.159058699878003, "learning_rate": 9.331556760287092e-06, "loss": 2.0873, "step": 19150 }, { "epoch": 2.01428, "grad_norm": 20.89478468984572, "learning_rate": 9.330747646863463e-06, "loss": 2.0587, "step": 19160 }, { "epoch": 2.01438, "grad_norm": 21.076930187425052, "learning_rate": 9.329938079164505e-06, "loss": 2.052, "step": 19170 }, { "epoch": 2.01448, "grad_norm": 21.75204925747861, "learning_rate": 9.329128057275138e-06, "loss": 2.1023, "step": 19180 }, { "epoch": 2.01458, "grad_norm": 22.14311656093646, "learning_rate": 9.328317581280331e-06, "loss": 2.0608, "step": 19190 }, { "epoch": 2.01468, "grad_norm": 21.56432177451539, "learning_rate": 9.327506651265096e-06, "loss": 2.0502, "step": 19200 }, { "epoch": 2.01478, "grad_norm": 20.440999742852938, "learning_rate": 9.326695267314499e-06, "loss": 2.0713, "step": 19210 }, { "epoch": 2.01488, "grad_norm": 18.250998277016034, "learning_rate": 9.32588342951365e-06, "loss": 2.082, "step": 19220 }, { "epoch": 2.01498, "grad_norm": 22.173662845114972, "learning_rate": 9.325071137947703e-06, "loss": 2.1051, "step": 19230 }, { "epoch": 2.01508, "grad_norm": 21.161390967091272, "learning_rate": 9.32425839270187e-06, "loss": 2.0707, "step": 19240 }, { "epoch": 2.01518, "grad_norm": 20.607965803653777, "learning_rate": 9.3234451938614e-06, "loss": 2.0428, "step": 19250 }, { "epoch": 2.01528, "grad_norm": 20.28728744218442, "learning_rate": 9.322631541511595e-06, "loss": 2.11, "step": 19260 }, { "epoch": 2.01538, "grad_norm": 21.179144490255545, "learning_rate": 9.321817435737805e-06, "loss": 2.1174, "step": 19270 }, { "epoch": 2.01548, "grad_norm": 23.581829572557734, "learning_rate": 9.321002876625422e-06, "loss": 2.1024, "step": 19280 }, { "epoch": 2.01558, "grad_norm": 20.326164176919665, "learning_rate": 9.320187864259894e-06, "loss": 2.0363, "step": 19290 }, { "epoch": 2.01568, "grad_norm": 20.857585639374037, "learning_rate": 9.319372398726707e-06, "loss": 2.0974, "step": 19300 }, { "epoch": 2.01578, "grad_norm": 19.338678595776315, "learning_rate": 9.318556480111402e-06, "loss": 2.0626, "step": 19310 }, { "epoch": 2.01588, "grad_norm": 21.22160703672688, "learning_rate": 9.317740108499567e-06, "loss": 2.0989, "step": 19320 }, { "epoch": 2.01598, "grad_norm": 21.047642984455802, "learning_rate": 9.316923283976834e-06, "loss": 2.0436, "step": 19330 }, { "epoch": 2.01608, "grad_norm": 19.434338579132202, "learning_rate": 9.316106006628882e-06, "loss": 2.0223, "step": 19340 }, { "epoch": 2.01618, "grad_norm": 21.442840509662076, "learning_rate": 9.315288276541444e-06, "loss": 2.0147, "step": 19350 }, { "epoch": 2.01628, "grad_norm": 23.47208757768788, "learning_rate": 9.314470093800291e-06, "loss": 2.0883, "step": 19360 }, { "epoch": 2.01638, "grad_norm": 19.130908333739097, "learning_rate": 9.313651458491251e-06, "loss": 2.0576, "step": 19370 }, { "epoch": 2.01648, "grad_norm": 22.57288878936349, "learning_rate": 9.312832370700192e-06, "loss": 2.0482, "step": 19380 }, { "epoch": 2.01658, "grad_norm": 21.63500490256675, "learning_rate": 9.312012830513036e-06, "loss": 2.1252, "step": 19390 }, { "epoch": 2.01668, "grad_norm": 20.300839626257837, "learning_rate": 9.311192838015746e-06, "loss": 2.0617, "step": 19400 }, { "epoch": 2.01678, "grad_norm": 23.797426875009972, "learning_rate": 9.310372393294336e-06, "loss": 2.1011, "step": 19410 }, { "epoch": 2.01688, "grad_norm": 20.572453444575416, "learning_rate": 9.309551496434866e-06, "loss": 2.0795, "step": 19420 }, { "epoch": 2.01698, "grad_norm": 20.89259048891207, "learning_rate": 9.308730147523446e-06, "loss": 2.0442, "step": 19430 }, { "epoch": 2.01708, "grad_norm": 19.88998958099447, "learning_rate": 9.307908346646231e-06, "loss": 2.097, "step": 19440 }, { "epoch": 2.01718, "grad_norm": 21.078443079906755, "learning_rate": 9.307086093889424e-06, "loss": 2.0487, "step": 19450 }, { "epoch": 2.01728, "grad_norm": 21.544982808222446, "learning_rate": 9.306263389339277e-06, "loss": 2.0251, "step": 19460 }, { "epoch": 2.01738, "grad_norm": 19.732246918103215, "learning_rate": 9.305440233082083e-06, "loss": 2.0919, "step": 19470 }, { "epoch": 2.01748, "grad_norm": 18.587442324463783, "learning_rate": 9.304616625204193e-06, "loss": 1.9877, "step": 19480 }, { "epoch": 2.01758, "grad_norm": 23.790818770030203, "learning_rate": 9.303792565791998e-06, "loss": 1.9866, "step": 19490 }, { "epoch": 2.01768, "grad_norm": 21.09141180964432, "learning_rate": 9.302968054931936e-06, "loss": 2.0524, "step": 19500 }, { "epoch": 2.01778, "grad_norm": 19.580683188034875, "learning_rate": 9.302143092710497e-06, "loss": 2.0013, "step": 19510 }, { "epoch": 2.01788, "grad_norm": 27.72084304408466, "learning_rate": 9.301317679214213e-06, "loss": 2.0747, "step": 19520 }, { "epoch": 2.01798, "grad_norm": 19.56020558108665, "learning_rate": 9.300491814529668e-06, "loss": 2.0568, "step": 19530 }, { "epoch": 2.01808, "grad_norm": 20.376276653919444, "learning_rate": 9.299665498743492e-06, "loss": 2.0602, "step": 19540 }, { "epoch": 2.01818, "grad_norm": 20.247063305532915, "learning_rate": 9.29883873194236e-06, "loss": 2.0923, "step": 19550 }, { "epoch": 2.01828, "grad_norm": 20.242520221626474, "learning_rate": 9.298011514212994e-06, "loss": 2.1055, "step": 19560 }, { "epoch": 2.01838, "grad_norm": 21.06386337019974, "learning_rate": 9.29718384564217e-06, "loss": 2.1315, "step": 19570 }, { "epoch": 2.01848, "grad_norm": 19.87633198163929, "learning_rate": 9.296355726316705e-06, "loss": 2.0626, "step": 19580 }, { "epoch": 2.01858, "grad_norm": 19.802021674909916, "learning_rate": 9.295527156323462e-06, "loss": 2.077, "step": 19590 }, { "epoch": 2.01868, "grad_norm": 20.58061778133727, "learning_rate": 9.29469813574936e-06, "loss": 2.0477, "step": 19600 }, { "epoch": 2.01878, "grad_norm": 18.28972958265427, "learning_rate": 9.293868664681353e-06, "loss": 1.9891, "step": 19610 }, { "epoch": 2.01888, "grad_norm": 18.488715223562775, "learning_rate": 9.293038743206454e-06, "loss": 2.1019, "step": 19620 }, { "epoch": 2.01898, "grad_norm": 19.122697802687178, "learning_rate": 9.292208371411714e-06, "loss": 2.0565, "step": 19630 }, { "epoch": 2.01908, "grad_norm": 19.144062218727807, "learning_rate": 9.291377549384236e-06, "loss": 2.0229, "step": 19640 }, { "epoch": 2.01918, "grad_norm": 19.375388683821782, "learning_rate": 9.29054627721117e-06, "loss": 2.0592, "step": 19650 }, { "epoch": 2.01928, "grad_norm": 20.77114282158543, "learning_rate": 9.289714554979714e-06, "loss": 2.076, "step": 19660 }, { "epoch": 2.01938, "grad_norm": 19.91710575988977, "learning_rate": 9.288882382777109e-06, "loss": 2.0715, "step": 19670 }, { "epoch": 2.01948, "grad_norm": 22.0027587202486, "learning_rate": 9.288049760690648e-06, "loss": 2.0298, "step": 19680 }, { "epoch": 2.01958, "grad_norm": 20.052807405762717, "learning_rate": 9.287216688807667e-06, "loss": 2.1062, "step": 19690 }, { "epoch": 2.01968, "grad_norm": 19.76990280747783, "learning_rate": 9.286383167215556e-06, "loss": 2.0205, "step": 19700 }, { "epoch": 2.01978, "grad_norm": 18.987724393593055, "learning_rate": 9.285549196001742e-06, "loss": 2.0172, "step": 19710 }, { "epoch": 2.01988, "grad_norm": 22.06353041110087, "learning_rate": 9.284714775253708e-06, "loss": 2.0471, "step": 19720 }, { "epoch": 2.01998, "grad_norm": 20.171124413673198, "learning_rate": 9.28387990505898e-06, "loss": 1.9949, "step": 19730 }, { "epoch": 2.02008, "grad_norm": 22.106374960956224, "learning_rate": 9.283044585505132e-06, "loss": 2.0037, "step": 19740 }, { "epoch": 2.02018, "grad_norm": 23.12448979545297, "learning_rate": 9.282208816679785e-06, "loss": 2.0123, "step": 19750 }, { "epoch": 2.02028, "grad_norm": 23.451943715763985, "learning_rate": 9.281372598670608e-06, "loss": 2.0262, "step": 19760 }, { "epoch": 2.02038, "grad_norm": 19.56411341192551, "learning_rate": 9.280535931565315e-06, "loss": 2.0885, "step": 19770 }, { "epoch": 2.02048, "grad_norm": 22.220223428504223, "learning_rate": 9.279698815451671e-06, "loss": 2.0504, "step": 19780 }, { "epoch": 2.02058, "grad_norm": 19.97438627932375, "learning_rate": 9.278861250417483e-06, "loss": 2.0521, "step": 19790 }, { "epoch": 2.02068, "grad_norm": 19.701915060067773, "learning_rate": 9.278023236550608e-06, "loss": 2.0649, "step": 19800 }, { "epoch": 2.02078, "grad_norm": 20.605828599318077, "learning_rate": 9.277184773938952e-06, "loss": 2.0757, "step": 19810 }, { "epoch": 2.02088, "grad_norm": 19.620654078663357, "learning_rate": 9.276345862670462e-06, "loss": 2.0109, "step": 19820 }, { "epoch": 2.0209799999999998, "grad_norm": 21.155582879520736, "learning_rate": 9.27550650283314e-06, "loss": 2.0516, "step": 19830 }, { "epoch": 2.02108, "grad_norm": 19.624304086316144, "learning_rate": 9.274666694515029e-06, "loss": 1.9555, "step": 19840 }, { "epoch": 2.02118, "grad_norm": 21.15854000985519, "learning_rate": 9.273826437804219e-06, "loss": 2.054, "step": 19850 }, { "epoch": 2.02128, "grad_norm": 21.155767231450454, "learning_rate": 9.272985732788853e-06, "loss": 1.9953, "step": 19860 }, { "epoch": 2.02138, "grad_norm": 22.1610771906197, "learning_rate": 9.272144579557114e-06, "loss": 2.1551, "step": 19870 }, { "epoch": 2.02148, "grad_norm": 20.699538005396892, "learning_rate": 9.271302978197238e-06, "loss": 2.0977, "step": 19880 }, { "epoch": 2.02158, "grad_norm": 23.358000654174315, "learning_rate": 9.270460928797503e-06, "loss": 1.9832, "step": 19890 }, { "epoch": 2.02168, "grad_norm": 19.684840076236597, "learning_rate": 9.269618431446236e-06, "loss": 2.1252, "step": 19900 }, { "epoch": 2.02178, "grad_norm": 20.671364239479484, "learning_rate": 9.268775486231811e-06, "loss": 2.086, "step": 19910 }, { "epoch": 2.02188, "grad_norm": 19.938755500486703, "learning_rate": 9.267932093242648e-06, "loss": 2.0411, "step": 19920 }, { "epoch": 2.02198, "grad_norm": 21.432435792032088, "learning_rate": 9.26708825256722e-06, "loss": 2.0385, "step": 19930 }, { "epoch": 2.02208, "grad_norm": 21.10652728254049, "learning_rate": 9.266243964294037e-06, "loss": 2.0618, "step": 19940 }, { "epoch": 2.02218, "grad_norm": 20.967015986037907, "learning_rate": 9.26539922851166e-06, "loss": 2.0108, "step": 19950 }, { "epoch": 2.02228, "grad_norm": 20.81836090106823, "learning_rate": 9.264554045308704e-06, "loss": 2.0242, "step": 19960 }, { "epoch": 2.02238, "grad_norm": 22.064353159953217, "learning_rate": 9.26370841477382e-06, "loss": 2.0688, "step": 19970 }, { "epoch": 2.02248, "grad_norm": 22.0576330186266, "learning_rate": 9.26286233699571e-06, "loss": 2.1248, "step": 19980 }, { "epoch": 2.02258, "grad_norm": 21.002873817644883, "learning_rate": 9.262015812063127e-06, "loss": 2.1182, "step": 19990 }, { "epoch": 2.02268, "grad_norm": 20.77636004102831, "learning_rate": 9.261168840064864e-06, "loss": 2.0893, "step": 20000 }, { "epoch": 2.02278, "grad_norm": 21.926425053762234, "learning_rate": 9.26032142108977e-06, "loss": 2.0801, "step": 20010 }, { "epoch": 2.02288, "grad_norm": 21.803390675962614, "learning_rate": 9.259473555226729e-06, "loss": 2.0853, "step": 20020 }, { "epoch": 2.02298, "grad_norm": 20.623366262465357, "learning_rate": 9.258625242564681e-06, "loss": 2.0834, "step": 20030 }, { "epoch": 2.02308, "grad_norm": 21.812859957898404, "learning_rate": 9.25777648319261e-06, "loss": 2.0773, "step": 20040 }, { "epoch": 2.02318, "grad_norm": 21.36480698119299, "learning_rate": 9.256927277199548e-06, "loss": 2.1326, "step": 20050 }, { "epoch": 2.02328, "grad_norm": 17.85610133739599, "learning_rate": 9.25607762467457e-06, "loss": 2.0974, "step": 20060 }, { "epoch": 2.02338, "grad_norm": 20.848153401303293, "learning_rate": 9.255227525706806e-06, "loss": 2.0813, "step": 20070 }, { "epoch": 2.02348, "grad_norm": 21.91382829883105, "learning_rate": 9.25437698038542e-06, "loss": 2.1364, "step": 20080 }, { "epoch": 2.02358, "grad_norm": 21.399240778052313, "learning_rate": 9.253525988799634e-06, "loss": 2.1213, "step": 20090 }, { "epoch": 2.02368, "grad_norm": 21.259319469987577, "learning_rate": 9.252674551038715e-06, "loss": 2.0885, "step": 20100 }, { "epoch": 2.02378, "grad_norm": 19.422880678880738, "learning_rate": 9.25182266719197e-06, "loss": 2.102, "step": 20110 }, { "epoch": 2.02388, "grad_norm": 20.47718661717349, "learning_rate": 9.250970337348762e-06, "loss": 2.0312, "step": 20120 }, { "epoch": 2.02398, "grad_norm": 20.674404953055493, "learning_rate": 9.250117561598495e-06, "loss": 2.0578, "step": 20130 }, { "epoch": 2.02408, "grad_norm": 19.667694844373333, "learning_rate": 9.249264340030622e-06, "loss": 2.025, "step": 20140 }, { "epoch": 2.02418, "grad_norm": 21.893061753776557, "learning_rate": 9.24841067273464e-06, "loss": 1.9571, "step": 20150 }, { "epoch": 2.02428, "grad_norm": 23.477590241387627, "learning_rate": 9.247556559800098e-06, "loss": 2.0231, "step": 20160 }, { "epoch": 2.02438, "grad_norm": 20.4888568515424, "learning_rate": 9.246702001316584e-06, "loss": 1.9806, "step": 20170 }, { "epoch": 2.02448, "grad_norm": 22.177195600440516, "learning_rate": 9.24584699737374e-06, "loss": 2.063, "step": 20180 }, { "epoch": 2.02458, "grad_norm": 19.640657550906973, "learning_rate": 9.244991548061251e-06, "loss": 2.0855, "step": 20190 }, { "epoch": 2.02468, "grad_norm": 20.203946701261927, "learning_rate": 9.244135653468852e-06, "loss": 2.094, "step": 20200 }, { "epoch": 2.02478, "grad_norm": 19.76538604462449, "learning_rate": 9.24327931368632e-06, "loss": 2.0534, "step": 20210 }, { "epoch": 2.02488, "grad_norm": 20.64879224114895, "learning_rate": 9.242422528803482e-06, "loss": 2.0935, "step": 20220 }, { "epoch": 2.02498, "grad_norm": 19.13628896969196, "learning_rate": 9.24156529891021e-06, "loss": 2.0449, "step": 20230 }, { "epoch": 2.02508, "grad_norm": 20.931916130872274, "learning_rate": 9.240707624096424e-06, "loss": 2.0457, "step": 20240 }, { "epoch": 2.02518, "grad_norm": 20.50006626111718, "learning_rate": 9.239849504452088e-06, "loss": 2.0104, "step": 20250 }, { "epoch": 2.02528, "grad_norm": 19.69247719543875, "learning_rate": 9.23899094006722e-06, "loss": 1.976, "step": 20260 }, { "epoch": 2.02538, "grad_norm": 20.7996978881385, "learning_rate": 9.238131931031873e-06, "loss": 2.0293, "step": 20270 }, { "epoch": 2.02548, "grad_norm": 18.911571616517982, "learning_rate": 9.237272477436159e-06, "loss": 2.0322, "step": 20280 }, { "epoch": 2.02558, "grad_norm": 20.33400688093196, "learning_rate": 9.236412579370226e-06, "loss": 1.96, "step": 20290 }, { "epoch": 2.02568, "grad_norm": 21.59090796173572, "learning_rate": 9.235552236924277e-06, "loss": 1.9987, "step": 20300 }, { "epoch": 2.02578, "grad_norm": 20.998121251573263, "learning_rate": 9.234691450188554e-06, "loss": 1.9916, "step": 20310 }, { "epoch": 2.02588, "grad_norm": 20.16927375336094, "learning_rate": 9.233830219253352e-06, "loss": 2.0109, "step": 20320 }, { "epoch": 2.02598, "grad_norm": 20.709410288912384, "learning_rate": 9.232968544209011e-06, "loss": 2.0008, "step": 20330 }, { "epoch": 2.02608, "grad_norm": 20.82935957704551, "learning_rate": 9.232106425145915e-06, "loss": 1.952, "step": 20340 }, { "epoch": 2.02618, "grad_norm": 21.260171909583455, "learning_rate": 9.231243862154498e-06, "loss": 2.0129, "step": 20350 }, { "epoch": 2.02628, "grad_norm": 20.46589741871073, "learning_rate": 9.230380855325236e-06, "loss": 2.001, "step": 20360 }, { "epoch": 2.02638, "grad_norm": 20.75380364821707, "learning_rate": 9.229517404748656e-06, "loss": 1.9837, "step": 20370 }, { "epoch": 2.02648, "grad_norm": 20.844621537412152, "learning_rate": 9.228653510515333e-06, "loss": 2.0962, "step": 20380 }, { "epoch": 2.02658, "grad_norm": 20.444005014104143, "learning_rate": 9.22778917271588e-06, "loss": 2.0269, "step": 20390 }, { "epoch": 2.02668, "grad_norm": 19.05829021058578, "learning_rate": 9.226924391440963e-06, "loss": 1.9043, "step": 20400 }, { "epoch": 2.02678, "grad_norm": 20.702973337177102, "learning_rate": 9.226059166781298e-06, "loss": 1.9681, "step": 20410 }, { "epoch": 2.02688, "grad_norm": 20.353545751027227, "learning_rate": 9.225193498827639e-06, "loss": 1.9245, "step": 20420 }, { "epoch": 2.02698, "grad_norm": 17.760111220548847, "learning_rate": 9.224327387670791e-06, "loss": 2.0643, "step": 20430 }, { "epoch": 2.02708, "grad_norm": 18.830344017272445, "learning_rate": 9.223460833401607e-06, "loss": 2.0016, "step": 20440 }, { "epoch": 2.02718, "grad_norm": 22.702007884285678, "learning_rate": 9.222593836110982e-06, "loss": 1.9652, "step": 20450 }, { "epoch": 2.02728, "grad_norm": 22.206380420650568, "learning_rate": 9.221726395889863e-06, "loss": 1.942, "step": 20460 }, { "epoch": 2.02738, "grad_norm": 21.654432979121967, "learning_rate": 9.220858512829237e-06, "loss": 1.9897, "step": 20470 }, { "epoch": 2.02748, "grad_norm": 20.117829253025075, "learning_rate": 9.219990187020144e-06, "loss": 1.9927, "step": 20480 }, { "epoch": 2.02758, "grad_norm": 19.376500027349753, "learning_rate": 9.219121418553666e-06, "loss": 2.0369, "step": 20490 }, { "epoch": 2.02768, "grad_norm": 19.759458304748197, "learning_rate": 9.21825220752093e-06, "loss": 1.9696, "step": 20500 }, { "epoch": 2.02778, "grad_norm": 20.007840789924998, "learning_rate": 9.217382554013117e-06, "loss": 2.0459, "step": 20510 }, { "epoch": 2.02788, "grad_norm": 20.546892566618258, "learning_rate": 9.216512458121449e-06, "loss": 1.9029, "step": 20520 }, { "epoch": 2.02798, "grad_norm": 19.44448590512423, "learning_rate": 9.215641919937192e-06, "loss": 1.9319, "step": 20530 }, { "epoch": 2.02808, "grad_norm": 20.30964128987178, "learning_rate": 9.214770939551662e-06, "loss": 1.9698, "step": 20540 }, { "epoch": 2.02818, "grad_norm": 20.08682467974906, "learning_rate": 9.213899517056223e-06, "loss": 2.021, "step": 20550 }, { "epoch": 2.02828, "grad_norm": 21.117879778326014, "learning_rate": 9.213027652542284e-06, "loss": 1.9744, "step": 20560 }, { "epoch": 2.02838, "grad_norm": 20.409758508346638, "learning_rate": 9.212155346101294e-06, "loss": 2.0064, "step": 20570 }, { "epoch": 2.02848, "grad_norm": 21.44499094087472, "learning_rate": 9.21128259782476e-06, "loss": 1.9833, "step": 20580 }, { "epoch": 2.02858, "grad_norm": 21.329746267225183, "learning_rate": 9.210409407804225e-06, "loss": 2.0148, "step": 20590 }, { "epoch": 2.02868, "grad_norm": 21.642834950134485, "learning_rate": 9.209535776131285e-06, "loss": 2.0079, "step": 20600 }, { "epoch": 2.02878, "grad_norm": 19.43147606842169, "learning_rate": 9.208661702897578e-06, "loss": 1.965, "step": 20610 }, { "epoch": 2.02888, "grad_norm": 23.735713216607934, "learning_rate": 9.207787188194793e-06, "loss": 1.9572, "step": 20620 }, { "epoch": 2.02898, "grad_norm": 22.98270866196439, "learning_rate": 9.20691223211466e-06, "loss": 2.0245, "step": 20630 }, { "epoch": 2.02908, "grad_norm": 22.05426290336925, "learning_rate": 9.206036834748958e-06, "loss": 1.9915, "step": 20640 }, { "epoch": 2.02918, "grad_norm": 21.746161664899585, "learning_rate": 9.205160996189513e-06, "loss": 2.0807, "step": 20650 }, { "epoch": 2.02928, "grad_norm": 21.2299502687362, "learning_rate": 9.204284716528198e-06, "loss": 2.0542, "step": 20660 }, { "epoch": 2.02938, "grad_norm": 19.04644278014629, "learning_rate": 9.203407995856927e-06, "loss": 2.0028, "step": 20670 }, { "epoch": 2.02948, "grad_norm": 18.7187249573083, "learning_rate": 9.202530834267669e-06, "loss": 1.9353, "step": 20680 }, { "epoch": 2.02958, "grad_norm": 19.16514693457818, "learning_rate": 9.201653231852428e-06, "loss": 2.0269, "step": 20690 }, { "epoch": 2.02968, "grad_norm": 18.777399400100403, "learning_rate": 9.200775188703264e-06, "loss": 1.9451, "step": 20700 }, { "epoch": 2.02978, "grad_norm": 22.592901016630584, "learning_rate": 9.19989670491228e-06, "loss": 2.0217, "step": 20710 }, { "epoch": 2.02988, "grad_norm": 21.39637524363026, "learning_rate": 9.199017780571621e-06, "loss": 1.9893, "step": 20720 }, { "epoch": 2.02998, "grad_norm": 20.67997319052783, "learning_rate": 9.19813841577349e-06, "loss": 1.8893, "step": 20730 }, { "epoch": 2.03008, "grad_norm": 21.416652880028703, "learning_rate": 9.19725861061012e-06, "loss": 1.8816, "step": 20740 }, { "epoch": 2.03018, "grad_norm": 21.19028656268079, "learning_rate": 9.196378365173802e-06, "loss": 1.9446, "step": 20750 }, { "epoch": 2.03028, "grad_norm": 22.05583406019122, "learning_rate": 9.19549767955687e-06, "loss": 1.9561, "step": 20760 }, { "epoch": 2.03038, "grad_norm": 21.920121779232847, "learning_rate": 9.194616553851704e-06, "loss": 1.9399, "step": 20770 }, { "epoch": 2.03048, "grad_norm": 21.57966071692057, "learning_rate": 9.19373498815073e-06, "loss": 1.9121, "step": 20780 }, { "epoch": 2.03058, "grad_norm": 19.685128649774114, "learning_rate": 9.192852982546416e-06, "loss": 1.9365, "step": 20790 }, { "epoch": 2.03068, "grad_norm": 20.713631048587988, "learning_rate": 9.191970537131285e-06, "loss": 1.9096, "step": 20800 }, { "epoch": 2.03078, "grad_norm": 22.153997920925825, "learning_rate": 9.191087651997902e-06, "loss": 1.8629, "step": 20810 }, { "epoch": 2.03088, "grad_norm": 21.153047318050547, "learning_rate": 9.190204327238875e-06, "loss": 1.9533, "step": 20820 }, { "epoch": 2.03098, "grad_norm": 20.45184981135462, "learning_rate": 9.189320562946862e-06, "loss": 1.9292, "step": 20830 }, { "epoch": 2.03108, "grad_norm": 19.726533455124894, "learning_rate": 9.188436359214564e-06, "loss": 1.8854, "step": 20840 }, { "epoch": 2.03118, "grad_norm": 19.460021286700126, "learning_rate": 9.187551716134731e-06, "loss": 1.9086, "step": 20850 }, { "epoch": 2.03128, "grad_norm": 20.857349562453255, "learning_rate": 9.186666633800158e-06, "loss": 1.9204, "step": 20860 }, { "epoch": 2.03138, "grad_norm": 20.041661006903087, "learning_rate": 9.185781112303686e-06, "loss": 1.9462, "step": 20870 }, { "epoch": 2.03148, "grad_norm": 19.57370319075407, "learning_rate": 9.184895151738204e-06, "loss": 1.9353, "step": 20880 }, { "epoch": 2.03158, "grad_norm": 20.375121036091425, "learning_rate": 9.184008752196642e-06, "loss": 1.8355, "step": 20890 }, { "epoch": 2.03168, "grad_norm": 23.744332061018607, "learning_rate": 9.18312191377198e-06, "loss": 1.8808, "step": 20900 }, { "epoch": 2.03178, "grad_norm": 20.190006859580407, "learning_rate": 9.182234636557245e-06, "loss": 1.8747, "step": 20910 }, { "epoch": 2.03188, "grad_norm": 20.510516482792113, "learning_rate": 9.181346920645506e-06, "loss": 1.9321, "step": 20920 }, { "epoch": 2.03198, "grad_norm": 19.312627839941772, "learning_rate": 9.18045876612988e-06, "loss": 1.8912, "step": 20930 }, { "epoch": 2.03208, "grad_norm": 22.904907039798392, "learning_rate": 9.179570173103533e-06, "loss": 1.9264, "step": 20940 }, { "epoch": 2.03218, "grad_norm": 21.033005367935296, "learning_rate": 9.178681141659673e-06, "loss": 1.9093, "step": 20950 }, { "epoch": 2.03228, "grad_norm": 18.32668456113916, "learning_rate": 9.177791671891556e-06, "loss": 1.8654, "step": 20960 }, { "epoch": 2.03238, "grad_norm": 20.264613735222035, "learning_rate": 9.17690176389248e-06, "loss": 1.895, "step": 20970 }, { "epoch": 2.03248, "grad_norm": 21.936482146531656, "learning_rate": 9.176011417755797e-06, "loss": 1.9339, "step": 20980 }, { "epoch": 2.03258, "grad_norm": 21.67022554335784, "learning_rate": 9.175120633574896e-06, "loss": 1.9588, "step": 20990 }, { "epoch": 2.03268, "grad_norm": 19.801444844396922, "learning_rate": 9.174229411443219e-06, "loss": 1.884, "step": 21000 }, { "epoch": 2.03278, "grad_norm": 20.665172588763, "learning_rate": 9.17333775145425e-06, "loss": 1.8761, "step": 21010 }, { "epoch": 2.03288, "grad_norm": 22.586674658397605, "learning_rate": 9.172445653701518e-06, "loss": 1.9423, "step": 21020 }, { "epoch": 2.03298, "grad_norm": 23.925702660699194, "learning_rate": 9.171553118278602e-06, "loss": 1.9637, "step": 21030 }, { "epoch": 2.03308, "grad_norm": 19.058767164613485, "learning_rate": 9.170660145279126e-06, "loss": 1.8904, "step": 21040 }, { "epoch": 2.03318, "grad_norm": 23.697785193008365, "learning_rate": 9.169766734796759e-06, "loss": 1.9957, "step": 21050 }, { "epoch": 2.03328, "grad_norm": 19.534783999784658, "learning_rate": 9.16887288692521e-06, "loss": 1.96, "step": 21060 }, { "epoch": 2.03338, "grad_norm": 20.49029068790327, "learning_rate": 9.167978601758248e-06, "loss": 1.9454, "step": 21070 }, { "epoch": 2.03348, "grad_norm": 20.614867141016738, "learning_rate": 9.167083879389671e-06, "loss": 1.993, "step": 21080 }, { "epoch": 2.03358, "grad_norm": 20.74252126961586, "learning_rate": 9.166188719913337e-06, "loss": 1.9403, "step": 21090 }, { "epoch": 2.03368, "grad_norm": 22.277736594618755, "learning_rate": 9.165293123423142e-06, "loss": 1.8641, "step": 21100 }, { "epoch": 2.03378, "grad_norm": 19.86531863886274, "learning_rate": 9.16439709001303e-06, "loss": 1.9561, "step": 21110 }, { "epoch": 2.03388, "grad_norm": 19.290370035780544, "learning_rate": 9.16350061977699e-06, "loss": 1.9116, "step": 21120 }, { "epoch": 2.03398, "grad_norm": 19.428566783821175, "learning_rate": 9.16260371280906e-06, "loss": 1.8608, "step": 21130 }, { "epoch": 2.03408, "grad_norm": 21.0942116517393, "learning_rate": 9.161706369203319e-06, "loss": 1.8547, "step": 21140 }, { "epoch": 2.03418, "grad_norm": 20.946345195281676, "learning_rate": 9.160808589053893e-06, "loss": 1.941, "step": 21150 }, { "epoch": 2.03428, "grad_norm": 21.980553228336507, "learning_rate": 9.159910372454958e-06, "loss": 1.9638, "step": 21160 }, { "epoch": 2.03438, "grad_norm": 19.60592024044914, "learning_rate": 9.159011719500733e-06, "loss": 1.8949, "step": 21170 }, { "epoch": 2.03448, "grad_norm": 20.536500543520628, "learning_rate": 9.15811263028548e-06, "loss": 1.8925, "step": 21180 }, { "epoch": 2.03458, "grad_norm": 21.017916680118862, "learning_rate": 9.157213104903513e-06, "loss": 1.8908, "step": 21190 }, { "epoch": 2.03468, "grad_norm": 19.56685030398099, "learning_rate": 9.156313143449183e-06, "loss": 1.8734, "step": 21200 }, { "epoch": 2.03478, "grad_norm": 18.539965629170606, "learning_rate": 9.155412746016894e-06, "loss": 1.8523, "step": 21210 }, { "epoch": 2.03488, "grad_norm": 19.900050201495166, "learning_rate": 9.154511912701094e-06, "loss": 1.8792, "step": 21220 }, { "epoch": 2.03498, "grad_norm": 21.696987770853635, "learning_rate": 9.153610643596278e-06, "loss": 1.8249, "step": 21230 }, { "epoch": 2.03508, "grad_norm": 24.092638850180336, "learning_rate": 9.152708938796981e-06, "loss": 1.8744, "step": 21240 }, { "epoch": 2.03518, "grad_norm": 22.451509580462638, "learning_rate": 9.151806798397792e-06, "loss": 1.9022, "step": 21250 }, { "epoch": 2.03528, "grad_norm": 22.272123376443606, "learning_rate": 9.150904222493336e-06, "loss": 1.9163, "step": 21260 }, { "epoch": 2.03538, "grad_norm": 23.258510274029355, "learning_rate": 9.150001211178294e-06, "loss": 1.9093, "step": 21270 }, { "epoch": 2.03548, "grad_norm": 22.09145596329092, "learning_rate": 9.149097764547385e-06, "loss": 1.9325, "step": 21280 }, { "epoch": 2.03558, "grad_norm": 21.282100688055337, "learning_rate": 9.148193882695376e-06, "loss": 1.843, "step": 21290 }, { "epoch": 2.03568, "grad_norm": 18.304035849542885, "learning_rate": 9.147289565717083e-06, "loss": 1.8714, "step": 21300 }, { "epoch": 2.03578, "grad_norm": 19.730418755482603, "learning_rate": 9.146384813707362e-06, "loss": 1.9236, "step": 21310 }, { "epoch": 2.03588, "grad_norm": 19.732563526619597, "learning_rate": 9.145479626761118e-06, "loss": 1.9334, "step": 21320 }, { "epoch": 2.03598, "grad_norm": 21.374380995103493, "learning_rate": 9.1445740049733e-06, "loss": 1.8049, "step": 21330 }, { "epoch": 2.03608, "grad_norm": 20.589561520409514, "learning_rate": 9.143667948438905e-06, "loss": 1.8965, "step": 21340 }, { "epoch": 2.03618, "grad_norm": 20.2863846409379, "learning_rate": 9.142761457252974e-06, "loss": 1.905, "step": 21350 }, { "epoch": 2.03628, "grad_norm": 20.761108001450626, "learning_rate": 9.141854531510593e-06, "loss": 1.9478, "step": 21360 }, { "epoch": 2.03638, "grad_norm": 18.77546590852578, "learning_rate": 9.140947171306895e-06, "loss": 1.8828, "step": 21370 }, { "epoch": 2.03648, "grad_norm": 20.83606846334676, "learning_rate": 9.140039376737057e-06, "loss": 1.8399, "step": 21380 }, { "epoch": 2.03658, "grad_norm": 21.85327912165669, "learning_rate": 9.139131147896303e-06, "loss": 1.9471, "step": 21390 }, { "epoch": 2.03668, "grad_norm": 22.808234171162194, "learning_rate": 9.138222484879901e-06, "loss": 1.9223, "step": 21400 }, { "epoch": 2.03678, "grad_norm": 21.676057364676918, "learning_rate": 9.13731338778317e-06, "loss": 1.9553, "step": 21410 }, { "epoch": 2.03688, "grad_norm": 21.985281864495388, "learning_rate": 9.136403856701462e-06, "loss": 1.8779, "step": 21420 }, { "epoch": 2.03698, "grad_norm": 20.314052283951067, "learning_rate": 9.13549389173019e-06, "loss": 1.8125, "step": 21430 }, { "epoch": 2.03708, "grad_norm": 22.88794338782358, "learning_rate": 9.1345834929648e-06, "loss": 1.9428, "step": 21440 }, { "epoch": 2.03718, "grad_norm": 20.113604454422095, "learning_rate": 9.133672660500792e-06, "loss": 2.0048, "step": 21450 }, { "epoch": 2.03728, "grad_norm": 21.660283555984176, "learning_rate": 9.132761394433708e-06, "loss": 1.9333, "step": 21460 }, { "epoch": 2.03738, "grad_norm": 20.501843163447006, "learning_rate": 9.131849694859132e-06, "loss": 1.9977, "step": 21470 }, { "epoch": 2.03748, "grad_norm": 18.677122577011584, "learning_rate": 9.130937561872701e-06, "loss": 1.974, "step": 21480 }, { "epoch": 2.03758, "grad_norm": 19.781460931899712, "learning_rate": 9.130024995570091e-06, "loss": 1.8921, "step": 21490 }, { "epoch": 2.03768, "grad_norm": 21.66574822327452, "learning_rate": 9.129111996047028e-06, "loss": 1.984, "step": 21500 }, { "epoch": 2.03778, "grad_norm": 21.067806181032402, "learning_rate": 9.12819856339928e-06, "loss": 1.9111, "step": 21510 }, { "epoch": 2.03788, "grad_norm": 19.633379037798022, "learning_rate": 9.127284697722663e-06, "loss": 1.8727, "step": 21520 }, { "epoch": 2.03798, "grad_norm": 22.908265711063137, "learning_rate": 9.126370399113035e-06, "loss": 1.8692, "step": 21530 }, { "epoch": 2.03808, "grad_norm": 21.085798987294602, "learning_rate": 9.125455667666304e-06, "loss": 1.8743, "step": 21540 }, { "epoch": 2.03818, "grad_norm": 21.94022076513874, "learning_rate": 9.12454050347842e-06, "loss": 1.8823, "step": 21550 }, { "epoch": 2.03828, "grad_norm": 19.53293530676506, "learning_rate": 9.12362490664538e-06, "loss": 1.9576, "step": 21560 }, { "epoch": 2.03838, "grad_norm": 21.0356019250674, "learning_rate": 9.122708877263226e-06, "loss": 1.9732, "step": 21570 }, { "epoch": 2.03848, "grad_norm": 21.508811801398814, "learning_rate": 9.121792415428045e-06, "loss": 1.9427, "step": 21580 }, { "epoch": 2.03858, "grad_norm": 18.220681782680245, "learning_rate": 9.12087552123597e-06, "loss": 1.8888, "step": 21590 }, { "epoch": 2.03868, "grad_norm": 20.381007863913254, "learning_rate": 9.119958194783177e-06, "loss": 1.8899, "step": 21600 }, { "epoch": 2.03878, "grad_norm": 21.107332270194274, "learning_rate": 9.11904043616589e-06, "loss": 1.971, "step": 21610 }, { "epoch": 2.03888, "grad_norm": 20.82467584042472, "learning_rate": 9.11812224548038e-06, "loss": 1.943, "step": 21620 }, { "epoch": 2.03898, "grad_norm": 19.90830339757223, "learning_rate": 9.11720362282296e-06, "loss": 1.9623, "step": 21630 }, { "epoch": 2.03908, "grad_norm": 21.225913039689022, "learning_rate": 9.116284568289987e-06, "loss": 1.9903, "step": 21640 }, { "epoch": 2.03918, "grad_norm": 19.671357484019197, "learning_rate": 9.115365081977866e-06, "loss": 1.8731, "step": 21650 }, { "epoch": 2.03928, "grad_norm": 20.04531228450161, "learning_rate": 9.11444516398305e-06, "loss": 1.8672, "step": 21660 }, { "epoch": 2.03938, "grad_norm": 20.821041613579737, "learning_rate": 9.113524814402032e-06, "loss": 1.841, "step": 21670 }, { "epoch": 2.03948, "grad_norm": 19.898024405206453, "learning_rate": 9.112604033331352e-06, "loss": 1.9182, "step": 21680 }, { "epoch": 2.03958, "grad_norm": 22.34069452530542, "learning_rate": 9.111682820867595e-06, "loss": 1.8626, "step": 21690 }, { "epoch": 2.03968, "grad_norm": 20.63213537215449, "learning_rate": 9.110761177107394e-06, "loss": 1.9642, "step": 21700 }, { "epoch": 2.03978, "grad_norm": 22.52626199939015, "learning_rate": 9.109839102147424e-06, "loss": 1.9294, "step": 21710 }, { "epoch": 2.03988, "grad_norm": 20.34320016253148, "learning_rate": 9.108916596084407e-06, "loss": 1.9243, "step": 21720 }, { "epoch": 2.03998, "grad_norm": 18.758613910323582, "learning_rate": 9.107993659015108e-06, "loss": 1.9227, "step": 21730 }, { "epoch": 2.04008, "grad_norm": 20.1330671444648, "learning_rate": 9.10707029103634e-06, "loss": 1.9338, "step": 21740 }, { "epoch": 2.04018, "grad_norm": 19.655099380292906, "learning_rate": 9.10614649224496e-06, "loss": 1.9425, "step": 21750 }, { "epoch": 2.04028, "grad_norm": 21.9346353274346, "learning_rate": 9.105222262737871e-06, "loss": 1.8821, "step": 21760 }, { "epoch": 2.04038, "grad_norm": 22.230769300472, "learning_rate": 9.10429760261202e-06, "loss": 1.9377, "step": 21770 }, { "epoch": 2.04048, "grad_norm": 20.990040328385405, "learning_rate": 9.1033725119644e-06, "loss": 1.9026, "step": 21780 }, { "epoch": 2.04058, "grad_norm": 21.754984822225914, "learning_rate": 9.102446990892044e-06, "loss": 1.8875, "step": 21790 }, { "epoch": 2.04068, "grad_norm": 20.25896649518612, "learning_rate": 9.101521039492042e-06, "loss": 1.862, "step": 21800 }, { "epoch": 2.04078, "grad_norm": 22.193714486949904, "learning_rate": 9.100594657861517e-06, "loss": 1.9021, "step": 21810 }, { "epoch": 2.04088, "grad_norm": 21.27902952747921, "learning_rate": 9.099667846097645e-06, "loss": 1.8783, "step": 21820 }, { "epoch": 2.04098, "grad_norm": 21.760633085001395, "learning_rate": 9.098740604297642e-06, "loss": 1.9797, "step": 21830 }, { "epoch": 2.04108, "grad_norm": 22.179884450489084, "learning_rate": 9.097812932558774e-06, "loss": 1.9494, "step": 21840 }, { "epoch": 2.04118, "grad_norm": 19.163300620134724, "learning_rate": 9.096884830978346e-06, "loss": 1.9149, "step": 21850 }, { "epoch": 2.04128, "grad_norm": 20.657246525820245, "learning_rate": 9.095956299653714e-06, "loss": 1.9506, "step": 21860 }, { "epoch": 2.04138, "grad_norm": 19.7347765150047, "learning_rate": 9.095027338682276e-06, "loss": 1.8572, "step": 21870 }, { "epoch": 2.04148, "grad_norm": 20.23535849545266, "learning_rate": 9.094097948161478e-06, "loss": 1.7822, "step": 21880 }, { "epoch": 2.04158, "grad_norm": 21.568058142839863, "learning_rate": 9.093168128188805e-06, "loss": 1.8748, "step": 21890 }, { "epoch": 2.04168, "grad_norm": 22.071090141700676, "learning_rate": 9.092237878861791e-06, "loss": 1.8824, "step": 21900 }, { "epoch": 2.04178, "grad_norm": 21.65747750513995, "learning_rate": 9.091307200278018e-06, "loss": 1.8356, "step": 21910 }, { "epoch": 2.04188, "grad_norm": 21.636304113202456, "learning_rate": 9.090376092535107e-06, "loss": 1.9667, "step": 21920 }, { "epoch": 2.04198, "grad_norm": 20.819234170130986, "learning_rate": 9.089444555730729e-06, "loss": 1.9138, "step": 21930 }, { "epoch": 2.04208, "grad_norm": 20.11585802982426, "learning_rate": 9.088512589962595e-06, "loss": 1.8071, "step": 21940 }, { "epoch": 2.04218, "grad_norm": 21.03465034708013, "learning_rate": 9.087580195328467e-06, "loss": 1.9626, "step": 21950 }, { "epoch": 2.04228, "grad_norm": 22.22866055106643, "learning_rate": 9.086647371926147e-06, "loss": 1.8992, "step": 21960 }, { "epoch": 2.04238, "grad_norm": 20.716638012631847, "learning_rate": 9.085714119853485e-06, "loss": 1.8825, "step": 21970 }, { "epoch": 2.04248, "grad_norm": 21.74061195465281, "learning_rate": 9.084780439208374e-06, "loss": 1.9845, "step": 21980 }, { "epoch": 2.04258, "grad_norm": 22.60503966748552, "learning_rate": 9.083846330088753e-06, "loss": 2.0055, "step": 21990 }, { "epoch": 2.04268, "grad_norm": 21.398205452430325, "learning_rate": 9.082911792592606e-06, "loss": 1.9203, "step": 22000 }, { "epoch": 2.04278, "grad_norm": 21.450596978557485, "learning_rate": 9.08197682681796e-06, "loss": 1.9723, "step": 22010 }, { "epoch": 2.04288, "grad_norm": 20.325156879934916, "learning_rate": 9.081041432862891e-06, "loss": 1.9119, "step": 22020 }, { "epoch": 2.04298, "grad_norm": 21.878900382245124, "learning_rate": 9.080105610825516e-06, "loss": 1.9168, "step": 22030 }, { "epoch": 2.04308, "grad_norm": 23.621788039137424, "learning_rate": 9.079169360803999e-06, "loss": 1.9239, "step": 22040 }, { "epoch": 2.04318, "grad_norm": 20.859908893806555, "learning_rate": 9.07823268289655e-06, "loss": 1.8876, "step": 22050 }, { "epoch": 2.04328, "grad_norm": 21.030841161238797, "learning_rate": 9.077295577201417e-06, "loss": 1.9049, "step": 22060 }, { "epoch": 2.04338, "grad_norm": 19.562289506588208, "learning_rate": 9.076358043816903e-06, "loss": 1.8908, "step": 22070 }, { "epoch": 2.04348, "grad_norm": 20.411170521397448, "learning_rate": 9.075420082841348e-06, "loss": 1.9896, "step": 22080 }, { "epoch": 2.04358, "grad_norm": 22.416930386694222, "learning_rate": 9.07448169437314e-06, "loss": 1.9128, "step": 22090 }, { "epoch": 2.04368, "grad_norm": 21.537958477233772, "learning_rate": 9.073542878510715e-06, "loss": 1.9301, "step": 22100 }, { "epoch": 2.04378, "grad_norm": 23.036210670220058, "learning_rate": 9.072603635352548e-06, "loss": 1.8955, "step": 22110 }, { "epoch": 2.04388, "grad_norm": 21.32239891071459, "learning_rate": 9.07166396499716e-06, "loss": 1.9112, "step": 22120 }, { "epoch": 2.04398, "grad_norm": 21.52216602444645, "learning_rate": 9.070723867543119e-06, "loss": 1.8757, "step": 22130 }, { "epoch": 2.04408, "grad_norm": 21.217810325204905, "learning_rate": 9.069783343089038e-06, "loss": 1.884, "step": 22140 }, { "epoch": 2.04418, "grad_norm": 20.291809081971852, "learning_rate": 9.068842391733572e-06, "loss": 1.941, "step": 22150 }, { "epoch": 2.04428, "grad_norm": 22.24437545236351, "learning_rate": 9.067901013575424e-06, "loss": 1.9515, "step": 22160 }, { "epoch": 2.04438, "grad_norm": 21.594615336237197, "learning_rate": 9.066959208713339e-06, "loss": 1.8828, "step": 22170 }, { "epoch": 2.04448, "grad_norm": 20.033958387922823, "learning_rate": 9.066016977246109e-06, "loss": 1.8628, "step": 22180 }, { "epoch": 2.04458, "grad_norm": 18.547830357198187, "learning_rate": 9.065074319272567e-06, "loss": 1.8992, "step": 22190 }, { "epoch": 2.04468, "grad_norm": 20.63819801984843, "learning_rate": 9.064131234891598e-06, "loss": 1.8705, "step": 22200 }, { "epoch": 2.04478, "grad_norm": 19.92764080163551, "learning_rate": 9.063187724202122e-06, "loss": 1.9158, "step": 22210 }, { "epoch": 2.04488, "grad_norm": 20.729423302829513, "learning_rate": 9.062243787303113e-06, "loss": 1.9135, "step": 22220 }, { "epoch": 2.04498, "grad_norm": 22.68050942825329, "learning_rate": 9.061299424293585e-06, "loss": 1.8903, "step": 22230 }, { "epoch": 2.04508, "grad_norm": 20.730377740097985, "learning_rate": 9.060354635272598e-06, "loss": 1.8716, "step": 22240 }, { "epoch": 2.04518, "grad_norm": 20.79814400393474, "learning_rate": 9.059409420339251e-06, "loss": 1.896, "step": 22250 }, { "epoch": 2.04528, "grad_norm": 20.178307407418785, "learning_rate": 9.058463779592696e-06, "loss": 1.9112, "step": 22260 }, { "epoch": 2.04538, "grad_norm": 19.537207659947608, "learning_rate": 9.05751771313213e-06, "loss": 1.899, "step": 22270 }, { "epoch": 2.04548, "grad_norm": 22.116242733109885, "learning_rate": 9.056571221056783e-06, "loss": 1.9196, "step": 22280 }, { "epoch": 2.04558, "grad_norm": 21.942430010640322, "learning_rate": 9.055624303465944e-06, "loss": 1.9511, "step": 22290 }, { "epoch": 2.04568, "grad_norm": 21.171102733314388, "learning_rate": 9.054676960458939e-06, "loss": 1.9511, "step": 22300 }, { "epoch": 2.04578, "grad_norm": 20.30052225323585, "learning_rate": 9.053729192135138e-06, "loss": 1.9218, "step": 22310 }, { "epoch": 2.04588, "grad_norm": 20.36752060432653, "learning_rate": 9.05278099859396e-06, "loss": 1.9221, "step": 22320 }, { "epoch": 2.04598, "grad_norm": 21.240440146332237, "learning_rate": 9.051832379934864e-06, "loss": 1.8759, "step": 22330 }, { "epoch": 2.04608, "grad_norm": 22.292312361337142, "learning_rate": 9.050883336257357e-06, "loss": 1.8349, "step": 22340 }, { "epoch": 2.04618, "grad_norm": 22.015155684897515, "learning_rate": 9.049933867660988e-06, "loss": 1.8826, "step": 22350 }, { "epoch": 2.04628, "grad_norm": 21.03381010739222, "learning_rate": 9.048983974245354e-06, "loss": 1.9026, "step": 22360 }, { "epoch": 2.04638, "grad_norm": 23.362565604000018, "learning_rate": 9.048033656110092e-06, "loss": 1.8887, "step": 22370 }, { "epoch": 2.04648, "grad_norm": 19.3304260167231, "learning_rate": 9.047082913354886e-06, "loss": 1.865, "step": 22380 }, { "epoch": 2.04658, "grad_norm": 22.718719294002817, "learning_rate": 9.046131746079467e-06, "loss": 1.8616, "step": 22390 }, { "epoch": 2.04668, "grad_norm": 21.438125536203362, "learning_rate": 9.045180154383606e-06, "loss": 1.8411, "step": 22400 }, { "epoch": 2.04678, "grad_norm": 22.685391981957444, "learning_rate": 9.044228138367122e-06, "loss": 1.8543, "step": 22410 }, { "epoch": 2.04688, "grad_norm": 22.050980585797113, "learning_rate": 9.043275698129875e-06, "loss": 1.8401, "step": 22420 }, { "epoch": 2.04698, "grad_norm": 23.299380077401864, "learning_rate": 9.042322833771772e-06, "loss": 1.9198, "step": 22430 }, { "epoch": 2.04708, "grad_norm": 23.088846787922947, "learning_rate": 9.041369545392764e-06, "loss": 1.8615, "step": 22440 }, { "epoch": 2.04718, "grad_norm": 21.832760453542136, "learning_rate": 9.04041583309285e-06, "loss": 1.9876, "step": 22450 }, { "epoch": 2.04728, "grad_norm": 20.271962146534513, "learning_rate": 9.039461696972066e-06, "loss": 1.8998, "step": 22460 }, { "epoch": 2.04738, "grad_norm": 19.731529533661973, "learning_rate": 9.0385071371305e-06, "loss": 1.8272, "step": 22470 }, { "epoch": 2.04748, "grad_norm": 20.78963601205681, "learning_rate": 9.037552153668273e-06, "loss": 1.905, "step": 22480 }, { "epoch": 2.04758, "grad_norm": 19.97212512679095, "learning_rate": 9.036596746685569e-06, "loss": 1.9359, "step": 22490 }, { "epoch": 2.04768, "grad_norm": 26.812813096766646, "learning_rate": 9.0356409162826e-06, "loss": 1.9668, "step": 22500 }, { "epoch": 2.04778, "grad_norm": 18.932431188679782, "learning_rate": 9.034684662559627e-06, "loss": 1.8401, "step": 22510 }, { "epoch": 2.04788, "grad_norm": 20.73530837918311, "learning_rate": 9.03372798561696e-06, "loss": 1.8551, "step": 22520 }, { "epoch": 2.04798, "grad_norm": 22.70669692562851, "learning_rate": 9.032770885554947e-06, "loss": 1.8831, "step": 22530 }, { "epoch": 2.04808, "grad_norm": 21.708788904482617, "learning_rate": 9.031813362473985e-06, "loss": 1.9121, "step": 22540 }, { "epoch": 2.04818, "grad_norm": 21.336649161424678, "learning_rate": 9.030855416474515e-06, "loss": 1.8733, "step": 22550 }, { "epoch": 2.04828, "grad_norm": 17.972536091393632, "learning_rate": 9.02989704765702e-06, "loss": 1.9367, "step": 22560 }, { "epoch": 2.04838, "grad_norm": 22.086803466152126, "learning_rate": 9.028938256122024e-06, "loss": 1.8789, "step": 22570 }, { "epoch": 2.04848, "grad_norm": 20.77585740034731, "learning_rate": 9.027979041970108e-06, "loss": 1.8775, "step": 22580 }, { "epoch": 2.04858, "grad_norm": 22.220040508833655, "learning_rate": 9.027019405301883e-06, "loss": 1.9243, "step": 22590 }, { "epoch": 2.04868, "grad_norm": 20.048934833146646, "learning_rate": 9.026059346218013e-06, "loss": 1.8842, "step": 22600 }, { "epoch": 2.04878, "grad_norm": 20.377986536517458, "learning_rate": 9.025098864819203e-06, "loss": 1.8845, "step": 22610 }, { "epoch": 2.04888, "grad_norm": 22.981116241245555, "learning_rate": 9.024137961206202e-06, "loss": 1.9166, "step": 22620 }, { "epoch": 2.04898, "grad_norm": 21.00142069736928, "learning_rate": 9.023176635479807e-06, "loss": 1.9079, "step": 22630 }, { "epoch": 2.04908, "grad_norm": 21.281143206996493, "learning_rate": 9.022214887740854e-06, "loss": 1.8568, "step": 22640 }, { "epoch": 2.04918, "grad_norm": 21.680602559565152, "learning_rate": 9.02125271809023e-06, "loss": 1.9304, "step": 22650 }, { "epoch": 2.04928, "grad_norm": 22.789157607636984, "learning_rate": 9.020290126628854e-06, "loss": 1.9464, "step": 22660 }, { "epoch": 2.04938, "grad_norm": 20.36833482563771, "learning_rate": 9.019327113457706e-06, "loss": 1.876, "step": 22670 }, { "epoch": 2.04948, "grad_norm": 19.997148099722416, "learning_rate": 9.018363678677797e-06, "loss": 1.8048, "step": 22680 }, { "epoch": 2.04958, "grad_norm": 21.284172500742628, "learning_rate": 9.017399822390189e-06, "loss": 1.7951, "step": 22690 }, { "epoch": 2.04968, "grad_norm": 22.22292893681337, "learning_rate": 9.016435544695983e-06, "loss": 1.9234, "step": 22700 }, { "epoch": 2.04978, "grad_norm": 21.38558893633165, "learning_rate": 9.015470845696331e-06, "loss": 1.8369, "step": 22710 }, { "epoch": 2.04988, "grad_norm": 22.813351099547074, "learning_rate": 9.014505725492422e-06, "loss": 1.8219, "step": 22720 }, { "epoch": 2.04998, "grad_norm": 21.019851431713892, "learning_rate": 9.013540184185497e-06, "loss": 1.8998, "step": 22730 }, { "epoch": 2.05008, "grad_norm": 22.278626337632456, "learning_rate": 9.012574221876831e-06, "loss": 1.8698, "step": 22740 }, { "epoch": 2.05018, "grad_norm": 20.350925579335165, "learning_rate": 9.011607838667753e-06, "loss": 1.8768, "step": 22750 }, { "epoch": 2.05028, "grad_norm": 20.3820207628597, "learning_rate": 9.010641034659631e-06, "loss": 1.886, "step": 22760 }, { "epoch": 2.05038, "grad_norm": 21.73070122212037, "learning_rate": 9.009673809953879e-06, "loss": 1.9535, "step": 22770 }, { "epoch": 2.05048, "grad_norm": 20.504173052502495, "learning_rate": 9.008706164651955e-06, "loss": 1.8444, "step": 22780 }, { "epoch": 2.05058, "grad_norm": 23.42467100443914, "learning_rate": 9.007738098855358e-06, "loss": 1.8885, "step": 22790 }, { "epoch": 2.05068, "grad_norm": 21.43811215294309, "learning_rate": 9.006769612665634e-06, "loss": 1.9928, "step": 22800 }, { "epoch": 2.05078, "grad_norm": 18.83701690518759, "learning_rate": 9.005800706184376e-06, "loss": 1.8943, "step": 22810 }, { "epoch": 2.05088, "grad_norm": 21.719462795691147, "learning_rate": 9.004831379513212e-06, "loss": 1.8668, "step": 22820 }, { "epoch": 2.05098, "grad_norm": 21.070058648345018, "learning_rate": 9.003861632753827e-06, "loss": 1.8488, "step": 22830 }, { "epoch": 2.05108, "grad_norm": 21.485498681451826, "learning_rate": 9.002891466007938e-06, "loss": 1.9595, "step": 22840 }, { "epoch": 2.05118, "grad_norm": 20.715685004321113, "learning_rate": 9.001920879377311e-06, "loss": 1.9046, "step": 22850 }, { "epoch": 2.05128, "grad_norm": 21.823676423682638, "learning_rate": 9.000949872963758e-06, "loss": 1.9691, "step": 22860 }, { "epoch": 2.05138, "grad_norm": 20.253850815617078, "learning_rate": 8.999978446869131e-06, "loss": 1.8006, "step": 22870 }, { "epoch": 2.05148, "grad_norm": 19.05204638475701, "learning_rate": 8.999006601195332e-06, "loss": 1.9283, "step": 22880 }, { "epoch": 2.05158, "grad_norm": 22.380314829824602, "learning_rate": 8.998034336044298e-06, "loss": 1.9108, "step": 22890 }, { "epoch": 2.05168, "grad_norm": 19.259408445065052, "learning_rate": 8.99706165151802e-06, "loss": 1.8624, "step": 22900 }, { "epoch": 2.05178, "grad_norm": 21.034477449630025, "learning_rate": 8.996088547718525e-06, "loss": 1.8942, "step": 22910 }, { "epoch": 2.05188, "grad_norm": 20.042441177149296, "learning_rate": 8.995115024747888e-06, "loss": 1.8353, "step": 22920 }, { "epoch": 2.05198, "grad_norm": 21.46582795398857, "learning_rate": 8.994141082708223e-06, "loss": 1.8507, "step": 22930 }, { "epoch": 2.05208, "grad_norm": 22.206482604686553, "learning_rate": 8.993166721701701e-06, "loss": 1.8861, "step": 22940 }, { "epoch": 2.05218, "grad_norm": 19.93644774937078, "learning_rate": 8.992191941830519e-06, "loss": 1.8483, "step": 22950 }, { "epoch": 2.05228, "grad_norm": 22.05952342879921, "learning_rate": 8.991216743196933e-06, "loss": 1.9547, "step": 22960 }, { "epoch": 2.05238, "grad_norm": 21.34322373591706, "learning_rate": 8.990241125903234e-06, "loss": 1.8954, "step": 22970 }, { "epoch": 2.05248, "grad_norm": 21.124340566045845, "learning_rate": 8.98926509005176e-06, "loss": 1.9026, "step": 22980 }, { "epoch": 2.05258, "grad_norm": 21.123050060501843, "learning_rate": 8.988288635744892e-06, "loss": 1.9066, "step": 22990 }, { "epoch": 2.05268, "grad_norm": 21.160685492925715, "learning_rate": 8.987311763085056e-06, "loss": 1.887, "step": 23000 }, { "epoch": 2.05278, "grad_norm": 20.037345150435186, "learning_rate": 8.986334472174724e-06, "loss": 1.8924, "step": 23010 }, { "epoch": 2.05288, "grad_norm": 20.87221841874739, "learning_rate": 8.985356763116406e-06, "loss": 1.8172, "step": 23020 }, { "epoch": 2.05298, "grad_norm": 22.440041785460796, "learning_rate": 8.984378636012659e-06, "loss": 1.9328, "step": 23030 }, { "epoch": 2.05308, "grad_norm": 23.113689536413705, "learning_rate": 8.983400090966086e-06, "loss": 1.8758, "step": 23040 }, { "epoch": 2.05318, "grad_norm": 20.5849448813577, "learning_rate": 8.98242112807933e-06, "loss": 1.8922, "step": 23050 }, { "epoch": 2.05328, "grad_norm": 20.264446454955547, "learning_rate": 8.981441747455084e-06, "loss": 1.9218, "step": 23060 }, { "epoch": 2.05338, "grad_norm": 20.306111709491923, "learning_rate": 8.980461949196073e-06, "loss": 1.8993, "step": 23070 }, { "epoch": 2.05348, "grad_norm": 18.96343748145045, "learning_rate": 8.97948173340508e-06, "loss": 1.8811, "step": 23080 }, { "epoch": 2.05358, "grad_norm": 21.112139825306773, "learning_rate": 8.978501100184922e-06, "loss": 1.9537, "step": 23090 }, { "epoch": 2.05368, "grad_norm": 19.87279133106826, "learning_rate": 8.977520049638463e-06, "loss": 1.9468, "step": 23100 }, { "epoch": 2.05378, "grad_norm": 20.936282825448263, "learning_rate": 8.97653858186861e-06, "loss": 1.9076, "step": 23110 }, { "epoch": 2.05388, "grad_norm": 19.639155072951016, "learning_rate": 8.975556696978319e-06, "loss": 1.9683, "step": 23120 }, { "epoch": 2.05398, "grad_norm": 21.115648050777537, "learning_rate": 8.974574395070579e-06, "loss": 1.938, "step": 23130 }, { "epoch": 2.05408, "grad_norm": 21.089937697336488, "learning_rate": 8.97359167624843e-06, "loss": 1.9507, "step": 23140 }, { "epoch": 2.05418, "grad_norm": 20.81597165675022, "learning_rate": 8.972608540614958e-06, "loss": 1.9278, "step": 23150 }, { "epoch": 2.05428, "grad_norm": 21.26067244670418, "learning_rate": 8.971624988273289e-06, "loss": 1.937, "step": 23160 }, { "epoch": 2.05438, "grad_norm": 21.49592752538835, "learning_rate": 8.970641019326589e-06, "loss": 1.9113, "step": 23170 }, { "epoch": 2.05448, "grad_norm": 21.174110344544644, "learning_rate": 8.969656633878076e-06, "loss": 1.8404, "step": 23180 }, { "epoch": 2.05458, "grad_norm": 20.840356636296423, "learning_rate": 8.968671832031004e-06, "loss": 1.8486, "step": 23190 }, { "epoch": 2.05468, "grad_norm": 22.870925535408354, "learning_rate": 8.967686613888677e-06, "loss": 1.8696, "step": 23200 }, { "epoch": 2.05478, "grad_norm": 19.95995247612801, "learning_rate": 8.966700979554442e-06, "loss": 1.9909, "step": 23210 }, { "epoch": 2.05488, "grad_norm": 20.728793254266872, "learning_rate": 8.96571492913168e-06, "loss": 1.8983, "step": 23220 }, { "epoch": 2.05498, "grad_norm": 23.28518422732344, "learning_rate": 8.964728462723829e-06, "loss": 1.8663, "step": 23230 }, { "epoch": 2.05508, "grad_norm": 21.680112729384863, "learning_rate": 8.963741580434363e-06, "loss": 1.8194, "step": 23240 }, { "epoch": 2.05518, "grad_norm": 20.91261489173409, "learning_rate": 8.9627542823668e-06, "loss": 1.8406, "step": 23250 }, { "epoch": 2.05528, "grad_norm": 22.077358767454257, "learning_rate": 8.961766568624708e-06, "loss": 1.9492, "step": 23260 }, { "epoch": 2.05538, "grad_norm": 20.91508941253775, "learning_rate": 8.960778439311689e-06, "loss": 1.8849, "step": 23270 }, { "epoch": 2.05548, "grad_norm": 22.063005986058727, "learning_rate": 8.959789894531392e-06, "loss": 1.9628, "step": 23280 }, { "epoch": 2.05558, "grad_norm": 21.375910639348255, "learning_rate": 8.958800934387512e-06, "loss": 1.9282, "step": 23290 }, { "epoch": 2.05568, "grad_norm": 20.92589357519, "learning_rate": 8.95781155898379e-06, "loss": 1.9537, "step": 23300 }, { "epoch": 2.05578, "grad_norm": 21.48743663759777, "learning_rate": 8.956821768424005e-06, "loss": 1.8941, "step": 23310 }, { "epoch": 2.05588, "grad_norm": 21.145526162846494, "learning_rate": 8.95583156281198e-06, "loss": 1.857, "step": 23320 }, { "epoch": 2.05598, "grad_norm": 21.015752247447953, "learning_rate": 8.954840942251582e-06, "loss": 1.9409, "step": 23330 }, { "epoch": 2.05608, "grad_norm": 22.512520695441967, "learning_rate": 8.953849906846726e-06, "loss": 1.891, "step": 23340 }, { "epoch": 2.05618, "grad_norm": 20.334357158195424, "learning_rate": 8.952858456701365e-06, "loss": 1.9178, "step": 23350 }, { "epoch": 2.05628, "grad_norm": 23.73614413542444, "learning_rate": 8.951866591919497e-06, "loss": 1.9036, "step": 23360 }, { "epoch": 2.05638, "grad_norm": 19.37484932119762, "learning_rate": 8.950874312605165e-06, "loss": 1.9066, "step": 23370 }, { "epoch": 2.05648, "grad_norm": 21.772382759053027, "learning_rate": 8.949881618862453e-06, "loss": 1.9421, "step": 23380 }, { "epoch": 2.05658, "grad_norm": 21.48843005847169, "learning_rate": 8.948888510795493e-06, "loss": 1.9468, "step": 23390 }, { "epoch": 2.05668, "grad_norm": 20.943449198144833, "learning_rate": 8.947894988508454e-06, "loss": 1.9308, "step": 23400 }, { "epoch": 2.05678, "grad_norm": 19.760373619410053, "learning_rate": 8.946901052105554e-06, "loss": 1.9102, "step": 23410 }, { "epoch": 2.05688, "grad_norm": 23.018433251111105, "learning_rate": 8.945906701691052e-06, "loss": 1.9622, "step": 23420 }, { "epoch": 2.05698, "grad_norm": 20.581240501947306, "learning_rate": 8.944911937369251e-06, "loss": 1.8956, "step": 23430 }, { "epoch": 2.05708, "grad_norm": 22.954605661424655, "learning_rate": 8.943916759244496e-06, "loss": 1.9051, "step": 23440 }, { "epoch": 2.05718, "grad_norm": 19.86793221030643, "learning_rate": 8.942921167421178e-06, "loss": 1.8889, "step": 23450 }, { "epoch": 2.05728, "grad_norm": 19.646079000587125, "learning_rate": 8.941925162003728e-06, "loss": 1.8932, "step": 23460 }, { "epoch": 2.05738, "grad_norm": 21.529135558399645, "learning_rate": 8.940928743096624e-06, "loss": 1.8658, "step": 23470 }, { "epoch": 2.05748, "grad_norm": 19.590359580700074, "learning_rate": 8.939931910804384e-06, "loss": 1.8737, "step": 23480 }, { "epoch": 2.05758, "grad_norm": 23.934987228973498, "learning_rate": 8.938934665231575e-06, "loss": 1.9063, "step": 23490 }, { "epoch": 2.05768, "grad_norm": 20.888720830819675, "learning_rate": 8.9379370064828e-06, "loss": 1.9499, "step": 23500 }, { "epoch": 2.05778, "grad_norm": 21.5779493862118, "learning_rate": 8.936938934662708e-06, "loss": 1.9697, "step": 23510 }, { "epoch": 2.05788, "grad_norm": 20.578366664621225, "learning_rate": 8.935940449875995e-06, "loss": 1.9712, "step": 23520 }, { "epoch": 2.05798, "grad_norm": 20.701768862731345, "learning_rate": 8.934941552227396e-06, "loss": 2.0402, "step": 23530 }, { "epoch": 2.05808, "grad_norm": 20.805160260598015, "learning_rate": 8.933942241821692e-06, "loss": 1.9609, "step": 23540 }, { "epoch": 2.05818, "grad_norm": 22.485077918637735, "learning_rate": 8.932942518763703e-06, "loss": 1.9609, "step": 23550 }, { "epoch": 2.05828, "grad_norm": 19.781071560043205, "learning_rate": 8.931942383158296e-06, "loss": 1.9313, "step": 23560 }, { "epoch": 2.05838, "grad_norm": 20.197083545690923, "learning_rate": 8.930941835110384e-06, "loss": 1.9312, "step": 23570 }, { "epoch": 2.05848, "grad_norm": 22.58875239696601, "learning_rate": 8.929940874724918e-06, "loss": 1.8807, "step": 23580 }, { "epoch": 2.05858, "grad_norm": 22.160374293420016, "learning_rate": 8.928939502106891e-06, "loss": 1.8605, "step": 23590 }, { "epoch": 2.05868, "grad_norm": 21.431605996448198, "learning_rate": 8.927937717361347e-06, "loss": 1.9101, "step": 23600 }, { "epoch": 2.05878, "grad_norm": 21.80376381135763, "learning_rate": 8.926935520593364e-06, "loss": 1.9296, "step": 23610 }, { "epoch": 2.05888, "grad_norm": 22.355897347449318, "learning_rate": 8.925932911908074e-06, "loss": 1.9304, "step": 23620 }, { "epoch": 2.05898, "grad_norm": 19.717474659804623, "learning_rate": 8.92492989141064e-06, "loss": 1.8358, "step": 23630 }, { "epoch": 2.05908, "grad_norm": 21.986274870921285, "learning_rate": 8.923926459206279e-06, "loss": 1.8531, "step": 23640 }, { "epoch": 2.05918, "grad_norm": 20.158795411228972, "learning_rate": 8.922922615400242e-06, "loss": 1.9742, "step": 23650 }, { "epoch": 2.05928, "grad_norm": 21.800170568842443, "learning_rate": 8.92191836009783e-06, "loss": 1.9293, "step": 23660 }, { "epoch": 2.05938, "grad_norm": 21.602505996019907, "learning_rate": 8.920913693404383e-06, "loss": 1.8927, "step": 23670 }, { "epoch": 2.05948, "grad_norm": 20.71329483884549, "learning_rate": 8.919908615425287e-06, "loss": 1.9489, "step": 23680 }, { "epoch": 2.05958, "grad_norm": 22.288229158143245, "learning_rate": 8.918903126265972e-06, "loss": 1.9246, "step": 23690 }, { "epoch": 2.05968, "grad_norm": 23.17414346338346, "learning_rate": 8.917897226031906e-06, "loss": 1.9146, "step": 23700 }, { "epoch": 2.05978, "grad_norm": 20.656453664326673, "learning_rate": 8.916890914828606e-06, "loss": 1.8714, "step": 23710 }, { "epoch": 2.05988, "grad_norm": 22.067900703667352, "learning_rate": 8.915884192761627e-06, "loss": 1.9639, "step": 23720 }, { "epoch": 2.05998, "grad_norm": 20.669862635245345, "learning_rate": 8.91487705993657e-06, "loss": 1.925, "step": 23730 }, { "epoch": 2.06008, "grad_norm": 19.945895401378294, "learning_rate": 8.91386951645908e-06, "loss": 1.8585, "step": 23740 }, { "epoch": 2.06018, "grad_norm": 21.450826495599664, "learning_rate": 8.91286156243484e-06, "loss": 1.9387, "step": 23750 }, { "epoch": 2.06028, "grad_norm": 21.721132370513473, "learning_rate": 8.911853197969584e-06, "loss": 1.921, "step": 23760 }, { "epoch": 2.06038, "grad_norm": 22.86726707530024, "learning_rate": 8.910844423169083e-06, "loss": 1.9442, "step": 23770 }, { "epoch": 2.06048, "grad_norm": 21.78607239938589, "learning_rate": 8.909835238139152e-06, "loss": 1.8432, "step": 23780 }, { "epoch": 2.06058, "grad_norm": 21.258860170805125, "learning_rate": 8.908825642985653e-06, "loss": 1.9324, "step": 23790 }, { "epoch": 2.06068, "grad_norm": 21.137362147093853, "learning_rate": 8.907815637814484e-06, "loss": 1.932, "step": 23800 }, { "epoch": 2.06078, "grad_norm": 18.913958106628083, "learning_rate": 8.906805222731589e-06, "loss": 1.8604, "step": 23810 }, { "epoch": 2.06088, "grad_norm": 19.32920933300572, "learning_rate": 8.905794397842962e-06, "loss": 1.972, "step": 23820 }, { "epoch": 2.06098, "grad_norm": 20.959361681964538, "learning_rate": 8.904783163254628e-06, "loss": 1.8892, "step": 23830 }, { "epoch": 2.06108, "grad_norm": 21.812217673648984, "learning_rate": 8.903771519072663e-06, "loss": 1.8309, "step": 23840 }, { "epoch": 2.06118, "grad_norm": 20.87697673587339, "learning_rate": 8.902759465403183e-06, "loss": 1.9197, "step": 23850 }, { "epoch": 2.06128, "grad_norm": 18.635044451187564, "learning_rate": 8.90174700235235e-06, "loss": 1.8647, "step": 23860 }, { "epoch": 2.06138, "grad_norm": 20.437561863423078, "learning_rate": 8.900734130026363e-06, "loss": 1.8079, "step": 23870 }, { "epoch": 2.06148, "grad_norm": 21.03388423997582, "learning_rate": 8.89972084853147e-06, "loss": 1.8653, "step": 23880 }, { "epoch": 2.06158, "grad_norm": 20.347706795612282, "learning_rate": 8.89870715797396e-06, "loss": 1.8203, "step": 23890 }, { "epoch": 2.06168, "grad_norm": 23.367002959010165, "learning_rate": 8.897693058460165e-06, "loss": 1.8523, "step": 23900 }, { "epoch": 2.06178, "grad_norm": 22.67481875917256, "learning_rate": 8.896678550096456e-06, "loss": 1.9652, "step": 23910 }, { "epoch": 2.06188, "grad_norm": 20.437753684436483, "learning_rate": 8.895663632989254e-06, "loss": 1.9103, "step": 23920 }, { "epoch": 2.06198, "grad_norm": 22.401980767196655, "learning_rate": 8.894648307245016e-06, "loss": 1.9245, "step": 23930 }, { "epoch": 2.06208, "grad_norm": 19.819823282575985, "learning_rate": 8.893632572970245e-06, "loss": 1.9027, "step": 23940 }, { "epoch": 2.06218, "grad_norm": 19.398384775790376, "learning_rate": 8.892616430271487e-06, "loss": 1.8681, "step": 23950 }, { "epoch": 2.06228, "grad_norm": 19.56829751580161, "learning_rate": 8.891599879255334e-06, "loss": 1.9184, "step": 23960 }, { "epoch": 2.06238, "grad_norm": 20.956162662092613, "learning_rate": 8.890582920028415e-06, "loss": 1.9164, "step": 23970 }, { "epoch": 2.06248, "grad_norm": 20.681094068766217, "learning_rate": 8.889565552697404e-06, "loss": 1.9512, "step": 23980 }, { "epoch": 2.06258, "grad_norm": 19.89008834524558, "learning_rate": 8.888547777369019e-06, "loss": 1.8651, "step": 23990 }, { "epoch": 2.06268, "grad_norm": 21.570317731377145, "learning_rate": 8.88752959415002e-06, "loss": 1.9352, "step": 24000 }, { "epoch": 2.06278, "grad_norm": 22.39514072748637, "learning_rate": 8.886511003147207e-06, "loss": 1.961, "step": 24010 }, { "epoch": 2.06288, "grad_norm": 21.35053644403113, "learning_rate": 8.885492004467431e-06, "loss": 1.8966, "step": 24020 }, { "epoch": 2.06298, "grad_norm": 20.607249005852807, "learning_rate": 8.884472598217574e-06, "loss": 1.8709, "step": 24030 }, { "epoch": 2.06308, "grad_norm": 22.438713550985746, "learning_rate": 8.883452784504571e-06, "loss": 1.8542, "step": 24040 }, { "epoch": 2.06318, "grad_norm": 20.986688288501153, "learning_rate": 8.882432563435394e-06, "loss": 1.8458, "step": 24050 }, { "epoch": 2.06328, "grad_norm": 23.204859985945863, "learning_rate": 8.88141193511706e-06, "loss": 1.9154, "step": 24060 }, { "epoch": 2.06338, "grad_norm": 24.691535997575816, "learning_rate": 8.880390899656629e-06, "loss": 1.9488, "step": 24070 }, { "epoch": 2.06348, "grad_norm": 20.65297553689106, "learning_rate": 8.8793694571612e-06, "loss": 1.8586, "step": 24080 }, { "epoch": 2.06358, "grad_norm": 21.450769554196373, "learning_rate": 8.878347607737923e-06, "loss": 1.8146, "step": 24090 }, { "epoch": 2.06368, "grad_norm": 22.101633305227473, "learning_rate": 8.87732535149398e-06, "loss": 1.9626, "step": 24100 }, { "epoch": 2.06378, "grad_norm": 20.421391192985805, "learning_rate": 8.876302688536602e-06, "loss": 1.9035, "step": 24110 }, { "epoch": 2.06388, "grad_norm": 19.59982116921165, "learning_rate": 8.875279618973064e-06, "loss": 1.8977, "step": 24120 }, { "epoch": 2.06398, "grad_norm": 17.992165104105794, "learning_rate": 8.87425614291068e-06, "loss": 1.8571, "step": 24130 }, { "epoch": 2.06408, "grad_norm": 20.998872662276167, "learning_rate": 8.873232260456806e-06, "loss": 1.849, "step": 24140 }, { "epoch": 2.06418, "grad_norm": 19.36869210686927, "learning_rate": 8.872207971718845e-06, "loss": 1.8838, "step": 24150 }, { "epoch": 2.06428, "grad_norm": 20.57298435760152, "learning_rate": 8.871183276804239e-06, "loss": 1.8855, "step": 24160 }, { "epoch": 2.06438, "grad_norm": 21.57833675708586, "learning_rate": 8.870158175820474e-06, "loss": 1.8262, "step": 24170 }, { "epoch": 2.06448, "grad_norm": 22.2181669590301, "learning_rate": 8.869132668875078e-06, "loss": 1.7928, "step": 24180 }, { "epoch": 2.06458, "grad_norm": 20.98788587969428, "learning_rate": 8.868106756075622e-06, "loss": 1.89, "step": 24190 }, { "epoch": 2.06468, "grad_norm": 19.059513568682284, "learning_rate": 8.867080437529721e-06, "loss": 1.9014, "step": 24200 }, { "epoch": 2.06478, "grad_norm": 22.799032487673898, "learning_rate": 8.866053713345029e-06, "loss": 1.9038, "step": 24210 }, { "epoch": 2.06488, "grad_norm": 21.62022081515715, "learning_rate": 8.865026583629245e-06, "loss": 1.871, "step": 24220 }, { "epoch": 2.06498, "grad_norm": 21.47369190877907, "learning_rate": 8.863999048490111e-06, "loss": 1.8524, "step": 24230 }, { "epoch": 2.06508, "grad_norm": 22.085357027409266, "learning_rate": 8.86297110803541e-06, "loss": 1.9204, "step": 24240 }, { "epoch": 2.06518, "grad_norm": 20.054646285324203, "learning_rate": 8.861942762372969e-06, "loss": 1.8269, "step": 24250 }, { "epoch": 2.06528, "grad_norm": 24.507200308583496, "learning_rate": 8.860914011610655e-06, "loss": 1.9157, "step": 24260 }, { "epoch": 2.06538, "grad_norm": 20.034606627253627, "learning_rate": 8.85988485585638e-06, "loss": 1.8777, "step": 24270 }, { "epoch": 2.06548, "grad_norm": 21.110521088954446, "learning_rate": 8.8588552952181e-06, "loss": 1.9078, "step": 24280 }, { "epoch": 2.06558, "grad_norm": 21.127945284662097, "learning_rate": 8.857825329803808e-06, "loss": 1.8925, "step": 24290 }, { "epoch": 2.06568, "grad_norm": 21.76243564989476, "learning_rate": 8.856794959721543e-06, "loss": 1.9142, "step": 24300 }, { "epoch": 2.06578, "grad_norm": 24.678300045921926, "learning_rate": 8.855764185079388e-06, "loss": 1.931, "step": 24310 }, { "epoch": 2.06588, "grad_norm": 21.161890427376996, "learning_rate": 8.854733005985464e-06, "loss": 1.9176, "step": 24320 }, { "epoch": 2.06598, "grad_norm": 21.57798647053683, "learning_rate": 8.853701422547938e-06, "loss": 1.9693, "step": 24330 }, { "epoch": 2.06608, "grad_norm": 22.087975479949556, "learning_rate": 8.852669434875019e-06, "loss": 1.9433, "step": 24340 }, { "epoch": 2.06618, "grad_norm": 20.47706796444351, "learning_rate": 8.851637043074957e-06, "loss": 1.8857, "step": 24350 }, { "epoch": 2.06628, "grad_norm": 22.389438120567434, "learning_rate": 8.850604247256043e-06, "loss": 1.929, "step": 24360 }, { "epoch": 2.06638, "grad_norm": 21.035283558455447, "learning_rate": 8.849571047526617e-06, "loss": 1.9679, "step": 24370 }, { "epoch": 2.06648, "grad_norm": 19.73009112001213, "learning_rate": 8.848537443995055e-06, "loss": 1.9358, "step": 24380 }, { "epoch": 2.06658, "grad_norm": 21.768920825218743, "learning_rate": 8.847503436769776e-06, "loss": 1.9542, "step": 24390 }, { "epoch": 2.06668, "grad_norm": 22.248689824263444, "learning_rate": 8.846469025959242e-06, "loss": 1.9227, "step": 24400 }, { "epoch": 2.06678, "grad_norm": 20.43324107613862, "learning_rate": 8.845434211671961e-06, "loss": 1.8166, "step": 24410 }, { "epoch": 2.06688, "grad_norm": 21.175185283418376, "learning_rate": 8.844398994016476e-06, "loss": 1.8535, "step": 24420 }, { "epoch": 2.06698, "grad_norm": 22.73555405793379, "learning_rate": 8.84336337310138e-06, "loss": 1.8724, "step": 24430 }, { "epoch": 2.06708, "grad_norm": 20.964719207338963, "learning_rate": 8.842327349035307e-06, "loss": 1.9187, "step": 24440 }, { "epoch": 2.06718, "grad_norm": 20.575143430815682, "learning_rate": 8.841290921926923e-06, "loss": 1.8881, "step": 24450 }, { "epoch": 2.0672800000000002, "grad_norm": 20.72971164116758, "learning_rate": 8.840254091884952e-06, "loss": 1.8832, "step": 24460 }, { "epoch": 2.06738, "grad_norm": 21.45920545607329, "learning_rate": 8.83921685901815e-06, "loss": 1.8996, "step": 24470 }, { "epoch": 2.06748, "grad_norm": 21.90513616510492, "learning_rate": 8.838179223435317e-06, "loss": 1.9055, "step": 24480 }, { "epoch": 2.06758, "grad_norm": 19.943980826009614, "learning_rate": 8.837141185245298e-06, "loss": 1.8755, "step": 24490 }, { "epoch": 2.06768, "grad_norm": 21.63048772470359, "learning_rate": 8.836102744556976e-06, "loss": 1.8234, "step": 24500 }, { "epoch": 2.06778, "grad_norm": 17.87883001865297, "learning_rate": 8.835063901479281e-06, "loss": 1.8233, "step": 24510 }, { "epoch": 2.06788, "grad_norm": 19.397832292843667, "learning_rate": 8.834024656121182e-06, "loss": 1.9523, "step": 24520 }, { "epoch": 2.06798, "grad_norm": 18.608857492866463, "learning_rate": 8.83298500859169e-06, "loss": 1.8564, "step": 24530 }, { "epoch": 2.06808, "grad_norm": 21.340376904338353, "learning_rate": 8.83194495899986e-06, "loss": 1.8656, "step": 24540 }, { "epoch": 2.06818, "grad_norm": 20.892910982347505, "learning_rate": 8.830904507454789e-06, "loss": 1.7833, "step": 24550 }, { "epoch": 2.06828, "grad_norm": 20.03123446212833, "learning_rate": 8.829863654065615e-06, "loss": 1.823, "step": 24560 }, { "epoch": 2.06838, "grad_norm": 20.656160184233684, "learning_rate": 8.82882239894152e-06, "loss": 1.8984, "step": 24570 }, { "epoch": 2.06848, "grad_norm": 20.85046481081614, "learning_rate": 8.827780742191723e-06, "loss": 1.8379, "step": 24580 }, { "epoch": 2.06858, "grad_norm": 19.92969222974569, "learning_rate": 8.826738683925491e-06, "loss": 1.8567, "step": 24590 }, { "epoch": 2.06868, "grad_norm": 21.75960930565021, "learning_rate": 8.825696224252134e-06, "loss": 1.865, "step": 24600 }, { "epoch": 2.06878, "grad_norm": 20.61607690215102, "learning_rate": 8.824653363280998e-06, "loss": 1.8727, "step": 24610 }, { "epoch": 2.06888, "grad_norm": 20.23716683725187, "learning_rate": 8.823610101121474e-06, "loss": 1.8649, "step": 24620 }, { "epoch": 2.06898, "grad_norm": 21.47769573435191, "learning_rate": 8.822566437882996e-06, "loss": 1.8344, "step": 24630 }, { "epoch": 2.06908, "grad_norm": 19.897103251182315, "learning_rate": 8.82152237367504e-06, "loss": 1.8002, "step": 24640 }, { "epoch": 2.06918, "grad_norm": 20.122662417090083, "learning_rate": 8.820477908607123e-06, "loss": 1.8573, "step": 24650 }, { "epoch": 2.06928, "grad_norm": 20.986001685168674, "learning_rate": 8.819433042788805e-06, "loss": 1.8616, "step": 24660 }, { "epoch": 2.0693799999999998, "grad_norm": 22.007691012837515, "learning_rate": 8.818387776329689e-06, "loss": 1.8867, "step": 24670 }, { "epoch": 2.06948, "grad_norm": 19.119206960919026, "learning_rate": 8.817342109339415e-06, "loss": 1.7639, "step": 24680 }, { "epoch": 2.06958, "grad_norm": 21.53167633650499, "learning_rate": 8.816296041927671e-06, "loss": 1.8295, "step": 24690 }, { "epoch": 2.06968, "grad_norm": 21.089152322380283, "learning_rate": 8.815249574204186e-06, "loss": 1.7872, "step": 24700 }, { "epoch": 2.06978, "grad_norm": 20.87163693674565, "learning_rate": 8.814202706278724e-06, "loss": 1.8265, "step": 24710 }, { "epoch": 2.06988, "grad_norm": 20.991109961574924, "learning_rate": 8.813155438261106e-06, "loss": 1.8806, "step": 24720 }, { "epoch": 2.06998, "grad_norm": 20.839691743264602, "learning_rate": 8.812107770261175e-06, "loss": 1.8054, "step": 24730 }, { "epoch": 2.07008, "grad_norm": 20.695488029838895, "learning_rate": 8.811059702388833e-06, "loss": 1.8861, "step": 24740 }, { "epoch": 2.07018, "grad_norm": 20.851071811646886, "learning_rate": 8.810011234754016e-06, "loss": 1.8149, "step": 24750 }, { "epoch": 2.07028, "grad_norm": 20.984734919145907, "learning_rate": 8.808962367466702e-06, "loss": 1.845, "step": 24760 }, { "epoch": 2.07038, "grad_norm": 22.189470778181196, "learning_rate": 8.807913100636916e-06, "loss": 1.8146, "step": 24770 }, { "epoch": 2.07048, "grad_norm": 23.33140374956899, "learning_rate": 8.806863434374717e-06, "loss": 1.8072, "step": 24780 }, { "epoch": 2.07058, "grad_norm": 23.279453745227418, "learning_rate": 8.805813368790212e-06, "loss": 1.8922, "step": 24790 }, { "epoch": 2.07068, "grad_norm": 21.825995080171165, "learning_rate": 8.804762903993548e-06, "loss": 1.8394, "step": 24800 }, { "epoch": 2.07078, "grad_norm": 21.58266182493819, "learning_rate": 8.803712040094916e-06, "loss": 1.9165, "step": 24810 }, { "epoch": 2.07088, "grad_norm": 20.25349322290842, "learning_rate": 8.802660777204544e-06, "loss": 1.8848, "step": 24820 }, { "epoch": 2.07098, "grad_norm": 22.656497448362828, "learning_rate": 8.801609115432703e-06, "loss": 1.8128, "step": 24830 }, { "epoch": 2.07108, "grad_norm": 21.090850540601057, "learning_rate": 8.80055705488971e-06, "loss": 1.8383, "step": 24840 }, { "epoch": 2.07118, "grad_norm": 19.140014053940643, "learning_rate": 8.799504595685922e-06, "loss": 1.8268, "step": 24850 }, { "epoch": 2.07128, "grad_norm": 21.40060227810121, "learning_rate": 8.798451737931734e-06, "loss": 1.849, "step": 24860 }, { "epoch": 2.07138, "grad_norm": 21.083234972183874, "learning_rate": 8.797398481737591e-06, "loss": 1.8314, "step": 24870 }, { "epoch": 2.07148, "grad_norm": 20.99336435469638, "learning_rate": 8.79634482721397e-06, "loss": 1.9216, "step": 24880 }, { "epoch": 2.07158, "grad_norm": 18.474529922349358, "learning_rate": 8.795290774471395e-06, "loss": 1.8068, "step": 24890 }, { "epoch": 2.07168, "grad_norm": 21.95918326626674, "learning_rate": 8.794236323620433e-06, "loss": 1.7907, "step": 24900 }, { "epoch": 2.07178, "grad_norm": 19.653564717693772, "learning_rate": 8.79318147477169e-06, "loss": 1.8572, "step": 24910 }, { "epoch": 2.07188, "grad_norm": 21.365641662855452, "learning_rate": 8.792126228035816e-06, "loss": 1.8049, "step": 24920 }, { "epoch": 2.07198, "grad_norm": 20.903632654636894, "learning_rate": 8.7910705835235e-06, "loss": 1.8785, "step": 24930 }, { "epoch": 2.07208, "grad_norm": 20.20807814769146, "learning_rate": 8.790014541345476e-06, "loss": 1.8031, "step": 24940 }, { "epoch": 2.07218, "grad_norm": 20.506918327384447, "learning_rate": 8.788958101612514e-06, "loss": 1.7479, "step": 24950 }, { "epoch": 2.07228, "grad_norm": 21.51284231950424, "learning_rate": 8.787901264435433e-06, "loss": 1.8168, "step": 24960 }, { "epoch": 2.07238, "grad_norm": 21.02635760219773, "learning_rate": 8.78684402992509e-06, "loss": 1.7784, "step": 24970 }, { "epoch": 2.07248, "grad_norm": 20.706586746048202, "learning_rate": 8.785786398192382e-06, "loss": 1.839, "step": 24980 }, { "epoch": 2.07258, "grad_norm": 22.713681340669226, "learning_rate": 8.784728369348255e-06, "loss": 1.8193, "step": 24990 }, { "epoch": 2.07268, "grad_norm": 21.39207011096826, "learning_rate": 8.783669943503685e-06, "loss": 1.7771, "step": 25000 }, { "epoch": 2.07278, "grad_norm": 22.17242495896561, "learning_rate": 8.782611120769698e-06, "loss": 1.9104, "step": 25010 }, { "epoch": 2.07288, "grad_norm": 19.685158720114583, "learning_rate": 8.78155190125736e-06, "loss": 1.8303, "step": 25020 }, { "epoch": 2.07298, "grad_norm": 22.78578412730259, "learning_rate": 8.78049228507778e-06, "loss": 1.8517, "step": 25030 }, { "epoch": 2.07308, "grad_norm": 21.39746692234897, "learning_rate": 8.779432272342103e-06, "loss": 1.8156, "step": 25040 }, { "epoch": 2.07318, "grad_norm": 19.191093688250433, "learning_rate": 8.778371863161523e-06, "loss": 1.8502, "step": 25050 }, { "epoch": 2.07328, "grad_norm": 21.98341492488227, "learning_rate": 8.777311057647272e-06, "loss": 1.8343, "step": 25060 }, { "epoch": 2.0733800000000002, "grad_norm": 22.847003970922575, "learning_rate": 8.77624985591062e-06, "loss": 1.8215, "step": 25070 }, { "epoch": 2.07348, "grad_norm": 21.473846285227523, "learning_rate": 8.775188258062883e-06, "loss": 1.8818, "step": 25080 }, { "epoch": 2.07358, "grad_norm": 22.069721123435606, "learning_rate": 8.774126264215422e-06, "loss": 1.7855, "step": 25090 }, { "epoch": 2.07368, "grad_norm": 22.640768983673567, "learning_rate": 8.77306387447963e-06, "loss": 1.8523, "step": 25100 }, { "epoch": 2.07378, "grad_norm": 19.770244875647364, "learning_rate": 8.772001088966948e-06, "loss": 1.7716, "step": 25110 }, { "epoch": 2.07388, "grad_norm": 21.53113340822945, "learning_rate": 8.77093790778886e-06, "loss": 1.7897, "step": 25120 }, { "epoch": 2.07398, "grad_norm": 22.097386355831333, "learning_rate": 8.769874331056887e-06, "loss": 1.7813, "step": 25130 }, { "epoch": 2.07408, "grad_norm": 22.05069873332912, "learning_rate": 8.768810358882592e-06, "loss": 1.8314, "step": 25140 }, { "epoch": 2.07418, "grad_norm": 21.449756513245433, "learning_rate": 8.767745991377582e-06, "loss": 1.8288, "step": 25150 }, { "epoch": 2.07428, "grad_norm": 22.27158793641129, "learning_rate": 8.766681228653507e-06, "loss": 1.8216, "step": 25160 }, { "epoch": 2.07438, "grad_norm": 24.73218458071132, "learning_rate": 8.76561607082205e-06, "loss": 1.847, "step": 25170 }, { "epoch": 2.07448, "grad_norm": 21.70071972309899, "learning_rate": 8.764550517994946e-06, "loss": 1.6935, "step": 25180 }, { "epoch": 2.07458, "grad_norm": 22.342580569705937, "learning_rate": 8.763484570283962e-06, "loss": 1.8585, "step": 25190 }, { "epoch": 2.07468, "grad_norm": 24.06884852595456, "learning_rate": 8.762418227800914e-06, "loss": 1.8647, "step": 25200 }, { "epoch": 2.07478, "grad_norm": 21.873051203499088, "learning_rate": 8.761351490657656e-06, "loss": 1.8107, "step": 25210 }, { "epoch": 2.07488, "grad_norm": 22.539180312920635, "learning_rate": 8.760284358966084e-06, "loss": 1.7679, "step": 25220 }, { "epoch": 2.07498, "grad_norm": 21.599039159935142, "learning_rate": 8.759216832838135e-06, "loss": 1.8159, "step": 25230 }, { "epoch": 2.07508, "grad_norm": 22.983009924958324, "learning_rate": 8.758148912385786e-06, "loss": 1.8359, "step": 25240 }, { "epoch": 2.07518, "grad_norm": 22.74019762511632, "learning_rate": 8.75708059772106e-06, "loss": 1.8476, "step": 25250 }, { "epoch": 2.07528, "grad_norm": 20.00967197263023, "learning_rate": 8.756011888956017e-06, "loss": 1.7254, "step": 25260 }, { "epoch": 2.07538, "grad_norm": 22.56983384946405, "learning_rate": 8.754942786202757e-06, "loss": 1.7706, "step": 25270 }, { "epoch": 2.07548, "grad_norm": 21.099004196033782, "learning_rate": 8.753873289573426e-06, "loss": 1.7994, "step": 25280 }, { "epoch": 2.07558, "grad_norm": 21.320537516508303, "learning_rate": 8.752803399180211e-06, "loss": 1.7596, "step": 25290 }, { "epoch": 2.07568, "grad_norm": 21.605611698970595, "learning_rate": 8.751733115135335e-06, "loss": 1.7834, "step": 25300 }, { "epoch": 2.07578, "grad_norm": 21.176804994376866, "learning_rate": 8.750662437551068e-06, "loss": 1.7308, "step": 25310 }, { "epoch": 2.07588, "grad_norm": 21.768825799764393, "learning_rate": 8.749591366539719e-06, "loss": 1.7519, "step": 25320 }, { "epoch": 2.07598, "grad_norm": 21.775549106907317, "learning_rate": 8.748519902213638e-06, "loss": 1.7923, "step": 25330 }, { "epoch": 2.07608, "grad_norm": 20.060777479014067, "learning_rate": 8.747448044685216e-06, "loss": 1.7753, "step": 25340 }, { "epoch": 2.07618, "grad_norm": 20.92899193731771, "learning_rate": 8.746375794066888e-06, "loss": 1.7072, "step": 25350 }, { "epoch": 2.07628, "grad_norm": 20.485794457222518, "learning_rate": 8.745303150471125e-06, "loss": 1.8307, "step": 25360 }, { "epoch": 2.07638, "grad_norm": 22.183853720970156, "learning_rate": 8.744230114010445e-06, "loss": 1.7482, "step": 25370 }, { "epoch": 2.07648, "grad_norm": 21.60103999580697, "learning_rate": 8.743156684797404e-06, "loss": 1.7748, "step": 25380 }, { "epoch": 2.07658, "grad_norm": 21.126464471650742, "learning_rate": 8.742082862944598e-06, "loss": 1.7272, "step": 25390 }, { "epoch": 2.07668, "grad_norm": 21.71812922295181, "learning_rate": 8.741008648564668e-06, "loss": 1.7114, "step": 25400 }, { "epoch": 2.07678, "grad_norm": 21.46310002673724, "learning_rate": 8.739934041770292e-06, "loss": 1.7347, "step": 25410 }, { "epoch": 2.07688, "grad_norm": 20.199478446249444, "learning_rate": 8.738859042674196e-06, "loss": 1.7484, "step": 25420 }, { "epoch": 2.07698, "grad_norm": 19.10253161893056, "learning_rate": 8.737783651389136e-06, "loss": 1.7634, "step": 25430 }, { "epoch": 2.07708, "grad_norm": 22.13564092372215, "learning_rate": 8.736707868027919e-06, "loss": 1.7732, "step": 25440 }, { "epoch": 2.07718, "grad_norm": 22.177321065113166, "learning_rate": 8.73563169270339e-06, "loss": 1.7452, "step": 25450 }, { "epoch": 2.07728, "grad_norm": 22.863638233541472, "learning_rate": 8.734555125528432e-06, "loss": 1.714, "step": 25460 }, { "epoch": 3.00006, "grad_norm": 25.508985066915848, "learning_rate": 8.733478166615977e-06, "loss": 1.9704, "step": 25470 }, { "epoch": 3.00016, "grad_norm": 22.41787835275307, "learning_rate": 8.732400816078987e-06, "loss": 1.8479, "step": 25480 }, { "epoch": 3.00026, "grad_norm": 20.034348255415107, "learning_rate": 8.731323074030475e-06, "loss": 1.8443, "step": 25490 }, { "epoch": 3.00036, "grad_norm": 20.80010599967522, "learning_rate": 8.730244940583493e-06, "loss": 1.7851, "step": 25500 }, { "epoch": 3.00046, "grad_norm": 22.270583530475527, "learning_rate": 8.729166415851125e-06, "loss": 1.8532, "step": 25510 }, { "epoch": 3.00056, "grad_norm": 21.093022145081846, "learning_rate": 8.728087499946511e-06, "loss": 1.9052, "step": 25520 }, { "epoch": 3.00066, "grad_norm": 24.025628376079368, "learning_rate": 8.727008192982818e-06, "loss": 1.9081, "step": 25530 }, { "epoch": 3.00076, "grad_norm": 21.27690351875233, "learning_rate": 8.725928495073265e-06, "loss": 1.7254, "step": 25540 }, { "epoch": 3.00086, "grad_norm": 21.675804556434663, "learning_rate": 8.724848406331105e-06, "loss": 1.851, "step": 25550 }, { "epoch": 3.00096, "grad_norm": 20.659837141953485, "learning_rate": 8.723767926869635e-06, "loss": 1.8138, "step": 25560 }, { "epoch": 3.00106, "grad_norm": 25.027462277993315, "learning_rate": 8.722687056802192e-06, "loss": 1.8758, "step": 25570 }, { "epoch": 3.00116, "grad_norm": 21.93722416411944, "learning_rate": 8.721605796242153e-06, "loss": 1.8111, "step": 25580 }, { "epoch": 3.00126, "grad_norm": 23.16139258224899, "learning_rate": 8.72052414530294e-06, "loss": 1.7887, "step": 25590 }, { "epoch": 3.00136, "grad_norm": 22.62913234257849, "learning_rate": 8.719442104098011e-06, "loss": 1.8317, "step": 25600 }, { "epoch": 3.00146, "grad_norm": 19.322984140972665, "learning_rate": 8.718359672740867e-06, "loss": 1.6971, "step": 25610 }, { "epoch": 3.00156, "grad_norm": 23.066215083343916, "learning_rate": 8.717276851345051e-06, "loss": 1.7979, "step": 25620 }, { "epoch": 3.00166, "grad_norm": 24.329096369623677, "learning_rate": 8.716193640024147e-06, "loss": 1.7878, "step": 25630 }, { "epoch": 3.00176, "grad_norm": 23.280710872560697, "learning_rate": 8.715110038891777e-06, "loss": 1.8424, "step": 25640 }, { "epoch": 3.00186, "grad_norm": 21.43754537719351, "learning_rate": 8.714026048061606e-06, "loss": 1.9121, "step": 25650 }, { "epoch": 3.00196, "grad_norm": 22.127923381534938, "learning_rate": 8.71294166764734e-06, "loss": 1.7732, "step": 25660 }, { "epoch": 3.00206, "grad_norm": 22.26697211251845, "learning_rate": 8.711856897762725e-06, "loss": 1.8602, "step": 25670 }, { "epoch": 3.00216, "grad_norm": 21.40481233976773, "learning_rate": 8.710771738521549e-06, "loss": 1.8159, "step": 25680 }, { "epoch": 3.00226, "grad_norm": 22.536897675001182, "learning_rate": 8.709686190037641e-06, "loss": 1.8167, "step": 25690 }, { "epoch": 3.00236, "grad_norm": 22.640560470684633, "learning_rate": 8.708600252424866e-06, "loss": 1.7881, "step": 25700 }, { "epoch": 3.00246, "grad_norm": 20.62674260679536, "learning_rate": 8.70751392579714e-06, "loss": 1.9269, "step": 25710 }, { "epoch": 3.00256, "grad_norm": 21.342302404854678, "learning_rate": 8.706427210268408e-06, "loss": 1.8239, "step": 25720 }, { "epoch": 3.00266, "grad_norm": 19.449020495580097, "learning_rate": 8.705340105952665e-06, "loss": 1.8426, "step": 25730 }, { "epoch": 3.00276, "grad_norm": 21.09847583283947, "learning_rate": 8.70425261296394e-06, "loss": 1.7696, "step": 25740 }, { "epoch": 3.00286, "grad_norm": 18.897231095090238, "learning_rate": 8.703164731416309e-06, "loss": 1.8346, "step": 25750 }, { "epoch": 3.00296, "grad_norm": 24.47971815173544, "learning_rate": 8.702076461423885e-06, "loss": 1.8567, "step": 25760 }, { "epoch": 3.00306, "grad_norm": 20.02271161913679, "learning_rate": 8.700987803100822e-06, "loss": 1.8414, "step": 25770 }, { "epoch": 3.00316, "grad_norm": 23.748460578132278, "learning_rate": 8.699898756561317e-06, "loss": 1.822, "step": 25780 }, { "epoch": 3.00326, "grad_norm": 21.495542482330528, "learning_rate": 8.698809321919602e-06, "loss": 1.7794, "step": 25790 }, { "epoch": 3.00336, "grad_norm": 21.478756868344345, "learning_rate": 8.697719499289956e-06, "loss": 1.8347, "step": 25800 }, { "epoch": 3.00346, "grad_norm": 21.991368955780104, "learning_rate": 8.696629288786695e-06, "loss": 1.8209, "step": 25810 }, { "epoch": 3.00356, "grad_norm": 19.339053738528747, "learning_rate": 8.695538690524178e-06, "loss": 1.788, "step": 25820 }, { "epoch": 3.00366, "grad_norm": 21.10856662959202, "learning_rate": 8.694447704616806e-06, "loss": 1.7861, "step": 25830 }, { "epoch": 3.00376, "grad_norm": 20.99445428760124, "learning_rate": 8.693356331179017e-06, "loss": 1.7849, "step": 25840 }, { "epoch": 3.00386, "grad_norm": 21.44929684142419, "learning_rate": 8.692264570325288e-06, "loss": 1.7437, "step": 25850 }, { "epoch": 3.00396, "grad_norm": 23.40588971342544, "learning_rate": 8.691172422170143e-06, "loss": 1.7437, "step": 25860 }, { "epoch": 3.00406, "grad_norm": 22.90980158596493, "learning_rate": 8.690079886828142e-06, "loss": 1.7755, "step": 25870 }, { "epoch": 3.00416, "grad_norm": 22.413724487133116, "learning_rate": 8.688986964413886e-06, "loss": 1.8728, "step": 25880 }, { "epoch": 3.00426, "grad_norm": 21.773590054536374, "learning_rate": 8.68789365504202e-06, "loss": 1.8243, "step": 25890 }, { "epoch": 3.00436, "grad_norm": 19.288665956423667, "learning_rate": 8.686799958827226e-06, "loss": 1.7712, "step": 25900 }, { "epoch": 3.00446, "grad_norm": 22.191925371394444, "learning_rate": 8.685705875884225e-06, "loss": 1.8131, "step": 25910 }, { "epoch": 3.00456, "grad_norm": 20.68831211807078, "learning_rate": 8.684611406327785e-06, "loss": 1.7783, "step": 25920 }, { "epoch": 3.00466, "grad_norm": 20.582888861551613, "learning_rate": 8.68351655027271e-06, "loss": 1.8161, "step": 25930 }, { "epoch": 3.00476, "grad_norm": 22.808096868181725, "learning_rate": 8.682421307833844e-06, "loss": 1.8094, "step": 25940 }, { "epoch": 3.00486, "grad_norm": 21.348720945102134, "learning_rate": 8.681325679126073e-06, "loss": 1.7847, "step": 25950 }, { "epoch": 3.00496, "grad_norm": 22.294762081819822, "learning_rate": 8.680229664264325e-06, "loss": 1.8045, "step": 25960 }, { "epoch": 3.00506, "grad_norm": 21.81805086656363, "learning_rate": 8.679133263363563e-06, "loss": 1.751, "step": 25970 }, { "epoch": 3.00516, "grad_norm": 22.44095431509747, "learning_rate": 8.6780364765388e-06, "loss": 1.7575, "step": 25980 }, { "epoch": 3.00526, "grad_norm": 23.13298733093787, "learning_rate": 8.67693930390508e-06, "loss": 1.9048, "step": 25990 }, { "epoch": 3.00536, "grad_norm": 21.603166459754107, "learning_rate": 8.675841745577492e-06, "loss": 1.8107, "step": 26000 }, { "epoch": 3.00546, "grad_norm": 20.599531877840274, "learning_rate": 8.674743801671164e-06, "loss": 1.8198, "step": 26010 }, { "epoch": 3.00556, "grad_norm": 21.110267836526337, "learning_rate": 8.67364547230127e-06, "loss": 1.8153, "step": 26020 }, { "epoch": 3.00566, "grad_norm": 21.961196599512135, "learning_rate": 8.672546757583012e-06, "loss": 1.875, "step": 26030 }, { "epoch": 3.00576, "grad_norm": 19.594240524759297, "learning_rate": 8.671447657631644e-06, "loss": 1.7378, "step": 26040 }, { "epoch": 3.00586, "grad_norm": 21.89195318402983, "learning_rate": 8.670348172562457e-06, "loss": 1.7318, "step": 26050 }, { "epoch": 3.00596, "grad_norm": 21.960919683788866, "learning_rate": 8.669248302490783e-06, "loss": 1.7875, "step": 26060 }, { "epoch": 3.00606, "grad_norm": 19.966809051952044, "learning_rate": 8.66814804753199e-06, "loss": 1.6898, "step": 26070 }, { "epoch": 3.00616, "grad_norm": 22.706538412288026, "learning_rate": 8.667047407801492e-06, "loss": 1.788, "step": 26080 }, { "epoch": 3.00626, "grad_norm": 22.74551812546531, "learning_rate": 8.66594638341474e-06, "loss": 1.7899, "step": 26090 }, { "epoch": 3.00636, "grad_norm": 24.213905282319505, "learning_rate": 8.664844974487226e-06, "loss": 1.7593, "step": 26100 }, { "epoch": 3.00646, "grad_norm": 19.982449569072312, "learning_rate": 8.663743181134483e-06, "loss": 1.7336, "step": 26110 }, { "epoch": 3.00656, "grad_norm": 21.359150553600468, "learning_rate": 8.662641003472086e-06, "loss": 1.7467, "step": 26120 }, { "epoch": 3.00666, "grad_norm": 22.7403410255154, "learning_rate": 8.661538441615644e-06, "loss": 1.8196, "step": 26130 }, { "epoch": 3.00676, "grad_norm": 19.54305309393355, "learning_rate": 8.660435495680816e-06, "loss": 1.7737, "step": 26140 }, { "epoch": 3.00686, "grad_norm": 21.70948651798592, "learning_rate": 8.659332165783292e-06, "loss": 1.8023, "step": 26150 }, { "epoch": 3.00696, "grad_norm": 23.129398355701536, "learning_rate": 8.658228452038807e-06, "loss": 1.8791, "step": 26160 }, { "epoch": 3.00706, "grad_norm": 22.292596818973976, "learning_rate": 8.657124354563136e-06, "loss": 1.8104, "step": 26170 }, { "epoch": 3.00716, "grad_norm": 20.924632338100512, "learning_rate": 8.656019873472093e-06, "loss": 1.7604, "step": 26180 }, { "epoch": 3.00726, "grad_norm": 19.57146011840822, "learning_rate": 8.654915008881532e-06, "loss": 1.8633, "step": 26190 }, { "epoch": 3.00736, "grad_norm": 18.622056079473925, "learning_rate": 8.653809760907352e-06, "loss": 1.8198, "step": 26200 }, { "epoch": 3.00746, "grad_norm": 21.30760417316008, "learning_rate": 8.652704129665486e-06, "loss": 1.7909, "step": 26210 }, { "epoch": 3.00756, "grad_norm": 21.816483056150364, "learning_rate": 8.651598115271909e-06, "loss": 1.691, "step": 26220 }, { "epoch": 3.00766, "grad_norm": 20.96505418496036, "learning_rate": 8.650491717842635e-06, "loss": 1.748, "step": 26230 }, { "epoch": 3.00776, "grad_norm": 20.83783777949372, "learning_rate": 8.649384937493725e-06, "loss": 1.7042, "step": 26240 }, { "epoch": 3.00786, "grad_norm": 24.005780538668862, "learning_rate": 8.648277774341271e-06, "loss": 1.7719, "step": 26250 }, { "epoch": 3.00796, "grad_norm": 25.155738704997148, "learning_rate": 8.647170228501412e-06, "loss": 1.7714, "step": 26260 }, { "epoch": 3.00806, "grad_norm": 23.1688303095586, "learning_rate": 8.646062300090322e-06, "loss": 1.7725, "step": 26270 }, { "epoch": 3.00816, "grad_norm": 21.790153924745383, "learning_rate": 8.644953989224218e-06, "loss": 1.8084, "step": 26280 }, { "epoch": 3.00826, "grad_norm": 21.833365781789116, "learning_rate": 8.64384529601936e-06, "loss": 1.7527, "step": 26290 }, { "epoch": 3.00836, "grad_norm": 21.017131476724117, "learning_rate": 8.642736220592038e-06, "loss": 1.7402, "step": 26300 }, { "epoch": 3.00846, "grad_norm": 19.843385800053994, "learning_rate": 8.641626763058597e-06, "loss": 1.7998, "step": 26310 }, { "epoch": 3.00856, "grad_norm": 20.726469972890612, "learning_rate": 8.640516923535408e-06, "loss": 1.8329, "step": 26320 }, { "epoch": 3.00866, "grad_norm": 22.29674507755735, "learning_rate": 8.639406702138889e-06, "loss": 1.8267, "step": 26330 }, { "epoch": 3.00876, "grad_norm": 20.707931988447093, "learning_rate": 8.638296098985499e-06, "loss": 1.8309, "step": 26340 }, { "epoch": 3.00886, "grad_norm": 20.165352645614263, "learning_rate": 8.637185114191734e-06, "loss": 1.7821, "step": 26350 }, { "epoch": 3.00896, "grad_norm": 21.412226262194327, "learning_rate": 8.636073747874132e-06, "loss": 1.8081, "step": 26360 }, { "epoch": 3.00906, "grad_norm": 19.645679110488487, "learning_rate": 8.634962000149268e-06, "loss": 1.7387, "step": 26370 }, { "epoch": 3.00916, "grad_norm": 22.516607425239823, "learning_rate": 8.633849871133761e-06, "loss": 1.7881, "step": 26380 }, { "epoch": 3.00926, "grad_norm": 22.912510886899796, "learning_rate": 8.632737360944268e-06, "loss": 1.6892, "step": 26390 }, { "epoch": 3.00936, "grad_norm": 23.012245437527014, "learning_rate": 8.631624469697485e-06, "loss": 1.7083, "step": 26400 }, { "epoch": 3.00946, "grad_norm": 22.166780432664435, "learning_rate": 8.630511197510151e-06, "loss": 1.7498, "step": 26410 }, { "epoch": 3.00956, "grad_norm": 21.83909541776352, "learning_rate": 8.629397544499042e-06, "loss": 1.8188, "step": 26420 }, { "epoch": 3.00966, "grad_norm": 20.416940769923915, "learning_rate": 8.628283510780974e-06, "loss": 1.7989, "step": 26430 }, { "epoch": 3.00976, "grad_norm": 20.069671442851906, "learning_rate": 8.627169096472807e-06, "loss": 1.8249, "step": 26440 }, { "epoch": 3.00986, "grad_norm": 21.352259342848896, "learning_rate": 8.626054301691436e-06, "loss": 1.7691, "step": 26450 }, { "epoch": 3.00996, "grad_norm": 19.961970152763467, "learning_rate": 8.624939126553798e-06, "loss": 1.7358, "step": 26460 }, { "epoch": 3.01006, "grad_norm": 24.184495797660578, "learning_rate": 8.623823571176868e-06, "loss": 1.6687, "step": 26470 }, { "epoch": 3.01016, "grad_norm": 19.6508711072149, "learning_rate": 8.622707635677663e-06, "loss": 1.8397, "step": 26480 }, { "epoch": 3.01026, "grad_norm": 21.805036414540783, "learning_rate": 8.62159132017324e-06, "loss": 1.8062, "step": 26490 }, { "epoch": 3.01036, "grad_norm": 22.497945184671178, "learning_rate": 8.620474624780701e-06, "loss": 1.8046, "step": 26500 }, { "epoch": 3.01046, "grad_norm": 21.24230656080384, "learning_rate": 8.619357549617174e-06, "loss": 1.8007, "step": 26510 }, { "epoch": 3.01056, "grad_norm": 21.094023289971286, "learning_rate": 8.618240094799837e-06, "loss": 1.7451, "step": 26520 }, { "epoch": 3.01066, "grad_norm": 20.28708342573338, "learning_rate": 8.617122260445907e-06, "loss": 1.8226, "step": 26530 }, { "epoch": 3.01076, "grad_norm": 21.351172515220803, "learning_rate": 8.61600404667264e-06, "loss": 1.6974, "step": 26540 }, { "epoch": 3.01086, "grad_norm": 22.746705638335623, "learning_rate": 8.614885453597331e-06, "loss": 1.783, "step": 26550 }, { "epoch": 3.01096, "grad_norm": 23.992359995056812, "learning_rate": 8.613766481337315e-06, "loss": 1.7655, "step": 26560 }, { "epoch": 3.01106, "grad_norm": 20.2989009823196, "learning_rate": 8.612647130009969e-06, "loss": 1.7046, "step": 26570 }, { "epoch": 3.01116, "grad_norm": 22.34075240120841, "learning_rate": 8.611527399732703e-06, "loss": 1.6809, "step": 26580 }, { "epoch": 3.01126, "grad_norm": 23.071858241836505, "learning_rate": 8.610407290622975e-06, "loss": 1.7111, "step": 26590 }, { "epoch": 3.01136, "grad_norm": 20.95590364500467, "learning_rate": 8.609286802798279e-06, "loss": 1.7616, "step": 26600 }, { "epoch": 3.01146, "grad_norm": 22.361092470025525, "learning_rate": 8.608165936376149e-06, "loss": 1.7867, "step": 26610 }, { "epoch": 3.01156, "grad_norm": 22.54955891081587, "learning_rate": 8.607044691474158e-06, "loss": 1.6927, "step": 26620 }, { "epoch": 3.01166, "grad_norm": 21.3410254061243, "learning_rate": 8.60592306820992e-06, "loss": 1.7188, "step": 26630 }, { "epoch": 3.01176, "grad_norm": 20.98519994413703, "learning_rate": 8.604801066701087e-06, "loss": 1.7194, "step": 26640 }, { "epoch": 3.01186, "grad_norm": 20.948235179715304, "learning_rate": 8.603678687065353e-06, "loss": 1.7738, "step": 26650 }, { "epoch": 3.01196, "grad_norm": 19.68885610544228, "learning_rate": 8.60255592942045e-06, "loss": 1.7278, "step": 26660 }, { "epoch": 3.01206, "grad_norm": 21.121785122012632, "learning_rate": 8.60143279388415e-06, "loss": 1.7824, "step": 26670 }, { "epoch": 3.01216, "grad_norm": 20.76825037250141, "learning_rate": 8.600309280574264e-06, "loss": 1.6684, "step": 26680 }, { "epoch": 3.01226, "grad_norm": 22.519375939968008, "learning_rate": 8.599185389608645e-06, "loss": 1.6281, "step": 26690 }, { "epoch": 3.01236, "grad_norm": 24.945083172655753, "learning_rate": 8.598061121105181e-06, "loss": 1.7959, "step": 26700 }, { "epoch": 3.01246, "grad_norm": 21.85533831628231, "learning_rate": 8.596936475181808e-06, "loss": 1.8573, "step": 26710 }, { "epoch": 3.01256, "grad_norm": 19.930674353793435, "learning_rate": 8.595811451956491e-06, "loss": 1.7112, "step": 26720 }, { "epoch": 3.01266, "grad_norm": 20.019662578810788, "learning_rate": 8.594686051547241e-06, "loss": 1.7604, "step": 26730 }, { "epoch": 3.01276, "grad_norm": 21.94826302499981, "learning_rate": 8.593560274072107e-06, "loss": 1.715, "step": 26740 }, { "epoch": 3.01286, "grad_norm": 21.233082402203244, "learning_rate": 8.59243411964918e-06, "loss": 1.7835, "step": 26750 }, { "epoch": 3.01296, "grad_norm": 23.030261859393597, "learning_rate": 8.591307588396584e-06, "loss": 1.7407, "step": 26760 }, { "epoch": 3.01306, "grad_norm": 21.006046754126746, "learning_rate": 8.590180680432493e-06, "loss": 1.7539, "step": 26770 }, { "epoch": 3.01316, "grad_norm": 19.915924898627374, "learning_rate": 8.589053395875109e-06, "loss": 1.7135, "step": 26780 }, { "epoch": 3.01326, "grad_norm": 22.026533291838653, "learning_rate": 8.587925734842683e-06, "loss": 1.7042, "step": 26790 }, { "epoch": 3.01336, "grad_norm": 21.449757138342786, "learning_rate": 8.586797697453497e-06, "loss": 1.7169, "step": 26800 }, { "epoch": 3.01346, "grad_norm": 23.00928787404459, "learning_rate": 8.585669283825882e-06, "loss": 1.6813, "step": 26810 }, { "epoch": 3.01356, "grad_norm": 21.356152410607873, "learning_rate": 8.5845404940782e-06, "loss": 1.749, "step": 26820 }, { "epoch": 3.01366, "grad_norm": 21.42961562228771, "learning_rate": 8.583411328328856e-06, "loss": 1.725, "step": 26830 }, { "epoch": 3.01376, "grad_norm": 23.493524118143288, "learning_rate": 8.582281786696294e-06, "loss": 1.7785, "step": 26840 }, { "epoch": 3.01386, "grad_norm": 21.452855043874717, "learning_rate": 8.581151869299e-06, "loss": 1.7047, "step": 26850 }, { "epoch": 3.01396, "grad_norm": 22.940059556198726, "learning_rate": 8.580021576255494e-06, "loss": 1.7141, "step": 26860 }, { "epoch": 3.01406, "grad_norm": 22.206201770301742, "learning_rate": 8.578890907684341e-06, "loss": 1.7059, "step": 26870 }, { "epoch": 3.01416, "grad_norm": 21.15417464887445, "learning_rate": 8.577759863704142e-06, "loss": 1.7445, "step": 26880 }, { "epoch": 3.01426, "grad_norm": 22.171435069722474, "learning_rate": 8.576628444433539e-06, "loss": 1.7656, "step": 26890 }, { "epoch": 3.01436, "grad_norm": 21.193648409381428, "learning_rate": 8.57549664999121e-06, "loss": 1.691, "step": 26900 }, { "epoch": 3.01446, "grad_norm": 21.69790525318581, "learning_rate": 8.574364480495879e-06, "loss": 1.7846, "step": 26910 }, { "epoch": 3.01456, "grad_norm": 20.655027138456436, "learning_rate": 8.573231936066303e-06, "loss": 1.8098, "step": 26920 }, { "epoch": 3.01466, "grad_norm": 20.641165145754222, "learning_rate": 8.572099016821278e-06, "loss": 1.7692, "step": 26930 }, { "epoch": 3.01476, "grad_norm": 19.2750773498018, "learning_rate": 8.570965722879648e-06, "loss": 1.669, "step": 26940 }, { "epoch": 3.01486, "grad_norm": 18.376228963409726, "learning_rate": 8.569832054360284e-06, "loss": 1.8152, "step": 26950 }, { "epoch": 3.01496, "grad_norm": 21.344379027223006, "learning_rate": 8.568698011382108e-06, "loss": 1.7767, "step": 26960 }, { "epoch": 3.01506, "grad_norm": 20.338872474078695, "learning_rate": 8.567563594064072e-06, "loss": 1.7257, "step": 26970 }, { "epoch": 3.01516, "grad_norm": 23.067614514804273, "learning_rate": 8.566428802525172e-06, "loss": 1.7461, "step": 26980 }, { "epoch": 3.01526, "grad_norm": 23.219277213358378, "learning_rate": 8.565293636884446e-06, "loss": 1.8553, "step": 26990 }, { "epoch": 3.01536, "grad_norm": 22.66178385647687, "learning_rate": 8.564158097260962e-06, "loss": 1.7409, "step": 27000 }, { "epoch": 3.01546, "grad_norm": 20.651690555123306, "learning_rate": 8.563022183773835e-06, "loss": 1.8508, "step": 27010 }, { "epoch": 3.01556, "grad_norm": 21.57566421531804, "learning_rate": 8.561885896542219e-06, "loss": 1.8021, "step": 27020 }, { "epoch": 3.01566, "grad_norm": 20.642909807113703, "learning_rate": 8.560749235685303e-06, "loss": 1.768, "step": 27030 }, { "epoch": 3.01576, "grad_norm": 20.05366257856564, "learning_rate": 8.559612201322318e-06, "loss": 1.767, "step": 27040 }, { "epoch": 3.01586, "grad_norm": 20.66532307330548, "learning_rate": 8.558474793572534e-06, "loss": 1.8021, "step": 27050 }, { "epoch": 3.01596, "grad_norm": 19.884334858122255, "learning_rate": 8.557337012555259e-06, "loss": 1.7592, "step": 27060 }, { "epoch": 3.01606, "grad_norm": 23.114051787804833, "learning_rate": 8.556198858389842e-06, "loss": 1.7511, "step": 27070 }, { "epoch": 3.01616, "grad_norm": 19.22458200330105, "learning_rate": 8.555060331195669e-06, "loss": 1.669, "step": 27080 }, { "epoch": 3.01626, "grad_norm": 21.902927217085548, "learning_rate": 8.553921431092167e-06, "loss": 1.7192, "step": 27090 }, { "epoch": 3.01636, "grad_norm": 23.631358147379384, "learning_rate": 8.552782158198803e-06, "loss": 1.764, "step": 27100 }, { "epoch": 3.01646, "grad_norm": 21.420337169404238, "learning_rate": 8.551642512635076e-06, "loss": 1.7555, "step": 27110 }, { "epoch": 3.01656, "grad_norm": 20.553236316277445, "learning_rate": 8.550502494520536e-06, "loss": 1.7661, "step": 27120 }, { "epoch": 3.01666, "grad_norm": 20.442110705362836, "learning_rate": 8.549362103974762e-06, "loss": 1.7479, "step": 27130 }, { "epoch": 3.01676, "grad_norm": 19.853562054367984, "learning_rate": 8.548221341117376e-06, "loss": 1.7626, "step": 27140 }, { "epoch": 3.01686, "grad_norm": 22.6203499094433, "learning_rate": 8.547080206068039e-06, "loss": 1.7951, "step": 27150 }, { "epoch": 3.01696, "grad_norm": 18.970095452372963, "learning_rate": 8.545938698946453e-06, "loss": 1.7559, "step": 27160 }, { "epoch": 3.01706, "grad_norm": 20.031089362944662, "learning_rate": 8.544796819872352e-06, "loss": 1.7623, "step": 27170 }, { "epoch": 3.01716, "grad_norm": 21.10765851366659, "learning_rate": 8.543654568965517e-06, "loss": 1.6997, "step": 27180 }, { "epoch": 3.01726, "grad_norm": 23.092736461794686, "learning_rate": 8.542511946345765e-06, "loss": 1.7362, "step": 27190 }, { "epoch": 3.01736, "grad_norm": 22.24502904788912, "learning_rate": 8.541368952132951e-06, "loss": 1.7332, "step": 27200 }, { "epoch": 3.01746, "grad_norm": 22.538896588717027, "learning_rate": 8.540225586446971e-06, "loss": 1.7133, "step": 27210 }, { "epoch": 3.01756, "grad_norm": 20.24623933120051, "learning_rate": 8.539081849407756e-06, "loss": 1.6499, "step": 27220 }, { "epoch": 3.01766, "grad_norm": 22.31398321269233, "learning_rate": 8.53793774113528e-06, "loss": 1.6869, "step": 27230 }, { "epoch": 3.01776, "grad_norm": 21.223401325857836, "learning_rate": 8.536793261749556e-06, "loss": 1.7405, "step": 27240 }, { "epoch": 3.01786, "grad_norm": 21.65938974410035, "learning_rate": 8.535648411370634e-06, "loss": 1.6693, "step": 27250 }, { "epoch": 3.01796, "grad_norm": 21.165400331120896, "learning_rate": 8.5345031901186e-06, "loss": 1.783, "step": 27260 }, { "epoch": 3.01806, "grad_norm": 22.94000140735604, "learning_rate": 8.533357598113589e-06, "loss": 1.7613, "step": 27270 }, { "epoch": 3.01816, "grad_norm": 22.145879282817305, "learning_rate": 8.532211635475765e-06, "loss": 1.7625, "step": 27280 }, { "epoch": 3.01826, "grad_norm": 20.822613860008794, "learning_rate": 8.531065302325334e-06, "loss": 1.7514, "step": 27290 }, { "epoch": 3.01836, "grad_norm": 20.10020608320012, "learning_rate": 8.52991859878254e-06, "loss": 1.8016, "step": 27300 }, { "epoch": 3.01846, "grad_norm": 20.48586952000503, "learning_rate": 8.528771524967666e-06, "loss": 1.718, "step": 27310 }, { "epoch": 3.01856, "grad_norm": 20.28529370732811, "learning_rate": 8.527624081001038e-06, "loss": 1.7958, "step": 27320 }, { "epoch": 3.01866, "grad_norm": 19.585182619812176, "learning_rate": 8.526476267003016e-06, "loss": 1.737, "step": 27330 }, { "epoch": 3.01876, "grad_norm": 22.426872742770833, "learning_rate": 8.525328083094003e-06, "loss": 1.6692, "step": 27340 }, { "epoch": 3.01886, "grad_norm": 23.559558964194423, "learning_rate": 8.524179529394431e-06, "loss": 1.7742, "step": 27350 }, { "epoch": 3.01896, "grad_norm": 24.298206867200644, "learning_rate": 8.523030606024786e-06, "loss": 1.7128, "step": 27360 }, { "epoch": 3.01906, "grad_norm": 21.88487823422896, "learning_rate": 8.52188131310558e-06, "loss": 1.7615, "step": 27370 }, { "epoch": 3.01916, "grad_norm": 23.730055969481477, "learning_rate": 8.52073165075737e-06, "loss": 1.7641, "step": 27380 }, { "epoch": 3.01926, "grad_norm": 22.09454502466763, "learning_rate": 8.51958161910075e-06, "loss": 1.7097, "step": 27390 }, { "epoch": 3.01936, "grad_norm": 20.94770150521808, "learning_rate": 8.518431218256353e-06, "loss": 1.797, "step": 27400 }, { "epoch": 3.01946, "grad_norm": 21.92626900658616, "learning_rate": 8.517280448344852e-06, "loss": 1.7089, "step": 27410 }, { "epoch": 3.01956, "grad_norm": 19.47081124125156, "learning_rate": 8.516129309486956e-06, "loss": 1.7789, "step": 27420 }, { "epoch": 3.01966, "grad_norm": 20.08706655566616, "learning_rate": 8.514977801803412e-06, "loss": 1.7274, "step": 27430 }, { "epoch": 3.0197599999999998, "grad_norm": 23.13152387926961, "learning_rate": 8.513825925415011e-06, "loss": 1.6777, "step": 27440 }, { "epoch": 3.01986, "grad_norm": 22.291190649301328, "learning_rate": 8.51267368044258e-06, "loss": 1.727, "step": 27450 }, { "epoch": 3.01996, "grad_norm": 21.444443056553137, "learning_rate": 8.511521067006983e-06, "loss": 1.7353, "step": 27460 }, { "epoch": 3.02006, "grad_norm": 20.16470716823579, "learning_rate": 8.51036808522912e-06, "loss": 1.7968, "step": 27470 }, { "epoch": 3.02016, "grad_norm": 19.756852055101874, "learning_rate": 8.509214735229941e-06, "loss": 1.8192, "step": 27480 }, { "epoch": 3.02026, "grad_norm": 19.21633702142266, "learning_rate": 8.50806101713042e-06, "loss": 1.6913, "step": 27490 }, { "epoch": 3.02036, "grad_norm": 20.722335649081092, "learning_rate": 8.506906931051581e-06, "loss": 1.7911, "step": 27500 }, { "epoch": 3.02046, "grad_norm": 21.047886527405986, "learning_rate": 8.505752477114484e-06, "loss": 1.7096, "step": 27510 }, { "epoch": 3.02056, "grad_norm": 21.869040455269854, "learning_rate": 8.504597655440219e-06, "loss": 1.7397, "step": 27520 }, { "epoch": 3.02066, "grad_norm": 19.444515533914675, "learning_rate": 8.503442466149926e-06, "loss": 1.7822, "step": 27530 }, { "epoch": 3.02076, "grad_norm": 20.62554419912082, "learning_rate": 8.502286909364779e-06, "loss": 1.7488, "step": 27540 }, { "epoch": 3.02086, "grad_norm": 20.124094416038346, "learning_rate": 8.501130985205989e-06, "loss": 1.7111, "step": 27550 }, { "epoch": 3.02096, "grad_norm": 22.928726929859263, "learning_rate": 8.499974693794808e-06, "loss": 1.7352, "step": 27560 }, { "epoch": 3.02106, "grad_norm": 22.205328250162935, "learning_rate": 8.498818035252527e-06, "loss": 1.7206, "step": 27570 }, { "epoch": 3.02116, "grad_norm": 22.571571137018022, "learning_rate": 8.49766100970047e-06, "loss": 1.7729, "step": 27580 }, { "epoch": 3.02126, "grad_norm": 19.944673810648062, "learning_rate": 8.496503617260008e-06, "loss": 1.7575, "step": 27590 }, { "epoch": 3.02136, "grad_norm": 21.893521832862238, "learning_rate": 8.495345858052544e-06, "loss": 1.7753, "step": 27600 }, { "epoch": 3.02146, "grad_norm": 22.559422708828972, "learning_rate": 8.494187732199522e-06, "loss": 1.7953, "step": 27610 }, { "epoch": 3.02156, "grad_norm": 21.761980992569487, "learning_rate": 8.493029239822423e-06, "loss": 1.8071, "step": 27620 }, { "epoch": 3.02166, "grad_norm": 22.01900262018133, "learning_rate": 8.491870381042771e-06, "loss": 1.8122, "step": 27630 }, { "epoch": 3.02176, "grad_norm": 19.978925023427564, "learning_rate": 8.49071115598212e-06, "loss": 1.7877, "step": 27640 }, { "epoch": 3.02186, "grad_norm": 21.267300326760434, "learning_rate": 8.489551564762069e-06, "loss": 1.7408, "step": 27650 }, { "epoch": 3.02196, "grad_norm": 20.751987357419775, "learning_rate": 8.488391607504256e-06, "loss": 1.6882, "step": 27660 }, { "epoch": 3.02206, "grad_norm": 20.39164456484309, "learning_rate": 8.487231284330353e-06, "loss": 1.7563, "step": 27670 }, { "epoch": 3.02216, "grad_norm": 22.993203015303326, "learning_rate": 8.486070595362072e-06, "loss": 1.7329, "step": 27680 }, { "epoch": 3.02226, "grad_norm": 22.19120663192919, "learning_rate": 8.484909540721167e-06, "loss": 1.761, "step": 27690 }, { "epoch": 3.02236, "grad_norm": 21.401029076279443, "learning_rate": 8.483748120529425e-06, "loss": 1.7567, "step": 27700 }, { "epoch": 3.02246, "grad_norm": 21.957362416564543, "learning_rate": 8.482586334908671e-06, "loss": 1.7765, "step": 27710 }, { "epoch": 3.02256, "grad_norm": 18.507444414792495, "learning_rate": 8.481424183980776e-06, "loss": 1.8334, "step": 27720 }, { "epoch": 3.02266, "grad_norm": 18.76606559294036, "learning_rate": 8.480261667867641e-06, "loss": 1.7961, "step": 27730 }, { "epoch": 3.02276, "grad_norm": 21.769561254265504, "learning_rate": 8.47909878669121e-06, "loss": 1.8185, "step": 27740 }, { "epoch": 3.02286, "grad_norm": 20.07435019352176, "learning_rate": 8.477935540573464e-06, "loss": 1.8252, "step": 27750 }, { "epoch": 3.02296, "grad_norm": 22.966804079868197, "learning_rate": 8.476771929636419e-06, "loss": 1.812, "step": 27760 }, { "epoch": 3.02306, "grad_norm": 22.66227820108647, "learning_rate": 8.475607954002136e-06, "loss": 1.8146, "step": 27770 }, { "epoch": 3.02316, "grad_norm": 19.723336360092866, "learning_rate": 8.47444361379271e-06, "loss": 1.7801, "step": 27780 }, { "epoch": 3.02326, "grad_norm": 22.60958858743952, "learning_rate": 8.473278909130273e-06, "loss": 1.7908, "step": 27790 }, { "epoch": 3.02336, "grad_norm": 20.836002246344883, "learning_rate": 8.472113840137e-06, "loss": 1.7622, "step": 27800 }, { "epoch": 3.02346, "grad_norm": 19.90226502835271, "learning_rate": 8.4709484069351e-06, "loss": 1.8043, "step": 27810 }, { "epoch": 3.02356, "grad_norm": 23.233551746596504, "learning_rate": 8.469782609646821e-06, "loss": 1.8198, "step": 27820 }, { "epoch": 3.02366, "grad_norm": 22.463691243611862, "learning_rate": 8.468616448394449e-06, "loss": 1.7605, "step": 27830 }, { "epoch": 3.02376, "grad_norm": 22.437787783002417, "learning_rate": 8.46744992330031e-06, "loss": 1.7589, "step": 27840 }, { "epoch": 3.02386, "grad_norm": 21.65484066935428, "learning_rate": 8.466283034486766e-06, "loss": 1.7583, "step": 27850 }, { "epoch": 3.02396, "grad_norm": 21.60047452007251, "learning_rate": 8.46511578207622e-06, "loss": 1.7447, "step": 27860 }, { "epoch": 3.02406, "grad_norm": 21.209692869619133, "learning_rate": 8.463948166191113e-06, "loss": 1.6981, "step": 27870 }, { "epoch": 3.02416, "grad_norm": 22.445611389514912, "learning_rate": 8.462780186953919e-06, "loss": 1.677, "step": 27880 }, { "epoch": 3.02426, "grad_norm": 22.7996973802479, "learning_rate": 8.461611844487151e-06, "loss": 1.7271, "step": 27890 }, { "epoch": 3.02436, "grad_norm": 22.165374769244476, "learning_rate": 8.46044313891337e-06, "loss": 1.7654, "step": 27900 }, { "epoch": 3.02446, "grad_norm": 20.796597281082665, "learning_rate": 8.459274070355164e-06, "loss": 1.7035, "step": 27910 }, { "epoch": 3.02456, "grad_norm": 20.418072870757342, "learning_rate": 8.458104638935162e-06, "loss": 1.807, "step": 27920 }, { "epoch": 3.02466, "grad_norm": 22.452227603243564, "learning_rate": 8.456934844776033e-06, "loss": 1.7711, "step": 27930 }, { "epoch": 3.02476, "grad_norm": 20.66369570005995, "learning_rate": 8.455764688000483e-06, "loss": 1.7506, "step": 27940 }, { "epoch": 3.02486, "grad_norm": 19.956344740874954, "learning_rate": 8.454594168731256e-06, "loss": 1.7049, "step": 27950 }, { "epoch": 3.02496, "grad_norm": 21.35074487885464, "learning_rate": 8.453423287091134e-06, "loss": 1.7242, "step": 27960 }, { "epoch": 3.02506, "grad_norm": 19.94304670083272, "learning_rate": 8.452252043202938e-06, "loss": 1.6932, "step": 27970 }, { "epoch": 3.02516, "grad_norm": 21.822279530098534, "learning_rate": 8.451080437189523e-06, "loss": 1.7162, "step": 27980 }, { "epoch": 3.02526, "grad_norm": 20.890091205212954, "learning_rate": 8.449908469173788e-06, "loss": 1.608, "step": 27990 }, { "epoch": 3.02536, "grad_norm": 24.06464041534843, "learning_rate": 8.448736139278665e-06, "loss": 1.7446, "step": 28000 }, { "epoch": 3.02546, "grad_norm": 24.52222250594515, "learning_rate": 8.44756344762713e-06, "loss": 1.7552, "step": 28010 }, { "epoch": 3.02556, "grad_norm": 20.35893139043707, "learning_rate": 8.446390394342188e-06, "loss": 1.7181, "step": 28020 }, { "epoch": 3.02566, "grad_norm": 20.65631682756198, "learning_rate": 8.445216979546888e-06, "loss": 1.6919, "step": 28030 }, { "epoch": 3.02576, "grad_norm": 22.715658743629305, "learning_rate": 8.444043203364318e-06, "loss": 1.6423, "step": 28040 }, { "epoch": 3.02586, "grad_norm": 21.80820502553151, "learning_rate": 8.4428690659176e-06, "loss": 1.7638, "step": 28050 }, { "epoch": 3.02596, "grad_norm": 21.054954536574705, "learning_rate": 8.441694567329896e-06, "loss": 1.7473, "step": 28060 }, { "epoch": 3.02606, "grad_norm": 22.726101945161773, "learning_rate": 8.440519707724407e-06, "loss": 1.6279, "step": 28070 }, { "epoch": 3.02616, "grad_norm": 22.31505852614239, "learning_rate": 8.439344487224365e-06, "loss": 1.7219, "step": 28080 }, { "epoch": 3.02626, "grad_norm": 20.670668196095964, "learning_rate": 8.438168905953051e-06, "loss": 1.6957, "step": 28090 }, { "epoch": 3.02636, "grad_norm": 22.89499988098515, "learning_rate": 8.436992964033775e-06, "loss": 1.6794, "step": 28100 }, { "epoch": 3.02646, "grad_norm": 22.589671636079135, "learning_rate": 8.435816661589888e-06, "loss": 1.7676, "step": 28110 }, { "epoch": 3.02656, "grad_norm": 22.537335570858847, "learning_rate": 8.434639998744782e-06, "loss": 1.7513, "step": 28120 }, { "epoch": 3.02666, "grad_norm": 20.76076096173321, "learning_rate": 8.433462975621877e-06, "loss": 1.6499, "step": 28130 }, { "epoch": 3.02676, "grad_norm": 22.29452074421268, "learning_rate": 8.432285592344644e-06, "loss": 1.6724, "step": 28140 }, { "epoch": 3.02686, "grad_norm": 22.79967984833594, "learning_rate": 8.431107849036582e-06, "loss": 1.6558, "step": 28150 }, { "epoch": 3.02696, "grad_norm": 22.74404196882653, "learning_rate": 8.429929745821229e-06, "loss": 1.7317, "step": 28160 }, { "epoch": 3.02706, "grad_norm": 20.964458179325582, "learning_rate": 8.428751282822166e-06, "loss": 1.6937, "step": 28170 }, { "epoch": 3.02716, "grad_norm": 21.12990240402949, "learning_rate": 8.427572460163006e-06, "loss": 1.6388, "step": 28180 }, { "epoch": 3.02726, "grad_norm": 20.905233064324353, "learning_rate": 8.426393277967403e-06, "loss": 1.636, "step": 28190 }, { "epoch": 3.02736, "grad_norm": 19.138222688310552, "learning_rate": 8.425213736359049e-06, "loss": 1.6762, "step": 28200 }, { "epoch": 3.02746, "grad_norm": 21.579626324678088, "learning_rate": 8.424033835461669e-06, "loss": 1.6649, "step": 28210 }, { "epoch": 3.02756, "grad_norm": 22.118916633374866, "learning_rate": 8.422853575399032e-06, "loss": 1.7033, "step": 28220 }, { "epoch": 3.02766, "grad_norm": 20.274219200579086, "learning_rate": 8.42167295629494e-06, "loss": 1.7121, "step": 28230 }, { "epoch": 3.02776, "grad_norm": 20.458365673035946, "learning_rate": 8.420491978273235e-06, "loss": 1.7573, "step": 28240 }, { "epoch": 3.02786, "grad_norm": 20.378660851255844, "learning_rate": 8.419310641457798e-06, "loss": 1.6863, "step": 28250 }, { "epoch": 3.02796, "grad_norm": 21.548827012446296, "learning_rate": 8.418128945972543e-06, "loss": 1.6786, "step": 28260 }, { "epoch": 3.02806, "grad_norm": 18.982326399113692, "learning_rate": 8.416946891941426e-06, "loss": 1.7027, "step": 28270 }, { "epoch": 3.02816, "grad_norm": 20.65206354691188, "learning_rate": 8.41576447948844e-06, "loss": 1.6819, "step": 28280 }, { "epoch": 3.02826, "grad_norm": 22.603484744908663, "learning_rate": 8.41458170873761e-06, "loss": 1.6635, "step": 28290 }, { "epoch": 3.02836, "grad_norm": 21.033344119501205, "learning_rate": 8.41339857981301e-06, "loss": 1.6898, "step": 28300 }, { "epoch": 3.02846, "grad_norm": 18.35741467701837, "learning_rate": 8.412215092838738e-06, "loss": 1.732, "step": 28310 }, { "epoch": 3.02856, "grad_norm": 18.770209741329076, "learning_rate": 8.411031247938942e-06, "loss": 1.7193, "step": 28320 }, { "epoch": 3.02866, "grad_norm": 21.812709508899957, "learning_rate": 8.409847045237798e-06, "loss": 1.6577, "step": 28330 }, { "epoch": 3.02876, "grad_norm": 19.990575843228616, "learning_rate": 8.408662484859524e-06, "loss": 1.6698, "step": 28340 }, { "epoch": 3.02886, "grad_norm": 20.397477872426172, "learning_rate": 8.407477566928377e-06, "loss": 1.6578, "step": 28350 }, { "epoch": 3.02896, "grad_norm": 19.611830934446342, "learning_rate": 8.406292291568645e-06, "loss": 1.7588, "step": 28360 }, { "epoch": 3.02906, "grad_norm": 23.280412881357016, "learning_rate": 8.405106658904663e-06, "loss": 1.6741, "step": 28370 }, { "epoch": 3.02916, "grad_norm": 28.375095929289277, "learning_rate": 8.403920669060794e-06, "loss": 1.6858, "step": 28380 }, { "epoch": 3.02926, "grad_norm": 21.72577417365837, "learning_rate": 8.402734322161446e-06, "loss": 1.7476, "step": 28390 }, { "epoch": 3.02936, "grad_norm": 21.739287276405413, "learning_rate": 8.401547618331062e-06, "loss": 1.6999, "step": 28400 }, { "epoch": 3.02946, "grad_norm": 20.11713050578066, "learning_rate": 8.400360557694118e-06, "loss": 1.655, "step": 28410 }, { "epoch": 3.02956, "grad_norm": 20.047451255817435, "learning_rate": 8.399173140375133e-06, "loss": 1.6149, "step": 28420 }, { "epoch": 3.02966, "grad_norm": 21.99288154400113, "learning_rate": 8.397985366498661e-06, "loss": 1.6598, "step": 28430 }, { "epoch": 3.02976, "grad_norm": 23.010420572016272, "learning_rate": 8.396797236189297e-06, "loss": 1.6717, "step": 28440 }, { "epoch": 3.02986, "grad_norm": 26.46515648528215, "learning_rate": 8.395608749571666e-06, "loss": 1.708, "step": 28450 }, { "epoch": 3.02996, "grad_norm": 22.81994424745803, "learning_rate": 8.394419906770436e-06, "loss": 1.6397, "step": 28460 }, { "epoch": 3.03006, "grad_norm": 25.22664364028834, "learning_rate": 8.393230707910313e-06, "loss": 1.6176, "step": 28470 }, { "epoch": 3.03016, "grad_norm": 25.231379133217676, "learning_rate": 8.39204115311604e-06, "loss": 1.6471, "step": 28480 }, { "epoch": 3.03026, "grad_norm": 23.799300980868846, "learning_rate": 8.39085124251239e-06, "loss": 1.6275, "step": 28490 }, { "epoch": 3.03036, "grad_norm": 21.586316682752567, "learning_rate": 8.38966097622418e-06, "loss": 1.6709, "step": 28500 }, { "epoch": 3.03046, "grad_norm": 20.91902074868421, "learning_rate": 8.38847035437627e-06, "loss": 1.5869, "step": 28510 }, { "epoch": 3.03056, "grad_norm": 21.763615445598713, "learning_rate": 8.387279377093543e-06, "loss": 1.6581, "step": 28520 }, { "epoch": 3.03066, "grad_norm": 20.89853287522081, "learning_rate": 8.386088044500932e-06, "loss": 1.6401, "step": 28530 }, { "epoch": 3.03076, "grad_norm": 20.315245523031383, "learning_rate": 8.3848963567234e-06, "loss": 1.5748, "step": 28540 }, { "epoch": 3.03086, "grad_norm": 19.24470614329896, "learning_rate": 8.38370431388595e-06, "loss": 1.6628, "step": 28550 }, { "epoch": 3.03096, "grad_norm": 20.97046882209617, "learning_rate": 8.38251191611362e-06, "loss": 1.6591, "step": 28560 }, { "epoch": 3.03106, "grad_norm": 20.0949735323826, "learning_rate": 8.38131916353149e-06, "loss": 1.5769, "step": 28570 }, { "epoch": 3.03116, "grad_norm": 20.99269709364024, "learning_rate": 8.380126056264674e-06, "loss": 1.6458, "step": 28580 }, { "epoch": 3.03126, "grad_norm": 21.65252025521125, "learning_rate": 8.37893259443832e-06, "loss": 1.6771, "step": 28590 }, { "epoch": 3.03136, "grad_norm": 21.70994330762107, "learning_rate": 8.37773877817762e-06, "loss": 1.6323, "step": 28600 }, { "epoch": 3.03146, "grad_norm": 19.89208342038132, "learning_rate": 8.3765446076078e-06, "loss": 1.5675, "step": 28610 }, { "epoch": 3.03156, "grad_norm": 21.285844985692624, "learning_rate": 8.375350082854121e-06, "loss": 1.5417, "step": 28620 }, { "epoch": 3.03166, "grad_norm": 21.195568477661954, "learning_rate": 8.374155204041882e-06, "loss": 1.5742, "step": 28630 }, { "epoch": 3.03176, "grad_norm": 23.47705701753617, "learning_rate": 8.372959971296425e-06, "loss": 1.5745, "step": 28640 }, { "epoch": 3.03186, "grad_norm": 21.134925214491968, "learning_rate": 8.37176438474312e-06, "loss": 1.652, "step": 28650 }, { "epoch": 3.03196, "grad_norm": 22.449701837102804, "learning_rate": 8.37056844450738e-06, "loss": 1.6195, "step": 28660 }, { "epoch": 3.03206, "grad_norm": 20.887137366804165, "learning_rate": 8.369372150714653e-06, "loss": 1.6171, "step": 28670 }, { "epoch": 3.03216, "grad_norm": 20.365049266276884, "learning_rate": 8.368175503490425e-06, "loss": 1.5867, "step": 28680 }, { "epoch": 3.03226, "grad_norm": 20.353606159462075, "learning_rate": 8.366978502960218e-06, "loss": 1.5904, "step": 28690 }, { "epoch": 3.03236, "grad_norm": 19.038264072712465, "learning_rate": 8.365781149249596e-06, "loss": 1.5595, "step": 28700 }, { "epoch": 3.03246, "grad_norm": 20.592804943478853, "learning_rate": 8.364583442484152e-06, "loss": 1.5466, "step": 28710 }, { "epoch": 3.03256, "grad_norm": 22.30498386900118, "learning_rate": 8.363385382789518e-06, "loss": 1.6029, "step": 28720 }, { "epoch": 3.03266, "grad_norm": 23.453032330822015, "learning_rate": 8.362186970291368e-06, "loss": 1.606, "step": 28730 }, { "epoch": 3.03276, "grad_norm": 21.05410526360274, "learning_rate": 8.360988205115409e-06, "loss": 1.5995, "step": 28740 }, { "epoch": 3.03286, "grad_norm": 19.826212161971117, "learning_rate": 8.359789087387387e-06, "loss": 1.5631, "step": 28750 }, { "epoch": 3.03296, "grad_norm": 22.08121741287819, "learning_rate": 8.358589617233081e-06, "loss": 1.6777, "step": 28760 }, { "epoch": 3.03306, "grad_norm": 22.500293476568274, "learning_rate": 8.357389794778315e-06, "loss": 1.6664, "step": 28770 }, { "epoch": 3.03316, "grad_norm": 21.874983916122638, "learning_rate": 8.35618962014894e-06, "loss": 1.6745, "step": 28780 }, { "epoch": 3.03326, "grad_norm": 21.921493934398192, "learning_rate": 8.35498909347085e-06, "loss": 1.6861, "step": 28790 }, { "epoch": 3.03336, "grad_norm": 19.651570722692437, "learning_rate": 8.353788214869975e-06, "loss": 1.6457, "step": 28800 }, { "epoch": 3.03346, "grad_norm": 21.96050863136269, "learning_rate": 8.352586984472282e-06, "loss": 1.685, "step": 28810 }, { "epoch": 3.03356, "grad_norm": 20.46388042943333, "learning_rate": 8.351385402403776e-06, "loss": 1.639, "step": 28820 }, { "epoch": 3.03366, "grad_norm": 17.952111971237205, "learning_rate": 8.350183468790493e-06, "loss": 1.4989, "step": 28830 }, { "epoch": 3.03376, "grad_norm": 22.033631577701485, "learning_rate": 8.348981183758513e-06, "loss": 1.6165, "step": 28840 }, { "epoch": 3.03386, "grad_norm": 21.64120000537166, "learning_rate": 8.347778547433951e-06, "loss": 1.6236, "step": 28850 }, { "epoch": 3.03396, "grad_norm": 22.316747810049474, "learning_rate": 8.346575559942957e-06, "loss": 1.5717, "step": 28860 }, { "epoch": 3.03406, "grad_norm": 21.99909584843179, "learning_rate": 8.345372221411719e-06, "loss": 1.6277, "step": 28870 }, { "epoch": 3.03416, "grad_norm": 22.361078492438846, "learning_rate": 8.344168531966459e-06, "loss": 1.6531, "step": 28880 }, { "epoch": 3.03426, "grad_norm": 19.012741213180387, "learning_rate": 8.342964491733444e-06, "loss": 1.6472, "step": 28890 }, { "epoch": 3.03436, "grad_norm": 19.958516926015776, "learning_rate": 8.341760100838967e-06, "loss": 1.5919, "step": 28900 }, { "epoch": 3.03446, "grad_norm": 21.11834916230186, "learning_rate": 8.340555359409364e-06, "loss": 1.6609, "step": 28910 }, { "epoch": 3.03456, "grad_norm": 20.843512143851573, "learning_rate": 8.33935026757101e-06, "loss": 1.6253, "step": 28920 }, { "epoch": 3.03466, "grad_norm": 18.482553558835527, "learning_rate": 8.33814482545031e-06, "loss": 1.5641, "step": 28930 }, { "epoch": 3.03476, "grad_norm": 18.938640471940758, "learning_rate": 8.336939033173711e-06, "loss": 1.5791, "step": 28940 }, { "epoch": 3.03486, "grad_norm": 19.535347678122466, "learning_rate": 8.335732890867695e-06, "loss": 1.5908, "step": 28950 }, { "epoch": 3.03496, "grad_norm": 23.5831717214132, "learning_rate": 8.33452639865878e-06, "loss": 1.6306, "step": 28960 }, { "epoch": 3.03506, "grad_norm": 20.30665103478643, "learning_rate": 8.333319556673521e-06, "loss": 1.5723, "step": 28970 }, { "epoch": 3.03516, "grad_norm": 22.630268195507178, "learning_rate": 8.332112365038512e-06, "loss": 1.5929, "step": 28980 }, { "epoch": 3.03526, "grad_norm": 20.777563084703143, "learning_rate": 8.330904823880382e-06, "loss": 1.5877, "step": 28990 }, { "epoch": 3.03536, "grad_norm": 20.933111698134617, "learning_rate": 8.329696933325792e-06, "loss": 1.5823, "step": 29000 }, { "epoch": 3.03546, "grad_norm": 22.45482847265105, "learning_rate": 8.32848869350145e-06, "loss": 1.6793, "step": 29010 }, { "epoch": 3.03556, "grad_norm": 21.091497120739835, "learning_rate": 8.32728010453409e-06, "loss": 1.5774, "step": 29020 }, { "epoch": 3.03566, "grad_norm": 21.562289589361452, "learning_rate": 8.326071166550491e-06, "loss": 1.5801, "step": 29030 }, { "epoch": 3.03576, "grad_norm": 22.538817168037898, "learning_rate": 8.324861879677462e-06, "loss": 1.5645, "step": 29040 }, { "epoch": 3.03586, "grad_norm": 20.781922864904335, "learning_rate": 8.323652244041854e-06, "loss": 1.6172, "step": 29050 }, { "epoch": 3.03596, "grad_norm": 23.214071587566185, "learning_rate": 8.322442259770551e-06, "loss": 1.5552, "step": 29060 }, { "epoch": 3.03606, "grad_norm": 22.414181959793407, "learning_rate": 8.321231926990476e-06, "loss": 1.5796, "step": 29070 }, { "epoch": 3.03616, "grad_norm": 21.540267918188196, "learning_rate": 8.320021245828584e-06, "loss": 1.5669, "step": 29080 }, { "epoch": 3.03626, "grad_norm": 22.543388064212515, "learning_rate": 8.318810216411873e-06, "loss": 1.6542, "step": 29090 }, { "epoch": 3.03636, "grad_norm": 20.890997292913447, "learning_rate": 8.317598838867373e-06, "loss": 1.657, "step": 29100 }, { "epoch": 3.03646, "grad_norm": 19.357246908583235, "learning_rate": 8.316387113322154e-06, "loss": 1.5969, "step": 29110 }, { "epoch": 3.03656, "grad_norm": 23.5753281838889, "learning_rate": 8.315175039903317e-06, "loss": 1.5983, "step": 29120 }, { "epoch": 3.03666, "grad_norm": 21.484440553879526, "learning_rate": 8.313962618738005e-06, "loss": 1.6932, "step": 29130 }, { "epoch": 3.03676, "grad_norm": 20.496497529224392, "learning_rate": 8.312749849953395e-06, "loss": 1.67, "step": 29140 }, { "epoch": 3.03686, "grad_norm": 20.852000888213592, "learning_rate": 8.3115367336767e-06, "loss": 1.6359, "step": 29150 }, { "epoch": 3.03696, "grad_norm": 20.713158770773546, "learning_rate": 8.310323270035171e-06, "loss": 1.5804, "step": 29160 }, { "epoch": 3.03706, "grad_norm": 22.935161361729907, "learning_rate": 8.309109459156096e-06, "loss": 1.699, "step": 29170 }, { "epoch": 3.03716, "grad_norm": 19.900552518710015, "learning_rate": 8.307895301166795e-06, "loss": 1.6528, "step": 29180 }, { "epoch": 3.03726, "grad_norm": 21.06770403578989, "learning_rate": 8.306680796194631e-06, "loss": 1.6124, "step": 29190 }, { "epoch": 3.03736, "grad_norm": 22.908383589030837, "learning_rate": 8.305465944366995e-06, "loss": 1.6307, "step": 29200 }, { "epoch": 3.03746, "grad_norm": 21.902360361737927, "learning_rate": 8.304250745811327e-06, "loss": 1.7444, "step": 29210 }, { "epoch": 3.03756, "grad_norm": 21.608243673728975, "learning_rate": 8.303035200655089e-06, "loss": 1.6625, "step": 29220 }, { "epoch": 3.03766, "grad_norm": 20.02734574618161, "learning_rate": 8.301819309025787e-06, "loss": 1.6629, "step": 29230 }, { "epoch": 3.03776, "grad_norm": 22.949545224256706, "learning_rate": 8.300603071050965e-06, "loss": 1.6645, "step": 29240 }, { "epoch": 3.03786, "grad_norm": 19.482339609468195, "learning_rate": 8.299386486858201e-06, "loss": 1.5833, "step": 29250 }, { "epoch": 3.03796, "grad_norm": 20.605650694074164, "learning_rate": 8.298169556575107e-06, "loss": 1.5931, "step": 29260 }, { "epoch": 3.03806, "grad_norm": 20.99756049157662, "learning_rate": 8.29695228032933e-06, "loss": 1.5415, "step": 29270 }, { "epoch": 3.03816, "grad_norm": 22.001678011682348, "learning_rate": 8.295734658248564e-06, "loss": 1.6347, "step": 29280 }, { "epoch": 3.03826, "grad_norm": 23.380906606224052, "learning_rate": 8.294516690460529e-06, "loss": 1.6413, "step": 29290 }, { "epoch": 3.03836, "grad_norm": 23.86128158908324, "learning_rate": 8.293298377092982e-06, "loss": 1.655, "step": 29300 }, { "epoch": 3.03846, "grad_norm": 19.55046524653931, "learning_rate": 8.29207971827372e-06, "loss": 1.651, "step": 29310 }, { "epoch": 3.03856, "grad_norm": 20.43651596579343, "learning_rate": 8.290860714130574e-06, "loss": 1.65, "step": 29320 }, { "epoch": 3.03866, "grad_norm": 22.051682732457227, "learning_rate": 8.289641364791413e-06, "loss": 1.6568, "step": 29330 }, { "epoch": 3.03876, "grad_norm": 18.902825185924442, "learning_rate": 8.28842167038414e-06, "loss": 1.6829, "step": 29340 }, { "epoch": 3.03886, "grad_norm": 22.010710633808852, "learning_rate": 8.287201631036697e-06, "loss": 1.6622, "step": 29350 }, { "epoch": 3.03896, "grad_norm": 25.891032758297637, "learning_rate": 8.285981246877058e-06, "loss": 1.6275, "step": 29360 }, { "epoch": 3.03906, "grad_norm": 19.978103358174874, "learning_rate": 8.28476051803324e-06, "loss": 1.6186, "step": 29370 }, { "epoch": 3.03916, "grad_norm": 21.24254785272336, "learning_rate": 8.283539444633284e-06, "loss": 1.6094, "step": 29380 }, { "epoch": 3.03926, "grad_norm": 21.37244057667097, "learning_rate": 8.282318026805282e-06, "loss": 1.6465, "step": 29390 }, { "epoch": 3.03936, "grad_norm": 20.782522148699613, "learning_rate": 8.281096264677352e-06, "loss": 1.5357, "step": 29400 }, { "epoch": 3.03946, "grad_norm": 23.47509048541154, "learning_rate": 8.279874158377654e-06, "loss": 1.6202, "step": 29410 }, { "epoch": 3.03956, "grad_norm": 23.010711865777335, "learning_rate": 8.278651708034378e-06, "loss": 1.632, "step": 29420 }, { "epoch": 3.03966, "grad_norm": 21.352280100035685, "learning_rate": 8.277428913775753e-06, "loss": 1.6447, "step": 29430 }, { "epoch": 3.03976, "grad_norm": 20.38340721695856, "learning_rate": 8.276205775730048e-06, "loss": 1.5696, "step": 29440 }, { "epoch": 3.03986, "grad_norm": 22.618706489545122, "learning_rate": 8.274982294025562e-06, "loss": 1.5903, "step": 29450 }, { "epoch": 3.03996, "grad_norm": 20.784002957966496, "learning_rate": 8.27375846879063e-06, "loss": 1.6834, "step": 29460 }, { "epoch": 3.04006, "grad_norm": 20.73897007712858, "learning_rate": 8.272534300153633e-06, "loss": 1.5996, "step": 29470 }, { "epoch": 3.04016, "grad_norm": 21.854589136924197, "learning_rate": 8.271309788242973e-06, "loss": 1.6574, "step": 29480 }, { "epoch": 3.04026, "grad_norm": 22.257740821090685, "learning_rate": 8.270084933187101e-06, "loss": 1.6566, "step": 29490 }, { "epoch": 3.04036, "grad_norm": 22.843539717385724, "learning_rate": 8.268859735114497e-06, "loss": 1.6541, "step": 29500 }, { "epoch": 3.04046, "grad_norm": 22.203472815626046, "learning_rate": 8.267634194153675e-06, "loss": 1.6565, "step": 29510 }, { "epoch": 3.04056, "grad_norm": 22.25505771756972, "learning_rate": 8.266408310433194e-06, "loss": 1.5688, "step": 29520 }, { "epoch": 3.04066, "grad_norm": 24.37887967015604, "learning_rate": 8.26518208408164e-06, "loss": 1.5675, "step": 29530 }, { "epoch": 3.04076, "grad_norm": 21.81861985766923, "learning_rate": 8.26395551522764e-06, "loss": 1.575, "step": 29540 }, { "epoch": 3.04086, "grad_norm": 20.635191043421873, "learning_rate": 8.262728603999857e-06, "loss": 1.6393, "step": 29550 }, { "epoch": 3.04096, "grad_norm": 24.32727044969092, "learning_rate": 8.261501350526984e-06, "loss": 1.6467, "step": 29560 }, { "epoch": 3.04106, "grad_norm": 21.112769586655983, "learning_rate": 8.260273754937756e-06, "loss": 1.7334, "step": 29570 }, { "epoch": 3.04116, "grad_norm": 22.636160541900953, "learning_rate": 8.259045817360942e-06, "loss": 1.6324, "step": 29580 }, { "epoch": 3.04126, "grad_norm": 20.42535583625274, "learning_rate": 8.257817537925348e-06, "loss": 1.6858, "step": 29590 }, { "epoch": 3.04136, "grad_norm": 21.86094120253192, "learning_rate": 8.256588916759816e-06, "loss": 1.631, "step": 29600 }, { "epoch": 3.04146, "grad_norm": 20.22673305249205, "learning_rate": 8.25535995399322e-06, "loss": 1.6386, "step": 29610 }, { "epoch": 3.04156, "grad_norm": 19.94446415655491, "learning_rate": 8.254130649754473e-06, "loss": 1.5817, "step": 29620 }, { "epoch": 3.04166, "grad_norm": 21.521710104462734, "learning_rate": 8.252901004172525e-06, "loss": 1.583, "step": 29630 }, { "epoch": 3.04176, "grad_norm": 23.162089033275123, "learning_rate": 8.251671017376357e-06, "loss": 1.607, "step": 29640 }, { "epoch": 3.04186, "grad_norm": 22.679907809370988, "learning_rate": 8.250440689494991e-06, "loss": 1.6525, "step": 29650 }, { "epoch": 3.04196, "grad_norm": 22.085252946684278, "learning_rate": 8.249210020657483e-06, "loss": 1.6012, "step": 29660 }, { "epoch": 3.04206, "grad_norm": 21.934062149639114, "learning_rate": 8.247979010992923e-06, "loss": 1.6001, "step": 29670 }, { "epoch": 3.04216, "grad_norm": 20.94187977416302, "learning_rate": 8.24674766063044e-06, "loss": 1.5469, "step": 29680 }, { "epoch": 3.04226, "grad_norm": 22.9895815260381, "learning_rate": 8.245515969699197e-06, "loss": 1.6885, "step": 29690 }, { "epoch": 3.04236, "grad_norm": 24.87943524024822, "learning_rate": 8.24428393832839e-06, "loss": 1.7328, "step": 29700 }, { "epoch": 3.04246, "grad_norm": 21.40631630490808, "learning_rate": 8.243051566647256e-06, "loss": 1.6793, "step": 29710 }, { "epoch": 3.04256, "grad_norm": 22.531845159883023, "learning_rate": 8.241818854785064e-06, "loss": 1.7271, "step": 29720 }, { "epoch": 3.04266, "grad_norm": 24.642678771729354, "learning_rate": 8.240585802871119e-06, "loss": 1.7012, "step": 29730 }, { "epoch": 3.04276, "grad_norm": 20.055879471419967, "learning_rate": 8.239352411034765e-06, "loss": 1.6478, "step": 29740 }, { "epoch": 3.04286, "grad_norm": 20.086607753892245, "learning_rate": 8.238118679405375e-06, "loss": 1.7071, "step": 29750 }, { "epoch": 3.04296, "grad_norm": 22.507498862747074, "learning_rate": 8.236884608112365e-06, "loss": 1.6925, "step": 29760 }, { "epoch": 3.04306, "grad_norm": 20.692477944273268, "learning_rate": 8.235650197285183e-06, "loss": 1.6014, "step": 29770 }, { "epoch": 3.04316, "grad_norm": 21.581419137652, "learning_rate": 8.234415447053313e-06, "loss": 1.6293, "step": 29780 }, { "epoch": 3.04326, "grad_norm": 23.053394153040486, "learning_rate": 8.233180357546273e-06, "loss": 1.5895, "step": 29790 }, { "epoch": 3.04336, "grad_norm": 22.786899480155643, "learning_rate": 8.23194492889362e-06, "loss": 1.5995, "step": 29800 }, { "epoch": 3.04346, "grad_norm": 21.28922910817076, "learning_rate": 8.230709161224943e-06, "loss": 1.6776, "step": 29810 }, { "epoch": 3.04356, "grad_norm": 20.450033252150803, "learning_rate": 8.22947305466987e-06, "loss": 1.6676, "step": 29820 }, { "epoch": 3.04366, "grad_norm": 21.458519614304024, "learning_rate": 8.22823660935806e-06, "loss": 1.584, "step": 29830 }, { "epoch": 3.04376, "grad_norm": 20.981708347635145, "learning_rate": 8.226999825419216e-06, "loss": 1.6391, "step": 29840 }, { "epoch": 3.04386, "grad_norm": 20.83156011691444, "learning_rate": 8.225762702983064e-06, "loss": 1.6313, "step": 29850 }, { "epoch": 3.04396, "grad_norm": 22.828149735959318, "learning_rate": 8.224525242179378e-06, "loss": 1.6148, "step": 29860 }, { "epoch": 3.04406, "grad_norm": 22.26463474456762, "learning_rate": 8.223287443137957e-06, "loss": 1.5981, "step": 29870 }, { "epoch": 3.04416, "grad_norm": 23.112758978137187, "learning_rate": 8.222049305988644e-06, "loss": 1.6022, "step": 29880 }, { "epoch": 3.04426, "grad_norm": 20.255260840338778, "learning_rate": 8.220810830861315e-06, "loss": 1.6215, "step": 29890 }, { "epoch": 3.04436, "grad_norm": 22.85236041827981, "learning_rate": 8.219572017885875e-06, "loss": 1.5773, "step": 29900 }, { "epoch": 3.04446, "grad_norm": 22.582040796182092, "learning_rate": 8.218332867192275e-06, "loss": 1.6561, "step": 29910 }, { "epoch": 3.04456, "grad_norm": 20.149300755636993, "learning_rate": 8.217093378910493e-06, "loss": 1.5948, "step": 29920 }, { "epoch": 3.04466, "grad_norm": 18.723753458945374, "learning_rate": 8.215853553170547e-06, "loss": 1.558, "step": 29930 }, { "epoch": 3.04476, "grad_norm": 21.51600838556977, "learning_rate": 8.214613390102487e-06, "loss": 1.5298, "step": 29940 }, { "epoch": 3.04486, "grad_norm": 22.55778298479445, "learning_rate": 8.213372889836404e-06, "loss": 1.5835, "step": 29950 }, { "epoch": 3.04496, "grad_norm": 27.02327425023058, "learning_rate": 8.212132052502417e-06, "loss": 1.6281, "step": 29960 }, { "epoch": 3.04506, "grad_norm": 20.80977420817297, "learning_rate": 8.210890878230686e-06, "loss": 1.6155, "step": 29970 }, { "epoch": 3.04516, "grad_norm": 22.660807408370342, "learning_rate": 8.209649367151404e-06, "loss": 1.6046, "step": 29980 }, { "epoch": 3.04526, "grad_norm": 22.60452786370365, "learning_rate": 8.208407519394801e-06, "loss": 1.6382, "step": 29990 }, { "epoch": 3.04536, "grad_norm": 22.91712810570764, "learning_rate": 8.207165335091139e-06, "loss": 1.5681, "step": 30000 }, { "epoch": 3.04546, "grad_norm": 20.545514310082865, "learning_rate": 8.205922814370717e-06, "loss": 1.6379, "step": 30010 }, { "epoch": 3.04556, "grad_norm": 20.698823449671817, "learning_rate": 8.204679957363873e-06, "loss": 1.6379, "step": 30020 }, { "epoch": 3.04566, "grad_norm": 18.532358954698616, "learning_rate": 8.203436764200974e-06, "loss": 1.6037, "step": 30030 }, { "epoch": 3.04576, "grad_norm": 21.547354311397605, "learning_rate": 8.202193235012425e-06, "loss": 1.5739, "step": 30040 }, { "epoch": 3.04586, "grad_norm": 21.634387205180435, "learning_rate": 8.20094936992867e-06, "loss": 1.6839, "step": 30050 }, { "epoch": 3.04596, "grad_norm": 24.238079182854733, "learning_rate": 8.199705169080181e-06, "loss": 1.6531, "step": 30060 }, { "epoch": 3.04606, "grad_norm": 20.98741956998053, "learning_rate": 8.19846063259747e-06, "loss": 1.5894, "step": 30070 }, { "epoch": 3.04616, "grad_norm": 18.756107231686308, "learning_rate": 8.197215760611083e-06, "loss": 1.5758, "step": 30080 }, { "epoch": 3.04626, "grad_norm": 21.960490727403414, "learning_rate": 8.195970553251601e-06, "loss": 1.5775, "step": 30090 }, { "epoch": 3.04636, "grad_norm": 21.524460191319708, "learning_rate": 8.194725010649642e-06, "loss": 1.6151, "step": 30100 }, { "epoch": 3.04646, "grad_norm": 21.59468809580747, "learning_rate": 8.193479132935856e-06, "loss": 1.6401, "step": 30110 }, { "epoch": 3.04656, "grad_norm": 20.35565438861903, "learning_rate": 8.19223292024093e-06, "loss": 1.5503, "step": 30120 }, { "epoch": 3.04666, "grad_norm": 22.6622594936616, "learning_rate": 8.190986372695586e-06, "loss": 1.5941, "step": 30130 }, { "epoch": 3.04676, "grad_norm": 20.89674226617324, "learning_rate": 8.189739490430581e-06, "loss": 1.5991, "step": 30140 }, { "epoch": 3.04686, "grad_norm": 21.834224836957603, "learning_rate": 8.18849227357671e-06, "loss": 1.5642, "step": 30150 }, { "epoch": 3.04696, "grad_norm": 21.855540751101096, "learning_rate": 8.187244722264795e-06, "loss": 1.5827, "step": 30160 }, { "epoch": 3.04706, "grad_norm": 22.128458427569452, "learning_rate": 8.185996836625702e-06, "loss": 1.5722, "step": 30170 }, { "epoch": 3.04716, "grad_norm": 22.21136445254874, "learning_rate": 8.184748616790328e-06, "loss": 1.6593, "step": 30180 }, { "epoch": 3.04726, "grad_norm": 22.260391413258983, "learning_rate": 8.183500062889605e-06, "loss": 1.6562, "step": 30190 }, { "epoch": 3.04736, "grad_norm": 20.25292303932025, "learning_rate": 8.182251175054499e-06, "loss": 1.563, "step": 30200 }, { "epoch": 3.04746, "grad_norm": 21.389258875280362, "learning_rate": 8.181001953416018e-06, "loss": 1.5898, "step": 30210 }, { "epoch": 3.04756, "grad_norm": 21.552471804191367, "learning_rate": 8.17975239810519e-06, "loss": 1.6474, "step": 30220 }, { "epoch": 3.04766, "grad_norm": 22.432480617259955, "learning_rate": 8.178502509253096e-06, "loss": 1.6323, "step": 30230 }, { "epoch": 3.04776, "grad_norm": 20.704594317906775, "learning_rate": 8.177252286990842e-06, "loss": 1.6162, "step": 30240 }, { "epoch": 3.04786, "grad_norm": 19.55215054380736, "learning_rate": 8.176001731449566e-06, "loss": 1.6027, "step": 30250 }, { "epoch": 3.04796, "grad_norm": 22.022138951915696, "learning_rate": 8.174750842760452e-06, "loss": 1.6364, "step": 30260 }, { "epoch": 3.04806, "grad_norm": 19.746689800366404, "learning_rate": 8.173499621054707e-06, "loss": 1.6005, "step": 30270 }, { "epoch": 3.04816, "grad_norm": 21.64249354165693, "learning_rate": 8.172248066463581e-06, "loss": 1.5892, "step": 30280 }, { "epoch": 3.04826, "grad_norm": 22.282142033081435, "learning_rate": 8.170996179118355e-06, "loss": 1.6406, "step": 30290 }, { "epoch": 3.04836, "grad_norm": 22.85591298771189, "learning_rate": 8.169743959150346e-06, "loss": 1.5545, "step": 30300 }, { "epoch": 3.04846, "grad_norm": 20.448949378827308, "learning_rate": 8.16849140669091e-06, "loss": 1.5327, "step": 30310 }, { "epoch": 3.04856, "grad_norm": 21.839017056671143, "learning_rate": 8.167238521871428e-06, "loss": 1.5393, "step": 30320 }, { "epoch": 3.04866, "grad_norm": 22.700856688708754, "learning_rate": 8.165985304823327e-06, "loss": 1.6154, "step": 30330 }, { "epoch": 3.04876, "grad_norm": 23.19693291710044, "learning_rate": 8.16473175567806e-06, "loss": 1.6109, "step": 30340 }, { "epoch": 3.04886, "grad_norm": 21.748503478927205, "learning_rate": 8.163477874567121e-06, "loss": 1.6335, "step": 30350 }, { "epoch": 3.04896, "grad_norm": 21.886725779112403, "learning_rate": 8.162223661622034e-06, "loss": 1.653, "step": 30360 }, { "epoch": 3.04906, "grad_norm": 22.889187726304392, "learning_rate": 8.16096911697436e-06, "loss": 1.6164, "step": 30370 }, { "epoch": 3.04916, "grad_norm": 21.551249447106247, "learning_rate": 8.159714240755697e-06, "loss": 1.6198, "step": 30380 }, { "epoch": 3.04926, "grad_norm": 21.614165167370906, "learning_rate": 8.158459033097676e-06, "loss": 1.608, "step": 30390 }, { "epoch": 3.04936, "grad_norm": 21.888928649146052, "learning_rate": 8.15720349413196e-06, "loss": 1.7041, "step": 30400 }, { "epoch": 3.04946, "grad_norm": 21.589888885900137, "learning_rate": 8.155947623990252e-06, "loss": 1.6149, "step": 30410 }, { "epoch": 3.04956, "grad_norm": 21.4099888615795, "learning_rate": 8.154691422804284e-06, "loss": 1.5661, "step": 30420 }, { "epoch": 3.04966, "grad_norm": 21.924924241645265, "learning_rate": 8.153434890705827e-06, "loss": 1.6292, "step": 30430 }, { "epoch": 3.04976, "grad_norm": 20.80891819204486, "learning_rate": 8.152178027826687e-06, "loss": 1.5597, "step": 30440 }, { "epoch": 3.04986, "grad_norm": 21.191973504579515, "learning_rate": 8.150920834298701e-06, "loss": 1.551, "step": 30450 }, { "epoch": 3.04996, "grad_norm": 21.64130902141465, "learning_rate": 8.149663310253743e-06, "loss": 1.5963, "step": 30460 }, { "epoch": 3.05006, "grad_norm": 22.559666031829234, "learning_rate": 8.148405455823723e-06, "loss": 1.5453, "step": 30470 }, { "epoch": 3.05016, "grad_norm": 21.785783891794594, "learning_rate": 8.147147271140582e-06, "loss": 1.5522, "step": 30480 }, { "epoch": 3.05026, "grad_norm": 22.938382851441347, "learning_rate": 8.145888756336298e-06, "loss": 1.6191, "step": 30490 }, { "epoch": 3.05036, "grad_norm": 21.794951356949582, "learning_rate": 8.144629911542883e-06, "loss": 1.5778, "step": 30500 }, { "epoch": 3.05046, "grad_norm": 19.94285788589658, "learning_rate": 8.143370736892386e-06, "loss": 1.5717, "step": 30510 }, { "epoch": 3.05056, "grad_norm": 21.88020880555108, "learning_rate": 8.142111232516888e-06, "loss": 1.5846, "step": 30520 }, { "epoch": 3.05066, "grad_norm": 22.792578128772032, "learning_rate": 8.140851398548505e-06, "loss": 1.5975, "step": 30530 }, { "epoch": 3.05076, "grad_norm": 19.901507937664167, "learning_rate": 8.139591235119386e-06, "loss": 1.7256, "step": 30540 }, { "epoch": 3.05086, "grad_norm": 20.603193734445256, "learning_rate": 8.13833074236172e-06, "loss": 1.6145, "step": 30550 }, { "epoch": 3.05096, "grad_norm": 19.878580600884984, "learning_rate": 8.137069920407722e-06, "loss": 1.5502, "step": 30560 }, { "epoch": 3.05106, "grad_norm": 23.877338152699824, "learning_rate": 8.135808769389651e-06, "loss": 1.7106, "step": 30570 }, { "epoch": 3.05116, "grad_norm": 22.323353983608552, "learning_rate": 8.134547289439794e-06, "loss": 1.6037, "step": 30580 }, { "epoch": 3.05126, "grad_norm": 23.7427442047656, "learning_rate": 8.133285480690472e-06, "loss": 1.6946, "step": 30590 }, { "epoch": 3.05136, "grad_norm": 22.13542957446964, "learning_rate": 8.132023343274046e-06, "loss": 1.5893, "step": 30600 }, { "epoch": 3.05146, "grad_norm": 22.052876295110938, "learning_rate": 8.130760877322909e-06, "loss": 1.7437, "step": 30610 }, { "epoch": 3.05156, "grad_norm": 20.281860422644975, "learning_rate": 8.129498082969485e-06, "loss": 1.6617, "step": 30620 }, { "epoch": 3.05166, "grad_norm": 21.786970092236956, "learning_rate": 8.128234960346238e-06, "loss": 1.6278, "step": 30630 }, { "epoch": 3.05176, "grad_norm": 20.934222885540606, "learning_rate": 8.126971509585659e-06, "loss": 1.5479, "step": 30640 }, { "epoch": 3.05186, "grad_norm": 20.74428088375456, "learning_rate": 8.125707730820284e-06, "loss": 1.614, "step": 30650 }, { "epoch": 3.05196, "grad_norm": 22.0160864158008, "learning_rate": 8.124443624182674e-06, "loss": 1.5366, "step": 30660 }, { "epoch": 3.05206, "grad_norm": 24.155042978224532, "learning_rate": 8.12317918980543e-06, "loss": 1.6525, "step": 30670 }, { "epoch": 3.05216, "grad_norm": 22.32026328593748, "learning_rate": 8.121914427821182e-06, "loss": 1.6038, "step": 30680 }, { "epoch": 3.05226, "grad_norm": 20.65949582500421, "learning_rate": 8.1206493383626e-06, "loss": 1.6093, "step": 30690 }, { "epoch": 3.05236, "grad_norm": 20.12678828134225, "learning_rate": 8.119383921562388e-06, "loss": 1.6383, "step": 30700 }, { "epoch": 3.05246, "grad_norm": 24.63512531412549, "learning_rate": 8.118118177553278e-06, "loss": 1.631, "step": 30710 }, { "epoch": 3.05256, "grad_norm": 22.13260226590077, "learning_rate": 8.116852106468044e-06, "loss": 1.6256, "step": 30720 }, { "epoch": 3.05266, "grad_norm": 22.585358205392676, "learning_rate": 8.11558570843949e-06, "loss": 1.6462, "step": 30730 }, { "epoch": 3.05276, "grad_norm": 20.40470897479948, "learning_rate": 8.114318983600454e-06, "loss": 1.6492, "step": 30740 }, { "epoch": 3.05286, "grad_norm": 20.251181613256616, "learning_rate": 8.113051932083811e-06, "loss": 1.6384, "step": 30750 }, { "epoch": 3.05296, "grad_norm": 20.283788060513842, "learning_rate": 8.111784554022469e-06, "loss": 1.6628, "step": 30760 }, { "epoch": 3.05306, "grad_norm": 19.597509270503107, "learning_rate": 8.110516849549369e-06, "loss": 1.5826, "step": 30770 }, { "epoch": 3.05316, "grad_norm": 20.183281402005154, "learning_rate": 8.109248818797487e-06, "loss": 1.5457, "step": 30780 }, { "epoch": 3.05326, "grad_norm": 20.636641948354246, "learning_rate": 8.107980461899836e-06, "loss": 1.6186, "step": 30790 }, { "epoch": 3.05336, "grad_norm": 22.41595863586398, "learning_rate": 8.106711778989457e-06, "loss": 1.6054, "step": 30800 }, { "epoch": 3.05346, "grad_norm": 19.96374748367921, "learning_rate": 8.105442770199434e-06, "loss": 1.6073, "step": 30810 }, { "epoch": 3.05356, "grad_norm": 24.280313461470556, "learning_rate": 8.104173435662876e-06, "loss": 1.6511, "step": 30820 }, { "epoch": 3.05366, "grad_norm": 21.94959293573951, "learning_rate": 8.10290377551293e-06, "loss": 1.639, "step": 30830 }, { "epoch": 3.05376, "grad_norm": 24.544761269363367, "learning_rate": 8.101633789882781e-06, "loss": 1.6315, "step": 30840 }, { "epoch": 3.05386, "grad_norm": 23.138112512268048, "learning_rate": 8.100363478905642e-06, "loss": 1.6478, "step": 30850 }, { "epoch": 3.05396, "grad_norm": 21.92505392717304, "learning_rate": 8.099092842714761e-06, "loss": 1.74, "step": 30860 }, { "epoch": 3.0540599999999998, "grad_norm": 20.666741716601, "learning_rate": 8.097821881443428e-06, "loss": 1.6582, "step": 30870 }, { "epoch": 3.05416, "grad_norm": 20.907951739348466, "learning_rate": 8.096550595224954e-06, "loss": 1.6144, "step": 30880 }, { "epoch": 3.05426, "grad_norm": 21.869582906967924, "learning_rate": 8.095278984192696e-06, "loss": 1.6456, "step": 30890 }, { "epoch": 3.05436, "grad_norm": 21.858520095983696, "learning_rate": 8.094007048480037e-06, "loss": 1.64, "step": 30900 }, { "epoch": 3.05446, "grad_norm": 21.585553717894534, "learning_rate": 8.0927347882204e-06, "loss": 1.5502, "step": 30910 }, { "epoch": 3.05456, "grad_norm": 20.944841766839616, "learning_rate": 8.091462203547236e-06, "loss": 1.6282, "step": 30920 }, { "epoch": 3.05466, "grad_norm": 20.78146073035544, "learning_rate": 8.090189294594035e-06, "loss": 1.612, "step": 30930 }, { "epoch": 3.05476, "grad_norm": 21.062389679122774, "learning_rate": 8.08891606149432e-06, "loss": 1.6579, "step": 30940 }, { "epoch": 3.05486, "grad_norm": 20.356246299234492, "learning_rate": 8.087642504381645e-06, "loss": 1.6651, "step": 30950 }, { "epoch": 3.05496, "grad_norm": 22.78204323496319, "learning_rate": 8.086368623389602e-06, "loss": 1.6256, "step": 30960 }, { "epoch": 3.05506, "grad_norm": 22.139743986254253, "learning_rate": 8.085094418651816e-06, "loss": 1.5552, "step": 30970 }, { "epoch": 3.05516, "grad_norm": 22.74631094243213, "learning_rate": 8.08381989030194e-06, "loss": 1.5981, "step": 30980 }, { "epoch": 3.05526, "grad_norm": 21.345853137806166, "learning_rate": 8.082545038473674e-06, "loss": 1.658, "step": 30990 }, { "epoch": 3.05536, "grad_norm": 21.61441007648645, "learning_rate": 8.081269863300739e-06, "loss": 1.6769, "step": 31000 }, { "epoch": 3.05546, "grad_norm": 21.263994537736707, "learning_rate": 8.079994364916894e-06, "loss": 1.6143, "step": 31010 }, { "epoch": 3.05556, "grad_norm": 22.04645674415649, "learning_rate": 8.078718543455936e-06, "loss": 1.6602, "step": 31020 }, { "epoch": 3.05566, "grad_norm": 20.574557978409644, "learning_rate": 8.077442399051692e-06, "loss": 1.6203, "step": 31030 }, { "epoch": 3.05576, "grad_norm": 20.2473393024003, "learning_rate": 8.076165931838023e-06, "loss": 1.6231, "step": 31040 }, { "epoch": 3.05586, "grad_norm": 20.483794624914633, "learning_rate": 8.074889141948825e-06, "loss": 1.6165, "step": 31050 }, { "epoch": 3.05596, "grad_norm": 22.002395009881262, "learning_rate": 8.073612029518027e-06, "loss": 1.641, "step": 31060 }, { "epoch": 3.05606, "grad_norm": 23.426613241202535, "learning_rate": 8.072334594679591e-06, "loss": 1.5642, "step": 31070 }, { "epoch": 3.05616, "grad_norm": 23.298303878079317, "learning_rate": 8.071056837567518e-06, "loss": 1.6246, "step": 31080 }, { "epoch": 3.05626, "grad_norm": 20.7478936022747, "learning_rate": 8.069778758315831e-06, "loss": 1.5875, "step": 31090 }, { "epoch": 3.05636, "grad_norm": 22.017525580298127, "learning_rate": 8.068500357058605e-06, "loss": 1.6796, "step": 31100 }, { "epoch": 3.05646, "grad_norm": 21.92263140230356, "learning_rate": 8.067221633929929e-06, "loss": 1.5884, "step": 31110 }, { "epoch": 3.05656, "grad_norm": 22.616191762180915, "learning_rate": 8.06594258906394e-06, "loss": 1.5866, "step": 31120 }, { "epoch": 3.05666, "grad_norm": 24.848148500734816, "learning_rate": 8.064663222594803e-06, "loss": 1.6482, "step": 31130 }, { "epoch": 3.05676, "grad_norm": 24.467486024045023, "learning_rate": 8.063383534656719e-06, "loss": 1.6282, "step": 31140 }, { "epoch": 3.05686, "grad_norm": 21.023116011070016, "learning_rate": 8.062103525383919e-06, "loss": 1.6662, "step": 31150 }, { "epoch": 3.05696, "grad_norm": 19.660086372579116, "learning_rate": 8.060823194910669e-06, "loss": 1.598, "step": 31160 }, { "epoch": 3.05706, "grad_norm": 26.004871641655306, "learning_rate": 8.059542543371274e-06, "loss": 1.6175, "step": 31170 }, { "epoch": 3.05716, "grad_norm": 21.54316328445214, "learning_rate": 8.058261570900065e-06, "loss": 1.5963, "step": 31180 }, { "epoch": 3.05726, "grad_norm": 22.131054375070043, "learning_rate": 8.05698027763141e-06, "loss": 1.6387, "step": 31190 }, { "epoch": 3.05736, "grad_norm": 22.267607636449917, "learning_rate": 8.055698663699714e-06, "loss": 1.6707, "step": 31200 }, { "epoch": 3.05746, "grad_norm": 21.863662358024694, "learning_rate": 8.054416729239407e-06, "loss": 1.638, "step": 31210 }, { "epoch": 3.05756, "grad_norm": 21.533643208475915, "learning_rate": 8.053134474384965e-06, "loss": 1.5787, "step": 31220 }, { "epoch": 3.05766, "grad_norm": 23.57193345125412, "learning_rate": 8.051851899270885e-06, "loss": 1.6405, "step": 31230 }, { "epoch": 3.05776, "grad_norm": 22.960757405434645, "learning_rate": 8.050569004031703e-06, "loss": 1.677, "step": 31240 }, { "epoch": 3.05786, "grad_norm": 26.049747838427912, "learning_rate": 8.049285788801992e-06, "loss": 1.656, "step": 31250 }, { "epoch": 3.05796, "grad_norm": 21.461325364968456, "learning_rate": 8.048002253716353e-06, "loss": 1.6957, "step": 31260 }, { "epoch": 3.0580600000000002, "grad_norm": 21.54440298902983, "learning_rate": 8.046718398909426e-06, "loss": 1.7232, "step": 31270 }, { "epoch": 3.05816, "grad_norm": 23.711205676631234, "learning_rate": 8.045434224515878e-06, "loss": 1.6313, "step": 31280 }, { "epoch": 3.05826, "grad_norm": 21.424877167406702, "learning_rate": 8.044149730670414e-06, "loss": 1.6431, "step": 31290 }, { "epoch": 3.05836, "grad_norm": 21.194709493835955, "learning_rate": 8.042864917507773e-06, "loss": 1.6491, "step": 31300 }, { "epoch": 3.05846, "grad_norm": 20.128935171467386, "learning_rate": 8.041579785162724e-06, "loss": 1.5888, "step": 31310 }, { "epoch": 3.05856, "grad_norm": 22.036095041953207, "learning_rate": 8.040294333770069e-06, "loss": 1.601, "step": 31320 }, { "epoch": 3.05866, "grad_norm": 23.542432288686854, "learning_rate": 8.039008563464652e-06, "loss": 1.6199, "step": 31330 }, { "epoch": 3.05876, "grad_norm": 23.086323651156935, "learning_rate": 8.03772247438134e-06, "loss": 1.6372, "step": 31340 }, { "epoch": 3.05886, "grad_norm": 22.387689849551045, "learning_rate": 8.03643606665504e-06, "loss": 1.6913, "step": 31350 }, { "epoch": 3.05896, "grad_norm": 25.183999480298, "learning_rate": 8.035149340420688e-06, "loss": 1.7081, "step": 31360 }, { "epoch": 3.05906, "grad_norm": 19.158118541318053, "learning_rate": 8.033862295813255e-06, "loss": 1.5373, "step": 31370 }, { "epoch": 3.05916, "grad_norm": 23.23065462576927, "learning_rate": 8.032574932967752e-06, "loss": 1.6605, "step": 31380 }, { "epoch": 3.05926, "grad_norm": 26.43030311744212, "learning_rate": 8.03128725201921e-06, "loss": 1.6681, "step": 31390 }, { "epoch": 3.05936, "grad_norm": 19.547150337582952, "learning_rate": 8.029999253102704e-06, "loss": 1.5981, "step": 31400 }, { "epoch": 3.05946, "grad_norm": 24.212409818788213, "learning_rate": 8.028710936353341e-06, "loss": 1.6622, "step": 31410 }, { "epoch": 3.05956, "grad_norm": 22.646416247186703, "learning_rate": 8.027422301906257e-06, "loss": 1.5857, "step": 31420 }, { "epoch": 3.05966, "grad_norm": 19.2682205767972, "learning_rate": 8.026133349896623e-06, "loss": 1.6565, "step": 31430 }, { "epoch": 3.05976, "grad_norm": 21.324612662549484, "learning_rate": 8.024844080459646e-06, "loss": 1.6627, "step": 31440 }, { "epoch": 3.05986, "grad_norm": 18.659512778583508, "learning_rate": 8.023554493730565e-06, "loss": 1.6636, "step": 31450 }, { "epoch": 3.05996, "grad_norm": 20.28441742936107, "learning_rate": 8.022264589844648e-06, "loss": 1.6027, "step": 31460 }, { "epoch": 3.06006, "grad_norm": 22.32312449776683, "learning_rate": 8.020974368937201e-06, "loss": 1.6128, "step": 31470 }, { "epoch": 3.06016, "grad_norm": 23.791398390802073, "learning_rate": 8.019683831143567e-06, "loss": 1.6018, "step": 31480 }, { "epoch": 3.06026, "grad_norm": 22.09479204662431, "learning_rate": 8.018392976599112e-06, "loss": 1.6782, "step": 31490 }, { "epoch": 3.06036, "grad_norm": 25.322955936191644, "learning_rate": 8.017101805439245e-06, "loss": 1.6693, "step": 31500 }, { "epoch": 3.06046, "grad_norm": 21.54694622992302, "learning_rate": 8.015810317799399e-06, "loss": 1.5688, "step": 31510 }, { "epoch": 3.06056, "grad_norm": 22.57474646890694, "learning_rate": 8.014518513815048e-06, "loss": 1.6102, "step": 31520 }, { "epoch": 3.06066, "grad_norm": 20.726398423953487, "learning_rate": 8.013226393621695e-06, "loss": 1.6726, "step": 31530 }, { "epoch": 3.06076, "grad_norm": 22.764709205688206, "learning_rate": 8.011933957354881e-06, "loss": 1.6116, "step": 31540 }, { "epoch": 3.06086, "grad_norm": 22.697643359629737, "learning_rate": 8.01064120515017e-06, "loss": 1.623, "step": 31550 }, { "epoch": 3.06096, "grad_norm": 21.92036839002209, "learning_rate": 8.009348137143172e-06, "loss": 1.5741, "step": 31560 }, { "epoch": 3.06106, "grad_norm": 22.758293260930266, "learning_rate": 8.008054753469518e-06, "loss": 1.5827, "step": 31570 }, { "epoch": 3.06116, "grad_norm": 22.407763856599058, "learning_rate": 8.006761054264884e-06, "loss": 1.6384, "step": 31580 }, { "epoch": 3.06126, "grad_norm": 20.519477232439844, "learning_rate": 8.00546703966497e-06, "loss": 1.5924, "step": 31590 }, { "epoch": 3.06136, "grad_norm": 26.3536420574151, "learning_rate": 8.004172709805513e-06, "loss": 1.5928, "step": 31600 }, { "epoch": 3.06146, "grad_norm": 23.298402222900467, "learning_rate": 8.002878064822279e-06, "loss": 1.6931, "step": 31610 }, { "epoch": 3.06156, "grad_norm": 22.064938240865565, "learning_rate": 8.001583104851074e-06, "loss": 1.6843, "step": 31620 }, { "epoch": 3.06166, "grad_norm": 21.91325474147412, "learning_rate": 8.000287830027734e-06, "loss": 1.5862, "step": 31630 }, { "epoch": 3.06176, "grad_norm": 22.58999264175039, "learning_rate": 7.998992240488125e-06, "loss": 1.6921, "step": 31640 }, { "epoch": 3.06186, "grad_norm": 22.41703225162181, "learning_rate": 7.997696336368146e-06, "loss": 1.6104, "step": 31650 }, { "epoch": 3.06196, "grad_norm": 22.487948076478133, "learning_rate": 7.996400117803735e-06, "loss": 1.6394, "step": 31660 }, { "epoch": 3.06206, "grad_norm": 21.5684666056958, "learning_rate": 7.995103584930859e-06, "loss": 1.64, "step": 31670 }, { "epoch": 3.06216, "grad_norm": 23.07649501760985, "learning_rate": 7.99380673788552e-06, "loss": 1.6153, "step": 31680 }, { "epoch": 3.06226, "grad_norm": 21.367827829923925, "learning_rate": 7.992509576803745e-06, "loss": 1.5986, "step": 31690 }, { "epoch": 3.06236, "grad_norm": 21.836023933406768, "learning_rate": 7.991212101821604e-06, "loss": 1.6503, "step": 31700 }, { "epoch": 3.06246, "grad_norm": 21.444738928528984, "learning_rate": 7.989914313075198e-06, "loss": 1.6896, "step": 31710 }, { "epoch": 3.06256, "grad_norm": 21.63149659267295, "learning_rate": 7.988616210700657e-06, "loss": 1.6072, "step": 31720 }, { "epoch": 3.06266, "grad_norm": 21.480622041327976, "learning_rate": 7.987317794834146e-06, "loss": 1.6826, "step": 31730 }, { "epoch": 3.06276, "grad_norm": 20.873578964010434, "learning_rate": 7.986019065611863e-06, "loss": 1.6653, "step": 31740 }, { "epoch": 3.06286, "grad_norm": 22.31247653338537, "learning_rate": 7.984720023170039e-06, "loss": 1.6113, "step": 31750 }, { "epoch": 3.06296, "grad_norm": 20.68281802949345, "learning_rate": 7.983420667644937e-06, "loss": 1.5733, "step": 31760 }, { "epoch": 3.06306, "grad_norm": 21.705240341545682, "learning_rate": 7.98212099917285e-06, "loss": 1.5712, "step": 31770 }, { "epoch": 3.06316, "grad_norm": 22.740980651042193, "learning_rate": 7.980821017890114e-06, "loss": 1.5415, "step": 31780 }, { "epoch": 3.06326, "grad_norm": 22.248152373488164, "learning_rate": 7.979520723933087e-06, "loss": 1.661, "step": 31790 }, { "epoch": 3.06336, "grad_norm": 21.827555457034382, "learning_rate": 7.978220117438166e-06, "loss": 1.5734, "step": 31800 }, { "epoch": 3.06346, "grad_norm": 23.701632109577336, "learning_rate": 7.976919198541775e-06, "loss": 1.6098, "step": 31810 }, { "epoch": 3.06356, "grad_norm": 22.521199652540844, "learning_rate": 7.975617967380378e-06, "loss": 1.548, "step": 31820 }, { "epoch": 3.06366, "grad_norm": 21.851173424539795, "learning_rate": 7.974316424090465e-06, "loss": 1.6397, "step": 31830 }, { "epoch": 3.06376, "grad_norm": 21.252395116400468, "learning_rate": 7.973014568808563e-06, "loss": 1.6946, "step": 31840 }, { "epoch": 3.06386, "grad_norm": 20.717425055834294, "learning_rate": 7.971712401671233e-06, "loss": 1.6268, "step": 31850 }, { "epoch": 3.06396, "grad_norm": 23.330751904294726, "learning_rate": 7.970409922815063e-06, "loss": 1.638, "step": 31860 }, { "epoch": 3.06406, "grad_norm": 20.792657709436668, "learning_rate": 7.969107132376678e-06, "loss": 1.5954, "step": 31870 }, { "epoch": 3.06416, "grad_norm": 20.622035852577973, "learning_rate": 7.967804030492735e-06, "loss": 1.6114, "step": 31880 }, { "epoch": 3.06426, "grad_norm": 22.75199457701128, "learning_rate": 7.966500617299923e-06, "loss": 1.6042, "step": 31890 }, { "epoch": 3.06436, "grad_norm": 22.0434761770981, "learning_rate": 7.965196892934964e-06, "loss": 1.5965, "step": 31900 }, { "epoch": 3.06446, "grad_norm": 19.99899333664099, "learning_rate": 7.963892857534614e-06, "loss": 1.5869, "step": 31910 }, { "epoch": 3.06456, "grad_norm": 20.701577251006277, "learning_rate": 7.962588511235657e-06, "loss": 1.6007, "step": 31920 }, { "epoch": 3.06466, "grad_norm": 22.410406851699364, "learning_rate": 7.961283854174917e-06, "loss": 1.6272, "step": 31930 }, { "epoch": 3.06476, "grad_norm": 22.55421268571699, "learning_rate": 7.959978886489244e-06, "loss": 1.608, "step": 31940 }, { "epoch": 3.06486, "grad_norm": 25.24452403072088, "learning_rate": 7.958673608315524e-06, "loss": 1.6255, "step": 31950 }, { "epoch": 3.06496, "grad_norm": 22.750721560209573, "learning_rate": 7.957368019790673e-06, "loss": 1.6273, "step": 31960 }, { "epoch": 3.06506, "grad_norm": 21.026242644828873, "learning_rate": 7.956062121051643e-06, "loss": 1.6287, "step": 31970 }, { "epoch": 3.06516, "grad_norm": 23.47596897457864, "learning_rate": 7.954755912235415e-06, "loss": 1.594, "step": 31980 }, { "epoch": 3.06526, "grad_norm": 26.71777885379747, "learning_rate": 7.953449393479006e-06, "loss": 1.6072, "step": 31990 }, { "epoch": 3.06536, "grad_norm": 23.020024121417773, "learning_rate": 7.952142564919462e-06, "loss": 1.6341, "step": 32000 }, { "epoch": 3.06546, "grad_norm": 21.72131934376594, "learning_rate": 7.950835426693866e-06, "loss": 1.5979, "step": 32010 }, { "epoch": 3.06556, "grad_norm": 21.248601532099972, "learning_rate": 7.94952797893933e-06, "loss": 1.6521, "step": 32020 }, { "epoch": 3.06566, "grad_norm": 21.91411083458503, "learning_rate": 7.948220221792995e-06, "loss": 1.5711, "step": 32030 }, { "epoch": 3.06576, "grad_norm": 22.944874970318814, "learning_rate": 7.946912155392044e-06, "loss": 1.6196, "step": 32040 }, { "epoch": 3.06586, "grad_norm": 21.674431601621645, "learning_rate": 7.945603779873682e-06, "loss": 1.5937, "step": 32050 }, { "epoch": 3.06596, "grad_norm": 22.268464453659842, "learning_rate": 7.944295095375159e-06, "loss": 1.5949, "step": 32060 }, { "epoch": 3.06606, "grad_norm": 21.687796736952553, "learning_rate": 7.942986102033743e-06, "loss": 1.6322, "step": 32070 }, { "epoch": 3.06616, "grad_norm": 23.059781855273613, "learning_rate": 7.941676799986743e-06, "loss": 1.6145, "step": 32080 }, { "epoch": 3.06626, "grad_norm": 21.723177723270968, "learning_rate": 7.9403671893715e-06, "loss": 1.6309, "step": 32090 }, { "epoch": 3.06636, "grad_norm": 21.398431118039603, "learning_rate": 7.939057270325385e-06, "loss": 1.6588, "step": 32100 }, { "epoch": 3.06646, "grad_norm": 23.818481326422937, "learning_rate": 7.937747042985804e-06, "loss": 1.6662, "step": 32110 }, { "epoch": 3.06656, "grad_norm": 23.91773961950298, "learning_rate": 7.936436507490193e-06, "loss": 1.6339, "step": 32120 }, { "epoch": 3.06666, "grad_norm": 22.11038869463377, "learning_rate": 7.93512566397602e-06, "loss": 1.6209, "step": 32130 }, { "epoch": 3.06676, "grad_norm": 23.40014762753317, "learning_rate": 7.933814512580787e-06, "loss": 1.5883, "step": 32140 }, { "epoch": 3.06686, "grad_norm": 22.34481194649828, "learning_rate": 7.932503053442028e-06, "loss": 1.6079, "step": 32150 }, { "epoch": 3.06696, "grad_norm": 22.14708865398858, "learning_rate": 7.93119128669731e-06, "loss": 1.5605, "step": 32160 }, { "epoch": 3.06706, "grad_norm": 22.206518378219155, "learning_rate": 7.929879212484228e-06, "loss": 1.6629, "step": 32170 }, { "epoch": 3.06716, "grad_norm": 23.6954550518028, "learning_rate": 7.928566830940417e-06, "loss": 1.6322, "step": 32180 }, { "epoch": 3.06726, "grad_norm": 23.579338559162338, "learning_rate": 7.927254142203535e-06, "loss": 1.5874, "step": 32190 }, { "epoch": 3.06736, "grad_norm": 22.535767747328716, "learning_rate": 7.92594114641128e-06, "loss": 1.6446, "step": 32200 }, { "epoch": 3.06746, "grad_norm": 21.803832372363864, "learning_rate": 7.92462784370138e-06, "loss": 1.6374, "step": 32210 }, { "epoch": 3.06756, "grad_norm": 26.00507936101173, "learning_rate": 7.92331423421159e-06, "loss": 1.6544, "step": 32220 }, { "epoch": 3.06766, "grad_norm": 20.46142686642296, "learning_rate": 7.922000318079706e-06, "loss": 1.5187, "step": 32230 }, { "epoch": 3.06776, "grad_norm": 23.770183344315576, "learning_rate": 7.92068609544355e-06, "loss": 1.5653, "step": 32240 }, { "epoch": 3.06786, "grad_norm": 23.622185525224165, "learning_rate": 7.91937156644098e-06, "loss": 1.6532, "step": 32250 }, { "epoch": 3.06796, "grad_norm": 24.432297111044797, "learning_rate": 7.918056731209878e-06, "loss": 1.6337, "step": 32260 }, { "epoch": 3.06806, "grad_norm": 21.47210794303764, "learning_rate": 7.91674158988817e-06, "loss": 1.5556, "step": 32270 }, { "epoch": 3.0681599999999998, "grad_norm": 24.357979281666683, "learning_rate": 7.915426142613807e-06, "loss": 1.5839, "step": 32280 }, { "epoch": 3.06826, "grad_norm": 21.152402078681764, "learning_rate": 7.914110389524772e-06, "loss": 1.6087, "step": 32290 }, { "epoch": 3.06836, "grad_norm": 23.93620319242678, "learning_rate": 7.912794330759082e-06, "loss": 1.5931, "step": 32300 }, { "epoch": 3.06846, "grad_norm": 21.669885280594702, "learning_rate": 7.911477966454783e-06, "loss": 1.6114, "step": 32310 }, { "epoch": 3.06856, "grad_norm": 21.29849573251636, "learning_rate": 7.91016129674996e-06, "loss": 1.5959, "step": 32320 }, { "epoch": 3.06866, "grad_norm": 24.042051073894335, "learning_rate": 7.908844321782724e-06, "loss": 1.5576, "step": 32330 }, { "epoch": 3.06876, "grad_norm": 22.176047396045696, "learning_rate": 7.907527041691218e-06, "loss": 1.5606, "step": 32340 }, { "epoch": 3.06886, "grad_norm": 19.566932599933995, "learning_rate": 7.90620945661362e-06, "loss": 1.4796, "step": 32350 }, { "epoch": 3.06896, "grad_norm": 25.895921320389068, "learning_rate": 7.904891566688138e-06, "loss": 1.5439, "step": 32360 }, { "epoch": 3.06906, "grad_norm": 21.330424331091308, "learning_rate": 7.903573372053015e-06, "loss": 1.5775, "step": 32370 }, { "epoch": 3.06916, "grad_norm": 22.415220166688645, "learning_rate": 7.902254872846518e-06, "loss": 1.6018, "step": 32380 }, { "epoch": 3.06926, "grad_norm": 21.423584740106936, "learning_rate": 7.900936069206955e-06, "loss": 1.6602, "step": 32390 }, { "epoch": 3.06936, "grad_norm": 24.506727778826992, "learning_rate": 7.899616961272662e-06, "loss": 1.6248, "step": 32400 }, { "epoch": 3.06946, "grad_norm": 21.587485108665646, "learning_rate": 7.898297549182008e-06, "loss": 1.5656, "step": 32410 }, { "epoch": 3.06956, "grad_norm": 20.935587062636174, "learning_rate": 7.896977833073393e-06, "loss": 1.5331, "step": 32420 }, { "epoch": 3.06966, "grad_norm": 20.811100721558397, "learning_rate": 7.895657813085248e-06, "loss": 1.5379, "step": 32430 }, { "epoch": 3.06976, "grad_norm": 22.231163781182186, "learning_rate": 7.894337489356037e-06, "loss": 1.5278, "step": 32440 }, { "epoch": 3.06986, "grad_norm": 24.165645536455425, "learning_rate": 7.893016862024256e-06, "loss": 1.5873, "step": 32450 }, { "epoch": 3.06996, "grad_norm": 22.551381944209137, "learning_rate": 7.891695931228435e-06, "loss": 1.6479, "step": 32460 }, { "epoch": 3.07006, "grad_norm": 19.54410970548744, "learning_rate": 7.89037469710713e-06, "loss": 1.538, "step": 32470 }, { "epoch": 3.07016, "grad_norm": 21.673413937690533, "learning_rate": 7.889053159798935e-06, "loss": 1.6234, "step": 32480 }, { "epoch": 3.07026, "grad_norm": 22.29674043420407, "learning_rate": 7.887731319442473e-06, "loss": 1.5611, "step": 32490 }, { "epoch": 3.07036, "grad_norm": 23.323212699554293, "learning_rate": 7.886409176176394e-06, "loss": 1.5737, "step": 32500 }, { "epoch": 3.07046, "grad_norm": 21.351347950813718, "learning_rate": 7.885086730139394e-06, "loss": 1.5243, "step": 32510 }, { "epoch": 3.07056, "grad_norm": 20.97605494838548, "learning_rate": 7.883763981470182e-06, "loss": 1.5293, "step": 32520 }, { "epoch": 3.07066, "grad_norm": 23.90590668658707, "learning_rate": 7.882440930307513e-06, "loss": 1.6288, "step": 32530 }, { "epoch": 3.07076, "grad_norm": 23.146023060747968, "learning_rate": 7.88111757679017e-06, "loss": 1.6566, "step": 32540 }, { "epoch": 3.07086, "grad_norm": 23.919597103143804, "learning_rate": 7.879793921056964e-06, "loss": 1.6312, "step": 32550 }, { "epoch": 3.07096, "grad_norm": 22.068864658436002, "learning_rate": 7.87846996324674e-06, "loss": 1.5964, "step": 32560 }, { "epoch": 3.07106, "grad_norm": 21.993765790039994, "learning_rate": 7.877145703498377e-06, "loss": 1.4588, "step": 32570 }, { "epoch": 3.07116, "grad_norm": 21.4123750173379, "learning_rate": 7.875821141950783e-06, "loss": 1.6452, "step": 32580 }, { "epoch": 3.07126, "grad_norm": 20.19182124418368, "learning_rate": 7.874496278742898e-06, "loss": 1.5935, "step": 32590 }, { "epoch": 3.07136, "grad_norm": 22.410867214663483, "learning_rate": 7.873171114013696e-06, "loss": 1.5327, "step": 32600 }, { "epoch": 3.07146, "grad_norm": 19.22602586115108, "learning_rate": 7.871845647902177e-06, "loss": 1.5706, "step": 32610 }, { "epoch": 3.07156, "grad_norm": 22.422287132144263, "learning_rate": 7.870519880547379e-06, "loss": 1.4693, "step": 32620 }, { "epoch": 3.07166, "grad_norm": 23.721756641557334, "learning_rate": 7.869193812088368e-06, "loss": 1.5007, "step": 32630 }, { "epoch": 3.07176, "grad_norm": 23.201910209600893, "learning_rate": 7.86786744266424e-06, "loss": 1.6125, "step": 32640 }, { "epoch": 3.07186, "grad_norm": 22.328467180308387, "learning_rate": 7.866540772414131e-06, "loss": 1.5167, "step": 32650 }, { "epoch": 3.07196, "grad_norm": 23.35385877096771, "learning_rate": 7.865213801477199e-06, "loss": 1.4548, "step": 32660 }, { "epoch": 3.07206, "grad_norm": 28.475977556567823, "learning_rate": 7.863886529992634e-06, "loss": 1.559, "step": 32670 }, { "epoch": 3.0721600000000002, "grad_norm": 21.91911388503643, "learning_rate": 7.862558958099666e-06, "loss": 1.5535, "step": 32680 }, { "epoch": 3.07226, "grad_norm": 22.425194291101267, "learning_rate": 7.861231085937549e-06, "loss": 1.5923, "step": 32690 }, { "epoch": 3.07236, "grad_norm": 20.31482393905907, "learning_rate": 7.859902913645571e-06, "loss": 1.5865, "step": 32700 }, { "epoch": 3.07246, "grad_norm": 21.629447327468974, "learning_rate": 7.85857444136305e-06, "loss": 1.5495, "step": 32710 }, { "epoch": 3.07256, "grad_norm": 24.04647703879316, "learning_rate": 7.857245669229335e-06, "loss": 1.5693, "step": 32720 }, { "epoch": 3.07266, "grad_norm": 24.12497555076957, "learning_rate": 7.855916597383812e-06, "loss": 1.5225, "step": 32730 }, { "epoch": 3.07276, "grad_norm": 23.2769652680979, "learning_rate": 7.854587225965894e-06, "loss": 1.5635, "step": 32740 }, { "epoch": 3.07286, "grad_norm": 23.08608195045423, "learning_rate": 7.853257555115021e-06, "loss": 1.5284, "step": 32750 }, { "epoch": 3.07296, "grad_norm": 24.127243167874393, "learning_rate": 7.851927584970676e-06, "loss": 1.5617, "step": 32760 }, { "epoch": 3.07306, "grad_norm": 20.880949313297474, "learning_rate": 7.850597315672362e-06, "loss": 1.527, "step": 32770 }, { "epoch": 3.07316, "grad_norm": 22.254250910064364, "learning_rate": 7.849266747359619e-06, "loss": 1.5657, "step": 32780 }, { "epoch": 3.07326, "grad_norm": 23.5928551254128, "learning_rate": 7.84793588017202e-06, "loss": 1.555, "step": 32790 }, { "epoch": 3.07336, "grad_norm": 23.19399209525167, "learning_rate": 7.846604714249164e-06, "loss": 1.5027, "step": 32800 }, { "epoch": 3.07346, "grad_norm": 20.786121936770144, "learning_rate": 7.845273249730685e-06, "loss": 1.5635, "step": 32810 }, { "epoch": 3.07356, "grad_norm": 24.462484367716335, "learning_rate": 7.843941486756247e-06, "loss": 1.5113, "step": 32820 }, { "epoch": 3.07366, "grad_norm": 22.172429038630987, "learning_rate": 7.842609425465545e-06, "loss": 1.5267, "step": 32830 }, { "epoch": 3.07376, "grad_norm": 23.340096063447874, "learning_rate": 7.84127706599831e-06, "loss": 1.531, "step": 32840 }, { "epoch": 3.07386, "grad_norm": 23.50651822467019, "learning_rate": 7.839944408494296e-06, "loss": 1.51, "step": 32850 }, { "epoch": 3.07396, "grad_norm": 22.15097712329124, "learning_rate": 7.838611453093294e-06, "loss": 1.515, "step": 32860 }, { "epoch": 3.07406, "grad_norm": 22.554646526154286, "learning_rate": 7.837278199935126e-06, "loss": 1.5581, "step": 32870 }, { "epoch": 3.07416, "grad_norm": 22.223905890189766, "learning_rate": 7.835944649159643e-06, "loss": 1.5115, "step": 32880 }, { "epoch": 3.07426, "grad_norm": 21.767490218507035, "learning_rate": 7.834610800906726e-06, "loss": 1.5746, "step": 32890 }, { "epoch": 3.07436, "grad_norm": 25.70701787297682, "learning_rate": 7.833276655316295e-06, "loss": 1.6282, "step": 32900 }, { "epoch": 3.07446, "grad_norm": 20.73763098783727, "learning_rate": 7.831942212528292e-06, "loss": 1.5087, "step": 32910 }, { "epoch": 3.07456, "grad_norm": 24.892499945605824, "learning_rate": 7.830607472682693e-06, "loss": 1.5896, "step": 32920 }, { "epoch": 3.07466, "grad_norm": 23.312500778021818, "learning_rate": 7.829272435919507e-06, "loss": 1.5255, "step": 32930 }, { "epoch": 3.07476, "grad_norm": 21.85748117176196, "learning_rate": 7.827937102378775e-06, "loss": 1.5021, "step": 32940 }, { "epoch": 3.07486, "grad_norm": 20.973383712346365, "learning_rate": 7.826601472200566e-06, "loss": 1.5573, "step": 32950 }, { "epoch": 3.07496, "grad_norm": 22.05228281186569, "learning_rate": 7.82526554552498e-06, "loss": 1.5421, "step": 32960 }, { "epoch": 3.07506, "grad_norm": 19.46437272989484, "learning_rate": 7.823929322492149e-06, "loss": 1.5341, "step": 32970 }, { "epoch": 3.07516, "grad_norm": 20.65959540037305, "learning_rate": 7.822592803242242e-06, "loss": 1.5358, "step": 32980 }, { "epoch": 3.07526, "grad_norm": 21.14647821313492, "learning_rate": 7.821255987915447e-06, "loss": 1.5135, "step": 32990 }, { "epoch": 3.07536, "grad_norm": 21.774943604195915, "learning_rate": 7.819918876651996e-06, "loss": 1.5277, "step": 33000 }, { "epoch": 3.07546, "grad_norm": 20.54647872692517, "learning_rate": 7.81858146959214e-06, "loss": 1.5072, "step": 33010 }, { "epoch": 3.07556, "grad_norm": 21.31076659043196, "learning_rate": 7.817243766876171e-06, "loss": 1.5456, "step": 33020 }, { "epoch": 3.07566, "grad_norm": 20.37265588790006, "learning_rate": 7.815905768644405e-06, "loss": 1.533, "step": 33030 }, { "epoch": 3.07576, "grad_norm": 24.002707489395725, "learning_rate": 7.814567475037191e-06, "loss": 1.5297, "step": 33040 }, { "epoch": 3.07586, "grad_norm": 22.791431741127386, "learning_rate": 7.813228886194914e-06, "loss": 1.4907, "step": 33050 }, { "epoch": 3.07596, "grad_norm": 21.179447542727882, "learning_rate": 7.811890002257983e-06, "loss": 1.5636, "step": 33060 }, { "epoch": 3.07606, "grad_norm": 21.70144636627567, "learning_rate": 7.810550823366841e-06, "loss": 1.4756, "step": 33070 }, { "epoch": 3.07616, "grad_norm": 21.896357604030268, "learning_rate": 7.809211349661963e-06, "loss": 1.4757, "step": 33080 }, { "epoch": 3.07626, "grad_norm": 22.49655326797683, "learning_rate": 7.80787158128385e-06, "loss": 1.5104, "step": 33090 }, { "epoch": 3.07636, "grad_norm": 20.584657305543455, "learning_rate": 7.80653151837304e-06, "loss": 1.4494, "step": 33100 }, { "epoch": 3.07646, "grad_norm": 22.90735083470913, "learning_rate": 7.805191161070103e-06, "loss": 1.5064, "step": 33110 }, { "epoch": 3.07656, "grad_norm": 22.031439187094353, "learning_rate": 7.803850509515629e-06, "loss": 1.4598, "step": 33120 }, { "epoch": 3.07666, "grad_norm": 21.443827986862004, "learning_rate": 7.802509563850251e-06, "loss": 1.4123, "step": 33130 }, { "epoch": 3.07676, "grad_norm": 23.491070045408254, "learning_rate": 7.801168324214627e-06, "loss": 1.4861, "step": 33140 }, { "epoch": 3.07686, "grad_norm": 20.91634858370723, "learning_rate": 7.799826790749445e-06, "loss": 1.435, "step": 33150 }, { "epoch": 3.07696, "grad_norm": 23.39053190888209, "learning_rate": 7.798484963595427e-06, "loss": 1.5588, "step": 33160 }, { "epoch": 3.07706, "grad_norm": 20.81647934367272, "learning_rate": 7.797142842893326e-06, "loss": 1.4837, "step": 33170 }, { "epoch": 3.07716, "grad_norm": 20.544089942657184, "learning_rate": 7.795800428783922e-06, "loss": 1.4624, "step": 33180 }, { "epoch": 3.07726, "grad_norm": 19.895994479058075, "learning_rate": 7.79445772140803e-06, "loss": 1.4273, "step": 33190 }, { "epoch": 4.00004, "grad_norm": 21.788492182634897, "learning_rate": 7.793114720906489e-06, "loss": 1.5785, "step": 33200 }, { "epoch": 4.00014, "grad_norm": 22.226517112850686, "learning_rate": 7.791771427420179e-06, "loss": 1.566, "step": 33210 }, { "epoch": 4.00024, "grad_norm": 23.6320212597196, "learning_rate": 7.790427841090003e-06, "loss": 1.5349, "step": 33220 }, { "epoch": 4.00034, "grad_norm": 23.71760788381119, "learning_rate": 7.789083962056897e-06, "loss": 1.5633, "step": 33230 }, { "epoch": 4.00044, "grad_norm": 24.269169366378392, "learning_rate": 7.787739790461831e-06, "loss": 1.6226, "step": 33240 }, { "epoch": 4.00054, "grad_norm": 22.403291575179342, "learning_rate": 7.786395326445795e-06, "loss": 1.6261, "step": 33250 }, { "epoch": 4.00064, "grad_norm": 21.43474353993906, "learning_rate": 7.785050570149824e-06, "loss": 1.6015, "step": 33260 }, { "epoch": 4.00074, "grad_norm": 21.101552307083193, "learning_rate": 7.783705521714973e-06, "loss": 1.5247, "step": 33270 }, { "epoch": 4.00084, "grad_norm": 19.841103327591533, "learning_rate": 7.782360181282333e-06, "loss": 1.5835, "step": 33280 }, { "epoch": 4.00094, "grad_norm": 23.424960453815192, "learning_rate": 7.781014548993023e-06, "loss": 1.5659, "step": 33290 }, { "epoch": 4.00104, "grad_norm": 22.83375821162665, "learning_rate": 7.779668624988192e-06, "loss": 1.6013, "step": 33300 }, { "epoch": 4.00114, "grad_norm": 22.403742706178054, "learning_rate": 7.778322409409026e-06, "loss": 1.6112, "step": 33310 }, { "epoch": 4.00124, "grad_norm": 22.109074281675863, "learning_rate": 7.776975902396734e-06, "loss": 1.5731, "step": 33320 }, { "epoch": 4.00134, "grad_norm": 21.621343268940983, "learning_rate": 7.775629104092555e-06, "loss": 1.5706, "step": 33330 }, { "epoch": 4.00144, "grad_norm": 18.767806636131496, "learning_rate": 7.774282014637766e-06, "loss": 1.5121, "step": 33340 }, { "epoch": 4.00154, "grad_norm": 22.721833463223177, "learning_rate": 7.77293463417367e-06, "loss": 1.468, "step": 33350 }, { "epoch": 4.00164, "grad_norm": 22.196150891563327, "learning_rate": 7.771586962841601e-06, "loss": 1.5616, "step": 33360 }, { "epoch": 4.00174, "grad_norm": 23.383269350054253, "learning_rate": 7.770239000782925e-06, "loss": 1.5091, "step": 33370 }, { "epoch": 4.00184, "grad_norm": 20.963362850068325, "learning_rate": 7.768890748139032e-06, "loss": 1.6067, "step": 33380 }, { "epoch": 4.00194, "grad_norm": 22.752218386846657, "learning_rate": 7.76754220505135e-06, "loss": 1.5233, "step": 33390 }, { "epoch": 4.00204, "grad_norm": 20.658035136965616, "learning_rate": 7.766193371661336e-06, "loss": 1.4898, "step": 33400 }, { "epoch": 4.00214, "grad_norm": 20.561774337965804, "learning_rate": 7.764844248110476e-06, "loss": 1.5471, "step": 33410 }, { "epoch": 4.00224, "grad_norm": 19.729037877613926, "learning_rate": 7.763494834540287e-06, "loss": 1.5942, "step": 33420 }, { "epoch": 4.00234, "grad_norm": 21.990022071040848, "learning_rate": 7.762145131092314e-06, "loss": 1.5784, "step": 33430 }, { "epoch": 4.00244, "grad_norm": 22.206790896497232, "learning_rate": 7.76079513790814e-06, "loss": 1.5734, "step": 33440 }, { "epoch": 4.00254, "grad_norm": 21.97476131160628, "learning_rate": 7.759444855129365e-06, "loss": 1.5098, "step": 33450 }, { "epoch": 4.00264, "grad_norm": 21.67136445599637, "learning_rate": 7.758094282897632e-06, "loss": 1.5304, "step": 33460 }, { "epoch": 4.00274, "grad_norm": 21.786092740028344, "learning_rate": 7.75674342135461e-06, "loss": 1.51, "step": 33470 }, { "epoch": 4.00284, "grad_norm": 22.647950330620493, "learning_rate": 7.755392270641999e-06, "loss": 1.5181, "step": 33480 }, { "epoch": 4.00294, "grad_norm": 22.02621384910114, "learning_rate": 7.754040830901527e-06, "loss": 1.4707, "step": 33490 }, { "epoch": 4.00304, "grad_norm": 23.918737139444673, "learning_rate": 7.752689102274954e-06, "loss": 1.5976, "step": 33500 }, { "epoch": 4.00314, "grad_norm": 23.261457768454957, "learning_rate": 7.751337084904066e-06, "loss": 1.518, "step": 33510 }, { "epoch": 4.00324, "grad_norm": 24.135111077183403, "learning_rate": 7.74998477893069e-06, "loss": 1.4681, "step": 33520 }, { "epoch": 4.00334, "grad_norm": 20.83675151927745, "learning_rate": 7.748632184496674e-06, "loss": 1.5593, "step": 33530 }, { "epoch": 4.00344, "grad_norm": 25.630549330603476, "learning_rate": 7.7472793017439e-06, "loss": 1.6404, "step": 33540 }, { "epoch": 4.00354, "grad_norm": 22.691552031190206, "learning_rate": 7.745926130814273e-06, "loss": 1.5514, "step": 33550 }, { "epoch": 4.00364, "grad_norm": 20.81359830670423, "learning_rate": 7.744572671849744e-06, "loss": 1.5576, "step": 33560 }, { "epoch": 4.00374, "grad_norm": 22.276384831659755, "learning_rate": 7.743218924992277e-06, "loss": 1.578, "step": 33570 }, { "epoch": 4.00384, "grad_norm": 22.37760317574486, "learning_rate": 7.741864890383876e-06, "loss": 1.5023, "step": 33580 }, { "epoch": 4.00394, "grad_norm": 23.015639055528368, "learning_rate": 7.740510568166578e-06, "loss": 1.5067, "step": 33590 }, { "epoch": 4.00404, "grad_norm": 20.965472930939164, "learning_rate": 7.739155958482434e-06, "loss": 1.5089, "step": 33600 }, { "epoch": 4.00414, "grad_norm": 24.432045695311448, "learning_rate": 7.737801061473548e-06, "loss": 1.5785, "step": 33610 }, { "epoch": 4.00424, "grad_norm": 19.76195683167377, "learning_rate": 7.736445877282034e-06, "loss": 1.6057, "step": 33620 }, { "epoch": 4.00434, "grad_norm": 23.232327890118732, "learning_rate": 7.735090406050049e-06, "loss": 1.6155, "step": 33630 }, { "epoch": 4.00444, "grad_norm": 20.423126052821086, "learning_rate": 7.733734647919774e-06, "loss": 1.5206, "step": 33640 }, { "epoch": 4.00454, "grad_norm": 20.877869625723783, "learning_rate": 7.732378603033422e-06, "loss": 1.4995, "step": 33650 }, { "epoch": 4.00464, "grad_norm": 22.82511649463471, "learning_rate": 7.731022271533238e-06, "loss": 1.4925, "step": 33660 }, { "epoch": 4.00474, "grad_norm": 20.604513792978562, "learning_rate": 7.729665653561489e-06, "loss": 1.5053, "step": 33670 }, { "epoch": 4.00484, "grad_norm": 22.523347433443476, "learning_rate": 7.728308749260483e-06, "loss": 1.496, "step": 33680 }, { "epoch": 4.00494, "grad_norm": 22.374305822349164, "learning_rate": 7.726951558772553e-06, "loss": 1.5007, "step": 33690 }, { "epoch": 4.00504, "grad_norm": 20.11087514983212, "learning_rate": 7.72559408224006e-06, "loss": 1.4711, "step": 33700 }, { "epoch": 4.00514, "grad_norm": 23.303222502779033, "learning_rate": 7.724236319805397e-06, "loss": 1.5274, "step": 33710 }, { "epoch": 4.00524, "grad_norm": 23.595987774534, "learning_rate": 7.72287827161099e-06, "loss": 1.5901, "step": 33720 }, { "epoch": 4.00534, "grad_norm": 22.28618766218241, "learning_rate": 7.721519937799286e-06, "loss": 1.5369, "step": 33730 }, { "epoch": 4.00544, "grad_norm": 20.624693268669546, "learning_rate": 7.720161318512773e-06, "loss": 1.53, "step": 33740 }, { "epoch": 4.00554, "grad_norm": 21.43608221151987, "learning_rate": 7.718802413893963e-06, "loss": 1.5699, "step": 33750 }, { "epoch": 4.00564, "grad_norm": 22.211364582674275, "learning_rate": 7.717443224085396e-06, "loss": 1.5899, "step": 33760 }, { "epoch": 4.00574, "grad_norm": 21.159165177686464, "learning_rate": 7.71608374922965e-06, "loss": 1.59, "step": 33770 }, { "epoch": 4.00584, "grad_norm": 18.488023723987116, "learning_rate": 7.714723989469324e-06, "loss": 1.4789, "step": 33780 }, { "epoch": 4.00594, "grad_norm": 22.476352248779552, "learning_rate": 7.713363944947049e-06, "loss": 1.487, "step": 33790 }, { "epoch": 4.00604, "grad_norm": 23.620759959394668, "learning_rate": 7.71200361580549e-06, "loss": 1.4322, "step": 33800 }, { "epoch": 4.00614, "grad_norm": 20.96096716397358, "learning_rate": 7.710643002187339e-06, "loss": 1.4395, "step": 33810 }, { "epoch": 4.00624, "grad_norm": 21.43116482716774, "learning_rate": 7.709282104235319e-06, "loss": 1.4926, "step": 33820 }, { "epoch": 4.00634, "grad_norm": 22.65624375898815, "learning_rate": 7.707920922092176e-06, "loss": 1.5093, "step": 33830 }, { "epoch": 4.00644, "grad_norm": 24.546122673435054, "learning_rate": 7.706559455900701e-06, "loss": 1.472, "step": 33840 }, { "epoch": 4.00654, "grad_norm": 24.022024866527232, "learning_rate": 7.705197705803697e-06, "loss": 1.5078, "step": 33850 }, { "epoch": 4.00664, "grad_norm": 22.809987120073906, "learning_rate": 7.703835671944012e-06, "loss": 1.5097, "step": 33860 }, { "epoch": 4.00674, "grad_norm": 22.50041491051627, "learning_rate": 7.702473354464512e-06, "loss": 1.5044, "step": 33870 }, { "epoch": 4.00684, "grad_norm": 23.2805795514464, "learning_rate": 7.701110753508098e-06, "loss": 1.5565, "step": 33880 }, { "epoch": 4.00694, "grad_norm": 21.923113376592976, "learning_rate": 7.699747869217704e-06, "loss": 1.5654, "step": 33890 }, { "epoch": 4.00704, "grad_norm": 24.207137536673955, "learning_rate": 7.69838470173629e-06, "loss": 1.5783, "step": 33900 }, { "epoch": 4.00714, "grad_norm": 21.232759680700315, "learning_rate": 7.697021251206841e-06, "loss": 1.4906, "step": 33910 }, { "epoch": 4.00724, "grad_norm": 22.644421673867875, "learning_rate": 7.695657517772383e-06, "loss": 1.5442, "step": 33920 }, { "epoch": 4.00734, "grad_norm": 21.199147685454246, "learning_rate": 7.694293501575961e-06, "loss": 1.568, "step": 33930 }, { "epoch": 4.00744, "grad_norm": 21.870013210763425, "learning_rate": 7.692929202760658e-06, "loss": 1.4901, "step": 33940 }, { "epoch": 4.00754, "grad_norm": 20.56242690343119, "learning_rate": 7.691564621469575e-06, "loss": 1.4513, "step": 33950 }, { "epoch": 4.00764, "grad_norm": 23.929194195046648, "learning_rate": 7.690199757845857e-06, "loss": 1.531, "step": 33960 }, { "epoch": 4.00774, "grad_norm": 20.471652411126897, "learning_rate": 7.688834612032672e-06, "loss": 1.428, "step": 33970 }, { "epoch": 4.00784, "grad_norm": 21.964881967761933, "learning_rate": 7.687469184173214e-06, "loss": 1.4682, "step": 33980 }, { "epoch": 4.00794, "grad_norm": 22.89916823728625, "learning_rate": 7.686103474410712e-06, "loss": 1.5018, "step": 33990 }, { "epoch": 4.00804, "grad_norm": 22.312640523701663, "learning_rate": 7.68473748288842e-06, "loss": 1.5839, "step": 34000 }, { "epoch": 4.00814, "grad_norm": 21.915008515565034, "learning_rate": 7.683371209749628e-06, "loss": 1.5211, "step": 34010 }, { "epoch": 4.00824, "grad_norm": 20.85993945931427, "learning_rate": 7.682004655137651e-06, "loss": 1.4972, "step": 34020 }, { "epoch": 4.00834, "grad_norm": 23.77445106777416, "learning_rate": 7.680637819195831e-06, "loss": 1.4588, "step": 34030 }, { "epoch": 4.00844, "grad_norm": 21.55701843652847, "learning_rate": 7.679270702067547e-06, "loss": 1.5217, "step": 34040 }, { "epoch": 4.00854, "grad_norm": 23.69101217463169, "learning_rate": 7.677903303896202e-06, "loss": 1.4972, "step": 34050 }, { "epoch": 4.00864, "grad_norm": 23.184077657190223, "learning_rate": 7.676535624825227e-06, "loss": 1.5456, "step": 34060 }, { "epoch": 4.00874, "grad_norm": 21.671487713917706, "learning_rate": 7.675167664998086e-06, "loss": 1.5149, "step": 34070 }, { "epoch": 4.00884, "grad_norm": 22.955400720509253, "learning_rate": 7.673799424558275e-06, "loss": 1.5236, "step": 34080 }, { "epoch": 4.00894, "grad_norm": 22.284003536744294, "learning_rate": 7.672430903649312e-06, "loss": 1.567, "step": 34090 }, { "epoch": 4.00904, "grad_norm": 21.836764034985364, "learning_rate": 7.671062102414753e-06, "loss": 1.4709, "step": 34100 }, { "epoch": 4.00914, "grad_norm": 22.19857087095689, "learning_rate": 7.669693020998175e-06, "loss": 1.5562, "step": 34110 }, { "epoch": 4.00924, "grad_norm": 22.691810986326157, "learning_rate": 7.668323659543189e-06, "loss": 1.4649, "step": 34120 }, { "epoch": 4.00934, "grad_norm": 21.401444420425868, "learning_rate": 7.666954018193435e-06, "loss": 1.5135, "step": 34130 }, { "epoch": 4.00944, "grad_norm": 24.471681920184746, "learning_rate": 7.665584097092583e-06, "loss": 1.4415, "step": 34140 }, { "epoch": 4.00954, "grad_norm": 20.860192698018288, "learning_rate": 7.66421389638433e-06, "loss": 1.4857, "step": 34150 }, { "epoch": 4.00964, "grad_norm": 21.706241723866338, "learning_rate": 7.662843416212405e-06, "loss": 1.5095, "step": 34160 }, { "epoch": 4.00974, "grad_norm": 28.138061772720068, "learning_rate": 7.661472656720563e-06, "loss": 1.4564, "step": 34170 }, { "epoch": 4.00984, "grad_norm": 22.091061304859103, "learning_rate": 7.660101618052592e-06, "loss": 1.5305, "step": 34180 }, { "epoch": 4.00994, "grad_norm": 24.980517343895276, "learning_rate": 7.658730300352308e-06, "loss": 1.5039, "step": 34190 }, { "epoch": 4.01004, "grad_norm": 21.352389847393578, "learning_rate": 7.657358703763556e-06, "loss": 1.4523, "step": 34200 }, { "epoch": 4.01014, "grad_norm": 24.195689966456666, "learning_rate": 7.655986828430206e-06, "loss": 1.5325, "step": 34210 }, { "epoch": 4.01024, "grad_norm": 19.99939648005553, "learning_rate": 7.654614674496167e-06, "loss": 1.5192, "step": 34220 }, { "epoch": 4.01034, "grad_norm": 20.351062222439538, "learning_rate": 7.653242242105369e-06, "loss": 1.5043, "step": 34230 }, { "epoch": 4.01044, "grad_norm": 21.40681356899588, "learning_rate": 7.651869531401772e-06, "loss": 1.5107, "step": 34240 }, { "epoch": 4.01054, "grad_norm": 20.845047092400687, "learning_rate": 7.650496542529371e-06, "loss": 1.5227, "step": 34250 }, { "epoch": 4.01064, "grad_norm": 23.0145687632217, "learning_rate": 7.649123275632182e-06, "loss": 1.5016, "step": 34260 }, { "epoch": 4.01074, "grad_norm": 21.105634405335604, "learning_rate": 7.647749730854258e-06, "loss": 1.4761, "step": 34270 }, { "epoch": 4.01084, "grad_norm": 22.973749798004018, "learning_rate": 7.646375908339676e-06, "loss": 1.4586, "step": 34280 }, { "epoch": 4.01094, "grad_norm": 20.34511102916892, "learning_rate": 7.645001808232541e-06, "loss": 1.4296, "step": 34290 }, { "epoch": 4.01104, "grad_norm": 19.539821188640644, "learning_rate": 7.643627430676995e-06, "loss": 1.3865, "step": 34300 }, { "epoch": 4.01114, "grad_norm": 23.94670821301097, "learning_rate": 7.6422527758172e-06, "loss": 1.5097, "step": 34310 }, { "epoch": 4.01124, "grad_norm": 18.671893803860502, "learning_rate": 7.640877843797351e-06, "loss": 1.379, "step": 34320 }, { "epoch": 4.01134, "grad_norm": 21.56196482083835, "learning_rate": 7.639502634761674e-06, "loss": 1.4474, "step": 34330 }, { "epoch": 4.01144, "grad_norm": 25.536534293439313, "learning_rate": 7.63812714885442e-06, "loss": 1.4714, "step": 34340 }, { "epoch": 4.01154, "grad_norm": 24.331115845654203, "learning_rate": 7.636751386219874e-06, "loss": 1.4206, "step": 34350 }, { "epoch": 4.01164, "grad_norm": 23.136343268991382, "learning_rate": 7.635375347002342e-06, "loss": 1.4251, "step": 34360 }, { "epoch": 4.01174, "grad_norm": 23.120754557090894, "learning_rate": 7.63399903134617e-06, "loss": 1.4276, "step": 34370 }, { "epoch": 4.01184, "grad_norm": 22.98792484167995, "learning_rate": 7.632622439395723e-06, "loss": 1.3786, "step": 34380 }, { "epoch": 4.01194, "grad_norm": 22.43814189703378, "learning_rate": 7.6312455712954e-06, "loss": 1.4625, "step": 34390 }, { "epoch": 4.01204, "grad_norm": 21.8874629886281, "learning_rate": 7.62986842718963e-06, "loss": 1.5064, "step": 34400 }, { "epoch": 4.01214, "grad_norm": 20.3698208386688, "learning_rate": 7.628491007222867e-06, "loss": 1.4193, "step": 34410 }, { "epoch": 4.01224, "grad_norm": 21.21717551426848, "learning_rate": 7.6271133115395955e-06, "loss": 1.4232, "step": 34420 }, { "epoch": 4.01234, "grad_norm": 22.766425222145415, "learning_rate": 7.625735340284332e-06, "loss": 1.485, "step": 34430 }, { "epoch": 4.01244, "grad_norm": 25.15540471548708, "learning_rate": 7.624357093601617e-06, "loss": 1.5397, "step": 34440 }, { "epoch": 4.01254, "grad_norm": 24.708796954952682, "learning_rate": 7.6229785716360235e-06, "loss": 1.5353, "step": 34450 }, { "epoch": 4.01264, "grad_norm": 22.776269417379936, "learning_rate": 7.621599774532151e-06, "loss": 1.428, "step": 34460 }, { "epoch": 4.01274, "grad_norm": 23.008699457835352, "learning_rate": 7.620220702434629e-06, "loss": 1.4397, "step": 34470 }, { "epoch": 4.01284, "grad_norm": 20.841873736528115, "learning_rate": 7.618841355488118e-06, "loss": 1.4448, "step": 34480 }, { "epoch": 4.01294, "grad_norm": 23.068303661452717, "learning_rate": 7.617461733837302e-06, "loss": 1.4965, "step": 34490 }, { "epoch": 4.01304, "grad_norm": 21.187963474996412, "learning_rate": 7.616081837626897e-06, "loss": 1.5058, "step": 34500 }, { "epoch": 4.01314, "grad_norm": 24.491636448212642, "learning_rate": 7.6147016670016495e-06, "loss": 1.4274, "step": 34510 }, { "epoch": 4.01324, "grad_norm": 23.248098329106377, "learning_rate": 7.613321222106331e-06, "loss": 1.478, "step": 34520 }, { "epoch": 4.01334, "grad_norm": 23.177465126564922, "learning_rate": 7.6119405030857485e-06, "loss": 1.5109, "step": 34530 }, { "epoch": 4.01344, "grad_norm": 21.693291678760602, "learning_rate": 7.610559510084727e-06, "loss": 1.4763, "step": 34540 }, { "epoch": 4.01354, "grad_norm": 22.112221133921828, "learning_rate": 7.60917824324813e-06, "loss": 1.4723, "step": 34550 }, { "epoch": 4.01364, "grad_norm": 21.13259434387763, "learning_rate": 7.607796702720844e-06, "loss": 1.478, "step": 34560 }, { "epoch": 4.01374, "grad_norm": 21.48003029796468, "learning_rate": 7.606414888647788e-06, "loss": 1.5498, "step": 34570 }, { "epoch": 4.01384, "grad_norm": 23.558281928055962, "learning_rate": 7.605032801173906e-06, "loss": 1.5143, "step": 34580 }, { "epoch": 4.01394, "grad_norm": 20.842094478321137, "learning_rate": 7.603650440444174e-06, "loss": 1.3945, "step": 34590 }, { "epoch": 4.01404, "grad_norm": 18.976522049200497, "learning_rate": 7.602267806603595e-06, "loss": 1.4095, "step": 34600 }, { "epoch": 4.01414, "grad_norm": 22.698426272752748, "learning_rate": 7.600884899797199e-06, "loss": 1.4923, "step": 34610 }, { "epoch": 4.01424, "grad_norm": 22.165597425864526, "learning_rate": 7.5995017201700505e-06, "loss": 1.4797, "step": 34620 }, { "epoch": 4.01434, "grad_norm": 22.066046476893312, "learning_rate": 7.598118267867235e-06, "loss": 1.5022, "step": 34630 }, { "epoch": 4.01444, "grad_norm": 22.46691623164771, "learning_rate": 7.596734543033871e-06, "loss": 1.5242, "step": 34640 }, { "epoch": 4.01454, "grad_norm": 20.34404325282992, "learning_rate": 7.5953505458151055e-06, "loss": 1.5033, "step": 34650 }, { "epoch": 4.01464, "grad_norm": 22.18051704733903, "learning_rate": 7.593966276356115e-06, "loss": 1.5212, "step": 34660 }, { "epoch": 4.01474, "grad_norm": 22.61676817186058, "learning_rate": 7.592581734802098e-06, "loss": 1.468, "step": 34670 }, { "epoch": 4.01484, "grad_norm": 19.102273161024847, "learning_rate": 7.5911969212982915e-06, "loss": 1.4427, "step": 34680 }, { "epoch": 4.01494, "grad_norm": 17.33130029539114, "learning_rate": 7.589811835989953e-06, "loss": 1.4063, "step": 34690 }, { "epoch": 4.01504, "grad_norm": 26.284518780720948, "learning_rate": 7.588426479022373e-06, "loss": 1.5113, "step": 34700 }, { "epoch": 4.01514, "grad_norm": 22.4813209223275, "learning_rate": 7.58704085054087e-06, "loss": 1.454, "step": 34710 }, { "epoch": 4.01524, "grad_norm": 22.754304288715453, "learning_rate": 7.585654950690786e-06, "loss": 1.5106, "step": 34720 }, { "epoch": 4.01534, "grad_norm": 25.701344933910605, "learning_rate": 7.5842687796175005e-06, "loss": 1.4546, "step": 34730 }, { "epoch": 4.01544, "grad_norm": 24.141443840754604, "learning_rate": 7.582882337466413e-06, "loss": 1.5606, "step": 34740 }, { "epoch": 4.01554, "grad_norm": 21.294902835762667, "learning_rate": 7.581495624382955e-06, "loss": 1.5052, "step": 34750 }, { "epoch": 4.01564, "grad_norm": 20.147870198918753, "learning_rate": 7.580108640512589e-06, "loss": 1.4956, "step": 34760 }, { "epoch": 4.01574, "grad_norm": 22.332720692050028, "learning_rate": 7.578721386000798e-06, "loss": 1.5381, "step": 34770 }, { "epoch": 4.01584, "grad_norm": 20.708682329623056, "learning_rate": 7.577333860993105e-06, "loss": 1.4996, "step": 34780 }, { "epoch": 4.01594, "grad_norm": 21.35775911382177, "learning_rate": 7.575946065635051e-06, "loss": 1.4711, "step": 34790 }, { "epoch": 4.01604, "grad_norm": 19.03937333495934, "learning_rate": 7.57455800007221e-06, "loss": 1.4447, "step": 34800 }, { "epoch": 4.01614, "grad_norm": 20.35737611896588, "learning_rate": 7.573169664450184e-06, "loss": 1.4288, "step": 34810 }, { "epoch": 4.01624, "grad_norm": 21.719843285453926, "learning_rate": 7.571781058914601e-06, "loss": 1.4254, "step": 34820 }, { "epoch": 4.01634, "grad_norm": 23.152173350414607, "learning_rate": 7.570392183611123e-06, "loss": 1.5148, "step": 34830 }, { "epoch": 4.01644, "grad_norm": 23.20777962916454, "learning_rate": 7.569003038685433e-06, "loss": 1.4365, "step": 34840 }, { "epoch": 4.01654, "grad_norm": 23.425424953747502, "learning_rate": 7.5676136242832475e-06, "loss": 1.5283, "step": 34850 }, { "epoch": 4.01664, "grad_norm": 22.887909927380136, "learning_rate": 7.56622394055031e-06, "loss": 1.5029, "step": 34860 }, { "epoch": 4.01674, "grad_norm": 21.67113263516007, "learning_rate": 7.56483398763239e-06, "loss": 1.4872, "step": 34870 }, { "epoch": 4.01684, "grad_norm": 22.06847228522114, "learning_rate": 7.563443765675292e-06, "loss": 1.4581, "step": 34880 }, { "epoch": 4.01694, "grad_norm": 21.026684890045686, "learning_rate": 7.562053274824836e-06, "loss": 1.4294, "step": 34890 }, { "epoch": 4.01704, "grad_norm": 23.46379420415784, "learning_rate": 7.5606625152268845e-06, "loss": 1.5113, "step": 34900 }, { "epoch": 4.01714, "grad_norm": 22.787451549620975, "learning_rate": 7.5592714870273196e-06, "loss": 1.5044, "step": 34910 }, { "epoch": 4.01724, "grad_norm": 22.06975022644403, "learning_rate": 7.557880190372054e-06, "loss": 1.4486, "step": 34920 }, { "epoch": 4.01734, "grad_norm": 22.735426093574873, "learning_rate": 7.5564886254070266e-06, "loss": 1.4778, "step": 34930 }, { "epoch": 4.01744, "grad_norm": 23.725380570739116, "learning_rate": 7.555096792278208e-06, "loss": 1.4862, "step": 34940 }, { "epoch": 4.01754, "grad_norm": 21.450975701766556, "learning_rate": 7.553704691131594e-06, "loss": 1.4183, "step": 34950 }, { "epoch": 4.01764, "grad_norm": 22.10738810333996, "learning_rate": 7.552312322113211e-06, "loss": 1.4254, "step": 34960 }, { "epoch": 4.01774, "grad_norm": 23.56725970071547, "learning_rate": 7.550919685369112e-06, "loss": 1.4544, "step": 34970 }, { "epoch": 4.01784, "grad_norm": 23.11850104108475, "learning_rate": 7.5495267810453755e-06, "loss": 1.4365, "step": 34980 }, { "epoch": 4.01794, "grad_norm": 20.22355365509863, "learning_rate": 7.548133609288113e-06, "loss": 1.4637, "step": 34990 }, { "epoch": 4.01804, "grad_norm": 21.38682401322333, "learning_rate": 7.546740170243461e-06, "loss": 1.4136, "step": 35000 }, { "epoch": 4.01814, "grad_norm": 24.22806927461363, "learning_rate": 7.545346464057584e-06, "loss": 1.4871, "step": 35010 }, { "epoch": 4.01824, "grad_norm": 22.12145920218103, "learning_rate": 7.543952490876678e-06, "loss": 1.4231, "step": 35020 }, { "epoch": 4.01834, "grad_norm": 23.86191936646446, "learning_rate": 7.5425582508469605e-06, "loss": 1.52, "step": 35030 }, { "epoch": 4.01844, "grad_norm": 20.284516890156624, "learning_rate": 7.541163744114684e-06, "loss": 1.4972, "step": 35040 }, { "epoch": 4.01854, "grad_norm": 20.345586609106636, "learning_rate": 7.539768970826125e-06, "loss": 1.4173, "step": 35050 }, { "epoch": 4.01864, "grad_norm": 21.22599151100371, "learning_rate": 7.538373931127587e-06, "loss": 1.4688, "step": 35060 }, { "epoch": 4.01874, "grad_norm": 23.4006160380967, "learning_rate": 7.536978625165407e-06, "loss": 1.4467, "step": 35070 }, { "epoch": 4.01884, "grad_norm": 22.407833445984316, "learning_rate": 7.535583053085942e-06, "loss": 1.4686, "step": 35080 }, { "epoch": 4.01894, "grad_norm": 23.539110528426416, "learning_rate": 7.534187215035582e-06, "loss": 1.4949, "step": 35090 }, { "epoch": 4.01904, "grad_norm": 23.765506056163797, "learning_rate": 7.532791111160746e-06, "loss": 1.4713, "step": 35100 }, { "epoch": 4.01914, "grad_norm": 21.86652057249665, "learning_rate": 7.531394741607878e-06, "loss": 1.4133, "step": 35110 }, { "epoch": 4.01924, "grad_norm": 23.292965823435175, "learning_rate": 7.529998106523448e-06, "loss": 1.4578, "step": 35120 }, { "epoch": 4.01934, "grad_norm": 22.73569933451098, "learning_rate": 7.528601206053961e-06, "loss": 1.4968, "step": 35130 }, { "epoch": 4.01944, "grad_norm": 21.53175549127423, "learning_rate": 7.527204040345941e-06, "loss": 1.4669, "step": 35140 }, { "epoch": 4.01954, "grad_norm": 24.483161767999455, "learning_rate": 7.525806609545948e-06, "loss": 1.517, "step": 35150 }, { "epoch": 4.01964, "grad_norm": 21.072093761334376, "learning_rate": 7.524408913800565e-06, "loss": 1.4763, "step": 35160 }, { "epoch": 4.01974, "grad_norm": 22.987216209053397, "learning_rate": 7.523010953256401e-06, "loss": 1.3765, "step": 35170 }, { "epoch": 4.01984, "grad_norm": 21.311977251753547, "learning_rate": 7.521612728060098e-06, "loss": 1.435, "step": 35180 }, { "epoch": 4.01994, "grad_norm": 23.972758728624264, "learning_rate": 7.520214238358324e-06, "loss": 1.441, "step": 35190 }, { "epoch": 4.02004, "grad_norm": 26.08595486204951, "learning_rate": 7.5188154842977725e-06, "loss": 1.4423, "step": 35200 }, { "epoch": 4.02014, "grad_norm": 20.394571155480136, "learning_rate": 7.517416466025167e-06, "loss": 1.4231, "step": 35210 }, { "epoch": 4.02024, "grad_norm": 21.80257482220723, "learning_rate": 7.516017183687256e-06, "loss": 1.447, "step": 35220 }, { "epoch": 4.02034, "grad_norm": 21.600094499546877, "learning_rate": 7.5146176374308225e-06, "loss": 1.3932, "step": 35230 }, { "epoch": 4.02044, "grad_norm": 20.047395703451567, "learning_rate": 7.513217827402669e-06, "loss": 1.4783, "step": 35240 }, { "epoch": 4.02054, "grad_norm": 22.724065203724876, "learning_rate": 7.5118177537496285e-06, "loss": 1.4664, "step": 35250 }, { "epoch": 4.02064, "grad_norm": 20.66326634619467, "learning_rate": 7.510417416618564e-06, "loss": 1.4442, "step": 35260 }, { "epoch": 4.02074, "grad_norm": 20.305920417986826, "learning_rate": 7.5090168161563644e-06, "loss": 1.432, "step": 35270 }, { "epoch": 4.02084, "grad_norm": 22.826703358755754, "learning_rate": 7.5076159525099445e-06, "loss": 1.4383, "step": 35280 }, { "epoch": 4.02094, "grad_norm": 22.860999245030875, "learning_rate": 7.506214825826249e-06, "loss": 1.4628, "step": 35290 }, { "epoch": 4.02104, "grad_norm": 24.02384132452627, "learning_rate": 7.50481343625225e-06, "loss": 1.4929, "step": 35300 }, { "epoch": 4.02114, "grad_norm": 22.14591075062157, "learning_rate": 7.5034117839349474e-06, "loss": 1.4784, "step": 35310 }, { "epoch": 4.02124, "grad_norm": 23.006035832991852, "learning_rate": 7.5020098690213686e-06, "loss": 1.4489, "step": 35320 }, { "epoch": 4.02134, "grad_norm": 20.57008486110147, "learning_rate": 7.500607691658565e-06, "loss": 1.4426, "step": 35330 }, { "epoch": 4.02144, "grad_norm": 22.71490870013227, "learning_rate": 7.499205251993621e-06, "loss": 1.4978, "step": 35340 }, { "epoch": 4.02154, "grad_norm": 20.51991244288622, "learning_rate": 7.497802550173645e-06, "loss": 1.4897, "step": 35350 }, { "epoch": 4.02164, "grad_norm": 19.556497374204756, "learning_rate": 7.496399586345773e-06, "loss": 1.4894, "step": 35360 }, { "epoch": 4.02174, "grad_norm": 21.306665680293232, "learning_rate": 7.494996360657172e-06, "loss": 1.5798, "step": 35370 }, { "epoch": 4.02184, "grad_norm": 20.20483471213513, "learning_rate": 7.493592873255032e-06, "loss": 1.5069, "step": 35380 }, { "epoch": 4.02194, "grad_norm": 20.71285364798382, "learning_rate": 7.492189124286573e-06, "loss": 1.4358, "step": 35390 }, { "epoch": 4.02204, "grad_norm": 20.404312600253537, "learning_rate": 7.490785113899041e-06, "loss": 1.4475, "step": 35400 }, { "epoch": 4.02214, "grad_norm": 22.565528498096548, "learning_rate": 7.48938084223971e-06, "loss": 1.5003, "step": 35410 }, { "epoch": 4.02224, "grad_norm": 24.318245145170447, "learning_rate": 7.487976309455883e-06, "loss": 1.5187, "step": 35420 }, { "epoch": 4.02234, "grad_norm": 23.015386225291255, "learning_rate": 7.486571515694888e-06, "loss": 1.5088, "step": 35430 }, { "epoch": 4.02244, "grad_norm": 21.222352745274893, "learning_rate": 7.485166461104083e-06, "loss": 1.4955, "step": 35440 }, { "epoch": 4.02254, "grad_norm": 21.4555207459489, "learning_rate": 7.483761145830848e-06, "loss": 1.5958, "step": 35450 }, { "epoch": 4.02264, "grad_norm": 24.61837947557202, "learning_rate": 7.482355570022597e-06, "loss": 1.506, "step": 35460 }, { "epoch": 4.02274, "grad_norm": 21.109693506485893, "learning_rate": 7.480949733826768e-06, "loss": 1.5032, "step": 35470 }, { "epoch": 4.02284, "grad_norm": 23.842243131028045, "learning_rate": 7.479543637390826e-06, "loss": 1.5483, "step": 35480 }, { "epoch": 4.02294, "grad_norm": 22.600051548866027, "learning_rate": 7.478137280862265e-06, "loss": 1.4558, "step": 35490 }, { "epoch": 4.02304, "grad_norm": 25.62480028619267, "learning_rate": 7.476730664388605e-06, "loss": 1.4988, "step": 35500 }, { "epoch": 4.02314, "grad_norm": 21.945505073186194, "learning_rate": 7.4753237881173925e-06, "loss": 1.5032, "step": 35510 }, { "epoch": 4.02324, "grad_norm": 21.29433939227313, "learning_rate": 7.473916652196203e-06, "loss": 1.44, "step": 35520 }, { "epoch": 4.02334, "grad_norm": 20.73689695040441, "learning_rate": 7.472509256772638e-06, "loss": 1.5178, "step": 35530 }, { "epoch": 4.02344, "grad_norm": 22.04869854898007, "learning_rate": 7.4711016019943274e-06, "loss": 1.5579, "step": 35540 }, { "epoch": 4.02354, "grad_norm": 22.204032703017116, "learning_rate": 7.4696936880089275e-06, "loss": 1.5343, "step": 35550 }, { "epoch": 4.02364, "grad_norm": 21.574709845473617, "learning_rate": 7.468285514964123e-06, "loss": 1.5762, "step": 35560 }, { "epoch": 4.02374, "grad_norm": 23.81814447050042, "learning_rate": 7.466877083007622e-06, "loss": 1.553, "step": 35570 }, { "epoch": 4.02384, "grad_norm": 23.85151150263337, "learning_rate": 7.465468392287165e-06, "loss": 1.5122, "step": 35580 }, { "epoch": 4.02394, "grad_norm": 23.6746629204186, "learning_rate": 7.464059442950515e-06, "loss": 1.5808, "step": 35590 }, { "epoch": 4.02404, "grad_norm": 22.189009266715672, "learning_rate": 7.4626502351454675e-06, "loss": 1.4858, "step": 35600 }, { "epoch": 4.02414, "grad_norm": 18.689697639048365, "learning_rate": 7.46124076901984e-06, "loss": 1.4259, "step": 35610 }, { "epoch": 4.02424, "grad_norm": 24.4115266361746, "learning_rate": 7.459831044721477e-06, "loss": 1.5026, "step": 35620 }, { "epoch": 4.02434, "grad_norm": 21.583415573704382, "learning_rate": 7.458421062398255e-06, "loss": 1.4111, "step": 35630 }, { "epoch": 4.02444, "grad_norm": 21.37435798379738, "learning_rate": 7.457010822198073e-06, "loss": 1.4752, "step": 35640 }, { "epoch": 4.02454, "grad_norm": 25.331351283474575, "learning_rate": 7.455600324268859e-06, "loss": 1.5179, "step": 35650 }, { "epoch": 4.02464, "grad_norm": 24.11266827148917, "learning_rate": 7.454189568758567e-06, "loss": 1.4759, "step": 35660 }, { "epoch": 4.02474, "grad_norm": 24.941002246656286, "learning_rate": 7.452778555815181e-06, "loss": 1.4964, "step": 35670 }, { "epoch": 4.02484, "grad_norm": 23.3351360475685, "learning_rate": 7.451367285586708e-06, "loss": 1.4711, "step": 35680 }, { "epoch": 4.02494, "grad_norm": 21.808252919551965, "learning_rate": 7.449955758221184e-06, "loss": 1.4805, "step": 35690 }, { "epoch": 4.02504, "grad_norm": 22.99328636864028, "learning_rate": 7.448543973866671e-06, "loss": 1.405, "step": 35700 }, { "epoch": 4.02514, "grad_norm": 21.45785422719127, "learning_rate": 7.447131932671259e-06, "loss": 1.447, "step": 35710 }, { "epoch": 4.02524, "grad_norm": 26.25654959527657, "learning_rate": 7.445719634783066e-06, "loss": 1.4154, "step": 35720 }, { "epoch": 4.02534, "grad_norm": 23.916662111074444, "learning_rate": 7.444307080350232e-06, "loss": 1.4896, "step": 35730 }, { "epoch": 4.02544, "grad_norm": 21.980905409220615, "learning_rate": 7.442894269520932e-06, "loss": 1.493, "step": 35740 }, { "epoch": 4.02554, "grad_norm": 23.62785616422147, "learning_rate": 7.441481202443359e-06, "loss": 1.4165, "step": 35750 }, { "epoch": 4.02564, "grad_norm": 20.547421121104392, "learning_rate": 7.44006787926574e-06, "loss": 1.4572, "step": 35760 }, { "epoch": 4.02574, "grad_norm": 20.84337835958458, "learning_rate": 7.438654300136325e-06, "loss": 1.4071, "step": 35770 }, { "epoch": 4.02584, "grad_norm": 22.22730554768144, "learning_rate": 7.437240465203392e-06, "loss": 1.4743, "step": 35780 }, { "epoch": 4.02594, "grad_norm": 22.268652461453264, "learning_rate": 7.435826374615246e-06, "loss": 1.4351, "step": 35790 }, { "epoch": 4.02604, "grad_norm": 24.919720705302982, "learning_rate": 7.434412028520219e-06, "loss": 1.4019, "step": 35800 }, { "epoch": 4.02614, "grad_norm": 21.489835130905007, "learning_rate": 7.432997427066668e-06, "loss": 1.4628, "step": 35810 }, { "epoch": 4.02624, "grad_norm": 19.242348867659885, "learning_rate": 7.431582570402979e-06, "loss": 1.4166, "step": 35820 }, { "epoch": 4.02634, "grad_norm": 23.230847489048106, "learning_rate": 7.430167458677563e-06, "loss": 1.4405, "step": 35830 }, { "epoch": 4.02644, "grad_norm": 20.580298778445382, "learning_rate": 7.428752092038859e-06, "loss": 1.4431, "step": 35840 }, { "epoch": 4.02654, "grad_norm": 21.61546207799083, "learning_rate": 7.427336470635334e-06, "loss": 1.4495, "step": 35850 }, { "epoch": 4.02664, "grad_norm": 22.46406788331943, "learning_rate": 7.425920594615479e-06, "loss": 1.472, "step": 35860 }, { "epoch": 4.02674, "grad_norm": 21.77639734260386, "learning_rate": 7.42450446412781e-06, "loss": 1.4583, "step": 35870 }, { "epoch": 4.02684, "grad_norm": 20.05439598084323, "learning_rate": 7.4230880793208796e-06, "loss": 1.4375, "step": 35880 }, { "epoch": 4.02694, "grad_norm": 24.722989456488044, "learning_rate": 7.421671440343252e-06, "loss": 1.4858, "step": 35890 }, { "epoch": 4.02704, "grad_norm": 25.009446495686674, "learning_rate": 7.4202545473435315e-06, "loss": 1.4599, "step": 35900 }, { "epoch": 4.02714, "grad_norm": 22.41699505918231, "learning_rate": 7.4188374004703415e-06, "loss": 1.417, "step": 35910 }, { "epoch": 4.02724, "grad_norm": 22.240116421073825, "learning_rate": 7.417419999872333e-06, "loss": 1.4457, "step": 35920 }, { "epoch": 4.02734, "grad_norm": 22.392827716558134, "learning_rate": 7.416002345698187e-06, "loss": 1.4715, "step": 35930 }, { "epoch": 4.02744, "grad_norm": 21.544302508184156, "learning_rate": 7.414584438096608e-06, "loss": 1.4233, "step": 35940 }, { "epoch": 4.02754, "grad_norm": 21.82465750343618, "learning_rate": 7.413166277216328e-06, "loss": 1.3711, "step": 35950 }, { "epoch": 4.02764, "grad_norm": 21.631979887815216, "learning_rate": 7.411747863206105e-06, "loss": 1.443, "step": 35960 }, { "epoch": 4.02774, "grad_norm": 21.114261173518205, "learning_rate": 7.410329196214725e-06, "loss": 1.4147, "step": 35970 }, { "epoch": 4.02784, "grad_norm": 22.063616121923403, "learning_rate": 7.408910276391e-06, "loss": 1.5019, "step": 35980 }, { "epoch": 4.02794, "grad_norm": 22.378928545398512, "learning_rate": 7.407491103883765e-06, "loss": 1.4103, "step": 35990 }, { "epoch": 4.02804, "grad_norm": 25.393928327735992, "learning_rate": 7.4060716788418865e-06, "loss": 1.3767, "step": 36000 }, { "epoch": 4.02814, "grad_norm": 22.286143536100923, "learning_rate": 7.404652001414256e-06, "loss": 1.4311, "step": 36010 }, { "epoch": 4.02824, "grad_norm": 22.617866556310812, "learning_rate": 7.403232071749793e-06, "loss": 1.3741, "step": 36020 }, { "epoch": 4.02834, "grad_norm": 21.957353620102907, "learning_rate": 7.4018118899974375e-06, "loss": 1.4271, "step": 36030 }, { "epoch": 4.02844, "grad_norm": 23.47811679118483, "learning_rate": 7.400391456306163e-06, "loss": 1.4101, "step": 36040 }, { "epoch": 4.02854, "grad_norm": 23.6214350822155, "learning_rate": 7.398970770824965e-06, "loss": 1.5026, "step": 36050 }, { "epoch": 4.02864, "grad_norm": 20.09797047638682, "learning_rate": 7.397549833702868e-06, "loss": 1.3896, "step": 36060 }, { "epoch": 4.02874, "grad_norm": 19.021129121506018, "learning_rate": 7.396128645088918e-06, "loss": 1.4284, "step": 36070 }, { "epoch": 4.02884, "grad_norm": 21.843082477122575, "learning_rate": 7.394707205132197e-06, "loss": 1.4526, "step": 36080 }, { "epoch": 4.02894, "grad_norm": 21.213474324035342, "learning_rate": 7.393285513981803e-06, "loss": 1.5025, "step": 36090 }, { "epoch": 4.02904, "grad_norm": 21.64268158955494, "learning_rate": 7.391863571786866e-06, "loss": 1.3852, "step": 36100 }, { "epoch": 4.02914, "grad_norm": 20.202497342756256, "learning_rate": 7.39044137869654e-06, "loss": 1.4283, "step": 36110 }, { "epoch": 4.02924, "grad_norm": 20.439344790674333, "learning_rate": 7.38901893486001e-06, "loss": 1.466, "step": 36120 }, { "epoch": 4.02934, "grad_norm": 21.415854285112236, "learning_rate": 7.38759624042648e-06, "loss": 1.4255, "step": 36130 }, { "epoch": 4.02944, "grad_norm": 24.69910558488707, "learning_rate": 7.386173295545187e-06, "loss": 1.496, "step": 36140 }, { "epoch": 4.02954, "grad_norm": 21.880686086677866, "learning_rate": 7.384750100365389e-06, "loss": 1.4426, "step": 36150 }, { "epoch": 4.02964, "grad_norm": 23.699645716346, "learning_rate": 7.383326655036373e-06, "loss": 1.4049, "step": 36160 }, { "epoch": 4.02974, "grad_norm": 22.420008919912704, "learning_rate": 7.381902959707454e-06, "loss": 1.4154, "step": 36170 }, { "epoch": 4.02984, "grad_norm": 21.64464707157544, "learning_rate": 7.380479014527967e-06, "loss": 1.4768, "step": 36180 }, { "epoch": 4.02994, "grad_norm": 21.795283909816284, "learning_rate": 7.37905481964728e-06, "loss": 1.3862, "step": 36190 }, { "epoch": 4.03004, "grad_norm": 23.47361864040612, "learning_rate": 7.377630375214785e-06, "loss": 1.352, "step": 36200 }, { "epoch": 4.03014, "grad_norm": 21.204243424981062, "learning_rate": 7.376205681379899e-06, "loss": 1.3399, "step": 36210 }, { "epoch": 4.03024, "grad_norm": 22.142277308688087, "learning_rate": 7.374780738292064e-06, "loss": 1.3663, "step": 36220 }, { "epoch": 4.03034, "grad_norm": 23.215602540797303, "learning_rate": 7.373355546100752e-06, "loss": 1.3852, "step": 36230 }, { "epoch": 4.03044, "grad_norm": 23.108325423174616, "learning_rate": 7.371930104955458e-06, "loss": 1.3362, "step": 36240 }, { "epoch": 4.03054, "grad_norm": 23.01513142132241, "learning_rate": 7.3705044150057035e-06, "loss": 1.4119, "step": 36250 }, { "epoch": 4.03064, "grad_norm": 22.833367176893855, "learning_rate": 7.36907847640104e-06, "loss": 1.4363, "step": 36260 }, { "epoch": 4.03074, "grad_norm": 22.214446739139287, "learning_rate": 7.367652289291039e-06, "loss": 1.409, "step": 36270 }, { "epoch": 4.03084, "grad_norm": 23.844827904110268, "learning_rate": 7.366225853825299e-06, "loss": 1.412, "step": 36280 }, { "epoch": 4.03094, "grad_norm": 23.851248557231816, "learning_rate": 7.364799170153451e-06, "loss": 1.4057, "step": 36290 }, { "epoch": 4.03104, "grad_norm": 21.54590575388939, "learning_rate": 7.363372238425145e-06, "loss": 1.3391, "step": 36300 }, { "epoch": 4.03114, "grad_norm": 22.562128054761796, "learning_rate": 7.361945058790061e-06, "loss": 1.3385, "step": 36310 }, { "epoch": 4.03124, "grad_norm": 22.367167179911235, "learning_rate": 7.360517631397901e-06, "loss": 1.391, "step": 36320 }, { "epoch": 4.03134, "grad_norm": 20.48055789378331, "learning_rate": 7.359089956398398e-06, "loss": 1.3707, "step": 36330 }, { "epoch": 4.03144, "grad_norm": 22.27614607672195, "learning_rate": 7.3576620339413066e-06, "loss": 1.3205, "step": 36340 }, { "epoch": 4.03154, "grad_norm": 21.72340416784781, "learning_rate": 7.35623386417641e-06, "loss": 1.2848, "step": 36350 }, { "epoch": 4.03164, "grad_norm": 21.908858974480285, "learning_rate": 7.354805447253517e-06, "loss": 1.2872, "step": 36360 }, { "epoch": 4.03174, "grad_norm": 21.49957182725432, "learning_rate": 7.353376783322461e-06, "loss": 1.3108, "step": 36370 }, { "epoch": 4.03184, "grad_norm": 24.415307149071392, "learning_rate": 7.351947872533104e-06, "loss": 1.3208, "step": 36380 }, { "epoch": 4.03194, "grad_norm": 22.889294735411408, "learning_rate": 7.350518715035331e-06, "loss": 1.3146, "step": 36390 }, { "epoch": 4.03204, "grad_norm": 25.018242351787187, "learning_rate": 7.3490893109790516e-06, "loss": 1.4047, "step": 36400 }, { "epoch": 4.03214, "grad_norm": 24.945858432249413, "learning_rate": 7.347659660514208e-06, "loss": 1.3677, "step": 36410 }, { "epoch": 4.03224, "grad_norm": 20.23264048965439, "learning_rate": 7.346229763790762e-06, "loss": 1.4605, "step": 36420 }, { "epoch": 4.03234, "grad_norm": 21.358974213186254, "learning_rate": 7.344799620958703e-06, "loss": 1.3636, "step": 36430 }, { "epoch": 4.03244, "grad_norm": 24.490594531931126, "learning_rate": 7.343369232168045e-06, "loss": 1.3613, "step": 36440 }, { "epoch": 4.03254, "grad_norm": 20.898115989771473, "learning_rate": 7.341938597568831e-06, "loss": 1.3462, "step": 36450 }, { "epoch": 4.03264, "grad_norm": 20.32137439457219, "learning_rate": 7.340507717311127e-06, "loss": 1.3093, "step": 36460 }, { "epoch": 4.03274, "grad_norm": 22.03615772570412, "learning_rate": 7.339076591545027e-06, "loss": 1.3329, "step": 36470 }, { "epoch": 4.03284, "grad_norm": 22.4854819391017, "learning_rate": 7.3376452204206476e-06, "loss": 1.3231, "step": 36480 }, { "epoch": 4.03294, "grad_norm": 20.773399917862363, "learning_rate": 7.336213604088135e-06, "loss": 1.3766, "step": 36490 }, { "epoch": 4.03304, "grad_norm": 20.43403659584754, "learning_rate": 7.3347817426976575e-06, "loss": 1.3934, "step": 36500 }, { "epoch": 4.03314, "grad_norm": 20.93407040617355, "learning_rate": 7.333349636399411e-06, "loss": 1.3644, "step": 36510 }, { "epoch": 4.03324, "grad_norm": 22.658507778773792, "learning_rate": 7.3319172853436175e-06, "loss": 1.3926, "step": 36520 }, { "epoch": 4.03334, "grad_norm": 20.339223527912726, "learning_rate": 7.330484689680523e-06, "loss": 1.3963, "step": 36530 }, { "epoch": 4.03344, "grad_norm": 21.399690517643844, "learning_rate": 7.3290518495604e-06, "loss": 1.328, "step": 36540 }, { "epoch": 4.03354, "grad_norm": 21.98058315242628, "learning_rate": 7.3276187651335484e-06, "loss": 1.3569, "step": 36550 }, { "epoch": 4.03364, "grad_norm": 21.262367831672115, "learning_rate": 7.326185436550292e-06, "loss": 1.3044, "step": 36560 }, { "epoch": 4.03374, "grad_norm": 21.925356609359717, "learning_rate": 7.324751863960979e-06, "loss": 1.3737, "step": 36570 }, { "epoch": 4.03384, "grad_norm": 24.763917799153432, "learning_rate": 7.3233180475159834e-06, "loss": 1.2969, "step": 36580 }, { "epoch": 4.03394, "grad_norm": 23.22287630567893, "learning_rate": 7.32188398736571e-06, "loss": 1.4199, "step": 36590 }, { "epoch": 4.03404, "grad_norm": 23.37637887674696, "learning_rate": 7.320449683660581e-06, "loss": 1.4258, "step": 36600 }, { "epoch": 4.03414, "grad_norm": 23.344839419699003, "learning_rate": 7.319015136551051e-06, "loss": 1.349, "step": 36610 }, { "epoch": 4.03424, "grad_norm": 22.472992240619217, "learning_rate": 7.317580346187594e-06, "loss": 1.3907, "step": 36620 }, { "epoch": 4.03434, "grad_norm": 22.473678194109585, "learning_rate": 7.316145312720716e-06, "loss": 1.3436, "step": 36630 }, { "epoch": 4.03444, "grad_norm": 21.479081882939777, "learning_rate": 7.314710036300944e-06, "loss": 1.3265, "step": 36640 }, { "epoch": 4.03454, "grad_norm": 21.672015824215624, "learning_rate": 7.313274517078832e-06, "loss": 1.3211, "step": 36650 }, { "epoch": 4.03464, "grad_norm": 23.365198738527383, "learning_rate": 7.31183875520496e-06, "loss": 1.3258, "step": 36660 }, { "epoch": 4.03474, "grad_norm": 24.371148479532156, "learning_rate": 7.310402750829932e-06, "loss": 1.3188, "step": 36670 }, { "epoch": 4.03484, "grad_norm": 23.473933925364268, "learning_rate": 7.308966504104378e-06, "loss": 1.4114, "step": 36680 }, { "epoch": 4.03494, "grad_norm": 24.404565055782783, "learning_rate": 7.307530015178954e-06, "loss": 1.374, "step": 36690 }, { "epoch": 4.03504, "grad_norm": 22.833287636500163, "learning_rate": 7.306093284204341e-06, "loss": 1.2651, "step": 36700 }, { "epoch": 4.03514, "grad_norm": 22.71008363187591, "learning_rate": 7.304656311331245e-06, "loss": 1.3379, "step": 36710 }, { "epoch": 4.03524, "grad_norm": 20.296912040297865, "learning_rate": 7.3032190967103985e-06, "loss": 1.3095, "step": 36720 }, { "epoch": 4.03534, "grad_norm": 22.025217942639053, "learning_rate": 7.301781640492558e-06, "loss": 1.2963, "step": 36730 }, { "epoch": 4.03544, "grad_norm": 20.500695843285943, "learning_rate": 7.300343942828507e-06, "loss": 1.3505, "step": 36740 }, { "epoch": 4.03554, "grad_norm": 26.2348589415802, "learning_rate": 7.298906003869053e-06, "loss": 1.3628, "step": 36750 }, { "epoch": 4.03564, "grad_norm": 21.23876055491755, "learning_rate": 7.297467823765027e-06, "loss": 1.2862, "step": 36760 }, { "epoch": 4.03574, "grad_norm": 23.623253957518795, "learning_rate": 7.2960294026672915e-06, "loss": 1.3272, "step": 36770 }, { "epoch": 4.03584, "grad_norm": 21.053033247834737, "learning_rate": 7.294590740726727e-06, "loss": 1.3416, "step": 36780 }, { "epoch": 4.03594, "grad_norm": 21.82613877836004, "learning_rate": 7.293151838094243e-06, "loss": 1.3521, "step": 36790 }, { "epoch": 4.03604, "grad_norm": 23.76756166801284, "learning_rate": 7.291712694920774e-06, "loss": 1.3612, "step": 36800 }, { "epoch": 4.03614, "grad_norm": 20.73197581721287, "learning_rate": 7.290273311357279e-06, "loss": 1.3268, "step": 36810 }, { "epoch": 4.03624, "grad_norm": 22.204302373612613, "learning_rate": 7.288833687554744e-06, "loss": 1.4211, "step": 36820 }, { "epoch": 4.03634, "grad_norm": 19.675494258233552, "learning_rate": 7.287393823664177e-06, "loss": 1.4011, "step": 36830 }, { "epoch": 4.03644, "grad_norm": 20.364916533775347, "learning_rate": 7.285953719836616e-06, "loss": 1.3294, "step": 36840 }, { "epoch": 4.03654, "grad_norm": 24.147011289093278, "learning_rate": 7.284513376223117e-06, "loss": 1.3179, "step": 36850 }, { "epoch": 4.03664, "grad_norm": 27.028751912042964, "learning_rate": 7.283072792974768e-06, "loss": 1.4197, "step": 36860 }, { "epoch": 4.03674, "grad_norm": 22.1096247643312, "learning_rate": 7.2816319702426805e-06, "loss": 1.3469, "step": 36870 }, { "epoch": 4.03684, "grad_norm": 22.297488724071645, "learning_rate": 7.280190908177988e-06, "loss": 1.3423, "step": 36880 }, { "epoch": 4.03694, "grad_norm": 23.986609897903904, "learning_rate": 7.278749606931851e-06, "loss": 1.3406, "step": 36890 }, { "epoch": 4.03704, "grad_norm": 23.080258667451027, "learning_rate": 7.277308066655456e-06, "loss": 1.3811, "step": 36900 }, { "epoch": 4.03714, "grad_norm": 23.94654858899779, "learning_rate": 7.275866287500015e-06, "loss": 1.3763, "step": 36910 }, { "epoch": 4.03724, "grad_norm": 22.355971429420894, "learning_rate": 7.2744242696167645e-06, "loss": 1.4109, "step": 36920 }, { "epoch": 4.03734, "grad_norm": 19.23699588940993, "learning_rate": 7.2729820131569615e-06, "loss": 1.3752, "step": 36930 }, { "epoch": 4.03744, "grad_norm": 25.163712199492537, "learning_rate": 7.271539518271897e-06, "loss": 1.4129, "step": 36940 }, { "epoch": 4.03754, "grad_norm": 20.067189227371966, "learning_rate": 7.270096785112878e-06, "loss": 1.355, "step": 36950 }, { "epoch": 4.03764, "grad_norm": 25.269035614924025, "learning_rate": 7.268653813831242e-06, "loss": 1.3284, "step": 36960 }, { "epoch": 4.03774, "grad_norm": 22.061919970734372, "learning_rate": 7.267210604578352e-06, "loss": 1.397, "step": 36970 }, { "epoch": 4.03784, "grad_norm": 22.51323614290972, "learning_rate": 7.26576715750559e-06, "loss": 1.3157, "step": 36980 }, { "epoch": 4.03794, "grad_norm": 21.19521590081352, "learning_rate": 7.264323472764369e-06, "loss": 1.3404, "step": 36990 }, { "epoch": 4.03804, "grad_norm": 23.04762941980875, "learning_rate": 7.262879550506124e-06, "loss": 1.304, "step": 37000 }, { "epoch": 4.03814, "grad_norm": 23.19609960591652, "learning_rate": 7.261435390882317e-06, "loss": 1.3349, "step": 37010 }, { "epoch": 4.03824, "grad_norm": 22.719755540385748, "learning_rate": 7.259990994044435e-06, "loss": 1.4209, "step": 37020 }, { "epoch": 4.03834, "grad_norm": 22.301058811766893, "learning_rate": 7.258546360143985e-06, "loss": 1.3347, "step": 37030 }, { "epoch": 4.03844, "grad_norm": 23.98797744240378, "learning_rate": 7.2571014893325055e-06, "loss": 1.3237, "step": 37040 }, { "epoch": 4.03854, "grad_norm": 21.75485512994387, "learning_rate": 7.2556563817615545e-06, "loss": 1.322, "step": 37050 }, { "epoch": 4.03864, "grad_norm": 24.273317904871732, "learning_rate": 7.254211037582717e-06, "loss": 1.3142, "step": 37060 }, { "epoch": 4.03874, "grad_norm": 23.45336595036661, "learning_rate": 7.252765456947605e-06, "loss": 1.3749, "step": 37070 }, { "epoch": 4.03884, "grad_norm": 23.387505662984065, "learning_rate": 7.25131964000785e-06, "loss": 1.3527, "step": 37080 }, { "epoch": 4.03894, "grad_norm": 20.494048246006212, "learning_rate": 7.249873586915115e-06, "loss": 1.3011, "step": 37090 }, { "epoch": 4.03904, "grad_norm": 22.66147940417146, "learning_rate": 7.248427297821083e-06, "loss": 1.42, "step": 37100 }, { "epoch": 4.03914, "grad_norm": 23.008628702597175, "learning_rate": 7.246980772877463e-06, "loss": 1.3215, "step": 37110 }, { "epoch": 4.03924, "grad_norm": 22.198690677715977, "learning_rate": 7.245534012235988e-06, "loss": 1.3985, "step": 37120 }, { "epoch": 4.03934, "grad_norm": 22.284741137063, "learning_rate": 7.2440870160484165e-06, "loss": 1.3049, "step": 37130 }, { "epoch": 4.03944, "grad_norm": 24.056352016439398, "learning_rate": 7.242639784466532e-06, "loss": 1.2856, "step": 37140 }, { "epoch": 4.03954, "grad_norm": 24.195486044565396, "learning_rate": 7.2411923176421435e-06, "loss": 1.4184, "step": 37150 }, { "epoch": 4.03964, "grad_norm": 23.14254341145783, "learning_rate": 7.239744615727081e-06, "loss": 1.3172, "step": 37160 }, { "epoch": 4.03974, "grad_norm": 31.252019205406718, "learning_rate": 7.238296678873205e-06, "loss": 1.3446, "step": 37170 }, { "epoch": 4.03984, "grad_norm": 23.237051943088172, "learning_rate": 7.236848507232394e-06, "loss": 1.3277, "step": 37180 }, { "epoch": 4.03994, "grad_norm": 19.425269347973334, "learning_rate": 7.235400100956558e-06, "loss": 1.389, "step": 37190 }, { "epoch": 4.04004, "grad_norm": 22.85632397101856, "learning_rate": 7.233951460197626e-06, "loss": 1.3711, "step": 37200 }, { "epoch": 4.04014, "grad_norm": 22.40978914793788, "learning_rate": 7.232502585107552e-06, "loss": 1.4001, "step": 37210 }, { "epoch": 4.04024, "grad_norm": 24.115668392407265, "learning_rate": 7.231053475838321e-06, "loss": 1.3914, "step": 37220 }, { "epoch": 4.04034, "grad_norm": 23.960727046256885, "learning_rate": 7.229604132541933e-06, "loss": 1.3804, "step": 37230 }, { "epoch": 4.04044, "grad_norm": 23.66947962069396, "learning_rate": 7.2281545553704195e-06, "loss": 1.4576, "step": 37240 }, { "epoch": 4.04054, "grad_norm": 18.967860975249966, "learning_rate": 7.226704744475834e-06, "loss": 1.2908, "step": 37250 }, { "epoch": 4.04064, "grad_norm": 23.09626991520196, "learning_rate": 7.2252547000102555e-06, "loss": 1.3764, "step": 37260 }, { "epoch": 4.0407399999999996, "grad_norm": 23.638454700524175, "learning_rate": 7.223804422125789e-06, "loss": 1.4534, "step": 37270 }, { "epoch": 4.04084, "grad_norm": 21.365504784452714, "learning_rate": 7.2223539109745575e-06, "loss": 1.3658, "step": 37280 }, { "epoch": 4.04094, "grad_norm": 21.208520652384408, "learning_rate": 7.220903166708715e-06, "loss": 1.428, "step": 37290 }, { "epoch": 4.04104, "grad_norm": 22.656154027724227, "learning_rate": 7.219452189480438e-06, "loss": 1.4177, "step": 37300 }, { "epoch": 4.04114, "grad_norm": 20.17332187925439, "learning_rate": 7.218000979441927e-06, "loss": 1.3452, "step": 37310 }, { "epoch": 4.04124, "grad_norm": 21.279284096329345, "learning_rate": 7.216549536745409e-06, "loss": 1.4655, "step": 37320 }, { "epoch": 4.04134, "grad_norm": 23.796581512297607, "learning_rate": 7.215097861543129e-06, "loss": 1.3251, "step": 37330 }, { "epoch": 4.04144, "grad_norm": 23.154857572268757, "learning_rate": 7.213645953987365e-06, "loss": 1.3318, "step": 37340 }, { "epoch": 4.04154, "grad_norm": 19.600434320604336, "learning_rate": 7.212193814230415e-06, "loss": 1.2981, "step": 37350 }, { "epoch": 4.04164, "grad_norm": 20.8160808003329, "learning_rate": 7.2107414424246e-06, "loss": 1.3383, "step": 37360 }, { "epoch": 4.04174, "grad_norm": 24.002019181789013, "learning_rate": 7.209288838722269e-06, "loss": 1.3674, "step": 37370 }, { "epoch": 4.04184, "grad_norm": 24.04794575994305, "learning_rate": 7.207836003275792e-06, "loss": 1.324, "step": 37380 }, { "epoch": 4.04194, "grad_norm": 21.29292742292871, "learning_rate": 7.206382936237566e-06, "loss": 1.3389, "step": 37390 }, { "epoch": 4.04204, "grad_norm": 20.92966652071526, "learning_rate": 7.20492963776001e-06, "loss": 1.3534, "step": 37400 }, { "epoch": 4.04214, "grad_norm": 22.02776407388357, "learning_rate": 7.203476107995567e-06, "loss": 1.3384, "step": 37410 }, { "epoch": 4.04224, "grad_norm": 22.335341104796402, "learning_rate": 7.202022347096708e-06, "loss": 1.4101, "step": 37420 }, { "epoch": 4.04234, "grad_norm": 23.936525547692508, "learning_rate": 7.200568355215925e-06, "loss": 1.3322, "step": 37430 }, { "epoch": 4.04244, "grad_norm": 22.708827821202878, "learning_rate": 7.199114132505733e-06, "loss": 1.3573, "step": 37440 }, { "epoch": 4.04254, "grad_norm": 22.672371604154964, "learning_rate": 7.197659679118676e-06, "loss": 1.4569, "step": 37450 }, { "epoch": 4.04264, "grad_norm": 21.27168524993786, "learning_rate": 7.196204995207318e-06, "loss": 1.3693, "step": 37460 }, { "epoch": 4.04274, "grad_norm": 22.713256961021678, "learning_rate": 7.19475008092425e-06, "loss": 1.3656, "step": 37470 }, { "epoch": 4.04284, "grad_norm": 21.40041928351554, "learning_rate": 7.193294936422084e-06, "loss": 1.4032, "step": 37480 }, { "epoch": 4.04294, "grad_norm": 21.52729034186894, "learning_rate": 7.191839561853459e-06, "loss": 1.3851, "step": 37490 }, { "epoch": 4.04304, "grad_norm": 24.057849290648637, "learning_rate": 7.190383957371037e-06, "loss": 1.3366, "step": 37500 }, { "epoch": 4.04314, "grad_norm": 22.04848349035122, "learning_rate": 7.188928123127501e-06, "loss": 1.4134, "step": 37510 }, { "epoch": 4.04324, "grad_norm": 21.68963830888399, "learning_rate": 7.187472059275565e-06, "loss": 1.3963, "step": 37520 }, { "epoch": 4.04334, "grad_norm": 21.929635845767184, "learning_rate": 7.1860157659679645e-06, "loss": 1.3478, "step": 37530 }, { "epoch": 4.04344, "grad_norm": 22.65703162786545, "learning_rate": 7.184559243357454e-06, "loss": 1.4025, "step": 37540 }, { "epoch": 4.04354, "grad_norm": 20.518180324757594, "learning_rate": 7.183102491596818e-06, "loss": 1.429, "step": 37550 }, { "epoch": 4.04364, "grad_norm": 22.954417355233684, "learning_rate": 7.181645510838864e-06, "loss": 1.4075, "step": 37560 }, { "epoch": 4.04374, "grad_norm": 18.97607458694037, "learning_rate": 7.18018830123642e-06, "loss": 1.3819, "step": 37570 }, { "epoch": 4.04384, "grad_norm": 24.675928737160127, "learning_rate": 7.178730862942341e-06, "loss": 1.4232, "step": 37580 }, { "epoch": 4.04394, "grad_norm": 22.616848402834428, "learning_rate": 7.1772731961095074e-06, "loss": 1.3629, "step": 37590 }, { "epoch": 4.04404, "grad_norm": 23.677397464079274, "learning_rate": 7.17581530089082e-06, "loss": 1.3802, "step": 37600 }, { "epoch": 4.04414, "grad_norm": 23.194339520295824, "learning_rate": 7.174357177439206e-06, "loss": 1.336, "step": 37610 }, { "epoch": 4.04424, "grad_norm": 23.034932389158477, "learning_rate": 7.172898825907615e-06, "loss": 1.3385, "step": 37620 }, { "epoch": 4.04434, "grad_norm": 19.61938651527134, "learning_rate": 7.171440246449024e-06, "loss": 1.3273, "step": 37630 }, { "epoch": 4.04444, "grad_norm": 23.443268855824556, "learning_rate": 7.169981439216427e-06, "loss": 1.4039, "step": 37640 }, { "epoch": 4.04454, "grad_norm": 23.731645908205838, "learning_rate": 7.168522404362849e-06, "loss": 1.3614, "step": 37650 }, { "epoch": 4.04464, "grad_norm": 23.102604875502486, "learning_rate": 7.167063142041336e-06, "loss": 1.3632, "step": 37660 }, { "epoch": 4.04474, "grad_norm": 22.642929538520125, "learning_rate": 7.165603652404955e-06, "loss": 1.3677, "step": 37670 }, { "epoch": 4.04484, "grad_norm": 23.135153609493358, "learning_rate": 7.164143935606803e-06, "loss": 1.2997, "step": 37680 }, { "epoch": 4.04494, "grad_norm": 20.3158259393856, "learning_rate": 7.1626839917999965e-06, "loss": 1.2774, "step": 37690 }, { "epoch": 4.04504, "grad_norm": 20.43965194882944, "learning_rate": 7.161223821137676e-06, "loss": 1.339, "step": 37700 }, { "epoch": 4.04514, "grad_norm": 24.711450825669495, "learning_rate": 7.159763423773009e-06, "loss": 1.3632, "step": 37710 }, { "epoch": 4.04524, "grad_norm": 22.931477808994284, "learning_rate": 7.1583027998591805e-06, "loss": 1.3423, "step": 37720 }, { "epoch": 4.04534, "grad_norm": 21.34443887540902, "learning_rate": 7.156841949549407e-06, "loss": 1.3541, "step": 37730 }, { "epoch": 4.04544, "grad_norm": 24.247946390721687, "learning_rate": 7.155380872996921e-06, "loss": 1.3711, "step": 37740 }, { "epoch": 4.04554, "grad_norm": 21.43311401723463, "learning_rate": 7.1539195703549854e-06, "loss": 1.4356, "step": 37750 }, { "epoch": 4.04564, "grad_norm": 19.482514544321756, "learning_rate": 7.152458041776885e-06, "loss": 1.3651, "step": 37760 }, { "epoch": 4.04574, "grad_norm": 22.033876911058076, "learning_rate": 7.150996287415926e-06, "loss": 1.3688, "step": 37770 }, { "epoch": 4.04584, "grad_norm": 18.641846631040895, "learning_rate": 7.1495343074254375e-06, "loss": 1.3425, "step": 37780 }, { "epoch": 4.04594, "grad_norm": 21.376969416057214, "learning_rate": 7.1480721019587775e-06, "loss": 1.3739, "step": 37790 }, { "epoch": 4.04604, "grad_norm": 22.04997324005254, "learning_rate": 7.146609671169322e-06, "loss": 1.3828, "step": 37800 }, { "epoch": 4.04614, "grad_norm": 22.07981655276969, "learning_rate": 7.145147015210477e-06, "loss": 1.3103, "step": 37810 }, { "epoch": 4.04624, "grad_norm": 25.386694599194556, "learning_rate": 7.143684134235666e-06, "loss": 1.4111, "step": 37820 }, { "epoch": 4.04634, "grad_norm": 25.44103311574642, "learning_rate": 7.142221028398338e-06, "loss": 1.3707, "step": 37830 }, { "epoch": 4.04644, "grad_norm": 22.38229190447389, "learning_rate": 7.140757697851966e-06, "loss": 1.3542, "step": 37840 }, { "epoch": 4.04654, "grad_norm": 20.36275073555127, "learning_rate": 7.139294142750047e-06, "loss": 1.3109, "step": 37850 }, { "epoch": 4.04664, "grad_norm": 23.39284571987198, "learning_rate": 7.137830363246101e-06, "loss": 1.3911, "step": 37860 }, { "epoch": 4.04674, "grad_norm": 22.025729437027593, "learning_rate": 7.136366359493672e-06, "loss": 1.3239, "step": 37870 }, { "epoch": 4.04684, "grad_norm": 23.276291325778185, "learning_rate": 7.134902131646329e-06, "loss": 1.313, "step": 37880 }, { "epoch": 4.04694, "grad_norm": 23.03612457305497, "learning_rate": 7.1334376798576574e-06, "loss": 1.3208, "step": 37890 }, { "epoch": 4.04704, "grad_norm": 24.010187688364432, "learning_rate": 7.131973004281276e-06, "loss": 1.3505, "step": 37900 }, { "epoch": 4.04714, "grad_norm": 21.16358437249349, "learning_rate": 7.130508105070824e-06, "loss": 1.3566, "step": 37910 }, { "epoch": 4.04724, "grad_norm": 22.42258898903194, "learning_rate": 7.129042982379957e-06, "loss": 1.2669, "step": 37920 }, { "epoch": 4.04734, "grad_norm": 19.98417510647732, "learning_rate": 7.127577636362364e-06, "loss": 1.3256, "step": 37930 }, { "epoch": 4.04744, "grad_norm": 19.99134942825241, "learning_rate": 7.126112067171749e-06, "loss": 1.3181, "step": 37940 }, { "epoch": 4.04754, "grad_norm": 22.864260532423675, "learning_rate": 7.124646274961847e-06, "loss": 1.3385, "step": 37950 }, { "epoch": 4.04764, "grad_norm": 20.56761662280175, "learning_rate": 7.123180259886411e-06, "loss": 1.3532, "step": 37960 }, { "epoch": 4.04774, "grad_norm": 23.439329415837435, "learning_rate": 7.12171402209922e-06, "loss": 1.385, "step": 37970 }, { "epoch": 4.04784, "grad_norm": 20.492029692225632, "learning_rate": 7.120247561754075e-06, "loss": 1.3637, "step": 37980 }, { "epoch": 4.04794, "grad_norm": 20.382225632572908, "learning_rate": 7.118780879004803e-06, "loss": 1.3754, "step": 37990 }, { "epoch": 4.04804, "grad_norm": 23.047577976654534, "learning_rate": 7.117313974005249e-06, "loss": 1.3272, "step": 38000 }, { "epoch": 4.04814, "grad_norm": 20.895425908253447, "learning_rate": 7.115846846909287e-06, "loss": 1.2952, "step": 38010 }, { "epoch": 4.04824, "grad_norm": 22.79384272138744, "learning_rate": 7.114379497870808e-06, "loss": 1.33, "step": 38020 }, { "epoch": 4.04834, "grad_norm": 24.173441946756668, "learning_rate": 7.1129119270437356e-06, "loss": 1.3952, "step": 38030 }, { "epoch": 4.04844, "grad_norm": 22.511757754539907, "learning_rate": 7.111444134582006e-06, "loss": 1.2947, "step": 38040 }, { "epoch": 4.04854, "grad_norm": 23.461417974429327, "learning_rate": 7.109976120639586e-06, "loss": 1.312, "step": 38050 }, { "epoch": 4.04864, "grad_norm": 24.141675727975862, "learning_rate": 7.108507885370467e-06, "loss": 1.3682, "step": 38060 }, { "epoch": 4.04874, "grad_norm": 22.36350361053493, "learning_rate": 7.107039428928655e-06, "loss": 1.3333, "step": 38070 }, { "epoch": 4.04884, "grad_norm": 22.279847041282324, "learning_rate": 7.105570751468185e-06, "loss": 1.3324, "step": 38080 }, { "epoch": 4.04894, "grad_norm": 22.949382019689367, "learning_rate": 7.104101853143117e-06, "loss": 1.3349, "step": 38090 }, { "epoch": 4.04904, "grad_norm": 23.398414311384393, "learning_rate": 7.102632734107528e-06, "loss": 1.3453, "step": 38100 }, { "epoch": 4.04914, "grad_norm": 21.629143051904855, "learning_rate": 7.101163394515527e-06, "loss": 1.285, "step": 38110 }, { "epoch": 4.04924, "grad_norm": 24.694938839515952, "learning_rate": 7.0996938345212355e-06, "loss": 1.4297, "step": 38120 }, { "epoch": 4.04934, "grad_norm": 25.34880979938056, "learning_rate": 7.0982240542788054e-06, "loss": 1.3773, "step": 38130 }, { "epoch": 4.04944, "grad_norm": 23.144783269126705, "learning_rate": 7.096754053942409e-06, "loss": 1.3983, "step": 38140 }, { "epoch": 4.04954, "grad_norm": 22.55984438414471, "learning_rate": 7.095283833666246e-06, "loss": 1.3545, "step": 38150 }, { "epoch": 4.04964, "grad_norm": 24.357565000767504, "learning_rate": 7.093813393604534e-06, "loss": 1.3357, "step": 38160 }, { "epoch": 4.04974, "grad_norm": 21.28102793711458, "learning_rate": 7.092342733911512e-06, "loss": 1.3997, "step": 38170 }, { "epoch": 4.04984, "grad_norm": 21.957279588053098, "learning_rate": 7.09087185474145e-06, "loss": 1.3701, "step": 38180 }, { "epoch": 4.04994, "grad_norm": 22.04222571533686, "learning_rate": 7.089400756248635e-06, "loss": 1.3897, "step": 38190 }, { "epoch": 4.05004, "grad_norm": 23.926538632451564, "learning_rate": 7.0879294385873765e-06, "loss": 1.3511, "step": 38200 }, { "epoch": 4.05014, "grad_norm": 23.208566508508664, "learning_rate": 7.086457901912011e-06, "loss": 1.3224, "step": 38210 }, { "epoch": 4.05024, "grad_norm": 24.3131233537944, "learning_rate": 7.0849861463768935e-06, "loss": 1.3851, "step": 38220 }, { "epoch": 4.05034, "grad_norm": 23.662480558897254, "learning_rate": 7.083514172136407e-06, "loss": 1.4779, "step": 38230 }, { "epoch": 4.05044, "grad_norm": 23.28502267649005, "learning_rate": 7.082041979344954e-06, "loss": 1.3957, "step": 38240 }, { "epoch": 4.05054, "grad_norm": 24.575588798772355, "learning_rate": 7.0805695681569605e-06, "loss": 1.4082, "step": 38250 }, { "epoch": 4.05064, "grad_norm": 23.583656131809104, "learning_rate": 7.079096938726876e-06, "loss": 1.4068, "step": 38260 }, { "epoch": 4.05074, "grad_norm": 23.444279528000063, "learning_rate": 7.077624091209172e-06, "loss": 1.378, "step": 38270 }, { "epoch": 4.05084, "grad_norm": 22.200625336287942, "learning_rate": 7.0761510257583425e-06, "loss": 1.368, "step": 38280 }, { "epoch": 4.05094, "grad_norm": 20.94691543865242, "learning_rate": 7.074677742528907e-06, "loss": 1.3547, "step": 38290 }, { "epoch": 4.05104, "grad_norm": 21.50221748773376, "learning_rate": 7.0732042416754045e-06, "loss": 1.3889, "step": 38300 }, { "epoch": 4.05114, "grad_norm": 22.274031083283823, "learning_rate": 7.071730523352399e-06, "loss": 1.4066, "step": 38310 }, { "epoch": 4.05124, "grad_norm": 22.50208900891088, "learning_rate": 7.070256587714477e-06, "loss": 1.3736, "step": 38320 }, { "epoch": 4.05134, "grad_norm": 20.38131918857985, "learning_rate": 7.068782434916247e-06, "loss": 1.3672, "step": 38330 }, { "epoch": 4.05144, "grad_norm": 24.10249442189663, "learning_rate": 7.067308065112342e-06, "loss": 1.3687, "step": 38340 }, { "epoch": 4.05154, "grad_norm": 25.066161717614925, "learning_rate": 7.065833478457415e-06, "loss": 1.4408, "step": 38350 }, { "epoch": 4.05164, "grad_norm": 21.87947659171191, "learning_rate": 7.064358675106143e-06, "loss": 1.3932, "step": 38360 }, { "epoch": 4.05174, "grad_norm": 21.441649743499294, "learning_rate": 7.062883655213228e-06, "loss": 1.302, "step": 38370 }, { "epoch": 4.05184, "grad_norm": 23.168492530151262, "learning_rate": 7.061408418933393e-06, "loss": 1.3648, "step": 38380 }, { "epoch": 4.05194, "grad_norm": 24.013397555351904, "learning_rate": 7.05993296642138e-06, "loss": 1.2624, "step": 38390 }, { "epoch": 4.05204, "grad_norm": 23.68374640905211, "learning_rate": 7.05845729783196e-06, "loss": 1.3631, "step": 38400 }, { "epoch": 4.05214, "grad_norm": 23.25567887710821, "learning_rate": 7.056981413319923e-06, "loss": 1.3261, "step": 38410 }, { "epoch": 4.05224, "grad_norm": 22.263233120702377, "learning_rate": 7.055505313040083e-06, "loss": 1.3292, "step": 38420 }, { "epoch": 4.05234, "grad_norm": 23.606426678477664, "learning_rate": 7.054028997147275e-06, "loss": 1.346, "step": 38430 }, { "epoch": 4.05244, "grad_norm": 22.365436624910583, "learning_rate": 7.05255246579636e-06, "loss": 1.3318, "step": 38440 }, { "epoch": 4.05254, "grad_norm": 26.157661879818754, "learning_rate": 7.051075719142215e-06, "loss": 1.3673, "step": 38450 }, { "epoch": 4.05264, "grad_norm": 25.622879704794233, "learning_rate": 7.049598757339747e-06, "loss": 1.4258, "step": 38460 }, { "epoch": 4.05274, "grad_norm": 25.33752833058769, "learning_rate": 7.048121580543883e-06, "loss": 1.3385, "step": 38470 }, { "epoch": 4.05284, "grad_norm": 22.971206294155305, "learning_rate": 7.04664418890957e-06, "loss": 1.358, "step": 38480 }, { "epoch": 4.05294, "grad_norm": 22.52105939105329, "learning_rate": 7.0451665825917804e-06, "loss": 1.3656, "step": 38490 }, { "epoch": 4.05304, "grad_norm": 22.81109426285412, "learning_rate": 7.043688761745508e-06, "loss": 1.3652, "step": 38500 }, { "epoch": 4.05314, "grad_norm": 25.048539762490133, "learning_rate": 7.042210726525769e-06, "loss": 1.316, "step": 38510 }, { "epoch": 4.05324, "grad_norm": 21.989976242820227, "learning_rate": 7.040732477087604e-06, "loss": 1.3734, "step": 38520 }, { "epoch": 4.05334, "grad_norm": 25.38417413608748, "learning_rate": 7.0392540135860736e-06, "loss": 1.3311, "step": 38530 }, { "epoch": 4.05344, "grad_norm": 21.783921209263923, "learning_rate": 7.037775336176262e-06, "loss": 1.3779, "step": 38540 }, { "epoch": 4.05354, "grad_norm": 19.966818130619746, "learning_rate": 7.0362964450132734e-06, "loss": 1.3485, "step": 38550 }, { "epoch": 4.05364, "grad_norm": 22.95075553560104, "learning_rate": 7.034817340252239e-06, "loss": 1.3965, "step": 38560 }, { "epoch": 4.05374, "grad_norm": 23.239302804279856, "learning_rate": 7.033338022048311e-06, "loss": 1.3768, "step": 38570 }, { "epoch": 4.05384, "grad_norm": 24.410464267190402, "learning_rate": 7.0318584905566586e-06, "loss": 1.4035, "step": 38580 }, { "epoch": 4.05394, "grad_norm": 22.424902159893936, "learning_rate": 7.030378745932482e-06, "loss": 1.358, "step": 38590 }, { "epoch": 4.05404, "grad_norm": 24.512672125644563, "learning_rate": 7.028898788331e-06, "loss": 1.3549, "step": 38600 }, { "epoch": 4.05414, "grad_norm": 25.077857707243783, "learning_rate": 7.027418617907451e-06, "loss": 1.4054, "step": 38610 }, { "epoch": 4.05424, "grad_norm": 23.021118978561816, "learning_rate": 7.025938234817098e-06, "loss": 1.4284, "step": 38620 }, { "epoch": 4.05434, "grad_norm": 26.15270285196354, "learning_rate": 7.024457639215226e-06, "loss": 1.4405, "step": 38630 }, { "epoch": 4.05444, "grad_norm": 25.750597367205344, "learning_rate": 7.022976831257145e-06, "loss": 1.3223, "step": 38640 }, { "epoch": 4.05454, "grad_norm": 21.565718388213867, "learning_rate": 7.021495811098183e-06, "loss": 1.3771, "step": 38650 }, { "epoch": 4.05464, "grad_norm": 22.12661861213002, "learning_rate": 7.020014578893692e-06, "loss": 1.4094, "step": 38660 }, { "epoch": 4.05474, "grad_norm": 22.23830301080659, "learning_rate": 7.018533134799046e-06, "loss": 1.3997, "step": 38670 }, { "epoch": 4.05484, "grad_norm": 20.705998444307994, "learning_rate": 7.017051478969643e-06, "loss": 1.4397, "step": 38680 }, { "epoch": 4.05494, "grad_norm": 23.190664519186402, "learning_rate": 7.015569611560902e-06, "loss": 1.3604, "step": 38690 }, { "epoch": 4.05504, "grad_norm": 22.43337049975622, "learning_rate": 7.014087532728263e-06, "loss": 1.3022, "step": 38700 }, { "epoch": 4.05514, "grad_norm": 21.698007328109476, "learning_rate": 7.012605242627189e-06, "loss": 1.3086, "step": 38710 }, { "epoch": 4.05524, "grad_norm": 22.983368807133708, "learning_rate": 7.0111227414131665e-06, "loss": 1.3075, "step": 38720 }, { "epoch": 4.05534, "grad_norm": 24.327263987043807, "learning_rate": 7.0096400292417026e-06, "loss": 1.3082, "step": 38730 }, { "epoch": 4.05544, "grad_norm": 22.90444825230238, "learning_rate": 7.008157106268325e-06, "loss": 1.3866, "step": 38740 }, { "epoch": 4.05554, "grad_norm": 24.20344024218742, "learning_rate": 7.006673972648588e-06, "loss": 1.3993, "step": 38750 }, { "epoch": 4.05564, "grad_norm": 21.782081055588005, "learning_rate": 7.005190628538065e-06, "loss": 1.3958, "step": 38760 }, { "epoch": 4.05574, "grad_norm": 23.34424621495763, "learning_rate": 7.003707074092352e-06, "loss": 1.3326, "step": 38770 }, { "epoch": 4.05584, "grad_norm": 22.339792042179152, "learning_rate": 7.002223309467064e-06, "loss": 1.3594, "step": 38780 }, { "epoch": 4.05594, "grad_norm": 21.94643071698594, "learning_rate": 7.000739334817845e-06, "loss": 1.3568, "step": 38790 }, { "epoch": 4.05604, "grad_norm": 20.9996149298656, "learning_rate": 6.999255150300357e-06, "loss": 1.2818, "step": 38800 }, { "epoch": 4.05614, "grad_norm": 22.415641834993572, "learning_rate": 6.99777075607028e-06, "loss": 1.3558, "step": 38810 }, { "epoch": 4.05624, "grad_norm": 26.894564270753314, "learning_rate": 6.996286152283325e-06, "loss": 1.3579, "step": 38820 }, { "epoch": 4.05634, "grad_norm": 23.599880341936704, "learning_rate": 6.994801339095216e-06, "loss": 1.3529, "step": 38830 }, { "epoch": 4.05644, "grad_norm": 24.334717024907857, "learning_rate": 6.993316316661705e-06, "loss": 1.4316, "step": 38840 }, { "epoch": 4.05654, "grad_norm": 24.01048569711147, "learning_rate": 6.991831085138563e-06, "loss": 1.3022, "step": 38850 }, { "epoch": 4.05664, "grad_norm": 23.55075103028641, "learning_rate": 6.990345644681584e-06, "loss": 1.3645, "step": 38860 }, { "epoch": 4.05674, "grad_norm": 24.853952999789158, "learning_rate": 6.988859995446586e-06, "loss": 1.4439, "step": 38870 }, { "epoch": 4.05684, "grad_norm": 23.687634889768887, "learning_rate": 6.987374137589405e-06, "loss": 1.3721, "step": 38880 }, { "epoch": 4.05694, "grad_norm": 25.347629274760692, "learning_rate": 6.9858880712659005e-06, "loss": 1.3591, "step": 38890 }, { "epoch": 4.05704, "grad_norm": 24.11858815543077, "learning_rate": 6.984401796631954e-06, "loss": 1.3079, "step": 38900 }, { "epoch": 4.05714, "grad_norm": 22.61871075947072, "learning_rate": 6.982915313843469e-06, "loss": 1.3542, "step": 38910 }, { "epoch": 4.05724, "grad_norm": 24.312487246433438, "learning_rate": 6.98142862305637e-06, "loss": 1.4023, "step": 38920 }, { "epoch": 4.05734, "grad_norm": 22.943161000183625, "learning_rate": 6.979941724426606e-06, "loss": 1.332, "step": 38930 }, { "epoch": 4.05744, "grad_norm": 23.490201767244887, "learning_rate": 6.978454618110142e-06, "loss": 1.3609, "step": 38940 }, { "epoch": 4.05754, "grad_norm": 25.496450205069905, "learning_rate": 6.976967304262973e-06, "loss": 1.3413, "step": 38950 }, { "epoch": 4.05764, "grad_norm": 21.18764864479274, "learning_rate": 6.975479783041108e-06, "loss": 1.3496, "step": 38960 }, { "epoch": 4.05774, "grad_norm": 24.674597747499256, "learning_rate": 6.973992054600583e-06, "loss": 1.401, "step": 38970 }, { "epoch": 4.05784, "grad_norm": 21.615116842878138, "learning_rate": 6.972504119097453e-06, "loss": 1.3384, "step": 38980 }, { "epoch": 4.05794, "grad_norm": 25.45686210084261, "learning_rate": 6.9710159766877935e-06, "loss": 1.3694, "step": 38990 }, { "epoch": 4.05804, "grad_norm": 29.13592088629628, "learning_rate": 6.9695276275277095e-06, "loss": 1.3925, "step": 39000 }, { "epoch": 4.05814, "grad_norm": 27.02530993161683, "learning_rate": 6.968039071773316e-06, "loss": 1.3911, "step": 39010 }, { "epoch": 4.05824, "grad_norm": 23.662555323282255, "learning_rate": 6.9665503095807575e-06, "loss": 1.3739, "step": 39020 }, { "epoch": 4.05834, "grad_norm": 22.041013368887274, "learning_rate": 6.965061341106198e-06, "loss": 1.358, "step": 39030 }, { "epoch": 4.05844, "grad_norm": 24.499271268065172, "learning_rate": 6.963572166505825e-06, "loss": 1.3637, "step": 39040 }, { "epoch": 4.05854, "grad_norm": 23.72971107518072, "learning_rate": 6.962082785935846e-06, "loss": 1.3003, "step": 39050 }, { "epoch": 4.05864, "grad_norm": 22.507323672310307, "learning_rate": 6.960593199552488e-06, "loss": 1.4151, "step": 39060 }, { "epoch": 4.05874, "grad_norm": 22.113533570847256, "learning_rate": 6.959103407512002e-06, "loss": 1.453, "step": 39070 }, { "epoch": 4.05884, "grad_norm": 23.80077959160841, "learning_rate": 6.957613409970662e-06, "loss": 1.495, "step": 39080 }, { "epoch": 4.05894, "grad_norm": 25.575133200958206, "learning_rate": 6.956123207084762e-06, "loss": 1.4539, "step": 39090 }, { "epoch": 4.05904, "grad_norm": 21.94969589243336, "learning_rate": 6.9546327990106145e-06, "loss": 1.3474, "step": 39100 }, { "epoch": 4.05914, "grad_norm": 21.793480453092407, "learning_rate": 6.953142185904558e-06, "loss": 1.4245, "step": 39110 }, { "epoch": 4.05924, "grad_norm": 21.333752311939556, "learning_rate": 6.951651367922951e-06, "loss": 1.3575, "step": 39120 }, { "epoch": 4.05934, "grad_norm": 25.37132075230678, "learning_rate": 6.950160345222175e-06, "loss": 1.4548, "step": 39130 }, { "epoch": 4.05944, "grad_norm": 22.83971558283741, "learning_rate": 6.9486691179586284e-06, "loss": 1.3592, "step": 39140 }, { "epoch": 4.05954, "grad_norm": 20.483697020232388, "learning_rate": 6.9471776862887365e-06, "loss": 1.3405, "step": 39150 }, { "epoch": 4.05964, "grad_norm": 23.141819927398682, "learning_rate": 6.9456860503689416e-06, "loss": 1.3752, "step": 39160 }, { "epoch": 4.05974, "grad_norm": 22.633497974871574, "learning_rate": 6.94419421035571e-06, "loss": 1.4158, "step": 39170 }, { "epoch": 4.05984, "grad_norm": 24.87355387907969, "learning_rate": 6.94270216640553e-06, "loss": 1.3954, "step": 39180 }, { "epoch": 4.05994, "grad_norm": 21.995056785046856, "learning_rate": 6.941209918674907e-06, "loss": 1.3615, "step": 39190 }, { "epoch": 4.06004, "grad_norm": 23.67501329202802, "learning_rate": 6.939717467320374e-06, "loss": 1.3392, "step": 39200 }, { "epoch": 4.06014, "grad_norm": 20.692133265034737, "learning_rate": 6.938224812498482e-06, "loss": 1.3043, "step": 39210 }, { "epoch": 4.06024, "grad_norm": 23.423374501936067, "learning_rate": 6.936731954365802e-06, "loss": 1.3658, "step": 39220 }, { "epoch": 4.06034, "grad_norm": 23.5790770547221, "learning_rate": 6.9352388930789296e-06, "loss": 1.3704, "step": 39230 }, { "epoch": 4.06044, "grad_norm": 26.007672122809044, "learning_rate": 6.933745628794478e-06, "loss": 1.3632, "step": 39240 }, { "epoch": 4.06054, "grad_norm": 22.170733235962174, "learning_rate": 6.932252161669085e-06, "loss": 1.3261, "step": 39250 }, { "epoch": 4.06064, "grad_norm": 24.716554824965662, "learning_rate": 6.930758491859408e-06, "loss": 1.3743, "step": 39260 }, { "epoch": 4.06074, "grad_norm": 22.602082464938817, "learning_rate": 6.929264619522127e-06, "loss": 1.4264, "step": 39270 }, { "epoch": 4.06084, "grad_norm": 26.70985395517453, "learning_rate": 6.927770544813941e-06, "loss": 1.3702, "step": 39280 }, { "epoch": 4.06094, "grad_norm": 26.360126566806333, "learning_rate": 6.9262762678915704e-06, "loss": 1.3575, "step": 39290 }, { "epoch": 4.06104, "grad_norm": 20.7579562888823, "learning_rate": 6.92478178891176e-06, "loss": 1.3935, "step": 39300 }, { "epoch": 4.06114, "grad_norm": 23.257638475224248, "learning_rate": 6.9232871080312755e-06, "loss": 1.3971, "step": 39310 }, { "epoch": 4.06124, "grad_norm": 27.386699721751437, "learning_rate": 6.921792225406898e-06, "loss": 1.4649, "step": 39320 }, { "epoch": 4.06134, "grad_norm": 20.93858763076683, "learning_rate": 6.920297141195435e-06, "loss": 1.2949, "step": 39330 }, { "epoch": 4.06144, "grad_norm": 22.634226157909982, "learning_rate": 6.918801855553715e-06, "loss": 1.4243, "step": 39340 }, { "epoch": 4.06154, "grad_norm": 22.58007719002702, "learning_rate": 6.917306368638584e-06, "loss": 1.4037, "step": 39350 }, { "epoch": 4.06164, "grad_norm": 23.334715992538612, "learning_rate": 6.915810680606916e-06, "loss": 1.4061, "step": 39360 }, { "epoch": 4.06174, "grad_norm": 23.81178083393517, "learning_rate": 6.914314791615597e-06, "loss": 1.3882, "step": 39370 }, { "epoch": 4.06184, "grad_norm": 23.543032141268753, "learning_rate": 6.9128187018215424e-06, "loss": 1.4291, "step": 39380 }, { "epoch": 4.06194, "grad_norm": 23.63431960536402, "learning_rate": 6.911322411381683e-06, "loss": 1.4086, "step": 39390 }, { "epoch": 4.06204, "grad_norm": 26.445831051591238, "learning_rate": 6.909825920452972e-06, "loss": 1.3476, "step": 39400 }, { "epoch": 4.06214, "grad_norm": 34.18188571000066, "learning_rate": 6.908329229192389e-06, "loss": 1.3177, "step": 39410 }, { "epoch": 4.06224, "grad_norm": 23.771545569870337, "learning_rate": 6.906832337756924e-06, "loss": 1.3566, "step": 39420 }, { "epoch": 4.06234, "grad_norm": 25.462241651296754, "learning_rate": 6.905335246303599e-06, "loss": 1.3667, "step": 39430 }, { "epoch": 4.06244, "grad_norm": 24.008191991587495, "learning_rate": 6.903837954989447e-06, "loss": 1.4312, "step": 39440 }, { "epoch": 4.06254, "grad_norm": 21.221834301907684, "learning_rate": 6.902340463971531e-06, "loss": 1.2984, "step": 39450 }, { "epoch": 4.06264, "grad_norm": 23.24894436235358, "learning_rate": 6.900842773406928e-06, "loss": 1.349, "step": 39460 }, { "epoch": 4.06274, "grad_norm": 24.036423337531392, "learning_rate": 6.899344883452742e-06, "loss": 1.3925, "step": 39470 }, { "epoch": 4.06284, "grad_norm": 23.41940553584725, "learning_rate": 6.897846794266092e-06, "loss": 1.3218, "step": 39480 }, { "epoch": 4.06294, "grad_norm": 23.42455015679741, "learning_rate": 6.8963485060041224e-06, "loss": 1.3228, "step": 39490 }, { "epoch": 4.06304, "grad_norm": 26.816228375540955, "learning_rate": 6.894850018823994e-06, "loss": 1.3801, "step": 39500 }, { "epoch": 4.06314, "grad_norm": 21.689113482247148, "learning_rate": 6.893351332882894e-06, "loss": 1.3285, "step": 39510 }, { "epoch": 4.06324, "grad_norm": 21.195122202479357, "learning_rate": 6.891852448338027e-06, "loss": 1.3908, "step": 39520 }, { "epoch": 4.06334, "grad_norm": 22.357839425556698, "learning_rate": 6.890353365346617e-06, "loss": 1.3198, "step": 39530 }, { "epoch": 4.06344, "grad_norm": 22.591953127567784, "learning_rate": 6.888854084065914e-06, "loss": 1.3962, "step": 39540 }, { "epoch": 4.06354, "grad_norm": 22.54310747946454, "learning_rate": 6.887354604653181e-06, "loss": 1.2954, "step": 39550 }, { "epoch": 4.06364, "grad_norm": 25.81565878605334, "learning_rate": 6.885854927265713e-06, "loss": 1.4352, "step": 39560 }, { "epoch": 4.06374, "grad_norm": 21.931032298429137, "learning_rate": 6.884355052060814e-06, "loss": 1.4075, "step": 39570 }, { "epoch": 4.06384, "grad_norm": 23.7945644492297, "learning_rate": 6.8828549791958154e-06, "loss": 1.3729, "step": 39580 }, { "epoch": 4.06394, "grad_norm": 23.21549490748316, "learning_rate": 6.88135470882807e-06, "loss": 1.3802, "step": 39590 }, { "epoch": 4.06404, "grad_norm": 23.809588752986183, "learning_rate": 6.879854241114945e-06, "loss": 1.2769, "step": 39600 }, { "epoch": 4.06414, "grad_norm": 23.89394705135227, "learning_rate": 6.878353576213836e-06, "loss": 1.2946, "step": 39610 }, { "epoch": 4.06424, "grad_norm": 24.53486137026213, "learning_rate": 6.876852714282154e-06, "loss": 1.3287, "step": 39620 }, { "epoch": 4.06434, "grad_norm": 23.228087540884594, "learning_rate": 6.8753516554773335e-06, "loss": 1.3769, "step": 39630 }, { "epoch": 4.06444, "grad_norm": 24.258754849514077, "learning_rate": 6.873850399956828e-06, "loss": 1.3129, "step": 39640 }, { "epoch": 4.06454, "grad_norm": 24.122034123359093, "learning_rate": 6.8723489478781115e-06, "loss": 1.3536, "step": 39650 }, { "epoch": 4.06464, "grad_norm": 22.946367490501967, "learning_rate": 6.870847299398681e-06, "loss": 1.3913, "step": 39660 }, { "epoch": 4.06474, "grad_norm": 25.135077859535578, "learning_rate": 6.86934545467605e-06, "loss": 1.3636, "step": 39670 }, { "epoch": 4.06484, "grad_norm": 24.423297183537947, "learning_rate": 6.8678434138677585e-06, "loss": 1.4068, "step": 39680 }, { "epoch": 4.06494, "grad_norm": 21.271826899294247, "learning_rate": 6.8663411771313625e-06, "loss": 1.3954, "step": 39690 }, { "epoch": 4.06504, "grad_norm": 23.21186005823681, "learning_rate": 6.864838744624438e-06, "loss": 1.3528, "step": 39700 }, { "epoch": 4.06514, "grad_norm": 22.88250386966482, "learning_rate": 6.863336116504585e-06, "loss": 1.3008, "step": 39710 }, { "epoch": 4.06524, "grad_norm": 24.234982494551492, "learning_rate": 6.86183329292942e-06, "loss": 1.3545, "step": 39720 }, { "epoch": 4.06534, "grad_norm": 21.576049752681573, "learning_rate": 6.860330274056583e-06, "loss": 1.2801, "step": 39730 }, { "epoch": 4.06544, "grad_norm": 25.69620803429456, "learning_rate": 6.858827060043737e-06, "loss": 1.3318, "step": 39740 }, { "epoch": 4.06554, "grad_norm": 24.547468279397215, "learning_rate": 6.857323651048558e-06, "loss": 1.375, "step": 39750 }, { "epoch": 4.06564, "grad_norm": 21.546536923926073, "learning_rate": 6.855820047228748e-06, "loss": 1.3523, "step": 39760 }, { "epoch": 4.06574, "grad_norm": 25.09336416762857, "learning_rate": 6.854316248742028e-06, "loss": 1.3322, "step": 39770 }, { "epoch": 4.06584, "grad_norm": 23.9817140297236, "learning_rate": 6.852812255746139e-06, "loss": 1.2927, "step": 39780 }, { "epoch": 4.06594, "grad_norm": 24.701505614365217, "learning_rate": 6.851308068398845e-06, "loss": 1.3561, "step": 39790 }, { "epoch": 4.06604, "grad_norm": 22.548042527982656, "learning_rate": 6.849803686857927e-06, "loss": 1.3612, "step": 39800 }, { "epoch": 4.06614, "grad_norm": 21.63029291538008, "learning_rate": 6.848299111281187e-06, "loss": 1.4106, "step": 39810 }, { "epoch": 4.06624, "grad_norm": 22.762325394734518, "learning_rate": 6.8467943418264474e-06, "loss": 1.3375, "step": 39820 }, { "epoch": 4.06634, "grad_norm": 27.105509581370605, "learning_rate": 6.845289378651553e-06, "loss": 1.4159, "step": 39830 }, { "epoch": 4.06644, "grad_norm": 22.054641102886144, "learning_rate": 6.843784221914368e-06, "loss": 1.3418, "step": 39840 }, { "epoch": 4.06654, "grad_norm": 22.306417509789533, "learning_rate": 6.842278871772774e-06, "loss": 1.3712, "step": 39850 }, { "epoch": 4.06664, "grad_norm": 21.79157534234874, "learning_rate": 6.840773328384678e-06, "loss": 1.3759, "step": 39860 }, { "epoch": 4.06674, "grad_norm": 21.91234964655845, "learning_rate": 6.839267591908003e-06, "loss": 1.3107, "step": 39870 }, { "epoch": 4.06684, "grad_norm": 24.985087162866982, "learning_rate": 6.837761662500692e-06, "loss": 1.2853, "step": 39880 }, { "epoch": 4.06694, "grad_norm": 24.22671864838291, "learning_rate": 6.836255540320716e-06, "loss": 1.3258, "step": 39890 }, { "epoch": 4.06704, "grad_norm": 22.2457580719886, "learning_rate": 6.834749225526052e-06, "loss": 1.3202, "step": 39900 }, { "epoch": 4.06714, "grad_norm": 23.34616106573, "learning_rate": 6.833242718274711e-06, "loss": 1.2709, "step": 39910 }, { "epoch": 4.06724, "grad_norm": 24.59830071990932, "learning_rate": 6.831736018724717e-06, "loss": 1.3281, "step": 39920 }, { "epoch": 4.06734, "grad_norm": 25.61335399876323, "learning_rate": 6.830229127034116e-06, "loss": 1.3399, "step": 39930 }, { "epoch": 4.06744, "grad_norm": 25.533959197469226, "learning_rate": 6.828722043360976e-06, "loss": 1.4246, "step": 39940 }, { "epoch": 4.06754, "grad_norm": 24.322200998106663, "learning_rate": 6.8272147678633785e-06, "loss": 1.3568, "step": 39950 }, { "epoch": 4.06764, "grad_norm": 23.371588460444904, "learning_rate": 6.825707300699433e-06, "loss": 1.337, "step": 39960 }, { "epoch": 4.06774, "grad_norm": 19.96064089916883, "learning_rate": 6.824199642027265e-06, "loss": 1.3403, "step": 39970 }, { "epoch": 4.06784, "grad_norm": 25.112830311663707, "learning_rate": 6.822691792005021e-06, "loss": 1.3946, "step": 39980 }, { "epoch": 4.06794, "grad_norm": 22.94287137128167, "learning_rate": 6.821183750790868e-06, "loss": 1.3616, "step": 39990 }, { "epoch": 4.06804, "grad_norm": 21.692892895350052, "learning_rate": 6.81967551854299e-06, "loss": 1.3485, "step": 40000 }, { "epoch": 4.06814, "grad_norm": 21.161705313625156, "learning_rate": 6.818167095419596e-06, "loss": 1.3377, "step": 40010 }, { "epoch": 4.06824, "grad_norm": 22.850734701420027, "learning_rate": 6.816658481578914e-06, "loss": 1.3513, "step": 40020 }, { "epoch": 4.06834, "grad_norm": 22.66728711840192, "learning_rate": 6.815149677179187e-06, "loss": 1.3456, "step": 40030 }, { "epoch": 4.06844, "grad_norm": 26.517880915385902, "learning_rate": 6.813640682378684e-06, "loss": 1.3537, "step": 40040 }, { "epoch": 4.06854, "grad_norm": 24.97687558481324, "learning_rate": 6.81213149733569e-06, "loss": 1.3031, "step": 40050 }, { "epoch": 4.06864, "grad_norm": 22.915563177850853, "learning_rate": 6.8106221222085125e-06, "loss": 1.2414, "step": 40060 }, { "epoch": 4.06874, "grad_norm": 22.37405885441566, "learning_rate": 6.8091125571554774e-06, "loss": 1.3551, "step": 40070 }, { "epoch": 4.06884, "grad_norm": 22.707301203456783, "learning_rate": 6.807602802334932e-06, "loss": 1.3023, "step": 40080 }, { "epoch": 4.06894, "grad_norm": 25.824874360675985, "learning_rate": 6.8060928579052425e-06, "loss": 1.3058, "step": 40090 }, { "epoch": 4.06904, "grad_norm": 23.84215859367365, "learning_rate": 6.804582724024794e-06, "loss": 1.3433, "step": 40100 }, { "epoch": 4.06914, "grad_norm": 25.03252414044392, "learning_rate": 6.803072400851993e-06, "loss": 1.3499, "step": 40110 }, { "epoch": 4.06924, "grad_norm": 25.754852100288637, "learning_rate": 6.8015618885452665e-06, "loss": 1.3682, "step": 40120 }, { "epoch": 4.06934, "grad_norm": 23.349029984280524, "learning_rate": 6.800051187263057e-06, "loss": 1.3705, "step": 40130 }, { "epoch": 4.06944, "grad_norm": 21.463309589845956, "learning_rate": 6.798540297163833e-06, "loss": 1.314, "step": 40140 }, { "epoch": 4.06954, "grad_norm": 22.772037844526075, "learning_rate": 6.797029218406081e-06, "loss": 1.2554, "step": 40150 }, { "epoch": 4.06964, "grad_norm": 20.585690783573636, "learning_rate": 6.795517951148303e-06, "loss": 1.2694, "step": 40160 }, { "epoch": 4.06974, "grad_norm": 22.98526323496721, "learning_rate": 6.794006495549024e-06, "loss": 1.2259, "step": 40170 }, { "epoch": 4.06984, "grad_norm": 24.438108121141795, "learning_rate": 6.792494851766791e-06, "loss": 1.2476, "step": 40180 }, { "epoch": 4.06994, "grad_norm": 25.793397359389935, "learning_rate": 6.790983019960167e-06, "loss": 1.3767, "step": 40190 }, { "epoch": 4.07004, "grad_norm": 22.94858317465858, "learning_rate": 6.789471000287737e-06, "loss": 1.386, "step": 40200 }, { "epoch": 4.07014, "grad_norm": 22.379198789906972, "learning_rate": 6.787958792908104e-06, "loss": 1.3227, "step": 40210 }, { "epoch": 4.07024, "grad_norm": 23.28225512856171, "learning_rate": 6.786446397979892e-06, "loss": 1.3167, "step": 40220 }, { "epoch": 4.07034, "grad_norm": 23.480944177680005, "learning_rate": 6.784933815661743e-06, "loss": 1.309, "step": 40230 }, { "epoch": 4.07044, "grad_norm": 27.017771634272886, "learning_rate": 6.783421046112322e-06, "loss": 1.2871, "step": 40240 }, { "epoch": 4.07054, "grad_norm": 24.055787133112727, "learning_rate": 6.781908089490309e-06, "loss": 1.2369, "step": 40250 }, { "epoch": 4.07064, "grad_norm": 24.382711310047572, "learning_rate": 6.780394945954408e-06, "loss": 1.332, "step": 40260 }, { "epoch": 4.07074, "grad_norm": 23.539469124034834, "learning_rate": 6.7788816156633415e-06, "loss": 1.3638, "step": 40270 }, { "epoch": 4.07084, "grad_norm": 25.900009725576023, "learning_rate": 6.777368098775848e-06, "loss": 1.3423, "step": 40280 }, { "epoch": 4.07094, "grad_norm": 24.289575224287685, "learning_rate": 6.775854395450691e-06, "loss": 1.3034, "step": 40290 }, { "epoch": 4.07104, "grad_norm": 24.00676491120084, "learning_rate": 6.7743405058466505e-06, "loss": 1.2745, "step": 40300 }, { "epoch": 4.07114, "grad_norm": 22.748592402481854, "learning_rate": 6.772826430122524e-06, "loss": 1.2905, "step": 40310 }, { "epoch": 4.07124, "grad_norm": 24.743686005678402, "learning_rate": 6.771312168437135e-06, "loss": 1.2753, "step": 40320 }, { "epoch": 4.07134, "grad_norm": 21.450839543974205, "learning_rate": 6.769797720949319e-06, "loss": 1.2527, "step": 40330 }, { "epoch": 4.07144, "grad_norm": 25.39662250042506, "learning_rate": 6.768283087817936e-06, "loss": 1.3323, "step": 40340 }, { "epoch": 4.07154, "grad_norm": 23.538493503821396, "learning_rate": 6.766768269201864e-06, "loss": 1.2973, "step": 40350 }, { "epoch": 4.07164, "grad_norm": 21.618366545802154, "learning_rate": 6.76525326526e-06, "loss": 1.2257, "step": 40360 }, { "epoch": 4.07174, "grad_norm": 26.84690073989012, "learning_rate": 6.763738076151263e-06, "loss": 1.2741, "step": 40370 }, { "epoch": 4.07184, "grad_norm": 26.54071420430054, "learning_rate": 6.762222702034587e-06, "loss": 1.3179, "step": 40380 }, { "epoch": 4.07194, "grad_norm": 24.07294793108617, "learning_rate": 6.760707143068926e-06, "loss": 1.2351, "step": 40390 }, { "epoch": 4.07204, "grad_norm": 25.288424406534087, "learning_rate": 6.759191399413261e-06, "loss": 1.3081, "step": 40400 }, { "epoch": 4.07214, "grad_norm": 22.745110245796198, "learning_rate": 6.75767547122658e-06, "loss": 1.2468, "step": 40410 }, { "epoch": 4.07224, "grad_norm": 22.478455724936055, "learning_rate": 6.756159358667901e-06, "loss": 1.3694, "step": 40420 }, { "epoch": 4.07234, "grad_norm": 22.845313452691396, "learning_rate": 6.754643061896255e-06, "loss": 1.2627, "step": 40430 }, { "epoch": 4.07244, "grad_norm": 25.292154280885523, "learning_rate": 6.753126581070695e-06, "loss": 1.3261, "step": 40440 }, { "epoch": 4.07254, "grad_norm": 20.56538813470339, "learning_rate": 6.7516099163502945e-06, "loss": 1.2643, "step": 40450 }, { "epoch": 4.07264, "grad_norm": 22.64757494038253, "learning_rate": 6.7500930678941436e-06, "loss": 1.3162, "step": 40460 }, { "epoch": 4.07274, "grad_norm": 32.74472228325291, "learning_rate": 6.748576035861351e-06, "loss": 1.3087, "step": 40470 }, { "epoch": 4.07284, "grad_norm": 355.22784349894835, "learning_rate": 6.7470588204110484e-06, "loss": 1.2941, "step": 40480 }, { "epoch": 4.07294, "grad_norm": 22.237299414481758, "learning_rate": 6.745541421702384e-06, "loss": 1.2627, "step": 40490 }, { "epoch": 4.07304, "grad_norm": 25.753281939793617, "learning_rate": 6.744023839894526e-06, "loss": 1.3125, "step": 40500 }, { "epoch": 4.07314, "grad_norm": 25.596785478665524, "learning_rate": 6.74250607514666e-06, "loss": 1.2686, "step": 40510 }, { "epoch": 4.07324, "grad_norm": 23.785633751653467, "learning_rate": 6.7409881276179945e-06, "loss": 1.2367, "step": 40520 }, { "epoch": 4.07334, "grad_norm": 22.099063638599677, "learning_rate": 6.739469997467754e-06, "loss": 1.3123, "step": 40530 }, { "epoch": 4.07344, "grad_norm": 25.87534663197247, "learning_rate": 6.737951684855185e-06, "loss": 1.3055, "step": 40540 }, { "epoch": 4.07354, "grad_norm": 23.34774951566463, "learning_rate": 6.7364331899395505e-06, "loss": 1.2752, "step": 40550 }, { "epoch": 4.07364, "grad_norm": 26.03859732050769, "learning_rate": 6.734914512880133e-06, "loss": 1.3076, "step": 40560 }, { "epoch": 4.07374, "grad_norm": 26.088433778935485, "learning_rate": 6.733395653836236e-06, "loss": 1.2608, "step": 40570 }, { "epoch": 4.07384, "grad_norm": 24.486490431616513, "learning_rate": 6.731876612967179e-06, "loss": 1.294, "step": 40580 }, { "epoch": 4.07394, "grad_norm": 21.849362523724686, "learning_rate": 6.7303573904323025e-06, "loss": 1.2394, "step": 40590 }, { "epoch": 4.07404, "grad_norm": 24.837072260913853, "learning_rate": 6.728837986390968e-06, "loss": 1.2403, "step": 40600 }, { "epoch": 4.07414, "grad_norm": 20.656523974888565, "learning_rate": 6.727318401002551e-06, "loss": 1.2639, "step": 40610 }, { "epoch": 4.07424, "grad_norm": 26.683576963245887, "learning_rate": 6.7257986344264506e-06, "loss": 1.2977, "step": 40620 }, { "epoch": 4.07434, "grad_norm": 23.86202536476196, "learning_rate": 6.724278686822085e-06, "loss": 1.2881, "step": 40630 }, { "epoch": 4.07444, "grad_norm": 23.022314870360617, "learning_rate": 6.722758558348886e-06, "loss": 1.2537, "step": 40640 }, { "epoch": 4.07454, "grad_norm": 22.654487348292307, "learning_rate": 6.721238249166311e-06, "loss": 1.2863, "step": 40650 }, { "epoch": 4.07464, "grad_norm": 24.91695401734844, "learning_rate": 6.719717759433832e-06, "loss": 1.2761, "step": 40660 }, { "epoch": 4.07474, "grad_norm": 22.935863909106686, "learning_rate": 6.71819708931094e-06, "loss": 1.2941, "step": 40670 }, { "epoch": 4.07484, "grad_norm": 22.751731821264542, "learning_rate": 6.71667623895715e-06, "loss": 1.2548, "step": 40680 }, { "epoch": 4.07494, "grad_norm": 23.551467616128555, "learning_rate": 6.715155208531988e-06, "loss": 1.2844, "step": 40690 }, { "epoch": 4.0750399999999996, "grad_norm": 24.27590574275284, "learning_rate": 6.713633998195005e-06, "loss": 1.2313, "step": 40700 }, { "epoch": 4.07514, "grad_norm": 26.359920090598077, "learning_rate": 6.7121126081057695e-06, "loss": 1.3062, "step": 40710 }, { "epoch": 4.07524, "grad_norm": 21.516138442521946, "learning_rate": 6.710591038423867e-06, "loss": 1.2465, "step": 40720 }, { "epoch": 4.07534, "grad_norm": 23.729612519657685, "learning_rate": 6.709069289308905e-06, "loss": 1.2697, "step": 40730 }, { "epoch": 4.07544, "grad_norm": 22.251982989033092, "learning_rate": 6.707547360920506e-06, "loss": 1.2017, "step": 40740 }, { "epoch": 4.07554, "grad_norm": 21.32862339218139, "learning_rate": 6.706025253418312e-06, "loss": 1.2731, "step": 40750 }, { "epoch": 4.07564, "grad_norm": 21.84380426711049, "learning_rate": 6.704502966961989e-06, "loss": 1.2369, "step": 40760 }, { "epoch": 4.07574, "grad_norm": 21.15876410652177, "learning_rate": 6.702980501711215e-06, "loss": 1.2289, "step": 40770 }, { "epoch": 4.07584, "grad_norm": 22.428277869001825, "learning_rate": 6.7014578578256894e-06, "loss": 1.1762, "step": 40780 }, { "epoch": 4.07594, "grad_norm": 20.868180219759083, "learning_rate": 6.699935035465131e-06, "loss": 1.2698, "step": 40790 }, { "epoch": 4.07604, "grad_norm": 19.634850275840435, "learning_rate": 6.698412034789279e-06, "loss": 1.2197, "step": 40800 }, { "epoch": 4.07614, "grad_norm": 23.76928650294024, "learning_rate": 6.696888855957886e-06, "loss": 1.2678, "step": 40810 }, { "epoch": 4.07624, "grad_norm": 24.578638128041305, "learning_rate": 6.6953654991307274e-06, "loss": 1.2451, "step": 40820 }, { "epoch": 4.07634, "grad_norm": 21.94686843478, "learning_rate": 6.6938419644675964e-06, "loss": 1.2765, "step": 40830 }, { "epoch": 4.07644, "grad_norm": 21.469717807240983, "learning_rate": 6.692318252128305e-06, "loss": 1.2297, "step": 40840 }, { "epoch": 4.07654, "grad_norm": 20.795395411301403, "learning_rate": 6.690794362272684e-06, "loss": 1.2406, "step": 40850 }, { "epoch": 4.07664, "grad_norm": 22.377872571838708, "learning_rate": 6.689270295060581e-06, "loss": 1.1488, "step": 40860 }, { "epoch": 4.07674, "grad_norm": 23.943559327129833, "learning_rate": 6.687746050651863e-06, "loss": 1.1985, "step": 40870 }, { "epoch": 4.07684, "grad_norm": 21.17958770225445, "learning_rate": 6.686221629206418e-06, "loss": 1.2083, "step": 40880 }, { "epoch": 4.07694, "grad_norm": 25.799017426956915, "learning_rate": 6.68469703088415e-06, "loss": 1.2654, "step": 40890 }, { "epoch": 4.07704, "grad_norm": 24.4714840125696, "learning_rate": 6.683172255844982e-06, "loss": 1.2391, "step": 40900 }, { "epoch": 4.07714, "grad_norm": 20.984138196170196, "learning_rate": 6.681647304248858e-06, "loss": 1.2001, "step": 40910 }, { "epoch": 4.07724, "grad_norm": 24.84985783688582, "learning_rate": 6.680122176255734e-06, "loss": 1.2339, "step": 40920 }, { "epoch": 5.00002, "grad_norm": 17.3605639613277, "learning_rate": 6.678596872025594e-06, "loss": 1.2211, "step": 40930 }, { "epoch": 5.00012, "grad_norm": 22.611505772406804, "learning_rate": 6.677071391718431e-06, "loss": 1.2803, "step": 40940 }, { "epoch": 5.00022, "grad_norm": 25.431852056717396, "learning_rate": 6.675545735494264e-06, "loss": 1.2215, "step": 40950 }, { "epoch": 5.00032, "grad_norm": 24.55610476062788, "learning_rate": 6.674019903513125e-06, "loss": 1.2458, "step": 40960 }, { "epoch": 5.00042, "grad_norm": 28.818547407714, "learning_rate": 6.672493895935066e-06, "loss": 1.3592, "step": 40970 }, { "epoch": 5.00052, "grad_norm": 27.313538511756008, "learning_rate": 6.670967712920162e-06, "loss": 1.2887, "step": 40980 }, { "epoch": 5.00062, "grad_norm": 25.5362579286886, "learning_rate": 6.669441354628498e-06, "loss": 1.3318, "step": 40990 }, { "epoch": 5.00072, "grad_norm": 26.894297377383214, "learning_rate": 6.667914821220186e-06, "loss": 1.2573, "step": 41000 }, { "epoch": 5.00082, "grad_norm": 24.211135411037418, "learning_rate": 6.6663881128553494e-06, "loss": 1.2648, "step": 41010 }, { "epoch": 5.00092, "grad_norm": 25.49528215391552, "learning_rate": 6.6648612296941335e-06, "loss": 1.4063, "step": 41020 }, { "epoch": 5.00102, "grad_norm": 23.413711603147355, "learning_rate": 6.663334171896702e-06, "loss": 1.3358, "step": 41030 }, { "epoch": 5.00112, "grad_norm": 23.417142964235886, "learning_rate": 6.661806939623235e-06, "loss": 1.2442, "step": 41040 }, { "epoch": 5.00122, "grad_norm": 25.25899720710899, "learning_rate": 6.660279533033933e-06, "loss": 1.295, "step": 41050 }, { "epoch": 5.00132, "grad_norm": 23.04214502703598, "learning_rate": 6.658751952289014e-06, "loss": 1.2811, "step": 41060 }, { "epoch": 5.00142, "grad_norm": 23.70963088087608, "learning_rate": 6.657224197548713e-06, "loss": 1.2945, "step": 41070 }, { "epoch": 5.00152, "grad_norm": 22.648398145732717, "learning_rate": 6.655696268973285e-06, "loss": 1.3023, "step": 41080 }, { "epoch": 5.00162, "grad_norm": 22.76016240562874, "learning_rate": 6.654168166723004e-06, "loss": 1.2856, "step": 41090 }, { "epoch": 5.00172, "grad_norm": 22.469015619855337, "learning_rate": 6.652639890958158e-06, "loss": 1.3202, "step": 41100 }, { "epoch": 5.00182, "grad_norm": 24.710754941462575, "learning_rate": 6.65111144183906e-06, "loss": 1.3086, "step": 41110 }, { "epoch": 5.00192, "grad_norm": 24.727466099212613, "learning_rate": 6.649582819526032e-06, "loss": 1.2614, "step": 41120 }, { "epoch": 5.00202, "grad_norm": 22.521438806364255, "learning_rate": 6.648054024179424e-06, "loss": 1.2415, "step": 41130 }, { "epoch": 5.00212, "grad_norm": 22.86993243323033, "learning_rate": 6.646525055959597e-06, "loss": 1.2536, "step": 41140 }, { "epoch": 5.00222, "grad_norm": 21.816493999062395, "learning_rate": 6.644995915026932e-06, "loss": 1.2562, "step": 41150 }, { "epoch": 5.00232, "grad_norm": 25.892731355218526, "learning_rate": 6.643466601541833e-06, "loss": 1.3249, "step": 41160 }, { "epoch": 5.00242, "grad_norm": 23.93612721536793, "learning_rate": 6.641937115664712e-06, "loss": 1.3256, "step": 41170 }, { "epoch": 5.00252, "grad_norm": 24.75246452636475, "learning_rate": 6.64040745755601e-06, "loss": 1.3439, "step": 41180 }, { "epoch": 5.00262, "grad_norm": 24.702859766109864, "learning_rate": 6.638877627376179e-06, "loss": 1.3405, "step": 41190 }, { "epoch": 5.00272, "grad_norm": 22.981319307003698, "learning_rate": 6.637347625285691e-06, "loss": 1.3185, "step": 41200 }, { "epoch": 5.00282, "grad_norm": 20.955770092686464, "learning_rate": 6.635817451445037e-06, "loss": 1.2806, "step": 41210 }, { "epoch": 5.00292, "grad_norm": 22.741931645115834, "learning_rate": 6.634287106014723e-06, "loss": 1.2934, "step": 41220 }, { "epoch": 5.00302, "grad_norm": 23.664193379596885, "learning_rate": 6.632756589155276e-06, "loss": 1.276, "step": 41230 }, { "epoch": 5.00312, "grad_norm": 20.777679182601712, "learning_rate": 6.631225901027241e-06, "loss": 1.3634, "step": 41240 }, { "epoch": 5.00322, "grad_norm": 21.92443747324613, "learning_rate": 6.629695041791181e-06, "loss": 1.2446, "step": 41250 }, { "epoch": 5.00332, "grad_norm": 22.612017989206066, "learning_rate": 6.628164011607675e-06, "loss": 1.2985, "step": 41260 }, { "epoch": 5.00342, "grad_norm": 24.02350039269511, "learning_rate": 6.626632810637319e-06, "loss": 1.3571, "step": 41270 }, { "epoch": 5.00352, "grad_norm": 22.30405996194042, "learning_rate": 6.625101439040732e-06, "loss": 1.236, "step": 41280 }, { "epoch": 5.00362, "grad_norm": 21.679280180006934, "learning_rate": 6.623569896978546e-06, "loss": 1.3335, "step": 41290 }, { "epoch": 5.00372, "grad_norm": 22.226692276314843, "learning_rate": 6.622038184611412e-06, "loss": 1.3302, "step": 41300 }, { "epoch": 5.00382, "grad_norm": 20.37426832589856, "learning_rate": 6.620506302100003e-06, "loss": 1.2234, "step": 41310 }, { "epoch": 5.00392, "grad_norm": 23.37639163367416, "learning_rate": 6.6189742496050015e-06, "loss": 1.2967, "step": 41320 }, { "epoch": 5.00402, "grad_norm": 23.11535214650993, "learning_rate": 6.617442027287115e-06, "loss": 1.257, "step": 41330 }, { "epoch": 5.00412, "grad_norm": 22.106302632109355, "learning_rate": 6.615909635307068e-06, "loss": 1.2899, "step": 41340 }, { "epoch": 5.00422, "grad_norm": 21.748425858333203, "learning_rate": 6.6143770738255995e-06, "loss": 1.277, "step": 41350 }, { "epoch": 5.00432, "grad_norm": 20.130155273242423, "learning_rate": 6.612844343003469e-06, "loss": 1.3013, "step": 41360 }, { "epoch": 5.00442, "grad_norm": 23.894556652389454, "learning_rate": 6.6113114430014525e-06, "loss": 1.2627, "step": 41370 }, { "epoch": 5.00452, "grad_norm": 23.661079909928883, "learning_rate": 6.609778373980343e-06, "loss": 1.222, "step": 41380 }, { "epoch": 5.00462, "grad_norm": 24.359500210459686, "learning_rate": 6.608245136100954e-06, "loss": 1.2511, "step": 41390 }, { "epoch": 5.00472, "grad_norm": 22.856173794976403, "learning_rate": 6.606711729524114e-06, "loss": 1.2683, "step": 41400 }, { "epoch": 5.00482, "grad_norm": 20.64474916249796, "learning_rate": 6.6051781544106696e-06, "loss": 1.2522, "step": 41410 }, { "epoch": 5.00492, "grad_norm": 20.92613730440595, "learning_rate": 6.603644410921488e-06, "loss": 1.2437, "step": 41420 }, { "epoch": 5.00502, "grad_norm": 23.609209874929242, "learning_rate": 6.60211049921745e-06, "loss": 1.2969, "step": 41430 }, { "epoch": 5.00512, "grad_norm": 23.340887037536096, "learning_rate": 6.600576419459457e-06, "loss": 1.2283, "step": 41440 }, { "epoch": 5.00522, "grad_norm": 24.10964701913, "learning_rate": 6.599042171808425e-06, "loss": 1.2616, "step": 41450 }, { "epoch": 5.00532, "grad_norm": 25.25804388505343, "learning_rate": 6.59750775642529e-06, "loss": 1.2887, "step": 41460 }, { "epoch": 5.00542, "grad_norm": 19.64397346588089, "learning_rate": 6.595973173471008e-06, "loss": 1.242, "step": 41470 }, { "epoch": 5.00552, "grad_norm": 26.13447562174418, "learning_rate": 6.594438423106545e-06, "loss": 1.3162, "step": 41480 }, { "epoch": 5.00562, "grad_norm": 24.885749828938533, "learning_rate": 6.592903505492892e-06, "loss": 1.3568, "step": 41490 }, { "epoch": 5.00572, "grad_norm": 17.809761907827053, "learning_rate": 6.5913684207910546e-06, "loss": 1.2301, "step": 41500 }, { "epoch": 5.00582, "grad_norm": 29.444980536885755, "learning_rate": 6.589833169162055e-06, "loss": 1.2626, "step": 41510 }, { "epoch": 5.00592, "grad_norm": 22.99443681844377, "learning_rate": 6.588297750766936e-06, "loss": 1.2186, "step": 41520 }, { "epoch": 5.00602, "grad_norm": 22.741153679290008, "learning_rate": 6.586762165766752e-06, "loss": 1.217, "step": 41530 }, { "epoch": 5.00612, "grad_norm": 18.86294779485918, "learning_rate": 6.585226414322584e-06, "loss": 1.2936, "step": 41540 }, { "epoch": 5.00622, "grad_norm": 22.028346744063427, "learning_rate": 6.5836904965955205e-06, "loss": 1.2286, "step": 41550 }, { "epoch": 5.00632, "grad_norm": 23.287385299100407, "learning_rate": 6.582154412746675e-06, "loss": 1.2566, "step": 41560 }, { "epoch": 5.00642, "grad_norm": 22.00919826345252, "learning_rate": 6.580618162937175e-06, "loss": 1.2109, "step": 41570 }, { "epoch": 5.00652, "grad_norm": 23.52862460141351, "learning_rate": 6.5790817473281645e-06, "loss": 1.2189, "step": 41580 }, { "epoch": 5.00662, "grad_norm": 19.06168169627471, "learning_rate": 6.577545166080807e-06, "loss": 1.2237, "step": 41590 }, { "epoch": 5.00672, "grad_norm": 23.075379583995264, "learning_rate": 6.576008419356284e-06, "loss": 1.2343, "step": 41600 }, { "epoch": 5.00682, "grad_norm": 26.00338279248607, "learning_rate": 6.574471507315791e-06, "loss": 1.2254, "step": 41610 }, { "epoch": 5.00692, "grad_norm": 21.133397404976556, "learning_rate": 6.572934430120548e-06, "loss": 1.2747, "step": 41620 }, { "epoch": 5.00702, "grad_norm": 24.566640883018966, "learning_rate": 6.571397187931781e-06, "loss": 1.2525, "step": 41630 }, { "epoch": 5.00712, "grad_norm": 21.025184882156783, "learning_rate": 6.5698597809107415e-06, "loss": 1.2816, "step": 41640 }, { "epoch": 5.00722, "grad_norm": 22.95597684044714, "learning_rate": 6.568322209218697e-06, "loss": 1.2314, "step": 41650 }, { "epoch": 5.00732, "grad_norm": 23.95213025243035, "learning_rate": 6.566784473016932e-06, "loss": 1.2646, "step": 41660 }, { "epoch": 5.00742, "grad_norm": 22.610166217281627, "learning_rate": 6.565246572466748e-06, "loss": 1.2683, "step": 41670 }, { "epoch": 5.00752, "grad_norm": 24.171034617630955, "learning_rate": 6.563708507729461e-06, "loss": 1.28, "step": 41680 }, { "epoch": 5.00762, "grad_norm": 24.391470720293285, "learning_rate": 6.562170278966411e-06, "loss": 1.199, "step": 41690 }, { "epoch": 5.00772, "grad_norm": 24.053721883019914, "learning_rate": 6.560631886338948e-06, "loss": 1.2015, "step": 41700 }, { "epoch": 5.00782, "grad_norm": 23.18817707904745, "learning_rate": 6.559093330008445e-06, "loss": 1.167, "step": 41710 }, { "epoch": 5.00792, "grad_norm": 24.235723983207816, "learning_rate": 6.557554610136287e-06, "loss": 1.292, "step": 41720 }, { "epoch": 5.00802, "grad_norm": 23.09663205077612, "learning_rate": 6.556015726883877e-06, "loss": 1.254, "step": 41730 }, { "epoch": 5.00812, "grad_norm": 21.891163650727147, "learning_rate": 6.554476680412641e-06, "loss": 1.2659, "step": 41740 }, { "epoch": 5.00822, "grad_norm": 23.968770465265845, "learning_rate": 6.552937470884016e-06, "loss": 1.2414, "step": 41750 }, { "epoch": 5.00832, "grad_norm": 23.32274376426434, "learning_rate": 6.551398098459457e-06, "loss": 1.1682, "step": 41760 }, { "epoch": 5.00842, "grad_norm": 25.160596443577656, "learning_rate": 6.549858563300438e-06, "loss": 1.234, "step": 41770 }, { "epoch": 5.00852, "grad_norm": 21.1474145749802, "learning_rate": 6.548318865568448e-06, "loss": 1.2755, "step": 41780 }, { "epoch": 5.00862, "grad_norm": 24.157109605162244, "learning_rate": 6.546779005424995e-06, "loss": 1.2546, "step": 41790 }, { "epoch": 5.00872, "grad_norm": 23.23621303159584, "learning_rate": 6.545238983031605e-06, "loss": 1.2584, "step": 41800 }, { "epoch": 5.00882, "grad_norm": 23.41169599237767, "learning_rate": 6.543698798549816e-06, "loss": 1.2406, "step": 41810 }, { "epoch": 5.00892, "grad_norm": 21.850982519709746, "learning_rate": 6.542158452141187e-06, "loss": 1.24, "step": 41820 }, { "epoch": 5.00902, "grad_norm": 24.232093773506953, "learning_rate": 6.540617943967296e-06, "loss": 1.2354, "step": 41830 }, { "epoch": 5.00912, "grad_norm": 23.048656670877794, "learning_rate": 6.5390772741897304e-06, "loss": 1.2999, "step": 41840 }, { "epoch": 5.00922, "grad_norm": 21.01419547782245, "learning_rate": 6.537536442970103e-06, "loss": 1.1616, "step": 41850 }, { "epoch": 5.00932, "grad_norm": 24.428778271840333, "learning_rate": 6.535995450470037e-06, "loss": 1.257, "step": 41860 }, { "epoch": 5.00942, "grad_norm": 21.71732584668655, "learning_rate": 6.534454296851179e-06, "loss": 1.1987, "step": 41870 }, { "epoch": 5.00952, "grad_norm": 23.596148131431928, "learning_rate": 6.532912982275185e-06, "loss": 1.1817, "step": 41880 }, { "epoch": 5.00962, "grad_norm": 23.710011916821827, "learning_rate": 6.531371506903736e-06, "loss": 1.2494, "step": 41890 }, { "epoch": 5.00972, "grad_norm": 26.847808250451006, "learning_rate": 6.529829870898523e-06, "loss": 1.2661, "step": 41900 }, { "epoch": 5.00982, "grad_norm": 23.277426729073788, "learning_rate": 6.528288074421255e-06, "loss": 1.341, "step": 41910 }, { "epoch": 5.00992, "grad_norm": 21.856346088984836, "learning_rate": 6.526746117633662e-06, "loss": 1.2066, "step": 41920 }, { "epoch": 5.01002, "grad_norm": 23.492701070441736, "learning_rate": 6.525204000697487e-06, "loss": 1.2255, "step": 41930 }, { "epoch": 5.01012, "grad_norm": 23.006988718630737, "learning_rate": 6.5236617237744906e-06, "loss": 1.1997, "step": 41940 }, { "epoch": 5.01022, "grad_norm": 22.71314812433456, "learning_rate": 6.522119287026452e-06, "loss": 1.2655, "step": 41950 }, { "epoch": 5.01032, "grad_norm": 22.82400046692311, "learning_rate": 6.520576690615163e-06, "loss": 1.2455, "step": 41960 }, { "epoch": 5.01042, "grad_norm": 24.3401425548907, "learning_rate": 6.519033934702437e-06, "loss": 1.2447, "step": 41970 }, { "epoch": 5.01052, "grad_norm": 25.18207760858508, "learning_rate": 6.517491019450103e-06, "loss": 1.2449, "step": 41980 }, { "epoch": 5.01062, "grad_norm": 23.210956757452664, "learning_rate": 6.515947945020001e-06, "loss": 1.2403, "step": 41990 }, { "epoch": 5.01072, "grad_norm": 24.400617421825327, "learning_rate": 6.514404711573998e-06, "loss": 1.2878, "step": 42000 }, { "epoch": 5.01082, "grad_norm": 23.365472115321623, "learning_rate": 6.5128613192739685e-06, "loss": 1.2186, "step": 42010 }, { "epoch": 5.01092, "grad_norm": 20.98196242838481, "learning_rate": 6.511317768281806e-06, "loss": 1.1882, "step": 42020 }, { "epoch": 5.01102, "grad_norm": 19.503974753070267, "learning_rate": 6.509774058759426e-06, "loss": 1.1479, "step": 42030 }, { "epoch": 5.01112, "grad_norm": 22.65851834958459, "learning_rate": 6.508230190868754e-06, "loss": 1.2614, "step": 42040 }, { "epoch": 5.01122, "grad_norm": 24.807059736847147, "learning_rate": 6.506686164771735e-06, "loss": 1.121, "step": 42050 }, { "epoch": 5.01132, "grad_norm": 23.612785936049995, "learning_rate": 6.50514198063033e-06, "loss": 1.208, "step": 42060 }, { "epoch": 5.01142, "grad_norm": 23.341978091700714, "learning_rate": 6.503597638606516e-06, "loss": 1.2494, "step": 42070 }, { "epoch": 5.01152, "grad_norm": 27.36397098298293, "learning_rate": 6.502053138862289e-06, "loss": 1.1864, "step": 42080 }, { "epoch": 5.01162, "grad_norm": 21.954332767840818, "learning_rate": 6.500508481559659e-06, "loss": 1.1748, "step": 42090 }, { "epoch": 5.01172, "grad_norm": 23.42400675405302, "learning_rate": 6.498963666860653e-06, "loss": 1.2184, "step": 42100 }, { "epoch": 5.01182, "grad_norm": 22.424952531312236, "learning_rate": 6.497418694927314e-06, "loss": 1.187, "step": 42110 }, { "epoch": 5.01192, "grad_norm": 21.583686316524574, "learning_rate": 6.4958735659217045e-06, "loss": 1.168, "step": 42120 }, { "epoch": 5.01202, "grad_norm": 23.29301723468967, "learning_rate": 6.494328280005901e-06, "loss": 1.2377, "step": 42130 }, { "epoch": 5.01212, "grad_norm": 21.879259991809857, "learning_rate": 6.4927828373419966e-06, "loss": 1.2907, "step": 42140 }, { "epoch": 5.01222, "grad_norm": 19.167449152248707, "learning_rate": 6.491237238092101e-06, "loss": 1.1587, "step": 42150 }, { "epoch": 5.01232, "grad_norm": 26.246179512837596, "learning_rate": 6.4896914824183395e-06, "loss": 1.303, "step": 42160 }, { "epoch": 5.01242, "grad_norm": 20.238460270879926, "learning_rate": 6.488145570482856e-06, "loss": 1.2449, "step": 42170 }, { "epoch": 5.01252, "grad_norm": 23.942381274481694, "learning_rate": 6.486599502447808e-06, "loss": 1.2718, "step": 42180 }, { "epoch": 5.01262, "grad_norm": 22.629745158153636, "learning_rate": 6.485053278475372e-06, "loss": 1.1952, "step": 42190 }, { "epoch": 5.01272, "grad_norm": 24.318969617876068, "learning_rate": 6.4835068987277395e-06, "loss": 1.1762, "step": 42200 }, { "epoch": 5.01282, "grad_norm": 23.407543058367327, "learning_rate": 6.481960363367118e-06, "loss": 1.2551, "step": 42210 }, { "epoch": 5.01292, "grad_norm": 22.158085908975885, "learning_rate": 6.480413672555732e-06, "loss": 1.2245, "step": 42220 }, { "epoch": 5.01302, "grad_norm": 22.00924439482966, "learning_rate": 6.4788668264558245e-06, "loss": 1.1923, "step": 42230 }, { "epoch": 5.01312, "grad_norm": 22.966354330179872, "learning_rate": 6.4773198252296485e-06, "loss": 1.1694, "step": 42240 }, { "epoch": 5.01322, "grad_norm": 24.505543486921276, "learning_rate": 6.475772669039481e-06, "loss": 1.2148, "step": 42250 }, { "epoch": 5.01332, "grad_norm": 23.61780408293282, "learning_rate": 6.474225358047609e-06, "loss": 1.2183, "step": 42260 }, { "epoch": 5.01342, "grad_norm": 22.983361061746358, "learning_rate": 6.472677892416339e-06, "loss": 1.2079, "step": 42270 }, { "epoch": 5.01352, "grad_norm": 21.34631634374378, "learning_rate": 6.471130272307994e-06, "loss": 1.2217, "step": 42280 }, { "epoch": 5.01362, "grad_norm": 22.591646751387675, "learning_rate": 6.469582497884909e-06, "loss": 1.2152, "step": 42290 }, { "epoch": 5.01372, "grad_norm": 21.424262478518994, "learning_rate": 6.468034569309441e-06, "loss": 1.2502, "step": 42300 }, { "epoch": 5.01382, "grad_norm": 21.489372442961137, "learning_rate": 6.466486486743961e-06, "loss": 1.2354, "step": 42310 }, { "epoch": 5.01392, "grad_norm": 25.354670174491435, "learning_rate": 6.464938250350854e-06, "loss": 1.1947, "step": 42320 }, { "epoch": 5.01402, "grad_norm": 20.888583136113724, "learning_rate": 6.463389860292525e-06, "loss": 1.1162, "step": 42330 }, { "epoch": 5.01412, "grad_norm": 20.325280975782896, "learning_rate": 6.46184131673139e-06, "loss": 1.2155, "step": 42340 }, { "epoch": 5.01422, "grad_norm": 20.791256660678524, "learning_rate": 6.460292619829886e-06, "loss": 1.2121, "step": 42350 }, { "epoch": 5.01432, "grad_norm": 22.468311258156067, "learning_rate": 6.458743769750464e-06, "loss": 1.2166, "step": 42360 }, { "epoch": 5.01442, "grad_norm": 24.26770660378827, "learning_rate": 6.4571947666555904e-06, "loss": 1.2656, "step": 42370 }, { "epoch": 5.01452, "grad_norm": 23.886505592977027, "learning_rate": 6.455645610707748e-06, "loss": 1.2235, "step": 42380 }, { "epoch": 5.01462, "grad_norm": 21.555994214251868, "learning_rate": 6.454096302069437e-06, "loss": 1.2403, "step": 42390 }, { "epoch": 5.01472, "grad_norm": 22.978951740934164, "learning_rate": 6.452546840903174e-06, "loss": 1.2143, "step": 42400 }, { "epoch": 5.01482, "grad_norm": 23.610643776428745, "learning_rate": 6.450997227371488e-06, "loss": 1.3206, "step": 42410 }, { "epoch": 5.01492, "grad_norm": 21.83806399379792, "learning_rate": 6.449447461636927e-06, "loss": 1.225, "step": 42420 }, { "epoch": 5.01502, "grad_norm": 22.004858292644883, "learning_rate": 6.447897543862055e-06, "loss": 1.1963, "step": 42430 }, { "epoch": 5.01512, "grad_norm": 23.449163371632498, "learning_rate": 6.44634747420945e-06, "loss": 1.2775, "step": 42440 }, { "epoch": 5.01522, "grad_norm": 25.22684108505738, "learning_rate": 6.444797252841708e-06, "loss": 1.2754, "step": 42450 }, { "epoch": 5.01532, "grad_norm": 25.32704615062123, "learning_rate": 6.443246879921439e-06, "loss": 1.2513, "step": 42460 }, { "epoch": 5.01542, "grad_norm": 19.148486215217222, "learning_rate": 6.441696355611271e-06, "loss": 1.2744, "step": 42470 }, { "epoch": 5.01552, "grad_norm": 24.908595695596286, "learning_rate": 6.440145680073847e-06, "loss": 1.3278, "step": 42480 }, { "epoch": 5.01562, "grad_norm": 22.87347128682893, "learning_rate": 6.438594853471825e-06, "loss": 1.2445, "step": 42490 }, { "epoch": 5.01572, "grad_norm": 21.40127211866771, "learning_rate": 6.437043875967878e-06, "loss": 1.2621, "step": 42500 }, { "epoch": 5.01582, "grad_norm": 24.374425311020307, "learning_rate": 6.4354927477247e-06, "loss": 1.1952, "step": 42510 }, { "epoch": 5.01592, "grad_norm": 22.00540341317859, "learning_rate": 6.433941468904994e-06, "loss": 1.2126, "step": 42520 }, { "epoch": 5.01602, "grad_norm": 21.663842529393605, "learning_rate": 6.432390039671483e-06, "loss": 1.1758, "step": 42530 }, { "epoch": 5.01612, "grad_norm": 22.065761501737477, "learning_rate": 6.430838460186906e-06, "loss": 1.2316, "step": 42540 }, { "epoch": 5.01622, "grad_norm": 22.39036317227215, "learning_rate": 6.429286730614014e-06, "loss": 1.2366, "step": 42550 }, { "epoch": 5.01632, "grad_norm": 22.270841598548465, "learning_rate": 6.427734851115577e-06, "loss": 1.1948, "step": 42560 }, { "epoch": 5.01642, "grad_norm": 23.746520395421935, "learning_rate": 6.426182821854381e-06, "loss": 1.1839, "step": 42570 }, { "epoch": 5.01652, "grad_norm": 24.9008843826623, "learning_rate": 6.424630642993226e-06, "loss": 1.2473, "step": 42580 }, { "epoch": 5.01662, "grad_norm": 25.662681774403143, "learning_rate": 6.423078314694929e-06, "loss": 1.3221, "step": 42590 }, { "epoch": 5.01672, "grad_norm": 25.64642647096655, "learning_rate": 6.421525837122321e-06, "loss": 1.2772, "step": 42600 }, { "epoch": 5.01682, "grad_norm": 22.907276801814085, "learning_rate": 6.419973210438252e-06, "loss": 1.1802, "step": 42610 }, { "epoch": 5.01692, "grad_norm": 20.960486634967005, "learning_rate": 6.418420434805581e-06, "loss": 1.2099, "step": 42620 }, { "epoch": 5.01702, "grad_norm": 21.655743911087452, "learning_rate": 6.41686751038719e-06, "loss": 1.1885, "step": 42630 }, { "epoch": 5.01712, "grad_norm": 23.056408259061577, "learning_rate": 6.415314437345974e-06, "loss": 1.2057, "step": 42640 }, { "epoch": 5.01722, "grad_norm": 22.19524235033613, "learning_rate": 6.413761215844842e-06, "loss": 1.2213, "step": 42650 }, { "epoch": 5.01732, "grad_norm": 24.829744240435847, "learning_rate": 6.41220784604672e-06, "loss": 1.1879, "step": 42660 }, { "epoch": 5.01742, "grad_norm": 21.21788990678579, "learning_rate": 6.410654328114548e-06, "loss": 1.2158, "step": 42670 }, { "epoch": 5.01752, "grad_norm": 22.35778258044918, "learning_rate": 6.409100662211285e-06, "loss": 1.2137, "step": 42680 }, { "epoch": 5.01762, "grad_norm": 23.36632767412862, "learning_rate": 6.4075468484999025e-06, "loss": 1.1862, "step": 42690 }, { "epoch": 5.01772, "grad_norm": 23.6250144415512, "learning_rate": 6.405992887143388e-06, "loss": 1.1938, "step": 42700 }, { "epoch": 5.01782, "grad_norm": 21.458981348418565, "learning_rate": 6.404438778304745e-06, "loss": 1.1191, "step": 42710 }, { "epoch": 5.01792, "grad_norm": 19.74096273611355, "learning_rate": 6.402884522146993e-06, "loss": 1.213, "step": 42720 }, { "epoch": 5.01802, "grad_norm": 20.48056199070827, "learning_rate": 6.401330118833165e-06, "loss": 1.1693, "step": 42730 }, { "epoch": 5.01812, "grad_norm": 20.504965175342004, "learning_rate": 6.399775568526311e-06, "loss": 1.1646, "step": 42740 }, { "epoch": 5.01822, "grad_norm": 23.261761517094385, "learning_rate": 6.398220871389495e-06, "loss": 1.1604, "step": 42750 }, { "epoch": 5.01832, "grad_norm": 23.805367417614008, "learning_rate": 6.396666027585803e-06, "loss": 1.2355, "step": 42760 }, { "epoch": 5.01842, "grad_norm": 21.49617954209649, "learning_rate": 6.395111037278323e-06, "loss": 1.2444, "step": 42770 }, { "epoch": 5.01852, "grad_norm": 20.75031727707638, "learning_rate": 6.393555900630172e-06, "loss": 1.1802, "step": 42780 }, { "epoch": 5.01862, "grad_norm": 23.687811914608243, "learning_rate": 6.392000617804475e-06, "loss": 1.2107, "step": 42790 }, { "epoch": 5.01872, "grad_norm": 23.02370523022832, "learning_rate": 6.390445188964372e-06, "loss": 1.1872, "step": 42800 }, { "epoch": 5.01882, "grad_norm": 22.806631897618015, "learning_rate": 6.388889614273023e-06, "loss": 1.1918, "step": 42810 }, { "epoch": 5.01892, "grad_norm": 24.807264900620595, "learning_rate": 6.387333893893599e-06, "loss": 1.186, "step": 42820 }, { "epoch": 5.01902, "grad_norm": 24.81927326779185, "learning_rate": 6.385778027989289e-06, "loss": 1.1722, "step": 42830 }, { "epoch": 5.01912, "grad_norm": 22.705586576869507, "learning_rate": 6.384222016723296e-06, "loss": 1.2164, "step": 42840 }, { "epoch": 5.01922, "grad_norm": 24.193786977148648, "learning_rate": 6.3826658602588365e-06, "loss": 1.1918, "step": 42850 }, { "epoch": 5.01932, "grad_norm": 23.268173280539937, "learning_rate": 6.381109558759146e-06, "loss": 1.1936, "step": 42860 }, { "epoch": 5.01942, "grad_norm": 23.90751727497973, "learning_rate": 6.3795531123874736e-06, "loss": 1.2112, "step": 42870 }, { "epoch": 5.01952, "grad_norm": 23.784745552660326, "learning_rate": 6.37799652130708e-06, "loss": 1.2019, "step": 42880 }, { "epoch": 5.01962, "grad_norm": 25.270793010770277, "learning_rate": 6.37643978568125e-06, "loss": 1.1916, "step": 42890 }, { "epoch": 5.01972, "grad_norm": 23.99233943750738, "learning_rate": 6.374882905673272e-06, "loss": 1.1558, "step": 42900 }, { "epoch": 5.01982, "grad_norm": 25.29535282243607, "learning_rate": 6.373325881446459e-06, "loss": 1.2216, "step": 42910 }, { "epoch": 5.01992, "grad_norm": 24.039543733122898, "learning_rate": 6.371768713164135e-06, "loss": 1.1918, "step": 42920 }, { "epoch": 5.02002, "grad_norm": 23.915398472351615, "learning_rate": 6.370211400989639e-06, "loss": 1.2107, "step": 42930 }, { "epoch": 5.02012, "grad_norm": 21.952433214076226, "learning_rate": 6.3686539450863286e-06, "loss": 1.1918, "step": 42940 }, { "epoch": 5.02022, "grad_norm": 23.98676127017123, "learning_rate": 6.367096345617569e-06, "loss": 1.1998, "step": 42950 }, { "epoch": 5.02032, "grad_norm": 21.780543433787702, "learning_rate": 6.365538602746748e-06, "loss": 1.1924, "step": 42960 }, { "epoch": 5.02042, "grad_norm": 23.588644659539636, "learning_rate": 6.363980716637267e-06, "loss": 1.2051, "step": 42970 }, { "epoch": 5.02052, "grad_norm": 22.170871814829635, "learning_rate": 6.362422687452538e-06, "loss": 1.2049, "step": 42980 }, { "epoch": 5.02062, "grad_norm": 26.358121103093907, "learning_rate": 6.3608645153559915e-06, "loss": 1.216, "step": 42990 }, { "epoch": 5.02072, "grad_norm": 22.67448325053101, "learning_rate": 6.359306200511074e-06, "loss": 1.2748, "step": 43000 }, { "epoch": 5.02082, "grad_norm": 24.353419756575477, "learning_rate": 6.357747743081244e-06, "loss": 1.2144, "step": 43010 }, { "epoch": 5.02092, "grad_norm": 22.918198543777248, "learning_rate": 6.356189143229979e-06, "loss": 1.2562, "step": 43020 }, { "epoch": 5.02102, "grad_norm": 20.838169751914663, "learning_rate": 6.354630401120766e-06, "loss": 1.2315, "step": 43030 }, { "epoch": 5.02112, "grad_norm": 23.607542666463697, "learning_rate": 6.353071516917112e-06, "loss": 1.1714, "step": 43040 }, { "epoch": 5.02122, "grad_norm": 23.20747853457971, "learning_rate": 6.351512490782534e-06, "loss": 1.2386, "step": 43050 }, { "epoch": 5.02132, "grad_norm": 24.473108803937116, "learning_rate": 6.34995332288057e-06, "loss": 1.1787, "step": 43060 }, { "epoch": 5.02142, "grad_norm": 23.88366263117835, "learning_rate": 6.348394013374768e-06, "loss": 1.1859, "step": 43070 }, { "epoch": 5.02152, "grad_norm": 23.421658967693972, "learning_rate": 6.34683456242869e-06, "loss": 1.1687, "step": 43080 }, { "epoch": 5.02162, "grad_norm": 22.17162162037918, "learning_rate": 6.345274970205918e-06, "loss": 1.2098, "step": 43090 }, { "epoch": 5.02172, "grad_norm": 23.72910285834958, "learning_rate": 6.343715236870045e-06, "loss": 1.2541, "step": 43100 }, { "epoch": 5.02182, "grad_norm": 23.757666874211385, "learning_rate": 6.342155362584681e-06, "loss": 1.2655, "step": 43110 }, { "epoch": 5.02192, "grad_norm": 21.24036947093891, "learning_rate": 6.340595347513449e-06, "loss": 1.1745, "step": 43120 }, { "epoch": 5.02202, "grad_norm": 24.417009748839956, "learning_rate": 6.3390351918199856e-06, "loss": 1.2508, "step": 43130 }, { "epoch": 5.02212, "grad_norm": 21.094184801304117, "learning_rate": 6.337474895667946e-06, "loss": 1.2081, "step": 43140 }, { "epoch": 5.02222, "grad_norm": 23.88564461114961, "learning_rate": 6.3359144592209975e-06, "loss": 1.2081, "step": 43150 }, { "epoch": 5.02232, "grad_norm": 25.47045472085474, "learning_rate": 6.334353882642821e-06, "loss": 1.2164, "step": 43160 }, { "epoch": 5.02242, "grad_norm": 24.20861939292341, "learning_rate": 6.3327931660971154e-06, "loss": 1.2323, "step": 43170 }, { "epoch": 5.02252, "grad_norm": 23.416077348991035, "learning_rate": 6.331232309747594e-06, "loss": 1.1846, "step": 43180 }, { "epoch": 5.02262, "grad_norm": 24.871989035394332, "learning_rate": 6.3296713137579795e-06, "loss": 1.2188, "step": 43190 }, { "epoch": 5.02272, "grad_norm": 28.137230599759235, "learning_rate": 6.328110178292017e-06, "loss": 1.2558, "step": 43200 }, { "epoch": 5.02282, "grad_norm": 24.09239727411606, "learning_rate": 6.326548903513461e-06, "loss": 1.2273, "step": 43210 }, { "epoch": 5.02292, "grad_norm": 22.282103142968644, "learning_rate": 6.324987489586082e-06, "loss": 1.2096, "step": 43220 }, { "epoch": 5.02302, "grad_norm": 25.57760863184832, "learning_rate": 6.323425936673663e-06, "loss": 1.2502, "step": 43230 }, { "epoch": 5.02312, "grad_norm": 27.739301766586095, "learning_rate": 6.321864244940006e-06, "loss": 1.2874, "step": 43240 }, { "epoch": 5.02322, "grad_norm": 23.74451478519033, "learning_rate": 6.3203024145489256e-06, "loss": 1.2053, "step": 43250 }, { "epoch": 5.02332, "grad_norm": 23.846722510755125, "learning_rate": 6.3187404456642486e-06, "loss": 1.2908, "step": 43260 }, { "epoch": 5.02342, "grad_norm": 24.110085934339956, "learning_rate": 6.31717833844982e-06, "loss": 1.2129, "step": 43270 }, { "epoch": 5.02352, "grad_norm": 24.20481359695791, "learning_rate": 6.3156160930694965e-06, "loss": 1.2204, "step": 43280 }, { "epoch": 5.02362, "grad_norm": 24.036918626620302, "learning_rate": 6.31405370968715e-06, "loss": 1.263, "step": 43290 }, { "epoch": 5.02372, "grad_norm": 21.62898174878675, "learning_rate": 6.312491188466671e-06, "loss": 1.246, "step": 43300 }, { "epoch": 5.02382, "grad_norm": 25.378598769099916, "learning_rate": 6.310928529571954e-06, "loss": 1.2854, "step": 43310 }, { "epoch": 5.02392, "grad_norm": 25.07256392159234, "learning_rate": 6.3093657331669224e-06, "loss": 1.2957, "step": 43320 }, { "epoch": 5.02402, "grad_norm": 25.13321377051187, "learning_rate": 6.307802799415499e-06, "loss": 1.2395, "step": 43330 }, { "epoch": 5.02412, "grad_norm": 24.50401242346228, "learning_rate": 6.306239728481633e-06, "loss": 1.2476, "step": 43340 }, { "epoch": 5.02422, "grad_norm": 24.291095810350857, "learning_rate": 6.304676520529282e-06, "loss": 1.204, "step": 43350 }, { "epoch": 5.02432, "grad_norm": 26.747200776031956, "learning_rate": 6.303113175722418e-06, "loss": 1.2178, "step": 43360 }, { "epoch": 5.02442, "grad_norm": 24.230633961063646, "learning_rate": 6.301549694225032e-06, "loss": 1.1895, "step": 43370 }, { "epoch": 5.02452, "grad_norm": 23.454941749377003, "learning_rate": 6.299986076201123e-06, "loss": 1.2081, "step": 43380 }, { "epoch": 5.02462, "grad_norm": 23.77018608158629, "learning_rate": 6.298422321814706e-06, "loss": 1.1501, "step": 43390 }, { "epoch": 5.02472, "grad_norm": 26.225236322452115, "learning_rate": 6.2968584312298164e-06, "loss": 1.2099, "step": 43400 }, { "epoch": 5.02482, "grad_norm": 23.510699499243724, "learning_rate": 6.295294404610495e-06, "loss": 1.2623, "step": 43410 }, { "epoch": 5.02492, "grad_norm": 21.54617232245423, "learning_rate": 6.293730242120801e-06, "loss": 1.207, "step": 43420 }, { "epoch": 5.02502, "grad_norm": 24.45504214537926, "learning_rate": 6.292165943924812e-06, "loss": 1.1514, "step": 43430 }, { "epoch": 5.02512, "grad_norm": 22.909593262511, "learning_rate": 6.29060151018661e-06, "loss": 1.1649, "step": 43440 }, { "epoch": 5.02522, "grad_norm": 23.472986517789934, "learning_rate": 6.2890369410703e-06, "loss": 1.1006, "step": 43450 }, { "epoch": 5.02532, "grad_norm": 22.09452101485283, "learning_rate": 6.287472236739998e-06, "loss": 1.1564, "step": 43460 }, { "epoch": 5.02542, "grad_norm": 25.640865939441188, "learning_rate": 6.285907397359833e-06, "loss": 1.1607, "step": 43470 }, { "epoch": 5.02552, "grad_norm": 24.008389593670046, "learning_rate": 6.284342423093951e-06, "loss": 1.1699, "step": 43480 }, { "epoch": 5.02562, "grad_norm": 22.526580516707135, "learning_rate": 6.282777314106509e-06, "loss": 1.1938, "step": 43490 }, { "epoch": 5.02572, "grad_norm": 23.084861588027014, "learning_rate": 6.281212070561682e-06, "loss": 1.1949, "step": 43500 }, { "epoch": 5.02582, "grad_norm": 23.186507672340447, "learning_rate": 6.279646692623653e-06, "loss": 1.1859, "step": 43510 }, { "epoch": 5.02592, "grad_norm": 22.079033887802396, "learning_rate": 6.278081180456625e-06, "loss": 1.1835, "step": 43520 }, { "epoch": 5.02602, "grad_norm": 25.035812706984235, "learning_rate": 6.276515534224813e-06, "loss": 1.1569, "step": 43530 }, { "epoch": 5.02612, "grad_norm": 24.514743300409794, "learning_rate": 6.2749497540924466e-06, "loss": 1.2621, "step": 43540 }, { "epoch": 5.02622, "grad_norm": 26.344261481946294, "learning_rate": 6.273383840223768e-06, "loss": 1.2273, "step": 43550 }, { "epoch": 5.02632, "grad_norm": 24.258367233688727, "learning_rate": 6.2718177927830345e-06, "loss": 1.1516, "step": 43560 }, { "epoch": 5.02642, "grad_norm": 24.16057474793522, "learning_rate": 6.270251611934517e-06, "loss": 1.2164, "step": 43570 }, { "epoch": 5.02652, "grad_norm": 23.688090342814935, "learning_rate": 6.268685297842503e-06, "loss": 1.1683, "step": 43580 }, { "epoch": 5.02662, "grad_norm": 19.08027323060075, "learning_rate": 6.267118850671286e-06, "loss": 1.1451, "step": 43590 }, { "epoch": 5.02672, "grad_norm": 23.870484913793575, "learning_rate": 6.265552270585186e-06, "loss": 1.149, "step": 43600 }, { "epoch": 5.02682, "grad_norm": 22.07005258309239, "learning_rate": 6.2639855577485235e-06, "loss": 1.1714, "step": 43610 }, { "epoch": 5.02692, "grad_norm": 23.31630462056947, "learning_rate": 6.262418712325643e-06, "loss": 1.1833, "step": 43620 }, { "epoch": 5.02702, "grad_norm": 24.583894672596635, "learning_rate": 6.260851734480898e-06, "loss": 1.2475, "step": 43630 }, { "epoch": 5.02712, "grad_norm": 23.071853168028127, "learning_rate": 6.259284624378659e-06, "loss": 1.1981, "step": 43640 }, { "epoch": 5.02722, "grad_norm": 24.16484776957138, "learning_rate": 6.257717382183308e-06, "loss": 1.1532, "step": 43650 }, { "epoch": 5.02732, "grad_norm": 23.906051619577788, "learning_rate": 6.25615000805924e-06, "loss": 1.2423, "step": 43660 }, { "epoch": 5.02742, "grad_norm": 23.419318909959284, "learning_rate": 6.254582502170865e-06, "loss": 1.1454, "step": 43670 }, { "epoch": 5.02752, "grad_norm": 22.85413111396266, "learning_rate": 6.253014864682611e-06, "loss": 1.122, "step": 43680 }, { "epoch": 5.02762, "grad_norm": 24.51619840758299, "learning_rate": 6.251447095758911e-06, "loss": 1.2087, "step": 43690 }, { "epoch": 5.02772, "grad_norm": 22.381083752216483, "learning_rate": 6.249879195564219e-06, "loss": 1.1578, "step": 43700 }, { "epoch": 5.02782, "grad_norm": 25.80219308926013, "learning_rate": 6.248311164263e-06, "loss": 1.1936, "step": 43710 }, { "epoch": 5.02792, "grad_norm": 25.851583524654313, "learning_rate": 6.246743002019734e-06, "loss": 1.2818, "step": 43720 }, { "epoch": 5.02802, "grad_norm": 21.118188393039155, "learning_rate": 6.245174708998914e-06, "loss": 1.1344, "step": 43730 }, { "epoch": 5.02812, "grad_norm": 22.957914398097763, "learning_rate": 6.243606285365045e-06, "loss": 1.188, "step": 43740 }, { "epoch": 5.02822, "grad_norm": 23.411610850480454, "learning_rate": 6.242037731282648e-06, "loss": 1.1628, "step": 43750 }, { "epoch": 5.02832, "grad_norm": 22.66301576811053, "learning_rate": 6.24046904691626e-06, "loss": 1.1145, "step": 43760 }, { "epoch": 5.02842, "grad_norm": 23.461258539699127, "learning_rate": 6.238900232430424e-06, "loss": 1.2152, "step": 43770 }, { "epoch": 5.02852, "grad_norm": 27.284360482547594, "learning_rate": 6.2373312879897056e-06, "loss": 1.147, "step": 43780 }, { "epoch": 5.02862, "grad_norm": 24.65511717515067, "learning_rate": 6.235762213758675e-06, "loss": 1.1822, "step": 43790 }, { "epoch": 5.02872, "grad_norm": 21.260641015789943, "learning_rate": 6.234193009901924e-06, "loss": 1.1821, "step": 43800 }, { "epoch": 5.02882, "grad_norm": 25.92121623982583, "learning_rate": 6.232623676584056e-06, "loss": 1.1495, "step": 43810 }, { "epoch": 5.02892, "grad_norm": 22.21412327000642, "learning_rate": 6.231054213969683e-06, "loss": 1.232, "step": 43820 }, { "epoch": 5.02902, "grad_norm": 23.288094150476415, "learning_rate": 6.2294846222234385e-06, "loss": 1.1444, "step": 43830 }, { "epoch": 5.02912, "grad_norm": 22.94457194340588, "learning_rate": 6.227914901509962e-06, "loss": 1.1499, "step": 43840 }, { "epoch": 5.02922, "grad_norm": 24.380281143861378, "learning_rate": 6.226345051993912e-06, "loss": 1.2715, "step": 43850 }, { "epoch": 5.02932, "grad_norm": 23.643769354147608, "learning_rate": 6.224775073839958e-06, "loss": 1.1977, "step": 43860 }, { "epoch": 5.02942, "grad_norm": 24.217729325983022, "learning_rate": 6.223204967212781e-06, "loss": 1.1692, "step": 43870 }, { "epoch": 5.02952, "grad_norm": 24.58631424317061, "learning_rate": 6.221634732277082e-06, "loss": 1.1598, "step": 43880 }, { "epoch": 5.02962, "grad_norm": 21.916942968342404, "learning_rate": 6.220064369197569e-06, "loss": 1.1271, "step": 43890 }, { "epoch": 5.02972, "grad_norm": 23.0935825227797, "learning_rate": 6.218493878138966e-06, "loss": 1.1858, "step": 43900 }, { "epoch": 5.02982, "grad_norm": 23.334066104638662, "learning_rate": 6.21692325926601e-06, "loss": 1.1713, "step": 43910 }, { "epoch": 5.02992, "grad_norm": 23.559980465237818, "learning_rate": 6.215352512743453e-06, "loss": 1.1404, "step": 43920 }, { "epoch": 5.03002, "grad_norm": 24.69508056011014, "learning_rate": 6.213781638736058e-06, "loss": 1.0636, "step": 43930 }, { "epoch": 5.03012, "grad_norm": 21.475705513160126, "learning_rate": 6.212210637408602e-06, "loss": 1.0528, "step": 43940 }, { "epoch": 5.03022, "grad_norm": 23.12262850591642, "learning_rate": 6.210639508925876e-06, "loss": 1.106, "step": 43950 }, { "epoch": 5.03032, "grad_norm": 21.441306654939005, "learning_rate": 6.209068253452684e-06, "loss": 1.1361, "step": 43960 }, { "epoch": 5.03042, "grad_norm": 24.976649458004257, "learning_rate": 6.2074968711538445e-06, "loss": 1.0993, "step": 43970 }, { "epoch": 5.03052, "grad_norm": 25.911576124354063, "learning_rate": 6.205925362194186e-06, "loss": 1.1769, "step": 43980 }, { "epoch": 5.03062, "grad_norm": 20.23277439742636, "learning_rate": 6.204353726738555e-06, "loss": 1.1213, "step": 43990 }, { "epoch": 5.03072, "grad_norm": 25.92605875511962, "learning_rate": 6.202781964951807e-06, "loss": 1.1702, "step": 44000 }, { "epoch": 5.03082, "grad_norm": 26.78905670909784, "learning_rate": 6.201210076998815e-06, "loss": 1.1666, "step": 44010 }, { "epoch": 5.03092, "grad_norm": 24.76004260498489, "learning_rate": 6.199638063044459e-06, "loss": 1.1577, "step": 44020 }, { "epoch": 5.03102, "grad_norm": 21.251717890849584, "learning_rate": 6.198065923253637e-06, "loss": 1.058, "step": 44030 }, { "epoch": 5.03112, "grad_norm": 24.135878098514652, "learning_rate": 6.196493657791263e-06, "loss": 1.0247, "step": 44040 }, { "epoch": 5.03122, "grad_norm": 20.51331174633403, "learning_rate": 6.194921266822256e-06, "loss": 1.1154, "step": 44050 }, { "epoch": 5.03132, "grad_norm": 23.01229896093519, "learning_rate": 6.193348750511553e-06, "loss": 1.1276, "step": 44060 }, { "epoch": 5.03142, "grad_norm": 23.671994430199234, "learning_rate": 6.191776109024104e-06, "loss": 1.0406, "step": 44070 }, { "epoch": 5.03152, "grad_norm": 19.218089359139043, "learning_rate": 6.190203342524874e-06, "loss": 1.0253, "step": 44080 }, { "epoch": 5.03162, "grad_norm": 28.611196365202932, "learning_rate": 6.188630451178837e-06, "loss": 1.0993, "step": 44090 }, { "epoch": 5.03172, "grad_norm": 27.32851259668026, "learning_rate": 6.187057435150983e-06, "loss": 1.0807, "step": 44100 }, { "epoch": 5.03182, "grad_norm": 23.35349402826335, "learning_rate": 6.185484294606313e-06, "loss": 1.1015, "step": 44110 }, { "epoch": 5.03192, "grad_norm": 25.943604458745643, "learning_rate": 6.183911029709842e-06, "loss": 1.1031, "step": 44120 }, { "epoch": 5.03202, "grad_norm": 25.127634340995357, "learning_rate": 6.182337640626599e-06, "loss": 1.0787, "step": 44130 }, { "epoch": 5.03212, "grad_norm": 23.08603560441645, "learning_rate": 6.180764127521624e-06, "loss": 1.0647, "step": 44140 }, { "epoch": 5.03222, "grad_norm": 23.06166987571145, "learning_rate": 6.179190490559973e-06, "loss": 1.0476, "step": 44150 }, { "epoch": 5.03232, "grad_norm": 21.313359359903018, "learning_rate": 6.177616729906712e-06, "loss": 1.0675, "step": 44160 }, { "epoch": 5.03242, "grad_norm": 20.341337762512048, "learning_rate": 6.176042845726922e-06, "loss": 0.9981, "step": 44170 }, { "epoch": 5.03252, "grad_norm": 24.024736558592405, "learning_rate": 6.174468838185694e-06, "loss": 1.0655, "step": 44180 }, { "epoch": 5.03262, "grad_norm": 23.0892346419284, "learning_rate": 6.172894707448138e-06, "loss": 1.0672, "step": 44190 }, { "epoch": 5.03272, "grad_norm": 21.452675233636125, "learning_rate": 6.1713204536793705e-06, "loss": 0.9815, "step": 44200 }, { "epoch": 5.03282, "grad_norm": 22.636182805550686, "learning_rate": 6.169746077044523e-06, "loss": 1.0531, "step": 44210 }, { "epoch": 5.03292, "grad_norm": 22.338715933826947, "learning_rate": 6.1681715777087415e-06, "loss": 1.016, "step": 44220 }, { "epoch": 5.03302, "grad_norm": 22.36665988487724, "learning_rate": 6.1665969558371816e-06, "loss": 1.0328, "step": 44230 }, { "epoch": 5.03312, "grad_norm": 21.620194986459477, "learning_rate": 6.1650222115950166e-06, "loss": 1.0798, "step": 44240 }, { "epoch": 5.03322, "grad_norm": 21.603499796531246, "learning_rate": 6.163447345147427e-06, "loss": 1.0795, "step": 44250 }, { "epoch": 5.03332, "grad_norm": 23.342946080078, "learning_rate": 6.161872356659614e-06, "loss": 1.0582, "step": 44260 }, { "epoch": 5.03342, "grad_norm": 24.474695591819682, "learning_rate": 6.1602972462967805e-06, "loss": 1.0638, "step": 44270 }, { "epoch": 5.03352, "grad_norm": 28.54025424453333, "learning_rate": 6.158722014224151e-06, "loss": 1.12, "step": 44280 }, { "epoch": 5.03362, "grad_norm": 20.053891660953937, "learning_rate": 6.157146660606962e-06, "loss": 1.0538, "step": 44290 }, { "epoch": 5.03372, "grad_norm": 23.647573117452968, "learning_rate": 6.1555711856104575e-06, "loss": 1.1078, "step": 44300 }, { "epoch": 5.03382, "grad_norm": 22.312626904563253, "learning_rate": 6.153995589399899e-06, "loss": 1.0327, "step": 44310 }, { "epoch": 5.03392, "grad_norm": 24.96911061354406, "learning_rate": 6.152419872140559e-06, "loss": 1.158, "step": 44320 }, { "epoch": 5.03402, "grad_norm": 23.88484462641076, "learning_rate": 6.150844033997721e-06, "loss": 1.034, "step": 44330 }, { "epoch": 5.03412, "grad_norm": 26.75223226067076, "learning_rate": 6.149268075136688e-06, "loss": 1.0897, "step": 44340 }, { "epoch": 5.03422, "grad_norm": 26.356701707757313, "learning_rate": 6.147691995722767e-06, "loss": 1.1377, "step": 44350 }, { "epoch": 5.03432, "grad_norm": 22.717330183907837, "learning_rate": 6.1461157959212814e-06, "loss": 1.0435, "step": 44360 }, { "epoch": 5.03442, "grad_norm": 27.043934102751752, "learning_rate": 6.144539475897568e-06, "loss": 1.153, "step": 44370 }, { "epoch": 5.03452, "grad_norm": 20.44604206275572, "learning_rate": 6.142963035816977e-06, "loss": 1.0518, "step": 44380 }, { "epoch": 5.03462, "grad_norm": 22.366688243717082, "learning_rate": 6.141386475844868e-06, "loss": 1.0704, "step": 44390 }, { "epoch": 5.03472, "grad_norm": 23.956840093561624, "learning_rate": 6.139809796146614e-06, "loss": 1.0641, "step": 44400 }, { "epoch": 5.03482, "grad_norm": 20.90464334698949, "learning_rate": 6.138232996887603e-06, "loss": 1.0723, "step": 44410 }, { "epoch": 5.03492, "grad_norm": 23.99861196498806, "learning_rate": 6.136656078233233e-06, "loss": 1.1091, "step": 44420 }, { "epoch": 5.03502, "grad_norm": 22.379415683653658, "learning_rate": 6.135079040348916e-06, "loss": 1.042, "step": 44430 }, { "epoch": 5.03512, "grad_norm": 22.428453259456372, "learning_rate": 6.133501883400076e-06, "loss": 1.1117, "step": 44440 }, { "epoch": 5.03522, "grad_norm": 21.901733653735914, "learning_rate": 6.131924607552149e-06, "loss": 1.0911, "step": 44450 }, { "epoch": 5.03532, "grad_norm": 23.92548987297259, "learning_rate": 6.130347212970583e-06, "loss": 1.0248, "step": 44460 }, { "epoch": 5.03542, "grad_norm": 21.56686065812154, "learning_rate": 6.128769699820843e-06, "loss": 1.0479, "step": 44470 }, { "epoch": 5.03552, "grad_norm": 23.679725721424344, "learning_rate": 6.127192068268397e-06, "loss": 1.0951, "step": 44480 }, { "epoch": 5.03562, "grad_norm": 23.21685312995051, "learning_rate": 6.125614318478738e-06, "loss": 1.0444, "step": 44490 }, { "epoch": 5.03572, "grad_norm": 23.31064732624331, "learning_rate": 6.1240364506173574e-06, "loss": 1.0798, "step": 44500 }, { "epoch": 5.03582, "grad_norm": 24.947815869111185, "learning_rate": 6.122458464849772e-06, "loss": 1.1129, "step": 44510 }, { "epoch": 5.03592, "grad_norm": 21.116485198542158, "learning_rate": 6.120880361341504e-06, "loss": 1.0629, "step": 44520 }, { "epoch": 5.03602, "grad_norm": 23.553530853487192, "learning_rate": 6.1193021402580864e-06, "loss": 1.0883, "step": 44530 }, { "epoch": 5.03612, "grad_norm": 25.98796755390105, "learning_rate": 6.117723801765071e-06, "loss": 1.074, "step": 44540 }, { "epoch": 5.03622, "grad_norm": 23.90447778545605, "learning_rate": 6.116145346028015e-06, "loss": 1.1139, "step": 44550 }, { "epoch": 5.03632, "grad_norm": 22.534808543587506, "learning_rate": 6.114566773212491e-06, "loss": 1.1447, "step": 44560 }, { "epoch": 5.03642, "grad_norm": 24.94829942312836, "learning_rate": 6.1129880834840885e-06, "loss": 1.0715, "step": 44570 }, { "epoch": 5.03652, "grad_norm": 22.422413017424258, "learning_rate": 6.1114092770084e-06, "loss": 1.0616, "step": 44580 }, { "epoch": 5.03662, "grad_norm": 23.22303455879833, "learning_rate": 6.1098303539510366e-06, "loss": 1.1494, "step": 44590 }, { "epoch": 5.03672, "grad_norm": 23.462062062017836, "learning_rate": 6.1082513144776205e-06, "loss": 1.1234, "step": 44600 }, { "epoch": 5.03682, "grad_norm": 24.764027584960516, "learning_rate": 6.1066721587537834e-06, "loss": 1.0647, "step": 44610 }, { "epoch": 5.03692, "grad_norm": 23.22460448698847, "learning_rate": 6.105092886945175e-06, "loss": 1.1042, "step": 44620 }, { "epoch": 5.03702, "grad_norm": 22.368292743680872, "learning_rate": 6.10351349921745e-06, "loss": 1.0577, "step": 44630 }, { "epoch": 5.03712, "grad_norm": 25.311609911015896, "learning_rate": 6.1019339957362814e-06, "loss": 1.0872, "step": 44640 }, { "epoch": 5.03722, "grad_norm": 23.13000066785688, "learning_rate": 6.1003543766673524e-06, "loss": 1.1161, "step": 44650 }, { "epoch": 5.03732, "grad_norm": 23.66270545019209, "learning_rate": 6.098774642176354e-06, "loss": 1.1277, "step": 44660 }, { "epoch": 5.03742, "grad_norm": 22.71496105492459, "learning_rate": 6.097194792428996e-06, "loss": 1.0728, "step": 44670 }, { "epoch": 5.03752, "grad_norm": 21.806102973997408, "learning_rate": 6.0956148275909955e-06, "loss": 1.0881, "step": 44680 }, { "epoch": 5.03762, "grad_norm": 25.720920521246928, "learning_rate": 6.094034747828085e-06, "loss": 1.0264, "step": 44690 }, { "epoch": 5.03772, "grad_norm": 24.64827054141134, "learning_rate": 6.092454553306009e-06, "loss": 1.1366, "step": 44700 }, { "epoch": 5.03782, "grad_norm": 28.940019536446627, "learning_rate": 6.090874244190518e-06, "loss": 1.1074, "step": 44710 }, { "epoch": 5.03792, "grad_norm": 27.136392596345058, "learning_rate": 6.089293820647384e-06, "loss": 1.1083, "step": 44720 }, { "epoch": 5.03802, "grad_norm": 24.210867523687803, "learning_rate": 6.087713282842381e-06, "loss": 1.0795, "step": 44730 }, { "epoch": 5.03812, "grad_norm": 22.507807918094763, "learning_rate": 6.086132630941305e-06, "loss": 1.0705, "step": 44740 }, { "epoch": 5.03822, "grad_norm": 24.313296216350935, "learning_rate": 6.084551865109956e-06, "loss": 1.1442, "step": 44750 }, { "epoch": 5.03832, "grad_norm": 19.51339074873464, "learning_rate": 6.082970985514151e-06, "loss": 1.0442, "step": 44760 }, { "epoch": 5.03842, "grad_norm": 22.496473599618078, "learning_rate": 6.081389992319714e-06, "loss": 1.0235, "step": 44770 }, { "epoch": 5.03852, "grad_norm": 23.2147700502269, "learning_rate": 6.0798088856924864e-06, "loss": 1.1297, "step": 44780 }, { "epoch": 5.03862, "grad_norm": 22.300253091807633, "learning_rate": 6.078227665798319e-06, "loss": 1.0947, "step": 44790 }, { "epoch": 5.03872, "grad_norm": 23.90851184319643, "learning_rate": 6.076646332803073e-06, "loss": 1.1084, "step": 44800 }, { "epoch": 5.03882, "grad_norm": 21.416546759061763, "learning_rate": 6.075064886872624e-06, "loss": 1.0705, "step": 44810 }, { "epoch": 5.03892, "grad_norm": 21.421984673978095, "learning_rate": 6.0734833281728574e-06, "loss": 1.0635, "step": 44820 }, { "epoch": 5.03902, "grad_norm": 25.618944539419015, "learning_rate": 6.071901656869672e-06, "loss": 1.0825, "step": 44830 }, { "epoch": 5.03912, "grad_norm": 20.393070304818583, "learning_rate": 6.070319873128978e-06, "loss": 1.0263, "step": 44840 }, { "epoch": 5.03922, "grad_norm": 25.72280597910421, "learning_rate": 6.068737977116697e-06, "loss": 1.1144, "step": 44850 }, { "epoch": 5.03932, "grad_norm": 21.319298293542396, "learning_rate": 6.067155968998762e-06, "loss": 1.0095, "step": 44860 }, { "epoch": 5.03942, "grad_norm": 19.868065780581503, "learning_rate": 6.065573848941119e-06, "loss": 1.0454, "step": 44870 }, { "epoch": 5.0395199999999996, "grad_norm": 21.876041523692482, "learning_rate": 6.063991617109723e-06, "loss": 1.1567, "step": 44880 }, { "epoch": 5.03962, "grad_norm": 21.48917747237734, "learning_rate": 6.062409273670547e-06, "loss": 1.0265, "step": 44890 }, { "epoch": 5.03972, "grad_norm": 24.0967626584939, "learning_rate": 6.060826818789567e-06, "loss": 1.1094, "step": 44900 }, { "epoch": 5.03982, "grad_norm": 23.892222184401767, "learning_rate": 6.059244252632779e-06, "loss": 1.1115, "step": 44910 }, { "epoch": 5.03992, "grad_norm": 22.807906287147212, "learning_rate": 6.057661575366184e-06, "loss": 1.1018, "step": 44920 }, { "epoch": 5.04002, "grad_norm": 22.57773410006828, "learning_rate": 6.056078787155798e-06, "loss": 1.1182, "step": 44930 }, { "epoch": 5.04012, "grad_norm": 23.00699019542958, "learning_rate": 6.054495888167649e-06, "loss": 1.0834, "step": 44940 }, { "epoch": 5.04022, "grad_norm": 23.066306936246193, "learning_rate": 6.052912878567775e-06, "loss": 1.1158, "step": 44950 }, { "epoch": 5.04032, "grad_norm": 24.959318978348236, "learning_rate": 6.051329758522228e-06, "loss": 1.0963, "step": 44960 }, { "epoch": 5.04042, "grad_norm": 24.0849211731409, "learning_rate": 6.0497465281970665e-06, "loss": 1.088, "step": 44970 }, { "epoch": 5.04052, "grad_norm": 18.795714548094818, "learning_rate": 6.048163187758369e-06, "loss": 1.0355, "step": 44980 }, { "epoch": 5.04062, "grad_norm": 23.62020237835883, "learning_rate": 6.046579737372215e-06, "loss": 1.0403, "step": 44990 }, { "epoch": 5.04072, "grad_norm": 22.26392975451705, "learning_rate": 6.044996177204707e-06, "loss": 1.0391, "step": 45000 } ], "logging_steps": 10, "max_steps": 100000, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "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": 7194899264045056.0, "train_batch_size": 2, "trial_name": null, "trial_params": null }