| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.000575926281436, | |
| "eval_steps": 326, | |
| "global_step": 1303, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0007679017085813016, | |
| "grad_norm": 1.3421963453292847, | |
| "learning_rate": 2e-05, | |
| "loss": 1.819, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0007679017085813016, | |
| "eval_loss": 1.6877703666687012, | |
| "eval_runtime": 19.7388, | |
| "eval_samples_per_second": 27.813, | |
| "eval_steps_per_second": 13.932, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0015358034171626032, | |
| "grad_norm": 1.0465058088302612, | |
| "learning_rate": 4e-05, | |
| "loss": 1.2514, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.002303705125743905, | |
| "grad_norm": 1.2153908014297485, | |
| "learning_rate": 6e-05, | |
| "loss": 1.6968, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0030716068343252063, | |
| "grad_norm": 1.4489881992340088, | |
| "learning_rate": 8e-05, | |
| "loss": 1.6527, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.003839508542906508, | |
| "grad_norm": 1.383280873298645, | |
| "learning_rate": 0.0001, | |
| "loss": 1.7778, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00460741025148781, | |
| "grad_norm": 1.1447311639785767, | |
| "learning_rate": 0.00012, | |
| "loss": 1.5621, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.005375311960069111, | |
| "grad_norm": 1.3640100955963135, | |
| "learning_rate": 0.00014, | |
| "loss": 1.5738, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.006143213668650413, | |
| "grad_norm": 1.419996738433838, | |
| "learning_rate": 0.00016, | |
| "loss": 1.3716, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.006911115377231714, | |
| "grad_norm": 1.2156635522842407, | |
| "learning_rate": 0.00018, | |
| "loss": 1.129, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.007679017085813016, | |
| "grad_norm": 1.6312865018844604, | |
| "learning_rate": 0.0002, | |
| "loss": 1.4983, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.008446918794394317, | |
| "grad_norm": 1.8767600059509277, | |
| "learning_rate": 0.00019999970482981582, | |
| "loss": 1.1722, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.00921482050297562, | |
| "grad_norm": 2.1750783920288086, | |
| "learning_rate": 0.0001999988193210057, | |
| "loss": 1.689, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.00998272221155692, | |
| "grad_norm": 2.1488802433013916, | |
| "learning_rate": 0.00019999734347879723, | |
| "loss": 1.3221, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.010750623920138222, | |
| "grad_norm": 2.133239269256592, | |
| "learning_rate": 0.0001999952773119029, | |
| "loss": 1.4366, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.011518525628719524, | |
| "grad_norm": 1.99974524974823, | |
| "learning_rate": 0.00019999262083252007, | |
| "loss": 1.2169, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.012286427337300825, | |
| "grad_norm": 1.5727359056472778, | |
| "learning_rate": 0.00019998937405633105, | |
| "loss": 1.1514, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.013054329045882128, | |
| "grad_norm": 2.484224557876587, | |
| "learning_rate": 0.00019998553700250284, | |
| "loss": 1.6751, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.013822230754463429, | |
| "grad_norm": 2.1434097290039062, | |
| "learning_rate": 0.00019998110969368717, | |
| "loss": 1.7869, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.01459013246304473, | |
| "grad_norm": 1.8864972591400146, | |
| "learning_rate": 0.00019997609215602019, | |
| "loss": 1.3473, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.015358034171626032, | |
| "grad_norm": 1.396660327911377, | |
| "learning_rate": 0.00019997048441912246, | |
| "loss": 1.0329, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.016125935880207335, | |
| "grad_norm": 1.8384391069412231, | |
| "learning_rate": 0.0001999642865160987, | |
| "loss": 1.5533, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.016893837588788634, | |
| "grad_norm": 1.6881628036499023, | |
| "learning_rate": 0.0001999574984835377, | |
| "loss": 1.4798, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.017661739297369937, | |
| "grad_norm": 1.6988723278045654, | |
| "learning_rate": 0.00019995012036151186, | |
| "loss": 1.3521, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.01842964100595124, | |
| "grad_norm": 1.5781642198562622, | |
| "learning_rate": 0.00019994215219357728, | |
| "loss": 1.1237, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01919754271453254, | |
| "grad_norm": 1.8065123558044434, | |
| "learning_rate": 0.00019993359402677323, | |
| "loss": 1.4728, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01996544442311384, | |
| "grad_norm": 1.6945093870162964, | |
| "learning_rate": 0.00019992444591162206, | |
| "loss": 1.2896, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.020733346131695144, | |
| "grad_norm": 1.4968225955963135, | |
| "learning_rate": 0.00019991470790212877, | |
| "loss": 1.3827, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.021501247840276443, | |
| "grad_norm": 1.4860451221466064, | |
| "learning_rate": 0.00019990438005578075, | |
| "loss": 1.4381, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.022269149548857746, | |
| "grad_norm": 1.543802261352539, | |
| "learning_rate": 0.00019989346243354746, | |
| "loss": 1.3484, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.02303705125743905, | |
| "grad_norm": 1.5513710975646973, | |
| "learning_rate": 0.00019988195509988005, | |
| "loss": 1.3842, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.023804952966020348, | |
| "grad_norm": 1.6369807720184326, | |
| "learning_rate": 0.00019986985812271092, | |
| "loss": 1.242, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.02457285467460165, | |
| "grad_norm": 1.7637338638305664, | |
| "learning_rate": 0.00019985717157345345, | |
| "loss": 1.1865, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.025340756383182953, | |
| "grad_norm": 1.4395427703857422, | |
| "learning_rate": 0.00019984389552700144, | |
| "loss": 1.2769, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.026108658091764256, | |
| "grad_norm": 1.8801194429397583, | |
| "learning_rate": 0.0001998300300617287, | |
| "loss": 1.7346, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.026876559800345555, | |
| "grad_norm": 1.7689722776412964, | |
| "learning_rate": 0.00019981557525948875, | |
| "loss": 1.4345, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.027644461508926858, | |
| "grad_norm": 1.1703026294708252, | |
| "learning_rate": 0.00019980053120561411, | |
| "loss": 0.8503, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.02841236321750816, | |
| "grad_norm": 1.4274190664291382, | |
| "learning_rate": 0.00019978489798891584, | |
| "loss": 1.2453, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02918026492608946, | |
| "grad_norm": 1.5258653163909912, | |
| "learning_rate": 0.00019976867570168318, | |
| "loss": 1.2976, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.029948166634670762, | |
| "grad_norm": 1.3230305910110474, | |
| "learning_rate": 0.00019975186443968286, | |
| "loss": 1.419, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.030716068343252065, | |
| "grad_norm": 1.4891058206558228, | |
| "learning_rate": 0.0001997344643021585, | |
| "loss": 1.2045, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.031483970051833364, | |
| "grad_norm": 1.3065136671066284, | |
| "learning_rate": 0.00019971647539183013, | |
| "loss": 1.2106, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.03225187176041467, | |
| "grad_norm": 1.5086220502853394, | |
| "learning_rate": 0.00019969789781489362, | |
| "loss": 1.4317, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.03301977346899597, | |
| "grad_norm": 1.7055323123931885, | |
| "learning_rate": 0.00019967873168101984, | |
| "loss": 1.5407, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.03378767517757727, | |
| "grad_norm": 1.1143574714660645, | |
| "learning_rate": 0.00019965897710335422, | |
| "loss": 0.9124, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.034555576886158575, | |
| "grad_norm": 1.7933096885681152, | |
| "learning_rate": 0.00019963863419851605, | |
| "loss": 1.5175, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.035323478594739874, | |
| "grad_norm": 1.963930368423462, | |
| "learning_rate": 0.00019961770308659767, | |
| "loss": 1.4704, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.03609138030332117, | |
| "grad_norm": 1.2341890335083008, | |
| "learning_rate": 0.00019959618389116387, | |
| "loss": 0.8867, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.03685928201190248, | |
| "grad_norm": 1.3576570749282837, | |
| "learning_rate": 0.0001995740767392512, | |
| "loss": 1.5176, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.03762718372048378, | |
| "grad_norm": 1.2808408737182617, | |
| "learning_rate": 0.0001995513817613671, | |
| "loss": 1.076, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.03839508542906508, | |
| "grad_norm": 1.597029209136963, | |
| "learning_rate": 0.00019952809909148914, | |
| "loss": 1.7766, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.039162987137646384, | |
| "grad_norm": 1.302276849746704, | |
| "learning_rate": 0.0001995042288670643, | |
| "loss": 1.1456, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.03993088884622768, | |
| "grad_norm": 1.564570665359497, | |
| "learning_rate": 0.00019947977122900822, | |
| "loss": 1.428, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.04069879055480898, | |
| "grad_norm": 1.4207552671432495, | |
| "learning_rate": 0.0001994547263217042, | |
| "loss": 1.4306, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.04146669226339029, | |
| "grad_norm": 1.530429720878601, | |
| "learning_rate": 0.00019942909429300238, | |
| "loss": 1.1629, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.04223459397197159, | |
| "grad_norm": 1.1206613779067993, | |
| "learning_rate": 0.00019940287529421902, | |
| "loss": 1.2482, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.04300249568055289, | |
| "grad_norm": 1.2022866010665894, | |
| "learning_rate": 0.00019937606948013548, | |
| "loss": 1.208, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.04377039738913419, | |
| "grad_norm": 1.7056037187576294, | |
| "learning_rate": 0.00019934867700899722, | |
| "loss": 1.2333, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.04453829909771549, | |
| "grad_norm": 1.274591088294983, | |
| "learning_rate": 0.00019932069804251312, | |
| "loss": 0.8844, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.04530620080629679, | |
| "grad_norm": 1.6336207389831543, | |
| "learning_rate": 0.0001992921327458543, | |
| "loss": 1.37, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0460741025148781, | |
| "grad_norm": 1.6932519674301147, | |
| "learning_rate": 0.00019926298128765323, | |
| "loss": 1.4952, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.046842004223459396, | |
| "grad_norm": 1.6268481016159058, | |
| "learning_rate": 0.00019923324384000276, | |
| "loss": 1.4606, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.047609905932040696, | |
| "grad_norm": 1.5920450687408447, | |
| "learning_rate": 0.00019920292057845499, | |
| "loss": 1.4074, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.048377807640622, | |
| "grad_norm": 1.5211634635925293, | |
| "learning_rate": 0.00019917201168202043, | |
| "loss": 1.4462, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0491457093492033, | |
| "grad_norm": 1.4284402132034302, | |
| "learning_rate": 0.00019914051733316678, | |
| "loss": 0.8046, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0499136110577846, | |
| "grad_norm": 1.6174309253692627, | |
| "learning_rate": 0.00019910843771781783, | |
| "loss": 1.5238, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.050681512766365906, | |
| "grad_norm": 1.4781291484832764, | |
| "learning_rate": 0.00019907577302535255, | |
| "loss": 1.3418, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.051449414474947205, | |
| "grad_norm": 1.597865343093872, | |
| "learning_rate": 0.00019904252344860382, | |
| "loss": 1.1921, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.05221731618352851, | |
| "grad_norm": 1.5983515977859497, | |
| "learning_rate": 0.00019900868918385726, | |
| "loss": 1.5806, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.05298521789210981, | |
| "grad_norm": 2.1672117710113525, | |
| "learning_rate": 0.00019897427043085022, | |
| "loss": 1.5998, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.05375311960069111, | |
| "grad_norm": 1.3775156736373901, | |
| "learning_rate": 0.0001989392673927705, | |
| "loss": 1.2913, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.054521021309272416, | |
| "grad_norm": 1.5657135248184204, | |
| "learning_rate": 0.00019890368027625517, | |
| "loss": 1.1062, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.055288923017853715, | |
| "grad_norm": 1.1934847831726074, | |
| "learning_rate": 0.00019886750929138934, | |
| "loss": 1.126, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.056056824726435014, | |
| "grad_norm": 1.964847207069397, | |
| "learning_rate": 0.0001988307546517049, | |
| "loss": 1.5169, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.05682472643501632, | |
| "grad_norm": 1.7718520164489746, | |
| "learning_rate": 0.00019879341657417935, | |
| "loss": 1.1245, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.05759262814359762, | |
| "grad_norm": 1.3038172721862793, | |
| "learning_rate": 0.00019875549527923449, | |
| "loss": 1.0887, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05836052985217892, | |
| "grad_norm": 1.4829697608947754, | |
| "learning_rate": 0.00019871699099073493, | |
| "loss": 1.2666, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.059128431560760225, | |
| "grad_norm": 1.5347778797149658, | |
| "learning_rate": 0.0001986779039359871, | |
| "loss": 1.2203, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.059896333269341524, | |
| "grad_norm": 1.6270524263381958, | |
| "learning_rate": 0.00019863823434573762, | |
| "loss": 1.4289, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.060664234977922823, | |
| "grad_norm": 1.3492023944854736, | |
| "learning_rate": 0.00019859798245417217, | |
| "loss": 1.2448, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.06143213668650413, | |
| "grad_norm": 1.7291414737701416, | |
| "learning_rate": 0.0001985571484989138, | |
| "loss": 1.2516, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06220003839508543, | |
| "grad_norm": 1.7101045846939087, | |
| "learning_rate": 0.00019851573272102195, | |
| "loss": 1.2677, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.06296794010366673, | |
| "grad_norm": 1.4374966621398926, | |
| "learning_rate": 0.0001984737353649906, | |
| "loss": 1.1178, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.06373584181224803, | |
| "grad_norm": 1.5133461952209473, | |
| "learning_rate": 0.00019843115667874707, | |
| "loss": 1.46, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.06450374352082934, | |
| "grad_norm": 1.594156265258789, | |
| "learning_rate": 0.00019838799691365065, | |
| "loss": 1.2635, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.06527164522941063, | |
| "grad_norm": 1.4688767194747925, | |
| "learning_rate": 0.00019834425632449075, | |
| "loss": 1.1498, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.06603954693799194, | |
| "grad_norm": 1.7303012609481812, | |
| "learning_rate": 0.00019829993516948577, | |
| "loss": 1.775, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.06680744864657324, | |
| "grad_norm": 1.2842633724212646, | |
| "learning_rate": 0.00019825503371028136, | |
| "loss": 1.1091, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.06757535035515454, | |
| "grad_norm": 1.4563853740692139, | |
| "learning_rate": 0.000198209552211949, | |
| "loss": 1.4345, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.06834325206373584, | |
| "grad_norm": 1.344085693359375, | |
| "learning_rate": 0.00019816349094298427, | |
| "loss": 1.4728, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.06911115377231715, | |
| "grad_norm": 1.4741275310516357, | |
| "learning_rate": 0.0001981168501753055, | |
| "loss": 1.4337, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.06987905548089844, | |
| "grad_norm": 1.4946739673614502, | |
| "learning_rate": 0.0001980696301842519, | |
| "loss": 1.4202, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.07064695718947975, | |
| "grad_norm": 1.2858799695968628, | |
| "learning_rate": 0.00019802183124858222, | |
| "loss": 0.9181, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.07141485889806105, | |
| "grad_norm": 1.789942979812622, | |
| "learning_rate": 0.00019797345365047284, | |
| "loss": 1.4772, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.07218276060664235, | |
| "grad_norm": 1.4283826351165771, | |
| "learning_rate": 0.0001979244976755162, | |
| "loss": 1.3747, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.07295066231522365, | |
| "grad_norm": 1.5858755111694336, | |
| "learning_rate": 0.00019787496361271925, | |
| "loss": 1.5726, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.07371856402380496, | |
| "grad_norm": 1.5366917848587036, | |
| "learning_rate": 0.00019782485175450155, | |
| "loss": 1.4846, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.07448646573238625, | |
| "grad_norm": 1.363489031791687, | |
| "learning_rate": 0.0001977741623966936, | |
| "loss": 1.211, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.07525436744096756, | |
| "grad_norm": 1.6271450519561768, | |
| "learning_rate": 0.00019772289583853514, | |
| "loss": 1.51, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.07602226914954886, | |
| "grad_norm": 1.333284854888916, | |
| "learning_rate": 0.00019767105238267338, | |
| "loss": 1.2552, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.07679017085813016, | |
| "grad_norm": 1.5619887113571167, | |
| "learning_rate": 0.00019761863233516117, | |
| "loss": 1.4492, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07755807256671146, | |
| "grad_norm": 1.3601068258285522, | |
| "learning_rate": 0.0001975656360054552, | |
| "loss": 1.2376, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.07832597427529277, | |
| "grad_norm": 1.6772176027297974, | |
| "learning_rate": 0.0001975120637064142, | |
| "loss": 1.4527, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.07909387598387406, | |
| "grad_norm": 1.5059335231781006, | |
| "learning_rate": 0.00019745791575429705, | |
| "loss": 1.3531, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.07986177769245537, | |
| "grad_norm": 1.6139791011810303, | |
| "learning_rate": 0.00019740319246876106, | |
| "loss": 1.5811, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.08062967940103667, | |
| "grad_norm": 1.6082731485366821, | |
| "learning_rate": 0.00019734789417285976, | |
| "loss": 1.3414, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.08139758110961796, | |
| "grad_norm": 1.5934547185897827, | |
| "learning_rate": 0.0001972920211930414, | |
| "loss": 1.3688, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.08216548281819927, | |
| "grad_norm": 1.2157509326934814, | |
| "learning_rate": 0.0001972355738591467, | |
| "loss": 0.8838, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.08293338452678058, | |
| "grad_norm": 1.6874890327453613, | |
| "learning_rate": 0.00019717855250440705, | |
| "loss": 1.5832, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.08370128623536187, | |
| "grad_norm": 1.4265177249908447, | |
| "learning_rate": 0.00019712095746544255, | |
| "loss": 1.3194, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.08446918794394317, | |
| "grad_norm": 1.457571268081665, | |
| "learning_rate": 0.00019706278908225992, | |
| "loss": 1.3425, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.08523708965252448, | |
| "grad_norm": 1.6059311628341675, | |
| "learning_rate": 0.00019700404769825068, | |
| "loss": 1.7976, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.08600499136110577, | |
| "grad_norm": 1.3891575336456299, | |
| "learning_rate": 0.00019694473366018887, | |
| "loss": 1.3154, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.08677289306968708, | |
| "grad_norm": 1.5132336616516113, | |
| "learning_rate": 0.00019688484731822923, | |
| "loss": 1.4888, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.08754079477826839, | |
| "grad_norm": 1.5129386186599731, | |
| "learning_rate": 0.00019682438902590498, | |
| "loss": 1.279, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.08830869648684968, | |
| "grad_norm": 1.4669959545135498, | |
| "learning_rate": 0.0001967633591401259, | |
| "loss": 1.1884, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.08907659819543098, | |
| "grad_norm": 1.3678569793701172, | |
| "learning_rate": 0.000196701758021176, | |
| "loss": 1.0954, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.08984449990401229, | |
| "grad_norm": 1.3512712717056274, | |
| "learning_rate": 0.00019663958603271148, | |
| "loss": 1.1583, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.09061240161259358, | |
| "grad_norm": 1.2493406534194946, | |
| "learning_rate": 0.0001965768435417588, | |
| "loss": 1.1847, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.09138030332117489, | |
| "grad_norm": 1.6582459211349487, | |
| "learning_rate": 0.00019651353091871215, | |
| "loss": 1.4876, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.0921482050297562, | |
| "grad_norm": 1.5519031286239624, | |
| "learning_rate": 0.00019644964853733152, | |
| "loss": 1.5264, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.09291610673833749, | |
| "grad_norm": 1.9167393445968628, | |
| "learning_rate": 0.0001963851967747404, | |
| "loss": 1.6469, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.09368400844691879, | |
| "grad_norm": 1.7159210443496704, | |
| "learning_rate": 0.00019632017601142355, | |
| "loss": 1.4076, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0944519101555001, | |
| "grad_norm": 1.5562142133712769, | |
| "learning_rate": 0.00019625458663122478, | |
| "loss": 1.0584, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.09521981186408139, | |
| "grad_norm": 1.5296403169631958, | |
| "learning_rate": 0.00019618842902134465, | |
| "loss": 1.3064, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0959877135726627, | |
| "grad_norm": 1.6723419427871704, | |
| "learning_rate": 0.00019612170357233836, | |
| "loss": 1.4708, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.096755615281244, | |
| "grad_norm": 1.6125822067260742, | |
| "learning_rate": 0.00019605441067811302, | |
| "loss": 1.7615, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0975235169898253, | |
| "grad_norm": 1.4108672142028809, | |
| "learning_rate": 0.00019598655073592585, | |
| "loss": 1.447, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.0982914186984066, | |
| "grad_norm": 1.373512864112854, | |
| "learning_rate": 0.0001959181241463814, | |
| "loss": 1.1841, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.09905932040698791, | |
| "grad_norm": 1.5692287683486938, | |
| "learning_rate": 0.00019584913131342953, | |
| "loss": 1.4497, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.0998272221155692, | |
| "grad_norm": 1.8311026096343994, | |
| "learning_rate": 0.0001957795726443628, | |
| "loss": 1.425, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.1005951238241505, | |
| "grad_norm": 1.5279914140701294, | |
| "learning_rate": 0.000195709448549814, | |
| "loss": 1.2395, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.10136302553273181, | |
| "grad_norm": 1.5802501440048218, | |
| "learning_rate": 0.00019563875944375407, | |
| "loss": 1.3774, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.10213092724131312, | |
| "grad_norm": 1.3959014415740967, | |
| "learning_rate": 0.0001955675057434893, | |
| "loss": 1.3567, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.10289882894989441, | |
| "grad_norm": 1.426458716392517, | |
| "learning_rate": 0.00019549568786965903, | |
| "loss": 1.26, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.10366673065847572, | |
| "grad_norm": 1.413362979888916, | |
| "learning_rate": 0.00019542330624623322, | |
| "loss": 1.3663, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.10443463236705702, | |
| "grad_norm": 1.4533791542053223, | |
| "learning_rate": 0.00019535036130050975, | |
| "loss": 1.3983, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.10520253407563832, | |
| "grad_norm": 1.3483699560165405, | |
| "learning_rate": 0.00019527685346311212, | |
| "loss": 1.0326, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.10597043578421962, | |
| "grad_norm": 1.591539740562439, | |
| "learning_rate": 0.0001952027831679867, | |
| "loss": 1.2265, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.10673833749280093, | |
| "grad_norm": 1.777915120124817, | |
| "learning_rate": 0.00019512815085240046, | |
| "loss": 1.6348, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.10750623920138222, | |
| "grad_norm": 1.4367990493774414, | |
| "learning_rate": 0.000195052956956938, | |
| "loss": 1.3012, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.10827414090996353, | |
| "grad_norm": 1.5832383632659912, | |
| "learning_rate": 0.00019497720192549926, | |
| "loss": 1.1717, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.10904204261854483, | |
| "grad_norm": 1.6121189594268799, | |
| "learning_rate": 0.00019490088620529678, | |
| "loss": 1.5101, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.10980994432712612, | |
| "grad_norm": 1.6623026132583618, | |
| "learning_rate": 0.00019482401024685308, | |
| "loss": 1.3055, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.11057784603570743, | |
| "grad_norm": 1.2664202451705933, | |
| "learning_rate": 0.0001947465745039979, | |
| "loss": 1.0602, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.11134574774428874, | |
| "grad_norm": 1.318278431892395, | |
| "learning_rate": 0.0001946685794338658, | |
| "loss": 1.285, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.11211364945287003, | |
| "grad_norm": 1.7018437385559082, | |
| "learning_rate": 0.00019459002549689308, | |
| "loss": 1.3439, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.11288155116145134, | |
| "grad_norm": 1.629821538925171, | |
| "learning_rate": 0.0001945109131568154, | |
| "loss": 1.2974, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.11364945287003264, | |
| "grad_norm": 1.416452407836914, | |
| "learning_rate": 0.00019443124288066475, | |
| "loss": 1.1124, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.11441735457861393, | |
| "grad_norm": 1.6793434619903564, | |
| "learning_rate": 0.00019435101513876703, | |
| "loss": 1.3027, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.11518525628719524, | |
| "grad_norm": 1.465809941291809, | |
| "learning_rate": 0.00019427023040473896, | |
| "loss": 0.9204, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11595315799577655, | |
| "grad_norm": 1.7174264192581177, | |
| "learning_rate": 0.0001941888891554854, | |
| "loss": 1.2307, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.11672105970435784, | |
| "grad_norm": 1.3627086877822876, | |
| "learning_rate": 0.00019410699187119663, | |
| "loss": 1.0926, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.11748896141293914, | |
| "grad_norm": 2.059274911880493, | |
| "learning_rate": 0.00019402453903534533, | |
| "loss": 1.5285, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.11825686312152045, | |
| "grad_norm": 1.3840800523757935, | |
| "learning_rate": 0.0001939415311346839, | |
| "loss": 1.2504, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.11902476483010174, | |
| "grad_norm": 1.6135592460632324, | |
| "learning_rate": 0.0001938579686592415, | |
| "loss": 1.3955, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.11979266653868305, | |
| "grad_norm": 1.4203635454177856, | |
| "learning_rate": 0.00019377385210232113, | |
| "loss": 1.2271, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.12056056824726435, | |
| "grad_norm": 1.4303590059280396, | |
| "learning_rate": 0.0001936891819604968, | |
| "loss": 1.5135, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.12132846995584565, | |
| "grad_norm": 1.3626837730407715, | |
| "learning_rate": 0.00019360395873361055, | |
| "loss": 1.1232, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.12209637166442695, | |
| "grad_norm": 1.2136698961257935, | |
| "learning_rate": 0.00019351818292476946, | |
| "loss": 1.2211, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.12286427337300826, | |
| "grad_norm": 1.463807463645935, | |
| "learning_rate": 0.00019343185504034277, | |
| "loss": 1.2924, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.12363217508158955, | |
| "grad_norm": 1.2477118968963623, | |
| "learning_rate": 0.0001933449755899588, | |
| "loss": 1.2391, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.12440007679017086, | |
| "grad_norm": 1.2382277250289917, | |
| "learning_rate": 0.0001932575450865021, | |
| "loss": 1.0537, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.12516797849875216, | |
| "grad_norm": 1.1444644927978516, | |
| "learning_rate": 0.00019316956404611012, | |
| "loss": 1.0164, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.12593588020733346, | |
| "grad_norm": 1.7438474893569946, | |
| "learning_rate": 0.00019308103298817052, | |
| "loss": 1.3299, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.12670378191591478, | |
| "grad_norm": 1.3742320537567139, | |
| "learning_rate": 0.00019299195243531792, | |
| "loss": 1.1848, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.12747168362449607, | |
| "grad_norm": 1.456499695777893, | |
| "learning_rate": 0.00019290232291343067, | |
| "loss": 1.5438, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.12823958533307736, | |
| "grad_norm": 1.0693509578704834, | |
| "learning_rate": 0.0001928121449516281, | |
| "loss": 1.0623, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.12900748704165868, | |
| "grad_norm": 1.7221637964248657, | |
| "learning_rate": 0.00019272141908226707, | |
| "loss": 1.1941, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.12977538875023997, | |
| "grad_norm": 1.3603307008743286, | |
| "learning_rate": 0.0001926301458409391, | |
| "loss": 1.0901, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.13054329045882127, | |
| "grad_norm": 1.498059868812561, | |
| "learning_rate": 0.00019253832576646688, | |
| "loss": 1.4565, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.13131119216740259, | |
| "grad_norm": 1.5251699686050415, | |
| "learning_rate": 0.00019244595940090143, | |
| "loss": 1.1852, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.13207909387598388, | |
| "grad_norm": 1.6394065618515015, | |
| "learning_rate": 0.00019235304728951866, | |
| "loss": 1.3357, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.13284699558456517, | |
| "grad_norm": 1.3979147672653198, | |
| "learning_rate": 0.00019225958998081633, | |
| "loss": 1.2176, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.1336148972931465, | |
| "grad_norm": 1.5247663259506226, | |
| "learning_rate": 0.0001921655880265106, | |
| "loss": 1.4776, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.13438279900172778, | |
| "grad_norm": 1.7664347887039185, | |
| "learning_rate": 0.00019207104198153295, | |
| "loss": 1.3425, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.13515070071030907, | |
| "grad_norm": 1.719008445739746, | |
| "learning_rate": 0.0001919759524040269, | |
| "loss": 1.5569, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.1359186024188904, | |
| "grad_norm": 1.5756323337554932, | |
| "learning_rate": 0.0001918803198553446, | |
| "loss": 1.4662, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.1366865041274717, | |
| "grad_norm": 1.1478333473205566, | |
| "learning_rate": 0.00019178414490004356, | |
| "loss": 0.8951, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.13745440583605298, | |
| "grad_norm": 1.549656629562378, | |
| "learning_rate": 0.00019168742810588335, | |
| "loss": 1.1626, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.1382223075446343, | |
| "grad_norm": 2.0804789066314697, | |
| "learning_rate": 0.00019159017004382234, | |
| "loss": 1.1341, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1389902092532156, | |
| "grad_norm": 1.3882943391799927, | |
| "learning_rate": 0.00019149237128801404, | |
| "loss": 1.0807, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.13975811096179688, | |
| "grad_norm": 1.5854548215866089, | |
| "learning_rate": 0.000191394032415804, | |
| "loss": 1.2928, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.1405260126703782, | |
| "grad_norm": 1.5255122184753418, | |
| "learning_rate": 0.00019129515400772635, | |
| "loss": 1.2089, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.1412939143789595, | |
| "grad_norm": 1.3562302589416504, | |
| "learning_rate": 0.00019119573664750018, | |
| "loss": 1.2578, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.1420618160875408, | |
| "grad_norm": 1.4164509773254395, | |
| "learning_rate": 0.00019109578092202628, | |
| "loss": 1.1642, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.1428297177961221, | |
| "grad_norm": 1.4175788164138794, | |
| "learning_rate": 0.00019099528742138371, | |
| "loss": 1.2517, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.1435976195047034, | |
| "grad_norm": 1.5696830749511719, | |
| "learning_rate": 0.00019089425673882615, | |
| "loss": 1.4897, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.1443655212132847, | |
| "grad_norm": 1.4855685234069824, | |
| "learning_rate": 0.0001907926894707785, | |
| "loss": 1.4277, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.145133422921866, | |
| "grad_norm": 1.6295796632766724, | |
| "learning_rate": 0.00019069058621683336, | |
| "loss": 1.3268, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.1459013246304473, | |
| "grad_norm": 1.5414029359817505, | |
| "learning_rate": 0.0001905879475797474, | |
| "loss": 1.4513, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1466692263390286, | |
| "grad_norm": 1.3137727975845337, | |
| "learning_rate": 0.00019048477416543801, | |
| "loss": 1.1609, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.14743712804760992, | |
| "grad_norm": 1.5763883590698242, | |
| "learning_rate": 0.00019038106658297944, | |
| "loss": 1.2035, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.1482050297561912, | |
| "grad_norm": 1.3297104835510254, | |
| "learning_rate": 0.00019027682544459947, | |
| "loss": 1.2346, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.1489729314647725, | |
| "grad_norm": 1.568937063217163, | |
| "learning_rate": 0.00019017205136567556, | |
| "loss": 1.3332, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.14974083317335382, | |
| "grad_norm": 1.6871936321258545, | |
| "learning_rate": 0.00019006674496473144, | |
| "loss": 1.0715, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1505087348819351, | |
| "grad_norm": 1.6157925128936768, | |
| "learning_rate": 0.00018996090686343328, | |
| "loss": 1.4985, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.1512766365905164, | |
| "grad_norm": 1.384244680404663, | |
| "learning_rate": 0.0001898545376865861, | |
| "loss": 0.9407, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.15204453829909773, | |
| "grad_norm": 1.5199958086013794, | |
| "learning_rate": 0.00018974763806213013, | |
| "loss": 1.459, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.15281244000767902, | |
| "grad_norm": 1.6747719049453735, | |
| "learning_rate": 0.000189640208621137, | |
| "loss": 1.475, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.1535803417162603, | |
| "grad_norm": 1.3858054876327515, | |
| "learning_rate": 0.00018953224999780605, | |
| "loss": 1.1565, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.15434824342484163, | |
| "grad_norm": 1.5445752143859863, | |
| "learning_rate": 0.00018942376282946066, | |
| "loss": 1.3588, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.15511614513342292, | |
| "grad_norm": 1.3727281093597412, | |
| "learning_rate": 0.0001893147477565443, | |
| "loss": 1.1978, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.15588404684200421, | |
| "grad_norm": 1.5093921422958374, | |
| "learning_rate": 0.000189205205422617, | |
| "loss": 1.2175, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.15665194855058553, | |
| "grad_norm": 1.3368033170700073, | |
| "learning_rate": 0.0001890951364743514, | |
| "loss": 1.0925, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.15741985025916683, | |
| "grad_norm": 1.3794496059417725, | |
| "learning_rate": 0.00018898454156152886, | |
| "loss": 1.3221, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.15818775196774812, | |
| "grad_norm": 1.8431724309921265, | |
| "learning_rate": 0.0001888734213370359, | |
| "loss": 1.4075, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.15895565367632944, | |
| "grad_norm": 1.592666745185852, | |
| "learning_rate": 0.00018876177645685998, | |
| "loss": 1.2406, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.15972355538491073, | |
| "grad_norm": 1.7013722658157349, | |
| "learning_rate": 0.00018864960758008592, | |
| "loss": 1.4999, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.16049145709349202, | |
| "grad_norm": 1.318893313407898, | |
| "learning_rate": 0.00018853691536889188, | |
| "loss": 0.9244, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.16125935880207334, | |
| "grad_norm": 1.1515171527862549, | |
| "learning_rate": 0.0001884237004885455, | |
| "loss": 0.9195, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.16202726051065464, | |
| "grad_norm": 1.6677429676055908, | |
| "learning_rate": 0.0001883099636073999, | |
| "loss": 1.315, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.16279516221923593, | |
| "grad_norm": 1.5073673725128174, | |
| "learning_rate": 0.0001881957053968898, | |
| "loss": 1.18, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.16356306392781725, | |
| "grad_norm": 1.6688820123672485, | |
| "learning_rate": 0.00018808092653152753, | |
| "loss": 1.4562, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.16433096563639854, | |
| "grad_norm": 1.4027730226516724, | |
| "learning_rate": 0.00018796562768889913, | |
| "loss": 1.2261, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.16509886734497983, | |
| "grad_norm": 1.699299931526184, | |
| "learning_rate": 0.0001878498095496601, | |
| "loss": 1.4926, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.16586676905356115, | |
| "grad_norm": 1.1611884832382202, | |
| "learning_rate": 0.00018773347279753177, | |
| "loss": 0.9826, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.16663467076214244, | |
| "grad_norm": 1.4166936874389648, | |
| "learning_rate": 0.00018761661811929686, | |
| "loss": 1.142, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.16740257247072374, | |
| "grad_norm": 1.652862310409546, | |
| "learning_rate": 0.00018749924620479585, | |
| "loss": 1.3107, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.16817047417930506, | |
| "grad_norm": 1.301758885383606, | |
| "learning_rate": 0.0001873813577469224, | |
| "loss": 1.0521, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.16893837588788635, | |
| "grad_norm": 1.6529386043548584, | |
| "learning_rate": 0.0001872629534416197, | |
| "loss": 1.1955, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.16970627759646764, | |
| "grad_norm": 1.254197359085083, | |
| "learning_rate": 0.0001871440339878762, | |
| "loss": 0.974, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.17047417930504896, | |
| "grad_norm": 1.7299309968948364, | |
| "learning_rate": 0.0001870246000877214, | |
| "loss": 1.4206, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.17124208101363025, | |
| "grad_norm": 1.4324923753738403, | |
| "learning_rate": 0.00018690465244622183, | |
| "loss": 1.1965, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.17200998272221155, | |
| "grad_norm": 1.4682673215866089, | |
| "learning_rate": 0.00018678419177147685, | |
| "loss": 1.2639, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.17277788443079287, | |
| "grad_norm": 1.4695028066635132, | |
| "learning_rate": 0.0001866632187746145, | |
| "loss": 1.2862, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.17354578613937416, | |
| "grad_norm": 1.2502456903457642, | |
| "learning_rate": 0.00018654173416978714, | |
| "loss": 1.3128, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.17431368784795545, | |
| "grad_norm": 1.443822979927063, | |
| "learning_rate": 0.0001864197386741674, | |
| "loss": 1.3217, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.17508158955653677, | |
| "grad_norm": 1.5240435600280762, | |
| "learning_rate": 0.00018629723300794408, | |
| "loss": 1.2936, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.17584949126511806, | |
| "grad_norm": 1.5708985328674316, | |
| "learning_rate": 0.00018617421789431747, | |
| "loss": 1.3001, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.17661739297369936, | |
| "grad_norm": 1.595597743988037, | |
| "learning_rate": 0.0001860506940594955, | |
| "loss": 1.2869, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.17738529468228068, | |
| "grad_norm": 1.3937422037124634, | |
| "learning_rate": 0.00018592666223268917, | |
| "loss": 1.1357, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.17815319639086197, | |
| "grad_norm": 1.6847139596939087, | |
| "learning_rate": 0.00018580212314610846, | |
| "loss": 1.6054, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.17892109809944326, | |
| "grad_norm": 2.011245012283325, | |
| "learning_rate": 0.0001856770775349579, | |
| "loss": 1.4489, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.17968899980802458, | |
| "grad_norm": 1.3544971942901611, | |
| "learning_rate": 0.00018555152613743215, | |
| "loss": 1.3519, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.18045690151660587, | |
| "grad_norm": 1.729085922241211, | |
| "learning_rate": 0.00018542546969471183, | |
| "loss": 1.5642, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.18122480322518716, | |
| "grad_norm": 1.8843064308166504, | |
| "learning_rate": 0.00018529890895095902, | |
| "loss": 1.7392, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.18199270493376848, | |
| "grad_norm": 1.3016268014907837, | |
| "learning_rate": 0.00018517184465331288, | |
| "loss": 1.0011, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.18276060664234978, | |
| "grad_norm": 1.5241748094558716, | |
| "learning_rate": 0.00018504427755188521, | |
| "loss": 1.0784, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.18352850835093107, | |
| "grad_norm": 1.42753267288208, | |
| "learning_rate": 0.00018491620839975617, | |
| "loss": 1.1811, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.1842964100595124, | |
| "grad_norm": 1.427199125289917, | |
| "learning_rate": 0.00018478763795296962, | |
| "loss": 1.0278, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.18506431176809368, | |
| "grad_norm": 1.4727647304534912, | |
| "learning_rate": 0.0001846585669705288, | |
| "loss": 1.4351, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.18583221347667497, | |
| "grad_norm": 1.6001156568527222, | |
| "learning_rate": 0.00018452899621439182, | |
| "loss": 1.4974, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.1866001151852563, | |
| "grad_norm": 1.8062998056411743, | |
| "learning_rate": 0.00018439892644946722, | |
| "loss": 1.4885, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.18736801689383759, | |
| "grad_norm": 1.3706297874450684, | |
| "learning_rate": 0.00018426835844360929, | |
| "loss": 1.2593, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.18813591860241888, | |
| "grad_norm": 1.557486653327942, | |
| "learning_rate": 0.00018413729296761364, | |
| "loss": 1.3712, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.1889038203110002, | |
| "grad_norm": 1.640881896018982, | |
| "learning_rate": 0.00018400573079521278, | |
| "loss": 1.4834, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.1896717220195815, | |
| "grad_norm": 1.6011335849761963, | |
| "learning_rate": 0.0001838736727030712, | |
| "loss": 1.2738, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.19043962372816278, | |
| "grad_norm": 1.2537946701049805, | |
| "learning_rate": 0.00018374111947078124, | |
| "loss": 1.0421, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.1912075254367441, | |
| "grad_norm": 1.5533615350723267, | |
| "learning_rate": 0.00018360807188085807, | |
| "loss": 1.2964, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.1919754271453254, | |
| "grad_norm": 1.967961072921753, | |
| "learning_rate": 0.00018347453071873536, | |
| "loss": 1.7348, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1927433288539067, | |
| "grad_norm": 1.4344468116760254, | |
| "learning_rate": 0.00018334049677276045, | |
| "loss": 1.1252, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.193511230562488, | |
| "grad_norm": 1.6143999099731445, | |
| "learning_rate": 0.0001832059708341899, | |
| "loss": 1.3829, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.1942791322710693, | |
| "grad_norm": 1.8929922580718994, | |
| "learning_rate": 0.00018307095369718456, | |
| "loss": 1.4921, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.1950470339796506, | |
| "grad_norm": 1.6425096988677979, | |
| "learning_rate": 0.00018293544615880517, | |
| "loss": 1.4332, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.1958149356882319, | |
| "grad_norm": 1.1571861505508423, | |
| "learning_rate": 0.00018279944901900737, | |
| "loss": 0.9519, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.1965828373968132, | |
| "grad_norm": 1.438349723815918, | |
| "learning_rate": 0.00018266296308063718, | |
| "loss": 1.2061, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.1973507391053945, | |
| "grad_norm": 1.3200489282608032, | |
| "learning_rate": 0.00018252598914942622, | |
| "loss": 1.4497, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.19811864081397582, | |
| "grad_norm": 1.596771240234375, | |
| "learning_rate": 0.00018238852803398689, | |
| "loss": 1.4469, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.1988865425225571, | |
| "grad_norm": 1.5184094905853271, | |
| "learning_rate": 0.00018225058054580765, | |
| "loss": 1.2613, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.1996544442311384, | |
| "grad_norm": 1.5872782468795776, | |
| "learning_rate": 0.0001821121474992482, | |
| "loss": 1.2036, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.20042234593971972, | |
| "grad_norm": 1.4941065311431885, | |
| "learning_rate": 0.00018197322971153467, | |
| "loss": 1.4401, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.201190247648301, | |
| "grad_norm": 1.490402102470398, | |
| "learning_rate": 0.0001818338280027549, | |
| "loss": 1.18, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.20195814935688233, | |
| "grad_norm": 1.4398747682571411, | |
| "learning_rate": 0.00018169394319585345, | |
| "loss": 1.4196, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.20272605106546362, | |
| "grad_norm": 1.7845875024795532, | |
| "learning_rate": 0.00018155357611662672, | |
| "loss": 1.7554, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.20349395277404492, | |
| "grad_norm": 1.4204550981521606, | |
| "learning_rate": 0.0001814127275937183, | |
| "loss": 1.1294, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.20426185448262624, | |
| "grad_norm": 1.406785011291504, | |
| "learning_rate": 0.0001812713984586139, | |
| "loss": 1.4107, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.20502975619120753, | |
| "grad_norm": 1.5216752290725708, | |
| "learning_rate": 0.00018112958954563646, | |
| "loss": 1.2575, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.20579765789978882, | |
| "grad_norm": 1.3387800455093384, | |
| "learning_rate": 0.00018098730169194117, | |
| "loss": 1.2524, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.20656555960837014, | |
| "grad_norm": 1.4416996240615845, | |
| "learning_rate": 0.00018084453573751072, | |
| "loss": 1.1916, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.20733346131695143, | |
| "grad_norm": 1.2977286577224731, | |
| "learning_rate": 0.00018070129252515014, | |
| "loss": 1.2981, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.20810136302553273, | |
| "grad_norm": 1.4423656463623047, | |
| "learning_rate": 0.00018055757290048202, | |
| "loss": 1.3222, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.20886926473411405, | |
| "grad_norm": 1.6373261213302612, | |
| "learning_rate": 0.00018041337771194121, | |
| "loss": 1.4657, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.20963716644269534, | |
| "grad_norm": 1.9930970668792725, | |
| "learning_rate": 0.0001802687078107702, | |
| "loss": 1.2641, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.21040506815127663, | |
| "grad_norm": 1.6621267795562744, | |
| "learning_rate": 0.0001801235640510138, | |
| "loss": 1.4616, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.21117296985985795, | |
| "grad_norm": 1.6404649019241333, | |
| "learning_rate": 0.0001799779472895142, | |
| "loss": 1.5701, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.21194087156843924, | |
| "grad_norm": 1.4635223150253296, | |
| "learning_rate": 0.00017983185838590587, | |
| "loss": 1.2226, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.21270877327702054, | |
| "grad_norm": 1.485813021659851, | |
| "learning_rate": 0.0001796852982026107, | |
| "loss": 1.5223, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.21347667498560186, | |
| "grad_norm": 1.6244434118270874, | |
| "learning_rate": 0.00017953826760483255, | |
| "loss": 1.189, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.21424457669418315, | |
| "grad_norm": 1.3490244150161743, | |
| "learning_rate": 0.00017939076746055239, | |
| "loss": 1.1965, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.21501247840276444, | |
| "grad_norm": 1.5216923952102661, | |
| "learning_rate": 0.00017924279864052313, | |
| "loss": 1.26, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.21578038011134576, | |
| "grad_norm": 1.4194529056549072, | |
| "learning_rate": 0.00017909436201826444, | |
| "loss": 1.1278, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.21654828181992705, | |
| "grad_norm": 1.5075751543045044, | |
| "learning_rate": 0.00017894545847005764, | |
| "loss": 1.2345, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.21731618352850834, | |
| "grad_norm": 1.5216305255889893, | |
| "learning_rate": 0.00017879608887494045, | |
| "loss": 1.2505, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.21808408523708966, | |
| "grad_norm": 1.0526906251907349, | |
| "learning_rate": 0.00017864625411470193, | |
| "loss": 0.9264, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.21885198694567096, | |
| "grad_norm": 1.7272154092788696, | |
| "learning_rate": 0.00017849595507387714, | |
| "loss": 1.3899, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.21961988865425225, | |
| "grad_norm": 1.4222748279571533, | |
| "learning_rate": 0.00017834519263974197, | |
| "loss": 1.2786, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.22038779036283357, | |
| "grad_norm": 1.5858385562896729, | |
| "learning_rate": 0.00017819396770230793, | |
| "loss": 1.58, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.22115569207141486, | |
| "grad_norm": 1.35157310962677, | |
| "learning_rate": 0.0001780422811543169, | |
| "loss": 1.0841, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.22192359377999615, | |
| "grad_norm": 1.4709627628326416, | |
| "learning_rate": 0.00017789013389123582, | |
| "loss": 1.3991, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.22269149548857747, | |
| "grad_norm": 1.6085643768310547, | |
| "learning_rate": 0.00017773752681125133, | |
| "loss": 1.2382, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.22345939719715877, | |
| "grad_norm": 1.9695161581039429, | |
| "learning_rate": 0.00017758446081526472, | |
| "loss": 1.408, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.22422729890574006, | |
| "grad_norm": 2.029320001602173, | |
| "learning_rate": 0.00017743093680688628, | |
| "loss": 1.4736, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.22499520061432138, | |
| "grad_norm": 1.4890549182891846, | |
| "learning_rate": 0.00017727695569243025, | |
| "loss": 1.1405, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.22576310232290267, | |
| "grad_norm": 1.3257640600204468, | |
| "learning_rate": 0.00017712251838090929, | |
| "loss": 1.0384, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.22653100403148396, | |
| "grad_norm": 1.1012932062149048, | |
| "learning_rate": 0.00017696762578402918, | |
| "loss": 0.7738, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.22729890574006528, | |
| "grad_norm": 1.72528874874115, | |
| "learning_rate": 0.0001768122788161835, | |
| "loss": 1.4075, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.22806680744864657, | |
| "grad_norm": 1.4007856845855713, | |
| "learning_rate": 0.00017665647839444808, | |
| "loss": 0.9999, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.22883470915722787, | |
| "grad_norm": 1.8628685474395752, | |
| "learning_rate": 0.0001765002254385757, | |
| "loss": 1.5956, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.2296026108658092, | |
| "grad_norm": 1.5395143032073975, | |
| "learning_rate": 0.0001763435208709906, | |
| "loss": 1.286, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.23037051257439048, | |
| "grad_norm": 1.555917739868164, | |
| "learning_rate": 0.00017618636561678316, | |
| "loss": 1.7136, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.23113841428297177, | |
| "grad_norm": 1.3042889833450317, | |
| "learning_rate": 0.0001760287606037043, | |
| "loss": 1.0877, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.2319063159915531, | |
| "grad_norm": 1.0165395736694336, | |
| "learning_rate": 0.00017587070676215993, | |
| "loss": 0.983, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.23267421770013438, | |
| "grad_norm": 1.264906406402588, | |
| "learning_rate": 0.0001757122050252058, | |
| "loss": 1.1153, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.23344211940871568, | |
| "grad_norm": 1.2855709791183472, | |
| "learning_rate": 0.0001755532563285416, | |
| "loss": 1.3091, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.234210021117297, | |
| "grad_norm": 1.8462400436401367, | |
| "learning_rate": 0.0001753938616105056, | |
| "loss": 1.4876, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.2349779228258783, | |
| "grad_norm": 1.6739815473556519, | |
| "learning_rate": 0.0001752340218120693, | |
| "loss": 1.4114, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.23574582453445958, | |
| "grad_norm": 1.4647902250289917, | |
| "learning_rate": 0.00017507373787683142, | |
| "loss": 1.2797, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.2365137262430409, | |
| "grad_norm": 1.4707120656967163, | |
| "learning_rate": 0.00017491301075101278, | |
| "loss": 1.3396, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.2372816279516222, | |
| "grad_norm": 1.462719202041626, | |
| "learning_rate": 0.0001747518413834505, | |
| "loss": 1.2757, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.23804952966020348, | |
| "grad_norm": 1.5493544340133667, | |
| "learning_rate": 0.0001745902307255924, | |
| "loss": 1.2215, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.2388174313687848, | |
| "grad_norm": 1.4878123998641968, | |
| "learning_rate": 0.00017442817973149145, | |
| "loss": 1.3161, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.2395853330773661, | |
| "grad_norm": 1.3560563325881958, | |
| "learning_rate": 0.0001742656893578001, | |
| "loss": 0.7846, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.2403532347859474, | |
| "grad_norm": 1.3380439281463623, | |
| "learning_rate": 0.00017410276056376456, | |
| "loss": 1.1383, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.2411211364945287, | |
| "grad_norm": 1.7162578105926514, | |
| "learning_rate": 0.00017393939431121933, | |
| "loss": 1.7719, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.24188903820311, | |
| "grad_norm": 1.621044397354126, | |
| "learning_rate": 0.00017377559156458132, | |
| "loss": 1.5056, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.2426569399116913, | |
| "grad_norm": 1.2746073007583618, | |
| "learning_rate": 0.00017361135329084428, | |
| "loss": 1.0386, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.24342484162027261, | |
| "grad_norm": 1.3122072219848633, | |
| "learning_rate": 0.00017344668045957305, | |
| "loss": 0.932, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.2441927433288539, | |
| "grad_norm": 1.3992735147476196, | |
| "learning_rate": 0.0001732815740428978, | |
| "loss": 1.2038, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.2449606450374352, | |
| "grad_norm": 1.61997389793396, | |
| "learning_rate": 0.00017311603501550838, | |
| "loss": 1.5754, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.24572854674601652, | |
| "grad_norm": 1.5020617246627808, | |
| "learning_rate": 0.00017295006435464848, | |
| "loss": 1.197, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.2464964484545978, | |
| "grad_norm": 1.4431666135787964, | |
| "learning_rate": 0.00017278366304010993, | |
| "loss": 1.5506, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.2472643501631791, | |
| "grad_norm": 1.6864901781082153, | |
| "learning_rate": 0.00017261683205422687, | |
| "loss": 1.478, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.24803225187176042, | |
| "grad_norm": 1.3386764526367188, | |
| "learning_rate": 0.00017244957238186993, | |
| "loss": 1.258, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.24880015358034172, | |
| "grad_norm": 1.4950274229049683, | |
| "learning_rate": 0.00017228188501044043, | |
| "loss": 1.2203, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.249568055288923, | |
| "grad_norm": 1.4726964235305786, | |
| "learning_rate": 0.00017211377092986476, | |
| "loss": 1.3879, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2503359569975043, | |
| "grad_norm": 1.3446931838989258, | |
| "learning_rate": 0.00017194523113258804, | |
| "loss": 1.3336, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.2503359569975043, | |
| "eval_loss": 1.2999707460403442, | |
| "eval_runtime": 19.7696, | |
| "eval_samples_per_second": 27.77, | |
| "eval_steps_per_second": 13.91, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.25110385870608565, | |
| "grad_norm": 1.1212903261184692, | |
| "learning_rate": 0.00017177626661356884, | |
| "loss": 0.9078, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.2518717604146669, | |
| "grad_norm": 1.5891238451004028, | |
| "learning_rate": 0.0001716068783702729, | |
| "loss": 1.3183, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.25263966212324823, | |
| "grad_norm": 1.8708879947662354, | |
| "learning_rate": 0.00017143706740266733, | |
| "loss": 1.6728, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.25340756383182955, | |
| "grad_norm": 1.304468035697937, | |
| "learning_rate": 0.00017126683471321494, | |
| "loss": 1.0729, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2541754655404108, | |
| "grad_norm": 2.041584014892578, | |
| "learning_rate": 0.00017109618130686793, | |
| "loss": 1.3812, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.25494336724899214, | |
| "grad_norm": 1.474676251411438, | |
| "learning_rate": 0.00017092510819106228, | |
| "loss": 1.2994, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.25571126895757346, | |
| "grad_norm": 2.100346088409424, | |
| "learning_rate": 0.00017075361637571164, | |
| "loss": 1.3583, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.2564791706661547, | |
| "grad_norm": 1.4596396684646606, | |
| "learning_rate": 0.00017058170687320144, | |
| "loss": 1.3277, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.25724707237473604, | |
| "grad_norm": 1.605150818824768, | |
| "learning_rate": 0.00017040938069838284, | |
| "loss": 1.2923, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.25801497408331736, | |
| "grad_norm": 1.727765679359436, | |
| "learning_rate": 0.00017023663886856681, | |
| "loss": 1.4675, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.2587828757918986, | |
| "grad_norm": 1.5267651081085205, | |
| "learning_rate": 0.0001700634824035182, | |
| "loss": 1.3337, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.25955077750047995, | |
| "grad_norm": 1.5880826711654663, | |
| "learning_rate": 0.00016988991232544943, | |
| "loss": 1.2193, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.26031867920906127, | |
| "grad_norm": 1.7486482858657837, | |
| "learning_rate": 0.00016971592965901472, | |
| "loss": 1.3199, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.26108658091764253, | |
| "grad_norm": 1.4829312562942505, | |
| "learning_rate": 0.00016954153543130405, | |
| "loss": 1.2846, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.26185448262622385, | |
| "grad_norm": 1.501550316810608, | |
| "learning_rate": 0.00016936673067183695, | |
| "loss": 1.2895, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.26262238433480517, | |
| "grad_norm": 1.6724597215652466, | |
| "learning_rate": 0.00016919151641255642, | |
| "loss": 1.7143, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.26339028604338643, | |
| "grad_norm": 1.6089798212051392, | |
| "learning_rate": 0.00016901589368782303, | |
| "loss": 1.4535, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.26415818775196775, | |
| "grad_norm": 1.5275441408157349, | |
| "learning_rate": 0.00016883986353440856, | |
| "loss": 1.2774, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.2649260894605491, | |
| "grad_norm": 1.6726583242416382, | |
| "learning_rate": 0.0001686634269914901, | |
| "loss": 1.3747, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.26569399116913034, | |
| "grad_norm": 1.340077519416809, | |
| "learning_rate": 0.00016848658510064377, | |
| "loss": 1.0331, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.26646189287771166, | |
| "grad_norm": 1.2696207761764526, | |
| "learning_rate": 0.00016830933890583865, | |
| "loss": 1.107, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.267229794586293, | |
| "grad_norm": 1.505295753479004, | |
| "learning_rate": 0.00016813168945343062, | |
| "loss": 1.3706, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.26799769629487424, | |
| "grad_norm": 1.503833293914795, | |
| "learning_rate": 0.000167953637792156, | |
| "loss": 1.3591, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.26876559800345556, | |
| "grad_norm": 1.5831080675125122, | |
| "learning_rate": 0.00016777518497312576, | |
| "loss": 1.4128, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2695334997120369, | |
| "grad_norm": 1.567592978477478, | |
| "learning_rate": 0.00016759633204981885, | |
| "loss": 1.3437, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.27030140142061815, | |
| "grad_norm": 1.4218353033065796, | |
| "learning_rate": 0.00016741708007807625, | |
| "loss": 1.1095, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.27106930312919947, | |
| "grad_norm": 1.5181939601898193, | |
| "learning_rate": 0.0001672374301160948, | |
| "loss": 1.3159, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.2718372048377808, | |
| "grad_norm": 1.3197153806686401, | |
| "learning_rate": 0.00016705738322442067, | |
| "loss": 1.2836, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.27260510654636205, | |
| "grad_norm": 1.5604197978973389, | |
| "learning_rate": 0.0001668769404659434, | |
| "loss": 1.525, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.2733730082549434, | |
| "grad_norm": 1.5216556787490845, | |
| "learning_rate": 0.00016669610290588938, | |
| "loss": 1.2548, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.2741409099635247, | |
| "grad_norm": 1.519769310951233, | |
| "learning_rate": 0.00016651487161181575, | |
| "loss": 1.4625, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.27490881167210596, | |
| "grad_norm": 1.584656834602356, | |
| "learning_rate": 0.00016633324765360404, | |
| "loss": 1.2258, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.2756767133806873, | |
| "grad_norm": 1.3976715803146362, | |
| "learning_rate": 0.00016615123210345374, | |
| "loss": 1.0869, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.2764446150892686, | |
| "grad_norm": 1.6249608993530273, | |
| "learning_rate": 0.00016596882603587613, | |
| "loss": 1.3358, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.27721251679784986, | |
| "grad_norm": 1.8045907020568848, | |
| "learning_rate": 0.00016578603052768787, | |
| "loss": 1.4165, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.2779804185064312, | |
| "grad_norm": 2.2819437980651855, | |
| "learning_rate": 0.00016560284665800463, | |
| "loss": 2.0465, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.2787483202150125, | |
| "grad_norm": 1.4787830114364624, | |
| "learning_rate": 0.00016541927550823475, | |
| "loss": 1.4912, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.27951622192359377, | |
| "grad_norm": 1.8598263263702393, | |
| "learning_rate": 0.00016523531816207285, | |
| "loss": 1.4541, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.2802841236321751, | |
| "grad_norm": 1.5561397075653076, | |
| "learning_rate": 0.00016505097570549334, | |
| "loss": 1.0468, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.2810520253407564, | |
| "grad_norm": 1.2687749862670898, | |
| "learning_rate": 0.00016486624922674423, | |
| "loss": 1.0835, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.28181992704933767, | |
| "grad_norm": 1.1787006855010986, | |
| "learning_rate": 0.0001646811398163405, | |
| "loss": 1.1177, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.282587828757919, | |
| "grad_norm": 1.8415557146072388, | |
| "learning_rate": 0.00016449564856705763, | |
| "loss": 1.5507, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.2833557304665003, | |
| "grad_norm": 1.6068737506866455, | |
| "learning_rate": 0.00016430977657392543, | |
| "loss": 1.1337, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.2841236321750816, | |
| "grad_norm": 1.4251381158828735, | |
| "learning_rate": 0.00016412352493422132, | |
| "loss": 1.1206, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2848915338836629, | |
| "grad_norm": 1.5951745510101318, | |
| "learning_rate": 0.00016393689474746383, | |
| "loss": 1.3409, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.2856594355922442, | |
| "grad_norm": 1.8638418912887573, | |
| "learning_rate": 0.00016374988711540634, | |
| "loss": 1.671, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.2864273373008255, | |
| "grad_norm": 1.7180877923965454, | |
| "learning_rate": 0.00016356250314203044, | |
| "loss": 1.2214, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.2871952390094068, | |
| "grad_norm": 1.66769540309906, | |
| "learning_rate": 0.00016337474393353932, | |
| "loss": 1.3264, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.2879631407179881, | |
| "grad_norm": 1.2491950988769531, | |
| "learning_rate": 0.00016318661059835135, | |
| "loss": 1.0643, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2887310424265694, | |
| "grad_norm": 1.6886341571807861, | |
| "learning_rate": 0.0001629981042470936, | |
| "loss": 1.4799, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.2894989441351507, | |
| "grad_norm": 1.3660120964050293, | |
| "learning_rate": 0.00016280922599259517, | |
| "loss": 1.0789, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.290266845843732, | |
| "grad_norm": 1.5407626628875732, | |
| "learning_rate": 0.00016261997694988064, | |
| "loss": 1.1531, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.2910347475523133, | |
| "grad_norm": 1.512680172920227, | |
| "learning_rate": 0.00016243035823616347, | |
| "loss": 1.3475, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.2918026492608946, | |
| "grad_norm": 1.6978341341018677, | |
| "learning_rate": 0.0001622403709708395, | |
| "loss": 1.4456, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.29257055096947593, | |
| "grad_norm": 1.3576867580413818, | |
| "learning_rate": 0.00016205001627548019, | |
| "loss": 1.2833, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.2933384526780572, | |
| "grad_norm": 1.6945931911468506, | |
| "learning_rate": 0.0001618592952738263, | |
| "loss": 1.2796, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.2941063543866385, | |
| "grad_norm": 1.592065691947937, | |
| "learning_rate": 0.00016166820909178074, | |
| "loss": 1.4813, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.29487425609521983, | |
| "grad_norm": 1.5409668684005737, | |
| "learning_rate": 0.00016147675885740242, | |
| "loss": 1.3686, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.2956421578038011, | |
| "grad_norm": 1.6272224187850952, | |
| "learning_rate": 0.00016128494570089944, | |
| "loss": 1.6073, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.2964100595123824, | |
| "grad_norm": 1.212295651435852, | |
| "learning_rate": 0.0001610927707546222, | |
| "loss": 1.1079, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.29717796122096374, | |
| "grad_norm": 1.5491012334823608, | |
| "learning_rate": 0.00016090023515305703, | |
| "loss": 1.1266, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.297945862929545, | |
| "grad_norm": 1.0519541501998901, | |
| "learning_rate": 0.0001607073400328193, | |
| "loss": 0.9496, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.2987137646381263, | |
| "grad_norm": 1.6982554197311401, | |
| "learning_rate": 0.00016051408653264675, | |
| "loss": 1.4332, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.29948166634670764, | |
| "grad_norm": 1.4598078727722168, | |
| "learning_rate": 0.00016032047579339287, | |
| "loss": 1.1593, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3002495680552889, | |
| "grad_norm": 1.446446180343628, | |
| "learning_rate": 0.00016012650895801995, | |
| "loss": 1.3672, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.3010174697638702, | |
| "grad_norm": 1.4834871292114258, | |
| "learning_rate": 0.00015993218717159254, | |
| "loss": 1.2994, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.30178537147245155, | |
| "grad_norm": 1.433243989944458, | |
| "learning_rate": 0.00015973751158127058, | |
| "loss": 1.083, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.3025532731810328, | |
| "grad_norm": 1.1852916479110718, | |
| "learning_rate": 0.00015954248333630266, | |
| "loss": 0.9979, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.30332117488961413, | |
| "grad_norm": 1.4138494729995728, | |
| "learning_rate": 0.0001593471035880193, | |
| "loss": 0.9835, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.30408907659819545, | |
| "grad_norm": 1.609925389289856, | |
| "learning_rate": 0.00015915137348982596, | |
| "loss": 1.2299, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.3048569783067767, | |
| "grad_norm": 1.60574471950531, | |
| "learning_rate": 0.00015895529419719643, | |
| "loss": 1.3246, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.30562488001535804, | |
| "grad_norm": 1.3476864099502563, | |
| "learning_rate": 0.00015875886686766597, | |
| "loss": 1.1798, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.30639278172393936, | |
| "grad_norm": 1.5925241708755493, | |
| "learning_rate": 0.00015856209266082436, | |
| "loss": 0.8936, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.3071606834325206, | |
| "grad_norm": 1.489413857460022, | |
| "learning_rate": 0.0001583649727383092, | |
| "loss": 1.1598, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.30792858514110194, | |
| "grad_norm": 1.3467170000076294, | |
| "learning_rate": 0.00015816750826379896, | |
| "loss": 1.0321, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.30869648684968326, | |
| "grad_norm": 1.8774082660675049, | |
| "learning_rate": 0.00015796970040300612, | |
| "loss": 1.3866, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.3094643885582645, | |
| "grad_norm": 1.35391104221344, | |
| "learning_rate": 0.0001577715503236704, | |
| "loss": 0.9064, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.31023229026684584, | |
| "grad_norm": 1.7187341451644897, | |
| "learning_rate": 0.00015757305919555164, | |
| "loss": 1.3035, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.31100019197542716, | |
| "grad_norm": 1.5372244119644165, | |
| "learning_rate": 0.00015737422819042313, | |
| "loss": 1.1807, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.31176809368400843, | |
| "grad_norm": 1.558975100517273, | |
| "learning_rate": 0.00015717505848206455, | |
| "loss": 1.273, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.31253599539258975, | |
| "grad_norm": 1.8539949655532837, | |
| "learning_rate": 0.00015697555124625508, | |
| "loss": 1.3312, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.31330389710117107, | |
| "grad_norm": 1.5257269144058228, | |
| "learning_rate": 0.00015677570766076652, | |
| "loss": 1.2561, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.31407179880975233, | |
| "grad_norm": 1.7116575241088867, | |
| "learning_rate": 0.0001565755289053562, | |
| "loss": 1.4553, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.31483970051833365, | |
| "grad_norm": 1.2616760730743408, | |
| "learning_rate": 0.00015637501616176005, | |
| "loss": 1.0677, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.315607602226915, | |
| "grad_norm": 1.6886520385742188, | |
| "learning_rate": 0.00015617417061368586, | |
| "loss": 1.3973, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.31637550393549624, | |
| "grad_norm": 1.5470880270004272, | |
| "learning_rate": 0.0001559729934468059, | |
| "loss": 1.2439, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.31714340564407756, | |
| "grad_norm": 1.5241177082061768, | |
| "learning_rate": 0.0001557714858487502, | |
| "loss": 1.1146, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.3179113073526589, | |
| "grad_norm": 1.266050100326538, | |
| "learning_rate": 0.00015556964900909952, | |
| "loss": 0.9626, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.31867920906124014, | |
| "grad_norm": 1.4324322938919067, | |
| "learning_rate": 0.00015536748411937814, | |
| "loss": 1.424, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.31944711076982146, | |
| "grad_norm": 1.3910094499588013, | |
| "learning_rate": 0.00015516499237304703, | |
| "loss": 1.0697, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.3202150124784028, | |
| "grad_norm": 1.579870343208313, | |
| "learning_rate": 0.00015496217496549673, | |
| "loss": 0.9477, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.32098291418698405, | |
| "grad_norm": 1.598908543586731, | |
| "learning_rate": 0.00015475903309404023, | |
| "loss": 1.3701, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.32175081589556537, | |
| "grad_norm": 1.9276705980300903, | |
| "learning_rate": 0.00015455556795790603, | |
| "loss": 1.268, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.3225187176041467, | |
| "grad_norm": 1.4769033193588257, | |
| "learning_rate": 0.000154351780758231, | |
| "loss": 1.315, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.32328661931272795, | |
| "grad_norm": 1.5065767765045166, | |
| "learning_rate": 0.00015414767269805317, | |
| "loss": 1.2895, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.32405452102130927, | |
| "grad_norm": 1.3060064315795898, | |
| "learning_rate": 0.00015394324498230487, | |
| "loss": 1.3985, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.3248224227298906, | |
| "grad_norm": 1.3480304479599, | |
| "learning_rate": 0.00015373849881780542, | |
| "loss": 1.1568, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.32559032443847186, | |
| "grad_norm": 1.3555710315704346, | |
| "learning_rate": 0.00015353343541325406, | |
| "loss": 1.2957, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.3263582261470532, | |
| "grad_norm": 1.3848627805709839, | |
| "learning_rate": 0.00015332805597922285, | |
| "loss": 1.1681, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.3271261278556345, | |
| "grad_norm": 1.3289916515350342, | |
| "learning_rate": 0.00015312236172814955, | |
| "loss": 1.0471, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.32789402956421576, | |
| "grad_norm": 1.392135739326477, | |
| "learning_rate": 0.0001529163538743303, | |
| "loss": 1.3595, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.3286619312727971, | |
| "grad_norm": 1.2764785289764404, | |
| "learning_rate": 0.00015271003363391268, | |
| "loss": 0.9874, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.3294298329813784, | |
| "grad_norm": 1.879889965057373, | |
| "learning_rate": 0.00015250340222488826, | |
| "loss": 1.4683, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.33019773468995967, | |
| "grad_norm": 1.6945278644561768, | |
| "learning_rate": 0.00015229646086708574, | |
| "loss": 1.2251, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.330965636398541, | |
| "grad_norm": 1.3778647184371948, | |
| "learning_rate": 0.0001520892107821635, | |
| "loss": 1.1494, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.3317335381071223, | |
| "grad_norm": 1.1476179361343384, | |
| "learning_rate": 0.0001518816531936024, | |
| "loss": 1.13, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.33250143981570357, | |
| "grad_norm": 1.8296000957489014, | |
| "learning_rate": 0.0001516737893266987, | |
| "loss": 1.4141, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.3332693415242849, | |
| "grad_norm": 1.7736458778381348, | |
| "learning_rate": 0.00015146562040855676, | |
| "loss": 1.0404, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.3340372432328662, | |
| "grad_norm": 2.155344009399414, | |
| "learning_rate": 0.00015125714766808167, | |
| "loss": 1.567, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.3348051449414475, | |
| "grad_norm": 1.3648680448532104, | |
| "learning_rate": 0.00015104837233597223, | |
| "loss": 1.1135, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.3355730466500288, | |
| "grad_norm": 1.4443109035491943, | |
| "learning_rate": 0.00015083929564471343, | |
| "loss": 1.2181, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.3363409483586101, | |
| "grad_norm": 1.52615225315094, | |
| "learning_rate": 0.00015062991882856946, | |
| "loss": 1.2578, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.3371088500671914, | |
| "grad_norm": 1.5846048593521118, | |
| "learning_rate": 0.00015042024312357616, | |
| "loss": 1.2116, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.3378767517757727, | |
| "grad_norm": 1.6490528583526611, | |
| "learning_rate": 0.00015021026976753385, | |
| "loss": 1.3501, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.338644653484354, | |
| "grad_norm": 1.5763674974441528, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 1.2794, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.3394125551929353, | |
| "grad_norm": 1.7455716133117676, | |
| "learning_rate": 0.000149789435062282, | |
| "loss": 1.3979, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.3401804569015166, | |
| "grad_norm": 1.2413816452026367, | |
| "learning_rate": 0.00014957857619742957, | |
| "loss": 1.0325, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.3409483586100979, | |
| "grad_norm": 1.4034167528152466, | |
| "learning_rate": 0.0001493674246502278, | |
| "loss": 1.0691, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.3417162603186792, | |
| "grad_norm": 1.4105448722839355, | |
| "learning_rate": 0.00014915598166718945, | |
| "loss": 1.4555, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.3424841620272605, | |
| "grad_norm": 1.4534122943878174, | |
| "learning_rate": 0.00014894424849654783, | |
| "loss": 1.2776, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.34325206373584183, | |
| "grad_norm": 1.3057913780212402, | |
| "learning_rate": 0.00014873222638824937, | |
| "loss": 1.121, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.3440199654444231, | |
| "grad_norm": 1.4911497831344604, | |
| "learning_rate": 0.0001485199165939461, | |
| "loss": 1.0317, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.3447878671530044, | |
| "grad_norm": 1.4716236591339111, | |
| "learning_rate": 0.00014830732036698845, | |
| "loss": 1.1425, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.34555576886158573, | |
| "grad_norm": 1.5776805877685547, | |
| "learning_rate": 0.0001480944389624178, | |
| "loss": 1.325, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.346323670570167, | |
| "grad_norm": 1.466970443725586, | |
| "learning_rate": 0.00014788127363695897, | |
| "loss": 1.4133, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.3470915722787483, | |
| "grad_norm": 1.6664904356002808, | |
| "learning_rate": 0.00014766782564901298, | |
| "loss": 1.358, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.34785947398732964, | |
| "grad_norm": 1.333856225013733, | |
| "learning_rate": 0.00014745409625864942, | |
| "loss": 1.1728, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.3486273756959109, | |
| "grad_norm": 1.8161249160766602, | |
| "learning_rate": 0.0001472400867275992, | |
| "loss": 1.2516, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.3493952774044922, | |
| "grad_norm": 1.9915804862976074, | |
| "learning_rate": 0.00014702579831924698, | |
| "loss": 1.6048, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.35016317911307354, | |
| "grad_norm": 1.4382892847061157, | |
| "learning_rate": 0.00014681123229862367, | |
| "loss": 1.2205, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.3509310808216548, | |
| "grad_norm": 1.7863125801086426, | |
| "learning_rate": 0.0001465963899323992, | |
| "loss": 1.3791, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.3516989825302361, | |
| "grad_norm": 1.5871827602386475, | |
| "learning_rate": 0.00014638127248887473, | |
| "loss": 1.2077, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.35246688423881745, | |
| "grad_norm": 1.621598720550537, | |
| "learning_rate": 0.00014616588123797535, | |
| "loss": 1.2283, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.3532347859473987, | |
| "grad_norm": 1.6450941562652588, | |
| "learning_rate": 0.0001459502174512426, | |
| "loss": 1.3417, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.35400268765598003, | |
| "grad_norm": 1.8434786796569824, | |
| "learning_rate": 0.0001457342824018269, | |
| "loss": 1.4919, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.35477058936456135, | |
| "grad_norm": 1.5148212909698486, | |
| "learning_rate": 0.00014551807736447995, | |
| "loss": 1.4935, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.3555384910731426, | |
| "grad_norm": 1.7087806463241577, | |
| "learning_rate": 0.0001453016036155474, | |
| "loss": 1.506, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.35630639278172394, | |
| "grad_norm": 1.463679552078247, | |
| "learning_rate": 0.00014508486243296122, | |
| "loss": 1.1589, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.35707429449030526, | |
| "grad_norm": 1.46501886844635, | |
| "learning_rate": 0.00014486785509623202, | |
| "loss": 1.5455, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.3578421961988865, | |
| "grad_norm": 1.5746129751205444, | |
| "learning_rate": 0.00014465058288644174, | |
| "loss": 1.4495, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.35861009790746784, | |
| "grad_norm": 1.6184135675430298, | |
| "learning_rate": 0.00014443304708623597, | |
| "loss": 1.0957, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.35937799961604916, | |
| "grad_norm": 1.584116816520691, | |
| "learning_rate": 0.00014421524897981637, | |
| "loss": 1.339, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.3601459013246304, | |
| "grad_norm": 1.3535794019699097, | |
| "learning_rate": 0.00014399718985293297, | |
| "loss": 1.0564, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.36091380303321174, | |
| "grad_norm": 1.1871492862701416, | |
| "learning_rate": 0.00014377887099287698, | |
| "loss": 0.9705, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.36168170474179306, | |
| "grad_norm": 1.4988235235214233, | |
| "learning_rate": 0.00014356029368847264, | |
| "loss": 1.4004, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.36244960645037433, | |
| "grad_norm": 1.3582288026809692, | |
| "learning_rate": 0.0001433414592300701, | |
| "loss": 1.1406, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.36321750815895565, | |
| "grad_norm": 1.7595349550247192, | |
| "learning_rate": 0.00014312236890953744, | |
| "loss": 1.3519, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.36398540986753697, | |
| "grad_norm": 1.5235776901245117, | |
| "learning_rate": 0.00014290302402025334, | |
| "loss": 1.0064, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.36475331157611823, | |
| "grad_norm": 2.060851812362671, | |
| "learning_rate": 0.00014268342585709913, | |
| "loss": 1.1825, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.36552121328469955, | |
| "grad_norm": 1.6202908754348755, | |
| "learning_rate": 0.00014246357571645152, | |
| "loss": 1.3926, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.3662891149932809, | |
| "grad_norm": 1.436496376991272, | |
| "learning_rate": 0.00014224347489617456, | |
| "loss": 1.4027, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.36705701670186214, | |
| "grad_norm": 1.5660499334335327, | |
| "learning_rate": 0.00014202312469561228, | |
| "loss": 1.2506, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.36782491841044346, | |
| "grad_norm": 1.473960041999817, | |
| "learning_rate": 0.00014180252641558084, | |
| "loss": 1.2378, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.3685928201190248, | |
| "grad_norm": 1.4252207279205322, | |
| "learning_rate": 0.00014158168135836095, | |
| "loss": 1.1854, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.36936072182760604, | |
| "grad_norm": 1.6253414154052734, | |
| "learning_rate": 0.00014136059082769017, | |
| "loss": 1.5745, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.37012862353618736, | |
| "grad_norm": 1.7292263507843018, | |
| "learning_rate": 0.00014113925612875512, | |
| "loss": 1.41, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.3708965252447687, | |
| "grad_norm": 1.7278509140014648, | |
| "learning_rate": 0.00014091767856818388, | |
| "loss": 1.3054, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.37166442695334995, | |
| "grad_norm": 1.8640563488006592, | |
| "learning_rate": 0.00014069585945403822, | |
| "loss": 1.5207, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.37243232866193127, | |
| "grad_norm": 1.6371948719024658, | |
| "learning_rate": 0.00014047380009580594, | |
| "loss": 1.3498, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.3732002303705126, | |
| "grad_norm": 1.2843165397644043, | |
| "learning_rate": 0.00014025150180439308, | |
| "loss": 1.0629, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.37396813207909385, | |
| "grad_norm": 1.539959192276001, | |
| "learning_rate": 0.00014002896589211618, | |
| "loss": 1.2718, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.37473603378767517, | |
| "grad_norm": 1.709976315498352, | |
| "learning_rate": 0.00013980619367269455, | |
| "loss": 1.1173, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.3755039354962565, | |
| "grad_norm": 1.6751576662063599, | |
| "learning_rate": 0.00013958318646124259, | |
| "loss": 1.5058, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.37627183720483776, | |
| "grad_norm": 1.8899919986724854, | |
| "learning_rate": 0.0001393599455742618, | |
| "loss": 1.5131, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3770397389134191, | |
| "grad_norm": 1.3949605226516724, | |
| "learning_rate": 0.00013913647232963332, | |
| "loss": 1.194, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.3778076406220004, | |
| "grad_norm": 1.3396071195602417, | |
| "learning_rate": 0.00013891276804660991, | |
| "loss": 1.178, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.37857554233058166, | |
| "grad_norm": 1.8949428796768188, | |
| "learning_rate": 0.00013868883404580823, | |
| "loss": 1.5164, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.379343444039163, | |
| "grad_norm": 1.6502594947814941, | |
| "learning_rate": 0.00013846467164920116, | |
| "loss": 1.1968, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.3801113457477443, | |
| "grad_norm": 1.6355159282684326, | |
| "learning_rate": 0.00013824028218010977, | |
| "loss": 1.4489, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.38087924745632556, | |
| "grad_norm": 1.509320855140686, | |
| "learning_rate": 0.00013801566696319562, | |
| "loss": 1.169, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.3816471491649069, | |
| "grad_norm": 1.206089735031128, | |
| "learning_rate": 0.0001377908273244531, | |
| "loss": 0.9691, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.3824150508734882, | |
| "grad_norm": 1.6482902765274048, | |
| "learning_rate": 0.0001375657645912014, | |
| "loss": 1.0815, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.38318295258206947, | |
| "grad_norm": 1.538040280342102, | |
| "learning_rate": 0.0001373404800920765, | |
| "loss": 1.528, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.3839508542906508, | |
| "grad_norm": 1.4498200416564941, | |
| "learning_rate": 0.00013711497515702398, | |
| "loss": 1.2253, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3847187559992321, | |
| "grad_norm": 1.4636632204055786, | |
| "learning_rate": 0.0001368892511172903, | |
| "loss": 1.3644, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.3854866577078134, | |
| "grad_norm": 1.5408172607421875, | |
| "learning_rate": 0.0001366633093054157, | |
| "loss": 1.3672, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.3862545594163947, | |
| "grad_norm": 1.4577535390853882, | |
| "learning_rate": 0.00013643715105522589, | |
| "loss": 1.2762, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.387022461124976, | |
| "grad_norm": 1.4348138570785522, | |
| "learning_rate": 0.0001362107777018243, | |
| "loss": 1.1911, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.3877903628335573, | |
| "grad_norm": 1.8007947206497192, | |
| "learning_rate": 0.0001359841905815842, | |
| "loss": 1.6611, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.3885582645421386, | |
| "grad_norm": 1.3817611932754517, | |
| "learning_rate": 0.00013575739103214088, | |
| "loss": 0.949, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.3893261662507199, | |
| "grad_norm": 1.5751131772994995, | |
| "learning_rate": 0.0001355303803923836, | |
| "loss": 1.4595, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.3900940679593012, | |
| "grad_norm": 1.4235345125198364, | |
| "learning_rate": 0.00013530316000244782, | |
| "loss": 1.1157, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.3908619696678825, | |
| "grad_norm": 1.4169321060180664, | |
| "learning_rate": 0.0001350757312037072, | |
| "loss": 1.1267, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.3916298713764638, | |
| "grad_norm": 1.4583044052124023, | |
| "learning_rate": 0.00013484809533876582, | |
| "loss": 1.0842, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.3923977730850451, | |
| "grad_norm": 1.3082489967346191, | |
| "learning_rate": 0.00013462025375145, | |
| "loss": 1.1793, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.3931656747936264, | |
| "grad_norm": 1.9846649169921875, | |
| "learning_rate": 0.00013439220778680067, | |
| "loss": 1.7083, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.3939335765022077, | |
| "grad_norm": 1.6533915996551514, | |
| "learning_rate": 0.00013416395879106515, | |
| "loss": 1.3144, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.394701478210789, | |
| "grad_norm": 1.5912500619888306, | |
| "learning_rate": 0.00013393550811168948, | |
| "loss": 1.3611, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.3954693799193703, | |
| "grad_norm": 1.2825086116790771, | |
| "learning_rate": 0.00013370685709731015, | |
| "loss": 1.2001, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.39623728162795163, | |
| "grad_norm": 1.493638515472412, | |
| "learning_rate": 0.00013347800709774652, | |
| "loss": 1.125, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.3970051833365329, | |
| "grad_norm": 1.3047711849212646, | |
| "learning_rate": 0.0001332489594639924, | |
| "loss": 1.031, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.3977730850451142, | |
| "grad_norm": 1.6641656160354614, | |
| "learning_rate": 0.00013301971554820853, | |
| "loss": 1.3513, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.39854098675369554, | |
| "grad_norm": 1.4950916767120361, | |
| "learning_rate": 0.00013279027670371426, | |
| "loss": 1.3485, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.3993088884622768, | |
| "grad_norm": 1.5947059392929077, | |
| "learning_rate": 0.00013256064428497966, | |
| "loss": 1.3895, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.4000767901708581, | |
| "grad_norm": 1.307298183441162, | |
| "learning_rate": 0.00013233081964761766, | |
| "loss": 1.3044, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.40084469187943944, | |
| "grad_norm": 1.5660213232040405, | |
| "learning_rate": 0.0001321008041483758, | |
| "loss": 1.6157, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.40161259358802076, | |
| "grad_norm": 1.3612569570541382, | |
| "learning_rate": 0.0001318705991451285, | |
| "loss": 1.226, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.402380495296602, | |
| "grad_norm": 1.351832628250122, | |
| "learning_rate": 0.00013164020599686882, | |
| "loss": 1.0547, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.40314839700518335, | |
| "grad_norm": 1.3935625553131104, | |
| "learning_rate": 0.00013140962606370048, | |
| "loss": 1.2035, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.40391629871376467, | |
| "grad_norm": 1.1423629522323608, | |
| "learning_rate": 0.0001311788607068299, | |
| "loss": 0.7527, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.40468420042234593, | |
| "grad_norm": 1.5600942373275757, | |
| "learning_rate": 0.00013094791128855814, | |
| "loss": 1.0371, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.40545210213092725, | |
| "grad_norm": 1.6167805194854736, | |
| "learning_rate": 0.0001307167791722729, | |
| "loss": 1.2223, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.40622000383950857, | |
| "grad_norm": 1.5059945583343506, | |
| "learning_rate": 0.00013048546572244036, | |
| "loss": 1.1128, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.40698790554808983, | |
| "grad_norm": 1.4017764329910278, | |
| "learning_rate": 0.0001302539723045971, | |
| "loss": 1.1994, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.40775580725667115, | |
| "grad_norm": 1.4649596214294434, | |
| "learning_rate": 0.00013002230028534234, | |
| "loss": 1.3015, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.4085237089652525, | |
| "grad_norm": 1.5319805145263672, | |
| "learning_rate": 0.00012979045103232945, | |
| "loss": 1.1543, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.40929161067383374, | |
| "grad_norm": 1.2714585065841675, | |
| "learning_rate": 0.00012955842591425818, | |
| "loss": 1.067, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.41005951238241506, | |
| "grad_norm": 1.4179991483688354, | |
| "learning_rate": 0.00012932622630086648, | |
| "loss": 1.247, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.4108274140909964, | |
| "grad_norm": 1.4584649801254272, | |
| "learning_rate": 0.0001290938535629224, | |
| "loss": 1.0901, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.41159531579957764, | |
| "grad_norm": 1.2720510959625244, | |
| "learning_rate": 0.00012886130907221603, | |
| "loss": 0.9964, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.41236321750815896, | |
| "grad_norm": 1.5172308683395386, | |
| "learning_rate": 0.00012862859420155134, | |
| "loss": 1.1805, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.4131311192167403, | |
| "grad_norm": 1.5134514570236206, | |
| "learning_rate": 0.00012839571032473814, | |
| "loss": 1.2426, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.41389902092532155, | |
| "grad_norm": 1.5913889408111572, | |
| "learning_rate": 0.00012816265881658405, | |
| "loss": 1.3287, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.41466692263390287, | |
| "grad_norm": 1.4839410781860352, | |
| "learning_rate": 0.00012792944105288612, | |
| "loss": 1.2787, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4154348243424842, | |
| "grad_norm": 1.7110356092453003, | |
| "learning_rate": 0.000127696058410423, | |
| "loss": 1.6284, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.41620272605106545, | |
| "grad_norm": 1.369132161140442, | |
| "learning_rate": 0.00012746251226694662, | |
| "loss": 1.3544, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.4169706277596468, | |
| "grad_norm": 1.293350338935852, | |
| "learning_rate": 0.00012722880400117413, | |
| "loss": 1.1775, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.4177385294682281, | |
| "grad_norm": 1.5114126205444336, | |
| "learning_rate": 0.00012699493499277983, | |
| "loss": 1.4465, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.41850643117680936, | |
| "grad_norm": 1.5067218542099, | |
| "learning_rate": 0.00012676090662238682, | |
| "loss": 1.3107, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.4192743328853907, | |
| "grad_norm": 1.2761341333389282, | |
| "learning_rate": 0.00012652672027155904, | |
| "loss": 1.0406, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.420042234593972, | |
| "grad_norm": 1.493939757347107, | |
| "learning_rate": 0.00012629237732279314, | |
| "loss": 1.2623, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.42081013630255326, | |
| "grad_norm": 1.484665870666504, | |
| "learning_rate": 0.0001260578791595101, | |
| "loss": 1.459, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.4215780380111346, | |
| "grad_norm": 1.2791699171066284, | |
| "learning_rate": 0.00012582322716604718, | |
| "loss": 0.9361, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.4223459397197159, | |
| "grad_norm": 1.5707510709762573, | |
| "learning_rate": 0.0001255884227276499, | |
| "loss": 1.1258, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.42311384142829717, | |
| "grad_norm": 1.6234004497528076, | |
| "learning_rate": 0.0001253534672304636, | |
| "loss": 1.2735, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.4238817431368785, | |
| "grad_norm": 1.4539543390274048, | |
| "learning_rate": 0.00012511836206152545, | |
| "loss": 1.2575, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.4246496448454598, | |
| "grad_norm": 1.4913365840911865, | |
| "learning_rate": 0.00012488310860875622, | |
| "loss": 1.4831, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.42541754655404107, | |
| "grad_norm": 1.9435288906097412, | |
| "learning_rate": 0.0001246477082609519, | |
| "loss": 0.9954, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.4261854482626224, | |
| "grad_norm": 1.5525418519973755, | |
| "learning_rate": 0.00012441216240777585, | |
| "loss": 1.2844, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.4269533499712037, | |
| "grad_norm": 1.238305926322937, | |
| "learning_rate": 0.0001241764724397503, | |
| "loss": 0.8637, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.427721251679785, | |
| "grad_norm": 1.7352509498596191, | |
| "learning_rate": 0.00012394063974824828, | |
| "loss": 1.3207, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.4284891533883663, | |
| "grad_norm": 1.5795037746429443, | |
| "learning_rate": 0.00012370466572548538, | |
| "loss": 1.2913, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.4292570550969476, | |
| "grad_norm": 1.6189255714416504, | |
| "learning_rate": 0.0001234685517645115, | |
| "loss": 1.154, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.4300249568055289, | |
| "grad_norm": 1.5668927431106567, | |
| "learning_rate": 0.00012323229925920273, | |
| "loss": 1.4944, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4307928585141102, | |
| "grad_norm": 1.6585174798965454, | |
| "learning_rate": 0.00012299590960425288, | |
| "loss": 1.2406, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.4315607602226915, | |
| "grad_norm": 1.6430768966674805, | |
| "learning_rate": 0.00012275938419516552, | |
| "loss": 1.2662, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.4323286619312728, | |
| "grad_norm": 1.5714935064315796, | |
| "learning_rate": 0.0001225227244282457, | |
| "loss": 1.2358, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.4330965636398541, | |
| "grad_norm": 1.551619291305542, | |
| "learning_rate": 0.00012228593170059151, | |
| "loss": 1.0941, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.4338644653484354, | |
| "grad_norm": 1.6610519886016846, | |
| "learning_rate": 0.000122049007410086, | |
| "loss": 1.3556, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.4346323670570167, | |
| "grad_norm": 1.4311244487762451, | |
| "learning_rate": 0.00012181195295538895, | |
| "loss": 1.1336, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.435400268765598, | |
| "grad_norm": 1.7137714624404907, | |
| "learning_rate": 0.00012157476973592842, | |
| "loss": 1.4885, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.43616817047417933, | |
| "grad_norm": 1.6225979328155518, | |
| "learning_rate": 0.00012133745915189278, | |
| "loss": 1.216, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.4369360721827606, | |
| "grad_norm": 1.5688353776931763, | |
| "learning_rate": 0.00012110002260422218, | |
| "loss": 1.2242, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.4377039738913419, | |
| "grad_norm": 1.724080204963684, | |
| "learning_rate": 0.00012086246149460038, | |
| "loss": 1.2706, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.43847187559992323, | |
| "grad_norm": 1.465526819229126, | |
| "learning_rate": 0.00012062477722544656, | |
| "loss": 1.0762, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.4392397773085045, | |
| "grad_norm": 1.3275350332260132, | |
| "learning_rate": 0.00012038697119990687, | |
| "loss": 1.2327, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.4400076790170858, | |
| "grad_norm": 1.5525063276290894, | |
| "learning_rate": 0.00012014904482184633, | |
| "loss": 1.3994, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.44077558072566714, | |
| "grad_norm": 1.318934679031372, | |
| "learning_rate": 0.00011991099949584032, | |
| "loss": 1.4836, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.4415434824342484, | |
| "grad_norm": 1.376668930053711, | |
| "learning_rate": 0.00011967283662716653, | |
| "loss": 1.2234, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.4423113841428297, | |
| "grad_norm": 1.4669725894927979, | |
| "learning_rate": 0.00011943455762179654, | |
| "loss": 1.3136, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.44307928585141104, | |
| "grad_norm": 1.3037981986999512, | |
| "learning_rate": 0.00011919616388638748, | |
| "loss": 1.2784, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.4438471875599923, | |
| "grad_norm": 1.8770321607589722, | |
| "learning_rate": 0.0001189576568282738, | |
| "loss": 1.4478, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.4446150892685736, | |
| "grad_norm": 1.633165717124939, | |
| "learning_rate": 0.00011871903785545897, | |
| "loss": 1.4224, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.44538299097715495, | |
| "grad_norm": 1.5755568742752075, | |
| "learning_rate": 0.00011848030837660709, | |
| "loss": 1.2581, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4461508926857362, | |
| "grad_norm": 1.5348775386810303, | |
| "learning_rate": 0.00011824146980103467, | |
| "loss": 1.0031, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.44691879439431753, | |
| "grad_norm": 1.3276475667953491, | |
| "learning_rate": 0.00011800252353870224, | |
| "loss": 1.0511, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.44768669610289885, | |
| "grad_norm": 2.070185661315918, | |
| "learning_rate": 0.00011776347100020602, | |
| "loss": 1.4314, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.4484545978114801, | |
| "grad_norm": 1.82218337059021, | |
| "learning_rate": 0.00011752431359676968, | |
| "loss": 1.5193, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.44922249952006144, | |
| "grad_norm": 1.479628086090088, | |
| "learning_rate": 0.00011728505274023584, | |
| "loss": 1.1324, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.44999040122864276, | |
| "grad_norm": 1.273962378501892, | |
| "learning_rate": 0.00011704568984305802, | |
| "loss": 1.0844, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.450758302937224, | |
| "grad_norm": 1.537895917892456, | |
| "learning_rate": 0.00011680622631829197, | |
| "loss": 1.2109, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.45152620464580534, | |
| "grad_norm": 1.5234670639038086, | |
| "learning_rate": 0.00011656666357958751, | |
| "loss": 1.3014, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.45229410635438666, | |
| "grad_norm": 1.6489896774291992, | |
| "learning_rate": 0.00011632700304118032, | |
| "loss": 1.4389, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.4530620080629679, | |
| "grad_norm": 1.419973373413086, | |
| "learning_rate": 0.0001160872461178832, | |
| "loss": 1.2462, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.45382990977154924, | |
| "grad_norm": 1.0888590812683105, | |
| "learning_rate": 0.00011584739422507804, | |
| "loss": 0.9298, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.45459781148013056, | |
| "grad_norm": 1.7613029479980469, | |
| "learning_rate": 0.00011560744877870748, | |
| "loss": 1.3252, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.45536571318871183, | |
| "grad_norm": 1.5147452354431152, | |
| "learning_rate": 0.00011536741119526628, | |
| "loss": 1.3277, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.45613361489729315, | |
| "grad_norm": 1.3303625583648682, | |
| "learning_rate": 0.00011512728289179323, | |
| "loss": 1.1962, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.45690151660587447, | |
| "grad_norm": 1.591471552848816, | |
| "learning_rate": 0.00011488706528586261, | |
| "loss": 1.324, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.45766941831445573, | |
| "grad_norm": 1.9963793754577637, | |
| "learning_rate": 0.00011464675979557593, | |
| "loss": 1.4472, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.45843732002303705, | |
| "grad_norm": 1.6944851875305176, | |
| "learning_rate": 0.00011440636783955356, | |
| "loss": 1.3831, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.4592052217316184, | |
| "grad_norm": 1.851412296295166, | |
| "learning_rate": 0.00011416589083692619, | |
| "loss": 1.3932, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.45997312344019964, | |
| "grad_norm": 1.5847994089126587, | |
| "learning_rate": 0.00011392533020732666, | |
| "loss": 1.1381, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.46074102514878096, | |
| "grad_norm": 1.6723082065582275, | |
| "learning_rate": 0.00011368468737088148, | |
| "loss": 1.5375, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.4615089268573623, | |
| "grad_norm": 1.5877834558486938, | |
| "learning_rate": 0.00011344396374820244, | |
| "loss": 1.3328, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.46227682856594354, | |
| "grad_norm": 1.797132968902588, | |
| "learning_rate": 0.0001132031607603783, | |
| "loss": 1.2852, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.46304473027452486, | |
| "grad_norm": 1.5253082513809204, | |
| "learning_rate": 0.0001129622798289663, | |
| "loss": 1.2521, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.4638126319831062, | |
| "grad_norm": 1.8899868726730347, | |
| "learning_rate": 0.00011272132237598376, | |
| "loss": 1.3255, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.46458053369168745, | |
| "grad_norm": 1.6422898769378662, | |
| "learning_rate": 0.00011248028982389989, | |
| "loss": 1.1699, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.46534843540026877, | |
| "grad_norm": 1.5534974336624146, | |
| "learning_rate": 0.00011223918359562708, | |
| "loss": 1.2355, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.4661163371088501, | |
| "grad_norm": 1.4911137819290161, | |
| "learning_rate": 0.00011199800511451273, | |
| "loss": 1.0746, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.46688423881743135, | |
| "grad_norm": 1.4544899463653564, | |
| "learning_rate": 0.0001117567558043308, | |
| "loss": 1.1927, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.46765214052601267, | |
| "grad_norm": 1.744524598121643, | |
| "learning_rate": 0.00011151543708927335, | |
| "loss": 1.6512, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.468420042234594, | |
| "grad_norm": 1.4004900455474854, | |
| "learning_rate": 0.00011127405039394216, | |
| "loss": 1.0693, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.46918794394317526, | |
| "grad_norm": 1.452285885810852, | |
| "learning_rate": 0.00011103259714334034, | |
| "loss": 1.0925, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.4699558456517566, | |
| "grad_norm": 1.1670202016830444, | |
| "learning_rate": 0.00011079107876286387, | |
| "loss": 0.9021, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.4707237473603379, | |
| "grad_norm": 1.809617042541504, | |
| "learning_rate": 0.0001105494966782933, | |
| "loss": 1.3879, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.47149164906891916, | |
| "grad_norm": 1.4755923748016357, | |
| "learning_rate": 0.0001103078523157852, | |
| "loss": 1.2053, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.4722595507775005, | |
| "grad_norm": 2.111751079559326, | |
| "learning_rate": 0.00011006614710186372, | |
| "loss": 1.5057, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.4730274524860818, | |
| "grad_norm": 1.3958059549331665, | |
| "learning_rate": 0.00010982438246341238, | |
| "loss": 1.196, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.47379535419466307, | |
| "grad_norm": 1.6343685388565063, | |
| "learning_rate": 0.00010958255982766538, | |
| "loss": 1.2389, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.4745632559032444, | |
| "grad_norm": 1.746578335762024, | |
| "learning_rate": 0.00010934068062219945, | |
| "loss": 1.3113, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.4753311576118257, | |
| "grad_norm": 1.474244475364685, | |
| "learning_rate": 0.0001090987462749251, | |
| "loss": 1.2434, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.47609905932040697, | |
| "grad_norm": 1.774789810180664, | |
| "learning_rate": 0.00010885675821407844, | |
| "loss": 1.3742, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.4768669610289883, | |
| "grad_norm": 1.5622092485427856, | |
| "learning_rate": 0.00010861471786821275, | |
| "loss": 1.2501, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.4776348627375696, | |
| "grad_norm": 1.8582671880722046, | |
| "learning_rate": 0.00010837262666618983, | |
| "loss": 1.3851, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.4784027644461509, | |
| "grad_norm": 1.37471342086792, | |
| "learning_rate": 0.00010813048603717182, | |
| "loss": 1.2147, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.4791706661547322, | |
| "grad_norm": 1.9024494886398315, | |
| "learning_rate": 0.0001078882974106126, | |
| "loss": 1.4406, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.4799385678633135, | |
| "grad_norm": 1.6293984651565552, | |
| "learning_rate": 0.00010764606221624933, | |
| "loss": 1.2317, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.4807064695718948, | |
| "grad_norm": 1.6357662677764893, | |
| "learning_rate": 0.00010740378188409426, | |
| "loss": 1.1184, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.4814743712804761, | |
| "grad_norm": 1.6468091011047363, | |
| "learning_rate": 0.00010716145784442593, | |
| "loss": 1.2806, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.4822422729890574, | |
| "grad_norm": 1.6838487386703491, | |
| "learning_rate": 0.00010691909152778094, | |
| "loss": 1.313, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.4830101746976387, | |
| "grad_norm": 1.7010256052017212, | |
| "learning_rate": 0.00010667668436494558, | |
| "loss": 1.2907, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.48377807640622, | |
| "grad_norm": 1.6259071826934814, | |
| "learning_rate": 0.00010643423778694712, | |
| "loss": 1.6014, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.4845459781148013, | |
| "grad_norm": 1.5683881044387817, | |
| "learning_rate": 0.0001061917532250456, | |
| "loss": 1.3454, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.4853138798233826, | |
| "grad_norm": 1.4420804977416992, | |
| "learning_rate": 0.00010594923211072532, | |
| "loss": 1.4059, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.4860817815319639, | |
| "grad_norm": 1.7529085874557495, | |
| "learning_rate": 0.00010570667587568626, | |
| "loss": 1.3871, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.48684968324054523, | |
| "grad_norm": 1.5442010164260864, | |
| "learning_rate": 0.00010546408595183578, | |
| "loss": 1.0737, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.4876175849491265, | |
| "grad_norm": 1.6395798921585083, | |
| "learning_rate": 0.00010522146377128021, | |
| "loss": 1.2025, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.4883854866577078, | |
| "grad_norm": 1.6505008935928345, | |
| "learning_rate": 0.00010497881076631615, | |
| "loss": 1.231, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.48915338836628913, | |
| "grad_norm": 1.7394682168960571, | |
| "learning_rate": 0.00010473612836942226, | |
| "loss": 1.6761, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.4899212900748704, | |
| "grad_norm": 1.410757303237915, | |
| "learning_rate": 0.00010449341801325073, | |
| "loss": 1.2553, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.4906891917834517, | |
| "grad_norm": 1.3731025457382202, | |
| "learning_rate": 0.00010425068113061873, | |
| "loss": 1.4168, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.49145709349203304, | |
| "grad_norm": 1.296189546585083, | |
| "learning_rate": 0.00010400791915450009, | |
| "loss": 1.105, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.4922249952006143, | |
| "grad_norm": 1.17586350440979, | |
| "learning_rate": 0.00010376513351801673, | |
| "loss": 0.9808, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.4929928969091956, | |
| "grad_norm": 2.6915199756622314, | |
| "learning_rate": 0.00010352232565443032, | |
| "loss": 1.4701, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.49376079861777694, | |
| "grad_norm": 1.942378044128418, | |
| "learning_rate": 0.00010327949699713366, | |
| "loss": 1.3895, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.4945287003263582, | |
| "grad_norm": 1.7854747772216797, | |
| "learning_rate": 0.00010303664897964232, | |
| "loss": 1.1927, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.4952966020349395, | |
| "grad_norm": 1.4492199420928955, | |
| "learning_rate": 0.00010279378303558624, | |
| "loss": 1.4889, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.49606450374352085, | |
| "grad_norm": 1.6698827743530273, | |
| "learning_rate": 0.00010255090059870107, | |
| "loss": 1.4768, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.4968324054521021, | |
| "grad_norm": 1.533857822418213, | |
| "learning_rate": 0.00010230800310281992, | |
| "loss": 1.3059, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.49760030716068343, | |
| "grad_norm": 1.816888689994812, | |
| "learning_rate": 0.00010206509198186476, | |
| "loss": 1.5369, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.49836820886926475, | |
| "grad_norm": 1.3209645748138428, | |
| "learning_rate": 0.00010182216866983796, | |
| "loss": 1.2222, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.499136110577846, | |
| "grad_norm": 1.441152811050415, | |
| "learning_rate": 0.00010157923460081394, | |
| "loss": 1.4053, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.49990401228642733, | |
| "grad_norm": 1.3601536750793457, | |
| "learning_rate": 0.00010133629120893055, | |
| "loss": 1.3098, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.5006719139950087, | |
| "grad_norm": 1.6354866027832031, | |
| "learning_rate": 0.00010109333992838072, | |
| "loss": 1.2891, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.5006719139950087, | |
| "eval_loss": 1.273740291595459, | |
| "eval_runtime": 19.7598, | |
| "eval_samples_per_second": 27.784, | |
| "eval_steps_per_second": 13.917, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.50143981570359, | |
| "grad_norm": 1.54447603225708, | |
| "learning_rate": 0.00010085038219340393, | |
| "loss": 1.296, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.5022077174121713, | |
| "grad_norm": 1.8366080522537231, | |
| "learning_rate": 0.00010060741943827776, | |
| "loss": 1.6111, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.5029756191207525, | |
| "grad_norm": 1.22926926612854, | |
| "learning_rate": 0.00010036445309730944, | |
| "loss": 1.0101, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.5037435208293338, | |
| "grad_norm": 1.0615041255950928, | |
| "learning_rate": 0.00010012148460482738, | |
| "loss": 0.8333, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.5045114225379151, | |
| "grad_norm": 1.7094695568084717, | |
| "learning_rate": 9.987851539517262e-05, | |
| "loss": 1.094, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.5052793242464965, | |
| "grad_norm": 1.4576027393341064, | |
| "learning_rate": 9.963554690269058e-05, | |
| "loss": 0.957, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.5060472259550778, | |
| "grad_norm": 1.4120879173278809, | |
| "learning_rate": 9.939258056172225e-05, | |
| "loss": 1.1924, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.5068151276636591, | |
| "grad_norm": 1.4560344219207764, | |
| "learning_rate": 9.914961780659609e-05, | |
| "loss": 1.0673, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5075830293722403, | |
| "grad_norm": 1.4893819093704224, | |
| "learning_rate": 9.890666007161929e-05, | |
| "loss": 1.1459, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.5083509310808216, | |
| "grad_norm": 1.5299136638641357, | |
| "learning_rate": 9.866370879106947e-05, | |
| "loss": 1.2983, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.509118832789403, | |
| "grad_norm": 1.6000056266784668, | |
| "learning_rate": 9.84207653991861e-05, | |
| "loss": 1.0902, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.5098867344979843, | |
| "grad_norm": 1.5117030143737793, | |
| "learning_rate": 9.817783133016206e-05, | |
| "loss": 1.2475, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.5106546362065656, | |
| "grad_norm": 1.7616103887557983, | |
| "learning_rate": 9.793490801813528e-05, | |
| "loss": 1.4153, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.5114225379151469, | |
| "grad_norm": 1.6097863912582397, | |
| "learning_rate": 9.769199689718009e-05, | |
| "loss": 1.3362, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.5121904396237281, | |
| "grad_norm": 1.431277871131897, | |
| "learning_rate": 9.744909940129895e-05, | |
| "loss": 1.2254, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.5129583413323094, | |
| "grad_norm": 1.4436259269714355, | |
| "learning_rate": 9.720621696441378e-05, | |
| "loss": 1.1335, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.5137262430408908, | |
| "grad_norm": 1.7994132041931152, | |
| "learning_rate": 9.69633510203577e-05, | |
| "loss": 1.3912, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.5144941447494721, | |
| "grad_norm": 1.828708529472351, | |
| "learning_rate": 9.672050300286636e-05, | |
| "loss": 1.5837, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.5152620464580534, | |
| "grad_norm": 1.592246651649475, | |
| "learning_rate": 9.64776743455697e-05, | |
| "loss": 1.3176, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.5160299481666347, | |
| "grad_norm": 1.7357617616653442, | |
| "learning_rate": 9.623486648198326e-05, | |
| "loss": 1.47, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.5167978498752159, | |
| "grad_norm": 1.6944161653518677, | |
| "learning_rate": 9.599208084549993e-05, | |
| "loss": 1.4586, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.5175657515837973, | |
| "grad_norm": 1.827109456062317, | |
| "learning_rate": 9.574931886938128e-05, | |
| "loss": 1.2451, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.5183336532923786, | |
| "grad_norm": 1.3678934574127197, | |
| "learning_rate": 9.550658198674931e-05, | |
| "loss": 1.2403, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.5191015550009599, | |
| "grad_norm": 0.994289755821228, | |
| "learning_rate": 9.526387163057777e-05, | |
| "loss": 0.8628, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.5198694567095412, | |
| "grad_norm": 1.3589410781860352, | |
| "learning_rate": 9.502118923368388e-05, | |
| "loss": 0.8397, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.5206373584181225, | |
| "grad_norm": 1.3971327543258667, | |
| "learning_rate": 9.477853622871984e-05, | |
| "loss": 1.1653, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.5214052601267037, | |
| "grad_norm": 1.3670016527175903, | |
| "learning_rate": 9.453591404816423e-05, | |
| "loss": 1.0544, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.5221731618352851, | |
| "grad_norm": 1.3910092115402222, | |
| "learning_rate": 9.429332412431377e-05, | |
| "loss": 1.2027, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5229410635438664, | |
| "grad_norm": 1.4565633535385132, | |
| "learning_rate": 9.405076788927469e-05, | |
| "loss": 1.3962, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.5237089652524477, | |
| "grad_norm": 1.4770212173461914, | |
| "learning_rate": 9.380824677495441e-05, | |
| "loss": 1.3789, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.524476866961029, | |
| "grad_norm": 1.5948761701583862, | |
| "learning_rate": 9.356576221305289e-05, | |
| "loss": 1.4457, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.5252447686696103, | |
| "grad_norm": 1.4264615774154663, | |
| "learning_rate": 9.332331563505444e-05, | |
| "loss": 1.0041, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.5260126703781915, | |
| "grad_norm": 1.5794204473495483, | |
| "learning_rate": 9.308090847221905e-05, | |
| "loss": 1.3674, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.5267805720867729, | |
| "grad_norm": 1.5819778442382812, | |
| "learning_rate": 9.283854215557409e-05, | |
| "loss": 1.5201, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.5275484737953542, | |
| "grad_norm": 1.3187499046325684, | |
| "learning_rate": 9.259621811590578e-05, | |
| "loss": 1.1523, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.5283163755039355, | |
| "grad_norm": 1.585579514503479, | |
| "learning_rate": 9.235393778375068e-05, | |
| "loss": 1.2612, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.5290842772125168, | |
| "grad_norm": 1.7242546081542969, | |
| "learning_rate": 9.211170258938747e-05, | |
| "loss": 1.374, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.5298521789210981, | |
| "grad_norm": 1.4706342220306396, | |
| "learning_rate": 9.18695139628282e-05, | |
| "loss": 1.264, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5306200806296794, | |
| "grad_norm": 1.520382046699524, | |
| "learning_rate": 9.162737333381019e-05, | |
| "loss": 1.4948, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.5313879823382607, | |
| "grad_norm": 1.598055362701416, | |
| "learning_rate": 9.138528213178727e-05, | |
| "loss": 1.4628, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.532155884046842, | |
| "grad_norm": 1.2860671281814575, | |
| "learning_rate": 9.11432417859216e-05, | |
| "loss": 1.0033, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.5329237857554233, | |
| "grad_norm": 1.4247864484786987, | |
| "learning_rate": 9.090125372507492e-05, | |
| "loss": 1.1321, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.5336916874640046, | |
| "grad_norm": 1.3131632804870605, | |
| "learning_rate": 9.065931937780059e-05, | |
| "loss": 0.9775, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.534459589172586, | |
| "grad_norm": 1.3943092823028564, | |
| "learning_rate": 9.041744017233462e-05, | |
| "loss": 1.1079, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.5352274908811672, | |
| "grad_norm": 1.3534505367279053, | |
| "learning_rate": 9.017561753658764e-05, | |
| "loss": 0.9166, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.5359953925897485, | |
| "grad_norm": 1.6207268238067627, | |
| "learning_rate": 8.993385289813627e-05, | |
| "loss": 1.1386, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.5367632942983298, | |
| "grad_norm": 1.4252740144729614, | |
| "learning_rate": 8.969214768421483e-05, | |
| "loss": 1.2191, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.5375311960069111, | |
| "grad_norm": 1.3587523698806763, | |
| "learning_rate": 8.945050332170672e-05, | |
| "loss": 1.1997, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5382990977154924, | |
| "grad_norm": 1.596205234527588, | |
| "learning_rate": 8.920892123713614e-05, | |
| "loss": 1.1552, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.5390669994240738, | |
| "grad_norm": 1.779982089996338, | |
| "learning_rate": 8.89674028566597e-05, | |
| "loss": 1.46, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.539834901132655, | |
| "grad_norm": 1.493944525718689, | |
| "learning_rate": 8.872594960605785e-05, | |
| "loss": 1.238, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.5406028028412363, | |
| "grad_norm": 1.5526620149612427, | |
| "learning_rate": 8.848456291072666e-05, | |
| "loss": 1.3413, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.5413707045498176, | |
| "grad_norm": 1.4706637859344482, | |
| "learning_rate": 8.82432441956692e-05, | |
| "loss": 1.2897, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.5421386062583989, | |
| "grad_norm": 1.2133674621582031, | |
| "learning_rate": 8.80019948854873e-05, | |
| "loss": 0.9593, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.5429065079669803, | |
| "grad_norm": 1.2873600721359253, | |
| "learning_rate": 8.776081640437294e-05, | |
| "loss": 1.118, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.5436744096755616, | |
| "grad_norm": 1.5350018739700317, | |
| "learning_rate": 8.751971017610012e-05, | |
| "loss": 1.3241, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.5444423113841428, | |
| "grad_norm": 1.4526116847991943, | |
| "learning_rate": 8.727867762401623e-05, | |
| "loss": 1.2734, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.5452102130927241, | |
| "grad_norm": 1.551858901977539, | |
| "learning_rate": 8.703772017103372e-05, | |
| "loss": 1.3445, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.5459781148013054, | |
| "grad_norm": 1.0594290494918823, | |
| "learning_rate": 8.679683923962174e-05, | |
| "loss": 0.891, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.5467460165098867, | |
| "grad_norm": 1.4673049449920654, | |
| "learning_rate": 8.655603625179759e-05, | |
| "loss": 1.3731, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.5475139182184681, | |
| "grad_norm": 1.2594002485275269, | |
| "learning_rate": 8.631531262911857e-05, | |
| "loss": 1.0889, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.5482818199270494, | |
| "grad_norm": 1.3199065923690796, | |
| "learning_rate": 8.607466979267338e-05, | |
| "loss": 1.1011, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.5490497216356306, | |
| "grad_norm": 1.752083659172058, | |
| "learning_rate": 8.583410916307386e-05, | |
| "loss": 1.4096, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.5498176233442119, | |
| "grad_norm": 1.4408888816833496, | |
| "learning_rate": 8.559363216044647e-05, | |
| "loss": 1.2785, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.5505855250527932, | |
| "grad_norm": 1.881710410118103, | |
| "learning_rate": 8.53532402044241e-05, | |
| "loss": 1.854, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.5513534267613746, | |
| "grad_norm": 1.3627890348434448, | |
| "learning_rate": 8.51129347141374e-05, | |
| "loss": 1.097, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.5521213284699559, | |
| "grad_norm": 1.4654157161712646, | |
| "learning_rate": 8.487271710820681e-05, | |
| "loss": 1.3027, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.5528892301785372, | |
| "grad_norm": 1.7268285751342773, | |
| "learning_rate": 8.463258880473373e-05, | |
| "loss": 1.3451, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5536571318871184, | |
| "grad_norm": 1.112960934638977, | |
| "learning_rate": 8.439255122129254e-05, | |
| "loss": 0.9304, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.5544250335956997, | |
| "grad_norm": 1.4320957660675049, | |
| "learning_rate": 8.415260577492195e-05, | |
| "loss": 1.2068, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.555192935304281, | |
| "grad_norm": 1.6056063175201416, | |
| "learning_rate": 8.391275388211684e-05, | |
| "loss": 1.1656, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.5559608370128624, | |
| "grad_norm": 1.415808081626892, | |
| "learning_rate": 8.367299695881973e-05, | |
| "loss": 1.3507, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.5567287387214437, | |
| "grad_norm": 1.5225753784179688, | |
| "learning_rate": 8.34333364204125e-05, | |
| "loss": 1.4994, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.557496640430025, | |
| "grad_norm": 1.4271262884140015, | |
| "learning_rate": 8.319377368170808e-05, | |
| "loss": 1.2742, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.5582645421386062, | |
| "grad_norm": 1.217391848564148, | |
| "learning_rate": 8.295431015694202e-05, | |
| "loss": 1.0471, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.5590324438471875, | |
| "grad_norm": 1.5793488025665283, | |
| "learning_rate": 8.271494725976418e-05, | |
| "loss": 1.3258, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.5598003455557689, | |
| "grad_norm": 1.4085965156555176, | |
| "learning_rate": 8.247568640323036e-05, | |
| "loss": 1.2015, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.5605682472643502, | |
| "grad_norm": 1.4623831510543823, | |
| "learning_rate": 8.223652899979402e-05, | |
| "loss": 1.1587, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5613361489729315, | |
| "grad_norm": 1.187821626663208, | |
| "learning_rate": 8.199747646129775e-05, | |
| "loss": 1.0068, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.5621040506815128, | |
| "grad_norm": 1.485092282295227, | |
| "learning_rate": 8.175853019896534e-05, | |
| "loss": 1.3343, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.562871952390094, | |
| "grad_norm": 1.3452789783477783, | |
| "learning_rate": 8.15196916233929e-05, | |
| "loss": 1.2402, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.5636398540986753, | |
| "grad_norm": 1.7040382623672485, | |
| "learning_rate": 8.128096214454105e-05, | |
| "loss": 1.1712, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.5644077558072567, | |
| "grad_norm": 1.5438884496688843, | |
| "learning_rate": 8.104234317172621e-05, | |
| "loss": 1.1095, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.565175657515838, | |
| "grad_norm": 1.173060417175293, | |
| "learning_rate": 8.080383611361254e-05, | |
| "loss": 0.9693, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.5659435592244193, | |
| "grad_norm": 1.4293849468231201, | |
| "learning_rate": 8.056544237820351e-05, | |
| "loss": 1.0731, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.5667114609330006, | |
| "grad_norm": 1.4654078483581543, | |
| "learning_rate": 8.03271633728335e-05, | |
| "loss": 1.4003, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.5674793626415818, | |
| "grad_norm": 1.4996401071548462, | |
| "learning_rate": 8.008900050415973e-05, | |
| "loss": 1.117, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.5682472643501632, | |
| "grad_norm": 1.5051236152648926, | |
| "learning_rate": 7.985095517815371e-05, | |
| "loss": 1.5012, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5690151660587445, | |
| "grad_norm": 1.4953557252883911, | |
| "learning_rate": 7.961302880009314e-05, | |
| "loss": 1.2279, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.5697830677673258, | |
| "grad_norm": 1.3291648626327515, | |
| "learning_rate": 7.937522277455343e-05, | |
| "loss": 1.1494, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.5705509694759071, | |
| "grad_norm": 1.496016025543213, | |
| "learning_rate": 7.913753850539964e-05, | |
| "loss": 1.2484, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.5713188711844884, | |
| "grad_norm": 1.4508578777313232, | |
| "learning_rate": 7.889997739577783e-05, | |
| "loss": 1.1512, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.5720867728930696, | |
| "grad_norm": 1.3858839273452759, | |
| "learning_rate": 7.866254084810724e-05, | |
| "loss": 1.3323, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.572854674601651, | |
| "grad_norm": 1.4042750597000122, | |
| "learning_rate": 7.842523026407159e-05, | |
| "loss": 0.9763, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.5736225763102323, | |
| "grad_norm": 1.4263511896133423, | |
| "learning_rate": 7.818804704461108e-05, | |
| "loss": 1.1, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.5743904780188136, | |
| "grad_norm": 1.4217872619628906, | |
| "learning_rate": 7.795099258991404e-05, | |
| "loss": 1.001, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.5751583797273949, | |
| "grad_norm": 1.5163894891738892, | |
| "learning_rate": 7.771406829940852e-05, | |
| "loss": 1.1633, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.5759262814359762, | |
| "grad_norm": 1.423274278640747, | |
| "learning_rate": 7.747727557175434e-05, | |
| "loss": 1.4725, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5766941831445574, | |
| "grad_norm": 1.5022209882736206, | |
| "learning_rate": 7.724061580483449e-05, | |
| "loss": 1.1757, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.5774620848531388, | |
| "grad_norm": 1.3035588264465332, | |
| "learning_rate": 7.700409039574717e-05, | |
| "loss": 1.0345, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.5782299865617201, | |
| "grad_norm": 1.6852291822433472, | |
| "learning_rate": 7.676770074079732e-05, | |
| "loss": 1.1958, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.5789978882703014, | |
| "grad_norm": 1.4031083583831787, | |
| "learning_rate": 7.653144823548852e-05, | |
| "loss": 1.3684, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.5797657899788827, | |
| "grad_norm": 1.4951286315917969, | |
| "learning_rate": 7.62953342745146e-05, | |
| "loss": 1.4351, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.580533691687464, | |
| "grad_norm": 1.273272156715393, | |
| "learning_rate": 7.605936025175174e-05, | |
| "loss": 1.1404, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.5813015933960453, | |
| "grad_norm": 1.4025629758834839, | |
| "learning_rate": 7.582352756024971e-05, | |
| "loss": 1.4564, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.5820694951046266, | |
| "grad_norm": 1.344117283821106, | |
| "learning_rate": 7.558783759222417e-05, | |
| "loss": 1.1624, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.5828373968132079, | |
| "grad_norm": 1.3355436325073242, | |
| "learning_rate": 7.535229173904811e-05, | |
| "loss": 1.043, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.5836052985217892, | |
| "grad_norm": 1.2501863241195679, | |
| "learning_rate": 7.511689139124382e-05, | |
| "loss": 1.2641, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5843732002303705, | |
| "grad_norm": 1.3736534118652344, | |
| "learning_rate": 7.488163793847458e-05, | |
| "loss": 1.1726, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.5851411019389519, | |
| "grad_norm": 1.429854154586792, | |
| "learning_rate": 7.464653276953643e-05, | |
| "loss": 1.2585, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.5859090036475331, | |
| "grad_norm": 1.7089576721191406, | |
| "learning_rate": 7.441157727235015e-05, | |
| "loss": 1.5036, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.5866769053561144, | |
| "grad_norm": 1.3912054300308228, | |
| "learning_rate": 7.417677283395284e-05, | |
| "loss": 1.0993, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.5874448070646957, | |
| "grad_norm": 1.7586991786956787, | |
| "learning_rate": 7.394212084048995e-05, | |
| "loss": 1.2544, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.588212708773277, | |
| "grad_norm": 1.2912665605545044, | |
| "learning_rate": 7.370762267720685e-05, | |
| "loss": 1.0896, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.5889806104818583, | |
| "grad_norm": 1.5114494562149048, | |
| "learning_rate": 7.347327972844096e-05, | |
| "loss": 1.1424, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.5897485121904397, | |
| "grad_norm": 1.5202648639678955, | |
| "learning_rate": 7.323909337761317e-05, | |
| "loss": 1.2709, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.5905164138990209, | |
| "grad_norm": 1.5682910680770874, | |
| "learning_rate": 7.30050650072202e-05, | |
| "loss": 1.5064, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.5912843156076022, | |
| "grad_norm": 1.7039157152175903, | |
| "learning_rate": 7.277119599882586e-05, | |
| "loss": 1.1595, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.5920522173161835, | |
| "grad_norm": 6.636146068572998, | |
| "learning_rate": 7.25374877330534e-05, | |
| "loss": 0.9431, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.5928201190247648, | |
| "grad_norm": 1.5151164531707764, | |
| "learning_rate": 7.230394158957705e-05, | |
| "loss": 1.2479, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.5935880207333462, | |
| "grad_norm": 1.6289341449737549, | |
| "learning_rate": 7.20705589471139e-05, | |
| "loss": 1.2524, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.5943559224419275, | |
| "grad_norm": 1.5332694053649902, | |
| "learning_rate": 7.1837341183416e-05, | |
| "loss": 1.4389, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.5951238241505087, | |
| "grad_norm": 1.4293265342712402, | |
| "learning_rate": 7.160428967526187e-05, | |
| "loss": 1.1761, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.59589172585909, | |
| "grad_norm": 1.1657414436340332, | |
| "learning_rate": 7.137140579844871e-05, | |
| "loss": 0.9015, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.5966596275676713, | |
| "grad_norm": 1.196415662765503, | |
| "learning_rate": 7.1138690927784e-05, | |
| "loss": 1.0494, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.5974275292762526, | |
| "grad_norm": 1.522498369216919, | |
| "learning_rate": 7.090614643707762e-05, | |
| "loss": 1.2778, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.598195430984834, | |
| "grad_norm": 1.7265686988830566, | |
| "learning_rate": 7.067377369913352e-05, | |
| "loss": 1.3857, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.5989633326934153, | |
| "grad_norm": 1.781242847442627, | |
| "learning_rate": 7.044157408574185e-05, | |
| "loss": 1.8548, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5997312344019965, | |
| "grad_norm": 1.2553101778030396, | |
| "learning_rate": 7.020954896767058e-05, | |
| "loss": 1.0755, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.6004991361105778, | |
| "grad_norm": 1.6484979391098022, | |
| "learning_rate": 6.997769971465769e-05, | |
| "loss": 1.3693, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.6012670378191591, | |
| "grad_norm": 2.0107421875, | |
| "learning_rate": 6.974602769540289e-05, | |
| "loss": 1.3331, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.6020349395277405, | |
| "grad_norm": 1.945434331893921, | |
| "learning_rate": 6.951453427755968e-05, | |
| "loss": 1.468, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.6028028412363218, | |
| "grad_norm": 1.637454628944397, | |
| "learning_rate": 6.928322082772712e-05, | |
| "loss": 1.3608, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.6035707429449031, | |
| "grad_norm": 1.5328805446624756, | |
| "learning_rate": 6.905208871144187e-05, | |
| "loss": 1.2353, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.6043386446534843, | |
| "grad_norm": 1.452931523323059, | |
| "learning_rate": 6.882113929317015e-05, | |
| "loss": 1.1362, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.6051065463620656, | |
| "grad_norm": 1.3005062341690063, | |
| "learning_rate": 6.859037393629957e-05, | |
| "loss": 1.0392, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.6058744480706469, | |
| "grad_norm": 1.3638523817062378, | |
| "learning_rate": 6.835979400313122e-05, | |
| "loss": 1.2327, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.6066423497792283, | |
| "grad_norm": 1.7218124866485596, | |
| "learning_rate": 6.81294008548715e-05, | |
| "loss": 1.3321, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6074102514878096, | |
| "grad_norm": 1.0565749406814575, | |
| "learning_rate": 6.789919585162423e-05, | |
| "loss": 0.9172, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.6081781531963909, | |
| "grad_norm": 1.1628111600875854, | |
| "learning_rate": 6.766918035238237e-05, | |
| "loss": 1.0269, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.6089460549049721, | |
| "grad_norm": 1.5882484912872314, | |
| "learning_rate": 6.743935571502038e-05, | |
| "loss": 1.281, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.6097139566135534, | |
| "grad_norm": 1.0573451519012451, | |
| "learning_rate": 6.720972329628577e-05, | |
| "loss": 0.681, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.6104818583221348, | |
| "grad_norm": 1.6153432130813599, | |
| "learning_rate": 6.698028445179148e-05, | |
| "loss": 1.3266, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.6112497600307161, | |
| "grad_norm": 1.3120388984680176, | |
| "learning_rate": 6.675104053600763e-05, | |
| "loss": 1.101, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.6120176617392974, | |
| "grad_norm": 1.544906735420227, | |
| "learning_rate": 6.65219929022535e-05, | |
| "loss": 1.2774, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.6127855634478787, | |
| "grad_norm": 1.5639188289642334, | |
| "learning_rate": 6.629314290268987e-05, | |
| "loss": 1.301, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.6135534651564599, | |
| "grad_norm": 1.116890549659729, | |
| "learning_rate": 6.606449188831057e-05, | |
| "loss": 1.0978, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.6143213668650412, | |
| "grad_norm": 1.459924578666687, | |
| "learning_rate": 6.583604120893488e-05, | |
| "loss": 1.1193, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6150892685736226, | |
| "grad_norm": 1.2557390928268433, | |
| "learning_rate": 6.560779221319938e-05, | |
| "loss": 1.0979, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.6158571702822039, | |
| "grad_norm": 1.281129240989685, | |
| "learning_rate": 6.537974624855003e-05, | |
| "loss": 1.3004, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.6166250719907852, | |
| "grad_norm": 1.4236611127853394, | |
| "learning_rate": 6.51519046612342e-05, | |
| "loss": 1.1254, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.6173929736993665, | |
| "grad_norm": 1.4232025146484375, | |
| "learning_rate": 6.492426879629282e-05, | |
| "loss": 1.0886, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.6181608754079477, | |
| "grad_norm": 1.569633960723877, | |
| "learning_rate": 6.46968399975522e-05, | |
| "loss": 1.1388, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.618928777116529, | |
| "grad_norm": 1.7293319702148438, | |
| "learning_rate": 6.446961960761643e-05, | |
| "loss": 1.4027, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.6196966788251104, | |
| "grad_norm": 1.5421292781829834, | |
| "learning_rate": 6.424260896785914e-05, | |
| "loss": 1.2518, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.6204645805336917, | |
| "grad_norm": 1.5703529119491577, | |
| "learning_rate": 6.40158094184158e-05, | |
| "loss": 1.2429, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.621232482242273, | |
| "grad_norm": 1.575032353401184, | |
| "learning_rate": 6.378922229817575e-05, | |
| "loss": 1.3882, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.6220003839508543, | |
| "grad_norm": 1.1647250652313232, | |
| "learning_rate": 6.356284894477412e-05, | |
| "loss": 1.031, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6227682856594355, | |
| "grad_norm": 1.8725273609161377, | |
| "learning_rate": 6.333669069458432e-05, | |
| "loss": 1.4888, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.6235361873680169, | |
| "grad_norm": 1.4924973249435425, | |
| "learning_rate": 6.311074888270971e-05, | |
| "loss": 1.3342, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.6243040890765982, | |
| "grad_norm": 1.2791497707366943, | |
| "learning_rate": 6.288502484297607e-05, | |
| "loss": 1.1041, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.6250719907851795, | |
| "grad_norm": 1.1811076402664185, | |
| "learning_rate": 6.265951990792347e-05, | |
| "loss": 0.838, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.6258398924937608, | |
| "grad_norm": 1.6005576848983765, | |
| "learning_rate": 6.243423540879865e-05, | |
| "loss": 1.3403, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.6266077942023421, | |
| "grad_norm": 1.2689980268478394, | |
| "learning_rate": 6.220917267554686e-05, | |
| "loss": 1.0595, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.6273756959109233, | |
| "grad_norm": 1.4566655158996582, | |
| "learning_rate": 6.198433303680439e-05, | |
| "loss": 1.2665, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.6281435976195047, | |
| "grad_norm": 1.9303637742996216, | |
| "learning_rate": 6.175971781989025e-05, | |
| "loss": 0.9127, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.628911499328086, | |
| "grad_norm": 1.7557438611984253, | |
| "learning_rate": 6.153532835079886e-05, | |
| "loss": 1.4541, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.6296794010366673, | |
| "grad_norm": 1.5808215141296387, | |
| "learning_rate": 6.131116595419178e-05, | |
| "loss": 1.4305, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6304473027452486, | |
| "grad_norm": 1.6107852458953857, | |
| "learning_rate": 6.108723195339011e-05, | |
| "loss": 1.1872, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.63121520445383, | |
| "grad_norm": 1.4639952182769775, | |
| "learning_rate": 6.086352767036673e-05, | |
| "loss": 0.8872, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.6319831061624112, | |
| "grad_norm": 1.5953764915466309, | |
| "learning_rate": 6.064005442573824e-05, | |
| "loss": 1.3455, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.6327510078709925, | |
| "grad_norm": 1.7272642850875854, | |
| "learning_rate": 6.041681353875746e-05, | |
| "loss": 1.273, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.6335189095795738, | |
| "grad_norm": 1.5354080200195312, | |
| "learning_rate": 6.019380632730546e-05, | |
| "loss": 1.1075, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.6342868112881551, | |
| "grad_norm": 1.4213950634002686, | |
| "learning_rate": 5.997103410788385e-05, | |
| "loss": 1.2971, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.6350547129967364, | |
| "grad_norm": 1.8925020694732666, | |
| "learning_rate": 5.9748498195606925e-05, | |
| "loss": 1.3272, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.6358226147053178, | |
| "grad_norm": 1.3079675436019897, | |
| "learning_rate": 5.952619990419408e-05, | |
| "loss": 1.1373, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.636590516413899, | |
| "grad_norm": 1.5753669738769531, | |
| "learning_rate": 5.9304140545961784e-05, | |
| "loss": 1.0857, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.6373584181224803, | |
| "grad_norm": 1.3841052055358887, | |
| "learning_rate": 5.9082321431816156e-05, | |
| "loss": 1.2683, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6381263198310616, | |
| "grad_norm": 1.3311750888824463, | |
| "learning_rate": 5.88607438712449e-05, | |
| "loss": 0.9628, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.6388942215396429, | |
| "grad_norm": 1.4135347604751587, | |
| "learning_rate": 5.863940917230986e-05, | |
| "loss": 1.0973, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.6396621232482242, | |
| "grad_norm": 1.665837049484253, | |
| "learning_rate": 5.841831864163909e-05, | |
| "loss": 1.5469, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.6404300249568056, | |
| "grad_norm": 1.3354551792144775, | |
| "learning_rate": 5.8197473584419184e-05, | |
| "loss": 1.1408, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.6411979266653869, | |
| "grad_norm": 1.3838506937026978, | |
| "learning_rate": 5.7976875304387756e-05, | |
| "loss": 1.0322, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.6419658283739681, | |
| "grad_norm": 1.6376174688339233, | |
| "learning_rate": 5.7756525103825474e-05, | |
| "loss": 1.4984, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.6427337300825494, | |
| "grad_norm": 1.2751094102859497, | |
| "learning_rate": 5.753642428354852e-05, | |
| "loss": 1.2019, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.6435016317911307, | |
| "grad_norm": 1.3204684257507324, | |
| "learning_rate": 5.731657414290085e-05, | |
| "loss": 1.1769, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.644269533499712, | |
| "grad_norm": 1.444831132888794, | |
| "learning_rate": 5.7096975979746704e-05, | |
| "loss": 1.2601, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.6450374352082934, | |
| "grad_norm": 1.5533994436264038, | |
| "learning_rate": 5.687763109046255e-05, | |
| "loss": 1.2037, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6458053369168747, | |
| "grad_norm": 1.3148772716522217, | |
| "learning_rate": 5.665854076992991e-05, | |
| "loss": 1.0888, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.6465732386254559, | |
| "grad_norm": 1.465082049369812, | |
| "learning_rate": 5.643970631152735e-05, | |
| "loss": 1.2246, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.6473411403340372, | |
| "grad_norm": 1.5530979633331299, | |
| "learning_rate": 5.622112900712304e-05, | |
| "loss": 1.4689, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.6481090420426185, | |
| "grad_norm": 1.1694343090057373, | |
| "learning_rate": 5.600281014706703e-05, | |
| "loss": 0.9216, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.6488769437511999, | |
| "grad_norm": 1.592970371246338, | |
| "learning_rate": 5.57847510201837e-05, | |
| "loss": 1.1406, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.6496448454597812, | |
| "grad_norm": 1.7581945657730103, | |
| "learning_rate": 5.556695291376406e-05, | |
| "loss": 1.4523, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.6504127471683625, | |
| "grad_norm": 1.7311755418777466, | |
| "learning_rate": 5.5349417113558254e-05, | |
| "loss": 1.3606, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.6511806488769437, | |
| "grad_norm": 1.291117548942566, | |
| "learning_rate": 5.5132144903768e-05, | |
| "loss": 1.0444, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.651948550585525, | |
| "grad_norm": 1.4802870750427246, | |
| "learning_rate": 5.491513756703881e-05, | |
| "loss": 1.1835, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.6527164522941064, | |
| "grad_norm": 1.0915478467941284, | |
| "learning_rate": 5.46983963844526e-05, | |
| "loss": 0.9936, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6534843540026877, | |
| "grad_norm": 1.842965006828308, | |
| "learning_rate": 5.448192263552006e-05, | |
| "loss": 1.2762, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.654252255711269, | |
| "grad_norm": 1.5491669178009033, | |
| "learning_rate": 5.426571759817314e-05, | |
| "loss": 1.2144, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.6550201574198503, | |
| "grad_norm": 1.6982448101043701, | |
| "learning_rate": 5.4049782548757386e-05, | |
| "loss": 1.0843, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.6557880591284315, | |
| "grad_norm": 1.5886989831924438, | |
| "learning_rate": 5.383411876202464e-05, | |
| "loss": 1.393, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.6565559608370128, | |
| "grad_norm": 1.3143181800842285, | |
| "learning_rate": 5.36187275111253e-05, | |
| "loss": 1.1188, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.6573238625455942, | |
| "grad_norm": 1.499030590057373, | |
| "learning_rate": 5.340361006760082e-05, | |
| "loss": 1.1549, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.6580917642541755, | |
| "grad_norm": 1.2520569562911987, | |
| "learning_rate": 5.318876770137634e-05, | |
| "loss": 1.1459, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.6588596659627568, | |
| "grad_norm": 1.5099189281463623, | |
| "learning_rate": 5.297420168075307e-05, | |
| "loss": 1.2704, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.6596275676713381, | |
| "grad_norm": 1.6787750720977783, | |
| "learning_rate": 5.275991327240082e-05, | |
| "loss": 1.2595, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.6603954693799193, | |
| "grad_norm": 1.695313811302185, | |
| "learning_rate": 5.254590374135058e-05, | |
| "loss": 1.3029, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6611633710885007, | |
| "grad_norm": 1.2656611204147339, | |
| "learning_rate": 5.233217435098707e-05, | |
| "loss": 1.2773, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.661931272797082, | |
| "grad_norm": 1.3123670816421509, | |
| "learning_rate": 5.2118726363041036e-05, | |
| "loss": 1.653, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.6626991745056633, | |
| "grad_norm": 1.6967337131500244, | |
| "learning_rate": 5.190556103758223e-05, | |
| "loss": 1.356, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.6634670762142446, | |
| "grad_norm": 1.0755648612976074, | |
| "learning_rate": 5.1692679633011564e-05, | |
| "loss": 0.8091, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.6642349779228259, | |
| "grad_norm": 1.0523223876953125, | |
| "learning_rate": 5.148008340605392e-05, | |
| "loss": 0.854, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.6650028796314071, | |
| "grad_norm": 1.3313195705413818, | |
| "learning_rate": 5.1267773611750624e-05, | |
| "loss": 1.0723, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.6657707813399885, | |
| "grad_norm": 1.5703046321868896, | |
| "learning_rate": 5.1055751503452144e-05, | |
| "loss": 1.3341, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.6665386830485698, | |
| "grad_norm": 1.6033220291137695, | |
| "learning_rate": 5.0844018332810594e-05, | |
| "loss": 1.4145, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.6673065847571511, | |
| "grad_norm": 1.1940321922302246, | |
| "learning_rate": 5.0632575349772225e-05, | |
| "loss": 1.0971, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.6680744864657324, | |
| "grad_norm": 1.534477710723877, | |
| "learning_rate": 5.0421423802570454e-05, | |
| "loss": 1.2956, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6688423881743137, | |
| "grad_norm": 1.457879900932312, | |
| "learning_rate": 5.0210564937718055e-05, | |
| "loss": 1.3399, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.669610289882895, | |
| "grad_norm": 1.6395093202590942, | |
| "learning_rate": 5.000000000000002e-05, | |
| "loss": 1.2267, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.6703781915914763, | |
| "grad_norm": 1.403594970703125, | |
| "learning_rate": 4.978973023246616e-05, | |
| "loss": 1.2561, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.6711460933000576, | |
| "grad_norm": 1.3514751195907593, | |
| "learning_rate": 4.957975687642389e-05, | |
| "loss": 0.8501, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.6719139950086389, | |
| "grad_norm": 1.7166774272918701, | |
| "learning_rate": 4.937008117143055e-05, | |
| "loss": 1.2769, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.6726818967172202, | |
| "grad_norm": 1.637560486793518, | |
| "learning_rate": 4.9160704355286577e-05, | |
| "loss": 1.2707, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.6734497984258015, | |
| "grad_norm": 1.7437028884887695, | |
| "learning_rate": 4.895162766402781e-05, | |
| "loss": 1.356, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.6742177001343828, | |
| "grad_norm": 1.4267345666885376, | |
| "learning_rate": 4.8742852331918364e-05, | |
| "loss": 1.2037, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.6749856018429641, | |
| "grad_norm": 2.0313918590545654, | |
| "learning_rate": 4.8534379591443246e-05, | |
| "loss": 1.5035, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.6757535035515454, | |
| "grad_norm": 1.4802058935165405, | |
| "learning_rate": 4.8326210673301284e-05, | |
| "loss": 1.1481, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6765214052601267, | |
| "grad_norm": 1.6689741611480713, | |
| "learning_rate": 4.811834680639765e-05, | |
| "loss": 1.2703, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.677289306968708, | |
| "grad_norm": 1.475077748298645, | |
| "learning_rate": 4.791078921783653e-05, | |
| "loss": 1.0892, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.6780572086772894, | |
| "grad_norm": 1.7580574750900269, | |
| "learning_rate": 4.770353913291428e-05, | |
| "loss": 1.2977, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.6788251103858706, | |
| "grad_norm": 1.482071876525879, | |
| "learning_rate": 4.749659777511177e-05, | |
| "loss": 1.2857, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.6795930120944519, | |
| "grad_norm": 1.2742493152618408, | |
| "learning_rate": 4.728996636608738e-05, | |
| "loss": 1.0448, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.6803609138030332, | |
| "grad_norm": 1.8095487356185913, | |
| "learning_rate": 4.708364612566969e-05, | |
| "loss": 1.4402, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.6811288155116145, | |
| "grad_norm": 1.4991480112075806, | |
| "learning_rate": 4.6877638271850485e-05, | |
| "loss": 1.0895, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.6818967172201958, | |
| "grad_norm": 1.588757038116455, | |
| "learning_rate": 4.667194402077714e-05, | |
| "loss": 1.472, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.6826646189287772, | |
| "grad_norm": 1.5407018661499023, | |
| "learning_rate": 4.646656458674595e-05, | |
| "loss": 1.0996, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.6834325206373584, | |
| "grad_norm": 1.4947824478149414, | |
| "learning_rate": 4.62615011821946e-05, | |
| "loss": 1.1829, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6842004223459397, | |
| "grad_norm": 1.6118358373641968, | |
| "learning_rate": 4.6056755017695155e-05, | |
| "loss": 1.26, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.684968324054521, | |
| "grad_norm": 1.4456740617752075, | |
| "learning_rate": 4.585232730194682e-05, | |
| "loss": 1.1356, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.6857362257631023, | |
| "grad_norm": 1.421918272972107, | |
| "learning_rate": 4.5648219241769054e-05, | |
| "loss": 1.1096, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.6865041274716837, | |
| "grad_norm": 1.182092547416687, | |
| "learning_rate": 4.5444432042093996e-05, | |
| "loss": 0.9751, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.687272029180265, | |
| "grad_norm": 1.425451636314392, | |
| "learning_rate": 4.524096690595978e-05, | |
| "loss": 1.3076, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.6880399308888462, | |
| "grad_norm": 1.116645097732544, | |
| "learning_rate": 4.5037825034503304e-05, | |
| "loss": 0.7194, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.6888078325974275, | |
| "grad_norm": 1.3570024967193604, | |
| "learning_rate": 4.4835007626953e-05, | |
| "loss": 0.9747, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.6895757343060088, | |
| "grad_norm": 1.3995678424835205, | |
| "learning_rate": 4.4632515880621894e-05, | |
| "loss": 1.0773, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.6903436360145901, | |
| "grad_norm": 1.5802603960037231, | |
| "learning_rate": 4.443035099090048e-05, | |
| "loss": 1.3399, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.6911115377231715, | |
| "grad_norm": 1.563367247581482, | |
| "learning_rate": 4.4228514151249825e-05, | |
| "loss": 1.0937, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6918794394317528, | |
| "grad_norm": 1.6263221502304077, | |
| "learning_rate": 4.4027006553194115e-05, | |
| "loss": 1.258, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.692647341140334, | |
| "grad_norm": 1.682242751121521, | |
| "learning_rate": 4.3825829386314166e-05, | |
| "loss": 1.3231, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.6934152428489153, | |
| "grad_norm": 1.3335479497909546, | |
| "learning_rate": 4.362498383823996e-05, | |
| "loss": 1.0193, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.6941831445574966, | |
| "grad_norm": 1.619031548500061, | |
| "learning_rate": 4.342447109464385e-05, | |
| "loss": 1.2696, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.694951046266078, | |
| "grad_norm": 1.4861383438110352, | |
| "learning_rate": 4.322429233923351e-05, | |
| "loss": 1.3105, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.6957189479746593, | |
| "grad_norm": 1.4006625413894653, | |
| "learning_rate": 4.3024448753744925e-05, | |
| "loss": 1.3573, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.6964868496832406, | |
| "grad_norm": 1.2410006523132324, | |
| "learning_rate": 4.282494151793548e-05, | |
| "loss": 0.9566, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.6972547513918218, | |
| "grad_norm": 1.0770704746246338, | |
| "learning_rate": 4.2625771809576874e-05, | |
| "loss": 0.7789, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.6980226531004031, | |
| "grad_norm": 1.2837942838668823, | |
| "learning_rate": 4.242694080444837e-05, | |
| "loss": 1.1047, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.6987905548089844, | |
| "grad_norm": 1.1040459871292114, | |
| "learning_rate": 4.2228449676329616e-05, | |
| "loss": 0.9763, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.6995584565175658, | |
| "grad_norm": 1.6829538345336914, | |
| "learning_rate": 4.2030299596993883e-05, | |
| "loss": 1.4803, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.7003263582261471, | |
| "grad_norm": 1.4828580617904663, | |
| "learning_rate": 4.1832491736201077e-05, | |
| "loss": 1.2561, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.7010942599347284, | |
| "grad_norm": 1.568756341934204, | |
| "learning_rate": 4.1635027261690827e-05, | |
| "loss": 1.4303, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.7018621616433096, | |
| "grad_norm": 1.3178819417953491, | |
| "learning_rate": 4.143790733917564e-05, | |
| "loss": 0.8952, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.7026300633518909, | |
| "grad_norm": 1.4817698001861572, | |
| "learning_rate": 4.124113313233404e-05, | |
| "loss": 1.3478, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.7033979650604723, | |
| "grad_norm": 1.4555271863937378, | |
| "learning_rate": 4.1044705802803574e-05, | |
| "loss": 1.3088, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.7041658667690536, | |
| "grad_norm": 1.4719526767730713, | |
| "learning_rate": 4.084862651017406e-05, | |
| "loss": 1.2125, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.7049337684776349, | |
| "grad_norm": 1.2816683053970337, | |
| "learning_rate": 4.065289641198073e-05, | |
| "loss": 1.208, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.7057016701862162, | |
| "grad_norm": 1.3525464534759521, | |
| "learning_rate": 4.045751666369736e-05, | |
| "loss": 1.1444, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.7064695718947974, | |
| "grad_norm": 1.4733887910842896, | |
| "learning_rate": 4.026248841872946e-05, | |
| "loss": 1.2619, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7072374736033787, | |
| "grad_norm": 1.3341400623321533, | |
| "learning_rate": 4.006781282840748e-05, | |
| "loss": 1.153, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.7080053753119601, | |
| "grad_norm": 1.5136420726776123, | |
| "learning_rate": 3.987349104198007e-05, | |
| "loss": 1.1891, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.7087732770205414, | |
| "grad_norm": 1.2748438119888306, | |
| "learning_rate": 3.9679524206607156e-05, | |
| "loss": 1.0631, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.7095411787291227, | |
| "grad_norm": 1.4721794128417969, | |
| "learning_rate": 3.948591346735325e-05, | |
| "loss": 1.2841, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.710309080437704, | |
| "grad_norm": 1.9709136486053467, | |
| "learning_rate": 3.929265996718072e-05, | |
| "loss": 1.478, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.7110769821462852, | |
| "grad_norm": 2.1271793842315674, | |
| "learning_rate": 3.9099764846943e-05, | |
| "loss": 1.1028, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.7118448838548666, | |
| "grad_norm": 1.5257985591888428, | |
| "learning_rate": 3.89072292453778e-05, | |
| "loss": 0.9435, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.7126127855634479, | |
| "grad_norm": 1.2246668338775635, | |
| "learning_rate": 3.871505429910057e-05, | |
| "loss": 0.9961, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.7133806872720292, | |
| "grad_norm": 1.3091059923171997, | |
| "learning_rate": 3.85232411425976e-05, | |
| "loss": 1.1994, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.7141485889806105, | |
| "grad_norm": 1.6306805610656738, | |
| "learning_rate": 3.833179090821929e-05, | |
| "loss": 1.4055, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7149164906891918, | |
| "grad_norm": 1.6316722631454468, | |
| "learning_rate": 3.814070472617375e-05, | |
| "loss": 1.4363, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.715684392397773, | |
| "grad_norm": 1.474591612815857, | |
| "learning_rate": 3.794998372451981e-05, | |
| "loss": 1.1251, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.7164522941063544, | |
| "grad_norm": 1.4235069751739502, | |
| "learning_rate": 3.775962902916056e-05, | |
| "loss": 1.0631, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.7172201958149357, | |
| "grad_norm": 1.580439567565918, | |
| "learning_rate": 3.756964176383655e-05, | |
| "loss": 1.3518, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.717988097523517, | |
| "grad_norm": 1.3070342540740967, | |
| "learning_rate": 3.7380023050119415e-05, | |
| "loss": 1.2074, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.7187559992320983, | |
| "grad_norm": 1.4719895124435425, | |
| "learning_rate": 3.7190774007404835e-05, | |
| "loss": 1.0298, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.7195239009406796, | |
| "grad_norm": 1.440446376800537, | |
| "learning_rate": 3.700189575290641e-05, | |
| "loss": 1.1669, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.7202918026492608, | |
| "grad_norm": 1.3623183965682983, | |
| "learning_rate": 3.681338940164868e-05, | |
| "loss": 1.1463, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.7210597043578422, | |
| "grad_norm": 1.303728461265564, | |
| "learning_rate": 3.6625256066460735e-05, | |
| "loss": 1.0379, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.7218276060664235, | |
| "grad_norm": 1.3376387357711792, | |
| "learning_rate": 3.6437496857969566e-05, | |
| "loss": 1.2042, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7225955077750048, | |
| "grad_norm": 1.6355212926864624, | |
| "learning_rate": 3.625011288459365e-05, | |
| "loss": 1.3353, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.7233634094835861, | |
| "grad_norm": 1.3309329748153687, | |
| "learning_rate": 3.606310525253621e-05, | |
| "loss": 0.765, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.7241313111921674, | |
| "grad_norm": 1.7754614353179932, | |
| "learning_rate": 3.5876475065778715e-05, | |
| "loss": 1.1589, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.7248992129007487, | |
| "grad_norm": 1.2315130233764648, | |
| "learning_rate": 3.5690223426074576e-05, | |
| "loss": 0.9012, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.72566711460933, | |
| "grad_norm": 1.3348158597946167, | |
| "learning_rate": 3.550435143294238e-05, | |
| "loss": 1.0467, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.7264350163179113, | |
| "grad_norm": 1.4577149152755737, | |
| "learning_rate": 3.531886018365954e-05, | |
| "loss": 1.0583, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.7272029180264926, | |
| "grad_norm": 1.927155613899231, | |
| "learning_rate": 3.513375077325575e-05, | |
| "loss": 1.2384, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.7279708197350739, | |
| "grad_norm": 2.0631394386291504, | |
| "learning_rate": 3.4949024294506674e-05, | |
| "loss": 1.3989, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.7287387214436553, | |
| "grad_norm": 1.7409682273864746, | |
| "learning_rate": 3.476468183792716e-05, | |
| "loss": 1.3022, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.7295066231522365, | |
| "grad_norm": 1.0105476379394531, | |
| "learning_rate": 3.458072449176525e-05, | |
| "loss": 0.902, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7302745248608178, | |
| "grad_norm": 1.6116336584091187, | |
| "learning_rate": 3.439715334199538e-05, | |
| "loss": 1.2317, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.7310424265693991, | |
| "grad_norm": 1.6682941913604736, | |
| "learning_rate": 3.4213969472312154e-05, | |
| "loss": 1.4236, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.7318103282779804, | |
| "grad_norm": 1.2750146389007568, | |
| "learning_rate": 3.403117396412391e-05, | |
| "loss": 0.8671, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.7325782299865617, | |
| "grad_norm": 1.2994352579116821, | |
| "learning_rate": 3.384876789654631e-05, | |
| "loss": 1.1212, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.7333461316951431, | |
| "grad_norm": 2.186424732208252, | |
| "learning_rate": 3.366675234639601e-05, | |
| "loss": 1.4304, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.7341140334037243, | |
| "grad_norm": 1.4062567949295044, | |
| "learning_rate": 3.348512838818425e-05, | |
| "loss": 1.1381, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.7348819351123056, | |
| "grad_norm": 1.4242571592330933, | |
| "learning_rate": 3.3303897094110636e-05, | |
| "loss": 1.0323, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.7356498368208869, | |
| "grad_norm": 1.2858667373657227, | |
| "learning_rate": 3.3123059534056634e-05, | |
| "loss": 1.0643, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.7364177385294682, | |
| "grad_norm": 1.6064081192016602, | |
| "learning_rate": 3.294261677557935e-05, | |
| "loss": 1.2303, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.7371856402380496, | |
| "grad_norm": 1.5934802293777466, | |
| "learning_rate": 3.2762569883905205e-05, | |
| "loss": 1.2093, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7379535419466309, | |
| "grad_norm": 1.4172946214675903, | |
| "learning_rate": 3.258291992192377e-05, | |
| "loss": 1.2843, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.7387214436552121, | |
| "grad_norm": 1.6268061399459839, | |
| "learning_rate": 3.240366795018117e-05, | |
| "loss": 1.4062, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.7394893453637934, | |
| "grad_norm": 1.4560757875442505, | |
| "learning_rate": 3.222481502687425e-05, | |
| "loss": 1.3195, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.7402572470723747, | |
| "grad_norm": 1.716341257095337, | |
| "learning_rate": 3.2046362207844e-05, | |
| "loss": 1.3937, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.741025148780956, | |
| "grad_norm": 1.4211591482162476, | |
| "learning_rate": 3.1868310546569424e-05, | |
| "loss": 1.3429, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.7417930504895374, | |
| "grad_norm": 1.4726101160049438, | |
| "learning_rate": 3.1690661094161364e-05, | |
| "loss": 1.2199, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.7425609521981187, | |
| "grad_norm": 1.6528246402740479, | |
| "learning_rate": 3.151341489935627e-05, | |
| "loss": 1.4508, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.7433288539066999, | |
| "grad_norm": 1.4311929941177368, | |
| "learning_rate": 3.133657300850995e-05, | |
| "loss": 1.3215, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.7440967556152812, | |
| "grad_norm": 1.6301379203796387, | |
| "learning_rate": 3.116013646559146e-05, | |
| "loss": 1.1288, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.7448646573238625, | |
| "grad_norm": 1.50910222530365, | |
| "learning_rate": 3.0984106312177e-05, | |
| "loss": 1.139, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7456325590324439, | |
| "grad_norm": 1.4490171670913696, | |
| "learning_rate": 3.0808483587443595e-05, | |
| "loss": 1.1863, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.7464004607410252, | |
| "grad_norm": 1.835550308227539, | |
| "learning_rate": 3.063326932816307e-05, | |
| "loss": 1.2192, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.7471683624496065, | |
| "grad_norm": 1.5930982828140259, | |
| "learning_rate": 3.045846456869592e-05, | |
| "loss": 1.148, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.7479362641581877, | |
| "grad_norm": 1.7530690431594849, | |
| "learning_rate": 3.0284070340985295e-05, | |
| "loss": 1.2803, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.748704165866769, | |
| "grad_norm": 1.397155523300171, | |
| "learning_rate": 3.011008767455059e-05, | |
| "loss": 1.2601, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.7494720675753503, | |
| "grad_norm": 1.3860681056976318, | |
| "learning_rate": 2.9936517596481818e-05, | |
| "loss": 1.314, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.7502399692839317, | |
| "grad_norm": 1.445178747177124, | |
| "learning_rate": 2.9763361131433208e-05, | |
| "loss": 0.9815, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.751007870992513, | |
| "grad_norm": 1.291642427444458, | |
| "learning_rate": 2.9590619301617183e-05, | |
| "loss": 1.0057, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.751007870992513, | |
| "eval_loss": 1.251602292060852, | |
| "eval_runtime": 19.752, | |
| "eval_samples_per_second": 27.795, | |
| "eval_steps_per_second": 13.923, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.7517757727010943, | |
| "grad_norm": 1.2883607149124146, | |
| "learning_rate": 2.9418293126798603e-05, | |
| "loss": 1.1005, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.7525436744096755, | |
| "grad_norm": 1.434813141822815, | |
| "learning_rate": 2.9246383624288387e-05, | |
| "loss": 1.201, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7533115761182568, | |
| "grad_norm": 2.126904249191284, | |
| "learning_rate": 2.9074891808937753e-05, | |
| "loss": 1.4774, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.7540794778268382, | |
| "grad_norm": 1.5914756059646606, | |
| "learning_rate": 2.8903818693132077e-05, | |
| "loss": 1.3087, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.7548473795354195, | |
| "grad_norm": 1.577763557434082, | |
| "learning_rate": 2.873316528678507e-05, | |
| "loss": 1.3263, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.7556152812440008, | |
| "grad_norm": 1.5671124458312988, | |
| "learning_rate": 2.856293259733266e-05, | |
| "loss": 1.2893, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.7563831829525821, | |
| "grad_norm": 1.6029667854309082, | |
| "learning_rate": 2.8393121629727138e-05, | |
| "loss": 1.1126, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.7571510846611633, | |
| "grad_norm": 1.4934439659118652, | |
| "learning_rate": 2.8223733386431185e-05, | |
| "loss": 1.2581, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.7579189863697446, | |
| "grad_norm": 1.327414870262146, | |
| "learning_rate": 2.8054768867411974e-05, | |
| "loss": 1.2253, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.758686888078326, | |
| "grad_norm": 1.5715781450271606, | |
| "learning_rate": 2.788622907013526e-05, | |
| "loss": 1.1577, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.7594547897869073, | |
| "grad_norm": 1.4220985174179077, | |
| "learning_rate": 2.7718114989559552e-05, | |
| "loss": 1.1358, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.7602226914954886, | |
| "grad_norm": 1.4818079471588135, | |
| "learning_rate": 2.7550427618130127e-05, | |
| "loss": 1.2783, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7609905932040699, | |
| "grad_norm": 1.394391655921936, | |
| "learning_rate": 2.738316794577315e-05, | |
| "loss": 1.2429, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.7617584949126511, | |
| "grad_norm": 1.7591447830200195, | |
| "learning_rate": 2.7216336959890076e-05, | |
| "loss": 1.5355, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.7625263966212324, | |
| "grad_norm": 1.226349949836731, | |
| "learning_rate": 2.704993564535152e-05, | |
| "loss": 1.0599, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.7632942983298138, | |
| "grad_norm": 1.3571503162384033, | |
| "learning_rate": 2.688396498449164e-05, | |
| "loss": 0.9554, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.7640622000383951, | |
| "grad_norm": 1.546478509902954, | |
| "learning_rate": 2.671842595710219e-05, | |
| "loss": 1.2609, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.7648301017469764, | |
| "grad_norm": 1.304118275642395, | |
| "learning_rate": 2.655331954042699e-05, | |
| "loss": 1.1655, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.7655980034555577, | |
| "grad_norm": 1.4672322273254395, | |
| "learning_rate": 2.638864670915572e-05, | |
| "loss": 1.1936, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.7663659051641389, | |
| "grad_norm": 1.1511002779006958, | |
| "learning_rate": 2.622440843541869e-05, | |
| "loss": 1.1561, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.7671338068727203, | |
| "grad_norm": 1.617389440536499, | |
| "learning_rate": 2.6060605688780694e-05, | |
| "loss": 1.2096, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.7679017085813016, | |
| "grad_norm": 1.7287838459014893, | |
| "learning_rate": 2.5897239436235466e-05, | |
| "loss": 1.3722, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7686696102898829, | |
| "grad_norm": 1.4887206554412842, | |
| "learning_rate": 2.5734310642199943e-05, | |
| "loss": 1.2082, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.7694375119984642, | |
| "grad_norm": 1.1259337663650513, | |
| "learning_rate": 2.557182026850855e-05, | |
| "loss": 0.937, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.7702054137070455, | |
| "grad_norm": 1.7888356447219849, | |
| "learning_rate": 2.5409769274407637e-05, | |
| "loss": 1.2036, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.7709733154156267, | |
| "grad_norm": 1.3678765296936035, | |
| "learning_rate": 2.524815861654952e-05, | |
| "loss": 1.183, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.7717412171242081, | |
| "grad_norm": 1.4759341478347778, | |
| "learning_rate": 2.5086989248987248e-05, | |
| "loss": 1.4874, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.7725091188327894, | |
| "grad_norm": 1.6217724084854126, | |
| "learning_rate": 2.492626212316862e-05, | |
| "loss": 1.3213, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.7732770205413707, | |
| "grad_norm": 1.735228180885315, | |
| "learning_rate": 2.476597818793075e-05, | |
| "loss": 1.4892, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.774044922249952, | |
| "grad_norm": 1.6505582332611084, | |
| "learning_rate": 2.460613838949437e-05, | |
| "loss": 1.3946, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.7748128239585333, | |
| "grad_norm": 1.5245342254638672, | |
| "learning_rate": 2.444674367145845e-05, | |
| "loss": 1.3344, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.7755807256671146, | |
| "grad_norm": 1.4345464706420898, | |
| "learning_rate": 2.4287794974794208e-05, | |
| "loss": 1.304, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.7763486273756959, | |
| "grad_norm": 1.120750069618225, | |
| "learning_rate": 2.4129293237840066e-05, | |
| "loss": 1.031, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.7771165290842772, | |
| "grad_norm": 1.4984389543533325, | |
| "learning_rate": 2.397123939629574e-05, | |
| "loss": 1.4389, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.7778844307928585, | |
| "grad_norm": 1.4427404403686523, | |
| "learning_rate": 2.3813634383216853e-05, | |
| "loss": 1.1667, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.7786523325014398, | |
| "grad_norm": 1.6551741361618042, | |
| "learning_rate": 2.3656479129009422e-05, | |
| "loss": 1.3006, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.7794202342100212, | |
| "grad_norm": 1.3143059015274048, | |
| "learning_rate": 2.3499774561424327e-05, | |
| "loss": 1.134, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.7801881359186024, | |
| "grad_norm": 1.7605934143066406, | |
| "learning_rate": 2.3343521605551967e-05, | |
| "loss": 1.1622, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.7809560376271837, | |
| "grad_norm": 1.5233832597732544, | |
| "learning_rate": 2.3187721183816503e-05, | |
| "loss": 1.2939, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.781723939335765, | |
| "grad_norm": 1.3118071556091309, | |
| "learning_rate": 2.303237421597082e-05, | |
| "loss": 0.9272, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.7824918410443463, | |
| "grad_norm": 1.559906005859375, | |
| "learning_rate": 2.2877481619090734e-05, | |
| "loss": 1.4749, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.7832597427529276, | |
| "grad_norm": 1.6087397336959839, | |
| "learning_rate": 2.2723044307569775e-05, | |
| "loss": 1.2059, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.784027644461509, | |
| "grad_norm": 1.5750290155410767, | |
| "learning_rate": 2.2569063193113716e-05, | |
| "loss": 1.2115, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.7847955461700902, | |
| "grad_norm": 1.4662597179412842, | |
| "learning_rate": 2.2415539184735323e-05, | |
| "loss": 0.9787, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.7855634478786715, | |
| "grad_norm": 1.5807639360427856, | |
| "learning_rate": 2.2262473188748667e-05, | |
| "loss": 1.5, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.7863313495872528, | |
| "grad_norm": 1.4712680578231812, | |
| "learning_rate": 2.210986610876421e-05, | |
| "loss": 1.1427, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.7870992512958341, | |
| "grad_norm": 1.7284588813781738, | |
| "learning_rate": 2.1957718845683106e-05, | |
| "loss": 0.9901, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.7878671530044155, | |
| "grad_norm": 1.7499277591705322, | |
| "learning_rate": 2.180603229769208e-05, | |
| "loss": 1.7659, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.7886350547129968, | |
| "grad_norm": 2.04580020904541, | |
| "learning_rate": 2.1654807360258068e-05, | |
| "loss": 1.4249, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.789402956421578, | |
| "grad_norm": 1.8796188831329346, | |
| "learning_rate": 2.15040449261229e-05, | |
| "loss": 1.4763, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.7901708581301593, | |
| "grad_norm": 1.7241255044937134, | |
| "learning_rate": 2.1353745885298103e-05, | |
| "loss": 1.4569, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.7909387598387406, | |
| "grad_norm": 1.2481935024261475, | |
| "learning_rate": 2.120391112505955e-05, | |
| "loss": 1.1742, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.7917066615473219, | |
| "grad_norm": 1.440290093421936, | |
| "learning_rate": 2.1054541529942374e-05, | |
| "loss": 1.243, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.7924745632559033, | |
| "grad_norm": 1.8699524402618408, | |
| "learning_rate": 2.090563798173557e-05, | |
| "loss": 1.3004, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.7932424649644846, | |
| "grad_norm": 1.7997124195098877, | |
| "learning_rate": 2.0757201359476884e-05, | |
| "loss": 1.5915, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.7940103666730658, | |
| "grad_norm": 1.4160553216934204, | |
| "learning_rate": 2.060923253944761e-05, | |
| "loss": 1.3386, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.7947782683816471, | |
| "grad_norm": 1.4589799642562866, | |
| "learning_rate": 2.0461732395167475e-05, | |
| "loss": 1.3019, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.7955461700902284, | |
| "grad_norm": 1.8283833265304565, | |
| "learning_rate": 2.03147017973893e-05, | |
| "loss": 1.3423, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.7963140717988098, | |
| "grad_norm": 1.4732909202575684, | |
| "learning_rate": 2.0168141614094126e-05, | |
| "loss": 1.2183, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.7970819735073911, | |
| "grad_norm": 1.5215457677841187, | |
| "learning_rate": 2.0022052710485874e-05, | |
| "loss": 1.2761, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.7978498752159724, | |
| "grad_norm": 1.812894344329834, | |
| "learning_rate": 1.9876435948986228e-05, | |
| "loss": 1.4416, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.7986177769245536, | |
| "grad_norm": 1.212895154953003, | |
| "learning_rate": 1.973129218922981e-05, | |
| "loss": 1.0733, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.7993856786331349, | |
| "grad_norm": 1.5380189418792725, | |
| "learning_rate": 1.95866222880588e-05, | |
| "loss": 1.4282, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.8001535803417162, | |
| "grad_norm": 1.671745777130127, | |
| "learning_rate": 1.9442427099518022e-05, | |
| "loss": 1.4183, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.8009214820502976, | |
| "grad_norm": 1.6439645290374756, | |
| "learning_rate": 1.9298707474849843e-05, | |
| "loss": 1.2435, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.8016893837588789, | |
| "grad_norm": 1.3178609609603882, | |
| "learning_rate": 1.9155464262489298e-05, | |
| "loss": 0.843, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.8024572854674602, | |
| "grad_norm": 1.7350953817367554, | |
| "learning_rate": 1.9012698308058852e-05, | |
| "loss": 1.4906, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.8032251871760415, | |
| "grad_norm": 1.2103064060211182, | |
| "learning_rate": 1.8870410454363573e-05, | |
| "loss": 0.9144, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.8039930888846227, | |
| "grad_norm": 1.1046971082687378, | |
| "learning_rate": 1.872860154138608e-05, | |
| "loss": 0.9619, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.804760990593204, | |
| "grad_norm": 1.832051157951355, | |
| "learning_rate": 1.858727240628171e-05, | |
| "loss": 1.4815, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.8055288923017854, | |
| "grad_norm": 1.1102668046951294, | |
| "learning_rate": 1.8446423883373286e-05, | |
| "loss": 0.9905, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.8062967940103667, | |
| "grad_norm": 1.155604600906372, | |
| "learning_rate": 1.8306056804146575e-05, | |
| "loss": 1.0172, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.807064695718948, | |
| "grad_norm": 1.4365675449371338, | |
| "learning_rate": 1.816617199724512e-05, | |
| "loss": 1.3202, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.8078325974275293, | |
| "grad_norm": 1.2660962343215942, | |
| "learning_rate": 1.8026770288465323e-05, | |
| "loss": 1.2372, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.8086004991361105, | |
| "grad_norm": 1.484471082687378, | |
| "learning_rate": 1.7887852500751822e-05, | |
| "loss": 1.2489, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.8093684008446919, | |
| "grad_norm": 1.62258780002594, | |
| "learning_rate": 1.7749419454192373e-05, | |
| "loss": 1.3631, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.8101363025532732, | |
| "grad_norm": 1.2742317914962769, | |
| "learning_rate": 1.7611471966013127e-05, | |
| "loss": 1.0303, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.8109042042618545, | |
| "grad_norm": 1.770555853843689, | |
| "learning_rate": 1.7474010850573775e-05, | |
| "loss": 1.4317, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.8116721059704358, | |
| "grad_norm": 1.3370784521102905, | |
| "learning_rate": 1.7337036919362827e-05, | |
| "loss": 1.1577, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.8124400076790171, | |
| "grad_norm": 1.4367464780807495, | |
| "learning_rate": 1.7200550980992647e-05, | |
| "loss": 1.3085, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.8132079093875983, | |
| "grad_norm": 1.363452434539795, | |
| "learning_rate": 1.706455384119485e-05, | |
| "loss": 1.1423, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.8139758110961797, | |
| "grad_norm": 1.3414223194122314, | |
| "learning_rate": 1.6929046302815443e-05, | |
| "loss": 1.2153, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.814743712804761, | |
| "grad_norm": 1.577257513999939, | |
| "learning_rate": 1.6794029165810133e-05, | |
| "loss": 1.357, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.8155116145133423, | |
| "grad_norm": 1.594196081161499, | |
| "learning_rate": 1.665950322723957e-05, | |
| "loss": 1.2597, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.8162795162219236, | |
| "grad_norm": 1.5210363864898682, | |
| "learning_rate": 1.652546928126466e-05, | |
| "loss": 1.2459, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.817047417930505, | |
| "grad_norm": 1.3744914531707764, | |
| "learning_rate": 1.6391928119141965e-05, | |
| "loss": 1.236, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.8178153196390862, | |
| "grad_norm": 1.56074857711792, | |
| "learning_rate": 1.625888052921878e-05, | |
| "loss": 1.2514, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.8185832213476675, | |
| "grad_norm": 1.3938485383987427, | |
| "learning_rate": 1.612632729692881e-05, | |
| "loss": 1.3724, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.8193511230562488, | |
| "grad_norm": 1.2004398107528687, | |
| "learning_rate": 1.599426920478726e-05, | |
| "loss": 0.8796, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.8201190247648301, | |
| "grad_norm": 1.6214113235473633, | |
| "learning_rate": 1.586270703238637e-05, | |
| "loss": 1.3691, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.8208869264734114, | |
| "grad_norm": 1.7719411849975586, | |
| "learning_rate": 1.573164155639073e-05, | |
| "loss": 1.5033, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.8216548281819928, | |
| "grad_norm": 1.2365126609802246, | |
| "learning_rate": 1.560107355053282e-05, | |
| "loss": 1.1611, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.822422729890574, | |
| "grad_norm": 1.804200291633606, | |
| "learning_rate": 1.5471003785608184e-05, | |
| "loss": 1.2228, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.8231906315991553, | |
| "grad_norm": 3.12967586517334, | |
| "learning_rate": 1.534143302947123e-05, | |
| "loss": 1.0455, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.8239585333077366, | |
| "grad_norm": 1.7391802072525024, | |
| "learning_rate": 1.5212362047030427e-05, | |
| "loss": 1.4632, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.8247264350163179, | |
| "grad_norm": 1.4144318103790283, | |
| "learning_rate": 1.5083791600243857e-05, | |
| "loss": 1.2948, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.8254943367248992, | |
| "grad_norm": 1.2028521299362183, | |
| "learning_rate": 1.4955722448114807e-05, | |
| "loss": 0.9203, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.8262622384334806, | |
| "grad_norm": 1.4866068363189697, | |
| "learning_rate": 1.4828155346687123e-05, | |
| "loss": 1.3486, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.8270301401420618, | |
| "grad_norm": 1.471641182899475, | |
| "learning_rate": 1.4701091049040994e-05, | |
| "loss": 1.1633, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.8277980418506431, | |
| "grad_norm": 1.3013348579406738, | |
| "learning_rate": 1.4574530305288158e-05, | |
| "loss": 0.9819, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.8285659435592244, | |
| "grad_norm": 1.5000224113464355, | |
| "learning_rate": 1.4448473862567857e-05, | |
| "loss": 1.1458, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.8293338452678057, | |
| "grad_norm": 1.4229687452316284, | |
| "learning_rate": 1.4322922465042132e-05, | |
| "loss": 1.3797, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8301017469763871, | |
| "grad_norm": 1.3966771364212036, | |
| "learning_rate": 1.4197876853891557e-05, | |
| "loss": 1.4436, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.8308696486849684, | |
| "grad_norm": 1.3111258745193481, | |
| "learning_rate": 1.4073337767310834e-05, | |
| "loss": 1.175, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.8316375503935496, | |
| "grad_norm": 1.4146270751953125, | |
| "learning_rate": 1.3949305940504541e-05, | |
| "loss": 1.1575, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.8324054521021309, | |
| "grad_norm": 1.2625914812088013, | |
| "learning_rate": 1.3825782105682527e-05, | |
| "loss": 0.984, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.8331733538107122, | |
| "grad_norm": 1.2037335634231567, | |
| "learning_rate": 1.3702766992055927e-05, | |
| "loss": 1.0908, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.8339412555192935, | |
| "grad_norm": 1.384108304977417, | |
| "learning_rate": 1.3580261325832578e-05, | |
| "loss": 1.4079, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.8347091572278749, | |
| "grad_norm": 1.0640181303024292, | |
| "learning_rate": 1.3458265830212891e-05, | |
| "loss": 0.68, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.8354770589364562, | |
| "grad_norm": 1.6578110456466675, | |
| "learning_rate": 1.333678122538553e-05, | |
| "loss": 1.1659, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.8362449606450374, | |
| "grad_norm": 1.3299002647399902, | |
| "learning_rate": 1.321580822852313e-05, | |
| "loss": 1.1961, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.8370128623536187, | |
| "grad_norm": 1.5801900625228882, | |
| "learning_rate": 1.3095347553778193e-05, | |
| "loss": 1.3358, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8377807640622, | |
| "grad_norm": 1.6045485734939575, | |
| "learning_rate": 1.2975399912278608e-05, | |
| "loss": 1.3452, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.8385486657707814, | |
| "grad_norm": 1.3222181797027588, | |
| "learning_rate": 1.2855966012123822e-05, | |
| "loss": 1.0153, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.8393165674793627, | |
| "grad_norm": 1.5065910816192627, | |
| "learning_rate": 1.2737046558380305e-05, | |
| "loss": 1.3556, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.840084469187944, | |
| "grad_norm": 1.6371604204177856, | |
| "learning_rate": 1.2618642253077628e-05, | |
| "loss": 1.3203, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.8408523708965252, | |
| "grad_norm": 1.801613688468933, | |
| "learning_rate": 1.2500753795204157e-05, | |
| "loss": 1.272, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.8416202726051065, | |
| "grad_norm": 1.6735864877700806, | |
| "learning_rate": 1.2383381880703138e-05, | |
| "loss": 1.2709, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.8423881743136878, | |
| "grad_norm": 1.3746118545532227, | |
| "learning_rate": 1.2266527202468248e-05, | |
| "loss": 1.164, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.8431560760222692, | |
| "grad_norm": 1.5882354974746704, | |
| "learning_rate": 1.2150190450339915e-05, | |
| "loss": 0.8384, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.8439239777308505, | |
| "grad_norm": 1.492803931236267, | |
| "learning_rate": 1.2034372311100905e-05, | |
| "loss": 1.4302, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.8446918794394318, | |
| "grad_norm": 1.868151307106018, | |
| "learning_rate": 1.1919073468472475e-05, | |
| "loss": 1.24, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.845459781148013, | |
| "grad_norm": 1.3878445625305176, | |
| "learning_rate": 1.1804294603110222e-05, | |
| "loss": 1.1959, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.8462276828565943, | |
| "grad_norm": 1.6517834663391113, | |
| "learning_rate": 1.1690036392600112e-05, | |
| "loss": 1.4946, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.8469955845651757, | |
| "grad_norm": 1.5959718227386475, | |
| "learning_rate": 1.1576299511454513e-05, | |
| "loss": 1.3349, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.847763486273757, | |
| "grad_norm": 1.3100310564041138, | |
| "learning_rate": 1.1463084631108101e-05, | |
| "loss": 1.0446, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.8485313879823383, | |
| "grad_norm": 1.5334558486938477, | |
| "learning_rate": 1.135039241991408e-05, | |
| "loss": 1.0935, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.8492992896909196, | |
| "grad_norm": 1.306308627128601, | |
| "learning_rate": 1.1238223543140024e-05, | |
| "loss": 1.0424, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.8500671913995008, | |
| "grad_norm": 1.5370351076126099, | |
| "learning_rate": 1.1126578662964115e-05, | |
| "loss": 1.1695, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.8508350931080821, | |
| "grad_norm": 1.5034894943237305, | |
| "learning_rate": 1.1015458438471116e-05, | |
| "loss": 1.3281, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.8516029948166635, | |
| "grad_norm": 1.815669059753418, | |
| "learning_rate": 1.0904863525648633e-05, | |
| "loss": 1.3144, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.8523708965252448, | |
| "grad_norm": 1.2194162607192993, | |
| "learning_rate": 1.0794794577383016e-05, | |
| "loss": 1.1239, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8531387982338261, | |
| "grad_norm": 1.3530551195144653, | |
| "learning_rate": 1.0685252243455712e-05, | |
| "loss": 1.1957, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.8539066999424074, | |
| "grad_norm": 1.437690019607544, | |
| "learning_rate": 1.0576237170539383e-05, | |
| "loss": 1.1088, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.8546746016509886, | |
| "grad_norm": 1.7501755952835083, | |
| "learning_rate": 1.0467750002193944e-05, | |
| "loss": 1.3246, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.85544250335957, | |
| "grad_norm": 1.587377905845642, | |
| "learning_rate": 1.0359791378863005e-05, | |
| "loss": 1.2558, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.8562104050681513, | |
| "grad_norm": 1.8887481689453125, | |
| "learning_rate": 1.025236193786987e-05, | |
| "loss": 1.523, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.8569783067767326, | |
| "grad_norm": 1.4728237390518188, | |
| "learning_rate": 1.014546231341391e-05, | |
| "loss": 1.4592, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.8577462084853139, | |
| "grad_norm": 1.5659328699111938, | |
| "learning_rate": 1.0039093136566735e-05, | |
| "loss": 1.2452, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.8585141101938952, | |
| "grad_norm": 1.501523733139038, | |
| "learning_rate": 9.933255035268574e-06, | |
| "loss": 1.1862, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.8592820119024764, | |
| "grad_norm": 1.4726933240890503, | |
| "learning_rate": 9.827948634324447e-06, | |
| "loss": 1.1691, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.8600499136110578, | |
| "grad_norm": 1.601131558418274, | |
| "learning_rate": 9.72317455540055e-06, | |
| "loss": 1.2678, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8608178153196391, | |
| "grad_norm": 1.5712275505065918, | |
| "learning_rate": 9.61893341702056e-06, | |
| "loss": 1.1582, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.8615857170282204, | |
| "grad_norm": 1.4145914316177368, | |
| "learning_rate": 9.515225834562003e-06, | |
| "loss": 0.9806, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.8623536187368017, | |
| "grad_norm": 1.4771418571472168, | |
| "learning_rate": 9.412052420252605e-06, | |
| "loss": 1.233, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.863121520445383, | |
| "grad_norm": 1.4225813150405884, | |
| "learning_rate": 9.309413783166654e-06, | |
| "loss": 0.8999, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.8638894221539642, | |
| "grad_norm": 1.6784931421279907, | |
| "learning_rate": 9.207310529221525e-06, | |
| "loss": 1.2208, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.8646573238625456, | |
| "grad_norm": 1.05604887008667, | |
| "learning_rate": 9.10574326117386e-06, | |
| "loss": 0.9784, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.8654252255711269, | |
| "grad_norm": 1.5101516246795654, | |
| "learning_rate": 9.004712578616304e-06, | |
| "loss": 1.2459, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.8661931272797082, | |
| "grad_norm": 1.5375887155532837, | |
| "learning_rate": 8.90421907797374e-06, | |
| "loss": 1.3941, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.8669610289882895, | |
| "grad_norm": 1.382124423980713, | |
| "learning_rate": 8.804263352499864e-06, | |
| "loss": 1.0871, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.8677289306968708, | |
| "grad_norm": 1.4245071411132812, | |
| "learning_rate": 8.70484599227367e-06, | |
| "loss": 1.0823, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8684968324054521, | |
| "grad_norm": 1.651361107826233, | |
| "learning_rate": 8.605967584195995e-06, | |
| "loss": 1.3007, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.8692647341140334, | |
| "grad_norm": 1.8151158094406128, | |
| "learning_rate": 8.507628711985983e-06, | |
| "loss": 1.3263, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.8700326358226147, | |
| "grad_norm": 1.6930426359176636, | |
| "learning_rate": 8.409829956177684e-06, | |
| "loss": 1.2812, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.870800537531196, | |
| "grad_norm": 1.6178615093231201, | |
| "learning_rate": 8.312571894116649e-06, | |
| "loss": 1.2922, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.8715684392397773, | |
| "grad_norm": 1.7102075815200806, | |
| "learning_rate": 8.215855099956472e-06, | |
| "loss": 1.1281, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.8723363409483587, | |
| "grad_norm": 1.4845244884490967, | |
| "learning_rate": 8.119680144655428e-06, | |
| "loss": 1.0213, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.8731042426569399, | |
| "grad_norm": 1.4863834381103516, | |
| "learning_rate": 8.024047595973095e-06, | |
| "loss": 1.0862, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.8738721443655212, | |
| "grad_norm": 1.2989436388015747, | |
| "learning_rate": 7.92895801846707e-06, | |
| "loss": 1.052, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.8746400460741025, | |
| "grad_norm": 1.3424791097640991, | |
| "learning_rate": 7.834411973489419e-06, | |
| "loss": 1.3913, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.8754079477826838, | |
| "grad_norm": 1.5543662309646606, | |
| "learning_rate": 7.740410019183697e-06, | |
| "loss": 1.4759, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.8761758494912651, | |
| "grad_norm": 1.499265193939209, | |
| "learning_rate": 7.646952710481336e-06, | |
| "loss": 1.3668, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.8769437511998465, | |
| "grad_norm": 1.6394531726837158, | |
| "learning_rate": 7.554040599098588e-06, | |
| "loss": 1.1713, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.8777116529084277, | |
| "grad_norm": 1.4457104206085205, | |
| "learning_rate": 7.461674233533123e-06, | |
| "loss": 1.3674, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.878479554617009, | |
| "grad_norm": 1.468809723854065, | |
| "learning_rate": 7.369854159060929e-06, | |
| "loss": 1.0444, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.8792474563255903, | |
| "grad_norm": 1.610374093055725, | |
| "learning_rate": 7.278580917732913e-06, | |
| "loss": 1.5037, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.8800153580341716, | |
| "grad_norm": 1.3369286060333252, | |
| "learning_rate": 7.187855048371917e-06, | |
| "loss": 1.129, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.880783259742753, | |
| "grad_norm": 1.4274189472198486, | |
| "learning_rate": 7.097677086569343e-06, | |
| "loss": 1.2629, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.8815511614513343, | |
| "grad_norm": 1.4373105764389038, | |
| "learning_rate": 7.008047564682119e-06, | |
| "loss": 1.2244, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.8823190631599155, | |
| "grad_norm": 1.5060944557189941, | |
| "learning_rate": 6.91896701182948e-06, | |
| "loss": 1.3279, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.8830869648684968, | |
| "grad_norm": 1.1090878248214722, | |
| "learning_rate": 6.83043595388988e-06, | |
| "loss": 0.9791, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.8838548665770781, | |
| "grad_norm": 1.6525593996047974, | |
| "learning_rate": 6.742454913497942e-06, | |
| "loss": 1.2751, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.8846227682856594, | |
| "grad_norm": 1.5659713745117188, | |
| "learning_rate": 6.6550244100412e-06, | |
| "loss": 1.3544, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.8853906699942408, | |
| "grad_norm": 1.4881879091262817, | |
| "learning_rate": 6.568144959657263e-06, | |
| "loss": 1.1873, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.8861585717028221, | |
| "grad_norm": 1.8786675930023193, | |
| "learning_rate": 6.481817075230567e-06, | |
| "loss": 1.3176, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.8869264734114033, | |
| "grad_norm": 1.4211797714233398, | |
| "learning_rate": 6.396041266389474e-06, | |
| "loss": 1.3475, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.8876943751199846, | |
| "grad_norm": 1.4349972009658813, | |
| "learning_rate": 6.3108180395031965e-06, | |
| "loss": 1.1019, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.8884622768285659, | |
| "grad_norm": 1.3336567878723145, | |
| "learning_rate": 6.22614789767888e-06, | |
| "loss": 1.1098, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.8892301785371473, | |
| "grad_norm": 1.4133919477462769, | |
| "learning_rate": 6.142031340758525e-06, | |
| "loss": 1.2118, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.8899980802457286, | |
| "grad_norm": 1.3893505334854126, | |
| "learning_rate": 6.058468865316102e-06, | |
| "loss": 1.1315, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.8907659819543099, | |
| "grad_norm": 1.786585807800293, | |
| "learning_rate": 5.975460964654689e-06, | |
| "loss": 1.6215, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.8915338836628911, | |
| "grad_norm": 1.3746907711029053, | |
| "learning_rate": 5.8930081288034014e-06, | |
| "loss": 1.0703, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.8923017853714724, | |
| "grad_norm": 1.420802116394043, | |
| "learning_rate": 5.8111108445146116e-06, | |
| "loss": 1.0426, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.8930696870800537, | |
| "grad_norm": 1.5010699033737183, | |
| "learning_rate": 5.72976959526107e-06, | |
| "loss": 1.3698, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.8938375887886351, | |
| "grad_norm": 1.3752045631408691, | |
| "learning_rate": 5.648984861232986e-06, | |
| "loss": 1.2014, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.8946054904972164, | |
| "grad_norm": 1.5743210315704346, | |
| "learning_rate": 5.568757119335244e-06, | |
| "loss": 1.4007, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.8953733922057977, | |
| "grad_norm": 1.4705897569656372, | |
| "learning_rate": 5.489086843184632e-06, | |
| "loss": 1.0485, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.8961412939143789, | |
| "grad_norm": 1.4664305448532104, | |
| "learning_rate": 5.40997450310693e-06, | |
| "loss": 1.3765, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.8969091956229602, | |
| "grad_norm": 1.6887331008911133, | |
| "learning_rate": 5.3314205661342155e-06, | |
| "loss": 1.3474, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.8976770973315416, | |
| "grad_norm": 1.5927088260650635, | |
| "learning_rate": 5.253425496002084e-06, | |
| "loss": 1.2915, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.8984449990401229, | |
| "grad_norm": 1.3032876253128052, | |
| "learning_rate": 5.175989753146948e-06, | |
| "loss": 1.2463, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.8992129007487042, | |
| "grad_norm": 1.2626043558120728, | |
| "learning_rate": 5.099113794703225e-06, | |
| "loss": 1.2223, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.8999808024572855, | |
| "grad_norm": 1.5485360622406006, | |
| "learning_rate": 5.0227980745007345e-06, | |
| "loss": 1.4437, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.9007487041658667, | |
| "grad_norm": 1.3756585121154785, | |
| "learning_rate": 4.947043043062016e-06, | |
| "loss": 1.3702, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.901516605874448, | |
| "grad_norm": 1.281550645828247, | |
| "learning_rate": 4.87184914759955e-06, | |
| "loss": 0.8686, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.9022845075830294, | |
| "grad_norm": 1.7812248468399048, | |
| "learning_rate": 4.7972168320132845e-06, | |
| "loss": 1.2599, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.9030524092916107, | |
| "grad_norm": 1.186799168586731, | |
| "learning_rate": 4.7231465368879124e-06, | |
| "loss": 1.0582, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.903820311000192, | |
| "grad_norm": 1.4105154275894165, | |
| "learning_rate": 4.649638699490266e-06, | |
| "loss": 1.1345, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.9045882127087733, | |
| "grad_norm": 1.4764769077301025, | |
| "learning_rate": 4.576693753766792e-06, | |
| "loss": 1.3989, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.9053561144173545, | |
| "grad_norm": 2.1660075187683105, | |
| "learning_rate": 4.5043121303409595e-06, | |
| "loss": 1.1665, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.9061240161259358, | |
| "grad_norm": 1.262603521347046, | |
| "learning_rate": 4.432494256510711e-06, | |
| "loss": 0.9425, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.9068919178345172, | |
| "grad_norm": 1.8108893632888794, | |
| "learning_rate": 4.361240556245938e-06, | |
| "loss": 1.2851, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.9076598195430985, | |
| "grad_norm": 1.5858032703399658, | |
| "learning_rate": 4.290551450185986e-06, | |
| "loss": 1.4577, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.9084277212516798, | |
| "grad_norm": 1.2271817922592163, | |
| "learning_rate": 4.220427355637224e-06, | |
| "loss": 1.1701, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.9091956229602611, | |
| "grad_norm": 1.280598759651184, | |
| "learning_rate": 4.150868686570464e-06, | |
| "loss": 1.2309, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.9099635246688423, | |
| "grad_norm": 1.40516996383667, | |
| "learning_rate": 4.081875853618588e-06, | |
| "loss": 1.0201, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.9107314263774237, | |
| "grad_norm": 1.4448282718658447, | |
| "learning_rate": 4.013449264074187e-06, | |
| "loss": 1.1834, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.911499328086005, | |
| "grad_norm": 1.2964657545089722, | |
| "learning_rate": 3.945589321886989e-06, | |
| "loss": 1.0199, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.9122672297945863, | |
| "grad_norm": 1.3470900058746338, | |
| "learning_rate": 3.878296427661676e-06, | |
| "loss": 1.0585, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.9130351315031676, | |
| "grad_norm": 1.3308759927749634, | |
| "learning_rate": 3.8115709786553435e-06, | |
| "loss": 1.3752, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.9138030332117489, | |
| "grad_norm": 1.4458931684494019, | |
| "learning_rate": 3.7454133687752524e-06, | |
| "loss": 1.1349, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9145709349203301, | |
| "grad_norm": 1.471204400062561, | |
| "learning_rate": 3.6798239885764806e-06, | |
| "loss": 1.126, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.9153388366289115, | |
| "grad_norm": 1.403232216835022, | |
| "learning_rate": 3.614803225259622e-06, | |
| "loss": 1.1377, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.9161067383374928, | |
| "grad_norm": 1.5859801769256592, | |
| "learning_rate": 3.550351462668489e-06, | |
| "loss": 1.2045, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.9168746400460741, | |
| "grad_norm": 1.2398918867111206, | |
| "learning_rate": 3.4864690812878688e-06, | |
| "loss": 1.111, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.9176425417546554, | |
| "grad_norm": 1.455277681350708, | |
| "learning_rate": 3.4231564582412167e-06, | |
| "loss": 1.1314, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.9184104434632367, | |
| "grad_norm": 1.5992294549942017, | |
| "learning_rate": 3.3604139672885227e-06, | |
| "loss": 1.2585, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.919178345171818, | |
| "grad_norm": 1.6772078275680542, | |
| "learning_rate": 3.298241978824046e-06, | |
| "loss": 1.0393, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.9199462468803993, | |
| "grad_norm": 1.202988862991333, | |
| "learning_rate": 3.2366408598741072e-06, | |
| "loss": 1.0763, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.9207141485889806, | |
| "grad_norm": 1.5782155990600586, | |
| "learning_rate": 3.175610974095011e-06, | |
| "loss": 1.3341, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.9214820502975619, | |
| "grad_norm": 1.611743450164795, | |
| "learning_rate": 3.115152681770783e-06, | |
| "loss": 1.0767, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9222499520061432, | |
| "grad_norm": 1.6371206045150757, | |
| "learning_rate": 3.055266339811147e-06, | |
| "loss": 0.7947, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.9230178537147246, | |
| "grad_norm": 1.335860252380371, | |
| "learning_rate": 2.9959523017493386e-06, | |
| "loss": 1.2236, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.9237857554233058, | |
| "grad_norm": 1.4622019529342651, | |
| "learning_rate": 2.9372109177400854e-06, | |
| "loss": 1.5577, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.9245536571318871, | |
| "grad_norm": 1.4906941652297974, | |
| "learning_rate": 2.8790425345574745e-06, | |
| "loss": 1.1759, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.9253215588404684, | |
| "grad_norm": 1.2302792072296143, | |
| "learning_rate": 2.821447495592977e-06, | |
| "loss": 0.7965, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.9260894605490497, | |
| "grad_norm": 1.4680209159851074, | |
| "learning_rate": 2.7644261408533155e-06, | |
| "loss": 1.213, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.926857362257631, | |
| "grad_norm": 1.737641453742981, | |
| "learning_rate": 2.707978806958611e-06, | |
| "loss": 1.1734, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.9276252639662124, | |
| "grad_norm": 1.386087417602539, | |
| "learning_rate": 2.6521058271402386e-06, | |
| "loss": 1.3075, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.9283931656747936, | |
| "grad_norm": 1.5818560123443604, | |
| "learning_rate": 2.596807531238965e-06, | |
| "loss": 1.3811, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.9291610673833749, | |
| "grad_norm": 1.2463423013687134, | |
| "learning_rate": 2.542084245702947e-06, | |
| "loss": 1.2488, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.9299289690919562, | |
| "grad_norm": 1.1760039329528809, | |
| "learning_rate": 2.487936293585813e-06, | |
| "loss": 0.9825, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.9306968708005375, | |
| "grad_norm": 1.2284053564071655, | |
| "learning_rate": 2.4343639945448306e-06, | |
| "loss": 1.12, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.9314647725091189, | |
| "grad_norm": 1.4763203859329224, | |
| "learning_rate": 2.3813676648388517e-06, | |
| "loss": 1.1244, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.9322326742177002, | |
| "grad_norm": 1.6483160257339478, | |
| "learning_rate": 2.3289476173266376e-06, | |
| "loss": 1.3215, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.9330005759262814, | |
| "grad_norm": 1.625486135482788, | |
| "learning_rate": 2.2771041614648825e-06, | |
| "loss": 1.578, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.9337684776348627, | |
| "grad_norm": 1.7281734943389893, | |
| "learning_rate": 2.2258376033064354e-06, | |
| "loss": 1.3752, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.934536379343444, | |
| "grad_norm": 1.1822775602340698, | |
| "learning_rate": 2.1751482454984706e-06, | |
| "loss": 0.9411, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.9353042810520253, | |
| "grad_norm": 1.5272676944732666, | |
| "learning_rate": 2.1250363872807655e-06, | |
| "loss": 1.3861, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.9360721827606067, | |
| "grad_norm": 1.6605676412582397, | |
| "learning_rate": 2.0755023244838136e-06, | |
| "loss": 1.5562, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.936840084469188, | |
| "grad_norm": 1.4125683307647705, | |
| "learning_rate": 2.026546349527181e-06, | |
| "loss": 1.0978, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9376079861777692, | |
| "grad_norm": 1.3512035608291626, | |
| "learning_rate": 1.978168751417786e-06, | |
| "loss": 0.9322, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.9383758878863505, | |
| "grad_norm": 1.7064318656921387, | |
| "learning_rate": 1.93036981574809e-06, | |
| "loss": 1.2816, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.9391437895949318, | |
| "grad_norm": 1.2211792469024658, | |
| "learning_rate": 1.8831498246945189e-06, | |
| "loss": 1.0071, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.9399116913035132, | |
| "grad_norm": 1.3540048599243164, | |
| "learning_rate": 1.836509057015734e-06, | |
| "loss": 1.202, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.9406795930120945, | |
| "grad_norm": 1.3373445272445679, | |
| "learning_rate": 1.7904477880510307e-06, | |
| "loss": 1.142, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.9414474947206758, | |
| "grad_norm": 1.38016939163208, | |
| "learning_rate": 1.7449662897186414e-06, | |
| "loss": 1.3193, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.942215396429257, | |
| "grad_norm": 1.3662525415420532, | |
| "learning_rate": 1.7000648305142364e-06, | |
| "loss": 1.0969, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 0.9429832981378383, | |
| "grad_norm": 1.7097259759902954, | |
| "learning_rate": 1.655743675509258e-06, | |
| "loss": 1.4585, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.9437511998464196, | |
| "grad_norm": 1.177868127822876, | |
| "learning_rate": 1.6120030863493674e-06, | |
| "loss": 0.9733, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 0.944519101555001, | |
| "grad_norm": 1.481663465499878, | |
| "learning_rate": 1.5688433212529107e-06, | |
| "loss": 1.0466, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9452870032635823, | |
| "grad_norm": 1.6859853267669678, | |
| "learning_rate": 1.5262646350094334e-06, | |
| "loss": 1.2808, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 0.9460549049721636, | |
| "grad_norm": 1.7730293273925781, | |
| "learning_rate": 1.4842672789780798e-06, | |
| "loss": 1.2563, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.9468228066807448, | |
| "grad_norm": 1.5149153470993042, | |
| "learning_rate": 1.4428515010861955e-06, | |
| "loss": 1.249, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 0.9475907083893261, | |
| "grad_norm": 1.8611276149749756, | |
| "learning_rate": 1.4020175458278607e-06, | |
| "loss": 1.5504, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.9483586100979075, | |
| "grad_norm": 1.5275269746780396, | |
| "learning_rate": 1.3617656542623813e-06, | |
| "loss": 1.3417, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.9491265118064888, | |
| "grad_norm": 1.3223843574523926, | |
| "learning_rate": 1.322096064012912e-06, | |
| "loss": 1.1907, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.9498944135150701, | |
| "grad_norm": 1.4529744386672974, | |
| "learning_rate": 1.2830090092650904e-06, | |
| "loss": 1.16, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 0.9506623152236514, | |
| "grad_norm": 1.4799784421920776, | |
| "learning_rate": 1.244504720765549e-06, | |
| "loss": 1.3583, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.9514302169322326, | |
| "grad_norm": 1.4964150190353394, | |
| "learning_rate": 1.20658342582064e-06, | |
| "loss": 1.2563, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 0.9521981186408139, | |
| "grad_norm": 1.318679690361023, | |
| "learning_rate": 1.1692453482951115e-06, | |
| "loss": 1.1699, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9529660203493953, | |
| "grad_norm": 1.2729301452636719, | |
| "learning_rate": 1.1324907086106895e-06, | |
| "loss": 1.0768, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 0.9537339220579766, | |
| "grad_norm": 1.4914193153381348, | |
| "learning_rate": 1.096319723744843e-06, | |
| "loss": 0.9835, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.9545018237665579, | |
| "grad_norm": 1.4499318599700928, | |
| "learning_rate": 1.0607326072295087e-06, | |
| "loss": 1.1102, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 0.9552697254751392, | |
| "grad_norm": 1.4721513986587524, | |
| "learning_rate": 1.0257295691497914e-06, | |
| "loss": 1.234, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.9560376271837204, | |
| "grad_norm": 1.553381323814392, | |
| "learning_rate": 9.913108161427543e-07, | |
| "loss": 1.2512, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.9568055288923017, | |
| "grad_norm": 1.384883165359497, | |
| "learning_rate": 9.57476551396197e-07, | |
| "loss": 1.2325, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.9575734306008831, | |
| "grad_norm": 1.6136502027511597, | |
| "learning_rate": 9.242269746474575e-07, | |
| "loss": 1.4892, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 0.9583413323094644, | |
| "grad_norm": 1.5573700666427612, | |
| "learning_rate": 8.915622821821789e-07, | |
| "loss": 1.447, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.9591092340180457, | |
| "grad_norm": 1.47049880027771, | |
| "learning_rate": 8.594826668332445e-07, | |
| "loss": 1.1597, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 0.959877135726627, | |
| "grad_norm": 0.9694231748580933, | |
| "learning_rate": 8.279883179795666e-07, | |
| "loss": 0.8528, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.9606450374352083, | |
| "grad_norm": 1.28829026222229, | |
| "learning_rate": 7.970794215450106e-07, | |
| "loss": 1.1936, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 0.9614129391437896, | |
| "grad_norm": 1.6050745248794556, | |
| "learning_rate": 7.667561599972505e-07, | |
| "loss": 1.3697, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.9621808408523709, | |
| "grad_norm": 1.1561931371688843, | |
| "learning_rate": 7.370187123467708e-07, | |
| "loss": 0.9132, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 0.9629487425609522, | |
| "grad_norm": 1.2682461738586426, | |
| "learning_rate": 7.078672541456999e-07, | |
| "loss": 1.1176, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.9637166442695335, | |
| "grad_norm": 1.4032613039016724, | |
| "learning_rate": 6.793019574868775e-07, | |
| "loss": 1.2812, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.9644845459781148, | |
| "grad_norm": 1.855252981185913, | |
| "learning_rate": 6.513229910027896e-07, | |
| "loss": 1.4774, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.9652524476866962, | |
| "grad_norm": 1.4534977674484253, | |
| "learning_rate": 6.239305198645462e-07, | |
| "loss": 1.097, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 0.9660203493952774, | |
| "grad_norm": 1.4636932611465454, | |
| "learning_rate": 5.971247057809826e-07, | |
| "loss": 1.2459, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.9667882511038587, | |
| "grad_norm": 1.318543553352356, | |
| "learning_rate": 5.709057069976265e-07, | |
| "loss": 1.342, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 0.96755615281244, | |
| "grad_norm": 1.2237138748168945, | |
| "learning_rate": 5.452736782958323e-07, | |
| "loss": 1.0311, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9683240545210213, | |
| "grad_norm": 1.3968307971954346, | |
| "learning_rate": 5.20228770991793e-07, | |
| "loss": 1.2152, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 0.9690919562296026, | |
| "grad_norm": 1.2888987064361572, | |
| "learning_rate": 4.957711329357073e-07, | |
| "loss": 0.9803, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.969859857938184, | |
| "grad_norm": 1.2455861568450928, | |
| "learning_rate": 4.7190090851090274e-07, | |
| "loss": 0.814, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 0.9706277596467652, | |
| "grad_norm": 1.263824462890625, | |
| "learning_rate": 4.4861823863292516e-07, | |
| "loss": 0.9433, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.9713956613553465, | |
| "grad_norm": 1.63260817527771, | |
| "learning_rate": 4.259232607487951e-07, | |
| "loss": 1.4563, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.9721635630639278, | |
| "grad_norm": 1.3201172351837158, | |
| "learning_rate": 4.038161088361192e-07, | |
| "loss": 1.0101, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.9729314647725091, | |
| "grad_norm": 1.456137776374817, | |
| "learning_rate": 3.8229691340234684e-07, | |
| "loss": 1.3377, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 0.9736993664810905, | |
| "grad_norm": 1.9307949542999268, | |
| "learning_rate": 3.613658014839594e-07, | |
| "loss": 1.5946, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.9744672681896718, | |
| "grad_norm": 1.0594274997711182, | |
| "learning_rate": 3.4102289664578177e-07, | |
| "loss": 1.0772, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 0.975235169898253, | |
| "grad_norm": 1.3033182621002197, | |
| "learning_rate": 3.212683189801724e-07, | |
| "loss": 1.2004, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.9760030716068343, | |
| "grad_norm": 1.574930191040039, | |
| "learning_rate": 3.021021851063899e-07, | |
| "loss": 1.266, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 0.9767709733154156, | |
| "grad_norm": 1.5406841039657593, | |
| "learning_rate": 2.8352460816986057e-07, | |
| "loss": 1.0827, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.977538875023997, | |
| "grad_norm": 1.1407910585403442, | |
| "learning_rate": 2.6553569784152357e-07, | |
| "loss": 0.8456, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 0.9783067767325783, | |
| "grad_norm": 1.4787224531173706, | |
| "learning_rate": 2.481355603171531e-07, | |
| "loss": 1.3375, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.9790746784411596, | |
| "grad_norm": 1.514699935913086, | |
| "learning_rate": 2.3132429831682622e-07, | |
| "loss": 1.4017, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.9798425801497408, | |
| "grad_norm": 1.372117519378662, | |
| "learning_rate": 2.1510201108416728e-07, | |
| "loss": 1.1619, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.9806104818583221, | |
| "grad_norm": 1.2351239919662476, | |
| "learning_rate": 1.9946879438592636e-07, | |
| "loss": 0.945, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 0.9813783835669034, | |
| "grad_norm": 1.4396629333496094, | |
| "learning_rate": 1.8442474051125757e-07, | |
| "loss": 0.9939, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.9821462852754848, | |
| "grad_norm": 1.9772976636886597, | |
| "learning_rate": 1.6996993827129715e-07, | |
| "loss": 1.4349, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 0.9829141869840661, | |
| "grad_norm": 1.2860651016235352, | |
| "learning_rate": 1.561044729985861e-07, | |
| "loss": 0.9433, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.9836820886926474, | |
| "grad_norm": 1.496124505996704, | |
| "learning_rate": 1.428284265465596e-07, | |
| "loss": 1.2376, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 0.9844499904012286, | |
| "grad_norm": 1.2294336557388306, | |
| "learning_rate": 1.3014187728906945e-07, | |
| "loss": 0.9644, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.9852178921098099, | |
| "grad_norm": 1.182978630065918, | |
| "learning_rate": 1.1804490011995129e-07, | |
| "loss": 1.2047, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 0.9859857938183912, | |
| "grad_norm": 1.6320767402648926, | |
| "learning_rate": 1.0653756645252477e-07, | |
| "loss": 1.4767, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.9867536955269726, | |
| "grad_norm": 1.482164978981018, | |
| "learning_rate": 9.561994421924958e-08, | |
| "loss": 1.21, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.9875215972355539, | |
| "grad_norm": 1.3324884176254272, | |
| "learning_rate": 8.529209787123682e-08, | |
| "loss": 1.0876, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.9882894989441352, | |
| "grad_norm": 1.530920386314392, | |
| "learning_rate": 7.555408837794931e-08, | |
| "loss": 1.1581, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 0.9890574006527164, | |
| "grad_norm": 1.4899532794952393, | |
| "learning_rate": 6.640597322677967e-08, | |
| "loss": 1.3396, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.9898253023612977, | |
| "grad_norm": 1.2358452081680298, | |
| "learning_rate": 5.784780642275056e-08, | |
| "loss": 1.0957, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 0.990593204069879, | |
| "grad_norm": 1.703139305114746, | |
| "learning_rate": 4.9879638488159465e-08, | |
| "loss": 1.2339, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.9913611057784604, | |
| "grad_norm": 1.3851401805877686, | |
| "learning_rate": 4.2501516462334356e-08, | |
| "loss": 1.4007, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 0.9921290074870417, | |
| "grad_norm": 1.52871572971344, | |
| "learning_rate": 3.5713483901300696e-08, | |
| "loss": 1.3904, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.992896909195623, | |
| "grad_norm": 1.9627262353897095, | |
| "learning_rate": 2.9515580877559346e-08, | |
| "loss": 1.2959, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 0.9936648109042042, | |
| "grad_norm": 2.1106338500976562, | |
| "learning_rate": 2.3907843979831257e-08, | |
| "loss": 1.7045, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.9944327126127855, | |
| "grad_norm": 1.3870468139648438, | |
| "learning_rate": 1.8890306312846495e-08, | |
| "loss": 1.2606, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.9952006143213669, | |
| "grad_norm": 1.4199986457824707, | |
| "learning_rate": 1.446299749716662e-08, | |
| "loss": 1.3145, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.9959685160299482, | |
| "grad_norm": 1.398864984512329, | |
| "learning_rate": 1.0625943668973736e-08, | |
| "loss": 1.3875, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 0.9967364177385295, | |
| "grad_norm": 1.4852126836776733, | |
| "learning_rate": 7.379167479948379e-09, | |
| "loss": 1.1144, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.9975043194471108, | |
| "grad_norm": 1.3623526096343994, | |
| "learning_rate": 4.722688097125172e-09, | |
| "loss": 1.3668, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 0.998272221155692, | |
| "grad_norm": 1.426559329032898, | |
| "learning_rate": 2.656521202770712e-09, | |
| "loss": 1.2487, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.9990401228642733, | |
| "grad_norm": 1.8108794689178467, | |
| "learning_rate": 1.1806789942947484e-09, | |
| "loss": 1.5596, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 0.9998080245728547, | |
| "grad_norm": 1.4854652881622314, | |
| "learning_rate": 2.9517018420577305e-10, | |
| "loss": 1.0427, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 1.000575926281436, | |
| "grad_norm": 2.9019205570220947, | |
| "learning_rate": 0.0, | |
| "loss": 2.1779, | |
| "step": 1303 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1303, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 326, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.5981287804174336e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |