{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 17188, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00011636025133814289, "grad_norm": 2.3270795345306396, "learning_rate": 0.0, "loss": 4.664092063903809, "step": 1 }, { "epoch": 0.00023272050267628578, "grad_norm": 4.2739434242248535, "learning_rate": 6.25e-06, "loss": 5.3881683349609375, "step": 2 }, { "epoch": 0.0003490807540144287, "grad_norm": 4.711091041564941, "learning_rate": 1.25e-05, "loss": 6.04703426361084, "step": 3 }, { "epoch": 0.00046544100535257155, "grad_norm": 3.3446109294891357, "learning_rate": 1.875e-05, "loss": 5.504456520080566, "step": 4 }, { "epoch": 0.0005818012566907144, "grad_norm": 2.297661066055298, "learning_rate": 2.5e-05, "loss": 5.29768180847168, "step": 5 }, { "epoch": 0.0006981615080288574, "grad_norm": 2.3992016315460205, "learning_rate": 3.125e-05, "loss": 4.789924621582031, "step": 6 }, { "epoch": 0.0008145217593670003, "grad_norm": 2.421863079071045, "learning_rate": 3.75e-05, "loss": 5.0406951904296875, "step": 7 }, { "epoch": 0.0009308820107051431, "grad_norm": 2.754455089569092, "learning_rate": 4.375e-05, "loss": 4.6764984130859375, "step": 8 }, { "epoch": 0.001047242262043286, "grad_norm": 2.382047414779663, "learning_rate": 5e-05, "loss": 3.885195255279541, "step": 9 }, { "epoch": 0.0011636025133814289, "grad_norm": 2.376626968383789, "learning_rate": 5.6250000000000005e-05, "loss": 5.107590198516846, "step": 10 }, { "epoch": 0.001279962764719572, "grad_norm": 2.345266103744507, "learning_rate": 6.25e-05, "loss": 4.641767978668213, "step": 11 }, { "epoch": 0.0013963230160577147, "grad_norm": 2.8511507511138916, "learning_rate": 6.875e-05, "loss": 4.798818111419678, "step": 12 }, { "epoch": 0.0015126832673958575, "grad_norm": 2.3071842193603516, "learning_rate": 7.5e-05, "loss": 4.683161735534668, "step": 13 }, { "epoch": 0.0016290435187340006, "grad_norm": 1.6372750997543335, "learning_rate": 8.125000000000001e-05, "loss": 4.592014312744141, "step": 14 }, { "epoch": 0.0017454037700721434, "grad_norm": 2.195841073989868, "learning_rate": 8.75e-05, "loss": 4.196272850036621, "step": 15 }, { "epoch": 0.0018617640214102862, "grad_norm": 2.146094799041748, "learning_rate": 9.375e-05, "loss": 4.769244194030762, "step": 16 }, { "epoch": 0.0019781242727484293, "grad_norm": 1.8519713878631592, "learning_rate": 0.0001, "loss": 4.861962795257568, "step": 17 }, { "epoch": 0.002094484524086572, "grad_norm": 2.474329710006714, "learning_rate": 0.00010625, "loss": 4.463995933532715, "step": 18 }, { "epoch": 0.002210844775424715, "grad_norm": 1.9596589803695679, "learning_rate": 0.00011250000000000001, "loss": 4.437412261962891, "step": 19 }, { "epoch": 0.0023272050267628577, "grad_norm": 3.325235605239868, "learning_rate": 0.00011875, "loss": 5.31038761138916, "step": 20 }, { "epoch": 0.0024435652781010005, "grad_norm": 2.658975839614868, "learning_rate": 0.000125, "loss": 4.4767351150512695, "step": 21 }, { "epoch": 0.002559925529439144, "grad_norm": 2.3103139400482178, "learning_rate": 0.00013125000000000002, "loss": 3.5791683197021484, "step": 22 }, { "epoch": 0.0026762857807772866, "grad_norm": 2.120814323425293, "learning_rate": 0.0001375, "loss": 4.207784652709961, "step": 23 }, { "epoch": 0.0027926460321154294, "grad_norm": 1.856181025505066, "learning_rate": 0.00014375, "loss": 4.25034236907959, "step": 24 }, { "epoch": 0.0029090062834535723, "grad_norm": 1.4976545572280884, "learning_rate": 0.00015, "loss": 4.858099460601807, "step": 25 }, { "epoch": 0.003025366534791715, "grad_norm": 1.6332848072052002, "learning_rate": 0.00015625, "loss": 4.276683807373047, "step": 26 }, { "epoch": 0.003141726786129858, "grad_norm": 2.4674794673919678, "learning_rate": 0.00016250000000000002, "loss": 4.386131286621094, "step": 27 }, { "epoch": 0.003258087037468001, "grad_norm": 1.944339632987976, "learning_rate": 0.00016875, "loss": 4.198155879974365, "step": 28 }, { "epoch": 0.003374447288806144, "grad_norm": 1.6072396039962769, "learning_rate": 0.000175, "loss": 4.25054931640625, "step": 29 }, { "epoch": 0.003490807540144287, "grad_norm": 2.2797884941101074, "learning_rate": 0.00018125, "loss": 3.4528462886810303, "step": 30 }, { "epoch": 0.0036071677914824296, "grad_norm": 2.28269624710083, "learning_rate": 0.0001875, "loss": 4.127889633178711, "step": 31 }, { "epoch": 0.0037235280428205724, "grad_norm": 1.5867356061935425, "learning_rate": 0.00019375000000000002, "loss": 5.007940292358398, "step": 32 }, { "epoch": 0.0038398882941587152, "grad_norm": 2.033514976501465, "learning_rate": 0.0002, "loss": 3.7485954761505127, "step": 33 }, { "epoch": 0.0039562485454968585, "grad_norm": 1.6474937200546265, "learning_rate": 0.00020625, "loss": 4.287443161010742, "step": 34 }, { "epoch": 0.004072608796835001, "grad_norm": 1.5173230171203613, "learning_rate": 0.0002125, "loss": 4.1473541259765625, "step": 35 }, { "epoch": 0.004188969048173144, "grad_norm": 3.491722345352173, "learning_rate": 0.00021875, "loss": 3.8269336223602295, "step": 36 }, { "epoch": 0.0043053292995112865, "grad_norm": 2.492915391921997, "learning_rate": 0.00022500000000000002, "loss": 4.052104949951172, "step": 37 }, { "epoch": 0.00442168955084943, "grad_norm": 1.8992524147033691, "learning_rate": 0.00023125, "loss": 4.166440963745117, "step": 38 }, { "epoch": 0.004538049802187573, "grad_norm": 1.9156253337860107, "learning_rate": 0.0002375, "loss": 4.084643363952637, "step": 39 }, { "epoch": 0.004654410053525715, "grad_norm": 1.9012948274612427, "learning_rate": 0.00024375, "loss": 3.6210577487945557, "step": 40 }, { "epoch": 0.004770770304863859, "grad_norm": 1.7792022228240967, "learning_rate": 0.00025, "loss": 3.967653751373291, "step": 41 }, { "epoch": 0.004887130556202001, "grad_norm": 1.781522274017334, "learning_rate": 0.00025624999999999997, "loss": 4.16281795501709, "step": 42 }, { "epoch": 0.005003490807540144, "grad_norm": 2.1171393394470215, "learning_rate": 0.00026250000000000004, "loss": 3.7069616317749023, "step": 43 }, { "epoch": 0.005119851058878288, "grad_norm": 2.101541757583618, "learning_rate": 0.00026875, "loss": 4.028191089630127, "step": 44 }, { "epoch": 0.00523621131021643, "grad_norm": 2.3833346366882324, "learning_rate": 0.000275, "loss": 3.8323793411254883, "step": 45 }, { "epoch": 0.005352571561554573, "grad_norm": 1.6171365976333618, "learning_rate": 0.00028125000000000003, "loss": 3.2606394290924072, "step": 46 }, { "epoch": 0.005468931812892716, "grad_norm": 1.901778221130371, "learning_rate": 0.0002875, "loss": 3.8342502117156982, "step": 47 }, { "epoch": 0.005585292064230859, "grad_norm": 1.880468726158142, "learning_rate": 0.00029375, "loss": 3.6157305240631104, "step": 48 }, { "epoch": 0.005701652315569001, "grad_norm": 4.323177337646484, "learning_rate": 0.0003, "loss": 5.339083194732666, "step": 49 }, { "epoch": 0.0058180125669071445, "grad_norm": 2.497553586959839, "learning_rate": 0.00030625000000000004, "loss": 3.39255428314209, "step": 50 }, { "epoch": 0.005934372818245288, "grad_norm": 1.8051750659942627, "learning_rate": 0.0003125, "loss": 3.904367685317993, "step": 51 }, { "epoch": 0.00605073306958343, "grad_norm": 3.2748138904571533, "learning_rate": 0.00031874999999999997, "loss": 3.3434672355651855, "step": 52 }, { "epoch": 0.006167093320921573, "grad_norm": 1.7216389179229736, "learning_rate": 0.00032500000000000004, "loss": 3.80228853225708, "step": 53 }, { "epoch": 0.006283453572259716, "grad_norm": 2.3394062519073486, "learning_rate": 0.00033125, "loss": 4.429328441619873, "step": 54 }, { "epoch": 0.006399813823597859, "grad_norm": 2.408318281173706, "learning_rate": 0.0003375, "loss": 3.7573800086975098, "step": 55 }, { "epoch": 0.006516174074936002, "grad_norm": 2.419487953186035, "learning_rate": 0.00034375, "loss": 3.4948878288269043, "step": 56 }, { "epoch": 0.006632534326274145, "grad_norm": 2.2088427543640137, "learning_rate": 0.00035, "loss": 4.232370376586914, "step": 57 }, { "epoch": 0.006748894577612288, "grad_norm": 1.7455617189407349, "learning_rate": 0.00035625, "loss": 3.908946990966797, "step": 58 }, { "epoch": 0.00686525482895043, "grad_norm": 1.7184514999389648, "learning_rate": 0.0003625, "loss": 3.8188416957855225, "step": 59 }, { "epoch": 0.006981615080288574, "grad_norm": 1.7980962991714478, "learning_rate": 0.00036875000000000005, "loss": 3.27487850189209, "step": 60 }, { "epoch": 0.007097975331626716, "grad_norm": 1.7895852327346802, "learning_rate": 0.000375, "loss": 3.745051622390747, "step": 61 }, { "epoch": 0.007214335582964859, "grad_norm": 2.210503339767456, "learning_rate": 0.00038124999999999997, "loss": 3.8740479946136475, "step": 62 }, { "epoch": 0.0073306958343030025, "grad_norm": 1.5991333723068237, "learning_rate": 0.00038750000000000004, "loss": 3.982822895050049, "step": 63 }, { "epoch": 0.007447056085641145, "grad_norm": 2.17825984954834, "learning_rate": 0.00039375, "loss": 3.9956729412078857, "step": 64 }, { "epoch": 0.007563416336979288, "grad_norm": 1.6471325159072876, "learning_rate": 0.0004, "loss": 4.050259113311768, "step": 65 }, { "epoch": 0.0076797765883174305, "grad_norm": 3.092813730239868, "learning_rate": 0.00040625000000000004, "loss": 3.9066579341888428, "step": 66 }, { "epoch": 0.007796136839655574, "grad_norm": 2.09673810005188, "learning_rate": 0.0004125, "loss": 3.73781681060791, "step": 67 }, { "epoch": 0.007912497090993717, "grad_norm": 1.6157270669937134, "learning_rate": 0.00041875, "loss": 3.2483882904052734, "step": 68 }, { "epoch": 0.00802885734233186, "grad_norm": 1.4974535703659058, "learning_rate": 0.000425, "loss": 4.483404159545898, "step": 69 }, { "epoch": 0.008145217593670002, "grad_norm": 2.725778818130493, "learning_rate": 0.00043125000000000005, "loss": 4.029339790344238, "step": 70 }, { "epoch": 0.008261577845008146, "grad_norm": 1.9228723049163818, "learning_rate": 0.0004375, "loss": 3.688366413116455, "step": 71 }, { "epoch": 0.008377938096346288, "grad_norm": 1.736736536026001, "learning_rate": 0.00044374999999999997, "loss": 4.281065940856934, "step": 72 }, { "epoch": 0.00849429834768443, "grad_norm": 2.0058302879333496, "learning_rate": 0.00045000000000000004, "loss": 4.036936283111572, "step": 73 }, { "epoch": 0.008610658599022573, "grad_norm": 2.336660861968994, "learning_rate": 0.00045625, "loss": 3.5911784172058105, "step": 74 }, { "epoch": 0.008727018850360717, "grad_norm": 1.9611707925796509, "learning_rate": 0.0004625, "loss": 3.6641268730163574, "step": 75 }, { "epoch": 0.00884337910169886, "grad_norm": 1.6011388301849365, "learning_rate": 0.00046875, "loss": 3.7479751110076904, "step": 76 }, { "epoch": 0.008959739353037002, "grad_norm": 2.1128592491149902, "learning_rate": 0.000475, "loss": 3.650224208831787, "step": 77 }, { "epoch": 0.009076099604375146, "grad_norm": 2.4485361576080322, "learning_rate": 0.00048125, "loss": 3.2231438159942627, "step": 78 }, { "epoch": 0.009192459855713288, "grad_norm": 1.850948452949524, "learning_rate": 0.0004875, "loss": 3.732471466064453, "step": 79 }, { "epoch": 0.00930882010705143, "grad_norm": 2.037260055541992, "learning_rate": 0.00049375, "loss": 4.370021343231201, "step": 80 }, { "epoch": 0.009425180358389575, "grad_norm": 2.1212971210479736, "learning_rate": 0.0005, "loss": 3.733323574066162, "step": 81 }, { "epoch": 0.009541540609727717, "grad_norm": 2.578944683074951, "learning_rate": 0.0005, "loss": 3.97304105758667, "step": 82 }, { "epoch": 0.00965790086106586, "grad_norm": 1.7424222230911255, "learning_rate": 0.0005, "loss": 3.1923420429229736, "step": 83 }, { "epoch": 0.009774261112404002, "grad_norm": 2.4231884479522705, "learning_rate": 0.0005, "loss": 3.958127498626709, "step": 84 }, { "epoch": 0.009890621363742146, "grad_norm": 2.073019027709961, "learning_rate": 0.0005, "loss": 3.7765989303588867, "step": 85 }, { "epoch": 0.010006981615080289, "grad_norm": 2.2059502601623535, "learning_rate": 0.0005, "loss": 3.9615578651428223, "step": 86 }, { "epoch": 0.010123341866418431, "grad_norm": 2.192765235900879, "learning_rate": 0.0005, "loss": 3.9804763793945312, "step": 87 }, { "epoch": 0.010239702117756575, "grad_norm": 2.352332830429077, "learning_rate": 0.0005, "loss": 4.077672004699707, "step": 88 }, { "epoch": 0.010356062369094718, "grad_norm": 2.2690980434417725, "learning_rate": 0.0005, "loss": 3.6995906829833984, "step": 89 }, { "epoch": 0.01047242262043286, "grad_norm": 2.408132314682007, "learning_rate": 0.0005, "loss": 3.8778107166290283, "step": 90 }, { "epoch": 0.010588782871771002, "grad_norm": 1.8011057376861572, "learning_rate": 0.0005, "loss": 4.164064407348633, "step": 91 }, { "epoch": 0.010705143123109146, "grad_norm": 1.8717453479766846, "learning_rate": 0.0005, "loss": 3.020747184753418, "step": 92 }, { "epoch": 0.010821503374447289, "grad_norm": 2.1343882083892822, "learning_rate": 0.0005, "loss": 3.626646041870117, "step": 93 }, { "epoch": 0.010937863625785431, "grad_norm": 1.8773666620254517, "learning_rate": 0.0005, "loss": 3.1198172569274902, "step": 94 }, { "epoch": 0.011054223877123575, "grad_norm": 2.233515739440918, "learning_rate": 0.0005, "loss": 4.591129302978516, "step": 95 }, { "epoch": 0.011170584128461718, "grad_norm": 1.9438142776489258, "learning_rate": 0.0005, "loss": 3.991184711456299, "step": 96 }, { "epoch": 0.01128694437979986, "grad_norm": 1.7457503080368042, "learning_rate": 0.0005, "loss": 3.8788294792175293, "step": 97 }, { "epoch": 0.011403304631138002, "grad_norm": 1.5909650325775146, "learning_rate": 0.0005, "loss": 3.3535990715026855, "step": 98 }, { "epoch": 0.011519664882476147, "grad_norm": 1.9996711015701294, "learning_rate": 0.0005, "loss": 3.4325928688049316, "step": 99 }, { "epoch": 0.011636025133814289, "grad_norm": 2.3346333503723145, "learning_rate": 0.0005, "loss": 4.457189083099365, "step": 100 }, { "epoch": 0.011752385385152431, "grad_norm": 2.545396327972412, "learning_rate": 0.0005, "loss": 4.230243682861328, "step": 101 }, { "epoch": 0.011868745636490576, "grad_norm": 2.019620656967163, "learning_rate": 0.0005, "loss": 3.652362823486328, "step": 102 }, { "epoch": 0.011985105887828718, "grad_norm": 2.2532145977020264, "learning_rate": 0.0005, "loss": 4.346138954162598, "step": 103 }, { "epoch": 0.01210146613916686, "grad_norm": 1.9524965286254883, "learning_rate": 0.0005, "loss": 3.7552125453948975, "step": 104 }, { "epoch": 0.012217826390505003, "grad_norm": 1.3078492879867554, "learning_rate": 0.0005, "loss": 4.28438663482666, "step": 105 }, { "epoch": 0.012334186641843147, "grad_norm": 1.8156719207763672, "learning_rate": 0.0005, "loss": 4.123618125915527, "step": 106 }, { "epoch": 0.01245054689318129, "grad_norm": 2.3359599113464355, "learning_rate": 0.0005, "loss": 3.3284788131713867, "step": 107 }, { "epoch": 0.012566907144519432, "grad_norm": 2.7451629638671875, "learning_rate": 0.0005, "loss": 4.593064785003662, "step": 108 }, { "epoch": 0.012683267395857576, "grad_norm": 2.430476427078247, "learning_rate": 0.0005, "loss": 3.0329718589782715, "step": 109 }, { "epoch": 0.012799627647195718, "grad_norm": 2.0515100955963135, "learning_rate": 0.0005, "loss": 3.9768154621124268, "step": 110 }, { "epoch": 0.01291598789853386, "grad_norm": 2.0789408683776855, "learning_rate": 0.0005, "loss": 3.9118361473083496, "step": 111 }, { "epoch": 0.013032348149872005, "grad_norm": 1.8252203464508057, "learning_rate": 0.0005, "loss": 3.8647868633270264, "step": 112 }, { "epoch": 0.013148708401210147, "grad_norm": 1.84706449508667, "learning_rate": 0.0005, "loss": 3.7198469638824463, "step": 113 }, { "epoch": 0.01326506865254829, "grad_norm": 2.5524415969848633, "learning_rate": 0.0005, "loss": 3.868863105773926, "step": 114 }, { "epoch": 0.013381428903886432, "grad_norm": 1.6786142587661743, "learning_rate": 0.0005, "loss": 3.440669059753418, "step": 115 }, { "epoch": 0.013497789155224576, "grad_norm": 2.5217416286468506, "learning_rate": 0.0005, "loss": 4.069984436035156, "step": 116 }, { "epoch": 0.013614149406562718, "grad_norm": 2.040886402130127, "learning_rate": 0.0005, "loss": 4.0957465171813965, "step": 117 }, { "epoch": 0.01373050965790086, "grad_norm": 1.8112701177597046, "learning_rate": 0.0005, "loss": 2.993703842163086, "step": 118 }, { "epoch": 0.013846869909239005, "grad_norm": 1.8502641916275024, "learning_rate": 0.0005, "loss": 3.5402815341949463, "step": 119 }, { "epoch": 0.013963230160577147, "grad_norm": 1.7063482999801636, "learning_rate": 0.0005, "loss": 3.9679503440856934, "step": 120 }, { "epoch": 0.01407959041191529, "grad_norm": 3.936837911605835, "learning_rate": 0.0005, "loss": 3.8166069984436035, "step": 121 }, { "epoch": 0.014195950663253432, "grad_norm": 3.7233831882476807, "learning_rate": 0.0005, "loss": 3.368717670440674, "step": 122 }, { "epoch": 0.014312310914591576, "grad_norm": 2.4355955123901367, "learning_rate": 0.0005, "loss": 3.188882827758789, "step": 123 }, { "epoch": 0.014428671165929718, "grad_norm": 1.9651821851730347, "learning_rate": 0.0005, "loss": 3.9231929779052734, "step": 124 }, { "epoch": 0.01454503141726786, "grad_norm": 2.1829628944396973, "learning_rate": 0.0005, "loss": 2.8831937313079834, "step": 125 }, { "epoch": 0.014661391668606005, "grad_norm": 2.429941415786743, "learning_rate": 0.0005, "loss": 3.7772510051727295, "step": 126 }, { "epoch": 0.014777751919944147, "grad_norm": 2.402390718460083, "learning_rate": 0.0005, "loss": 3.4786787033081055, "step": 127 }, { "epoch": 0.01489411217128229, "grad_norm": 3.7051613330841064, "learning_rate": 0.0005, "loss": 4.706839561462402, "step": 128 }, { "epoch": 0.015010472422620432, "grad_norm": 1.9175933599472046, "learning_rate": 0.0005, "loss": 3.9048757553100586, "step": 129 }, { "epoch": 0.015126832673958576, "grad_norm": 1.553305745124817, "learning_rate": 0.0005, "loss": 3.561711311340332, "step": 130 }, { "epoch": 0.015243192925296719, "grad_norm": 1.4520928859710693, "learning_rate": 0.0005, "loss": 3.175926923751831, "step": 131 }, { "epoch": 0.015359553176634861, "grad_norm": 1.5617194175720215, "learning_rate": 0.0005, "loss": 3.0025200843811035, "step": 132 }, { "epoch": 0.015475913427973005, "grad_norm": 3.1706314086914062, "learning_rate": 0.0005, "loss": 3.0104150772094727, "step": 133 }, { "epoch": 0.015592273679311148, "grad_norm": 1.7094146013259888, "learning_rate": 0.0005, "loss": 4.183825492858887, "step": 134 }, { "epoch": 0.01570863393064929, "grad_norm": 1.7199894189834595, "learning_rate": 0.0005, "loss": 3.42240309715271, "step": 135 }, { "epoch": 0.015824994181987434, "grad_norm": 1.613256812095642, "learning_rate": 0.0005, "loss": 3.6226272583007812, "step": 136 }, { "epoch": 0.015941354433325575, "grad_norm": 1.8295347690582275, "learning_rate": 0.0005, "loss": 4.03933572769165, "step": 137 }, { "epoch": 0.01605771468466372, "grad_norm": 3.048163414001465, "learning_rate": 0.0005, "loss": 3.4586079120635986, "step": 138 }, { "epoch": 0.016174074936001863, "grad_norm": 1.940755844116211, "learning_rate": 0.0005, "loss": 3.5647263526916504, "step": 139 }, { "epoch": 0.016290435187340004, "grad_norm": 1.9861328601837158, "learning_rate": 0.0005, "loss": 3.487105131149292, "step": 140 }, { "epoch": 0.016406795438678148, "grad_norm": 2.647684335708618, "learning_rate": 0.0005, "loss": 3.5664196014404297, "step": 141 }, { "epoch": 0.016523155690016292, "grad_norm": 2.110851764678955, "learning_rate": 0.0005, "loss": 4.029186248779297, "step": 142 }, { "epoch": 0.016639515941354432, "grad_norm": 1.5830297470092773, "learning_rate": 0.0005, "loss": 3.0674257278442383, "step": 143 }, { "epoch": 0.016755876192692577, "grad_norm": 2.455854654312134, "learning_rate": 0.0005, "loss": 3.74483060836792, "step": 144 }, { "epoch": 0.01687223644403072, "grad_norm": 2.164848804473877, "learning_rate": 0.0005, "loss": 2.819584846496582, "step": 145 }, { "epoch": 0.01698859669536886, "grad_norm": 1.633680820465088, "learning_rate": 0.0005, "loss": 3.31184983253479, "step": 146 }, { "epoch": 0.017104956946707005, "grad_norm": 2.655198812484741, "learning_rate": 0.0005, "loss": 4.527023792266846, "step": 147 }, { "epoch": 0.017221317198045146, "grad_norm": 1.681672215461731, "learning_rate": 0.0005, "loss": 3.747264862060547, "step": 148 }, { "epoch": 0.01733767744938329, "grad_norm": 8.31913948059082, "learning_rate": 0.0005, "loss": 4.150959014892578, "step": 149 }, { "epoch": 0.017454037700721434, "grad_norm": 2.008788585662842, "learning_rate": 0.0005, "loss": 4.216483116149902, "step": 150 }, { "epoch": 0.017570397952059575, "grad_norm": 1.5784648656845093, "learning_rate": 0.0005, "loss": 3.8373310565948486, "step": 151 }, { "epoch": 0.01768675820339772, "grad_norm": 1.8064918518066406, "learning_rate": 0.0005, "loss": 3.555936574935913, "step": 152 }, { "epoch": 0.017803118454735863, "grad_norm": 2.3040595054626465, "learning_rate": 0.0005, "loss": 4.041842937469482, "step": 153 }, { "epoch": 0.017919478706074004, "grad_norm": 2.443654775619507, "learning_rate": 0.0005, "loss": 3.460151433944702, "step": 154 }, { "epoch": 0.018035838957412148, "grad_norm": 1.5721330642700195, "learning_rate": 0.0005, "loss": 4.0308732986450195, "step": 155 }, { "epoch": 0.018152199208750292, "grad_norm": 2.076934814453125, "learning_rate": 0.0005, "loss": 3.6663131713867188, "step": 156 }, { "epoch": 0.018268559460088433, "grad_norm": 1.7271630764007568, "learning_rate": 0.0005, "loss": 2.905573606491089, "step": 157 }, { "epoch": 0.018384919711426577, "grad_norm": 5.649632453918457, "learning_rate": 0.0005, "loss": 4.059811592102051, "step": 158 }, { "epoch": 0.01850127996276472, "grad_norm": 2.289559841156006, "learning_rate": 0.0005, "loss": 3.3454155921936035, "step": 159 }, { "epoch": 0.01861764021410286, "grad_norm": 1.90410578250885, "learning_rate": 0.0005, "loss": 3.277409076690674, "step": 160 }, { "epoch": 0.018734000465441006, "grad_norm": 1.7329320907592773, "learning_rate": 0.0005, "loss": 3.6504218578338623, "step": 161 }, { "epoch": 0.01885036071677915, "grad_norm": 1.8936841487884521, "learning_rate": 0.0005, "loss": 3.655198812484741, "step": 162 }, { "epoch": 0.01896672096811729, "grad_norm": 2.771662473678589, "learning_rate": 0.0005, "loss": 3.481839656829834, "step": 163 }, { "epoch": 0.019083081219455435, "grad_norm": 2.1885321140289307, "learning_rate": 0.0005, "loss": 3.6950628757476807, "step": 164 }, { "epoch": 0.019199441470793575, "grad_norm": 1.7044739723205566, "learning_rate": 0.0005, "loss": 4.093265533447266, "step": 165 }, { "epoch": 0.01931580172213172, "grad_norm": 2.5684478282928467, "learning_rate": 0.0005, "loss": 4.113027095794678, "step": 166 }, { "epoch": 0.019432161973469864, "grad_norm": 1.309643268585205, "learning_rate": 0.0005, "loss": 2.995182752609253, "step": 167 }, { "epoch": 0.019548522224808004, "grad_norm": 2.350935697555542, "learning_rate": 0.0005, "loss": 3.6764352321624756, "step": 168 }, { "epoch": 0.01966488247614615, "grad_norm": 2.0914840698242188, "learning_rate": 0.0005, "loss": 4.077611923217773, "step": 169 }, { "epoch": 0.019781242727484293, "grad_norm": 1.530475378036499, "learning_rate": 0.0005, "loss": 2.9924421310424805, "step": 170 }, { "epoch": 0.019897602978822433, "grad_norm": 1.9877121448516846, "learning_rate": 0.0005, "loss": 3.302217960357666, "step": 171 }, { "epoch": 0.020013963230160577, "grad_norm": 3.6492860317230225, "learning_rate": 0.0005, "loss": 3.449582576751709, "step": 172 }, { "epoch": 0.02013032348149872, "grad_norm": 2.205078125, "learning_rate": 0.0005, "loss": 2.713923454284668, "step": 173 }, { "epoch": 0.020246683732836862, "grad_norm": 1.5383967161178589, "learning_rate": 0.0005, "loss": 3.2576491832733154, "step": 174 }, { "epoch": 0.020363043984175006, "grad_norm": 1.6856515407562256, "learning_rate": 0.0005, "loss": 3.7118659019470215, "step": 175 }, { "epoch": 0.02047940423551315, "grad_norm": 2.104196071624756, "learning_rate": 0.0005, "loss": 3.886768341064453, "step": 176 }, { "epoch": 0.02059576448685129, "grad_norm": 2.476850748062134, "learning_rate": 0.0005, "loss": 3.519711494445801, "step": 177 }, { "epoch": 0.020712124738189435, "grad_norm": 1.8402421474456787, "learning_rate": 0.0005, "loss": 3.5248279571533203, "step": 178 }, { "epoch": 0.020828484989527576, "grad_norm": 2.2989280223846436, "learning_rate": 0.0005, "loss": 3.55511736869812, "step": 179 }, { "epoch": 0.02094484524086572, "grad_norm": 2.2887275218963623, "learning_rate": 0.0005, "loss": 3.530306339263916, "step": 180 }, { "epoch": 0.021061205492203864, "grad_norm": 1.8339380025863647, "learning_rate": 0.0005, "loss": 4.2307844161987305, "step": 181 }, { "epoch": 0.021177565743542005, "grad_norm": 1.5380574464797974, "learning_rate": 0.0005, "loss": 3.3386037349700928, "step": 182 }, { "epoch": 0.02129392599488015, "grad_norm": 1.9606529474258423, "learning_rate": 0.0005, "loss": 3.4952609539031982, "step": 183 }, { "epoch": 0.021410286246218293, "grad_norm": 1.992826223373413, "learning_rate": 0.0005, "loss": 2.9490480422973633, "step": 184 }, { "epoch": 0.021526646497556434, "grad_norm": 3.010685443878174, "learning_rate": 0.0005, "loss": 3.116447925567627, "step": 185 }, { "epoch": 0.021643006748894578, "grad_norm": 2.516221284866333, "learning_rate": 0.0005, "loss": 3.307588815689087, "step": 186 }, { "epoch": 0.021759367000232722, "grad_norm": 2.29280161857605, "learning_rate": 0.0005, "loss": 4.00446891784668, "step": 187 }, { "epoch": 0.021875727251570862, "grad_norm": 1.5280998945236206, "learning_rate": 0.0005, "loss": 3.9846413135528564, "step": 188 }, { "epoch": 0.021992087502909007, "grad_norm": 1.3532257080078125, "learning_rate": 0.0005, "loss": 2.9143171310424805, "step": 189 }, { "epoch": 0.02210844775424715, "grad_norm": 2.8569555282592773, "learning_rate": 0.0005, "loss": 3.204385757446289, "step": 190 }, { "epoch": 0.02222480800558529, "grad_norm": 2.016827344894409, "learning_rate": 0.0005, "loss": 2.9477667808532715, "step": 191 }, { "epoch": 0.022341168256923435, "grad_norm": 2.351505756378174, "learning_rate": 0.0005, "loss": 3.692361831665039, "step": 192 }, { "epoch": 0.02245752850826158, "grad_norm": 2.235283613204956, "learning_rate": 0.0005, "loss": 3.3754796981811523, "step": 193 }, { "epoch": 0.02257388875959972, "grad_norm": 1.7794203758239746, "learning_rate": 0.0005, "loss": 3.5726568698883057, "step": 194 }, { "epoch": 0.022690249010937864, "grad_norm": 1.7505621910095215, "learning_rate": 0.0005, "loss": 4.155007839202881, "step": 195 }, { "epoch": 0.022806609262276005, "grad_norm": 2.241335391998291, "learning_rate": 0.0005, "loss": 3.8679096698760986, "step": 196 }, { "epoch": 0.02292296951361415, "grad_norm": 2.0992484092712402, "learning_rate": 0.0005, "loss": 3.6948742866516113, "step": 197 }, { "epoch": 0.023039329764952293, "grad_norm": 2.4658660888671875, "learning_rate": 0.0005, "loss": 3.137082576751709, "step": 198 }, { "epoch": 0.023155690016290434, "grad_norm": 1.620656967163086, "learning_rate": 0.0005, "loss": 3.516057014465332, "step": 199 }, { "epoch": 0.023272050267628578, "grad_norm": 2.3436696529388428, "learning_rate": 0.0005, "loss": 3.076921224594116, "step": 200 }, { "epoch": 0.023388410518966722, "grad_norm": 2.2382054328918457, "learning_rate": 0.0005, "loss": 3.0241057872772217, "step": 201 }, { "epoch": 0.023504770770304863, "grad_norm": 2.0275142192840576, "learning_rate": 0.0005, "loss": 3.478590488433838, "step": 202 }, { "epoch": 0.023621131021643007, "grad_norm": 1.7969965934753418, "learning_rate": 0.0005, "loss": 2.8156936168670654, "step": 203 }, { "epoch": 0.02373749127298115, "grad_norm": 2.538156509399414, "learning_rate": 0.0005, "loss": 3.649552583694458, "step": 204 }, { "epoch": 0.02385385152431929, "grad_norm": 2.4898972511291504, "learning_rate": 0.0005, "loss": 3.7990610599517822, "step": 205 }, { "epoch": 0.023970211775657436, "grad_norm": 1.6089762449264526, "learning_rate": 0.0005, "loss": 3.4810543060302734, "step": 206 }, { "epoch": 0.02408657202699558, "grad_norm": 2.0157508850097656, "learning_rate": 0.0005, "loss": 3.4316680431365967, "step": 207 }, { "epoch": 0.02420293227833372, "grad_norm": 1.937638521194458, "learning_rate": 0.0005, "loss": 3.4051473140716553, "step": 208 }, { "epoch": 0.024319292529671865, "grad_norm": 1.4578909873962402, "learning_rate": 0.0005, "loss": 3.1191747188568115, "step": 209 }, { "epoch": 0.024435652781010005, "grad_norm": 2.4279251098632812, "learning_rate": 0.0005, "loss": 4.347732067108154, "step": 210 }, { "epoch": 0.02455201303234815, "grad_norm": 1.9447154998779297, "learning_rate": 0.0005, "loss": 3.5582425594329834, "step": 211 }, { "epoch": 0.024668373283686294, "grad_norm": 2.1601240634918213, "learning_rate": 0.0005, "loss": 4.176887512207031, "step": 212 }, { "epoch": 0.024784733535024434, "grad_norm": 1.8674569129943848, "learning_rate": 0.0005, "loss": 3.361980438232422, "step": 213 }, { "epoch": 0.02490109378636258, "grad_norm": 2.1197094917297363, "learning_rate": 0.0005, "loss": 3.4261906147003174, "step": 214 }, { "epoch": 0.025017454037700722, "grad_norm": 2.0017499923706055, "learning_rate": 0.0005, "loss": 3.0080409049987793, "step": 215 }, { "epoch": 0.025133814289038863, "grad_norm": 1.377832055091858, "learning_rate": 0.0005, "loss": 3.544893741607666, "step": 216 }, { "epoch": 0.025250174540377007, "grad_norm": 1.4383478164672852, "learning_rate": 0.0005, "loss": 3.661233425140381, "step": 217 }, { "epoch": 0.02536653479171515, "grad_norm": 2.0526912212371826, "learning_rate": 0.0005, "loss": 4.077496528625488, "step": 218 }, { "epoch": 0.025482895043053292, "grad_norm": 1.6757892370224, "learning_rate": 0.0005, "loss": 3.9890756607055664, "step": 219 }, { "epoch": 0.025599255294391436, "grad_norm": 1.5053119659423828, "learning_rate": 0.0005, "loss": 4.068739891052246, "step": 220 }, { "epoch": 0.02571561554572958, "grad_norm": 1.7343138456344604, "learning_rate": 0.0005, "loss": 3.3113672733306885, "step": 221 }, { "epoch": 0.02583197579706772, "grad_norm": 1.9327360391616821, "learning_rate": 0.0005, "loss": 3.5671095848083496, "step": 222 }, { "epoch": 0.025948336048405865, "grad_norm": 1.7923086881637573, "learning_rate": 0.0005, "loss": 3.713487148284912, "step": 223 }, { "epoch": 0.02606469629974401, "grad_norm": 1.935158133506775, "learning_rate": 0.0005, "loss": 3.885714530944824, "step": 224 }, { "epoch": 0.02618105655108215, "grad_norm": 1.4025973081588745, "learning_rate": 0.0005, "loss": 3.379624605178833, "step": 225 }, { "epoch": 0.026297416802420294, "grad_norm": 1.3266505002975464, "learning_rate": 0.0005, "loss": 2.5918965339660645, "step": 226 }, { "epoch": 0.026413777053758435, "grad_norm": 1.7964364290237427, "learning_rate": 0.0005, "loss": 3.7153799533843994, "step": 227 }, { "epoch": 0.02653013730509658, "grad_norm": 1.582588791847229, "learning_rate": 0.0005, "loss": 2.7141780853271484, "step": 228 }, { "epoch": 0.026646497556434723, "grad_norm": 1.4241458177566528, "learning_rate": 0.0005, "loss": 3.6072163581848145, "step": 229 }, { "epoch": 0.026762857807772863, "grad_norm": 1.891135811805725, "learning_rate": 0.0005, "loss": 3.3247642517089844, "step": 230 }, { "epoch": 0.026879218059111008, "grad_norm": 2.0601117610931396, "learning_rate": 0.0005, "loss": 3.782503128051758, "step": 231 }, { "epoch": 0.026995578310449152, "grad_norm": 1.60047447681427, "learning_rate": 0.0005, "loss": 3.1661689281463623, "step": 232 }, { "epoch": 0.027111938561787292, "grad_norm": 1.400134801864624, "learning_rate": 0.0005, "loss": 3.164323329925537, "step": 233 }, { "epoch": 0.027228298813125437, "grad_norm": 2.005354881286621, "learning_rate": 0.0005, "loss": 4.210352420806885, "step": 234 }, { "epoch": 0.02734465906446358, "grad_norm": 2.0026698112487793, "learning_rate": 0.0005, "loss": 3.349792957305908, "step": 235 }, { "epoch": 0.02746101931580172, "grad_norm": 2.3687798976898193, "learning_rate": 0.0005, "loss": 3.2963531017303467, "step": 236 }, { "epoch": 0.027577379567139865, "grad_norm": 2.1688194274902344, "learning_rate": 0.0005, "loss": 3.5364627838134766, "step": 237 }, { "epoch": 0.02769373981847801, "grad_norm": 1.503375768661499, "learning_rate": 0.0005, "loss": 2.743271827697754, "step": 238 }, { "epoch": 0.02781010006981615, "grad_norm": 4.40216588973999, "learning_rate": 0.0005, "loss": 3.852424144744873, "step": 239 }, { "epoch": 0.027926460321154294, "grad_norm": 2.05436635017395, "learning_rate": 0.0005, "loss": 3.396801471710205, "step": 240 }, { "epoch": 0.028042820572492435, "grad_norm": 1.4515416622161865, "learning_rate": 0.0005, "loss": 3.335935115814209, "step": 241 }, { "epoch": 0.02815918082383058, "grad_norm": 1.5833300352096558, "learning_rate": 0.0005, "loss": 3.0267014503479004, "step": 242 }, { "epoch": 0.028275541075168723, "grad_norm": 2.060011625289917, "learning_rate": 0.0005, "loss": 3.5615031719207764, "step": 243 }, { "epoch": 0.028391901326506864, "grad_norm": 1.668498158454895, "learning_rate": 0.0005, "loss": 3.663360357284546, "step": 244 }, { "epoch": 0.028508261577845008, "grad_norm": 2.0780112743377686, "learning_rate": 0.0005, "loss": 2.763594388961792, "step": 245 }, { "epoch": 0.028624621829183152, "grad_norm": 1.6862883567810059, "learning_rate": 0.0005, "loss": 3.2206618785858154, "step": 246 }, { "epoch": 0.028740982080521293, "grad_norm": 1.98220694065094, "learning_rate": 0.0005, "loss": 3.3943674564361572, "step": 247 }, { "epoch": 0.028857342331859437, "grad_norm": 1.8367375135421753, "learning_rate": 0.0005, "loss": 3.56044864654541, "step": 248 }, { "epoch": 0.02897370258319758, "grad_norm": 2.1576037406921387, "learning_rate": 0.0005, "loss": 4.121304035186768, "step": 249 }, { "epoch": 0.02909006283453572, "grad_norm": 1.8120437860488892, "learning_rate": 0.0005, "loss": 3.2609028816223145, "step": 250 }, { "epoch": 0.029206423085873866, "grad_norm": 2.323681116104126, "learning_rate": 0.0005, "loss": 2.8932785987854004, "step": 251 }, { "epoch": 0.02932278333721201, "grad_norm": 1.5645674467086792, "learning_rate": 0.0005, "loss": 3.400346279144287, "step": 252 }, { "epoch": 0.02943914358855015, "grad_norm": 2.400132179260254, "learning_rate": 0.0005, "loss": 4.239377498626709, "step": 253 }, { "epoch": 0.029555503839888295, "grad_norm": 1.6156092882156372, "learning_rate": 0.0005, "loss": 3.8542375564575195, "step": 254 }, { "epoch": 0.02967186409122644, "grad_norm": 1.5001672506332397, "learning_rate": 0.0005, "loss": 2.8913474082946777, "step": 255 }, { "epoch": 0.02978822434256458, "grad_norm": 1.4340436458587646, "learning_rate": 0.0005, "loss": 2.8856024742126465, "step": 256 }, { "epoch": 0.029904584593902724, "grad_norm": 1.7479887008666992, "learning_rate": 0.0005, "loss": 2.9978580474853516, "step": 257 }, { "epoch": 0.030020944845240864, "grad_norm": 1.4569412469863892, "learning_rate": 0.0005, "loss": 3.5454962253570557, "step": 258 }, { "epoch": 0.03013730509657901, "grad_norm": 1.8081955909729004, "learning_rate": 0.0005, "loss": 3.369647264480591, "step": 259 }, { "epoch": 0.030253665347917152, "grad_norm": 2.0181567668914795, "learning_rate": 0.0005, "loss": 3.241596221923828, "step": 260 }, { "epoch": 0.030370025599255293, "grad_norm": 1.6037163734436035, "learning_rate": 0.0005, "loss": 2.5720067024230957, "step": 261 }, { "epoch": 0.030486385850593437, "grad_norm": 1.661336898803711, "learning_rate": 0.0005, "loss": 3.6325578689575195, "step": 262 }, { "epoch": 0.03060274610193158, "grad_norm": 1.8369412422180176, "learning_rate": 0.0005, "loss": 2.87227463722229, "step": 263 }, { "epoch": 0.030719106353269722, "grad_norm": 3.7766435146331787, "learning_rate": 0.0005, "loss": 3.642364501953125, "step": 264 }, { "epoch": 0.030835466604607866, "grad_norm": 2.163064956665039, "learning_rate": 0.0005, "loss": 3.5229129791259766, "step": 265 }, { "epoch": 0.03095182685594601, "grad_norm": 2.2613253593444824, "learning_rate": 0.0005, "loss": 3.473998546600342, "step": 266 }, { "epoch": 0.03106818710728415, "grad_norm": 2.20828914642334, "learning_rate": 0.0005, "loss": 3.5958282947540283, "step": 267 }, { "epoch": 0.031184547358622295, "grad_norm": 1.9009125232696533, "learning_rate": 0.0005, "loss": 3.3943991661071777, "step": 268 }, { "epoch": 0.03130090760996044, "grad_norm": 1.9757460355758667, "learning_rate": 0.0005, "loss": 3.1702382564544678, "step": 269 }, { "epoch": 0.03141726786129858, "grad_norm": 1.6446518898010254, "learning_rate": 0.0005, "loss": 4.146110534667969, "step": 270 }, { "epoch": 0.03153362811263672, "grad_norm": 1.7341471910476685, "learning_rate": 0.0005, "loss": 3.7601895332336426, "step": 271 }, { "epoch": 0.03164998836397487, "grad_norm": 1.950677752494812, "learning_rate": 0.0005, "loss": 3.0910911560058594, "step": 272 }, { "epoch": 0.03176634861531301, "grad_norm": 2.335789680480957, "learning_rate": 0.0005, "loss": 3.029651165008545, "step": 273 }, { "epoch": 0.03188270886665115, "grad_norm": 1.9113010168075562, "learning_rate": 0.0005, "loss": 3.3589234352111816, "step": 274 }, { "epoch": 0.0319990691179893, "grad_norm": 1.6719666719436646, "learning_rate": 0.0005, "loss": 3.016921043395996, "step": 275 }, { "epoch": 0.03211542936932744, "grad_norm": 1.834572672843933, "learning_rate": 0.0005, "loss": 3.4725661277770996, "step": 276 }, { "epoch": 0.03223178962066558, "grad_norm": 2.3888392448425293, "learning_rate": 0.0005, "loss": 3.9266915321350098, "step": 277 }, { "epoch": 0.032348149872003726, "grad_norm": 1.7387276887893677, "learning_rate": 0.0005, "loss": 3.2578535079956055, "step": 278 }, { "epoch": 0.032464510123341866, "grad_norm": 1.6121336221694946, "learning_rate": 0.0005, "loss": 3.284609794616699, "step": 279 }, { "epoch": 0.03258087037468001, "grad_norm": 2.325873613357544, "learning_rate": 0.0005, "loss": 3.570427417755127, "step": 280 }, { "epoch": 0.032697230626018155, "grad_norm": 1.9467570781707764, "learning_rate": 0.0005, "loss": 3.969043493270874, "step": 281 }, { "epoch": 0.032813590877356295, "grad_norm": 2.9408328533172607, "learning_rate": 0.0005, "loss": 3.307749032974243, "step": 282 }, { "epoch": 0.032929951128694436, "grad_norm": 2.0161333084106445, "learning_rate": 0.0005, "loss": 3.6865768432617188, "step": 283 }, { "epoch": 0.033046311380032584, "grad_norm": 1.6475714445114136, "learning_rate": 0.0005, "loss": 2.2002439498901367, "step": 284 }, { "epoch": 0.033162671631370724, "grad_norm": 2.237626075744629, "learning_rate": 0.0005, "loss": 3.165590286254883, "step": 285 }, { "epoch": 0.033279031882708865, "grad_norm": 2.0483765602111816, "learning_rate": 0.0005, "loss": 3.245053768157959, "step": 286 }, { "epoch": 0.03339539213404701, "grad_norm": 1.888586163520813, "learning_rate": 0.0005, "loss": 3.2045059204101562, "step": 287 }, { "epoch": 0.03351175238538515, "grad_norm": 1.416321873664856, "learning_rate": 0.0005, "loss": 3.52538800239563, "step": 288 }, { "epoch": 0.033628112636723294, "grad_norm": 1.7762115001678467, "learning_rate": 0.0005, "loss": 3.1291518211364746, "step": 289 }, { "epoch": 0.03374447288806144, "grad_norm": 1.6938666105270386, "learning_rate": 0.0005, "loss": 4.047937393188477, "step": 290 }, { "epoch": 0.03386083313939958, "grad_norm": 1.8781445026397705, "learning_rate": 0.0005, "loss": 3.5358128547668457, "step": 291 }, { "epoch": 0.03397719339073772, "grad_norm": 1.561174988746643, "learning_rate": 0.0005, "loss": 2.7117204666137695, "step": 292 }, { "epoch": 0.03409355364207587, "grad_norm": 1.695386290550232, "learning_rate": 0.0005, "loss": 3.3731942176818848, "step": 293 }, { "epoch": 0.03420991389341401, "grad_norm": 1.6718072891235352, "learning_rate": 0.0005, "loss": 2.835005760192871, "step": 294 }, { "epoch": 0.03432627414475215, "grad_norm": 1.5992778539657593, "learning_rate": 0.0005, "loss": 2.6825766563415527, "step": 295 }, { "epoch": 0.03444263439609029, "grad_norm": 1.7454643249511719, "learning_rate": 0.0005, "loss": 3.166196346282959, "step": 296 }, { "epoch": 0.03455899464742844, "grad_norm": 1.6452158689498901, "learning_rate": 0.0005, "loss": 3.5247769355773926, "step": 297 }, { "epoch": 0.03467535489876658, "grad_norm": 1.7459310293197632, "learning_rate": 0.0005, "loss": 3.4689934253692627, "step": 298 }, { "epoch": 0.03479171515010472, "grad_norm": 1.6518104076385498, "learning_rate": 0.0005, "loss": 2.5621025562286377, "step": 299 }, { "epoch": 0.03490807540144287, "grad_norm": 1.551615595817566, "learning_rate": 0.0005, "loss": 3.390077590942383, "step": 300 }, { "epoch": 0.03502443565278101, "grad_norm": 1.451642394065857, "learning_rate": 0.0005, "loss": 2.855961799621582, "step": 301 }, { "epoch": 0.03514079590411915, "grad_norm": 1.7851136922836304, "learning_rate": 0.0005, "loss": 3.6936700344085693, "step": 302 }, { "epoch": 0.0352571561554573, "grad_norm": 1.6012567281723022, "learning_rate": 0.0005, "loss": 3.787328004837036, "step": 303 }, { "epoch": 0.03537351640679544, "grad_norm": 2.1887619495391846, "learning_rate": 0.0005, "loss": 2.954280138015747, "step": 304 }, { "epoch": 0.03548987665813358, "grad_norm": 1.3418368101119995, "learning_rate": 0.0005, "loss": 3.0066609382629395, "step": 305 }, { "epoch": 0.03560623690947173, "grad_norm": 1.5297883749008179, "learning_rate": 0.0005, "loss": 3.8918185234069824, "step": 306 }, { "epoch": 0.03572259716080987, "grad_norm": 1.3925807476043701, "learning_rate": 0.0005, "loss": 3.66349720954895, "step": 307 }, { "epoch": 0.03583895741214801, "grad_norm": 1.429685354232788, "learning_rate": 0.0005, "loss": 3.9246230125427246, "step": 308 }, { "epoch": 0.035955317663486155, "grad_norm": 2.2384862899780273, "learning_rate": 0.0005, "loss": 2.994626045227051, "step": 309 }, { "epoch": 0.036071677914824296, "grad_norm": 1.8900842666625977, "learning_rate": 0.0005, "loss": 3.0729317665100098, "step": 310 }, { "epoch": 0.03618803816616244, "grad_norm": 1.865025520324707, "learning_rate": 0.0005, "loss": 3.716552257537842, "step": 311 }, { "epoch": 0.036304398417500584, "grad_norm": 3.327409029006958, "learning_rate": 0.0005, "loss": 3.4357995986938477, "step": 312 }, { "epoch": 0.036420758668838725, "grad_norm": 2.056077241897583, "learning_rate": 0.0005, "loss": 3.618192195892334, "step": 313 }, { "epoch": 0.036537118920176866, "grad_norm": 1.6302822828292847, "learning_rate": 0.0005, "loss": 3.540818214416504, "step": 314 }, { "epoch": 0.03665347917151501, "grad_norm": 1.7387845516204834, "learning_rate": 0.0005, "loss": 2.443150043487549, "step": 315 }, { "epoch": 0.036769839422853154, "grad_norm": 1.0551483631134033, "learning_rate": 0.0005, "loss": 1.8818137645721436, "step": 316 }, { "epoch": 0.036886199674191295, "grad_norm": 2.329772472381592, "learning_rate": 0.0005, "loss": 2.565884590148926, "step": 317 }, { "epoch": 0.03700255992552944, "grad_norm": 1.3989498615264893, "learning_rate": 0.0005, "loss": 2.847046136856079, "step": 318 }, { "epoch": 0.03711892017686758, "grad_norm": 2.0156683921813965, "learning_rate": 0.0005, "loss": 3.371687412261963, "step": 319 }, { "epoch": 0.03723528042820572, "grad_norm": 2.0975027084350586, "learning_rate": 0.0005, "loss": 2.6371710300445557, "step": 320 }, { "epoch": 0.03735164067954387, "grad_norm": 1.4888262748718262, "learning_rate": 0.0005, "loss": 3.117061138153076, "step": 321 }, { "epoch": 0.03746800093088201, "grad_norm": 2.8325681686401367, "learning_rate": 0.0005, "loss": 3.720219373703003, "step": 322 }, { "epoch": 0.03758436118222015, "grad_norm": 1.840483546257019, "learning_rate": 0.0005, "loss": 4.04099702835083, "step": 323 }, { "epoch": 0.0377007214335583, "grad_norm": 2.3202757835388184, "learning_rate": 0.0005, "loss": 3.292710781097412, "step": 324 }, { "epoch": 0.03781708168489644, "grad_norm": 1.9127817153930664, "learning_rate": 0.0005, "loss": 2.0486133098602295, "step": 325 }, { "epoch": 0.03793344193623458, "grad_norm": 2.6126456260681152, "learning_rate": 0.0005, "loss": 3.6742546558380127, "step": 326 }, { "epoch": 0.03804980218757272, "grad_norm": 2.4446022510528564, "learning_rate": 0.0005, "loss": 3.8946831226348877, "step": 327 }, { "epoch": 0.03816616243891087, "grad_norm": 3.4044363498687744, "learning_rate": 0.0005, "loss": 3.8770015239715576, "step": 328 }, { "epoch": 0.03828252269024901, "grad_norm": 1.8986557722091675, "learning_rate": 0.0005, "loss": 4.155683994293213, "step": 329 }, { "epoch": 0.03839888294158715, "grad_norm": 1.718415379524231, "learning_rate": 0.0005, "loss": 2.9904794692993164, "step": 330 }, { "epoch": 0.0385152431929253, "grad_norm": 2.6138546466827393, "learning_rate": 0.0005, "loss": 2.811829090118408, "step": 331 }, { "epoch": 0.03863160344426344, "grad_norm": 2.844193935394287, "learning_rate": 0.0005, "loss": 4.213522911071777, "step": 332 }, { "epoch": 0.03874796369560158, "grad_norm": 1.7184765338897705, "learning_rate": 0.0005, "loss": 2.8187813758850098, "step": 333 }, { "epoch": 0.03886432394693973, "grad_norm": 1.5009312629699707, "learning_rate": 0.0005, "loss": 3.3561434745788574, "step": 334 }, { "epoch": 0.03898068419827787, "grad_norm": 1.2533422708511353, "learning_rate": 0.0005, "loss": 2.3871712684631348, "step": 335 }, { "epoch": 0.03909704444961601, "grad_norm": 1.5095778703689575, "learning_rate": 0.0005, "loss": 3.1033082008361816, "step": 336 }, { "epoch": 0.039213404700954156, "grad_norm": 2.2524242401123047, "learning_rate": 0.0005, "loss": 3.5060369968414307, "step": 337 }, { "epoch": 0.0393297649522923, "grad_norm": 2.3214504718780518, "learning_rate": 0.0005, "loss": 3.470369338989258, "step": 338 }, { "epoch": 0.03944612520363044, "grad_norm": 1.8249326944351196, "learning_rate": 0.0005, "loss": 3.2349112033843994, "step": 339 }, { "epoch": 0.039562485454968585, "grad_norm": 1.7378160953521729, "learning_rate": 0.0005, "loss": 3.114673614501953, "step": 340 }, { "epoch": 0.039678845706306726, "grad_norm": 1.6449459791183472, "learning_rate": 0.0005, "loss": 3.197432041168213, "step": 341 }, { "epoch": 0.039795205957644866, "grad_norm": 1.4831427335739136, "learning_rate": 0.0005, "loss": 3.2968220710754395, "step": 342 }, { "epoch": 0.039911566208983014, "grad_norm": 2.355278730392456, "learning_rate": 0.0005, "loss": 3.749598979949951, "step": 343 }, { "epoch": 0.040027926460321155, "grad_norm": 1.8098030090332031, "learning_rate": 0.0005, "loss": 3.2660388946533203, "step": 344 }, { "epoch": 0.040144286711659295, "grad_norm": 1.410232663154602, "learning_rate": 0.0005, "loss": 3.371715784072876, "step": 345 }, { "epoch": 0.04026064696299744, "grad_norm": 2.131173849105835, "learning_rate": 0.0005, "loss": 3.6835319995880127, "step": 346 }, { "epoch": 0.040377007214335583, "grad_norm": 3.0996365547180176, "learning_rate": 0.0005, "loss": 3.58111572265625, "step": 347 }, { "epoch": 0.040493367465673724, "grad_norm": 1.8829933404922485, "learning_rate": 0.0005, "loss": 3.056480884552002, "step": 348 }, { "epoch": 0.04060972771701187, "grad_norm": 2.1676464080810547, "learning_rate": 0.0005, "loss": 3.4561870098114014, "step": 349 }, { "epoch": 0.04072608796835001, "grad_norm": 1.8915196657180786, "learning_rate": 0.0005, "loss": 3.5927491188049316, "step": 350 }, { "epoch": 0.04084244821968815, "grad_norm": 1.7564036846160889, "learning_rate": 0.0005, "loss": 3.7707347869873047, "step": 351 }, { "epoch": 0.0409588084710263, "grad_norm": 2.0689449310302734, "learning_rate": 0.0005, "loss": 3.181535005569458, "step": 352 }, { "epoch": 0.04107516872236444, "grad_norm": 1.860761046409607, "learning_rate": 0.0005, "loss": 3.5962235927581787, "step": 353 }, { "epoch": 0.04119152897370258, "grad_norm": 1.75383722782135, "learning_rate": 0.0005, "loss": 4.2043914794921875, "step": 354 }, { "epoch": 0.04130788922504073, "grad_norm": 1.9676024913787842, "learning_rate": 0.0005, "loss": 2.3361310958862305, "step": 355 }, { "epoch": 0.04142424947637887, "grad_norm": 1.9723052978515625, "learning_rate": 0.0005, "loss": 3.9214138984680176, "step": 356 }, { "epoch": 0.04154060972771701, "grad_norm": 1.4494774341583252, "learning_rate": 0.0005, "loss": 2.928427219390869, "step": 357 }, { "epoch": 0.04165696997905515, "grad_norm": 2.0751025676727295, "learning_rate": 0.0005, "loss": 3.074841260910034, "step": 358 }, { "epoch": 0.0417733302303933, "grad_norm": 2.3734183311462402, "learning_rate": 0.0005, "loss": 3.723914861679077, "step": 359 }, { "epoch": 0.04188969048173144, "grad_norm": 2.171003818511963, "learning_rate": 0.0005, "loss": 2.9220473766326904, "step": 360 }, { "epoch": 0.04200605073306958, "grad_norm": 1.7370340824127197, "learning_rate": 0.0005, "loss": 3.754694938659668, "step": 361 }, { "epoch": 0.04212241098440773, "grad_norm": 2.344961643218994, "learning_rate": 0.0005, "loss": 3.1057097911834717, "step": 362 }, { "epoch": 0.04223877123574587, "grad_norm": 1.6656569242477417, "learning_rate": 0.0005, "loss": 2.633056640625, "step": 363 }, { "epoch": 0.04235513148708401, "grad_norm": 2.285907030105591, "learning_rate": 0.0005, "loss": 3.4979796409606934, "step": 364 }, { "epoch": 0.04247149173842216, "grad_norm": 1.5720155239105225, "learning_rate": 0.0005, "loss": 2.9646077156066895, "step": 365 }, { "epoch": 0.0425878519897603, "grad_norm": 2.2808997631073, "learning_rate": 0.0005, "loss": 2.902860641479492, "step": 366 }, { "epoch": 0.04270421224109844, "grad_norm": 2.3016817569732666, "learning_rate": 0.0005, "loss": 3.0591742992401123, "step": 367 }, { "epoch": 0.042820572492436586, "grad_norm": 1.772385835647583, "learning_rate": 0.0005, "loss": 3.0446410179138184, "step": 368 }, { "epoch": 0.042936932743774726, "grad_norm": 2.6846063137054443, "learning_rate": 0.0005, "loss": 3.541226387023926, "step": 369 }, { "epoch": 0.04305329299511287, "grad_norm": 1.3265177011489868, "learning_rate": 0.0005, "loss": 3.103463649749756, "step": 370 }, { "epoch": 0.043169653246451015, "grad_norm": 1.6230732202529907, "learning_rate": 0.0005, "loss": 2.785961866378784, "step": 371 }, { "epoch": 0.043286013497789155, "grad_norm": 3.099057197570801, "learning_rate": 0.0005, "loss": 3.6006219387054443, "step": 372 }, { "epoch": 0.043402373749127296, "grad_norm": 2.5798728466033936, "learning_rate": 0.0005, "loss": 3.201540946960449, "step": 373 }, { "epoch": 0.043518734000465444, "grad_norm": 2.022095203399658, "learning_rate": 0.0005, "loss": 4.235955238342285, "step": 374 }, { "epoch": 0.043635094251803584, "grad_norm": 1.8526825904846191, "learning_rate": 0.0005, "loss": 3.23081111907959, "step": 375 }, { "epoch": 0.043751454503141725, "grad_norm": 1.549649715423584, "learning_rate": 0.0005, "loss": 3.3168485164642334, "step": 376 }, { "epoch": 0.04386781475447987, "grad_norm": 2.3796887397766113, "learning_rate": 0.0005, "loss": 3.9643049240112305, "step": 377 }, { "epoch": 0.04398417500581801, "grad_norm": 1.9559346437454224, "learning_rate": 0.0005, "loss": 3.827578067779541, "step": 378 }, { "epoch": 0.044100535257156154, "grad_norm": 2.627467632293701, "learning_rate": 0.0005, "loss": 2.956955909729004, "step": 379 }, { "epoch": 0.0442168955084943, "grad_norm": 1.6777039766311646, "learning_rate": 0.0005, "loss": 3.5659215450286865, "step": 380 }, { "epoch": 0.04433325575983244, "grad_norm": 1.725409746170044, "learning_rate": 0.0005, "loss": 4.138806343078613, "step": 381 }, { "epoch": 0.04444961601117058, "grad_norm": 1.8499876260757446, "learning_rate": 0.0005, "loss": 4.190670013427734, "step": 382 }, { "epoch": 0.04456597626250873, "grad_norm": 1.4864085912704468, "learning_rate": 0.0005, "loss": 3.5968177318573, "step": 383 }, { "epoch": 0.04468233651384687, "grad_norm": 2.5939066410064697, "learning_rate": 0.0005, "loss": 3.53652286529541, "step": 384 }, { "epoch": 0.04479869676518501, "grad_norm": 1.6897478103637695, "learning_rate": 0.0005, "loss": 2.903923988342285, "step": 385 }, { "epoch": 0.04491505701652316, "grad_norm": 2.2688419818878174, "learning_rate": 0.0005, "loss": 2.8996009826660156, "step": 386 }, { "epoch": 0.0450314172678613, "grad_norm": 1.5222731828689575, "learning_rate": 0.0005, "loss": 3.394651412963867, "step": 387 }, { "epoch": 0.04514777751919944, "grad_norm": 0.9576128125190735, "learning_rate": 0.0005, "loss": 1.6515159606933594, "step": 388 }, { "epoch": 0.04526413777053758, "grad_norm": 1.3553264141082764, "learning_rate": 0.0005, "loss": 3.4357588291168213, "step": 389 }, { "epoch": 0.04538049802187573, "grad_norm": 1.5127384662628174, "learning_rate": 0.0005, "loss": 2.814365863800049, "step": 390 }, { "epoch": 0.04549685827321387, "grad_norm": 1.7682687044143677, "learning_rate": 0.0005, "loss": 3.7842488288879395, "step": 391 }, { "epoch": 0.04561321852455201, "grad_norm": 1.9444671869277954, "learning_rate": 0.0005, "loss": 3.6807122230529785, "step": 392 }, { "epoch": 0.04572957877589016, "grad_norm": 1.6389590501785278, "learning_rate": 0.0005, "loss": 3.048198699951172, "step": 393 }, { "epoch": 0.0458459390272283, "grad_norm": 1.7809596061706543, "learning_rate": 0.0005, "loss": 3.203494071960449, "step": 394 }, { "epoch": 0.04596229927856644, "grad_norm": 2.208263397216797, "learning_rate": 0.0005, "loss": 3.5481247901916504, "step": 395 }, { "epoch": 0.046078659529904586, "grad_norm": 1.5472679138183594, "learning_rate": 0.0005, "loss": 3.183647632598877, "step": 396 }, { "epoch": 0.04619501978124273, "grad_norm": 2.460883855819702, "learning_rate": 0.0005, "loss": 2.895902395248413, "step": 397 }, { "epoch": 0.04631138003258087, "grad_norm": 1.951831579208374, "learning_rate": 0.0005, "loss": 3.2309730052948, "step": 398 }, { "epoch": 0.046427740283919015, "grad_norm": 2.228562116622925, "learning_rate": 0.0005, "loss": 3.1395068168640137, "step": 399 }, { "epoch": 0.046544100535257156, "grad_norm": 1.788228154182434, "learning_rate": 0.0005, "loss": 3.042717218399048, "step": 400 }, { "epoch": 0.0466604607865953, "grad_norm": 1.8629422187805176, "learning_rate": 0.0005, "loss": 3.7521958351135254, "step": 401 }, { "epoch": 0.046776821037933444, "grad_norm": 1.667325496673584, "learning_rate": 0.0005, "loss": 3.0018463134765625, "step": 402 }, { "epoch": 0.046893181289271585, "grad_norm": 1.604515790939331, "learning_rate": 0.0005, "loss": 3.252298355102539, "step": 403 }, { "epoch": 0.047009541540609726, "grad_norm": 2.317946672439575, "learning_rate": 0.0005, "loss": 3.5539016723632812, "step": 404 }, { "epoch": 0.04712590179194787, "grad_norm": 1.5454347133636475, "learning_rate": 0.0005, "loss": 3.019568920135498, "step": 405 }, { "epoch": 0.047242262043286014, "grad_norm": 2.6272056102752686, "learning_rate": 0.0005, "loss": 2.9907007217407227, "step": 406 }, { "epoch": 0.047358622294624154, "grad_norm": 2.150782585144043, "learning_rate": 0.0005, "loss": 4.156610488891602, "step": 407 }, { "epoch": 0.0474749825459623, "grad_norm": 1.6859955787658691, "learning_rate": 0.0005, "loss": 3.2569713592529297, "step": 408 }, { "epoch": 0.04759134279730044, "grad_norm": 1.3025444746017456, "learning_rate": 0.0005, "loss": 3.35189151763916, "step": 409 }, { "epoch": 0.04770770304863858, "grad_norm": 1.0640041828155518, "learning_rate": 0.0005, "loss": 2.058751106262207, "step": 410 }, { "epoch": 0.04782406329997673, "grad_norm": 1.8709415197372437, "learning_rate": 0.0005, "loss": 3.1988091468811035, "step": 411 }, { "epoch": 0.04794042355131487, "grad_norm": 1.562076210975647, "learning_rate": 0.0005, "loss": 2.227816104888916, "step": 412 }, { "epoch": 0.04805678380265301, "grad_norm": 1.820129156112671, "learning_rate": 0.0005, "loss": 3.388315439224243, "step": 413 }, { "epoch": 0.04817314405399116, "grad_norm": 1.4140963554382324, "learning_rate": 0.0005, "loss": 2.1194663047790527, "step": 414 }, { "epoch": 0.0482895043053293, "grad_norm": 6.935140132904053, "learning_rate": 0.0005, "loss": 3.257274627685547, "step": 415 }, { "epoch": 0.04840586455666744, "grad_norm": 1.4338597059249878, "learning_rate": 0.0005, "loss": 3.0323057174682617, "step": 416 }, { "epoch": 0.04852222480800559, "grad_norm": 1.6365386247634888, "learning_rate": 0.0005, "loss": 2.571798801422119, "step": 417 }, { "epoch": 0.04863858505934373, "grad_norm": 1.9374221563339233, "learning_rate": 0.0005, "loss": 3.540569305419922, "step": 418 }, { "epoch": 0.04875494531068187, "grad_norm": 1.9780375957489014, "learning_rate": 0.0005, "loss": 3.4317686557769775, "step": 419 }, { "epoch": 0.04887130556202001, "grad_norm": 1.5831632614135742, "learning_rate": 0.0005, "loss": 3.3174080848693848, "step": 420 }, { "epoch": 0.04898766581335816, "grad_norm": 1.6168988943099976, "learning_rate": 0.0005, "loss": 4.011370658874512, "step": 421 }, { "epoch": 0.0491040260646963, "grad_norm": 1.66060209274292, "learning_rate": 0.0005, "loss": 4.232476711273193, "step": 422 }, { "epoch": 0.04922038631603444, "grad_norm": 2.119701862335205, "learning_rate": 0.0005, "loss": 3.266170024871826, "step": 423 }, { "epoch": 0.04933674656737259, "grad_norm": 1.8871358633041382, "learning_rate": 0.0005, "loss": 3.2320430278778076, "step": 424 }, { "epoch": 0.04945310681871073, "grad_norm": 1.933849811553955, "learning_rate": 0.0005, "loss": 2.955183982849121, "step": 425 }, { "epoch": 0.04956946707004887, "grad_norm": 1.7607662677764893, "learning_rate": 0.0005, "loss": 3.1926300525665283, "step": 426 }, { "epoch": 0.049685827321387016, "grad_norm": 1.837754249572754, "learning_rate": 0.0005, "loss": 3.130188465118408, "step": 427 }, { "epoch": 0.04980218757272516, "grad_norm": 1.2374534606933594, "learning_rate": 0.0005, "loss": 1.8723636865615845, "step": 428 }, { "epoch": 0.0499185478240633, "grad_norm": 1.8212019205093384, "learning_rate": 0.0005, "loss": 3.05529522895813, "step": 429 }, { "epoch": 0.050034908075401445, "grad_norm": 2.0811734199523926, "learning_rate": 0.0005, "loss": 3.8807897567749023, "step": 430 }, { "epoch": 0.050151268326739586, "grad_norm": 1.9269137382507324, "learning_rate": 0.0005, "loss": 3.514103412628174, "step": 431 }, { "epoch": 0.050267628578077726, "grad_norm": 1.7810970544815063, "learning_rate": 0.0005, "loss": 2.605012893676758, "step": 432 }, { "epoch": 0.050383988829415874, "grad_norm": 1.2134294509887695, "learning_rate": 0.0005, "loss": 2.3884456157684326, "step": 433 }, { "epoch": 0.050500349080754015, "grad_norm": 1.535073161125183, "learning_rate": 0.0005, "loss": 2.9038186073303223, "step": 434 }, { "epoch": 0.050616709332092155, "grad_norm": 2.2614948749542236, "learning_rate": 0.0005, "loss": 4.334193229675293, "step": 435 }, { "epoch": 0.0507330695834303, "grad_norm": 1.5800001621246338, "learning_rate": 0.0005, "loss": 3.067885398864746, "step": 436 }, { "epoch": 0.05084942983476844, "grad_norm": 1.892672061920166, "learning_rate": 0.0005, "loss": 3.8382060527801514, "step": 437 }, { "epoch": 0.050965790086106584, "grad_norm": 2.8276898860931396, "learning_rate": 0.0005, "loss": 2.960033893585205, "step": 438 }, { "epoch": 0.05108215033744473, "grad_norm": 2.9279513359069824, "learning_rate": 0.0005, "loss": 2.7959799766540527, "step": 439 }, { "epoch": 0.05119851058878287, "grad_norm": 1.7614431381225586, "learning_rate": 0.0005, "loss": 2.729748010635376, "step": 440 }, { "epoch": 0.05131487084012101, "grad_norm": 2.5381829738616943, "learning_rate": 0.0005, "loss": 3.1744344234466553, "step": 441 }, { "epoch": 0.05143123109145916, "grad_norm": 1.975001573562622, "learning_rate": 0.0005, "loss": 3.1008896827697754, "step": 442 }, { "epoch": 0.0515475913427973, "grad_norm": 1.6013413667678833, "learning_rate": 0.0005, "loss": 3.72288179397583, "step": 443 }, { "epoch": 0.05166395159413544, "grad_norm": 2.002793073654175, "learning_rate": 0.0005, "loss": 2.543729066848755, "step": 444 }, { "epoch": 0.05178031184547359, "grad_norm": 1.3989185094833374, "learning_rate": 0.0005, "loss": 2.282432794570923, "step": 445 }, { "epoch": 0.05189667209681173, "grad_norm": 1.8271533250808716, "learning_rate": 0.0005, "loss": 3.22914457321167, "step": 446 }, { "epoch": 0.05201303234814987, "grad_norm": 1.7439924478530884, "learning_rate": 0.0005, "loss": 3.180853843688965, "step": 447 }, { "epoch": 0.05212939259948802, "grad_norm": 1.9898873567581177, "learning_rate": 0.0005, "loss": 2.9342660903930664, "step": 448 }, { "epoch": 0.05224575285082616, "grad_norm": 1.7579214572906494, "learning_rate": 0.0005, "loss": 3.9572112560272217, "step": 449 }, { "epoch": 0.0523621131021643, "grad_norm": 2.2179744243621826, "learning_rate": 0.0005, "loss": 3.414027214050293, "step": 450 }, { "epoch": 0.05247847335350244, "grad_norm": 1.774378776550293, "learning_rate": 0.0005, "loss": 3.443589210510254, "step": 451 }, { "epoch": 0.05259483360484059, "grad_norm": 1.6467825174331665, "learning_rate": 0.0005, "loss": 3.826615333557129, "step": 452 }, { "epoch": 0.05271119385617873, "grad_norm": 2.0204641819000244, "learning_rate": 0.0005, "loss": 3.1003453731536865, "step": 453 }, { "epoch": 0.05282755410751687, "grad_norm": 1.2595069408416748, "learning_rate": 0.0005, "loss": 2.8761799335479736, "step": 454 }, { "epoch": 0.05294391435885502, "grad_norm": 1.8468117713928223, "learning_rate": 0.0005, "loss": 4.253689765930176, "step": 455 }, { "epoch": 0.05306027461019316, "grad_norm": 1.6120482683181763, "learning_rate": 0.0005, "loss": 3.471392869949341, "step": 456 }, { "epoch": 0.0531766348615313, "grad_norm": 1.9237167835235596, "learning_rate": 0.0005, "loss": 2.9400811195373535, "step": 457 }, { "epoch": 0.053292995112869446, "grad_norm": 1.494133472442627, "learning_rate": 0.0005, "loss": 3.288273334503174, "step": 458 }, { "epoch": 0.053409355364207586, "grad_norm": 1.578689455986023, "learning_rate": 0.0005, "loss": 2.8085389137268066, "step": 459 }, { "epoch": 0.05352571561554573, "grad_norm": 1.7372816801071167, "learning_rate": 0.0005, "loss": 3.472651243209839, "step": 460 }, { "epoch": 0.053642075866883875, "grad_norm": 2.053668260574341, "learning_rate": 0.0005, "loss": 3.6036758422851562, "step": 461 }, { "epoch": 0.053758436118222015, "grad_norm": 1.6073429584503174, "learning_rate": 0.0005, "loss": 2.557689666748047, "step": 462 }, { "epoch": 0.053874796369560156, "grad_norm": 2.5039191246032715, "learning_rate": 0.0005, "loss": 3.5916390419006348, "step": 463 }, { "epoch": 0.053991156620898303, "grad_norm": 1.936462640762329, "learning_rate": 0.0005, "loss": 3.149838447570801, "step": 464 }, { "epoch": 0.054107516872236444, "grad_norm": 1.098185420036316, "learning_rate": 0.0005, "loss": 1.848844289779663, "step": 465 }, { "epoch": 0.054223877123574585, "grad_norm": 1.5905792713165283, "learning_rate": 0.0005, "loss": 2.6846909523010254, "step": 466 }, { "epoch": 0.05434023737491273, "grad_norm": 2.112776517868042, "learning_rate": 0.0005, "loss": 3.3275413513183594, "step": 467 }, { "epoch": 0.05445659762625087, "grad_norm": 2.248877763748169, "learning_rate": 0.0005, "loss": 3.279857635498047, "step": 468 }, { "epoch": 0.054572957877589014, "grad_norm": 2.8101935386657715, "learning_rate": 0.0005, "loss": 3.3499035835266113, "step": 469 }, { "epoch": 0.05468931812892716, "grad_norm": 1.9310647249221802, "learning_rate": 0.0005, "loss": 3.839895486831665, "step": 470 }, { "epoch": 0.0548056783802653, "grad_norm": 1.689351201057434, "learning_rate": 0.0005, "loss": 3.3809471130371094, "step": 471 }, { "epoch": 0.05492203863160344, "grad_norm": 2.3190314769744873, "learning_rate": 0.0005, "loss": 3.2616195678710938, "step": 472 }, { "epoch": 0.05503839888294159, "grad_norm": 1.6566301584243774, "learning_rate": 0.0005, "loss": 3.1284408569335938, "step": 473 }, { "epoch": 0.05515475913427973, "grad_norm": 2.10628342628479, "learning_rate": 0.0005, "loss": 3.8369569778442383, "step": 474 }, { "epoch": 0.05527111938561787, "grad_norm": 2.1915929317474365, "learning_rate": 0.0005, "loss": 3.5141525268554688, "step": 475 }, { "epoch": 0.05538747963695602, "grad_norm": 1.4264603853225708, "learning_rate": 0.0005, "loss": 2.8088645935058594, "step": 476 }, { "epoch": 0.05550383988829416, "grad_norm": 1.3649042844772339, "learning_rate": 0.0005, "loss": 3.124664783477783, "step": 477 }, { "epoch": 0.0556202001396323, "grad_norm": 1.3659725189208984, "learning_rate": 0.0005, "loss": 2.4217164516448975, "step": 478 }, { "epoch": 0.05573656039097045, "grad_norm": 3.0164239406585693, "learning_rate": 0.0005, "loss": 2.2775678634643555, "step": 479 }, { "epoch": 0.05585292064230859, "grad_norm": 1.7277549505233765, "learning_rate": 0.0005, "loss": 3.1249642372131348, "step": 480 }, { "epoch": 0.05596928089364673, "grad_norm": 2.1073060035705566, "learning_rate": 0.0005, "loss": 2.9835104942321777, "step": 481 }, { "epoch": 0.05608564114498487, "grad_norm": 1.5284918546676636, "learning_rate": 0.0005, "loss": 2.657083511352539, "step": 482 }, { "epoch": 0.05620200139632302, "grad_norm": 1.8046748638153076, "learning_rate": 0.0005, "loss": 3.667206287384033, "step": 483 }, { "epoch": 0.05631836164766116, "grad_norm": 1.8371268510818481, "learning_rate": 0.0005, "loss": 3.16513991355896, "step": 484 }, { "epoch": 0.0564347218989993, "grad_norm": 2.6210103034973145, "learning_rate": 0.0005, "loss": 3.782196521759033, "step": 485 }, { "epoch": 0.056551082150337446, "grad_norm": 2.1008450984954834, "learning_rate": 0.0005, "loss": 3.497231960296631, "step": 486 }, { "epoch": 0.05666744240167559, "grad_norm": 2.2701117992401123, "learning_rate": 0.0005, "loss": 2.8440346717834473, "step": 487 }, { "epoch": 0.05678380265301373, "grad_norm": 2.1577630043029785, "learning_rate": 0.0005, "loss": 2.945573329925537, "step": 488 }, { "epoch": 0.056900162904351875, "grad_norm": 1.642511010169983, "learning_rate": 0.0005, "loss": 2.7997751235961914, "step": 489 }, { "epoch": 0.057016523155690016, "grad_norm": 1.7151869535446167, "learning_rate": 0.0005, "loss": 2.711364984512329, "step": 490 }, { "epoch": 0.05713288340702816, "grad_norm": 5.402647495269775, "learning_rate": 0.0005, "loss": 3.7909462451934814, "step": 491 }, { "epoch": 0.057249243658366304, "grad_norm": 1.4016932249069214, "learning_rate": 0.0005, "loss": 2.083296537399292, "step": 492 }, { "epoch": 0.057365603909704445, "grad_norm": 1.5389763116836548, "learning_rate": 0.0005, "loss": 3.436645746231079, "step": 493 }, { "epoch": 0.057481964161042585, "grad_norm": 1.6894044876098633, "learning_rate": 0.0005, "loss": 2.7783219814300537, "step": 494 }, { "epoch": 0.05759832441238073, "grad_norm": 1.1143381595611572, "learning_rate": 0.0005, "loss": 1.960021734237671, "step": 495 }, { "epoch": 0.057714684663718874, "grad_norm": 1.5775984525680542, "learning_rate": 0.0005, "loss": 3.7584586143493652, "step": 496 }, { "epoch": 0.057831044915057014, "grad_norm": 2.623002290725708, "learning_rate": 0.0005, "loss": 3.075967788696289, "step": 497 }, { "epoch": 0.05794740516639516, "grad_norm": 1.8509860038757324, "learning_rate": 0.0005, "loss": 3.339221239089966, "step": 498 }, { "epoch": 0.0580637654177333, "grad_norm": 3.786482572555542, "learning_rate": 0.0005, "loss": 3.5668725967407227, "step": 499 }, { "epoch": 0.05818012566907144, "grad_norm": 1.6442499160766602, "learning_rate": 0.0005, "loss": 2.019258737564087, "step": 500 }, { "epoch": 0.05829648592040959, "grad_norm": 1.4240667819976807, "learning_rate": 0.0005, "loss": 2.174107789993286, "step": 501 }, { "epoch": 0.05841284617174773, "grad_norm": 1.4882062673568726, "learning_rate": 0.0005, "loss": 2.5903377532958984, "step": 502 }, { "epoch": 0.05852920642308587, "grad_norm": 2.5138680934906006, "learning_rate": 0.0005, "loss": 3.0949370861053467, "step": 503 }, { "epoch": 0.05864556667442402, "grad_norm": 1.9994254112243652, "learning_rate": 0.0005, "loss": 2.6283648014068604, "step": 504 }, { "epoch": 0.05876192692576216, "grad_norm": 2.8019778728485107, "learning_rate": 0.0005, "loss": 3.725496292114258, "step": 505 }, { "epoch": 0.0588782871771003, "grad_norm": 2.0267393589019775, "learning_rate": 0.0005, "loss": 3.1837873458862305, "step": 506 }, { "epoch": 0.05899464742843845, "grad_norm": 1.676829218864441, "learning_rate": 0.0005, "loss": 2.2666049003601074, "step": 507 }, { "epoch": 0.05911100767977659, "grad_norm": 2.5197079181671143, "learning_rate": 0.0005, "loss": 2.9303061962127686, "step": 508 }, { "epoch": 0.05922736793111473, "grad_norm": 1.5661803483963013, "learning_rate": 0.0005, "loss": 3.8241186141967773, "step": 509 }, { "epoch": 0.05934372818245288, "grad_norm": 1.9689704179763794, "learning_rate": 0.0005, "loss": 3.239025354385376, "step": 510 }, { "epoch": 0.05946008843379102, "grad_norm": 1.8026964664459229, "learning_rate": 0.0005, "loss": 3.1819896697998047, "step": 511 }, { "epoch": 0.05957644868512916, "grad_norm": 1.3198778629302979, "learning_rate": 0.0005, "loss": 3.3170223236083984, "step": 512 }, { "epoch": 0.0596928089364673, "grad_norm": 2.1444313526153564, "learning_rate": 0.0005, "loss": 2.943448066711426, "step": 513 }, { "epoch": 0.05980916918780545, "grad_norm": 1.4797515869140625, "learning_rate": 0.0005, "loss": 2.2373509407043457, "step": 514 }, { "epoch": 0.05992552943914359, "grad_norm": 2.4529833793640137, "learning_rate": 0.0005, "loss": 3.2510952949523926, "step": 515 }, { "epoch": 0.06004188969048173, "grad_norm": 1.3874645233154297, "learning_rate": 0.0005, "loss": 3.751375436782837, "step": 516 }, { "epoch": 0.060158249941819876, "grad_norm": 1.9671306610107422, "learning_rate": 0.0005, "loss": 3.062037229537964, "step": 517 }, { "epoch": 0.06027461019315802, "grad_norm": 2.772260904312134, "learning_rate": 0.0005, "loss": 4.004115581512451, "step": 518 }, { "epoch": 0.06039097044449616, "grad_norm": 1.2643851041793823, "learning_rate": 0.0005, "loss": 2.910762071609497, "step": 519 }, { "epoch": 0.060507330695834305, "grad_norm": 2.123593807220459, "learning_rate": 0.0005, "loss": 4.159757614135742, "step": 520 }, { "epoch": 0.060623690947172446, "grad_norm": 2.6264872550964355, "learning_rate": 0.0005, "loss": 3.5089430809020996, "step": 521 }, { "epoch": 0.060740051198510586, "grad_norm": 1.7371543645858765, "learning_rate": 0.0005, "loss": 3.5447030067443848, "step": 522 }, { "epoch": 0.060856411449848734, "grad_norm": 1.7882312536239624, "learning_rate": 0.0005, "loss": 2.993882656097412, "step": 523 }, { "epoch": 0.060972771701186874, "grad_norm": 2.1465280055999756, "learning_rate": 0.0005, "loss": 3.8945910930633545, "step": 524 }, { "epoch": 0.061089131952525015, "grad_norm": 1.8911898136138916, "learning_rate": 0.0005, "loss": 3.359140157699585, "step": 525 }, { "epoch": 0.06120549220386316, "grad_norm": 2.58127760887146, "learning_rate": 0.0005, "loss": 2.8034720420837402, "step": 526 }, { "epoch": 0.0613218524552013, "grad_norm": 1.2243887186050415, "learning_rate": 0.0005, "loss": 3.2250216007232666, "step": 527 }, { "epoch": 0.061438212706539444, "grad_norm": 1.7526781558990479, "learning_rate": 0.0005, "loss": 3.011910915374756, "step": 528 }, { "epoch": 0.06155457295787759, "grad_norm": 1.4113941192626953, "learning_rate": 0.0005, "loss": 2.95430850982666, "step": 529 }, { "epoch": 0.06167093320921573, "grad_norm": 1.4275161027908325, "learning_rate": 0.0005, "loss": 2.8535590171813965, "step": 530 }, { "epoch": 0.06178729346055387, "grad_norm": 1.0982990264892578, "learning_rate": 0.0005, "loss": 2.8440277576446533, "step": 531 }, { "epoch": 0.06190365371189202, "grad_norm": 1.9567222595214844, "learning_rate": 0.0005, "loss": 3.1605231761932373, "step": 532 }, { "epoch": 0.06202001396323016, "grad_norm": 1.7263126373291016, "learning_rate": 0.0005, "loss": 3.9205880165100098, "step": 533 }, { "epoch": 0.0621363742145683, "grad_norm": 2.6187446117401123, "learning_rate": 0.0005, "loss": 2.772944450378418, "step": 534 }, { "epoch": 0.06225273446590645, "grad_norm": 1.9049394130706787, "learning_rate": 0.0005, "loss": 4.008223056793213, "step": 535 }, { "epoch": 0.06236909471724459, "grad_norm": 1.5199333429336548, "learning_rate": 0.0005, "loss": 2.722832679748535, "step": 536 }, { "epoch": 0.06248545496858273, "grad_norm": 1.6098222732543945, "learning_rate": 0.0005, "loss": 2.7739970684051514, "step": 537 }, { "epoch": 0.06260181521992088, "grad_norm": 2.544825792312622, "learning_rate": 0.0005, "loss": 3.2491865158081055, "step": 538 }, { "epoch": 0.06271817547125902, "grad_norm": 1.668768048286438, "learning_rate": 0.0005, "loss": 4.037559986114502, "step": 539 }, { "epoch": 0.06283453572259716, "grad_norm": 1.9319044351577759, "learning_rate": 0.0005, "loss": 2.1417202949523926, "step": 540 }, { "epoch": 0.0629508959739353, "grad_norm": 1.5686439275741577, "learning_rate": 0.0005, "loss": 3.6350860595703125, "step": 541 }, { "epoch": 0.06306725622527344, "grad_norm": 1.5513801574707031, "learning_rate": 0.0005, "loss": 3.669403076171875, "step": 542 }, { "epoch": 0.0631836164766116, "grad_norm": 1.4380167722702026, "learning_rate": 0.0005, "loss": 3.951080322265625, "step": 543 }, { "epoch": 0.06329997672794974, "grad_norm": 1.5344616174697876, "learning_rate": 0.0005, "loss": 2.706566333770752, "step": 544 }, { "epoch": 0.06341633697928788, "grad_norm": 1.6789888143539429, "learning_rate": 0.0005, "loss": 3.2319068908691406, "step": 545 }, { "epoch": 0.06353269723062602, "grad_norm": 2.024277448654175, "learning_rate": 0.0005, "loss": 2.772101402282715, "step": 546 }, { "epoch": 0.06364905748196416, "grad_norm": 1.960309624671936, "learning_rate": 0.0005, "loss": 2.895967960357666, "step": 547 }, { "epoch": 0.0637654177333023, "grad_norm": 1.758994460105896, "learning_rate": 0.0005, "loss": 3.8223626613616943, "step": 548 }, { "epoch": 0.06388177798464045, "grad_norm": 2.3166897296905518, "learning_rate": 0.0005, "loss": 3.1613926887512207, "step": 549 }, { "epoch": 0.0639981382359786, "grad_norm": 2.4812169075012207, "learning_rate": 0.0005, "loss": 3.802072525024414, "step": 550 }, { "epoch": 0.06411449848731673, "grad_norm": 1.571794867515564, "learning_rate": 0.0005, "loss": 3.479586601257324, "step": 551 }, { "epoch": 0.06423085873865488, "grad_norm": 1.7565480470657349, "learning_rate": 0.0005, "loss": 3.2894442081451416, "step": 552 }, { "epoch": 0.06434721898999302, "grad_norm": 1.39332914352417, "learning_rate": 0.0005, "loss": 3.6631453037261963, "step": 553 }, { "epoch": 0.06446357924133116, "grad_norm": 1.3827078342437744, "learning_rate": 0.0005, "loss": 2.5752511024475098, "step": 554 }, { "epoch": 0.06457993949266931, "grad_norm": 1.7936327457427979, "learning_rate": 0.0005, "loss": 3.856935501098633, "step": 555 }, { "epoch": 0.06469629974400745, "grad_norm": 1.7493606805801392, "learning_rate": 0.0005, "loss": 4.30413818359375, "step": 556 }, { "epoch": 0.06481265999534559, "grad_norm": 1.242283821105957, "learning_rate": 0.0005, "loss": 3.26005482673645, "step": 557 }, { "epoch": 0.06492902024668373, "grad_norm": 2.50891375541687, "learning_rate": 0.0005, "loss": 2.844532012939453, "step": 558 }, { "epoch": 0.06504538049802187, "grad_norm": 1.7875151634216309, "learning_rate": 0.0005, "loss": 2.2734639644622803, "step": 559 }, { "epoch": 0.06516174074936001, "grad_norm": 4.068731307983398, "learning_rate": 0.0005, "loss": 4.090875148773193, "step": 560 }, { "epoch": 0.06527810100069815, "grad_norm": 1.9121017456054688, "learning_rate": 0.0005, "loss": 2.9523324966430664, "step": 561 }, { "epoch": 0.06539446125203631, "grad_norm": 1.6863412857055664, "learning_rate": 0.0005, "loss": 3.7112438678741455, "step": 562 }, { "epoch": 0.06551082150337445, "grad_norm": 1.750466227531433, "learning_rate": 0.0005, "loss": 3.1193346977233887, "step": 563 }, { "epoch": 0.06562718175471259, "grad_norm": 1.9035143852233887, "learning_rate": 0.0005, "loss": 3.155970335006714, "step": 564 }, { "epoch": 0.06574354200605073, "grad_norm": 3.0242702960968018, "learning_rate": 0.0005, "loss": 3.6384501457214355, "step": 565 }, { "epoch": 0.06585990225738887, "grad_norm": 1.524357557296753, "learning_rate": 0.0005, "loss": 3.427905797958374, "step": 566 }, { "epoch": 0.06597626250872701, "grad_norm": 1.7608522176742554, "learning_rate": 0.0005, "loss": 2.7603366374969482, "step": 567 }, { "epoch": 0.06609262276006517, "grad_norm": 2.1584534645080566, "learning_rate": 0.0005, "loss": 3.490264654159546, "step": 568 }, { "epoch": 0.06620898301140331, "grad_norm": 1.6764492988586426, "learning_rate": 0.0005, "loss": 3.594144821166992, "step": 569 }, { "epoch": 0.06632534326274145, "grad_norm": 1.6048740148544312, "learning_rate": 0.0005, "loss": 3.416121244430542, "step": 570 }, { "epoch": 0.06644170351407959, "grad_norm": 2.219259738922119, "learning_rate": 0.0005, "loss": 2.983992099761963, "step": 571 }, { "epoch": 0.06655806376541773, "grad_norm": 2.670984983444214, "learning_rate": 0.0005, "loss": 3.1803860664367676, "step": 572 }, { "epoch": 0.06667442401675587, "grad_norm": 1.2197760343551636, "learning_rate": 0.0005, "loss": 2.2362024784088135, "step": 573 }, { "epoch": 0.06679078426809403, "grad_norm": 1.9848850965499878, "learning_rate": 0.0005, "loss": 3.4220666885375977, "step": 574 }, { "epoch": 0.06690714451943217, "grad_norm": 1.1488796472549438, "learning_rate": 0.0005, "loss": 2.747370719909668, "step": 575 }, { "epoch": 0.0670235047707703, "grad_norm": 1.4045239686965942, "learning_rate": 0.0005, "loss": 3.521786689758301, "step": 576 }, { "epoch": 0.06713986502210845, "grad_norm": 1.833404779434204, "learning_rate": 0.0005, "loss": 2.9314889907836914, "step": 577 }, { "epoch": 0.06725622527344659, "grad_norm": 1.675803303718567, "learning_rate": 0.0005, "loss": 3.4676170349121094, "step": 578 }, { "epoch": 0.06737258552478473, "grad_norm": 1.7247964143753052, "learning_rate": 0.0005, "loss": 3.9219319820404053, "step": 579 }, { "epoch": 0.06748894577612288, "grad_norm": 1.680809736251831, "learning_rate": 0.0005, "loss": 3.1786985397338867, "step": 580 }, { "epoch": 0.06760530602746102, "grad_norm": 1.715903878211975, "learning_rate": 0.0005, "loss": 4.2086334228515625, "step": 581 }, { "epoch": 0.06772166627879916, "grad_norm": 1.9448446035385132, "learning_rate": 0.0005, "loss": 3.34690260887146, "step": 582 }, { "epoch": 0.0678380265301373, "grad_norm": 1.4990628957748413, "learning_rate": 0.0005, "loss": 3.1398727893829346, "step": 583 }, { "epoch": 0.06795438678147545, "grad_norm": 1.5092887878417969, "learning_rate": 0.0005, "loss": 3.3123416900634766, "step": 584 }, { "epoch": 0.06807074703281359, "grad_norm": 2.5267560482025146, "learning_rate": 0.0005, "loss": 3.8615074157714844, "step": 585 }, { "epoch": 0.06818710728415174, "grad_norm": 2.3368992805480957, "learning_rate": 0.0005, "loss": 3.7028422355651855, "step": 586 }, { "epoch": 0.06830346753548988, "grad_norm": 1.7596198320388794, "learning_rate": 0.0005, "loss": 2.5074777603149414, "step": 587 }, { "epoch": 0.06841982778682802, "grad_norm": 1.6722785234451294, "learning_rate": 0.0005, "loss": 3.540940761566162, "step": 588 }, { "epoch": 0.06853618803816616, "grad_norm": 1.7394963502883911, "learning_rate": 0.0005, "loss": 2.997788429260254, "step": 589 }, { "epoch": 0.0686525482895043, "grad_norm": 1.4594110250473022, "learning_rate": 0.0005, "loss": 3.4011290073394775, "step": 590 }, { "epoch": 0.06876890854084244, "grad_norm": 1.923892855644226, "learning_rate": 0.0005, "loss": 3.1975769996643066, "step": 591 }, { "epoch": 0.06888526879218058, "grad_norm": 1.622883915901184, "learning_rate": 0.0005, "loss": 2.6702170372009277, "step": 592 }, { "epoch": 0.06900162904351874, "grad_norm": 1.5262458324432373, "learning_rate": 0.0005, "loss": 3.8132965564727783, "step": 593 }, { "epoch": 0.06911798929485688, "grad_norm": 1.4834305047988892, "learning_rate": 0.0005, "loss": 2.630295753479004, "step": 594 }, { "epoch": 0.06923434954619502, "grad_norm": 1.4987618923187256, "learning_rate": 0.0005, "loss": 2.2964303493499756, "step": 595 }, { "epoch": 0.06935070979753316, "grad_norm": 2.1706669330596924, "learning_rate": 0.0005, "loss": 3.3533666133880615, "step": 596 }, { "epoch": 0.0694670700488713, "grad_norm": 2.12739634513855, "learning_rate": 0.0005, "loss": 3.0015459060668945, "step": 597 }, { "epoch": 0.06958343030020944, "grad_norm": 1.6977829933166504, "learning_rate": 0.0005, "loss": 2.9558730125427246, "step": 598 }, { "epoch": 0.0696997905515476, "grad_norm": 2.674265146255493, "learning_rate": 0.0005, "loss": 2.741140842437744, "step": 599 }, { "epoch": 0.06981615080288574, "grad_norm": 1.8517578840255737, "learning_rate": 0.0005, "loss": 4.0843024253845215, "step": 600 }, { "epoch": 0.06993251105422388, "grad_norm": 1.3020217418670654, "learning_rate": 0.0005, "loss": 3.060413122177124, "step": 601 }, { "epoch": 0.07004887130556202, "grad_norm": 1.3139289617538452, "learning_rate": 0.0005, "loss": 2.5821242332458496, "step": 602 }, { "epoch": 0.07016523155690016, "grad_norm": 2.087655544281006, "learning_rate": 0.0005, "loss": 3.8784947395324707, "step": 603 }, { "epoch": 0.0702815918082383, "grad_norm": 1.5585155487060547, "learning_rate": 0.0005, "loss": 3.463740587234497, "step": 604 }, { "epoch": 0.07039795205957645, "grad_norm": 1.577074646949768, "learning_rate": 0.0005, "loss": 2.7823128700256348, "step": 605 }, { "epoch": 0.0705143123109146, "grad_norm": 3.796245813369751, "learning_rate": 0.0005, "loss": 2.815418243408203, "step": 606 }, { "epoch": 0.07063067256225274, "grad_norm": 2.136244773864746, "learning_rate": 0.0005, "loss": 3.4236085414886475, "step": 607 }, { "epoch": 0.07074703281359088, "grad_norm": 1.991894006729126, "learning_rate": 0.0005, "loss": 3.8296730518341064, "step": 608 }, { "epoch": 0.07086339306492902, "grad_norm": 1.613943338394165, "learning_rate": 0.0005, "loss": 3.853527307510376, "step": 609 }, { "epoch": 0.07097975331626716, "grad_norm": 1.1950289011001587, "learning_rate": 0.0005, "loss": 2.3029799461364746, "step": 610 }, { "epoch": 0.07109611356760531, "grad_norm": 1.4296890497207642, "learning_rate": 0.0005, "loss": 4.088702201843262, "step": 611 }, { "epoch": 0.07121247381894345, "grad_norm": 3.0486092567443848, "learning_rate": 0.0005, "loss": 3.2301549911499023, "step": 612 }, { "epoch": 0.0713288340702816, "grad_norm": 1.283438801765442, "learning_rate": 0.0005, "loss": 2.180316925048828, "step": 613 }, { "epoch": 0.07144519432161973, "grad_norm": 1.503531813621521, "learning_rate": 0.0005, "loss": 1.798720359802246, "step": 614 }, { "epoch": 0.07156155457295788, "grad_norm": 2.2085161209106445, "learning_rate": 0.0005, "loss": 4.2315568923950195, "step": 615 }, { "epoch": 0.07167791482429602, "grad_norm": 2.473137855529785, "learning_rate": 0.0005, "loss": 3.5920610427856445, "step": 616 }, { "epoch": 0.07179427507563417, "grad_norm": 1.4687808752059937, "learning_rate": 0.0005, "loss": 2.2607927322387695, "step": 617 }, { "epoch": 0.07191063532697231, "grad_norm": 1.8344619274139404, "learning_rate": 0.0005, "loss": 3.202641487121582, "step": 618 }, { "epoch": 0.07202699557831045, "grad_norm": 1.757636547088623, "learning_rate": 0.0005, "loss": 2.8701934814453125, "step": 619 }, { "epoch": 0.07214335582964859, "grad_norm": 1.9222596883773804, "learning_rate": 0.0005, "loss": 3.997591495513916, "step": 620 }, { "epoch": 0.07225971608098673, "grad_norm": 1.4464876651763916, "learning_rate": 0.0005, "loss": 3.881986141204834, "step": 621 }, { "epoch": 0.07237607633232487, "grad_norm": 1.7945464849472046, "learning_rate": 0.0005, "loss": 2.752440929412842, "step": 622 }, { "epoch": 0.07249243658366301, "grad_norm": 1.3058348894119263, "learning_rate": 0.0005, "loss": 2.662376880645752, "step": 623 }, { "epoch": 0.07260879683500117, "grad_norm": 2.109873056411743, "learning_rate": 0.0005, "loss": 2.704580783843994, "step": 624 }, { "epoch": 0.07272515708633931, "grad_norm": 2.6356263160705566, "learning_rate": 0.0005, "loss": 3.0806679725646973, "step": 625 }, { "epoch": 0.07284151733767745, "grad_norm": 0.9339062571525574, "learning_rate": 0.0005, "loss": 1.5451171398162842, "step": 626 }, { "epoch": 0.07295787758901559, "grad_norm": 1.6331520080566406, "learning_rate": 0.0005, "loss": 3.7608790397644043, "step": 627 }, { "epoch": 0.07307423784035373, "grad_norm": 1.9393126964569092, "learning_rate": 0.0005, "loss": 3.772765636444092, "step": 628 }, { "epoch": 0.07319059809169187, "grad_norm": 1.2523902654647827, "learning_rate": 0.0005, "loss": 2.7019848823547363, "step": 629 }, { "epoch": 0.07330695834303003, "grad_norm": 1.464639663696289, "learning_rate": 0.0005, "loss": 3.026366949081421, "step": 630 }, { "epoch": 0.07342331859436817, "grad_norm": 1.2059028148651123, "learning_rate": 0.0005, "loss": 2.794416904449463, "step": 631 }, { "epoch": 0.07353967884570631, "grad_norm": 2.268174886703491, "learning_rate": 0.0005, "loss": 2.814908504486084, "step": 632 }, { "epoch": 0.07365603909704445, "grad_norm": 1.374658226966858, "learning_rate": 0.0005, "loss": 2.0688791275024414, "step": 633 }, { "epoch": 0.07377239934838259, "grad_norm": 1.8397772312164307, "learning_rate": 0.0005, "loss": 3.5317013263702393, "step": 634 }, { "epoch": 0.07388875959972073, "grad_norm": 1.529707670211792, "learning_rate": 0.0005, "loss": 3.533769369125366, "step": 635 }, { "epoch": 0.07400511985105888, "grad_norm": 1.8425740003585815, "learning_rate": 0.0005, "loss": 3.926928758621216, "step": 636 }, { "epoch": 0.07412148010239702, "grad_norm": 1.603253722190857, "learning_rate": 0.0005, "loss": 3.6449975967407227, "step": 637 }, { "epoch": 0.07423784035373517, "grad_norm": 1.4652858972549438, "learning_rate": 0.0005, "loss": 2.449411630630493, "step": 638 }, { "epoch": 0.0743542006050733, "grad_norm": 1.3199950456619263, "learning_rate": 0.0005, "loss": 2.074410915374756, "step": 639 }, { "epoch": 0.07447056085641145, "grad_norm": 1.6672178506851196, "learning_rate": 0.0005, "loss": 3.1190345287323, "step": 640 }, { "epoch": 0.07458692110774959, "grad_norm": 1.7021634578704834, "learning_rate": 0.0005, "loss": 3.214585065841675, "step": 641 }, { "epoch": 0.07470328135908774, "grad_norm": 2.6320741176605225, "learning_rate": 0.0005, "loss": 3.6410655975341797, "step": 642 }, { "epoch": 0.07481964161042588, "grad_norm": 2.1700291633605957, "learning_rate": 0.0005, "loss": 2.375516653060913, "step": 643 }, { "epoch": 0.07493600186176402, "grad_norm": 1.7750126123428345, "learning_rate": 0.0005, "loss": 2.8452513217926025, "step": 644 }, { "epoch": 0.07505236211310216, "grad_norm": 1.2051197290420532, "learning_rate": 0.0005, "loss": 2.2486820220947266, "step": 645 }, { "epoch": 0.0751687223644403, "grad_norm": 2.058738946914673, "learning_rate": 0.0005, "loss": 2.9783596992492676, "step": 646 }, { "epoch": 0.07528508261577845, "grad_norm": 2.606109142303467, "learning_rate": 0.0005, "loss": 2.5572509765625, "step": 647 }, { "epoch": 0.0754014428671166, "grad_norm": 2.0330073833465576, "learning_rate": 0.0005, "loss": 3.7334017753601074, "step": 648 }, { "epoch": 0.07551780311845474, "grad_norm": 1.967951774597168, "learning_rate": 0.0005, "loss": 3.085742950439453, "step": 649 }, { "epoch": 0.07563416336979288, "grad_norm": 1.5936894416809082, "learning_rate": 0.0005, "loss": 2.9539544582366943, "step": 650 }, { "epoch": 0.07575052362113102, "grad_norm": 2.5575129985809326, "learning_rate": 0.0005, "loss": 3.344768524169922, "step": 651 }, { "epoch": 0.07586688387246916, "grad_norm": 1.7511110305786133, "learning_rate": 0.0005, "loss": 3.7356090545654297, "step": 652 }, { "epoch": 0.0759832441238073, "grad_norm": 1.5879371166229248, "learning_rate": 0.0005, "loss": 3.7727255821228027, "step": 653 }, { "epoch": 0.07609960437514544, "grad_norm": 1.6060837507247925, "learning_rate": 0.0005, "loss": 3.486220359802246, "step": 654 }, { "epoch": 0.0762159646264836, "grad_norm": 1.9142886400222778, "learning_rate": 0.0005, "loss": 2.3164782524108887, "step": 655 }, { "epoch": 0.07633232487782174, "grad_norm": 1.9732738733291626, "learning_rate": 0.0005, "loss": 4.025639533996582, "step": 656 }, { "epoch": 0.07644868512915988, "grad_norm": 3.4225144386291504, "learning_rate": 0.0005, "loss": 4.073407173156738, "step": 657 }, { "epoch": 0.07656504538049802, "grad_norm": 2.0828733444213867, "learning_rate": 0.0005, "loss": 3.063100576400757, "step": 658 }, { "epoch": 0.07668140563183616, "grad_norm": 1.7873866558074951, "learning_rate": 0.0005, "loss": 3.8411850929260254, "step": 659 }, { "epoch": 0.0767977658831743, "grad_norm": 2.470547676086426, "learning_rate": 0.0005, "loss": 3.6190710067749023, "step": 660 }, { "epoch": 0.07691412613451246, "grad_norm": 1.2185571193695068, "learning_rate": 0.0005, "loss": 2.281996965408325, "step": 661 }, { "epoch": 0.0770304863858506, "grad_norm": 2.7431299686431885, "learning_rate": 0.0005, "loss": 3.718797206878662, "step": 662 }, { "epoch": 0.07714684663718874, "grad_norm": 1.6315995454788208, "learning_rate": 0.0005, "loss": 2.76790714263916, "step": 663 }, { "epoch": 0.07726320688852688, "grad_norm": 1.7318239212036133, "learning_rate": 0.0005, "loss": 3.8966565132141113, "step": 664 }, { "epoch": 0.07737956713986502, "grad_norm": 2.277390241622925, "learning_rate": 0.0005, "loss": 3.7403297424316406, "step": 665 }, { "epoch": 0.07749592739120316, "grad_norm": 1.5544275045394897, "learning_rate": 0.0005, "loss": 3.3816494941711426, "step": 666 }, { "epoch": 0.07761228764254131, "grad_norm": 1.8975651264190674, "learning_rate": 0.0005, "loss": 3.6339728832244873, "step": 667 }, { "epoch": 0.07772864789387945, "grad_norm": 1.6856626272201538, "learning_rate": 0.0005, "loss": 3.1142702102661133, "step": 668 }, { "epoch": 0.0778450081452176, "grad_norm": 2.5432493686676025, "learning_rate": 0.0005, "loss": 3.66203236579895, "step": 669 }, { "epoch": 0.07796136839655574, "grad_norm": 1.5592471361160278, "learning_rate": 0.0005, "loss": 2.746598243713379, "step": 670 }, { "epoch": 0.07807772864789388, "grad_norm": 2.0901286602020264, "learning_rate": 0.0005, "loss": 3.441352605819702, "step": 671 }, { "epoch": 0.07819408889923202, "grad_norm": 1.3084555864334106, "learning_rate": 0.0005, "loss": 3.0658583641052246, "step": 672 }, { "epoch": 0.07831044915057017, "grad_norm": 3.336930274963379, "learning_rate": 0.0005, "loss": 2.801029682159424, "step": 673 }, { "epoch": 0.07842680940190831, "grad_norm": 1.9137037992477417, "learning_rate": 0.0005, "loss": 3.5025198459625244, "step": 674 }, { "epoch": 0.07854316965324645, "grad_norm": 1.788578987121582, "learning_rate": 0.0005, "loss": 2.5123348236083984, "step": 675 }, { "epoch": 0.0786595299045846, "grad_norm": 1.956906795501709, "learning_rate": 0.0005, "loss": 3.323348045349121, "step": 676 }, { "epoch": 0.07877589015592273, "grad_norm": 1.6862564086914062, "learning_rate": 0.0005, "loss": 3.025200843811035, "step": 677 }, { "epoch": 0.07889225040726087, "grad_norm": 1.1005070209503174, "learning_rate": 0.0005, "loss": 1.987252950668335, "step": 678 }, { "epoch": 0.07900861065859903, "grad_norm": 1.4738399982452393, "learning_rate": 0.0005, "loss": 1.9015471935272217, "step": 679 }, { "epoch": 0.07912497090993717, "grad_norm": 1.6238611936569214, "learning_rate": 0.0005, "loss": 2.650779962539673, "step": 680 }, { "epoch": 0.07924133116127531, "grad_norm": 1.6954472064971924, "learning_rate": 0.0005, "loss": 3.359719753265381, "step": 681 }, { "epoch": 0.07935769141261345, "grad_norm": 2.210923433303833, "learning_rate": 0.0005, "loss": 2.815011978149414, "step": 682 }, { "epoch": 0.07947405166395159, "grad_norm": 2.638267755508423, "learning_rate": 0.0005, "loss": 2.3777830600738525, "step": 683 }, { "epoch": 0.07959041191528973, "grad_norm": 1.503888726234436, "learning_rate": 0.0005, "loss": 2.6173017024993896, "step": 684 }, { "epoch": 0.07970677216662787, "grad_norm": 1.975439429283142, "learning_rate": 0.0005, "loss": 2.976177215576172, "step": 685 }, { "epoch": 0.07982313241796603, "grad_norm": 2.2635161876678467, "learning_rate": 0.0005, "loss": 4.521445274353027, "step": 686 }, { "epoch": 0.07993949266930417, "grad_norm": 1.9588831663131714, "learning_rate": 0.0005, "loss": 3.152374744415283, "step": 687 }, { "epoch": 0.08005585292064231, "grad_norm": 2.5386223793029785, "learning_rate": 0.0005, "loss": 3.415569543838501, "step": 688 }, { "epoch": 0.08017221317198045, "grad_norm": 1.65004563331604, "learning_rate": 0.0005, "loss": 3.176684617996216, "step": 689 }, { "epoch": 0.08028857342331859, "grad_norm": 1.2136907577514648, "learning_rate": 0.0005, "loss": 2.577134132385254, "step": 690 }, { "epoch": 0.08040493367465673, "grad_norm": 1.3360830545425415, "learning_rate": 0.0005, "loss": 2.974538803100586, "step": 691 }, { "epoch": 0.08052129392599489, "grad_norm": 1.6344270706176758, "learning_rate": 0.0005, "loss": 4.206530570983887, "step": 692 }, { "epoch": 0.08063765417733303, "grad_norm": 1.4718836545944214, "learning_rate": 0.0005, "loss": 4.1452836990356445, "step": 693 }, { "epoch": 0.08075401442867117, "grad_norm": 1.2823169231414795, "learning_rate": 0.0005, "loss": 2.8530125617980957, "step": 694 }, { "epoch": 0.08087037468000931, "grad_norm": 1.2002909183502197, "learning_rate": 0.0005, "loss": 3.270141839981079, "step": 695 }, { "epoch": 0.08098673493134745, "grad_norm": 1.2297921180725098, "learning_rate": 0.0005, "loss": 2.5223350524902344, "step": 696 }, { "epoch": 0.08110309518268559, "grad_norm": 1.8321577310562134, "learning_rate": 0.0005, "loss": 3.9343087673187256, "step": 697 }, { "epoch": 0.08121945543402374, "grad_norm": 1.8546967506408691, "learning_rate": 0.0005, "loss": 3.2482974529266357, "step": 698 }, { "epoch": 0.08133581568536188, "grad_norm": 1.489845871925354, "learning_rate": 0.0005, "loss": 3.308948040008545, "step": 699 }, { "epoch": 0.08145217593670002, "grad_norm": 1.5045605897903442, "learning_rate": 0.0005, "loss": 2.7246339321136475, "step": 700 }, { "epoch": 0.08156853618803817, "grad_norm": 1.2971184253692627, "learning_rate": 0.0005, "loss": 3.2889633178710938, "step": 701 }, { "epoch": 0.0816848964393763, "grad_norm": 2.1439971923828125, "learning_rate": 0.0005, "loss": 3.902674674987793, "step": 702 }, { "epoch": 0.08180125669071445, "grad_norm": 1.2181849479675293, "learning_rate": 0.0005, "loss": 2.7522802352905273, "step": 703 }, { "epoch": 0.0819176169420526, "grad_norm": 1.957573652267456, "learning_rate": 0.0005, "loss": 2.5573854446411133, "step": 704 }, { "epoch": 0.08203397719339074, "grad_norm": 2.37092661857605, "learning_rate": 0.0005, "loss": 3.396116018295288, "step": 705 }, { "epoch": 0.08215033744472888, "grad_norm": 1.453513264656067, "learning_rate": 0.0005, "loss": 3.8109397888183594, "step": 706 }, { "epoch": 0.08226669769606702, "grad_norm": 2.1474392414093018, "learning_rate": 0.0005, "loss": 3.0200695991516113, "step": 707 }, { "epoch": 0.08238305794740516, "grad_norm": 1.3773717880249023, "learning_rate": 0.0005, "loss": 3.5507748126983643, "step": 708 }, { "epoch": 0.0824994181987433, "grad_norm": 1.0975871086120605, "learning_rate": 0.0005, "loss": 2.0131468772888184, "step": 709 }, { "epoch": 0.08261577845008146, "grad_norm": 1.7778373956680298, "learning_rate": 0.0005, "loss": 3.8256936073303223, "step": 710 }, { "epoch": 0.0827321387014196, "grad_norm": 1.6379785537719727, "learning_rate": 0.0005, "loss": 2.943480968475342, "step": 711 }, { "epoch": 0.08284849895275774, "grad_norm": 1.3786803483963013, "learning_rate": 0.0005, "loss": 2.8375158309936523, "step": 712 }, { "epoch": 0.08296485920409588, "grad_norm": 1.7703783512115479, "learning_rate": 0.0005, "loss": 3.285118579864502, "step": 713 }, { "epoch": 0.08308121945543402, "grad_norm": 0.9833521246910095, "learning_rate": 0.0005, "loss": 2.1259970664978027, "step": 714 }, { "epoch": 0.08319757970677216, "grad_norm": 1.5147203207015991, "learning_rate": 0.0005, "loss": 2.7074692249298096, "step": 715 }, { "epoch": 0.0833139399581103, "grad_norm": 1.3885997533798218, "learning_rate": 0.0005, "loss": 2.0931036472320557, "step": 716 }, { "epoch": 0.08343030020944846, "grad_norm": 1.9560216665267944, "learning_rate": 0.0005, "loss": 3.5742790699005127, "step": 717 }, { "epoch": 0.0835466604607866, "grad_norm": 1.5588157176971436, "learning_rate": 0.0005, "loss": 2.280756950378418, "step": 718 }, { "epoch": 0.08366302071212474, "grad_norm": 1.71077299118042, "learning_rate": 0.0005, "loss": 2.877591371536255, "step": 719 }, { "epoch": 0.08377938096346288, "grad_norm": 2.022705554962158, "learning_rate": 0.0005, "loss": 4.065557956695557, "step": 720 }, { "epoch": 0.08389574121480102, "grad_norm": 1.9451713562011719, "learning_rate": 0.0005, "loss": 3.351302146911621, "step": 721 }, { "epoch": 0.08401210146613916, "grad_norm": 1.8547897338867188, "learning_rate": 0.0005, "loss": 2.6655702590942383, "step": 722 }, { "epoch": 0.08412846171747732, "grad_norm": 2.829834461212158, "learning_rate": 0.0005, "loss": 3.7597312927246094, "step": 723 }, { "epoch": 0.08424482196881546, "grad_norm": 1.7473607063293457, "learning_rate": 0.0005, "loss": 2.0108895301818848, "step": 724 }, { "epoch": 0.0843611822201536, "grad_norm": 1.7468092441558838, "learning_rate": 0.0005, "loss": 2.2822341918945312, "step": 725 }, { "epoch": 0.08447754247149174, "grad_norm": 1.2311584949493408, "learning_rate": 0.0005, "loss": 3.082156181335449, "step": 726 }, { "epoch": 0.08459390272282988, "grad_norm": 1.6082040071487427, "learning_rate": 0.0005, "loss": 3.3876891136169434, "step": 727 }, { "epoch": 0.08471026297416802, "grad_norm": 2.692915201187134, "learning_rate": 0.0005, "loss": 3.962205410003662, "step": 728 }, { "epoch": 0.08482662322550617, "grad_norm": 1.4318583011627197, "learning_rate": 0.0005, "loss": 3.2999582290649414, "step": 729 }, { "epoch": 0.08494298347684431, "grad_norm": 2.5312674045562744, "learning_rate": 0.0005, "loss": 3.67673921585083, "step": 730 }, { "epoch": 0.08505934372818245, "grad_norm": 1.695825219154358, "learning_rate": 0.0005, "loss": 3.713311195373535, "step": 731 }, { "epoch": 0.0851757039795206, "grad_norm": 1.31784188747406, "learning_rate": 0.0005, "loss": 3.0759949684143066, "step": 732 }, { "epoch": 0.08529206423085874, "grad_norm": 1.7572649717330933, "learning_rate": 0.0005, "loss": 2.671841859817505, "step": 733 }, { "epoch": 0.08540842448219688, "grad_norm": 2.400822639465332, "learning_rate": 0.0005, "loss": 3.3713254928588867, "step": 734 }, { "epoch": 0.08552478473353503, "grad_norm": 2.1245410442352295, "learning_rate": 0.0005, "loss": 3.733680248260498, "step": 735 }, { "epoch": 0.08564114498487317, "grad_norm": 2.2849390506744385, "learning_rate": 0.0005, "loss": 2.849623680114746, "step": 736 }, { "epoch": 0.08575750523621131, "grad_norm": 2.4699697494506836, "learning_rate": 0.0005, "loss": 3.6925621032714844, "step": 737 }, { "epoch": 0.08587386548754945, "grad_norm": 2.306849718093872, "learning_rate": 0.0005, "loss": 3.0433144569396973, "step": 738 }, { "epoch": 0.0859902257388876, "grad_norm": 2.0580992698669434, "learning_rate": 0.0005, "loss": 2.4186055660247803, "step": 739 }, { "epoch": 0.08610658599022573, "grad_norm": 3.195651054382324, "learning_rate": 0.0005, "loss": 3.9038286209106445, "step": 740 }, { "epoch": 0.08622294624156389, "grad_norm": 1.8194226026535034, "learning_rate": 0.0005, "loss": 3.682857036590576, "step": 741 }, { "epoch": 0.08633930649290203, "grad_norm": 1.9056700468063354, "learning_rate": 0.0005, "loss": 3.183387517929077, "step": 742 }, { "epoch": 0.08645566674424017, "grad_norm": 1.680816650390625, "learning_rate": 0.0005, "loss": 2.9027249813079834, "step": 743 }, { "epoch": 0.08657202699557831, "grad_norm": 1.890523910522461, "learning_rate": 0.0005, "loss": 2.61527419090271, "step": 744 }, { "epoch": 0.08668838724691645, "grad_norm": 1.419066309928894, "learning_rate": 0.0005, "loss": 3.107663631439209, "step": 745 }, { "epoch": 0.08680474749825459, "grad_norm": 1.623015284538269, "learning_rate": 0.0005, "loss": 2.5960285663604736, "step": 746 }, { "epoch": 0.08692110774959273, "grad_norm": 1.5655137300491333, "learning_rate": 0.0005, "loss": 2.751488447189331, "step": 747 }, { "epoch": 0.08703746800093089, "grad_norm": 1.4016993045806885, "learning_rate": 0.0005, "loss": 3.7522945404052734, "step": 748 }, { "epoch": 0.08715382825226903, "grad_norm": 2.4269213676452637, "learning_rate": 0.0005, "loss": 3.504145622253418, "step": 749 }, { "epoch": 0.08727018850360717, "grad_norm": 2.6252152919769287, "learning_rate": 0.0005, "loss": 3.3145124912261963, "step": 750 }, { "epoch": 0.08738654875494531, "grad_norm": 1.6844309568405151, "learning_rate": 0.0005, "loss": 2.7570528984069824, "step": 751 }, { "epoch": 0.08750290900628345, "grad_norm": 1.2655315399169922, "learning_rate": 0.0005, "loss": 2.8523988723754883, "step": 752 }, { "epoch": 0.08761926925762159, "grad_norm": 1.5085628032684326, "learning_rate": 0.0005, "loss": 2.509293794631958, "step": 753 }, { "epoch": 0.08773562950895974, "grad_norm": 1.7988539934158325, "learning_rate": 0.0005, "loss": 2.7581276893615723, "step": 754 }, { "epoch": 0.08785198976029789, "grad_norm": 1.6905429363250732, "learning_rate": 0.0005, "loss": 2.77895450592041, "step": 755 }, { "epoch": 0.08796835001163603, "grad_norm": 2.3257999420166016, "learning_rate": 0.0005, "loss": 3.556565284729004, "step": 756 }, { "epoch": 0.08808471026297417, "grad_norm": 1.4315767288208008, "learning_rate": 0.0005, "loss": 2.582134246826172, "step": 757 }, { "epoch": 0.08820107051431231, "grad_norm": 1.5850844383239746, "learning_rate": 0.0005, "loss": 2.889511823654175, "step": 758 }, { "epoch": 0.08831743076565045, "grad_norm": 1.1270239353179932, "learning_rate": 0.0005, "loss": 2.5133485794067383, "step": 759 }, { "epoch": 0.0884337910169886, "grad_norm": 1.6629329919815063, "learning_rate": 0.0005, "loss": 3.430764675140381, "step": 760 }, { "epoch": 0.08855015126832674, "grad_norm": 1.7818522453308105, "learning_rate": 0.0005, "loss": 4.094860076904297, "step": 761 }, { "epoch": 0.08866651151966488, "grad_norm": 1.4433670043945312, "learning_rate": 0.0005, "loss": 2.621364116668701, "step": 762 }, { "epoch": 0.08878287177100302, "grad_norm": 1.8176960945129395, "learning_rate": 0.0005, "loss": 2.7510223388671875, "step": 763 }, { "epoch": 0.08889923202234117, "grad_norm": 1.4143214225769043, "learning_rate": 0.0005, "loss": 2.2197189331054688, "step": 764 }, { "epoch": 0.0890155922736793, "grad_norm": 1.9742403030395508, "learning_rate": 0.0005, "loss": 4.044788837432861, "step": 765 }, { "epoch": 0.08913195252501746, "grad_norm": 3.5309665203094482, "learning_rate": 0.0005, "loss": 3.2278380393981934, "step": 766 }, { "epoch": 0.0892483127763556, "grad_norm": 1.6474153995513916, "learning_rate": 0.0005, "loss": 3.10652232170105, "step": 767 }, { "epoch": 0.08936467302769374, "grad_norm": 2.0138731002807617, "learning_rate": 0.0005, "loss": 3.891369342803955, "step": 768 }, { "epoch": 0.08948103327903188, "grad_norm": 1.3474961519241333, "learning_rate": 0.0005, "loss": 2.029085159301758, "step": 769 }, { "epoch": 0.08959739353037002, "grad_norm": 1.6477543115615845, "learning_rate": 0.0005, "loss": 3.746704578399658, "step": 770 }, { "epoch": 0.08971375378170816, "grad_norm": 1.201422095298767, "learning_rate": 0.0005, "loss": 2.7016263008117676, "step": 771 }, { "epoch": 0.08983011403304632, "grad_norm": 1.7232636213302612, "learning_rate": 0.0005, "loss": 3.06325364112854, "step": 772 }, { "epoch": 0.08994647428438446, "grad_norm": 1.442062497138977, "learning_rate": 0.0005, "loss": 2.6072611808776855, "step": 773 }, { "epoch": 0.0900628345357226, "grad_norm": 2.408005714416504, "learning_rate": 0.0005, "loss": 3.2239084243774414, "step": 774 }, { "epoch": 0.09017919478706074, "grad_norm": 1.3267821073532104, "learning_rate": 0.0005, "loss": 3.6604766845703125, "step": 775 }, { "epoch": 0.09029555503839888, "grad_norm": 1.7993947267532349, "learning_rate": 0.0005, "loss": 4.131452560424805, "step": 776 }, { "epoch": 0.09041191528973702, "grad_norm": 1.9943432807922363, "learning_rate": 0.0005, "loss": 3.219383955001831, "step": 777 }, { "epoch": 0.09052827554107516, "grad_norm": 1.6310880184173584, "learning_rate": 0.0005, "loss": 3.364603042602539, "step": 778 }, { "epoch": 0.09064463579241332, "grad_norm": 1.6887151002883911, "learning_rate": 0.0005, "loss": 2.469487190246582, "step": 779 }, { "epoch": 0.09076099604375146, "grad_norm": 1.207228422164917, "learning_rate": 0.0005, "loss": 3.4078617095947266, "step": 780 }, { "epoch": 0.0908773562950896, "grad_norm": 3.8354780673980713, "learning_rate": 0.0005, "loss": 3.8475513458251953, "step": 781 }, { "epoch": 0.09099371654642774, "grad_norm": 4.1029052734375, "learning_rate": 0.0005, "loss": 3.78294038772583, "step": 782 }, { "epoch": 0.09111007679776588, "grad_norm": 1.4152204990386963, "learning_rate": 0.0005, "loss": 2.768890619277954, "step": 783 }, { "epoch": 0.09122643704910402, "grad_norm": 1.9851429462432861, "learning_rate": 0.0005, "loss": 3.7106173038482666, "step": 784 }, { "epoch": 0.09134279730044217, "grad_norm": 1.3434042930603027, "learning_rate": 0.0005, "loss": 3.152622699737549, "step": 785 }, { "epoch": 0.09145915755178032, "grad_norm": 1.3356359004974365, "learning_rate": 0.0005, "loss": 2.4228687286376953, "step": 786 }, { "epoch": 0.09157551780311846, "grad_norm": 1.3350874185562134, "learning_rate": 0.0005, "loss": 2.473385810852051, "step": 787 }, { "epoch": 0.0916918780544566, "grad_norm": 1.7762045860290527, "learning_rate": 0.0005, "loss": 2.7415695190429688, "step": 788 }, { "epoch": 0.09180823830579474, "grad_norm": 1.5718176364898682, "learning_rate": 0.0005, "loss": 2.530411720275879, "step": 789 }, { "epoch": 0.09192459855713288, "grad_norm": 1.4305745363235474, "learning_rate": 0.0005, "loss": 2.494821071624756, "step": 790 }, { "epoch": 0.09204095880847103, "grad_norm": 2.2562570571899414, "learning_rate": 0.0005, "loss": 3.0369534492492676, "step": 791 }, { "epoch": 0.09215731905980917, "grad_norm": 2.147120237350464, "learning_rate": 0.0005, "loss": 4.353631019592285, "step": 792 }, { "epoch": 0.09227367931114731, "grad_norm": 2.279519557952881, "learning_rate": 0.0005, "loss": 3.1917996406555176, "step": 793 }, { "epoch": 0.09239003956248545, "grad_norm": 1.555631399154663, "learning_rate": 0.0005, "loss": 2.87555193901062, "step": 794 }, { "epoch": 0.0925063998138236, "grad_norm": 1.9062976837158203, "learning_rate": 0.0005, "loss": 3.035515308380127, "step": 795 }, { "epoch": 0.09262276006516174, "grad_norm": 1.7990491390228271, "learning_rate": 0.0005, "loss": 3.915710926055908, "step": 796 }, { "epoch": 0.09273912031649989, "grad_norm": 1.8126208782196045, "learning_rate": 0.0005, "loss": 2.7894814014434814, "step": 797 }, { "epoch": 0.09285548056783803, "grad_norm": 1.3588335514068604, "learning_rate": 0.0005, "loss": 2.828728199005127, "step": 798 }, { "epoch": 0.09297184081917617, "grad_norm": 1.323169231414795, "learning_rate": 0.0005, "loss": 3.2432520389556885, "step": 799 }, { "epoch": 0.09308820107051431, "grad_norm": 2.437084913253784, "learning_rate": 0.0005, "loss": 2.9401988983154297, "step": 800 }, { "epoch": 0.09320456132185245, "grad_norm": 1.6640892028808594, "learning_rate": 0.0005, "loss": 2.689986228942871, "step": 801 }, { "epoch": 0.0933209215731906, "grad_norm": 1.6990158557891846, "learning_rate": 0.0005, "loss": 2.771259069442749, "step": 802 }, { "epoch": 0.09343728182452875, "grad_norm": 1.3437937498092651, "learning_rate": 0.0005, "loss": 2.7231853008270264, "step": 803 }, { "epoch": 0.09355364207586689, "grad_norm": 1.5425901412963867, "learning_rate": 0.0005, "loss": 3.2400970458984375, "step": 804 }, { "epoch": 0.09367000232720503, "grad_norm": 2.796128511428833, "learning_rate": 0.0005, "loss": 3.579082489013672, "step": 805 }, { "epoch": 0.09378636257854317, "grad_norm": 2.186485528945923, "learning_rate": 0.0005, "loss": 2.9923620223999023, "step": 806 }, { "epoch": 0.09390272282988131, "grad_norm": 1.5912902355194092, "learning_rate": 0.0005, "loss": 3.010887861251831, "step": 807 }, { "epoch": 0.09401908308121945, "grad_norm": 1.9585703611373901, "learning_rate": 0.0005, "loss": 2.6353209018707275, "step": 808 }, { "epoch": 0.09413544333255759, "grad_norm": 2.2377939224243164, "learning_rate": 0.0005, "loss": 4.004065036773682, "step": 809 }, { "epoch": 0.09425180358389575, "grad_norm": 1.2034333944320679, "learning_rate": 0.0005, "loss": 3.045992851257324, "step": 810 }, { "epoch": 0.09436816383523389, "grad_norm": 1.2418042421340942, "learning_rate": 0.0005, "loss": 3.9973509311676025, "step": 811 }, { "epoch": 0.09448452408657203, "grad_norm": 1.6094856262207031, "learning_rate": 0.0005, "loss": 3.7394025325775146, "step": 812 }, { "epoch": 0.09460088433791017, "grad_norm": 1.320062518119812, "learning_rate": 0.0005, "loss": 2.326014757156372, "step": 813 }, { "epoch": 0.09471724458924831, "grad_norm": 0.9458709955215454, "learning_rate": 0.0005, "loss": 1.9655994176864624, "step": 814 }, { "epoch": 0.09483360484058645, "grad_norm": 1.424687147140503, "learning_rate": 0.0005, "loss": 2.1257224082946777, "step": 815 }, { "epoch": 0.0949499650919246, "grad_norm": 1.5389761924743652, "learning_rate": 0.0005, "loss": 2.8777685165405273, "step": 816 }, { "epoch": 0.09506632534326274, "grad_norm": 1.6755635738372803, "learning_rate": 0.0005, "loss": 2.945033073425293, "step": 817 }, { "epoch": 0.09518268559460089, "grad_norm": 1.193750023841858, "learning_rate": 0.0005, "loss": 2.8448433876037598, "step": 818 }, { "epoch": 0.09529904584593903, "grad_norm": 1.695380687713623, "learning_rate": 0.0005, "loss": 3.3993682861328125, "step": 819 }, { "epoch": 0.09541540609727717, "grad_norm": 2.445915699005127, "learning_rate": 0.0005, "loss": 3.2858195304870605, "step": 820 }, { "epoch": 0.09553176634861531, "grad_norm": 2.5163726806640625, "learning_rate": 0.0005, "loss": 3.1831893920898438, "step": 821 }, { "epoch": 0.09564812659995346, "grad_norm": 2.7517011165618896, "learning_rate": 0.0005, "loss": 3.867307186126709, "step": 822 }, { "epoch": 0.0957644868512916, "grad_norm": 1.8447110652923584, "learning_rate": 0.0005, "loss": 2.5974674224853516, "step": 823 }, { "epoch": 0.09588084710262974, "grad_norm": 1.3565363883972168, "learning_rate": 0.0005, "loss": 3.084824562072754, "step": 824 }, { "epoch": 0.09599720735396788, "grad_norm": 3.050870418548584, "learning_rate": 0.0005, "loss": 3.3214778900146484, "step": 825 }, { "epoch": 0.09611356760530602, "grad_norm": 1.5361419916152954, "learning_rate": 0.0005, "loss": 3.495621681213379, "step": 826 }, { "epoch": 0.09622992785664417, "grad_norm": 1.1021103858947754, "learning_rate": 0.0005, "loss": 2.639112949371338, "step": 827 }, { "epoch": 0.09634628810798232, "grad_norm": 1.2833772897720337, "learning_rate": 0.0005, "loss": 2.8036186695098877, "step": 828 }, { "epoch": 0.09646264835932046, "grad_norm": 1.3886353969573975, "learning_rate": 0.0005, "loss": 3.523324966430664, "step": 829 }, { "epoch": 0.0965790086106586, "grad_norm": 1.631870150566101, "learning_rate": 0.0005, "loss": 3.1132586002349854, "step": 830 }, { "epoch": 0.09669536886199674, "grad_norm": 1.594771385192871, "learning_rate": 0.0005, "loss": 3.3520395755767822, "step": 831 }, { "epoch": 0.09681172911333488, "grad_norm": 4.565903663635254, "learning_rate": 0.0005, "loss": 3.596346616744995, "step": 832 }, { "epoch": 0.09692808936467302, "grad_norm": 1.6841791868209839, "learning_rate": 0.0005, "loss": 2.5075268745422363, "step": 833 }, { "epoch": 0.09704444961601118, "grad_norm": 1.6824424266815186, "learning_rate": 0.0005, "loss": 2.993873119354248, "step": 834 }, { "epoch": 0.09716080986734932, "grad_norm": 1.694780945777893, "learning_rate": 0.0005, "loss": 2.791520118713379, "step": 835 }, { "epoch": 0.09727717011868746, "grad_norm": 1.7953431606292725, "learning_rate": 0.0005, "loss": 3.7534971237182617, "step": 836 }, { "epoch": 0.0973935303700256, "grad_norm": 2.1428442001342773, "learning_rate": 0.0005, "loss": 2.4285242557525635, "step": 837 }, { "epoch": 0.09750989062136374, "grad_norm": 1.6770631074905396, "learning_rate": 0.0005, "loss": 3.0793004035949707, "step": 838 }, { "epoch": 0.09762625087270188, "grad_norm": 2.7367684841156006, "learning_rate": 0.0005, "loss": 2.6045327186584473, "step": 839 }, { "epoch": 0.09774261112404002, "grad_norm": 1.9546453952789307, "learning_rate": 0.0005, "loss": 4.115060329437256, "step": 840 }, { "epoch": 0.09785897137537818, "grad_norm": 1.681657314300537, "learning_rate": 0.0005, "loss": 3.123420238494873, "step": 841 }, { "epoch": 0.09797533162671632, "grad_norm": 1.8087105751037598, "learning_rate": 0.0005, "loss": 2.598615884780884, "step": 842 }, { "epoch": 0.09809169187805446, "grad_norm": 1.5020184516906738, "learning_rate": 0.0005, "loss": 3.7295045852661133, "step": 843 }, { "epoch": 0.0982080521293926, "grad_norm": 2.2411515712738037, "learning_rate": 0.0005, "loss": 3.8406238555908203, "step": 844 }, { "epoch": 0.09832441238073074, "grad_norm": 1.0366839170455933, "learning_rate": 0.0005, "loss": 1.7204281091690063, "step": 845 }, { "epoch": 0.09844077263206888, "grad_norm": 4.672757625579834, "learning_rate": 0.0005, "loss": 2.897643566131592, "step": 846 }, { "epoch": 0.09855713288340703, "grad_norm": 1.7928950786590576, "learning_rate": 0.0005, "loss": 2.569140672683716, "step": 847 }, { "epoch": 0.09867349313474517, "grad_norm": 1.56406831741333, "learning_rate": 0.0005, "loss": 3.1948745250701904, "step": 848 }, { "epoch": 0.09878985338608332, "grad_norm": 1.6442477703094482, "learning_rate": 0.0005, "loss": 3.346688747406006, "step": 849 }, { "epoch": 0.09890621363742146, "grad_norm": 2.2209360599517822, "learning_rate": 0.0005, "loss": 2.741847038269043, "step": 850 }, { "epoch": 0.0990225738887596, "grad_norm": 2.3332817554473877, "learning_rate": 0.0005, "loss": 4.423539638519287, "step": 851 }, { "epoch": 0.09913893414009774, "grad_norm": 1.586690902709961, "learning_rate": 0.0005, "loss": 2.2620973587036133, "step": 852 }, { "epoch": 0.09925529439143589, "grad_norm": 1.3709179162979126, "learning_rate": 0.0005, "loss": 3.1547393798828125, "step": 853 }, { "epoch": 0.09937165464277403, "grad_norm": 1.9019063711166382, "learning_rate": 0.0005, "loss": 3.992757797241211, "step": 854 }, { "epoch": 0.09948801489411217, "grad_norm": 2.5156421661376953, "learning_rate": 0.0005, "loss": 3.5231213569641113, "step": 855 }, { "epoch": 0.09960437514545031, "grad_norm": 1.4733223915100098, "learning_rate": 0.0005, "loss": 2.7983665466308594, "step": 856 }, { "epoch": 0.09972073539678845, "grad_norm": 1.4298638105392456, "learning_rate": 0.0005, "loss": 2.4823663234710693, "step": 857 }, { "epoch": 0.0998370956481266, "grad_norm": 1.3547927141189575, "learning_rate": 0.0005, "loss": 2.9664692878723145, "step": 858 }, { "epoch": 0.09995345589946475, "grad_norm": 1.4633407592773438, "learning_rate": 0.0005, "loss": 2.1505115032196045, "step": 859 }, { "epoch": 0.10006981615080289, "grad_norm": 1.351035714149475, "learning_rate": 0.0005, "loss": 2.4983973503112793, "step": 860 }, { "epoch": 0.10018617640214103, "grad_norm": 1.3994934558868408, "learning_rate": 0.0005, "loss": 3.0840649604797363, "step": 861 }, { "epoch": 0.10030253665347917, "grad_norm": 1.588141679763794, "learning_rate": 0.0005, "loss": 2.8606395721435547, "step": 862 }, { "epoch": 0.10041889690481731, "grad_norm": 2.7809112071990967, "learning_rate": 0.0005, "loss": 3.59307861328125, "step": 863 }, { "epoch": 0.10053525715615545, "grad_norm": 2.056478977203369, "learning_rate": 0.0005, "loss": 2.6208643913269043, "step": 864 }, { "epoch": 0.10065161740749361, "grad_norm": 1.7787386178970337, "learning_rate": 0.0005, "loss": 2.415426254272461, "step": 865 }, { "epoch": 0.10076797765883175, "grad_norm": 1.0749552249908447, "learning_rate": 0.0005, "loss": 2.9270224571228027, "step": 866 }, { "epoch": 0.10088433791016989, "grad_norm": 1.5145859718322754, "learning_rate": 0.0005, "loss": 3.31779146194458, "step": 867 }, { "epoch": 0.10100069816150803, "grad_norm": 1.966573715209961, "learning_rate": 0.0005, "loss": 3.414325714111328, "step": 868 }, { "epoch": 0.10111705841284617, "grad_norm": 1.479109287261963, "learning_rate": 0.0005, "loss": 2.513913154602051, "step": 869 }, { "epoch": 0.10123341866418431, "grad_norm": 1.879659652709961, "learning_rate": 0.0005, "loss": 2.3279736042022705, "step": 870 }, { "epoch": 0.10134977891552245, "grad_norm": 1.4992876052856445, "learning_rate": 0.0005, "loss": 3.7287344932556152, "step": 871 }, { "epoch": 0.1014661391668606, "grad_norm": 1.1490381956100464, "learning_rate": 0.0005, "loss": 2.6628434658050537, "step": 872 }, { "epoch": 0.10158249941819875, "grad_norm": 1.1030324697494507, "learning_rate": 0.0005, "loss": 2.5495805740356445, "step": 873 }, { "epoch": 0.10169885966953689, "grad_norm": 1.8332682847976685, "learning_rate": 0.0005, "loss": 2.7109909057617188, "step": 874 }, { "epoch": 0.10181521992087503, "grad_norm": 1.8166879415512085, "learning_rate": 0.0005, "loss": 2.589210033416748, "step": 875 }, { "epoch": 0.10193158017221317, "grad_norm": 1.615565299987793, "learning_rate": 0.0005, "loss": 3.323798656463623, "step": 876 }, { "epoch": 0.10204794042355131, "grad_norm": 2.5411581993103027, "learning_rate": 0.0005, "loss": 3.5650887489318848, "step": 877 }, { "epoch": 0.10216430067488946, "grad_norm": 2.014106512069702, "learning_rate": 0.0005, "loss": 3.346585273742676, "step": 878 }, { "epoch": 0.1022806609262276, "grad_norm": 1.7886332273483276, "learning_rate": 0.0005, "loss": 3.4142189025878906, "step": 879 }, { "epoch": 0.10239702117756574, "grad_norm": 1.7743351459503174, "learning_rate": 0.0005, "loss": 3.862621545791626, "step": 880 }, { "epoch": 0.10251338142890389, "grad_norm": 1.328055500984192, "learning_rate": 0.0005, "loss": 2.683272361755371, "step": 881 }, { "epoch": 0.10262974168024203, "grad_norm": 1.5332611799240112, "learning_rate": 0.0005, "loss": 3.228286027908325, "step": 882 }, { "epoch": 0.10274610193158017, "grad_norm": 1.6539602279663086, "learning_rate": 0.0005, "loss": 2.055504560470581, "step": 883 }, { "epoch": 0.10286246218291832, "grad_norm": 2.008864641189575, "learning_rate": 0.0005, "loss": 2.4561824798583984, "step": 884 }, { "epoch": 0.10297882243425646, "grad_norm": 1.3289179801940918, "learning_rate": 0.0005, "loss": 3.4013147354125977, "step": 885 }, { "epoch": 0.1030951826855946, "grad_norm": 1.726192593574524, "learning_rate": 0.0005, "loss": 3.8569483757019043, "step": 886 }, { "epoch": 0.10321154293693274, "grad_norm": 2.0185952186584473, "learning_rate": 0.0005, "loss": 3.297654151916504, "step": 887 }, { "epoch": 0.10332790318827088, "grad_norm": 1.9545679092407227, "learning_rate": 0.0005, "loss": 2.8988442420959473, "step": 888 }, { "epoch": 0.10344426343960902, "grad_norm": 1.1710586547851562, "learning_rate": 0.0005, "loss": 2.172929048538208, "step": 889 }, { "epoch": 0.10356062369094718, "grad_norm": 2.0080983638763428, "learning_rate": 0.0005, "loss": 2.1447677612304688, "step": 890 }, { "epoch": 0.10367698394228532, "grad_norm": 1.6266226768493652, "learning_rate": 0.0005, "loss": 2.402242660522461, "step": 891 }, { "epoch": 0.10379334419362346, "grad_norm": 1.200952410697937, "learning_rate": 0.0005, "loss": 2.433271884918213, "step": 892 }, { "epoch": 0.1039097044449616, "grad_norm": 1.523972749710083, "learning_rate": 0.0005, "loss": 2.3372397422790527, "step": 893 }, { "epoch": 0.10402606469629974, "grad_norm": 1.5483864545822144, "learning_rate": 0.0005, "loss": 2.5077528953552246, "step": 894 }, { "epoch": 0.10414242494763788, "grad_norm": 2.995103597640991, "learning_rate": 0.0005, "loss": 4.4650559425354, "step": 895 }, { "epoch": 0.10425878519897604, "grad_norm": 2.19272518157959, "learning_rate": 0.0005, "loss": 3.323420524597168, "step": 896 }, { "epoch": 0.10437514545031418, "grad_norm": 1.7430497407913208, "learning_rate": 0.0005, "loss": 2.1454381942749023, "step": 897 }, { "epoch": 0.10449150570165232, "grad_norm": 1.98826003074646, "learning_rate": 0.0005, "loss": 3.8278391361236572, "step": 898 }, { "epoch": 0.10460786595299046, "grad_norm": 1.5622836351394653, "learning_rate": 0.0005, "loss": 2.9732165336608887, "step": 899 }, { "epoch": 0.1047242262043286, "grad_norm": 2.9477651119232178, "learning_rate": 0.0005, "loss": 2.1289470195770264, "step": 900 }, { "epoch": 0.10484058645566674, "grad_norm": 1.229727864265442, "learning_rate": 0.0005, "loss": 2.6064343452453613, "step": 901 }, { "epoch": 0.10495694670700488, "grad_norm": 2.226315975189209, "learning_rate": 0.0005, "loss": 3.542391538619995, "step": 902 }, { "epoch": 0.10507330695834304, "grad_norm": 1.4582396745681763, "learning_rate": 0.0005, "loss": 2.3531956672668457, "step": 903 }, { "epoch": 0.10518966720968118, "grad_norm": 2.109858274459839, "learning_rate": 0.0005, "loss": 2.664353370666504, "step": 904 }, { "epoch": 0.10530602746101932, "grad_norm": 1.5863817930221558, "learning_rate": 0.0005, "loss": 1.9692188501358032, "step": 905 }, { "epoch": 0.10542238771235746, "grad_norm": 1.80977201461792, "learning_rate": 0.0005, "loss": 2.3862743377685547, "step": 906 }, { "epoch": 0.1055387479636956, "grad_norm": 1.8480857610702515, "learning_rate": 0.0005, "loss": 2.280177116394043, "step": 907 }, { "epoch": 0.10565510821503374, "grad_norm": 2.1234350204467773, "learning_rate": 0.0005, "loss": 3.4788899421691895, "step": 908 }, { "epoch": 0.10577146846637189, "grad_norm": 0.9453167915344238, "learning_rate": 0.0005, "loss": 1.498813271522522, "step": 909 }, { "epoch": 0.10588782871771003, "grad_norm": 1.003427267074585, "learning_rate": 0.0005, "loss": 1.933509111404419, "step": 910 }, { "epoch": 0.10600418896904817, "grad_norm": 1.3794406652450562, "learning_rate": 0.0005, "loss": 3.0505547523498535, "step": 911 }, { "epoch": 0.10612054922038631, "grad_norm": 1.7293777465820312, "learning_rate": 0.0005, "loss": 3.1006736755371094, "step": 912 }, { "epoch": 0.10623690947172446, "grad_norm": 2.2421088218688965, "learning_rate": 0.0005, "loss": 3.8465218544006348, "step": 913 }, { "epoch": 0.1063532697230626, "grad_norm": 1.6412031650543213, "learning_rate": 0.0005, "loss": 3.146040439605713, "step": 914 }, { "epoch": 0.10646962997440075, "grad_norm": 1.2343034744262695, "learning_rate": 0.0005, "loss": 2.653201103210449, "step": 915 }, { "epoch": 0.10658599022573889, "grad_norm": 1.4522041082382202, "learning_rate": 0.0005, "loss": 2.3166239261627197, "step": 916 }, { "epoch": 0.10670235047707703, "grad_norm": 2.0722403526306152, "learning_rate": 0.0005, "loss": 3.696424961090088, "step": 917 }, { "epoch": 0.10681871072841517, "grad_norm": 1.5374417304992676, "learning_rate": 0.0005, "loss": 3.0583362579345703, "step": 918 }, { "epoch": 0.10693507097975331, "grad_norm": 1.0053869485855103, "learning_rate": 0.0005, "loss": 2.3425965309143066, "step": 919 }, { "epoch": 0.10705143123109145, "grad_norm": 2.3791682720184326, "learning_rate": 0.0005, "loss": 3.088862180709839, "step": 920 }, { "epoch": 0.10716779148242961, "grad_norm": 2.145245313644409, "learning_rate": 0.0005, "loss": 3.4446511268615723, "step": 921 }, { "epoch": 0.10728415173376775, "grad_norm": 1.9289110898971558, "learning_rate": 0.0005, "loss": 3.81783390045166, "step": 922 }, { "epoch": 0.10740051198510589, "grad_norm": 1.7962710857391357, "learning_rate": 0.0005, "loss": 2.6617789268493652, "step": 923 }, { "epoch": 0.10751687223644403, "grad_norm": 1.8578293323516846, "learning_rate": 0.0005, "loss": 2.8854591846466064, "step": 924 }, { "epoch": 0.10763323248778217, "grad_norm": 1.819614052772522, "learning_rate": 0.0005, "loss": 3.0718610286712646, "step": 925 }, { "epoch": 0.10774959273912031, "grad_norm": 1.2544841766357422, "learning_rate": 0.0005, "loss": 2.215911388397217, "step": 926 }, { "epoch": 0.10786595299045847, "grad_norm": 1.203364610671997, "learning_rate": 0.0005, "loss": 2.159878730773926, "step": 927 }, { "epoch": 0.10798231324179661, "grad_norm": 3.2898902893066406, "learning_rate": 0.0005, "loss": 2.903494358062744, "step": 928 }, { "epoch": 0.10809867349313475, "grad_norm": 1.679819941520691, "learning_rate": 0.0005, "loss": 2.7370638847351074, "step": 929 }, { "epoch": 0.10821503374447289, "grad_norm": 1.2175593376159668, "learning_rate": 0.0005, "loss": 2.0513994693756104, "step": 930 }, { "epoch": 0.10833139399581103, "grad_norm": 1.3255860805511475, "learning_rate": 0.0005, "loss": 3.1115951538085938, "step": 931 }, { "epoch": 0.10844775424714917, "grad_norm": 1.0445244312286377, "learning_rate": 0.0005, "loss": 2.0108563899993896, "step": 932 }, { "epoch": 0.10856411449848731, "grad_norm": 1.3611421585083008, "learning_rate": 0.0005, "loss": 2.467533588409424, "step": 933 }, { "epoch": 0.10868047474982546, "grad_norm": 1.9333813190460205, "learning_rate": 0.0005, "loss": 2.833292007446289, "step": 934 }, { "epoch": 0.1087968350011636, "grad_norm": 1.7357898950576782, "learning_rate": 0.0005, "loss": 3.978862762451172, "step": 935 }, { "epoch": 0.10891319525250175, "grad_norm": 1.4424545764923096, "learning_rate": 0.0005, "loss": 2.8271429538726807, "step": 936 }, { "epoch": 0.10902955550383989, "grad_norm": 1.3981579542160034, "learning_rate": 0.0005, "loss": 3.1698079109191895, "step": 937 }, { "epoch": 0.10914591575517803, "grad_norm": 1.439723253250122, "learning_rate": 0.0005, "loss": 3.165797710418701, "step": 938 }, { "epoch": 0.10926227600651617, "grad_norm": 1.5086069107055664, "learning_rate": 0.0005, "loss": 2.839114189147949, "step": 939 }, { "epoch": 0.10937863625785432, "grad_norm": 2.1540017127990723, "learning_rate": 0.0005, "loss": 3.392300605773926, "step": 940 }, { "epoch": 0.10949499650919246, "grad_norm": 1.379219889640808, "learning_rate": 0.0005, "loss": 4.934439659118652, "step": 941 }, { "epoch": 0.1096113567605306, "grad_norm": 1.8470816612243652, "learning_rate": 0.0005, "loss": 3.050569772720337, "step": 942 }, { "epoch": 0.10972771701186874, "grad_norm": 1.690447449684143, "learning_rate": 0.0005, "loss": 3.0211946964263916, "step": 943 }, { "epoch": 0.10984407726320689, "grad_norm": 2.081968307495117, "learning_rate": 0.0005, "loss": 3.21653413772583, "step": 944 }, { "epoch": 0.10996043751454503, "grad_norm": 1.376754641532898, "learning_rate": 0.0005, "loss": 2.592280864715576, "step": 945 }, { "epoch": 0.11007679776588318, "grad_norm": 1.8861291408538818, "learning_rate": 0.0005, "loss": 2.220724582672119, "step": 946 }, { "epoch": 0.11019315801722132, "grad_norm": 1.4525763988494873, "learning_rate": 0.0005, "loss": 3.1068921089172363, "step": 947 }, { "epoch": 0.11030951826855946, "grad_norm": 1.6248224973678589, "learning_rate": 0.0005, "loss": 3.4502501487731934, "step": 948 }, { "epoch": 0.1104258785198976, "grad_norm": 1.8883899450302124, "learning_rate": 0.0005, "loss": 3.5865511894226074, "step": 949 }, { "epoch": 0.11054223877123574, "grad_norm": 1.5915381908416748, "learning_rate": 0.0005, "loss": 2.844007968902588, "step": 950 }, { "epoch": 0.11065859902257388, "grad_norm": 3.055647611618042, "learning_rate": 0.0005, "loss": 3.809166431427002, "step": 951 }, { "epoch": 0.11077495927391204, "grad_norm": 1.3520376682281494, "learning_rate": 0.0005, "loss": 2.6316280364990234, "step": 952 }, { "epoch": 0.11089131952525018, "grad_norm": 1.7658408880233765, "learning_rate": 0.0005, "loss": 2.749207019805908, "step": 953 }, { "epoch": 0.11100767977658832, "grad_norm": 1.6544650793075562, "learning_rate": 0.0005, "loss": 4.269392490386963, "step": 954 }, { "epoch": 0.11112404002792646, "grad_norm": 1.4064911603927612, "learning_rate": 0.0005, "loss": 1.852072834968567, "step": 955 }, { "epoch": 0.1112404002792646, "grad_norm": 1.8112343549728394, "learning_rate": 0.0005, "loss": 3.0463006496429443, "step": 956 }, { "epoch": 0.11135676053060274, "grad_norm": 1.2533773183822632, "learning_rate": 0.0005, "loss": 3.1538121700286865, "step": 957 }, { "epoch": 0.1114731207819409, "grad_norm": 1.7545520067214966, "learning_rate": 0.0005, "loss": 3.6736721992492676, "step": 958 }, { "epoch": 0.11158948103327904, "grad_norm": 1.3052898645401, "learning_rate": 0.0005, "loss": 3.4465456008911133, "step": 959 }, { "epoch": 0.11170584128461718, "grad_norm": 1.07697594165802, "learning_rate": 0.0005, "loss": 3.0751020908355713, "step": 960 }, { "epoch": 0.11182220153595532, "grad_norm": 1.5695326328277588, "learning_rate": 0.0005, "loss": 3.368069648742676, "step": 961 }, { "epoch": 0.11193856178729346, "grad_norm": 1.4163048267364502, "learning_rate": 0.0005, "loss": 2.6051881313323975, "step": 962 }, { "epoch": 0.1120549220386316, "grad_norm": 1.5914324522018433, "learning_rate": 0.0005, "loss": 3.686537265777588, "step": 963 }, { "epoch": 0.11217128228996974, "grad_norm": 1.3910282850265503, "learning_rate": 0.0005, "loss": 3.3043057918548584, "step": 964 }, { "epoch": 0.1122876425413079, "grad_norm": 1.4753278493881226, "learning_rate": 0.0005, "loss": 2.9810891151428223, "step": 965 }, { "epoch": 0.11240400279264604, "grad_norm": 1.1816507577896118, "learning_rate": 0.0005, "loss": 3.8752493858337402, "step": 966 }, { "epoch": 0.11252036304398418, "grad_norm": 1.743349552154541, "learning_rate": 0.0005, "loss": 2.630587577819824, "step": 967 }, { "epoch": 0.11263672329532232, "grad_norm": 2.534611701965332, "learning_rate": 0.0005, "loss": 3.111593246459961, "step": 968 }, { "epoch": 0.11275308354666046, "grad_norm": 3.2426347732543945, "learning_rate": 0.0005, "loss": 3.535088300704956, "step": 969 }, { "epoch": 0.1128694437979986, "grad_norm": 1.6830507516860962, "learning_rate": 0.0005, "loss": 2.846632480621338, "step": 970 }, { "epoch": 0.11298580404933675, "grad_norm": 1.781771183013916, "learning_rate": 0.0005, "loss": 2.2154808044433594, "step": 971 }, { "epoch": 0.11310216430067489, "grad_norm": 2.349761486053467, "learning_rate": 0.0005, "loss": 3.994173049926758, "step": 972 }, { "epoch": 0.11321852455201303, "grad_norm": 1.9365158081054688, "learning_rate": 0.0005, "loss": 3.0403404235839844, "step": 973 }, { "epoch": 0.11333488480335117, "grad_norm": 1.163610577583313, "learning_rate": 0.0005, "loss": 2.3089287281036377, "step": 974 }, { "epoch": 0.11345124505468931, "grad_norm": 1.3388068675994873, "learning_rate": 0.0005, "loss": 2.8586413860321045, "step": 975 }, { "epoch": 0.11356760530602746, "grad_norm": 1.586579442024231, "learning_rate": 0.0005, "loss": 2.6769509315490723, "step": 976 }, { "epoch": 0.11368396555736561, "grad_norm": 1.1204932928085327, "learning_rate": 0.0005, "loss": 2.5634894371032715, "step": 977 }, { "epoch": 0.11380032580870375, "grad_norm": 2.1098368167877197, "learning_rate": 0.0005, "loss": 3.94299578666687, "step": 978 }, { "epoch": 0.11391668606004189, "grad_norm": 2.089972496032715, "learning_rate": 0.0005, "loss": 3.320288896560669, "step": 979 }, { "epoch": 0.11403304631138003, "grad_norm": 1.4209843873977661, "learning_rate": 0.0005, "loss": 3.641594648361206, "step": 980 }, { "epoch": 0.11414940656271817, "grad_norm": 1.541580080986023, "learning_rate": 0.0005, "loss": 3.7820916175842285, "step": 981 }, { "epoch": 0.11426576681405631, "grad_norm": 1.521388292312622, "learning_rate": 0.0005, "loss": 2.166182279586792, "step": 982 }, { "epoch": 0.11438212706539447, "grad_norm": 1.4705302715301514, "learning_rate": 0.0005, "loss": 2.1767752170562744, "step": 983 }, { "epoch": 0.11449848731673261, "grad_norm": 0.921963632106781, "learning_rate": 0.0005, "loss": 2.001033306121826, "step": 984 }, { "epoch": 0.11461484756807075, "grad_norm": 1.8604081869125366, "learning_rate": 0.0005, "loss": 3.1111278533935547, "step": 985 }, { "epoch": 0.11473120781940889, "grad_norm": 2.743561029434204, "learning_rate": 0.0005, "loss": 3.489243984222412, "step": 986 }, { "epoch": 0.11484756807074703, "grad_norm": 1.8716753721237183, "learning_rate": 0.0005, "loss": 3.013861894607544, "step": 987 }, { "epoch": 0.11496392832208517, "grad_norm": 1.0212959051132202, "learning_rate": 0.0005, "loss": 1.540539264678955, "step": 988 }, { "epoch": 0.11508028857342333, "grad_norm": 1.6473785638809204, "learning_rate": 0.0005, "loss": 2.9196062088012695, "step": 989 }, { "epoch": 0.11519664882476147, "grad_norm": 1.9215309619903564, "learning_rate": 0.0005, "loss": 2.3225438594818115, "step": 990 }, { "epoch": 0.1153130090760996, "grad_norm": 1.8997764587402344, "learning_rate": 0.0005, "loss": 3.0220704078674316, "step": 991 }, { "epoch": 0.11542936932743775, "grad_norm": 2.1402080059051514, "learning_rate": 0.0005, "loss": 3.720740795135498, "step": 992 }, { "epoch": 0.11554572957877589, "grad_norm": 1.9110528230667114, "learning_rate": 0.0005, "loss": 3.6200413703918457, "step": 993 }, { "epoch": 0.11566208983011403, "grad_norm": 1.1250066757202148, "learning_rate": 0.0005, "loss": 3.1409831047058105, "step": 994 }, { "epoch": 0.11577845008145217, "grad_norm": 1.9343516826629639, "learning_rate": 0.0005, "loss": 3.407529354095459, "step": 995 }, { "epoch": 0.11589481033279032, "grad_norm": 1.6144293546676636, "learning_rate": 0.0005, "loss": 4.040177345275879, "step": 996 }, { "epoch": 0.11601117058412846, "grad_norm": 1.4076095819473267, "learning_rate": 0.0005, "loss": 2.6073315143585205, "step": 997 }, { "epoch": 0.1161275308354666, "grad_norm": 2.29235577583313, "learning_rate": 0.0005, "loss": 3.2853200435638428, "step": 998 }, { "epoch": 0.11624389108680475, "grad_norm": 1.6054328680038452, "learning_rate": 0.0005, "loss": 4.155269622802734, "step": 999 }, { "epoch": 0.11636025133814289, "grad_norm": 1.4023011922836304, "learning_rate": 0.0005, "loss": 2.301267147064209, "step": 1000 }, { "epoch": 0.11647661158948103, "grad_norm": 1.3838239908218384, "learning_rate": 0.0005, "loss": 3.0546984672546387, "step": 1001 }, { "epoch": 0.11659297184081918, "grad_norm": 3.3050038814544678, "learning_rate": 0.0005, "loss": 3.876274824142456, "step": 1002 }, { "epoch": 0.11670933209215732, "grad_norm": 2.2697999477386475, "learning_rate": 0.0005, "loss": 3.8324172496795654, "step": 1003 }, { "epoch": 0.11682569234349546, "grad_norm": 0.9514665007591248, "learning_rate": 0.0005, "loss": 2.1612257957458496, "step": 1004 }, { "epoch": 0.1169420525948336, "grad_norm": 1.4117294549942017, "learning_rate": 0.0005, "loss": 2.6187305450439453, "step": 1005 }, { "epoch": 0.11705841284617174, "grad_norm": 1.5039042234420776, "learning_rate": 0.0005, "loss": 2.7158994674682617, "step": 1006 }, { "epoch": 0.11717477309750989, "grad_norm": 1.9607737064361572, "learning_rate": 0.0005, "loss": 2.9681482315063477, "step": 1007 }, { "epoch": 0.11729113334884804, "grad_norm": 0.9345232844352722, "learning_rate": 0.0005, "loss": 1.9494991302490234, "step": 1008 }, { "epoch": 0.11740749360018618, "grad_norm": 1.800058364868164, "learning_rate": 0.0005, "loss": 2.859384059906006, "step": 1009 }, { "epoch": 0.11752385385152432, "grad_norm": 1.3666576147079468, "learning_rate": 0.0005, "loss": 3.357834815979004, "step": 1010 }, { "epoch": 0.11764021410286246, "grad_norm": 1.5345925092697144, "learning_rate": 0.0005, "loss": 3.1048574447631836, "step": 1011 }, { "epoch": 0.1177565743542006, "grad_norm": 1.3965930938720703, "learning_rate": 0.0005, "loss": 3.3453316688537598, "step": 1012 }, { "epoch": 0.11787293460553874, "grad_norm": 1.9470493793487549, "learning_rate": 0.0005, "loss": 3.420276641845703, "step": 1013 }, { "epoch": 0.1179892948568769, "grad_norm": 1.2639479637145996, "learning_rate": 0.0005, "loss": 2.168743133544922, "step": 1014 }, { "epoch": 0.11810565510821504, "grad_norm": 1.8157931566238403, "learning_rate": 0.0005, "loss": 3.077977180480957, "step": 1015 }, { "epoch": 0.11822201535955318, "grad_norm": 1.5362200736999512, "learning_rate": 0.0005, "loss": 3.2560362815856934, "step": 1016 }, { "epoch": 0.11833837561089132, "grad_norm": 1.2419425249099731, "learning_rate": 0.0005, "loss": 2.8936479091644287, "step": 1017 }, { "epoch": 0.11845473586222946, "grad_norm": 1.9777984619140625, "learning_rate": 0.0005, "loss": 4.522703170776367, "step": 1018 }, { "epoch": 0.1185710961135676, "grad_norm": 1.7144265174865723, "learning_rate": 0.0005, "loss": 3.5970418453216553, "step": 1019 }, { "epoch": 0.11868745636490576, "grad_norm": 1.4444557428359985, "learning_rate": 0.0005, "loss": 2.8353822231292725, "step": 1020 }, { "epoch": 0.1188038166162439, "grad_norm": 1.5350289344787598, "learning_rate": 0.0005, "loss": 4.122879981994629, "step": 1021 }, { "epoch": 0.11892017686758204, "grad_norm": 1.5863550901412964, "learning_rate": 0.0005, "loss": 3.0912837982177734, "step": 1022 }, { "epoch": 0.11903653711892018, "grad_norm": 1.0356968641281128, "learning_rate": 0.0005, "loss": 2.719604015350342, "step": 1023 }, { "epoch": 0.11915289737025832, "grad_norm": 2.172177314758301, "learning_rate": 0.0005, "loss": 3.609288215637207, "step": 1024 }, { "epoch": 0.11926925762159646, "grad_norm": 1.3519492149353027, "learning_rate": 0.0005, "loss": 2.6706302165985107, "step": 1025 }, { "epoch": 0.1193856178729346, "grad_norm": 1.377647876739502, "learning_rate": 0.0005, "loss": 3.6153063774108887, "step": 1026 }, { "epoch": 0.11950197812427275, "grad_norm": 1.1637029647827148, "learning_rate": 0.0005, "loss": 2.258821487426758, "step": 1027 }, { "epoch": 0.1196183383756109, "grad_norm": 1.2497912645339966, "learning_rate": 0.0005, "loss": 3.6315183639526367, "step": 1028 }, { "epoch": 0.11973469862694903, "grad_norm": 1.5153230428695679, "learning_rate": 0.0005, "loss": 3.6919784545898438, "step": 1029 }, { "epoch": 0.11985105887828718, "grad_norm": 1.378698706626892, "learning_rate": 0.0005, "loss": 2.5557796955108643, "step": 1030 }, { "epoch": 0.11996741912962532, "grad_norm": 2.082376480102539, "learning_rate": 0.0005, "loss": 3.329050064086914, "step": 1031 }, { "epoch": 0.12008377938096346, "grad_norm": 1.7622196674346924, "learning_rate": 0.0005, "loss": 3.1830897331237793, "step": 1032 }, { "epoch": 0.12020013963230161, "grad_norm": 2.0747225284576416, "learning_rate": 0.0005, "loss": 3.093928575515747, "step": 1033 }, { "epoch": 0.12031649988363975, "grad_norm": 1.7712424993515015, "learning_rate": 0.0005, "loss": 4.139501571655273, "step": 1034 }, { "epoch": 0.12043286013497789, "grad_norm": 1.9454553127288818, "learning_rate": 0.0005, "loss": 3.251155376434326, "step": 1035 }, { "epoch": 0.12054922038631603, "grad_norm": 1.608168601989746, "learning_rate": 0.0005, "loss": 2.667921543121338, "step": 1036 }, { "epoch": 0.12066558063765417, "grad_norm": 1.7343401908874512, "learning_rate": 0.0005, "loss": 3.142469644546509, "step": 1037 }, { "epoch": 0.12078194088899231, "grad_norm": 1.7670176029205322, "learning_rate": 0.0005, "loss": 2.7775723934173584, "step": 1038 }, { "epoch": 0.12089830114033047, "grad_norm": 2.016369581222534, "learning_rate": 0.0005, "loss": 3.2641024589538574, "step": 1039 }, { "epoch": 0.12101466139166861, "grad_norm": 1.2174581289291382, "learning_rate": 0.0005, "loss": 2.1977808475494385, "step": 1040 }, { "epoch": 0.12113102164300675, "grad_norm": 1.5736727714538574, "learning_rate": 0.0005, "loss": 2.372089385986328, "step": 1041 }, { "epoch": 0.12124738189434489, "grad_norm": 2.5300724506378174, "learning_rate": 0.0005, "loss": 3.588496685028076, "step": 1042 }, { "epoch": 0.12136374214568303, "grad_norm": 1.8736765384674072, "learning_rate": 0.0005, "loss": 1.9062526226043701, "step": 1043 }, { "epoch": 0.12148010239702117, "grad_norm": 1.256761074066162, "learning_rate": 0.0005, "loss": 2.692701816558838, "step": 1044 }, { "epoch": 0.12159646264835933, "grad_norm": 1.667963981628418, "learning_rate": 0.0005, "loss": 3.0356156826019287, "step": 1045 }, { "epoch": 0.12171282289969747, "grad_norm": 1.4766185283660889, "learning_rate": 0.0005, "loss": 3.1742007732391357, "step": 1046 }, { "epoch": 0.12182918315103561, "grad_norm": 1.0169633626937866, "learning_rate": 0.0005, "loss": 2.508636713027954, "step": 1047 }, { "epoch": 0.12194554340237375, "grad_norm": 2.8421478271484375, "learning_rate": 0.0005, "loss": 2.0381369590759277, "step": 1048 }, { "epoch": 0.12206190365371189, "grad_norm": 1.9338887929916382, "learning_rate": 0.0005, "loss": 3.768437623977661, "step": 1049 }, { "epoch": 0.12217826390505003, "grad_norm": 0.8075014352798462, "learning_rate": 0.0005, "loss": 1.382634162902832, "step": 1050 }, { "epoch": 0.12229462415638818, "grad_norm": 2.194216012954712, "learning_rate": 0.0005, "loss": 1.71846604347229, "step": 1051 }, { "epoch": 0.12241098440772633, "grad_norm": 2.062145948410034, "learning_rate": 0.0005, "loss": 4.008426666259766, "step": 1052 }, { "epoch": 0.12252734465906447, "grad_norm": 1.3940693140029907, "learning_rate": 0.0005, "loss": 3.351222038269043, "step": 1053 }, { "epoch": 0.1226437049104026, "grad_norm": 1.5155006647109985, "learning_rate": 0.0005, "loss": 2.0707178115844727, "step": 1054 }, { "epoch": 0.12276006516174075, "grad_norm": 1.6117401123046875, "learning_rate": 0.0005, "loss": 2.585775375366211, "step": 1055 }, { "epoch": 0.12287642541307889, "grad_norm": 1.1991280317306519, "learning_rate": 0.0005, "loss": 1.9371531009674072, "step": 1056 }, { "epoch": 0.12299278566441703, "grad_norm": 1.5323009490966797, "learning_rate": 0.0005, "loss": 3.1001412868499756, "step": 1057 }, { "epoch": 0.12310914591575518, "grad_norm": 1.986441731452942, "learning_rate": 0.0005, "loss": 2.6905858516693115, "step": 1058 }, { "epoch": 0.12322550616709332, "grad_norm": 2.2832887172698975, "learning_rate": 0.0005, "loss": 2.6036858558654785, "step": 1059 }, { "epoch": 0.12334186641843146, "grad_norm": 1.422617793083191, "learning_rate": 0.0005, "loss": 3.585294485092163, "step": 1060 }, { "epoch": 0.1234582266697696, "grad_norm": 1.746978998184204, "learning_rate": 0.0005, "loss": 3.089482545852661, "step": 1061 }, { "epoch": 0.12357458692110775, "grad_norm": 1.2223461866378784, "learning_rate": 0.0005, "loss": 2.382389783859253, "step": 1062 }, { "epoch": 0.12369094717244589, "grad_norm": 1.7279739379882812, "learning_rate": 0.0005, "loss": 3.6751441955566406, "step": 1063 }, { "epoch": 0.12380730742378404, "grad_norm": 1.8719154596328735, "learning_rate": 0.0005, "loss": 3.256155014038086, "step": 1064 }, { "epoch": 0.12392366767512218, "grad_norm": 1.4087225198745728, "learning_rate": 0.0005, "loss": 2.6658053398132324, "step": 1065 }, { "epoch": 0.12404002792646032, "grad_norm": 1.1156642436981201, "learning_rate": 0.0005, "loss": 2.2761971950531006, "step": 1066 }, { "epoch": 0.12415638817779846, "grad_norm": 1.0727815628051758, "learning_rate": 0.0005, "loss": 2.2938649654388428, "step": 1067 }, { "epoch": 0.1242727484291366, "grad_norm": 1.3325526714324951, "learning_rate": 0.0005, "loss": 2.807668924331665, "step": 1068 }, { "epoch": 0.12438910868047474, "grad_norm": 0.9774489998817444, "learning_rate": 0.0005, "loss": 2.080841541290283, "step": 1069 }, { "epoch": 0.1245054689318129, "grad_norm": 1.5690456628799438, "learning_rate": 0.0005, "loss": 3.203453779220581, "step": 1070 }, { "epoch": 0.12462182918315104, "grad_norm": 1.3476996421813965, "learning_rate": 0.0005, "loss": 2.806140184402466, "step": 1071 }, { "epoch": 0.12473818943448918, "grad_norm": 1.3371068239212036, "learning_rate": 0.0005, "loss": 2.0817503929138184, "step": 1072 }, { "epoch": 0.12485454968582732, "grad_norm": 2.681757688522339, "learning_rate": 0.0005, "loss": 3.257814884185791, "step": 1073 }, { "epoch": 0.12497090993716546, "grad_norm": 2.1182851791381836, "learning_rate": 0.0005, "loss": 2.198174238204956, "step": 1074 }, { "epoch": 0.1250872701885036, "grad_norm": 1.470100998878479, "learning_rate": 0.0005, "loss": 4.320055961608887, "step": 1075 }, { "epoch": 0.12520363043984176, "grad_norm": 1.596453309059143, "learning_rate": 0.0005, "loss": 3.165013551712036, "step": 1076 }, { "epoch": 0.12531999069117988, "grad_norm": 2.0208895206451416, "learning_rate": 0.0005, "loss": 3.122788667678833, "step": 1077 }, { "epoch": 0.12543635094251804, "grad_norm": 2.0725417137145996, "learning_rate": 0.0005, "loss": 2.7057957649230957, "step": 1078 }, { "epoch": 0.1255527111938562, "grad_norm": 1.212181568145752, "learning_rate": 0.0005, "loss": 2.585014820098877, "step": 1079 }, { "epoch": 0.12566907144519432, "grad_norm": 1.9876421689987183, "learning_rate": 0.0005, "loss": 3.517090082168579, "step": 1080 }, { "epoch": 0.12578543169653247, "grad_norm": 0.872463047504425, "learning_rate": 0.0005, "loss": 1.6714601516723633, "step": 1081 }, { "epoch": 0.1259017919478706, "grad_norm": 1.1492938995361328, "learning_rate": 0.0005, "loss": 2.5008206367492676, "step": 1082 }, { "epoch": 0.12601815219920875, "grad_norm": 1.8576844930648804, "learning_rate": 0.0005, "loss": 4.771984100341797, "step": 1083 }, { "epoch": 0.12613451245054688, "grad_norm": 2.704396963119507, "learning_rate": 0.0005, "loss": 3.301356554031372, "step": 1084 }, { "epoch": 0.12625087270188504, "grad_norm": 1.9163872003555298, "learning_rate": 0.0005, "loss": 3.590040683746338, "step": 1085 }, { "epoch": 0.1263672329532232, "grad_norm": 2.3230762481689453, "learning_rate": 0.0005, "loss": 2.9822380542755127, "step": 1086 }, { "epoch": 0.12648359320456132, "grad_norm": 1.253488540649414, "learning_rate": 0.0005, "loss": 2.5862879753112793, "step": 1087 }, { "epoch": 0.12659995345589947, "grad_norm": 1.9784696102142334, "learning_rate": 0.0005, "loss": 3.2833995819091797, "step": 1088 }, { "epoch": 0.1267163137072376, "grad_norm": 1.2571723461151123, "learning_rate": 0.0005, "loss": 1.8440098762512207, "step": 1089 }, { "epoch": 0.12683267395857575, "grad_norm": 1.7770789861679077, "learning_rate": 0.0005, "loss": 3.579794406890869, "step": 1090 }, { "epoch": 0.12694903420991388, "grad_norm": 2.48284912109375, "learning_rate": 0.0005, "loss": 2.5404350757598877, "step": 1091 }, { "epoch": 0.12706539446125203, "grad_norm": 1.6762226819992065, "learning_rate": 0.0005, "loss": 2.8777670860290527, "step": 1092 }, { "epoch": 0.1271817547125902, "grad_norm": 1.5756412744522095, "learning_rate": 0.0005, "loss": 3.625800371170044, "step": 1093 }, { "epoch": 0.12729811496392832, "grad_norm": 1.884124755859375, "learning_rate": 0.0005, "loss": 3.579761505126953, "step": 1094 }, { "epoch": 0.12741447521526647, "grad_norm": 4.41705846786499, "learning_rate": 0.0005, "loss": 3.1959991455078125, "step": 1095 }, { "epoch": 0.1275308354666046, "grad_norm": 2.2252936363220215, "learning_rate": 0.0005, "loss": 2.6779181957244873, "step": 1096 }, { "epoch": 0.12764719571794275, "grad_norm": 1.8912218809127808, "learning_rate": 0.0005, "loss": 3.3932645320892334, "step": 1097 }, { "epoch": 0.1277635559692809, "grad_norm": 1.4573801755905151, "learning_rate": 0.0005, "loss": 2.2925734519958496, "step": 1098 }, { "epoch": 0.12787991622061903, "grad_norm": 1.7965749502182007, "learning_rate": 0.0005, "loss": 1.8604106903076172, "step": 1099 }, { "epoch": 0.1279962764719572, "grad_norm": 1.2848044633865356, "learning_rate": 0.0005, "loss": 2.4487767219543457, "step": 1100 }, { "epoch": 0.12811263672329531, "grad_norm": 1.6706624031066895, "learning_rate": 0.0005, "loss": 3.776745319366455, "step": 1101 }, { "epoch": 0.12822899697463347, "grad_norm": 1.5049617290496826, "learning_rate": 0.0005, "loss": 3.6659774780273438, "step": 1102 }, { "epoch": 0.1283453572259716, "grad_norm": 1.7841410636901855, "learning_rate": 0.0005, "loss": 2.8089218139648438, "step": 1103 }, { "epoch": 0.12846171747730975, "grad_norm": 1.376047134399414, "learning_rate": 0.0005, "loss": 3.8535003662109375, "step": 1104 }, { "epoch": 0.1285780777286479, "grad_norm": 1.5424031019210815, "learning_rate": 0.0005, "loss": 3.133074998855591, "step": 1105 }, { "epoch": 0.12869443797998603, "grad_norm": 1.3813899755477905, "learning_rate": 0.0005, "loss": 2.414869785308838, "step": 1106 }, { "epoch": 0.1288107982313242, "grad_norm": 0.8265230059623718, "learning_rate": 0.0005, "loss": 2.3144187927246094, "step": 1107 }, { "epoch": 0.1289271584826623, "grad_norm": 1.6895943880081177, "learning_rate": 0.0005, "loss": 3.361924648284912, "step": 1108 }, { "epoch": 0.12904351873400047, "grad_norm": 1.4926340579986572, "learning_rate": 0.0005, "loss": 2.5380194187164307, "step": 1109 }, { "epoch": 0.12915987898533862, "grad_norm": 1.9021977186203003, "learning_rate": 0.0005, "loss": 3.2349610328674316, "step": 1110 }, { "epoch": 0.12927623923667675, "grad_norm": 1.8899376392364502, "learning_rate": 0.0005, "loss": 3.061781883239746, "step": 1111 }, { "epoch": 0.1293925994880149, "grad_norm": 1.7694050073623657, "learning_rate": 0.0005, "loss": 2.7730209827423096, "step": 1112 }, { "epoch": 0.12950895973935303, "grad_norm": 0.9099882245063782, "learning_rate": 0.0005, "loss": 1.4797767400741577, "step": 1113 }, { "epoch": 0.12962531999069118, "grad_norm": 2.595682144165039, "learning_rate": 0.0005, "loss": 3.4624133110046387, "step": 1114 }, { "epoch": 0.1297416802420293, "grad_norm": 1.6250889301300049, "learning_rate": 0.0005, "loss": 3.476848602294922, "step": 1115 }, { "epoch": 0.12985804049336747, "grad_norm": 1.9602813720703125, "learning_rate": 0.0005, "loss": 3.056128740310669, "step": 1116 }, { "epoch": 0.12997440074470562, "grad_norm": 1.5075271129608154, "learning_rate": 0.0005, "loss": 3.0311081409454346, "step": 1117 }, { "epoch": 0.13009076099604375, "grad_norm": 1.7049750089645386, "learning_rate": 0.0005, "loss": 2.829407215118408, "step": 1118 }, { "epoch": 0.1302071212473819, "grad_norm": 1.4644451141357422, "learning_rate": 0.0005, "loss": 2.3161308765411377, "step": 1119 }, { "epoch": 0.13032348149872003, "grad_norm": 1.9653124809265137, "learning_rate": 0.0005, "loss": 2.446566104888916, "step": 1120 }, { "epoch": 0.13043984175005818, "grad_norm": 1.6319701671600342, "learning_rate": 0.0005, "loss": 2.752246618270874, "step": 1121 }, { "epoch": 0.1305562020013963, "grad_norm": 1.6741734743118286, "learning_rate": 0.0005, "loss": 3.223961353302002, "step": 1122 }, { "epoch": 0.13067256225273446, "grad_norm": 1.3498221635818481, "learning_rate": 0.0005, "loss": 3.074716567993164, "step": 1123 }, { "epoch": 0.13078892250407262, "grad_norm": 1.9328330755233765, "learning_rate": 0.0005, "loss": 2.4898064136505127, "step": 1124 }, { "epoch": 0.13090528275541075, "grad_norm": 1.2622218132019043, "learning_rate": 0.0005, "loss": 2.7193846702575684, "step": 1125 }, { "epoch": 0.1310216430067489, "grad_norm": 1.5713133811950684, "learning_rate": 0.0005, "loss": 2.8694024085998535, "step": 1126 }, { "epoch": 0.13113800325808703, "grad_norm": 1.6782160997390747, "learning_rate": 0.0005, "loss": 2.281383991241455, "step": 1127 }, { "epoch": 0.13125436350942518, "grad_norm": 1.77689790725708, "learning_rate": 0.0005, "loss": 2.8840081691741943, "step": 1128 }, { "epoch": 0.13137072376076334, "grad_norm": 1.8946973085403442, "learning_rate": 0.0005, "loss": 3.8964810371398926, "step": 1129 }, { "epoch": 0.13148708401210146, "grad_norm": 1.3771755695343018, "learning_rate": 0.0005, "loss": 3.3567938804626465, "step": 1130 }, { "epoch": 0.13160344426343962, "grad_norm": 1.330999493598938, "learning_rate": 0.0005, "loss": 2.7207140922546387, "step": 1131 }, { "epoch": 0.13171980451477774, "grad_norm": 1.5680346488952637, "learning_rate": 0.0005, "loss": 2.032546043395996, "step": 1132 }, { "epoch": 0.1318361647661159, "grad_norm": 1.4260971546173096, "learning_rate": 0.0005, "loss": 2.4420042037963867, "step": 1133 }, { "epoch": 0.13195252501745403, "grad_norm": 1.7026287317276, "learning_rate": 0.0005, "loss": 3.684373140335083, "step": 1134 }, { "epoch": 0.13206888526879218, "grad_norm": 1.1632437705993652, "learning_rate": 0.0005, "loss": 2.2935402393341064, "step": 1135 }, { "epoch": 0.13218524552013033, "grad_norm": 1.8171706199645996, "learning_rate": 0.0005, "loss": 4.207464218139648, "step": 1136 }, { "epoch": 0.13230160577146846, "grad_norm": 1.4419505596160889, "learning_rate": 0.0005, "loss": 3.3875489234924316, "step": 1137 }, { "epoch": 0.13241796602280662, "grad_norm": 1.5426961183547974, "learning_rate": 0.0005, "loss": 3.7569475173950195, "step": 1138 }, { "epoch": 0.13253432627414474, "grad_norm": 1.067246437072754, "learning_rate": 0.0005, "loss": 2.1240367889404297, "step": 1139 }, { "epoch": 0.1326506865254829, "grad_norm": 1.1351290941238403, "learning_rate": 0.0005, "loss": 2.350966691970825, "step": 1140 }, { "epoch": 0.13276704677682105, "grad_norm": 1.7333426475524902, "learning_rate": 0.0005, "loss": 3.4311509132385254, "step": 1141 }, { "epoch": 0.13288340702815918, "grad_norm": 1.2694453001022339, "learning_rate": 0.0005, "loss": 2.7840213775634766, "step": 1142 }, { "epoch": 0.13299976727949733, "grad_norm": 2.127816677093506, "learning_rate": 0.0005, "loss": 3.960578441619873, "step": 1143 }, { "epoch": 0.13311612753083546, "grad_norm": 2.010190963745117, "learning_rate": 0.0005, "loss": 2.7434768676757812, "step": 1144 }, { "epoch": 0.13323248778217361, "grad_norm": 2.371790647506714, "learning_rate": 0.0005, "loss": 2.733095645904541, "step": 1145 }, { "epoch": 0.13334884803351174, "grad_norm": 1.756447672843933, "learning_rate": 0.0005, "loss": 4.146889686584473, "step": 1146 }, { "epoch": 0.1334652082848499, "grad_norm": 1.6134003400802612, "learning_rate": 0.0005, "loss": 2.791886329650879, "step": 1147 }, { "epoch": 0.13358156853618805, "grad_norm": 2.8080477714538574, "learning_rate": 0.0005, "loss": 3.1325736045837402, "step": 1148 }, { "epoch": 0.13369792878752618, "grad_norm": 2.023634433746338, "learning_rate": 0.0005, "loss": 2.4574999809265137, "step": 1149 }, { "epoch": 0.13381428903886433, "grad_norm": 1.9781432151794434, "learning_rate": 0.0005, "loss": 2.598574161529541, "step": 1150 }, { "epoch": 0.13393064929020246, "grad_norm": 1.5541309118270874, "learning_rate": 0.0005, "loss": 3.628246784210205, "step": 1151 }, { "epoch": 0.1340470095415406, "grad_norm": 1.6861976385116577, "learning_rate": 0.0005, "loss": 3.3636622428894043, "step": 1152 }, { "epoch": 0.13416336979287874, "grad_norm": 2.2556164264678955, "learning_rate": 0.0005, "loss": 2.9142799377441406, "step": 1153 }, { "epoch": 0.1342797300442169, "grad_norm": 0.8996792435646057, "learning_rate": 0.0005, "loss": 2.1275436878204346, "step": 1154 }, { "epoch": 0.13439609029555505, "grad_norm": 1.5120410919189453, "learning_rate": 0.0005, "loss": 3.1611504554748535, "step": 1155 }, { "epoch": 0.13451245054689318, "grad_norm": 1.4231352806091309, "learning_rate": 0.0005, "loss": 3.4921951293945312, "step": 1156 }, { "epoch": 0.13462881079823133, "grad_norm": 2.3820877075195312, "learning_rate": 0.0005, "loss": 4.140105247497559, "step": 1157 }, { "epoch": 0.13474517104956946, "grad_norm": 1.5047739744186401, "learning_rate": 0.0005, "loss": 2.7057833671569824, "step": 1158 }, { "epoch": 0.1348615313009076, "grad_norm": 1.1977043151855469, "learning_rate": 0.0005, "loss": 3.0814976692199707, "step": 1159 }, { "epoch": 0.13497789155224577, "grad_norm": 1.9171512126922607, "learning_rate": 0.0005, "loss": 3.9093594551086426, "step": 1160 }, { "epoch": 0.1350942518035839, "grad_norm": 1.3418214321136475, "learning_rate": 0.0005, "loss": 2.394282579421997, "step": 1161 }, { "epoch": 0.13521061205492205, "grad_norm": 2.0847253799438477, "learning_rate": 0.0005, "loss": 3.5093741416931152, "step": 1162 }, { "epoch": 0.13532697230626017, "grad_norm": 1.3462049961090088, "learning_rate": 0.0005, "loss": 3.0904433727264404, "step": 1163 }, { "epoch": 0.13544333255759833, "grad_norm": 1.816955327987671, "learning_rate": 0.0005, "loss": 2.9889590740203857, "step": 1164 }, { "epoch": 0.13555969280893646, "grad_norm": 1.476987600326538, "learning_rate": 0.0005, "loss": 2.4116899967193604, "step": 1165 }, { "epoch": 0.1356760530602746, "grad_norm": 1.3721040487289429, "learning_rate": 0.0005, "loss": 3.229428768157959, "step": 1166 }, { "epoch": 0.13579241331161276, "grad_norm": 0.9296485781669617, "learning_rate": 0.0005, "loss": 2.033315420150757, "step": 1167 }, { "epoch": 0.1359087735629509, "grad_norm": 1.0089081525802612, "learning_rate": 0.0005, "loss": 1.904714822769165, "step": 1168 }, { "epoch": 0.13602513381428905, "grad_norm": 2.0622074604034424, "learning_rate": 0.0005, "loss": 3.188436985015869, "step": 1169 }, { "epoch": 0.13614149406562717, "grad_norm": 1.851173996925354, "learning_rate": 0.0005, "loss": 3.3018503189086914, "step": 1170 }, { "epoch": 0.13625785431696533, "grad_norm": 1.9734801054000854, "learning_rate": 0.0005, "loss": 3.003459930419922, "step": 1171 }, { "epoch": 0.13637421456830348, "grad_norm": 1.269294023513794, "learning_rate": 0.0005, "loss": 2.3125576972961426, "step": 1172 }, { "epoch": 0.1364905748196416, "grad_norm": 2.1898059844970703, "learning_rate": 0.0005, "loss": 3.665323257446289, "step": 1173 }, { "epoch": 0.13660693507097976, "grad_norm": 1.5357303619384766, "learning_rate": 0.0005, "loss": 3.0341696739196777, "step": 1174 }, { "epoch": 0.1367232953223179, "grad_norm": 1.619503378868103, "learning_rate": 0.0005, "loss": 3.735258102416992, "step": 1175 }, { "epoch": 0.13683965557365604, "grad_norm": 7.921723365783691, "learning_rate": 0.0005, "loss": 3.0735220909118652, "step": 1176 }, { "epoch": 0.13695601582499417, "grad_norm": 1.878461480140686, "learning_rate": 0.0005, "loss": 3.0641326904296875, "step": 1177 }, { "epoch": 0.13707237607633233, "grad_norm": 8.305521011352539, "learning_rate": 0.0005, "loss": 3.0711889266967773, "step": 1178 }, { "epoch": 0.13718873632767048, "grad_norm": 2.0845401287078857, "learning_rate": 0.0005, "loss": 4.104194641113281, "step": 1179 }, { "epoch": 0.1373050965790086, "grad_norm": 1.5992029905319214, "learning_rate": 0.0005, "loss": 3.6514153480529785, "step": 1180 }, { "epoch": 0.13742145683034676, "grad_norm": 1.2901184558868408, "learning_rate": 0.0005, "loss": 2.957993984222412, "step": 1181 }, { "epoch": 0.1375378170816849, "grad_norm": 1.4515324831008911, "learning_rate": 0.0005, "loss": 3.2105650901794434, "step": 1182 }, { "epoch": 0.13765417733302304, "grad_norm": 1.4065589904785156, "learning_rate": 0.0005, "loss": 3.7577736377716064, "step": 1183 }, { "epoch": 0.13777053758436117, "grad_norm": 0.862323522567749, "learning_rate": 0.0005, "loss": 1.684601068496704, "step": 1184 }, { "epoch": 0.13788689783569932, "grad_norm": 1.9608614444732666, "learning_rate": 0.0005, "loss": 2.8668031692504883, "step": 1185 }, { "epoch": 0.13800325808703748, "grad_norm": 1.0436049699783325, "learning_rate": 0.0005, "loss": 1.8255882263183594, "step": 1186 }, { "epoch": 0.1381196183383756, "grad_norm": 2.5870065689086914, "learning_rate": 0.0005, "loss": 3.2135934829711914, "step": 1187 }, { "epoch": 0.13823597858971376, "grad_norm": 1.178883671760559, "learning_rate": 0.0005, "loss": 2.297147274017334, "step": 1188 }, { "epoch": 0.1383523388410519, "grad_norm": 3.2632176876068115, "learning_rate": 0.0005, "loss": 3.1775453090667725, "step": 1189 }, { "epoch": 0.13846869909239004, "grad_norm": 1.6470800638198853, "learning_rate": 0.0005, "loss": 2.3925247192382812, "step": 1190 }, { "epoch": 0.1385850593437282, "grad_norm": 1.6014783382415771, "learning_rate": 0.0005, "loss": 2.790741443634033, "step": 1191 }, { "epoch": 0.13870141959506632, "grad_norm": 1.4252674579620361, "learning_rate": 0.0005, "loss": 2.384843111038208, "step": 1192 }, { "epoch": 0.13881777984640448, "grad_norm": 2.3991658687591553, "learning_rate": 0.0005, "loss": 2.605086326599121, "step": 1193 }, { "epoch": 0.1389341400977426, "grad_norm": 1.395447850227356, "learning_rate": 0.0005, "loss": 2.9023306369781494, "step": 1194 }, { "epoch": 0.13905050034908076, "grad_norm": 1.7108843326568604, "learning_rate": 0.0005, "loss": 3.0074853897094727, "step": 1195 }, { "epoch": 0.13916686060041888, "grad_norm": 1.2883665561676025, "learning_rate": 0.0005, "loss": 2.892524242401123, "step": 1196 }, { "epoch": 0.13928322085175704, "grad_norm": 2.4545257091522217, "learning_rate": 0.0005, "loss": 2.7241249084472656, "step": 1197 }, { "epoch": 0.1393995811030952, "grad_norm": 1.708423137664795, "learning_rate": 0.0005, "loss": 3.6791491508483887, "step": 1198 }, { "epoch": 0.13951594135443332, "grad_norm": 1.4178217649459839, "learning_rate": 0.0005, "loss": 3.5168814659118652, "step": 1199 }, { "epoch": 0.13963230160577147, "grad_norm": 1.275586485862732, "learning_rate": 0.0005, "loss": 2.859586238861084, "step": 1200 }, { "epoch": 0.1397486618571096, "grad_norm": 2.2532153129577637, "learning_rate": 0.0005, "loss": 2.5464158058166504, "step": 1201 }, { "epoch": 0.13986502210844776, "grad_norm": 1.3914402723312378, "learning_rate": 0.0005, "loss": 3.134002923965454, "step": 1202 }, { "epoch": 0.1399813823597859, "grad_norm": 1.7723350524902344, "learning_rate": 0.0005, "loss": 3.0881760120391846, "step": 1203 }, { "epoch": 0.14009774261112404, "grad_norm": 1.3245432376861572, "learning_rate": 0.0005, "loss": 3.6850833892822266, "step": 1204 }, { "epoch": 0.1402141028624622, "grad_norm": 1.6044765710830688, "learning_rate": 0.0005, "loss": 2.41326642036438, "step": 1205 }, { "epoch": 0.14033046311380032, "grad_norm": 1.3161282539367676, "learning_rate": 0.0005, "loss": 3.090850353240967, "step": 1206 }, { "epoch": 0.14044682336513847, "grad_norm": 1.5421135425567627, "learning_rate": 0.0005, "loss": 2.742361068725586, "step": 1207 }, { "epoch": 0.1405631836164766, "grad_norm": 1.9081957340240479, "learning_rate": 0.0005, "loss": 3.238370895385742, "step": 1208 }, { "epoch": 0.14067954386781475, "grad_norm": 1.5849430561065674, "learning_rate": 0.0005, "loss": 2.78751277923584, "step": 1209 }, { "epoch": 0.1407959041191529, "grad_norm": 1.311338186264038, "learning_rate": 0.0005, "loss": 2.8128676414489746, "step": 1210 }, { "epoch": 0.14091226437049104, "grad_norm": 1.2825626134872437, "learning_rate": 0.0005, "loss": 2.665228843688965, "step": 1211 }, { "epoch": 0.1410286246218292, "grad_norm": 0.6258381009101868, "learning_rate": 0.0005, "loss": 1.3320430517196655, "step": 1212 }, { "epoch": 0.14114498487316732, "grad_norm": 1.966069221496582, "learning_rate": 0.0005, "loss": 3.545051336288452, "step": 1213 }, { "epoch": 0.14126134512450547, "grad_norm": 1.3731516599655151, "learning_rate": 0.0005, "loss": 2.502843141555786, "step": 1214 }, { "epoch": 0.1413777053758436, "grad_norm": 0.9002246260643005, "learning_rate": 0.0005, "loss": 1.9443126916885376, "step": 1215 }, { "epoch": 0.14149406562718175, "grad_norm": 1.859285831451416, "learning_rate": 0.0005, "loss": 2.293400764465332, "step": 1216 }, { "epoch": 0.1416104258785199, "grad_norm": 1.2354788780212402, "learning_rate": 0.0005, "loss": 2.348202705383301, "step": 1217 }, { "epoch": 0.14172678612985803, "grad_norm": 1.5294746160507202, "learning_rate": 0.0005, "loss": 2.613985061645508, "step": 1218 }, { "epoch": 0.1418431463811962, "grad_norm": 3.2677001953125, "learning_rate": 0.0005, "loss": 3.5854227542877197, "step": 1219 }, { "epoch": 0.14195950663253432, "grad_norm": 1.7121351957321167, "learning_rate": 0.0005, "loss": 3.439518928527832, "step": 1220 }, { "epoch": 0.14207586688387247, "grad_norm": 1.7701886892318726, "learning_rate": 0.0005, "loss": 3.0910511016845703, "step": 1221 }, { "epoch": 0.14219222713521062, "grad_norm": 1.8444422483444214, "learning_rate": 0.0005, "loss": 3.8709914684295654, "step": 1222 }, { "epoch": 0.14230858738654875, "grad_norm": 2.0785586833953857, "learning_rate": 0.0005, "loss": 2.7421324253082275, "step": 1223 }, { "epoch": 0.1424249476378869, "grad_norm": 1.4021011590957642, "learning_rate": 0.0005, "loss": 3.1966538429260254, "step": 1224 }, { "epoch": 0.14254130788922503, "grad_norm": 1.9097285270690918, "learning_rate": 0.0005, "loss": 2.972684144973755, "step": 1225 }, { "epoch": 0.1426576681405632, "grad_norm": 2.166123390197754, "learning_rate": 0.0005, "loss": 2.4546780586242676, "step": 1226 }, { "epoch": 0.14277402839190131, "grad_norm": 0.9084411859512329, "learning_rate": 0.0005, "loss": 1.8117471933364868, "step": 1227 }, { "epoch": 0.14289038864323947, "grad_norm": 1.478908658027649, "learning_rate": 0.0005, "loss": 2.685188055038452, "step": 1228 }, { "epoch": 0.14300674889457762, "grad_norm": 1.7235755920410156, "learning_rate": 0.0005, "loss": 3.4969124794006348, "step": 1229 }, { "epoch": 0.14312310914591575, "grad_norm": 1.2877141237258911, "learning_rate": 0.0005, "loss": 3.0323026180267334, "step": 1230 }, { "epoch": 0.1432394693972539, "grad_norm": 2.0948610305786133, "learning_rate": 0.0005, "loss": 3.298945188522339, "step": 1231 }, { "epoch": 0.14335582964859203, "grad_norm": 2.4535610675811768, "learning_rate": 0.0005, "loss": 3.3652923107147217, "step": 1232 }, { "epoch": 0.14347218989993019, "grad_norm": 1.8281275033950806, "learning_rate": 0.0005, "loss": 2.9214978218078613, "step": 1233 }, { "epoch": 0.14358855015126834, "grad_norm": 2.0797476768493652, "learning_rate": 0.0005, "loss": 2.721787691116333, "step": 1234 }, { "epoch": 0.14370491040260647, "grad_norm": 1.755271077156067, "learning_rate": 0.0005, "loss": 2.5869803428649902, "step": 1235 }, { "epoch": 0.14382127065394462, "grad_norm": 1.967810034751892, "learning_rate": 0.0005, "loss": 2.5328800678253174, "step": 1236 }, { "epoch": 0.14393763090528275, "grad_norm": 1.418915867805481, "learning_rate": 0.0005, "loss": 3.138594150543213, "step": 1237 }, { "epoch": 0.1440539911566209, "grad_norm": 2.0494024753570557, "learning_rate": 0.0005, "loss": 2.811922073364258, "step": 1238 }, { "epoch": 0.14417035140795903, "grad_norm": 1.4688292741775513, "learning_rate": 0.0005, "loss": 2.4981894493103027, "step": 1239 }, { "epoch": 0.14428671165929718, "grad_norm": 1.1861572265625, "learning_rate": 0.0005, "loss": 3.438173770904541, "step": 1240 }, { "epoch": 0.14440307191063534, "grad_norm": 2.86098051071167, "learning_rate": 0.0005, "loss": 3.557823657989502, "step": 1241 }, { "epoch": 0.14451943216197347, "grad_norm": 1.6491986513137817, "learning_rate": 0.0005, "loss": 2.5738720893859863, "step": 1242 }, { "epoch": 0.14463579241331162, "grad_norm": 1.6674221754074097, "learning_rate": 0.0005, "loss": 3.066058874130249, "step": 1243 }, { "epoch": 0.14475215266464975, "grad_norm": 1.6578329801559448, "learning_rate": 0.0005, "loss": 3.0054805278778076, "step": 1244 }, { "epoch": 0.1448685129159879, "grad_norm": 1.382940411567688, "learning_rate": 0.0005, "loss": 2.3849568367004395, "step": 1245 }, { "epoch": 0.14498487316732603, "grad_norm": 1.2261521816253662, "learning_rate": 0.0005, "loss": 2.0165224075317383, "step": 1246 }, { "epoch": 0.14510123341866418, "grad_norm": 1.3374019861221313, "learning_rate": 0.0005, "loss": 3.1273012161254883, "step": 1247 }, { "epoch": 0.14521759367000234, "grad_norm": 1.196035623550415, "learning_rate": 0.0005, "loss": 2.6971304416656494, "step": 1248 }, { "epoch": 0.14533395392134046, "grad_norm": 3.173363208770752, "learning_rate": 0.0005, "loss": 3.7674179077148438, "step": 1249 }, { "epoch": 0.14545031417267862, "grad_norm": 4.669440269470215, "learning_rate": 0.0005, "loss": 3.1285829544067383, "step": 1250 }, { "epoch": 0.14556667442401675, "grad_norm": 1.3394029140472412, "learning_rate": 0.0005, "loss": 3.353351593017578, "step": 1251 }, { "epoch": 0.1456830346753549, "grad_norm": 1.781904935836792, "learning_rate": 0.0005, "loss": 3.6915361881256104, "step": 1252 }, { "epoch": 0.14579939492669305, "grad_norm": 1.7420424222946167, "learning_rate": 0.0005, "loss": 2.5233306884765625, "step": 1253 }, { "epoch": 0.14591575517803118, "grad_norm": 1.734012246131897, "learning_rate": 0.0005, "loss": 2.2260446548461914, "step": 1254 }, { "epoch": 0.14603211542936934, "grad_norm": 1.6469981670379639, "learning_rate": 0.0005, "loss": 2.173892021179199, "step": 1255 }, { "epoch": 0.14614847568070746, "grad_norm": 1.3761146068572998, "learning_rate": 0.0005, "loss": 2.994448184967041, "step": 1256 }, { "epoch": 0.14626483593204562, "grad_norm": 1.5525376796722412, "learning_rate": 0.0005, "loss": 2.912632942199707, "step": 1257 }, { "epoch": 0.14638119618338374, "grad_norm": 1.8030638694763184, "learning_rate": 0.0005, "loss": 3.6671142578125, "step": 1258 }, { "epoch": 0.1464975564347219, "grad_norm": 19.72126007080078, "learning_rate": 0.0005, "loss": 2.801976203918457, "step": 1259 }, { "epoch": 0.14661391668606005, "grad_norm": 1.1294535398483276, "learning_rate": 0.0005, "loss": 2.993107557296753, "step": 1260 }, { "epoch": 0.14673027693739818, "grad_norm": 3.8078596591949463, "learning_rate": 0.0005, "loss": 2.4098703861236572, "step": 1261 }, { "epoch": 0.14684663718873633, "grad_norm": 1.3452529907226562, "learning_rate": 0.0005, "loss": 3.3955330848693848, "step": 1262 }, { "epoch": 0.14696299744007446, "grad_norm": 2.1339457035064697, "learning_rate": 0.0005, "loss": 2.2044169902801514, "step": 1263 }, { "epoch": 0.14707935769141262, "grad_norm": 1.5673366785049438, "learning_rate": 0.0005, "loss": 2.3710591793060303, "step": 1264 }, { "epoch": 0.14719571794275077, "grad_norm": 2.1346547603607178, "learning_rate": 0.0005, "loss": 3.1265602111816406, "step": 1265 }, { "epoch": 0.1473120781940889, "grad_norm": 1.6953681707382202, "learning_rate": 0.0005, "loss": 3.8638486862182617, "step": 1266 }, { "epoch": 0.14742843844542705, "grad_norm": 0.8903780579566956, "learning_rate": 0.0005, "loss": 1.4374836683273315, "step": 1267 }, { "epoch": 0.14754479869676518, "grad_norm": 1.0994645357131958, "learning_rate": 0.0005, "loss": 2.9737610816955566, "step": 1268 }, { "epoch": 0.14766115894810333, "grad_norm": 1.9598175287246704, "learning_rate": 0.0005, "loss": 4.09943962097168, "step": 1269 }, { "epoch": 0.14777751919944146, "grad_norm": 1.4369163513183594, "learning_rate": 0.0005, "loss": 2.9355521202087402, "step": 1270 }, { "epoch": 0.14789387945077961, "grad_norm": 1.7758277654647827, "learning_rate": 0.0005, "loss": 2.9586892127990723, "step": 1271 }, { "epoch": 0.14801023970211777, "grad_norm": 3.605473041534424, "learning_rate": 0.0005, "loss": 2.520864248275757, "step": 1272 }, { "epoch": 0.1481265999534559, "grad_norm": 1.8423339128494263, "learning_rate": 0.0005, "loss": 2.3219118118286133, "step": 1273 }, { "epoch": 0.14824296020479405, "grad_norm": 1.659783124923706, "learning_rate": 0.0005, "loss": 3.270984172821045, "step": 1274 }, { "epoch": 0.14835932045613218, "grad_norm": 1.2317039966583252, "learning_rate": 0.0005, "loss": 2.7383346557617188, "step": 1275 }, { "epoch": 0.14847568070747033, "grad_norm": 1.7409366369247437, "learning_rate": 0.0005, "loss": 2.898653268814087, "step": 1276 }, { "epoch": 0.14859204095880846, "grad_norm": 1.8966269493103027, "learning_rate": 0.0005, "loss": 2.189739227294922, "step": 1277 }, { "epoch": 0.1487084012101466, "grad_norm": 1.287936806678772, "learning_rate": 0.0005, "loss": 2.5472230911254883, "step": 1278 }, { "epoch": 0.14882476146148477, "grad_norm": 1.972786545753479, "learning_rate": 0.0005, "loss": 2.6829545497894287, "step": 1279 }, { "epoch": 0.1489411217128229, "grad_norm": 1.2525224685668945, "learning_rate": 0.0005, "loss": 2.5142836570739746, "step": 1280 }, { "epoch": 0.14905748196416105, "grad_norm": 1.147956132888794, "learning_rate": 0.0005, "loss": 2.5233707427978516, "step": 1281 }, { "epoch": 0.14917384221549918, "grad_norm": 2.623988151550293, "learning_rate": 0.0005, "loss": 3.352041482925415, "step": 1282 }, { "epoch": 0.14929020246683733, "grad_norm": 1.8976205587387085, "learning_rate": 0.0005, "loss": 3.203500747680664, "step": 1283 }, { "epoch": 0.14940656271817548, "grad_norm": 1.4914804697036743, "learning_rate": 0.0005, "loss": 2.5038938522338867, "step": 1284 }, { "epoch": 0.1495229229695136, "grad_norm": 1.4500045776367188, "learning_rate": 0.0005, "loss": 3.9151408672332764, "step": 1285 }, { "epoch": 0.14963928322085177, "grad_norm": 1.5322238206863403, "learning_rate": 0.0005, "loss": 2.337580442428589, "step": 1286 }, { "epoch": 0.1497556434721899, "grad_norm": 1.5508679151535034, "learning_rate": 0.0005, "loss": 3.8949365615844727, "step": 1287 }, { "epoch": 0.14987200372352805, "grad_norm": 2.976147174835205, "learning_rate": 0.0005, "loss": 2.6961655616760254, "step": 1288 }, { "epoch": 0.14998836397486617, "grad_norm": 2.0389015674591064, "learning_rate": 0.0005, "loss": 3.187824010848999, "step": 1289 }, { "epoch": 0.15010472422620433, "grad_norm": 2.8986711502075195, "learning_rate": 0.0005, "loss": 1.9778069257736206, "step": 1290 }, { "epoch": 0.15022108447754248, "grad_norm": 1.8678762912750244, "learning_rate": 0.0005, "loss": 2.586475372314453, "step": 1291 }, { "epoch": 0.1503374447288806, "grad_norm": 2.049560308456421, "learning_rate": 0.0005, "loss": 4.0162506103515625, "step": 1292 }, { "epoch": 0.15045380498021876, "grad_norm": 1.5162851810455322, "learning_rate": 0.0005, "loss": 3.1607704162597656, "step": 1293 }, { "epoch": 0.1505701652315569, "grad_norm": 1.547920823097229, "learning_rate": 0.0005, "loss": 3.124577045440674, "step": 1294 }, { "epoch": 0.15068652548289505, "grad_norm": 1.8933708667755127, "learning_rate": 0.0005, "loss": 2.8053348064422607, "step": 1295 }, { "epoch": 0.1508028857342332, "grad_norm": 1.5213063955307007, "learning_rate": 0.0005, "loss": 2.029113531112671, "step": 1296 }, { "epoch": 0.15091924598557133, "grad_norm": 2.024714946746826, "learning_rate": 0.0005, "loss": 2.9739389419555664, "step": 1297 }, { "epoch": 0.15103560623690948, "grad_norm": 1.6427042484283447, "learning_rate": 0.0005, "loss": 3.305607318878174, "step": 1298 }, { "epoch": 0.1511519664882476, "grad_norm": 1.324811577796936, "learning_rate": 0.0005, "loss": 3.5781755447387695, "step": 1299 }, { "epoch": 0.15126832673958576, "grad_norm": 2.173323631286621, "learning_rate": 0.0005, "loss": 3.2347488403320312, "step": 1300 }, { "epoch": 0.1513846869909239, "grad_norm": 1.7351391315460205, "learning_rate": 0.0005, "loss": 3.944035530090332, "step": 1301 }, { "epoch": 0.15150104724226204, "grad_norm": 2.0918638706207275, "learning_rate": 0.0005, "loss": 3.3886654376983643, "step": 1302 }, { "epoch": 0.1516174074936002, "grad_norm": 1.5249494314193726, "learning_rate": 0.0005, "loss": 2.472740411758423, "step": 1303 }, { "epoch": 0.15173376774493832, "grad_norm": 2.104567527770996, "learning_rate": 0.0005, "loss": 4.347761154174805, "step": 1304 }, { "epoch": 0.15185012799627648, "grad_norm": 1.7103352546691895, "learning_rate": 0.0005, "loss": 3.2459394931793213, "step": 1305 }, { "epoch": 0.1519664882476146, "grad_norm": 0.8469365239143372, "learning_rate": 0.0005, "loss": 1.4820451736450195, "step": 1306 }, { "epoch": 0.15208284849895276, "grad_norm": 1.3834853172302246, "learning_rate": 0.0005, "loss": 3.624624013900757, "step": 1307 }, { "epoch": 0.1521992087502909, "grad_norm": 1.790384292602539, "learning_rate": 0.0005, "loss": 3.292355537414551, "step": 1308 }, { "epoch": 0.15231556900162904, "grad_norm": 0.9100643992424011, "learning_rate": 0.0005, "loss": 1.9803869724273682, "step": 1309 }, { "epoch": 0.1524319292529672, "grad_norm": 2.0783488750457764, "learning_rate": 0.0005, "loss": 3.289945363998413, "step": 1310 }, { "epoch": 0.15254828950430532, "grad_norm": 1.84855318069458, "learning_rate": 0.0005, "loss": 2.7415311336517334, "step": 1311 }, { "epoch": 0.15266464975564348, "grad_norm": 1.2660781145095825, "learning_rate": 0.0005, "loss": 2.594771385192871, "step": 1312 }, { "epoch": 0.1527810100069816, "grad_norm": 1.7110408544540405, "learning_rate": 0.0005, "loss": 3.5807089805603027, "step": 1313 }, { "epoch": 0.15289737025831976, "grad_norm": 1.6697880029678345, "learning_rate": 0.0005, "loss": 2.2786433696746826, "step": 1314 }, { "epoch": 0.1530137305096579, "grad_norm": 1.717991828918457, "learning_rate": 0.0005, "loss": 3.3842997550964355, "step": 1315 }, { "epoch": 0.15313009076099604, "grad_norm": 2.1903977394104004, "learning_rate": 0.0005, "loss": 2.8325047492980957, "step": 1316 }, { "epoch": 0.1532464510123342, "grad_norm": 2.913641929626465, "learning_rate": 0.0005, "loss": 3.492891311645508, "step": 1317 }, { "epoch": 0.15336281126367232, "grad_norm": 1.5471830368041992, "learning_rate": 0.0005, "loss": 2.742673635482788, "step": 1318 }, { "epoch": 0.15347917151501048, "grad_norm": 1.4664736986160278, "learning_rate": 0.0005, "loss": 3.0773305892944336, "step": 1319 }, { "epoch": 0.1535955317663486, "grad_norm": 1.7548848390579224, "learning_rate": 0.0005, "loss": 2.9083333015441895, "step": 1320 }, { "epoch": 0.15371189201768676, "grad_norm": 1.2225292921066284, "learning_rate": 0.0005, "loss": 3.0440971851348877, "step": 1321 }, { "epoch": 0.1538282522690249, "grad_norm": 2.3297953605651855, "learning_rate": 0.0005, "loss": 2.7947421073913574, "step": 1322 }, { "epoch": 0.15394461252036304, "grad_norm": 1.63193678855896, "learning_rate": 0.0005, "loss": 3.087986946105957, "step": 1323 }, { "epoch": 0.1540609727717012, "grad_norm": 2.537323474884033, "learning_rate": 0.0005, "loss": 3.534731149673462, "step": 1324 }, { "epoch": 0.15417733302303932, "grad_norm": 1.7080413103103638, "learning_rate": 0.0005, "loss": 2.8519208431243896, "step": 1325 }, { "epoch": 0.15429369327437747, "grad_norm": 1.1475238800048828, "learning_rate": 0.0005, "loss": 2.3788626194000244, "step": 1326 }, { "epoch": 0.15441005352571563, "grad_norm": 1.1574487686157227, "learning_rate": 0.0005, "loss": 2.1436734199523926, "step": 1327 }, { "epoch": 0.15452641377705376, "grad_norm": 1.5362590551376343, "learning_rate": 0.0005, "loss": 2.564072847366333, "step": 1328 }, { "epoch": 0.1546427740283919, "grad_norm": 1.7839981317520142, "learning_rate": 0.0005, "loss": 1.7229506969451904, "step": 1329 }, { "epoch": 0.15475913427973004, "grad_norm": 1.233977198600769, "learning_rate": 0.0005, "loss": 2.9341063499450684, "step": 1330 }, { "epoch": 0.1548754945310682, "grad_norm": 1.3291810750961304, "learning_rate": 0.0005, "loss": 2.6130118370056152, "step": 1331 }, { "epoch": 0.15499185478240632, "grad_norm": 1.1918210983276367, "learning_rate": 0.0005, "loss": 2.2348086833953857, "step": 1332 }, { "epoch": 0.15510821503374447, "grad_norm": 1.693474531173706, "learning_rate": 0.0005, "loss": 3.156536102294922, "step": 1333 }, { "epoch": 0.15522457528508263, "grad_norm": 1.2583674192428589, "learning_rate": 0.0005, "loss": 2.4873456954956055, "step": 1334 }, { "epoch": 0.15534093553642075, "grad_norm": 1.8100290298461914, "learning_rate": 0.0005, "loss": 2.286477565765381, "step": 1335 }, { "epoch": 0.1554572957877589, "grad_norm": 1.9051156044006348, "learning_rate": 0.0005, "loss": 3.234467029571533, "step": 1336 }, { "epoch": 0.15557365603909704, "grad_norm": 1.156856656074524, "learning_rate": 0.0005, "loss": 2.5138285160064697, "step": 1337 }, { "epoch": 0.1556900162904352, "grad_norm": 1.896978735923767, "learning_rate": 0.0005, "loss": 4.166560173034668, "step": 1338 }, { "epoch": 0.15580637654177332, "grad_norm": 2.017758369445801, "learning_rate": 0.0005, "loss": 2.968233346939087, "step": 1339 }, { "epoch": 0.15592273679311147, "grad_norm": 4.065412521362305, "learning_rate": 0.0005, "loss": 3.664830207824707, "step": 1340 }, { "epoch": 0.15603909704444963, "grad_norm": 1.8338733911514282, "learning_rate": 0.0005, "loss": 3.2729926109313965, "step": 1341 }, { "epoch": 0.15615545729578775, "grad_norm": 2.032975912094116, "learning_rate": 0.0005, "loss": 2.022468090057373, "step": 1342 }, { "epoch": 0.1562718175471259, "grad_norm": 1.534416675567627, "learning_rate": 0.0005, "loss": 2.920464277267456, "step": 1343 }, { "epoch": 0.15638817779846403, "grad_norm": 1.663926362991333, "learning_rate": 0.0005, "loss": 2.9153389930725098, "step": 1344 }, { "epoch": 0.1565045380498022, "grad_norm": 1.6292798519134521, "learning_rate": 0.0005, "loss": 2.9979052543640137, "step": 1345 }, { "epoch": 0.15662089830114034, "grad_norm": 1.7019487619400024, "learning_rate": 0.0005, "loss": 3.447366952896118, "step": 1346 }, { "epoch": 0.15673725855247847, "grad_norm": 1.6501426696777344, "learning_rate": 0.0005, "loss": 4.028041839599609, "step": 1347 }, { "epoch": 0.15685361880381662, "grad_norm": 1.010796070098877, "learning_rate": 0.0005, "loss": 1.870465874671936, "step": 1348 }, { "epoch": 0.15696997905515475, "grad_norm": 1.6604260206222534, "learning_rate": 0.0005, "loss": 3.4489848613739014, "step": 1349 }, { "epoch": 0.1570863393064929, "grad_norm": 0.9412803053855896, "learning_rate": 0.0005, "loss": 2.086319923400879, "step": 1350 }, { "epoch": 0.15720269955783103, "grad_norm": 2.3271617889404297, "learning_rate": 0.0005, "loss": 4.183286190032959, "step": 1351 }, { "epoch": 0.1573190598091692, "grad_norm": 2.056978702545166, "learning_rate": 0.0005, "loss": 3.4285430908203125, "step": 1352 }, { "epoch": 0.15743542006050734, "grad_norm": 2.2147440910339355, "learning_rate": 0.0005, "loss": 3.609374523162842, "step": 1353 }, { "epoch": 0.15755178031184547, "grad_norm": 2.2341301441192627, "learning_rate": 0.0005, "loss": 3.445528984069824, "step": 1354 }, { "epoch": 0.15766814056318362, "grad_norm": 1.6581374406814575, "learning_rate": 0.0005, "loss": 4.094567775726318, "step": 1355 }, { "epoch": 0.15778450081452175, "grad_norm": 1.8910102844238281, "learning_rate": 0.0005, "loss": 3.1655092239379883, "step": 1356 }, { "epoch": 0.1579008610658599, "grad_norm": 2.050414562225342, "learning_rate": 0.0005, "loss": 3.1970431804656982, "step": 1357 }, { "epoch": 0.15801722131719806, "grad_norm": 1.35541832447052, "learning_rate": 0.0005, "loss": 2.3604414463043213, "step": 1358 }, { "epoch": 0.15813358156853619, "grad_norm": 1.262420415878296, "learning_rate": 0.0005, "loss": 2.0857696533203125, "step": 1359 }, { "epoch": 0.15824994181987434, "grad_norm": 1.8871090412139893, "learning_rate": 0.0005, "loss": 2.732142210006714, "step": 1360 }, { "epoch": 0.15836630207121247, "grad_norm": 1.3894951343536377, "learning_rate": 0.0005, "loss": 3.446260452270508, "step": 1361 }, { "epoch": 0.15848266232255062, "grad_norm": 1.2063063383102417, "learning_rate": 0.0005, "loss": 3.9659230709075928, "step": 1362 }, { "epoch": 0.15859902257388875, "grad_norm": 5.101456165313721, "learning_rate": 0.0005, "loss": 3.0295233726501465, "step": 1363 }, { "epoch": 0.1587153828252269, "grad_norm": 1.9869158267974854, "learning_rate": 0.0005, "loss": 3.3561506271362305, "step": 1364 }, { "epoch": 0.15883174307656506, "grad_norm": 1.1268388032913208, "learning_rate": 0.0005, "loss": 2.563530445098877, "step": 1365 }, { "epoch": 0.15894810332790318, "grad_norm": 2.828127861022949, "learning_rate": 0.0005, "loss": 3.5199408531188965, "step": 1366 }, { "epoch": 0.15906446357924134, "grad_norm": 1.897702932357788, "learning_rate": 0.0005, "loss": 3.2608938217163086, "step": 1367 }, { "epoch": 0.15918082383057947, "grad_norm": 1.766129493713379, "learning_rate": 0.0005, "loss": 3.434648036956787, "step": 1368 }, { "epoch": 0.15929718408191762, "grad_norm": 1.5537935495376587, "learning_rate": 0.0005, "loss": 1.6779277324676514, "step": 1369 }, { "epoch": 0.15941354433325575, "grad_norm": 1.7792454957962036, "learning_rate": 0.0005, "loss": 2.907231330871582, "step": 1370 }, { "epoch": 0.1595299045845939, "grad_norm": 1.2074503898620605, "learning_rate": 0.0005, "loss": 2.6433823108673096, "step": 1371 }, { "epoch": 0.15964626483593206, "grad_norm": 1.5636208057403564, "learning_rate": 0.0005, "loss": 2.200359344482422, "step": 1372 }, { "epoch": 0.15976262508727018, "grad_norm": 1.4633656740188599, "learning_rate": 0.0005, "loss": 3.3690710067749023, "step": 1373 }, { "epoch": 0.15987898533860834, "grad_norm": 6.160268306732178, "learning_rate": 0.0005, "loss": 3.8811495304107666, "step": 1374 }, { "epoch": 0.15999534558994646, "grad_norm": 1.4317172765731812, "learning_rate": 0.0005, "loss": 2.3823766708374023, "step": 1375 }, { "epoch": 0.16011170584128462, "grad_norm": 1.4829127788543701, "learning_rate": 0.0005, "loss": 3.2713770866394043, "step": 1376 }, { "epoch": 0.16022806609262277, "grad_norm": 1.4022045135498047, "learning_rate": 0.0005, "loss": 3.0519769191741943, "step": 1377 }, { "epoch": 0.1603444263439609, "grad_norm": 1.3219099044799805, "learning_rate": 0.0005, "loss": 2.342106342315674, "step": 1378 }, { "epoch": 0.16046078659529905, "grad_norm": 2.9754385948181152, "learning_rate": 0.0005, "loss": 2.9224636554718018, "step": 1379 }, { "epoch": 0.16057714684663718, "grad_norm": 1.2617545127868652, "learning_rate": 0.0005, "loss": 1.9056116342544556, "step": 1380 }, { "epoch": 0.16069350709797534, "grad_norm": 1.7720472812652588, "learning_rate": 0.0005, "loss": 2.469356060028076, "step": 1381 }, { "epoch": 0.16080986734931346, "grad_norm": 1.213746428489685, "learning_rate": 0.0005, "loss": 3.5587410926818848, "step": 1382 }, { "epoch": 0.16092622760065162, "grad_norm": 1.3939893245697021, "learning_rate": 0.0005, "loss": 2.7187511920928955, "step": 1383 }, { "epoch": 0.16104258785198977, "grad_norm": 1.4700466394424438, "learning_rate": 0.0005, "loss": 3.827080488204956, "step": 1384 }, { "epoch": 0.1611589481033279, "grad_norm": 1.405349612236023, "learning_rate": 0.0005, "loss": 2.8724775314331055, "step": 1385 }, { "epoch": 0.16127530835466605, "grad_norm": 1.4146230220794678, "learning_rate": 0.0005, "loss": 3.159381628036499, "step": 1386 }, { "epoch": 0.16139166860600418, "grad_norm": 1.6272644996643066, "learning_rate": 0.0005, "loss": 3.3305954933166504, "step": 1387 }, { "epoch": 0.16150802885734233, "grad_norm": 2.151742935180664, "learning_rate": 0.0005, "loss": 2.4662044048309326, "step": 1388 }, { "epoch": 0.1616243891086805, "grad_norm": 1.7764644622802734, "learning_rate": 0.0005, "loss": 2.7324020862579346, "step": 1389 }, { "epoch": 0.16174074936001862, "grad_norm": 1.520949125289917, "learning_rate": 0.0005, "loss": 2.9593400955200195, "step": 1390 }, { "epoch": 0.16185710961135677, "grad_norm": 2.3112192153930664, "learning_rate": 0.0005, "loss": 3.695084571838379, "step": 1391 }, { "epoch": 0.1619734698626949, "grad_norm": 1.5174751281738281, "learning_rate": 0.0005, "loss": 2.6583542823791504, "step": 1392 }, { "epoch": 0.16208983011403305, "grad_norm": 1.3576359748840332, "learning_rate": 0.0005, "loss": 1.6618975400924683, "step": 1393 }, { "epoch": 0.16220619036537118, "grad_norm": 2.3932788372039795, "learning_rate": 0.0005, "loss": 2.973386287689209, "step": 1394 }, { "epoch": 0.16232255061670933, "grad_norm": 2.399538040161133, "learning_rate": 0.0005, "loss": 2.8678083419799805, "step": 1395 }, { "epoch": 0.1624389108680475, "grad_norm": 1.2087746858596802, "learning_rate": 0.0005, "loss": 2.294276237487793, "step": 1396 }, { "epoch": 0.1625552711193856, "grad_norm": 2.8302743434906006, "learning_rate": 0.0005, "loss": 3.752187728881836, "step": 1397 }, { "epoch": 0.16267163137072377, "grad_norm": 1.035844326019287, "learning_rate": 0.0005, "loss": 2.140115737915039, "step": 1398 }, { "epoch": 0.1627879916220619, "grad_norm": 1.9365355968475342, "learning_rate": 0.0005, "loss": 3.304382085800171, "step": 1399 }, { "epoch": 0.16290435187340005, "grad_norm": 2.9993107318878174, "learning_rate": 0.0005, "loss": 3.2162985801696777, "step": 1400 }, { "epoch": 0.16302071212473818, "grad_norm": 2.067537546157837, "learning_rate": 0.0005, "loss": 2.8493947982788086, "step": 1401 }, { "epoch": 0.16313707237607633, "grad_norm": 1.5614358186721802, "learning_rate": 0.0005, "loss": 3.7486913204193115, "step": 1402 }, { "epoch": 0.16325343262741449, "grad_norm": 1.2116942405700684, "learning_rate": 0.0005, "loss": 2.6942806243896484, "step": 1403 }, { "epoch": 0.1633697928787526, "grad_norm": 1.6736952066421509, "learning_rate": 0.0005, "loss": 3.696082830429077, "step": 1404 }, { "epoch": 0.16348615313009077, "grad_norm": 1.490970492362976, "learning_rate": 0.0005, "loss": 3.003026008605957, "step": 1405 }, { "epoch": 0.1636025133814289, "grad_norm": 1.6171547174453735, "learning_rate": 0.0005, "loss": 3.034665822982788, "step": 1406 }, { "epoch": 0.16371887363276705, "grad_norm": 1.0337468385696411, "learning_rate": 0.0005, "loss": 1.6207304000854492, "step": 1407 }, { "epoch": 0.1638352338841052, "grad_norm": 1.5478510856628418, "learning_rate": 0.0005, "loss": 2.919145107269287, "step": 1408 }, { "epoch": 0.16395159413544333, "grad_norm": 2.5393731594085693, "learning_rate": 0.0005, "loss": 3.2190349102020264, "step": 1409 }, { "epoch": 0.16406795438678148, "grad_norm": 1.7188713550567627, "learning_rate": 0.0005, "loss": 1.8893166780471802, "step": 1410 }, { "epoch": 0.1641843146381196, "grad_norm": 1.0342037677764893, "learning_rate": 0.0005, "loss": 2.0404927730560303, "step": 1411 }, { "epoch": 0.16430067488945777, "grad_norm": 1.0972559452056885, "learning_rate": 0.0005, "loss": 2.1731925010681152, "step": 1412 }, { "epoch": 0.1644170351407959, "grad_norm": 0.8737489581108093, "learning_rate": 0.0005, "loss": 2.140598773956299, "step": 1413 }, { "epoch": 0.16453339539213405, "grad_norm": 1.9075500965118408, "learning_rate": 0.0005, "loss": 3.335751533508301, "step": 1414 }, { "epoch": 0.1646497556434722, "grad_norm": 1.7429486513137817, "learning_rate": 0.0005, "loss": 3.459628105163574, "step": 1415 }, { "epoch": 0.16476611589481033, "grad_norm": 1.2397856712341309, "learning_rate": 0.0005, "loss": 2.7744741439819336, "step": 1416 }, { "epoch": 0.16488247614614848, "grad_norm": 1.4284775257110596, "learning_rate": 0.0005, "loss": 3.677063465118408, "step": 1417 }, { "epoch": 0.1649988363974866, "grad_norm": 1.3169736862182617, "learning_rate": 0.0005, "loss": 3.654026746749878, "step": 1418 }, { "epoch": 0.16511519664882476, "grad_norm": 1.7373757362365723, "learning_rate": 0.0005, "loss": 2.908578872680664, "step": 1419 }, { "epoch": 0.16523155690016292, "grad_norm": 1.6205613613128662, "learning_rate": 0.0005, "loss": 2.5863990783691406, "step": 1420 }, { "epoch": 0.16534791715150104, "grad_norm": 1.7445828914642334, "learning_rate": 0.0005, "loss": 2.468249797821045, "step": 1421 }, { "epoch": 0.1654642774028392, "grad_norm": 1.6066784858703613, "learning_rate": 0.0005, "loss": 2.9056692123413086, "step": 1422 }, { "epoch": 0.16558063765417733, "grad_norm": 1.2865722179412842, "learning_rate": 0.0005, "loss": 2.235666036605835, "step": 1423 }, { "epoch": 0.16569699790551548, "grad_norm": 1.013777256011963, "learning_rate": 0.0005, "loss": 1.8795599937438965, "step": 1424 }, { "epoch": 0.1658133581568536, "grad_norm": 0.9780606031417847, "learning_rate": 0.0005, "loss": 2.211960554122925, "step": 1425 }, { "epoch": 0.16592971840819176, "grad_norm": 1.4850162267684937, "learning_rate": 0.0005, "loss": 2.5737226009368896, "step": 1426 }, { "epoch": 0.16604607865952992, "grad_norm": 1.6445766687393188, "learning_rate": 0.0005, "loss": 2.3139002323150635, "step": 1427 }, { "epoch": 0.16616243891086804, "grad_norm": 1.3171616792678833, "learning_rate": 0.0005, "loss": 2.619267463684082, "step": 1428 }, { "epoch": 0.1662787991622062, "grad_norm": 1.686381459236145, "learning_rate": 0.0005, "loss": 3.904081344604492, "step": 1429 }, { "epoch": 0.16639515941354432, "grad_norm": 1.9897873401641846, "learning_rate": 0.0005, "loss": 2.954641580581665, "step": 1430 }, { "epoch": 0.16651151966488248, "grad_norm": 1.667549729347229, "learning_rate": 0.0005, "loss": 2.0618205070495605, "step": 1431 }, { "epoch": 0.1666278799162206, "grad_norm": 1.3557779788970947, "learning_rate": 0.0005, "loss": 3.195913553237915, "step": 1432 }, { "epoch": 0.16674424016755876, "grad_norm": 2.1088457107543945, "learning_rate": 0.0005, "loss": 2.2364089488983154, "step": 1433 }, { "epoch": 0.16686060041889691, "grad_norm": 1.499916672706604, "learning_rate": 0.0005, "loss": 2.4878950119018555, "step": 1434 }, { "epoch": 0.16697696067023504, "grad_norm": 1.4484236240386963, "learning_rate": 0.0005, "loss": 2.502836227416992, "step": 1435 }, { "epoch": 0.1670933209215732, "grad_norm": 2.1612579822540283, "learning_rate": 0.0005, "loss": 3.711602210998535, "step": 1436 }, { "epoch": 0.16720968117291132, "grad_norm": 1.4933667182922363, "learning_rate": 0.0005, "loss": 3.0740509033203125, "step": 1437 }, { "epoch": 0.16732604142424948, "grad_norm": 1.4591602087020874, "learning_rate": 0.0005, "loss": 2.826570510864258, "step": 1438 }, { "epoch": 0.16744240167558763, "grad_norm": 1.3743723630905151, "learning_rate": 0.0005, "loss": 3.204713821411133, "step": 1439 }, { "epoch": 0.16755876192692576, "grad_norm": 1.8053860664367676, "learning_rate": 0.0005, "loss": 2.7400527000427246, "step": 1440 }, { "epoch": 0.1676751221782639, "grad_norm": 2.0670547485351562, "learning_rate": 0.0005, "loss": 3.5610861778259277, "step": 1441 }, { "epoch": 0.16779148242960204, "grad_norm": 2.5712852478027344, "learning_rate": 0.0005, "loss": 3.0195512771606445, "step": 1442 }, { "epoch": 0.1679078426809402, "grad_norm": 1.4336730241775513, "learning_rate": 0.0005, "loss": 2.3379387855529785, "step": 1443 }, { "epoch": 0.16802420293227832, "grad_norm": 1.8327536582946777, "learning_rate": 0.0005, "loss": 3.0091168880462646, "step": 1444 }, { "epoch": 0.16814056318361648, "grad_norm": 1.9712021350860596, "learning_rate": 0.0005, "loss": 3.2275307178497314, "step": 1445 }, { "epoch": 0.16825692343495463, "grad_norm": 1.5096906423568726, "learning_rate": 0.0005, "loss": 2.480403423309326, "step": 1446 }, { "epoch": 0.16837328368629276, "grad_norm": 1.3620197772979736, "learning_rate": 0.0005, "loss": 2.2459280490875244, "step": 1447 }, { "epoch": 0.1684896439376309, "grad_norm": 2.042745590209961, "learning_rate": 0.0005, "loss": 3.555440664291382, "step": 1448 }, { "epoch": 0.16860600418896904, "grad_norm": 1.426956295967102, "learning_rate": 0.0005, "loss": 3.474616765975952, "step": 1449 }, { "epoch": 0.1687223644403072, "grad_norm": 1.9527841806411743, "learning_rate": 0.0005, "loss": 3.9862475395202637, "step": 1450 }, { "epoch": 0.16883872469164535, "grad_norm": 2.593047857284546, "learning_rate": 0.0005, "loss": 4.245871543884277, "step": 1451 }, { "epoch": 0.16895508494298347, "grad_norm": 1.5615897178649902, "learning_rate": 0.0005, "loss": 3.2123823165893555, "step": 1452 }, { "epoch": 0.16907144519432163, "grad_norm": 1.802713394165039, "learning_rate": 0.0005, "loss": 2.578110694885254, "step": 1453 }, { "epoch": 0.16918780544565976, "grad_norm": 1.8803844451904297, "learning_rate": 0.0005, "loss": 2.6028809547424316, "step": 1454 }, { "epoch": 0.1693041656969979, "grad_norm": 1.4082305431365967, "learning_rate": 0.0005, "loss": 2.5046329498291016, "step": 1455 }, { "epoch": 0.16942052594833604, "grad_norm": 1.2124643325805664, "learning_rate": 0.0005, "loss": 2.5002756118774414, "step": 1456 }, { "epoch": 0.1695368861996742, "grad_norm": 2.097496509552002, "learning_rate": 0.0005, "loss": 3.4137606620788574, "step": 1457 }, { "epoch": 0.16965324645101235, "grad_norm": 2.2536113262176514, "learning_rate": 0.0005, "loss": 3.184433937072754, "step": 1458 }, { "epoch": 0.16976960670235047, "grad_norm": 1.452745795249939, "learning_rate": 0.0005, "loss": 2.3949594497680664, "step": 1459 }, { "epoch": 0.16988596695368863, "grad_norm": 2.5337326526641846, "learning_rate": 0.0005, "loss": 3.393552303314209, "step": 1460 }, { "epoch": 0.17000232720502675, "grad_norm": 1.36369788646698, "learning_rate": 0.0005, "loss": 2.6352248191833496, "step": 1461 }, { "epoch": 0.1701186874563649, "grad_norm": 1.9773494005203247, "learning_rate": 0.0005, "loss": 2.3195154666900635, "step": 1462 }, { "epoch": 0.17023504770770304, "grad_norm": 1.601611614227295, "learning_rate": 0.0005, "loss": 2.8296570777893066, "step": 1463 }, { "epoch": 0.1703514079590412, "grad_norm": 1.0864402055740356, "learning_rate": 0.0005, "loss": 3.204437732696533, "step": 1464 }, { "epoch": 0.17046776821037934, "grad_norm": 1.489680528640747, "learning_rate": 0.0005, "loss": 3.1911234855651855, "step": 1465 }, { "epoch": 0.17058412846171747, "grad_norm": 1.6226658821105957, "learning_rate": 0.0005, "loss": 2.879490375518799, "step": 1466 }, { "epoch": 0.17070048871305563, "grad_norm": 0.8138511180877686, "learning_rate": 0.0005, "loss": 1.3340725898742676, "step": 1467 }, { "epoch": 0.17081684896439375, "grad_norm": 1.4335951805114746, "learning_rate": 0.0005, "loss": 2.845364570617676, "step": 1468 }, { "epoch": 0.1709332092157319, "grad_norm": 1.012242317199707, "learning_rate": 0.0005, "loss": 1.9538934230804443, "step": 1469 }, { "epoch": 0.17104956946707006, "grad_norm": 2.439274549484253, "learning_rate": 0.0005, "loss": 3.378216028213501, "step": 1470 }, { "epoch": 0.1711659297184082, "grad_norm": 1.3114206790924072, "learning_rate": 0.0005, "loss": 1.7214866876602173, "step": 1471 }, { "epoch": 0.17128228996974634, "grad_norm": 1.6228433847427368, "learning_rate": 0.0005, "loss": 3.026310443878174, "step": 1472 }, { "epoch": 0.17139865022108447, "grad_norm": 1.684665322303772, "learning_rate": 0.0005, "loss": 3.624849796295166, "step": 1473 }, { "epoch": 0.17151501047242262, "grad_norm": 3.607868194580078, "learning_rate": 0.0005, "loss": 3.0068318843841553, "step": 1474 }, { "epoch": 0.17163137072376075, "grad_norm": 1.8449021577835083, "learning_rate": 0.0005, "loss": 1.9642173051834106, "step": 1475 }, { "epoch": 0.1717477309750989, "grad_norm": 1.5289379358291626, "learning_rate": 0.0005, "loss": 3.214555025100708, "step": 1476 }, { "epoch": 0.17186409122643706, "grad_norm": 1.9012815952301025, "learning_rate": 0.0005, "loss": 2.667783737182617, "step": 1477 }, { "epoch": 0.1719804514777752, "grad_norm": 2.41347599029541, "learning_rate": 0.0005, "loss": 3.583077907562256, "step": 1478 }, { "epoch": 0.17209681172911334, "grad_norm": 4.490588665008545, "learning_rate": 0.0005, "loss": 2.5592968463897705, "step": 1479 }, { "epoch": 0.17221317198045147, "grad_norm": 1.2721092700958252, "learning_rate": 0.0005, "loss": 2.0470123291015625, "step": 1480 }, { "epoch": 0.17232953223178962, "grad_norm": 1.6962995529174805, "learning_rate": 0.0005, "loss": 2.483807325363159, "step": 1481 }, { "epoch": 0.17244589248312778, "grad_norm": 1.4440605640411377, "learning_rate": 0.0005, "loss": 3.469964027404785, "step": 1482 }, { "epoch": 0.1725622527344659, "grad_norm": 1.4539761543273926, "learning_rate": 0.0005, "loss": 3.0663421154022217, "step": 1483 }, { "epoch": 0.17267861298580406, "grad_norm": 1.504542350769043, "learning_rate": 0.0005, "loss": 2.5268242359161377, "step": 1484 }, { "epoch": 0.17279497323714219, "grad_norm": 1.480657935142517, "learning_rate": 0.0005, "loss": 2.985982894897461, "step": 1485 }, { "epoch": 0.17291133348848034, "grad_norm": 1.28091561794281, "learning_rate": 0.0005, "loss": 3.2114152908325195, "step": 1486 }, { "epoch": 0.17302769373981847, "grad_norm": 1.807437539100647, "learning_rate": 0.0005, "loss": 3.452399253845215, "step": 1487 }, { "epoch": 0.17314405399115662, "grad_norm": 3.8111679553985596, "learning_rate": 0.0005, "loss": 3.43290376663208, "step": 1488 }, { "epoch": 0.17326041424249478, "grad_norm": 1.9241238832473755, "learning_rate": 0.0005, "loss": 2.87683367729187, "step": 1489 }, { "epoch": 0.1733767744938329, "grad_norm": 1.6204004287719727, "learning_rate": 0.0005, "loss": 3.4770865440368652, "step": 1490 }, { "epoch": 0.17349313474517106, "grad_norm": 1.3759082555770874, "learning_rate": 0.0005, "loss": 2.4330408573150635, "step": 1491 }, { "epoch": 0.17360949499650918, "grad_norm": 2.072864532470703, "learning_rate": 0.0005, "loss": 2.5648391246795654, "step": 1492 }, { "epoch": 0.17372585524784734, "grad_norm": 1.405783772468567, "learning_rate": 0.0005, "loss": 3.1451573371887207, "step": 1493 }, { "epoch": 0.17384221549918547, "grad_norm": 1.04197096824646, "learning_rate": 0.0005, "loss": 1.645844578742981, "step": 1494 }, { "epoch": 0.17395857575052362, "grad_norm": 2.301907539367676, "learning_rate": 0.0005, "loss": 2.8152284622192383, "step": 1495 }, { "epoch": 0.17407493600186177, "grad_norm": 1.1484410762786865, "learning_rate": 0.0005, "loss": 1.6510181427001953, "step": 1496 }, { "epoch": 0.1741912962531999, "grad_norm": 1.6281095743179321, "learning_rate": 0.0005, "loss": 3.052417039871216, "step": 1497 }, { "epoch": 0.17430765650453806, "grad_norm": 1.6002663373947144, "learning_rate": 0.0005, "loss": 2.6216883659362793, "step": 1498 }, { "epoch": 0.17442401675587618, "grad_norm": 1.287427306175232, "learning_rate": 0.0005, "loss": 1.9535906314849854, "step": 1499 }, { "epoch": 0.17454037700721434, "grad_norm": 1.7763299942016602, "learning_rate": 0.0005, "loss": 3.9799599647521973, "step": 1500 }, { "epoch": 0.1746567372585525, "grad_norm": 1.5630847215652466, "learning_rate": 0.0005, "loss": 3.2351725101470947, "step": 1501 }, { "epoch": 0.17477309750989062, "grad_norm": 1.1004085540771484, "learning_rate": 0.0005, "loss": 1.9574580192565918, "step": 1502 }, { "epoch": 0.17488945776122877, "grad_norm": 0.9255499243736267, "learning_rate": 0.0005, "loss": 2.0636966228485107, "step": 1503 }, { "epoch": 0.1750058180125669, "grad_norm": 1.8931958675384521, "learning_rate": 0.0005, "loss": 3.498870849609375, "step": 1504 }, { "epoch": 0.17512217826390505, "grad_norm": 1.294753909111023, "learning_rate": 0.0005, "loss": 3.0458922386169434, "step": 1505 }, { "epoch": 0.17523853851524318, "grad_norm": 1.380699872970581, "learning_rate": 0.0005, "loss": 2.413391351699829, "step": 1506 }, { "epoch": 0.17535489876658134, "grad_norm": 2.184823989868164, "learning_rate": 0.0005, "loss": 2.8411874771118164, "step": 1507 }, { "epoch": 0.1754712590179195, "grad_norm": 1.8466076850891113, "learning_rate": 0.0005, "loss": 2.975172758102417, "step": 1508 }, { "epoch": 0.17558761926925762, "grad_norm": 2.6130824089050293, "learning_rate": 0.0005, "loss": 3.7624034881591797, "step": 1509 }, { "epoch": 0.17570397952059577, "grad_norm": 1.7579545974731445, "learning_rate": 0.0005, "loss": 2.302267074584961, "step": 1510 }, { "epoch": 0.1758203397719339, "grad_norm": 1.6161837577819824, "learning_rate": 0.0005, "loss": 3.8970565795898438, "step": 1511 }, { "epoch": 0.17593670002327205, "grad_norm": 1.140446662902832, "learning_rate": 0.0005, "loss": 2.3299217224121094, "step": 1512 }, { "epoch": 0.1760530602746102, "grad_norm": 1.2595301866531372, "learning_rate": 0.0005, "loss": 1.931819200515747, "step": 1513 }, { "epoch": 0.17616942052594833, "grad_norm": 1.2201507091522217, "learning_rate": 0.0005, "loss": 2.5715670585632324, "step": 1514 }, { "epoch": 0.1762857807772865, "grad_norm": 1.5325111150741577, "learning_rate": 0.0005, "loss": 2.8312809467315674, "step": 1515 }, { "epoch": 0.17640214102862461, "grad_norm": 2.229252815246582, "learning_rate": 0.0005, "loss": 2.517174243927002, "step": 1516 }, { "epoch": 0.17651850127996277, "grad_norm": 1.0776259899139404, "learning_rate": 0.0005, "loss": 1.749908685684204, "step": 1517 }, { "epoch": 0.1766348615313009, "grad_norm": 1.7815333604812622, "learning_rate": 0.0005, "loss": 3.7552506923675537, "step": 1518 }, { "epoch": 0.17675122178263905, "grad_norm": 1.7792633771896362, "learning_rate": 0.0005, "loss": 3.0207724571228027, "step": 1519 }, { "epoch": 0.1768675820339772, "grad_norm": 2.360957384109497, "learning_rate": 0.0005, "loss": 2.8549838066101074, "step": 1520 }, { "epoch": 0.17698394228531533, "grad_norm": 1.7688671350479126, "learning_rate": 0.0005, "loss": 2.391195774078369, "step": 1521 }, { "epoch": 0.1771003025366535, "grad_norm": 1.2361857891082764, "learning_rate": 0.0005, "loss": 2.9692459106445312, "step": 1522 }, { "epoch": 0.1772166627879916, "grad_norm": 2.311533212661743, "learning_rate": 0.0005, "loss": 3.2226223945617676, "step": 1523 }, { "epoch": 0.17733302303932977, "grad_norm": 1.9745410680770874, "learning_rate": 0.0005, "loss": 3.394294023513794, "step": 1524 }, { "epoch": 0.1774493832906679, "grad_norm": 1.0501673221588135, "learning_rate": 0.0005, "loss": 1.8704475164413452, "step": 1525 }, { "epoch": 0.17756574354200605, "grad_norm": 2.059622049331665, "learning_rate": 0.0005, "loss": 3.6858370304107666, "step": 1526 }, { "epoch": 0.1776821037933442, "grad_norm": 2.0972752571105957, "learning_rate": 0.0005, "loss": 3.9662418365478516, "step": 1527 }, { "epoch": 0.17779846404468233, "grad_norm": 1.504744052886963, "learning_rate": 0.0005, "loss": 4.009502410888672, "step": 1528 }, { "epoch": 0.17791482429602049, "grad_norm": 1.7932515144348145, "learning_rate": 0.0005, "loss": 2.776583433151245, "step": 1529 }, { "epoch": 0.1780311845473586, "grad_norm": 1.9170491695404053, "learning_rate": 0.0005, "loss": 2.851410388946533, "step": 1530 }, { "epoch": 0.17814754479869677, "grad_norm": 2.129634380340576, "learning_rate": 0.0005, "loss": 2.751840829849243, "step": 1531 }, { "epoch": 0.17826390505003492, "grad_norm": 0.9957258105278015, "learning_rate": 0.0005, "loss": 1.871185302734375, "step": 1532 }, { "epoch": 0.17838026530137305, "grad_norm": 1.691717505455017, "learning_rate": 0.0005, "loss": 3.195402145385742, "step": 1533 }, { "epoch": 0.1784966255527112, "grad_norm": 1.6905596256256104, "learning_rate": 0.0005, "loss": 3.0639569759368896, "step": 1534 }, { "epoch": 0.17861298580404933, "grad_norm": 6.4529643058776855, "learning_rate": 0.0005, "loss": 2.8491828441619873, "step": 1535 }, { "epoch": 0.17872934605538748, "grad_norm": 1.725303053855896, "learning_rate": 0.0005, "loss": 3.754985809326172, "step": 1536 }, { "epoch": 0.1788457063067256, "grad_norm": 1.4363908767700195, "learning_rate": 0.0005, "loss": 2.888406276702881, "step": 1537 }, { "epoch": 0.17896206655806376, "grad_norm": 1.5449804067611694, "learning_rate": 0.0005, "loss": 2.039641857147217, "step": 1538 }, { "epoch": 0.17907842680940192, "grad_norm": 1.0573865175247192, "learning_rate": 0.0005, "loss": 2.7607665061950684, "step": 1539 }, { "epoch": 0.17919478706074005, "grad_norm": 1.6121631860733032, "learning_rate": 0.0005, "loss": 3.9521210193634033, "step": 1540 }, { "epoch": 0.1793111473120782, "grad_norm": 0.9397016763687134, "learning_rate": 0.0005, "loss": 2.311253547668457, "step": 1541 }, { "epoch": 0.17942750756341633, "grad_norm": 1.3720083236694336, "learning_rate": 0.0005, "loss": 2.2401561737060547, "step": 1542 }, { "epoch": 0.17954386781475448, "grad_norm": 1.4266612529754639, "learning_rate": 0.0005, "loss": 3.4270622730255127, "step": 1543 }, { "epoch": 0.17966022806609264, "grad_norm": 1.1338788270950317, "learning_rate": 0.0005, "loss": 2.6742701530456543, "step": 1544 }, { "epoch": 0.17977658831743076, "grad_norm": 1.4141156673431396, "learning_rate": 0.0005, "loss": 2.9147894382476807, "step": 1545 }, { "epoch": 0.17989294856876892, "grad_norm": 4.176743030548096, "learning_rate": 0.0005, "loss": 4.455652236938477, "step": 1546 }, { "epoch": 0.18000930882010704, "grad_norm": 1.4285649061203003, "learning_rate": 0.0005, "loss": 3.146941661834717, "step": 1547 }, { "epoch": 0.1801256690714452, "grad_norm": 1.610248327255249, "learning_rate": 0.0005, "loss": 2.5839452743530273, "step": 1548 }, { "epoch": 0.18024202932278333, "grad_norm": 1.7294764518737793, "learning_rate": 0.0005, "loss": 1.934983730316162, "step": 1549 }, { "epoch": 0.18035838957412148, "grad_norm": 1.4228395223617554, "learning_rate": 0.0005, "loss": 2.3472743034362793, "step": 1550 }, { "epoch": 0.18047474982545963, "grad_norm": 1.5658857822418213, "learning_rate": 0.0005, "loss": 2.821896553039551, "step": 1551 }, { "epoch": 0.18059111007679776, "grad_norm": 1.0804157257080078, "learning_rate": 0.0005, "loss": 2.9113574028015137, "step": 1552 }, { "epoch": 0.18070747032813592, "grad_norm": 2.0070888996124268, "learning_rate": 0.0005, "loss": 2.7742555141448975, "step": 1553 }, { "epoch": 0.18082383057947404, "grad_norm": 1.321006178855896, "learning_rate": 0.0005, "loss": 2.1081156730651855, "step": 1554 }, { "epoch": 0.1809401908308122, "grad_norm": 2.281219244003296, "learning_rate": 0.0005, "loss": 2.9678664207458496, "step": 1555 }, { "epoch": 0.18105655108215032, "grad_norm": 1.914966106414795, "learning_rate": 0.0005, "loss": 2.4390885829925537, "step": 1556 }, { "epoch": 0.18117291133348848, "grad_norm": 1.1239347457885742, "learning_rate": 0.0005, "loss": 3.0517401695251465, "step": 1557 }, { "epoch": 0.18128927158482663, "grad_norm": 0.902466893196106, "learning_rate": 0.0005, "loss": 1.391784906387329, "step": 1558 }, { "epoch": 0.18140563183616476, "grad_norm": 1.2388030290603638, "learning_rate": 0.0005, "loss": 2.327179431915283, "step": 1559 }, { "epoch": 0.18152199208750291, "grad_norm": 2.9799728393554688, "learning_rate": 0.0005, "loss": 3.9864511489868164, "step": 1560 }, { "epoch": 0.18163835233884104, "grad_norm": 1.3410004377365112, "learning_rate": 0.0005, "loss": 2.661050796508789, "step": 1561 }, { "epoch": 0.1817547125901792, "grad_norm": 1.2754926681518555, "learning_rate": 0.0005, "loss": 2.3581206798553467, "step": 1562 }, { "epoch": 0.18187107284151735, "grad_norm": 1.5720113515853882, "learning_rate": 0.0005, "loss": 3.917184591293335, "step": 1563 }, { "epoch": 0.18198743309285548, "grad_norm": 1.212867021560669, "learning_rate": 0.0005, "loss": 2.3370022773742676, "step": 1564 }, { "epoch": 0.18210379334419363, "grad_norm": 1.5742199420928955, "learning_rate": 0.0005, "loss": 2.817272663116455, "step": 1565 }, { "epoch": 0.18222015359553176, "grad_norm": 1.7014492750167847, "learning_rate": 0.0005, "loss": 3.5983290672302246, "step": 1566 }, { "epoch": 0.1823365138468699, "grad_norm": 2.048076629638672, "learning_rate": 0.0005, "loss": 2.614335060119629, "step": 1567 }, { "epoch": 0.18245287409820804, "grad_norm": 1.828538417816162, "learning_rate": 0.0005, "loss": 2.6226730346679688, "step": 1568 }, { "epoch": 0.1825692343495462, "grad_norm": 1.212674617767334, "learning_rate": 0.0005, "loss": 2.782590627670288, "step": 1569 }, { "epoch": 0.18268559460088435, "grad_norm": 1.9470751285552979, "learning_rate": 0.0005, "loss": 3.1103107929229736, "step": 1570 }, { "epoch": 0.18280195485222248, "grad_norm": 1.987728476524353, "learning_rate": 0.0005, "loss": 2.9427547454833984, "step": 1571 }, { "epoch": 0.18291831510356063, "grad_norm": 1.2937275171279907, "learning_rate": 0.0005, "loss": 2.7122457027435303, "step": 1572 }, { "epoch": 0.18303467535489876, "grad_norm": 1.8273974657058716, "learning_rate": 0.0005, "loss": 2.364450693130493, "step": 1573 }, { "epoch": 0.1831510356062369, "grad_norm": 1.4408247470855713, "learning_rate": 0.0005, "loss": 3.362272024154663, "step": 1574 }, { "epoch": 0.18326739585757507, "grad_norm": 1.1846312284469604, "learning_rate": 0.0005, "loss": 3.0362539291381836, "step": 1575 }, { "epoch": 0.1833837561089132, "grad_norm": 2.1256279945373535, "learning_rate": 0.0005, "loss": 3.2828545570373535, "step": 1576 }, { "epoch": 0.18350011636025135, "grad_norm": 1.5360885858535767, "learning_rate": 0.0005, "loss": 2.830533504486084, "step": 1577 }, { "epoch": 0.18361647661158947, "grad_norm": 2.103891372680664, "learning_rate": 0.0005, "loss": 2.9329097270965576, "step": 1578 }, { "epoch": 0.18373283686292763, "grad_norm": 1.665924310684204, "learning_rate": 0.0005, "loss": 3.146228790283203, "step": 1579 }, { "epoch": 0.18384919711426576, "grad_norm": 1.2857015132904053, "learning_rate": 0.0005, "loss": 3.5446956157684326, "step": 1580 }, { "epoch": 0.1839655573656039, "grad_norm": 1.5124201774597168, "learning_rate": 0.0005, "loss": 2.599911689758301, "step": 1581 }, { "epoch": 0.18408191761694206, "grad_norm": 1.6088682413101196, "learning_rate": 0.0005, "loss": 2.947530746459961, "step": 1582 }, { "epoch": 0.1841982778682802, "grad_norm": 1.0488728284835815, "learning_rate": 0.0005, "loss": 1.6863478422164917, "step": 1583 }, { "epoch": 0.18431463811961835, "grad_norm": 2.4778740406036377, "learning_rate": 0.0005, "loss": 2.9477357864379883, "step": 1584 }, { "epoch": 0.18443099837095647, "grad_norm": 1.3350037336349487, "learning_rate": 0.0005, "loss": 2.163242816925049, "step": 1585 }, { "epoch": 0.18454735862229463, "grad_norm": 1.8083953857421875, "learning_rate": 0.0005, "loss": 2.3700077533721924, "step": 1586 }, { "epoch": 0.18466371887363275, "grad_norm": 1.5869735479354858, "learning_rate": 0.0005, "loss": 2.5002264976501465, "step": 1587 }, { "epoch": 0.1847800791249709, "grad_norm": 1.386444091796875, "learning_rate": 0.0005, "loss": 1.8626261949539185, "step": 1588 }, { "epoch": 0.18489643937630906, "grad_norm": 1.267214298248291, "learning_rate": 0.0005, "loss": 2.703801393508911, "step": 1589 }, { "epoch": 0.1850127996276472, "grad_norm": 1.489113450050354, "learning_rate": 0.0005, "loss": 2.5209946632385254, "step": 1590 }, { "epoch": 0.18512915987898534, "grad_norm": 1.3151209354400635, "learning_rate": 0.0005, "loss": 2.5216379165649414, "step": 1591 }, { "epoch": 0.18524552013032347, "grad_norm": 1.0069496631622314, "learning_rate": 0.0005, "loss": 2.403298854827881, "step": 1592 }, { "epoch": 0.18536188038166163, "grad_norm": 2.170194625854492, "learning_rate": 0.0005, "loss": 2.600533962249756, "step": 1593 }, { "epoch": 0.18547824063299978, "grad_norm": 2.004777193069458, "learning_rate": 0.0005, "loss": 3.2060844898223877, "step": 1594 }, { "epoch": 0.1855946008843379, "grad_norm": 1.7647672891616821, "learning_rate": 0.0005, "loss": 3.738450765609741, "step": 1595 }, { "epoch": 0.18571096113567606, "grad_norm": 1.5606639385223389, "learning_rate": 0.0005, "loss": 3.164621114730835, "step": 1596 }, { "epoch": 0.1858273213870142, "grad_norm": 1.9015321731567383, "learning_rate": 0.0005, "loss": 3.142904281616211, "step": 1597 }, { "epoch": 0.18594368163835234, "grad_norm": 1.9272286891937256, "learning_rate": 0.0005, "loss": 3.238274574279785, "step": 1598 }, { "epoch": 0.18606004188969047, "grad_norm": 1.2955111265182495, "learning_rate": 0.0005, "loss": 3.4182791709899902, "step": 1599 }, { "epoch": 0.18617640214102862, "grad_norm": 2.2565884590148926, "learning_rate": 0.0005, "loss": 3.5823922157287598, "step": 1600 }, { "epoch": 0.18629276239236678, "grad_norm": 1.787440299987793, "learning_rate": 0.0005, "loss": 2.8005871772766113, "step": 1601 }, { "epoch": 0.1864091226437049, "grad_norm": 1.1005488634109497, "learning_rate": 0.0005, "loss": 3.246279239654541, "step": 1602 }, { "epoch": 0.18652548289504306, "grad_norm": 1.4864059686660767, "learning_rate": 0.0005, "loss": 2.224607229232788, "step": 1603 }, { "epoch": 0.1866418431463812, "grad_norm": 1.411382794380188, "learning_rate": 0.0005, "loss": 2.2138803005218506, "step": 1604 }, { "epoch": 0.18675820339771934, "grad_norm": 1.475378155708313, "learning_rate": 0.0005, "loss": 3.399970531463623, "step": 1605 }, { "epoch": 0.1868745636490575, "grad_norm": 1.2786372900009155, "learning_rate": 0.0005, "loss": 2.32413387298584, "step": 1606 }, { "epoch": 0.18699092390039562, "grad_norm": 1.5238559246063232, "learning_rate": 0.0005, "loss": 2.835562229156494, "step": 1607 }, { "epoch": 0.18710728415173378, "grad_norm": 1.9395548105239868, "learning_rate": 0.0005, "loss": 3.4226760864257812, "step": 1608 }, { "epoch": 0.1872236444030719, "grad_norm": 1.3805917501449585, "learning_rate": 0.0005, "loss": 2.6330857276916504, "step": 1609 }, { "epoch": 0.18734000465441006, "grad_norm": 1.9825735092163086, "learning_rate": 0.0005, "loss": 4.06669282913208, "step": 1610 }, { "epoch": 0.18745636490574819, "grad_norm": 1.193631649017334, "learning_rate": 0.0005, "loss": 2.330601692199707, "step": 1611 }, { "epoch": 0.18757272515708634, "grad_norm": 1.6639126539230347, "learning_rate": 0.0005, "loss": 2.7565431594848633, "step": 1612 }, { "epoch": 0.1876890854084245, "grad_norm": 1.6904487609863281, "learning_rate": 0.0005, "loss": 2.654419422149658, "step": 1613 }, { "epoch": 0.18780544565976262, "grad_norm": 1.564922571182251, "learning_rate": 0.0005, "loss": 3.279993772506714, "step": 1614 }, { "epoch": 0.18792180591110078, "grad_norm": 1.5226348638534546, "learning_rate": 0.0005, "loss": 2.415048122406006, "step": 1615 }, { "epoch": 0.1880381661624389, "grad_norm": 2.9624593257904053, "learning_rate": 0.0005, "loss": 3.8881092071533203, "step": 1616 }, { "epoch": 0.18815452641377706, "grad_norm": 1.3856958150863647, "learning_rate": 0.0005, "loss": 3.531363010406494, "step": 1617 }, { "epoch": 0.18827088666511518, "grad_norm": 1.53312087059021, "learning_rate": 0.0005, "loss": 4.26216459274292, "step": 1618 }, { "epoch": 0.18838724691645334, "grad_norm": 2.2466373443603516, "learning_rate": 0.0005, "loss": 3.853468179702759, "step": 1619 }, { "epoch": 0.1885036071677915, "grad_norm": 1.4851162433624268, "learning_rate": 0.0005, "loss": 2.9782791137695312, "step": 1620 }, { "epoch": 0.18861996741912962, "grad_norm": 1.1105725765228271, "learning_rate": 0.0005, "loss": 2.3737614154815674, "step": 1621 }, { "epoch": 0.18873632767046777, "grad_norm": 1.4159029722213745, "learning_rate": 0.0005, "loss": 3.2892651557922363, "step": 1622 }, { "epoch": 0.1888526879218059, "grad_norm": 1.677006721496582, "learning_rate": 0.0005, "loss": 2.8898425102233887, "step": 1623 }, { "epoch": 0.18896904817314406, "grad_norm": 1.8943504095077515, "learning_rate": 0.0005, "loss": 3.1267356872558594, "step": 1624 }, { "epoch": 0.1890854084244822, "grad_norm": 1.1940680742263794, "learning_rate": 0.0005, "loss": 2.304142475128174, "step": 1625 }, { "epoch": 0.18920176867582034, "grad_norm": 2.3673760890960693, "learning_rate": 0.0005, "loss": 3.0375711917877197, "step": 1626 }, { "epoch": 0.1893181289271585, "grad_norm": 1.9268386363983154, "learning_rate": 0.0005, "loss": 3.238056182861328, "step": 1627 }, { "epoch": 0.18943448917849662, "grad_norm": 0.63846355676651, "learning_rate": 0.0005, "loss": 1.135921835899353, "step": 1628 }, { "epoch": 0.18955084942983477, "grad_norm": 1.1679712533950806, "learning_rate": 0.0005, "loss": 2.565617561340332, "step": 1629 }, { "epoch": 0.1896672096811729, "grad_norm": 1.2331748008728027, "learning_rate": 0.0005, "loss": 2.268787145614624, "step": 1630 }, { "epoch": 0.18978356993251105, "grad_norm": 1.4928064346313477, "learning_rate": 0.0005, "loss": 2.8620481491088867, "step": 1631 }, { "epoch": 0.1898999301838492, "grad_norm": 2.5678892135620117, "learning_rate": 0.0005, "loss": 3.0931320190429688, "step": 1632 }, { "epoch": 0.19001629043518733, "grad_norm": 1.299848198890686, "learning_rate": 0.0005, "loss": 2.3552744388580322, "step": 1633 }, { "epoch": 0.1901326506865255, "grad_norm": 1.9726548194885254, "learning_rate": 0.0005, "loss": 2.900641441345215, "step": 1634 }, { "epoch": 0.19024901093786362, "grad_norm": 1.830936074256897, "learning_rate": 0.0005, "loss": 3.151479959487915, "step": 1635 }, { "epoch": 0.19036537118920177, "grad_norm": 1.4455053806304932, "learning_rate": 0.0005, "loss": 2.891433000564575, "step": 1636 }, { "epoch": 0.19048173144053993, "grad_norm": 1.0830836296081543, "learning_rate": 0.0005, "loss": 1.9576151371002197, "step": 1637 }, { "epoch": 0.19059809169187805, "grad_norm": 1.3093061447143555, "learning_rate": 0.0005, "loss": 2.9249062538146973, "step": 1638 }, { "epoch": 0.1907144519432162, "grad_norm": 1.246644139289856, "learning_rate": 0.0005, "loss": 2.8987839221954346, "step": 1639 }, { "epoch": 0.19083081219455433, "grad_norm": 1.6269499063491821, "learning_rate": 0.0005, "loss": 3.5398130416870117, "step": 1640 }, { "epoch": 0.1909471724458925, "grad_norm": 1.3785462379455566, "learning_rate": 0.0005, "loss": 2.175717830657959, "step": 1641 }, { "epoch": 0.19106353269723061, "grad_norm": 1.3575260639190674, "learning_rate": 0.0005, "loss": 3.117095947265625, "step": 1642 }, { "epoch": 0.19117989294856877, "grad_norm": 1.7634257078170776, "learning_rate": 0.0005, "loss": 3.169039249420166, "step": 1643 }, { "epoch": 0.19129625319990692, "grad_norm": 1.2840557098388672, "learning_rate": 0.0005, "loss": 2.790682554244995, "step": 1644 }, { "epoch": 0.19141261345124505, "grad_norm": 1.181034803390503, "learning_rate": 0.0005, "loss": 2.189145565032959, "step": 1645 }, { "epoch": 0.1915289737025832, "grad_norm": 1.2201951742172241, "learning_rate": 0.0005, "loss": 3.2793445587158203, "step": 1646 }, { "epoch": 0.19164533395392133, "grad_norm": 1.2325091361999512, "learning_rate": 0.0005, "loss": 2.519509792327881, "step": 1647 }, { "epoch": 0.1917616942052595, "grad_norm": 0.9576650857925415, "learning_rate": 0.0005, "loss": 2.132822036743164, "step": 1648 }, { "epoch": 0.1918780544565976, "grad_norm": 2.1394803524017334, "learning_rate": 0.0005, "loss": 3.27437424659729, "step": 1649 }, { "epoch": 0.19199441470793577, "grad_norm": 1.1753777265548706, "learning_rate": 0.0005, "loss": 2.220182418823242, "step": 1650 }, { "epoch": 0.19211077495927392, "grad_norm": 2.7055819034576416, "learning_rate": 0.0005, "loss": 3.2750539779663086, "step": 1651 }, { "epoch": 0.19222713521061205, "grad_norm": 1.3921533823013306, "learning_rate": 0.0005, "loss": 2.625317096710205, "step": 1652 }, { "epoch": 0.1923434954619502, "grad_norm": 1.4192805290222168, "learning_rate": 0.0005, "loss": 2.3302903175354004, "step": 1653 }, { "epoch": 0.19245985571328833, "grad_norm": 2.3318936824798584, "learning_rate": 0.0005, "loss": 3.2085652351379395, "step": 1654 }, { "epoch": 0.19257621596462648, "grad_norm": 1.3645410537719727, "learning_rate": 0.0005, "loss": 3.0575456619262695, "step": 1655 }, { "epoch": 0.19269257621596464, "grad_norm": 1.4087344408035278, "learning_rate": 0.0005, "loss": 2.64758038520813, "step": 1656 }, { "epoch": 0.19280893646730277, "grad_norm": 0.9424048662185669, "learning_rate": 0.0005, "loss": 2.4090018272399902, "step": 1657 }, { "epoch": 0.19292529671864092, "grad_norm": 1.502576470375061, "learning_rate": 0.0005, "loss": 3.785881757736206, "step": 1658 }, { "epoch": 0.19304165696997905, "grad_norm": 1.2246469259262085, "learning_rate": 0.0005, "loss": 2.144075393676758, "step": 1659 }, { "epoch": 0.1931580172213172, "grad_norm": 1.8561298847198486, "learning_rate": 0.0005, "loss": 3.671271800994873, "step": 1660 }, { "epoch": 0.19327437747265533, "grad_norm": 1.1900311708450317, "learning_rate": 0.0005, "loss": 2.0986521244049072, "step": 1661 }, { "epoch": 0.19339073772399348, "grad_norm": 3.6832971572875977, "learning_rate": 0.0005, "loss": 3.9330391883850098, "step": 1662 }, { "epoch": 0.19350709797533164, "grad_norm": 1.7420308589935303, "learning_rate": 0.0005, "loss": 3.138671636581421, "step": 1663 }, { "epoch": 0.19362345822666976, "grad_norm": 1.5761183500289917, "learning_rate": 0.0005, "loss": 3.275052070617676, "step": 1664 }, { "epoch": 0.19373981847800792, "grad_norm": 1.6678380966186523, "learning_rate": 0.0005, "loss": 2.739856719970703, "step": 1665 }, { "epoch": 0.19385617872934605, "grad_norm": 1.534190058708191, "learning_rate": 0.0005, "loss": 2.966789722442627, "step": 1666 }, { "epoch": 0.1939725389806842, "grad_norm": 1.9310766458511353, "learning_rate": 0.0005, "loss": 2.616766929626465, "step": 1667 }, { "epoch": 0.19408889923202235, "grad_norm": 1.5271670818328857, "learning_rate": 0.0005, "loss": 2.2870845794677734, "step": 1668 }, { "epoch": 0.19420525948336048, "grad_norm": 4.598545551300049, "learning_rate": 0.0005, "loss": 3.8539676666259766, "step": 1669 }, { "epoch": 0.19432161973469864, "grad_norm": 2.0305068492889404, "learning_rate": 0.0005, "loss": 3.068175792694092, "step": 1670 }, { "epoch": 0.19443797998603676, "grad_norm": 1.0424830913543701, "learning_rate": 0.0005, "loss": 2.046687364578247, "step": 1671 }, { "epoch": 0.19455434023737492, "grad_norm": 2.0043652057647705, "learning_rate": 0.0005, "loss": 3.33341646194458, "step": 1672 }, { "epoch": 0.19467070048871304, "grad_norm": 6.135672092437744, "learning_rate": 0.0005, "loss": 2.215456008911133, "step": 1673 }, { "epoch": 0.1947870607400512, "grad_norm": 2.1624085903167725, "learning_rate": 0.0005, "loss": 3.2841639518737793, "step": 1674 }, { "epoch": 0.19490342099138935, "grad_norm": 1.6624716520309448, "learning_rate": 0.0005, "loss": 2.605182409286499, "step": 1675 }, { "epoch": 0.19501978124272748, "grad_norm": 1.53553307056427, "learning_rate": 0.0005, "loss": 3.826169490814209, "step": 1676 }, { "epoch": 0.19513614149406563, "grad_norm": 1.6859859228134155, "learning_rate": 0.0005, "loss": 2.9106874465942383, "step": 1677 }, { "epoch": 0.19525250174540376, "grad_norm": 1.8571475744247437, "learning_rate": 0.0005, "loss": 2.315690755844116, "step": 1678 }, { "epoch": 0.19536886199674192, "grad_norm": 1.3419268131256104, "learning_rate": 0.0005, "loss": 2.7121527194976807, "step": 1679 }, { "epoch": 0.19548522224808004, "grad_norm": 2.2769737243652344, "learning_rate": 0.0005, "loss": 3.363197088241577, "step": 1680 }, { "epoch": 0.1956015824994182, "grad_norm": 1.8802883625030518, "learning_rate": 0.0005, "loss": 4.086695194244385, "step": 1681 }, { "epoch": 0.19571794275075635, "grad_norm": 1.9346365928649902, "learning_rate": 0.0005, "loss": 2.6896326541900635, "step": 1682 }, { "epoch": 0.19583430300209448, "grad_norm": 1.2865386009216309, "learning_rate": 0.0005, "loss": 2.5458946228027344, "step": 1683 }, { "epoch": 0.19595066325343263, "grad_norm": 2.5728468894958496, "learning_rate": 0.0005, "loss": 2.1754181385040283, "step": 1684 }, { "epoch": 0.19606702350477076, "grad_norm": 1.2677364349365234, "learning_rate": 0.0005, "loss": 2.645263671875, "step": 1685 }, { "epoch": 0.19618338375610891, "grad_norm": 0.9646199941635132, "learning_rate": 0.0005, "loss": 1.7773340940475464, "step": 1686 }, { "epoch": 0.19629974400744707, "grad_norm": 1.594507098197937, "learning_rate": 0.0005, "loss": 2.837803602218628, "step": 1687 }, { "epoch": 0.1964161042587852, "grad_norm": 1.1746665239334106, "learning_rate": 0.0005, "loss": 2.2761380672454834, "step": 1688 }, { "epoch": 0.19653246451012335, "grad_norm": 1.7234675884246826, "learning_rate": 0.0005, "loss": 2.5248215198516846, "step": 1689 }, { "epoch": 0.19664882476146148, "grad_norm": 1.5194926261901855, "learning_rate": 0.0005, "loss": 2.6174774169921875, "step": 1690 }, { "epoch": 0.19676518501279963, "grad_norm": 2.248971939086914, "learning_rate": 0.0005, "loss": 2.8099241256713867, "step": 1691 }, { "epoch": 0.19688154526413776, "grad_norm": 1.25732421875, "learning_rate": 0.0005, "loss": 2.7136754989624023, "step": 1692 }, { "epoch": 0.1969979055154759, "grad_norm": 1.5237469673156738, "learning_rate": 0.0005, "loss": 2.0478622913360596, "step": 1693 }, { "epoch": 0.19711426576681407, "grad_norm": 1.6109638214111328, "learning_rate": 0.0005, "loss": 3.3402037620544434, "step": 1694 }, { "epoch": 0.1972306260181522, "grad_norm": 1.8891627788543701, "learning_rate": 0.0005, "loss": 2.2865688800811768, "step": 1695 }, { "epoch": 0.19734698626949035, "grad_norm": 1.3814154863357544, "learning_rate": 0.0005, "loss": 3.286846160888672, "step": 1696 }, { "epoch": 0.19746334652082848, "grad_norm": 1.8648093938827515, "learning_rate": 0.0005, "loss": 3.104793071746826, "step": 1697 }, { "epoch": 0.19757970677216663, "grad_norm": 1.5154322385787964, "learning_rate": 0.0005, "loss": 3.0192155838012695, "step": 1698 }, { "epoch": 0.19769606702350478, "grad_norm": 2.183286666870117, "learning_rate": 0.0005, "loss": 2.6160054206848145, "step": 1699 }, { "epoch": 0.1978124272748429, "grad_norm": 1.0699459314346313, "learning_rate": 0.0005, "loss": 2.9315268993377686, "step": 1700 }, { "epoch": 0.19792878752618107, "grad_norm": 2.2494611740112305, "learning_rate": 0.0005, "loss": 2.396627902984619, "step": 1701 }, { "epoch": 0.1980451477775192, "grad_norm": 1.2647851705551147, "learning_rate": 0.0005, "loss": 2.7122316360473633, "step": 1702 }, { "epoch": 0.19816150802885735, "grad_norm": 1.4541184902191162, "learning_rate": 0.0005, "loss": 1.7797297239303589, "step": 1703 }, { "epoch": 0.19827786828019547, "grad_norm": 1.4424564838409424, "learning_rate": 0.0005, "loss": 2.639265537261963, "step": 1704 }, { "epoch": 0.19839422853153363, "grad_norm": 1.7941181659698486, "learning_rate": 0.0005, "loss": 2.9089670181274414, "step": 1705 }, { "epoch": 0.19851058878287178, "grad_norm": 2.137197732925415, "learning_rate": 0.0005, "loss": 3.209092617034912, "step": 1706 }, { "epoch": 0.1986269490342099, "grad_norm": 2.5306460857391357, "learning_rate": 0.0005, "loss": 2.9348320960998535, "step": 1707 }, { "epoch": 0.19874330928554806, "grad_norm": 1.4414724111557007, "learning_rate": 0.0005, "loss": 3.1772758960723877, "step": 1708 }, { "epoch": 0.1988596695368862, "grad_norm": 1.4296311140060425, "learning_rate": 0.0005, "loss": 2.464325428009033, "step": 1709 }, { "epoch": 0.19897602978822435, "grad_norm": 1.6348072290420532, "learning_rate": 0.0005, "loss": 2.3475492000579834, "step": 1710 }, { "epoch": 0.19909239003956247, "grad_norm": 1.3367416858673096, "learning_rate": 0.0005, "loss": 1.9706370830535889, "step": 1711 }, { "epoch": 0.19920875029090063, "grad_norm": 1.8869922161102295, "learning_rate": 0.0005, "loss": 3.059877872467041, "step": 1712 }, { "epoch": 0.19932511054223878, "grad_norm": 1.07173752784729, "learning_rate": 0.0005, "loss": 3.4000258445739746, "step": 1713 }, { "epoch": 0.1994414707935769, "grad_norm": 1.7631334066390991, "learning_rate": 0.0005, "loss": 3.665440797805786, "step": 1714 }, { "epoch": 0.19955783104491506, "grad_norm": 1.5389635562896729, "learning_rate": 0.0005, "loss": 2.6482222080230713, "step": 1715 }, { "epoch": 0.1996741912962532, "grad_norm": 1.8891563415527344, "learning_rate": 0.0005, "loss": 3.670891284942627, "step": 1716 }, { "epoch": 0.19979055154759134, "grad_norm": 1.380836009979248, "learning_rate": 0.0005, "loss": 3.1061854362487793, "step": 1717 }, { "epoch": 0.1999069117989295, "grad_norm": 1.7142726182937622, "learning_rate": 0.0005, "loss": 2.3804383277893066, "step": 1718 }, { "epoch": 0.20002327205026763, "grad_norm": 1.1218377351760864, "learning_rate": 0.0005, "loss": 1.8241729736328125, "step": 1719 }, { "epoch": 0.20013963230160578, "grad_norm": 1.5476360321044922, "learning_rate": 0.0005, "loss": 2.7537217140197754, "step": 1720 }, { "epoch": 0.2002559925529439, "grad_norm": 1.8176738023757935, "learning_rate": 0.0005, "loss": 3.2070324420928955, "step": 1721 }, { "epoch": 0.20037235280428206, "grad_norm": 1.6759697198867798, "learning_rate": 0.0005, "loss": 3.0808610916137695, "step": 1722 }, { "epoch": 0.2004887130556202, "grad_norm": 1.4842325448989868, "learning_rate": 0.0005, "loss": 2.5921506881713867, "step": 1723 }, { "epoch": 0.20060507330695834, "grad_norm": 1.1884864568710327, "learning_rate": 0.0005, "loss": 3.067511558532715, "step": 1724 }, { "epoch": 0.2007214335582965, "grad_norm": 1.7752630710601807, "learning_rate": 0.0005, "loss": 2.713313341140747, "step": 1725 }, { "epoch": 0.20083779380963462, "grad_norm": 1.5853452682495117, "learning_rate": 0.0005, "loss": 3.284952163696289, "step": 1726 }, { "epoch": 0.20095415406097278, "grad_norm": 1.3767107725143433, "learning_rate": 0.0005, "loss": 2.9645471572875977, "step": 1727 }, { "epoch": 0.2010705143123109, "grad_norm": 1.8448277711868286, "learning_rate": 0.0005, "loss": 2.6187081336975098, "step": 1728 }, { "epoch": 0.20118687456364906, "grad_norm": 14.003458976745605, "learning_rate": 0.0005, "loss": 2.861008644104004, "step": 1729 }, { "epoch": 0.20130323481498721, "grad_norm": 2.0199601650238037, "learning_rate": 0.0005, "loss": 2.5558547973632812, "step": 1730 }, { "epoch": 0.20141959506632534, "grad_norm": 1.5531307458877563, "learning_rate": 0.0005, "loss": 2.523899793624878, "step": 1731 }, { "epoch": 0.2015359553176635, "grad_norm": 1.5873080492019653, "learning_rate": 0.0005, "loss": 2.8417491912841797, "step": 1732 }, { "epoch": 0.20165231556900162, "grad_norm": 1.8507355451583862, "learning_rate": 0.0005, "loss": 3.214693546295166, "step": 1733 }, { "epoch": 0.20176867582033978, "grad_norm": 1.7685776948928833, "learning_rate": 0.0005, "loss": 2.746452808380127, "step": 1734 }, { "epoch": 0.2018850360716779, "grad_norm": 2.2717466354370117, "learning_rate": 0.0005, "loss": 3.9883785247802734, "step": 1735 }, { "epoch": 0.20200139632301606, "grad_norm": 1.3811739683151245, "learning_rate": 0.0005, "loss": 2.5096230506896973, "step": 1736 }, { "epoch": 0.2021177565743542, "grad_norm": 1.7659660577774048, "learning_rate": 0.0005, "loss": 2.4564552307128906, "step": 1737 }, { "epoch": 0.20223411682569234, "grad_norm": 1.190426230430603, "learning_rate": 0.0005, "loss": 2.266406774520874, "step": 1738 }, { "epoch": 0.2023504770770305, "grad_norm": 0.9761633276939392, "learning_rate": 0.0005, "loss": 1.9096988439559937, "step": 1739 }, { "epoch": 0.20246683732836862, "grad_norm": 1.1674213409423828, "learning_rate": 0.0005, "loss": 2.1912941932678223, "step": 1740 }, { "epoch": 0.20258319757970678, "grad_norm": 1.5798462629318237, "learning_rate": 0.0005, "loss": 3.544511318206787, "step": 1741 }, { "epoch": 0.2026995578310449, "grad_norm": 2.1035497188568115, "learning_rate": 0.0005, "loss": 3.1832809448242188, "step": 1742 }, { "epoch": 0.20281591808238306, "grad_norm": 1.4307563304901123, "learning_rate": 0.0005, "loss": 2.48376727104187, "step": 1743 }, { "epoch": 0.2029322783337212, "grad_norm": 1.1687531471252441, "learning_rate": 0.0005, "loss": 2.3344180583953857, "step": 1744 }, { "epoch": 0.20304863858505934, "grad_norm": 3.579416513442993, "learning_rate": 0.0005, "loss": 4.552857398986816, "step": 1745 }, { "epoch": 0.2031649988363975, "grad_norm": 1.7406665086746216, "learning_rate": 0.0005, "loss": 3.0044548511505127, "step": 1746 }, { "epoch": 0.20328135908773562, "grad_norm": 2.017698049545288, "learning_rate": 0.0005, "loss": 2.720414161682129, "step": 1747 }, { "epoch": 0.20339771933907377, "grad_norm": 1.8057676553726196, "learning_rate": 0.0005, "loss": 2.9958419799804688, "step": 1748 }, { "epoch": 0.20351407959041193, "grad_norm": 1.8474323749542236, "learning_rate": 0.0005, "loss": 3.167360782623291, "step": 1749 }, { "epoch": 0.20363043984175005, "grad_norm": 3.3042283058166504, "learning_rate": 0.0005, "loss": 3.65144419670105, "step": 1750 }, { "epoch": 0.2037468000930882, "grad_norm": 2.054702043533325, "learning_rate": 0.0005, "loss": 3.6581029891967773, "step": 1751 }, { "epoch": 0.20386316034442634, "grad_norm": 1.1436667442321777, "learning_rate": 0.0005, "loss": 2.7128899097442627, "step": 1752 }, { "epoch": 0.2039795205957645, "grad_norm": 1.0829439163208008, "learning_rate": 0.0005, "loss": 2.3562498092651367, "step": 1753 }, { "epoch": 0.20409588084710262, "grad_norm": 1.6966544389724731, "learning_rate": 0.0005, "loss": 3.974552631378174, "step": 1754 }, { "epoch": 0.20421224109844077, "grad_norm": 1.2024133205413818, "learning_rate": 0.0005, "loss": 2.4700279235839844, "step": 1755 }, { "epoch": 0.20432860134977893, "grad_norm": 1.1451996564865112, "learning_rate": 0.0005, "loss": 1.738276720046997, "step": 1756 }, { "epoch": 0.20444496160111705, "grad_norm": 5.4202423095703125, "learning_rate": 0.0005, "loss": 3.744413375854492, "step": 1757 }, { "epoch": 0.2045613218524552, "grad_norm": 1.3992565870285034, "learning_rate": 0.0005, "loss": 3.788541316986084, "step": 1758 }, { "epoch": 0.20467768210379333, "grad_norm": 1.542879343032837, "learning_rate": 0.0005, "loss": 2.668656826019287, "step": 1759 }, { "epoch": 0.2047940423551315, "grad_norm": 1.1784700155258179, "learning_rate": 0.0005, "loss": 2.0196468830108643, "step": 1760 }, { "epoch": 0.20491040260646964, "grad_norm": 1.7103854417800903, "learning_rate": 0.0005, "loss": 2.4799318313598633, "step": 1761 }, { "epoch": 0.20502676285780777, "grad_norm": 1.0304756164550781, "learning_rate": 0.0005, "loss": 2.706636428833008, "step": 1762 }, { "epoch": 0.20514312310914593, "grad_norm": 1.489940881729126, "learning_rate": 0.0005, "loss": 3.092426300048828, "step": 1763 }, { "epoch": 0.20525948336048405, "grad_norm": 1.1130279302597046, "learning_rate": 0.0005, "loss": 2.286238431930542, "step": 1764 }, { "epoch": 0.2053758436118222, "grad_norm": 2.257755756378174, "learning_rate": 0.0005, "loss": 3.2333216667175293, "step": 1765 }, { "epoch": 0.20549220386316033, "grad_norm": 1.7450517416000366, "learning_rate": 0.0005, "loss": 3.1049728393554688, "step": 1766 }, { "epoch": 0.2056085641144985, "grad_norm": 1.8733328580856323, "learning_rate": 0.0005, "loss": 3.6814403533935547, "step": 1767 }, { "epoch": 0.20572492436583664, "grad_norm": 1.0874592065811157, "learning_rate": 0.0005, "loss": 2.5849318504333496, "step": 1768 }, { "epoch": 0.20584128461717477, "grad_norm": 1.8687841892242432, "learning_rate": 0.0005, "loss": 2.3563783168792725, "step": 1769 }, { "epoch": 0.20595764486851292, "grad_norm": 1.4830468893051147, "learning_rate": 0.0005, "loss": 2.7771263122558594, "step": 1770 }, { "epoch": 0.20607400511985105, "grad_norm": 1.0466889142990112, "learning_rate": 0.0005, "loss": 2.295929193496704, "step": 1771 }, { "epoch": 0.2061903653711892, "grad_norm": 2.1012094020843506, "learning_rate": 0.0005, "loss": 3.154038906097412, "step": 1772 }, { "epoch": 0.20630672562252733, "grad_norm": 1.820001482963562, "learning_rate": 0.0005, "loss": 2.5873680114746094, "step": 1773 }, { "epoch": 0.2064230858738655, "grad_norm": 0.824118971824646, "learning_rate": 0.0005, "loss": 1.6975750923156738, "step": 1774 }, { "epoch": 0.20653944612520364, "grad_norm": 2.073272228240967, "learning_rate": 0.0005, "loss": 2.7135610580444336, "step": 1775 }, { "epoch": 0.20665580637654177, "grad_norm": 1.4530901908874512, "learning_rate": 0.0005, "loss": 2.7803611755371094, "step": 1776 }, { "epoch": 0.20677216662787992, "grad_norm": 1.7434327602386475, "learning_rate": 0.0005, "loss": 2.9192256927490234, "step": 1777 }, { "epoch": 0.20688852687921805, "grad_norm": 0.938852846622467, "learning_rate": 0.0005, "loss": 2.2381184101104736, "step": 1778 }, { "epoch": 0.2070048871305562, "grad_norm": 1.1611731052398682, "learning_rate": 0.0005, "loss": 2.6013424396514893, "step": 1779 }, { "epoch": 0.20712124738189436, "grad_norm": 1.1238296031951904, "learning_rate": 0.0005, "loss": 2.247361660003662, "step": 1780 }, { "epoch": 0.20723760763323248, "grad_norm": 0.9096644520759583, "learning_rate": 0.0005, "loss": 2.0229744911193848, "step": 1781 }, { "epoch": 0.20735396788457064, "grad_norm": 1.3605525493621826, "learning_rate": 0.0005, "loss": 2.260359287261963, "step": 1782 }, { "epoch": 0.20747032813590877, "grad_norm": 1.261316180229187, "learning_rate": 0.0005, "loss": 1.9990429878234863, "step": 1783 }, { "epoch": 0.20758668838724692, "grad_norm": 1.9120123386383057, "learning_rate": 0.0005, "loss": 3.118513822555542, "step": 1784 }, { "epoch": 0.20770304863858505, "grad_norm": 1.3453010320663452, "learning_rate": 0.0005, "loss": 1.649718999862671, "step": 1785 }, { "epoch": 0.2078194088899232, "grad_norm": 1.4551962614059448, "learning_rate": 0.0005, "loss": 1.8741028308868408, "step": 1786 }, { "epoch": 0.20793576914126136, "grad_norm": 1.8477280139923096, "learning_rate": 0.0005, "loss": 2.4574403762817383, "step": 1787 }, { "epoch": 0.20805212939259948, "grad_norm": 2.5272841453552246, "learning_rate": 0.0005, "loss": 2.769925355911255, "step": 1788 }, { "epoch": 0.20816848964393764, "grad_norm": 1.6665104627609253, "learning_rate": 0.0005, "loss": 3.122305393218994, "step": 1789 }, { "epoch": 0.20828484989527576, "grad_norm": 1.4327303171157837, "learning_rate": 0.0005, "loss": 3.5265145301818848, "step": 1790 }, { "epoch": 0.20840121014661392, "grad_norm": 1.2969571352005005, "learning_rate": 0.0005, "loss": 3.294512987136841, "step": 1791 }, { "epoch": 0.20851757039795207, "grad_norm": 1.894049048423767, "learning_rate": 0.0005, "loss": 2.914736270904541, "step": 1792 }, { "epoch": 0.2086339306492902, "grad_norm": 1.3299120664596558, "learning_rate": 0.0005, "loss": 1.5527993440628052, "step": 1793 }, { "epoch": 0.20875029090062835, "grad_norm": 1.5812400579452515, "learning_rate": 0.0005, "loss": 2.282374143600464, "step": 1794 }, { "epoch": 0.20886665115196648, "grad_norm": 1.4055771827697754, "learning_rate": 0.0005, "loss": 1.9367930889129639, "step": 1795 }, { "epoch": 0.20898301140330464, "grad_norm": 1.1389321088790894, "learning_rate": 0.0005, "loss": 2.3263728618621826, "step": 1796 }, { "epoch": 0.20909937165464276, "grad_norm": 0.992232620716095, "learning_rate": 0.0005, "loss": 2.1385750770568848, "step": 1797 }, { "epoch": 0.20921573190598092, "grad_norm": 1.170425295829773, "learning_rate": 0.0005, "loss": 2.3765881061553955, "step": 1798 }, { "epoch": 0.20933209215731907, "grad_norm": 1.5558797121047974, "learning_rate": 0.0005, "loss": 2.73748517036438, "step": 1799 }, { "epoch": 0.2094484524086572, "grad_norm": 1.4725152254104614, "learning_rate": 0.0005, "loss": 2.281399726867676, "step": 1800 }, { "epoch": 0.20956481265999535, "grad_norm": 1.5058929920196533, "learning_rate": 0.0005, "loss": 2.7640933990478516, "step": 1801 }, { "epoch": 0.20968117291133348, "grad_norm": 1.6797609329223633, "learning_rate": 0.0005, "loss": 2.6772077083587646, "step": 1802 }, { "epoch": 0.20979753316267163, "grad_norm": 2.4325199127197266, "learning_rate": 0.0005, "loss": 2.2778539657592773, "step": 1803 }, { "epoch": 0.20991389341400976, "grad_norm": 1.438475251197815, "learning_rate": 0.0005, "loss": 2.1986405849456787, "step": 1804 }, { "epoch": 0.21003025366534792, "grad_norm": 2.664564847946167, "learning_rate": 0.0005, "loss": 2.792201519012451, "step": 1805 }, { "epoch": 0.21014661391668607, "grad_norm": 1.9777635335922241, "learning_rate": 0.0005, "loss": 2.813058853149414, "step": 1806 }, { "epoch": 0.2102629741680242, "grad_norm": 1.7879353761672974, "learning_rate": 0.0005, "loss": 2.8519577980041504, "step": 1807 }, { "epoch": 0.21037933441936235, "grad_norm": 1.604702115058899, "learning_rate": 0.0005, "loss": 2.149984359741211, "step": 1808 }, { "epoch": 0.21049569467070048, "grad_norm": 1.3922832012176514, "learning_rate": 0.0005, "loss": 2.413398027420044, "step": 1809 }, { "epoch": 0.21061205492203863, "grad_norm": 1.04544997215271, "learning_rate": 0.0005, "loss": 3.287208080291748, "step": 1810 }, { "epoch": 0.2107284151733768, "grad_norm": 1.0555675029754639, "learning_rate": 0.0005, "loss": 3.06223464012146, "step": 1811 }, { "epoch": 0.21084477542471491, "grad_norm": 2.527863025665283, "learning_rate": 0.0005, "loss": 2.3153417110443115, "step": 1812 }, { "epoch": 0.21096113567605307, "grad_norm": 1.5073294639587402, "learning_rate": 0.0005, "loss": 3.597105026245117, "step": 1813 }, { "epoch": 0.2110774959273912, "grad_norm": 1.893275260925293, "learning_rate": 0.0005, "loss": 3.3674492835998535, "step": 1814 }, { "epoch": 0.21119385617872935, "grad_norm": 1.684256911277771, "learning_rate": 0.0005, "loss": 3.3083391189575195, "step": 1815 }, { "epoch": 0.21131021643006748, "grad_norm": 1.2630271911621094, "learning_rate": 0.0005, "loss": 3.024655342102051, "step": 1816 }, { "epoch": 0.21142657668140563, "grad_norm": 2.590808629989624, "learning_rate": 0.0005, "loss": 3.1055092811584473, "step": 1817 }, { "epoch": 0.21154293693274379, "grad_norm": 2.5623714923858643, "learning_rate": 0.0005, "loss": 4.102719783782959, "step": 1818 }, { "epoch": 0.2116592971840819, "grad_norm": 2.818323850631714, "learning_rate": 0.0005, "loss": 3.0502243041992188, "step": 1819 }, { "epoch": 0.21177565743542007, "grad_norm": 2.0595133304595947, "learning_rate": 0.0005, "loss": 3.089907169342041, "step": 1820 }, { "epoch": 0.2118920176867582, "grad_norm": 1.6325641870498657, "learning_rate": 0.0005, "loss": 2.552997350692749, "step": 1821 }, { "epoch": 0.21200837793809635, "grad_norm": 3.444408416748047, "learning_rate": 0.0005, "loss": 2.1208841800689697, "step": 1822 }, { "epoch": 0.2121247381894345, "grad_norm": 1.4986313581466675, "learning_rate": 0.0005, "loss": 2.6409003734588623, "step": 1823 }, { "epoch": 0.21224109844077263, "grad_norm": 1.2863872051239014, "learning_rate": 0.0005, "loss": 2.860323667526245, "step": 1824 }, { "epoch": 0.21235745869211078, "grad_norm": 1.3644609451293945, "learning_rate": 0.0005, "loss": 3.009158134460449, "step": 1825 }, { "epoch": 0.2124738189434489, "grad_norm": 1.535202980041504, "learning_rate": 0.0005, "loss": 4.181195259094238, "step": 1826 }, { "epoch": 0.21259017919478707, "grad_norm": 1.642413854598999, "learning_rate": 0.0005, "loss": 3.1619396209716797, "step": 1827 }, { "epoch": 0.2127065394461252, "grad_norm": 2.2178914546966553, "learning_rate": 0.0005, "loss": 2.2819814682006836, "step": 1828 }, { "epoch": 0.21282289969746335, "grad_norm": 1.6859033107757568, "learning_rate": 0.0005, "loss": 2.688703775405884, "step": 1829 }, { "epoch": 0.2129392599488015, "grad_norm": 1.7083638906478882, "learning_rate": 0.0005, "loss": 2.7603073120117188, "step": 1830 }, { "epoch": 0.21305562020013963, "grad_norm": 1.3833396434783936, "learning_rate": 0.0005, "loss": 3.0665946006774902, "step": 1831 }, { "epoch": 0.21317198045147778, "grad_norm": 1.7270359992980957, "learning_rate": 0.0005, "loss": 3.206141710281372, "step": 1832 }, { "epoch": 0.2132883407028159, "grad_norm": 1.7981091737747192, "learning_rate": 0.0005, "loss": 3.036869525909424, "step": 1833 }, { "epoch": 0.21340470095415406, "grad_norm": 2.018650770187378, "learning_rate": 0.0005, "loss": 2.837920665740967, "step": 1834 }, { "epoch": 0.2135210612054922, "grad_norm": 1.3460259437561035, "learning_rate": 0.0005, "loss": 2.5300230979919434, "step": 1835 }, { "epoch": 0.21363742145683035, "grad_norm": 1.5560133457183838, "learning_rate": 0.0005, "loss": 2.588992118835449, "step": 1836 }, { "epoch": 0.2137537817081685, "grad_norm": 1.796059489250183, "learning_rate": 0.0005, "loss": 4.17003870010376, "step": 1837 }, { "epoch": 0.21387014195950663, "grad_norm": 1.4388108253479004, "learning_rate": 0.0005, "loss": 3.372680187225342, "step": 1838 }, { "epoch": 0.21398650221084478, "grad_norm": 1.4929641485214233, "learning_rate": 0.0005, "loss": 2.827019214630127, "step": 1839 }, { "epoch": 0.2141028624621829, "grad_norm": 1.5839581489562988, "learning_rate": 0.0005, "loss": 3.0295796394348145, "step": 1840 }, { "epoch": 0.21421922271352106, "grad_norm": 3.3849735260009766, "learning_rate": 0.0005, "loss": 3.0454330444335938, "step": 1841 }, { "epoch": 0.21433558296485922, "grad_norm": 1.0872520208358765, "learning_rate": 0.0005, "loss": 2.1773664951324463, "step": 1842 }, { "epoch": 0.21445194321619734, "grad_norm": 1.9034572839736938, "learning_rate": 0.0005, "loss": 2.9046125411987305, "step": 1843 }, { "epoch": 0.2145683034675355, "grad_norm": 2.661412477493286, "learning_rate": 0.0005, "loss": 3.1922171115875244, "step": 1844 }, { "epoch": 0.21468466371887363, "grad_norm": 1.07418954372406, "learning_rate": 0.0005, "loss": 3.3317251205444336, "step": 1845 }, { "epoch": 0.21480102397021178, "grad_norm": 1.466397762298584, "learning_rate": 0.0005, "loss": 3.9917402267456055, "step": 1846 }, { "epoch": 0.2149173842215499, "grad_norm": 1.6631861925125122, "learning_rate": 0.0005, "loss": 2.6725220680236816, "step": 1847 }, { "epoch": 0.21503374447288806, "grad_norm": 1.6594804525375366, "learning_rate": 0.0005, "loss": 3.1515002250671387, "step": 1848 }, { "epoch": 0.21515010472422622, "grad_norm": 0.9890098571777344, "learning_rate": 0.0005, "loss": 2.3900399208068848, "step": 1849 }, { "epoch": 0.21526646497556434, "grad_norm": 1.819234848022461, "learning_rate": 0.0005, "loss": 2.788347005844116, "step": 1850 }, { "epoch": 0.2153828252269025, "grad_norm": 2.101435899734497, "learning_rate": 0.0005, "loss": 2.595958709716797, "step": 1851 }, { "epoch": 0.21549918547824062, "grad_norm": 1.217645525932312, "learning_rate": 0.0005, "loss": 3.331216335296631, "step": 1852 }, { "epoch": 0.21561554572957878, "grad_norm": 2.122714042663574, "learning_rate": 0.0005, "loss": 3.3463010787963867, "step": 1853 }, { "epoch": 0.21573190598091693, "grad_norm": 1.3241801261901855, "learning_rate": 0.0005, "loss": 2.742036819458008, "step": 1854 }, { "epoch": 0.21584826623225506, "grad_norm": 1.3490850925445557, "learning_rate": 0.0005, "loss": 3.2554636001586914, "step": 1855 }, { "epoch": 0.21596462648359321, "grad_norm": 1.1063563823699951, "learning_rate": 0.0005, "loss": 1.8724479675292969, "step": 1856 }, { "epoch": 0.21608098673493134, "grad_norm": 1.7175042629241943, "learning_rate": 0.0005, "loss": 2.790452480316162, "step": 1857 }, { "epoch": 0.2161973469862695, "grad_norm": 1.803737998008728, "learning_rate": 0.0005, "loss": 2.734097719192505, "step": 1858 }, { "epoch": 0.21631370723760762, "grad_norm": 1.329321026802063, "learning_rate": 0.0005, "loss": 3.3589584827423096, "step": 1859 }, { "epoch": 0.21643006748894578, "grad_norm": 1.1778243780136108, "learning_rate": 0.0005, "loss": 2.7449264526367188, "step": 1860 }, { "epoch": 0.21654642774028393, "grad_norm": 1.5053176879882812, "learning_rate": 0.0005, "loss": 1.4929859638214111, "step": 1861 }, { "epoch": 0.21666278799162206, "grad_norm": 2.2740354537963867, "learning_rate": 0.0005, "loss": 3.0929853916168213, "step": 1862 }, { "epoch": 0.2167791482429602, "grad_norm": 2.0512239933013916, "learning_rate": 0.0005, "loss": 3.267381191253662, "step": 1863 }, { "epoch": 0.21689550849429834, "grad_norm": 1.306561827659607, "learning_rate": 0.0005, "loss": 2.6178348064422607, "step": 1864 }, { "epoch": 0.2170118687456365, "grad_norm": 0.9434206485748291, "learning_rate": 0.0005, "loss": 2.3729639053344727, "step": 1865 }, { "epoch": 0.21712822899697462, "grad_norm": 1.1772795915603638, "learning_rate": 0.0005, "loss": 2.8758509159088135, "step": 1866 }, { "epoch": 0.21724458924831277, "grad_norm": 2.4374117851257324, "learning_rate": 0.0005, "loss": 2.6253609657287598, "step": 1867 }, { "epoch": 0.21736094949965093, "grad_norm": 1.2731140851974487, "learning_rate": 0.0005, "loss": 2.3374581336975098, "step": 1868 }, { "epoch": 0.21747730975098906, "grad_norm": 2.4684953689575195, "learning_rate": 0.0005, "loss": 3.316681385040283, "step": 1869 }, { "epoch": 0.2175936700023272, "grad_norm": 1.5461894273757935, "learning_rate": 0.0005, "loss": 2.775940418243408, "step": 1870 }, { "epoch": 0.21771003025366534, "grad_norm": 2.470789909362793, "learning_rate": 0.0005, "loss": 3.0568838119506836, "step": 1871 }, { "epoch": 0.2178263905050035, "grad_norm": 0.9992440342903137, "learning_rate": 0.0005, "loss": 1.7191905975341797, "step": 1872 }, { "epoch": 0.21794275075634165, "grad_norm": 0.8057507872581482, "learning_rate": 0.0005, "loss": 1.4898128509521484, "step": 1873 }, { "epoch": 0.21805911100767977, "grad_norm": 2.0177128314971924, "learning_rate": 0.0005, "loss": 2.971724510192871, "step": 1874 }, { "epoch": 0.21817547125901793, "grad_norm": 2.4672253131866455, "learning_rate": 0.0005, "loss": 3.4289965629577637, "step": 1875 }, { "epoch": 0.21829183151035605, "grad_norm": 1.460632085800171, "learning_rate": 0.0005, "loss": 2.7185111045837402, "step": 1876 }, { "epoch": 0.2184081917616942, "grad_norm": 1.4028069972991943, "learning_rate": 0.0005, "loss": 3.3400015830993652, "step": 1877 }, { "epoch": 0.21852455201303234, "grad_norm": 0.8378100991249084, "learning_rate": 0.0005, "loss": 2.7947804927825928, "step": 1878 }, { "epoch": 0.2186409122643705, "grad_norm": 1.8938777446746826, "learning_rate": 0.0005, "loss": 2.942072868347168, "step": 1879 }, { "epoch": 0.21875727251570865, "grad_norm": 1.6219985485076904, "learning_rate": 0.0005, "loss": 3.915417194366455, "step": 1880 }, { "epoch": 0.21887363276704677, "grad_norm": 1.5737286806106567, "learning_rate": 0.0005, "loss": 3.529770612716675, "step": 1881 }, { "epoch": 0.21898999301838493, "grad_norm": 0.8699261546134949, "learning_rate": 0.0005, "loss": 1.505272388458252, "step": 1882 }, { "epoch": 0.21910635326972305, "grad_norm": 1.0247377157211304, "learning_rate": 0.0005, "loss": 3.3551342487335205, "step": 1883 }, { "epoch": 0.2192227135210612, "grad_norm": 1.4704298973083496, "learning_rate": 0.0005, "loss": 3.6647069454193115, "step": 1884 }, { "epoch": 0.21933907377239936, "grad_norm": 1.4831422567367554, "learning_rate": 0.0005, "loss": 3.4518465995788574, "step": 1885 }, { "epoch": 0.2194554340237375, "grad_norm": 1.3866097927093506, "learning_rate": 0.0005, "loss": 3.2278099060058594, "step": 1886 }, { "epoch": 0.21957179427507564, "grad_norm": 1.1245250701904297, "learning_rate": 0.0005, "loss": 2.6513748168945312, "step": 1887 }, { "epoch": 0.21968815452641377, "grad_norm": 1.5698460340499878, "learning_rate": 0.0005, "loss": 3.049520969390869, "step": 1888 }, { "epoch": 0.21980451477775192, "grad_norm": 1.3877687454223633, "learning_rate": 0.0005, "loss": 4.084006309509277, "step": 1889 }, { "epoch": 0.21992087502909005, "grad_norm": 1.6831880807876587, "learning_rate": 0.0005, "loss": 2.4963231086730957, "step": 1890 }, { "epoch": 0.2200372352804282, "grad_norm": 1.473719835281372, "learning_rate": 0.0005, "loss": 2.4043357372283936, "step": 1891 }, { "epoch": 0.22015359553176636, "grad_norm": 1.5157462358474731, "learning_rate": 0.0005, "loss": 2.8022313117980957, "step": 1892 }, { "epoch": 0.2202699557831045, "grad_norm": 2.059622287750244, "learning_rate": 0.0005, "loss": 3.5610599517822266, "step": 1893 }, { "epoch": 0.22038631603444264, "grad_norm": 1.2204917669296265, "learning_rate": 0.0005, "loss": 2.4152138233184814, "step": 1894 }, { "epoch": 0.22050267628578077, "grad_norm": 1.9463512897491455, "learning_rate": 0.0005, "loss": 2.854403257369995, "step": 1895 }, { "epoch": 0.22061903653711892, "grad_norm": 1.1644350290298462, "learning_rate": 0.0005, "loss": 2.6105005741119385, "step": 1896 }, { "epoch": 0.22073539678845705, "grad_norm": 1.6353756189346313, "learning_rate": 0.0005, "loss": 3.343406915664673, "step": 1897 }, { "epoch": 0.2208517570397952, "grad_norm": 0.9246922731399536, "learning_rate": 0.0005, "loss": 2.035646438598633, "step": 1898 }, { "epoch": 0.22096811729113336, "grad_norm": 1.1097502708435059, "learning_rate": 0.0005, "loss": 2.3795833587646484, "step": 1899 }, { "epoch": 0.22108447754247149, "grad_norm": 1.6585819721221924, "learning_rate": 0.0005, "loss": 2.7984232902526855, "step": 1900 }, { "epoch": 0.22120083779380964, "grad_norm": 1.6707987785339355, "learning_rate": 0.0005, "loss": 3.4430124759674072, "step": 1901 }, { "epoch": 0.22131719804514777, "grad_norm": 1.3047199249267578, "learning_rate": 0.0005, "loss": 2.4524338245391846, "step": 1902 }, { "epoch": 0.22143355829648592, "grad_norm": 1.027960181236267, "learning_rate": 0.0005, "loss": 1.9055365324020386, "step": 1903 }, { "epoch": 0.22154991854782408, "grad_norm": 1.3529175519943237, "learning_rate": 0.0005, "loss": 2.865999698638916, "step": 1904 }, { "epoch": 0.2216662787991622, "grad_norm": 1.3573176860809326, "learning_rate": 0.0005, "loss": 2.9344372749328613, "step": 1905 }, { "epoch": 0.22178263905050036, "grad_norm": 1.3330446481704712, "learning_rate": 0.0005, "loss": 3.0312066078186035, "step": 1906 }, { "epoch": 0.22189899930183848, "grad_norm": 1.6583548784255981, "learning_rate": 0.0005, "loss": 2.7864863872528076, "step": 1907 }, { "epoch": 0.22201535955317664, "grad_norm": 1.2352924346923828, "learning_rate": 0.0005, "loss": 2.7064011096954346, "step": 1908 }, { "epoch": 0.22213171980451477, "grad_norm": 1.7297879457473755, "learning_rate": 0.0005, "loss": 2.531543254852295, "step": 1909 }, { "epoch": 0.22224808005585292, "grad_norm": 2.9444479942321777, "learning_rate": 0.0005, "loss": 3.602180004119873, "step": 1910 }, { "epoch": 0.22236444030719107, "grad_norm": 2.1488730907440186, "learning_rate": 0.0005, "loss": 3.3208401203155518, "step": 1911 }, { "epoch": 0.2224808005585292, "grad_norm": 1.758061170578003, "learning_rate": 0.0005, "loss": 2.865919589996338, "step": 1912 }, { "epoch": 0.22259716080986736, "grad_norm": 1.1200721263885498, "learning_rate": 0.0005, "loss": 1.8514150381088257, "step": 1913 }, { "epoch": 0.22271352106120548, "grad_norm": 1.4694349765777588, "learning_rate": 0.0005, "loss": 3.3310205936431885, "step": 1914 }, { "epoch": 0.22282988131254364, "grad_norm": 1.2346962690353394, "learning_rate": 0.0005, "loss": 3.165090560913086, "step": 1915 }, { "epoch": 0.2229462415638818, "grad_norm": 2.165590763092041, "learning_rate": 0.0005, "loss": 3.212001323699951, "step": 1916 }, { "epoch": 0.22306260181521992, "grad_norm": 1.3876787424087524, "learning_rate": 0.0005, "loss": 3.250732421875, "step": 1917 }, { "epoch": 0.22317896206655807, "grad_norm": 2.4383292198181152, "learning_rate": 0.0005, "loss": 3.9888803958892822, "step": 1918 }, { "epoch": 0.2232953223178962, "grad_norm": 2.2620084285736084, "learning_rate": 0.0005, "loss": 3.7929792404174805, "step": 1919 }, { "epoch": 0.22341168256923435, "grad_norm": 1.6101529598236084, "learning_rate": 0.0005, "loss": 3.030651569366455, "step": 1920 }, { "epoch": 0.22352804282057248, "grad_norm": 1.7306513786315918, "learning_rate": 0.0005, "loss": 2.255587577819824, "step": 1921 }, { "epoch": 0.22364440307191064, "grad_norm": 1.1321943998336792, "learning_rate": 0.0005, "loss": 3.016376495361328, "step": 1922 }, { "epoch": 0.2237607633232488, "grad_norm": 0.9583985209465027, "learning_rate": 0.0005, "loss": 1.8740531206130981, "step": 1923 }, { "epoch": 0.22387712357458692, "grad_norm": 2.5417215824127197, "learning_rate": 0.0005, "loss": 3.253241777420044, "step": 1924 }, { "epoch": 0.22399348382592507, "grad_norm": 2.1620943546295166, "learning_rate": 0.0005, "loss": 3.386707305908203, "step": 1925 }, { "epoch": 0.2241098440772632, "grad_norm": 1.9717833995819092, "learning_rate": 0.0005, "loss": 2.276956558227539, "step": 1926 }, { "epoch": 0.22422620432860135, "grad_norm": 1.3741368055343628, "learning_rate": 0.0005, "loss": 2.2694454193115234, "step": 1927 }, { "epoch": 0.22434256457993948, "grad_norm": 1.380843162536621, "learning_rate": 0.0005, "loss": 2.2942774295806885, "step": 1928 }, { "epoch": 0.22445892483127763, "grad_norm": 2.055002212524414, "learning_rate": 0.0005, "loss": 3.3147740364074707, "step": 1929 }, { "epoch": 0.2245752850826158, "grad_norm": 0.9967233538627625, "learning_rate": 0.0005, "loss": 2.11016583442688, "step": 1930 }, { "epoch": 0.22469164533395392, "grad_norm": 1.2809038162231445, "learning_rate": 0.0005, "loss": 2.904726505279541, "step": 1931 }, { "epoch": 0.22480800558529207, "grad_norm": 1.2346303462982178, "learning_rate": 0.0005, "loss": 3.320587396621704, "step": 1932 }, { "epoch": 0.2249243658366302, "grad_norm": 3.1294362545013428, "learning_rate": 0.0005, "loss": 2.3210387229919434, "step": 1933 }, { "epoch": 0.22504072608796835, "grad_norm": 1.980728030204773, "learning_rate": 0.0005, "loss": 2.3783140182495117, "step": 1934 }, { "epoch": 0.2251570863393065, "grad_norm": 4.748810291290283, "learning_rate": 0.0005, "loss": 3.6052958965301514, "step": 1935 }, { "epoch": 0.22527344659064463, "grad_norm": 1.4236210584640503, "learning_rate": 0.0005, "loss": 2.850560188293457, "step": 1936 }, { "epoch": 0.2253898068419828, "grad_norm": 1.4556043148040771, "learning_rate": 0.0005, "loss": 2.894773483276367, "step": 1937 }, { "epoch": 0.22550616709332091, "grad_norm": 1.5484142303466797, "learning_rate": 0.0005, "loss": 2.742433547973633, "step": 1938 }, { "epoch": 0.22562252734465907, "grad_norm": 1.1518751382827759, "learning_rate": 0.0005, "loss": 2.8193953037261963, "step": 1939 }, { "epoch": 0.2257388875959972, "grad_norm": 0.6955416202545166, "learning_rate": 0.0005, "loss": 1.713208556175232, "step": 1940 }, { "epoch": 0.22585524784733535, "grad_norm": 2.1258955001831055, "learning_rate": 0.0005, "loss": 3.6144895553588867, "step": 1941 }, { "epoch": 0.2259716080986735, "grad_norm": 2.093395471572876, "learning_rate": 0.0005, "loss": 3.3913631439208984, "step": 1942 }, { "epoch": 0.22608796835001163, "grad_norm": 1.2382853031158447, "learning_rate": 0.0005, "loss": 1.70235276222229, "step": 1943 }, { "epoch": 0.22620432860134979, "grad_norm": 0.664587140083313, "learning_rate": 0.0005, "loss": 1.1767754554748535, "step": 1944 }, { "epoch": 0.2263206888526879, "grad_norm": 1.3460766077041626, "learning_rate": 0.0005, "loss": 2.302358627319336, "step": 1945 }, { "epoch": 0.22643704910402607, "grad_norm": 0.999992847442627, "learning_rate": 0.0005, "loss": 2.5645365715026855, "step": 1946 }, { "epoch": 0.22655340935536422, "grad_norm": 2.162304162979126, "learning_rate": 0.0005, "loss": 3.313053607940674, "step": 1947 }, { "epoch": 0.22666976960670235, "grad_norm": 0.9508485794067383, "learning_rate": 0.0005, "loss": 1.9269522428512573, "step": 1948 }, { "epoch": 0.2267861298580405, "grad_norm": 1.0091514587402344, "learning_rate": 0.0005, "loss": 2.592957019805908, "step": 1949 }, { "epoch": 0.22690249010937863, "grad_norm": 1.8960503339767456, "learning_rate": 0.0005, "loss": 3.3396213054656982, "step": 1950 }, { "epoch": 0.22701885036071678, "grad_norm": 1.0093789100646973, "learning_rate": 0.0005, "loss": 2.1668882369995117, "step": 1951 }, { "epoch": 0.2271352106120549, "grad_norm": 1.8258392810821533, "learning_rate": 0.0005, "loss": 2.997786045074463, "step": 1952 }, { "epoch": 0.22725157086339307, "grad_norm": 2.5867514610290527, "learning_rate": 0.0005, "loss": 2.846627712249756, "step": 1953 }, { "epoch": 0.22736793111473122, "grad_norm": 1.870158314704895, "learning_rate": 0.0005, "loss": 3.2447283267974854, "step": 1954 }, { "epoch": 0.22748429136606935, "grad_norm": 1.6012407541275024, "learning_rate": 0.0005, "loss": 2.6426572799682617, "step": 1955 }, { "epoch": 0.2276006516174075, "grad_norm": 2.2913525104522705, "learning_rate": 0.0005, "loss": 2.8346238136291504, "step": 1956 }, { "epoch": 0.22771701186874563, "grad_norm": 1.8972386121749878, "learning_rate": 0.0005, "loss": 2.818377733230591, "step": 1957 }, { "epoch": 0.22783337212008378, "grad_norm": 1.3989150524139404, "learning_rate": 0.0005, "loss": 2.8084542751312256, "step": 1958 }, { "epoch": 0.2279497323714219, "grad_norm": 1.8632177114486694, "learning_rate": 0.0005, "loss": 3.1197621822357178, "step": 1959 }, { "epoch": 0.22806609262276006, "grad_norm": 1.3721699714660645, "learning_rate": 0.0005, "loss": 2.8177058696746826, "step": 1960 }, { "epoch": 0.22818245287409822, "grad_norm": 1.5446667671203613, "learning_rate": 0.0005, "loss": 2.194685459136963, "step": 1961 }, { "epoch": 0.22829881312543635, "grad_norm": 1.83095121383667, "learning_rate": 0.0005, "loss": 4.4959940910339355, "step": 1962 }, { "epoch": 0.2284151733767745, "grad_norm": 2.5814359188079834, "learning_rate": 0.0005, "loss": 2.9630696773529053, "step": 1963 }, { "epoch": 0.22853153362811263, "grad_norm": 2.393503427505493, "learning_rate": 0.0005, "loss": 3.365973711013794, "step": 1964 }, { "epoch": 0.22864789387945078, "grad_norm": 1.1047462224960327, "learning_rate": 0.0005, "loss": 2.2186379432678223, "step": 1965 }, { "epoch": 0.22876425413078894, "grad_norm": 1.5829651355743408, "learning_rate": 0.0005, "loss": 2.376771926879883, "step": 1966 }, { "epoch": 0.22888061438212706, "grad_norm": 0.9126771092414856, "learning_rate": 0.0005, "loss": 1.911980390548706, "step": 1967 }, { "epoch": 0.22899697463346522, "grad_norm": 1.9452184438705444, "learning_rate": 0.0005, "loss": 2.6055681705474854, "step": 1968 }, { "epoch": 0.22911333488480334, "grad_norm": 1.4070595502853394, "learning_rate": 0.0005, "loss": 2.7665843963623047, "step": 1969 }, { "epoch": 0.2292296951361415, "grad_norm": 1.196320652961731, "learning_rate": 0.0005, "loss": 2.479368209838867, "step": 1970 }, { "epoch": 0.22934605538747962, "grad_norm": 1.985961675643921, "learning_rate": 0.0005, "loss": 2.4689207077026367, "step": 1971 }, { "epoch": 0.22946241563881778, "grad_norm": 1.296837329864502, "learning_rate": 0.0005, "loss": 2.8341643810272217, "step": 1972 }, { "epoch": 0.22957877589015593, "grad_norm": 1.194145679473877, "learning_rate": 0.0005, "loss": 3.0795881748199463, "step": 1973 }, { "epoch": 0.22969513614149406, "grad_norm": 2.1702260971069336, "learning_rate": 0.0005, "loss": 3.3988192081451416, "step": 1974 }, { "epoch": 0.22981149639283222, "grad_norm": 1.6786917448043823, "learning_rate": 0.0005, "loss": 3.548382043838501, "step": 1975 }, { "epoch": 0.22992785664417034, "grad_norm": 1.5254594087600708, "learning_rate": 0.0005, "loss": 2.386417865753174, "step": 1976 }, { "epoch": 0.2300442168955085, "grad_norm": 2.944565773010254, "learning_rate": 0.0005, "loss": 2.3027801513671875, "step": 1977 }, { "epoch": 0.23016057714684665, "grad_norm": 1.1963938474655151, "learning_rate": 0.0005, "loss": 2.2427632808685303, "step": 1978 }, { "epoch": 0.23027693739818478, "grad_norm": 1.460768699645996, "learning_rate": 0.0005, "loss": 3.1595520973205566, "step": 1979 }, { "epoch": 0.23039329764952293, "grad_norm": 1.1512938737869263, "learning_rate": 0.0005, "loss": 3.037415027618408, "step": 1980 }, { "epoch": 0.23050965790086106, "grad_norm": 2.1783220767974854, "learning_rate": 0.0005, "loss": 2.798811197280884, "step": 1981 }, { "epoch": 0.2306260181521992, "grad_norm": 1.5940921306610107, "learning_rate": 0.0005, "loss": 2.5909736156463623, "step": 1982 }, { "epoch": 0.23074237840353734, "grad_norm": 1.3779934644699097, "learning_rate": 0.0005, "loss": 3.024296283721924, "step": 1983 }, { "epoch": 0.2308587386548755, "grad_norm": 1.51166570186615, "learning_rate": 0.0005, "loss": 2.7177553176879883, "step": 1984 }, { "epoch": 0.23097509890621365, "grad_norm": 0.8711497783660889, "learning_rate": 0.0005, "loss": 1.4486808776855469, "step": 1985 }, { "epoch": 0.23109145915755178, "grad_norm": 1.944928526878357, "learning_rate": 0.0005, "loss": 2.5176615715026855, "step": 1986 }, { "epoch": 0.23120781940888993, "grad_norm": 1.6162831783294678, "learning_rate": 0.0005, "loss": 3.6867294311523438, "step": 1987 }, { "epoch": 0.23132417966022806, "grad_norm": 1.3684206008911133, "learning_rate": 0.0005, "loss": 2.5284998416900635, "step": 1988 }, { "epoch": 0.2314405399115662, "grad_norm": 1.7091127634048462, "learning_rate": 0.0005, "loss": 3.1685941219329834, "step": 1989 }, { "epoch": 0.23155690016290434, "grad_norm": 1.5162988901138306, "learning_rate": 0.0005, "loss": 1.8163009881973267, "step": 1990 }, { "epoch": 0.2316732604142425, "grad_norm": 1.5385851860046387, "learning_rate": 0.0005, "loss": 2.57222843170166, "step": 1991 }, { "epoch": 0.23178962066558065, "grad_norm": 1.5276060104370117, "learning_rate": 0.0005, "loss": 2.1820921897888184, "step": 1992 }, { "epoch": 0.23190598091691877, "grad_norm": 1.6749486923217773, "learning_rate": 0.0005, "loss": 1.8496737480163574, "step": 1993 }, { "epoch": 0.23202234116825693, "grad_norm": 1.504396677017212, "learning_rate": 0.0005, "loss": 2.7035298347473145, "step": 1994 }, { "epoch": 0.23213870141959506, "grad_norm": 1.7893058061599731, "learning_rate": 0.0005, "loss": 4.082459449768066, "step": 1995 }, { "epoch": 0.2322550616709332, "grad_norm": 1.0803303718566895, "learning_rate": 0.0005, "loss": 2.542609691619873, "step": 1996 }, { "epoch": 0.23237142192227137, "grad_norm": 0.8738937377929688, "learning_rate": 0.0005, "loss": 1.8285503387451172, "step": 1997 }, { "epoch": 0.2324877821736095, "grad_norm": 1.1583905220031738, "learning_rate": 0.0005, "loss": 2.5323104858398438, "step": 1998 }, { "epoch": 0.23260414242494765, "grad_norm": 1.0875431299209595, "learning_rate": 0.0005, "loss": 1.7663723230361938, "step": 1999 }, { "epoch": 0.23272050267628577, "grad_norm": 1.7441260814666748, "learning_rate": 0.0005, "loss": 3.934447765350342, "step": 2000 }, { "epoch": 0.23283686292762393, "grad_norm": 1.4424196481704712, "learning_rate": 0.0005, "loss": 2.737191677093506, "step": 2001 }, { "epoch": 0.23295322317896205, "grad_norm": 1.1108288764953613, "learning_rate": 0.0005, "loss": 2.9903852939605713, "step": 2002 }, { "epoch": 0.2330695834303002, "grad_norm": 1.813579797744751, "learning_rate": 0.0005, "loss": 2.674086570739746, "step": 2003 }, { "epoch": 0.23318594368163836, "grad_norm": 1.800194263458252, "learning_rate": 0.0005, "loss": 2.3742475509643555, "step": 2004 }, { "epoch": 0.2333023039329765, "grad_norm": 2.4165728092193604, "learning_rate": 0.0005, "loss": 2.003248929977417, "step": 2005 }, { "epoch": 0.23341866418431464, "grad_norm": 1.7828632593154907, "learning_rate": 0.0005, "loss": 3.6325058937072754, "step": 2006 }, { "epoch": 0.23353502443565277, "grad_norm": 2.2356231212615967, "learning_rate": 0.0005, "loss": 3.4995765686035156, "step": 2007 }, { "epoch": 0.23365138468699093, "grad_norm": 1.6767470836639404, "learning_rate": 0.0005, "loss": 2.9249353408813477, "step": 2008 }, { "epoch": 0.23376774493832908, "grad_norm": 1.6398428678512573, "learning_rate": 0.0005, "loss": 3.0443811416625977, "step": 2009 }, { "epoch": 0.2338841051896672, "grad_norm": 1.6286615133285522, "learning_rate": 0.0005, "loss": 3.466829776763916, "step": 2010 }, { "epoch": 0.23400046544100536, "grad_norm": 1.5557527542114258, "learning_rate": 0.0005, "loss": 2.351898670196533, "step": 2011 }, { "epoch": 0.2341168256923435, "grad_norm": 1.6570842266082764, "learning_rate": 0.0005, "loss": 2.676041841506958, "step": 2012 }, { "epoch": 0.23423318594368164, "grad_norm": 1.2421975135803223, "learning_rate": 0.0005, "loss": 2.84279727935791, "step": 2013 }, { "epoch": 0.23434954619501977, "grad_norm": 1.2473797798156738, "learning_rate": 0.0005, "loss": 2.5398130416870117, "step": 2014 }, { "epoch": 0.23446590644635792, "grad_norm": 1.5818469524383545, "learning_rate": 0.0005, "loss": 2.8039135932922363, "step": 2015 }, { "epoch": 0.23458226669769608, "grad_norm": 2.0686750411987305, "learning_rate": 0.0005, "loss": 3.9820556640625, "step": 2016 }, { "epoch": 0.2346986269490342, "grad_norm": 1.617677927017212, "learning_rate": 0.0005, "loss": 3.157148838043213, "step": 2017 }, { "epoch": 0.23481498720037236, "grad_norm": 2.2605152130126953, "learning_rate": 0.0005, "loss": 3.6337337493896484, "step": 2018 }, { "epoch": 0.2349313474517105, "grad_norm": 1.2125521898269653, "learning_rate": 0.0005, "loss": 2.9888882637023926, "step": 2019 }, { "epoch": 0.23504770770304864, "grad_norm": 1.5487085580825806, "learning_rate": 0.0005, "loss": 2.147709846496582, "step": 2020 }, { "epoch": 0.23516406795438677, "grad_norm": 1.2510932683944702, "learning_rate": 0.0005, "loss": 2.6125597953796387, "step": 2021 }, { "epoch": 0.23528042820572492, "grad_norm": 4.69822883605957, "learning_rate": 0.0005, "loss": 2.071094036102295, "step": 2022 }, { "epoch": 0.23539678845706308, "grad_norm": 1.1846556663513184, "learning_rate": 0.0005, "loss": 3.1416797637939453, "step": 2023 }, { "epoch": 0.2355131487084012, "grad_norm": 1.3511103391647339, "learning_rate": 0.0005, "loss": 3.2031617164611816, "step": 2024 }, { "epoch": 0.23562950895973936, "grad_norm": 0.9705579280853271, "learning_rate": 0.0005, "loss": 2.227652072906494, "step": 2025 }, { "epoch": 0.23574586921107749, "grad_norm": 1.4569745063781738, "learning_rate": 0.0005, "loss": 3.5486092567443848, "step": 2026 }, { "epoch": 0.23586222946241564, "grad_norm": 1.6750099658966064, "learning_rate": 0.0005, "loss": 2.7901082038879395, "step": 2027 }, { "epoch": 0.2359785897137538, "grad_norm": 2.067143201828003, "learning_rate": 0.0005, "loss": 2.2974653244018555, "step": 2028 }, { "epoch": 0.23609494996509192, "grad_norm": 2.2380402088165283, "learning_rate": 0.0005, "loss": 2.6009624004364014, "step": 2029 }, { "epoch": 0.23621131021643008, "grad_norm": 1.546571135520935, "learning_rate": 0.0005, "loss": 2.2926931381225586, "step": 2030 }, { "epoch": 0.2363276704677682, "grad_norm": 0.804892897605896, "learning_rate": 0.0005, "loss": 1.9639016389846802, "step": 2031 }, { "epoch": 0.23644403071910636, "grad_norm": 1.0015301704406738, "learning_rate": 0.0005, "loss": 2.2443928718566895, "step": 2032 }, { "epoch": 0.23656039097044448, "grad_norm": 1.4648278951644897, "learning_rate": 0.0005, "loss": 2.302785873413086, "step": 2033 }, { "epoch": 0.23667675122178264, "grad_norm": 1.3597394227981567, "learning_rate": 0.0005, "loss": 2.978607177734375, "step": 2034 }, { "epoch": 0.2367931114731208, "grad_norm": 1.7540919780731201, "learning_rate": 0.0005, "loss": 2.8231053352355957, "step": 2035 }, { "epoch": 0.23690947172445892, "grad_norm": 1.565554141998291, "learning_rate": 0.0005, "loss": 2.782198905944824, "step": 2036 }, { "epoch": 0.23702583197579707, "grad_norm": 1.6597596406936646, "learning_rate": 0.0005, "loss": 1.5503568649291992, "step": 2037 }, { "epoch": 0.2371421922271352, "grad_norm": 1.33372163772583, "learning_rate": 0.0005, "loss": 2.7543110847473145, "step": 2038 }, { "epoch": 0.23725855247847336, "grad_norm": 1.940869927406311, "learning_rate": 0.0005, "loss": 3.251628875732422, "step": 2039 }, { "epoch": 0.2373749127298115, "grad_norm": 1.7158352136611938, "learning_rate": 0.0005, "loss": 2.6279373168945312, "step": 2040 }, { "epoch": 0.23749127298114964, "grad_norm": 1.394553780555725, "learning_rate": 0.0005, "loss": 2.2641663551330566, "step": 2041 }, { "epoch": 0.2376076332324878, "grad_norm": 1.1550694704055786, "learning_rate": 0.0005, "loss": 2.6385855674743652, "step": 2042 }, { "epoch": 0.23772399348382592, "grad_norm": 2.0380189418792725, "learning_rate": 0.0005, "loss": 3.4522383213043213, "step": 2043 }, { "epoch": 0.23784035373516407, "grad_norm": 1.7468080520629883, "learning_rate": 0.0005, "loss": 3.0060677528381348, "step": 2044 }, { "epoch": 0.2379567139865022, "grad_norm": 0.890819251537323, "learning_rate": 0.0005, "loss": 2.5808825492858887, "step": 2045 }, { "epoch": 0.23807307423784035, "grad_norm": 1.0221949815750122, "learning_rate": 0.0005, "loss": 1.9644633531570435, "step": 2046 }, { "epoch": 0.2381894344891785, "grad_norm": 1.2143158912658691, "learning_rate": 0.0005, "loss": 2.100233554840088, "step": 2047 }, { "epoch": 0.23830579474051664, "grad_norm": 1.8257936239242554, "learning_rate": 0.0005, "loss": 3.2014331817626953, "step": 2048 }, { "epoch": 0.2384221549918548, "grad_norm": 1.7131969928741455, "learning_rate": 0.0005, "loss": 3.5367321968078613, "step": 2049 }, { "epoch": 0.23853851524319292, "grad_norm": 1.2958118915557861, "learning_rate": 0.0005, "loss": 2.0508737564086914, "step": 2050 }, { "epoch": 0.23865487549453107, "grad_norm": 1.912474513053894, "learning_rate": 0.0005, "loss": 2.1284842491149902, "step": 2051 }, { "epoch": 0.2387712357458692, "grad_norm": 1.3245501518249512, "learning_rate": 0.0005, "loss": 2.5556211471557617, "step": 2052 }, { "epoch": 0.23888759599720735, "grad_norm": 5.3542160987854, "learning_rate": 0.0005, "loss": 3.2344882488250732, "step": 2053 }, { "epoch": 0.2390039562485455, "grad_norm": 1.8259129524230957, "learning_rate": 0.0005, "loss": 2.7968263626098633, "step": 2054 }, { "epoch": 0.23912031649988363, "grad_norm": 1.4553518295288086, "learning_rate": 0.0005, "loss": 3.1522204875946045, "step": 2055 }, { "epoch": 0.2392366767512218, "grad_norm": 1.796276330947876, "learning_rate": 0.0005, "loss": 2.1088762283325195, "step": 2056 }, { "epoch": 0.23935303700255992, "grad_norm": 1.4609057903289795, "learning_rate": 0.0005, "loss": 2.980403423309326, "step": 2057 }, { "epoch": 0.23946939725389807, "grad_norm": 1.1301411390304565, "learning_rate": 0.0005, "loss": 2.5562729835510254, "step": 2058 }, { "epoch": 0.23958575750523622, "grad_norm": 1.6680655479431152, "learning_rate": 0.0005, "loss": 3.71170973777771, "step": 2059 }, { "epoch": 0.23970211775657435, "grad_norm": 1.6958831548690796, "learning_rate": 0.0005, "loss": 3.260883092880249, "step": 2060 }, { "epoch": 0.2398184780079125, "grad_norm": 0.9068015217781067, "learning_rate": 0.0005, "loss": 1.9898293018341064, "step": 2061 }, { "epoch": 0.23993483825925063, "grad_norm": 1.3619521856307983, "learning_rate": 0.0005, "loss": 2.7686173915863037, "step": 2062 }, { "epoch": 0.2400511985105888, "grad_norm": 8.117517471313477, "learning_rate": 0.0005, "loss": 4.280611515045166, "step": 2063 }, { "epoch": 0.2401675587619269, "grad_norm": 6.086554527282715, "learning_rate": 0.0005, "loss": 2.463881015777588, "step": 2064 }, { "epoch": 0.24028391901326507, "grad_norm": 1.4168068170547485, "learning_rate": 0.0005, "loss": 3.048549175262451, "step": 2065 }, { "epoch": 0.24040027926460322, "grad_norm": 1.141838788986206, "learning_rate": 0.0005, "loss": 2.4659900665283203, "step": 2066 }, { "epoch": 0.24051663951594135, "grad_norm": 1.1214213371276855, "learning_rate": 0.0005, "loss": 2.225743532180786, "step": 2067 }, { "epoch": 0.2406329997672795, "grad_norm": 1.2745879888534546, "learning_rate": 0.0005, "loss": 3.5886142253875732, "step": 2068 }, { "epoch": 0.24074936001861763, "grad_norm": 1.9410568475723267, "learning_rate": 0.0005, "loss": 3.4791746139526367, "step": 2069 }, { "epoch": 0.24086572026995579, "grad_norm": 1.5294867753982544, "learning_rate": 0.0005, "loss": 3.2688839435577393, "step": 2070 }, { "epoch": 0.24098208052129394, "grad_norm": 1.0719656944274902, "learning_rate": 0.0005, "loss": 3.0434982776641846, "step": 2071 }, { "epoch": 0.24109844077263207, "grad_norm": 1.3178482055664062, "learning_rate": 0.0005, "loss": 2.8162424564361572, "step": 2072 }, { "epoch": 0.24121480102397022, "grad_norm": 1.00355863571167, "learning_rate": 0.0005, "loss": 2.230412483215332, "step": 2073 }, { "epoch": 0.24133116127530835, "grad_norm": 1.6818498373031616, "learning_rate": 0.0005, "loss": 3.1771883964538574, "step": 2074 }, { "epoch": 0.2414475215266465, "grad_norm": 1.3334813117980957, "learning_rate": 0.0005, "loss": 2.4136693477630615, "step": 2075 }, { "epoch": 0.24156388177798463, "grad_norm": 1.236397624015808, "learning_rate": 0.0005, "loss": 3.137477159500122, "step": 2076 }, { "epoch": 0.24168024202932278, "grad_norm": 1.9257491827011108, "learning_rate": 0.0005, "loss": 2.5615463256835938, "step": 2077 }, { "epoch": 0.24179660228066094, "grad_norm": 1.6507651805877686, "learning_rate": 0.0005, "loss": 3.1983230113983154, "step": 2078 }, { "epoch": 0.24191296253199907, "grad_norm": 1.180038571357727, "learning_rate": 0.0005, "loss": 2.376162528991699, "step": 2079 }, { "epoch": 0.24202932278333722, "grad_norm": 1.4578036069869995, "learning_rate": 0.0005, "loss": 3.2097678184509277, "step": 2080 }, { "epoch": 0.24214568303467535, "grad_norm": 1.4947612285614014, "learning_rate": 0.0005, "loss": 2.304799795150757, "step": 2081 }, { "epoch": 0.2422620432860135, "grad_norm": 1.2784031629562378, "learning_rate": 0.0005, "loss": 2.223257064819336, "step": 2082 }, { "epoch": 0.24237840353735163, "grad_norm": 1.1404836177825928, "learning_rate": 0.0005, "loss": 2.5058488845825195, "step": 2083 }, { "epoch": 0.24249476378868978, "grad_norm": 1.3033844232559204, "learning_rate": 0.0005, "loss": 2.700371265411377, "step": 2084 }, { "epoch": 0.24261112404002794, "grad_norm": 1.6774605512619019, "learning_rate": 0.0005, "loss": 2.7761940956115723, "step": 2085 }, { "epoch": 0.24272748429136606, "grad_norm": 1.5184510946273804, "learning_rate": 0.0005, "loss": 2.7305264472961426, "step": 2086 }, { "epoch": 0.24284384454270422, "grad_norm": 1.4441016912460327, "learning_rate": 0.0005, "loss": 1.6713067293167114, "step": 2087 }, { "epoch": 0.24296020479404234, "grad_norm": 1.2861828804016113, "learning_rate": 0.0005, "loss": 3.003587007522583, "step": 2088 }, { "epoch": 0.2430765650453805, "grad_norm": 1.695221185684204, "learning_rate": 0.0005, "loss": 2.6247520446777344, "step": 2089 }, { "epoch": 0.24319292529671865, "grad_norm": 1.6894747018814087, "learning_rate": 0.0005, "loss": 2.6688907146453857, "step": 2090 }, { "epoch": 0.24330928554805678, "grad_norm": 1.8660062551498413, "learning_rate": 0.0005, "loss": 3.5948376655578613, "step": 2091 }, { "epoch": 0.24342564579939494, "grad_norm": 1.0014532804489136, "learning_rate": 0.0005, "loss": 1.9742662906646729, "step": 2092 }, { "epoch": 0.24354200605073306, "grad_norm": 1.0387051105499268, "learning_rate": 0.0005, "loss": 2.3384106159210205, "step": 2093 }, { "epoch": 0.24365836630207122, "grad_norm": 1.123916506767273, "learning_rate": 0.0005, "loss": 2.633666515350342, "step": 2094 }, { "epoch": 0.24377472655340934, "grad_norm": 1.4681531190872192, "learning_rate": 0.0005, "loss": 2.8046722412109375, "step": 2095 }, { "epoch": 0.2438910868047475, "grad_norm": 1.5275856256484985, "learning_rate": 0.0005, "loss": 3.1696949005126953, "step": 2096 }, { "epoch": 0.24400744705608565, "grad_norm": 1.287976861000061, "learning_rate": 0.0005, "loss": 1.7997314929962158, "step": 2097 }, { "epoch": 0.24412380730742378, "grad_norm": 1.8472884893417358, "learning_rate": 0.0005, "loss": 1.974682331085205, "step": 2098 }, { "epoch": 0.24424016755876193, "grad_norm": 1.0962353944778442, "learning_rate": 0.0005, "loss": 3.014498472213745, "step": 2099 }, { "epoch": 0.24435652781010006, "grad_norm": 1.3348381519317627, "learning_rate": 0.0005, "loss": 2.894181251525879, "step": 2100 }, { "epoch": 0.24447288806143821, "grad_norm": 1.5656325817108154, "learning_rate": 0.0005, "loss": 2.2860684394836426, "step": 2101 }, { "epoch": 0.24458924831277637, "grad_norm": 1.5738155841827393, "learning_rate": 0.0005, "loss": 1.7422624826431274, "step": 2102 }, { "epoch": 0.2447056085641145, "grad_norm": 1.6365022659301758, "learning_rate": 0.0005, "loss": 2.7288713455200195, "step": 2103 }, { "epoch": 0.24482196881545265, "grad_norm": 1.3621160984039307, "learning_rate": 0.0005, "loss": 2.674382209777832, "step": 2104 }, { "epoch": 0.24493832906679078, "grad_norm": 1.6373966932296753, "learning_rate": 0.0005, "loss": 2.76271915435791, "step": 2105 }, { "epoch": 0.24505468931812893, "grad_norm": 1.562830924987793, "learning_rate": 0.0005, "loss": 3.9520506858825684, "step": 2106 }, { "epoch": 0.24517104956946706, "grad_norm": 1.2259135246276855, "learning_rate": 0.0005, "loss": 2.046985387802124, "step": 2107 }, { "epoch": 0.2452874098208052, "grad_norm": 1.0393617153167725, "learning_rate": 0.0005, "loss": 2.7006142139434814, "step": 2108 }, { "epoch": 0.24540377007214337, "grad_norm": 1.433167815208435, "learning_rate": 0.0005, "loss": 2.717881202697754, "step": 2109 }, { "epoch": 0.2455201303234815, "grad_norm": 1.8185474872589111, "learning_rate": 0.0005, "loss": 3.6641902923583984, "step": 2110 }, { "epoch": 0.24563649057481965, "grad_norm": 1.6998789310455322, "learning_rate": 0.0005, "loss": 2.150144100189209, "step": 2111 }, { "epoch": 0.24575285082615778, "grad_norm": 1.1129181385040283, "learning_rate": 0.0005, "loss": 2.1264190673828125, "step": 2112 }, { "epoch": 0.24586921107749593, "grad_norm": 1.2357577085494995, "learning_rate": 0.0005, "loss": 2.707909107208252, "step": 2113 }, { "epoch": 0.24598557132883406, "grad_norm": 1.8896933794021606, "learning_rate": 0.0005, "loss": 3.328057289123535, "step": 2114 }, { "epoch": 0.2461019315801722, "grad_norm": 1.093569040298462, "learning_rate": 0.0005, "loss": 2.3096461296081543, "step": 2115 }, { "epoch": 0.24621829183151037, "grad_norm": 1.9536887407302856, "learning_rate": 0.0005, "loss": 3.9018807411193848, "step": 2116 }, { "epoch": 0.2463346520828485, "grad_norm": 1.2529191970825195, "learning_rate": 0.0005, "loss": 2.8819143772125244, "step": 2117 }, { "epoch": 0.24645101233418665, "grad_norm": 2.0078821182250977, "learning_rate": 0.0005, "loss": 2.974693775177002, "step": 2118 }, { "epoch": 0.24656737258552477, "grad_norm": 1.6234769821166992, "learning_rate": 0.0005, "loss": 2.0220279693603516, "step": 2119 }, { "epoch": 0.24668373283686293, "grad_norm": 2.033837080001831, "learning_rate": 0.0005, "loss": 3.012603759765625, "step": 2120 }, { "epoch": 0.24680009308820108, "grad_norm": 0.6883848905563354, "learning_rate": 0.0005, "loss": 1.812535047531128, "step": 2121 }, { "epoch": 0.2469164533395392, "grad_norm": 0.9112616181373596, "learning_rate": 0.0005, "loss": 1.8315794467926025, "step": 2122 }, { "epoch": 0.24703281359087736, "grad_norm": 0.9417629837989807, "learning_rate": 0.0005, "loss": 2.1727664470672607, "step": 2123 }, { "epoch": 0.2471491738422155, "grad_norm": 1.5923807621002197, "learning_rate": 0.0005, "loss": 2.8767378330230713, "step": 2124 }, { "epoch": 0.24726553409355365, "grad_norm": 1.2109942436218262, "learning_rate": 0.0005, "loss": 2.937532663345337, "step": 2125 }, { "epoch": 0.24738189434489177, "grad_norm": 3.8141701221466064, "learning_rate": 0.0005, "loss": 1.9437958002090454, "step": 2126 }, { "epoch": 0.24749825459622993, "grad_norm": 1.5298882722854614, "learning_rate": 0.0005, "loss": 3.2672858238220215, "step": 2127 }, { "epoch": 0.24761461484756808, "grad_norm": 2.005837917327881, "learning_rate": 0.0005, "loss": 3.815385103225708, "step": 2128 }, { "epoch": 0.2477309750989062, "grad_norm": 1.699508786201477, "learning_rate": 0.0005, "loss": 2.36061429977417, "step": 2129 }, { "epoch": 0.24784733535024436, "grad_norm": 2.060215950012207, "learning_rate": 0.0005, "loss": 2.6919312477111816, "step": 2130 }, { "epoch": 0.2479636956015825, "grad_norm": 1.5009678602218628, "learning_rate": 0.0005, "loss": 2.8910741806030273, "step": 2131 }, { "epoch": 0.24808005585292064, "grad_norm": 2.031731367111206, "learning_rate": 0.0005, "loss": 2.886824131011963, "step": 2132 }, { "epoch": 0.2481964161042588, "grad_norm": 1.3450860977172852, "learning_rate": 0.0005, "loss": 2.4826085567474365, "step": 2133 }, { "epoch": 0.24831277635559693, "grad_norm": 1.1536837816238403, "learning_rate": 0.0005, "loss": 2.446611166000366, "step": 2134 }, { "epoch": 0.24842913660693508, "grad_norm": 0.9362629055976868, "learning_rate": 0.0005, "loss": 2.4325475692749023, "step": 2135 }, { "epoch": 0.2485454968582732, "grad_norm": 1.4991313219070435, "learning_rate": 0.0005, "loss": 3.666242837905884, "step": 2136 }, { "epoch": 0.24866185710961136, "grad_norm": 1.9971085786819458, "learning_rate": 0.0005, "loss": 2.772329807281494, "step": 2137 }, { "epoch": 0.2487782173609495, "grad_norm": 0.961291491985321, "learning_rate": 0.0005, "loss": 2.7894325256347656, "step": 2138 }, { "epoch": 0.24889457761228764, "grad_norm": 1.2724031209945679, "learning_rate": 0.0005, "loss": 1.9004285335540771, "step": 2139 }, { "epoch": 0.2490109378636258, "grad_norm": 1.7678052186965942, "learning_rate": 0.0005, "loss": 3.1715941429138184, "step": 2140 }, { "epoch": 0.24912729811496392, "grad_norm": 0.9880622029304504, "learning_rate": 0.0005, "loss": 2.8770904541015625, "step": 2141 }, { "epoch": 0.24924365836630208, "grad_norm": 1.7205471992492676, "learning_rate": 0.0005, "loss": 2.9095406532287598, "step": 2142 }, { "epoch": 0.2493600186176402, "grad_norm": 6.2325358390808105, "learning_rate": 0.0005, "loss": 2.206610918045044, "step": 2143 }, { "epoch": 0.24947637886897836, "grad_norm": 1.2832204103469849, "learning_rate": 0.0005, "loss": 2.40303111076355, "step": 2144 }, { "epoch": 0.2495927391203165, "grad_norm": 1.33913254737854, "learning_rate": 0.0005, "loss": 2.8140273094177246, "step": 2145 }, { "epoch": 0.24970909937165464, "grad_norm": 2.0613744258880615, "learning_rate": 0.0005, "loss": 2.5648183822631836, "step": 2146 }, { "epoch": 0.2498254596229928, "grad_norm": 4.07171630859375, "learning_rate": 0.0005, "loss": 2.717742919921875, "step": 2147 }, { "epoch": 0.24994181987433092, "grad_norm": 1.8263185024261475, "learning_rate": 0.0005, "loss": 2.6709718704223633, "step": 2148 }, { "epoch": 0.25005818012566905, "grad_norm": 1.9293456077575684, "learning_rate": 0.0005, "loss": 2.657449960708618, "step": 2149 }, { "epoch": 0.2501745403770072, "grad_norm": 1.4114089012145996, "learning_rate": 0.0005, "loss": 2.2342991828918457, "step": 2150 }, { "epoch": 0.25029090062834536, "grad_norm": 1.574230670928955, "learning_rate": 0.0005, "loss": 2.937946081161499, "step": 2151 }, { "epoch": 0.2504072608796835, "grad_norm": 1.1408923864364624, "learning_rate": 0.0005, "loss": 1.5038785934448242, "step": 2152 }, { "epoch": 0.25052362113102167, "grad_norm": 1.1989797353744507, "learning_rate": 0.0005, "loss": 2.6589274406433105, "step": 2153 }, { "epoch": 0.25063998138235977, "grad_norm": 1.9136905670166016, "learning_rate": 0.0005, "loss": 3.4918386936187744, "step": 2154 }, { "epoch": 0.2507563416336979, "grad_norm": 1.6882892847061157, "learning_rate": 0.0005, "loss": 2.6009159088134766, "step": 2155 }, { "epoch": 0.2508727018850361, "grad_norm": 1.7210443019866943, "learning_rate": 0.0005, "loss": 3.0088276863098145, "step": 2156 }, { "epoch": 0.25098906213637423, "grad_norm": 1.3189992904663086, "learning_rate": 0.0005, "loss": 3.3152883052825928, "step": 2157 }, { "epoch": 0.2511054223877124, "grad_norm": 1.0968436002731323, "learning_rate": 0.0005, "loss": 1.8610306978225708, "step": 2158 }, { "epoch": 0.2512217826390505, "grad_norm": 1.1963720321655273, "learning_rate": 0.0005, "loss": 2.8454880714416504, "step": 2159 }, { "epoch": 0.25133814289038864, "grad_norm": 1.6971409320831299, "learning_rate": 0.0005, "loss": 3.4606385231018066, "step": 2160 }, { "epoch": 0.2514545031417268, "grad_norm": 2.2361066341400146, "learning_rate": 0.0005, "loss": 1.9998196363449097, "step": 2161 }, { "epoch": 0.25157086339306495, "grad_norm": 1.4471428394317627, "learning_rate": 0.0005, "loss": 3.0970616340637207, "step": 2162 }, { "epoch": 0.25168722364440305, "grad_norm": 1.1804625988006592, "learning_rate": 0.0005, "loss": 2.1578891277313232, "step": 2163 }, { "epoch": 0.2518035838957412, "grad_norm": 1.225951075553894, "learning_rate": 0.0005, "loss": 1.6714935302734375, "step": 2164 }, { "epoch": 0.25191994414707936, "grad_norm": 1.214389681816101, "learning_rate": 0.0005, "loss": 1.997008204460144, "step": 2165 }, { "epoch": 0.2520363043984175, "grad_norm": 1.5054218769073486, "learning_rate": 0.0005, "loss": 1.7113761901855469, "step": 2166 }, { "epoch": 0.25215266464975566, "grad_norm": 2.064084768295288, "learning_rate": 0.0005, "loss": 2.9902822971343994, "step": 2167 }, { "epoch": 0.25226902490109376, "grad_norm": 1.7990894317626953, "learning_rate": 0.0005, "loss": 3.601569652557373, "step": 2168 }, { "epoch": 0.2523853851524319, "grad_norm": 1.6451548337936401, "learning_rate": 0.0005, "loss": 2.627444267272949, "step": 2169 }, { "epoch": 0.2525017454037701, "grad_norm": 1.4669774770736694, "learning_rate": 0.0005, "loss": 2.8141744136810303, "step": 2170 }, { "epoch": 0.2526181056551082, "grad_norm": 1.0086795091629028, "learning_rate": 0.0005, "loss": 1.7088446617126465, "step": 2171 }, { "epoch": 0.2527344659064464, "grad_norm": 1.295405626296997, "learning_rate": 0.0005, "loss": 2.0579309463500977, "step": 2172 }, { "epoch": 0.2528508261577845, "grad_norm": 0.7545479536056519, "learning_rate": 0.0005, "loss": 1.6116490364074707, "step": 2173 }, { "epoch": 0.25296718640912264, "grad_norm": 1.6981996297836304, "learning_rate": 0.0005, "loss": 3.4763059616088867, "step": 2174 }, { "epoch": 0.2530835466604608, "grad_norm": 1.1721436977386475, "learning_rate": 0.0005, "loss": 2.1977574825286865, "step": 2175 }, { "epoch": 0.25319990691179894, "grad_norm": 1.3661613464355469, "learning_rate": 0.0005, "loss": 2.3354759216308594, "step": 2176 }, { "epoch": 0.2533162671631371, "grad_norm": 1.4790822267532349, "learning_rate": 0.0005, "loss": 2.2930643558502197, "step": 2177 }, { "epoch": 0.2534326274144752, "grad_norm": 1.5073710680007935, "learning_rate": 0.0005, "loss": 2.754939556121826, "step": 2178 }, { "epoch": 0.25354898766581335, "grad_norm": 1.6988463401794434, "learning_rate": 0.0005, "loss": 2.7624478340148926, "step": 2179 }, { "epoch": 0.2536653479171515, "grad_norm": 1.576993465423584, "learning_rate": 0.0005, "loss": 3.6553843021392822, "step": 2180 }, { "epoch": 0.25378170816848966, "grad_norm": 1.2421926259994507, "learning_rate": 0.0005, "loss": 1.6035676002502441, "step": 2181 }, { "epoch": 0.25389806841982776, "grad_norm": 2.364396572113037, "learning_rate": 0.0005, "loss": 4.127711296081543, "step": 2182 }, { "epoch": 0.2540144286711659, "grad_norm": 1.5517669916152954, "learning_rate": 0.0005, "loss": 2.624337673187256, "step": 2183 }, { "epoch": 0.25413078892250407, "grad_norm": 2.4032645225524902, "learning_rate": 0.0005, "loss": 3.361825942993164, "step": 2184 }, { "epoch": 0.2542471491738422, "grad_norm": 2.1858887672424316, "learning_rate": 0.0005, "loss": 3.6529061794281006, "step": 2185 }, { "epoch": 0.2543635094251804, "grad_norm": 1.4026062488555908, "learning_rate": 0.0005, "loss": 2.771111488342285, "step": 2186 }, { "epoch": 0.2544798696765185, "grad_norm": 1.4449752569198608, "learning_rate": 0.0005, "loss": 3.224984645843506, "step": 2187 }, { "epoch": 0.25459622992785663, "grad_norm": 1.1041408777236938, "learning_rate": 0.0005, "loss": 1.767478108406067, "step": 2188 }, { "epoch": 0.2547125901791948, "grad_norm": 1.0958892107009888, "learning_rate": 0.0005, "loss": 2.643385171890259, "step": 2189 }, { "epoch": 0.25482895043053294, "grad_norm": 2.305292844772339, "learning_rate": 0.0005, "loss": 3.069176435470581, "step": 2190 }, { "epoch": 0.2549453106818711, "grad_norm": 1.3163992166519165, "learning_rate": 0.0005, "loss": 2.8810789585113525, "step": 2191 }, { "epoch": 0.2550616709332092, "grad_norm": 2.1279428005218506, "learning_rate": 0.0005, "loss": 2.8745477199554443, "step": 2192 }, { "epoch": 0.25517803118454735, "grad_norm": 1.217730164527893, "learning_rate": 0.0005, "loss": 4.153157711029053, "step": 2193 }, { "epoch": 0.2552943914358855, "grad_norm": 0.8701626658439636, "learning_rate": 0.0005, "loss": 2.269773483276367, "step": 2194 }, { "epoch": 0.25541075168722366, "grad_norm": 1.6205908060073853, "learning_rate": 0.0005, "loss": 2.741386890411377, "step": 2195 }, { "epoch": 0.2555271119385618, "grad_norm": 1.2644745111465454, "learning_rate": 0.0005, "loss": 3.0928244590759277, "step": 2196 }, { "epoch": 0.2556434721898999, "grad_norm": 1.0826194286346436, "learning_rate": 0.0005, "loss": 1.8008732795715332, "step": 2197 }, { "epoch": 0.25575983244123807, "grad_norm": 1.115295648574829, "learning_rate": 0.0005, "loss": 2.7416679859161377, "step": 2198 }, { "epoch": 0.2558761926925762, "grad_norm": 2.2582974433898926, "learning_rate": 0.0005, "loss": 4.041973114013672, "step": 2199 }, { "epoch": 0.2559925529439144, "grad_norm": 6.185363292694092, "learning_rate": 0.0005, "loss": 2.571857213973999, "step": 2200 }, { "epoch": 0.2561089131952525, "grad_norm": 1.7113964557647705, "learning_rate": 0.0005, "loss": 2.8243601322174072, "step": 2201 }, { "epoch": 0.25622527344659063, "grad_norm": 1.544701337814331, "learning_rate": 0.0005, "loss": 2.2163124084472656, "step": 2202 }, { "epoch": 0.2563416336979288, "grad_norm": 1.153214693069458, "learning_rate": 0.0005, "loss": 2.597377061843872, "step": 2203 }, { "epoch": 0.25645799394926694, "grad_norm": 1.6176050901412964, "learning_rate": 0.0005, "loss": 3.297924280166626, "step": 2204 }, { "epoch": 0.2565743542006051, "grad_norm": 1.9178531169891357, "learning_rate": 0.0005, "loss": 2.8155322074890137, "step": 2205 }, { "epoch": 0.2566907144519432, "grad_norm": 0.9451261758804321, "learning_rate": 0.0005, "loss": 2.4330976009368896, "step": 2206 }, { "epoch": 0.25680707470328135, "grad_norm": 1.2226014137268066, "learning_rate": 0.0005, "loss": 3.3329808712005615, "step": 2207 }, { "epoch": 0.2569234349546195, "grad_norm": 1.3697822093963623, "learning_rate": 0.0005, "loss": 3.580698251724243, "step": 2208 }, { "epoch": 0.25703979520595766, "grad_norm": 2.4249613285064697, "learning_rate": 0.0005, "loss": 3.3706588745117188, "step": 2209 }, { "epoch": 0.2571561554572958, "grad_norm": 1.1686193943023682, "learning_rate": 0.0005, "loss": 2.393584728240967, "step": 2210 }, { "epoch": 0.2572725157086339, "grad_norm": 1.3727835416793823, "learning_rate": 0.0005, "loss": 3.5356528759002686, "step": 2211 }, { "epoch": 0.25738887595997206, "grad_norm": 0.7075715065002441, "learning_rate": 0.0005, "loss": 1.3301877975463867, "step": 2212 }, { "epoch": 0.2575052362113102, "grad_norm": 1.4742674827575684, "learning_rate": 0.0005, "loss": 3.3933444023132324, "step": 2213 }, { "epoch": 0.2576215964626484, "grad_norm": 1.030596375465393, "learning_rate": 0.0005, "loss": 1.406200885772705, "step": 2214 }, { "epoch": 0.2577379567139865, "grad_norm": 1.3205608129501343, "learning_rate": 0.0005, "loss": 2.2637782096862793, "step": 2215 }, { "epoch": 0.2578543169653246, "grad_norm": 1.023240089416504, "learning_rate": 0.0005, "loss": 2.5101027488708496, "step": 2216 }, { "epoch": 0.2579706772166628, "grad_norm": 1.7749645709991455, "learning_rate": 0.0005, "loss": 3.6373395919799805, "step": 2217 }, { "epoch": 0.25808703746800093, "grad_norm": 1.0497139692306519, "learning_rate": 0.0005, "loss": 2.0535130500793457, "step": 2218 }, { "epoch": 0.2582033977193391, "grad_norm": 1.5768603086471558, "learning_rate": 0.0005, "loss": 3.557297945022583, "step": 2219 }, { "epoch": 0.25831975797067724, "grad_norm": 1.30990731716156, "learning_rate": 0.0005, "loss": 2.8038039207458496, "step": 2220 }, { "epoch": 0.25843611822201534, "grad_norm": 0.9034984707832336, "learning_rate": 0.0005, "loss": 1.31401789188385, "step": 2221 }, { "epoch": 0.2585524784733535, "grad_norm": 0.9242690801620483, "learning_rate": 0.0005, "loss": 1.8637890815734863, "step": 2222 }, { "epoch": 0.25866883872469165, "grad_norm": 2.1080281734466553, "learning_rate": 0.0005, "loss": 2.6427040100097656, "step": 2223 }, { "epoch": 0.2587851989760298, "grad_norm": 1.2570947408676147, "learning_rate": 0.0005, "loss": 3.2344415187835693, "step": 2224 }, { "epoch": 0.2589015592273679, "grad_norm": 0.9660317301750183, "learning_rate": 0.0005, "loss": 1.946629285812378, "step": 2225 }, { "epoch": 0.25901791947870606, "grad_norm": 2.3558406829833984, "learning_rate": 0.0005, "loss": 2.7554941177368164, "step": 2226 }, { "epoch": 0.2591342797300442, "grad_norm": 1.3186414241790771, "learning_rate": 0.0005, "loss": 2.9396700859069824, "step": 2227 }, { "epoch": 0.25925063998138237, "grad_norm": 1.9196650981903076, "learning_rate": 0.0005, "loss": 2.427178144454956, "step": 2228 }, { "epoch": 0.2593670002327205, "grad_norm": 1.0641847848892212, "learning_rate": 0.0005, "loss": 2.4706344604492188, "step": 2229 }, { "epoch": 0.2594833604840586, "grad_norm": 1.3208322525024414, "learning_rate": 0.0005, "loss": 2.20458984375, "step": 2230 }, { "epoch": 0.2595997207353968, "grad_norm": 1.5761752128601074, "learning_rate": 0.0005, "loss": 3.096165657043457, "step": 2231 }, { "epoch": 0.25971608098673493, "grad_norm": 1.2709949016571045, "learning_rate": 0.0005, "loss": 2.4944958686828613, "step": 2232 }, { "epoch": 0.2598324412380731, "grad_norm": 1.5620756149291992, "learning_rate": 0.0005, "loss": 2.2586867809295654, "step": 2233 }, { "epoch": 0.25994880148941124, "grad_norm": 1.209309697151184, "learning_rate": 0.0005, "loss": 2.082977056503296, "step": 2234 }, { "epoch": 0.26006516174074934, "grad_norm": 2.7942981719970703, "learning_rate": 0.0005, "loss": 3.1208910942077637, "step": 2235 }, { "epoch": 0.2601815219920875, "grad_norm": 0.9608127474784851, "learning_rate": 0.0005, "loss": 2.568215847015381, "step": 2236 }, { "epoch": 0.26029788224342565, "grad_norm": 2.9614343643188477, "learning_rate": 0.0005, "loss": 2.818135976791382, "step": 2237 }, { "epoch": 0.2604142424947638, "grad_norm": 2.3632936477661133, "learning_rate": 0.0005, "loss": 2.9353837966918945, "step": 2238 }, { "epoch": 0.26053060274610196, "grad_norm": 1.307018518447876, "learning_rate": 0.0005, "loss": 2.836360454559326, "step": 2239 }, { "epoch": 0.26064696299744006, "grad_norm": 6.1915082931518555, "learning_rate": 0.0005, "loss": 1.8238506317138672, "step": 2240 }, { "epoch": 0.2607633232487782, "grad_norm": 1.032593011856079, "learning_rate": 0.0005, "loss": 2.414036512374878, "step": 2241 }, { "epoch": 0.26087968350011637, "grad_norm": 2.2977051734924316, "learning_rate": 0.0005, "loss": 2.7538986206054688, "step": 2242 }, { "epoch": 0.2609960437514545, "grad_norm": 1.4614907503128052, "learning_rate": 0.0005, "loss": 2.8736042976379395, "step": 2243 }, { "epoch": 0.2611124040027926, "grad_norm": 1.5539355278015137, "learning_rate": 0.0005, "loss": 3.4405298233032227, "step": 2244 }, { "epoch": 0.2612287642541308, "grad_norm": 1.7117472887039185, "learning_rate": 0.0005, "loss": 2.6517391204833984, "step": 2245 }, { "epoch": 0.26134512450546893, "grad_norm": 1.0968302488327026, "learning_rate": 0.0005, "loss": 2.2154297828674316, "step": 2246 }, { "epoch": 0.2614614847568071, "grad_norm": 1.805533528327942, "learning_rate": 0.0005, "loss": 2.4481935501098633, "step": 2247 }, { "epoch": 0.26157784500814524, "grad_norm": 1.4077866077423096, "learning_rate": 0.0005, "loss": 2.3907694816589355, "step": 2248 }, { "epoch": 0.26169420525948334, "grad_norm": 1.6360067129135132, "learning_rate": 0.0005, "loss": 1.7993937730789185, "step": 2249 }, { "epoch": 0.2618105655108215, "grad_norm": 1.9274269342422485, "learning_rate": 0.0005, "loss": 3.5332727432250977, "step": 2250 }, { "epoch": 0.26192692576215965, "grad_norm": 1.8422876596450806, "learning_rate": 0.0005, "loss": 3.065063953399658, "step": 2251 }, { "epoch": 0.2620432860134978, "grad_norm": 1.3616223335266113, "learning_rate": 0.0005, "loss": 3.1809937953948975, "step": 2252 }, { "epoch": 0.26215964626483595, "grad_norm": 0.5567875504493713, "learning_rate": 0.0005, "loss": 1.2368584871292114, "step": 2253 }, { "epoch": 0.26227600651617405, "grad_norm": 1.4063295125961304, "learning_rate": 0.0005, "loss": 3.197387456893921, "step": 2254 }, { "epoch": 0.2623923667675122, "grad_norm": 0.9255122542381287, "learning_rate": 0.0005, "loss": 1.8653100728988647, "step": 2255 }, { "epoch": 0.26250872701885036, "grad_norm": 1.363680124282837, "learning_rate": 0.0005, "loss": 2.3528614044189453, "step": 2256 }, { "epoch": 0.2626250872701885, "grad_norm": 2.7286312580108643, "learning_rate": 0.0005, "loss": 3.1632750034332275, "step": 2257 }, { "epoch": 0.26274144752152667, "grad_norm": 1.2888352870941162, "learning_rate": 0.0005, "loss": 2.7973549365997314, "step": 2258 }, { "epoch": 0.26285780777286477, "grad_norm": 1.72642183303833, "learning_rate": 0.0005, "loss": 3.8713064193725586, "step": 2259 }, { "epoch": 0.2629741680242029, "grad_norm": 1.070989966392517, "learning_rate": 0.0005, "loss": 1.865256428718567, "step": 2260 }, { "epoch": 0.2630905282755411, "grad_norm": 1.3640937805175781, "learning_rate": 0.0005, "loss": 1.9642022848129272, "step": 2261 }, { "epoch": 0.26320688852687923, "grad_norm": 2.1354517936706543, "learning_rate": 0.0005, "loss": 3.1463396549224854, "step": 2262 }, { "epoch": 0.26332324877821733, "grad_norm": 2.043248176574707, "learning_rate": 0.0005, "loss": 3.1307878494262695, "step": 2263 }, { "epoch": 0.2634396090295555, "grad_norm": 1.9859758615493774, "learning_rate": 0.0005, "loss": 2.428422451019287, "step": 2264 }, { "epoch": 0.26355596928089364, "grad_norm": 1.4534987211227417, "learning_rate": 0.0005, "loss": 2.353203535079956, "step": 2265 }, { "epoch": 0.2636723295322318, "grad_norm": 2.0681514739990234, "learning_rate": 0.0005, "loss": 3.499990701675415, "step": 2266 }, { "epoch": 0.26378868978356995, "grad_norm": 1.3867833614349365, "learning_rate": 0.0005, "loss": 2.3377184867858887, "step": 2267 }, { "epoch": 0.26390505003490805, "grad_norm": 2.5064103603363037, "learning_rate": 0.0005, "loss": 2.9297611713409424, "step": 2268 }, { "epoch": 0.2640214102862462, "grad_norm": 1.3860751390457153, "learning_rate": 0.0005, "loss": 2.5749826431274414, "step": 2269 }, { "epoch": 0.26413777053758436, "grad_norm": 1.3660567998886108, "learning_rate": 0.0005, "loss": 2.995704174041748, "step": 2270 }, { "epoch": 0.2642541307889225, "grad_norm": 1.666011095046997, "learning_rate": 0.0005, "loss": 3.133185386657715, "step": 2271 }, { "epoch": 0.26437049104026067, "grad_norm": 1.7205379009246826, "learning_rate": 0.0005, "loss": 3.1294455528259277, "step": 2272 }, { "epoch": 0.26448685129159877, "grad_norm": 1.2746440172195435, "learning_rate": 0.0005, "loss": 2.8068065643310547, "step": 2273 }, { "epoch": 0.2646032115429369, "grad_norm": 1.1170271635055542, "learning_rate": 0.0005, "loss": 2.337553024291992, "step": 2274 }, { "epoch": 0.2647195717942751, "grad_norm": 1.7196991443634033, "learning_rate": 0.0005, "loss": 3.079047203063965, "step": 2275 }, { "epoch": 0.26483593204561323, "grad_norm": 1.7608554363250732, "learning_rate": 0.0005, "loss": 2.6888227462768555, "step": 2276 }, { "epoch": 0.2649522922969514, "grad_norm": 1.672956943511963, "learning_rate": 0.0005, "loss": 2.893537998199463, "step": 2277 }, { "epoch": 0.2650686525482895, "grad_norm": 1.2046961784362793, "learning_rate": 0.0005, "loss": 2.4167423248291016, "step": 2278 }, { "epoch": 0.26518501279962764, "grad_norm": 1.6891365051269531, "learning_rate": 0.0005, "loss": 3.0620861053466797, "step": 2279 }, { "epoch": 0.2653013730509658, "grad_norm": 1.8527193069458008, "learning_rate": 0.0005, "loss": 3.1655166149139404, "step": 2280 }, { "epoch": 0.26541773330230395, "grad_norm": 1.1756476163864136, "learning_rate": 0.0005, "loss": 3.008115768432617, "step": 2281 }, { "epoch": 0.2655340935536421, "grad_norm": 1.3614046573638916, "learning_rate": 0.0005, "loss": 3.181379795074463, "step": 2282 }, { "epoch": 0.2656504538049802, "grad_norm": 1.0315521955490112, "learning_rate": 0.0005, "loss": 2.127427577972412, "step": 2283 }, { "epoch": 0.26576681405631836, "grad_norm": 1.8387107849121094, "learning_rate": 0.0005, "loss": 2.731966733932495, "step": 2284 }, { "epoch": 0.2658831743076565, "grad_norm": 1.7757594585418701, "learning_rate": 0.0005, "loss": 2.508212089538574, "step": 2285 }, { "epoch": 0.26599953455899467, "grad_norm": 1.5802122354507446, "learning_rate": 0.0005, "loss": 2.7584707736968994, "step": 2286 }, { "epoch": 0.26611589481033276, "grad_norm": 1.6123102903366089, "learning_rate": 0.0005, "loss": 3.5356104373931885, "step": 2287 }, { "epoch": 0.2662322550616709, "grad_norm": 0.9850979447364807, "learning_rate": 0.0005, "loss": 2.417801856994629, "step": 2288 }, { "epoch": 0.2663486153130091, "grad_norm": 1.262558937072754, "learning_rate": 0.0005, "loss": 1.8131434917449951, "step": 2289 }, { "epoch": 0.26646497556434723, "grad_norm": 1.9280685186386108, "learning_rate": 0.0005, "loss": 3.265929698944092, "step": 2290 }, { "epoch": 0.2665813358156854, "grad_norm": 2.865764617919922, "learning_rate": 0.0005, "loss": 2.7157225608825684, "step": 2291 }, { "epoch": 0.2666976960670235, "grad_norm": 1.3160172700881958, "learning_rate": 0.0005, "loss": 2.7282302379608154, "step": 2292 }, { "epoch": 0.26681405631836164, "grad_norm": 1.439740777015686, "learning_rate": 0.0005, "loss": 3.363124370574951, "step": 2293 }, { "epoch": 0.2669304165696998, "grad_norm": 1.252800703048706, "learning_rate": 0.0005, "loss": 1.7166250944137573, "step": 2294 }, { "epoch": 0.26704677682103795, "grad_norm": 1.2298887968063354, "learning_rate": 0.0005, "loss": 3.030384063720703, "step": 2295 }, { "epoch": 0.2671631370723761, "grad_norm": 1.8280768394470215, "learning_rate": 0.0005, "loss": 2.1882927417755127, "step": 2296 }, { "epoch": 0.2672794973237142, "grad_norm": 2.1266069412231445, "learning_rate": 0.0005, "loss": 3.564023017883301, "step": 2297 }, { "epoch": 0.26739585757505235, "grad_norm": 1.2622485160827637, "learning_rate": 0.0005, "loss": 2.4557852745056152, "step": 2298 }, { "epoch": 0.2675122178263905, "grad_norm": 1.6099462509155273, "learning_rate": 0.0005, "loss": 3.34446120262146, "step": 2299 }, { "epoch": 0.26762857807772866, "grad_norm": 1.2090251445770264, "learning_rate": 0.0005, "loss": 2.534593343734741, "step": 2300 }, { "epoch": 0.2677449383290668, "grad_norm": 0.8156647086143494, "learning_rate": 0.0005, "loss": 1.8928722143173218, "step": 2301 }, { "epoch": 0.2678612985804049, "grad_norm": 1.2040759325027466, "learning_rate": 0.0005, "loss": 2.3425142765045166, "step": 2302 }, { "epoch": 0.26797765883174307, "grad_norm": 1.486663579940796, "learning_rate": 0.0005, "loss": 2.3727047443389893, "step": 2303 }, { "epoch": 0.2680940190830812, "grad_norm": 1.2319350242614746, "learning_rate": 0.0005, "loss": 1.8491780757904053, "step": 2304 }, { "epoch": 0.2682103793344194, "grad_norm": 1.989894151687622, "learning_rate": 0.0005, "loss": 3.3948473930358887, "step": 2305 }, { "epoch": 0.2683267395857575, "grad_norm": 1.4446219205856323, "learning_rate": 0.0005, "loss": 2.480699062347412, "step": 2306 }, { "epoch": 0.26844309983709563, "grad_norm": 3.262650489807129, "learning_rate": 0.0005, "loss": 3.9739737510681152, "step": 2307 }, { "epoch": 0.2685594600884338, "grad_norm": 2.2766966819763184, "learning_rate": 0.0005, "loss": 3.155083417892456, "step": 2308 }, { "epoch": 0.26867582033977194, "grad_norm": 2.7468831539154053, "learning_rate": 0.0005, "loss": 3.2276053428649902, "step": 2309 }, { "epoch": 0.2687921805911101, "grad_norm": 1.5636627674102783, "learning_rate": 0.0005, "loss": 2.350529193878174, "step": 2310 }, { "epoch": 0.2689085408424482, "grad_norm": 1.2976549863815308, "learning_rate": 0.0005, "loss": 2.712686538696289, "step": 2311 }, { "epoch": 0.26902490109378635, "grad_norm": 1.362465739250183, "learning_rate": 0.0005, "loss": 2.3187832832336426, "step": 2312 }, { "epoch": 0.2691412613451245, "grad_norm": 1.0866199731826782, "learning_rate": 0.0005, "loss": 1.7341907024383545, "step": 2313 }, { "epoch": 0.26925762159646266, "grad_norm": 1.5962648391723633, "learning_rate": 0.0005, "loss": 3.3290393352508545, "step": 2314 }, { "epoch": 0.2693739818478008, "grad_norm": 1.8659870624542236, "learning_rate": 0.0005, "loss": 2.9215140342712402, "step": 2315 }, { "epoch": 0.2694903420991389, "grad_norm": 1.2047208547592163, "learning_rate": 0.0005, "loss": 2.11991548538208, "step": 2316 }, { "epoch": 0.26960670235047707, "grad_norm": 1.422917127609253, "learning_rate": 0.0005, "loss": 2.147891044616699, "step": 2317 }, { "epoch": 0.2697230626018152, "grad_norm": 0.7762160301208496, "learning_rate": 0.0005, "loss": 1.8990976810455322, "step": 2318 }, { "epoch": 0.2698394228531534, "grad_norm": 2.843705654144287, "learning_rate": 0.0005, "loss": 2.496994972229004, "step": 2319 }, { "epoch": 0.26995578310449153, "grad_norm": 1.3083949089050293, "learning_rate": 0.0005, "loss": 2.6174449920654297, "step": 2320 }, { "epoch": 0.27007214335582963, "grad_norm": 1.551568865776062, "learning_rate": 0.0005, "loss": 2.6062517166137695, "step": 2321 }, { "epoch": 0.2701885036071678, "grad_norm": 1.297574758529663, "learning_rate": 0.0005, "loss": 2.5186686515808105, "step": 2322 }, { "epoch": 0.27030486385850594, "grad_norm": 1.4705876111984253, "learning_rate": 0.0005, "loss": 3.3869614601135254, "step": 2323 }, { "epoch": 0.2704212241098441, "grad_norm": 1.3503700494766235, "learning_rate": 0.0005, "loss": 3.1638216972351074, "step": 2324 }, { "epoch": 0.2705375843611822, "grad_norm": 1.129490613937378, "learning_rate": 0.0005, "loss": 2.313063621520996, "step": 2325 }, { "epoch": 0.27065394461252035, "grad_norm": 1.793400764465332, "learning_rate": 0.0005, "loss": 3.539377212524414, "step": 2326 }, { "epoch": 0.2707703048638585, "grad_norm": 2.109463691711426, "learning_rate": 0.0005, "loss": 3.1947789192199707, "step": 2327 }, { "epoch": 0.27088666511519666, "grad_norm": 1.668148398399353, "learning_rate": 0.0005, "loss": 2.995589256286621, "step": 2328 }, { "epoch": 0.2710030253665348, "grad_norm": 1.7280677556991577, "learning_rate": 0.0005, "loss": 2.780503749847412, "step": 2329 }, { "epoch": 0.2711193856178729, "grad_norm": 1.4188508987426758, "learning_rate": 0.0005, "loss": 2.4868550300598145, "step": 2330 }, { "epoch": 0.27123574586921106, "grad_norm": 2.5874719619750977, "learning_rate": 0.0005, "loss": 2.675096273422241, "step": 2331 }, { "epoch": 0.2713521061205492, "grad_norm": 2.003934621810913, "learning_rate": 0.0005, "loss": 3.5251991748809814, "step": 2332 }, { "epoch": 0.2714684663718874, "grad_norm": 1.4105827808380127, "learning_rate": 0.0005, "loss": 3.311394453048706, "step": 2333 }, { "epoch": 0.27158482662322553, "grad_norm": 3.680148124694824, "learning_rate": 0.0005, "loss": 2.095341205596924, "step": 2334 }, { "epoch": 0.2717011868745636, "grad_norm": 1.6284512281417847, "learning_rate": 0.0005, "loss": 2.818167209625244, "step": 2335 }, { "epoch": 0.2718175471259018, "grad_norm": 2.0093257427215576, "learning_rate": 0.0005, "loss": 2.345085620880127, "step": 2336 }, { "epoch": 0.27193390737723994, "grad_norm": 1.5369048118591309, "learning_rate": 0.0005, "loss": 2.414436101913452, "step": 2337 }, { "epoch": 0.2720502676285781, "grad_norm": 1.9060949087142944, "learning_rate": 0.0005, "loss": 3.025021553039551, "step": 2338 }, { "epoch": 0.27216662787991625, "grad_norm": 2.028846502304077, "learning_rate": 0.0005, "loss": 2.936556816101074, "step": 2339 }, { "epoch": 0.27228298813125434, "grad_norm": 1.122997522354126, "learning_rate": 0.0005, "loss": 1.9879724979400635, "step": 2340 }, { "epoch": 0.2723993483825925, "grad_norm": 1.552412986755371, "learning_rate": 0.0005, "loss": 2.803253412246704, "step": 2341 }, { "epoch": 0.27251570863393065, "grad_norm": 2.0010786056518555, "learning_rate": 0.0005, "loss": 2.3969082832336426, "step": 2342 }, { "epoch": 0.2726320688852688, "grad_norm": 1.4600192308425903, "learning_rate": 0.0005, "loss": 2.5029141902923584, "step": 2343 }, { "epoch": 0.27274842913660696, "grad_norm": 2.4274961948394775, "learning_rate": 0.0005, "loss": 2.786479949951172, "step": 2344 }, { "epoch": 0.27286478938794506, "grad_norm": 1.4402852058410645, "learning_rate": 0.0005, "loss": 1.9102654457092285, "step": 2345 }, { "epoch": 0.2729811496392832, "grad_norm": 1.641672134399414, "learning_rate": 0.0005, "loss": 2.684943675994873, "step": 2346 }, { "epoch": 0.27309750989062137, "grad_norm": 1.967423439025879, "learning_rate": 0.0005, "loss": 3.4378414154052734, "step": 2347 }, { "epoch": 0.2732138701419595, "grad_norm": 1.8476804494857788, "learning_rate": 0.0005, "loss": 3.5742921829223633, "step": 2348 }, { "epoch": 0.2733302303932976, "grad_norm": 1.6979557275772095, "learning_rate": 0.0005, "loss": 3.577089786529541, "step": 2349 }, { "epoch": 0.2734465906446358, "grad_norm": 1.6919467449188232, "learning_rate": 0.0005, "loss": 2.402505874633789, "step": 2350 }, { "epoch": 0.27356295089597393, "grad_norm": 0.9580860733985901, "learning_rate": 0.0005, "loss": 2.4863486289978027, "step": 2351 }, { "epoch": 0.2736793111473121, "grad_norm": 1.9464561939239502, "learning_rate": 0.0005, "loss": 3.4161336421966553, "step": 2352 }, { "epoch": 0.27379567139865024, "grad_norm": 1.664357304573059, "learning_rate": 0.0005, "loss": 2.6866941452026367, "step": 2353 }, { "epoch": 0.27391203164998834, "grad_norm": 1.3955535888671875, "learning_rate": 0.0005, "loss": 2.720895767211914, "step": 2354 }, { "epoch": 0.2740283919013265, "grad_norm": 0.7550809979438782, "learning_rate": 0.0005, "loss": 1.610878586769104, "step": 2355 }, { "epoch": 0.27414475215266465, "grad_norm": 1.3940068483352661, "learning_rate": 0.0005, "loss": 2.367497205734253, "step": 2356 }, { "epoch": 0.2742611124040028, "grad_norm": 1.0940685272216797, "learning_rate": 0.0005, "loss": 1.9999922513961792, "step": 2357 }, { "epoch": 0.27437747265534096, "grad_norm": 1.3337523937225342, "learning_rate": 0.0005, "loss": 1.9550998210906982, "step": 2358 }, { "epoch": 0.27449383290667906, "grad_norm": 1.3600640296936035, "learning_rate": 0.0005, "loss": 2.501619815826416, "step": 2359 }, { "epoch": 0.2746101931580172, "grad_norm": 1.1857712268829346, "learning_rate": 0.0005, "loss": 3.051889419555664, "step": 2360 }, { "epoch": 0.27472655340935537, "grad_norm": 1.0503605604171753, "learning_rate": 0.0005, "loss": 1.8696305751800537, "step": 2361 }, { "epoch": 0.2748429136606935, "grad_norm": 0.963606595993042, "learning_rate": 0.0005, "loss": 2.2179999351501465, "step": 2362 }, { "epoch": 0.2749592739120317, "grad_norm": 1.11416757106781, "learning_rate": 0.0005, "loss": 2.8604280948638916, "step": 2363 }, { "epoch": 0.2750756341633698, "grad_norm": 2.9421780109405518, "learning_rate": 0.0005, "loss": 1.9808225631713867, "step": 2364 }, { "epoch": 0.27519199441470793, "grad_norm": 3.241112232208252, "learning_rate": 0.0005, "loss": 2.0402274131774902, "step": 2365 }, { "epoch": 0.2753083546660461, "grad_norm": 1.812779426574707, "learning_rate": 0.0005, "loss": 3.7871532440185547, "step": 2366 }, { "epoch": 0.27542471491738424, "grad_norm": 1.354154348373413, "learning_rate": 0.0005, "loss": 3.5417261123657227, "step": 2367 }, { "epoch": 0.27554107516872234, "grad_norm": 2.092130422592163, "learning_rate": 0.0005, "loss": 2.542367935180664, "step": 2368 }, { "epoch": 0.2756574354200605, "grad_norm": 1.7333362102508545, "learning_rate": 0.0005, "loss": 2.5952725410461426, "step": 2369 }, { "epoch": 0.27577379567139865, "grad_norm": 5.104503631591797, "learning_rate": 0.0005, "loss": 4.119900703430176, "step": 2370 }, { "epoch": 0.2758901559227368, "grad_norm": 5.936033248901367, "learning_rate": 0.0005, "loss": 3.836961507797241, "step": 2371 }, { "epoch": 0.27600651617407496, "grad_norm": 2.029782772064209, "learning_rate": 0.0005, "loss": 2.1609935760498047, "step": 2372 }, { "epoch": 0.27612287642541306, "grad_norm": 1.8178685903549194, "learning_rate": 0.0005, "loss": 2.992570161819458, "step": 2373 }, { "epoch": 0.2762392366767512, "grad_norm": 1.984687089920044, "learning_rate": 0.0005, "loss": 3.105215072631836, "step": 2374 }, { "epoch": 0.27635559692808936, "grad_norm": 1.1644110679626465, "learning_rate": 0.0005, "loss": 2.727598190307617, "step": 2375 }, { "epoch": 0.2764719571794275, "grad_norm": 1.111915111541748, "learning_rate": 0.0005, "loss": 2.3679280281066895, "step": 2376 }, { "epoch": 0.2765883174307657, "grad_norm": 1.0093337297439575, "learning_rate": 0.0005, "loss": 2.7766549587249756, "step": 2377 }, { "epoch": 0.2767046776821038, "grad_norm": 1.2093461751937866, "learning_rate": 0.0005, "loss": 3.12442684173584, "step": 2378 }, { "epoch": 0.2768210379334419, "grad_norm": 0.850453794002533, "learning_rate": 0.0005, "loss": 1.7424817085266113, "step": 2379 }, { "epoch": 0.2769373981847801, "grad_norm": 2.358706474304199, "learning_rate": 0.0005, "loss": 3.947498083114624, "step": 2380 }, { "epoch": 0.27705375843611824, "grad_norm": 2.1216325759887695, "learning_rate": 0.0005, "loss": 2.1949687004089355, "step": 2381 }, { "epoch": 0.2771701186874564, "grad_norm": 1.2754383087158203, "learning_rate": 0.0005, "loss": 2.251441717147827, "step": 2382 }, { "epoch": 0.2772864789387945, "grad_norm": 1.3789246082305908, "learning_rate": 0.0005, "loss": 3.6014866828918457, "step": 2383 }, { "epoch": 0.27740283919013264, "grad_norm": 1.36449134349823, "learning_rate": 0.0005, "loss": 2.9394354820251465, "step": 2384 }, { "epoch": 0.2775191994414708, "grad_norm": 1.0153462886810303, "learning_rate": 0.0005, "loss": 1.5383481979370117, "step": 2385 }, { "epoch": 0.27763555969280895, "grad_norm": 1.0582243204116821, "learning_rate": 0.0005, "loss": 2.783369779586792, "step": 2386 }, { "epoch": 0.27775191994414705, "grad_norm": 1.567341685295105, "learning_rate": 0.0005, "loss": 2.568563938140869, "step": 2387 }, { "epoch": 0.2778682801954852, "grad_norm": 1.279762864112854, "learning_rate": 0.0005, "loss": 2.327620506286621, "step": 2388 }, { "epoch": 0.27798464044682336, "grad_norm": 1.2387917041778564, "learning_rate": 0.0005, "loss": 2.46622896194458, "step": 2389 }, { "epoch": 0.2781010006981615, "grad_norm": 1.2700644731521606, "learning_rate": 0.0005, "loss": 2.2805328369140625, "step": 2390 }, { "epoch": 0.27821736094949967, "grad_norm": 0.6846035122871399, "learning_rate": 0.0005, "loss": 1.575728178024292, "step": 2391 }, { "epoch": 0.27833372120083777, "grad_norm": 2.475693702697754, "learning_rate": 0.0005, "loss": 3.1706669330596924, "step": 2392 }, { "epoch": 0.2784500814521759, "grad_norm": 1.727538824081421, "learning_rate": 0.0005, "loss": 3.8519458770751953, "step": 2393 }, { "epoch": 0.2785664417035141, "grad_norm": 1.0771523714065552, "learning_rate": 0.0005, "loss": 2.4134092330932617, "step": 2394 }, { "epoch": 0.27868280195485223, "grad_norm": 1.6915178298950195, "learning_rate": 0.0005, "loss": 3.4921395778656006, "step": 2395 }, { "epoch": 0.2787991622061904, "grad_norm": 1.4357593059539795, "learning_rate": 0.0005, "loss": 3.1721243858337402, "step": 2396 }, { "epoch": 0.2789155224575285, "grad_norm": 1.3952869176864624, "learning_rate": 0.0005, "loss": 2.160432815551758, "step": 2397 }, { "epoch": 0.27903188270886664, "grad_norm": 1.176511526107788, "learning_rate": 0.0005, "loss": 2.0073275566101074, "step": 2398 }, { "epoch": 0.2791482429602048, "grad_norm": 1.7176597118377686, "learning_rate": 0.0005, "loss": 2.5214478969573975, "step": 2399 }, { "epoch": 0.27926460321154295, "grad_norm": 1.121134877204895, "learning_rate": 0.0005, "loss": 2.5954883098602295, "step": 2400 }, { "epoch": 0.2793809634628811, "grad_norm": 1.2429463863372803, "learning_rate": 0.0005, "loss": 2.073822021484375, "step": 2401 }, { "epoch": 0.2794973237142192, "grad_norm": 1.061019778251648, "learning_rate": 0.0005, "loss": 2.1417691707611084, "step": 2402 }, { "epoch": 0.27961368396555736, "grad_norm": 1.2311038970947266, "learning_rate": 0.0005, "loss": 2.206151008605957, "step": 2403 }, { "epoch": 0.2797300442168955, "grad_norm": 0.6522687077522278, "learning_rate": 0.0005, "loss": 1.4739010334014893, "step": 2404 }, { "epoch": 0.27984640446823367, "grad_norm": 1.5635076761245728, "learning_rate": 0.0005, "loss": 2.439008951187134, "step": 2405 }, { "epoch": 0.2799627647195718, "grad_norm": 1.4748753309249878, "learning_rate": 0.0005, "loss": 2.886033773422241, "step": 2406 }, { "epoch": 0.2800791249709099, "grad_norm": 1.6384035348892212, "learning_rate": 0.0005, "loss": 2.1270968914031982, "step": 2407 }, { "epoch": 0.2801954852222481, "grad_norm": 2.7767093181610107, "learning_rate": 0.0005, "loss": 2.7331790924072266, "step": 2408 }, { "epoch": 0.28031184547358623, "grad_norm": 1.9978128671646118, "learning_rate": 0.0005, "loss": 3.4547135829925537, "step": 2409 }, { "epoch": 0.2804282057249244, "grad_norm": 1.6905481815338135, "learning_rate": 0.0005, "loss": 3.4339332580566406, "step": 2410 }, { "epoch": 0.2805445659762625, "grad_norm": 1.6521128416061401, "learning_rate": 0.0005, "loss": 3.2180895805358887, "step": 2411 }, { "epoch": 0.28066092622760064, "grad_norm": 2.091822862625122, "learning_rate": 0.0005, "loss": 2.28914475440979, "step": 2412 }, { "epoch": 0.2807772864789388, "grad_norm": 1.003705620765686, "learning_rate": 0.0005, "loss": 1.8211804628372192, "step": 2413 }, { "epoch": 0.28089364673027695, "grad_norm": 1.5207388401031494, "learning_rate": 0.0005, "loss": 3.3904571533203125, "step": 2414 }, { "epoch": 0.2810100069816151, "grad_norm": 1.6382559537887573, "learning_rate": 0.0005, "loss": 3.4160428047180176, "step": 2415 }, { "epoch": 0.2811263672329532, "grad_norm": 1.0836353302001953, "learning_rate": 0.0005, "loss": 2.61165714263916, "step": 2416 }, { "epoch": 0.28124272748429135, "grad_norm": 1.1308635473251343, "learning_rate": 0.0005, "loss": 2.213981866836548, "step": 2417 }, { "epoch": 0.2813590877356295, "grad_norm": 1.2577544450759888, "learning_rate": 0.0005, "loss": 2.4727652072906494, "step": 2418 }, { "epoch": 0.28147544798696766, "grad_norm": 2.0769259929656982, "learning_rate": 0.0005, "loss": 3.960310935974121, "step": 2419 }, { "epoch": 0.2815918082383058, "grad_norm": 1.4697868824005127, "learning_rate": 0.0005, "loss": 3.573946237564087, "step": 2420 }, { "epoch": 0.2817081684896439, "grad_norm": 1.1495205163955688, "learning_rate": 0.0005, "loss": 3.3030004501342773, "step": 2421 }, { "epoch": 0.28182452874098207, "grad_norm": 1.5205227136611938, "learning_rate": 0.0005, "loss": 2.822360038757324, "step": 2422 }, { "epoch": 0.2819408889923202, "grad_norm": 0.8972822427749634, "learning_rate": 0.0005, "loss": 1.6900570392608643, "step": 2423 }, { "epoch": 0.2820572492436584, "grad_norm": 1.3498270511627197, "learning_rate": 0.0005, "loss": 2.6439294815063477, "step": 2424 }, { "epoch": 0.28217360949499654, "grad_norm": 1.412627100944519, "learning_rate": 0.0005, "loss": 2.7492666244506836, "step": 2425 }, { "epoch": 0.28228996974633463, "grad_norm": 1.3352324962615967, "learning_rate": 0.0005, "loss": 2.5228209495544434, "step": 2426 }, { "epoch": 0.2824063299976728, "grad_norm": 1.5460655689239502, "learning_rate": 0.0005, "loss": 2.900681495666504, "step": 2427 }, { "epoch": 0.28252269024901094, "grad_norm": 0.911779522895813, "learning_rate": 0.0005, "loss": 1.398646593093872, "step": 2428 }, { "epoch": 0.2826390505003491, "grad_norm": 3.6651523113250732, "learning_rate": 0.0005, "loss": 3.2218523025512695, "step": 2429 }, { "epoch": 0.2827554107516872, "grad_norm": 1.440851092338562, "learning_rate": 0.0005, "loss": 2.5891964435577393, "step": 2430 }, { "epoch": 0.28287177100302535, "grad_norm": 2.4030356407165527, "learning_rate": 0.0005, "loss": 2.938314914703369, "step": 2431 }, { "epoch": 0.2829881312543635, "grad_norm": 1.8511189222335815, "learning_rate": 0.0005, "loss": 3.0371270179748535, "step": 2432 }, { "epoch": 0.28310449150570166, "grad_norm": 4.741107940673828, "learning_rate": 0.0005, "loss": 1.7189381122589111, "step": 2433 }, { "epoch": 0.2832208517570398, "grad_norm": 1.6781667470932007, "learning_rate": 0.0005, "loss": 3.6804192066192627, "step": 2434 }, { "epoch": 0.2833372120083779, "grad_norm": 1.3910778760910034, "learning_rate": 0.0005, "loss": 3.2753942012786865, "step": 2435 }, { "epoch": 0.28345357225971607, "grad_norm": 4.2115044593811035, "learning_rate": 0.0005, "loss": 2.7252821922302246, "step": 2436 }, { "epoch": 0.2835699325110542, "grad_norm": 1.2471601963043213, "learning_rate": 0.0005, "loss": 1.662340760231018, "step": 2437 }, { "epoch": 0.2836862927623924, "grad_norm": 1.3652156591415405, "learning_rate": 0.0005, "loss": 2.7400412559509277, "step": 2438 }, { "epoch": 0.28380265301373053, "grad_norm": 1.317334771156311, "learning_rate": 0.0005, "loss": 2.2411415576934814, "step": 2439 }, { "epoch": 0.28391901326506863, "grad_norm": 1.6604535579681396, "learning_rate": 0.0005, "loss": 3.2319021224975586, "step": 2440 }, { "epoch": 0.2840353735164068, "grad_norm": 1.5832558870315552, "learning_rate": 0.0005, "loss": 3.279193162918091, "step": 2441 }, { "epoch": 0.28415173376774494, "grad_norm": 0.9973189234733582, "learning_rate": 0.0005, "loss": 1.7270740270614624, "step": 2442 }, { "epoch": 0.2842680940190831, "grad_norm": 1.9050670862197876, "learning_rate": 0.0005, "loss": 2.273967981338501, "step": 2443 }, { "epoch": 0.28438445427042125, "grad_norm": 1.038280963897705, "learning_rate": 0.0005, "loss": 2.122197151184082, "step": 2444 }, { "epoch": 0.28450081452175935, "grad_norm": 0.8672648668289185, "learning_rate": 0.0005, "loss": 1.9910117387771606, "step": 2445 }, { "epoch": 0.2846171747730975, "grad_norm": 1.3335837125778198, "learning_rate": 0.0005, "loss": 2.369875907897949, "step": 2446 }, { "epoch": 0.28473353502443566, "grad_norm": 1.1421165466308594, "learning_rate": 0.0005, "loss": 2.1907413005828857, "step": 2447 }, { "epoch": 0.2848498952757738, "grad_norm": 1.5192935466766357, "learning_rate": 0.0005, "loss": 2.8070294857025146, "step": 2448 }, { "epoch": 0.2849662555271119, "grad_norm": 1.622707486152649, "learning_rate": 0.0005, "loss": 3.1650304794311523, "step": 2449 }, { "epoch": 0.28508261577845007, "grad_norm": 1.9940264225006104, "learning_rate": 0.0005, "loss": 3.35630464553833, "step": 2450 }, { "epoch": 0.2851989760297882, "grad_norm": 2.0135438442230225, "learning_rate": 0.0005, "loss": 3.537013292312622, "step": 2451 }, { "epoch": 0.2853153362811264, "grad_norm": 1.588438868522644, "learning_rate": 0.0005, "loss": 2.456580877304077, "step": 2452 }, { "epoch": 0.28543169653246453, "grad_norm": 1.006256103515625, "learning_rate": 0.0005, "loss": 1.778764247894287, "step": 2453 }, { "epoch": 0.28554805678380263, "grad_norm": 1.5613702535629272, "learning_rate": 0.0005, "loss": 3.1598780155181885, "step": 2454 }, { "epoch": 0.2856644170351408, "grad_norm": 2.6361513137817383, "learning_rate": 0.0005, "loss": 2.916727304458618, "step": 2455 }, { "epoch": 0.28578077728647894, "grad_norm": 1.5185465812683105, "learning_rate": 0.0005, "loss": 2.4189250469207764, "step": 2456 }, { "epoch": 0.2858971375378171, "grad_norm": 1.609674334526062, "learning_rate": 0.0005, "loss": 2.06524395942688, "step": 2457 }, { "epoch": 0.28601349778915525, "grad_norm": 2.0887725353240967, "learning_rate": 0.0005, "loss": 3.0165815353393555, "step": 2458 }, { "epoch": 0.28612985804049335, "grad_norm": 2.0775675773620605, "learning_rate": 0.0005, "loss": 3.5800299644470215, "step": 2459 }, { "epoch": 0.2862462182918315, "grad_norm": 1.4683773517608643, "learning_rate": 0.0005, "loss": 1.7664196491241455, "step": 2460 }, { "epoch": 0.28636257854316965, "grad_norm": 1.6459877490997314, "learning_rate": 0.0005, "loss": 2.9374172687530518, "step": 2461 }, { "epoch": 0.2864789387945078, "grad_norm": 1.4395381212234497, "learning_rate": 0.0005, "loss": 2.447594165802002, "step": 2462 }, { "epoch": 0.28659529904584596, "grad_norm": 1.3127238750457764, "learning_rate": 0.0005, "loss": 2.3739328384399414, "step": 2463 }, { "epoch": 0.28671165929718406, "grad_norm": 1.2773821353912354, "learning_rate": 0.0005, "loss": 2.745105028152466, "step": 2464 }, { "epoch": 0.2868280195485222, "grad_norm": 0.5498765707015991, "learning_rate": 0.0005, "loss": 1.211961269378662, "step": 2465 }, { "epoch": 0.28694437979986037, "grad_norm": 1.2886196374893188, "learning_rate": 0.0005, "loss": 2.481452703475952, "step": 2466 }, { "epoch": 0.2870607400511985, "grad_norm": 1.4197895526885986, "learning_rate": 0.0005, "loss": 3.1099982261657715, "step": 2467 }, { "epoch": 0.2871771003025367, "grad_norm": 2.2366724014282227, "learning_rate": 0.0005, "loss": 2.6795406341552734, "step": 2468 }, { "epoch": 0.2872934605538748, "grad_norm": 1.2103521823883057, "learning_rate": 0.0005, "loss": 2.287747383117676, "step": 2469 }, { "epoch": 0.28740982080521293, "grad_norm": 1.363227128982544, "learning_rate": 0.0005, "loss": 2.7927792072296143, "step": 2470 }, { "epoch": 0.2875261810565511, "grad_norm": 2.80922532081604, "learning_rate": 0.0005, "loss": 2.393932819366455, "step": 2471 }, { "epoch": 0.28764254130788924, "grad_norm": 3.0367472171783447, "learning_rate": 0.0005, "loss": 2.3473868370056152, "step": 2472 }, { "epoch": 0.28775890155922734, "grad_norm": 1.1200406551361084, "learning_rate": 0.0005, "loss": 2.531217575073242, "step": 2473 }, { "epoch": 0.2878752618105655, "grad_norm": 1.4522632360458374, "learning_rate": 0.0005, "loss": 3.1691086292266846, "step": 2474 }, { "epoch": 0.28799162206190365, "grad_norm": 1.3909413814544678, "learning_rate": 0.0005, "loss": 2.6199026107788086, "step": 2475 }, { "epoch": 0.2881079823132418, "grad_norm": 1.4043647050857544, "learning_rate": 0.0005, "loss": 2.707162857055664, "step": 2476 }, { "epoch": 0.28822434256457996, "grad_norm": 1.4751300811767578, "learning_rate": 0.0005, "loss": 3.012415647506714, "step": 2477 }, { "epoch": 0.28834070281591806, "grad_norm": 1.4401347637176514, "learning_rate": 0.0005, "loss": 2.383439540863037, "step": 2478 }, { "epoch": 0.2884570630672562, "grad_norm": 0.8067825436592102, "learning_rate": 0.0005, "loss": 1.7751681804656982, "step": 2479 }, { "epoch": 0.28857342331859437, "grad_norm": 1.0631362199783325, "learning_rate": 0.0005, "loss": 1.9992094039916992, "step": 2480 }, { "epoch": 0.2886897835699325, "grad_norm": 1.8766798973083496, "learning_rate": 0.0005, "loss": 3.345137596130371, "step": 2481 }, { "epoch": 0.2888061438212707, "grad_norm": 1.735108733177185, "learning_rate": 0.0005, "loss": 2.3946359157562256, "step": 2482 }, { "epoch": 0.2889225040726088, "grad_norm": 1.4618643522262573, "learning_rate": 0.0005, "loss": 2.475102663040161, "step": 2483 }, { "epoch": 0.28903886432394693, "grad_norm": 1.0468015670776367, "learning_rate": 0.0005, "loss": 1.8328969478607178, "step": 2484 }, { "epoch": 0.2891552245752851, "grad_norm": 1.1738418340682983, "learning_rate": 0.0005, "loss": 2.7747960090637207, "step": 2485 }, { "epoch": 0.28927158482662324, "grad_norm": 1.2422451972961426, "learning_rate": 0.0005, "loss": 2.274622678756714, "step": 2486 }, { "epoch": 0.2893879450779614, "grad_norm": 1.4065170288085938, "learning_rate": 0.0005, "loss": 2.8981127738952637, "step": 2487 }, { "epoch": 0.2895043053292995, "grad_norm": 0.9370867609977722, "learning_rate": 0.0005, "loss": 1.5970709323883057, "step": 2488 }, { "epoch": 0.28962066558063765, "grad_norm": 1.5022742748260498, "learning_rate": 0.0005, "loss": 2.209639072418213, "step": 2489 }, { "epoch": 0.2897370258319758, "grad_norm": 1.6341339349746704, "learning_rate": 0.0005, "loss": 2.0256776809692383, "step": 2490 }, { "epoch": 0.28985338608331396, "grad_norm": 1.5109732151031494, "learning_rate": 0.0005, "loss": 2.9172353744506836, "step": 2491 }, { "epoch": 0.28996974633465206, "grad_norm": 1.4283204078674316, "learning_rate": 0.0005, "loss": 3.26835298538208, "step": 2492 }, { "epoch": 0.2900861065859902, "grad_norm": 1.207871675491333, "learning_rate": 0.0005, "loss": 2.213735818862915, "step": 2493 }, { "epoch": 0.29020246683732837, "grad_norm": 2.328056812286377, "learning_rate": 0.0005, "loss": 3.2641756534576416, "step": 2494 }, { "epoch": 0.2903188270886665, "grad_norm": 1.7080339193344116, "learning_rate": 0.0005, "loss": 3.301349639892578, "step": 2495 }, { "epoch": 0.2904351873400047, "grad_norm": 1.4758973121643066, "learning_rate": 0.0005, "loss": 2.910513162612915, "step": 2496 }, { "epoch": 0.2905515475913428, "grad_norm": 1.5071208477020264, "learning_rate": 0.0005, "loss": 3.5107650756835938, "step": 2497 }, { "epoch": 0.29066790784268093, "grad_norm": 1.3258721828460693, "learning_rate": 0.0005, "loss": 2.920543670654297, "step": 2498 }, { "epoch": 0.2907842680940191, "grad_norm": 2.013155937194824, "learning_rate": 0.0005, "loss": 1.9604774713516235, "step": 2499 }, { "epoch": 0.29090062834535724, "grad_norm": 1.0111708641052246, "learning_rate": 0.0005, "loss": 2.8931822776794434, "step": 2500 }, { "epoch": 0.2910169885966954, "grad_norm": 1.6118136644363403, "learning_rate": 0.0005, "loss": 2.628938674926758, "step": 2501 }, { "epoch": 0.2911333488480335, "grad_norm": 1.4301656484603882, "learning_rate": 0.0005, "loss": 2.009669303894043, "step": 2502 }, { "epoch": 0.29124970909937165, "grad_norm": 1.406503438949585, "learning_rate": 0.0005, "loss": 2.753964900970459, "step": 2503 }, { "epoch": 0.2913660693507098, "grad_norm": 1.4985982179641724, "learning_rate": 0.0005, "loss": 3.1609668731689453, "step": 2504 }, { "epoch": 0.29148242960204795, "grad_norm": 3.016528367996216, "learning_rate": 0.0005, "loss": 4.373642921447754, "step": 2505 }, { "epoch": 0.2915987898533861, "grad_norm": 1.5975544452667236, "learning_rate": 0.0005, "loss": 3.3085358142852783, "step": 2506 }, { "epoch": 0.2917151501047242, "grad_norm": 1.508824348449707, "learning_rate": 0.0005, "loss": 2.804561138153076, "step": 2507 }, { "epoch": 0.29183151035606236, "grad_norm": 1.622753381729126, "learning_rate": 0.0005, "loss": 3.2366576194763184, "step": 2508 }, { "epoch": 0.2919478706074005, "grad_norm": 1.3585233688354492, "learning_rate": 0.0005, "loss": 3.6507797241210938, "step": 2509 }, { "epoch": 0.29206423085873867, "grad_norm": 1.9994440078735352, "learning_rate": 0.0005, "loss": 2.181509017944336, "step": 2510 }, { "epoch": 0.29218059111007677, "grad_norm": 1.1355371475219727, "learning_rate": 0.0005, "loss": 2.352661371231079, "step": 2511 }, { "epoch": 0.2922969513614149, "grad_norm": 1.3721851110458374, "learning_rate": 0.0005, "loss": 3.12626314163208, "step": 2512 }, { "epoch": 0.2924133116127531, "grad_norm": 1.6032084226608276, "learning_rate": 0.0005, "loss": 2.5076375007629395, "step": 2513 }, { "epoch": 0.29252967186409123, "grad_norm": 1.9457002878189087, "learning_rate": 0.0005, "loss": 2.392595052719116, "step": 2514 }, { "epoch": 0.2926460321154294, "grad_norm": 2.3311104774475098, "learning_rate": 0.0005, "loss": 2.979477882385254, "step": 2515 }, { "epoch": 0.2927623923667675, "grad_norm": 1.8346070051193237, "learning_rate": 0.0005, "loss": 2.1506307125091553, "step": 2516 }, { "epoch": 0.29287875261810564, "grad_norm": 1.3796058893203735, "learning_rate": 0.0005, "loss": 2.506303310394287, "step": 2517 }, { "epoch": 0.2929951128694438, "grad_norm": 1.5052775144577026, "learning_rate": 0.0005, "loss": 2.3841302394866943, "step": 2518 }, { "epoch": 0.29311147312078195, "grad_norm": 0.9918193221092224, "learning_rate": 0.0005, "loss": 1.9463863372802734, "step": 2519 }, { "epoch": 0.2932278333721201, "grad_norm": 4.823192119598389, "learning_rate": 0.0005, "loss": 3.6671297550201416, "step": 2520 }, { "epoch": 0.2933441936234582, "grad_norm": 2.495576858520508, "learning_rate": 0.0005, "loss": 3.556142807006836, "step": 2521 }, { "epoch": 0.29346055387479636, "grad_norm": 1.6686450242996216, "learning_rate": 0.0005, "loss": 2.534771203994751, "step": 2522 }, { "epoch": 0.2935769141261345, "grad_norm": 2.3938350677490234, "learning_rate": 0.0005, "loss": 3.0803916454315186, "step": 2523 }, { "epoch": 0.29369327437747267, "grad_norm": 1.2823899984359741, "learning_rate": 0.0005, "loss": 1.968970537185669, "step": 2524 }, { "epoch": 0.2938096346288108, "grad_norm": 2.59246563911438, "learning_rate": 0.0005, "loss": 3.0195372104644775, "step": 2525 }, { "epoch": 0.2939259948801489, "grad_norm": 1.2961246967315674, "learning_rate": 0.0005, "loss": 2.7586724758148193, "step": 2526 }, { "epoch": 0.2940423551314871, "grad_norm": 1.2787731885910034, "learning_rate": 0.0005, "loss": 2.570160388946533, "step": 2527 }, { "epoch": 0.29415871538282523, "grad_norm": 1.509345293045044, "learning_rate": 0.0005, "loss": 2.9619698524475098, "step": 2528 }, { "epoch": 0.2942750756341634, "grad_norm": 1.668727159500122, "learning_rate": 0.0005, "loss": 2.8333239555358887, "step": 2529 }, { "epoch": 0.29439143588550154, "grad_norm": 1.8741761445999146, "learning_rate": 0.0005, "loss": 3.8022267818450928, "step": 2530 }, { "epoch": 0.29450779613683964, "grad_norm": 1.0314371585845947, "learning_rate": 0.0005, "loss": 2.7540247440338135, "step": 2531 }, { "epoch": 0.2946241563881778, "grad_norm": 1.3579773902893066, "learning_rate": 0.0005, "loss": 2.5256760120391846, "step": 2532 }, { "epoch": 0.29474051663951595, "grad_norm": 4.914595603942871, "learning_rate": 0.0005, "loss": 4.322915077209473, "step": 2533 }, { "epoch": 0.2948568768908541, "grad_norm": 1.6230322122573853, "learning_rate": 0.0005, "loss": 2.8137640953063965, "step": 2534 }, { "epoch": 0.2949732371421922, "grad_norm": 1.4046807289123535, "learning_rate": 0.0005, "loss": 3.4730722904205322, "step": 2535 }, { "epoch": 0.29508959739353036, "grad_norm": 0.9442599415779114, "learning_rate": 0.0005, "loss": 2.238646984100342, "step": 2536 }, { "epoch": 0.2952059576448685, "grad_norm": 1.7586169242858887, "learning_rate": 0.0005, "loss": 4.07412052154541, "step": 2537 }, { "epoch": 0.29532231789620667, "grad_norm": 1.8519909381866455, "learning_rate": 0.0005, "loss": 3.5612270832061768, "step": 2538 }, { "epoch": 0.2954386781475448, "grad_norm": 1.7727736234664917, "learning_rate": 0.0005, "loss": 3.1617438793182373, "step": 2539 }, { "epoch": 0.2955550383988829, "grad_norm": 1.4596920013427734, "learning_rate": 0.0005, "loss": 3.1291344165802, "step": 2540 }, { "epoch": 0.2956713986502211, "grad_norm": 1.6592257022857666, "learning_rate": 0.0005, "loss": 3.654707431793213, "step": 2541 }, { "epoch": 0.29578775890155923, "grad_norm": 0.734359085559845, "learning_rate": 0.0005, "loss": 1.3374007940292358, "step": 2542 }, { "epoch": 0.2959041191528974, "grad_norm": 1.3209655284881592, "learning_rate": 0.0005, "loss": 2.997495651245117, "step": 2543 }, { "epoch": 0.29602047940423554, "grad_norm": 1.1017769575119019, "learning_rate": 0.0005, "loss": 2.0792064666748047, "step": 2544 }, { "epoch": 0.29613683965557364, "grad_norm": 2.652202606201172, "learning_rate": 0.0005, "loss": 2.788332223892212, "step": 2545 }, { "epoch": 0.2962531999069118, "grad_norm": 1.8970659971237183, "learning_rate": 0.0005, "loss": 1.988734483718872, "step": 2546 }, { "epoch": 0.29636956015824995, "grad_norm": 1.1975091695785522, "learning_rate": 0.0005, "loss": 2.617582321166992, "step": 2547 }, { "epoch": 0.2964859204095881, "grad_norm": 1.545166254043579, "learning_rate": 0.0005, "loss": 3.3212733268737793, "step": 2548 }, { "epoch": 0.29660228066092625, "grad_norm": 1.5507986545562744, "learning_rate": 0.0005, "loss": 2.352349281311035, "step": 2549 }, { "epoch": 0.29671864091226435, "grad_norm": 3.1652746200561523, "learning_rate": 0.0005, "loss": 2.5404775142669678, "step": 2550 }, { "epoch": 0.2968350011636025, "grad_norm": 1.1637811660766602, "learning_rate": 0.0005, "loss": 1.960606336593628, "step": 2551 }, { "epoch": 0.29695136141494066, "grad_norm": 2.418177604675293, "learning_rate": 0.0005, "loss": 2.8136847019195557, "step": 2552 }, { "epoch": 0.2970677216662788, "grad_norm": 1.2113529443740845, "learning_rate": 0.0005, "loss": 3.3891496658325195, "step": 2553 }, { "epoch": 0.2971840819176169, "grad_norm": 1.2118968963623047, "learning_rate": 0.0005, "loss": 3.3337292671203613, "step": 2554 }, { "epoch": 0.29730044216895507, "grad_norm": 1.7687140703201294, "learning_rate": 0.0005, "loss": 2.3171322345733643, "step": 2555 }, { "epoch": 0.2974168024202932, "grad_norm": 1.7009458541870117, "learning_rate": 0.0005, "loss": 2.6379470825195312, "step": 2556 }, { "epoch": 0.2975331626716314, "grad_norm": 1.4262168407440186, "learning_rate": 0.0005, "loss": 3.2571704387664795, "step": 2557 }, { "epoch": 0.29764952292296953, "grad_norm": 1.0538151264190674, "learning_rate": 0.0005, "loss": 2.801589250564575, "step": 2558 }, { "epoch": 0.29776588317430763, "grad_norm": 1.1299984455108643, "learning_rate": 0.0005, "loss": 2.8100814819335938, "step": 2559 }, { "epoch": 0.2978822434256458, "grad_norm": 1.0064573287963867, "learning_rate": 0.0005, "loss": 2.967296600341797, "step": 2560 }, { "epoch": 0.29799860367698394, "grad_norm": 1.0620896816253662, "learning_rate": 0.0005, "loss": 2.7316699028015137, "step": 2561 }, { "epoch": 0.2981149639283221, "grad_norm": 1.671363353729248, "learning_rate": 0.0005, "loss": 2.7049946784973145, "step": 2562 }, { "epoch": 0.29823132417966025, "grad_norm": 1.0764485597610474, "learning_rate": 0.0005, "loss": 2.263392925262451, "step": 2563 }, { "epoch": 0.29834768443099835, "grad_norm": 1.1731793880462646, "learning_rate": 0.0005, "loss": 2.108036518096924, "step": 2564 }, { "epoch": 0.2984640446823365, "grad_norm": 2.0012829303741455, "learning_rate": 0.0005, "loss": 2.800246238708496, "step": 2565 }, { "epoch": 0.29858040493367466, "grad_norm": 1.3421759605407715, "learning_rate": 0.0005, "loss": 2.6729896068573, "step": 2566 }, { "epoch": 0.2986967651850128, "grad_norm": 1.8303272724151611, "learning_rate": 0.0005, "loss": 2.889739513397217, "step": 2567 }, { "epoch": 0.29881312543635097, "grad_norm": 1.1305625438690186, "learning_rate": 0.0005, "loss": 2.4032983779907227, "step": 2568 }, { "epoch": 0.29892948568768907, "grad_norm": 1.8583776950836182, "learning_rate": 0.0005, "loss": 3.1246118545532227, "step": 2569 }, { "epoch": 0.2990458459390272, "grad_norm": 1.4893176555633545, "learning_rate": 0.0005, "loss": 1.987287998199463, "step": 2570 }, { "epoch": 0.2991622061903654, "grad_norm": 1.8642005920410156, "learning_rate": 0.0005, "loss": 2.916667938232422, "step": 2571 }, { "epoch": 0.29927856644170353, "grad_norm": 1.423061728477478, "learning_rate": 0.0005, "loss": 2.498445749282837, "step": 2572 }, { "epoch": 0.29939492669304163, "grad_norm": 1.7093839645385742, "learning_rate": 0.0005, "loss": 1.770568609237671, "step": 2573 }, { "epoch": 0.2995112869443798, "grad_norm": 2.3141496181488037, "learning_rate": 0.0005, "loss": 3.2796812057495117, "step": 2574 }, { "epoch": 0.29962764719571794, "grad_norm": 1.9289778470993042, "learning_rate": 0.0005, "loss": 3.7220396995544434, "step": 2575 }, { "epoch": 0.2997440074470561, "grad_norm": 2.0913338661193848, "learning_rate": 0.0005, "loss": 2.209686756134033, "step": 2576 }, { "epoch": 0.29986036769839425, "grad_norm": 1.653732419013977, "learning_rate": 0.0005, "loss": 2.628824234008789, "step": 2577 }, { "epoch": 0.29997672794973235, "grad_norm": 1.5918190479278564, "learning_rate": 0.0005, "loss": 3.4249768257141113, "step": 2578 }, { "epoch": 0.3000930882010705, "grad_norm": 1.4635164737701416, "learning_rate": 0.0005, "loss": 1.9988638162612915, "step": 2579 }, { "epoch": 0.30020944845240866, "grad_norm": 1.4820092916488647, "learning_rate": 0.0005, "loss": 2.9385569095611572, "step": 2580 }, { "epoch": 0.3003258087037468, "grad_norm": 1.3759599924087524, "learning_rate": 0.0005, "loss": 4.049814701080322, "step": 2581 }, { "epoch": 0.30044216895508497, "grad_norm": 0.6971772909164429, "learning_rate": 0.0005, "loss": 1.4238539934158325, "step": 2582 }, { "epoch": 0.30055852920642306, "grad_norm": 1.2120821475982666, "learning_rate": 0.0005, "loss": 3.1462903022766113, "step": 2583 }, { "epoch": 0.3006748894577612, "grad_norm": 1.4945987462997437, "learning_rate": 0.0005, "loss": 1.8240286111831665, "step": 2584 }, { "epoch": 0.3007912497090994, "grad_norm": 1.944807529449463, "learning_rate": 0.0005, "loss": 2.8372178077697754, "step": 2585 }, { "epoch": 0.3009076099604375, "grad_norm": 2.07163143157959, "learning_rate": 0.0005, "loss": 2.8738083839416504, "step": 2586 }, { "epoch": 0.3010239702117757, "grad_norm": 1.8340388536453247, "learning_rate": 0.0005, "loss": 2.5620789527893066, "step": 2587 }, { "epoch": 0.3011403304631138, "grad_norm": 1.6004481315612793, "learning_rate": 0.0005, "loss": 2.0775864124298096, "step": 2588 }, { "epoch": 0.30125669071445194, "grad_norm": 1.740578055381775, "learning_rate": 0.0005, "loss": 2.7603585720062256, "step": 2589 }, { "epoch": 0.3013730509657901, "grad_norm": 1.2970950603485107, "learning_rate": 0.0005, "loss": 2.649691581726074, "step": 2590 }, { "epoch": 0.30148941121712824, "grad_norm": 1.6633912324905396, "learning_rate": 0.0005, "loss": 2.6754302978515625, "step": 2591 }, { "epoch": 0.3016057714684664, "grad_norm": 1.4030988216400146, "learning_rate": 0.0005, "loss": 2.7899820804595947, "step": 2592 }, { "epoch": 0.3017221317198045, "grad_norm": 2.213146209716797, "learning_rate": 0.0005, "loss": 3.246814250946045, "step": 2593 }, { "epoch": 0.30183849197114265, "grad_norm": 1.8076905012130737, "learning_rate": 0.0005, "loss": 3.5188488960266113, "step": 2594 }, { "epoch": 0.3019548522224808, "grad_norm": 1.3961155414581299, "learning_rate": 0.0005, "loss": 2.8871657848358154, "step": 2595 }, { "epoch": 0.30207121247381896, "grad_norm": 1.0884641408920288, "learning_rate": 0.0005, "loss": 3.285460948944092, "step": 2596 }, { "epoch": 0.30218757272515706, "grad_norm": 1.4089772701263428, "learning_rate": 0.0005, "loss": 2.88446307182312, "step": 2597 }, { "epoch": 0.3023039329764952, "grad_norm": 1.8198028802871704, "learning_rate": 0.0005, "loss": 3.5161874294281006, "step": 2598 }, { "epoch": 0.30242029322783337, "grad_norm": 1.2309684753417969, "learning_rate": 0.0005, "loss": 2.409395933151245, "step": 2599 }, { "epoch": 0.3025366534791715, "grad_norm": 5.410592079162598, "learning_rate": 0.0005, "loss": 2.592890739440918, "step": 2600 }, { "epoch": 0.3026530137305097, "grad_norm": 0.8168410658836365, "learning_rate": 0.0005, "loss": 1.2963062524795532, "step": 2601 }, { "epoch": 0.3027693739818478, "grad_norm": 1.6332334280014038, "learning_rate": 0.0005, "loss": 3.3155972957611084, "step": 2602 }, { "epoch": 0.30288573423318593, "grad_norm": 0.9004361033439636, "learning_rate": 0.0005, "loss": 2.198108673095703, "step": 2603 }, { "epoch": 0.3030020944845241, "grad_norm": 1.1056987047195435, "learning_rate": 0.0005, "loss": 1.7668213844299316, "step": 2604 }, { "epoch": 0.30311845473586224, "grad_norm": 3.052408218383789, "learning_rate": 0.0005, "loss": 2.53009033203125, "step": 2605 }, { "epoch": 0.3032348149872004, "grad_norm": 1.0984711647033691, "learning_rate": 0.0005, "loss": 2.423490285873413, "step": 2606 }, { "epoch": 0.3033511752385385, "grad_norm": 1.4042905569076538, "learning_rate": 0.0005, "loss": 2.6018643379211426, "step": 2607 }, { "epoch": 0.30346753548987665, "grad_norm": 1.9607641696929932, "learning_rate": 0.0005, "loss": 3.289897918701172, "step": 2608 }, { "epoch": 0.3035838957412148, "grad_norm": 1.608778953552246, "learning_rate": 0.0005, "loss": 2.9865951538085938, "step": 2609 }, { "epoch": 0.30370025599255296, "grad_norm": 0.8711371421813965, "learning_rate": 0.0005, "loss": 2.214155673980713, "step": 2610 }, { "epoch": 0.3038166162438911, "grad_norm": 1.7817353010177612, "learning_rate": 0.0005, "loss": 3.386536121368408, "step": 2611 }, { "epoch": 0.3039329764952292, "grad_norm": 0.6715436577796936, "learning_rate": 0.0005, "loss": 1.7564767599105835, "step": 2612 }, { "epoch": 0.30404933674656737, "grad_norm": 1.164249062538147, "learning_rate": 0.0005, "loss": 3.063964366912842, "step": 2613 }, { "epoch": 0.3041656969979055, "grad_norm": 1.0094205141067505, "learning_rate": 0.0005, "loss": 1.640662431716919, "step": 2614 }, { "epoch": 0.3042820572492437, "grad_norm": 1.5936671495437622, "learning_rate": 0.0005, "loss": 3.4277162551879883, "step": 2615 }, { "epoch": 0.3043984175005818, "grad_norm": 1.3902015686035156, "learning_rate": 0.0005, "loss": 2.378671407699585, "step": 2616 }, { "epoch": 0.30451477775191993, "grad_norm": 1.1714025735855103, "learning_rate": 0.0005, "loss": 2.780519962310791, "step": 2617 }, { "epoch": 0.3046311380032581, "grad_norm": 2.641470193862915, "learning_rate": 0.0005, "loss": 2.587191581726074, "step": 2618 }, { "epoch": 0.30474749825459624, "grad_norm": 1.7196552753448486, "learning_rate": 0.0005, "loss": 2.7826807498931885, "step": 2619 }, { "epoch": 0.3048638585059344, "grad_norm": 1.1146475076675415, "learning_rate": 0.0005, "loss": 2.2365455627441406, "step": 2620 }, { "epoch": 0.3049802187572725, "grad_norm": 0.9511536955833435, "learning_rate": 0.0005, "loss": 1.8118383884429932, "step": 2621 }, { "epoch": 0.30509657900861065, "grad_norm": 1.3791271448135376, "learning_rate": 0.0005, "loss": 2.877645492553711, "step": 2622 }, { "epoch": 0.3052129392599488, "grad_norm": 1.348244309425354, "learning_rate": 0.0005, "loss": 2.92271089553833, "step": 2623 }, { "epoch": 0.30532929951128696, "grad_norm": 1.4904946088790894, "learning_rate": 0.0005, "loss": 2.6622509956359863, "step": 2624 }, { "epoch": 0.3054456597626251, "grad_norm": 1.495985746383667, "learning_rate": 0.0005, "loss": 1.6295089721679688, "step": 2625 }, { "epoch": 0.3055620200139632, "grad_norm": 0.9856493473052979, "learning_rate": 0.0005, "loss": 2.3606815338134766, "step": 2626 }, { "epoch": 0.30567838026530136, "grad_norm": 3.0611140727996826, "learning_rate": 0.0005, "loss": 3.9103591442108154, "step": 2627 }, { "epoch": 0.3057947405166395, "grad_norm": 1.3078091144561768, "learning_rate": 0.0005, "loss": 2.579293727874756, "step": 2628 }, { "epoch": 0.3059111007679777, "grad_norm": 1.5084903240203857, "learning_rate": 0.0005, "loss": 1.8716679811477661, "step": 2629 }, { "epoch": 0.3060274610193158, "grad_norm": 1.7944217920303345, "learning_rate": 0.0005, "loss": 3.0049891471862793, "step": 2630 }, { "epoch": 0.3061438212706539, "grad_norm": 1.9800302982330322, "learning_rate": 0.0005, "loss": 2.748002529144287, "step": 2631 }, { "epoch": 0.3062601815219921, "grad_norm": 1.2246177196502686, "learning_rate": 0.0005, "loss": 2.6972880363464355, "step": 2632 }, { "epoch": 0.30637654177333024, "grad_norm": 1.6672301292419434, "learning_rate": 0.0005, "loss": 2.640988349914551, "step": 2633 }, { "epoch": 0.3064929020246684, "grad_norm": 1.184309482574463, "learning_rate": 0.0005, "loss": 2.449720859527588, "step": 2634 }, { "epoch": 0.3066092622760065, "grad_norm": 1.4502148628234863, "learning_rate": 0.0005, "loss": 3.4184205532073975, "step": 2635 }, { "epoch": 0.30672562252734464, "grad_norm": 1.608622431755066, "learning_rate": 0.0005, "loss": 2.8657991886138916, "step": 2636 }, { "epoch": 0.3068419827786828, "grad_norm": 1.2282804250717163, "learning_rate": 0.0005, "loss": 2.023055076599121, "step": 2637 }, { "epoch": 0.30695834303002095, "grad_norm": 2.921027183532715, "learning_rate": 0.0005, "loss": 2.8409948348999023, "step": 2638 }, { "epoch": 0.3070747032813591, "grad_norm": 1.304185152053833, "learning_rate": 0.0005, "loss": 1.8253693580627441, "step": 2639 }, { "epoch": 0.3071910635326972, "grad_norm": 1.0451529026031494, "learning_rate": 0.0005, "loss": 1.885663628578186, "step": 2640 }, { "epoch": 0.30730742378403536, "grad_norm": 1.965948462486267, "learning_rate": 0.0005, "loss": 2.5058512687683105, "step": 2641 }, { "epoch": 0.3074237840353735, "grad_norm": 1.4247303009033203, "learning_rate": 0.0005, "loss": 2.179577350616455, "step": 2642 }, { "epoch": 0.30754014428671167, "grad_norm": 1.67776620388031, "learning_rate": 0.0005, "loss": 3.420281410217285, "step": 2643 }, { "epoch": 0.3076565045380498, "grad_norm": 1.116676926612854, "learning_rate": 0.0005, "loss": 2.9199514389038086, "step": 2644 }, { "epoch": 0.3077728647893879, "grad_norm": 1.3518067598342896, "learning_rate": 0.0005, "loss": 2.0732672214508057, "step": 2645 }, { "epoch": 0.3078892250407261, "grad_norm": 1.6510145664215088, "learning_rate": 0.0005, "loss": 2.955486536026001, "step": 2646 }, { "epoch": 0.30800558529206423, "grad_norm": 0.9021035432815552, "learning_rate": 0.0005, "loss": 1.7718074321746826, "step": 2647 }, { "epoch": 0.3081219455434024, "grad_norm": 1.1616393327713013, "learning_rate": 0.0005, "loss": 3.0905559062957764, "step": 2648 }, { "epoch": 0.30823830579474054, "grad_norm": 2.0010225772857666, "learning_rate": 0.0005, "loss": 2.251816749572754, "step": 2649 }, { "epoch": 0.30835466604607864, "grad_norm": 1.3249309062957764, "learning_rate": 0.0005, "loss": 2.762444019317627, "step": 2650 }, { "epoch": 0.3084710262974168, "grad_norm": 2.70540189743042, "learning_rate": 0.0005, "loss": 3.1722168922424316, "step": 2651 }, { "epoch": 0.30858738654875495, "grad_norm": 2.63199782371521, "learning_rate": 0.0005, "loss": 3.8662211894989014, "step": 2652 }, { "epoch": 0.3087037468000931, "grad_norm": 8.13338565826416, "learning_rate": 0.0005, "loss": 2.375678062438965, "step": 2653 }, { "epoch": 0.30882010705143126, "grad_norm": 1.4726545810699463, "learning_rate": 0.0005, "loss": 3.28302001953125, "step": 2654 }, { "epoch": 0.30893646730276936, "grad_norm": 1.420387864112854, "learning_rate": 0.0005, "loss": 1.8984508514404297, "step": 2655 }, { "epoch": 0.3090528275541075, "grad_norm": 1.1909576654434204, "learning_rate": 0.0005, "loss": 2.9398512840270996, "step": 2656 }, { "epoch": 0.30916918780544567, "grad_norm": 2.0916213989257812, "learning_rate": 0.0005, "loss": 2.77227783203125, "step": 2657 }, { "epoch": 0.3092855480567838, "grad_norm": 1.3272796869277954, "learning_rate": 0.0005, "loss": 2.70514178276062, "step": 2658 }, { "epoch": 0.3094019083081219, "grad_norm": 2.597667694091797, "learning_rate": 0.0005, "loss": 2.7196044921875, "step": 2659 }, { "epoch": 0.3095182685594601, "grad_norm": 1.9500954151153564, "learning_rate": 0.0005, "loss": 3.9217724800109863, "step": 2660 }, { "epoch": 0.30963462881079823, "grad_norm": 1.078900694847107, "learning_rate": 0.0005, "loss": 1.7501407861709595, "step": 2661 }, { "epoch": 0.3097509890621364, "grad_norm": 1.8047492504119873, "learning_rate": 0.0005, "loss": 3.3581714630126953, "step": 2662 }, { "epoch": 0.30986734931347454, "grad_norm": 1.8419286012649536, "learning_rate": 0.0005, "loss": 3.3411171436309814, "step": 2663 }, { "epoch": 0.30998370956481264, "grad_norm": 1.5390262603759766, "learning_rate": 0.0005, "loss": 3.443376302719116, "step": 2664 }, { "epoch": 0.3101000698161508, "grad_norm": 1.77194344997406, "learning_rate": 0.0005, "loss": 2.5353190898895264, "step": 2665 }, { "epoch": 0.31021643006748895, "grad_norm": 1.8262406587600708, "learning_rate": 0.0005, "loss": 2.880417585372925, "step": 2666 }, { "epoch": 0.3103327903188271, "grad_norm": 1.5437978506088257, "learning_rate": 0.0005, "loss": 2.6522250175476074, "step": 2667 }, { "epoch": 0.31044915057016526, "grad_norm": 1.547623634338379, "learning_rate": 0.0005, "loss": 2.443972587585449, "step": 2668 }, { "epoch": 0.31056551082150335, "grad_norm": 1.1474658250808716, "learning_rate": 0.0005, "loss": 2.0153651237487793, "step": 2669 }, { "epoch": 0.3106818710728415, "grad_norm": 1.1222329139709473, "learning_rate": 0.0005, "loss": 1.769317388534546, "step": 2670 }, { "epoch": 0.31079823132417966, "grad_norm": 1.570754051208496, "learning_rate": 0.0005, "loss": 2.7083468437194824, "step": 2671 }, { "epoch": 0.3109145915755178, "grad_norm": 1.1080241203308105, "learning_rate": 0.0005, "loss": 2.8226888179779053, "step": 2672 }, { "epoch": 0.311030951826856, "grad_norm": 0.6522505879402161, "learning_rate": 0.0005, "loss": 1.527038812637329, "step": 2673 }, { "epoch": 0.31114731207819407, "grad_norm": 1.4882867336273193, "learning_rate": 0.0005, "loss": 3.023973226547241, "step": 2674 }, { "epoch": 0.3112636723295322, "grad_norm": 5.056358814239502, "learning_rate": 0.0005, "loss": 4.379003047943115, "step": 2675 }, { "epoch": 0.3113800325808704, "grad_norm": 1.4847445487976074, "learning_rate": 0.0005, "loss": 2.9933788776397705, "step": 2676 }, { "epoch": 0.31149639283220854, "grad_norm": 1.3976844549179077, "learning_rate": 0.0005, "loss": 3.5131263732910156, "step": 2677 }, { "epoch": 0.31161275308354663, "grad_norm": 1.3015083074569702, "learning_rate": 0.0005, "loss": 2.90081787109375, "step": 2678 }, { "epoch": 0.3117291133348848, "grad_norm": 1.7967826128005981, "learning_rate": 0.0005, "loss": 3.367605209350586, "step": 2679 }, { "epoch": 0.31184547358622294, "grad_norm": 3.2052366733551025, "learning_rate": 0.0005, "loss": 2.963588237762451, "step": 2680 }, { "epoch": 0.3119618338375611, "grad_norm": 1.2908661365509033, "learning_rate": 0.0005, "loss": 3.0489766597747803, "step": 2681 }, { "epoch": 0.31207819408889925, "grad_norm": 1.537253975868225, "learning_rate": 0.0005, "loss": 3.9992377758026123, "step": 2682 }, { "epoch": 0.31219455434023735, "grad_norm": 1.09306001663208, "learning_rate": 0.0005, "loss": 2.096548080444336, "step": 2683 }, { "epoch": 0.3123109145915755, "grad_norm": 1.1079188585281372, "learning_rate": 0.0005, "loss": 2.2732982635498047, "step": 2684 }, { "epoch": 0.31242727484291366, "grad_norm": 1.049542784690857, "learning_rate": 0.0005, "loss": 2.434772491455078, "step": 2685 }, { "epoch": 0.3125436350942518, "grad_norm": 1.0374869108200073, "learning_rate": 0.0005, "loss": 2.1649765968322754, "step": 2686 }, { "epoch": 0.31265999534558997, "grad_norm": 1.7500662803649902, "learning_rate": 0.0005, "loss": 3.381432056427002, "step": 2687 }, { "epoch": 0.31277635559692807, "grad_norm": 2.0761806964874268, "learning_rate": 0.0005, "loss": 3.3655402660369873, "step": 2688 }, { "epoch": 0.3128927158482662, "grad_norm": 1.7947094440460205, "learning_rate": 0.0005, "loss": 2.2043304443359375, "step": 2689 }, { "epoch": 0.3130090760996044, "grad_norm": 3.1569032669067383, "learning_rate": 0.0005, "loss": 2.997452735900879, "step": 2690 }, { "epoch": 0.31312543635094253, "grad_norm": 1.1087287664413452, "learning_rate": 0.0005, "loss": 1.2472913265228271, "step": 2691 }, { "epoch": 0.3132417966022807, "grad_norm": 1.137703776359558, "learning_rate": 0.0005, "loss": 2.8862757682800293, "step": 2692 }, { "epoch": 0.3133581568536188, "grad_norm": 1.2599291801452637, "learning_rate": 0.0005, "loss": 2.0187792778015137, "step": 2693 }, { "epoch": 0.31347451710495694, "grad_norm": 1.366416573524475, "learning_rate": 0.0005, "loss": 2.0738673210144043, "step": 2694 }, { "epoch": 0.3135908773562951, "grad_norm": 1.697181224822998, "learning_rate": 0.0005, "loss": 2.681551694869995, "step": 2695 }, { "epoch": 0.31370723760763325, "grad_norm": 1.8391103744506836, "learning_rate": 0.0005, "loss": 3.473078727722168, "step": 2696 }, { "epoch": 0.31382359785897135, "grad_norm": 1.0756993293762207, "learning_rate": 0.0005, "loss": 2.4561853408813477, "step": 2697 }, { "epoch": 0.3139399581103095, "grad_norm": 1.891585350036621, "learning_rate": 0.0005, "loss": 2.6269519329071045, "step": 2698 }, { "epoch": 0.31405631836164766, "grad_norm": 1.4118800163269043, "learning_rate": 0.0005, "loss": 2.572838306427002, "step": 2699 }, { "epoch": 0.3141726786129858, "grad_norm": 1.479665756225586, "learning_rate": 0.0005, "loss": 3.0450000762939453, "step": 2700 }, { "epoch": 0.31428903886432397, "grad_norm": 1.4392988681793213, "learning_rate": 0.0005, "loss": 3.2205681800842285, "step": 2701 }, { "epoch": 0.31440539911566207, "grad_norm": 1.6179747581481934, "learning_rate": 0.0005, "loss": 3.3391361236572266, "step": 2702 }, { "epoch": 0.3145217593670002, "grad_norm": 2.318920612335205, "learning_rate": 0.0005, "loss": 2.7779242992401123, "step": 2703 }, { "epoch": 0.3146381196183384, "grad_norm": 0.9971416592597961, "learning_rate": 0.0005, "loss": 1.3413132429122925, "step": 2704 }, { "epoch": 0.31475447986967653, "grad_norm": 1.0209873914718628, "learning_rate": 0.0005, "loss": 2.283982515335083, "step": 2705 }, { "epoch": 0.3148708401210147, "grad_norm": 1.1039974689483643, "learning_rate": 0.0005, "loss": 3.400420665740967, "step": 2706 }, { "epoch": 0.3149872003723528, "grad_norm": 1.5598567724227905, "learning_rate": 0.0005, "loss": 3.3991169929504395, "step": 2707 }, { "epoch": 0.31510356062369094, "grad_norm": 1.184046983718872, "learning_rate": 0.0005, "loss": 2.205728054046631, "step": 2708 }, { "epoch": 0.3152199208750291, "grad_norm": 1.6391711235046387, "learning_rate": 0.0005, "loss": 2.558366298675537, "step": 2709 }, { "epoch": 0.31533628112636725, "grad_norm": 2.1706531047821045, "learning_rate": 0.0005, "loss": 2.786032199859619, "step": 2710 }, { "epoch": 0.3154526413777054, "grad_norm": 2.0836427211761475, "learning_rate": 0.0005, "loss": 2.2149875164031982, "step": 2711 }, { "epoch": 0.3155690016290435, "grad_norm": 1.9387314319610596, "learning_rate": 0.0005, "loss": 2.5848395824432373, "step": 2712 }, { "epoch": 0.31568536188038165, "grad_norm": 1.5400722026824951, "learning_rate": 0.0005, "loss": 1.6333752870559692, "step": 2713 }, { "epoch": 0.3158017221317198, "grad_norm": 1.9029853343963623, "learning_rate": 0.0005, "loss": 3.779604911804199, "step": 2714 }, { "epoch": 0.31591808238305796, "grad_norm": 2.2818915843963623, "learning_rate": 0.0005, "loss": 3.380012035369873, "step": 2715 }, { "epoch": 0.3160344426343961, "grad_norm": 3.21740460395813, "learning_rate": 0.0005, "loss": 3.2102465629577637, "step": 2716 }, { "epoch": 0.3161508028857342, "grad_norm": 1.3268345594406128, "learning_rate": 0.0005, "loss": 2.559609889984131, "step": 2717 }, { "epoch": 0.31626716313707237, "grad_norm": 2.3766367435455322, "learning_rate": 0.0005, "loss": 3.194746494293213, "step": 2718 }, { "epoch": 0.3163835233884105, "grad_norm": 1.246375560760498, "learning_rate": 0.0005, "loss": 2.8856730461120605, "step": 2719 }, { "epoch": 0.3164998836397487, "grad_norm": 1.229235291481018, "learning_rate": 0.0005, "loss": 2.027296543121338, "step": 2720 }, { "epoch": 0.3166162438910868, "grad_norm": 2.4699530601501465, "learning_rate": 0.0005, "loss": 2.453206777572632, "step": 2721 }, { "epoch": 0.31673260414242493, "grad_norm": 2.0303232669830322, "learning_rate": 0.0005, "loss": 2.9476146697998047, "step": 2722 }, { "epoch": 0.3168489643937631, "grad_norm": 1.1473535299301147, "learning_rate": 0.0005, "loss": 2.873598098754883, "step": 2723 }, { "epoch": 0.31696532464510124, "grad_norm": 1.7114776372909546, "learning_rate": 0.0005, "loss": 3.229490280151367, "step": 2724 }, { "epoch": 0.3170816848964394, "grad_norm": 1.540689468383789, "learning_rate": 0.0005, "loss": 2.226670026779175, "step": 2725 }, { "epoch": 0.3171980451477775, "grad_norm": 1.4991974830627441, "learning_rate": 0.0005, "loss": 3.618332624435425, "step": 2726 }, { "epoch": 0.31731440539911565, "grad_norm": 2.375962972640991, "learning_rate": 0.0005, "loss": 3.4658608436584473, "step": 2727 }, { "epoch": 0.3174307656504538, "grad_norm": 1.5093399286270142, "learning_rate": 0.0005, "loss": 2.3834991455078125, "step": 2728 }, { "epoch": 0.31754712590179196, "grad_norm": 1.2474769353866577, "learning_rate": 0.0005, "loss": 2.2652392387390137, "step": 2729 }, { "epoch": 0.3176634861531301, "grad_norm": 0.8926898241043091, "learning_rate": 0.0005, "loss": 2.0693352222442627, "step": 2730 }, { "epoch": 0.3177798464044682, "grad_norm": 2.023954391479492, "learning_rate": 0.0005, "loss": 3.2011570930480957, "step": 2731 }, { "epoch": 0.31789620665580637, "grad_norm": 0.9671939611434937, "learning_rate": 0.0005, "loss": 1.1864405870437622, "step": 2732 }, { "epoch": 0.3180125669071445, "grad_norm": 1.3918437957763672, "learning_rate": 0.0005, "loss": 3.1129395961761475, "step": 2733 }, { "epoch": 0.3181289271584827, "grad_norm": 1.8146287202835083, "learning_rate": 0.0005, "loss": 3.9951300621032715, "step": 2734 }, { "epoch": 0.31824528740982083, "grad_norm": 1.1069053411483765, "learning_rate": 0.0005, "loss": 1.8819986581802368, "step": 2735 }, { "epoch": 0.31836164766115893, "grad_norm": 1.2267266511917114, "learning_rate": 0.0005, "loss": 3.118350028991699, "step": 2736 }, { "epoch": 0.3184780079124971, "grad_norm": 0.564929187297821, "learning_rate": 0.0005, "loss": 1.101633071899414, "step": 2737 }, { "epoch": 0.31859436816383524, "grad_norm": 1.4097847938537598, "learning_rate": 0.0005, "loss": 2.6462624073028564, "step": 2738 }, { "epoch": 0.3187107284151734, "grad_norm": 1.3417928218841553, "learning_rate": 0.0005, "loss": 3.7151222229003906, "step": 2739 }, { "epoch": 0.3188270886665115, "grad_norm": 1.7835251092910767, "learning_rate": 0.0005, "loss": 1.905583143234253, "step": 2740 }, { "epoch": 0.31894344891784965, "grad_norm": 1.6253830194473267, "learning_rate": 0.0005, "loss": 2.345700979232788, "step": 2741 }, { "epoch": 0.3190598091691878, "grad_norm": 1.6236863136291504, "learning_rate": 0.0005, "loss": 2.8550848960876465, "step": 2742 }, { "epoch": 0.31917616942052596, "grad_norm": 1.507595419883728, "learning_rate": 0.0005, "loss": 3.6029303073883057, "step": 2743 }, { "epoch": 0.3192925296718641, "grad_norm": 1.2733125686645508, "learning_rate": 0.0005, "loss": 2.203300714492798, "step": 2744 }, { "epoch": 0.3194088899232022, "grad_norm": 1.1780147552490234, "learning_rate": 0.0005, "loss": 1.949838399887085, "step": 2745 }, { "epoch": 0.31952525017454037, "grad_norm": 1.910536766052246, "learning_rate": 0.0005, "loss": 2.455843210220337, "step": 2746 }, { "epoch": 0.3196416104258785, "grad_norm": 0.9390857219696045, "learning_rate": 0.0005, "loss": 2.6759822368621826, "step": 2747 }, { "epoch": 0.3197579706772167, "grad_norm": 0.9112976789474487, "learning_rate": 0.0005, "loss": 1.7976751327514648, "step": 2748 }, { "epoch": 0.31987433092855483, "grad_norm": 0.8712980151176453, "learning_rate": 0.0005, "loss": 2.640842914581299, "step": 2749 }, { "epoch": 0.3199906911798929, "grad_norm": 1.0632745027542114, "learning_rate": 0.0005, "loss": 2.6537561416625977, "step": 2750 }, { "epoch": 0.3201070514312311, "grad_norm": 1.097491979598999, "learning_rate": 0.0005, "loss": 2.339694023132324, "step": 2751 }, { "epoch": 0.32022341168256924, "grad_norm": 1.4686355590820312, "learning_rate": 0.0005, "loss": 2.6321797370910645, "step": 2752 }, { "epoch": 0.3203397719339074, "grad_norm": 1.8641924858093262, "learning_rate": 0.0005, "loss": 2.6741433143615723, "step": 2753 }, { "epoch": 0.32045613218524555, "grad_norm": 1.6832281351089478, "learning_rate": 0.0005, "loss": 4.177314758300781, "step": 2754 }, { "epoch": 0.32057249243658364, "grad_norm": 2.092519521713257, "learning_rate": 0.0005, "loss": 2.650723457336426, "step": 2755 }, { "epoch": 0.3206888526879218, "grad_norm": 2.2761178016662598, "learning_rate": 0.0005, "loss": 3.1623663902282715, "step": 2756 }, { "epoch": 0.32080521293925995, "grad_norm": 2.0712034702301025, "learning_rate": 0.0005, "loss": 3.445685386657715, "step": 2757 }, { "epoch": 0.3209215731905981, "grad_norm": 2.056272268295288, "learning_rate": 0.0005, "loss": 2.2768025398254395, "step": 2758 }, { "epoch": 0.3210379334419362, "grad_norm": 0.9176197052001953, "learning_rate": 0.0005, "loss": 1.8057924509048462, "step": 2759 }, { "epoch": 0.32115429369327436, "grad_norm": 1.5996547937393188, "learning_rate": 0.0005, "loss": 2.8868794441223145, "step": 2760 }, { "epoch": 0.3212706539446125, "grad_norm": 2.2621164321899414, "learning_rate": 0.0005, "loss": 2.7930731773376465, "step": 2761 }, { "epoch": 0.32138701419595067, "grad_norm": 2.3069536685943604, "learning_rate": 0.0005, "loss": 3.56276273727417, "step": 2762 }, { "epoch": 0.3215033744472888, "grad_norm": 3.4266374111175537, "learning_rate": 0.0005, "loss": 2.403801918029785, "step": 2763 }, { "epoch": 0.3216197346986269, "grad_norm": 1.1924471855163574, "learning_rate": 0.0005, "loss": 1.9130830764770508, "step": 2764 }, { "epoch": 0.3217360949499651, "grad_norm": 1.020106315612793, "learning_rate": 0.0005, "loss": 1.8525168895721436, "step": 2765 }, { "epoch": 0.32185245520130323, "grad_norm": 1.911275029182434, "learning_rate": 0.0005, "loss": 3.1127305030822754, "step": 2766 }, { "epoch": 0.3219688154526414, "grad_norm": 3.076185464859009, "learning_rate": 0.0005, "loss": 2.8352303504943848, "step": 2767 }, { "epoch": 0.32208517570397954, "grad_norm": 0.7619187235832214, "learning_rate": 0.0005, "loss": 1.9441629648208618, "step": 2768 }, { "epoch": 0.32220153595531764, "grad_norm": 1.4736870527267456, "learning_rate": 0.0005, "loss": 2.754626512527466, "step": 2769 }, { "epoch": 0.3223178962066558, "grad_norm": 1.9170572757720947, "learning_rate": 0.0005, "loss": 3.3695521354675293, "step": 2770 }, { "epoch": 0.32243425645799395, "grad_norm": 2.3196861743927, "learning_rate": 0.0005, "loss": 2.752880573272705, "step": 2771 }, { "epoch": 0.3225506167093321, "grad_norm": 1.2419006824493408, "learning_rate": 0.0005, "loss": 2.1331071853637695, "step": 2772 }, { "epoch": 0.32266697696067026, "grad_norm": 0.9429122805595398, "learning_rate": 0.0005, "loss": 2.244014263153076, "step": 2773 }, { "epoch": 0.32278333721200836, "grad_norm": 1.6781415939331055, "learning_rate": 0.0005, "loss": 3.656585216522217, "step": 2774 }, { "epoch": 0.3228996974633465, "grad_norm": 1.3770055770874023, "learning_rate": 0.0005, "loss": 2.411693572998047, "step": 2775 }, { "epoch": 0.32301605771468467, "grad_norm": 1.4240407943725586, "learning_rate": 0.0005, "loss": 2.732296943664551, "step": 2776 }, { "epoch": 0.3231324179660228, "grad_norm": 1.3203983306884766, "learning_rate": 0.0005, "loss": 2.8892321586608887, "step": 2777 }, { "epoch": 0.323248778217361, "grad_norm": 1.380839228630066, "learning_rate": 0.0005, "loss": 2.8692309856414795, "step": 2778 }, { "epoch": 0.3233651384686991, "grad_norm": 1.8802239894866943, "learning_rate": 0.0005, "loss": 3.580350399017334, "step": 2779 }, { "epoch": 0.32348149872003723, "grad_norm": 1.8625292778015137, "learning_rate": 0.0005, "loss": 2.514815330505371, "step": 2780 }, { "epoch": 0.3235978589713754, "grad_norm": 1.9206985235214233, "learning_rate": 0.0005, "loss": 2.4406769275665283, "step": 2781 }, { "epoch": 0.32371421922271354, "grad_norm": 1.8505829572677612, "learning_rate": 0.0005, "loss": 3.3187131881713867, "step": 2782 }, { "epoch": 0.32383057947405164, "grad_norm": 1.6806622743606567, "learning_rate": 0.0005, "loss": 2.6963682174682617, "step": 2783 }, { "epoch": 0.3239469397253898, "grad_norm": 2.459476947784424, "learning_rate": 0.0005, "loss": 3.1147918701171875, "step": 2784 }, { "epoch": 0.32406329997672795, "grad_norm": 1.4250311851501465, "learning_rate": 0.0005, "loss": 2.8786745071411133, "step": 2785 }, { "epoch": 0.3241796602280661, "grad_norm": 1.0962566137313843, "learning_rate": 0.0005, "loss": 2.789114475250244, "step": 2786 }, { "epoch": 0.32429602047940426, "grad_norm": 1.5413535833358765, "learning_rate": 0.0005, "loss": 3.5628128051757812, "step": 2787 }, { "epoch": 0.32441238073074236, "grad_norm": 0.9026380777359009, "learning_rate": 0.0005, "loss": 1.9926005601882935, "step": 2788 }, { "epoch": 0.3245287409820805, "grad_norm": 4.310874938964844, "learning_rate": 0.0005, "loss": 3.3826377391815186, "step": 2789 }, { "epoch": 0.32464510123341866, "grad_norm": 1.7723416090011597, "learning_rate": 0.0005, "loss": 2.4639151096343994, "step": 2790 }, { "epoch": 0.3247614614847568, "grad_norm": 1.716262936592102, "learning_rate": 0.0005, "loss": 3.147006034851074, "step": 2791 }, { "epoch": 0.324877821736095, "grad_norm": 1.085876226425171, "learning_rate": 0.0005, "loss": 1.858000636100769, "step": 2792 }, { "epoch": 0.3249941819874331, "grad_norm": 1.3956224918365479, "learning_rate": 0.0005, "loss": 2.7238259315490723, "step": 2793 }, { "epoch": 0.3251105422387712, "grad_norm": 1.8903944492340088, "learning_rate": 0.0005, "loss": 2.8434255123138428, "step": 2794 }, { "epoch": 0.3252269024901094, "grad_norm": 1.462646245956421, "learning_rate": 0.0005, "loss": 3.7472808361053467, "step": 2795 }, { "epoch": 0.32534326274144754, "grad_norm": 0.9962533712387085, "learning_rate": 0.0005, "loss": 2.7065834999084473, "step": 2796 }, { "epoch": 0.3254596229927857, "grad_norm": 0.7003095149993896, "learning_rate": 0.0005, "loss": 1.1886428594589233, "step": 2797 }, { "epoch": 0.3255759832441238, "grad_norm": 1.6805988550186157, "learning_rate": 0.0005, "loss": 3.4876599311828613, "step": 2798 }, { "epoch": 0.32569234349546194, "grad_norm": 2.030129909515381, "learning_rate": 0.0005, "loss": 3.639166831970215, "step": 2799 }, { "epoch": 0.3258087037468001, "grad_norm": 1.6068239212036133, "learning_rate": 0.0005, "loss": 3.379589557647705, "step": 2800 }, { "epoch": 0.32592506399813825, "grad_norm": 1.452810287475586, "learning_rate": 0.0005, "loss": 2.6411752700805664, "step": 2801 }, { "epoch": 0.32604142424947635, "grad_norm": 1.181993007659912, "learning_rate": 0.0005, "loss": 3.4723925590515137, "step": 2802 }, { "epoch": 0.3261577845008145, "grad_norm": 1.6994913816452026, "learning_rate": 0.0005, "loss": 2.3823368549346924, "step": 2803 }, { "epoch": 0.32627414475215266, "grad_norm": 1.4606236219406128, "learning_rate": 0.0005, "loss": 2.735367774963379, "step": 2804 }, { "epoch": 0.3263905050034908, "grad_norm": 1.4436842203140259, "learning_rate": 0.0005, "loss": 1.8679931163787842, "step": 2805 }, { "epoch": 0.32650686525482897, "grad_norm": 1.1098309755325317, "learning_rate": 0.0005, "loss": 2.603278636932373, "step": 2806 }, { "epoch": 0.32662322550616707, "grad_norm": 1.074942946434021, "learning_rate": 0.0005, "loss": 2.1665093898773193, "step": 2807 }, { "epoch": 0.3267395857575052, "grad_norm": 1.0715240240097046, "learning_rate": 0.0005, "loss": 2.2566256523132324, "step": 2808 }, { "epoch": 0.3268559460088434, "grad_norm": 1.282082200050354, "learning_rate": 0.0005, "loss": 3.0622429847717285, "step": 2809 }, { "epoch": 0.32697230626018153, "grad_norm": 1.3025521039962769, "learning_rate": 0.0005, "loss": 2.995208263397217, "step": 2810 }, { "epoch": 0.3270886665115197, "grad_norm": 1.603096604347229, "learning_rate": 0.0005, "loss": 3.4792990684509277, "step": 2811 }, { "epoch": 0.3272050267628578, "grad_norm": 3.1562795639038086, "learning_rate": 0.0005, "loss": 2.7990128993988037, "step": 2812 }, { "epoch": 0.32732138701419594, "grad_norm": 1.8531442880630493, "learning_rate": 0.0005, "loss": 2.8864622116088867, "step": 2813 }, { "epoch": 0.3274377472655341, "grad_norm": 0.8382186889648438, "learning_rate": 0.0005, "loss": 1.7596009969711304, "step": 2814 }, { "epoch": 0.32755410751687225, "grad_norm": 1.0323878526687622, "learning_rate": 0.0005, "loss": 2.288422107696533, "step": 2815 }, { "epoch": 0.3276704677682104, "grad_norm": 1.898597002029419, "learning_rate": 0.0005, "loss": 2.595719337463379, "step": 2816 }, { "epoch": 0.3277868280195485, "grad_norm": 1.2955125570297241, "learning_rate": 0.0005, "loss": 2.64011549949646, "step": 2817 }, { "epoch": 0.32790318827088666, "grad_norm": 1.6683999300003052, "learning_rate": 0.0005, "loss": 3.1606030464172363, "step": 2818 }, { "epoch": 0.3280195485222248, "grad_norm": 1.572951316833496, "learning_rate": 0.0005, "loss": 2.4243314266204834, "step": 2819 }, { "epoch": 0.32813590877356297, "grad_norm": 1.0931260585784912, "learning_rate": 0.0005, "loss": 2.2355051040649414, "step": 2820 }, { "epoch": 0.32825226902490107, "grad_norm": 1.8687793016433716, "learning_rate": 0.0005, "loss": 3.4266116619110107, "step": 2821 }, { "epoch": 0.3283686292762392, "grad_norm": 1.889883041381836, "learning_rate": 0.0005, "loss": 2.422729969024658, "step": 2822 }, { "epoch": 0.3284849895275774, "grad_norm": 1.0995755195617676, "learning_rate": 0.0005, "loss": 2.512910842895508, "step": 2823 }, { "epoch": 0.32860134977891553, "grad_norm": 1.159250020980835, "learning_rate": 0.0005, "loss": 2.18934965133667, "step": 2824 }, { "epoch": 0.3287177100302537, "grad_norm": 1.8553835153579712, "learning_rate": 0.0005, "loss": 2.5010743141174316, "step": 2825 }, { "epoch": 0.3288340702815918, "grad_norm": 1.9726134538650513, "learning_rate": 0.0005, "loss": 3.8682589530944824, "step": 2826 }, { "epoch": 0.32895043053292994, "grad_norm": 1.2067551612854004, "learning_rate": 0.0005, "loss": 3.3901705741882324, "step": 2827 }, { "epoch": 0.3290667907842681, "grad_norm": 0.9696656465530396, "learning_rate": 0.0005, "loss": 2.010073184967041, "step": 2828 }, { "epoch": 0.32918315103560625, "grad_norm": 1.6666064262390137, "learning_rate": 0.0005, "loss": 2.9784088134765625, "step": 2829 }, { "epoch": 0.3292995112869444, "grad_norm": 1.2990270853042603, "learning_rate": 0.0005, "loss": 2.181421995162964, "step": 2830 }, { "epoch": 0.3294158715382825, "grad_norm": 1.2282400131225586, "learning_rate": 0.0005, "loss": 2.718949317932129, "step": 2831 }, { "epoch": 0.32953223178962066, "grad_norm": 1.0623844861984253, "learning_rate": 0.0005, "loss": 2.1283016204833984, "step": 2832 }, { "epoch": 0.3296485920409588, "grad_norm": 1.5657587051391602, "learning_rate": 0.0005, "loss": 2.379387855529785, "step": 2833 }, { "epoch": 0.32976495229229696, "grad_norm": 2.012876033782959, "learning_rate": 0.0005, "loss": 2.454606056213379, "step": 2834 }, { "epoch": 0.3298813125436351, "grad_norm": 1.5148799419403076, "learning_rate": 0.0005, "loss": 2.7991929054260254, "step": 2835 }, { "epoch": 0.3299976727949732, "grad_norm": 1.5306731462478638, "learning_rate": 0.0005, "loss": 2.8792545795440674, "step": 2836 }, { "epoch": 0.3301140330463114, "grad_norm": 1.287975788116455, "learning_rate": 0.0005, "loss": 2.5030012130737305, "step": 2837 }, { "epoch": 0.3302303932976495, "grad_norm": 1.285010576248169, "learning_rate": 0.0005, "loss": 2.910898208618164, "step": 2838 }, { "epoch": 0.3303467535489877, "grad_norm": 1.2351547479629517, "learning_rate": 0.0005, "loss": 3.063960552215576, "step": 2839 }, { "epoch": 0.33046311380032584, "grad_norm": 1.6049305200576782, "learning_rate": 0.0005, "loss": 3.0198636054992676, "step": 2840 }, { "epoch": 0.33057947405166394, "grad_norm": 1.1076563596725464, "learning_rate": 0.0005, "loss": 2.0715575218200684, "step": 2841 }, { "epoch": 0.3306958343030021, "grad_norm": 0.9857180714607239, "learning_rate": 0.0005, "loss": 2.307642936706543, "step": 2842 }, { "epoch": 0.33081219455434024, "grad_norm": 2.148733377456665, "learning_rate": 0.0005, "loss": 3.0658669471740723, "step": 2843 }, { "epoch": 0.3309285548056784, "grad_norm": 1.2772865295410156, "learning_rate": 0.0005, "loss": 1.7670236825942993, "step": 2844 }, { "epoch": 0.3310449150570165, "grad_norm": 2.0172460079193115, "learning_rate": 0.0005, "loss": 2.440373420715332, "step": 2845 }, { "epoch": 0.33116127530835465, "grad_norm": 1.6722975969314575, "learning_rate": 0.0005, "loss": 3.302797794342041, "step": 2846 }, { "epoch": 0.3312776355596928, "grad_norm": 1.5863209962844849, "learning_rate": 0.0005, "loss": 2.9593732357025146, "step": 2847 }, { "epoch": 0.33139399581103096, "grad_norm": 1.844501256942749, "learning_rate": 0.0005, "loss": 3.561051607131958, "step": 2848 }, { "epoch": 0.3315103560623691, "grad_norm": 1.2991217374801636, "learning_rate": 0.0005, "loss": 3.7562496662139893, "step": 2849 }, { "epoch": 0.3316267163137072, "grad_norm": 2.545119047164917, "learning_rate": 0.0005, "loss": 2.677546501159668, "step": 2850 }, { "epoch": 0.33174307656504537, "grad_norm": 2.5674874782562256, "learning_rate": 0.0005, "loss": 3.2128071784973145, "step": 2851 }, { "epoch": 0.3318594368163835, "grad_norm": 1.4581595659255981, "learning_rate": 0.0005, "loss": 3.1153769493103027, "step": 2852 }, { "epoch": 0.3319757970677217, "grad_norm": 1.6446795463562012, "learning_rate": 0.0005, "loss": 2.434072971343994, "step": 2853 }, { "epoch": 0.33209215731905983, "grad_norm": 1.2342830896377563, "learning_rate": 0.0005, "loss": 2.272557258605957, "step": 2854 }, { "epoch": 0.33220851757039793, "grad_norm": 0.8784463405609131, "learning_rate": 0.0005, "loss": 1.8242074251174927, "step": 2855 }, { "epoch": 0.3323248778217361, "grad_norm": 1.2833701372146606, "learning_rate": 0.0005, "loss": 2.225203514099121, "step": 2856 }, { "epoch": 0.33244123807307424, "grad_norm": 1.7819154262542725, "learning_rate": 0.0005, "loss": 2.891244888305664, "step": 2857 }, { "epoch": 0.3325575983244124, "grad_norm": 1.6208535432815552, "learning_rate": 0.0005, "loss": 2.673750638961792, "step": 2858 }, { "epoch": 0.33267395857575055, "grad_norm": 1.6529090404510498, "learning_rate": 0.0005, "loss": 2.840512752532959, "step": 2859 }, { "epoch": 0.33279031882708865, "grad_norm": 1.1896681785583496, "learning_rate": 0.0005, "loss": 1.5689491033554077, "step": 2860 }, { "epoch": 0.3329066790784268, "grad_norm": 1.1071110963821411, "learning_rate": 0.0005, "loss": 2.168527841567993, "step": 2861 }, { "epoch": 0.33302303932976496, "grad_norm": 1.3983420133590698, "learning_rate": 0.0005, "loss": 2.852468490600586, "step": 2862 }, { "epoch": 0.3331393995811031, "grad_norm": 2.799621820449829, "learning_rate": 0.0005, "loss": 3.495389938354492, "step": 2863 }, { "epoch": 0.3332557598324412, "grad_norm": 1.3844236135482788, "learning_rate": 0.0005, "loss": 2.640805721282959, "step": 2864 }, { "epoch": 0.33337212008377937, "grad_norm": 1.9818646907806396, "learning_rate": 0.0005, "loss": 2.347865581512451, "step": 2865 }, { "epoch": 0.3334884803351175, "grad_norm": 1.7879208326339722, "learning_rate": 0.0005, "loss": 2.5062248706817627, "step": 2866 }, { "epoch": 0.3336048405864557, "grad_norm": 2.1741528511047363, "learning_rate": 0.0005, "loss": 2.449131965637207, "step": 2867 }, { "epoch": 0.33372120083779383, "grad_norm": 1.4019737243652344, "learning_rate": 0.0005, "loss": 2.28448224067688, "step": 2868 }, { "epoch": 0.33383756108913193, "grad_norm": 1.5710723400115967, "learning_rate": 0.0005, "loss": 2.265650987625122, "step": 2869 }, { "epoch": 0.3339539213404701, "grad_norm": 1.560561180114746, "learning_rate": 0.0005, "loss": 2.6908559799194336, "step": 2870 }, { "epoch": 0.33407028159180824, "grad_norm": 1.3046687841415405, "learning_rate": 0.0005, "loss": 2.9764890670776367, "step": 2871 }, { "epoch": 0.3341866418431464, "grad_norm": 1.2960386276245117, "learning_rate": 0.0005, "loss": 3.36417293548584, "step": 2872 }, { "epoch": 0.33430300209448455, "grad_norm": 1.6370900869369507, "learning_rate": 0.0005, "loss": 2.379417896270752, "step": 2873 }, { "epoch": 0.33441936234582265, "grad_norm": 0.6214933395385742, "learning_rate": 0.0005, "loss": 1.1057665348052979, "step": 2874 }, { "epoch": 0.3345357225971608, "grad_norm": 1.2923089265823364, "learning_rate": 0.0005, "loss": 1.5704307556152344, "step": 2875 }, { "epoch": 0.33465208284849896, "grad_norm": 1.544688105583191, "learning_rate": 0.0005, "loss": 1.4515628814697266, "step": 2876 }, { "epoch": 0.3347684430998371, "grad_norm": 2.3899097442626953, "learning_rate": 0.0005, "loss": 2.6480588912963867, "step": 2877 }, { "epoch": 0.33488480335117526, "grad_norm": 1.7117280960083008, "learning_rate": 0.0005, "loss": 2.2440953254699707, "step": 2878 }, { "epoch": 0.33500116360251336, "grad_norm": 1.8162496089935303, "learning_rate": 0.0005, "loss": 3.1255149841308594, "step": 2879 }, { "epoch": 0.3351175238538515, "grad_norm": 2.103555917739868, "learning_rate": 0.0005, "loss": 2.4500985145568848, "step": 2880 }, { "epoch": 0.3352338841051897, "grad_norm": 1.6446324586868286, "learning_rate": 0.0005, "loss": 2.6399567127227783, "step": 2881 }, { "epoch": 0.3353502443565278, "grad_norm": 1.304978609085083, "learning_rate": 0.0005, "loss": 2.7298593521118164, "step": 2882 }, { "epoch": 0.3354666046078659, "grad_norm": 1.7910853624343872, "learning_rate": 0.0005, "loss": 3.2920241355895996, "step": 2883 }, { "epoch": 0.3355829648592041, "grad_norm": 1.432845115661621, "learning_rate": 0.0005, "loss": 3.0726094245910645, "step": 2884 }, { "epoch": 0.33569932511054223, "grad_norm": 1.5736920833587646, "learning_rate": 0.0005, "loss": 3.5056755542755127, "step": 2885 }, { "epoch": 0.3358156853618804, "grad_norm": 1.430181860923767, "learning_rate": 0.0005, "loss": 2.51558256149292, "step": 2886 }, { "epoch": 0.33593204561321854, "grad_norm": 1.0992467403411865, "learning_rate": 0.0005, "loss": 2.4737136363983154, "step": 2887 }, { "epoch": 0.33604840586455664, "grad_norm": 0.9051283001899719, "learning_rate": 0.0005, "loss": 2.475252151489258, "step": 2888 }, { "epoch": 0.3361647661158948, "grad_norm": 1.4668736457824707, "learning_rate": 0.0005, "loss": 3.279414176940918, "step": 2889 }, { "epoch": 0.33628112636723295, "grad_norm": 1.25912606716156, "learning_rate": 0.0005, "loss": 2.670360803604126, "step": 2890 }, { "epoch": 0.3363974866185711, "grad_norm": 2.330204486846924, "learning_rate": 0.0005, "loss": 3.8524270057678223, "step": 2891 }, { "epoch": 0.33651384686990926, "grad_norm": 1.9864660501480103, "learning_rate": 0.0005, "loss": 3.642017364501953, "step": 2892 }, { "epoch": 0.33663020712124736, "grad_norm": 1.3472670316696167, "learning_rate": 0.0005, "loss": 1.995323896408081, "step": 2893 }, { "epoch": 0.3367465673725855, "grad_norm": 1.1032425165176392, "learning_rate": 0.0005, "loss": 1.786393642425537, "step": 2894 }, { "epoch": 0.33686292762392367, "grad_norm": 0.6894261837005615, "learning_rate": 0.0005, "loss": 1.5946160554885864, "step": 2895 }, { "epoch": 0.3369792878752618, "grad_norm": 0.8875985741615295, "learning_rate": 0.0005, "loss": 2.4375529289245605, "step": 2896 }, { "epoch": 0.3370956481266, "grad_norm": 1.1174476146697998, "learning_rate": 0.0005, "loss": 2.716517925262451, "step": 2897 }, { "epoch": 0.3372120083779381, "grad_norm": 1.3502695560455322, "learning_rate": 0.0005, "loss": 2.864655017852783, "step": 2898 }, { "epoch": 0.33732836862927623, "grad_norm": 1.3580279350280762, "learning_rate": 0.0005, "loss": 1.761278510093689, "step": 2899 }, { "epoch": 0.3374447288806144, "grad_norm": 1.0366215705871582, "learning_rate": 0.0005, "loss": 2.6888275146484375, "step": 2900 }, { "epoch": 0.33756108913195254, "grad_norm": 1.124558925628662, "learning_rate": 0.0005, "loss": 2.582498073577881, "step": 2901 }, { "epoch": 0.3376774493832907, "grad_norm": 1.1829286813735962, "learning_rate": 0.0005, "loss": 3.8433706760406494, "step": 2902 }, { "epoch": 0.3377938096346288, "grad_norm": 1.4785802364349365, "learning_rate": 0.0005, "loss": 3.2235207557678223, "step": 2903 }, { "epoch": 0.33791016988596695, "grad_norm": 1.0855166912078857, "learning_rate": 0.0005, "loss": 3.249429941177368, "step": 2904 }, { "epoch": 0.3380265301373051, "grad_norm": 1.8304834365844727, "learning_rate": 0.0005, "loss": 3.4858808517456055, "step": 2905 }, { "epoch": 0.33814289038864326, "grad_norm": 1.4619642496109009, "learning_rate": 0.0005, "loss": 2.988734245300293, "step": 2906 }, { "epoch": 0.33825925063998136, "grad_norm": 1.5678050518035889, "learning_rate": 0.0005, "loss": 2.0641732215881348, "step": 2907 }, { "epoch": 0.3383756108913195, "grad_norm": 1.0468074083328247, "learning_rate": 0.0005, "loss": 1.7403926849365234, "step": 2908 }, { "epoch": 0.33849197114265767, "grad_norm": 1.4521793127059937, "learning_rate": 0.0005, "loss": 2.6058802604675293, "step": 2909 }, { "epoch": 0.3386083313939958, "grad_norm": 2.0861055850982666, "learning_rate": 0.0005, "loss": 2.8547897338867188, "step": 2910 }, { "epoch": 0.338724691645334, "grad_norm": 4.064158916473389, "learning_rate": 0.0005, "loss": 2.4632649421691895, "step": 2911 }, { "epoch": 0.3388410518966721, "grad_norm": 1.3528908491134644, "learning_rate": 0.0005, "loss": 2.8684885501861572, "step": 2912 }, { "epoch": 0.33895741214801023, "grad_norm": 0.9958127737045288, "learning_rate": 0.0005, "loss": 1.982079267501831, "step": 2913 }, { "epoch": 0.3390737723993484, "grad_norm": 1.6661376953125, "learning_rate": 0.0005, "loss": 2.7018842697143555, "step": 2914 }, { "epoch": 0.33919013265068654, "grad_norm": 1.3360669612884521, "learning_rate": 0.0005, "loss": 2.3657455444335938, "step": 2915 }, { "epoch": 0.3393064929020247, "grad_norm": 1.4571404457092285, "learning_rate": 0.0005, "loss": 2.657528877258301, "step": 2916 }, { "epoch": 0.3394228531533628, "grad_norm": 1.544730305671692, "learning_rate": 0.0005, "loss": 3.0376129150390625, "step": 2917 }, { "epoch": 0.33953921340470095, "grad_norm": 2.0095038414001465, "learning_rate": 0.0005, "loss": 2.0336570739746094, "step": 2918 }, { "epoch": 0.3396555736560391, "grad_norm": 1.9854782819747925, "learning_rate": 0.0005, "loss": 2.9180569648742676, "step": 2919 }, { "epoch": 0.33977193390737725, "grad_norm": 3.925788164138794, "learning_rate": 0.0005, "loss": 3.848236083984375, "step": 2920 }, { "epoch": 0.3398882941587154, "grad_norm": 1.8800134658813477, "learning_rate": 0.0005, "loss": 2.680290699005127, "step": 2921 }, { "epoch": 0.3400046544100535, "grad_norm": 1.188132882118225, "learning_rate": 0.0005, "loss": 2.3637475967407227, "step": 2922 }, { "epoch": 0.34012101466139166, "grad_norm": 1.0945101976394653, "learning_rate": 0.0005, "loss": 1.7846801280975342, "step": 2923 }, { "epoch": 0.3402373749127298, "grad_norm": 1.2039880752563477, "learning_rate": 0.0005, "loss": 2.8828542232513428, "step": 2924 }, { "epoch": 0.34035373516406797, "grad_norm": 1.9092034101486206, "learning_rate": 0.0005, "loss": 2.825481653213501, "step": 2925 }, { "epoch": 0.34047009541540607, "grad_norm": 1.21201491355896, "learning_rate": 0.0005, "loss": 1.502392053604126, "step": 2926 }, { "epoch": 0.3405864556667442, "grad_norm": 1.6543183326721191, "learning_rate": 0.0005, "loss": 2.814441680908203, "step": 2927 }, { "epoch": 0.3407028159180824, "grad_norm": 1.3422067165374756, "learning_rate": 0.0005, "loss": 2.8564743995666504, "step": 2928 }, { "epoch": 0.34081917616942053, "grad_norm": 1.8472952842712402, "learning_rate": 0.0005, "loss": 2.8743433952331543, "step": 2929 }, { "epoch": 0.3409355364207587, "grad_norm": 1.047316551208496, "learning_rate": 0.0005, "loss": 1.5712780952453613, "step": 2930 }, { "epoch": 0.3410518966720968, "grad_norm": 0.9632250070571899, "learning_rate": 0.0005, "loss": 1.6859527826309204, "step": 2931 }, { "epoch": 0.34116825692343494, "grad_norm": 1.5218621492385864, "learning_rate": 0.0005, "loss": 2.5814602375030518, "step": 2932 }, { "epoch": 0.3412846171747731, "grad_norm": 1.1557586193084717, "learning_rate": 0.0005, "loss": 2.5148305892944336, "step": 2933 }, { "epoch": 0.34140097742611125, "grad_norm": 2.3094558715820312, "learning_rate": 0.0005, "loss": 2.9872288703918457, "step": 2934 }, { "epoch": 0.3415173376774494, "grad_norm": 1.5847439765930176, "learning_rate": 0.0005, "loss": 2.49344539642334, "step": 2935 }, { "epoch": 0.3416336979287875, "grad_norm": 1.6706825494766235, "learning_rate": 0.0005, "loss": 2.341388702392578, "step": 2936 }, { "epoch": 0.34175005818012566, "grad_norm": 1.4062751531600952, "learning_rate": 0.0005, "loss": 2.1513688564300537, "step": 2937 }, { "epoch": 0.3418664184314638, "grad_norm": 1.7478421926498413, "learning_rate": 0.0005, "loss": 2.5704433917999268, "step": 2938 }, { "epoch": 0.34198277868280197, "grad_norm": 1.2973188161849976, "learning_rate": 0.0005, "loss": 3.0232186317443848, "step": 2939 }, { "epoch": 0.3420991389341401, "grad_norm": 1.093546986579895, "learning_rate": 0.0005, "loss": 2.0287604331970215, "step": 2940 }, { "epoch": 0.3422154991854782, "grad_norm": 1.8917518854141235, "learning_rate": 0.0005, "loss": 2.3899474143981934, "step": 2941 }, { "epoch": 0.3423318594368164, "grad_norm": 1.2180324792861938, "learning_rate": 0.0005, "loss": 1.7609930038452148, "step": 2942 }, { "epoch": 0.34244821968815453, "grad_norm": 1.4107860326766968, "learning_rate": 0.0005, "loss": 2.2487685680389404, "step": 2943 }, { "epoch": 0.3425645799394927, "grad_norm": 2.1507415771484375, "learning_rate": 0.0005, "loss": 3.069791078567505, "step": 2944 }, { "epoch": 0.3426809401908308, "grad_norm": 1.2666289806365967, "learning_rate": 0.0005, "loss": 2.0909032821655273, "step": 2945 }, { "epoch": 0.34279730044216894, "grad_norm": 1.0024373531341553, "learning_rate": 0.0005, "loss": 2.029921293258667, "step": 2946 }, { "epoch": 0.3429136606935071, "grad_norm": 0.8922961354255676, "learning_rate": 0.0005, "loss": 2.53261661529541, "step": 2947 }, { "epoch": 0.34303002094484525, "grad_norm": 1.516330599784851, "learning_rate": 0.0005, "loss": 2.587677240371704, "step": 2948 }, { "epoch": 0.3431463811961834, "grad_norm": 1.3750272989273071, "learning_rate": 0.0005, "loss": 2.012977123260498, "step": 2949 }, { "epoch": 0.3432627414475215, "grad_norm": 1.685670256614685, "learning_rate": 0.0005, "loss": 2.363251209259033, "step": 2950 }, { "epoch": 0.34337910169885966, "grad_norm": 0.809118926525116, "learning_rate": 0.0005, "loss": 1.8410810232162476, "step": 2951 }, { "epoch": 0.3434954619501978, "grad_norm": 1.0503942966461182, "learning_rate": 0.0005, "loss": 2.3877828121185303, "step": 2952 }, { "epoch": 0.34361182220153597, "grad_norm": 1.6126598119735718, "learning_rate": 0.0005, "loss": 3.39559006690979, "step": 2953 }, { "epoch": 0.3437281824528741, "grad_norm": 1.6962260007858276, "learning_rate": 0.0005, "loss": 2.745046615600586, "step": 2954 }, { "epoch": 0.3438445427042122, "grad_norm": 2.0560104846954346, "learning_rate": 0.0005, "loss": 2.625004291534424, "step": 2955 }, { "epoch": 0.3439609029555504, "grad_norm": 1.4277141094207764, "learning_rate": 0.0005, "loss": 2.8202767372131348, "step": 2956 }, { "epoch": 0.34407726320688853, "grad_norm": 1.0009845495224, "learning_rate": 0.0005, "loss": 2.2462637424468994, "step": 2957 }, { "epoch": 0.3441936234582267, "grad_norm": 1.8820592164993286, "learning_rate": 0.0005, "loss": 3.1096277236938477, "step": 2958 }, { "epoch": 0.34430998370956484, "grad_norm": 1.8607518672943115, "learning_rate": 0.0005, "loss": 2.9097280502319336, "step": 2959 }, { "epoch": 0.34442634396090294, "grad_norm": 1.0905319452285767, "learning_rate": 0.0005, "loss": 1.6704540252685547, "step": 2960 }, { "epoch": 0.3445427042122411, "grad_norm": 1.4704850912094116, "learning_rate": 0.0005, "loss": 3.475127935409546, "step": 2961 }, { "epoch": 0.34465906446357925, "grad_norm": 1.8093321323394775, "learning_rate": 0.0005, "loss": 3.0328845977783203, "step": 2962 }, { "epoch": 0.3447754247149174, "grad_norm": 2.014650583267212, "learning_rate": 0.0005, "loss": 3.3983941078186035, "step": 2963 }, { "epoch": 0.34489178496625555, "grad_norm": 1.1026443243026733, "learning_rate": 0.0005, "loss": 1.7854214906692505, "step": 2964 }, { "epoch": 0.34500814521759365, "grad_norm": 0.7875393629074097, "learning_rate": 0.0005, "loss": 1.6579363346099854, "step": 2965 }, { "epoch": 0.3451245054689318, "grad_norm": 1.3083842992782593, "learning_rate": 0.0005, "loss": 2.6111958026885986, "step": 2966 }, { "epoch": 0.34524086572026996, "grad_norm": 1.0008652210235596, "learning_rate": 0.0005, "loss": 2.067040205001831, "step": 2967 }, { "epoch": 0.3453572259716081, "grad_norm": 1.6238112449645996, "learning_rate": 0.0005, "loss": 2.9308266639709473, "step": 2968 }, { "epoch": 0.3454735862229462, "grad_norm": 1.9766372442245483, "learning_rate": 0.0005, "loss": 3.5341854095458984, "step": 2969 }, { "epoch": 0.34558994647428437, "grad_norm": 2.717575788497925, "learning_rate": 0.0005, "loss": 3.3207473754882812, "step": 2970 }, { "epoch": 0.3457063067256225, "grad_norm": 0.7414844632148743, "learning_rate": 0.0005, "loss": 1.575366735458374, "step": 2971 }, { "epoch": 0.3458226669769607, "grad_norm": 1.0225918292999268, "learning_rate": 0.0005, "loss": 2.13913893699646, "step": 2972 }, { "epoch": 0.34593902722829883, "grad_norm": 1.1538358926773071, "learning_rate": 0.0005, "loss": 1.9435033798217773, "step": 2973 }, { "epoch": 0.34605538747963693, "grad_norm": 2.408158540725708, "learning_rate": 0.0005, "loss": 3.118253707885742, "step": 2974 }, { "epoch": 0.3461717477309751, "grad_norm": 1.5970255136489868, "learning_rate": 0.0005, "loss": 2.5011391639709473, "step": 2975 }, { "epoch": 0.34628810798231324, "grad_norm": 1.6854162216186523, "learning_rate": 0.0005, "loss": 3.0599112510681152, "step": 2976 }, { "epoch": 0.3464044682336514, "grad_norm": 1.5270098447799683, "learning_rate": 0.0005, "loss": 3.1917309761047363, "step": 2977 }, { "epoch": 0.34652082848498955, "grad_norm": 1.8015533685684204, "learning_rate": 0.0005, "loss": 3.3456387519836426, "step": 2978 }, { "epoch": 0.34663718873632765, "grad_norm": 1.2841064929962158, "learning_rate": 0.0005, "loss": 2.6989221572875977, "step": 2979 }, { "epoch": 0.3467535489876658, "grad_norm": 0.9392946362495422, "learning_rate": 0.0005, "loss": 2.6175427436828613, "step": 2980 }, { "epoch": 0.34686990923900396, "grad_norm": 3.568803071975708, "learning_rate": 0.0005, "loss": 2.799546957015991, "step": 2981 }, { "epoch": 0.3469862694903421, "grad_norm": 1.6655851602554321, "learning_rate": 0.0005, "loss": 3.2728967666625977, "step": 2982 }, { "epoch": 0.34710262974168027, "grad_norm": 0.39786267280578613, "learning_rate": 0.0005, "loss": 1.0187511444091797, "step": 2983 }, { "epoch": 0.34721898999301837, "grad_norm": 1.8484247922897339, "learning_rate": 0.0005, "loss": 2.87412166595459, "step": 2984 }, { "epoch": 0.3473353502443565, "grad_norm": 1.5981767177581787, "learning_rate": 0.0005, "loss": 2.700479030609131, "step": 2985 }, { "epoch": 0.3474517104956947, "grad_norm": 1.021470546722412, "learning_rate": 0.0005, "loss": 2.080101251602173, "step": 2986 }, { "epoch": 0.34756807074703283, "grad_norm": 1.2716999053955078, "learning_rate": 0.0005, "loss": 2.1129989624023438, "step": 2987 }, { "epoch": 0.34768443099837093, "grad_norm": 1.0076149702072144, "learning_rate": 0.0005, "loss": 1.757420301437378, "step": 2988 }, { "epoch": 0.3478007912497091, "grad_norm": 1.4586327075958252, "learning_rate": 0.0005, "loss": 3.0373144149780273, "step": 2989 }, { "epoch": 0.34791715150104724, "grad_norm": 0.998043417930603, "learning_rate": 0.0005, "loss": 1.7671722173690796, "step": 2990 }, { "epoch": 0.3480335117523854, "grad_norm": 1.7384237051010132, "learning_rate": 0.0005, "loss": 2.376800537109375, "step": 2991 }, { "epoch": 0.34814987200372355, "grad_norm": 1.2216213941574097, "learning_rate": 0.0005, "loss": 2.22506046295166, "step": 2992 }, { "epoch": 0.34826623225506165, "grad_norm": 2.110884666442871, "learning_rate": 0.0005, "loss": 3.223494529724121, "step": 2993 }, { "epoch": 0.3483825925063998, "grad_norm": 1.7431440353393555, "learning_rate": 0.0005, "loss": 2.2658567428588867, "step": 2994 }, { "epoch": 0.34849895275773796, "grad_norm": 0.8443347811698914, "learning_rate": 0.0005, "loss": 2.5266802310943604, "step": 2995 }, { "epoch": 0.3486153130090761, "grad_norm": 1.8561553955078125, "learning_rate": 0.0005, "loss": 2.894610643386841, "step": 2996 }, { "epoch": 0.34873167326041427, "grad_norm": 1.2443711757659912, "learning_rate": 0.0005, "loss": 2.253227710723877, "step": 2997 }, { "epoch": 0.34884803351175236, "grad_norm": 1.1391549110412598, "learning_rate": 0.0005, "loss": 2.5525076389312744, "step": 2998 }, { "epoch": 0.3489643937630905, "grad_norm": 1.9862288236618042, "learning_rate": 0.0005, "loss": 2.04661226272583, "step": 2999 }, { "epoch": 0.3490807540144287, "grad_norm": 1.3669002056121826, "learning_rate": 0.0005, "loss": 2.948152542114258, "step": 3000 }, { "epoch": 0.34919711426576683, "grad_norm": 1.2040235996246338, "learning_rate": 0.0005, "loss": 1.7628713846206665, "step": 3001 }, { "epoch": 0.349313474517105, "grad_norm": 1.4237653017044067, "learning_rate": 0.0005, "loss": 2.5484466552734375, "step": 3002 }, { "epoch": 0.3494298347684431, "grad_norm": 1.7230074405670166, "learning_rate": 0.0005, "loss": 3.0358521938323975, "step": 3003 }, { "epoch": 0.34954619501978124, "grad_norm": 1.1154574155807495, "learning_rate": 0.0005, "loss": 2.316718578338623, "step": 3004 }, { "epoch": 0.3496625552711194, "grad_norm": 1.96422278881073, "learning_rate": 0.0005, "loss": 2.4674925804138184, "step": 3005 }, { "epoch": 0.34977891552245755, "grad_norm": 1.925690770149231, "learning_rate": 0.0005, "loss": 3.7361879348754883, "step": 3006 }, { "epoch": 0.34989527577379564, "grad_norm": 1.4150640964508057, "learning_rate": 0.0005, "loss": 2.340567111968994, "step": 3007 }, { "epoch": 0.3500116360251338, "grad_norm": 0.9099397659301758, "learning_rate": 0.0005, "loss": 1.9478461742401123, "step": 3008 }, { "epoch": 0.35012799627647195, "grad_norm": 1.1375606060028076, "learning_rate": 0.0005, "loss": 2.078488826751709, "step": 3009 }, { "epoch": 0.3502443565278101, "grad_norm": 1.1717045307159424, "learning_rate": 0.0005, "loss": 2.628192663192749, "step": 3010 }, { "epoch": 0.35036071677914826, "grad_norm": 1.386407732963562, "learning_rate": 0.0005, "loss": 2.3421592712402344, "step": 3011 }, { "epoch": 0.35047707703048636, "grad_norm": 1.3287742137908936, "learning_rate": 0.0005, "loss": 3.616209030151367, "step": 3012 }, { "epoch": 0.3505934372818245, "grad_norm": 0.8424655795097351, "learning_rate": 0.0005, "loss": 1.9509861469268799, "step": 3013 }, { "epoch": 0.35070979753316267, "grad_norm": 1.5231273174285889, "learning_rate": 0.0005, "loss": 2.426969051361084, "step": 3014 }, { "epoch": 0.3508261577845008, "grad_norm": 1.3960320949554443, "learning_rate": 0.0005, "loss": 2.2875280380249023, "step": 3015 }, { "epoch": 0.350942518035839, "grad_norm": 1.888317584991455, "learning_rate": 0.0005, "loss": 2.9634182453155518, "step": 3016 }, { "epoch": 0.3510588782871771, "grad_norm": 1.890015721321106, "learning_rate": 0.0005, "loss": 2.516666889190674, "step": 3017 }, { "epoch": 0.35117523853851523, "grad_norm": 1.218410849571228, "learning_rate": 0.0005, "loss": 2.7453606128692627, "step": 3018 }, { "epoch": 0.3512915987898534, "grad_norm": 1.6589404344558716, "learning_rate": 0.0005, "loss": 3.0091981887817383, "step": 3019 }, { "epoch": 0.35140795904119154, "grad_norm": 15.203161239624023, "learning_rate": 0.0005, "loss": 3.6607306003570557, "step": 3020 }, { "epoch": 0.3515243192925297, "grad_norm": 0.7026939392089844, "learning_rate": 0.0005, "loss": 1.698649525642395, "step": 3021 }, { "epoch": 0.3516406795438678, "grad_norm": 1.451849102973938, "learning_rate": 0.0005, "loss": 1.9517972469329834, "step": 3022 }, { "epoch": 0.35175703979520595, "grad_norm": 1.2297614812850952, "learning_rate": 0.0005, "loss": 3.1033191680908203, "step": 3023 }, { "epoch": 0.3518734000465441, "grad_norm": 1.0284736156463623, "learning_rate": 0.0005, "loss": 2.573246717453003, "step": 3024 }, { "epoch": 0.35198976029788226, "grad_norm": 1.1512006521224976, "learning_rate": 0.0005, "loss": 2.1328635215759277, "step": 3025 }, { "epoch": 0.3521061205492204, "grad_norm": 1.4276303052902222, "learning_rate": 0.0005, "loss": 3.300783157348633, "step": 3026 }, { "epoch": 0.3522224808005585, "grad_norm": 0.5015394687652588, "learning_rate": 0.0005, "loss": 1.2084758281707764, "step": 3027 }, { "epoch": 0.35233884105189667, "grad_norm": 1.4822404384613037, "learning_rate": 0.0005, "loss": 1.9274852275848389, "step": 3028 }, { "epoch": 0.3524552013032348, "grad_norm": 1.0382603406906128, "learning_rate": 0.0005, "loss": 2.831270933151245, "step": 3029 }, { "epoch": 0.352571561554573, "grad_norm": 1.0264822244644165, "learning_rate": 0.0005, "loss": 1.9495770931243896, "step": 3030 }, { "epoch": 0.3526879218059111, "grad_norm": 1.870711088180542, "learning_rate": 0.0005, "loss": 2.6475236415863037, "step": 3031 }, { "epoch": 0.35280428205724923, "grad_norm": 2.3391826152801514, "learning_rate": 0.0005, "loss": 2.3440089225769043, "step": 3032 }, { "epoch": 0.3529206423085874, "grad_norm": 1.4966154098510742, "learning_rate": 0.0005, "loss": 2.690262794494629, "step": 3033 }, { "epoch": 0.35303700255992554, "grad_norm": 1.0604422092437744, "learning_rate": 0.0005, "loss": 2.1999597549438477, "step": 3034 }, { "epoch": 0.3531533628112637, "grad_norm": 1.5636577606201172, "learning_rate": 0.0005, "loss": 2.446199417114258, "step": 3035 }, { "epoch": 0.3532697230626018, "grad_norm": 1.2286434173583984, "learning_rate": 0.0005, "loss": 2.1186130046844482, "step": 3036 }, { "epoch": 0.35338608331393995, "grad_norm": 1.7589337825775146, "learning_rate": 0.0005, "loss": 3.0397558212280273, "step": 3037 }, { "epoch": 0.3535024435652781, "grad_norm": 1.8177653551101685, "learning_rate": 0.0005, "loss": 2.587388515472412, "step": 3038 }, { "epoch": 0.35361880381661626, "grad_norm": 1.851494550704956, "learning_rate": 0.0005, "loss": 3.0548605918884277, "step": 3039 }, { "epoch": 0.3537351640679544, "grad_norm": 1.4174697399139404, "learning_rate": 0.0005, "loss": 2.5628018379211426, "step": 3040 }, { "epoch": 0.3538515243192925, "grad_norm": 1.7550221681594849, "learning_rate": 0.0005, "loss": 2.849271774291992, "step": 3041 }, { "epoch": 0.35396788457063066, "grad_norm": 2.318831205368042, "learning_rate": 0.0005, "loss": 2.907066822052002, "step": 3042 }, { "epoch": 0.3540842448219688, "grad_norm": 1.6663320064544678, "learning_rate": 0.0005, "loss": 3.1352124214172363, "step": 3043 }, { "epoch": 0.354200605073307, "grad_norm": 1.6014577150344849, "learning_rate": 0.0005, "loss": 2.9012928009033203, "step": 3044 }, { "epoch": 0.35431696532464513, "grad_norm": 1.8002325296401978, "learning_rate": 0.0005, "loss": 3.2477707862854004, "step": 3045 }, { "epoch": 0.3544333255759832, "grad_norm": 2.076986074447632, "learning_rate": 0.0005, "loss": 2.503159999847412, "step": 3046 }, { "epoch": 0.3545496858273214, "grad_norm": 1.3765915632247925, "learning_rate": 0.0005, "loss": 3.2396817207336426, "step": 3047 }, { "epoch": 0.35466604607865954, "grad_norm": 1.164252519607544, "learning_rate": 0.0005, "loss": 2.99468994140625, "step": 3048 }, { "epoch": 0.3547824063299977, "grad_norm": 1.3281246423721313, "learning_rate": 0.0005, "loss": 3.2799758911132812, "step": 3049 }, { "epoch": 0.3548987665813358, "grad_norm": 1.5338579416275024, "learning_rate": 0.0005, "loss": 2.1964609622955322, "step": 3050 }, { "epoch": 0.35501512683267394, "grad_norm": 1.3188880681991577, "learning_rate": 0.0005, "loss": 2.525714635848999, "step": 3051 }, { "epoch": 0.3551314870840121, "grad_norm": 1.4012632369995117, "learning_rate": 0.0005, "loss": 2.72499942779541, "step": 3052 }, { "epoch": 0.35524784733535025, "grad_norm": 1.2067627906799316, "learning_rate": 0.0005, "loss": 2.186424732208252, "step": 3053 }, { "epoch": 0.3553642075866884, "grad_norm": 0.9627065062522888, "learning_rate": 0.0005, "loss": 2.197223663330078, "step": 3054 }, { "epoch": 0.3554805678380265, "grad_norm": 1.241629719734192, "learning_rate": 0.0005, "loss": 1.9185073375701904, "step": 3055 }, { "epoch": 0.35559692808936466, "grad_norm": 1.0796494483947754, "learning_rate": 0.0005, "loss": 2.9748191833496094, "step": 3056 }, { "epoch": 0.3557132883407028, "grad_norm": 1.2651246786117554, "learning_rate": 0.0005, "loss": 2.4327521324157715, "step": 3057 }, { "epoch": 0.35582964859204097, "grad_norm": 1.5054744482040405, "learning_rate": 0.0005, "loss": 2.5522985458374023, "step": 3058 }, { "epoch": 0.3559460088433791, "grad_norm": 0.9516842365264893, "learning_rate": 0.0005, "loss": 2.114375591278076, "step": 3059 }, { "epoch": 0.3560623690947172, "grad_norm": 1.5034960508346558, "learning_rate": 0.0005, "loss": 3.384666919708252, "step": 3060 }, { "epoch": 0.3561787293460554, "grad_norm": 1.319442629814148, "learning_rate": 0.0005, "loss": 2.9777369499206543, "step": 3061 }, { "epoch": 0.35629508959739353, "grad_norm": 4.426334857940674, "learning_rate": 0.0005, "loss": 3.264613628387451, "step": 3062 }, { "epoch": 0.3564114498487317, "grad_norm": 1.1562718152999878, "learning_rate": 0.0005, "loss": 2.681633949279785, "step": 3063 }, { "epoch": 0.35652781010006984, "grad_norm": 2.2017905712127686, "learning_rate": 0.0005, "loss": 2.364470958709717, "step": 3064 }, { "epoch": 0.35664417035140794, "grad_norm": 1.236924409866333, "learning_rate": 0.0005, "loss": 2.429760217666626, "step": 3065 }, { "epoch": 0.3567605306027461, "grad_norm": 1.9834462404251099, "learning_rate": 0.0005, "loss": 2.4115917682647705, "step": 3066 }, { "epoch": 0.35687689085408425, "grad_norm": 1.6693408489227295, "learning_rate": 0.0005, "loss": 2.463613510131836, "step": 3067 }, { "epoch": 0.3569932511054224, "grad_norm": 1.7513805627822876, "learning_rate": 0.0005, "loss": 2.115003824234009, "step": 3068 }, { "epoch": 0.3571096113567605, "grad_norm": 1.807765245437622, "learning_rate": 0.0005, "loss": 2.9850258827209473, "step": 3069 }, { "epoch": 0.35722597160809866, "grad_norm": 1.3950563669204712, "learning_rate": 0.0005, "loss": 3.051558494567871, "step": 3070 }, { "epoch": 0.3573423318594368, "grad_norm": 1.8692152500152588, "learning_rate": 0.0005, "loss": 3.4496452808380127, "step": 3071 }, { "epoch": 0.35745869211077497, "grad_norm": 1.3726673126220703, "learning_rate": 0.0005, "loss": 2.770068645477295, "step": 3072 }, { "epoch": 0.3575750523621131, "grad_norm": 2.1729938983917236, "learning_rate": 0.0005, "loss": 3.3099210262298584, "step": 3073 }, { "epoch": 0.3576914126134512, "grad_norm": 2.2374303340911865, "learning_rate": 0.0005, "loss": 2.667543888092041, "step": 3074 }, { "epoch": 0.3578077728647894, "grad_norm": 0.7193471789360046, "learning_rate": 0.0005, "loss": 1.8144699335098267, "step": 3075 }, { "epoch": 0.35792413311612753, "grad_norm": 1.4941900968551636, "learning_rate": 0.0005, "loss": 2.492218494415283, "step": 3076 }, { "epoch": 0.3580404933674657, "grad_norm": 1.149609923362732, "learning_rate": 0.0005, "loss": 1.832329511642456, "step": 3077 }, { "epoch": 0.35815685361880384, "grad_norm": 3.0229709148406982, "learning_rate": 0.0005, "loss": 2.8418095111846924, "step": 3078 }, { "epoch": 0.35827321387014194, "grad_norm": 2.021929979324341, "learning_rate": 0.0005, "loss": 3.966431140899658, "step": 3079 }, { "epoch": 0.3583895741214801, "grad_norm": 1.3220722675323486, "learning_rate": 0.0005, "loss": 3.3719193935394287, "step": 3080 }, { "epoch": 0.35850593437281825, "grad_norm": 1.3948800563812256, "learning_rate": 0.0005, "loss": 2.6744987964630127, "step": 3081 }, { "epoch": 0.3586222946241564, "grad_norm": 1.4655275344848633, "learning_rate": 0.0005, "loss": 3.0903830528259277, "step": 3082 }, { "epoch": 0.35873865487549456, "grad_norm": 1.8566228151321411, "learning_rate": 0.0005, "loss": 2.906064510345459, "step": 3083 }, { "epoch": 0.35885501512683265, "grad_norm": 2.0068650245666504, "learning_rate": 0.0005, "loss": 2.8548169136047363, "step": 3084 }, { "epoch": 0.3589713753781708, "grad_norm": 1.443968415260315, "learning_rate": 0.0005, "loss": 2.8030879497528076, "step": 3085 }, { "epoch": 0.35908773562950896, "grad_norm": 1.361948013305664, "learning_rate": 0.0005, "loss": 2.96710205078125, "step": 3086 }, { "epoch": 0.3592040958808471, "grad_norm": 1.1117860078811646, "learning_rate": 0.0005, "loss": 1.974592924118042, "step": 3087 }, { "epoch": 0.3593204561321853, "grad_norm": 1.600521206855774, "learning_rate": 0.0005, "loss": 2.447422981262207, "step": 3088 }, { "epoch": 0.35943681638352337, "grad_norm": 1.2970507144927979, "learning_rate": 0.0005, "loss": 2.0793564319610596, "step": 3089 }, { "epoch": 0.3595531766348615, "grad_norm": 1.0663632154464722, "learning_rate": 0.0005, "loss": 2.7150356769561768, "step": 3090 }, { "epoch": 0.3596695368861997, "grad_norm": 1.1647123098373413, "learning_rate": 0.0005, "loss": 2.1399447917938232, "step": 3091 }, { "epoch": 0.35978589713753784, "grad_norm": 1.5958245992660522, "learning_rate": 0.0005, "loss": 1.9472153186798096, "step": 3092 }, { "epoch": 0.35990225738887593, "grad_norm": 1.783907413482666, "learning_rate": 0.0005, "loss": 1.8758773803710938, "step": 3093 }, { "epoch": 0.3600186176402141, "grad_norm": 1.3494147062301636, "learning_rate": 0.0005, "loss": 2.287809133529663, "step": 3094 }, { "epoch": 0.36013497789155224, "grad_norm": 1.502247929573059, "learning_rate": 0.0005, "loss": 1.9088956117630005, "step": 3095 }, { "epoch": 0.3602513381428904, "grad_norm": 1.8741579055786133, "learning_rate": 0.0005, "loss": 2.44535756111145, "step": 3096 }, { "epoch": 0.36036769839422855, "grad_norm": 1.532774567604065, "learning_rate": 0.0005, "loss": 2.546396255493164, "step": 3097 }, { "epoch": 0.36048405864556665, "grad_norm": 1.3164883852005005, "learning_rate": 0.0005, "loss": 2.261289596557617, "step": 3098 }, { "epoch": 0.3606004188969048, "grad_norm": 1.6733803749084473, "learning_rate": 0.0005, "loss": 1.788901686668396, "step": 3099 }, { "epoch": 0.36071677914824296, "grad_norm": 3.285952091217041, "learning_rate": 0.0005, "loss": 2.3055572509765625, "step": 3100 }, { "epoch": 0.3608331393995811, "grad_norm": 2.045719861984253, "learning_rate": 0.0005, "loss": 2.7666845321655273, "step": 3101 }, { "epoch": 0.36094949965091927, "grad_norm": 1.3899496793746948, "learning_rate": 0.0005, "loss": 2.398068428039551, "step": 3102 }, { "epoch": 0.36106585990225737, "grad_norm": 1.9148805141448975, "learning_rate": 0.0005, "loss": 3.43938946723938, "step": 3103 }, { "epoch": 0.3611822201535955, "grad_norm": 1.595762014389038, "learning_rate": 0.0005, "loss": 2.9504051208496094, "step": 3104 }, { "epoch": 0.3612985804049337, "grad_norm": 1.8277374505996704, "learning_rate": 0.0005, "loss": 3.001951217651367, "step": 3105 }, { "epoch": 0.36141494065627183, "grad_norm": 1.1896295547485352, "learning_rate": 0.0005, "loss": 2.620943307876587, "step": 3106 }, { "epoch": 0.36153130090761, "grad_norm": 1.655053734779358, "learning_rate": 0.0005, "loss": 3.2287964820861816, "step": 3107 }, { "epoch": 0.3616476611589481, "grad_norm": 1.3822404146194458, "learning_rate": 0.0005, "loss": 2.9649782180786133, "step": 3108 }, { "epoch": 0.36176402141028624, "grad_norm": 1.0789315700531006, "learning_rate": 0.0005, "loss": 2.307163715362549, "step": 3109 }, { "epoch": 0.3618803816616244, "grad_norm": 1.0697643756866455, "learning_rate": 0.0005, "loss": 1.7559897899627686, "step": 3110 }, { "epoch": 0.36199674191296255, "grad_norm": 2.0463602542877197, "learning_rate": 0.0005, "loss": 3.4693877696990967, "step": 3111 }, { "epoch": 0.36211310216430065, "grad_norm": 1.2660478353500366, "learning_rate": 0.0005, "loss": 3.795474052429199, "step": 3112 }, { "epoch": 0.3622294624156388, "grad_norm": 1.4183602333068848, "learning_rate": 0.0005, "loss": 2.823497772216797, "step": 3113 }, { "epoch": 0.36234582266697696, "grad_norm": 1.3478935956954956, "learning_rate": 0.0005, "loss": 1.9020377397537231, "step": 3114 }, { "epoch": 0.3624621829183151, "grad_norm": 1.7399929761886597, "learning_rate": 0.0005, "loss": 2.739558219909668, "step": 3115 }, { "epoch": 0.36257854316965327, "grad_norm": 2.045186996459961, "learning_rate": 0.0005, "loss": 3.022213935852051, "step": 3116 }, { "epoch": 0.36269490342099137, "grad_norm": 1.2892272472381592, "learning_rate": 0.0005, "loss": 2.762312412261963, "step": 3117 }, { "epoch": 0.3628112636723295, "grad_norm": 1.6572771072387695, "learning_rate": 0.0005, "loss": 2.888458251953125, "step": 3118 }, { "epoch": 0.3629276239236677, "grad_norm": 3.6897268295288086, "learning_rate": 0.0005, "loss": 3.4750049114227295, "step": 3119 }, { "epoch": 0.36304398417500583, "grad_norm": 0.7384099960327148, "learning_rate": 0.0005, "loss": 1.6925240755081177, "step": 3120 }, { "epoch": 0.363160344426344, "grad_norm": 1.9694050550460815, "learning_rate": 0.0005, "loss": 2.143711805343628, "step": 3121 }, { "epoch": 0.3632767046776821, "grad_norm": 2.2813899517059326, "learning_rate": 0.0005, "loss": 2.65510892868042, "step": 3122 }, { "epoch": 0.36339306492902024, "grad_norm": 0.9770374298095703, "learning_rate": 0.0005, "loss": 2.409367561340332, "step": 3123 }, { "epoch": 0.3635094251803584, "grad_norm": 3.029136896133423, "learning_rate": 0.0005, "loss": 2.8595457077026367, "step": 3124 }, { "epoch": 0.36362578543169655, "grad_norm": 1.5843770503997803, "learning_rate": 0.0005, "loss": 3.854678153991699, "step": 3125 }, { "epoch": 0.3637421456830347, "grad_norm": 2.297682046890259, "learning_rate": 0.0005, "loss": 1.6448009014129639, "step": 3126 }, { "epoch": 0.3638585059343728, "grad_norm": 1.9124424457550049, "learning_rate": 0.0005, "loss": 4.042426109313965, "step": 3127 }, { "epoch": 0.36397486618571095, "grad_norm": 1.7574961185455322, "learning_rate": 0.0005, "loss": 2.9083566665649414, "step": 3128 }, { "epoch": 0.3640912264370491, "grad_norm": 2.4926207065582275, "learning_rate": 0.0005, "loss": 3.069089412689209, "step": 3129 }, { "epoch": 0.36420758668838726, "grad_norm": 1.2219856977462769, "learning_rate": 0.0005, "loss": 1.990375280380249, "step": 3130 }, { "epoch": 0.36432394693972536, "grad_norm": 1.0767751932144165, "learning_rate": 0.0005, "loss": 2.0843753814697266, "step": 3131 }, { "epoch": 0.3644403071910635, "grad_norm": 1.4445991516113281, "learning_rate": 0.0005, "loss": 3.4116549491882324, "step": 3132 }, { "epoch": 0.36455666744240167, "grad_norm": 1.6090490818023682, "learning_rate": 0.0005, "loss": 3.7519564628601074, "step": 3133 }, { "epoch": 0.3646730276937398, "grad_norm": 1.1228892803192139, "learning_rate": 0.0005, "loss": 3.7043538093566895, "step": 3134 }, { "epoch": 0.364789387945078, "grad_norm": 1.2705051898956299, "learning_rate": 0.0005, "loss": 2.6350138187408447, "step": 3135 }, { "epoch": 0.3649057481964161, "grad_norm": 0.9408599138259888, "learning_rate": 0.0005, "loss": 1.8148839473724365, "step": 3136 }, { "epoch": 0.36502210844775423, "grad_norm": 1.6269267797470093, "learning_rate": 0.0005, "loss": 3.8412299156188965, "step": 3137 }, { "epoch": 0.3651384686990924, "grad_norm": 1.524914264678955, "learning_rate": 0.0005, "loss": 2.685807704925537, "step": 3138 }, { "epoch": 0.36525482895043054, "grad_norm": 1.6133837699890137, "learning_rate": 0.0005, "loss": 2.936030387878418, "step": 3139 }, { "epoch": 0.3653711892017687, "grad_norm": 1.9375150203704834, "learning_rate": 0.0005, "loss": 1.7746086120605469, "step": 3140 }, { "epoch": 0.3654875494531068, "grad_norm": 1.6503480672836304, "learning_rate": 0.0005, "loss": 3.4027392864227295, "step": 3141 }, { "epoch": 0.36560390970444495, "grad_norm": 1.253235101699829, "learning_rate": 0.0005, "loss": 2.131485939025879, "step": 3142 }, { "epoch": 0.3657202699557831, "grad_norm": 0.9807434678077698, "learning_rate": 0.0005, "loss": 1.5544824600219727, "step": 3143 }, { "epoch": 0.36583663020712126, "grad_norm": 1.0916540622711182, "learning_rate": 0.0005, "loss": 2.501192331314087, "step": 3144 }, { "epoch": 0.3659529904584594, "grad_norm": 1.8602125644683838, "learning_rate": 0.0005, "loss": 2.990915298461914, "step": 3145 }, { "epoch": 0.3660693507097975, "grad_norm": 1.263630986213684, "learning_rate": 0.0005, "loss": 2.668837070465088, "step": 3146 }, { "epoch": 0.36618571096113567, "grad_norm": 1.6819634437561035, "learning_rate": 0.0005, "loss": 2.4505715370178223, "step": 3147 }, { "epoch": 0.3663020712124738, "grad_norm": 1.4709641933441162, "learning_rate": 0.0005, "loss": 2.8521625995635986, "step": 3148 }, { "epoch": 0.366418431463812, "grad_norm": 2.106933116912842, "learning_rate": 0.0005, "loss": 3.31318736076355, "step": 3149 }, { "epoch": 0.36653479171515013, "grad_norm": 1.0591099262237549, "learning_rate": 0.0005, "loss": 2.6404478549957275, "step": 3150 }, { "epoch": 0.36665115196648823, "grad_norm": 1.1185370683670044, "learning_rate": 0.0005, "loss": 2.3891236782073975, "step": 3151 }, { "epoch": 0.3667675122178264, "grad_norm": 1.9200925827026367, "learning_rate": 0.0005, "loss": 2.8567616939544678, "step": 3152 }, { "epoch": 0.36688387246916454, "grad_norm": 2.0250654220581055, "learning_rate": 0.0005, "loss": 2.951409339904785, "step": 3153 }, { "epoch": 0.3670002327205027, "grad_norm": 1.6650478839874268, "learning_rate": 0.0005, "loss": 3.0021984577178955, "step": 3154 }, { "epoch": 0.3671165929718408, "grad_norm": 1.3949061632156372, "learning_rate": 0.0005, "loss": 1.9479811191558838, "step": 3155 }, { "epoch": 0.36723295322317895, "grad_norm": 1.1458547115325928, "learning_rate": 0.0005, "loss": 1.9515020847320557, "step": 3156 }, { "epoch": 0.3673493134745171, "grad_norm": 1.386411428451538, "learning_rate": 0.0005, "loss": 2.05092716217041, "step": 3157 }, { "epoch": 0.36746567372585526, "grad_norm": 2.0472207069396973, "learning_rate": 0.0005, "loss": 3.3292713165283203, "step": 3158 }, { "epoch": 0.3675820339771934, "grad_norm": 1.3544964790344238, "learning_rate": 0.0005, "loss": 2.5584185123443604, "step": 3159 }, { "epoch": 0.3676983942285315, "grad_norm": 3.558715343475342, "learning_rate": 0.0005, "loss": 2.184621572494507, "step": 3160 }, { "epoch": 0.36781475447986967, "grad_norm": 1.6310577392578125, "learning_rate": 0.0005, "loss": 2.365367889404297, "step": 3161 }, { "epoch": 0.3679311147312078, "grad_norm": 1.3927433490753174, "learning_rate": 0.0005, "loss": 2.6325740814208984, "step": 3162 }, { "epoch": 0.368047474982546, "grad_norm": 1.0674755573272705, "learning_rate": 0.0005, "loss": 1.9844672679901123, "step": 3163 }, { "epoch": 0.36816383523388413, "grad_norm": 1.0715298652648926, "learning_rate": 0.0005, "loss": 2.063133716583252, "step": 3164 }, { "epoch": 0.36828019548522223, "grad_norm": 1.95490300655365, "learning_rate": 0.0005, "loss": 2.4225995540618896, "step": 3165 }, { "epoch": 0.3683965557365604, "grad_norm": 0.8938838839530945, "learning_rate": 0.0005, "loss": 1.5459187030792236, "step": 3166 }, { "epoch": 0.36851291598789854, "grad_norm": 0.9863972067832947, "learning_rate": 0.0005, "loss": 2.099534511566162, "step": 3167 }, { "epoch": 0.3686292762392367, "grad_norm": 1.3626079559326172, "learning_rate": 0.0005, "loss": 3.0503592491149902, "step": 3168 }, { "epoch": 0.36874563649057485, "grad_norm": 0.729357123374939, "learning_rate": 0.0005, "loss": 1.694570779800415, "step": 3169 }, { "epoch": 0.36886199674191295, "grad_norm": 1.9291496276855469, "learning_rate": 0.0005, "loss": 2.2854065895080566, "step": 3170 }, { "epoch": 0.3689783569932511, "grad_norm": 0.6695313453674316, "learning_rate": 0.0005, "loss": 1.72551429271698, "step": 3171 }, { "epoch": 0.36909471724458925, "grad_norm": 1.0317682027816772, "learning_rate": 0.0005, "loss": 2.6067707538604736, "step": 3172 }, { "epoch": 0.3692110774959274, "grad_norm": 1.7483527660369873, "learning_rate": 0.0005, "loss": 3.2939510345458984, "step": 3173 }, { "epoch": 0.3693274377472655, "grad_norm": 1.0811083316802979, "learning_rate": 0.0005, "loss": 1.6100130081176758, "step": 3174 }, { "epoch": 0.36944379799860366, "grad_norm": 1.3665343523025513, "learning_rate": 0.0005, "loss": 3.0822372436523438, "step": 3175 }, { "epoch": 0.3695601582499418, "grad_norm": 2.1977014541625977, "learning_rate": 0.0005, "loss": 3.229870080947876, "step": 3176 }, { "epoch": 0.36967651850127997, "grad_norm": 1.8912177085876465, "learning_rate": 0.0005, "loss": 3.389862060546875, "step": 3177 }, { "epoch": 0.3697928787526181, "grad_norm": 0.8960926532745361, "learning_rate": 0.0005, "loss": 1.6586334705352783, "step": 3178 }, { "epoch": 0.3699092390039562, "grad_norm": 1.2830039262771606, "learning_rate": 0.0005, "loss": 2.0173604488372803, "step": 3179 }, { "epoch": 0.3700255992552944, "grad_norm": 0.6745855808258057, "learning_rate": 0.0005, "loss": 1.610183596611023, "step": 3180 }, { "epoch": 0.37014195950663253, "grad_norm": 1.4135466814041138, "learning_rate": 0.0005, "loss": 3.065079927444458, "step": 3181 }, { "epoch": 0.3702583197579707, "grad_norm": 1.160635232925415, "learning_rate": 0.0005, "loss": 2.1657145023345947, "step": 3182 }, { "epoch": 0.37037468000930884, "grad_norm": 1.693410873413086, "learning_rate": 0.0005, "loss": 2.2704319953918457, "step": 3183 }, { "epoch": 0.37049104026064694, "grad_norm": 1.755423903465271, "learning_rate": 0.0005, "loss": 3.373669385910034, "step": 3184 }, { "epoch": 0.3706074005119851, "grad_norm": 1.7602242231369019, "learning_rate": 0.0005, "loss": 3.495438575744629, "step": 3185 }, { "epoch": 0.37072376076332325, "grad_norm": 1.2198472023010254, "learning_rate": 0.0005, "loss": 2.200364589691162, "step": 3186 }, { "epoch": 0.3708401210146614, "grad_norm": 1.0612224340438843, "learning_rate": 0.0005, "loss": 1.8409147262573242, "step": 3187 }, { "epoch": 0.37095648126599956, "grad_norm": 1.3568469285964966, "learning_rate": 0.0005, "loss": 2.2718849182128906, "step": 3188 }, { "epoch": 0.37107284151733766, "grad_norm": 1.8733888864517212, "learning_rate": 0.0005, "loss": 2.9325623512268066, "step": 3189 }, { "epoch": 0.3711892017686758, "grad_norm": 1.497044563293457, "learning_rate": 0.0005, "loss": 2.7272167205810547, "step": 3190 }, { "epoch": 0.37130556202001397, "grad_norm": 1.6147260665893555, "learning_rate": 0.0005, "loss": 2.650674819946289, "step": 3191 }, { "epoch": 0.3714219222713521, "grad_norm": 2.445720672607422, "learning_rate": 0.0005, "loss": 2.2290430068969727, "step": 3192 }, { "epoch": 0.3715382825226902, "grad_norm": 2.0396311283111572, "learning_rate": 0.0005, "loss": 2.3020081520080566, "step": 3193 }, { "epoch": 0.3716546427740284, "grad_norm": 1.7193384170532227, "learning_rate": 0.0005, "loss": 2.65971302986145, "step": 3194 }, { "epoch": 0.37177100302536653, "grad_norm": 2.513704299926758, "learning_rate": 0.0005, "loss": 2.282853841781616, "step": 3195 }, { "epoch": 0.3718873632767047, "grad_norm": 1.4194144010543823, "learning_rate": 0.0005, "loss": 2.9563748836517334, "step": 3196 }, { "epoch": 0.37200372352804284, "grad_norm": 1.5537056922912598, "learning_rate": 0.0005, "loss": 3.0333542823791504, "step": 3197 }, { "epoch": 0.37212008377938094, "grad_norm": 1.3795382976531982, "learning_rate": 0.0005, "loss": 2.5769152641296387, "step": 3198 }, { "epoch": 0.3722364440307191, "grad_norm": 1.6483690738677979, "learning_rate": 0.0005, "loss": 2.3164448738098145, "step": 3199 }, { "epoch": 0.37235280428205725, "grad_norm": 1.0133957862854004, "learning_rate": 0.0005, "loss": 2.709606647491455, "step": 3200 }, { "epoch": 0.3724691645333954, "grad_norm": 1.7157118320465088, "learning_rate": 0.0005, "loss": 2.001587152481079, "step": 3201 }, { "epoch": 0.37258552478473356, "grad_norm": 2.1324076652526855, "learning_rate": 0.0005, "loss": 2.789797067642212, "step": 3202 }, { "epoch": 0.37270188503607166, "grad_norm": 2.660930633544922, "learning_rate": 0.0005, "loss": 3.195608615875244, "step": 3203 }, { "epoch": 0.3728182452874098, "grad_norm": 1.5125526189804077, "learning_rate": 0.0005, "loss": 3.7048277854919434, "step": 3204 }, { "epoch": 0.37293460553874797, "grad_norm": 1.1519604921340942, "learning_rate": 0.0005, "loss": 2.1730480194091797, "step": 3205 }, { "epoch": 0.3730509657900861, "grad_norm": 1.1809570789337158, "learning_rate": 0.0005, "loss": 2.4546895027160645, "step": 3206 }, { "epoch": 0.3731673260414243, "grad_norm": 1.0777201652526855, "learning_rate": 0.0005, "loss": 2.421049118041992, "step": 3207 }, { "epoch": 0.3732836862927624, "grad_norm": 1.6985023021697998, "learning_rate": 0.0005, "loss": 2.198044776916504, "step": 3208 }, { "epoch": 0.37340004654410053, "grad_norm": 1.3259015083312988, "learning_rate": 0.0005, "loss": 2.977273464202881, "step": 3209 }, { "epoch": 0.3735164067954387, "grad_norm": 1.2551778554916382, "learning_rate": 0.0005, "loss": 2.351804733276367, "step": 3210 }, { "epoch": 0.37363276704677684, "grad_norm": 1.7675495147705078, "learning_rate": 0.0005, "loss": 2.845426082611084, "step": 3211 }, { "epoch": 0.373749127298115, "grad_norm": 1.4050899744033813, "learning_rate": 0.0005, "loss": 2.9959535598754883, "step": 3212 }, { "epoch": 0.3738654875494531, "grad_norm": 2.1820857524871826, "learning_rate": 0.0005, "loss": 3.4796276092529297, "step": 3213 }, { "epoch": 0.37398184780079125, "grad_norm": 1.2893497943878174, "learning_rate": 0.0005, "loss": 2.388251781463623, "step": 3214 }, { "epoch": 0.3740982080521294, "grad_norm": 1.1316425800323486, "learning_rate": 0.0005, "loss": 2.1198315620422363, "step": 3215 }, { "epoch": 0.37421456830346755, "grad_norm": 2.3388755321502686, "learning_rate": 0.0005, "loss": 2.5533719062805176, "step": 3216 }, { "epoch": 0.37433092855480565, "grad_norm": 1.4531197547912598, "learning_rate": 0.0005, "loss": 2.2204370498657227, "step": 3217 }, { "epoch": 0.3744472888061438, "grad_norm": 1.5658084154129028, "learning_rate": 0.0005, "loss": 2.3171162605285645, "step": 3218 }, { "epoch": 0.37456364905748196, "grad_norm": 2.871286630630493, "learning_rate": 0.0005, "loss": 3.0947606563568115, "step": 3219 }, { "epoch": 0.3746800093088201, "grad_norm": 1.2683714628219604, "learning_rate": 0.0005, "loss": 3.0864102840423584, "step": 3220 }, { "epoch": 0.37479636956015827, "grad_norm": 1.5502575635910034, "learning_rate": 0.0005, "loss": 2.7100260257720947, "step": 3221 }, { "epoch": 0.37491272981149637, "grad_norm": 1.8806076049804688, "learning_rate": 0.0005, "loss": 2.876983165740967, "step": 3222 }, { "epoch": 0.3750290900628345, "grad_norm": 1.724228858947754, "learning_rate": 0.0005, "loss": 2.6847403049468994, "step": 3223 }, { "epoch": 0.3751454503141727, "grad_norm": 0.9829881191253662, "learning_rate": 0.0005, "loss": 2.3621444702148438, "step": 3224 }, { "epoch": 0.37526181056551083, "grad_norm": 1.1912838220596313, "learning_rate": 0.0005, "loss": 2.6748454570770264, "step": 3225 }, { "epoch": 0.375378170816849, "grad_norm": 1.2570134401321411, "learning_rate": 0.0005, "loss": 3.531742572784424, "step": 3226 }, { "epoch": 0.3754945310681871, "grad_norm": 1.7681480646133423, "learning_rate": 0.0005, "loss": 1.6907962560653687, "step": 3227 }, { "epoch": 0.37561089131952524, "grad_norm": 0.9964762330055237, "learning_rate": 0.0005, "loss": 2.1974334716796875, "step": 3228 }, { "epoch": 0.3757272515708634, "grad_norm": 2.5735721588134766, "learning_rate": 0.0005, "loss": 3.218729019165039, "step": 3229 }, { "epoch": 0.37584361182220155, "grad_norm": 1.4276965856552124, "learning_rate": 0.0005, "loss": 2.5386571884155273, "step": 3230 }, { "epoch": 0.3759599720735397, "grad_norm": 1.0793124437332153, "learning_rate": 0.0005, "loss": 2.3670554161071777, "step": 3231 }, { "epoch": 0.3760763323248778, "grad_norm": 1.6259068250656128, "learning_rate": 0.0005, "loss": 2.084653377532959, "step": 3232 }, { "epoch": 0.37619269257621596, "grad_norm": 1.3811135292053223, "learning_rate": 0.0005, "loss": 3.4263253211975098, "step": 3233 }, { "epoch": 0.3763090528275541, "grad_norm": 1.7420010566711426, "learning_rate": 0.0005, "loss": 2.7588772773742676, "step": 3234 }, { "epoch": 0.37642541307889227, "grad_norm": 1.6470860242843628, "learning_rate": 0.0005, "loss": 3.1414499282836914, "step": 3235 }, { "epoch": 0.37654177333023037, "grad_norm": 1.2940081357955933, "learning_rate": 0.0005, "loss": 2.9520423412323, "step": 3236 }, { "epoch": 0.3766581335815685, "grad_norm": 1.604627013206482, "learning_rate": 0.0005, "loss": 1.8454442024230957, "step": 3237 }, { "epoch": 0.3767744938329067, "grad_norm": 1.4423238039016724, "learning_rate": 0.0005, "loss": 2.56174373626709, "step": 3238 }, { "epoch": 0.37689085408424483, "grad_norm": 0.9076018929481506, "learning_rate": 0.0005, "loss": 1.9932212829589844, "step": 3239 }, { "epoch": 0.377007214335583, "grad_norm": 1.3100192546844482, "learning_rate": 0.0005, "loss": 2.944173812866211, "step": 3240 }, { "epoch": 0.3771235745869211, "grad_norm": 1.2176942825317383, "learning_rate": 0.0005, "loss": 2.4187238216400146, "step": 3241 }, { "epoch": 0.37723993483825924, "grad_norm": 1.0215306282043457, "learning_rate": 0.0005, "loss": 3.115713119506836, "step": 3242 }, { "epoch": 0.3773562950895974, "grad_norm": 1.6091545820236206, "learning_rate": 0.0005, "loss": 4.116889953613281, "step": 3243 }, { "epoch": 0.37747265534093555, "grad_norm": 2.2433276176452637, "learning_rate": 0.0005, "loss": 3.0358331203460693, "step": 3244 }, { "epoch": 0.3775890155922737, "grad_norm": 1.786867380142212, "learning_rate": 0.0005, "loss": 2.369856834411621, "step": 3245 }, { "epoch": 0.3777053758436118, "grad_norm": 1.4746720790863037, "learning_rate": 0.0005, "loss": 2.228281259536743, "step": 3246 }, { "epoch": 0.37782173609494996, "grad_norm": 1.2226921319961548, "learning_rate": 0.0005, "loss": 2.4830150604248047, "step": 3247 }, { "epoch": 0.3779380963462881, "grad_norm": 2.017260789871216, "learning_rate": 0.0005, "loss": 3.0948915481567383, "step": 3248 }, { "epoch": 0.37805445659762626, "grad_norm": 1.4430809020996094, "learning_rate": 0.0005, "loss": 2.556942939758301, "step": 3249 }, { "epoch": 0.3781708168489644, "grad_norm": 34.94797897338867, "learning_rate": 0.0005, "loss": 3.0685508251190186, "step": 3250 }, { "epoch": 0.3782871771003025, "grad_norm": 1.5354552268981934, "learning_rate": 0.0005, "loss": 2.6570677757263184, "step": 3251 }, { "epoch": 0.3784035373516407, "grad_norm": 1.263974905014038, "learning_rate": 0.0005, "loss": 2.434661865234375, "step": 3252 }, { "epoch": 0.3785198976029788, "grad_norm": 1.3193647861480713, "learning_rate": 0.0005, "loss": 2.3753557205200195, "step": 3253 }, { "epoch": 0.378636257854317, "grad_norm": 1.2339500188827515, "learning_rate": 0.0005, "loss": 2.349794387817383, "step": 3254 }, { "epoch": 0.3787526181056551, "grad_norm": 1.422671914100647, "learning_rate": 0.0005, "loss": 2.387908697128296, "step": 3255 }, { "epoch": 0.37886897835699324, "grad_norm": 1.670433759689331, "learning_rate": 0.0005, "loss": 3.4576468467712402, "step": 3256 }, { "epoch": 0.3789853386083314, "grad_norm": 1.6536065340042114, "learning_rate": 0.0005, "loss": 1.9894909858703613, "step": 3257 }, { "epoch": 0.37910169885966954, "grad_norm": 0.8055899739265442, "learning_rate": 0.0005, "loss": 1.7764737606048584, "step": 3258 }, { "epoch": 0.3792180591110077, "grad_norm": 1.710353136062622, "learning_rate": 0.0005, "loss": 2.814730167388916, "step": 3259 }, { "epoch": 0.3793344193623458, "grad_norm": 1.0159252882003784, "learning_rate": 0.0005, "loss": 2.1511611938476562, "step": 3260 }, { "epoch": 0.37945077961368395, "grad_norm": 1.5717432498931885, "learning_rate": 0.0005, "loss": 2.300391674041748, "step": 3261 }, { "epoch": 0.3795671398650221, "grad_norm": 1.213296890258789, "learning_rate": 0.0005, "loss": 1.8021808862686157, "step": 3262 }, { "epoch": 0.37968350011636026, "grad_norm": 1.494771957397461, "learning_rate": 0.0005, "loss": 3.203777551651001, "step": 3263 }, { "epoch": 0.3797998603676984, "grad_norm": 1.7262907028198242, "learning_rate": 0.0005, "loss": 2.135740280151367, "step": 3264 }, { "epoch": 0.3799162206190365, "grad_norm": 1.9751158952713013, "learning_rate": 0.0005, "loss": 3.4517321586608887, "step": 3265 }, { "epoch": 0.38003258087037467, "grad_norm": 1.6193809509277344, "learning_rate": 0.0005, "loss": 3.496342897415161, "step": 3266 }, { "epoch": 0.3801489411217128, "grad_norm": 2.0919506549835205, "learning_rate": 0.0005, "loss": 4.219329833984375, "step": 3267 }, { "epoch": 0.380265301373051, "grad_norm": 2.314491033554077, "learning_rate": 0.0005, "loss": 2.4574122428894043, "step": 3268 }, { "epoch": 0.38038166162438913, "grad_norm": 1.553682804107666, "learning_rate": 0.0005, "loss": 3.032177209854126, "step": 3269 }, { "epoch": 0.38049802187572723, "grad_norm": 1.5196443796157837, "learning_rate": 0.0005, "loss": 2.7567334175109863, "step": 3270 }, { "epoch": 0.3806143821270654, "grad_norm": 0.8624895811080933, "learning_rate": 0.0005, "loss": 1.1056307554244995, "step": 3271 }, { "epoch": 0.38073074237840354, "grad_norm": 1.7111860513687134, "learning_rate": 0.0005, "loss": 2.812154769897461, "step": 3272 }, { "epoch": 0.3808471026297417, "grad_norm": 1.151489496231079, "learning_rate": 0.0005, "loss": 2.3859081268310547, "step": 3273 }, { "epoch": 0.38096346288107985, "grad_norm": 1.5905297994613647, "learning_rate": 0.0005, "loss": 2.050797462463379, "step": 3274 }, { "epoch": 0.38107982313241795, "grad_norm": 1.2374035120010376, "learning_rate": 0.0005, "loss": 2.60170841217041, "step": 3275 }, { "epoch": 0.3811961833837561, "grad_norm": 1.191928744316101, "learning_rate": 0.0005, "loss": 3.090287685394287, "step": 3276 }, { "epoch": 0.38131254363509426, "grad_norm": 1.2651338577270508, "learning_rate": 0.0005, "loss": 2.907167434692383, "step": 3277 }, { "epoch": 0.3814289038864324, "grad_norm": 1.9423199892044067, "learning_rate": 0.0005, "loss": 2.2232189178466797, "step": 3278 }, { "epoch": 0.3815452641377705, "grad_norm": 1.0527278184890747, "learning_rate": 0.0005, "loss": 2.5200273990631104, "step": 3279 }, { "epoch": 0.38166162438910867, "grad_norm": 1.4305201768875122, "learning_rate": 0.0005, "loss": 2.9491453170776367, "step": 3280 }, { "epoch": 0.3817779846404468, "grad_norm": 1.4369839429855347, "learning_rate": 0.0005, "loss": 2.545851707458496, "step": 3281 }, { "epoch": 0.381894344891785, "grad_norm": 1.1331886053085327, "learning_rate": 0.0005, "loss": 3.201007843017578, "step": 3282 }, { "epoch": 0.38201070514312313, "grad_norm": 1.1980009078979492, "learning_rate": 0.0005, "loss": 3.143057346343994, "step": 3283 }, { "epoch": 0.38212706539446123, "grad_norm": 1.5399960279464722, "learning_rate": 0.0005, "loss": 2.2588515281677246, "step": 3284 }, { "epoch": 0.3822434256457994, "grad_norm": 1.421566128730774, "learning_rate": 0.0005, "loss": 2.6674718856811523, "step": 3285 }, { "epoch": 0.38235978589713754, "grad_norm": 1.3489373922348022, "learning_rate": 0.0005, "loss": 2.4353528022766113, "step": 3286 }, { "epoch": 0.3824761461484757, "grad_norm": 1.6512593030929565, "learning_rate": 0.0005, "loss": 2.8771703243255615, "step": 3287 }, { "epoch": 0.38259250639981385, "grad_norm": 1.463977575302124, "learning_rate": 0.0005, "loss": 2.2223920822143555, "step": 3288 }, { "epoch": 0.38270886665115195, "grad_norm": 0.8639407157897949, "learning_rate": 0.0005, "loss": 2.6230227947235107, "step": 3289 }, { "epoch": 0.3828252269024901, "grad_norm": 1.2061233520507812, "learning_rate": 0.0005, "loss": 2.290301561355591, "step": 3290 }, { "epoch": 0.38294158715382826, "grad_norm": 1.893208622932434, "learning_rate": 0.0005, "loss": 2.3583147525787354, "step": 3291 }, { "epoch": 0.3830579474051664, "grad_norm": 2.3961691856384277, "learning_rate": 0.0005, "loss": 2.8860769271850586, "step": 3292 }, { "epoch": 0.38317430765650456, "grad_norm": 1.8326953649520874, "learning_rate": 0.0005, "loss": 3.522494316101074, "step": 3293 }, { "epoch": 0.38329066790784266, "grad_norm": 0.7918187975883484, "learning_rate": 0.0005, "loss": 2.30910062789917, "step": 3294 }, { "epoch": 0.3834070281591808, "grad_norm": 2.540083408355713, "learning_rate": 0.0005, "loss": 2.872596263885498, "step": 3295 }, { "epoch": 0.383523388410519, "grad_norm": 1.3636590242385864, "learning_rate": 0.0005, "loss": 2.5388705730438232, "step": 3296 }, { "epoch": 0.3836397486618571, "grad_norm": 1.6398018598556519, "learning_rate": 0.0005, "loss": 3.953704833984375, "step": 3297 }, { "epoch": 0.3837561089131952, "grad_norm": 1.393274188041687, "learning_rate": 0.0005, "loss": 2.305649995803833, "step": 3298 }, { "epoch": 0.3838724691645334, "grad_norm": 1.1428015232086182, "learning_rate": 0.0005, "loss": 2.729010820388794, "step": 3299 }, { "epoch": 0.38398882941587154, "grad_norm": 1.2856144905090332, "learning_rate": 0.0005, "loss": 2.4065732955932617, "step": 3300 }, { "epoch": 0.3841051896672097, "grad_norm": 1.0811141729354858, "learning_rate": 0.0005, "loss": 1.6548352241516113, "step": 3301 }, { "epoch": 0.38422154991854784, "grad_norm": 1.4413946866989136, "learning_rate": 0.0005, "loss": 3.2535204887390137, "step": 3302 }, { "epoch": 0.38433791016988594, "grad_norm": 1.8024569749832153, "learning_rate": 0.0005, "loss": 3.1564455032348633, "step": 3303 }, { "epoch": 0.3844542704212241, "grad_norm": 1.686610221862793, "learning_rate": 0.0005, "loss": 3.8046388626098633, "step": 3304 }, { "epoch": 0.38457063067256225, "grad_norm": 1.063227653503418, "learning_rate": 0.0005, "loss": 2.0539002418518066, "step": 3305 }, { "epoch": 0.3846869909239004, "grad_norm": 1.3899345397949219, "learning_rate": 0.0005, "loss": 3.097939968109131, "step": 3306 }, { "epoch": 0.38480335117523856, "grad_norm": 1.6079728603363037, "learning_rate": 0.0005, "loss": 2.6695098876953125, "step": 3307 }, { "epoch": 0.38491971142657666, "grad_norm": 1.8983001708984375, "learning_rate": 0.0005, "loss": 2.719095230102539, "step": 3308 }, { "epoch": 0.3850360716779148, "grad_norm": 1.1821783781051636, "learning_rate": 0.0005, "loss": 1.875855803489685, "step": 3309 }, { "epoch": 0.38515243192925297, "grad_norm": 2.308514356613159, "learning_rate": 0.0005, "loss": 2.9177920818328857, "step": 3310 }, { "epoch": 0.3852687921805911, "grad_norm": 1.2073110342025757, "learning_rate": 0.0005, "loss": 2.1953701972961426, "step": 3311 }, { "epoch": 0.3853851524319293, "grad_norm": 1.6922868490219116, "learning_rate": 0.0005, "loss": 2.7369089126586914, "step": 3312 }, { "epoch": 0.3855015126832674, "grad_norm": 2.7573082447052, "learning_rate": 0.0005, "loss": 2.231031656265259, "step": 3313 }, { "epoch": 0.38561787293460553, "grad_norm": 1.4847898483276367, "learning_rate": 0.0005, "loss": 2.2087531089782715, "step": 3314 }, { "epoch": 0.3857342331859437, "grad_norm": 1.3035799264907837, "learning_rate": 0.0005, "loss": 1.9667243957519531, "step": 3315 }, { "epoch": 0.38585059343728184, "grad_norm": 0.6045926809310913, "learning_rate": 0.0005, "loss": 1.4128059148788452, "step": 3316 }, { "epoch": 0.38596695368861994, "grad_norm": 1.717873215675354, "learning_rate": 0.0005, "loss": 3.3118770122528076, "step": 3317 }, { "epoch": 0.3860833139399581, "grad_norm": 2.4034578800201416, "learning_rate": 0.0005, "loss": 2.5906898975372314, "step": 3318 }, { "epoch": 0.38619967419129625, "grad_norm": 1.375089406967163, "learning_rate": 0.0005, "loss": 2.5266475677490234, "step": 3319 }, { "epoch": 0.3863160344426344, "grad_norm": 1.2764822244644165, "learning_rate": 0.0005, "loss": 2.2778584957122803, "step": 3320 }, { "epoch": 0.38643239469397256, "grad_norm": 1.2803852558135986, "learning_rate": 0.0005, "loss": 1.8855321407318115, "step": 3321 }, { "epoch": 0.38654875494531066, "grad_norm": 1.5482816696166992, "learning_rate": 0.0005, "loss": 2.5934853553771973, "step": 3322 }, { "epoch": 0.3866651151966488, "grad_norm": 1.2215721607208252, "learning_rate": 0.0005, "loss": 2.765746593475342, "step": 3323 }, { "epoch": 0.38678147544798697, "grad_norm": 1.2810348272323608, "learning_rate": 0.0005, "loss": 3.4475436210632324, "step": 3324 }, { "epoch": 0.3868978356993251, "grad_norm": 1.365927815437317, "learning_rate": 0.0005, "loss": 2.8762083053588867, "step": 3325 }, { "epoch": 0.3870141959506633, "grad_norm": 1.7180488109588623, "learning_rate": 0.0005, "loss": 2.7531754970550537, "step": 3326 }, { "epoch": 0.3871305562020014, "grad_norm": 1.5110126733779907, "learning_rate": 0.0005, "loss": 3.4491708278656006, "step": 3327 }, { "epoch": 0.38724691645333953, "grad_norm": 2.673285484313965, "learning_rate": 0.0005, "loss": 2.871432304382324, "step": 3328 }, { "epoch": 0.3873632767046777, "grad_norm": 2.398615598678589, "learning_rate": 0.0005, "loss": 3.16805100440979, "step": 3329 }, { "epoch": 0.38747963695601584, "grad_norm": 5.47760009765625, "learning_rate": 0.0005, "loss": 3.2541959285736084, "step": 3330 }, { "epoch": 0.387595997207354, "grad_norm": 1.7670905590057373, "learning_rate": 0.0005, "loss": 3.58915376663208, "step": 3331 }, { "epoch": 0.3877123574586921, "grad_norm": 1.1149946451187134, "learning_rate": 0.0005, "loss": 2.1194515228271484, "step": 3332 }, { "epoch": 0.38782871771003025, "grad_norm": 1.7077581882476807, "learning_rate": 0.0005, "loss": 3.380627393722534, "step": 3333 }, { "epoch": 0.3879450779613684, "grad_norm": 1.212534785270691, "learning_rate": 0.0005, "loss": 2.9512383937835693, "step": 3334 }, { "epoch": 0.38806143821270656, "grad_norm": 1.6803135871887207, "learning_rate": 0.0005, "loss": 3.256087064743042, "step": 3335 }, { "epoch": 0.3881777984640447, "grad_norm": 1.1955254077911377, "learning_rate": 0.0005, "loss": 2.53957200050354, "step": 3336 }, { "epoch": 0.3882941587153828, "grad_norm": 0.9226422309875488, "learning_rate": 0.0005, "loss": 2.795681953430176, "step": 3337 }, { "epoch": 0.38841051896672096, "grad_norm": 1.29548180103302, "learning_rate": 0.0005, "loss": 2.6528356075286865, "step": 3338 }, { "epoch": 0.3885268792180591, "grad_norm": 2.5031418800354004, "learning_rate": 0.0005, "loss": 2.216808795928955, "step": 3339 }, { "epoch": 0.3886432394693973, "grad_norm": 0.9231769442558289, "learning_rate": 0.0005, "loss": 2.2161240577697754, "step": 3340 }, { "epoch": 0.38875959972073537, "grad_norm": 1.1056973934173584, "learning_rate": 0.0005, "loss": 2.6601474285125732, "step": 3341 }, { "epoch": 0.3888759599720735, "grad_norm": 1.5073076486587524, "learning_rate": 0.0005, "loss": 2.126774787902832, "step": 3342 }, { "epoch": 0.3889923202234117, "grad_norm": 0.5192974805831909, "learning_rate": 0.0005, "loss": 1.106445074081421, "step": 3343 }, { "epoch": 0.38910868047474984, "grad_norm": 2.6399970054626465, "learning_rate": 0.0005, "loss": 2.9461538791656494, "step": 3344 }, { "epoch": 0.389225040726088, "grad_norm": 1.6880971193313599, "learning_rate": 0.0005, "loss": 2.9830212593078613, "step": 3345 }, { "epoch": 0.3893414009774261, "grad_norm": 1.2661397457122803, "learning_rate": 0.0005, "loss": 2.084136486053467, "step": 3346 }, { "epoch": 0.38945776122876424, "grad_norm": 1.4375282526016235, "learning_rate": 0.0005, "loss": 2.709254026412964, "step": 3347 }, { "epoch": 0.3895741214801024, "grad_norm": 2.0208656787872314, "learning_rate": 0.0005, "loss": 3.358051300048828, "step": 3348 }, { "epoch": 0.38969048173144055, "grad_norm": 1.5681500434875488, "learning_rate": 0.0005, "loss": 3.7466092109680176, "step": 3349 }, { "epoch": 0.3898068419827787, "grad_norm": 1.227840542793274, "learning_rate": 0.0005, "loss": 2.2459514141082764, "step": 3350 }, { "epoch": 0.3899232022341168, "grad_norm": 1.8841266632080078, "learning_rate": 0.0005, "loss": 2.5867388248443604, "step": 3351 }, { "epoch": 0.39003956248545496, "grad_norm": 1.5512434244155884, "learning_rate": 0.0005, "loss": 2.9423863887786865, "step": 3352 }, { "epoch": 0.3901559227367931, "grad_norm": 1.6758058071136475, "learning_rate": 0.0005, "loss": 3.1185402870178223, "step": 3353 }, { "epoch": 0.39027228298813127, "grad_norm": 1.5452909469604492, "learning_rate": 0.0005, "loss": 1.960686445236206, "step": 3354 }, { "epoch": 0.3903886432394694, "grad_norm": 1.2716560363769531, "learning_rate": 0.0005, "loss": 1.768651008605957, "step": 3355 }, { "epoch": 0.3905050034908075, "grad_norm": 0.9913520812988281, "learning_rate": 0.0005, "loss": 1.500036358833313, "step": 3356 }, { "epoch": 0.3906213637421457, "grad_norm": 1.3961656093597412, "learning_rate": 0.0005, "loss": 3.336775302886963, "step": 3357 }, { "epoch": 0.39073772399348383, "grad_norm": 2.0020768642425537, "learning_rate": 0.0005, "loss": 3.038952112197876, "step": 3358 }, { "epoch": 0.390854084244822, "grad_norm": 1.6859855651855469, "learning_rate": 0.0005, "loss": 2.5230607986450195, "step": 3359 }, { "epoch": 0.3909704444961601, "grad_norm": 1.5448724031448364, "learning_rate": 0.0005, "loss": 2.8719370365142822, "step": 3360 }, { "epoch": 0.39108680474749824, "grad_norm": 1.1680206060409546, "learning_rate": 0.0005, "loss": 3.4148857593536377, "step": 3361 }, { "epoch": 0.3912031649988364, "grad_norm": 2.050448179244995, "learning_rate": 0.0005, "loss": 3.0502421855926514, "step": 3362 }, { "epoch": 0.39131952525017455, "grad_norm": 1.3098413944244385, "learning_rate": 0.0005, "loss": 3.044595241546631, "step": 3363 }, { "epoch": 0.3914358855015127, "grad_norm": 1.7544670104980469, "learning_rate": 0.0005, "loss": 3.136399269104004, "step": 3364 }, { "epoch": 0.3915522457528508, "grad_norm": 1.5589927434921265, "learning_rate": 0.0005, "loss": 2.7611541748046875, "step": 3365 }, { "epoch": 0.39166860600418896, "grad_norm": 7.39522123336792, "learning_rate": 0.0005, "loss": 2.376380443572998, "step": 3366 }, { "epoch": 0.3917849662555271, "grad_norm": 2.083742618560791, "learning_rate": 0.0005, "loss": 3.9864983558654785, "step": 3367 }, { "epoch": 0.39190132650686527, "grad_norm": 1.5739364624023438, "learning_rate": 0.0005, "loss": 2.5408899784088135, "step": 3368 }, { "epoch": 0.3920176867582034, "grad_norm": 1.0730429887771606, "learning_rate": 0.0005, "loss": 2.150644540786743, "step": 3369 }, { "epoch": 0.3921340470095415, "grad_norm": 1.8521438837051392, "learning_rate": 0.0005, "loss": 2.6319093704223633, "step": 3370 }, { "epoch": 0.3922504072608797, "grad_norm": 1.3573013544082642, "learning_rate": 0.0005, "loss": 2.649813413619995, "step": 3371 }, { "epoch": 0.39236676751221783, "grad_norm": 1.3013778924942017, "learning_rate": 0.0005, "loss": 2.178807020187378, "step": 3372 }, { "epoch": 0.392483127763556, "grad_norm": 1.3561475276947021, "learning_rate": 0.0005, "loss": 2.8381125926971436, "step": 3373 }, { "epoch": 0.39259948801489414, "grad_norm": 1.1891461610794067, "learning_rate": 0.0005, "loss": 2.8282482624053955, "step": 3374 }, { "epoch": 0.39271584826623224, "grad_norm": 1.0243922472000122, "learning_rate": 0.0005, "loss": 1.5395817756652832, "step": 3375 }, { "epoch": 0.3928322085175704, "grad_norm": 1.4980254173278809, "learning_rate": 0.0005, "loss": 3.0745673179626465, "step": 3376 }, { "epoch": 0.39294856876890855, "grad_norm": 1.7316150665283203, "learning_rate": 0.0005, "loss": 2.889720916748047, "step": 3377 }, { "epoch": 0.3930649290202467, "grad_norm": 1.1786075830459595, "learning_rate": 0.0005, "loss": 2.749344825744629, "step": 3378 }, { "epoch": 0.3931812892715848, "grad_norm": 0.7534823417663574, "learning_rate": 0.0005, "loss": 2.0557034015655518, "step": 3379 }, { "epoch": 0.39329764952292295, "grad_norm": 0.8283966779708862, "learning_rate": 0.0005, "loss": 2.2033610343933105, "step": 3380 }, { "epoch": 0.3934140097742611, "grad_norm": 1.0982571840286255, "learning_rate": 0.0005, "loss": 2.5483946800231934, "step": 3381 }, { "epoch": 0.39353037002559926, "grad_norm": 1.6472724676132202, "learning_rate": 0.0005, "loss": 3.1055431365966797, "step": 3382 }, { "epoch": 0.3936467302769374, "grad_norm": 1.8078585863113403, "learning_rate": 0.0005, "loss": 3.041828155517578, "step": 3383 }, { "epoch": 0.3937630905282755, "grad_norm": 1.2917276620864868, "learning_rate": 0.0005, "loss": 1.8430582284927368, "step": 3384 }, { "epoch": 0.39387945077961367, "grad_norm": 1.2840255498886108, "learning_rate": 0.0005, "loss": 2.220165729522705, "step": 3385 }, { "epoch": 0.3939958110309518, "grad_norm": 1.5264852046966553, "learning_rate": 0.0005, "loss": 3.56655216217041, "step": 3386 }, { "epoch": 0.39411217128229, "grad_norm": 1.7819008827209473, "learning_rate": 0.0005, "loss": 3.1831560134887695, "step": 3387 }, { "epoch": 0.39422853153362813, "grad_norm": 1.487494707107544, "learning_rate": 0.0005, "loss": 2.472952365875244, "step": 3388 }, { "epoch": 0.39434489178496623, "grad_norm": 1.3228964805603027, "learning_rate": 0.0005, "loss": 2.7731504440307617, "step": 3389 }, { "epoch": 0.3944612520363044, "grad_norm": 1.0465631484985352, "learning_rate": 0.0005, "loss": 2.286100149154663, "step": 3390 }, { "epoch": 0.39457761228764254, "grad_norm": 0.8504453897476196, "learning_rate": 0.0005, "loss": 1.7997734546661377, "step": 3391 }, { "epoch": 0.3946939725389807, "grad_norm": 1.5533539056777954, "learning_rate": 0.0005, "loss": 3.1434879302978516, "step": 3392 }, { "epoch": 0.39481033279031885, "grad_norm": 1.4940106868743896, "learning_rate": 0.0005, "loss": 2.748807907104492, "step": 3393 }, { "epoch": 0.39492669304165695, "grad_norm": 1.2592008113861084, "learning_rate": 0.0005, "loss": 2.379411458969116, "step": 3394 }, { "epoch": 0.3950430532929951, "grad_norm": 1.3027132749557495, "learning_rate": 0.0005, "loss": 2.198409080505371, "step": 3395 }, { "epoch": 0.39515941354433326, "grad_norm": 1.4180073738098145, "learning_rate": 0.0005, "loss": 2.113865613937378, "step": 3396 }, { "epoch": 0.3952757737956714, "grad_norm": 1.5723797082901, "learning_rate": 0.0005, "loss": 2.565751075744629, "step": 3397 }, { "epoch": 0.39539213404700957, "grad_norm": 1.1439145803451538, "learning_rate": 0.0005, "loss": 1.6774330139160156, "step": 3398 }, { "epoch": 0.39550849429834767, "grad_norm": 1.7916325330734253, "learning_rate": 0.0005, "loss": 2.5726349353790283, "step": 3399 }, { "epoch": 0.3956248545496858, "grad_norm": 1.6232117414474487, "learning_rate": 0.0005, "loss": 2.5760021209716797, "step": 3400 }, { "epoch": 0.395741214801024, "grad_norm": 1.702970266342163, "learning_rate": 0.0005, "loss": 2.869702100753784, "step": 3401 }, { "epoch": 0.39585757505236213, "grad_norm": 1.3714243173599243, "learning_rate": 0.0005, "loss": 1.7660820484161377, "step": 3402 }, { "epoch": 0.39597393530370023, "grad_norm": 2.127356767654419, "learning_rate": 0.0005, "loss": 2.3185229301452637, "step": 3403 }, { "epoch": 0.3960902955550384, "grad_norm": 2.0642690658569336, "learning_rate": 0.0005, "loss": 3.177259683609009, "step": 3404 }, { "epoch": 0.39620665580637654, "grad_norm": 1.5234936475753784, "learning_rate": 0.0005, "loss": 2.4310224056243896, "step": 3405 }, { "epoch": 0.3963230160577147, "grad_norm": 1.3244378566741943, "learning_rate": 0.0005, "loss": 2.2249913215637207, "step": 3406 }, { "epoch": 0.39643937630905285, "grad_norm": 3.3445370197296143, "learning_rate": 0.0005, "loss": 3.3662471771240234, "step": 3407 }, { "epoch": 0.39655573656039095, "grad_norm": 1.8305838108062744, "learning_rate": 0.0005, "loss": 2.844545364379883, "step": 3408 }, { "epoch": 0.3966720968117291, "grad_norm": 1.1622802019119263, "learning_rate": 0.0005, "loss": 2.6375584602355957, "step": 3409 }, { "epoch": 0.39678845706306726, "grad_norm": 1.6729676723480225, "learning_rate": 0.0005, "loss": 2.7383928298950195, "step": 3410 }, { "epoch": 0.3969048173144054, "grad_norm": 1.313934087753296, "learning_rate": 0.0005, "loss": 1.5456488132476807, "step": 3411 }, { "epoch": 0.39702117756574357, "grad_norm": 1.908255696296692, "learning_rate": 0.0005, "loss": 3.5285086631774902, "step": 3412 }, { "epoch": 0.39713753781708167, "grad_norm": 1.9210495948791504, "learning_rate": 0.0005, "loss": 2.6785717010498047, "step": 3413 }, { "epoch": 0.3972538980684198, "grad_norm": 1.813922643661499, "learning_rate": 0.0005, "loss": 3.3032047748565674, "step": 3414 }, { "epoch": 0.397370258319758, "grad_norm": 2.035856246948242, "learning_rate": 0.0005, "loss": 2.7580065727233887, "step": 3415 }, { "epoch": 0.39748661857109613, "grad_norm": 1.442378044128418, "learning_rate": 0.0005, "loss": 3.1659302711486816, "step": 3416 }, { "epoch": 0.3976029788224343, "grad_norm": 1.6292821168899536, "learning_rate": 0.0005, "loss": 1.9547920227050781, "step": 3417 }, { "epoch": 0.3977193390737724, "grad_norm": 1.3669437170028687, "learning_rate": 0.0005, "loss": 3.4412622451782227, "step": 3418 }, { "epoch": 0.39783569932511054, "grad_norm": 1.3516480922698975, "learning_rate": 0.0005, "loss": 2.651048183441162, "step": 3419 }, { "epoch": 0.3979520595764487, "grad_norm": 1.1234502792358398, "learning_rate": 0.0005, "loss": 2.787200927734375, "step": 3420 }, { "epoch": 0.39806841982778685, "grad_norm": 1.929244041442871, "learning_rate": 0.0005, "loss": 2.9637913703918457, "step": 3421 }, { "epoch": 0.39818478007912494, "grad_norm": 1.8211137056350708, "learning_rate": 0.0005, "loss": 3.2795841693878174, "step": 3422 }, { "epoch": 0.3983011403304631, "grad_norm": 1.081864356994629, "learning_rate": 0.0005, "loss": 1.299156665802002, "step": 3423 }, { "epoch": 0.39841750058180125, "grad_norm": 1.3618429899215698, "learning_rate": 0.0005, "loss": 2.736276149749756, "step": 3424 }, { "epoch": 0.3985338608331394, "grad_norm": 1.3607277870178223, "learning_rate": 0.0005, "loss": 2.5468740463256836, "step": 3425 }, { "epoch": 0.39865022108447756, "grad_norm": 1.4168821573257446, "learning_rate": 0.0005, "loss": 3.0259952545166016, "step": 3426 }, { "epoch": 0.39876658133581566, "grad_norm": 1.25534188747406, "learning_rate": 0.0005, "loss": 2.949432134628296, "step": 3427 }, { "epoch": 0.3988829415871538, "grad_norm": 2.533391237258911, "learning_rate": 0.0005, "loss": 3.3374505043029785, "step": 3428 }, { "epoch": 0.39899930183849197, "grad_norm": 1.2896090745925903, "learning_rate": 0.0005, "loss": 2.2782304286956787, "step": 3429 }, { "epoch": 0.3991156620898301, "grad_norm": 4.1741108894348145, "learning_rate": 0.0005, "loss": 3.252542018890381, "step": 3430 }, { "epoch": 0.3992320223411683, "grad_norm": 2.945859670639038, "learning_rate": 0.0005, "loss": 2.7835538387298584, "step": 3431 }, { "epoch": 0.3993483825925064, "grad_norm": 1.1230741739273071, "learning_rate": 0.0005, "loss": 3.1505632400512695, "step": 3432 }, { "epoch": 0.39946474284384453, "grad_norm": 1.2300323247909546, "learning_rate": 0.0005, "loss": 2.9122095108032227, "step": 3433 }, { "epoch": 0.3995811030951827, "grad_norm": 1.5090608596801758, "learning_rate": 0.0005, "loss": 3.4133358001708984, "step": 3434 }, { "epoch": 0.39969746334652084, "grad_norm": 1.7952786684036255, "learning_rate": 0.0005, "loss": 2.1586716175079346, "step": 3435 }, { "epoch": 0.399813823597859, "grad_norm": 1.7487845420837402, "learning_rate": 0.0005, "loss": 3.1825814247131348, "step": 3436 }, { "epoch": 0.3999301838491971, "grad_norm": 1.055110216140747, "learning_rate": 0.0005, "loss": 1.7151544094085693, "step": 3437 }, { "epoch": 0.40004654410053525, "grad_norm": 1.6782658100128174, "learning_rate": 0.0005, "loss": 3.471585988998413, "step": 3438 }, { "epoch": 0.4001629043518734, "grad_norm": 1.4465539455413818, "learning_rate": 0.0005, "loss": 2.7387535572052, "step": 3439 }, { "epoch": 0.40027926460321156, "grad_norm": 1.172430157661438, "learning_rate": 0.0005, "loss": 2.314775228500366, "step": 3440 }, { "epoch": 0.40039562485454966, "grad_norm": 1.4048672914505005, "learning_rate": 0.0005, "loss": 2.030975818634033, "step": 3441 }, { "epoch": 0.4005119851058878, "grad_norm": 2.0316860675811768, "learning_rate": 0.0005, "loss": 3.7966957092285156, "step": 3442 }, { "epoch": 0.40062834535722597, "grad_norm": 2.1140594482421875, "learning_rate": 0.0005, "loss": 1.8890271186828613, "step": 3443 }, { "epoch": 0.4007447056085641, "grad_norm": 1.8714858293533325, "learning_rate": 0.0005, "loss": 2.8035099506378174, "step": 3444 }, { "epoch": 0.4008610658599023, "grad_norm": 1.4464157819747925, "learning_rate": 0.0005, "loss": 3.244283676147461, "step": 3445 }, { "epoch": 0.4009774261112404, "grad_norm": 1.0294463634490967, "learning_rate": 0.0005, "loss": 2.1293630599975586, "step": 3446 }, { "epoch": 0.40109378636257853, "grad_norm": 3.4289536476135254, "learning_rate": 0.0005, "loss": 3.0816221237182617, "step": 3447 }, { "epoch": 0.4012101466139167, "grad_norm": 2.055652141571045, "learning_rate": 0.0005, "loss": 2.575242042541504, "step": 3448 }, { "epoch": 0.40132650686525484, "grad_norm": 1.6380213499069214, "learning_rate": 0.0005, "loss": 3.3373007774353027, "step": 3449 }, { "epoch": 0.401442867116593, "grad_norm": 1.7738184928894043, "learning_rate": 0.0005, "loss": 3.0413818359375, "step": 3450 }, { "epoch": 0.4015592273679311, "grad_norm": 0.9443990588188171, "learning_rate": 0.0005, "loss": 1.5930901765823364, "step": 3451 }, { "epoch": 0.40167558761926925, "grad_norm": 2.099717140197754, "learning_rate": 0.0005, "loss": 2.08731746673584, "step": 3452 }, { "epoch": 0.4017919478706074, "grad_norm": 1.5863937139511108, "learning_rate": 0.0005, "loss": 2.27655029296875, "step": 3453 }, { "epoch": 0.40190830812194556, "grad_norm": 1.5136626958847046, "learning_rate": 0.0005, "loss": 2.215426445007324, "step": 3454 }, { "epoch": 0.4020246683732837, "grad_norm": 1.5118188858032227, "learning_rate": 0.0005, "loss": 2.811495304107666, "step": 3455 }, { "epoch": 0.4021410286246218, "grad_norm": 1.3811120986938477, "learning_rate": 0.0005, "loss": 2.526489734649658, "step": 3456 }, { "epoch": 0.40225738887595996, "grad_norm": 1.3132346868515015, "learning_rate": 0.0005, "loss": 2.216233730316162, "step": 3457 }, { "epoch": 0.4023737491272981, "grad_norm": 1.2756102085113525, "learning_rate": 0.0005, "loss": 2.3719160556793213, "step": 3458 }, { "epoch": 0.4024901093786363, "grad_norm": 1.9183651208877563, "learning_rate": 0.0005, "loss": 3.2441258430480957, "step": 3459 }, { "epoch": 0.40260646962997443, "grad_norm": 1.8066562414169312, "learning_rate": 0.0005, "loss": 2.9284415245056152, "step": 3460 }, { "epoch": 0.4027228298813125, "grad_norm": 1.25031578540802, "learning_rate": 0.0005, "loss": 2.2672972679138184, "step": 3461 }, { "epoch": 0.4028391901326507, "grad_norm": 1.3227845430374146, "learning_rate": 0.0005, "loss": 2.751176595687866, "step": 3462 }, { "epoch": 0.40295555038398884, "grad_norm": 1.5983668565750122, "learning_rate": 0.0005, "loss": 2.439910411834717, "step": 3463 }, { "epoch": 0.403071910635327, "grad_norm": 1.419299840927124, "learning_rate": 0.0005, "loss": 3.2501394748687744, "step": 3464 }, { "epoch": 0.4031882708866651, "grad_norm": 1.4388278722763062, "learning_rate": 0.0005, "loss": 3.2354469299316406, "step": 3465 }, { "epoch": 0.40330463113800324, "grad_norm": 1.365867257118225, "learning_rate": 0.0005, "loss": 2.5611283779144287, "step": 3466 }, { "epoch": 0.4034209913893414, "grad_norm": 1.2442957162857056, "learning_rate": 0.0005, "loss": 1.9734270572662354, "step": 3467 }, { "epoch": 0.40353735164067955, "grad_norm": 1.5537896156311035, "learning_rate": 0.0005, "loss": 2.3390417098999023, "step": 3468 }, { "epoch": 0.4036537118920177, "grad_norm": 1.3991401195526123, "learning_rate": 0.0005, "loss": 2.0110058784484863, "step": 3469 }, { "epoch": 0.4037700721433558, "grad_norm": 1.1634231805801392, "learning_rate": 0.0005, "loss": 2.440565586090088, "step": 3470 }, { "epoch": 0.40388643239469396, "grad_norm": 1.1711775064468384, "learning_rate": 0.0005, "loss": 1.6491062641143799, "step": 3471 }, { "epoch": 0.4040027926460321, "grad_norm": 1.429988980293274, "learning_rate": 0.0005, "loss": 3.800370216369629, "step": 3472 }, { "epoch": 0.40411915289737027, "grad_norm": 1.1159085035324097, "learning_rate": 0.0005, "loss": 1.5777630805969238, "step": 3473 }, { "epoch": 0.4042355131487084, "grad_norm": 1.2396644353866577, "learning_rate": 0.0005, "loss": 3.563307285308838, "step": 3474 }, { "epoch": 0.4043518734000465, "grad_norm": 0.8534581065177917, "learning_rate": 0.0005, "loss": 1.8689137697219849, "step": 3475 }, { "epoch": 0.4044682336513847, "grad_norm": 1.4965249300003052, "learning_rate": 0.0005, "loss": 3.060573101043701, "step": 3476 }, { "epoch": 0.40458459390272283, "grad_norm": 1.0251760482788086, "learning_rate": 0.0005, "loss": 2.0125808715820312, "step": 3477 }, { "epoch": 0.404700954154061, "grad_norm": 1.0081892013549805, "learning_rate": 0.0005, "loss": 2.2923247814178467, "step": 3478 }, { "epoch": 0.40481731440539914, "grad_norm": 1.3935294151306152, "learning_rate": 0.0005, "loss": 1.9999675750732422, "step": 3479 }, { "epoch": 0.40493367465673724, "grad_norm": 1.415260910987854, "learning_rate": 0.0005, "loss": 2.996288776397705, "step": 3480 }, { "epoch": 0.4050500349080754, "grad_norm": 2.030099630355835, "learning_rate": 0.0005, "loss": 2.9420671463012695, "step": 3481 }, { "epoch": 0.40516639515941355, "grad_norm": 1.7251441478729248, "learning_rate": 0.0005, "loss": 1.7402946949005127, "step": 3482 }, { "epoch": 0.4052827554107517, "grad_norm": 2.2009220123291016, "learning_rate": 0.0005, "loss": 3.0408623218536377, "step": 3483 }, { "epoch": 0.4053991156620898, "grad_norm": 1.7132964134216309, "learning_rate": 0.0005, "loss": 2.4259486198425293, "step": 3484 }, { "epoch": 0.40551547591342796, "grad_norm": 1.3458764553070068, "learning_rate": 0.0005, "loss": 2.9775471687316895, "step": 3485 }, { "epoch": 0.4056318361647661, "grad_norm": 1.6004714965820312, "learning_rate": 0.0005, "loss": 2.8444042205810547, "step": 3486 }, { "epoch": 0.40574819641610427, "grad_norm": 2.0514039993286133, "learning_rate": 0.0005, "loss": 2.3752403259277344, "step": 3487 }, { "epoch": 0.4058645566674424, "grad_norm": 0.9381067156791687, "learning_rate": 0.0005, "loss": 2.256075382232666, "step": 3488 }, { "epoch": 0.4059809169187805, "grad_norm": 1.4619406461715698, "learning_rate": 0.0005, "loss": 3.2130205631256104, "step": 3489 }, { "epoch": 0.4060972771701187, "grad_norm": 1.5200763940811157, "learning_rate": 0.0005, "loss": 2.135531425476074, "step": 3490 }, { "epoch": 0.40621363742145683, "grad_norm": 2.256272792816162, "learning_rate": 0.0005, "loss": 1.9476186037063599, "step": 3491 }, { "epoch": 0.406329997672795, "grad_norm": 2.034168243408203, "learning_rate": 0.0005, "loss": 2.8924460411071777, "step": 3492 }, { "epoch": 0.40644635792413314, "grad_norm": 1.9229998588562012, "learning_rate": 0.0005, "loss": 2.8357880115509033, "step": 3493 }, { "epoch": 0.40656271817547124, "grad_norm": 1.314518928527832, "learning_rate": 0.0005, "loss": 3.097031593322754, "step": 3494 }, { "epoch": 0.4066790784268094, "grad_norm": 1.8849035501480103, "learning_rate": 0.0005, "loss": 3.565446615219116, "step": 3495 }, { "epoch": 0.40679543867814755, "grad_norm": 1.7028270959854126, "learning_rate": 0.0005, "loss": 2.2790184020996094, "step": 3496 }, { "epoch": 0.4069117989294857, "grad_norm": 1.590895414352417, "learning_rate": 0.0005, "loss": 1.5502769947052002, "step": 3497 }, { "epoch": 0.40702815918082386, "grad_norm": 1.4986270666122437, "learning_rate": 0.0005, "loss": 2.575504779815674, "step": 3498 }, { "epoch": 0.40714451943216196, "grad_norm": 1.4316084384918213, "learning_rate": 0.0005, "loss": 2.9476194381713867, "step": 3499 }, { "epoch": 0.4072608796835001, "grad_norm": 1.3804514408111572, "learning_rate": 0.0005, "loss": 3.5693275928497314, "step": 3500 }, { "epoch": 0.40737723993483826, "grad_norm": 1.5379316806793213, "learning_rate": 0.0005, "loss": 3.542019844055176, "step": 3501 }, { "epoch": 0.4074936001861764, "grad_norm": 1.2096043825149536, "learning_rate": 0.0005, "loss": 2.5188403129577637, "step": 3502 }, { "epoch": 0.4076099604375145, "grad_norm": 1.6531848907470703, "learning_rate": 0.0005, "loss": 2.2543585300445557, "step": 3503 }, { "epoch": 0.4077263206888527, "grad_norm": 1.2272229194641113, "learning_rate": 0.0005, "loss": 1.7143685817718506, "step": 3504 }, { "epoch": 0.4078426809401908, "grad_norm": 1.4841606616973877, "learning_rate": 0.0005, "loss": 2.24491548538208, "step": 3505 }, { "epoch": 0.407959041191529, "grad_norm": 0.7320953011512756, "learning_rate": 0.0005, "loss": 1.2203388214111328, "step": 3506 }, { "epoch": 0.40807540144286714, "grad_norm": 1.8650641441345215, "learning_rate": 0.0005, "loss": 2.801809310913086, "step": 3507 }, { "epoch": 0.40819176169420524, "grad_norm": 2.1569042205810547, "learning_rate": 0.0005, "loss": 2.592510461807251, "step": 3508 }, { "epoch": 0.4083081219455434, "grad_norm": 1.1470247507095337, "learning_rate": 0.0005, "loss": 2.7674431800842285, "step": 3509 }, { "epoch": 0.40842448219688154, "grad_norm": 2.0646238327026367, "learning_rate": 0.0005, "loss": 2.996159315109253, "step": 3510 }, { "epoch": 0.4085408424482197, "grad_norm": 1.2491809129714966, "learning_rate": 0.0005, "loss": 2.595968723297119, "step": 3511 }, { "epoch": 0.40865720269955785, "grad_norm": 1.2473047971725464, "learning_rate": 0.0005, "loss": 2.94221830368042, "step": 3512 }, { "epoch": 0.40877356295089595, "grad_norm": 1.4643296003341675, "learning_rate": 0.0005, "loss": 2.6252408027648926, "step": 3513 }, { "epoch": 0.4088899232022341, "grad_norm": 2.109937906265259, "learning_rate": 0.0005, "loss": 2.6409008502960205, "step": 3514 }, { "epoch": 0.40900628345357226, "grad_norm": 0.8623331189155579, "learning_rate": 0.0005, "loss": 1.864322304725647, "step": 3515 }, { "epoch": 0.4091226437049104, "grad_norm": 1.2388205528259277, "learning_rate": 0.0005, "loss": 2.1657779216766357, "step": 3516 }, { "epoch": 0.40923900395624857, "grad_norm": 1.750732421875, "learning_rate": 0.0005, "loss": 2.707200527191162, "step": 3517 }, { "epoch": 0.40935536420758667, "grad_norm": 1.1383450031280518, "learning_rate": 0.0005, "loss": 1.9319121837615967, "step": 3518 }, { "epoch": 0.4094717244589248, "grad_norm": 2.0286970138549805, "learning_rate": 0.0005, "loss": 3.1740405559539795, "step": 3519 }, { "epoch": 0.409588084710263, "grad_norm": 0.9973896741867065, "learning_rate": 0.0005, "loss": 2.248905658721924, "step": 3520 }, { "epoch": 0.40970444496160113, "grad_norm": 1.4808131456375122, "learning_rate": 0.0005, "loss": 3.422138214111328, "step": 3521 }, { "epoch": 0.4098208052129393, "grad_norm": 1.424363136291504, "learning_rate": 0.0005, "loss": 2.6180131435394287, "step": 3522 }, { "epoch": 0.4099371654642774, "grad_norm": 0.7695171236991882, "learning_rate": 0.0005, "loss": 2.1648173332214355, "step": 3523 }, { "epoch": 0.41005352571561554, "grad_norm": 1.5796315670013428, "learning_rate": 0.0005, "loss": 2.792572021484375, "step": 3524 }, { "epoch": 0.4101698859669537, "grad_norm": 1.1115000247955322, "learning_rate": 0.0005, "loss": 1.977445125579834, "step": 3525 }, { "epoch": 0.41028624621829185, "grad_norm": 1.5181083679199219, "learning_rate": 0.0005, "loss": 2.845797061920166, "step": 3526 }, { "epoch": 0.41040260646962995, "grad_norm": 1.0156209468841553, "learning_rate": 0.0005, "loss": 2.336177110671997, "step": 3527 }, { "epoch": 0.4105189667209681, "grad_norm": 1.5638571977615356, "learning_rate": 0.0005, "loss": 1.873497724533081, "step": 3528 }, { "epoch": 0.41063532697230626, "grad_norm": 2.0202696323394775, "learning_rate": 0.0005, "loss": 2.7848024368286133, "step": 3529 }, { "epoch": 0.4107516872236444, "grad_norm": 1.7973953485488892, "learning_rate": 0.0005, "loss": 2.9848949909210205, "step": 3530 }, { "epoch": 0.41086804747498257, "grad_norm": 2.6115145683288574, "learning_rate": 0.0005, "loss": 2.241391658782959, "step": 3531 }, { "epoch": 0.41098440772632067, "grad_norm": 1.2067991495132446, "learning_rate": 0.0005, "loss": 2.7484350204467773, "step": 3532 }, { "epoch": 0.4111007679776588, "grad_norm": 1.6918928623199463, "learning_rate": 0.0005, "loss": 2.808673143386841, "step": 3533 }, { "epoch": 0.411217128228997, "grad_norm": 1.4854204654693604, "learning_rate": 0.0005, "loss": 2.973616123199463, "step": 3534 }, { "epoch": 0.41133348848033513, "grad_norm": 2.7242751121520996, "learning_rate": 0.0005, "loss": 3.010812759399414, "step": 3535 }, { "epoch": 0.4114498487316733, "grad_norm": 1.7647991180419922, "learning_rate": 0.0005, "loss": 2.9339113235473633, "step": 3536 }, { "epoch": 0.4115662089830114, "grad_norm": 1.2123749256134033, "learning_rate": 0.0005, "loss": 3.2288122177124023, "step": 3537 }, { "epoch": 0.41168256923434954, "grad_norm": 1.9699888229370117, "learning_rate": 0.0005, "loss": 2.5830869674682617, "step": 3538 }, { "epoch": 0.4117989294856877, "grad_norm": 1.4758180379867554, "learning_rate": 0.0005, "loss": 2.7381110191345215, "step": 3539 }, { "epoch": 0.41191528973702585, "grad_norm": 2.2389466762542725, "learning_rate": 0.0005, "loss": 2.87027645111084, "step": 3540 }, { "epoch": 0.412031649988364, "grad_norm": 1.2605321407318115, "learning_rate": 0.0005, "loss": 1.6688376665115356, "step": 3541 }, { "epoch": 0.4121480102397021, "grad_norm": 1.26254141330719, "learning_rate": 0.0005, "loss": 2.48075532913208, "step": 3542 }, { "epoch": 0.41226437049104026, "grad_norm": 2.2717854976654053, "learning_rate": 0.0005, "loss": 2.6551918983459473, "step": 3543 }, { "epoch": 0.4123807307423784, "grad_norm": 1.3070719242095947, "learning_rate": 0.0005, "loss": 2.3657145500183105, "step": 3544 }, { "epoch": 0.41249709099371656, "grad_norm": 1.527243971824646, "learning_rate": 0.0005, "loss": 2.2619824409484863, "step": 3545 }, { "epoch": 0.41261345124505466, "grad_norm": 1.7900511026382446, "learning_rate": 0.0005, "loss": 2.6733598709106445, "step": 3546 }, { "epoch": 0.4127298114963928, "grad_norm": 1.5168720483779907, "learning_rate": 0.0005, "loss": 2.742802858352661, "step": 3547 }, { "epoch": 0.412846171747731, "grad_norm": 1.095089077949524, "learning_rate": 0.0005, "loss": 2.9814319610595703, "step": 3548 }, { "epoch": 0.4129625319990691, "grad_norm": 2.113933801651001, "learning_rate": 0.0005, "loss": 3.125924587249756, "step": 3549 }, { "epoch": 0.4130788922504073, "grad_norm": 1.5247474908828735, "learning_rate": 0.0005, "loss": 2.3548786640167236, "step": 3550 }, { "epoch": 0.4131952525017454, "grad_norm": 1.3845670223236084, "learning_rate": 0.0005, "loss": 2.483227252960205, "step": 3551 }, { "epoch": 0.41331161275308353, "grad_norm": 2.1676180362701416, "learning_rate": 0.0005, "loss": 2.637173652648926, "step": 3552 }, { "epoch": 0.4134279730044217, "grad_norm": 1.1807361841201782, "learning_rate": 0.0005, "loss": 2.2067461013793945, "step": 3553 }, { "epoch": 0.41354433325575984, "grad_norm": 1.184105396270752, "learning_rate": 0.0005, "loss": 3.390657424926758, "step": 3554 }, { "epoch": 0.413660693507098, "grad_norm": 1.8353511095046997, "learning_rate": 0.0005, "loss": 3.6757688522338867, "step": 3555 }, { "epoch": 0.4137770537584361, "grad_norm": 1.3246452808380127, "learning_rate": 0.0005, "loss": 2.2440967559814453, "step": 3556 }, { "epoch": 0.41389341400977425, "grad_norm": 1.9684422016143799, "learning_rate": 0.0005, "loss": 2.9433085918426514, "step": 3557 }, { "epoch": 0.4140097742611124, "grad_norm": 1.3999030590057373, "learning_rate": 0.0005, "loss": 2.9788053035736084, "step": 3558 }, { "epoch": 0.41412613451245056, "grad_norm": 3.8911449909210205, "learning_rate": 0.0005, "loss": 1.966365098953247, "step": 3559 }, { "epoch": 0.4142424947637887, "grad_norm": 1.8955718278884888, "learning_rate": 0.0005, "loss": 2.3041434288024902, "step": 3560 }, { "epoch": 0.4143588550151268, "grad_norm": 1.9681140184402466, "learning_rate": 0.0005, "loss": 2.3662779331207275, "step": 3561 }, { "epoch": 0.41447521526646497, "grad_norm": 1.1951572895050049, "learning_rate": 0.0005, "loss": 1.662512183189392, "step": 3562 }, { "epoch": 0.4145915755178031, "grad_norm": 1.5490957498550415, "learning_rate": 0.0005, "loss": 2.1748223304748535, "step": 3563 }, { "epoch": 0.4147079357691413, "grad_norm": 1.2144601345062256, "learning_rate": 0.0005, "loss": 3.3996100425720215, "step": 3564 }, { "epoch": 0.4148242960204794, "grad_norm": 0.8861011266708374, "learning_rate": 0.0005, "loss": 1.7157007455825806, "step": 3565 }, { "epoch": 0.41494065627181753, "grad_norm": 1.2893201112747192, "learning_rate": 0.0005, "loss": 2.070387125015259, "step": 3566 }, { "epoch": 0.4150570165231557, "grad_norm": 1.5782955884933472, "learning_rate": 0.0005, "loss": 2.5110971927642822, "step": 3567 }, { "epoch": 0.41517337677449384, "grad_norm": 1.4998327493667603, "learning_rate": 0.0005, "loss": 2.17586350440979, "step": 3568 }, { "epoch": 0.415289737025832, "grad_norm": 1.5425536632537842, "learning_rate": 0.0005, "loss": 2.948335886001587, "step": 3569 }, { "epoch": 0.4154060972771701, "grad_norm": 1.2172563076019287, "learning_rate": 0.0005, "loss": 2.5723624229431152, "step": 3570 }, { "epoch": 0.41552245752850825, "grad_norm": 1.6359590291976929, "learning_rate": 0.0005, "loss": 2.5797085762023926, "step": 3571 }, { "epoch": 0.4156388177798464, "grad_norm": 1.8565596342086792, "learning_rate": 0.0005, "loss": 2.228938102722168, "step": 3572 }, { "epoch": 0.41575517803118456, "grad_norm": 1.8172850608825684, "learning_rate": 0.0005, "loss": 2.9924135208129883, "step": 3573 }, { "epoch": 0.4158715382825227, "grad_norm": 1.3641587495803833, "learning_rate": 0.0005, "loss": 2.3789782524108887, "step": 3574 }, { "epoch": 0.4159878985338608, "grad_norm": 1.5178261995315552, "learning_rate": 0.0005, "loss": 2.370457172393799, "step": 3575 }, { "epoch": 0.41610425878519897, "grad_norm": 1.705771565437317, "learning_rate": 0.0005, "loss": 3.3448948860168457, "step": 3576 }, { "epoch": 0.4162206190365371, "grad_norm": 1.1034334897994995, "learning_rate": 0.0005, "loss": 2.58807373046875, "step": 3577 }, { "epoch": 0.4163369792878753, "grad_norm": 1.4296990633010864, "learning_rate": 0.0005, "loss": 2.3195230960845947, "step": 3578 }, { "epoch": 0.41645333953921343, "grad_norm": 0.8533658385276794, "learning_rate": 0.0005, "loss": 2.6584677696228027, "step": 3579 }, { "epoch": 0.41656969979055153, "grad_norm": 2.1617791652679443, "learning_rate": 0.0005, "loss": 2.583261013031006, "step": 3580 }, { "epoch": 0.4166860600418897, "grad_norm": 0.8590340614318848, "learning_rate": 0.0005, "loss": 1.911232590675354, "step": 3581 }, { "epoch": 0.41680242029322784, "grad_norm": 0.9799509644508362, "learning_rate": 0.0005, "loss": 2.28127121925354, "step": 3582 }, { "epoch": 0.416918780544566, "grad_norm": 2.7000386714935303, "learning_rate": 0.0005, "loss": 3.508164644241333, "step": 3583 }, { "epoch": 0.41703514079590415, "grad_norm": 1.4346486330032349, "learning_rate": 0.0005, "loss": 3.116610527038574, "step": 3584 }, { "epoch": 0.41715150104724225, "grad_norm": 1.582090139389038, "learning_rate": 0.0005, "loss": 2.3776612281799316, "step": 3585 }, { "epoch": 0.4172678612985804, "grad_norm": 2.3102362155914307, "learning_rate": 0.0005, "loss": 3.3677186965942383, "step": 3586 }, { "epoch": 0.41738422154991855, "grad_norm": 1.3214284181594849, "learning_rate": 0.0005, "loss": 2.491597890853882, "step": 3587 }, { "epoch": 0.4175005818012567, "grad_norm": 1.3614765405654907, "learning_rate": 0.0005, "loss": 2.6559863090515137, "step": 3588 }, { "epoch": 0.4176169420525948, "grad_norm": 0.7504866123199463, "learning_rate": 0.0005, "loss": 1.9781677722930908, "step": 3589 }, { "epoch": 0.41773330230393296, "grad_norm": 1.1819809675216675, "learning_rate": 0.0005, "loss": 2.8996753692626953, "step": 3590 }, { "epoch": 0.4178496625552711, "grad_norm": 1.909773349761963, "learning_rate": 0.0005, "loss": 2.6821818351745605, "step": 3591 }, { "epoch": 0.41796602280660927, "grad_norm": 1.5054854154586792, "learning_rate": 0.0005, "loss": 2.9628522396087646, "step": 3592 }, { "epoch": 0.4180823830579474, "grad_norm": 1.1736552715301514, "learning_rate": 0.0005, "loss": 2.4873199462890625, "step": 3593 }, { "epoch": 0.4181987433092855, "grad_norm": 0.8674280047416687, "learning_rate": 0.0005, "loss": 1.512953519821167, "step": 3594 }, { "epoch": 0.4183151035606237, "grad_norm": 1.504006266593933, "learning_rate": 0.0005, "loss": 2.5647482872009277, "step": 3595 }, { "epoch": 0.41843146381196183, "grad_norm": 1.185025691986084, "learning_rate": 0.0005, "loss": 2.341102123260498, "step": 3596 }, { "epoch": 0.4185478240633, "grad_norm": 1.5841528177261353, "learning_rate": 0.0005, "loss": 3.618896961212158, "step": 3597 }, { "epoch": 0.41866418431463814, "grad_norm": 1.5419975519180298, "learning_rate": 0.0005, "loss": 2.897462844848633, "step": 3598 }, { "epoch": 0.41878054456597624, "grad_norm": 1.0961004495620728, "learning_rate": 0.0005, "loss": 2.4912588596343994, "step": 3599 }, { "epoch": 0.4188969048173144, "grad_norm": 1.169926404953003, "learning_rate": 0.0005, "loss": 2.977165937423706, "step": 3600 }, { "epoch": 0.41901326506865255, "grad_norm": 1.3080471754074097, "learning_rate": 0.0005, "loss": 1.9986965656280518, "step": 3601 }, { "epoch": 0.4191296253199907, "grad_norm": 1.2221800088882446, "learning_rate": 0.0005, "loss": 2.4210925102233887, "step": 3602 }, { "epoch": 0.41924598557132886, "grad_norm": 1.3224883079528809, "learning_rate": 0.0005, "loss": 2.7394845485687256, "step": 3603 }, { "epoch": 0.41936234582266696, "grad_norm": 1.226483702659607, "learning_rate": 0.0005, "loss": 2.685628652572632, "step": 3604 }, { "epoch": 0.4194787060740051, "grad_norm": 1.2874228954315186, "learning_rate": 0.0005, "loss": 2.0431387424468994, "step": 3605 }, { "epoch": 0.41959506632534327, "grad_norm": 1.3712419271469116, "learning_rate": 0.0005, "loss": 3.2047247886657715, "step": 3606 }, { "epoch": 0.4197114265766814, "grad_norm": 1.2295993566513062, "learning_rate": 0.0005, "loss": 3.0296878814697266, "step": 3607 }, { "epoch": 0.4198277868280195, "grad_norm": 1.2701144218444824, "learning_rate": 0.0005, "loss": 1.9794080257415771, "step": 3608 }, { "epoch": 0.4199441470793577, "grad_norm": 1.904185175895691, "learning_rate": 0.0005, "loss": 3.034351348876953, "step": 3609 }, { "epoch": 0.42006050733069583, "grad_norm": 2.7056307792663574, "learning_rate": 0.0005, "loss": 2.4514007568359375, "step": 3610 }, { "epoch": 0.420176867582034, "grad_norm": 1.5293413400650024, "learning_rate": 0.0005, "loss": 2.165339469909668, "step": 3611 }, { "epoch": 0.42029322783337214, "grad_norm": 0.4274660348892212, "learning_rate": 0.0005, "loss": 1.0596938133239746, "step": 3612 }, { "epoch": 0.42040958808471024, "grad_norm": 2.458397388458252, "learning_rate": 0.0005, "loss": 2.33536958694458, "step": 3613 }, { "epoch": 0.4205259483360484, "grad_norm": 1.9890453815460205, "learning_rate": 0.0005, "loss": 2.6575493812561035, "step": 3614 }, { "epoch": 0.42064230858738655, "grad_norm": 1.0464259386062622, "learning_rate": 0.0005, "loss": 1.3921021223068237, "step": 3615 }, { "epoch": 0.4207586688387247, "grad_norm": 1.7667717933654785, "learning_rate": 0.0005, "loss": 3.3766393661499023, "step": 3616 }, { "epoch": 0.42087502909006286, "grad_norm": 1.723868727684021, "learning_rate": 0.0005, "loss": 2.9594104290008545, "step": 3617 }, { "epoch": 0.42099138934140096, "grad_norm": 1.617846965789795, "learning_rate": 0.0005, "loss": 2.8117966651916504, "step": 3618 }, { "epoch": 0.4211077495927391, "grad_norm": 1.2414406538009644, "learning_rate": 0.0005, "loss": 1.689602255821228, "step": 3619 }, { "epoch": 0.42122410984407727, "grad_norm": 1.401458740234375, "learning_rate": 0.0005, "loss": 2.576211929321289, "step": 3620 }, { "epoch": 0.4213404700954154, "grad_norm": 1.8264671564102173, "learning_rate": 0.0005, "loss": 2.7954070568084717, "step": 3621 }, { "epoch": 0.4214568303467536, "grad_norm": 1.6356277465820312, "learning_rate": 0.0005, "loss": 3.3127527236938477, "step": 3622 }, { "epoch": 0.4215731905980917, "grad_norm": 1.4639359712600708, "learning_rate": 0.0005, "loss": 3.6319375038146973, "step": 3623 }, { "epoch": 0.42168955084942983, "grad_norm": 1.9099329710006714, "learning_rate": 0.0005, "loss": 2.9397594928741455, "step": 3624 }, { "epoch": 0.421805911100768, "grad_norm": 1.4313874244689941, "learning_rate": 0.0005, "loss": 3.1826539039611816, "step": 3625 }, { "epoch": 0.42192227135210614, "grad_norm": 2.038041830062866, "learning_rate": 0.0005, "loss": 3.695927143096924, "step": 3626 }, { "epoch": 0.42203863160344424, "grad_norm": 1.4965035915374756, "learning_rate": 0.0005, "loss": 3.6060404777526855, "step": 3627 }, { "epoch": 0.4221549918547824, "grad_norm": 1.8160654306411743, "learning_rate": 0.0005, "loss": 2.7220845222473145, "step": 3628 }, { "epoch": 0.42227135210612055, "grad_norm": 1.489425539970398, "learning_rate": 0.0005, "loss": 3.287504196166992, "step": 3629 }, { "epoch": 0.4223877123574587, "grad_norm": 1.9554033279418945, "learning_rate": 0.0005, "loss": 3.019564628601074, "step": 3630 }, { "epoch": 0.42250407260879685, "grad_norm": 1.3467458486557007, "learning_rate": 0.0005, "loss": 3.2945470809936523, "step": 3631 }, { "epoch": 0.42262043286013495, "grad_norm": 0.7948794364929199, "learning_rate": 0.0005, "loss": 1.7366547584533691, "step": 3632 }, { "epoch": 0.4227367931114731, "grad_norm": 1.2662813663482666, "learning_rate": 0.0005, "loss": 3.066089153289795, "step": 3633 }, { "epoch": 0.42285315336281126, "grad_norm": 1.809550166130066, "learning_rate": 0.0005, "loss": 2.703516960144043, "step": 3634 }, { "epoch": 0.4229695136141494, "grad_norm": 1.0288972854614258, "learning_rate": 0.0005, "loss": 1.8528103828430176, "step": 3635 }, { "epoch": 0.42308587386548757, "grad_norm": 0.9857488870620728, "learning_rate": 0.0005, "loss": 1.9716130495071411, "step": 3636 }, { "epoch": 0.42320223411682567, "grad_norm": 3.043346643447876, "learning_rate": 0.0005, "loss": 3.3724241256713867, "step": 3637 }, { "epoch": 0.4233185943681638, "grad_norm": 1.5917909145355225, "learning_rate": 0.0005, "loss": 3.2433018684387207, "step": 3638 }, { "epoch": 0.423434954619502, "grad_norm": 1.6312276124954224, "learning_rate": 0.0005, "loss": 2.684969902038574, "step": 3639 }, { "epoch": 0.42355131487084013, "grad_norm": 1.6320081949234009, "learning_rate": 0.0005, "loss": 2.876173734664917, "step": 3640 }, { "epoch": 0.4236676751221783, "grad_norm": 1.4431082010269165, "learning_rate": 0.0005, "loss": 2.7278342247009277, "step": 3641 }, { "epoch": 0.4237840353735164, "grad_norm": 1.554728388786316, "learning_rate": 0.0005, "loss": 2.786761999130249, "step": 3642 }, { "epoch": 0.42390039562485454, "grad_norm": 1.5864334106445312, "learning_rate": 0.0005, "loss": 2.7503933906555176, "step": 3643 }, { "epoch": 0.4240167558761927, "grad_norm": 1.1946243047714233, "learning_rate": 0.0005, "loss": 2.6281378269195557, "step": 3644 }, { "epoch": 0.42413311612753085, "grad_norm": 0.6528477072715759, "learning_rate": 0.0005, "loss": 1.4663050174713135, "step": 3645 }, { "epoch": 0.424249476378869, "grad_norm": 0.9928905963897705, "learning_rate": 0.0005, "loss": 2.0969371795654297, "step": 3646 }, { "epoch": 0.4243658366302071, "grad_norm": 1.8008027076721191, "learning_rate": 0.0005, "loss": 2.7008633613586426, "step": 3647 }, { "epoch": 0.42448219688154526, "grad_norm": 1.2571237087249756, "learning_rate": 0.0005, "loss": 2.6728668212890625, "step": 3648 }, { "epoch": 0.4245985571328834, "grad_norm": 0.7195775508880615, "learning_rate": 0.0005, "loss": 1.993428349494934, "step": 3649 }, { "epoch": 0.42471491738422157, "grad_norm": 1.2190779447555542, "learning_rate": 0.0005, "loss": 2.1115670204162598, "step": 3650 }, { "epoch": 0.42483127763555967, "grad_norm": 1.4059826135635376, "learning_rate": 0.0005, "loss": 2.9094271659851074, "step": 3651 }, { "epoch": 0.4249476378868978, "grad_norm": 1.0463223457336426, "learning_rate": 0.0005, "loss": 1.9074562788009644, "step": 3652 }, { "epoch": 0.425063998138236, "grad_norm": 1.156882643699646, "learning_rate": 0.0005, "loss": 2.8695883750915527, "step": 3653 }, { "epoch": 0.42518035838957413, "grad_norm": 1.3313803672790527, "learning_rate": 0.0005, "loss": 2.949838161468506, "step": 3654 }, { "epoch": 0.4252967186409123, "grad_norm": 0.8142552971839905, "learning_rate": 0.0005, "loss": 2.207890510559082, "step": 3655 }, { "epoch": 0.4254130788922504, "grad_norm": 0.7936264276504517, "learning_rate": 0.0005, "loss": 1.8728162050247192, "step": 3656 }, { "epoch": 0.42552943914358854, "grad_norm": 1.2206989526748657, "learning_rate": 0.0005, "loss": 2.345040798187256, "step": 3657 }, { "epoch": 0.4256457993949267, "grad_norm": 1.550908088684082, "learning_rate": 0.0005, "loss": 2.6253180503845215, "step": 3658 }, { "epoch": 0.42576215964626485, "grad_norm": 1.4484202861785889, "learning_rate": 0.0005, "loss": 2.5280203819274902, "step": 3659 }, { "epoch": 0.425878519897603, "grad_norm": 3.371542453765869, "learning_rate": 0.0005, "loss": 1.9407641887664795, "step": 3660 }, { "epoch": 0.4259948801489411, "grad_norm": 1.8102643489837646, "learning_rate": 0.0005, "loss": 2.3917629718780518, "step": 3661 }, { "epoch": 0.42611124040027926, "grad_norm": 1.534664273262024, "learning_rate": 0.0005, "loss": 2.9506468772888184, "step": 3662 }, { "epoch": 0.4262276006516174, "grad_norm": 1.9613505601882935, "learning_rate": 0.0005, "loss": 3.6236960887908936, "step": 3663 }, { "epoch": 0.42634396090295557, "grad_norm": 1.6049607992172241, "learning_rate": 0.0005, "loss": 2.5563101768493652, "step": 3664 }, { "epoch": 0.4264603211542937, "grad_norm": 1.2006189823150635, "learning_rate": 0.0005, "loss": 2.2360522747039795, "step": 3665 }, { "epoch": 0.4265766814056318, "grad_norm": 1.5209678411483765, "learning_rate": 0.0005, "loss": 2.830078125, "step": 3666 }, { "epoch": 0.42669304165697, "grad_norm": 2.397660493850708, "learning_rate": 0.0005, "loss": 2.221479654312134, "step": 3667 }, { "epoch": 0.42680940190830813, "grad_norm": 1.9745205640792847, "learning_rate": 0.0005, "loss": 2.5051109790802, "step": 3668 }, { "epoch": 0.4269257621596463, "grad_norm": 1.709424376487732, "learning_rate": 0.0005, "loss": 1.7027350664138794, "step": 3669 }, { "epoch": 0.4270421224109844, "grad_norm": 0.8886062502861023, "learning_rate": 0.0005, "loss": 1.695128321647644, "step": 3670 }, { "epoch": 0.42715848266232254, "grad_norm": 1.4019606113433838, "learning_rate": 0.0005, "loss": 3.8063549995422363, "step": 3671 }, { "epoch": 0.4272748429136607, "grad_norm": 1.6107099056243896, "learning_rate": 0.0005, "loss": 2.7721550464630127, "step": 3672 }, { "epoch": 0.42739120316499885, "grad_norm": 1.2567472457885742, "learning_rate": 0.0005, "loss": 2.744962215423584, "step": 3673 }, { "epoch": 0.427507563416337, "grad_norm": 1.2993204593658447, "learning_rate": 0.0005, "loss": 3.2468366622924805, "step": 3674 }, { "epoch": 0.4276239236676751, "grad_norm": 1.2157272100448608, "learning_rate": 0.0005, "loss": 2.759719133377075, "step": 3675 }, { "epoch": 0.42774028391901325, "grad_norm": 1.3966912031173706, "learning_rate": 0.0005, "loss": 3.0014352798461914, "step": 3676 }, { "epoch": 0.4278566441703514, "grad_norm": 0.9185543656349182, "learning_rate": 0.0005, "loss": 2.0358290672302246, "step": 3677 }, { "epoch": 0.42797300442168956, "grad_norm": 1.5540215969085693, "learning_rate": 0.0005, "loss": 2.3407299518585205, "step": 3678 }, { "epoch": 0.4280893646730277, "grad_norm": 2.089446783065796, "learning_rate": 0.0005, "loss": 2.0532965660095215, "step": 3679 }, { "epoch": 0.4282057249243658, "grad_norm": 1.3941940069198608, "learning_rate": 0.0005, "loss": 2.681015968322754, "step": 3680 }, { "epoch": 0.42832208517570397, "grad_norm": 1.4089813232421875, "learning_rate": 0.0005, "loss": 1.7332348823547363, "step": 3681 }, { "epoch": 0.4284384454270421, "grad_norm": 1.7346802949905396, "learning_rate": 0.0005, "loss": 2.4943153858184814, "step": 3682 }, { "epoch": 0.4285548056783803, "grad_norm": 2.115389823913574, "learning_rate": 0.0005, "loss": 2.1699328422546387, "step": 3683 }, { "epoch": 0.42867116592971843, "grad_norm": 1.6138417720794678, "learning_rate": 0.0005, "loss": 3.0387978553771973, "step": 3684 }, { "epoch": 0.42878752618105653, "grad_norm": 2.149381160736084, "learning_rate": 0.0005, "loss": 2.0581798553466797, "step": 3685 }, { "epoch": 0.4289038864323947, "grad_norm": 1.51939857006073, "learning_rate": 0.0005, "loss": 2.8367881774902344, "step": 3686 }, { "epoch": 0.42902024668373284, "grad_norm": 1.596615195274353, "learning_rate": 0.0005, "loss": 2.658177375793457, "step": 3687 }, { "epoch": 0.429136606935071, "grad_norm": 1.6306709051132202, "learning_rate": 0.0005, "loss": 3.5806922912597656, "step": 3688 }, { "epoch": 0.4292529671864091, "grad_norm": 1.8310528993606567, "learning_rate": 0.0005, "loss": 3.9797425270080566, "step": 3689 }, { "epoch": 0.42936932743774725, "grad_norm": 1.2711997032165527, "learning_rate": 0.0005, "loss": 2.982862949371338, "step": 3690 }, { "epoch": 0.4294856876890854, "grad_norm": 2.49755597114563, "learning_rate": 0.0005, "loss": 3.0518012046813965, "step": 3691 }, { "epoch": 0.42960204794042356, "grad_norm": 1.3145815134048462, "learning_rate": 0.0005, "loss": 3.1835904121398926, "step": 3692 }, { "epoch": 0.4297184081917617, "grad_norm": 2.0115675926208496, "learning_rate": 0.0005, "loss": 2.7568130493164062, "step": 3693 }, { "epoch": 0.4298347684430998, "grad_norm": 1.341585636138916, "learning_rate": 0.0005, "loss": 1.7585442066192627, "step": 3694 }, { "epoch": 0.42995112869443797, "grad_norm": 1.7537519931793213, "learning_rate": 0.0005, "loss": 2.637637138366699, "step": 3695 }, { "epoch": 0.4300674889457761, "grad_norm": 1.0002870559692383, "learning_rate": 0.0005, "loss": 2.0058584213256836, "step": 3696 }, { "epoch": 0.4301838491971143, "grad_norm": 1.3972324132919312, "learning_rate": 0.0005, "loss": 2.1669068336486816, "step": 3697 }, { "epoch": 0.43030020944845243, "grad_norm": 1.2309833765029907, "learning_rate": 0.0005, "loss": 2.587634563446045, "step": 3698 }, { "epoch": 0.43041656969979053, "grad_norm": 1.3375413417816162, "learning_rate": 0.0005, "loss": 1.9408516883850098, "step": 3699 }, { "epoch": 0.4305329299511287, "grad_norm": 0.9838714599609375, "learning_rate": 0.0005, "loss": 1.7276105880737305, "step": 3700 }, { "epoch": 0.43064929020246684, "grad_norm": 1.448132038116455, "learning_rate": 0.0005, "loss": 2.9772891998291016, "step": 3701 }, { "epoch": 0.430765650453805, "grad_norm": 1.6756401062011719, "learning_rate": 0.0005, "loss": 3.0532679557800293, "step": 3702 }, { "epoch": 0.43088201070514315, "grad_norm": 1.517676591873169, "learning_rate": 0.0005, "loss": 2.698204517364502, "step": 3703 }, { "epoch": 0.43099837095648125, "grad_norm": 1.4468528032302856, "learning_rate": 0.0005, "loss": 2.226351737976074, "step": 3704 }, { "epoch": 0.4311147312078194, "grad_norm": 1.866858720779419, "learning_rate": 0.0005, "loss": 3.0890982151031494, "step": 3705 }, { "epoch": 0.43123109145915756, "grad_norm": 1.614814043045044, "learning_rate": 0.0005, "loss": 3.026437759399414, "step": 3706 }, { "epoch": 0.4313474517104957, "grad_norm": 1.2483257055282593, "learning_rate": 0.0005, "loss": 2.891122817993164, "step": 3707 }, { "epoch": 0.43146381196183387, "grad_norm": 0.9207040071487427, "learning_rate": 0.0005, "loss": 2.448378086090088, "step": 3708 }, { "epoch": 0.43158017221317196, "grad_norm": 1.1644073724746704, "learning_rate": 0.0005, "loss": 2.476518154144287, "step": 3709 }, { "epoch": 0.4316965324645101, "grad_norm": 2.203155517578125, "learning_rate": 0.0005, "loss": 2.9047560691833496, "step": 3710 }, { "epoch": 0.4318128927158483, "grad_norm": 1.2973066568374634, "learning_rate": 0.0005, "loss": 2.8553245067596436, "step": 3711 }, { "epoch": 0.43192925296718643, "grad_norm": 1.6444945335388184, "learning_rate": 0.0005, "loss": 3.2775821685791016, "step": 3712 }, { "epoch": 0.4320456132185245, "grad_norm": 0.8720571398735046, "learning_rate": 0.0005, "loss": 1.7631771564483643, "step": 3713 }, { "epoch": 0.4321619734698627, "grad_norm": 1.413110613822937, "learning_rate": 0.0005, "loss": 2.479773759841919, "step": 3714 }, { "epoch": 0.43227833372120084, "grad_norm": 1.2867331504821777, "learning_rate": 0.0005, "loss": 2.8120408058166504, "step": 3715 }, { "epoch": 0.432394693972539, "grad_norm": 1.9121387004852295, "learning_rate": 0.0005, "loss": 3.393521785736084, "step": 3716 }, { "epoch": 0.43251105422387714, "grad_norm": 1.5640455484390259, "learning_rate": 0.0005, "loss": 1.9650139808654785, "step": 3717 }, { "epoch": 0.43262741447521524, "grad_norm": 1.5569171905517578, "learning_rate": 0.0005, "loss": 2.248872756958008, "step": 3718 }, { "epoch": 0.4327437747265534, "grad_norm": 1.6100847721099854, "learning_rate": 0.0005, "loss": 3.7679169178009033, "step": 3719 }, { "epoch": 0.43286013497789155, "grad_norm": 2.456345558166504, "learning_rate": 0.0005, "loss": 3.117689609527588, "step": 3720 }, { "epoch": 0.4329764952292297, "grad_norm": 2.1089835166931152, "learning_rate": 0.0005, "loss": 2.4681553840637207, "step": 3721 }, { "epoch": 0.43309285548056786, "grad_norm": 0.8219454884529114, "learning_rate": 0.0005, "loss": 1.9488284587860107, "step": 3722 }, { "epoch": 0.43320921573190596, "grad_norm": 1.5674567222595215, "learning_rate": 0.0005, "loss": 3.245345115661621, "step": 3723 }, { "epoch": 0.4333255759832441, "grad_norm": 1.393102765083313, "learning_rate": 0.0005, "loss": 1.8400027751922607, "step": 3724 }, { "epoch": 0.43344193623458227, "grad_norm": 1.812125563621521, "learning_rate": 0.0005, "loss": 1.7487504482269287, "step": 3725 }, { "epoch": 0.4335582964859204, "grad_norm": 1.3555387258529663, "learning_rate": 0.0005, "loss": 2.7794113159179688, "step": 3726 }, { "epoch": 0.4336746567372586, "grad_norm": 1.259437918663025, "learning_rate": 0.0005, "loss": 2.2281858921051025, "step": 3727 }, { "epoch": 0.4337910169885967, "grad_norm": 1.4789035320281982, "learning_rate": 0.0005, "loss": 3.3063108921051025, "step": 3728 }, { "epoch": 0.43390737723993483, "grad_norm": 1.3517882823944092, "learning_rate": 0.0005, "loss": 2.348689556121826, "step": 3729 }, { "epoch": 0.434023737491273, "grad_norm": 1.3030098676681519, "learning_rate": 0.0005, "loss": 2.589365243911743, "step": 3730 }, { "epoch": 0.43414009774261114, "grad_norm": 1.3407851457595825, "learning_rate": 0.0005, "loss": 3.108969211578369, "step": 3731 }, { "epoch": 0.43425645799394924, "grad_norm": 1.6658273935317993, "learning_rate": 0.0005, "loss": 3.196077823638916, "step": 3732 }, { "epoch": 0.4343728182452874, "grad_norm": 1.443517804145813, "learning_rate": 0.0005, "loss": 2.4180960655212402, "step": 3733 }, { "epoch": 0.43448917849662555, "grad_norm": 1.2797372341156006, "learning_rate": 0.0005, "loss": 3.472383499145508, "step": 3734 }, { "epoch": 0.4346055387479637, "grad_norm": 1.9671756029129028, "learning_rate": 0.0005, "loss": 2.530095338821411, "step": 3735 }, { "epoch": 0.43472189899930186, "grad_norm": 2.2589223384857178, "learning_rate": 0.0005, "loss": 3.0656542778015137, "step": 3736 }, { "epoch": 0.43483825925063996, "grad_norm": 1.3518811464309692, "learning_rate": 0.0005, "loss": 1.8901716470718384, "step": 3737 }, { "epoch": 0.4349546195019781, "grad_norm": 1.0175700187683105, "learning_rate": 0.0005, "loss": 3.000602960586548, "step": 3738 }, { "epoch": 0.43507097975331627, "grad_norm": 1.241011619567871, "learning_rate": 0.0005, "loss": 2.959178924560547, "step": 3739 }, { "epoch": 0.4351873400046544, "grad_norm": 1.1562161445617676, "learning_rate": 0.0005, "loss": 1.693913459777832, "step": 3740 }, { "epoch": 0.4353037002559926, "grad_norm": 1.8160563707351685, "learning_rate": 0.0005, "loss": 1.7714872360229492, "step": 3741 }, { "epoch": 0.4354200605073307, "grad_norm": 0.7907192707061768, "learning_rate": 0.0005, "loss": 1.6665279865264893, "step": 3742 }, { "epoch": 0.43553642075866883, "grad_norm": 2.0314111709594727, "learning_rate": 0.0005, "loss": 2.440213203430176, "step": 3743 }, { "epoch": 0.435652781010007, "grad_norm": 0.9403818845748901, "learning_rate": 0.0005, "loss": 1.8908802270889282, "step": 3744 }, { "epoch": 0.43576914126134514, "grad_norm": 1.4760035276412964, "learning_rate": 0.0005, "loss": 2.62630558013916, "step": 3745 }, { "epoch": 0.4358855015126833, "grad_norm": 2.5968973636627197, "learning_rate": 0.0005, "loss": 2.875030040740967, "step": 3746 }, { "epoch": 0.4360018617640214, "grad_norm": 2.369858741760254, "learning_rate": 0.0005, "loss": 3.6406397819519043, "step": 3747 }, { "epoch": 0.43611822201535955, "grad_norm": 1.523825764656067, "learning_rate": 0.0005, "loss": 2.5834710597991943, "step": 3748 }, { "epoch": 0.4362345822666977, "grad_norm": 1.7151767015457153, "learning_rate": 0.0005, "loss": 3.1058285236358643, "step": 3749 }, { "epoch": 0.43635094251803586, "grad_norm": 1.5671602487564087, "learning_rate": 0.0005, "loss": 2.876619577407837, "step": 3750 }, { "epoch": 0.43646730276937395, "grad_norm": 1.2375978231430054, "learning_rate": 0.0005, "loss": 2.071162700653076, "step": 3751 }, { "epoch": 0.4365836630207121, "grad_norm": 1.5285863876342773, "learning_rate": 0.0005, "loss": 2.395183801651001, "step": 3752 }, { "epoch": 0.43670002327205026, "grad_norm": 1.7376606464385986, "learning_rate": 0.0005, "loss": 2.336923122406006, "step": 3753 }, { "epoch": 0.4368163835233884, "grad_norm": 1.4028922319412231, "learning_rate": 0.0005, "loss": 3.109294891357422, "step": 3754 }, { "epoch": 0.4369327437747266, "grad_norm": 1.2089054584503174, "learning_rate": 0.0005, "loss": 2.8014822006225586, "step": 3755 }, { "epoch": 0.43704910402606467, "grad_norm": 1.0257374048233032, "learning_rate": 0.0005, "loss": 1.4293714761734009, "step": 3756 }, { "epoch": 0.4371654642774028, "grad_norm": 1.0161652565002441, "learning_rate": 0.0005, "loss": 2.404771327972412, "step": 3757 }, { "epoch": 0.437281824528741, "grad_norm": 1.538580298423767, "learning_rate": 0.0005, "loss": 2.089806079864502, "step": 3758 }, { "epoch": 0.43739818478007914, "grad_norm": 1.608263373374939, "learning_rate": 0.0005, "loss": 2.9308390617370605, "step": 3759 }, { "epoch": 0.4375145450314173, "grad_norm": 0.9898412227630615, "learning_rate": 0.0005, "loss": 2.052945613861084, "step": 3760 }, { "epoch": 0.4376309052827554, "grad_norm": 3.2244677543640137, "learning_rate": 0.0005, "loss": 3.2447609901428223, "step": 3761 }, { "epoch": 0.43774726553409354, "grad_norm": 2.3715128898620605, "learning_rate": 0.0005, "loss": 3.3618879318237305, "step": 3762 }, { "epoch": 0.4378636257854317, "grad_norm": 5.322961807250977, "learning_rate": 0.0005, "loss": 3.4009056091308594, "step": 3763 }, { "epoch": 0.43797998603676985, "grad_norm": 0.8852713704109192, "learning_rate": 0.0005, "loss": 1.7957402467727661, "step": 3764 }, { "epoch": 0.438096346288108, "grad_norm": 1.5381077527999878, "learning_rate": 0.0005, "loss": 1.7435479164123535, "step": 3765 }, { "epoch": 0.4382127065394461, "grad_norm": 1.0055738687515259, "learning_rate": 0.0005, "loss": 2.379533290863037, "step": 3766 }, { "epoch": 0.43832906679078426, "grad_norm": 0.6095143556594849, "learning_rate": 0.0005, "loss": 1.7157403230667114, "step": 3767 }, { "epoch": 0.4384454270421224, "grad_norm": 2.027879238128662, "learning_rate": 0.0005, "loss": 2.900397777557373, "step": 3768 }, { "epoch": 0.43856178729346057, "grad_norm": 1.8677657842636108, "learning_rate": 0.0005, "loss": 2.7619242668151855, "step": 3769 }, { "epoch": 0.4386781475447987, "grad_norm": 0.9166868329048157, "learning_rate": 0.0005, "loss": 2.5517735481262207, "step": 3770 }, { "epoch": 0.4387945077961368, "grad_norm": 1.9550862312316895, "learning_rate": 0.0005, "loss": 2.43300461769104, "step": 3771 }, { "epoch": 0.438910868047475, "grad_norm": 1.6404510736465454, "learning_rate": 0.0005, "loss": 3.667019844055176, "step": 3772 }, { "epoch": 0.43902722829881313, "grad_norm": 1.7760021686553955, "learning_rate": 0.0005, "loss": 2.2917563915252686, "step": 3773 }, { "epoch": 0.4391435885501513, "grad_norm": 1.3378634452819824, "learning_rate": 0.0005, "loss": 2.526778221130371, "step": 3774 }, { "epoch": 0.4392599488014894, "grad_norm": 1.6741548776626587, "learning_rate": 0.0005, "loss": 2.7608587741851807, "step": 3775 }, { "epoch": 0.43937630905282754, "grad_norm": 1.4950852394104004, "learning_rate": 0.0005, "loss": 2.6527230739593506, "step": 3776 }, { "epoch": 0.4394926693041657, "grad_norm": 2.0475921630859375, "learning_rate": 0.0005, "loss": 2.8747153282165527, "step": 3777 }, { "epoch": 0.43960902955550385, "grad_norm": 1.3786699771881104, "learning_rate": 0.0005, "loss": 2.582080841064453, "step": 3778 }, { "epoch": 0.439725389806842, "grad_norm": 1.6718294620513916, "learning_rate": 0.0005, "loss": 3.225754737854004, "step": 3779 }, { "epoch": 0.4398417500581801, "grad_norm": 1.6164419651031494, "learning_rate": 0.0005, "loss": 2.3472981452941895, "step": 3780 }, { "epoch": 0.43995811030951826, "grad_norm": 1.3810216188430786, "learning_rate": 0.0005, "loss": 3.813232898712158, "step": 3781 }, { "epoch": 0.4400744705608564, "grad_norm": 1.3663448095321655, "learning_rate": 0.0005, "loss": 3.3444530963897705, "step": 3782 }, { "epoch": 0.44019083081219457, "grad_norm": 1.35324227809906, "learning_rate": 0.0005, "loss": 1.9909446239471436, "step": 3783 }, { "epoch": 0.4403071910635327, "grad_norm": 1.743386149406433, "learning_rate": 0.0005, "loss": 2.4179062843322754, "step": 3784 }, { "epoch": 0.4404235513148708, "grad_norm": 2.643895387649536, "learning_rate": 0.0005, "loss": 3.8018698692321777, "step": 3785 }, { "epoch": 0.440539911566209, "grad_norm": 1.695999026298523, "learning_rate": 0.0005, "loss": 2.7326765060424805, "step": 3786 }, { "epoch": 0.44065627181754713, "grad_norm": 1.4406812191009521, "learning_rate": 0.0005, "loss": 2.112440824508667, "step": 3787 }, { "epoch": 0.4407726320688853, "grad_norm": 1.1761054992675781, "learning_rate": 0.0005, "loss": 2.160396099090576, "step": 3788 }, { "epoch": 0.44088899232022344, "grad_norm": 1.3019315004348755, "learning_rate": 0.0005, "loss": 2.701781749725342, "step": 3789 }, { "epoch": 0.44100535257156154, "grad_norm": 1.459629774093628, "learning_rate": 0.0005, "loss": 1.7989581823349, "step": 3790 }, { "epoch": 0.4411217128228997, "grad_norm": 1.9425632953643799, "learning_rate": 0.0005, "loss": 3.0135278701782227, "step": 3791 }, { "epoch": 0.44123807307423785, "grad_norm": 0.5000950694084167, "learning_rate": 0.0005, "loss": 1.1141836643218994, "step": 3792 }, { "epoch": 0.441354433325576, "grad_norm": 1.5874359607696533, "learning_rate": 0.0005, "loss": 2.7878143787384033, "step": 3793 }, { "epoch": 0.4414707935769141, "grad_norm": 2.160815477371216, "learning_rate": 0.0005, "loss": 3.0234007835388184, "step": 3794 }, { "epoch": 0.44158715382825225, "grad_norm": 1.5341631174087524, "learning_rate": 0.0005, "loss": 2.6152729988098145, "step": 3795 }, { "epoch": 0.4417035140795904, "grad_norm": 0.7011716365814209, "learning_rate": 0.0005, "loss": 1.657287836074829, "step": 3796 }, { "epoch": 0.44181987433092856, "grad_norm": 1.2601912021636963, "learning_rate": 0.0005, "loss": 2.0890650749206543, "step": 3797 }, { "epoch": 0.4419362345822667, "grad_norm": 1.7668694257736206, "learning_rate": 0.0005, "loss": 2.2390623092651367, "step": 3798 }, { "epoch": 0.4420525948336048, "grad_norm": 0.4646679162979126, "learning_rate": 0.0005, "loss": 1.6186201572418213, "step": 3799 }, { "epoch": 0.44216895508494297, "grad_norm": 1.8121527433395386, "learning_rate": 0.0005, "loss": 2.4753408432006836, "step": 3800 }, { "epoch": 0.4422853153362811, "grad_norm": 0.7891700863838196, "learning_rate": 0.0005, "loss": 2.084686040878296, "step": 3801 }, { "epoch": 0.4424016755876193, "grad_norm": 3.124638795852661, "learning_rate": 0.0005, "loss": 2.8169655799865723, "step": 3802 }, { "epoch": 0.44251803583895744, "grad_norm": 1.1901624202728271, "learning_rate": 0.0005, "loss": 1.7499321699142456, "step": 3803 }, { "epoch": 0.44263439609029553, "grad_norm": 1.2739356756210327, "learning_rate": 0.0005, "loss": 2.5791826248168945, "step": 3804 }, { "epoch": 0.4427507563416337, "grad_norm": 1.392619013786316, "learning_rate": 0.0005, "loss": 2.6982779502868652, "step": 3805 }, { "epoch": 0.44286711659297184, "grad_norm": 1.282881259918213, "learning_rate": 0.0005, "loss": 2.8662521839141846, "step": 3806 }, { "epoch": 0.44298347684431, "grad_norm": 1.5924996137619019, "learning_rate": 0.0005, "loss": 2.466454029083252, "step": 3807 }, { "epoch": 0.44309983709564815, "grad_norm": 2.2630131244659424, "learning_rate": 0.0005, "loss": 3.31905460357666, "step": 3808 }, { "epoch": 0.44321619734698625, "grad_norm": 3.473078489303589, "learning_rate": 0.0005, "loss": 3.008331537246704, "step": 3809 }, { "epoch": 0.4433325575983244, "grad_norm": 1.5166399478912354, "learning_rate": 0.0005, "loss": 3.051215171813965, "step": 3810 }, { "epoch": 0.44344891784966256, "grad_norm": 0.6209268569946289, "learning_rate": 0.0005, "loss": 1.176092267036438, "step": 3811 }, { "epoch": 0.4435652781010007, "grad_norm": 0.9516848921775818, "learning_rate": 0.0005, "loss": 2.070474863052368, "step": 3812 }, { "epoch": 0.4436816383523388, "grad_norm": 2.334339141845703, "learning_rate": 0.0005, "loss": 2.444795608520508, "step": 3813 }, { "epoch": 0.44379799860367697, "grad_norm": 1.3456065654754639, "learning_rate": 0.0005, "loss": 2.3626656532287598, "step": 3814 }, { "epoch": 0.4439143588550151, "grad_norm": 1.698976993560791, "learning_rate": 0.0005, "loss": 2.076326608657837, "step": 3815 }, { "epoch": 0.4440307191063533, "grad_norm": 1.7257075309753418, "learning_rate": 0.0005, "loss": 2.791896104812622, "step": 3816 }, { "epoch": 0.44414707935769143, "grad_norm": 1.4517180919647217, "learning_rate": 0.0005, "loss": 2.5073156356811523, "step": 3817 }, { "epoch": 0.44426343960902953, "grad_norm": 1.5914207696914673, "learning_rate": 0.0005, "loss": 2.5034916400909424, "step": 3818 }, { "epoch": 0.4443797998603677, "grad_norm": 0.7599960565567017, "learning_rate": 0.0005, "loss": 1.2384662628173828, "step": 3819 }, { "epoch": 0.44449616011170584, "grad_norm": 1.1748127937316895, "learning_rate": 0.0005, "loss": 2.4136390686035156, "step": 3820 }, { "epoch": 0.444612520363044, "grad_norm": 2.6533215045928955, "learning_rate": 0.0005, "loss": 3.2994132041931152, "step": 3821 }, { "epoch": 0.44472888061438215, "grad_norm": 1.1073694229125977, "learning_rate": 0.0005, "loss": 1.908607006072998, "step": 3822 }, { "epoch": 0.44484524086572025, "grad_norm": 1.3513935804367065, "learning_rate": 0.0005, "loss": 2.503049850463867, "step": 3823 }, { "epoch": 0.4449616011170584, "grad_norm": 1.8774641752243042, "learning_rate": 0.0005, "loss": 2.8038291931152344, "step": 3824 }, { "epoch": 0.44507796136839656, "grad_norm": 1.0012606382369995, "learning_rate": 0.0005, "loss": 2.035458564758301, "step": 3825 }, { "epoch": 0.4451943216197347, "grad_norm": 1.6287617683410645, "learning_rate": 0.0005, "loss": 3.0121042728424072, "step": 3826 }, { "epoch": 0.44531068187107287, "grad_norm": 2.4387760162353516, "learning_rate": 0.0005, "loss": 2.535383701324463, "step": 3827 }, { "epoch": 0.44542704212241097, "grad_norm": 1.3675419092178345, "learning_rate": 0.0005, "loss": 2.7130632400512695, "step": 3828 }, { "epoch": 0.4455434023737491, "grad_norm": 1.5810918807983398, "learning_rate": 0.0005, "loss": 2.6638503074645996, "step": 3829 }, { "epoch": 0.4456597626250873, "grad_norm": 1.4959328174591064, "learning_rate": 0.0005, "loss": 3.251227378845215, "step": 3830 }, { "epoch": 0.44577612287642543, "grad_norm": 1.4380756616592407, "learning_rate": 0.0005, "loss": 2.579336643218994, "step": 3831 }, { "epoch": 0.4458924831277636, "grad_norm": 1.5108494758605957, "learning_rate": 0.0005, "loss": 2.6755831241607666, "step": 3832 }, { "epoch": 0.4460088433791017, "grad_norm": 0.8516874313354492, "learning_rate": 0.0005, "loss": 2.398226737976074, "step": 3833 }, { "epoch": 0.44612520363043984, "grad_norm": 1.1936826705932617, "learning_rate": 0.0005, "loss": 1.9928336143493652, "step": 3834 }, { "epoch": 0.446241563881778, "grad_norm": 1.2207374572753906, "learning_rate": 0.0005, "loss": 2.950514793395996, "step": 3835 }, { "epoch": 0.44635792413311615, "grad_norm": 1.606623888015747, "learning_rate": 0.0005, "loss": 2.47652530670166, "step": 3836 }, { "epoch": 0.44647428438445425, "grad_norm": 1.8277336359024048, "learning_rate": 0.0005, "loss": 2.4901537895202637, "step": 3837 }, { "epoch": 0.4465906446357924, "grad_norm": 1.1695401668548584, "learning_rate": 0.0005, "loss": 2.683304786682129, "step": 3838 }, { "epoch": 0.44670700488713055, "grad_norm": 1.474777340888977, "learning_rate": 0.0005, "loss": 2.0759694576263428, "step": 3839 }, { "epoch": 0.4468233651384687, "grad_norm": 1.6701345443725586, "learning_rate": 0.0005, "loss": 2.1121697425842285, "step": 3840 }, { "epoch": 0.44693972538980686, "grad_norm": 2.2469372749328613, "learning_rate": 0.0005, "loss": 3.045462131500244, "step": 3841 }, { "epoch": 0.44705608564114496, "grad_norm": 1.3195899724960327, "learning_rate": 0.0005, "loss": 3.0569097995758057, "step": 3842 }, { "epoch": 0.4471724458924831, "grad_norm": 1.7840185165405273, "learning_rate": 0.0005, "loss": 1.7244970798492432, "step": 3843 }, { "epoch": 0.44728880614382127, "grad_norm": 1.334754228591919, "learning_rate": 0.0005, "loss": 3.0618507862091064, "step": 3844 }, { "epoch": 0.4474051663951594, "grad_norm": 1.8251495361328125, "learning_rate": 0.0005, "loss": 1.9142812490463257, "step": 3845 }, { "epoch": 0.4475215266464976, "grad_norm": 1.2908732891082764, "learning_rate": 0.0005, "loss": 3.3070967197418213, "step": 3846 }, { "epoch": 0.4476378868978357, "grad_norm": 0.8707017302513123, "learning_rate": 0.0005, "loss": 2.3521313667297363, "step": 3847 }, { "epoch": 0.44775424714917383, "grad_norm": 1.5911743640899658, "learning_rate": 0.0005, "loss": 2.1927428245544434, "step": 3848 }, { "epoch": 0.447870607400512, "grad_norm": 1.3994948863983154, "learning_rate": 0.0005, "loss": 2.1069931983947754, "step": 3849 }, { "epoch": 0.44798696765185014, "grad_norm": 1.3289681673049927, "learning_rate": 0.0005, "loss": 2.2443714141845703, "step": 3850 }, { "epoch": 0.4481033279031883, "grad_norm": 1.1576192378997803, "learning_rate": 0.0005, "loss": 2.891091823577881, "step": 3851 }, { "epoch": 0.4482196881545264, "grad_norm": 1.4341884851455688, "learning_rate": 0.0005, "loss": 1.9368788003921509, "step": 3852 }, { "epoch": 0.44833604840586455, "grad_norm": 1.829474925994873, "learning_rate": 0.0005, "loss": 2.872988700866699, "step": 3853 }, { "epoch": 0.4484524086572027, "grad_norm": 2.1711180210113525, "learning_rate": 0.0005, "loss": 2.4326183795928955, "step": 3854 }, { "epoch": 0.44856876890854086, "grad_norm": 1.8769035339355469, "learning_rate": 0.0005, "loss": 2.7869720458984375, "step": 3855 }, { "epoch": 0.44868512915987896, "grad_norm": 1.0257426500320435, "learning_rate": 0.0005, "loss": 2.7340989112854004, "step": 3856 }, { "epoch": 0.4488014894112171, "grad_norm": 0.5559507012367249, "learning_rate": 0.0005, "loss": 1.6317726373672485, "step": 3857 }, { "epoch": 0.44891784966255527, "grad_norm": 1.3072463274002075, "learning_rate": 0.0005, "loss": 2.778852701187134, "step": 3858 }, { "epoch": 0.4490342099138934, "grad_norm": 1.3886841535568237, "learning_rate": 0.0005, "loss": 1.5941667556762695, "step": 3859 }, { "epoch": 0.4491505701652316, "grad_norm": 1.6541683673858643, "learning_rate": 0.0005, "loss": 1.8876805305480957, "step": 3860 }, { "epoch": 0.4492669304165697, "grad_norm": 1.5118316411972046, "learning_rate": 0.0005, "loss": 3.7175886631011963, "step": 3861 }, { "epoch": 0.44938329066790783, "grad_norm": 1.7197390794754028, "learning_rate": 0.0005, "loss": 2.1305441856384277, "step": 3862 }, { "epoch": 0.449499650919246, "grad_norm": 1.4251545667648315, "learning_rate": 0.0005, "loss": 2.2338240146636963, "step": 3863 }, { "epoch": 0.44961601117058414, "grad_norm": 1.2340164184570312, "learning_rate": 0.0005, "loss": 1.3594391345977783, "step": 3864 }, { "epoch": 0.4497323714219223, "grad_norm": 1.1522345542907715, "learning_rate": 0.0005, "loss": 2.0190794467926025, "step": 3865 }, { "epoch": 0.4498487316732604, "grad_norm": 1.1936609745025635, "learning_rate": 0.0005, "loss": 2.1832356452941895, "step": 3866 }, { "epoch": 0.44996509192459855, "grad_norm": 1.269629716873169, "learning_rate": 0.0005, "loss": 2.232147693634033, "step": 3867 }, { "epoch": 0.4500814521759367, "grad_norm": 0.9248690605163574, "learning_rate": 0.0005, "loss": 1.8185771703720093, "step": 3868 }, { "epoch": 0.45019781242727486, "grad_norm": 1.2021273374557495, "learning_rate": 0.0005, "loss": 2.743216037750244, "step": 3869 }, { "epoch": 0.450314172678613, "grad_norm": 1.8474992513656616, "learning_rate": 0.0005, "loss": 3.3545584678649902, "step": 3870 }, { "epoch": 0.4504305329299511, "grad_norm": 1.2867085933685303, "learning_rate": 0.0005, "loss": 2.888381004333496, "step": 3871 }, { "epoch": 0.45054689318128927, "grad_norm": 1.9193731546401978, "learning_rate": 0.0005, "loss": 3.4203412532806396, "step": 3872 }, { "epoch": 0.4506632534326274, "grad_norm": 1.5484026670455933, "learning_rate": 0.0005, "loss": 3.093942165374756, "step": 3873 }, { "epoch": 0.4507796136839656, "grad_norm": 1.578008770942688, "learning_rate": 0.0005, "loss": 2.4189553260803223, "step": 3874 }, { "epoch": 0.4508959739353037, "grad_norm": 2.451927900314331, "learning_rate": 0.0005, "loss": 3.2657530307769775, "step": 3875 }, { "epoch": 0.45101233418664183, "grad_norm": 1.3392601013183594, "learning_rate": 0.0005, "loss": 2.7719240188598633, "step": 3876 }, { "epoch": 0.45112869443798, "grad_norm": 0.8405636548995972, "learning_rate": 0.0005, "loss": 2.206634998321533, "step": 3877 }, { "epoch": 0.45124505468931814, "grad_norm": 2.465076208114624, "learning_rate": 0.0005, "loss": 2.8937103748321533, "step": 3878 }, { "epoch": 0.4513614149406563, "grad_norm": 1.2117451429367065, "learning_rate": 0.0005, "loss": 2.2278122901916504, "step": 3879 }, { "epoch": 0.4514777751919944, "grad_norm": 1.101820707321167, "learning_rate": 0.0005, "loss": 2.505089282989502, "step": 3880 }, { "epoch": 0.45159413544333254, "grad_norm": 1.430679202079773, "learning_rate": 0.0005, "loss": 2.698082208633423, "step": 3881 }, { "epoch": 0.4517104956946707, "grad_norm": 1.5195356607437134, "learning_rate": 0.0005, "loss": 2.9590671062469482, "step": 3882 }, { "epoch": 0.45182685594600885, "grad_norm": 0.8636601567268372, "learning_rate": 0.0005, "loss": 1.51756751537323, "step": 3883 }, { "epoch": 0.451943216197347, "grad_norm": 0.7989457845687866, "learning_rate": 0.0005, "loss": 2.1074378490448, "step": 3884 }, { "epoch": 0.4520595764486851, "grad_norm": 1.2098910808563232, "learning_rate": 0.0005, "loss": 2.14910888671875, "step": 3885 }, { "epoch": 0.45217593670002326, "grad_norm": 1.2435628175735474, "learning_rate": 0.0005, "loss": 2.826406478881836, "step": 3886 }, { "epoch": 0.4522922969513614, "grad_norm": 1.576339602470398, "learning_rate": 0.0005, "loss": 3.015951156616211, "step": 3887 }, { "epoch": 0.45240865720269957, "grad_norm": 0.9153954386711121, "learning_rate": 0.0005, "loss": 2.6731343269348145, "step": 3888 }, { "epoch": 0.4525250174540377, "grad_norm": 1.4030303955078125, "learning_rate": 0.0005, "loss": 2.3960232734680176, "step": 3889 }, { "epoch": 0.4526413777053758, "grad_norm": 1.251288652420044, "learning_rate": 0.0005, "loss": 2.423197031021118, "step": 3890 }, { "epoch": 0.452757737956714, "grad_norm": 1.235622763633728, "learning_rate": 0.0005, "loss": 2.741419553756714, "step": 3891 }, { "epoch": 0.45287409820805213, "grad_norm": 0.7862246036529541, "learning_rate": 0.0005, "loss": 1.7586584091186523, "step": 3892 }, { "epoch": 0.4529904584593903, "grad_norm": 1.4740369319915771, "learning_rate": 0.0005, "loss": 2.4986000061035156, "step": 3893 }, { "epoch": 0.45310681871072844, "grad_norm": 1.9904474020004272, "learning_rate": 0.0005, "loss": 3.3016324043273926, "step": 3894 }, { "epoch": 0.45322317896206654, "grad_norm": 1.2479850053787231, "learning_rate": 0.0005, "loss": 3.644240617752075, "step": 3895 }, { "epoch": 0.4533395392134047, "grad_norm": 1.6395941972732544, "learning_rate": 0.0005, "loss": 1.9345331192016602, "step": 3896 }, { "epoch": 0.45345589946474285, "grad_norm": 1.310549259185791, "learning_rate": 0.0005, "loss": 2.5953662395477295, "step": 3897 }, { "epoch": 0.453572259716081, "grad_norm": 1.2828127145767212, "learning_rate": 0.0005, "loss": 2.5026931762695312, "step": 3898 }, { "epoch": 0.4536886199674191, "grad_norm": 1.654996395111084, "learning_rate": 0.0005, "loss": 2.160017967224121, "step": 3899 }, { "epoch": 0.45380498021875726, "grad_norm": 1.5740511417388916, "learning_rate": 0.0005, "loss": 1.8403854370117188, "step": 3900 }, { "epoch": 0.4539213404700954, "grad_norm": 1.6868181228637695, "learning_rate": 0.0005, "loss": 3.4835000038146973, "step": 3901 }, { "epoch": 0.45403770072143357, "grad_norm": 2.063915967941284, "learning_rate": 0.0005, "loss": 2.4150915145874023, "step": 3902 }, { "epoch": 0.4541540609727717, "grad_norm": 0.9169086813926697, "learning_rate": 0.0005, "loss": 1.4104418754577637, "step": 3903 }, { "epoch": 0.4542704212241098, "grad_norm": 1.857008695602417, "learning_rate": 0.0005, "loss": 1.984849214553833, "step": 3904 }, { "epoch": 0.454386781475448, "grad_norm": 1.5319795608520508, "learning_rate": 0.0005, "loss": 1.9844564199447632, "step": 3905 }, { "epoch": 0.45450314172678613, "grad_norm": 1.1319804191589355, "learning_rate": 0.0005, "loss": 2.3635761737823486, "step": 3906 }, { "epoch": 0.4546195019781243, "grad_norm": 2.6112256050109863, "learning_rate": 0.0005, "loss": 2.7837390899658203, "step": 3907 }, { "epoch": 0.45473586222946244, "grad_norm": 2.2536256313323975, "learning_rate": 0.0005, "loss": 2.3824963569641113, "step": 3908 }, { "epoch": 0.45485222248080054, "grad_norm": 2.4871604442596436, "learning_rate": 0.0005, "loss": 2.959545135498047, "step": 3909 }, { "epoch": 0.4549685827321387, "grad_norm": 1.47477126121521, "learning_rate": 0.0005, "loss": 2.959825038909912, "step": 3910 }, { "epoch": 0.45508494298347685, "grad_norm": 1.4615318775177002, "learning_rate": 0.0005, "loss": 2.7939419746398926, "step": 3911 }, { "epoch": 0.455201303234815, "grad_norm": 2.1743712425231934, "learning_rate": 0.0005, "loss": 3.1192069053649902, "step": 3912 }, { "epoch": 0.45531766348615316, "grad_norm": 35.9581413269043, "learning_rate": 0.0005, "loss": 2.6254634857177734, "step": 3913 }, { "epoch": 0.45543402373749126, "grad_norm": 1.4408986568450928, "learning_rate": 0.0005, "loss": 2.1598899364471436, "step": 3914 }, { "epoch": 0.4555503839888294, "grad_norm": 1.692901849746704, "learning_rate": 0.0005, "loss": 3.4105961322784424, "step": 3915 }, { "epoch": 0.45566674424016756, "grad_norm": 1.2319523096084595, "learning_rate": 0.0005, "loss": 3.151297092437744, "step": 3916 }, { "epoch": 0.4557831044915057, "grad_norm": 1.756495475769043, "learning_rate": 0.0005, "loss": 2.936511516571045, "step": 3917 }, { "epoch": 0.4558994647428438, "grad_norm": 1.5119680166244507, "learning_rate": 0.0005, "loss": 2.5678343772888184, "step": 3918 }, { "epoch": 0.456015824994182, "grad_norm": 1.6710854768753052, "learning_rate": 0.0005, "loss": 2.508145809173584, "step": 3919 }, { "epoch": 0.4561321852455201, "grad_norm": 5.277055263519287, "learning_rate": 0.0005, "loss": 2.4335663318634033, "step": 3920 }, { "epoch": 0.4562485454968583, "grad_norm": 1.5996265411376953, "learning_rate": 0.0005, "loss": 2.1710376739501953, "step": 3921 }, { "epoch": 0.45636490574819644, "grad_norm": 1.0147737264633179, "learning_rate": 0.0005, "loss": 1.9691987037658691, "step": 3922 }, { "epoch": 0.45648126599953454, "grad_norm": 1.7188160419464111, "learning_rate": 0.0005, "loss": 2.196521282196045, "step": 3923 }, { "epoch": 0.4565976262508727, "grad_norm": 1.0221537351608276, "learning_rate": 0.0005, "loss": 2.079132556915283, "step": 3924 }, { "epoch": 0.45671398650221084, "grad_norm": 0.8560405373573303, "learning_rate": 0.0005, "loss": 1.9780857563018799, "step": 3925 }, { "epoch": 0.456830346753549, "grad_norm": 1.6735775470733643, "learning_rate": 0.0005, "loss": 3.23325777053833, "step": 3926 }, { "epoch": 0.45694670700488715, "grad_norm": 1.5003691911697388, "learning_rate": 0.0005, "loss": 2.5186309814453125, "step": 3927 }, { "epoch": 0.45706306725622525, "grad_norm": 1.8929688930511475, "learning_rate": 0.0005, "loss": 2.5110867023468018, "step": 3928 }, { "epoch": 0.4571794275075634, "grad_norm": 1.3396176099777222, "learning_rate": 0.0005, "loss": 1.7574928998947144, "step": 3929 }, { "epoch": 0.45729578775890156, "grad_norm": 1.2199913263320923, "learning_rate": 0.0005, "loss": 2.229992389678955, "step": 3930 }, { "epoch": 0.4574121480102397, "grad_norm": 0.9087095856666565, "learning_rate": 0.0005, "loss": 2.6716971397399902, "step": 3931 }, { "epoch": 0.45752850826157787, "grad_norm": 1.6573411226272583, "learning_rate": 0.0005, "loss": 2.907465934753418, "step": 3932 }, { "epoch": 0.45764486851291597, "grad_norm": 1.4989802837371826, "learning_rate": 0.0005, "loss": 2.8462464809417725, "step": 3933 }, { "epoch": 0.4577612287642541, "grad_norm": 1.4698665142059326, "learning_rate": 0.0005, "loss": 1.923499584197998, "step": 3934 }, { "epoch": 0.4578775890155923, "grad_norm": 1.5603020191192627, "learning_rate": 0.0005, "loss": 2.7389941215515137, "step": 3935 }, { "epoch": 0.45799394926693043, "grad_norm": 1.8147433996200562, "learning_rate": 0.0005, "loss": 3.038226842880249, "step": 3936 }, { "epoch": 0.45811030951826853, "grad_norm": 1.3640012741088867, "learning_rate": 0.0005, "loss": 2.6388564109802246, "step": 3937 }, { "epoch": 0.4582266697696067, "grad_norm": 6.325087070465088, "learning_rate": 0.0005, "loss": 3.6794838905334473, "step": 3938 }, { "epoch": 0.45834303002094484, "grad_norm": 1.0279159545898438, "learning_rate": 0.0005, "loss": 2.6988685131073, "step": 3939 }, { "epoch": 0.458459390272283, "grad_norm": 1.3479756116867065, "learning_rate": 0.0005, "loss": 2.763646125793457, "step": 3940 }, { "epoch": 0.45857575052362115, "grad_norm": 1.0145291090011597, "learning_rate": 0.0005, "loss": 1.6065394878387451, "step": 3941 }, { "epoch": 0.45869211077495925, "grad_norm": 1.4403934478759766, "learning_rate": 0.0005, "loss": 2.20001482963562, "step": 3942 }, { "epoch": 0.4588084710262974, "grad_norm": 2.1104910373687744, "learning_rate": 0.0005, "loss": 2.9735512733459473, "step": 3943 }, { "epoch": 0.45892483127763556, "grad_norm": 1.2970165014266968, "learning_rate": 0.0005, "loss": 2.3288838863372803, "step": 3944 }, { "epoch": 0.4590411915289737, "grad_norm": 1.7753108739852905, "learning_rate": 0.0005, "loss": 4.049365043640137, "step": 3945 }, { "epoch": 0.45915755178031187, "grad_norm": 1.4476279020309448, "learning_rate": 0.0005, "loss": 2.4435672760009766, "step": 3946 }, { "epoch": 0.45927391203164997, "grad_norm": 1.3320841789245605, "learning_rate": 0.0005, "loss": 2.3569037914276123, "step": 3947 }, { "epoch": 0.4593902722829881, "grad_norm": 1.511350393295288, "learning_rate": 0.0005, "loss": 2.421490430831909, "step": 3948 }, { "epoch": 0.4595066325343263, "grad_norm": 0.9724201560020447, "learning_rate": 0.0005, "loss": 2.1911182403564453, "step": 3949 }, { "epoch": 0.45962299278566443, "grad_norm": 2.0965445041656494, "learning_rate": 0.0005, "loss": 2.6088855266571045, "step": 3950 }, { "epoch": 0.4597393530370026, "grad_norm": 1.317309021949768, "learning_rate": 0.0005, "loss": 2.392970085144043, "step": 3951 }, { "epoch": 0.4598557132883407, "grad_norm": 1.9167712926864624, "learning_rate": 0.0005, "loss": 1.6929118633270264, "step": 3952 }, { "epoch": 0.45997207353967884, "grad_norm": 2.30942440032959, "learning_rate": 0.0005, "loss": 2.9546737670898438, "step": 3953 }, { "epoch": 0.460088433791017, "grad_norm": 2.1162147521972656, "learning_rate": 0.0005, "loss": 2.128385543823242, "step": 3954 }, { "epoch": 0.46020479404235515, "grad_norm": 1.4353752136230469, "learning_rate": 0.0005, "loss": 2.623103618621826, "step": 3955 }, { "epoch": 0.4603211542936933, "grad_norm": 2.211336612701416, "learning_rate": 0.0005, "loss": 3.479616165161133, "step": 3956 }, { "epoch": 0.4604375145450314, "grad_norm": 0.7261670827865601, "learning_rate": 0.0005, "loss": 1.421196460723877, "step": 3957 }, { "epoch": 0.46055387479636956, "grad_norm": 1.4391309022903442, "learning_rate": 0.0005, "loss": 2.3284175395965576, "step": 3958 }, { "epoch": 0.4606702350477077, "grad_norm": 1.4234877824783325, "learning_rate": 0.0005, "loss": 1.9932109117507935, "step": 3959 }, { "epoch": 0.46078659529904586, "grad_norm": 1.7848683595657349, "learning_rate": 0.0005, "loss": 2.794682741165161, "step": 3960 }, { "epoch": 0.46090295555038396, "grad_norm": 1.7245599031448364, "learning_rate": 0.0005, "loss": 2.466590404510498, "step": 3961 }, { "epoch": 0.4610193158017221, "grad_norm": 1.3050463199615479, "learning_rate": 0.0005, "loss": 3.512357711791992, "step": 3962 }, { "epoch": 0.4611356760530603, "grad_norm": 1.6029680967330933, "learning_rate": 0.0005, "loss": 2.871967315673828, "step": 3963 }, { "epoch": 0.4612520363043984, "grad_norm": 1.3509345054626465, "learning_rate": 0.0005, "loss": 2.6306262016296387, "step": 3964 }, { "epoch": 0.4613683965557366, "grad_norm": 1.4930750131607056, "learning_rate": 0.0005, "loss": 3.0362205505371094, "step": 3965 }, { "epoch": 0.4614847568070747, "grad_norm": 1.4562134742736816, "learning_rate": 0.0005, "loss": 2.24025821685791, "step": 3966 }, { "epoch": 0.46160111705841284, "grad_norm": 1.4487005472183228, "learning_rate": 0.0005, "loss": 3.584383964538574, "step": 3967 }, { "epoch": 0.461717477309751, "grad_norm": 1.695540428161621, "learning_rate": 0.0005, "loss": 3.2994885444641113, "step": 3968 }, { "epoch": 0.46183383756108914, "grad_norm": 1.146511197090149, "learning_rate": 0.0005, "loss": 2.6087863445281982, "step": 3969 }, { "epoch": 0.4619501978124273, "grad_norm": 0.7350478768348694, "learning_rate": 0.0005, "loss": 1.501478910446167, "step": 3970 }, { "epoch": 0.4620665580637654, "grad_norm": 1.7914540767669678, "learning_rate": 0.0005, "loss": 3.200026035308838, "step": 3971 }, { "epoch": 0.46218291831510355, "grad_norm": 1.2641594409942627, "learning_rate": 0.0005, "loss": 2.171241044998169, "step": 3972 }, { "epoch": 0.4622992785664417, "grad_norm": 1.0315186977386475, "learning_rate": 0.0005, "loss": 2.1142170429229736, "step": 3973 }, { "epoch": 0.46241563881777986, "grad_norm": 0.8066403865814209, "learning_rate": 0.0005, "loss": 1.6576570272445679, "step": 3974 }, { "epoch": 0.462531999069118, "grad_norm": 1.7998607158660889, "learning_rate": 0.0005, "loss": 2.6011219024658203, "step": 3975 }, { "epoch": 0.4626483593204561, "grad_norm": 1.273092269897461, "learning_rate": 0.0005, "loss": 2.8946597576141357, "step": 3976 }, { "epoch": 0.46276471957179427, "grad_norm": 1.4942173957824707, "learning_rate": 0.0005, "loss": 2.2698302268981934, "step": 3977 }, { "epoch": 0.4628810798231324, "grad_norm": 0.9947460889816284, "learning_rate": 0.0005, "loss": 2.2462127208709717, "step": 3978 }, { "epoch": 0.4629974400744706, "grad_norm": 1.7785618305206299, "learning_rate": 0.0005, "loss": 2.150933265686035, "step": 3979 }, { "epoch": 0.4631138003258087, "grad_norm": 1.0740437507629395, "learning_rate": 0.0005, "loss": 1.9293735027313232, "step": 3980 }, { "epoch": 0.46323016057714683, "grad_norm": 1.8517075777053833, "learning_rate": 0.0005, "loss": 3.283118724822998, "step": 3981 }, { "epoch": 0.463346520828485, "grad_norm": 0.929166316986084, "learning_rate": 0.0005, "loss": 2.7491097450256348, "step": 3982 }, { "epoch": 0.46346288107982314, "grad_norm": 1.6538422107696533, "learning_rate": 0.0005, "loss": 2.411181926727295, "step": 3983 }, { "epoch": 0.4635792413311613, "grad_norm": 0.5351723432540894, "learning_rate": 0.0005, "loss": 1.5036590099334717, "step": 3984 }, { "epoch": 0.4636956015824994, "grad_norm": 1.3444228172302246, "learning_rate": 0.0005, "loss": 2.2757794857025146, "step": 3985 }, { "epoch": 0.46381196183383755, "grad_norm": 1.3253432512283325, "learning_rate": 0.0005, "loss": 3.0397095680236816, "step": 3986 }, { "epoch": 0.4639283220851757, "grad_norm": 2.6588408946990967, "learning_rate": 0.0005, "loss": 1.489176869392395, "step": 3987 }, { "epoch": 0.46404468233651386, "grad_norm": 1.4168545007705688, "learning_rate": 0.0005, "loss": 2.1785497665405273, "step": 3988 }, { "epoch": 0.464161042587852, "grad_norm": 1.759217381477356, "learning_rate": 0.0005, "loss": 2.6198058128356934, "step": 3989 }, { "epoch": 0.4642774028391901, "grad_norm": 1.265694260597229, "learning_rate": 0.0005, "loss": 2.771308422088623, "step": 3990 }, { "epoch": 0.46439376309052827, "grad_norm": 1.5493489503860474, "learning_rate": 0.0005, "loss": 2.4073970317840576, "step": 3991 }, { "epoch": 0.4645101233418664, "grad_norm": 1.4466813802719116, "learning_rate": 0.0005, "loss": 3.0983147621154785, "step": 3992 }, { "epoch": 0.4646264835932046, "grad_norm": 0.7715907692909241, "learning_rate": 0.0005, "loss": 1.3082823753356934, "step": 3993 }, { "epoch": 0.46474284384454273, "grad_norm": 1.6616424322128296, "learning_rate": 0.0005, "loss": 3.3255841732025146, "step": 3994 }, { "epoch": 0.46485920409588083, "grad_norm": 1.3368093967437744, "learning_rate": 0.0005, "loss": 1.7089667320251465, "step": 3995 }, { "epoch": 0.464975564347219, "grad_norm": 1.1063296794891357, "learning_rate": 0.0005, "loss": 2.256890296936035, "step": 3996 }, { "epoch": 0.46509192459855714, "grad_norm": 0.9363062977790833, "learning_rate": 0.0005, "loss": 2.42172908782959, "step": 3997 }, { "epoch": 0.4652082848498953, "grad_norm": 1.2969213724136353, "learning_rate": 0.0005, "loss": 2.5303149223327637, "step": 3998 }, { "epoch": 0.4653246451012334, "grad_norm": 1.583664059638977, "learning_rate": 0.0005, "loss": 2.62338924407959, "step": 3999 }, { "epoch": 0.46544100535257155, "grad_norm": 1.424845576286316, "learning_rate": 0.0005, "loss": 3.3770370483398438, "step": 4000 }, { "epoch": 0.4655573656039097, "grad_norm": 1.8867411613464355, "learning_rate": 0.0005, "loss": 3.3422529697418213, "step": 4001 }, { "epoch": 0.46567372585524786, "grad_norm": 1.342769980430603, "learning_rate": 0.0005, "loss": 2.912065267562866, "step": 4002 }, { "epoch": 0.465790086106586, "grad_norm": 1.4388351440429688, "learning_rate": 0.0005, "loss": 2.758969306945801, "step": 4003 }, { "epoch": 0.4659064463579241, "grad_norm": 0.7484511137008667, "learning_rate": 0.0005, "loss": 1.6208796501159668, "step": 4004 }, { "epoch": 0.46602280660926226, "grad_norm": 1.7892600297927856, "learning_rate": 0.0005, "loss": 2.52400279045105, "step": 4005 }, { "epoch": 0.4661391668606004, "grad_norm": 1.1935358047485352, "learning_rate": 0.0005, "loss": 2.039818286895752, "step": 4006 }, { "epoch": 0.4662555271119386, "grad_norm": 1.1233073472976685, "learning_rate": 0.0005, "loss": 2.0606589317321777, "step": 4007 }, { "epoch": 0.4663718873632767, "grad_norm": 1.577512264251709, "learning_rate": 0.0005, "loss": 2.1131820678710938, "step": 4008 }, { "epoch": 0.4664882476146148, "grad_norm": 1.1708693504333496, "learning_rate": 0.0005, "loss": 2.0958809852600098, "step": 4009 }, { "epoch": 0.466604607865953, "grad_norm": 1.7563081979751587, "learning_rate": 0.0005, "loss": 3.1057825088500977, "step": 4010 }, { "epoch": 0.46672096811729114, "grad_norm": 1.5398458242416382, "learning_rate": 0.0005, "loss": 3.735867977142334, "step": 4011 }, { "epoch": 0.4668373283686293, "grad_norm": 1.871461033821106, "learning_rate": 0.0005, "loss": 2.6183574199676514, "step": 4012 }, { "epoch": 0.46695368861996744, "grad_norm": 1.1521246433258057, "learning_rate": 0.0005, "loss": 3.0808558464050293, "step": 4013 }, { "epoch": 0.46707004887130554, "grad_norm": 1.6277447938919067, "learning_rate": 0.0005, "loss": 2.991525173187256, "step": 4014 }, { "epoch": 0.4671864091226437, "grad_norm": 1.1717617511749268, "learning_rate": 0.0005, "loss": 2.7980990409851074, "step": 4015 }, { "epoch": 0.46730276937398185, "grad_norm": 0.9147955179214478, "learning_rate": 0.0005, "loss": 2.02030611038208, "step": 4016 }, { "epoch": 0.46741912962532, "grad_norm": 1.5176557302474976, "learning_rate": 0.0005, "loss": 3.0989537239074707, "step": 4017 }, { "epoch": 0.46753548987665816, "grad_norm": 2.271960735321045, "learning_rate": 0.0005, "loss": 2.2671971321105957, "step": 4018 }, { "epoch": 0.46765185012799626, "grad_norm": 1.0249013900756836, "learning_rate": 0.0005, "loss": 1.971011996269226, "step": 4019 }, { "epoch": 0.4677682103793344, "grad_norm": 3.4268879890441895, "learning_rate": 0.0005, "loss": 3.076098680496216, "step": 4020 }, { "epoch": 0.46788457063067257, "grad_norm": 2.418198347091675, "learning_rate": 0.0005, "loss": 3.04357647895813, "step": 4021 }, { "epoch": 0.4680009308820107, "grad_norm": 1.2964661121368408, "learning_rate": 0.0005, "loss": 2.4886319637298584, "step": 4022 }, { "epoch": 0.4681172911333488, "grad_norm": 0.9168747663497925, "learning_rate": 0.0005, "loss": 1.5786147117614746, "step": 4023 }, { "epoch": 0.468233651384687, "grad_norm": 1.5382720232009888, "learning_rate": 0.0005, "loss": 2.321056842803955, "step": 4024 }, { "epoch": 0.46835001163602513, "grad_norm": 1.050465703010559, "learning_rate": 0.0005, "loss": 2.1942763328552246, "step": 4025 }, { "epoch": 0.4684663718873633, "grad_norm": 1.653356909751892, "learning_rate": 0.0005, "loss": 3.2379493713378906, "step": 4026 }, { "epoch": 0.46858273213870144, "grad_norm": 1.6701655387878418, "learning_rate": 0.0005, "loss": 3.116506576538086, "step": 4027 }, { "epoch": 0.46869909239003954, "grad_norm": 2.527437210083008, "learning_rate": 0.0005, "loss": 3.045609951019287, "step": 4028 }, { "epoch": 0.4688154526413777, "grad_norm": 0.6113051176071167, "learning_rate": 0.0005, "loss": 1.8353407382965088, "step": 4029 }, { "epoch": 0.46893181289271585, "grad_norm": 1.2681769132614136, "learning_rate": 0.0005, "loss": 1.664600133895874, "step": 4030 }, { "epoch": 0.469048173144054, "grad_norm": 1.5006153583526611, "learning_rate": 0.0005, "loss": 2.2328433990478516, "step": 4031 }, { "epoch": 0.46916453339539216, "grad_norm": 1.1369341611862183, "learning_rate": 0.0005, "loss": 2.9613637924194336, "step": 4032 }, { "epoch": 0.46928089364673026, "grad_norm": 1.336114525794983, "learning_rate": 0.0005, "loss": 2.3377315998077393, "step": 4033 }, { "epoch": 0.4693972538980684, "grad_norm": 1.569944977760315, "learning_rate": 0.0005, "loss": 2.8940813541412354, "step": 4034 }, { "epoch": 0.46951361414940657, "grad_norm": 1.6009762287139893, "learning_rate": 0.0005, "loss": 2.4510886669158936, "step": 4035 }, { "epoch": 0.4696299744007447, "grad_norm": 1.434658408164978, "learning_rate": 0.0005, "loss": 1.7245761156082153, "step": 4036 }, { "epoch": 0.4697463346520829, "grad_norm": 1.8331400156021118, "learning_rate": 0.0005, "loss": 2.213352680206299, "step": 4037 }, { "epoch": 0.469862694903421, "grad_norm": 1.244423747062683, "learning_rate": 0.0005, "loss": 2.1125450134277344, "step": 4038 }, { "epoch": 0.46997905515475913, "grad_norm": 1.614142894744873, "learning_rate": 0.0005, "loss": 2.3449881076812744, "step": 4039 }, { "epoch": 0.4700954154060973, "grad_norm": 1.2868056297302246, "learning_rate": 0.0005, "loss": 1.9132676124572754, "step": 4040 }, { "epoch": 0.47021177565743544, "grad_norm": 1.1193654537200928, "learning_rate": 0.0005, "loss": 3.1829442977905273, "step": 4041 }, { "epoch": 0.47032813590877354, "grad_norm": 1.0374406576156616, "learning_rate": 0.0005, "loss": 2.0068087577819824, "step": 4042 }, { "epoch": 0.4704444961601117, "grad_norm": 1.6590590476989746, "learning_rate": 0.0005, "loss": 3.18304705619812, "step": 4043 }, { "epoch": 0.47056085641144985, "grad_norm": 1.2156819105148315, "learning_rate": 0.0005, "loss": 3.297147274017334, "step": 4044 }, { "epoch": 0.470677216662788, "grad_norm": 2.2081458568573, "learning_rate": 0.0005, "loss": 2.6977853775024414, "step": 4045 }, { "epoch": 0.47079357691412616, "grad_norm": 3.881134033203125, "learning_rate": 0.0005, "loss": 3.2071738243103027, "step": 4046 }, { "epoch": 0.47090993716546425, "grad_norm": 1.5063174962997437, "learning_rate": 0.0005, "loss": 2.02816104888916, "step": 4047 }, { "epoch": 0.4710262974168024, "grad_norm": 1.0431678295135498, "learning_rate": 0.0005, "loss": 1.9687576293945312, "step": 4048 }, { "epoch": 0.47114265766814056, "grad_norm": 1.8978022336959839, "learning_rate": 0.0005, "loss": 3.0870542526245117, "step": 4049 }, { "epoch": 0.4712590179194787, "grad_norm": 1.1492213010787964, "learning_rate": 0.0005, "loss": 2.723727226257324, "step": 4050 }, { "epoch": 0.47137537817081687, "grad_norm": 1.1181035041809082, "learning_rate": 0.0005, "loss": 2.547436237335205, "step": 4051 }, { "epoch": 0.47149173842215497, "grad_norm": 1.5658222436904907, "learning_rate": 0.0005, "loss": 3.0025882720947266, "step": 4052 }, { "epoch": 0.4716080986734931, "grad_norm": 1.991661787033081, "learning_rate": 0.0005, "loss": 2.160310745239258, "step": 4053 }, { "epoch": 0.4717244589248313, "grad_norm": 5.8120880126953125, "learning_rate": 0.0005, "loss": 2.8021888732910156, "step": 4054 }, { "epoch": 0.47184081917616943, "grad_norm": 0.6056155562400818, "learning_rate": 0.0005, "loss": 1.1722545623779297, "step": 4055 }, { "epoch": 0.4719571794275076, "grad_norm": 1.7299776077270508, "learning_rate": 0.0005, "loss": 2.679342746734619, "step": 4056 }, { "epoch": 0.4720735396788457, "grad_norm": 3.361541748046875, "learning_rate": 0.0005, "loss": 2.6155619621276855, "step": 4057 }, { "epoch": 0.47218989993018384, "grad_norm": 1.6226961612701416, "learning_rate": 0.0005, "loss": 2.9239368438720703, "step": 4058 }, { "epoch": 0.472306260181522, "grad_norm": 1.9383913278579712, "learning_rate": 0.0005, "loss": 3.0905497074127197, "step": 4059 }, { "epoch": 0.47242262043286015, "grad_norm": 1.837311029434204, "learning_rate": 0.0005, "loss": 2.4823532104492188, "step": 4060 }, { "epoch": 0.47253898068419825, "grad_norm": 1.1777087450027466, "learning_rate": 0.0005, "loss": 1.8181602954864502, "step": 4061 }, { "epoch": 0.4726553409355364, "grad_norm": 1.4577535390853882, "learning_rate": 0.0005, "loss": 2.842209815979004, "step": 4062 }, { "epoch": 0.47277170118687456, "grad_norm": 1.4292466640472412, "learning_rate": 0.0005, "loss": 2.987757682800293, "step": 4063 }, { "epoch": 0.4728880614382127, "grad_norm": 0.9269856810569763, "learning_rate": 0.0005, "loss": 2.155261278152466, "step": 4064 }, { "epoch": 0.47300442168955087, "grad_norm": 1.5201843976974487, "learning_rate": 0.0005, "loss": 3.4791314601898193, "step": 4065 }, { "epoch": 0.47312078194088897, "grad_norm": 1.9190831184387207, "learning_rate": 0.0005, "loss": 3.078578472137451, "step": 4066 }, { "epoch": 0.4732371421922271, "grad_norm": 0.68155837059021, "learning_rate": 0.0005, "loss": 1.9079807996749878, "step": 4067 }, { "epoch": 0.4733535024435653, "grad_norm": 2.869124412536621, "learning_rate": 0.0005, "loss": 2.744690418243408, "step": 4068 }, { "epoch": 0.47346986269490343, "grad_norm": 1.1134613752365112, "learning_rate": 0.0005, "loss": 2.6562981605529785, "step": 4069 }, { "epoch": 0.4735862229462416, "grad_norm": 1.7165461778640747, "learning_rate": 0.0005, "loss": 3.2373859882354736, "step": 4070 }, { "epoch": 0.4737025831975797, "grad_norm": 1.3220659494400024, "learning_rate": 0.0005, "loss": 2.241727828979492, "step": 4071 }, { "epoch": 0.47381894344891784, "grad_norm": 1.6266759634017944, "learning_rate": 0.0005, "loss": 2.6648831367492676, "step": 4072 }, { "epoch": 0.473935303700256, "grad_norm": 1.4763867855072021, "learning_rate": 0.0005, "loss": 3.001347541809082, "step": 4073 }, { "epoch": 0.47405166395159415, "grad_norm": 2.037285327911377, "learning_rate": 0.0005, "loss": 2.775043249130249, "step": 4074 }, { "epoch": 0.4741680242029323, "grad_norm": 0.8923532962799072, "learning_rate": 0.0005, "loss": 2.191007614135742, "step": 4075 }, { "epoch": 0.4742843844542704, "grad_norm": 1.1030582189559937, "learning_rate": 0.0005, "loss": 2.9059016704559326, "step": 4076 }, { "epoch": 0.47440074470560856, "grad_norm": 1.146384358406067, "learning_rate": 0.0005, "loss": 1.9642585515975952, "step": 4077 }, { "epoch": 0.4745171049569467, "grad_norm": 1.3954664468765259, "learning_rate": 0.0005, "loss": 1.6901707649230957, "step": 4078 }, { "epoch": 0.47463346520828487, "grad_norm": 0.911091148853302, "learning_rate": 0.0005, "loss": 2.057325601577759, "step": 4079 }, { "epoch": 0.474749825459623, "grad_norm": 1.4216269254684448, "learning_rate": 0.0005, "loss": 1.9843528270721436, "step": 4080 }, { "epoch": 0.4748661857109611, "grad_norm": 1.234220027923584, "learning_rate": 0.0005, "loss": 2.172354221343994, "step": 4081 }, { "epoch": 0.4749825459622993, "grad_norm": 1.5681416988372803, "learning_rate": 0.0005, "loss": 3.1116156578063965, "step": 4082 }, { "epoch": 0.47509890621363743, "grad_norm": 0.616345226764679, "learning_rate": 0.0005, "loss": 1.6687142848968506, "step": 4083 }, { "epoch": 0.4752152664649756, "grad_norm": 1.0498785972595215, "learning_rate": 0.0005, "loss": 2.1791656017303467, "step": 4084 }, { "epoch": 0.4753316267163137, "grad_norm": 1.051698923110962, "learning_rate": 0.0005, "loss": 1.8962801694869995, "step": 4085 }, { "epoch": 0.47544798696765184, "grad_norm": 1.1990541219711304, "learning_rate": 0.0005, "loss": 3.079745054244995, "step": 4086 }, { "epoch": 0.47556434721899, "grad_norm": 1.2905497550964355, "learning_rate": 0.0005, "loss": 2.470269203186035, "step": 4087 }, { "epoch": 0.47568070747032815, "grad_norm": 1.6923763751983643, "learning_rate": 0.0005, "loss": 2.760251760482788, "step": 4088 }, { "epoch": 0.4757970677216663, "grad_norm": 1.1103086471557617, "learning_rate": 0.0005, "loss": 2.567735195159912, "step": 4089 }, { "epoch": 0.4759134279730044, "grad_norm": 1.1522648334503174, "learning_rate": 0.0005, "loss": 1.7990397214889526, "step": 4090 }, { "epoch": 0.47602978822434255, "grad_norm": 1.7239704132080078, "learning_rate": 0.0005, "loss": 3.713188648223877, "step": 4091 }, { "epoch": 0.4761461484756807, "grad_norm": 1.4375358819961548, "learning_rate": 0.0005, "loss": 3.0670642852783203, "step": 4092 }, { "epoch": 0.47626250872701886, "grad_norm": 1.3876526355743408, "learning_rate": 0.0005, "loss": 3.453862190246582, "step": 4093 }, { "epoch": 0.476378868978357, "grad_norm": 1.6521697044372559, "learning_rate": 0.0005, "loss": 3.7187647819519043, "step": 4094 }, { "epoch": 0.4764952292296951, "grad_norm": 1.7893645763397217, "learning_rate": 0.0005, "loss": 3.072652816772461, "step": 4095 }, { "epoch": 0.47661158948103327, "grad_norm": 1.4629476070404053, "learning_rate": 0.0005, "loss": 2.443800449371338, "step": 4096 }, { "epoch": 0.4767279497323714, "grad_norm": 1.4026843309402466, "learning_rate": 0.0005, "loss": 1.9757994413375854, "step": 4097 }, { "epoch": 0.4768443099837096, "grad_norm": 1.5686686038970947, "learning_rate": 0.0005, "loss": 2.3409478664398193, "step": 4098 }, { "epoch": 0.47696067023504773, "grad_norm": 1.7387734651565552, "learning_rate": 0.0005, "loss": 2.0833935737609863, "step": 4099 }, { "epoch": 0.47707703048638583, "grad_norm": 1.2909200191497803, "learning_rate": 0.0005, "loss": 2.8012144565582275, "step": 4100 }, { "epoch": 0.477193390737724, "grad_norm": 1.9844845533370972, "learning_rate": 0.0005, "loss": 3.497922658920288, "step": 4101 }, { "epoch": 0.47730975098906214, "grad_norm": 2.24552845954895, "learning_rate": 0.0005, "loss": 2.4987597465515137, "step": 4102 }, { "epoch": 0.4774261112404003, "grad_norm": 1.7096978425979614, "learning_rate": 0.0005, "loss": 3.2529354095458984, "step": 4103 }, { "epoch": 0.4775424714917384, "grad_norm": 0.963129460811615, "learning_rate": 0.0005, "loss": 1.7962645292282104, "step": 4104 }, { "epoch": 0.47765883174307655, "grad_norm": 2.0189735889434814, "learning_rate": 0.0005, "loss": 3.6613171100616455, "step": 4105 }, { "epoch": 0.4777751919944147, "grad_norm": 3.0924012660980225, "learning_rate": 0.0005, "loss": 2.9493329524993896, "step": 4106 }, { "epoch": 0.47789155224575286, "grad_norm": 1.4587485790252686, "learning_rate": 0.0005, "loss": 2.907156467437744, "step": 4107 }, { "epoch": 0.478007912497091, "grad_norm": 0.9636668562889099, "learning_rate": 0.0005, "loss": 2.059835433959961, "step": 4108 }, { "epoch": 0.4781242727484291, "grad_norm": 1.6072510480880737, "learning_rate": 0.0005, "loss": 3.313129425048828, "step": 4109 }, { "epoch": 0.47824063299976727, "grad_norm": 1.6510123014450073, "learning_rate": 0.0005, "loss": 3.3616108894348145, "step": 4110 }, { "epoch": 0.4783569932511054, "grad_norm": 0.665504515171051, "learning_rate": 0.0005, "loss": 1.883048415184021, "step": 4111 }, { "epoch": 0.4784733535024436, "grad_norm": 1.5324718952178955, "learning_rate": 0.0005, "loss": 2.669980525970459, "step": 4112 }, { "epoch": 0.47858971375378173, "grad_norm": 1.4090367555618286, "learning_rate": 0.0005, "loss": 2.658444881439209, "step": 4113 }, { "epoch": 0.47870607400511983, "grad_norm": 1.6729121208190918, "learning_rate": 0.0005, "loss": 3.443936824798584, "step": 4114 }, { "epoch": 0.478822434256458, "grad_norm": 10.615257263183594, "learning_rate": 0.0005, "loss": 2.198882579803467, "step": 4115 }, { "epoch": 0.47893879450779614, "grad_norm": 1.6148366928100586, "learning_rate": 0.0005, "loss": 2.2260279655456543, "step": 4116 }, { "epoch": 0.4790551547591343, "grad_norm": 2.3209609985351562, "learning_rate": 0.0005, "loss": 2.5519211292266846, "step": 4117 }, { "epoch": 0.47917151501047245, "grad_norm": 1.4247139692306519, "learning_rate": 0.0005, "loss": 2.15736722946167, "step": 4118 }, { "epoch": 0.47928787526181055, "grad_norm": 1.3411707878112793, "learning_rate": 0.0005, "loss": 2.052170991897583, "step": 4119 }, { "epoch": 0.4794042355131487, "grad_norm": 2.073439836502075, "learning_rate": 0.0005, "loss": 2.49983549118042, "step": 4120 }, { "epoch": 0.47952059576448686, "grad_norm": 1.1485110521316528, "learning_rate": 0.0005, "loss": 2.06125807762146, "step": 4121 }, { "epoch": 0.479636956015825, "grad_norm": 1.021728515625, "learning_rate": 0.0005, "loss": 1.760751485824585, "step": 4122 }, { "epoch": 0.4797533162671631, "grad_norm": 0.8547714352607727, "learning_rate": 0.0005, "loss": 1.6664589643478394, "step": 4123 }, { "epoch": 0.47986967651850126, "grad_norm": 1.6232200860977173, "learning_rate": 0.0005, "loss": 2.9666175842285156, "step": 4124 }, { "epoch": 0.4799860367698394, "grad_norm": 2.381831407546997, "learning_rate": 0.0005, "loss": 3.0976402759552, "step": 4125 }, { "epoch": 0.4801023970211776, "grad_norm": 2.0837059020996094, "learning_rate": 0.0005, "loss": 3.1338276863098145, "step": 4126 }, { "epoch": 0.48021875727251573, "grad_norm": 1.4040296077728271, "learning_rate": 0.0005, "loss": 2.3998565673828125, "step": 4127 }, { "epoch": 0.4803351175238538, "grad_norm": 1.7058953046798706, "learning_rate": 0.0005, "loss": 2.2842698097229004, "step": 4128 }, { "epoch": 0.480451477775192, "grad_norm": 1.6561944484710693, "learning_rate": 0.0005, "loss": 2.612415313720703, "step": 4129 }, { "epoch": 0.48056783802653014, "grad_norm": 1.4450079202651978, "learning_rate": 0.0005, "loss": 3.5095396041870117, "step": 4130 }, { "epoch": 0.4806841982778683, "grad_norm": 1.498691439628601, "learning_rate": 0.0005, "loss": 2.574190855026245, "step": 4131 }, { "epoch": 0.48080055852920645, "grad_norm": 1.476399540901184, "learning_rate": 0.0005, "loss": 3.021662712097168, "step": 4132 }, { "epoch": 0.48091691878054454, "grad_norm": 2.913567543029785, "learning_rate": 0.0005, "loss": 3.2052597999572754, "step": 4133 }, { "epoch": 0.4810332790318827, "grad_norm": 1.5442357063293457, "learning_rate": 0.0005, "loss": 2.411917209625244, "step": 4134 }, { "epoch": 0.48114963928322085, "grad_norm": 1.835145115852356, "learning_rate": 0.0005, "loss": 2.9444305896759033, "step": 4135 }, { "epoch": 0.481265999534559, "grad_norm": 1.3409042358398438, "learning_rate": 0.0005, "loss": 2.9562644958496094, "step": 4136 }, { "epoch": 0.48138235978589716, "grad_norm": 2.988974094390869, "learning_rate": 0.0005, "loss": 3.5597457885742188, "step": 4137 }, { "epoch": 0.48149872003723526, "grad_norm": 1.3429757356643677, "learning_rate": 0.0005, "loss": 3.0237832069396973, "step": 4138 }, { "epoch": 0.4816150802885734, "grad_norm": 2.04575514793396, "learning_rate": 0.0005, "loss": 2.547605037689209, "step": 4139 }, { "epoch": 0.48173144053991157, "grad_norm": 1.066178798675537, "learning_rate": 0.0005, "loss": 2.389616012573242, "step": 4140 }, { "epoch": 0.4818478007912497, "grad_norm": 1.496209979057312, "learning_rate": 0.0005, "loss": 2.379438877105713, "step": 4141 }, { "epoch": 0.4819641610425879, "grad_norm": 1.7507011890411377, "learning_rate": 0.0005, "loss": 2.8129994869232178, "step": 4142 }, { "epoch": 0.482080521293926, "grad_norm": 1.5486092567443848, "learning_rate": 0.0005, "loss": 2.9605278968811035, "step": 4143 }, { "epoch": 0.48219688154526413, "grad_norm": 0.8396853804588318, "learning_rate": 0.0005, "loss": 1.9496320486068726, "step": 4144 }, { "epoch": 0.4823132417966023, "grad_norm": 1.7434070110321045, "learning_rate": 0.0005, "loss": 2.023106336593628, "step": 4145 }, { "epoch": 0.48242960204794044, "grad_norm": 1.5310877561569214, "learning_rate": 0.0005, "loss": 2.962451219558716, "step": 4146 }, { "epoch": 0.48254596229927854, "grad_norm": 0.7042495012283325, "learning_rate": 0.0005, "loss": 1.7199989557266235, "step": 4147 }, { "epoch": 0.4826623225506167, "grad_norm": 1.4420840740203857, "learning_rate": 0.0005, "loss": 2.7382283210754395, "step": 4148 }, { "epoch": 0.48277868280195485, "grad_norm": 2.0353035926818848, "learning_rate": 0.0005, "loss": 2.5755207538604736, "step": 4149 }, { "epoch": 0.482895043053293, "grad_norm": 1.7504761219024658, "learning_rate": 0.0005, "loss": 2.220349073410034, "step": 4150 }, { "epoch": 0.48301140330463116, "grad_norm": 1.8590048551559448, "learning_rate": 0.0005, "loss": 2.2734217643737793, "step": 4151 }, { "epoch": 0.48312776355596926, "grad_norm": 1.5705417394638062, "learning_rate": 0.0005, "loss": 2.8191471099853516, "step": 4152 }, { "epoch": 0.4832441238073074, "grad_norm": 1.6683181524276733, "learning_rate": 0.0005, "loss": 2.8280744552612305, "step": 4153 }, { "epoch": 0.48336048405864557, "grad_norm": 1.0268020629882812, "learning_rate": 0.0005, "loss": 2.8873887062072754, "step": 4154 }, { "epoch": 0.4834768443099837, "grad_norm": 1.3892539739608765, "learning_rate": 0.0005, "loss": 3.24395751953125, "step": 4155 }, { "epoch": 0.4835932045613219, "grad_norm": 1.3641951084136963, "learning_rate": 0.0005, "loss": 2.963520050048828, "step": 4156 }, { "epoch": 0.48370956481266, "grad_norm": 1.4103639125823975, "learning_rate": 0.0005, "loss": 2.2395739555358887, "step": 4157 }, { "epoch": 0.48382592506399813, "grad_norm": 1.0421756505966187, "learning_rate": 0.0005, "loss": 2.4847090244293213, "step": 4158 }, { "epoch": 0.4839422853153363, "grad_norm": 1.6348321437835693, "learning_rate": 0.0005, "loss": 2.858574151992798, "step": 4159 }, { "epoch": 0.48405864556667444, "grad_norm": 1.0761163234710693, "learning_rate": 0.0005, "loss": 2.8634305000305176, "step": 4160 }, { "epoch": 0.4841750058180126, "grad_norm": 2.0337936878204346, "learning_rate": 0.0005, "loss": 3.2497692108154297, "step": 4161 }, { "epoch": 0.4842913660693507, "grad_norm": 0.9996312856674194, "learning_rate": 0.0005, "loss": 2.4556877613067627, "step": 4162 }, { "epoch": 0.48440772632068885, "grad_norm": 1.3405101299285889, "learning_rate": 0.0005, "loss": 1.7556114196777344, "step": 4163 }, { "epoch": 0.484524086572027, "grad_norm": 1.3061480522155762, "learning_rate": 0.0005, "loss": 1.8343474864959717, "step": 4164 }, { "epoch": 0.48464044682336516, "grad_norm": 1.3087385892868042, "learning_rate": 0.0005, "loss": 2.844256639480591, "step": 4165 }, { "epoch": 0.48475680707470326, "grad_norm": 1.2032428979873657, "learning_rate": 0.0005, "loss": 2.111981153488159, "step": 4166 }, { "epoch": 0.4848731673260414, "grad_norm": 1.6945830583572388, "learning_rate": 0.0005, "loss": 3.0469236373901367, "step": 4167 }, { "epoch": 0.48498952757737956, "grad_norm": 1.968410849571228, "learning_rate": 0.0005, "loss": 2.937554121017456, "step": 4168 }, { "epoch": 0.4851058878287177, "grad_norm": 1.6447138786315918, "learning_rate": 0.0005, "loss": 2.7763209342956543, "step": 4169 }, { "epoch": 0.4852222480800559, "grad_norm": 1.9397023916244507, "learning_rate": 0.0005, "loss": 3.480860710144043, "step": 4170 }, { "epoch": 0.485338608331394, "grad_norm": 1.454775094985962, "learning_rate": 0.0005, "loss": 1.918351411819458, "step": 4171 }, { "epoch": 0.4854549685827321, "grad_norm": 0.8257770538330078, "learning_rate": 0.0005, "loss": 2.352583408355713, "step": 4172 }, { "epoch": 0.4855713288340703, "grad_norm": 1.7360268831253052, "learning_rate": 0.0005, "loss": 3.375326633453369, "step": 4173 }, { "epoch": 0.48568768908540844, "grad_norm": 1.0658442974090576, "learning_rate": 0.0005, "loss": 1.8844330310821533, "step": 4174 }, { "epoch": 0.4858040493367466, "grad_norm": 1.1943316459655762, "learning_rate": 0.0005, "loss": 2.1270594596862793, "step": 4175 }, { "epoch": 0.4859204095880847, "grad_norm": 2.5520622730255127, "learning_rate": 0.0005, "loss": 2.5688552856445312, "step": 4176 }, { "epoch": 0.48603676983942284, "grad_norm": 0.9193720817565918, "learning_rate": 0.0005, "loss": 2.6461946964263916, "step": 4177 }, { "epoch": 0.486153130090761, "grad_norm": 0.9877060651779175, "learning_rate": 0.0005, "loss": 2.534675121307373, "step": 4178 }, { "epoch": 0.48626949034209915, "grad_norm": 2.340284824371338, "learning_rate": 0.0005, "loss": 3.4338340759277344, "step": 4179 }, { "epoch": 0.4863858505934373, "grad_norm": 1.807217001914978, "learning_rate": 0.0005, "loss": 2.9356656074523926, "step": 4180 }, { "epoch": 0.4865022108447754, "grad_norm": 1.282764196395874, "learning_rate": 0.0005, "loss": 1.7078940868377686, "step": 4181 }, { "epoch": 0.48661857109611356, "grad_norm": 1.537466049194336, "learning_rate": 0.0005, "loss": 2.4502878189086914, "step": 4182 }, { "epoch": 0.4867349313474517, "grad_norm": 1.590583086013794, "learning_rate": 0.0005, "loss": 2.79282283782959, "step": 4183 }, { "epoch": 0.48685129159878987, "grad_norm": 1.5159574747085571, "learning_rate": 0.0005, "loss": 2.9404945373535156, "step": 4184 }, { "epoch": 0.48696765185012797, "grad_norm": 1.37322199344635, "learning_rate": 0.0005, "loss": 1.8317064046859741, "step": 4185 }, { "epoch": 0.4870840121014661, "grad_norm": 1.1967835426330566, "learning_rate": 0.0005, "loss": 1.9113200902938843, "step": 4186 }, { "epoch": 0.4872003723528043, "grad_norm": 1.5156038999557495, "learning_rate": 0.0005, "loss": 3.2137093544006348, "step": 4187 }, { "epoch": 0.48731673260414243, "grad_norm": 1.6462702751159668, "learning_rate": 0.0005, "loss": 2.140162467956543, "step": 4188 }, { "epoch": 0.4874330928554806, "grad_norm": 1.2082815170288086, "learning_rate": 0.0005, "loss": 2.764993190765381, "step": 4189 }, { "epoch": 0.4875494531068187, "grad_norm": 2.4354729652404785, "learning_rate": 0.0005, "loss": 2.965150833129883, "step": 4190 }, { "epoch": 0.48766581335815684, "grad_norm": 1.6633617877960205, "learning_rate": 0.0005, "loss": 2.8134212493896484, "step": 4191 }, { "epoch": 0.487782173609495, "grad_norm": 1.6433606147766113, "learning_rate": 0.0005, "loss": 1.847128987312317, "step": 4192 }, { "epoch": 0.48789853386083315, "grad_norm": 1.4789931774139404, "learning_rate": 0.0005, "loss": 2.5215566158294678, "step": 4193 }, { "epoch": 0.4880148941121713, "grad_norm": 1.4089851379394531, "learning_rate": 0.0005, "loss": 2.677415370941162, "step": 4194 }, { "epoch": 0.4881312543635094, "grad_norm": 1.1470872163772583, "learning_rate": 0.0005, "loss": 2.9172520637512207, "step": 4195 }, { "epoch": 0.48824761461484756, "grad_norm": 1.4295463562011719, "learning_rate": 0.0005, "loss": 2.107783079147339, "step": 4196 }, { "epoch": 0.4883639748661857, "grad_norm": 1.5711979866027832, "learning_rate": 0.0005, "loss": 2.8437671661376953, "step": 4197 }, { "epoch": 0.48848033511752387, "grad_norm": 1.4439148902893066, "learning_rate": 0.0005, "loss": 2.90124773979187, "step": 4198 }, { "epoch": 0.488596695368862, "grad_norm": 1.3836263418197632, "learning_rate": 0.0005, "loss": 2.490079164505005, "step": 4199 }, { "epoch": 0.4887130556202001, "grad_norm": 2.8426666259765625, "learning_rate": 0.0005, "loss": 3.5252134799957275, "step": 4200 }, { "epoch": 0.4888294158715383, "grad_norm": 1.568039059638977, "learning_rate": 0.0005, "loss": 2.2513339519500732, "step": 4201 }, { "epoch": 0.48894577612287643, "grad_norm": 1.353098750114441, "learning_rate": 0.0005, "loss": 2.037794589996338, "step": 4202 }, { "epoch": 0.4890621363742146, "grad_norm": 1.2291001081466675, "learning_rate": 0.0005, "loss": 1.7313801050186157, "step": 4203 }, { "epoch": 0.48917849662555274, "grad_norm": 1.433970332145691, "learning_rate": 0.0005, "loss": 2.7533674240112305, "step": 4204 }, { "epoch": 0.48929485687689084, "grad_norm": 1.4187190532684326, "learning_rate": 0.0005, "loss": 2.0333333015441895, "step": 4205 }, { "epoch": 0.489411217128229, "grad_norm": 1.4579784870147705, "learning_rate": 0.0005, "loss": 2.348107099533081, "step": 4206 }, { "epoch": 0.48952757737956715, "grad_norm": 1.6726996898651123, "learning_rate": 0.0005, "loss": 3.130567789077759, "step": 4207 }, { "epoch": 0.4896439376309053, "grad_norm": 1.23531174659729, "learning_rate": 0.0005, "loss": 2.8656182289123535, "step": 4208 }, { "epoch": 0.4897602978822434, "grad_norm": 1.459234595298767, "learning_rate": 0.0005, "loss": 2.195051670074463, "step": 4209 }, { "epoch": 0.48987665813358156, "grad_norm": 1.343945026397705, "learning_rate": 0.0005, "loss": 2.503082036972046, "step": 4210 }, { "epoch": 0.4899930183849197, "grad_norm": 1.5992622375488281, "learning_rate": 0.0005, "loss": 2.2033162117004395, "step": 4211 }, { "epoch": 0.49010937863625786, "grad_norm": 1.3503273725509644, "learning_rate": 0.0005, "loss": 2.485166072845459, "step": 4212 }, { "epoch": 0.490225738887596, "grad_norm": 1.8647288084030151, "learning_rate": 0.0005, "loss": 2.9769461154937744, "step": 4213 }, { "epoch": 0.4903420991389341, "grad_norm": 1.8084328174591064, "learning_rate": 0.0005, "loss": 3.558412551879883, "step": 4214 }, { "epoch": 0.49045845939027227, "grad_norm": 1.296066403388977, "learning_rate": 0.0005, "loss": 2.127403736114502, "step": 4215 }, { "epoch": 0.4905748196416104, "grad_norm": 1.5551295280456543, "learning_rate": 0.0005, "loss": 3.1592965126037598, "step": 4216 }, { "epoch": 0.4906911798929486, "grad_norm": 1.5602864027023315, "learning_rate": 0.0005, "loss": 2.4852707386016846, "step": 4217 }, { "epoch": 0.49080754014428674, "grad_norm": 1.3147205114364624, "learning_rate": 0.0005, "loss": 3.4033362865448, "step": 4218 }, { "epoch": 0.49092390039562483, "grad_norm": 1.2919620275497437, "learning_rate": 0.0005, "loss": 2.7133870124816895, "step": 4219 }, { "epoch": 0.491040260646963, "grad_norm": 0.9582745432853699, "learning_rate": 0.0005, "loss": 2.4812405109405518, "step": 4220 }, { "epoch": 0.49115662089830114, "grad_norm": 1.8236918449401855, "learning_rate": 0.0005, "loss": 2.8541946411132812, "step": 4221 }, { "epoch": 0.4912729811496393, "grad_norm": 1.5975697040557861, "learning_rate": 0.0005, "loss": 2.4625191688537598, "step": 4222 }, { "epoch": 0.49138934140097745, "grad_norm": 1.2094051837921143, "learning_rate": 0.0005, "loss": 1.9791693687438965, "step": 4223 }, { "epoch": 0.49150570165231555, "grad_norm": 1.4884952306747437, "learning_rate": 0.0005, "loss": 2.6773016452789307, "step": 4224 }, { "epoch": 0.4916220619036537, "grad_norm": 1.5250227451324463, "learning_rate": 0.0005, "loss": 2.575742721557617, "step": 4225 }, { "epoch": 0.49173842215499186, "grad_norm": 1.8630925416946411, "learning_rate": 0.0005, "loss": 3.2060434818267822, "step": 4226 }, { "epoch": 0.49185478240633, "grad_norm": 1.7407087087631226, "learning_rate": 0.0005, "loss": 1.588289737701416, "step": 4227 }, { "epoch": 0.4919711426576681, "grad_norm": 0.8899808526039124, "learning_rate": 0.0005, "loss": 1.361098051071167, "step": 4228 }, { "epoch": 0.49208750290900627, "grad_norm": 0.9927741885185242, "learning_rate": 0.0005, "loss": 1.758208155632019, "step": 4229 }, { "epoch": 0.4922038631603444, "grad_norm": 1.7565895318984985, "learning_rate": 0.0005, "loss": 3.3541908264160156, "step": 4230 }, { "epoch": 0.4923202234116826, "grad_norm": 1.9105944633483887, "learning_rate": 0.0005, "loss": 3.3497984409332275, "step": 4231 }, { "epoch": 0.49243658366302073, "grad_norm": 2.575422763824463, "learning_rate": 0.0005, "loss": 3.5881028175354004, "step": 4232 }, { "epoch": 0.49255294391435883, "grad_norm": 1.6813477277755737, "learning_rate": 0.0005, "loss": 3.214581251144409, "step": 4233 }, { "epoch": 0.492669304165697, "grad_norm": 1.3173099756240845, "learning_rate": 0.0005, "loss": 2.4709510803222656, "step": 4234 }, { "epoch": 0.49278566441703514, "grad_norm": 1.3049832582473755, "learning_rate": 0.0005, "loss": 1.8651328086853027, "step": 4235 }, { "epoch": 0.4929020246683733, "grad_norm": 1.2949748039245605, "learning_rate": 0.0005, "loss": 3.704664707183838, "step": 4236 }, { "epoch": 0.49301838491971145, "grad_norm": 2.764672040939331, "learning_rate": 0.0005, "loss": 2.3390748500823975, "step": 4237 }, { "epoch": 0.49313474517104955, "grad_norm": 1.7432652711868286, "learning_rate": 0.0005, "loss": 3.1023032665252686, "step": 4238 }, { "epoch": 0.4932511054223877, "grad_norm": 1.8511908054351807, "learning_rate": 0.0005, "loss": 2.640681266784668, "step": 4239 }, { "epoch": 0.49336746567372586, "grad_norm": 2.0726711750030518, "learning_rate": 0.0005, "loss": 1.8516051769256592, "step": 4240 }, { "epoch": 0.493483825925064, "grad_norm": 0.9656800031661987, "learning_rate": 0.0005, "loss": 2.3563058376312256, "step": 4241 }, { "epoch": 0.49360018617640217, "grad_norm": 1.3363420963287354, "learning_rate": 0.0005, "loss": 2.124089241027832, "step": 4242 }, { "epoch": 0.49371654642774027, "grad_norm": 2.26653790473938, "learning_rate": 0.0005, "loss": 3.4103903770446777, "step": 4243 }, { "epoch": 0.4938329066790784, "grad_norm": 1.6640828847885132, "learning_rate": 0.0005, "loss": 3.6913485527038574, "step": 4244 }, { "epoch": 0.4939492669304166, "grad_norm": 0.762977123260498, "learning_rate": 0.0005, "loss": 1.3196009397506714, "step": 4245 }, { "epoch": 0.49406562718175473, "grad_norm": 1.733230471611023, "learning_rate": 0.0005, "loss": 2.745753288269043, "step": 4246 }, { "epoch": 0.49418198743309283, "grad_norm": 1.6656098365783691, "learning_rate": 0.0005, "loss": 3.0565290451049805, "step": 4247 }, { "epoch": 0.494298347684431, "grad_norm": 1.0865321159362793, "learning_rate": 0.0005, "loss": 2.5784401893615723, "step": 4248 }, { "epoch": 0.49441470793576914, "grad_norm": 1.586242914199829, "learning_rate": 0.0005, "loss": 2.3382396697998047, "step": 4249 }, { "epoch": 0.4945310681871073, "grad_norm": 0.9334681034088135, "learning_rate": 0.0005, "loss": 1.7738022804260254, "step": 4250 }, { "epoch": 0.49464742843844545, "grad_norm": 1.7821484804153442, "learning_rate": 0.0005, "loss": 3.6903700828552246, "step": 4251 }, { "epoch": 0.49476378868978355, "grad_norm": 1.3861855268478394, "learning_rate": 0.0005, "loss": 2.0523877143859863, "step": 4252 }, { "epoch": 0.4948801489411217, "grad_norm": 1.218277931213379, "learning_rate": 0.0005, "loss": 2.9783859252929688, "step": 4253 }, { "epoch": 0.49499650919245985, "grad_norm": 1.5563364028930664, "learning_rate": 0.0005, "loss": 2.3685550689697266, "step": 4254 }, { "epoch": 0.495112869443798, "grad_norm": 1.1002699136734009, "learning_rate": 0.0005, "loss": 2.8541839122772217, "step": 4255 }, { "epoch": 0.49522922969513616, "grad_norm": 1.4941718578338623, "learning_rate": 0.0005, "loss": 1.9573556184768677, "step": 4256 }, { "epoch": 0.49534558994647426, "grad_norm": 1.3255953788757324, "learning_rate": 0.0005, "loss": 2.2813210487365723, "step": 4257 }, { "epoch": 0.4954619501978124, "grad_norm": 0.921146035194397, "learning_rate": 0.0005, "loss": 2.2623844146728516, "step": 4258 }, { "epoch": 0.49557831044915057, "grad_norm": 1.257340431213379, "learning_rate": 0.0005, "loss": 2.578965187072754, "step": 4259 }, { "epoch": 0.4956946707004887, "grad_norm": 1.9023772478103638, "learning_rate": 0.0005, "loss": 2.5567309856414795, "step": 4260 }, { "epoch": 0.4958110309518269, "grad_norm": 2.038750171661377, "learning_rate": 0.0005, "loss": 2.300724983215332, "step": 4261 }, { "epoch": 0.495927391203165, "grad_norm": 1.2723140716552734, "learning_rate": 0.0005, "loss": 2.9824020862579346, "step": 4262 }, { "epoch": 0.49604375145450313, "grad_norm": 1.601718544960022, "learning_rate": 0.0005, "loss": 2.5086424350738525, "step": 4263 }, { "epoch": 0.4961601117058413, "grad_norm": 0.7780627608299255, "learning_rate": 0.0005, "loss": 1.4024338722229004, "step": 4264 }, { "epoch": 0.49627647195717944, "grad_norm": 1.471816897392273, "learning_rate": 0.0005, "loss": 3.1004157066345215, "step": 4265 }, { "epoch": 0.4963928322085176, "grad_norm": 2.332413911819458, "learning_rate": 0.0005, "loss": 2.604769229888916, "step": 4266 }, { "epoch": 0.4965091924598557, "grad_norm": 1.3438984155654907, "learning_rate": 0.0005, "loss": 2.476053476333618, "step": 4267 }, { "epoch": 0.49662555271119385, "grad_norm": 1.8793271780014038, "learning_rate": 0.0005, "loss": 2.9725165367126465, "step": 4268 }, { "epoch": 0.496741912962532, "grad_norm": 1.264726161956787, "learning_rate": 0.0005, "loss": 2.4925713539123535, "step": 4269 }, { "epoch": 0.49685827321387016, "grad_norm": 0.7504220604896545, "learning_rate": 0.0005, "loss": 1.847270131111145, "step": 4270 }, { "epoch": 0.49697463346520826, "grad_norm": 0.7614720463752747, "learning_rate": 0.0005, "loss": 2.116511821746826, "step": 4271 }, { "epoch": 0.4970909937165464, "grad_norm": 1.630852460861206, "learning_rate": 0.0005, "loss": 3.228053092956543, "step": 4272 }, { "epoch": 0.49720735396788457, "grad_norm": 1.1017518043518066, "learning_rate": 0.0005, "loss": 2.4080021381378174, "step": 4273 }, { "epoch": 0.4973237142192227, "grad_norm": 0.6819878816604614, "learning_rate": 0.0005, "loss": 1.610389232635498, "step": 4274 }, { "epoch": 0.4974400744705609, "grad_norm": 2.532632827758789, "learning_rate": 0.0005, "loss": 2.9697699546813965, "step": 4275 }, { "epoch": 0.497556434721899, "grad_norm": 1.9287813901901245, "learning_rate": 0.0005, "loss": 2.379671096801758, "step": 4276 }, { "epoch": 0.49767279497323713, "grad_norm": 1.285784363746643, "learning_rate": 0.0005, "loss": 2.8883185386657715, "step": 4277 }, { "epoch": 0.4977891552245753, "grad_norm": 1.5826822519302368, "learning_rate": 0.0005, "loss": 2.7485783100128174, "step": 4278 }, { "epoch": 0.49790551547591344, "grad_norm": 1.6777145862579346, "learning_rate": 0.0005, "loss": 3.238656520843506, "step": 4279 }, { "epoch": 0.4980218757272516, "grad_norm": 1.8114548921585083, "learning_rate": 0.0005, "loss": 2.7610373497009277, "step": 4280 }, { "epoch": 0.4981382359785897, "grad_norm": 1.505791425704956, "learning_rate": 0.0005, "loss": 2.5101165771484375, "step": 4281 }, { "epoch": 0.49825459622992785, "grad_norm": 1.1923967599868774, "learning_rate": 0.0005, "loss": 2.5225768089294434, "step": 4282 }, { "epoch": 0.498370956481266, "grad_norm": 3.0954036712646484, "learning_rate": 0.0005, "loss": 2.3982086181640625, "step": 4283 }, { "epoch": 0.49848731673260416, "grad_norm": 1.4173287153244019, "learning_rate": 0.0005, "loss": 3.017137289047241, "step": 4284 }, { "epoch": 0.4986036769839423, "grad_norm": 1.0551044940948486, "learning_rate": 0.0005, "loss": 1.7130136489868164, "step": 4285 }, { "epoch": 0.4987200372352804, "grad_norm": 1.5201116800308228, "learning_rate": 0.0005, "loss": 2.852463960647583, "step": 4286 }, { "epoch": 0.49883639748661857, "grad_norm": 1.3231415748596191, "learning_rate": 0.0005, "loss": 1.8216161727905273, "step": 4287 }, { "epoch": 0.4989527577379567, "grad_norm": 1.4097390174865723, "learning_rate": 0.0005, "loss": 2.218219757080078, "step": 4288 }, { "epoch": 0.4990691179892949, "grad_norm": 1.4492145776748657, "learning_rate": 0.0005, "loss": 3.4025070667266846, "step": 4289 }, { "epoch": 0.499185478240633, "grad_norm": 1.5015431642532349, "learning_rate": 0.0005, "loss": 2.308976888656616, "step": 4290 }, { "epoch": 0.49930183849197113, "grad_norm": 2.021812915802002, "learning_rate": 0.0005, "loss": 2.9478654861450195, "step": 4291 }, { "epoch": 0.4994181987433093, "grad_norm": 1.38611900806427, "learning_rate": 0.0005, "loss": 2.6963963508605957, "step": 4292 }, { "epoch": 0.49953455899464744, "grad_norm": 0.8640482425689697, "learning_rate": 0.0005, "loss": 1.5134563446044922, "step": 4293 }, { "epoch": 0.4996509192459856, "grad_norm": 2.000326156616211, "learning_rate": 0.0005, "loss": 2.6734728813171387, "step": 4294 }, { "epoch": 0.4997672794973237, "grad_norm": 1.6556589603424072, "learning_rate": 0.0005, "loss": 1.5573055744171143, "step": 4295 }, { "epoch": 0.49988363974866185, "grad_norm": 1.5715916156768799, "learning_rate": 0.0005, "loss": 3.513840675354004, "step": 4296 }, { "epoch": 0.5, "grad_norm": 0.7529677748680115, "learning_rate": 0.0005, "loss": 1.6043522357940674, "step": 4297 }, { "epoch": 0.5001163602513381, "grad_norm": 1.445060133934021, "learning_rate": 0.0005, "loss": 2.3190231323242188, "step": 4298 }, { "epoch": 0.5002327205026763, "grad_norm": 1.3905092477798462, "learning_rate": 0.0005, "loss": 2.7421317100524902, "step": 4299 }, { "epoch": 0.5003490807540144, "grad_norm": 1.6013423204421997, "learning_rate": 0.0005, "loss": 2.037733554840088, "step": 4300 }, { "epoch": 0.5004654410053526, "grad_norm": 1.0699832439422607, "learning_rate": 0.0005, "loss": 2.036468267440796, "step": 4301 }, { "epoch": 0.5005818012566907, "grad_norm": 1.7912687063217163, "learning_rate": 0.0005, "loss": 2.8137869834899902, "step": 4302 }, { "epoch": 0.5006981615080288, "grad_norm": 1.8515831232070923, "learning_rate": 0.0005, "loss": 2.9837613105773926, "step": 4303 }, { "epoch": 0.500814521759367, "grad_norm": 1.6791424751281738, "learning_rate": 0.0005, "loss": 2.454643726348877, "step": 4304 }, { "epoch": 0.5009308820107051, "grad_norm": 0.9154033064842224, "learning_rate": 0.0005, "loss": 1.8021129369735718, "step": 4305 }, { "epoch": 0.5010472422620433, "grad_norm": 1.653742790222168, "learning_rate": 0.0005, "loss": 2.847588539123535, "step": 4306 }, { "epoch": 0.5011636025133814, "grad_norm": 1.1242668628692627, "learning_rate": 0.0005, "loss": 2.458972692489624, "step": 4307 }, { "epoch": 0.5012799627647195, "grad_norm": 1.3550723791122437, "learning_rate": 0.0005, "loss": 2.4367833137512207, "step": 4308 }, { "epoch": 0.5013963230160577, "grad_norm": 1.1209900379180908, "learning_rate": 0.0005, "loss": 2.483053684234619, "step": 4309 }, { "epoch": 0.5015126832673958, "grad_norm": 1.54885995388031, "learning_rate": 0.0005, "loss": 2.366821765899658, "step": 4310 }, { "epoch": 0.501629043518734, "grad_norm": 2.300008773803711, "learning_rate": 0.0005, "loss": 2.182558298110962, "step": 4311 }, { "epoch": 0.5017454037700722, "grad_norm": 2.10150408744812, "learning_rate": 0.0005, "loss": 2.5544309616088867, "step": 4312 }, { "epoch": 0.5018617640214103, "grad_norm": 1.1077055931091309, "learning_rate": 0.0005, "loss": 2.4516987800598145, "step": 4313 }, { "epoch": 0.5019781242727485, "grad_norm": 0.9392231702804565, "learning_rate": 0.0005, "loss": 2.0085718631744385, "step": 4314 }, { "epoch": 0.5020944845240866, "grad_norm": 1.5466656684875488, "learning_rate": 0.0005, "loss": 2.764435291290283, "step": 4315 }, { "epoch": 0.5022108447754248, "grad_norm": 1.232458472251892, "learning_rate": 0.0005, "loss": 2.2724575996398926, "step": 4316 }, { "epoch": 0.5023272050267629, "grad_norm": 1.0138405561447144, "learning_rate": 0.0005, "loss": 1.759016513824463, "step": 4317 }, { "epoch": 0.502443565278101, "grad_norm": 1.2544325590133667, "learning_rate": 0.0005, "loss": 3.482625961303711, "step": 4318 }, { "epoch": 0.5025599255294392, "grad_norm": 1.6006461381912231, "learning_rate": 0.0005, "loss": 2.9154350757598877, "step": 4319 }, { "epoch": 0.5026762857807773, "grad_norm": 1.2648411989212036, "learning_rate": 0.0005, "loss": 1.8014332056045532, "step": 4320 }, { "epoch": 0.5027926460321154, "grad_norm": 1.756784200668335, "learning_rate": 0.0005, "loss": 2.6039962768554688, "step": 4321 }, { "epoch": 0.5029090062834536, "grad_norm": 1.3359638452529907, "learning_rate": 0.0005, "loss": 2.6326706409454346, "step": 4322 }, { "epoch": 0.5030253665347917, "grad_norm": 1.6321698427200317, "learning_rate": 0.0005, "loss": 2.6451117992401123, "step": 4323 }, { "epoch": 0.5031417267861299, "grad_norm": 1.2180019617080688, "learning_rate": 0.0005, "loss": 3.2421045303344727, "step": 4324 }, { "epoch": 0.503258087037468, "grad_norm": 1.6632449626922607, "learning_rate": 0.0005, "loss": 2.3821637630462646, "step": 4325 }, { "epoch": 0.5033744472888061, "grad_norm": 0.5251306891441345, "learning_rate": 0.0005, "loss": 1.0541353225708008, "step": 4326 }, { "epoch": 0.5034908075401443, "grad_norm": 1.0429275035858154, "learning_rate": 0.0005, "loss": 1.8849785327911377, "step": 4327 }, { "epoch": 0.5036071677914824, "grad_norm": 1.2568315267562866, "learning_rate": 0.0005, "loss": 2.051234722137451, "step": 4328 }, { "epoch": 0.5037235280428206, "grad_norm": 1.1803728342056274, "learning_rate": 0.0005, "loss": 2.986225128173828, "step": 4329 }, { "epoch": 0.5038398882941587, "grad_norm": 1.264846682548523, "learning_rate": 0.0005, "loss": 3.596257448196411, "step": 4330 }, { "epoch": 0.5039562485454968, "grad_norm": 1.6771217584609985, "learning_rate": 0.0005, "loss": 2.778069496154785, "step": 4331 }, { "epoch": 0.504072608796835, "grad_norm": 1.7411271333694458, "learning_rate": 0.0005, "loss": 2.742675304412842, "step": 4332 }, { "epoch": 0.5041889690481731, "grad_norm": 1.1162751913070679, "learning_rate": 0.0005, "loss": 2.7423930168151855, "step": 4333 }, { "epoch": 0.5043053292995113, "grad_norm": 1.4541422128677368, "learning_rate": 0.0005, "loss": 2.886833906173706, "step": 4334 }, { "epoch": 0.5044216895508494, "grad_norm": 1.7373820543289185, "learning_rate": 0.0005, "loss": 2.187844753265381, "step": 4335 }, { "epoch": 0.5045380498021875, "grad_norm": 1.966188669204712, "learning_rate": 0.0005, "loss": 2.5260162353515625, "step": 4336 }, { "epoch": 0.5046544100535257, "grad_norm": 1.2754170894622803, "learning_rate": 0.0005, "loss": 2.0483105182647705, "step": 4337 }, { "epoch": 0.5047707703048638, "grad_norm": 1.6568262577056885, "learning_rate": 0.0005, "loss": 2.1297030448913574, "step": 4338 }, { "epoch": 0.504887130556202, "grad_norm": 1.2242165803909302, "learning_rate": 0.0005, "loss": 2.8315930366516113, "step": 4339 }, { "epoch": 0.5050034908075401, "grad_norm": 1.2661627531051636, "learning_rate": 0.0005, "loss": 2.581486701965332, "step": 4340 }, { "epoch": 0.5051198510588782, "grad_norm": 1.0880348682403564, "learning_rate": 0.0005, "loss": 2.4204154014587402, "step": 4341 }, { "epoch": 0.5052362113102165, "grad_norm": 1.6096090078353882, "learning_rate": 0.0005, "loss": 2.3570635318756104, "step": 4342 }, { "epoch": 0.5053525715615546, "grad_norm": 1.7945795059204102, "learning_rate": 0.0005, "loss": 3.031529664993286, "step": 4343 }, { "epoch": 0.5054689318128928, "grad_norm": 1.5504841804504395, "learning_rate": 0.0005, "loss": 2.6368350982666016, "step": 4344 }, { "epoch": 0.5055852920642309, "grad_norm": 1.6264886856079102, "learning_rate": 0.0005, "loss": 2.6076316833496094, "step": 4345 }, { "epoch": 0.505701652315569, "grad_norm": 1.6578874588012695, "learning_rate": 0.0005, "loss": 2.340104103088379, "step": 4346 }, { "epoch": 0.5058180125669072, "grad_norm": 1.3695777654647827, "learning_rate": 0.0005, "loss": 2.199185371398926, "step": 4347 }, { "epoch": 0.5059343728182453, "grad_norm": 1.2245659828186035, "learning_rate": 0.0005, "loss": 1.9055910110473633, "step": 4348 }, { "epoch": 0.5060507330695835, "grad_norm": 2.688965082168579, "learning_rate": 0.0005, "loss": 2.47554087638855, "step": 4349 }, { "epoch": 0.5061670933209216, "grad_norm": 1.2648447751998901, "learning_rate": 0.0005, "loss": 2.0037734508514404, "step": 4350 }, { "epoch": 0.5062834535722597, "grad_norm": 1.436112880706787, "learning_rate": 0.0005, "loss": 1.9377944469451904, "step": 4351 }, { "epoch": 0.5063998138235979, "grad_norm": 1.449105978012085, "learning_rate": 0.0005, "loss": 2.789668560028076, "step": 4352 }, { "epoch": 0.506516174074936, "grad_norm": 1.564715027809143, "learning_rate": 0.0005, "loss": 2.3997764587402344, "step": 4353 }, { "epoch": 0.5066325343262742, "grad_norm": 1.2843036651611328, "learning_rate": 0.0005, "loss": 2.4280734062194824, "step": 4354 }, { "epoch": 0.5067488945776123, "grad_norm": 2.4824233055114746, "learning_rate": 0.0005, "loss": 1.249854564666748, "step": 4355 }, { "epoch": 0.5068652548289504, "grad_norm": 0.9486432671546936, "learning_rate": 0.0005, "loss": 1.6533870697021484, "step": 4356 }, { "epoch": 0.5069816150802886, "grad_norm": 2.17350172996521, "learning_rate": 0.0005, "loss": 2.940241813659668, "step": 4357 }, { "epoch": 0.5070979753316267, "grad_norm": 2.169586420059204, "learning_rate": 0.0005, "loss": 3.192417621612549, "step": 4358 }, { "epoch": 0.5072143355829648, "grad_norm": 0.7101598381996155, "learning_rate": 0.0005, "loss": 1.923311710357666, "step": 4359 }, { "epoch": 0.507330695834303, "grad_norm": 1.892429232597351, "learning_rate": 0.0005, "loss": 2.7399282455444336, "step": 4360 }, { "epoch": 0.5074470560856411, "grad_norm": 1.4079537391662598, "learning_rate": 0.0005, "loss": 2.182076930999756, "step": 4361 }, { "epoch": 0.5075634163369793, "grad_norm": 1.406116008758545, "learning_rate": 0.0005, "loss": 2.9780333042144775, "step": 4362 }, { "epoch": 0.5076797765883174, "grad_norm": 1.514574408531189, "learning_rate": 0.0005, "loss": 2.935410261154175, "step": 4363 }, { "epoch": 0.5077961368396555, "grad_norm": 1.913508415222168, "learning_rate": 0.0005, "loss": 2.8085994720458984, "step": 4364 }, { "epoch": 0.5079124970909937, "grad_norm": 1.7448464632034302, "learning_rate": 0.0005, "loss": 3.2663955688476562, "step": 4365 }, { "epoch": 0.5080288573423318, "grad_norm": 1.799521565437317, "learning_rate": 0.0005, "loss": 2.4637482166290283, "step": 4366 }, { "epoch": 0.50814521759367, "grad_norm": 1.2880854606628418, "learning_rate": 0.0005, "loss": 2.8667807579040527, "step": 4367 }, { "epoch": 0.5082615778450081, "grad_norm": 1.5337433815002441, "learning_rate": 0.0005, "loss": 3.112079381942749, "step": 4368 }, { "epoch": 0.5083779380963462, "grad_norm": 2.3982956409454346, "learning_rate": 0.0005, "loss": 2.6121840476989746, "step": 4369 }, { "epoch": 0.5084942983476844, "grad_norm": 1.4822932481765747, "learning_rate": 0.0005, "loss": 3.1612534523010254, "step": 4370 }, { "epoch": 0.5086106585990225, "grad_norm": 1.6687933206558228, "learning_rate": 0.0005, "loss": 2.13140869140625, "step": 4371 }, { "epoch": 0.5087270188503608, "grad_norm": 1.6832702159881592, "learning_rate": 0.0005, "loss": 3.0333855152130127, "step": 4372 }, { "epoch": 0.5088433791016989, "grad_norm": 1.737233281135559, "learning_rate": 0.0005, "loss": 2.838132858276367, "step": 4373 }, { "epoch": 0.508959739353037, "grad_norm": 1.4593639373779297, "learning_rate": 0.0005, "loss": 2.2542777061462402, "step": 4374 }, { "epoch": 0.5090760996043752, "grad_norm": 1.790753960609436, "learning_rate": 0.0005, "loss": 2.983762264251709, "step": 4375 }, { "epoch": 0.5091924598557133, "grad_norm": 1.0193418264389038, "learning_rate": 0.0005, "loss": 2.694983959197998, "step": 4376 }, { "epoch": 0.5093088201070515, "grad_norm": 1.9684956073760986, "learning_rate": 0.0005, "loss": 2.7775020599365234, "step": 4377 }, { "epoch": 0.5094251803583896, "grad_norm": 1.5240273475646973, "learning_rate": 0.0005, "loss": 2.795546054840088, "step": 4378 }, { "epoch": 0.5095415406097277, "grad_norm": 1.360435128211975, "learning_rate": 0.0005, "loss": 2.5835094451904297, "step": 4379 }, { "epoch": 0.5096579008610659, "grad_norm": 0.9372875094413757, "learning_rate": 0.0005, "loss": 2.1210014820098877, "step": 4380 }, { "epoch": 0.509774261112404, "grad_norm": 1.7640143632888794, "learning_rate": 0.0005, "loss": 3.4902431964874268, "step": 4381 }, { "epoch": 0.5098906213637422, "grad_norm": 1.2442305088043213, "learning_rate": 0.0005, "loss": 2.3188552856445312, "step": 4382 }, { "epoch": 0.5100069816150803, "grad_norm": 1.0300590991973877, "learning_rate": 0.0005, "loss": 1.6499950885772705, "step": 4383 }, { "epoch": 0.5101233418664184, "grad_norm": 1.675739049911499, "learning_rate": 0.0005, "loss": 2.927805185317993, "step": 4384 }, { "epoch": 0.5102397021177566, "grad_norm": 1.138453722000122, "learning_rate": 0.0005, "loss": 2.2259836196899414, "step": 4385 }, { "epoch": 0.5103560623690947, "grad_norm": 1.6376146078109741, "learning_rate": 0.0005, "loss": 2.87837553024292, "step": 4386 }, { "epoch": 0.5104724226204329, "grad_norm": 1.8917533159255981, "learning_rate": 0.0005, "loss": 2.2747411727905273, "step": 4387 }, { "epoch": 0.510588782871771, "grad_norm": 1.3844925165176392, "learning_rate": 0.0005, "loss": 1.876295566558838, "step": 4388 }, { "epoch": 0.5107051431231091, "grad_norm": 1.9536889791488647, "learning_rate": 0.0005, "loss": 2.0410709381103516, "step": 4389 }, { "epoch": 0.5108215033744473, "grad_norm": 1.5971417427062988, "learning_rate": 0.0005, "loss": 2.518174648284912, "step": 4390 }, { "epoch": 0.5109378636257854, "grad_norm": 1.6623553037643433, "learning_rate": 0.0005, "loss": 2.266361951828003, "step": 4391 }, { "epoch": 0.5110542238771236, "grad_norm": 1.2042380571365356, "learning_rate": 0.0005, "loss": 2.019437074661255, "step": 4392 }, { "epoch": 0.5111705841284617, "grad_norm": 2.09104061126709, "learning_rate": 0.0005, "loss": 2.9267373085021973, "step": 4393 }, { "epoch": 0.5112869443797998, "grad_norm": 0.9783738851547241, "learning_rate": 0.0005, "loss": 2.575603485107422, "step": 4394 }, { "epoch": 0.511403304631138, "grad_norm": 1.022363305091858, "learning_rate": 0.0005, "loss": 2.268000602722168, "step": 4395 }, { "epoch": 0.5115196648824761, "grad_norm": 1.3046116828918457, "learning_rate": 0.0005, "loss": 3.0696916580200195, "step": 4396 }, { "epoch": 0.5116360251338143, "grad_norm": 1.85861337184906, "learning_rate": 0.0005, "loss": 2.771963119506836, "step": 4397 }, { "epoch": 0.5117523853851524, "grad_norm": 1.006986379623413, "learning_rate": 0.0005, "loss": 2.2723727226257324, "step": 4398 }, { "epoch": 0.5118687456364905, "grad_norm": 1.2176982164382935, "learning_rate": 0.0005, "loss": 2.574941635131836, "step": 4399 }, { "epoch": 0.5119851058878288, "grad_norm": 1.3324333429336548, "learning_rate": 0.0005, "loss": 2.1500258445739746, "step": 4400 }, { "epoch": 0.5121014661391668, "grad_norm": 1.1759792566299438, "learning_rate": 0.0005, "loss": 2.7284157276153564, "step": 4401 }, { "epoch": 0.512217826390505, "grad_norm": 4.489726543426514, "learning_rate": 0.0005, "loss": 3.0056891441345215, "step": 4402 }, { "epoch": 0.5123341866418432, "grad_norm": 1.3425837755203247, "learning_rate": 0.0005, "loss": 2.754396438598633, "step": 4403 }, { "epoch": 0.5124505468931813, "grad_norm": 1.4545295238494873, "learning_rate": 0.0005, "loss": 2.212843894958496, "step": 4404 }, { "epoch": 0.5125669071445195, "grad_norm": 1.1476550102233887, "learning_rate": 0.0005, "loss": 3.1626949310302734, "step": 4405 }, { "epoch": 0.5126832673958576, "grad_norm": 1.2852405309677124, "learning_rate": 0.0005, "loss": 2.646486282348633, "step": 4406 }, { "epoch": 0.5127996276471957, "grad_norm": 1.5811433792114258, "learning_rate": 0.0005, "loss": 2.404158592224121, "step": 4407 }, { "epoch": 0.5129159878985339, "grad_norm": 0.8584755063056946, "learning_rate": 0.0005, "loss": 2.0783023834228516, "step": 4408 }, { "epoch": 0.513032348149872, "grad_norm": 0.8993431925773621, "learning_rate": 0.0005, "loss": 1.9903907775878906, "step": 4409 }, { "epoch": 0.5131487084012102, "grad_norm": 2.3549909591674805, "learning_rate": 0.0005, "loss": 2.073184013366699, "step": 4410 }, { "epoch": 0.5132650686525483, "grad_norm": 1.1792997121810913, "learning_rate": 0.0005, "loss": 2.803773880004883, "step": 4411 }, { "epoch": 0.5133814289038864, "grad_norm": 1.0668175220489502, "learning_rate": 0.0005, "loss": 2.4033563137054443, "step": 4412 }, { "epoch": 0.5134977891552246, "grad_norm": 0.8903965353965759, "learning_rate": 0.0005, "loss": 2.0185906887054443, "step": 4413 }, { "epoch": 0.5136141494065627, "grad_norm": 1.6379824876785278, "learning_rate": 0.0005, "loss": 3.3637375831604004, "step": 4414 }, { "epoch": 0.5137305096579009, "grad_norm": 1.0746347904205322, "learning_rate": 0.0005, "loss": 1.4811522960662842, "step": 4415 }, { "epoch": 0.513846869909239, "grad_norm": 1.6694687604904175, "learning_rate": 0.0005, "loss": 2.4404921531677246, "step": 4416 }, { "epoch": 0.5139632301605771, "grad_norm": 1.056274652481079, "learning_rate": 0.0005, "loss": 2.554901599884033, "step": 4417 }, { "epoch": 0.5140795904119153, "grad_norm": 1.7402416467666626, "learning_rate": 0.0005, "loss": 2.962339162826538, "step": 4418 }, { "epoch": 0.5141959506632534, "grad_norm": 1.056902527809143, "learning_rate": 0.0005, "loss": 2.1494767665863037, "step": 4419 }, { "epoch": 0.5143123109145916, "grad_norm": 1.5063605308532715, "learning_rate": 0.0005, "loss": 2.422481060028076, "step": 4420 }, { "epoch": 0.5144286711659297, "grad_norm": 0.942192792892456, "learning_rate": 0.0005, "loss": 2.3879458904266357, "step": 4421 }, { "epoch": 0.5145450314172678, "grad_norm": 1.1707170009613037, "learning_rate": 0.0005, "loss": 3.1713438034057617, "step": 4422 }, { "epoch": 0.514661391668606, "grad_norm": 1.6253688335418701, "learning_rate": 0.0005, "loss": 2.7905874252319336, "step": 4423 }, { "epoch": 0.5147777519199441, "grad_norm": 1.439302682876587, "learning_rate": 0.0005, "loss": 3.0935187339782715, "step": 4424 }, { "epoch": 0.5148941121712823, "grad_norm": 1.6724705696105957, "learning_rate": 0.0005, "loss": 3.1596555709838867, "step": 4425 }, { "epoch": 0.5150104724226204, "grad_norm": 1.498511552810669, "learning_rate": 0.0005, "loss": 3.9977293014526367, "step": 4426 }, { "epoch": 0.5151268326739585, "grad_norm": 1.2774509191513062, "learning_rate": 0.0005, "loss": 2.3385682106018066, "step": 4427 }, { "epoch": 0.5152431929252967, "grad_norm": 1.517087459564209, "learning_rate": 0.0005, "loss": 3.4175257682800293, "step": 4428 }, { "epoch": 0.5153595531766348, "grad_norm": 1.4216724634170532, "learning_rate": 0.0005, "loss": 1.978283405303955, "step": 4429 }, { "epoch": 0.515475913427973, "grad_norm": 1.3506606817245483, "learning_rate": 0.0005, "loss": 2.582263946533203, "step": 4430 }, { "epoch": 0.5155922736793112, "grad_norm": 1.4131494760513306, "learning_rate": 0.0005, "loss": 3.5091452598571777, "step": 4431 }, { "epoch": 0.5157086339306493, "grad_norm": 1.3084310293197632, "learning_rate": 0.0005, "loss": 2.038046360015869, "step": 4432 }, { "epoch": 0.5158249941819875, "grad_norm": 1.4067527055740356, "learning_rate": 0.0005, "loss": 2.2282509803771973, "step": 4433 }, { "epoch": 0.5159413544333256, "grad_norm": 1.7248032093048096, "learning_rate": 0.0005, "loss": 3.2741668224334717, "step": 4434 }, { "epoch": 0.5160577146846638, "grad_norm": 2.1222474575042725, "learning_rate": 0.0005, "loss": 3.402484655380249, "step": 4435 }, { "epoch": 0.5161740749360019, "grad_norm": 3.781834602355957, "learning_rate": 0.0005, "loss": 2.3399572372436523, "step": 4436 }, { "epoch": 0.51629043518734, "grad_norm": 1.6863731145858765, "learning_rate": 0.0005, "loss": 2.7257020473480225, "step": 4437 }, { "epoch": 0.5164067954386782, "grad_norm": 1.6786997318267822, "learning_rate": 0.0005, "loss": 2.596874237060547, "step": 4438 }, { "epoch": 0.5165231556900163, "grad_norm": 1.323737382888794, "learning_rate": 0.0005, "loss": 2.02144718170166, "step": 4439 }, { "epoch": 0.5166395159413545, "grad_norm": 1.5659974813461304, "learning_rate": 0.0005, "loss": 2.6544690132141113, "step": 4440 }, { "epoch": 0.5167558761926926, "grad_norm": 1.2275422811508179, "learning_rate": 0.0005, "loss": 2.4772491455078125, "step": 4441 }, { "epoch": 0.5168722364440307, "grad_norm": 1.4184529781341553, "learning_rate": 0.0005, "loss": 2.4523186683654785, "step": 4442 }, { "epoch": 0.5169885966953689, "grad_norm": 3.2533464431762695, "learning_rate": 0.0005, "loss": 2.884518623352051, "step": 4443 }, { "epoch": 0.517104956946707, "grad_norm": 1.0492863655090332, "learning_rate": 0.0005, "loss": 2.9992151260375977, "step": 4444 }, { "epoch": 0.5172213171980451, "grad_norm": 1.4365826845169067, "learning_rate": 0.0005, "loss": 2.5032827854156494, "step": 4445 }, { "epoch": 0.5173376774493833, "grad_norm": 1.1935601234436035, "learning_rate": 0.0005, "loss": 2.9887919425964355, "step": 4446 }, { "epoch": 0.5174540377007214, "grad_norm": 1.4232165813446045, "learning_rate": 0.0005, "loss": 3.127922296524048, "step": 4447 }, { "epoch": 0.5175703979520596, "grad_norm": 2.3181772232055664, "learning_rate": 0.0005, "loss": 2.069485902786255, "step": 4448 }, { "epoch": 0.5176867582033977, "grad_norm": 2.3150227069854736, "learning_rate": 0.0005, "loss": 2.1671695709228516, "step": 4449 }, { "epoch": 0.5178031184547358, "grad_norm": 1.094756841659546, "learning_rate": 0.0005, "loss": 1.3568062782287598, "step": 4450 }, { "epoch": 0.517919478706074, "grad_norm": 1.564326524734497, "learning_rate": 0.0005, "loss": 3.0787739753723145, "step": 4451 }, { "epoch": 0.5180358389574121, "grad_norm": 0.7401697039604187, "learning_rate": 0.0005, "loss": 1.6668174266815186, "step": 4452 }, { "epoch": 0.5181521992087503, "grad_norm": 2.259364128112793, "learning_rate": 0.0005, "loss": 2.5221705436706543, "step": 4453 }, { "epoch": 0.5182685594600884, "grad_norm": 1.340566635131836, "learning_rate": 0.0005, "loss": 2.7646820545196533, "step": 4454 }, { "epoch": 0.5183849197114265, "grad_norm": 1.0804712772369385, "learning_rate": 0.0005, "loss": 2.3097081184387207, "step": 4455 }, { "epoch": 0.5185012799627647, "grad_norm": 1.127339482307434, "learning_rate": 0.0005, "loss": 3.074251174926758, "step": 4456 }, { "epoch": 0.5186176402141028, "grad_norm": 1.4461160898208618, "learning_rate": 0.0005, "loss": 1.661721110343933, "step": 4457 }, { "epoch": 0.518734000465441, "grad_norm": 1.2073640823364258, "learning_rate": 0.0005, "loss": 3.2192649841308594, "step": 4458 }, { "epoch": 0.5188503607167791, "grad_norm": 4.683888912200928, "learning_rate": 0.0005, "loss": 2.7211155891418457, "step": 4459 }, { "epoch": 0.5189667209681172, "grad_norm": 1.8337070941925049, "learning_rate": 0.0005, "loss": 3.152644157409668, "step": 4460 }, { "epoch": 0.5190830812194555, "grad_norm": 1.3867348432540894, "learning_rate": 0.0005, "loss": 2.2916576862335205, "step": 4461 }, { "epoch": 0.5191994414707936, "grad_norm": 1.7012728452682495, "learning_rate": 0.0005, "loss": 3.5153913497924805, "step": 4462 }, { "epoch": 0.5193158017221318, "grad_norm": 0.6888840794563293, "learning_rate": 0.0005, "loss": 1.740912675857544, "step": 4463 }, { "epoch": 0.5194321619734699, "grad_norm": 1.9418309926986694, "learning_rate": 0.0005, "loss": 2.884401321411133, "step": 4464 }, { "epoch": 0.519548522224808, "grad_norm": 0.8473296761512756, "learning_rate": 0.0005, "loss": 1.7487537860870361, "step": 4465 }, { "epoch": 0.5196648824761462, "grad_norm": 1.1856926679611206, "learning_rate": 0.0005, "loss": 2.5355827808380127, "step": 4466 }, { "epoch": 0.5197812427274843, "grad_norm": 1.234639048576355, "learning_rate": 0.0005, "loss": 2.8008313179016113, "step": 4467 }, { "epoch": 0.5198976029788225, "grad_norm": 0.9823858737945557, "learning_rate": 0.0005, "loss": 2.341416835784912, "step": 4468 }, { "epoch": 0.5200139632301606, "grad_norm": 1.5185195207595825, "learning_rate": 0.0005, "loss": 3.4534080028533936, "step": 4469 }, { "epoch": 0.5201303234814987, "grad_norm": 1.1770074367523193, "learning_rate": 0.0005, "loss": 2.0372395515441895, "step": 4470 }, { "epoch": 0.5202466837328369, "grad_norm": 3.3727359771728516, "learning_rate": 0.0005, "loss": 2.933311939239502, "step": 4471 }, { "epoch": 0.520363043984175, "grad_norm": 1.1983212232589722, "learning_rate": 0.0005, "loss": 1.6332341432571411, "step": 4472 }, { "epoch": 0.5204794042355132, "grad_norm": 1.407438039779663, "learning_rate": 0.0005, "loss": 2.697768211364746, "step": 4473 }, { "epoch": 0.5205957644868513, "grad_norm": 0.9749016761779785, "learning_rate": 0.0005, "loss": 1.7696199417114258, "step": 4474 }, { "epoch": 0.5207121247381894, "grad_norm": 2.4016053676605225, "learning_rate": 0.0005, "loss": 2.8625967502593994, "step": 4475 }, { "epoch": 0.5208284849895276, "grad_norm": 1.7813080549240112, "learning_rate": 0.0005, "loss": 2.7342052459716797, "step": 4476 }, { "epoch": 0.5209448452408657, "grad_norm": 1.0035480260849, "learning_rate": 0.0005, "loss": 1.8196334838867188, "step": 4477 }, { "epoch": 0.5210612054922039, "grad_norm": 1.3858311176300049, "learning_rate": 0.0005, "loss": 2.258845567703247, "step": 4478 }, { "epoch": 0.521177565743542, "grad_norm": 0.7454908490180969, "learning_rate": 0.0005, "loss": 1.6429543495178223, "step": 4479 }, { "epoch": 0.5212939259948801, "grad_norm": 2.6323485374450684, "learning_rate": 0.0005, "loss": 3.2094132900238037, "step": 4480 }, { "epoch": 0.5214102862462183, "grad_norm": 1.4172775745391846, "learning_rate": 0.0005, "loss": 2.990307092666626, "step": 4481 }, { "epoch": 0.5215266464975564, "grad_norm": 1.1295671463012695, "learning_rate": 0.0005, "loss": 2.1988203525543213, "step": 4482 }, { "epoch": 0.5216430067488945, "grad_norm": 1.3452074527740479, "learning_rate": 0.0005, "loss": 2.147055149078369, "step": 4483 }, { "epoch": 0.5217593670002327, "grad_norm": 1.1919653415679932, "learning_rate": 0.0005, "loss": 2.272684335708618, "step": 4484 }, { "epoch": 0.5218757272515708, "grad_norm": 1.4366843700408936, "learning_rate": 0.0005, "loss": 3.360694408416748, "step": 4485 }, { "epoch": 0.521992087502909, "grad_norm": 1.1762393712997437, "learning_rate": 0.0005, "loss": 1.9947789907455444, "step": 4486 }, { "epoch": 0.5221084477542471, "grad_norm": 3.306978225708008, "learning_rate": 0.0005, "loss": 2.6699090003967285, "step": 4487 }, { "epoch": 0.5222248080055852, "grad_norm": 1.1134713888168335, "learning_rate": 0.0005, "loss": 3.4890248775482178, "step": 4488 }, { "epoch": 0.5223411682569234, "grad_norm": 0.4324381649494171, "learning_rate": 0.0005, "loss": 1.1240675449371338, "step": 4489 }, { "epoch": 0.5224575285082615, "grad_norm": 1.1833322048187256, "learning_rate": 0.0005, "loss": 2.2753515243530273, "step": 4490 }, { "epoch": 0.5225738887595998, "grad_norm": 2.11090350151062, "learning_rate": 0.0005, "loss": 2.189272403717041, "step": 4491 }, { "epoch": 0.5226902490109379, "grad_norm": 2.069639205932617, "learning_rate": 0.0005, "loss": 2.7071056365966797, "step": 4492 }, { "epoch": 0.522806609262276, "grad_norm": 1.8091833591461182, "learning_rate": 0.0005, "loss": 2.419377326965332, "step": 4493 }, { "epoch": 0.5229229695136142, "grad_norm": 1.921238660812378, "learning_rate": 0.0005, "loss": 2.9879965782165527, "step": 4494 }, { "epoch": 0.5230393297649523, "grad_norm": 1.5321344137191772, "learning_rate": 0.0005, "loss": 2.0439343452453613, "step": 4495 }, { "epoch": 0.5231556900162905, "grad_norm": 1.08760666847229, "learning_rate": 0.0005, "loss": 3.0379104614257812, "step": 4496 }, { "epoch": 0.5232720502676286, "grad_norm": 1.460315227508545, "learning_rate": 0.0005, "loss": 2.8210389614105225, "step": 4497 }, { "epoch": 0.5233884105189667, "grad_norm": 1.3835095167160034, "learning_rate": 0.0005, "loss": 2.160369396209717, "step": 4498 }, { "epoch": 0.5235047707703049, "grad_norm": 1.9382325410842896, "learning_rate": 0.0005, "loss": 3.9299349784851074, "step": 4499 }, { "epoch": 0.523621131021643, "grad_norm": 1.1962717771530151, "learning_rate": 0.0005, "loss": 2.1801228523254395, "step": 4500 }, { "epoch": 0.5237374912729812, "grad_norm": 1.5621376037597656, "learning_rate": 0.0005, "loss": 3.2903013229370117, "step": 4501 }, { "epoch": 0.5238538515243193, "grad_norm": 1.2975343465805054, "learning_rate": 0.0005, "loss": 2.773277759552002, "step": 4502 }, { "epoch": 0.5239702117756574, "grad_norm": 1.5228608846664429, "learning_rate": 0.0005, "loss": 2.4706592559814453, "step": 4503 }, { "epoch": 0.5240865720269956, "grad_norm": 1.5708156824111938, "learning_rate": 0.0005, "loss": 2.808809995651245, "step": 4504 }, { "epoch": 0.5242029322783337, "grad_norm": 1.5702348947525024, "learning_rate": 0.0005, "loss": 2.2328739166259766, "step": 4505 }, { "epoch": 0.5243192925296719, "grad_norm": 1.2933580875396729, "learning_rate": 0.0005, "loss": 2.3472414016723633, "step": 4506 }, { "epoch": 0.52443565278101, "grad_norm": 1.1389187574386597, "learning_rate": 0.0005, "loss": 2.579010009765625, "step": 4507 }, { "epoch": 0.5245520130323481, "grad_norm": 1.439256191253662, "learning_rate": 0.0005, "loss": 2.521049976348877, "step": 4508 }, { "epoch": 0.5246683732836863, "grad_norm": 0.9560222029685974, "learning_rate": 0.0005, "loss": 2.3985958099365234, "step": 4509 }, { "epoch": 0.5247847335350244, "grad_norm": 1.7222777605056763, "learning_rate": 0.0005, "loss": 2.834956645965576, "step": 4510 }, { "epoch": 0.5249010937863626, "grad_norm": 1.4962608814239502, "learning_rate": 0.0005, "loss": 1.9038724899291992, "step": 4511 }, { "epoch": 0.5250174540377007, "grad_norm": 1.4584640264511108, "learning_rate": 0.0005, "loss": 2.710200309753418, "step": 4512 }, { "epoch": 0.5251338142890388, "grad_norm": 2.0484421253204346, "learning_rate": 0.0005, "loss": 3.14176082611084, "step": 4513 }, { "epoch": 0.525250174540377, "grad_norm": 1.7944780588150024, "learning_rate": 0.0005, "loss": 2.8463807106018066, "step": 4514 }, { "epoch": 0.5253665347917151, "grad_norm": 1.1947914361953735, "learning_rate": 0.0005, "loss": 2.0904877185821533, "step": 4515 }, { "epoch": 0.5254828950430533, "grad_norm": 1.8214833736419678, "learning_rate": 0.0005, "loss": 2.5735228061676025, "step": 4516 }, { "epoch": 0.5255992552943914, "grad_norm": 1.93600332736969, "learning_rate": 0.0005, "loss": 2.9120006561279297, "step": 4517 }, { "epoch": 0.5257156155457295, "grad_norm": 1.5923758745193481, "learning_rate": 0.0005, "loss": 2.9357852935791016, "step": 4518 }, { "epoch": 0.5258319757970678, "grad_norm": 1.6170650720596313, "learning_rate": 0.0005, "loss": 2.21364688873291, "step": 4519 }, { "epoch": 0.5259483360484059, "grad_norm": 2.119196653366089, "learning_rate": 0.0005, "loss": 2.6635916233062744, "step": 4520 }, { "epoch": 0.5260646962997441, "grad_norm": 1.4183626174926758, "learning_rate": 0.0005, "loss": 2.9466724395751953, "step": 4521 }, { "epoch": 0.5261810565510822, "grad_norm": 1.7947163581848145, "learning_rate": 0.0005, "loss": 2.3965353965759277, "step": 4522 }, { "epoch": 0.5262974168024203, "grad_norm": 1.0889025926589966, "learning_rate": 0.0005, "loss": 3.0247645378112793, "step": 4523 }, { "epoch": 0.5264137770537585, "grad_norm": 1.9751441478729248, "learning_rate": 0.0005, "loss": 3.505237102508545, "step": 4524 }, { "epoch": 0.5265301373050966, "grad_norm": 1.3246890306472778, "learning_rate": 0.0005, "loss": 2.234658718109131, "step": 4525 }, { "epoch": 0.5266464975564347, "grad_norm": 1.082008719444275, "learning_rate": 0.0005, "loss": 1.4275522232055664, "step": 4526 }, { "epoch": 0.5267628578077729, "grad_norm": 1.1467478275299072, "learning_rate": 0.0005, "loss": 2.1709728240966797, "step": 4527 }, { "epoch": 0.526879218059111, "grad_norm": 1.5852186679840088, "learning_rate": 0.0005, "loss": 2.723227024078369, "step": 4528 }, { "epoch": 0.5269955783104492, "grad_norm": 1.8492118120193481, "learning_rate": 0.0005, "loss": 3.3129825592041016, "step": 4529 }, { "epoch": 0.5271119385617873, "grad_norm": 1.2913249731063843, "learning_rate": 0.0005, "loss": 2.86460280418396, "step": 4530 }, { "epoch": 0.5272282988131254, "grad_norm": 1.4646602869033813, "learning_rate": 0.0005, "loss": 2.929072380065918, "step": 4531 }, { "epoch": 0.5273446590644636, "grad_norm": 1.066957712173462, "learning_rate": 0.0005, "loss": 2.6642448902130127, "step": 4532 }, { "epoch": 0.5274610193158017, "grad_norm": 1.5363305807113647, "learning_rate": 0.0005, "loss": 3.810154914855957, "step": 4533 }, { "epoch": 0.5275773795671399, "grad_norm": 1.9873384237289429, "learning_rate": 0.0005, "loss": 2.51760196685791, "step": 4534 }, { "epoch": 0.527693739818478, "grad_norm": 0.7255672216415405, "learning_rate": 0.0005, "loss": 1.6936161518096924, "step": 4535 }, { "epoch": 0.5278101000698161, "grad_norm": 1.3753688335418701, "learning_rate": 0.0005, "loss": 3.063084602355957, "step": 4536 }, { "epoch": 0.5279264603211543, "grad_norm": 1.6502898931503296, "learning_rate": 0.0005, "loss": 3.102102756500244, "step": 4537 }, { "epoch": 0.5280428205724924, "grad_norm": 1.1016534566879272, "learning_rate": 0.0005, "loss": 1.738572597503662, "step": 4538 }, { "epoch": 0.5281591808238306, "grad_norm": 1.3378649950027466, "learning_rate": 0.0005, "loss": 3.565957546234131, "step": 4539 }, { "epoch": 0.5282755410751687, "grad_norm": 1.8594001531600952, "learning_rate": 0.0005, "loss": 3.2644145488739014, "step": 4540 }, { "epoch": 0.5283919013265068, "grad_norm": 1.3674787282943726, "learning_rate": 0.0005, "loss": 1.7000643014907837, "step": 4541 }, { "epoch": 0.528508261577845, "grad_norm": 1.6711913347244263, "learning_rate": 0.0005, "loss": 2.5747568607330322, "step": 4542 }, { "epoch": 0.5286246218291831, "grad_norm": 1.0622020959854126, "learning_rate": 0.0005, "loss": 2.234046459197998, "step": 4543 }, { "epoch": 0.5287409820805213, "grad_norm": 2.164855480194092, "learning_rate": 0.0005, "loss": 2.3578052520751953, "step": 4544 }, { "epoch": 0.5288573423318594, "grad_norm": 1.742938756942749, "learning_rate": 0.0005, "loss": 2.668780565261841, "step": 4545 }, { "epoch": 0.5289737025831975, "grad_norm": 1.545100450515747, "learning_rate": 0.0005, "loss": 1.950742483139038, "step": 4546 }, { "epoch": 0.5290900628345357, "grad_norm": 6.527382850646973, "learning_rate": 0.0005, "loss": 2.8164892196655273, "step": 4547 }, { "epoch": 0.5292064230858738, "grad_norm": 1.7952311038970947, "learning_rate": 0.0005, "loss": 2.1696009635925293, "step": 4548 }, { "epoch": 0.529322783337212, "grad_norm": 1.2568684816360474, "learning_rate": 0.0005, "loss": 2.5424094200134277, "step": 4549 }, { "epoch": 0.5294391435885502, "grad_norm": 1.6504714488983154, "learning_rate": 0.0005, "loss": 3.2880821228027344, "step": 4550 }, { "epoch": 0.5295555038398883, "grad_norm": 1.8508111238479614, "learning_rate": 0.0005, "loss": 2.8931965827941895, "step": 4551 }, { "epoch": 0.5296718640912265, "grad_norm": 1.6002839803695679, "learning_rate": 0.0005, "loss": 2.260873794555664, "step": 4552 }, { "epoch": 0.5297882243425646, "grad_norm": 1.7297054529190063, "learning_rate": 0.0005, "loss": 3.1887290477752686, "step": 4553 }, { "epoch": 0.5299045845939028, "grad_norm": 1.635890245437622, "learning_rate": 0.0005, "loss": 3.189302444458008, "step": 4554 }, { "epoch": 0.5300209448452409, "grad_norm": 1.346612811088562, "learning_rate": 0.0005, "loss": 2.068639039993286, "step": 4555 }, { "epoch": 0.530137305096579, "grad_norm": 1.4841821193695068, "learning_rate": 0.0005, "loss": 2.2632994651794434, "step": 4556 }, { "epoch": 0.5302536653479172, "grad_norm": 2.6203012466430664, "learning_rate": 0.0005, "loss": 2.64551043510437, "step": 4557 }, { "epoch": 0.5303700255992553, "grad_norm": 1.5962157249450684, "learning_rate": 0.0005, "loss": 3.036379337310791, "step": 4558 }, { "epoch": 0.5304863858505935, "grad_norm": 0.8845282793045044, "learning_rate": 0.0005, "loss": 1.6630752086639404, "step": 4559 }, { "epoch": 0.5306027461019316, "grad_norm": 1.5768355131149292, "learning_rate": 0.0005, "loss": 1.9300847053527832, "step": 4560 }, { "epoch": 0.5307191063532697, "grad_norm": 1.4439525604248047, "learning_rate": 0.0005, "loss": 3.080341100692749, "step": 4561 }, { "epoch": 0.5308354666046079, "grad_norm": 1.1414897441864014, "learning_rate": 0.0005, "loss": 2.8017523288726807, "step": 4562 }, { "epoch": 0.530951826855946, "grad_norm": 1.2289785146713257, "learning_rate": 0.0005, "loss": 1.8794044256210327, "step": 4563 }, { "epoch": 0.5310681871072842, "grad_norm": 1.687277913093567, "learning_rate": 0.0005, "loss": 2.3535282611846924, "step": 4564 }, { "epoch": 0.5311845473586223, "grad_norm": 1.6043264865875244, "learning_rate": 0.0005, "loss": 2.543463706970215, "step": 4565 }, { "epoch": 0.5313009076099604, "grad_norm": 1.7232370376586914, "learning_rate": 0.0005, "loss": 2.406209945678711, "step": 4566 }, { "epoch": 0.5314172678612986, "grad_norm": 2.1062374114990234, "learning_rate": 0.0005, "loss": 3.0015649795532227, "step": 4567 }, { "epoch": 0.5315336281126367, "grad_norm": 0.8645972609519958, "learning_rate": 0.0005, "loss": 1.6659951210021973, "step": 4568 }, { "epoch": 0.5316499883639748, "grad_norm": 1.6256579160690308, "learning_rate": 0.0005, "loss": 3.4537084102630615, "step": 4569 }, { "epoch": 0.531766348615313, "grad_norm": 1.4839580059051514, "learning_rate": 0.0005, "loss": 2.540355920791626, "step": 4570 }, { "epoch": 0.5318827088666511, "grad_norm": 1.147976279258728, "learning_rate": 0.0005, "loss": 3.0602221488952637, "step": 4571 }, { "epoch": 0.5319990691179893, "grad_norm": 1.1821304559707642, "learning_rate": 0.0005, "loss": 2.7771553993225098, "step": 4572 }, { "epoch": 0.5321154293693274, "grad_norm": 1.3375201225280762, "learning_rate": 0.0005, "loss": 3.096522331237793, "step": 4573 }, { "epoch": 0.5322317896206655, "grad_norm": 0.9750698804855347, "learning_rate": 0.0005, "loss": 1.8856773376464844, "step": 4574 }, { "epoch": 0.5323481498720037, "grad_norm": 1.7677679061889648, "learning_rate": 0.0005, "loss": 3.2892117500305176, "step": 4575 }, { "epoch": 0.5324645101233418, "grad_norm": 1.7228654623031616, "learning_rate": 0.0005, "loss": 3.0292816162109375, "step": 4576 }, { "epoch": 0.53258087037468, "grad_norm": 1.298804759979248, "learning_rate": 0.0005, "loss": 2.2394609451293945, "step": 4577 }, { "epoch": 0.5326972306260181, "grad_norm": 2.4984323978424072, "learning_rate": 0.0005, "loss": 2.5325088500976562, "step": 4578 }, { "epoch": 0.5328135908773562, "grad_norm": 1.815006971359253, "learning_rate": 0.0005, "loss": 2.7996325492858887, "step": 4579 }, { "epoch": 0.5329299511286945, "grad_norm": 2.945361375808716, "learning_rate": 0.0005, "loss": 3.295010566711426, "step": 4580 }, { "epoch": 0.5330463113800326, "grad_norm": 1.007115125656128, "learning_rate": 0.0005, "loss": 2.2916717529296875, "step": 4581 }, { "epoch": 0.5331626716313708, "grad_norm": 1.2158935070037842, "learning_rate": 0.0005, "loss": 2.449481248855591, "step": 4582 }, { "epoch": 0.5332790318827089, "grad_norm": 3.231476306915283, "learning_rate": 0.0005, "loss": 3.4945993423461914, "step": 4583 }, { "epoch": 0.533395392134047, "grad_norm": 1.4557452201843262, "learning_rate": 0.0005, "loss": 2.0950565338134766, "step": 4584 }, { "epoch": 0.5335117523853852, "grad_norm": 1.5199249982833862, "learning_rate": 0.0005, "loss": 1.9910211563110352, "step": 4585 }, { "epoch": 0.5336281126367233, "grad_norm": 1.467010259628296, "learning_rate": 0.0005, "loss": 2.692786693572998, "step": 4586 }, { "epoch": 0.5337444728880615, "grad_norm": 0.9294857978820801, "learning_rate": 0.0005, "loss": 2.312464952468872, "step": 4587 }, { "epoch": 0.5338608331393996, "grad_norm": 1.4258774518966675, "learning_rate": 0.0005, "loss": 3.4528000354766846, "step": 4588 }, { "epoch": 0.5339771933907377, "grad_norm": 1.0235490798950195, "learning_rate": 0.0005, "loss": 1.4951053857803345, "step": 4589 }, { "epoch": 0.5340935536420759, "grad_norm": 1.9264123439788818, "learning_rate": 0.0005, "loss": 3.5881690979003906, "step": 4590 }, { "epoch": 0.534209913893414, "grad_norm": 1.284294843673706, "learning_rate": 0.0005, "loss": 3.4475185871124268, "step": 4591 }, { "epoch": 0.5343262741447522, "grad_norm": 1.9108046293258667, "learning_rate": 0.0005, "loss": 3.1278579235076904, "step": 4592 }, { "epoch": 0.5344426343960903, "grad_norm": 1.6362006664276123, "learning_rate": 0.0005, "loss": 2.840012788772583, "step": 4593 }, { "epoch": 0.5345589946474284, "grad_norm": 1.244134545326233, "learning_rate": 0.0005, "loss": 1.9649338722229004, "step": 4594 }, { "epoch": 0.5346753548987666, "grad_norm": 0.9653854370117188, "learning_rate": 0.0005, "loss": 1.9962208271026611, "step": 4595 }, { "epoch": 0.5347917151501047, "grad_norm": 1.5848264694213867, "learning_rate": 0.0005, "loss": 2.8712780475616455, "step": 4596 }, { "epoch": 0.5349080754014429, "grad_norm": 1.0435951948165894, "learning_rate": 0.0005, "loss": 2.4244384765625, "step": 4597 }, { "epoch": 0.535024435652781, "grad_norm": 1.3469160795211792, "learning_rate": 0.0005, "loss": 1.67329740524292, "step": 4598 }, { "epoch": 0.5351407959041191, "grad_norm": 3.215723991394043, "learning_rate": 0.0005, "loss": 2.679391622543335, "step": 4599 }, { "epoch": 0.5352571561554573, "grad_norm": 1.6303367614746094, "learning_rate": 0.0005, "loss": 1.7883601188659668, "step": 4600 }, { "epoch": 0.5353735164067954, "grad_norm": 1.9131972789764404, "learning_rate": 0.0005, "loss": 2.3239963054656982, "step": 4601 }, { "epoch": 0.5354898766581336, "grad_norm": 1.5162878036499023, "learning_rate": 0.0005, "loss": 2.2891552448272705, "step": 4602 }, { "epoch": 0.5356062369094717, "grad_norm": 1.270595669746399, "learning_rate": 0.0005, "loss": 3.5679686069488525, "step": 4603 }, { "epoch": 0.5357225971608098, "grad_norm": 1.6006745100021362, "learning_rate": 0.0005, "loss": 3.4120097160339355, "step": 4604 }, { "epoch": 0.535838957412148, "grad_norm": 0.91862952709198, "learning_rate": 0.0005, "loss": 1.7184529304504395, "step": 4605 }, { "epoch": 0.5359553176634861, "grad_norm": 1.49038565158844, "learning_rate": 0.0005, "loss": 3.0559768676757812, "step": 4606 }, { "epoch": 0.5360716779148242, "grad_norm": 1.2980023622512817, "learning_rate": 0.0005, "loss": 2.1730499267578125, "step": 4607 }, { "epoch": 0.5361880381661625, "grad_norm": 2.0537686347961426, "learning_rate": 0.0005, "loss": 2.8481621742248535, "step": 4608 }, { "epoch": 0.5363043984175005, "grad_norm": 1.5333473682403564, "learning_rate": 0.0005, "loss": 2.852675437927246, "step": 4609 }, { "epoch": 0.5364207586688388, "grad_norm": 0.924637496471405, "learning_rate": 0.0005, "loss": 2.023625135421753, "step": 4610 }, { "epoch": 0.5365371189201769, "grad_norm": 1.6962664127349854, "learning_rate": 0.0005, "loss": 2.608797550201416, "step": 4611 }, { "epoch": 0.536653479171515, "grad_norm": 1.4114402532577515, "learning_rate": 0.0005, "loss": 3.2159886360168457, "step": 4612 }, { "epoch": 0.5367698394228532, "grad_norm": 0.999940037727356, "learning_rate": 0.0005, "loss": 2.3816275596618652, "step": 4613 }, { "epoch": 0.5368861996741913, "grad_norm": 1.6455925703048706, "learning_rate": 0.0005, "loss": 2.6580445766448975, "step": 4614 }, { "epoch": 0.5370025599255295, "grad_norm": 1.4309312105178833, "learning_rate": 0.0005, "loss": 3.353527545928955, "step": 4615 }, { "epoch": 0.5371189201768676, "grad_norm": 1.2949343919754028, "learning_rate": 0.0005, "loss": 2.390223264694214, "step": 4616 }, { "epoch": 0.5372352804282057, "grad_norm": 1.7519936561584473, "learning_rate": 0.0005, "loss": 2.4301459789276123, "step": 4617 }, { "epoch": 0.5373516406795439, "grad_norm": 2.2293901443481445, "learning_rate": 0.0005, "loss": 3.6291141510009766, "step": 4618 }, { "epoch": 0.537468000930882, "grad_norm": 1.555812120437622, "learning_rate": 0.0005, "loss": 2.8628406524658203, "step": 4619 }, { "epoch": 0.5375843611822202, "grad_norm": 1.460936188697815, "learning_rate": 0.0005, "loss": 2.0162105560302734, "step": 4620 }, { "epoch": 0.5377007214335583, "grad_norm": 1.6351145505905151, "learning_rate": 0.0005, "loss": 3.2465202808380127, "step": 4621 }, { "epoch": 0.5378170816848964, "grad_norm": 1.792602300643921, "learning_rate": 0.0005, "loss": 2.858830451965332, "step": 4622 }, { "epoch": 0.5379334419362346, "grad_norm": 2.7402641773223877, "learning_rate": 0.0005, "loss": 2.9975008964538574, "step": 4623 }, { "epoch": 0.5380498021875727, "grad_norm": 1.0141304731369019, "learning_rate": 0.0005, "loss": 2.0667388439178467, "step": 4624 }, { "epoch": 0.5381661624389109, "grad_norm": 1.2372537851333618, "learning_rate": 0.0005, "loss": 2.976145029067993, "step": 4625 }, { "epoch": 0.538282522690249, "grad_norm": 1.4099297523498535, "learning_rate": 0.0005, "loss": 2.557432174682617, "step": 4626 }, { "epoch": 0.5383988829415871, "grad_norm": 1.4304903745651245, "learning_rate": 0.0005, "loss": 2.333962917327881, "step": 4627 }, { "epoch": 0.5385152431929253, "grad_norm": 1.3492478132247925, "learning_rate": 0.0005, "loss": 2.3431882858276367, "step": 4628 }, { "epoch": 0.5386316034442634, "grad_norm": 1.581080675125122, "learning_rate": 0.0005, "loss": 3.2398934364318848, "step": 4629 }, { "epoch": 0.5387479636956016, "grad_norm": 1.9075977802276611, "learning_rate": 0.0005, "loss": 3.2915797233581543, "step": 4630 }, { "epoch": 0.5388643239469397, "grad_norm": 1.0173527002334595, "learning_rate": 0.0005, "loss": 2.3508970737457275, "step": 4631 }, { "epoch": 0.5389806841982778, "grad_norm": 2.7357070446014404, "learning_rate": 0.0005, "loss": 1.5808745622634888, "step": 4632 }, { "epoch": 0.539097044449616, "grad_norm": 1.130028247833252, "learning_rate": 0.0005, "loss": 2.484889030456543, "step": 4633 }, { "epoch": 0.5392134047009541, "grad_norm": 1.0820249319076538, "learning_rate": 0.0005, "loss": 2.712693214416504, "step": 4634 }, { "epoch": 0.5393297649522923, "grad_norm": 1.5070823431015015, "learning_rate": 0.0005, "loss": 3.1221697330474854, "step": 4635 }, { "epoch": 0.5394461252036304, "grad_norm": 1.940471887588501, "learning_rate": 0.0005, "loss": 2.190335273742676, "step": 4636 }, { "epoch": 0.5395624854549685, "grad_norm": 0.9617466926574707, "learning_rate": 0.0005, "loss": 1.8163225650787354, "step": 4637 }, { "epoch": 0.5396788457063068, "grad_norm": 1.0776264667510986, "learning_rate": 0.0005, "loss": 2.11458683013916, "step": 4638 }, { "epoch": 0.5397952059576449, "grad_norm": 1.0368319749832153, "learning_rate": 0.0005, "loss": 2.094566583633423, "step": 4639 }, { "epoch": 0.5399115662089831, "grad_norm": 1.0699291229248047, "learning_rate": 0.0005, "loss": 2.419229507446289, "step": 4640 }, { "epoch": 0.5400279264603212, "grad_norm": 1.756104588508606, "learning_rate": 0.0005, "loss": 3.142711639404297, "step": 4641 }, { "epoch": 0.5401442867116593, "grad_norm": 1.6149893999099731, "learning_rate": 0.0005, "loss": 2.005899667739868, "step": 4642 }, { "epoch": 0.5402606469629975, "grad_norm": 1.9131115674972534, "learning_rate": 0.0005, "loss": 2.3511605262756348, "step": 4643 }, { "epoch": 0.5403770072143356, "grad_norm": 1.5622345209121704, "learning_rate": 0.0005, "loss": 2.8124732971191406, "step": 4644 }, { "epoch": 0.5404933674656738, "grad_norm": 1.7990520000457764, "learning_rate": 0.0005, "loss": 3.692871332168579, "step": 4645 }, { "epoch": 0.5406097277170119, "grad_norm": 1.8872359991073608, "learning_rate": 0.0005, "loss": 2.194669485092163, "step": 4646 }, { "epoch": 0.54072608796835, "grad_norm": 4.890352249145508, "learning_rate": 0.0005, "loss": 3.681067705154419, "step": 4647 }, { "epoch": 0.5408424482196882, "grad_norm": 1.393047571182251, "learning_rate": 0.0005, "loss": 2.541379928588867, "step": 4648 }, { "epoch": 0.5409588084710263, "grad_norm": 1.631504774093628, "learning_rate": 0.0005, "loss": 2.1481456756591797, "step": 4649 }, { "epoch": 0.5410751687223644, "grad_norm": 1.2616146802902222, "learning_rate": 0.0005, "loss": 3.3353090286254883, "step": 4650 }, { "epoch": 0.5411915289737026, "grad_norm": 1.083571434020996, "learning_rate": 0.0005, "loss": 1.9328850507736206, "step": 4651 }, { "epoch": 0.5413078892250407, "grad_norm": 1.68014657497406, "learning_rate": 0.0005, "loss": 2.36240816116333, "step": 4652 }, { "epoch": 0.5414242494763789, "grad_norm": 1.2621294260025024, "learning_rate": 0.0005, "loss": 3.229987621307373, "step": 4653 }, { "epoch": 0.541540609727717, "grad_norm": 2.3906185626983643, "learning_rate": 0.0005, "loss": 2.8445329666137695, "step": 4654 }, { "epoch": 0.5416569699790551, "grad_norm": 1.2647498846054077, "learning_rate": 0.0005, "loss": 2.9626264572143555, "step": 4655 }, { "epoch": 0.5417733302303933, "grad_norm": 1.6735234260559082, "learning_rate": 0.0005, "loss": 2.626636505126953, "step": 4656 }, { "epoch": 0.5418896904817314, "grad_norm": 1.445993185043335, "learning_rate": 0.0005, "loss": 1.6347485780715942, "step": 4657 }, { "epoch": 0.5420060507330696, "grad_norm": 1.2595140933990479, "learning_rate": 0.0005, "loss": 3.027644634246826, "step": 4658 }, { "epoch": 0.5421224109844077, "grad_norm": 0.688371479511261, "learning_rate": 0.0005, "loss": 1.6313705444335938, "step": 4659 }, { "epoch": 0.5422387712357458, "grad_norm": 1.8262202739715576, "learning_rate": 0.0005, "loss": 2.2589995861053467, "step": 4660 }, { "epoch": 0.542355131487084, "grad_norm": 1.683814287185669, "learning_rate": 0.0005, "loss": 2.5642566680908203, "step": 4661 }, { "epoch": 0.5424714917384221, "grad_norm": 1.284733533859253, "learning_rate": 0.0005, "loss": 1.9217677116394043, "step": 4662 }, { "epoch": 0.5425878519897603, "grad_norm": 1.6482062339782715, "learning_rate": 0.0005, "loss": 1.9932758808135986, "step": 4663 }, { "epoch": 0.5427042122410984, "grad_norm": 1.7953264713287354, "learning_rate": 0.0005, "loss": 2.861314296722412, "step": 4664 }, { "epoch": 0.5428205724924365, "grad_norm": 1.4344340562820435, "learning_rate": 0.0005, "loss": 3.657987117767334, "step": 4665 }, { "epoch": 0.5429369327437747, "grad_norm": 1.02976393699646, "learning_rate": 0.0005, "loss": 1.7762664556503296, "step": 4666 }, { "epoch": 0.5430532929951128, "grad_norm": 1.1423542499542236, "learning_rate": 0.0005, "loss": 1.851978063583374, "step": 4667 }, { "epoch": 0.5431696532464511, "grad_norm": 1.3261969089508057, "learning_rate": 0.0005, "loss": 2.845630168914795, "step": 4668 }, { "epoch": 0.5432860134977892, "grad_norm": 1.270348072052002, "learning_rate": 0.0005, "loss": 2.5462236404418945, "step": 4669 }, { "epoch": 0.5434023737491273, "grad_norm": 1.1213932037353516, "learning_rate": 0.0005, "loss": 1.702886700630188, "step": 4670 }, { "epoch": 0.5435187340004655, "grad_norm": 1.6043760776519775, "learning_rate": 0.0005, "loss": 2.8238487243652344, "step": 4671 }, { "epoch": 0.5436350942518036, "grad_norm": 1.8992674350738525, "learning_rate": 0.0005, "loss": 3.3418145179748535, "step": 4672 }, { "epoch": 0.5437514545031418, "grad_norm": 1.9586501121520996, "learning_rate": 0.0005, "loss": 3.0320310592651367, "step": 4673 }, { "epoch": 0.5438678147544799, "grad_norm": 1.1880695819854736, "learning_rate": 0.0005, "loss": 2.3786282539367676, "step": 4674 }, { "epoch": 0.543984175005818, "grad_norm": 2.8157145977020264, "learning_rate": 0.0005, "loss": 2.5104801654815674, "step": 4675 }, { "epoch": 0.5441005352571562, "grad_norm": 1.4394649267196655, "learning_rate": 0.0005, "loss": 2.277287483215332, "step": 4676 }, { "epoch": 0.5442168955084943, "grad_norm": 1.236939787864685, "learning_rate": 0.0005, "loss": 2.6455719470977783, "step": 4677 }, { "epoch": 0.5443332557598325, "grad_norm": 1.7619330883026123, "learning_rate": 0.0005, "loss": 2.512152671813965, "step": 4678 }, { "epoch": 0.5444496160111706, "grad_norm": 0.9359850287437439, "learning_rate": 0.0005, "loss": 1.782721996307373, "step": 4679 }, { "epoch": 0.5445659762625087, "grad_norm": 1.7370190620422363, "learning_rate": 0.0005, "loss": 2.93643856048584, "step": 4680 }, { "epoch": 0.5446823365138469, "grad_norm": 1.6302342414855957, "learning_rate": 0.0005, "loss": 1.9980716705322266, "step": 4681 }, { "epoch": 0.544798696765185, "grad_norm": 1.6747212409973145, "learning_rate": 0.0005, "loss": 2.728402614593506, "step": 4682 }, { "epoch": 0.5449150570165232, "grad_norm": 1.3561816215515137, "learning_rate": 0.0005, "loss": 2.4419984817504883, "step": 4683 }, { "epoch": 0.5450314172678613, "grad_norm": 1.2224955558776855, "learning_rate": 0.0005, "loss": 1.9796901941299438, "step": 4684 }, { "epoch": 0.5451477775191994, "grad_norm": 1.331196665763855, "learning_rate": 0.0005, "loss": 1.6213021278381348, "step": 4685 }, { "epoch": 0.5452641377705376, "grad_norm": 1.2906631231307983, "learning_rate": 0.0005, "loss": 2.3946566581726074, "step": 4686 }, { "epoch": 0.5453804980218757, "grad_norm": 1.9213730096817017, "learning_rate": 0.0005, "loss": 2.0792288780212402, "step": 4687 }, { "epoch": 0.5454968582732139, "grad_norm": 1.6613374948501587, "learning_rate": 0.0005, "loss": 3.126863718032837, "step": 4688 }, { "epoch": 0.545613218524552, "grad_norm": 1.05916428565979, "learning_rate": 0.0005, "loss": 2.585440158843994, "step": 4689 }, { "epoch": 0.5457295787758901, "grad_norm": 1.181397557258606, "learning_rate": 0.0005, "loss": 2.545321464538574, "step": 4690 }, { "epoch": 0.5458459390272283, "grad_norm": 2.2501437664031982, "learning_rate": 0.0005, "loss": 3.3803412914276123, "step": 4691 }, { "epoch": 0.5459622992785664, "grad_norm": 0.8494386076927185, "learning_rate": 0.0005, "loss": 2.5630884170532227, "step": 4692 }, { "epoch": 0.5460786595299045, "grad_norm": 1.351651668548584, "learning_rate": 0.0005, "loss": 3.009610176086426, "step": 4693 }, { "epoch": 0.5461950197812427, "grad_norm": 0.8625035881996155, "learning_rate": 0.0005, "loss": 1.3228225708007812, "step": 4694 }, { "epoch": 0.5463113800325808, "grad_norm": 0.9746277332305908, "learning_rate": 0.0005, "loss": 2.5936198234558105, "step": 4695 }, { "epoch": 0.546427740283919, "grad_norm": 2.5963075160980225, "learning_rate": 0.0005, "loss": 2.9537041187286377, "step": 4696 }, { "epoch": 0.5465441005352571, "grad_norm": 2.3002631664276123, "learning_rate": 0.0005, "loss": 2.435398817062378, "step": 4697 }, { "epoch": 0.5466604607865952, "grad_norm": 1.6570161581039429, "learning_rate": 0.0005, "loss": 2.269314765930176, "step": 4698 }, { "epoch": 0.5467768210379335, "grad_norm": 1.1198991537094116, "learning_rate": 0.0005, "loss": 2.4513890743255615, "step": 4699 }, { "epoch": 0.5468931812892716, "grad_norm": 1.3546128273010254, "learning_rate": 0.0005, "loss": 3.144097328186035, "step": 4700 }, { "epoch": 0.5470095415406098, "grad_norm": 1.1722497940063477, "learning_rate": 0.0005, "loss": 3.0885417461395264, "step": 4701 }, { "epoch": 0.5471259017919479, "grad_norm": 1.4159928560256958, "learning_rate": 0.0005, "loss": 1.9426586627960205, "step": 4702 }, { "epoch": 0.547242262043286, "grad_norm": 1.3249820470809937, "learning_rate": 0.0005, "loss": 1.8401508331298828, "step": 4703 }, { "epoch": 0.5473586222946242, "grad_norm": 1.380046010017395, "learning_rate": 0.0005, "loss": 2.1822586059570312, "step": 4704 }, { "epoch": 0.5474749825459623, "grad_norm": 1.1642084121704102, "learning_rate": 0.0005, "loss": 2.5953116416931152, "step": 4705 }, { "epoch": 0.5475913427973005, "grad_norm": 1.396435260772705, "learning_rate": 0.0005, "loss": 2.645169496536255, "step": 4706 }, { "epoch": 0.5477077030486386, "grad_norm": 1.058157205581665, "learning_rate": 0.0005, "loss": 2.621427536010742, "step": 4707 }, { "epoch": 0.5478240632999767, "grad_norm": 0.901872992515564, "learning_rate": 0.0005, "loss": 1.2660861015319824, "step": 4708 }, { "epoch": 0.5479404235513149, "grad_norm": 1.5672606229782104, "learning_rate": 0.0005, "loss": 2.338560104370117, "step": 4709 }, { "epoch": 0.548056783802653, "grad_norm": 1.3621840476989746, "learning_rate": 0.0005, "loss": 2.1648333072662354, "step": 4710 }, { "epoch": 0.5481731440539912, "grad_norm": 0.8467158079147339, "learning_rate": 0.0005, "loss": 1.6237750053405762, "step": 4711 }, { "epoch": 0.5482895043053293, "grad_norm": 0.8805700540542603, "learning_rate": 0.0005, "loss": 1.7264587879180908, "step": 4712 }, { "epoch": 0.5484058645566674, "grad_norm": 1.6683118343353271, "learning_rate": 0.0005, "loss": 2.26058292388916, "step": 4713 }, { "epoch": 0.5485222248080056, "grad_norm": 1.3111170530319214, "learning_rate": 0.0005, "loss": 2.0000452995300293, "step": 4714 }, { "epoch": 0.5486385850593437, "grad_norm": 1.6845202445983887, "learning_rate": 0.0005, "loss": 2.3729915618896484, "step": 4715 }, { "epoch": 0.5487549453106819, "grad_norm": 0.9376471638679504, "learning_rate": 0.0005, "loss": 1.7146551609039307, "step": 4716 }, { "epoch": 0.54887130556202, "grad_norm": 1.9743807315826416, "learning_rate": 0.0005, "loss": 3.493961811065674, "step": 4717 }, { "epoch": 0.5489876658133581, "grad_norm": 1.48805570602417, "learning_rate": 0.0005, "loss": 2.8152904510498047, "step": 4718 }, { "epoch": 0.5491040260646963, "grad_norm": 1.2669204473495483, "learning_rate": 0.0005, "loss": 2.055957317352295, "step": 4719 }, { "epoch": 0.5492203863160344, "grad_norm": 2.2073819637298584, "learning_rate": 0.0005, "loss": 3.486467123031616, "step": 4720 }, { "epoch": 0.5493367465673726, "grad_norm": 1.8074837923049927, "learning_rate": 0.0005, "loss": 2.4345240592956543, "step": 4721 }, { "epoch": 0.5494531068187107, "grad_norm": 1.420491099357605, "learning_rate": 0.0005, "loss": 2.7943766117095947, "step": 4722 }, { "epoch": 0.5495694670700488, "grad_norm": 1.3839600086212158, "learning_rate": 0.0005, "loss": 2.6646738052368164, "step": 4723 }, { "epoch": 0.549685827321387, "grad_norm": 1.3106794357299805, "learning_rate": 0.0005, "loss": 3.2556798458099365, "step": 4724 }, { "epoch": 0.5498021875727251, "grad_norm": 1.5425095558166504, "learning_rate": 0.0005, "loss": 3.027130126953125, "step": 4725 }, { "epoch": 0.5499185478240634, "grad_norm": 2.5946764945983887, "learning_rate": 0.0005, "loss": 2.911595582962036, "step": 4726 }, { "epoch": 0.5500349080754015, "grad_norm": 1.1880532503128052, "learning_rate": 0.0005, "loss": 1.4558064937591553, "step": 4727 }, { "epoch": 0.5501512683267396, "grad_norm": 0.868094801902771, "learning_rate": 0.0005, "loss": 2.3843774795532227, "step": 4728 }, { "epoch": 0.5502676285780778, "grad_norm": 3.140235424041748, "learning_rate": 0.0005, "loss": 4.0257673263549805, "step": 4729 }, { "epoch": 0.5503839888294159, "grad_norm": 1.0731799602508545, "learning_rate": 0.0005, "loss": 2.2500545978546143, "step": 4730 }, { "epoch": 0.550500349080754, "grad_norm": 1.1007195711135864, "learning_rate": 0.0005, "loss": 1.6505789756774902, "step": 4731 }, { "epoch": 0.5506167093320922, "grad_norm": 1.7402747869491577, "learning_rate": 0.0005, "loss": 2.507660388946533, "step": 4732 }, { "epoch": 0.5507330695834303, "grad_norm": 2.005901575088501, "learning_rate": 0.0005, "loss": 2.434999942779541, "step": 4733 }, { "epoch": 0.5508494298347685, "grad_norm": 2.400378465652466, "learning_rate": 0.0005, "loss": 3.0730128288269043, "step": 4734 }, { "epoch": 0.5509657900861066, "grad_norm": 1.6545559167861938, "learning_rate": 0.0005, "loss": 1.5234594345092773, "step": 4735 }, { "epoch": 0.5510821503374447, "grad_norm": 1.1446928977966309, "learning_rate": 0.0005, "loss": 1.8829759359359741, "step": 4736 }, { "epoch": 0.5511985105887829, "grad_norm": 1.7438087463378906, "learning_rate": 0.0005, "loss": 2.1179943084716797, "step": 4737 }, { "epoch": 0.551314870840121, "grad_norm": 0.904407799243927, "learning_rate": 0.0005, "loss": 1.7796661853790283, "step": 4738 }, { "epoch": 0.5514312310914592, "grad_norm": 1.710686206817627, "learning_rate": 0.0005, "loss": 3.826533317565918, "step": 4739 }, { "epoch": 0.5515475913427973, "grad_norm": 1.2056918144226074, "learning_rate": 0.0005, "loss": 2.053255081176758, "step": 4740 }, { "epoch": 0.5516639515941354, "grad_norm": 2.975501298904419, "learning_rate": 0.0005, "loss": 4.8430328369140625, "step": 4741 }, { "epoch": 0.5517803118454736, "grad_norm": 1.3058350086212158, "learning_rate": 0.0005, "loss": 1.9685405492782593, "step": 4742 }, { "epoch": 0.5518966720968117, "grad_norm": 0.8434354066848755, "learning_rate": 0.0005, "loss": 1.2466609477996826, "step": 4743 }, { "epoch": 0.5520130323481499, "grad_norm": 1.6686536073684692, "learning_rate": 0.0005, "loss": 2.573115587234497, "step": 4744 }, { "epoch": 0.552129392599488, "grad_norm": 1.5462876558303833, "learning_rate": 0.0005, "loss": 2.6664085388183594, "step": 4745 }, { "epoch": 0.5522457528508261, "grad_norm": 1.7295880317687988, "learning_rate": 0.0005, "loss": 2.363430976867676, "step": 4746 }, { "epoch": 0.5523621131021643, "grad_norm": 0.9030128121376038, "learning_rate": 0.0005, "loss": 1.6661432981491089, "step": 4747 }, { "epoch": 0.5524784733535024, "grad_norm": 1.5508487224578857, "learning_rate": 0.0005, "loss": 3.5411055088043213, "step": 4748 }, { "epoch": 0.5525948336048406, "grad_norm": 1.83445143699646, "learning_rate": 0.0005, "loss": 2.6502532958984375, "step": 4749 }, { "epoch": 0.5527111938561787, "grad_norm": 1.1280885934829712, "learning_rate": 0.0005, "loss": 2.2338223457336426, "step": 4750 }, { "epoch": 0.5528275541075168, "grad_norm": 1.7896637916564941, "learning_rate": 0.0005, "loss": 2.8091773986816406, "step": 4751 }, { "epoch": 0.552943914358855, "grad_norm": 1.3935750722885132, "learning_rate": 0.0005, "loss": 2.4620981216430664, "step": 4752 }, { "epoch": 0.5530602746101931, "grad_norm": 1.1643747091293335, "learning_rate": 0.0005, "loss": 2.073348045349121, "step": 4753 }, { "epoch": 0.5531766348615313, "grad_norm": 0.7583476305007935, "learning_rate": 0.0005, "loss": 2.1265296936035156, "step": 4754 }, { "epoch": 0.5532929951128694, "grad_norm": 2.8004050254821777, "learning_rate": 0.0005, "loss": 3.2758333683013916, "step": 4755 }, { "epoch": 0.5534093553642075, "grad_norm": 2.051466703414917, "learning_rate": 0.0005, "loss": 3.3042030334472656, "step": 4756 }, { "epoch": 0.5535257156155458, "grad_norm": 1.2096549272537231, "learning_rate": 0.0005, "loss": 2.9157965183258057, "step": 4757 }, { "epoch": 0.5536420758668839, "grad_norm": 0.9563805460929871, "learning_rate": 0.0005, "loss": 2.612178325653076, "step": 4758 }, { "epoch": 0.5537584361182221, "grad_norm": 1.4326714277267456, "learning_rate": 0.0005, "loss": 3.0210866928100586, "step": 4759 }, { "epoch": 0.5538747963695602, "grad_norm": 1.9430723190307617, "learning_rate": 0.0005, "loss": 2.6131577491760254, "step": 4760 }, { "epoch": 0.5539911566208983, "grad_norm": 1.5519375801086426, "learning_rate": 0.0005, "loss": 2.996753215789795, "step": 4761 }, { "epoch": 0.5541075168722365, "grad_norm": 1.0838764905929565, "learning_rate": 0.0005, "loss": 1.9147770404815674, "step": 4762 }, { "epoch": 0.5542238771235746, "grad_norm": 1.1335053443908691, "learning_rate": 0.0005, "loss": 2.2963552474975586, "step": 4763 }, { "epoch": 0.5543402373749128, "grad_norm": 1.5778361558914185, "learning_rate": 0.0005, "loss": 3.0778274536132812, "step": 4764 }, { "epoch": 0.5544565976262509, "grad_norm": 1.5437172651290894, "learning_rate": 0.0005, "loss": 2.5487587451934814, "step": 4765 }, { "epoch": 0.554572957877589, "grad_norm": 1.2653053998947144, "learning_rate": 0.0005, "loss": 1.9482501745224, "step": 4766 }, { "epoch": 0.5546893181289272, "grad_norm": 1.341903805732727, "learning_rate": 0.0005, "loss": 2.2100234031677246, "step": 4767 }, { "epoch": 0.5548056783802653, "grad_norm": 1.3230767250061035, "learning_rate": 0.0005, "loss": 2.968607187271118, "step": 4768 }, { "epoch": 0.5549220386316035, "grad_norm": 1.6338937282562256, "learning_rate": 0.0005, "loss": 2.868431568145752, "step": 4769 }, { "epoch": 0.5550383988829416, "grad_norm": 0.8688406348228455, "learning_rate": 0.0005, "loss": 1.552404761314392, "step": 4770 }, { "epoch": 0.5551547591342797, "grad_norm": 1.0598176717758179, "learning_rate": 0.0005, "loss": 2.288104295730591, "step": 4771 }, { "epoch": 0.5552711193856179, "grad_norm": 1.0976827144622803, "learning_rate": 0.0005, "loss": 2.1741414070129395, "step": 4772 }, { "epoch": 0.555387479636956, "grad_norm": 1.4431453943252563, "learning_rate": 0.0005, "loss": 2.7720632553100586, "step": 4773 }, { "epoch": 0.5555038398882941, "grad_norm": 1.807641625404358, "learning_rate": 0.0005, "loss": 1.8374052047729492, "step": 4774 }, { "epoch": 0.5556202001396323, "grad_norm": 1.5405501127243042, "learning_rate": 0.0005, "loss": 2.142531633377075, "step": 4775 }, { "epoch": 0.5557365603909704, "grad_norm": 1.305023431777954, "learning_rate": 0.0005, "loss": 2.3962862491607666, "step": 4776 }, { "epoch": 0.5558529206423086, "grad_norm": 2.584116220474243, "learning_rate": 0.0005, "loss": 2.853222131729126, "step": 4777 }, { "epoch": 0.5559692808936467, "grad_norm": 1.9922573566436768, "learning_rate": 0.0005, "loss": 2.751317024230957, "step": 4778 }, { "epoch": 0.5560856411449848, "grad_norm": 0.7171551585197449, "learning_rate": 0.0005, "loss": 1.548488974571228, "step": 4779 }, { "epoch": 0.556202001396323, "grad_norm": 1.2904433012008667, "learning_rate": 0.0005, "loss": 2.9951910972595215, "step": 4780 }, { "epoch": 0.5563183616476611, "grad_norm": 1.4146969318389893, "learning_rate": 0.0005, "loss": 2.4372427463531494, "step": 4781 }, { "epoch": 0.5564347218989993, "grad_norm": 1.158100962638855, "learning_rate": 0.0005, "loss": 2.5857746601104736, "step": 4782 }, { "epoch": 0.5565510821503374, "grad_norm": 1.7133333683013916, "learning_rate": 0.0005, "loss": 2.7365529537200928, "step": 4783 }, { "epoch": 0.5566674424016755, "grad_norm": 1.8376455307006836, "learning_rate": 0.0005, "loss": 1.818216323852539, "step": 4784 }, { "epoch": 0.5567838026530137, "grad_norm": 1.743504285812378, "learning_rate": 0.0005, "loss": 2.4689369201660156, "step": 4785 }, { "epoch": 0.5569001629043518, "grad_norm": 0.8973091840744019, "learning_rate": 0.0005, "loss": 1.667027235031128, "step": 4786 }, { "epoch": 0.5570165231556901, "grad_norm": 1.0718705654144287, "learning_rate": 0.0005, "loss": 1.35703444480896, "step": 4787 }, { "epoch": 0.5571328834070282, "grad_norm": 1.5608980655670166, "learning_rate": 0.0005, "loss": 3.3856520652770996, "step": 4788 }, { "epoch": 0.5572492436583663, "grad_norm": 2.498779296875, "learning_rate": 0.0005, "loss": 2.7842164039611816, "step": 4789 }, { "epoch": 0.5573656039097045, "grad_norm": 1.6602811813354492, "learning_rate": 0.0005, "loss": 2.549142360687256, "step": 4790 }, { "epoch": 0.5574819641610426, "grad_norm": 1.8900034427642822, "learning_rate": 0.0005, "loss": 2.569545269012451, "step": 4791 }, { "epoch": 0.5575983244123808, "grad_norm": 1.3680917024612427, "learning_rate": 0.0005, "loss": 2.670279026031494, "step": 4792 }, { "epoch": 0.5577146846637189, "grad_norm": 1.1918189525604248, "learning_rate": 0.0005, "loss": 1.6972808837890625, "step": 4793 }, { "epoch": 0.557831044915057, "grad_norm": 1.2624326944351196, "learning_rate": 0.0005, "loss": 2.5146408081054688, "step": 4794 }, { "epoch": 0.5579474051663952, "grad_norm": 1.9032295942306519, "learning_rate": 0.0005, "loss": 2.9333765506744385, "step": 4795 }, { "epoch": 0.5580637654177333, "grad_norm": 0.9831367135047913, "learning_rate": 0.0005, "loss": 2.184671401977539, "step": 4796 }, { "epoch": 0.5581801256690715, "grad_norm": 0.8516638278961182, "learning_rate": 0.0005, "loss": 1.9091598987579346, "step": 4797 }, { "epoch": 0.5582964859204096, "grad_norm": 3.7051658630371094, "learning_rate": 0.0005, "loss": 3.0258548259735107, "step": 4798 }, { "epoch": 0.5584128461717477, "grad_norm": 1.8111814260482788, "learning_rate": 0.0005, "loss": 3.443185329437256, "step": 4799 }, { "epoch": 0.5585292064230859, "grad_norm": 1.2352865934371948, "learning_rate": 0.0005, "loss": 2.398383617401123, "step": 4800 }, { "epoch": 0.558645566674424, "grad_norm": 1.3222638368606567, "learning_rate": 0.0005, "loss": 2.136911392211914, "step": 4801 }, { "epoch": 0.5587619269257622, "grad_norm": 1.9073820114135742, "learning_rate": 0.0005, "loss": 3.01823091506958, "step": 4802 }, { "epoch": 0.5588782871771003, "grad_norm": 1.7783491611480713, "learning_rate": 0.0005, "loss": 2.9618537425994873, "step": 4803 }, { "epoch": 0.5589946474284384, "grad_norm": 1.866321086883545, "learning_rate": 0.0005, "loss": 2.1261000633239746, "step": 4804 }, { "epoch": 0.5591110076797766, "grad_norm": 1.0632203817367554, "learning_rate": 0.0005, "loss": 2.2777841091156006, "step": 4805 }, { "epoch": 0.5592273679311147, "grad_norm": 1.6169179677963257, "learning_rate": 0.0005, "loss": 2.502795696258545, "step": 4806 }, { "epoch": 0.5593437281824529, "grad_norm": 1.3568980693817139, "learning_rate": 0.0005, "loss": 2.7209830284118652, "step": 4807 }, { "epoch": 0.559460088433791, "grad_norm": 1.2002520561218262, "learning_rate": 0.0005, "loss": 2.538886308670044, "step": 4808 }, { "epoch": 0.5595764486851291, "grad_norm": 1.4999717473983765, "learning_rate": 0.0005, "loss": 2.3492908477783203, "step": 4809 }, { "epoch": 0.5596928089364673, "grad_norm": 2.8492257595062256, "learning_rate": 0.0005, "loss": 3.313591957092285, "step": 4810 }, { "epoch": 0.5598091691878054, "grad_norm": 0.9924427270889282, "learning_rate": 0.0005, "loss": 2.753622055053711, "step": 4811 }, { "epoch": 0.5599255294391436, "grad_norm": 0.911246120929718, "learning_rate": 0.0005, "loss": 1.6199729442596436, "step": 4812 }, { "epoch": 0.5600418896904817, "grad_norm": 2.457887887954712, "learning_rate": 0.0005, "loss": 2.330669403076172, "step": 4813 }, { "epoch": 0.5601582499418198, "grad_norm": 1.1893154382705688, "learning_rate": 0.0005, "loss": 1.6117483377456665, "step": 4814 }, { "epoch": 0.560274610193158, "grad_norm": 1.2545132637023926, "learning_rate": 0.0005, "loss": 2.2704737186431885, "step": 4815 }, { "epoch": 0.5603909704444962, "grad_norm": 1.8501702547073364, "learning_rate": 0.0005, "loss": 2.4062459468841553, "step": 4816 }, { "epoch": 0.5605073306958342, "grad_norm": 0.9066752791404724, "learning_rate": 0.0005, "loss": 1.4696524143218994, "step": 4817 }, { "epoch": 0.5606236909471725, "grad_norm": 1.6003347635269165, "learning_rate": 0.0005, "loss": 2.836158037185669, "step": 4818 }, { "epoch": 0.5607400511985106, "grad_norm": 1.2220618724822998, "learning_rate": 0.0005, "loss": 2.1352033615112305, "step": 4819 }, { "epoch": 0.5608564114498488, "grad_norm": 2.798163890838623, "learning_rate": 0.0005, "loss": 2.5816378593444824, "step": 4820 }, { "epoch": 0.5609727717011869, "grad_norm": 1.6927250623703003, "learning_rate": 0.0005, "loss": 3.089127779006958, "step": 4821 }, { "epoch": 0.561089131952525, "grad_norm": 1.9472298622131348, "learning_rate": 0.0005, "loss": 2.9854471683502197, "step": 4822 }, { "epoch": 0.5612054922038632, "grad_norm": 1.7768803834915161, "learning_rate": 0.0005, "loss": 3.1430234909057617, "step": 4823 }, { "epoch": 0.5613218524552013, "grad_norm": 1.3609347343444824, "learning_rate": 0.0005, "loss": 2.1992597579956055, "step": 4824 }, { "epoch": 0.5614382127065395, "grad_norm": 0.877399206161499, "learning_rate": 0.0005, "loss": 1.6426095962524414, "step": 4825 }, { "epoch": 0.5615545729578776, "grad_norm": 1.9929436445236206, "learning_rate": 0.0005, "loss": 2.499567985534668, "step": 4826 }, { "epoch": 0.5616709332092157, "grad_norm": 1.9371901750564575, "learning_rate": 0.0005, "loss": 2.897019386291504, "step": 4827 }, { "epoch": 0.5617872934605539, "grad_norm": 1.0947823524475098, "learning_rate": 0.0005, "loss": 2.4774575233459473, "step": 4828 }, { "epoch": 0.561903653711892, "grad_norm": 1.2317867279052734, "learning_rate": 0.0005, "loss": 2.575922966003418, "step": 4829 }, { "epoch": 0.5620200139632302, "grad_norm": 1.3366806507110596, "learning_rate": 0.0005, "loss": 2.3284544944763184, "step": 4830 }, { "epoch": 0.5621363742145683, "grad_norm": 1.4715003967285156, "learning_rate": 0.0005, "loss": 2.454683780670166, "step": 4831 }, { "epoch": 0.5622527344659064, "grad_norm": 1.5548241138458252, "learning_rate": 0.0005, "loss": 2.8873648643493652, "step": 4832 }, { "epoch": 0.5623690947172446, "grad_norm": 1.9960204362869263, "learning_rate": 0.0005, "loss": 2.2471916675567627, "step": 4833 }, { "epoch": 0.5624854549685827, "grad_norm": 1.2378345727920532, "learning_rate": 0.0005, "loss": 2.6072187423706055, "step": 4834 }, { "epoch": 0.5626018152199209, "grad_norm": 1.6438862085342407, "learning_rate": 0.0005, "loss": 2.4394073486328125, "step": 4835 }, { "epoch": 0.562718175471259, "grad_norm": 1.2881187200546265, "learning_rate": 0.0005, "loss": 2.296243667602539, "step": 4836 }, { "epoch": 0.5628345357225971, "grad_norm": 1.7862131595611572, "learning_rate": 0.0005, "loss": 3.2375435829162598, "step": 4837 }, { "epoch": 0.5629508959739353, "grad_norm": 1.6000518798828125, "learning_rate": 0.0005, "loss": 3.1118836402893066, "step": 4838 }, { "epoch": 0.5630672562252734, "grad_norm": 1.1782450675964355, "learning_rate": 0.0005, "loss": 3.148289680480957, "step": 4839 }, { "epoch": 0.5631836164766116, "grad_norm": 1.452722430229187, "learning_rate": 0.0005, "loss": 3.1147360801696777, "step": 4840 }, { "epoch": 0.5632999767279497, "grad_norm": 1.1900864839553833, "learning_rate": 0.0005, "loss": 3.1077160835266113, "step": 4841 }, { "epoch": 0.5634163369792878, "grad_norm": 1.4437841176986694, "learning_rate": 0.0005, "loss": 2.12973690032959, "step": 4842 }, { "epoch": 0.563532697230626, "grad_norm": 1.2658696174621582, "learning_rate": 0.0005, "loss": 2.314436912536621, "step": 4843 }, { "epoch": 0.5636490574819641, "grad_norm": 1.2216458320617676, "learning_rate": 0.0005, "loss": 2.6905808448791504, "step": 4844 }, { "epoch": 0.5637654177333024, "grad_norm": 1.4414135217666626, "learning_rate": 0.0005, "loss": 2.4663748741149902, "step": 4845 }, { "epoch": 0.5638817779846405, "grad_norm": 0.9267218708992004, "learning_rate": 0.0005, "loss": 2.212174892425537, "step": 4846 }, { "epoch": 0.5639981382359786, "grad_norm": 1.853723168373108, "learning_rate": 0.0005, "loss": 2.7852163314819336, "step": 4847 }, { "epoch": 0.5641144984873168, "grad_norm": 0.8549034595489502, "learning_rate": 0.0005, "loss": 2.3463239669799805, "step": 4848 }, { "epoch": 0.5642308587386549, "grad_norm": 1.225159764289856, "learning_rate": 0.0005, "loss": 3.079272985458374, "step": 4849 }, { "epoch": 0.5643472189899931, "grad_norm": 1.0935674905776978, "learning_rate": 0.0005, "loss": 2.37117338180542, "step": 4850 }, { "epoch": 0.5644635792413312, "grad_norm": 2.6001505851745605, "learning_rate": 0.0005, "loss": 3.3329780101776123, "step": 4851 }, { "epoch": 0.5645799394926693, "grad_norm": 2.385823965072632, "learning_rate": 0.0005, "loss": 2.5903263092041016, "step": 4852 }, { "epoch": 0.5646962997440075, "grad_norm": 2.54761004447937, "learning_rate": 0.0005, "loss": 2.7618680000305176, "step": 4853 }, { "epoch": 0.5648126599953456, "grad_norm": 1.845916986465454, "learning_rate": 0.0005, "loss": 2.4571776390075684, "step": 4854 }, { "epoch": 0.5649290202466837, "grad_norm": 1.3029682636260986, "learning_rate": 0.0005, "loss": 2.801243782043457, "step": 4855 }, { "epoch": 0.5650453804980219, "grad_norm": 1.0553041696548462, "learning_rate": 0.0005, "loss": 2.2067971229553223, "step": 4856 }, { "epoch": 0.56516174074936, "grad_norm": 1.6492561101913452, "learning_rate": 0.0005, "loss": 2.851356029510498, "step": 4857 }, { "epoch": 0.5652781010006982, "grad_norm": 1.5204108953475952, "learning_rate": 0.0005, "loss": 3.0912845134735107, "step": 4858 }, { "epoch": 0.5653944612520363, "grad_norm": 1.0123320817947388, "learning_rate": 0.0005, "loss": 2.4736552238464355, "step": 4859 }, { "epoch": 0.5655108215033744, "grad_norm": 1.8515509366989136, "learning_rate": 0.0005, "loss": 2.487217903137207, "step": 4860 }, { "epoch": 0.5656271817547126, "grad_norm": 1.1870793104171753, "learning_rate": 0.0005, "loss": 3.0944631099700928, "step": 4861 }, { "epoch": 0.5657435420060507, "grad_norm": 1.0852141380310059, "learning_rate": 0.0005, "loss": 2.012455940246582, "step": 4862 }, { "epoch": 0.5658599022573889, "grad_norm": 1.328415036201477, "learning_rate": 0.0005, "loss": 2.639265775680542, "step": 4863 }, { "epoch": 0.565976262508727, "grad_norm": 1.6393014192581177, "learning_rate": 0.0005, "loss": 3.0331315994262695, "step": 4864 }, { "epoch": 0.5660926227600651, "grad_norm": 1.3268343210220337, "learning_rate": 0.0005, "loss": 2.681870460510254, "step": 4865 }, { "epoch": 0.5662089830114033, "grad_norm": 1.2368320226669312, "learning_rate": 0.0005, "loss": 1.8708739280700684, "step": 4866 }, { "epoch": 0.5663253432627414, "grad_norm": 1.1742199659347534, "learning_rate": 0.0005, "loss": 2.418867826461792, "step": 4867 }, { "epoch": 0.5664417035140796, "grad_norm": 2.0380289554595947, "learning_rate": 0.0005, "loss": 2.299896240234375, "step": 4868 }, { "epoch": 0.5665580637654177, "grad_norm": 1.1209410429000854, "learning_rate": 0.0005, "loss": 1.8743764162063599, "step": 4869 }, { "epoch": 0.5666744240167558, "grad_norm": 0.6740163564682007, "learning_rate": 0.0005, "loss": 1.2901862859725952, "step": 4870 }, { "epoch": 0.566790784268094, "grad_norm": 1.364804983139038, "learning_rate": 0.0005, "loss": 2.8688366413116455, "step": 4871 }, { "epoch": 0.5669071445194321, "grad_norm": 1.793728232383728, "learning_rate": 0.0005, "loss": 2.2999234199523926, "step": 4872 }, { "epoch": 0.5670235047707703, "grad_norm": 1.3923252820968628, "learning_rate": 0.0005, "loss": 2.6786742210388184, "step": 4873 }, { "epoch": 0.5671398650221084, "grad_norm": 1.7407772541046143, "learning_rate": 0.0005, "loss": 3.4543776512145996, "step": 4874 }, { "epoch": 0.5672562252734465, "grad_norm": 1.7159181833267212, "learning_rate": 0.0005, "loss": 2.604829788208008, "step": 4875 }, { "epoch": 0.5673725855247848, "grad_norm": 1.6872724294662476, "learning_rate": 0.0005, "loss": 2.1674909591674805, "step": 4876 }, { "epoch": 0.5674889457761229, "grad_norm": 1.3245044946670532, "learning_rate": 0.0005, "loss": 1.8471219539642334, "step": 4877 }, { "epoch": 0.5676053060274611, "grad_norm": 2.071049213409424, "learning_rate": 0.0005, "loss": 3.2832722663879395, "step": 4878 }, { "epoch": 0.5677216662787992, "grad_norm": 1.1707481145858765, "learning_rate": 0.0005, "loss": 2.5870718955993652, "step": 4879 }, { "epoch": 0.5678380265301373, "grad_norm": 1.5294156074523926, "learning_rate": 0.0005, "loss": 1.9001858234405518, "step": 4880 }, { "epoch": 0.5679543867814755, "grad_norm": 1.3911908864974976, "learning_rate": 0.0005, "loss": 2.3428163528442383, "step": 4881 }, { "epoch": 0.5680707470328136, "grad_norm": 1.0049487352371216, "learning_rate": 0.0005, "loss": 2.3538565635681152, "step": 4882 }, { "epoch": 0.5681871072841518, "grad_norm": 3.0148041248321533, "learning_rate": 0.0005, "loss": 2.9947509765625, "step": 4883 }, { "epoch": 0.5683034675354899, "grad_norm": 2.096938371658325, "learning_rate": 0.0005, "loss": 2.9240102767944336, "step": 4884 }, { "epoch": 0.568419827786828, "grad_norm": 1.0930123329162598, "learning_rate": 0.0005, "loss": 2.1497769355773926, "step": 4885 }, { "epoch": 0.5685361880381662, "grad_norm": 1.3288073539733887, "learning_rate": 0.0005, "loss": 2.4005918502807617, "step": 4886 }, { "epoch": 0.5686525482895043, "grad_norm": 0.7585673332214355, "learning_rate": 0.0005, "loss": 2.2344202995300293, "step": 4887 }, { "epoch": 0.5687689085408425, "grad_norm": 1.0444575548171997, "learning_rate": 0.0005, "loss": 1.9566208124160767, "step": 4888 }, { "epoch": 0.5688852687921806, "grad_norm": 1.313940405845642, "learning_rate": 0.0005, "loss": 3.453479766845703, "step": 4889 }, { "epoch": 0.5690016290435187, "grad_norm": 1.0241423845291138, "learning_rate": 0.0005, "loss": 2.1326193809509277, "step": 4890 }, { "epoch": 0.5691179892948569, "grad_norm": 1.4499733448028564, "learning_rate": 0.0005, "loss": 2.3935000896453857, "step": 4891 }, { "epoch": 0.569234349546195, "grad_norm": 1.8834282159805298, "learning_rate": 0.0005, "loss": 2.1500682830810547, "step": 4892 }, { "epoch": 0.5693507097975332, "grad_norm": 1.4572968482971191, "learning_rate": 0.0005, "loss": 2.8461713790893555, "step": 4893 }, { "epoch": 0.5694670700488713, "grad_norm": 1.8554418087005615, "learning_rate": 0.0005, "loss": 2.432218074798584, "step": 4894 }, { "epoch": 0.5695834303002094, "grad_norm": 1.410446286201477, "learning_rate": 0.0005, "loss": 2.5967116355895996, "step": 4895 }, { "epoch": 0.5696997905515476, "grad_norm": 1.2011823654174805, "learning_rate": 0.0005, "loss": 2.4488987922668457, "step": 4896 }, { "epoch": 0.5698161508028857, "grad_norm": 0.5086477398872375, "learning_rate": 0.0005, "loss": 1.4287803173065186, "step": 4897 }, { "epoch": 0.5699325110542238, "grad_norm": 1.7086306810379028, "learning_rate": 0.0005, "loss": 2.1405375003814697, "step": 4898 }, { "epoch": 0.570048871305562, "grad_norm": 1.012854814529419, "learning_rate": 0.0005, "loss": 1.2217514514923096, "step": 4899 }, { "epoch": 0.5701652315569001, "grad_norm": 1.589971661567688, "learning_rate": 0.0005, "loss": 2.0847771167755127, "step": 4900 }, { "epoch": 0.5702815918082383, "grad_norm": 2.2472078800201416, "learning_rate": 0.0005, "loss": 2.2461037635803223, "step": 4901 }, { "epoch": 0.5703979520595764, "grad_norm": 1.2256159782409668, "learning_rate": 0.0005, "loss": 2.3330554962158203, "step": 4902 }, { "epoch": 0.5705143123109145, "grad_norm": 1.315436840057373, "learning_rate": 0.0005, "loss": 2.3656959533691406, "step": 4903 }, { "epoch": 0.5706306725622527, "grad_norm": 1.2906938791275024, "learning_rate": 0.0005, "loss": 2.475182294845581, "step": 4904 }, { "epoch": 0.5707470328135908, "grad_norm": 0.8402225375175476, "learning_rate": 0.0005, "loss": 2.3192100524902344, "step": 4905 }, { "epoch": 0.5708633930649291, "grad_norm": 1.35415518283844, "learning_rate": 0.0005, "loss": 2.7129201889038086, "step": 4906 }, { "epoch": 0.5709797533162672, "grad_norm": 0.43275296688079834, "learning_rate": 0.0005, "loss": 1.0352528095245361, "step": 4907 }, { "epoch": 0.5710961135676053, "grad_norm": 2.9181747436523438, "learning_rate": 0.0005, "loss": 2.092064380645752, "step": 4908 }, { "epoch": 0.5712124738189435, "grad_norm": 1.4404429197311401, "learning_rate": 0.0005, "loss": 2.934720993041992, "step": 4909 }, { "epoch": 0.5713288340702816, "grad_norm": 1.985600233078003, "learning_rate": 0.0005, "loss": 2.5751590728759766, "step": 4910 }, { "epoch": 0.5714451943216198, "grad_norm": 1.0230649709701538, "learning_rate": 0.0005, "loss": 2.6787607669830322, "step": 4911 }, { "epoch": 0.5715615545729579, "grad_norm": 1.2477991580963135, "learning_rate": 0.0005, "loss": 3.3954920768737793, "step": 4912 }, { "epoch": 0.571677914824296, "grad_norm": 2.5958495140075684, "learning_rate": 0.0005, "loss": 2.2553367614746094, "step": 4913 }, { "epoch": 0.5717942750756342, "grad_norm": 1.789640188217163, "learning_rate": 0.0005, "loss": 2.8795342445373535, "step": 4914 }, { "epoch": 0.5719106353269723, "grad_norm": 1.8013149499893188, "learning_rate": 0.0005, "loss": 1.8819596767425537, "step": 4915 }, { "epoch": 0.5720269955783105, "grad_norm": 1.223562240600586, "learning_rate": 0.0005, "loss": 2.5418813228607178, "step": 4916 }, { "epoch": 0.5721433558296486, "grad_norm": 1.2683085203170776, "learning_rate": 0.0005, "loss": 1.3697214126586914, "step": 4917 }, { "epoch": 0.5722597160809867, "grad_norm": 1.1339257955551147, "learning_rate": 0.0005, "loss": 1.817455768585205, "step": 4918 }, { "epoch": 0.5723760763323249, "grad_norm": 1.211586356163025, "learning_rate": 0.0005, "loss": 2.6574416160583496, "step": 4919 }, { "epoch": 0.572492436583663, "grad_norm": 1.4101237058639526, "learning_rate": 0.0005, "loss": 2.8686299324035645, "step": 4920 }, { "epoch": 0.5726087968350012, "grad_norm": 1.1456592082977295, "learning_rate": 0.0005, "loss": 1.7700586318969727, "step": 4921 }, { "epoch": 0.5727251570863393, "grad_norm": 0.9551551938056946, "learning_rate": 0.0005, "loss": 1.9488619565963745, "step": 4922 }, { "epoch": 0.5728415173376774, "grad_norm": 0.87574303150177, "learning_rate": 0.0005, "loss": 1.4609779119491577, "step": 4923 }, { "epoch": 0.5729578775890156, "grad_norm": 1.4417614936828613, "learning_rate": 0.0005, "loss": 2.616542339324951, "step": 4924 }, { "epoch": 0.5730742378403537, "grad_norm": 1.6723157167434692, "learning_rate": 0.0005, "loss": 3.245985984802246, "step": 4925 }, { "epoch": 0.5731905980916919, "grad_norm": 1.0749274492263794, "learning_rate": 0.0005, "loss": 1.887345314025879, "step": 4926 }, { "epoch": 0.57330695834303, "grad_norm": 1.5027215480804443, "learning_rate": 0.0005, "loss": 2.7959914207458496, "step": 4927 }, { "epoch": 0.5734233185943681, "grad_norm": 1.6203738451004028, "learning_rate": 0.0005, "loss": 2.5909500122070312, "step": 4928 }, { "epoch": 0.5735396788457063, "grad_norm": 1.6124529838562012, "learning_rate": 0.0005, "loss": 1.9175217151641846, "step": 4929 }, { "epoch": 0.5736560390970444, "grad_norm": 1.569394826889038, "learning_rate": 0.0005, "loss": 2.5984890460968018, "step": 4930 }, { "epoch": 0.5737723993483826, "grad_norm": 1.5829331874847412, "learning_rate": 0.0005, "loss": 3.280135154724121, "step": 4931 }, { "epoch": 0.5738887595997207, "grad_norm": 1.9749072790145874, "learning_rate": 0.0005, "loss": 2.989574909210205, "step": 4932 }, { "epoch": 0.5740051198510588, "grad_norm": 1.0348141193389893, "learning_rate": 0.0005, "loss": 2.2694883346557617, "step": 4933 }, { "epoch": 0.574121480102397, "grad_norm": 1.4454681873321533, "learning_rate": 0.0005, "loss": 3.465512275695801, "step": 4934 }, { "epoch": 0.5742378403537352, "grad_norm": 1.3538146018981934, "learning_rate": 0.0005, "loss": 2.345607280731201, "step": 4935 }, { "epoch": 0.5743542006050734, "grad_norm": 1.4551613330841064, "learning_rate": 0.0005, "loss": 2.9802756309509277, "step": 4936 }, { "epoch": 0.5744705608564115, "grad_norm": 1.2866463661193848, "learning_rate": 0.0005, "loss": 2.7225890159606934, "step": 4937 }, { "epoch": 0.5745869211077496, "grad_norm": 1.0351178646087646, "learning_rate": 0.0005, "loss": 2.648735761642456, "step": 4938 }, { "epoch": 0.5747032813590878, "grad_norm": 1.9435597658157349, "learning_rate": 0.0005, "loss": 3.0593385696411133, "step": 4939 }, { "epoch": 0.5748196416104259, "grad_norm": 0.9806981086730957, "learning_rate": 0.0005, "loss": 2.892704486846924, "step": 4940 }, { "epoch": 0.574936001861764, "grad_norm": 2.065488576889038, "learning_rate": 0.0005, "loss": 2.384037971496582, "step": 4941 }, { "epoch": 0.5750523621131022, "grad_norm": 1.2335695028305054, "learning_rate": 0.0005, "loss": 2.186225414276123, "step": 4942 }, { "epoch": 0.5751687223644403, "grad_norm": 1.6215466260910034, "learning_rate": 0.0005, "loss": 2.511404514312744, "step": 4943 }, { "epoch": 0.5752850826157785, "grad_norm": 1.042648196220398, "learning_rate": 0.0005, "loss": 2.0830318927764893, "step": 4944 }, { "epoch": 0.5754014428671166, "grad_norm": 1.2620859146118164, "learning_rate": 0.0005, "loss": 2.9070959091186523, "step": 4945 }, { "epoch": 0.5755178031184547, "grad_norm": 1.4356342554092407, "learning_rate": 0.0005, "loss": 2.467231273651123, "step": 4946 }, { "epoch": 0.5756341633697929, "grad_norm": 1.157885193824768, "learning_rate": 0.0005, "loss": 2.4069039821624756, "step": 4947 }, { "epoch": 0.575750523621131, "grad_norm": 1.4271405935287476, "learning_rate": 0.0005, "loss": 2.480518341064453, "step": 4948 }, { "epoch": 0.5758668838724692, "grad_norm": 1.2410887479782104, "learning_rate": 0.0005, "loss": 1.6032145023345947, "step": 4949 }, { "epoch": 0.5759832441238073, "grad_norm": 1.3486655950546265, "learning_rate": 0.0005, "loss": 2.9200592041015625, "step": 4950 }, { "epoch": 0.5760996043751454, "grad_norm": 1.835410714149475, "learning_rate": 0.0005, "loss": 2.841064453125, "step": 4951 }, { "epoch": 0.5762159646264836, "grad_norm": 1.15981125831604, "learning_rate": 0.0005, "loss": 2.7603516578674316, "step": 4952 }, { "epoch": 0.5763323248778217, "grad_norm": 1.5184402465820312, "learning_rate": 0.0005, "loss": 2.616903305053711, "step": 4953 }, { "epoch": 0.5764486851291599, "grad_norm": 0.9741116762161255, "learning_rate": 0.0005, "loss": 1.3571604490280151, "step": 4954 }, { "epoch": 0.576565045380498, "grad_norm": 1.410571575164795, "learning_rate": 0.0005, "loss": 2.7878165245056152, "step": 4955 }, { "epoch": 0.5766814056318361, "grad_norm": 1.1485401391983032, "learning_rate": 0.0005, "loss": 2.0977425575256348, "step": 4956 }, { "epoch": 0.5767977658831743, "grad_norm": 1.7251616716384888, "learning_rate": 0.0005, "loss": 2.350526809692383, "step": 4957 }, { "epoch": 0.5769141261345124, "grad_norm": 1.701641321182251, "learning_rate": 0.0005, "loss": 1.8512868881225586, "step": 4958 }, { "epoch": 0.5770304863858506, "grad_norm": 1.367621898651123, "learning_rate": 0.0005, "loss": 2.8719606399536133, "step": 4959 }, { "epoch": 0.5771468466371887, "grad_norm": 2.4832959175109863, "learning_rate": 0.0005, "loss": 3.104555606842041, "step": 4960 }, { "epoch": 0.5772632068885268, "grad_norm": 1.240811824798584, "learning_rate": 0.0005, "loss": 2.0360729694366455, "step": 4961 }, { "epoch": 0.577379567139865, "grad_norm": 1.145084261894226, "learning_rate": 0.0005, "loss": 1.8199812173843384, "step": 4962 }, { "epoch": 0.5774959273912031, "grad_norm": 2.0879645347595215, "learning_rate": 0.0005, "loss": 2.1551318168640137, "step": 4963 }, { "epoch": 0.5776122876425414, "grad_norm": 0.910757303237915, "learning_rate": 0.0005, "loss": 2.2879819869995117, "step": 4964 }, { "epoch": 0.5777286478938795, "grad_norm": 1.4903372526168823, "learning_rate": 0.0005, "loss": 2.3554317951202393, "step": 4965 }, { "epoch": 0.5778450081452176, "grad_norm": 1.4423542022705078, "learning_rate": 0.0005, "loss": 3.1156511306762695, "step": 4966 }, { "epoch": 0.5779613683965558, "grad_norm": 1.5228215456008911, "learning_rate": 0.0005, "loss": 2.5398192405700684, "step": 4967 }, { "epoch": 0.5780777286478939, "grad_norm": 1.6450353860855103, "learning_rate": 0.0005, "loss": 2.587242603302002, "step": 4968 }, { "epoch": 0.5781940888992321, "grad_norm": 1.6016349792480469, "learning_rate": 0.0005, "loss": 2.014415740966797, "step": 4969 }, { "epoch": 0.5783104491505702, "grad_norm": 0.913890540599823, "learning_rate": 0.0005, "loss": 2.7929630279541016, "step": 4970 }, { "epoch": 0.5784268094019083, "grad_norm": 1.6308192014694214, "learning_rate": 0.0005, "loss": 2.023110866546631, "step": 4971 }, { "epoch": 0.5785431696532465, "grad_norm": 1.2993108034133911, "learning_rate": 0.0005, "loss": 2.0043246746063232, "step": 4972 }, { "epoch": 0.5786595299045846, "grad_norm": 1.4863706827163696, "learning_rate": 0.0005, "loss": 2.597893714904785, "step": 4973 }, { "epoch": 0.5787758901559228, "grad_norm": 1.3436055183410645, "learning_rate": 0.0005, "loss": 2.5329174995422363, "step": 4974 }, { "epoch": 0.5788922504072609, "grad_norm": 1.417954444885254, "learning_rate": 0.0005, "loss": 2.3820791244506836, "step": 4975 }, { "epoch": 0.579008610658599, "grad_norm": 1.1481807231903076, "learning_rate": 0.0005, "loss": 2.5241987705230713, "step": 4976 }, { "epoch": 0.5791249709099372, "grad_norm": 1.7531474828720093, "learning_rate": 0.0005, "loss": 3.7979326248168945, "step": 4977 }, { "epoch": 0.5792413311612753, "grad_norm": 0.8364459872245789, "learning_rate": 0.0005, "loss": 2.006645679473877, "step": 4978 }, { "epoch": 0.5793576914126134, "grad_norm": 1.0812067985534668, "learning_rate": 0.0005, "loss": 2.911501169204712, "step": 4979 }, { "epoch": 0.5794740516639516, "grad_norm": 1.2621183395385742, "learning_rate": 0.0005, "loss": 2.5286614894866943, "step": 4980 }, { "epoch": 0.5795904119152897, "grad_norm": 1.5641049146652222, "learning_rate": 0.0005, "loss": 2.942079544067383, "step": 4981 }, { "epoch": 0.5797067721666279, "grad_norm": 1.4974230527877808, "learning_rate": 0.0005, "loss": 2.9546172618865967, "step": 4982 }, { "epoch": 0.579823132417966, "grad_norm": 1.5020745992660522, "learning_rate": 0.0005, "loss": 1.9558343887329102, "step": 4983 }, { "epoch": 0.5799394926693041, "grad_norm": 1.6500147581100464, "learning_rate": 0.0005, "loss": 2.075864791870117, "step": 4984 }, { "epoch": 0.5800558529206423, "grad_norm": 1.2140018939971924, "learning_rate": 0.0005, "loss": 2.2317285537719727, "step": 4985 }, { "epoch": 0.5801722131719804, "grad_norm": 0.7338871359825134, "learning_rate": 0.0005, "loss": 1.3027153015136719, "step": 4986 }, { "epoch": 0.5802885734233186, "grad_norm": 1.6731774806976318, "learning_rate": 0.0005, "loss": 2.7063651084899902, "step": 4987 }, { "epoch": 0.5804049336746567, "grad_norm": 2.3416662216186523, "learning_rate": 0.0005, "loss": 2.6441116333007812, "step": 4988 }, { "epoch": 0.5805212939259948, "grad_norm": 2.2519030570983887, "learning_rate": 0.0005, "loss": 3.6426074504852295, "step": 4989 }, { "epoch": 0.580637654177333, "grad_norm": 1.0944790840148926, "learning_rate": 0.0005, "loss": 2.2277326583862305, "step": 4990 }, { "epoch": 0.5807540144286711, "grad_norm": 2.087804079055786, "learning_rate": 0.0005, "loss": 3.0558762550354004, "step": 4991 }, { "epoch": 0.5808703746800093, "grad_norm": 1.4735772609710693, "learning_rate": 0.0005, "loss": 3.121201515197754, "step": 4992 }, { "epoch": 0.5809867349313474, "grad_norm": 1.844943642616272, "learning_rate": 0.0005, "loss": 3.3606934547424316, "step": 4993 }, { "epoch": 0.5811030951826855, "grad_norm": 1.4473878145217896, "learning_rate": 0.0005, "loss": 1.8771843910217285, "step": 4994 }, { "epoch": 0.5812194554340238, "grad_norm": 2.297982692718506, "learning_rate": 0.0005, "loss": 3.043546199798584, "step": 4995 }, { "epoch": 0.5813358156853619, "grad_norm": 1.735012173652649, "learning_rate": 0.0005, "loss": 2.6157569885253906, "step": 4996 }, { "epoch": 0.5814521759367001, "grad_norm": 0.9629035592079163, "learning_rate": 0.0005, "loss": 2.1525373458862305, "step": 4997 }, { "epoch": 0.5815685361880382, "grad_norm": 1.6683212518692017, "learning_rate": 0.0005, "loss": 2.872270107269287, "step": 4998 }, { "epoch": 0.5816848964393763, "grad_norm": 1.5511947870254517, "learning_rate": 0.0005, "loss": 1.8359591960906982, "step": 4999 }, { "epoch": 0.5818012566907145, "grad_norm": 1.269931435585022, "learning_rate": 0.0005, "loss": 2.0013270378112793, "step": 5000 }, { "epoch": 0.5819176169420526, "grad_norm": 1.767072319984436, "learning_rate": 0.0005, "loss": 2.2255818843841553, "step": 5001 }, { "epoch": 0.5820339771933908, "grad_norm": 0.8274176120758057, "learning_rate": 0.0005, "loss": 1.5439183712005615, "step": 5002 }, { "epoch": 0.5821503374447289, "grad_norm": 1.4383193254470825, "learning_rate": 0.0005, "loss": 3.1258745193481445, "step": 5003 }, { "epoch": 0.582266697696067, "grad_norm": 1.6174858808517456, "learning_rate": 0.0005, "loss": 2.9028360843658447, "step": 5004 }, { "epoch": 0.5823830579474052, "grad_norm": 1.6505807638168335, "learning_rate": 0.0005, "loss": 2.552182674407959, "step": 5005 }, { "epoch": 0.5824994181987433, "grad_norm": 1.3180702924728394, "learning_rate": 0.0005, "loss": 1.9013383388519287, "step": 5006 }, { "epoch": 0.5826157784500815, "grad_norm": 1.4734636545181274, "learning_rate": 0.0005, "loss": 2.922947883605957, "step": 5007 }, { "epoch": 0.5827321387014196, "grad_norm": 1.6085399389266968, "learning_rate": 0.0005, "loss": 2.306851863861084, "step": 5008 }, { "epoch": 0.5828484989527577, "grad_norm": 1.3922656774520874, "learning_rate": 0.0005, "loss": 2.5028388500213623, "step": 5009 }, { "epoch": 0.5829648592040959, "grad_norm": 1.71601402759552, "learning_rate": 0.0005, "loss": 2.373126983642578, "step": 5010 }, { "epoch": 0.583081219455434, "grad_norm": 1.055856466293335, "learning_rate": 0.0005, "loss": 2.100015640258789, "step": 5011 }, { "epoch": 0.5831975797067722, "grad_norm": 1.3765647411346436, "learning_rate": 0.0005, "loss": 3.0045666694641113, "step": 5012 }, { "epoch": 0.5833139399581103, "grad_norm": 1.4065295457839966, "learning_rate": 0.0005, "loss": 3.077594757080078, "step": 5013 }, { "epoch": 0.5834303002094484, "grad_norm": 0.9240631461143494, "learning_rate": 0.0005, "loss": 1.7625703811645508, "step": 5014 }, { "epoch": 0.5835466604607866, "grad_norm": 1.153354287147522, "learning_rate": 0.0005, "loss": 2.4715185165405273, "step": 5015 }, { "epoch": 0.5836630207121247, "grad_norm": 0.9798492789268494, "learning_rate": 0.0005, "loss": 2.0149388313293457, "step": 5016 }, { "epoch": 0.5837793809634629, "grad_norm": 1.1843664646148682, "learning_rate": 0.0005, "loss": 1.7578063011169434, "step": 5017 }, { "epoch": 0.583895741214801, "grad_norm": 1.5901914834976196, "learning_rate": 0.0005, "loss": 3.1878793239593506, "step": 5018 }, { "epoch": 0.5840121014661391, "grad_norm": 1.4409494400024414, "learning_rate": 0.0005, "loss": 2.1912317276000977, "step": 5019 }, { "epoch": 0.5841284617174773, "grad_norm": 1.499985694885254, "learning_rate": 0.0005, "loss": 2.713500499725342, "step": 5020 }, { "epoch": 0.5842448219688154, "grad_norm": 1.6622966527938843, "learning_rate": 0.0005, "loss": 1.5847244262695312, "step": 5021 }, { "epoch": 0.5843611822201535, "grad_norm": 1.2655712366104126, "learning_rate": 0.0005, "loss": 2.634491443634033, "step": 5022 }, { "epoch": 0.5844775424714918, "grad_norm": 1.26528799533844, "learning_rate": 0.0005, "loss": 3.0936646461486816, "step": 5023 }, { "epoch": 0.5845939027228299, "grad_norm": 0.9899635314941406, "learning_rate": 0.0005, "loss": 2.1171681880950928, "step": 5024 }, { "epoch": 0.5847102629741681, "grad_norm": 1.6655668020248413, "learning_rate": 0.0005, "loss": 2.1918749809265137, "step": 5025 }, { "epoch": 0.5848266232255062, "grad_norm": 1.2583637237548828, "learning_rate": 0.0005, "loss": 2.625831127166748, "step": 5026 }, { "epoch": 0.5849429834768443, "grad_norm": 2.0013692378997803, "learning_rate": 0.0005, "loss": 3.323798656463623, "step": 5027 }, { "epoch": 0.5850593437281825, "grad_norm": 1.3996020555496216, "learning_rate": 0.0005, "loss": 2.4556081295013428, "step": 5028 }, { "epoch": 0.5851757039795206, "grad_norm": 3.2950751781463623, "learning_rate": 0.0005, "loss": 2.92557954788208, "step": 5029 }, { "epoch": 0.5852920642308588, "grad_norm": 1.6182010173797607, "learning_rate": 0.0005, "loss": 2.6270101070404053, "step": 5030 }, { "epoch": 0.5854084244821969, "grad_norm": 1.2164239883422852, "learning_rate": 0.0005, "loss": 3.7892534732818604, "step": 5031 }, { "epoch": 0.585524784733535, "grad_norm": 1.0639077425003052, "learning_rate": 0.0005, "loss": 2.732184648513794, "step": 5032 }, { "epoch": 0.5856411449848732, "grad_norm": 1.7808469533920288, "learning_rate": 0.0005, "loss": 2.712800979614258, "step": 5033 }, { "epoch": 0.5857575052362113, "grad_norm": 1.7240866422653198, "learning_rate": 0.0005, "loss": 2.919154167175293, "step": 5034 }, { "epoch": 0.5858738654875495, "grad_norm": 1.3513909578323364, "learning_rate": 0.0005, "loss": 3.0359420776367188, "step": 5035 }, { "epoch": 0.5859902257388876, "grad_norm": 0.8505710363388062, "learning_rate": 0.0005, "loss": 1.8749760389328003, "step": 5036 }, { "epoch": 0.5861065859902257, "grad_norm": 2.2458419799804688, "learning_rate": 0.0005, "loss": 1.587962031364441, "step": 5037 }, { "epoch": 0.5862229462415639, "grad_norm": 1.0332890748977661, "learning_rate": 0.0005, "loss": 1.6433296203613281, "step": 5038 }, { "epoch": 0.586339306492902, "grad_norm": 1.4532285928726196, "learning_rate": 0.0005, "loss": 2.612037181854248, "step": 5039 }, { "epoch": 0.5864556667442402, "grad_norm": 1.1623588800430298, "learning_rate": 0.0005, "loss": 1.6762487888336182, "step": 5040 }, { "epoch": 0.5865720269955783, "grad_norm": 1.7923990488052368, "learning_rate": 0.0005, "loss": 3.0968804359436035, "step": 5041 }, { "epoch": 0.5866883872469164, "grad_norm": 1.46500563621521, "learning_rate": 0.0005, "loss": 2.5378611087799072, "step": 5042 }, { "epoch": 0.5868047474982546, "grad_norm": 1.8507412672042847, "learning_rate": 0.0005, "loss": 2.9861159324645996, "step": 5043 }, { "epoch": 0.5869211077495927, "grad_norm": 1.897735357284546, "learning_rate": 0.0005, "loss": 2.612125873565674, "step": 5044 }, { "epoch": 0.5870374680009309, "grad_norm": 1.457945466041565, "learning_rate": 0.0005, "loss": 2.979635238647461, "step": 5045 }, { "epoch": 0.587153828252269, "grad_norm": 1.3537876605987549, "learning_rate": 0.0005, "loss": 2.924753189086914, "step": 5046 }, { "epoch": 0.5872701885036071, "grad_norm": 3.0764312744140625, "learning_rate": 0.0005, "loss": 2.6950483322143555, "step": 5047 }, { "epoch": 0.5873865487549453, "grad_norm": 1.5560956001281738, "learning_rate": 0.0005, "loss": 3.104255199432373, "step": 5048 }, { "epoch": 0.5875029090062834, "grad_norm": 1.4037270545959473, "learning_rate": 0.0005, "loss": 1.7274398803710938, "step": 5049 }, { "epoch": 0.5876192692576216, "grad_norm": 1.6068952083587646, "learning_rate": 0.0005, "loss": 2.419921875, "step": 5050 }, { "epoch": 0.5877356295089597, "grad_norm": 1.3659743070602417, "learning_rate": 0.0005, "loss": 2.3002071380615234, "step": 5051 }, { "epoch": 0.5878519897602978, "grad_norm": 1.0272595882415771, "learning_rate": 0.0005, "loss": 1.6720077991485596, "step": 5052 }, { "epoch": 0.587968350011636, "grad_norm": 1.426501750946045, "learning_rate": 0.0005, "loss": 2.571115493774414, "step": 5053 }, { "epoch": 0.5880847102629742, "grad_norm": 1.6531898975372314, "learning_rate": 0.0005, "loss": 2.4988951683044434, "step": 5054 }, { "epoch": 0.5882010705143124, "grad_norm": 1.505996584892273, "learning_rate": 0.0005, "loss": 2.23286771774292, "step": 5055 }, { "epoch": 0.5883174307656505, "grad_norm": 1.1225775480270386, "learning_rate": 0.0005, "loss": 2.673847198486328, "step": 5056 }, { "epoch": 0.5884337910169886, "grad_norm": 1.5167350769042969, "learning_rate": 0.0005, "loss": 2.0511302947998047, "step": 5057 }, { "epoch": 0.5885501512683268, "grad_norm": 1.0009028911590576, "learning_rate": 0.0005, "loss": 2.113766670227051, "step": 5058 }, { "epoch": 0.5886665115196649, "grad_norm": 1.052918553352356, "learning_rate": 0.0005, "loss": 2.0840108394622803, "step": 5059 }, { "epoch": 0.5887828717710031, "grad_norm": 1.1520318984985352, "learning_rate": 0.0005, "loss": 3.06908917427063, "step": 5060 }, { "epoch": 0.5888992320223412, "grad_norm": 1.7140936851501465, "learning_rate": 0.0005, "loss": 2.829334259033203, "step": 5061 }, { "epoch": 0.5890155922736793, "grad_norm": 1.5019304752349854, "learning_rate": 0.0005, "loss": 2.5441181659698486, "step": 5062 }, { "epoch": 0.5891319525250175, "grad_norm": 1.2942986488342285, "learning_rate": 0.0005, "loss": 2.2314257621765137, "step": 5063 }, { "epoch": 0.5892483127763556, "grad_norm": 1.2796809673309326, "learning_rate": 0.0005, "loss": 2.376492977142334, "step": 5064 }, { "epoch": 0.5893646730276937, "grad_norm": 1.2508115768432617, "learning_rate": 0.0005, "loss": 2.2067954540252686, "step": 5065 }, { "epoch": 0.5894810332790319, "grad_norm": 2.0413081645965576, "learning_rate": 0.0005, "loss": 2.999391555786133, "step": 5066 }, { "epoch": 0.58959739353037, "grad_norm": 0.748653769493103, "learning_rate": 0.0005, "loss": 2.1388659477233887, "step": 5067 }, { "epoch": 0.5897137537817082, "grad_norm": 2.70459246635437, "learning_rate": 0.0005, "loss": 1.90003502368927, "step": 5068 }, { "epoch": 0.5898301140330463, "grad_norm": 1.9605598449707031, "learning_rate": 0.0005, "loss": 3.1541810035705566, "step": 5069 }, { "epoch": 0.5899464742843844, "grad_norm": 1.529555082321167, "learning_rate": 0.0005, "loss": 2.780978202819824, "step": 5070 }, { "epoch": 0.5900628345357226, "grad_norm": 1.1463209390640259, "learning_rate": 0.0005, "loss": 2.4813389778137207, "step": 5071 }, { "epoch": 0.5901791947870607, "grad_norm": 1.4633355140686035, "learning_rate": 0.0005, "loss": 2.8787131309509277, "step": 5072 }, { "epoch": 0.5902955550383989, "grad_norm": 1.5168083906173706, "learning_rate": 0.0005, "loss": 2.549635410308838, "step": 5073 }, { "epoch": 0.590411915289737, "grad_norm": 1.6418076753616333, "learning_rate": 0.0005, "loss": 2.5028271675109863, "step": 5074 }, { "epoch": 0.5905282755410751, "grad_norm": 1.463559865951538, "learning_rate": 0.0005, "loss": 2.9398388862609863, "step": 5075 }, { "epoch": 0.5906446357924133, "grad_norm": 1.1458790302276611, "learning_rate": 0.0005, "loss": 1.500706434249878, "step": 5076 }, { "epoch": 0.5907609960437514, "grad_norm": 0.9446449875831604, "learning_rate": 0.0005, "loss": 2.4676766395568848, "step": 5077 }, { "epoch": 0.5908773562950896, "grad_norm": 1.0870634317398071, "learning_rate": 0.0005, "loss": 1.9068598747253418, "step": 5078 }, { "epoch": 0.5909937165464277, "grad_norm": 1.3767292499542236, "learning_rate": 0.0005, "loss": 2.5029687881469727, "step": 5079 }, { "epoch": 0.5911100767977658, "grad_norm": 1.121969223022461, "learning_rate": 0.0005, "loss": 2.206693649291992, "step": 5080 }, { "epoch": 0.591226437049104, "grad_norm": 1.290236234664917, "learning_rate": 0.0005, "loss": 2.8644397258758545, "step": 5081 }, { "epoch": 0.5913427973004421, "grad_norm": 1.7708739042282104, "learning_rate": 0.0005, "loss": 2.7913601398468018, "step": 5082 }, { "epoch": 0.5914591575517804, "grad_norm": 1.748870849609375, "learning_rate": 0.0005, "loss": 2.918095111846924, "step": 5083 }, { "epoch": 0.5915755178031185, "grad_norm": 1.174122929573059, "learning_rate": 0.0005, "loss": 2.342764377593994, "step": 5084 }, { "epoch": 0.5916918780544566, "grad_norm": 1.014850378036499, "learning_rate": 0.0005, "loss": 2.7236547470092773, "step": 5085 }, { "epoch": 0.5918082383057948, "grad_norm": 0.8637403845787048, "learning_rate": 0.0005, "loss": 2.1215672492980957, "step": 5086 }, { "epoch": 0.5919245985571329, "grad_norm": 0.7163143157958984, "learning_rate": 0.0005, "loss": 1.5656734704971313, "step": 5087 }, { "epoch": 0.5920409588084711, "grad_norm": 1.4391037225723267, "learning_rate": 0.0005, "loss": 2.3813161849975586, "step": 5088 }, { "epoch": 0.5921573190598092, "grad_norm": 0.7158318161964417, "learning_rate": 0.0005, "loss": 1.2981910705566406, "step": 5089 }, { "epoch": 0.5922736793111473, "grad_norm": 1.2499535083770752, "learning_rate": 0.0005, "loss": 2.9655184745788574, "step": 5090 }, { "epoch": 0.5923900395624855, "grad_norm": 1.376986026763916, "learning_rate": 0.0005, "loss": 1.996962547302246, "step": 5091 }, { "epoch": 0.5925063998138236, "grad_norm": 1.0380785465240479, "learning_rate": 0.0005, "loss": 2.3145346641540527, "step": 5092 }, { "epoch": 0.5926227600651618, "grad_norm": 2.443202018737793, "learning_rate": 0.0005, "loss": 2.691089153289795, "step": 5093 }, { "epoch": 0.5927391203164999, "grad_norm": 1.4663634300231934, "learning_rate": 0.0005, "loss": 2.3481404781341553, "step": 5094 }, { "epoch": 0.592855480567838, "grad_norm": 1.7498301267623901, "learning_rate": 0.0005, "loss": 2.5210120677948, "step": 5095 }, { "epoch": 0.5929718408191762, "grad_norm": 1.1654716730117798, "learning_rate": 0.0005, "loss": 2.0264406204223633, "step": 5096 }, { "epoch": 0.5930882010705143, "grad_norm": 1.47840416431427, "learning_rate": 0.0005, "loss": 2.3797225952148438, "step": 5097 }, { "epoch": 0.5932045613218525, "grad_norm": 1.4016270637512207, "learning_rate": 0.0005, "loss": 2.5236244201660156, "step": 5098 }, { "epoch": 0.5933209215731906, "grad_norm": 1.29625403881073, "learning_rate": 0.0005, "loss": 1.924552321434021, "step": 5099 }, { "epoch": 0.5934372818245287, "grad_norm": 0.9530450701713562, "learning_rate": 0.0005, "loss": 2.0391430854797363, "step": 5100 }, { "epoch": 0.5935536420758669, "grad_norm": 1.1683448553085327, "learning_rate": 0.0005, "loss": 2.782862663269043, "step": 5101 }, { "epoch": 0.593670002327205, "grad_norm": 0.8621286153793335, "learning_rate": 0.0005, "loss": 2.705662727355957, "step": 5102 }, { "epoch": 0.5937863625785431, "grad_norm": 1.8735748529434204, "learning_rate": 0.0005, "loss": 1.6897999048233032, "step": 5103 }, { "epoch": 0.5939027228298813, "grad_norm": 1.3538495302200317, "learning_rate": 0.0005, "loss": 1.8515243530273438, "step": 5104 }, { "epoch": 0.5940190830812194, "grad_norm": 1.692826747894287, "learning_rate": 0.0005, "loss": 2.112826347351074, "step": 5105 }, { "epoch": 0.5941354433325576, "grad_norm": 1.909965991973877, "learning_rate": 0.0005, "loss": 3.3775107860565186, "step": 5106 }, { "epoch": 0.5942518035838957, "grad_norm": 1.5803085565567017, "learning_rate": 0.0005, "loss": 2.4118237495422363, "step": 5107 }, { "epoch": 0.5943681638352338, "grad_norm": 1.469330072402954, "learning_rate": 0.0005, "loss": 3.4067251682281494, "step": 5108 }, { "epoch": 0.594484524086572, "grad_norm": 1.7345937490463257, "learning_rate": 0.0005, "loss": 2.2519009113311768, "step": 5109 }, { "epoch": 0.5946008843379101, "grad_norm": 1.5009886026382446, "learning_rate": 0.0005, "loss": 2.859884738922119, "step": 5110 }, { "epoch": 0.5947172445892484, "grad_norm": 2.432373523712158, "learning_rate": 0.0005, "loss": 3.4679853916168213, "step": 5111 }, { "epoch": 0.5948336048405864, "grad_norm": 0.906930148601532, "learning_rate": 0.0005, "loss": 2.0961074829101562, "step": 5112 }, { "epoch": 0.5949499650919245, "grad_norm": 1.682231068611145, "learning_rate": 0.0005, "loss": 3.130770206451416, "step": 5113 }, { "epoch": 0.5950663253432628, "grad_norm": 1.922986626625061, "learning_rate": 0.0005, "loss": 2.8774209022521973, "step": 5114 }, { "epoch": 0.5951826855946009, "grad_norm": 1.7594610452651978, "learning_rate": 0.0005, "loss": 2.4586076736450195, "step": 5115 }, { "epoch": 0.5952990458459391, "grad_norm": 1.188852071762085, "learning_rate": 0.0005, "loss": 2.752115249633789, "step": 5116 }, { "epoch": 0.5954154060972772, "grad_norm": 1.1858896017074585, "learning_rate": 0.0005, "loss": 1.826213002204895, "step": 5117 }, { "epoch": 0.5955317663486153, "grad_norm": 1.5326265096664429, "learning_rate": 0.0005, "loss": 2.413940668106079, "step": 5118 }, { "epoch": 0.5956481265999535, "grad_norm": 2.482356071472168, "learning_rate": 0.0005, "loss": 2.6949281692504883, "step": 5119 }, { "epoch": 0.5957644868512916, "grad_norm": 1.3419358730316162, "learning_rate": 0.0005, "loss": 2.7718780040740967, "step": 5120 }, { "epoch": 0.5958808471026298, "grad_norm": 1.025955080986023, "learning_rate": 0.0005, "loss": 2.385849952697754, "step": 5121 }, { "epoch": 0.5959972073539679, "grad_norm": 1.0634984970092773, "learning_rate": 0.0005, "loss": 2.0367884635925293, "step": 5122 }, { "epoch": 0.596113567605306, "grad_norm": 1.3467639684677124, "learning_rate": 0.0005, "loss": 2.665120840072632, "step": 5123 }, { "epoch": 0.5962299278566442, "grad_norm": 1.0516636371612549, "learning_rate": 0.0005, "loss": 2.163856029510498, "step": 5124 }, { "epoch": 0.5963462881079823, "grad_norm": 1.7683302164077759, "learning_rate": 0.0005, "loss": 2.708740234375, "step": 5125 }, { "epoch": 0.5964626483593205, "grad_norm": 1.3448009490966797, "learning_rate": 0.0005, "loss": 2.5113675594329834, "step": 5126 }, { "epoch": 0.5965790086106586, "grad_norm": 1.7273319959640503, "learning_rate": 0.0005, "loss": 2.93701171875, "step": 5127 }, { "epoch": 0.5966953688619967, "grad_norm": 1.3662163019180298, "learning_rate": 0.0005, "loss": 2.724219799041748, "step": 5128 }, { "epoch": 0.5968117291133349, "grad_norm": 1.7960257530212402, "learning_rate": 0.0005, "loss": 3.560330629348755, "step": 5129 }, { "epoch": 0.596928089364673, "grad_norm": 1.3652925491333008, "learning_rate": 0.0005, "loss": 2.5370357036590576, "step": 5130 }, { "epoch": 0.5970444496160112, "grad_norm": 1.1189544200897217, "learning_rate": 0.0005, "loss": 1.8669666051864624, "step": 5131 }, { "epoch": 0.5971608098673493, "grad_norm": 2.271319627761841, "learning_rate": 0.0005, "loss": 2.7055587768554688, "step": 5132 }, { "epoch": 0.5972771701186874, "grad_norm": 1.517158031463623, "learning_rate": 0.0005, "loss": 1.9942647218704224, "step": 5133 }, { "epoch": 0.5973935303700256, "grad_norm": 0.7540388107299805, "learning_rate": 0.0005, "loss": 1.6335095167160034, "step": 5134 }, { "epoch": 0.5975098906213637, "grad_norm": 1.7912129163742065, "learning_rate": 0.0005, "loss": 2.5996718406677246, "step": 5135 }, { "epoch": 0.5976262508727019, "grad_norm": 0.9380338191986084, "learning_rate": 0.0005, "loss": 2.0395708084106445, "step": 5136 }, { "epoch": 0.59774261112404, "grad_norm": 1.3968234062194824, "learning_rate": 0.0005, "loss": 3.501502752304077, "step": 5137 }, { "epoch": 0.5978589713753781, "grad_norm": 0.918893039226532, "learning_rate": 0.0005, "loss": 1.9752676486968994, "step": 5138 }, { "epoch": 0.5979753316267163, "grad_norm": 1.2113451957702637, "learning_rate": 0.0005, "loss": 1.7639614343643188, "step": 5139 }, { "epoch": 0.5980916918780544, "grad_norm": 1.167879343032837, "learning_rate": 0.0005, "loss": 2.3618760108947754, "step": 5140 }, { "epoch": 0.5982080521293927, "grad_norm": 1.2606980800628662, "learning_rate": 0.0005, "loss": 2.278271198272705, "step": 5141 }, { "epoch": 0.5983244123807308, "grad_norm": 1.9460985660552979, "learning_rate": 0.0005, "loss": 3.3464908599853516, "step": 5142 }, { "epoch": 0.5984407726320689, "grad_norm": 1.7190008163452148, "learning_rate": 0.0005, "loss": 3.797760248184204, "step": 5143 }, { "epoch": 0.5985571328834071, "grad_norm": 1.894939661026001, "learning_rate": 0.0005, "loss": 2.6874518394470215, "step": 5144 }, { "epoch": 0.5986734931347452, "grad_norm": 1.752074122428894, "learning_rate": 0.0005, "loss": 2.9609360694885254, "step": 5145 }, { "epoch": 0.5987898533860833, "grad_norm": 1.3750838041305542, "learning_rate": 0.0005, "loss": 1.8122353553771973, "step": 5146 }, { "epoch": 0.5989062136374215, "grad_norm": 1.75764000415802, "learning_rate": 0.0005, "loss": 2.5599048137664795, "step": 5147 }, { "epoch": 0.5990225738887596, "grad_norm": 1.4836392402648926, "learning_rate": 0.0005, "loss": 3.3361330032348633, "step": 5148 }, { "epoch": 0.5991389341400978, "grad_norm": 0.8876492381095886, "learning_rate": 0.0005, "loss": 1.9311275482177734, "step": 5149 }, { "epoch": 0.5992552943914359, "grad_norm": 1.3876925706863403, "learning_rate": 0.0005, "loss": 2.0416088104248047, "step": 5150 }, { "epoch": 0.599371654642774, "grad_norm": 2.2497446537017822, "learning_rate": 0.0005, "loss": 2.9633569717407227, "step": 5151 }, { "epoch": 0.5994880148941122, "grad_norm": 1.7735451459884644, "learning_rate": 0.0005, "loss": 2.5103797912597656, "step": 5152 }, { "epoch": 0.5996043751454503, "grad_norm": 1.205041766166687, "learning_rate": 0.0005, "loss": 2.751722812652588, "step": 5153 }, { "epoch": 0.5997207353967885, "grad_norm": 1.4910317659378052, "learning_rate": 0.0005, "loss": 2.203331232070923, "step": 5154 }, { "epoch": 0.5998370956481266, "grad_norm": 1.212788701057434, "learning_rate": 0.0005, "loss": 2.5392401218414307, "step": 5155 }, { "epoch": 0.5999534558994647, "grad_norm": 1.0930861234664917, "learning_rate": 0.0005, "loss": 1.7995555400848389, "step": 5156 }, { "epoch": 0.6000698161508029, "grad_norm": 1.3595596551895142, "learning_rate": 0.0005, "loss": 2.3707776069641113, "step": 5157 }, { "epoch": 0.600186176402141, "grad_norm": 1.7580429315567017, "learning_rate": 0.0005, "loss": 1.853244662284851, "step": 5158 }, { "epoch": 0.6003025366534792, "grad_norm": 2.1270198822021484, "learning_rate": 0.0005, "loss": 3.044673442840576, "step": 5159 }, { "epoch": 0.6004188969048173, "grad_norm": 1.0707335472106934, "learning_rate": 0.0005, "loss": 2.509476661682129, "step": 5160 }, { "epoch": 0.6005352571561554, "grad_norm": 1.0410090684890747, "learning_rate": 0.0005, "loss": 2.3422112464904785, "step": 5161 }, { "epoch": 0.6006516174074936, "grad_norm": 1.0685222148895264, "learning_rate": 0.0005, "loss": 1.4284451007843018, "step": 5162 }, { "epoch": 0.6007679776588317, "grad_norm": 1.3607637882232666, "learning_rate": 0.0005, "loss": 2.051712989807129, "step": 5163 }, { "epoch": 0.6008843379101699, "grad_norm": 1.104966163635254, "learning_rate": 0.0005, "loss": 2.9377529621124268, "step": 5164 }, { "epoch": 0.601000698161508, "grad_norm": 1.4222643375396729, "learning_rate": 0.0005, "loss": 2.873788833618164, "step": 5165 }, { "epoch": 0.6011170584128461, "grad_norm": 1.5039167404174805, "learning_rate": 0.0005, "loss": 1.780956745147705, "step": 5166 }, { "epoch": 0.6012334186641843, "grad_norm": 0.8790713548660278, "learning_rate": 0.0005, "loss": 2.0415618419647217, "step": 5167 }, { "epoch": 0.6013497789155224, "grad_norm": 1.3924710750579834, "learning_rate": 0.0005, "loss": 2.4181084632873535, "step": 5168 }, { "epoch": 0.6014661391668606, "grad_norm": 1.8411635160446167, "learning_rate": 0.0005, "loss": 2.3365750312805176, "step": 5169 }, { "epoch": 0.6015824994181987, "grad_norm": 1.9734920263290405, "learning_rate": 0.0005, "loss": 2.2037458419799805, "step": 5170 }, { "epoch": 0.6016988596695368, "grad_norm": 1.3872599601745605, "learning_rate": 0.0005, "loss": 3.286898612976074, "step": 5171 }, { "epoch": 0.601815219920875, "grad_norm": 1.871841549873352, "learning_rate": 0.0005, "loss": 2.300548553466797, "step": 5172 }, { "epoch": 0.6019315801722132, "grad_norm": 1.4430339336395264, "learning_rate": 0.0005, "loss": 2.8988757133483887, "step": 5173 }, { "epoch": 0.6020479404235514, "grad_norm": 0.8924757838249207, "learning_rate": 0.0005, "loss": 1.9088397026062012, "step": 5174 }, { "epoch": 0.6021643006748895, "grad_norm": 1.2923663854599, "learning_rate": 0.0005, "loss": 2.158839225769043, "step": 5175 }, { "epoch": 0.6022806609262276, "grad_norm": 1.9514260292053223, "learning_rate": 0.0005, "loss": 2.846282958984375, "step": 5176 }, { "epoch": 0.6023970211775658, "grad_norm": 2.087754964828491, "learning_rate": 0.0005, "loss": 3.3485212326049805, "step": 5177 }, { "epoch": 0.6025133814289039, "grad_norm": 1.7382527589797974, "learning_rate": 0.0005, "loss": 2.336761713027954, "step": 5178 }, { "epoch": 0.6026297416802421, "grad_norm": 1.6387939453125, "learning_rate": 0.0005, "loss": 3.007108449935913, "step": 5179 }, { "epoch": 0.6027461019315802, "grad_norm": 1.0551998615264893, "learning_rate": 0.0005, "loss": 2.0328762531280518, "step": 5180 }, { "epoch": 0.6028624621829183, "grad_norm": 1.2118923664093018, "learning_rate": 0.0005, "loss": 2.047125816345215, "step": 5181 }, { "epoch": 0.6029788224342565, "grad_norm": 1.321945071220398, "learning_rate": 0.0005, "loss": 2.6572132110595703, "step": 5182 }, { "epoch": 0.6030951826855946, "grad_norm": 1.442402958869934, "learning_rate": 0.0005, "loss": 2.337623119354248, "step": 5183 }, { "epoch": 0.6032115429369328, "grad_norm": 1.9335740804672241, "learning_rate": 0.0005, "loss": 3.480422019958496, "step": 5184 }, { "epoch": 0.6033279031882709, "grad_norm": 1.4461275339126587, "learning_rate": 0.0005, "loss": 1.5388312339782715, "step": 5185 }, { "epoch": 0.603444263439609, "grad_norm": 1.408394694328308, "learning_rate": 0.0005, "loss": 2.7043235301971436, "step": 5186 }, { "epoch": 0.6035606236909472, "grad_norm": 1.607194185256958, "learning_rate": 0.0005, "loss": 3.219985008239746, "step": 5187 }, { "epoch": 0.6036769839422853, "grad_norm": 1.6522157192230225, "learning_rate": 0.0005, "loss": 2.372631072998047, "step": 5188 }, { "epoch": 0.6037933441936234, "grad_norm": 0.47076722979545593, "learning_rate": 0.0005, "loss": 1.0933587551116943, "step": 5189 }, { "epoch": 0.6039097044449616, "grad_norm": 2.863219976425171, "learning_rate": 0.0005, "loss": 3.1811442375183105, "step": 5190 }, { "epoch": 0.6040260646962997, "grad_norm": 1.1978203058242798, "learning_rate": 0.0005, "loss": 3.0272953510284424, "step": 5191 }, { "epoch": 0.6041424249476379, "grad_norm": 1.659914493560791, "learning_rate": 0.0005, "loss": 3.2136130332946777, "step": 5192 }, { "epoch": 0.604258785198976, "grad_norm": 1.057526707649231, "learning_rate": 0.0005, "loss": 2.36348819732666, "step": 5193 }, { "epoch": 0.6043751454503141, "grad_norm": 1.158929467201233, "learning_rate": 0.0005, "loss": 3.0145270824432373, "step": 5194 }, { "epoch": 0.6044915057016523, "grad_norm": 1.7114137411117554, "learning_rate": 0.0005, "loss": 2.239680290222168, "step": 5195 }, { "epoch": 0.6046078659529904, "grad_norm": 1.3065251111984253, "learning_rate": 0.0005, "loss": 2.9173667430877686, "step": 5196 }, { "epoch": 0.6047242262043286, "grad_norm": 1.9239592552185059, "learning_rate": 0.0005, "loss": 2.946532964706421, "step": 5197 }, { "epoch": 0.6048405864556667, "grad_norm": 1.1981637477874756, "learning_rate": 0.0005, "loss": 2.4119207859039307, "step": 5198 }, { "epoch": 0.6049569467070048, "grad_norm": 2.291775941848755, "learning_rate": 0.0005, "loss": 2.104060411453247, "step": 5199 }, { "epoch": 0.605073306958343, "grad_norm": 1.5759357213974, "learning_rate": 0.0005, "loss": 3.2216248512268066, "step": 5200 }, { "epoch": 0.6051896672096811, "grad_norm": 1.0694431066513062, "learning_rate": 0.0005, "loss": 1.828710913658142, "step": 5201 }, { "epoch": 0.6053060274610194, "grad_norm": 1.4723762273788452, "learning_rate": 0.0005, "loss": 2.3345718383789062, "step": 5202 }, { "epoch": 0.6054223877123575, "grad_norm": 1.4065526723861694, "learning_rate": 0.0005, "loss": 3.085747241973877, "step": 5203 }, { "epoch": 0.6055387479636956, "grad_norm": 2.9312775135040283, "learning_rate": 0.0005, "loss": 3.2438673973083496, "step": 5204 }, { "epoch": 0.6056551082150338, "grad_norm": 2.0263609886169434, "learning_rate": 0.0005, "loss": 2.4216504096984863, "step": 5205 }, { "epoch": 0.6057714684663719, "grad_norm": 1.304914951324463, "learning_rate": 0.0005, "loss": 2.924553394317627, "step": 5206 }, { "epoch": 0.6058878287177101, "grad_norm": 1.1499189138412476, "learning_rate": 0.0005, "loss": 2.454916477203369, "step": 5207 }, { "epoch": 0.6060041889690482, "grad_norm": 0.6693520545959473, "learning_rate": 0.0005, "loss": 1.1669846773147583, "step": 5208 }, { "epoch": 0.6061205492203863, "grad_norm": 1.3910253047943115, "learning_rate": 0.0005, "loss": 2.3429477214813232, "step": 5209 }, { "epoch": 0.6062369094717245, "grad_norm": 1.4340721368789673, "learning_rate": 0.0005, "loss": 2.5851166248321533, "step": 5210 }, { "epoch": 0.6063532697230626, "grad_norm": 1.2564059495925903, "learning_rate": 0.0005, "loss": 2.2318115234375, "step": 5211 }, { "epoch": 0.6064696299744008, "grad_norm": 1.0559040307998657, "learning_rate": 0.0005, "loss": 1.9353563785552979, "step": 5212 }, { "epoch": 0.6065859902257389, "grad_norm": 1.243233323097229, "learning_rate": 0.0005, "loss": 3.234445333480835, "step": 5213 }, { "epoch": 0.606702350477077, "grad_norm": 0.9008641242980957, "learning_rate": 0.0005, "loss": 1.6344223022460938, "step": 5214 }, { "epoch": 0.6068187107284152, "grad_norm": 2.8459317684173584, "learning_rate": 0.0005, "loss": 2.959181785583496, "step": 5215 }, { "epoch": 0.6069350709797533, "grad_norm": 1.6725645065307617, "learning_rate": 0.0005, "loss": 3.27079176902771, "step": 5216 }, { "epoch": 0.6070514312310915, "grad_norm": 1.2226532697677612, "learning_rate": 0.0005, "loss": 2.621673583984375, "step": 5217 }, { "epoch": 0.6071677914824296, "grad_norm": 6.0427703857421875, "learning_rate": 0.0005, "loss": 2.4901137351989746, "step": 5218 }, { "epoch": 0.6072841517337677, "grad_norm": 1.4969794750213623, "learning_rate": 0.0005, "loss": 2.218449115753174, "step": 5219 }, { "epoch": 0.6074005119851059, "grad_norm": 1.4446555376052856, "learning_rate": 0.0005, "loss": 3.0430855751037598, "step": 5220 }, { "epoch": 0.607516872236444, "grad_norm": 0.8215150237083435, "learning_rate": 0.0005, "loss": 1.7268999814987183, "step": 5221 }, { "epoch": 0.6076332324877822, "grad_norm": 1.9783045053482056, "learning_rate": 0.0005, "loss": 2.812525749206543, "step": 5222 }, { "epoch": 0.6077495927391203, "grad_norm": 2.483097553253174, "learning_rate": 0.0005, "loss": 2.6956000328063965, "step": 5223 }, { "epoch": 0.6078659529904584, "grad_norm": 1.0986542701721191, "learning_rate": 0.0005, "loss": 3.315286636352539, "step": 5224 }, { "epoch": 0.6079823132417966, "grad_norm": 1.065183401107788, "learning_rate": 0.0005, "loss": 2.2782070636749268, "step": 5225 }, { "epoch": 0.6080986734931347, "grad_norm": 1.3753275871276855, "learning_rate": 0.0005, "loss": 2.8564743995666504, "step": 5226 }, { "epoch": 0.6082150337444728, "grad_norm": 1.7052563428878784, "learning_rate": 0.0005, "loss": 3.9230988025665283, "step": 5227 }, { "epoch": 0.608331393995811, "grad_norm": 1.5688972473144531, "learning_rate": 0.0005, "loss": 2.754883289337158, "step": 5228 }, { "epoch": 0.6084477542471491, "grad_norm": 2.0382256507873535, "learning_rate": 0.0005, "loss": 2.690640449523926, "step": 5229 }, { "epoch": 0.6085641144984874, "grad_norm": 1.6723573207855225, "learning_rate": 0.0005, "loss": 2.8904051780700684, "step": 5230 }, { "epoch": 0.6086804747498255, "grad_norm": 1.707916259765625, "learning_rate": 0.0005, "loss": 3.3955445289611816, "step": 5231 }, { "epoch": 0.6087968350011635, "grad_norm": 5.296551704406738, "learning_rate": 0.0005, "loss": 1.448533058166504, "step": 5232 }, { "epoch": 0.6089131952525018, "grad_norm": 1.7329515218734741, "learning_rate": 0.0005, "loss": 2.407646656036377, "step": 5233 }, { "epoch": 0.6090295555038399, "grad_norm": 1.2253268957138062, "learning_rate": 0.0005, "loss": 2.830103874206543, "step": 5234 }, { "epoch": 0.6091459157551781, "grad_norm": 1.6342580318450928, "learning_rate": 0.0005, "loss": 2.521653890609741, "step": 5235 }, { "epoch": 0.6092622760065162, "grad_norm": 1.3964024782180786, "learning_rate": 0.0005, "loss": 2.515718936920166, "step": 5236 }, { "epoch": 0.6093786362578543, "grad_norm": 1.6206413507461548, "learning_rate": 0.0005, "loss": 2.8591692447662354, "step": 5237 }, { "epoch": 0.6094949965091925, "grad_norm": 3.7448034286499023, "learning_rate": 0.0005, "loss": 2.969120740890503, "step": 5238 }, { "epoch": 0.6096113567605306, "grad_norm": 1.6275883913040161, "learning_rate": 0.0005, "loss": 2.497588872909546, "step": 5239 }, { "epoch": 0.6097277170118688, "grad_norm": 0.5788981914520264, "learning_rate": 0.0005, "loss": 2.0184807777404785, "step": 5240 }, { "epoch": 0.6098440772632069, "grad_norm": 1.80013108253479, "learning_rate": 0.0005, "loss": 3.2796218395233154, "step": 5241 }, { "epoch": 0.609960437514545, "grad_norm": 1.432180404663086, "learning_rate": 0.0005, "loss": 2.4737839698791504, "step": 5242 }, { "epoch": 0.6100767977658832, "grad_norm": 0.9168245792388916, "learning_rate": 0.0005, "loss": 2.471597909927368, "step": 5243 }, { "epoch": 0.6101931580172213, "grad_norm": 1.6827131509780884, "learning_rate": 0.0005, "loss": 2.91086483001709, "step": 5244 }, { "epoch": 0.6103095182685595, "grad_norm": 1.8704482316970825, "learning_rate": 0.0005, "loss": 2.7587738037109375, "step": 5245 }, { "epoch": 0.6104258785198976, "grad_norm": 1.3364454507827759, "learning_rate": 0.0005, "loss": 2.5255346298217773, "step": 5246 }, { "epoch": 0.6105422387712357, "grad_norm": 1.306249976158142, "learning_rate": 0.0005, "loss": 2.3421146869659424, "step": 5247 }, { "epoch": 0.6106585990225739, "grad_norm": 2.062415838241577, "learning_rate": 0.0005, "loss": 3.2812557220458984, "step": 5248 }, { "epoch": 0.610774959273912, "grad_norm": 1.1858261823654175, "learning_rate": 0.0005, "loss": 1.988518238067627, "step": 5249 }, { "epoch": 0.6108913195252502, "grad_norm": 1.8886786699295044, "learning_rate": 0.0005, "loss": 3.108872413635254, "step": 5250 }, { "epoch": 0.6110076797765883, "grad_norm": 1.4035848379135132, "learning_rate": 0.0005, "loss": 2.9094038009643555, "step": 5251 }, { "epoch": 0.6111240400279264, "grad_norm": 3.6342787742614746, "learning_rate": 0.0005, "loss": 2.7938578128814697, "step": 5252 }, { "epoch": 0.6112404002792646, "grad_norm": 1.4750075340270996, "learning_rate": 0.0005, "loss": 2.115145683288574, "step": 5253 }, { "epoch": 0.6113567605306027, "grad_norm": 3.8560848236083984, "learning_rate": 0.0005, "loss": 3.0487329959869385, "step": 5254 }, { "epoch": 0.6114731207819409, "grad_norm": 1.379031777381897, "learning_rate": 0.0005, "loss": 2.826061725616455, "step": 5255 }, { "epoch": 0.611589481033279, "grad_norm": 1.9704877138137817, "learning_rate": 0.0005, "loss": 1.8760603666305542, "step": 5256 }, { "epoch": 0.6117058412846171, "grad_norm": 1.0499753952026367, "learning_rate": 0.0005, "loss": 2.625805377960205, "step": 5257 }, { "epoch": 0.6118222015359553, "grad_norm": 2.2204062938690186, "learning_rate": 0.0005, "loss": 2.5332131385803223, "step": 5258 }, { "epoch": 0.6119385617872934, "grad_norm": 1.5021547079086304, "learning_rate": 0.0005, "loss": 1.9008101224899292, "step": 5259 }, { "epoch": 0.6120549220386317, "grad_norm": 0.6519118547439575, "learning_rate": 0.0005, "loss": 1.6875615119934082, "step": 5260 }, { "epoch": 0.6121712822899698, "grad_norm": 1.4249463081359863, "learning_rate": 0.0005, "loss": 2.876464366912842, "step": 5261 }, { "epoch": 0.6122876425413079, "grad_norm": 1.6739047765731812, "learning_rate": 0.0005, "loss": 3.209235668182373, "step": 5262 }, { "epoch": 0.6124040027926461, "grad_norm": 0.8171105980873108, "learning_rate": 0.0005, "loss": 1.578187108039856, "step": 5263 }, { "epoch": 0.6125203630439842, "grad_norm": 0.86603844165802, "learning_rate": 0.0005, "loss": 1.358441948890686, "step": 5264 }, { "epoch": 0.6126367232953224, "grad_norm": 0.5203656554222107, "learning_rate": 0.0005, "loss": 1.1228697299957275, "step": 5265 }, { "epoch": 0.6127530835466605, "grad_norm": 1.2673395872116089, "learning_rate": 0.0005, "loss": 3.121049404144287, "step": 5266 }, { "epoch": 0.6128694437979986, "grad_norm": 1.3844090700149536, "learning_rate": 0.0005, "loss": 3.0172836780548096, "step": 5267 }, { "epoch": 0.6129858040493368, "grad_norm": 0.9232192039489746, "learning_rate": 0.0005, "loss": 2.0456061363220215, "step": 5268 }, { "epoch": 0.6131021643006749, "grad_norm": 9.654717445373535, "learning_rate": 0.0005, "loss": 2.9546453952789307, "step": 5269 }, { "epoch": 0.613218524552013, "grad_norm": 0.800971269607544, "learning_rate": 0.0005, "loss": 1.886244773864746, "step": 5270 }, { "epoch": 0.6133348848033512, "grad_norm": 1.6506283283233643, "learning_rate": 0.0005, "loss": 2.305172920227051, "step": 5271 }, { "epoch": 0.6134512450546893, "grad_norm": 4.471978187561035, "learning_rate": 0.0005, "loss": 2.6563923358917236, "step": 5272 }, { "epoch": 0.6135676053060275, "grad_norm": 1.3601417541503906, "learning_rate": 0.0005, "loss": 2.1783604621887207, "step": 5273 }, { "epoch": 0.6136839655573656, "grad_norm": 1.1865193843841553, "learning_rate": 0.0005, "loss": 2.657214879989624, "step": 5274 }, { "epoch": 0.6138003258087037, "grad_norm": 0.9605000019073486, "learning_rate": 0.0005, "loss": 1.7428271770477295, "step": 5275 }, { "epoch": 0.6139166860600419, "grad_norm": 1.2911688089370728, "learning_rate": 0.0005, "loss": 1.9989027976989746, "step": 5276 }, { "epoch": 0.61403304631138, "grad_norm": 1.6344481706619263, "learning_rate": 0.0005, "loss": 3.412588357925415, "step": 5277 }, { "epoch": 0.6141494065627182, "grad_norm": 1.7385483980178833, "learning_rate": 0.0005, "loss": 2.9290859699249268, "step": 5278 }, { "epoch": 0.6142657668140563, "grad_norm": 1.1762820482254028, "learning_rate": 0.0005, "loss": 2.141111135482788, "step": 5279 }, { "epoch": 0.6143821270653944, "grad_norm": 1.8503022193908691, "learning_rate": 0.0005, "loss": 2.8946542739868164, "step": 5280 }, { "epoch": 0.6144984873167326, "grad_norm": 1.5836780071258545, "learning_rate": 0.0005, "loss": 2.6634774208068848, "step": 5281 }, { "epoch": 0.6146148475680707, "grad_norm": 1.2386016845703125, "learning_rate": 0.0005, "loss": 2.594003438949585, "step": 5282 }, { "epoch": 0.6147312078194089, "grad_norm": 1.2293461561203003, "learning_rate": 0.0005, "loss": 2.5058236122131348, "step": 5283 }, { "epoch": 0.614847568070747, "grad_norm": 1.4496899843215942, "learning_rate": 0.0005, "loss": 2.1195616722106934, "step": 5284 }, { "epoch": 0.6149639283220851, "grad_norm": 3.0264477729797363, "learning_rate": 0.0005, "loss": 3.1983754634857178, "step": 5285 }, { "epoch": 0.6150802885734233, "grad_norm": 1.5007191896438599, "learning_rate": 0.0005, "loss": 1.8477258682250977, "step": 5286 }, { "epoch": 0.6151966488247614, "grad_norm": 1.232607364654541, "learning_rate": 0.0005, "loss": 2.5199155807495117, "step": 5287 }, { "epoch": 0.6153130090760996, "grad_norm": 1.2209645509719849, "learning_rate": 0.0005, "loss": 3.0510501861572266, "step": 5288 }, { "epoch": 0.6154293693274377, "grad_norm": 1.2269104719161987, "learning_rate": 0.0005, "loss": 1.9746164083480835, "step": 5289 }, { "epoch": 0.6155457295787758, "grad_norm": 1.7267402410507202, "learning_rate": 0.0005, "loss": 2.7370963096618652, "step": 5290 }, { "epoch": 0.6156620898301141, "grad_norm": 1.3672682046890259, "learning_rate": 0.0005, "loss": 2.542611598968506, "step": 5291 }, { "epoch": 0.6157784500814522, "grad_norm": 1.5537338256835938, "learning_rate": 0.0005, "loss": 1.7342839241027832, "step": 5292 }, { "epoch": 0.6158948103327904, "grad_norm": 1.9146873950958252, "learning_rate": 0.0005, "loss": 1.8477046489715576, "step": 5293 }, { "epoch": 0.6160111705841285, "grad_norm": 1.4541550874710083, "learning_rate": 0.0005, "loss": 2.1815004348754883, "step": 5294 }, { "epoch": 0.6161275308354666, "grad_norm": 1.5404025316238403, "learning_rate": 0.0005, "loss": 2.398552656173706, "step": 5295 }, { "epoch": 0.6162438910868048, "grad_norm": 1.9004243612289429, "learning_rate": 0.0005, "loss": 2.334563970565796, "step": 5296 }, { "epoch": 0.6163602513381429, "grad_norm": 1.288313388824463, "learning_rate": 0.0005, "loss": 2.456362724304199, "step": 5297 }, { "epoch": 0.6164766115894811, "grad_norm": 1.509105920791626, "learning_rate": 0.0005, "loss": 2.267759084701538, "step": 5298 }, { "epoch": 0.6165929718408192, "grad_norm": 1.4242560863494873, "learning_rate": 0.0005, "loss": 1.7569398880004883, "step": 5299 }, { "epoch": 0.6167093320921573, "grad_norm": 1.9810552597045898, "learning_rate": 0.0005, "loss": 1.9499508142471313, "step": 5300 }, { "epoch": 0.6168256923434955, "grad_norm": 0.5770711898803711, "learning_rate": 0.0005, "loss": 1.7961726188659668, "step": 5301 }, { "epoch": 0.6169420525948336, "grad_norm": 3.921043634414673, "learning_rate": 0.0005, "loss": 2.595071315765381, "step": 5302 }, { "epoch": 0.6170584128461718, "grad_norm": 1.2552266120910645, "learning_rate": 0.0005, "loss": 2.5771050453186035, "step": 5303 }, { "epoch": 0.6171747730975099, "grad_norm": 1.382308840751648, "learning_rate": 0.0005, "loss": 3.7147021293640137, "step": 5304 }, { "epoch": 0.617291133348848, "grad_norm": 1.348252773284912, "learning_rate": 0.0005, "loss": 2.9940481185913086, "step": 5305 }, { "epoch": 0.6174074936001862, "grad_norm": 0.8775129318237305, "learning_rate": 0.0005, "loss": 1.889519214630127, "step": 5306 }, { "epoch": 0.6175238538515243, "grad_norm": 1.2305539846420288, "learning_rate": 0.0005, "loss": 2.4011054039001465, "step": 5307 }, { "epoch": 0.6176402141028625, "grad_norm": 1.6554173231124878, "learning_rate": 0.0005, "loss": 2.8048384189605713, "step": 5308 }, { "epoch": 0.6177565743542006, "grad_norm": 1.0209856033325195, "learning_rate": 0.0005, "loss": 1.4987951517105103, "step": 5309 }, { "epoch": 0.6178729346055387, "grad_norm": 1.652217149734497, "learning_rate": 0.0005, "loss": 2.9289352893829346, "step": 5310 }, { "epoch": 0.6179892948568769, "grad_norm": 1.0700031518936157, "learning_rate": 0.0005, "loss": 1.7832980155944824, "step": 5311 }, { "epoch": 0.618105655108215, "grad_norm": 2.0653703212738037, "learning_rate": 0.0005, "loss": 2.7345972061157227, "step": 5312 }, { "epoch": 0.6182220153595531, "grad_norm": 1.3724251985549927, "learning_rate": 0.0005, "loss": 2.058830976486206, "step": 5313 }, { "epoch": 0.6183383756108913, "grad_norm": 1.4176279306411743, "learning_rate": 0.0005, "loss": 2.182858943939209, "step": 5314 }, { "epoch": 0.6184547358622294, "grad_norm": 1.2067744731903076, "learning_rate": 0.0005, "loss": 2.4249627590179443, "step": 5315 }, { "epoch": 0.6185710961135676, "grad_norm": 1.4271585941314697, "learning_rate": 0.0005, "loss": 2.967941999435425, "step": 5316 }, { "epoch": 0.6186874563649057, "grad_norm": 1.4700309038162231, "learning_rate": 0.0005, "loss": 2.1077263355255127, "step": 5317 }, { "epoch": 0.6188038166162438, "grad_norm": 1.330775260925293, "learning_rate": 0.0005, "loss": 2.4736647605895996, "step": 5318 }, { "epoch": 0.618920176867582, "grad_norm": 1.1779768466949463, "learning_rate": 0.0005, "loss": 2.292316436767578, "step": 5319 }, { "epoch": 0.6190365371189201, "grad_norm": 1.0776690244674683, "learning_rate": 0.0005, "loss": 3.036881446838379, "step": 5320 }, { "epoch": 0.6191528973702584, "grad_norm": 1.954795241355896, "learning_rate": 0.0005, "loss": 2.535712718963623, "step": 5321 }, { "epoch": 0.6192692576215965, "grad_norm": 1.1292097568511963, "learning_rate": 0.0005, "loss": 2.0062451362609863, "step": 5322 }, { "epoch": 0.6193856178729346, "grad_norm": 1.5376098155975342, "learning_rate": 0.0005, "loss": 3.1183815002441406, "step": 5323 }, { "epoch": 0.6195019781242728, "grad_norm": 1.7092353105545044, "learning_rate": 0.0005, "loss": 2.8610517978668213, "step": 5324 }, { "epoch": 0.6196183383756109, "grad_norm": 1.3123958110809326, "learning_rate": 0.0005, "loss": 2.558438777923584, "step": 5325 }, { "epoch": 0.6197346986269491, "grad_norm": 1.3916703462600708, "learning_rate": 0.0005, "loss": 2.7101640701293945, "step": 5326 }, { "epoch": 0.6198510588782872, "grad_norm": 1.6244062185287476, "learning_rate": 0.0005, "loss": 2.4639220237731934, "step": 5327 }, { "epoch": 0.6199674191296253, "grad_norm": 1.6791664361953735, "learning_rate": 0.0005, "loss": 2.989579200744629, "step": 5328 }, { "epoch": 0.6200837793809635, "grad_norm": 0.841510534286499, "learning_rate": 0.0005, "loss": 1.4614756107330322, "step": 5329 }, { "epoch": 0.6202001396323016, "grad_norm": 2.7648820877075195, "learning_rate": 0.0005, "loss": 3.108748435974121, "step": 5330 }, { "epoch": 0.6203164998836398, "grad_norm": 1.7593051195144653, "learning_rate": 0.0005, "loss": 2.0547916889190674, "step": 5331 }, { "epoch": 0.6204328601349779, "grad_norm": 1.6879924535751343, "learning_rate": 0.0005, "loss": 2.7232892513275146, "step": 5332 }, { "epoch": 0.620549220386316, "grad_norm": 1.5041217803955078, "learning_rate": 0.0005, "loss": 2.9976842403411865, "step": 5333 }, { "epoch": 0.6206655806376542, "grad_norm": 1.7140053510665894, "learning_rate": 0.0005, "loss": 2.439809560775757, "step": 5334 }, { "epoch": 0.6207819408889923, "grad_norm": 2.440791606903076, "learning_rate": 0.0005, "loss": 3.0342295169830322, "step": 5335 }, { "epoch": 0.6208983011403305, "grad_norm": 1.1126720905303955, "learning_rate": 0.0005, "loss": 1.481452226638794, "step": 5336 }, { "epoch": 0.6210146613916686, "grad_norm": 2.19067120552063, "learning_rate": 0.0005, "loss": 2.5260729789733887, "step": 5337 }, { "epoch": 0.6211310216430067, "grad_norm": 1.1310224533081055, "learning_rate": 0.0005, "loss": 1.8932127952575684, "step": 5338 }, { "epoch": 0.6212473818943449, "grad_norm": 1.4343364238739014, "learning_rate": 0.0005, "loss": 3.577232837677002, "step": 5339 }, { "epoch": 0.621363742145683, "grad_norm": 0.9807645082473755, "learning_rate": 0.0005, "loss": 1.6243228912353516, "step": 5340 }, { "epoch": 0.6214801023970212, "grad_norm": 1.7419676780700684, "learning_rate": 0.0005, "loss": 2.7483267784118652, "step": 5341 }, { "epoch": 0.6215964626483593, "grad_norm": 1.5138227939605713, "learning_rate": 0.0005, "loss": 3.3821845054626465, "step": 5342 }, { "epoch": 0.6217128228996974, "grad_norm": 1.5726256370544434, "learning_rate": 0.0005, "loss": 2.8864705562591553, "step": 5343 }, { "epoch": 0.6218291831510356, "grad_norm": 2.2593436241149902, "learning_rate": 0.0005, "loss": 2.673086643218994, "step": 5344 }, { "epoch": 0.6219455434023737, "grad_norm": 0.597499668598175, "learning_rate": 0.0005, "loss": 1.5867819786071777, "step": 5345 }, { "epoch": 0.622061903653712, "grad_norm": 1.257448434829712, "learning_rate": 0.0005, "loss": 2.142279863357544, "step": 5346 }, { "epoch": 0.62217826390505, "grad_norm": 1.5536563396453857, "learning_rate": 0.0005, "loss": 2.415231943130493, "step": 5347 }, { "epoch": 0.6222946241563881, "grad_norm": 0.6918880343437195, "learning_rate": 0.0005, "loss": 1.2552522420883179, "step": 5348 }, { "epoch": 0.6224109844077264, "grad_norm": 0.9340885877609253, "learning_rate": 0.0005, "loss": 2.0757689476013184, "step": 5349 }, { "epoch": 0.6225273446590645, "grad_norm": 1.5397212505340576, "learning_rate": 0.0005, "loss": 2.7977445125579834, "step": 5350 }, { "epoch": 0.6226437049104026, "grad_norm": 1.3888609409332275, "learning_rate": 0.0005, "loss": 2.2711315155029297, "step": 5351 }, { "epoch": 0.6227600651617408, "grad_norm": 1.5047460794448853, "learning_rate": 0.0005, "loss": 3.2323591709136963, "step": 5352 }, { "epoch": 0.6228764254130789, "grad_norm": 1.114488959312439, "learning_rate": 0.0005, "loss": 2.877295732498169, "step": 5353 }, { "epoch": 0.6229927856644171, "grad_norm": 4.897356033325195, "learning_rate": 0.0005, "loss": 2.560955047607422, "step": 5354 }, { "epoch": 0.6231091459157552, "grad_norm": 0.9870287775993347, "learning_rate": 0.0005, "loss": 2.403343677520752, "step": 5355 }, { "epoch": 0.6232255061670933, "grad_norm": 1.95140540599823, "learning_rate": 0.0005, "loss": 3.3149824142456055, "step": 5356 }, { "epoch": 0.6233418664184315, "grad_norm": 1.4690890312194824, "learning_rate": 0.0005, "loss": 2.6616992950439453, "step": 5357 }, { "epoch": 0.6234582266697696, "grad_norm": 1.4521533250808716, "learning_rate": 0.0005, "loss": 2.45041561126709, "step": 5358 }, { "epoch": 0.6235745869211078, "grad_norm": 1.667739748954773, "learning_rate": 0.0005, "loss": 2.3554844856262207, "step": 5359 }, { "epoch": 0.6236909471724459, "grad_norm": 1.3470280170440674, "learning_rate": 0.0005, "loss": 2.4683420658111572, "step": 5360 }, { "epoch": 0.623807307423784, "grad_norm": 1.2201565504074097, "learning_rate": 0.0005, "loss": 2.4450008869171143, "step": 5361 }, { "epoch": 0.6239236676751222, "grad_norm": 1.2794480323791504, "learning_rate": 0.0005, "loss": 2.2684876918792725, "step": 5362 }, { "epoch": 0.6240400279264603, "grad_norm": 1.1652661561965942, "learning_rate": 0.0005, "loss": 2.735715866088867, "step": 5363 }, { "epoch": 0.6241563881777985, "grad_norm": 1.5949180126190186, "learning_rate": 0.0005, "loss": 3.3189210891723633, "step": 5364 }, { "epoch": 0.6242727484291366, "grad_norm": 2.0873184204101562, "learning_rate": 0.0005, "loss": 2.9415431022644043, "step": 5365 }, { "epoch": 0.6243891086804747, "grad_norm": 1.2308382987976074, "learning_rate": 0.0005, "loss": 2.521322250366211, "step": 5366 }, { "epoch": 0.6245054689318129, "grad_norm": 1.2319703102111816, "learning_rate": 0.0005, "loss": 2.896510601043701, "step": 5367 }, { "epoch": 0.624621829183151, "grad_norm": 1.1204620599746704, "learning_rate": 0.0005, "loss": 1.950224757194519, "step": 5368 }, { "epoch": 0.6247381894344892, "grad_norm": 1.3614693880081177, "learning_rate": 0.0005, "loss": 3.846735715866089, "step": 5369 }, { "epoch": 0.6248545496858273, "grad_norm": 1.0930753946304321, "learning_rate": 0.0005, "loss": 1.729522466659546, "step": 5370 }, { "epoch": 0.6249709099371654, "grad_norm": 2.874504327774048, "learning_rate": 0.0005, "loss": 2.112518548965454, "step": 5371 }, { "epoch": 0.6250872701885036, "grad_norm": 5.248315811157227, "learning_rate": 0.0005, "loss": 2.130772590637207, "step": 5372 }, { "epoch": 0.6252036304398417, "grad_norm": 1.4010642766952515, "learning_rate": 0.0005, "loss": 2.3932838439941406, "step": 5373 }, { "epoch": 0.6253199906911799, "grad_norm": 1.3949087858200073, "learning_rate": 0.0005, "loss": 2.418799877166748, "step": 5374 }, { "epoch": 0.625436350942518, "grad_norm": 1.7356928586959839, "learning_rate": 0.0005, "loss": 3.1385111808776855, "step": 5375 }, { "epoch": 0.6255527111938561, "grad_norm": 1.2332154512405396, "learning_rate": 0.0005, "loss": 3.5852761268615723, "step": 5376 }, { "epoch": 0.6256690714451943, "grad_norm": 1.6192249059677124, "learning_rate": 0.0005, "loss": 2.0604891777038574, "step": 5377 }, { "epoch": 0.6257854316965324, "grad_norm": 1.139983057975769, "learning_rate": 0.0005, "loss": 3.1344921588897705, "step": 5378 }, { "epoch": 0.6259017919478707, "grad_norm": 1.799254059791565, "learning_rate": 0.0005, "loss": 2.5380754470825195, "step": 5379 }, { "epoch": 0.6260181521992088, "grad_norm": 0.8958700895309448, "learning_rate": 0.0005, "loss": 2.424804210662842, "step": 5380 }, { "epoch": 0.6261345124505469, "grad_norm": 1.3052349090576172, "learning_rate": 0.0005, "loss": 2.7790751457214355, "step": 5381 }, { "epoch": 0.6262508727018851, "grad_norm": 0.9012979865074158, "learning_rate": 0.0005, "loss": 2.2533552646636963, "step": 5382 }, { "epoch": 0.6263672329532232, "grad_norm": 1.2972609996795654, "learning_rate": 0.0005, "loss": 2.074383020401001, "step": 5383 }, { "epoch": 0.6264835932045614, "grad_norm": 0.6304476857185364, "learning_rate": 0.0005, "loss": 1.6583843231201172, "step": 5384 }, { "epoch": 0.6265999534558995, "grad_norm": 1.0269663333892822, "learning_rate": 0.0005, "loss": 2.113935947418213, "step": 5385 }, { "epoch": 0.6267163137072376, "grad_norm": 2.1654272079467773, "learning_rate": 0.0005, "loss": 3.0487594604492188, "step": 5386 }, { "epoch": 0.6268326739585758, "grad_norm": 1.0617663860321045, "learning_rate": 0.0005, "loss": 2.039759635925293, "step": 5387 }, { "epoch": 0.6269490342099139, "grad_norm": 1.825823426246643, "learning_rate": 0.0005, "loss": 1.4844043254852295, "step": 5388 }, { "epoch": 0.6270653944612521, "grad_norm": 1.5450621843338013, "learning_rate": 0.0005, "loss": 2.2480320930480957, "step": 5389 }, { "epoch": 0.6271817547125902, "grad_norm": 1.8583258390426636, "learning_rate": 0.0005, "loss": 2.819993257522583, "step": 5390 }, { "epoch": 0.6272981149639283, "grad_norm": 2.1484615802764893, "learning_rate": 0.0005, "loss": 2.262815475463867, "step": 5391 }, { "epoch": 0.6274144752152665, "grad_norm": 1.4797896146774292, "learning_rate": 0.0005, "loss": 2.254258155822754, "step": 5392 }, { "epoch": 0.6275308354666046, "grad_norm": 1.9330984354019165, "learning_rate": 0.0005, "loss": 1.8133201599121094, "step": 5393 }, { "epoch": 0.6276471957179427, "grad_norm": 1.2776557207107544, "learning_rate": 0.0005, "loss": 2.3507909774780273, "step": 5394 }, { "epoch": 0.6277635559692809, "grad_norm": 1.4337457418441772, "learning_rate": 0.0005, "loss": 2.2285680770874023, "step": 5395 }, { "epoch": 0.627879916220619, "grad_norm": 1.4049469232559204, "learning_rate": 0.0005, "loss": 1.204460620880127, "step": 5396 }, { "epoch": 0.6279962764719572, "grad_norm": 2.4895994663238525, "learning_rate": 0.0005, "loss": 2.298177719116211, "step": 5397 }, { "epoch": 0.6281126367232953, "grad_norm": 1.6714824438095093, "learning_rate": 0.0005, "loss": 2.4408349990844727, "step": 5398 }, { "epoch": 0.6282289969746334, "grad_norm": 1.3147624731063843, "learning_rate": 0.0005, "loss": 2.877305030822754, "step": 5399 }, { "epoch": 0.6283453572259716, "grad_norm": 1.2782235145568848, "learning_rate": 0.0005, "loss": 2.0586304664611816, "step": 5400 }, { "epoch": 0.6284617174773097, "grad_norm": 1.0839107036590576, "learning_rate": 0.0005, "loss": 2.22206449508667, "step": 5401 }, { "epoch": 0.6285780777286479, "grad_norm": 1.341463565826416, "learning_rate": 0.0005, "loss": 2.4543612003326416, "step": 5402 }, { "epoch": 0.628694437979986, "grad_norm": 1.5129022598266602, "learning_rate": 0.0005, "loss": 2.8005571365356445, "step": 5403 }, { "epoch": 0.6288107982313241, "grad_norm": 0.8266580104827881, "learning_rate": 0.0005, "loss": 1.5953648090362549, "step": 5404 }, { "epoch": 0.6289271584826623, "grad_norm": 2.213916063308716, "learning_rate": 0.0005, "loss": 2.1808834075927734, "step": 5405 }, { "epoch": 0.6290435187340004, "grad_norm": 2.161968469619751, "learning_rate": 0.0005, "loss": 3.2361490726470947, "step": 5406 }, { "epoch": 0.6291598789853386, "grad_norm": 1.5317140817642212, "learning_rate": 0.0005, "loss": 2.3122668266296387, "step": 5407 }, { "epoch": 0.6292762392366767, "grad_norm": 1.3881347179412842, "learning_rate": 0.0005, "loss": 2.1866230964660645, "step": 5408 }, { "epoch": 0.6293925994880148, "grad_norm": 1.284340739250183, "learning_rate": 0.0005, "loss": 2.6188607215881348, "step": 5409 }, { "epoch": 0.6295089597393531, "grad_norm": 1.5053927898406982, "learning_rate": 0.0005, "loss": 3.576383590698242, "step": 5410 }, { "epoch": 0.6296253199906912, "grad_norm": 1.0612058639526367, "learning_rate": 0.0005, "loss": 2.1248843669891357, "step": 5411 }, { "epoch": 0.6297416802420294, "grad_norm": 1.4051803350448608, "learning_rate": 0.0005, "loss": 2.207024097442627, "step": 5412 }, { "epoch": 0.6298580404933675, "grad_norm": 1.646437406539917, "learning_rate": 0.0005, "loss": 2.259505271911621, "step": 5413 }, { "epoch": 0.6299744007447056, "grad_norm": 1.344152808189392, "learning_rate": 0.0005, "loss": 2.0721054077148438, "step": 5414 }, { "epoch": 0.6300907609960438, "grad_norm": 1.337250828742981, "learning_rate": 0.0005, "loss": 2.795590877532959, "step": 5415 }, { "epoch": 0.6302071212473819, "grad_norm": 1.1409014463424683, "learning_rate": 0.0005, "loss": 2.35129451751709, "step": 5416 }, { "epoch": 0.6303234814987201, "grad_norm": 1.3203755617141724, "learning_rate": 0.0005, "loss": 1.6704816818237305, "step": 5417 }, { "epoch": 0.6304398417500582, "grad_norm": 0.7926871180534363, "learning_rate": 0.0005, "loss": 1.9069095849990845, "step": 5418 }, { "epoch": 0.6305562020013963, "grad_norm": 1.4277501106262207, "learning_rate": 0.0005, "loss": 2.340785503387451, "step": 5419 }, { "epoch": 0.6306725622527345, "grad_norm": 1.5067076683044434, "learning_rate": 0.0005, "loss": 2.009988784790039, "step": 5420 }, { "epoch": 0.6307889225040726, "grad_norm": 1.160112977027893, "learning_rate": 0.0005, "loss": 1.693293571472168, "step": 5421 }, { "epoch": 0.6309052827554108, "grad_norm": 1.2260123491287231, "learning_rate": 0.0005, "loss": 2.938692569732666, "step": 5422 }, { "epoch": 0.6310216430067489, "grad_norm": 1.0901216268539429, "learning_rate": 0.0005, "loss": 2.507643222808838, "step": 5423 }, { "epoch": 0.631138003258087, "grad_norm": 1.3419755697250366, "learning_rate": 0.0005, "loss": 2.3818299770355225, "step": 5424 }, { "epoch": 0.6312543635094252, "grad_norm": 1.2695049047470093, "learning_rate": 0.0005, "loss": 3.1774654388427734, "step": 5425 }, { "epoch": 0.6313707237607633, "grad_norm": 1.8366730213165283, "learning_rate": 0.0005, "loss": 1.9544228315353394, "step": 5426 }, { "epoch": 0.6314870840121015, "grad_norm": 2.0092504024505615, "learning_rate": 0.0005, "loss": 2.562798261642456, "step": 5427 }, { "epoch": 0.6316034442634396, "grad_norm": 1.570718765258789, "learning_rate": 0.0005, "loss": 2.929607391357422, "step": 5428 }, { "epoch": 0.6317198045147777, "grad_norm": 2.0780739784240723, "learning_rate": 0.0005, "loss": 2.4371535778045654, "step": 5429 }, { "epoch": 0.6318361647661159, "grad_norm": 1.0199909210205078, "learning_rate": 0.0005, "loss": 2.105681896209717, "step": 5430 }, { "epoch": 0.631952525017454, "grad_norm": 1.1084750890731812, "learning_rate": 0.0005, "loss": 2.5612311363220215, "step": 5431 }, { "epoch": 0.6320688852687922, "grad_norm": 1.1708623170852661, "learning_rate": 0.0005, "loss": 1.9048068523406982, "step": 5432 }, { "epoch": 0.6321852455201303, "grad_norm": 1.1956225633621216, "learning_rate": 0.0005, "loss": 2.7833454608917236, "step": 5433 }, { "epoch": 0.6323016057714684, "grad_norm": 1.4885585308074951, "learning_rate": 0.0005, "loss": 2.9302046298980713, "step": 5434 }, { "epoch": 0.6324179660228066, "grad_norm": 1.6131162643432617, "learning_rate": 0.0005, "loss": 2.0886576175689697, "step": 5435 }, { "epoch": 0.6325343262741447, "grad_norm": 1.8681529760360718, "learning_rate": 0.0005, "loss": 3.096536636352539, "step": 5436 }, { "epoch": 0.6326506865254828, "grad_norm": 0.885040819644928, "learning_rate": 0.0005, "loss": 2.183475971221924, "step": 5437 }, { "epoch": 0.632767046776821, "grad_norm": 1.1914259195327759, "learning_rate": 0.0005, "loss": 2.526925563812256, "step": 5438 }, { "epoch": 0.6328834070281592, "grad_norm": 2.1324827671051025, "learning_rate": 0.0005, "loss": 2.62294864654541, "step": 5439 }, { "epoch": 0.6329997672794974, "grad_norm": 1.28070867061615, "learning_rate": 0.0005, "loss": 2.0633695125579834, "step": 5440 }, { "epoch": 0.6331161275308355, "grad_norm": 1.378190279006958, "learning_rate": 0.0005, "loss": 2.515725612640381, "step": 5441 }, { "epoch": 0.6332324877821736, "grad_norm": 2.0792837142944336, "learning_rate": 0.0005, "loss": 3.1140635013580322, "step": 5442 }, { "epoch": 0.6333488480335118, "grad_norm": 2.1470766067504883, "learning_rate": 0.0005, "loss": 2.8081729412078857, "step": 5443 }, { "epoch": 0.6334652082848499, "grad_norm": 1.0734959840774536, "learning_rate": 0.0005, "loss": 2.3764877319335938, "step": 5444 }, { "epoch": 0.6335815685361881, "grad_norm": 1.1487302780151367, "learning_rate": 0.0005, "loss": 2.6464667320251465, "step": 5445 }, { "epoch": 0.6336979287875262, "grad_norm": 1.2341057062149048, "learning_rate": 0.0005, "loss": 2.203106641769409, "step": 5446 }, { "epoch": 0.6338142890388643, "grad_norm": 2.7656302452087402, "learning_rate": 0.0005, "loss": 3.5202417373657227, "step": 5447 }, { "epoch": 0.6339306492902025, "grad_norm": 0.9464788436889648, "learning_rate": 0.0005, "loss": 1.4924671649932861, "step": 5448 }, { "epoch": 0.6340470095415406, "grad_norm": 1.5450561046600342, "learning_rate": 0.0005, "loss": 2.873889446258545, "step": 5449 }, { "epoch": 0.6341633697928788, "grad_norm": 1.819283366203308, "learning_rate": 0.0005, "loss": 2.615805149078369, "step": 5450 }, { "epoch": 0.6342797300442169, "grad_norm": 1.485187292098999, "learning_rate": 0.0005, "loss": 2.9621376991271973, "step": 5451 }, { "epoch": 0.634396090295555, "grad_norm": 1.0038560628890991, "learning_rate": 0.0005, "loss": 2.818443775177002, "step": 5452 }, { "epoch": 0.6345124505468932, "grad_norm": 1.7000319957733154, "learning_rate": 0.0005, "loss": 3.0773634910583496, "step": 5453 }, { "epoch": 0.6346288107982313, "grad_norm": 1.1255650520324707, "learning_rate": 0.0005, "loss": 1.8520500659942627, "step": 5454 }, { "epoch": 0.6347451710495695, "grad_norm": 1.2436275482177734, "learning_rate": 0.0005, "loss": 2.6239218711853027, "step": 5455 }, { "epoch": 0.6348615313009076, "grad_norm": 1.0549730062484741, "learning_rate": 0.0005, "loss": 1.9497560262680054, "step": 5456 }, { "epoch": 0.6349778915522457, "grad_norm": 1.7664998769760132, "learning_rate": 0.0005, "loss": 2.5310821533203125, "step": 5457 }, { "epoch": 0.6350942518035839, "grad_norm": 1.3169333934783936, "learning_rate": 0.0005, "loss": 2.069561243057251, "step": 5458 }, { "epoch": 0.635210612054922, "grad_norm": 1.1783640384674072, "learning_rate": 0.0005, "loss": 3.4822373390197754, "step": 5459 }, { "epoch": 0.6353269723062602, "grad_norm": 1.886864423751831, "learning_rate": 0.0005, "loss": 2.68096661567688, "step": 5460 }, { "epoch": 0.6354433325575983, "grad_norm": 0.9343472719192505, "learning_rate": 0.0005, "loss": 2.2435944080352783, "step": 5461 }, { "epoch": 0.6355596928089364, "grad_norm": 1.1028385162353516, "learning_rate": 0.0005, "loss": 2.3563613891601562, "step": 5462 }, { "epoch": 0.6356760530602746, "grad_norm": 0.8444628119468689, "learning_rate": 0.0005, "loss": 2.092012643814087, "step": 5463 }, { "epoch": 0.6357924133116127, "grad_norm": 1.632556438446045, "learning_rate": 0.0005, "loss": 1.5690786838531494, "step": 5464 }, { "epoch": 0.635908773562951, "grad_norm": 1.625706672668457, "learning_rate": 0.0005, "loss": 2.596437454223633, "step": 5465 }, { "epoch": 0.636025133814289, "grad_norm": 1.2379117012023926, "learning_rate": 0.0005, "loss": 3.3788163661956787, "step": 5466 }, { "epoch": 0.6361414940656271, "grad_norm": 1.114140272140503, "learning_rate": 0.0005, "loss": 2.288041830062866, "step": 5467 }, { "epoch": 0.6362578543169654, "grad_norm": 1.945538878440857, "learning_rate": 0.0005, "loss": 3.4635934829711914, "step": 5468 }, { "epoch": 0.6363742145683035, "grad_norm": 3.6667864322662354, "learning_rate": 0.0005, "loss": 3.765352725982666, "step": 5469 }, { "epoch": 0.6364905748196417, "grad_norm": 1.3198049068450928, "learning_rate": 0.0005, "loss": 3.164614677429199, "step": 5470 }, { "epoch": 0.6366069350709798, "grad_norm": 1.4433565139770508, "learning_rate": 0.0005, "loss": 2.8009750843048096, "step": 5471 }, { "epoch": 0.6367232953223179, "grad_norm": 1.1550955772399902, "learning_rate": 0.0005, "loss": 2.4600579738616943, "step": 5472 }, { "epoch": 0.6368396555736561, "grad_norm": 1.302188754081726, "learning_rate": 0.0005, "loss": 2.3945631980895996, "step": 5473 }, { "epoch": 0.6369560158249942, "grad_norm": 1.4428725242614746, "learning_rate": 0.0005, "loss": 3.3069100379943848, "step": 5474 }, { "epoch": 0.6370723760763323, "grad_norm": 1.3783235549926758, "learning_rate": 0.0005, "loss": 3.2795419692993164, "step": 5475 }, { "epoch": 0.6371887363276705, "grad_norm": 1.0291111469268799, "learning_rate": 0.0005, "loss": 2.2037572860717773, "step": 5476 }, { "epoch": 0.6373050965790086, "grad_norm": 1.070785403251648, "learning_rate": 0.0005, "loss": 2.395562171936035, "step": 5477 }, { "epoch": 0.6374214568303468, "grad_norm": 1.161563515663147, "learning_rate": 0.0005, "loss": 2.2167587280273438, "step": 5478 }, { "epoch": 0.6375378170816849, "grad_norm": 0.8779720067977905, "learning_rate": 0.0005, "loss": 1.1073228120803833, "step": 5479 }, { "epoch": 0.637654177333023, "grad_norm": 0.9097484946250916, "learning_rate": 0.0005, "loss": 2.295854330062866, "step": 5480 }, { "epoch": 0.6377705375843612, "grad_norm": 1.6816214323043823, "learning_rate": 0.0005, "loss": 2.2294843196868896, "step": 5481 }, { "epoch": 0.6378868978356993, "grad_norm": 1.4592560529708862, "learning_rate": 0.0005, "loss": 1.9789834022521973, "step": 5482 }, { "epoch": 0.6380032580870375, "grad_norm": 1.3654992580413818, "learning_rate": 0.0005, "loss": 2.7708897590637207, "step": 5483 }, { "epoch": 0.6381196183383756, "grad_norm": 1.6425316333770752, "learning_rate": 0.0005, "loss": 2.8925700187683105, "step": 5484 }, { "epoch": 0.6382359785897137, "grad_norm": 1.211214542388916, "learning_rate": 0.0005, "loss": 2.42625093460083, "step": 5485 }, { "epoch": 0.6383523388410519, "grad_norm": 1.4542173147201538, "learning_rate": 0.0005, "loss": 2.304443359375, "step": 5486 }, { "epoch": 0.63846869909239, "grad_norm": 0.999933123588562, "learning_rate": 0.0005, "loss": 1.9527487754821777, "step": 5487 }, { "epoch": 0.6385850593437282, "grad_norm": 1.2825465202331543, "learning_rate": 0.0005, "loss": 1.7758703231811523, "step": 5488 }, { "epoch": 0.6387014195950663, "grad_norm": 2.1102287769317627, "learning_rate": 0.0005, "loss": 2.0855655670166016, "step": 5489 }, { "epoch": 0.6388177798464044, "grad_norm": 1.7667145729064941, "learning_rate": 0.0005, "loss": 2.0990264415740967, "step": 5490 }, { "epoch": 0.6389341400977426, "grad_norm": 1.6271870136260986, "learning_rate": 0.0005, "loss": 2.6789391040802, "step": 5491 }, { "epoch": 0.6390505003490807, "grad_norm": 1.182381272315979, "learning_rate": 0.0005, "loss": 2.4623074531555176, "step": 5492 }, { "epoch": 0.6391668606004189, "grad_norm": 1.045552372932434, "learning_rate": 0.0005, "loss": 1.2350287437438965, "step": 5493 }, { "epoch": 0.639283220851757, "grad_norm": 1.076607584953308, "learning_rate": 0.0005, "loss": 1.7626500129699707, "step": 5494 }, { "epoch": 0.6393995811030951, "grad_norm": 0.692025899887085, "learning_rate": 0.0005, "loss": 1.371267318725586, "step": 5495 }, { "epoch": 0.6395159413544333, "grad_norm": 1.3307139873504639, "learning_rate": 0.0005, "loss": 1.8603646755218506, "step": 5496 }, { "epoch": 0.6396323016057714, "grad_norm": 2.049118995666504, "learning_rate": 0.0005, "loss": 2.353472948074341, "step": 5497 }, { "epoch": 0.6397486618571097, "grad_norm": 2.045945644378662, "learning_rate": 0.0005, "loss": 2.5677318572998047, "step": 5498 }, { "epoch": 0.6398650221084478, "grad_norm": 0.61854487657547, "learning_rate": 0.0005, "loss": 1.6986167430877686, "step": 5499 }, { "epoch": 0.6399813823597859, "grad_norm": 1.490334153175354, "learning_rate": 0.0005, "loss": 3.1206586360931396, "step": 5500 }, { "epoch": 0.6400977426111241, "grad_norm": 1.090012788772583, "learning_rate": 0.0005, "loss": 2.9214539527893066, "step": 5501 }, { "epoch": 0.6402141028624622, "grad_norm": 4.249142646789551, "learning_rate": 0.0005, "loss": 3.127805471420288, "step": 5502 }, { "epoch": 0.6403304631138004, "grad_norm": 2.1357250213623047, "learning_rate": 0.0005, "loss": 2.4220457077026367, "step": 5503 }, { "epoch": 0.6404468233651385, "grad_norm": 1.2168083190917969, "learning_rate": 0.0005, "loss": 2.166501998901367, "step": 5504 }, { "epoch": 0.6405631836164766, "grad_norm": 1.6403719186782837, "learning_rate": 0.0005, "loss": 2.6197221279144287, "step": 5505 }, { "epoch": 0.6406795438678148, "grad_norm": 1.167574167251587, "learning_rate": 0.0005, "loss": 2.6558003425598145, "step": 5506 }, { "epoch": 0.6407959041191529, "grad_norm": 1.483120322227478, "learning_rate": 0.0005, "loss": 2.873439311981201, "step": 5507 }, { "epoch": 0.6409122643704911, "grad_norm": 2.601393938064575, "learning_rate": 0.0005, "loss": 3.2989659309387207, "step": 5508 }, { "epoch": 0.6410286246218292, "grad_norm": 1.731353998184204, "learning_rate": 0.0005, "loss": 2.142894744873047, "step": 5509 }, { "epoch": 0.6411449848731673, "grad_norm": 2.024791717529297, "learning_rate": 0.0005, "loss": 2.3636436462402344, "step": 5510 }, { "epoch": 0.6412613451245055, "grad_norm": 1.8229752779006958, "learning_rate": 0.0005, "loss": 3.252775192260742, "step": 5511 }, { "epoch": 0.6413777053758436, "grad_norm": 1.76633620262146, "learning_rate": 0.0005, "loss": 3.641278028488159, "step": 5512 }, { "epoch": 0.6414940656271818, "grad_norm": 1.2692341804504395, "learning_rate": 0.0005, "loss": 2.2210514545440674, "step": 5513 }, { "epoch": 0.6416104258785199, "grad_norm": 1.7533719539642334, "learning_rate": 0.0005, "loss": 2.1272053718566895, "step": 5514 }, { "epoch": 0.641726786129858, "grad_norm": 1.377568244934082, "learning_rate": 0.0005, "loss": 1.7335505485534668, "step": 5515 }, { "epoch": 0.6418431463811962, "grad_norm": 0.9266469478607178, "learning_rate": 0.0005, "loss": 1.5685639381408691, "step": 5516 }, { "epoch": 0.6419595066325343, "grad_norm": 1.1407990455627441, "learning_rate": 0.0005, "loss": 1.5187416076660156, "step": 5517 }, { "epoch": 0.6420758668838724, "grad_norm": 1.8127139806747437, "learning_rate": 0.0005, "loss": 2.7269277572631836, "step": 5518 }, { "epoch": 0.6421922271352106, "grad_norm": 1.076920747756958, "learning_rate": 0.0005, "loss": 1.404825210571289, "step": 5519 }, { "epoch": 0.6423085873865487, "grad_norm": 0.9167554378509521, "learning_rate": 0.0005, "loss": 1.8205679655075073, "step": 5520 }, { "epoch": 0.6424249476378869, "grad_norm": 1.5450571775436401, "learning_rate": 0.0005, "loss": 2.3586807250976562, "step": 5521 }, { "epoch": 0.642541307889225, "grad_norm": 2.028151512145996, "learning_rate": 0.0005, "loss": 3.0181021690368652, "step": 5522 }, { "epoch": 0.6426576681405631, "grad_norm": 1.5637892484664917, "learning_rate": 0.0005, "loss": 2.3896992206573486, "step": 5523 }, { "epoch": 0.6427740283919013, "grad_norm": 1.563101053237915, "learning_rate": 0.0005, "loss": 2.203324794769287, "step": 5524 }, { "epoch": 0.6428903886432394, "grad_norm": 1.2941244840621948, "learning_rate": 0.0005, "loss": 2.0843610763549805, "step": 5525 }, { "epoch": 0.6430067488945777, "grad_norm": 1.5148353576660156, "learning_rate": 0.0005, "loss": 2.6751532554626465, "step": 5526 }, { "epoch": 0.6431231091459158, "grad_norm": 2.372504949569702, "learning_rate": 0.0005, "loss": 2.423517942428589, "step": 5527 }, { "epoch": 0.6432394693972538, "grad_norm": 6.703436374664307, "learning_rate": 0.0005, "loss": 2.901775360107422, "step": 5528 }, { "epoch": 0.6433558296485921, "grad_norm": 1.2978520393371582, "learning_rate": 0.0005, "loss": 2.7341580390930176, "step": 5529 }, { "epoch": 0.6434721898999302, "grad_norm": 1.7117339372634888, "learning_rate": 0.0005, "loss": 3.229315996170044, "step": 5530 }, { "epoch": 0.6435885501512684, "grad_norm": 2.353780746459961, "learning_rate": 0.0005, "loss": 2.9406585693359375, "step": 5531 }, { "epoch": 0.6437049104026065, "grad_norm": 1.587532877922058, "learning_rate": 0.0005, "loss": 2.374152660369873, "step": 5532 }, { "epoch": 0.6438212706539446, "grad_norm": 2.240144968032837, "learning_rate": 0.0005, "loss": 2.3798909187316895, "step": 5533 }, { "epoch": 0.6439376309052828, "grad_norm": 2.0516202449798584, "learning_rate": 0.0005, "loss": 3.1522834300994873, "step": 5534 }, { "epoch": 0.6440539911566209, "grad_norm": 1.3859606981277466, "learning_rate": 0.0005, "loss": 2.3930206298828125, "step": 5535 }, { "epoch": 0.6441703514079591, "grad_norm": 1.423614501953125, "learning_rate": 0.0005, "loss": 2.862271785736084, "step": 5536 }, { "epoch": 0.6442867116592972, "grad_norm": 0.6527055501937866, "learning_rate": 0.0005, "loss": 1.7394261360168457, "step": 5537 }, { "epoch": 0.6444030719106353, "grad_norm": 1.2659662961959839, "learning_rate": 0.0005, "loss": 2.334049940109253, "step": 5538 }, { "epoch": 0.6445194321619735, "grad_norm": 0.8341386914253235, "learning_rate": 0.0005, "loss": 1.4201934337615967, "step": 5539 }, { "epoch": 0.6446357924133116, "grad_norm": 1.5035897493362427, "learning_rate": 0.0005, "loss": 2.7478857040405273, "step": 5540 }, { "epoch": 0.6447521526646498, "grad_norm": 1.278772234916687, "learning_rate": 0.0005, "loss": 2.088588237762451, "step": 5541 }, { "epoch": 0.6448685129159879, "grad_norm": 1.8402221202850342, "learning_rate": 0.0005, "loss": 2.869990825653076, "step": 5542 }, { "epoch": 0.644984873167326, "grad_norm": 1.1730800867080688, "learning_rate": 0.0005, "loss": 2.6623435020446777, "step": 5543 }, { "epoch": 0.6451012334186642, "grad_norm": 0.8559513688087463, "learning_rate": 0.0005, "loss": 1.8787888288497925, "step": 5544 }, { "epoch": 0.6452175936700023, "grad_norm": 2.154114007949829, "learning_rate": 0.0005, "loss": 2.1015453338623047, "step": 5545 }, { "epoch": 0.6453339539213405, "grad_norm": 1.1141066551208496, "learning_rate": 0.0005, "loss": 2.6142444610595703, "step": 5546 }, { "epoch": 0.6454503141726786, "grad_norm": 0.7668408155441284, "learning_rate": 0.0005, "loss": 1.4960944652557373, "step": 5547 }, { "epoch": 0.6455666744240167, "grad_norm": 0.6550959348678589, "learning_rate": 0.0005, "loss": 1.8185069561004639, "step": 5548 }, { "epoch": 0.6456830346753549, "grad_norm": 1.4161157608032227, "learning_rate": 0.0005, "loss": 2.5728063583374023, "step": 5549 }, { "epoch": 0.645799394926693, "grad_norm": 1.0175859928131104, "learning_rate": 0.0005, "loss": 2.366726875305176, "step": 5550 }, { "epoch": 0.6459157551780312, "grad_norm": 1.5263642072677612, "learning_rate": 0.0005, "loss": 2.478652238845825, "step": 5551 }, { "epoch": 0.6460321154293693, "grad_norm": 1.7838023900985718, "learning_rate": 0.0005, "loss": 3.0157833099365234, "step": 5552 }, { "epoch": 0.6461484756807074, "grad_norm": 1.5322932004928589, "learning_rate": 0.0005, "loss": 1.6767261028289795, "step": 5553 }, { "epoch": 0.6462648359320456, "grad_norm": 0.927945077419281, "learning_rate": 0.0005, "loss": 1.4860596656799316, "step": 5554 }, { "epoch": 0.6463811961833837, "grad_norm": 2.450650453567505, "learning_rate": 0.0005, "loss": 3.0204906463623047, "step": 5555 }, { "epoch": 0.646497556434722, "grad_norm": 2.452650547027588, "learning_rate": 0.0005, "loss": 2.4730844497680664, "step": 5556 }, { "epoch": 0.64661391668606, "grad_norm": 2.179575204849243, "learning_rate": 0.0005, "loss": 2.369947910308838, "step": 5557 }, { "epoch": 0.6467302769373982, "grad_norm": 1.5872398614883423, "learning_rate": 0.0005, "loss": 2.26212739944458, "step": 5558 }, { "epoch": 0.6468466371887364, "grad_norm": 0.7170474529266357, "learning_rate": 0.0005, "loss": 1.5047667026519775, "step": 5559 }, { "epoch": 0.6469629974400745, "grad_norm": 1.3160711526870728, "learning_rate": 0.0005, "loss": 2.3786487579345703, "step": 5560 }, { "epoch": 0.6470793576914126, "grad_norm": 1.132677435874939, "learning_rate": 0.0005, "loss": 1.9000811576843262, "step": 5561 }, { "epoch": 0.6471957179427508, "grad_norm": 1.2165226936340332, "learning_rate": 0.0005, "loss": 2.6936540603637695, "step": 5562 }, { "epoch": 0.6473120781940889, "grad_norm": 1.2829755544662476, "learning_rate": 0.0005, "loss": 2.6372570991516113, "step": 5563 }, { "epoch": 0.6474284384454271, "grad_norm": 1.8329343795776367, "learning_rate": 0.0005, "loss": 2.1102919578552246, "step": 5564 }, { "epoch": 0.6475447986967652, "grad_norm": 0.9572914838790894, "learning_rate": 0.0005, "loss": 2.5326294898986816, "step": 5565 }, { "epoch": 0.6476611589481033, "grad_norm": 0.7712112665176392, "learning_rate": 0.0005, "loss": 2.108440637588501, "step": 5566 }, { "epoch": 0.6477775191994415, "grad_norm": 2.0280168056488037, "learning_rate": 0.0005, "loss": 2.4192490577697754, "step": 5567 }, { "epoch": 0.6478938794507796, "grad_norm": 1.176944613456726, "learning_rate": 0.0005, "loss": 2.1139907836914062, "step": 5568 }, { "epoch": 0.6480102397021178, "grad_norm": 1.9272764921188354, "learning_rate": 0.0005, "loss": 1.7868469953536987, "step": 5569 }, { "epoch": 0.6481265999534559, "grad_norm": 1.1184720993041992, "learning_rate": 0.0005, "loss": 2.6319713592529297, "step": 5570 }, { "epoch": 0.648242960204794, "grad_norm": 1.1083967685699463, "learning_rate": 0.0005, "loss": 1.9617698192596436, "step": 5571 }, { "epoch": 0.6483593204561322, "grad_norm": 1.3263771533966064, "learning_rate": 0.0005, "loss": 2.5347864627838135, "step": 5572 }, { "epoch": 0.6484756807074703, "grad_norm": 2.519571542739868, "learning_rate": 0.0005, "loss": 2.0795578956604004, "step": 5573 }, { "epoch": 0.6485920409588085, "grad_norm": 2.7391774654388428, "learning_rate": 0.0005, "loss": 2.640451669692993, "step": 5574 }, { "epoch": 0.6487084012101466, "grad_norm": 1.0607221126556396, "learning_rate": 0.0005, "loss": 2.3653063774108887, "step": 5575 }, { "epoch": 0.6488247614614847, "grad_norm": 1.5046817064285278, "learning_rate": 0.0005, "loss": 2.0317161083221436, "step": 5576 }, { "epoch": 0.6489411217128229, "grad_norm": 1.4368423223495483, "learning_rate": 0.0005, "loss": 2.0097367763519287, "step": 5577 }, { "epoch": 0.649057481964161, "grad_norm": 1.8748005628585815, "learning_rate": 0.0005, "loss": 2.3871710300445557, "step": 5578 }, { "epoch": 0.6491738422154992, "grad_norm": 1.7629176378250122, "learning_rate": 0.0005, "loss": 2.4702720642089844, "step": 5579 }, { "epoch": 0.6492902024668373, "grad_norm": 1.83213210105896, "learning_rate": 0.0005, "loss": 3.1859982013702393, "step": 5580 }, { "epoch": 0.6494065627181754, "grad_norm": 0.6653947234153748, "learning_rate": 0.0005, "loss": 1.432347059249878, "step": 5581 }, { "epoch": 0.6495229229695136, "grad_norm": 1.2735834121704102, "learning_rate": 0.0005, "loss": 2.1255409717559814, "step": 5582 }, { "epoch": 0.6496392832208517, "grad_norm": 1.194770097732544, "learning_rate": 0.0005, "loss": 2.377490520477295, "step": 5583 }, { "epoch": 0.64975564347219, "grad_norm": 1.7285901308059692, "learning_rate": 0.0005, "loss": 2.4821224212646484, "step": 5584 }, { "epoch": 0.649872003723528, "grad_norm": 1.7901241779327393, "learning_rate": 0.0005, "loss": 2.6813626289367676, "step": 5585 }, { "epoch": 0.6499883639748661, "grad_norm": 1.9324554204940796, "learning_rate": 0.0005, "loss": 3.0868794918060303, "step": 5586 }, { "epoch": 0.6501047242262044, "grad_norm": 1.102738380432129, "learning_rate": 0.0005, "loss": 1.6776988506317139, "step": 5587 }, { "epoch": 0.6502210844775425, "grad_norm": 0.8774121403694153, "learning_rate": 0.0005, "loss": 2.1690335273742676, "step": 5588 }, { "epoch": 0.6503374447288807, "grad_norm": 1.1589785814285278, "learning_rate": 0.0005, "loss": 2.0649237632751465, "step": 5589 }, { "epoch": 0.6504538049802188, "grad_norm": 1.2601573467254639, "learning_rate": 0.0005, "loss": 2.982462167739868, "step": 5590 }, { "epoch": 0.6505701652315569, "grad_norm": 1.238213300704956, "learning_rate": 0.0005, "loss": 2.475130081176758, "step": 5591 }, { "epoch": 0.6506865254828951, "grad_norm": 0.9183129072189331, "learning_rate": 0.0005, "loss": 2.0838444232940674, "step": 5592 }, { "epoch": 0.6508028857342332, "grad_norm": 1.5290653705596924, "learning_rate": 0.0005, "loss": 2.314480781555176, "step": 5593 }, { "epoch": 0.6509192459855714, "grad_norm": 1.3732936382293701, "learning_rate": 0.0005, "loss": 2.6803762912750244, "step": 5594 }, { "epoch": 0.6510356062369095, "grad_norm": 2.3242671489715576, "learning_rate": 0.0005, "loss": 2.3670148849487305, "step": 5595 }, { "epoch": 0.6511519664882476, "grad_norm": 1.49930739402771, "learning_rate": 0.0005, "loss": 2.7654333114624023, "step": 5596 }, { "epoch": 0.6512683267395858, "grad_norm": 1.075416088104248, "learning_rate": 0.0005, "loss": 1.744248390197754, "step": 5597 }, { "epoch": 0.6513846869909239, "grad_norm": 1.053237795829773, "learning_rate": 0.0005, "loss": 2.308453321456909, "step": 5598 }, { "epoch": 0.651501047242262, "grad_norm": 1.915985345840454, "learning_rate": 0.0005, "loss": 2.3213133811950684, "step": 5599 }, { "epoch": 0.6516174074936002, "grad_norm": 1.9433226585388184, "learning_rate": 0.0005, "loss": 3.131119966506958, "step": 5600 }, { "epoch": 0.6517337677449383, "grad_norm": 2.183072328567505, "learning_rate": 0.0005, "loss": 2.309349536895752, "step": 5601 }, { "epoch": 0.6518501279962765, "grad_norm": 2.2012479305267334, "learning_rate": 0.0005, "loss": 2.2729530334472656, "step": 5602 }, { "epoch": 0.6519664882476146, "grad_norm": 1.4166812896728516, "learning_rate": 0.0005, "loss": 2.1421170234680176, "step": 5603 }, { "epoch": 0.6520828484989527, "grad_norm": 1.7200343608856201, "learning_rate": 0.0005, "loss": 3.01808762550354, "step": 5604 }, { "epoch": 0.6521992087502909, "grad_norm": 1.0286365747451782, "learning_rate": 0.0005, "loss": 2.317417860031128, "step": 5605 }, { "epoch": 0.652315569001629, "grad_norm": 1.552214503288269, "learning_rate": 0.0005, "loss": 2.263665199279785, "step": 5606 }, { "epoch": 0.6524319292529672, "grad_norm": 1.2509444952011108, "learning_rate": 0.0005, "loss": 2.502389430999756, "step": 5607 }, { "epoch": 0.6525482895043053, "grad_norm": 0.9797378778457642, "learning_rate": 0.0005, "loss": 2.0703110694885254, "step": 5608 }, { "epoch": 0.6526646497556434, "grad_norm": 1.136204719543457, "learning_rate": 0.0005, "loss": 2.572810411453247, "step": 5609 }, { "epoch": 0.6527810100069816, "grad_norm": 1.439085841178894, "learning_rate": 0.0005, "loss": 2.286384105682373, "step": 5610 }, { "epoch": 0.6528973702583197, "grad_norm": 1.3448232412338257, "learning_rate": 0.0005, "loss": 3.5827231407165527, "step": 5611 }, { "epoch": 0.6530137305096579, "grad_norm": 1.6113407611846924, "learning_rate": 0.0005, "loss": 2.5550365447998047, "step": 5612 }, { "epoch": 0.653130090760996, "grad_norm": 1.2730004787445068, "learning_rate": 0.0005, "loss": 2.321112632751465, "step": 5613 }, { "epoch": 0.6532464510123341, "grad_norm": 1.1046783924102783, "learning_rate": 0.0005, "loss": 2.4245760440826416, "step": 5614 }, { "epoch": 0.6533628112636723, "grad_norm": 1.922178864479065, "learning_rate": 0.0005, "loss": 2.8833374977111816, "step": 5615 }, { "epoch": 0.6534791715150104, "grad_norm": 2.6631436347961426, "learning_rate": 0.0005, "loss": 3.325392961502075, "step": 5616 }, { "epoch": 0.6535955317663487, "grad_norm": 1.3358204364776611, "learning_rate": 0.0005, "loss": 1.6402335166931152, "step": 5617 }, { "epoch": 0.6537118920176868, "grad_norm": 1.707427978515625, "learning_rate": 0.0005, "loss": 2.9684648513793945, "step": 5618 }, { "epoch": 0.6538282522690249, "grad_norm": 2.038914918899536, "learning_rate": 0.0005, "loss": 3.0023765563964844, "step": 5619 }, { "epoch": 0.6539446125203631, "grad_norm": 1.1787234544754028, "learning_rate": 0.0005, "loss": 2.298264503479004, "step": 5620 }, { "epoch": 0.6540609727717012, "grad_norm": 1.3751068115234375, "learning_rate": 0.0005, "loss": 3.6961922645568848, "step": 5621 }, { "epoch": 0.6541773330230394, "grad_norm": 1.6822348833084106, "learning_rate": 0.0005, "loss": 3.430241823196411, "step": 5622 }, { "epoch": 0.6542936932743775, "grad_norm": 0.9143469333648682, "learning_rate": 0.0005, "loss": 1.5993541479110718, "step": 5623 }, { "epoch": 0.6544100535257156, "grad_norm": 1.0982732772827148, "learning_rate": 0.0005, "loss": 1.83981192111969, "step": 5624 }, { "epoch": 0.6545264137770538, "grad_norm": 3.281662940979004, "learning_rate": 0.0005, "loss": 2.5092787742614746, "step": 5625 }, { "epoch": 0.6546427740283919, "grad_norm": 1.1765549182891846, "learning_rate": 0.0005, "loss": 1.851693868637085, "step": 5626 }, { "epoch": 0.6547591342797301, "grad_norm": 2.091970205307007, "learning_rate": 0.0005, "loss": 2.524385929107666, "step": 5627 }, { "epoch": 0.6548754945310682, "grad_norm": 1.207135558128357, "learning_rate": 0.0005, "loss": 2.5755341053009033, "step": 5628 }, { "epoch": 0.6549918547824063, "grad_norm": 1.983873724937439, "learning_rate": 0.0005, "loss": 2.525883674621582, "step": 5629 }, { "epoch": 0.6551082150337445, "grad_norm": 0.7767796516418457, "learning_rate": 0.0005, "loss": 1.9071638584136963, "step": 5630 }, { "epoch": 0.6552245752850826, "grad_norm": 1.0699549913406372, "learning_rate": 0.0005, "loss": 2.080716609954834, "step": 5631 }, { "epoch": 0.6553409355364208, "grad_norm": 1.8318109512329102, "learning_rate": 0.0005, "loss": 2.437793731689453, "step": 5632 }, { "epoch": 0.6554572957877589, "grad_norm": 1.0597747564315796, "learning_rate": 0.0005, "loss": 1.8443673849105835, "step": 5633 }, { "epoch": 0.655573656039097, "grad_norm": 1.7595731019973755, "learning_rate": 0.0005, "loss": 2.8139286041259766, "step": 5634 }, { "epoch": 0.6556900162904352, "grad_norm": 2.3552606105804443, "learning_rate": 0.0005, "loss": 2.6371583938598633, "step": 5635 }, { "epoch": 0.6558063765417733, "grad_norm": 0.9979475736618042, "learning_rate": 0.0005, "loss": 2.2722227573394775, "step": 5636 }, { "epoch": 0.6559227367931115, "grad_norm": 1.3571394681930542, "learning_rate": 0.0005, "loss": 2.4545602798461914, "step": 5637 }, { "epoch": 0.6560390970444496, "grad_norm": 1.3332420587539673, "learning_rate": 0.0005, "loss": 2.3792333602905273, "step": 5638 }, { "epoch": 0.6561554572957877, "grad_norm": 1.2036510705947876, "learning_rate": 0.0005, "loss": 2.07873272895813, "step": 5639 }, { "epoch": 0.6562718175471259, "grad_norm": 1.0596387386322021, "learning_rate": 0.0005, "loss": 1.7810261249542236, "step": 5640 }, { "epoch": 0.656388177798464, "grad_norm": 1.5308281183242798, "learning_rate": 0.0005, "loss": 2.393366813659668, "step": 5641 }, { "epoch": 0.6565045380498021, "grad_norm": 1.5264487266540527, "learning_rate": 0.0005, "loss": 2.461829662322998, "step": 5642 }, { "epoch": 0.6566208983011403, "grad_norm": 2.019965887069702, "learning_rate": 0.0005, "loss": 2.0661838054656982, "step": 5643 }, { "epoch": 0.6567372585524784, "grad_norm": 1.1774219274520874, "learning_rate": 0.0005, "loss": 1.7609596252441406, "step": 5644 }, { "epoch": 0.6568536188038167, "grad_norm": 1.7408252954483032, "learning_rate": 0.0005, "loss": 2.1253445148468018, "step": 5645 }, { "epoch": 0.6569699790551548, "grad_norm": 1.203044056892395, "learning_rate": 0.0005, "loss": 3.0907845497131348, "step": 5646 }, { "epoch": 0.6570863393064929, "grad_norm": 1.6693137884140015, "learning_rate": 0.0005, "loss": 2.6207025051116943, "step": 5647 }, { "epoch": 0.6572026995578311, "grad_norm": 1.3766251802444458, "learning_rate": 0.0005, "loss": 2.756986618041992, "step": 5648 }, { "epoch": 0.6573190598091692, "grad_norm": 1.3407267332077026, "learning_rate": 0.0005, "loss": 3.292870283126831, "step": 5649 }, { "epoch": 0.6574354200605074, "grad_norm": 0.9130280613899231, "learning_rate": 0.0005, "loss": 2.291822910308838, "step": 5650 }, { "epoch": 0.6575517803118455, "grad_norm": 1.4231085777282715, "learning_rate": 0.0005, "loss": 2.6988449096679688, "step": 5651 }, { "epoch": 0.6576681405631836, "grad_norm": 0.6273180246353149, "learning_rate": 0.0005, "loss": 1.469455599784851, "step": 5652 }, { "epoch": 0.6577845008145218, "grad_norm": 1.5139445066452026, "learning_rate": 0.0005, "loss": 2.5145375728607178, "step": 5653 }, { "epoch": 0.6579008610658599, "grad_norm": 1.16476571559906, "learning_rate": 0.0005, "loss": 1.627827763557434, "step": 5654 }, { "epoch": 0.6580172213171981, "grad_norm": 1.470228910446167, "learning_rate": 0.0005, "loss": 1.7984328269958496, "step": 5655 }, { "epoch": 0.6581335815685362, "grad_norm": 1.6672089099884033, "learning_rate": 0.0005, "loss": 2.3409616947174072, "step": 5656 }, { "epoch": 0.6582499418198743, "grad_norm": 1.9837889671325684, "learning_rate": 0.0005, "loss": 3.119098663330078, "step": 5657 }, { "epoch": 0.6583663020712125, "grad_norm": 0.958683967590332, "learning_rate": 0.0005, "loss": 1.1157760620117188, "step": 5658 }, { "epoch": 0.6584826623225506, "grad_norm": 1.4619495868682861, "learning_rate": 0.0005, "loss": 1.9775922298431396, "step": 5659 }, { "epoch": 0.6585990225738888, "grad_norm": 1.4258556365966797, "learning_rate": 0.0005, "loss": 1.8974308967590332, "step": 5660 }, { "epoch": 0.6587153828252269, "grad_norm": 1.0526845455169678, "learning_rate": 0.0005, "loss": 1.9709832668304443, "step": 5661 }, { "epoch": 0.658831743076565, "grad_norm": 2.8531649112701416, "learning_rate": 0.0005, "loss": 2.6477723121643066, "step": 5662 }, { "epoch": 0.6589481033279032, "grad_norm": 1.7247601747512817, "learning_rate": 0.0005, "loss": 1.9942657947540283, "step": 5663 }, { "epoch": 0.6590644635792413, "grad_norm": 1.5546835660934448, "learning_rate": 0.0005, "loss": 2.6744842529296875, "step": 5664 }, { "epoch": 0.6591808238305795, "grad_norm": 1.0904150009155273, "learning_rate": 0.0005, "loss": 2.1859934329986572, "step": 5665 }, { "epoch": 0.6592971840819176, "grad_norm": 2.2009801864624023, "learning_rate": 0.0005, "loss": 2.7074856758117676, "step": 5666 }, { "epoch": 0.6594135443332557, "grad_norm": 1.6098301410675049, "learning_rate": 0.0005, "loss": 2.453077554702759, "step": 5667 }, { "epoch": 0.6595299045845939, "grad_norm": 1.6067955493927002, "learning_rate": 0.0005, "loss": 2.836712598800659, "step": 5668 }, { "epoch": 0.659646264835932, "grad_norm": 1.821433186531067, "learning_rate": 0.0005, "loss": 2.111643075942993, "step": 5669 }, { "epoch": 0.6597626250872702, "grad_norm": 1.0411711931228638, "learning_rate": 0.0005, "loss": 2.372368812561035, "step": 5670 }, { "epoch": 0.6598789853386083, "grad_norm": 1.7291765213012695, "learning_rate": 0.0005, "loss": 2.153728723526001, "step": 5671 }, { "epoch": 0.6599953455899464, "grad_norm": 1.1156178712844849, "learning_rate": 0.0005, "loss": 2.5053184032440186, "step": 5672 }, { "epoch": 0.6601117058412846, "grad_norm": 1.3065062761306763, "learning_rate": 0.0005, "loss": 2.1645727157592773, "step": 5673 }, { "epoch": 0.6602280660926227, "grad_norm": 1.1734585762023926, "learning_rate": 0.0005, "loss": 2.761899471282959, "step": 5674 }, { "epoch": 0.660344426343961, "grad_norm": 0.6386162042617798, "learning_rate": 0.0005, "loss": 1.5203063488006592, "step": 5675 }, { "epoch": 0.660460786595299, "grad_norm": 1.290156602859497, "learning_rate": 0.0005, "loss": 1.992274522781372, "step": 5676 }, { "epoch": 0.6605771468466372, "grad_norm": 1.861250400543213, "learning_rate": 0.0005, "loss": 2.6778249740600586, "step": 5677 }, { "epoch": 0.6606935070979754, "grad_norm": 1.2103826999664307, "learning_rate": 0.0005, "loss": 2.0726261138916016, "step": 5678 }, { "epoch": 0.6608098673493135, "grad_norm": 1.207003116607666, "learning_rate": 0.0005, "loss": 2.0196280479431152, "step": 5679 }, { "epoch": 0.6609262276006517, "grad_norm": 1.2115585803985596, "learning_rate": 0.0005, "loss": 2.5195796489715576, "step": 5680 }, { "epoch": 0.6610425878519898, "grad_norm": 1.5592243671417236, "learning_rate": 0.0005, "loss": 2.7698421478271484, "step": 5681 }, { "epoch": 0.6611589481033279, "grad_norm": 0.8829861283302307, "learning_rate": 0.0005, "loss": 1.4042489528656006, "step": 5682 }, { "epoch": 0.6612753083546661, "grad_norm": 1.5661845207214355, "learning_rate": 0.0005, "loss": 2.907644748687744, "step": 5683 }, { "epoch": 0.6613916686060042, "grad_norm": 2.0639848709106445, "learning_rate": 0.0005, "loss": 3.093067169189453, "step": 5684 }, { "epoch": 0.6615080288573423, "grad_norm": 1.6714746952056885, "learning_rate": 0.0005, "loss": 2.3509349822998047, "step": 5685 }, { "epoch": 0.6616243891086805, "grad_norm": 1.7759662866592407, "learning_rate": 0.0005, "loss": 2.659655809402466, "step": 5686 }, { "epoch": 0.6617407493600186, "grad_norm": 1.0913764238357544, "learning_rate": 0.0005, "loss": 1.7220702171325684, "step": 5687 }, { "epoch": 0.6618571096113568, "grad_norm": 1.081551194190979, "learning_rate": 0.0005, "loss": 2.300365686416626, "step": 5688 }, { "epoch": 0.6619734698626949, "grad_norm": 1.6283924579620361, "learning_rate": 0.0005, "loss": 2.6531379222869873, "step": 5689 }, { "epoch": 0.662089830114033, "grad_norm": 0.8076973557472229, "learning_rate": 0.0005, "loss": 2.604203224182129, "step": 5690 }, { "epoch": 0.6622061903653712, "grad_norm": 1.309563159942627, "learning_rate": 0.0005, "loss": 2.377775192260742, "step": 5691 }, { "epoch": 0.6623225506167093, "grad_norm": 1.23996102809906, "learning_rate": 0.0005, "loss": 1.7795391082763672, "step": 5692 }, { "epoch": 0.6624389108680475, "grad_norm": 1.7768393754959106, "learning_rate": 0.0005, "loss": 1.4167388677597046, "step": 5693 }, { "epoch": 0.6625552711193856, "grad_norm": 1.2587451934814453, "learning_rate": 0.0005, "loss": 2.5132346153259277, "step": 5694 }, { "epoch": 0.6626716313707237, "grad_norm": 1.5794011354446411, "learning_rate": 0.0005, "loss": 2.5143232345581055, "step": 5695 }, { "epoch": 0.6627879916220619, "grad_norm": 0.8735095262527466, "learning_rate": 0.0005, "loss": 1.8324527740478516, "step": 5696 }, { "epoch": 0.6629043518734, "grad_norm": 0.9635817408561707, "learning_rate": 0.0005, "loss": 2.013892650604248, "step": 5697 }, { "epoch": 0.6630207121247382, "grad_norm": 1.8199394941329956, "learning_rate": 0.0005, "loss": 2.486295461654663, "step": 5698 }, { "epoch": 0.6631370723760763, "grad_norm": 1.4316266775131226, "learning_rate": 0.0005, "loss": 2.5526065826416016, "step": 5699 }, { "epoch": 0.6632534326274144, "grad_norm": 1.6785597801208496, "learning_rate": 0.0005, "loss": 2.7765896320343018, "step": 5700 }, { "epoch": 0.6633697928787526, "grad_norm": 1.4159091711044312, "learning_rate": 0.0005, "loss": 2.1127684116363525, "step": 5701 }, { "epoch": 0.6634861531300907, "grad_norm": 0.6767757534980774, "learning_rate": 0.0005, "loss": 1.492402195930481, "step": 5702 }, { "epoch": 0.663602513381429, "grad_norm": 1.802455186843872, "learning_rate": 0.0005, "loss": 2.0452466011047363, "step": 5703 }, { "epoch": 0.663718873632767, "grad_norm": 2.7301721572875977, "learning_rate": 0.0005, "loss": 3.2262072563171387, "step": 5704 }, { "epoch": 0.6638352338841051, "grad_norm": 2.4275307655334473, "learning_rate": 0.0005, "loss": 2.5680525302886963, "step": 5705 }, { "epoch": 0.6639515941354434, "grad_norm": 1.3930442333221436, "learning_rate": 0.0005, "loss": 2.4596915245056152, "step": 5706 }, { "epoch": 0.6640679543867815, "grad_norm": 1.425702452659607, "learning_rate": 0.0005, "loss": 2.2423670291900635, "step": 5707 }, { "epoch": 0.6641843146381197, "grad_norm": 1.142111897468567, "learning_rate": 0.0005, "loss": 2.3915951251983643, "step": 5708 }, { "epoch": 0.6643006748894578, "grad_norm": 1.1765652894973755, "learning_rate": 0.0005, "loss": 3.2259912490844727, "step": 5709 }, { "epoch": 0.6644170351407959, "grad_norm": 1.2616244554519653, "learning_rate": 0.0005, "loss": 2.3798742294311523, "step": 5710 }, { "epoch": 0.6645333953921341, "grad_norm": 1.1961570978164673, "learning_rate": 0.0005, "loss": 2.267897129058838, "step": 5711 }, { "epoch": 0.6646497556434722, "grad_norm": 1.6746591329574585, "learning_rate": 0.0005, "loss": 3.3513612747192383, "step": 5712 }, { "epoch": 0.6647661158948104, "grad_norm": 2.671159029006958, "learning_rate": 0.0005, "loss": 2.618293523788452, "step": 5713 }, { "epoch": 0.6648824761461485, "grad_norm": 0.9922807216644287, "learning_rate": 0.0005, "loss": 2.132539987564087, "step": 5714 }, { "epoch": 0.6649988363974866, "grad_norm": 2.037588357925415, "learning_rate": 0.0005, "loss": 2.199341297149658, "step": 5715 }, { "epoch": 0.6651151966488248, "grad_norm": 1.2731008529663086, "learning_rate": 0.0005, "loss": 2.78779673576355, "step": 5716 }, { "epoch": 0.6652315569001629, "grad_norm": 1.3793734312057495, "learning_rate": 0.0005, "loss": 3.043950080871582, "step": 5717 }, { "epoch": 0.6653479171515011, "grad_norm": 1.632625937461853, "learning_rate": 0.0005, "loss": 2.3752453327178955, "step": 5718 }, { "epoch": 0.6654642774028392, "grad_norm": 9.490448951721191, "learning_rate": 0.0005, "loss": 2.773534059524536, "step": 5719 }, { "epoch": 0.6655806376541773, "grad_norm": 1.4992599487304688, "learning_rate": 0.0005, "loss": 1.9446555376052856, "step": 5720 }, { "epoch": 0.6656969979055155, "grad_norm": 1.4715821743011475, "learning_rate": 0.0005, "loss": 2.9320077896118164, "step": 5721 }, { "epoch": 0.6658133581568536, "grad_norm": 1.316563606262207, "learning_rate": 0.0005, "loss": 3.6872968673706055, "step": 5722 }, { "epoch": 0.6659297184081917, "grad_norm": 1.6042295694351196, "learning_rate": 0.0005, "loss": 2.639457941055298, "step": 5723 }, { "epoch": 0.6660460786595299, "grad_norm": 1.5006715059280396, "learning_rate": 0.0005, "loss": 2.0061793327331543, "step": 5724 }, { "epoch": 0.666162438910868, "grad_norm": 0.7448312044143677, "learning_rate": 0.0005, "loss": 2.4358935356140137, "step": 5725 }, { "epoch": 0.6662787991622062, "grad_norm": 1.8771144151687622, "learning_rate": 0.0005, "loss": 2.4703845977783203, "step": 5726 }, { "epoch": 0.6663951594135443, "grad_norm": 1.5003478527069092, "learning_rate": 0.0005, "loss": 3.3099937438964844, "step": 5727 }, { "epoch": 0.6665115196648824, "grad_norm": 1.3008407354354858, "learning_rate": 0.0005, "loss": 2.0909955501556396, "step": 5728 }, { "epoch": 0.6666278799162206, "grad_norm": 1.9492356777191162, "learning_rate": 0.0005, "loss": 2.6907076835632324, "step": 5729 }, { "epoch": 0.6667442401675587, "grad_norm": 1.125031590461731, "learning_rate": 0.0005, "loss": 2.4307594299316406, "step": 5730 }, { "epoch": 0.6668606004188969, "grad_norm": 1.3238037824630737, "learning_rate": 0.0005, "loss": 2.722625255584717, "step": 5731 }, { "epoch": 0.666976960670235, "grad_norm": 0.5682061314582825, "learning_rate": 0.0005, "loss": 1.4098623991012573, "step": 5732 }, { "epoch": 0.6670933209215731, "grad_norm": 2.1923136711120605, "learning_rate": 0.0005, "loss": 2.17044734954834, "step": 5733 }, { "epoch": 0.6672096811729114, "grad_norm": 0.8237733244895935, "learning_rate": 0.0005, "loss": 1.4613220691680908, "step": 5734 }, { "epoch": 0.6673260414242494, "grad_norm": 0.9701855182647705, "learning_rate": 0.0005, "loss": 1.2138338088989258, "step": 5735 }, { "epoch": 0.6674424016755877, "grad_norm": 3.0381014347076416, "learning_rate": 0.0005, "loss": 3.143331527709961, "step": 5736 }, { "epoch": 0.6675587619269258, "grad_norm": 2.479865312576294, "learning_rate": 0.0005, "loss": 3.068100690841675, "step": 5737 }, { "epoch": 0.6676751221782639, "grad_norm": 1.162544846534729, "learning_rate": 0.0005, "loss": 1.9333747625350952, "step": 5738 }, { "epoch": 0.6677914824296021, "grad_norm": 1.5891757011413574, "learning_rate": 0.0005, "loss": 3.092679262161255, "step": 5739 }, { "epoch": 0.6679078426809402, "grad_norm": 2.6761043071746826, "learning_rate": 0.0005, "loss": 2.2846872806549072, "step": 5740 }, { "epoch": 0.6680242029322784, "grad_norm": 1.3205876350402832, "learning_rate": 0.0005, "loss": 2.4580941200256348, "step": 5741 }, { "epoch": 0.6681405631836165, "grad_norm": 1.2387853860855103, "learning_rate": 0.0005, "loss": 2.1380763053894043, "step": 5742 }, { "epoch": 0.6682569234349546, "grad_norm": 0.9122348427772522, "learning_rate": 0.0005, "loss": 2.2097620964050293, "step": 5743 }, { "epoch": 0.6683732836862928, "grad_norm": 1.6006042957305908, "learning_rate": 0.0005, "loss": 2.365990161895752, "step": 5744 }, { "epoch": 0.6684896439376309, "grad_norm": 1.0948026180267334, "learning_rate": 0.0005, "loss": 2.3380846977233887, "step": 5745 }, { "epoch": 0.6686060041889691, "grad_norm": 1.9390292167663574, "learning_rate": 0.0005, "loss": 2.1520166397094727, "step": 5746 }, { "epoch": 0.6687223644403072, "grad_norm": 2.664945602416992, "learning_rate": 0.0005, "loss": 2.9017744064331055, "step": 5747 }, { "epoch": 0.6688387246916453, "grad_norm": 1.4571577310562134, "learning_rate": 0.0005, "loss": 2.9049859046936035, "step": 5748 }, { "epoch": 0.6689550849429835, "grad_norm": 2.0032618045806885, "learning_rate": 0.0005, "loss": 2.9576220512390137, "step": 5749 }, { "epoch": 0.6690714451943216, "grad_norm": 1.0622599124908447, "learning_rate": 0.0005, "loss": 2.6318907737731934, "step": 5750 }, { "epoch": 0.6691878054456598, "grad_norm": 1.776455283164978, "learning_rate": 0.0005, "loss": 2.4034738540649414, "step": 5751 }, { "epoch": 0.6693041656969979, "grad_norm": 1.2307560443878174, "learning_rate": 0.0005, "loss": 2.0465664863586426, "step": 5752 }, { "epoch": 0.669420525948336, "grad_norm": 1.4980920553207397, "learning_rate": 0.0005, "loss": 2.4828124046325684, "step": 5753 }, { "epoch": 0.6695368861996742, "grad_norm": 1.3547381162643433, "learning_rate": 0.0005, "loss": 2.8728017807006836, "step": 5754 }, { "epoch": 0.6696532464510123, "grad_norm": 0.8561180830001831, "learning_rate": 0.0005, "loss": 2.2945899963378906, "step": 5755 }, { "epoch": 0.6697696067023505, "grad_norm": 4.293191432952881, "learning_rate": 0.0005, "loss": 2.223884105682373, "step": 5756 }, { "epoch": 0.6698859669536886, "grad_norm": 1.3386472463607788, "learning_rate": 0.0005, "loss": 2.6463491916656494, "step": 5757 }, { "epoch": 0.6700023272050267, "grad_norm": 1.5046498775482178, "learning_rate": 0.0005, "loss": 2.688232898712158, "step": 5758 }, { "epoch": 0.6701186874563649, "grad_norm": 1.8077701330184937, "learning_rate": 0.0005, "loss": 2.888493061065674, "step": 5759 }, { "epoch": 0.670235047707703, "grad_norm": 1.4021424055099487, "learning_rate": 0.0005, "loss": 2.7793455123901367, "step": 5760 }, { "epoch": 0.6703514079590412, "grad_norm": 1.4038985967636108, "learning_rate": 0.0005, "loss": 3.179593563079834, "step": 5761 }, { "epoch": 0.6704677682103793, "grad_norm": 0.9418932795524597, "learning_rate": 0.0005, "loss": 2.2524185180664062, "step": 5762 }, { "epoch": 0.6705841284617174, "grad_norm": 1.8276652097702026, "learning_rate": 0.0005, "loss": 2.4313971996307373, "step": 5763 }, { "epoch": 0.6707004887130557, "grad_norm": 0.8912972211837769, "learning_rate": 0.0005, "loss": 1.4931929111480713, "step": 5764 }, { "epoch": 0.6708168489643938, "grad_norm": 0.9382560849189758, "learning_rate": 0.0005, "loss": 2.4464759826660156, "step": 5765 }, { "epoch": 0.6709332092157319, "grad_norm": 1.496046543121338, "learning_rate": 0.0005, "loss": 2.9420225620269775, "step": 5766 }, { "epoch": 0.6710495694670701, "grad_norm": 2.0338308811187744, "learning_rate": 0.0005, "loss": 2.225388526916504, "step": 5767 }, { "epoch": 0.6711659297184082, "grad_norm": 1.4972375631332397, "learning_rate": 0.0005, "loss": 3.235133647918701, "step": 5768 }, { "epoch": 0.6712822899697464, "grad_norm": 1.4230402708053589, "learning_rate": 0.0005, "loss": 2.6572461128234863, "step": 5769 }, { "epoch": 0.6713986502210845, "grad_norm": 1.6773850917816162, "learning_rate": 0.0005, "loss": 2.728079319000244, "step": 5770 }, { "epoch": 0.6715150104724226, "grad_norm": 1.1584497690200806, "learning_rate": 0.0005, "loss": 1.836025357246399, "step": 5771 }, { "epoch": 0.6716313707237608, "grad_norm": 1.2796638011932373, "learning_rate": 0.0005, "loss": 2.1267988681793213, "step": 5772 }, { "epoch": 0.6717477309750989, "grad_norm": 1.4723702669143677, "learning_rate": 0.0005, "loss": 2.6200122833251953, "step": 5773 }, { "epoch": 0.6718640912264371, "grad_norm": 1.1014084815979004, "learning_rate": 0.0005, "loss": 1.6213977336883545, "step": 5774 }, { "epoch": 0.6719804514777752, "grad_norm": 1.295823097229004, "learning_rate": 0.0005, "loss": 1.621316909790039, "step": 5775 }, { "epoch": 0.6720968117291133, "grad_norm": 1.6711491346359253, "learning_rate": 0.0005, "loss": 2.8985233306884766, "step": 5776 }, { "epoch": 0.6722131719804515, "grad_norm": 2.176978349685669, "learning_rate": 0.0005, "loss": 2.7809157371520996, "step": 5777 }, { "epoch": 0.6723295322317896, "grad_norm": 1.6082488298416138, "learning_rate": 0.0005, "loss": 3.0930399894714355, "step": 5778 }, { "epoch": 0.6724458924831278, "grad_norm": 1.3668484687805176, "learning_rate": 0.0005, "loss": 3.0126376152038574, "step": 5779 }, { "epoch": 0.6725622527344659, "grad_norm": 0.9418849349021912, "learning_rate": 0.0005, "loss": 1.9763529300689697, "step": 5780 }, { "epoch": 0.672678612985804, "grad_norm": 6.575294494628906, "learning_rate": 0.0005, "loss": 2.287993907928467, "step": 5781 }, { "epoch": 0.6727949732371422, "grad_norm": 1.7731804847717285, "learning_rate": 0.0005, "loss": 2.9669463634490967, "step": 5782 }, { "epoch": 0.6729113334884803, "grad_norm": 1.8387751579284668, "learning_rate": 0.0005, "loss": 3.0139451026916504, "step": 5783 }, { "epoch": 0.6730276937398185, "grad_norm": 1.5072792768478394, "learning_rate": 0.0005, "loss": 2.1487114429473877, "step": 5784 }, { "epoch": 0.6731440539911566, "grad_norm": 1.4223910570144653, "learning_rate": 0.0005, "loss": 2.8678765296936035, "step": 5785 }, { "epoch": 0.6732604142424947, "grad_norm": 1.6723741292953491, "learning_rate": 0.0005, "loss": 3.137526512145996, "step": 5786 }, { "epoch": 0.6733767744938329, "grad_norm": 1.9263240098953247, "learning_rate": 0.0005, "loss": 2.582573413848877, "step": 5787 }, { "epoch": 0.673493134745171, "grad_norm": 0.8587229251861572, "learning_rate": 0.0005, "loss": 1.8790829181671143, "step": 5788 }, { "epoch": 0.6736094949965092, "grad_norm": 1.3214880228042603, "learning_rate": 0.0005, "loss": 2.977937698364258, "step": 5789 }, { "epoch": 0.6737258552478473, "grad_norm": 1.5467467308044434, "learning_rate": 0.0005, "loss": 3.041999340057373, "step": 5790 }, { "epoch": 0.6738422154991854, "grad_norm": 0.590505838394165, "learning_rate": 0.0005, "loss": 1.344329595565796, "step": 5791 }, { "epoch": 0.6739585757505236, "grad_norm": 1.2818717956542969, "learning_rate": 0.0005, "loss": 2.401700019836426, "step": 5792 }, { "epoch": 0.6740749360018617, "grad_norm": 1.7029088735580444, "learning_rate": 0.0005, "loss": 2.915191411972046, "step": 5793 }, { "epoch": 0.6741912962532, "grad_norm": 1.0936570167541504, "learning_rate": 0.0005, "loss": 2.709324836730957, "step": 5794 }, { "epoch": 0.674307656504538, "grad_norm": 1.6952438354492188, "learning_rate": 0.0005, "loss": 3.254394054412842, "step": 5795 }, { "epoch": 0.6744240167558762, "grad_norm": 1.758971095085144, "learning_rate": 0.0005, "loss": 3.2515904903411865, "step": 5796 }, { "epoch": 0.6745403770072144, "grad_norm": 1.3557534217834473, "learning_rate": 0.0005, "loss": 2.2611942291259766, "step": 5797 }, { "epoch": 0.6746567372585525, "grad_norm": 2.219444751739502, "learning_rate": 0.0005, "loss": 2.3025386333465576, "step": 5798 }, { "epoch": 0.6747730975098907, "grad_norm": 1.5273324251174927, "learning_rate": 0.0005, "loss": 1.8515408039093018, "step": 5799 }, { "epoch": 0.6748894577612288, "grad_norm": 1.6265332698822021, "learning_rate": 0.0005, "loss": 2.7330482006073, "step": 5800 }, { "epoch": 0.6750058180125669, "grad_norm": 1.9439046382904053, "learning_rate": 0.0005, "loss": 2.2500510215759277, "step": 5801 }, { "epoch": 0.6751221782639051, "grad_norm": 1.368984341621399, "learning_rate": 0.0005, "loss": 2.932974338531494, "step": 5802 }, { "epoch": 0.6752385385152432, "grad_norm": 1.1821233034133911, "learning_rate": 0.0005, "loss": 2.401472568511963, "step": 5803 }, { "epoch": 0.6753548987665814, "grad_norm": 1.2220174074172974, "learning_rate": 0.0005, "loss": 2.048309564590454, "step": 5804 }, { "epoch": 0.6754712590179195, "grad_norm": 1.7159602642059326, "learning_rate": 0.0005, "loss": 3.021820306777954, "step": 5805 }, { "epoch": 0.6755876192692576, "grad_norm": 1.5191447734832764, "learning_rate": 0.0005, "loss": 2.298950672149658, "step": 5806 }, { "epoch": 0.6757039795205958, "grad_norm": 1.2807308435440063, "learning_rate": 0.0005, "loss": 2.4771852493286133, "step": 5807 }, { "epoch": 0.6758203397719339, "grad_norm": 0.7002722024917603, "learning_rate": 0.0005, "loss": 1.5449538230895996, "step": 5808 }, { "epoch": 0.675936700023272, "grad_norm": 1.5921756029129028, "learning_rate": 0.0005, "loss": 2.0702974796295166, "step": 5809 }, { "epoch": 0.6760530602746102, "grad_norm": 2.537916421890259, "learning_rate": 0.0005, "loss": 2.629786968231201, "step": 5810 }, { "epoch": 0.6761694205259483, "grad_norm": 1.342359185218811, "learning_rate": 0.0005, "loss": 2.1851232051849365, "step": 5811 }, { "epoch": 0.6762857807772865, "grad_norm": 1.3019535541534424, "learning_rate": 0.0005, "loss": 2.5536603927612305, "step": 5812 }, { "epoch": 0.6764021410286246, "grad_norm": 1.2408889532089233, "learning_rate": 0.0005, "loss": 2.6040873527526855, "step": 5813 }, { "epoch": 0.6765185012799627, "grad_norm": 1.8935165405273438, "learning_rate": 0.0005, "loss": 2.612715244293213, "step": 5814 }, { "epoch": 0.6766348615313009, "grad_norm": 1.6923856735229492, "learning_rate": 0.0005, "loss": 2.6793971061706543, "step": 5815 }, { "epoch": 0.676751221782639, "grad_norm": 1.4440909624099731, "learning_rate": 0.0005, "loss": 1.6895403861999512, "step": 5816 }, { "epoch": 0.6768675820339772, "grad_norm": 1.0212359428405762, "learning_rate": 0.0005, "loss": 2.1510701179504395, "step": 5817 }, { "epoch": 0.6769839422853153, "grad_norm": 1.7115870714187622, "learning_rate": 0.0005, "loss": 2.335576295852661, "step": 5818 }, { "epoch": 0.6771003025366534, "grad_norm": 1.556394338607788, "learning_rate": 0.0005, "loss": 2.439974784851074, "step": 5819 }, { "epoch": 0.6772166627879916, "grad_norm": 2.030867576599121, "learning_rate": 0.0005, "loss": 2.7316107749938965, "step": 5820 }, { "epoch": 0.6773330230393297, "grad_norm": 1.6879005432128906, "learning_rate": 0.0005, "loss": 2.5427393913269043, "step": 5821 }, { "epoch": 0.677449383290668, "grad_norm": 1.4189468622207642, "learning_rate": 0.0005, "loss": 2.719058036804199, "step": 5822 }, { "epoch": 0.677565743542006, "grad_norm": 0.9950214624404907, "learning_rate": 0.0005, "loss": 1.5734405517578125, "step": 5823 }, { "epoch": 0.6776821037933441, "grad_norm": 1.0112310647964478, "learning_rate": 0.0005, "loss": 2.1694657802581787, "step": 5824 }, { "epoch": 0.6777984640446824, "grad_norm": 1.1840986013412476, "learning_rate": 0.0005, "loss": 2.2462964057922363, "step": 5825 }, { "epoch": 0.6779148242960205, "grad_norm": 0.9646816253662109, "learning_rate": 0.0005, "loss": 2.94950532913208, "step": 5826 }, { "epoch": 0.6780311845473587, "grad_norm": 1.3945322036743164, "learning_rate": 0.0005, "loss": 2.20387864112854, "step": 5827 }, { "epoch": 0.6781475447986968, "grad_norm": 1.0788975954055786, "learning_rate": 0.0005, "loss": 2.2149744033813477, "step": 5828 }, { "epoch": 0.6782639050500349, "grad_norm": 1.5778807401657104, "learning_rate": 0.0005, "loss": 2.9753613471984863, "step": 5829 }, { "epoch": 0.6783802653013731, "grad_norm": 1.0964319705963135, "learning_rate": 0.0005, "loss": 1.5518062114715576, "step": 5830 }, { "epoch": 0.6784966255527112, "grad_norm": 0.8357813954353333, "learning_rate": 0.0005, "loss": 1.989546537399292, "step": 5831 }, { "epoch": 0.6786129858040494, "grad_norm": 1.1219457387924194, "learning_rate": 0.0005, "loss": 1.8898413181304932, "step": 5832 }, { "epoch": 0.6787293460553875, "grad_norm": 1.3890869617462158, "learning_rate": 0.0005, "loss": 3.1236696243286133, "step": 5833 }, { "epoch": 0.6788457063067256, "grad_norm": 1.4758809804916382, "learning_rate": 0.0005, "loss": 2.9093031883239746, "step": 5834 }, { "epoch": 0.6789620665580638, "grad_norm": 3.440474271774292, "learning_rate": 0.0005, "loss": 2.923182487487793, "step": 5835 }, { "epoch": 0.6790784268094019, "grad_norm": 1.2479203939437866, "learning_rate": 0.0005, "loss": 2.0942649841308594, "step": 5836 }, { "epoch": 0.6791947870607401, "grad_norm": 1.00099778175354, "learning_rate": 0.0005, "loss": 2.54494571685791, "step": 5837 }, { "epoch": 0.6793111473120782, "grad_norm": 1.3278270959854126, "learning_rate": 0.0005, "loss": 2.9605565071105957, "step": 5838 }, { "epoch": 0.6794275075634163, "grad_norm": 1.4562578201293945, "learning_rate": 0.0005, "loss": 2.418830633163452, "step": 5839 }, { "epoch": 0.6795438678147545, "grad_norm": 0.4296720623970032, "learning_rate": 0.0005, "loss": 1.1900572776794434, "step": 5840 }, { "epoch": 0.6796602280660926, "grad_norm": 1.4692912101745605, "learning_rate": 0.0005, "loss": 3.0633726119995117, "step": 5841 }, { "epoch": 0.6797765883174308, "grad_norm": 1.7014390230178833, "learning_rate": 0.0005, "loss": 2.621772527694702, "step": 5842 }, { "epoch": 0.6798929485687689, "grad_norm": 1.3804104328155518, "learning_rate": 0.0005, "loss": 2.159552574157715, "step": 5843 }, { "epoch": 0.680009308820107, "grad_norm": 1.2492953538894653, "learning_rate": 0.0005, "loss": 2.9713337421417236, "step": 5844 }, { "epoch": 0.6801256690714452, "grad_norm": 1.4795887470245361, "learning_rate": 0.0005, "loss": 2.8478479385375977, "step": 5845 }, { "epoch": 0.6802420293227833, "grad_norm": 1.6529765129089355, "learning_rate": 0.0005, "loss": 2.7835941314697266, "step": 5846 }, { "epoch": 0.6803583895741214, "grad_norm": 2.3043053150177, "learning_rate": 0.0005, "loss": 2.9661824703216553, "step": 5847 }, { "epoch": 0.6804747498254596, "grad_norm": 1.0531642436981201, "learning_rate": 0.0005, "loss": 1.9603431224822998, "step": 5848 }, { "epoch": 0.6805911100767977, "grad_norm": 1.3867627382278442, "learning_rate": 0.0005, "loss": 2.654092788696289, "step": 5849 }, { "epoch": 0.6807074703281359, "grad_norm": 1.8866138458251953, "learning_rate": 0.0005, "loss": 3.1615982055664062, "step": 5850 }, { "epoch": 0.680823830579474, "grad_norm": 1.2940906286239624, "learning_rate": 0.0005, "loss": 2.483088493347168, "step": 5851 }, { "epoch": 0.6809401908308121, "grad_norm": 1.5540351867675781, "learning_rate": 0.0005, "loss": 2.9620819091796875, "step": 5852 }, { "epoch": 0.6810565510821504, "grad_norm": 1.8478572368621826, "learning_rate": 0.0005, "loss": 2.2621893882751465, "step": 5853 }, { "epoch": 0.6811729113334885, "grad_norm": 1.324630856513977, "learning_rate": 0.0005, "loss": 1.8164398670196533, "step": 5854 }, { "epoch": 0.6812892715848267, "grad_norm": 1.8967022895812988, "learning_rate": 0.0005, "loss": 2.3709096908569336, "step": 5855 }, { "epoch": 0.6814056318361648, "grad_norm": 1.3802231550216675, "learning_rate": 0.0005, "loss": 2.012486457824707, "step": 5856 }, { "epoch": 0.6815219920875029, "grad_norm": 1.5615581274032593, "learning_rate": 0.0005, "loss": 2.518498420715332, "step": 5857 }, { "epoch": 0.6816383523388411, "grad_norm": 1.6586326360702515, "learning_rate": 0.0005, "loss": 2.5999526977539062, "step": 5858 }, { "epoch": 0.6817547125901792, "grad_norm": 1.4318983554840088, "learning_rate": 0.0005, "loss": 2.570573091506958, "step": 5859 }, { "epoch": 0.6818710728415174, "grad_norm": 3.920058012008667, "learning_rate": 0.0005, "loss": 2.4058914184570312, "step": 5860 }, { "epoch": 0.6819874330928555, "grad_norm": 1.7468093633651733, "learning_rate": 0.0005, "loss": 3.0631723403930664, "step": 5861 }, { "epoch": 0.6821037933441936, "grad_norm": 2.2558345794677734, "learning_rate": 0.0005, "loss": 2.7134931087493896, "step": 5862 }, { "epoch": 0.6822201535955318, "grad_norm": 1.8738856315612793, "learning_rate": 0.0005, "loss": 2.537838935852051, "step": 5863 }, { "epoch": 0.6823365138468699, "grad_norm": 2.680204391479492, "learning_rate": 0.0005, "loss": 2.6248483657836914, "step": 5864 }, { "epoch": 0.6824528740982081, "grad_norm": 1.1344175338745117, "learning_rate": 0.0005, "loss": 1.8947899341583252, "step": 5865 }, { "epoch": 0.6825692343495462, "grad_norm": 2.116976261138916, "learning_rate": 0.0005, "loss": 2.699202060699463, "step": 5866 }, { "epoch": 0.6826855946008843, "grad_norm": 2.2565701007843018, "learning_rate": 0.0005, "loss": 2.4187369346618652, "step": 5867 }, { "epoch": 0.6828019548522225, "grad_norm": 1.6081483364105225, "learning_rate": 0.0005, "loss": 2.6570067405700684, "step": 5868 }, { "epoch": 0.6829183151035606, "grad_norm": 2.1385252475738525, "learning_rate": 0.0005, "loss": 3.322460889816284, "step": 5869 }, { "epoch": 0.6830346753548988, "grad_norm": 2.993427038192749, "learning_rate": 0.0005, "loss": 3.653501510620117, "step": 5870 }, { "epoch": 0.6831510356062369, "grad_norm": 1.4087588787078857, "learning_rate": 0.0005, "loss": 2.3167898654937744, "step": 5871 }, { "epoch": 0.683267395857575, "grad_norm": 1.4455721378326416, "learning_rate": 0.0005, "loss": 2.6799259185791016, "step": 5872 }, { "epoch": 0.6833837561089132, "grad_norm": 1.5301705598831177, "learning_rate": 0.0005, "loss": 1.6906282901763916, "step": 5873 }, { "epoch": 0.6835001163602513, "grad_norm": 1.9841523170471191, "learning_rate": 0.0005, "loss": 2.5631906986236572, "step": 5874 }, { "epoch": 0.6836164766115895, "grad_norm": 1.4993690252304077, "learning_rate": 0.0005, "loss": 2.0895798206329346, "step": 5875 }, { "epoch": 0.6837328368629276, "grad_norm": 2.1941702365875244, "learning_rate": 0.0005, "loss": 2.9731059074401855, "step": 5876 }, { "epoch": 0.6838491971142657, "grad_norm": 1.5068162679672241, "learning_rate": 0.0005, "loss": 3.112147092819214, "step": 5877 }, { "epoch": 0.6839655573656039, "grad_norm": 1.5360915660858154, "learning_rate": 0.0005, "loss": 2.239748954772949, "step": 5878 }, { "epoch": 0.684081917616942, "grad_norm": 1.2252835035324097, "learning_rate": 0.0005, "loss": 2.661499261856079, "step": 5879 }, { "epoch": 0.6841982778682802, "grad_norm": 2.500978708267212, "learning_rate": 0.0005, "loss": 2.7722580432891846, "step": 5880 }, { "epoch": 0.6843146381196183, "grad_norm": 1.6202787160873413, "learning_rate": 0.0005, "loss": 2.787306785583496, "step": 5881 }, { "epoch": 0.6844309983709564, "grad_norm": 0.5401080250740051, "learning_rate": 0.0005, "loss": 1.5158467292785645, "step": 5882 }, { "epoch": 0.6845473586222947, "grad_norm": 1.0660529136657715, "learning_rate": 0.0005, "loss": 1.6761846542358398, "step": 5883 }, { "epoch": 0.6846637188736328, "grad_norm": 1.3933814764022827, "learning_rate": 0.0005, "loss": 2.6263651847839355, "step": 5884 }, { "epoch": 0.684780079124971, "grad_norm": 1.3089683055877686, "learning_rate": 0.0005, "loss": 1.604140043258667, "step": 5885 }, { "epoch": 0.6848964393763091, "grad_norm": 0.8784886598587036, "learning_rate": 0.0005, "loss": 1.9430776834487915, "step": 5886 }, { "epoch": 0.6850127996276472, "grad_norm": 1.1612101793289185, "learning_rate": 0.0005, "loss": 2.1294643878936768, "step": 5887 }, { "epoch": 0.6851291598789854, "grad_norm": 1.9186514616012573, "learning_rate": 0.0005, "loss": 2.6659433841705322, "step": 5888 }, { "epoch": 0.6852455201303235, "grad_norm": 1.477444052696228, "learning_rate": 0.0005, "loss": 2.3009464740753174, "step": 5889 }, { "epoch": 0.6853618803816616, "grad_norm": 0.8024328351020813, "learning_rate": 0.0005, "loss": 1.3502027988433838, "step": 5890 }, { "epoch": 0.6854782406329998, "grad_norm": 1.6738908290863037, "learning_rate": 0.0005, "loss": 2.2155938148498535, "step": 5891 }, { "epoch": 0.6855946008843379, "grad_norm": 1.1363096237182617, "learning_rate": 0.0005, "loss": 2.0935816764831543, "step": 5892 }, { "epoch": 0.6857109611356761, "grad_norm": 1.6262520551681519, "learning_rate": 0.0005, "loss": 2.5624942779541016, "step": 5893 }, { "epoch": 0.6858273213870142, "grad_norm": 1.419840931892395, "learning_rate": 0.0005, "loss": 2.2350785732269287, "step": 5894 }, { "epoch": 0.6859436816383523, "grad_norm": 3.5597519874572754, "learning_rate": 0.0005, "loss": 2.892277240753174, "step": 5895 }, { "epoch": 0.6860600418896905, "grad_norm": 0.98110431432724, "learning_rate": 0.0005, "loss": 1.442115068435669, "step": 5896 }, { "epoch": 0.6861764021410286, "grad_norm": 1.8107197284698486, "learning_rate": 0.0005, "loss": 2.7755649089813232, "step": 5897 }, { "epoch": 0.6862927623923668, "grad_norm": 1.5904796123504639, "learning_rate": 0.0005, "loss": 2.463531494140625, "step": 5898 }, { "epoch": 0.6864091226437049, "grad_norm": 0.9551846981048584, "learning_rate": 0.0005, "loss": 2.0036253929138184, "step": 5899 }, { "epoch": 0.686525482895043, "grad_norm": 1.555245041847229, "learning_rate": 0.0005, "loss": 3.3955936431884766, "step": 5900 }, { "epoch": 0.6866418431463812, "grad_norm": 0.7832594513893127, "learning_rate": 0.0005, "loss": 1.4400014877319336, "step": 5901 }, { "epoch": 0.6867582033977193, "grad_norm": 1.4286037683486938, "learning_rate": 0.0005, "loss": 2.3751466274261475, "step": 5902 }, { "epoch": 0.6868745636490575, "grad_norm": 0.8479102253913879, "learning_rate": 0.0005, "loss": 1.528092622756958, "step": 5903 }, { "epoch": 0.6869909239003956, "grad_norm": 1.4065452814102173, "learning_rate": 0.0005, "loss": 2.3759214878082275, "step": 5904 }, { "epoch": 0.6871072841517337, "grad_norm": 1.9709367752075195, "learning_rate": 0.0005, "loss": 3.1597795486450195, "step": 5905 }, { "epoch": 0.6872236444030719, "grad_norm": 1.0205103158950806, "learning_rate": 0.0005, "loss": 2.1638009548187256, "step": 5906 }, { "epoch": 0.68734000465441, "grad_norm": 2.570578098297119, "learning_rate": 0.0005, "loss": 2.7100634574890137, "step": 5907 }, { "epoch": 0.6874563649057482, "grad_norm": 2.2404375076293945, "learning_rate": 0.0005, "loss": 2.6850342750549316, "step": 5908 }, { "epoch": 0.6875727251570863, "grad_norm": 1.266059398651123, "learning_rate": 0.0005, "loss": 2.355720043182373, "step": 5909 }, { "epoch": 0.6876890854084244, "grad_norm": 1.6083931922912598, "learning_rate": 0.0005, "loss": 2.441169500350952, "step": 5910 }, { "epoch": 0.6878054456597626, "grad_norm": 1.802720308303833, "learning_rate": 0.0005, "loss": 2.3157951831817627, "step": 5911 }, { "epoch": 0.6879218059111007, "grad_norm": 1.4173146486282349, "learning_rate": 0.0005, "loss": 2.4384117126464844, "step": 5912 }, { "epoch": 0.688038166162439, "grad_norm": 1.5321682691574097, "learning_rate": 0.0005, "loss": 2.508512258529663, "step": 5913 }, { "epoch": 0.6881545264137771, "grad_norm": 1.3895548582077026, "learning_rate": 0.0005, "loss": 2.9651880264282227, "step": 5914 }, { "epoch": 0.6882708866651152, "grad_norm": 0.794955849647522, "learning_rate": 0.0005, "loss": 1.5272051095962524, "step": 5915 }, { "epoch": 0.6883872469164534, "grad_norm": 0.8213961124420166, "learning_rate": 0.0005, "loss": 1.4333548545837402, "step": 5916 }, { "epoch": 0.6885036071677915, "grad_norm": 2.5309693813323975, "learning_rate": 0.0005, "loss": 2.545588493347168, "step": 5917 }, { "epoch": 0.6886199674191297, "grad_norm": 0.8910328149795532, "learning_rate": 0.0005, "loss": 2.0618300437927246, "step": 5918 }, { "epoch": 0.6887363276704678, "grad_norm": 3.5512428283691406, "learning_rate": 0.0005, "loss": 2.3538904190063477, "step": 5919 }, { "epoch": 0.6888526879218059, "grad_norm": 1.053330421447754, "learning_rate": 0.0005, "loss": 2.13797664642334, "step": 5920 }, { "epoch": 0.6889690481731441, "grad_norm": 1.1737395524978638, "learning_rate": 0.0005, "loss": 3.008228302001953, "step": 5921 }, { "epoch": 0.6890854084244822, "grad_norm": 1.0994199514389038, "learning_rate": 0.0005, "loss": 2.5500102043151855, "step": 5922 }, { "epoch": 0.6892017686758204, "grad_norm": 0.951357364654541, "learning_rate": 0.0005, "loss": 1.8940402269363403, "step": 5923 }, { "epoch": 0.6893181289271585, "grad_norm": 1.481738567352295, "learning_rate": 0.0005, "loss": 2.363107442855835, "step": 5924 }, { "epoch": 0.6894344891784966, "grad_norm": 1.2032790184020996, "learning_rate": 0.0005, "loss": 2.148000717163086, "step": 5925 }, { "epoch": 0.6895508494298348, "grad_norm": 1.6748126745224, "learning_rate": 0.0005, "loss": 2.6163947582244873, "step": 5926 }, { "epoch": 0.6896672096811729, "grad_norm": 1.514495611190796, "learning_rate": 0.0005, "loss": 2.8593406677246094, "step": 5927 }, { "epoch": 0.6897835699325111, "grad_norm": 1.9211064577102661, "learning_rate": 0.0005, "loss": 2.892190933227539, "step": 5928 }, { "epoch": 0.6898999301838492, "grad_norm": 1.4445286989212036, "learning_rate": 0.0005, "loss": 2.6406168937683105, "step": 5929 }, { "epoch": 0.6900162904351873, "grad_norm": 1.4631818532943726, "learning_rate": 0.0005, "loss": 2.2416322231292725, "step": 5930 }, { "epoch": 0.6901326506865255, "grad_norm": 1.229177474975586, "learning_rate": 0.0005, "loss": 2.4587042331695557, "step": 5931 }, { "epoch": 0.6902490109378636, "grad_norm": 1.5905272960662842, "learning_rate": 0.0005, "loss": 2.824410915374756, "step": 5932 }, { "epoch": 0.6903653711892017, "grad_norm": 1.3809694051742554, "learning_rate": 0.0005, "loss": 2.2896759510040283, "step": 5933 }, { "epoch": 0.6904817314405399, "grad_norm": 1.8084416389465332, "learning_rate": 0.0005, "loss": 3.092419385910034, "step": 5934 }, { "epoch": 0.690598091691878, "grad_norm": 1.0655261278152466, "learning_rate": 0.0005, "loss": 2.860598564147949, "step": 5935 }, { "epoch": 0.6907144519432162, "grad_norm": 1.335268497467041, "learning_rate": 0.0005, "loss": 2.603301525115967, "step": 5936 }, { "epoch": 0.6908308121945543, "grad_norm": 0.7929680943489075, "learning_rate": 0.0005, "loss": 1.9470775127410889, "step": 5937 }, { "epoch": 0.6909471724458924, "grad_norm": 2.2568533420562744, "learning_rate": 0.0005, "loss": 2.935372829437256, "step": 5938 }, { "epoch": 0.6910635326972306, "grad_norm": 1.5772067308425903, "learning_rate": 0.0005, "loss": 2.539625406265259, "step": 5939 }, { "epoch": 0.6911798929485687, "grad_norm": 0.9914002418518066, "learning_rate": 0.0005, "loss": 1.7660880088806152, "step": 5940 }, { "epoch": 0.691296253199907, "grad_norm": 0.8500457406044006, "learning_rate": 0.0005, "loss": 2.1770589351654053, "step": 5941 }, { "epoch": 0.691412613451245, "grad_norm": 1.9860116243362427, "learning_rate": 0.0005, "loss": 2.813594341278076, "step": 5942 }, { "epoch": 0.6915289737025831, "grad_norm": 1.3013404607772827, "learning_rate": 0.0005, "loss": 2.2539567947387695, "step": 5943 }, { "epoch": 0.6916453339539214, "grad_norm": 1.1923818588256836, "learning_rate": 0.0005, "loss": 2.306234359741211, "step": 5944 }, { "epoch": 0.6917616942052595, "grad_norm": 1.4701259136199951, "learning_rate": 0.0005, "loss": 2.690040111541748, "step": 5945 }, { "epoch": 0.6918780544565977, "grad_norm": 1.6312838792800903, "learning_rate": 0.0005, "loss": 2.472292900085449, "step": 5946 }, { "epoch": 0.6919944147079358, "grad_norm": 2.1037943363189697, "learning_rate": 0.0005, "loss": 2.767679214477539, "step": 5947 }, { "epoch": 0.6921107749592739, "grad_norm": 1.6543244123458862, "learning_rate": 0.0005, "loss": 2.5396759510040283, "step": 5948 }, { "epoch": 0.6922271352106121, "grad_norm": 1.1377652883529663, "learning_rate": 0.0005, "loss": 2.7152371406555176, "step": 5949 }, { "epoch": 0.6923434954619502, "grad_norm": 1.8394851684570312, "learning_rate": 0.0005, "loss": 2.464202880859375, "step": 5950 }, { "epoch": 0.6924598557132884, "grad_norm": 1.617811918258667, "learning_rate": 0.0005, "loss": 2.358795166015625, "step": 5951 }, { "epoch": 0.6925762159646265, "grad_norm": 1.922484040260315, "learning_rate": 0.0005, "loss": 3.0807347297668457, "step": 5952 }, { "epoch": 0.6926925762159646, "grad_norm": 1.0673787593841553, "learning_rate": 0.0005, "loss": 2.8282947540283203, "step": 5953 }, { "epoch": 0.6928089364673028, "grad_norm": 0.6294093728065491, "learning_rate": 0.0005, "loss": 1.7050623893737793, "step": 5954 }, { "epoch": 0.6929252967186409, "grad_norm": 0.7682778239250183, "learning_rate": 0.0005, "loss": 2.5516324043273926, "step": 5955 }, { "epoch": 0.6930416569699791, "grad_norm": 1.752000331878662, "learning_rate": 0.0005, "loss": 2.475597381591797, "step": 5956 }, { "epoch": 0.6931580172213172, "grad_norm": 1.5037742853164673, "learning_rate": 0.0005, "loss": 2.340517044067383, "step": 5957 }, { "epoch": 0.6932743774726553, "grad_norm": 1.0898534059524536, "learning_rate": 0.0005, "loss": 2.2470107078552246, "step": 5958 }, { "epoch": 0.6933907377239935, "grad_norm": 1.1271734237670898, "learning_rate": 0.0005, "loss": 2.8152875900268555, "step": 5959 }, { "epoch": 0.6935070979753316, "grad_norm": 1.3760102987289429, "learning_rate": 0.0005, "loss": 3.1306402683258057, "step": 5960 }, { "epoch": 0.6936234582266698, "grad_norm": 1.6449787616729736, "learning_rate": 0.0005, "loss": 2.5461459159851074, "step": 5961 }, { "epoch": 0.6937398184780079, "grad_norm": 1.231703758239746, "learning_rate": 0.0005, "loss": 2.495398998260498, "step": 5962 }, { "epoch": 0.693856178729346, "grad_norm": 1.4724735021591187, "learning_rate": 0.0005, "loss": 2.329688549041748, "step": 5963 }, { "epoch": 0.6939725389806842, "grad_norm": 2.665243625640869, "learning_rate": 0.0005, "loss": 1.6679989099502563, "step": 5964 }, { "epoch": 0.6940888992320223, "grad_norm": 1.7936514616012573, "learning_rate": 0.0005, "loss": 2.8526625633239746, "step": 5965 }, { "epoch": 0.6942052594833605, "grad_norm": 2.1308250427246094, "learning_rate": 0.0005, "loss": 2.907616138458252, "step": 5966 }, { "epoch": 0.6943216197346986, "grad_norm": 1.6300785541534424, "learning_rate": 0.0005, "loss": 2.4341824054718018, "step": 5967 }, { "epoch": 0.6944379799860367, "grad_norm": 0.935076892375946, "learning_rate": 0.0005, "loss": 1.4625368118286133, "step": 5968 }, { "epoch": 0.694554340237375, "grad_norm": 1.2352511882781982, "learning_rate": 0.0005, "loss": 2.2271265983581543, "step": 5969 }, { "epoch": 0.694670700488713, "grad_norm": 1.5270664691925049, "learning_rate": 0.0005, "loss": 2.1695003509521484, "step": 5970 }, { "epoch": 0.6947870607400511, "grad_norm": 1.2835191488265991, "learning_rate": 0.0005, "loss": 2.87851881980896, "step": 5971 }, { "epoch": 0.6949034209913894, "grad_norm": 1.6306633949279785, "learning_rate": 0.0005, "loss": 2.2174205780029297, "step": 5972 }, { "epoch": 0.6950197812427275, "grad_norm": 0.8608205318450928, "learning_rate": 0.0005, "loss": 2.1465423107147217, "step": 5973 }, { "epoch": 0.6951361414940657, "grad_norm": 1.5844593048095703, "learning_rate": 0.0005, "loss": 3.161649703979492, "step": 5974 }, { "epoch": 0.6952525017454038, "grad_norm": 3.063802719116211, "learning_rate": 0.0005, "loss": 3.255727767944336, "step": 5975 }, { "epoch": 0.6953688619967419, "grad_norm": 1.3834360837936401, "learning_rate": 0.0005, "loss": 2.130295753479004, "step": 5976 }, { "epoch": 0.6954852222480801, "grad_norm": 1.907336711883545, "learning_rate": 0.0005, "loss": 2.182619094848633, "step": 5977 }, { "epoch": 0.6956015824994182, "grad_norm": 0.8350352048873901, "learning_rate": 0.0005, "loss": 1.5304515361785889, "step": 5978 }, { "epoch": 0.6957179427507564, "grad_norm": 1.4188247919082642, "learning_rate": 0.0005, "loss": 2.076530933380127, "step": 5979 }, { "epoch": 0.6958343030020945, "grad_norm": 1.2554702758789062, "learning_rate": 0.0005, "loss": 2.203875780105591, "step": 5980 }, { "epoch": 0.6959506632534326, "grad_norm": 1.6570264101028442, "learning_rate": 0.0005, "loss": 2.3842830657958984, "step": 5981 }, { "epoch": 0.6960670235047708, "grad_norm": 1.1876636743545532, "learning_rate": 0.0005, "loss": 1.9967749118804932, "step": 5982 }, { "epoch": 0.6961833837561089, "grad_norm": 1.8385467529296875, "learning_rate": 0.0005, "loss": 2.2731099128723145, "step": 5983 }, { "epoch": 0.6962997440074471, "grad_norm": 0.9421830177307129, "learning_rate": 0.0005, "loss": 1.6027764081954956, "step": 5984 }, { "epoch": 0.6964161042587852, "grad_norm": 1.3387091159820557, "learning_rate": 0.0005, "loss": 2.4021658897399902, "step": 5985 }, { "epoch": 0.6965324645101233, "grad_norm": 1.4544432163238525, "learning_rate": 0.0005, "loss": 2.2143330574035645, "step": 5986 }, { "epoch": 0.6966488247614615, "grad_norm": 1.8509852886199951, "learning_rate": 0.0005, "loss": 2.2682101726531982, "step": 5987 }, { "epoch": 0.6967651850127996, "grad_norm": 1.2008808851242065, "learning_rate": 0.0005, "loss": 1.9313271045684814, "step": 5988 }, { "epoch": 0.6968815452641378, "grad_norm": 1.6428011655807495, "learning_rate": 0.0005, "loss": 2.598947286605835, "step": 5989 }, { "epoch": 0.6969979055154759, "grad_norm": 1.6773881912231445, "learning_rate": 0.0005, "loss": 2.838533878326416, "step": 5990 }, { "epoch": 0.697114265766814, "grad_norm": 0.9896075129508972, "learning_rate": 0.0005, "loss": 2.4567043781280518, "step": 5991 }, { "epoch": 0.6972306260181522, "grad_norm": 1.3662759065628052, "learning_rate": 0.0005, "loss": 3.139981508255005, "step": 5992 }, { "epoch": 0.6973469862694903, "grad_norm": 1.2778316736221313, "learning_rate": 0.0005, "loss": 2.175778865814209, "step": 5993 }, { "epoch": 0.6974633465208285, "grad_norm": 1.8210952281951904, "learning_rate": 0.0005, "loss": 3.356072425842285, "step": 5994 }, { "epoch": 0.6975797067721666, "grad_norm": 1.4049360752105713, "learning_rate": 0.0005, "loss": 2.3289055824279785, "step": 5995 }, { "epoch": 0.6976960670235047, "grad_norm": 0.9693872332572937, "learning_rate": 0.0005, "loss": 1.6214261054992676, "step": 5996 }, { "epoch": 0.6978124272748429, "grad_norm": 2.3735222816467285, "learning_rate": 0.0005, "loss": 2.2656679153442383, "step": 5997 }, { "epoch": 0.697928787526181, "grad_norm": 1.498732089996338, "learning_rate": 0.0005, "loss": 3.3032310009002686, "step": 5998 }, { "epoch": 0.6980451477775192, "grad_norm": 1.607500672340393, "learning_rate": 0.0005, "loss": 3.460350275039673, "step": 5999 }, { "epoch": 0.6981615080288573, "grad_norm": 0.9576772451400757, "learning_rate": 0.0005, "loss": 2.4021897315979004, "step": 6000 }, { "epoch": 0.6982778682801954, "grad_norm": 1.5067192316055298, "learning_rate": 0.0005, "loss": 2.7401206493377686, "step": 6001 }, { "epoch": 0.6983942285315337, "grad_norm": 1.0756802558898926, "learning_rate": 0.0005, "loss": 1.315260648727417, "step": 6002 }, { "epoch": 0.6985105887828718, "grad_norm": 1.3054972887039185, "learning_rate": 0.0005, "loss": 2.7328882217407227, "step": 6003 }, { "epoch": 0.69862694903421, "grad_norm": 1.2105051279067993, "learning_rate": 0.0005, "loss": 2.883401870727539, "step": 6004 }, { "epoch": 0.6987433092855481, "grad_norm": 1.1554924249649048, "learning_rate": 0.0005, "loss": 1.981079339981079, "step": 6005 }, { "epoch": 0.6988596695368862, "grad_norm": 1.2384835481643677, "learning_rate": 0.0005, "loss": 2.8445329666137695, "step": 6006 }, { "epoch": 0.6989760297882244, "grad_norm": 2.09493088722229, "learning_rate": 0.0005, "loss": 2.505587100982666, "step": 6007 }, { "epoch": 0.6990923900395625, "grad_norm": 1.4198592901229858, "learning_rate": 0.0005, "loss": 2.9672627449035645, "step": 6008 }, { "epoch": 0.6992087502909007, "grad_norm": 1.5167189836502075, "learning_rate": 0.0005, "loss": 2.4338178634643555, "step": 6009 }, { "epoch": 0.6993251105422388, "grad_norm": 1.4299405813217163, "learning_rate": 0.0005, "loss": 2.711125373840332, "step": 6010 }, { "epoch": 0.6994414707935769, "grad_norm": 1.8460235595703125, "learning_rate": 0.0005, "loss": 1.7921459674835205, "step": 6011 }, { "epoch": 0.6995578310449151, "grad_norm": 1.5603699684143066, "learning_rate": 0.0005, "loss": 2.205544948577881, "step": 6012 }, { "epoch": 0.6996741912962532, "grad_norm": 1.8517025709152222, "learning_rate": 0.0005, "loss": 2.1833951473236084, "step": 6013 }, { "epoch": 0.6997905515475913, "grad_norm": 1.907700538635254, "learning_rate": 0.0005, "loss": 2.8758609294891357, "step": 6014 }, { "epoch": 0.6999069117989295, "grad_norm": 0.9467036724090576, "learning_rate": 0.0005, "loss": 2.242992401123047, "step": 6015 }, { "epoch": 0.7000232720502676, "grad_norm": 1.1614235639572144, "learning_rate": 0.0005, "loss": 2.078096866607666, "step": 6016 }, { "epoch": 0.7001396323016058, "grad_norm": 1.6674795150756836, "learning_rate": 0.0005, "loss": 2.831552267074585, "step": 6017 }, { "epoch": 0.7002559925529439, "grad_norm": 1.328433871269226, "learning_rate": 0.0005, "loss": 2.3444347381591797, "step": 6018 }, { "epoch": 0.700372352804282, "grad_norm": 1.737181544303894, "learning_rate": 0.0005, "loss": 2.497631549835205, "step": 6019 }, { "epoch": 0.7004887130556202, "grad_norm": 2.0384912490844727, "learning_rate": 0.0005, "loss": 2.91593074798584, "step": 6020 }, { "epoch": 0.7006050733069583, "grad_norm": 1.128587007522583, "learning_rate": 0.0005, "loss": 1.5485999584197998, "step": 6021 }, { "epoch": 0.7007214335582965, "grad_norm": 1.2275300025939941, "learning_rate": 0.0005, "loss": 2.4307713508605957, "step": 6022 }, { "epoch": 0.7008377938096346, "grad_norm": 1.3132097721099854, "learning_rate": 0.0005, "loss": 2.276031732559204, "step": 6023 }, { "epoch": 0.7009541540609727, "grad_norm": 1.6512175798416138, "learning_rate": 0.0005, "loss": 2.5689496994018555, "step": 6024 }, { "epoch": 0.7010705143123109, "grad_norm": 0.7823888659477234, "learning_rate": 0.0005, "loss": 1.7162203788757324, "step": 6025 }, { "epoch": 0.701186874563649, "grad_norm": 1.3020271062850952, "learning_rate": 0.0005, "loss": 2.7085132598876953, "step": 6026 }, { "epoch": 0.7013032348149872, "grad_norm": 1.1995576620101929, "learning_rate": 0.0005, "loss": 2.3974485397338867, "step": 6027 }, { "epoch": 0.7014195950663253, "grad_norm": 1.2534310817718506, "learning_rate": 0.0005, "loss": 2.2569398880004883, "step": 6028 }, { "epoch": 0.7015359553176634, "grad_norm": 1.2099671363830566, "learning_rate": 0.0005, "loss": 1.739292860031128, "step": 6029 }, { "epoch": 0.7016523155690016, "grad_norm": 1.1741859912872314, "learning_rate": 0.0005, "loss": 2.319197654724121, "step": 6030 }, { "epoch": 0.7017686758203397, "grad_norm": 1.4474129676818848, "learning_rate": 0.0005, "loss": 2.3268487453460693, "step": 6031 }, { "epoch": 0.701885036071678, "grad_norm": 1.4032062292099, "learning_rate": 0.0005, "loss": 2.2076845169067383, "step": 6032 }, { "epoch": 0.7020013963230161, "grad_norm": 1.5324387550354004, "learning_rate": 0.0005, "loss": 3.3264834880828857, "step": 6033 }, { "epoch": 0.7021177565743542, "grad_norm": 1.1664276123046875, "learning_rate": 0.0005, "loss": 1.8626863956451416, "step": 6034 }, { "epoch": 0.7022341168256924, "grad_norm": 1.3202979564666748, "learning_rate": 0.0005, "loss": 3.2990288734436035, "step": 6035 }, { "epoch": 0.7023504770770305, "grad_norm": 1.0094172954559326, "learning_rate": 0.0005, "loss": 2.0181305408477783, "step": 6036 }, { "epoch": 0.7024668373283687, "grad_norm": 1.490119218826294, "learning_rate": 0.0005, "loss": 2.488954782485962, "step": 6037 }, { "epoch": 0.7025831975797068, "grad_norm": 1.324607491493225, "learning_rate": 0.0005, "loss": 2.2909350395202637, "step": 6038 }, { "epoch": 0.7026995578310449, "grad_norm": 1.5712872743606567, "learning_rate": 0.0005, "loss": 2.4589591026306152, "step": 6039 }, { "epoch": 0.7028159180823831, "grad_norm": 1.4950883388519287, "learning_rate": 0.0005, "loss": 2.587160348892212, "step": 6040 }, { "epoch": 0.7029322783337212, "grad_norm": 1.2174339294433594, "learning_rate": 0.0005, "loss": 2.5303525924682617, "step": 6041 }, { "epoch": 0.7030486385850594, "grad_norm": 1.2746613025665283, "learning_rate": 0.0005, "loss": 2.936981678009033, "step": 6042 }, { "epoch": 0.7031649988363975, "grad_norm": 1.7107276916503906, "learning_rate": 0.0005, "loss": 2.3888559341430664, "step": 6043 }, { "epoch": 0.7032813590877356, "grad_norm": 0.660270631313324, "learning_rate": 0.0005, "loss": 1.8134855031967163, "step": 6044 }, { "epoch": 0.7033977193390738, "grad_norm": 1.3211641311645508, "learning_rate": 0.0005, "loss": 2.532224178314209, "step": 6045 }, { "epoch": 0.7035140795904119, "grad_norm": 1.6983201503753662, "learning_rate": 0.0005, "loss": 3.0008370876312256, "step": 6046 }, { "epoch": 0.7036304398417501, "grad_norm": 2.1168925762176514, "learning_rate": 0.0005, "loss": 2.850315570831299, "step": 6047 }, { "epoch": 0.7037468000930882, "grad_norm": 1.5589466094970703, "learning_rate": 0.0005, "loss": 2.4341883659362793, "step": 6048 }, { "epoch": 0.7038631603444263, "grad_norm": 1.0382479429244995, "learning_rate": 0.0005, "loss": 1.7170652151107788, "step": 6049 }, { "epoch": 0.7039795205957645, "grad_norm": 0.9652741551399231, "learning_rate": 0.0005, "loss": 1.1256647109985352, "step": 6050 }, { "epoch": 0.7040958808471026, "grad_norm": 1.1624537706375122, "learning_rate": 0.0005, "loss": 2.6837246417999268, "step": 6051 }, { "epoch": 0.7042122410984408, "grad_norm": 2.890404462814331, "learning_rate": 0.0005, "loss": 3.2982075214385986, "step": 6052 }, { "epoch": 0.7043286013497789, "grad_norm": 1.9730148315429688, "learning_rate": 0.0005, "loss": 2.557267665863037, "step": 6053 }, { "epoch": 0.704444961601117, "grad_norm": 1.2510899305343628, "learning_rate": 0.0005, "loss": 2.211514472961426, "step": 6054 }, { "epoch": 0.7045613218524552, "grad_norm": 1.794007420539856, "learning_rate": 0.0005, "loss": 1.9295709133148193, "step": 6055 }, { "epoch": 0.7046776821037933, "grad_norm": 1.2762247323989868, "learning_rate": 0.0005, "loss": 2.3526628017425537, "step": 6056 }, { "epoch": 0.7047940423551314, "grad_norm": 1.5760128498077393, "learning_rate": 0.0005, "loss": 2.0919198989868164, "step": 6057 }, { "epoch": 0.7049104026064696, "grad_norm": 1.6867765188217163, "learning_rate": 0.0005, "loss": 2.8119029998779297, "step": 6058 }, { "epoch": 0.7050267628578077, "grad_norm": 0.7371355295181274, "learning_rate": 0.0005, "loss": 1.5997552871704102, "step": 6059 }, { "epoch": 0.705143123109146, "grad_norm": 1.3404488563537598, "learning_rate": 0.0005, "loss": 2.75705885887146, "step": 6060 }, { "epoch": 0.705259483360484, "grad_norm": 1.067837119102478, "learning_rate": 0.0005, "loss": 1.8581674098968506, "step": 6061 }, { "epoch": 0.7053758436118222, "grad_norm": 2.587698459625244, "learning_rate": 0.0005, "loss": 2.405916452407837, "step": 6062 }, { "epoch": 0.7054922038631604, "grad_norm": 1.6892502307891846, "learning_rate": 0.0005, "loss": 2.882833242416382, "step": 6063 }, { "epoch": 0.7056085641144985, "grad_norm": 1.6639163494110107, "learning_rate": 0.0005, "loss": 2.0103909969329834, "step": 6064 }, { "epoch": 0.7057249243658367, "grad_norm": 1.626293659210205, "learning_rate": 0.0005, "loss": 2.654724597930908, "step": 6065 }, { "epoch": 0.7058412846171748, "grad_norm": 2.57058048248291, "learning_rate": 0.0005, "loss": 2.653036117553711, "step": 6066 }, { "epoch": 0.7059576448685129, "grad_norm": 1.3350692987442017, "learning_rate": 0.0005, "loss": 2.89613676071167, "step": 6067 }, { "epoch": 0.7060740051198511, "grad_norm": 1.3181530237197876, "learning_rate": 0.0005, "loss": 2.257178783416748, "step": 6068 }, { "epoch": 0.7061903653711892, "grad_norm": 1.6671525239944458, "learning_rate": 0.0005, "loss": 2.425002098083496, "step": 6069 }, { "epoch": 0.7063067256225274, "grad_norm": 1.4165101051330566, "learning_rate": 0.0005, "loss": 2.1993777751922607, "step": 6070 }, { "epoch": 0.7064230858738655, "grad_norm": 1.5112229585647583, "learning_rate": 0.0005, "loss": 1.8871104717254639, "step": 6071 }, { "epoch": 0.7065394461252036, "grad_norm": 1.241260051727295, "learning_rate": 0.0005, "loss": 2.422762632369995, "step": 6072 }, { "epoch": 0.7066558063765418, "grad_norm": 1.712924838066101, "learning_rate": 0.0005, "loss": 2.8039746284484863, "step": 6073 }, { "epoch": 0.7067721666278799, "grad_norm": 1.8088185787200928, "learning_rate": 0.0005, "loss": 2.3982152938842773, "step": 6074 }, { "epoch": 0.7068885268792181, "grad_norm": 1.1519981622695923, "learning_rate": 0.0005, "loss": 2.4922292232513428, "step": 6075 }, { "epoch": 0.7070048871305562, "grad_norm": 1.0395088195800781, "learning_rate": 0.0005, "loss": 2.1039860248565674, "step": 6076 }, { "epoch": 0.7071212473818943, "grad_norm": 2.689101457595825, "learning_rate": 0.0005, "loss": 2.6702699661254883, "step": 6077 }, { "epoch": 0.7072376076332325, "grad_norm": 1.5295230150222778, "learning_rate": 0.0005, "loss": 2.1991519927978516, "step": 6078 }, { "epoch": 0.7073539678845706, "grad_norm": 1.1893165111541748, "learning_rate": 0.0005, "loss": 1.8600106239318848, "step": 6079 }, { "epoch": 0.7074703281359088, "grad_norm": 2.838994264602661, "learning_rate": 0.0005, "loss": 2.5340194702148438, "step": 6080 }, { "epoch": 0.7075866883872469, "grad_norm": 1.2339450120925903, "learning_rate": 0.0005, "loss": 2.0565412044525146, "step": 6081 }, { "epoch": 0.707703048638585, "grad_norm": 0.9857081174850464, "learning_rate": 0.0005, "loss": 1.9755761623382568, "step": 6082 }, { "epoch": 0.7078194088899232, "grad_norm": 1.2268552780151367, "learning_rate": 0.0005, "loss": 2.498042106628418, "step": 6083 }, { "epoch": 0.7079357691412613, "grad_norm": 1.6570947170257568, "learning_rate": 0.0005, "loss": 2.5830531120300293, "step": 6084 }, { "epoch": 0.7080521293925995, "grad_norm": 1.830531358718872, "learning_rate": 0.0005, "loss": 2.7227935791015625, "step": 6085 }, { "epoch": 0.7081684896439376, "grad_norm": 1.984724998474121, "learning_rate": 0.0005, "loss": 2.8749234676361084, "step": 6086 }, { "epoch": 0.7082848498952757, "grad_norm": 1.33474862575531, "learning_rate": 0.0005, "loss": 1.6938629150390625, "step": 6087 }, { "epoch": 0.708401210146614, "grad_norm": 1.5716264247894287, "learning_rate": 0.0005, "loss": 2.4222116470336914, "step": 6088 }, { "epoch": 0.708517570397952, "grad_norm": 2.2776870727539062, "learning_rate": 0.0005, "loss": 2.1564204692840576, "step": 6089 }, { "epoch": 0.7086339306492903, "grad_norm": 1.0423123836517334, "learning_rate": 0.0005, "loss": 2.2668330669403076, "step": 6090 }, { "epoch": 0.7087502909006284, "grad_norm": 1.5385138988494873, "learning_rate": 0.0005, "loss": 2.3367843627929688, "step": 6091 }, { "epoch": 0.7088666511519665, "grad_norm": 1.1591346263885498, "learning_rate": 0.0005, "loss": 2.0824708938598633, "step": 6092 }, { "epoch": 0.7089830114033047, "grad_norm": 2.3525099754333496, "learning_rate": 0.0005, "loss": 2.8577427864074707, "step": 6093 }, { "epoch": 0.7090993716546428, "grad_norm": 1.5109291076660156, "learning_rate": 0.0005, "loss": 2.384289264678955, "step": 6094 }, { "epoch": 0.7092157319059809, "grad_norm": 1.2839449644088745, "learning_rate": 0.0005, "loss": 1.8785158395767212, "step": 6095 }, { "epoch": 0.7093320921573191, "grad_norm": 1.8416862487792969, "learning_rate": 0.0005, "loss": 2.699087619781494, "step": 6096 }, { "epoch": 0.7094484524086572, "grad_norm": 1.6620374917984009, "learning_rate": 0.0005, "loss": 3.0417306423187256, "step": 6097 }, { "epoch": 0.7095648126599954, "grad_norm": 1.0909563302993774, "learning_rate": 0.0005, "loss": 2.149203062057495, "step": 6098 }, { "epoch": 0.7096811729113335, "grad_norm": 1.2774466276168823, "learning_rate": 0.0005, "loss": 2.3371315002441406, "step": 6099 }, { "epoch": 0.7097975331626716, "grad_norm": 1.50645911693573, "learning_rate": 0.0005, "loss": 2.1538610458374023, "step": 6100 }, { "epoch": 0.7099138934140098, "grad_norm": 1.7886486053466797, "learning_rate": 0.0005, "loss": 2.2538936138153076, "step": 6101 }, { "epoch": 0.7100302536653479, "grad_norm": 1.9938572645187378, "learning_rate": 0.0005, "loss": 3.1543593406677246, "step": 6102 }, { "epoch": 0.7101466139166861, "grad_norm": 1.776045322418213, "learning_rate": 0.0005, "loss": 2.2238340377807617, "step": 6103 }, { "epoch": 0.7102629741680242, "grad_norm": 1.3725814819335938, "learning_rate": 0.0005, "loss": 2.7760229110717773, "step": 6104 }, { "epoch": 0.7103793344193623, "grad_norm": 1.4414188861846924, "learning_rate": 0.0005, "loss": 1.9663798809051514, "step": 6105 }, { "epoch": 0.7104956946707005, "grad_norm": 1.403536081314087, "learning_rate": 0.0005, "loss": 1.6744053363800049, "step": 6106 }, { "epoch": 0.7106120549220386, "grad_norm": 1.0381215810775757, "learning_rate": 0.0005, "loss": 1.7031606435775757, "step": 6107 }, { "epoch": 0.7107284151733768, "grad_norm": 3.1330742835998535, "learning_rate": 0.0005, "loss": 2.8786301612854004, "step": 6108 }, { "epoch": 0.7108447754247149, "grad_norm": 1.4081693887710571, "learning_rate": 0.0005, "loss": 1.9545409679412842, "step": 6109 }, { "epoch": 0.710961135676053, "grad_norm": 1.5246388912200928, "learning_rate": 0.0005, "loss": 1.6894848346710205, "step": 6110 }, { "epoch": 0.7110774959273912, "grad_norm": 2.1587276458740234, "learning_rate": 0.0005, "loss": 3.283191680908203, "step": 6111 }, { "epoch": 0.7111938561787293, "grad_norm": 1.122330665588379, "learning_rate": 0.0005, "loss": 1.900623083114624, "step": 6112 }, { "epoch": 0.7113102164300675, "grad_norm": 0.9051558375358582, "learning_rate": 0.0005, "loss": 1.178837776184082, "step": 6113 }, { "epoch": 0.7114265766814056, "grad_norm": 1.5998010635375977, "learning_rate": 0.0005, "loss": 2.924682855606079, "step": 6114 }, { "epoch": 0.7115429369327437, "grad_norm": 1.3021914958953857, "learning_rate": 0.0005, "loss": 2.9813621044158936, "step": 6115 }, { "epoch": 0.7116592971840819, "grad_norm": 1.048576831817627, "learning_rate": 0.0005, "loss": 2.6250014305114746, "step": 6116 }, { "epoch": 0.71177565743542, "grad_norm": 1.1606603860855103, "learning_rate": 0.0005, "loss": 1.9892140626907349, "step": 6117 }, { "epoch": 0.7118920176867582, "grad_norm": 0.9648784399032593, "learning_rate": 0.0005, "loss": 2.0783045291900635, "step": 6118 }, { "epoch": 0.7120083779380963, "grad_norm": 1.76430082321167, "learning_rate": 0.0005, "loss": 2.642871856689453, "step": 6119 }, { "epoch": 0.7121247381894344, "grad_norm": 0.9705331921577454, "learning_rate": 0.0005, "loss": 3.00077223777771, "step": 6120 }, { "epoch": 0.7122410984407727, "grad_norm": 0.7958101630210876, "learning_rate": 0.0005, "loss": 1.9573516845703125, "step": 6121 }, { "epoch": 0.7123574586921108, "grad_norm": 1.6832222938537598, "learning_rate": 0.0005, "loss": 1.8324068784713745, "step": 6122 }, { "epoch": 0.712473818943449, "grad_norm": 1.4669724702835083, "learning_rate": 0.0005, "loss": 2.03525710105896, "step": 6123 }, { "epoch": 0.7125901791947871, "grad_norm": 1.3407840728759766, "learning_rate": 0.0005, "loss": 3.1122145652770996, "step": 6124 }, { "epoch": 0.7127065394461252, "grad_norm": 2.8432607650756836, "learning_rate": 0.0005, "loss": 2.489755153656006, "step": 6125 }, { "epoch": 0.7128228996974634, "grad_norm": 1.1400277614593506, "learning_rate": 0.0005, "loss": 1.8448922634124756, "step": 6126 }, { "epoch": 0.7129392599488015, "grad_norm": 1.3021022081375122, "learning_rate": 0.0005, "loss": 1.9678452014923096, "step": 6127 }, { "epoch": 0.7130556202001397, "grad_norm": 1.3991358280181885, "learning_rate": 0.0005, "loss": 2.046193838119507, "step": 6128 }, { "epoch": 0.7131719804514778, "grad_norm": 1.1883879899978638, "learning_rate": 0.0005, "loss": 2.098172426223755, "step": 6129 }, { "epoch": 0.7132883407028159, "grad_norm": 0.9998819828033447, "learning_rate": 0.0005, "loss": 1.8855985403060913, "step": 6130 }, { "epoch": 0.7134047009541541, "grad_norm": 1.593044400215149, "learning_rate": 0.0005, "loss": 2.0549018383026123, "step": 6131 }, { "epoch": 0.7135210612054922, "grad_norm": 2.6130826473236084, "learning_rate": 0.0005, "loss": 2.0905442237854004, "step": 6132 }, { "epoch": 0.7136374214568304, "grad_norm": 1.1514747142791748, "learning_rate": 0.0005, "loss": 1.3164148330688477, "step": 6133 }, { "epoch": 0.7137537817081685, "grad_norm": 0.9952547550201416, "learning_rate": 0.0005, "loss": 2.1326799392700195, "step": 6134 }, { "epoch": 0.7138701419595066, "grad_norm": 1.4609507322311401, "learning_rate": 0.0005, "loss": 2.4439034461975098, "step": 6135 }, { "epoch": 0.7139865022108448, "grad_norm": 1.3363300561904907, "learning_rate": 0.0005, "loss": 1.8468228578567505, "step": 6136 }, { "epoch": 0.7141028624621829, "grad_norm": 1.6760669946670532, "learning_rate": 0.0005, "loss": 2.9848828315734863, "step": 6137 }, { "epoch": 0.714219222713521, "grad_norm": 1.1536870002746582, "learning_rate": 0.0005, "loss": 2.761260986328125, "step": 6138 }, { "epoch": 0.7143355829648592, "grad_norm": 1.3064615726470947, "learning_rate": 0.0005, "loss": 2.5229809284210205, "step": 6139 }, { "epoch": 0.7144519432161973, "grad_norm": 0.860102653503418, "learning_rate": 0.0005, "loss": 1.4399335384368896, "step": 6140 }, { "epoch": 0.7145683034675355, "grad_norm": 1.7216836214065552, "learning_rate": 0.0005, "loss": 1.882699966430664, "step": 6141 }, { "epoch": 0.7146846637188736, "grad_norm": 1.350157618522644, "learning_rate": 0.0005, "loss": 2.2274389266967773, "step": 6142 }, { "epoch": 0.7148010239702117, "grad_norm": 1.1707820892333984, "learning_rate": 0.0005, "loss": 2.5057711601257324, "step": 6143 }, { "epoch": 0.7149173842215499, "grad_norm": 1.1056342124938965, "learning_rate": 0.0005, "loss": 2.4528768062591553, "step": 6144 }, { "epoch": 0.715033744472888, "grad_norm": 1.0591439008712769, "learning_rate": 0.0005, "loss": 1.7958663702011108, "step": 6145 }, { "epoch": 0.7151501047242262, "grad_norm": 1.879913330078125, "learning_rate": 0.0005, "loss": 2.427699565887451, "step": 6146 }, { "epoch": 0.7152664649755643, "grad_norm": 2.0386860370635986, "learning_rate": 0.0005, "loss": 2.484736442565918, "step": 6147 }, { "epoch": 0.7153828252269024, "grad_norm": 1.0199817419052124, "learning_rate": 0.0005, "loss": 3.0075457096099854, "step": 6148 }, { "epoch": 0.7154991854782407, "grad_norm": 2.1948792934417725, "learning_rate": 0.0005, "loss": 2.162837505340576, "step": 6149 }, { "epoch": 0.7156155457295788, "grad_norm": 1.1630821228027344, "learning_rate": 0.0005, "loss": 2.277812957763672, "step": 6150 }, { "epoch": 0.715731905980917, "grad_norm": 14.091378211975098, "learning_rate": 0.0005, "loss": 3.2234063148498535, "step": 6151 }, { "epoch": 0.7158482662322551, "grad_norm": 1.3886439800262451, "learning_rate": 0.0005, "loss": 1.8721038103103638, "step": 6152 }, { "epoch": 0.7159646264835932, "grad_norm": 1.4394787549972534, "learning_rate": 0.0005, "loss": 1.8006511926651, "step": 6153 }, { "epoch": 0.7160809867349314, "grad_norm": 1.4708024263381958, "learning_rate": 0.0005, "loss": 2.4004123210906982, "step": 6154 }, { "epoch": 0.7161973469862695, "grad_norm": 1.0472912788391113, "learning_rate": 0.0005, "loss": 1.8775222301483154, "step": 6155 }, { "epoch": 0.7163137072376077, "grad_norm": 1.3332887887954712, "learning_rate": 0.0005, "loss": 2.5708746910095215, "step": 6156 }, { "epoch": 0.7164300674889458, "grad_norm": 0.8760904669761658, "learning_rate": 0.0005, "loss": 2.265622138977051, "step": 6157 }, { "epoch": 0.7165464277402839, "grad_norm": 1.304702639579773, "learning_rate": 0.0005, "loss": 2.1793222427368164, "step": 6158 }, { "epoch": 0.7166627879916221, "grad_norm": 1.0349783897399902, "learning_rate": 0.0005, "loss": 1.5173654556274414, "step": 6159 }, { "epoch": 0.7167791482429602, "grad_norm": 1.7733097076416016, "learning_rate": 0.0005, "loss": 3.2065494060516357, "step": 6160 }, { "epoch": 0.7168955084942984, "grad_norm": 1.841853141784668, "learning_rate": 0.0005, "loss": 2.446024179458618, "step": 6161 }, { "epoch": 0.7170118687456365, "grad_norm": 1.1771408319473267, "learning_rate": 0.0005, "loss": 2.3161237239837646, "step": 6162 }, { "epoch": 0.7171282289969746, "grad_norm": 1.2945020198822021, "learning_rate": 0.0005, "loss": 2.6739118099212646, "step": 6163 }, { "epoch": 0.7172445892483128, "grad_norm": 1.7168726921081543, "learning_rate": 0.0005, "loss": 2.8160605430603027, "step": 6164 }, { "epoch": 0.7173609494996509, "grad_norm": 1.0900431871414185, "learning_rate": 0.0005, "loss": 1.7534778118133545, "step": 6165 }, { "epoch": 0.7174773097509891, "grad_norm": 1.0464482307434082, "learning_rate": 0.0005, "loss": 2.2080531120300293, "step": 6166 }, { "epoch": 0.7175936700023272, "grad_norm": 1.360775113105774, "learning_rate": 0.0005, "loss": 1.9029340744018555, "step": 6167 }, { "epoch": 0.7177100302536653, "grad_norm": 1.6633448600769043, "learning_rate": 0.0005, "loss": 2.205216407775879, "step": 6168 }, { "epoch": 0.7178263905050035, "grad_norm": 2.2854368686676025, "learning_rate": 0.0005, "loss": 2.6214027404785156, "step": 6169 }, { "epoch": 0.7179427507563416, "grad_norm": 1.2272580862045288, "learning_rate": 0.0005, "loss": 3.153984785079956, "step": 6170 }, { "epoch": 0.7180591110076798, "grad_norm": 0.9904224872589111, "learning_rate": 0.0005, "loss": 2.4371485710144043, "step": 6171 }, { "epoch": 0.7181754712590179, "grad_norm": 1.4806219339370728, "learning_rate": 0.0005, "loss": 3.163214683532715, "step": 6172 }, { "epoch": 0.718291831510356, "grad_norm": 1.4181827306747437, "learning_rate": 0.0005, "loss": 1.8469855785369873, "step": 6173 }, { "epoch": 0.7184081917616942, "grad_norm": 1.4103814363479614, "learning_rate": 0.0005, "loss": 2.318305015563965, "step": 6174 }, { "epoch": 0.7185245520130323, "grad_norm": 1.4548084735870361, "learning_rate": 0.0005, "loss": 2.643772602081299, "step": 6175 }, { "epoch": 0.7186409122643705, "grad_norm": 1.1660045385360718, "learning_rate": 0.0005, "loss": 2.829052209854126, "step": 6176 }, { "epoch": 0.7187572725157086, "grad_norm": 1.9237515926361084, "learning_rate": 0.0005, "loss": 1.8869683742523193, "step": 6177 }, { "epoch": 0.7188736327670467, "grad_norm": 1.194944977760315, "learning_rate": 0.0005, "loss": 2.5324206352233887, "step": 6178 }, { "epoch": 0.718989993018385, "grad_norm": 1.0597407817840576, "learning_rate": 0.0005, "loss": 1.9197096824645996, "step": 6179 }, { "epoch": 0.719106353269723, "grad_norm": 0.7562021017074585, "learning_rate": 0.0005, "loss": 2.138674736022949, "step": 6180 }, { "epoch": 0.7192227135210612, "grad_norm": 0.9871069192886353, "learning_rate": 0.0005, "loss": 1.9848556518554688, "step": 6181 }, { "epoch": 0.7193390737723994, "grad_norm": 1.9970941543579102, "learning_rate": 0.0005, "loss": 2.6152114868164062, "step": 6182 }, { "epoch": 0.7194554340237375, "grad_norm": 2.42211651802063, "learning_rate": 0.0005, "loss": 3.0784554481506348, "step": 6183 }, { "epoch": 0.7195717942750757, "grad_norm": 1.9181673526763916, "learning_rate": 0.0005, "loss": 2.69508957862854, "step": 6184 }, { "epoch": 0.7196881545264138, "grad_norm": 1.5393604040145874, "learning_rate": 0.0005, "loss": 2.4738805294036865, "step": 6185 }, { "epoch": 0.7198045147777519, "grad_norm": 1.3630603551864624, "learning_rate": 0.0005, "loss": 2.4975967407226562, "step": 6186 }, { "epoch": 0.7199208750290901, "grad_norm": 0.8479555249214172, "learning_rate": 0.0005, "loss": 2.809809446334839, "step": 6187 }, { "epoch": 0.7200372352804282, "grad_norm": 1.0919438600540161, "learning_rate": 0.0005, "loss": 2.4834065437316895, "step": 6188 }, { "epoch": 0.7201535955317664, "grad_norm": 1.484466314315796, "learning_rate": 0.0005, "loss": 2.2658448219299316, "step": 6189 }, { "epoch": 0.7202699557831045, "grad_norm": 1.5535448789596558, "learning_rate": 0.0005, "loss": 2.2082056999206543, "step": 6190 }, { "epoch": 0.7203863160344426, "grad_norm": 1.716178059577942, "learning_rate": 0.0005, "loss": 2.632277727127075, "step": 6191 }, { "epoch": 0.7205026762857808, "grad_norm": 2.1765763759613037, "learning_rate": 0.0005, "loss": 2.8283796310424805, "step": 6192 }, { "epoch": 0.7206190365371189, "grad_norm": 1.1594127416610718, "learning_rate": 0.0005, "loss": 1.967573881149292, "step": 6193 }, { "epoch": 0.7207353967884571, "grad_norm": 0.8460395932197571, "learning_rate": 0.0005, "loss": 1.732749104499817, "step": 6194 }, { "epoch": 0.7208517570397952, "grad_norm": 0.9980353713035583, "learning_rate": 0.0005, "loss": 1.995784878730774, "step": 6195 }, { "epoch": 0.7209681172911333, "grad_norm": 1.8812328577041626, "learning_rate": 0.0005, "loss": 2.78641414642334, "step": 6196 }, { "epoch": 0.7210844775424715, "grad_norm": 1.811010479927063, "learning_rate": 0.0005, "loss": 3.1008291244506836, "step": 6197 }, { "epoch": 0.7212008377938096, "grad_norm": 1.1173813343048096, "learning_rate": 0.0005, "loss": 2.0273423194885254, "step": 6198 }, { "epoch": 0.7213171980451478, "grad_norm": 1.5117700099945068, "learning_rate": 0.0005, "loss": 2.6087498664855957, "step": 6199 }, { "epoch": 0.7214335582964859, "grad_norm": 1.7582508325576782, "learning_rate": 0.0005, "loss": 2.6630468368530273, "step": 6200 }, { "epoch": 0.721549918547824, "grad_norm": 1.3138481378555298, "learning_rate": 0.0005, "loss": 1.8540546894073486, "step": 6201 }, { "epoch": 0.7216662787991622, "grad_norm": 1.523010015487671, "learning_rate": 0.0005, "loss": 1.677042007446289, "step": 6202 }, { "epoch": 0.7217826390505003, "grad_norm": 1.731858730316162, "learning_rate": 0.0005, "loss": 2.557037353515625, "step": 6203 }, { "epoch": 0.7218989993018385, "grad_norm": 1.531911849975586, "learning_rate": 0.0005, "loss": 2.9218339920043945, "step": 6204 }, { "epoch": 0.7220153595531766, "grad_norm": 1.2957439422607422, "learning_rate": 0.0005, "loss": 2.3458495140075684, "step": 6205 }, { "epoch": 0.7221317198045147, "grad_norm": 1.3042354583740234, "learning_rate": 0.0005, "loss": 2.7793009281158447, "step": 6206 }, { "epoch": 0.722248080055853, "grad_norm": 1.0100061893463135, "learning_rate": 0.0005, "loss": 2.3992667198181152, "step": 6207 }, { "epoch": 0.722364440307191, "grad_norm": 1.6616400480270386, "learning_rate": 0.0005, "loss": 3.287998914718628, "step": 6208 }, { "epoch": 0.7224808005585293, "grad_norm": 1.3385024070739746, "learning_rate": 0.0005, "loss": 3.0704126358032227, "step": 6209 }, { "epoch": 0.7225971608098674, "grad_norm": 1.3042683601379395, "learning_rate": 0.0005, "loss": 2.611279010772705, "step": 6210 }, { "epoch": 0.7227135210612055, "grad_norm": 1.8110318183898926, "learning_rate": 0.0005, "loss": 2.1165757179260254, "step": 6211 }, { "epoch": 0.7228298813125437, "grad_norm": 1.2282015085220337, "learning_rate": 0.0005, "loss": 2.7601630687713623, "step": 6212 }, { "epoch": 0.7229462415638818, "grad_norm": 1.2580056190490723, "learning_rate": 0.0005, "loss": 2.1346304416656494, "step": 6213 }, { "epoch": 0.72306260181522, "grad_norm": 2.1392338275909424, "learning_rate": 0.0005, "loss": 3.3145856857299805, "step": 6214 }, { "epoch": 0.7231789620665581, "grad_norm": 1.16697359085083, "learning_rate": 0.0005, "loss": 3.126357078552246, "step": 6215 }, { "epoch": 0.7232953223178962, "grad_norm": 1.3645893335342407, "learning_rate": 0.0005, "loss": 2.7468442916870117, "step": 6216 }, { "epoch": 0.7234116825692344, "grad_norm": 1.8291047811508179, "learning_rate": 0.0005, "loss": 3.1184983253479004, "step": 6217 }, { "epoch": 0.7235280428205725, "grad_norm": 1.0115174055099487, "learning_rate": 0.0005, "loss": 2.376460075378418, "step": 6218 }, { "epoch": 0.7236444030719106, "grad_norm": 1.5174888372421265, "learning_rate": 0.0005, "loss": 2.1472134590148926, "step": 6219 }, { "epoch": 0.7237607633232488, "grad_norm": 1.7663583755493164, "learning_rate": 0.0005, "loss": 4.216102600097656, "step": 6220 }, { "epoch": 0.7238771235745869, "grad_norm": 1.001153826713562, "learning_rate": 0.0005, "loss": 1.963836431503296, "step": 6221 }, { "epoch": 0.7239934838259251, "grad_norm": 1.16970956325531, "learning_rate": 0.0005, "loss": 2.645996570587158, "step": 6222 }, { "epoch": 0.7241098440772632, "grad_norm": 1.0311764478683472, "learning_rate": 0.0005, "loss": 2.077036142349243, "step": 6223 }, { "epoch": 0.7242262043286013, "grad_norm": 1.4247909784317017, "learning_rate": 0.0005, "loss": 2.8965015411376953, "step": 6224 }, { "epoch": 0.7243425645799395, "grad_norm": 1.6179760694503784, "learning_rate": 0.0005, "loss": 3.167388916015625, "step": 6225 }, { "epoch": 0.7244589248312776, "grad_norm": 1.1301507949829102, "learning_rate": 0.0005, "loss": 1.726069688796997, "step": 6226 }, { "epoch": 0.7245752850826158, "grad_norm": 1.411033034324646, "learning_rate": 0.0005, "loss": 2.5038933753967285, "step": 6227 }, { "epoch": 0.7246916453339539, "grad_norm": 0.6820347309112549, "learning_rate": 0.0005, "loss": 1.6930716037750244, "step": 6228 }, { "epoch": 0.724808005585292, "grad_norm": 1.6020936965942383, "learning_rate": 0.0005, "loss": 2.8691630363464355, "step": 6229 }, { "epoch": 0.7249243658366302, "grad_norm": 1.2394988536834717, "learning_rate": 0.0005, "loss": 2.079618453979492, "step": 6230 }, { "epoch": 0.7250407260879683, "grad_norm": 1.565054178237915, "learning_rate": 0.0005, "loss": 2.794632911682129, "step": 6231 }, { "epoch": 0.7251570863393065, "grad_norm": 1.423925757408142, "learning_rate": 0.0005, "loss": 2.1653385162353516, "step": 6232 }, { "epoch": 0.7252734465906446, "grad_norm": 1.1943589448928833, "learning_rate": 0.0005, "loss": 2.576927900314331, "step": 6233 }, { "epoch": 0.7253898068419827, "grad_norm": 1.2999205589294434, "learning_rate": 0.0005, "loss": 2.465359926223755, "step": 6234 }, { "epoch": 0.7255061670933209, "grad_norm": 1.6339727640151978, "learning_rate": 0.0005, "loss": 1.9804353713989258, "step": 6235 }, { "epoch": 0.725622527344659, "grad_norm": 1.6582833528518677, "learning_rate": 0.0005, "loss": 2.5932013988494873, "step": 6236 }, { "epoch": 0.7257388875959973, "grad_norm": 0.9675233960151672, "learning_rate": 0.0005, "loss": 1.626112937927246, "step": 6237 }, { "epoch": 0.7258552478473353, "grad_norm": 0.8204558491706848, "learning_rate": 0.0005, "loss": 2.2189254760742188, "step": 6238 }, { "epoch": 0.7259716080986734, "grad_norm": 1.4014114141464233, "learning_rate": 0.0005, "loss": 2.2565035820007324, "step": 6239 }, { "epoch": 0.7260879683500117, "grad_norm": 1.1786469221115112, "learning_rate": 0.0005, "loss": 2.6379144191741943, "step": 6240 }, { "epoch": 0.7262043286013498, "grad_norm": 1.409699559211731, "learning_rate": 0.0005, "loss": 2.739225149154663, "step": 6241 }, { "epoch": 0.726320688852688, "grad_norm": 0.8463601469993591, "learning_rate": 0.0005, "loss": 2.064831495285034, "step": 6242 }, { "epoch": 0.7264370491040261, "grad_norm": 0.8304766416549683, "learning_rate": 0.0005, "loss": 2.383678436279297, "step": 6243 }, { "epoch": 0.7265534093553642, "grad_norm": 1.1742831468582153, "learning_rate": 0.0005, "loss": 3.534060001373291, "step": 6244 }, { "epoch": 0.7266697696067024, "grad_norm": 7.5518107414245605, "learning_rate": 0.0005, "loss": 2.2247390747070312, "step": 6245 }, { "epoch": 0.7267861298580405, "grad_norm": 1.6375116109848022, "learning_rate": 0.0005, "loss": 2.8047935962677, "step": 6246 }, { "epoch": 0.7269024901093787, "grad_norm": 1.4216396808624268, "learning_rate": 0.0005, "loss": 2.189537525177002, "step": 6247 }, { "epoch": 0.7270188503607168, "grad_norm": 2.614933490753174, "learning_rate": 0.0005, "loss": 2.56618595123291, "step": 6248 }, { "epoch": 0.7271352106120549, "grad_norm": 1.564860224723816, "learning_rate": 0.0005, "loss": 2.7500271797180176, "step": 6249 }, { "epoch": 0.7272515708633931, "grad_norm": 1.2043426036834717, "learning_rate": 0.0005, "loss": 2.851351261138916, "step": 6250 }, { "epoch": 0.7273679311147312, "grad_norm": 1.501053810119629, "learning_rate": 0.0005, "loss": 1.9899396896362305, "step": 6251 }, { "epoch": 0.7274842913660694, "grad_norm": 1.1399434804916382, "learning_rate": 0.0005, "loss": 2.186159372329712, "step": 6252 }, { "epoch": 0.7276006516174075, "grad_norm": 1.460216760635376, "learning_rate": 0.0005, "loss": 2.3288660049438477, "step": 6253 }, { "epoch": 0.7277170118687456, "grad_norm": 1.5306932926177979, "learning_rate": 0.0005, "loss": 2.1122758388519287, "step": 6254 }, { "epoch": 0.7278333721200838, "grad_norm": 1.287614345550537, "learning_rate": 0.0005, "loss": 2.2806851863861084, "step": 6255 }, { "epoch": 0.7279497323714219, "grad_norm": 1.517438292503357, "learning_rate": 0.0005, "loss": 2.148003101348877, "step": 6256 }, { "epoch": 0.7280660926227601, "grad_norm": 1.3513669967651367, "learning_rate": 0.0005, "loss": 3.041064977645874, "step": 6257 }, { "epoch": 0.7281824528740982, "grad_norm": 2.880159616470337, "learning_rate": 0.0005, "loss": 3.3885068893432617, "step": 6258 }, { "epoch": 0.7282988131254363, "grad_norm": 3.561749219894409, "learning_rate": 0.0005, "loss": 2.2998361587524414, "step": 6259 }, { "epoch": 0.7284151733767745, "grad_norm": 1.1810212135314941, "learning_rate": 0.0005, "loss": 2.245114326477051, "step": 6260 }, { "epoch": 0.7285315336281126, "grad_norm": 0.7619054913520813, "learning_rate": 0.0005, "loss": 1.4110386371612549, "step": 6261 }, { "epoch": 0.7286478938794507, "grad_norm": 1.3432152271270752, "learning_rate": 0.0005, "loss": 2.5766396522521973, "step": 6262 }, { "epoch": 0.7287642541307889, "grad_norm": 2.1570637226104736, "learning_rate": 0.0005, "loss": 3.121072292327881, "step": 6263 }, { "epoch": 0.728880614382127, "grad_norm": 1.015032172203064, "learning_rate": 0.0005, "loss": 2.486233711242676, "step": 6264 }, { "epoch": 0.7289969746334652, "grad_norm": 1.517364263534546, "learning_rate": 0.0005, "loss": 2.1824445724487305, "step": 6265 }, { "epoch": 0.7291133348848033, "grad_norm": 1.5550109148025513, "learning_rate": 0.0005, "loss": 2.9203686714172363, "step": 6266 }, { "epoch": 0.7292296951361414, "grad_norm": 1.2005012035369873, "learning_rate": 0.0005, "loss": 2.2634549140930176, "step": 6267 }, { "epoch": 0.7293460553874797, "grad_norm": 0.9621649384498596, "learning_rate": 0.0005, "loss": 1.9740657806396484, "step": 6268 }, { "epoch": 0.7294624156388178, "grad_norm": 1.1628385782241821, "learning_rate": 0.0005, "loss": 1.911617398262024, "step": 6269 }, { "epoch": 0.729578775890156, "grad_norm": 1.3223544359207153, "learning_rate": 0.0005, "loss": 2.592068910598755, "step": 6270 }, { "epoch": 0.7296951361414941, "grad_norm": 1.8886663913726807, "learning_rate": 0.0005, "loss": 3.406796932220459, "step": 6271 }, { "epoch": 0.7298114963928322, "grad_norm": 1.6573237180709839, "learning_rate": 0.0005, "loss": 2.5641584396362305, "step": 6272 }, { "epoch": 0.7299278566441704, "grad_norm": 0.7958207130432129, "learning_rate": 0.0005, "loss": 2.1414384841918945, "step": 6273 }, { "epoch": 0.7300442168955085, "grad_norm": 1.550085425376892, "learning_rate": 0.0005, "loss": 2.6858043670654297, "step": 6274 }, { "epoch": 0.7301605771468467, "grad_norm": 0.9186850786209106, "learning_rate": 0.0005, "loss": 1.5623009204864502, "step": 6275 }, { "epoch": 0.7302769373981848, "grad_norm": 1.685020923614502, "learning_rate": 0.0005, "loss": 2.037118673324585, "step": 6276 }, { "epoch": 0.7303932976495229, "grad_norm": 1.374502182006836, "learning_rate": 0.0005, "loss": 2.5935466289520264, "step": 6277 }, { "epoch": 0.7305096579008611, "grad_norm": 1.6404894590377808, "learning_rate": 0.0005, "loss": 2.678900718688965, "step": 6278 }, { "epoch": 0.7306260181521992, "grad_norm": 1.445905089378357, "learning_rate": 0.0005, "loss": 2.7032222747802734, "step": 6279 }, { "epoch": 0.7307423784035374, "grad_norm": 1.4417102336883545, "learning_rate": 0.0005, "loss": 2.711216449737549, "step": 6280 }, { "epoch": 0.7308587386548755, "grad_norm": 2.723754405975342, "learning_rate": 0.0005, "loss": 2.482797622680664, "step": 6281 }, { "epoch": 0.7309750989062136, "grad_norm": 1.8283226490020752, "learning_rate": 0.0005, "loss": 2.5389246940612793, "step": 6282 }, { "epoch": 0.7310914591575518, "grad_norm": 1.3800828456878662, "learning_rate": 0.0005, "loss": 2.4765548706054688, "step": 6283 }, { "epoch": 0.7312078194088899, "grad_norm": 4.178553581237793, "learning_rate": 0.0005, "loss": 2.5874834060668945, "step": 6284 }, { "epoch": 0.7313241796602281, "grad_norm": 0.9896007776260376, "learning_rate": 0.0005, "loss": 2.618053913116455, "step": 6285 }, { "epoch": 0.7314405399115662, "grad_norm": 1.8816442489624023, "learning_rate": 0.0005, "loss": 2.731630802154541, "step": 6286 }, { "epoch": 0.7315569001629043, "grad_norm": 1.340142846107483, "learning_rate": 0.0005, "loss": 2.6808066368103027, "step": 6287 }, { "epoch": 0.7316732604142425, "grad_norm": 0.8448675870895386, "learning_rate": 0.0005, "loss": 2.3512675762176514, "step": 6288 }, { "epoch": 0.7317896206655806, "grad_norm": 0.9681573510169983, "learning_rate": 0.0005, "loss": 2.172475576400757, "step": 6289 }, { "epoch": 0.7319059809169188, "grad_norm": 1.8762953281402588, "learning_rate": 0.0005, "loss": 2.3621392250061035, "step": 6290 }, { "epoch": 0.7320223411682569, "grad_norm": 1.3360600471496582, "learning_rate": 0.0005, "loss": 1.8168123960494995, "step": 6291 }, { "epoch": 0.732138701419595, "grad_norm": 1.3279759883880615, "learning_rate": 0.0005, "loss": 2.150977849960327, "step": 6292 }, { "epoch": 0.7322550616709332, "grad_norm": 1.3856558799743652, "learning_rate": 0.0005, "loss": 2.476984977722168, "step": 6293 }, { "epoch": 0.7323714219222713, "grad_norm": 1.391826868057251, "learning_rate": 0.0005, "loss": 2.3250489234924316, "step": 6294 }, { "epoch": 0.7324877821736095, "grad_norm": 1.4337204694747925, "learning_rate": 0.0005, "loss": 2.181878089904785, "step": 6295 }, { "epoch": 0.7326041424249476, "grad_norm": 1.1544533967971802, "learning_rate": 0.0005, "loss": 1.601982593536377, "step": 6296 }, { "epoch": 0.7327205026762857, "grad_norm": 1.2074675559997559, "learning_rate": 0.0005, "loss": 2.7359442710876465, "step": 6297 }, { "epoch": 0.732836862927624, "grad_norm": 1.3116973638534546, "learning_rate": 0.0005, "loss": 1.719559907913208, "step": 6298 }, { "epoch": 0.732953223178962, "grad_norm": 1.8199113607406616, "learning_rate": 0.0005, "loss": 3.153552532196045, "step": 6299 }, { "epoch": 0.7330695834303003, "grad_norm": 1.120459794998169, "learning_rate": 0.0005, "loss": 1.8944215774536133, "step": 6300 }, { "epoch": 0.7331859436816384, "grad_norm": 1.0956372022628784, "learning_rate": 0.0005, "loss": 1.9854590892791748, "step": 6301 }, { "epoch": 0.7333023039329765, "grad_norm": 0.8585531711578369, "learning_rate": 0.0005, "loss": 2.4264941215515137, "step": 6302 }, { "epoch": 0.7334186641843147, "grad_norm": 1.8044734001159668, "learning_rate": 0.0005, "loss": 3.4302356243133545, "step": 6303 }, { "epoch": 0.7335350244356528, "grad_norm": 1.1656688451766968, "learning_rate": 0.0005, "loss": 2.757380962371826, "step": 6304 }, { "epoch": 0.7336513846869909, "grad_norm": 1.9104232788085938, "learning_rate": 0.0005, "loss": 3.2517406940460205, "step": 6305 }, { "epoch": 0.7337677449383291, "grad_norm": 3.822662353515625, "learning_rate": 0.0005, "loss": 3.078284502029419, "step": 6306 }, { "epoch": 0.7338841051896672, "grad_norm": 0.9048203229904175, "learning_rate": 0.0005, "loss": 2.185750722885132, "step": 6307 }, { "epoch": 0.7340004654410054, "grad_norm": 1.0712453126907349, "learning_rate": 0.0005, "loss": 3.150550365447998, "step": 6308 }, { "epoch": 0.7341168256923435, "grad_norm": 1.3029602766036987, "learning_rate": 0.0005, "loss": 2.5142345428466797, "step": 6309 }, { "epoch": 0.7342331859436816, "grad_norm": 1.2440619468688965, "learning_rate": 0.0005, "loss": 2.1055006980895996, "step": 6310 }, { "epoch": 0.7343495461950198, "grad_norm": 1.6604783535003662, "learning_rate": 0.0005, "loss": 2.1209065914154053, "step": 6311 }, { "epoch": 0.7344659064463579, "grad_norm": 1.125120997428894, "learning_rate": 0.0005, "loss": 2.1717801094055176, "step": 6312 }, { "epoch": 0.7345822666976961, "grad_norm": 1.0237345695495605, "learning_rate": 0.0005, "loss": 2.2165818214416504, "step": 6313 }, { "epoch": 0.7346986269490342, "grad_norm": 1.7790241241455078, "learning_rate": 0.0005, "loss": 2.821025848388672, "step": 6314 }, { "epoch": 0.7348149872003723, "grad_norm": 1.524976372718811, "learning_rate": 0.0005, "loss": 2.579505205154419, "step": 6315 }, { "epoch": 0.7349313474517105, "grad_norm": 1.3778678178787231, "learning_rate": 0.0005, "loss": 2.672024726867676, "step": 6316 }, { "epoch": 0.7350477077030486, "grad_norm": 1.5246832370758057, "learning_rate": 0.0005, "loss": 2.3050403594970703, "step": 6317 }, { "epoch": 0.7351640679543868, "grad_norm": 1.0995678901672363, "learning_rate": 0.0005, "loss": 2.513136863708496, "step": 6318 }, { "epoch": 0.7352804282057249, "grad_norm": 1.198954701423645, "learning_rate": 0.0005, "loss": 1.5466816425323486, "step": 6319 }, { "epoch": 0.735396788457063, "grad_norm": 0.8200907707214355, "learning_rate": 0.0005, "loss": 1.8401809930801392, "step": 6320 }, { "epoch": 0.7355131487084012, "grad_norm": 1.4758809804916382, "learning_rate": 0.0005, "loss": 2.7973849773406982, "step": 6321 }, { "epoch": 0.7356295089597393, "grad_norm": 1.2302827835083008, "learning_rate": 0.0005, "loss": 1.922503113746643, "step": 6322 }, { "epoch": 0.7357458692110775, "grad_norm": 0.48459625244140625, "learning_rate": 0.0005, "loss": 1.0987555980682373, "step": 6323 }, { "epoch": 0.7358622294624156, "grad_norm": 1.4671434164047241, "learning_rate": 0.0005, "loss": 2.6945035457611084, "step": 6324 }, { "epoch": 0.7359785897137537, "grad_norm": 0.8137004375457764, "learning_rate": 0.0005, "loss": 1.224367618560791, "step": 6325 }, { "epoch": 0.736094949965092, "grad_norm": 1.6796958446502686, "learning_rate": 0.0005, "loss": 2.5699880123138428, "step": 6326 }, { "epoch": 0.73621131021643, "grad_norm": 1.4795112609863281, "learning_rate": 0.0005, "loss": 2.4239578247070312, "step": 6327 }, { "epoch": 0.7363276704677683, "grad_norm": 1.450789451599121, "learning_rate": 0.0005, "loss": 1.8649808168411255, "step": 6328 }, { "epoch": 0.7364440307191064, "grad_norm": 2.2773866653442383, "learning_rate": 0.0005, "loss": 3.7979748249053955, "step": 6329 }, { "epoch": 0.7365603909704445, "grad_norm": 3.943708658218384, "learning_rate": 0.0005, "loss": 2.6438822746276855, "step": 6330 }, { "epoch": 0.7366767512217827, "grad_norm": 1.5453505516052246, "learning_rate": 0.0005, "loss": 2.045604944229126, "step": 6331 }, { "epoch": 0.7367931114731208, "grad_norm": 1.9880644083023071, "learning_rate": 0.0005, "loss": 3.1801655292510986, "step": 6332 }, { "epoch": 0.736909471724459, "grad_norm": 1.2147326469421387, "learning_rate": 0.0005, "loss": 2.4590773582458496, "step": 6333 }, { "epoch": 0.7370258319757971, "grad_norm": 1.3679816722869873, "learning_rate": 0.0005, "loss": 2.1700949668884277, "step": 6334 }, { "epoch": 0.7371421922271352, "grad_norm": 3.5410077571868896, "learning_rate": 0.0005, "loss": 2.490422248840332, "step": 6335 }, { "epoch": 0.7372585524784734, "grad_norm": 1.3609936237335205, "learning_rate": 0.0005, "loss": 2.5255630016326904, "step": 6336 }, { "epoch": 0.7373749127298115, "grad_norm": 1.825018286705017, "learning_rate": 0.0005, "loss": 3.0954108238220215, "step": 6337 }, { "epoch": 0.7374912729811497, "grad_norm": 1.4771778583526611, "learning_rate": 0.0005, "loss": 2.5251717567443848, "step": 6338 }, { "epoch": 0.7376076332324878, "grad_norm": 2.024014711380005, "learning_rate": 0.0005, "loss": 1.745739459991455, "step": 6339 }, { "epoch": 0.7377239934838259, "grad_norm": 1.5779333114624023, "learning_rate": 0.0005, "loss": 2.6097805500030518, "step": 6340 }, { "epoch": 0.7378403537351641, "grad_norm": 1.495733380317688, "learning_rate": 0.0005, "loss": 2.1579771041870117, "step": 6341 }, { "epoch": 0.7379567139865022, "grad_norm": 1.6200989484786987, "learning_rate": 0.0005, "loss": 2.6256866455078125, "step": 6342 }, { "epoch": 0.7380730742378403, "grad_norm": 1.2832568883895874, "learning_rate": 0.0005, "loss": 2.152259588241577, "step": 6343 }, { "epoch": 0.7381894344891785, "grad_norm": 1.7491185665130615, "learning_rate": 0.0005, "loss": 2.7592649459838867, "step": 6344 }, { "epoch": 0.7383057947405166, "grad_norm": 1.0792045593261719, "learning_rate": 0.0005, "loss": 1.3468296527862549, "step": 6345 }, { "epoch": 0.7384221549918548, "grad_norm": 1.4042797088623047, "learning_rate": 0.0005, "loss": 2.883815288543701, "step": 6346 }, { "epoch": 0.7385385152431929, "grad_norm": 0.9555338621139526, "learning_rate": 0.0005, "loss": 2.1424713134765625, "step": 6347 }, { "epoch": 0.738654875494531, "grad_norm": 1.3816651105880737, "learning_rate": 0.0005, "loss": 1.4182628393173218, "step": 6348 }, { "epoch": 0.7387712357458692, "grad_norm": 0.9509002566337585, "learning_rate": 0.0005, "loss": 1.839932918548584, "step": 6349 }, { "epoch": 0.7388875959972073, "grad_norm": 1.2065932750701904, "learning_rate": 0.0005, "loss": 1.9536197185516357, "step": 6350 }, { "epoch": 0.7390039562485455, "grad_norm": 1.4103518724441528, "learning_rate": 0.0005, "loss": 2.487985610961914, "step": 6351 }, { "epoch": 0.7391203164998836, "grad_norm": 1.6689242124557495, "learning_rate": 0.0005, "loss": 2.2737972736358643, "step": 6352 }, { "epoch": 0.7392366767512217, "grad_norm": 1.6171599626541138, "learning_rate": 0.0005, "loss": 2.4528069496154785, "step": 6353 }, { "epoch": 0.7393530370025599, "grad_norm": 0.7537352442741394, "learning_rate": 0.0005, "loss": 1.1510233879089355, "step": 6354 }, { "epoch": 0.739469397253898, "grad_norm": 2.645179510116577, "learning_rate": 0.0005, "loss": 2.7234277725219727, "step": 6355 }, { "epoch": 0.7395857575052363, "grad_norm": 1.5757379531860352, "learning_rate": 0.0005, "loss": 2.6404318809509277, "step": 6356 }, { "epoch": 0.7397021177565744, "grad_norm": 2.125092029571533, "learning_rate": 0.0005, "loss": 3.6386990547180176, "step": 6357 }, { "epoch": 0.7398184780079125, "grad_norm": 2.851680278778076, "learning_rate": 0.0005, "loss": 2.243523120880127, "step": 6358 }, { "epoch": 0.7399348382592507, "grad_norm": 2.2324700355529785, "learning_rate": 0.0005, "loss": 2.391298294067383, "step": 6359 }, { "epoch": 0.7400511985105888, "grad_norm": 1.34650719165802, "learning_rate": 0.0005, "loss": 2.2571778297424316, "step": 6360 }, { "epoch": 0.740167558761927, "grad_norm": 2.286496162414551, "learning_rate": 0.0005, "loss": 2.6633193492889404, "step": 6361 }, { "epoch": 0.7402839190132651, "grad_norm": 1.7588191032409668, "learning_rate": 0.0005, "loss": 1.6712383031845093, "step": 6362 }, { "epoch": 0.7404002792646032, "grad_norm": 1.8932249546051025, "learning_rate": 0.0005, "loss": 3.2338924407958984, "step": 6363 }, { "epoch": 0.7405166395159414, "grad_norm": 1.4789056777954102, "learning_rate": 0.0005, "loss": 1.9246774911880493, "step": 6364 }, { "epoch": 0.7406329997672795, "grad_norm": 1.4730654954910278, "learning_rate": 0.0005, "loss": 2.985384941101074, "step": 6365 }, { "epoch": 0.7407493600186177, "grad_norm": 1.400878667831421, "learning_rate": 0.0005, "loss": 3.0106213092803955, "step": 6366 }, { "epoch": 0.7408657202699558, "grad_norm": 1.2277170419692993, "learning_rate": 0.0005, "loss": 2.5376780033111572, "step": 6367 }, { "epoch": 0.7409820805212939, "grad_norm": 1.2687698602676392, "learning_rate": 0.0005, "loss": 2.2991585731506348, "step": 6368 }, { "epoch": 0.7410984407726321, "grad_norm": 2.5882248878479004, "learning_rate": 0.0005, "loss": 2.8386316299438477, "step": 6369 }, { "epoch": 0.7412148010239702, "grad_norm": 1.3150346279144287, "learning_rate": 0.0005, "loss": 2.040709972381592, "step": 6370 }, { "epoch": 0.7413311612753084, "grad_norm": 1.2768348455429077, "learning_rate": 0.0005, "loss": 1.9227641820907593, "step": 6371 }, { "epoch": 0.7414475215266465, "grad_norm": 0.9381568431854248, "learning_rate": 0.0005, "loss": 2.1570487022399902, "step": 6372 }, { "epoch": 0.7415638817779846, "grad_norm": 1.2941489219665527, "learning_rate": 0.0005, "loss": 1.8123865127563477, "step": 6373 }, { "epoch": 0.7416802420293228, "grad_norm": 1.1698569059371948, "learning_rate": 0.0005, "loss": 2.459099769592285, "step": 6374 }, { "epoch": 0.7417966022806609, "grad_norm": 1.7424073219299316, "learning_rate": 0.0005, "loss": 2.844114303588867, "step": 6375 }, { "epoch": 0.7419129625319991, "grad_norm": 2.155552625656128, "learning_rate": 0.0005, "loss": 3.282095193862915, "step": 6376 }, { "epoch": 0.7420293227833372, "grad_norm": 1.3389251232147217, "learning_rate": 0.0005, "loss": 1.9007935523986816, "step": 6377 }, { "epoch": 0.7421456830346753, "grad_norm": 1.7445029020309448, "learning_rate": 0.0005, "loss": 2.6954293251037598, "step": 6378 }, { "epoch": 0.7422620432860135, "grad_norm": 1.953654170036316, "learning_rate": 0.0005, "loss": 2.431307315826416, "step": 6379 }, { "epoch": 0.7423784035373516, "grad_norm": 1.4097398519515991, "learning_rate": 0.0005, "loss": 2.799126148223877, "step": 6380 }, { "epoch": 0.7424947637886898, "grad_norm": 1.2536914348602295, "learning_rate": 0.0005, "loss": 1.6995627880096436, "step": 6381 }, { "epoch": 0.7426111240400279, "grad_norm": 1.2572290897369385, "learning_rate": 0.0005, "loss": 2.0045132637023926, "step": 6382 }, { "epoch": 0.742727484291366, "grad_norm": 1.9730318784713745, "learning_rate": 0.0005, "loss": 2.2043375968933105, "step": 6383 }, { "epoch": 0.7428438445427042, "grad_norm": 1.7407126426696777, "learning_rate": 0.0005, "loss": 2.885683536529541, "step": 6384 }, { "epoch": 0.7429602047940423, "grad_norm": 1.2363234758377075, "learning_rate": 0.0005, "loss": 2.245814323425293, "step": 6385 }, { "epoch": 0.7430765650453804, "grad_norm": 1.412895917892456, "learning_rate": 0.0005, "loss": 2.572404384613037, "step": 6386 }, { "epoch": 0.7431929252967187, "grad_norm": 0.8851895332336426, "learning_rate": 0.0005, "loss": 1.1985385417938232, "step": 6387 }, { "epoch": 0.7433092855480568, "grad_norm": 0.8261929154396057, "learning_rate": 0.0005, "loss": 1.6990360021591187, "step": 6388 }, { "epoch": 0.743425645799395, "grad_norm": 1.3208268880844116, "learning_rate": 0.0005, "loss": 1.7564936876296997, "step": 6389 }, { "epoch": 0.7435420060507331, "grad_norm": 1.5694520473480225, "learning_rate": 0.0005, "loss": 2.45833420753479, "step": 6390 }, { "epoch": 0.7436583663020712, "grad_norm": 1.8320045471191406, "learning_rate": 0.0005, "loss": 2.495865821838379, "step": 6391 }, { "epoch": 0.7437747265534094, "grad_norm": 1.8472342491149902, "learning_rate": 0.0005, "loss": 2.754317283630371, "step": 6392 }, { "epoch": 0.7438910868047475, "grad_norm": 2.264242649078369, "learning_rate": 0.0005, "loss": 2.9917593002319336, "step": 6393 }, { "epoch": 0.7440074470560857, "grad_norm": 1.1904070377349854, "learning_rate": 0.0005, "loss": 1.8578671216964722, "step": 6394 }, { "epoch": 0.7441238073074238, "grad_norm": 1.2273720502853394, "learning_rate": 0.0005, "loss": 2.5351078510284424, "step": 6395 }, { "epoch": 0.7442401675587619, "grad_norm": 1.6202729940414429, "learning_rate": 0.0005, "loss": 1.9778766632080078, "step": 6396 }, { "epoch": 0.7443565278101001, "grad_norm": 0.8278363347053528, "learning_rate": 0.0005, "loss": 2.447511911392212, "step": 6397 }, { "epoch": 0.7444728880614382, "grad_norm": 1.5709787607192993, "learning_rate": 0.0005, "loss": 2.5996041297912598, "step": 6398 }, { "epoch": 0.7445892483127764, "grad_norm": 1.5813226699829102, "learning_rate": 0.0005, "loss": 2.5846924781799316, "step": 6399 }, { "epoch": 0.7447056085641145, "grad_norm": 2.613647222518921, "learning_rate": 0.0005, "loss": 3.8376693725585938, "step": 6400 }, { "epoch": 0.7448219688154526, "grad_norm": 1.2882378101348877, "learning_rate": 0.0005, "loss": 1.9396114349365234, "step": 6401 }, { "epoch": 0.7449383290667908, "grad_norm": 1.4931856393814087, "learning_rate": 0.0005, "loss": 1.8770778179168701, "step": 6402 }, { "epoch": 0.7450546893181289, "grad_norm": 1.5164846181869507, "learning_rate": 0.0005, "loss": 3.0487256050109863, "step": 6403 }, { "epoch": 0.7451710495694671, "grad_norm": 0.9596115946769714, "learning_rate": 0.0005, "loss": 2.0795867443084717, "step": 6404 }, { "epoch": 0.7452874098208052, "grad_norm": 1.9549190998077393, "learning_rate": 0.0005, "loss": 2.7126574516296387, "step": 6405 }, { "epoch": 0.7454037700721433, "grad_norm": 1.361585021018982, "learning_rate": 0.0005, "loss": 2.9417660236358643, "step": 6406 }, { "epoch": 0.7455201303234815, "grad_norm": 0.9344819188117981, "learning_rate": 0.0005, "loss": 1.9273825883865356, "step": 6407 }, { "epoch": 0.7456364905748196, "grad_norm": 2.0541656017303467, "learning_rate": 0.0005, "loss": 2.6710283756256104, "step": 6408 }, { "epoch": 0.7457528508261578, "grad_norm": 1.3595110177993774, "learning_rate": 0.0005, "loss": 2.5624735355377197, "step": 6409 }, { "epoch": 0.7458692110774959, "grad_norm": 1.2734416723251343, "learning_rate": 0.0005, "loss": 2.261059284210205, "step": 6410 }, { "epoch": 0.745985571328834, "grad_norm": 1.3261176347732544, "learning_rate": 0.0005, "loss": 2.8460044860839844, "step": 6411 }, { "epoch": 0.7461019315801722, "grad_norm": 1.2165181636810303, "learning_rate": 0.0005, "loss": 2.7829465866088867, "step": 6412 }, { "epoch": 0.7462182918315103, "grad_norm": 1.2679786682128906, "learning_rate": 0.0005, "loss": 2.722322463989258, "step": 6413 }, { "epoch": 0.7463346520828485, "grad_norm": 1.4325768947601318, "learning_rate": 0.0005, "loss": 2.170466423034668, "step": 6414 }, { "epoch": 0.7464510123341866, "grad_norm": 1.2458933591842651, "learning_rate": 0.0005, "loss": 1.8081130981445312, "step": 6415 }, { "epoch": 0.7465673725855247, "grad_norm": 1.0132416486740112, "learning_rate": 0.0005, "loss": 1.4809153079986572, "step": 6416 }, { "epoch": 0.746683732836863, "grad_norm": 1.4003684520721436, "learning_rate": 0.0005, "loss": 2.0325427055358887, "step": 6417 }, { "epoch": 0.7468000930882011, "grad_norm": 1.466229796409607, "learning_rate": 0.0005, "loss": 2.0571532249450684, "step": 6418 }, { "epoch": 0.7469164533395393, "grad_norm": 1.642175316810608, "learning_rate": 0.0005, "loss": 3.2135419845581055, "step": 6419 }, { "epoch": 0.7470328135908774, "grad_norm": 1.7317759990692139, "learning_rate": 0.0005, "loss": 2.704939365386963, "step": 6420 }, { "epoch": 0.7471491738422155, "grad_norm": 1.7794225215911865, "learning_rate": 0.0005, "loss": 2.7074592113494873, "step": 6421 }, { "epoch": 0.7472655340935537, "grad_norm": 2.6530113220214844, "learning_rate": 0.0005, "loss": 2.7600526809692383, "step": 6422 }, { "epoch": 0.7473818943448918, "grad_norm": 0.9717121720314026, "learning_rate": 0.0005, "loss": 1.9161970615386963, "step": 6423 }, { "epoch": 0.74749825459623, "grad_norm": 1.1640828847885132, "learning_rate": 0.0005, "loss": 2.1096272468566895, "step": 6424 }, { "epoch": 0.7476146148475681, "grad_norm": 1.3820154666900635, "learning_rate": 0.0005, "loss": 3.3158435821533203, "step": 6425 }, { "epoch": 0.7477309750989062, "grad_norm": 2.039572238922119, "learning_rate": 0.0005, "loss": 3.379930257797241, "step": 6426 }, { "epoch": 0.7478473353502444, "grad_norm": 1.370453953742981, "learning_rate": 0.0005, "loss": 2.8834681510925293, "step": 6427 }, { "epoch": 0.7479636956015825, "grad_norm": 1.344191312789917, "learning_rate": 0.0005, "loss": 2.227224349975586, "step": 6428 }, { "epoch": 0.7480800558529206, "grad_norm": 1.8066247701644897, "learning_rate": 0.0005, "loss": 2.8497164249420166, "step": 6429 }, { "epoch": 0.7481964161042588, "grad_norm": 1.1632041931152344, "learning_rate": 0.0005, "loss": 2.948301076889038, "step": 6430 }, { "epoch": 0.7483127763555969, "grad_norm": 1.0332608222961426, "learning_rate": 0.0005, "loss": 2.8136305809020996, "step": 6431 }, { "epoch": 0.7484291366069351, "grad_norm": 1.751071572303772, "learning_rate": 0.0005, "loss": 2.9381513595581055, "step": 6432 }, { "epoch": 0.7485454968582732, "grad_norm": 0.6714087724685669, "learning_rate": 0.0005, "loss": 1.3473711013793945, "step": 6433 }, { "epoch": 0.7486618571096113, "grad_norm": 1.406733512878418, "learning_rate": 0.0005, "loss": 2.225144863128662, "step": 6434 }, { "epoch": 0.7487782173609495, "grad_norm": 1.0529701709747314, "learning_rate": 0.0005, "loss": 1.972161889076233, "step": 6435 }, { "epoch": 0.7488945776122876, "grad_norm": 1.0887507200241089, "learning_rate": 0.0005, "loss": 2.2437620162963867, "step": 6436 }, { "epoch": 0.7490109378636258, "grad_norm": 1.3925511837005615, "learning_rate": 0.0005, "loss": 3.04331111907959, "step": 6437 }, { "epoch": 0.7491272981149639, "grad_norm": 1.002661108970642, "learning_rate": 0.0005, "loss": 2.4719252586364746, "step": 6438 }, { "epoch": 0.749243658366302, "grad_norm": 1.4376240968704224, "learning_rate": 0.0005, "loss": 2.267253875732422, "step": 6439 }, { "epoch": 0.7493600186176402, "grad_norm": 1.7121764421463013, "learning_rate": 0.0005, "loss": 2.433648109436035, "step": 6440 }, { "epoch": 0.7494763788689783, "grad_norm": 1.4870953559875488, "learning_rate": 0.0005, "loss": 2.536313533782959, "step": 6441 }, { "epoch": 0.7495927391203165, "grad_norm": 1.5500493049621582, "learning_rate": 0.0005, "loss": 3.1862220764160156, "step": 6442 }, { "epoch": 0.7497090993716546, "grad_norm": 1.5171666145324707, "learning_rate": 0.0005, "loss": 2.342827081680298, "step": 6443 }, { "epoch": 0.7498254596229927, "grad_norm": 1.6136353015899658, "learning_rate": 0.0005, "loss": 2.8868768215179443, "step": 6444 }, { "epoch": 0.749941819874331, "grad_norm": 1.0841704607009888, "learning_rate": 0.0005, "loss": 2.503387451171875, "step": 6445 }, { "epoch": 0.750058180125669, "grad_norm": 0.96712726354599, "learning_rate": 0.0005, "loss": 1.3406858444213867, "step": 6446 }, { "epoch": 0.7501745403770073, "grad_norm": 0.9767398238182068, "learning_rate": 0.0005, "loss": 2.199378728866577, "step": 6447 }, { "epoch": 0.7502909006283454, "grad_norm": 2.148742914199829, "learning_rate": 0.0005, "loss": 2.9760568141937256, "step": 6448 }, { "epoch": 0.7504072608796835, "grad_norm": 3.2088265419006348, "learning_rate": 0.0005, "loss": 1.7971693277359009, "step": 6449 }, { "epoch": 0.7505236211310217, "grad_norm": 1.2024962902069092, "learning_rate": 0.0005, "loss": 2.924820899963379, "step": 6450 }, { "epoch": 0.7506399813823598, "grad_norm": 0.6930612921714783, "learning_rate": 0.0005, "loss": 1.7736839056015015, "step": 6451 }, { "epoch": 0.750756341633698, "grad_norm": 0.6132990717887878, "learning_rate": 0.0005, "loss": 1.3287701606750488, "step": 6452 }, { "epoch": 0.7508727018850361, "grad_norm": 1.2174361944198608, "learning_rate": 0.0005, "loss": 1.8227970600128174, "step": 6453 }, { "epoch": 0.7509890621363742, "grad_norm": 1.6945855617523193, "learning_rate": 0.0005, "loss": 2.918015480041504, "step": 6454 }, { "epoch": 0.7511054223877124, "grad_norm": 2.041717290878296, "learning_rate": 0.0005, "loss": 3.1634888648986816, "step": 6455 }, { "epoch": 0.7512217826390505, "grad_norm": 2.1607375144958496, "learning_rate": 0.0005, "loss": 2.6664741039276123, "step": 6456 }, { "epoch": 0.7513381428903887, "grad_norm": 1.3948255777359009, "learning_rate": 0.0005, "loss": 2.707347869873047, "step": 6457 }, { "epoch": 0.7514545031417268, "grad_norm": 0.8751952648162842, "learning_rate": 0.0005, "loss": 1.676638126373291, "step": 6458 }, { "epoch": 0.7515708633930649, "grad_norm": 1.985247254371643, "learning_rate": 0.0005, "loss": 2.5301876068115234, "step": 6459 }, { "epoch": 0.7516872236444031, "grad_norm": 1.806390643119812, "learning_rate": 0.0005, "loss": 2.330979585647583, "step": 6460 }, { "epoch": 0.7518035838957412, "grad_norm": 2.029679298400879, "learning_rate": 0.0005, "loss": 2.940577745437622, "step": 6461 }, { "epoch": 0.7519199441470794, "grad_norm": 1.2048559188842773, "learning_rate": 0.0005, "loss": 2.2119157314300537, "step": 6462 }, { "epoch": 0.7520363043984175, "grad_norm": 1.2023564577102661, "learning_rate": 0.0005, "loss": 2.5548133850097656, "step": 6463 }, { "epoch": 0.7521526646497556, "grad_norm": 1.499509334564209, "learning_rate": 0.0005, "loss": 2.528398275375366, "step": 6464 }, { "epoch": 0.7522690249010938, "grad_norm": 1.327142357826233, "learning_rate": 0.0005, "loss": 3.25126576423645, "step": 6465 }, { "epoch": 0.7523853851524319, "grad_norm": 1.368011236190796, "learning_rate": 0.0005, "loss": 1.9712966680526733, "step": 6466 }, { "epoch": 0.75250174540377, "grad_norm": 1.026809811592102, "learning_rate": 0.0005, "loss": 2.073751449584961, "step": 6467 }, { "epoch": 0.7526181056551082, "grad_norm": 1.01758873462677, "learning_rate": 0.0005, "loss": 1.7017195224761963, "step": 6468 }, { "epoch": 0.7527344659064463, "grad_norm": 1.5233421325683594, "learning_rate": 0.0005, "loss": 2.7213282585144043, "step": 6469 }, { "epoch": 0.7528508261577845, "grad_norm": 3.1522507667541504, "learning_rate": 0.0005, "loss": 2.230900287628174, "step": 6470 }, { "epoch": 0.7529671864091226, "grad_norm": 2.017583131790161, "learning_rate": 0.0005, "loss": 2.355905771255493, "step": 6471 }, { "epoch": 0.7530835466604607, "grad_norm": 1.5102159976959229, "learning_rate": 0.0005, "loss": 2.1955175399780273, "step": 6472 }, { "epoch": 0.7531999069117989, "grad_norm": 1.2266889810562134, "learning_rate": 0.0005, "loss": 2.3803579807281494, "step": 6473 }, { "epoch": 0.753316267163137, "grad_norm": 1.5342425107955933, "learning_rate": 0.0005, "loss": 2.1091296672821045, "step": 6474 }, { "epoch": 0.7534326274144753, "grad_norm": 1.5155290365219116, "learning_rate": 0.0005, "loss": 1.8094573020935059, "step": 6475 }, { "epoch": 0.7535489876658134, "grad_norm": 1.9218260049819946, "learning_rate": 0.0005, "loss": 2.0726706981658936, "step": 6476 }, { "epoch": 0.7536653479171515, "grad_norm": 1.3243250846862793, "learning_rate": 0.0005, "loss": 2.3259332180023193, "step": 6477 }, { "epoch": 0.7537817081684897, "grad_norm": 1.9845136404037476, "learning_rate": 0.0005, "loss": 2.0008957386016846, "step": 6478 }, { "epoch": 0.7538980684198278, "grad_norm": 0.716528594493866, "learning_rate": 0.0005, "loss": 1.6515039205551147, "step": 6479 }, { "epoch": 0.754014428671166, "grad_norm": 1.5130876302719116, "learning_rate": 0.0005, "loss": 1.615256905555725, "step": 6480 }, { "epoch": 0.7541307889225041, "grad_norm": 1.6193736791610718, "learning_rate": 0.0005, "loss": 2.8856871128082275, "step": 6481 }, { "epoch": 0.7542471491738422, "grad_norm": 15.0086669921875, "learning_rate": 0.0005, "loss": 2.716242790222168, "step": 6482 }, { "epoch": 0.7543635094251804, "grad_norm": 1.4803274869918823, "learning_rate": 0.0005, "loss": 2.2105231285095215, "step": 6483 }, { "epoch": 0.7544798696765185, "grad_norm": 1.8721063137054443, "learning_rate": 0.0005, "loss": 2.485093593597412, "step": 6484 }, { "epoch": 0.7545962299278567, "grad_norm": 1.9517498016357422, "learning_rate": 0.0005, "loss": 1.9893168210983276, "step": 6485 }, { "epoch": 0.7547125901791948, "grad_norm": 0.9295997619628906, "learning_rate": 0.0005, "loss": 2.102144241333008, "step": 6486 }, { "epoch": 0.7548289504305329, "grad_norm": 1.121806263923645, "learning_rate": 0.0005, "loss": 1.7321298122406006, "step": 6487 }, { "epoch": 0.7549453106818711, "grad_norm": 0.9399700164794922, "learning_rate": 0.0005, "loss": 1.7729681730270386, "step": 6488 }, { "epoch": 0.7550616709332092, "grad_norm": 1.559400200843811, "learning_rate": 0.0005, "loss": 2.6147594451904297, "step": 6489 }, { "epoch": 0.7551780311845474, "grad_norm": 1.8293863534927368, "learning_rate": 0.0005, "loss": 2.8737564086914062, "step": 6490 }, { "epoch": 0.7552943914358855, "grad_norm": 1.271485686302185, "learning_rate": 0.0005, "loss": 2.192941904067993, "step": 6491 }, { "epoch": 0.7554107516872236, "grad_norm": 0.8124676942825317, "learning_rate": 0.0005, "loss": 1.959005355834961, "step": 6492 }, { "epoch": 0.7555271119385618, "grad_norm": 1.5315673351287842, "learning_rate": 0.0005, "loss": 2.397555351257324, "step": 6493 }, { "epoch": 0.7556434721898999, "grad_norm": 1.8151473999023438, "learning_rate": 0.0005, "loss": 2.716442346572876, "step": 6494 }, { "epoch": 0.7557598324412381, "grad_norm": 1.38180410861969, "learning_rate": 0.0005, "loss": 2.3072922229766846, "step": 6495 }, { "epoch": 0.7558761926925762, "grad_norm": 2.1972055435180664, "learning_rate": 0.0005, "loss": 3.0741167068481445, "step": 6496 }, { "epoch": 0.7559925529439143, "grad_norm": 1.8608815670013428, "learning_rate": 0.0005, "loss": 2.5317344665527344, "step": 6497 }, { "epoch": 0.7561089131952525, "grad_norm": 2.6849777698516846, "learning_rate": 0.0005, "loss": 2.3790979385375977, "step": 6498 }, { "epoch": 0.7562252734465906, "grad_norm": 1.3366425037384033, "learning_rate": 0.0005, "loss": 2.725107192993164, "step": 6499 }, { "epoch": 0.7563416336979288, "grad_norm": 0.8777281641960144, "learning_rate": 0.0005, "loss": 2.352268695831299, "step": 6500 }, { "epoch": 0.7564579939492669, "grad_norm": 1.6278526782989502, "learning_rate": 0.0005, "loss": 2.419281005859375, "step": 6501 }, { "epoch": 0.756574354200605, "grad_norm": 2.0308594703674316, "learning_rate": 0.0005, "loss": 2.9892730712890625, "step": 6502 }, { "epoch": 0.7566907144519432, "grad_norm": 1.8134223222732544, "learning_rate": 0.0005, "loss": 2.381532907485962, "step": 6503 }, { "epoch": 0.7568070747032813, "grad_norm": 1.5216456651687622, "learning_rate": 0.0005, "loss": 2.940260887145996, "step": 6504 }, { "epoch": 0.7569234349546196, "grad_norm": 1.654281735420227, "learning_rate": 0.0005, "loss": 3.055276870727539, "step": 6505 }, { "epoch": 0.7570397952059577, "grad_norm": 1.0349814891815186, "learning_rate": 0.0005, "loss": 2.0459840297698975, "step": 6506 }, { "epoch": 0.7571561554572958, "grad_norm": 1.0752143859863281, "learning_rate": 0.0005, "loss": 1.953561782836914, "step": 6507 }, { "epoch": 0.757272515708634, "grad_norm": 2.077186107635498, "learning_rate": 0.0005, "loss": 2.118417501449585, "step": 6508 }, { "epoch": 0.7573888759599721, "grad_norm": 1.387843370437622, "learning_rate": 0.0005, "loss": 2.0454554557800293, "step": 6509 }, { "epoch": 0.7575052362113102, "grad_norm": 1.185533881187439, "learning_rate": 0.0005, "loss": 2.242657423019409, "step": 6510 }, { "epoch": 0.7576215964626484, "grad_norm": 0.9749278426170349, "learning_rate": 0.0005, "loss": 2.3684580326080322, "step": 6511 }, { "epoch": 0.7577379567139865, "grad_norm": 1.8797252178192139, "learning_rate": 0.0005, "loss": 2.0195250511169434, "step": 6512 }, { "epoch": 0.7578543169653247, "grad_norm": 1.2402280569076538, "learning_rate": 0.0005, "loss": 2.997610092163086, "step": 6513 }, { "epoch": 0.7579706772166628, "grad_norm": 1.2400702238082886, "learning_rate": 0.0005, "loss": 2.883394479751587, "step": 6514 }, { "epoch": 0.7580870374680009, "grad_norm": 1.2854028940200806, "learning_rate": 0.0005, "loss": 2.548086404800415, "step": 6515 }, { "epoch": 0.7582033977193391, "grad_norm": 1.3869577646255493, "learning_rate": 0.0005, "loss": 2.7325034141540527, "step": 6516 }, { "epoch": 0.7583197579706772, "grad_norm": 1.552189588546753, "learning_rate": 0.0005, "loss": 2.418854236602783, "step": 6517 }, { "epoch": 0.7584361182220154, "grad_norm": 0.9230902194976807, "learning_rate": 0.0005, "loss": 2.3353564739227295, "step": 6518 }, { "epoch": 0.7585524784733535, "grad_norm": 1.2980306148529053, "learning_rate": 0.0005, "loss": 2.822758197784424, "step": 6519 }, { "epoch": 0.7586688387246916, "grad_norm": 2.0789310932159424, "learning_rate": 0.0005, "loss": 2.9602150917053223, "step": 6520 }, { "epoch": 0.7587851989760298, "grad_norm": 0.6756019592285156, "learning_rate": 0.0005, "loss": 1.762083888053894, "step": 6521 }, { "epoch": 0.7589015592273679, "grad_norm": 1.0843614339828491, "learning_rate": 0.0005, "loss": 2.785208225250244, "step": 6522 }, { "epoch": 0.7590179194787061, "grad_norm": 0.9753291606903076, "learning_rate": 0.0005, "loss": 2.329352617263794, "step": 6523 }, { "epoch": 0.7591342797300442, "grad_norm": 1.2192602157592773, "learning_rate": 0.0005, "loss": 2.1923205852508545, "step": 6524 }, { "epoch": 0.7592506399813823, "grad_norm": 1.5193066596984863, "learning_rate": 0.0005, "loss": 2.159729480743408, "step": 6525 }, { "epoch": 0.7593670002327205, "grad_norm": 1.4600938558578491, "learning_rate": 0.0005, "loss": 2.366823196411133, "step": 6526 }, { "epoch": 0.7594833604840586, "grad_norm": 2.6593384742736816, "learning_rate": 0.0005, "loss": 2.792412757873535, "step": 6527 }, { "epoch": 0.7595997207353968, "grad_norm": 1.2839841842651367, "learning_rate": 0.0005, "loss": 2.070755958557129, "step": 6528 }, { "epoch": 0.7597160809867349, "grad_norm": 1.4619472026824951, "learning_rate": 0.0005, "loss": 2.4211623668670654, "step": 6529 }, { "epoch": 0.759832441238073, "grad_norm": 1.3719288110733032, "learning_rate": 0.0005, "loss": 2.562445878982544, "step": 6530 }, { "epoch": 0.7599488014894112, "grad_norm": 1.981100082397461, "learning_rate": 0.0005, "loss": 2.650597095489502, "step": 6531 }, { "epoch": 0.7600651617407493, "grad_norm": 1.421478271484375, "learning_rate": 0.0005, "loss": 2.797104835510254, "step": 6532 }, { "epoch": 0.7601815219920875, "grad_norm": 1.3474892377853394, "learning_rate": 0.0005, "loss": 2.786017894744873, "step": 6533 }, { "epoch": 0.7602978822434256, "grad_norm": 1.3906599283218384, "learning_rate": 0.0005, "loss": 3.487746000289917, "step": 6534 }, { "epoch": 0.7604142424947637, "grad_norm": 1.3178150653839111, "learning_rate": 0.0005, "loss": 1.5082758665084839, "step": 6535 }, { "epoch": 0.760530602746102, "grad_norm": 1.6164528131484985, "learning_rate": 0.0005, "loss": 2.1698482036590576, "step": 6536 }, { "epoch": 0.7606469629974401, "grad_norm": 1.7868841886520386, "learning_rate": 0.0005, "loss": 2.346409559249878, "step": 6537 }, { "epoch": 0.7607633232487783, "grad_norm": 1.5369727611541748, "learning_rate": 0.0005, "loss": 1.9165714979171753, "step": 6538 }, { "epoch": 0.7608796835001164, "grad_norm": 1.7866986989974976, "learning_rate": 0.0005, "loss": 3.0761232376098633, "step": 6539 }, { "epoch": 0.7609960437514545, "grad_norm": 2.183962106704712, "learning_rate": 0.0005, "loss": 1.9211682081222534, "step": 6540 }, { "epoch": 0.7611124040027927, "grad_norm": 1.3221046924591064, "learning_rate": 0.0005, "loss": 2.1835854053497314, "step": 6541 }, { "epoch": 0.7612287642541308, "grad_norm": 2.087190866470337, "learning_rate": 0.0005, "loss": 2.264843463897705, "step": 6542 }, { "epoch": 0.761345124505469, "grad_norm": 1.3496204614639282, "learning_rate": 0.0005, "loss": 1.867402195930481, "step": 6543 }, { "epoch": 0.7614614847568071, "grad_norm": 1.9012055397033691, "learning_rate": 0.0005, "loss": 2.679211139678955, "step": 6544 }, { "epoch": 0.7615778450081452, "grad_norm": 1.643144130706787, "learning_rate": 0.0005, "loss": 1.8865621089935303, "step": 6545 }, { "epoch": 0.7616942052594834, "grad_norm": 1.243139624595642, "learning_rate": 0.0005, "loss": 2.582245111465454, "step": 6546 }, { "epoch": 0.7618105655108215, "grad_norm": 1.3206837177276611, "learning_rate": 0.0005, "loss": 2.7951340675354004, "step": 6547 }, { "epoch": 0.7619269257621597, "grad_norm": 1.3155817985534668, "learning_rate": 0.0005, "loss": 1.6831269264221191, "step": 6548 }, { "epoch": 0.7620432860134978, "grad_norm": 1.6920537948608398, "learning_rate": 0.0005, "loss": 2.76589298248291, "step": 6549 }, { "epoch": 0.7621596462648359, "grad_norm": 1.0751286745071411, "learning_rate": 0.0005, "loss": 2.0342061519622803, "step": 6550 }, { "epoch": 0.7622760065161741, "grad_norm": 1.4531430006027222, "learning_rate": 0.0005, "loss": 2.04874587059021, "step": 6551 }, { "epoch": 0.7623923667675122, "grad_norm": 1.9632481336593628, "learning_rate": 0.0005, "loss": 2.726877212524414, "step": 6552 }, { "epoch": 0.7625087270188503, "grad_norm": 0.6389368772506714, "learning_rate": 0.0005, "loss": 1.5802085399627686, "step": 6553 }, { "epoch": 0.7626250872701885, "grad_norm": 2.025306463241577, "learning_rate": 0.0005, "loss": 3.015847682952881, "step": 6554 }, { "epoch": 0.7627414475215266, "grad_norm": 1.543422818183899, "learning_rate": 0.0005, "loss": 2.6206064224243164, "step": 6555 }, { "epoch": 0.7628578077728648, "grad_norm": 1.4921793937683105, "learning_rate": 0.0005, "loss": 2.548741340637207, "step": 6556 }, { "epoch": 0.7629741680242029, "grad_norm": 0.5540951490402222, "learning_rate": 0.0005, "loss": 1.0942447185516357, "step": 6557 }, { "epoch": 0.763090528275541, "grad_norm": 1.3840875625610352, "learning_rate": 0.0005, "loss": 2.3521041870117188, "step": 6558 }, { "epoch": 0.7632068885268792, "grad_norm": 1.5327931642532349, "learning_rate": 0.0005, "loss": 2.659482955932617, "step": 6559 }, { "epoch": 0.7633232487782173, "grad_norm": 1.1821174621582031, "learning_rate": 0.0005, "loss": 1.570805311203003, "step": 6560 }, { "epoch": 0.7634396090295555, "grad_norm": 1.5932202339172363, "learning_rate": 0.0005, "loss": 2.614126682281494, "step": 6561 }, { "epoch": 0.7635559692808936, "grad_norm": 1.6031512022018433, "learning_rate": 0.0005, "loss": 2.2690770626068115, "step": 6562 }, { "epoch": 0.7636723295322317, "grad_norm": 1.561739206314087, "learning_rate": 0.0005, "loss": 2.1866188049316406, "step": 6563 }, { "epoch": 0.76378868978357, "grad_norm": 0.7158169746398926, "learning_rate": 0.0005, "loss": 2.031118154525757, "step": 6564 }, { "epoch": 0.763905050034908, "grad_norm": 1.462701678276062, "learning_rate": 0.0005, "loss": 3.0644633769989014, "step": 6565 }, { "epoch": 0.7640214102862463, "grad_norm": 1.5624433755874634, "learning_rate": 0.0005, "loss": 1.9419076442718506, "step": 6566 }, { "epoch": 0.7641377705375844, "grad_norm": 1.4401471614837646, "learning_rate": 0.0005, "loss": 3.4493985176086426, "step": 6567 }, { "epoch": 0.7642541307889225, "grad_norm": 1.1668007373809814, "learning_rate": 0.0005, "loss": 2.0674045085906982, "step": 6568 }, { "epoch": 0.7643704910402607, "grad_norm": 1.430259346961975, "learning_rate": 0.0005, "loss": 2.3104634284973145, "step": 6569 }, { "epoch": 0.7644868512915988, "grad_norm": 1.350536823272705, "learning_rate": 0.0005, "loss": 4.026108741760254, "step": 6570 }, { "epoch": 0.764603211542937, "grad_norm": 8.464850425720215, "learning_rate": 0.0005, "loss": 3.2302427291870117, "step": 6571 }, { "epoch": 0.7647195717942751, "grad_norm": 1.438124418258667, "learning_rate": 0.0005, "loss": 2.332164764404297, "step": 6572 }, { "epoch": 0.7648359320456132, "grad_norm": 1.7644026279449463, "learning_rate": 0.0005, "loss": 1.8061174154281616, "step": 6573 }, { "epoch": 0.7649522922969514, "grad_norm": 1.6257153749465942, "learning_rate": 0.0005, "loss": 3.2533750534057617, "step": 6574 }, { "epoch": 0.7650686525482895, "grad_norm": 1.4908843040466309, "learning_rate": 0.0005, "loss": 2.2645370960235596, "step": 6575 }, { "epoch": 0.7651850127996277, "grad_norm": 2.017049551010132, "learning_rate": 0.0005, "loss": 2.8385186195373535, "step": 6576 }, { "epoch": 0.7653013730509658, "grad_norm": 1.6944893598556519, "learning_rate": 0.0005, "loss": 2.23663592338562, "step": 6577 }, { "epoch": 0.7654177333023039, "grad_norm": 1.5807539224624634, "learning_rate": 0.0005, "loss": 3.623323678970337, "step": 6578 }, { "epoch": 0.7655340935536421, "grad_norm": 1.0320266485214233, "learning_rate": 0.0005, "loss": 2.2995831966400146, "step": 6579 }, { "epoch": 0.7656504538049802, "grad_norm": 0.9594892859458923, "learning_rate": 0.0005, "loss": 2.380121946334839, "step": 6580 }, { "epoch": 0.7657668140563184, "grad_norm": 1.0869085788726807, "learning_rate": 0.0005, "loss": 2.915482521057129, "step": 6581 }, { "epoch": 0.7658831743076565, "grad_norm": 1.160295009613037, "learning_rate": 0.0005, "loss": 2.370718240737915, "step": 6582 }, { "epoch": 0.7659995345589946, "grad_norm": 1.6814733743667603, "learning_rate": 0.0005, "loss": 3.1593918800354004, "step": 6583 }, { "epoch": 0.7661158948103328, "grad_norm": 1.1129424571990967, "learning_rate": 0.0005, "loss": 1.927170753479004, "step": 6584 }, { "epoch": 0.7662322550616709, "grad_norm": 1.303275465965271, "learning_rate": 0.0005, "loss": 2.836569309234619, "step": 6585 }, { "epoch": 0.7663486153130091, "grad_norm": 1.1350849866867065, "learning_rate": 0.0005, "loss": 1.8996373414993286, "step": 6586 }, { "epoch": 0.7664649755643472, "grad_norm": 1.1354761123657227, "learning_rate": 0.0005, "loss": 1.9449641704559326, "step": 6587 }, { "epoch": 0.7665813358156853, "grad_norm": 1.1075444221496582, "learning_rate": 0.0005, "loss": 2.146987199783325, "step": 6588 }, { "epoch": 0.7666976960670235, "grad_norm": 1.2440886497497559, "learning_rate": 0.0005, "loss": 1.8924734592437744, "step": 6589 }, { "epoch": 0.7668140563183616, "grad_norm": 1.1137667894363403, "learning_rate": 0.0005, "loss": 1.8646209239959717, "step": 6590 }, { "epoch": 0.7669304165696997, "grad_norm": 1.0824904441833496, "learning_rate": 0.0005, "loss": 2.1795873641967773, "step": 6591 }, { "epoch": 0.767046776821038, "grad_norm": 1.0040233135223389, "learning_rate": 0.0005, "loss": 1.9751266241073608, "step": 6592 }, { "epoch": 0.767163137072376, "grad_norm": 1.1053733825683594, "learning_rate": 0.0005, "loss": 1.7668054103851318, "step": 6593 }, { "epoch": 0.7672794973237143, "grad_norm": 0.9969053864479065, "learning_rate": 0.0005, "loss": 2.2229504585266113, "step": 6594 }, { "epoch": 0.7673958575750524, "grad_norm": 1.2345597743988037, "learning_rate": 0.0005, "loss": 2.6868627071380615, "step": 6595 }, { "epoch": 0.7675122178263905, "grad_norm": 1.2313069105148315, "learning_rate": 0.0005, "loss": 2.5943474769592285, "step": 6596 }, { "epoch": 0.7676285780777287, "grad_norm": 0.9513518214225769, "learning_rate": 0.0005, "loss": 1.1606254577636719, "step": 6597 }, { "epoch": 0.7677449383290668, "grad_norm": 1.9240528345108032, "learning_rate": 0.0005, "loss": 3.029237747192383, "step": 6598 }, { "epoch": 0.767861298580405, "grad_norm": 2.7035393714904785, "learning_rate": 0.0005, "loss": 2.031770706176758, "step": 6599 }, { "epoch": 0.7679776588317431, "grad_norm": 1.1754133701324463, "learning_rate": 0.0005, "loss": 1.6298222541809082, "step": 6600 }, { "epoch": 0.7680940190830812, "grad_norm": 1.3530550003051758, "learning_rate": 0.0005, "loss": 2.4979217052459717, "step": 6601 }, { "epoch": 0.7682103793344194, "grad_norm": 1.0313117504119873, "learning_rate": 0.0005, "loss": 2.298213005065918, "step": 6602 }, { "epoch": 0.7683267395857575, "grad_norm": 1.454910159111023, "learning_rate": 0.0005, "loss": 2.9498047828674316, "step": 6603 }, { "epoch": 0.7684430998370957, "grad_norm": 1.5671076774597168, "learning_rate": 0.0005, "loss": 1.7665458917617798, "step": 6604 }, { "epoch": 0.7685594600884338, "grad_norm": 2.1456539630889893, "learning_rate": 0.0005, "loss": 1.9880177974700928, "step": 6605 }, { "epoch": 0.7686758203397719, "grad_norm": 0.7046297788619995, "learning_rate": 0.0005, "loss": 2.4200081825256348, "step": 6606 }, { "epoch": 0.7687921805911101, "grad_norm": 1.1645746231079102, "learning_rate": 0.0005, "loss": 2.315180540084839, "step": 6607 }, { "epoch": 0.7689085408424482, "grad_norm": 1.298017978668213, "learning_rate": 0.0005, "loss": 1.4113537073135376, "step": 6608 }, { "epoch": 0.7690249010937864, "grad_norm": 1.8540581464767456, "learning_rate": 0.0005, "loss": 2.8061084747314453, "step": 6609 }, { "epoch": 0.7691412613451245, "grad_norm": 1.2293227910995483, "learning_rate": 0.0005, "loss": 2.582231283187866, "step": 6610 }, { "epoch": 0.7692576215964626, "grad_norm": 1.101189374923706, "learning_rate": 0.0005, "loss": 1.835717797279358, "step": 6611 }, { "epoch": 0.7693739818478008, "grad_norm": 1.2415825128555298, "learning_rate": 0.0005, "loss": 2.2755777835845947, "step": 6612 }, { "epoch": 0.7694903420991389, "grad_norm": 1.3753582239151, "learning_rate": 0.0005, "loss": 2.4977951049804688, "step": 6613 }, { "epoch": 0.7696067023504771, "grad_norm": 1.187341332435608, "learning_rate": 0.0005, "loss": 1.6758778095245361, "step": 6614 }, { "epoch": 0.7697230626018152, "grad_norm": 0.8229736089706421, "learning_rate": 0.0005, "loss": 1.5562183856964111, "step": 6615 }, { "epoch": 0.7698394228531533, "grad_norm": 1.6420483589172363, "learning_rate": 0.0005, "loss": 1.810360312461853, "step": 6616 }, { "epoch": 0.7699557831044915, "grad_norm": 1.3010671138763428, "learning_rate": 0.0005, "loss": 2.4818930625915527, "step": 6617 }, { "epoch": 0.7700721433558296, "grad_norm": 1.7325748205184937, "learning_rate": 0.0005, "loss": 3.464158535003662, "step": 6618 }, { "epoch": 0.7701885036071678, "grad_norm": 0.9241393804550171, "learning_rate": 0.0005, "loss": 2.017214298248291, "step": 6619 }, { "epoch": 0.7703048638585059, "grad_norm": 1.9638856649398804, "learning_rate": 0.0005, "loss": 3.179739236831665, "step": 6620 }, { "epoch": 0.770421224109844, "grad_norm": 1.5262128114700317, "learning_rate": 0.0005, "loss": 2.641115188598633, "step": 6621 }, { "epoch": 0.7705375843611822, "grad_norm": 1.8322036266326904, "learning_rate": 0.0005, "loss": 2.221442222595215, "step": 6622 }, { "epoch": 0.7706539446125203, "grad_norm": 1.345701813697815, "learning_rate": 0.0005, "loss": 2.704756021499634, "step": 6623 }, { "epoch": 0.7707703048638586, "grad_norm": 0.6927066445350647, "learning_rate": 0.0005, "loss": 1.7946014404296875, "step": 6624 }, { "epoch": 0.7708866651151967, "grad_norm": 0.9140785336494446, "learning_rate": 0.0005, "loss": 1.3095874786376953, "step": 6625 }, { "epoch": 0.7710030253665348, "grad_norm": 2.319554328918457, "learning_rate": 0.0005, "loss": 3.3185906410217285, "step": 6626 }, { "epoch": 0.771119385617873, "grad_norm": 1.551537036895752, "learning_rate": 0.0005, "loss": 2.605886459350586, "step": 6627 }, { "epoch": 0.7712357458692111, "grad_norm": 1.50904381275177, "learning_rate": 0.0005, "loss": 2.5172414779663086, "step": 6628 }, { "epoch": 0.7713521061205493, "grad_norm": 0.8678286671638489, "learning_rate": 0.0005, "loss": 2.2915053367614746, "step": 6629 }, { "epoch": 0.7714684663718874, "grad_norm": 2.084821939468384, "learning_rate": 0.0005, "loss": 2.390920877456665, "step": 6630 }, { "epoch": 0.7715848266232255, "grad_norm": 1.6195504665374756, "learning_rate": 0.0005, "loss": 2.672321081161499, "step": 6631 }, { "epoch": 0.7717011868745637, "grad_norm": 1.3020882606506348, "learning_rate": 0.0005, "loss": 1.9042099714279175, "step": 6632 }, { "epoch": 0.7718175471259018, "grad_norm": 1.4621565341949463, "learning_rate": 0.0005, "loss": 1.9792245626449585, "step": 6633 }, { "epoch": 0.7719339073772399, "grad_norm": 1.1904605627059937, "learning_rate": 0.0005, "loss": 2.319490909576416, "step": 6634 }, { "epoch": 0.7720502676285781, "grad_norm": 1.2707573175430298, "learning_rate": 0.0005, "loss": 2.39631986618042, "step": 6635 }, { "epoch": 0.7721666278799162, "grad_norm": 1.221987247467041, "learning_rate": 0.0005, "loss": 2.3538122177124023, "step": 6636 }, { "epoch": 0.7722829881312544, "grad_norm": 2.113539695739746, "learning_rate": 0.0005, "loss": 2.683354616165161, "step": 6637 }, { "epoch": 0.7723993483825925, "grad_norm": 1.76398503780365, "learning_rate": 0.0005, "loss": 3.393886089324951, "step": 6638 }, { "epoch": 0.7725157086339306, "grad_norm": 2.0269994735717773, "learning_rate": 0.0005, "loss": 2.0368175506591797, "step": 6639 }, { "epoch": 0.7726320688852688, "grad_norm": 1.657485842704773, "learning_rate": 0.0005, "loss": 1.9822520017623901, "step": 6640 }, { "epoch": 0.7727484291366069, "grad_norm": 0.6362349390983582, "learning_rate": 0.0005, "loss": 1.4018131494522095, "step": 6641 }, { "epoch": 0.7728647893879451, "grad_norm": 0.8374517560005188, "learning_rate": 0.0005, "loss": 1.7059088945388794, "step": 6642 }, { "epoch": 0.7729811496392832, "grad_norm": 3.391843318939209, "learning_rate": 0.0005, "loss": 2.1605491638183594, "step": 6643 }, { "epoch": 0.7730975098906213, "grad_norm": 1.1479891538619995, "learning_rate": 0.0005, "loss": 1.6187047958374023, "step": 6644 }, { "epoch": 0.7732138701419595, "grad_norm": 1.4031606912612915, "learning_rate": 0.0005, "loss": 2.1857759952545166, "step": 6645 }, { "epoch": 0.7733302303932976, "grad_norm": 1.069178819656372, "learning_rate": 0.0005, "loss": 2.2222752571105957, "step": 6646 }, { "epoch": 0.7734465906446358, "grad_norm": 0.842315673828125, "learning_rate": 0.0005, "loss": 1.8875856399536133, "step": 6647 }, { "epoch": 0.7735629508959739, "grad_norm": 1.1627907752990723, "learning_rate": 0.0005, "loss": 2.0708212852478027, "step": 6648 }, { "epoch": 0.773679311147312, "grad_norm": 1.0748428106307983, "learning_rate": 0.0005, "loss": 2.8982982635498047, "step": 6649 }, { "epoch": 0.7737956713986502, "grad_norm": 1.6061577796936035, "learning_rate": 0.0005, "loss": 2.6597554683685303, "step": 6650 }, { "epoch": 0.7739120316499883, "grad_norm": 1.3667000532150269, "learning_rate": 0.0005, "loss": 2.690099000930786, "step": 6651 }, { "epoch": 0.7740283919013266, "grad_norm": 1.2164368629455566, "learning_rate": 0.0005, "loss": 1.9558510780334473, "step": 6652 }, { "epoch": 0.7741447521526647, "grad_norm": 0.9446958303451538, "learning_rate": 0.0005, "loss": 1.706611156463623, "step": 6653 }, { "epoch": 0.7742611124040027, "grad_norm": 1.3514692783355713, "learning_rate": 0.0005, "loss": 2.1837868690490723, "step": 6654 }, { "epoch": 0.774377472655341, "grad_norm": 1.7816083431243896, "learning_rate": 0.0005, "loss": 2.8063511848449707, "step": 6655 }, { "epoch": 0.7744938329066791, "grad_norm": 1.347769856452942, "learning_rate": 0.0005, "loss": 2.1877193450927734, "step": 6656 }, { "epoch": 0.7746101931580173, "grad_norm": 1.9048134088516235, "learning_rate": 0.0005, "loss": 3.044379472732544, "step": 6657 }, { "epoch": 0.7747265534093554, "grad_norm": 0.8033535480499268, "learning_rate": 0.0005, "loss": 1.9421132802963257, "step": 6658 }, { "epoch": 0.7748429136606935, "grad_norm": 1.5957096815109253, "learning_rate": 0.0005, "loss": 2.684288740158081, "step": 6659 }, { "epoch": 0.7749592739120317, "grad_norm": 1.4053784608840942, "learning_rate": 0.0005, "loss": 2.4471540451049805, "step": 6660 }, { "epoch": 0.7750756341633698, "grad_norm": 1.8661881685256958, "learning_rate": 0.0005, "loss": 2.3846654891967773, "step": 6661 }, { "epoch": 0.775191994414708, "grad_norm": 1.2817000150680542, "learning_rate": 0.0005, "loss": 2.299286365509033, "step": 6662 }, { "epoch": 0.7753083546660461, "grad_norm": 0.8688260912895203, "learning_rate": 0.0005, "loss": 1.691602349281311, "step": 6663 }, { "epoch": 0.7754247149173842, "grad_norm": 1.2279331684112549, "learning_rate": 0.0005, "loss": 2.8948984146118164, "step": 6664 }, { "epoch": 0.7755410751687224, "grad_norm": 1.2845252752304077, "learning_rate": 0.0005, "loss": 1.9423238039016724, "step": 6665 }, { "epoch": 0.7756574354200605, "grad_norm": 2.2908170223236084, "learning_rate": 0.0005, "loss": 3.0766801834106445, "step": 6666 }, { "epoch": 0.7757737956713987, "grad_norm": 1.191168189048767, "learning_rate": 0.0005, "loss": 1.941011667251587, "step": 6667 }, { "epoch": 0.7758901559227368, "grad_norm": 1.2244173288345337, "learning_rate": 0.0005, "loss": 1.734710454940796, "step": 6668 }, { "epoch": 0.7760065161740749, "grad_norm": 1.7671022415161133, "learning_rate": 0.0005, "loss": 2.4323842525482178, "step": 6669 }, { "epoch": 0.7761228764254131, "grad_norm": 1.5946084260940552, "learning_rate": 0.0005, "loss": 3.0969247817993164, "step": 6670 }, { "epoch": 0.7762392366767512, "grad_norm": 1.2738854885101318, "learning_rate": 0.0005, "loss": 2.176875352859497, "step": 6671 }, { "epoch": 0.7763555969280894, "grad_norm": 1.5104352235794067, "learning_rate": 0.0005, "loss": 2.4704720973968506, "step": 6672 }, { "epoch": 0.7764719571794275, "grad_norm": 1.2377256155014038, "learning_rate": 0.0005, "loss": 2.7482006549835205, "step": 6673 }, { "epoch": 0.7765883174307656, "grad_norm": 1.3818076848983765, "learning_rate": 0.0005, "loss": 2.3905062675476074, "step": 6674 }, { "epoch": 0.7767046776821038, "grad_norm": 1.2934315204620361, "learning_rate": 0.0005, "loss": 2.2143585681915283, "step": 6675 }, { "epoch": 0.7768210379334419, "grad_norm": 2.1366918087005615, "learning_rate": 0.0005, "loss": 2.4037625789642334, "step": 6676 }, { "epoch": 0.77693739818478, "grad_norm": 2.6409311294555664, "learning_rate": 0.0005, "loss": 3.7400827407836914, "step": 6677 }, { "epoch": 0.7770537584361182, "grad_norm": 1.4833396673202515, "learning_rate": 0.0005, "loss": 2.093245506286621, "step": 6678 }, { "epoch": 0.7771701186874563, "grad_norm": 1.514613151550293, "learning_rate": 0.0005, "loss": 1.963051199913025, "step": 6679 }, { "epoch": 0.7772864789387945, "grad_norm": 1.7286967039108276, "learning_rate": 0.0005, "loss": 2.3710923194885254, "step": 6680 }, { "epoch": 0.7774028391901326, "grad_norm": 0.9628117680549622, "learning_rate": 0.0005, "loss": 1.9240952730178833, "step": 6681 }, { "epoch": 0.7775191994414707, "grad_norm": 1.160936713218689, "learning_rate": 0.0005, "loss": 2.380887269973755, "step": 6682 }, { "epoch": 0.777635559692809, "grad_norm": 1.106406331062317, "learning_rate": 0.0005, "loss": 2.1237387657165527, "step": 6683 }, { "epoch": 0.777751919944147, "grad_norm": 9.030311584472656, "learning_rate": 0.0005, "loss": 2.2093749046325684, "step": 6684 }, { "epoch": 0.7778682801954853, "grad_norm": 0.93793123960495, "learning_rate": 0.0005, "loss": 1.7900779247283936, "step": 6685 }, { "epoch": 0.7779846404468234, "grad_norm": 1.3077672719955444, "learning_rate": 0.0005, "loss": 2.0899465084075928, "step": 6686 }, { "epoch": 0.7781010006981615, "grad_norm": 1.278946876525879, "learning_rate": 0.0005, "loss": 2.741029977798462, "step": 6687 }, { "epoch": 0.7782173609494997, "grad_norm": 1.7153486013412476, "learning_rate": 0.0005, "loss": 2.6018009185791016, "step": 6688 }, { "epoch": 0.7783337212008378, "grad_norm": 1.1012866497039795, "learning_rate": 0.0005, "loss": 2.4365851879119873, "step": 6689 }, { "epoch": 0.778450081452176, "grad_norm": 1.4532099962234497, "learning_rate": 0.0005, "loss": 1.5267080068588257, "step": 6690 }, { "epoch": 0.7785664417035141, "grad_norm": 1.9278507232666016, "learning_rate": 0.0005, "loss": 2.5718929767608643, "step": 6691 }, { "epoch": 0.7786828019548522, "grad_norm": 1.3877726793289185, "learning_rate": 0.0005, "loss": 2.8088021278381348, "step": 6692 }, { "epoch": 0.7787991622061904, "grad_norm": 1.223380446434021, "learning_rate": 0.0005, "loss": 1.8781863451004028, "step": 6693 }, { "epoch": 0.7789155224575285, "grad_norm": 1.1323318481445312, "learning_rate": 0.0005, "loss": 2.1458520889282227, "step": 6694 }, { "epoch": 0.7790318827088667, "grad_norm": 1.0938585996627808, "learning_rate": 0.0005, "loss": 2.12605619430542, "step": 6695 }, { "epoch": 0.7791482429602048, "grad_norm": 1.8255871534347534, "learning_rate": 0.0005, "loss": 2.645554542541504, "step": 6696 }, { "epoch": 0.7792646032115429, "grad_norm": 1.7596429586410522, "learning_rate": 0.0005, "loss": 2.7390542030334473, "step": 6697 }, { "epoch": 0.7793809634628811, "grad_norm": 0.8240475654602051, "learning_rate": 0.0005, "loss": 2.026571273803711, "step": 6698 }, { "epoch": 0.7794973237142192, "grad_norm": 1.4546476602554321, "learning_rate": 0.0005, "loss": 2.6701366901397705, "step": 6699 }, { "epoch": 0.7796136839655574, "grad_norm": 0.9539596438407898, "learning_rate": 0.0005, "loss": 1.9722493886947632, "step": 6700 }, { "epoch": 0.7797300442168955, "grad_norm": 0.7128525376319885, "learning_rate": 0.0005, "loss": 1.35613214969635, "step": 6701 }, { "epoch": 0.7798464044682336, "grad_norm": 1.0509402751922607, "learning_rate": 0.0005, "loss": 2.879660129547119, "step": 6702 }, { "epoch": 0.7799627647195718, "grad_norm": 1.5026761293411255, "learning_rate": 0.0005, "loss": 2.245363473892212, "step": 6703 }, { "epoch": 0.7800791249709099, "grad_norm": 1.5733740329742432, "learning_rate": 0.0005, "loss": 2.3079471588134766, "step": 6704 }, { "epoch": 0.7801954852222481, "grad_norm": 1.5719194412231445, "learning_rate": 0.0005, "loss": 2.212083101272583, "step": 6705 }, { "epoch": 0.7803118454735862, "grad_norm": 2.0520870685577393, "learning_rate": 0.0005, "loss": 2.8384275436401367, "step": 6706 }, { "epoch": 0.7804282057249243, "grad_norm": 1.8045371770858765, "learning_rate": 0.0005, "loss": 2.527707576751709, "step": 6707 }, { "epoch": 0.7805445659762625, "grad_norm": 1.6453688144683838, "learning_rate": 0.0005, "loss": 2.705047130584717, "step": 6708 }, { "epoch": 0.7806609262276006, "grad_norm": 0.9347249865531921, "learning_rate": 0.0005, "loss": 1.5353128910064697, "step": 6709 }, { "epoch": 0.7807772864789388, "grad_norm": 1.4382003545761108, "learning_rate": 0.0005, "loss": 2.4754576683044434, "step": 6710 }, { "epoch": 0.780893646730277, "grad_norm": 1.3260337114334106, "learning_rate": 0.0005, "loss": 2.294766664505005, "step": 6711 }, { "epoch": 0.781010006981615, "grad_norm": 1.7544699907302856, "learning_rate": 0.0005, "loss": 2.773099184036255, "step": 6712 }, { "epoch": 0.7811263672329533, "grad_norm": 1.3791823387145996, "learning_rate": 0.0005, "loss": 2.5151004791259766, "step": 6713 }, { "epoch": 0.7812427274842914, "grad_norm": 0.7112414836883545, "learning_rate": 0.0005, "loss": 1.4344501495361328, "step": 6714 }, { "epoch": 0.7813590877356295, "grad_norm": 1.7966293096542358, "learning_rate": 0.0005, "loss": 2.955489158630371, "step": 6715 }, { "epoch": 0.7814754479869677, "grad_norm": 1.4937143325805664, "learning_rate": 0.0005, "loss": 2.4721906185150146, "step": 6716 }, { "epoch": 0.7815918082383058, "grad_norm": 1.7277195453643799, "learning_rate": 0.0005, "loss": 2.8892245292663574, "step": 6717 }, { "epoch": 0.781708168489644, "grad_norm": 1.5338399410247803, "learning_rate": 0.0005, "loss": 2.0465002059936523, "step": 6718 }, { "epoch": 0.7818245287409821, "grad_norm": 1.2639800310134888, "learning_rate": 0.0005, "loss": 2.718071460723877, "step": 6719 }, { "epoch": 0.7819408889923202, "grad_norm": 1.33017897605896, "learning_rate": 0.0005, "loss": 2.798335075378418, "step": 6720 }, { "epoch": 0.7820572492436584, "grad_norm": 1.6600985527038574, "learning_rate": 0.0005, "loss": 2.648730993270874, "step": 6721 }, { "epoch": 0.7821736094949965, "grad_norm": 1.5046567916870117, "learning_rate": 0.0005, "loss": 2.409125804901123, "step": 6722 }, { "epoch": 0.7822899697463347, "grad_norm": 1.4196909666061401, "learning_rate": 0.0005, "loss": 2.6186604499816895, "step": 6723 }, { "epoch": 0.7824063299976728, "grad_norm": 1.4789166450500488, "learning_rate": 0.0005, "loss": 2.974740982055664, "step": 6724 }, { "epoch": 0.7825226902490109, "grad_norm": 1.7636475563049316, "learning_rate": 0.0005, "loss": 2.3096508979797363, "step": 6725 }, { "epoch": 0.7826390505003491, "grad_norm": 1.501306176185608, "learning_rate": 0.0005, "loss": 1.7053947448730469, "step": 6726 }, { "epoch": 0.7827554107516872, "grad_norm": 1.730636477470398, "learning_rate": 0.0005, "loss": 2.65917706489563, "step": 6727 }, { "epoch": 0.7828717710030254, "grad_norm": 1.4151710271835327, "learning_rate": 0.0005, "loss": 2.443145275115967, "step": 6728 }, { "epoch": 0.7829881312543635, "grad_norm": 0.9942631125450134, "learning_rate": 0.0005, "loss": 1.5605785846710205, "step": 6729 }, { "epoch": 0.7831044915057016, "grad_norm": 1.2143299579620361, "learning_rate": 0.0005, "loss": 1.8863142728805542, "step": 6730 }, { "epoch": 0.7832208517570398, "grad_norm": 1.147915005683899, "learning_rate": 0.0005, "loss": 2.5544564723968506, "step": 6731 }, { "epoch": 0.7833372120083779, "grad_norm": 1.3690294027328491, "learning_rate": 0.0005, "loss": 2.572099208831787, "step": 6732 }, { "epoch": 0.7834535722597161, "grad_norm": 1.3497899770736694, "learning_rate": 0.0005, "loss": 1.9391775131225586, "step": 6733 }, { "epoch": 0.7835699325110542, "grad_norm": 1.2064470052719116, "learning_rate": 0.0005, "loss": 2.0496513843536377, "step": 6734 }, { "epoch": 0.7836862927623923, "grad_norm": 1.8124465942382812, "learning_rate": 0.0005, "loss": 1.4908453226089478, "step": 6735 }, { "epoch": 0.7838026530137305, "grad_norm": 1.3273684978485107, "learning_rate": 0.0005, "loss": 1.9245188236236572, "step": 6736 }, { "epoch": 0.7839190132650686, "grad_norm": 1.2730597257614136, "learning_rate": 0.0005, "loss": 2.5025486946105957, "step": 6737 }, { "epoch": 0.7840353735164068, "grad_norm": 1.5037500858306885, "learning_rate": 0.0005, "loss": 2.383152961730957, "step": 6738 }, { "epoch": 0.7841517337677449, "grad_norm": 2.3662548065185547, "learning_rate": 0.0005, "loss": 1.6600520610809326, "step": 6739 }, { "epoch": 0.784268094019083, "grad_norm": 2.7695353031158447, "learning_rate": 0.0005, "loss": 3.2135324478149414, "step": 6740 }, { "epoch": 0.7843844542704212, "grad_norm": 1.8049970865249634, "learning_rate": 0.0005, "loss": 2.570204496383667, "step": 6741 }, { "epoch": 0.7845008145217593, "grad_norm": 1.5934549570083618, "learning_rate": 0.0005, "loss": 2.9032998085021973, "step": 6742 }, { "epoch": 0.7846171747730976, "grad_norm": 2.202214241027832, "learning_rate": 0.0005, "loss": 2.467749834060669, "step": 6743 }, { "epoch": 0.7847335350244357, "grad_norm": 1.2631548643112183, "learning_rate": 0.0005, "loss": 2.076289653778076, "step": 6744 }, { "epoch": 0.7848498952757738, "grad_norm": 0.9915600419044495, "learning_rate": 0.0005, "loss": 2.3657917976379395, "step": 6745 }, { "epoch": 0.784966255527112, "grad_norm": 0.9935624599456787, "learning_rate": 0.0005, "loss": 2.334374189376831, "step": 6746 }, { "epoch": 0.7850826157784501, "grad_norm": 1.1616294384002686, "learning_rate": 0.0005, "loss": 2.0240554809570312, "step": 6747 }, { "epoch": 0.7851989760297883, "grad_norm": 1.2805352210998535, "learning_rate": 0.0005, "loss": 1.9862291812896729, "step": 6748 }, { "epoch": 0.7853153362811264, "grad_norm": 1.103901982307434, "learning_rate": 0.0005, "loss": 2.0697648525238037, "step": 6749 }, { "epoch": 0.7854316965324645, "grad_norm": 1.3941456079483032, "learning_rate": 0.0005, "loss": 2.8119583129882812, "step": 6750 }, { "epoch": 0.7855480567838027, "grad_norm": 1.644770622253418, "learning_rate": 0.0005, "loss": 2.5924837589263916, "step": 6751 }, { "epoch": 0.7856644170351408, "grad_norm": 1.1814570426940918, "learning_rate": 0.0005, "loss": 1.5847878456115723, "step": 6752 }, { "epoch": 0.785780777286479, "grad_norm": 1.2599806785583496, "learning_rate": 0.0005, "loss": 2.3496294021606445, "step": 6753 }, { "epoch": 0.7858971375378171, "grad_norm": 0.7313160300254822, "learning_rate": 0.0005, "loss": 2.016456127166748, "step": 6754 }, { "epoch": 0.7860134977891552, "grad_norm": 1.210172414779663, "learning_rate": 0.0005, "loss": 2.660076141357422, "step": 6755 }, { "epoch": 0.7861298580404934, "grad_norm": 1.2443575859069824, "learning_rate": 0.0005, "loss": 2.3076696395874023, "step": 6756 }, { "epoch": 0.7862462182918315, "grad_norm": 2.1875412464141846, "learning_rate": 0.0005, "loss": 1.6028668880462646, "step": 6757 }, { "epoch": 0.7863625785431696, "grad_norm": 1.3263931274414062, "learning_rate": 0.0005, "loss": 3.0583858489990234, "step": 6758 }, { "epoch": 0.7864789387945078, "grad_norm": 1.5563448667526245, "learning_rate": 0.0005, "loss": 3.8978593349456787, "step": 6759 }, { "epoch": 0.7865952990458459, "grad_norm": 2.753436803817749, "learning_rate": 0.0005, "loss": 1.9590741395950317, "step": 6760 }, { "epoch": 0.7867116592971841, "grad_norm": 1.322919487953186, "learning_rate": 0.0005, "loss": 2.780784845352173, "step": 6761 }, { "epoch": 0.7868280195485222, "grad_norm": 1.2651621103286743, "learning_rate": 0.0005, "loss": 2.3989100456237793, "step": 6762 }, { "epoch": 0.7869443797998603, "grad_norm": 1.6778672933578491, "learning_rate": 0.0005, "loss": 2.170388698577881, "step": 6763 }, { "epoch": 0.7870607400511985, "grad_norm": 1.5347321033477783, "learning_rate": 0.0005, "loss": 3.975445032119751, "step": 6764 }, { "epoch": 0.7871771003025366, "grad_norm": 1.2822518348693848, "learning_rate": 0.0005, "loss": 2.2495622634887695, "step": 6765 }, { "epoch": 0.7872934605538748, "grad_norm": 1.5166056156158447, "learning_rate": 0.0005, "loss": 2.7783427238464355, "step": 6766 }, { "epoch": 0.7874098208052129, "grad_norm": 1.6590622663497925, "learning_rate": 0.0005, "loss": 1.8989663124084473, "step": 6767 }, { "epoch": 0.787526181056551, "grad_norm": 1.6903462409973145, "learning_rate": 0.0005, "loss": 2.481417179107666, "step": 6768 }, { "epoch": 0.7876425413078892, "grad_norm": 1.0861846208572388, "learning_rate": 0.0005, "loss": 1.5938634872436523, "step": 6769 }, { "epoch": 0.7877589015592273, "grad_norm": 1.4957647323608398, "learning_rate": 0.0005, "loss": 1.307134747505188, "step": 6770 }, { "epoch": 0.7878752618105656, "grad_norm": 3.8311891555786133, "learning_rate": 0.0005, "loss": 2.653261423110962, "step": 6771 }, { "epoch": 0.7879916220619037, "grad_norm": 1.6160966157913208, "learning_rate": 0.0005, "loss": 1.7142095565795898, "step": 6772 }, { "epoch": 0.7881079823132418, "grad_norm": 1.5127819776535034, "learning_rate": 0.0005, "loss": 2.1808791160583496, "step": 6773 }, { "epoch": 0.78822434256458, "grad_norm": 1.7711145877838135, "learning_rate": 0.0005, "loss": 3.216188430786133, "step": 6774 }, { "epoch": 0.7883407028159181, "grad_norm": 1.3988568782806396, "learning_rate": 0.0005, "loss": 2.776702880859375, "step": 6775 }, { "epoch": 0.7884570630672563, "grad_norm": 1.7854281663894653, "learning_rate": 0.0005, "loss": 2.772651195526123, "step": 6776 }, { "epoch": 0.7885734233185944, "grad_norm": 8.881709098815918, "learning_rate": 0.0005, "loss": 3.665534019470215, "step": 6777 }, { "epoch": 0.7886897835699325, "grad_norm": 1.417786955833435, "learning_rate": 0.0005, "loss": 2.5972633361816406, "step": 6778 }, { "epoch": 0.7888061438212707, "grad_norm": 1.0039488077163696, "learning_rate": 0.0005, "loss": 1.6281180381774902, "step": 6779 }, { "epoch": 0.7889225040726088, "grad_norm": 1.1648036241531372, "learning_rate": 0.0005, "loss": 1.8318647146224976, "step": 6780 }, { "epoch": 0.789038864323947, "grad_norm": 2.05019474029541, "learning_rate": 0.0005, "loss": 2.881925582885742, "step": 6781 }, { "epoch": 0.7891552245752851, "grad_norm": 1.8270504474639893, "learning_rate": 0.0005, "loss": 2.921717643737793, "step": 6782 }, { "epoch": 0.7892715848266232, "grad_norm": 1.3043205738067627, "learning_rate": 0.0005, "loss": 2.0527467727661133, "step": 6783 }, { "epoch": 0.7893879450779614, "grad_norm": 3.2029709815979004, "learning_rate": 0.0005, "loss": 2.277644157409668, "step": 6784 }, { "epoch": 0.7895043053292995, "grad_norm": 1.4927762746810913, "learning_rate": 0.0005, "loss": 2.8135499954223633, "step": 6785 }, { "epoch": 0.7896206655806377, "grad_norm": 1.519026517868042, "learning_rate": 0.0005, "loss": 2.060697555541992, "step": 6786 }, { "epoch": 0.7897370258319758, "grad_norm": 1.670053243637085, "learning_rate": 0.0005, "loss": 2.989266872406006, "step": 6787 }, { "epoch": 0.7898533860833139, "grad_norm": 4.136087894439697, "learning_rate": 0.0005, "loss": 2.0917515754699707, "step": 6788 }, { "epoch": 0.7899697463346521, "grad_norm": 2.0708484649658203, "learning_rate": 0.0005, "loss": 2.7650628089904785, "step": 6789 }, { "epoch": 0.7900861065859902, "grad_norm": 0.8713362216949463, "learning_rate": 0.0005, "loss": 1.9397480487823486, "step": 6790 }, { "epoch": 0.7902024668373284, "grad_norm": 0.9895966053009033, "learning_rate": 0.0005, "loss": 1.6541802883148193, "step": 6791 }, { "epoch": 0.7903188270886665, "grad_norm": 1.5480461120605469, "learning_rate": 0.0005, "loss": 2.378096103668213, "step": 6792 }, { "epoch": 0.7904351873400046, "grad_norm": 0.9013834595680237, "learning_rate": 0.0005, "loss": 1.9631646871566772, "step": 6793 }, { "epoch": 0.7905515475913428, "grad_norm": 1.6703413724899292, "learning_rate": 0.0005, "loss": 2.5473473072052, "step": 6794 }, { "epoch": 0.7906679078426809, "grad_norm": 1.5532788038253784, "learning_rate": 0.0005, "loss": 1.8300776481628418, "step": 6795 }, { "epoch": 0.7907842680940191, "grad_norm": 1.1481420993804932, "learning_rate": 0.0005, "loss": 1.7941372394561768, "step": 6796 }, { "epoch": 0.7909006283453572, "grad_norm": 1.6368921995162964, "learning_rate": 0.0005, "loss": 2.2175588607788086, "step": 6797 }, { "epoch": 0.7910169885966953, "grad_norm": 1.6862839460372925, "learning_rate": 0.0005, "loss": 3.2936458587646484, "step": 6798 }, { "epoch": 0.7911333488480335, "grad_norm": 0.7672051191329956, "learning_rate": 0.0005, "loss": 1.8339869976043701, "step": 6799 }, { "epoch": 0.7912497090993716, "grad_norm": 1.1110107898712158, "learning_rate": 0.0005, "loss": 2.450544834136963, "step": 6800 }, { "epoch": 0.7913660693507097, "grad_norm": 1.4339540004730225, "learning_rate": 0.0005, "loss": 2.6280112266540527, "step": 6801 }, { "epoch": 0.791482429602048, "grad_norm": 1.1594489812850952, "learning_rate": 0.0005, "loss": 1.85927152633667, "step": 6802 }, { "epoch": 0.791598789853386, "grad_norm": 1.7650305032730103, "learning_rate": 0.0005, "loss": 3.2927021980285645, "step": 6803 }, { "epoch": 0.7917151501047243, "grad_norm": 1.6212831735610962, "learning_rate": 0.0005, "loss": 3.2759768962860107, "step": 6804 }, { "epoch": 0.7918315103560624, "grad_norm": 1.3943806886672974, "learning_rate": 0.0005, "loss": 1.9544754028320312, "step": 6805 }, { "epoch": 0.7919478706074005, "grad_norm": 1.971595048904419, "learning_rate": 0.0005, "loss": 2.4100236892700195, "step": 6806 }, { "epoch": 0.7920642308587387, "grad_norm": 1.5203782320022583, "learning_rate": 0.0005, "loss": 2.8139195442199707, "step": 6807 }, { "epoch": 0.7921805911100768, "grad_norm": 1.1461626291275024, "learning_rate": 0.0005, "loss": 2.544888973236084, "step": 6808 }, { "epoch": 0.792296951361415, "grad_norm": 2.070955276489258, "learning_rate": 0.0005, "loss": 3.052840232849121, "step": 6809 }, { "epoch": 0.7924133116127531, "grad_norm": 1.3840439319610596, "learning_rate": 0.0005, "loss": 2.1589112281799316, "step": 6810 }, { "epoch": 0.7925296718640912, "grad_norm": 1.5714340209960938, "learning_rate": 0.0005, "loss": 2.5838263034820557, "step": 6811 }, { "epoch": 0.7926460321154294, "grad_norm": 0.777932345867157, "learning_rate": 0.0005, "loss": 1.7630109786987305, "step": 6812 }, { "epoch": 0.7927623923667675, "grad_norm": 1.3230175971984863, "learning_rate": 0.0005, "loss": 1.9108343124389648, "step": 6813 }, { "epoch": 0.7928787526181057, "grad_norm": 0.7821425795555115, "learning_rate": 0.0005, "loss": 1.5496000051498413, "step": 6814 }, { "epoch": 0.7929951128694438, "grad_norm": 1.2139053344726562, "learning_rate": 0.0005, "loss": 2.0598628520965576, "step": 6815 }, { "epoch": 0.7931114731207819, "grad_norm": 1.2065454721450806, "learning_rate": 0.0005, "loss": 2.4997897148132324, "step": 6816 }, { "epoch": 0.7932278333721201, "grad_norm": 1.37099289894104, "learning_rate": 0.0005, "loss": 1.601628303527832, "step": 6817 }, { "epoch": 0.7933441936234582, "grad_norm": 0.35934194922447205, "learning_rate": 0.0005, "loss": 0.9131976962089539, "step": 6818 }, { "epoch": 0.7934605538747964, "grad_norm": 1.3134008646011353, "learning_rate": 0.0005, "loss": 2.8858087062835693, "step": 6819 }, { "epoch": 0.7935769141261345, "grad_norm": 1.2313716411590576, "learning_rate": 0.0005, "loss": 1.9310505390167236, "step": 6820 }, { "epoch": 0.7936932743774726, "grad_norm": 1.7222596406936646, "learning_rate": 0.0005, "loss": 2.984508991241455, "step": 6821 }, { "epoch": 0.7938096346288108, "grad_norm": 1.8410876989364624, "learning_rate": 0.0005, "loss": 3.112532138824463, "step": 6822 }, { "epoch": 0.7939259948801489, "grad_norm": 1.6647151708602905, "learning_rate": 0.0005, "loss": 2.179453134536743, "step": 6823 }, { "epoch": 0.7940423551314871, "grad_norm": 1.4634723663330078, "learning_rate": 0.0005, "loss": 2.5193796157836914, "step": 6824 }, { "epoch": 0.7941587153828252, "grad_norm": 1.253619909286499, "learning_rate": 0.0005, "loss": 2.831022262573242, "step": 6825 }, { "epoch": 0.7942750756341633, "grad_norm": 1.3788819313049316, "learning_rate": 0.0005, "loss": 2.8820314407348633, "step": 6826 }, { "epoch": 0.7943914358855015, "grad_norm": 2.8543972969055176, "learning_rate": 0.0005, "loss": 2.3960061073303223, "step": 6827 }, { "epoch": 0.7945077961368396, "grad_norm": 1.1340653896331787, "learning_rate": 0.0005, "loss": 2.2410435676574707, "step": 6828 }, { "epoch": 0.7946241563881778, "grad_norm": 2.1942782402038574, "learning_rate": 0.0005, "loss": 2.4908435344696045, "step": 6829 }, { "epoch": 0.794740516639516, "grad_norm": 1.1056764125823975, "learning_rate": 0.0005, "loss": 1.7558214664459229, "step": 6830 }, { "epoch": 0.794856876890854, "grad_norm": 1.3870463371276855, "learning_rate": 0.0005, "loss": 2.1475203037261963, "step": 6831 }, { "epoch": 0.7949732371421923, "grad_norm": 1.4608420133590698, "learning_rate": 0.0005, "loss": 2.0979461669921875, "step": 6832 }, { "epoch": 0.7950895973935304, "grad_norm": 1.6142181158065796, "learning_rate": 0.0005, "loss": 2.848756790161133, "step": 6833 }, { "epoch": 0.7952059576448686, "grad_norm": 1.5299683809280396, "learning_rate": 0.0005, "loss": 3.2351412773132324, "step": 6834 }, { "epoch": 0.7953223178962067, "grad_norm": 1.4849522113800049, "learning_rate": 0.0005, "loss": 1.9918603897094727, "step": 6835 }, { "epoch": 0.7954386781475448, "grad_norm": 1.3805937767028809, "learning_rate": 0.0005, "loss": 2.118666648864746, "step": 6836 }, { "epoch": 0.795555038398883, "grad_norm": 1.4525189399719238, "learning_rate": 0.0005, "loss": 2.280637264251709, "step": 6837 }, { "epoch": 0.7956713986502211, "grad_norm": 0.9657095074653625, "learning_rate": 0.0005, "loss": 2.428952217102051, "step": 6838 }, { "epoch": 0.7957877589015592, "grad_norm": 2.16355562210083, "learning_rate": 0.0005, "loss": 2.998279094696045, "step": 6839 }, { "epoch": 0.7959041191528974, "grad_norm": 1.3933964967727661, "learning_rate": 0.0005, "loss": 2.5693483352661133, "step": 6840 }, { "epoch": 0.7960204794042355, "grad_norm": 1.4262800216674805, "learning_rate": 0.0005, "loss": 2.117178440093994, "step": 6841 }, { "epoch": 0.7961368396555737, "grad_norm": 1.9724730253219604, "learning_rate": 0.0005, "loss": 2.604687213897705, "step": 6842 }, { "epoch": 0.7962531999069118, "grad_norm": 1.2019985914230347, "learning_rate": 0.0005, "loss": 1.656766414642334, "step": 6843 }, { "epoch": 0.7963695601582499, "grad_norm": 1.4682735204696655, "learning_rate": 0.0005, "loss": 2.3518309593200684, "step": 6844 }, { "epoch": 0.7964859204095881, "grad_norm": 1.7378884553909302, "learning_rate": 0.0005, "loss": 2.6870250701904297, "step": 6845 }, { "epoch": 0.7966022806609262, "grad_norm": 1.2282088994979858, "learning_rate": 0.0005, "loss": 2.087444305419922, "step": 6846 }, { "epoch": 0.7967186409122644, "grad_norm": 1.263033390045166, "learning_rate": 0.0005, "loss": 2.5904202461242676, "step": 6847 }, { "epoch": 0.7968350011636025, "grad_norm": 1.0466628074645996, "learning_rate": 0.0005, "loss": 1.4244993925094604, "step": 6848 }, { "epoch": 0.7969513614149406, "grad_norm": 0.9651027321815491, "learning_rate": 0.0005, "loss": 2.9514074325561523, "step": 6849 }, { "epoch": 0.7970677216662788, "grad_norm": 1.3985309600830078, "learning_rate": 0.0005, "loss": 2.9853200912475586, "step": 6850 }, { "epoch": 0.7971840819176169, "grad_norm": 1.5470738410949707, "learning_rate": 0.0005, "loss": 2.859579086303711, "step": 6851 }, { "epoch": 0.7973004421689551, "grad_norm": 0.506679356098175, "learning_rate": 0.0005, "loss": 1.1206305027008057, "step": 6852 }, { "epoch": 0.7974168024202932, "grad_norm": 1.0638351440429688, "learning_rate": 0.0005, "loss": 1.8459526300430298, "step": 6853 }, { "epoch": 0.7975331626716313, "grad_norm": 1.2139389514923096, "learning_rate": 0.0005, "loss": 2.1872496604919434, "step": 6854 }, { "epoch": 0.7976495229229695, "grad_norm": 1.4278264045715332, "learning_rate": 0.0005, "loss": 2.416912078857422, "step": 6855 }, { "epoch": 0.7977658831743076, "grad_norm": 0.9132272005081177, "learning_rate": 0.0005, "loss": 1.8230596780776978, "step": 6856 }, { "epoch": 0.7978822434256458, "grad_norm": 1.094401240348816, "learning_rate": 0.0005, "loss": 2.2579474449157715, "step": 6857 }, { "epoch": 0.7979986036769839, "grad_norm": 1.5321608781814575, "learning_rate": 0.0005, "loss": 2.9554128646850586, "step": 6858 }, { "epoch": 0.798114963928322, "grad_norm": 1.090861439704895, "learning_rate": 0.0005, "loss": 2.436234712600708, "step": 6859 }, { "epoch": 0.7982313241796603, "grad_norm": 0.9183866381645203, "learning_rate": 0.0005, "loss": 2.272815465927124, "step": 6860 }, { "epoch": 0.7983476844309984, "grad_norm": 1.3130574226379395, "learning_rate": 0.0005, "loss": 3.448904037475586, "step": 6861 }, { "epoch": 0.7984640446823366, "grad_norm": 2.001431941986084, "learning_rate": 0.0005, "loss": 2.180551052093506, "step": 6862 }, { "epoch": 0.7985804049336747, "grad_norm": 1.1273359060287476, "learning_rate": 0.0005, "loss": 2.5870985984802246, "step": 6863 }, { "epoch": 0.7986967651850128, "grad_norm": 1.4130092859268188, "learning_rate": 0.0005, "loss": 2.3858530521392822, "step": 6864 }, { "epoch": 0.798813125436351, "grad_norm": 1.3945364952087402, "learning_rate": 0.0005, "loss": 2.5340702533721924, "step": 6865 }, { "epoch": 0.7989294856876891, "grad_norm": 1.3951534032821655, "learning_rate": 0.0005, "loss": 2.60835337638855, "step": 6866 }, { "epoch": 0.7990458459390273, "grad_norm": 0.8409401774406433, "learning_rate": 0.0005, "loss": 1.9582555294036865, "step": 6867 }, { "epoch": 0.7991622061903654, "grad_norm": 1.4871562719345093, "learning_rate": 0.0005, "loss": 2.8834705352783203, "step": 6868 }, { "epoch": 0.7992785664417035, "grad_norm": 1.6257288455963135, "learning_rate": 0.0005, "loss": 2.727015495300293, "step": 6869 }, { "epoch": 0.7993949266930417, "grad_norm": 1.2827656269073486, "learning_rate": 0.0005, "loss": 1.5776078701019287, "step": 6870 }, { "epoch": 0.7995112869443798, "grad_norm": 1.9179037809371948, "learning_rate": 0.0005, "loss": 2.368312358856201, "step": 6871 }, { "epoch": 0.799627647195718, "grad_norm": 1.6516995429992676, "learning_rate": 0.0005, "loss": 2.4699015617370605, "step": 6872 }, { "epoch": 0.7997440074470561, "grad_norm": 4.074286460876465, "learning_rate": 0.0005, "loss": 1.7305855751037598, "step": 6873 }, { "epoch": 0.7998603676983942, "grad_norm": 1.6323548555374146, "learning_rate": 0.0005, "loss": 2.366365432739258, "step": 6874 }, { "epoch": 0.7999767279497324, "grad_norm": 1.5892506837844849, "learning_rate": 0.0005, "loss": 2.2026126384735107, "step": 6875 }, { "epoch": 0.8000930882010705, "grad_norm": 1.2756043672561646, "learning_rate": 0.0005, "loss": 1.8969991207122803, "step": 6876 }, { "epoch": 0.8002094484524087, "grad_norm": 1.859039306640625, "learning_rate": 0.0005, "loss": 3.1885898113250732, "step": 6877 }, { "epoch": 0.8003258087037468, "grad_norm": 1.1921340227127075, "learning_rate": 0.0005, "loss": 3.4727470874786377, "step": 6878 }, { "epoch": 0.8004421689550849, "grad_norm": 2.1105523109436035, "learning_rate": 0.0005, "loss": 2.825507402420044, "step": 6879 }, { "epoch": 0.8005585292064231, "grad_norm": 1.2696000337600708, "learning_rate": 0.0005, "loss": 2.4215362071990967, "step": 6880 }, { "epoch": 0.8006748894577612, "grad_norm": 2.045733690261841, "learning_rate": 0.0005, "loss": 2.6540098190307617, "step": 6881 }, { "epoch": 0.8007912497090993, "grad_norm": 2.433635711669922, "learning_rate": 0.0005, "loss": 2.9209399223327637, "step": 6882 }, { "epoch": 0.8009076099604375, "grad_norm": 1.5866488218307495, "learning_rate": 0.0005, "loss": 2.727099895477295, "step": 6883 }, { "epoch": 0.8010239702117756, "grad_norm": 1.46147620677948, "learning_rate": 0.0005, "loss": 2.5948479175567627, "step": 6884 }, { "epoch": 0.8011403304631138, "grad_norm": 1.0325003862380981, "learning_rate": 0.0005, "loss": 2.3963255882263184, "step": 6885 }, { "epoch": 0.8012566907144519, "grad_norm": 1.2473903894424438, "learning_rate": 0.0005, "loss": 2.037407398223877, "step": 6886 }, { "epoch": 0.80137305096579, "grad_norm": 1.2148414850234985, "learning_rate": 0.0005, "loss": 1.9509714841842651, "step": 6887 }, { "epoch": 0.8014894112171282, "grad_norm": 0.36242374777793884, "learning_rate": 0.0005, "loss": 0.9573647975921631, "step": 6888 }, { "epoch": 0.8016057714684663, "grad_norm": 1.4156004190444946, "learning_rate": 0.0005, "loss": 2.541994571685791, "step": 6889 }, { "epoch": 0.8017221317198046, "grad_norm": 1.6205782890319824, "learning_rate": 0.0005, "loss": 2.098543167114258, "step": 6890 }, { "epoch": 0.8018384919711427, "grad_norm": 5.234562873840332, "learning_rate": 0.0005, "loss": 2.421602487564087, "step": 6891 }, { "epoch": 0.8019548522224808, "grad_norm": 0.6807900667190552, "learning_rate": 0.0005, "loss": 1.4689310789108276, "step": 6892 }, { "epoch": 0.802071212473819, "grad_norm": 0.8421406149864197, "learning_rate": 0.0005, "loss": 1.178560733795166, "step": 6893 }, { "epoch": 0.8021875727251571, "grad_norm": 1.4846322536468506, "learning_rate": 0.0005, "loss": 2.2985894680023193, "step": 6894 }, { "epoch": 0.8023039329764953, "grad_norm": 1.684720754623413, "learning_rate": 0.0005, "loss": 2.689749240875244, "step": 6895 }, { "epoch": 0.8024202932278334, "grad_norm": 1.421723484992981, "learning_rate": 0.0005, "loss": 2.6838881969451904, "step": 6896 }, { "epoch": 0.8025366534791715, "grad_norm": 0.9393020868301392, "learning_rate": 0.0005, "loss": 2.0667314529418945, "step": 6897 }, { "epoch": 0.8026530137305097, "grad_norm": 1.3079415559768677, "learning_rate": 0.0005, "loss": 2.141230583190918, "step": 6898 }, { "epoch": 0.8027693739818478, "grad_norm": 0.6420184969902039, "learning_rate": 0.0005, "loss": 1.0829408168792725, "step": 6899 }, { "epoch": 0.802885734233186, "grad_norm": 2.7368552684783936, "learning_rate": 0.0005, "loss": 2.314253807067871, "step": 6900 }, { "epoch": 0.8030020944845241, "grad_norm": 2.003553628921509, "learning_rate": 0.0005, "loss": 3.0612552165985107, "step": 6901 }, { "epoch": 0.8031184547358622, "grad_norm": 1.6678414344787598, "learning_rate": 0.0005, "loss": 2.7044568061828613, "step": 6902 }, { "epoch": 0.8032348149872004, "grad_norm": 1.163870930671692, "learning_rate": 0.0005, "loss": 2.559142589569092, "step": 6903 }, { "epoch": 0.8033511752385385, "grad_norm": 1.2052510976791382, "learning_rate": 0.0005, "loss": 2.2835476398468018, "step": 6904 }, { "epoch": 0.8034675354898767, "grad_norm": 2.060441017150879, "learning_rate": 0.0005, "loss": 2.811776638031006, "step": 6905 }, { "epoch": 0.8035838957412148, "grad_norm": 1.5533373355865479, "learning_rate": 0.0005, "loss": 2.0443365573883057, "step": 6906 }, { "epoch": 0.8037002559925529, "grad_norm": 0.8943369388580322, "learning_rate": 0.0005, "loss": 1.6311694383621216, "step": 6907 }, { "epoch": 0.8038166162438911, "grad_norm": 0.7655029296875, "learning_rate": 0.0005, "loss": 1.7803150415420532, "step": 6908 }, { "epoch": 0.8039329764952292, "grad_norm": 1.0836045742034912, "learning_rate": 0.0005, "loss": 1.7522704601287842, "step": 6909 }, { "epoch": 0.8040493367465674, "grad_norm": 1.063887596130371, "learning_rate": 0.0005, "loss": 2.180950164794922, "step": 6910 }, { "epoch": 0.8041656969979055, "grad_norm": 1.1044062376022339, "learning_rate": 0.0005, "loss": 1.6311655044555664, "step": 6911 }, { "epoch": 0.8042820572492436, "grad_norm": 1.7446521520614624, "learning_rate": 0.0005, "loss": 2.8029980659484863, "step": 6912 }, { "epoch": 0.8043984175005818, "grad_norm": 1.6022124290466309, "learning_rate": 0.0005, "loss": 2.694348096847534, "step": 6913 }, { "epoch": 0.8045147777519199, "grad_norm": 1.1252559423446655, "learning_rate": 0.0005, "loss": 1.3472739458084106, "step": 6914 }, { "epoch": 0.8046311380032581, "grad_norm": 1.54173743724823, "learning_rate": 0.0005, "loss": 1.9308900833129883, "step": 6915 }, { "epoch": 0.8047474982545962, "grad_norm": 1.3176937103271484, "learning_rate": 0.0005, "loss": 1.9920525550842285, "step": 6916 }, { "epoch": 0.8048638585059343, "grad_norm": 1.1235672235488892, "learning_rate": 0.0005, "loss": 1.535516381263733, "step": 6917 }, { "epoch": 0.8049802187572725, "grad_norm": 1.9744664430618286, "learning_rate": 0.0005, "loss": 2.6237378120422363, "step": 6918 }, { "epoch": 0.8050965790086106, "grad_norm": 1.6218153238296509, "learning_rate": 0.0005, "loss": 1.9922819137573242, "step": 6919 }, { "epoch": 0.8052129392599489, "grad_norm": 1.3716628551483154, "learning_rate": 0.0005, "loss": 2.568366527557373, "step": 6920 }, { "epoch": 0.805329299511287, "grad_norm": 1.0602279901504517, "learning_rate": 0.0005, "loss": 1.7749075889587402, "step": 6921 }, { "epoch": 0.805445659762625, "grad_norm": 1.1760283708572388, "learning_rate": 0.0005, "loss": 2.0434751510620117, "step": 6922 }, { "epoch": 0.8055620200139633, "grad_norm": 1.7635252475738525, "learning_rate": 0.0005, "loss": 2.3528308868408203, "step": 6923 }, { "epoch": 0.8056783802653014, "grad_norm": 1.238690733909607, "learning_rate": 0.0005, "loss": 2.052236795425415, "step": 6924 }, { "epoch": 0.8057947405166395, "grad_norm": 1.0034528970718384, "learning_rate": 0.0005, "loss": 3.1093173027038574, "step": 6925 }, { "epoch": 0.8059111007679777, "grad_norm": 1.3216723203659058, "learning_rate": 0.0005, "loss": 1.360349178314209, "step": 6926 }, { "epoch": 0.8060274610193158, "grad_norm": 1.7245230674743652, "learning_rate": 0.0005, "loss": 2.018603801727295, "step": 6927 }, { "epoch": 0.806143821270654, "grad_norm": 0.990485429763794, "learning_rate": 0.0005, "loss": 2.5202112197875977, "step": 6928 }, { "epoch": 0.8062601815219921, "grad_norm": 1.5502524375915527, "learning_rate": 0.0005, "loss": 2.29268217086792, "step": 6929 }, { "epoch": 0.8063765417733302, "grad_norm": 2.319065809249878, "learning_rate": 0.0005, "loss": 2.7171072959899902, "step": 6930 }, { "epoch": 0.8064929020246684, "grad_norm": 1.068354606628418, "learning_rate": 0.0005, "loss": 1.9482839107513428, "step": 6931 }, { "epoch": 0.8066092622760065, "grad_norm": 1.1434650421142578, "learning_rate": 0.0005, "loss": 1.7774782180786133, "step": 6932 }, { "epoch": 0.8067256225273447, "grad_norm": 1.0849964618682861, "learning_rate": 0.0005, "loss": 2.2439422607421875, "step": 6933 }, { "epoch": 0.8068419827786828, "grad_norm": 1.305965542793274, "learning_rate": 0.0005, "loss": 1.6113544702529907, "step": 6934 }, { "epoch": 0.8069583430300209, "grad_norm": 1.5992611646652222, "learning_rate": 0.0005, "loss": 2.110424757003784, "step": 6935 }, { "epoch": 0.8070747032813591, "grad_norm": 1.384178638458252, "learning_rate": 0.0005, "loss": 2.63877534866333, "step": 6936 }, { "epoch": 0.8071910635326972, "grad_norm": 2.117448091506958, "learning_rate": 0.0005, "loss": 2.0681354999542236, "step": 6937 }, { "epoch": 0.8073074237840354, "grad_norm": 0.6779744029045105, "learning_rate": 0.0005, "loss": 1.8995248079299927, "step": 6938 }, { "epoch": 0.8074237840353735, "grad_norm": 0.8723366856575012, "learning_rate": 0.0005, "loss": 1.8367375135421753, "step": 6939 }, { "epoch": 0.8075401442867116, "grad_norm": 1.7168492078781128, "learning_rate": 0.0005, "loss": 2.545048236846924, "step": 6940 }, { "epoch": 0.8076565045380498, "grad_norm": 0.5985177159309387, "learning_rate": 0.0005, "loss": 1.403778076171875, "step": 6941 }, { "epoch": 0.8077728647893879, "grad_norm": 1.39649498462677, "learning_rate": 0.0005, "loss": 1.7772021293640137, "step": 6942 }, { "epoch": 0.8078892250407261, "grad_norm": 1.84030020236969, "learning_rate": 0.0005, "loss": 3.2715890407562256, "step": 6943 }, { "epoch": 0.8080055852920642, "grad_norm": 0.9733898639678955, "learning_rate": 0.0005, "loss": 2.0819993019104004, "step": 6944 }, { "epoch": 0.8081219455434023, "grad_norm": 1.5225170850753784, "learning_rate": 0.0005, "loss": 2.158815383911133, "step": 6945 }, { "epoch": 0.8082383057947405, "grad_norm": 1.1779961585998535, "learning_rate": 0.0005, "loss": 1.2916291952133179, "step": 6946 }, { "epoch": 0.8083546660460786, "grad_norm": 1.005102276802063, "learning_rate": 0.0005, "loss": 1.96743905544281, "step": 6947 }, { "epoch": 0.8084710262974169, "grad_norm": 1.2971954345703125, "learning_rate": 0.0005, "loss": 1.9108307361602783, "step": 6948 }, { "epoch": 0.808587386548755, "grad_norm": 1.0782716274261475, "learning_rate": 0.0005, "loss": 2.9787673950195312, "step": 6949 }, { "epoch": 0.808703746800093, "grad_norm": 1.4376734495162964, "learning_rate": 0.0005, "loss": 2.019914150238037, "step": 6950 }, { "epoch": 0.8088201070514313, "grad_norm": 1.3333897590637207, "learning_rate": 0.0005, "loss": 2.1854000091552734, "step": 6951 }, { "epoch": 0.8089364673027694, "grad_norm": 0.9109049439430237, "learning_rate": 0.0005, "loss": 1.7734962701797485, "step": 6952 }, { "epoch": 0.8090528275541076, "grad_norm": 1.0315349102020264, "learning_rate": 0.0005, "loss": 2.604729175567627, "step": 6953 }, { "epoch": 0.8091691878054457, "grad_norm": 0.6656447649002075, "learning_rate": 0.0005, "loss": 1.9073458909988403, "step": 6954 }, { "epoch": 0.8092855480567838, "grad_norm": 1.522691249847412, "learning_rate": 0.0005, "loss": 2.071748971939087, "step": 6955 }, { "epoch": 0.809401908308122, "grad_norm": 1.4103491306304932, "learning_rate": 0.0005, "loss": 2.755079746246338, "step": 6956 }, { "epoch": 0.8095182685594601, "grad_norm": 1.096112847328186, "learning_rate": 0.0005, "loss": 1.8440394401550293, "step": 6957 }, { "epoch": 0.8096346288107983, "grad_norm": 1.5992958545684814, "learning_rate": 0.0005, "loss": 2.984157085418701, "step": 6958 }, { "epoch": 0.8097509890621364, "grad_norm": 1.2641409635543823, "learning_rate": 0.0005, "loss": 2.620591163635254, "step": 6959 }, { "epoch": 0.8098673493134745, "grad_norm": 4.9728288650512695, "learning_rate": 0.0005, "loss": 3.0946316719055176, "step": 6960 }, { "epoch": 0.8099837095648127, "grad_norm": 0.8341188430786133, "learning_rate": 0.0005, "loss": 1.6963927745819092, "step": 6961 }, { "epoch": 0.8101000698161508, "grad_norm": 1.9163998365402222, "learning_rate": 0.0005, "loss": 2.010127067565918, "step": 6962 }, { "epoch": 0.8102164300674889, "grad_norm": 2.563565254211426, "learning_rate": 0.0005, "loss": 2.872481346130371, "step": 6963 }, { "epoch": 0.8103327903188271, "grad_norm": 0.6125648021697998, "learning_rate": 0.0005, "loss": 1.647904634475708, "step": 6964 }, { "epoch": 0.8104491505701652, "grad_norm": 1.6658415794372559, "learning_rate": 0.0005, "loss": 2.244547128677368, "step": 6965 }, { "epoch": 0.8105655108215034, "grad_norm": 3.7453103065490723, "learning_rate": 0.0005, "loss": 2.7662763595581055, "step": 6966 }, { "epoch": 0.8106818710728415, "grad_norm": 1.28369140625, "learning_rate": 0.0005, "loss": 2.2921957969665527, "step": 6967 }, { "epoch": 0.8107982313241796, "grad_norm": 1.5897258520126343, "learning_rate": 0.0005, "loss": 2.444948673248291, "step": 6968 }, { "epoch": 0.8109145915755178, "grad_norm": 1.2277436256408691, "learning_rate": 0.0005, "loss": 1.6680281162261963, "step": 6969 }, { "epoch": 0.8110309518268559, "grad_norm": 1.472629427909851, "learning_rate": 0.0005, "loss": 1.5699267387390137, "step": 6970 }, { "epoch": 0.8111473120781941, "grad_norm": 1.2638862133026123, "learning_rate": 0.0005, "loss": 2.2764718532562256, "step": 6971 }, { "epoch": 0.8112636723295322, "grad_norm": 0.5734549760818481, "learning_rate": 0.0005, "loss": 1.6351791620254517, "step": 6972 }, { "epoch": 0.8113800325808703, "grad_norm": 1.7361775636672974, "learning_rate": 0.0005, "loss": 3.113187313079834, "step": 6973 }, { "epoch": 0.8114963928322085, "grad_norm": 1.9521801471710205, "learning_rate": 0.0005, "loss": 3.3017213344573975, "step": 6974 }, { "epoch": 0.8116127530835466, "grad_norm": 1.3718149662017822, "learning_rate": 0.0005, "loss": 1.9087040424346924, "step": 6975 }, { "epoch": 0.8117291133348848, "grad_norm": 2.220245361328125, "learning_rate": 0.0005, "loss": 3.3480873107910156, "step": 6976 }, { "epoch": 0.8118454735862229, "grad_norm": 0.7940687537193298, "learning_rate": 0.0005, "loss": 1.1553682088851929, "step": 6977 }, { "epoch": 0.811961833837561, "grad_norm": 1.4117863178253174, "learning_rate": 0.0005, "loss": 2.1862633228302, "step": 6978 }, { "epoch": 0.8120781940888993, "grad_norm": 0.6740559339523315, "learning_rate": 0.0005, "loss": 1.6007651090621948, "step": 6979 }, { "epoch": 0.8121945543402374, "grad_norm": 1.3237190246582031, "learning_rate": 0.0005, "loss": 2.3904953002929688, "step": 6980 }, { "epoch": 0.8123109145915756, "grad_norm": 1.212873935699463, "learning_rate": 0.0005, "loss": 1.4462271928787231, "step": 6981 }, { "epoch": 0.8124272748429137, "grad_norm": 1.3186661005020142, "learning_rate": 0.0005, "loss": 2.4469540119171143, "step": 6982 }, { "epoch": 0.8125436350942518, "grad_norm": 1.5590521097183228, "learning_rate": 0.0005, "loss": 3.1055569648742676, "step": 6983 }, { "epoch": 0.81265999534559, "grad_norm": 1.292542576789856, "learning_rate": 0.0005, "loss": 2.277764320373535, "step": 6984 }, { "epoch": 0.8127763555969281, "grad_norm": 2.1060516834259033, "learning_rate": 0.0005, "loss": 2.823176383972168, "step": 6985 }, { "epoch": 0.8128927158482663, "grad_norm": 1.7483909130096436, "learning_rate": 0.0005, "loss": 2.752882957458496, "step": 6986 }, { "epoch": 0.8130090760996044, "grad_norm": 1.188499927520752, "learning_rate": 0.0005, "loss": 2.1238088607788086, "step": 6987 }, { "epoch": 0.8131254363509425, "grad_norm": 1.3119730949401855, "learning_rate": 0.0005, "loss": 2.6661572456359863, "step": 6988 }, { "epoch": 0.8132417966022807, "grad_norm": 0.9360121488571167, "learning_rate": 0.0005, "loss": 2.373720407485962, "step": 6989 }, { "epoch": 0.8133581568536188, "grad_norm": 1.0826630592346191, "learning_rate": 0.0005, "loss": 1.5190212726593018, "step": 6990 }, { "epoch": 0.813474517104957, "grad_norm": 1.5421303510665894, "learning_rate": 0.0005, "loss": 2.3752193450927734, "step": 6991 }, { "epoch": 0.8135908773562951, "grad_norm": 1.8102037906646729, "learning_rate": 0.0005, "loss": 2.4120302200317383, "step": 6992 }, { "epoch": 0.8137072376076332, "grad_norm": 1.0433604717254639, "learning_rate": 0.0005, "loss": 2.2721171379089355, "step": 6993 }, { "epoch": 0.8138235978589714, "grad_norm": 1.5146124362945557, "learning_rate": 0.0005, "loss": 2.1530909538269043, "step": 6994 }, { "epoch": 0.8139399581103095, "grad_norm": 1.8597546815872192, "learning_rate": 0.0005, "loss": 3.0996346473693848, "step": 6995 }, { "epoch": 0.8140563183616477, "grad_norm": 1.5634381771087646, "learning_rate": 0.0005, "loss": 2.2338809967041016, "step": 6996 }, { "epoch": 0.8141726786129858, "grad_norm": 2.035428285598755, "learning_rate": 0.0005, "loss": 3.0881967544555664, "step": 6997 }, { "epoch": 0.8142890388643239, "grad_norm": 1.4894263744354248, "learning_rate": 0.0005, "loss": 2.4698596000671387, "step": 6998 }, { "epoch": 0.8144053991156621, "grad_norm": 0.5923909544944763, "learning_rate": 0.0005, "loss": 1.994606375694275, "step": 6999 }, { "epoch": 0.8145217593670002, "grad_norm": 1.7427936792373657, "learning_rate": 0.0005, "loss": 3.0963473320007324, "step": 7000 }, { "epoch": 0.8146381196183384, "grad_norm": 1.3740311861038208, "learning_rate": 0.0005, "loss": 3.2217354774475098, "step": 7001 }, { "epoch": 0.8147544798696765, "grad_norm": 1.0545254945755005, "learning_rate": 0.0005, "loss": 2.275407314300537, "step": 7002 }, { "epoch": 0.8148708401210146, "grad_norm": 1.188371181488037, "learning_rate": 0.0005, "loss": 3.5698728561401367, "step": 7003 }, { "epoch": 0.8149872003723528, "grad_norm": 3.434650421142578, "learning_rate": 0.0005, "loss": 2.3290469646453857, "step": 7004 }, { "epoch": 0.8151035606236909, "grad_norm": 1.1625018119812012, "learning_rate": 0.0005, "loss": 1.7425575256347656, "step": 7005 }, { "epoch": 0.815219920875029, "grad_norm": 1.4033159017562866, "learning_rate": 0.0005, "loss": 2.709113121032715, "step": 7006 }, { "epoch": 0.8153362811263672, "grad_norm": 1.4492976665496826, "learning_rate": 0.0005, "loss": 2.6421773433685303, "step": 7007 }, { "epoch": 0.8154526413777053, "grad_norm": 2.0155868530273438, "learning_rate": 0.0005, "loss": 2.159426689147949, "step": 7008 }, { "epoch": 0.8155690016290436, "grad_norm": 1.6672320365905762, "learning_rate": 0.0005, "loss": 1.7721070051193237, "step": 7009 }, { "epoch": 0.8156853618803817, "grad_norm": 3.466909408569336, "learning_rate": 0.0005, "loss": 2.556586980819702, "step": 7010 }, { "epoch": 0.8158017221317198, "grad_norm": 1.8316643238067627, "learning_rate": 0.0005, "loss": 2.6085479259490967, "step": 7011 }, { "epoch": 0.815918082383058, "grad_norm": 0.7895799875259399, "learning_rate": 0.0005, "loss": 2.02968168258667, "step": 7012 }, { "epoch": 0.8160344426343961, "grad_norm": 1.2805049419403076, "learning_rate": 0.0005, "loss": 2.6691970825195312, "step": 7013 }, { "epoch": 0.8161508028857343, "grad_norm": 3.2894551753997803, "learning_rate": 0.0005, "loss": 3.299450397491455, "step": 7014 }, { "epoch": 0.8162671631370724, "grad_norm": 1.4889250993728638, "learning_rate": 0.0005, "loss": 2.575328826904297, "step": 7015 }, { "epoch": 0.8163835233884105, "grad_norm": 2.2189533710479736, "learning_rate": 0.0005, "loss": 2.5562055110931396, "step": 7016 }, { "epoch": 0.8164998836397487, "grad_norm": 1.7351969480514526, "learning_rate": 0.0005, "loss": 2.702991485595703, "step": 7017 }, { "epoch": 0.8166162438910868, "grad_norm": 1.289649486541748, "learning_rate": 0.0005, "loss": 2.4116365909576416, "step": 7018 }, { "epoch": 0.816732604142425, "grad_norm": 1.262345552444458, "learning_rate": 0.0005, "loss": 1.8672337532043457, "step": 7019 }, { "epoch": 0.8168489643937631, "grad_norm": 1.4697672128677368, "learning_rate": 0.0005, "loss": 1.875687599182129, "step": 7020 }, { "epoch": 0.8169653246451012, "grad_norm": 1.2107690572738647, "learning_rate": 0.0005, "loss": 1.751511812210083, "step": 7021 }, { "epoch": 0.8170816848964394, "grad_norm": 1.228724479675293, "learning_rate": 0.0005, "loss": 2.380002975463867, "step": 7022 }, { "epoch": 0.8171980451477775, "grad_norm": 1.4898037910461426, "learning_rate": 0.0005, "loss": 2.574831485748291, "step": 7023 }, { "epoch": 0.8173144053991157, "grad_norm": 0.8144104480743408, "learning_rate": 0.0005, "loss": 1.1597893238067627, "step": 7024 }, { "epoch": 0.8174307656504538, "grad_norm": 2.5799765586853027, "learning_rate": 0.0005, "loss": 2.5085129737854004, "step": 7025 }, { "epoch": 0.8175471259017919, "grad_norm": 1.6092332601547241, "learning_rate": 0.0005, "loss": 2.2301034927368164, "step": 7026 }, { "epoch": 0.8176634861531301, "grad_norm": 1.4061518907546997, "learning_rate": 0.0005, "loss": 1.8759080171585083, "step": 7027 }, { "epoch": 0.8177798464044682, "grad_norm": 1.0438076257705688, "learning_rate": 0.0005, "loss": 2.132586717605591, "step": 7028 }, { "epoch": 0.8178962066558064, "grad_norm": 1.550542950630188, "learning_rate": 0.0005, "loss": 2.700739622116089, "step": 7029 }, { "epoch": 0.8180125669071445, "grad_norm": 1.404826283454895, "learning_rate": 0.0005, "loss": 3.4124677181243896, "step": 7030 }, { "epoch": 0.8181289271584826, "grad_norm": 0.9342209100723267, "learning_rate": 0.0005, "loss": 1.9187872409820557, "step": 7031 }, { "epoch": 0.8182452874098208, "grad_norm": 1.7036761045455933, "learning_rate": 0.0005, "loss": 2.6868410110473633, "step": 7032 }, { "epoch": 0.8183616476611589, "grad_norm": 1.9473196268081665, "learning_rate": 0.0005, "loss": 3.1449296474456787, "step": 7033 }, { "epoch": 0.8184780079124971, "grad_norm": 1.3733934164047241, "learning_rate": 0.0005, "loss": 2.849956750869751, "step": 7034 }, { "epoch": 0.8185943681638352, "grad_norm": 0.8277425765991211, "learning_rate": 0.0005, "loss": 1.8878910541534424, "step": 7035 }, { "epoch": 0.8187107284151733, "grad_norm": 1.4776737689971924, "learning_rate": 0.0005, "loss": 2.348966121673584, "step": 7036 }, { "epoch": 0.8188270886665115, "grad_norm": 1.0911648273468018, "learning_rate": 0.0005, "loss": 2.650846481323242, "step": 7037 }, { "epoch": 0.8189434489178496, "grad_norm": 0.45215967297554016, "learning_rate": 0.0005, "loss": 1.0214262008666992, "step": 7038 }, { "epoch": 0.8190598091691879, "grad_norm": 1.5630888938903809, "learning_rate": 0.0005, "loss": 2.1858766078948975, "step": 7039 }, { "epoch": 0.819176169420526, "grad_norm": 1.2356282472610474, "learning_rate": 0.0005, "loss": 2.5209410190582275, "step": 7040 }, { "epoch": 0.8192925296718641, "grad_norm": 1.3399187326431274, "learning_rate": 0.0005, "loss": 2.3136892318725586, "step": 7041 }, { "epoch": 0.8194088899232023, "grad_norm": 1.1317670345306396, "learning_rate": 0.0005, "loss": 2.5372848510742188, "step": 7042 }, { "epoch": 0.8195252501745404, "grad_norm": 0.887861967086792, "learning_rate": 0.0005, "loss": 2.225745677947998, "step": 7043 }, { "epoch": 0.8196416104258786, "grad_norm": 1.125356674194336, "learning_rate": 0.0005, "loss": 2.784539222717285, "step": 7044 }, { "epoch": 0.8197579706772167, "grad_norm": 1.2585556507110596, "learning_rate": 0.0005, "loss": 2.18384051322937, "step": 7045 }, { "epoch": 0.8198743309285548, "grad_norm": 0.7201582789421082, "learning_rate": 0.0005, "loss": 1.6372740268707275, "step": 7046 }, { "epoch": 0.819990691179893, "grad_norm": 1.3644202947616577, "learning_rate": 0.0005, "loss": 2.2509241104125977, "step": 7047 }, { "epoch": 0.8201070514312311, "grad_norm": 1.7804518938064575, "learning_rate": 0.0005, "loss": 3.20373797416687, "step": 7048 }, { "epoch": 0.8202234116825692, "grad_norm": 1.7661731243133545, "learning_rate": 0.0005, "loss": 2.8886256217956543, "step": 7049 }, { "epoch": 0.8203397719339074, "grad_norm": 0.9143977165222168, "learning_rate": 0.0005, "loss": 1.9833920001983643, "step": 7050 }, { "epoch": 0.8204561321852455, "grad_norm": 1.4660269021987915, "learning_rate": 0.0005, "loss": 2.6047563552856445, "step": 7051 }, { "epoch": 0.8205724924365837, "grad_norm": 1.5415560007095337, "learning_rate": 0.0005, "loss": 3.0353846549987793, "step": 7052 }, { "epoch": 0.8206888526879218, "grad_norm": 0.5620847344398499, "learning_rate": 0.0005, "loss": 1.5593979358673096, "step": 7053 }, { "epoch": 0.8208052129392599, "grad_norm": 1.8045378923416138, "learning_rate": 0.0005, "loss": 3.0278449058532715, "step": 7054 }, { "epoch": 0.8209215731905981, "grad_norm": 1.6359893083572388, "learning_rate": 0.0005, "loss": 3.37131929397583, "step": 7055 }, { "epoch": 0.8210379334419362, "grad_norm": 1.5106090307235718, "learning_rate": 0.0005, "loss": 2.165022373199463, "step": 7056 }, { "epoch": 0.8211542936932744, "grad_norm": 1.2309035062789917, "learning_rate": 0.0005, "loss": 2.8625199794769287, "step": 7057 }, { "epoch": 0.8212706539446125, "grad_norm": 1.532605767250061, "learning_rate": 0.0005, "loss": 1.8649730682373047, "step": 7058 }, { "epoch": 0.8213870141959506, "grad_norm": 1.5131617784500122, "learning_rate": 0.0005, "loss": 2.041109800338745, "step": 7059 }, { "epoch": 0.8215033744472888, "grad_norm": 1.3169225454330444, "learning_rate": 0.0005, "loss": 2.889509677886963, "step": 7060 }, { "epoch": 0.8216197346986269, "grad_norm": 1.0212024450302124, "learning_rate": 0.0005, "loss": 1.6333427429199219, "step": 7061 }, { "epoch": 0.8217360949499651, "grad_norm": 1.1943660974502563, "learning_rate": 0.0005, "loss": 2.38461971282959, "step": 7062 }, { "epoch": 0.8218524552013032, "grad_norm": 2.017672061920166, "learning_rate": 0.0005, "loss": 2.89083194732666, "step": 7063 }, { "epoch": 0.8219688154526413, "grad_norm": 1.3822003602981567, "learning_rate": 0.0005, "loss": 2.239253044128418, "step": 7064 }, { "epoch": 0.8220851757039795, "grad_norm": 1.2223752737045288, "learning_rate": 0.0005, "loss": 1.878115177154541, "step": 7065 }, { "epoch": 0.8222015359553176, "grad_norm": 1.2241442203521729, "learning_rate": 0.0005, "loss": 2.795959949493408, "step": 7066 }, { "epoch": 0.8223178962066559, "grad_norm": 1.8860759735107422, "learning_rate": 0.0005, "loss": 2.549145221710205, "step": 7067 }, { "epoch": 0.822434256457994, "grad_norm": 1.7202826738357544, "learning_rate": 0.0005, "loss": 2.7335166931152344, "step": 7068 }, { "epoch": 0.822550616709332, "grad_norm": 1.445870280265808, "learning_rate": 0.0005, "loss": 2.977536201477051, "step": 7069 }, { "epoch": 0.8226669769606703, "grad_norm": 1.2936115264892578, "learning_rate": 0.0005, "loss": 3.1017887592315674, "step": 7070 }, { "epoch": 0.8227833372120084, "grad_norm": 1.0506242513656616, "learning_rate": 0.0005, "loss": 1.6827929019927979, "step": 7071 }, { "epoch": 0.8228996974633466, "grad_norm": 1.3519690036773682, "learning_rate": 0.0005, "loss": 2.7163453102111816, "step": 7072 }, { "epoch": 0.8230160577146847, "grad_norm": 0.62905353307724, "learning_rate": 0.0005, "loss": 1.8328912258148193, "step": 7073 }, { "epoch": 0.8231324179660228, "grad_norm": 2.722548007965088, "learning_rate": 0.0005, "loss": 2.7023301124572754, "step": 7074 }, { "epoch": 0.823248778217361, "grad_norm": 1.0718868970870972, "learning_rate": 0.0005, "loss": 1.8084019422531128, "step": 7075 }, { "epoch": 0.8233651384686991, "grad_norm": 0.8307222723960876, "learning_rate": 0.0005, "loss": 2.0315985679626465, "step": 7076 }, { "epoch": 0.8234814987200373, "grad_norm": 0.7362029552459717, "learning_rate": 0.0005, "loss": 1.264190912246704, "step": 7077 }, { "epoch": 0.8235978589713754, "grad_norm": 2.9280598163604736, "learning_rate": 0.0005, "loss": 2.4371283054351807, "step": 7078 }, { "epoch": 0.8237142192227135, "grad_norm": 0.8848613500595093, "learning_rate": 0.0005, "loss": 2.219860076904297, "step": 7079 }, { "epoch": 0.8238305794740517, "grad_norm": 1.0778688192367554, "learning_rate": 0.0005, "loss": 1.9723587036132812, "step": 7080 }, { "epoch": 0.8239469397253898, "grad_norm": 0.9263066053390503, "learning_rate": 0.0005, "loss": 2.1052205562591553, "step": 7081 }, { "epoch": 0.824063299976728, "grad_norm": 1.295380711555481, "learning_rate": 0.0005, "loss": 2.0978918075561523, "step": 7082 }, { "epoch": 0.8241796602280661, "grad_norm": 1.2636300325393677, "learning_rate": 0.0005, "loss": 2.337963342666626, "step": 7083 }, { "epoch": 0.8242960204794042, "grad_norm": 0.8852530121803284, "learning_rate": 0.0005, "loss": 1.738633394241333, "step": 7084 }, { "epoch": 0.8244123807307424, "grad_norm": 1.1593632698059082, "learning_rate": 0.0005, "loss": 1.9787418842315674, "step": 7085 }, { "epoch": 0.8245287409820805, "grad_norm": 1.439606785774231, "learning_rate": 0.0005, "loss": 2.6632204055786133, "step": 7086 }, { "epoch": 0.8246451012334186, "grad_norm": 1.1051346063613892, "learning_rate": 0.0005, "loss": 2.307553768157959, "step": 7087 }, { "epoch": 0.8247614614847568, "grad_norm": 0.6484727263450623, "learning_rate": 0.0005, "loss": 1.5950316190719604, "step": 7088 }, { "epoch": 0.8248778217360949, "grad_norm": 0.8146952390670776, "learning_rate": 0.0005, "loss": 2.198065757751465, "step": 7089 }, { "epoch": 0.8249941819874331, "grad_norm": 1.3430312871932983, "learning_rate": 0.0005, "loss": 2.199305295944214, "step": 7090 }, { "epoch": 0.8251105422387712, "grad_norm": 1.1330604553222656, "learning_rate": 0.0005, "loss": 2.963200092315674, "step": 7091 }, { "epoch": 0.8252269024901093, "grad_norm": 1.721061110496521, "learning_rate": 0.0005, "loss": 3.117523670196533, "step": 7092 }, { "epoch": 0.8253432627414475, "grad_norm": 1.0777792930603027, "learning_rate": 0.0005, "loss": 2.1244029998779297, "step": 7093 }, { "epoch": 0.8254596229927856, "grad_norm": 1.050182580947876, "learning_rate": 0.0005, "loss": 2.8099849224090576, "step": 7094 }, { "epoch": 0.8255759832441238, "grad_norm": 1.2041364908218384, "learning_rate": 0.0005, "loss": 2.286886215209961, "step": 7095 }, { "epoch": 0.825692343495462, "grad_norm": 1.6505794525146484, "learning_rate": 0.0005, "loss": 2.1097967624664307, "step": 7096 }, { "epoch": 0.8258087037468, "grad_norm": 1.047904372215271, "learning_rate": 0.0005, "loss": 2.6061036586761475, "step": 7097 }, { "epoch": 0.8259250639981383, "grad_norm": 1.0773142576217651, "learning_rate": 0.0005, "loss": 1.9442157745361328, "step": 7098 }, { "epoch": 0.8260414242494764, "grad_norm": 1.3587507009506226, "learning_rate": 0.0005, "loss": 1.6533188819885254, "step": 7099 }, { "epoch": 0.8261577845008146, "grad_norm": 0.9235113859176636, "learning_rate": 0.0005, "loss": 1.248069167137146, "step": 7100 }, { "epoch": 0.8262741447521527, "grad_norm": 0.9985796213150024, "learning_rate": 0.0005, "loss": 1.9132680892944336, "step": 7101 }, { "epoch": 0.8263905050034908, "grad_norm": 1.0674326419830322, "learning_rate": 0.0005, "loss": 2.2091352939605713, "step": 7102 }, { "epoch": 0.826506865254829, "grad_norm": 1.3332091569900513, "learning_rate": 0.0005, "loss": 1.700578212738037, "step": 7103 }, { "epoch": 0.8266232255061671, "grad_norm": 1.179943323135376, "learning_rate": 0.0005, "loss": 3.0885260105133057, "step": 7104 }, { "epoch": 0.8267395857575053, "grad_norm": 1.7343922853469849, "learning_rate": 0.0005, "loss": 3.1497995853424072, "step": 7105 }, { "epoch": 0.8268559460088434, "grad_norm": 1.6929913759231567, "learning_rate": 0.0005, "loss": 2.7159032821655273, "step": 7106 }, { "epoch": 0.8269723062601815, "grad_norm": 1.4409523010253906, "learning_rate": 0.0005, "loss": 2.5230770111083984, "step": 7107 }, { "epoch": 0.8270886665115197, "grad_norm": 1.5940618515014648, "learning_rate": 0.0005, "loss": 2.7112298011779785, "step": 7108 }, { "epoch": 0.8272050267628578, "grad_norm": 1.5755163431167603, "learning_rate": 0.0005, "loss": 2.493800163269043, "step": 7109 }, { "epoch": 0.827321387014196, "grad_norm": 1.270740032196045, "learning_rate": 0.0005, "loss": 2.305177688598633, "step": 7110 }, { "epoch": 0.8274377472655341, "grad_norm": 1.3615232706069946, "learning_rate": 0.0005, "loss": 2.4689254760742188, "step": 7111 }, { "epoch": 0.8275541075168722, "grad_norm": 1.266920566558838, "learning_rate": 0.0005, "loss": 1.9293932914733887, "step": 7112 }, { "epoch": 0.8276704677682104, "grad_norm": 1.385542631149292, "learning_rate": 0.0005, "loss": 1.893470287322998, "step": 7113 }, { "epoch": 0.8277868280195485, "grad_norm": 2.4953460693359375, "learning_rate": 0.0005, "loss": 3.3077940940856934, "step": 7114 }, { "epoch": 0.8279031882708867, "grad_norm": 1.7172435522079468, "learning_rate": 0.0005, "loss": 2.9281373023986816, "step": 7115 }, { "epoch": 0.8280195485222248, "grad_norm": 1.313448429107666, "learning_rate": 0.0005, "loss": 1.882954478263855, "step": 7116 }, { "epoch": 0.8281359087735629, "grad_norm": 1.5484652519226074, "learning_rate": 0.0005, "loss": 2.1163060665130615, "step": 7117 }, { "epoch": 0.8282522690249011, "grad_norm": 0.5703946352005005, "learning_rate": 0.0005, "loss": 1.0866649150848389, "step": 7118 }, { "epoch": 0.8283686292762392, "grad_norm": 0.9973875284194946, "learning_rate": 0.0005, "loss": 1.7161338329315186, "step": 7119 }, { "epoch": 0.8284849895275774, "grad_norm": 1.3992012739181519, "learning_rate": 0.0005, "loss": 2.5492663383483887, "step": 7120 }, { "epoch": 0.8286013497789155, "grad_norm": 1.3497521877288818, "learning_rate": 0.0005, "loss": 2.7082712650299072, "step": 7121 }, { "epoch": 0.8287177100302536, "grad_norm": 0.8852019309997559, "learning_rate": 0.0005, "loss": 2.0489699840545654, "step": 7122 }, { "epoch": 0.8288340702815918, "grad_norm": 3.272456407546997, "learning_rate": 0.0005, "loss": 2.9158055782318115, "step": 7123 }, { "epoch": 0.8289504305329299, "grad_norm": 2.3982295989990234, "learning_rate": 0.0005, "loss": 2.2909045219421387, "step": 7124 }, { "epoch": 0.8290667907842681, "grad_norm": 1.6032079458236694, "learning_rate": 0.0005, "loss": 2.790508270263672, "step": 7125 }, { "epoch": 0.8291831510356062, "grad_norm": 1.4078112840652466, "learning_rate": 0.0005, "loss": 2.825437307357788, "step": 7126 }, { "epoch": 0.8292995112869443, "grad_norm": 1.1161376237869263, "learning_rate": 0.0005, "loss": 3.0987730026245117, "step": 7127 }, { "epoch": 0.8294158715382826, "grad_norm": 1.7460917234420776, "learning_rate": 0.0005, "loss": 3.4202606678009033, "step": 7128 }, { "epoch": 0.8295322317896207, "grad_norm": 0.9163888692855835, "learning_rate": 0.0005, "loss": 2.7025861740112305, "step": 7129 }, { "epoch": 0.8296485920409588, "grad_norm": 1.058613657951355, "learning_rate": 0.0005, "loss": 1.8627710342407227, "step": 7130 }, { "epoch": 0.829764952292297, "grad_norm": 2.5131242275238037, "learning_rate": 0.0005, "loss": 2.9620201587677, "step": 7131 }, { "epoch": 0.8298813125436351, "grad_norm": 1.262831449508667, "learning_rate": 0.0005, "loss": 2.521150588989258, "step": 7132 }, { "epoch": 0.8299976727949733, "grad_norm": 1.6379066705703735, "learning_rate": 0.0005, "loss": 2.760106325149536, "step": 7133 }, { "epoch": 0.8301140330463114, "grad_norm": 1.4704134464263916, "learning_rate": 0.0005, "loss": 3.531337261199951, "step": 7134 }, { "epoch": 0.8302303932976495, "grad_norm": 1.0331366062164307, "learning_rate": 0.0005, "loss": 1.736060619354248, "step": 7135 }, { "epoch": 0.8303467535489877, "grad_norm": 1.0355315208435059, "learning_rate": 0.0005, "loss": 2.4481263160705566, "step": 7136 }, { "epoch": 0.8304631138003258, "grad_norm": 0.7969640493392944, "learning_rate": 0.0005, "loss": 1.493196964263916, "step": 7137 }, { "epoch": 0.830579474051664, "grad_norm": 1.382011890411377, "learning_rate": 0.0005, "loss": 2.3457577228546143, "step": 7138 }, { "epoch": 0.8306958343030021, "grad_norm": 1.6070162057876587, "learning_rate": 0.0005, "loss": 2.249438762664795, "step": 7139 }, { "epoch": 0.8308121945543402, "grad_norm": 1.6092966794967651, "learning_rate": 0.0005, "loss": 2.4575600624084473, "step": 7140 }, { "epoch": 0.8309285548056784, "grad_norm": 1.1684931516647339, "learning_rate": 0.0005, "loss": 2.1728193759918213, "step": 7141 }, { "epoch": 0.8310449150570165, "grad_norm": 1.3347275257110596, "learning_rate": 0.0005, "loss": 1.9064826965332031, "step": 7142 }, { "epoch": 0.8311612753083547, "grad_norm": 1.3321881294250488, "learning_rate": 0.0005, "loss": 1.8040845394134521, "step": 7143 }, { "epoch": 0.8312776355596928, "grad_norm": 1.511112928390503, "learning_rate": 0.0005, "loss": 2.164616584777832, "step": 7144 }, { "epoch": 0.8313939958110309, "grad_norm": 1.5049506425857544, "learning_rate": 0.0005, "loss": 2.216261386871338, "step": 7145 }, { "epoch": 0.8315103560623691, "grad_norm": 1.2834945917129517, "learning_rate": 0.0005, "loss": 1.8791561126708984, "step": 7146 }, { "epoch": 0.8316267163137072, "grad_norm": 1.349661946296692, "learning_rate": 0.0005, "loss": 2.913200616836548, "step": 7147 }, { "epoch": 0.8317430765650454, "grad_norm": 2.018167734146118, "learning_rate": 0.0005, "loss": 2.017949104309082, "step": 7148 }, { "epoch": 0.8318594368163835, "grad_norm": 1.4160137176513672, "learning_rate": 0.0005, "loss": 1.720937967300415, "step": 7149 }, { "epoch": 0.8319757970677216, "grad_norm": 1.2729108333587646, "learning_rate": 0.0005, "loss": 2.7972218990325928, "step": 7150 }, { "epoch": 0.8320921573190598, "grad_norm": 1.6946947574615479, "learning_rate": 0.0005, "loss": 2.7510480880737305, "step": 7151 }, { "epoch": 0.8322085175703979, "grad_norm": 1.2747946977615356, "learning_rate": 0.0005, "loss": 1.8626092672348022, "step": 7152 }, { "epoch": 0.8323248778217361, "grad_norm": 1.3882524967193604, "learning_rate": 0.0005, "loss": 2.2560248374938965, "step": 7153 }, { "epoch": 0.8324412380730742, "grad_norm": 1.898413896560669, "learning_rate": 0.0005, "loss": 2.909013032913208, "step": 7154 }, { "epoch": 0.8325575983244123, "grad_norm": 1.7770923376083374, "learning_rate": 0.0005, "loss": 2.679194927215576, "step": 7155 }, { "epoch": 0.8326739585757506, "grad_norm": 1.3247777223587036, "learning_rate": 0.0005, "loss": 3.0848093032836914, "step": 7156 }, { "epoch": 0.8327903188270886, "grad_norm": 1.717225432395935, "learning_rate": 0.0005, "loss": 1.9930224418640137, "step": 7157 }, { "epoch": 0.8329066790784269, "grad_norm": 1.4592241048812866, "learning_rate": 0.0005, "loss": 2.1450414657592773, "step": 7158 }, { "epoch": 0.833023039329765, "grad_norm": 1.8933430910110474, "learning_rate": 0.0005, "loss": 2.4284043312072754, "step": 7159 }, { "epoch": 0.8331393995811031, "grad_norm": 1.5718965530395508, "learning_rate": 0.0005, "loss": 1.7448866367340088, "step": 7160 }, { "epoch": 0.8332557598324413, "grad_norm": 1.0139873027801514, "learning_rate": 0.0005, "loss": 1.768707275390625, "step": 7161 }, { "epoch": 0.8333721200837794, "grad_norm": 1.7229325771331787, "learning_rate": 0.0005, "loss": 2.2773618698120117, "step": 7162 }, { "epoch": 0.8334884803351176, "grad_norm": 0.8128993511199951, "learning_rate": 0.0005, "loss": 1.2742533683776855, "step": 7163 }, { "epoch": 0.8336048405864557, "grad_norm": 1.282706618309021, "learning_rate": 0.0005, "loss": 1.9534310102462769, "step": 7164 }, { "epoch": 0.8337212008377938, "grad_norm": 2.448305368423462, "learning_rate": 0.0005, "loss": 2.978102922439575, "step": 7165 }, { "epoch": 0.833837561089132, "grad_norm": 0.9336016774177551, "learning_rate": 0.0005, "loss": 1.9547499418258667, "step": 7166 }, { "epoch": 0.8339539213404701, "grad_norm": 2.848606586456299, "learning_rate": 0.0005, "loss": 2.314222574234009, "step": 7167 }, { "epoch": 0.8340702815918083, "grad_norm": 1.4495707750320435, "learning_rate": 0.0005, "loss": 2.0064759254455566, "step": 7168 }, { "epoch": 0.8341866418431464, "grad_norm": 1.0385042428970337, "learning_rate": 0.0005, "loss": 2.1381728649139404, "step": 7169 }, { "epoch": 0.8343030020944845, "grad_norm": 2.8694729804992676, "learning_rate": 0.0005, "loss": 3.345879077911377, "step": 7170 }, { "epoch": 0.8344193623458227, "grad_norm": 1.5141167640686035, "learning_rate": 0.0005, "loss": 2.1040079593658447, "step": 7171 }, { "epoch": 0.8345357225971608, "grad_norm": 1.153153419494629, "learning_rate": 0.0005, "loss": 2.3298792839050293, "step": 7172 }, { "epoch": 0.8346520828484989, "grad_norm": 1.954288363456726, "learning_rate": 0.0005, "loss": 2.5547423362731934, "step": 7173 }, { "epoch": 0.8347684430998371, "grad_norm": 0.8079365491867065, "learning_rate": 0.0005, "loss": 1.5330579280853271, "step": 7174 }, { "epoch": 0.8348848033511752, "grad_norm": 1.4723296165466309, "learning_rate": 0.0005, "loss": 2.540076494216919, "step": 7175 }, { "epoch": 0.8350011636025134, "grad_norm": 1.5651168823242188, "learning_rate": 0.0005, "loss": 2.969594955444336, "step": 7176 }, { "epoch": 0.8351175238538515, "grad_norm": 1.1025395393371582, "learning_rate": 0.0005, "loss": 2.289682626724243, "step": 7177 }, { "epoch": 0.8352338841051896, "grad_norm": 1.8671562671661377, "learning_rate": 0.0005, "loss": 1.9936113357543945, "step": 7178 }, { "epoch": 0.8353502443565278, "grad_norm": 1.5300028324127197, "learning_rate": 0.0005, "loss": 3.114220142364502, "step": 7179 }, { "epoch": 0.8354666046078659, "grad_norm": 1.8775746822357178, "learning_rate": 0.0005, "loss": 2.4781179428100586, "step": 7180 }, { "epoch": 0.8355829648592041, "grad_norm": 1.9882957935333252, "learning_rate": 0.0005, "loss": 2.8179802894592285, "step": 7181 }, { "epoch": 0.8356993251105422, "grad_norm": 2.2057809829711914, "learning_rate": 0.0005, "loss": 2.5045623779296875, "step": 7182 }, { "epoch": 0.8358156853618803, "grad_norm": 1.0819768905639648, "learning_rate": 0.0005, "loss": 1.8770086765289307, "step": 7183 }, { "epoch": 0.8359320456132185, "grad_norm": 1.6872144937515259, "learning_rate": 0.0005, "loss": 2.29026460647583, "step": 7184 }, { "epoch": 0.8360484058645566, "grad_norm": 1.4438096284866333, "learning_rate": 0.0005, "loss": 2.5322823524475098, "step": 7185 }, { "epoch": 0.8361647661158949, "grad_norm": 1.2644881010055542, "learning_rate": 0.0005, "loss": 2.6206979751586914, "step": 7186 }, { "epoch": 0.836281126367233, "grad_norm": 1.538777232170105, "learning_rate": 0.0005, "loss": 2.5225636959075928, "step": 7187 }, { "epoch": 0.836397486618571, "grad_norm": 1.6736621856689453, "learning_rate": 0.0005, "loss": 2.9460060596466064, "step": 7188 }, { "epoch": 0.8365138468699093, "grad_norm": 1.182669758796692, "learning_rate": 0.0005, "loss": 2.3157453536987305, "step": 7189 }, { "epoch": 0.8366302071212474, "grad_norm": 1.5039441585540771, "learning_rate": 0.0005, "loss": 2.520069122314453, "step": 7190 }, { "epoch": 0.8367465673725856, "grad_norm": 1.4831101894378662, "learning_rate": 0.0005, "loss": 2.28912353515625, "step": 7191 }, { "epoch": 0.8368629276239237, "grad_norm": 1.2016595602035522, "learning_rate": 0.0005, "loss": 2.17520809173584, "step": 7192 }, { "epoch": 0.8369792878752618, "grad_norm": 1.8380005359649658, "learning_rate": 0.0005, "loss": 2.6658096313476562, "step": 7193 }, { "epoch": 0.8370956481266, "grad_norm": 2.056126356124878, "learning_rate": 0.0005, "loss": 2.017397403717041, "step": 7194 }, { "epoch": 0.8372120083779381, "grad_norm": 1.756319522857666, "learning_rate": 0.0005, "loss": 2.5244383811950684, "step": 7195 }, { "epoch": 0.8373283686292763, "grad_norm": 1.8791868686676025, "learning_rate": 0.0005, "loss": 2.476640224456787, "step": 7196 }, { "epoch": 0.8374447288806144, "grad_norm": 1.5600095987319946, "learning_rate": 0.0005, "loss": 2.0468311309814453, "step": 7197 }, { "epoch": 0.8375610891319525, "grad_norm": 1.2134236097335815, "learning_rate": 0.0005, "loss": 2.518693685531616, "step": 7198 }, { "epoch": 0.8376774493832907, "grad_norm": 1.7397950887680054, "learning_rate": 0.0005, "loss": 2.7853145599365234, "step": 7199 }, { "epoch": 0.8377938096346288, "grad_norm": 0.5180177688598633, "learning_rate": 0.0005, "loss": 1.0943127870559692, "step": 7200 }, { "epoch": 0.837910169885967, "grad_norm": 1.1787573099136353, "learning_rate": 0.0005, "loss": 1.8835822343826294, "step": 7201 }, { "epoch": 0.8380265301373051, "grad_norm": 1.232109546661377, "learning_rate": 0.0005, "loss": 1.9473202228546143, "step": 7202 }, { "epoch": 0.8381428903886432, "grad_norm": 1.583120584487915, "learning_rate": 0.0005, "loss": 1.5241284370422363, "step": 7203 }, { "epoch": 0.8382592506399814, "grad_norm": 1.2476497888565063, "learning_rate": 0.0005, "loss": 2.0865120887756348, "step": 7204 }, { "epoch": 0.8383756108913195, "grad_norm": 1.2361143827438354, "learning_rate": 0.0005, "loss": 2.24980092048645, "step": 7205 }, { "epoch": 0.8384919711426577, "grad_norm": 1.4632983207702637, "learning_rate": 0.0005, "loss": 1.909865379333496, "step": 7206 }, { "epoch": 0.8386083313939958, "grad_norm": 1.4117417335510254, "learning_rate": 0.0005, "loss": 1.748267650604248, "step": 7207 }, { "epoch": 0.8387246916453339, "grad_norm": 2.198690176010132, "learning_rate": 0.0005, "loss": 2.323390483856201, "step": 7208 }, { "epoch": 0.8388410518966721, "grad_norm": 0.9254676103591919, "learning_rate": 0.0005, "loss": 2.709764242172241, "step": 7209 }, { "epoch": 0.8389574121480102, "grad_norm": 1.2935070991516113, "learning_rate": 0.0005, "loss": 2.5961060523986816, "step": 7210 }, { "epoch": 0.8390737723993483, "grad_norm": 1.933647632598877, "learning_rate": 0.0005, "loss": 2.824476957321167, "step": 7211 }, { "epoch": 0.8391901326506865, "grad_norm": 1.2986048460006714, "learning_rate": 0.0005, "loss": 2.1342153549194336, "step": 7212 }, { "epoch": 0.8393064929020246, "grad_norm": 1.4772944450378418, "learning_rate": 0.0005, "loss": 3.242335319519043, "step": 7213 }, { "epoch": 0.8394228531533628, "grad_norm": 1.2324172258377075, "learning_rate": 0.0005, "loss": 1.8647892475128174, "step": 7214 }, { "epoch": 0.839539213404701, "grad_norm": 2.0456042289733887, "learning_rate": 0.0005, "loss": 1.9676761627197266, "step": 7215 }, { "epoch": 0.839655573656039, "grad_norm": 1.0394953489303589, "learning_rate": 0.0005, "loss": 2.1339221000671387, "step": 7216 }, { "epoch": 0.8397719339073773, "grad_norm": 1.3268777132034302, "learning_rate": 0.0005, "loss": 2.087648868560791, "step": 7217 }, { "epoch": 0.8398882941587154, "grad_norm": 1.140565037727356, "learning_rate": 0.0005, "loss": 2.3633840084075928, "step": 7218 }, { "epoch": 0.8400046544100536, "grad_norm": 0.952840268611908, "learning_rate": 0.0005, "loss": 2.1997079849243164, "step": 7219 }, { "epoch": 0.8401210146613917, "grad_norm": 0.9648594260215759, "learning_rate": 0.0005, "loss": 2.1947402954101562, "step": 7220 }, { "epoch": 0.8402373749127298, "grad_norm": 1.2272069454193115, "learning_rate": 0.0005, "loss": 2.4298031330108643, "step": 7221 }, { "epoch": 0.840353735164068, "grad_norm": 1.292668342590332, "learning_rate": 0.0005, "loss": 2.7970356941223145, "step": 7222 }, { "epoch": 0.8404700954154061, "grad_norm": 0.8012850880622864, "learning_rate": 0.0005, "loss": 1.810701608657837, "step": 7223 }, { "epoch": 0.8405864556667443, "grad_norm": 1.5792750120162964, "learning_rate": 0.0005, "loss": 1.8212939500808716, "step": 7224 }, { "epoch": 0.8407028159180824, "grad_norm": 1.3014273643493652, "learning_rate": 0.0005, "loss": 2.6775689125061035, "step": 7225 }, { "epoch": 0.8408191761694205, "grad_norm": 1.2217856645584106, "learning_rate": 0.0005, "loss": 2.6903879642486572, "step": 7226 }, { "epoch": 0.8409355364207587, "grad_norm": 1.2393276691436768, "learning_rate": 0.0005, "loss": 2.8859806060791016, "step": 7227 }, { "epoch": 0.8410518966720968, "grad_norm": 1.2128945589065552, "learning_rate": 0.0005, "loss": 1.832677960395813, "step": 7228 }, { "epoch": 0.841168256923435, "grad_norm": 1.2150315046310425, "learning_rate": 0.0005, "loss": 2.3815646171569824, "step": 7229 }, { "epoch": 0.8412846171747731, "grad_norm": 1.7171316146850586, "learning_rate": 0.0005, "loss": 2.3364198207855225, "step": 7230 }, { "epoch": 0.8414009774261112, "grad_norm": 1.1350462436676025, "learning_rate": 0.0005, "loss": 2.1683382987976074, "step": 7231 }, { "epoch": 0.8415173376774494, "grad_norm": 1.4148280620574951, "learning_rate": 0.0005, "loss": 1.8079999685287476, "step": 7232 }, { "epoch": 0.8416336979287875, "grad_norm": 1.120171308517456, "learning_rate": 0.0005, "loss": 2.2986388206481934, "step": 7233 }, { "epoch": 0.8417500581801257, "grad_norm": 1.420190691947937, "learning_rate": 0.0005, "loss": 2.539703845977783, "step": 7234 }, { "epoch": 0.8418664184314638, "grad_norm": 0.9660479426383972, "learning_rate": 0.0005, "loss": 2.2168703079223633, "step": 7235 }, { "epoch": 0.8419827786828019, "grad_norm": 1.1973906755447388, "learning_rate": 0.0005, "loss": 2.7644948959350586, "step": 7236 }, { "epoch": 0.8420991389341401, "grad_norm": 1.6416306495666504, "learning_rate": 0.0005, "loss": 2.2100939750671387, "step": 7237 }, { "epoch": 0.8422154991854782, "grad_norm": 1.3905656337738037, "learning_rate": 0.0005, "loss": 1.752725601196289, "step": 7238 }, { "epoch": 0.8423318594368164, "grad_norm": 0.804436206817627, "learning_rate": 0.0005, "loss": 1.355208396911621, "step": 7239 }, { "epoch": 0.8424482196881545, "grad_norm": 1.2542742490768433, "learning_rate": 0.0005, "loss": 2.408107280731201, "step": 7240 }, { "epoch": 0.8425645799394926, "grad_norm": 2.7105469703674316, "learning_rate": 0.0005, "loss": 2.457111358642578, "step": 7241 }, { "epoch": 0.8426809401908308, "grad_norm": 0.8489802479743958, "learning_rate": 0.0005, "loss": 2.759269952774048, "step": 7242 }, { "epoch": 0.8427973004421689, "grad_norm": 1.4307667016983032, "learning_rate": 0.0005, "loss": 3.00181245803833, "step": 7243 }, { "epoch": 0.8429136606935071, "grad_norm": 4.676695346832275, "learning_rate": 0.0005, "loss": 2.467134475708008, "step": 7244 }, { "epoch": 0.8430300209448452, "grad_norm": 1.53572678565979, "learning_rate": 0.0005, "loss": 2.6774024963378906, "step": 7245 }, { "epoch": 0.8431463811961833, "grad_norm": 1.9980276823043823, "learning_rate": 0.0005, "loss": 2.697667121887207, "step": 7246 }, { "epoch": 0.8432627414475216, "grad_norm": 1.2776896953582764, "learning_rate": 0.0005, "loss": 1.3161230087280273, "step": 7247 }, { "epoch": 0.8433791016988597, "grad_norm": 1.5445088148117065, "learning_rate": 0.0005, "loss": 2.5823793411254883, "step": 7248 }, { "epoch": 0.8434954619501979, "grad_norm": 1.2268372774124146, "learning_rate": 0.0005, "loss": 2.4809274673461914, "step": 7249 }, { "epoch": 0.843611822201536, "grad_norm": 0.7606256604194641, "learning_rate": 0.0005, "loss": 1.1360459327697754, "step": 7250 }, { "epoch": 0.8437281824528741, "grad_norm": 1.9570094347000122, "learning_rate": 0.0005, "loss": 2.4331648349761963, "step": 7251 }, { "epoch": 0.8438445427042123, "grad_norm": 1.1292356252670288, "learning_rate": 0.0005, "loss": 2.216275215148926, "step": 7252 }, { "epoch": 0.8439609029555504, "grad_norm": 1.2923243045806885, "learning_rate": 0.0005, "loss": 2.6075680255889893, "step": 7253 }, { "epoch": 0.8440772632068885, "grad_norm": 0.9561182260513306, "learning_rate": 0.0005, "loss": 2.0270767211914062, "step": 7254 }, { "epoch": 0.8441936234582267, "grad_norm": 1.4231712818145752, "learning_rate": 0.0005, "loss": 2.0843427181243896, "step": 7255 }, { "epoch": 0.8443099837095648, "grad_norm": 1.3594845533370972, "learning_rate": 0.0005, "loss": 2.359987735748291, "step": 7256 }, { "epoch": 0.844426343960903, "grad_norm": 3.210935354232788, "learning_rate": 0.0005, "loss": 2.714705467224121, "step": 7257 }, { "epoch": 0.8445427042122411, "grad_norm": 1.3267649412155151, "learning_rate": 0.0005, "loss": 2.462167978286743, "step": 7258 }, { "epoch": 0.8446590644635792, "grad_norm": 0.981423556804657, "learning_rate": 0.0005, "loss": 1.6584815979003906, "step": 7259 }, { "epoch": 0.8447754247149174, "grad_norm": 1.551095962524414, "learning_rate": 0.0005, "loss": 2.327941417694092, "step": 7260 }, { "epoch": 0.8448917849662555, "grad_norm": 1.0690362453460693, "learning_rate": 0.0005, "loss": 1.9656391143798828, "step": 7261 }, { "epoch": 0.8450081452175937, "grad_norm": 1.4513180255889893, "learning_rate": 0.0005, "loss": 1.8614962100982666, "step": 7262 }, { "epoch": 0.8451245054689318, "grad_norm": 1.4109522104263306, "learning_rate": 0.0005, "loss": 2.05942440032959, "step": 7263 }, { "epoch": 0.8452408657202699, "grad_norm": 1.8859246969223022, "learning_rate": 0.0005, "loss": 3.0957283973693848, "step": 7264 }, { "epoch": 0.8453572259716081, "grad_norm": 1.61604905128479, "learning_rate": 0.0005, "loss": 2.9572272300720215, "step": 7265 }, { "epoch": 0.8454735862229462, "grad_norm": 1.7690647840499878, "learning_rate": 0.0005, "loss": 2.672975540161133, "step": 7266 }, { "epoch": 0.8455899464742844, "grad_norm": 1.1729378700256348, "learning_rate": 0.0005, "loss": 2.6009154319763184, "step": 7267 }, { "epoch": 0.8457063067256225, "grad_norm": 1.5625404119491577, "learning_rate": 0.0005, "loss": 2.4277281761169434, "step": 7268 }, { "epoch": 0.8458226669769606, "grad_norm": 1.4549027681350708, "learning_rate": 0.0005, "loss": 2.4344472885131836, "step": 7269 }, { "epoch": 0.8459390272282988, "grad_norm": 1.2889211177825928, "learning_rate": 0.0005, "loss": 2.062178611755371, "step": 7270 }, { "epoch": 0.8460553874796369, "grad_norm": 0.7684811949729919, "learning_rate": 0.0005, "loss": 1.8457649946212769, "step": 7271 }, { "epoch": 0.8461717477309751, "grad_norm": 1.9838310480117798, "learning_rate": 0.0005, "loss": 2.4478349685668945, "step": 7272 }, { "epoch": 0.8462881079823132, "grad_norm": 1.7230451107025146, "learning_rate": 0.0005, "loss": 2.5580039024353027, "step": 7273 }, { "epoch": 0.8464044682336513, "grad_norm": 1.5093069076538086, "learning_rate": 0.0005, "loss": 3.047706365585327, "step": 7274 }, { "epoch": 0.8465208284849896, "grad_norm": 0.7731769680976868, "learning_rate": 0.0005, "loss": 1.97607421875, "step": 7275 }, { "epoch": 0.8466371887363277, "grad_norm": 3.537811756134033, "learning_rate": 0.0005, "loss": 2.696404457092285, "step": 7276 }, { "epoch": 0.8467535489876659, "grad_norm": 1.3304028511047363, "learning_rate": 0.0005, "loss": 1.9827277660369873, "step": 7277 }, { "epoch": 0.846869909239004, "grad_norm": 1.7163864374160767, "learning_rate": 0.0005, "loss": 3.5278143882751465, "step": 7278 }, { "epoch": 0.8469862694903421, "grad_norm": 0.782795250415802, "learning_rate": 0.0005, "loss": 1.5615402460098267, "step": 7279 }, { "epoch": 0.8471026297416803, "grad_norm": 1.4799104928970337, "learning_rate": 0.0005, "loss": 2.1370949745178223, "step": 7280 }, { "epoch": 0.8472189899930184, "grad_norm": 0.9017885327339172, "learning_rate": 0.0005, "loss": 1.9913411140441895, "step": 7281 }, { "epoch": 0.8473353502443566, "grad_norm": 0.8976240158081055, "learning_rate": 0.0005, "loss": 1.4252334833145142, "step": 7282 }, { "epoch": 0.8474517104956947, "grad_norm": 1.9185298681259155, "learning_rate": 0.0005, "loss": 1.429037094116211, "step": 7283 }, { "epoch": 0.8475680707470328, "grad_norm": 0.8844945430755615, "learning_rate": 0.0005, "loss": 2.1599812507629395, "step": 7284 }, { "epoch": 0.847684430998371, "grad_norm": 1.110540747642517, "learning_rate": 0.0005, "loss": 2.1555356979370117, "step": 7285 }, { "epoch": 0.8478007912497091, "grad_norm": 1.2461997270584106, "learning_rate": 0.0005, "loss": 2.6421420574188232, "step": 7286 }, { "epoch": 0.8479171515010473, "grad_norm": 0.9884178042411804, "learning_rate": 0.0005, "loss": 2.47169828414917, "step": 7287 }, { "epoch": 0.8480335117523854, "grad_norm": 1.3715448379516602, "learning_rate": 0.0005, "loss": 2.831817388534546, "step": 7288 }, { "epoch": 0.8481498720037235, "grad_norm": 0.9852806925773621, "learning_rate": 0.0005, "loss": 1.8303420543670654, "step": 7289 }, { "epoch": 0.8482662322550617, "grad_norm": 1.246607780456543, "learning_rate": 0.0005, "loss": 2.42331600189209, "step": 7290 }, { "epoch": 0.8483825925063998, "grad_norm": 1.1572457551956177, "learning_rate": 0.0005, "loss": 2.1116232872009277, "step": 7291 }, { "epoch": 0.848498952757738, "grad_norm": 2.032175302505493, "learning_rate": 0.0005, "loss": 2.1904778480529785, "step": 7292 }, { "epoch": 0.8486153130090761, "grad_norm": 1.0355300903320312, "learning_rate": 0.0005, "loss": 2.1225481033325195, "step": 7293 }, { "epoch": 0.8487316732604142, "grad_norm": 1.051830768585205, "learning_rate": 0.0005, "loss": 2.44350528717041, "step": 7294 }, { "epoch": 0.8488480335117524, "grad_norm": 1.3846749067306519, "learning_rate": 0.0005, "loss": 2.524538993835449, "step": 7295 }, { "epoch": 0.8489643937630905, "grad_norm": 1.2628613710403442, "learning_rate": 0.0005, "loss": 1.8858897686004639, "step": 7296 }, { "epoch": 0.8490807540144286, "grad_norm": 0.6717677116394043, "learning_rate": 0.0005, "loss": 1.2113819122314453, "step": 7297 }, { "epoch": 0.8491971142657668, "grad_norm": 1.126994013786316, "learning_rate": 0.0005, "loss": 2.236440658569336, "step": 7298 }, { "epoch": 0.8493134745171049, "grad_norm": 1.3738186359405518, "learning_rate": 0.0005, "loss": 1.647007942199707, "step": 7299 }, { "epoch": 0.8494298347684431, "grad_norm": 1.2195475101470947, "learning_rate": 0.0005, "loss": 2.165083885192871, "step": 7300 }, { "epoch": 0.8495461950197812, "grad_norm": 1.162175178527832, "learning_rate": 0.0005, "loss": 2.3006911277770996, "step": 7301 }, { "epoch": 0.8496625552711193, "grad_norm": 1.5338112115859985, "learning_rate": 0.0005, "loss": 2.6254844665527344, "step": 7302 }, { "epoch": 0.8497789155224575, "grad_norm": 1.5297173261642456, "learning_rate": 0.0005, "loss": 1.9591673612594604, "step": 7303 }, { "epoch": 0.8498952757737956, "grad_norm": 1.394924283027649, "learning_rate": 0.0005, "loss": 2.2558434009552, "step": 7304 }, { "epoch": 0.8500116360251339, "grad_norm": 1.4988679885864258, "learning_rate": 0.0005, "loss": 2.8732311725616455, "step": 7305 }, { "epoch": 0.850127996276472, "grad_norm": 1.1908595561981201, "learning_rate": 0.0005, "loss": 2.170912742614746, "step": 7306 }, { "epoch": 0.85024435652781, "grad_norm": 1.4442332983016968, "learning_rate": 0.0005, "loss": 2.3796277046203613, "step": 7307 }, { "epoch": 0.8503607167791483, "grad_norm": 1.3786120414733887, "learning_rate": 0.0005, "loss": 1.931466817855835, "step": 7308 }, { "epoch": 0.8504770770304864, "grad_norm": 1.3706926107406616, "learning_rate": 0.0005, "loss": 3.793943166732788, "step": 7309 }, { "epoch": 0.8505934372818246, "grad_norm": 1.983642816543579, "learning_rate": 0.0005, "loss": 2.970552444458008, "step": 7310 }, { "epoch": 0.8507097975331627, "grad_norm": 1.6619069576263428, "learning_rate": 0.0005, "loss": 2.8158512115478516, "step": 7311 }, { "epoch": 0.8508261577845008, "grad_norm": 0.8528446555137634, "learning_rate": 0.0005, "loss": 2.1290829181671143, "step": 7312 }, { "epoch": 0.850942518035839, "grad_norm": 1.195029377937317, "learning_rate": 0.0005, "loss": 2.215364456176758, "step": 7313 }, { "epoch": 0.8510588782871771, "grad_norm": 1.3905590772628784, "learning_rate": 0.0005, "loss": 1.9964231252670288, "step": 7314 }, { "epoch": 0.8511752385385153, "grad_norm": 1.3477405309677124, "learning_rate": 0.0005, "loss": 2.147467613220215, "step": 7315 }, { "epoch": 0.8512915987898534, "grad_norm": 1.2671890258789062, "learning_rate": 0.0005, "loss": 1.6817983388900757, "step": 7316 }, { "epoch": 0.8514079590411915, "grad_norm": 1.1408965587615967, "learning_rate": 0.0005, "loss": 2.667942523956299, "step": 7317 }, { "epoch": 0.8515243192925297, "grad_norm": 1.4045153856277466, "learning_rate": 0.0005, "loss": 2.1822378635406494, "step": 7318 }, { "epoch": 0.8516406795438678, "grad_norm": 1.6409215927124023, "learning_rate": 0.0005, "loss": 2.5953056812286377, "step": 7319 }, { "epoch": 0.851757039795206, "grad_norm": 1.5329681634902954, "learning_rate": 0.0005, "loss": 1.7559280395507812, "step": 7320 }, { "epoch": 0.8518734000465441, "grad_norm": 0.9751830697059631, "learning_rate": 0.0005, "loss": 1.8563519716262817, "step": 7321 }, { "epoch": 0.8519897602978822, "grad_norm": 2.0468530654907227, "learning_rate": 0.0005, "loss": 2.1976261138916016, "step": 7322 }, { "epoch": 0.8521061205492204, "grad_norm": 0.9272493124008179, "learning_rate": 0.0005, "loss": 2.3773629665374756, "step": 7323 }, { "epoch": 0.8522224808005585, "grad_norm": 1.2822504043579102, "learning_rate": 0.0005, "loss": 2.5060267448425293, "step": 7324 }, { "epoch": 0.8523388410518967, "grad_norm": 1.5266422033309937, "learning_rate": 0.0005, "loss": 2.5864248275756836, "step": 7325 }, { "epoch": 0.8524552013032348, "grad_norm": 1.1513853073120117, "learning_rate": 0.0005, "loss": 2.107321262359619, "step": 7326 }, { "epoch": 0.8525715615545729, "grad_norm": 1.048236608505249, "learning_rate": 0.0005, "loss": 1.8990954160690308, "step": 7327 }, { "epoch": 0.8526879218059111, "grad_norm": 1.1985284090042114, "learning_rate": 0.0005, "loss": 3.3093743324279785, "step": 7328 }, { "epoch": 0.8528042820572492, "grad_norm": 1.7030162811279297, "learning_rate": 0.0005, "loss": 2.0666589736938477, "step": 7329 }, { "epoch": 0.8529206423085874, "grad_norm": 1.1483346223831177, "learning_rate": 0.0005, "loss": 3.0572400093078613, "step": 7330 }, { "epoch": 0.8530370025599255, "grad_norm": 1.7414566278457642, "learning_rate": 0.0005, "loss": 2.847698926925659, "step": 7331 }, { "epoch": 0.8531533628112636, "grad_norm": 1.3016647100448608, "learning_rate": 0.0005, "loss": 1.8599169254302979, "step": 7332 }, { "epoch": 0.8532697230626018, "grad_norm": 2.9228203296661377, "learning_rate": 0.0005, "loss": 3.176288604736328, "step": 7333 }, { "epoch": 0.85338608331394, "grad_norm": 1.311795711517334, "learning_rate": 0.0005, "loss": 2.1495041847229004, "step": 7334 }, { "epoch": 0.853502443565278, "grad_norm": 2.2205395698547363, "learning_rate": 0.0005, "loss": 2.6887102127075195, "step": 7335 }, { "epoch": 0.8536188038166163, "grad_norm": 1.7952438592910767, "learning_rate": 0.0005, "loss": 2.898383617401123, "step": 7336 }, { "epoch": 0.8537351640679544, "grad_norm": 1.948012113571167, "learning_rate": 0.0005, "loss": 2.016232490539551, "step": 7337 }, { "epoch": 0.8538515243192926, "grad_norm": 1.2450350522994995, "learning_rate": 0.0005, "loss": 2.1239945888519287, "step": 7338 }, { "epoch": 0.8539678845706307, "grad_norm": 1.4152717590332031, "learning_rate": 0.0005, "loss": 2.5391509532928467, "step": 7339 }, { "epoch": 0.8540842448219688, "grad_norm": 2.043558359146118, "learning_rate": 0.0005, "loss": 2.5449161529541016, "step": 7340 }, { "epoch": 0.854200605073307, "grad_norm": 0.8196771144866943, "learning_rate": 0.0005, "loss": 1.3082969188690186, "step": 7341 }, { "epoch": 0.8543169653246451, "grad_norm": 2.2368826866149902, "learning_rate": 0.0005, "loss": 2.6135611534118652, "step": 7342 }, { "epoch": 0.8544333255759833, "grad_norm": 1.2601827383041382, "learning_rate": 0.0005, "loss": 1.7397525310516357, "step": 7343 }, { "epoch": 0.8545496858273214, "grad_norm": 2.0919134616851807, "learning_rate": 0.0005, "loss": 2.826953649520874, "step": 7344 }, { "epoch": 0.8546660460786595, "grad_norm": 1.7846157550811768, "learning_rate": 0.0005, "loss": 1.9743337631225586, "step": 7345 }, { "epoch": 0.8547824063299977, "grad_norm": 2.089385747909546, "learning_rate": 0.0005, "loss": 2.518531322479248, "step": 7346 }, { "epoch": 0.8548987665813358, "grad_norm": 1.1455364227294922, "learning_rate": 0.0005, "loss": 2.4761533737182617, "step": 7347 }, { "epoch": 0.855015126832674, "grad_norm": 0.9546332955360413, "learning_rate": 0.0005, "loss": 2.0490400791168213, "step": 7348 }, { "epoch": 0.8551314870840121, "grad_norm": 1.7211815118789673, "learning_rate": 0.0005, "loss": 2.520174264907837, "step": 7349 }, { "epoch": 0.8552478473353502, "grad_norm": 1.719606876373291, "learning_rate": 0.0005, "loss": 2.471306324005127, "step": 7350 }, { "epoch": 0.8553642075866884, "grad_norm": 1.2098228931427002, "learning_rate": 0.0005, "loss": 2.0061376094818115, "step": 7351 }, { "epoch": 0.8554805678380265, "grad_norm": 2.0228347778320312, "learning_rate": 0.0005, "loss": 2.511622428894043, "step": 7352 }, { "epoch": 0.8555969280893647, "grad_norm": 1.7927120923995972, "learning_rate": 0.0005, "loss": 2.544503688812256, "step": 7353 }, { "epoch": 0.8557132883407028, "grad_norm": 2.3867099285125732, "learning_rate": 0.0005, "loss": 2.8809032440185547, "step": 7354 }, { "epoch": 0.8558296485920409, "grad_norm": 1.6710952520370483, "learning_rate": 0.0005, "loss": 2.5150675773620605, "step": 7355 }, { "epoch": 0.8559460088433791, "grad_norm": 2.7728686332702637, "learning_rate": 0.0005, "loss": 2.7812037467956543, "step": 7356 }, { "epoch": 0.8560623690947172, "grad_norm": 0.7622725367546082, "learning_rate": 0.0005, "loss": 1.8958749771118164, "step": 7357 }, { "epoch": 0.8561787293460554, "grad_norm": 1.1952533721923828, "learning_rate": 0.0005, "loss": 1.874843955039978, "step": 7358 }, { "epoch": 0.8562950895973935, "grad_norm": 1.2479636669158936, "learning_rate": 0.0005, "loss": 1.9980766773223877, "step": 7359 }, { "epoch": 0.8564114498487316, "grad_norm": 1.5067106485366821, "learning_rate": 0.0005, "loss": 2.7871828079223633, "step": 7360 }, { "epoch": 0.8565278101000698, "grad_norm": 1.2906702756881714, "learning_rate": 0.0005, "loss": 1.6264472007751465, "step": 7361 }, { "epoch": 0.8566441703514079, "grad_norm": 0.9911531209945679, "learning_rate": 0.0005, "loss": 2.0278382301330566, "step": 7362 }, { "epoch": 0.8567605306027462, "grad_norm": 1.1151337623596191, "learning_rate": 0.0005, "loss": 2.653186082839966, "step": 7363 }, { "epoch": 0.8568768908540842, "grad_norm": 1.4761502742767334, "learning_rate": 0.0005, "loss": 2.241825819015503, "step": 7364 }, { "epoch": 0.8569932511054223, "grad_norm": 0.8125339150428772, "learning_rate": 0.0005, "loss": 1.6073980331420898, "step": 7365 }, { "epoch": 0.8571096113567606, "grad_norm": 1.6586612462997437, "learning_rate": 0.0005, "loss": 2.5303421020507812, "step": 7366 }, { "epoch": 0.8572259716080987, "grad_norm": 2.002060890197754, "learning_rate": 0.0005, "loss": 2.3311126232147217, "step": 7367 }, { "epoch": 0.8573423318594369, "grad_norm": 0.910335123538971, "learning_rate": 0.0005, "loss": 2.4170541763305664, "step": 7368 }, { "epoch": 0.857458692110775, "grad_norm": 1.4475531578063965, "learning_rate": 0.0005, "loss": 2.8409032821655273, "step": 7369 }, { "epoch": 0.8575750523621131, "grad_norm": 0.8939231038093567, "learning_rate": 0.0005, "loss": 1.229626178741455, "step": 7370 }, { "epoch": 0.8576914126134513, "grad_norm": 0.8376776576042175, "learning_rate": 0.0005, "loss": 2.1473498344421387, "step": 7371 }, { "epoch": 0.8578077728647894, "grad_norm": 1.251231074333191, "learning_rate": 0.0005, "loss": 2.488067150115967, "step": 7372 }, { "epoch": 0.8579241331161276, "grad_norm": 1.7592638731002808, "learning_rate": 0.0005, "loss": 3.664572238922119, "step": 7373 }, { "epoch": 0.8580404933674657, "grad_norm": 1.218505620956421, "learning_rate": 0.0005, "loss": 2.4762990474700928, "step": 7374 }, { "epoch": 0.8581568536188038, "grad_norm": 1.428534746170044, "learning_rate": 0.0005, "loss": 3.1036932468414307, "step": 7375 }, { "epoch": 0.858273213870142, "grad_norm": 1.5141639709472656, "learning_rate": 0.0005, "loss": 2.484060764312744, "step": 7376 }, { "epoch": 0.8583895741214801, "grad_norm": 1.227901577949524, "learning_rate": 0.0005, "loss": 2.5841617584228516, "step": 7377 }, { "epoch": 0.8585059343728182, "grad_norm": 1.1901785135269165, "learning_rate": 0.0005, "loss": 2.2211475372314453, "step": 7378 }, { "epoch": 0.8586222946241564, "grad_norm": 1.4300209283828735, "learning_rate": 0.0005, "loss": 2.543771982192993, "step": 7379 }, { "epoch": 0.8587386548754945, "grad_norm": 1.772518515586853, "learning_rate": 0.0005, "loss": 2.0135796070098877, "step": 7380 }, { "epoch": 0.8588550151268327, "grad_norm": 1.1356319189071655, "learning_rate": 0.0005, "loss": 1.6645936965942383, "step": 7381 }, { "epoch": 0.8589713753781708, "grad_norm": 1.5838823318481445, "learning_rate": 0.0005, "loss": 2.6822509765625, "step": 7382 }, { "epoch": 0.8590877356295089, "grad_norm": 1.2055302858352661, "learning_rate": 0.0005, "loss": 1.7815945148468018, "step": 7383 }, { "epoch": 0.8592040958808471, "grad_norm": 1.4362872838974, "learning_rate": 0.0005, "loss": 2.075634002685547, "step": 7384 }, { "epoch": 0.8593204561321852, "grad_norm": 3.5295374393463135, "learning_rate": 0.0005, "loss": 2.5495080947875977, "step": 7385 }, { "epoch": 0.8594368163835234, "grad_norm": 1.7152513265609741, "learning_rate": 0.0005, "loss": 2.8836677074432373, "step": 7386 }, { "epoch": 0.8595531766348615, "grad_norm": 2.4477131366729736, "learning_rate": 0.0005, "loss": 2.695890426635742, "step": 7387 }, { "epoch": 0.8596695368861996, "grad_norm": 1.8506239652633667, "learning_rate": 0.0005, "loss": 3.1418676376342773, "step": 7388 }, { "epoch": 0.8597858971375378, "grad_norm": 1.4769601821899414, "learning_rate": 0.0005, "loss": 2.793297290802002, "step": 7389 }, { "epoch": 0.8599022573888759, "grad_norm": 1.2433916330337524, "learning_rate": 0.0005, "loss": 2.053267478942871, "step": 7390 }, { "epoch": 0.8600186176402141, "grad_norm": 2.6509292125701904, "learning_rate": 0.0005, "loss": 2.6453664302825928, "step": 7391 }, { "epoch": 0.8601349778915522, "grad_norm": 21.573993682861328, "learning_rate": 0.0005, "loss": 3.5023016929626465, "step": 7392 }, { "epoch": 0.8602513381428903, "grad_norm": 1.3858014345169067, "learning_rate": 0.0005, "loss": 2.367248058319092, "step": 7393 }, { "epoch": 0.8603676983942286, "grad_norm": 1.5112857818603516, "learning_rate": 0.0005, "loss": 2.5382602214813232, "step": 7394 }, { "epoch": 0.8604840586455667, "grad_norm": 1.6584181785583496, "learning_rate": 0.0005, "loss": 1.8203556537628174, "step": 7395 }, { "epoch": 0.8606004188969049, "grad_norm": 1.7776777744293213, "learning_rate": 0.0005, "loss": 2.3593621253967285, "step": 7396 }, { "epoch": 0.860716779148243, "grad_norm": 2.767014503479004, "learning_rate": 0.0005, "loss": 3.535219669342041, "step": 7397 }, { "epoch": 0.8608331393995811, "grad_norm": 2.0850534439086914, "learning_rate": 0.0005, "loss": 2.2505669593811035, "step": 7398 }, { "epoch": 0.8609494996509193, "grad_norm": 1.4178167581558228, "learning_rate": 0.0005, "loss": 2.3406975269317627, "step": 7399 }, { "epoch": 0.8610658599022574, "grad_norm": 2.0089633464813232, "learning_rate": 0.0005, "loss": 1.9569001197814941, "step": 7400 }, { "epoch": 0.8611822201535956, "grad_norm": 1.9799917936325073, "learning_rate": 0.0005, "loss": 2.362318515777588, "step": 7401 }, { "epoch": 0.8612985804049337, "grad_norm": 0.9783197641372681, "learning_rate": 0.0005, "loss": 2.0698091983795166, "step": 7402 }, { "epoch": 0.8614149406562718, "grad_norm": 0.8839658498764038, "learning_rate": 0.0005, "loss": 2.0148422718048096, "step": 7403 }, { "epoch": 0.86153130090761, "grad_norm": 1.4991214275360107, "learning_rate": 0.0005, "loss": 1.7811741828918457, "step": 7404 }, { "epoch": 0.8616476611589481, "grad_norm": 2.4252026081085205, "learning_rate": 0.0005, "loss": 3.1556732654571533, "step": 7405 }, { "epoch": 0.8617640214102863, "grad_norm": 0.9636747241020203, "learning_rate": 0.0005, "loss": 1.6295093297958374, "step": 7406 }, { "epoch": 0.8618803816616244, "grad_norm": 1.1272910833358765, "learning_rate": 0.0005, "loss": 1.8618462085723877, "step": 7407 }, { "epoch": 0.8619967419129625, "grad_norm": 1.3672295808792114, "learning_rate": 0.0005, "loss": 2.3672711849212646, "step": 7408 }, { "epoch": 0.8621131021643007, "grad_norm": 1.447733759880066, "learning_rate": 0.0005, "loss": 2.4873530864715576, "step": 7409 }, { "epoch": 0.8622294624156388, "grad_norm": 0.8154493570327759, "learning_rate": 0.0005, "loss": 1.9124138355255127, "step": 7410 }, { "epoch": 0.862345822666977, "grad_norm": 1.5855971574783325, "learning_rate": 0.0005, "loss": 3.0137758255004883, "step": 7411 }, { "epoch": 0.8624621829183151, "grad_norm": 1.099121332168579, "learning_rate": 0.0005, "loss": 2.3086259365081787, "step": 7412 }, { "epoch": 0.8625785431696532, "grad_norm": 0.9502010941505432, "learning_rate": 0.0005, "loss": 1.7936303615570068, "step": 7413 }, { "epoch": 0.8626949034209914, "grad_norm": 0.8685739636421204, "learning_rate": 0.0005, "loss": 2.1016156673431396, "step": 7414 }, { "epoch": 0.8628112636723295, "grad_norm": 1.1668963432312012, "learning_rate": 0.0005, "loss": 2.7766590118408203, "step": 7415 }, { "epoch": 0.8629276239236677, "grad_norm": 0.7794076800346375, "learning_rate": 0.0005, "loss": 1.6414121389389038, "step": 7416 }, { "epoch": 0.8630439841750058, "grad_norm": 0.8204219937324524, "learning_rate": 0.0005, "loss": 1.822816014289856, "step": 7417 }, { "epoch": 0.8631603444263439, "grad_norm": 1.4319376945495605, "learning_rate": 0.0005, "loss": 3.310063123703003, "step": 7418 }, { "epoch": 0.8632767046776821, "grad_norm": 1.138564944267273, "learning_rate": 0.0005, "loss": 2.5497007369995117, "step": 7419 }, { "epoch": 0.8633930649290202, "grad_norm": 1.0620466470718384, "learning_rate": 0.0005, "loss": 2.2167093753814697, "step": 7420 }, { "epoch": 0.8635094251803583, "grad_norm": 1.1952142715454102, "learning_rate": 0.0005, "loss": 1.6922788619995117, "step": 7421 }, { "epoch": 0.8636257854316965, "grad_norm": 1.5402023792266846, "learning_rate": 0.0005, "loss": 2.6348395347595215, "step": 7422 }, { "epoch": 0.8637421456830346, "grad_norm": 1.4082905054092407, "learning_rate": 0.0005, "loss": 2.9523284435272217, "step": 7423 }, { "epoch": 0.8638585059343729, "grad_norm": 1.0551034212112427, "learning_rate": 0.0005, "loss": 2.5934576988220215, "step": 7424 }, { "epoch": 0.863974866185711, "grad_norm": 1.4106476306915283, "learning_rate": 0.0005, "loss": 1.8469256162643433, "step": 7425 }, { "epoch": 0.864091226437049, "grad_norm": 0.8970038294792175, "learning_rate": 0.0005, "loss": 1.830808401107788, "step": 7426 }, { "epoch": 0.8642075866883873, "grad_norm": 1.6599029302597046, "learning_rate": 0.0005, "loss": 2.461170196533203, "step": 7427 }, { "epoch": 0.8643239469397254, "grad_norm": 1.066637635231018, "learning_rate": 0.0005, "loss": 2.529684066772461, "step": 7428 }, { "epoch": 0.8644403071910636, "grad_norm": 1.4417098760604858, "learning_rate": 0.0005, "loss": 2.371821880340576, "step": 7429 }, { "epoch": 0.8645566674424017, "grad_norm": 1.3539937734603882, "learning_rate": 0.0005, "loss": 2.875504493713379, "step": 7430 }, { "epoch": 0.8646730276937398, "grad_norm": 2.2023162841796875, "learning_rate": 0.0005, "loss": 2.665374279022217, "step": 7431 }, { "epoch": 0.864789387945078, "grad_norm": 1.2194329500198364, "learning_rate": 0.0005, "loss": 2.0112955570220947, "step": 7432 }, { "epoch": 0.8649057481964161, "grad_norm": 1.7602094411849976, "learning_rate": 0.0005, "loss": 3.2617125511169434, "step": 7433 }, { "epoch": 0.8650221084477543, "grad_norm": 1.3276101350784302, "learning_rate": 0.0005, "loss": 1.8738481998443604, "step": 7434 }, { "epoch": 0.8651384686990924, "grad_norm": 1.8504289388656616, "learning_rate": 0.0005, "loss": 2.6197662353515625, "step": 7435 }, { "epoch": 0.8652548289504305, "grad_norm": 1.117930293083191, "learning_rate": 0.0005, "loss": 2.194193124771118, "step": 7436 }, { "epoch": 0.8653711892017687, "grad_norm": 1.250488042831421, "learning_rate": 0.0005, "loss": 2.6384105682373047, "step": 7437 }, { "epoch": 0.8654875494531068, "grad_norm": 1.2093896865844727, "learning_rate": 0.0005, "loss": 1.8301167488098145, "step": 7438 }, { "epoch": 0.865603909704445, "grad_norm": 1.6260991096496582, "learning_rate": 0.0005, "loss": 2.860560178756714, "step": 7439 }, { "epoch": 0.8657202699557831, "grad_norm": 1.1152675151824951, "learning_rate": 0.0005, "loss": 1.873138427734375, "step": 7440 }, { "epoch": 0.8658366302071212, "grad_norm": 1.7157469987869263, "learning_rate": 0.0005, "loss": 2.920558452606201, "step": 7441 }, { "epoch": 0.8659529904584594, "grad_norm": 1.4451056718826294, "learning_rate": 0.0005, "loss": 3.1294941902160645, "step": 7442 }, { "epoch": 0.8660693507097975, "grad_norm": 1.1824122667312622, "learning_rate": 0.0005, "loss": 2.9618282318115234, "step": 7443 }, { "epoch": 0.8661857109611357, "grad_norm": 4.272015571594238, "learning_rate": 0.0005, "loss": 2.640583038330078, "step": 7444 }, { "epoch": 0.8663020712124738, "grad_norm": 1.3476123809814453, "learning_rate": 0.0005, "loss": 2.1835708618164062, "step": 7445 }, { "epoch": 0.8664184314638119, "grad_norm": 1.2234010696411133, "learning_rate": 0.0005, "loss": 2.4223012924194336, "step": 7446 }, { "epoch": 0.8665347917151501, "grad_norm": 1.6115639209747314, "learning_rate": 0.0005, "loss": 2.2652525901794434, "step": 7447 }, { "epoch": 0.8666511519664882, "grad_norm": 2.5585339069366455, "learning_rate": 0.0005, "loss": 3.7158641815185547, "step": 7448 }, { "epoch": 0.8667675122178264, "grad_norm": 1.227504014968872, "learning_rate": 0.0005, "loss": 2.2398762702941895, "step": 7449 }, { "epoch": 0.8668838724691645, "grad_norm": 1.2373822927474976, "learning_rate": 0.0005, "loss": 2.028163194656372, "step": 7450 }, { "epoch": 0.8670002327205026, "grad_norm": 0.820438027381897, "learning_rate": 0.0005, "loss": 1.4985196590423584, "step": 7451 }, { "epoch": 0.8671165929718408, "grad_norm": 1.0703589916229248, "learning_rate": 0.0005, "loss": 2.0853092670440674, "step": 7452 }, { "epoch": 0.867232953223179, "grad_norm": 1.3042060136795044, "learning_rate": 0.0005, "loss": 2.1953365802764893, "step": 7453 }, { "epoch": 0.8673493134745172, "grad_norm": 1.6257069110870361, "learning_rate": 0.0005, "loss": 2.7896909713745117, "step": 7454 }, { "epoch": 0.8674656737258553, "grad_norm": 1.2153270244598389, "learning_rate": 0.0005, "loss": 2.667573928833008, "step": 7455 }, { "epoch": 0.8675820339771934, "grad_norm": 2.092632532119751, "learning_rate": 0.0005, "loss": 2.562654733657837, "step": 7456 }, { "epoch": 0.8676983942285316, "grad_norm": 0.85234534740448, "learning_rate": 0.0005, "loss": 1.5290451049804688, "step": 7457 }, { "epoch": 0.8678147544798697, "grad_norm": 2.146806001663208, "learning_rate": 0.0005, "loss": 2.6604976654052734, "step": 7458 }, { "epoch": 0.8679311147312078, "grad_norm": 1.2860462665557861, "learning_rate": 0.0005, "loss": 2.579073429107666, "step": 7459 }, { "epoch": 0.868047474982546, "grad_norm": 1.33024263381958, "learning_rate": 0.0005, "loss": 2.371837615966797, "step": 7460 }, { "epoch": 0.8681638352338841, "grad_norm": 1.6998203992843628, "learning_rate": 0.0005, "loss": 2.204598903656006, "step": 7461 }, { "epoch": 0.8682801954852223, "grad_norm": 0.8870180249214172, "learning_rate": 0.0005, "loss": 1.6543526649475098, "step": 7462 }, { "epoch": 0.8683965557365604, "grad_norm": 1.66090726852417, "learning_rate": 0.0005, "loss": 2.636910915374756, "step": 7463 }, { "epoch": 0.8685129159878985, "grad_norm": 0.9343963861465454, "learning_rate": 0.0005, "loss": 1.7893280982971191, "step": 7464 }, { "epoch": 0.8686292762392367, "grad_norm": 1.3147834539413452, "learning_rate": 0.0005, "loss": 1.8488389253616333, "step": 7465 }, { "epoch": 0.8687456364905748, "grad_norm": 0.5621454119682312, "learning_rate": 0.0005, "loss": 1.130142092704773, "step": 7466 }, { "epoch": 0.868861996741913, "grad_norm": 0.9769304990768433, "learning_rate": 0.0005, "loss": 1.6979827880859375, "step": 7467 }, { "epoch": 0.8689783569932511, "grad_norm": 1.3610620498657227, "learning_rate": 0.0005, "loss": 2.2474608421325684, "step": 7468 }, { "epoch": 0.8690947172445892, "grad_norm": 0.974434494972229, "learning_rate": 0.0005, "loss": 1.691630482673645, "step": 7469 }, { "epoch": 0.8692110774959274, "grad_norm": 1.2432397603988647, "learning_rate": 0.0005, "loss": 2.793074369430542, "step": 7470 }, { "epoch": 0.8693274377472655, "grad_norm": 1.1343811750411987, "learning_rate": 0.0005, "loss": 1.767604947090149, "step": 7471 }, { "epoch": 0.8694437979986037, "grad_norm": 2.2297840118408203, "learning_rate": 0.0005, "loss": 2.174696445465088, "step": 7472 }, { "epoch": 0.8695601582499418, "grad_norm": 1.9543460607528687, "learning_rate": 0.0005, "loss": 2.272503137588501, "step": 7473 }, { "epoch": 0.8696765185012799, "grad_norm": 1.419333577156067, "learning_rate": 0.0005, "loss": 1.3611555099487305, "step": 7474 }, { "epoch": 0.8697928787526181, "grad_norm": 1.6241589784622192, "learning_rate": 0.0005, "loss": 2.7703423500061035, "step": 7475 }, { "epoch": 0.8699092390039562, "grad_norm": 1.7264245748519897, "learning_rate": 0.0005, "loss": 2.804145336151123, "step": 7476 }, { "epoch": 0.8700255992552944, "grad_norm": 1.6816891431808472, "learning_rate": 0.0005, "loss": 2.2423548698425293, "step": 7477 }, { "epoch": 0.8701419595066325, "grad_norm": 1.8037954568862915, "learning_rate": 0.0005, "loss": 1.7593297958374023, "step": 7478 }, { "epoch": 0.8702583197579706, "grad_norm": 0.8603208661079407, "learning_rate": 0.0005, "loss": 1.6702470779418945, "step": 7479 }, { "epoch": 0.8703746800093088, "grad_norm": 1.1671890020370483, "learning_rate": 0.0005, "loss": 2.184232711791992, "step": 7480 }, { "epoch": 0.8704910402606469, "grad_norm": 1.1566872596740723, "learning_rate": 0.0005, "loss": 2.546337127685547, "step": 7481 }, { "epoch": 0.8706074005119852, "grad_norm": 0.8317961692810059, "learning_rate": 0.0005, "loss": 1.4659088850021362, "step": 7482 }, { "epoch": 0.8707237607633233, "grad_norm": 1.3266754150390625, "learning_rate": 0.0005, "loss": 2.704472541809082, "step": 7483 }, { "epoch": 0.8708401210146614, "grad_norm": 1.1467777490615845, "learning_rate": 0.0005, "loss": 1.8815088272094727, "step": 7484 }, { "epoch": 0.8709564812659996, "grad_norm": 1.691009521484375, "learning_rate": 0.0005, "loss": 2.435115337371826, "step": 7485 }, { "epoch": 0.8710728415173377, "grad_norm": 1.5943281650543213, "learning_rate": 0.0005, "loss": 3.1273417472839355, "step": 7486 }, { "epoch": 0.8711892017686759, "grad_norm": 1.6962312459945679, "learning_rate": 0.0005, "loss": 2.706110954284668, "step": 7487 }, { "epoch": 0.871305562020014, "grad_norm": 2.4870078563690186, "learning_rate": 0.0005, "loss": 2.434384822845459, "step": 7488 }, { "epoch": 0.8714219222713521, "grad_norm": 1.9588547945022583, "learning_rate": 0.0005, "loss": 2.525993585586548, "step": 7489 }, { "epoch": 0.8715382825226903, "grad_norm": 1.129354476928711, "learning_rate": 0.0005, "loss": 2.584346294403076, "step": 7490 }, { "epoch": 0.8716546427740284, "grad_norm": 1.8777365684509277, "learning_rate": 0.0005, "loss": 2.964650869369507, "step": 7491 }, { "epoch": 0.8717710030253666, "grad_norm": 1.054437279701233, "learning_rate": 0.0005, "loss": 2.5022950172424316, "step": 7492 }, { "epoch": 0.8718873632767047, "grad_norm": 1.269099235534668, "learning_rate": 0.0005, "loss": 1.7879326343536377, "step": 7493 }, { "epoch": 0.8720037235280428, "grad_norm": 1.2545605897903442, "learning_rate": 0.0005, "loss": 1.9234315156936646, "step": 7494 }, { "epoch": 0.872120083779381, "grad_norm": 1.896484136581421, "learning_rate": 0.0005, "loss": 2.838390588760376, "step": 7495 }, { "epoch": 0.8722364440307191, "grad_norm": 1.6052870750427246, "learning_rate": 0.0005, "loss": 2.7203803062438965, "step": 7496 }, { "epoch": 0.8723528042820573, "grad_norm": 1.2210806608200073, "learning_rate": 0.0005, "loss": 2.644153118133545, "step": 7497 }, { "epoch": 0.8724691645333954, "grad_norm": 0.8145812749862671, "learning_rate": 0.0005, "loss": 1.7619259357452393, "step": 7498 }, { "epoch": 0.8725855247847335, "grad_norm": 2.4471936225891113, "learning_rate": 0.0005, "loss": 2.250641345977783, "step": 7499 }, { "epoch": 0.8727018850360717, "grad_norm": 1.6567739248275757, "learning_rate": 0.0005, "loss": 2.9869384765625, "step": 7500 }, { "epoch": 0.8728182452874098, "grad_norm": 1.545512318611145, "learning_rate": 0.0005, "loss": 2.8902976512908936, "step": 7501 }, { "epoch": 0.8729346055387479, "grad_norm": 0.5587537884712219, "learning_rate": 0.0005, "loss": 1.3718949556350708, "step": 7502 }, { "epoch": 0.8730509657900861, "grad_norm": 1.5455025434494019, "learning_rate": 0.0005, "loss": 2.708691358566284, "step": 7503 }, { "epoch": 0.8731673260414242, "grad_norm": 1.454641580581665, "learning_rate": 0.0005, "loss": 1.8936020135879517, "step": 7504 }, { "epoch": 0.8732836862927624, "grad_norm": 1.6343741416931152, "learning_rate": 0.0005, "loss": 2.994525194168091, "step": 7505 }, { "epoch": 0.8734000465441005, "grad_norm": 1.0994819402694702, "learning_rate": 0.0005, "loss": 2.0385069847106934, "step": 7506 }, { "epoch": 0.8735164067954386, "grad_norm": 0.8928747773170471, "learning_rate": 0.0005, "loss": 1.48801589012146, "step": 7507 }, { "epoch": 0.8736327670467768, "grad_norm": 1.5342358350753784, "learning_rate": 0.0005, "loss": 3.2185654640197754, "step": 7508 }, { "epoch": 0.8737491272981149, "grad_norm": 1.7559059858322144, "learning_rate": 0.0005, "loss": 4.026601791381836, "step": 7509 }, { "epoch": 0.8738654875494531, "grad_norm": 1.097216010093689, "learning_rate": 0.0005, "loss": 1.7757186889648438, "step": 7510 }, { "epoch": 0.8739818478007912, "grad_norm": 1.19537353515625, "learning_rate": 0.0005, "loss": 2.5444414615631104, "step": 7511 }, { "epoch": 0.8740982080521293, "grad_norm": 1.1260825395584106, "learning_rate": 0.0005, "loss": 2.7773334980010986, "step": 7512 }, { "epoch": 0.8742145683034676, "grad_norm": 1.5612316131591797, "learning_rate": 0.0005, "loss": 2.3869645595550537, "step": 7513 }, { "epoch": 0.8743309285548057, "grad_norm": 1.0451366901397705, "learning_rate": 0.0005, "loss": 2.316765308380127, "step": 7514 }, { "epoch": 0.8744472888061439, "grad_norm": 1.0250269174575806, "learning_rate": 0.0005, "loss": 2.3729686737060547, "step": 7515 }, { "epoch": 0.874563649057482, "grad_norm": 0.5963291525840759, "learning_rate": 0.0005, "loss": 1.0929369926452637, "step": 7516 }, { "epoch": 0.8746800093088201, "grad_norm": 0.7391331791877747, "learning_rate": 0.0005, "loss": 1.659428358078003, "step": 7517 }, { "epoch": 0.8747963695601583, "grad_norm": 2.9208929538726807, "learning_rate": 0.0005, "loss": 2.246293544769287, "step": 7518 }, { "epoch": 0.8749127298114964, "grad_norm": 2.8228485584259033, "learning_rate": 0.0005, "loss": 2.148181438446045, "step": 7519 }, { "epoch": 0.8750290900628346, "grad_norm": 1.7937575578689575, "learning_rate": 0.0005, "loss": 2.6455421447753906, "step": 7520 }, { "epoch": 0.8751454503141727, "grad_norm": 1.1776864528656006, "learning_rate": 0.0005, "loss": 1.4777305126190186, "step": 7521 }, { "epoch": 0.8752618105655108, "grad_norm": 1.5552092790603638, "learning_rate": 0.0005, "loss": 3.02337646484375, "step": 7522 }, { "epoch": 0.875378170816849, "grad_norm": 1.366343379020691, "learning_rate": 0.0005, "loss": 1.8336889743804932, "step": 7523 }, { "epoch": 0.8754945310681871, "grad_norm": 1.48308527469635, "learning_rate": 0.0005, "loss": 2.497908592224121, "step": 7524 }, { "epoch": 0.8756108913195253, "grad_norm": 1.4118342399597168, "learning_rate": 0.0005, "loss": 1.968435525894165, "step": 7525 }, { "epoch": 0.8757272515708634, "grad_norm": 1.5390958786010742, "learning_rate": 0.0005, "loss": 2.0478291511535645, "step": 7526 }, { "epoch": 0.8758436118222015, "grad_norm": 1.362062692642212, "learning_rate": 0.0005, "loss": 2.644242286682129, "step": 7527 }, { "epoch": 0.8759599720735397, "grad_norm": 0.6519679427146912, "learning_rate": 0.0005, "loss": 1.447988510131836, "step": 7528 }, { "epoch": 0.8760763323248778, "grad_norm": 0.733338475227356, "learning_rate": 0.0005, "loss": 1.4478821754455566, "step": 7529 }, { "epoch": 0.876192692576216, "grad_norm": 0.9789804816246033, "learning_rate": 0.0005, "loss": 1.8738739490509033, "step": 7530 }, { "epoch": 0.8763090528275541, "grad_norm": 1.340279221534729, "learning_rate": 0.0005, "loss": 1.9769842624664307, "step": 7531 }, { "epoch": 0.8764254130788922, "grad_norm": 1.8023453950881958, "learning_rate": 0.0005, "loss": 1.9448492527008057, "step": 7532 }, { "epoch": 0.8765417733302304, "grad_norm": 1.1699355840682983, "learning_rate": 0.0005, "loss": 2.562237024307251, "step": 7533 }, { "epoch": 0.8766581335815685, "grad_norm": 1.4323787689208984, "learning_rate": 0.0005, "loss": 3.07399582862854, "step": 7534 }, { "epoch": 0.8767744938329067, "grad_norm": 1.3571686744689941, "learning_rate": 0.0005, "loss": 1.4635775089263916, "step": 7535 }, { "epoch": 0.8768908540842448, "grad_norm": 1.090855360031128, "learning_rate": 0.0005, "loss": 2.4337446689605713, "step": 7536 }, { "epoch": 0.8770072143355829, "grad_norm": 1.043703317642212, "learning_rate": 0.0005, "loss": 1.8302669525146484, "step": 7537 }, { "epoch": 0.8771235745869211, "grad_norm": 2.1554765701293945, "learning_rate": 0.0005, "loss": 2.0906107425689697, "step": 7538 }, { "epoch": 0.8772399348382592, "grad_norm": 1.1198290586471558, "learning_rate": 0.0005, "loss": 2.6434309482574463, "step": 7539 }, { "epoch": 0.8773562950895974, "grad_norm": 1.286954641342163, "learning_rate": 0.0005, "loss": 2.0591533184051514, "step": 7540 }, { "epoch": 0.8774726553409355, "grad_norm": 1.489282488822937, "learning_rate": 0.0005, "loss": 2.688603401184082, "step": 7541 }, { "epoch": 0.8775890155922736, "grad_norm": 2.0174496173858643, "learning_rate": 0.0005, "loss": 2.3364174365997314, "step": 7542 }, { "epoch": 0.8777053758436119, "grad_norm": 1.9846857786178589, "learning_rate": 0.0005, "loss": 2.7865686416625977, "step": 7543 }, { "epoch": 0.87782173609495, "grad_norm": 1.2823373079299927, "learning_rate": 0.0005, "loss": 1.7457609176635742, "step": 7544 }, { "epoch": 0.877938096346288, "grad_norm": 1.5149736404418945, "learning_rate": 0.0005, "loss": 2.052527904510498, "step": 7545 }, { "epoch": 0.8780544565976263, "grad_norm": 1.3440476655960083, "learning_rate": 0.0005, "loss": 2.7102081775665283, "step": 7546 }, { "epoch": 0.8781708168489644, "grad_norm": 2.2056901454925537, "learning_rate": 0.0005, "loss": 2.5511631965637207, "step": 7547 }, { "epoch": 0.8782871771003026, "grad_norm": 1.3654868602752686, "learning_rate": 0.0005, "loss": 2.6894805431365967, "step": 7548 }, { "epoch": 0.8784035373516407, "grad_norm": 1.273969054222107, "learning_rate": 0.0005, "loss": 2.0915699005126953, "step": 7549 }, { "epoch": 0.8785198976029788, "grad_norm": 2.091421127319336, "learning_rate": 0.0005, "loss": 2.4726574420928955, "step": 7550 }, { "epoch": 0.878636257854317, "grad_norm": 1.3275110721588135, "learning_rate": 0.0005, "loss": 2.5534732341766357, "step": 7551 }, { "epoch": 0.8787526181056551, "grad_norm": 1.3663805723190308, "learning_rate": 0.0005, "loss": 2.4930381774902344, "step": 7552 }, { "epoch": 0.8788689783569933, "grad_norm": 1.022444486618042, "learning_rate": 0.0005, "loss": 1.4740307331085205, "step": 7553 }, { "epoch": 0.8789853386083314, "grad_norm": 1.2048794031143188, "learning_rate": 0.0005, "loss": 1.697634220123291, "step": 7554 }, { "epoch": 0.8791016988596695, "grad_norm": 0.61223304271698, "learning_rate": 0.0005, "loss": 1.6571018695831299, "step": 7555 }, { "epoch": 0.8792180591110077, "grad_norm": 1.3198397159576416, "learning_rate": 0.0005, "loss": 2.2289257049560547, "step": 7556 }, { "epoch": 0.8793344193623458, "grad_norm": 1.2193418741226196, "learning_rate": 0.0005, "loss": 2.0220963954925537, "step": 7557 }, { "epoch": 0.879450779613684, "grad_norm": 1.496264934539795, "learning_rate": 0.0005, "loss": 2.6834630966186523, "step": 7558 }, { "epoch": 0.8795671398650221, "grad_norm": 2.09137225151062, "learning_rate": 0.0005, "loss": 2.1111690998077393, "step": 7559 }, { "epoch": 0.8796835001163602, "grad_norm": 1.7990614175796509, "learning_rate": 0.0005, "loss": 2.806558609008789, "step": 7560 }, { "epoch": 0.8797998603676984, "grad_norm": 1.438984990119934, "learning_rate": 0.0005, "loss": 2.520688056945801, "step": 7561 }, { "epoch": 0.8799162206190365, "grad_norm": 1.292348861694336, "learning_rate": 0.0005, "loss": 2.791412353515625, "step": 7562 }, { "epoch": 0.8800325808703747, "grad_norm": 1.1297862529754639, "learning_rate": 0.0005, "loss": 2.5622057914733887, "step": 7563 }, { "epoch": 0.8801489411217128, "grad_norm": 1.9071013927459717, "learning_rate": 0.0005, "loss": 3.068544387817383, "step": 7564 }, { "epoch": 0.8802653013730509, "grad_norm": 1.5242595672607422, "learning_rate": 0.0005, "loss": 2.5415995121002197, "step": 7565 }, { "epoch": 0.8803816616243891, "grad_norm": 2.3393938541412354, "learning_rate": 0.0005, "loss": 2.8422422409057617, "step": 7566 }, { "epoch": 0.8804980218757272, "grad_norm": 1.239562749862671, "learning_rate": 0.0005, "loss": 1.7063689231872559, "step": 7567 }, { "epoch": 0.8806143821270654, "grad_norm": 1.2254582643508911, "learning_rate": 0.0005, "loss": 2.9893758296966553, "step": 7568 }, { "epoch": 0.8807307423784035, "grad_norm": 1.2592151165008545, "learning_rate": 0.0005, "loss": 2.0257070064544678, "step": 7569 }, { "epoch": 0.8808471026297416, "grad_norm": 1.2295900583267212, "learning_rate": 0.0005, "loss": 2.8434104919433594, "step": 7570 }, { "epoch": 0.8809634628810799, "grad_norm": 1.5729076862335205, "learning_rate": 0.0005, "loss": 2.56978702545166, "step": 7571 }, { "epoch": 0.881079823132418, "grad_norm": 1.5198302268981934, "learning_rate": 0.0005, "loss": 2.7385435104370117, "step": 7572 }, { "epoch": 0.8811961833837562, "grad_norm": 1.6081937551498413, "learning_rate": 0.0005, "loss": 2.3521342277526855, "step": 7573 }, { "epoch": 0.8813125436350943, "grad_norm": 0.7478731274604797, "learning_rate": 0.0005, "loss": 2.067654848098755, "step": 7574 }, { "epoch": 0.8814289038864324, "grad_norm": 0.9145020842552185, "learning_rate": 0.0005, "loss": 2.2973132133483887, "step": 7575 }, { "epoch": 0.8815452641377706, "grad_norm": 1.6396502256393433, "learning_rate": 0.0005, "loss": 2.621490955352783, "step": 7576 }, { "epoch": 0.8816616243891087, "grad_norm": 1.087424635887146, "learning_rate": 0.0005, "loss": 2.3484487533569336, "step": 7577 }, { "epoch": 0.8817779846404469, "grad_norm": 1.4651542901992798, "learning_rate": 0.0005, "loss": 2.5206809043884277, "step": 7578 }, { "epoch": 0.881894344891785, "grad_norm": 1.3096851110458374, "learning_rate": 0.0005, "loss": 2.80265474319458, "step": 7579 }, { "epoch": 0.8820107051431231, "grad_norm": 1.4618489742279053, "learning_rate": 0.0005, "loss": 2.86993408203125, "step": 7580 }, { "epoch": 0.8821270653944613, "grad_norm": 2.840264320373535, "learning_rate": 0.0005, "loss": 3.1843929290771484, "step": 7581 }, { "epoch": 0.8822434256457994, "grad_norm": 1.707365870475769, "learning_rate": 0.0005, "loss": 1.7961952686309814, "step": 7582 }, { "epoch": 0.8823597858971375, "grad_norm": 1.2709020376205444, "learning_rate": 0.0005, "loss": 1.9197465181350708, "step": 7583 }, { "epoch": 0.8824761461484757, "grad_norm": 1.0309407711029053, "learning_rate": 0.0005, "loss": 1.5525052547454834, "step": 7584 }, { "epoch": 0.8825925063998138, "grad_norm": 1.8501654863357544, "learning_rate": 0.0005, "loss": 2.0668411254882812, "step": 7585 }, { "epoch": 0.882708866651152, "grad_norm": 1.0544023513793945, "learning_rate": 0.0005, "loss": 1.7206133604049683, "step": 7586 }, { "epoch": 0.8828252269024901, "grad_norm": 1.7557587623596191, "learning_rate": 0.0005, "loss": 3.0555334091186523, "step": 7587 }, { "epoch": 0.8829415871538282, "grad_norm": 2.377202033996582, "learning_rate": 0.0005, "loss": 2.7300562858581543, "step": 7588 }, { "epoch": 0.8830579474051664, "grad_norm": 0.8553043603897095, "learning_rate": 0.0005, "loss": 2.1620545387268066, "step": 7589 }, { "epoch": 0.8831743076565045, "grad_norm": 1.5956419706344604, "learning_rate": 0.0005, "loss": 2.4754185676574707, "step": 7590 }, { "epoch": 0.8832906679078427, "grad_norm": 1.5890963077545166, "learning_rate": 0.0005, "loss": 2.615668773651123, "step": 7591 }, { "epoch": 0.8834070281591808, "grad_norm": 1.131912350654602, "learning_rate": 0.0005, "loss": 1.5541536808013916, "step": 7592 }, { "epoch": 0.8835233884105189, "grad_norm": 1.6965010166168213, "learning_rate": 0.0005, "loss": 1.6109833717346191, "step": 7593 }, { "epoch": 0.8836397486618571, "grad_norm": 1.1860867738723755, "learning_rate": 0.0005, "loss": 2.479269027709961, "step": 7594 }, { "epoch": 0.8837561089131952, "grad_norm": 1.0285496711730957, "learning_rate": 0.0005, "loss": 2.025484085083008, "step": 7595 }, { "epoch": 0.8838724691645334, "grad_norm": 2.2502620220184326, "learning_rate": 0.0005, "loss": 3.93892502784729, "step": 7596 }, { "epoch": 0.8839888294158715, "grad_norm": 2.075653076171875, "learning_rate": 0.0005, "loss": 2.5718159675598145, "step": 7597 }, { "epoch": 0.8841051896672096, "grad_norm": 41.718021392822266, "learning_rate": 0.0005, "loss": 2.2594945430755615, "step": 7598 }, { "epoch": 0.8842215499185478, "grad_norm": 1.5182478427886963, "learning_rate": 0.0005, "loss": 2.3418054580688477, "step": 7599 }, { "epoch": 0.8843379101698859, "grad_norm": 1.1640805006027222, "learning_rate": 0.0005, "loss": 1.6629555225372314, "step": 7600 }, { "epoch": 0.8844542704212242, "grad_norm": 1.5043467283248901, "learning_rate": 0.0005, "loss": 2.069178581237793, "step": 7601 }, { "epoch": 0.8845706306725623, "grad_norm": 1.489148736000061, "learning_rate": 0.0005, "loss": 3.2146148681640625, "step": 7602 }, { "epoch": 0.8846869909239004, "grad_norm": 1.6556379795074463, "learning_rate": 0.0005, "loss": 2.351588726043701, "step": 7603 }, { "epoch": 0.8848033511752386, "grad_norm": 0.7688738107681274, "learning_rate": 0.0005, "loss": 2.191821813583374, "step": 7604 }, { "epoch": 0.8849197114265767, "grad_norm": 0.7986741065979004, "learning_rate": 0.0005, "loss": 1.5103987455368042, "step": 7605 }, { "epoch": 0.8850360716779149, "grad_norm": 1.6662285327911377, "learning_rate": 0.0005, "loss": 2.004929542541504, "step": 7606 }, { "epoch": 0.885152431929253, "grad_norm": 0.9175357818603516, "learning_rate": 0.0005, "loss": 1.3038275241851807, "step": 7607 }, { "epoch": 0.8852687921805911, "grad_norm": 1.7825208902359009, "learning_rate": 0.0005, "loss": 2.088756561279297, "step": 7608 }, { "epoch": 0.8853851524319293, "grad_norm": 1.6790274381637573, "learning_rate": 0.0005, "loss": 1.8432807922363281, "step": 7609 }, { "epoch": 0.8855015126832674, "grad_norm": 1.6883631944656372, "learning_rate": 0.0005, "loss": 2.920741558074951, "step": 7610 }, { "epoch": 0.8856178729346056, "grad_norm": 1.2082051038742065, "learning_rate": 0.0005, "loss": 2.2494900226593018, "step": 7611 }, { "epoch": 0.8857342331859437, "grad_norm": 0.9402722120285034, "learning_rate": 0.0005, "loss": 2.467503547668457, "step": 7612 }, { "epoch": 0.8858505934372818, "grad_norm": 1.2917677164077759, "learning_rate": 0.0005, "loss": 2.553617000579834, "step": 7613 }, { "epoch": 0.88596695368862, "grad_norm": 1.420607566833496, "learning_rate": 0.0005, "loss": 3.0674867630004883, "step": 7614 }, { "epoch": 0.8860833139399581, "grad_norm": 1.465802788734436, "learning_rate": 0.0005, "loss": 2.1950511932373047, "step": 7615 }, { "epoch": 0.8861996741912963, "grad_norm": 1.1932967901229858, "learning_rate": 0.0005, "loss": 1.9550352096557617, "step": 7616 }, { "epoch": 0.8863160344426344, "grad_norm": 1.7882511615753174, "learning_rate": 0.0005, "loss": 2.5256028175354004, "step": 7617 }, { "epoch": 0.8864323946939725, "grad_norm": 1.758410930633545, "learning_rate": 0.0005, "loss": 2.467986583709717, "step": 7618 }, { "epoch": 0.8865487549453107, "grad_norm": 1.647929072380066, "learning_rate": 0.0005, "loss": 1.8724274635314941, "step": 7619 }, { "epoch": 0.8866651151966488, "grad_norm": 0.942543625831604, "learning_rate": 0.0005, "loss": 1.4572038650512695, "step": 7620 }, { "epoch": 0.886781475447987, "grad_norm": 1.0083767175674438, "learning_rate": 0.0005, "loss": 3.936140537261963, "step": 7621 }, { "epoch": 0.8868978356993251, "grad_norm": 2.6221048831939697, "learning_rate": 0.0005, "loss": 1.6594977378845215, "step": 7622 }, { "epoch": 0.8870141959506632, "grad_norm": 1.6928577423095703, "learning_rate": 0.0005, "loss": 3.1067676544189453, "step": 7623 }, { "epoch": 0.8871305562020014, "grad_norm": 2.0457727909088135, "learning_rate": 0.0005, "loss": 3.151747703552246, "step": 7624 }, { "epoch": 0.8872469164533395, "grad_norm": 1.2879018783569336, "learning_rate": 0.0005, "loss": 1.8450522422790527, "step": 7625 }, { "epoch": 0.8873632767046776, "grad_norm": 1.0021945238113403, "learning_rate": 0.0005, "loss": 2.0449838638305664, "step": 7626 }, { "epoch": 0.8874796369560158, "grad_norm": 1.1823898553848267, "learning_rate": 0.0005, "loss": 1.8217394351959229, "step": 7627 }, { "epoch": 0.8875959972073539, "grad_norm": 1.459945559501648, "learning_rate": 0.0005, "loss": 2.59242844581604, "step": 7628 }, { "epoch": 0.8877123574586921, "grad_norm": 0.7351071834564209, "learning_rate": 0.0005, "loss": 1.5753982067108154, "step": 7629 }, { "epoch": 0.8878287177100302, "grad_norm": 1.246698021888733, "learning_rate": 0.0005, "loss": 1.6578035354614258, "step": 7630 }, { "epoch": 0.8879450779613683, "grad_norm": 1.8988116979599, "learning_rate": 0.0005, "loss": 3.0709636211395264, "step": 7631 }, { "epoch": 0.8880614382127066, "grad_norm": 1.41653311252594, "learning_rate": 0.0005, "loss": 2.009960174560547, "step": 7632 }, { "epoch": 0.8881777984640447, "grad_norm": 0.8465510606765747, "learning_rate": 0.0005, "loss": 1.9574371576309204, "step": 7633 }, { "epoch": 0.8882941587153829, "grad_norm": 1.3666266202926636, "learning_rate": 0.0005, "loss": 2.001889228820801, "step": 7634 }, { "epoch": 0.888410518966721, "grad_norm": 1.3032468557357788, "learning_rate": 0.0005, "loss": 2.133683204650879, "step": 7635 }, { "epoch": 0.8885268792180591, "grad_norm": 1.0615426301956177, "learning_rate": 0.0005, "loss": 1.5989503860473633, "step": 7636 }, { "epoch": 0.8886432394693973, "grad_norm": 1.007817029953003, "learning_rate": 0.0005, "loss": 2.7354302406311035, "step": 7637 }, { "epoch": 0.8887595997207354, "grad_norm": 1.456581711769104, "learning_rate": 0.0005, "loss": 2.9933371543884277, "step": 7638 }, { "epoch": 0.8888759599720736, "grad_norm": 1.2057504653930664, "learning_rate": 0.0005, "loss": 2.0931143760681152, "step": 7639 }, { "epoch": 0.8889923202234117, "grad_norm": 0.8804028630256653, "learning_rate": 0.0005, "loss": 2.329162836074829, "step": 7640 }, { "epoch": 0.8891086804747498, "grad_norm": 1.2052080631256104, "learning_rate": 0.0005, "loss": 1.582434892654419, "step": 7641 }, { "epoch": 0.889225040726088, "grad_norm": 1.5024408102035522, "learning_rate": 0.0005, "loss": 1.9097057580947876, "step": 7642 }, { "epoch": 0.8893414009774261, "grad_norm": 2.680724620819092, "learning_rate": 0.0005, "loss": 2.558096408843994, "step": 7643 }, { "epoch": 0.8894577612287643, "grad_norm": 2.1397509574890137, "learning_rate": 0.0005, "loss": 2.7137057781219482, "step": 7644 }, { "epoch": 0.8895741214801024, "grad_norm": 2.1871840953826904, "learning_rate": 0.0005, "loss": 1.9990921020507812, "step": 7645 }, { "epoch": 0.8896904817314405, "grad_norm": 5.773715496063232, "learning_rate": 0.0005, "loss": 2.187309741973877, "step": 7646 }, { "epoch": 0.8898068419827787, "grad_norm": 1.222832202911377, "learning_rate": 0.0005, "loss": 2.2590880393981934, "step": 7647 }, { "epoch": 0.8899232022341168, "grad_norm": 1.4895002841949463, "learning_rate": 0.0005, "loss": 3.0328872203826904, "step": 7648 }, { "epoch": 0.890039562485455, "grad_norm": 1.434665560722351, "learning_rate": 0.0005, "loss": 2.2004873752593994, "step": 7649 }, { "epoch": 0.8901559227367931, "grad_norm": 1.7937543392181396, "learning_rate": 0.0005, "loss": 2.309227228164673, "step": 7650 }, { "epoch": 0.8902722829881312, "grad_norm": 1.042385220527649, "learning_rate": 0.0005, "loss": 1.8751957416534424, "step": 7651 }, { "epoch": 0.8903886432394694, "grad_norm": 0.9196656942367554, "learning_rate": 0.0005, "loss": 2.1209094524383545, "step": 7652 }, { "epoch": 0.8905050034908075, "grad_norm": 1.4911237955093384, "learning_rate": 0.0005, "loss": 2.082942247390747, "step": 7653 }, { "epoch": 0.8906213637421457, "grad_norm": 1.3911982774734497, "learning_rate": 0.0005, "loss": 3.304647207260132, "step": 7654 }, { "epoch": 0.8907377239934838, "grad_norm": 0.8571603298187256, "learning_rate": 0.0005, "loss": 2.079939842224121, "step": 7655 }, { "epoch": 0.8908540842448219, "grad_norm": 0.9364422559738159, "learning_rate": 0.0005, "loss": 1.4415912628173828, "step": 7656 }, { "epoch": 0.8909704444961601, "grad_norm": 1.6187163591384888, "learning_rate": 0.0005, "loss": 2.850632667541504, "step": 7657 }, { "epoch": 0.8910868047474982, "grad_norm": 1.2150907516479492, "learning_rate": 0.0005, "loss": 2.609764575958252, "step": 7658 }, { "epoch": 0.8912031649988365, "grad_norm": 0.820106565952301, "learning_rate": 0.0005, "loss": 2.321582555770874, "step": 7659 }, { "epoch": 0.8913195252501745, "grad_norm": 0.33704209327697754, "learning_rate": 0.0005, "loss": 0.9417343139648438, "step": 7660 }, { "epoch": 0.8914358855015126, "grad_norm": 1.4488266706466675, "learning_rate": 0.0005, "loss": 1.8341259956359863, "step": 7661 }, { "epoch": 0.8915522457528509, "grad_norm": 1.1867504119873047, "learning_rate": 0.0005, "loss": 1.582308053970337, "step": 7662 }, { "epoch": 0.891668606004189, "grad_norm": 2.668788194656372, "learning_rate": 0.0005, "loss": 2.0015814304351807, "step": 7663 }, { "epoch": 0.8917849662555272, "grad_norm": 1.2584623098373413, "learning_rate": 0.0005, "loss": 2.2022712230682373, "step": 7664 }, { "epoch": 0.8919013265068653, "grad_norm": 1.1648211479187012, "learning_rate": 0.0005, "loss": 2.1584272384643555, "step": 7665 }, { "epoch": 0.8920176867582034, "grad_norm": 1.304031252861023, "learning_rate": 0.0005, "loss": 2.795212745666504, "step": 7666 }, { "epoch": 0.8921340470095416, "grad_norm": 1.172315239906311, "learning_rate": 0.0005, "loss": 2.0774765014648438, "step": 7667 }, { "epoch": 0.8922504072608797, "grad_norm": 1.6944899559020996, "learning_rate": 0.0005, "loss": 2.4053502082824707, "step": 7668 }, { "epoch": 0.8923667675122178, "grad_norm": 1.6467466354370117, "learning_rate": 0.0005, "loss": 2.365152359008789, "step": 7669 }, { "epoch": 0.892483127763556, "grad_norm": 1.5398633480072021, "learning_rate": 0.0005, "loss": 2.3378665447235107, "step": 7670 }, { "epoch": 0.8925994880148941, "grad_norm": 1.4344935417175293, "learning_rate": 0.0005, "loss": 3.0024139881134033, "step": 7671 }, { "epoch": 0.8927158482662323, "grad_norm": 1.6209956407546997, "learning_rate": 0.0005, "loss": 1.938354253768921, "step": 7672 }, { "epoch": 0.8928322085175704, "grad_norm": 1.067208170890808, "learning_rate": 0.0005, "loss": 1.9985425472259521, "step": 7673 }, { "epoch": 0.8929485687689085, "grad_norm": 1.4670541286468506, "learning_rate": 0.0005, "loss": 2.5415830612182617, "step": 7674 }, { "epoch": 0.8930649290202467, "grad_norm": 1.484474778175354, "learning_rate": 0.0005, "loss": 2.634573221206665, "step": 7675 }, { "epoch": 0.8931812892715848, "grad_norm": 1.2112970352172852, "learning_rate": 0.0005, "loss": 2.382870674133301, "step": 7676 }, { "epoch": 0.893297649522923, "grad_norm": 2.1048154830932617, "learning_rate": 0.0005, "loss": 2.463542938232422, "step": 7677 }, { "epoch": 0.8934140097742611, "grad_norm": 0.8986136317253113, "learning_rate": 0.0005, "loss": 2.240537405014038, "step": 7678 }, { "epoch": 0.8935303700255992, "grad_norm": 1.741166591644287, "learning_rate": 0.0005, "loss": 2.4567716121673584, "step": 7679 }, { "epoch": 0.8936467302769374, "grad_norm": 1.0781829357147217, "learning_rate": 0.0005, "loss": 2.419788360595703, "step": 7680 }, { "epoch": 0.8937630905282755, "grad_norm": 1.2632942199707031, "learning_rate": 0.0005, "loss": 2.7765445709228516, "step": 7681 }, { "epoch": 0.8938794507796137, "grad_norm": 1.0851377248764038, "learning_rate": 0.0005, "loss": 2.0873055458068848, "step": 7682 }, { "epoch": 0.8939958110309518, "grad_norm": 1.9299523830413818, "learning_rate": 0.0005, "loss": 1.56343412399292, "step": 7683 }, { "epoch": 0.8941121712822899, "grad_norm": 1.792458415031433, "learning_rate": 0.0005, "loss": 2.0444014072418213, "step": 7684 }, { "epoch": 0.8942285315336281, "grad_norm": 0.6366916298866272, "learning_rate": 0.0005, "loss": 1.8608803749084473, "step": 7685 }, { "epoch": 0.8943448917849662, "grad_norm": 0.6828328371047974, "learning_rate": 0.0005, "loss": 1.393094778060913, "step": 7686 }, { "epoch": 0.8944612520363044, "grad_norm": 1.2452219724655151, "learning_rate": 0.0005, "loss": 2.5579845905303955, "step": 7687 }, { "epoch": 0.8945776122876425, "grad_norm": 2.3818955421447754, "learning_rate": 0.0005, "loss": 2.6244266033172607, "step": 7688 }, { "epoch": 0.8946939725389806, "grad_norm": 1.3593318462371826, "learning_rate": 0.0005, "loss": 2.609127998352051, "step": 7689 }, { "epoch": 0.8948103327903189, "grad_norm": 1.1927608251571655, "learning_rate": 0.0005, "loss": 1.491858720779419, "step": 7690 }, { "epoch": 0.894926693041657, "grad_norm": 1.4794926643371582, "learning_rate": 0.0005, "loss": 2.329092025756836, "step": 7691 }, { "epoch": 0.8950430532929952, "grad_norm": 1.2129005193710327, "learning_rate": 0.0005, "loss": 1.89430570602417, "step": 7692 }, { "epoch": 0.8951594135443333, "grad_norm": 1.5095036029815674, "learning_rate": 0.0005, "loss": 2.5477490425109863, "step": 7693 }, { "epoch": 0.8952757737956714, "grad_norm": 1.8497079610824585, "learning_rate": 0.0005, "loss": 2.544752597808838, "step": 7694 }, { "epoch": 0.8953921340470096, "grad_norm": 1.8015389442443848, "learning_rate": 0.0005, "loss": 3.0106592178344727, "step": 7695 }, { "epoch": 0.8955084942983477, "grad_norm": 1.3325510025024414, "learning_rate": 0.0005, "loss": 2.3927106857299805, "step": 7696 }, { "epoch": 0.8956248545496859, "grad_norm": 1.7876619100570679, "learning_rate": 0.0005, "loss": 2.7772603034973145, "step": 7697 }, { "epoch": 0.895741214801024, "grad_norm": 0.7582190632820129, "learning_rate": 0.0005, "loss": 1.226806402206421, "step": 7698 }, { "epoch": 0.8958575750523621, "grad_norm": 1.2032396793365479, "learning_rate": 0.0005, "loss": 2.553712844848633, "step": 7699 }, { "epoch": 0.8959739353037003, "grad_norm": 1.8331615924835205, "learning_rate": 0.0005, "loss": 3.0225706100463867, "step": 7700 }, { "epoch": 0.8960902955550384, "grad_norm": 1.3381401300430298, "learning_rate": 0.0005, "loss": 2.099544048309326, "step": 7701 }, { "epoch": 0.8962066558063766, "grad_norm": 1.4375418424606323, "learning_rate": 0.0005, "loss": 2.2142691612243652, "step": 7702 }, { "epoch": 0.8963230160577147, "grad_norm": 0.9737322926521301, "learning_rate": 0.0005, "loss": 1.3805632591247559, "step": 7703 }, { "epoch": 0.8964393763090528, "grad_norm": 1.2927488088607788, "learning_rate": 0.0005, "loss": 1.6807470321655273, "step": 7704 }, { "epoch": 0.896555736560391, "grad_norm": 1.610863447189331, "learning_rate": 0.0005, "loss": 2.2609519958496094, "step": 7705 }, { "epoch": 0.8966720968117291, "grad_norm": 1.3024046421051025, "learning_rate": 0.0005, "loss": 1.789779543876648, "step": 7706 }, { "epoch": 0.8967884570630672, "grad_norm": 1.2885360717773438, "learning_rate": 0.0005, "loss": 3.0776586532592773, "step": 7707 }, { "epoch": 0.8969048173144054, "grad_norm": 1.513112187385559, "learning_rate": 0.0005, "loss": 1.5915992259979248, "step": 7708 }, { "epoch": 0.8970211775657435, "grad_norm": 1.2572146654129028, "learning_rate": 0.0005, "loss": 2.0133087635040283, "step": 7709 }, { "epoch": 0.8971375378170817, "grad_norm": 1.2159494161605835, "learning_rate": 0.0005, "loss": 2.5688118934631348, "step": 7710 }, { "epoch": 0.8972538980684198, "grad_norm": 1.2642279863357544, "learning_rate": 0.0005, "loss": 2.6157991886138916, "step": 7711 }, { "epoch": 0.8973702583197579, "grad_norm": 0.977454662322998, "learning_rate": 0.0005, "loss": 1.8443872928619385, "step": 7712 }, { "epoch": 0.8974866185710961, "grad_norm": 1.710494875907898, "learning_rate": 0.0005, "loss": 2.898120164871216, "step": 7713 }, { "epoch": 0.8976029788224342, "grad_norm": 1.670717477798462, "learning_rate": 0.0005, "loss": 2.9593663215637207, "step": 7714 }, { "epoch": 0.8977193390737724, "grad_norm": 1.5798063278198242, "learning_rate": 0.0005, "loss": 2.895726442337036, "step": 7715 }, { "epoch": 0.8978356993251105, "grad_norm": 0.8931432366371155, "learning_rate": 0.0005, "loss": 2.0106699466705322, "step": 7716 }, { "epoch": 0.8979520595764486, "grad_norm": 1.6394760608673096, "learning_rate": 0.0005, "loss": 2.943021535873413, "step": 7717 }, { "epoch": 0.8980684198277868, "grad_norm": 1.2528084516525269, "learning_rate": 0.0005, "loss": 2.743001937866211, "step": 7718 }, { "epoch": 0.898184780079125, "grad_norm": 2.021969795227051, "learning_rate": 0.0005, "loss": 2.8001608848571777, "step": 7719 }, { "epoch": 0.8983011403304632, "grad_norm": 1.1150232553482056, "learning_rate": 0.0005, "loss": 1.9169257879257202, "step": 7720 }, { "epoch": 0.8984175005818013, "grad_norm": 1.4593708515167236, "learning_rate": 0.0005, "loss": 2.333621025085449, "step": 7721 }, { "epoch": 0.8985338608331394, "grad_norm": 4.706109523773193, "learning_rate": 0.0005, "loss": 2.8206582069396973, "step": 7722 }, { "epoch": 0.8986502210844776, "grad_norm": 1.5154404640197754, "learning_rate": 0.0005, "loss": 1.7769383192062378, "step": 7723 }, { "epoch": 0.8987665813358157, "grad_norm": 2.4046270847320557, "learning_rate": 0.0005, "loss": 2.8178324699401855, "step": 7724 }, { "epoch": 0.8988829415871539, "grad_norm": 1.6977378129959106, "learning_rate": 0.0005, "loss": 2.210519552230835, "step": 7725 }, { "epoch": 0.898999301838492, "grad_norm": 1.3541258573532104, "learning_rate": 0.0005, "loss": 2.9615774154663086, "step": 7726 }, { "epoch": 0.8991156620898301, "grad_norm": 0.7388682961463928, "learning_rate": 0.0005, "loss": 1.0382118225097656, "step": 7727 }, { "epoch": 0.8992320223411683, "grad_norm": 1.6121841669082642, "learning_rate": 0.0005, "loss": 2.5216732025146484, "step": 7728 }, { "epoch": 0.8993483825925064, "grad_norm": 1.2263565063476562, "learning_rate": 0.0005, "loss": 3.194185972213745, "step": 7729 }, { "epoch": 0.8994647428438446, "grad_norm": 1.4205753803253174, "learning_rate": 0.0005, "loss": 2.713764190673828, "step": 7730 }, { "epoch": 0.8995811030951827, "grad_norm": 0.9463260769844055, "learning_rate": 0.0005, "loss": 1.6647930145263672, "step": 7731 }, { "epoch": 0.8996974633465208, "grad_norm": 1.019063115119934, "learning_rate": 0.0005, "loss": 2.2586417198181152, "step": 7732 }, { "epoch": 0.899813823597859, "grad_norm": 1.2836503982543945, "learning_rate": 0.0005, "loss": 1.7689152956008911, "step": 7733 }, { "epoch": 0.8999301838491971, "grad_norm": 1.4252876043319702, "learning_rate": 0.0005, "loss": 3.2223691940307617, "step": 7734 }, { "epoch": 0.9000465441005353, "grad_norm": 0.9318607449531555, "learning_rate": 0.0005, "loss": 1.6710282564163208, "step": 7735 }, { "epoch": 0.9001629043518734, "grad_norm": 1.3712565898895264, "learning_rate": 0.0005, "loss": 2.7299952507019043, "step": 7736 }, { "epoch": 0.9002792646032115, "grad_norm": 1.5178343057632446, "learning_rate": 0.0005, "loss": 2.556766986846924, "step": 7737 }, { "epoch": 0.9003956248545497, "grad_norm": 1.410753607749939, "learning_rate": 0.0005, "loss": 1.877881646156311, "step": 7738 }, { "epoch": 0.9005119851058878, "grad_norm": 0.9493579864501953, "learning_rate": 0.0005, "loss": 1.9536139965057373, "step": 7739 }, { "epoch": 0.900628345357226, "grad_norm": 1.6137560606002808, "learning_rate": 0.0005, "loss": 2.289862632751465, "step": 7740 }, { "epoch": 0.9007447056085641, "grad_norm": 1.0327463150024414, "learning_rate": 0.0005, "loss": 2.111292839050293, "step": 7741 }, { "epoch": 0.9008610658599022, "grad_norm": 1.2827900648117065, "learning_rate": 0.0005, "loss": 2.667017936706543, "step": 7742 }, { "epoch": 0.9009774261112404, "grad_norm": 1.099672794342041, "learning_rate": 0.0005, "loss": 1.6819161176681519, "step": 7743 }, { "epoch": 0.9010937863625785, "grad_norm": 1.0963764190673828, "learning_rate": 0.0005, "loss": 1.9790380001068115, "step": 7744 }, { "epoch": 0.9012101466139167, "grad_norm": 1.5260701179504395, "learning_rate": 0.0005, "loss": 1.988919973373413, "step": 7745 }, { "epoch": 0.9013265068652548, "grad_norm": 0.9897924661636353, "learning_rate": 0.0005, "loss": 2.0670969486236572, "step": 7746 }, { "epoch": 0.9014428671165929, "grad_norm": 1.3010462522506714, "learning_rate": 0.0005, "loss": 1.7943426370620728, "step": 7747 }, { "epoch": 0.9015592273679311, "grad_norm": 1.274890422821045, "learning_rate": 0.0005, "loss": 2.517035484313965, "step": 7748 }, { "epoch": 0.9016755876192692, "grad_norm": 1.4681559801101685, "learning_rate": 0.0005, "loss": 2.522340774536133, "step": 7749 }, { "epoch": 0.9017919478706073, "grad_norm": 1.9290223121643066, "learning_rate": 0.0005, "loss": 2.5583291053771973, "step": 7750 }, { "epoch": 0.9019083081219456, "grad_norm": 2.136167287826538, "learning_rate": 0.0005, "loss": 2.2961697578430176, "step": 7751 }, { "epoch": 0.9020246683732837, "grad_norm": 2.08320689201355, "learning_rate": 0.0005, "loss": 3.254070997238159, "step": 7752 }, { "epoch": 0.9021410286246219, "grad_norm": 1.2094444036483765, "learning_rate": 0.0005, "loss": 2.169820785522461, "step": 7753 }, { "epoch": 0.90225738887596, "grad_norm": 2.503545045852661, "learning_rate": 0.0005, "loss": 2.4733853340148926, "step": 7754 }, { "epoch": 0.9023737491272981, "grad_norm": 2.0761284828186035, "learning_rate": 0.0005, "loss": 2.047490119934082, "step": 7755 }, { "epoch": 0.9024901093786363, "grad_norm": 2.180610179901123, "learning_rate": 0.0005, "loss": 2.558295249938965, "step": 7756 }, { "epoch": 0.9026064696299744, "grad_norm": 0.9887974262237549, "learning_rate": 0.0005, "loss": 2.523120403289795, "step": 7757 }, { "epoch": 0.9027228298813126, "grad_norm": 1.4893473386764526, "learning_rate": 0.0005, "loss": 2.856152057647705, "step": 7758 }, { "epoch": 0.9028391901326507, "grad_norm": 0.9644671082496643, "learning_rate": 0.0005, "loss": 2.360637903213501, "step": 7759 }, { "epoch": 0.9029555503839888, "grad_norm": 0.6932117342948914, "learning_rate": 0.0005, "loss": 1.6822870969772339, "step": 7760 }, { "epoch": 0.903071910635327, "grad_norm": 1.712518572807312, "learning_rate": 0.0005, "loss": 2.5784873962402344, "step": 7761 }, { "epoch": 0.9031882708866651, "grad_norm": 1.1048915386199951, "learning_rate": 0.0005, "loss": 2.0239474773406982, "step": 7762 }, { "epoch": 0.9033046311380033, "grad_norm": 1.610830545425415, "learning_rate": 0.0005, "loss": 1.5662753582000732, "step": 7763 }, { "epoch": 0.9034209913893414, "grad_norm": 1.1057645082473755, "learning_rate": 0.0005, "loss": 2.1492738723754883, "step": 7764 }, { "epoch": 0.9035373516406795, "grad_norm": 1.6655524969100952, "learning_rate": 0.0005, "loss": 2.6669018268585205, "step": 7765 }, { "epoch": 0.9036537118920177, "grad_norm": 1.5596888065338135, "learning_rate": 0.0005, "loss": 2.149606227874756, "step": 7766 }, { "epoch": 0.9037700721433558, "grad_norm": 1.3916609287261963, "learning_rate": 0.0005, "loss": 2.216360092163086, "step": 7767 }, { "epoch": 0.903886432394694, "grad_norm": 1.6689940690994263, "learning_rate": 0.0005, "loss": 3.078747034072876, "step": 7768 }, { "epoch": 0.9040027926460321, "grad_norm": 1.0615670680999756, "learning_rate": 0.0005, "loss": 2.2094831466674805, "step": 7769 }, { "epoch": 0.9041191528973702, "grad_norm": 0.964573860168457, "learning_rate": 0.0005, "loss": 1.9848618507385254, "step": 7770 }, { "epoch": 0.9042355131487084, "grad_norm": 0.9135136604309082, "learning_rate": 0.0005, "loss": 1.6542692184448242, "step": 7771 }, { "epoch": 0.9043518734000465, "grad_norm": 1.956060767173767, "learning_rate": 0.0005, "loss": 1.9326584339141846, "step": 7772 }, { "epoch": 0.9044682336513847, "grad_norm": 1.117048978805542, "learning_rate": 0.0005, "loss": 1.4901847839355469, "step": 7773 }, { "epoch": 0.9045845939027228, "grad_norm": 1.2113900184631348, "learning_rate": 0.0005, "loss": 1.7024061679840088, "step": 7774 }, { "epoch": 0.9047009541540609, "grad_norm": 1.365248680114746, "learning_rate": 0.0005, "loss": 2.6941077709198, "step": 7775 }, { "epoch": 0.9048173144053991, "grad_norm": 1.1828632354736328, "learning_rate": 0.0005, "loss": 2.4152207374572754, "step": 7776 }, { "epoch": 0.9049336746567372, "grad_norm": 1.2761445045471191, "learning_rate": 0.0005, "loss": 2.578293800354004, "step": 7777 }, { "epoch": 0.9050500349080755, "grad_norm": 1.2195779085159302, "learning_rate": 0.0005, "loss": 2.2650599479675293, "step": 7778 }, { "epoch": 0.9051663951594136, "grad_norm": 1.098904013633728, "learning_rate": 0.0005, "loss": 2.265671730041504, "step": 7779 }, { "epoch": 0.9052827554107516, "grad_norm": 0.9455636739730835, "learning_rate": 0.0005, "loss": 1.94551682472229, "step": 7780 }, { "epoch": 0.9053991156620899, "grad_norm": 1.8619732856750488, "learning_rate": 0.0005, "loss": 2.471768856048584, "step": 7781 }, { "epoch": 0.905515475913428, "grad_norm": 1.0485469102859497, "learning_rate": 0.0005, "loss": 1.9606127738952637, "step": 7782 }, { "epoch": 0.9056318361647662, "grad_norm": 1.7153773307800293, "learning_rate": 0.0005, "loss": 2.862943172454834, "step": 7783 }, { "epoch": 0.9057481964161043, "grad_norm": 3.2281172275543213, "learning_rate": 0.0005, "loss": 2.472696542739868, "step": 7784 }, { "epoch": 0.9058645566674424, "grad_norm": 1.3400816917419434, "learning_rate": 0.0005, "loss": 3.115612506866455, "step": 7785 }, { "epoch": 0.9059809169187806, "grad_norm": 2.1034374237060547, "learning_rate": 0.0005, "loss": 1.6712770462036133, "step": 7786 }, { "epoch": 0.9060972771701187, "grad_norm": 1.6288315057754517, "learning_rate": 0.0005, "loss": 2.377290725708008, "step": 7787 }, { "epoch": 0.9062136374214569, "grad_norm": 1.2654024362564087, "learning_rate": 0.0005, "loss": 2.1189849376678467, "step": 7788 }, { "epoch": 0.906329997672795, "grad_norm": 1.409403920173645, "learning_rate": 0.0005, "loss": 2.443258285522461, "step": 7789 }, { "epoch": 0.9064463579241331, "grad_norm": 1.464422583580017, "learning_rate": 0.0005, "loss": 2.5469412803649902, "step": 7790 }, { "epoch": 0.9065627181754713, "grad_norm": 0.7676106095314026, "learning_rate": 0.0005, "loss": 1.991841197013855, "step": 7791 }, { "epoch": 0.9066790784268094, "grad_norm": 1.3301478624343872, "learning_rate": 0.0005, "loss": 2.1924378871917725, "step": 7792 }, { "epoch": 0.9067954386781475, "grad_norm": 0.8145026564598083, "learning_rate": 0.0005, "loss": 1.5616767406463623, "step": 7793 }, { "epoch": 0.9069117989294857, "grad_norm": 1.2461026906967163, "learning_rate": 0.0005, "loss": 2.1940784454345703, "step": 7794 }, { "epoch": 0.9070281591808238, "grad_norm": 1.2083044052124023, "learning_rate": 0.0005, "loss": 2.5786619186401367, "step": 7795 }, { "epoch": 0.907144519432162, "grad_norm": 1.3539817333221436, "learning_rate": 0.0005, "loss": 2.2450714111328125, "step": 7796 }, { "epoch": 0.9072608796835001, "grad_norm": 2.6883907318115234, "learning_rate": 0.0005, "loss": 2.3602616786956787, "step": 7797 }, { "epoch": 0.9073772399348382, "grad_norm": 3.3641440868377686, "learning_rate": 0.0005, "loss": 2.575056552886963, "step": 7798 }, { "epoch": 0.9074936001861764, "grad_norm": 1.6828571557998657, "learning_rate": 0.0005, "loss": 2.452026605606079, "step": 7799 }, { "epoch": 0.9076099604375145, "grad_norm": 1.2184853553771973, "learning_rate": 0.0005, "loss": 1.9411611557006836, "step": 7800 }, { "epoch": 0.9077263206888527, "grad_norm": 1.1310466527938843, "learning_rate": 0.0005, "loss": 1.4608161449432373, "step": 7801 }, { "epoch": 0.9078426809401908, "grad_norm": 1.305931806564331, "learning_rate": 0.0005, "loss": 2.379061698913574, "step": 7802 }, { "epoch": 0.9079590411915289, "grad_norm": 0.5682767629623413, "learning_rate": 0.0005, "loss": 1.3227289915084839, "step": 7803 }, { "epoch": 0.9080754014428671, "grad_norm": 1.4724552631378174, "learning_rate": 0.0005, "loss": 3.3158693313598633, "step": 7804 }, { "epoch": 0.9081917616942052, "grad_norm": 1.8116309642791748, "learning_rate": 0.0005, "loss": 2.2361690998077393, "step": 7805 }, { "epoch": 0.9083081219455434, "grad_norm": 1.413339376449585, "learning_rate": 0.0005, "loss": 2.205223560333252, "step": 7806 }, { "epoch": 0.9084244821968815, "grad_norm": 0.9619995355606079, "learning_rate": 0.0005, "loss": 1.3546820878982544, "step": 7807 }, { "epoch": 0.9085408424482196, "grad_norm": 1.3101413249969482, "learning_rate": 0.0005, "loss": 1.8030190467834473, "step": 7808 }, { "epoch": 0.9086572026995579, "grad_norm": 1.3495460748672485, "learning_rate": 0.0005, "loss": 1.7902086973190308, "step": 7809 }, { "epoch": 0.908773562950896, "grad_norm": 1.1567630767822266, "learning_rate": 0.0005, "loss": 2.132756471633911, "step": 7810 }, { "epoch": 0.9088899232022342, "grad_norm": 0.7884332537651062, "learning_rate": 0.0005, "loss": 2.020937204360962, "step": 7811 }, { "epoch": 0.9090062834535723, "grad_norm": 1.662343144416809, "learning_rate": 0.0005, "loss": 2.5023021697998047, "step": 7812 }, { "epoch": 0.9091226437049104, "grad_norm": 2.1301774978637695, "learning_rate": 0.0005, "loss": 2.717963695526123, "step": 7813 }, { "epoch": 0.9092390039562486, "grad_norm": 2.060157060623169, "learning_rate": 0.0005, "loss": 2.898914337158203, "step": 7814 }, { "epoch": 0.9093553642075867, "grad_norm": 1.6057766675949097, "learning_rate": 0.0005, "loss": 2.607811450958252, "step": 7815 }, { "epoch": 0.9094717244589249, "grad_norm": 0.9504224061965942, "learning_rate": 0.0005, "loss": 1.7362236976623535, "step": 7816 }, { "epoch": 0.909588084710263, "grad_norm": 1.5860202312469482, "learning_rate": 0.0005, "loss": 2.607564687728882, "step": 7817 }, { "epoch": 0.9097044449616011, "grad_norm": 1.3436251878738403, "learning_rate": 0.0005, "loss": 2.4269442558288574, "step": 7818 }, { "epoch": 0.9098208052129393, "grad_norm": 1.165657877922058, "learning_rate": 0.0005, "loss": 1.8927216529846191, "step": 7819 }, { "epoch": 0.9099371654642774, "grad_norm": 0.7947813272476196, "learning_rate": 0.0005, "loss": 1.5979735851287842, "step": 7820 }, { "epoch": 0.9100535257156156, "grad_norm": 1.2166670560836792, "learning_rate": 0.0005, "loss": 2.198632001876831, "step": 7821 }, { "epoch": 0.9101698859669537, "grad_norm": 1.7101174592971802, "learning_rate": 0.0005, "loss": 2.810184955596924, "step": 7822 }, { "epoch": 0.9102862462182918, "grad_norm": 1.0171107053756714, "learning_rate": 0.0005, "loss": 2.0847795009613037, "step": 7823 }, { "epoch": 0.91040260646963, "grad_norm": 1.0106645822525024, "learning_rate": 0.0005, "loss": 1.5463991165161133, "step": 7824 }, { "epoch": 0.9105189667209681, "grad_norm": 0.41644081473350525, "learning_rate": 0.0005, "loss": 0.9660901427268982, "step": 7825 }, { "epoch": 0.9106353269723063, "grad_norm": 1.2650762796401978, "learning_rate": 0.0005, "loss": 2.012941837310791, "step": 7826 }, { "epoch": 0.9107516872236444, "grad_norm": 2.3911325931549072, "learning_rate": 0.0005, "loss": 2.6052567958831787, "step": 7827 }, { "epoch": 0.9108680474749825, "grad_norm": 1.7502548694610596, "learning_rate": 0.0005, "loss": 2.963916540145874, "step": 7828 }, { "epoch": 0.9109844077263207, "grad_norm": 2.054952383041382, "learning_rate": 0.0005, "loss": 2.79154109954834, "step": 7829 }, { "epoch": 0.9111007679776588, "grad_norm": 1.4667819738388062, "learning_rate": 0.0005, "loss": 2.798443078994751, "step": 7830 }, { "epoch": 0.9112171282289969, "grad_norm": 0.9564381241798401, "learning_rate": 0.0005, "loss": 1.9635623693466187, "step": 7831 }, { "epoch": 0.9113334884803351, "grad_norm": 1.153086543083191, "learning_rate": 0.0005, "loss": 2.245356559753418, "step": 7832 }, { "epoch": 0.9114498487316732, "grad_norm": 1.267159104347229, "learning_rate": 0.0005, "loss": 2.510110378265381, "step": 7833 }, { "epoch": 0.9115662089830114, "grad_norm": 0.9126370549201965, "learning_rate": 0.0005, "loss": 2.5260024070739746, "step": 7834 }, { "epoch": 0.9116825692343495, "grad_norm": 1.9457497596740723, "learning_rate": 0.0005, "loss": 1.880854606628418, "step": 7835 }, { "epoch": 0.9117989294856876, "grad_norm": 1.2542705535888672, "learning_rate": 0.0005, "loss": 2.6664280891418457, "step": 7836 }, { "epoch": 0.9119152897370258, "grad_norm": 1.2880427837371826, "learning_rate": 0.0005, "loss": 1.8903837203979492, "step": 7837 }, { "epoch": 0.912031649988364, "grad_norm": 1.4931247234344482, "learning_rate": 0.0005, "loss": 2.759892463684082, "step": 7838 }, { "epoch": 0.9121480102397022, "grad_norm": 1.7082719802856445, "learning_rate": 0.0005, "loss": 2.8556511402130127, "step": 7839 }, { "epoch": 0.9122643704910403, "grad_norm": 1.3048232793807983, "learning_rate": 0.0005, "loss": 2.549225330352783, "step": 7840 }, { "epoch": 0.9123807307423784, "grad_norm": 1.0352473258972168, "learning_rate": 0.0005, "loss": 2.026124954223633, "step": 7841 }, { "epoch": 0.9124970909937166, "grad_norm": 1.5457940101623535, "learning_rate": 0.0005, "loss": 2.637765407562256, "step": 7842 }, { "epoch": 0.9126134512450547, "grad_norm": 1.5868288278579712, "learning_rate": 0.0005, "loss": 2.341646671295166, "step": 7843 }, { "epoch": 0.9127298114963929, "grad_norm": 1.4443681240081787, "learning_rate": 0.0005, "loss": 2.430232048034668, "step": 7844 }, { "epoch": 0.912846171747731, "grad_norm": 2.2599098682403564, "learning_rate": 0.0005, "loss": 2.001443862915039, "step": 7845 }, { "epoch": 0.9129625319990691, "grad_norm": 0.9373155236244202, "learning_rate": 0.0005, "loss": 1.7221837043762207, "step": 7846 }, { "epoch": 0.9130788922504073, "grad_norm": 1.1662018299102783, "learning_rate": 0.0005, "loss": 2.153010368347168, "step": 7847 }, { "epoch": 0.9131952525017454, "grad_norm": 1.6142146587371826, "learning_rate": 0.0005, "loss": 2.6163508892059326, "step": 7848 }, { "epoch": 0.9133116127530836, "grad_norm": 1.4501653909683228, "learning_rate": 0.0005, "loss": 2.9715609550476074, "step": 7849 }, { "epoch": 0.9134279730044217, "grad_norm": 1.3125818967819214, "learning_rate": 0.0005, "loss": 2.1256260871887207, "step": 7850 }, { "epoch": 0.9135443332557598, "grad_norm": 1.4377180337905884, "learning_rate": 0.0005, "loss": 2.6294174194335938, "step": 7851 }, { "epoch": 0.913660693507098, "grad_norm": 1.7066487073898315, "learning_rate": 0.0005, "loss": 2.8899452686309814, "step": 7852 }, { "epoch": 0.9137770537584361, "grad_norm": 1.235555648803711, "learning_rate": 0.0005, "loss": 2.5555827617645264, "step": 7853 }, { "epoch": 0.9138934140097743, "grad_norm": 0.8811997175216675, "learning_rate": 0.0005, "loss": 1.6276769638061523, "step": 7854 }, { "epoch": 0.9140097742611124, "grad_norm": 0.8827686905860901, "learning_rate": 0.0005, "loss": 2.1139438152313232, "step": 7855 }, { "epoch": 0.9141261345124505, "grad_norm": 1.6273890733718872, "learning_rate": 0.0005, "loss": 2.478085994720459, "step": 7856 }, { "epoch": 0.9142424947637887, "grad_norm": 1.1803492307662964, "learning_rate": 0.0005, "loss": 2.588137149810791, "step": 7857 }, { "epoch": 0.9143588550151268, "grad_norm": 1.7377103567123413, "learning_rate": 0.0005, "loss": 1.7513656616210938, "step": 7858 }, { "epoch": 0.914475215266465, "grad_norm": 1.1200133562088013, "learning_rate": 0.0005, "loss": 1.8242889642715454, "step": 7859 }, { "epoch": 0.9145915755178031, "grad_norm": 1.4206129312515259, "learning_rate": 0.0005, "loss": 2.6379554271698, "step": 7860 }, { "epoch": 0.9147079357691412, "grad_norm": 1.5959436893463135, "learning_rate": 0.0005, "loss": 2.2135984897613525, "step": 7861 }, { "epoch": 0.9148242960204794, "grad_norm": 1.7172664403915405, "learning_rate": 0.0005, "loss": 2.394996166229248, "step": 7862 }, { "epoch": 0.9149406562718175, "grad_norm": 0.7885422706604004, "learning_rate": 0.0005, "loss": 1.8437843322753906, "step": 7863 }, { "epoch": 0.9150570165231557, "grad_norm": 1.5080747604370117, "learning_rate": 0.0005, "loss": 2.1518163681030273, "step": 7864 }, { "epoch": 0.9151733767744938, "grad_norm": 1.4102715253829956, "learning_rate": 0.0005, "loss": 2.869239568710327, "step": 7865 }, { "epoch": 0.9152897370258319, "grad_norm": 1.0609835386276245, "learning_rate": 0.0005, "loss": 1.6419146060943604, "step": 7866 }, { "epoch": 0.9154060972771701, "grad_norm": 1.0410116910934448, "learning_rate": 0.0005, "loss": 2.3942744731903076, "step": 7867 }, { "epoch": 0.9155224575285082, "grad_norm": 5.010270595550537, "learning_rate": 0.0005, "loss": 2.712155818939209, "step": 7868 }, { "epoch": 0.9156388177798465, "grad_norm": 1.1551274061203003, "learning_rate": 0.0005, "loss": 1.6054913997650146, "step": 7869 }, { "epoch": 0.9157551780311846, "grad_norm": 1.4802310466766357, "learning_rate": 0.0005, "loss": 3.2224323749542236, "step": 7870 }, { "epoch": 0.9158715382825227, "grad_norm": 1.7216582298278809, "learning_rate": 0.0005, "loss": 2.789783477783203, "step": 7871 }, { "epoch": 0.9159878985338609, "grad_norm": 1.3999096155166626, "learning_rate": 0.0005, "loss": 2.394667625427246, "step": 7872 }, { "epoch": 0.916104258785199, "grad_norm": 1.544496774673462, "learning_rate": 0.0005, "loss": 2.600404977798462, "step": 7873 }, { "epoch": 0.9162206190365371, "grad_norm": 1.3317385911941528, "learning_rate": 0.0005, "loss": 3.126887798309326, "step": 7874 }, { "epoch": 0.9163369792878753, "grad_norm": 0.7745746374130249, "learning_rate": 0.0005, "loss": 1.6938154697418213, "step": 7875 }, { "epoch": 0.9164533395392134, "grad_norm": 1.3023412227630615, "learning_rate": 0.0005, "loss": 2.1091055870056152, "step": 7876 }, { "epoch": 0.9165696997905516, "grad_norm": 1.0736613273620605, "learning_rate": 0.0005, "loss": 2.021300792694092, "step": 7877 }, { "epoch": 0.9166860600418897, "grad_norm": 1.0074118375778198, "learning_rate": 0.0005, "loss": 2.4508161544799805, "step": 7878 }, { "epoch": 0.9168024202932278, "grad_norm": 1.298190951347351, "learning_rate": 0.0005, "loss": 2.403280735015869, "step": 7879 }, { "epoch": 0.916918780544566, "grad_norm": 0.9747894406318665, "learning_rate": 0.0005, "loss": 2.0294501781463623, "step": 7880 }, { "epoch": 0.9170351407959041, "grad_norm": 1.683060884475708, "learning_rate": 0.0005, "loss": 2.3799760341644287, "step": 7881 }, { "epoch": 0.9171515010472423, "grad_norm": 1.4742499589920044, "learning_rate": 0.0005, "loss": 2.63266921043396, "step": 7882 }, { "epoch": 0.9172678612985804, "grad_norm": 1.2447148561477661, "learning_rate": 0.0005, "loss": 1.2498807907104492, "step": 7883 }, { "epoch": 0.9173842215499185, "grad_norm": 1.5221344232559204, "learning_rate": 0.0005, "loss": 2.239020347595215, "step": 7884 }, { "epoch": 0.9175005818012567, "grad_norm": 1.155176043510437, "learning_rate": 0.0005, "loss": 1.6302651166915894, "step": 7885 }, { "epoch": 0.9176169420525948, "grad_norm": 1.415170431137085, "learning_rate": 0.0005, "loss": 1.8990538120269775, "step": 7886 }, { "epoch": 0.917733302303933, "grad_norm": 0.5588191747665405, "learning_rate": 0.0005, "loss": 0.9959374666213989, "step": 7887 }, { "epoch": 0.9178496625552711, "grad_norm": 1.4285645484924316, "learning_rate": 0.0005, "loss": 1.7942259311676025, "step": 7888 }, { "epoch": 0.9179660228066092, "grad_norm": 1.584407091140747, "learning_rate": 0.0005, "loss": 2.5648179054260254, "step": 7889 }, { "epoch": 0.9180823830579474, "grad_norm": 1.2176965475082397, "learning_rate": 0.0005, "loss": 2.2022650241851807, "step": 7890 }, { "epoch": 0.9181987433092855, "grad_norm": 1.169215440750122, "learning_rate": 0.0005, "loss": 1.837673306465149, "step": 7891 }, { "epoch": 0.9183151035606237, "grad_norm": 1.6591061353683472, "learning_rate": 0.0005, "loss": 2.7872490882873535, "step": 7892 }, { "epoch": 0.9184314638119618, "grad_norm": 0.9041929841041565, "learning_rate": 0.0005, "loss": 1.663388967514038, "step": 7893 }, { "epoch": 0.9185478240632999, "grad_norm": 2.9507765769958496, "learning_rate": 0.0005, "loss": 2.112616539001465, "step": 7894 }, { "epoch": 0.9186641843146381, "grad_norm": 1.7273930311203003, "learning_rate": 0.0005, "loss": 2.4978384971618652, "step": 7895 }, { "epoch": 0.9187805445659762, "grad_norm": 1.0550800561904907, "learning_rate": 0.0005, "loss": 1.5190156698226929, "step": 7896 }, { "epoch": 0.9188969048173145, "grad_norm": 2.0748212337493896, "learning_rate": 0.0005, "loss": 2.839153289794922, "step": 7897 }, { "epoch": 0.9190132650686526, "grad_norm": 1.3227111101150513, "learning_rate": 0.0005, "loss": 2.0869946479797363, "step": 7898 }, { "epoch": 0.9191296253199907, "grad_norm": 1.1659348011016846, "learning_rate": 0.0005, "loss": 1.99351966381073, "step": 7899 }, { "epoch": 0.9192459855713289, "grad_norm": 1.0951869487762451, "learning_rate": 0.0005, "loss": 3.068171977996826, "step": 7900 }, { "epoch": 0.919362345822667, "grad_norm": 1.222900629043579, "learning_rate": 0.0005, "loss": 2.0099070072174072, "step": 7901 }, { "epoch": 0.9194787060740052, "grad_norm": 1.4045213460922241, "learning_rate": 0.0005, "loss": 2.7941298484802246, "step": 7902 }, { "epoch": 0.9195950663253433, "grad_norm": 1.0433682203292847, "learning_rate": 0.0005, "loss": 1.4857884645462036, "step": 7903 }, { "epoch": 0.9197114265766814, "grad_norm": 1.1134611368179321, "learning_rate": 0.0005, "loss": 2.016007900238037, "step": 7904 }, { "epoch": 0.9198277868280196, "grad_norm": 1.3766201734542847, "learning_rate": 0.0005, "loss": 2.3197519779205322, "step": 7905 }, { "epoch": 0.9199441470793577, "grad_norm": 1.6175565719604492, "learning_rate": 0.0005, "loss": 2.635725498199463, "step": 7906 }, { "epoch": 0.9200605073306959, "grad_norm": 0.9244587421417236, "learning_rate": 0.0005, "loss": 1.8892738819122314, "step": 7907 }, { "epoch": 0.920176867582034, "grad_norm": 1.295203685760498, "learning_rate": 0.0005, "loss": 2.4943647384643555, "step": 7908 }, { "epoch": 0.9202932278333721, "grad_norm": 1.2980422973632812, "learning_rate": 0.0005, "loss": 2.8164918422698975, "step": 7909 }, { "epoch": 0.9204095880847103, "grad_norm": 1.8328222036361694, "learning_rate": 0.0005, "loss": 3.351097583770752, "step": 7910 }, { "epoch": 0.9205259483360484, "grad_norm": 1.302154541015625, "learning_rate": 0.0005, "loss": 2.670520782470703, "step": 7911 }, { "epoch": 0.9206423085873866, "grad_norm": 1.6182756423950195, "learning_rate": 0.0005, "loss": 1.737347960472107, "step": 7912 }, { "epoch": 0.9207586688387247, "grad_norm": 1.9272401332855225, "learning_rate": 0.0005, "loss": 2.1443724632263184, "step": 7913 }, { "epoch": 0.9208750290900628, "grad_norm": 1.4737545251846313, "learning_rate": 0.0005, "loss": 2.4565367698669434, "step": 7914 }, { "epoch": 0.920991389341401, "grad_norm": 1.468254804611206, "learning_rate": 0.0005, "loss": 2.2701122760772705, "step": 7915 }, { "epoch": 0.9211077495927391, "grad_norm": 1.157025933265686, "learning_rate": 0.0005, "loss": 2.225588321685791, "step": 7916 }, { "epoch": 0.9212241098440772, "grad_norm": 0.9220951795578003, "learning_rate": 0.0005, "loss": 2.0793943405151367, "step": 7917 }, { "epoch": 0.9213404700954154, "grad_norm": 1.435103416442871, "learning_rate": 0.0005, "loss": 1.6521605253219604, "step": 7918 }, { "epoch": 0.9214568303467535, "grad_norm": 0.9067822098731995, "learning_rate": 0.0005, "loss": 2.161710739135742, "step": 7919 }, { "epoch": 0.9215731905980917, "grad_norm": 1.1046348810195923, "learning_rate": 0.0005, "loss": 1.8821552991867065, "step": 7920 }, { "epoch": 0.9216895508494298, "grad_norm": 1.380983591079712, "learning_rate": 0.0005, "loss": 2.3051583766937256, "step": 7921 }, { "epoch": 0.9218059111007679, "grad_norm": 1.1929126977920532, "learning_rate": 0.0005, "loss": 2.514782428741455, "step": 7922 }, { "epoch": 0.9219222713521061, "grad_norm": 1.0344158411026, "learning_rate": 0.0005, "loss": 2.3907413482666016, "step": 7923 }, { "epoch": 0.9220386316034442, "grad_norm": 0.9366905093193054, "learning_rate": 0.0005, "loss": 1.968892216682434, "step": 7924 }, { "epoch": 0.9221549918547824, "grad_norm": 1.0016310214996338, "learning_rate": 0.0005, "loss": 1.3674347400665283, "step": 7925 }, { "epoch": 0.9222713521061205, "grad_norm": 1.3224754333496094, "learning_rate": 0.0005, "loss": 2.6073873043060303, "step": 7926 }, { "epoch": 0.9223877123574586, "grad_norm": 1.3024834394454956, "learning_rate": 0.0005, "loss": 2.0384020805358887, "step": 7927 }, { "epoch": 0.9225040726087969, "grad_norm": 1.3412392139434814, "learning_rate": 0.0005, "loss": 2.392632007598877, "step": 7928 }, { "epoch": 0.922620432860135, "grad_norm": 1.9718292951583862, "learning_rate": 0.0005, "loss": 3.5621137619018555, "step": 7929 }, { "epoch": 0.9227367931114732, "grad_norm": 1.4455864429473877, "learning_rate": 0.0005, "loss": 2.2435379028320312, "step": 7930 }, { "epoch": 0.9228531533628113, "grad_norm": 0.739625096321106, "learning_rate": 0.0005, "loss": 1.4438492059707642, "step": 7931 }, { "epoch": 0.9229695136141494, "grad_norm": 0.8028134703636169, "learning_rate": 0.0005, "loss": 1.5121688842773438, "step": 7932 }, { "epoch": 0.9230858738654876, "grad_norm": 1.3385062217712402, "learning_rate": 0.0005, "loss": 2.1771388053894043, "step": 7933 }, { "epoch": 0.9232022341168257, "grad_norm": 1.562947392463684, "learning_rate": 0.0005, "loss": 2.3517394065856934, "step": 7934 }, { "epoch": 0.9233185943681639, "grad_norm": 1.4997247457504272, "learning_rate": 0.0005, "loss": 2.258352279663086, "step": 7935 }, { "epoch": 0.923434954619502, "grad_norm": 1.4728261232376099, "learning_rate": 0.0005, "loss": 1.6051030158996582, "step": 7936 }, { "epoch": 0.9235513148708401, "grad_norm": 0.7201387286186218, "learning_rate": 0.0005, "loss": 1.7774616479873657, "step": 7937 }, { "epoch": 0.9236676751221783, "grad_norm": 1.1027393341064453, "learning_rate": 0.0005, "loss": 2.051197052001953, "step": 7938 }, { "epoch": 0.9237840353735164, "grad_norm": 1.7573504447937012, "learning_rate": 0.0005, "loss": 2.6965718269348145, "step": 7939 }, { "epoch": 0.9239003956248546, "grad_norm": 1.1371808052062988, "learning_rate": 0.0005, "loss": 2.015845775604248, "step": 7940 }, { "epoch": 0.9240167558761927, "grad_norm": 1.1987591981887817, "learning_rate": 0.0005, "loss": 1.7752023935317993, "step": 7941 }, { "epoch": 0.9241331161275308, "grad_norm": 1.640711784362793, "learning_rate": 0.0005, "loss": 2.9789373874664307, "step": 7942 }, { "epoch": 0.924249476378869, "grad_norm": 0.9468284249305725, "learning_rate": 0.0005, "loss": 1.3635408878326416, "step": 7943 }, { "epoch": 0.9243658366302071, "grad_norm": 1.6185044050216675, "learning_rate": 0.0005, "loss": 2.3175699710845947, "step": 7944 }, { "epoch": 0.9244821968815453, "grad_norm": 1.3276090621948242, "learning_rate": 0.0005, "loss": 1.8894535303115845, "step": 7945 }, { "epoch": 0.9245985571328834, "grad_norm": 1.4153269529342651, "learning_rate": 0.0005, "loss": 2.1267309188842773, "step": 7946 }, { "epoch": 0.9247149173842215, "grad_norm": 1.0332897901535034, "learning_rate": 0.0005, "loss": 2.0734262466430664, "step": 7947 }, { "epoch": 0.9248312776355597, "grad_norm": 0.5670360326766968, "learning_rate": 0.0005, "loss": 1.3504676818847656, "step": 7948 }, { "epoch": 0.9249476378868978, "grad_norm": 1.259696125984192, "learning_rate": 0.0005, "loss": 1.8972495794296265, "step": 7949 }, { "epoch": 0.925063998138236, "grad_norm": 1.6681650876998901, "learning_rate": 0.0005, "loss": 2.5474724769592285, "step": 7950 }, { "epoch": 0.9251803583895741, "grad_norm": 1.2674554586410522, "learning_rate": 0.0005, "loss": 2.7576398849487305, "step": 7951 }, { "epoch": 0.9252967186409122, "grad_norm": 1.4361062049865723, "learning_rate": 0.0005, "loss": 2.3114492893218994, "step": 7952 }, { "epoch": 0.9254130788922504, "grad_norm": 1.4006232023239136, "learning_rate": 0.0005, "loss": 3.085853338241577, "step": 7953 }, { "epoch": 0.9255294391435885, "grad_norm": 2.1292059421539307, "learning_rate": 0.0005, "loss": 1.656507968902588, "step": 7954 }, { "epoch": 0.9256457993949266, "grad_norm": 1.0939292907714844, "learning_rate": 0.0005, "loss": 1.1117134094238281, "step": 7955 }, { "epoch": 0.9257621596462648, "grad_norm": 1.7763365507125854, "learning_rate": 0.0005, "loss": 2.4132330417633057, "step": 7956 }, { "epoch": 0.925878519897603, "grad_norm": 1.9848698377609253, "learning_rate": 0.0005, "loss": 2.444822072982788, "step": 7957 }, { "epoch": 0.9259948801489412, "grad_norm": 1.5796507596969604, "learning_rate": 0.0005, "loss": 2.2331762313842773, "step": 7958 }, { "epoch": 0.9261112404002793, "grad_norm": 2.0298006534576416, "learning_rate": 0.0005, "loss": 3.0356428623199463, "step": 7959 }, { "epoch": 0.9262276006516174, "grad_norm": 1.6211832761764526, "learning_rate": 0.0005, "loss": 2.37406325340271, "step": 7960 }, { "epoch": 0.9263439609029556, "grad_norm": 1.402051329612732, "learning_rate": 0.0005, "loss": 1.618458867073059, "step": 7961 }, { "epoch": 0.9264603211542937, "grad_norm": 1.5651153326034546, "learning_rate": 0.0005, "loss": 1.793090581893921, "step": 7962 }, { "epoch": 0.9265766814056319, "grad_norm": 1.4541511535644531, "learning_rate": 0.0005, "loss": 2.492798328399658, "step": 7963 }, { "epoch": 0.92669304165697, "grad_norm": 1.831954002380371, "learning_rate": 0.0005, "loss": 2.294468641281128, "step": 7964 }, { "epoch": 0.9268094019083081, "grad_norm": 1.0062834024429321, "learning_rate": 0.0005, "loss": 1.9241845607757568, "step": 7965 }, { "epoch": 0.9269257621596463, "grad_norm": 1.4209301471710205, "learning_rate": 0.0005, "loss": 1.9262844324111938, "step": 7966 }, { "epoch": 0.9270421224109844, "grad_norm": 1.0525826215744019, "learning_rate": 0.0005, "loss": 1.3758095502853394, "step": 7967 }, { "epoch": 0.9271584826623226, "grad_norm": 2.2169480323791504, "learning_rate": 0.0005, "loss": 3.0660505294799805, "step": 7968 }, { "epoch": 0.9272748429136607, "grad_norm": 1.1179267168045044, "learning_rate": 0.0005, "loss": 2.3700807094573975, "step": 7969 }, { "epoch": 0.9273912031649988, "grad_norm": 1.0164048671722412, "learning_rate": 0.0005, "loss": 2.3286688327789307, "step": 7970 }, { "epoch": 0.927507563416337, "grad_norm": 0.9436336755752563, "learning_rate": 0.0005, "loss": 1.5322386026382446, "step": 7971 }, { "epoch": 0.9276239236676751, "grad_norm": 0.9265375137329102, "learning_rate": 0.0005, "loss": 2.21825909614563, "step": 7972 }, { "epoch": 0.9277402839190133, "grad_norm": 1.2922953367233276, "learning_rate": 0.0005, "loss": 2.059060573577881, "step": 7973 }, { "epoch": 0.9278566441703514, "grad_norm": 1.3632869720458984, "learning_rate": 0.0005, "loss": 3.1313271522521973, "step": 7974 }, { "epoch": 0.9279730044216895, "grad_norm": 0.9602912664413452, "learning_rate": 0.0005, "loss": 2.136303186416626, "step": 7975 }, { "epoch": 0.9280893646730277, "grad_norm": 1.1103312969207764, "learning_rate": 0.0005, "loss": 2.0771851539611816, "step": 7976 }, { "epoch": 0.9282057249243658, "grad_norm": 2.3041632175445557, "learning_rate": 0.0005, "loss": 2.252229928970337, "step": 7977 }, { "epoch": 0.928322085175704, "grad_norm": 1.4265375137329102, "learning_rate": 0.0005, "loss": 1.8496999740600586, "step": 7978 }, { "epoch": 0.9284384454270421, "grad_norm": 1.5993640422821045, "learning_rate": 0.0005, "loss": 1.7988696098327637, "step": 7979 }, { "epoch": 0.9285548056783802, "grad_norm": 0.8760966658592224, "learning_rate": 0.0005, "loss": 2.083573579788208, "step": 7980 }, { "epoch": 0.9286711659297184, "grad_norm": 1.735673189163208, "learning_rate": 0.0005, "loss": 1.9350943565368652, "step": 7981 }, { "epoch": 0.9287875261810565, "grad_norm": 1.0909628868103027, "learning_rate": 0.0005, "loss": 2.1469857692718506, "step": 7982 }, { "epoch": 0.9289038864323947, "grad_norm": 1.4960325956344604, "learning_rate": 0.0005, "loss": 1.918235182762146, "step": 7983 }, { "epoch": 0.9290202466837328, "grad_norm": 1.1457828283309937, "learning_rate": 0.0005, "loss": 2.2201087474823, "step": 7984 }, { "epoch": 0.9291366069350709, "grad_norm": 0.777927577495575, "learning_rate": 0.0005, "loss": 2.0248606204986572, "step": 7985 }, { "epoch": 0.9292529671864092, "grad_norm": 1.1064690351486206, "learning_rate": 0.0005, "loss": 1.2809985876083374, "step": 7986 }, { "epoch": 0.9293693274377473, "grad_norm": 5.34758186340332, "learning_rate": 0.0005, "loss": 2.474663734436035, "step": 7987 }, { "epoch": 0.9294856876890855, "grad_norm": 1.2782727479934692, "learning_rate": 0.0005, "loss": 2.5541484355926514, "step": 7988 }, { "epoch": 0.9296020479404236, "grad_norm": 1.7378790378570557, "learning_rate": 0.0005, "loss": 1.9350955486297607, "step": 7989 }, { "epoch": 0.9297184081917617, "grad_norm": 1.5421085357666016, "learning_rate": 0.0005, "loss": 2.1700572967529297, "step": 7990 }, { "epoch": 0.9298347684430999, "grad_norm": 0.4986864924430847, "learning_rate": 0.0005, "loss": 1.1417315006256104, "step": 7991 }, { "epoch": 0.929951128694438, "grad_norm": 1.0833475589752197, "learning_rate": 0.0005, "loss": 1.9874229431152344, "step": 7992 }, { "epoch": 0.9300674889457762, "grad_norm": 2.1489529609680176, "learning_rate": 0.0005, "loss": 2.7426791191101074, "step": 7993 }, { "epoch": 0.9301838491971143, "grad_norm": 0.9612453579902649, "learning_rate": 0.0005, "loss": 1.7184557914733887, "step": 7994 }, { "epoch": 0.9303002094484524, "grad_norm": 1.1211706399917603, "learning_rate": 0.0005, "loss": 1.9768770933151245, "step": 7995 }, { "epoch": 0.9304165696997906, "grad_norm": 1.23808753490448, "learning_rate": 0.0005, "loss": 2.492774724960327, "step": 7996 }, { "epoch": 0.9305329299511287, "grad_norm": 1.5497349500656128, "learning_rate": 0.0005, "loss": 2.0618884563446045, "step": 7997 }, { "epoch": 0.9306492902024668, "grad_norm": 5.064155578613281, "learning_rate": 0.0005, "loss": 1.9453555345535278, "step": 7998 }, { "epoch": 0.930765650453805, "grad_norm": 2.059150218963623, "learning_rate": 0.0005, "loss": 2.9128050804138184, "step": 7999 }, { "epoch": 0.9308820107051431, "grad_norm": 1.168267011642456, "learning_rate": 0.0005, "loss": 1.9447160959243774, "step": 8000 }, { "epoch": 0.9309983709564813, "grad_norm": 1.2287940979003906, "learning_rate": 0.0005, "loss": 2.3155696392059326, "step": 8001 }, { "epoch": 0.9311147312078194, "grad_norm": 1.0088343620300293, "learning_rate": 0.0005, "loss": 1.738856315612793, "step": 8002 }, { "epoch": 0.9312310914591575, "grad_norm": 1.3859232664108276, "learning_rate": 0.0005, "loss": 2.480546712875366, "step": 8003 }, { "epoch": 0.9313474517104957, "grad_norm": 1.8519229888916016, "learning_rate": 0.0005, "loss": 3.45658016204834, "step": 8004 }, { "epoch": 0.9314638119618338, "grad_norm": 2.0722997188568115, "learning_rate": 0.0005, "loss": 2.9777755737304688, "step": 8005 }, { "epoch": 0.931580172213172, "grad_norm": 2.3196802139282227, "learning_rate": 0.0005, "loss": 2.7046751976013184, "step": 8006 }, { "epoch": 0.9316965324645101, "grad_norm": 1.24957275390625, "learning_rate": 0.0005, "loss": 2.4473447799682617, "step": 8007 }, { "epoch": 0.9318128927158482, "grad_norm": 1.2787144184112549, "learning_rate": 0.0005, "loss": 2.731079339981079, "step": 8008 }, { "epoch": 0.9319292529671864, "grad_norm": 2.0119469165802, "learning_rate": 0.0005, "loss": 3.4203531742095947, "step": 8009 }, { "epoch": 0.9320456132185245, "grad_norm": 1.9253424406051636, "learning_rate": 0.0005, "loss": 2.5563220977783203, "step": 8010 }, { "epoch": 0.9321619734698627, "grad_norm": 1.6343131065368652, "learning_rate": 0.0005, "loss": 2.576709270477295, "step": 8011 }, { "epoch": 0.9322783337212008, "grad_norm": 1.1225422620773315, "learning_rate": 0.0005, "loss": 2.4081857204437256, "step": 8012 }, { "epoch": 0.9323946939725389, "grad_norm": 1.7140882015228271, "learning_rate": 0.0005, "loss": 2.896088123321533, "step": 8013 }, { "epoch": 0.9325110542238771, "grad_norm": 1.2709155082702637, "learning_rate": 0.0005, "loss": 2.1097583770751953, "step": 8014 }, { "epoch": 0.9326274144752152, "grad_norm": 1.3108773231506348, "learning_rate": 0.0005, "loss": 3.436774730682373, "step": 8015 }, { "epoch": 0.9327437747265535, "grad_norm": 0.7744138836860657, "learning_rate": 0.0005, "loss": 1.1239250898361206, "step": 8016 }, { "epoch": 0.9328601349778916, "grad_norm": 1.804494857788086, "learning_rate": 0.0005, "loss": 3.2447967529296875, "step": 8017 }, { "epoch": 0.9329764952292297, "grad_norm": 1.4839116334915161, "learning_rate": 0.0005, "loss": 2.2716503143310547, "step": 8018 }, { "epoch": 0.9330928554805679, "grad_norm": 1.4953453540802002, "learning_rate": 0.0005, "loss": 2.2103023529052734, "step": 8019 }, { "epoch": 0.933209215731906, "grad_norm": 1.3089699745178223, "learning_rate": 0.0005, "loss": 2.2724967002868652, "step": 8020 }, { "epoch": 0.9333255759832442, "grad_norm": 1.5596482753753662, "learning_rate": 0.0005, "loss": 3.0825393199920654, "step": 8021 }, { "epoch": 0.9334419362345823, "grad_norm": 3.6082606315612793, "learning_rate": 0.0005, "loss": 2.314466953277588, "step": 8022 }, { "epoch": 0.9335582964859204, "grad_norm": 1.9218778610229492, "learning_rate": 0.0005, "loss": 2.9786458015441895, "step": 8023 }, { "epoch": 0.9336746567372586, "grad_norm": 1.162336826324463, "learning_rate": 0.0005, "loss": 2.2055411338806152, "step": 8024 }, { "epoch": 0.9337910169885967, "grad_norm": 1.015403389930725, "learning_rate": 0.0005, "loss": 1.8615045547485352, "step": 8025 }, { "epoch": 0.9339073772399349, "grad_norm": 1.436041235923767, "learning_rate": 0.0005, "loss": 2.577018976211548, "step": 8026 }, { "epoch": 0.934023737491273, "grad_norm": 1.3035471439361572, "learning_rate": 0.0005, "loss": 2.523040294647217, "step": 8027 }, { "epoch": 0.9341400977426111, "grad_norm": 2.8291995525360107, "learning_rate": 0.0005, "loss": 2.3473398685455322, "step": 8028 }, { "epoch": 0.9342564579939493, "grad_norm": 1.491720199584961, "learning_rate": 0.0005, "loss": 1.7853379249572754, "step": 8029 }, { "epoch": 0.9343728182452874, "grad_norm": 1.2213528156280518, "learning_rate": 0.0005, "loss": 2.696871757507324, "step": 8030 }, { "epoch": 0.9344891784966256, "grad_norm": 1.3288979530334473, "learning_rate": 0.0005, "loss": 2.4911231994628906, "step": 8031 }, { "epoch": 0.9346055387479637, "grad_norm": 0.9525869488716125, "learning_rate": 0.0005, "loss": 2.594578504562378, "step": 8032 }, { "epoch": 0.9347218989993018, "grad_norm": 1.6778039932250977, "learning_rate": 0.0005, "loss": 3.369734764099121, "step": 8033 }, { "epoch": 0.93483825925064, "grad_norm": 1.5928691625595093, "learning_rate": 0.0005, "loss": 2.136568307876587, "step": 8034 }, { "epoch": 0.9349546195019781, "grad_norm": 1.4038749933242798, "learning_rate": 0.0005, "loss": 2.8653523921966553, "step": 8035 }, { "epoch": 0.9350709797533163, "grad_norm": 1.0125408172607422, "learning_rate": 0.0005, "loss": 2.6200079917907715, "step": 8036 }, { "epoch": 0.9351873400046544, "grad_norm": 1.6465785503387451, "learning_rate": 0.0005, "loss": 2.4553310871124268, "step": 8037 }, { "epoch": 0.9353037002559925, "grad_norm": 1.1951411962509155, "learning_rate": 0.0005, "loss": 2.001246929168701, "step": 8038 }, { "epoch": 0.9354200605073307, "grad_norm": 2.034644603729248, "learning_rate": 0.0005, "loss": 2.578768253326416, "step": 8039 }, { "epoch": 0.9355364207586688, "grad_norm": 1.234755039215088, "learning_rate": 0.0005, "loss": 2.264418840408325, "step": 8040 }, { "epoch": 0.9356527810100069, "grad_norm": 1.3806818723678589, "learning_rate": 0.0005, "loss": 2.4578351974487305, "step": 8041 }, { "epoch": 0.9357691412613451, "grad_norm": 1.5250245332717896, "learning_rate": 0.0005, "loss": 3.1222596168518066, "step": 8042 }, { "epoch": 0.9358855015126832, "grad_norm": 1.7893788814544678, "learning_rate": 0.0005, "loss": 2.261129856109619, "step": 8043 }, { "epoch": 0.9360018617640214, "grad_norm": 1.42664635181427, "learning_rate": 0.0005, "loss": 2.8126213550567627, "step": 8044 }, { "epoch": 0.9361182220153595, "grad_norm": 1.1467779874801636, "learning_rate": 0.0005, "loss": 2.254837989807129, "step": 8045 }, { "epoch": 0.9362345822666976, "grad_norm": 1.5349225997924805, "learning_rate": 0.0005, "loss": 2.439302682876587, "step": 8046 }, { "epoch": 0.9363509425180359, "grad_norm": 1.1708660125732422, "learning_rate": 0.0005, "loss": 2.828420639038086, "step": 8047 }, { "epoch": 0.936467302769374, "grad_norm": 3.2474892139434814, "learning_rate": 0.0005, "loss": 2.6335818767547607, "step": 8048 }, { "epoch": 0.9365836630207122, "grad_norm": 1.5289669036865234, "learning_rate": 0.0005, "loss": 2.427675247192383, "step": 8049 }, { "epoch": 0.9367000232720503, "grad_norm": 1.5959579944610596, "learning_rate": 0.0005, "loss": 3.2012484073638916, "step": 8050 }, { "epoch": 0.9368163835233884, "grad_norm": 0.9074914455413818, "learning_rate": 0.0005, "loss": 1.8876256942749023, "step": 8051 }, { "epoch": 0.9369327437747266, "grad_norm": 1.2385284900665283, "learning_rate": 0.0005, "loss": 2.3072402477264404, "step": 8052 }, { "epoch": 0.9370491040260647, "grad_norm": 1.389464020729065, "learning_rate": 0.0005, "loss": 2.098609447479248, "step": 8053 }, { "epoch": 0.9371654642774029, "grad_norm": 1.2359110116958618, "learning_rate": 0.0005, "loss": 1.8525029420852661, "step": 8054 }, { "epoch": 0.937281824528741, "grad_norm": 1.1281852722167969, "learning_rate": 0.0005, "loss": 2.173699140548706, "step": 8055 }, { "epoch": 0.9373981847800791, "grad_norm": 1.6058903932571411, "learning_rate": 0.0005, "loss": 2.6948492527008057, "step": 8056 }, { "epoch": 0.9375145450314173, "grad_norm": 1.52690851688385, "learning_rate": 0.0005, "loss": 2.5095577239990234, "step": 8057 }, { "epoch": 0.9376309052827554, "grad_norm": 1.9776264429092407, "learning_rate": 0.0005, "loss": 2.8522090911865234, "step": 8058 }, { "epoch": 0.9377472655340936, "grad_norm": 1.057157039642334, "learning_rate": 0.0005, "loss": 2.4933900833129883, "step": 8059 }, { "epoch": 0.9378636257854317, "grad_norm": 1.3994266986846924, "learning_rate": 0.0005, "loss": 2.7054171562194824, "step": 8060 }, { "epoch": 0.9379799860367698, "grad_norm": 1.818786859512329, "learning_rate": 0.0005, "loss": 2.7896065711975098, "step": 8061 }, { "epoch": 0.938096346288108, "grad_norm": 1.8734581470489502, "learning_rate": 0.0005, "loss": 2.320857048034668, "step": 8062 }, { "epoch": 0.9382127065394461, "grad_norm": 1.6564141511917114, "learning_rate": 0.0005, "loss": 1.9913554191589355, "step": 8063 }, { "epoch": 0.9383290667907843, "grad_norm": 0.5537933707237244, "learning_rate": 0.0005, "loss": 0.9790866374969482, "step": 8064 }, { "epoch": 0.9384454270421224, "grad_norm": 1.3738281726837158, "learning_rate": 0.0005, "loss": 2.0174195766448975, "step": 8065 }, { "epoch": 0.9385617872934605, "grad_norm": 1.1667003631591797, "learning_rate": 0.0005, "loss": 2.083500385284424, "step": 8066 }, { "epoch": 0.9386781475447987, "grad_norm": 1.1265079975128174, "learning_rate": 0.0005, "loss": 2.1154232025146484, "step": 8067 }, { "epoch": 0.9387945077961368, "grad_norm": 1.541893482208252, "learning_rate": 0.0005, "loss": 1.4545600414276123, "step": 8068 }, { "epoch": 0.938910868047475, "grad_norm": 0.6318364143371582, "learning_rate": 0.0005, "loss": 1.2141821384429932, "step": 8069 }, { "epoch": 0.9390272282988131, "grad_norm": 1.157607913017273, "learning_rate": 0.0005, "loss": 2.3440356254577637, "step": 8070 }, { "epoch": 0.9391435885501512, "grad_norm": 1.127759337425232, "learning_rate": 0.0005, "loss": 2.42446231842041, "step": 8071 }, { "epoch": 0.9392599488014894, "grad_norm": 1.4028942584991455, "learning_rate": 0.0005, "loss": 2.154669761657715, "step": 8072 }, { "epoch": 0.9393763090528275, "grad_norm": 0.9851990938186646, "learning_rate": 0.0005, "loss": 2.1659445762634277, "step": 8073 }, { "epoch": 0.9394926693041658, "grad_norm": 1.3776366710662842, "learning_rate": 0.0005, "loss": 2.7160816192626953, "step": 8074 }, { "epoch": 0.9396090295555038, "grad_norm": 1.5348029136657715, "learning_rate": 0.0005, "loss": 1.8756624460220337, "step": 8075 }, { "epoch": 0.939725389806842, "grad_norm": 1.8056890964508057, "learning_rate": 0.0005, "loss": 2.8196816444396973, "step": 8076 }, { "epoch": 0.9398417500581802, "grad_norm": 1.3427270650863647, "learning_rate": 0.0005, "loss": 2.7638516426086426, "step": 8077 }, { "epoch": 0.9399581103095183, "grad_norm": 0.8360802531242371, "learning_rate": 0.0005, "loss": 1.5013742446899414, "step": 8078 }, { "epoch": 0.9400744705608564, "grad_norm": 1.547441840171814, "learning_rate": 0.0005, "loss": 2.505423069000244, "step": 8079 }, { "epoch": 0.9401908308121946, "grad_norm": 1.088336706161499, "learning_rate": 0.0005, "loss": 2.5997133255004883, "step": 8080 }, { "epoch": 0.9403071910635327, "grad_norm": 1.248031735420227, "learning_rate": 0.0005, "loss": 1.46366548538208, "step": 8081 }, { "epoch": 0.9404235513148709, "grad_norm": 1.114066243171692, "learning_rate": 0.0005, "loss": 3.146364212036133, "step": 8082 }, { "epoch": 0.940539911566209, "grad_norm": 2.192082643508911, "learning_rate": 0.0005, "loss": 3.2203927040100098, "step": 8083 }, { "epoch": 0.9406562718175471, "grad_norm": 1.4742646217346191, "learning_rate": 0.0005, "loss": 2.509246349334717, "step": 8084 }, { "epoch": 0.9407726320688853, "grad_norm": 1.4586421251296997, "learning_rate": 0.0005, "loss": 2.6309378147125244, "step": 8085 }, { "epoch": 0.9408889923202234, "grad_norm": 0.5948104858398438, "learning_rate": 0.0005, "loss": 1.2438488006591797, "step": 8086 }, { "epoch": 0.9410053525715616, "grad_norm": 2.0933339595794678, "learning_rate": 0.0005, "loss": 2.509718656539917, "step": 8087 }, { "epoch": 0.9411217128228997, "grad_norm": 1.0132516622543335, "learning_rate": 0.0005, "loss": 1.7378801107406616, "step": 8088 }, { "epoch": 0.9412380730742378, "grad_norm": 1.0623281002044678, "learning_rate": 0.0005, "loss": 2.232067108154297, "step": 8089 }, { "epoch": 0.941354433325576, "grad_norm": 1.5761628150939941, "learning_rate": 0.0005, "loss": 2.4040675163269043, "step": 8090 }, { "epoch": 0.9414707935769141, "grad_norm": 1.07224702835083, "learning_rate": 0.0005, "loss": 2.1813995838165283, "step": 8091 }, { "epoch": 0.9415871538282523, "grad_norm": 2.734583616256714, "learning_rate": 0.0005, "loss": 2.375822067260742, "step": 8092 }, { "epoch": 0.9417035140795904, "grad_norm": 1.3859273195266724, "learning_rate": 0.0005, "loss": 2.582231044769287, "step": 8093 }, { "epoch": 0.9418198743309285, "grad_norm": 1.3699851036071777, "learning_rate": 0.0005, "loss": 2.0441582202911377, "step": 8094 }, { "epoch": 0.9419362345822667, "grad_norm": 1.053688645362854, "learning_rate": 0.0005, "loss": 2.009413480758667, "step": 8095 }, { "epoch": 0.9420525948336048, "grad_norm": 1.6310094594955444, "learning_rate": 0.0005, "loss": 2.295224666595459, "step": 8096 }, { "epoch": 0.942168955084943, "grad_norm": 1.7348121404647827, "learning_rate": 0.0005, "loss": 2.374227523803711, "step": 8097 }, { "epoch": 0.9422853153362811, "grad_norm": 1.124811053276062, "learning_rate": 0.0005, "loss": 3.090214729309082, "step": 8098 }, { "epoch": 0.9424016755876192, "grad_norm": 1.5781210660934448, "learning_rate": 0.0005, "loss": 3.321155071258545, "step": 8099 }, { "epoch": 0.9425180358389574, "grad_norm": 1.3021968603134155, "learning_rate": 0.0005, "loss": 2.2700185775756836, "step": 8100 }, { "epoch": 0.9426343960902955, "grad_norm": 0.9564859867095947, "learning_rate": 0.0005, "loss": 1.3234670162200928, "step": 8101 }, { "epoch": 0.9427507563416337, "grad_norm": 1.1381617784500122, "learning_rate": 0.0005, "loss": 1.8593792915344238, "step": 8102 }, { "epoch": 0.9428671165929718, "grad_norm": 0.9163731336593628, "learning_rate": 0.0005, "loss": 2.5456442832946777, "step": 8103 }, { "epoch": 0.9429834768443099, "grad_norm": 1.2128074169158936, "learning_rate": 0.0005, "loss": 2.031832695007324, "step": 8104 }, { "epoch": 0.9430998370956482, "grad_norm": 1.8852685689926147, "learning_rate": 0.0005, "loss": 1.7523120641708374, "step": 8105 }, { "epoch": 0.9432161973469863, "grad_norm": 2.296839714050293, "learning_rate": 0.0005, "loss": 3.037172317504883, "step": 8106 }, { "epoch": 0.9433325575983245, "grad_norm": 1.511763334274292, "learning_rate": 0.0005, "loss": 2.791130542755127, "step": 8107 }, { "epoch": 0.9434489178496626, "grad_norm": 0.9572844505310059, "learning_rate": 0.0005, "loss": 1.7540063858032227, "step": 8108 }, { "epoch": 0.9435652781010007, "grad_norm": 0.9182978868484497, "learning_rate": 0.0005, "loss": 1.7836700677871704, "step": 8109 }, { "epoch": 0.9436816383523389, "grad_norm": 1.5438027381896973, "learning_rate": 0.0005, "loss": 1.9851744174957275, "step": 8110 }, { "epoch": 0.943797998603677, "grad_norm": 1.141404151916504, "learning_rate": 0.0005, "loss": 1.9951609373092651, "step": 8111 }, { "epoch": 0.9439143588550152, "grad_norm": 1.5026566982269287, "learning_rate": 0.0005, "loss": 3.1001224517822266, "step": 8112 }, { "epoch": 0.9440307191063533, "grad_norm": 1.347379207611084, "learning_rate": 0.0005, "loss": 2.5464072227478027, "step": 8113 }, { "epoch": 0.9441470793576914, "grad_norm": 1.2072392702102661, "learning_rate": 0.0005, "loss": 1.5603700876235962, "step": 8114 }, { "epoch": 0.9442634396090296, "grad_norm": 1.916426181793213, "learning_rate": 0.0005, "loss": 2.7851481437683105, "step": 8115 }, { "epoch": 0.9443797998603677, "grad_norm": 1.577457070350647, "learning_rate": 0.0005, "loss": 2.4159297943115234, "step": 8116 }, { "epoch": 0.9444961601117059, "grad_norm": 1.353455901145935, "learning_rate": 0.0005, "loss": 2.147517204284668, "step": 8117 }, { "epoch": 0.944612520363044, "grad_norm": 1.1172245740890503, "learning_rate": 0.0005, "loss": 2.000875473022461, "step": 8118 }, { "epoch": 0.9447288806143821, "grad_norm": 2.4558892250061035, "learning_rate": 0.0005, "loss": 2.357309579849243, "step": 8119 }, { "epoch": 0.9448452408657203, "grad_norm": 1.2292125225067139, "learning_rate": 0.0005, "loss": 2.9396047592163086, "step": 8120 }, { "epoch": 0.9449616011170584, "grad_norm": 1.90151047706604, "learning_rate": 0.0005, "loss": 3.1661081314086914, "step": 8121 }, { "epoch": 0.9450779613683965, "grad_norm": 1.3206993341445923, "learning_rate": 0.0005, "loss": 2.4106640815734863, "step": 8122 }, { "epoch": 0.9451943216197347, "grad_norm": 1.305453896522522, "learning_rate": 0.0005, "loss": 2.078666925430298, "step": 8123 }, { "epoch": 0.9453106818710728, "grad_norm": 0.8703522682189941, "learning_rate": 0.0005, "loss": 1.8537160158157349, "step": 8124 }, { "epoch": 0.945427042122411, "grad_norm": 1.2120722532272339, "learning_rate": 0.0005, "loss": 2.032566547393799, "step": 8125 }, { "epoch": 0.9455434023737491, "grad_norm": 0.7935473322868347, "learning_rate": 0.0005, "loss": 2.0110390186309814, "step": 8126 }, { "epoch": 0.9456597626250872, "grad_norm": 1.6253165006637573, "learning_rate": 0.0005, "loss": 1.7475372552871704, "step": 8127 }, { "epoch": 0.9457761228764254, "grad_norm": 2.317927837371826, "learning_rate": 0.0005, "loss": 2.266810655593872, "step": 8128 }, { "epoch": 0.9458924831277635, "grad_norm": 2.1073973178863525, "learning_rate": 0.0005, "loss": 2.1131386756896973, "step": 8129 }, { "epoch": 0.9460088433791017, "grad_norm": 1.333376169204712, "learning_rate": 0.0005, "loss": 2.4028444290161133, "step": 8130 }, { "epoch": 0.9461252036304398, "grad_norm": 0.8847634196281433, "learning_rate": 0.0005, "loss": 2.064653158187866, "step": 8131 }, { "epoch": 0.9462415638817779, "grad_norm": 1.1967895030975342, "learning_rate": 0.0005, "loss": 2.3063137531280518, "step": 8132 }, { "epoch": 0.9463579241331161, "grad_norm": 1.115301489830017, "learning_rate": 0.0005, "loss": 1.8463242053985596, "step": 8133 }, { "epoch": 0.9464742843844542, "grad_norm": 1.4797910451889038, "learning_rate": 0.0005, "loss": 2.480726957321167, "step": 8134 }, { "epoch": 0.9465906446357925, "grad_norm": 1.0766513347625732, "learning_rate": 0.0005, "loss": 2.300509214401245, "step": 8135 }, { "epoch": 0.9467070048871306, "grad_norm": 1.6335573196411133, "learning_rate": 0.0005, "loss": 1.994673490524292, "step": 8136 }, { "epoch": 0.9468233651384687, "grad_norm": 1.5056668519973755, "learning_rate": 0.0005, "loss": 2.447380781173706, "step": 8137 }, { "epoch": 0.9469397253898069, "grad_norm": 1.172281265258789, "learning_rate": 0.0005, "loss": 2.3008062839508057, "step": 8138 }, { "epoch": 0.947056085641145, "grad_norm": 1.221229910850525, "learning_rate": 0.0005, "loss": 2.3869454860687256, "step": 8139 }, { "epoch": 0.9471724458924832, "grad_norm": 1.3055627346038818, "learning_rate": 0.0005, "loss": 1.9172229766845703, "step": 8140 }, { "epoch": 0.9472888061438213, "grad_norm": 1.6186867952346802, "learning_rate": 0.0005, "loss": 2.136861562728882, "step": 8141 }, { "epoch": 0.9474051663951594, "grad_norm": 1.5306637287139893, "learning_rate": 0.0005, "loss": 2.221696138381958, "step": 8142 }, { "epoch": 0.9475215266464976, "grad_norm": 1.382035255432129, "learning_rate": 0.0005, "loss": 2.8874025344848633, "step": 8143 }, { "epoch": 0.9476378868978357, "grad_norm": 1.1357890367507935, "learning_rate": 0.0005, "loss": 1.6880877017974854, "step": 8144 }, { "epoch": 0.9477542471491739, "grad_norm": 1.4519938230514526, "learning_rate": 0.0005, "loss": 2.6045889854431152, "step": 8145 }, { "epoch": 0.947870607400512, "grad_norm": 1.2101848125457764, "learning_rate": 0.0005, "loss": 2.232419967651367, "step": 8146 }, { "epoch": 0.9479869676518501, "grad_norm": 0.6410360932350159, "learning_rate": 0.0005, "loss": 1.337556004524231, "step": 8147 }, { "epoch": 0.9481033279031883, "grad_norm": 1.665164589881897, "learning_rate": 0.0005, "loss": 2.8644375801086426, "step": 8148 }, { "epoch": 0.9482196881545264, "grad_norm": 1.5215450525283813, "learning_rate": 0.0005, "loss": 3.1247975826263428, "step": 8149 }, { "epoch": 0.9483360484058646, "grad_norm": 1.0266170501708984, "learning_rate": 0.0005, "loss": 2.0314865112304688, "step": 8150 }, { "epoch": 0.9484524086572027, "grad_norm": 1.2766965627670288, "learning_rate": 0.0005, "loss": 2.7174947261810303, "step": 8151 }, { "epoch": 0.9485687689085408, "grad_norm": 1.4922796487808228, "learning_rate": 0.0005, "loss": 2.574782371520996, "step": 8152 }, { "epoch": 0.948685129159879, "grad_norm": 2.2906012535095215, "learning_rate": 0.0005, "loss": 2.885355234146118, "step": 8153 }, { "epoch": 0.9488014894112171, "grad_norm": 0.9833163619041443, "learning_rate": 0.0005, "loss": 2.2260591983795166, "step": 8154 }, { "epoch": 0.9489178496625553, "grad_norm": 1.3001528978347778, "learning_rate": 0.0005, "loss": 2.1303868293762207, "step": 8155 }, { "epoch": 0.9490342099138934, "grad_norm": 1.5073456764221191, "learning_rate": 0.0005, "loss": 2.577174425125122, "step": 8156 }, { "epoch": 0.9491505701652315, "grad_norm": 1.6661653518676758, "learning_rate": 0.0005, "loss": 2.312969207763672, "step": 8157 }, { "epoch": 0.9492669304165697, "grad_norm": 0.3551325500011444, "learning_rate": 0.0005, "loss": 0.9306035041809082, "step": 8158 }, { "epoch": 0.9493832906679078, "grad_norm": 1.6870218515396118, "learning_rate": 0.0005, "loss": 2.439803123474121, "step": 8159 }, { "epoch": 0.949499650919246, "grad_norm": 1.4803677797317505, "learning_rate": 0.0005, "loss": 1.631882905960083, "step": 8160 }, { "epoch": 0.9496160111705841, "grad_norm": 2.0224366188049316, "learning_rate": 0.0005, "loss": 2.3774094581604004, "step": 8161 }, { "epoch": 0.9497323714219222, "grad_norm": 1.7340528964996338, "learning_rate": 0.0005, "loss": 1.8864399194717407, "step": 8162 }, { "epoch": 0.9498487316732604, "grad_norm": 2.0643787384033203, "learning_rate": 0.0005, "loss": 2.758915424346924, "step": 8163 }, { "epoch": 0.9499650919245985, "grad_norm": 1.3667254447937012, "learning_rate": 0.0005, "loss": 2.0422234535217285, "step": 8164 }, { "epoch": 0.9500814521759366, "grad_norm": 1.2078015804290771, "learning_rate": 0.0005, "loss": 2.161496162414551, "step": 8165 }, { "epoch": 0.9501978124272749, "grad_norm": 1.1088541746139526, "learning_rate": 0.0005, "loss": 2.696061849594116, "step": 8166 }, { "epoch": 0.950314172678613, "grad_norm": 1.402261734008789, "learning_rate": 0.0005, "loss": 1.652144193649292, "step": 8167 }, { "epoch": 0.9504305329299512, "grad_norm": 1.4777568578720093, "learning_rate": 0.0005, "loss": 2.6258902549743652, "step": 8168 }, { "epoch": 0.9505468931812893, "grad_norm": 1.6210554838180542, "learning_rate": 0.0005, "loss": 3.07802152633667, "step": 8169 }, { "epoch": 0.9506632534326274, "grad_norm": 2.372985363006592, "learning_rate": 0.0005, "loss": 2.611900806427002, "step": 8170 }, { "epoch": 0.9507796136839656, "grad_norm": 1.390943169593811, "learning_rate": 0.0005, "loss": 3.222658634185791, "step": 8171 }, { "epoch": 0.9508959739353037, "grad_norm": 2.938267707824707, "learning_rate": 0.0005, "loss": 2.7953038215637207, "step": 8172 }, { "epoch": 0.9510123341866419, "grad_norm": 1.1475584506988525, "learning_rate": 0.0005, "loss": 2.6936583518981934, "step": 8173 }, { "epoch": 0.95112869443798, "grad_norm": 1.017111897468567, "learning_rate": 0.0005, "loss": 1.8936519622802734, "step": 8174 }, { "epoch": 0.9512450546893181, "grad_norm": 1.0414472818374634, "learning_rate": 0.0005, "loss": 1.51473069190979, "step": 8175 }, { "epoch": 0.9513614149406563, "grad_norm": 1.688220739364624, "learning_rate": 0.0005, "loss": 2.486527442932129, "step": 8176 }, { "epoch": 0.9514777751919944, "grad_norm": 1.385155439376831, "learning_rate": 0.0005, "loss": 1.8582630157470703, "step": 8177 }, { "epoch": 0.9515941354433326, "grad_norm": 1.3396137952804565, "learning_rate": 0.0005, "loss": 2.678056478500366, "step": 8178 }, { "epoch": 0.9517104956946707, "grad_norm": 1.2312586307525635, "learning_rate": 0.0005, "loss": 2.2893176078796387, "step": 8179 }, { "epoch": 0.9518268559460088, "grad_norm": 1.4217698574066162, "learning_rate": 0.0005, "loss": 2.894876480102539, "step": 8180 }, { "epoch": 0.951943216197347, "grad_norm": 1.9136919975280762, "learning_rate": 0.0005, "loss": 3.14644718170166, "step": 8181 }, { "epoch": 0.9520595764486851, "grad_norm": 1.0634764432907104, "learning_rate": 0.0005, "loss": 1.6133519411087036, "step": 8182 }, { "epoch": 0.9521759367000233, "grad_norm": 1.22176194190979, "learning_rate": 0.0005, "loss": 2.0592238903045654, "step": 8183 }, { "epoch": 0.9522922969513614, "grad_norm": 0.8287064433097839, "learning_rate": 0.0005, "loss": 1.6930073499679565, "step": 8184 }, { "epoch": 0.9524086572026995, "grad_norm": 1.3253042697906494, "learning_rate": 0.0005, "loss": 2.0513851642608643, "step": 8185 }, { "epoch": 0.9525250174540377, "grad_norm": 1.2100486755371094, "learning_rate": 0.0005, "loss": 1.2929866313934326, "step": 8186 }, { "epoch": 0.9526413777053758, "grad_norm": 1.594687581062317, "learning_rate": 0.0005, "loss": 2.549517869949341, "step": 8187 }, { "epoch": 0.952757737956714, "grad_norm": 1.076639175415039, "learning_rate": 0.0005, "loss": 2.174142837524414, "step": 8188 }, { "epoch": 0.9528740982080521, "grad_norm": 0.3855164051055908, "learning_rate": 0.0005, "loss": 1.4961568117141724, "step": 8189 }, { "epoch": 0.9529904584593902, "grad_norm": 1.6198011636734009, "learning_rate": 0.0005, "loss": 1.4859544038772583, "step": 8190 }, { "epoch": 0.9531068187107284, "grad_norm": 1.2501864433288574, "learning_rate": 0.0005, "loss": 2.5111708641052246, "step": 8191 }, { "epoch": 0.9532231789620665, "grad_norm": 0.9035991430282593, "learning_rate": 0.0005, "loss": 1.414825439453125, "step": 8192 }, { "epoch": 0.9533395392134048, "grad_norm": 1.3227512836456299, "learning_rate": 0.0005, "loss": 2.5736160278320312, "step": 8193 }, { "epoch": 0.9534558994647429, "grad_norm": 2.744373083114624, "learning_rate": 0.0005, "loss": 2.317662239074707, "step": 8194 }, { "epoch": 0.953572259716081, "grad_norm": 2.6748321056365967, "learning_rate": 0.0005, "loss": 2.8868141174316406, "step": 8195 }, { "epoch": 0.9536886199674192, "grad_norm": 2.034926414489746, "learning_rate": 0.0005, "loss": 2.9070494174957275, "step": 8196 }, { "epoch": 0.9538049802187573, "grad_norm": 1.7226359844207764, "learning_rate": 0.0005, "loss": 2.321932077407837, "step": 8197 }, { "epoch": 0.9539213404700955, "grad_norm": 1.4686262607574463, "learning_rate": 0.0005, "loss": 3.0282540321350098, "step": 8198 }, { "epoch": 0.9540377007214336, "grad_norm": 1.3520323038101196, "learning_rate": 0.0005, "loss": 2.429363250732422, "step": 8199 }, { "epoch": 0.9541540609727717, "grad_norm": 1.6433398723602295, "learning_rate": 0.0005, "loss": 2.350595474243164, "step": 8200 }, { "epoch": 0.9542704212241099, "grad_norm": 0.6041705012321472, "learning_rate": 0.0005, "loss": 1.4241187572479248, "step": 8201 }, { "epoch": 0.954386781475448, "grad_norm": 1.2658576965332031, "learning_rate": 0.0005, "loss": 2.1825480461120605, "step": 8202 }, { "epoch": 0.9545031417267861, "grad_norm": 1.3434234857559204, "learning_rate": 0.0005, "loss": 1.9967153072357178, "step": 8203 }, { "epoch": 0.9546195019781243, "grad_norm": 1.1930309534072876, "learning_rate": 0.0005, "loss": 1.6068732738494873, "step": 8204 }, { "epoch": 0.9547358622294624, "grad_norm": 0.8816052675247192, "learning_rate": 0.0005, "loss": 2.4241514205932617, "step": 8205 }, { "epoch": 0.9548522224808006, "grad_norm": 1.3086539506912231, "learning_rate": 0.0005, "loss": 1.9784173965454102, "step": 8206 }, { "epoch": 0.9549685827321387, "grad_norm": 1.033158302307129, "learning_rate": 0.0005, "loss": 1.759436845779419, "step": 8207 }, { "epoch": 0.9550849429834768, "grad_norm": 1.119530200958252, "learning_rate": 0.0005, "loss": 2.2197389602661133, "step": 8208 }, { "epoch": 0.955201303234815, "grad_norm": 1.0763853788375854, "learning_rate": 0.0005, "loss": 2.0172529220581055, "step": 8209 }, { "epoch": 0.9553176634861531, "grad_norm": 2.064204454421997, "learning_rate": 0.0005, "loss": 2.6612277030944824, "step": 8210 }, { "epoch": 0.9554340237374913, "grad_norm": 1.8502073287963867, "learning_rate": 0.0005, "loss": 2.5086605548858643, "step": 8211 }, { "epoch": 0.9555503839888294, "grad_norm": 1.3534903526306152, "learning_rate": 0.0005, "loss": 2.358797550201416, "step": 8212 }, { "epoch": 0.9556667442401675, "grad_norm": 1.1412622928619385, "learning_rate": 0.0005, "loss": 2.0634303092956543, "step": 8213 }, { "epoch": 0.9557831044915057, "grad_norm": 1.1502821445465088, "learning_rate": 0.0005, "loss": 2.0255837440490723, "step": 8214 }, { "epoch": 0.9558994647428438, "grad_norm": 1.4690138101577759, "learning_rate": 0.0005, "loss": 1.8031322956085205, "step": 8215 }, { "epoch": 0.956015824994182, "grad_norm": 1.1253246068954468, "learning_rate": 0.0005, "loss": 1.976327657699585, "step": 8216 }, { "epoch": 0.9561321852455201, "grad_norm": 1.0375360250473022, "learning_rate": 0.0005, "loss": 2.3601861000061035, "step": 8217 }, { "epoch": 0.9562485454968582, "grad_norm": 1.4807827472686768, "learning_rate": 0.0005, "loss": 3.2359633445739746, "step": 8218 }, { "epoch": 0.9563649057481964, "grad_norm": 1.2706468105316162, "learning_rate": 0.0005, "loss": 2.181316375732422, "step": 8219 }, { "epoch": 0.9564812659995345, "grad_norm": 3.9845056533813477, "learning_rate": 0.0005, "loss": 4.050662040710449, "step": 8220 }, { "epoch": 0.9565976262508727, "grad_norm": 1.749723196029663, "learning_rate": 0.0005, "loss": 2.491283655166626, "step": 8221 }, { "epoch": 0.9567139865022108, "grad_norm": 0.9383131861686707, "learning_rate": 0.0005, "loss": 1.7174558639526367, "step": 8222 }, { "epoch": 0.9568303467535489, "grad_norm": 1.4827525615692139, "learning_rate": 0.0005, "loss": 3.06158447265625, "step": 8223 }, { "epoch": 0.9569467070048872, "grad_norm": 0.737318217754364, "learning_rate": 0.0005, "loss": 1.868417739868164, "step": 8224 }, { "epoch": 0.9570630672562253, "grad_norm": 1.5450239181518555, "learning_rate": 0.0005, "loss": 3.05193829536438, "step": 8225 }, { "epoch": 0.9571794275075635, "grad_norm": 0.7556057572364807, "learning_rate": 0.0005, "loss": 1.8879281282424927, "step": 8226 }, { "epoch": 0.9572957877589016, "grad_norm": 2.2028019428253174, "learning_rate": 0.0005, "loss": 3.18154239654541, "step": 8227 }, { "epoch": 0.9574121480102397, "grad_norm": 1.2785379886627197, "learning_rate": 0.0005, "loss": 2.6605958938598633, "step": 8228 }, { "epoch": 0.9575285082615779, "grad_norm": 5.043343544006348, "learning_rate": 0.0005, "loss": 1.511953353881836, "step": 8229 }, { "epoch": 0.957644868512916, "grad_norm": 1.2331606149673462, "learning_rate": 0.0005, "loss": 1.4675332307815552, "step": 8230 }, { "epoch": 0.9577612287642542, "grad_norm": 2.0563955307006836, "learning_rate": 0.0005, "loss": 3.2661824226379395, "step": 8231 }, { "epoch": 0.9578775890155923, "grad_norm": 1.9780913591384888, "learning_rate": 0.0005, "loss": 2.3850996494293213, "step": 8232 }, { "epoch": 0.9579939492669304, "grad_norm": 1.561988115310669, "learning_rate": 0.0005, "loss": 2.526785373687744, "step": 8233 }, { "epoch": 0.9581103095182686, "grad_norm": 1.1410670280456543, "learning_rate": 0.0005, "loss": 3.083484649658203, "step": 8234 }, { "epoch": 0.9582266697696067, "grad_norm": 2.0469882488250732, "learning_rate": 0.0005, "loss": 2.645002841949463, "step": 8235 }, { "epoch": 0.9583430300209449, "grad_norm": 0.7675408124923706, "learning_rate": 0.0005, "loss": 1.3065924644470215, "step": 8236 }, { "epoch": 0.958459390272283, "grad_norm": 1.3546569347381592, "learning_rate": 0.0005, "loss": 1.6790881156921387, "step": 8237 }, { "epoch": 0.9585757505236211, "grad_norm": 2.7549312114715576, "learning_rate": 0.0005, "loss": 2.6671886444091797, "step": 8238 }, { "epoch": 0.9586921107749593, "grad_norm": 1.249519944190979, "learning_rate": 0.0005, "loss": 2.2621593475341797, "step": 8239 }, { "epoch": 0.9588084710262974, "grad_norm": 1.4012819528579712, "learning_rate": 0.0005, "loss": 2.265615463256836, "step": 8240 }, { "epoch": 0.9589248312776356, "grad_norm": 1.5739529132843018, "learning_rate": 0.0005, "loss": 3.568451404571533, "step": 8241 }, { "epoch": 0.9590411915289737, "grad_norm": 1.6527175903320312, "learning_rate": 0.0005, "loss": 3.385403871536255, "step": 8242 }, { "epoch": 0.9591575517803118, "grad_norm": 1.3322614431381226, "learning_rate": 0.0005, "loss": 2.776689291000366, "step": 8243 }, { "epoch": 0.95927391203165, "grad_norm": 1.7235517501831055, "learning_rate": 0.0005, "loss": 2.2323801517486572, "step": 8244 }, { "epoch": 0.9593902722829881, "grad_norm": 2.0748422145843506, "learning_rate": 0.0005, "loss": 3.228489398956299, "step": 8245 }, { "epoch": 0.9595066325343262, "grad_norm": 1.6424740552902222, "learning_rate": 0.0005, "loss": 1.9263322353363037, "step": 8246 }, { "epoch": 0.9596229927856644, "grad_norm": 1.156777024269104, "learning_rate": 0.0005, "loss": 2.4459996223449707, "step": 8247 }, { "epoch": 0.9597393530370025, "grad_norm": 1.3193798065185547, "learning_rate": 0.0005, "loss": 1.8463189601898193, "step": 8248 }, { "epoch": 0.9598557132883407, "grad_norm": 1.482870101928711, "learning_rate": 0.0005, "loss": 3.2143025398254395, "step": 8249 }, { "epoch": 0.9599720735396788, "grad_norm": 1.4459657669067383, "learning_rate": 0.0005, "loss": 2.6028175354003906, "step": 8250 }, { "epoch": 0.9600884337910169, "grad_norm": 0.9630603790283203, "learning_rate": 0.0005, "loss": 1.2938567399978638, "step": 8251 }, { "epoch": 0.9602047940423551, "grad_norm": 1.4013137817382812, "learning_rate": 0.0005, "loss": 1.9703871011734009, "step": 8252 }, { "epoch": 0.9603211542936932, "grad_norm": 1.4994158744812012, "learning_rate": 0.0005, "loss": 2.5482869148254395, "step": 8253 }, { "epoch": 0.9604375145450315, "grad_norm": 1.357055425643921, "learning_rate": 0.0005, "loss": 2.345231294631958, "step": 8254 }, { "epoch": 0.9605538747963696, "grad_norm": 1.273889422416687, "learning_rate": 0.0005, "loss": 2.2018964290618896, "step": 8255 }, { "epoch": 0.9606702350477077, "grad_norm": 1.5154597759246826, "learning_rate": 0.0005, "loss": 3.0571069717407227, "step": 8256 }, { "epoch": 0.9607865952990459, "grad_norm": 1.1944289207458496, "learning_rate": 0.0005, "loss": 1.4664196968078613, "step": 8257 }, { "epoch": 0.960902955550384, "grad_norm": 1.8630976676940918, "learning_rate": 0.0005, "loss": 2.990590810775757, "step": 8258 }, { "epoch": 0.9610193158017222, "grad_norm": 1.6992380619049072, "learning_rate": 0.0005, "loss": 2.471540689468384, "step": 8259 }, { "epoch": 0.9611356760530603, "grad_norm": 1.6810072660446167, "learning_rate": 0.0005, "loss": 1.9560649394989014, "step": 8260 }, { "epoch": 0.9612520363043984, "grad_norm": 1.0258169174194336, "learning_rate": 0.0005, "loss": 2.8538014888763428, "step": 8261 }, { "epoch": 0.9613683965557366, "grad_norm": 1.8861578702926636, "learning_rate": 0.0005, "loss": 2.5621581077575684, "step": 8262 }, { "epoch": 0.9614847568070747, "grad_norm": 1.3345551490783691, "learning_rate": 0.0005, "loss": 2.0483617782592773, "step": 8263 }, { "epoch": 0.9616011170584129, "grad_norm": 1.1991380453109741, "learning_rate": 0.0005, "loss": 2.8562874794006348, "step": 8264 }, { "epoch": 0.961717477309751, "grad_norm": 1.513325810432434, "learning_rate": 0.0005, "loss": 2.982405662536621, "step": 8265 }, { "epoch": 0.9618338375610891, "grad_norm": 0.9097127318382263, "learning_rate": 0.0005, "loss": 1.8477227687835693, "step": 8266 }, { "epoch": 0.9619501978124273, "grad_norm": 1.4588844776153564, "learning_rate": 0.0005, "loss": 2.575052499771118, "step": 8267 }, { "epoch": 0.9620665580637654, "grad_norm": 1.3818832635879517, "learning_rate": 0.0005, "loss": 2.6691431999206543, "step": 8268 }, { "epoch": 0.9621829183151036, "grad_norm": 1.1026880741119385, "learning_rate": 0.0005, "loss": 2.4383432865142822, "step": 8269 }, { "epoch": 0.9622992785664417, "grad_norm": 1.9443250894546509, "learning_rate": 0.0005, "loss": 3.1931309700012207, "step": 8270 }, { "epoch": 0.9624156388177798, "grad_norm": 1.539969801902771, "learning_rate": 0.0005, "loss": 2.429986000061035, "step": 8271 }, { "epoch": 0.962531999069118, "grad_norm": 1.4838213920593262, "learning_rate": 0.0005, "loss": 2.2338619232177734, "step": 8272 }, { "epoch": 0.9626483593204561, "grad_norm": 1.2490062713623047, "learning_rate": 0.0005, "loss": 1.9494062662124634, "step": 8273 }, { "epoch": 0.9627647195717943, "grad_norm": 1.261742353439331, "learning_rate": 0.0005, "loss": 2.739670515060425, "step": 8274 }, { "epoch": 0.9628810798231324, "grad_norm": 0.37422457337379456, "learning_rate": 0.0005, "loss": 0.9345466494560242, "step": 8275 }, { "epoch": 0.9629974400744705, "grad_norm": 1.0941416025161743, "learning_rate": 0.0005, "loss": 2.551978588104248, "step": 8276 }, { "epoch": 0.9631138003258087, "grad_norm": 0.9036052227020264, "learning_rate": 0.0005, "loss": 1.5645979642868042, "step": 8277 }, { "epoch": 0.9632301605771468, "grad_norm": 1.6785513162612915, "learning_rate": 0.0005, "loss": 2.3052754402160645, "step": 8278 }, { "epoch": 0.963346520828485, "grad_norm": 1.6914994716644287, "learning_rate": 0.0005, "loss": 1.4882569313049316, "step": 8279 }, { "epoch": 0.9634628810798231, "grad_norm": 0.8341985940933228, "learning_rate": 0.0005, "loss": 1.395103096961975, "step": 8280 }, { "epoch": 0.9635792413311612, "grad_norm": 1.7948896884918213, "learning_rate": 0.0005, "loss": 2.545542001724243, "step": 8281 }, { "epoch": 0.9636956015824995, "grad_norm": 1.1620999574661255, "learning_rate": 0.0005, "loss": 2.9632272720336914, "step": 8282 }, { "epoch": 0.9638119618338375, "grad_norm": 0.8297519087791443, "learning_rate": 0.0005, "loss": 1.6452159881591797, "step": 8283 }, { "epoch": 0.9639283220851758, "grad_norm": 2.287346363067627, "learning_rate": 0.0005, "loss": 2.1931114196777344, "step": 8284 }, { "epoch": 0.9640446823365139, "grad_norm": 1.9929403066635132, "learning_rate": 0.0005, "loss": 2.403745174407959, "step": 8285 }, { "epoch": 0.964161042587852, "grad_norm": 1.2645171880722046, "learning_rate": 0.0005, "loss": 2.3414053916931152, "step": 8286 }, { "epoch": 0.9642774028391902, "grad_norm": 0.9724118113517761, "learning_rate": 0.0005, "loss": 2.552615165710449, "step": 8287 }, { "epoch": 0.9643937630905283, "grad_norm": 1.4174097776412964, "learning_rate": 0.0005, "loss": 2.4070117473602295, "step": 8288 }, { "epoch": 0.9645101233418664, "grad_norm": 1.4317445755004883, "learning_rate": 0.0005, "loss": 2.329355478286743, "step": 8289 }, { "epoch": 0.9646264835932046, "grad_norm": 0.8762933015823364, "learning_rate": 0.0005, "loss": 2.129798412322998, "step": 8290 }, { "epoch": 0.9647428438445427, "grad_norm": 1.8476933240890503, "learning_rate": 0.0005, "loss": 2.4283008575439453, "step": 8291 }, { "epoch": 0.9648592040958809, "grad_norm": 1.7988072633743286, "learning_rate": 0.0005, "loss": 2.874330520629883, "step": 8292 }, { "epoch": 0.964975564347219, "grad_norm": 0.8091671466827393, "learning_rate": 0.0005, "loss": 2.1209003925323486, "step": 8293 }, { "epoch": 0.9650919245985571, "grad_norm": 1.1570667028427124, "learning_rate": 0.0005, "loss": 2.0046749114990234, "step": 8294 }, { "epoch": 0.9652082848498953, "grad_norm": 0.7728087306022644, "learning_rate": 0.0005, "loss": 2.0603904724121094, "step": 8295 }, { "epoch": 0.9653246451012334, "grad_norm": 1.400789737701416, "learning_rate": 0.0005, "loss": 1.9781179428100586, "step": 8296 }, { "epoch": 0.9654410053525716, "grad_norm": 0.8727428913116455, "learning_rate": 0.0005, "loss": 1.8767609596252441, "step": 8297 }, { "epoch": 0.9655573656039097, "grad_norm": 1.2695672512054443, "learning_rate": 0.0005, "loss": 2.535342216491699, "step": 8298 }, { "epoch": 0.9656737258552478, "grad_norm": 1.1405686140060425, "learning_rate": 0.0005, "loss": 2.1267788410186768, "step": 8299 }, { "epoch": 0.965790086106586, "grad_norm": 1.4840810298919678, "learning_rate": 0.0005, "loss": 1.953453779220581, "step": 8300 }, { "epoch": 0.9659064463579241, "grad_norm": 1.7221981287002563, "learning_rate": 0.0005, "loss": 2.3101089000701904, "step": 8301 }, { "epoch": 0.9660228066092623, "grad_norm": 1.4167461395263672, "learning_rate": 0.0005, "loss": 1.9726191759109497, "step": 8302 }, { "epoch": 0.9661391668606004, "grad_norm": 1.694231629371643, "learning_rate": 0.0005, "loss": 2.9112167358398438, "step": 8303 }, { "epoch": 0.9662555271119385, "grad_norm": 1.559240698814392, "learning_rate": 0.0005, "loss": 1.39994215965271, "step": 8304 }, { "epoch": 0.9663718873632767, "grad_norm": 1.6464133262634277, "learning_rate": 0.0005, "loss": 2.6734111309051514, "step": 8305 }, { "epoch": 0.9664882476146148, "grad_norm": 0.8258013725280762, "learning_rate": 0.0005, "loss": 1.0553617477416992, "step": 8306 }, { "epoch": 0.966604607865953, "grad_norm": 0.9362924695014954, "learning_rate": 0.0005, "loss": 1.5544737577438354, "step": 8307 }, { "epoch": 0.9667209681172911, "grad_norm": 1.215668797492981, "learning_rate": 0.0005, "loss": 1.4781566858291626, "step": 8308 }, { "epoch": 0.9668373283686292, "grad_norm": 1.2451051473617554, "learning_rate": 0.0005, "loss": 2.3132386207580566, "step": 8309 }, { "epoch": 0.9669536886199674, "grad_norm": 1.3965532779693604, "learning_rate": 0.0005, "loss": 2.1615307331085205, "step": 8310 }, { "epoch": 0.9670700488713055, "grad_norm": 3.5584700107574463, "learning_rate": 0.0005, "loss": 3.196525812149048, "step": 8311 }, { "epoch": 0.9671864091226438, "grad_norm": 0.4681856036186218, "learning_rate": 0.0005, "loss": 1.000003457069397, "step": 8312 }, { "epoch": 0.9673027693739819, "grad_norm": 0.9295578002929688, "learning_rate": 0.0005, "loss": 1.8624110221862793, "step": 8313 }, { "epoch": 0.96741912962532, "grad_norm": 1.5224521160125732, "learning_rate": 0.0005, "loss": 2.0396416187286377, "step": 8314 }, { "epoch": 0.9675354898766582, "grad_norm": 0.7605894804000854, "learning_rate": 0.0005, "loss": 2.269069194793701, "step": 8315 }, { "epoch": 0.9676518501279963, "grad_norm": 1.327048897743225, "learning_rate": 0.0005, "loss": 2.4935545921325684, "step": 8316 }, { "epoch": 0.9677682103793345, "grad_norm": 2.912140369415283, "learning_rate": 0.0005, "loss": 2.388054609298706, "step": 8317 }, { "epoch": 0.9678845706306726, "grad_norm": 1.776699185371399, "learning_rate": 0.0005, "loss": 2.4790453910827637, "step": 8318 }, { "epoch": 0.9680009308820107, "grad_norm": 1.0341042280197144, "learning_rate": 0.0005, "loss": 1.9441579580307007, "step": 8319 }, { "epoch": 0.9681172911333489, "grad_norm": 1.3241328001022339, "learning_rate": 0.0005, "loss": 2.0315499305725098, "step": 8320 }, { "epoch": 0.968233651384687, "grad_norm": 1.118796467781067, "learning_rate": 0.0005, "loss": 2.0183584690093994, "step": 8321 }, { "epoch": 0.9683500116360252, "grad_norm": 1.4363137483596802, "learning_rate": 0.0005, "loss": 2.1994752883911133, "step": 8322 }, { "epoch": 0.9684663718873633, "grad_norm": 1.1454683542251587, "learning_rate": 0.0005, "loss": 2.4396822452545166, "step": 8323 }, { "epoch": 0.9685827321387014, "grad_norm": 1.3860313892364502, "learning_rate": 0.0005, "loss": 2.371591567993164, "step": 8324 }, { "epoch": 0.9686990923900396, "grad_norm": 1.5586504936218262, "learning_rate": 0.0005, "loss": 2.714719295501709, "step": 8325 }, { "epoch": 0.9688154526413777, "grad_norm": 1.2350281476974487, "learning_rate": 0.0005, "loss": 1.4153578281402588, "step": 8326 }, { "epoch": 0.9689318128927158, "grad_norm": 1.3755418062210083, "learning_rate": 0.0005, "loss": 3.344059467315674, "step": 8327 }, { "epoch": 0.969048173144054, "grad_norm": 1.615715742111206, "learning_rate": 0.0005, "loss": 2.2753634452819824, "step": 8328 }, { "epoch": 0.9691645333953921, "grad_norm": 1.09220552444458, "learning_rate": 0.0005, "loss": 1.604887843132019, "step": 8329 }, { "epoch": 0.9692808936467303, "grad_norm": 1.3517464399337769, "learning_rate": 0.0005, "loss": 2.4504661560058594, "step": 8330 }, { "epoch": 0.9693972538980684, "grad_norm": 1.038333535194397, "learning_rate": 0.0005, "loss": 1.8745074272155762, "step": 8331 }, { "epoch": 0.9695136141494065, "grad_norm": 2.118900775909424, "learning_rate": 0.0005, "loss": 2.809264659881592, "step": 8332 }, { "epoch": 0.9696299744007447, "grad_norm": 1.1341938972473145, "learning_rate": 0.0005, "loss": 2.4522628784179688, "step": 8333 }, { "epoch": 0.9697463346520828, "grad_norm": 1.3363324403762817, "learning_rate": 0.0005, "loss": 2.093341827392578, "step": 8334 }, { "epoch": 0.969862694903421, "grad_norm": 2.067568778991699, "learning_rate": 0.0005, "loss": 2.46567964553833, "step": 8335 }, { "epoch": 0.9699790551547591, "grad_norm": 1.2977925539016724, "learning_rate": 0.0005, "loss": 2.5221128463745117, "step": 8336 }, { "epoch": 0.9700954154060972, "grad_norm": 0.5649584531784058, "learning_rate": 0.0005, "loss": 1.3689067363739014, "step": 8337 }, { "epoch": 0.9702117756574354, "grad_norm": 0.9402199983596802, "learning_rate": 0.0005, "loss": 1.6523637771606445, "step": 8338 }, { "epoch": 0.9703281359087735, "grad_norm": 1.764949917793274, "learning_rate": 0.0005, "loss": 2.267214775085449, "step": 8339 }, { "epoch": 0.9704444961601117, "grad_norm": 1.3320010900497437, "learning_rate": 0.0005, "loss": 2.369642734527588, "step": 8340 }, { "epoch": 0.9705608564114498, "grad_norm": 0.8533387184143066, "learning_rate": 0.0005, "loss": 1.8654645681381226, "step": 8341 }, { "epoch": 0.970677216662788, "grad_norm": 0.8761659264564514, "learning_rate": 0.0005, "loss": 1.2488956451416016, "step": 8342 }, { "epoch": 0.9707935769141262, "grad_norm": 1.575355887413025, "learning_rate": 0.0005, "loss": 2.7376108169555664, "step": 8343 }, { "epoch": 0.9709099371654643, "grad_norm": 2.832170009613037, "learning_rate": 0.0005, "loss": 2.871546745300293, "step": 8344 }, { "epoch": 0.9710262974168025, "grad_norm": 1.5963596105575562, "learning_rate": 0.0005, "loss": 1.8596925735473633, "step": 8345 }, { "epoch": 0.9711426576681406, "grad_norm": 1.8671255111694336, "learning_rate": 0.0005, "loss": 2.2524795532226562, "step": 8346 }, { "epoch": 0.9712590179194787, "grad_norm": 1.2162848711013794, "learning_rate": 0.0005, "loss": 2.1520214080810547, "step": 8347 }, { "epoch": 0.9713753781708169, "grad_norm": 1.1010810136795044, "learning_rate": 0.0005, "loss": 1.5770490169525146, "step": 8348 }, { "epoch": 0.971491738422155, "grad_norm": 1.5313001871109009, "learning_rate": 0.0005, "loss": 2.267192840576172, "step": 8349 }, { "epoch": 0.9716080986734932, "grad_norm": 1.1449270248413086, "learning_rate": 0.0005, "loss": 2.175487518310547, "step": 8350 }, { "epoch": 0.9717244589248313, "grad_norm": 1.0984153747558594, "learning_rate": 0.0005, "loss": 1.8961178064346313, "step": 8351 }, { "epoch": 0.9718408191761694, "grad_norm": 1.0764312744140625, "learning_rate": 0.0005, "loss": 1.6994576454162598, "step": 8352 }, { "epoch": 0.9719571794275076, "grad_norm": 1.1862900257110596, "learning_rate": 0.0005, "loss": 2.147343158721924, "step": 8353 }, { "epoch": 0.9720735396788457, "grad_norm": 1.4481306076049805, "learning_rate": 0.0005, "loss": 1.3795599937438965, "step": 8354 }, { "epoch": 0.9721898999301839, "grad_norm": 1.0535895824432373, "learning_rate": 0.0005, "loss": 2.2070369720458984, "step": 8355 }, { "epoch": 0.972306260181522, "grad_norm": 1.2935724258422852, "learning_rate": 0.0005, "loss": 2.15393328666687, "step": 8356 }, { "epoch": 0.9724226204328601, "grad_norm": 1.6755404472351074, "learning_rate": 0.0005, "loss": 1.9314631223678589, "step": 8357 }, { "epoch": 0.9725389806841983, "grad_norm": 1.8525938987731934, "learning_rate": 0.0005, "loss": 2.953155040740967, "step": 8358 }, { "epoch": 0.9726553409355364, "grad_norm": 2.495138645172119, "learning_rate": 0.0005, "loss": 2.437224864959717, "step": 8359 }, { "epoch": 0.9727717011868746, "grad_norm": 1.1039650440216064, "learning_rate": 0.0005, "loss": 2.4598984718322754, "step": 8360 }, { "epoch": 0.9728880614382127, "grad_norm": 1.4981458187103271, "learning_rate": 0.0005, "loss": 2.883026599884033, "step": 8361 }, { "epoch": 0.9730044216895508, "grad_norm": 1.6783055067062378, "learning_rate": 0.0005, "loss": 2.2832136154174805, "step": 8362 }, { "epoch": 0.973120781940889, "grad_norm": 1.4859015941619873, "learning_rate": 0.0005, "loss": 2.4576122760772705, "step": 8363 }, { "epoch": 0.9732371421922271, "grad_norm": 1.3777594566345215, "learning_rate": 0.0005, "loss": 2.694765090942383, "step": 8364 }, { "epoch": 0.9733535024435653, "grad_norm": 2.7122538089752197, "learning_rate": 0.0005, "loss": 1.8159536123275757, "step": 8365 }, { "epoch": 0.9734698626949034, "grad_norm": 1.2296490669250488, "learning_rate": 0.0005, "loss": 2.577476739883423, "step": 8366 }, { "epoch": 0.9735862229462415, "grad_norm": 1.1503578424453735, "learning_rate": 0.0005, "loss": 2.5221452713012695, "step": 8367 }, { "epoch": 0.9737025831975797, "grad_norm": 1.3419164419174194, "learning_rate": 0.0005, "loss": 2.972611904144287, "step": 8368 }, { "epoch": 0.9738189434489178, "grad_norm": 1.7471024990081787, "learning_rate": 0.0005, "loss": 2.0560808181762695, "step": 8369 }, { "epoch": 0.9739353037002559, "grad_norm": 0.7319504618644714, "learning_rate": 0.0005, "loss": 1.713529348373413, "step": 8370 }, { "epoch": 0.9740516639515941, "grad_norm": 1.588768720626831, "learning_rate": 0.0005, "loss": 2.594008684158325, "step": 8371 }, { "epoch": 0.9741680242029322, "grad_norm": 1.507180094718933, "learning_rate": 0.0005, "loss": 1.8803530931472778, "step": 8372 }, { "epoch": 0.9742843844542705, "grad_norm": 1.429048776626587, "learning_rate": 0.0005, "loss": 1.9590680599212646, "step": 8373 }, { "epoch": 0.9744007447056086, "grad_norm": 1.6720348596572876, "learning_rate": 0.0005, "loss": 2.273656129837036, "step": 8374 }, { "epoch": 0.9745171049569467, "grad_norm": 1.2060139179229736, "learning_rate": 0.0005, "loss": 1.7573342323303223, "step": 8375 }, { "epoch": 0.9746334652082849, "grad_norm": 1.481887698173523, "learning_rate": 0.0005, "loss": 2.1494596004486084, "step": 8376 }, { "epoch": 0.974749825459623, "grad_norm": 1.7153993844985962, "learning_rate": 0.0005, "loss": 2.4745349884033203, "step": 8377 }, { "epoch": 0.9748661857109612, "grad_norm": 1.1700432300567627, "learning_rate": 0.0005, "loss": 2.049304246902466, "step": 8378 }, { "epoch": 0.9749825459622993, "grad_norm": 1.3461183309555054, "learning_rate": 0.0005, "loss": 2.2186367511749268, "step": 8379 }, { "epoch": 0.9750989062136374, "grad_norm": 7.996060371398926, "learning_rate": 0.0005, "loss": 2.5843448638916016, "step": 8380 }, { "epoch": 0.9752152664649756, "grad_norm": 0.951648473739624, "learning_rate": 0.0005, "loss": 2.069568157196045, "step": 8381 }, { "epoch": 0.9753316267163137, "grad_norm": 1.9040476083755493, "learning_rate": 0.0005, "loss": 1.7472736835479736, "step": 8382 }, { "epoch": 0.9754479869676519, "grad_norm": 1.0858840942382812, "learning_rate": 0.0005, "loss": 2.8595471382141113, "step": 8383 }, { "epoch": 0.97556434721899, "grad_norm": 1.5962177515029907, "learning_rate": 0.0005, "loss": 2.282198667526245, "step": 8384 }, { "epoch": 0.9756807074703281, "grad_norm": 1.2735780477523804, "learning_rate": 0.0005, "loss": 3.1542673110961914, "step": 8385 }, { "epoch": 0.9757970677216663, "grad_norm": 1.5326085090637207, "learning_rate": 0.0005, "loss": 2.203777313232422, "step": 8386 }, { "epoch": 0.9759134279730044, "grad_norm": 1.1285526752471924, "learning_rate": 0.0005, "loss": 2.7371420860290527, "step": 8387 }, { "epoch": 0.9760297882243426, "grad_norm": 1.545606255531311, "learning_rate": 0.0005, "loss": 3.0103580951690674, "step": 8388 }, { "epoch": 0.9761461484756807, "grad_norm": 0.9024639129638672, "learning_rate": 0.0005, "loss": 2.154592514038086, "step": 8389 }, { "epoch": 0.9762625087270188, "grad_norm": 1.3901969194412231, "learning_rate": 0.0005, "loss": 3.2100348472595215, "step": 8390 }, { "epoch": 0.976378868978357, "grad_norm": 1.5647482872009277, "learning_rate": 0.0005, "loss": 2.942890167236328, "step": 8391 }, { "epoch": 0.9764952292296951, "grad_norm": 1.0509840250015259, "learning_rate": 0.0005, "loss": 2.694816827774048, "step": 8392 }, { "epoch": 0.9766115894810333, "grad_norm": 1.5144579410552979, "learning_rate": 0.0005, "loss": 2.957937240600586, "step": 8393 }, { "epoch": 0.9767279497323714, "grad_norm": 1.3192825317382812, "learning_rate": 0.0005, "loss": 3.0199007987976074, "step": 8394 }, { "epoch": 0.9768443099837095, "grad_norm": 0.8639818429946899, "learning_rate": 0.0005, "loss": 2.2802319526672363, "step": 8395 }, { "epoch": 0.9769606702350477, "grad_norm": 1.359416127204895, "learning_rate": 0.0005, "loss": 1.6369184255599976, "step": 8396 }, { "epoch": 0.9770770304863858, "grad_norm": 1.2584805488586426, "learning_rate": 0.0005, "loss": 1.630000114440918, "step": 8397 }, { "epoch": 0.977193390737724, "grad_norm": 1.9059982299804688, "learning_rate": 0.0005, "loss": 2.4665846824645996, "step": 8398 }, { "epoch": 0.9773097509890621, "grad_norm": 1.1578845977783203, "learning_rate": 0.0005, "loss": 1.9031364917755127, "step": 8399 }, { "epoch": 0.9774261112404002, "grad_norm": 1.192850947380066, "learning_rate": 0.0005, "loss": 1.8460783958435059, "step": 8400 }, { "epoch": 0.9775424714917385, "grad_norm": 1.5878750085830688, "learning_rate": 0.0005, "loss": 1.7472131252288818, "step": 8401 }, { "epoch": 0.9776588317430766, "grad_norm": 1.7112866640090942, "learning_rate": 0.0005, "loss": 2.1546034812927246, "step": 8402 }, { "epoch": 0.9777751919944148, "grad_norm": 3.65033221244812, "learning_rate": 0.0005, "loss": 2.5236597061157227, "step": 8403 }, { "epoch": 0.9778915522457529, "grad_norm": 1.1291711330413818, "learning_rate": 0.0005, "loss": 2.093755006790161, "step": 8404 }, { "epoch": 0.978007912497091, "grad_norm": 0.837774395942688, "learning_rate": 0.0005, "loss": 1.535224199295044, "step": 8405 }, { "epoch": 0.9781242727484292, "grad_norm": 1.6786127090454102, "learning_rate": 0.0005, "loss": 3.085170269012451, "step": 8406 }, { "epoch": 0.9782406329997673, "grad_norm": 1.4253735542297363, "learning_rate": 0.0005, "loss": 3.5231451988220215, "step": 8407 }, { "epoch": 0.9783569932511055, "grad_norm": 1.7069554328918457, "learning_rate": 0.0005, "loss": 2.9870004653930664, "step": 8408 }, { "epoch": 0.9784733535024436, "grad_norm": 1.3461625576019287, "learning_rate": 0.0005, "loss": 2.1110219955444336, "step": 8409 }, { "epoch": 0.9785897137537817, "grad_norm": 0.5630580186843872, "learning_rate": 0.0005, "loss": 1.037619709968567, "step": 8410 }, { "epoch": 0.9787060740051199, "grad_norm": 1.0892961025238037, "learning_rate": 0.0005, "loss": 1.7738577127456665, "step": 8411 }, { "epoch": 0.978822434256458, "grad_norm": 0.6922491192817688, "learning_rate": 0.0005, "loss": 1.5472733974456787, "step": 8412 }, { "epoch": 0.9789387945077961, "grad_norm": 1.6127526760101318, "learning_rate": 0.0005, "loss": 2.1778225898742676, "step": 8413 }, { "epoch": 0.9790551547591343, "grad_norm": 1.3672386407852173, "learning_rate": 0.0005, "loss": 2.194051504135132, "step": 8414 }, { "epoch": 0.9791715150104724, "grad_norm": 2.78948974609375, "learning_rate": 0.0005, "loss": 2.554262638092041, "step": 8415 }, { "epoch": 0.9792878752618106, "grad_norm": 0.959382951259613, "learning_rate": 0.0005, "loss": 1.9135055541992188, "step": 8416 }, { "epoch": 0.9794042355131487, "grad_norm": 1.3948332071304321, "learning_rate": 0.0005, "loss": 2.595950126647949, "step": 8417 }, { "epoch": 0.9795205957644868, "grad_norm": 1.2715269327163696, "learning_rate": 0.0005, "loss": 2.4283595085144043, "step": 8418 }, { "epoch": 0.979636956015825, "grad_norm": 0.43070724606513977, "learning_rate": 0.0005, "loss": 1.0169334411621094, "step": 8419 }, { "epoch": 0.9797533162671631, "grad_norm": 1.5885465145111084, "learning_rate": 0.0005, "loss": 2.2486979961395264, "step": 8420 }, { "epoch": 0.9798696765185013, "grad_norm": 0.6540311574935913, "learning_rate": 0.0005, "loss": 1.8287379741668701, "step": 8421 }, { "epoch": 0.9799860367698394, "grad_norm": 1.1199250221252441, "learning_rate": 0.0005, "loss": 2.1172332763671875, "step": 8422 }, { "epoch": 0.9801023970211775, "grad_norm": 1.5052434206008911, "learning_rate": 0.0005, "loss": 2.480842113494873, "step": 8423 }, { "epoch": 0.9802187572725157, "grad_norm": 1.486821174621582, "learning_rate": 0.0005, "loss": 2.945465326309204, "step": 8424 }, { "epoch": 0.9803351175238538, "grad_norm": 0.946513831615448, "learning_rate": 0.0005, "loss": 1.829217553138733, "step": 8425 }, { "epoch": 0.980451477775192, "grad_norm": 1.3780475854873657, "learning_rate": 0.0005, "loss": 2.2150518894195557, "step": 8426 }, { "epoch": 0.9805678380265301, "grad_norm": 1.172834038734436, "learning_rate": 0.0005, "loss": 2.0062308311462402, "step": 8427 }, { "epoch": 0.9806841982778682, "grad_norm": 2.544900417327881, "learning_rate": 0.0005, "loss": 2.0610218048095703, "step": 8428 }, { "epoch": 0.9808005585292064, "grad_norm": 2.24282169342041, "learning_rate": 0.0005, "loss": 2.262432813644409, "step": 8429 }, { "epoch": 0.9809169187805445, "grad_norm": 1.4951560497283936, "learning_rate": 0.0005, "loss": 2.2314934730529785, "step": 8430 }, { "epoch": 0.9810332790318828, "grad_norm": 1.2337146997451782, "learning_rate": 0.0005, "loss": 2.508002758026123, "step": 8431 }, { "epoch": 0.9811496392832209, "grad_norm": 1.6300562620162964, "learning_rate": 0.0005, "loss": 2.07824444770813, "step": 8432 }, { "epoch": 0.981265999534559, "grad_norm": 0.9703794717788696, "learning_rate": 0.0005, "loss": 1.3008469343185425, "step": 8433 }, { "epoch": 0.9813823597858972, "grad_norm": 1.469765543937683, "learning_rate": 0.0005, "loss": 2.627145290374756, "step": 8434 }, { "epoch": 0.9814987200372353, "grad_norm": 1.6007649898529053, "learning_rate": 0.0005, "loss": 1.9492359161376953, "step": 8435 }, { "epoch": 0.9816150802885735, "grad_norm": 1.1897581815719604, "learning_rate": 0.0005, "loss": 2.421438694000244, "step": 8436 }, { "epoch": 0.9817314405399116, "grad_norm": 1.4647340774536133, "learning_rate": 0.0005, "loss": 2.565735340118408, "step": 8437 }, { "epoch": 0.9818478007912497, "grad_norm": 1.0411372184753418, "learning_rate": 0.0005, "loss": 2.1666855812072754, "step": 8438 }, { "epoch": 0.9819641610425879, "grad_norm": 1.1206769943237305, "learning_rate": 0.0005, "loss": 1.7462995052337646, "step": 8439 }, { "epoch": 0.982080521293926, "grad_norm": 1.50767982006073, "learning_rate": 0.0005, "loss": 2.2844743728637695, "step": 8440 }, { "epoch": 0.9821968815452642, "grad_norm": 1.752713680267334, "learning_rate": 0.0005, "loss": 2.412078857421875, "step": 8441 }, { "epoch": 0.9823132417966023, "grad_norm": 1.420791745185852, "learning_rate": 0.0005, "loss": 1.518194317817688, "step": 8442 }, { "epoch": 0.9824296020479404, "grad_norm": 1.1758925914764404, "learning_rate": 0.0005, "loss": 1.5550265312194824, "step": 8443 }, { "epoch": 0.9825459622992786, "grad_norm": 1.276394009590149, "learning_rate": 0.0005, "loss": 2.0480191707611084, "step": 8444 }, { "epoch": 0.9826623225506167, "grad_norm": 1.387951374053955, "learning_rate": 0.0005, "loss": 2.3161890506744385, "step": 8445 }, { "epoch": 0.9827786828019549, "grad_norm": 1.5471723079681396, "learning_rate": 0.0005, "loss": 1.546267032623291, "step": 8446 }, { "epoch": 0.982895043053293, "grad_norm": 1.4456690549850464, "learning_rate": 0.0005, "loss": 3.1563220024108887, "step": 8447 }, { "epoch": 0.9830114033046311, "grad_norm": 1.957698941230774, "learning_rate": 0.0005, "loss": 2.9570207595825195, "step": 8448 }, { "epoch": 0.9831277635559693, "grad_norm": 1.569164752960205, "learning_rate": 0.0005, "loss": 1.6564263105392456, "step": 8449 }, { "epoch": 0.9832441238073074, "grad_norm": 1.3781540393829346, "learning_rate": 0.0005, "loss": 2.6202428340911865, "step": 8450 }, { "epoch": 0.9833604840586455, "grad_norm": 1.9002392292022705, "learning_rate": 0.0005, "loss": 1.7814018726348877, "step": 8451 }, { "epoch": 0.9834768443099837, "grad_norm": 4.364488124847412, "learning_rate": 0.0005, "loss": 2.766218900680542, "step": 8452 }, { "epoch": 0.9835932045613218, "grad_norm": 1.0108619928359985, "learning_rate": 0.0005, "loss": 1.930306315422058, "step": 8453 }, { "epoch": 0.98370956481266, "grad_norm": 1.5050007104873657, "learning_rate": 0.0005, "loss": 2.6144142150878906, "step": 8454 }, { "epoch": 0.9838259250639981, "grad_norm": 1.2674486637115479, "learning_rate": 0.0005, "loss": 1.6530325412750244, "step": 8455 }, { "epoch": 0.9839422853153362, "grad_norm": 2.178548574447632, "learning_rate": 0.0005, "loss": 2.7208046913146973, "step": 8456 }, { "epoch": 0.9840586455666744, "grad_norm": 1.524245023727417, "learning_rate": 0.0005, "loss": 1.9172981977462769, "step": 8457 }, { "epoch": 0.9841750058180125, "grad_norm": 1.425323486328125, "learning_rate": 0.0005, "loss": 2.255108594894409, "step": 8458 }, { "epoch": 0.9842913660693507, "grad_norm": 1.2235069274902344, "learning_rate": 0.0005, "loss": 2.7595927715301514, "step": 8459 }, { "epoch": 0.9844077263206888, "grad_norm": 1.4516620635986328, "learning_rate": 0.0005, "loss": 2.034550666809082, "step": 8460 }, { "epoch": 0.984524086572027, "grad_norm": 1.4103469848632812, "learning_rate": 0.0005, "loss": 2.0785393714904785, "step": 8461 }, { "epoch": 0.9846404468233652, "grad_norm": 1.4333552122116089, "learning_rate": 0.0005, "loss": 2.0828680992126465, "step": 8462 }, { "epoch": 0.9847568070747033, "grad_norm": 1.3803032636642456, "learning_rate": 0.0005, "loss": 2.815239429473877, "step": 8463 }, { "epoch": 0.9848731673260415, "grad_norm": 0.7846501469612122, "learning_rate": 0.0005, "loss": 1.4462201595306396, "step": 8464 }, { "epoch": 0.9849895275773796, "grad_norm": 1.41922926902771, "learning_rate": 0.0005, "loss": 2.300629138946533, "step": 8465 }, { "epoch": 0.9851058878287177, "grad_norm": 1.4646527767181396, "learning_rate": 0.0005, "loss": 2.9397521018981934, "step": 8466 }, { "epoch": 0.9852222480800559, "grad_norm": 1.167974829673767, "learning_rate": 0.0005, "loss": 2.1452410221099854, "step": 8467 }, { "epoch": 0.985338608331394, "grad_norm": 1.5670138597488403, "learning_rate": 0.0005, "loss": 3.761043071746826, "step": 8468 }, { "epoch": 0.9854549685827322, "grad_norm": 1.3862519264221191, "learning_rate": 0.0005, "loss": 2.4291768074035645, "step": 8469 }, { "epoch": 0.9855713288340703, "grad_norm": 1.0143330097198486, "learning_rate": 0.0005, "loss": 2.3365304470062256, "step": 8470 }, { "epoch": 0.9856876890854084, "grad_norm": 1.0550545454025269, "learning_rate": 0.0005, "loss": 1.8339009284973145, "step": 8471 }, { "epoch": 0.9858040493367466, "grad_norm": 1.3220423460006714, "learning_rate": 0.0005, "loss": 1.8118923902511597, "step": 8472 }, { "epoch": 0.9859204095880847, "grad_norm": 0.9794461131095886, "learning_rate": 0.0005, "loss": 1.9161791801452637, "step": 8473 }, { "epoch": 0.9860367698394229, "grad_norm": 1.8932536840438843, "learning_rate": 0.0005, "loss": 3.0095367431640625, "step": 8474 }, { "epoch": 0.986153130090761, "grad_norm": 1.330293893814087, "learning_rate": 0.0005, "loss": 2.437512159347534, "step": 8475 }, { "epoch": 0.9862694903420991, "grad_norm": 2.3702340126037598, "learning_rate": 0.0005, "loss": 2.8898019790649414, "step": 8476 }, { "epoch": 0.9863858505934373, "grad_norm": 1.2699897289276123, "learning_rate": 0.0005, "loss": 2.239974021911621, "step": 8477 }, { "epoch": 0.9865022108447754, "grad_norm": 1.3219884634017944, "learning_rate": 0.0005, "loss": 2.956752300262451, "step": 8478 }, { "epoch": 0.9866185710961136, "grad_norm": 1.6430721282958984, "learning_rate": 0.0005, "loss": 3.171687126159668, "step": 8479 }, { "epoch": 0.9867349313474517, "grad_norm": 1.776432991027832, "learning_rate": 0.0005, "loss": 2.5041208267211914, "step": 8480 }, { "epoch": 0.9868512915987898, "grad_norm": 1.281627893447876, "learning_rate": 0.0005, "loss": 2.2104623317718506, "step": 8481 }, { "epoch": 0.986967651850128, "grad_norm": 1.3769556283950806, "learning_rate": 0.0005, "loss": 1.9914122819900513, "step": 8482 }, { "epoch": 0.9870840121014661, "grad_norm": 1.0783520936965942, "learning_rate": 0.0005, "loss": 1.9898232221603394, "step": 8483 }, { "epoch": 0.9872003723528043, "grad_norm": 2.165598154067993, "learning_rate": 0.0005, "loss": 2.7512331008911133, "step": 8484 }, { "epoch": 0.9873167326041424, "grad_norm": 2.6012609004974365, "learning_rate": 0.0005, "loss": 2.8952016830444336, "step": 8485 }, { "epoch": 0.9874330928554805, "grad_norm": 0.9038764238357544, "learning_rate": 0.0005, "loss": 2.1375017166137695, "step": 8486 }, { "epoch": 0.9875494531068187, "grad_norm": 1.5647006034851074, "learning_rate": 0.0005, "loss": 2.1986913681030273, "step": 8487 }, { "epoch": 0.9876658133581568, "grad_norm": 1.6795214414596558, "learning_rate": 0.0005, "loss": 2.089207649230957, "step": 8488 }, { "epoch": 0.987782173609495, "grad_norm": 1.7930564880371094, "learning_rate": 0.0005, "loss": 2.8490424156188965, "step": 8489 }, { "epoch": 0.9878985338608332, "grad_norm": 0.9577872157096863, "learning_rate": 0.0005, "loss": 1.6734557151794434, "step": 8490 }, { "epoch": 0.9880148941121712, "grad_norm": 1.3845068216323853, "learning_rate": 0.0005, "loss": 3.150906562805176, "step": 8491 }, { "epoch": 0.9881312543635095, "grad_norm": 1.3951343297958374, "learning_rate": 0.0005, "loss": 2.3167948722839355, "step": 8492 }, { "epoch": 0.9882476146148476, "grad_norm": 1.2608897686004639, "learning_rate": 0.0005, "loss": 2.017733573913574, "step": 8493 }, { "epoch": 0.9883639748661857, "grad_norm": 1.9346811771392822, "learning_rate": 0.0005, "loss": 1.7709991931915283, "step": 8494 }, { "epoch": 0.9884803351175239, "grad_norm": 1.4555699825286865, "learning_rate": 0.0005, "loss": 2.1149790287017822, "step": 8495 }, { "epoch": 0.988596695368862, "grad_norm": 1.2034289836883545, "learning_rate": 0.0005, "loss": 2.628302812576294, "step": 8496 }, { "epoch": 0.9887130556202002, "grad_norm": 1.1077970266342163, "learning_rate": 0.0005, "loss": 2.045208215713501, "step": 8497 }, { "epoch": 0.9888294158715383, "grad_norm": 0.9105047583580017, "learning_rate": 0.0005, "loss": 1.3609662055969238, "step": 8498 }, { "epoch": 0.9889457761228764, "grad_norm": 1.7454859018325806, "learning_rate": 0.0005, "loss": 2.894627571105957, "step": 8499 }, { "epoch": 0.9890621363742146, "grad_norm": 1.2298765182495117, "learning_rate": 0.0005, "loss": 2.295849561691284, "step": 8500 }, { "epoch": 0.9891784966255527, "grad_norm": 1.358257532119751, "learning_rate": 0.0005, "loss": 2.6022696495056152, "step": 8501 }, { "epoch": 0.9892948568768909, "grad_norm": 1.6760727167129517, "learning_rate": 0.0005, "loss": 2.4940900802612305, "step": 8502 }, { "epoch": 0.989411217128229, "grad_norm": 1.4169002771377563, "learning_rate": 0.0005, "loss": 2.4976487159729004, "step": 8503 }, { "epoch": 0.9895275773795671, "grad_norm": 1.202072262763977, "learning_rate": 0.0005, "loss": 2.091728687286377, "step": 8504 }, { "epoch": 0.9896439376309053, "grad_norm": 0.9111424684524536, "learning_rate": 0.0005, "loss": 1.5058279037475586, "step": 8505 }, { "epoch": 0.9897602978822434, "grad_norm": 1.0988264083862305, "learning_rate": 0.0005, "loss": 2.298107147216797, "step": 8506 }, { "epoch": 0.9898766581335816, "grad_norm": 2.9353668689727783, "learning_rate": 0.0005, "loss": 2.59952449798584, "step": 8507 }, { "epoch": 0.9899930183849197, "grad_norm": 1.4536044597625732, "learning_rate": 0.0005, "loss": 1.7522832155227661, "step": 8508 }, { "epoch": 0.9901093786362578, "grad_norm": 1.5553910732269287, "learning_rate": 0.0005, "loss": 2.225588321685791, "step": 8509 }, { "epoch": 0.990225738887596, "grad_norm": 1.420372486114502, "learning_rate": 0.0005, "loss": 2.376674175262451, "step": 8510 }, { "epoch": 0.9903420991389341, "grad_norm": 2.122561454772949, "learning_rate": 0.0005, "loss": 2.3635876178741455, "step": 8511 }, { "epoch": 0.9904584593902723, "grad_norm": 1.2679250240325928, "learning_rate": 0.0005, "loss": 1.7341771125793457, "step": 8512 }, { "epoch": 0.9905748196416104, "grad_norm": 1.4818904399871826, "learning_rate": 0.0005, "loss": 2.162169933319092, "step": 8513 }, { "epoch": 0.9906911798929485, "grad_norm": 0.5699349045753479, "learning_rate": 0.0005, "loss": 1.5317285060882568, "step": 8514 }, { "epoch": 0.9908075401442867, "grad_norm": 0.8544589877128601, "learning_rate": 0.0005, "loss": 1.6421020030975342, "step": 8515 }, { "epoch": 0.9909239003956248, "grad_norm": 1.4486192464828491, "learning_rate": 0.0005, "loss": 2.676225423812866, "step": 8516 }, { "epoch": 0.991040260646963, "grad_norm": 1.0939923524856567, "learning_rate": 0.0005, "loss": 1.8683185577392578, "step": 8517 }, { "epoch": 0.9911566208983011, "grad_norm": 1.30144202709198, "learning_rate": 0.0005, "loss": 2.05844783782959, "step": 8518 }, { "epoch": 0.9912729811496392, "grad_norm": 1.650685429573059, "learning_rate": 0.0005, "loss": 2.876561164855957, "step": 8519 }, { "epoch": 0.9913893414009775, "grad_norm": 1.5273405313491821, "learning_rate": 0.0005, "loss": 2.319894313812256, "step": 8520 }, { "epoch": 0.9915057016523156, "grad_norm": 1.7452799081802368, "learning_rate": 0.0005, "loss": 2.0713136196136475, "step": 8521 }, { "epoch": 0.9916220619036538, "grad_norm": 2.250670909881592, "learning_rate": 0.0005, "loss": 2.596705436706543, "step": 8522 }, { "epoch": 0.9917384221549919, "grad_norm": 1.3136003017425537, "learning_rate": 0.0005, "loss": 2.7398416996002197, "step": 8523 }, { "epoch": 0.99185478240633, "grad_norm": 1.1445156335830688, "learning_rate": 0.0005, "loss": 2.625241756439209, "step": 8524 }, { "epoch": 0.9919711426576682, "grad_norm": 1.3811922073364258, "learning_rate": 0.0005, "loss": 2.3686838150024414, "step": 8525 }, { "epoch": 0.9920875029090063, "grad_norm": 1.0956529378890991, "learning_rate": 0.0005, "loss": 2.5148751735687256, "step": 8526 }, { "epoch": 0.9922038631603445, "grad_norm": 0.8637689352035522, "learning_rate": 0.0005, "loss": 1.9858818054199219, "step": 8527 }, { "epoch": 0.9923202234116826, "grad_norm": 1.1867289543151855, "learning_rate": 0.0005, "loss": 2.1937992572784424, "step": 8528 }, { "epoch": 0.9924365836630207, "grad_norm": 0.9377933144569397, "learning_rate": 0.0005, "loss": 2.90942645072937, "step": 8529 }, { "epoch": 0.9925529439143589, "grad_norm": 1.2069053649902344, "learning_rate": 0.0005, "loss": 2.7070038318634033, "step": 8530 }, { "epoch": 0.992669304165697, "grad_norm": 1.2282620668411255, "learning_rate": 0.0005, "loss": 2.456449031829834, "step": 8531 }, { "epoch": 0.9927856644170352, "grad_norm": 2.104077100753784, "learning_rate": 0.0005, "loss": 2.59214448928833, "step": 8532 }, { "epoch": 0.9929020246683733, "grad_norm": 1.5036197900772095, "learning_rate": 0.0005, "loss": 3.610454559326172, "step": 8533 }, { "epoch": 0.9930183849197114, "grad_norm": 1.4107730388641357, "learning_rate": 0.0005, "loss": 2.2542052268981934, "step": 8534 }, { "epoch": 0.9931347451710496, "grad_norm": 1.569549798965454, "learning_rate": 0.0005, "loss": 2.0917534828186035, "step": 8535 }, { "epoch": 0.9932511054223877, "grad_norm": 1.070616602897644, "learning_rate": 0.0005, "loss": 2.9036307334899902, "step": 8536 }, { "epoch": 0.9933674656737258, "grad_norm": 1.505021095275879, "learning_rate": 0.0005, "loss": 2.091285228729248, "step": 8537 }, { "epoch": 0.993483825925064, "grad_norm": 1.865270733833313, "learning_rate": 0.0005, "loss": 2.1734468936920166, "step": 8538 }, { "epoch": 0.9936001861764021, "grad_norm": 1.1008858680725098, "learning_rate": 0.0005, "loss": 3.042654514312744, "step": 8539 }, { "epoch": 0.9937165464277403, "grad_norm": 1.8221298456192017, "learning_rate": 0.0005, "loss": 1.8449718952178955, "step": 8540 }, { "epoch": 0.9938329066790784, "grad_norm": 1.0664631128311157, "learning_rate": 0.0005, "loss": 2.747563600540161, "step": 8541 }, { "epoch": 0.9939492669304165, "grad_norm": 0.5861731767654419, "learning_rate": 0.0005, "loss": 1.2630770206451416, "step": 8542 }, { "epoch": 0.9940656271817547, "grad_norm": 0.8854292035102844, "learning_rate": 0.0005, "loss": 2.0712594985961914, "step": 8543 }, { "epoch": 0.9941819874330928, "grad_norm": 1.7331691980361938, "learning_rate": 0.0005, "loss": 2.0726566314697266, "step": 8544 }, { "epoch": 0.994298347684431, "grad_norm": 2.3810269832611084, "learning_rate": 0.0005, "loss": 3.073369026184082, "step": 8545 }, { "epoch": 0.9944147079357691, "grad_norm": 1.4121155738830566, "learning_rate": 0.0005, "loss": 2.3503570556640625, "step": 8546 }, { "epoch": 0.9945310681871072, "grad_norm": 0.8945354223251343, "learning_rate": 0.0005, "loss": 1.803870677947998, "step": 8547 }, { "epoch": 0.9946474284384454, "grad_norm": 3.217602252960205, "learning_rate": 0.0005, "loss": 2.9705073833465576, "step": 8548 }, { "epoch": 0.9947637886897835, "grad_norm": 1.3855957984924316, "learning_rate": 0.0005, "loss": 3.5806331634521484, "step": 8549 }, { "epoch": 0.9948801489411218, "grad_norm": 1.582412600517273, "learning_rate": 0.0005, "loss": 2.2328131198883057, "step": 8550 }, { "epoch": 0.9949965091924599, "grad_norm": 2.0078983306884766, "learning_rate": 0.0005, "loss": 2.356987953186035, "step": 8551 }, { "epoch": 0.995112869443798, "grad_norm": 1.7502830028533936, "learning_rate": 0.0005, "loss": 1.390505075454712, "step": 8552 }, { "epoch": 0.9952292296951362, "grad_norm": 2.0701189041137695, "learning_rate": 0.0005, "loss": 2.328401803970337, "step": 8553 }, { "epoch": 0.9953455899464743, "grad_norm": 1.6540741920471191, "learning_rate": 0.0005, "loss": 1.8283740282058716, "step": 8554 }, { "epoch": 0.9954619501978125, "grad_norm": 1.1771917343139648, "learning_rate": 0.0005, "loss": 1.7733724117279053, "step": 8555 }, { "epoch": 0.9955783104491506, "grad_norm": 1.2513154745101929, "learning_rate": 0.0005, "loss": 2.354334831237793, "step": 8556 }, { "epoch": 0.9956946707004887, "grad_norm": 2.0222764015197754, "learning_rate": 0.0005, "loss": 2.9384984970092773, "step": 8557 }, { "epoch": 0.9958110309518269, "grad_norm": 1.1461888551712036, "learning_rate": 0.0005, "loss": 1.6566791534423828, "step": 8558 }, { "epoch": 0.995927391203165, "grad_norm": 1.3644092082977295, "learning_rate": 0.0005, "loss": 3.1498172283172607, "step": 8559 }, { "epoch": 0.9960437514545032, "grad_norm": 1.111903190612793, "learning_rate": 0.0005, "loss": 1.8964877128601074, "step": 8560 }, { "epoch": 0.9961601117058413, "grad_norm": 1.9627017974853516, "learning_rate": 0.0005, "loss": 2.8406155109405518, "step": 8561 }, { "epoch": 0.9962764719571794, "grad_norm": 1.6437625885009766, "learning_rate": 0.0005, "loss": 2.0485031604766846, "step": 8562 }, { "epoch": 0.9963928322085176, "grad_norm": 1.318956971168518, "learning_rate": 0.0005, "loss": 2.8009839057922363, "step": 8563 }, { "epoch": 0.9965091924598557, "grad_norm": 1.089129090309143, "learning_rate": 0.0005, "loss": 3.012232780456543, "step": 8564 }, { "epoch": 0.9966255527111939, "grad_norm": 1.660347819328308, "learning_rate": 0.0005, "loss": 2.2216856479644775, "step": 8565 }, { "epoch": 0.996741912962532, "grad_norm": 2.626246452331543, "learning_rate": 0.0005, "loss": 2.1011581420898438, "step": 8566 }, { "epoch": 0.9968582732138701, "grad_norm": 1.7256790399551392, "learning_rate": 0.0005, "loss": 3.947274684906006, "step": 8567 }, { "epoch": 0.9969746334652083, "grad_norm": 1.5172854661941528, "learning_rate": 0.0005, "loss": 1.9441925287246704, "step": 8568 }, { "epoch": 0.9970909937165464, "grad_norm": 1.0459494590759277, "learning_rate": 0.0005, "loss": 2.3643245697021484, "step": 8569 }, { "epoch": 0.9972073539678846, "grad_norm": 1.6418695449829102, "learning_rate": 0.0005, "loss": 2.3978078365325928, "step": 8570 }, { "epoch": 0.9973237142192227, "grad_norm": 12.499951362609863, "learning_rate": 0.0005, "loss": 2.223775625228882, "step": 8571 }, { "epoch": 0.9974400744705608, "grad_norm": 1.3143454790115356, "learning_rate": 0.0005, "loss": 1.8339803218841553, "step": 8572 }, { "epoch": 0.997556434721899, "grad_norm": 2.29852032661438, "learning_rate": 0.0005, "loss": 2.1877427101135254, "step": 8573 }, { "epoch": 0.9976727949732371, "grad_norm": 1.6288357973098755, "learning_rate": 0.0005, "loss": 2.6842122077941895, "step": 8574 }, { "epoch": 0.9977891552245752, "grad_norm": 0.9790574908256531, "learning_rate": 0.0005, "loss": 1.898557424545288, "step": 8575 }, { "epoch": 0.9979055154759134, "grad_norm": 0.8451627492904663, "learning_rate": 0.0005, "loss": 2.0692901611328125, "step": 8576 }, { "epoch": 0.9980218757272515, "grad_norm": 1.5996900796890259, "learning_rate": 0.0005, "loss": 2.830179452896118, "step": 8577 }, { "epoch": 0.9981382359785897, "grad_norm": 1.3970175981521606, "learning_rate": 0.0005, "loss": 2.6256208419799805, "step": 8578 }, { "epoch": 0.9982545962299278, "grad_norm": 0.7769176959991455, "learning_rate": 0.0005, "loss": 1.3154147863388062, "step": 8579 }, { "epoch": 0.998370956481266, "grad_norm": 1.625506043434143, "learning_rate": 0.0005, "loss": 2.3969614505767822, "step": 8580 }, { "epoch": 0.9984873167326042, "grad_norm": 0.9042901992797852, "learning_rate": 0.0005, "loss": 1.395573616027832, "step": 8581 }, { "epoch": 0.9986036769839423, "grad_norm": 1.7361637353897095, "learning_rate": 0.0005, "loss": 2.6145575046539307, "step": 8582 }, { "epoch": 0.9987200372352805, "grad_norm": 2.247596502304077, "learning_rate": 0.0005, "loss": 2.752142906188965, "step": 8583 }, { "epoch": 0.9988363974866186, "grad_norm": 0.824201226234436, "learning_rate": 0.0005, "loss": 1.9293750524520874, "step": 8584 }, { "epoch": 0.9989527577379567, "grad_norm": 1.3047575950622559, "learning_rate": 0.0005, "loss": 1.8930463790893555, "step": 8585 }, { "epoch": 0.9990691179892949, "grad_norm": 2.2846667766571045, "learning_rate": 0.0005, "loss": 2.7817435264587402, "step": 8586 }, { "epoch": 0.999185478240633, "grad_norm": 1.314111590385437, "learning_rate": 0.0005, "loss": 2.4107816219329834, "step": 8587 }, { "epoch": 0.9993018384919712, "grad_norm": 1.075466513633728, "learning_rate": 0.0005, "loss": 1.9913363456726074, "step": 8588 }, { "epoch": 0.9994181987433093, "grad_norm": 1.296169400215149, "learning_rate": 0.0005, "loss": 3.0176548957824707, "step": 8589 }, { "epoch": 0.9995345589946474, "grad_norm": 2.1130127906799316, "learning_rate": 0.0005, "loss": 3.673046588897705, "step": 8590 }, { "epoch": 0.9996509192459856, "grad_norm": 1.4867082834243774, "learning_rate": 0.0005, "loss": 1.7893174886703491, "step": 8591 }, { "epoch": 0.9997672794973237, "grad_norm": 1.0750170946121216, "learning_rate": 0.0005, "loss": 1.7260987758636475, "step": 8592 }, { "epoch": 0.9998836397486619, "grad_norm": 1.9390910863876343, "learning_rate": 0.0005, "loss": 1.7581543922424316, "step": 8593 }, { "epoch": 1.0, "grad_norm": 1.3791906833648682, "learning_rate": 0.0005, "loss": 1.913263201713562, "step": 8594 }, { "epoch": 1.000116360251338, "grad_norm": 1.4784801006317139, "learning_rate": 0.0005, "loss": 2.089498281478882, "step": 8595 }, { "epoch": 1.0002327205026762, "grad_norm": 1.0084612369537354, "learning_rate": 0.0005, "loss": 1.8812183141708374, "step": 8596 }, { "epoch": 1.0003490807540145, "grad_norm": 1.0821057558059692, "learning_rate": 0.0005, "loss": 2.210146427154541, "step": 8597 }, { "epoch": 1.0004654410053526, "grad_norm": 0.9458748698234558, "learning_rate": 0.0005, "loss": 1.9154887199401855, "step": 8598 }, { "epoch": 1.0005818012566907, "grad_norm": 0.7688151001930237, "learning_rate": 0.0005, "loss": 1.2854609489440918, "step": 8599 }, { "epoch": 1.0006981615080288, "grad_norm": 0.9018763899803162, "learning_rate": 0.0005, "loss": 2.4703006744384766, "step": 8600 }, { "epoch": 1.000814521759367, "grad_norm": 1.1192121505737305, "learning_rate": 0.0005, "loss": 1.7873140573501587, "step": 8601 }, { "epoch": 1.0009308820107052, "grad_norm": 1.1412138938903809, "learning_rate": 0.0005, "loss": 1.2010703086853027, "step": 8602 }, { "epoch": 1.0010472422620433, "grad_norm": 12.401172637939453, "learning_rate": 0.0005, "loss": 1.1983299255371094, "step": 8603 }, { "epoch": 1.0011636025133814, "grad_norm": 0.8513967394828796, "learning_rate": 0.0005, "loss": 1.5826022624969482, "step": 8604 }, { "epoch": 1.0012799627647195, "grad_norm": 1.0101861953735352, "learning_rate": 0.0005, "loss": 1.4341447353363037, "step": 8605 }, { "epoch": 1.0013963230160576, "grad_norm": 1.036468267440796, "learning_rate": 0.0005, "loss": 2.338243007659912, "step": 8606 }, { "epoch": 1.001512683267396, "grad_norm": 0.9143533706665039, "learning_rate": 0.0005, "loss": 1.9046106338500977, "step": 8607 }, { "epoch": 1.001629043518734, "grad_norm": 1.445552945137024, "learning_rate": 0.0005, "loss": 2.260295867919922, "step": 8608 }, { "epoch": 1.0017454037700722, "grad_norm": 0.8579748868942261, "learning_rate": 0.0005, "loss": 1.3083688020706177, "step": 8609 }, { "epoch": 1.0018617640214103, "grad_norm": 1.2907248735427856, "learning_rate": 0.0005, "loss": 2.5121467113494873, "step": 8610 }, { "epoch": 1.0019781242727483, "grad_norm": 1.0901682376861572, "learning_rate": 0.0005, "loss": 1.6218559741973877, "step": 8611 }, { "epoch": 1.0020944845240867, "grad_norm": 1.232764720916748, "learning_rate": 0.0005, "loss": 1.8248836994171143, "step": 8612 }, { "epoch": 1.0022108447754248, "grad_norm": 1.1971676349639893, "learning_rate": 0.0005, "loss": 2.2208194732666016, "step": 8613 }, { "epoch": 1.0023272050267629, "grad_norm": 1.406782865524292, "learning_rate": 0.0005, "loss": 2.094517707824707, "step": 8614 }, { "epoch": 1.002443565278101, "grad_norm": 1.5085443258285522, "learning_rate": 0.0005, "loss": 1.7971786260604858, "step": 8615 }, { "epoch": 1.002559925529439, "grad_norm": 1.1169966459274292, "learning_rate": 0.0005, "loss": 1.885339617729187, "step": 8616 }, { "epoch": 1.0026762857807774, "grad_norm": 1.0140808820724487, "learning_rate": 0.0005, "loss": 2.0703887939453125, "step": 8617 }, { "epoch": 1.0027926460321155, "grad_norm": 1.0831760168075562, "learning_rate": 0.0005, "loss": 1.8773835897445679, "step": 8618 }, { "epoch": 1.0029090062834536, "grad_norm": 1.6315836906433105, "learning_rate": 0.0005, "loss": 2.4794557094573975, "step": 8619 }, { "epoch": 1.0030253665347917, "grad_norm": 0.752508282661438, "learning_rate": 0.0005, "loss": 1.6468546390533447, "step": 8620 }, { "epoch": 1.0031417267861298, "grad_norm": 1.0011351108551025, "learning_rate": 0.0005, "loss": 1.8006532192230225, "step": 8621 }, { "epoch": 1.003258087037468, "grad_norm": 1.1344443559646606, "learning_rate": 0.0005, "loss": 1.8083168268203735, "step": 8622 }, { "epoch": 1.0033744472888062, "grad_norm": 1.4112050533294678, "learning_rate": 0.0005, "loss": 1.5763869285583496, "step": 8623 }, { "epoch": 1.0034908075401443, "grad_norm": 0.9821398854255676, "learning_rate": 0.0005, "loss": 1.6885689496994019, "step": 8624 }, { "epoch": 1.0036071677914824, "grad_norm": 0.9837414026260376, "learning_rate": 0.0005, "loss": 1.7832403182983398, "step": 8625 }, { "epoch": 1.0037235280428205, "grad_norm": 1.1403721570968628, "learning_rate": 0.0005, "loss": 2.739882469177246, "step": 8626 }, { "epoch": 1.0038398882941588, "grad_norm": 1.7240382432937622, "learning_rate": 0.0005, "loss": 2.5084567070007324, "step": 8627 }, { "epoch": 1.003956248545497, "grad_norm": 1.1906569004058838, "learning_rate": 0.0005, "loss": 2.1137020587921143, "step": 8628 }, { "epoch": 1.004072608796835, "grad_norm": 0.8140192627906799, "learning_rate": 0.0005, "loss": 1.939173698425293, "step": 8629 }, { "epoch": 1.0041889690481731, "grad_norm": 0.7634296417236328, "learning_rate": 0.0005, "loss": 1.4724268913269043, "step": 8630 }, { "epoch": 1.0043053292995112, "grad_norm": 1.083869457244873, "learning_rate": 0.0005, "loss": 1.8294706344604492, "step": 8631 }, { "epoch": 1.0044216895508495, "grad_norm": 0.6402466297149658, "learning_rate": 0.0005, "loss": 1.2018861770629883, "step": 8632 }, { "epoch": 1.0045380498021876, "grad_norm": 1.3686035871505737, "learning_rate": 0.0005, "loss": 1.9797050952911377, "step": 8633 }, { "epoch": 1.0046544100535257, "grad_norm": 1.6791260242462158, "learning_rate": 0.0005, "loss": 1.8600480556488037, "step": 8634 }, { "epoch": 1.0047707703048638, "grad_norm": 0.7784068584442139, "learning_rate": 0.0005, "loss": 1.5644724369049072, "step": 8635 }, { "epoch": 1.004887130556202, "grad_norm": 1.0245431661605835, "learning_rate": 0.0005, "loss": 2.0580403804779053, "step": 8636 }, { "epoch": 1.00500349080754, "grad_norm": 1.4136879444122314, "learning_rate": 0.0005, "loss": 2.588200092315674, "step": 8637 }, { "epoch": 1.0051198510588784, "grad_norm": 1.3309476375579834, "learning_rate": 0.0005, "loss": 2.360858678817749, "step": 8638 }, { "epoch": 1.0052362113102165, "grad_norm": 0.7167325019836426, "learning_rate": 0.0005, "loss": 1.7758733034133911, "step": 8639 }, { "epoch": 1.0053525715615546, "grad_norm": 1.0489951372146606, "learning_rate": 0.0005, "loss": 1.2116127014160156, "step": 8640 }, { "epoch": 1.0054689318128927, "grad_norm": 1.4329299926757812, "learning_rate": 0.0005, "loss": 2.2745120525360107, "step": 8641 }, { "epoch": 1.0055852920642308, "grad_norm": 1.6213538646697998, "learning_rate": 0.0005, "loss": 1.8908096551895142, "step": 8642 }, { "epoch": 1.005701652315569, "grad_norm": 1.544118046760559, "learning_rate": 0.0005, "loss": 1.8484864234924316, "step": 8643 }, { "epoch": 1.0058180125669072, "grad_norm": 1.05573570728302, "learning_rate": 0.0005, "loss": 1.8792577981948853, "step": 8644 }, { "epoch": 1.0059343728182453, "grad_norm": 1.0772285461425781, "learning_rate": 0.0005, "loss": 1.734480857849121, "step": 8645 }, { "epoch": 1.0060507330695834, "grad_norm": 1.672518253326416, "learning_rate": 0.0005, "loss": 2.564279317855835, "step": 8646 }, { "epoch": 1.0061670933209215, "grad_norm": 1.172057032585144, "learning_rate": 0.0005, "loss": 1.4828405380249023, "step": 8647 }, { "epoch": 1.0062834535722598, "grad_norm": 1.0135252475738525, "learning_rate": 0.0005, "loss": 1.3798084259033203, "step": 8648 }, { "epoch": 1.0063998138235979, "grad_norm": 1.0678081512451172, "learning_rate": 0.0005, "loss": 1.7856513261795044, "step": 8649 }, { "epoch": 1.006516174074936, "grad_norm": 1.1433831453323364, "learning_rate": 0.0005, "loss": 2.0991368293762207, "step": 8650 }, { "epoch": 1.006632534326274, "grad_norm": 1.4030914306640625, "learning_rate": 0.0005, "loss": 2.4820971488952637, "step": 8651 }, { "epoch": 1.0067488945776122, "grad_norm": 0.8124790191650391, "learning_rate": 0.0005, "loss": 1.3363840579986572, "step": 8652 }, { "epoch": 1.0068652548289505, "grad_norm": 1.4870339632034302, "learning_rate": 0.0005, "loss": 1.4260919094085693, "step": 8653 }, { "epoch": 1.0069816150802886, "grad_norm": 0.9527716040611267, "learning_rate": 0.0005, "loss": 1.3541133403778076, "step": 8654 }, { "epoch": 1.0070979753316267, "grad_norm": 1.7323896884918213, "learning_rate": 0.0005, "loss": 1.8154234886169434, "step": 8655 }, { "epoch": 1.0072143355829648, "grad_norm": 1.43209707736969, "learning_rate": 0.0005, "loss": 1.9789164066314697, "step": 8656 }, { "epoch": 1.007330695834303, "grad_norm": 0.7963276505470276, "learning_rate": 0.0005, "loss": 1.1593081951141357, "step": 8657 }, { "epoch": 1.0074470560856412, "grad_norm": 1.0491944551467896, "learning_rate": 0.0005, "loss": 1.574669361114502, "step": 8658 }, { "epoch": 1.0075634163369793, "grad_norm": 0.9466221928596497, "learning_rate": 0.0005, "loss": 2.189608097076416, "step": 8659 }, { "epoch": 1.0076797765883174, "grad_norm": 0.8815527558326721, "learning_rate": 0.0005, "loss": 1.8772833347320557, "step": 8660 }, { "epoch": 1.0077961368396555, "grad_norm": 0.7320405840873718, "learning_rate": 0.0005, "loss": 1.475663423538208, "step": 8661 }, { "epoch": 1.0079124970909936, "grad_norm": 1.0770598649978638, "learning_rate": 0.0005, "loss": 1.7522759437561035, "step": 8662 }, { "epoch": 1.008028857342332, "grad_norm": 1.2383664846420288, "learning_rate": 0.0005, "loss": 2.663433074951172, "step": 8663 }, { "epoch": 1.00814521759367, "grad_norm": 0.7933303713798523, "learning_rate": 0.0005, "loss": 2.0730490684509277, "step": 8664 }, { "epoch": 1.0082615778450081, "grad_norm": 1.5062651634216309, "learning_rate": 0.0005, "loss": 1.8605327606201172, "step": 8665 }, { "epoch": 1.0083779380963462, "grad_norm": 1.0197950601577759, "learning_rate": 0.0005, "loss": 2.2041804790496826, "step": 8666 }, { "epoch": 1.0084942983476843, "grad_norm": 1.0494917631149292, "learning_rate": 0.0005, "loss": 1.3081092834472656, "step": 8667 }, { "epoch": 1.0086106585990227, "grad_norm": 0.6561743021011353, "learning_rate": 0.0005, "loss": 1.232776165008545, "step": 8668 }, { "epoch": 1.0087270188503608, "grad_norm": 1.999536395072937, "learning_rate": 0.0005, "loss": 2.3426897525787354, "step": 8669 }, { "epoch": 1.0088433791016989, "grad_norm": 1.071685791015625, "learning_rate": 0.0005, "loss": 1.7770004272460938, "step": 8670 }, { "epoch": 1.008959739353037, "grad_norm": 0.896064817905426, "learning_rate": 0.0005, "loss": 1.7289965152740479, "step": 8671 }, { "epoch": 1.009076099604375, "grad_norm": 0.9037473797798157, "learning_rate": 0.0005, "loss": 1.6049493551254272, "step": 8672 }, { "epoch": 1.0091924598557134, "grad_norm": 1.3773072957992554, "learning_rate": 0.0005, "loss": 1.8338056802749634, "step": 8673 }, { "epoch": 1.0093088201070515, "grad_norm": 1.0337374210357666, "learning_rate": 0.0005, "loss": 2.5715982913970947, "step": 8674 }, { "epoch": 1.0094251803583896, "grad_norm": 1.1627863645553589, "learning_rate": 0.0005, "loss": 2.438251495361328, "step": 8675 }, { "epoch": 1.0095415406097277, "grad_norm": 1.1415516138076782, "learning_rate": 0.0005, "loss": 2.260920286178589, "step": 8676 }, { "epoch": 1.0096579008610658, "grad_norm": 0.6460487842559814, "learning_rate": 0.0005, "loss": 1.5141600370407104, "step": 8677 }, { "epoch": 1.009774261112404, "grad_norm": 0.9403098225593567, "learning_rate": 0.0005, "loss": 1.7847824096679688, "step": 8678 }, { "epoch": 1.0098906213637422, "grad_norm": 1.0006693601608276, "learning_rate": 0.0005, "loss": 1.5979787111282349, "step": 8679 }, { "epoch": 1.0100069816150803, "grad_norm": 1.1897728443145752, "learning_rate": 0.0005, "loss": 2.110442638397217, "step": 8680 }, { "epoch": 1.0101233418664184, "grad_norm": 0.9779231548309326, "learning_rate": 0.0005, "loss": 1.4128843545913696, "step": 8681 }, { "epoch": 1.0102397021177565, "grad_norm": 1.7459807395935059, "learning_rate": 0.0005, "loss": 2.1749773025512695, "step": 8682 }, { "epoch": 1.0103560623690948, "grad_norm": 1.175876498222351, "learning_rate": 0.0005, "loss": 2.431959629058838, "step": 8683 }, { "epoch": 1.010472422620433, "grad_norm": 0.8913065791130066, "learning_rate": 0.0005, "loss": 1.5940736532211304, "step": 8684 }, { "epoch": 1.010588782871771, "grad_norm": 1.3051278591156006, "learning_rate": 0.0005, "loss": 2.5383384227752686, "step": 8685 }, { "epoch": 1.010705143123109, "grad_norm": 0.7066026926040649, "learning_rate": 0.0005, "loss": 1.5829156637191772, "step": 8686 }, { "epoch": 1.0108215033744472, "grad_norm": 1.5507988929748535, "learning_rate": 0.0005, "loss": 2.226163864135742, "step": 8687 }, { "epoch": 1.0109378636257855, "grad_norm": 0.577959418296814, "learning_rate": 0.0005, "loss": 1.2997554540634155, "step": 8688 }, { "epoch": 1.0110542238771236, "grad_norm": 1.4607073068618774, "learning_rate": 0.0005, "loss": 2.10215163230896, "step": 8689 }, { "epoch": 1.0111705841284617, "grad_norm": 2.749601364135742, "learning_rate": 0.0005, "loss": 2.1032791137695312, "step": 8690 }, { "epoch": 1.0112869443797998, "grad_norm": 0.9403076171875, "learning_rate": 0.0005, "loss": 1.6642305850982666, "step": 8691 }, { "epoch": 1.011403304631138, "grad_norm": 1.2796220779418945, "learning_rate": 0.0005, "loss": 2.1422951221466064, "step": 8692 }, { "epoch": 1.0115196648824762, "grad_norm": 1.2958098649978638, "learning_rate": 0.0005, "loss": 1.5136497020721436, "step": 8693 }, { "epoch": 1.0116360251338143, "grad_norm": 0.7152499556541443, "learning_rate": 0.0005, "loss": 1.9211314916610718, "step": 8694 }, { "epoch": 1.0117523853851524, "grad_norm": 1.046525001525879, "learning_rate": 0.0005, "loss": 1.9907455444335938, "step": 8695 }, { "epoch": 1.0118687456364905, "grad_norm": 1.20867121219635, "learning_rate": 0.0005, "loss": 2.3350226879119873, "step": 8696 }, { "epoch": 1.0119851058878286, "grad_norm": 1.9845356941223145, "learning_rate": 0.0005, "loss": 2.14241361618042, "step": 8697 }, { "epoch": 1.012101466139167, "grad_norm": 1.4156523942947388, "learning_rate": 0.0005, "loss": 1.4378612041473389, "step": 8698 }, { "epoch": 1.012217826390505, "grad_norm": 1.263974666595459, "learning_rate": 0.0005, "loss": 1.4603254795074463, "step": 8699 }, { "epoch": 1.0123341866418432, "grad_norm": 1.3269107341766357, "learning_rate": 0.0005, "loss": 1.8596928119659424, "step": 8700 }, { "epoch": 1.0124505468931813, "grad_norm": 1.5318222045898438, "learning_rate": 0.0005, "loss": 2.383314609527588, "step": 8701 }, { "epoch": 1.0125669071445194, "grad_norm": 0.8257607221603394, "learning_rate": 0.0005, "loss": 2.214027166366577, "step": 8702 }, { "epoch": 1.0126832673958577, "grad_norm": 1.070818305015564, "learning_rate": 0.0005, "loss": 1.7044954299926758, "step": 8703 }, { "epoch": 1.0127996276471958, "grad_norm": 1.251967191696167, "learning_rate": 0.0005, "loss": 2.5318703651428223, "step": 8704 }, { "epoch": 1.0129159878985339, "grad_norm": 1.0153473615646362, "learning_rate": 0.0005, "loss": 1.9997806549072266, "step": 8705 }, { "epoch": 1.013032348149872, "grad_norm": 0.9613564610481262, "learning_rate": 0.0005, "loss": 2.446380138397217, "step": 8706 }, { "epoch": 1.01314870840121, "grad_norm": 0.9035724401473999, "learning_rate": 0.0005, "loss": 1.762995719909668, "step": 8707 }, { "epoch": 1.0132650686525484, "grad_norm": 0.8695595860481262, "learning_rate": 0.0005, "loss": 1.2219059467315674, "step": 8708 }, { "epoch": 1.0133814289038865, "grad_norm": 1.084744930267334, "learning_rate": 0.0005, "loss": 1.782224178314209, "step": 8709 }, { "epoch": 1.0134977891552246, "grad_norm": 1.2793264389038086, "learning_rate": 0.0005, "loss": 2.18569278717041, "step": 8710 }, { "epoch": 1.0136141494065627, "grad_norm": 1.0176461935043335, "learning_rate": 0.0005, "loss": 2.189924716949463, "step": 8711 }, { "epoch": 1.0137305096579008, "grad_norm": 0.8664209842681885, "learning_rate": 0.0005, "loss": 1.5532000064849854, "step": 8712 }, { "epoch": 1.0138468699092391, "grad_norm": 1.3514589071273804, "learning_rate": 0.0005, "loss": 1.3524699211120605, "step": 8713 }, { "epoch": 1.0139632301605772, "grad_norm": 0.7960562109947205, "learning_rate": 0.0005, "loss": 1.6593425273895264, "step": 8714 }, { "epoch": 1.0140795904119153, "grad_norm": 1.5115633010864258, "learning_rate": 0.0005, "loss": 2.50799560546875, "step": 8715 }, { "epoch": 1.0141959506632534, "grad_norm": 1.6448404788970947, "learning_rate": 0.0005, "loss": 2.0033888816833496, "step": 8716 }, { "epoch": 1.0143123109145915, "grad_norm": 0.8753646612167358, "learning_rate": 0.0005, "loss": 2.385709047317505, "step": 8717 }, { "epoch": 1.0144286711659296, "grad_norm": 1.2111570835113525, "learning_rate": 0.0005, "loss": 2.9162850379943848, "step": 8718 }, { "epoch": 1.014545031417268, "grad_norm": 1.9438579082489014, "learning_rate": 0.0005, "loss": 2.691680431365967, "step": 8719 }, { "epoch": 1.014661391668606, "grad_norm": 1.3877811431884766, "learning_rate": 0.0005, "loss": 1.8563013076782227, "step": 8720 }, { "epoch": 1.0147777519199441, "grad_norm": 0.9769173264503479, "learning_rate": 0.0005, "loss": 2.03837251663208, "step": 8721 }, { "epoch": 1.0148941121712822, "grad_norm": 1.2963042259216309, "learning_rate": 0.0005, "loss": 2.71354341506958, "step": 8722 }, { "epoch": 1.0150104724226203, "grad_norm": 1.5252835750579834, "learning_rate": 0.0005, "loss": 2.054509401321411, "step": 8723 }, { "epoch": 1.0151268326739586, "grad_norm": 1.1245940923690796, "learning_rate": 0.0005, "loss": 1.9857139587402344, "step": 8724 }, { "epoch": 1.0152431929252967, "grad_norm": 1.0362112522125244, "learning_rate": 0.0005, "loss": 1.6597120761871338, "step": 8725 }, { "epoch": 1.0153595531766348, "grad_norm": 1.1934704780578613, "learning_rate": 0.0005, "loss": 1.5690693855285645, "step": 8726 }, { "epoch": 1.015475913427973, "grad_norm": 0.928954005241394, "learning_rate": 0.0005, "loss": 1.6348634958267212, "step": 8727 }, { "epoch": 1.015592273679311, "grad_norm": 1.4124282598495483, "learning_rate": 0.0005, "loss": 2.660780906677246, "step": 8728 }, { "epoch": 1.0157086339306494, "grad_norm": 1.0052192211151123, "learning_rate": 0.0005, "loss": 1.609391450881958, "step": 8729 }, { "epoch": 1.0158249941819875, "grad_norm": 1.0079346895217896, "learning_rate": 0.0005, "loss": 1.7518408298492432, "step": 8730 }, { "epoch": 1.0159413544333256, "grad_norm": 0.8931841254234314, "learning_rate": 0.0005, "loss": 1.7896604537963867, "step": 8731 }, { "epoch": 1.0160577146846637, "grad_norm": 1.2201777696609497, "learning_rate": 0.0005, "loss": 2.455334186553955, "step": 8732 }, { "epoch": 1.0161740749360018, "grad_norm": 1.3120982646942139, "learning_rate": 0.0005, "loss": 1.9759997129440308, "step": 8733 }, { "epoch": 1.01629043518734, "grad_norm": 1.020248293876648, "learning_rate": 0.0005, "loss": 1.4932067394256592, "step": 8734 }, { "epoch": 1.0164067954386782, "grad_norm": 1.0267874002456665, "learning_rate": 0.0005, "loss": 1.4976286888122559, "step": 8735 }, { "epoch": 1.0165231556900163, "grad_norm": 0.8302260041236877, "learning_rate": 0.0005, "loss": 1.8382363319396973, "step": 8736 }, { "epoch": 1.0166395159413544, "grad_norm": 1.1493616104125977, "learning_rate": 0.0005, "loss": 1.8365684747695923, "step": 8737 }, { "epoch": 1.0167558761926925, "grad_norm": 3.68015456199646, "learning_rate": 0.0005, "loss": 1.31101655960083, "step": 8738 }, { "epoch": 1.0168722364440308, "grad_norm": 0.9189639091491699, "learning_rate": 0.0005, "loss": 1.9765942096710205, "step": 8739 }, { "epoch": 1.016988596695369, "grad_norm": 0.5918366312980652, "learning_rate": 0.0005, "loss": 1.236654281616211, "step": 8740 }, { "epoch": 1.017104956946707, "grad_norm": 1.161927580833435, "learning_rate": 0.0005, "loss": 1.7697420120239258, "step": 8741 }, { "epoch": 1.017221317198045, "grad_norm": 1.1517665386199951, "learning_rate": 0.0005, "loss": 1.5232305526733398, "step": 8742 }, { "epoch": 1.0173376774493832, "grad_norm": 1.3843375444412231, "learning_rate": 0.0005, "loss": 2.4086601734161377, "step": 8743 }, { "epoch": 1.0174540377007215, "grad_norm": 1.030274510383606, "learning_rate": 0.0005, "loss": 1.940367579460144, "step": 8744 }, { "epoch": 1.0175703979520596, "grad_norm": 1.0073418617248535, "learning_rate": 0.0005, "loss": 2.4770455360412598, "step": 8745 }, { "epoch": 1.0176867582033977, "grad_norm": 1.2838090658187866, "learning_rate": 0.0005, "loss": 2.397442102432251, "step": 8746 }, { "epoch": 1.0178031184547358, "grad_norm": 1.255005121231079, "learning_rate": 0.0005, "loss": 2.4686615467071533, "step": 8747 }, { "epoch": 1.017919478706074, "grad_norm": 0.7242392897605896, "learning_rate": 0.0005, "loss": 1.4493787288665771, "step": 8748 }, { "epoch": 1.0180358389574122, "grad_norm": 1.0156829357147217, "learning_rate": 0.0005, "loss": 2.0730700492858887, "step": 8749 }, { "epoch": 1.0181521992087503, "grad_norm": 1.715233564376831, "learning_rate": 0.0005, "loss": 2.6165213584899902, "step": 8750 }, { "epoch": 1.0182685594600884, "grad_norm": 1.1859203577041626, "learning_rate": 0.0005, "loss": 2.129706621170044, "step": 8751 }, { "epoch": 1.0183849197114265, "grad_norm": 1.2070728540420532, "learning_rate": 0.0005, "loss": 2.206031560897827, "step": 8752 }, { "epoch": 1.0185012799627646, "grad_norm": 1.1673563718795776, "learning_rate": 0.0005, "loss": 1.8033857345581055, "step": 8753 }, { "epoch": 1.018617640214103, "grad_norm": 0.7357473373413086, "learning_rate": 0.0005, "loss": 1.6044224500656128, "step": 8754 }, { "epoch": 1.018734000465441, "grad_norm": 2.804664373397827, "learning_rate": 0.0005, "loss": 1.7982168197631836, "step": 8755 }, { "epoch": 1.0188503607167791, "grad_norm": 1.2115191221237183, "learning_rate": 0.0005, "loss": 2.156132698059082, "step": 8756 }, { "epoch": 1.0189667209681172, "grad_norm": 1.2767751216888428, "learning_rate": 0.0005, "loss": 1.4498610496520996, "step": 8757 }, { "epoch": 1.0190830812194553, "grad_norm": 1.1251317262649536, "learning_rate": 0.0005, "loss": 2.5033516883850098, "step": 8758 }, { "epoch": 1.0191994414707937, "grad_norm": 2.0853524208068848, "learning_rate": 0.0005, "loss": 1.985145926475525, "step": 8759 }, { "epoch": 1.0193158017221318, "grad_norm": 1.3107337951660156, "learning_rate": 0.0005, "loss": 1.99714195728302, "step": 8760 }, { "epoch": 1.0194321619734699, "grad_norm": 0.7997744083404541, "learning_rate": 0.0005, "loss": 1.2573943138122559, "step": 8761 }, { "epoch": 1.019548522224808, "grad_norm": 1.437821388244629, "learning_rate": 0.0005, "loss": 1.8080451488494873, "step": 8762 }, { "epoch": 1.019664882476146, "grad_norm": 1.1929548978805542, "learning_rate": 0.0005, "loss": 2.104738712310791, "step": 8763 }, { "epoch": 1.0197812427274844, "grad_norm": 1.0615884065628052, "learning_rate": 0.0005, "loss": 1.8231792449951172, "step": 8764 }, { "epoch": 1.0198976029788225, "grad_norm": 2.372457981109619, "learning_rate": 0.0005, "loss": 1.5628235340118408, "step": 8765 }, { "epoch": 1.0200139632301606, "grad_norm": 1.1181837320327759, "learning_rate": 0.0005, "loss": 2.0632100105285645, "step": 8766 }, { "epoch": 1.0201303234814987, "grad_norm": 1.0236198902130127, "learning_rate": 0.0005, "loss": 2.3677525520324707, "step": 8767 }, { "epoch": 1.0202466837328368, "grad_norm": 1.1470674276351929, "learning_rate": 0.0005, "loss": 2.1053647994995117, "step": 8768 }, { "epoch": 1.020363043984175, "grad_norm": 1.1694879531860352, "learning_rate": 0.0005, "loss": 2.538064479827881, "step": 8769 }, { "epoch": 1.0204794042355132, "grad_norm": 1.2776293754577637, "learning_rate": 0.0005, "loss": 2.770717144012451, "step": 8770 }, { "epoch": 1.0205957644868513, "grad_norm": 1.1600664854049683, "learning_rate": 0.0005, "loss": 2.459319591522217, "step": 8771 }, { "epoch": 1.0207121247381894, "grad_norm": 0.8850054144859314, "learning_rate": 0.0005, "loss": 1.932140588760376, "step": 8772 }, { "epoch": 1.0208284849895275, "grad_norm": 1.034053087234497, "learning_rate": 0.0005, "loss": 1.9327040910720825, "step": 8773 }, { "epoch": 1.0209448452408658, "grad_norm": 0.9826325178146362, "learning_rate": 0.0005, "loss": 2.081477642059326, "step": 8774 }, { "epoch": 1.021061205492204, "grad_norm": 1.6845725774765015, "learning_rate": 0.0005, "loss": 1.5385066270828247, "step": 8775 }, { "epoch": 1.021177565743542, "grad_norm": 0.6910210847854614, "learning_rate": 0.0005, "loss": 1.3746885061264038, "step": 8776 }, { "epoch": 1.0212939259948801, "grad_norm": 1.583825707435608, "learning_rate": 0.0005, "loss": 2.1157243251800537, "step": 8777 }, { "epoch": 1.0214102862462182, "grad_norm": 1.7314786911010742, "learning_rate": 0.0005, "loss": 2.252131938934326, "step": 8778 }, { "epoch": 1.0215266464975565, "grad_norm": 1.3113096952438354, "learning_rate": 0.0005, "loss": 1.9676892757415771, "step": 8779 }, { "epoch": 1.0216430067488946, "grad_norm": 1.0287972688674927, "learning_rate": 0.0005, "loss": 1.593350887298584, "step": 8780 }, { "epoch": 1.0217593670002327, "grad_norm": 1.3785991668701172, "learning_rate": 0.0005, "loss": 2.1322402954101562, "step": 8781 }, { "epoch": 1.0218757272515708, "grad_norm": 0.9721459746360779, "learning_rate": 0.0005, "loss": 1.7145373821258545, "step": 8782 }, { "epoch": 1.021992087502909, "grad_norm": 0.9026914834976196, "learning_rate": 0.0005, "loss": 2.3413655757904053, "step": 8783 }, { "epoch": 1.0221084477542473, "grad_norm": 0.9586659073829651, "learning_rate": 0.0005, "loss": 2.3669180870056152, "step": 8784 }, { "epoch": 1.0222248080055854, "grad_norm": 1.0438166856765747, "learning_rate": 0.0005, "loss": 1.599652647972107, "step": 8785 }, { "epoch": 1.0223411682569234, "grad_norm": 2.07757568359375, "learning_rate": 0.0005, "loss": 1.2400827407836914, "step": 8786 }, { "epoch": 1.0224575285082615, "grad_norm": 1.1167513132095337, "learning_rate": 0.0005, "loss": 1.9161468744277954, "step": 8787 }, { "epoch": 1.0225738887595996, "grad_norm": 1.073239803314209, "learning_rate": 0.0005, "loss": 2.2548346519470215, "step": 8788 }, { "epoch": 1.022690249010938, "grad_norm": 1.1663821935653687, "learning_rate": 0.0005, "loss": 2.3975071907043457, "step": 8789 }, { "epoch": 1.022806609262276, "grad_norm": 2.086670160293579, "learning_rate": 0.0005, "loss": 1.9309742450714111, "step": 8790 }, { "epoch": 1.0229229695136142, "grad_norm": 1.215800166130066, "learning_rate": 0.0005, "loss": 2.2451114654541016, "step": 8791 }, { "epoch": 1.0230393297649523, "grad_norm": 1.462756633758545, "learning_rate": 0.0005, "loss": 2.1268649101257324, "step": 8792 }, { "epoch": 1.0231556900162904, "grad_norm": 0.7173197865486145, "learning_rate": 0.0005, "loss": 1.4622894525527954, "step": 8793 }, { "epoch": 1.0232720502676287, "grad_norm": 0.9832972288131714, "learning_rate": 0.0005, "loss": 2.0124640464782715, "step": 8794 }, { "epoch": 1.0233884105189668, "grad_norm": 1.561681866645813, "learning_rate": 0.0005, "loss": 2.228015661239624, "step": 8795 }, { "epoch": 1.0235047707703049, "grad_norm": 0.9020715951919556, "learning_rate": 0.0005, "loss": 1.4742202758789062, "step": 8796 }, { "epoch": 1.023621131021643, "grad_norm": 0.7762868404388428, "learning_rate": 0.0005, "loss": 1.9052300453186035, "step": 8797 }, { "epoch": 1.023737491272981, "grad_norm": 1.0279121398925781, "learning_rate": 0.0005, "loss": 1.2964351177215576, "step": 8798 }, { "epoch": 1.0238538515243194, "grad_norm": 0.8904324173927307, "learning_rate": 0.0005, "loss": 2.3838090896606445, "step": 8799 }, { "epoch": 1.0239702117756575, "grad_norm": 1.0678473711013794, "learning_rate": 0.0005, "loss": 2.0473904609680176, "step": 8800 }, { "epoch": 1.0240865720269956, "grad_norm": 2.810816526412964, "learning_rate": 0.0005, "loss": 2.3486549854278564, "step": 8801 }, { "epoch": 1.0242029322783337, "grad_norm": 1.1278778314590454, "learning_rate": 0.0005, "loss": 1.5400663614273071, "step": 8802 }, { "epoch": 1.0243192925296718, "grad_norm": 1.770579218864441, "learning_rate": 0.0005, "loss": 2.0290651321411133, "step": 8803 }, { "epoch": 1.02443565278101, "grad_norm": 1.2005168199539185, "learning_rate": 0.0005, "loss": 2.4923572540283203, "step": 8804 }, { "epoch": 1.0245520130323482, "grad_norm": 1.1470297574996948, "learning_rate": 0.0005, "loss": 2.096336841583252, "step": 8805 }, { "epoch": 1.0246683732836863, "grad_norm": 0.801318347454071, "learning_rate": 0.0005, "loss": 1.7084563970565796, "step": 8806 }, { "epoch": 1.0247847335350244, "grad_norm": 0.7798870205879211, "learning_rate": 0.0005, "loss": 1.690786600112915, "step": 8807 }, { "epoch": 1.0249010937863625, "grad_norm": 1.2002754211425781, "learning_rate": 0.0005, "loss": 2.519439697265625, "step": 8808 }, { "epoch": 1.0250174540377006, "grad_norm": 0.7680034637451172, "learning_rate": 0.0005, "loss": 2.237950563430786, "step": 8809 }, { "epoch": 1.025133814289039, "grad_norm": 1.177504539489746, "learning_rate": 0.0005, "loss": 2.9929237365722656, "step": 8810 }, { "epoch": 1.025250174540377, "grad_norm": 0.8747479319572449, "learning_rate": 0.0005, "loss": 1.7215301990509033, "step": 8811 }, { "epoch": 1.0253665347917151, "grad_norm": 1.3239479064941406, "learning_rate": 0.0005, "loss": 2.54545259475708, "step": 8812 }, { "epoch": 1.0254828950430532, "grad_norm": 0.9428069591522217, "learning_rate": 0.0005, "loss": 1.9939765930175781, "step": 8813 }, { "epoch": 1.0255992552943913, "grad_norm": 1.1058461666107178, "learning_rate": 0.0005, "loss": 2.2701432704925537, "step": 8814 }, { "epoch": 1.0257156155457297, "grad_norm": 1.2364360094070435, "learning_rate": 0.0005, "loss": 2.2008309364318848, "step": 8815 }, { "epoch": 1.0258319757970678, "grad_norm": 0.8598096370697021, "learning_rate": 0.0005, "loss": 2.2772216796875, "step": 8816 }, { "epoch": 1.0259483360484059, "grad_norm": 1.1691920757293701, "learning_rate": 0.0005, "loss": 1.6720010042190552, "step": 8817 }, { "epoch": 1.026064696299744, "grad_norm": 1.1421003341674805, "learning_rate": 0.0005, "loss": 1.8047802448272705, "step": 8818 }, { "epoch": 1.026181056551082, "grad_norm": 1.1103969812393188, "learning_rate": 0.0005, "loss": 2.1459436416625977, "step": 8819 }, { "epoch": 1.0262974168024204, "grad_norm": 1.345664381980896, "learning_rate": 0.0005, "loss": 2.2174980640411377, "step": 8820 }, { "epoch": 1.0264137770537585, "grad_norm": 1.2948657274246216, "learning_rate": 0.0005, "loss": 1.9031428098678589, "step": 8821 }, { "epoch": 1.0265301373050966, "grad_norm": 1.3974299430847168, "learning_rate": 0.0005, "loss": 2.041713237762451, "step": 8822 }, { "epoch": 1.0266464975564347, "grad_norm": 1.4217824935913086, "learning_rate": 0.0005, "loss": 2.031106948852539, "step": 8823 }, { "epoch": 1.0267628578077728, "grad_norm": 2.3034164905548096, "learning_rate": 0.0005, "loss": 2.0274271965026855, "step": 8824 }, { "epoch": 1.026879218059111, "grad_norm": 1.5382938385009766, "learning_rate": 0.0005, "loss": 2.7892613410949707, "step": 8825 }, { "epoch": 1.0269955783104492, "grad_norm": 3.005507230758667, "learning_rate": 0.0005, "loss": 2.0923638343811035, "step": 8826 }, { "epoch": 1.0271119385617873, "grad_norm": 1.2319368124008179, "learning_rate": 0.0005, "loss": 2.0757219791412354, "step": 8827 }, { "epoch": 1.0272282988131254, "grad_norm": 1.2654407024383545, "learning_rate": 0.0005, "loss": 2.131578207015991, "step": 8828 }, { "epoch": 1.0273446590644635, "grad_norm": 0.9965867400169373, "learning_rate": 0.0005, "loss": 1.5651514530181885, "step": 8829 }, { "epoch": 1.0274610193158018, "grad_norm": 0.9853618741035461, "learning_rate": 0.0005, "loss": 1.7292814254760742, "step": 8830 }, { "epoch": 1.02757737956714, "grad_norm": 0.5346775650978088, "learning_rate": 0.0005, "loss": 1.1507762670516968, "step": 8831 }, { "epoch": 1.027693739818478, "grad_norm": 0.9194213151931763, "learning_rate": 0.0005, "loss": 1.7498944997787476, "step": 8832 }, { "epoch": 1.027810100069816, "grad_norm": 0.7719211578369141, "learning_rate": 0.0005, "loss": 1.8470546007156372, "step": 8833 }, { "epoch": 1.0279264603211542, "grad_norm": 1.0575878620147705, "learning_rate": 0.0005, "loss": 1.5483752489089966, "step": 8834 }, { "epoch": 1.0280428205724925, "grad_norm": 1.1471678018569946, "learning_rate": 0.0005, "loss": 2.145608425140381, "step": 8835 }, { "epoch": 1.0281591808238306, "grad_norm": 1.5784385204315186, "learning_rate": 0.0005, "loss": 1.344956398010254, "step": 8836 }, { "epoch": 1.0282755410751687, "grad_norm": 1.0702850818634033, "learning_rate": 0.0005, "loss": 2.614536762237549, "step": 8837 }, { "epoch": 1.0283919013265068, "grad_norm": 1.2670351266860962, "learning_rate": 0.0005, "loss": 1.710679531097412, "step": 8838 }, { "epoch": 1.028508261577845, "grad_norm": 1.7835029363632202, "learning_rate": 0.0005, "loss": 2.0829262733459473, "step": 8839 }, { "epoch": 1.0286246218291832, "grad_norm": 1.438394546508789, "learning_rate": 0.0005, "loss": 2.299283027648926, "step": 8840 }, { "epoch": 1.0287409820805213, "grad_norm": 1.3322503566741943, "learning_rate": 0.0005, "loss": 1.790391445159912, "step": 8841 }, { "epoch": 1.0288573423318594, "grad_norm": 1.4952430725097656, "learning_rate": 0.0005, "loss": 1.3037514686584473, "step": 8842 }, { "epoch": 1.0289737025831975, "grad_norm": 1.0672101974487305, "learning_rate": 0.0005, "loss": 1.4624000787734985, "step": 8843 }, { "epoch": 1.0290900628345356, "grad_norm": 1.2400400638580322, "learning_rate": 0.0005, "loss": 1.8472192287445068, "step": 8844 }, { "epoch": 1.029206423085874, "grad_norm": 0.8994589447975159, "learning_rate": 0.0005, "loss": 1.8713786602020264, "step": 8845 }, { "epoch": 1.029322783337212, "grad_norm": 0.9060514569282532, "learning_rate": 0.0005, "loss": 2.6754651069641113, "step": 8846 }, { "epoch": 1.0294391435885502, "grad_norm": 6.830418109893799, "learning_rate": 0.0005, "loss": 2.2399368286132812, "step": 8847 }, { "epoch": 1.0295555038398883, "grad_norm": 0.9463400840759277, "learning_rate": 0.0005, "loss": 1.5439739227294922, "step": 8848 }, { "epoch": 1.0296718640912264, "grad_norm": 0.8200202584266663, "learning_rate": 0.0005, "loss": 2.069061279296875, "step": 8849 }, { "epoch": 1.0297882243425647, "grad_norm": 1.1884373426437378, "learning_rate": 0.0005, "loss": 2.553317070007324, "step": 8850 }, { "epoch": 1.0299045845939028, "grad_norm": 3.156261682510376, "learning_rate": 0.0005, "loss": 2.3792216777801514, "step": 8851 }, { "epoch": 1.0300209448452409, "grad_norm": 1.3002748489379883, "learning_rate": 0.0005, "loss": 2.0174546241760254, "step": 8852 }, { "epoch": 1.030137305096579, "grad_norm": 0.8897709846496582, "learning_rate": 0.0005, "loss": 1.6750621795654297, "step": 8853 }, { "epoch": 1.030253665347917, "grad_norm": 1.9937121868133545, "learning_rate": 0.0005, "loss": 2.6447677612304688, "step": 8854 }, { "epoch": 1.0303700255992554, "grad_norm": 1.2707326412200928, "learning_rate": 0.0005, "loss": 1.1736412048339844, "step": 8855 }, { "epoch": 1.0304863858505935, "grad_norm": 1.1320126056671143, "learning_rate": 0.0005, "loss": 1.9519805908203125, "step": 8856 }, { "epoch": 1.0306027461019316, "grad_norm": 1.3525100946426392, "learning_rate": 0.0005, "loss": 2.1031360626220703, "step": 8857 }, { "epoch": 1.0307191063532697, "grad_norm": 1.19904363155365, "learning_rate": 0.0005, "loss": 2.0536112785339355, "step": 8858 }, { "epoch": 1.0308354666046078, "grad_norm": 1.5085654258728027, "learning_rate": 0.0005, "loss": 1.9930529594421387, "step": 8859 }, { "epoch": 1.030951826855946, "grad_norm": 1.4100874662399292, "learning_rate": 0.0005, "loss": 2.8186442852020264, "step": 8860 }, { "epoch": 1.0310681871072842, "grad_norm": 1.0176242589950562, "learning_rate": 0.0005, "loss": 1.8553783893585205, "step": 8861 }, { "epoch": 1.0311845473586223, "grad_norm": 0.9677698612213135, "learning_rate": 0.0005, "loss": 1.2305841445922852, "step": 8862 }, { "epoch": 1.0313009076099604, "grad_norm": 1.0545644760131836, "learning_rate": 0.0005, "loss": 2.2466413974761963, "step": 8863 }, { "epoch": 1.0314172678612985, "grad_norm": 1.346238136291504, "learning_rate": 0.0005, "loss": 2.5175914764404297, "step": 8864 }, { "epoch": 1.0315336281126368, "grad_norm": 0.5149098634719849, "learning_rate": 0.0005, "loss": 1.335688829421997, "step": 8865 }, { "epoch": 1.031649988363975, "grad_norm": 0.6412691473960876, "learning_rate": 0.0005, "loss": 1.4227948188781738, "step": 8866 }, { "epoch": 1.031766348615313, "grad_norm": 0.8594722151756287, "learning_rate": 0.0005, "loss": 2.1816482543945312, "step": 8867 }, { "epoch": 1.0318827088666511, "grad_norm": 0.7406394481658936, "learning_rate": 0.0005, "loss": 1.154698371887207, "step": 8868 }, { "epoch": 1.0319990691179892, "grad_norm": 0.8583954572677612, "learning_rate": 0.0005, "loss": 2.0251801013946533, "step": 8869 }, { "epoch": 1.0321154293693275, "grad_norm": 1.290236473083496, "learning_rate": 0.0005, "loss": 2.4426980018615723, "step": 8870 }, { "epoch": 1.0322317896206656, "grad_norm": 1.2926950454711914, "learning_rate": 0.0005, "loss": 2.5695815086364746, "step": 8871 }, { "epoch": 1.0323481498720037, "grad_norm": 0.7954239249229431, "learning_rate": 0.0005, "loss": 2.273815631866455, "step": 8872 }, { "epoch": 1.0324645101233418, "grad_norm": 1.3536527156829834, "learning_rate": 0.0005, "loss": 2.082700490951538, "step": 8873 }, { "epoch": 1.03258087037468, "grad_norm": 1.8165581226348877, "learning_rate": 0.0005, "loss": 2.1972639560699463, "step": 8874 }, { "epoch": 1.0326972306260183, "grad_norm": 1.4935382604599, "learning_rate": 0.0005, "loss": 1.3503942489624023, "step": 8875 }, { "epoch": 1.0328135908773564, "grad_norm": 1.2258541584014893, "learning_rate": 0.0005, "loss": 1.7986700534820557, "step": 8876 }, { "epoch": 1.0329299511286945, "grad_norm": 1.0997990369796753, "learning_rate": 0.0005, "loss": 2.6963419914245605, "step": 8877 }, { "epoch": 1.0330463113800326, "grad_norm": 1.3939651250839233, "learning_rate": 0.0005, "loss": 2.3661813735961914, "step": 8878 }, { "epoch": 1.0331626716313707, "grad_norm": 1.2127236127853394, "learning_rate": 0.0005, "loss": 1.2543631792068481, "step": 8879 }, { "epoch": 1.0332790318827088, "grad_norm": 1.6395808458328247, "learning_rate": 0.0005, "loss": 1.8936917781829834, "step": 8880 }, { "epoch": 1.033395392134047, "grad_norm": 1.4654942750930786, "learning_rate": 0.0005, "loss": 2.0434482097625732, "step": 8881 }, { "epoch": 1.0335117523853852, "grad_norm": 1.0720324516296387, "learning_rate": 0.0005, "loss": 2.362231492996216, "step": 8882 }, { "epoch": 1.0336281126367233, "grad_norm": 0.9954423308372498, "learning_rate": 0.0005, "loss": 2.0800204277038574, "step": 8883 }, { "epoch": 1.0337444728880614, "grad_norm": 1.042142391204834, "learning_rate": 0.0005, "loss": 1.438946008682251, "step": 8884 }, { "epoch": 1.0338608331393995, "grad_norm": 0.8838745355606079, "learning_rate": 0.0005, "loss": 2.400132417678833, "step": 8885 }, { "epoch": 1.0339771933907378, "grad_norm": 1.1061841249465942, "learning_rate": 0.0005, "loss": 1.768646001815796, "step": 8886 }, { "epoch": 1.034093553642076, "grad_norm": 1.4769954681396484, "learning_rate": 0.0005, "loss": 2.0478382110595703, "step": 8887 }, { "epoch": 1.034209913893414, "grad_norm": 1.1962800025939941, "learning_rate": 0.0005, "loss": 2.2718825340270996, "step": 8888 }, { "epoch": 1.034326274144752, "grad_norm": 0.9796128869056702, "learning_rate": 0.0005, "loss": 2.194976329803467, "step": 8889 }, { "epoch": 1.0344426343960902, "grad_norm": 0.4523630738258362, "learning_rate": 0.0005, "loss": 0.993510901927948, "step": 8890 }, { "epoch": 1.0345589946474285, "grad_norm": 1.7653694152832031, "learning_rate": 0.0005, "loss": 2.7415237426757812, "step": 8891 }, { "epoch": 1.0346753548987666, "grad_norm": 1.4686323404312134, "learning_rate": 0.0005, "loss": 2.254905939102173, "step": 8892 }, { "epoch": 1.0347917151501047, "grad_norm": 2.123372793197632, "learning_rate": 0.0005, "loss": 1.919787049293518, "step": 8893 }, { "epoch": 1.0349080754014428, "grad_norm": 1.191489338874817, "learning_rate": 0.0005, "loss": 1.8079826831817627, "step": 8894 }, { "epoch": 1.035024435652781, "grad_norm": 1.3363782167434692, "learning_rate": 0.0005, "loss": 1.9060821533203125, "step": 8895 }, { "epoch": 1.0351407959041192, "grad_norm": 1.315615177154541, "learning_rate": 0.0005, "loss": 1.9175063371658325, "step": 8896 }, { "epoch": 1.0352571561554573, "grad_norm": 1.0641460418701172, "learning_rate": 0.0005, "loss": 2.2965219020843506, "step": 8897 }, { "epoch": 1.0353735164067954, "grad_norm": 1.948502779006958, "learning_rate": 0.0005, "loss": 2.289139747619629, "step": 8898 }, { "epoch": 1.0354898766581335, "grad_norm": 0.893547773361206, "learning_rate": 0.0005, "loss": 1.5084844827651978, "step": 8899 }, { "epoch": 1.0356062369094716, "grad_norm": 1.4230011701583862, "learning_rate": 0.0005, "loss": 1.7947001457214355, "step": 8900 }, { "epoch": 1.03572259716081, "grad_norm": 0.995140016078949, "learning_rate": 0.0005, "loss": 1.6059019565582275, "step": 8901 }, { "epoch": 1.035838957412148, "grad_norm": 1.5038926601409912, "learning_rate": 0.0005, "loss": 2.1721034049987793, "step": 8902 }, { "epoch": 1.0359553176634861, "grad_norm": 0.9034818410873413, "learning_rate": 0.0005, "loss": 1.5531823635101318, "step": 8903 }, { "epoch": 1.0360716779148242, "grad_norm": 1.4934998750686646, "learning_rate": 0.0005, "loss": 2.138857126235962, "step": 8904 }, { "epoch": 1.0361880381661623, "grad_norm": 1.3840968608856201, "learning_rate": 0.0005, "loss": 2.3546700477600098, "step": 8905 }, { "epoch": 1.0363043984175007, "grad_norm": 0.7848337292671204, "learning_rate": 0.0005, "loss": 2.3429696559906006, "step": 8906 }, { "epoch": 1.0364207586688388, "grad_norm": 1.3613313436508179, "learning_rate": 0.0005, "loss": 2.2362587451934814, "step": 8907 }, { "epoch": 1.0365371189201769, "grad_norm": 1.2251988649368286, "learning_rate": 0.0005, "loss": 1.924525499343872, "step": 8908 }, { "epoch": 1.036653479171515, "grad_norm": 1.0238932371139526, "learning_rate": 0.0005, "loss": 1.7790517807006836, "step": 8909 }, { "epoch": 1.036769839422853, "grad_norm": 1.3066328763961792, "learning_rate": 0.0005, "loss": 2.107693672180176, "step": 8910 }, { "epoch": 1.0368861996741914, "grad_norm": 1.4811784029006958, "learning_rate": 0.0005, "loss": 2.361103057861328, "step": 8911 }, { "epoch": 1.0370025599255295, "grad_norm": 0.8730705380439758, "learning_rate": 0.0005, "loss": 2.079131603240967, "step": 8912 }, { "epoch": 1.0371189201768676, "grad_norm": 1.6872785091400146, "learning_rate": 0.0005, "loss": 2.0297691822052, "step": 8913 }, { "epoch": 1.0372352804282057, "grad_norm": 0.8442443013191223, "learning_rate": 0.0005, "loss": 1.3102357387542725, "step": 8914 }, { "epoch": 1.0373516406795438, "grad_norm": 1.3887591361999512, "learning_rate": 0.0005, "loss": 2.5169248580932617, "step": 8915 }, { "epoch": 1.037468000930882, "grad_norm": 1.1515567302703857, "learning_rate": 0.0005, "loss": 2.177870035171509, "step": 8916 }, { "epoch": 1.0375843611822202, "grad_norm": 1.0343254804611206, "learning_rate": 0.0005, "loss": 1.8982032537460327, "step": 8917 }, { "epoch": 1.0377007214335583, "grad_norm": 0.9829516410827637, "learning_rate": 0.0005, "loss": 2.057396650314331, "step": 8918 }, { "epoch": 1.0378170816848964, "grad_norm": 1.0375769138336182, "learning_rate": 0.0005, "loss": 1.7712738513946533, "step": 8919 }, { "epoch": 1.0379334419362345, "grad_norm": 0.8974060416221619, "learning_rate": 0.0005, "loss": 1.635155439376831, "step": 8920 }, { "epoch": 1.0380498021875728, "grad_norm": 1.7585930824279785, "learning_rate": 0.0005, "loss": 1.5327311754226685, "step": 8921 }, { "epoch": 1.038166162438911, "grad_norm": 0.8584159016609192, "learning_rate": 0.0005, "loss": 1.9720215797424316, "step": 8922 }, { "epoch": 1.038282522690249, "grad_norm": 1.3062098026275635, "learning_rate": 0.0005, "loss": 1.624097466468811, "step": 8923 }, { "epoch": 1.038398882941587, "grad_norm": 1.0334399938583374, "learning_rate": 0.0005, "loss": 1.795872449874878, "step": 8924 }, { "epoch": 1.0385152431929252, "grad_norm": 1.2119972705841064, "learning_rate": 0.0005, "loss": 2.3161842823028564, "step": 8925 }, { "epoch": 1.0386316034442635, "grad_norm": 1.2655246257781982, "learning_rate": 0.0005, "loss": 1.88089120388031, "step": 8926 }, { "epoch": 1.0387479636956016, "grad_norm": 0.979088544845581, "learning_rate": 0.0005, "loss": 2.0419416427612305, "step": 8927 }, { "epoch": 1.0388643239469397, "grad_norm": 0.8207539319992065, "learning_rate": 0.0005, "loss": 1.8842034339904785, "step": 8928 }, { "epoch": 1.0389806841982778, "grad_norm": 1.0524228811264038, "learning_rate": 0.0005, "loss": 1.876441240310669, "step": 8929 }, { "epoch": 1.039097044449616, "grad_norm": 1.4035223722457886, "learning_rate": 0.0005, "loss": 2.2686522006988525, "step": 8930 }, { "epoch": 1.0392134047009542, "grad_norm": 1.0763758420944214, "learning_rate": 0.0005, "loss": 2.181586503982544, "step": 8931 }, { "epoch": 1.0393297649522923, "grad_norm": 0.9282170534133911, "learning_rate": 0.0005, "loss": 2.5325570106506348, "step": 8932 }, { "epoch": 1.0394461252036304, "grad_norm": 0.9416851997375488, "learning_rate": 0.0005, "loss": 2.2999398708343506, "step": 8933 }, { "epoch": 1.0395624854549685, "grad_norm": 1.5136266946792603, "learning_rate": 0.0005, "loss": 2.0596230030059814, "step": 8934 }, { "epoch": 1.0396788457063066, "grad_norm": 0.7610506415367126, "learning_rate": 0.0005, "loss": 1.5583405494689941, "step": 8935 }, { "epoch": 1.039795205957645, "grad_norm": 1.0098391771316528, "learning_rate": 0.0005, "loss": 1.5193250179290771, "step": 8936 }, { "epoch": 1.039911566208983, "grad_norm": 1.3513649702072144, "learning_rate": 0.0005, "loss": 1.8073149919509888, "step": 8937 }, { "epoch": 1.0400279264603212, "grad_norm": 0.9541348814964294, "learning_rate": 0.0005, "loss": 1.7703889608383179, "step": 8938 }, { "epoch": 1.0401442867116593, "grad_norm": 0.85873943567276, "learning_rate": 0.0005, "loss": 1.3854272365570068, "step": 8939 }, { "epoch": 1.0402606469629974, "grad_norm": 0.8521105647087097, "learning_rate": 0.0005, "loss": 1.499258041381836, "step": 8940 }, { "epoch": 1.0403770072143357, "grad_norm": 1.4911388158798218, "learning_rate": 0.0005, "loss": 2.4495816230773926, "step": 8941 }, { "epoch": 1.0404933674656738, "grad_norm": 1.0007203817367554, "learning_rate": 0.0005, "loss": 2.4942753314971924, "step": 8942 }, { "epoch": 1.0406097277170119, "grad_norm": 1.3227137327194214, "learning_rate": 0.0005, "loss": 2.7487802505493164, "step": 8943 }, { "epoch": 1.04072608796835, "grad_norm": 3.757406234741211, "learning_rate": 0.0005, "loss": 2.4780826568603516, "step": 8944 }, { "epoch": 1.040842448219688, "grad_norm": 1.5529893636703491, "learning_rate": 0.0005, "loss": 2.22046160697937, "step": 8945 }, { "epoch": 1.0409588084710264, "grad_norm": 1.5308417081832886, "learning_rate": 0.0005, "loss": 1.8281128406524658, "step": 8946 }, { "epoch": 1.0410751687223645, "grad_norm": 1.3354125022888184, "learning_rate": 0.0005, "loss": 2.498471736907959, "step": 8947 }, { "epoch": 1.0411915289737026, "grad_norm": 0.854310154914856, "learning_rate": 0.0005, "loss": 1.9929336309432983, "step": 8948 }, { "epoch": 1.0413078892250407, "grad_norm": 2.202669382095337, "learning_rate": 0.0005, "loss": 1.308229923248291, "step": 8949 }, { "epoch": 1.0414242494763788, "grad_norm": 1.3068608045578003, "learning_rate": 0.0005, "loss": 2.410658121109009, "step": 8950 }, { "epoch": 1.0415406097277171, "grad_norm": 1.237275242805481, "learning_rate": 0.0005, "loss": 2.434530735015869, "step": 8951 }, { "epoch": 1.0416569699790552, "grad_norm": 1.0886156558990479, "learning_rate": 0.0005, "loss": 1.8910900354385376, "step": 8952 }, { "epoch": 1.0417733302303933, "grad_norm": 1.2079931497573853, "learning_rate": 0.0005, "loss": 2.6749792098999023, "step": 8953 }, { "epoch": 1.0418896904817314, "grad_norm": 1.1215225458145142, "learning_rate": 0.0005, "loss": 2.1525838375091553, "step": 8954 }, { "epoch": 1.0420060507330695, "grad_norm": 1.0919109582901, "learning_rate": 0.0005, "loss": 2.076422691345215, "step": 8955 }, { "epoch": 1.0421224109844078, "grad_norm": 1.268187403678894, "learning_rate": 0.0005, "loss": 2.015674114227295, "step": 8956 }, { "epoch": 1.042238771235746, "grad_norm": 1.148767113685608, "learning_rate": 0.0005, "loss": 2.9269609451293945, "step": 8957 }, { "epoch": 1.042355131487084, "grad_norm": 0.9695676565170288, "learning_rate": 0.0005, "loss": 1.519567608833313, "step": 8958 }, { "epoch": 1.0424714917384221, "grad_norm": 0.5770400762557983, "learning_rate": 0.0005, "loss": 1.4350223541259766, "step": 8959 }, { "epoch": 1.0425878519897602, "grad_norm": 0.8510339856147766, "learning_rate": 0.0005, "loss": 1.9412485361099243, "step": 8960 }, { "epoch": 1.0427042122410985, "grad_norm": 1.275396466255188, "learning_rate": 0.0005, "loss": 1.8252265453338623, "step": 8961 }, { "epoch": 1.0428205724924366, "grad_norm": 0.9928138852119446, "learning_rate": 0.0005, "loss": 2.3974034786224365, "step": 8962 }, { "epoch": 1.0429369327437747, "grad_norm": 1.1193854808807373, "learning_rate": 0.0005, "loss": 1.4839606285095215, "step": 8963 }, { "epoch": 1.0430532929951128, "grad_norm": 1.4068971872329712, "learning_rate": 0.0005, "loss": 2.220130443572998, "step": 8964 }, { "epoch": 1.043169653246451, "grad_norm": 1.4145872592926025, "learning_rate": 0.0005, "loss": 2.0020551681518555, "step": 8965 }, { "epoch": 1.0432860134977893, "grad_norm": 1.3211784362792969, "learning_rate": 0.0005, "loss": 2.2440576553344727, "step": 8966 }, { "epoch": 1.0434023737491274, "grad_norm": 2.4865217208862305, "learning_rate": 0.0005, "loss": 1.75936758518219, "step": 8967 }, { "epoch": 1.0435187340004655, "grad_norm": 1.6345752477645874, "learning_rate": 0.0005, "loss": 2.0925498008728027, "step": 8968 }, { "epoch": 1.0436350942518036, "grad_norm": 1.0643749237060547, "learning_rate": 0.0005, "loss": 1.9423316717147827, "step": 8969 }, { "epoch": 1.0437514545031417, "grad_norm": 1.0640476942062378, "learning_rate": 0.0005, "loss": 1.811255693435669, "step": 8970 }, { "epoch": 1.04386781475448, "grad_norm": 1.2316577434539795, "learning_rate": 0.0005, "loss": 2.0191893577575684, "step": 8971 }, { "epoch": 1.043984175005818, "grad_norm": 0.872414767742157, "learning_rate": 0.0005, "loss": 2.1128039360046387, "step": 8972 }, { "epoch": 1.0441005352571562, "grad_norm": 1.0821479558944702, "learning_rate": 0.0005, "loss": 1.4844523668289185, "step": 8973 }, { "epoch": 1.0442168955084943, "grad_norm": 0.9422394633293152, "learning_rate": 0.0005, "loss": 1.3482810258865356, "step": 8974 }, { "epoch": 1.0443332557598324, "grad_norm": 1.3151803016662598, "learning_rate": 0.0005, "loss": 2.1830551624298096, "step": 8975 }, { "epoch": 1.0444496160111705, "grad_norm": 2.806490421295166, "learning_rate": 0.0005, "loss": 1.9264545440673828, "step": 8976 }, { "epoch": 1.0445659762625088, "grad_norm": 1.4851983785629272, "learning_rate": 0.0005, "loss": 2.1300430297851562, "step": 8977 }, { "epoch": 1.044682336513847, "grad_norm": 1.1320768594741821, "learning_rate": 0.0005, "loss": 2.803356170654297, "step": 8978 }, { "epoch": 1.044798696765185, "grad_norm": 0.9532802104949951, "learning_rate": 0.0005, "loss": 2.1873979568481445, "step": 8979 }, { "epoch": 1.044915057016523, "grad_norm": 1.1215258836746216, "learning_rate": 0.0005, "loss": 2.321207284927368, "step": 8980 }, { "epoch": 1.0450314172678612, "grad_norm": 0.3137792944908142, "learning_rate": 0.0005, "loss": 0.9562162160873413, "step": 8981 }, { "epoch": 1.0451477775191995, "grad_norm": 0.9891328811645508, "learning_rate": 0.0005, "loss": 1.4678795337677002, "step": 8982 }, { "epoch": 1.0452641377705376, "grad_norm": 0.8354882597923279, "learning_rate": 0.0005, "loss": 1.8356142044067383, "step": 8983 }, { "epoch": 1.0453804980218757, "grad_norm": 0.833428680896759, "learning_rate": 0.0005, "loss": 1.1600289344787598, "step": 8984 }, { "epoch": 1.0454968582732138, "grad_norm": 1.242569923400879, "learning_rate": 0.0005, "loss": 2.358283519744873, "step": 8985 }, { "epoch": 1.045613218524552, "grad_norm": 1.1146270036697388, "learning_rate": 0.0005, "loss": 1.9834579229354858, "step": 8986 }, { "epoch": 1.0457295787758902, "grad_norm": 0.7271438837051392, "learning_rate": 0.0005, "loss": 1.3684258460998535, "step": 8987 }, { "epoch": 1.0458459390272283, "grad_norm": 1.487100601196289, "learning_rate": 0.0005, "loss": 2.0581371784210205, "step": 8988 }, { "epoch": 1.0459622992785664, "grad_norm": 1.058923363685608, "learning_rate": 0.0005, "loss": 2.399409770965576, "step": 8989 }, { "epoch": 1.0460786595299045, "grad_norm": 1.2278709411621094, "learning_rate": 0.0005, "loss": 2.0132503509521484, "step": 8990 }, { "epoch": 1.0461950197812426, "grad_norm": 1.022546648979187, "learning_rate": 0.0005, "loss": 1.8027560710906982, "step": 8991 }, { "epoch": 1.046311380032581, "grad_norm": 1.4520412683486938, "learning_rate": 0.0005, "loss": 2.4026546478271484, "step": 8992 }, { "epoch": 1.046427740283919, "grad_norm": 0.6911661028862, "learning_rate": 0.0005, "loss": 1.4113883972167969, "step": 8993 }, { "epoch": 1.0465441005352571, "grad_norm": 1.1189382076263428, "learning_rate": 0.0005, "loss": 2.4184961318969727, "step": 8994 }, { "epoch": 1.0466604607865952, "grad_norm": 0.958798348903656, "learning_rate": 0.0005, "loss": 1.5483825206756592, "step": 8995 }, { "epoch": 1.0467768210379333, "grad_norm": 1.4312719106674194, "learning_rate": 0.0005, "loss": 2.281799793243408, "step": 8996 }, { "epoch": 1.0468931812892717, "grad_norm": 0.9583410024642944, "learning_rate": 0.0005, "loss": 1.5983357429504395, "step": 8997 }, { "epoch": 1.0470095415406098, "grad_norm": 0.8340486288070679, "learning_rate": 0.0005, "loss": 1.8615646362304688, "step": 8998 }, { "epoch": 1.0471259017919479, "grad_norm": 0.6170753240585327, "learning_rate": 0.0005, "loss": 1.1412838697433472, "step": 8999 }, { "epoch": 1.047242262043286, "grad_norm": 0.7778415083885193, "learning_rate": 0.0005, "loss": 1.6640836000442505, "step": 9000 }, { "epoch": 1.047358622294624, "grad_norm": 1.149247646331787, "learning_rate": 0.0005, "loss": 1.7264249324798584, "step": 9001 }, { "epoch": 1.0474749825459624, "grad_norm": 1.3603330850601196, "learning_rate": 0.0005, "loss": 2.1776885986328125, "step": 9002 }, { "epoch": 1.0475913427973005, "grad_norm": 0.8932783603668213, "learning_rate": 0.0005, "loss": 2.0622451305389404, "step": 9003 }, { "epoch": 1.0477077030486386, "grad_norm": 1.8535897731781006, "learning_rate": 0.0005, "loss": 2.007054328918457, "step": 9004 }, { "epoch": 1.0478240632999767, "grad_norm": 1.87935471534729, "learning_rate": 0.0005, "loss": 2.052096366882324, "step": 9005 }, { "epoch": 1.0479404235513148, "grad_norm": 0.7152826189994812, "learning_rate": 0.0005, "loss": 1.6624538898468018, "step": 9006 }, { "epoch": 1.048056783802653, "grad_norm": 1.10069739818573, "learning_rate": 0.0005, "loss": 1.8682072162628174, "step": 9007 }, { "epoch": 1.0481731440539912, "grad_norm": 1.0042407512664795, "learning_rate": 0.0005, "loss": 2.043294906616211, "step": 9008 }, { "epoch": 1.0482895043053293, "grad_norm": 1.0974748134613037, "learning_rate": 0.0005, "loss": 2.529745101928711, "step": 9009 }, { "epoch": 1.0484058645566674, "grad_norm": 1.0968083143234253, "learning_rate": 0.0005, "loss": 1.6068764925003052, "step": 9010 }, { "epoch": 1.0485222248080055, "grad_norm": 0.7890809178352356, "learning_rate": 0.0005, "loss": 1.5696669816970825, "step": 9011 }, { "epoch": 1.0486385850593438, "grad_norm": 1.2476818561553955, "learning_rate": 0.0005, "loss": 1.8985623121261597, "step": 9012 }, { "epoch": 1.048754945310682, "grad_norm": 1.5293684005737305, "learning_rate": 0.0005, "loss": 2.7323508262634277, "step": 9013 }, { "epoch": 1.04887130556202, "grad_norm": 1.512650728225708, "learning_rate": 0.0005, "loss": 2.509366989135742, "step": 9014 }, { "epoch": 1.0489876658133581, "grad_norm": 0.7261092066764832, "learning_rate": 0.0005, "loss": 1.617976188659668, "step": 9015 }, { "epoch": 1.0491040260646962, "grad_norm": 1.7618004083633423, "learning_rate": 0.0005, "loss": 2.836778163909912, "step": 9016 }, { "epoch": 1.0492203863160345, "grad_norm": 1.166358232498169, "learning_rate": 0.0005, "loss": 2.139292001724243, "step": 9017 }, { "epoch": 1.0493367465673726, "grad_norm": 1.2574807405471802, "learning_rate": 0.0005, "loss": 2.239534854888916, "step": 9018 }, { "epoch": 1.0494531068187107, "grad_norm": 1.0064187049865723, "learning_rate": 0.0005, "loss": 1.9264686107635498, "step": 9019 }, { "epoch": 1.0495694670700488, "grad_norm": 0.9096444845199585, "learning_rate": 0.0005, "loss": 2.2273097038269043, "step": 9020 }, { "epoch": 1.049685827321387, "grad_norm": 0.5904374122619629, "learning_rate": 0.0005, "loss": 1.373968243598938, "step": 9021 }, { "epoch": 1.0498021875727253, "grad_norm": 0.961370050907135, "learning_rate": 0.0005, "loss": 2.058730125427246, "step": 9022 }, { "epoch": 1.0499185478240634, "grad_norm": 0.6577151417732239, "learning_rate": 0.0005, "loss": 1.0796756744384766, "step": 9023 }, { "epoch": 1.0500349080754015, "grad_norm": 1.0522756576538086, "learning_rate": 0.0005, "loss": 2.2489399909973145, "step": 9024 }, { "epoch": 1.0501512683267396, "grad_norm": 0.9810872077941895, "learning_rate": 0.0005, "loss": 2.3011109828948975, "step": 9025 }, { "epoch": 1.0502676285780776, "grad_norm": 1.252941608428955, "learning_rate": 0.0005, "loss": 2.7559731006622314, "step": 9026 }, { "epoch": 1.050383988829416, "grad_norm": 1.1222127676010132, "learning_rate": 0.0005, "loss": 2.140282154083252, "step": 9027 }, { "epoch": 1.050500349080754, "grad_norm": 1.5326017141342163, "learning_rate": 0.0005, "loss": 2.2692863941192627, "step": 9028 }, { "epoch": 1.0506167093320922, "grad_norm": 1.1997790336608887, "learning_rate": 0.0005, "loss": 2.260364294052124, "step": 9029 }, { "epoch": 1.0507330695834303, "grad_norm": 1.343194603919983, "learning_rate": 0.0005, "loss": 1.7464648485183716, "step": 9030 }, { "epoch": 1.0508494298347684, "grad_norm": 0.854306697845459, "learning_rate": 0.0005, "loss": 1.2007083892822266, "step": 9031 }, { "epoch": 1.0509657900861067, "grad_norm": 1.413901686668396, "learning_rate": 0.0005, "loss": 2.1089634895324707, "step": 9032 }, { "epoch": 1.0510821503374448, "grad_norm": 1.397491216659546, "learning_rate": 0.0005, "loss": 2.495889186859131, "step": 9033 }, { "epoch": 1.0511985105887829, "grad_norm": 0.9267702698707581, "learning_rate": 0.0005, "loss": 1.7849462032318115, "step": 9034 }, { "epoch": 1.051314870840121, "grad_norm": 1.2596757411956787, "learning_rate": 0.0005, "loss": 1.8818038702011108, "step": 9035 }, { "epoch": 1.051431231091459, "grad_norm": 0.9453951716423035, "learning_rate": 0.0005, "loss": 1.5267651081085205, "step": 9036 }, { "epoch": 1.0515475913427974, "grad_norm": 1.2756104469299316, "learning_rate": 0.0005, "loss": 2.197099208831787, "step": 9037 }, { "epoch": 1.0516639515941355, "grad_norm": 1.20002019405365, "learning_rate": 0.0005, "loss": 2.1002159118652344, "step": 9038 }, { "epoch": 1.0517803118454736, "grad_norm": 0.7347091436386108, "learning_rate": 0.0005, "loss": 1.071623682975769, "step": 9039 }, { "epoch": 1.0518966720968117, "grad_norm": 0.7768446803092957, "learning_rate": 0.0005, "loss": 1.175012469291687, "step": 9040 }, { "epoch": 1.0520130323481498, "grad_norm": 1.288610577583313, "learning_rate": 0.0005, "loss": 2.3766164779663086, "step": 9041 }, { "epoch": 1.0521293925994881, "grad_norm": 0.8763245344161987, "learning_rate": 0.0005, "loss": 1.7863430976867676, "step": 9042 }, { "epoch": 1.0522457528508262, "grad_norm": 1.0267263650894165, "learning_rate": 0.0005, "loss": 2.0148396492004395, "step": 9043 }, { "epoch": 1.0523621131021643, "grad_norm": 1.2953559160232544, "learning_rate": 0.0005, "loss": 1.8681838512420654, "step": 9044 }, { "epoch": 1.0524784733535024, "grad_norm": 1.0198988914489746, "learning_rate": 0.0005, "loss": 2.6294779777526855, "step": 9045 }, { "epoch": 1.0525948336048405, "grad_norm": 1.0128458738327026, "learning_rate": 0.0005, "loss": 2.1861770153045654, "step": 9046 }, { "epoch": 1.0527111938561786, "grad_norm": 2.4994287490844727, "learning_rate": 0.0005, "loss": 2.328523874282837, "step": 9047 }, { "epoch": 1.052827554107517, "grad_norm": 1.5045089721679688, "learning_rate": 0.0005, "loss": 2.1210427284240723, "step": 9048 }, { "epoch": 1.052943914358855, "grad_norm": 1.2310386896133423, "learning_rate": 0.0005, "loss": 1.6881670951843262, "step": 9049 }, { "epoch": 1.0530602746101931, "grad_norm": 1.3255949020385742, "learning_rate": 0.0005, "loss": 2.33811616897583, "step": 9050 }, { "epoch": 1.0531766348615312, "grad_norm": 0.92902672290802, "learning_rate": 0.0005, "loss": 1.6960277557373047, "step": 9051 }, { "epoch": 1.0532929951128693, "grad_norm": 1.3968002796173096, "learning_rate": 0.0005, "loss": 1.8462921380996704, "step": 9052 }, { "epoch": 1.0534093553642077, "grad_norm": 1.1291935443878174, "learning_rate": 0.0005, "loss": 1.7822297811508179, "step": 9053 }, { "epoch": 1.0535257156155458, "grad_norm": 0.936245858669281, "learning_rate": 0.0005, "loss": 1.8305063247680664, "step": 9054 }, { "epoch": 1.0536420758668839, "grad_norm": 1.1208934783935547, "learning_rate": 0.0005, "loss": 1.3680222034454346, "step": 9055 }, { "epoch": 1.053758436118222, "grad_norm": 1.105034589767456, "learning_rate": 0.0005, "loss": 1.5507649183273315, "step": 9056 }, { "epoch": 1.05387479636956, "grad_norm": 1.2646186351776123, "learning_rate": 0.0005, "loss": 2.26302433013916, "step": 9057 }, { "epoch": 1.0539911566208984, "grad_norm": 1.0732109546661377, "learning_rate": 0.0005, "loss": 2.3953568935394287, "step": 9058 }, { "epoch": 1.0541075168722365, "grad_norm": 1.2404365539550781, "learning_rate": 0.0005, "loss": 2.266279458999634, "step": 9059 }, { "epoch": 1.0542238771235746, "grad_norm": 0.6930710673332214, "learning_rate": 0.0005, "loss": 1.7524240016937256, "step": 9060 }, { "epoch": 1.0543402373749127, "grad_norm": 1.1784192323684692, "learning_rate": 0.0005, "loss": 1.6954931020736694, "step": 9061 }, { "epoch": 1.0544565976262508, "grad_norm": 0.9809881448745728, "learning_rate": 0.0005, "loss": 1.8108861446380615, "step": 9062 }, { "epoch": 1.054572957877589, "grad_norm": 1.130616307258606, "learning_rate": 0.0005, "loss": 2.127993106842041, "step": 9063 }, { "epoch": 1.0546893181289272, "grad_norm": 0.8666200041770935, "learning_rate": 0.0005, "loss": 1.7976138591766357, "step": 9064 }, { "epoch": 1.0548056783802653, "grad_norm": 0.7315996885299683, "learning_rate": 0.0005, "loss": 1.7114956378936768, "step": 9065 }, { "epoch": 1.0549220386316034, "grad_norm": 1.3109023571014404, "learning_rate": 0.0005, "loss": 1.2608071565628052, "step": 9066 }, { "epoch": 1.0550383988829415, "grad_norm": 2.6671040058135986, "learning_rate": 0.0005, "loss": 2.197314739227295, "step": 9067 }, { "epoch": 1.0551547591342798, "grad_norm": 1.251289963722229, "learning_rate": 0.0005, "loss": 2.1272099018096924, "step": 9068 }, { "epoch": 1.055271119385618, "grad_norm": 1.1831669807434082, "learning_rate": 0.0005, "loss": 1.6095871925354004, "step": 9069 }, { "epoch": 1.055387479636956, "grad_norm": 1.4816908836364746, "learning_rate": 0.0005, "loss": 2.788092613220215, "step": 9070 }, { "epoch": 1.055503839888294, "grad_norm": 1.1910090446472168, "learning_rate": 0.0005, "loss": 1.8410398960113525, "step": 9071 }, { "epoch": 1.0556202001396322, "grad_norm": 1.0350499153137207, "learning_rate": 0.0005, "loss": 1.8735946416854858, "step": 9072 }, { "epoch": 1.0557365603909705, "grad_norm": 2.1400389671325684, "learning_rate": 0.0005, "loss": 1.7545884847640991, "step": 9073 }, { "epoch": 1.0558529206423086, "grad_norm": 1.2541823387145996, "learning_rate": 0.0005, "loss": 3.0809073448181152, "step": 9074 }, { "epoch": 1.0559692808936467, "grad_norm": 0.7149894833564758, "learning_rate": 0.0005, "loss": 1.106373906135559, "step": 9075 }, { "epoch": 1.0560856411449848, "grad_norm": 1.3020869493484497, "learning_rate": 0.0005, "loss": 2.725482940673828, "step": 9076 }, { "epoch": 1.056202001396323, "grad_norm": 1.4250437021255493, "learning_rate": 0.0005, "loss": 2.2092227935791016, "step": 9077 }, { "epoch": 1.0563183616476612, "grad_norm": 1.0772334337234497, "learning_rate": 0.0005, "loss": 1.8611490726470947, "step": 9078 }, { "epoch": 1.0564347218989993, "grad_norm": 0.7746367454528809, "learning_rate": 0.0005, "loss": 1.6942906379699707, "step": 9079 }, { "epoch": 1.0565510821503374, "grad_norm": 1.2991195917129517, "learning_rate": 0.0005, "loss": 2.430759906768799, "step": 9080 }, { "epoch": 1.0566674424016755, "grad_norm": 1.1414010524749756, "learning_rate": 0.0005, "loss": 1.8732937574386597, "step": 9081 }, { "epoch": 1.0567838026530136, "grad_norm": 1.0856343507766724, "learning_rate": 0.0005, "loss": 1.6667448282241821, "step": 9082 }, { "epoch": 1.056900162904352, "grad_norm": 0.8374168872833252, "learning_rate": 0.0005, "loss": 2.0536513328552246, "step": 9083 }, { "epoch": 1.05701652315569, "grad_norm": 1.0571889877319336, "learning_rate": 0.0005, "loss": 2.3490614891052246, "step": 9084 }, { "epoch": 1.0571328834070282, "grad_norm": 0.9388861656188965, "learning_rate": 0.0005, "loss": 2.6135005950927734, "step": 9085 }, { "epoch": 1.0572492436583663, "grad_norm": 1.1617012023925781, "learning_rate": 0.0005, "loss": 1.8693206310272217, "step": 9086 }, { "epoch": 1.0573656039097044, "grad_norm": 1.125982642173767, "learning_rate": 0.0005, "loss": 1.5282732248306274, "step": 9087 }, { "epoch": 1.0574819641610427, "grad_norm": 3.434926986694336, "learning_rate": 0.0005, "loss": 3.058222770690918, "step": 9088 }, { "epoch": 1.0575983244123808, "grad_norm": 0.7748910188674927, "learning_rate": 0.0005, "loss": 1.7968918085098267, "step": 9089 }, { "epoch": 1.0577146846637189, "grad_norm": 0.9576210975646973, "learning_rate": 0.0005, "loss": 2.106365203857422, "step": 9090 }, { "epoch": 1.057831044915057, "grad_norm": 1.0818862915039062, "learning_rate": 0.0005, "loss": 1.7280478477478027, "step": 9091 }, { "epoch": 1.057947405166395, "grad_norm": 1.656177282333374, "learning_rate": 0.0005, "loss": 2.1467225551605225, "step": 9092 }, { "epoch": 1.0580637654177334, "grad_norm": 0.8732380867004395, "learning_rate": 0.0005, "loss": 1.652734398841858, "step": 9093 }, { "epoch": 1.0581801256690715, "grad_norm": 1.2865636348724365, "learning_rate": 0.0005, "loss": 1.7486274242401123, "step": 9094 }, { "epoch": 1.0582964859204096, "grad_norm": 0.9267597198486328, "learning_rate": 0.0005, "loss": 1.7218410968780518, "step": 9095 }, { "epoch": 1.0584128461717477, "grad_norm": 1.1611164808273315, "learning_rate": 0.0005, "loss": 1.990232229232788, "step": 9096 }, { "epoch": 1.0585292064230858, "grad_norm": 0.7501029968261719, "learning_rate": 0.0005, "loss": 2.0399832725524902, "step": 9097 }, { "epoch": 1.058645566674424, "grad_norm": 0.6999855041503906, "learning_rate": 0.0005, "loss": 1.2222259044647217, "step": 9098 }, { "epoch": 1.0587619269257622, "grad_norm": 0.9725911617279053, "learning_rate": 0.0005, "loss": 2.234823703765869, "step": 9099 }, { "epoch": 1.0588782871771003, "grad_norm": 0.8000269532203674, "learning_rate": 0.0005, "loss": 1.3932814598083496, "step": 9100 }, { "epoch": 1.0589946474284384, "grad_norm": 1.5104296207427979, "learning_rate": 0.0005, "loss": 1.8508927822113037, "step": 9101 }, { "epoch": 1.0591110076797765, "grad_norm": 1.8924649953842163, "learning_rate": 0.0005, "loss": 2.109375, "step": 9102 }, { "epoch": 1.0592273679311148, "grad_norm": 0.8291494250297546, "learning_rate": 0.0005, "loss": 1.821577787399292, "step": 9103 }, { "epoch": 1.059343728182453, "grad_norm": 1.7388789653778076, "learning_rate": 0.0005, "loss": 1.907006025314331, "step": 9104 }, { "epoch": 1.059460088433791, "grad_norm": 0.8452038168907166, "learning_rate": 0.0005, "loss": 1.3683180809020996, "step": 9105 }, { "epoch": 1.0595764486851291, "grad_norm": 1.436282753944397, "learning_rate": 0.0005, "loss": 3.0061144828796387, "step": 9106 }, { "epoch": 1.0596928089364672, "grad_norm": 1.7267355918884277, "learning_rate": 0.0005, "loss": 2.089860439300537, "step": 9107 }, { "epoch": 1.0598091691878055, "grad_norm": 0.7900183200836182, "learning_rate": 0.0005, "loss": 2.0595428943634033, "step": 9108 }, { "epoch": 1.0599255294391436, "grad_norm": 1.4144281148910522, "learning_rate": 0.0005, "loss": 1.7351552248001099, "step": 9109 }, { "epoch": 1.0600418896904817, "grad_norm": 1.9161701202392578, "learning_rate": 0.0005, "loss": 2.2358710765838623, "step": 9110 }, { "epoch": 1.0601582499418198, "grad_norm": 1.1599445343017578, "learning_rate": 0.0005, "loss": 1.9187983274459839, "step": 9111 }, { "epoch": 1.060274610193158, "grad_norm": 0.8520712852478027, "learning_rate": 0.0005, "loss": 1.8329288959503174, "step": 9112 }, { "epoch": 1.0603909704444963, "grad_norm": 1.3191574811935425, "learning_rate": 0.0005, "loss": 1.5999960899353027, "step": 9113 }, { "epoch": 1.0605073306958344, "grad_norm": 0.8952183723449707, "learning_rate": 0.0005, "loss": 1.9626250267028809, "step": 9114 }, { "epoch": 1.0606236909471725, "grad_norm": 1.187366247177124, "learning_rate": 0.0005, "loss": 2.394340753555298, "step": 9115 }, { "epoch": 1.0607400511985106, "grad_norm": 1.0586974620819092, "learning_rate": 0.0005, "loss": 1.862027645111084, "step": 9116 }, { "epoch": 1.0608564114498487, "grad_norm": 1.102661371231079, "learning_rate": 0.0005, "loss": 2.3821637630462646, "step": 9117 }, { "epoch": 1.060972771701187, "grad_norm": 0.7565105557441711, "learning_rate": 0.0005, "loss": 1.7123847007751465, "step": 9118 }, { "epoch": 1.061089131952525, "grad_norm": 1.0681066513061523, "learning_rate": 0.0005, "loss": 2.0557217597961426, "step": 9119 }, { "epoch": 1.0612054922038632, "grad_norm": 1.1346862316131592, "learning_rate": 0.0005, "loss": 2.269864797592163, "step": 9120 }, { "epoch": 1.0613218524552013, "grad_norm": 1.7128944396972656, "learning_rate": 0.0005, "loss": 1.4949926137924194, "step": 9121 }, { "epoch": 1.0614382127065394, "grad_norm": 1.2131068706512451, "learning_rate": 0.0005, "loss": 2.0384817123413086, "step": 9122 }, { "epoch": 1.0615545729578777, "grad_norm": 1.3952491283416748, "learning_rate": 0.0005, "loss": 1.8030885457992554, "step": 9123 }, { "epoch": 1.0616709332092158, "grad_norm": 1.0572279691696167, "learning_rate": 0.0005, "loss": 1.71590256690979, "step": 9124 }, { "epoch": 1.061787293460554, "grad_norm": 1.1606073379516602, "learning_rate": 0.0005, "loss": 2.0241847038269043, "step": 9125 }, { "epoch": 1.061903653711892, "grad_norm": 0.4781564176082611, "learning_rate": 0.0005, "loss": 1.115894079208374, "step": 9126 }, { "epoch": 1.06202001396323, "grad_norm": 1.4487881660461426, "learning_rate": 0.0005, "loss": 2.7534713745117188, "step": 9127 }, { "epoch": 1.0621363742145684, "grad_norm": 1.1288878917694092, "learning_rate": 0.0005, "loss": 1.8103811740875244, "step": 9128 }, { "epoch": 1.0622527344659065, "grad_norm": 1.6903843879699707, "learning_rate": 0.0005, "loss": 2.585094928741455, "step": 9129 }, { "epoch": 1.0623690947172446, "grad_norm": 1.3072834014892578, "learning_rate": 0.0005, "loss": 2.0090744495391846, "step": 9130 }, { "epoch": 1.0624854549685827, "grad_norm": 1.5287528038024902, "learning_rate": 0.0005, "loss": 1.7503880262374878, "step": 9131 }, { "epoch": 1.0626018152199208, "grad_norm": 1.250717043876648, "learning_rate": 0.0005, "loss": 2.686561107635498, "step": 9132 }, { "epoch": 1.0627181754712591, "grad_norm": 1.172607183456421, "learning_rate": 0.0005, "loss": 2.1434531211853027, "step": 9133 }, { "epoch": 1.0628345357225972, "grad_norm": 1.1168811321258545, "learning_rate": 0.0005, "loss": 1.9093034267425537, "step": 9134 }, { "epoch": 1.0629508959739353, "grad_norm": 0.8644958138465881, "learning_rate": 0.0005, "loss": 1.6328020095825195, "step": 9135 }, { "epoch": 1.0630672562252734, "grad_norm": 0.8671873211860657, "learning_rate": 0.0005, "loss": 1.9599089622497559, "step": 9136 }, { "epoch": 1.0631836164766115, "grad_norm": 0.8496830463409424, "learning_rate": 0.0005, "loss": 1.6234625577926636, "step": 9137 }, { "epoch": 1.0632999767279498, "grad_norm": 1.9470595121383667, "learning_rate": 0.0005, "loss": 1.8334903717041016, "step": 9138 }, { "epoch": 1.063416336979288, "grad_norm": 1.1676782369613647, "learning_rate": 0.0005, "loss": 2.3793063163757324, "step": 9139 }, { "epoch": 1.063532697230626, "grad_norm": 6.540235996246338, "learning_rate": 0.0005, "loss": 2.1622536182403564, "step": 9140 }, { "epoch": 1.0636490574819641, "grad_norm": 1.2376312017440796, "learning_rate": 0.0005, "loss": 1.845414638519287, "step": 9141 }, { "epoch": 1.0637654177333022, "grad_norm": 1.0674324035644531, "learning_rate": 0.0005, "loss": 2.374614715576172, "step": 9142 }, { "epoch": 1.0638817779846406, "grad_norm": 1.125860571861267, "learning_rate": 0.0005, "loss": 1.649761438369751, "step": 9143 }, { "epoch": 1.0639981382359787, "grad_norm": 1.4496676921844482, "learning_rate": 0.0005, "loss": 1.9092044830322266, "step": 9144 }, { "epoch": 1.0641144984873168, "grad_norm": 1.1431471109390259, "learning_rate": 0.0005, "loss": 2.080338478088379, "step": 9145 }, { "epoch": 1.0642308587386549, "grad_norm": 0.9863559007644653, "learning_rate": 0.0005, "loss": 2.030121326446533, "step": 9146 }, { "epoch": 1.064347218989993, "grad_norm": 1.5694477558135986, "learning_rate": 0.0005, "loss": 2.4721145629882812, "step": 9147 }, { "epoch": 1.064463579241331, "grad_norm": 0.5667486786842346, "learning_rate": 0.0005, "loss": 1.2753827571868896, "step": 9148 }, { "epoch": 1.0645799394926694, "grad_norm": 1.1309504508972168, "learning_rate": 0.0005, "loss": 1.6261982917785645, "step": 9149 }, { "epoch": 1.0646962997440075, "grad_norm": 1.8283928632736206, "learning_rate": 0.0005, "loss": 1.8377271890640259, "step": 9150 }, { "epoch": 1.0648126599953456, "grad_norm": 1.0097832679748535, "learning_rate": 0.0005, "loss": 2.39788818359375, "step": 9151 }, { "epoch": 1.0649290202466837, "grad_norm": 1.3652154207229614, "learning_rate": 0.0005, "loss": 2.48830509185791, "step": 9152 }, { "epoch": 1.0650453804980218, "grad_norm": 1.0252068042755127, "learning_rate": 0.0005, "loss": 1.6562530994415283, "step": 9153 }, { "epoch": 1.06516174074936, "grad_norm": 0.776473879814148, "learning_rate": 0.0005, "loss": 1.5622594356536865, "step": 9154 }, { "epoch": 1.0652781010006982, "grad_norm": 0.9163984656333923, "learning_rate": 0.0005, "loss": 1.5565993785858154, "step": 9155 }, { "epoch": 1.0653944612520363, "grad_norm": 1.3345227241516113, "learning_rate": 0.0005, "loss": 2.8328659534454346, "step": 9156 }, { "epoch": 1.0655108215033744, "grad_norm": 0.9745703935623169, "learning_rate": 0.0005, "loss": 2.018430471420288, "step": 9157 }, { "epoch": 1.0656271817547125, "grad_norm": 1.0792800188064575, "learning_rate": 0.0005, "loss": 1.2850418090820312, "step": 9158 }, { "epoch": 1.0657435420060508, "grad_norm": 1.5765761137008667, "learning_rate": 0.0005, "loss": 1.923816204071045, "step": 9159 }, { "epoch": 1.065859902257389, "grad_norm": 1.139602780342102, "learning_rate": 0.0005, "loss": 1.521726131439209, "step": 9160 }, { "epoch": 1.065976262508727, "grad_norm": 1.0639901161193848, "learning_rate": 0.0005, "loss": 1.697416067123413, "step": 9161 }, { "epoch": 1.0660926227600651, "grad_norm": 1.8330347537994385, "learning_rate": 0.0005, "loss": 2.3906707763671875, "step": 9162 }, { "epoch": 1.0662089830114032, "grad_norm": 1.336807131767273, "learning_rate": 0.0005, "loss": 2.061674118041992, "step": 9163 }, { "epoch": 1.0663253432627415, "grad_norm": 1.7420322895050049, "learning_rate": 0.0005, "loss": 1.8060238361358643, "step": 9164 }, { "epoch": 1.0664417035140796, "grad_norm": 1.4799803495407104, "learning_rate": 0.0005, "loss": 2.0200724601745605, "step": 9165 }, { "epoch": 1.0665580637654177, "grad_norm": 1.0569814443588257, "learning_rate": 0.0005, "loss": 2.2210047245025635, "step": 9166 }, { "epoch": 1.0666744240167558, "grad_norm": 0.7238220572471619, "learning_rate": 0.0005, "loss": 2.07967209815979, "step": 9167 }, { "epoch": 1.066790784268094, "grad_norm": 2.1799378395080566, "learning_rate": 0.0005, "loss": 1.8619928359985352, "step": 9168 }, { "epoch": 1.0669071445194322, "grad_norm": 1.383553147315979, "learning_rate": 0.0005, "loss": 1.7538233995437622, "step": 9169 }, { "epoch": 1.0670235047707703, "grad_norm": 1.2522876262664795, "learning_rate": 0.0005, "loss": 2.633547782897949, "step": 9170 }, { "epoch": 1.0671398650221084, "grad_norm": 1.2017115354537964, "learning_rate": 0.0005, "loss": 1.850675106048584, "step": 9171 }, { "epoch": 1.0672562252734465, "grad_norm": 1.017093539237976, "learning_rate": 0.0005, "loss": 2.439980983734131, "step": 9172 }, { "epoch": 1.0673725855247846, "grad_norm": 1.2583627700805664, "learning_rate": 0.0005, "loss": 2.3144173622131348, "step": 9173 }, { "epoch": 1.067488945776123, "grad_norm": 0.8232220411300659, "learning_rate": 0.0005, "loss": 2.092757225036621, "step": 9174 }, { "epoch": 1.067605306027461, "grad_norm": 0.9103623032569885, "learning_rate": 0.0005, "loss": 1.3645548820495605, "step": 9175 }, { "epoch": 1.0677216662787992, "grad_norm": 1.247363567352295, "learning_rate": 0.0005, "loss": 2.588719367980957, "step": 9176 }, { "epoch": 1.0678380265301373, "grad_norm": 0.7388882040977478, "learning_rate": 0.0005, "loss": 2.1497440338134766, "step": 9177 }, { "epoch": 1.0679543867814754, "grad_norm": 1.5061570405960083, "learning_rate": 0.0005, "loss": 2.431875228881836, "step": 9178 }, { "epoch": 1.0680707470328137, "grad_norm": 1.1434065103530884, "learning_rate": 0.0005, "loss": 2.413703203201294, "step": 9179 }, { "epoch": 1.0681871072841518, "grad_norm": 1.5579807758331299, "learning_rate": 0.0005, "loss": 1.6855500936508179, "step": 9180 }, { "epoch": 1.0683034675354899, "grad_norm": 0.8806154131889343, "learning_rate": 0.0005, "loss": 1.3644578456878662, "step": 9181 }, { "epoch": 1.068419827786828, "grad_norm": 1.4288787841796875, "learning_rate": 0.0005, "loss": 2.315413236618042, "step": 9182 }, { "epoch": 1.068536188038166, "grad_norm": 1.0924538373947144, "learning_rate": 0.0005, "loss": 1.9547905921936035, "step": 9183 }, { "epoch": 1.0686525482895044, "grad_norm": 0.8573681712150574, "learning_rate": 0.0005, "loss": 1.4140002727508545, "step": 9184 }, { "epoch": 1.0687689085408425, "grad_norm": 1.374976396560669, "learning_rate": 0.0005, "loss": 2.4309158325195312, "step": 9185 }, { "epoch": 1.0688852687921806, "grad_norm": 1.2125650644302368, "learning_rate": 0.0005, "loss": 2.1772611141204834, "step": 9186 }, { "epoch": 1.0690016290435187, "grad_norm": 1.3125523328781128, "learning_rate": 0.0005, "loss": 2.3021864891052246, "step": 9187 }, { "epoch": 1.0691179892948568, "grad_norm": 0.9291238784790039, "learning_rate": 0.0005, "loss": 1.400355339050293, "step": 9188 }, { "epoch": 1.0692343495461951, "grad_norm": 1.321377158164978, "learning_rate": 0.0005, "loss": 1.9004664421081543, "step": 9189 }, { "epoch": 1.0693507097975332, "grad_norm": 1.1296924352645874, "learning_rate": 0.0005, "loss": 1.746154546737671, "step": 9190 }, { "epoch": 1.0694670700488713, "grad_norm": 0.9810980558395386, "learning_rate": 0.0005, "loss": 1.4977567195892334, "step": 9191 }, { "epoch": 1.0695834303002094, "grad_norm": 1.3437939882278442, "learning_rate": 0.0005, "loss": 2.0895028114318848, "step": 9192 }, { "epoch": 1.0696997905515475, "grad_norm": 1.3820011615753174, "learning_rate": 0.0005, "loss": 1.8456099033355713, "step": 9193 }, { "epoch": 1.0698161508028858, "grad_norm": 1.2075246572494507, "learning_rate": 0.0005, "loss": 1.756895899772644, "step": 9194 }, { "epoch": 1.069932511054224, "grad_norm": 1.3525735139846802, "learning_rate": 0.0005, "loss": 1.8599910736083984, "step": 9195 }, { "epoch": 1.070048871305562, "grad_norm": 0.8951424360275269, "learning_rate": 0.0005, "loss": 2.6853678226470947, "step": 9196 }, { "epoch": 1.0701652315569001, "grad_norm": 0.8038378357887268, "learning_rate": 0.0005, "loss": 1.318502426147461, "step": 9197 }, { "epoch": 1.0702815918082382, "grad_norm": 1.0446289777755737, "learning_rate": 0.0005, "loss": 2.0040173530578613, "step": 9198 }, { "epoch": 1.0703979520595766, "grad_norm": 0.9816330075263977, "learning_rate": 0.0005, "loss": 2.5232129096984863, "step": 9199 }, { "epoch": 1.0705143123109147, "grad_norm": 1.0734602212905884, "learning_rate": 0.0005, "loss": 2.1594557762145996, "step": 9200 }, { "epoch": 1.0706306725622527, "grad_norm": 1.8100136518478394, "learning_rate": 0.0005, "loss": 2.7988061904907227, "step": 9201 }, { "epoch": 1.0707470328135908, "grad_norm": 0.9303094744682312, "learning_rate": 0.0005, "loss": 1.8176034688949585, "step": 9202 }, { "epoch": 1.070863393064929, "grad_norm": 1.133840799331665, "learning_rate": 0.0005, "loss": 1.5787644386291504, "step": 9203 }, { "epoch": 1.070979753316267, "grad_norm": 1.0719703435897827, "learning_rate": 0.0005, "loss": 1.8142826557159424, "step": 9204 }, { "epoch": 1.0710961135676054, "grad_norm": 0.7842122316360474, "learning_rate": 0.0005, "loss": 1.9754782915115356, "step": 9205 }, { "epoch": 1.0712124738189435, "grad_norm": 1.0248299837112427, "learning_rate": 0.0005, "loss": 2.138888359069824, "step": 9206 }, { "epoch": 1.0713288340702816, "grad_norm": 1.2025021314620972, "learning_rate": 0.0005, "loss": 2.0058469772338867, "step": 9207 }, { "epoch": 1.0714451943216197, "grad_norm": 1.370762825012207, "learning_rate": 0.0005, "loss": 2.387622833251953, "step": 9208 }, { "epoch": 1.0715615545729578, "grad_norm": 1.2037523984909058, "learning_rate": 0.0005, "loss": 2.423976421356201, "step": 9209 }, { "epoch": 1.071677914824296, "grad_norm": 1.0053452253341675, "learning_rate": 0.0005, "loss": 1.6369580030441284, "step": 9210 }, { "epoch": 1.0717942750756342, "grad_norm": 1.0221017599105835, "learning_rate": 0.0005, "loss": 2.5532779693603516, "step": 9211 }, { "epoch": 1.0719106353269723, "grad_norm": 0.7609215974807739, "learning_rate": 0.0005, "loss": 1.9155795574188232, "step": 9212 }, { "epoch": 1.0720269955783104, "grad_norm": 1.1075001955032349, "learning_rate": 0.0005, "loss": 1.5035345554351807, "step": 9213 }, { "epoch": 1.0721433558296485, "grad_norm": 1.102100133895874, "learning_rate": 0.0005, "loss": 1.831598162651062, "step": 9214 }, { "epoch": 1.0722597160809868, "grad_norm": 1.1743453741073608, "learning_rate": 0.0005, "loss": 1.798053503036499, "step": 9215 }, { "epoch": 1.072376076332325, "grad_norm": 3.8629345893859863, "learning_rate": 0.0005, "loss": 2.4596080780029297, "step": 9216 }, { "epoch": 1.072492436583663, "grad_norm": 1.1436728239059448, "learning_rate": 0.0005, "loss": 2.1139261722564697, "step": 9217 }, { "epoch": 1.072608796835001, "grad_norm": 1.6834180355072021, "learning_rate": 0.0005, "loss": 1.8086745738983154, "step": 9218 }, { "epoch": 1.0727251570863392, "grad_norm": 1.4968360662460327, "learning_rate": 0.0005, "loss": 1.9917622804641724, "step": 9219 }, { "epoch": 1.0728415173376775, "grad_norm": 0.8853338360786438, "learning_rate": 0.0005, "loss": 1.4795485734939575, "step": 9220 }, { "epoch": 1.0729578775890156, "grad_norm": 1.0355005264282227, "learning_rate": 0.0005, "loss": 1.9152169227600098, "step": 9221 }, { "epoch": 1.0730742378403537, "grad_norm": 0.9048589468002319, "learning_rate": 0.0005, "loss": 1.7875458002090454, "step": 9222 }, { "epoch": 1.0731905980916918, "grad_norm": 1.0662107467651367, "learning_rate": 0.0005, "loss": 1.573157548904419, "step": 9223 }, { "epoch": 1.07330695834303, "grad_norm": 1.2879308462142944, "learning_rate": 0.0005, "loss": 1.411513090133667, "step": 9224 }, { "epoch": 1.0734233185943682, "grad_norm": 1.5133788585662842, "learning_rate": 0.0005, "loss": 1.3684465885162354, "step": 9225 }, { "epoch": 1.0735396788457063, "grad_norm": 0.8202557563781738, "learning_rate": 0.0005, "loss": 2.0530543327331543, "step": 9226 }, { "epoch": 1.0736560390970444, "grad_norm": 1.4437196254730225, "learning_rate": 0.0005, "loss": 2.3703606128692627, "step": 9227 }, { "epoch": 1.0737723993483825, "grad_norm": 0.9604954719543457, "learning_rate": 0.0005, "loss": 2.074700117111206, "step": 9228 }, { "epoch": 1.0738887595997206, "grad_norm": 1.1108500957489014, "learning_rate": 0.0005, "loss": 2.2408204078674316, "step": 9229 }, { "epoch": 1.074005119851059, "grad_norm": 3.0743308067321777, "learning_rate": 0.0005, "loss": 1.8168566226959229, "step": 9230 }, { "epoch": 1.074121480102397, "grad_norm": 0.7066875696182251, "learning_rate": 0.0005, "loss": 1.2001444101333618, "step": 9231 }, { "epoch": 1.0742378403537352, "grad_norm": 0.9670792818069458, "learning_rate": 0.0005, "loss": 1.7049062252044678, "step": 9232 }, { "epoch": 1.0743542006050733, "grad_norm": 0.9725689888000488, "learning_rate": 0.0005, "loss": 1.9277945756912231, "step": 9233 }, { "epoch": 1.0744705608564113, "grad_norm": 1.5355950593948364, "learning_rate": 0.0005, "loss": 2.3039345741271973, "step": 9234 }, { "epoch": 1.0745869211077497, "grad_norm": 1.1854697465896606, "learning_rate": 0.0005, "loss": 1.442258358001709, "step": 9235 }, { "epoch": 1.0747032813590878, "grad_norm": 0.8887911438941956, "learning_rate": 0.0005, "loss": 2.768550395965576, "step": 9236 }, { "epoch": 1.0748196416104259, "grad_norm": 0.995191216468811, "learning_rate": 0.0005, "loss": 1.702726125717163, "step": 9237 }, { "epoch": 1.074936001861764, "grad_norm": 0.8320786952972412, "learning_rate": 0.0005, "loss": 1.373389720916748, "step": 9238 }, { "epoch": 1.075052362113102, "grad_norm": 0.642936646938324, "learning_rate": 0.0005, "loss": 1.296529769897461, "step": 9239 }, { "epoch": 1.0751687223644404, "grad_norm": 0.9440909028053284, "learning_rate": 0.0005, "loss": 1.4652386903762817, "step": 9240 }, { "epoch": 1.0752850826157785, "grad_norm": 2.0476317405700684, "learning_rate": 0.0005, "loss": 1.8404167890548706, "step": 9241 }, { "epoch": 1.0754014428671166, "grad_norm": 1.3387726545333862, "learning_rate": 0.0005, "loss": 2.127552032470703, "step": 9242 }, { "epoch": 1.0755178031184547, "grad_norm": 1.0192418098449707, "learning_rate": 0.0005, "loss": 1.8447884321212769, "step": 9243 }, { "epoch": 1.0756341633697928, "grad_norm": 1.5565599203109741, "learning_rate": 0.0005, "loss": 2.2797369956970215, "step": 9244 }, { "epoch": 1.075750523621131, "grad_norm": 1.0344011783599854, "learning_rate": 0.0005, "loss": 1.6167960166931152, "step": 9245 }, { "epoch": 1.0758668838724692, "grad_norm": 0.9522687196731567, "learning_rate": 0.0005, "loss": 1.722809076309204, "step": 9246 }, { "epoch": 1.0759832441238073, "grad_norm": 1.0354410409927368, "learning_rate": 0.0005, "loss": 2.296131134033203, "step": 9247 }, { "epoch": 1.0760996043751454, "grad_norm": 1.1766215562820435, "learning_rate": 0.0005, "loss": 1.8381240367889404, "step": 9248 }, { "epoch": 1.0762159646264835, "grad_norm": 1.1684502363204956, "learning_rate": 0.0005, "loss": 2.0635666847229004, "step": 9249 }, { "epoch": 1.0763323248778218, "grad_norm": 1.2323590517044067, "learning_rate": 0.0005, "loss": 2.415050506591797, "step": 9250 }, { "epoch": 1.07644868512916, "grad_norm": 1.107252597808838, "learning_rate": 0.0005, "loss": 1.8311872482299805, "step": 9251 }, { "epoch": 1.076565045380498, "grad_norm": 1.2139065265655518, "learning_rate": 0.0005, "loss": 2.39973521232605, "step": 9252 }, { "epoch": 1.0766814056318361, "grad_norm": 0.3646998405456543, "learning_rate": 0.0005, "loss": 0.9325999021530151, "step": 9253 }, { "epoch": 1.0767977658831742, "grad_norm": 1.4963057041168213, "learning_rate": 0.0005, "loss": 1.5244224071502686, "step": 9254 }, { "epoch": 1.0769141261345125, "grad_norm": 0.6515117883682251, "learning_rate": 0.0005, "loss": 1.124185562133789, "step": 9255 }, { "epoch": 1.0770304863858506, "grad_norm": 2.244330883026123, "learning_rate": 0.0005, "loss": 2.451580047607422, "step": 9256 }, { "epoch": 1.0771468466371887, "grad_norm": 0.9733558893203735, "learning_rate": 0.0005, "loss": 2.190636157989502, "step": 9257 }, { "epoch": 1.0772632068885268, "grad_norm": 0.9199249148368835, "learning_rate": 0.0005, "loss": 1.492815375328064, "step": 9258 }, { "epoch": 1.077379567139865, "grad_norm": 1.2024096250534058, "learning_rate": 0.0005, "loss": 1.8016934394836426, "step": 9259 }, { "epoch": 1.0774959273912033, "grad_norm": 1.3357137441635132, "learning_rate": 0.0005, "loss": 2.2681431770324707, "step": 9260 }, { "epoch": 1.0776122876425414, "grad_norm": 1.518349528312683, "learning_rate": 0.0005, "loss": 2.0711112022399902, "step": 9261 }, { "epoch": 1.0777286478938795, "grad_norm": 1.512905240058899, "learning_rate": 0.0005, "loss": 1.4159631729125977, "step": 9262 }, { "epoch": 1.0778450081452176, "grad_norm": 1.0190116167068481, "learning_rate": 0.0005, "loss": 2.250704288482666, "step": 9263 }, { "epoch": 1.0779613683965557, "grad_norm": 1.0337159633636475, "learning_rate": 0.0005, "loss": 1.850014567375183, "step": 9264 }, { "epoch": 1.078077728647894, "grad_norm": 0.9582556486129761, "learning_rate": 0.0005, "loss": 1.685032606124878, "step": 9265 }, { "epoch": 1.078194088899232, "grad_norm": 1.2150167226791382, "learning_rate": 0.0005, "loss": 2.0085129737854004, "step": 9266 }, { "epoch": 1.0783104491505702, "grad_norm": 1.5358940362930298, "learning_rate": 0.0005, "loss": 2.8094959259033203, "step": 9267 }, { "epoch": 1.0784268094019083, "grad_norm": 1.0563644170761108, "learning_rate": 0.0005, "loss": 1.8204303979873657, "step": 9268 }, { "epoch": 1.0785431696532464, "grad_norm": 1.0003384351730347, "learning_rate": 0.0005, "loss": 2.4012858867645264, "step": 9269 }, { "epoch": 1.0786595299045847, "grad_norm": 1.460273265838623, "learning_rate": 0.0005, "loss": 1.3152294158935547, "step": 9270 }, { "epoch": 1.0787758901559228, "grad_norm": 0.4513404369354248, "learning_rate": 0.0005, "loss": 0.9530991315841675, "step": 9271 }, { "epoch": 1.0788922504072609, "grad_norm": 3.230276346206665, "learning_rate": 0.0005, "loss": 2.194382905960083, "step": 9272 }, { "epoch": 1.079008610658599, "grad_norm": 0.9478812217712402, "learning_rate": 0.0005, "loss": 2.0477070808410645, "step": 9273 }, { "epoch": 1.079124970909937, "grad_norm": 1.1894160509109497, "learning_rate": 0.0005, "loss": 2.214611530303955, "step": 9274 }, { "epoch": 1.0792413311612754, "grad_norm": 0.9201321005821228, "learning_rate": 0.0005, "loss": 1.4495525360107422, "step": 9275 }, { "epoch": 1.0793576914126135, "grad_norm": 1.2866454124450684, "learning_rate": 0.0005, "loss": 2.260497570037842, "step": 9276 }, { "epoch": 1.0794740516639516, "grad_norm": 1.0283267498016357, "learning_rate": 0.0005, "loss": 1.433741807937622, "step": 9277 }, { "epoch": 1.0795904119152897, "grad_norm": 1.496648907661438, "learning_rate": 0.0005, "loss": 2.724242925643921, "step": 9278 }, { "epoch": 1.0797067721666278, "grad_norm": 0.829460084438324, "learning_rate": 0.0005, "loss": 2.3784196376800537, "step": 9279 }, { "epoch": 1.0798231324179661, "grad_norm": 1.6216752529144287, "learning_rate": 0.0005, "loss": 2.277350425720215, "step": 9280 }, { "epoch": 1.0799394926693042, "grad_norm": 0.7403020858764648, "learning_rate": 0.0005, "loss": 1.9070405960083008, "step": 9281 }, { "epoch": 1.0800558529206423, "grad_norm": 1.187227725982666, "learning_rate": 0.0005, "loss": 2.1893601417541504, "step": 9282 }, { "epoch": 1.0801722131719804, "grad_norm": 0.8103721737861633, "learning_rate": 0.0005, "loss": 1.5875790119171143, "step": 9283 }, { "epoch": 1.0802885734233185, "grad_norm": 0.9669319987297058, "learning_rate": 0.0005, "loss": 2.282731056213379, "step": 9284 }, { "epoch": 1.0804049336746568, "grad_norm": 1.1936062574386597, "learning_rate": 0.0005, "loss": 1.8732433319091797, "step": 9285 }, { "epoch": 1.080521293925995, "grad_norm": 0.8701842427253723, "learning_rate": 0.0005, "loss": 2.087212085723877, "step": 9286 }, { "epoch": 1.080637654177333, "grad_norm": 1.7580995559692383, "learning_rate": 0.0005, "loss": 1.9682033061981201, "step": 9287 }, { "epoch": 1.0807540144286711, "grad_norm": 0.9427087903022766, "learning_rate": 0.0005, "loss": 1.8766827583312988, "step": 9288 }, { "epoch": 1.0808703746800092, "grad_norm": 0.9887258410453796, "learning_rate": 0.0005, "loss": 2.0618679523468018, "step": 9289 }, { "epoch": 1.0809867349313476, "grad_norm": 0.9686588048934937, "learning_rate": 0.0005, "loss": 2.4237897396087646, "step": 9290 }, { "epoch": 1.0811030951826857, "grad_norm": 1.0295144319534302, "learning_rate": 0.0005, "loss": 2.220268487930298, "step": 9291 }, { "epoch": 1.0812194554340238, "grad_norm": 1.2282627820968628, "learning_rate": 0.0005, "loss": 1.652869701385498, "step": 9292 }, { "epoch": 1.0813358156853619, "grad_norm": 0.9347949624061584, "learning_rate": 0.0005, "loss": 1.5319700241088867, "step": 9293 }, { "epoch": 1.0814521759367, "grad_norm": 1.0156689882278442, "learning_rate": 0.0005, "loss": 1.52812659740448, "step": 9294 }, { "epoch": 1.0815685361880383, "grad_norm": 1.0567197799682617, "learning_rate": 0.0005, "loss": 2.3664937019348145, "step": 9295 }, { "epoch": 1.0816848964393764, "grad_norm": 0.9540097713470459, "learning_rate": 0.0005, "loss": 2.440005302429199, "step": 9296 }, { "epoch": 1.0818012566907145, "grad_norm": 1.2374132871627808, "learning_rate": 0.0005, "loss": 1.9314777851104736, "step": 9297 }, { "epoch": 1.0819176169420526, "grad_norm": 0.7888651490211487, "learning_rate": 0.0005, "loss": 2.0152015686035156, "step": 9298 }, { "epoch": 1.0820339771933907, "grad_norm": 1.0766505002975464, "learning_rate": 0.0005, "loss": 1.7154288291931152, "step": 9299 }, { "epoch": 1.082150337444729, "grad_norm": 6.203220367431641, "learning_rate": 0.0005, "loss": 2.0943198204040527, "step": 9300 }, { "epoch": 1.082266697696067, "grad_norm": 1.0187796354293823, "learning_rate": 0.0005, "loss": 1.9566113948822021, "step": 9301 }, { "epoch": 1.0823830579474052, "grad_norm": 1.1387040615081787, "learning_rate": 0.0005, "loss": 2.0589308738708496, "step": 9302 }, { "epoch": 1.0824994181987433, "grad_norm": 0.5397745370864868, "learning_rate": 0.0005, "loss": 1.1624839305877686, "step": 9303 }, { "epoch": 1.0826157784500814, "grad_norm": 1.3207614421844482, "learning_rate": 0.0005, "loss": 2.4838366508483887, "step": 9304 }, { "epoch": 1.0827321387014197, "grad_norm": 0.9188575744628906, "learning_rate": 0.0005, "loss": 1.7001261711120605, "step": 9305 }, { "epoch": 1.0828484989527578, "grad_norm": 0.7705788016319275, "learning_rate": 0.0005, "loss": 1.3597482442855835, "step": 9306 }, { "epoch": 1.082964859204096, "grad_norm": 1.3939889669418335, "learning_rate": 0.0005, "loss": 1.7698428630828857, "step": 9307 }, { "epoch": 1.083081219455434, "grad_norm": 1.0502731800079346, "learning_rate": 0.0005, "loss": 2.3406565189361572, "step": 9308 }, { "epoch": 1.083197579706772, "grad_norm": 1.2091853618621826, "learning_rate": 0.0005, "loss": 2.4753470420837402, "step": 9309 }, { "epoch": 1.0833139399581102, "grad_norm": 1.261705756187439, "learning_rate": 0.0005, "loss": 1.9443233013153076, "step": 9310 }, { "epoch": 1.0834303002094485, "grad_norm": 2.206871509552002, "learning_rate": 0.0005, "loss": 2.36982798576355, "step": 9311 }, { "epoch": 1.0835466604607866, "grad_norm": 1.1144464015960693, "learning_rate": 0.0005, "loss": 1.8467506170272827, "step": 9312 }, { "epoch": 1.0836630207121247, "grad_norm": 1.5111415386199951, "learning_rate": 0.0005, "loss": 2.3172717094421387, "step": 9313 }, { "epoch": 1.0837793809634628, "grad_norm": 1.1793431043624878, "learning_rate": 0.0005, "loss": 2.3502440452575684, "step": 9314 }, { "epoch": 1.083895741214801, "grad_norm": 0.6859694123268127, "learning_rate": 0.0005, "loss": 1.5973645448684692, "step": 9315 }, { "epoch": 1.0840121014661392, "grad_norm": 1.2291016578674316, "learning_rate": 0.0005, "loss": 1.6154487133026123, "step": 9316 }, { "epoch": 1.0841284617174773, "grad_norm": 0.48438331484794617, "learning_rate": 0.0005, "loss": 1.105214238166809, "step": 9317 }, { "epoch": 1.0842448219688154, "grad_norm": 1.1969763040542603, "learning_rate": 0.0005, "loss": 1.9043577909469604, "step": 9318 }, { "epoch": 1.0843611822201535, "grad_norm": 0.6927200555801392, "learning_rate": 0.0005, "loss": 2.010026693344116, "step": 9319 }, { "epoch": 1.0844775424714916, "grad_norm": 1.0600626468658447, "learning_rate": 0.0005, "loss": 1.7713371515274048, "step": 9320 }, { "epoch": 1.08459390272283, "grad_norm": 1.3569904565811157, "learning_rate": 0.0005, "loss": 1.6065846681594849, "step": 9321 }, { "epoch": 1.084710262974168, "grad_norm": 1.4104193449020386, "learning_rate": 0.0005, "loss": 2.02117919921875, "step": 9322 }, { "epoch": 1.0848266232255062, "grad_norm": 1.2255719900131226, "learning_rate": 0.0005, "loss": 1.7737114429473877, "step": 9323 }, { "epoch": 1.0849429834768443, "grad_norm": 0.8629629611968994, "learning_rate": 0.0005, "loss": 2.249484062194824, "step": 9324 }, { "epoch": 1.0850593437281824, "grad_norm": 1.5200870037078857, "learning_rate": 0.0005, "loss": 2.4893646240234375, "step": 9325 }, { "epoch": 1.0851757039795207, "grad_norm": 1.5048140287399292, "learning_rate": 0.0005, "loss": 2.7397894859313965, "step": 9326 }, { "epoch": 1.0852920642308588, "grad_norm": 2.0297908782958984, "learning_rate": 0.0005, "loss": 2.611995220184326, "step": 9327 }, { "epoch": 1.0854084244821969, "grad_norm": 0.9555137753486633, "learning_rate": 0.0005, "loss": 1.7068288326263428, "step": 9328 }, { "epoch": 1.085524784733535, "grad_norm": 0.7876639366149902, "learning_rate": 0.0005, "loss": 2.1797266006469727, "step": 9329 }, { "epoch": 1.085641144984873, "grad_norm": 1.3501951694488525, "learning_rate": 0.0005, "loss": 2.7183897495269775, "step": 9330 }, { "epoch": 1.0857575052362114, "grad_norm": 1.4179258346557617, "learning_rate": 0.0005, "loss": 2.0693862438201904, "step": 9331 }, { "epoch": 1.0858738654875495, "grad_norm": 0.9069746136665344, "learning_rate": 0.0005, "loss": 1.9539239406585693, "step": 9332 }, { "epoch": 1.0859902257388876, "grad_norm": 1.285771131515503, "learning_rate": 0.0005, "loss": 2.3838486671447754, "step": 9333 }, { "epoch": 1.0861065859902257, "grad_norm": 1.084672451019287, "learning_rate": 0.0005, "loss": 1.6271203756332397, "step": 9334 }, { "epoch": 1.0862229462415638, "grad_norm": 0.9070120453834534, "learning_rate": 0.0005, "loss": 1.9947402477264404, "step": 9335 }, { "epoch": 1.0863393064929021, "grad_norm": 1.1212080717086792, "learning_rate": 0.0005, "loss": 1.9389392137527466, "step": 9336 }, { "epoch": 1.0864556667442402, "grad_norm": 1.2969454526901245, "learning_rate": 0.0005, "loss": 2.480348825454712, "step": 9337 }, { "epoch": 1.0865720269955783, "grad_norm": 1.4853448867797852, "learning_rate": 0.0005, "loss": 2.368061065673828, "step": 9338 }, { "epoch": 1.0866883872469164, "grad_norm": 1.112667202949524, "learning_rate": 0.0005, "loss": 1.8436524868011475, "step": 9339 }, { "epoch": 1.0868047474982545, "grad_norm": 0.8970672488212585, "learning_rate": 0.0005, "loss": 2.532866954803467, "step": 9340 }, { "epoch": 1.0869211077495928, "grad_norm": 0.9606540203094482, "learning_rate": 0.0005, "loss": 1.731102466583252, "step": 9341 }, { "epoch": 1.087037468000931, "grad_norm": 0.9427567720413208, "learning_rate": 0.0005, "loss": 1.4928358793258667, "step": 9342 }, { "epoch": 1.087153828252269, "grad_norm": 1.8893067836761475, "learning_rate": 0.0005, "loss": 1.5382310152053833, "step": 9343 }, { "epoch": 1.0872701885036071, "grad_norm": 1.1161328554153442, "learning_rate": 0.0005, "loss": 2.387981414794922, "step": 9344 }, { "epoch": 1.0873865487549452, "grad_norm": 1.1826798915863037, "learning_rate": 0.0005, "loss": 2.2822070121765137, "step": 9345 }, { "epoch": 1.0875029090062835, "grad_norm": 1.3304353952407837, "learning_rate": 0.0005, "loss": 2.2807929515838623, "step": 9346 }, { "epoch": 1.0876192692576216, "grad_norm": 1.088201880455017, "learning_rate": 0.0005, "loss": 1.9185762405395508, "step": 9347 }, { "epoch": 1.0877356295089597, "grad_norm": 1.518739104270935, "learning_rate": 0.0005, "loss": 1.92606782913208, "step": 9348 }, { "epoch": 1.0878519897602978, "grad_norm": 0.936935305595398, "learning_rate": 0.0005, "loss": 2.1730940341949463, "step": 9349 }, { "epoch": 1.087968350011636, "grad_norm": 0.7458550930023193, "learning_rate": 0.0005, "loss": 1.6320730447769165, "step": 9350 }, { "epoch": 1.0880847102629743, "grad_norm": 0.6810176968574524, "learning_rate": 0.0005, "loss": 1.1052348613739014, "step": 9351 }, { "epoch": 1.0882010705143124, "grad_norm": 1.151843547821045, "learning_rate": 0.0005, "loss": 1.6718511581420898, "step": 9352 }, { "epoch": 1.0883174307656505, "grad_norm": 1.296975016593933, "learning_rate": 0.0005, "loss": 1.8611408472061157, "step": 9353 }, { "epoch": 1.0884337910169886, "grad_norm": 0.8767242431640625, "learning_rate": 0.0005, "loss": 2.200237989425659, "step": 9354 }, { "epoch": 1.0885501512683267, "grad_norm": 0.7668787837028503, "learning_rate": 0.0005, "loss": 1.3979413509368896, "step": 9355 }, { "epoch": 1.088666511519665, "grad_norm": 1.2696685791015625, "learning_rate": 0.0005, "loss": 2.7932779788970947, "step": 9356 }, { "epoch": 1.088782871771003, "grad_norm": 1.0972237586975098, "learning_rate": 0.0005, "loss": 2.2518558502197266, "step": 9357 }, { "epoch": 1.0888992320223412, "grad_norm": 1.782698631286621, "learning_rate": 0.0005, "loss": 1.3485583066940308, "step": 9358 }, { "epoch": 1.0890155922736793, "grad_norm": 1.2834917306900024, "learning_rate": 0.0005, "loss": 1.5802841186523438, "step": 9359 }, { "epoch": 1.0891319525250174, "grad_norm": 0.9680922627449036, "learning_rate": 0.0005, "loss": 1.572605848312378, "step": 9360 }, { "epoch": 1.0892483127763557, "grad_norm": 1.2298489809036255, "learning_rate": 0.0005, "loss": 2.3158817291259766, "step": 9361 }, { "epoch": 1.0893646730276938, "grad_norm": 0.9239411950111389, "learning_rate": 0.0005, "loss": 1.7014731168746948, "step": 9362 }, { "epoch": 1.089481033279032, "grad_norm": 0.8683110475540161, "learning_rate": 0.0005, "loss": 1.8552241325378418, "step": 9363 }, { "epoch": 1.08959739353037, "grad_norm": 1.0210686922073364, "learning_rate": 0.0005, "loss": 1.3574442863464355, "step": 9364 }, { "epoch": 1.089713753781708, "grad_norm": 1.105169415473938, "learning_rate": 0.0005, "loss": 2.4722518920898438, "step": 9365 }, { "epoch": 1.0898301140330464, "grad_norm": 1.0000773668289185, "learning_rate": 0.0005, "loss": 1.991880178451538, "step": 9366 }, { "epoch": 1.0899464742843845, "grad_norm": 1.3729195594787598, "learning_rate": 0.0005, "loss": 1.9007246494293213, "step": 9367 }, { "epoch": 1.0900628345357226, "grad_norm": 0.8687191009521484, "learning_rate": 0.0005, "loss": 1.334022045135498, "step": 9368 }, { "epoch": 1.0901791947870607, "grad_norm": 0.9038997888565063, "learning_rate": 0.0005, "loss": 1.8330588340759277, "step": 9369 }, { "epoch": 1.0902955550383988, "grad_norm": 1.098620891571045, "learning_rate": 0.0005, "loss": 1.9423091411590576, "step": 9370 }, { "epoch": 1.090411915289737, "grad_norm": 1.3355776071548462, "learning_rate": 0.0005, "loss": 1.6317203044891357, "step": 9371 }, { "epoch": 1.0905282755410752, "grad_norm": 1.1786441802978516, "learning_rate": 0.0005, "loss": 2.000685930252075, "step": 9372 }, { "epoch": 1.0906446357924133, "grad_norm": 1.4836921691894531, "learning_rate": 0.0005, "loss": 2.185248374938965, "step": 9373 }, { "epoch": 1.0907609960437514, "grad_norm": 1.8455965518951416, "learning_rate": 0.0005, "loss": 1.7026093006134033, "step": 9374 }, { "epoch": 1.0908773562950895, "grad_norm": 0.823588490486145, "learning_rate": 0.0005, "loss": 1.9326114654541016, "step": 9375 }, { "epoch": 1.0909937165464276, "grad_norm": 1.2549502849578857, "learning_rate": 0.0005, "loss": 2.0740041732788086, "step": 9376 }, { "epoch": 1.091110076797766, "grad_norm": 0.5545835494995117, "learning_rate": 0.0005, "loss": 1.0811954736709595, "step": 9377 }, { "epoch": 1.091226437049104, "grad_norm": 2.2479496002197266, "learning_rate": 0.0005, "loss": 2.107618570327759, "step": 9378 }, { "epoch": 1.0913427973004421, "grad_norm": 0.9384542107582092, "learning_rate": 0.0005, "loss": 2.740295648574829, "step": 9379 }, { "epoch": 1.0914591575517802, "grad_norm": 0.8864178657531738, "learning_rate": 0.0005, "loss": 1.6864542961120605, "step": 9380 }, { "epoch": 1.0915755178031183, "grad_norm": 1.2108021974563599, "learning_rate": 0.0005, "loss": 2.5143003463745117, "step": 9381 }, { "epoch": 1.0916918780544567, "grad_norm": 1.2413431406021118, "learning_rate": 0.0005, "loss": 2.111727476119995, "step": 9382 }, { "epoch": 1.0918082383057948, "grad_norm": 0.868870735168457, "learning_rate": 0.0005, "loss": 1.981273889541626, "step": 9383 }, { "epoch": 1.0919245985571329, "grad_norm": 1.2557897567749023, "learning_rate": 0.0005, "loss": 2.4912025928497314, "step": 9384 }, { "epoch": 1.092040958808471, "grad_norm": 1.370296835899353, "learning_rate": 0.0005, "loss": 2.5701725482940674, "step": 9385 }, { "epoch": 1.092157319059809, "grad_norm": 0.9308842420578003, "learning_rate": 0.0005, "loss": 1.9316706657409668, "step": 9386 }, { "epoch": 1.0922736793111474, "grad_norm": 1.0253095626831055, "learning_rate": 0.0005, "loss": 2.3599634170532227, "step": 9387 }, { "epoch": 1.0923900395624855, "grad_norm": 1.4036922454833984, "learning_rate": 0.0005, "loss": 2.088611364364624, "step": 9388 }, { "epoch": 1.0925063998138236, "grad_norm": 1.031123161315918, "learning_rate": 0.0005, "loss": 1.6507855653762817, "step": 9389 }, { "epoch": 1.0926227600651617, "grad_norm": 1.1255812644958496, "learning_rate": 0.0005, "loss": 1.8039237260818481, "step": 9390 }, { "epoch": 1.0927391203164998, "grad_norm": 0.908493161201477, "learning_rate": 0.0005, "loss": 1.8895280361175537, "step": 9391 }, { "epoch": 1.092855480567838, "grad_norm": 0.9245963096618652, "learning_rate": 0.0005, "loss": 1.972374439239502, "step": 9392 }, { "epoch": 1.0929718408191762, "grad_norm": 1.0119956731796265, "learning_rate": 0.0005, "loss": 1.9216711521148682, "step": 9393 }, { "epoch": 1.0930882010705143, "grad_norm": 0.9332067966461182, "learning_rate": 0.0005, "loss": 1.9714947938919067, "step": 9394 }, { "epoch": 1.0932045613218524, "grad_norm": 0.8340306282043457, "learning_rate": 0.0005, "loss": 1.5351831912994385, "step": 9395 }, { "epoch": 1.0933209215731905, "grad_norm": 1.438627004623413, "learning_rate": 0.0005, "loss": 1.683664321899414, "step": 9396 }, { "epoch": 1.0934372818245288, "grad_norm": 1.2165900468826294, "learning_rate": 0.0005, "loss": 1.4127535820007324, "step": 9397 }, { "epoch": 1.093553642075867, "grad_norm": 0.8877065181732178, "learning_rate": 0.0005, "loss": 1.5826421976089478, "step": 9398 }, { "epoch": 1.093670002327205, "grad_norm": 1.88226318359375, "learning_rate": 0.0005, "loss": 2.423757553100586, "step": 9399 }, { "epoch": 1.0937863625785431, "grad_norm": 1.2628506422042847, "learning_rate": 0.0005, "loss": 1.7773350477218628, "step": 9400 }, { "epoch": 1.0939027228298812, "grad_norm": 2.132678747177124, "learning_rate": 0.0005, "loss": 1.7589914798736572, "step": 9401 }, { "epoch": 1.0940190830812195, "grad_norm": 1.569007396697998, "learning_rate": 0.0005, "loss": 2.058671236038208, "step": 9402 }, { "epoch": 1.0941354433325576, "grad_norm": 1.9606150388717651, "learning_rate": 0.0005, "loss": 1.8691414594650269, "step": 9403 }, { "epoch": 1.0942518035838957, "grad_norm": 0.7780658006668091, "learning_rate": 0.0005, "loss": 1.2039943933486938, "step": 9404 }, { "epoch": 1.0943681638352338, "grad_norm": 1.1196446418762207, "learning_rate": 0.0005, "loss": 1.7550454139709473, "step": 9405 }, { "epoch": 1.094484524086572, "grad_norm": 1.33487868309021, "learning_rate": 0.0005, "loss": 2.3585286140441895, "step": 9406 }, { "epoch": 1.0946008843379103, "grad_norm": 1.1816680431365967, "learning_rate": 0.0005, "loss": 1.9389715194702148, "step": 9407 }, { "epoch": 1.0947172445892484, "grad_norm": 1.8424162864685059, "learning_rate": 0.0005, "loss": 1.815563678741455, "step": 9408 }, { "epoch": 1.0948336048405864, "grad_norm": 1.2672398090362549, "learning_rate": 0.0005, "loss": 2.348116397857666, "step": 9409 }, { "epoch": 1.0949499650919245, "grad_norm": 0.746070146560669, "learning_rate": 0.0005, "loss": 1.3991458415985107, "step": 9410 }, { "epoch": 1.0950663253432626, "grad_norm": 1.9469536542892456, "learning_rate": 0.0005, "loss": 2.667856216430664, "step": 9411 }, { "epoch": 1.095182685594601, "grad_norm": 3.505506992340088, "learning_rate": 0.0005, "loss": 2.5024189949035645, "step": 9412 }, { "epoch": 1.095299045845939, "grad_norm": 0.8399046659469604, "learning_rate": 0.0005, "loss": 1.8479161262512207, "step": 9413 }, { "epoch": 1.0954154060972772, "grad_norm": 0.8662971258163452, "learning_rate": 0.0005, "loss": 1.5732700824737549, "step": 9414 }, { "epoch": 1.0955317663486153, "grad_norm": 1.243898868560791, "learning_rate": 0.0005, "loss": 1.9495055675506592, "step": 9415 }, { "epoch": 1.0956481265999534, "grad_norm": 1.0101585388183594, "learning_rate": 0.0005, "loss": 1.301190972328186, "step": 9416 }, { "epoch": 1.0957644868512917, "grad_norm": 1.1871213912963867, "learning_rate": 0.0005, "loss": 1.5871200561523438, "step": 9417 }, { "epoch": 1.0958808471026298, "grad_norm": 2.040257453918457, "learning_rate": 0.0005, "loss": 1.7471420764923096, "step": 9418 }, { "epoch": 1.0959972073539679, "grad_norm": 1.244551181793213, "learning_rate": 0.0005, "loss": 1.4093482494354248, "step": 9419 }, { "epoch": 1.096113567605306, "grad_norm": 0.9999156594276428, "learning_rate": 0.0005, "loss": 1.9193956851959229, "step": 9420 }, { "epoch": 1.096229927856644, "grad_norm": 1.0750696659088135, "learning_rate": 0.0005, "loss": 2.0017504692077637, "step": 9421 }, { "epoch": 1.0963462881079824, "grad_norm": 1.3237472772598267, "learning_rate": 0.0005, "loss": 1.4930145740509033, "step": 9422 }, { "epoch": 1.0964626483593205, "grad_norm": 1.2417526245117188, "learning_rate": 0.0005, "loss": 2.64033842086792, "step": 9423 }, { "epoch": 1.0965790086106586, "grad_norm": 1.2806153297424316, "learning_rate": 0.0005, "loss": 2.3259835243225098, "step": 9424 }, { "epoch": 1.0966953688619967, "grad_norm": 1.0825899839401245, "learning_rate": 0.0005, "loss": 1.6301953792572021, "step": 9425 }, { "epoch": 1.0968117291133348, "grad_norm": 1.1612842082977295, "learning_rate": 0.0005, "loss": 2.043140411376953, "step": 9426 }, { "epoch": 1.0969280893646731, "grad_norm": 0.8152163028717041, "learning_rate": 0.0005, "loss": 1.7622994184494019, "step": 9427 }, { "epoch": 1.0970444496160112, "grad_norm": 0.6243313550949097, "learning_rate": 0.0005, "loss": 1.5455979108810425, "step": 9428 }, { "epoch": 1.0971608098673493, "grad_norm": 0.765069305896759, "learning_rate": 0.0005, "loss": 1.2600945234298706, "step": 9429 }, { "epoch": 1.0972771701186874, "grad_norm": 0.8874971866607666, "learning_rate": 0.0005, "loss": 1.6493895053863525, "step": 9430 }, { "epoch": 1.0973935303700255, "grad_norm": 1.3341513872146606, "learning_rate": 0.0005, "loss": 2.031862258911133, "step": 9431 }, { "epoch": 1.0975098906213638, "grad_norm": 0.5701166391372681, "learning_rate": 0.0005, "loss": 1.563816785812378, "step": 9432 }, { "epoch": 1.097626250872702, "grad_norm": 1.0729211568832397, "learning_rate": 0.0005, "loss": 1.5375187397003174, "step": 9433 }, { "epoch": 1.09774261112404, "grad_norm": 1.4217921495437622, "learning_rate": 0.0005, "loss": 1.9575304985046387, "step": 9434 }, { "epoch": 1.0978589713753781, "grad_norm": 1.645717740058899, "learning_rate": 0.0005, "loss": 2.0397088527679443, "step": 9435 }, { "epoch": 1.0979753316267162, "grad_norm": 1.5555658340454102, "learning_rate": 0.0005, "loss": 1.9715827703475952, "step": 9436 }, { "epoch": 1.0980916918780546, "grad_norm": 1.4771124124526978, "learning_rate": 0.0005, "loss": 2.1965956687927246, "step": 9437 }, { "epoch": 1.0982080521293927, "grad_norm": 0.8130070567131042, "learning_rate": 0.0005, "loss": 1.850013256072998, "step": 9438 }, { "epoch": 1.0983244123807308, "grad_norm": 1.0844203233718872, "learning_rate": 0.0005, "loss": 2.3103508949279785, "step": 9439 }, { "epoch": 1.0984407726320689, "grad_norm": 1.1094506978988647, "learning_rate": 0.0005, "loss": 1.9754459857940674, "step": 9440 }, { "epoch": 1.098557132883407, "grad_norm": 0.7950576543807983, "learning_rate": 0.0005, "loss": 1.541324257850647, "step": 9441 }, { "epoch": 1.0986734931347453, "grad_norm": 1.121370553970337, "learning_rate": 0.0005, "loss": 1.7886509895324707, "step": 9442 }, { "epoch": 1.0987898533860834, "grad_norm": 0.6775095462799072, "learning_rate": 0.0005, "loss": 1.4003872871398926, "step": 9443 }, { "epoch": 1.0989062136374215, "grad_norm": 1.03663969039917, "learning_rate": 0.0005, "loss": 1.888411045074463, "step": 9444 }, { "epoch": 1.0990225738887596, "grad_norm": 0.826994776725769, "learning_rate": 0.0005, "loss": 2.2926149368286133, "step": 9445 }, { "epoch": 1.0991389341400977, "grad_norm": 0.9144778251647949, "learning_rate": 0.0005, "loss": 1.729108214378357, "step": 9446 }, { "epoch": 1.099255294391436, "grad_norm": 1.4459705352783203, "learning_rate": 0.0005, "loss": 2.8369500637054443, "step": 9447 }, { "epoch": 1.099371654642774, "grad_norm": 0.8886168599128723, "learning_rate": 0.0005, "loss": 1.2786420583724976, "step": 9448 }, { "epoch": 1.0994880148941122, "grad_norm": 1.098629117012024, "learning_rate": 0.0005, "loss": 2.0116336345672607, "step": 9449 }, { "epoch": 1.0996043751454503, "grad_norm": 1.1630300283432007, "learning_rate": 0.0005, "loss": 2.112926959991455, "step": 9450 }, { "epoch": 1.0997207353967884, "grad_norm": 1.2924787998199463, "learning_rate": 0.0005, "loss": 1.8408595323562622, "step": 9451 }, { "epoch": 1.0998370956481267, "grad_norm": 0.9005879163742065, "learning_rate": 0.0005, "loss": 1.6174554824829102, "step": 9452 }, { "epoch": 1.0999534558994648, "grad_norm": 2.5682449340820312, "learning_rate": 0.0005, "loss": 2.294429302215576, "step": 9453 }, { "epoch": 1.100069816150803, "grad_norm": 1.1684800386428833, "learning_rate": 0.0005, "loss": 1.6284973621368408, "step": 9454 }, { "epoch": 1.100186176402141, "grad_norm": 1.0127198696136475, "learning_rate": 0.0005, "loss": 1.5470118522644043, "step": 9455 }, { "epoch": 1.100302536653479, "grad_norm": 1.6003000736236572, "learning_rate": 0.0005, "loss": 2.8039560317993164, "step": 9456 }, { "epoch": 1.1004188969048174, "grad_norm": 0.9358761310577393, "learning_rate": 0.0005, "loss": 1.6171159744262695, "step": 9457 }, { "epoch": 1.1005352571561555, "grad_norm": 1.4434771537780762, "learning_rate": 0.0005, "loss": 2.3940141201019287, "step": 9458 }, { "epoch": 1.1006516174074936, "grad_norm": 1.1822395324707031, "learning_rate": 0.0005, "loss": 1.8620206117630005, "step": 9459 }, { "epoch": 1.1007679776588317, "grad_norm": 0.9132100939750671, "learning_rate": 0.0005, "loss": 1.7750493288040161, "step": 9460 }, { "epoch": 1.1008843379101698, "grad_norm": 1.0618401765823364, "learning_rate": 0.0005, "loss": 2.361114740371704, "step": 9461 }, { "epoch": 1.1010006981615081, "grad_norm": 2.6066927909851074, "learning_rate": 0.0005, "loss": 2.052000045776367, "step": 9462 }, { "epoch": 1.1011170584128462, "grad_norm": 1.0051138401031494, "learning_rate": 0.0005, "loss": 1.6455307006835938, "step": 9463 }, { "epoch": 1.1012334186641843, "grad_norm": 1.265527367591858, "learning_rate": 0.0005, "loss": 1.5157010555267334, "step": 9464 }, { "epoch": 1.1013497789155224, "grad_norm": 0.8755699396133423, "learning_rate": 0.0005, "loss": 1.6895349025726318, "step": 9465 }, { "epoch": 1.1014661391668605, "grad_norm": 1.7388015985488892, "learning_rate": 0.0005, "loss": 2.6754674911499023, "step": 9466 }, { "epoch": 1.1015824994181989, "grad_norm": 0.9075922966003418, "learning_rate": 0.0005, "loss": 1.6900436878204346, "step": 9467 }, { "epoch": 1.101698859669537, "grad_norm": 1.1186670064926147, "learning_rate": 0.0005, "loss": 1.777724027633667, "step": 9468 }, { "epoch": 1.101815219920875, "grad_norm": 3.5636239051818848, "learning_rate": 0.0005, "loss": 1.5317301750183105, "step": 9469 }, { "epoch": 1.1019315801722132, "grad_norm": 1.238358974456787, "learning_rate": 0.0005, "loss": 2.3722145557403564, "step": 9470 }, { "epoch": 1.1020479404235513, "grad_norm": 1.2278367280960083, "learning_rate": 0.0005, "loss": 1.8723831176757812, "step": 9471 }, { "epoch": 1.1021643006748896, "grad_norm": 1.4161818027496338, "learning_rate": 0.0005, "loss": 1.4968911409378052, "step": 9472 }, { "epoch": 1.1022806609262277, "grad_norm": 2.528656482696533, "learning_rate": 0.0005, "loss": 2.036264419555664, "step": 9473 }, { "epoch": 1.1023970211775658, "grad_norm": 0.9541603922843933, "learning_rate": 0.0005, "loss": 1.752571702003479, "step": 9474 }, { "epoch": 1.1025133814289039, "grad_norm": 1.921437382698059, "learning_rate": 0.0005, "loss": 2.172595977783203, "step": 9475 }, { "epoch": 1.102629741680242, "grad_norm": 0.9761614799499512, "learning_rate": 0.0005, "loss": 1.486332654953003, "step": 9476 }, { "epoch": 1.10274610193158, "grad_norm": 1.3876923322677612, "learning_rate": 0.0005, "loss": 1.9172115325927734, "step": 9477 }, { "epoch": 1.1028624621829184, "grad_norm": 0.9561255574226379, "learning_rate": 0.0005, "loss": 2.253260612487793, "step": 9478 }, { "epoch": 1.1029788224342565, "grad_norm": 1.4647119045257568, "learning_rate": 0.0005, "loss": 2.2080159187316895, "step": 9479 }, { "epoch": 1.1030951826855946, "grad_norm": 0.9234036207199097, "learning_rate": 0.0005, "loss": 2.1615540981292725, "step": 9480 }, { "epoch": 1.1032115429369327, "grad_norm": 1.1522040367126465, "learning_rate": 0.0005, "loss": 1.8399412631988525, "step": 9481 }, { "epoch": 1.1033279031882708, "grad_norm": 1.044779658317566, "learning_rate": 0.0005, "loss": 2.185694694519043, "step": 9482 }, { "epoch": 1.103444263439609, "grad_norm": 1.3881345987319946, "learning_rate": 0.0005, "loss": 1.7542877197265625, "step": 9483 }, { "epoch": 1.1035606236909472, "grad_norm": 1.3348536491394043, "learning_rate": 0.0005, "loss": 2.097078323364258, "step": 9484 }, { "epoch": 1.1036769839422853, "grad_norm": 0.7923707962036133, "learning_rate": 0.0005, "loss": 1.8436263799667358, "step": 9485 }, { "epoch": 1.1037933441936234, "grad_norm": 1.244972586631775, "learning_rate": 0.0005, "loss": 2.408827304840088, "step": 9486 }, { "epoch": 1.1039097044449615, "grad_norm": 1.1304290294647217, "learning_rate": 0.0005, "loss": 2.627041816711426, "step": 9487 }, { "epoch": 1.1040260646962998, "grad_norm": 0.6774085164070129, "learning_rate": 0.0005, "loss": 1.804731011390686, "step": 9488 }, { "epoch": 1.104142424947638, "grad_norm": 1.2579948902130127, "learning_rate": 0.0005, "loss": 2.0198984146118164, "step": 9489 }, { "epoch": 1.104258785198976, "grad_norm": 0.9405115842819214, "learning_rate": 0.0005, "loss": 2.2535550594329834, "step": 9490 }, { "epoch": 1.1043751454503141, "grad_norm": 1.0733940601348877, "learning_rate": 0.0005, "loss": 1.4778125286102295, "step": 9491 }, { "epoch": 1.1044915057016522, "grad_norm": 1.1383854150772095, "learning_rate": 0.0005, "loss": 1.820893406867981, "step": 9492 }, { "epoch": 1.1046078659529905, "grad_norm": 1.2832413911819458, "learning_rate": 0.0005, "loss": 2.1714611053466797, "step": 9493 }, { "epoch": 1.1047242262043286, "grad_norm": 1.193089246749878, "learning_rate": 0.0005, "loss": 1.8382568359375, "step": 9494 }, { "epoch": 1.1048405864556667, "grad_norm": 0.8988745212554932, "learning_rate": 0.0005, "loss": 1.8104808330535889, "step": 9495 }, { "epoch": 1.1049569467070048, "grad_norm": 1.183150291442871, "learning_rate": 0.0005, "loss": 2.4746110439300537, "step": 9496 }, { "epoch": 1.105073306958343, "grad_norm": 1.5234720706939697, "learning_rate": 0.0005, "loss": 1.7254340648651123, "step": 9497 }, { "epoch": 1.1051896672096813, "grad_norm": 1.215813398361206, "learning_rate": 0.0005, "loss": 2.594190835952759, "step": 9498 }, { "epoch": 1.1053060274610194, "grad_norm": 1.2114869356155396, "learning_rate": 0.0005, "loss": 2.691619873046875, "step": 9499 }, { "epoch": 1.1054223877123575, "grad_norm": 1.28435480594635, "learning_rate": 0.0005, "loss": 1.8231858015060425, "step": 9500 }, { "epoch": 1.1055387479636956, "grad_norm": 1.1617740392684937, "learning_rate": 0.0005, "loss": 2.478809118270874, "step": 9501 }, { "epoch": 1.1056551082150337, "grad_norm": 1.5518012046813965, "learning_rate": 0.0005, "loss": 2.202930450439453, "step": 9502 }, { "epoch": 1.105771468466372, "grad_norm": 0.919980525970459, "learning_rate": 0.0005, "loss": 1.4427472352981567, "step": 9503 }, { "epoch": 1.10588782871771, "grad_norm": 1.1496520042419434, "learning_rate": 0.0005, "loss": 2.1421799659729004, "step": 9504 }, { "epoch": 1.1060041889690482, "grad_norm": 1.52555251121521, "learning_rate": 0.0005, "loss": 2.049650192260742, "step": 9505 }, { "epoch": 1.1061205492203863, "grad_norm": 1.3073278665542603, "learning_rate": 0.0005, "loss": 1.9373961687088013, "step": 9506 }, { "epoch": 1.1062369094717244, "grad_norm": 1.6462360620498657, "learning_rate": 0.0005, "loss": 2.1035289764404297, "step": 9507 }, { "epoch": 1.1063532697230627, "grad_norm": 1.351086139678955, "learning_rate": 0.0005, "loss": 1.6553586721420288, "step": 9508 }, { "epoch": 1.1064696299744008, "grad_norm": 1.5010815858840942, "learning_rate": 0.0005, "loss": 2.332411766052246, "step": 9509 }, { "epoch": 1.106585990225739, "grad_norm": 0.627585768699646, "learning_rate": 0.0005, "loss": 1.433072566986084, "step": 9510 }, { "epoch": 1.106702350477077, "grad_norm": 1.5669097900390625, "learning_rate": 0.0005, "loss": 2.7238106727600098, "step": 9511 }, { "epoch": 1.106818710728415, "grad_norm": 0.9262981414794922, "learning_rate": 0.0005, "loss": 2.0351340770721436, "step": 9512 }, { "epoch": 1.1069350709797534, "grad_norm": 1.470159649848938, "learning_rate": 0.0005, "loss": 2.1954808235168457, "step": 9513 }, { "epoch": 1.1070514312310915, "grad_norm": 1.1437549591064453, "learning_rate": 0.0005, "loss": 2.464630603790283, "step": 9514 }, { "epoch": 1.1071677914824296, "grad_norm": 1.3204699754714966, "learning_rate": 0.0005, "loss": 2.1516952514648438, "step": 9515 }, { "epoch": 1.1072841517337677, "grad_norm": 0.7113813757896423, "learning_rate": 0.0005, "loss": 1.5322133302688599, "step": 9516 }, { "epoch": 1.1074005119851058, "grad_norm": 1.4112787246704102, "learning_rate": 0.0005, "loss": 1.4041037559509277, "step": 9517 }, { "epoch": 1.1075168722364441, "grad_norm": 1.4721423387527466, "learning_rate": 0.0005, "loss": 2.701725721359253, "step": 9518 }, { "epoch": 1.1076332324877822, "grad_norm": 0.7294268012046814, "learning_rate": 0.0005, "loss": 1.1480519771575928, "step": 9519 }, { "epoch": 1.1077495927391203, "grad_norm": 1.0716166496276855, "learning_rate": 0.0005, "loss": 1.603980541229248, "step": 9520 }, { "epoch": 1.1078659529904584, "grad_norm": 1.3474712371826172, "learning_rate": 0.0005, "loss": 2.5211002826690674, "step": 9521 }, { "epoch": 1.1079823132417965, "grad_norm": 1.4058655500411987, "learning_rate": 0.0005, "loss": 2.6181931495666504, "step": 9522 }, { "epoch": 1.1080986734931348, "grad_norm": 1.280264973640442, "learning_rate": 0.0005, "loss": 2.2786710262298584, "step": 9523 }, { "epoch": 1.108215033744473, "grad_norm": 1.326615333557129, "learning_rate": 0.0005, "loss": 2.098674774169922, "step": 9524 }, { "epoch": 1.108331393995811, "grad_norm": 1.28035569190979, "learning_rate": 0.0005, "loss": 1.986964225769043, "step": 9525 }, { "epoch": 1.1084477542471491, "grad_norm": 1.6297880411148071, "learning_rate": 0.0005, "loss": 2.2240519523620605, "step": 9526 }, { "epoch": 1.1085641144984872, "grad_norm": 1.3774583339691162, "learning_rate": 0.0005, "loss": 2.430384635925293, "step": 9527 }, { "epoch": 1.1086804747498256, "grad_norm": 1.5267080068588257, "learning_rate": 0.0005, "loss": 2.246694803237915, "step": 9528 }, { "epoch": 1.1087968350011637, "grad_norm": 1.1564477682113647, "learning_rate": 0.0005, "loss": 2.2080140113830566, "step": 9529 }, { "epoch": 1.1089131952525018, "grad_norm": 1.2321895360946655, "learning_rate": 0.0005, "loss": 2.1477372646331787, "step": 9530 }, { "epoch": 1.1090295555038399, "grad_norm": 1.039984941482544, "learning_rate": 0.0005, "loss": 1.990073561668396, "step": 9531 }, { "epoch": 1.109145915755178, "grad_norm": 1.1812337636947632, "learning_rate": 0.0005, "loss": 1.887352705001831, "step": 9532 }, { "epoch": 1.109262276006516, "grad_norm": 1.2566043138504028, "learning_rate": 0.0005, "loss": 2.1692426204681396, "step": 9533 }, { "epoch": 1.1093786362578544, "grad_norm": 1.7376971244812012, "learning_rate": 0.0005, "loss": 2.3834481239318848, "step": 9534 }, { "epoch": 1.1094949965091925, "grad_norm": 1.445085883140564, "learning_rate": 0.0005, "loss": 1.6455707550048828, "step": 9535 }, { "epoch": 1.1096113567605306, "grad_norm": 0.6167716979980469, "learning_rate": 0.0005, "loss": 1.231130838394165, "step": 9536 }, { "epoch": 1.1097277170118687, "grad_norm": 1.0870625972747803, "learning_rate": 0.0005, "loss": 1.6211870908737183, "step": 9537 }, { "epoch": 1.1098440772632068, "grad_norm": 1.1712121963500977, "learning_rate": 0.0005, "loss": 1.910984992980957, "step": 9538 }, { "epoch": 1.109960437514545, "grad_norm": 1.3203585147857666, "learning_rate": 0.0005, "loss": 2.350475788116455, "step": 9539 }, { "epoch": 1.1100767977658832, "grad_norm": 3.962529182434082, "learning_rate": 0.0005, "loss": 2.0705366134643555, "step": 9540 }, { "epoch": 1.1101931580172213, "grad_norm": 0.8366161584854126, "learning_rate": 0.0005, "loss": 1.4788726568222046, "step": 9541 }, { "epoch": 1.1103095182685594, "grad_norm": 1.5323362350463867, "learning_rate": 0.0005, "loss": 2.0425045490264893, "step": 9542 }, { "epoch": 1.1104258785198975, "grad_norm": 0.8835008144378662, "learning_rate": 0.0005, "loss": 1.9024581909179688, "step": 9543 }, { "epoch": 1.1105422387712358, "grad_norm": 0.7895865440368652, "learning_rate": 0.0005, "loss": 1.528193473815918, "step": 9544 }, { "epoch": 1.110658599022574, "grad_norm": 1.450738787651062, "learning_rate": 0.0005, "loss": 1.5240497589111328, "step": 9545 }, { "epoch": 1.110774959273912, "grad_norm": 0.949970543384552, "learning_rate": 0.0005, "loss": 1.2044086456298828, "step": 9546 }, { "epoch": 1.11089131952525, "grad_norm": 0.9663376808166504, "learning_rate": 0.0005, "loss": 1.7603055238723755, "step": 9547 }, { "epoch": 1.1110076797765882, "grad_norm": 0.8468471765518188, "learning_rate": 0.0005, "loss": 1.8574966192245483, "step": 9548 }, { "epoch": 1.1111240400279265, "grad_norm": 1.5330396890640259, "learning_rate": 0.0005, "loss": 2.1343178749084473, "step": 9549 }, { "epoch": 1.1112404002792646, "grad_norm": 1.1190928220748901, "learning_rate": 0.0005, "loss": 1.974509835243225, "step": 9550 }, { "epoch": 1.1113567605306027, "grad_norm": 0.9580017924308777, "learning_rate": 0.0005, "loss": 1.5718467235565186, "step": 9551 }, { "epoch": 1.1114731207819408, "grad_norm": 1.4313818216323853, "learning_rate": 0.0005, "loss": 1.561964511871338, "step": 9552 }, { "epoch": 1.111589481033279, "grad_norm": 1.2156388759613037, "learning_rate": 0.0005, "loss": 1.765836238861084, "step": 9553 }, { "epoch": 1.1117058412846172, "grad_norm": 1.515864610671997, "learning_rate": 0.0005, "loss": 2.243814468383789, "step": 9554 }, { "epoch": 1.1118222015359553, "grad_norm": 1.474794626235962, "learning_rate": 0.0005, "loss": 1.4805877208709717, "step": 9555 }, { "epoch": 1.1119385617872934, "grad_norm": 0.7794340252876282, "learning_rate": 0.0005, "loss": 2.4301249980926514, "step": 9556 }, { "epoch": 1.1120549220386315, "grad_norm": 0.7781127691268921, "learning_rate": 0.0005, "loss": 1.3726798295974731, "step": 9557 }, { "epoch": 1.1121712822899696, "grad_norm": 0.5850581526756287, "learning_rate": 0.0005, "loss": 1.6849473714828491, "step": 9558 }, { "epoch": 1.112287642541308, "grad_norm": 1.3557305335998535, "learning_rate": 0.0005, "loss": 2.529710292816162, "step": 9559 }, { "epoch": 1.112404002792646, "grad_norm": 0.9342251420021057, "learning_rate": 0.0005, "loss": 2.3590564727783203, "step": 9560 }, { "epoch": 1.1125203630439842, "grad_norm": 1.6744073629379272, "learning_rate": 0.0005, "loss": 1.720915675163269, "step": 9561 }, { "epoch": 1.1126367232953223, "grad_norm": 1.4017163515090942, "learning_rate": 0.0005, "loss": 2.5079169273376465, "step": 9562 }, { "epoch": 1.1127530835466604, "grad_norm": 0.9953608512878418, "learning_rate": 0.0005, "loss": 1.970841646194458, "step": 9563 }, { "epoch": 1.1128694437979987, "grad_norm": 0.7512888312339783, "learning_rate": 0.0005, "loss": 1.513314962387085, "step": 9564 }, { "epoch": 1.1129858040493368, "grad_norm": 1.227353811264038, "learning_rate": 0.0005, "loss": 2.023216962814331, "step": 9565 }, { "epoch": 1.1131021643006749, "grad_norm": 1.2534574270248413, "learning_rate": 0.0005, "loss": 2.468566656112671, "step": 9566 }, { "epoch": 1.113218524552013, "grad_norm": 1.4886187314987183, "learning_rate": 0.0005, "loss": 2.3047029972076416, "step": 9567 }, { "epoch": 1.113334884803351, "grad_norm": 1.572172999382019, "learning_rate": 0.0005, "loss": 2.2880725860595703, "step": 9568 }, { "epoch": 1.1134512450546894, "grad_norm": 1.2898170948028564, "learning_rate": 0.0005, "loss": 1.8598413467407227, "step": 9569 }, { "epoch": 1.1135676053060275, "grad_norm": 1.2249761819839478, "learning_rate": 0.0005, "loss": 2.1721692085266113, "step": 9570 }, { "epoch": 1.1136839655573656, "grad_norm": 1.39962637424469, "learning_rate": 0.0005, "loss": 2.2780556678771973, "step": 9571 }, { "epoch": 1.1138003258087037, "grad_norm": 1.1677072048187256, "learning_rate": 0.0005, "loss": 1.4452667236328125, "step": 9572 }, { "epoch": 1.1139166860600418, "grad_norm": 1.3929376602172852, "learning_rate": 0.0005, "loss": 2.0950355529785156, "step": 9573 }, { "epoch": 1.1140330463113801, "grad_norm": 0.7982338666915894, "learning_rate": 0.0005, "loss": 1.1770422458648682, "step": 9574 }, { "epoch": 1.1141494065627182, "grad_norm": 1.233730673789978, "learning_rate": 0.0005, "loss": 1.7152149677276611, "step": 9575 }, { "epoch": 1.1142657668140563, "grad_norm": 1.268812894821167, "learning_rate": 0.0005, "loss": 1.9522427320480347, "step": 9576 }, { "epoch": 1.1143821270653944, "grad_norm": 1.326626181602478, "learning_rate": 0.0005, "loss": 2.0864925384521484, "step": 9577 }, { "epoch": 1.1144984873167325, "grad_norm": 1.396943211555481, "learning_rate": 0.0005, "loss": 2.355268716812134, "step": 9578 }, { "epoch": 1.1146148475680708, "grad_norm": 0.9646427631378174, "learning_rate": 0.0005, "loss": 1.9906284809112549, "step": 9579 }, { "epoch": 1.114731207819409, "grad_norm": 1.2591172456741333, "learning_rate": 0.0005, "loss": 2.3527235984802246, "step": 9580 }, { "epoch": 1.114847568070747, "grad_norm": 0.7840797305107117, "learning_rate": 0.0005, "loss": 2.6747350692749023, "step": 9581 }, { "epoch": 1.1149639283220851, "grad_norm": 1.0357887744903564, "learning_rate": 0.0005, "loss": 2.253408908843994, "step": 9582 }, { "epoch": 1.1150802885734232, "grad_norm": 0.9002862572669983, "learning_rate": 0.0005, "loss": 1.959031581878662, "step": 9583 }, { "epoch": 1.1151966488247615, "grad_norm": 1.0661823749542236, "learning_rate": 0.0005, "loss": 1.4967541694641113, "step": 9584 }, { "epoch": 1.1153130090760996, "grad_norm": 1.341519832611084, "learning_rate": 0.0005, "loss": 1.820549726486206, "step": 9585 }, { "epoch": 1.1154293693274377, "grad_norm": 1.3668984174728394, "learning_rate": 0.0005, "loss": 2.090017795562744, "step": 9586 }, { "epoch": 1.1155457295787758, "grad_norm": 1.4521437883377075, "learning_rate": 0.0005, "loss": 2.720831871032715, "step": 9587 }, { "epoch": 1.115662089830114, "grad_norm": 1.3480403423309326, "learning_rate": 0.0005, "loss": 2.5192453861236572, "step": 9588 }, { "epoch": 1.1157784500814523, "grad_norm": 1.1289082765579224, "learning_rate": 0.0005, "loss": 2.5353927612304688, "step": 9589 }, { "epoch": 1.1158948103327904, "grad_norm": 1.0818278789520264, "learning_rate": 0.0005, "loss": 2.296983242034912, "step": 9590 }, { "epoch": 1.1160111705841285, "grad_norm": 0.8000326752662659, "learning_rate": 0.0005, "loss": 1.6311132907867432, "step": 9591 }, { "epoch": 1.1161275308354666, "grad_norm": 0.7451955676078796, "learning_rate": 0.0005, "loss": 1.721139907836914, "step": 9592 }, { "epoch": 1.1162438910868047, "grad_norm": 1.1367369890213013, "learning_rate": 0.0005, "loss": 2.5681490898132324, "step": 9593 }, { "epoch": 1.116360251338143, "grad_norm": 1.275295376777649, "learning_rate": 0.0005, "loss": 2.7190561294555664, "step": 9594 }, { "epoch": 1.116476611589481, "grad_norm": 0.970905601978302, "learning_rate": 0.0005, "loss": 1.9580461978912354, "step": 9595 }, { "epoch": 1.1165929718408192, "grad_norm": 0.918265163898468, "learning_rate": 0.0005, "loss": 1.9834977388381958, "step": 9596 }, { "epoch": 1.1167093320921573, "grad_norm": 0.937432587146759, "learning_rate": 0.0005, "loss": 1.977963924407959, "step": 9597 }, { "epoch": 1.1168256923434954, "grad_norm": 1.3884791135787964, "learning_rate": 0.0005, "loss": 1.935837745666504, "step": 9598 }, { "epoch": 1.1169420525948337, "grad_norm": 1.292780876159668, "learning_rate": 0.0005, "loss": 1.7744839191436768, "step": 9599 }, { "epoch": 1.1170584128461718, "grad_norm": 0.7816658020019531, "learning_rate": 0.0005, "loss": 1.273045539855957, "step": 9600 }, { "epoch": 1.11717477309751, "grad_norm": 1.0657329559326172, "learning_rate": 0.0005, "loss": 1.9882192611694336, "step": 9601 }, { "epoch": 1.117291133348848, "grad_norm": 0.6842100620269775, "learning_rate": 0.0005, "loss": 1.211108684539795, "step": 9602 }, { "epoch": 1.117407493600186, "grad_norm": 1.1283609867095947, "learning_rate": 0.0005, "loss": 2.184231996536255, "step": 9603 }, { "epoch": 1.1175238538515244, "grad_norm": 0.6660596132278442, "learning_rate": 0.0005, "loss": 1.4666626453399658, "step": 9604 }, { "epoch": 1.1176402141028625, "grad_norm": 1.3382843732833862, "learning_rate": 0.0005, "loss": 1.5917894840240479, "step": 9605 }, { "epoch": 1.1177565743542006, "grad_norm": 1.2478246688842773, "learning_rate": 0.0005, "loss": 1.9631907939910889, "step": 9606 }, { "epoch": 1.1178729346055387, "grad_norm": 1.3613699674606323, "learning_rate": 0.0005, "loss": 1.6659690141677856, "step": 9607 }, { "epoch": 1.1179892948568768, "grad_norm": 0.8595877289772034, "learning_rate": 0.0005, "loss": 2.1483001708984375, "step": 9608 }, { "epoch": 1.1181056551082151, "grad_norm": 0.9672869443893433, "learning_rate": 0.0005, "loss": 1.5265202522277832, "step": 9609 }, { "epoch": 1.1182220153595532, "grad_norm": 0.7106420993804932, "learning_rate": 0.0005, "loss": 2.520597457885742, "step": 9610 }, { "epoch": 1.1183383756108913, "grad_norm": 1.070279836654663, "learning_rate": 0.0005, "loss": 2.102039098739624, "step": 9611 }, { "epoch": 1.1184547358622294, "grad_norm": 1.0323973894119263, "learning_rate": 0.0005, "loss": 1.476513385772705, "step": 9612 }, { "epoch": 1.1185710961135675, "grad_norm": 1.2694215774536133, "learning_rate": 0.0005, "loss": 2.0311386585235596, "step": 9613 }, { "epoch": 1.1186874563649059, "grad_norm": 1.5536811351776123, "learning_rate": 0.0005, "loss": 2.2495763301849365, "step": 9614 }, { "epoch": 1.118803816616244, "grad_norm": 0.9308979511260986, "learning_rate": 0.0005, "loss": 1.5263748168945312, "step": 9615 }, { "epoch": 1.118920176867582, "grad_norm": 1.6874133348464966, "learning_rate": 0.0005, "loss": 2.5965921878814697, "step": 9616 }, { "epoch": 1.1190365371189201, "grad_norm": 0.7266198396682739, "learning_rate": 0.0005, "loss": 1.5477263927459717, "step": 9617 }, { "epoch": 1.1191528973702582, "grad_norm": 1.2013992071151733, "learning_rate": 0.0005, "loss": 1.4617809057235718, "step": 9618 }, { "epoch": 1.1192692576215966, "grad_norm": 0.9315011501312256, "learning_rate": 0.0005, "loss": 2.134899139404297, "step": 9619 }, { "epoch": 1.1193856178729347, "grad_norm": 1.0311317443847656, "learning_rate": 0.0005, "loss": 1.8085209131240845, "step": 9620 }, { "epoch": 1.1195019781242728, "grad_norm": 1.3976058959960938, "learning_rate": 0.0005, "loss": 1.8308796882629395, "step": 9621 }, { "epoch": 1.1196183383756109, "grad_norm": 1.0376033782958984, "learning_rate": 0.0005, "loss": 1.6673364639282227, "step": 9622 }, { "epoch": 1.119734698626949, "grad_norm": 1.3954272270202637, "learning_rate": 0.0005, "loss": 2.0595898628234863, "step": 9623 }, { "epoch": 1.1198510588782873, "grad_norm": 1.1546294689178467, "learning_rate": 0.0005, "loss": 1.9127485752105713, "step": 9624 }, { "epoch": 1.1199674191296254, "grad_norm": 0.9772277474403381, "learning_rate": 0.0005, "loss": 2.515883445739746, "step": 9625 }, { "epoch": 1.1200837793809635, "grad_norm": 0.9108421802520752, "learning_rate": 0.0005, "loss": 1.5828931331634521, "step": 9626 }, { "epoch": 1.1202001396323016, "grad_norm": 1.0503710508346558, "learning_rate": 0.0005, "loss": 1.6858813762664795, "step": 9627 }, { "epoch": 1.1203164998836397, "grad_norm": 0.9051483869552612, "learning_rate": 0.0005, "loss": 1.903854250907898, "step": 9628 }, { "epoch": 1.120432860134978, "grad_norm": 1.3161330223083496, "learning_rate": 0.0005, "loss": 3.3569114208221436, "step": 9629 }, { "epoch": 1.120549220386316, "grad_norm": 1.4961944818496704, "learning_rate": 0.0005, "loss": 1.7562299966812134, "step": 9630 }, { "epoch": 1.1206655806376542, "grad_norm": 1.3201982975006104, "learning_rate": 0.0005, "loss": 1.7763336896896362, "step": 9631 }, { "epoch": 1.1207819408889923, "grad_norm": 0.9548272490501404, "learning_rate": 0.0005, "loss": 1.5998635292053223, "step": 9632 }, { "epoch": 1.1208983011403304, "grad_norm": 1.50835120677948, "learning_rate": 0.0005, "loss": 1.7871326208114624, "step": 9633 }, { "epoch": 1.1210146613916687, "grad_norm": 0.9236119389533997, "learning_rate": 0.0005, "loss": 2.1695194244384766, "step": 9634 }, { "epoch": 1.1211310216430068, "grad_norm": 1.7362847328186035, "learning_rate": 0.0005, "loss": 1.603776216506958, "step": 9635 }, { "epoch": 1.121247381894345, "grad_norm": 1.2928285598754883, "learning_rate": 0.0005, "loss": 2.688170909881592, "step": 9636 }, { "epoch": 1.121363742145683, "grad_norm": 0.9352211952209473, "learning_rate": 0.0005, "loss": 2.2473859786987305, "step": 9637 }, { "epoch": 1.1214801023970211, "grad_norm": 1.3800855875015259, "learning_rate": 0.0005, "loss": 2.0683186054229736, "step": 9638 }, { "epoch": 1.1215964626483594, "grad_norm": 1.8481254577636719, "learning_rate": 0.0005, "loss": 1.9774606227874756, "step": 9639 }, { "epoch": 1.1217128228996975, "grad_norm": 0.9561550617218018, "learning_rate": 0.0005, "loss": 1.9049829244613647, "step": 9640 }, { "epoch": 1.1218291831510356, "grad_norm": 1.0091465711593628, "learning_rate": 0.0005, "loss": 1.8804950714111328, "step": 9641 }, { "epoch": 1.1219455434023737, "grad_norm": 1.0819685459136963, "learning_rate": 0.0005, "loss": 2.3314733505249023, "step": 9642 }, { "epoch": 1.1220619036537118, "grad_norm": 0.6482192277908325, "learning_rate": 0.0005, "loss": 1.0916554927825928, "step": 9643 }, { "epoch": 1.12217826390505, "grad_norm": 1.5434389114379883, "learning_rate": 0.0005, "loss": 2.1718664169311523, "step": 9644 }, { "epoch": 1.1222946241563883, "grad_norm": 0.8997504711151123, "learning_rate": 0.0005, "loss": 1.6131649017333984, "step": 9645 }, { "epoch": 1.1224109844077264, "grad_norm": 1.259385347366333, "learning_rate": 0.0005, "loss": 2.513845443725586, "step": 9646 }, { "epoch": 1.1225273446590645, "grad_norm": 0.7172149419784546, "learning_rate": 0.0005, "loss": 1.3780546188354492, "step": 9647 }, { "epoch": 1.1226437049104026, "grad_norm": 0.7403724789619446, "learning_rate": 0.0005, "loss": 1.6690069437026978, "step": 9648 }, { "epoch": 1.1227600651617407, "grad_norm": 1.2156928777694702, "learning_rate": 0.0005, "loss": 1.5173957347869873, "step": 9649 }, { "epoch": 1.122876425413079, "grad_norm": 1.3038549423217773, "learning_rate": 0.0005, "loss": 2.0100622177124023, "step": 9650 }, { "epoch": 1.122992785664417, "grad_norm": 0.8049188852310181, "learning_rate": 0.0005, "loss": 1.7458281517028809, "step": 9651 }, { "epoch": 1.1231091459157552, "grad_norm": 1.2459006309509277, "learning_rate": 0.0005, "loss": 2.4952993392944336, "step": 9652 }, { "epoch": 1.1232255061670933, "grad_norm": 0.38414186239242554, "learning_rate": 0.0005, "loss": 1.001937747001648, "step": 9653 }, { "epoch": 1.1233418664184314, "grad_norm": 0.8307240009307861, "learning_rate": 0.0005, "loss": 1.5865702629089355, "step": 9654 }, { "epoch": 1.1234582266697697, "grad_norm": 0.7620081305503845, "learning_rate": 0.0005, "loss": 1.2294578552246094, "step": 9655 }, { "epoch": 1.1235745869211078, "grad_norm": 1.6069104671478271, "learning_rate": 0.0005, "loss": 2.556802749633789, "step": 9656 }, { "epoch": 1.1236909471724459, "grad_norm": 1.6239066123962402, "learning_rate": 0.0005, "loss": 1.6542277336120605, "step": 9657 }, { "epoch": 1.123807307423784, "grad_norm": 0.8190427422523499, "learning_rate": 0.0005, "loss": 1.4669272899627686, "step": 9658 }, { "epoch": 1.123923667675122, "grad_norm": 1.2073997259140015, "learning_rate": 0.0005, "loss": 1.998917818069458, "step": 9659 }, { "epoch": 1.1240400279264604, "grad_norm": 1.0876169204711914, "learning_rate": 0.0005, "loss": 2.079914093017578, "step": 9660 }, { "epoch": 1.1241563881777985, "grad_norm": 1.236571192741394, "learning_rate": 0.0005, "loss": 1.6293740272521973, "step": 9661 }, { "epoch": 1.1242727484291366, "grad_norm": 0.9869362115859985, "learning_rate": 0.0005, "loss": 1.4703543186187744, "step": 9662 }, { "epoch": 1.1243891086804747, "grad_norm": 0.7483663558959961, "learning_rate": 0.0005, "loss": 1.5017063617706299, "step": 9663 }, { "epoch": 1.1245054689318128, "grad_norm": 1.4962447881698608, "learning_rate": 0.0005, "loss": 1.8047854900360107, "step": 9664 }, { "epoch": 1.1246218291831511, "grad_norm": 2.6974782943725586, "learning_rate": 0.0005, "loss": 2.768033027648926, "step": 9665 }, { "epoch": 1.1247381894344892, "grad_norm": 0.8847116827964783, "learning_rate": 0.0005, "loss": 2.1853742599487305, "step": 9666 }, { "epoch": 1.1248545496858273, "grad_norm": 1.1763948202133179, "learning_rate": 0.0005, "loss": 2.1101789474487305, "step": 9667 }, { "epoch": 1.1249709099371654, "grad_norm": 1.6359261274337769, "learning_rate": 0.0005, "loss": 2.5628714561462402, "step": 9668 }, { "epoch": 1.1250872701885035, "grad_norm": 0.9070422649383545, "learning_rate": 0.0005, "loss": 1.8790968656539917, "step": 9669 }, { "epoch": 1.1252036304398418, "grad_norm": 1.3340792655944824, "learning_rate": 0.0005, "loss": 1.9802005290985107, "step": 9670 }, { "epoch": 1.12531999069118, "grad_norm": 0.8162907958030701, "learning_rate": 0.0005, "loss": 1.6325634717941284, "step": 9671 }, { "epoch": 1.125436350942518, "grad_norm": 1.1179351806640625, "learning_rate": 0.0005, "loss": 2.2910239696502686, "step": 9672 }, { "epoch": 1.1255527111938561, "grad_norm": 0.9591838717460632, "learning_rate": 0.0005, "loss": 1.8546062707901, "step": 9673 }, { "epoch": 1.1256690714451942, "grad_norm": 1.7100988626480103, "learning_rate": 0.0005, "loss": 1.7272851467132568, "step": 9674 }, { "epoch": 1.1257854316965326, "grad_norm": 1.201399564743042, "learning_rate": 0.0005, "loss": 1.6586943864822388, "step": 9675 }, { "epoch": 1.1259017919478707, "grad_norm": 1.5193339586257935, "learning_rate": 0.0005, "loss": 2.1383771896362305, "step": 9676 }, { "epoch": 1.1260181521992088, "grad_norm": 0.9399427771568298, "learning_rate": 0.0005, "loss": 2.302752733230591, "step": 9677 }, { "epoch": 1.1261345124505469, "grad_norm": 1.172909140586853, "learning_rate": 0.0005, "loss": 1.5448493957519531, "step": 9678 }, { "epoch": 1.126250872701885, "grad_norm": 11.439671516418457, "learning_rate": 0.0005, "loss": 1.6043577194213867, "step": 9679 }, { "epoch": 1.1263672329532233, "grad_norm": 1.3733181953430176, "learning_rate": 0.0005, "loss": 2.0459835529327393, "step": 9680 }, { "epoch": 1.1264835932045614, "grad_norm": 1.1937997341156006, "learning_rate": 0.0005, "loss": 1.6074007749557495, "step": 9681 }, { "epoch": 1.1265999534558995, "grad_norm": 0.9245234131813049, "learning_rate": 0.0005, "loss": 1.667734146118164, "step": 9682 }, { "epoch": 1.1267163137072376, "grad_norm": 0.9464077353477478, "learning_rate": 0.0005, "loss": 1.636077642440796, "step": 9683 }, { "epoch": 1.1268326739585757, "grad_norm": 1.3766963481903076, "learning_rate": 0.0005, "loss": 2.272078514099121, "step": 9684 }, { "epoch": 1.1269490342099138, "grad_norm": 1.2862818241119385, "learning_rate": 0.0005, "loss": 1.9346586465835571, "step": 9685 }, { "epoch": 1.127065394461252, "grad_norm": 0.6764764189720154, "learning_rate": 0.0005, "loss": 1.1758216619491577, "step": 9686 }, { "epoch": 1.1271817547125902, "grad_norm": 1.1305440664291382, "learning_rate": 0.0005, "loss": 2.000314712524414, "step": 9687 }, { "epoch": 1.1272981149639283, "grad_norm": 1.5923449993133545, "learning_rate": 0.0005, "loss": 1.8435330390930176, "step": 9688 }, { "epoch": 1.1274144752152664, "grad_norm": 0.9637640118598938, "learning_rate": 0.0005, "loss": 2.0629048347473145, "step": 9689 }, { "epoch": 1.1275308354666045, "grad_norm": 1.1773650646209717, "learning_rate": 0.0005, "loss": 2.007204532623291, "step": 9690 }, { "epoch": 1.1276471957179428, "grad_norm": 0.5921338200569153, "learning_rate": 0.0005, "loss": 1.1124831438064575, "step": 9691 }, { "epoch": 1.127763555969281, "grad_norm": 1.2639694213867188, "learning_rate": 0.0005, "loss": 2.1562094688415527, "step": 9692 }, { "epoch": 1.127879916220619, "grad_norm": 1.1864933967590332, "learning_rate": 0.0005, "loss": 2.8481101989746094, "step": 9693 }, { "epoch": 1.127996276471957, "grad_norm": 1.3580663204193115, "learning_rate": 0.0005, "loss": 1.686387300491333, "step": 9694 }, { "epoch": 1.1281126367232952, "grad_norm": 0.994886040687561, "learning_rate": 0.0005, "loss": 1.8929994106292725, "step": 9695 }, { "epoch": 1.1282289969746335, "grad_norm": 1.427294135093689, "learning_rate": 0.0005, "loss": 1.7688219547271729, "step": 9696 }, { "epoch": 1.1283453572259716, "grad_norm": 1.432830572128296, "learning_rate": 0.0005, "loss": 2.3083577156066895, "step": 9697 }, { "epoch": 1.1284617174773097, "grad_norm": 0.8924670219421387, "learning_rate": 0.0005, "loss": 1.572977066040039, "step": 9698 }, { "epoch": 1.1285780777286478, "grad_norm": 0.5993791818618774, "learning_rate": 0.0005, "loss": 1.1575357913970947, "step": 9699 }, { "epoch": 1.128694437979986, "grad_norm": 0.7319068908691406, "learning_rate": 0.0005, "loss": 1.4982526302337646, "step": 9700 }, { "epoch": 1.1288107982313242, "grad_norm": 4.060693740844727, "learning_rate": 0.0005, "loss": 2.0611441135406494, "step": 9701 }, { "epoch": 1.1289271584826623, "grad_norm": 0.48351579904556274, "learning_rate": 0.0005, "loss": 1.21573805809021, "step": 9702 }, { "epoch": 1.1290435187340004, "grad_norm": 1.3068708181381226, "learning_rate": 0.0005, "loss": 2.014235734939575, "step": 9703 }, { "epoch": 1.1291598789853385, "grad_norm": 1.2493337392807007, "learning_rate": 0.0005, "loss": 2.2169384956359863, "step": 9704 }, { "epoch": 1.1292762392366766, "grad_norm": 1.1114418506622314, "learning_rate": 0.0005, "loss": 2.033161163330078, "step": 9705 }, { "epoch": 1.129392599488015, "grad_norm": 1.460833191871643, "learning_rate": 0.0005, "loss": 2.344870090484619, "step": 9706 }, { "epoch": 1.129508959739353, "grad_norm": 1.0870829820632935, "learning_rate": 0.0005, "loss": 2.126328468322754, "step": 9707 }, { "epoch": 1.1296253199906912, "grad_norm": 0.9892815947532654, "learning_rate": 0.0005, "loss": 1.8747384548187256, "step": 9708 }, { "epoch": 1.1297416802420293, "grad_norm": 1.0077216625213623, "learning_rate": 0.0005, "loss": 1.580617904663086, "step": 9709 }, { "epoch": 1.1298580404933674, "grad_norm": 1.5128475427627563, "learning_rate": 0.0005, "loss": 1.58515465259552, "step": 9710 }, { "epoch": 1.1299744007447057, "grad_norm": 1.1271286010742188, "learning_rate": 0.0005, "loss": 1.5857727527618408, "step": 9711 }, { "epoch": 1.1300907609960438, "grad_norm": 1.2546762228012085, "learning_rate": 0.0005, "loss": 2.0521669387817383, "step": 9712 }, { "epoch": 1.1302071212473819, "grad_norm": 1.362680435180664, "learning_rate": 0.0005, "loss": 2.25398588180542, "step": 9713 }, { "epoch": 1.13032348149872, "grad_norm": 1.7238037586212158, "learning_rate": 0.0005, "loss": 2.469813346862793, "step": 9714 }, { "epoch": 1.130439841750058, "grad_norm": 1.200558066368103, "learning_rate": 0.0005, "loss": 2.2048370838165283, "step": 9715 }, { "epoch": 1.1305562020013964, "grad_norm": 0.7683531641960144, "learning_rate": 0.0005, "loss": 1.6655158996582031, "step": 9716 }, { "epoch": 1.1306725622527345, "grad_norm": 1.873674988746643, "learning_rate": 0.0005, "loss": 2.549983024597168, "step": 9717 }, { "epoch": 1.1307889225040726, "grad_norm": 1.2420700788497925, "learning_rate": 0.0005, "loss": 2.7966151237487793, "step": 9718 }, { "epoch": 1.1309052827554107, "grad_norm": 1.189638376235962, "learning_rate": 0.0005, "loss": 1.7139158248901367, "step": 9719 }, { "epoch": 1.1310216430067488, "grad_norm": 0.9257033467292786, "learning_rate": 0.0005, "loss": 1.7985150814056396, "step": 9720 }, { "epoch": 1.131138003258087, "grad_norm": 0.9332889914512634, "learning_rate": 0.0005, "loss": 2.0541510581970215, "step": 9721 }, { "epoch": 1.1312543635094252, "grad_norm": 1.5236890316009521, "learning_rate": 0.0005, "loss": 1.7194924354553223, "step": 9722 }, { "epoch": 1.1313707237607633, "grad_norm": 0.8871192932128906, "learning_rate": 0.0005, "loss": 1.9419856071472168, "step": 9723 }, { "epoch": 1.1314870840121014, "grad_norm": 0.9380793571472168, "learning_rate": 0.0005, "loss": 1.5498839616775513, "step": 9724 }, { "epoch": 1.1316034442634395, "grad_norm": 0.986492931842804, "learning_rate": 0.0005, "loss": 1.566847801208496, "step": 9725 }, { "epoch": 1.1317198045147778, "grad_norm": 1.1801704168319702, "learning_rate": 0.0005, "loss": 2.56211256980896, "step": 9726 }, { "epoch": 1.131836164766116, "grad_norm": 1.5722599029541016, "learning_rate": 0.0005, "loss": 1.502285122871399, "step": 9727 }, { "epoch": 1.131952525017454, "grad_norm": 0.9750563502311707, "learning_rate": 0.0005, "loss": 2.0880556106567383, "step": 9728 }, { "epoch": 1.1320688852687921, "grad_norm": 0.9634859561920166, "learning_rate": 0.0005, "loss": 1.6657986640930176, "step": 9729 }, { "epoch": 1.1321852455201302, "grad_norm": 1.2102779150009155, "learning_rate": 0.0005, "loss": 1.940477728843689, "step": 9730 }, { "epoch": 1.1323016057714685, "grad_norm": 1.125681757926941, "learning_rate": 0.0005, "loss": 1.9343578815460205, "step": 9731 }, { "epoch": 1.1324179660228066, "grad_norm": 0.8477080464363098, "learning_rate": 0.0005, "loss": 1.9358175992965698, "step": 9732 }, { "epoch": 1.1325343262741447, "grad_norm": 1.5787509679794312, "learning_rate": 0.0005, "loss": 1.6403201818466187, "step": 9733 }, { "epoch": 1.1326506865254828, "grad_norm": 0.9039691090583801, "learning_rate": 0.0005, "loss": 1.4013944864273071, "step": 9734 }, { "epoch": 1.132767046776821, "grad_norm": 1.515966773033142, "learning_rate": 0.0005, "loss": 2.3212943077087402, "step": 9735 }, { "epoch": 1.1328834070281593, "grad_norm": 1.0970019102096558, "learning_rate": 0.0005, "loss": 2.2146363258361816, "step": 9736 }, { "epoch": 1.1329997672794974, "grad_norm": 1.3696398735046387, "learning_rate": 0.0005, "loss": 2.369476795196533, "step": 9737 }, { "epoch": 1.1331161275308355, "grad_norm": 1.5284579992294312, "learning_rate": 0.0005, "loss": 2.2599406242370605, "step": 9738 }, { "epoch": 1.1332324877821736, "grad_norm": 1.0939977169036865, "learning_rate": 0.0005, "loss": 1.8717522621154785, "step": 9739 }, { "epoch": 1.1333488480335117, "grad_norm": 1.4901108741760254, "learning_rate": 0.0005, "loss": 1.9946802854537964, "step": 9740 }, { "epoch": 1.13346520828485, "grad_norm": 0.7270864248275757, "learning_rate": 0.0005, "loss": 1.5480139255523682, "step": 9741 }, { "epoch": 1.133581568536188, "grad_norm": 1.9097391366958618, "learning_rate": 0.0005, "loss": 3.0461764335632324, "step": 9742 }, { "epoch": 1.1336979287875262, "grad_norm": 1.180654764175415, "learning_rate": 0.0005, "loss": 2.005608558654785, "step": 9743 }, { "epoch": 1.1338142890388643, "grad_norm": 1.2728699445724487, "learning_rate": 0.0005, "loss": 1.675351858139038, "step": 9744 }, { "epoch": 1.1339306492902024, "grad_norm": 1.314315676689148, "learning_rate": 0.0005, "loss": 1.6109620332717896, "step": 9745 }, { "epoch": 1.1340470095415407, "grad_norm": 1.032898187637329, "learning_rate": 0.0005, "loss": 2.0727920532226562, "step": 9746 }, { "epoch": 1.1341633697928788, "grad_norm": 1.1780402660369873, "learning_rate": 0.0005, "loss": 1.399658203125, "step": 9747 }, { "epoch": 1.134279730044217, "grad_norm": 0.9941393136978149, "learning_rate": 0.0005, "loss": 1.4632267951965332, "step": 9748 }, { "epoch": 1.134396090295555, "grad_norm": 0.7329675555229187, "learning_rate": 0.0005, "loss": 1.32142174243927, "step": 9749 }, { "epoch": 1.134512450546893, "grad_norm": 1.1157443523406982, "learning_rate": 0.0005, "loss": 1.4847002029418945, "step": 9750 }, { "epoch": 1.1346288107982314, "grad_norm": 1.4235674142837524, "learning_rate": 0.0005, "loss": 1.6521506309509277, "step": 9751 }, { "epoch": 1.1347451710495695, "grad_norm": 1.4674592018127441, "learning_rate": 0.0005, "loss": 1.8779475688934326, "step": 9752 }, { "epoch": 1.1348615313009076, "grad_norm": 0.8537184000015259, "learning_rate": 0.0005, "loss": 1.3271863460540771, "step": 9753 }, { "epoch": 1.1349778915522457, "grad_norm": 1.046595811843872, "learning_rate": 0.0005, "loss": 1.4296462535858154, "step": 9754 }, { "epoch": 1.1350942518035838, "grad_norm": 1.8889497518539429, "learning_rate": 0.0005, "loss": 3.1882858276367188, "step": 9755 }, { "epoch": 1.1352106120549221, "grad_norm": 1.1068633794784546, "learning_rate": 0.0005, "loss": 1.8821965456008911, "step": 9756 }, { "epoch": 1.1353269723062602, "grad_norm": 1.236063003540039, "learning_rate": 0.0005, "loss": 1.5030953884124756, "step": 9757 }, { "epoch": 1.1354433325575983, "grad_norm": 1.1985588073730469, "learning_rate": 0.0005, "loss": 1.6996219158172607, "step": 9758 }, { "epoch": 1.1355596928089364, "grad_norm": 0.8035512566566467, "learning_rate": 0.0005, "loss": 1.3895015716552734, "step": 9759 }, { "epoch": 1.1356760530602745, "grad_norm": 1.1790772676467896, "learning_rate": 0.0005, "loss": 1.7151401042938232, "step": 9760 }, { "epoch": 1.1357924133116128, "grad_norm": 1.5829496383666992, "learning_rate": 0.0005, "loss": 1.8590571880340576, "step": 9761 }, { "epoch": 1.135908773562951, "grad_norm": 2.923159599304199, "learning_rate": 0.0005, "loss": 1.2727656364440918, "step": 9762 }, { "epoch": 1.136025133814289, "grad_norm": 1.2414222955703735, "learning_rate": 0.0005, "loss": 2.4112019538879395, "step": 9763 }, { "epoch": 1.1361414940656271, "grad_norm": 0.919888436794281, "learning_rate": 0.0005, "loss": 2.219402313232422, "step": 9764 }, { "epoch": 1.1362578543169652, "grad_norm": 1.5442380905151367, "learning_rate": 0.0005, "loss": 2.314396858215332, "step": 9765 }, { "epoch": 1.1363742145683036, "grad_norm": 1.4225972890853882, "learning_rate": 0.0005, "loss": 2.2015457153320312, "step": 9766 }, { "epoch": 1.1364905748196417, "grad_norm": 0.6516631245613098, "learning_rate": 0.0005, "loss": 1.9120522737503052, "step": 9767 }, { "epoch": 1.1366069350709798, "grad_norm": 0.9055142998695374, "learning_rate": 0.0005, "loss": 2.0147175788879395, "step": 9768 }, { "epoch": 1.1367232953223179, "grad_norm": 1.4454095363616943, "learning_rate": 0.0005, "loss": 1.7439396381378174, "step": 9769 }, { "epoch": 1.136839655573656, "grad_norm": 0.5266249775886536, "learning_rate": 0.0005, "loss": 1.17266845703125, "step": 9770 }, { "epoch": 1.1369560158249943, "grad_norm": 1.083418607711792, "learning_rate": 0.0005, "loss": 2.1240105628967285, "step": 9771 }, { "epoch": 1.1370723760763324, "grad_norm": 0.9407041668891907, "learning_rate": 0.0005, "loss": 1.7612828016281128, "step": 9772 }, { "epoch": 1.1371887363276705, "grad_norm": 1.0713589191436768, "learning_rate": 0.0005, "loss": 2.3779280185699463, "step": 9773 }, { "epoch": 1.1373050965790086, "grad_norm": 1.5390052795410156, "learning_rate": 0.0005, "loss": 2.1733837127685547, "step": 9774 }, { "epoch": 1.1374214568303467, "grad_norm": 0.9457529783248901, "learning_rate": 0.0005, "loss": 1.8186390399932861, "step": 9775 }, { "epoch": 1.137537817081685, "grad_norm": 1.6122990846633911, "learning_rate": 0.0005, "loss": 1.7573466300964355, "step": 9776 }, { "epoch": 1.137654177333023, "grad_norm": 1.3542360067367554, "learning_rate": 0.0005, "loss": 2.1380059719085693, "step": 9777 }, { "epoch": 1.1377705375843612, "grad_norm": 14.755228996276855, "learning_rate": 0.0005, "loss": 1.5388422012329102, "step": 9778 }, { "epoch": 1.1378868978356993, "grad_norm": 0.8375025987625122, "learning_rate": 0.0005, "loss": 2.0532407760620117, "step": 9779 }, { "epoch": 1.1380032580870374, "grad_norm": 0.9476334452629089, "learning_rate": 0.0005, "loss": 2.253931760787964, "step": 9780 }, { "epoch": 1.1381196183383757, "grad_norm": 1.559467077255249, "learning_rate": 0.0005, "loss": 2.3739256858825684, "step": 9781 }, { "epoch": 1.1382359785897138, "grad_norm": 0.8439242243766785, "learning_rate": 0.0005, "loss": 2.099916458129883, "step": 9782 }, { "epoch": 1.138352338841052, "grad_norm": 1.5743499994277954, "learning_rate": 0.0005, "loss": 2.01066255569458, "step": 9783 }, { "epoch": 1.13846869909239, "grad_norm": 0.9997684359550476, "learning_rate": 0.0005, "loss": 1.7985029220581055, "step": 9784 }, { "epoch": 1.1385850593437281, "grad_norm": 0.6247989535331726, "learning_rate": 0.0005, "loss": 1.392967700958252, "step": 9785 }, { "epoch": 1.1387014195950664, "grad_norm": 1.3620952367782593, "learning_rate": 0.0005, "loss": 2.3571789264678955, "step": 9786 }, { "epoch": 1.1388177798464045, "grad_norm": 1.3647396564483643, "learning_rate": 0.0005, "loss": 2.076201915740967, "step": 9787 }, { "epoch": 1.1389341400977426, "grad_norm": 1.1674998998641968, "learning_rate": 0.0005, "loss": 2.589503526687622, "step": 9788 }, { "epoch": 1.1390505003490807, "grad_norm": 1.6064139604568481, "learning_rate": 0.0005, "loss": 2.1840686798095703, "step": 9789 }, { "epoch": 1.1391668606004188, "grad_norm": 1.7664812803268433, "learning_rate": 0.0005, "loss": 2.382967948913574, "step": 9790 }, { "epoch": 1.1392832208517572, "grad_norm": 1.2675061225891113, "learning_rate": 0.0005, "loss": 1.7883681058883667, "step": 9791 }, { "epoch": 1.1393995811030952, "grad_norm": 1.465924859046936, "learning_rate": 0.0005, "loss": 2.040971279144287, "step": 9792 }, { "epoch": 1.1395159413544333, "grad_norm": 0.8815763592720032, "learning_rate": 0.0005, "loss": 2.1650166511535645, "step": 9793 }, { "epoch": 1.1396323016057714, "grad_norm": 2.09008526802063, "learning_rate": 0.0005, "loss": 2.4136624336242676, "step": 9794 }, { "epoch": 1.1397486618571095, "grad_norm": 1.0155493021011353, "learning_rate": 0.0005, "loss": 1.7215193510055542, "step": 9795 }, { "epoch": 1.1398650221084479, "grad_norm": 1.1466140747070312, "learning_rate": 0.0005, "loss": 2.016530752182007, "step": 9796 }, { "epoch": 1.139981382359786, "grad_norm": 1.1784186363220215, "learning_rate": 0.0005, "loss": 1.4063408374786377, "step": 9797 }, { "epoch": 1.140097742611124, "grad_norm": 0.9552853107452393, "learning_rate": 0.0005, "loss": 1.9232847690582275, "step": 9798 }, { "epoch": 1.1402141028624622, "grad_norm": 1.2603379487991333, "learning_rate": 0.0005, "loss": 2.33552885055542, "step": 9799 }, { "epoch": 1.1403304631138003, "grad_norm": 2.0889034271240234, "learning_rate": 0.0005, "loss": 1.5696252584457397, "step": 9800 }, { "epoch": 1.1404468233651386, "grad_norm": 1.120712161064148, "learning_rate": 0.0005, "loss": 1.3597121238708496, "step": 9801 }, { "epoch": 1.1405631836164767, "grad_norm": 1.3769702911376953, "learning_rate": 0.0005, "loss": 1.8746492862701416, "step": 9802 }, { "epoch": 1.1406795438678148, "grad_norm": 1.4028385877609253, "learning_rate": 0.0005, "loss": 2.959367275238037, "step": 9803 }, { "epoch": 1.1407959041191529, "grad_norm": 1.0713202953338623, "learning_rate": 0.0005, "loss": 1.5398770570755005, "step": 9804 }, { "epoch": 1.140912264370491, "grad_norm": 1.1046451330184937, "learning_rate": 0.0005, "loss": 1.733201265335083, "step": 9805 }, { "epoch": 1.1410286246218293, "grad_norm": 1.1702433824539185, "learning_rate": 0.0005, "loss": 2.222543716430664, "step": 9806 }, { "epoch": 1.1411449848731674, "grad_norm": 0.9376416802406311, "learning_rate": 0.0005, "loss": 1.2330888509750366, "step": 9807 }, { "epoch": 1.1412613451245055, "grad_norm": 0.9741355776786804, "learning_rate": 0.0005, "loss": 1.5630884170532227, "step": 9808 }, { "epoch": 1.1413777053758436, "grad_norm": 1.0496082305908203, "learning_rate": 0.0005, "loss": 1.7369976043701172, "step": 9809 }, { "epoch": 1.1414940656271817, "grad_norm": 1.33891761302948, "learning_rate": 0.0005, "loss": 2.268725872039795, "step": 9810 }, { "epoch": 1.14161042587852, "grad_norm": 1.4874342679977417, "learning_rate": 0.0005, "loss": 2.0148699283599854, "step": 9811 }, { "epoch": 1.1417267861298581, "grad_norm": 0.9631057381629944, "learning_rate": 0.0005, "loss": 1.6854993104934692, "step": 9812 }, { "epoch": 1.1418431463811962, "grad_norm": 1.1991603374481201, "learning_rate": 0.0005, "loss": 3.2543869018554688, "step": 9813 }, { "epoch": 1.1419595066325343, "grad_norm": 1.4550567865371704, "learning_rate": 0.0005, "loss": 2.3900551795959473, "step": 9814 }, { "epoch": 1.1420758668838724, "grad_norm": 1.2184761762619019, "learning_rate": 0.0005, "loss": 2.11645770072937, "step": 9815 }, { "epoch": 1.1421922271352107, "grad_norm": 1.413215160369873, "learning_rate": 0.0005, "loss": 1.7548437118530273, "step": 9816 }, { "epoch": 1.1423085873865488, "grad_norm": 0.88514244556427, "learning_rate": 0.0005, "loss": 1.7575749158859253, "step": 9817 }, { "epoch": 1.142424947637887, "grad_norm": 1.3221279382705688, "learning_rate": 0.0005, "loss": 1.8531410694122314, "step": 9818 }, { "epoch": 1.142541307889225, "grad_norm": 1.113596796989441, "learning_rate": 0.0005, "loss": 1.6303560733795166, "step": 9819 }, { "epoch": 1.1426576681405631, "grad_norm": 0.9403806924819946, "learning_rate": 0.0005, "loss": 1.5859484672546387, "step": 9820 }, { "epoch": 1.1427740283919012, "grad_norm": 0.9528055191040039, "learning_rate": 0.0005, "loss": 1.5211520195007324, "step": 9821 }, { "epoch": 1.1428903886432396, "grad_norm": 0.8082841634750366, "learning_rate": 0.0005, "loss": 2.0999298095703125, "step": 9822 }, { "epoch": 1.1430067488945777, "grad_norm": 1.0306822061538696, "learning_rate": 0.0005, "loss": 2.1724510192871094, "step": 9823 }, { "epoch": 1.1431231091459158, "grad_norm": 0.8978529572486877, "learning_rate": 0.0005, "loss": 1.368501901626587, "step": 9824 }, { "epoch": 1.1432394693972538, "grad_norm": 1.0048747062683105, "learning_rate": 0.0005, "loss": 1.974041223526001, "step": 9825 }, { "epoch": 1.143355829648592, "grad_norm": 0.9734175205230713, "learning_rate": 0.0005, "loss": 1.9317359924316406, "step": 9826 }, { "epoch": 1.1434721898999303, "grad_norm": 1.2763476371765137, "learning_rate": 0.0005, "loss": 2.3267650604248047, "step": 9827 }, { "epoch": 1.1435885501512684, "grad_norm": 1.3630695343017578, "learning_rate": 0.0005, "loss": 2.5809974670410156, "step": 9828 }, { "epoch": 1.1437049104026065, "grad_norm": 1.8058949708938599, "learning_rate": 0.0005, "loss": 2.1115007400512695, "step": 9829 }, { "epoch": 1.1438212706539446, "grad_norm": 0.844076931476593, "learning_rate": 0.0005, "loss": 1.6367487907409668, "step": 9830 }, { "epoch": 1.1439376309052827, "grad_norm": 1.2672728300094604, "learning_rate": 0.0005, "loss": 2.163909673690796, "step": 9831 }, { "epoch": 1.144053991156621, "grad_norm": 1.1905200481414795, "learning_rate": 0.0005, "loss": 2.2521004676818848, "step": 9832 }, { "epoch": 1.144170351407959, "grad_norm": 1.234753131866455, "learning_rate": 0.0005, "loss": 1.7714574337005615, "step": 9833 }, { "epoch": 1.1442867116592972, "grad_norm": 1.5145525932312012, "learning_rate": 0.0005, "loss": 1.4721927642822266, "step": 9834 }, { "epoch": 1.1444030719106353, "grad_norm": 0.9524219632148743, "learning_rate": 0.0005, "loss": 1.8913943767547607, "step": 9835 }, { "epoch": 1.1445194321619734, "grad_norm": 1.8911921977996826, "learning_rate": 0.0005, "loss": 1.861358404159546, "step": 9836 }, { "epoch": 1.1446357924133117, "grad_norm": 1.0664061307907104, "learning_rate": 0.0005, "loss": 2.1525511741638184, "step": 9837 }, { "epoch": 1.1447521526646498, "grad_norm": 1.0896670818328857, "learning_rate": 0.0005, "loss": 1.9449577331542969, "step": 9838 }, { "epoch": 1.144868512915988, "grad_norm": 1.031850814819336, "learning_rate": 0.0005, "loss": 1.9337711334228516, "step": 9839 }, { "epoch": 1.144984873167326, "grad_norm": 1.3857401609420776, "learning_rate": 0.0005, "loss": 1.8655579090118408, "step": 9840 }, { "epoch": 1.145101233418664, "grad_norm": 1.3530516624450684, "learning_rate": 0.0005, "loss": 1.868156909942627, "step": 9841 }, { "epoch": 1.1452175936700024, "grad_norm": 1.7396456003189087, "learning_rate": 0.0005, "loss": 2.5511727333068848, "step": 9842 }, { "epoch": 1.1453339539213405, "grad_norm": 1.6619749069213867, "learning_rate": 0.0005, "loss": 2.6518380641937256, "step": 9843 }, { "epoch": 1.1454503141726786, "grad_norm": 1.1328312158584595, "learning_rate": 0.0005, "loss": 2.151862621307373, "step": 9844 }, { "epoch": 1.1455666744240167, "grad_norm": 1.3277099132537842, "learning_rate": 0.0005, "loss": 1.912729263305664, "step": 9845 }, { "epoch": 1.1456830346753548, "grad_norm": 0.9978898167610168, "learning_rate": 0.0005, "loss": 1.633498191833496, "step": 9846 }, { "epoch": 1.1457993949266931, "grad_norm": 1.709738850593567, "learning_rate": 0.0005, "loss": 2.9427084922790527, "step": 9847 }, { "epoch": 1.1459157551780312, "grad_norm": 1.00983464717865, "learning_rate": 0.0005, "loss": 1.5968573093414307, "step": 9848 }, { "epoch": 1.1460321154293693, "grad_norm": 1.3762935400009155, "learning_rate": 0.0005, "loss": 1.7843520641326904, "step": 9849 }, { "epoch": 1.1461484756807074, "grad_norm": 1.5460612773895264, "learning_rate": 0.0005, "loss": 2.406435489654541, "step": 9850 }, { "epoch": 1.1462648359320455, "grad_norm": 0.8522940278053284, "learning_rate": 0.0005, "loss": 1.6660221815109253, "step": 9851 }, { "epoch": 1.1463811961833836, "grad_norm": 1.4815675020217896, "learning_rate": 0.0005, "loss": 2.1056458950042725, "step": 9852 }, { "epoch": 1.146497556434722, "grad_norm": 1.0047097206115723, "learning_rate": 0.0005, "loss": 1.576993703842163, "step": 9853 }, { "epoch": 1.14661391668606, "grad_norm": 1.1893614530563354, "learning_rate": 0.0005, "loss": 1.4629600048065186, "step": 9854 }, { "epoch": 1.1467302769373982, "grad_norm": 1.1252567768096924, "learning_rate": 0.0005, "loss": 2.1884772777557373, "step": 9855 }, { "epoch": 1.1468466371887363, "grad_norm": 1.6421409845352173, "learning_rate": 0.0005, "loss": 2.0118765830993652, "step": 9856 }, { "epoch": 1.1469629974400743, "grad_norm": 0.8174896240234375, "learning_rate": 0.0005, "loss": 2.0277814865112305, "step": 9857 }, { "epoch": 1.1470793576914127, "grad_norm": 0.8891903162002563, "learning_rate": 0.0005, "loss": 1.9023616313934326, "step": 9858 }, { "epoch": 1.1471957179427508, "grad_norm": 1.1126786470413208, "learning_rate": 0.0005, "loss": 1.6132698059082031, "step": 9859 }, { "epoch": 1.1473120781940889, "grad_norm": 0.5844160318374634, "learning_rate": 0.0005, "loss": 1.3710334300994873, "step": 9860 }, { "epoch": 1.147428438445427, "grad_norm": 1.7591220140457153, "learning_rate": 0.0005, "loss": 2.093163013458252, "step": 9861 }, { "epoch": 1.147544798696765, "grad_norm": 1.3544938564300537, "learning_rate": 0.0005, "loss": 2.7682647705078125, "step": 9862 }, { "epoch": 1.1476611589481034, "grad_norm": 1.0884047746658325, "learning_rate": 0.0005, "loss": 1.6667722463607788, "step": 9863 }, { "epoch": 1.1477775191994415, "grad_norm": 1.1437088251113892, "learning_rate": 0.0005, "loss": 1.917694330215454, "step": 9864 }, { "epoch": 1.1478938794507796, "grad_norm": 1.2468430995941162, "learning_rate": 0.0005, "loss": 2.5176243782043457, "step": 9865 }, { "epoch": 1.1480102397021177, "grad_norm": 1.405680775642395, "learning_rate": 0.0005, "loss": 2.5886576175689697, "step": 9866 }, { "epoch": 1.1481265999534558, "grad_norm": 1.336422085762024, "learning_rate": 0.0005, "loss": 2.752473831176758, "step": 9867 }, { "epoch": 1.148242960204794, "grad_norm": 1.1389156579971313, "learning_rate": 0.0005, "loss": 1.7215372323989868, "step": 9868 }, { "epoch": 1.1483593204561322, "grad_norm": 1.3295797109603882, "learning_rate": 0.0005, "loss": 2.4077658653259277, "step": 9869 }, { "epoch": 1.1484756807074703, "grad_norm": 1.1711689233779907, "learning_rate": 0.0005, "loss": 1.7799986600875854, "step": 9870 }, { "epoch": 1.1485920409588084, "grad_norm": 2.133561849594116, "learning_rate": 0.0005, "loss": 1.5058180093765259, "step": 9871 }, { "epoch": 1.1487084012101465, "grad_norm": 1.3318730592727661, "learning_rate": 0.0005, "loss": 2.122535467147827, "step": 9872 }, { "epoch": 1.1488247614614848, "grad_norm": 0.949949324131012, "learning_rate": 0.0005, "loss": 1.2276593446731567, "step": 9873 }, { "epoch": 1.148941121712823, "grad_norm": 1.532450795173645, "learning_rate": 0.0005, "loss": 2.282440185546875, "step": 9874 }, { "epoch": 1.149057481964161, "grad_norm": 0.956377387046814, "learning_rate": 0.0005, "loss": 2.0405328273773193, "step": 9875 }, { "epoch": 1.1491738422154991, "grad_norm": 1.0474656820297241, "learning_rate": 0.0005, "loss": 2.3781120777130127, "step": 9876 }, { "epoch": 1.1492902024668372, "grad_norm": 1.5914373397827148, "learning_rate": 0.0005, "loss": 1.753712773323059, "step": 9877 }, { "epoch": 1.1494065627181755, "grad_norm": 1.220468282699585, "learning_rate": 0.0005, "loss": 2.6538784503936768, "step": 9878 }, { "epoch": 1.1495229229695136, "grad_norm": 1.0380616188049316, "learning_rate": 0.0005, "loss": 2.159984827041626, "step": 9879 }, { "epoch": 1.1496392832208517, "grad_norm": 0.8874942660331726, "learning_rate": 0.0005, "loss": 1.4935111999511719, "step": 9880 }, { "epoch": 1.1497556434721898, "grad_norm": 0.8616886734962463, "learning_rate": 0.0005, "loss": 1.7459841966629028, "step": 9881 }, { "epoch": 1.149872003723528, "grad_norm": 1.6759281158447266, "learning_rate": 0.0005, "loss": 2.2613508701324463, "step": 9882 }, { "epoch": 1.1499883639748663, "grad_norm": 2.0842630863189697, "learning_rate": 0.0005, "loss": 2.073941707611084, "step": 9883 }, { "epoch": 1.1501047242262044, "grad_norm": 1.1422364711761475, "learning_rate": 0.0005, "loss": 1.9808663129806519, "step": 9884 }, { "epoch": 1.1502210844775425, "grad_norm": 1.4874777793884277, "learning_rate": 0.0005, "loss": 2.2214596271514893, "step": 9885 }, { "epoch": 1.1503374447288806, "grad_norm": 1.4146119356155396, "learning_rate": 0.0005, "loss": 1.34614896774292, "step": 9886 }, { "epoch": 1.1504538049802187, "grad_norm": 1.5611727237701416, "learning_rate": 0.0005, "loss": 1.9106773138046265, "step": 9887 }, { "epoch": 1.150570165231557, "grad_norm": 0.8603689670562744, "learning_rate": 0.0005, "loss": 2.0175368785858154, "step": 9888 }, { "epoch": 1.150686525482895, "grad_norm": 0.986005961894989, "learning_rate": 0.0005, "loss": 1.4657646417617798, "step": 9889 }, { "epoch": 1.1508028857342332, "grad_norm": 1.1678972244262695, "learning_rate": 0.0005, "loss": 2.192519187927246, "step": 9890 }, { "epoch": 1.1509192459855713, "grad_norm": 1.4882677793502808, "learning_rate": 0.0005, "loss": 2.9829931259155273, "step": 9891 }, { "epoch": 1.1510356062369094, "grad_norm": 1.1829416751861572, "learning_rate": 0.0005, "loss": 2.64827561378479, "step": 9892 }, { "epoch": 1.1511519664882477, "grad_norm": 1.241912841796875, "learning_rate": 0.0005, "loss": 1.7941536903381348, "step": 9893 }, { "epoch": 1.1512683267395858, "grad_norm": 1.5847519636154175, "learning_rate": 0.0005, "loss": 1.366138219833374, "step": 9894 }, { "epoch": 1.151384686990924, "grad_norm": 1.2031383514404297, "learning_rate": 0.0005, "loss": 2.1274917125701904, "step": 9895 }, { "epoch": 1.151501047242262, "grad_norm": 1.4562002420425415, "learning_rate": 0.0005, "loss": 2.7272400856018066, "step": 9896 }, { "epoch": 1.1516174074936, "grad_norm": 1.0617260932922363, "learning_rate": 0.0005, "loss": 2.555389642715454, "step": 9897 }, { "epoch": 1.1517337677449384, "grad_norm": 1.6892036199569702, "learning_rate": 0.0005, "loss": 3.171008586883545, "step": 9898 }, { "epoch": 1.1518501279962765, "grad_norm": 1.0480271577835083, "learning_rate": 0.0005, "loss": 1.6878273487091064, "step": 9899 }, { "epoch": 1.1519664882476146, "grad_norm": 1.6798914670944214, "learning_rate": 0.0005, "loss": 1.8098888397216797, "step": 9900 }, { "epoch": 1.1520828484989527, "grad_norm": 1.0952717065811157, "learning_rate": 0.0005, "loss": 3.047691822052002, "step": 9901 }, { "epoch": 1.1521992087502908, "grad_norm": 1.0732611417770386, "learning_rate": 0.0005, "loss": 2.1317827701568604, "step": 9902 }, { "epoch": 1.1523155690016291, "grad_norm": 0.7860086560249329, "learning_rate": 0.0005, "loss": 1.2165963649749756, "step": 9903 }, { "epoch": 1.1524319292529672, "grad_norm": 1.1199760437011719, "learning_rate": 0.0005, "loss": 2.5234856605529785, "step": 9904 }, { "epoch": 1.1525482895043053, "grad_norm": 1.397545576095581, "learning_rate": 0.0005, "loss": 2.36689829826355, "step": 9905 }, { "epoch": 1.1526646497556434, "grad_norm": 1.0831313133239746, "learning_rate": 0.0005, "loss": 2.502432107925415, "step": 9906 }, { "epoch": 1.1527810100069815, "grad_norm": 1.219292163848877, "learning_rate": 0.0005, "loss": 2.162642478942871, "step": 9907 }, { "epoch": 1.1528973702583198, "grad_norm": 1.3055479526519775, "learning_rate": 0.0005, "loss": 2.2972211837768555, "step": 9908 }, { "epoch": 1.153013730509658, "grad_norm": 1.217087984085083, "learning_rate": 0.0005, "loss": 1.8481347560882568, "step": 9909 }, { "epoch": 1.153130090760996, "grad_norm": 1.2068785429000854, "learning_rate": 0.0005, "loss": 1.6852296590805054, "step": 9910 }, { "epoch": 1.1532464510123341, "grad_norm": 1.3835541009902954, "learning_rate": 0.0005, "loss": 1.5508406162261963, "step": 9911 }, { "epoch": 1.1533628112636722, "grad_norm": 1.4498746395111084, "learning_rate": 0.0005, "loss": 2.696284294128418, "step": 9912 }, { "epoch": 1.1534791715150106, "grad_norm": 1.015337347984314, "learning_rate": 0.0005, "loss": 1.8376119136810303, "step": 9913 }, { "epoch": 1.1535955317663487, "grad_norm": 1.5791728496551514, "learning_rate": 0.0005, "loss": 1.7649681568145752, "step": 9914 }, { "epoch": 1.1537118920176868, "grad_norm": 0.9820184111595154, "learning_rate": 0.0005, "loss": 1.601088047027588, "step": 9915 }, { "epoch": 1.1538282522690249, "grad_norm": 1.166978120803833, "learning_rate": 0.0005, "loss": 1.7823129892349243, "step": 9916 }, { "epoch": 1.153944612520363, "grad_norm": 1.1088217496871948, "learning_rate": 0.0005, "loss": 1.7382152080535889, "step": 9917 }, { "epoch": 1.1540609727717013, "grad_norm": 0.921246349811554, "learning_rate": 0.0005, "loss": 2.0624237060546875, "step": 9918 }, { "epoch": 1.1541773330230394, "grad_norm": 0.6956495642662048, "learning_rate": 0.0005, "loss": 1.152828574180603, "step": 9919 }, { "epoch": 1.1542936932743775, "grad_norm": 0.6190876960754395, "learning_rate": 0.0005, "loss": 1.5749421119689941, "step": 9920 }, { "epoch": 1.1544100535257156, "grad_norm": 0.9670872688293457, "learning_rate": 0.0005, "loss": 2.294179916381836, "step": 9921 }, { "epoch": 1.1545264137770537, "grad_norm": 0.7926026582717896, "learning_rate": 0.0005, "loss": 1.8004255294799805, "step": 9922 }, { "epoch": 1.154642774028392, "grad_norm": 1.1014256477355957, "learning_rate": 0.0005, "loss": 2.1380844116210938, "step": 9923 }, { "epoch": 1.15475913427973, "grad_norm": 0.9642512798309326, "learning_rate": 0.0005, "loss": 1.7777246236801147, "step": 9924 }, { "epoch": 1.1548754945310682, "grad_norm": 1.393215298652649, "learning_rate": 0.0005, "loss": 2.2335357666015625, "step": 9925 }, { "epoch": 1.1549918547824063, "grad_norm": 0.9461647868156433, "learning_rate": 0.0005, "loss": 1.7488672733306885, "step": 9926 }, { "epoch": 1.1551082150337444, "grad_norm": 1.1269882917404175, "learning_rate": 0.0005, "loss": 1.958705186843872, "step": 9927 }, { "epoch": 1.1552245752850827, "grad_norm": 1.2952426671981812, "learning_rate": 0.0005, "loss": 1.416235089302063, "step": 9928 }, { "epoch": 1.1553409355364208, "grad_norm": 1.5405796766281128, "learning_rate": 0.0005, "loss": 2.4629688262939453, "step": 9929 }, { "epoch": 1.155457295787759, "grad_norm": 0.9175857305526733, "learning_rate": 0.0005, "loss": 1.811286211013794, "step": 9930 }, { "epoch": 1.155573656039097, "grad_norm": 0.5258617997169495, "learning_rate": 0.0005, "loss": 1.2330613136291504, "step": 9931 }, { "epoch": 1.155690016290435, "grad_norm": 1.0508601665496826, "learning_rate": 0.0005, "loss": 1.6376726627349854, "step": 9932 }, { "epoch": 1.1558063765417734, "grad_norm": 0.7156040668487549, "learning_rate": 0.0005, "loss": 1.6885141134262085, "step": 9933 }, { "epoch": 1.1559227367931115, "grad_norm": 1.616471529006958, "learning_rate": 0.0005, "loss": 2.505185604095459, "step": 9934 }, { "epoch": 1.1560390970444496, "grad_norm": 0.8617608547210693, "learning_rate": 0.0005, "loss": 1.9168975353240967, "step": 9935 }, { "epoch": 1.1561554572957877, "grad_norm": 1.249693512916565, "learning_rate": 0.0005, "loss": 2.1498870849609375, "step": 9936 }, { "epoch": 1.1562718175471258, "grad_norm": 1.6213793754577637, "learning_rate": 0.0005, "loss": 2.2374002933502197, "step": 9937 }, { "epoch": 1.1563881777984641, "grad_norm": 1.1126629114151, "learning_rate": 0.0005, "loss": 2.2296841144561768, "step": 9938 }, { "epoch": 1.1565045380498022, "grad_norm": 0.6656932234764099, "learning_rate": 0.0005, "loss": 1.4386940002441406, "step": 9939 }, { "epoch": 1.1566208983011403, "grad_norm": 0.9229215979576111, "learning_rate": 0.0005, "loss": 1.564726710319519, "step": 9940 }, { "epoch": 1.1567372585524784, "grad_norm": 1.1179932355880737, "learning_rate": 0.0005, "loss": 2.096741199493408, "step": 9941 }, { "epoch": 1.1568536188038165, "grad_norm": 0.873839795589447, "learning_rate": 0.0005, "loss": 2.2447166442871094, "step": 9942 }, { "epoch": 1.1569699790551549, "grad_norm": 0.9504138231277466, "learning_rate": 0.0005, "loss": 2.6330714225769043, "step": 9943 }, { "epoch": 1.157086339306493, "grad_norm": 0.9235151410102844, "learning_rate": 0.0005, "loss": 2.3937063217163086, "step": 9944 }, { "epoch": 1.157202699557831, "grad_norm": 1.3128174543380737, "learning_rate": 0.0005, "loss": 1.6644362211227417, "step": 9945 }, { "epoch": 1.1573190598091692, "grad_norm": 1.7268847227096558, "learning_rate": 0.0005, "loss": 3.091139793395996, "step": 9946 }, { "epoch": 1.1574354200605073, "grad_norm": 0.9957143664360046, "learning_rate": 0.0005, "loss": 1.8107863664627075, "step": 9947 }, { "epoch": 1.1575517803118456, "grad_norm": 0.989453911781311, "learning_rate": 0.0005, "loss": 1.40719473361969, "step": 9948 }, { "epoch": 1.1576681405631837, "grad_norm": 1.2607502937316895, "learning_rate": 0.0005, "loss": 1.7661564350128174, "step": 9949 }, { "epoch": 1.1577845008145218, "grad_norm": 1.0499217510223389, "learning_rate": 0.0005, "loss": 2.060586929321289, "step": 9950 }, { "epoch": 1.1579008610658599, "grad_norm": 1.3734999895095825, "learning_rate": 0.0005, "loss": 2.51320743560791, "step": 9951 }, { "epoch": 1.158017221317198, "grad_norm": 1.8079586029052734, "learning_rate": 0.0005, "loss": 1.8219711780548096, "step": 9952 }, { "epoch": 1.1581335815685363, "grad_norm": 0.6285274028778076, "learning_rate": 0.0005, "loss": 1.0857908725738525, "step": 9953 }, { "epoch": 1.1582499418198744, "grad_norm": 1.6809700727462769, "learning_rate": 0.0005, "loss": 1.8699477910995483, "step": 9954 }, { "epoch": 1.1583663020712125, "grad_norm": 0.6491721868515015, "learning_rate": 0.0005, "loss": 1.3438154458999634, "step": 9955 }, { "epoch": 1.1584826623225506, "grad_norm": 1.1465657949447632, "learning_rate": 0.0005, "loss": 1.6422104835510254, "step": 9956 }, { "epoch": 1.1585990225738887, "grad_norm": 1.2853800058364868, "learning_rate": 0.0005, "loss": 2.0643553733825684, "step": 9957 }, { "epoch": 1.158715382825227, "grad_norm": 1.318521499633789, "learning_rate": 0.0005, "loss": 2.286546230316162, "step": 9958 }, { "epoch": 1.1588317430765651, "grad_norm": 1.0002834796905518, "learning_rate": 0.0005, "loss": 1.5519626140594482, "step": 9959 }, { "epoch": 1.1589481033279032, "grad_norm": 1.2960023880004883, "learning_rate": 0.0005, "loss": 1.259387493133545, "step": 9960 }, { "epoch": 1.1590644635792413, "grad_norm": 1.030916690826416, "learning_rate": 0.0005, "loss": 1.7648534774780273, "step": 9961 }, { "epoch": 1.1591808238305794, "grad_norm": 0.9925736784934998, "learning_rate": 0.0005, "loss": 2.0338869094848633, "step": 9962 }, { "epoch": 1.1592971840819177, "grad_norm": 1.3750653266906738, "learning_rate": 0.0005, "loss": 2.1466832160949707, "step": 9963 }, { "epoch": 1.1594135443332558, "grad_norm": 1.0470341444015503, "learning_rate": 0.0005, "loss": 1.672456979751587, "step": 9964 }, { "epoch": 1.159529904584594, "grad_norm": 1.625411033630371, "learning_rate": 0.0005, "loss": 1.8079605102539062, "step": 9965 }, { "epoch": 1.159646264835932, "grad_norm": 2.632993698120117, "learning_rate": 0.0005, "loss": 1.6847846508026123, "step": 9966 }, { "epoch": 1.1597626250872701, "grad_norm": 1.2595887184143066, "learning_rate": 0.0005, "loss": 2.422739028930664, "step": 9967 }, { "epoch": 1.1598789853386084, "grad_norm": 1.1207211017608643, "learning_rate": 0.0005, "loss": 2.7561471462249756, "step": 9968 }, { "epoch": 1.1599953455899465, "grad_norm": 1.0967382192611694, "learning_rate": 0.0005, "loss": 1.755763053894043, "step": 9969 }, { "epoch": 1.1601117058412846, "grad_norm": 0.8465824723243713, "learning_rate": 0.0005, "loss": 1.3870731592178345, "step": 9970 }, { "epoch": 1.1602280660926227, "grad_norm": 1.647214651107788, "learning_rate": 0.0005, "loss": 2.0047993659973145, "step": 9971 }, { "epoch": 1.1603444263439608, "grad_norm": 1.0797380208969116, "learning_rate": 0.0005, "loss": 2.1478347778320312, "step": 9972 }, { "epoch": 1.1604607865952992, "grad_norm": 0.9592166543006897, "learning_rate": 0.0005, "loss": 2.127328395843506, "step": 9973 }, { "epoch": 1.1605771468466373, "grad_norm": 2.6981394290924072, "learning_rate": 0.0005, "loss": 1.949255347251892, "step": 9974 }, { "epoch": 1.1606935070979754, "grad_norm": 1.340811848640442, "learning_rate": 0.0005, "loss": 2.248929500579834, "step": 9975 }, { "epoch": 1.1608098673493135, "grad_norm": 1.5999064445495605, "learning_rate": 0.0005, "loss": 1.6782113313674927, "step": 9976 }, { "epoch": 1.1609262276006516, "grad_norm": 1.2260380983352661, "learning_rate": 0.0005, "loss": 2.484144687652588, "step": 9977 }, { "epoch": 1.1610425878519899, "grad_norm": 0.9808769822120667, "learning_rate": 0.0005, "loss": 1.350714921951294, "step": 9978 }, { "epoch": 1.161158948103328, "grad_norm": 1.0997624397277832, "learning_rate": 0.0005, "loss": 1.7900639772415161, "step": 9979 }, { "epoch": 1.161275308354666, "grad_norm": 1.5289218425750732, "learning_rate": 0.0005, "loss": 2.1387834548950195, "step": 9980 }, { "epoch": 1.1613916686060042, "grad_norm": 1.2859218120574951, "learning_rate": 0.0005, "loss": 2.030892848968506, "step": 9981 }, { "epoch": 1.1615080288573423, "grad_norm": 0.961258590221405, "learning_rate": 0.0005, "loss": 2.0726122856140137, "step": 9982 }, { "epoch": 1.1616243891086806, "grad_norm": 0.8155167698860168, "learning_rate": 0.0005, "loss": 1.709172248840332, "step": 9983 }, { "epoch": 1.1617407493600187, "grad_norm": 1.0254820585250854, "learning_rate": 0.0005, "loss": 2.009746551513672, "step": 9984 }, { "epoch": 1.1618571096113568, "grad_norm": 1.1207512617111206, "learning_rate": 0.0005, "loss": 1.4166487455368042, "step": 9985 }, { "epoch": 1.161973469862695, "grad_norm": 0.6127386093139648, "learning_rate": 0.0005, "loss": 1.2050737142562866, "step": 9986 }, { "epoch": 1.162089830114033, "grad_norm": 1.1455349922180176, "learning_rate": 0.0005, "loss": 2.216684579849243, "step": 9987 }, { "epoch": 1.162206190365371, "grad_norm": 1.2960073947906494, "learning_rate": 0.0005, "loss": 2.5021324157714844, "step": 9988 }, { "epoch": 1.1623225506167094, "grad_norm": 0.9101362824440002, "learning_rate": 0.0005, "loss": 1.663290023803711, "step": 9989 }, { "epoch": 1.1624389108680475, "grad_norm": 0.8798077702522278, "learning_rate": 0.0005, "loss": 1.9808334112167358, "step": 9990 }, { "epoch": 1.1625552711193856, "grad_norm": 2.472926378250122, "learning_rate": 0.0005, "loss": 2.3908331394195557, "step": 9991 }, { "epoch": 1.1626716313707237, "grad_norm": 1.1692886352539062, "learning_rate": 0.0005, "loss": 1.8593034744262695, "step": 9992 }, { "epoch": 1.1627879916220618, "grad_norm": 1.3573857545852661, "learning_rate": 0.0005, "loss": 1.4820656776428223, "step": 9993 }, { "epoch": 1.1629043518734001, "grad_norm": 0.7647511959075928, "learning_rate": 0.0005, "loss": 1.6567065715789795, "step": 9994 }, { "epoch": 1.1630207121247382, "grad_norm": 1.7413220405578613, "learning_rate": 0.0005, "loss": 2.241612434387207, "step": 9995 }, { "epoch": 1.1631370723760763, "grad_norm": 0.8036602139472961, "learning_rate": 0.0005, "loss": 1.8106086254119873, "step": 9996 }, { "epoch": 1.1632534326274144, "grad_norm": 0.9402632117271423, "learning_rate": 0.0005, "loss": 1.4670727252960205, "step": 9997 }, { "epoch": 1.1633697928787525, "grad_norm": 1.3700315952301025, "learning_rate": 0.0005, "loss": 2.5103657245635986, "step": 9998 }, { "epoch": 1.1634861531300908, "grad_norm": 1.63005793094635, "learning_rate": 0.0005, "loss": 2.0280256271362305, "step": 9999 }, { "epoch": 1.163602513381429, "grad_norm": 1.2929631471633911, "learning_rate": 0.0005, "loss": 1.9074574708938599, "step": 10000 }, { "epoch": 1.163718873632767, "grad_norm": 1.087392807006836, "learning_rate": 0.0005, "loss": 1.824012279510498, "step": 10001 }, { "epoch": 1.1638352338841051, "grad_norm": 1.0648138523101807, "learning_rate": 0.0005, "loss": 1.737447738647461, "step": 10002 }, { "epoch": 1.1639515941354432, "grad_norm": 1.3043230772018433, "learning_rate": 0.0005, "loss": 2.254399299621582, "step": 10003 }, { "epoch": 1.1640679543867816, "grad_norm": 1.0197432041168213, "learning_rate": 0.0005, "loss": 1.8758313655853271, "step": 10004 }, { "epoch": 1.1641843146381197, "grad_norm": 1.3057407140731812, "learning_rate": 0.0005, "loss": 2.8432059288024902, "step": 10005 }, { "epoch": 1.1643006748894578, "grad_norm": 0.9566653966903687, "learning_rate": 0.0005, "loss": 1.370056390762329, "step": 10006 }, { "epoch": 1.1644170351407959, "grad_norm": 0.9842144846916199, "learning_rate": 0.0005, "loss": 2.1025166511535645, "step": 10007 }, { "epoch": 1.164533395392134, "grad_norm": 1.6047874689102173, "learning_rate": 0.0005, "loss": 1.7875362634658813, "step": 10008 }, { "epoch": 1.1646497556434723, "grad_norm": 1.158656358718872, "learning_rate": 0.0005, "loss": 1.8411033153533936, "step": 10009 }, { "epoch": 1.1647661158948104, "grad_norm": 0.8119416236877441, "learning_rate": 0.0005, "loss": 1.589455246925354, "step": 10010 }, { "epoch": 1.1648824761461485, "grad_norm": 1.2955440282821655, "learning_rate": 0.0005, "loss": 2.4313831329345703, "step": 10011 }, { "epoch": 1.1649988363974866, "grad_norm": 1.1797475814819336, "learning_rate": 0.0005, "loss": 2.900116205215454, "step": 10012 }, { "epoch": 1.1651151966488247, "grad_norm": 1.3567657470703125, "learning_rate": 0.0005, "loss": 2.075202465057373, "step": 10013 }, { "epoch": 1.165231556900163, "grad_norm": 0.7464427351951599, "learning_rate": 0.0005, "loss": 1.6567611694335938, "step": 10014 }, { "epoch": 1.165347917151501, "grad_norm": 11.55996322631836, "learning_rate": 0.0005, "loss": 1.5918686389923096, "step": 10015 }, { "epoch": 1.1654642774028392, "grad_norm": 1.1003870964050293, "learning_rate": 0.0005, "loss": 1.3709644079208374, "step": 10016 }, { "epoch": 1.1655806376541773, "grad_norm": 0.870420515537262, "learning_rate": 0.0005, "loss": 1.5193195343017578, "step": 10017 }, { "epoch": 1.1656969979055154, "grad_norm": 1.3583511114120483, "learning_rate": 0.0005, "loss": 1.9738264083862305, "step": 10018 }, { "epoch": 1.1658133581568535, "grad_norm": 1.0216912031173706, "learning_rate": 0.0005, "loss": 1.4219175577163696, "step": 10019 }, { "epoch": 1.1659297184081918, "grad_norm": 1.1969218254089355, "learning_rate": 0.0005, "loss": 2.293391227722168, "step": 10020 }, { "epoch": 1.16604607865953, "grad_norm": 0.7055753469467163, "learning_rate": 0.0005, "loss": 1.2960926294326782, "step": 10021 }, { "epoch": 1.166162438910868, "grad_norm": 1.0547945499420166, "learning_rate": 0.0005, "loss": 2.013195753097534, "step": 10022 }, { "epoch": 1.1662787991622061, "grad_norm": 1.4668190479278564, "learning_rate": 0.0005, "loss": 1.532376766204834, "step": 10023 }, { "epoch": 1.1663951594135442, "grad_norm": 2.3368098735809326, "learning_rate": 0.0005, "loss": 1.7192245721817017, "step": 10024 }, { "epoch": 1.1665115196648825, "grad_norm": 1.053276538848877, "learning_rate": 0.0005, "loss": 1.7288004159927368, "step": 10025 }, { "epoch": 1.1666278799162206, "grad_norm": 1.1081204414367676, "learning_rate": 0.0005, "loss": 1.5547029972076416, "step": 10026 }, { "epoch": 1.1667442401675587, "grad_norm": 1.12199866771698, "learning_rate": 0.0005, "loss": 1.9045922756195068, "step": 10027 }, { "epoch": 1.1668606004188968, "grad_norm": 0.966611385345459, "learning_rate": 0.0005, "loss": 2.3404598236083984, "step": 10028 }, { "epoch": 1.166976960670235, "grad_norm": 1.124335765838623, "learning_rate": 0.0005, "loss": 1.6809229850769043, "step": 10029 }, { "epoch": 1.1670933209215733, "grad_norm": 1.214233160018921, "learning_rate": 0.0005, "loss": 2.2741446495056152, "step": 10030 }, { "epoch": 1.1672096811729114, "grad_norm": 1.4190129041671753, "learning_rate": 0.0005, "loss": 2.0731234550476074, "step": 10031 }, { "epoch": 1.1673260414242494, "grad_norm": 1.7698277235031128, "learning_rate": 0.0005, "loss": 2.0095419883728027, "step": 10032 }, { "epoch": 1.1674424016755875, "grad_norm": 1.2537986040115356, "learning_rate": 0.0005, "loss": 2.6412744522094727, "step": 10033 }, { "epoch": 1.1675587619269256, "grad_norm": 1.2340925931930542, "learning_rate": 0.0005, "loss": 3.470844268798828, "step": 10034 }, { "epoch": 1.167675122178264, "grad_norm": 1.6709985733032227, "learning_rate": 0.0005, "loss": 1.8580207824707031, "step": 10035 }, { "epoch": 1.167791482429602, "grad_norm": 0.8760534524917603, "learning_rate": 0.0005, "loss": 2.1386775970458984, "step": 10036 }, { "epoch": 1.1679078426809402, "grad_norm": 0.4241819977760315, "learning_rate": 0.0005, "loss": 0.9306986927986145, "step": 10037 }, { "epoch": 1.1680242029322783, "grad_norm": 0.7903595566749573, "learning_rate": 0.0005, "loss": 1.7562220096588135, "step": 10038 }, { "epoch": 1.1681405631836164, "grad_norm": 0.8513551950454712, "learning_rate": 0.0005, "loss": 1.8723170757293701, "step": 10039 }, { "epoch": 1.1682569234349547, "grad_norm": 1.0944337844848633, "learning_rate": 0.0005, "loss": 2.1120641231536865, "step": 10040 }, { "epoch": 1.1683732836862928, "grad_norm": 1.0342031717300415, "learning_rate": 0.0005, "loss": 2.0439321994781494, "step": 10041 }, { "epoch": 1.1684896439376309, "grad_norm": 1.1314038038253784, "learning_rate": 0.0005, "loss": 1.728562831878662, "step": 10042 }, { "epoch": 1.168606004188969, "grad_norm": 2.571902275085449, "learning_rate": 0.0005, "loss": 2.0996623039245605, "step": 10043 }, { "epoch": 1.168722364440307, "grad_norm": 1.3775882720947266, "learning_rate": 0.0005, "loss": 2.1960182189941406, "step": 10044 }, { "epoch": 1.1688387246916454, "grad_norm": 1.5146971940994263, "learning_rate": 0.0005, "loss": 2.401437997817993, "step": 10045 }, { "epoch": 1.1689550849429835, "grad_norm": 0.9231436252593994, "learning_rate": 0.0005, "loss": 1.7647852897644043, "step": 10046 }, { "epoch": 1.1690714451943216, "grad_norm": 1.2774982452392578, "learning_rate": 0.0005, "loss": 2.661731243133545, "step": 10047 }, { "epoch": 1.1691878054456597, "grad_norm": 0.5279671549797058, "learning_rate": 0.0005, "loss": 1.4997036457061768, "step": 10048 }, { "epoch": 1.1693041656969978, "grad_norm": 0.9532074332237244, "learning_rate": 0.0005, "loss": 2.123168468475342, "step": 10049 }, { "epoch": 1.1694205259483361, "grad_norm": 1.4092732667922974, "learning_rate": 0.0005, "loss": 1.911286473274231, "step": 10050 }, { "epoch": 1.1695368861996742, "grad_norm": 1.5982320308685303, "learning_rate": 0.0005, "loss": 1.817012071609497, "step": 10051 }, { "epoch": 1.1696532464510123, "grad_norm": 1.337907075881958, "learning_rate": 0.0005, "loss": 2.1859514713287354, "step": 10052 }, { "epoch": 1.1697696067023504, "grad_norm": 0.9427022933959961, "learning_rate": 0.0005, "loss": 2.1032426357269287, "step": 10053 }, { "epoch": 1.1698859669536885, "grad_norm": 2.0304791927337646, "learning_rate": 0.0005, "loss": 2.241022825241089, "step": 10054 }, { "epoch": 1.1700023272050268, "grad_norm": 1.3724370002746582, "learning_rate": 0.0005, "loss": 2.4613265991210938, "step": 10055 }, { "epoch": 1.170118687456365, "grad_norm": 2.1299164295196533, "learning_rate": 0.0005, "loss": 1.393695592880249, "step": 10056 }, { "epoch": 1.170235047707703, "grad_norm": 1.0399763584136963, "learning_rate": 0.0005, "loss": 1.7769861221313477, "step": 10057 }, { "epoch": 1.1703514079590411, "grad_norm": 1.2456146478652954, "learning_rate": 0.0005, "loss": 1.6789922714233398, "step": 10058 }, { "epoch": 1.1704677682103792, "grad_norm": 1.3321576118469238, "learning_rate": 0.0005, "loss": 2.6390483379364014, "step": 10059 }, { "epoch": 1.1705841284617176, "grad_norm": 1.1936299800872803, "learning_rate": 0.0005, "loss": 2.1035990715026855, "step": 10060 }, { "epoch": 1.1707004887130557, "grad_norm": 0.755016565322876, "learning_rate": 0.0005, "loss": 1.2418198585510254, "step": 10061 }, { "epoch": 1.1708168489643938, "grad_norm": 0.9636359214782715, "learning_rate": 0.0005, "loss": 1.9731014966964722, "step": 10062 }, { "epoch": 1.1709332092157319, "grad_norm": 0.955162525177002, "learning_rate": 0.0005, "loss": 2.1477208137512207, "step": 10063 }, { "epoch": 1.17104956946707, "grad_norm": 0.8859733939170837, "learning_rate": 0.0005, "loss": 2.0500335693359375, "step": 10064 }, { "epoch": 1.1711659297184083, "grad_norm": 1.268118977546692, "learning_rate": 0.0005, "loss": 2.901817798614502, "step": 10065 }, { "epoch": 1.1712822899697464, "grad_norm": 0.7049152255058289, "learning_rate": 0.0005, "loss": 2.01263427734375, "step": 10066 }, { "epoch": 1.1713986502210845, "grad_norm": 1.3640166521072388, "learning_rate": 0.0005, "loss": 1.796898603439331, "step": 10067 }, { "epoch": 1.1715150104724226, "grad_norm": 1.125411868095398, "learning_rate": 0.0005, "loss": 1.3836112022399902, "step": 10068 }, { "epoch": 1.1716313707237607, "grad_norm": 1.2337489128112793, "learning_rate": 0.0005, "loss": 2.1934289932250977, "step": 10069 }, { "epoch": 1.171747730975099, "grad_norm": 1.2071211338043213, "learning_rate": 0.0005, "loss": 2.469076633453369, "step": 10070 }, { "epoch": 1.171864091226437, "grad_norm": 1.08464777469635, "learning_rate": 0.0005, "loss": 1.8296597003936768, "step": 10071 }, { "epoch": 1.1719804514777752, "grad_norm": 1.343115210533142, "learning_rate": 0.0005, "loss": 2.210921049118042, "step": 10072 }, { "epoch": 1.1720968117291133, "grad_norm": 1.2192180156707764, "learning_rate": 0.0005, "loss": 1.7950743436813354, "step": 10073 }, { "epoch": 1.1722131719804514, "grad_norm": 0.9897035956382751, "learning_rate": 0.0005, "loss": 1.9600679874420166, "step": 10074 }, { "epoch": 1.1723295322317897, "grad_norm": 0.8127139210700989, "learning_rate": 0.0005, "loss": 1.714432716369629, "step": 10075 }, { "epoch": 1.1724458924831278, "grad_norm": 1.0828298330307007, "learning_rate": 0.0005, "loss": 2.101489543914795, "step": 10076 }, { "epoch": 1.172562252734466, "grad_norm": 0.660459578037262, "learning_rate": 0.0005, "loss": 1.6327648162841797, "step": 10077 }, { "epoch": 1.172678612985804, "grad_norm": 1.5027599334716797, "learning_rate": 0.0005, "loss": 2.5630741119384766, "step": 10078 }, { "epoch": 1.172794973237142, "grad_norm": 5.521785736083984, "learning_rate": 0.0005, "loss": 1.9876959323883057, "step": 10079 }, { "epoch": 1.1729113334884804, "grad_norm": 1.2794551849365234, "learning_rate": 0.0005, "loss": 1.9214781522750854, "step": 10080 }, { "epoch": 1.1730276937398185, "grad_norm": 1.0589439868927002, "learning_rate": 0.0005, "loss": 2.0748019218444824, "step": 10081 }, { "epoch": 1.1731440539911566, "grad_norm": 1.1308470964431763, "learning_rate": 0.0005, "loss": 2.550112247467041, "step": 10082 }, { "epoch": 1.1732604142424947, "grad_norm": 1.5458256006240845, "learning_rate": 0.0005, "loss": 2.553222179412842, "step": 10083 }, { "epoch": 1.1733767744938328, "grad_norm": 1.1355130672454834, "learning_rate": 0.0005, "loss": 2.099569797515869, "step": 10084 }, { "epoch": 1.1734931347451711, "grad_norm": 1.0726587772369385, "learning_rate": 0.0005, "loss": 1.9596662521362305, "step": 10085 }, { "epoch": 1.1736094949965092, "grad_norm": 1.3546092510223389, "learning_rate": 0.0005, "loss": 2.6076412200927734, "step": 10086 }, { "epoch": 1.1737258552478473, "grad_norm": 1.474587082862854, "learning_rate": 0.0005, "loss": 1.9548949003219604, "step": 10087 }, { "epoch": 1.1738422154991854, "grad_norm": 1.2180789709091187, "learning_rate": 0.0005, "loss": 1.6339218616485596, "step": 10088 }, { "epoch": 1.1739585757505235, "grad_norm": 0.6213049292564392, "learning_rate": 0.0005, "loss": 1.3344072103500366, "step": 10089 }, { "epoch": 1.1740749360018619, "grad_norm": 1.0286709070205688, "learning_rate": 0.0005, "loss": 1.7970366477966309, "step": 10090 }, { "epoch": 1.1741912962532, "grad_norm": 1.07451331615448, "learning_rate": 0.0005, "loss": 2.2952497005462646, "step": 10091 }, { "epoch": 1.174307656504538, "grad_norm": 0.7669099569320679, "learning_rate": 0.0005, "loss": 1.5954301357269287, "step": 10092 }, { "epoch": 1.1744240167558762, "grad_norm": 1.163613200187683, "learning_rate": 0.0005, "loss": 2.3807826042175293, "step": 10093 }, { "epoch": 1.1745403770072143, "grad_norm": 0.7647632360458374, "learning_rate": 0.0005, "loss": 2.1313915252685547, "step": 10094 }, { "epoch": 1.1746567372585526, "grad_norm": 0.9942143559455872, "learning_rate": 0.0005, "loss": 1.7838776111602783, "step": 10095 }, { "epoch": 1.1747730975098907, "grad_norm": 1.3287585973739624, "learning_rate": 0.0005, "loss": 2.7306506633758545, "step": 10096 }, { "epoch": 1.1748894577612288, "grad_norm": 1.8463537693023682, "learning_rate": 0.0005, "loss": 2.91581392288208, "step": 10097 }, { "epoch": 1.1750058180125669, "grad_norm": 0.7633119821548462, "learning_rate": 0.0005, "loss": 2.1875290870666504, "step": 10098 }, { "epoch": 1.175122178263905, "grad_norm": 0.8912979364395142, "learning_rate": 0.0005, "loss": 1.2106820344924927, "step": 10099 }, { "epoch": 1.1752385385152433, "grad_norm": 1.1307170391082764, "learning_rate": 0.0005, "loss": 1.8987244367599487, "step": 10100 }, { "epoch": 1.1753548987665814, "grad_norm": 1.6735897064208984, "learning_rate": 0.0005, "loss": 2.5183005332946777, "step": 10101 }, { "epoch": 1.1754712590179195, "grad_norm": 1.6953867673873901, "learning_rate": 0.0005, "loss": 2.024041175842285, "step": 10102 }, { "epoch": 1.1755876192692576, "grad_norm": 1.436585545539856, "learning_rate": 0.0005, "loss": 1.8008089065551758, "step": 10103 }, { "epoch": 1.1757039795205957, "grad_norm": 1.3288192749023438, "learning_rate": 0.0005, "loss": 2.710808753967285, "step": 10104 }, { "epoch": 1.175820339771934, "grad_norm": 2.216810703277588, "learning_rate": 0.0005, "loss": 2.5499484539031982, "step": 10105 }, { "epoch": 1.175936700023272, "grad_norm": 2.5502774715423584, "learning_rate": 0.0005, "loss": 2.084836721420288, "step": 10106 }, { "epoch": 1.1760530602746102, "grad_norm": 0.988595724105835, "learning_rate": 0.0005, "loss": 2.225438117980957, "step": 10107 }, { "epoch": 1.1761694205259483, "grad_norm": 1.0562243461608887, "learning_rate": 0.0005, "loss": 2.0951602458953857, "step": 10108 }, { "epoch": 1.1762857807772864, "grad_norm": 1.0127589702606201, "learning_rate": 0.0005, "loss": 1.9077434539794922, "step": 10109 }, { "epoch": 1.1764021410286247, "grad_norm": 1.0410854816436768, "learning_rate": 0.0005, "loss": 2.126594305038452, "step": 10110 }, { "epoch": 1.1765185012799628, "grad_norm": 1.184259057044983, "learning_rate": 0.0005, "loss": 2.0759127140045166, "step": 10111 }, { "epoch": 1.176634861531301, "grad_norm": 1.486992359161377, "learning_rate": 0.0005, "loss": 2.2036306858062744, "step": 10112 }, { "epoch": 1.176751221782639, "grad_norm": 1.3469700813293457, "learning_rate": 0.0005, "loss": 2.3481240272521973, "step": 10113 }, { "epoch": 1.1768675820339771, "grad_norm": 1.4133490324020386, "learning_rate": 0.0005, "loss": 1.6582961082458496, "step": 10114 }, { "epoch": 1.1769839422853154, "grad_norm": 0.9900507926940918, "learning_rate": 0.0005, "loss": 1.620147705078125, "step": 10115 }, { "epoch": 1.1771003025366535, "grad_norm": 0.9584294557571411, "learning_rate": 0.0005, "loss": 1.9522604942321777, "step": 10116 }, { "epoch": 1.1772166627879916, "grad_norm": 1.245924711227417, "learning_rate": 0.0005, "loss": 2.477231025695801, "step": 10117 }, { "epoch": 1.1773330230393297, "grad_norm": 0.8643444776535034, "learning_rate": 0.0005, "loss": 2.7499499320983887, "step": 10118 }, { "epoch": 1.1774493832906678, "grad_norm": 1.2205687761306763, "learning_rate": 0.0005, "loss": 2.0031230449676514, "step": 10119 }, { "epoch": 1.1775657435420062, "grad_norm": 1.1862363815307617, "learning_rate": 0.0005, "loss": 2.507082939147949, "step": 10120 }, { "epoch": 1.1776821037933443, "grad_norm": 0.7782881259918213, "learning_rate": 0.0005, "loss": 1.4821128845214844, "step": 10121 }, { "epoch": 1.1777984640446824, "grad_norm": 1.0823867321014404, "learning_rate": 0.0005, "loss": 1.4199880361557007, "step": 10122 }, { "epoch": 1.1779148242960205, "grad_norm": 0.7970598340034485, "learning_rate": 0.0005, "loss": 1.0686028003692627, "step": 10123 }, { "epoch": 1.1780311845473586, "grad_norm": 1.1242724657058716, "learning_rate": 0.0005, "loss": 2.278174638748169, "step": 10124 }, { "epoch": 1.1781475447986969, "grad_norm": 1.3821858167648315, "learning_rate": 0.0005, "loss": 1.819665551185608, "step": 10125 }, { "epoch": 1.178263905050035, "grad_norm": 1.0453697443008423, "learning_rate": 0.0005, "loss": 1.7742117643356323, "step": 10126 }, { "epoch": 1.178380265301373, "grad_norm": 1.169540286064148, "learning_rate": 0.0005, "loss": 1.8713594675064087, "step": 10127 }, { "epoch": 1.1784966255527112, "grad_norm": 1.5227895975112915, "learning_rate": 0.0005, "loss": 1.866529941558838, "step": 10128 }, { "epoch": 1.1786129858040493, "grad_norm": 0.5404368042945862, "learning_rate": 0.0005, "loss": 0.9854614734649658, "step": 10129 }, { "epoch": 1.1787293460553876, "grad_norm": 1.6172982454299927, "learning_rate": 0.0005, "loss": 2.6782331466674805, "step": 10130 }, { "epoch": 1.1788457063067257, "grad_norm": 1.0094722509384155, "learning_rate": 0.0005, "loss": 2.032352924346924, "step": 10131 }, { "epoch": 1.1789620665580638, "grad_norm": 1.4109081029891968, "learning_rate": 0.0005, "loss": 2.910095691680908, "step": 10132 }, { "epoch": 1.179078426809402, "grad_norm": 1.1600803136825562, "learning_rate": 0.0005, "loss": 1.7627758979797363, "step": 10133 }, { "epoch": 1.17919478706074, "grad_norm": 1.1682120561599731, "learning_rate": 0.0005, "loss": 1.7206166982650757, "step": 10134 }, { "epoch": 1.1793111473120783, "grad_norm": 0.5669963359832764, "learning_rate": 0.0005, "loss": 1.2508604526519775, "step": 10135 }, { "epoch": 1.1794275075634164, "grad_norm": 1.6322482824325562, "learning_rate": 0.0005, "loss": 2.2777469158172607, "step": 10136 }, { "epoch": 1.1795438678147545, "grad_norm": 1.1195639371871948, "learning_rate": 0.0005, "loss": 1.5507793426513672, "step": 10137 }, { "epoch": 1.1796602280660926, "grad_norm": 2.135307550430298, "learning_rate": 0.0005, "loss": 1.7711485624313354, "step": 10138 }, { "epoch": 1.1797765883174307, "grad_norm": 1.3480709791183472, "learning_rate": 0.0005, "loss": 1.9096623659133911, "step": 10139 }, { "epoch": 1.179892948568769, "grad_norm": 1.1772611141204834, "learning_rate": 0.0005, "loss": 2.2667651176452637, "step": 10140 }, { "epoch": 1.1800093088201071, "grad_norm": 1.0932339429855347, "learning_rate": 0.0005, "loss": 2.5364291667938232, "step": 10141 }, { "epoch": 1.1801256690714452, "grad_norm": 0.8939746022224426, "learning_rate": 0.0005, "loss": 1.9480576515197754, "step": 10142 }, { "epoch": 1.1802420293227833, "grad_norm": 1.5781508684158325, "learning_rate": 0.0005, "loss": 1.684323787689209, "step": 10143 }, { "epoch": 1.1803583895741214, "grad_norm": 1.2645045518875122, "learning_rate": 0.0005, "loss": 1.8885881900787354, "step": 10144 }, { "epoch": 1.1804747498254597, "grad_norm": 1.4182337522506714, "learning_rate": 0.0005, "loss": 2.2873783111572266, "step": 10145 }, { "epoch": 1.1805911100767978, "grad_norm": 0.7250911593437195, "learning_rate": 0.0005, "loss": 1.730963110923767, "step": 10146 }, { "epoch": 1.180707470328136, "grad_norm": 1.4904450178146362, "learning_rate": 0.0005, "loss": 2.463172674179077, "step": 10147 }, { "epoch": 1.180823830579474, "grad_norm": 1.2443045377731323, "learning_rate": 0.0005, "loss": 1.861814260482788, "step": 10148 }, { "epoch": 1.1809401908308121, "grad_norm": 2.0588200092315674, "learning_rate": 0.0005, "loss": 2.0180277824401855, "step": 10149 }, { "epoch": 1.1810565510821502, "grad_norm": 1.2827796936035156, "learning_rate": 0.0005, "loss": 1.902746319770813, "step": 10150 }, { "epoch": 1.1811729113334886, "grad_norm": 0.9820755124092102, "learning_rate": 0.0005, "loss": 1.4686486721038818, "step": 10151 }, { "epoch": 1.1812892715848267, "grad_norm": 1.4158556461334229, "learning_rate": 0.0005, "loss": 2.0525755882263184, "step": 10152 }, { "epoch": 1.1814056318361648, "grad_norm": 0.9518614411354065, "learning_rate": 0.0005, "loss": 1.8450058698654175, "step": 10153 }, { "epoch": 1.1815219920875029, "grad_norm": 1.2269476652145386, "learning_rate": 0.0005, "loss": 2.2542998790740967, "step": 10154 }, { "epoch": 1.181638352338841, "grad_norm": 3.1338350772857666, "learning_rate": 0.0005, "loss": 2.143050193786621, "step": 10155 }, { "epoch": 1.1817547125901793, "grad_norm": 1.2468904256820679, "learning_rate": 0.0005, "loss": 2.0317039489746094, "step": 10156 }, { "epoch": 1.1818710728415174, "grad_norm": 1.083358645439148, "learning_rate": 0.0005, "loss": 2.1470441818237305, "step": 10157 }, { "epoch": 1.1819874330928555, "grad_norm": 1.1941637992858887, "learning_rate": 0.0005, "loss": 1.977297306060791, "step": 10158 }, { "epoch": 1.1821037933441936, "grad_norm": 1.4215017557144165, "learning_rate": 0.0005, "loss": 2.6959800720214844, "step": 10159 }, { "epoch": 1.1822201535955317, "grad_norm": 0.9994685053825378, "learning_rate": 0.0005, "loss": 2.1762900352478027, "step": 10160 }, { "epoch": 1.18233651384687, "grad_norm": 1.3048064708709717, "learning_rate": 0.0005, "loss": 1.6264139413833618, "step": 10161 }, { "epoch": 1.182452874098208, "grad_norm": 1.312304139137268, "learning_rate": 0.0005, "loss": 2.049431085586548, "step": 10162 }, { "epoch": 1.1825692343495462, "grad_norm": 1.1969432830810547, "learning_rate": 0.0005, "loss": 2.0934700965881348, "step": 10163 }, { "epoch": 1.1826855946008843, "grad_norm": 1.6441800594329834, "learning_rate": 0.0005, "loss": 2.622451066970825, "step": 10164 }, { "epoch": 1.1828019548522224, "grad_norm": 1.0459765195846558, "learning_rate": 0.0005, "loss": 1.847062587738037, "step": 10165 }, { "epoch": 1.1829183151035607, "grad_norm": 0.5784925818443298, "learning_rate": 0.0005, "loss": 1.0653358697891235, "step": 10166 }, { "epoch": 1.1830346753548988, "grad_norm": 2.975794792175293, "learning_rate": 0.0005, "loss": 1.756762146949768, "step": 10167 }, { "epoch": 1.183151035606237, "grad_norm": 1.9479823112487793, "learning_rate": 0.0005, "loss": 2.181816577911377, "step": 10168 }, { "epoch": 1.183267395857575, "grad_norm": 1.3760977983474731, "learning_rate": 0.0005, "loss": 2.2387924194335938, "step": 10169 }, { "epoch": 1.183383756108913, "grad_norm": 1.2636698484420776, "learning_rate": 0.0005, "loss": 2.3461861610412598, "step": 10170 }, { "epoch": 1.1835001163602514, "grad_norm": 1.7522019147872925, "learning_rate": 0.0005, "loss": 2.081468105316162, "step": 10171 }, { "epoch": 1.1836164766115895, "grad_norm": 0.8650615215301514, "learning_rate": 0.0005, "loss": 1.9660753011703491, "step": 10172 }, { "epoch": 1.1837328368629276, "grad_norm": 1.0509686470031738, "learning_rate": 0.0005, "loss": 1.7577272653579712, "step": 10173 }, { "epoch": 1.1838491971142657, "grad_norm": 0.8445509076118469, "learning_rate": 0.0005, "loss": 1.3997752666473389, "step": 10174 }, { "epoch": 1.1839655573656038, "grad_norm": 1.1906241178512573, "learning_rate": 0.0005, "loss": 2.34843111038208, "step": 10175 }, { "epoch": 1.1840819176169421, "grad_norm": 1.3548251390457153, "learning_rate": 0.0005, "loss": 2.5600218772888184, "step": 10176 }, { "epoch": 1.1841982778682802, "grad_norm": 1.2648729085922241, "learning_rate": 0.0005, "loss": 2.5123066902160645, "step": 10177 }, { "epoch": 1.1843146381196183, "grad_norm": 1.4853416681289673, "learning_rate": 0.0005, "loss": 1.911637306213379, "step": 10178 }, { "epoch": 1.1844309983709564, "grad_norm": 1.1376234292984009, "learning_rate": 0.0005, "loss": 1.7578611373901367, "step": 10179 }, { "epoch": 1.1845473586222945, "grad_norm": 0.9223757982254028, "learning_rate": 0.0005, "loss": 1.8916314840316772, "step": 10180 }, { "epoch": 1.1846637188736326, "grad_norm": 0.9424601793289185, "learning_rate": 0.0005, "loss": 2.2139949798583984, "step": 10181 }, { "epoch": 1.184780079124971, "grad_norm": 0.932019054889679, "learning_rate": 0.0005, "loss": 1.3337211608886719, "step": 10182 }, { "epoch": 1.184896439376309, "grad_norm": 1.7611513137817383, "learning_rate": 0.0005, "loss": 1.8702107667922974, "step": 10183 }, { "epoch": 1.1850127996276472, "grad_norm": 0.831818699836731, "learning_rate": 0.0005, "loss": 2.038856267929077, "step": 10184 }, { "epoch": 1.1851291598789853, "grad_norm": 0.8171558976173401, "learning_rate": 0.0005, "loss": 1.709890604019165, "step": 10185 }, { "epoch": 1.1852455201303234, "grad_norm": 1.6951218843460083, "learning_rate": 0.0005, "loss": 2.4270973205566406, "step": 10186 }, { "epoch": 1.1853618803816617, "grad_norm": 1.3429253101348877, "learning_rate": 0.0005, "loss": 1.3934657573699951, "step": 10187 }, { "epoch": 1.1854782406329998, "grad_norm": 0.9380197525024414, "learning_rate": 0.0005, "loss": 2.6542460918426514, "step": 10188 }, { "epoch": 1.1855946008843379, "grad_norm": 0.8312566876411438, "learning_rate": 0.0005, "loss": 1.6017651557922363, "step": 10189 }, { "epoch": 1.185710961135676, "grad_norm": 1.3956457376480103, "learning_rate": 0.0005, "loss": 1.6493475437164307, "step": 10190 }, { "epoch": 1.185827321387014, "grad_norm": 1.03964102268219, "learning_rate": 0.0005, "loss": 1.8655989170074463, "step": 10191 }, { "epoch": 1.1859436816383524, "grad_norm": 1.4259424209594727, "learning_rate": 0.0005, "loss": 1.456343173980713, "step": 10192 }, { "epoch": 1.1860600418896905, "grad_norm": 1.0810356140136719, "learning_rate": 0.0005, "loss": 1.7311238050460815, "step": 10193 }, { "epoch": 1.1861764021410286, "grad_norm": 0.7431973814964294, "learning_rate": 0.0005, "loss": 1.2358667850494385, "step": 10194 }, { "epoch": 1.1862927623923667, "grad_norm": 1.9362952709197998, "learning_rate": 0.0005, "loss": 2.290849208831787, "step": 10195 }, { "epoch": 1.1864091226437048, "grad_norm": 1.156564474105835, "learning_rate": 0.0005, "loss": 1.6119266748428345, "step": 10196 }, { "epoch": 1.1865254828950431, "grad_norm": 2.501645565032959, "learning_rate": 0.0005, "loss": 2.3283259868621826, "step": 10197 }, { "epoch": 1.1866418431463812, "grad_norm": 1.14970862865448, "learning_rate": 0.0005, "loss": 1.3398503065109253, "step": 10198 }, { "epoch": 1.1867582033977193, "grad_norm": 1.0335843563079834, "learning_rate": 0.0005, "loss": 2.238807201385498, "step": 10199 }, { "epoch": 1.1868745636490574, "grad_norm": 1.747766375541687, "learning_rate": 0.0005, "loss": 1.8724777698516846, "step": 10200 }, { "epoch": 1.1869909239003955, "grad_norm": 1.299264907836914, "learning_rate": 0.0005, "loss": 1.1547795534133911, "step": 10201 }, { "epoch": 1.1871072841517338, "grad_norm": 2.4452219009399414, "learning_rate": 0.0005, "loss": 1.9535808563232422, "step": 10202 }, { "epoch": 1.187223644403072, "grad_norm": 0.810932457447052, "learning_rate": 0.0005, "loss": 1.288163661956787, "step": 10203 }, { "epoch": 1.18734000465441, "grad_norm": 1.0648261308670044, "learning_rate": 0.0005, "loss": 2.0369274616241455, "step": 10204 }, { "epoch": 1.1874563649057481, "grad_norm": 1.93411123752594, "learning_rate": 0.0005, "loss": 2.0624349117279053, "step": 10205 }, { "epoch": 1.1875727251570862, "grad_norm": 1.4711976051330566, "learning_rate": 0.0005, "loss": 2.254075050354004, "step": 10206 }, { "epoch": 1.1876890854084245, "grad_norm": 1.3297476768493652, "learning_rate": 0.0005, "loss": 2.3658194541931152, "step": 10207 }, { "epoch": 1.1878054456597626, "grad_norm": 0.8464184403419495, "learning_rate": 0.0005, "loss": 1.9910874366760254, "step": 10208 }, { "epoch": 1.1879218059111007, "grad_norm": 1.337659239768982, "learning_rate": 0.0005, "loss": 2.1644930839538574, "step": 10209 }, { "epoch": 1.1880381661624388, "grad_norm": 1.3909950256347656, "learning_rate": 0.0005, "loss": 1.803555965423584, "step": 10210 }, { "epoch": 1.188154526413777, "grad_norm": 1.1900368928909302, "learning_rate": 0.0005, "loss": 1.8881412744522095, "step": 10211 }, { "epoch": 1.1882708866651153, "grad_norm": 1.058497428894043, "learning_rate": 0.0005, "loss": 2.218348979949951, "step": 10212 }, { "epoch": 1.1883872469164534, "grad_norm": 0.5113038420677185, "learning_rate": 0.0005, "loss": 1.1917632818222046, "step": 10213 }, { "epoch": 1.1885036071677915, "grad_norm": 1.2567692995071411, "learning_rate": 0.0005, "loss": 2.6669559478759766, "step": 10214 }, { "epoch": 1.1886199674191296, "grad_norm": 1.0981439352035522, "learning_rate": 0.0005, "loss": 2.121859550476074, "step": 10215 }, { "epoch": 1.1887363276704677, "grad_norm": 1.288995385169983, "learning_rate": 0.0005, "loss": 2.076397657394409, "step": 10216 }, { "epoch": 1.188852687921806, "grad_norm": 1.2841720581054688, "learning_rate": 0.0005, "loss": 2.0753087997436523, "step": 10217 }, { "epoch": 1.188969048173144, "grad_norm": 1.0716170072555542, "learning_rate": 0.0005, "loss": 2.0950751304626465, "step": 10218 }, { "epoch": 1.1890854084244822, "grad_norm": 1.0167800188064575, "learning_rate": 0.0005, "loss": 1.272529125213623, "step": 10219 }, { "epoch": 1.1892017686758203, "grad_norm": 1.563816785812378, "learning_rate": 0.0005, "loss": 2.316235065460205, "step": 10220 }, { "epoch": 1.1893181289271584, "grad_norm": 0.7736263871192932, "learning_rate": 0.0005, "loss": 1.8615505695343018, "step": 10221 }, { "epoch": 1.1894344891784967, "grad_norm": 0.9780483245849609, "learning_rate": 0.0005, "loss": 1.4340541362762451, "step": 10222 }, { "epoch": 1.1895508494298348, "grad_norm": 1.5843089818954468, "learning_rate": 0.0005, "loss": 2.006972312927246, "step": 10223 }, { "epoch": 1.189667209681173, "grad_norm": 0.9694278240203857, "learning_rate": 0.0005, "loss": 1.4555866718292236, "step": 10224 }, { "epoch": 1.189783569932511, "grad_norm": 1.40117609500885, "learning_rate": 0.0005, "loss": 2.192453145980835, "step": 10225 }, { "epoch": 1.189899930183849, "grad_norm": 1.3642628192901611, "learning_rate": 0.0005, "loss": 2.078735828399658, "step": 10226 }, { "epoch": 1.1900162904351874, "grad_norm": 1.1597622632980347, "learning_rate": 0.0005, "loss": 2.5845251083374023, "step": 10227 }, { "epoch": 1.1901326506865255, "grad_norm": 1.6494207382202148, "learning_rate": 0.0005, "loss": 1.503348708152771, "step": 10228 }, { "epoch": 1.1902490109378636, "grad_norm": 1.1106665134429932, "learning_rate": 0.0005, "loss": 1.828223466873169, "step": 10229 }, { "epoch": 1.1903653711892017, "grad_norm": 0.8509787321090698, "learning_rate": 0.0005, "loss": 2.1480677127838135, "step": 10230 }, { "epoch": 1.1904817314405398, "grad_norm": 1.0364550352096558, "learning_rate": 0.0005, "loss": 1.9677186012268066, "step": 10231 }, { "epoch": 1.1905980916918781, "grad_norm": 1.1815346479415894, "learning_rate": 0.0005, "loss": 2.884749412536621, "step": 10232 }, { "epoch": 1.1907144519432162, "grad_norm": 0.7474707365036011, "learning_rate": 0.0005, "loss": 1.590612530708313, "step": 10233 }, { "epoch": 1.1908308121945543, "grad_norm": 1.0320920944213867, "learning_rate": 0.0005, "loss": 1.3435653448104858, "step": 10234 }, { "epoch": 1.1909471724458924, "grad_norm": 0.7918128371238708, "learning_rate": 0.0005, "loss": 1.4765113592147827, "step": 10235 }, { "epoch": 1.1910635326972305, "grad_norm": 1.1432336568832397, "learning_rate": 0.0005, "loss": 1.842780351638794, "step": 10236 }, { "epoch": 1.1911798929485689, "grad_norm": 1.2087496519088745, "learning_rate": 0.0005, "loss": 1.5253381729125977, "step": 10237 }, { "epoch": 1.191296253199907, "grad_norm": 1.4569374322891235, "learning_rate": 0.0005, "loss": 1.8053292036056519, "step": 10238 }, { "epoch": 1.191412613451245, "grad_norm": 1.2288986444473267, "learning_rate": 0.0005, "loss": 2.456026315689087, "step": 10239 }, { "epoch": 1.1915289737025831, "grad_norm": 1.2005101442337036, "learning_rate": 0.0005, "loss": 2.5374412536621094, "step": 10240 }, { "epoch": 1.1916453339539212, "grad_norm": 0.6788423657417297, "learning_rate": 0.0005, "loss": 1.7439088821411133, "step": 10241 }, { "epoch": 1.1917616942052596, "grad_norm": 1.0961697101593018, "learning_rate": 0.0005, "loss": 1.527714729309082, "step": 10242 }, { "epoch": 1.1918780544565977, "grad_norm": 1.176461100578308, "learning_rate": 0.0005, "loss": 2.1837854385375977, "step": 10243 }, { "epoch": 1.1919944147079358, "grad_norm": 1.5263792276382446, "learning_rate": 0.0005, "loss": 2.338993549346924, "step": 10244 }, { "epoch": 1.1921107749592739, "grad_norm": 0.7738340497016907, "learning_rate": 0.0005, "loss": 1.4225867986679077, "step": 10245 }, { "epoch": 1.192227135210612, "grad_norm": 1.0775495767593384, "learning_rate": 0.0005, "loss": 1.6714704036712646, "step": 10246 }, { "epoch": 1.1923434954619503, "grad_norm": 0.8412376046180725, "learning_rate": 0.0005, "loss": 1.667636513710022, "step": 10247 }, { "epoch": 1.1924598557132884, "grad_norm": 1.2254875898361206, "learning_rate": 0.0005, "loss": 2.1260831356048584, "step": 10248 }, { "epoch": 1.1925762159646265, "grad_norm": 1.3067476749420166, "learning_rate": 0.0005, "loss": 2.498068332672119, "step": 10249 }, { "epoch": 1.1926925762159646, "grad_norm": 1.5035665035247803, "learning_rate": 0.0005, "loss": 2.391174793243408, "step": 10250 }, { "epoch": 1.1928089364673027, "grad_norm": 0.9876210689544678, "learning_rate": 0.0005, "loss": 2.0093188285827637, "step": 10251 }, { "epoch": 1.192925296718641, "grad_norm": 1.1053625345230103, "learning_rate": 0.0005, "loss": 1.497464656829834, "step": 10252 }, { "epoch": 1.193041656969979, "grad_norm": 0.9194111227989197, "learning_rate": 0.0005, "loss": 1.4228966236114502, "step": 10253 }, { "epoch": 1.1931580172213172, "grad_norm": 2.339163064956665, "learning_rate": 0.0005, "loss": 1.8249160051345825, "step": 10254 }, { "epoch": 1.1932743774726553, "grad_norm": 0.9037461280822754, "learning_rate": 0.0005, "loss": 2.1338746547698975, "step": 10255 }, { "epoch": 1.1933907377239934, "grad_norm": 0.9077068567276001, "learning_rate": 0.0005, "loss": 1.551738977432251, "step": 10256 }, { "epoch": 1.1935070979753317, "grad_norm": 1.2568624019622803, "learning_rate": 0.0005, "loss": 2.366424560546875, "step": 10257 }, { "epoch": 1.1936234582266698, "grad_norm": 0.5580053925514221, "learning_rate": 0.0005, "loss": 1.3194842338562012, "step": 10258 }, { "epoch": 1.193739818478008, "grad_norm": 1.4857739210128784, "learning_rate": 0.0005, "loss": 2.2890517711639404, "step": 10259 }, { "epoch": 1.193856178729346, "grad_norm": 1.2117621898651123, "learning_rate": 0.0005, "loss": 2.044339656829834, "step": 10260 }, { "epoch": 1.1939725389806841, "grad_norm": 1.765119194984436, "learning_rate": 0.0005, "loss": 2.4958271980285645, "step": 10261 }, { "epoch": 1.1940888992320224, "grad_norm": 1.2069774866104126, "learning_rate": 0.0005, "loss": 2.8921947479248047, "step": 10262 }, { "epoch": 1.1942052594833605, "grad_norm": 1.118843674659729, "learning_rate": 0.0005, "loss": 2.1711087226867676, "step": 10263 }, { "epoch": 1.1943216197346986, "grad_norm": 0.9700037240982056, "learning_rate": 0.0005, "loss": 1.549013614654541, "step": 10264 }, { "epoch": 1.1944379799860367, "grad_norm": 0.9518982768058777, "learning_rate": 0.0005, "loss": 1.7379437685012817, "step": 10265 }, { "epoch": 1.1945543402373748, "grad_norm": 1.4053465127944946, "learning_rate": 0.0005, "loss": 3.007300615310669, "step": 10266 }, { "epoch": 1.1946707004887132, "grad_norm": 1.512799859046936, "learning_rate": 0.0005, "loss": 1.9881013631820679, "step": 10267 }, { "epoch": 1.1947870607400513, "grad_norm": 1.8298332691192627, "learning_rate": 0.0005, "loss": 2.342233657836914, "step": 10268 }, { "epoch": 1.1949034209913894, "grad_norm": 1.2932116985321045, "learning_rate": 0.0005, "loss": 2.522097587585449, "step": 10269 }, { "epoch": 1.1950197812427275, "grad_norm": 0.6292586326599121, "learning_rate": 0.0005, "loss": 1.5396242141723633, "step": 10270 }, { "epoch": 1.1951361414940656, "grad_norm": 0.9963943362236023, "learning_rate": 0.0005, "loss": 1.704347014427185, "step": 10271 }, { "epoch": 1.1952525017454039, "grad_norm": 1.4701852798461914, "learning_rate": 0.0005, "loss": 2.095527172088623, "step": 10272 }, { "epoch": 1.195368861996742, "grad_norm": 1.0773156881332397, "learning_rate": 0.0005, "loss": 1.469578504562378, "step": 10273 }, { "epoch": 1.19548522224808, "grad_norm": 1.4091250896453857, "learning_rate": 0.0005, "loss": 1.8022451400756836, "step": 10274 }, { "epoch": 1.1956015824994182, "grad_norm": 1.4913891553878784, "learning_rate": 0.0005, "loss": 3.029582977294922, "step": 10275 }, { "epoch": 1.1957179427507563, "grad_norm": 0.8965685367584229, "learning_rate": 0.0005, "loss": 2.0466370582580566, "step": 10276 }, { "epoch": 1.1958343030020946, "grad_norm": 1.5945807695388794, "learning_rate": 0.0005, "loss": 1.7971861362457275, "step": 10277 }, { "epoch": 1.1959506632534327, "grad_norm": 1.252509355545044, "learning_rate": 0.0005, "loss": 1.710733413696289, "step": 10278 }, { "epoch": 1.1960670235047708, "grad_norm": 1.715193748474121, "learning_rate": 0.0005, "loss": 2.2828927040100098, "step": 10279 }, { "epoch": 1.1961833837561089, "grad_norm": 1.248490810394287, "learning_rate": 0.0005, "loss": 2.946056842803955, "step": 10280 }, { "epoch": 1.196299744007447, "grad_norm": 0.8840658068656921, "learning_rate": 0.0005, "loss": 2.0724477767944336, "step": 10281 }, { "epoch": 1.1964161042587853, "grad_norm": 1.1781394481658936, "learning_rate": 0.0005, "loss": 1.8866046667099, "step": 10282 }, { "epoch": 1.1965324645101234, "grad_norm": 1.3650623559951782, "learning_rate": 0.0005, "loss": 1.9544011354446411, "step": 10283 }, { "epoch": 1.1966488247614615, "grad_norm": 0.9493118524551392, "learning_rate": 0.0005, "loss": 1.5022271871566772, "step": 10284 }, { "epoch": 1.1967651850127996, "grad_norm": 1.4770869016647339, "learning_rate": 0.0005, "loss": 2.739778995513916, "step": 10285 }, { "epoch": 1.1968815452641377, "grad_norm": 1.2095606327056885, "learning_rate": 0.0005, "loss": 1.815913438796997, "step": 10286 }, { "epoch": 1.196997905515476, "grad_norm": 0.56517493724823, "learning_rate": 0.0005, "loss": 1.7370623350143433, "step": 10287 }, { "epoch": 1.1971142657668141, "grad_norm": 1.1367846727371216, "learning_rate": 0.0005, "loss": 2.293910503387451, "step": 10288 }, { "epoch": 1.1972306260181522, "grad_norm": 1.3725963830947876, "learning_rate": 0.0005, "loss": 2.0123045444488525, "step": 10289 }, { "epoch": 1.1973469862694903, "grad_norm": 1.5879607200622559, "learning_rate": 0.0005, "loss": 2.7683327198028564, "step": 10290 }, { "epoch": 1.1974633465208284, "grad_norm": 1.2307530641555786, "learning_rate": 0.0005, "loss": 2.178572177886963, "step": 10291 }, { "epoch": 1.1975797067721667, "grad_norm": 1.3225733041763306, "learning_rate": 0.0005, "loss": 1.7206201553344727, "step": 10292 }, { "epoch": 1.1976960670235048, "grad_norm": 1.0949734449386597, "learning_rate": 0.0005, "loss": 1.9729520082473755, "step": 10293 }, { "epoch": 1.197812427274843, "grad_norm": 1.1990996599197388, "learning_rate": 0.0005, "loss": 2.4467170238494873, "step": 10294 }, { "epoch": 1.197928787526181, "grad_norm": 1.2694398164749146, "learning_rate": 0.0005, "loss": 2.0036139488220215, "step": 10295 }, { "epoch": 1.1980451477775191, "grad_norm": 0.5465975999832153, "learning_rate": 0.0005, "loss": 1.3070569038391113, "step": 10296 }, { "epoch": 1.1981615080288575, "grad_norm": 0.9212431311607361, "learning_rate": 0.0005, "loss": 1.292372465133667, "step": 10297 }, { "epoch": 1.1982778682801956, "grad_norm": 1.0696570873260498, "learning_rate": 0.0005, "loss": 1.681322455406189, "step": 10298 }, { "epoch": 1.1983942285315337, "grad_norm": 1.1477837562561035, "learning_rate": 0.0005, "loss": 2.5288233757019043, "step": 10299 }, { "epoch": 1.1985105887828718, "grad_norm": 1.0013378858566284, "learning_rate": 0.0005, "loss": 2.0207624435424805, "step": 10300 }, { "epoch": 1.1986269490342099, "grad_norm": 2.0424115657806396, "learning_rate": 0.0005, "loss": 1.996492862701416, "step": 10301 }, { "epoch": 1.1987433092855482, "grad_norm": 1.1346886157989502, "learning_rate": 0.0005, "loss": 1.6683655977249146, "step": 10302 }, { "epoch": 1.1988596695368863, "grad_norm": 0.9526774287223816, "learning_rate": 0.0005, "loss": 1.9624755382537842, "step": 10303 }, { "epoch": 1.1989760297882244, "grad_norm": 0.7191323637962341, "learning_rate": 0.0005, "loss": 1.656299114227295, "step": 10304 }, { "epoch": 1.1990923900395625, "grad_norm": 1.9861969947814941, "learning_rate": 0.0005, "loss": 2.3663344383239746, "step": 10305 }, { "epoch": 1.1992087502909006, "grad_norm": 1.6841049194335938, "learning_rate": 0.0005, "loss": 1.9131250381469727, "step": 10306 }, { "epoch": 1.199325110542239, "grad_norm": 1.3402378559112549, "learning_rate": 0.0005, "loss": 2.4108664989471436, "step": 10307 }, { "epoch": 1.199441470793577, "grad_norm": 1.120797038078308, "learning_rate": 0.0005, "loss": 1.5492165088653564, "step": 10308 }, { "epoch": 1.199557831044915, "grad_norm": 0.9548761248588562, "learning_rate": 0.0005, "loss": 1.9881808757781982, "step": 10309 }, { "epoch": 1.1996741912962532, "grad_norm": 0.8895648717880249, "learning_rate": 0.0005, "loss": 2.2837798595428467, "step": 10310 }, { "epoch": 1.1997905515475913, "grad_norm": 1.360836386680603, "learning_rate": 0.0005, "loss": 1.6528217792510986, "step": 10311 }, { "epoch": 1.1999069117989296, "grad_norm": 1.4993376731872559, "learning_rate": 0.0005, "loss": 2.357900381088257, "step": 10312 }, { "epoch": 1.2000232720502677, "grad_norm": 1.080084204673767, "learning_rate": 0.0005, "loss": 1.520756483078003, "step": 10313 }, { "epoch": 1.2001396323016058, "grad_norm": 1.1335357427597046, "learning_rate": 0.0005, "loss": 1.1884937286376953, "step": 10314 }, { "epoch": 1.200255992552944, "grad_norm": 1.139799952507019, "learning_rate": 0.0005, "loss": 2.1991710662841797, "step": 10315 }, { "epoch": 1.200372352804282, "grad_norm": 1.0058811902999878, "learning_rate": 0.0005, "loss": 1.8101474046707153, "step": 10316 }, { "epoch": 1.20048871305562, "grad_norm": 1.631212592124939, "learning_rate": 0.0005, "loss": 2.025012493133545, "step": 10317 }, { "epoch": 1.2006050733069584, "grad_norm": 1.0909227132797241, "learning_rate": 0.0005, "loss": 1.8261586427688599, "step": 10318 }, { "epoch": 1.2007214335582965, "grad_norm": 1.0800178050994873, "learning_rate": 0.0005, "loss": 1.9131872653961182, "step": 10319 }, { "epoch": 1.2008377938096346, "grad_norm": 0.6398373246192932, "learning_rate": 0.0005, "loss": 1.2720363140106201, "step": 10320 }, { "epoch": 1.2009541540609727, "grad_norm": 1.0977466106414795, "learning_rate": 0.0005, "loss": 2.097768545150757, "step": 10321 }, { "epoch": 1.2010705143123108, "grad_norm": 1.472525954246521, "learning_rate": 0.0005, "loss": 1.7907159328460693, "step": 10322 }, { "epoch": 1.2011868745636491, "grad_norm": 0.8810976147651672, "learning_rate": 0.0005, "loss": 1.2442121505737305, "step": 10323 }, { "epoch": 1.2013032348149872, "grad_norm": 1.3330024480819702, "learning_rate": 0.0005, "loss": 1.683829665184021, "step": 10324 }, { "epoch": 1.2014195950663253, "grad_norm": 1.3822885751724243, "learning_rate": 0.0005, "loss": 2.1612308025360107, "step": 10325 }, { "epoch": 1.2015359553176634, "grad_norm": 1.375839352607727, "learning_rate": 0.0005, "loss": 1.7300529479980469, "step": 10326 }, { "epoch": 1.2016523155690015, "grad_norm": 1.3546861410140991, "learning_rate": 0.0005, "loss": 2.346039295196533, "step": 10327 }, { "epoch": 1.2017686758203399, "grad_norm": 0.6553544402122498, "learning_rate": 0.0005, "loss": 1.136789083480835, "step": 10328 }, { "epoch": 1.201885036071678, "grad_norm": 1.474177598953247, "learning_rate": 0.0005, "loss": 1.9310132265090942, "step": 10329 }, { "epoch": 1.202001396323016, "grad_norm": 1.9273625612258911, "learning_rate": 0.0005, "loss": 1.9899370670318604, "step": 10330 }, { "epoch": 1.2021177565743542, "grad_norm": 1.1270778179168701, "learning_rate": 0.0005, "loss": 2.205730438232422, "step": 10331 }, { "epoch": 1.2022341168256923, "grad_norm": 0.8084068894386292, "learning_rate": 0.0005, "loss": 1.9156817197799683, "step": 10332 }, { "epoch": 1.2023504770770306, "grad_norm": 1.0973141193389893, "learning_rate": 0.0005, "loss": 2.0229880809783936, "step": 10333 }, { "epoch": 1.2024668373283687, "grad_norm": 1.0472025871276855, "learning_rate": 0.0005, "loss": 1.6240932941436768, "step": 10334 }, { "epoch": 1.2025831975797068, "grad_norm": 1.3578567504882812, "learning_rate": 0.0005, "loss": 2.721902370452881, "step": 10335 }, { "epoch": 1.2026995578310449, "grad_norm": 1.8921624422073364, "learning_rate": 0.0005, "loss": 2.234605073928833, "step": 10336 }, { "epoch": 1.202815918082383, "grad_norm": 1.6272435188293457, "learning_rate": 0.0005, "loss": 1.7358982563018799, "step": 10337 }, { "epoch": 1.2029322783337213, "grad_norm": 0.987489640712738, "learning_rate": 0.0005, "loss": 1.7463626861572266, "step": 10338 }, { "epoch": 1.2030486385850594, "grad_norm": 1.2665870189666748, "learning_rate": 0.0005, "loss": 1.8267393112182617, "step": 10339 }, { "epoch": 1.2031649988363975, "grad_norm": 1.3112472295761108, "learning_rate": 0.0005, "loss": 2.475203037261963, "step": 10340 }, { "epoch": 1.2032813590877356, "grad_norm": 1.3203121423721313, "learning_rate": 0.0005, "loss": 2.176293134689331, "step": 10341 }, { "epoch": 1.2033977193390737, "grad_norm": 0.8735288977622986, "learning_rate": 0.0005, "loss": 2.076598644256592, "step": 10342 }, { "epoch": 1.203514079590412, "grad_norm": 1.0052292346954346, "learning_rate": 0.0005, "loss": 2.0013885498046875, "step": 10343 }, { "epoch": 1.20363043984175, "grad_norm": 1.654747724533081, "learning_rate": 0.0005, "loss": 2.109435558319092, "step": 10344 }, { "epoch": 1.2037468000930882, "grad_norm": 0.971899688243866, "learning_rate": 0.0005, "loss": 2.0225415229797363, "step": 10345 }, { "epoch": 1.2038631603444263, "grad_norm": 1.5697520971298218, "learning_rate": 0.0005, "loss": 2.310011863708496, "step": 10346 }, { "epoch": 1.2039795205957644, "grad_norm": 1.0814478397369385, "learning_rate": 0.0005, "loss": 1.7263635396957397, "step": 10347 }, { "epoch": 1.2040958808471025, "grad_norm": 1.0967416763305664, "learning_rate": 0.0005, "loss": 2.248044490814209, "step": 10348 }, { "epoch": 1.2042122410984408, "grad_norm": 1.1367003917694092, "learning_rate": 0.0005, "loss": 2.9196434020996094, "step": 10349 }, { "epoch": 1.204328601349779, "grad_norm": 1.3075065612792969, "learning_rate": 0.0005, "loss": 2.2940354347229004, "step": 10350 }, { "epoch": 1.204444961601117, "grad_norm": 1.8938499689102173, "learning_rate": 0.0005, "loss": 2.465970993041992, "step": 10351 }, { "epoch": 1.2045613218524551, "grad_norm": 1.1088603734970093, "learning_rate": 0.0005, "loss": 2.0948410034179688, "step": 10352 }, { "epoch": 1.2046776821037932, "grad_norm": 1.3135877847671509, "learning_rate": 0.0005, "loss": 1.740980625152588, "step": 10353 }, { "epoch": 1.2047940423551315, "grad_norm": 1.3062599897384644, "learning_rate": 0.0005, "loss": 2.034911632537842, "step": 10354 }, { "epoch": 1.2049104026064696, "grad_norm": 1.1343920230865479, "learning_rate": 0.0005, "loss": 1.7564810514450073, "step": 10355 }, { "epoch": 1.2050267628578077, "grad_norm": 1.2891422510147095, "learning_rate": 0.0005, "loss": 2.364187717437744, "step": 10356 }, { "epoch": 1.2051431231091458, "grad_norm": 1.0400151014328003, "learning_rate": 0.0005, "loss": 1.8252134323120117, "step": 10357 }, { "epoch": 1.205259483360484, "grad_norm": 2.72186279296875, "learning_rate": 0.0005, "loss": 2.0041110515594482, "step": 10358 }, { "epoch": 1.2053758436118223, "grad_norm": 1.0981141328811646, "learning_rate": 0.0005, "loss": 1.8455874919891357, "step": 10359 }, { "epoch": 1.2054922038631604, "grad_norm": 0.6592535376548767, "learning_rate": 0.0005, "loss": 1.2707934379577637, "step": 10360 }, { "epoch": 1.2056085641144985, "grad_norm": 0.633973240852356, "learning_rate": 0.0005, "loss": 1.3553341627120972, "step": 10361 }, { "epoch": 1.2057249243658366, "grad_norm": 1.0365064144134521, "learning_rate": 0.0005, "loss": 2.0711519718170166, "step": 10362 }, { "epoch": 1.2058412846171747, "grad_norm": 1.4292221069335938, "learning_rate": 0.0005, "loss": 2.0461041927337646, "step": 10363 }, { "epoch": 1.205957644868513, "grad_norm": 1.0910183191299438, "learning_rate": 0.0005, "loss": 1.7031116485595703, "step": 10364 }, { "epoch": 1.206074005119851, "grad_norm": 1.6294091939926147, "learning_rate": 0.0005, "loss": 2.4597389698028564, "step": 10365 }, { "epoch": 1.2061903653711892, "grad_norm": 1.1339406967163086, "learning_rate": 0.0005, "loss": 1.5657566785812378, "step": 10366 }, { "epoch": 1.2063067256225273, "grad_norm": 1.0660500526428223, "learning_rate": 0.0005, "loss": 2.2617087364196777, "step": 10367 }, { "epoch": 1.2064230858738654, "grad_norm": 1.2305760383605957, "learning_rate": 0.0005, "loss": 2.0727832317352295, "step": 10368 }, { "epoch": 1.2065394461252037, "grad_norm": 1.0652849674224854, "learning_rate": 0.0005, "loss": 1.8661001920700073, "step": 10369 }, { "epoch": 1.2066558063765418, "grad_norm": 1.0203571319580078, "learning_rate": 0.0005, "loss": 1.884656310081482, "step": 10370 }, { "epoch": 1.20677216662788, "grad_norm": 1.2349071502685547, "learning_rate": 0.0005, "loss": 1.6948294639587402, "step": 10371 }, { "epoch": 1.206888526879218, "grad_norm": 1.1917662620544434, "learning_rate": 0.0005, "loss": 2.027245044708252, "step": 10372 }, { "epoch": 1.207004887130556, "grad_norm": 1.4880353212356567, "learning_rate": 0.0005, "loss": 2.0560195446014404, "step": 10373 }, { "epoch": 1.2071212473818944, "grad_norm": 0.8572818040847778, "learning_rate": 0.0005, "loss": 1.9197816848754883, "step": 10374 }, { "epoch": 1.2072376076332325, "grad_norm": 1.4328718185424805, "learning_rate": 0.0005, "loss": 2.226714611053467, "step": 10375 }, { "epoch": 1.2073539678845706, "grad_norm": 1.4623957872390747, "learning_rate": 0.0005, "loss": 1.933973789215088, "step": 10376 }, { "epoch": 1.2074703281359087, "grad_norm": 1.20992910861969, "learning_rate": 0.0005, "loss": 1.8914008140563965, "step": 10377 }, { "epoch": 1.2075866883872468, "grad_norm": 1.07823646068573, "learning_rate": 0.0005, "loss": 1.9570155143737793, "step": 10378 }, { "epoch": 1.2077030486385851, "grad_norm": 1.3646711111068726, "learning_rate": 0.0005, "loss": 1.6626043319702148, "step": 10379 }, { "epoch": 1.2078194088899232, "grad_norm": 1.6090682744979858, "learning_rate": 0.0005, "loss": 1.329785943031311, "step": 10380 }, { "epoch": 1.2079357691412613, "grad_norm": 1.022599697113037, "learning_rate": 0.0005, "loss": 2.407886266708374, "step": 10381 }, { "epoch": 1.2080521293925994, "grad_norm": 1.0771992206573486, "learning_rate": 0.0005, "loss": 1.9432324171066284, "step": 10382 }, { "epoch": 1.2081684896439375, "grad_norm": 1.3811417818069458, "learning_rate": 0.0005, "loss": 2.2601711750030518, "step": 10383 }, { "epoch": 1.2082848498952758, "grad_norm": 1.4416375160217285, "learning_rate": 0.0005, "loss": 2.3890573978424072, "step": 10384 }, { "epoch": 1.208401210146614, "grad_norm": 1.1193211078643799, "learning_rate": 0.0005, "loss": 2.5398457050323486, "step": 10385 }, { "epoch": 1.208517570397952, "grad_norm": 1.4429982900619507, "learning_rate": 0.0005, "loss": 2.792074203491211, "step": 10386 }, { "epoch": 1.2086339306492901, "grad_norm": 1.4163388013839722, "learning_rate": 0.0005, "loss": 2.3542542457580566, "step": 10387 }, { "epoch": 1.2087502909006282, "grad_norm": 0.7988329529762268, "learning_rate": 0.0005, "loss": 1.2851765155792236, "step": 10388 }, { "epoch": 1.2088666511519666, "grad_norm": 1.3523809909820557, "learning_rate": 0.0005, "loss": 1.595705270767212, "step": 10389 }, { "epoch": 1.2089830114033047, "grad_norm": 0.8852105736732483, "learning_rate": 0.0005, "loss": 1.7089364528656006, "step": 10390 }, { "epoch": 1.2090993716546428, "grad_norm": 0.909953236579895, "learning_rate": 0.0005, "loss": 1.5827455520629883, "step": 10391 }, { "epoch": 1.2092157319059809, "grad_norm": 1.6679340600967407, "learning_rate": 0.0005, "loss": 1.985612154006958, "step": 10392 }, { "epoch": 1.209332092157319, "grad_norm": 1.393795132637024, "learning_rate": 0.0005, "loss": 2.0372533798217773, "step": 10393 }, { "epoch": 1.2094484524086573, "grad_norm": 1.210584044456482, "learning_rate": 0.0005, "loss": 2.1538949012756348, "step": 10394 }, { "epoch": 1.2095648126599954, "grad_norm": 1.240349531173706, "learning_rate": 0.0005, "loss": 1.7945809364318848, "step": 10395 }, { "epoch": 1.2096811729113335, "grad_norm": 1.5341780185699463, "learning_rate": 0.0005, "loss": 2.0765671730041504, "step": 10396 }, { "epoch": 1.2097975331626716, "grad_norm": 1.5936251878738403, "learning_rate": 0.0005, "loss": 3.0347535610198975, "step": 10397 }, { "epoch": 1.2099138934140097, "grad_norm": 1.2892895936965942, "learning_rate": 0.0005, "loss": 2.075798988342285, "step": 10398 }, { "epoch": 1.210030253665348, "grad_norm": 1.046851634979248, "learning_rate": 0.0005, "loss": 1.8438339233398438, "step": 10399 }, { "epoch": 1.210146613916686, "grad_norm": 1.4682949781417847, "learning_rate": 0.0005, "loss": 2.285794973373413, "step": 10400 }, { "epoch": 1.2102629741680242, "grad_norm": 0.7808605432510376, "learning_rate": 0.0005, "loss": 2.0245819091796875, "step": 10401 }, { "epoch": 1.2103793344193623, "grad_norm": 1.2796231508255005, "learning_rate": 0.0005, "loss": 2.2449944019317627, "step": 10402 }, { "epoch": 1.2104956946707004, "grad_norm": 1.249682903289795, "learning_rate": 0.0005, "loss": 2.2831757068634033, "step": 10403 }, { "epoch": 1.2106120549220387, "grad_norm": 1.6582143306732178, "learning_rate": 0.0005, "loss": 1.9773483276367188, "step": 10404 }, { "epoch": 1.2107284151733768, "grad_norm": 0.9049345254898071, "learning_rate": 0.0005, "loss": 1.6236639022827148, "step": 10405 }, { "epoch": 1.210844775424715, "grad_norm": 1.017163634300232, "learning_rate": 0.0005, "loss": 2.4607725143432617, "step": 10406 }, { "epoch": 1.210961135676053, "grad_norm": 1.5130876302719116, "learning_rate": 0.0005, "loss": 2.450204372406006, "step": 10407 }, { "epoch": 1.2110774959273911, "grad_norm": 0.8718411922454834, "learning_rate": 0.0005, "loss": 1.3583908081054688, "step": 10408 }, { "epoch": 1.2111938561787294, "grad_norm": 1.6403088569641113, "learning_rate": 0.0005, "loss": 2.100496768951416, "step": 10409 }, { "epoch": 1.2113102164300675, "grad_norm": 1.1670979261398315, "learning_rate": 0.0005, "loss": 1.964707851409912, "step": 10410 }, { "epoch": 1.2114265766814056, "grad_norm": 1.4231315851211548, "learning_rate": 0.0005, "loss": 1.4156928062438965, "step": 10411 }, { "epoch": 1.2115429369327437, "grad_norm": 1.480196237564087, "learning_rate": 0.0005, "loss": 2.2908711433410645, "step": 10412 }, { "epoch": 1.2116592971840818, "grad_norm": 1.1313191652297974, "learning_rate": 0.0005, "loss": 1.1927047967910767, "step": 10413 }, { "epoch": 1.2117756574354202, "grad_norm": 1.4195611476898193, "learning_rate": 0.0005, "loss": 2.071941375732422, "step": 10414 }, { "epoch": 1.2118920176867582, "grad_norm": 1.230150818824768, "learning_rate": 0.0005, "loss": 1.568406343460083, "step": 10415 }, { "epoch": 1.2120083779380963, "grad_norm": 4.720644950866699, "learning_rate": 0.0005, "loss": 1.4146795272827148, "step": 10416 }, { "epoch": 1.2121247381894344, "grad_norm": 1.3064653873443604, "learning_rate": 0.0005, "loss": 2.1511971950531006, "step": 10417 }, { "epoch": 1.2122410984407725, "grad_norm": 1.4137206077575684, "learning_rate": 0.0005, "loss": 2.173067092895508, "step": 10418 }, { "epoch": 1.2123574586921109, "grad_norm": 1.3871256113052368, "learning_rate": 0.0005, "loss": 2.010624885559082, "step": 10419 }, { "epoch": 1.212473818943449, "grad_norm": 1.3255990743637085, "learning_rate": 0.0005, "loss": 1.837528944015503, "step": 10420 }, { "epoch": 1.212590179194787, "grad_norm": 0.8170570731163025, "learning_rate": 0.0005, "loss": 1.7183719873428345, "step": 10421 }, { "epoch": 1.2127065394461252, "grad_norm": 1.83701491355896, "learning_rate": 0.0005, "loss": 1.8129959106445312, "step": 10422 }, { "epoch": 1.2128228996974633, "grad_norm": 1.4036073684692383, "learning_rate": 0.0005, "loss": 2.0325305461883545, "step": 10423 }, { "epoch": 1.2129392599488016, "grad_norm": 0.9640429615974426, "learning_rate": 0.0005, "loss": 2.113063335418701, "step": 10424 }, { "epoch": 1.2130556202001397, "grad_norm": 2.0365586280822754, "learning_rate": 0.0005, "loss": 1.7744559049606323, "step": 10425 }, { "epoch": 1.2131719804514778, "grad_norm": 0.6474990248680115, "learning_rate": 0.0005, "loss": 1.4387017488479614, "step": 10426 }, { "epoch": 1.2132883407028159, "grad_norm": 1.1768244504928589, "learning_rate": 0.0005, "loss": 1.8008276224136353, "step": 10427 }, { "epoch": 1.213404700954154, "grad_norm": 1.3564660549163818, "learning_rate": 0.0005, "loss": 1.8692829608917236, "step": 10428 }, { "epoch": 1.2135210612054923, "grad_norm": 1.3594907522201538, "learning_rate": 0.0005, "loss": 2.235617160797119, "step": 10429 }, { "epoch": 1.2136374214568304, "grad_norm": 0.8994200825691223, "learning_rate": 0.0005, "loss": 1.8111588954925537, "step": 10430 }, { "epoch": 1.2137537817081685, "grad_norm": 1.533177137374878, "learning_rate": 0.0005, "loss": 2.1848678588867188, "step": 10431 }, { "epoch": 1.2138701419595066, "grad_norm": 1.0862082242965698, "learning_rate": 0.0005, "loss": 2.197932243347168, "step": 10432 }, { "epoch": 1.2139865022108447, "grad_norm": 1.2785214185714722, "learning_rate": 0.0005, "loss": 1.5335280895233154, "step": 10433 }, { "epoch": 1.214102862462183, "grad_norm": 1.1672463417053223, "learning_rate": 0.0005, "loss": 1.75981867313385, "step": 10434 }, { "epoch": 1.2142192227135211, "grad_norm": 1.266136646270752, "learning_rate": 0.0005, "loss": 1.7529680728912354, "step": 10435 }, { "epoch": 1.2143355829648592, "grad_norm": 1.4713890552520752, "learning_rate": 0.0005, "loss": 2.2166686058044434, "step": 10436 }, { "epoch": 1.2144519432161973, "grad_norm": 5.489457130432129, "learning_rate": 0.0005, "loss": 1.812070369720459, "step": 10437 }, { "epoch": 1.2145683034675354, "grad_norm": 1.071390151977539, "learning_rate": 0.0005, "loss": 1.9944857358932495, "step": 10438 }, { "epoch": 1.2146846637188737, "grad_norm": 0.8430845737457275, "learning_rate": 0.0005, "loss": 1.9642002582550049, "step": 10439 }, { "epoch": 1.2148010239702118, "grad_norm": 1.095118761062622, "learning_rate": 0.0005, "loss": 1.6355388164520264, "step": 10440 }, { "epoch": 1.21491738422155, "grad_norm": 1.0875660181045532, "learning_rate": 0.0005, "loss": 2.141052722930908, "step": 10441 }, { "epoch": 1.215033744472888, "grad_norm": 1.474953055381775, "learning_rate": 0.0005, "loss": 1.4335273504257202, "step": 10442 }, { "epoch": 1.2151501047242261, "grad_norm": 1.0980381965637207, "learning_rate": 0.0005, "loss": 1.7066881656646729, "step": 10443 }, { "epoch": 1.2152664649755645, "grad_norm": 1.4894061088562012, "learning_rate": 0.0005, "loss": 2.03416109085083, "step": 10444 }, { "epoch": 1.2153828252269026, "grad_norm": 1.4642436504364014, "learning_rate": 0.0005, "loss": 2.035590648651123, "step": 10445 }, { "epoch": 1.2154991854782407, "grad_norm": 1.6404145956039429, "learning_rate": 0.0005, "loss": 2.0002036094665527, "step": 10446 }, { "epoch": 1.2156155457295788, "grad_norm": 1.0189769268035889, "learning_rate": 0.0005, "loss": 1.6705431938171387, "step": 10447 }, { "epoch": 1.2157319059809168, "grad_norm": 1.2819995880126953, "learning_rate": 0.0005, "loss": 1.752587914466858, "step": 10448 }, { "epoch": 1.2158482662322552, "grad_norm": 1.6536064147949219, "learning_rate": 0.0005, "loss": 2.590620756149292, "step": 10449 }, { "epoch": 1.2159646264835933, "grad_norm": 1.1664665937423706, "learning_rate": 0.0005, "loss": 2.211825132369995, "step": 10450 }, { "epoch": 1.2160809867349314, "grad_norm": 1.5782300233840942, "learning_rate": 0.0005, "loss": 1.9274895191192627, "step": 10451 }, { "epoch": 1.2161973469862695, "grad_norm": 0.5720386505126953, "learning_rate": 0.0005, "loss": 1.3688583374023438, "step": 10452 }, { "epoch": 1.2163137072376076, "grad_norm": 4.084535121917725, "learning_rate": 0.0005, "loss": 1.8880995512008667, "step": 10453 }, { "epoch": 1.2164300674889459, "grad_norm": 0.7576525211334229, "learning_rate": 0.0005, "loss": 1.255316138267517, "step": 10454 }, { "epoch": 1.216546427740284, "grad_norm": 1.5220001935958862, "learning_rate": 0.0005, "loss": 2.2753686904907227, "step": 10455 }, { "epoch": 1.216662787991622, "grad_norm": 1.2455525398254395, "learning_rate": 0.0005, "loss": 1.367994785308838, "step": 10456 }, { "epoch": 1.2167791482429602, "grad_norm": 1.2816530466079712, "learning_rate": 0.0005, "loss": 1.504845142364502, "step": 10457 }, { "epoch": 1.2168955084942983, "grad_norm": 2.1449503898620605, "learning_rate": 0.0005, "loss": 1.7810313701629639, "step": 10458 }, { "epoch": 1.2170118687456366, "grad_norm": 1.2896366119384766, "learning_rate": 0.0005, "loss": 1.221789002418518, "step": 10459 }, { "epoch": 1.2171282289969747, "grad_norm": 1.5716936588287354, "learning_rate": 0.0005, "loss": 2.0631961822509766, "step": 10460 }, { "epoch": 1.2172445892483128, "grad_norm": 2.017409563064575, "learning_rate": 0.0005, "loss": 2.119821786880493, "step": 10461 }, { "epoch": 1.217360949499651, "grad_norm": 3.2965424060821533, "learning_rate": 0.0005, "loss": 1.9359912872314453, "step": 10462 }, { "epoch": 1.217477309750989, "grad_norm": 0.8697468042373657, "learning_rate": 0.0005, "loss": 2.0831692218780518, "step": 10463 }, { "epoch": 1.2175936700023273, "grad_norm": 2.0204122066497803, "learning_rate": 0.0005, "loss": 2.1938846111297607, "step": 10464 }, { "epoch": 1.2177100302536654, "grad_norm": 1.029181957244873, "learning_rate": 0.0005, "loss": 1.7075448036193848, "step": 10465 }, { "epoch": 1.2178263905050035, "grad_norm": 1.3664711713790894, "learning_rate": 0.0005, "loss": 2.365255117416382, "step": 10466 }, { "epoch": 1.2179427507563416, "grad_norm": 1.0803951025009155, "learning_rate": 0.0005, "loss": 1.7259879112243652, "step": 10467 }, { "epoch": 1.2180591110076797, "grad_norm": 0.9662436842918396, "learning_rate": 0.0005, "loss": 2.3153598308563232, "step": 10468 }, { "epoch": 1.218175471259018, "grad_norm": 1.1729676723480225, "learning_rate": 0.0005, "loss": 1.8918712139129639, "step": 10469 }, { "epoch": 1.2182918315103561, "grad_norm": 1.7527810335159302, "learning_rate": 0.0005, "loss": 2.146106481552124, "step": 10470 }, { "epoch": 1.2184081917616942, "grad_norm": 1.1322734355926514, "learning_rate": 0.0005, "loss": 2.512629985809326, "step": 10471 }, { "epoch": 1.2185245520130323, "grad_norm": 1.099434733390808, "learning_rate": 0.0005, "loss": 2.0040035247802734, "step": 10472 }, { "epoch": 1.2186409122643704, "grad_norm": 1.0756381750106812, "learning_rate": 0.0005, "loss": 2.4816765785217285, "step": 10473 }, { "epoch": 1.2187572725157088, "grad_norm": 1.2678966522216797, "learning_rate": 0.0005, "loss": 3.3574788570404053, "step": 10474 }, { "epoch": 1.2188736327670469, "grad_norm": 1.1143913269042969, "learning_rate": 0.0005, "loss": 2.1231493949890137, "step": 10475 }, { "epoch": 1.218989993018385, "grad_norm": 0.6369521021842957, "learning_rate": 0.0005, "loss": 1.6680241823196411, "step": 10476 }, { "epoch": 1.219106353269723, "grad_norm": 1.152558445930481, "learning_rate": 0.0005, "loss": 1.5835416316986084, "step": 10477 }, { "epoch": 1.2192227135210612, "grad_norm": 1.160557746887207, "learning_rate": 0.0005, "loss": 2.4013781547546387, "step": 10478 }, { "epoch": 1.2193390737723995, "grad_norm": 1.1225272417068481, "learning_rate": 0.0005, "loss": 1.3825275897979736, "step": 10479 }, { "epoch": 1.2194554340237376, "grad_norm": 1.010054111480713, "learning_rate": 0.0005, "loss": 1.803912878036499, "step": 10480 }, { "epoch": 1.2195717942750757, "grad_norm": 1.317000389099121, "learning_rate": 0.0005, "loss": 2.110137939453125, "step": 10481 }, { "epoch": 1.2196881545264138, "grad_norm": 0.9692714810371399, "learning_rate": 0.0005, "loss": 1.1413004398345947, "step": 10482 }, { "epoch": 1.2198045147777519, "grad_norm": 1.100303292274475, "learning_rate": 0.0005, "loss": 2.437323570251465, "step": 10483 }, { "epoch": 1.21992087502909, "grad_norm": 1.2283737659454346, "learning_rate": 0.0005, "loss": 1.65447998046875, "step": 10484 }, { "epoch": 1.2200372352804283, "grad_norm": 1.2862999439239502, "learning_rate": 0.0005, "loss": 2.524996280670166, "step": 10485 }, { "epoch": 1.2201535955317664, "grad_norm": 1.5916684865951538, "learning_rate": 0.0005, "loss": 2.036562919616699, "step": 10486 }, { "epoch": 1.2202699557831045, "grad_norm": 1.293800711631775, "learning_rate": 0.0005, "loss": 2.1756067276000977, "step": 10487 }, { "epoch": 1.2203863160344426, "grad_norm": 1.794132113456726, "learning_rate": 0.0005, "loss": 2.183021306991577, "step": 10488 }, { "epoch": 1.2205026762857807, "grad_norm": 1.066373348236084, "learning_rate": 0.0005, "loss": 1.7949504852294922, "step": 10489 }, { "epoch": 1.220619036537119, "grad_norm": 0.7413613200187683, "learning_rate": 0.0005, "loss": 1.578911542892456, "step": 10490 }, { "epoch": 1.220735396788457, "grad_norm": 1.6255625486373901, "learning_rate": 0.0005, "loss": 2.007628917694092, "step": 10491 }, { "epoch": 1.2208517570397952, "grad_norm": 3.1884443759918213, "learning_rate": 0.0005, "loss": 1.9625663757324219, "step": 10492 }, { "epoch": 1.2209681172911333, "grad_norm": 1.1321507692337036, "learning_rate": 0.0005, "loss": 2.320410966873169, "step": 10493 }, { "epoch": 1.2210844775424714, "grad_norm": 2.0943591594696045, "learning_rate": 0.0005, "loss": 1.9994217157363892, "step": 10494 }, { "epoch": 1.2212008377938097, "grad_norm": 1.3325010538101196, "learning_rate": 0.0005, "loss": 2.139110803604126, "step": 10495 }, { "epoch": 1.2213171980451478, "grad_norm": 1.2730894088745117, "learning_rate": 0.0005, "loss": 1.3010201454162598, "step": 10496 }, { "epoch": 1.221433558296486, "grad_norm": 1.2376528978347778, "learning_rate": 0.0005, "loss": 2.4355950355529785, "step": 10497 }, { "epoch": 1.221549918547824, "grad_norm": 1.5222162008285522, "learning_rate": 0.0005, "loss": 2.6496877670288086, "step": 10498 }, { "epoch": 1.2216662787991621, "grad_norm": 0.9386829137802124, "learning_rate": 0.0005, "loss": 1.9276593923568726, "step": 10499 }, { "epoch": 1.2217826390505004, "grad_norm": 1.2743685245513916, "learning_rate": 0.0005, "loss": 1.5720114707946777, "step": 10500 }, { "epoch": 1.2218989993018385, "grad_norm": 1.136067509651184, "learning_rate": 0.0005, "loss": 1.729670524597168, "step": 10501 }, { "epoch": 1.2220153595531766, "grad_norm": 2.3542709350585938, "learning_rate": 0.0005, "loss": 1.8539528846740723, "step": 10502 }, { "epoch": 1.2221317198045147, "grad_norm": 1.236571192741394, "learning_rate": 0.0005, "loss": 1.7165002822875977, "step": 10503 }, { "epoch": 1.2222480800558528, "grad_norm": 1.3584986925125122, "learning_rate": 0.0005, "loss": 1.473103642463684, "step": 10504 }, { "epoch": 1.2223644403071912, "grad_norm": 1.640668272972107, "learning_rate": 0.0005, "loss": 2.003398895263672, "step": 10505 }, { "epoch": 1.2224808005585293, "grad_norm": 2.37709641456604, "learning_rate": 0.0005, "loss": 2.047459840774536, "step": 10506 }, { "epoch": 1.2225971608098674, "grad_norm": 1.5528218746185303, "learning_rate": 0.0005, "loss": 2.5840282440185547, "step": 10507 }, { "epoch": 1.2227135210612055, "grad_norm": 0.9147042036056519, "learning_rate": 0.0005, "loss": 1.6153644323349, "step": 10508 }, { "epoch": 1.2228298813125436, "grad_norm": 1.299268364906311, "learning_rate": 0.0005, "loss": 1.9985437393188477, "step": 10509 }, { "epoch": 1.2229462415638819, "grad_norm": 1.5860116481781006, "learning_rate": 0.0005, "loss": 2.2392945289611816, "step": 10510 }, { "epoch": 1.22306260181522, "grad_norm": 0.7453405261039734, "learning_rate": 0.0005, "loss": 1.6407662630081177, "step": 10511 }, { "epoch": 1.223178962066558, "grad_norm": 0.9350400567054749, "learning_rate": 0.0005, "loss": 1.7022037506103516, "step": 10512 }, { "epoch": 1.2232953223178962, "grad_norm": 1.302735447883606, "learning_rate": 0.0005, "loss": 2.0652804374694824, "step": 10513 }, { "epoch": 1.2234116825692343, "grad_norm": 2.8509719371795654, "learning_rate": 0.0005, "loss": 2.0306758880615234, "step": 10514 }, { "epoch": 1.2235280428205724, "grad_norm": 0.985202968120575, "learning_rate": 0.0005, "loss": 1.905517816543579, "step": 10515 }, { "epoch": 1.2236444030719107, "grad_norm": 0.9496476054191589, "learning_rate": 0.0005, "loss": 1.6428382396697998, "step": 10516 }, { "epoch": 1.2237607633232488, "grad_norm": 1.0233290195465088, "learning_rate": 0.0005, "loss": 1.6160669326782227, "step": 10517 }, { "epoch": 1.223877123574587, "grad_norm": 1.372199535369873, "learning_rate": 0.0005, "loss": 2.488537311553955, "step": 10518 }, { "epoch": 1.223993483825925, "grad_norm": 1.3624783754348755, "learning_rate": 0.0005, "loss": 1.8312971591949463, "step": 10519 }, { "epoch": 1.224109844077263, "grad_norm": 1.2689346075057983, "learning_rate": 0.0005, "loss": 1.8563470840454102, "step": 10520 }, { "epoch": 1.2242262043286014, "grad_norm": 1.0961707830429077, "learning_rate": 0.0005, "loss": 1.8340259790420532, "step": 10521 }, { "epoch": 1.2243425645799395, "grad_norm": 1.0304365158081055, "learning_rate": 0.0005, "loss": 1.7178592681884766, "step": 10522 }, { "epoch": 1.2244589248312776, "grad_norm": 1.5092581510543823, "learning_rate": 0.0005, "loss": 2.030806541442871, "step": 10523 }, { "epoch": 1.2245752850826157, "grad_norm": 0.8468675017356873, "learning_rate": 0.0005, "loss": 1.930159568786621, "step": 10524 }, { "epoch": 1.2246916453339538, "grad_norm": 0.8709773421287537, "learning_rate": 0.0005, "loss": 1.1532939672470093, "step": 10525 }, { "epoch": 1.2248080055852921, "grad_norm": 1.0518457889556885, "learning_rate": 0.0005, "loss": 1.541756272315979, "step": 10526 }, { "epoch": 1.2249243658366302, "grad_norm": 0.7377468347549438, "learning_rate": 0.0005, "loss": 1.7319109439849854, "step": 10527 }, { "epoch": 1.2250407260879683, "grad_norm": 1.2666983604431152, "learning_rate": 0.0005, "loss": 2.6434059143066406, "step": 10528 }, { "epoch": 1.2251570863393064, "grad_norm": 1.1084038019180298, "learning_rate": 0.0005, "loss": 2.723093032836914, "step": 10529 }, { "epoch": 1.2252734465906445, "grad_norm": 1.6747616529464722, "learning_rate": 0.0005, "loss": 2.4983367919921875, "step": 10530 }, { "epoch": 1.2253898068419828, "grad_norm": 0.8019246459007263, "learning_rate": 0.0005, "loss": 1.274432897567749, "step": 10531 }, { "epoch": 1.225506167093321, "grad_norm": 1.3663780689239502, "learning_rate": 0.0005, "loss": 1.6675859689712524, "step": 10532 }, { "epoch": 1.225622527344659, "grad_norm": 1.5443490743637085, "learning_rate": 0.0005, "loss": 2.2293143272399902, "step": 10533 }, { "epoch": 1.2257388875959971, "grad_norm": 0.9129527807235718, "learning_rate": 0.0005, "loss": 1.8257051706314087, "step": 10534 }, { "epoch": 1.2258552478473352, "grad_norm": 0.803429901599884, "learning_rate": 0.0005, "loss": 1.2870595455169678, "step": 10535 }, { "epoch": 1.2259716080986736, "grad_norm": 0.4566649794578552, "learning_rate": 0.0005, "loss": 1.0361660718917847, "step": 10536 }, { "epoch": 1.2260879683500117, "grad_norm": 1.3723704814910889, "learning_rate": 0.0005, "loss": 2.326686143875122, "step": 10537 }, { "epoch": 1.2262043286013498, "grad_norm": 1.1840550899505615, "learning_rate": 0.0005, "loss": 2.331308364868164, "step": 10538 }, { "epoch": 1.2263206888526879, "grad_norm": 0.7647539973258972, "learning_rate": 0.0005, "loss": 1.3797874450683594, "step": 10539 }, { "epoch": 1.226437049104026, "grad_norm": 0.7755753993988037, "learning_rate": 0.0005, "loss": 1.795701026916504, "step": 10540 }, { "epoch": 1.2265534093553643, "grad_norm": 0.9830312728881836, "learning_rate": 0.0005, "loss": 1.7055461406707764, "step": 10541 }, { "epoch": 1.2266697696067024, "grad_norm": 1.8648622035980225, "learning_rate": 0.0005, "loss": 2.902271032333374, "step": 10542 }, { "epoch": 1.2267861298580405, "grad_norm": 0.9308325052261353, "learning_rate": 0.0005, "loss": 1.6047484874725342, "step": 10543 }, { "epoch": 1.2269024901093786, "grad_norm": 0.8869661688804626, "learning_rate": 0.0005, "loss": 1.307309627532959, "step": 10544 }, { "epoch": 1.2270188503607167, "grad_norm": 0.7585084438323975, "learning_rate": 0.0005, "loss": 1.8555774688720703, "step": 10545 }, { "epoch": 1.227135210612055, "grad_norm": 1.1129809617996216, "learning_rate": 0.0005, "loss": 1.5780527591705322, "step": 10546 }, { "epoch": 1.227251570863393, "grad_norm": 1.1846754550933838, "learning_rate": 0.0005, "loss": 1.9160676002502441, "step": 10547 }, { "epoch": 1.2273679311147312, "grad_norm": 0.9390838146209717, "learning_rate": 0.0005, "loss": 1.9065766334533691, "step": 10548 }, { "epoch": 1.2274842913660693, "grad_norm": 1.3339686393737793, "learning_rate": 0.0005, "loss": 1.8991373777389526, "step": 10549 }, { "epoch": 1.2276006516174074, "grad_norm": 1.0772137641906738, "learning_rate": 0.0005, "loss": 2.0638067722320557, "step": 10550 }, { "epoch": 1.2277170118687457, "grad_norm": 0.8628330230712891, "learning_rate": 0.0005, "loss": 1.7258225679397583, "step": 10551 }, { "epoch": 1.2278333721200838, "grad_norm": 1.0189980268478394, "learning_rate": 0.0005, "loss": 1.4717721939086914, "step": 10552 }, { "epoch": 1.227949732371422, "grad_norm": 0.9790884852409363, "learning_rate": 0.0005, "loss": 1.6528096199035645, "step": 10553 }, { "epoch": 1.22806609262276, "grad_norm": 1.511123776435852, "learning_rate": 0.0005, "loss": 2.7654523849487305, "step": 10554 }, { "epoch": 1.228182452874098, "grad_norm": 0.8731256127357483, "learning_rate": 0.0005, "loss": 1.9276856184005737, "step": 10555 }, { "epoch": 1.2282988131254364, "grad_norm": 1.0525637865066528, "learning_rate": 0.0005, "loss": 2.2012619972229004, "step": 10556 }, { "epoch": 1.2284151733767745, "grad_norm": 1.0502569675445557, "learning_rate": 0.0005, "loss": 1.8504135608673096, "step": 10557 }, { "epoch": 1.2285315336281126, "grad_norm": 1.0219578742980957, "learning_rate": 0.0005, "loss": 2.044602394104004, "step": 10558 }, { "epoch": 1.2286478938794507, "grad_norm": 1.0972288846969604, "learning_rate": 0.0005, "loss": 2.179960250854492, "step": 10559 }, { "epoch": 1.2287642541307888, "grad_norm": 0.9717214107513428, "learning_rate": 0.0005, "loss": 1.962622880935669, "step": 10560 }, { "epoch": 1.2288806143821271, "grad_norm": 1.1743539571762085, "learning_rate": 0.0005, "loss": 2.369752883911133, "step": 10561 }, { "epoch": 1.2289969746334652, "grad_norm": 0.896686315536499, "learning_rate": 0.0005, "loss": 1.5735459327697754, "step": 10562 }, { "epoch": 1.2291133348848033, "grad_norm": 1.4970144033432007, "learning_rate": 0.0005, "loss": 1.890181541442871, "step": 10563 }, { "epoch": 1.2292296951361414, "grad_norm": 1.7351244688034058, "learning_rate": 0.0005, "loss": 2.670252799987793, "step": 10564 }, { "epoch": 1.2293460553874795, "grad_norm": 0.9009762406349182, "learning_rate": 0.0005, "loss": 1.6617815494537354, "step": 10565 }, { "epoch": 1.2294624156388179, "grad_norm": 1.1985081434249878, "learning_rate": 0.0005, "loss": 2.129622459411621, "step": 10566 }, { "epoch": 1.229578775890156, "grad_norm": 1.2678980827331543, "learning_rate": 0.0005, "loss": 2.8197383880615234, "step": 10567 }, { "epoch": 1.229695136141494, "grad_norm": 1.5034149885177612, "learning_rate": 0.0005, "loss": 1.7272143363952637, "step": 10568 }, { "epoch": 1.2298114963928322, "grad_norm": 1.4754787683486938, "learning_rate": 0.0005, "loss": 2.018749237060547, "step": 10569 }, { "epoch": 1.2299278566441703, "grad_norm": 1.4600249528884888, "learning_rate": 0.0005, "loss": 2.566769599914551, "step": 10570 }, { "epoch": 1.2300442168955086, "grad_norm": 1.369888424873352, "learning_rate": 0.0005, "loss": 2.5481162071228027, "step": 10571 }, { "epoch": 1.2301605771468467, "grad_norm": 0.9078864455223083, "learning_rate": 0.0005, "loss": 2.031170606613159, "step": 10572 }, { "epoch": 1.2302769373981848, "grad_norm": 1.364242434501648, "learning_rate": 0.0005, "loss": 2.5231051445007324, "step": 10573 }, { "epoch": 1.2303932976495229, "grad_norm": 1.2704511880874634, "learning_rate": 0.0005, "loss": 1.5065743923187256, "step": 10574 }, { "epoch": 1.230509657900861, "grad_norm": 0.9194371104240417, "learning_rate": 0.0005, "loss": 2.5719852447509766, "step": 10575 }, { "epoch": 1.2306260181521993, "grad_norm": 1.0102717876434326, "learning_rate": 0.0005, "loss": 2.3339648246765137, "step": 10576 }, { "epoch": 1.2307423784035374, "grad_norm": 0.9810610413551331, "learning_rate": 0.0005, "loss": 1.9012401103973389, "step": 10577 }, { "epoch": 1.2308587386548755, "grad_norm": 1.5742146968841553, "learning_rate": 0.0005, "loss": 2.3165581226348877, "step": 10578 }, { "epoch": 1.2309750989062136, "grad_norm": 1.264999270439148, "learning_rate": 0.0005, "loss": 1.9168413877487183, "step": 10579 }, { "epoch": 1.2310914591575517, "grad_norm": 0.8432804942131042, "learning_rate": 0.0005, "loss": 1.8179693222045898, "step": 10580 }, { "epoch": 1.23120781940889, "grad_norm": 1.0557528734207153, "learning_rate": 0.0005, "loss": 2.151578187942505, "step": 10581 }, { "epoch": 1.2313241796602281, "grad_norm": 1.2547855377197266, "learning_rate": 0.0005, "loss": 1.7311089038848877, "step": 10582 }, { "epoch": 1.2314405399115662, "grad_norm": 0.5194126963615417, "learning_rate": 0.0005, "loss": 1.177873969078064, "step": 10583 }, { "epoch": 1.2315569001629043, "grad_norm": 0.9850232601165771, "learning_rate": 0.0005, "loss": 2.254899024963379, "step": 10584 }, { "epoch": 1.2316732604142424, "grad_norm": 0.9136621356010437, "learning_rate": 0.0005, "loss": 1.9568394422531128, "step": 10585 }, { "epoch": 1.2317896206655807, "grad_norm": 1.0083650350570679, "learning_rate": 0.0005, "loss": 1.8174986839294434, "step": 10586 }, { "epoch": 1.2319059809169188, "grad_norm": 1.3728593587875366, "learning_rate": 0.0005, "loss": 2.098860740661621, "step": 10587 }, { "epoch": 1.232022341168257, "grad_norm": 1.136602759361267, "learning_rate": 0.0005, "loss": 1.7992291450500488, "step": 10588 }, { "epoch": 1.232138701419595, "grad_norm": 9.932621955871582, "learning_rate": 0.0005, "loss": 2.1626362800598145, "step": 10589 }, { "epoch": 1.2322550616709331, "grad_norm": 1.338680386543274, "learning_rate": 0.0005, "loss": 2.6315813064575195, "step": 10590 }, { "epoch": 1.2323714219222714, "grad_norm": 0.888100266456604, "learning_rate": 0.0005, "loss": 2.1714601516723633, "step": 10591 }, { "epoch": 1.2324877821736095, "grad_norm": 1.9019447565078735, "learning_rate": 0.0005, "loss": 3.0057902336120605, "step": 10592 }, { "epoch": 1.2326041424249476, "grad_norm": 0.9023690223693848, "learning_rate": 0.0005, "loss": 2.1250414848327637, "step": 10593 }, { "epoch": 1.2327205026762857, "grad_norm": 1.2503999471664429, "learning_rate": 0.0005, "loss": 1.5898241996765137, "step": 10594 }, { "epoch": 1.2328368629276238, "grad_norm": 1.2144041061401367, "learning_rate": 0.0005, "loss": 2.5176548957824707, "step": 10595 }, { "epoch": 1.2329532231789622, "grad_norm": 1.0503138303756714, "learning_rate": 0.0005, "loss": 1.7747063636779785, "step": 10596 }, { "epoch": 1.2330695834303003, "grad_norm": 2.4390487670898438, "learning_rate": 0.0005, "loss": 2.5972347259521484, "step": 10597 }, { "epoch": 1.2331859436816384, "grad_norm": 1.240062952041626, "learning_rate": 0.0005, "loss": 1.6034016609191895, "step": 10598 }, { "epoch": 1.2333023039329765, "grad_norm": 1.1068164110183716, "learning_rate": 0.0005, "loss": 1.8488540649414062, "step": 10599 }, { "epoch": 1.2334186641843146, "grad_norm": 1.1215049028396606, "learning_rate": 0.0005, "loss": 2.13991641998291, "step": 10600 }, { "epoch": 1.2335350244356529, "grad_norm": 1.1983850002288818, "learning_rate": 0.0005, "loss": 1.7625995874404907, "step": 10601 }, { "epoch": 1.233651384686991, "grad_norm": 1.04949951171875, "learning_rate": 0.0005, "loss": 2.203113555908203, "step": 10602 }, { "epoch": 1.233767744938329, "grad_norm": 1.7387144565582275, "learning_rate": 0.0005, "loss": 1.5301963090896606, "step": 10603 }, { "epoch": 1.2338841051896672, "grad_norm": 0.9263464212417603, "learning_rate": 0.0005, "loss": 1.884187936782837, "step": 10604 }, { "epoch": 1.2340004654410053, "grad_norm": 0.9334235191345215, "learning_rate": 0.0005, "loss": 2.1431312561035156, "step": 10605 }, { "epoch": 1.2341168256923436, "grad_norm": 0.7816097736358643, "learning_rate": 0.0005, "loss": 1.518441915512085, "step": 10606 }, { "epoch": 1.2342331859436817, "grad_norm": 1.0650807619094849, "learning_rate": 0.0005, "loss": 1.6861870288848877, "step": 10607 }, { "epoch": 1.2343495461950198, "grad_norm": 1.6246856451034546, "learning_rate": 0.0005, "loss": 2.6578614711761475, "step": 10608 }, { "epoch": 1.234465906446358, "grad_norm": 0.9400997161865234, "learning_rate": 0.0005, "loss": 1.968318223953247, "step": 10609 }, { "epoch": 1.234582266697696, "grad_norm": 1.1693885326385498, "learning_rate": 0.0005, "loss": 2.2314696311950684, "step": 10610 }, { "epoch": 1.2346986269490343, "grad_norm": 0.803843080997467, "learning_rate": 0.0005, "loss": 1.7979986667633057, "step": 10611 }, { "epoch": 1.2348149872003724, "grad_norm": 0.5636619329452515, "learning_rate": 0.0005, "loss": 1.1483564376831055, "step": 10612 }, { "epoch": 1.2349313474517105, "grad_norm": 1.1776221990585327, "learning_rate": 0.0005, "loss": 2.2634124755859375, "step": 10613 }, { "epoch": 1.2350477077030486, "grad_norm": 0.9145669937133789, "learning_rate": 0.0005, "loss": 1.38772714138031, "step": 10614 }, { "epoch": 1.2351640679543867, "grad_norm": 1.699491262435913, "learning_rate": 0.0005, "loss": 2.051365375518799, "step": 10615 }, { "epoch": 1.235280428205725, "grad_norm": 0.890022873878479, "learning_rate": 0.0005, "loss": 1.3570936918258667, "step": 10616 }, { "epoch": 1.2353967884570631, "grad_norm": 1.3214008808135986, "learning_rate": 0.0005, "loss": 1.723008155822754, "step": 10617 }, { "epoch": 1.2355131487084012, "grad_norm": 1.3229668140411377, "learning_rate": 0.0005, "loss": 1.4817882776260376, "step": 10618 }, { "epoch": 1.2356295089597393, "grad_norm": 0.9429250359535217, "learning_rate": 0.0005, "loss": 1.9645562171936035, "step": 10619 }, { "epoch": 1.2357458692110774, "grad_norm": 0.9307271838188171, "learning_rate": 0.0005, "loss": 2.0825648307800293, "step": 10620 }, { "epoch": 1.2358622294624158, "grad_norm": 2.2349557876586914, "learning_rate": 0.0005, "loss": 2.0914483070373535, "step": 10621 }, { "epoch": 1.2359785897137539, "grad_norm": 0.9479789137840271, "learning_rate": 0.0005, "loss": 1.4663467407226562, "step": 10622 }, { "epoch": 1.236094949965092, "grad_norm": 1.086710810661316, "learning_rate": 0.0005, "loss": 1.7543916702270508, "step": 10623 }, { "epoch": 1.23621131021643, "grad_norm": 0.8877580165863037, "learning_rate": 0.0005, "loss": 1.9935386180877686, "step": 10624 }, { "epoch": 1.2363276704677681, "grad_norm": 0.8191852569580078, "learning_rate": 0.0005, "loss": 2.0804390907287598, "step": 10625 }, { "epoch": 1.2364440307191065, "grad_norm": 1.194007158279419, "learning_rate": 0.0005, "loss": 2.2116482257843018, "step": 10626 }, { "epoch": 1.2365603909704446, "grad_norm": 1.399574637413025, "learning_rate": 0.0005, "loss": 2.0388662815093994, "step": 10627 }, { "epoch": 1.2366767512217827, "grad_norm": 1.33253812789917, "learning_rate": 0.0005, "loss": 1.8400123119354248, "step": 10628 }, { "epoch": 1.2367931114731208, "grad_norm": 1.4425106048583984, "learning_rate": 0.0005, "loss": 1.6156078577041626, "step": 10629 }, { "epoch": 1.2369094717244589, "grad_norm": 1.8031303882598877, "learning_rate": 0.0005, "loss": 1.9392757415771484, "step": 10630 }, { "epoch": 1.2370258319757972, "grad_norm": 1.2713665962219238, "learning_rate": 0.0005, "loss": 1.9083322286605835, "step": 10631 }, { "epoch": 1.2371421922271353, "grad_norm": 1.2755413055419922, "learning_rate": 0.0005, "loss": 2.0452017784118652, "step": 10632 }, { "epoch": 1.2372585524784734, "grad_norm": 1.808545470237732, "learning_rate": 0.0005, "loss": 2.255502223968506, "step": 10633 }, { "epoch": 1.2373749127298115, "grad_norm": 1.1694759130477905, "learning_rate": 0.0005, "loss": 2.1363067626953125, "step": 10634 }, { "epoch": 1.2374912729811496, "grad_norm": 1.057582139968872, "learning_rate": 0.0005, "loss": 2.5557188987731934, "step": 10635 }, { "epoch": 1.237607633232488, "grad_norm": 1.891402006149292, "learning_rate": 0.0005, "loss": 2.150165557861328, "step": 10636 }, { "epoch": 1.237723993483826, "grad_norm": 1.6830053329467773, "learning_rate": 0.0005, "loss": 2.670194149017334, "step": 10637 }, { "epoch": 1.237840353735164, "grad_norm": 1.037611722946167, "learning_rate": 0.0005, "loss": 1.705291748046875, "step": 10638 }, { "epoch": 1.2379567139865022, "grad_norm": 0.9923727512359619, "learning_rate": 0.0005, "loss": 2.0888524055480957, "step": 10639 }, { "epoch": 1.2380730742378403, "grad_norm": 0.7803424000740051, "learning_rate": 0.0005, "loss": 2.2467334270477295, "step": 10640 }, { "epoch": 1.2381894344891786, "grad_norm": 1.4633679389953613, "learning_rate": 0.0005, "loss": 1.736403226852417, "step": 10641 }, { "epoch": 1.2383057947405167, "grad_norm": 1.3840912580490112, "learning_rate": 0.0005, "loss": 1.842054843902588, "step": 10642 }, { "epoch": 1.2384221549918548, "grad_norm": 0.9696892499923706, "learning_rate": 0.0005, "loss": 1.5880135297775269, "step": 10643 }, { "epoch": 1.238538515243193, "grad_norm": 0.9318137168884277, "learning_rate": 0.0005, "loss": 1.8710938692092896, "step": 10644 }, { "epoch": 1.238654875494531, "grad_norm": 1.6923387050628662, "learning_rate": 0.0005, "loss": 2.1603798866271973, "step": 10645 }, { "epoch": 1.2387712357458691, "grad_norm": 1.2079559564590454, "learning_rate": 0.0005, "loss": 2.0749282836914062, "step": 10646 }, { "epoch": 1.2388875959972074, "grad_norm": 1.6791088581085205, "learning_rate": 0.0005, "loss": 1.9486134052276611, "step": 10647 }, { "epoch": 1.2390039562485455, "grad_norm": 1.082169532775879, "learning_rate": 0.0005, "loss": 1.5791468620300293, "step": 10648 }, { "epoch": 1.2391203164998836, "grad_norm": 0.965815007686615, "learning_rate": 0.0005, "loss": 1.7906194925308228, "step": 10649 }, { "epoch": 1.2392366767512217, "grad_norm": 1.0933783054351807, "learning_rate": 0.0005, "loss": 1.9128541946411133, "step": 10650 }, { "epoch": 1.2393530370025598, "grad_norm": 1.4072922468185425, "learning_rate": 0.0005, "loss": 2.9560914039611816, "step": 10651 }, { "epoch": 1.2394693972538982, "grad_norm": 1.0081771612167358, "learning_rate": 0.0005, "loss": 1.9633204936981201, "step": 10652 }, { "epoch": 1.2395857575052363, "grad_norm": 1.0325947999954224, "learning_rate": 0.0005, "loss": 1.5675086975097656, "step": 10653 }, { "epoch": 1.2397021177565744, "grad_norm": 1.1893972158432007, "learning_rate": 0.0005, "loss": 2.0797948837280273, "step": 10654 }, { "epoch": 1.2398184780079125, "grad_norm": 1.0815637111663818, "learning_rate": 0.0005, "loss": 2.619859218597412, "step": 10655 }, { "epoch": 1.2399348382592505, "grad_norm": 1.4273982048034668, "learning_rate": 0.0005, "loss": 1.837040662765503, "step": 10656 }, { "epoch": 1.2400511985105889, "grad_norm": 2.0146260261535645, "learning_rate": 0.0005, "loss": 2.0618484020233154, "step": 10657 }, { "epoch": 1.240167558761927, "grad_norm": 1.4266114234924316, "learning_rate": 0.0005, "loss": 2.6141161918640137, "step": 10658 }, { "epoch": 1.240283919013265, "grad_norm": 1.1720572710037231, "learning_rate": 0.0005, "loss": 2.2606871128082275, "step": 10659 }, { "epoch": 1.2404002792646032, "grad_norm": 0.8846367597579956, "learning_rate": 0.0005, "loss": 1.5244477987289429, "step": 10660 }, { "epoch": 1.2405166395159413, "grad_norm": 1.28355073928833, "learning_rate": 0.0005, "loss": 2.727750301361084, "step": 10661 }, { "epoch": 1.2406329997672796, "grad_norm": 1.1881663799285889, "learning_rate": 0.0005, "loss": 1.7373151779174805, "step": 10662 }, { "epoch": 1.2407493600186177, "grad_norm": 1.2182676792144775, "learning_rate": 0.0005, "loss": 1.6725925207138062, "step": 10663 }, { "epoch": 1.2408657202699558, "grad_norm": 1.0120083093643188, "learning_rate": 0.0005, "loss": 2.106504201889038, "step": 10664 }, { "epoch": 1.2409820805212939, "grad_norm": 1.8951663970947266, "learning_rate": 0.0005, "loss": 1.8863435983657837, "step": 10665 }, { "epoch": 1.241098440772632, "grad_norm": 0.8730097413063049, "learning_rate": 0.0005, "loss": 1.6147856712341309, "step": 10666 }, { "epoch": 1.2412148010239703, "grad_norm": 0.9774824380874634, "learning_rate": 0.0005, "loss": 2.0038065910339355, "step": 10667 }, { "epoch": 1.2413311612753084, "grad_norm": 1.2094601392745972, "learning_rate": 0.0005, "loss": 1.9061261415481567, "step": 10668 }, { "epoch": 1.2414475215266465, "grad_norm": 1.1258227825164795, "learning_rate": 0.0005, "loss": 1.467590093612671, "step": 10669 }, { "epoch": 1.2415638817779846, "grad_norm": 0.9420789480209351, "learning_rate": 0.0005, "loss": 1.763461947441101, "step": 10670 }, { "epoch": 1.2416802420293227, "grad_norm": 0.6858430504798889, "learning_rate": 0.0005, "loss": 1.4819703102111816, "step": 10671 }, { "epoch": 1.241796602280661, "grad_norm": 1.1447114944458008, "learning_rate": 0.0005, "loss": 1.8038064241409302, "step": 10672 }, { "epoch": 1.2419129625319991, "grad_norm": 0.942231297492981, "learning_rate": 0.0005, "loss": 1.8729349374771118, "step": 10673 }, { "epoch": 1.2420293227833372, "grad_norm": 1.100998044013977, "learning_rate": 0.0005, "loss": 2.246750831604004, "step": 10674 }, { "epoch": 1.2421456830346753, "grad_norm": 1.5131428241729736, "learning_rate": 0.0005, "loss": 3.3262805938720703, "step": 10675 }, { "epoch": 1.2422620432860134, "grad_norm": 1.4055922031402588, "learning_rate": 0.0005, "loss": 2.2585015296936035, "step": 10676 }, { "epoch": 1.2423784035373515, "grad_norm": 0.9310246109962463, "learning_rate": 0.0005, "loss": 1.755237340927124, "step": 10677 }, { "epoch": 1.2424947637886898, "grad_norm": 0.8963416218757629, "learning_rate": 0.0005, "loss": 1.4499101638793945, "step": 10678 }, { "epoch": 1.242611124040028, "grad_norm": 1.4287196397781372, "learning_rate": 0.0005, "loss": 2.7283618450164795, "step": 10679 }, { "epoch": 1.242727484291366, "grad_norm": 0.6185722947120667, "learning_rate": 0.0005, "loss": 1.0529077053070068, "step": 10680 }, { "epoch": 1.2428438445427041, "grad_norm": 1.4978872537612915, "learning_rate": 0.0005, "loss": 2.073793649673462, "step": 10681 }, { "epoch": 1.2429602047940422, "grad_norm": 0.8119426369667053, "learning_rate": 0.0005, "loss": 1.6583465337753296, "step": 10682 }, { "epoch": 1.2430765650453806, "grad_norm": 1.2283275127410889, "learning_rate": 0.0005, "loss": 2.157804489135742, "step": 10683 }, { "epoch": 1.2431929252967187, "grad_norm": 1.3082671165466309, "learning_rate": 0.0005, "loss": 1.8461108207702637, "step": 10684 }, { "epoch": 1.2433092855480568, "grad_norm": 1.2107950448989868, "learning_rate": 0.0005, "loss": 2.4983720779418945, "step": 10685 }, { "epoch": 1.2434256457993949, "grad_norm": 1.5430527925491333, "learning_rate": 0.0005, "loss": 1.5770153999328613, "step": 10686 }, { "epoch": 1.243542006050733, "grad_norm": 0.7716774344444275, "learning_rate": 0.0005, "loss": 1.7151668071746826, "step": 10687 }, { "epoch": 1.2436583663020713, "grad_norm": 1.5816994905471802, "learning_rate": 0.0005, "loss": 1.8669676780700684, "step": 10688 }, { "epoch": 1.2437747265534094, "grad_norm": 1.1082583665847778, "learning_rate": 0.0005, "loss": 2.8741276264190674, "step": 10689 }, { "epoch": 1.2438910868047475, "grad_norm": 1.527762770652771, "learning_rate": 0.0005, "loss": 2.4300734996795654, "step": 10690 }, { "epoch": 1.2440074470560856, "grad_norm": 1.4216322898864746, "learning_rate": 0.0005, "loss": 1.8315846920013428, "step": 10691 }, { "epoch": 1.2441238073074237, "grad_norm": 0.8706177473068237, "learning_rate": 0.0005, "loss": 1.9421026706695557, "step": 10692 }, { "epoch": 1.244240167558762, "grad_norm": 1.7105882167816162, "learning_rate": 0.0005, "loss": 2.762700319290161, "step": 10693 }, { "epoch": 1.2443565278101, "grad_norm": 1.7280153036117554, "learning_rate": 0.0005, "loss": 2.3255553245544434, "step": 10694 }, { "epoch": 1.2444728880614382, "grad_norm": 1.144067406654358, "learning_rate": 0.0005, "loss": 2.135058879852295, "step": 10695 }, { "epoch": 1.2445892483127763, "grad_norm": 1.252049446105957, "learning_rate": 0.0005, "loss": 1.807680606842041, "step": 10696 }, { "epoch": 1.2447056085641144, "grad_norm": 1.1476277112960815, "learning_rate": 0.0005, "loss": 2.0801196098327637, "step": 10697 }, { "epoch": 1.2448219688154527, "grad_norm": 0.9203476905822754, "learning_rate": 0.0005, "loss": 2.0039525032043457, "step": 10698 }, { "epoch": 1.2449383290667908, "grad_norm": 0.8783137798309326, "learning_rate": 0.0005, "loss": 2.2203376293182373, "step": 10699 }, { "epoch": 1.245054689318129, "grad_norm": 1.0495296716690063, "learning_rate": 0.0005, "loss": 2.1955933570861816, "step": 10700 }, { "epoch": 1.245171049569467, "grad_norm": 0.7573689818382263, "learning_rate": 0.0005, "loss": 1.7984085083007812, "step": 10701 }, { "epoch": 1.245287409820805, "grad_norm": 2.4941635131835938, "learning_rate": 0.0005, "loss": 2.497530937194824, "step": 10702 }, { "epoch": 1.2454037700721434, "grad_norm": 1.7983005046844482, "learning_rate": 0.0005, "loss": 2.2247695922851562, "step": 10703 }, { "epoch": 1.2455201303234815, "grad_norm": 1.1886540651321411, "learning_rate": 0.0005, "loss": 1.849385380744934, "step": 10704 }, { "epoch": 1.2456364905748196, "grad_norm": 1.1791765689849854, "learning_rate": 0.0005, "loss": 1.7710769176483154, "step": 10705 }, { "epoch": 1.2457528508261577, "grad_norm": 1.2109981775283813, "learning_rate": 0.0005, "loss": 2.2882699966430664, "step": 10706 }, { "epoch": 1.2458692110774958, "grad_norm": 0.9066065549850464, "learning_rate": 0.0005, "loss": 1.850663423538208, "step": 10707 }, { "epoch": 1.2459855713288341, "grad_norm": 1.3799195289611816, "learning_rate": 0.0005, "loss": 2.0853431224823, "step": 10708 }, { "epoch": 1.2461019315801722, "grad_norm": 1.583990454673767, "learning_rate": 0.0005, "loss": 2.785825252532959, "step": 10709 }, { "epoch": 1.2462182918315103, "grad_norm": 1.2505896091461182, "learning_rate": 0.0005, "loss": 2.37164306640625, "step": 10710 }, { "epoch": 1.2463346520828484, "grad_norm": 0.9791228771209717, "learning_rate": 0.0005, "loss": 2.017240524291992, "step": 10711 }, { "epoch": 1.2464510123341865, "grad_norm": 0.9519674181938171, "learning_rate": 0.0005, "loss": 1.8898489475250244, "step": 10712 }, { "epoch": 1.2465673725855249, "grad_norm": 1.3666822910308838, "learning_rate": 0.0005, "loss": 1.9780043363571167, "step": 10713 }, { "epoch": 1.246683732836863, "grad_norm": 11.458443641662598, "learning_rate": 0.0005, "loss": 2.888423442840576, "step": 10714 }, { "epoch": 1.246800093088201, "grad_norm": 1.0902273654937744, "learning_rate": 0.0005, "loss": 1.7410399913787842, "step": 10715 }, { "epoch": 1.2469164533395392, "grad_norm": 0.7720955014228821, "learning_rate": 0.0005, "loss": 1.3199758529663086, "step": 10716 }, { "epoch": 1.2470328135908773, "grad_norm": 1.203869104385376, "learning_rate": 0.0005, "loss": 1.7412421703338623, "step": 10717 }, { "epoch": 1.2471491738422156, "grad_norm": 0.791059136390686, "learning_rate": 0.0005, "loss": 1.8418798446655273, "step": 10718 }, { "epoch": 1.2472655340935537, "grad_norm": 0.938888669013977, "learning_rate": 0.0005, "loss": 1.4885311126708984, "step": 10719 }, { "epoch": 1.2473818943448918, "grad_norm": 1.1155023574829102, "learning_rate": 0.0005, "loss": 1.5685367584228516, "step": 10720 }, { "epoch": 1.2474982545962299, "grad_norm": 0.9705535173416138, "learning_rate": 0.0005, "loss": 2.102059841156006, "step": 10721 }, { "epoch": 1.247614614847568, "grad_norm": 1.1524043083190918, "learning_rate": 0.0005, "loss": 1.8795796632766724, "step": 10722 }, { "epoch": 1.2477309750989063, "grad_norm": 2.008716344833374, "learning_rate": 0.0005, "loss": 2.131988763809204, "step": 10723 }, { "epoch": 1.2478473353502444, "grad_norm": 1.0413485765457153, "learning_rate": 0.0005, "loss": 1.7177047729492188, "step": 10724 }, { "epoch": 1.2479636956015825, "grad_norm": 1.2927299737930298, "learning_rate": 0.0005, "loss": 1.624929428100586, "step": 10725 }, { "epoch": 1.2480800558529206, "grad_norm": 0.8215634226799011, "learning_rate": 0.0005, "loss": 1.1569634675979614, "step": 10726 }, { "epoch": 1.2481964161042587, "grad_norm": 0.7417754530906677, "learning_rate": 0.0005, "loss": 1.4854943752288818, "step": 10727 }, { "epoch": 1.248312776355597, "grad_norm": 0.9166190028190613, "learning_rate": 0.0005, "loss": 2.470574378967285, "step": 10728 }, { "epoch": 1.248429136606935, "grad_norm": 1.035698652267456, "learning_rate": 0.0005, "loss": 1.9486701488494873, "step": 10729 }, { "epoch": 1.2485454968582732, "grad_norm": 0.8587124943733215, "learning_rate": 0.0005, "loss": 2.1205050945281982, "step": 10730 }, { "epoch": 1.2486618571096113, "grad_norm": 1.2332383394241333, "learning_rate": 0.0005, "loss": 2.4972715377807617, "step": 10731 }, { "epoch": 1.2487782173609494, "grad_norm": 1.064106822013855, "learning_rate": 0.0005, "loss": 2.070939064025879, "step": 10732 }, { "epoch": 1.2488945776122877, "grad_norm": 1.1373977661132812, "learning_rate": 0.0005, "loss": 2.37766695022583, "step": 10733 }, { "epoch": 1.2490109378636258, "grad_norm": 0.9773362874984741, "learning_rate": 0.0005, "loss": 1.4132922887802124, "step": 10734 }, { "epoch": 1.249127298114964, "grad_norm": 1.290120005607605, "learning_rate": 0.0005, "loss": 1.8215274810791016, "step": 10735 }, { "epoch": 1.249243658366302, "grad_norm": 1.5954474210739136, "learning_rate": 0.0005, "loss": 1.826418399810791, "step": 10736 }, { "epoch": 1.2493600186176401, "grad_norm": 1.1137903928756714, "learning_rate": 0.0005, "loss": 1.7620797157287598, "step": 10737 }, { "epoch": 1.2494763788689784, "grad_norm": 1.0888051986694336, "learning_rate": 0.0005, "loss": 1.7874813079833984, "step": 10738 }, { "epoch": 1.2495927391203165, "grad_norm": 0.8482913970947266, "learning_rate": 0.0005, "loss": 1.9603965282440186, "step": 10739 }, { "epoch": 1.2497090993716546, "grad_norm": 0.8064431548118591, "learning_rate": 0.0005, "loss": 2.20316481590271, "step": 10740 }, { "epoch": 1.2498254596229927, "grad_norm": 3.9406015872955322, "learning_rate": 0.0005, "loss": 1.9747642278671265, "step": 10741 }, { "epoch": 1.2499418198743308, "grad_norm": 1.293616771697998, "learning_rate": 0.0005, "loss": 2.488983154296875, "step": 10742 }, { "epoch": 1.2500581801256692, "grad_norm": 0.9252203702926636, "learning_rate": 0.0005, "loss": 2.1724584102630615, "step": 10743 }, { "epoch": 1.2501745403770073, "grad_norm": 0.9726313352584839, "learning_rate": 0.0005, "loss": 1.9951249361038208, "step": 10744 }, { "epoch": 1.2502909006283454, "grad_norm": 1.5021566152572632, "learning_rate": 0.0005, "loss": 2.0595669746398926, "step": 10745 }, { "epoch": 1.2504072608796835, "grad_norm": 1.0407346487045288, "learning_rate": 0.0005, "loss": 2.04594087600708, "step": 10746 }, { "epoch": 1.2505236211310216, "grad_norm": 1.1324572563171387, "learning_rate": 0.0005, "loss": 2.289559841156006, "step": 10747 }, { "epoch": 1.2506399813823599, "grad_norm": 0.8461036086082458, "learning_rate": 0.0005, "loss": 2.374612808227539, "step": 10748 }, { "epoch": 1.250756341633698, "grad_norm": 1.0347940921783447, "learning_rate": 0.0005, "loss": 2.2082433700561523, "step": 10749 }, { "epoch": 1.250872701885036, "grad_norm": 1.4383772611618042, "learning_rate": 0.0005, "loss": 1.7109826803207397, "step": 10750 }, { "epoch": 1.2509890621363742, "grad_norm": 1.1522151231765747, "learning_rate": 0.0005, "loss": 2.596895933151245, "step": 10751 }, { "epoch": 1.2511054223877123, "grad_norm": 0.8548830151557922, "learning_rate": 0.0005, "loss": 1.4900662899017334, "step": 10752 }, { "epoch": 1.2512217826390506, "grad_norm": 1.424659013748169, "learning_rate": 0.0005, "loss": 2.791733741760254, "step": 10753 }, { "epoch": 1.2513381428903887, "grad_norm": 0.762791097164154, "learning_rate": 0.0005, "loss": 1.8366913795471191, "step": 10754 }, { "epoch": 1.2514545031417268, "grad_norm": 1.2086181640625, "learning_rate": 0.0005, "loss": 2.1744213104248047, "step": 10755 }, { "epoch": 1.251570863393065, "grad_norm": 0.7562536001205444, "learning_rate": 0.0005, "loss": 1.3098710775375366, "step": 10756 }, { "epoch": 1.251687223644403, "grad_norm": 0.9622958302497864, "learning_rate": 0.0005, "loss": 1.8874502182006836, "step": 10757 }, { "epoch": 1.2518035838957413, "grad_norm": 0.7428821921348572, "learning_rate": 0.0005, "loss": 1.8719418048858643, "step": 10758 }, { "epoch": 1.2519199441470794, "grad_norm": 1.6100671291351318, "learning_rate": 0.0005, "loss": 2.295886516571045, "step": 10759 }, { "epoch": 1.2520363043984175, "grad_norm": 1.131281852722168, "learning_rate": 0.0005, "loss": 2.0090763568878174, "step": 10760 }, { "epoch": 1.2521526646497556, "grad_norm": 0.8465152382850647, "learning_rate": 0.0005, "loss": 1.2019069194793701, "step": 10761 }, { "epoch": 1.2522690249010937, "grad_norm": 1.6020792722702026, "learning_rate": 0.0005, "loss": 2.7345900535583496, "step": 10762 }, { "epoch": 1.252385385152432, "grad_norm": 1.2315412759780884, "learning_rate": 0.0005, "loss": 1.5871282815933228, "step": 10763 }, { "epoch": 1.2525017454037701, "grad_norm": 0.6948709487915039, "learning_rate": 0.0005, "loss": 1.5113365650177002, "step": 10764 }, { "epoch": 1.2526181056551082, "grad_norm": 1.0071797370910645, "learning_rate": 0.0005, "loss": 1.5249617099761963, "step": 10765 }, { "epoch": 1.2527344659064463, "grad_norm": 0.9697378873825073, "learning_rate": 0.0005, "loss": 2.613804340362549, "step": 10766 }, { "epoch": 1.2528508261577844, "grad_norm": 1.0542198419570923, "learning_rate": 0.0005, "loss": 2.06347393989563, "step": 10767 }, { "epoch": 1.2529671864091227, "grad_norm": 1.2321423292160034, "learning_rate": 0.0005, "loss": 1.592813491821289, "step": 10768 }, { "epoch": 1.2530835466604608, "grad_norm": 0.9845329523086548, "learning_rate": 0.0005, "loss": 1.7349073886871338, "step": 10769 }, { "epoch": 1.253199906911799, "grad_norm": 1.2024115324020386, "learning_rate": 0.0005, "loss": 2.2804617881774902, "step": 10770 }, { "epoch": 1.253316267163137, "grad_norm": 1.175663948059082, "learning_rate": 0.0005, "loss": 2.1043436527252197, "step": 10771 }, { "epoch": 1.2534326274144751, "grad_norm": 0.6183297038078308, "learning_rate": 0.0005, "loss": 1.7141788005828857, "step": 10772 }, { "epoch": 1.2535489876658135, "grad_norm": 1.2363014221191406, "learning_rate": 0.0005, "loss": 1.4771246910095215, "step": 10773 }, { "epoch": 1.2536653479171516, "grad_norm": 0.9642414450645447, "learning_rate": 0.0005, "loss": 2.1458418369293213, "step": 10774 }, { "epoch": 1.2537817081684897, "grad_norm": 0.7936850786209106, "learning_rate": 0.0005, "loss": 1.8290648460388184, "step": 10775 }, { "epoch": 1.2538980684198278, "grad_norm": 1.5439237356185913, "learning_rate": 0.0005, "loss": 2.1437907218933105, "step": 10776 }, { "epoch": 1.2540144286711659, "grad_norm": 0.8139208555221558, "learning_rate": 0.0005, "loss": 2.088493824005127, "step": 10777 }, { "epoch": 1.2541307889225042, "grad_norm": 1.102912425994873, "learning_rate": 0.0005, "loss": 2.276052474975586, "step": 10778 }, { "epoch": 1.2542471491738423, "grad_norm": 0.9027138948440552, "learning_rate": 0.0005, "loss": 1.8323345184326172, "step": 10779 }, { "epoch": 1.2543635094251804, "grad_norm": 1.820221185684204, "learning_rate": 0.0005, "loss": 1.9600341320037842, "step": 10780 }, { "epoch": 1.2544798696765185, "grad_norm": 0.9025999903678894, "learning_rate": 0.0005, "loss": 1.5460231304168701, "step": 10781 }, { "epoch": 1.2545962299278566, "grad_norm": 0.8800266981124878, "learning_rate": 0.0005, "loss": 2.0665810108184814, "step": 10782 }, { "epoch": 1.254712590179195, "grad_norm": 0.6513659358024597, "learning_rate": 0.0005, "loss": 1.328381061553955, "step": 10783 }, { "epoch": 1.254828950430533, "grad_norm": 1.3504196405410767, "learning_rate": 0.0005, "loss": 1.8260555267333984, "step": 10784 }, { "epoch": 1.254945310681871, "grad_norm": 1.3057235479354858, "learning_rate": 0.0005, "loss": 1.9834827184677124, "step": 10785 }, { "epoch": 1.2550616709332092, "grad_norm": 1.169983983039856, "learning_rate": 0.0005, "loss": 1.8227626085281372, "step": 10786 }, { "epoch": 1.2551780311845473, "grad_norm": 1.1451282501220703, "learning_rate": 0.0005, "loss": 1.8504600524902344, "step": 10787 }, { "epoch": 1.2552943914358856, "grad_norm": 1.204629898071289, "learning_rate": 0.0005, "loss": 2.3389029502868652, "step": 10788 }, { "epoch": 1.2554107516872237, "grad_norm": 1.200154423713684, "learning_rate": 0.0005, "loss": 1.7957395315170288, "step": 10789 }, { "epoch": 1.2555271119385618, "grad_norm": 0.7842393517494202, "learning_rate": 0.0005, "loss": 1.4797595739364624, "step": 10790 }, { "epoch": 1.2556434721899, "grad_norm": 0.9013908505439758, "learning_rate": 0.0005, "loss": 1.4821677207946777, "step": 10791 }, { "epoch": 1.255759832441238, "grad_norm": 1.2986191511154175, "learning_rate": 0.0005, "loss": 2.211843252182007, "step": 10792 }, { "epoch": 1.2558761926925763, "grad_norm": 1.1701637506484985, "learning_rate": 0.0005, "loss": 2.3505916595458984, "step": 10793 }, { "epoch": 1.2559925529439144, "grad_norm": 0.9784746766090393, "learning_rate": 0.0005, "loss": 2.000521659851074, "step": 10794 }, { "epoch": 1.2561089131952525, "grad_norm": 0.9335739016532898, "learning_rate": 0.0005, "loss": 1.6411601305007935, "step": 10795 }, { "epoch": 1.2562252734465906, "grad_norm": 1.1329314708709717, "learning_rate": 0.0005, "loss": 1.833336353302002, "step": 10796 }, { "epoch": 1.2563416336979287, "grad_norm": 1.1258639097213745, "learning_rate": 0.0005, "loss": 2.418792486190796, "step": 10797 }, { "epoch": 1.256457993949267, "grad_norm": 2.6906380653381348, "learning_rate": 0.0005, "loss": 1.7826056480407715, "step": 10798 }, { "epoch": 1.2565743542006051, "grad_norm": 0.9328961372375488, "learning_rate": 0.0005, "loss": 2.30678129196167, "step": 10799 }, { "epoch": 1.2566907144519432, "grad_norm": 0.8483196496963501, "learning_rate": 0.0005, "loss": 2.430872678756714, "step": 10800 }, { "epoch": 1.2568070747032813, "grad_norm": 1.1994946002960205, "learning_rate": 0.0005, "loss": 1.1578214168548584, "step": 10801 }, { "epoch": 1.2569234349546194, "grad_norm": 1.1093603372573853, "learning_rate": 0.0005, "loss": 2.0280420780181885, "step": 10802 }, { "epoch": 1.2570397952059578, "grad_norm": 1.2644948959350586, "learning_rate": 0.0005, "loss": 1.7320773601531982, "step": 10803 }, { "epoch": 1.2571561554572959, "grad_norm": 0.8941187858581543, "learning_rate": 0.0005, "loss": 2.6990432739257812, "step": 10804 }, { "epoch": 1.257272515708634, "grad_norm": 1.1531606912612915, "learning_rate": 0.0005, "loss": 2.194221258163452, "step": 10805 }, { "epoch": 1.257388875959972, "grad_norm": 1.017342448234558, "learning_rate": 0.0005, "loss": 1.6895360946655273, "step": 10806 }, { "epoch": 1.2575052362113102, "grad_norm": 1.0452969074249268, "learning_rate": 0.0005, "loss": 2.151000738143921, "step": 10807 }, { "epoch": 1.2576215964626485, "grad_norm": 1.1688016653060913, "learning_rate": 0.0005, "loss": 1.50986909866333, "step": 10808 }, { "epoch": 1.2577379567139866, "grad_norm": 0.6953150033950806, "learning_rate": 0.0005, "loss": 1.579459547996521, "step": 10809 }, { "epoch": 1.2578543169653247, "grad_norm": 1.5074321031570435, "learning_rate": 0.0005, "loss": 1.7418776750564575, "step": 10810 }, { "epoch": 1.2579706772166628, "grad_norm": 0.9744343757629395, "learning_rate": 0.0005, "loss": 2.039276123046875, "step": 10811 }, { "epoch": 1.2580870374680009, "grad_norm": 1.6505004167556763, "learning_rate": 0.0005, "loss": 1.9137474298477173, "step": 10812 }, { "epoch": 1.2582033977193392, "grad_norm": 1.7615176439285278, "learning_rate": 0.0005, "loss": 2.301356315612793, "step": 10813 }, { "epoch": 1.2583197579706773, "grad_norm": 1.3485146760940552, "learning_rate": 0.0005, "loss": 2.3214190006256104, "step": 10814 }, { "epoch": 1.2584361182220154, "grad_norm": 0.5422003269195557, "learning_rate": 0.0005, "loss": 1.1087877750396729, "step": 10815 }, { "epoch": 1.2585524784733535, "grad_norm": 1.1227158308029175, "learning_rate": 0.0005, "loss": 1.7194418907165527, "step": 10816 }, { "epoch": 1.2586688387246916, "grad_norm": 0.9596409201622009, "learning_rate": 0.0005, "loss": 1.872048258781433, "step": 10817 }, { "epoch": 1.25878519897603, "grad_norm": 1.162835717201233, "learning_rate": 0.0005, "loss": 1.6566259860992432, "step": 10818 }, { "epoch": 1.2589015592273678, "grad_norm": 1.0755679607391357, "learning_rate": 0.0005, "loss": 2.7308785915374756, "step": 10819 }, { "epoch": 1.2590179194787061, "grad_norm": 0.898806095123291, "learning_rate": 0.0005, "loss": 1.4678523540496826, "step": 10820 }, { "epoch": 1.2591342797300442, "grad_norm": 1.6302310228347778, "learning_rate": 0.0005, "loss": 1.947554111480713, "step": 10821 }, { "epoch": 1.2592506399813823, "grad_norm": 0.9294211864471436, "learning_rate": 0.0005, "loss": 2.103734254837036, "step": 10822 }, { "epoch": 1.2593670002327206, "grad_norm": 0.5123127698898315, "learning_rate": 0.0005, "loss": 1.0186986923217773, "step": 10823 }, { "epoch": 1.2594833604840585, "grad_norm": 1.1395368576049805, "learning_rate": 0.0005, "loss": 1.5030157566070557, "step": 10824 }, { "epoch": 1.2595997207353968, "grad_norm": 0.5713496208190918, "learning_rate": 0.0005, "loss": 1.4487109184265137, "step": 10825 }, { "epoch": 1.259716080986735, "grad_norm": 1.3798109292984009, "learning_rate": 0.0005, "loss": 2.570922613143921, "step": 10826 }, { "epoch": 1.259832441238073, "grad_norm": 0.9075044393539429, "learning_rate": 0.0005, "loss": 2.210387706756592, "step": 10827 }, { "epoch": 1.2599488014894114, "grad_norm": 1.3945233821868896, "learning_rate": 0.0005, "loss": 2.3491735458374023, "step": 10828 }, { "epoch": 1.2600651617407492, "grad_norm": 2.402723550796509, "learning_rate": 0.0005, "loss": 1.7483158111572266, "step": 10829 }, { "epoch": 1.2601815219920875, "grad_norm": 1.1656301021575928, "learning_rate": 0.0005, "loss": 2.262251377105713, "step": 10830 }, { "epoch": 1.2602978822434256, "grad_norm": 0.970143735408783, "learning_rate": 0.0005, "loss": 1.561525821685791, "step": 10831 }, { "epoch": 1.2604142424947637, "grad_norm": 1.3127154111862183, "learning_rate": 0.0005, "loss": 2.432464122772217, "step": 10832 }, { "epoch": 1.260530602746102, "grad_norm": 1.1535320281982422, "learning_rate": 0.0005, "loss": 1.8810853958129883, "step": 10833 }, { "epoch": 1.26064696299744, "grad_norm": 1.010544776916504, "learning_rate": 0.0005, "loss": 1.5915820598602295, "step": 10834 }, { "epoch": 1.2607633232487783, "grad_norm": 1.0780532360076904, "learning_rate": 0.0005, "loss": 1.9111078977584839, "step": 10835 }, { "epoch": 1.2608796835001164, "grad_norm": 1.1661638021469116, "learning_rate": 0.0005, "loss": 2.006978988647461, "step": 10836 }, { "epoch": 1.2609960437514545, "grad_norm": 1.1471151113510132, "learning_rate": 0.0005, "loss": 2.3764166831970215, "step": 10837 }, { "epoch": 1.2611124040027926, "grad_norm": 1.2004609107971191, "learning_rate": 0.0005, "loss": 2.393688201904297, "step": 10838 }, { "epoch": 1.2612287642541307, "grad_norm": 0.8065153360366821, "learning_rate": 0.0005, "loss": 2.2208151817321777, "step": 10839 }, { "epoch": 1.261345124505469, "grad_norm": 1.1365336179733276, "learning_rate": 0.0005, "loss": 1.4456400871276855, "step": 10840 }, { "epoch": 1.261461484756807, "grad_norm": 1.2730953693389893, "learning_rate": 0.0005, "loss": 2.3917195796966553, "step": 10841 }, { "epoch": 1.2615778450081452, "grad_norm": 0.9399621486663818, "learning_rate": 0.0005, "loss": 1.529982089996338, "step": 10842 }, { "epoch": 1.2616942052594833, "grad_norm": 1.0711405277252197, "learning_rate": 0.0005, "loss": 1.9237632751464844, "step": 10843 }, { "epoch": 1.2618105655108214, "grad_norm": 0.9206799864768982, "learning_rate": 0.0005, "loss": 1.7706289291381836, "step": 10844 }, { "epoch": 1.2619269257621597, "grad_norm": 1.0350241661071777, "learning_rate": 0.0005, "loss": 1.1491488218307495, "step": 10845 }, { "epoch": 1.2620432860134978, "grad_norm": 0.5604760050773621, "learning_rate": 0.0005, "loss": 1.2910207509994507, "step": 10846 }, { "epoch": 1.262159646264836, "grad_norm": 0.9041705131530762, "learning_rate": 0.0005, "loss": 2.0255966186523438, "step": 10847 }, { "epoch": 1.262276006516174, "grad_norm": 1.014450192451477, "learning_rate": 0.0005, "loss": 1.2543480396270752, "step": 10848 }, { "epoch": 1.262392366767512, "grad_norm": 1.1616222858428955, "learning_rate": 0.0005, "loss": 2.102940082550049, "step": 10849 }, { "epoch": 1.2625087270188504, "grad_norm": 0.7052053809165955, "learning_rate": 0.0005, "loss": 1.325563669204712, "step": 10850 }, { "epoch": 1.2626250872701885, "grad_norm": 1.899920105934143, "learning_rate": 0.0005, "loss": 2.34407377243042, "step": 10851 }, { "epoch": 1.2627414475215266, "grad_norm": 1.3417941331863403, "learning_rate": 0.0005, "loss": 1.720426321029663, "step": 10852 }, { "epoch": 1.2628578077728647, "grad_norm": 1.3750413656234741, "learning_rate": 0.0005, "loss": 2.103555679321289, "step": 10853 }, { "epoch": 1.2629741680242028, "grad_norm": 0.7982633709907532, "learning_rate": 0.0005, "loss": 1.9054481983184814, "step": 10854 }, { "epoch": 1.2630905282755411, "grad_norm": 1.8896235227584839, "learning_rate": 0.0005, "loss": 1.8895947933197021, "step": 10855 }, { "epoch": 1.2632068885268792, "grad_norm": 1.1659139394760132, "learning_rate": 0.0005, "loss": 1.707836389541626, "step": 10856 }, { "epoch": 1.2633232487782173, "grad_norm": 1.9398705959320068, "learning_rate": 0.0005, "loss": 2.1882500648498535, "step": 10857 }, { "epoch": 1.2634396090295554, "grad_norm": 1.5682568550109863, "learning_rate": 0.0005, "loss": 2.147120714187622, "step": 10858 }, { "epoch": 1.2635559692808935, "grad_norm": 1.4398683309555054, "learning_rate": 0.0005, "loss": 2.1022403240203857, "step": 10859 }, { "epoch": 1.2636723295322319, "grad_norm": 1.0644999742507935, "learning_rate": 0.0005, "loss": 2.2641351222991943, "step": 10860 }, { "epoch": 1.26378868978357, "grad_norm": 0.7769869565963745, "learning_rate": 0.0005, "loss": 1.455309510231018, "step": 10861 }, { "epoch": 1.263905050034908, "grad_norm": 1.1972715854644775, "learning_rate": 0.0005, "loss": 2.406358242034912, "step": 10862 }, { "epoch": 1.2640214102862461, "grad_norm": 0.9160026907920837, "learning_rate": 0.0005, "loss": 2.1277084350585938, "step": 10863 }, { "epoch": 1.2641377705375842, "grad_norm": 0.7895997762680054, "learning_rate": 0.0005, "loss": 1.3155969381332397, "step": 10864 }, { "epoch": 1.2642541307889226, "grad_norm": 1.2461084127426147, "learning_rate": 0.0005, "loss": 1.5752391815185547, "step": 10865 }, { "epoch": 1.2643704910402607, "grad_norm": 1.0870991945266724, "learning_rate": 0.0005, "loss": 2.0673084259033203, "step": 10866 }, { "epoch": 1.2644868512915988, "grad_norm": 0.852364182472229, "learning_rate": 0.0005, "loss": 1.7849793434143066, "step": 10867 }, { "epoch": 1.2646032115429369, "grad_norm": 0.8306906819343567, "learning_rate": 0.0005, "loss": 1.7588554620742798, "step": 10868 }, { "epoch": 1.264719571794275, "grad_norm": 0.8531947731971741, "learning_rate": 0.0005, "loss": 1.5615071058273315, "step": 10869 }, { "epoch": 1.2648359320456133, "grad_norm": 0.8942677974700928, "learning_rate": 0.0005, "loss": 1.8064179420471191, "step": 10870 }, { "epoch": 1.2649522922969514, "grad_norm": 1.4357056617736816, "learning_rate": 0.0005, "loss": 2.730705738067627, "step": 10871 }, { "epoch": 1.2650686525482895, "grad_norm": 1.0957472324371338, "learning_rate": 0.0005, "loss": 1.6258416175842285, "step": 10872 }, { "epoch": 1.2651850127996276, "grad_norm": 1.1091911792755127, "learning_rate": 0.0005, "loss": 2.162848711013794, "step": 10873 }, { "epoch": 1.2653013730509657, "grad_norm": 1.389946460723877, "learning_rate": 0.0005, "loss": 2.8106160163879395, "step": 10874 }, { "epoch": 1.265417733302304, "grad_norm": 1.272669792175293, "learning_rate": 0.0005, "loss": 1.6688451766967773, "step": 10875 }, { "epoch": 1.265534093553642, "grad_norm": 1.5153456926345825, "learning_rate": 0.0005, "loss": 2.09093976020813, "step": 10876 }, { "epoch": 1.2656504538049802, "grad_norm": 1.4293911457061768, "learning_rate": 0.0005, "loss": 2.0709710121154785, "step": 10877 }, { "epoch": 1.2657668140563183, "grad_norm": 1.4318468570709229, "learning_rate": 0.0005, "loss": 1.576329231262207, "step": 10878 }, { "epoch": 1.2658831743076564, "grad_norm": 0.6585202813148499, "learning_rate": 0.0005, "loss": 1.08507239818573, "step": 10879 }, { "epoch": 1.2659995345589947, "grad_norm": 1.2616561651229858, "learning_rate": 0.0005, "loss": 2.633429527282715, "step": 10880 }, { "epoch": 1.2661158948103328, "grad_norm": 1.5174344778060913, "learning_rate": 0.0005, "loss": 2.135117530822754, "step": 10881 }, { "epoch": 1.266232255061671, "grad_norm": 2.334583044052124, "learning_rate": 0.0005, "loss": 3.1754236221313477, "step": 10882 }, { "epoch": 1.266348615313009, "grad_norm": 0.9901371598243713, "learning_rate": 0.0005, "loss": 2.0861897468566895, "step": 10883 }, { "epoch": 1.2664649755643471, "grad_norm": 1.2184789180755615, "learning_rate": 0.0005, "loss": 3.048276424407959, "step": 10884 }, { "epoch": 1.2665813358156854, "grad_norm": 1.0657802820205688, "learning_rate": 0.0005, "loss": 1.5226545333862305, "step": 10885 }, { "epoch": 1.2666976960670235, "grad_norm": 1.3759721517562866, "learning_rate": 0.0005, "loss": 2.4903526306152344, "step": 10886 }, { "epoch": 1.2668140563183616, "grad_norm": 0.7391406297683716, "learning_rate": 0.0005, "loss": 1.97274911403656, "step": 10887 }, { "epoch": 1.2669304165696997, "grad_norm": 1.3891535997390747, "learning_rate": 0.0005, "loss": 1.9462273120880127, "step": 10888 }, { "epoch": 1.2670467768210378, "grad_norm": 1.0485261678695679, "learning_rate": 0.0005, "loss": 1.680767297744751, "step": 10889 }, { "epoch": 1.2671631370723762, "grad_norm": 1.658693552017212, "learning_rate": 0.0005, "loss": 2.234771251678467, "step": 10890 }, { "epoch": 1.2672794973237143, "grad_norm": 0.9684266448020935, "learning_rate": 0.0005, "loss": 2.6083056926727295, "step": 10891 }, { "epoch": 1.2673958575750524, "grad_norm": 1.1486327648162842, "learning_rate": 0.0005, "loss": 1.9722681045532227, "step": 10892 }, { "epoch": 1.2675122178263905, "grad_norm": 1.2332024574279785, "learning_rate": 0.0005, "loss": 1.652959942817688, "step": 10893 }, { "epoch": 1.2676285780777286, "grad_norm": 1.7323096990585327, "learning_rate": 0.0005, "loss": 2.578552722930908, "step": 10894 }, { "epoch": 1.2677449383290669, "grad_norm": 1.1343798637390137, "learning_rate": 0.0005, "loss": 2.484502077102661, "step": 10895 }, { "epoch": 1.267861298580405, "grad_norm": 1.011356234550476, "learning_rate": 0.0005, "loss": 1.6003986597061157, "step": 10896 }, { "epoch": 1.267977658831743, "grad_norm": 1.184365153312683, "learning_rate": 0.0005, "loss": 2.066749334335327, "step": 10897 }, { "epoch": 1.2680940190830812, "grad_norm": 1.3390597105026245, "learning_rate": 0.0005, "loss": 1.9855808019638062, "step": 10898 }, { "epoch": 1.2682103793344193, "grad_norm": 0.9622694253921509, "learning_rate": 0.0005, "loss": 1.6014751195907593, "step": 10899 }, { "epoch": 1.2683267395857576, "grad_norm": 1.352283000946045, "learning_rate": 0.0005, "loss": 2.1966285705566406, "step": 10900 }, { "epoch": 1.2684430998370957, "grad_norm": 1.7489449977874756, "learning_rate": 0.0005, "loss": 2.452425479888916, "step": 10901 }, { "epoch": 1.2685594600884338, "grad_norm": 1.2584024667739868, "learning_rate": 0.0005, "loss": 1.392282247543335, "step": 10902 }, { "epoch": 1.2686758203397719, "grad_norm": 1.4860849380493164, "learning_rate": 0.0005, "loss": 2.41804838180542, "step": 10903 }, { "epoch": 1.26879218059111, "grad_norm": 1.106600284576416, "learning_rate": 0.0005, "loss": 2.3495421409606934, "step": 10904 }, { "epoch": 1.2689085408424483, "grad_norm": 0.8897204995155334, "learning_rate": 0.0005, "loss": 1.7709498405456543, "step": 10905 }, { "epoch": 1.2690249010937864, "grad_norm": 1.1501777172088623, "learning_rate": 0.0005, "loss": 2.268036365509033, "step": 10906 }, { "epoch": 1.2691412613451245, "grad_norm": 1.7151110172271729, "learning_rate": 0.0005, "loss": 2.087393283843994, "step": 10907 }, { "epoch": 1.2692576215964626, "grad_norm": 1.1385674476623535, "learning_rate": 0.0005, "loss": 1.890616774559021, "step": 10908 }, { "epoch": 1.2693739818478007, "grad_norm": 1.303398847579956, "learning_rate": 0.0005, "loss": 2.192110300064087, "step": 10909 }, { "epoch": 1.269490342099139, "grad_norm": 1.3128522634506226, "learning_rate": 0.0005, "loss": 1.2675985097885132, "step": 10910 }, { "epoch": 1.2696067023504771, "grad_norm": 0.7769038677215576, "learning_rate": 0.0005, "loss": 2.6146059036254883, "step": 10911 }, { "epoch": 1.2697230626018152, "grad_norm": 1.152052879333496, "learning_rate": 0.0005, "loss": 2.1862289905548096, "step": 10912 }, { "epoch": 1.2698394228531533, "grad_norm": 1.1077510118484497, "learning_rate": 0.0005, "loss": 1.8995970487594604, "step": 10913 }, { "epoch": 1.2699557831044914, "grad_norm": 1.019486427307129, "learning_rate": 0.0005, "loss": 1.9658164978027344, "step": 10914 }, { "epoch": 1.2700721433558297, "grad_norm": 1.6731270551681519, "learning_rate": 0.0005, "loss": 2.3912572860717773, "step": 10915 }, { "epoch": 1.2701885036071678, "grad_norm": 1.4301036596298218, "learning_rate": 0.0005, "loss": 1.8208122253417969, "step": 10916 }, { "epoch": 1.270304863858506, "grad_norm": 2.6268792152404785, "learning_rate": 0.0005, "loss": 1.9753520488739014, "step": 10917 }, { "epoch": 1.270421224109844, "grad_norm": 0.6791011691093445, "learning_rate": 0.0005, "loss": 2.0168590545654297, "step": 10918 }, { "epoch": 1.2705375843611821, "grad_norm": 1.1315959692001343, "learning_rate": 0.0005, "loss": 2.003288507461548, "step": 10919 }, { "epoch": 1.2706539446125205, "grad_norm": 0.9174259305000305, "learning_rate": 0.0005, "loss": 1.3632317781448364, "step": 10920 }, { "epoch": 1.2707703048638586, "grad_norm": 1.343131184577942, "learning_rate": 0.0005, "loss": 1.8259649276733398, "step": 10921 }, { "epoch": 1.2708866651151967, "grad_norm": 1.2057050466537476, "learning_rate": 0.0005, "loss": 1.3505289554595947, "step": 10922 }, { "epoch": 1.2710030253665348, "grad_norm": 1.4390991926193237, "learning_rate": 0.0005, "loss": 1.9679330587387085, "step": 10923 }, { "epoch": 1.2711193856178729, "grad_norm": 0.8198075890541077, "learning_rate": 0.0005, "loss": 2.3547680377960205, "step": 10924 }, { "epoch": 1.2712357458692112, "grad_norm": 1.003391146659851, "learning_rate": 0.0005, "loss": 1.7380530834197998, "step": 10925 }, { "epoch": 1.2713521061205493, "grad_norm": 1.1680455207824707, "learning_rate": 0.0005, "loss": 1.9190753698349, "step": 10926 }, { "epoch": 1.2714684663718874, "grad_norm": 1.5516074895858765, "learning_rate": 0.0005, "loss": 1.7390391826629639, "step": 10927 }, { "epoch": 1.2715848266232255, "grad_norm": 1.5982378721237183, "learning_rate": 0.0005, "loss": 2.2877020835876465, "step": 10928 }, { "epoch": 1.2717011868745636, "grad_norm": 1.9920127391815186, "learning_rate": 0.0005, "loss": 2.013455390930176, "step": 10929 }, { "epoch": 1.271817547125902, "grad_norm": 1.0484511852264404, "learning_rate": 0.0005, "loss": 1.748866319656372, "step": 10930 }, { "epoch": 1.27193390737724, "grad_norm": 0.7988470792770386, "learning_rate": 0.0005, "loss": 1.280669927597046, "step": 10931 }, { "epoch": 1.272050267628578, "grad_norm": 1.0362646579742432, "learning_rate": 0.0005, "loss": 1.7671799659729004, "step": 10932 }, { "epoch": 1.2721666278799162, "grad_norm": 1.3331987857818604, "learning_rate": 0.0005, "loss": 2.0539755821228027, "step": 10933 }, { "epoch": 1.2722829881312543, "grad_norm": 0.7970008850097656, "learning_rate": 0.0005, "loss": 2.254361152648926, "step": 10934 }, { "epoch": 1.2723993483825926, "grad_norm": 1.3469294309616089, "learning_rate": 0.0005, "loss": 2.3893327713012695, "step": 10935 }, { "epoch": 1.2725157086339307, "grad_norm": 0.9478334188461304, "learning_rate": 0.0005, "loss": 1.8759150505065918, "step": 10936 }, { "epoch": 1.2726320688852688, "grad_norm": 1.3257428407669067, "learning_rate": 0.0005, "loss": 1.9677636623382568, "step": 10937 }, { "epoch": 1.272748429136607, "grad_norm": 1.281353235244751, "learning_rate": 0.0005, "loss": 2.080946683883667, "step": 10938 }, { "epoch": 1.272864789387945, "grad_norm": 0.8603822588920593, "learning_rate": 0.0005, "loss": 2.2432761192321777, "step": 10939 }, { "epoch": 1.2729811496392833, "grad_norm": 1.4377919435501099, "learning_rate": 0.0005, "loss": 2.2235183715820312, "step": 10940 }, { "epoch": 1.2730975098906214, "grad_norm": 1.4015095233917236, "learning_rate": 0.0005, "loss": 2.4417777061462402, "step": 10941 }, { "epoch": 1.2732138701419595, "grad_norm": 1.5388193130493164, "learning_rate": 0.0005, "loss": 1.9837005138397217, "step": 10942 }, { "epoch": 1.2733302303932976, "grad_norm": 2.8504483699798584, "learning_rate": 0.0005, "loss": 2.3802742958068848, "step": 10943 }, { "epoch": 1.2734465906446357, "grad_norm": 1.1503483057022095, "learning_rate": 0.0005, "loss": 2.379761219024658, "step": 10944 }, { "epoch": 1.273562950895974, "grad_norm": 2.271575450897217, "learning_rate": 0.0005, "loss": 1.6333372592926025, "step": 10945 }, { "epoch": 1.2736793111473121, "grad_norm": 1.160601019859314, "learning_rate": 0.0005, "loss": 1.938281536102295, "step": 10946 }, { "epoch": 1.2737956713986502, "grad_norm": 1.5191289186477661, "learning_rate": 0.0005, "loss": 2.19492244720459, "step": 10947 }, { "epoch": 1.2739120316499883, "grad_norm": 1.1615376472473145, "learning_rate": 0.0005, "loss": 1.85281240940094, "step": 10948 }, { "epoch": 1.2740283919013264, "grad_norm": 1.5285565853118896, "learning_rate": 0.0005, "loss": 1.6475164890289307, "step": 10949 }, { "epoch": 1.2741447521526648, "grad_norm": 1.4279614686965942, "learning_rate": 0.0005, "loss": 2.3975048065185547, "step": 10950 }, { "epoch": 1.2742611124040029, "grad_norm": 1.6131523847579956, "learning_rate": 0.0005, "loss": 1.945613980293274, "step": 10951 }, { "epoch": 1.274377472655341, "grad_norm": 1.3414835929870605, "learning_rate": 0.0005, "loss": 2.3148717880249023, "step": 10952 }, { "epoch": 1.274493832906679, "grad_norm": 0.7582719922065735, "learning_rate": 0.0005, "loss": 1.750354290008545, "step": 10953 }, { "epoch": 1.2746101931580172, "grad_norm": 0.7787777781486511, "learning_rate": 0.0005, "loss": 1.4361252784729004, "step": 10954 }, { "epoch": 1.2747265534093555, "grad_norm": 1.0168293714523315, "learning_rate": 0.0005, "loss": 2.8918473720550537, "step": 10955 }, { "epoch": 1.2748429136606936, "grad_norm": 1.0903931856155396, "learning_rate": 0.0005, "loss": 1.9733545780181885, "step": 10956 }, { "epoch": 1.2749592739120317, "grad_norm": 1.1843647956848145, "learning_rate": 0.0005, "loss": 1.7153644561767578, "step": 10957 }, { "epoch": 1.2750756341633698, "grad_norm": 6.284883499145508, "learning_rate": 0.0005, "loss": 2.046560287475586, "step": 10958 }, { "epoch": 1.2751919944147079, "grad_norm": 1.6447175741195679, "learning_rate": 0.0005, "loss": 2.1663479804992676, "step": 10959 }, { "epoch": 1.2753083546660462, "grad_norm": 1.4681226015090942, "learning_rate": 0.0005, "loss": 2.11480712890625, "step": 10960 }, { "epoch": 1.2754247149173843, "grad_norm": 1.1971118450164795, "learning_rate": 0.0005, "loss": 1.8762980699539185, "step": 10961 }, { "epoch": 1.2755410751687224, "grad_norm": 0.7330619692802429, "learning_rate": 0.0005, "loss": 2.053424835205078, "step": 10962 }, { "epoch": 1.2756574354200605, "grad_norm": 1.438230037689209, "learning_rate": 0.0005, "loss": 2.217540740966797, "step": 10963 }, { "epoch": 1.2757737956713986, "grad_norm": 0.6009076833724976, "learning_rate": 0.0005, "loss": 1.4405431747436523, "step": 10964 }, { "epoch": 1.275890155922737, "grad_norm": 1.650202751159668, "learning_rate": 0.0005, "loss": 2.310220718383789, "step": 10965 }, { "epoch": 1.276006516174075, "grad_norm": 0.6977154016494751, "learning_rate": 0.0005, "loss": 1.1174430847167969, "step": 10966 }, { "epoch": 1.276122876425413, "grad_norm": 0.8470923900604248, "learning_rate": 0.0005, "loss": 1.9584261178970337, "step": 10967 }, { "epoch": 1.2762392366767512, "grad_norm": 0.8547884821891785, "learning_rate": 0.0005, "loss": 1.7092223167419434, "step": 10968 }, { "epoch": 1.2763555969280893, "grad_norm": 1.8953144550323486, "learning_rate": 0.0005, "loss": 1.9865388870239258, "step": 10969 }, { "epoch": 1.2764719571794276, "grad_norm": 1.173439383506775, "learning_rate": 0.0005, "loss": 1.828176736831665, "step": 10970 }, { "epoch": 1.2765883174307657, "grad_norm": 0.9141744375228882, "learning_rate": 0.0005, "loss": 2.594372272491455, "step": 10971 }, { "epoch": 1.2767046776821038, "grad_norm": 1.0148136615753174, "learning_rate": 0.0005, "loss": 1.7051007747650146, "step": 10972 }, { "epoch": 1.276821037933442, "grad_norm": 1.3417972326278687, "learning_rate": 0.0005, "loss": 1.9584989547729492, "step": 10973 }, { "epoch": 1.27693739818478, "grad_norm": 1.1568224430084229, "learning_rate": 0.0005, "loss": 2.364525556564331, "step": 10974 }, { "epoch": 1.2770537584361183, "grad_norm": 1.537711501121521, "learning_rate": 0.0005, "loss": 2.1514320373535156, "step": 10975 }, { "epoch": 1.2771701186874564, "grad_norm": 0.7711365222930908, "learning_rate": 0.0005, "loss": 1.298633098602295, "step": 10976 }, { "epoch": 1.2772864789387945, "grad_norm": 0.9740407466888428, "learning_rate": 0.0005, "loss": 1.619814395904541, "step": 10977 }, { "epoch": 1.2774028391901326, "grad_norm": 0.9473000168800354, "learning_rate": 0.0005, "loss": 2.1056487560272217, "step": 10978 }, { "epoch": 1.2775191994414707, "grad_norm": 1.0910433530807495, "learning_rate": 0.0005, "loss": 1.892881989479065, "step": 10979 }, { "epoch": 1.277635559692809, "grad_norm": 1.2232660055160522, "learning_rate": 0.0005, "loss": 2.2768161296844482, "step": 10980 }, { "epoch": 1.277751919944147, "grad_norm": 0.8768250346183777, "learning_rate": 0.0005, "loss": 1.7315860986709595, "step": 10981 }, { "epoch": 1.2778682801954853, "grad_norm": 1.0002119541168213, "learning_rate": 0.0005, "loss": 1.8413196802139282, "step": 10982 }, { "epoch": 1.2779846404468234, "grad_norm": 1.0166419744491577, "learning_rate": 0.0005, "loss": 2.2225143909454346, "step": 10983 }, { "epoch": 1.2781010006981615, "grad_norm": 1.143162488937378, "learning_rate": 0.0005, "loss": 1.866348385810852, "step": 10984 }, { "epoch": 1.2782173609494998, "grad_norm": 2.855402708053589, "learning_rate": 0.0005, "loss": 2.0097873210906982, "step": 10985 }, { "epoch": 1.2783337212008377, "grad_norm": 1.0759986639022827, "learning_rate": 0.0005, "loss": 1.9318469762802124, "step": 10986 }, { "epoch": 1.278450081452176, "grad_norm": 0.9534398317337036, "learning_rate": 0.0005, "loss": 2.3322882652282715, "step": 10987 }, { "epoch": 1.278566441703514, "grad_norm": 1.1736611127853394, "learning_rate": 0.0005, "loss": 2.0704379081726074, "step": 10988 }, { "epoch": 1.2786828019548522, "grad_norm": 1.1953964233398438, "learning_rate": 0.0005, "loss": 2.27897310256958, "step": 10989 }, { "epoch": 1.2787991622061905, "grad_norm": 1.0420018434524536, "learning_rate": 0.0005, "loss": 1.3143205642700195, "step": 10990 }, { "epoch": 1.2789155224575284, "grad_norm": 1.2112934589385986, "learning_rate": 0.0005, "loss": 2.2550041675567627, "step": 10991 }, { "epoch": 1.2790318827088667, "grad_norm": 1.3811122179031372, "learning_rate": 0.0005, "loss": 1.9655039310455322, "step": 10992 }, { "epoch": 1.2791482429602048, "grad_norm": 1.2410048246383667, "learning_rate": 0.0005, "loss": 2.445401668548584, "step": 10993 }, { "epoch": 1.279264603211543, "grad_norm": 1.0672271251678467, "learning_rate": 0.0005, "loss": 1.9714828729629517, "step": 10994 }, { "epoch": 1.2793809634628812, "grad_norm": 1.0602691173553467, "learning_rate": 0.0005, "loss": 1.5485509634017944, "step": 10995 }, { "epoch": 1.279497323714219, "grad_norm": 1.2603402137756348, "learning_rate": 0.0005, "loss": 2.430490016937256, "step": 10996 }, { "epoch": 1.2796136839655574, "grad_norm": 1.574507236480713, "learning_rate": 0.0005, "loss": 2.202857732772827, "step": 10997 }, { "epoch": 1.2797300442168955, "grad_norm": 1.3395968675613403, "learning_rate": 0.0005, "loss": 2.7439842224121094, "step": 10998 }, { "epoch": 1.2798464044682336, "grad_norm": 1.2725645303726196, "learning_rate": 0.0005, "loss": 1.55928635597229, "step": 10999 }, { "epoch": 1.279962764719572, "grad_norm": 1.3728859424591064, "learning_rate": 0.0005, "loss": 2.644777774810791, "step": 11000 }, { "epoch": 1.2800791249709098, "grad_norm": 0.6683398485183716, "learning_rate": 0.0005, "loss": 1.3511052131652832, "step": 11001 }, { "epoch": 1.2801954852222481, "grad_norm": 1.1048457622528076, "learning_rate": 0.0005, "loss": 2.066143274307251, "step": 11002 }, { "epoch": 1.2803118454735862, "grad_norm": 1.115784764289856, "learning_rate": 0.0005, "loss": 2.040163040161133, "step": 11003 }, { "epoch": 1.2804282057249243, "grad_norm": 1.6022084951400757, "learning_rate": 0.0005, "loss": 2.139298439025879, "step": 11004 }, { "epoch": 1.2805445659762624, "grad_norm": 1.5627999305725098, "learning_rate": 0.0005, "loss": 1.8545138835906982, "step": 11005 }, { "epoch": 1.2806609262276005, "grad_norm": 1.3110240697860718, "learning_rate": 0.0005, "loss": 2.6795482635498047, "step": 11006 }, { "epoch": 1.2807772864789388, "grad_norm": 1.4388129711151123, "learning_rate": 0.0005, "loss": 2.294713020324707, "step": 11007 }, { "epoch": 1.280893646730277, "grad_norm": 1.207331657409668, "learning_rate": 0.0005, "loss": 1.930589199066162, "step": 11008 }, { "epoch": 1.281010006981615, "grad_norm": 1.3730517625808716, "learning_rate": 0.0005, "loss": 1.7335808277130127, "step": 11009 }, { "epoch": 1.2811263672329531, "grad_norm": 1.2890667915344238, "learning_rate": 0.0005, "loss": 1.915895700454712, "step": 11010 }, { "epoch": 1.2812427274842912, "grad_norm": 1.3265966176986694, "learning_rate": 0.0005, "loss": 2.6385560035705566, "step": 11011 }, { "epoch": 1.2813590877356296, "grad_norm": 1.1786051988601685, "learning_rate": 0.0005, "loss": 2.0183398723602295, "step": 11012 }, { "epoch": 1.2814754479869677, "grad_norm": 0.9313045740127563, "learning_rate": 0.0005, "loss": 1.948270559310913, "step": 11013 }, { "epoch": 1.2815918082383058, "grad_norm": 0.9687692523002625, "learning_rate": 0.0005, "loss": 1.857085943222046, "step": 11014 }, { "epoch": 1.2817081684896439, "grad_norm": 1.9872649908065796, "learning_rate": 0.0005, "loss": 1.5475776195526123, "step": 11015 }, { "epoch": 1.281824528740982, "grad_norm": 0.8436844348907471, "learning_rate": 0.0005, "loss": 1.3595318794250488, "step": 11016 }, { "epoch": 1.2819408889923203, "grad_norm": 1.1975406408309937, "learning_rate": 0.0005, "loss": 1.7661877870559692, "step": 11017 }, { "epoch": 1.2820572492436584, "grad_norm": 0.6395766139030457, "learning_rate": 0.0005, "loss": 1.2836096286773682, "step": 11018 }, { "epoch": 1.2821736094949965, "grad_norm": 0.8396804928779602, "learning_rate": 0.0005, "loss": 1.303257703781128, "step": 11019 }, { "epoch": 1.2822899697463346, "grad_norm": 1.1028696298599243, "learning_rate": 0.0005, "loss": 2.109938621520996, "step": 11020 }, { "epoch": 1.2824063299976727, "grad_norm": 1.3271018266677856, "learning_rate": 0.0005, "loss": 1.8769707679748535, "step": 11021 }, { "epoch": 1.282522690249011, "grad_norm": 0.9336903691291809, "learning_rate": 0.0005, "loss": 1.9760297536849976, "step": 11022 }, { "epoch": 1.282639050500349, "grad_norm": 1.4539827108383179, "learning_rate": 0.0005, "loss": 1.6879984140396118, "step": 11023 }, { "epoch": 1.2827554107516872, "grad_norm": 1.2387555837631226, "learning_rate": 0.0005, "loss": 1.49698007106781, "step": 11024 }, { "epoch": 1.2828717710030253, "grad_norm": 0.8496610522270203, "learning_rate": 0.0005, "loss": 1.5572881698608398, "step": 11025 }, { "epoch": 1.2829881312543634, "grad_norm": 0.7794922590255737, "learning_rate": 0.0005, "loss": 2.0384647846221924, "step": 11026 }, { "epoch": 1.2831044915057017, "grad_norm": 0.8923865556716919, "learning_rate": 0.0005, "loss": 2.6626195907592773, "step": 11027 }, { "epoch": 1.2832208517570398, "grad_norm": 1.4157249927520752, "learning_rate": 0.0005, "loss": 2.1536850929260254, "step": 11028 }, { "epoch": 1.283337212008378, "grad_norm": 1.10107421875, "learning_rate": 0.0005, "loss": 2.5919203758239746, "step": 11029 }, { "epoch": 1.283453572259716, "grad_norm": 1.508456826210022, "learning_rate": 0.0005, "loss": 2.465628147125244, "step": 11030 }, { "epoch": 1.2835699325110541, "grad_norm": 1.1534111499786377, "learning_rate": 0.0005, "loss": 2.1384756565093994, "step": 11031 }, { "epoch": 1.2836862927623924, "grad_norm": 1.4373047351837158, "learning_rate": 0.0005, "loss": 2.133620262145996, "step": 11032 }, { "epoch": 1.2838026530137305, "grad_norm": 1.0633975267410278, "learning_rate": 0.0005, "loss": 2.1803011894226074, "step": 11033 }, { "epoch": 1.2839190132650686, "grad_norm": 1.1319154500961304, "learning_rate": 0.0005, "loss": 1.8094260692596436, "step": 11034 }, { "epoch": 1.2840353735164067, "grad_norm": 1.5578216314315796, "learning_rate": 0.0005, "loss": 1.298094630241394, "step": 11035 }, { "epoch": 1.2841517337677448, "grad_norm": 1.2823787927627563, "learning_rate": 0.0005, "loss": 1.6646502017974854, "step": 11036 }, { "epoch": 1.2842680940190832, "grad_norm": 1.753293514251709, "learning_rate": 0.0005, "loss": 2.146275043487549, "step": 11037 }, { "epoch": 1.2843844542704212, "grad_norm": 0.7592232823371887, "learning_rate": 0.0005, "loss": 1.9887948036193848, "step": 11038 }, { "epoch": 1.2845008145217593, "grad_norm": 1.3892568349838257, "learning_rate": 0.0005, "loss": 2.4549739360809326, "step": 11039 }, { "epoch": 1.2846171747730974, "grad_norm": 1.3121200799942017, "learning_rate": 0.0005, "loss": 1.739731788635254, "step": 11040 }, { "epoch": 1.2847335350244355, "grad_norm": 1.422210931777954, "learning_rate": 0.0005, "loss": 2.685215473175049, "step": 11041 }, { "epoch": 1.2848498952757739, "grad_norm": 1.4148229360580444, "learning_rate": 0.0005, "loss": 2.746518135070801, "step": 11042 }, { "epoch": 1.284966255527112, "grad_norm": 1.339976191520691, "learning_rate": 0.0005, "loss": 1.3105039596557617, "step": 11043 }, { "epoch": 1.28508261577845, "grad_norm": 0.6942470669746399, "learning_rate": 0.0005, "loss": 1.7003428936004639, "step": 11044 }, { "epoch": 1.2851989760297882, "grad_norm": 1.153382658958435, "learning_rate": 0.0005, "loss": 2.520720958709717, "step": 11045 }, { "epoch": 1.2853153362811263, "grad_norm": 1.4177525043487549, "learning_rate": 0.0005, "loss": 2.1479549407958984, "step": 11046 }, { "epoch": 1.2854316965324646, "grad_norm": 1.1716086864471436, "learning_rate": 0.0005, "loss": 1.919037938117981, "step": 11047 }, { "epoch": 1.2855480567838027, "grad_norm": 0.6003502011299133, "learning_rate": 0.0005, "loss": 1.0696685314178467, "step": 11048 }, { "epoch": 1.2856644170351408, "grad_norm": 1.542962908744812, "learning_rate": 0.0005, "loss": 2.4026107788085938, "step": 11049 }, { "epoch": 1.2857807772864789, "grad_norm": 0.9535843133926392, "learning_rate": 0.0005, "loss": 2.579495429992676, "step": 11050 }, { "epoch": 1.285897137537817, "grad_norm": 1.3338141441345215, "learning_rate": 0.0005, "loss": 2.0558431148529053, "step": 11051 }, { "epoch": 1.2860134977891553, "grad_norm": 1.1214160919189453, "learning_rate": 0.0005, "loss": 2.4490816593170166, "step": 11052 }, { "epoch": 1.2861298580404934, "grad_norm": 1.2037910223007202, "learning_rate": 0.0005, "loss": 1.9227588176727295, "step": 11053 }, { "epoch": 1.2862462182918315, "grad_norm": 0.6843492388725281, "learning_rate": 0.0005, "loss": 1.4044264554977417, "step": 11054 }, { "epoch": 1.2863625785431696, "grad_norm": 2.716191291809082, "learning_rate": 0.0005, "loss": 2.1289048194885254, "step": 11055 }, { "epoch": 1.2864789387945077, "grad_norm": 1.36166512966156, "learning_rate": 0.0005, "loss": 2.2067205905914307, "step": 11056 }, { "epoch": 1.286595299045846, "grad_norm": 1.337166428565979, "learning_rate": 0.0005, "loss": 1.849493145942688, "step": 11057 }, { "epoch": 1.2867116592971841, "grad_norm": 0.8929842114448547, "learning_rate": 0.0005, "loss": 1.7192738056182861, "step": 11058 }, { "epoch": 1.2868280195485222, "grad_norm": 1.2291218042373657, "learning_rate": 0.0005, "loss": 2.470076084136963, "step": 11059 }, { "epoch": 1.2869443797998603, "grad_norm": 1.328392505645752, "learning_rate": 0.0005, "loss": 1.92333984375, "step": 11060 }, { "epoch": 1.2870607400511984, "grad_norm": 1.7154388427734375, "learning_rate": 0.0005, "loss": 2.1912195682525635, "step": 11061 }, { "epoch": 1.2871771003025367, "grad_norm": 1.0716146230697632, "learning_rate": 0.0005, "loss": 2.0264058113098145, "step": 11062 }, { "epoch": 1.2872934605538748, "grad_norm": 0.6628087162971497, "learning_rate": 0.0005, "loss": 1.7836155891418457, "step": 11063 }, { "epoch": 1.287409820805213, "grad_norm": 1.359189748764038, "learning_rate": 0.0005, "loss": 2.448242664337158, "step": 11064 }, { "epoch": 1.287526181056551, "grad_norm": 1.0326170921325684, "learning_rate": 0.0005, "loss": 2.2989087104797363, "step": 11065 }, { "epoch": 1.2876425413078891, "grad_norm": 1.8085535764694214, "learning_rate": 0.0005, "loss": 1.9371989965438843, "step": 11066 }, { "epoch": 1.2877589015592275, "grad_norm": 1.3689796924591064, "learning_rate": 0.0005, "loss": 2.266911268234253, "step": 11067 }, { "epoch": 1.2878752618105656, "grad_norm": 0.9855753779411316, "learning_rate": 0.0005, "loss": 2.5039641857147217, "step": 11068 }, { "epoch": 1.2879916220619037, "grad_norm": 1.2515028715133667, "learning_rate": 0.0005, "loss": 1.9570152759552002, "step": 11069 }, { "epoch": 1.2881079823132418, "grad_norm": 1.0988982915878296, "learning_rate": 0.0005, "loss": 2.131054162979126, "step": 11070 }, { "epoch": 1.2882243425645798, "grad_norm": 2.5281529426574707, "learning_rate": 0.0005, "loss": 2.671880006790161, "step": 11071 }, { "epoch": 1.2883407028159182, "grad_norm": 1.3439923524856567, "learning_rate": 0.0005, "loss": 1.984567403793335, "step": 11072 }, { "epoch": 1.2884570630672563, "grad_norm": 1.2878602743148804, "learning_rate": 0.0005, "loss": 2.576706886291504, "step": 11073 }, { "epoch": 1.2885734233185944, "grad_norm": 0.9711268544197083, "learning_rate": 0.0005, "loss": 1.876978874206543, "step": 11074 }, { "epoch": 1.2886897835699325, "grad_norm": 1.0265228748321533, "learning_rate": 0.0005, "loss": 1.8442472219467163, "step": 11075 }, { "epoch": 1.2888061438212706, "grad_norm": 1.0282596349716187, "learning_rate": 0.0005, "loss": 2.3141040802001953, "step": 11076 }, { "epoch": 1.2889225040726089, "grad_norm": 0.5356454253196716, "learning_rate": 0.0005, "loss": 1.2506083250045776, "step": 11077 }, { "epoch": 1.289038864323947, "grad_norm": 0.8415369391441345, "learning_rate": 0.0005, "loss": 1.3998591899871826, "step": 11078 }, { "epoch": 1.289155224575285, "grad_norm": 1.555646538734436, "learning_rate": 0.0005, "loss": 1.6014009714126587, "step": 11079 }, { "epoch": 1.2892715848266232, "grad_norm": 0.7623177766799927, "learning_rate": 0.0005, "loss": 1.4545189142227173, "step": 11080 }, { "epoch": 1.2893879450779613, "grad_norm": 1.0284955501556396, "learning_rate": 0.0005, "loss": 1.9859566688537598, "step": 11081 }, { "epoch": 1.2895043053292996, "grad_norm": 0.6191939115524292, "learning_rate": 0.0005, "loss": 2.2451868057250977, "step": 11082 }, { "epoch": 1.2896206655806377, "grad_norm": 0.9198442697525024, "learning_rate": 0.0005, "loss": 1.9655561447143555, "step": 11083 }, { "epoch": 1.2897370258319758, "grad_norm": 0.8290823101997375, "learning_rate": 0.0005, "loss": 1.5586893558502197, "step": 11084 }, { "epoch": 1.289853386083314, "grad_norm": 0.9690024852752686, "learning_rate": 0.0005, "loss": 2.233894109725952, "step": 11085 }, { "epoch": 1.289969746334652, "grad_norm": 1.497807502746582, "learning_rate": 0.0005, "loss": 2.0689785480499268, "step": 11086 }, { "epoch": 1.2900861065859903, "grad_norm": 0.6972648501396179, "learning_rate": 0.0005, "loss": 1.4019033908843994, "step": 11087 }, { "epoch": 1.2902024668373284, "grad_norm": 1.3599210977554321, "learning_rate": 0.0005, "loss": 2.4092254638671875, "step": 11088 }, { "epoch": 1.2903188270886665, "grad_norm": 1.3486729860305786, "learning_rate": 0.0005, "loss": 1.699631690979004, "step": 11089 }, { "epoch": 1.2904351873400046, "grad_norm": 1.1551071405410767, "learning_rate": 0.0005, "loss": 2.465686082839966, "step": 11090 }, { "epoch": 1.2905515475913427, "grad_norm": 1.0661211013793945, "learning_rate": 0.0005, "loss": 1.737756609916687, "step": 11091 }, { "epoch": 1.290667907842681, "grad_norm": 1.1282765865325928, "learning_rate": 0.0005, "loss": 2.005851984024048, "step": 11092 }, { "epoch": 1.2907842680940191, "grad_norm": 0.7380380630493164, "learning_rate": 0.0005, "loss": 1.5923469066619873, "step": 11093 }, { "epoch": 1.2909006283453572, "grad_norm": 1.6563528776168823, "learning_rate": 0.0005, "loss": 2.0305402278900146, "step": 11094 }, { "epoch": 1.2910169885966953, "grad_norm": 0.7436327338218689, "learning_rate": 0.0005, "loss": 1.914921522140503, "step": 11095 }, { "epoch": 1.2911333488480334, "grad_norm": 1.1363426446914673, "learning_rate": 0.0005, "loss": 2.248537063598633, "step": 11096 }, { "epoch": 1.2912497090993718, "grad_norm": 1.411702275276184, "learning_rate": 0.0005, "loss": 1.728280782699585, "step": 11097 }, { "epoch": 1.2913660693507099, "grad_norm": 1.955894112586975, "learning_rate": 0.0005, "loss": 1.7304409742355347, "step": 11098 }, { "epoch": 1.291482429602048, "grad_norm": 0.7456801533699036, "learning_rate": 0.0005, "loss": 1.6291565895080566, "step": 11099 }, { "epoch": 1.291598789853386, "grad_norm": 0.9754871129989624, "learning_rate": 0.0005, "loss": 1.610508918762207, "step": 11100 }, { "epoch": 1.2917151501047242, "grad_norm": 1.0180875062942505, "learning_rate": 0.0005, "loss": 1.902433156967163, "step": 11101 }, { "epoch": 1.2918315103560625, "grad_norm": 2.239637851715088, "learning_rate": 0.0005, "loss": 2.2454209327697754, "step": 11102 }, { "epoch": 1.2919478706074006, "grad_norm": 1.000935673713684, "learning_rate": 0.0005, "loss": 1.5413540601730347, "step": 11103 }, { "epoch": 1.2920642308587387, "grad_norm": 1.162548303604126, "learning_rate": 0.0005, "loss": 1.8115750551223755, "step": 11104 }, { "epoch": 1.2921805911100768, "grad_norm": 1.366681456565857, "learning_rate": 0.0005, "loss": 2.254607677459717, "step": 11105 }, { "epoch": 1.2922969513614149, "grad_norm": 3.455767869949341, "learning_rate": 0.0005, "loss": 1.8894294500350952, "step": 11106 }, { "epoch": 1.2924133116127532, "grad_norm": 0.9527636766433716, "learning_rate": 0.0005, "loss": 1.1777119636535645, "step": 11107 }, { "epoch": 1.2925296718640913, "grad_norm": 1.1742439270019531, "learning_rate": 0.0005, "loss": 2.219971179962158, "step": 11108 }, { "epoch": 1.2926460321154294, "grad_norm": 1.3803868293762207, "learning_rate": 0.0005, "loss": 2.075249671936035, "step": 11109 }, { "epoch": 1.2927623923667675, "grad_norm": 1.137900710105896, "learning_rate": 0.0005, "loss": 2.336132526397705, "step": 11110 }, { "epoch": 1.2928787526181056, "grad_norm": 1.725006103515625, "learning_rate": 0.0005, "loss": 2.1828746795654297, "step": 11111 }, { "epoch": 1.292995112869444, "grad_norm": 0.9694578051567078, "learning_rate": 0.0005, "loss": 1.1481902599334717, "step": 11112 }, { "epoch": 1.293111473120782, "grad_norm": 0.8424460291862488, "learning_rate": 0.0005, "loss": 1.5990628004074097, "step": 11113 }, { "epoch": 1.29322783337212, "grad_norm": 1.2377681732177734, "learning_rate": 0.0005, "loss": 2.6539387702941895, "step": 11114 }, { "epoch": 1.2933441936234582, "grad_norm": 2.0051393508911133, "learning_rate": 0.0005, "loss": 1.9152441024780273, "step": 11115 }, { "epoch": 1.2934605538747963, "grad_norm": 2.0044116973876953, "learning_rate": 0.0005, "loss": 2.5314407348632812, "step": 11116 }, { "epoch": 1.2935769141261346, "grad_norm": 1.0937916040420532, "learning_rate": 0.0005, "loss": 2.0440425872802734, "step": 11117 }, { "epoch": 1.2936932743774727, "grad_norm": 1.3810147047042847, "learning_rate": 0.0005, "loss": 1.9918608665466309, "step": 11118 }, { "epoch": 1.2938096346288108, "grad_norm": 3.9888436794281006, "learning_rate": 0.0005, "loss": 1.6006927490234375, "step": 11119 }, { "epoch": 1.293925994880149, "grad_norm": 0.6667367815971375, "learning_rate": 0.0005, "loss": 1.2970153093338013, "step": 11120 }, { "epoch": 1.294042355131487, "grad_norm": 1.1402987241744995, "learning_rate": 0.0005, "loss": 2.242539167404175, "step": 11121 }, { "epoch": 1.2941587153828253, "grad_norm": 1.1365299224853516, "learning_rate": 0.0005, "loss": 2.3395962715148926, "step": 11122 }, { "epoch": 1.2942750756341634, "grad_norm": 0.7889721989631653, "learning_rate": 0.0005, "loss": 1.5788302421569824, "step": 11123 }, { "epoch": 1.2943914358855015, "grad_norm": 0.7960143685340881, "learning_rate": 0.0005, "loss": 2.123051166534424, "step": 11124 }, { "epoch": 1.2945077961368396, "grad_norm": 1.0614856481552124, "learning_rate": 0.0005, "loss": 2.183041572570801, "step": 11125 }, { "epoch": 1.2946241563881777, "grad_norm": 0.8280754685401917, "learning_rate": 0.0005, "loss": 1.6973363161087036, "step": 11126 }, { "epoch": 1.294740516639516, "grad_norm": 0.8692798018455505, "learning_rate": 0.0005, "loss": 2.199054002761841, "step": 11127 }, { "epoch": 1.2948568768908542, "grad_norm": 1.258409857749939, "learning_rate": 0.0005, "loss": 1.7059367895126343, "step": 11128 }, { "epoch": 1.2949732371421923, "grad_norm": 0.9310052990913391, "learning_rate": 0.0005, "loss": 1.8775959014892578, "step": 11129 }, { "epoch": 1.2950895973935304, "grad_norm": 1.2536442279815674, "learning_rate": 0.0005, "loss": 1.8085482120513916, "step": 11130 }, { "epoch": 1.2952059576448685, "grad_norm": 1.0838823318481445, "learning_rate": 0.0005, "loss": 2.412876605987549, "step": 11131 }, { "epoch": 1.2953223178962068, "grad_norm": 0.8702659606933594, "learning_rate": 0.0005, "loss": 1.5294814109802246, "step": 11132 }, { "epoch": 1.2954386781475449, "grad_norm": 2.5118255615234375, "learning_rate": 0.0005, "loss": 1.8794877529144287, "step": 11133 }, { "epoch": 1.295555038398883, "grad_norm": 1.0368866920471191, "learning_rate": 0.0005, "loss": 2.0648412704467773, "step": 11134 }, { "epoch": 1.295671398650221, "grad_norm": 1.2972570657730103, "learning_rate": 0.0005, "loss": 2.019878387451172, "step": 11135 }, { "epoch": 1.2957877589015592, "grad_norm": 1.0250840187072754, "learning_rate": 0.0005, "loss": 1.720800757408142, "step": 11136 }, { "epoch": 1.2959041191528975, "grad_norm": 1.3586313724517822, "learning_rate": 0.0005, "loss": 2.5711283683776855, "step": 11137 }, { "epoch": 1.2960204794042356, "grad_norm": 1.0505242347717285, "learning_rate": 0.0005, "loss": 1.4035838842391968, "step": 11138 }, { "epoch": 1.2961368396555737, "grad_norm": 1.1313273906707764, "learning_rate": 0.0005, "loss": 2.088132858276367, "step": 11139 }, { "epoch": 1.2962531999069118, "grad_norm": 1.2076427936553955, "learning_rate": 0.0005, "loss": 1.9343029260635376, "step": 11140 }, { "epoch": 1.29636956015825, "grad_norm": 0.5785985589027405, "learning_rate": 0.0005, "loss": 1.4280856847763062, "step": 11141 }, { "epoch": 1.2964859204095882, "grad_norm": 1.300264596939087, "learning_rate": 0.0005, "loss": 2.617514133453369, "step": 11142 }, { "epoch": 1.2966022806609263, "grad_norm": 1.4461606740951538, "learning_rate": 0.0005, "loss": 1.8436062335968018, "step": 11143 }, { "epoch": 1.2967186409122644, "grad_norm": 1.2857661247253418, "learning_rate": 0.0005, "loss": 1.9045720100402832, "step": 11144 }, { "epoch": 1.2968350011636025, "grad_norm": 1.3972983360290527, "learning_rate": 0.0005, "loss": 2.041881561279297, "step": 11145 }, { "epoch": 1.2969513614149406, "grad_norm": 1.1442714929580688, "learning_rate": 0.0005, "loss": 2.2174320220947266, "step": 11146 }, { "epoch": 1.297067721666279, "grad_norm": 1.2683191299438477, "learning_rate": 0.0005, "loss": 1.8833223581314087, "step": 11147 }, { "epoch": 1.2971840819176168, "grad_norm": 1.3041915893554688, "learning_rate": 0.0005, "loss": 2.006944179534912, "step": 11148 }, { "epoch": 1.2973004421689551, "grad_norm": 1.384435772895813, "learning_rate": 0.0005, "loss": 1.9660223722457886, "step": 11149 }, { "epoch": 1.2974168024202932, "grad_norm": 0.9420487284660339, "learning_rate": 0.0005, "loss": 1.27189302444458, "step": 11150 }, { "epoch": 1.2975331626716313, "grad_norm": 1.4678566455841064, "learning_rate": 0.0005, "loss": 2.30389142036438, "step": 11151 }, { "epoch": 1.2976495229229696, "grad_norm": 1.1090954542160034, "learning_rate": 0.0005, "loss": 1.7775015830993652, "step": 11152 }, { "epoch": 1.2977658831743075, "grad_norm": 0.8123235106468201, "learning_rate": 0.0005, "loss": 1.3829073905944824, "step": 11153 }, { "epoch": 1.2978822434256458, "grad_norm": 0.8402891159057617, "learning_rate": 0.0005, "loss": 1.6805698871612549, "step": 11154 }, { "epoch": 1.297998603676984, "grad_norm": 1.0378127098083496, "learning_rate": 0.0005, "loss": 1.9200108051300049, "step": 11155 }, { "epoch": 1.298114963928322, "grad_norm": 0.9818351864814758, "learning_rate": 0.0005, "loss": 1.7637426853179932, "step": 11156 }, { "epoch": 1.2982313241796604, "grad_norm": 0.8270947933197021, "learning_rate": 0.0005, "loss": 1.6933807134628296, "step": 11157 }, { "epoch": 1.2983476844309982, "grad_norm": 0.9598032236099243, "learning_rate": 0.0005, "loss": 1.9312756061553955, "step": 11158 }, { "epoch": 1.2984640446823366, "grad_norm": 1.5778957605361938, "learning_rate": 0.0005, "loss": 1.9865045547485352, "step": 11159 }, { "epoch": 1.2985804049336747, "grad_norm": 1.069789171218872, "learning_rate": 0.0005, "loss": 2.638589382171631, "step": 11160 }, { "epoch": 1.2986967651850128, "grad_norm": 1.1314148902893066, "learning_rate": 0.0005, "loss": 2.2325267791748047, "step": 11161 }, { "epoch": 1.298813125436351, "grad_norm": 0.9122942090034485, "learning_rate": 0.0005, "loss": 2.296536922454834, "step": 11162 }, { "epoch": 1.298929485687689, "grad_norm": 1.2917474508285522, "learning_rate": 0.0005, "loss": 1.4432440996170044, "step": 11163 }, { "epoch": 1.2990458459390273, "grad_norm": 0.7540499567985535, "learning_rate": 0.0005, "loss": 1.76572585105896, "step": 11164 }, { "epoch": 1.2991622061903654, "grad_norm": 1.1946176290512085, "learning_rate": 0.0005, "loss": 2.1684093475341797, "step": 11165 }, { "epoch": 1.2992785664417035, "grad_norm": 0.8726640343666077, "learning_rate": 0.0005, "loss": 1.6701059341430664, "step": 11166 }, { "epoch": 1.2993949266930416, "grad_norm": 1.0559989213943481, "learning_rate": 0.0005, "loss": 2.001556634902954, "step": 11167 }, { "epoch": 1.2995112869443797, "grad_norm": 1.8629953861236572, "learning_rate": 0.0005, "loss": 2.1948022842407227, "step": 11168 }, { "epoch": 1.299627647195718, "grad_norm": 1.3331180810928345, "learning_rate": 0.0005, "loss": 2.11887264251709, "step": 11169 }, { "epoch": 1.299744007447056, "grad_norm": 0.9062345623970032, "learning_rate": 0.0005, "loss": 1.037922739982605, "step": 11170 }, { "epoch": 1.2998603676983942, "grad_norm": 1.5536024570465088, "learning_rate": 0.0005, "loss": 2.004974603652954, "step": 11171 }, { "epoch": 1.2999767279497323, "grad_norm": 1.3733717203140259, "learning_rate": 0.0005, "loss": 1.7006809711456299, "step": 11172 }, { "epoch": 1.3000930882010704, "grad_norm": 0.8134461045265198, "learning_rate": 0.0005, "loss": 1.253133773803711, "step": 11173 }, { "epoch": 1.3002094484524087, "grad_norm": 1.0196833610534668, "learning_rate": 0.0005, "loss": 1.8618800640106201, "step": 11174 }, { "epoch": 1.3003258087037468, "grad_norm": 0.7886137962341309, "learning_rate": 0.0005, "loss": 1.7072641849517822, "step": 11175 }, { "epoch": 1.300442168955085, "grad_norm": 1.0170589685440063, "learning_rate": 0.0005, "loss": 1.3437128067016602, "step": 11176 }, { "epoch": 1.300558529206423, "grad_norm": 0.7507055997848511, "learning_rate": 0.0005, "loss": 1.4984240531921387, "step": 11177 }, { "epoch": 1.300674889457761, "grad_norm": 1.0848612785339355, "learning_rate": 0.0005, "loss": 1.589930534362793, "step": 11178 }, { "epoch": 1.3007912497090994, "grad_norm": 1.3200207948684692, "learning_rate": 0.0005, "loss": 2.4629294872283936, "step": 11179 }, { "epoch": 1.3009076099604375, "grad_norm": 1.1264578104019165, "learning_rate": 0.0005, "loss": 1.941878080368042, "step": 11180 }, { "epoch": 1.3010239702117756, "grad_norm": 1.8757721185684204, "learning_rate": 0.0005, "loss": 1.700989007949829, "step": 11181 }, { "epoch": 1.3011403304631137, "grad_norm": 1.6063079833984375, "learning_rate": 0.0005, "loss": 2.558849334716797, "step": 11182 }, { "epoch": 1.3012566907144518, "grad_norm": 0.9538173079490662, "learning_rate": 0.0005, "loss": 1.986445426940918, "step": 11183 }, { "epoch": 1.3013730509657901, "grad_norm": 1.457890272140503, "learning_rate": 0.0005, "loss": 2.112481117248535, "step": 11184 }, { "epoch": 1.3014894112171282, "grad_norm": 0.9360113143920898, "learning_rate": 0.0005, "loss": 1.541674017906189, "step": 11185 }, { "epoch": 1.3016057714684663, "grad_norm": 1.4005663394927979, "learning_rate": 0.0005, "loss": 1.8933097124099731, "step": 11186 }, { "epoch": 1.3017221317198044, "grad_norm": 1.0398002862930298, "learning_rate": 0.0005, "loss": 2.0373692512512207, "step": 11187 }, { "epoch": 1.3018384919711425, "grad_norm": 1.0459401607513428, "learning_rate": 0.0005, "loss": 1.4503257274627686, "step": 11188 }, { "epoch": 1.3019548522224809, "grad_norm": 1.1575921773910522, "learning_rate": 0.0005, "loss": 1.9697858095169067, "step": 11189 }, { "epoch": 1.302071212473819, "grad_norm": 1.3035351037979126, "learning_rate": 0.0005, "loss": 2.0554070472717285, "step": 11190 }, { "epoch": 1.302187572725157, "grad_norm": 0.9767264127731323, "learning_rate": 0.0005, "loss": 1.8470903635025024, "step": 11191 }, { "epoch": 1.3023039329764952, "grad_norm": 0.681975245475769, "learning_rate": 0.0005, "loss": 1.845841646194458, "step": 11192 }, { "epoch": 1.3024202932278333, "grad_norm": 1.0590437650680542, "learning_rate": 0.0005, "loss": 2.168403387069702, "step": 11193 }, { "epoch": 1.3025366534791716, "grad_norm": 0.6859256625175476, "learning_rate": 0.0005, "loss": 1.610762119293213, "step": 11194 }, { "epoch": 1.3026530137305097, "grad_norm": 0.8845553398132324, "learning_rate": 0.0005, "loss": 2.0528101921081543, "step": 11195 }, { "epoch": 1.3027693739818478, "grad_norm": 1.0277214050292969, "learning_rate": 0.0005, "loss": 1.7273316383361816, "step": 11196 }, { "epoch": 1.3028857342331859, "grad_norm": 0.93158358335495, "learning_rate": 0.0005, "loss": 2.07344126701355, "step": 11197 }, { "epoch": 1.303002094484524, "grad_norm": 1.0571759939193726, "learning_rate": 0.0005, "loss": 1.5308425426483154, "step": 11198 }, { "epoch": 1.3031184547358623, "grad_norm": 1.0058327913284302, "learning_rate": 0.0005, "loss": 2.1779189109802246, "step": 11199 }, { "epoch": 1.3032348149872004, "grad_norm": 1.006142020225525, "learning_rate": 0.0005, "loss": 2.0867011547088623, "step": 11200 }, { "epoch": 1.3033511752385385, "grad_norm": 1.2259225845336914, "learning_rate": 0.0005, "loss": 1.3396856784820557, "step": 11201 }, { "epoch": 1.3034675354898766, "grad_norm": 1.4202227592468262, "learning_rate": 0.0005, "loss": 1.7859575748443604, "step": 11202 }, { "epoch": 1.3035838957412147, "grad_norm": 1.5575003623962402, "learning_rate": 0.0005, "loss": 2.527618885040283, "step": 11203 }, { "epoch": 1.303700255992553, "grad_norm": 1.1999796628952026, "learning_rate": 0.0005, "loss": 2.553347110748291, "step": 11204 }, { "epoch": 1.3038166162438911, "grad_norm": 1.2574889659881592, "learning_rate": 0.0005, "loss": 2.0285937786102295, "step": 11205 }, { "epoch": 1.3039329764952292, "grad_norm": 1.349931240081787, "learning_rate": 0.0005, "loss": 2.330578327178955, "step": 11206 }, { "epoch": 1.3040493367465673, "grad_norm": 0.7875637412071228, "learning_rate": 0.0005, "loss": 1.1294262409210205, "step": 11207 }, { "epoch": 1.3041656969979054, "grad_norm": 1.4256089925765991, "learning_rate": 0.0005, "loss": 2.0741915702819824, "step": 11208 }, { "epoch": 1.3042820572492437, "grad_norm": 0.7103700041770935, "learning_rate": 0.0005, "loss": 1.4917340278625488, "step": 11209 }, { "epoch": 1.3043984175005818, "grad_norm": 0.7212035655975342, "learning_rate": 0.0005, "loss": 1.580299973487854, "step": 11210 }, { "epoch": 1.30451477775192, "grad_norm": 1.2497873306274414, "learning_rate": 0.0005, "loss": 1.8581583499908447, "step": 11211 }, { "epoch": 1.304631138003258, "grad_norm": 1.2600629329681396, "learning_rate": 0.0005, "loss": 1.6006994247436523, "step": 11212 }, { "epoch": 1.3047474982545961, "grad_norm": 0.8358891010284424, "learning_rate": 0.0005, "loss": 1.221240520477295, "step": 11213 }, { "epoch": 1.3048638585059344, "grad_norm": 1.1761212348937988, "learning_rate": 0.0005, "loss": 1.743048906326294, "step": 11214 }, { "epoch": 1.3049802187572725, "grad_norm": 0.6319395899772644, "learning_rate": 0.0005, "loss": 1.3333330154418945, "step": 11215 }, { "epoch": 1.3050965790086106, "grad_norm": 1.158789873123169, "learning_rate": 0.0005, "loss": 2.1532764434814453, "step": 11216 }, { "epoch": 1.3052129392599487, "grad_norm": 1.3788098096847534, "learning_rate": 0.0005, "loss": 2.075185775756836, "step": 11217 }, { "epoch": 1.3053292995112868, "grad_norm": 1.0195478200912476, "learning_rate": 0.0005, "loss": 1.9188969135284424, "step": 11218 }, { "epoch": 1.3054456597626252, "grad_norm": 0.7832414507865906, "learning_rate": 0.0005, "loss": 1.60001540184021, "step": 11219 }, { "epoch": 1.3055620200139633, "grad_norm": 1.007925033569336, "learning_rate": 0.0005, "loss": 1.7748692035675049, "step": 11220 }, { "epoch": 1.3056783802653014, "grad_norm": 0.8224172592163086, "learning_rate": 0.0005, "loss": 1.7460366487503052, "step": 11221 }, { "epoch": 1.3057947405166395, "grad_norm": 0.8997535109519958, "learning_rate": 0.0005, "loss": 2.2618491649627686, "step": 11222 }, { "epoch": 1.3059111007679776, "grad_norm": 0.9820994734764099, "learning_rate": 0.0005, "loss": 2.52685809135437, "step": 11223 }, { "epoch": 1.3060274610193159, "grad_norm": 1.5621623992919922, "learning_rate": 0.0005, "loss": 2.1400532722473145, "step": 11224 }, { "epoch": 1.306143821270654, "grad_norm": 0.9628254771232605, "learning_rate": 0.0005, "loss": 1.6153868436813354, "step": 11225 }, { "epoch": 1.306260181521992, "grad_norm": 1.2862480878829956, "learning_rate": 0.0005, "loss": 1.879880428314209, "step": 11226 }, { "epoch": 1.3063765417733302, "grad_norm": 1.1771506071090698, "learning_rate": 0.0005, "loss": 1.8232309818267822, "step": 11227 }, { "epoch": 1.3064929020246683, "grad_norm": 1.1690154075622559, "learning_rate": 0.0005, "loss": 1.646592617034912, "step": 11228 }, { "epoch": 1.3066092622760066, "grad_norm": 1.943041443824768, "learning_rate": 0.0005, "loss": 2.86184024810791, "step": 11229 }, { "epoch": 1.3067256225273447, "grad_norm": 0.8042564392089844, "learning_rate": 0.0005, "loss": 1.1663494110107422, "step": 11230 }, { "epoch": 1.3068419827786828, "grad_norm": 1.334364652633667, "learning_rate": 0.0005, "loss": 2.1619296073913574, "step": 11231 }, { "epoch": 1.306958343030021, "grad_norm": 0.564383864402771, "learning_rate": 0.0005, "loss": 1.4411518573760986, "step": 11232 }, { "epoch": 1.307074703281359, "grad_norm": 1.0105077028274536, "learning_rate": 0.0005, "loss": 1.608534336090088, "step": 11233 }, { "epoch": 1.3071910635326973, "grad_norm": 1.2792925834655762, "learning_rate": 0.0005, "loss": 1.7285563945770264, "step": 11234 }, { "epoch": 1.3073074237840354, "grad_norm": 1.7306973934173584, "learning_rate": 0.0005, "loss": 2.4420828819274902, "step": 11235 }, { "epoch": 1.3074237840353735, "grad_norm": 1.1358145475387573, "learning_rate": 0.0005, "loss": 1.8718235492706299, "step": 11236 }, { "epoch": 1.3075401442867116, "grad_norm": 1.2711207866668701, "learning_rate": 0.0005, "loss": 1.629845380783081, "step": 11237 }, { "epoch": 1.3076565045380497, "grad_norm": 0.8050060868263245, "learning_rate": 0.0005, "loss": 1.565683364868164, "step": 11238 }, { "epoch": 1.307772864789388, "grad_norm": 0.9745016694068909, "learning_rate": 0.0005, "loss": 1.7201581001281738, "step": 11239 }, { "epoch": 1.3078892250407261, "grad_norm": 1.0988579988479614, "learning_rate": 0.0005, "loss": 2.1614608764648438, "step": 11240 }, { "epoch": 1.3080055852920642, "grad_norm": 1.1048318147659302, "learning_rate": 0.0005, "loss": 1.9940471649169922, "step": 11241 }, { "epoch": 1.3081219455434023, "grad_norm": 1.3539921045303345, "learning_rate": 0.0005, "loss": 2.276726245880127, "step": 11242 }, { "epoch": 1.3082383057947404, "grad_norm": 0.877596914768219, "learning_rate": 0.0005, "loss": 1.7183125019073486, "step": 11243 }, { "epoch": 1.3083546660460788, "grad_norm": 0.8403649926185608, "learning_rate": 0.0005, "loss": 1.6896860599517822, "step": 11244 }, { "epoch": 1.3084710262974169, "grad_norm": 0.8941880464553833, "learning_rate": 0.0005, "loss": 1.827102780342102, "step": 11245 }, { "epoch": 1.308587386548755, "grad_norm": 0.49209359288215637, "learning_rate": 0.0005, "loss": 1.2141289710998535, "step": 11246 }, { "epoch": 1.308703746800093, "grad_norm": 1.1271365880966187, "learning_rate": 0.0005, "loss": 3.070436716079712, "step": 11247 }, { "epoch": 1.3088201070514311, "grad_norm": 0.7958762049674988, "learning_rate": 0.0005, "loss": 1.4508908987045288, "step": 11248 }, { "epoch": 1.3089364673027695, "grad_norm": 1.221114993095398, "learning_rate": 0.0005, "loss": 1.8777251243591309, "step": 11249 }, { "epoch": 1.3090528275541076, "grad_norm": 1.219569444656372, "learning_rate": 0.0005, "loss": 2.278195858001709, "step": 11250 }, { "epoch": 1.3091691878054457, "grad_norm": 1.0888934135437012, "learning_rate": 0.0005, "loss": 1.9967057704925537, "step": 11251 }, { "epoch": 1.3092855480567838, "grad_norm": 6.731247901916504, "learning_rate": 0.0005, "loss": 1.6312559843063354, "step": 11252 }, { "epoch": 1.3094019083081219, "grad_norm": 0.778605043888092, "learning_rate": 0.0005, "loss": 1.7406576871871948, "step": 11253 }, { "epoch": 1.3095182685594602, "grad_norm": 1.196773648262024, "learning_rate": 0.0005, "loss": 1.9573558568954468, "step": 11254 }, { "epoch": 1.3096346288107983, "grad_norm": 0.6423141360282898, "learning_rate": 0.0005, "loss": 1.4460004568099976, "step": 11255 }, { "epoch": 1.3097509890621364, "grad_norm": 1.2210413217544556, "learning_rate": 0.0005, "loss": 1.8399271965026855, "step": 11256 }, { "epoch": 1.3098673493134745, "grad_norm": 0.9633110761642456, "learning_rate": 0.0005, "loss": 1.3721401691436768, "step": 11257 }, { "epoch": 1.3099837095648126, "grad_norm": 1.5673083066940308, "learning_rate": 0.0005, "loss": 1.7339820861816406, "step": 11258 }, { "epoch": 1.310100069816151, "grad_norm": 0.7400236129760742, "learning_rate": 0.0005, "loss": 1.5123560428619385, "step": 11259 }, { "epoch": 1.310216430067489, "grad_norm": 0.9214052557945251, "learning_rate": 0.0005, "loss": 1.380326271057129, "step": 11260 }, { "epoch": 1.310332790318827, "grad_norm": 1.1668872833251953, "learning_rate": 0.0005, "loss": 1.5747395753860474, "step": 11261 }, { "epoch": 1.3104491505701652, "grad_norm": 0.9119603037834167, "learning_rate": 0.0005, "loss": 1.5993387699127197, "step": 11262 }, { "epoch": 1.3105655108215033, "grad_norm": 1.040382742881775, "learning_rate": 0.0005, "loss": 1.6786353588104248, "step": 11263 }, { "epoch": 1.3106818710728416, "grad_norm": 0.8929489254951477, "learning_rate": 0.0005, "loss": 1.526535153388977, "step": 11264 }, { "epoch": 1.3107982313241797, "grad_norm": 1.2987725734710693, "learning_rate": 0.0005, "loss": 2.619908571243286, "step": 11265 }, { "epoch": 1.3109145915755178, "grad_norm": 1.1169489622116089, "learning_rate": 0.0005, "loss": 2.182473659515381, "step": 11266 }, { "epoch": 1.311030951826856, "grad_norm": 0.9278358221054077, "learning_rate": 0.0005, "loss": 1.1819210052490234, "step": 11267 }, { "epoch": 1.311147312078194, "grad_norm": 1.2536098957061768, "learning_rate": 0.0005, "loss": 1.9791308641433716, "step": 11268 }, { "epoch": 1.3112636723295323, "grad_norm": 0.5893057584762573, "learning_rate": 0.0005, "loss": 1.0561177730560303, "step": 11269 }, { "epoch": 1.3113800325808704, "grad_norm": 1.2940231561660767, "learning_rate": 0.0005, "loss": 1.457635760307312, "step": 11270 }, { "epoch": 1.3114963928322085, "grad_norm": 1.2029527425765991, "learning_rate": 0.0005, "loss": 1.8230743408203125, "step": 11271 }, { "epoch": 1.3116127530835466, "grad_norm": 0.9192609786987305, "learning_rate": 0.0005, "loss": 1.4216279983520508, "step": 11272 }, { "epoch": 1.3117291133348847, "grad_norm": 1.4040687084197998, "learning_rate": 0.0005, "loss": 1.9172269105911255, "step": 11273 }, { "epoch": 1.311845473586223, "grad_norm": 1.0533400774002075, "learning_rate": 0.0005, "loss": 2.2675604820251465, "step": 11274 }, { "epoch": 1.3119618338375612, "grad_norm": 1.7625585794448853, "learning_rate": 0.0005, "loss": 1.9285162687301636, "step": 11275 }, { "epoch": 1.3120781940888993, "grad_norm": 1.2946220636367798, "learning_rate": 0.0005, "loss": 1.9192862510681152, "step": 11276 }, { "epoch": 1.3121945543402374, "grad_norm": 0.8768684267997742, "learning_rate": 0.0005, "loss": 1.5589675903320312, "step": 11277 }, { "epoch": 1.3123109145915755, "grad_norm": 1.0907907485961914, "learning_rate": 0.0005, "loss": 2.1532533168792725, "step": 11278 }, { "epoch": 1.3124272748429138, "grad_norm": 0.754148006439209, "learning_rate": 0.0005, "loss": 1.7619425058364868, "step": 11279 }, { "epoch": 1.3125436350942519, "grad_norm": 1.4928747415542603, "learning_rate": 0.0005, "loss": 2.173555374145508, "step": 11280 }, { "epoch": 1.31265999534559, "grad_norm": 1.840938925743103, "learning_rate": 0.0005, "loss": 1.8324644565582275, "step": 11281 }, { "epoch": 1.312776355596928, "grad_norm": 0.9187030792236328, "learning_rate": 0.0005, "loss": 2.2729337215423584, "step": 11282 }, { "epoch": 1.3128927158482662, "grad_norm": 1.0335197448730469, "learning_rate": 0.0005, "loss": 2.1692302227020264, "step": 11283 }, { "epoch": 1.3130090760996045, "grad_norm": 1.2132742404937744, "learning_rate": 0.0005, "loss": 2.2241406440734863, "step": 11284 }, { "epoch": 1.3131254363509426, "grad_norm": 0.832695484161377, "learning_rate": 0.0005, "loss": 1.3156030178070068, "step": 11285 }, { "epoch": 1.3132417966022807, "grad_norm": 1.3875066041946411, "learning_rate": 0.0005, "loss": 2.442866802215576, "step": 11286 }, { "epoch": 1.3133581568536188, "grad_norm": 1.1860822439193726, "learning_rate": 0.0005, "loss": 1.4694385528564453, "step": 11287 }, { "epoch": 1.3134745171049569, "grad_norm": 0.9293099641799927, "learning_rate": 0.0005, "loss": 1.3498575687408447, "step": 11288 }, { "epoch": 1.3135908773562952, "grad_norm": 1.5588023662567139, "learning_rate": 0.0005, "loss": 2.211792230606079, "step": 11289 }, { "epoch": 1.3137072376076333, "grad_norm": 1.564500331878662, "learning_rate": 0.0005, "loss": 2.9097061157226562, "step": 11290 }, { "epoch": 1.3138235978589714, "grad_norm": 0.6920228004455566, "learning_rate": 0.0005, "loss": 1.1608846187591553, "step": 11291 }, { "epoch": 1.3139399581103095, "grad_norm": 1.466722846031189, "learning_rate": 0.0005, "loss": 2.069798469543457, "step": 11292 }, { "epoch": 1.3140563183616476, "grad_norm": 1.029483437538147, "learning_rate": 0.0005, "loss": 1.314671516418457, "step": 11293 }, { "epoch": 1.314172678612986, "grad_norm": 1.3587738275527954, "learning_rate": 0.0005, "loss": 2.4105732440948486, "step": 11294 }, { "epoch": 1.314289038864324, "grad_norm": 1.5583546161651611, "learning_rate": 0.0005, "loss": 2.001678943634033, "step": 11295 }, { "epoch": 1.3144053991156621, "grad_norm": 0.8375001549720764, "learning_rate": 0.0005, "loss": 2.0108211040496826, "step": 11296 }, { "epoch": 1.3145217593670002, "grad_norm": 1.2070204019546509, "learning_rate": 0.0005, "loss": 1.9103742837905884, "step": 11297 }, { "epoch": 1.3146381196183383, "grad_norm": 1.1708639860153198, "learning_rate": 0.0005, "loss": 2.5934669971466064, "step": 11298 }, { "epoch": 1.3147544798696766, "grad_norm": 1.5622004270553589, "learning_rate": 0.0005, "loss": 1.8857264518737793, "step": 11299 }, { "epoch": 1.3148708401210147, "grad_norm": 0.8505969643592834, "learning_rate": 0.0005, "loss": 1.3826987743377686, "step": 11300 }, { "epoch": 1.3149872003723528, "grad_norm": 1.089503288269043, "learning_rate": 0.0005, "loss": 1.8455785512924194, "step": 11301 }, { "epoch": 1.315103560623691, "grad_norm": 1.00845205783844, "learning_rate": 0.0005, "loss": 1.77675461769104, "step": 11302 }, { "epoch": 1.315219920875029, "grad_norm": 1.2754969596862793, "learning_rate": 0.0005, "loss": 2.0048298835754395, "step": 11303 }, { "epoch": 1.3153362811263674, "grad_norm": 1.1967486143112183, "learning_rate": 0.0005, "loss": 1.4771430492401123, "step": 11304 }, { "epoch": 1.3154526413777055, "grad_norm": 1.2359344959259033, "learning_rate": 0.0005, "loss": 2.1799795627593994, "step": 11305 }, { "epoch": 1.3155690016290436, "grad_norm": 1.26243257522583, "learning_rate": 0.0005, "loss": 1.732219934463501, "step": 11306 }, { "epoch": 1.3156853618803817, "grad_norm": 0.9690139293670654, "learning_rate": 0.0005, "loss": 2.232593536376953, "step": 11307 }, { "epoch": 1.3158017221317198, "grad_norm": 0.9696044921875, "learning_rate": 0.0005, "loss": 2.0990712642669678, "step": 11308 }, { "epoch": 1.315918082383058, "grad_norm": 1.2688031196594238, "learning_rate": 0.0005, "loss": 2.1882472038269043, "step": 11309 }, { "epoch": 1.3160344426343962, "grad_norm": 2.1095263957977295, "learning_rate": 0.0005, "loss": 2.4219799041748047, "step": 11310 }, { "epoch": 1.3161508028857343, "grad_norm": 0.9126394987106323, "learning_rate": 0.0005, "loss": 1.7633609771728516, "step": 11311 }, { "epoch": 1.3162671631370724, "grad_norm": 0.7471426129341125, "learning_rate": 0.0005, "loss": 1.3741507530212402, "step": 11312 }, { "epoch": 1.3163835233884105, "grad_norm": 0.8027582168579102, "learning_rate": 0.0005, "loss": 1.386271357536316, "step": 11313 }, { "epoch": 1.3164998836397488, "grad_norm": 1.4919390678405762, "learning_rate": 0.0005, "loss": 2.7316231727600098, "step": 11314 }, { "epoch": 1.3166162438910867, "grad_norm": 1.2266051769256592, "learning_rate": 0.0005, "loss": 2.31829833984375, "step": 11315 }, { "epoch": 1.316732604142425, "grad_norm": 1.221930742263794, "learning_rate": 0.0005, "loss": 2.3712525367736816, "step": 11316 }, { "epoch": 1.316848964393763, "grad_norm": 1.7164443731307983, "learning_rate": 0.0005, "loss": 1.9871827363967896, "step": 11317 }, { "epoch": 1.3169653246451012, "grad_norm": 1.022786259651184, "learning_rate": 0.0005, "loss": 2.38775634765625, "step": 11318 }, { "epoch": 1.3170816848964395, "grad_norm": 1.2601886987686157, "learning_rate": 0.0005, "loss": 1.4480277299880981, "step": 11319 }, { "epoch": 1.3171980451477774, "grad_norm": 0.9908168911933899, "learning_rate": 0.0005, "loss": 2.0400776863098145, "step": 11320 }, { "epoch": 1.3173144053991157, "grad_norm": 1.6246368885040283, "learning_rate": 0.0005, "loss": 1.379340648651123, "step": 11321 }, { "epoch": 1.3174307656504538, "grad_norm": 0.8852670192718506, "learning_rate": 0.0005, "loss": 1.979797124862671, "step": 11322 }, { "epoch": 1.317547125901792, "grad_norm": 1.6219061613082886, "learning_rate": 0.0005, "loss": 1.5894330739974976, "step": 11323 }, { "epoch": 1.3176634861531302, "grad_norm": 1.3580937385559082, "learning_rate": 0.0005, "loss": 1.4044551849365234, "step": 11324 }, { "epoch": 1.317779846404468, "grad_norm": 1.0581454038619995, "learning_rate": 0.0005, "loss": 1.9685239791870117, "step": 11325 }, { "epoch": 1.3178962066558064, "grad_norm": 0.9090920090675354, "learning_rate": 0.0005, "loss": 1.9400055408477783, "step": 11326 }, { "epoch": 1.3180125669071445, "grad_norm": 0.8712654709815979, "learning_rate": 0.0005, "loss": 1.7409255504608154, "step": 11327 }, { "epoch": 1.3181289271584826, "grad_norm": 1.9222873449325562, "learning_rate": 0.0005, "loss": 1.5038777589797974, "step": 11328 }, { "epoch": 1.318245287409821, "grad_norm": 1.5261270999908447, "learning_rate": 0.0005, "loss": 1.8916385173797607, "step": 11329 }, { "epoch": 1.3183616476611588, "grad_norm": 0.8956916332244873, "learning_rate": 0.0005, "loss": 1.9544169902801514, "step": 11330 }, { "epoch": 1.3184780079124971, "grad_norm": 0.9766532182693481, "learning_rate": 0.0005, "loss": 2.320720672607422, "step": 11331 }, { "epoch": 1.3185943681638352, "grad_norm": 1.0762662887573242, "learning_rate": 0.0005, "loss": 1.8956170082092285, "step": 11332 }, { "epoch": 1.3187107284151733, "grad_norm": 1.1246461868286133, "learning_rate": 0.0005, "loss": 1.960564136505127, "step": 11333 }, { "epoch": 1.3188270886665114, "grad_norm": 1.0599435567855835, "learning_rate": 0.0005, "loss": 2.136566162109375, "step": 11334 }, { "epoch": 1.3189434489178495, "grad_norm": 1.649035930633545, "learning_rate": 0.0005, "loss": 1.8835668563842773, "step": 11335 }, { "epoch": 1.3190598091691879, "grad_norm": 1.202438235282898, "learning_rate": 0.0005, "loss": 3.081071376800537, "step": 11336 }, { "epoch": 1.319176169420526, "grad_norm": 1.6160693168640137, "learning_rate": 0.0005, "loss": 1.7941042184829712, "step": 11337 }, { "epoch": 1.319292529671864, "grad_norm": 1.2017887830734253, "learning_rate": 0.0005, "loss": 1.5853486061096191, "step": 11338 }, { "epoch": 1.3194088899232022, "grad_norm": 1.237770915031433, "learning_rate": 0.0005, "loss": 2.012206554412842, "step": 11339 }, { "epoch": 1.3195252501745403, "grad_norm": 1.1433833837509155, "learning_rate": 0.0005, "loss": 1.8082377910614014, "step": 11340 }, { "epoch": 1.3196416104258786, "grad_norm": 1.2877339124679565, "learning_rate": 0.0005, "loss": 2.2200748920440674, "step": 11341 }, { "epoch": 1.3197579706772167, "grad_norm": 1.2265784740447998, "learning_rate": 0.0005, "loss": 2.478959560394287, "step": 11342 }, { "epoch": 1.3198743309285548, "grad_norm": 1.0762065649032593, "learning_rate": 0.0005, "loss": 2.43312931060791, "step": 11343 }, { "epoch": 1.3199906911798929, "grad_norm": 0.98589688539505, "learning_rate": 0.0005, "loss": 1.959273099899292, "step": 11344 }, { "epoch": 1.320107051431231, "grad_norm": 0.8932684659957886, "learning_rate": 0.0005, "loss": 1.5842987298965454, "step": 11345 }, { "epoch": 1.3202234116825693, "grad_norm": 1.699581503868103, "learning_rate": 0.0005, "loss": 2.320850133895874, "step": 11346 }, { "epoch": 1.3203397719339074, "grad_norm": 1.127363920211792, "learning_rate": 0.0005, "loss": 1.7165818214416504, "step": 11347 }, { "epoch": 1.3204561321852455, "grad_norm": 1.2518069744110107, "learning_rate": 0.0005, "loss": 1.8272322416305542, "step": 11348 }, { "epoch": 1.3205724924365836, "grad_norm": 1.2945202589035034, "learning_rate": 0.0005, "loss": 1.8930249214172363, "step": 11349 }, { "epoch": 1.3206888526879217, "grad_norm": 0.7280187606811523, "learning_rate": 0.0005, "loss": 1.528795838356018, "step": 11350 }, { "epoch": 1.32080521293926, "grad_norm": 1.038051724433899, "learning_rate": 0.0005, "loss": 1.7578598260879517, "step": 11351 }, { "epoch": 1.320921573190598, "grad_norm": 0.9455552101135254, "learning_rate": 0.0005, "loss": 1.5385711193084717, "step": 11352 }, { "epoch": 1.3210379334419362, "grad_norm": 1.480900764465332, "learning_rate": 0.0005, "loss": 1.834336280822754, "step": 11353 }, { "epoch": 1.3211542936932743, "grad_norm": 0.8637471795082092, "learning_rate": 0.0005, "loss": 1.6551332473754883, "step": 11354 }, { "epoch": 1.3212706539446124, "grad_norm": 1.328007459640503, "learning_rate": 0.0005, "loss": 2.2883214950561523, "step": 11355 }, { "epoch": 1.3213870141959507, "grad_norm": 1.0034645795822144, "learning_rate": 0.0005, "loss": 1.934211254119873, "step": 11356 }, { "epoch": 1.3215033744472888, "grad_norm": 1.4270153045654297, "learning_rate": 0.0005, "loss": 1.6411621570587158, "step": 11357 }, { "epoch": 1.321619734698627, "grad_norm": 1.5610986948013306, "learning_rate": 0.0005, "loss": 2.5687103271484375, "step": 11358 }, { "epoch": 1.321736094949965, "grad_norm": 1.1648024320602417, "learning_rate": 0.0005, "loss": 2.1453185081481934, "step": 11359 }, { "epoch": 1.3218524552013031, "grad_norm": 1.5146344900131226, "learning_rate": 0.0005, "loss": 2.026371717453003, "step": 11360 }, { "epoch": 1.3219688154526414, "grad_norm": 1.5782917737960815, "learning_rate": 0.0005, "loss": 2.049856424331665, "step": 11361 }, { "epoch": 1.3220851757039795, "grad_norm": 1.2995821237564087, "learning_rate": 0.0005, "loss": 1.998220443725586, "step": 11362 }, { "epoch": 1.3222015359553176, "grad_norm": 1.2735285758972168, "learning_rate": 0.0005, "loss": 2.438066005706787, "step": 11363 }, { "epoch": 1.3223178962066557, "grad_norm": 0.7371091842651367, "learning_rate": 0.0005, "loss": 1.727522611618042, "step": 11364 }, { "epoch": 1.3224342564579938, "grad_norm": 1.6992982625961304, "learning_rate": 0.0005, "loss": 1.694197654724121, "step": 11365 }, { "epoch": 1.3225506167093322, "grad_norm": 0.819244384765625, "learning_rate": 0.0005, "loss": 1.386925458908081, "step": 11366 }, { "epoch": 1.3226669769606703, "grad_norm": 1.1038333177566528, "learning_rate": 0.0005, "loss": 1.5862540006637573, "step": 11367 }, { "epoch": 1.3227833372120084, "grad_norm": 1.166908860206604, "learning_rate": 0.0005, "loss": 2.729182243347168, "step": 11368 }, { "epoch": 1.3228996974633465, "grad_norm": 0.6245754957199097, "learning_rate": 0.0005, "loss": 1.298136591911316, "step": 11369 }, { "epoch": 1.3230160577146846, "grad_norm": 0.6631609201431274, "learning_rate": 0.0005, "loss": 1.6166698932647705, "step": 11370 }, { "epoch": 1.3231324179660229, "grad_norm": 2.6503357887268066, "learning_rate": 0.0005, "loss": 2.68295955657959, "step": 11371 }, { "epoch": 1.323248778217361, "grad_norm": 2.343196392059326, "learning_rate": 0.0005, "loss": 2.293205976486206, "step": 11372 }, { "epoch": 1.323365138468699, "grad_norm": 1.330976963043213, "learning_rate": 0.0005, "loss": 2.5203263759613037, "step": 11373 }, { "epoch": 1.3234814987200372, "grad_norm": 0.8057281374931335, "learning_rate": 0.0005, "loss": 1.531319260597229, "step": 11374 }, { "epoch": 1.3235978589713753, "grad_norm": 1.2611228227615356, "learning_rate": 0.0005, "loss": 1.768959879875183, "step": 11375 }, { "epoch": 1.3237142192227136, "grad_norm": 0.5307143926620483, "learning_rate": 0.0005, "loss": 1.3890128135681152, "step": 11376 }, { "epoch": 1.3238305794740517, "grad_norm": 0.5570175051689148, "learning_rate": 0.0005, "loss": 1.3147597312927246, "step": 11377 }, { "epoch": 1.3239469397253898, "grad_norm": 1.0570110082626343, "learning_rate": 0.0005, "loss": 1.48966383934021, "step": 11378 }, { "epoch": 1.324063299976728, "grad_norm": 1.8118666410446167, "learning_rate": 0.0005, "loss": 2.3024582862854004, "step": 11379 }, { "epoch": 1.324179660228066, "grad_norm": 1.297208309173584, "learning_rate": 0.0005, "loss": 2.3243203163146973, "step": 11380 }, { "epoch": 1.3242960204794043, "grad_norm": 0.7299213409423828, "learning_rate": 0.0005, "loss": 1.242922067642212, "step": 11381 }, { "epoch": 1.3244123807307424, "grad_norm": 1.0624637603759766, "learning_rate": 0.0005, "loss": 2.4746036529541016, "step": 11382 }, { "epoch": 1.3245287409820805, "grad_norm": 0.789862334728241, "learning_rate": 0.0005, "loss": 1.7613863945007324, "step": 11383 }, { "epoch": 1.3246451012334186, "grad_norm": 1.424683928489685, "learning_rate": 0.0005, "loss": 1.7665139436721802, "step": 11384 }, { "epoch": 1.3247614614847567, "grad_norm": 1.4495506286621094, "learning_rate": 0.0005, "loss": 2.0651469230651855, "step": 11385 }, { "epoch": 1.324877821736095, "grad_norm": 1.1576099395751953, "learning_rate": 0.0005, "loss": 2.1460142135620117, "step": 11386 }, { "epoch": 1.3249941819874331, "grad_norm": 1.3653627634048462, "learning_rate": 0.0005, "loss": 2.26113224029541, "step": 11387 }, { "epoch": 1.3251105422387712, "grad_norm": 1.3971203565597534, "learning_rate": 0.0005, "loss": 1.8335562944412231, "step": 11388 }, { "epoch": 1.3252269024901093, "grad_norm": 1.1667534112930298, "learning_rate": 0.0005, "loss": 2.145073890686035, "step": 11389 }, { "epoch": 1.3253432627414474, "grad_norm": 1.5062307119369507, "learning_rate": 0.0005, "loss": 2.3284261226654053, "step": 11390 }, { "epoch": 1.3254596229927857, "grad_norm": 1.2172305583953857, "learning_rate": 0.0005, "loss": 2.349785804748535, "step": 11391 }, { "epoch": 1.3255759832441238, "grad_norm": 1.418199896812439, "learning_rate": 0.0005, "loss": 1.6930263042449951, "step": 11392 }, { "epoch": 1.325692343495462, "grad_norm": 1.2169960737228394, "learning_rate": 0.0005, "loss": 2.5957412719726562, "step": 11393 }, { "epoch": 1.3258087037468, "grad_norm": 1.5176087617874146, "learning_rate": 0.0005, "loss": 1.8421380519866943, "step": 11394 }, { "epoch": 1.3259250639981381, "grad_norm": 0.650344729423523, "learning_rate": 0.0005, "loss": 1.1761384010314941, "step": 11395 }, { "epoch": 1.3260414242494765, "grad_norm": 1.3189523220062256, "learning_rate": 0.0005, "loss": 1.6792956590652466, "step": 11396 }, { "epoch": 1.3261577845008146, "grad_norm": 0.9652311205863953, "learning_rate": 0.0005, "loss": 1.9188826084136963, "step": 11397 }, { "epoch": 1.3262741447521527, "grad_norm": 1.0860354900360107, "learning_rate": 0.0005, "loss": 1.5640586614608765, "step": 11398 }, { "epoch": 1.3263905050034908, "grad_norm": 0.8710451126098633, "learning_rate": 0.0005, "loss": 1.7025138139724731, "step": 11399 }, { "epoch": 1.3265068652548289, "grad_norm": 1.3893256187438965, "learning_rate": 0.0005, "loss": 2.692641019821167, "step": 11400 }, { "epoch": 1.3266232255061672, "grad_norm": 0.978234589099884, "learning_rate": 0.0005, "loss": 1.6472923755645752, "step": 11401 }, { "epoch": 1.3267395857575053, "grad_norm": 0.8367610573768616, "learning_rate": 0.0005, "loss": 1.710538387298584, "step": 11402 }, { "epoch": 1.3268559460088434, "grad_norm": 1.2551796436309814, "learning_rate": 0.0005, "loss": 1.8203800916671753, "step": 11403 }, { "epoch": 1.3269723062601815, "grad_norm": 0.9576721787452698, "learning_rate": 0.0005, "loss": 1.6973814964294434, "step": 11404 }, { "epoch": 1.3270886665115196, "grad_norm": 1.4197524785995483, "learning_rate": 0.0005, "loss": 2.303500175476074, "step": 11405 }, { "epoch": 1.327205026762858, "grad_norm": 0.9026312232017517, "learning_rate": 0.0005, "loss": 1.514302134513855, "step": 11406 }, { "epoch": 1.327321387014196, "grad_norm": 0.6750333309173584, "learning_rate": 0.0005, "loss": 1.2240442037582397, "step": 11407 }, { "epoch": 1.327437747265534, "grad_norm": 0.9021749496459961, "learning_rate": 0.0005, "loss": 2.031759738922119, "step": 11408 }, { "epoch": 1.3275541075168722, "grad_norm": 1.3031600713729858, "learning_rate": 0.0005, "loss": 2.4617650508880615, "step": 11409 }, { "epoch": 1.3276704677682103, "grad_norm": 0.9840618968009949, "learning_rate": 0.0005, "loss": 2.2096428871154785, "step": 11410 }, { "epoch": 1.3277868280195486, "grad_norm": 1.4480739831924438, "learning_rate": 0.0005, "loss": 2.4501748085021973, "step": 11411 }, { "epoch": 1.3279031882708867, "grad_norm": 1.3535100221633911, "learning_rate": 0.0005, "loss": 2.376363754272461, "step": 11412 }, { "epoch": 1.3280195485222248, "grad_norm": 1.2807846069335938, "learning_rate": 0.0005, "loss": 1.7375870943069458, "step": 11413 }, { "epoch": 1.328135908773563, "grad_norm": 0.8934677243232727, "learning_rate": 0.0005, "loss": 1.4200552701950073, "step": 11414 }, { "epoch": 1.328252269024901, "grad_norm": 0.6259615421295166, "learning_rate": 0.0005, "loss": 1.7384743690490723, "step": 11415 }, { "epoch": 1.3283686292762393, "grad_norm": 0.8557577133178711, "learning_rate": 0.0005, "loss": 1.572320580482483, "step": 11416 }, { "epoch": 1.3284849895275774, "grad_norm": 0.3862861096858978, "learning_rate": 0.0005, "loss": 1.0333530902862549, "step": 11417 }, { "epoch": 1.3286013497789155, "grad_norm": 1.3606667518615723, "learning_rate": 0.0005, "loss": 2.181276559829712, "step": 11418 }, { "epoch": 1.3287177100302536, "grad_norm": 1.2223066091537476, "learning_rate": 0.0005, "loss": 1.8986608982086182, "step": 11419 }, { "epoch": 1.3288340702815917, "grad_norm": 1.0743045806884766, "learning_rate": 0.0005, "loss": 2.7242321968078613, "step": 11420 }, { "epoch": 1.32895043053293, "grad_norm": 1.3276972770690918, "learning_rate": 0.0005, "loss": 2.153075933456421, "step": 11421 }, { "epoch": 1.3290667907842681, "grad_norm": 0.7493404150009155, "learning_rate": 0.0005, "loss": 2.0898404121398926, "step": 11422 }, { "epoch": 1.3291831510356062, "grad_norm": 1.2818827629089355, "learning_rate": 0.0005, "loss": 2.2606453895568848, "step": 11423 }, { "epoch": 1.3292995112869443, "grad_norm": 1.0610744953155518, "learning_rate": 0.0005, "loss": 1.722292184829712, "step": 11424 }, { "epoch": 1.3294158715382824, "grad_norm": 1.4276984930038452, "learning_rate": 0.0005, "loss": 2.1223878860473633, "step": 11425 }, { "epoch": 1.3295322317896208, "grad_norm": 1.5785408020019531, "learning_rate": 0.0005, "loss": 2.39739727973938, "step": 11426 }, { "epoch": 1.3296485920409589, "grad_norm": 0.9582995176315308, "learning_rate": 0.0005, "loss": 1.0453598499298096, "step": 11427 }, { "epoch": 1.329764952292297, "grad_norm": 1.165109634399414, "learning_rate": 0.0005, "loss": 1.9556024074554443, "step": 11428 }, { "epoch": 1.329881312543635, "grad_norm": 1.220124363899231, "learning_rate": 0.0005, "loss": 2.2741928100585938, "step": 11429 }, { "epoch": 1.3299976727949732, "grad_norm": 0.8022972941398621, "learning_rate": 0.0005, "loss": 2.1620287895202637, "step": 11430 }, { "epoch": 1.3301140330463115, "grad_norm": 1.2160121202468872, "learning_rate": 0.0005, "loss": 2.3087244033813477, "step": 11431 }, { "epoch": 1.3302303932976496, "grad_norm": 0.7212191224098206, "learning_rate": 0.0005, "loss": 1.7733783721923828, "step": 11432 }, { "epoch": 1.3303467535489877, "grad_norm": 0.730388879776001, "learning_rate": 0.0005, "loss": 1.3242833614349365, "step": 11433 }, { "epoch": 1.3304631138003258, "grad_norm": 0.8845384120941162, "learning_rate": 0.0005, "loss": 1.7224137783050537, "step": 11434 }, { "epoch": 1.3305794740516639, "grad_norm": 1.5492066144943237, "learning_rate": 0.0005, "loss": 1.9687669277191162, "step": 11435 }, { "epoch": 1.3306958343030022, "grad_norm": 0.9512984156608582, "learning_rate": 0.0005, "loss": 1.3092420101165771, "step": 11436 }, { "epoch": 1.3308121945543403, "grad_norm": 1.048248291015625, "learning_rate": 0.0005, "loss": 2.080695867538452, "step": 11437 }, { "epoch": 1.3309285548056784, "grad_norm": 1.0317018032073975, "learning_rate": 0.0005, "loss": 1.79573655128479, "step": 11438 }, { "epoch": 1.3310449150570165, "grad_norm": 0.5598382949829102, "learning_rate": 0.0005, "loss": 1.4274688959121704, "step": 11439 }, { "epoch": 1.3311612753083546, "grad_norm": 0.6717821359634399, "learning_rate": 0.0005, "loss": 1.3137385845184326, "step": 11440 }, { "epoch": 1.331277635559693, "grad_norm": 0.8810651302337646, "learning_rate": 0.0005, "loss": 1.9758186340332031, "step": 11441 }, { "epoch": 1.331393995811031, "grad_norm": 0.9786748290061951, "learning_rate": 0.0005, "loss": 1.824836015701294, "step": 11442 }, { "epoch": 1.3315103560623691, "grad_norm": 0.9143431782722473, "learning_rate": 0.0005, "loss": 2.3929123878479004, "step": 11443 }, { "epoch": 1.3316267163137072, "grad_norm": 1.0159703493118286, "learning_rate": 0.0005, "loss": 2.101280689239502, "step": 11444 }, { "epoch": 1.3317430765650453, "grad_norm": 1.1451940536499023, "learning_rate": 0.0005, "loss": 2.4981486797332764, "step": 11445 }, { "epoch": 1.3318594368163836, "grad_norm": 0.6481580138206482, "learning_rate": 0.0005, "loss": 1.9389538764953613, "step": 11446 }, { "epoch": 1.3319757970677217, "grad_norm": 1.375891089439392, "learning_rate": 0.0005, "loss": 1.7070057392120361, "step": 11447 }, { "epoch": 1.3320921573190598, "grad_norm": 0.9314395189285278, "learning_rate": 0.0005, "loss": 2.2449700832366943, "step": 11448 }, { "epoch": 1.332208517570398, "grad_norm": 1.5013949871063232, "learning_rate": 0.0005, "loss": 2.1944093704223633, "step": 11449 }, { "epoch": 1.332324877821736, "grad_norm": 1.206976294517517, "learning_rate": 0.0005, "loss": 2.314302921295166, "step": 11450 }, { "epoch": 1.3324412380730744, "grad_norm": 0.996589720249176, "learning_rate": 0.0005, "loss": 1.904826283454895, "step": 11451 }, { "epoch": 1.3325575983244125, "grad_norm": 0.8287182450294495, "learning_rate": 0.0005, "loss": 1.5063176155090332, "step": 11452 }, { "epoch": 1.3326739585757506, "grad_norm": 1.4837701320648193, "learning_rate": 0.0005, "loss": 2.7189319133758545, "step": 11453 }, { "epoch": 1.3327903188270886, "grad_norm": 1.0736033916473389, "learning_rate": 0.0005, "loss": 2.2978177070617676, "step": 11454 }, { "epoch": 1.3329066790784267, "grad_norm": 0.8872838616371155, "learning_rate": 0.0005, "loss": 2.0896995067596436, "step": 11455 }, { "epoch": 1.333023039329765, "grad_norm": 0.860453724861145, "learning_rate": 0.0005, "loss": 1.659766435623169, "step": 11456 }, { "epoch": 1.3331393995811032, "grad_norm": 0.5498002171516418, "learning_rate": 0.0005, "loss": 1.0303932428359985, "step": 11457 }, { "epoch": 1.3332557598324413, "grad_norm": 1.2551504373550415, "learning_rate": 0.0005, "loss": 1.8024218082427979, "step": 11458 }, { "epoch": 1.3333721200837794, "grad_norm": 1.1298503875732422, "learning_rate": 0.0005, "loss": 1.432288408279419, "step": 11459 }, { "epoch": 1.3334884803351175, "grad_norm": 0.9159054160118103, "learning_rate": 0.0005, "loss": 1.8123199939727783, "step": 11460 }, { "epoch": 1.3336048405864558, "grad_norm": 0.9012481570243835, "learning_rate": 0.0005, "loss": 1.6139476299285889, "step": 11461 }, { "epoch": 1.3337212008377939, "grad_norm": 0.9098804593086243, "learning_rate": 0.0005, "loss": 1.904930830001831, "step": 11462 }, { "epoch": 1.333837561089132, "grad_norm": 2.3619258403778076, "learning_rate": 0.0005, "loss": 1.9159550666809082, "step": 11463 }, { "epoch": 1.33395392134047, "grad_norm": 0.4104361832141876, "learning_rate": 0.0005, "loss": 1.0158839225769043, "step": 11464 }, { "epoch": 1.3340702815918082, "grad_norm": 1.1010454893112183, "learning_rate": 0.0005, "loss": 1.2712281942367554, "step": 11465 }, { "epoch": 1.3341866418431465, "grad_norm": 5.001888275146484, "learning_rate": 0.0005, "loss": 2.3005900382995605, "step": 11466 }, { "epoch": 1.3343030020944846, "grad_norm": 1.3850802183151245, "learning_rate": 0.0005, "loss": 1.8875821828842163, "step": 11467 }, { "epoch": 1.3344193623458227, "grad_norm": 0.5849584937095642, "learning_rate": 0.0005, "loss": 1.6346542835235596, "step": 11468 }, { "epoch": 1.3345357225971608, "grad_norm": 1.582040786743164, "learning_rate": 0.0005, "loss": 2.4304065704345703, "step": 11469 }, { "epoch": 1.334652082848499, "grad_norm": 0.8797380328178406, "learning_rate": 0.0005, "loss": 1.665693998336792, "step": 11470 }, { "epoch": 1.3347684430998372, "grad_norm": 1.1820530891418457, "learning_rate": 0.0005, "loss": 2.559278964996338, "step": 11471 }, { "epoch": 1.3348848033511753, "grad_norm": 0.9280850291252136, "learning_rate": 0.0005, "loss": 1.718156099319458, "step": 11472 }, { "epoch": 1.3350011636025134, "grad_norm": 1.3752623796463013, "learning_rate": 0.0005, "loss": 2.483041763305664, "step": 11473 }, { "epoch": 1.3351175238538515, "grad_norm": 1.1759604215621948, "learning_rate": 0.0005, "loss": 2.257713794708252, "step": 11474 }, { "epoch": 1.3352338841051896, "grad_norm": 1.3835760354995728, "learning_rate": 0.0005, "loss": 2.5593178272247314, "step": 11475 }, { "epoch": 1.335350244356528, "grad_norm": 0.958020806312561, "learning_rate": 0.0005, "loss": 1.2930988073349, "step": 11476 }, { "epoch": 1.3354666046078658, "grad_norm": 1.3735902309417725, "learning_rate": 0.0005, "loss": 2.165940999984741, "step": 11477 }, { "epoch": 1.3355829648592041, "grad_norm": 1.3299214839935303, "learning_rate": 0.0005, "loss": 1.5998635292053223, "step": 11478 }, { "epoch": 1.3356993251105422, "grad_norm": 1.78052818775177, "learning_rate": 0.0005, "loss": 1.8548657894134521, "step": 11479 }, { "epoch": 1.3358156853618803, "grad_norm": 1.317622184753418, "learning_rate": 0.0005, "loss": 2.4494595527648926, "step": 11480 }, { "epoch": 1.3359320456132187, "grad_norm": 1.1421321630477905, "learning_rate": 0.0005, "loss": 1.933159351348877, "step": 11481 }, { "epoch": 1.3360484058645565, "grad_norm": 0.8157709240913391, "learning_rate": 0.0005, "loss": 1.4549585580825806, "step": 11482 }, { "epoch": 1.3361647661158949, "grad_norm": 0.9699364304542542, "learning_rate": 0.0005, "loss": 1.579247236251831, "step": 11483 }, { "epoch": 1.336281126367233, "grad_norm": 0.6302566528320312, "learning_rate": 0.0005, "loss": 1.1354758739471436, "step": 11484 }, { "epoch": 1.336397486618571, "grad_norm": 1.5664808750152588, "learning_rate": 0.0005, "loss": 1.9560794830322266, "step": 11485 }, { "epoch": 1.3365138468699094, "grad_norm": 0.7952525019645691, "learning_rate": 0.0005, "loss": 1.7313172817230225, "step": 11486 }, { "epoch": 1.3366302071212472, "grad_norm": 0.7079125046730042, "learning_rate": 0.0005, "loss": 1.3508046865463257, "step": 11487 }, { "epoch": 1.3367465673725856, "grad_norm": 1.358850121498108, "learning_rate": 0.0005, "loss": 2.384688377380371, "step": 11488 }, { "epoch": 1.3368629276239237, "grad_norm": 1.4535856246948242, "learning_rate": 0.0005, "loss": 1.4615793228149414, "step": 11489 }, { "epoch": 1.3369792878752618, "grad_norm": 1.7332099676132202, "learning_rate": 0.0005, "loss": 1.1650813817977905, "step": 11490 }, { "epoch": 1.3370956481266, "grad_norm": 1.1538206338882446, "learning_rate": 0.0005, "loss": 2.0489120483398438, "step": 11491 }, { "epoch": 1.337212008377938, "grad_norm": 1.8012349605560303, "learning_rate": 0.0005, "loss": 2.2616782188415527, "step": 11492 }, { "epoch": 1.3373283686292763, "grad_norm": 1.2810288667678833, "learning_rate": 0.0005, "loss": 1.9716624021530151, "step": 11493 }, { "epoch": 1.3374447288806144, "grad_norm": 1.1258386373519897, "learning_rate": 0.0005, "loss": 1.7196857929229736, "step": 11494 }, { "epoch": 1.3375610891319525, "grad_norm": 1.0015056133270264, "learning_rate": 0.0005, "loss": 2.5244364738464355, "step": 11495 }, { "epoch": 1.3376774493832908, "grad_norm": 1.0457329750061035, "learning_rate": 0.0005, "loss": 1.7765285968780518, "step": 11496 }, { "epoch": 1.3377938096346287, "grad_norm": 0.9111689329147339, "learning_rate": 0.0005, "loss": 1.8800172805786133, "step": 11497 }, { "epoch": 1.337910169885967, "grad_norm": 1.0797337293624878, "learning_rate": 0.0005, "loss": 2.7802133560180664, "step": 11498 }, { "epoch": 1.338026530137305, "grad_norm": 1.912630319595337, "learning_rate": 0.0005, "loss": 2.648739814758301, "step": 11499 }, { "epoch": 1.3381428903886432, "grad_norm": 1.3730978965759277, "learning_rate": 0.0005, "loss": 2.151261806488037, "step": 11500 }, { "epoch": 1.3382592506399813, "grad_norm": 0.7888169884681702, "learning_rate": 0.0005, "loss": 2.1745352745056152, "step": 11501 }, { "epoch": 1.3383756108913194, "grad_norm": 0.7067517042160034, "learning_rate": 0.0005, "loss": 1.9278233051300049, "step": 11502 }, { "epoch": 1.3384919711426577, "grad_norm": 1.189268708229065, "learning_rate": 0.0005, "loss": 2.359409809112549, "step": 11503 }, { "epoch": 1.3386083313939958, "grad_norm": 1.316586971282959, "learning_rate": 0.0005, "loss": 2.3791730403900146, "step": 11504 }, { "epoch": 1.338724691645334, "grad_norm": 1.0991075038909912, "learning_rate": 0.0005, "loss": 1.586973786354065, "step": 11505 }, { "epoch": 1.338841051896672, "grad_norm": 1.2014614343643188, "learning_rate": 0.0005, "loss": 1.5192203521728516, "step": 11506 }, { "epoch": 1.3389574121480101, "grad_norm": 1.2473905086517334, "learning_rate": 0.0005, "loss": 1.9230467081069946, "step": 11507 }, { "epoch": 1.3390737723993484, "grad_norm": 1.00663423538208, "learning_rate": 0.0005, "loss": 1.418710708618164, "step": 11508 }, { "epoch": 1.3391901326506865, "grad_norm": 1.4933239221572876, "learning_rate": 0.0005, "loss": 2.133169651031494, "step": 11509 }, { "epoch": 1.3393064929020246, "grad_norm": 0.8673821091651917, "learning_rate": 0.0005, "loss": 1.4809834957122803, "step": 11510 }, { "epoch": 1.3394228531533627, "grad_norm": 1.1693295240402222, "learning_rate": 0.0005, "loss": 1.7983195781707764, "step": 11511 }, { "epoch": 1.3395392134047008, "grad_norm": 0.9690908193588257, "learning_rate": 0.0005, "loss": 2.3482069969177246, "step": 11512 }, { "epoch": 1.3396555736560392, "grad_norm": 1.324051856994629, "learning_rate": 0.0005, "loss": 2.282909870147705, "step": 11513 }, { "epoch": 1.3397719339073773, "grad_norm": 0.9392535090446472, "learning_rate": 0.0005, "loss": 1.7487996816635132, "step": 11514 }, { "epoch": 1.3398882941587154, "grad_norm": 1.1202220916748047, "learning_rate": 0.0005, "loss": 1.7468280792236328, "step": 11515 }, { "epoch": 1.3400046544100535, "grad_norm": 0.9618825912475586, "learning_rate": 0.0005, "loss": 2.205364465713501, "step": 11516 }, { "epoch": 1.3401210146613916, "grad_norm": 0.8825957179069519, "learning_rate": 0.0005, "loss": 1.6130584478378296, "step": 11517 }, { "epoch": 1.3402373749127299, "grad_norm": 0.7934668064117432, "learning_rate": 0.0005, "loss": 1.5153049230575562, "step": 11518 }, { "epoch": 1.340353735164068, "grad_norm": 0.7957031726837158, "learning_rate": 0.0005, "loss": 1.3956626653671265, "step": 11519 }, { "epoch": 1.340470095415406, "grad_norm": 1.3383369445800781, "learning_rate": 0.0005, "loss": 2.4417049884796143, "step": 11520 }, { "epoch": 1.3405864556667442, "grad_norm": 1.8373984098434448, "learning_rate": 0.0005, "loss": 2.1890311241149902, "step": 11521 }, { "epoch": 1.3407028159180823, "grad_norm": 1.0072362422943115, "learning_rate": 0.0005, "loss": 1.687620759010315, "step": 11522 }, { "epoch": 1.3408191761694206, "grad_norm": 1.0225073099136353, "learning_rate": 0.0005, "loss": 1.90548574924469, "step": 11523 }, { "epoch": 1.3409355364207587, "grad_norm": 1.3991538286209106, "learning_rate": 0.0005, "loss": 1.5556273460388184, "step": 11524 }, { "epoch": 1.3410518966720968, "grad_norm": 2.1027872562408447, "learning_rate": 0.0005, "loss": 2.606692314147949, "step": 11525 }, { "epoch": 1.3411682569234349, "grad_norm": 0.5865132212638855, "learning_rate": 0.0005, "loss": 1.022632360458374, "step": 11526 }, { "epoch": 1.341284617174773, "grad_norm": 1.4224061965942383, "learning_rate": 0.0005, "loss": 1.7334001064300537, "step": 11527 }, { "epoch": 1.3414009774261113, "grad_norm": 1.831514596939087, "learning_rate": 0.0005, "loss": 1.701606273651123, "step": 11528 }, { "epoch": 1.3415173376774494, "grad_norm": 1.3508965969085693, "learning_rate": 0.0005, "loss": 2.2982709407806396, "step": 11529 }, { "epoch": 1.3416336979287875, "grad_norm": 0.7040659189224243, "learning_rate": 0.0005, "loss": 1.9136017560958862, "step": 11530 }, { "epoch": 1.3417500581801256, "grad_norm": 0.965501070022583, "learning_rate": 0.0005, "loss": 2.1895978450775146, "step": 11531 }, { "epoch": 1.3418664184314637, "grad_norm": 1.0354270935058594, "learning_rate": 0.0005, "loss": 2.462672233581543, "step": 11532 }, { "epoch": 1.341982778682802, "grad_norm": 1.0955872535705566, "learning_rate": 0.0005, "loss": 2.5014073848724365, "step": 11533 }, { "epoch": 1.3420991389341401, "grad_norm": 1.0713132619857788, "learning_rate": 0.0005, "loss": 1.9163846969604492, "step": 11534 }, { "epoch": 1.3422154991854782, "grad_norm": 0.8689844012260437, "learning_rate": 0.0005, "loss": 1.8492205142974854, "step": 11535 }, { "epoch": 1.3423318594368163, "grad_norm": 2.144301652908325, "learning_rate": 0.0005, "loss": 2.1427981853485107, "step": 11536 }, { "epoch": 1.3424482196881544, "grad_norm": 1.1400463581085205, "learning_rate": 0.0005, "loss": 1.9212266206741333, "step": 11537 }, { "epoch": 1.3425645799394927, "grad_norm": 0.8286324143409729, "learning_rate": 0.0005, "loss": 1.288745403289795, "step": 11538 }, { "epoch": 1.3426809401908308, "grad_norm": 1.2424273490905762, "learning_rate": 0.0005, "loss": 1.6024280786514282, "step": 11539 }, { "epoch": 1.342797300442169, "grad_norm": 2.1453049182891846, "learning_rate": 0.0005, "loss": 1.8160457611083984, "step": 11540 }, { "epoch": 1.342913660693507, "grad_norm": 1.2531402111053467, "learning_rate": 0.0005, "loss": 2.292309522628784, "step": 11541 }, { "epoch": 1.3430300209448451, "grad_norm": 0.7360814213752747, "learning_rate": 0.0005, "loss": 1.6569976806640625, "step": 11542 }, { "epoch": 1.3431463811961835, "grad_norm": 1.3134297132492065, "learning_rate": 0.0005, "loss": 2.804379463195801, "step": 11543 }, { "epoch": 1.3432627414475216, "grad_norm": 1.1279345750808716, "learning_rate": 0.0005, "loss": 1.8621110916137695, "step": 11544 }, { "epoch": 1.3433791016988597, "grad_norm": 1.9158210754394531, "learning_rate": 0.0005, "loss": 1.8663907051086426, "step": 11545 }, { "epoch": 1.3434954619501978, "grad_norm": 1.264367699623108, "learning_rate": 0.0005, "loss": 1.7288818359375, "step": 11546 }, { "epoch": 1.3436118222015359, "grad_norm": 1.1036279201507568, "learning_rate": 0.0005, "loss": 1.9597737789154053, "step": 11547 }, { "epoch": 1.3437281824528742, "grad_norm": 1.192800760269165, "learning_rate": 0.0005, "loss": 1.4859352111816406, "step": 11548 }, { "epoch": 1.3438445427042123, "grad_norm": 1.1807374954223633, "learning_rate": 0.0005, "loss": 1.6565017700195312, "step": 11549 }, { "epoch": 1.3439609029555504, "grad_norm": 1.115087628364563, "learning_rate": 0.0005, "loss": 2.319355010986328, "step": 11550 }, { "epoch": 1.3440772632068885, "grad_norm": 0.9621771574020386, "learning_rate": 0.0005, "loss": 2.1389694213867188, "step": 11551 }, { "epoch": 1.3441936234582266, "grad_norm": 1.152222752571106, "learning_rate": 0.0005, "loss": 2.177173137664795, "step": 11552 }, { "epoch": 1.344309983709565, "grad_norm": 1.0072839260101318, "learning_rate": 0.0005, "loss": 1.9172608852386475, "step": 11553 }, { "epoch": 1.344426343960903, "grad_norm": 1.3792927265167236, "learning_rate": 0.0005, "loss": 2.4846839904785156, "step": 11554 }, { "epoch": 1.344542704212241, "grad_norm": 1.1914399862289429, "learning_rate": 0.0005, "loss": 2.2113919258117676, "step": 11555 }, { "epoch": 1.3446590644635792, "grad_norm": 1.331739068031311, "learning_rate": 0.0005, "loss": 2.0256412029266357, "step": 11556 }, { "epoch": 1.3447754247149173, "grad_norm": 0.8849365711212158, "learning_rate": 0.0005, "loss": 2.452371597290039, "step": 11557 }, { "epoch": 1.3448917849662556, "grad_norm": 1.205116629600525, "learning_rate": 0.0005, "loss": 1.73978853225708, "step": 11558 }, { "epoch": 1.3450081452175937, "grad_norm": 3.142997980117798, "learning_rate": 0.0005, "loss": 2.0542185306549072, "step": 11559 }, { "epoch": 1.3451245054689318, "grad_norm": 2.077453851699829, "learning_rate": 0.0005, "loss": 2.2678282260894775, "step": 11560 }, { "epoch": 1.34524086572027, "grad_norm": 1.022445797920227, "learning_rate": 0.0005, "loss": 1.3577492237091064, "step": 11561 }, { "epoch": 1.345357225971608, "grad_norm": 1.6524860858917236, "learning_rate": 0.0005, "loss": 2.406024932861328, "step": 11562 }, { "epoch": 1.3454735862229463, "grad_norm": 0.3984715938568115, "learning_rate": 0.0005, "loss": 1.0038585662841797, "step": 11563 }, { "epoch": 1.3455899464742844, "grad_norm": 1.2458105087280273, "learning_rate": 0.0005, "loss": 1.7026803493499756, "step": 11564 }, { "epoch": 1.3457063067256225, "grad_norm": 0.8510987758636475, "learning_rate": 0.0005, "loss": 1.637824535369873, "step": 11565 }, { "epoch": 1.3458226669769606, "grad_norm": 1.4186547994613647, "learning_rate": 0.0005, "loss": 2.5759663581848145, "step": 11566 }, { "epoch": 1.3459390272282987, "grad_norm": 1.2315014600753784, "learning_rate": 0.0005, "loss": 1.7653738260269165, "step": 11567 }, { "epoch": 1.346055387479637, "grad_norm": 1.242887258529663, "learning_rate": 0.0005, "loss": 2.2152199745178223, "step": 11568 }, { "epoch": 1.3461717477309751, "grad_norm": 0.8651190996170044, "learning_rate": 0.0005, "loss": 1.9764716625213623, "step": 11569 }, { "epoch": 1.3462881079823132, "grad_norm": 1.3075289726257324, "learning_rate": 0.0005, "loss": 2.3125789165496826, "step": 11570 }, { "epoch": 1.3464044682336513, "grad_norm": 2.21716046333313, "learning_rate": 0.0005, "loss": 1.3391633033752441, "step": 11571 }, { "epoch": 1.3465208284849894, "grad_norm": 0.9200653433799744, "learning_rate": 0.0005, "loss": 1.6453903913497925, "step": 11572 }, { "epoch": 1.3466371887363278, "grad_norm": 0.9456970691680908, "learning_rate": 0.0005, "loss": 1.7310943603515625, "step": 11573 }, { "epoch": 1.3467535489876659, "grad_norm": 1.4991471767425537, "learning_rate": 0.0005, "loss": 2.0700440406799316, "step": 11574 }, { "epoch": 1.346869909239004, "grad_norm": 1.3733932971954346, "learning_rate": 0.0005, "loss": 2.3917739391326904, "step": 11575 }, { "epoch": 1.346986269490342, "grad_norm": 1.10379958152771, "learning_rate": 0.0005, "loss": 1.2992595434188843, "step": 11576 }, { "epoch": 1.3471026297416802, "grad_norm": 1.4092406034469604, "learning_rate": 0.0005, "loss": 1.8172494173049927, "step": 11577 }, { "epoch": 1.3472189899930185, "grad_norm": 0.9679336547851562, "learning_rate": 0.0005, "loss": 1.9167051315307617, "step": 11578 }, { "epoch": 1.3473353502443566, "grad_norm": 1.0203920602798462, "learning_rate": 0.0005, "loss": 2.7022361755371094, "step": 11579 }, { "epoch": 1.3474517104956947, "grad_norm": 1.686693787574768, "learning_rate": 0.0005, "loss": 2.329127311706543, "step": 11580 }, { "epoch": 1.3475680707470328, "grad_norm": 1.270320177078247, "learning_rate": 0.0005, "loss": 1.88031005859375, "step": 11581 }, { "epoch": 1.3476844309983709, "grad_norm": 0.899612545967102, "learning_rate": 0.0005, "loss": 1.9922139644622803, "step": 11582 }, { "epoch": 1.3478007912497092, "grad_norm": 0.9876620769500732, "learning_rate": 0.0005, "loss": 1.5240811109542847, "step": 11583 }, { "epoch": 1.3479171515010473, "grad_norm": 1.2386800050735474, "learning_rate": 0.0005, "loss": 1.84739351272583, "step": 11584 }, { "epoch": 1.3480335117523854, "grad_norm": 1.0892609357833862, "learning_rate": 0.0005, "loss": 1.9328808784484863, "step": 11585 }, { "epoch": 1.3481498720037235, "grad_norm": 1.0001951456069946, "learning_rate": 0.0005, "loss": 1.575221300125122, "step": 11586 }, { "epoch": 1.3482662322550616, "grad_norm": 0.7968599200248718, "learning_rate": 0.0005, "loss": 1.6555461883544922, "step": 11587 }, { "epoch": 1.3483825925064, "grad_norm": 0.9391235113143921, "learning_rate": 0.0005, "loss": 1.4258553981781006, "step": 11588 }, { "epoch": 1.348498952757738, "grad_norm": 1.175125241279602, "learning_rate": 0.0005, "loss": 2.2302846908569336, "step": 11589 }, { "epoch": 1.348615313009076, "grad_norm": 1.281917929649353, "learning_rate": 0.0005, "loss": 1.8482670783996582, "step": 11590 }, { "epoch": 1.3487316732604142, "grad_norm": 1.6306790113449097, "learning_rate": 0.0005, "loss": 2.2555131912231445, "step": 11591 }, { "epoch": 1.3488480335117523, "grad_norm": 1.3414146900177002, "learning_rate": 0.0005, "loss": 1.765419602394104, "step": 11592 }, { "epoch": 1.3489643937630906, "grad_norm": 0.9726435542106628, "learning_rate": 0.0005, "loss": 2.2484490871429443, "step": 11593 }, { "epoch": 1.3490807540144287, "grad_norm": 1.9104257822036743, "learning_rate": 0.0005, "loss": 1.9973429441452026, "step": 11594 }, { "epoch": 1.3491971142657668, "grad_norm": 1.6286475658416748, "learning_rate": 0.0005, "loss": 3.270216464996338, "step": 11595 }, { "epoch": 1.349313474517105, "grad_norm": 1.3363910913467407, "learning_rate": 0.0005, "loss": 2.223452091217041, "step": 11596 }, { "epoch": 1.349429834768443, "grad_norm": 1.3809524774551392, "learning_rate": 0.0005, "loss": 1.9090464115142822, "step": 11597 }, { "epoch": 1.3495461950197813, "grad_norm": 0.795587420463562, "learning_rate": 0.0005, "loss": 1.5203593969345093, "step": 11598 }, { "epoch": 1.3496625552711194, "grad_norm": 0.5843817591667175, "learning_rate": 0.0005, "loss": 1.5450931787490845, "step": 11599 }, { "epoch": 1.3497789155224575, "grad_norm": 1.110974669456482, "learning_rate": 0.0005, "loss": 1.67055344581604, "step": 11600 }, { "epoch": 1.3498952757737956, "grad_norm": 1.7457572221755981, "learning_rate": 0.0005, "loss": 1.9814391136169434, "step": 11601 }, { "epoch": 1.3500116360251337, "grad_norm": 1.1514816284179688, "learning_rate": 0.0005, "loss": 2.454941749572754, "step": 11602 }, { "epoch": 1.350127996276472, "grad_norm": 0.6464001536369324, "learning_rate": 0.0005, "loss": 1.6855714321136475, "step": 11603 }, { "epoch": 1.3502443565278102, "grad_norm": 1.4445984363555908, "learning_rate": 0.0005, "loss": 1.99588942527771, "step": 11604 }, { "epoch": 1.3503607167791483, "grad_norm": 1.2397639751434326, "learning_rate": 0.0005, "loss": 2.082415819168091, "step": 11605 }, { "epoch": 1.3504770770304864, "grad_norm": 1.000783920288086, "learning_rate": 0.0005, "loss": 1.8906781673431396, "step": 11606 }, { "epoch": 1.3505934372818245, "grad_norm": 1.226485013961792, "learning_rate": 0.0005, "loss": 1.578745722770691, "step": 11607 }, { "epoch": 1.3507097975331628, "grad_norm": 2.5692965984344482, "learning_rate": 0.0005, "loss": 1.8316540718078613, "step": 11608 }, { "epoch": 1.3508261577845009, "grad_norm": 1.101272463798523, "learning_rate": 0.0005, "loss": 2.4809107780456543, "step": 11609 }, { "epoch": 1.350942518035839, "grad_norm": 1.668611764907837, "learning_rate": 0.0005, "loss": 1.537848949432373, "step": 11610 }, { "epoch": 1.351058878287177, "grad_norm": 0.9091026782989502, "learning_rate": 0.0005, "loss": 1.80901038646698, "step": 11611 }, { "epoch": 1.3511752385385152, "grad_norm": 1.433764100074768, "learning_rate": 0.0005, "loss": 1.7129623889923096, "step": 11612 }, { "epoch": 1.3512915987898535, "grad_norm": 1.1238219738006592, "learning_rate": 0.0005, "loss": 2.848639965057373, "step": 11613 }, { "epoch": 1.3514079590411916, "grad_norm": 1.1142828464508057, "learning_rate": 0.0005, "loss": 1.469266653060913, "step": 11614 }, { "epoch": 1.3515243192925297, "grad_norm": 1.8111200332641602, "learning_rate": 0.0005, "loss": 1.9701319932937622, "step": 11615 }, { "epoch": 1.3516406795438678, "grad_norm": 1.1812052726745605, "learning_rate": 0.0005, "loss": 1.8243954181671143, "step": 11616 }, { "epoch": 1.351757039795206, "grad_norm": 0.9291090965270996, "learning_rate": 0.0005, "loss": 1.2412666082382202, "step": 11617 }, { "epoch": 1.3518734000465442, "grad_norm": 1.072884440422058, "learning_rate": 0.0005, "loss": 1.9641735553741455, "step": 11618 }, { "epoch": 1.3519897602978823, "grad_norm": 1.6472413539886475, "learning_rate": 0.0005, "loss": 2.1848435401916504, "step": 11619 }, { "epoch": 1.3521061205492204, "grad_norm": 1.1737806797027588, "learning_rate": 0.0005, "loss": 1.8132860660552979, "step": 11620 }, { "epoch": 1.3522224808005585, "grad_norm": 1.2625230550765991, "learning_rate": 0.0005, "loss": 2.171579360961914, "step": 11621 }, { "epoch": 1.3523388410518966, "grad_norm": 0.7687036395072937, "learning_rate": 0.0005, "loss": 1.390962839126587, "step": 11622 }, { "epoch": 1.352455201303235, "grad_norm": 0.9679911732673645, "learning_rate": 0.0005, "loss": 2.094130516052246, "step": 11623 }, { "epoch": 1.352571561554573, "grad_norm": 0.7127274870872498, "learning_rate": 0.0005, "loss": 1.093032717704773, "step": 11624 }, { "epoch": 1.3526879218059111, "grad_norm": 1.3034992218017578, "learning_rate": 0.0005, "loss": 1.981529712677002, "step": 11625 }, { "epoch": 1.3528042820572492, "grad_norm": 1.119035243988037, "learning_rate": 0.0005, "loss": 2.2750449180603027, "step": 11626 }, { "epoch": 1.3529206423085873, "grad_norm": 0.7865897417068481, "learning_rate": 0.0005, "loss": 1.2158284187316895, "step": 11627 }, { "epoch": 1.3530370025599257, "grad_norm": 0.9297236204147339, "learning_rate": 0.0005, "loss": 1.4163312911987305, "step": 11628 }, { "epoch": 1.3531533628112637, "grad_norm": 0.9853190183639526, "learning_rate": 0.0005, "loss": 2.242741584777832, "step": 11629 }, { "epoch": 1.3532697230626018, "grad_norm": 1.2614861726760864, "learning_rate": 0.0005, "loss": 1.7381222248077393, "step": 11630 }, { "epoch": 1.35338608331394, "grad_norm": 1.187402606010437, "learning_rate": 0.0005, "loss": 1.8724603652954102, "step": 11631 }, { "epoch": 1.353502443565278, "grad_norm": 0.8840911984443665, "learning_rate": 0.0005, "loss": 1.867314338684082, "step": 11632 }, { "epoch": 1.3536188038166164, "grad_norm": 1.291624903678894, "learning_rate": 0.0005, "loss": 1.7851402759552002, "step": 11633 }, { "epoch": 1.3537351640679545, "grad_norm": 1.0250537395477295, "learning_rate": 0.0005, "loss": 1.8244624137878418, "step": 11634 }, { "epoch": 1.3538515243192926, "grad_norm": 1.3738600015640259, "learning_rate": 0.0005, "loss": 1.971299171447754, "step": 11635 }, { "epoch": 1.3539678845706307, "grad_norm": 1.1756638288497925, "learning_rate": 0.0005, "loss": 1.5827398300170898, "step": 11636 }, { "epoch": 1.3540842448219688, "grad_norm": 6.8727126121521, "learning_rate": 0.0005, "loss": 1.6679776906967163, "step": 11637 }, { "epoch": 1.354200605073307, "grad_norm": 1.1044244766235352, "learning_rate": 0.0005, "loss": 1.9114458560943604, "step": 11638 }, { "epoch": 1.3543169653246452, "grad_norm": 1.1021193265914917, "learning_rate": 0.0005, "loss": 1.9776862859725952, "step": 11639 }, { "epoch": 1.3544333255759833, "grad_norm": 0.9361802935600281, "learning_rate": 0.0005, "loss": 2.310417413711548, "step": 11640 }, { "epoch": 1.3545496858273214, "grad_norm": 0.7288767695426941, "learning_rate": 0.0005, "loss": 1.5150015354156494, "step": 11641 }, { "epoch": 1.3546660460786595, "grad_norm": 1.0133352279663086, "learning_rate": 0.0005, "loss": 1.616944670677185, "step": 11642 }, { "epoch": 1.3547824063299978, "grad_norm": 1.5362595319747925, "learning_rate": 0.0005, "loss": 2.3504109382629395, "step": 11643 }, { "epoch": 1.3548987665813357, "grad_norm": 1.0868074893951416, "learning_rate": 0.0005, "loss": 2.3260273933410645, "step": 11644 }, { "epoch": 1.355015126832674, "grad_norm": 1.5249704122543335, "learning_rate": 0.0005, "loss": 2.2998855113983154, "step": 11645 }, { "epoch": 1.355131487084012, "grad_norm": 1.2543224096298218, "learning_rate": 0.0005, "loss": 1.7696173191070557, "step": 11646 }, { "epoch": 1.3552478473353502, "grad_norm": 1.3255306482315063, "learning_rate": 0.0005, "loss": 1.9389891624450684, "step": 11647 }, { "epoch": 1.3553642075866885, "grad_norm": 1.1999841928482056, "learning_rate": 0.0005, "loss": 2.0737719535827637, "step": 11648 }, { "epoch": 1.3554805678380264, "grad_norm": 1.0845410823822021, "learning_rate": 0.0005, "loss": 1.1794320344924927, "step": 11649 }, { "epoch": 1.3555969280893647, "grad_norm": 1.3057795763015747, "learning_rate": 0.0005, "loss": 2.2197036743164062, "step": 11650 }, { "epoch": 1.3557132883407028, "grad_norm": 1.1872800588607788, "learning_rate": 0.0005, "loss": 2.004096746444702, "step": 11651 }, { "epoch": 1.355829648592041, "grad_norm": 1.127331018447876, "learning_rate": 0.0005, "loss": 1.8591569662094116, "step": 11652 }, { "epoch": 1.3559460088433792, "grad_norm": 1.482977271080017, "learning_rate": 0.0005, "loss": 2.6223068237304688, "step": 11653 }, { "epoch": 1.3560623690947171, "grad_norm": 1.1407657861709595, "learning_rate": 0.0005, "loss": 1.5874037742614746, "step": 11654 }, { "epoch": 1.3561787293460554, "grad_norm": 1.3294752836227417, "learning_rate": 0.0005, "loss": 1.5463870763778687, "step": 11655 }, { "epoch": 1.3562950895973935, "grad_norm": 1.1670451164245605, "learning_rate": 0.0005, "loss": 1.88541841506958, "step": 11656 }, { "epoch": 1.3564114498487316, "grad_norm": 1.0186965465545654, "learning_rate": 0.0005, "loss": 1.4485070705413818, "step": 11657 }, { "epoch": 1.35652781010007, "grad_norm": 1.0091475248336792, "learning_rate": 0.0005, "loss": 2.555389404296875, "step": 11658 }, { "epoch": 1.3566441703514078, "grad_norm": 1.212510347366333, "learning_rate": 0.0005, "loss": 1.786928415298462, "step": 11659 }, { "epoch": 1.3567605306027462, "grad_norm": 1.2433518171310425, "learning_rate": 0.0005, "loss": 1.9890971183776855, "step": 11660 }, { "epoch": 1.3568768908540842, "grad_norm": 1.3265737295150757, "learning_rate": 0.0005, "loss": 1.6643348932266235, "step": 11661 }, { "epoch": 1.3569932511054223, "grad_norm": 0.9902440905570984, "learning_rate": 0.0005, "loss": 2.1131021976470947, "step": 11662 }, { "epoch": 1.3571096113567604, "grad_norm": 0.9269476532936096, "learning_rate": 0.0005, "loss": 1.510652780532837, "step": 11663 }, { "epoch": 1.3572259716080985, "grad_norm": 0.761917233467102, "learning_rate": 0.0005, "loss": 1.216744303703308, "step": 11664 }, { "epoch": 1.3573423318594369, "grad_norm": 1.0756076574325562, "learning_rate": 0.0005, "loss": 2.2872495651245117, "step": 11665 }, { "epoch": 1.357458692110775, "grad_norm": 0.8033173680305481, "learning_rate": 0.0005, "loss": 1.6158794164657593, "step": 11666 }, { "epoch": 1.357575052362113, "grad_norm": 2.0566482543945312, "learning_rate": 0.0005, "loss": 2.069370985031128, "step": 11667 }, { "epoch": 1.3576914126134512, "grad_norm": 0.9397344589233398, "learning_rate": 0.0005, "loss": 1.832924723625183, "step": 11668 }, { "epoch": 1.3578077728647893, "grad_norm": 1.029049038887024, "learning_rate": 0.0005, "loss": 1.5944602489471436, "step": 11669 }, { "epoch": 1.3579241331161276, "grad_norm": 1.0780208110809326, "learning_rate": 0.0005, "loss": 1.411102056503296, "step": 11670 }, { "epoch": 1.3580404933674657, "grad_norm": 1.5481383800506592, "learning_rate": 0.0005, "loss": 2.2773730754852295, "step": 11671 }, { "epoch": 1.3581568536188038, "grad_norm": 1.9321755170822144, "learning_rate": 0.0005, "loss": 2.347160577774048, "step": 11672 }, { "epoch": 1.3582732138701419, "grad_norm": 0.9275152683258057, "learning_rate": 0.0005, "loss": 1.5897620916366577, "step": 11673 }, { "epoch": 1.35838957412148, "grad_norm": 0.30560022592544556, "learning_rate": 0.0005, "loss": 0.92965167760849, "step": 11674 }, { "epoch": 1.3585059343728183, "grad_norm": 0.880066990852356, "learning_rate": 0.0005, "loss": 1.767623782157898, "step": 11675 }, { "epoch": 1.3586222946241564, "grad_norm": 1.0569815635681152, "learning_rate": 0.0005, "loss": 1.7722810506820679, "step": 11676 }, { "epoch": 1.3587386548754945, "grad_norm": 0.9045547842979431, "learning_rate": 0.0005, "loss": 1.2476153373718262, "step": 11677 }, { "epoch": 1.3588550151268326, "grad_norm": 0.8713786005973816, "learning_rate": 0.0005, "loss": 1.5272948741912842, "step": 11678 }, { "epoch": 1.3589713753781707, "grad_norm": 1.0427378416061401, "learning_rate": 0.0005, "loss": 1.714680552482605, "step": 11679 }, { "epoch": 1.359087735629509, "grad_norm": 1.1338820457458496, "learning_rate": 0.0005, "loss": 2.476165771484375, "step": 11680 }, { "epoch": 1.3592040958808471, "grad_norm": 1.4170302152633667, "learning_rate": 0.0005, "loss": 2.35726261138916, "step": 11681 }, { "epoch": 1.3593204561321852, "grad_norm": 0.8982783555984497, "learning_rate": 0.0005, "loss": 1.342639684677124, "step": 11682 }, { "epoch": 1.3594368163835233, "grad_norm": 1.143707513809204, "learning_rate": 0.0005, "loss": 2.03944730758667, "step": 11683 }, { "epoch": 1.3595531766348614, "grad_norm": 1.271514654159546, "learning_rate": 0.0005, "loss": 2.1343421936035156, "step": 11684 }, { "epoch": 1.3596695368861997, "grad_norm": 1.1861993074417114, "learning_rate": 0.0005, "loss": 2.0793395042419434, "step": 11685 }, { "epoch": 1.3597858971375378, "grad_norm": 1.067350149154663, "learning_rate": 0.0005, "loss": 1.9897527694702148, "step": 11686 }, { "epoch": 1.359902257388876, "grad_norm": 0.9684837460517883, "learning_rate": 0.0005, "loss": 1.6867411136627197, "step": 11687 }, { "epoch": 1.360018617640214, "grad_norm": 1.283090353012085, "learning_rate": 0.0005, "loss": 1.7539126873016357, "step": 11688 }, { "epoch": 1.3601349778915521, "grad_norm": 1.0082231760025024, "learning_rate": 0.0005, "loss": 2.144847869873047, "step": 11689 }, { "epoch": 1.3602513381428905, "grad_norm": 1.0359299182891846, "learning_rate": 0.0005, "loss": 1.1579726934432983, "step": 11690 }, { "epoch": 1.3603676983942286, "grad_norm": 1.0857272148132324, "learning_rate": 0.0005, "loss": 1.7838830947875977, "step": 11691 }, { "epoch": 1.3604840586455667, "grad_norm": 2.63114595413208, "learning_rate": 0.0005, "loss": 1.8663972616195679, "step": 11692 }, { "epoch": 1.3606004188969048, "grad_norm": 1.1528161764144897, "learning_rate": 0.0005, "loss": 2.239753246307373, "step": 11693 }, { "epoch": 1.3607167791482428, "grad_norm": 1.1749452352523804, "learning_rate": 0.0005, "loss": 1.7130509614944458, "step": 11694 }, { "epoch": 1.3608331393995812, "grad_norm": 1.248290777206421, "learning_rate": 0.0005, "loss": 1.8653409481048584, "step": 11695 }, { "epoch": 1.3609494996509193, "grad_norm": 1.5703004598617554, "learning_rate": 0.0005, "loss": 1.804959774017334, "step": 11696 }, { "epoch": 1.3610658599022574, "grad_norm": 1.67190682888031, "learning_rate": 0.0005, "loss": 1.455277442932129, "step": 11697 }, { "epoch": 1.3611822201535955, "grad_norm": 1.1258611679077148, "learning_rate": 0.0005, "loss": 1.9517121315002441, "step": 11698 }, { "epoch": 1.3612985804049336, "grad_norm": 0.9099571704864502, "learning_rate": 0.0005, "loss": 2.2503786087036133, "step": 11699 }, { "epoch": 1.3614149406562719, "grad_norm": 0.9190863966941833, "learning_rate": 0.0005, "loss": 1.6422960758209229, "step": 11700 }, { "epoch": 1.36153130090761, "grad_norm": 1.319130539894104, "learning_rate": 0.0005, "loss": 3.0655717849731445, "step": 11701 }, { "epoch": 1.361647661158948, "grad_norm": 0.8706028461456299, "learning_rate": 0.0005, "loss": 1.5865672826766968, "step": 11702 }, { "epoch": 1.3617640214102862, "grad_norm": 1.1688934564590454, "learning_rate": 0.0005, "loss": 1.7512922286987305, "step": 11703 }, { "epoch": 1.3618803816616243, "grad_norm": 1.059430718421936, "learning_rate": 0.0005, "loss": 1.6819732189178467, "step": 11704 }, { "epoch": 1.3619967419129626, "grad_norm": 1.4117642641067505, "learning_rate": 0.0005, "loss": 2.364445209503174, "step": 11705 }, { "epoch": 1.3621131021643007, "grad_norm": 1.5627375841140747, "learning_rate": 0.0005, "loss": 1.5554287433624268, "step": 11706 }, { "epoch": 1.3622294624156388, "grad_norm": 1.022154450416565, "learning_rate": 0.0005, "loss": 1.4497809410095215, "step": 11707 }, { "epoch": 1.362345822666977, "grad_norm": 1.2311573028564453, "learning_rate": 0.0005, "loss": 2.3815855979919434, "step": 11708 }, { "epoch": 1.362462182918315, "grad_norm": 1.3323285579681396, "learning_rate": 0.0005, "loss": 2.4170284271240234, "step": 11709 }, { "epoch": 1.3625785431696533, "grad_norm": 1.0546224117279053, "learning_rate": 0.0005, "loss": 1.9030039310455322, "step": 11710 }, { "epoch": 1.3626949034209914, "grad_norm": 1.2778630256652832, "learning_rate": 0.0005, "loss": 1.865764856338501, "step": 11711 }, { "epoch": 1.3628112636723295, "grad_norm": 1.4055277109146118, "learning_rate": 0.0005, "loss": 2.632662057876587, "step": 11712 }, { "epoch": 1.3629276239236676, "grad_norm": 1.1943520307540894, "learning_rate": 0.0005, "loss": 1.8492830991744995, "step": 11713 }, { "epoch": 1.3630439841750057, "grad_norm": 0.8963519334793091, "learning_rate": 0.0005, "loss": 1.8355685472488403, "step": 11714 }, { "epoch": 1.363160344426344, "grad_norm": 1.2165303230285645, "learning_rate": 0.0005, "loss": 2.6053099632263184, "step": 11715 }, { "epoch": 1.3632767046776821, "grad_norm": 1.0217347145080566, "learning_rate": 0.0005, "loss": 2.0464587211608887, "step": 11716 }, { "epoch": 1.3633930649290202, "grad_norm": 0.859032928943634, "learning_rate": 0.0005, "loss": 1.9244604110717773, "step": 11717 }, { "epoch": 1.3635094251803583, "grad_norm": 1.1879762411117554, "learning_rate": 0.0005, "loss": 2.211796283721924, "step": 11718 }, { "epoch": 1.3636257854316964, "grad_norm": 1.3409777879714966, "learning_rate": 0.0005, "loss": 2.1782045364379883, "step": 11719 }, { "epoch": 1.3637421456830348, "grad_norm": 1.1558986902236938, "learning_rate": 0.0005, "loss": 1.857067584991455, "step": 11720 }, { "epoch": 1.3638585059343729, "grad_norm": 1.598249077796936, "learning_rate": 0.0005, "loss": 1.7167860269546509, "step": 11721 }, { "epoch": 1.363974866185711, "grad_norm": 0.929888129234314, "learning_rate": 0.0005, "loss": 2.9370200634002686, "step": 11722 }, { "epoch": 1.364091226437049, "grad_norm": 1.7398431301116943, "learning_rate": 0.0005, "loss": 1.8702760934829712, "step": 11723 }, { "epoch": 1.3642075866883872, "grad_norm": 1.1604835987091064, "learning_rate": 0.0005, "loss": 1.847554087638855, "step": 11724 }, { "epoch": 1.3643239469397255, "grad_norm": 0.7439526915550232, "learning_rate": 0.0005, "loss": 1.7693970203399658, "step": 11725 }, { "epoch": 1.3644403071910636, "grad_norm": 1.0029878616333008, "learning_rate": 0.0005, "loss": 1.487391710281372, "step": 11726 }, { "epoch": 1.3645566674424017, "grad_norm": 1.3291343450546265, "learning_rate": 0.0005, "loss": 2.3417162895202637, "step": 11727 }, { "epoch": 1.3646730276937398, "grad_norm": 1.0452476739883423, "learning_rate": 0.0005, "loss": 1.6604909896850586, "step": 11728 }, { "epoch": 1.3647893879450779, "grad_norm": 1.6970609426498413, "learning_rate": 0.0005, "loss": 1.971353530883789, "step": 11729 }, { "epoch": 1.3649057481964162, "grad_norm": 0.9160407185554504, "learning_rate": 0.0005, "loss": 1.0909221172332764, "step": 11730 }, { "epoch": 1.3650221084477543, "grad_norm": 1.1929429769515991, "learning_rate": 0.0005, "loss": 1.9173794984817505, "step": 11731 }, { "epoch": 1.3651384686990924, "grad_norm": 1.1459698677062988, "learning_rate": 0.0005, "loss": 2.265899658203125, "step": 11732 }, { "epoch": 1.3652548289504305, "grad_norm": 1.080796241760254, "learning_rate": 0.0005, "loss": 1.631643295288086, "step": 11733 }, { "epoch": 1.3653711892017686, "grad_norm": 0.9587097764015198, "learning_rate": 0.0005, "loss": 1.3533821105957031, "step": 11734 }, { "epoch": 1.365487549453107, "grad_norm": 1.2540034055709839, "learning_rate": 0.0005, "loss": 1.185797095298767, "step": 11735 }, { "epoch": 1.365603909704445, "grad_norm": 1.6191074848175049, "learning_rate": 0.0005, "loss": 1.7136900424957275, "step": 11736 }, { "epoch": 1.365720269955783, "grad_norm": 1.532837986946106, "learning_rate": 0.0005, "loss": 2.2434139251708984, "step": 11737 }, { "epoch": 1.3658366302071212, "grad_norm": 1.350029706954956, "learning_rate": 0.0005, "loss": 1.8917717933654785, "step": 11738 }, { "epoch": 1.3659529904584593, "grad_norm": 0.9541910886764526, "learning_rate": 0.0005, "loss": 1.9920332431793213, "step": 11739 }, { "epoch": 1.3660693507097976, "grad_norm": 1.4077574014663696, "learning_rate": 0.0005, "loss": 2.2544968128204346, "step": 11740 }, { "epoch": 1.3661857109611357, "grad_norm": 1.2205148935317993, "learning_rate": 0.0005, "loss": 1.7520462274551392, "step": 11741 }, { "epoch": 1.3663020712124738, "grad_norm": 1.142215609550476, "learning_rate": 0.0005, "loss": 2.3698792457580566, "step": 11742 }, { "epoch": 1.366418431463812, "grad_norm": 1.6111875772476196, "learning_rate": 0.0005, "loss": 2.3455655574798584, "step": 11743 }, { "epoch": 1.36653479171515, "grad_norm": 1.1820805072784424, "learning_rate": 0.0005, "loss": 1.241700530052185, "step": 11744 }, { "epoch": 1.3666511519664883, "grad_norm": 0.693078339099884, "learning_rate": 0.0005, "loss": 1.2805143594741821, "step": 11745 }, { "epoch": 1.3667675122178264, "grad_norm": 1.2895493507385254, "learning_rate": 0.0005, "loss": 2.3363726139068604, "step": 11746 }, { "epoch": 1.3668838724691645, "grad_norm": 0.856740415096283, "learning_rate": 0.0005, "loss": 1.3277931213378906, "step": 11747 }, { "epoch": 1.3670002327205026, "grad_norm": 1.1103715896606445, "learning_rate": 0.0005, "loss": 2.083134889602661, "step": 11748 }, { "epoch": 1.3671165929718407, "grad_norm": 1.0692541599273682, "learning_rate": 0.0005, "loss": 1.731025218963623, "step": 11749 }, { "epoch": 1.367232953223179, "grad_norm": 0.9373646378517151, "learning_rate": 0.0005, "loss": 1.933678150177002, "step": 11750 }, { "epoch": 1.3673493134745172, "grad_norm": 0.8752546906471252, "learning_rate": 0.0005, "loss": 1.52069091796875, "step": 11751 }, { "epoch": 1.3674656737258553, "grad_norm": 1.0721220970153809, "learning_rate": 0.0005, "loss": 1.7302274703979492, "step": 11752 }, { "epoch": 1.3675820339771934, "grad_norm": 1.6433714628219604, "learning_rate": 0.0005, "loss": 2.7112128734588623, "step": 11753 }, { "epoch": 1.3676983942285315, "grad_norm": 1.3020979166030884, "learning_rate": 0.0005, "loss": 1.5637531280517578, "step": 11754 }, { "epoch": 1.3678147544798698, "grad_norm": 0.8232213258743286, "learning_rate": 0.0005, "loss": 1.7009212970733643, "step": 11755 }, { "epoch": 1.3679311147312079, "grad_norm": 1.3124479055404663, "learning_rate": 0.0005, "loss": 2.481598138809204, "step": 11756 }, { "epoch": 1.368047474982546, "grad_norm": 1.7558858394622803, "learning_rate": 0.0005, "loss": 1.7060438394546509, "step": 11757 }, { "epoch": 1.368163835233884, "grad_norm": 1.8383187055587769, "learning_rate": 0.0005, "loss": 2.35420560836792, "step": 11758 }, { "epoch": 1.3682801954852222, "grad_norm": 0.6437139511108398, "learning_rate": 0.0005, "loss": 1.2004075050354004, "step": 11759 }, { "epoch": 1.3683965557365605, "grad_norm": 0.7527384757995605, "learning_rate": 0.0005, "loss": 1.7385387420654297, "step": 11760 }, { "epoch": 1.3685129159878986, "grad_norm": 0.9418526887893677, "learning_rate": 0.0005, "loss": 2.1793999671936035, "step": 11761 }, { "epoch": 1.3686292762392367, "grad_norm": 1.1199718713760376, "learning_rate": 0.0005, "loss": 2.0922160148620605, "step": 11762 }, { "epoch": 1.3687456364905748, "grad_norm": 0.9470729827880859, "learning_rate": 0.0005, "loss": 2.166790723800659, "step": 11763 }, { "epoch": 1.368861996741913, "grad_norm": 1.1783305406570435, "learning_rate": 0.0005, "loss": 2.3523197174072266, "step": 11764 }, { "epoch": 1.3689783569932512, "grad_norm": 1.0241585969924927, "learning_rate": 0.0005, "loss": 2.0786356925964355, "step": 11765 }, { "epoch": 1.3690947172445893, "grad_norm": 0.8532206416130066, "learning_rate": 0.0005, "loss": 2.4139885902404785, "step": 11766 }, { "epoch": 1.3692110774959274, "grad_norm": 1.482969045639038, "learning_rate": 0.0005, "loss": 2.2364487648010254, "step": 11767 }, { "epoch": 1.3693274377472655, "grad_norm": 1.1498218774795532, "learning_rate": 0.0005, "loss": 1.8500837087631226, "step": 11768 }, { "epoch": 1.3694437979986036, "grad_norm": 0.9833841919898987, "learning_rate": 0.0005, "loss": 1.7993005514144897, "step": 11769 }, { "epoch": 1.369560158249942, "grad_norm": 0.8823962807655334, "learning_rate": 0.0005, "loss": 1.4148955345153809, "step": 11770 }, { "epoch": 1.36967651850128, "grad_norm": 1.175379991531372, "learning_rate": 0.0005, "loss": 2.3581271171569824, "step": 11771 }, { "epoch": 1.3697928787526181, "grad_norm": 1.112627387046814, "learning_rate": 0.0005, "loss": 2.1966183185577393, "step": 11772 }, { "epoch": 1.3699092390039562, "grad_norm": 1.2754253149032593, "learning_rate": 0.0005, "loss": 1.5738245248794556, "step": 11773 }, { "epoch": 1.3700255992552943, "grad_norm": 1.1071960926055908, "learning_rate": 0.0005, "loss": 2.224205493927002, "step": 11774 }, { "epoch": 1.3701419595066326, "grad_norm": 1.2233761548995972, "learning_rate": 0.0005, "loss": 2.070193290710449, "step": 11775 }, { "epoch": 1.3702583197579707, "grad_norm": 1.0025511980056763, "learning_rate": 0.0005, "loss": 1.8337143659591675, "step": 11776 }, { "epoch": 1.3703746800093088, "grad_norm": 1.623692512512207, "learning_rate": 0.0005, "loss": 2.342778444290161, "step": 11777 }, { "epoch": 1.370491040260647, "grad_norm": 1.222359299659729, "learning_rate": 0.0005, "loss": 1.9794938564300537, "step": 11778 }, { "epoch": 1.370607400511985, "grad_norm": 1.533342719078064, "learning_rate": 0.0005, "loss": 1.8620308637619019, "step": 11779 }, { "epoch": 1.3707237607633234, "grad_norm": 0.6030145287513733, "learning_rate": 0.0005, "loss": 1.708937168121338, "step": 11780 }, { "epoch": 1.3708401210146615, "grad_norm": 1.3141101598739624, "learning_rate": 0.0005, "loss": 1.6157435178756714, "step": 11781 }, { "epoch": 1.3709564812659996, "grad_norm": 1.3729912042617798, "learning_rate": 0.0005, "loss": 1.7936967611312866, "step": 11782 }, { "epoch": 1.3710728415173377, "grad_norm": 1.1309235095977783, "learning_rate": 0.0005, "loss": 1.7150962352752686, "step": 11783 }, { "epoch": 1.3711892017686758, "grad_norm": 1.3335413932800293, "learning_rate": 0.0005, "loss": 2.068246364593506, "step": 11784 }, { "epoch": 1.371305562020014, "grad_norm": 0.9055866003036499, "learning_rate": 0.0005, "loss": 1.6027209758758545, "step": 11785 }, { "epoch": 1.3714219222713522, "grad_norm": 1.3103781938552856, "learning_rate": 0.0005, "loss": 1.7625470161437988, "step": 11786 }, { "epoch": 1.3715382825226903, "grad_norm": 1.1455222368240356, "learning_rate": 0.0005, "loss": 2.0630362033843994, "step": 11787 }, { "epoch": 1.3716546427740284, "grad_norm": 1.0711508989334106, "learning_rate": 0.0005, "loss": 2.071739673614502, "step": 11788 }, { "epoch": 1.3717710030253665, "grad_norm": 0.7893151044845581, "learning_rate": 0.0005, "loss": 1.663654088973999, "step": 11789 }, { "epoch": 1.3718873632767048, "grad_norm": 1.1602293252944946, "learning_rate": 0.0005, "loss": 1.5331529378890991, "step": 11790 }, { "epoch": 1.372003723528043, "grad_norm": 1.6971566677093506, "learning_rate": 0.0005, "loss": 1.5985329151153564, "step": 11791 }, { "epoch": 1.372120083779381, "grad_norm": 0.6983996033668518, "learning_rate": 0.0005, "loss": 1.4629114866256714, "step": 11792 }, { "epoch": 1.372236444030719, "grad_norm": 1.1445306539535522, "learning_rate": 0.0005, "loss": 2.3736300468444824, "step": 11793 }, { "epoch": 1.3723528042820572, "grad_norm": 0.6872996687889099, "learning_rate": 0.0005, "loss": 1.3038312196731567, "step": 11794 }, { "epoch": 1.3724691645333955, "grad_norm": 1.6609492301940918, "learning_rate": 0.0005, "loss": 2.058274984359741, "step": 11795 }, { "epoch": 1.3725855247847336, "grad_norm": 0.8715193271636963, "learning_rate": 0.0005, "loss": 2.4426381587982178, "step": 11796 }, { "epoch": 1.3727018850360717, "grad_norm": 1.3595601320266724, "learning_rate": 0.0005, "loss": 1.8825550079345703, "step": 11797 }, { "epoch": 1.3728182452874098, "grad_norm": 1.785218596458435, "learning_rate": 0.0005, "loss": 2.5659775733947754, "step": 11798 }, { "epoch": 1.372934605538748, "grad_norm": 1.6555447578430176, "learning_rate": 0.0005, "loss": 2.353297233581543, "step": 11799 }, { "epoch": 1.3730509657900862, "grad_norm": 0.7008298635482788, "learning_rate": 0.0005, "loss": 1.489565134048462, "step": 11800 }, { "epoch": 1.3731673260414243, "grad_norm": 1.6264511346817017, "learning_rate": 0.0005, "loss": 1.7075369358062744, "step": 11801 }, { "epoch": 1.3732836862927624, "grad_norm": 1.2127156257629395, "learning_rate": 0.0005, "loss": 1.5064725875854492, "step": 11802 }, { "epoch": 1.3734000465441005, "grad_norm": 1.349082350730896, "learning_rate": 0.0005, "loss": 2.445798635482788, "step": 11803 }, { "epoch": 1.3735164067954386, "grad_norm": 0.9166944026947021, "learning_rate": 0.0005, "loss": 1.3762238025665283, "step": 11804 }, { "epoch": 1.373632767046777, "grad_norm": 1.4475597143173218, "learning_rate": 0.0005, "loss": 2.1244797706604004, "step": 11805 }, { "epoch": 1.373749127298115, "grad_norm": 1.3431288003921509, "learning_rate": 0.0005, "loss": 2.113187789916992, "step": 11806 }, { "epoch": 1.3738654875494531, "grad_norm": 1.2321346998214722, "learning_rate": 0.0005, "loss": 1.240330696105957, "step": 11807 }, { "epoch": 1.3739818478007912, "grad_norm": 1.063262939453125, "learning_rate": 0.0005, "loss": 1.4910516738891602, "step": 11808 }, { "epoch": 1.3740982080521293, "grad_norm": 1.8875912427902222, "learning_rate": 0.0005, "loss": 2.20766544342041, "step": 11809 }, { "epoch": 1.3742145683034677, "grad_norm": 1.5770936012268066, "learning_rate": 0.0005, "loss": 1.6515699625015259, "step": 11810 }, { "epoch": 1.3743309285548055, "grad_norm": 1.1242648363113403, "learning_rate": 0.0005, "loss": 1.5256526470184326, "step": 11811 }, { "epoch": 1.3744472888061439, "grad_norm": 1.5709701776504517, "learning_rate": 0.0005, "loss": 2.3924150466918945, "step": 11812 }, { "epoch": 1.374563649057482, "grad_norm": 0.9171364903450012, "learning_rate": 0.0005, "loss": 1.4629662036895752, "step": 11813 }, { "epoch": 1.37468000930882, "grad_norm": 1.3460983037948608, "learning_rate": 0.0005, "loss": 1.8169028759002686, "step": 11814 }, { "epoch": 1.3747963695601584, "grad_norm": 1.2894147634506226, "learning_rate": 0.0005, "loss": 2.1364593505859375, "step": 11815 }, { "epoch": 1.3749127298114963, "grad_norm": 1.1374061107635498, "learning_rate": 0.0005, "loss": 2.088622570037842, "step": 11816 }, { "epoch": 1.3750290900628346, "grad_norm": 1.036241888999939, "learning_rate": 0.0005, "loss": 1.401261806488037, "step": 11817 }, { "epoch": 1.3751454503141727, "grad_norm": 1.2080100774765015, "learning_rate": 0.0005, "loss": 1.527977705001831, "step": 11818 }, { "epoch": 1.3752618105655108, "grad_norm": 1.4106972217559814, "learning_rate": 0.0005, "loss": 2.0174639225006104, "step": 11819 }, { "epoch": 1.375378170816849, "grad_norm": 1.1126725673675537, "learning_rate": 0.0005, "loss": 2.210232734680176, "step": 11820 }, { "epoch": 1.375494531068187, "grad_norm": 0.6061375737190247, "learning_rate": 0.0005, "loss": 1.773882508277893, "step": 11821 }, { "epoch": 1.3756108913195253, "grad_norm": 1.1144397258758545, "learning_rate": 0.0005, "loss": 2.2114577293395996, "step": 11822 }, { "epoch": 1.3757272515708634, "grad_norm": 1.112950325012207, "learning_rate": 0.0005, "loss": 2.218702793121338, "step": 11823 }, { "epoch": 1.3758436118222015, "grad_norm": 0.6065866947174072, "learning_rate": 0.0005, "loss": 1.3032885789871216, "step": 11824 }, { "epoch": 1.3759599720735398, "grad_norm": 1.204195261001587, "learning_rate": 0.0005, "loss": 1.7801685333251953, "step": 11825 }, { "epoch": 1.3760763323248777, "grad_norm": 1.2748397588729858, "learning_rate": 0.0005, "loss": 1.75257408618927, "step": 11826 }, { "epoch": 1.376192692576216, "grad_norm": 0.9069448709487915, "learning_rate": 0.0005, "loss": 2.165742874145508, "step": 11827 }, { "epoch": 1.3763090528275541, "grad_norm": 1.1764131784439087, "learning_rate": 0.0005, "loss": 1.904630184173584, "step": 11828 }, { "epoch": 1.3764254130788922, "grad_norm": 3.1256892681121826, "learning_rate": 0.0005, "loss": 1.6596288681030273, "step": 11829 }, { "epoch": 1.3765417733302303, "grad_norm": 0.8232569694519043, "learning_rate": 0.0005, "loss": 1.505860686302185, "step": 11830 }, { "epoch": 1.3766581335815684, "grad_norm": 0.39947718381881714, "learning_rate": 0.0005, "loss": 1.096903920173645, "step": 11831 }, { "epoch": 1.3767744938329067, "grad_norm": 1.0207363367080688, "learning_rate": 0.0005, "loss": 1.6841306686401367, "step": 11832 }, { "epoch": 1.3768908540842448, "grad_norm": 1.1187118291854858, "learning_rate": 0.0005, "loss": 1.8587157726287842, "step": 11833 }, { "epoch": 1.377007214335583, "grad_norm": 0.996056854724884, "learning_rate": 0.0005, "loss": 1.4075307846069336, "step": 11834 }, { "epoch": 1.377123574586921, "grad_norm": 1.2382104396820068, "learning_rate": 0.0005, "loss": 2.2321009635925293, "step": 11835 }, { "epoch": 1.3772399348382591, "grad_norm": 1.3789530992507935, "learning_rate": 0.0005, "loss": 2.2363574504852295, "step": 11836 }, { "epoch": 1.3773562950895974, "grad_norm": 1.3466449975967407, "learning_rate": 0.0005, "loss": 2.0623364448547363, "step": 11837 }, { "epoch": 1.3774726553409355, "grad_norm": 1.2999048233032227, "learning_rate": 0.0005, "loss": 2.024858236312866, "step": 11838 }, { "epoch": 1.3775890155922736, "grad_norm": 1.9538636207580566, "learning_rate": 0.0005, "loss": 2.6926794052124023, "step": 11839 }, { "epoch": 1.3777053758436117, "grad_norm": 1.777700662612915, "learning_rate": 0.0005, "loss": 2.1548705101013184, "step": 11840 }, { "epoch": 1.3778217360949498, "grad_norm": 1.480005145072937, "learning_rate": 0.0005, "loss": 1.6347460746765137, "step": 11841 }, { "epoch": 1.3779380963462882, "grad_norm": 1.113806128501892, "learning_rate": 0.0005, "loss": 3.1344666481018066, "step": 11842 }, { "epoch": 1.3780544565976263, "grad_norm": 0.43992534279823303, "learning_rate": 0.0005, "loss": 0.9437655210494995, "step": 11843 }, { "epoch": 1.3781708168489644, "grad_norm": 1.1096936464309692, "learning_rate": 0.0005, "loss": 2.4106292724609375, "step": 11844 }, { "epoch": 1.3782871771003025, "grad_norm": 0.9173380732536316, "learning_rate": 0.0005, "loss": 2.3414833545684814, "step": 11845 }, { "epoch": 1.3784035373516406, "grad_norm": 1.1763052940368652, "learning_rate": 0.0005, "loss": 2.8080434799194336, "step": 11846 }, { "epoch": 1.3785198976029789, "grad_norm": 1.1484712362289429, "learning_rate": 0.0005, "loss": 1.3176121711730957, "step": 11847 }, { "epoch": 1.378636257854317, "grad_norm": 0.8972233533859253, "learning_rate": 0.0005, "loss": 1.7698702812194824, "step": 11848 }, { "epoch": 1.378752618105655, "grad_norm": 1.9353331327438354, "learning_rate": 0.0005, "loss": 1.8518282175064087, "step": 11849 }, { "epoch": 1.3788689783569932, "grad_norm": 0.8840627074241638, "learning_rate": 0.0005, "loss": 1.504813551902771, "step": 11850 }, { "epoch": 1.3789853386083313, "grad_norm": 2.8247857093811035, "learning_rate": 0.0005, "loss": 2.4809913635253906, "step": 11851 }, { "epoch": 1.3791016988596696, "grad_norm": 1.2213644981384277, "learning_rate": 0.0005, "loss": 1.8224380016326904, "step": 11852 }, { "epoch": 1.3792180591110077, "grad_norm": 0.7767882347106934, "learning_rate": 0.0005, "loss": 1.4028418064117432, "step": 11853 }, { "epoch": 1.3793344193623458, "grad_norm": 1.0080574750900269, "learning_rate": 0.0005, "loss": 1.9607384204864502, "step": 11854 }, { "epoch": 1.379450779613684, "grad_norm": 1.3009843826293945, "learning_rate": 0.0005, "loss": 2.7420735359191895, "step": 11855 }, { "epoch": 1.379567139865022, "grad_norm": 2.7974603176116943, "learning_rate": 0.0005, "loss": 2.613166570663452, "step": 11856 }, { "epoch": 1.3796835001163603, "grad_norm": 1.4049984216690063, "learning_rate": 0.0005, "loss": 1.930394172668457, "step": 11857 }, { "epoch": 1.3797998603676984, "grad_norm": 0.9964177012443542, "learning_rate": 0.0005, "loss": 1.590271234512329, "step": 11858 }, { "epoch": 1.3799162206190365, "grad_norm": 1.013662338256836, "learning_rate": 0.0005, "loss": 1.8244624137878418, "step": 11859 }, { "epoch": 1.3800325808703746, "grad_norm": 1.0963436365127563, "learning_rate": 0.0005, "loss": 1.37080979347229, "step": 11860 }, { "epoch": 1.3801489411217127, "grad_norm": 0.9959232211112976, "learning_rate": 0.0005, "loss": 2.233501434326172, "step": 11861 }, { "epoch": 1.380265301373051, "grad_norm": 1.830162525177002, "learning_rate": 0.0005, "loss": 1.798860788345337, "step": 11862 }, { "epoch": 1.3803816616243891, "grad_norm": 1.1401746273040771, "learning_rate": 0.0005, "loss": 1.503232717514038, "step": 11863 }, { "epoch": 1.3804980218757272, "grad_norm": 1.2496882677078247, "learning_rate": 0.0005, "loss": 1.719982385635376, "step": 11864 }, { "epoch": 1.3806143821270653, "grad_norm": 0.8142140507698059, "learning_rate": 0.0005, "loss": 1.3814738988876343, "step": 11865 }, { "epoch": 1.3807307423784034, "grad_norm": 1.531303882598877, "learning_rate": 0.0005, "loss": 2.615635395050049, "step": 11866 }, { "epoch": 1.3808471026297418, "grad_norm": 0.9889023900032043, "learning_rate": 0.0005, "loss": 2.2160489559173584, "step": 11867 }, { "epoch": 1.3809634628810799, "grad_norm": 1.1200827360153198, "learning_rate": 0.0005, "loss": 2.951592445373535, "step": 11868 }, { "epoch": 1.381079823132418, "grad_norm": 1.233803629875183, "learning_rate": 0.0005, "loss": 1.78910493850708, "step": 11869 }, { "epoch": 1.381196183383756, "grad_norm": 7.9498114585876465, "learning_rate": 0.0005, "loss": 2.2581374645233154, "step": 11870 }, { "epoch": 1.3813125436350941, "grad_norm": 1.9978948831558228, "learning_rate": 0.0005, "loss": 2.2208786010742188, "step": 11871 }, { "epoch": 1.3814289038864325, "grad_norm": 1.1709489822387695, "learning_rate": 0.0005, "loss": 2.8966188430786133, "step": 11872 }, { "epoch": 1.3815452641377706, "grad_norm": 0.9133275151252747, "learning_rate": 0.0005, "loss": 1.8318524360656738, "step": 11873 }, { "epoch": 1.3816616243891087, "grad_norm": 1.4232704639434814, "learning_rate": 0.0005, "loss": 1.4562242031097412, "step": 11874 }, { "epoch": 1.3817779846404468, "grad_norm": 1.4252524375915527, "learning_rate": 0.0005, "loss": 1.8779966831207275, "step": 11875 }, { "epoch": 1.3818943448917849, "grad_norm": 1.4271546602249146, "learning_rate": 0.0005, "loss": 2.891622543334961, "step": 11876 }, { "epoch": 1.3820107051431232, "grad_norm": 1.0049411058425903, "learning_rate": 0.0005, "loss": 1.852600336074829, "step": 11877 }, { "epoch": 1.3821270653944613, "grad_norm": 0.5802159309387207, "learning_rate": 0.0005, "loss": 1.4544117450714111, "step": 11878 }, { "epoch": 1.3822434256457994, "grad_norm": 1.111852765083313, "learning_rate": 0.0005, "loss": 1.9649897813796997, "step": 11879 }, { "epoch": 1.3823597858971375, "grad_norm": 1.0433259010314941, "learning_rate": 0.0005, "loss": 1.8469762802124023, "step": 11880 }, { "epoch": 1.3824761461484756, "grad_norm": 0.9881801009178162, "learning_rate": 0.0005, "loss": 2.200871229171753, "step": 11881 }, { "epoch": 1.382592506399814, "grad_norm": 2.599362373352051, "learning_rate": 0.0005, "loss": 1.827871322631836, "step": 11882 }, { "epoch": 1.382708866651152, "grad_norm": 1.1467317342758179, "learning_rate": 0.0005, "loss": 2.149223566055298, "step": 11883 }, { "epoch": 1.38282522690249, "grad_norm": 0.6385611891746521, "learning_rate": 0.0005, "loss": 1.3661847114562988, "step": 11884 }, { "epoch": 1.3829415871538282, "grad_norm": 1.2278107404708862, "learning_rate": 0.0005, "loss": 2.379110336303711, "step": 11885 }, { "epoch": 1.3830579474051663, "grad_norm": 0.8731955885887146, "learning_rate": 0.0005, "loss": 1.798393964767456, "step": 11886 }, { "epoch": 1.3831743076565046, "grad_norm": 1.1264793872833252, "learning_rate": 0.0005, "loss": 2.9376308917999268, "step": 11887 }, { "epoch": 1.3832906679078427, "grad_norm": 1.171359896659851, "learning_rate": 0.0005, "loss": 2.467837333679199, "step": 11888 }, { "epoch": 1.3834070281591808, "grad_norm": 1.4650509357452393, "learning_rate": 0.0005, "loss": 1.5724977254867554, "step": 11889 }, { "epoch": 1.383523388410519, "grad_norm": 1.0182946920394897, "learning_rate": 0.0005, "loss": 1.5724422931671143, "step": 11890 }, { "epoch": 1.383639748661857, "grad_norm": 0.8760738372802734, "learning_rate": 0.0005, "loss": 2.009697198867798, "step": 11891 }, { "epoch": 1.3837561089131953, "grad_norm": 1.0814807415008545, "learning_rate": 0.0005, "loss": 2.3306941986083984, "step": 11892 }, { "epoch": 1.3838724691645334, "grad_norm": 1.3746849298477173, "learning_rate": 0.0005, "loss": 2.1581168174743652, "step": 11893 }, { "epoch": 1.3839888294158715, "grad_norm": 0.91953444480896, "learning_rate": 0.0005, "loss": 2.417616605758667, "step": 11894 }, { "epoch": 1.3841051896672096, "grad_norm": 1.476165771484375, "learning_rate": 0.0005, "loss": 1.984039068222046, "step": 11895 }, { "epoch": 1.3842215499185477, "grad_norm": 1.4018850326538086, "learning_rate": 0.0005, "loss": 1.439657211303711, "step": 11896 }, { "epoch": 1.384337910169886, "grad_norm": 1.0154058933258057, "learning_rate": 0.0005, "loss": 2.6438183784484863, "step": 11897 }, { "epoch": 1.3844542704212242, "grad_norm": 0.9869478940963745, "learning_rate": 0.0005, "loss": 1.4106935262680054, "step": 11898 }, { "epoch": 1.3845706306725623, "grad_norm": 1.1275423765182495, "learning_rate": 0.0005, "loss": 2.6146445274353027, "step": 11899 }, { "epoch": 1.3846869909239004, "grad_norm": 0.8800764679908752, "learning_rate": 0.0005, "loss": 1.9502569437026978, "step": 11900 }, { "epoch": 1.3848033511752385, "grad_norm": 1.4584813117980957, "learning_rate": 0.0005, "loss": 2.146519660949707, "step": 11901 }, { "epoch": 1.3849197114265768, "grad_norm": 1.410073161125183, "learning_rate": 0.0005, "loss": 2.4170804023742676, "step": 11902 }, { "epoch": 1.3850360716779149, "grad_norm": 1.2059510946273804, "learning_rate": 0.0005, "loss": 1.7780369520187378, "step": 11903 }, { "epoch": 1.385152431929253, "grad_norm": 1.1745429039001465, "learning_rate": 0.0005, "loss": 1.64730703830719, "step": 11904 }, { "epoch": 1.385268792180591, "grad_norm": 0.9795333743095398, "learning_rate": 0.0005, "loss": 1.4026219844818115, "step": 11905 }, { "epoch": 1.3853851524319292, "grad_norm": 1.1382938623428345, "learning_rate": 0.0005, "loss": 1.6417293548583984, "step": 11906 }, { "epoch": 1.3855015126832675, "grad_norm": 1.1834521293640137, "learning_rate": 0.0005, "loss": 2.458547830581665, "step": 11907 }, { "epoch": 1.3856178729346056, "grad_norm": 0.7760950326919556, "learning_rate": 0.0005, "loss": 1.335730791091919, "step": 11908 }, { "epoch": 1.3857342331859437, "grad_norm": 0.7100525498390198, "learning_rate": 0.0005, "loss": 1.359922170639038, "step": 11909 }, { "epoch": 1.3858505934372818, "grad_norm": 0.9058483242988586, "learning_rate": 0.0005, "loss": 2.094893455505371, "step": 11910 }, { "epoch": 1.3859669536886199, "grad_norm": 1.2039037942886353, "learning_rate": 0.0005, "loss": 1.6358877420425415, "step": 11911 }, { "epoch": 1.3860833139399582, "grad_norm": 0.7513308525085449, "learning_rate": 0.0005, "loss": 1.9123709201812744, "step": 11912 }, { "epoch": 1.3861996741912963, "grad_norm": 0.9468382596969604, "learning_rate": 0.0005, "loss": 1.8556832075119019, "step": 11913 }, { "epoch": 1.3863160344426344, "grad_norm": 1.1003639698028564, "learning_rate": 0.0005, "loss": 1.7267613410949707, "step": 11914 }, { "epoch": 1.3864323946939725, "grad_norm": 1.1744366884231567, "learning_rate": 0.0005, "loss": 2.1340808868408203, "step": 11915 }, { "epoch": 1.3865487549453106, "grad_norm": 1.4020920991897583, "learning_rate": 0.0005, "loss": 1.8951250314712524, "step": 11916 }, { "epoch": 1.386665115196649, "grad_norm": 1.091356635093689, "learning_rate": 0.0005, "loss": 1.54898202419281, "step": 11917 }, { "epoch": 1.386781475447987, "grad_norm": 0.9583336114883423, "learning_rate": 0.0005, "loss": 2.4969582557678223, "step": 11918 }, { "epoch": 1.3868978356993251, "grad_norm": 1.252681851387024, "learning_rate": 0.0005, "loss": 1.437934160232544, "step": 11919 }, { "epoch": 1.3870141959506632, "grad_norm": 0.9273810386657715, "learning_rate": 0.0005, "loss": 2.1469056606292725, "step": 11920 }, { "epoch": 1.3871305562020013, "grad_norm": 1.5234088897705078, "learning_rate": 0.0005, "loss": 2.4957807064056396, "step": 11921 }, { "epoch": 1.3872469164533396, "grad_norm": 1.3048185110092163, "learning_rate": 0.0005, "loss": 1.9390652179718018, "step": 11922 }, { "epoch": 1.3873632767046777, "grad_norm": 1.0680241584777832, "learning_rate": 0.0005, "loss": 1.8202025890350342, "step": 11923 }, { "epoch": 1.3874796369560158, "grad_norm": 1.4491708278656006, "learning_rate": 0.0005, "loss": 1.6277031898498535, "step": 11924 }, { "epoch": 1.387595997207354, "grad_norm": 1.343401312828064, "learning_rate": 0.0005, "loss": 1.911959171295166, "step": 11925 }, { "epoch": 1.387712357458692, "grad_norm": 1.2529202699661255, "learning_rate": 0.0005, "loss": 2.3216776847839355, "step": 11926 }, { "epoch": 1.3878287177100304, "grad_norm": 0.8095467686653137, "learning_rate": 0.0005, "loss": 1.283520221710205, "step": 11927 }, { "epoch": 1.3879450779613685, "grad_norm": 1.6883599758148193, "learning_rate": 0.0005, "loss": 2.028148651123047, "step": 11928 }, { "epoch": 1.3880614382127066, "grad_norm": 0.9949870109558105, "learning_rate": 0.0005, "loss": 1.7444508075714111, "step": 11929 }, { "epoch": 1.3881777984640447, "grad_norm": 0.7558394074440002, "learning_rate": 0.0005, "loss": 1.5252256393432617, "step": 11930 }, { "epoch": 1.3882941587153828, "grad_norm": 0.9806411862373352, "learning_rate": 0.0005, "loss": 1.7883944511413574, "step": 11931 }, { "epoch": 1.388410518966721, "grad_norm": 1.069045901298523, "learning_rate": 0.0005, "loss": 2.1703670024871826, "step": 11932 }, { "epoch": 1.3885268792180592, "grad_norm": 1.4263174533843994, "learning_rate": 0.0005, "loss": 2.536525249481201, "step": 11933 }, { "epoch": 1.3886432394693973, "grad_norm": 1.3299920558929443, "learning_rate": 0.0005, "loss": 1.5145342350006104, "step": 11934 }, { "epoch": 1.3887595997207354, "grad_norm": 1.234621524810791, "learning_rate": 0.0005, "loss": 1.8521523475646973, "step": 11935 }, { "epoch": 1.3888759599720735, "grad_norm": 1.1324453353881836, "learning_rate": 0.0005, "loss": 2.5065369606018066, "step": 11936 }, { "epoch": 1.3889923202234118, "grad_norm": 1.1199842691421509, "learning_rate": 0.0005, "loss": 2.127836227416992, "step": 11937 }, { "epoch": 1.38910868047475, "grad_norm": 1.3506780862808228, "learning_rate": 0.0005, "loss": 2.3035848140716553, "step": 11938 }, { "epoch": 1.389225040726088, "grad_norm": 1.0164363384246826, "learning_rate": 0.0005, "loss": 1.9190986156463623, "step": 11939 }, { "epoch": 1.389341400977426, "grad_norm": 1.1370177268981934, "learning_rate": 0.0005, "loss": 1.663459062576294, "step": 11940 }, { "epoch": 1.3894577612287642, "grad_norm": 1.1260266304016113, "learning_rate": 0.0005, "loss": 1.592320442199707, "step": 11941 }, { "epoch": 1.3895741214801025, "grad_norm": 1.1052275896072388, "learning_rate": 0.0005, "loss": 2.1609625816345215, "step": 11942 }, { "epoch": 1.3896904817314406, "grad_norm": 1.1028691530227661, "learning_rate": 0.0005, "loss": 2.024048328399658, "step": 11943 }, { "epoch": 1.3898068419827787, "grad_norm": 1.847106695175171, "learning_rate": 0.0005, "loss": 2.732456922531128, "step": 11944 }, { "epoch": 1.3899232022341168, "grad_norm": 1.0511105060577393, "learning_rate": 0.0005, "loss": 2.2647595405578613, "step": 11945 }, { "epoch": 1.390039562485455, "grad_norm": 1.191409707069397, "learning_rate": 0.0005, "loss": 1.5830209255218506, "step": 11946 }, { "epoch": 1.3901559227367932, "grad_norm": 1.2400330305099487, "learning_rate": 0.0005, "loss": 2.124666452407837, "step": 11947 }, { "epoch": 1.3902722829881313, "grad_norm": 1.1908024549484253, "learning_rate": 0.0005, "loss": 2.5612330436706543, "step": 11948 }, { "epoch": 1.3903886432394694, "grad_norm": 1.5732766389846802, "learning_rate": 0.0005, "loss": 1.9544365406036377, "step": 11949 }, { "epoch": 1.3905050034908075, "grad_norm": 1.237664818763733, "learning_rate": 0.0005, "loss": 1.9063191413879395, "step": 11950 }, { "epoch": 1.3906213637421456, "grad_norm": 0.6773868203163147, "learning_rate": 0.0005, "loss": 1.5704255104064941, "step": 11951 }, { "epoch": 1.390737723993484, "grad_norm": 0.8625361919403076, "learning_rate": 0.0005, "loss": 1.7415423393249512, "step": 11952 }, { "epoch": 1.390854084244822, "grad_norm": 0.8999035358428955, "learning_rate": 0.0005, "loss": 2.1202101707458496, "step": 11953 }, { "epoch": 1.3909704444961601, "grad_norm": 1.2531852722167969, "learning_rate": 0.0005, "loss": 1.7600469589233398, "step": 11954 }, { "epoch": 1.3910868047474982, "grad_norm": 1.0090725421905518, "learning_rate": 0.0005, "loss": 1.6845232248306274, "step": 11955 }, { "epoch": 1.3912031649988363, "grad_norm": 0.7016327977180481, "learning_rate": 0.0005, "loss": 1.134850025177002, "step": 11956 }, { "epoch": 1.3913195252501747, "grad_norm": 0.9417238235473633, "learning_rate": 0.0005, "loss": 2.085151433944702, "step": 11957 }, { "epoch": 1.3914358855015128, "grad_norm": 1.5802664756774902, "learning_rate": 0.0005, "loss": 1.7451022863388062, "step": 11958 }, { "epoch": 1.3915522457528509, "grad_norm": 0.8027724623680115, "learning_rate": 0.0005, "loss": 1.442313551902771, "step": 11959 }, { "epoch": 1.391668606004189, "grad_norm": 1.343784213066101, "learning_rate": 0.0005, "loss": 2.3389694690704346, "step": 11960 }, { "epoch": 1.391784966255527, "grad_norm": 0.8329843282699585, "learning_rate": 0.0005, "loss": 1.4882829189300537, "step": 11961 }, { "epoch": 1.3919013265068654, "grad_norm": 1.5346266031265259, "learning_rate": 0.0005, "loss": 2.1743533611297607, "step": 11962 }, { "epoch": 1.3920176867582035, "grad_norm": 1.0113006830215454, "learning_rate": 0.0005, "loss": 1.862162709236145, "step": 11963 }, { "epoch": 1.3921340470095416, "grad_norm": 1.354536771774292, "learning_rate": 0.0005, "loss": 2.174539566040039, "step": 11964 }, { "epoch": 1.3922504072608797, "grad_norm": 1.1745929718017578, "learning_rate": 0.0005, "loss": 1.9483258724212646, "step": 11965 }, { "epoch": 1.3923667675122178, "grad_norm": 0.779075562953949, "learning_rate": 0.0005, "loss": 1.692340612411499, "step": 11966 }, { "epoch": 1.392483127763556, "grad_norm": 1.939949870109558, "learning_rate": 0.0005, "loss": 2.8352670669555664, "step": 11967 }, { "epoch": 1.3925994880148942, "grad_norm": 1.1946640014648438, "learning_rate": 0.0005, "loss": 1.3252007961273193, "step": 11968 }, { "epoch": 1.3927158482662323, "grad_norm": 1.2136150598526, "learning_rate": 0.0005, "loss": 2.3594188690185547, "step": 11969 }, { "epoch": 1.3928322085175704, "grad_norm": 1.6490933895111084, "learning_rate": 0.0005, "loss": 2.308574676513672, "step": 11970 }, { "epoch": 1.3929485687689085, "grad_norm": 1.1391048431396484, "learning_rate": 0.0005, "loss": 1.781606912612915, "step": 11971 }, { "epoch": 1.3930649290202468, "grad_norm": 1.0331848859786987, "learning_rate": 0.0005, "loss": 1.4664198160171509, "step": 11972 }, { "epoch": 1.3931812892715847, "grad_norm": 1.3248491287231445, "learning_rate": 0.0005, "loss": 1.9602915048599243, "step": 11973 }, { "epoch": 1.393297649522923, "grad_norm": 1.0540060997009277, "learning_rate": 0.0005, "loss": 2.183720350265503, "step": 11974 }, { "epoch": 1.393414009774261, "grad_norm": 1.2076009511947632, "learning_rate": 0.0005, "loss": 2.262721538543701, "step": 11975 }, { "epoch": 1.3935303700255992, "grad_norm": 1.6045856475830078, "learning_rate": 0.0005, "loss": 1.7156949043273926, "step": 11976 }, { "epoch": 1.3936467302769375, "grad_norm": 1.2677124738693237, "learning_rate": 0.0005, "loss": 2.4171218872070312, "step": 11977 }, { "epoch": 1.3937630905282754, "grad_norm": 0.9937112331390381, "learning_rate": 0.0005, "loss": 2.3409664630889893, "step": 11978 }, { "epoch": 1.3938794507796137, "grad_norm": 1.290388822555542, "learning_rate": 0.0005, "loss": 1.7441316843032837, "step": 11979 }, { "epoch": 1.3939958110309518, "grad_norm": 1.25343656539917, "learning_rate": 0.0005, "loss": 1.9881473779678345, "step": 11980 }, { "epoch": 1.39411217128229, "grad_norm": 0.8321143388748169, "learning_rate": 0.0005, "loss": 2.265692710876465, "step": 11981 }, { "epoch": 1.3942285315336282, "grad_norm": 1.4058876037597656, "learning_rate": 0.0005, "loss": 1.801269292831421, "step": 11982 }, { "epoch": 1.3943448917849661, "grad_norm": 1.0298131704330444, "learning_rate": 0.0005, "loss": 1.8672220706939697, "step": 11983 }, { "epoch": 1.3944612520363044, "grad_norm": 1.0453847646713257, "learning_rate": 0.0005, "loss": 1.556809425354004, "step": 11984 }, { "epoch": 1.3945776122876425, "grad_norm": 1.0307250022888184, "learning_rate": 0.0005, "loss": 1.697913408279419, "step": 11985 }, { "epoch": 1.3946939725389806, "grad_norm": 1.2402907609939575, "learning_rate": 0.0005, "loss": 2.249636173248291, "step": 11986 }, { "epoch": 1.394810332790319, "grad_norm": 0.8960959315299988, "learning_rate": 0.0005, "loss": 1.9576280117034912, "step": 11987 }, { "epoch": 1.3949266930416568, "grad_norm": 1.059768557548523, "learning_rate": 0.0005, "loss": 2.084439754486084, "step": 11988 }, { "epoch": 1.3950430532929952, "grad_norm": 1.4089714288711548, "learning_rate": 0.0005, "loss": 2.3830933570861816, "step": 11989 }, { "epoch": 1.3951594135443333, "grad_norm": 1.0707647800445557, "learning_rate": 0.0005, "loss": 1.5801513195037842, "step": 11990 }, { "epoch": 1.3952757737956714, "grad_norm": 1.0105184316635132, "learning_rate": 0.0005, "loss": 1.4684257507324219, "step": 11991 }, { "epoch": 1.3953921340470097, "grad_norm": 1.0189205408096313, "learning_rate": 0.0005, "loss": 1.5515620708465576, "step": 11992 }, { "epoch": 1.3955084942983476, "grad_norm": 0.7883567810058594, "learning_rate": 0.0005, "loss": 1.3360183238983154, "step": 11993 }, { "epoch": 1.3956248545496859, "grad_norm": 1.254320740699768, "learning_rate": 0.0005, "loss": 2.164595603942871, "step": 11994 }, { "epoch": 1.395741214801024, "grad_norm": 1.2064870595932007, "learning_rate": 0.0005, "loss": 1.9016172885894775, "step": 11995 }, { "epoch": 1.395857575052362, "grad_norm": 0.9401193261146545, "learning_rate": 0.0005, "loss": 2.1764473915100098, "step": 11996 }, { "epoch": 1.3959739353037002, "grad_norm": 1.1994835138320923, "learning_rate": 0.0005, "loss": 1.9713656902313232, "step": 11997 }, { "epoch": 1.3960902955550383, "grad_norm": 1.1504145860671997, "learning_rate": 0.0005, "loss": 2.2404584884643555, "step": 11998 }, { "epoch": 1.3962066558063766, "grad_norm": 1.2680425643920898, "learning_rate": 0.0005, "loss": 2.4471683502197266, "step": 11999 }, { "epoch": 1.3963230160577147, "grad_norm": 1.3957003355026245, "learning_rate": 0.0005, "loss": 1.9281659126281738, "step": 12000 }, { "epoch": 1.3964393763090528, "grad_norm": 1.6675509214401245, "learning_rate": 0.0005, "loss": 1.954600214958191, "step": 12001 }, { "epoch": 1.396555736560391, "grad_norm": 0.7330928444862366, "learning_rate": 0.0005, "loss": 1.478177785873413, "step": 12002 }, { "epoch": 1.396672096811729, "grad_norm": 0.6690261960029602, "learning_rate": 0.0005, "loss": 1.2008721828460693, "step": 12003 }, { "epoch": 1.3967884570630673, "grad_norm": 1.211679458618164, "learning_rate": 0.0005, "loss": 2.115722179412842, "step": 12004 }, { "epoch": 1.3969048173144054, "grad_norm": 1.0510543584823608, "learning_rate": 0.0005, "loss": 1.8216075897216797, "step": 12005 }, { "epoch": 1.3970211775657435, "grad_norm": 0.9752001762390137, "learning_rate": 0.0005, "loss": 2.0201399326324463, "step": 12006 }, { "epoch": 1.3971375378170816, "grad_norm": 1.0996999740600586, "learning_rate": 0.0005, "loss": 1.655881643295288, "step": 12007 }, { "epoch": 1.3972538980684197, "grad_norm": 1.2959092855453491, "learning_rate": 0.0005, "loss": 1.102354645729065, "step": 12008 }, { "epoch": 1.397370258319758, "grad_norm": 1.224789023399353, "learning_rate": 0.0005, "loss": 1.7399578094482422, "step": 12009 }, { "epoch": 1.3974866185710961, "grad_norm": 1.2319303750991821, "learning_rate": 0.0005, "loss": 2.250805139541626, "step": 12010 }, { "epoch": 1.3976029788224342, "grad_norm": 1.1484339237213135, "learning_rate": 0.0005, "loss": 1.8208301067352295, "step": 12011 }, { "epoch": 1.3977193390737723, "grad_norm": 1.2771399021148682, "learning_rate": 0.0005, "loss": 2.2881805896759033, "step": 12012 }, { "epoch": 1.3978356993251104, "grad_norm": 1.670776605606079, "learning_rate": 0.0005, "loss": 2.1810832023620605, "step": 12013 }, { "epoch": 1.3979520595764487, "grad_norm": 0.8037741780281067, "learning_rate": 0.0005, "loss": 1.4146090745925903, "step": 12014 }, { "epoch": 1.3980684198277868, "grad_norm": 0.930718183517456, "learning_rate": 0.0005, "loss": 1.4644745588302612, "step": 12015 }, { "epoch": 1.398184780079125, "grad_norm": 1.0791630744934082, "learning_rate": 0.0005, "loss": 1.5443823337554932, "step": 12016 }, { "epoch": 1.398301140330463, "grad_norm": 1.240224003791809, "learning_rate": 0.0005, "loss": 1.8970409631729126, "step": 12017 }, { "epoch": 1.3984175005818011, "grad_norm": 1.569014549255371, "learning_rate": 0.0005, "loss": 1.7325043678283691, "step": 12018 }, { "epoch": 1.3985338608331395, "grad_norm": 1.326664924621582, "learning_rate": 0.0005, "loss": 2.6486082077026367, "step": 12019 }, { "epoch": 1.3986502210844776, "grad_norm": 1.0783722400665283, "learning_rate": 0.0005, "loss": 1.7797768115997314, "step": 12020 }, { "epoch": 1.3987665813358157, "grad_norm": 1.980368733406067, "learning_rate": 0.0005, "loss": 2.512147903442383, "step": 12021 }, { "epoch": 1.3988829415871538, "grad_norm": 1.092635989189148, "learning_rate": 0.0005, "loss": 1.6803090572357178, "step": 12022 }, { "epoch": 1.3989993018384919, "grad_norm": 0.9758747220039368, "learning_rate": 0.0005, "loss": 1.8595404624938965, "step": 12023 }, { "epoch": 1.3991156620898302, "grad_norm": 1.1137796640396118, "learning_rate": 0.0005, "loss": 2.9118080139160156, "step": 12024 }, { "epoch": 1.3992320223411683, "grad_norm": 1.0414702892303467, "learning_rate": 0.0005, "loss": 1.5581331253051758, "step": 12025 }, { "epoch": 1.3993483825925064, "grad_norm": 0.9389981031417847, "learning_rate": 0.0005, "loss": 2.022270679473877, "step": 12026 }, { "epoch": 1.3994647428438445, "grad_norm": 0.9783743023872375, "learning_rate": 0.0005, "loss": 1.8954441547393799, "step": 12027 }, { "epoch": 1.3995811030951826, "grad_norm": 0.8908609747886658, "learning_rate": 0.0005, "loss": 1.6078696250915527, "step": 12028 }, { "epoch": 1.399697463346521, "grad_norm": 1.312913417816162, "learning_rate": 0.0005, "loss": 1.9314789772033691, "step": 12029 }, { "epoch": 1.399813823597859, "grad_norm": 1.469462513923645, "learning_rate": 0.0005, "loss": 1.9481348991394043, "step": 12030 }, { "epoch": 1.399930183849197, "grad_norm": 0.7068536281585693, "learning_rate": 0.0005, "loss": 1.9512343406677246, "step": 12031 }, { "epoch": 1.4000465441005352, "grad_norm": 1.0480456352233887, "learning_rate": 0.0005, "loss": 1.550156593322754, "step": 12032 }, { "epoch": 1.4001629043518733, "grad_norm": 0.6075075268745422, "learning_rate": 0.0005, "loss": 1.714604377746582, "step": 12033 }, { "epoch": 1.4002792646032116, "grad_norm": 1.1116406917572021, "learning_rate": 0.0005, "loss": 2.079896926879883, "step": 12034 }, { "epoch": 1.4003956248545497, "grad_norm": 1.5226664543151855, "learning_rate": 0.0005, "loss": 2.519930362701416, "step": 12035 }, { "epoch": 1.4005119851058878, "grad_norm": 1.229640007019043, "learning_rate": 0.0005, "loss": 1.2638604640960693, "step": 12036 }, { "epoch": 1.400628345357226, "grad_norm": 0.8791353106498718, "learning_rate": 0.0005, "loss": 1.6832149028778076, "step": 12037 }, { "epoch": 1.400744705608564, "grad_norm": 0.9421720504760742, "learning_rate": 0.0005, "loss": 1.5964473485946655, "step": 12038 }, { "epoch": 1.4008610658599023, "grad_norm": 0.8015968203544617, "learning_rate": 0.0005, "loss": 1.7369108200073242, "step": 12039 }, { "epoch": 1.4009774261112404, "grad_norm": 1.1159332990646362, "learning_rate": 0.0005, "loss": 2.1418745517730713, "step": 12040 }, { "epoch": 1.4010937863625785, "grad_norm": 0.9798673987388611, "learning_rate": 0.0005, "loss": 1.8692426681518555, "step": 12041 }, { "epoch": 1.4012101466139166, "grad_norm": 1.6181155443191528, "learning_rate": 0.0005, "loss": 2.542470932006836, "step": 12042 }, { "epoch": 1.4013265068652547, "grad_norm": 1.0773603916168213, "learning_rate": 0.0005, "loss": 1.1238619089126587, "step": 12043 }, { "epoch": 1.401442867116593, "grad_norm": 0.9041081070899963, "learning_rate": 0.0005, "loss": 1.7653677463531494, "step": 12044 }, { "epoch": 1.4015592273679311, "grad_norm": 1.1074291467666626, "learning_rate": 0.0005, "loss": 2.745911121368408, "step": 12045 }, { "epoch": 1.4016755876192692, "grad_norm": 0.9068001508712769, "learning_rate": 0.0005, "loss": 1.920554280281067, "step": 12046 }, { "epoch": 1.4017919478706073, "grad_norm": 1.256693720817566, "learning_rate": 0.0005, "loss": 1.4329084157943726, "step": 12047 }, { "epoch": 1.4019083081219454, "grad_norm": 0.8812597393989563, "learning_rate": 0.0005, "loss": 1.7009079456329346, "step": 12048 }, { "epoch": 1.4020246683732838, "grad_norm": 1.2643101215362549, "learning_rate": 0.0005, "loss": 2.5262413024902344, "step": 12049 }, { "epoch": 1.4021410286246219, "grad_norm": 0.9655269980430603, "learning_rate": 0.0005, "loss": 2.1953861713409424, "step": 12050 }, { "epoch": 1.40225738887596, "grad_norm": 1.3486851453781128, "learning_rate": 0.0005, "loss": 2.364236831665039, "step": 12051 }, { "epoch": 1.402373749127298, "grad_norm": 1.305309772491455, "learning_rate": 0.0005, "loss": 2.7701306343078613, "step": 12052 }, { "epoch": 1.4024901093786362, "grad_norm": 1.6399526596069336, "learning_rate": 0.0005, "loss": 2.3106751441955566, "step": 12053 }, { "epoch": 1.4026064696299745, "grad_norm": 0.9403480887413025, "learning_rate": 0.0005, "loss": 1.4920785427093506, "step": 12054 }, { "epoch": 1.4027228298813126, "grad_norm": 1.518635869026184, "learning_rate": 0.0005, "loss": 2.344604015350342, "step": 12055 }, { "epoch": 1.4028391901326507, "grad_norm": 1.1341352462768555, "learning_rate": 0.0005, "loss": 1.648530125617981, "step": 12056 }, { "epoch": 1.4029555503839888, "grad_norm": 1.1873760223388672, "learning_rate": 0.0005, "loss": 2.181230068206787, "step": 12057 }, { "epoch": 1.4030719106353269, "grad_norm": 1.5043294429779053, "learning_rate": 0.0005, "loss": 2.3407607078552246, "step": 12058 }, { "epoch": 1.4031882708866652, "grad_norm": 1.3988677263259888, "learning_rate": 0.0005, "loss": 2.8669204711914062, "step": 12059 }, { "epoch": 1.4033046311380033, "grad_norm": 1.1511496305465698, "learning_rate": 0.0005, "loss": 2.675457000732422, "step": 12060 }, { "epoch": 1.4034209913893414, "grad_norm": 1.3531033992767334, "learning_rate": 0.0005, "loss": 2.164429187774658, "step": 12061 }, { "epoch": 1.4035373516406795, "grad_norm": 1.1844252347946167, "learning_rate": 0.0005, "loss": 1.772273063659668, "step": 12062 }, { "epoch": 1.4036537118920176, "grad_norm": 0.9040512442588806, "learning_rate": 0.0005, "loss": 1.353577733039856, "step": 12063 }, { "epoch": 1.403770072143356, "grad_norm": 3.701115608215332, "learning_rate": 0.0005, "loss": 1.6489111185073853, "step": 12064 }, { "epoch": 1.403886432394694, "grad_norm": 1.015814185142517, "learning_rate": 0.0005, "loss": 1.7216682434082031, "step": 12065 }, { "epoch": 1.4040027926460321, "grad_norm": 1.3203325271606445, "learning_rate": 0.0005, "loss": 2.053494453430176, "step": 12066 }, { "epoch": 1.4041191528973702, "grad_norm": 1.800125241279602, "learning_rate": 0.0005, "loss": 1.9217886924743652, "step": 12067 }, { "epoch": 1.4042355131487083, "grad_norm": 1.6732643842697144, "learning_rate": 0.0005, "loss": 2.116933584213257, "step": 12068 }, { "epoch": 1.4043518734000466, "grad_norm": 1.0310338735580444, "learning_rate": 0.0005, "loss": 1.4266788959503174, "step": 12069 }, { "epoch": 1.4044682336513847, "grad_norm": 1.1116251945495605, "learning_rate": 0.0005, "loss": 1.7382864952087402, "step": 12070 }, { "epoch": 1.4045845939027228, "grad_norm": 1.3751416206359863, "learning_rate": 0.0005, "loss": 1.8503791093826294, "step": 12071 }, { "epoch": 1.404700954154061, "grad_norm": 0.9378432631492615, "learning_rate": 0.0005, "loss": 2.177137851715088, "step": 12072 }, { "epoch": 1.404817314405399, "grad_norm": 1.1198060512542725, "learning_rate": 0.0005, "loss": 1.971813678741455, "step": 12073 }, { "epoch": 1.4049336746567374, "grad_norm": 1.0926165580749512, "learning_rate": 0.0005, "loss": 1.6705360412597656, "step": 12074 }, { "epoch": 1.4050500349080755, "grad_norm": 1.4355130195617676, "learning_rate": 0.0005, "loss": 2.6340489387512207, "step": 12075 }, { "epoch": 1.4051663951594136, "grad_norm": 1.3184378147125244, "learning_rate": 0.0005, "loss": 1.603945016860962, "step": 12076 }, { "epoch": 1.4052827554107516, "grad_norm": 1.2864052057266235, "learning_rate": 0.0005, "loss": 1.6501460075378418, "step": 12077 }, { "epoch": 1.4053991156620897, "grad_norm": 1.6797610521316528, "learning_rate": 0.0005, "loss": 2.3844261169433594, "step": 12078 }, { "epoch": 1.405515475913428, "grad_norm": 1.1621578931808472, "learning_rate": 0.0005, "loss": 1.7014728784561157, "step": 12079 }, { "epoch": 1.4056318361647662, "grad_norm": 0.8221945762634277, "learning_rate": 0.0005, "loss": 1.470465898513794, "step": 12080 }, { "epoch": 1.4057481964161043, "grad_norm": 1.258758783340454, "learning_rate": 0.0005, "loss": 1.427485704421997, "step": 12081 }, { "epoch": 1.4058645566674424, "grad_norm": 1.3001352548599243, "learning_rate": 0.0005, "loss": 1.8873217105865479, "step": 12082 }, { "epoch": 1.4059809169187805, "grad_norm": 0.9362208843231201, "learning_rate": 0.0005, "loss": 1.9239486455917358, "step": 12083 }, { "epoch": 1.4060972771701188, "grad_norm": 1.917797565460205, "learning_rate": 0.0005, "loss": 2.2598114013671875, "step": 12084 }, { "epoch": 1.4062136374214569, "grad_norm": 1.3327878713607788, "learning_rate": 0.0005, "loss": 2.30684232711792, "step": 12085 }, { "epoch": 1.406329997672795, "grad_norm": 1.0873613357543945, "learning_rate": 0.0005, "loss": 2.1584038734436035, "step": 12086 }, { "epoch": 1.406446357924133, "grad_norm": 1.238641381263733, "learning_rate": 0.0005, "loss": 2.7347218990325928, "step": 12087 }, { "epoch": 1.4065627181754712, "grad_norm": 0.7917787432670593, "learning_rate": 0.0005, "loss": 2.3635170459747314, "step": 12088 }, { "epoch": 1.4066790784268095, "grad_norm": 0.8864940404891968, "learning_rate": 0.0005, "loss": 1.5845905542373657, "step": 12089 }, { "epoch": 1.4067954386781476, "grad_norm": 0.9848189949989319, "learning_rate": 0.0005, "loss": 1.9412388801574707, "step": 12090 }, { "epoch": 1.4069117989294857, "grad_norm": 3.727609872817993, "learning_rate": 0.0005, "loss": 2.570268392562866, "step": 12091 }, { "epoch": 1.4070281591808238, "grad_norm": 1.2233392000198364, "learning_rate": 0.0005, "loss": 1.9344059228897095, "step": 12092 }, { "epoch": 1.407144519432162, "grad_norm": 0.9612720608711243, "learning_rate": 0.0005, "loss": 2.47961163520813, "step": 12093 }, { "epoch": 1.4072608796835002, "grad_norm": 0.9026976227760315, "learning_rate": 0.0005, "loss": 1.2848680019378662, "step": 12094 }, { "epoch": 1.4073772399348383, "grad_norm": 1.2665915489196777, "learning_rate": 0.0005, "loss": 1.5537197589874268, "step": 12095 }, { "epoch": 1.4074936001861764, "grad_norm": 1.3298161029815674, "learning_rate": 0.0005, "loss": 1.9834589958190918, "step": 12096 }, { "epoch": 1.4076099604375145, "grad_norm": 1.6238125562667847, "learning_rate": 0.0005, "loss": 1.6133544445037842, "step": 12097 }, { "epoch": 1.4077263206888526, "grad_norm": 1.7361334562301636, "learning_rate": 0.0005, "loss": 2.0762779712677, "step": 12098 }, { "epoch": 1.407842680940191, "grad_norm": 1.1732900142669678, "learning_rate": 0.0005, "loss": 1.8732775449752808, "step": 12099 }, { "epoch": 1.407959041191529, "grad_norm": 0.9937006235122681, "learning_rate": 0.0005, "loss": 1.9021793603897095, "step": 12100 }, { "epoch": 1.4080754014428671, "grad_norm": 1.4388935565948486, "learning_rate": 0.0005, "loss": 2.104135751724243, "step": 12101 }, { "epoch": 1.4081917616942052, "grad_norm": 1.007148027420044, "learning_rate": 0.0005, "loss": 1.6726380586624146, "step": 12102 }, { "epoch": 1.4083081219455433, "grad_norm": 1.3500638008117676, "learning_rate": 0.0005, "loss": 2.2477712631225586, "step": 12103 }, { "epoch": 1.4084244821968817, "grad_norm": 1.7120227813720703, "learning_rate": 0.0005, "loss": 1.9200412034988403, "step": 12104 }, { "epoch": 1.4085408424482198, "grad_norm": 0.7935181260108948, "learning_rate": 0.0005, "loss": 2.009615659713745, "step": 12105 }, { "epoch": 1.4086572026995579, "grad_norm": 1.1576272249221802, "learning_rate": 0.0005, "loss": 2.1440377235412598, "step": 12106 }, { "epoch": 1.408773562950896, "grad_norm": 0.9774397015571594, "learning_rate": 0.0005, "loss": 1.9490560293197632, "step": 12107 }, { "epoch": 1.408889923202234, "grad_norm": 2.24808406829834, "learning_rate": 0.0005, "loss": 2.2971694469451904, "step": 12108 }, { "epoch": 1.4090062834535724, "grad_norm": 1.3740575313568115, "learning_rate": 0.0005, "loss": 1.6101181507110596, "step": 12109 }, { "epoch": 1.4091226437049105, "grad_norm": 1.1515628099441528, "learning_rate": 0.0005, "loss": 1.749976634979248, "step": 12110 }, { "epoch": 1.4092390039562486, "grad_norm": 1.6773033142089844, "learning_rate": 0.0005, "loss": 1.6361823081970215, "step": 12111 }, { "epoch": 1.4093553642075867, "grad_norm": 1.4641927480697632, "learning_rate": 0.0005, "loss": 2.1467039585113525, "step": 12112 }, { "epoch": 1.4094717244589248, "grad_norm": 1.494902491569519, "learning_rate": 0.0005, "loss": 2.0349440574645996, "step": 12113 }, { "epoch": 1.409588084710263, "grad_norm": 1.5905495882034302, "learning_rate": 0.0005, "loss": 1.5343914031982422, "step": 12114 }, { "epoch": 1.4097044449616012, "grad_norm": 0.969770610332489, "learning_rate": 0.0005, "loss": 1.7817306518554688, "step": 12115 }, { "epoch": 1.4098208052129393, "grad_norm": 2.262301206588745, "learning_rate": 0.0005, "loss": 2.4312238693237305, "step": 12116 }, { "epoch": 1.4099371654642774, "grad_norm": 1.9819836616516113, "learning_rate": 0.0005, "loss": 2.5981690883636475, "step": 12117 }, { "epoch": 1.4100535257156155, "grad_norm": 0.7891995310783386, "learning_rate": 0.0005, "loss": 2.288317918777466, "step": 12118 }, { "epoch": 1.4101698859669538, "grad_norm": 1.3583028316497803, "learning_rate": 0.0005, "loss": 1.9489694833755493, "step": 12119 }, { "epoch": 1.410286246218292, "grad_norm": 1.2246248722076416, "learning_rate": 0.0005, "loss": 2.114014148712158, "step": 12120 }, { "epoch": 1.41040260646963, "grad_norm": 1.2533289194107056, "learning_rate": 0.0005, "loss": 1.9902011156082153, "step": 12121 }, { "epoch": 1.410518966720968, "grad_norm": 0.719921350479126, "learning_rate": 0.0005, "loss": 1.640593409538269, "step": 12122 }, { "epoch": 1.4106353269723062, "grad_norm": 1.2405998706817627, "learning_rate": 0.0005, "loss": 2.358706474304199, "step": 12123 }, { "epoch": 1.4107516872236445, "grad_norm": 0.9207081198692322, "learning_rate": 0.0005, "loss": 1.4591587781906128, "step": 12124 }, { "epoch": 1.4108680474749826, "grad_norm": 0.6369417309761047, "learning_rate": 0.0005, "loss": 1.1221222877502441, "step": 12125 }, { "epoch": 1.4109844077263207, "grad_norm": 1.2074464559555054, "learning_rate": 0.0005, "loss": 1.9095852375030518, "step": 12126 }, { "epoch": 1.4111007679776588, "grad_norm": 1.2509315013885498, "learning_rate": 0.0005, "loss": 1.6370973587036133, "step": 12127 }, { "epoch": 1.411217128228997, "grad_norm": 0.9554930329322815, "learning_rate": 0.0005, "loss": 1.8937827348709106, "step": 12128 }, { "epoch": 1.4113334884803352, "grad_norm": 0.9745476245880127, "learning_rate": 0.0005, "loss": 1.7779712677001953, "step": 12129 }, { "epoch": 1.4114498487316733, "grad_norm": 0.8856155872344971, "learning_rate": 0.0005, "loss": 1.7794079780578613, "step": 12130 }, { "epoch": 1.4115662089830114, "grad_norm": 1.0824981927871704, "learning_rate": 0.0005, "loss": 1.424018383026123, "step": 12131 }, { "epoch": 1.4116825692343495, "grad_norm": 1.0365955829620361, "learning_rate": 0.0005, "loss": 1.809126853942871, "step": 12132 }, { "epoch": 1.4117989294856876, "grad_norm": 0.8642493486404419, "learning_rate": 0.0005, "loss": 1.5740931034088135, "step": 12133 }, { "epoch": 1.411915289737026, "grad_norm": 1.3254185914993286, "learning_rate": 0.0005, "loss": 2.442486047744751, "step": 12134 }, { "epoch": 1.412031649988364, "grad_norm": 1.7727659940719604, "learning_rate": 0.0005, "loss": 1.918960452079773, "step": 12135 }, { "epoch": 1.4121480102397022, "grad_norm": 1.01762855052948, "learning_rate": 0.0005, "loss": 2.1236815452575684, "step": 12136 }, { "epoch": 1.4122643704910403, "grad_norm": 0.8536914587020874, "learning_rate": 0.0005, "loss": 1.351784110069275, "step": 12137 }, { "epoch": 1.4123807307423784, "grad_norm": 1.0565077066421509, "learning_rate": 0.0005, "loss": 1.962350606918335, "step": 12138 }, { "epoch": 1.4124970909937167, "grad_norm": 1.137204885482788, "learning_rate": 0.0005, "loss": 1.8645825386047363, "step": 12139 }, { "epoch": 1.4126134512450546, "grad_norm": 1.6209120750427246, "learning_rate": 0.0005, "loss": 2.0757808685302734, "step": 12140 }, { "epoch": 1.4127298114963929, "grad_norm": 0.7414000630378723, "learning_rate": 0.0005, "loss": 1.4123475551605225, "step": 12141 }, { "epoch": 1.412846171747731, "grad_norm": 0.7091333866119385, "learning_rate": 0.0005, "loss": 1.9685742855072021, "step": 12142 }, { "epoch": 1.412962531999069, "grad_norm": 1.0253477096557617, "learning_rate": 0.0005, "loss": 1.6093077659606934, "step": 12143 }, { "epoch": 1.4130788922504074, "grad_norm": 2.36154842376709, "learning_rate": 0.0005, "loss": 1.5177114009857178, "step": 12144 }, { "epoch": 1.4131952525017453, "grad_norm": 1.0812782049179077, "learning_rate": 0.0005, "loss": 1.2469780445098877, "step": 12145 }, { "epoch": 1.4133116127530836, "grad_norm": 1.272281527519226, "learning_rate": 0.0005, "loss": 2.169800281524658, "step": 12146 }, { "epoch": 1.4134279730044217, "grad_norm": 1.0615003108978271, "learning_rate": 0.0005, "loss": 1.6994320154190063, "step": 12147 }, { "epoch": 1.4135443332557598, "grad_norm": 1.2321772575378418, "learning_rate": 0.0005, "loss": 2.521224021911621, "step": 12148 }, { "epoch": 1.413660693507098, "grad_norm": 1.3072172403335571, "learning_rate": 0.0005, "loss": 1.5879427194595337, "step": 12149 }, { "epoch": 1.413777053758436, "grad_norm": 1.5915127992630005, "learning_rate": 0.0005, "loss": 1.7222990989685059, "step": 12150 }, { "epoch": 1.4138934140097743, "grad_norm": 0.4656679928302765, "learning_rate": 0.0005, "loss": 1.1634725332260132, "step": 12151 }, { "epoch": 1.4140097742611124, "grad_norm": 1.3946479558944702, "learning_rate": 0.0005, "loss": 2.238349437713623, "step": 12152 }, { "epoch": 1.4141261345124505, "grad_norm": 0.7779880166053772, "learning_rate": 0.0005, "loss": 2.369011878967285, "step": 12153 }, { "epoch": 1.4142424947637888, "grad_norm": 1.101557731628418, "learning_rate": 0.0005, "loss": 1.8332526683807373, "step": 12154 }, { "epoch": 1.4143588550151267, "grad_norm": 0.7353130578994751, "learning_rate": 0.0005, "loss": 2.1427817344665527, "step": 12155 }, { "epoch": 1.414475215266465, "grad_norm": 0.7116214036941528, "learning_rate": 0.0005, "loss": 1.4250307083129883, "step": 12156 }, { "epoch": 1.4145915755178031, "grad_norm": 1.3031799793243408, "learning_rate": 0.0005, "loss": 2.2870774269104004, "step": 12157 }, { "epoch": 1.4147079357691412, "grad_norm": 1.5815882682800293, "learning_rate": 0.0005, "loss": 2.5684969425201416, "step": 12158 }, { "epoch": 1.4148242960204793, "grad_norm": 1.0847030878067017, "learning_rate": 0.0005, "loss": 1.9273104667663574, "step": 12159 }, { "epoch": 1.4149406562718174, "grad_norm": 1.2486531734466553, "learning_rate": 0.0005, "loss": 2.337503433227539, "step": 12160 }, { "epoch": 1.4150570165231557, "grad_norm": 0.9711251854896545, "learning_rate": 0.0005, "loss": 2.0065197944641113, "step": 12161 }, { "epoch": 1.4151733767744938, "grad_norm": 1.1353230476379395, "learning_rate": 0.0005, "loss": 1.9230995178222656, "step": 12162 }, { "epoch": 1.415289737025832, "grad_norm": 1.2685943841934204, "learning_rate": 0.0005, "loss": 2.364230155944824, "step": 12163 }, { "epoch": 1.41540609727717, "grad_norm": 2.632675886154175, "learning_rate": 0.0005, "loss": 2.145188570022583, "step": 12164 }, { "epoch": 1.4155224575285081, "grad_norm": 0.915299117565155, "learning_rate": 0.0005, "loss": 1.3613407611846924, "step": 12165 }, { "epoch": 1.4156388177798465, "grad_norm": 1.0946214199066162, "learning_rate": 0.0005, "loss": 2.1385858058929443, "step": 12166 }, { "epoch": 1.4157551780311846, "grad_norm": 1.640870451927185, "learning_rate": 0.0005, "loss": 2.129939079284668, "step": 12167 }, { "epoch": 1.4158715382825227, "grad_norm": 1.293994426727295, "learning_rate": 0.0005, "loss": 1.9051175117492676, "step": 12168 }, { "epoch": 1.4159878985338608, "grad_norm": 0.7556346654891968, "learning_rate": 0.0005, "loss": 1.6484732627868652, "step": 12169 }, { "epoch": 1.4161042587851989, "grad_norm": 1.4024412631988525, "learning_rate": 0.0005, "loss": 2.1855359077453613, "step": 12170 }, { "epoch": 1.4162206190365372, "grad_norm": 1.762896180152893, "learning_rate": 0.0005, "loss": 2.1079916954040527, "step": 12171 }, { "epoch": 1.4163369792878753, "grad_norm": 0.951777458190918, "learning_rate": 0.0005, "loss": 1.6471896171569824, "step": 12172 }, { "epoch": 1.4164533395392134, "grad_norm": 0.9397629499435425, "learning_rate": 0.0005, "loss": 1.7878581285476685, "step": 12173 }, { "epoch": 1.4165696997905515, "grad_norm": 0.724154531955719, "learning_rate": 0.0005, "loss": 1.1138660907745361, "step": 12174 }, { "epoch": 1.4166860600418896, "grad_norm": 1.4506664276123047, "learning_rate": 0.0005, "loss": 2.0735220909118652, "step": 12175 }, { "epoch": 1.416802420293228, "grad_norm": 1.0546890497207642, "learning_rate": 0.0005, "loss": 1.6223797798156738, "step": 12176 }, { "epoch": 1.416918780544566, "grad_norm": 0.7483145594596863, "learning_rate": 0.0005, "loss": 1.9475011825561523, "step": 12177 }, { "epoch": 1.417035140795904, "grad_norm": 0.7731388807296753, "learning_rate": 0.0005, "loss": 2.43577241897583, "step": 12178 }, { "epoch": 1.4171515010472422, "grad_norm": 0.8463970422744751, "learning_rate": 0.0005, "loss": 1.3355319499969482, "step": 12179 }, { "epoch": 1.4172678612985803, "grad_norm": 1.8829007148742676, "learning_rate": 0.0005, "loss": 1.8013007640838623, "step": 12180 }, { "epoch": 1.4173842215499186, "grad_norm": 0.9989438056945801, "learning_rate": 0.0005, "loss": 1.3242979049682617, "step": 12181 }, { "epoch": 1.4175005818012567, "grad_norm": 1.3285434246063232, "learning_rate": 0.0005, "loss": 1.8664854764938354, "step": 12182 }, { "epoch": 1.4176169420525948, "grad_norm": 0.9541193246841431, "learning_rate": 0.0005, "loss": 1.4445436000823975, "step": 12183 }, { "epoch": 1.417733302303933, "grad_norm": 1.214941143989563, "learning_rate": 0.0005, "loss": 2.0729875564575195, "step": 12184 }, { "epoch": 1.417849662555271, "grad_norm": 1.0154743194580078, "learning_rate": 0.0005, "loss": 2.548874855041504, "step": 12185 }, { "epoch": 1.4179660228066093, "grad_norm": 1.2389843463897705, "learning_rate": 0.0005, "loss": 1.747293472290039, "step": 12186 }, { "epoch": 1.4180823830579474, "grad_norm": 1.1500070095062256, "learning_rate": 0.0005, "loss": 2.2148585319519043, "step": 12187 }, { "epoch": 1.4181987433092855, "grad_norm": 1.7248581647872925, "learning_rate": 0.0005, "loss": 3.212766408920288, "step": 12188 }, { "epoch": 1.4183151035606236, "grad_norm": 0.7142119407653809, "learning_rate": 0.0005, "loss": 1.346698522567749, "step": 12189 }, { "epoch": 1.4184314638119617, "grad_norm": 0.8810511827468872, "learning_rate": 0.0005, "loss": 1.6757922172546387, "step": 12190 }, { "epoch": 1.4185478240633, "grad_norm": 1.3615810871124268, "learning_rate": 0.0005, "loss": 2.5125374794006348, "step": 12191 }, { "epoch": 1.4186641843146381, "grad_norm": 1.319856882095337, "learning_rate": 0.0005, "loss": 3.102458953857422, "step": 12192 }, { "epoch": 1.4187805445659762, "grad_norm": 1.419551134109497, "learning_rate": 0.0005, "loss": 1.3627541065216064, "step": 12193 }, { "epoch": 1.4188969048173143, "grad_norm": 2.6540398597717285, "learning_rate": 0.0005, "loss": 1.782155990600586, "step": 12194 }, { "epoch": 1.4190132650686524, "grad_norm": 1.2302236557006836, "learning_rate": 0.0005, "loss": 1.809291124343872, "step": 12195 }, { "epoch": 1.4191296253199908, "grad_norm": 1.275692343711853, "learning_rate": 0.0005, "loss": 1.986853837966919, "step": 12196 }, { "epoch": 1.4192459855713289, "grad_norm": 1.4016298055648804, "learning_rate": 0.0005, "loss": 2.217531204223633, "step": 12197 }, { "epoch": 1.419362345822667, "grad_norm": 1.1760350465774536, "learning_rate": 0.0005, "loss": 2.063649892807007, "step": 12198 }, { "epoch": 1.419478706074005, "grad_norm": 0.9572426676750183, "learning_rate": 0.0005, "loss": 1.5629019737243652, "step": 12199 }, { "epoch": 1.4195950663253432, "grad_norm": 0.9579935669898987, "learning_rate": 0.0005, "loss": 2.56689190864563, "step": 12200 }, { "epoch": 1.4197114265766815, "grad_norm": 1.0813039541244507, "learning_rate": 0.0005, "loss": 2.125699758529663, "step": 12201 }, { "epoch": 1.4198277868280196, "grad_norm": 0.8589019775390625, "learning_rate": 0.0005, "loss": 1.4213238954544067, "step": 12202 }, { "epoch": 1.4199441470793577, "grad_norm": 1.8761383295059204, "learning_rate": 0.0005, "loss": 1.8524976968765259, "step": 12203 }, { "epoch": 1.4200605073306958, "grad_norm": 1.2899504899978638, "learning_rate": 0.0005, "loss": 2.1075737476348877, "step": 12204 }, { "epoch": 1.4201768675820339, "grad_norm": 1.2354553937911987, "learning_rate": 0.0005, "loss": 1.4881017208099365, "step": 12205 }, { "epoch": 1.4202932278333722, "grad_norm": 0.9148956537246704, "learning_rate": 0.0005, "loss": 1.683005690574646, "step": 12206 }, { "epoch": 1.4204095880847103, "grad_norm": 1.0777666568756104, "learning_rate": 0.0005, "loss": 1.4242260456085205, "step": 12207 }, { "epoch": 1.4205259483360484, "grad_norm": 1.6591465473175049, "learning_rate": 0.0005, "loss": 1.8798763751983643, "step": 12208 }, { "epoch": 1.4206423085873865, "grad_norm": 0.7976081371307373, "learning_rate": 0.0005, "loss": 1.3123571872711182, "step": 12209 }, { "epoch": 1.4207586688387246, "grad_norm": 0.630119264125824, "learning_rate": 0.0005, "loss": 1.429111123085022, "step": 12210 }, { "epoch": 1.420875029090063, "grad_norm": 0.6084225177764893, "learning_rate": 0.0005, "loss": 1.637976884841919, "step": 12211 }, { "epoch": 1.420991389341401, "grad_norm": 1.0078858137130737, "learning_rate": 0.0005, "loss": 2.2300844192504883, "step": 12212 }, { "epoch": 1.4211077495927391, "grad_norm": 1.085740327835083, "learning_rate": 0.0005, "loss": 1.779141902923584, "step": 12213 }, { "epoch": 1.4212241098440772, "grad_norm": 1.4350522756576538, "learning_rate": 0.0005, "loss": 1.7553837299346924, "step": 12214 }, { "epoch": 1.4213404700954153, "grad_norm": 0.9937778115272522, "learning_rate": 0.0005, "loss": 1.588820457458496, "step": 12215 }, { "epoch": 1.4214568303467536, "grad_norm": 1.020695447921753, "learning_rate": 0.0005, "loss": 2.2653098106384277, "step": 12216 }, { "epoch": 1.4215731905980917, "grad_norm": 1.064577579498291, "learning_rate": 0.0005, "loss": 2.0495786666870117, "step": 12217 }, { "epoch": 1.4216895508494298, "grad_norm": 0.7125756144523621, "learning_rate": 0.0005, "loss": 1.3379065990447998, "step": 12218 }, { "epoch": 1.421805911100768, "grad_norm": 0.9999560713768005, "learning_rate": 0.0005, "loss": 1.743283748626709, "step": 12219 }, { "epoch": 1.421922271352106, "grad_norm": 1.2147938013076782, "learning_rate": 0.0005, "loss": 1.8861777782440186, "step": 12220 }, { "epoch": 1.4220386316034443, "grad_norm": 0.854105532169342, "learning_rate": 0.0005, "loss": 1.5136733055114746, "step": 12221 }, { "epoch": 1.4221549918547824, "grad_norm": 1.9569460153579712, "learning_rate": 0.0005, "loss": 1.611897349357605, "step": 12222 }, { "epoch": 1.4222713521061205, "grad_norm": 1.6704448461532593, "learning_rate": 0.0005, "loss": 2.1347899436950684, "step": 12223 }, { "epoch": 1.4223877123574586, "grad_norm": 1.340993046760559, "learning_rate": 0.0005, "loss": 1.9573876857757568, "step": 12224 }, { "epoch": 1.4225040726087967, "grad_norm": 0.7215344309806824, "learning_rate": 0.0005, "loss": 1.166568636894226, "step": 12225 }, { "epoch": 1.422620432860135, "grad_norm": 1.2791554927825928, "learning_rate": 0.0005, "loss": 2.7768197059631348, "step": 12226 }, { "epoch": 1.4227367931114732, "grad_norm": 1.2142252922058105, "learning_rate": 0.0005, "loss": 1.4930853843688965, "step": 12227 }, { "epoch": 1.4228531533628113, "grad_norm": 1.1295725107192993, "learning_rate": 0.0005, "loss": 1.8238239288330078, "step": 12228 }, { "epoch": 1.4229695136141494, "grad_norm": 0.9243558049201965, "learning_rate": 0.0005, "loss": 1.8662546873092651, "step": 12229 }, { "epoch": 1.4230858738654875, "grad_norm": 0.9282456040382385, "learning_rate": 0.0005, "loss": 1.528051733970642, "step": 12230 }, { "epoch": 1.4232022341168258, "grad_norm": 1.3392475843429565, "learning_rate": 0.0005, "loss": 2.8500566482543945, "step": 12231 }, { "epoch": 1.4233185943681639, "grad_norm": 1.144636631011963, "learning_rate": 0.0005, "loss": 2.278494358062744, "step": 12232 }, { "epoch": 1.423434954619502, "grad_norm": 1.1595540046691895, "learning_rate": 0.0005, "loss": 2.2733702659606934, "step": 12233 }, { "epoch": 1.42355131487084, "grad_norm": 1.4962284564971924, "learning_rate": 0.0005, "loss": 2.286803960800171, "step": 12234 }, { "epoch": 1.4236676751221782, "grad_norm": 1.481938123703003, "learning_rate": 0.0005, "loss": 1.795685052871704, "step": 12235 }, { "epoch": 1.4237840353735165, "grad_norm": 1.2224570512771606, "learning_rate": 0.0005, "loss": 2.591442346572876, "step": 12236 }, { "epoch": 1.4239003956248546, "grad_norm": 0.7690354585647583, "learning_rate": 0.0005, "loss": 1.7207050323486328, "step": 12237 }, { "epoch": 1.4240167558761927, "grad_norm": 0.8933542370796204, "learning_rate": 0.0005, "loss": 1.8542675971984863, "step": 12238 }, { "epoch": 1.4241331161275308, "grad_norm": 1.2660101652145386, "learning_rate": 0.0005, "loss": 2.4382829666137695, "step": 12239 }, { "epoch": 1.424249476378869, "grad_norm": 1.0124293565750122, "learning_rate": 0.0005, "loss": 1.741805076599121, "step": 12240 }, { "epoch": 1.4243658366302072, "grad_norm": 1.3846111297607422, "learning_rate": 0.0005, "loss": 1.46071195602417, "step": 12241 }, { "epoch": 1.4244821968815453, "grad_norm": 0.7842039465904236, "learning_rate": 0.0005, "loss": 1.5414831638336182, "step": 12242 }, { "epoch": 1.4245985571328834, "grad_norm": 0.8069117069244385, "learning_rate": 0.0005, "loss": 1.6152430772781372, "step": 12243 }, { "epoch": 1.4247149173842215, "grad_norm": 1.1771947145462036, "learning_rate": 0.0005, "loss": 1.9557729959487915, "step": 12244 }, { "epoch": 1.4248312776355596, "grad_norm": 1.5866913795471191, "learning_rate": 0.0005, "loss": 2.180410146713257, "step": 12245 }, { "epoch": 1.424947637886898, "grad_norm": 0.7480349540710449, "learning_rate": 0.0005, "loss": 1.5421512126922607, "step": 12246 }, { "epoch": 1.425063998138236, "grad_norm": 1.3184572458267212, "learning_rate": 0.0005, "loss": 2.6540722846984863, "step": 12247 }, { "epoch": 1.4251803583895741, "grad_norm": 0.8704939484596252, "learning_rate": 0.0005, "loss": 1.5582749843597412, "step": 12248 }, { "epoch": 1.4252967186409122, "grad_norm": 1.230206847190857, "learning_rate": 0.0005, "loss": 2.3994405269622803, "step": 12249 }, { "epoch": 1.4254130788922503, "grad_norm": 0.7638249397277832, "learning_rate": 0.0005, "loss": 1.1352863311767578, "step": 12250 }, { "epoch": 1.4255294391435887, "grad_norm": 1.0078887939453125, "learning_rate": 0.0005, "loss": 1.2171146869659424, "step": 12251 }, { "epoch": 1.4256457993949267, "grad_norm": 1.1761049032211304, "learning_rate": 0.0005, "loss": 1.7304195165634155, "step": 12252 }, { "epoch": 1.4257621596462648, "grad_norm": 1.814791202545166, "learning_rate": 0.0005, "loss": 2.148360013961792, "step": 12253 }, { "epoch": 1.425878519897603, "grad_norm": 1.3996890783309937, "learning_rate": 0.0005, "loss": 1.598686695098877, "step": 12254 }, { "epoch": 1.425994880148941, "grad_norm": 1.334843635559082, "learning_rate": 0.0005, "loss": 2.0124094486236572, "step": 12255 }, { "epoch": 1.4261112404002794, "grad_norm": 1.5127856731414795, "learning_rate": 0.0005, "loss": 1.9297852516174316, "step": 12256 }, { "epoch": 1.4262276006516175, "grad_norm": 0.9704545736312866, "learning_rate": 0.0005, "loss": 1.5929832458496094, "step": 12257 }, { "epoch": 1.4263439609029556, "grad_norm": 1.4755690097808838, "learning_rate": 0.0005, "loss": 1.7975951433181763, "step": 12258 }, { "epoch": 1.4264603211542937, "grad_norm": 1.0947036743164062, "learning_rate": 0.0005, "loss": 1.5641005039215088, "step": 12259 }, { "epoch": 1.4265766814056318, "grad_norm": 1.1377896070480347, "learning_rate": 0.0005, "loss": 1.9069072008132935, "step": 12260 }, { "epoch": 1.42669304165697, "grad_norm": 2.554590940475464, "learning_rate": 0.0005, "loss": 2.4049441814422607, "step": 12261 }, { "epoch": 1.4268094019083082, "grad_norm": 0.8600170016288757, "learning_rate": 0.0005, "loss": 1.6818434000015259, "step": 12262 }, { "epoch": 1.4269257621596463, "grad_norm": 1.2425745725631714, "learning_rate": 0.0005, "loss": 2.3304238319396973, "step": 12263 }, { "epoch": 1.4270421224109844, "grad_norm": 1.1002404689788818, "learning_rate": 0.0005, "loss": 1.500516653060913, "step": 12264 }, { "epoch": 1.4271584826623225, "grad_norm": 1.4503475427627563, "learning_rate": 0.0005, "loss": 1.592154622077942, "step": 12265 }, { "epoch": 1.4272748429136608, "grad_norm": 1.221510887145996, "learning_rate": 0.0005, "loss": 2.045905590057373, "step": 12266 }, { "epoch": 1.427391203164999, "grad_norm": 1.0605100393295288, "learning_rate": 0.0005, "loss": 2.0664234161376953, "step": 12267 }, { "epoch": 1.427507563416337, "grad_norm": 1.4224607944488525, "learning_rate": 0.0005, "loss": 2.7947049140930176, "step": 12268 }, { "epoch": 1.427623923667675, "grad_norm": 0.9678479433059692, "learning_rate": 0.0005, "loss": 1.640215516090393, "step": 12269 }, { "epoch": 1.4277402839190132, "grad_norm": 0.44928550720214844, "learning_rate": 0.0005, "loss": 1.3596832752227783, "step": 12270 }, { "epoch": 1.4278566441703515, "grad_norm": 0.9510323405265808, "learning_rate": 0.0005, "loss": 1.9578421115875244, "step": 12271 }, { "epoch": 1.4279730044216896, "grad_norm": 1.3291845321655273, "learning_rate": 0.0005, "loss": 1.9425461292266846, "step": 12272 }, { "epoch": 1.4280893646730277, "grad_norm": 1.1873993873596191, "learning_rate": 0.0005, "loss": 1.9422962665557861, "step": 12273 }, { "epoch": 1.4282057249243658, "grad_norm": 1.646215558052063, "learning_rate": 0.0005, "loss": 2.0367045402526855, "step": 12274 }, { "epoch": 1.428322085175704, "grad_norm": 1.353356122970581, "learning_rate": 0.0005, "loss": 1.7601723670959473, "step": 12275 }, { "epoch": 1.4284384454270422, "grad_norm": 0.8840089440345764, "learning_rate": 0.0005, "loss": 1.1652007102966309, "step": 12276 }, { "epoch": 1.4285548056783803, "grad_norm": 0.9660678505897522, "learning_rate": 0.0005, "loss": 2.1152124404907227, "step": 12277 }, { "epoch": 1.4286711659297184, "grad_norm": 0.9658466577529907, "learning_rate": 0.0005, "loss": 1.1588115692138672, "step": 12278 }, { "epoch": 1.4287875261810565, "grad_norm": 1.5250273942947388, "learning_rate": 0.0005, "loss": 2.1556499004364014, "step": 12279 }, { "epoch": 1.4289038864323946, "grad_norm": 1.1884442567825317, "learning_rate": 0.0005, "loss": 1.8950929641723633, "step": 12280 }, { "epoch": 1.429020246683733, "grad_norm": 0.9785860776901245, "learning_rate": 0.0005, "loss": 1.5685538053512573, "step": 12281 }, { "epoch": 1.429136606935071, "grad_norm": 1.0484013557434082, "learning_rate": 0.0005, "loss": 2.857140064239502, "step": 12282 }, { "epoch": 1.4292529671864092, "grad_norm": 1.313450813293457, "learning_rate": 0.0005, "loss": 1.7016091346740723, "step": 12283 }, { "epoch": 1.4293693274377473, "grad_norm": 1.3102390766143799, "learning_rate": 0.0005, "loss": 1.364179015159607, "step": 12284 }, { "epoch": 1.4294856876890853, "grad_norm": 1.7065054178237915, "learning_rate": 0.0005, "loss": 1.7414076328277588, "step": 12285 }, { "epoch": 1.4296020479404237, "grad_norm": 0.9512113332748413, "learning_rate": 0.0005, "loss": 1.9043591022491455, "step": 12286 }, { "epoch": 1.4297184081917618, "grad_norm": 1.2124189138412476, "learning_rate": 0.0005, "loss": 2.4265966415405273, "step": 12287 }, { "epoch": 1.4298347684430999, "grad_norm": 1.0842114686965942, "learning_rate": 0.0005, "loss": 1.5842434167861938, "step": 12288 }, { "epoch": 1.429951128694438, "grad_norm": 0.6670258045196533, "learning_rate": 0.0005, "loss": 1.4204250574111938, "step": 12289 }, { "epoch": 1.430067488945776, "grad_norm": 0.7571166753768921, "learning_rate": 0.0005, "loss": 1.5342103242874146, "step": 12290 }, { "epoch": 1.4301838491971144, "grad_norm": 1.4539164304733276, "learning_rate": 0.0005, "loss": 2.185481071472168, "step": 12291 }, { "epoch": 1.4303002094484525, "grad_norm": 1.185924768447876, "learning_rate": 0.0005, "loss": 1.88893723487854, "step": 12292 }, { "epoch": 1.4304165696997906, "grad_norm": 1.138829231262207, "learning_rate": 0.0005, "loss": 1.784146785736084, "step": 12293 }, { "epoch": 1.4305329299511287, "grad_norm": 0.9251687526702881, "learning_rate": 0.0005, "loss": 1.0305511951446533, "step": 12294 }, { "epoch": 1.4306492902024668, "grad_norm": 1.6801016330718994, "learning_rate": 0.0005, "loss": 1.5612088441848755, "step": 12295 }, { "epoch": 1.430765650453805, "grad_norm": 1.4525024890899658, "learning_rate": 0.0005, "loss": 1.9775488376617432, "step": 12296 }, { "epoch": 1.4308820107051432, "grad_norm": 1.2818832397460938, "learning_rate": 0.0005, "loss": 1.8495378494262695, "step": 12297 }, { "epoch": 1.4309983709564813, "grad_norm": 1.0268402099609375, "learning_rate": 0.0005, "loss": 1.8319287300109863, "step": 12298 }, { "epoch": 1.4311147312078194, "grad_norm": 1.410725474357605, "learning_rate": 0.0005, "loss": 1.6879981756210327, "step": 12299 }, { "epoch": 1.4312310914591575, "grad_norm": 1.2619441747665405, "learning_rate": 0.0005, "loss": 2.250384569168091, "step": 12300 }, { "epoch": 1.4313474517104958, "grad_norm": 1.616732120513916, "learning_rate": 0.0005, "loss": 1.6916438341140747, "step": 12301 }, { "epoch": 1.431463811961834, "grad_norm": 1.4861270189285278, "learning_rate": 0.0005, "loss": 1.777665376663208, "step": 12302 }, { "epoch": 1.431580172213172, "grad_norm": 1.2046113014221191, "learning_rate": 0.0005, "loss": 1.8887271881103516, "step": 12303 }, { "epoch": 1.4316965324645101, "grad_norm": 1.0437581539154053, "learning_rate": 0.0005, "loss": 1.9444364309310913, "step": 12304 }, { "epoch": 1.4318128927158482, "grad_norm": 1.7062602043151855, "learning_rate": 0.0005, "loss": 2.712435007095337, "step": 12305 }, { "epoch": 1.4319292529671865, "grad_norm": 1.2799763679504395, "learning_rate": 0.0005, "loss": 1.5272622108459473, "step": 12306 }, { "epoch": 1.4320456132185244, "grad_norm": 1.4785391092300415, "learning_rate": 0.0005, "loss": 1.7140089273452759, "step": 12307 }, { "epoch": 1.4321619734698627, "grad_norm": 1.3340178728103638, "learning_rate": 0.0005, "loss": 2.1698949337005615, "step": 12308 }, { "epoch": 1.4322783337212008, "grad_norm": 1.072081446647644, "learning_rate": 0.0005, "loss": 2.230733871459961, "step": 12309 }, { "epoch": 1.432394693972539, "grad_norm": 0.6289306879043579, "learning_rate": 0.0005, "loss": 1.3375763893127441, "step": 12310 }, { "epoch": 1.4325110542238773, "grad_norm": 0.8786226511001587, "learning_rate": 0.0005, "loss": 2.170062303543091, "step": 12311 }, { "epoch": 1.4326274144752151, "grad_norm": 1.2747135162353516, "learning_rate": 0.0005, "loss": 2.2372257709503174, "step": 12312 }, { "epoch": 1.4327437747265535, "grad_norm": 1.0212087631225586, "learning_rate": 0.0005, "loss": 2.0346686840057373, "step": 12313 }, { "epoch": 1.4328601349778916, "grad_norm": 1.41177237033844, "learning_rate": 0.0005, "loss": 2.1880197525024414, "step": 12314 }, { "epoch": 1.4329764952292297, "grad_norm": 1.121295690536499, "learning_rate": 0.0005, "loss": 2.2466115951538086, "step": 12315 }, { "epoch": 1.433092855480568, "grad_norm": 1.204454779624939, "learning_rate": 0.0005, "loss": 2.2833216190338135, "step": 12316 }, { "epoch": 1.4332092157319059, "grad_norm": 1.1684521436691284, "learning_rate": 0.0005, "loss": 2.1862196922302246, "step": 12317 }, { "epoch": 1.4333255759832442, "grad_norm": 0.7585222125053406, "learning_rate": 0.0005, "loss": 1.4707386493682861, "step": 12318 }, { "epoch": 1.4334419362345823, "grad_norm": 1.0928606986999512, "learning_rate": 0.0005, "loss": 1.8380846977233887, "step": 12319 }, { "epoch": 1.4335582964859204, "grad_norm": 1.6188148260116577, "learning_rate": 0.0005, "loss": 2.053424596786499, "step": 12320 }, { "epoch": 1.4336746567372587, "grad_norm": 0.9779770374298096, "learning_rate": 0.0005, "loss": 1.7885421514511108, "step": 12321 }, { "epoch": 1.4337910169885966, "grad_norm": 0.8516079783439636, "learning_rate": 0.0005, "loss": 1.8264508247375488, "step": 12322 }, { "epoch": 1.4339073772399349, "grad_norm": 1.0146267414093018, "learning_rate": 0.0005, "loss": 1.6951676607131958, "step": 12323 }, { "epoch": 1.434023737491273, "grad_norm": 2.5021865367889404, "learning_rate": 0.0005, "loss": 2.3914546966552734, "step": 12324 }, { "epoch": 1.434140097742611, "grad_norm": 1.0362759828567505, "learning_rate": 0.0005, "loss": 1.4353973865509033, "step": 12325 }, { "epoch": 1.4342564579939492, "grad_norm": 0.8067347407341003, "learning_rate": 0.0005, "loss": 1.3624253273010254, "step": 12326 }, { "epoch": 1.4343728182452873, "grad_norm": 0.8310686945915222, "learning_rate": 0.0005, "loss": 1.5503454208374023, "step": 12327 }, { "epoch": 1.4344891784966256, "grad_norm": 1.054508924484253, "learning_rate": 0.0005, "loss": 2.2135093212127686, "step": 12328 }, { "epoch": 1.4346055387479637, "grad_norm": 1.312955617904663, "learning_rate": 0.0005, "loss": 2.3185102939605713, "step": 12329 }, { "epoch": 1.4347218989993018, "grad_norm": 2.535299301147461, "learning_rate": 0.0005, "loss": 2.0690999031066895, "step": 12330 }, { "epoch": 1.43483825925064, "grad_norm": 0.9847354292869568, "learning_rate": 0.0005, "loss": 1.3007776737213135, "step": 12331 }, { "epoch": 1.434954619501978, "grad_norm": 1.1990952491760254, "learning_rate": 0.0005, "loss": 1.6163257360458374, "step": 12332 }, { "epoch": 1.4350709797533163, "grad_norm": 1.9171265363693237, "learning_rate": 0.0005, "loss": 1.5860003232955933, "step": 12333 }, { "epoch": 1.4351873400046544, "grad_norm": 0.744211733341217, "learning_rate": 0.0005, "loss": 1.1203970909118652, "step": 12334 }, { "epoch": 1.4353037002559925, "grad_norm": 0.5149848461151123, "learning_rate": 0.0005, "loss": 1.2975974082946777, "step": 12335 }, { "epoch": 1.4354200605073306, "grad_norm": 0.9176421165466309, "learning_rate": 0.0005, "loss": 2.019864797592163, "step": 12336 }, { "epoch": 1.4355364207586687, "grad_norm": 0.8582545518875122, "learning_rate": 0.0005, "loss": 1.4953171014785767, "step": 12337 }, { "epoch": 1.435652781010007, "grad_norm": 1.4188485145568848, "learning_rate": 0.0005, "loss": 1.7768430709838867, "step": 12338 }, { "epoch": 1.4357691412613451, "grad_norm": 0.8855640292167664, "learning_rate": 0.0005, "loss": 1.6645115613937378, "step": 12339 }, { "epoch": 1.4358855015126832, "grad_norm": 0.39724594354629517, "learning_rate": 0.0005, "loss": 0.9255553483963013, "step": 12340 }, { "epoch": 1.4360018617640213, "grad_norm": 1.1931474208831787, "learning_rate": 0.0005, "loss": 1.5483014583587646, "step": 12341 }, { "epoch": 1.4361182220153594, "grad_norm": 0.9997895956039429, "learning_rate": 0.0005, "loss": 1.9329583644866943, "step": 12342 }, { "epoch": 1.4362345822666978, "grad_norm": 0.6765666604042053, "learning_rate": 0.0005, "loss": 1.109067440032959, "step": 12343 }, { "epoch": 1.4363509425180359, "grad_norm": 1.1409803628921509, "learning_rate": 0.0005, "loss": 1.9498121738433838, "step": 12344 }, { "epoch": 1.436467302769374, "grad_norm": 1.2620874643325806, "learning_rate": 0.0005, "loss": 2.041964530944824, "step": 12345 }, { "epoch": 1.436583663020712, "grad_norm": 1.2409545183181763, "learning_rate": 0.0005, "loss": 1.8363786935806274, "step": 12346 }, { "epoch": 1.4367000232720502, "grad_norm": 1.0261532068252563, "learning_rate": 0.0005, "loss": 2.193643569946289, "step": 12347 }, { "epoch": 1.4368163835233885, "grad_norm": 0.7158399820327759, "learning_rate": 0.0005, "loss": 1.2352358102798462, "step": 12348 }, { "epoch": 1.4369327437747266, "grad_norm": 1.328229546546936, "learning_rate": 0.0005, "loss": 1.721822738647461, "step": 12349 }, { "epoch": 1.4370491040260647, "grad_norm": 1.4201823472976685, "learning_rate": 0.0005, "loss": 2.5921945571899414, "step": 12350 }, { "epoch": 1.4371654642774028, "grad_norm": 1.7878496646881104, "learning_rate": 0.0005, "loss": 2.561997652053833, "step": 12351 }, { "epoch": 1.4372818245287409, "grad_norm": 0.7528570294380188, "learning_rate": 0.0005, "loss": 1.398885726928711, "step": 12352 }, { "epoch": 1.4373981847800792, "grad_norm": 1.5097495317459106, "learning_rate": 0.0005, "loss": 1.6782511472702026, "step": 12353 }, { "epoch": 1.4375145450314173, "grad_norm": 2.04727840423584, "learning_rate": 0.0005, "loss": 1.5539014339447021, "step": 12354 }, { "epoch": 1.4376309052827554, "grad_norm": 0.5996906161308289, "learning_rate": 0.0005, "loss": 1.3297009468078613, "step": 12355 }, { "epoch": 1.4377472655340935, "grad_norm": 0.799140453338623, "learning_rate": 0.0005, "loss": 1.3840464353561401, "step": 12356 }, { "epoch": 1.4378636257854316, "grad_norm": 0.938310980796814, "learning_rate": 0.0005, "loss": 1.6212310791015625, "step": 12357 }, { "epoch": 1.43797998603677, "grad_norm": 1.2617037296295166, "learning_rate": 0.0005, "loss": 1.5416648387908936, "step": 12358 }, { "epoch": 1.438096346288108, "grad_norm": 1.3211753368377686, "learning_rate": 0.0005, "loss": 1.7567415237426758, "step": 12359 }, { "epoch": 1.438212706539446, "grad_norm": 1.1672207117080688, "learning_rate": 0.0005, "loss": 2.8219871520996094, "step": 12360 }, { "epoch": 1.4383290667907842, "grad_norm": 0.9742942452430725, "learning_rate": 0.0005, "loss": 2.0151479244232178, "step": 12361 }, { "epoch": 1.4384454270421223, "grad_norm": 2.374464273452759, "learning_rate": 0.0005, "loss": 1.6647738218307495, "step": 12362 }, { "epoch": 1.4385617872934606, "grad_norm": 1.0375947952270508, "learning_rate": 0.0005, "loss": 2.2425668239593506, "step": 12363 }, { "epoch": 1.4386781475447987, "grad_norm": 1.3335765600204468, "learning_rate": 0.0005, "loss": 1.8294707536697388, "step": 12364 }, { "epoch": 1.4387945077961368, "grad_norm": 0.6751587390899658, "learning_rate": 0.0005, "loss": 1.70885169506073, "step": 12365 }, { "epoch": 1.438910868047475, "grad_norm": 1.0801920890808105, "learning_rate": 0.0005, "loss": 2.2120726108551025, "step": 12366 }, { "epoch": 1.439027228298813, "grad_norm": 0.5116971135139465, "learning_rate": 0.0005, "loss": 1.375396490097046, "step": 12367 }, { "epoch": 1.4391435885501513, "grad_norm": 2.3652398586273193, "learning_rate": 0.0005, "loss": 1.9033758640289307, "step": 12368 }, { "epoch": 1.4392599488014894, "grad_norm": 1.6641759872436523, "learning_rate": 0.0005, "loss": 1.974169373512268, "step": 12369 }, { "epoch": 1.4393763090528275, "grad_norm": 1.4907702207565308, "learning_rate": 0.0005, "loss": 1.6551029682159424, "step": 12370 }, { "epoch": 1.4394926693041656, "grad_norm": 1.0156232118606567, "learning_rate": 0.0005, "loss": 1.9268372058868408, "step": 12371 }, { "epoch": 1.4396090295555037, "grad_norm": 1.2448725700378418, "learning_rate": 0.0005, "loss": 2.555978298187256, "step": 12372 }, { "epoch": 1.439725389806842, "grad_norm": 1.0307778120040894, "learning_rate": 0.0005, "loss": 1.6299530267715454, "step": 12373 }, { "epoch": 1.4398417500581802, "grad_norm": 1.0364255905151367, "learning_rate": 0.0005, "loss": 2.2263314723968506, "step": 12374 }, { "epoch": 1.4399581103095183, "grad_norm": 1.165479302406311, "learning_rate": 0.0005, "loss": 1.955948829650879, "step": 12375 }, { "epoch": 1.4400744705608564, "grad_norm": 0.9643038511276245, "learning_rate": 0.0005, "loss": 2.524998903274536, "step": 12376 }, { "epoch": 1.4401908308121945, "grad_norm": 0.9106166958808899, "learning_rate": 0.0005, "loss": 1.4315321445465088, "step": 12377 }, { "epoch": 1.4403071910635328, "grad_norm": 1.289610505104065, "learning_rate": 0.0005, "loss": 1.4352939128875732, "step": 12378 }, { "epoch": 1.4404235513148709, "grad_norm": 1.1911622285842896, "learning_rate": 0.0005, "loss": 2.4307072162628174, "step": 12379 }, { "epoch": 1.440539911566209, "grad_norm": 2.332700490951538, "learning_rate": 0.0005, "loss": 2.273721694946289, "step": 12380 }, { "epoch": 1.440656271817547, "grad_norm": 1.4755960702896118, "learning_rate": 0.0005, "loss": 2.2764384746551514, "step": 12381 }, { "epoch": 1.4407726320688852, "grad_norm": 0.6581947803497314, "learning_rate": 0.0005, "loss": 1.5200260877609253, "step": 12382 }, { "epoch": 1.4408889923202235, "grad_norm": 1.4128587245941162, "learning_rate": 0.0005, "loss": 1.8589903116226196, "step": 12383 }, { "epoch": 1.4410053525715616, "grad_norm": 0.9399765133857727, "learning_rate": 0.0005, "loss": 1.3916925191879272, "step": 12384 }, { "epoch": 1.4411217128228997, "grad_norm": 1.0917854309082031, "learning_rate": 0.0005, "loss": 2.825836658477783, "step": 12385 }, { "epoch": 1.4412380730742378, "grad_norm": 0.5433253049850464, "learning_rate": 0.0005, "loss": 1.132089376449585, "step": 12386 }, { "epoch": 1.441354433325576, "grad_norm": 1.2037053108215332, "learning_rate": 0.0005, "loss": 1.6721930503845215, "step": 12387 }, { "epoch": 1.4414707935769142, "grad_norm": 1.0828412771224976, "learning_rate": 0.0005, "loss": 1.9523104429244995, "step": 12388 }, { "epoch": 1.4415871538282523, "grad_norm": 1.1051688194274902, "learning_rate": 0.0005, "loss": 2.0191593170166016, "step": 12389 }, { "epoch": 1.4417035140795904, "grad_norm": 0.9688310623168945, "learning_rate": 0.0005, "loss": 1.998904824256897, "step": 12390 }, { "epoch": 1.4418198743309285, "grad_norm": 1.0098116397857666, "learning_rate": 0.0005, "loss": 1.3407971858978271, "step": 12391 }, { "epoch": 1.4419362345822666, "grad_norm": 1.0319020748138428, "learning_rate": 0.0005, "loss": 1.437157154083252, "step": 12392 }, { "epoch": 1.442052594833605, "grad_norm": 1.0372116565704346, "learning_rate": 0.0005, "loss": 2.246976375579834, "step": 12393 }, { "epoch": 1.442168955084943, "grad_norm": 0.9596474170684814, "learning_rate": 0.0005, "loss": 2.3804736137390137, "step": 12394 }, { "epoch": 1.4422853153362811, "grad_norm": 0.8292890191078186, "learning_rate": 0.0005, "loss": 1.877824068069458, "step": 12395 }, { "epoch": 1.4424016755876192, "grad_norm": 0.7867659330368042, "learning_rate": 0.0005, "loss": 1.2369868755340576, "step": 12396 }, { "epoch": 1.4425180358389573, "grad_norm": 1.2719142436981201, "learning_rate": 0.0005, "loss": 2.577353000640869, "step": 12397 }, { "epoch": 1.4426343960902956, "grad_norm": 0.7861093282699585, "learning_rate": 0.0005, "loss": 1.7778507471084595, "step": 12398 }, { "epoch": 1.4427507563416337, "grad_norm": 0.8802685737609863, "learning_rate": 0.0005, "loss": 1.648980736732483, "step": 12399 }, { "epoch": 1.4428671165929718, "grad_norm": 1.3811124563217163, "learning_rate": 0.0005, "loss": 2.0919885635375977, "step": 12400 }, { "epoch": 1.44298347684431, "grad_norm": 0.9448145627975464, "learning_rate": 0.0005, "loss": 1.4787511825561523, "step": 12401 }, { "epoch": 1.443099837095648, "grad_norm": 1.8861721754074097, "learning_rate": 0.0005, "loss": 2.1126549243927, "step": 12402 }, { "epoch": 1.4432161973469864, "grad_norm": 1.022936463356018, "learning_rate": 0.0005, "loss": 1.9821372032165527, "step": 12403 }, { "epoch": 1.4433325575983245, "grad_norm": 1.2131491899490356, "learning_rate": 0.0005, "loss": 2.5277488231658936, "step": 12404 }, { "epoch": 1.4434489178496626, "grad_norm": 1.4033639430999756, "learning_rate": 0.0005, "loss": 2.0207924842834473, "step": 12405 }, { "epoch": 1.4435652781010007, "grad_norm": 1.6060878038406372, "learning_rate": 0.0005, "loss": 1.7036736011505127, "step": 12406 }, { "epoch": 1.4436816383523388, "grad_norm": 1.158766269683838, "learning_rate": 0.0005, "loss": 2.222221612930298, "step": 12407 }, { "epoch": 1.443797998603677, "grad_norm": 0.8371630311012268, "learning_rate": 0.0005, "loss": 1.511613368988037, "step": 12408 }, { "epoch": 1.4439143588550152, "grad_norm": 1.477290153503418, "learning_rate": 0.0005, "loss": 1.8587329387664795, "step": 12409 }, { "epoch": 1.4440307191063533, "grad_norm": 1.4292190074920654, "learning_rate": 0.0005, "loss": 2.1063709259033203, "step": 12410 }, { "epoch": 1.4441470793576914, "grad_norm": 0.974929928779602, "learning_rate": 0.0005, "loss": 1.7832393646240234, "step": 12411 }, { "epoch": 1.4442634396090295, "grad_norm": 0.5056607127189636, "learning_rate": 0.0005, "loss": 1.1877386569976807, "step": 12412 }, { "epoch": 1.4443797998603678, "grad_norm": 0.9790412187576294, "learning_rate": 0.0005, "loss": 1.7913615703582764, "step": 12413 }, { "epoch": 1.444496160111706, "grad_norm": 1.2496447563171387, "learning_rate": 0.0005, "loss": 1.7007026672363281, "step": 12414 }, { "epoch": 1.444612520363044, "grad_norm": 1.1101353168487549, "learning_rate": 0.0005, "loss": 1.7430176734924316, "step": 12415 }, { "epoch": 1.444728880614382, "grad_norm": 1.147391438484192, "learning_rate": 0.0005, "loss": 2.3548531532287598, "step": 12416 }, { "epoch": 1.4448452408657202, "grad_norm": 1.5359042882919312, "learning_rate": 0.0005, "loss": 2.3740148544311523, "step": 12417 }, { "epoch": 1.4449616011170585, "grad_norm": 1.5699480772018433, "learning_rate": 0.0005, "loss": 2.3315799236297607, "step": 12418 }, { "epoch": 1.4450779613683966, "grad_norm": 1.0310605764389038, "learning_rate": 0.0005, "loss": 2.0686051845550537, "step": 12419 }, { "epoch": 1.4451943216197347, "grad_norm": 1.111596703529358, "learning_rate": 0.0005, "loss": 1.9293655157089233, "step": 12420 }, { "epoch": 1.4453106818710728, "grad_norm": 0.7003515362739563, "learning_rate": 0.0005, "loss": 1.7966046333312988, "step": 12421 }, { "epoch": 1.445427042122411, "grad_norm": 0.815872311592102, "learning_rate": 0.0005, "loss": 1.6834746599197388, "step": 12422 }, { "epoch": 1.4455434023737492, "grad_norm": 1.0036320686340332, "learning_rate": 0.0005, "loss": 1.9132492542266846, "step": 12423 }, { "epoch": 1.4456597626250873, "grad_norm": 0.43652212619781494, "learning_rate": 0.0005, "loss": 0.9612328410148621, "step": 12424 }, { "epoch": 1.4457761228764254, "grad_norm": 0.9698758721351624, "learning_rate": 0.0005, "loss": 1.8886117935180664, "step": 12425 }, { "epoch": 1.4458924831277635, "grad_norm": 1.3675414323806763, "learning_rate": 0.0005, "loss": 2.2309257984161377, "step": 12426 }, { "epoch": 1.4460088433791016, "grad_norm": 1.1376781463623047, "learning_rate": 0.0005, "loss": 1.620912790298462, "step": 12427 }, { "epoch": 1.44612520363044, "grad_norm": 0.9272350668907166, "learning_rate": 0.0005, "loss": 2.2277297973632812, "step": 12428 }, { "epoch": 1.446241563881778, "grad_norm": 1.4537837505340576, "learning_rate": 0.0005, "loss": 1.6804978847503662, "step": 12429 }, { "epoch": 1.4463579241331161, "grad_norm": 1.2840179204940796, "learning_rate": 0.0005, "loss": 1.8132579326629639, "step": 12430 }, { "epoch": 1.4464742843844542, "grad_norm": 0.8456588387489319, "learning_rate": 0.0005, "loss": 1.539232611656189, "step": 12431 }, { "epoch": 1.4465906446357923, "grad_norm": 1.0909461975097656, "learning_rate": 0.0005, "loss": 1.6016062498092651, "step": 12432 }, { "epoch": 1.4467070048871307, "grad_norm": 1.1339651346206665, "learning_rate": 0.0005, "loss": 1.5502976179122925, "step": 12433 }, { "epoch": 1.4468233651384688, "grad_norm": 1.0723514556884766, "learning_rate": 0.0005, "loss": 1.9372698068618774, "step": 12434 }, { "epoch": 1.4469397253898069, "grad_norm": 2.326110601425171, "learning_rate": 0.0005, "loss": 2.1343369483947754, "step": 12435 }, { "epoch": 1.447056085641145, "grad_norm": 2.2689242362976074, "learning_rate": 0.0005, "loss": 2.60172700881958, "step": 12436 }, { "epoch": 1.447172445892483, "grad_norm": 0.9958102107048035, "learning_rate": 0.0005, "loss": 1.5901250839233398, "step": 12437 }, { "epoch": 1.4472888061438214, "grad_norm": 1.7724982500076294, "learning_rate": 0.0005, "loss": 2.4867825508117676, "step": 12438 }, { "epoch": 1.4474051663951595, "grad_norm": 1.0858252048492432, "learning_rate": 0.0005, "loss": 2.050698757171631, "step": 12439 }, { "epoch": 1.4475215266464976, "grad_norm": 1.2516320943832397, "learning_rate": 0.0005, "loss": 1.927006483078003, "step": 12440 }, { "epoch": 1.4476378868978357, "grad_norm": 1.0104897022247314, "learning_rate": 0.0005, "loss": 2.1315674781799316, "step": 12441 }, { "epoch": 1.4477542471491738, "grad_norm": 1.073191523551941, "learning_rate": 0.0005, "loss": 1.717366099357605, "step": 12442 }, { "epoch": 1.447870607400512, "grad_norm": 0.7672179937362671, "learning_rate": 0.0005, "loss": 1.3558770418167114, "step": 12443 }, { "epoch": 1.4479869676518502, "grad_norm": 1.2386759519577026, "learning_rate": 0.0005, "loss": 2.277554512023926, "step": 12444 }, { "epoch": 1.4481033279031883, "grad_norm": 1.0269325971603394, "learning_rate": 0.0005, "loss": 1.7198575735092163, "step": 12445 }, { "epoch": 1.4482196881545264, "grad_norm": 2.540377140045166, "learning_rate": 0.0005, "loss": 1.4748964309692383, "step": 12446 }, { "epoch": 1.4483360484058645, "grad_norm": 0.9589330554008484, "learning_rate": 0.0005, "loss": 2.1646194458007812, "step": 12447 }, { "epoch": 1.4484524086572028, "grad_norm": 1.2790005207061768, "learning_rate": 0.0005, "loss": 2.3993701934814453, "step": 12448 }, { "epoch": 1.448568768908541, "grad_norm": 0.673856258392334, "learning_rate": 0.0005, "loss": 1.0397475957870483, "step": 12449 }, { "epoch": 1.448685129159879, "grad_norm": 0.7968916296958923, "learning_rate": 0.0005, "loss": 1.1612436771392822, "step": 12450 }, { "epoch": 1.4488014894112171, "grad_norm": 1.0696213245391846, "learning_rate": 0.0005, "loss": 2.1889524459838867, "step": 12451 }, { "epoch": 1.4489178496625552, "grad_norm": 0.8785756230354309, "learning_rate": 0.0005, "loss": 1.3723292350769043, "step": 12452 }, { "epoch": 1.4490342099138935, "grad_norm": 1.051623821258545, "learning_rate": 0.0005, "loss": 1.537470817565918, "step": 12453 }, { "epoch": 1.4491505701652316, "grad_norm": 0.9813816547393799, "learning_rate": 0.0005, "loss": 1.7869629859924316, "step": 12454 }, { "epoch": 1.4492669304165697, "grad_norm": 1.012731671333313, "learning_rate": 0.0005, "loss": 2.226323127746582, "step": 12455 }, { "epoch": 1.4493832906679078, "grad_norm": 1.0814462900161743, "learning_rate": 0.0005, "loss": 2.105372905731201, "step": 12456 }, { "epoch": 1.449499650919246, "grad_norm": 1.1126844882965088, "learning_rate": 0.0005, "loss": 1.8149183988571167, "step": 12457 }, { "epoch": 1.4496160111705843, "grad_norm": 0.978208065032959, "learning_rate": 0.0005, "loss": 1.1669836044311523, "step": 12458 }, { "epoch": 1.4497323714219224, "grad_norm": 1.1040138006210327, "learning_rate": 0.0005, "loss": 2.089466094970703, "step": 12459 }, { "epoch": 1.4498487316732604, "grad_norm": 1.313637375831604, "learning_rate": 0.0005, "loss": 2.158947229385376, "step": 12460 }, { "epoch": 1.4499650919245985, "grad_norm": 1.7385612726211548, "learning_rate": 0.0005, "loss": 1.672187328338623, "step": 12461 }, { "epoch": 1.4500814521759366, "grad_norm": 0.9080255627632141, "learning_rate": 0.0005, "loss": 1.365574598312378, "step": 12462 }, { "epoch": 1.450197812427275, "grad_norm": 1.079602599143982, "learning_rate": 0.0005, "loss": 2.0354719161987305, "step": 12463 }, { "epoch": 1.450314172678613, "grad_norm": 0.823197066783905, "learning_rate": 0.0005, "loss": 1.794910192489624, "step": 12464 }, { "epoch": 1.4504305329299512, "grad_norm": 1.054993987083435, "learning_rate": 0.0005, "loss": 1.9706804752349854, "step": 12465 }, { "epoch": 1.4505468931812893, "grad_norm": 1.004539966583252, "learning_rate": 0.0005, "loss": 1.5395941734313965, "step": 12466 }, { "epoch": 1.4506632534326274, "grad_norm": 0.5082913637161255, "learning_rate": 0.0005, "loss": 1.0589981079101562, "step": 12467 }, { "epoch": 1.4507796136839657, "grad_norm": 1.1406810283660889, "learning_rate": 0.0005, "loss": 2.2323925495147705, "step": 12468 }, { "epoch": 1.4508959739353036, "grad_norm": 1.4353471994400024, "learning_rate": 0.0005, "loss": 2.0923705101013184, "step": 12469 }, { "epoch": 1.4510123341866419, "grad_norm": 1.9054383039474487, "learning_rate": 0.0005, "loss": 2.5612411499023438, "step": 12470 }, { "epoch": 1.45112869443798, "grad_norm": 0.9034773707389832, "learning_rate": 0.0005, "loss": 1.8566038608551025, "step": 12471 }, { "epoch": 1.451245054689318, "grad_norm": 1.7035974264144897, "learning_rate": 0.0005, "loss": 2.3597333431243896, "step": 12472 }, { "epoch": 1.4513614149406564, "grad_norm": 1.1515398025512695, "learning_rate": 0.0005, "loss": 1.5469794273376465, "step": 12473 }, { "epoch": 1.4514777751919943, "grad_norm": 0.9692118763923645, "learning_rate": 0.0005, "loss": 1.3783892393112183, "step": 12474 }, { "epoch": 1.4515941354433326, "grad_norm": 1.6646369695663452, "learning_rate": 0.0005, "loss": 2.633147954940796, "step": 12475 }, { "epoch": 1.4517104956946707, "grad_norm": 1.3927767276763916, "learning_rate": 0.0005, "loss": 1.7677874565124512, "step": 12476 }, { "epoch": 1.4518268559460088, "grad_norm": 1.4149606227874756, "learning_rate": 0.0005, "loss": 1.6879324913024902, "step": 12477 }, { "epoch": 1.4519432161973471, "grad_norm": 1.2250256538391113, "learning_rate": 0.0005, "loss": 3.0016701221466064, "step": 12478 }, { "epoch": 1.452059576448685, "grad_norm": 1.3309801816940308, "learning_rate": 0.0005, "loss": 2.038743495941162, "step": 12479 }, { "epoch": 1.4521759367000233, "grad_norm": 2.1949524879455566, "learning_rate": 0.0005, "loss": 1.90818452835083, "step": 12480 }, { "epoch": 1.4522922969513614, "grad_norm": 1.0454329252243042, "learning_rate": 0.0005, "loss": 1.833867073059082, "step": 12481 }, { "epoch": 1.4524086572026995, "grad_norm": 1.050533652305603, "learning_rate": 0.0005, "loss": 2.5581469535827637, "step": 12482 }, { "epoch": 1.4525250174540378, "grad_norm": 1.2361295223236084, "learning_rate": 0.0005, "loss": 1.964252233505249, "step": 12483 }, { "epoch": 1.4526413777053757, "grad_norm": 1.1212784051895142, "learning_rate": 0.0005, "loss": 2.1271257400512695, "step": 12484 }, { "epoch": 1.452757737956714, "grad_norm": 1.140943169593811, "learning_rate": 0.0005, "loss": 2.383305311203003, "step": 12485 }, { "epoch": 1.4528740982080521, "grad_norm": 1.5947706699371338, "learning_rate": 0.0005, "loss": 2.25561785697937, "step": 12486 }, { "epoch": 1.4529904584593902, "grad_norm": 1.1148465871810913, "learning_rate": 0.0005, "loss": 2.512683153152466, "step": 12487 }, { "epoch": 1.4531068187107286, "grad_norm": 1.4373908042907715, "learning_rate": 0.0005, "loss": 1.4717097282409668, "step": 12488 }, { "epoch": 1.4532231789620664, "grad_norm": 0.8811745643615723, "learning_rate": 0.0005, "loss": 2.076228618621826, "step": 12489 }, { "epoch": 1.4533395392134048, "grad_norm": 1.3355461359024048, "learning_rate": 0.0005, "loss": 2.224987745285034, "step": 12490 }, { "epoch": 1.4534558994647429, "grad_norm": 1.3178420066833496, "learning_rate": 0.0005, "loss": 2.3757307529449463, "step": 12491 }, { "epoch": 1.453572259716081, "grad_norm": 1.080371618270874, "learning_rate": 0.0005, "loss": 1.8903496265411377, "step": 12492 }, { "epoch": 1.453688619967419, "grad_norm": 1.1755175590515137, "learning_rate": 0.0005, "loss": 2.1640403270721436, "step": 12493 }, { "epoch": 1.4538049802187571, "grad_norm": 1.3521255254745483, "learning_rate": 0.0005, "loss": 1.9926221370697021, "step": 12494 }, { "epoch": 1.4539213404700955, "grad_norm": 1.3209434747695923, "learning_rate": 0.0005, "loss": 1.4721519947052002, "step": 12495 }, { "epoch": 1.4540377007214336, "grad_norm": 1.3329063653945923, "learning_rate": 0.0005, "loss": 2.0262603759765625, "step": 12496 }, { "epoch": 1.4541540609727717, "grad_norm": 1.0402027368545532, "learning_rate": 0.0005, "loss": 1.3878028392791748, "step": 12497 }, { "epoch": 1.4542704212241098, "grad_norm": 0.8477565050125122, "learning_rate": 0.0005, "loss": 1.6123700141906738, "step": 12498 }, { "epoch": 1.4543867814754479, "grad_norm": 1.1184990406036377, "learning_rate": 0.0005, "loss": 1.4208192825317383, "step": 12499 }, { "epoch": 1.4545031417267862, "grad_norm": 1.1456881761550903, "learning_rate": 0.0005, "loss": 1.8124409914016724, "step": 12500 }, { "epoch": 1.4546195019781243, "grad_norm": 1.0952916145324707, "learning_rate": 0.0005, "loss": 2.4511749744415283, "step": 12501 }, { "epoch": 1.4547358622294624, "grad_norm": 1.0249441862106323, "learning_rate": 0.0005, "loss": 1.751025915145874, "step": 12502 }, { "epoch": 1.4548522224808005, "grad_norm": 0.5044689774513245, "learning_rate": 0.0005, "loss": 1.0149736404418945, "step": 12503 }, { "epoch": 1.4549685827321386, "grad_norm": 1.1652501821517944, "learning_rate": 0.0005, "loss": 2.6068878173828125, "step": 12504 }, { "epoch": 1.455084942983477, "grad_norm": 0.8658007383346558, "learning_rate": 0.0005, "loss": 1.690042495727539, "step": 12505 }, { "epoch": 1.455201303234815, "grad_norm": 1.2599365711212158, "learning_rate": 0.0005, "loss": 1.9530625343322754, "step": 12506 }, { "epoch": 1.455317663486153, "grad_norm": 0.9805769920349121, "learning_rate": 0.0005, "loss": 1.9064276218414307, "step": 12507 }, { "epoch": 1.4554340237374912, "grad_norm": 0.9279906749725342, "learning_rate": 0.0005, "loss": 2.183323860168457, "step": 12508 }, { "epoch": 1.4555503839888293, "grad_norm": 1.1817432641983032, "learning_rate": 0.0005, "loss": 1.7541003227233887, "step": 12509 }, { "epoch": 1.4556667442401676, "grad_norm": 1.0399459600448608, "learning_rate": 0.0005, "loss": 2.3318967819213867, "step": 12510 }, { "epoch": 1.4557831044915057, "grad_norm": 1.1171455383300781, "learning_rate": 0.0005, "loss": 2.5545802116394043, "step": 12511 }, { "epoch": 1.4558994647428438, "grad_norm": 1.1883387565612793, "learning_rate": 0.0005, "loss": 2.1316964626312256, "step": 12512 }, { "epoch": 1.456015824994182, "grad_norm": 0.6613398194313049, "learning_rate": 0.0005, "loss": 1.6582863330841064, "step": 12513 }, { "epoch": 1.45613218524552, "grad_norm": 1.5469433069229126, "learning_rate": 0.0005, "loss": 2.0395755767822266, "step": 12514 }, { "epoch": 1.4562485454968583, "grad_norm": 1.9749504327774048, "learning_rate": 0.0005, "loss": 2.2214760780334473, "step": 12515 }, { "epoch": 1.4563649057481964, "grad_norm": 1.3938759565353394, "learning_rate": 0.0005, "loss": 1.571035623550415, "step": 12516 }, { "epoch": 1.4564812659995345, "grad_norm": 1.1720504760742188, "learning_rate": 0.0005, "loss": 1.5725438594818115, "step": 12517 }, { "epoch": 1.4565976262508726, "grad_norm": 1.2500665187835693, "learning_rate": 0.0005, "loss": 1.7779598236083984, "step": 12518 }, { "epoch": 1.4567139865022107, "grad_norm": 1.6138609647750854, "learning_rate": 0.0005, "loss": 2.1427059173583984, "step": 12519 }, { "epoch": 1.456830346753549, "grad_norm": 1.1343845129013062, "learning_rate": 0.0005, "loss": 1.456611156463623, "step": 12520 }, { "epoch": 1.4569467070048872, "grad_norm": 0.9284986257553101, "learning_rate": 0.0005, "loss": 1.8491750955581665, "step": 12521 }, { "epoch": 1.4570630672562253, "grad_norm": 1.1168878078460693, "learning_rate": 0.0005, "loss": 2.3425841331481934, "step": 12522 }, { "epoch": 1.4571794275075634, "grad_norm": 1.3312318325042725, "learning_rate": 0.0005, "loss": 1.396224021911621, "step": 12523 }, { "epoch": 1.4572957877589015, "grad_norm": 1.2474348545074463, "learning_rate": 0.0005, "loss": 1.5074810981750488, "step": 12524 }, { "epoch": 1.4574121480102398, "grad_norm": 1.2698439359664917, "learning_rate": 0.0005, "loss": 1.8733638525009155, "step": 12525 }, { "epoch": 1.4575285082615779, "grad_norm": 0.7511066198348999, "learning_rate": 0.0005, "loss": 1.2330189943313599, "step": 12526 }, { "epoch": 1.457644868512916, "grad_norm": 1.1816412210464478, "learning_rate": 0.0005, "loss": 2.454442262649536, "step": 12527 }, { "epoch": 1.457761228764254, "grad_norm": 1.0512564182281494, "learning_rate": 0.0005, "loss": 2.313612222671509, "step": 12528 }, { "epoch": 1.4578775890155922, "grad_norm": 1.2716244459152222, "learning_rate": 0.0005, "loss": 2.046264410018921, "step": 12529 }, { "epoch": 1.4579939492669305, "grad_norm": 0.9325674772262573, "learning_rate": 0.0005, "loss": 2.0546658039093018, "step": 12530 }, { "epoch": 1.4581103095182686, "grad_norm": 1.655846118927002, "learning_rate": 0.0005, "loss": 2.8664143085479736, "step": 12531 }, { "epoch": 1.4582266697696067, "grad_norm": 0.7767854928970337, "learning_rate": 0.0005, "loss": 1.9400500059127808, "step": 12532 }, { "epoch": 1.4583430300209448, "grad_norm": 1.1859700679779053, "learning_rate": 0.0005, "loss": 1.9138906002044678, "step": 12533 }, { "epoch": 1.4584593902722829, "grad_norm": 0.8274246454238892, "learning_rate": 0.0005, "loss": 1.364737629890442, "step": 12534 }, { "epoch": 1.4585757505236212, "grad_norm": 0.9275668859481812, "learning_rate": 0.0005, "loss": 2.165126323699951, "step": 12535 }, { "epoch": 1.4586921107749593, "grad_norm": 1.8264672756195068, "learning_rate": 0.0005, "loss": 2.2100563049316406, "step": 12536 }, { "epoch": 1.4588084710262974, "grad_norm": 1.247198462486267, "learning_rate": 0.0005, "loss": 2.140733242034912, "step": 12537 }, { "epoch": 1.4589248312776355, "grad_norm": 0.9824536442756653, "learning_rate": 0.0005, "loss": 2.1961989402770996, "step": 12538 }, { "epoch": 1.4590411915289736, "grad_norm": 0.6244957447052002, "learning_rate": 0.0005, "loss": 1.274274468421936, "step": 12539 }, { "epoch": 1.459157551780312, "grad_norm": 0.831852376461029, "learning_rate": 0.0005, "loss": 1.7199249267578125, "step": 12540 }, { "epoch": 1.45927391203165, "grad_norm": 1.1339868307113647, "learning_rate": 0.0005, "loss": 2.6363070011138916, "step": 12541 }, { "epoch": 1.4593902722829881, "grad_norm": 1.2927279472351074, "learning_rate": 0.0005, "loss": 1.7398083209991455, "step": 12542 }, { "epoch": 1.4595066325343262, "grad_norm": 1.026332974433899, "learning_rate": 0.0005, "loss": 1.6764719486236572, "step": 12543 }, { "epoch": 1.4596229927856643, "grad_norm": 1.9403201341629028, "learning_rate": 0.0005, "loss": 1.9293336868286133, "step": 12544 }, { "epoch": 1.4597393530370026, "grad_norm": 1.8261388540267944, "learning_rate": 0.0005, "loss": 2.5536346435546875, "step": 12545 }, { "epoch": 1.4598557132883407, "grad_norm": 0.9384727478027344, "learning_rate": 0.0005, "loss": 1.6804149150848389, "step": 12546 }, { "epoch": 1.4599720735396788, "grad_norm": 1.2748221158981323, "learning_rate": 0.0005, "loss": 2.627115488052368, "step": 12547 }, { "epoch": 1.460088433791017, "grad_norm": 1.2156550884246826, "learning_rate": 0.0005, "loss": 2.310120105743408, "step": 12548 }, { "epoch": 1.460204794042355, "grad_norm": 0.8979206085205078, "learning_rate": 0.0005, "loss": 1.7969129085540771, "step": 12549 }, { "epoch": 1.4603211542936934, "grad_norm": 1.0619564056396484, "learning_rate": 0.0005, "loss": 2.285170078277588, "step": 12550 }, { "epoch": 1.4604375145450315, "grad_norm": 1.7921090126037598, "learning_rate": 0.0005, "loss": 2.7432777881622314, "step": 12551 }, { "epoch": 1.4605538747963696, "grad_norm": 0.7817202210426331, "learning_rate": 0.0005, "loss": 1.4858438968658447, "step": 12552 }, { "epoch": 1.4606702350477077, "grad_norm": 0.7993564009666443, "learning_rate": 0.0005, "loss": 1.207211971282959, "step": 12553 }, { "epoch": 1.4607865952990458, "grad_norm": 1.2839592695236206, "learning_rate": 0.0005, "loss": 2.0208966732025146, "step": 12554 }, { "epoch": 1.460902955550384, "grad_norm": 1.550074577331543, "learning_rate": 0.0005, "loss": 2.325939655303955, "step": 12555 }, { "epoch": 1.4610193158017222, "grad_norm": 1.663559079170227, "learning_rate": 0.0005, "loss": 1.9034662246704102, "step": 12556 }, { "epoch": 1.4611356760530603, "grad_norm": 1.0287461280822754, "learning_rate": 0.0005, "loss": 1.9502179622650146, "step": 12557 }, { "epoch": 1.4612520363043984, "grad_norm": 0.9695848226547241, "learning_rate": 0.0005, "loss": 1.4978649616241455, "step": 12558 }, { "epoch": 1.4613683965557365, "grad_norm": 1.390615701675415, "learning_rate": 0.0005, "loss": 1.8861254453659058, "step": 12559 }, { "epoch": 1.4614847568070748, "grad_norm": 1.5366661548614502, "learning_rate": 0.0005, "loss": 2.1681156158447266, "step": 12560 }, { "epoch": 1.461601117058413, "grad_norm": 1.2592309713363647, "learning_rate": 0.0005, "loss": 2.149354934692383, "step": 12561 }, { "epoch": 1.461717477309751, "grad_norm": 1.0414565801620483, "learning_rate": 0.0005, "loss": 1.7981243133544922, "step": 12562 }, { "epoch": 1.461833837561089, "grad_norm": 0.6561988592147827, "learning_rate": 0.0005, "loss": 1.3376400470733643, "step": 12563 }, { "epoch": 1.4619501978124272, "grad_norm": 1.092366337776184, "learning_rate": 0.0005, "loss": 2.3992042541503906, "step": 12564 }, { "epoch": 1.4620665580637655, "grad_norm": 1.2083237171173096, "learning_rate": 0.0005, "loss": 2.03493070602417, "step": 12565 }, { "epoch": 1.4621829183151036, "grad_norm": 0.7993194460868835, "learning_rate": 0.0005, "loss": 1.6342557668685913, "step": 12566 }, { "epoch": 1.4622992785664417, "grad_norm": 1.9770833253860474, "learning_rate": 0.0005, "loss": 1.8692467212677002, "step": 12567 }, { "epoch": 1.4624156388177798, "grad_norm": 1.3372387886047363, "learning_rate": 0.0005, "loss": 1.605173110961914, "step": 12568 }, { "epoch": 1.462531999069118, "grad_norm": 1.0282082557678223, "learning_rate": 0.0005, "loss": 2.153510093688965, "step": 12569 }, { "epoch": 1.4626483593204562, "grad_norm": 1.4478455781936646, "learning_rate": 0.0005, "loss": 2.142307758331299, "step": 12570 }, { "epoch": 1.4627647195717943, "grad_norm": 1.5016462802886963, "learning_rate": 0.0005, "loss": 1.9129886627197266, "step": 12571 }, { "epoch": 1.4628810798231324, "grad_norm": 2.156653642654419, "learning_rate": 0.0005, "loss": 2.3606958389282227, "step": 12572 }, { "epoch": 1.4629974400744705, "grad_norm": 1.2742595672607422, "learning_rate": 0.0005, "loss": 1.6134419441223145, "step": 12573 }, { "epoch": 1.4631138003258086, "grad_norm": 1.2270104885101318, "learning_rate": 0.0005, "loss": 1.8793755769729614, "step": 12574 }, { "epoch": 1.463230160577147, "grad_norm": 0.897611141204834, "learning_rate": 0.0005, "loss": 1.820864200592041, "step": 12575 }, { "epoch": 1.463346520828485, "grad_norm": 0.9908652305603027, "learning_rate": 0.0005, "loss": 1.5269808769226074, "step": 12576 }, { "epoch": 1.4634628810798231, "grad_norm": 1.4788440465927124, "learning_rate": 0.0005, "loss": 2.2585480213165283, "step": 12577 }, { "epoch": 1.4635792413311612, "grad_norm": 0.9008258581161499, "learning_rate": 0.0005, "loss": 2.029979705810547, "step": 12578 }, { "epoch": 1.4636956015824993, "grad_norm": 1.1011512279510498, "learning_rate": 0.0005, "loss": 1.5886716842651367, "step": 12579 }, { "epoch": 1.4638119618338377, "grad_norm": 1.4231866598129272, "learning_rate": 0.0005, "loss": 1.4914038181304932, "step": 12580 }, { "epoch": 1.4639283220851758, "grad_norm": 1.478178858757019, "learning_rate": 0.0005, "loss": 2.4197516441345215, "step": 12581 }, { "epoch": 1.4640446823365139, "grad_norm": 1.2141331434249878, "learning_rate": 0.0005, "loss": 1.8685961961746216, "step": 12582 }, { "epoch": 1.464161042587852, "grad_norm": 1.1194413900375366, "learning_rate": 0.0005, "loss": 2.2345714569091797, "step": 12583 }, { "epoch": 1.46427740283919, "grad_norm": 1.4274400472640991, "learning_rate": 0.0005, "loss": 2.5011239051818848, "step": 12584 }, { "epoch": 1.4643937630905284, "grad_norm": 1.337330937385559, "learning_rate": 0.0005, "loss": 1.9546579122543335, "step": 12585 }, { "epoch": 1.4645101233418665, "grad_norm": 2.377819061279297, "learning_rate": 0.0005, "loss": 2.242433786392212, "step": 12586 }, { "epoch": 1.4646264835932046, "grad_norm": 1.0140544176101685, "learning_rate": 0.0005, "loss": 1.5581543445587158, "step": 12587 }, { "epoch": 1.4647428438445427, "grad_norm": 2.4339072704315186, "learning_rate": 0.0005, "loss": 2.7074403762817383, "step": 12588 }, { "epoch": 1.4648592040958808, "grad_norm": 0.9438770413398743, "learning_rate": 0.0005, "loss": 1.3377493619918823, "step": 12589 }, { "epoch": 1.464975564347219, "grad_norm": 1.0387502908706665, "learning_rate": 0.0005, "loss": 2.6741557121276855, "step": 12590 }, { "epoch": 1.4650919245985572, "grad_norm": 1.0022341012954712, "learning_rate": 0.0005, "loss": 1.9924551248550415, "step": 12591 }, { "epoch": 1.4652082848498953, "grad_norm": 0.8549808263778687, "learning_rate": 0.0005, "loss": 2.4067325592041016, "step": 12592 }, { "epoch": 1.4653246451012334, "grad_norm": 1.167618751525879, "learning_rate": 0.0005, "loss": 2.2478444576263428, "step": 12593 }, { "epoch": 1.4654410053525715, "grad_norm": 1.5635110139846802, "learning_rate": 0.0005, "loss": 2.530505895614624, "step": 12594 }, { "epoch": 1.4655573656039098, "grad_norm": 1.083410382270813, "learning_rate": 0.0005, "loss": 2.4268672466278076, "step": 12595 }, { "epoch": 1.465673725855248, "grad_norm": 0.6669396758079529, "learning_rate": 0.0005, "loss": 1.7126245498657227, "step": 12596 }, { "epoch": 1.465790086106586, "grad_norm": 1.216465950012207, "learning_rate": 0.0005, "loss": 1.7618792057037354, "step": 12597 }, { "epoch": 1.465906446357924, "grad_norm": 0.9824515581130981, "learning_rate": 0.0005, "loss": 1.7130756378173828, "step": 12598 }, { "epoch": 1.4660228066092622, "grad_norm": 1.0403687953948975, "learning_rate": 0.0005, "loss": 1.8703116178512573, "step": 12599 }, { "epoch": 1.4661391668606005, "grad_norm": 0.7812127470970154, "learning_rate": 0.0005, "loss": 1.4522068500518799, "step": 12600 }, { "epoch": 1.4662555271119386, "grad_norm": 1.1391092538833618, "learning_rate": 0.0005, "loss": 1.9037129878997803, "step": 12601 }, { "epoch": 1.4663718873632767, "grad_norm": 1.286102533340454, "learning_rate": 0.0005, "loss": 2.454622745513916, "step": 12602 }, { "epoch": 1.4664882476146148, "grad_norm": 1.1582344770431519, "learning_rate": 0.0005, "loss": 1.6628893613815308, "step": 12603 }, { "epoch": 1.466604607865953, "grad_norm": 1.034296989440918, "learning_rate": 0.0005, "loss": 1.5512616634368896, "step": 12604 }, { "epoch": 1.4667209681172912, "grad_norm": 0.7004542350769043, "learning_rate": 0.0005, "loss": 1.2758228778839111, "step": 12605 }, { "epoch": 1.4668373283686293, "grad_norm": 1.441772699356079, "learning_rate": 0.0005, "loss": 2.354649066925049, "step": 12606 }, { "epoch": 1.4669536886199674, "grad_norm": 0.8806883692741394, "learning_rate": 0.0005, "loss": 1.4331177473068237, "step": 12607 }, { "epoch": 1.4670700488713055, "grad_norm": 0.6754941344261169, "learning_rate": 0.0005, "loss": 1.3656504154205322, "step": 12608 }, { "epoch": 1.4671864091226436, "grad_norm": 1.0179640054702759, "learning_rate": 0.0005, "loss": 2.2935714721679688, "step": 12609 }, { "epoch": 1.467302769373982, "grad_norm": 1.7182201147079468, "learning_rate": 0.0005, "loss": 2.0603156089782715, "step": 12610 }, { "epoch": 1.46741912962532, "grad_norm": 0.7718592882156372, "learning_rate": 0.0005, "loss": 1.7920211553573608, "step": 12611 }, { "epoch": 1.4675354898766582, "grad_norm": 0.9368522763252258, "learning_rate": 0.0005, "loss": 1.3233901262283325, "step": 12612 }, { "epoch": 1.4676518501279963, "grad_norm": 0.712476909160614, "learning_rate": 0.0005, "loss": 1.5254311561584473, "step": 12613 }, { "epoch": 1.4677682103793344, "grad_norm": 1.4357036352157593, "learning_rate": 0.0005, "loss": 2.687030553817749, "step": 12614 }, { "epoch": 1.4678845706306727, "grad_norm": 1.1789486408233643, "learning_rate": 0.0005, "loss": 2.14076828956604, "step": 12615 }, { "epoch": 1.4680009308820108, "grad_norm": 0.8721911907196045, "learning_rate": 0.0005, "loss": 1.89845871925354, "step": 12616 }, { "epoch": 1.4681172911333489, "grad_norm": 1.4436219930648804, "learning_rate": 0.0005, "loss": 1.9633398056030273, "step": 12617 }, { "epoch": 1.468233651384687, "grad_norm": 1.0364561080932617, "learning_rate": 0.0005, "loss": 1.6850444078445435, "step": 12618 }, { "epoch": 1.468350011636025, "grad_norm": 1.3509352207183838, "learning_rate": 0.0005, "loss": 1.9930455684661865, "step": 12619 }, { "epoch": 1.4684663718873634, "grad_norm": 1.1207733154296875, "learning_rate": 0.0005, "loss": 1.8943558931350708, "step": 12620 }, { "epoch": 1.4685827321387015, "grad_norm": 1.0770714282989502, "learning_rate": 0.0005, "loss": 2.120218276977539, "step": 12621 }, { "epoch": 1.4686990923900396, "grad_norm": 1.0561151504516602, "learning_rate": 0.0005, "loss": 2.061497688293457, "step": 12622 }, { "epoch": 1.4688154526413777, "grad_norm": 1.3065494298934937, "learning_rate": 0.0005, "loss": 2.6118030548095703, "step": 12623 }, { "epoch": 1.4689318128927158, "grad_norm": 1.0868029594421387, "learning_rate": 0.0005, "loss": 2.0289764404296875, "step": 12624 }, { "epoch": 1.4690481731440541, "grad_norm": 1.6646852493286133, "learning_rate": 0.0005, "loss": 2.6810336112976074, "step": 12625 }, { "epoch": 1.4691645333953922, "grad_norm": 2.491870641708374, "learning_rate": 0.0005, "loss": 1.6517192125320435, "step": 12626 }, { "epoch": 1.4692808936467303, "grad_norm": 1.296068549156189, "learning_rate": 0.0005, "loss": 1.8707563877105713, "step": 12627 }, { "epoch": 1.4693972538980684, "grad_norm": 1.1538445949554443, "learning_rate": 0.0005, "loss": 1.8474922180175781, "step": 12628 }, { "epoch": 1.4695136141494065, "grad_norm": 1.1832146644592285, "learning_rate": 0.0005, "loss": 2.5454070568084717, "step": 12629 }, { "epoch": 1.4696299744007448, "grad_norm": 1.3450655937194824, "learning_rate": 0.0005, "loss": 2.078565835952759, "step": 12630 }, { "epoch": 1.469746334652083, "grad_norm": 1.2205226421356201, "learning_rate": 0.0005, "loss": 2.1124846935272217, "step": 12631 }, { "epoch": 1.469862694903421, "grad_norm": 6.644641876220703, "learning_rate": 0.0005, "loss": 1.8313685655593872, "step": 12632 }, { "epoch": 1.4699790551547591, "grad_norm": 1.1322715282440186, "learning_rate": 0.0005, "loss": 2.310673475265503, "step": 12633 }, { "epoch": 1.4700954154060972, "grad_norm": 1.5746337175369263, "learning_rate": 0.0005, "loss": 1.6970529556274414, "step": 12634 }, { "epoch": 1.4702117756574355, "grad_norm": 1.1258935928344727, "learning_rate": 0.0005, "loss": 1.9714088439941406, "step": 12635 }, { "epoch": 1.4703281359087734, "grad_norm": 1.646490454673767, "learning_rate": 0.0005, "loss": 1.9561959505081177, "step": 12636 }, { "epoch": 1.4704444961601117, "grad_norm": 1.5234699249267578, "learning_rate": 0.0005, "loss": 2.0247528553009033, "step": 12637 }, { "epoch": 1.4705608564114498, "grad_norm": 1.0338901281356812, "learning_rate": 0.0005, "loss": 2.241011619567871, "step": 12638 }, { "epoch": 1.470677216662788, "grad_norm": 1.2577646970748901, "learning_rate": 0.0005, "loss": 1.7308951616287231, "step": 12639 }, { "epoch": 1.4707935769141263, "grad_norm": 1.0518240928649902, "learning_rate": 0.0005, "loss": 1.808823823928833, "step": 12640 }, { "epoch": 1.4709099371654641, "grad_norm": 0.8366705775260925, "learning_rate": 0.0005, "loss": 1.385127067565918, "step": 12641 }, { "epoch": 1.4710262974168025, "grad_norm": 0.51191246509552, "learning_rate": 0.0005, "loss": 1.002781629562378, "step": 12642 }, { "epoch": 1.4711426576681406, "grad_norm": 1.192986249923706, "learning_rate": 0.0005, "loss": 1.807340145111084, "step": 12643 }, { "epoch": 1.4712590179194787, "grad_norm": 1.599333643913269, "learning_rate": 0.0005, "loss": 2.2778632640838623, "step": 12644 }, { "epoch": 1.471375378170817, "grad_norm": 1.239410400390625, "learning_rate": 0.0005, "loss": 2.469501495361328, "step": 12645 }, { "epoch": 1.4714917384221549, "grad_norm": 1.1092723608016968, "learning_rate": 0.0005, "loss": 1.6615983247756958, "step": 12646 }, { "epoch": 1.4716080986734932, "grad_norm": 0.9600158929824829, "learning_rate": 0.0005, "loss": 2.116076707839966, "step": 12647 }, { "epoch": 1.4717244589248313, "grad_norm": 1.549352765083313, "learning_rate": 0.0005, "loss": 2.586184501647949, "step": 12648 }, { "epoch": 1.4718408191761694, "grad_norm": 0.8285616040229797, "learning_rate": 0.0005, "loss": 1.6330639123916626, "step": 12649 }, { "epoch": 1.4719571794275077, "grad_norm": 1.1226310729980469, "learning_rate": 0.0005, "loss": 1.3986331224441528, "step": 12650 }, { "epoch": 1.4720735396788456, "grad_norm": 1.0023311376571655, "learning_rate": 0.0005, "loss": 1.8212857246398926, "step": 12651 }, { "epoch": 1.472189899930184, "grad_norm": 1.0067049264907837, "learning_rate": 0.0005, "loss": 1.643120288848877, "step": 12652 }, { "epoch": 1.472306260181522, "grad_norm": 2.2402048110961914, "learning_rate": 0.0005, "loss": 2.0007801055908203, "step": 12653 }, { "epoch": 1.47242262043286, "grad_norm": 1.2885740995407104, "learning_rate": 0.0005, "loss": 2.07962965965271, "step": 12654 }, { "epoch": 1.4725389806841982, "grad_norm": 1.2039973735809326, "learning_rate": 0.0005, "loss": 2.2871499061584473, "step": 12655 }, { "epoch": 1.4726553409355363, "grad_norm": 1.0583611726760864, "learning_rate": 0.0005, "loss": 2.107112169265747, "step": 12656 }, { "epoch": 1.4727717011868746, "grad_norm": 0.9918095469474792, "learning_rate": 0.0005, "loss": 1.8864834308624268, "step": 12657 }, { "epoch": 1.4728880614382127, "grad_norm": 1.353052020072937, "learning_rate": 0.0005, "loss": 1.519935131072998, "step": 12658 }, { "epoch": 1.4730044216895508, "grad_norm": 0.8842751383781433, "learning_rate": 0.0005, "loss": 1.848447561264038, "step": 12659 }, { "epoch": 1.473120781940889, "grad_norm": 1.2141382694244385, "learning_rate": 0.0005, "loss": 1.9825273752212524, "step": 12660 }, { "epoch": 1.473237142192227, "grad_norm": 1.0798182487487793, "learning_rate": 0.0005, "loss": 1.5645387172698975, "step": 12661 }, { "epoch": 1.4733535024435653, "grad_norm": 0.6491854786872864, "learning_rate": 0.0005, "loss": 1.061662197113037, "step": 12662 }, { "epoch": 1.4734698626949034, "grad_norm": 1.3354324102401733, "learning_rate": 0.0005, "loss": 2.3992719650268555, "step": 12663 }, { "epoch": 1.4735862229462415, "grad_norm": 0.8569203615188599, "learning_rate": 0.0005, "loss": 2.354412317276001, "step": 12664 }, { "epoch": 1.4737025831975796, "grad_norm": 1.161618947982788, "learning_rate": 0.0005, "loss": 1.826291799545288, "step": 12665 }, { "epoch": 1.4738189434489177, "grad_norm": 0.8616560697555542, "learning_rate": 0.0005, "loss": 1.9610198736190796, "step": 12666 }, { "epoch": 1.473935303700256, "grad_norm": 0.9708389043807983, "learning_rate": 0.0005, "loss": 1.5058059692382812, "step": 12667 }, { "epoch": 1.4740516639515941, "grad_norm": 1.2297027111053467, "learning_rate": 0.0005, "loss": 1.3890910148620605, "step": 12668 }, { "epoch": 1.4741680242029322, "grad_norm": 0.9085683822631836, "learning_rate": 0.0005, "loss": 1.2469472885131836, "step": 12669 }, { "epoch": 1.4742843844542703, "grad_norm": 1.8464958667755127, "learning_rate": 0.0005, "loss": 1.4689233303070068, "step": 12670 }, { "epoch": 1.4744007447056084, "grad_norm": 1.2226289510726929, "learning_rate": 0.0005, "loss": 2.1684536933898926, "step": 12671 }, { "epoch": 1.4745171049569468, "grad_norm": 0.8201925158500671, "learning_rate": 0.0005, "loss": 2.0009946823120117, "step": 12672 }, { "epoch": 1.4746334652082849, "grad_norm": 0.721131443977356, "learning_rate": 0.0005, "loss": 1.4621343612670898, "step": 12673 }, { "epoch": 1.474749825459623, "grad_norm": 1.143966794013977, "learning_rate": 0.0005, "loss": 2.5352067947387695, "step": 12674 }, { "epoch": 1.474866185710961, "grad_norm": 2.1493325233459473, "learning_rate": 0.0005, "loss": 2.609361171722412, "step": 12675 }, { "epoch": 1.4749825459622992, "grad_norm": 0.5964694023132324, "learning_rate": 0.0005, "loss": 1.1361926794052124, "step": 12676 }, { "epoch": 1.4750989062136375, "grad_norm": 1.1077669858932495, "learning_rate": 0.0005, "loss": 2.3581175804138184, "step": 12677 }, { "epoch": 1.4752152664649756, "grad_norm": 0.9648036956787109, "learning_rate": 0.0005, "loss": 2.038940906524658, "step": 12678 }, { "epoch": 1.4753316267163137, "grad_norm": 1.152962565422058, "learning_rate": 0.0005, "loss": 2.265316963195801, "step": 12679 }, { "epoch": 1.4754479869676518, "grad_norm": 1.2466044425964355, "learning_rate": 0.0005, "loss": 2.2007765769958496, "step": 12680 }, { "epoch": 1.4755643472189899, "grad_norm": 0.37743833661079407, "learning_rate": 0.0005, "loss": 0.9145251512527466, "step": 12681 }, { "epoch": 1.4756807074703282, "grad_norm": 1.0690317153930664, "learning_rate": 0.0005, "loss": 1.9505722522735596, "step": 12682 }, { "epoch": 1.4757970677216663, "grad_norm": 1.343558669090271, "learning_rate": 0.0005, "loss": 2.152003765106201, "step": 12683 }, { "epoch": 1.4759134279730044, "grad_norm": 1.2154580354690552, "learning_rate": 0.0005, "loss": 2.099958896636963, "step": 12684 }, { "epoch": 1.4760297882243425, "grad_norm": 1.4292594194412231, "learning_rate": 0.0005, "loss": 2.123915910720825, "step": 12685 }, { "epoch": 1.4761461484756806, "grad_norm": 1.2301180362701416, "learning_rate": 0.0005, "loss": 2.459848403930664, "step": 12686 }, { "epoch": 1.476262508727019, "grad_norm": 0.8470762968063354, "learning_rate": 0.0005, "loss": 1.5748525857925415, "step": 12687 }, { "epoch": 1.476378868978357, "grad_norm": 0.8218814134597778, "learning_rate": 0.0005, "loss": 1.2978730201721191, "step": 12688 }, { "epoch": 1.4764952292296951, "grad_norm": 1.3199834823608398, "learning_rate": 0.0005, "loss": 2.1552839279174805, "step": 12689 }, { "epoch": 1.4766115894810332, "grad_norm": 1.5168012380599976, "learning_rate": 0.0005, "loss": 2.368170738220215, "step": 12690 }, { "epoch": 1.4767279497323713, "grad_norm": 0.7856726050376892, "learning_rate": 0.0005, "loss": 1.5461585521697998, "step": 12691 }, { "epoch": 1.4768443099837096, "grad_norm": 0.9531592130661011, "learning_rate": 0.0005, "loss": 2.1988956928253174, "step": 12692 }, { "epoch": 1.4769606702350477, "grad_norm": 1.3478981256484985, "learning_rate": 0.0005, "loss": 1.923888921737671, "step": 12693 }, { "epoch": 1.4770770304863858, "grad_norm": 1.1499145030975342, "learning_rate": 0.0005, "loss": 2.8505349159240723, "step": 12694 }, { "epoch": 1.477193390737724, "grad_norm": 1.0722235441207886, "learning_rate": 0.0005, "loss": 1.9974162578582764, "step": 12695 }, { "epoch": 1.477309750989062, "grad_norm": 0.894441545009613, "learning_rate": 0.0005, "loss": 1.6733083724975586, "step": 12696 }, { "epoch": 1.4774261112404004, "grad_norm": 0.7318077683448792, "learning_rate": 0.0005, "loss": 1.3498049974441528, "step": 12697 }, { "epoch": 1.4775424714917385, "grad_norm": 1.2487905025482178, "learning_rate": 0.0005, "loss": 1.8416999578475952, "step": 12698 }, { "epoch": 1.4776588317430766, "grad_norm": 0.997003436088562, "learning_rate": 0.0005, "loss": 2.90168833732605, "step": 12699 }, { "epoch": 1.4777751919944146, "grad_norm": 0.8051272034645081, "learning_rate": 0.0005, "loss": 1.8440957069396973, "step": 12700 }, { "epoch": 1.4778915522457527, "grad_norm": 0.9385614991188049, "learning_rate": 0.0005, "loss": 1.7508351802825928, "step": 12701 }, { "epoch": 1.478007912497091, "grad_norm": 0.7925062775611877, "learning_rate": 0.0005, "loss": 1.2725352048873901, "step": 12702 }, { "epoch": 1.4781242727484292, "grad_norm": 1.167256474494934, "learning_rate": 0.0005, "loss": 2.473815441131592, "step": 12703 }, { "epoch": 1.4782406329997673, "grad_norm": 1.3304059505462646, "learning_rate": 0.0005, "loss": 1.5263004302978516, "step": 12704 }, { "epoch": 1.4783569932511054, "grad_norm": 0.9918673038482666, "learning_rate": 0.0005, "loss": 2.2474045753479004, "step": 12705 }, { "epoch": 1.4784733535024435, "grad_norm": 1.1441744565963745, "learning_rate": 0.0005, "loss": 1.9086780548095703, "step": 12706 }, { "epoch": 1.4785897137537818, "grad_norm": 1.0756938457489014, "learning_rate": 0.0005, "loss": 1.693434476852417, "step": 12707 }, { "epoch": 1.4787060740051199, "grad_norm": 0.9739733934402466, "learning_rate": 0.0005, "loss": 1.8381311893463135, "step": 12708 }, { "epoch": 1.478822434256458, "grad_norm": 1.9442933797836304, "learning_rate": 0.0005, "loss": 2.0139057636260986, "step": 12709 }, { "epoch": 1.478938794507796, "grad_norm": 0.7663922905921936, "learning_rate": 0.0005, "loss": 1.3609895706176758, "step": 12710 }, { "epoch": 1.4790551547591342, "grad_norm": 1.0463773012161255, "learning_rate": 0.0005, "loss": 1.2282395362854004, "step": 12711 }, { "epoch": 1.4791715150104725, "grad_norm": 1.3064693212509155, "learning_rate": 0.0005, "loss": 1.9696942567825317, "step": 12712 }, { "epoch": 1.4792878752618106, "grad_norm": 0.8025947213172913, "learning_rate": 0.0005, "loss": 1.6235153675079346, "step": 12713 }, { "epoch": 1.4794042355131487, "grad_norm": 0.9537453055381775, "learning_rate": 0.0005, "loss": 1.6217632293701172, "step": 12714 }, { "epoch": 1.4795205957644868, "grad_norm": 0.7086408138275146, "learning_rate": 0.0005, "loss": 1.3548533916473389, "step": 12715 }, { "epoch": 1.479636956015825, "grad_norm": 0.9256346821784973, "learning_rate": 0.0005, "loss": 1.7691525220870972, "step": 12716 }, { "epoch": 1.4797533162671632, "grad_norm": 1.0672647953033447, "learning_rate": 0.0005, "loss": 2.501767635345459, "step": 12717 }, { "epoch": 1.4798696765185013, "grad_norm": 0.739907443523407, "learning_rate": 0.0005, "loss": 1.4832322597503662, "step": 12718 }, { "epoch": 1.4799860367698394, "grad_norm": 1.5883350372314453, "learning_rate": 0.0005, "loss": 1.919236421585083, "step": 12719 }, { "epoch": 1.4801023970211775, "grad_norm": 0.6225197911262512, "learning_rate": 0.0005, "loss": 1.2328197956085205, "step": 12720 }, { "epoch": 1.4802187572725156, "grad_norm": 1.1586617231369019, "learning_rate": 0.0005, "loss": 2.2939016819000244, "step": 12721 }, { "epoch": 1.480335117523854, "grad_norm": 1.453865647315979, "learning_rate": 0.0005, "loss": 1.6781225204467773, "step": 12722 }, { "epoch": 1.480451477775192, "grad_norm": 1.8443361520767212, "learning_rate": 0.0005, "loss": 2.392758369445801, "step": 12723 }, { "epoch": 1.4805678380265301, "grad_norm": 0.7191759943962097, "learning_rate": 0.0005, "loss": 1.6345674991607666, "step": 12724 }, { "epoch": 1.4806841982778682, "grad_norm": 1.6498143672943115, "learning_rate": 0.0005, "loss": 2.5765457153320312, "step": 12725 }, { "epoch": 1.4808005585292063, "grad_norm": 0.9552630186080933, "learning_rate": 0.0005, "loss": 1.8545236587524414, "step": 12726 }, { "epoch": 1.4809169187805447, "grad_norm": 0.8044872879981995, "learning_rate": 0.0005, "loss": 1.509950876235962, "step": 12727 }, { "epoch": 1.4810332790318828, "grad_norm": 0.9443849325180054, "learning_rate": 0.0005, "loss": 1.9333341121673584, "step": 12728 }, { "epoch": 1.4811496392832209, "grad_norm": 1.263087511062622, "learning_rate": 0.0005, "loss": 2.2696285247802734, "step": 12729 }, { "epoch": 1.481265999534559, "grad_norm": 0.902316689491272, "learning_rate": 0.0005, "loss": 1.857918620109558, "step": 12730 }, { "epoch": 1.481382359785897, "grad_norm": 0.859641969203949, "learning_rate": 0.0005, "loss": 1.3204326629638672, "step": 12731 }, { "epoch": 1.4814987200372354, "grad_norm": 0.9310330152511597, "learning_rate": 0.0005, "loss": 1.5045647621154785, "step": 12732 }, { "epoch": 1.4816150802885735, "grad_norm": 0.7654628753662109, "learning_rate": 0.0005, "loss": 1.1356661319732666, "step": 12733 }, { "epoch": 1.4817314405399116, "grad_norm": 1.1395245790481567, "learning_rate": 0.0005, "loss": 2.2217531204223633, "step": 12734 }, { "epoch": 1.4818478007912497, "grad_norm": 1.0900700092315674, "learning_rate": 0.0005, "loss": 2.7297303676605225, "step": 12735 }, { "epoch": 1.4819641610425878, "grad_norm": 0.8678656220436096, "learning_rate": 0.0005, "loss": 1.472778558731079, "step": 12736 }, { "epoch": 1.482080521293926, "grad_norm": 1.2206084728240967, "learning_rate": 0.0005, "loss": 2.2426514625549316, "step": 12737 }, { "epoch": 1.4821968815452642, "grad_norm": 0.9458897709846497, "learning_rate": 0.0005, "loss": 1.5066497325897217, "step": 12738 }, { "epoch": 1.4823132417966023, "grad_norm": 0.9943762421607971, "learning_rate": 0.0005, "loss": 2.15592098236084, "step": 12739 }, { "epoch": 1.4824296020479404, "grad_norm": 0.8609460592269897, "learning_rate": 0.0005, "loss": 2.1112375259399414, "step": 12740 }, { "epoch": 1.4825459622992785, "grad_norm": 0.8983215093612671, "learning_rate": 0.0005, "loss": 1.8538672924041748, "step": 12741 }, { "epoch": 1.4826623225506168, "grad_norm": 1.114200234413147, "learning_rate": 0.0005, "loss": 1.8859584331512451, "step": 12742 }, { "epoch": 1.482778682801955, "grad_norm": 1.850893497467041, "learning_rate": 0.0005, "loss": 1.9768593311309814, "step": 12743 }, { "epoch": 1.482895043053293, "grad_norm": 0.844947874546051, "learning_rate": 0.0005, "loss": 2.2605013847351074, "step": 12744 }, { "epoch": 1.483011403304631, "grad_norm": 1.0468968152999878, "learning_rate": 0.0005, "loss": 2.3921687602996826, "step": 12745 }, { "epoch": 1.4831277635559692, "grad_norm": 1.322537899017334, "learning_rate": 0.0005, "loss": 2.0603208541870117, "step": 12746 }, { "epoch": 1.4832441238073075, "grad_norm": 1.4175999164581299, "learning_rate": 0.0005, "loss": 2.6828203201293945, "step": 12747 }, { "epoch": 1.4833604840586456, "grad_norm": 0.7385936975479126, "learning_rate": 0.0005, "loss": 1.1344895362854004, "step": 12748 }, { "epoch": 1.4834768443099837, "grad_norm": 1.3076797723770142, "learning_rate": 0.0005, "loss": 2.3991808891296387, "step": 12749 }, { "epoch": 1.4835932045613218, "grad_norm": 1.097069501876831, "learning_rate": 0.0005, "loss": 1.9272539615631104, "step": 12750 }, { "epoch": 1.48370956481266, "grad_norm": 0.5908493399620056, "learning_rate": 0.0005, "loss": 1.458785891532898, "step": 12751 }, { "epoch": 1.4838259250639982, "grad_norm": 0.7139748334884644, "learning_rate": 0.0005, "loss": 1.5958690643310547, "step": 12752 }, { "epoch": 1.4839422853153363, "grad_norm": 0.8191168308258057, "learning_rate": 0.0005, "loss": 1.4059020280838013, "step": 12753 }, { "epoch": 1.4840586455666744, "grad_norm": 1.3284173011779785, "learning_rate": 0.0005, "loss": 1.4244383573532104, "step": 12754 }, { "epoch": 1.4841750058180125, "grad_norm": 1.3170422315597534, "learning_rate": 0.0005, "loss": 2.723722219467163, "step": 12755 }, { "epoch": 1.4842913660693506, "grad_norm": 0.4707847535610199, "learning_rate": 0.0005, "loss": 1.1390917301177979, "step": 12756 }, { "epoch": 1.484407726320689, "grad_norm": 1.0524044036865234, "learning_rate": 0.0005, "loss": 1.3645081520080566, "step": 12757 }, { "epoch": 1.484524086572027, "grad_norm": 0.8619306087493896, "learning_rate": 0.0005, "loss": 1.1097912788391113, "step": 12758 }, { "epoch": 1.4846404468233652, "grad_norm": 1.0904979705810547, "learning_rate": 0.0005, "loss": 1.8940722942352295, "step": 12759 }, { "epoch": 1.4847568070747033, "grad_norm": 1.144209623336792, "learning_rate": 0.0005, "loss": 1.7185826301574707, "step": 12760 }, { "epoch": 1.4848731673260414, "grad_norm": 0.9441536664962769, "learning_rate": 0.0005, "loss": 1.4340765476226807, "step": 12761 }, { "epoch": 1.4849895275773797, "grad_norm": 1.0660713911056519, "learning_rate": 0.0005, "loss": 2.010897636413574, "step": 12762 }, { "epoch": 1.4851058878287178, "grad_norm": 0.751002311706543, "learning_rate": 0.0005, "loss": 1.6239147186279297, "step": 12763 }, { "epoch": 1.4852222480800559, "grad_norm": 0.7924482226371765, "learning_rate": 0.0005, "loss": 1.617734670639038, "step": 12764 }, { "epoch": 1.485338608331394, "grad_norm": 0.9454014897346497, "learning_rate": 0.0005, "loss": 1.926232099533081, "step": 12765 }, { "epoch": 1.485454968582732, "grad_norm": 1.232241153717041, "learning_rate": 0.0005, "loss": 2.0040903091430664, "step": 12766 }, { "epoch": 1.4855713288340704, "grad_norm": 0.7493155598640442, "learning_rate": 0.0005, "loss": 1.983202338218689, "step": 12767 }, { "epoch": 1.4856876890854085, "grad_norm": 1.352948784828186, "learning_rate": 0.0005, "loss": 1.6822757720947266, "step": 12768 }, { "epoch": 1.4858040493367466, "grad_norm": 1.447739839553833, "learning_rate": 0.0005, "loss": 2.1359236240386963, "step": 12769 }, { "epoch": 1.4859204095880847, "grad_norm": 1.179587721824646, "learning_rate": 0.0005, "loss": 2.348531723022461, "step": 12770 }, { "epoch": 1.4860367698394228, "grad_norm": 0.664903461933136, "learning_rate": 0.0005, "loss": 1.310742974281311, "step": 12771 }, { "epoch": 1.486153130090761, "grad_norm": 0.7918275594711304, "learning_rate": 0.0005, "loss": 1.6550531387329102, "step": 12772 }, { "epoch": 1.4862694903420992, "grad_norm": 0.7542226910591125, "learning_rate": 0.0005, "loss": 1.6215651035308838, "step": 12773 }, { "epoch": 1.4863858505934373, "grad_norm": 2.573298215866089, "learning_rate": 0.0005, "loss": 2.352135181427002, "step": 12774 }, { "epoch": 1.4865022108447754, "grad_norm": 1.3382982015609741, "learning_rate": 0.0005, "loss": 1.3683863878250122, "step": 12775 }, { "epoch": 1.4866185710961135, "grad_norm": 1.718429446220398, "learning_rate": 0.0005, "loss": 2.1293439865112305, "step": 12776 }, { "epoch": 1.4867349313474518, "grad_norm": 1.0170578956604004, "learning_rate": 0.0005, "loss": 2.115628957748413, "step": 12777 }, { "epoch": 1.48685129159879, "grad_norm": 0.5132238268852234, "learning_rate": 0.0005, "loss": 1.3064358234405518, "step": 12778 }, { "epoch": 1.486967651850128, "grad_norm": 0.7401127815246582, "learning_rate": 0.0005, "loss": 2.1169285774230957, "step": 12779 }, { "epoch": 1.4870840121014661, "grad_norm": 0.8173818588256836, "learning_rate": 0.0005, "loss": 1.6192091703414917, "step": 12780 }, { "epoch": 1.4872003723528042, "grad_norm": 1.4148263931274414, "learning_rate": 0.0005, "loss": 1.6120129823684692, "step": 12781 }, { "epoch": 1.4873167326041425, "grad_norm": 0.970488965511322, "learning_rate": 0.0005, "loss": 2.1259031295776367, "step": 12782 }, { "epoch": 1.4874330928554806, "grad_norm": 0.8504739999771118, "learning_rate": 0.0005, "loss": 1.755554437637329, "step": 12783 }, { "epoch": 1.4875494531068187, "grad_norm": 0.7170439958572388, "learning_rate": 0.0005, "loss": 1.5987675189971924, "step": 12784 }, { "epoch": 1.4876658133581568, "grad_norm": 1.2579708099365234, "learning_rate": 0.0005, "loss": 2.0454814434051514, "step": 12785 }, { "epoch": 1.487782173609495, "grad_norm": 1.2845954895019531, "learning_rate": 0.0005, "loss": 1.6173937320709229, "step": 12786 }, { "epoch": 1.4878985338608333, "grad_norm": 1.259865164756775, "learning_rate": 0.0005, "loss": 2.141357660293579, "step": 12787 }, { "epoch": 1.4880148941121714, "grad_norm": 1.1948250532150269, "learning_rate": 0.0005, "loss": 2.207895040512085, "step": 12788 }, { "epoch": 1.4881312543635095, "grad_norm": 0.8902191519737244, "learning_rate": 0.0005, "loss": 2.0574755668640137, "step": 12789 }, { "epoch": 1.4882476146148476, "grad_norm": 1.123421549797058, "learning_rate": 0.0005, "loss": 1.9331727027893066, "step": 12790 }, { "epoch": 1.4883639748661857, "grad_norm": 1.0916163921356201, "learning_rate": 0.0005, "loss": 1.548256278038025, "step": 12791 }, { "epoch": 1.488480335117524, "grad_norm": 1.698090672492981, "learning_rate": 0.0005, "loss": 2.2958736419677734, "step": 12792 }, { "epoch": 1.488596695368862, "grad_norm": 1.3574234247207642, "learning_rate": 0.0005, "loss": 1.6092113256454468, "step": 12793 }, { "epoch": 1.4887130556202002, "grad_norm": 1.612139105796814, "learning_rate": 0.0005, "loss": 1.881466031074524, "step": 12794 }, { "epoch": 1.4888294158715383, "grad_norm": 1.4832135438919067, "learning_rate": 0.0005, "loss": 1.8819599151611328, "step": 12795 }, { "epoch": 1.4889457761228764, "grad_norm": 1.3512476682662964, "learning_rate": 0.0005, "loss": 2.443495035171509, "step": 12796 }, { "epoch": 1.4890621363742147, "grad_norm": 0.8472211360931396, "learning_rate": 0.0005, "loss": 1.704561710357666, "step": 12797 }, { "epoch": 1.4891784966255528, "grad_norm": 1.3834922313690186, "learning_rate": 0.0005, "loss": 2.255598545074463, "step": 12798 }, { "epoch": 1.489294856876891, "grad_norm": 0.8393954038619995, "learning_rate": 0.0005, "loss": 1.6202497482299805, "step": 12799 }, { "epoch": 1.489411217128229, "grad_norm": 1.178252100944519, "learning_rate": 0.0005, "loss": 1.6187639236450195, "step": 12800 }, { "epoch": 1.489527577379567, "grad_norm": 0.9308677315711975, "learning_rate": 0.0005, "loss": 1.8939341306686401, "step": 12801 }, { "epoch": 1.4896439376309054, "grad_norm": 1.0641964673995972, "learning_rate": 0.0005, "loss": 1.5846261978149414, "step": 12802 }, { "epoch": 1.4897602978822433, "grad_norm": 1.0409188270568848, "learning_rate": 0.0005, "loss": 1.7695255279541016, "step": 12803 }, { "epoch": 1.4898766581335816, "grad_norm": 1.0995168685913086, "learning_rate": 0.0005, "loss": 1.5830967426300049, "step": 12804 }, { "epoch": 1.4899930183849197, "grad_norm": 1.2793563604354858, "learning_rate": 0.0005, "loss": 2.400514602661133, "step": 12805 }, { "epoch": 1.4901093786362578, "grad_norm": 1.5435240268707275, "learning_rate": 0.0005, "loss": 2.155407428741455, "step": 12806 }, { "epoch": 1.4902257388875961, "grad_norm": 1.0663602352142334, "learning_rate": 0.0005, "loss": 1.6419316530227661, "step": 12807 }, { "epoch": 1.490342099138934, "grad_norm": 1.5293489694595337, "learning_rate": 0.0005, "loss": 1.8595244884490967, "step": 12808 }, { "epoch": 1.4904584593902723, "grad_norm": 1.269901990890503, "learning_rate": 0.0005, "loss": 2.0964906215667725, "step": 12809 }, { "epoch": 1.4905748196416104, "grad_norm": 1.2171834707260132, "learning_rate": 0.0005, "loss": 2.230886459350586, "step": 12810 }, { "epoch": 1.4906911798929485, "grad_norm": 1.3304650783538818, "learning_rate": 0.0005, "loss": 2.2939891815185547, "step": 12811 }, { "epoch": 1.4908075401442868, "grad_norm": 1.3860656023025513, "learning_rate": 0.0005, "loss": 1.825670838356018, "step": 12812 }, { "epoch": 1.4909239003956247, "grad_norm": 1.4076493978500366, "learning_rate": 0.0005, "loss": 1.4894704818725586, "step": 12813 }, { "epoch": 1.491040260646963, "grad_norm": 1.166322112083435, "learning_rate": 0.0005, "loss": 2.0611915588378906, "step": 12814 }, { "epoch": 1.4911566208983011, "grad_norm": 1.096909999847412, "learning_rate": 0.0005, "loss": 1.8179388046264648, "step": 12815 }, { "epoch": 1.4912729811496392, "grad_norm": 1.3512952327728271, "learning_rate": 0.0005, "loss": 2.2376320362091064, "step": 12816 }, { "epoch": 1.4913893414009776, "grad_norm": 1.2315903902053833, "learning_rate": 0.0005, "loss": 1.6675306558609009, "step": 12817 }, { "epoch": 1.4915057016523154, "grad_norm": 0.9167574048042297, "learning_rate": 0.0005, "loss": 1.5352351665496826, "step": 12818 }, { "epoch": 1.4916220619036538, "grad_norm": 0.9701797962188721, "learning_rate": 0.0005, "loss": 1.6750471591949463, "step": 12819 }, { "epoch": 1.4917384221549919, "grad_norm": 1.1420354843139648, "learning_rate": 0.0005, "loss": 2.1684823036193848, "step": 12820 }, { "epoch": 1.49185478240633, "grad_norm": 1.1313432455062866, "learning_rate": 0.0005, "loss": 1.8829983472824097, "step": 12821 }, { "epoch": 1.491971142657668, "grad_norm": 0.7944203615188599, "learning_rate": 0.0005, "loss": 1.4473440647125244, "step": 12822 }, { "epoch": 1.4920875029090062, "grad_norm": 0.7967546582221985, "learning_rate": 0.0005, "loss": 1.4152181148529053, "step": 12823 }, { "epoch": 1.4922038631603445, "grad_norm": 0.7778344750404358, "learning_rate": 0.0005, "loss": 1.4305920600891113, "step": 12824 }, { "epoch": 1.4923202234116826, "grad_norm": 1.0088919401168823, "learning_rate": 0.0005, "loss": 2.0039772987365723, "step": 12825 }, { "epoch": 1.4924365836630207, "grad_norm": 0.9475328922271729, "learning_rate": 0.0005, "loss": 1.518534541130066, "step": 12826 }, { "epoch": 1.4925529439143588, "grad_norm": 0.7753596901893616, "learning_rate": 0.0005, "loss": 1.4122984409332275, "step": 12827 }, { "epoch": 1.4926693041656969, "grad_norm": 1.2627284526824951, "learning_rate": 0.0005, "loss": 2.3052871227264404, "step": 12828 }, { "epoch": 1.4927856644170352, "grad_norm": 0.7581610679626465, "learning_rate": 0.0005, "loss": 1.5799601078033447, "step": 12829 }, { "epoch": 1.4929020246683733, "grad_norm": 0.8381916284561157, "learning_rate": 0.0005, "loss": 1.2758865356445312, "step": 12830 }, { "epoch": 1.4930183849197114, "grad_norm": 1.014390230178833, "learning_rate": 0.0005, "loss": 1.696779727935791, "step": 12831 }, { "epoch": 1.4931347451710495, "grad_norm": 1.2820992469787598, "learning_rate": 0.0005, "loss": 2.406456708908081, "step": 12832 }, { "epoch": 1.4932511054223876, "grad_norm": 1.2236015796661377, "learning_rate": 0.0005, "loss": 2.0239696502685547, "step": 12833 }, { "epoch": 1.493367465673726, "grad_norm": 0.7942686080932617, "learning_rate": 0.0005, "loss": 1.1789109706878662, "step": 12834 }, { "epoch": 1.493483825925064, "grad_norm": 1.1726797819137573, "learning_rate": 0.0005, "loss": 1.6937918663024902, "step": 12835 }, { "epoch": 1.4936001861764021, "grad_norm": 1.059348225593567, "learning_rate": 0.0005, "loss": 2.2286486625671387, "step": 12836 }, { "epoch": 1.4937165464277402, "grad_norm": 2.7495758533477783, "learning_rate": 0.0005, "loss": 2.4615285396575928, "step": 12837 }, { "epoch": 1.4938329066790783, "grad_norm": 1.0417581796646118, "learning_rate": 0.0005, "loss": 1.883426308631897, "step": 12838 }, { "epoch": 1.4939492669304166, "grad_norm": 1.6610561609268188, "learning_rate": 0.0005, "loss": 1.9980242252349854, "step": 12839 }, { "epoch": 1.4940656271817547, "grad_norm": 0.9365177154541016, "learning_rate": 0.0005, "loss": 2.484961748123169, "step": 12840 }, { "epoch": 1.4941819874330928, "grad_norm": 1.4458410739898682, "learning_rate": 0.0005, "loss": 1.636946678161621, "step": 12841 }, { "epoch": 1.494298347684431, "grad_norm": 1.4155290126800537, "learning_rate": 0.0005, "loss": 1.872786521911621, "step": 12842 }, { "epoch": 1.494414707935769, "grad_norm": 0.7784556746482849, "learning_rate": 0.0005, "loss": 1.7922990322113037, "step": 12843 }, { "epoch": 1.4945310681871073, "grad_norm": 0.778678834438324, "learning_rate": 0.0005, "loss": 1.3280882835388184, "step": 12844 }, { "epoch": 1.4946474284384454, "grad_norm": 1.000952124595642, "learning_rate": 0.0005, "loss": 1.3988744020462036, "step": 12845 }, { "epoch": 1.4947637886897835, "grad_norm": 1.3764253854751587, "learning_rate": 0.0005, "loss": 1.785738468170166, "step": 12846 }, { "epoch": 1.4948801489411216, "grad_norm": 1.0947682857513428, "learning_rate": 0.0005, "loss": 2.3032338619232178, "step": 12847 }, { "epoch": 1.4949965091924597, "grad_norm": 1.6964761018753052, "learning_rate": 0.0005, "loss": 2.636434555053711, "step": 12848 }, { "epoch": 1.495112869443798, "grad_norm": 0.8995680809020996, "learning_rate": 0.0005, "loss": 1.5953178405761719, "step": 12849 }, { "epoch": 1.4952292296951362, "grad_norm": 0.9850319027900696, "learning_rate": 0.0005, "loss": 2.1286816596984863, "step": 12850 }, { "epoch": 1.4953455899464743, "grad_norm": 0.7224987149238586, "learning_rate": 0.0005, "loss": 1.1353092193603516, "step": 12851 }, { "epoch": 1.4954619501978124, "grad_norm": 1.0484082698822021, "learning_rate": 0.0005, "loss": 2.019076347351074, "step": 12852 }, { "epoch": 1.4955783104491505, "grad_norm": 0.9754845499992371, "learning_rate": 0.0005, "loss": 1.959235668182373, "step": 12853 }, { "epoch": 1.4956946707004888, "grad_norm": 1.536221981048584, "learning_rate": 0.0005, "loss": 2.0629167556762695, "step": 12854 }, { "epoch": 1.4958110309518269, "grad_norm": 0.8725597262382507, "learning_rate": 0.0005, "loss": 1.673326849937439, "step": 12855 }, { "epoch": 1.495927391203165, "grad_norm": 1.2446600198745728, "learning_rate": 0.0005, "loss": 1.8019802570343018, "step": 12856 }, { "epoch": 1.496043751454503, "grad_norm": 1.3261922597885132, "learning_rate": 0.0005, "loss": 2.538674831390381, "step": 12857 }, { "epoch": 1.4961601117058412, "grad_norm": 1.9899507761001587, "learning_rate": 0.0005, "loss": 2.5527281761169434, "step": 12858 }, { "epoch": 1.4962764719571795, "grad_norm": 0.9472229480743408, "learning_rate": 0.0005, "loss": 2.092588424682617, "step": 12859 }, { "epoch": 1.4963928322085176, "grad_norm": 1.3940949440002441, "learning_rate": 0.0005, "loss": 1.6799030303955078, "step": 12860 }, { "epoch": 1.4965091924598557, "grad_norm": 2.0425753593444824, "learning_rate": 0.0005, "loss": 1.5696444511413574, "step": 12861 }, { "epoch": 1.4966255527111938, "grad_norm": 1.105877161026001, "learning_rate": 0.0005, "loss": 1.9196102619171143, "step": 12862 }, { "epoch": 1.496741912962532, "grad_norm": 1.1024121046066284, "learning_rate": 0.0005, "loss": 1.6006531715393066, "step": 12863 }, { "epoch": 1.4968582732138702, "grad_norm": 1.1014803647994995, "learning_rate": 0.0005, "loss": 2.2792012691497803, "step": 12864 }, { "epoch": 1.4969746334652083, "grad_norm": 1.1007652282714844, "learning_rate": 0.0005, "loss": 2.26475191116333, "step": 12865 }, { "epoch": 1.4970909937165464, "grad_norm": 1.0689414739608765, "learning_rate": 0.0005, "loss": 1.7800360918045044, "step": 12866 }, { "epoch": 1.4972073539678845, "grad_norm": 1.2534728050231934, "learning_rate": 0.0005, "loss": 2.7848498821258545, "step": 12867 }, { "epoch": 1.4973237142192226, "grad_norm": 1.3872877359390259, "learning_rate": 0.0005, "loss": 2.1959383487701416, "step": 12868 }, { "epoch": 1.497440074470561, "grad_norm": 1.0405532121658325, "learning_rate": 0.0005, "loss": 2.031409740447998, "step": 12869 }, { "epoch": 1.497556434721899, "grad_norm": 1.315024495124817, "learning_rate": 0.0005, "loss": 1.4110268354415894, "step": 12870 }, { "epoch": 1.4976727949732371, "grad_norm": 0.6134066581726074, "learning_rate": 0.0005, "loss": 1.1366164684295654, "step": 12871 }, { "epoch": 1.4977891552245752, "grad_norm": 1.759540319442749, "learning_rate": 0.0005, "loss": 2.4279253482818604, "step": 12872 }, { "epoch": 1.4979055154759133, "grad_norm": 1.05043363571167, "learning_rate": 0.0005, "loss": 1.7911162376403809, "step": 12873 }, { "epoch": 1.4980218757272517, "grad_norm": 1.9113726615905762, "learning_rate": 0.0005, "loss": 1.807624101638794, "step": 12874 }, { "epoch": 1.4981382359785897, "grad_norm": 0.9425702691078186, "learning_rate": 0.0005, "loss": 1.5084412097930908, "step": 12875 }, { "epoch": 1.4982545962299278, "grad_norm": 0.9875898957252502, "learning_rate": 0.0005, "loss": 1.6019338369369507, "step": 12876 }, { "epoch": 1.498370956481266, "grad_norm": 1.3942978382110596, "learning_rate": 0.0005, "loss": 2.344714879989624, "step": 12877 }, { "epoch": 1.498487316732604, "grad_norm": 1.3472450971603394, "learning_rate": 0.0005, "loss": 1.9780739545822144, "step": 12878 }, { "epoch": 1.4986036769839424, "grad_norm": 0.9233900904655457, "learning_rate": 0.0005, "loss": 1.5320364236831665, "step": 12879 }, { "epoch": 1.4987200372352805, "grad_norm": 1.1178346872329712, "learning_rate": 0.0005, "loss": 2.0004377365112305, "step": 12880 }, { "epoch": 1.4988363974866186, "grad_norm": 1.2510117292404175, "learning_rate": 0.0005, "loss": 2.6579270362854004, "step": 12881 }, { "epoch": 1.4989527577379567, "grad_norm": 0.9708411693572998, "learning_rate": 0.0005, "loss": 1.7956454753875732, "step": 12882 }, { "epoch": 1.4990691179892948, "grad_norm": 1.7860538959503174, "learning_rate": 0.0005, "loss": 1.8993396759033203, "step": 12883 }, { "epoch": 1.499185478240633, "grad_norm": 1.0739866495132446, "learning_rate": 0.0005, "loss": 2.3851351737976074, "step": 12884 }, { "epoch": 1.4993018384919712, "grad_norm": 3.1430275440216064, "learning_rate": 0.0005, "loss": 1.411879539489746, "step": 12885 }, { "epoch": 1.4994181987433093, "grad_norm": 3.342329740524292, "learning_rate": 0.0005, "loss": 1.90804123878479, "step": 12886 }, { "epoch": 1.4995345589946474, "grad_norm": 0.7628262042999268, "learning_rate": 0.0005, "loss": 1.4869261980056763, "step": 12887 }, { "epoch": 1.4996509192459855, "grad_norm": 1.8298128843307495, "learning_rate": 0.0005, "loss": 1.7980399131774902, "step": 12888 }, { "epoch": 1.4997672794973238, "grad_norm": 1.6800628900527954, "learning_rate": 0.0005, "loss": 2.233205556869507, "step": 12889 }, { "epoch": 1.499883639748662, "grad_norm": 0.6642251014709473, "learning_rate": 0.0005, "loss": 1.4216738939285278, "step": 12890 }, { "epoch": 1.5, "grad_norm": 0.6716475486755371, "learning_rate": 0.0005, "loss": 1.742722511291504, "step": 12891 }, { "epoch": 1.500116360251338, "grad_norm": 0.6246010661125183, "learning_rate": 0.0005, "loss": 1.815373182296753, "step": 12892 }, { "epoch": 1.5002327205026762, "grad_norm": 1.4504927396774292, "learning_rate": 0.0005, "loss": 2.2436208724975586, "step": 12893 }, { "epoch": 1.5003490807540145, "grad_norm": 0.8827069401741028, "learning_rate": 0.0005, "loss": 1.7953916788101196, "step": 12894 }, { "epoch": 1.5004654410053526, "grad_norm": 1.1005868911743164, "learning_rate": 0.0005, "loss": 1.8436816930770874, "step": 12895 }, { "epoch": 1.5005818012566907, "grad_norm": 1.4865069389343262, "learning_rate": 0.0005, "loss": 2.1920647621154785, "step": 12896 }, { "epoch": 1.5006981615080288, "grad_norm": 1.046993613243103, "learning_rate": 0.0005, "loss": 2.2111573219299316, "step": 12897 }, { "epoch": 1.500814521759367, "grad_norm": 1.1062934398651123, "learning_rate": 0.0005, "loss": 2.248584032058716, "step": 12898 }, { "epoch": 1.5009308820107052, "grad_norm": 0.8091353178024292, "learning_rate": 0.0005, "loss": 1.3410944938659668, "step": 12899 }, { "epoch": 1.5010472422620433, "grad_norm": 1.3645763397216797, "learning_rate": 0.0005, "loss": 2.4530768394470215, "step": 12900 }, { "epoch": 1.5011636025133814, "grad_norm": 1.7230266332626343, "learning_rate": 0.0005, "loss": 1.86527681350708, "step": 12901 }, { "epoch": 1.5012799627647195, "grad_norm": 1.3326696157455444, "learning_rate": 0.0005, "loss": 2.148743152618408, "step": 12902 }, { "epoch": 1.5013963230160576, "grad_norm": 1.0628491640090942, "learning_rate": 0.0005, "loss": 2.2427351474761963, "step": 12903 }, { "epoch": 1.501512683267396, "grad_norm": 0.6977850794792175, "learning_rate": 0.0005, "loss": 1.6091740131378174, "step": 12904 }, { "epoch": 1.501629043518734, "grad_norm": 1.4979636669158936, "learning_rate": 0.0005, "loss": 1.3399382829666138, "step": 12905 }, { "epoch": 1.5017454037700722, "grad_norm": 1.7036923170089722, "learning_rate": 0.0005, "loss": 2.397406816482544, "step": 12906 }, { "epoch": 1.5018617640214103, "grad_norm": 1.7580536603927612, "learning_rate": 0.0005, "loss": 2.1673927307128906, "step": 12907 }, { "epoch": 1.5019781242727483, "grad_norm": 1.1092737913131714, "learning_rate": 0.0005, "loss": 1.5515259504318237, "step": 12908 }, { "epoch": 1.5020944845240867, "grad_norm": 1.0152407884597778, "learning_rate": 0.0005, "loss": 1.7504022121429443, "step": 12909 }, { "epoch": 1.5022108447754248, "grad_norm": 0.7931157946586609, "learning_rate": 0.0005, "loss": 2.0192344188690186, "step": 12910 }, { "epoch": 1.5023272050267629, "grad_norm": 1.7201945781707764, "learning_rate": 0.0005, "loss": 2.746335029602051, "step": 12911 }, { "epoch": 1.502443565278101, "grad_norm": 1.1590304374694824, "learning_rate": 0.0005, "loss": 2.2241439819335938, "step": 12912 }, { "epoch": 1.502559925529439, "grad_norm": 0.7249947190284729, "learning_rate": 0.0005, "loss": 2.0549333095550537, "step": 12913 }, { "epoch": 1.5026762857807774, "grad_norm": 1.2464267015457153, "learning_rate": 0.0005, "loss": 2.8762378692626953, "step": 12914 }, { "epoch": 1.5027926460321153, "grad_norm": 0.7748426795005798, "learning_rate": 0.0005, "loss": 1.4320992231369019, "step": 12915 }, { "epoch": 1.5029090062834536, "grad_norm": 1.0185216665267944, "learning_rate": 0.0005, "loss": 2.0388898849487305, "step": 12916 }, { "epoch": 1.5030253665347917, "grad_norm": 1.673548936843872, "learning_rate": 0.0005, "loss": 2.2714791297912598, "step": 12917 }, { "epoch": 1.5031417267861298, "grad_norm": 1.3453800678253174, "learning_rate": 0.0005, "loss": 2.0352253913879395, "step": 12918 }, { "epoch": 1.503258087037468, "grad_norm": 0.8736308813095093, "learning_rate": 0.0005, "loss": 2.201931953430176, "step": 12919 }, { "epoch": 1.503374447288806, "grad_norm": 1.1317296028137207, "learning_rate": 0.0005, "loss": 1.7835392951965332, "step": 12920 }, { "epoch": 1.5034908075401443, "grad_norm": 1.7463874816894531, "learning_rate": 0.0005, "loss": 2.2648000717163086, "step": 12921 }, { "epoch": 1.5036071677914824, "grad_norm": 1.3727121353149414, "learning_rate": 0.0005, "loss": 1.5501090288162231, "step": 12922 }, { "epoch": 1.5037235280428205, "grad_norm": 1.1342123746871948, "learning_rate": 0.0005, "loss": 2.156676769256592, "step": 12923 }, { "epoch": 1.5038398882941588, "grad_norm": 1.1954751014709473, "learning_rate": 0.0005, "loss": 1.9124066829681396, "step": 12924 }, { "epoch": 1.5039562485454967, "grad_norm": 0.830142617225647, "learning_rate": 0.0005, "loss": 1.5563770532608032, "step": 12925 }, { "epoch": 1.504072608796835, "grad_norm": 1.110835075378418, "learning_rate": 0.0005, "loss": 2.066359758377075, "step": 12926 }, { "epoch": 1.5041889690481731, "grad_norm": 1.0486526489257812, "learning_rate": 0.0005, "loss": 1.6862062215805054, "step": 12927 }, { "epoch": 1.5043053292995112, "grad_norm": 1.1958527565002441, "learning_rate": 0.0005, "loss": 1.5163183212280273, "step": 12928 }, { "epoch": 1.5044216895508495, "grad_norm": 1.4914915561676025, "learning_rate": 0.0005, "loss": 2.436412811279297, "step": 12929 }, { "epoch": 1.5045380498021874, "grad_norm": 1.10056734085083, "learning_rate": 0.0005, "loss": 2.1400866508483887, "step": 12930 }, { "epoch": 1.5046544100535257, "grad_norm": 1.2133290767669678, "learning_rate": 0.0005, "loss": 2.696962594985962, "step": 12931 }, { "epoch": 1.5047707703048638, "grad_norm": 0.560185968875885, "learning_rate": 0.0005, "loss": 1.1180355548858643, "step": 12932 }, { "epoch": 1.504887130556202, "grad_norm": 1.2051723003387451, "learning_rate": 0.0005, "loss": 2.4631619453430176, "step": 12933 }, { "epoch": 1.5050034908075403, "grad_norm": 0.8277959823608398, "learning_rate": 0.0005, "loss": 1.221150279045105, "step": 12934 }, { "epoch": 1.5051198510588781, "grad_norm": 1.17118239402771, "learning_rate": 0.0005, "loss": 2.257901191711426, "step": 12935 }, { "epoch": 1.5052362113102165, "grad_norm": 1.1061376333236694, "learning_rate": 0.0005, "loss": 1.9028589725494385, "step": 12936 }, { "epoch": 1.5053525715615546, "grad_norm": 1.0934388637542725, "learning_rate": 0.0005, "loss": 1.9440841674804688, "step": 12937 }, { "epoch": 1.5054689318128927, "grad_norm": 0.8479641079902649, "learning_rate": 0.0005, "loss": 1.6070486307144165, "step": 12938 }, { "epoch": 1.505585292064231, "grad_norm": 0.9313420057296753, "learning_rate": 0.0005, "loss": 1.3660008907318115, "step": 12939 }, { "epoch": 1.5057016523155689, "grad_norm": 0.8962718844413757, "learning_rate": 0.0005, "loss": 1.9282528162002563, "step": 12940 }, { "epoch": 1.5058180125669072, "grad_norm": 1.213234782218933, "learning_rate": 0.0005, "loss": 1.4668617248535156, "step": 12941 }, { "epoch": 1.5059343728182453, "grad_norm": 0.7038624882698059, "learning_rate": 0.0005, "loss": 1.4817901849746704, "step": 12942 }, { "epoch": 1.5060507330695834, "grad_norm": 1.2797664403915405, "learning_rate": 0.0005, "loss": 2.3642191886901855, "step": 12943 }, { "epoch": 1.5061670933209217, "grad_norm": 0.8087707757949829, "learning_rate": 0.0005, "loss": 1.342618465423584, "step": 12944 }, { "epoch": 1.5062834535722596, "grad_norm": 1.4055125713348389, "learning_rate": 0.0005, "loss": 2.759225845336914, "step": 12945 }, { "epoch": 1.5063998138235979, "grad_norm": 0.8875362873077393, "learning_rate": 0.0005, "loss": 1.4570531845092773, "step": 12946 }, { "epoch": 1.506516174074936, "grad_norm": 1.0063186883926392, "learning_rate": 0.0005, "loss": 2.4837613105773926, "step": 12947 }, { "epoch": 1.506632534326274, "grad_norm": 1.192739725112915, "learning_rate": 0.0005, "loss": 1.9038352966308594, "step": 12948 }, { "epoch": 1.5067488945776124, "grad_norm": 0.5714936852455139, "learning_rate": 0.0005, "loss": 1.376719355583191, "step": 12949 }, { "epoch": 1.5068652548289503, "grad_norm": 1.973392128944397, "learning_rate": 0.0005, "loss": 2.5144848823547363, "step": 12950 }, { "epoch": 1.5069816150802886, "grad_norm": 0.9593464732170105, "learning_rate": 0.0005, "loss": 2.067595958709717, "step": 12951 }, { "epoch": 1.5070979753316267, "grad_norm": 0.8358908891677856, "learning_rate": 0.0005, "loss": 1.8906974792480469, "step": 12952 }, { "epoch": 1.5072143355829648, "grad_norm": 1.3374929428100586, "learning_rate": 0.0005, "loss": 1.8871266841888428, "step": 12953 }, { "epoch": 1.5073306958343031, "grad_norm": 1.0024405717849731, "learning_rate": 0.0005, "loss": 1.9011070728302002, "step": 12954 }, { "epoch": 1.507447056085641, "grad_norm": 2.000203847885132, "learning_rate": 0.0005, "loss": 1.8391969203948975, "step": 12955 }, { "epoch": 1.5075634163369793, "grad_norm": 1.2641832828521729, "learning_rate": 0.0005, "loss": 2.4064993858337402, "step": 12956 }, { "epoch": 1.5076797765883174, "grad_norm": 1.3633992671966553, "learning_rate": 0.0005, "loss": 1.3892122507095337, "step": 12957 }, { "epoch": 1.5077961368396555, "grad_norm": 1.2914267778396606, "learning_rate": 0.0005, "loss": 2.2309930324554443, "step": 12958 }, { "epoch": 1.5079124970909938, "grad_norm": 1.0712109804153442, "learning_rate": 0.0005, "loss": 1.873348593711853, "step": 12959 }, { "epoch": 1.5080288573423317, "grad_norm": 1.1584703922271729, "learning_rate": 0.0005, "loss": 2.348281145095825, "step": 12960 }, { "epoch": 1.50814521759367, "grad_norm": 1.0620931386947632, "learning_rate": 0.0005, "loss": 2.2062950134277344, "step": 12961 }, { "epoch": 1.5082615778450081, "grad_norm": 0.634336531162262, "learning_rate": 0.0005, "loss": 1.560336709022522, "step": 12962 }, { "epoch": 1.5083779380963462, "grad_norm": 1.8578311204910278, "learning_rate": 0.0005, "loss": 1.6755406856536865, "step": 12963 }, { "epoch": 1.5084942983476846, "grad_norm": 0.7908370494842529, "learning_rate": 0.0005, "loss": 1.774591326713562, "step": 12964 }, { "epoch": 1.5086106585990224, "grad_norm": 0.9824550747871399, "learning_rate": 0.0005, "loss": 1.9416626691818237, "step": 12965 }, { "epoch": 1.5087270188503608, "grad_norm": 1.3044084310531616, "learning_rate": 0.0005, "loss": 1.7556431293487549, "step": 12966 }, { "epoch": 1.5088433791016989, "grad_norm": 1.357508659362793, "learning_rate": 0.0005, "loss": 1.9336168766021729, "step": 12967 }, { "epoch": 1.508959739353037, "grad_norm": 1.2953248023986816, "learning_rate": 0.0005, "loss": 1.9662525653839111, "step": 12968 }, { "epoch": 1.5090760996043753, "grad_norm": 1.1700665950775146, "learning_rate": 0.0005, "loss": 2.1998512744903564, "step": 12969 }, { "epoch": 1.5091924598557132, "grad_norm": 2.0585851669311523, "learning_rate": 0.0005, "loss": 1.8277385234832764, "step": 12970 }, { "epoch": 1.5093088201070515, "grad_norm": 0.8556792736053467, "learning_rate": 0.0005, "loss": 1.545178771018982, "step": 12971 }, { "epoch": 1.5094251803583896, "grad_norm": 0.8271819353103638, "learning_rate": 0.0005, "loss": 1.6035513877868652, "step": 12972 }, { "epoch": 1.5095415406097277, "grad_norm": 1.0078343152999878, "learning_rate": 0.0005, "loss": 1.715747594833374, "step": 12973 }, { "epoch": 1.509657900861066, "grad_norm": 1.986899733543396, "learning_rate": 0.0005, "loss": 2.0293848514556885, "step": 12974 }, { "epoch": 1.5097742611124039, "grad_norm": 1.0316656827926636, "learning_rate": 0.0005, "loss": 1.4550058841705322, "step": 12975 }, { "epoch": 1.5098906213637422, "grad_norm": 1.4163705110549927, "learning_rate": 0.0005, "loss": 2.1438283920288086, "step": 12976 }, { "epoch": 1.5100069816150803, "grad_norm": 1.5853582620620728, "learning_rate": 0.0005, "loss": 2.304034471511841, "step": 12977 }, { "epoch": 1.5101233418664184, "grad_norm": 0.8131425976753235, "learning_rate": 0.0005, "loss": 1.339812994003296, "step": 12978 }, { "epoch": 1.5102397021177567, "grad_norm": 0.8157157301902771, "learning_rate": 0.0005, "loss": 2.1542181968688965, "step": 12979 }, { "epoch": 1.5103560623690946, "grad_norm": 1.4795942306518555, "learning_rate": 0.0005, "loss": 2.861318588256836, "step": 12980 }, { "epoch": 1.510472422620433, "grad_norm": 1.033497929573059, "learning_rate": 0.0005, "loss": 2.3455801010131836, "step": 12981 }, { "epoch": 1.510588782871771, "grad_norm": 0.704266369342804, "learning_rate": 0.0005, "loss": 1.6269006729125977, "step": 12982 }, { "epoch": 1.510705143123109, "grad_norm": 0.8778473734855652, "learning_rate": 0.0005, "loss": 2.1705403327941895, "step": 12983 }, { "epoch": 1.5108215033744474, "grad_norm": 0.9774231314659119, "learning_rate": 0.0005, "loss": 1.5894474983215332, "step": 12984 }, { "epoch": 1.5109378636257853, "grad_norm": 1.2244940996170044, "learning_rate": 0.0005, "loss": 2.445465564727783, "step": 12985 }, { "epoch": 1.5110542238771236, "grad_norm": 1.2442175149917603, "learning_rate": 0.0005, "loss": 1.6893304586410522, "step": 12986 }, { "epoch": 1.5111705841284617, "grad_norm": 1.1472712755203247, "learning_rate": 0.0005, "loss": 1.8108841180801392, "step": 12987 }, { "epoch": 1.5112869443797998, "grad_norm": 0.8054238557815552, "learning_rate": 0.0005, "loss": 1.6774420738220215, "step": 12988 }, { "epoch": 1.5114033046311381, "grad_norm": 1.80202317237854, "learning_rate": 0.0005, "loss": 1.9051915407180786, "step": 12989 }, { "epoch": 1.511519664882476, "grad_norm": 0.8857640027999878, "learning_rate": 0.0005, "loss": 1.5239442586898804, "step": 12990 }, { "epoch": 1.5116360251338143, "grad_norm": 0.9103928208351135, "learning_rate": 0.0005, "loss": 1.8623273372650146, "step": 12991 }, { "epoch": 1.5117523853851524, "grad_norm": 1.2433412075042725, "learning_rate": 0.0005, "loss": 2.7150955200195312, "step": 12992 }, { "epoch": 1.5118687456364905, "grad_norm": 0.7581937909126282, "learning_rate": 0.0005, "loss": 2.294326066970825, "step": 12993 }, { "epoch": 1.5119851058878289, "grad_norm": 1.1133017539978027, "learning_rate": 0.0005, "loss": 2.5824437141418457, "step": 12994 }, { "epoch": 1.5121014661391667, "grad_norm": 1.2688839435577393, "learning_rate": 0.0005, "loss": 2.4564638137817383, "step": 12995 }, { "epoch": 1.512217826390505, "grad_norm": 0.7382099032402039, "learning_rate": 0.0005, "loss": 1.6577799320220947, "step": 12996 }, { "epoch": 1.5123341866418432, "grad_norm": 1.1049163341522217, "learning_rate": 0.0005, "loss": 2.0688228607177734, "step": 12997 }, { "epoch": 1.5124505468931813, "grad_norm": 1.1304665803909302, "learning_rate": 0.0005, "loss": 1.449019193649292, "step": 12998 }, { "epoch": 1.5125669071445196, "grad_norm": 0.8971614241600037, "learning_rate": 0.0005, "loss": 1.9427367448806763, "step": 12999 }, { "epoch": 1.5126832673958575, "grad_norm": 1.064632534980774, "learning_rate": 0.0005, "loss": 2.0315656661987305, "step": 13000 }, { "epoch": 1.5127996276471958, "grad_norm": 0.5087819695472717, "learning_rate": 0.0005, "loss": 1.4095699787139893, "step": 13001 }, { "epoch": 1.5129159878985339, "grad_norm": 0.8610946536064148, "learning_rate": 0.0005, "loss": 1.8892022371292114, "step": 13002 }, { "epoch": 1.513032348149872, "grad_norm": 1.0251057147979736, "learning_rate": 0.0005, "loss": 1.3096282482147217, "step": 13003 }, { "epoch": 1.5131487084012103, "grad_norm": 1.2771449089050293, "learning_rate": 0.0005, "loss": 1.8976882696151733, "step": 13004 }, { "epoch": 1.5132650686525482, "grad_norm": 1.4443727731704712, "learning_rate": 0.0005, "loss": 2.0855884552001953, "step": 13005 }, { "epoch": 1.5133814289038865, "grad_norm": 1.5409528017044067, "learning_rate": 0.0005, "loss": 2.1601362228393555, "step": 13006 }, { "epoch": 1.5134977891552246, "grad_norm": 0.7917913794517517, "learning_rate": 0.0005, "loss": 1.4511051177978516, "step": 13007 }, { "epoch": 1.5136141494065627, "grad_norm": 1.062237024307251, "learning_rate": 0.0005, "loss": 2.1118221282958984, "step": 13008 }, { "epoch": 1.513730509657901, "grad_norm": 2.0585098266601562, "learning_rate": 0.0005, "loss": 2.380399227142334, "step": 13009 }, { "epoch": 1.513846869909239, "grad_norm": 1.353498101234436, "learning_rate": 0.0005, "loss": 2.0269346237182617, "step": 13010 }, { "epoch": 1.5139632301605772, "grad_norm": 0.8583902716636658, "learning_rate": 0.0005, "loss": 1.0889308452606201, "step": 13011 }, { "epoch": 1.5140795904119153, "grad_norm": 0.6771736741065979, "learning_rate": 0.0005, "loss": 1.8233766555786133, "step": 13012 }, { "epoch": 1.5141959506632534, "grad_norm": 0.9939432144165039, "learning_rate": 0.0005, "loss": 1.6713694334030151, "step": 13013 }, { "epoch": 1.5143123109145917, "grad_norm": 1.2447301149368286, "learning_rate": 0.0005, "loss": 1.906128168106079, "step": 13014 }, { "epoch": 1.5144286711659296, "grad_norm": 1.0294665098190308, "learning_rate": 0.0005, "loss": 1.6968426704406738, "step": 13015 }, { "epoch": 1.514545031417268, "grad_norm": 1.2450485229492188, "learning_rate": 0.0005, "loss": 2.278916835784912, "step": 13016 }, { "epoch": 1.514661391668606, "grad_norm": 1.4411232471466064, "learning_rate": 0.0005, "loss": 1.9722620248794556, "step": 13017 }, { "epoch": 1.5147777519199441, "grad_norm": 1.1613906621932983, "learning_rate": 0.0005, "loss": 2.0097408294677734, "step": 13018 }, { "epoch": 1.5148941121712824, "grad_norm": 1.6738804578781128, "learning_rate": 0.0005, "loss": 2.4562106132507324, "step": 13019 }, { "epoch": 1.5150104724226203, "grad_norm": 0.8648136258125305, "learning_rate": 0.0005, "loss": 1.6857491731643677, "step": 13020 }, { "epoch": 1.5151268326739586, "grad_norm": 1.4647817611694336, "learning_rate": 0.0005, "loss": 2.2519025802612305, "step": 13021 }, { "epoch": 1.5152431929252967, "grad_norm": 0.586189329624176, "learning_rate": 0.0005, "loss": 1.4953291416168213, "step": 13022 }, { "epoch": 1.5153595531766348, "grad_norm": 1.0467197895050049, "learning_rate": 0.0005, "loss": 1.6941697597503662, "step": 13023 }, { "epoch": 1.5154759134279732, "grad_norm": 0.831559956073761, "learning_rate": 0.0005, "loss": 2.136958599090576, "step": 13024 }, { "epoch": 1.515592273679311, "grad_norm": 0.9928175210952759, "learning_rate": 0.0005, "loss": 2.265062093734741, "step": 13025 }, { "epoch": 1.5157086339306494, "grad_norm": 1.3334752321243286, "learning_rate": 0.0005, "loss": 2.0367488861083984, "step": 13026 }, { "epoch": 1.5158249941819875, "grad_norm": 1.279244065284729, "learning_rate": 0.0005, "loss": 1.9221482276916504, "step": 13027 }, { "epoch": 1.5159413544333256, "grad_norm": 0.7726930379867554, "learning_rate": 0.0005, "loss": 1.5631046295166016, "step": 13028 }, { "epoch": 1.5160577146846639, "grad_norm": 1.1363437175750732, "learning_rate": 0.0005, "loss": 2.530057907104492, "step": 13029 }, { "epoch": 1.5161740749360018, "grad_norm": 0.8229116201400757, "learning_rate": 0.0005, "loss": 1.4361512660980225, "step": 13030 }, { "epoch": 1.51629043518734, "grad_norm": 1.1783946752548218, "learning_rate": 0.0005, "loss": 2.3235321044921875, "step": 13031 }, { "epoch": 1.5164067954386782, "grad_norm": 1.4067797660827637, "learning_rate": 0.0005, "loss": 1.6143171787261963, "step": 13032 }, { "epoch": 1.5165231556900163, "grad_norm": 0.9108263850212097, "learning_rate": 0.0005, "loss": 1.9980024099349976, "step": 13033 }, { "epoch": 1.5166395159413546, "grad_norm": 0.8845124244689941, "learning_rate": 0.0005, "loss": 1.8734430074691772, "step": 13034 }, { "epoch": 1.5167558761926925, "grad_norm": 0.6742126941680908, "learning_rate": 0.0005, "loss": 1.9177896976470947, "step": 13035 }, { "epoch": 1.5168722364440308, "grad_norm": 2.0537681579589844, "learning_rate": 0.0005, "loss": 3.2076363563537598, "step": 13036 }, { "epoch": 1.516988596695369, "grad_norm": 1.1446387767791748, "learning_rate": 0.0005, "loss": 2.0415053367614746, "step": 13037 }, { "epoch": 1.517104956946707, "grad_norm": 1.102054238319397, "learning_rate": 0.0005, "loss": 2.9232420921325684, "step": 13038 }, { "epoch": 1.517221317198045, "grad_norm": 1.376743197441101, "learning_rate": 0.0005, "loss": 1.463308334350586, "step": 13039 }, { "epoch": 1.5173376774493832, "grad_norm": 1.4462695121765137, "learning_rate": 0.0005, "loss": 1.8976064920425415, "step": 13040 }, { "epoch": 1.5174540377007215, "grad_norm": 1.335237741470337, "learning_rate": 0.0005, "loss": 2.361826181411743, "step": 13041 }, { "epoch": 1.5175703979520596, "grad_norm": 1.0049089193344116, "learning_rate": 0.0005, "loss": 1.2758711576461792, "step": 13042 }, { "epoch": 1.5176867582033977, "grad_norm": 2.0050501823425293, "learning_rate": 0.0005, "loss": 2.3556995391845703, "step": 13043 }, { "epoch": 1.5178031184547358, "grad_norm": 1.0609163045883179, "learning_rate": 0.0005, "loss": 2.2338550090789795, "step": 13044 }, { "epoch": 1.517919478706074, "grad_norm": 1.504128336906433, "learning_rate": 0.0005, "loss": 2.517139434814453, "step": 13045 }, { "epoch": 1.5180358389574122, "grad_norm": 1.2688276767730713, "learning_rate": 0.0005, "loss": 2.0642330646514893, "step": 13046 }, { "epoch": 1.5181521992087503, "grad_norm": 2.27494478225708, "learning_rate": 0.0005, "loss": 1.9723105430603027, "step": 13047 }, { "epoch": 1.5182685594600884, "grad_norm": 1.3006480932235718, "learning_rate": 0.0005, "loss": 1.0930765867233276, "step": 13048 }, { "epoch": 1.5183849197114265, "grad_norm": 0.8116711974143982, "learning_rate": 0.0005, "loss": 2.197772741317749, "step": 13049 }, { "epoch": 1.5185012799627646, "grad_norm": 1.0584543943405151, "learning_rate": 0.0005, "loss": 1.9189724922180176, "step": 13050 }, { "epoch": 1.518617640214103, "grad_norm": 0.7362178564071655, "learning_rate": 0.0005, "loss": 2.1670687198638916, "step": 13051 }, { "epoch": 1.518734000465441, "grad_norm": 1.1963063478469849, "learning_rate": 0.0005, "loss": 1.7466397285461426, "step": 13052 }, { "epoch": 1.5188503607167791, "grad_norm": 1.0073654651641846, "learning_rate": 0.0005, "loss": 2.328007221221924, "step": 13053 }, { "epoch": 1.5189667209681172, "grad_norm": 1.723456859588623, "learning_rate": 0.0005, "loss": 1.7579257488250732, "step": 13054 }, { "epoch": 1.5190830812194553, "grad_norm": 1.2948247194290161, "learning_rate": 0.0005, "loss": 1.980304479598999, "step": 13055 }, { "epoch": 1.5191994414707937, "grad_norm": 1.4259889125823975, "learning_rate": 0.0005, "loss": 1.8264529705047607, "step": 13056 }, { "epoch": 1.5193158017221318, "grad_norm": 1.2967658042907715, "learning_rate": 0.0005, "loss": 2.6811814308166504, "step": 13057 }, { "epoch": 1.5194321619734699, "grad_norm": 1.0734357833862305, "learning_rate": 0.0005, "loss": 2.424551010131836, "step": 13058 }, { "epoch": 1.519548522224808, "grad_norm": 0.6439323425292969, "learning_rate": 0.0005, "loss": 1.473130464553833, "step": 13059 }, { "epoch": 1.519664882476146, "grad_norm": 0.748991072177887, "learning_rate": 0.0005, "loss": 1.2116572856903076, "step": 13060 }, { "epoch": 1.5197812427274844, "grad_norm": 0.6125790476799011, "learning_rate": 0.0005, "loss": 1.1753498315811157, "step": 13061 }, { "epoch": 1.5198976029788225, "grad_norm": 0.9303100109100342, "learning_rate": 0.0005, "loss": 1.8602728843688965, "step": 13062 }, { "epoch": 1.5200139632301606, "grad_norm": 2.0092198848724365, "learning_rate": 0.0005, "loss": 1.8594144582748413, "step": 13063 }, { "epoch": 1.5201303234814987, "grad_norm": 1.3664252758026123, "learning_rate": 0.0005, "loss": 1.4351491928100586, "step": 13064 }, { "epoch": 1.5202466837328368, "grad_norm": 1.439168095588684, "learning_rate": 0.0005, "loss": 2.624298095703125, "step": 13065 }, { "epoch": 1.520363043984175, "grad_norm": 0.9794280529022217, "learning_rate": 0.0005, "loss": 1.7907755374908447, "step": 13066 }, { "epoch": 1.5204794042355132, "grad_norm": 1.1477785110473633, "learning_rate": 0.0005, "loss": 2.375100612640381, "step": 13067 }, { "epoch": 1.5205957644868513, "grad_norm": 1.1454490423202515, "learning_rate": 0.0005, "loss": 1.359199047088623, "step": 13068 }, { "epoch": 1.5207121247381894, "grad_norm": 1.5682122707366943, "learning_rate": 0.0005, "loss": 1.6409941911697388, "step": 13069 }, { "epoch": 1.5208284849895275, "grad_norm": 0.877663254737854, "learning_rate": 0.0005, "loss": 1.8285510540008545, "step": 13070 }, { "epoch": 1.5209448452408658, "grad_norm": 1.2521394491195679, "learning_rate": 0.0005, "loss": 2.2974154949188232, "step": 13071 }, { "epoch": 1.521061205492204, "grad_norm": 1.5889533758163452, "learning_rate": 0.0005, "loss": 2.5685553550720215, "step": 13072 }, { "epoch": 1.521177565743542, "grad_norm": 0.8180505633354187, "learning_rate": 0.0005, "loss": 1.147841215133667, "step": 13073 }, { "epoch": 1.5212939259948801, "grad_norm": 0.8876619935035706, "learning_rate": 0.0005, "loss": 1.5567126274108887, "step": 13074 }, { "epoch": 1.5214102862462182, "grad_norm": 1.072040319442749, "learning_rate": 0.0005, "loss": 1.737986445426941, "step": 13075 }, { "epoch": 1.5215266464975565, "grad_norm": 2.422921657562256, "learning_rate": 0.0005, "loss": 1.806630253791809, "step": 13076 }, { "epoch": 1.5216430067488944, "grad_norm": 1.524727463722229, "learning_rate": 0.0005, "loss": 2.4496548175811768, "step": 13077 }, { "epoch": 1.5217593670002327, "grad_norm": 3.020573377609253, "learning_rate": 0.0005, "loss": 1.3476696014404297, "step": 13078 }, { "epoch": 1.5218757272515708, "grad_norm": 1.4608385562896729, "learning_rate": 0.0005, "loss": 1.9981950521469116, "step": 13079 }, { "epoch": 1.521992087502909, "grad_norm": 0.8186982274055481, "learning_rate": 0.0005, "loss": 1.8533227443695068, "step": 13080 }, { "epoch": 1.5221084477542473, "grad_norm": 1.704621434211731, "learning_rate": 0.0005, "loss": 2.174576759338379, "step": 13081 }, { "epoch": 1.5222248080055851, "grad_norm": 0.8923514485359192, "learning_rate": 0.0005, "loss": 2.0208027362823486, "step": 13082 }, { "epoch": 1.5223411682569234, "grad_norm": 0.8818320631980896, "learning_rate": 0.0005, "loss": 2.146944046020508, "step": 13083 }, { "epoch": 1.5224575285082615, "grad_norm": 1.349792242050171, "learning_rate": 0.0005, "loss": 2.660616874694824, "step": 13084 }, { "epoch": 1.5225738887595996, "grad_norm": 1.1459625959396362, "learning_rate": 0.0005, "loss": 2.9769158363342285, "step": 13085 }, { "epoch": 1.522690249010938, "grad_norm": 0.6909345984458923, "learning_rate": 0.0005, "loss": 1.276219367980957, "step": 13086 }, { "epoch": 1.5228066092622758, "grad_norm": 1.1178231239318848, "learning_rate": 0.0005, "loss": 2.2252750396728516, "step": 13087 }, { "epoch": 1.5229229695136142, "grad_norm": 1.2592599391937256, "learning_rate": 0.0005, "loss": 2.8633921146392822, "step": 13088 }, { "epoch": 1.5230393297649523, "grad_norm": 2.652040719985962, "learning_rate": 0.0005, "loss": 1.8442933559417725, "step": 13089 }, { "epoch": 1.5231556900162904, "grad_norm": 1.3680078983306885, "learning_rate": 0.0005, "loss": 2.264801502227783, "step": 13090 }, { "epoch": 1.5232720502676287, "grad_norm": 1.0535683631896973, "learning_rate": 0.0005, "loss": 1.9399479627609253, "step": 13091 }, { "epoch": 1.5233884105189666, "grad_norm": 1.33506441116333, "learning_rate": 0.0005, "loss": 2.106492042541504, "step": 13092 }, { "epoch": 1.5235047707703049, "grad_norm": 0.9130874276161194, "learning_rate": 0.0005, "loss": 1.8360027074813843, "step": 13093 }, { "epoch": 1.523621131021643, "grad_norm": 1.0957415103912354, "learning_rate": 0.0005, "loss": 2.0314083099365234, "step": 13094 }, { "epoch": 1.523737491272981, "grad_norm": 0.5070493221282959, "learning_rate": 0.0005, "loss": 1.1387901306152344, "step": 13095 }, { "epoch": 1.5238538515243194, "grad_norm": 0.9109091758728027, "learning_rate": 0.0005, "loss": 1.469468355178833, "step": 13096 }, { "epoch": 1.5239702117756573, "grad_norm": 1.3575507402420044, "learning_rate": 0.0005, "loss": 2.1869027614593506, "step": 13097 }, { "epoch": 1.5240865720269956, "grad_norm": 0.6569625735282898, "learning_rate": 0.0005, "loss": 1.1254340410232544, "step": 13098 }, { "epoch": 1.5242029322783337, "grad_norm": 0.6343892812728882, "learning_rate": 0.0005, "loss": 1.4321117401123047, "step": 13099 }, { "epoch": 1.5243192925296718, "grad_norm": 1.61677086353302, "learning_rate": 0.0005, "loss": 1.8635406494140625, "step": 13100 }, { "epoch": 1.5244356527810101, "grad_norm": 0.7977615594863892, "learning_rate": 0.0005, "loss": 1.6343433856964111, "step": 13101 }, { "epoch": 1.524552013032348, "grad_norm": 0.9578171372413635, "learning_rate": 0.0005, "loss": 1.7919576168060303, "step": 13102 }, { "epoch": 1.5246683732836863, "grad_norm": 1.1766502857208252, "learning_rate": 0.0005, "loss": 2.1427557468414307, "step": 13103 }, { "epoch": 1.5247847335350244, "grad_norm": 1.108213186264038, "learning_rate": 0.0005, "loss": 2.2040133476257324, "step": 13104 }, { "epoch": 1.5249010937863625, "grad_norm": 0.9935802817344666, "learning_rate": 0.0005, "loss": 3.017852783203125, "step": 13105 }, { "epoch": 1.5250174540377008, "grad_norm": 1.4672653675079346, "learning_rate": 0.0005, "loss": 1.814913034439087, "step": 13106 }, { "epoch": 1.5251338142890387, "grad_norm": 1.3724383115768433, "learning_rate": 0.0005, "loss": 1.9716949462890625, "step": 13107 }, { "epoch": 1.525250174540377, "grad_norm": 1.3892309665679932, "learning_rate": 0.0005, "loss": 2.0816545486450195, "step": 13108 }, { "epoch": 1.5253665347917151, "grad_norm": 1.0610336065292358, "learning_rate": 0.0005, "loss": 2.3681859970092773, "step": 13109 }, { "epoch": 1.5254828950430532, "grad_norm": 1.0044574737548828, "learning_rate": 0.0005, "loss": 2.1794981956481934, "step": 13110 }, { "epoch": 1.5255992552943916, "grad_norm": 1.4900486469268799, "learning_rate": 0.0005, "loss": 2.1227307319641113, "step": 13111 }, { "epoch": 1.5257156155457294, "grad_norm": 1.3826072216033936, "learning_rate": 0.0005, "loss": 2.9119832515716553, "step": 13112 }, { "epoch": 1.5258319757970678, "grad_norm": 1.1080100536346436, "learning_rate": 0.0005, "loss": 1.9088926315307617, "step": 13113 }, { "epoch": 1.5259483360484059, "grad_norm": 1.8262650966644287, "learning_rate": 0.0005, "loss": 2.561265468597412, "step": 13114 }, { "epoch": 1.526064696299744, "grad_norm": 0.677436351776123, "learning_rate": 0.0005, "loss": 1.3845014572143555, "step": 13115 }, { "epoch": 1.5261810565510823, "grad_norm": 0.9759583473205566, "learning_rate": 0.0005, "loss": 1.8073397874832153, "step": 13116 }, { "epoch": 1.5262974168024201, "grad_norm": 0.8516537547111511, "learning_rate": 0.0005, "loss": 1.633980631828308, "step": 13117 }, { "epoch": 1.5264137770537585, "grad_norm": 1.2692104578018188, "learning_rate": 0.0005, "loss": 2.2473092079162598, "step": 13118 }, { "epoch": 1.5265301373050966, "grad_norm": 1.245039939880371, "learning_rate": 0.0005, "loss": 2.1195449829101562, "step": 13119 }, { "epoch": 1.5266464975564347, "grad_norm": 0.8392482399940491, "learning_rate": 0.0005, "loss": 1.3718225955963135, "step": 13120 }, { "epoch": 1.526762857807773, "grad_norm": 2.1493022441864014, "learning_rate": 0.0005, "loss": 1.964179515838623, "step": 13121 }, { "epoch": 1.5268792180591109, "grad_norm": 1.2393169403076172, "learning_rate": 0.0005, "loss": 2.3901467323303223, "step": 13122 }, { "epoch": 1.5269955783104492, "grad_norm": 1.3689504861831665, "learning_rate": 0.0005, "loss": 1.8380041122436523, "step": 13123 }, { "epoch": 1.5271119385617873, "grad_norm": 0.94551682472229, "learning_rate": 0.0005, "loss": 1.8633942604064941, "step": 13124 }, { "epoch": 1.5272282988131254, "grad_norm": 1.4329988956451416, "learning_rate": 0.0005, "loss": 2.5175697803497314, "step": 13125 }, { "epoch": 1.5273446590644637, "grad_norm": 1.0549134016036987, "learning_rate": 0.0005, "loss": 2.1842007637023926, "step": 13126 }, { "epoch": 1.5274610193158016, "grad_norm": 2.5255298614501953, "learning_rate": 0.0005, "loss": 1.9956637620925903, "step": 13127 }, { "epoch": 1.52757737956714, "grad_norm": 1.5954203605651855, "learning_rate": 0.0005, "loss": 1.819834589958191, "step": 13128 }, { "epoch": 1.527693739818478, "grad_norm": 0.9386032819747925, "learning_rate": 0.0005, "loss": 1.6596240997314453, "step": 13129 }, { "epoch": 1.527810100069816, "grad_norm": 1.352478265762329, "learning_rate": 0.0005, "loss": 2.085341453552246, "step": 13130 }, { "epoch": 1.5279264603211544, "grad_norm": 1.2286020517349243, "learning_rate": 0.0005, "loss": 1.860278844833374, "step": 13131 }, { "epoch": 1.5280428205724923, "grad_norm": 1.2448266744613647, "learning_rate": 0.0005, "loss": 1.872786283493042, "step": 13132 }, { "epoch": 1.5281591808238306, "grad_norm": 0.6594110727310181, "learning_rate": 0.0005, "loss": 1.240410566329956, "step": 13133 }, { "epoch": 1.5282755410751687, "grad_norm": 0.7008416056632996, "learning_rate": 0.0005, "loss": 1.3636415004730225, "step": 13134 }, { "epoch": 1.5283919013265068, "grad_norm": 0.7508533000946045, "learning_rate": 0.0005, "loss": 1.619173526763916, "step": 13135 }, { "epoch": 1.5285082615778451, "grad_norm": 0.6438082456588745, "learning_rate": 0.0005, "loss": 1.3507494926452637, "step": 13136 }, { "epoch": 1.528624621829183, "grad_norm": 1.0750118494033813, "learning_rate": 0.0005, "loss": 2.1324942111968994, "step": 13137 }, { "epoch": 1.5287409820805213, "grad_norm": 0.9422860741615295, "learning_rate": 0.0005, "loss": 1.8549678325653076, "step": 13138 }, { "epoch": 1.5288573423318594, "grad_norm": 2.581674098968506, "learning_rate": 0.0005, "loss": 3.1230030059814453, "step": 13139 }, { "epoch": 1.5289737025831975, "grad_norm": 1.2552191019058228, "learning_rate": 0.0005, "loss": 1.7000319957733154, "step": 13140 }, { "epoch": 1.5290900628345359, "grad_norm": 4.697878360748291, "learning_rate": 0.0005, "loss": 1.458488941192627, "step": 13141 }, { "epoch": 1.5292064230858737, "grad_norm": 1.3001959323883057, "learning_rate": 0.0005, "loss": 2.4961469173431396, "step": 13142 }, { "epoch": 1.529322783337212, "grad_norm": 1.8687477111816406, "learning_rate": 0.0005, "loss": 2.0677695274353027, "step": 13143 }, { "epoch": 1.5294391435885502, "grad_norm": 1.1639046669006348, "learning_rate": 0.0005, "loss": 2.124748468399048, "step": 13144 }, { "epoch": 1.5295555038398883, "grad_norm": 1.703965663909912, "learning_rate": 0.0005, "loss": 2.4203953742980957, "step": 13145 }, { "epoch": 1.5296718640912266, "grad_norm": 1.2180970907211304, "learning_rate": 0.0005, "loss": 2.070446491241455, "step": 13146 }, { "epoch": 1.5297882243425645, "grad_norm": 1.0268763303756714, "learning_rate": 0.0005, "loss": 2.189746856689453, "step": 13147 }, { "epoch": 1.5299045845939028, "grad_norm": 0.7887564897537231, "learning_rate": 0.0005, "loss": 2.510585308074951, "step": 13148 }, { "epoch": 1.5300209448452409, "grad_norm": 2.0710372924804688, "learning_rate": 0.0005, "loss": 2.039607048034668, "step": 13149 }, { "epoch": 1.530137305096579, "grad_norm": 1.1114230155944824, "learning_rate": 0.0005, "loss": 1.699315071105957, "step": 13150 }, { "epoch": 1.5302536653479173, "grad_norm": 0.6714921593666077, "learning_rate": 0.0005, "loss": 1.7353367805480957, "step": 13151 }, { "epoch": 1.5303700255992552, "grad_norm": 0.7899265885353088, "learning_rate": 0.0005, "loss": 1.9763565063476562, "step": 13152 }, { "epoch": 1.5304863858505935, "grad_norm": 1.1327520608901978, "learning_rate": 0.0005, "loss": 2.4286561012268066, "step": 13153 }, { "epoch": 1.5306027461019316, "grad_norm": 0.9457978010177612, "learning_rate": 0.0005, "loss": 2.1444458961486816, "step": 13154 }, { "epoch": 1.5307191063532697, "grad_norm": 0.8270923495292664, "learning_rate": 0.0005, "loss": 2.0808045864105225, "step": 13155 }, { "epoch": 1.530835466604608, "grad_norm": 0.7629337906837463, "learning_rate": 0.0005, "loss": 1.5454367399215698, "step": 13156 }, { "epoch": 1.5309518268559459, "grad_norm": 0.7782924175262451, "learning_rate": 0.0005, "loss": 1.5010660886764526, "step": 13157 }, { "epoch": 1.5310681871072842, "grad_norm": 0.824578583240509, "learning_rate": 0.0005, "loss": 1.6524064540863037, "step": 13158 }, { "epoch": 1.5311845473586223, "grad_norm": 1.041152000427246, "learning_rate": 0.0005, "loss": 1.7434074878692627, "step": 13159 }, { "epoch": 1.5313009076099604, "grad_norm": 0.7069413065910339, "learning_rate": 0.0005, "loss": 1.6489510536193848, "step": 13160 }, { "epoch": 1.5314172678612987, "grad_norm": 0.8182671666145325, "learning_rate": 0.0005, "loss": 1.9710158109664917, "step": 13161 }, { "epoch": 1.5315336281126366, "grad_norm": 1.3982802629470825, "learning_rate": 0.0005, "loss": 1.2420275211334229, "step": 13162 }, { "epoch": 1.531649988363975, "grad_norm": 1.1816442012786865, "learning_rate": 0.0005, "loss": 2.1382508277893066, "step": 13163 }, { "epoch": 1.531766348615313, "grad_norm": 0.9578937292098999, "learning_rate": 0.0005, "loss": 1.3374347686767578, "step": 13164 }, { "epoch": 1.5318827088666511, "grad_norm": 1.4795254468917847, "learning_rate": 0.0005, "loss": 2.061164140701294, "step": 13165 }, { "epoch": 1.5319990691179894, "grad_norm": 3.1241703033447266, "learning_rate": 0.0005, "loss": 1.2299230098724365, "step": 13166 }, { "epoch": 1.5321154293693273, "grad_norm": 0.8786327242851257, "learning_rate": 0.0005, "loss": 1.1493414640426636, "step": 13167 }, { "epoch": 1.5322317896206656, "grad_norm": 1.2020118236541748, "learning_rate": 0.0005, "loss": 1.6759992837905884, "step": 13168 }, { "epoch": 1.5323481498720037, "grad_norm": 0.9839704632759094, "learning_rate": 0.0005, "loss": 1.6228954792022705, "step": 13169 }, { "epoch": 1.5324645101233418, "grad_norm": 0.9962419271469116, "learning_rate": 0.0005, "loss": 2.071746587753296, "step": 13170 }, { "epoch": 1.5325808703746802, "grad_norm": 0.8149795532226562, "learning_rate": 0.0005, "loss": 1.4216300249099731, "step": 13171 }, { "epoch": 1.532697230626018, "grad_norm": 0.8513684868812561, "learning_rate": 0.0005, "loss": 1.8249664306640625, "step": 13172 }, { "epoch": 1.5328135908773564, "grad_norm": 0.7372353076934814, "learning_rate": 0.0005, "loss": 1.6882472038269043, "step": 13173 }, { "epoch": 1.5329299511286945, "grad_norm": 1.170937418937683, "learning_rate": 0.0005, "loss": 2.8874783515930176, "step": 13174 }, { "epoch": 1.5330463113800326, "grad_norm": 1.100407361984253, "learning_rate": 0.0005, "loss": 1.5005202293395996, "step": 13175 }, { "epoch": 1.5331626716313709, "grad_norm": 0.8969219326972961, "learning_rate": 0.0005, "loss": 1.4422005414962769, "step": 13176 }, { "epoch": 1.5332790318827088, "grad_norm": 0.9277841448783875, "learning_rate": 0.0005, "loss": 1.8054940700531006, "step": 13177 }, { "epoch": 1.533395392134047, "grad_norm": 1.1544418334960938, "learning_rate": 0.0005, "loss": 1.4390836954116821, "step": 13178 }, { "epoch": 1.5335117523853852, "grad_norm": 1.3001139163970947, "learning_rate": 0.0005, "loss": 2.4837846755981445, "step": 13179 }, { "epoch": 1.5336281126367233, "grad_norm": 1.2460983991622925, "learning_rate": 0.0005, "loss": 2.127976179122925, "step": 13180 }, { "epoch": 1.5337444728880616, "grad_norm": 1.1353613138198853, "learning_rate": 0.0005, "loss": 2.243070363998413, "step": 13181 }, { "epoch": 1.5338608331393995, "grad_norm": 1.343986988067627, "learning_rate": 0.0005, "loss": 2.0467402935028076, "step": 13182 }, { "epoch": 1.5339771933907378, "grad_norm": 1.385937213897705, "learning_rate": 0.0005, "loss": 1.3199071884155273, "step": 13183 }, { "epoch": 1.534093553642076, "grad_norm": 1.0912556648254395, "learning_rate": 0.0005, "loss": 1.5874651670455933, "step": 13184 }, { "epoch": 1.534209913893414, "grad_norm": 1.1178070306777954, "learning_rate": 0.0005, "loss": 1.335010051727295, "step": 13185 }, { "epoch": 1.5343262741447523, "grad_norm": 1.1269927024841309, "learning_rate": 0.0005, "loss": 2.137108564376831, "step": 13186 }, { "epoch": 1.5344426343960902, "grad_norm": 0.9644860029220581, "learning_rate": 0.0005, "loss": 2.002610683441162, "step": 13187 }, { "epoch": 1.5345589946474285, "grad_norm": 1.113131046295166, "learning_rate": 0.0005, "loss": 1.4334230422973633, "step": 13188 }, { "epoch": 1.5346753548987666, "grad_norm": 1.759937047958374, "learning_rate": 0.0005, "loss": 2.408421277999878, "step": 13189 }, { "epoch": 1.5347917151501047, "grad_norm": 1.0537617206573486, "learning_rate": 0.0005, "loss": 2.2719194889068604, "step": 13190 }, { "epoch": 1.534908075401443, "grad_norm": 1.3661179542541504, "learning_rate": 0.0005, "loss": 2.368386745452881, "step": 13191 }, { "epoch": 1.535024435652781, "grad_norm": 1.2368513345718384, "learning_rate": 0.0005, "loss": 2.4327592849731445, "step": 13192 }, { "epoch": 1.5351407959041192, "grad_norm": 1.0461771488189697, "learning_rate": 0.0005, "loss": 1.6269187927246094, "step": 13193 }, { "epoch": 1.5352571561554573, "grad_norm": 0.82329922914505, "learning_rate": 0.0005, "loss": 1.598503589630127, "step": 13194 }, { "epoch": 1.5353735164067954, "grad_norm": 0.8746817111968994, "learning_rate": 0.0005, "loss": 1.437325358390808, "step": 13195 }, { "epoch": 1.5354898766581337, "grad_norm": 1.009078025817871, "learning_rate": 0.0005, "loss": 2.293243408203125, "step": 13196 }, { "epoch": 1.5356062369094716, "grad_norm": 0.8826369643211365, "learning_rate": 0.0005, "loss": 1.525221824645996, "step": 13197 }, { "epoch": 1.53572259716081, "grad_norm": 0.4684056341648102, "learning_rate": 0.0005, "loss": 1.1846905946731567, "step": 13198 }, { "epoch": 1.535838957412148, "grad_norm": 0.8096669912338257, "learning_rate": 0.0005, "loss": 1.4473527669906616, "step": 13199 }, { "epoch": 1.5359553176634861, "grad_norm": 1.1579574346542358, "learning_rate": 0.0005, "loss": 1.8217538595199585, "step": 13200 }, { "epoch": 1.5360716779148242, "grad_norm": 0.7194176912307739, "learning_rate": 0.0005, "loss": 1.4477238655090332, "step": 13201 }, { "epoch": 1.5361880381661623, "grad_norm": 2.2806034088134766, "learning_rate": 0.0005, "loss": 1.8993664979934692, "step": 13202 }, { "epoch": 1.5363043984175007, "grad_norm": 1.155308485031128, "learning_rate": 0.0005, "loss": 2.1336822509765625, "step": 13203 }, { "epoch": 1.5364207586688388, "grad_norm": 1.0101473331451416, "learning_rate": 0.0005, "loss": 2.4709904193878174, "step": 13204 }, { "epoch": 1.5365371189201769, "grad_norm": 0.9708017706871033, "learning_rate": 0.0005, "loss": 2.006147861480713, "step": 13205 }, { "epoch": 1.536653479171515, "grad_norm": 1.1144678592681885, "learning_rate": 0.0005, "loss": 1.9470224380493164, "step": 13206 }, { "epoch": 1.536769839422853, "grad_norm": 0.9101101160049438, "learning_rate": 0.0005, "loss": 2.1562604904174805, "step": 13207 }, { "epoch": 1.5368861996741914, "grad_norm": 0.8229733109474182, "learning_rate": 0.0005, "loss": 1.5814952850341797, "step": 13208 }, { "epoch": 1.5370025599255295, "grad_norm": 1.8914698362350464, "learning_rate": 0.0005, "loss": 2.4294371604919434, "step": 13209 }, { "epoch": 1.5371189201768676, "grad_norm": 1.1456310749053955, "learning_rate": 0.0005, "loss": 2.3608129024505615, "step": 13210 }, { "epoch": 1.5372352804282057, "grad_norm": 1.1278663873672485, "learning_rate": 0.0005, "loss": 1.9181795120239258, "step": 13211 }, { "epoch": 1.5373516406795438, "grad_norm": 2.8387503623962402, "learning_rate": 0.0005, "loss": 2.1747019290924072, "step": 13212 }, { "epoch": 1.537468000930882, "grad_norm": 0.9544283151626587, "learning_rate": 0.0005, "loss": 2.0164244174957275, "step": 13213 }, { "epoch": 1.5375843611822202, "grad_norm": 1.2718981504440308, "learning_rate": 0.0005, "loss": 1.7452021837234497, "step": 13214 }, { "epoch": 1.5377007214335583, "grad_norm": 1.3491363525390625, "learning_rate": 0.0005, "loss": 2.596773624420166, "step": 13215 }, { "epoch": 1.5378170816848964, "grad_norm": 1.0638476610183716, "learning_rate": 0.0005, "loss": 1.88517427444458, "step": 13216 }, { "epoch": 1.5379334419362345, "grad_norm": 1.5456143617630005, "learning_rate": 0.0005, "loss": 1.7544995546340942, "step": 13217 }, { "epoch": 1.5380498021875728, "grad_norm": 1.1312326192855835, "learning_rate": 0.0005, "loss": 2.556312322616577, "step": 13218 }, { "epoch": 1.538166162438911, "grad_norm": 1.2545390129089355, "learning_rate": 0.0005, "loss": 1.944018840789795, "step": 13219 }, { "epoch": 1.538282522690249, "grad_norm": 1.2742851972579956, "learning_rate": 0.0005, "loss": 2.3590054512023926, "step": 13220 }, { "epoch": 1.538398882941587, "grad_norm": 2.0635459423065186, "learning_rate": 0.0005, "loss": 2.2963039875030518, "step": 13221 }, { "epoch": 1.5385152431929252, "grad_norm": 0.9299890398979187, "learning_rate": 0.0005, "loss": 1.6821335554122925, "step": 13222 }, { "epoch": 1.5386316034442635, "grad_norm": 1.913616418838501, "learning_rate": 0.0005, "loss": 2.0125436782836914, "step": 13223 }, { "epoch": 1.5387479636956016, "grad_norm": 1.2595654726028442, "learning_rate": 0.0005, "loss": 2.578352451324463, "step": 13224 }, { "epoch": 1.5388643239469397, "grad_norm": 1.7835956811904907, "learning_rate": 0.0005, "loss": 2.376112937927246, "step": 13225 }, { "epoch": 1.5389806841982778, "grad_norm": 1.0109848976135254, "learning_rate": 0.0005, "loss": 1.9899730682373047, "step": 13226 }, { "epoch": 1.539097044449616, "grad_norm": 0.949701726436615, "learning_rate": 0.0005, "loss": 1.8424561023712158, "step": 13227 }, { "epoch": 1.5392134047009542, "grad_norm": 1.5286874771118164, "learning_rate": 0.0005, "loss": 2.300652027130127, "step": 13228 }, { "epoch": 1.5393297649522923, "grad_norm": 1.3859140872955322, "learning_rate": 0.0005, "loss": 1.9988373517990112, "step": 13229 }, { "epoch": 1.5394461252036304, "grad_norm": 1.2099214792251587, "learning_rate": 0.0005, "loss": 2.2671256065368652, "step": 13230 }, { "epoch": 1.5395624854549685, "grad_norm": 1.0285670757293701, "learning_rate": 0.0005, "loss": 1.4053356647491455, "step": 13231 }, { "epoch": 1.5396788457063066, "grad_norm": 1.3005317449569702, "learning_rate": 0.0005, "loss": 2.1077089309692383, "step": 13232 }, { "epoch": 1.539795205957645, "grad_norm": 1.3393560647964478, "learning_rate": 0.0005, "loss": 2.5829696655273438, "step": 13233 }, { "epoch": 1.539911566208983, "grad_norm": 0.7644079327583313, "learning_rate": 0.0005, "loss": 1.4956092834472656, "step": 13234 }, { "epoch": 1.5400279264603212, "grad_norm": 1.9912241697311401, "learning_rate": 0.0005, "loss": 1.4939939975738525, "step": 13235 }, { "epoch": 1.5401442867116593, "grad_norm": 0.9770375490188599, "learning_rate": 0.0005, "loss": 1.4780421257019043, "step": 13236 }, { "epoch": 1.5402606469629974, "grad_norm": 1.6671950817108154, "learning_rate": 0.0005, "loss": 2.263579845428467, "step": 13237 }, { "epoch": 1.5403770072143357, "grad_norm": 1.1782578229904175, "learning_rate": 0.0005, "loss": 2.529179811477661, "step": 13238 }, { "epoch": 1.5404933674656738, "grad_norm": 1.3724626302719116, "learning_rate": 0.0005, "loss": 1.9082518815994263, "step": 13239 }, { "epoch": 1.5406097277170119, "grad_norm": 0.9424728155136108, "learning_rate": 0.0005, "loss": 1.6812899112701416, "step": 13240 }, { "epoch": 1.54072608796835, "grad_norm": 0.8167920708656311, "learning_rate": 0.0005, "loss": 2.183164119720459, "step": 13241 }, { "epoch": 1.540842448219688, "grad_norm": 1.6692179441452026, "learning_rate": 0.0005, "loss": 2.1498355865478516, "step": 13242 }, { "epoch": 1.5409588084710264, "grad_norm": 0.8674074411392212, "learning_rate": 0.0005, "loss": 2.2684638500213623, "step": 13243 }, { "epoch": 1.5410751687223643, "grad_norm": 1.123455286026001, "learning_rate": 0.0005, "loss": 2.22959303855896, "step": 13244 }, { "epoch": 1.5411915289737026, "grad_norm": 0.9628140330314636, "learning_rate": 0.0005, "loss": 2.0702242851257324, "step": 13245 }, { "epoch": 1.5413078892250407, "grad_norm": 1.3643773794174194, "learning_rate": 0.0005, "loss": 2.779557943344116, "step": 13246 }, { "epoch": 1.5414242494763788, "grad_norm": 1.4067070484161377, "learning_rate": 0.0005, "loss": 2.4297049045562744, "step": 13247 }, { "epoch": 1.5415406097277171, "grad_norm": 1.0659254789352417, "learning_rate": 0.0005, "loss": 1.8420706987380981, "step": 13248 }, { "epoch": 1.541656969979055, "grad_norm": 1.6722724437713623, "learning_rate": 0.0005, "loss": 1.5559675693511963, "step": 13249 }, { "epoch": 1.5417733302303933, "grad_norm": 1.204235315322876, "learning_rate": 0.0005, "loss": 2.634798049926758, "step": 13250 }, { "epoch": 1.5418896904817314, "grad_norm": 1.0193039178848267, "learning_rate": 0.0005, "loss": 1.4548085927963257, "step": 13251 }, { "epoch": 1.5420060507330695, "grad_norm": 0.7767723798751831, "learning_rate": 0.0005, "loss": 2.3740360736846924, "step": 13252 }, { "epoch": 1.5421224109844078, "grad_norm": 1.1915910243988037, "learning_rate": 0.0005, "loss": 2.560133457183838, "step": 13253 }, { "epoch": 1.5422387712357457, "grad_norm": 0.9726634621620178, "learning_rate": 0.0005, "loss": 1.5749709606170654, "step": 13254 }, { "epoch": 1.542355131487084, "grad_norm": 1.0420774221420288, "learning_rate": 0.0005, "loss": 1.7153348922729492, "step": 13255 }, { "epoch": 1.5424714917384221, "grad_norm": 1.3750579357147217, "learning_rate": 0.0005, "loss": 1.7273402214050293, "step": 13256 }, { "epoch": 1.5425878519897602, "grad_norm": 2.218778371810913, "learning_rate": 0.0005, "loss": 1.7972097396850586, "step": 13257 }, { "epoch": 1.5427042122410985, "grad_norm": 1.3602490425109863, "learning_rate": 0.0005, "loss": 1.4528875350952148, "step": 13258 }, { "epoch": 1.5428205724924364, "grad_norm": 1.5046226978302002, "learning_rate": 0.0005, "loss": 2.1099438667297363, "step": 13259 }, { "epoch": 1.5429369327437747, "grad_norm": 1.0792096853256226, "learning_rate": 0.0005, "loss": 1.5802668333053589, "step": 13260 }, { "epoch": 1.5430532929951128, "grad_norm": 1.2610677480697632, "learning_rate": 0.0005, "loss": 2.0077459812164307, "step": 13261 }, { "epoch": 1.543169653246451, "grad_norm": 1.0237083435058594, "learning_rate": 0.0005, "loss": 1.4225279092788696, "step": 13262 }, { "epoch": 1.5432860134977893, "grad_norm": 1.1302486658096313, "learning_rate": 0.0005, "loss": 2.9069647789001465, "step": 13263 }, { "epoch": 1.5434023737491271, "grad_norm": 1.1085835695266724, "learning_rate": 0.0005, "loss": 1.4715569019317627, "step": 13264 }, { "epoch": 1.5435187340004655, "grad_norm": 1.3826007843017578, "learning_rate": 0.0005, "loss": 2.534320831298828, "step": 13265 }, { "epoch": 1.5436350942518036, "grad_norm": 0.8972004652023315, "learning_rate": 0.0005, "loss": 1.9687511920928955, "step": 13266 }, { "epoch": 1.5437514545031417, "grad_norm": 1.2093735933303833, "learning_rate": 0.0005, "loss": 1.6759473085403442, "step": 13267 }, { "epoch": 1.54386781475448, "grad_norm": 1.2737687826156616, "learning_rate": 0.0005, "loss": 2.2161366939544678, "step": 13268 }, { "epoch": 1.5439841750058179, "grad_norm": 1.3010278940200806, "learning_rate": 0.0005, "loss": 2.0168495178222656, "step": 13269 }, { "epoch": 1.5441005352571562, "grad_norm": 0.5889127254486084, "learning_rate": 0.0005, "loss": 1.4482548236846924, "step": 13270 }, { "epoch": 1.5442168955084943, "grad_norm": 1.7278785705566406, "learning_rate": 0.0005, "loss": 1.5390937328338623, "step": 13271 }, { "epoch": 1.5443332557598324, "grad_norm": 1.0744649171829224, "learning_rate": 0.0005, "loss": 1.5781328678131104, "step": 13272 }, { "epoch": 1.5444496160111707, "grad_norm": 1.3399741649627686, "learning_rate": 0.0005, "loss": 2.223504066467285, "step": 13273 }, { "epoch": 1.5445659762625086, "grad_norm": 1.0034472942352295, "learning_rate": 0.0005, "loss": 2.0995326042175293, "step": 13274 }, { "epoch": 1.544682336513847, "grad_norm": 0.9629292488098145, "learning_rate": 0.0005, "loss": 1.5779461860656738, "step": 13275 }, { "epoch": 1.544798696765185, "grad_norm": 0.8459042310714722, "learning_rate": 0.0005, "loss": 1.809389591217041, "step": 13276 }, { "epoch": 1.544915057016523, "grad_norm": 1.4229038953781128, "learning_rate": 0.0005, "loss": 2.0794153213500977, "step": 13277 }, { "epoch": 1.5450314172678614, "grad_norm": 1.093532681465149, "learning_rate": 0.0005, "loss": 2.1743669509887695, "step": 13278 }, { "epoch": 1.5451477775191993, "grad_norm": 0.9711503386497498, "learning_rate": 0.0005, "loss": 1.7280821800231934, "step": 13279 }, { "epoch": 1.5452641377705376, "grad_norm": 0.7824082970619202, "learning_rate": 0.0005, "loss": 1.4578166007995605, "step": 13280 }, { "epoch": 1.5453804980218757, "grad_norm": 1.7019219398498535, "learning_rate": 0.0005, "loss": 1.7713775634765625, "step": 13281 }, { "epoch": 1.5454968582732138, "grad_norm": 1.605067253112793, "learning_rate": 0.0005, "loss": 2.47581148147583, "step": 13282 }, { "epoch": 1.5456132185245521, "grad_norm": 1.1028335094451904, "learning_rate": 0.0005, "loss": 1.4558616876602173, "step": 13283 }, { "epoch": 1.54572957877589, "grad_norm": 1.6883244514465332, "learning_rate": 0.0005, "loss": 1.8903592824935913, "step": 13284 }, { "epoch": 1.5458459390272283, "grad_norm": 0.7136910557746887, "learning_rate": 0.0005, "loss": 1.4737811088562012, "step": 13285 }, { "epoch": 1.5459622992785664, "grad_norm": 1.4284801483154297, "learning_rate": 0.0005, "loss": 1.7196474075317383, "step": 13286 }, { "epoch": 1.5460786595299045, "grad_norm": 1.534422516822815, "learning_rate": 0.0005, "loss": 2.415004253387451, "step": 13287 }, { "epoch": 1.5461950197812429, "grad_norm": 1.7456550598144531, "learning_rate": 0.0005, "loss": 2.0235438346862793, "step": 13288 }, { "epoch": 1.5463113800325807, "grad_norm": 1.239378809928894, "learning_rate": 0.0005, "loss": 1.7288453578948975, "step": 13289 }, { "epoch": 1.546427740283919, "grad_norm": 1.3916685581207275, "learning_rate": 0.0005, "loss": 2.011857032775879, "step": 13290 }, { "epoch": 1.5465441005352571, "grad_norm": 1.1719396114349365, "learning_rate": 0.0005, "loss": 1.9288140535354614, "step": 13291 }, { "epoch": 1.5466604607865952, "grad_norm": 0.7773336172103882, "learning_rate": 0.0005, "loss": 1.1223995685577393, "step": 13292 }, { "epoch": 1.5467768210379336, "grad_norm": 1.3678418397903442, "learning_rate": 0.0005, "loss": 2.52860164642334, "step": 13293 }, { "epoch": 1.5468931812892714, "grad_norm": 1.2037633657455444, "learning_rate": 0.0005, "loss": 2.5923542976379395, "step": 13294 }, { "epoch": 1.5470095415406098, "grad_norm": 0.8533692955970764, "learning_rate": 0.0005, "loss": 1.2002124786376953, "step": 13295 }, { "epoch": 1.5471259017919479, "grad_norm": 1.1368739604949951, "learning_rate": 0.0005, "loss": 2.1559247970581055, "step": 13296 }, { "epoch": 1.547242262043286, "grad_norm": 0.8933596611022949, "learning_rate": 0.0005, "loss": 2.245573043823242, "step": 13297 }, { "epoch": 1.5473586222946243, "grad_norm": 1.3220634460449219, "learning_rate": 0.0005, "loss": 2.0053582191467285, "step": 13298 }, { "epoch": 1.5474749825459622, "grad_norm": 1.819369912147522, "learning_rate": 0.0005, "loss": 2.2823033332824707, "step": 13299 }, { "epoch": 1.5475913427973005, "grad_norm": 1.2073243856430054, "learning_rate": 0.0005, "loss": 1.5832833051681519, "step": 13300 }, { "epoch": 1.5477077030486386, "grad_norm": 0.7897509336471558, "learning_rate": 0.0005, "loss": 1.3715431690216064, "step": 13301 }, { "epoch": 1.5478240632999767, "grad_norm": 0.823590874671936, "learning_rate": 0.0005, "loss": 1.742770791053772, "step": 13302 }, { "epoch": 1.547940423551315, "grad_norm": 0.7903034090995789, "learning_rate": 0.0005, "loss": 1.7930481433868408, "step": 13303 }, { "epoch": 1.5480567838026529, "grad_norm": 1.013572335243225, "learning_rate": 0.0005, "loss": 1.329007625579834, "step": 13304 }, { "epoch": 1.5481731440539912, "grad_norm": 1.4199447631835938, "learning_rate": 0.0005, "loss": 1.8595393896102905, "step": 13305 }, { "epoch": 1.5482895043053293, "grad_norm": 1.4168298244476318, "learning_rate": 0.0005, "loss": 2.8879499435424805, "step": 13306 }, { "epoch": 1.5484058645566674, "grad_norm": 0.8483012318611145, "learning_rate": 0.0005, "loss": 1.3579483032226562, "step": 13307 }, { "epoch": 1.5485222248080057, "grad_norm": 1.3381882905960083, "learning_rate": 0.0005, "loss": 2.3574790954589844, "step": 13308 }, { "epoch": 1.5486385850593436, "grad_norm": 0.5443614721298218, "learning_rate": 0.0005, "loss": 1.4233696460723877, "step": 13309 }, { "epoch": 1.548754945310682, "grad_norm": 0.569031834602356, "learning_rate": 0.0005, "loss": 1.1448835134506226, "step": 13310 }, { "epoch": 1.54887130556202, "grad_norm": 0.6455227732658386, "learning_rate": 0.0005, "loss": 1.350058913230896, "step": 13311 }, { "epoch": 1.5489876658133581, "grad_norm": 1.040940761566162, "learning_rate": 0.0005, "loss": 1.492593765258789, "step": 13312 }, { "epoch": 1.5491040260646964, "grad_norm": 0.6478472948074341, "learning_rate": 0.0005, "loss": 1.3337767124176025, "step": 13313 }, { "epoch": 1.5492203863160343, "grad_norm": 1.3523250818252563, "learning_rate": 0.0005, "loss": 2.383523941040039, "step": 13314 }, { "epoch": 1.5493367465673726, "grad_norm": 1.2439031600952148, "learning_rate": 0.0005, "loss": 2.3585429191589355, "step": 13315 }, { "epoch": 1.5494531068187107, "grad_norm": 1.3324165344238281, "learning_rate": 0.0005, "loss": 1.9138479232788086, "step": 13316 }, { "epoch": 1.5495694670700488, "grad_norm": 0.9060428142547607, "learning_rate": 0.0005, "loss": 1.861074686050415, "step": 13317 }, { "epoch": 1.5496858273213872, "grad_norm": 1.7827908992767334, "learning_rate": 0.0005, "loss": 2.1972036361694336, "step": 13318 }, { "epoch": 1.549802187572725, "grad_norm": 1.4889289140701294, "learning_rate": 0.0005, "loss": 2.143169403076172, "step": 13319 }, { "epoch": 1.5499185478240634, "grad_norm": 0.9619382619857788, "learning_rate": 0.0005, "loss": 2.3845748901367188, "step": 13320 }, { "epoch": 1.5500349080754015, "grad_norm": 1.407141923904419, "learning_rate": 0.0005, "loss": 1.7300770282745361, "step": 13321 }, { "epoch": 1.5501512683267396, "grad_norm": 1.0534225702285767, "learning_rate": 0.0005, "loss": 1.9745230674743652, "step": 13322 }, { "epoch": 1.5502676285780779, "grad_norm": 1.2940235137939453, "learning_rate": 0.0005, "loss": 2.4387075901031494, "step": 13323 }, { "epoch": 1.5503839888294157, "grad_norm": 1.036082148551941, "learning_rate": 0.0005, "loss": 1.9945032596588135, "step": 13324 }, { "epoch": 1.550500349080754, "grad_norm": 1.4478702545166016, "learning_rate": 0.0005, "loss": 1.5340774059295654, "step": 13325 }, { "epoch": 1.5506167093320922, "grad_norm": 0.7833636403083801, "learning_rate": 0.0005, "loss": 1.3674476146697998, "step": 13326 }, { "epoch": 1.5507330695834303, "grad_norm": 1.0010900497436523, "learning_rate": 0.0005, "loss": 2.309081554412842, "step": 13327 }, { "epoch": 1.5508494298347686, "grad_norm": 1.4405362606048584, "learning_rate": 0.0005, "loss": 1.648270606994629, "step": 13328 }, { "epoch": 1.5509657900861065, "grad_norm": 0.8130527138710022, "learning_rate": 0.0005, "loss": 1.8742661476135254, "step": 13329 }, { "epoch": 1.5510821503374448, "grad_norm": 3.7745256423950195, "learning_rate": 0.0005, "loss": 2.366387367248535, "step": 13330 }, { "epoch": 1.5511985105887829, "grad_norm": 1.4281059503555298, "learning_rate": 0.0005, "loss": 1.868434190750122, "step": 13331 }, { "epoch": 1.551314870840121, "grad_norm": 1.018465280532837, "learning_rate": 0.0005, "loss": 2.420638084411621, "step": 13332 }, { "epoch": 1.5514312310914593, "grad_norm": 1.276800513267517, "learning_rate": 0.0005, "loss": 2.072742462158203, "step": 13333 }, { "epoch": 1.5515475913427972, "grad_norm": 1.3359261751174927, "learning_rate": 0.0005, "loss": 2.461235523223877, "step": 13334 }, { "epoch": 1.5516639515941355, "grad_norm": 1.6829570531845093, "learning_rate": 0.0005, "loss": 2.19366717338562, "step": 13335 }, { "epoch": 1.5517803118454736, "grad_norm": 1.0100178718566895, "learning_rate": 0.0005, "loss": 1.9718880653381348, "step": 13336 }, { "epoch": 1.5518966720968117, "grad_norm": 1.0683972835540771, "learning_rate": 0.0005, "loss": 1.6923820972442627, "step": 13337 }, { "epoch": 1.55201303234815, "grad_norm": 1.221129298210144, "learning_rate": 0.0005, "loss": 1.7619554996490479, "step": 13338 }, { "epoch": 1.552129392599488, "grad_norm": 1.4339423179626465, "learning_rate": 0.0005, "loss": 2.1085801124572754, "step": 13339 }, { "epoch": 1.5522457528508262, "grad_norm": 1.631085991859436, "learning_rate": 0.0005, "loss": 2.447519302368164, "step": 13340 }, { "epoch": 1.5523621131021643, "grad_norm": 0.8733460307121277, "learning_rate": 0.0005, "loss": 1.4083397388458252, "step": 13341 }, { "epoch": 1.5524784733535024, "grad_norm": 1.3493095636367798, "learning_rate": 0.0005, "loss": 2.136427402496338, "step": 13342 }, { "epoch": 1.5525948336048407, "grad_norm": 0.6069682836532593, "learning_rate": 0.0005, "loss": 1.5324500799179077, "step": 13343 }, { "epoch": 1.5527111938561786, "grad_norm": 1.2323086261749268, "learning_rate": 0.0005, "loss": 2.5468268394470215, "step": 13344 }, { "epoch": 1.552827554107517, "grad_norm": 1.4171851873397827, "learning_rate": 0.0005, "loss": 1.899442434310913, "step": 13345 }, { "epoch": 1.552943914358855, "grad_norm": 0.7733126282691956, "learning_rate": 0.0005, "loss": 1.7444368600845337, "step": 13346 }, { "epoch": 1.5530602746101931, "grad_norm": 1.303585410118103, "learning_rate": 0.0005, "loss": 2.071661949157715, "step": 13347 }, { "epoch": 1.5531766348615315, "grad_norm": 0.9117871522903442, "learning_rate": 0.0005, "loss": 1.92360520362854, "step": 13348 }, { "epoch": 1.5532929951128693, "grad_norm": 0.8046948909759521, "learning_rate": 0.0005, "loss": 1.8507894277572632, "step": 13349 }, { "epoch": 1.5534093553642077, "grad_norm": 1.0958443880081177, "learning_rate": 0.0005, "loss": 2.1964383125305176, "step": 13350 }, { "epoch": 1.5535257156155458, "grad_norm": 1.8167870044708252, "learning_rate": 0.0005, "loss": 1.8987104892730713, "step": 13351 }, { "epoch": 1.5536420758668839, "grad_norm": 0.8821601271629333, "learning_rate": 0.0005, "loss": 1.4442119598388672, "step": 13352 }, { "epoch": 1.5537584361182222, "grad_norm": 1.2758264541625977, "learning_rate": 0.0005, "loss": 2.4094080924987793, "step": 13353 }, { "epoch": 1.55387479636956, "grad_norm": 0.7109474539756775, "learning_rate": 0.0005, "loss": 1.7454593181610107, "step": 13354 }, { "epoch": 1.5539911566208984, "grad_norm": 1.494210958480835, "learning_rate": 0.0005, "loss": 2.129063367843628, "step": 13355 }, { "epoch": 1.5541075168722365, "grad_norm": 0.9679514169692993, "learning_rate": 0.0005, "loss": 2.106473445892334, "step": 13356 }, { "epoch": 1.5542238771235746, "grad_norm": 0.9337621331214905, "learning_rate": 0.0005, "loss": 2.027677059173584, "step": 13357 }, { "epoch": 1.554340237374913, "grad_norm": 1.5740408897399902, "learning_rate": 0.0005, "loss": 1.6788079738616943, "step": 13358 }, { "epoch": 1.5544565976262508, "grad_norm": 0.842499852180481, "learning_rate": 0.0005, "loss": 1.5612106323242188, "step": 13359 }, { "epoch": 1.554572957877589, "grad_norm": 1.3684030771255493, "learning_rate": 0.0005, "loss": 1.7012696266174316, "step": 13360 }, { "epoch": 1.5546893181289272, "grad_norm": 2.005270004272461, "learning_rate": 0.0005, "loss": 1.686572551727295, "step": 13361 }, { "epoch": 1.5548056783802653, "grad_norm": 0.866980254650116, "learning_rate": 0.0005, "loss": 1.9574415683746338, "step": 13362 }, { "epoch": 1.5549220386316036, "grad_norm": 0.7890490293502808, "learning_rate": 0.0005, "loss": 1.648974895477295, "step": 13363 }, { "epoch": 1.5550383988829415, "grad_norm": 2.4138264656066895, "learning_rate": 0.0005, "loss": 1.83514404296875, "step": 13364 }, { "epoch": 1.5551547591342798, "grad_norm": 0.9079474210739136, "learning_rate": 0.0005, "loss": 1.8454334735870361, "step": 13365 }, { "epoch": 1.555271119385618, "grad_norm": 1.0443692207336426, "learning_rate": 0.0005, "loss": 1.607058048248291, "step": 13366 }, { "epoch": 1.555387479636956, "grad_norm": 1.0635004043579102, "learning_rate": 0.0005, "loss": 1.6936860084533691, "step": 13367 }, { "epoch": 1.555503839888294, "grad_norm": 1.0577380657196045, "learning_rate": 0.0005, "loss": 2.1727468967437744, "step": 13368 }, { "epoch": 1.5556202001396322, "grad_norm": 0.662837028503418, "learning_rate": 0.0005, "loss": 1.7374004125595093, "step": 13369 }, { "epoch": 1.5557365603909705, "grad_norm": 1.2184513807296753, "learning_rate": 0.0005, "loss": 1.7782022953033447, "step": 13370 }, { "epoch": 1.5558529206423086, "grad_norm": 1.259631633758545, "learning_rate": 0.0005, "loss": 2.8155956268310547, "step": 13371 }, { "epoch": 1.5559692808936467, "grad_norm": 1.020951271057129, "learning_rate": 0.0005, "loss": 1.598610281944275, "step": 13372 }, { "epoch": 1.5560856411449848, "grad_norm": 0.7786713242530823, "learning_rate": 0.0005, "loss": 1.945142388343811, "step": 13373 }, { "epoch": 1.556202001396323, "grad_norm": 1.132153034210205, "learning_rate": 0.0005, "loss": 1.9518013000488281, "step": 13374 }, { "epoch": 1.5563183616476612, "grad_norm": 0.7900897860527039, "learning_rate": 0.0005, "loss": 1.3228223323822021, "step": 13375 }, { "epoch": 1.5564347218989993, "grad_norm": 0.6807382106781006, "learning_rate": 0.0005, "loss": 1.1912466287612915, "step": 13376 }, { "epoch": 1.5565510821503374, "grad_norm": 1.091141700744629, "learning_rate": 0.0005, "loss": 1.8240268230438232, "step": 13377 }, { "epoch": 1.5566674424016755, "grad_norm": 0.8910931348800659, "learning_rate": 0.0005, "loss": 1.9552721977233887, "step": 13378 }, { "epoch": 1.5567838026530136, "grad_norm": 0.8946311473846436, "learning_rate": 0.0005, "loss": 1.5568598508834839, "step": 13379 }, { "epoch": 1.556900162904352, "grad_norm": 0.6883222460746765, "learning_rate": 0.0005, "loss": 1.89918851852417, "step": 13380 }, { "epoch": 1.55701652315569, "grad_norm": 1.2773858308792114, "learning_rate": 0.0005, "loss": 1.8346154689788818, "step": 13381 }, { "epoch": 1.5571328834070282, "grad_norm": 1.0555673837661743, "learning_rate": 0.0005, "loss": 1.8229289054870605, "step": 13382 }, { "epoch": 1.5572492436583663, "grad_norm": 0.7149641513824463, "learning_rate": 0.0005, "loss": 1.1630828380584717, "step": 13383 }, { "epoch": 1.5573656039097044, "grad_norm": 0.7705469727516174, "learning_rate": 0.0005, "loss": 1.500745415687561, "step": 13384 }, { "epoch": 1.5574819641610427, "grad_norm": 1.631252408027649, "learning_rate": 0.0005, "loss": 2.264714241027832, "step": 13385 }, { "epoch": 1.5575983244123808, "grad_norm": 1.202771544456482, "learning_rate": 0.0005, "loss": 2.3026304244995117, "step": 13386 }, { "epoch": 1.5577146846637189, "grad_norm": 0.9383032917976379, "learning_rate": 0.0005, "loss": 2.1001462936401367, "step": 13387 }, { "epoch": 1.557831044915057, "grad_norm": 1.0502835512161255, "learning_rate": 0.0005, "loss": 2.460094451904297, "step": 13388 }, { "epoch": 1.557947405166395, "grad_norm": 1.4293586015701294, "learning_rate": 0.0005, "loss": 1.75203275680542, "step": 13389 }, { "epoch": 1.5580637654177334, "grad_norm": 1.2006964683532715, "learning_rate": 0.0005, "loss": 2.8147244453430176, "step": 13390 }, { "epoch": 1.5581801256690715, "grad_norm": 1.0520365238189697, "learning_rate": 0.0005, "loss": 1.763310194015503, "step": 13391 }, { "epoch": 1.5582964859204096, "grad_norm": 1.3563328981399536, "learning_rate": 0.0005, "loss": 1.951303482055664, "step": 13392 }, { "epoch": 1.5584128461717477, "grad_norm": 1.1467872858047485, "learning_rate": 0.0005, "loss": 2.1671252250671387, "step": 13393 }, { "epoch": 1.5585292064230858, "grad_norm": 0.6281328201293945, "learning_rate": 0.0005, "loss": 1.558570384979248, "step": 13394 }, { "epoch": 1.558645566674424, "grad_norm": 1.0948612689971924, "learning_rate": 0.0005, "loss": 2.15837025642395, "step": 13395 }, { "epoch": 1.5587619269257622, "grad_norm": 1.3617610931396484, "learning_rate": 0.0005, "loss": 1.8126569986343384, "step": 13396 }, { "epoch": 1.5588782871771003, "grad_norm": 1.1120103597640991, "learning_rate": 0.0005, "loss": 1.4925189018249512, "step": 13397 }, { "epoch": 1.5589946474284384, "grad_norm": 0.8354700803756714, "learning_rate": 0.0005, "loss": 1.7452524900436401, "step": 13398 }, { "epoch": 1.5591110076797765, "grad_norm": 1.3837264776229858, "learning_rate": 0.0005, "loss": 2.386826992034912, "step": 13399 }, { "epoch": 1.5592273679311148, "grad_norm": 0.6794087290763855, "learning_rate": 0.0005, "loss": 1.8071014881134033, "step": 13400 }, { "epoch": 1.559343728182453, "grad_norm": 1.608285903930664, "learning_rate": 0.0005, "loss": 1.648902416229248, "step": 13401 }, { "epoch": 1.559460088433791, "grad_norm": 1.0467476844787598, "learning_rate": 0.0005, "loss": 1.6532073020935059, "step": 13402 }, { "epoch": 1.5595764486851291, "grad_norm": 0.9102752804756165, "learning_rate": 0.0005, "loss": 1.2278296947479248, "step": 13403 }, { "epoch": 1.5596928089364672, "grad_norm": 1.1103276014328003, "learning_rate": 0.0005, "loss": 1.5172221660614014, "step": 13404 }, { "epoch": 1.5598091691878055, "grad_norm": 1.0834531784057617, "learning_rate": 0.0005, "loss": 1.4374501705169678, "step": 13405 }, { "epoch": 1.5599255294391436, "grad_norm": 1.6056857109069824, "learning_rate": 0.0005, "loss": 1.7665491104125977, "step": 13406 }, { "epoch": 1.5600418896904817, "grad_norm": 2.036007881164551, "learning_rate": 0.0005, "loss": 1.5043656826019287, "step": 13407 }, { "epoch": 1.5601582499418198, "grad_norm": 1.115470051765442, "learning_rate": 0.0005, "loss": 1.6029791831970215, "step": 13408 }, { "epoch": 1.560274610193158, "grad_norm": 0.5908070802688599, "learning_rate": 0.0005, "loss": 1.2694958448410034, "step": 13409 }, { "epoch": 1.5603909704444963, "grad_norm": 1.1399301290512085, "learning_rate": 0.0005, "loss": 1.298102617263794, "step": 13410 }, { "epoch": 1.5605073306958341, "grad_norm": 1.0427889823913574, "learning_rate": 0.0005, "loss": 2.193683624267578, "step": 13411 }, { "epoch": 1.5606236909471725, "grad_norm": 1.0882161855697632, "learning_rate": 0.0005, "loss": 2.058957576751709, "step": 13412 }, { "epoch": 1.5607400511985106, "grad_norm": 1.1526145935058594, "learning_rate": 0.0005, "loss": 2.0063886642456055, "step": 13413 }, { "epoch": 1.5608564114498487, "grad_norm": 1.1158140897750854, "learning_rate": 0.0005, "loss": 1.7482906579971313, "step": 13414 }, { "epoch": 1.560972771701187, "grad_norm": 1.2551597356796265, "learning_rate": 0.0005, "loss": 1.5346274375915527, "step": 13415 }, { "epoch": 1.5610891319525249, "grad_norm": 0.9628999829292297, "learning_rate": 0.0005, "loss": 1.8062748908996582, "step": 13416 }, { "epoch": 1.5612054922038632, "grad_norm": 1.6600894927978516, "learning_rate": 0.0005, "loss": 2.2484591007232666, "step": 13417 }, { "epoch": 1.5613218524552013, "grad_norm": 1.2717245817184448, "learning_rate": 0.0005, "loss": 2.8069698810577393, "step": 13418 }, { "epoch": 1.5614382127065394, "grad_norm": 1.2826149463653564, "learning_rate": 0.0005, "loss": 1.7969642877578735, "step": 13419 }, { "epoch": 1.5615545729578777, "grad_norm": 1.2530595064163208, "learning_rate": 0.0005, "loss": 2.2318601608276367, "step": 13420 }, { "epoch": 1.5616709332092156, "grad_norm": 0.8061347007751465, "learning_rate": 0.0005, "loss": 1.0716743469238281, "step": 13421 }, { "epoch": 1.561787293460554, "grad_norm": 0.7167201042175293, "learning_rate": 0.0005, "loss": 1.650748372077942, "step": 13422 }, { "epoch": 1.561903653711892, "grad_norm": 1.0064510107040405, "learning_rate": 0.0005, "loss": 2.703313112258911, "step": 13423 }, { "epoch": 1.56202001396323, "grad_norm": 1.0348572731018066, "learning_rate": 0.0005, "loss": 1.2793060541152954, "step": 13424 }, { "epoch": 1.5621363742145684, "grad_norm": 1.139517903327942, "learning_rate": 0.0005, "loss": 2.4643659591674805, "step": 13425 }, { "epoch": 1.5622527344659063, "grad_norm": 0.9620540738105774, "learning_rate": 0.0005, "loss": 1.66542387008667, "step": 13426 }, { "epoch": 1.5623690947172446, "grad_norm": 1.4876078367233276, "learning_rate": 0.0005, "loss": 2.5621280670166016, "step": 13427 }, { "epoch": 1.5624854549685827, "grad_norm": 1.3628982305526733, "learning_rate": 0.0005, "loss": 1.6396677494049072, "step": 13428 }, { "epoch": 1.5626018152199208, "grad_norm": 1.01966392993927, "learning_rate": 0.0005, "loss": 1.4828879833221436, "step": 13429 }, { "epoch": 1.5627181754712591, "grad_norm": 1.2260606288909912, "learning_rate": 0.0005, "loss": 1.8150484561920166, "step": 13430 }, { "epoch": 1.562834535722597, "grad_norm": 1.0788357257843018, "learning_rate": 0.0005, "loss": 1.490462064743042, "step": 13431 }, { "epoch": 1.5629508959739353, "grad_norm": 5.551595687866211, "learning_rate": 0.0005, "loss": 2.606605052947998, "step": 13432 }, { "epoch": 1.5630672562252734, "grad_norm": 0.9590635299682617, "learning_rate": 0.0005, "loss": 1.99387788772583, "step": 13433 }, { "epoch": 1.5631836164766115, "grad_norm": 1.6789007186889648, "learning_rate": 0.0005, "loss": 2.632704734802246, "step": 13434 }, { "epoch": 1.5632999767279498, "grad_norm": 0.8600485920906067, "learning_rate": 0.0005, "loss": 1.8815913200378418, "step": 13435 }, { "epoch": 1.5634163369792877, "grad_norm": 0.8257424831390381, "learning_rate": 0.0005, "loss": 1.807075023651123, "step": 13436 }, { "epoch": 1.563532697230626, "grad_norm": 0.7819545865058899, "learning_rate": 0.0005, "loss": 1.987661361694336, "step": 13437 }, { "epoch": 1.5636490574819641, "grad_norm": 1.4036574363708496, "learning_rate": 0.0005, "loss": 2.2984132766723633, "step": 13438 }, { "epoch": 1.5637654177333022, "grad_norm": 1.174048900604248, "learning_rate": 0.0005, "loss": 2.2598371505737305, "step": 13439 }, { "epoch": 1.5638817779846406, "grad_norm": 1.105915904045105, "learning_rate": 0.0005, "loss": 2.168992757797241, "step": 13440 }, { "epoch": 1.5639981382359784, "grad_norm": 1.3198922872543335, "learning_rate": 0.0005, "loss": 1.9225856065750122, "step": 13441 }, { "epoch": 1.5641144984873168, "grad_norm": 0.8434134721755981, "learning_rate": 0.0005, "loss": 1.3176220655441284, "step": 13442 }, { "epoch": 1.5642308587386549, "grad_norm": 6.066948890686035, "learning_rate": 0.0005, "loss": 1.7493669986724854, "step": 13443 }, { "epoch": 1.564347218989993, "grad_norm": 1.4680583477020264, "learning_rate": 0.0005, "loss": 1.9249640703201294, "step": 13444 }, { "epoch": 1.5644635792413313, "grad_norm": 1.3378076553344727, "learning_rate": 0.0005, "loss": 1.8137996196746826, "step": 13445 }, { "epoch": 1.5645799394926692, "grad_norm": 0.7458644509315491, "learning_rate": 0.0005, "loss": 1.4204310178756714, "step": 13446 }, { "epoch": 1.5646962997440075, "grad_norm": 1.1054682731628418, "learning_rate": 0.0005, "loss": 1.2911031246185303, "step": 13447 }, { "epoch": 1.5648126599953456, "grad_norm": 1.2662204504013062, "learning_rate": 0.0005, "loss": 1.897702932357788, "step": 13448 }, { "epoch": 1.5649290202466837, "grad_norm": 1.1436296701431274, "learning_rate": 0.0005, "loss": 2.43746280670166, "step": 13449 }, { "epoch": 1.565045380498022, "grad_norm": 0.8577805161476135, "learning_rate": 0.0005, "loss": 1.6178667545318604, "step": 13450 }, { "epoch": 1.5651617407493599, "grad_norm": 1.7089890241622925, "learning_rate": 0.0005, "loss": 2.132049560546875, "step": 13451 }, { "epoch": 1.5652781010006982, "grad_norm": 0.9578372240066528, "learning_rate": 0.0005, "loss": 2.1151108741760254, "step": 13452 }, { "epoch": 1.5653944612520363, "grad_norm": 1.5566624402999878, "learning_rate": 0.0005, "loss": 2.1251330375671387, "step": 13453 }, { "epoch": 1.5655108215033744, "grad_norm": 0.8233059644699097, "learning_rate": 0.0005, "loss": 1.3511216640472412, "step": 13454 }, { "epoch": 1.5656271817547127, "grad_norm": 1.290695071220398, "learning_rate": 0.0005, "loss": 1.4868974685668945, "step": 13455 }, { "epoch": 1.5657435420060506, "grad_norm": 1.012717843055725, "learning_rate": 0.0005, "loss": 1.6561572551727295, "step": 13456 }, { "epoch": 1.565859902257389, "grad_norm": 1.3437026739120483, "learning_rate": 0.0005, "loss": 1.4966304302215576, "step": 13457 }, { "epoch": 1.565976262508727, "grad_norm": 0.7912753820419312, "learning_rate": 0.0005, "loss": 1.4073796272277832, "step": 13458 }, { "epoch": 1.5660926227600651, "grad_norm": 1.9143500328063965, "learning_rate": 0.0005, "loss": 2.7599127292633057, "step": 13459 }, { "epoch": 1.5662089830114034, "grad_norm": 0.7610607147216797, "learning_rate": 0.0005, "loss": 1.670809268951416, "step": 13460 }, { "epoch": 1.5663253432627413, "grad_norm": 1.1737666130065918, "learning_rate": 0.0005, "loss": 1.747753381729126, "step": 13461 }, { "epoch": 1.5664417035140796, "grad_norm": 1.0737377405166626, "learning_rate": 0.0005, "loss": 2.4170026779174805, "step": 13462 }, { "epoch": 1.5665580637654177, "grad_norm": 1.3637341260910034, "learning_rate": 0.0005, "loss": 2.369389057159424, "step": 13463 }, { "epoch": 1.5666744240167558, "grad_norm": 1.3508002758026123, "learning_rate": 0.0005, "loss": 2.0344457626342773, "step": 13464 }, { "epoch": 1.5667907842680941, "grad_norm": 1.1158772706985474, "learning_rate": 0.0005, "loss": 2.3391170501708984, "step": 13465 }, { "epoch": 1.566907144519432, "grad_norm": 1.0637359619140625, "learning_rate": 0.0005, "loss": 1.6107532978057861, "step": 13466 }, { "epoch": 1.5670235047707703, "grad_norm": 1.216426134109497, "learning_rate": 0.0005, "loss": 1.7851158380508423, "step": 13467 }, { "epoch": 1.5671398650221084, "grad_norm": 1.8696619272232056, "learning_rate": 0.0005, "loss": 2.50482177734375, "step": 13468 }, { "epoch": 1.5672562252734465, "grad_norm": 1.1664814949035645, "learning_rate": 0.0005, "loss": 1.881981372833252, "step": 13469 }, { "epoch": 1.5673725855247849, "grad_norm": 1.4620686769485474, "learning_rate": 0.0005, "loss": 2.084503173828125, "step": 13470 }, { "epoch": 1.5674889457761227, "grad_norm": 1.0206691026687622, "learning_rate": 0.0005, "loss": 2.0089592933654785, "step": 13471 }, { "epoch": 1.567605306027461, "grad_norm": 0.9446032047271729, "learning_rate": 0.0005, "loss": 1.8479082584381104, "step": 13472 }, { "epoch": 1.5677216662787992, "grad_norm": 0.967515230178833, "learning_rate": 0.0005, "loss": 1.4657268524169922, "step": 13473 }, { "epoch": 1.5678380265301373, "grad_norm": 0.8882481455802917, "learning_rate": 0.0005, "loss": 2.14762282371521, "step": 13474 }, { "epoch": 1.5679543867814756, "grad_norm": 0.7344369888305664, "learning_rate": 0.0005, "loss": 1.5891640186309814, "step": 13475 }, { "epoch": 1.5680707470328135, "grad_norm": 1.3619180917739868, "learning_rate": 0.0005, "loss": 2.5746912956237793, "step": 13476 }, { "epoch": 1.5681871072841518, "grad_norm": 1.0660514831542969, "learning_rate": 0.0005, "loss": 1.7719519138336182, "step": 13477 }, { "epoch": 1.5683034675354899, "grad_norm": 1.3238054513931274, "learning_rate": 0.0005, "loss": 2.1862094402313232, "step": 13478 }, { "epoch": 1.568419827786828, "grad_norm": 1.3274132013320923, "learning_rate": 0.0005, "loss": 2.1250991821289062, "step": 13479 }, { "epoch": 1.5685361880381663, "grad_norm": 0.5011571645736694, "learning_rate": 0.0005, "loss": 1.0018339157104492, "step": 13480 }, { "epoch": 1.5686525482895042, "grad_norm": 1.4226597547531128, "learning_rate": 0.0005, "loss": 2.4702398777008057, "step": 13481 }, { "epoch": 1.5687689085408425, "grad_norm": 1.3729994297027588, "learning_rate": 0.0005, "loss": 2.0484585762023926, "step": 13482 }, { "epoch": 1.5688852687921806, "grad_norm": 1.490882396697998, "learning_rate": 0.0005, "loss": 2.176910400390625, "step": 13483 }, { "epoch": 1.5690016290435187, "grad_norm": 1.6872036457061768, "learning_rate": 0.0005, "loss": 2.3091323375701904, "step": 13484 }, { "epoch": 1.569117989294857, "grad_norm": 0.8860781788825989, "learning_rate": 0.0005, "loss": 1.3726577758789062, "step": 13485 }, { "epoch": 1.569234349546195, "grad_norm": 1.2393808364868164, "learning_rate": 0.0005, "loss": 1.5124105215072632, "step": 13486 }, { "epoch": 1.5693507097975332, "grad_norm": 0.566139817237854, "learning_rate": 0.0005, "loss": 1.0458004474639893, "step": 13487 }, { "epoch": 1.5694670700488713, "grad_norm": 0.5775531530380249, "learning_rate": 0.0005, "loss": 1.1241209506988525, "step": 13488 }, { "epoch": 1.5695834303002094, "grad_norm": 0.9410028457641602, "learning_rate": 0.0005, "loss": 1.9429118633270264, "step": 13489 }, { "epoch": 1.5696997905515477, "grad_norm": 0.944892168045044, "learning_rate": 0.0005, "loss": 1.5697276592254639, "step": 13490 }, { "epoch": 1.5698161508028856, "grad_norm": 0.864639163017273, "learning_rate": 0.0005, "loss": 1.7255624532699585, "step": 13491 }, { "epoch": 1.569932511054224, "grad_norm": 1.549058198928833, "learning_rate": 0.0005, "loss": 1.4766309261322021, "step": 13492 }, { "epoch": 1.570048871305562, "grad_norm": 1.2250255346298218, "learning_rate": 0.0005, "loss": 2.318397283554077, "step": 13493 }, { "epoch": 1.5701652315569001, "grad_norm": 1.0180964469909668, "learning_rate": 0.0005, "loss": 2.208958864212036, "step": 13494 }, { "epoch": 1.5702815918082385, "grad_norm": 1.589888572692871, "learning_rate": 0.0005, "loss": 2.382352828979492, "step": 13495 }, { "epoch": 1.5703979520595763, "grad_norm": 1.0874558687210083, "learning_rate": 0.0005, "loss": 1.367088794708252, "step": 13496 }, { "epoch": 1.5705143123109147, "grad_norm": 1.4268513917922974, "learning_rate": 0.0005, "loss": 2.204068660736084, "step": 13497 }, { "epoch": 1.5706306725622527, "grad_norm": 1.1621427536010742, "learning_rate": 0.0005, "loss": 1.8033467531204224, "step": 13498 }, { "epoch": 1.5707470328135908, "grad_norm": 1.564673662185669, "learning_rate": 0.0005, "loss": 3.027947425842285, "step": 13499 }, { "epoch": 1.5708633930649292, "grad_norm": 0.7608010768890381, "learning_rate": 0.0005, "loss": 1.666323184967041, "step": 13500 }, { "epoch": 1.570979753316267, "grad_norm": 0.477786660194397, "learning_rate": 0.0005, "loss": 0.9807422161102295, "step": 13501 }, { "epoch": 1.5710961135676054, "grad_norm": 1.0152332782745361, "learning_rate": 0.0005, "loss": 2.099299907684326, "step": 13502 }, { "epoch": 1.5712124738189435, "grad_norm": 1.3328807353973389, "learning_rate": 0.0005, "loss": 2.1775617599487305, "step": 13503 }, { "epoch": 1.5713288340702816, "grad_norm": 1.2323102951049805, "learning_rate": 0.0005, "loss": 2.163588047027588, "step": 13504 }, { "epoch": 1.5714451943216199, "grad_norm": 0.6427290439605713, "learning_rate": 0.0005, "loss": 1.7374887466430664, "step": 13505 }, { "epoch": 1.5715615545729578, "grad_norm": 1.1253403425216675, "learning_rate": 0.0005, "loss": 1.7184159755706787, "step": 13506 }, { "epoch": 1.571677914824296, "grad_norm": 1.1821691989898682, "learning_rate": 0.0005, "loss": 1.7641899585723877, "step": 13507 }, { "epoch": 1.5717942750756342, "grad_norm": 1.0484830141067505, "learning_rate": 0.0005, "loss": 1.9996459484100342, "step": 13508 }, { "epoch": 1.5719106353269723, "grad_norm": 1.0443276166915894, "learning_rate": 0.0005, "loss": 2.4233999252319336, "step": 13509 }, { "epoch": 1.5720269955783106, "grad_norm": 1.395384430885315, "learning_rate": 0.0005, "loss": 1.9664061069488525, "step": 13510 }, { "epoch": 1.5721433558296485, "grad_norm": 1.2538132667541504, "learning_rate": 0.0005, "loss": 1.5161125659942627, "step": 13511 }, { "epoch": 1.5722597160809868, "grad_norm": 1.1817564964294434, "learning_rate": 0.0005, "loss": 2.558058738708496, "step": 13512 }, { "epoch": 1.572376076332325, "grad_norm": 1.6477621793746948, "learning_rate": 0.0005, "loss": 1.9475061893463135, "step": 13513 }, { "epoch": 1.572492436583663, "grad_norm": 1.6046277284622192, "learning_rate": 0.0005, "loss": 2.130819320678711, "step": 13514 }, { "epoch": 1.5726087968350013, "grad_norm": 0.4552499055862427, "learning_rate": 0.0005, "loss": 1.6100842952728271, "step": 13515 }, { "epoch": 1.5727251570863392, "grad_norm": 1.2382804155349731, "learning_rate": 0.0005, "loss": 1.9627653360366821, "step": 13516 }, { "epoch": 1.5728415173376775, "grad_norm": 1.2166556119918823, "learning_rate": 0.0005, "loss": 2.0711636543273926, "step": 13517 }, { "epoch": 1.5729578775890156, "grad_norm": 1.707837700843811, "learning_rate": 0.0005, "loss": 1.1716047525405884, "step": 13518 }, { "epoch": 1.5730742378403537, "grad_norm": 1.3748780488967896, "learning_rate": 0.0005, "loss": 2.2688469886779785, "step": 13519 }, { "epoch": 1.573190598091692, "grad_norm": 1.5116225481033325, "learning_rate": 0.0005, "loss": 1.8931467533111572, "step": 13520 }, { "epoch": 1.57330695834303, "grad_norm": 1.0616616010665894, "learning_rate": 0.0005, "loss": 1.4892066717147827, "step": 13521 }, { "epoch": 1.5734233185943682, "grad_norm": 0.8041079640388489, "learning_rate": 0.0005, "loss": 1.6676547527313232, "step": 13522 }, { "epoch": 1.5735396788457063, "grad_norm": 1.1143494844436646, "learning_rate": 0.0005, "loss": 2.013998508453369, "step": 13523 }, { "epoch": 1.5736560390970444, "grad_norm": 1.0973681211471558, "learning_rate": 0.0005, "loss": 2.217284679412842, "step": 13524 }, { "epoch": 1.5737723993483828, "grad_norm": 1.1724047660827637, "learning_rate": 0.0005, "loss": 1.6056569814682007, "step": 13525 }, { "epoch": 1.5738887595997206, "grad_norm": 1.4301810264587402, "learning_rate": 0.0005, "loss": 2.2049028873443604, "step": 13526 }, { "epoch": 1.574005119851059, "grad_norm": 1.0446792840957642, "learning_rate": 0.0005, "loss": 2.3901267051696777, "step": 13527 }, { "epoch": 1.574121480102397, "grad_norm": 0.9534007906913757, "learning_rate": 0.0005, "loss": 2.9664201736450195, "step": 13528 }, { "epoch": 1.5742378403537352, "grad_norm": 0.6575694680213928, "learning_rate": 0.0005, "loss": 1.222182035446167, "step": 13529 }, { "epoch": 1.5743542006050735, "grad_norm": 1.711262822151184, "learning_rate": 0.0005, "loss": 1.6494300365447998, "step": 13530 }, { "epoch": 1.5744705608564113, "grad_norm": 1.001658320426941, "learning_rate": 0.0005, "loss": 1.8113648891448975, "step": 13531 }, { "epoch": 1.5745869211077497, "grad_norm": 0.9716507792472839, "learning_rate": 0.0005, "loss": 2.135845184326172, "step": 13532 }, { "epoch": 1.5747032813590878, "grad_norm": 1.354407548904419, "learning_rate": 0.0005, "loss": 2.176185369491577, "step": 13533 }, { "epoch": 1.5748196416104259, "grad_norm": 1.1706676483154297, "learning_rate": 0.0005, "loss": 1.9955368041992188, "step": 13534 }, { "epoch": 1.574936001861764, "grad_norm": 1.0361155271530151, "learning_rate": 0.0005, "loss": 1.303954839706421, "step": 13535 }, { "epoch": 1.575052362113102, "grad_norm": 1.158568263053894, "learning_rate": 0.0005, "loss": 2.392956018447876, "step": 13536 }, { "epoch": 1.5751687223644404, "grad_norm": 1.2797489166259766, "learning_rate": 0.0005, "loss": 1.8454759120941162, "step": 13537 }, { "epoch": 1.5752850826157785, "grad_norm": 1.5284157991409302, "learning_rate": 0.0005, "loss": 2.352281093597412, "step": 13538 }, { "epoch": 1.5754014428671166, "grad_norm": 1.1816799640655518, "learning_rate": 0.0005, "loss": 1.8867149353027344, "step": 13539 }, { "epoch": 1.5755178031184547, "grad_norm": 0.9609493017196655, "learning_rate": 0.0005, "loss": 1.863981008529663, "step": 13540 }, { "epoch": 1.5756341633697928, "grad_norm": 1.2250621318817139, "learning_rate": 0.0005, "loss": 2.509066581726074, "step": 13541 }, { "epoch": 1.575750523621131, "grad_norm": 1.4234120845794678, "learning_rate": 0.0005, "loss": 2.2205986976623535, "step": 13542 }, { "epoch": 1.5758668838724692, "grad_norm": 0.9446897506713867, "learning_rate": 0.0005, "loss": 2.241032600402832, "step": 13543 }, { "epoch": 1.5759832441238073, "grad_norm": 1.2777328491210938, "learning_rate": 0.0005, "loss": 1.5663098096847534, "step": 13544 }, { "epoch": 1.5760996043751454, "grad_norm": 1.0392497777938843, "learning_rate": 0.0005, "loss": 2.2895987033843994, "step": 13545 }, { "epoch": 1.5762159646264835, "grad_norm": 1.4733458757400513, "learning_rate": 0.0005, "loss": 1.7647110223770142, "step": 13546 }, { "epoch": 1.5763323248778218, "grad_norm": 0.9712660908699036, "learning_rate": 0.0005, "loss": 1.5160319805145264, "step": 13547 }, { "epoch": 1.57644868512916, "grad_norm": 1.372991919517517, "learning_rate": 0.0005, "loss": 1.8354439735412598, "step": 13548 }, { "epoch": 1.576565045380498, "grad_norm": 1.6725245714187622, "learning_rate": 0.0005, "loss": 2.009131908416748, "step": 13549 }, { "epoch": 1.5766814056318361, "grad_norm": 1.1423993110656738, "learning_rate": 0.0005, "loss": 2.237440347671509, "step": 13550 }, { "epoch": 1.5767977658831742, "grad_norm": 1.0679973363876343, "learning_rate": 0.0005, "loss": 1.4904816150665283, "step": 13551 }, { "epoch": 1.5769141261345125, "grad_norm": 2.2561943531036377, "learning_rate": 0.0005, "loss": 1.8002058267593384, "step": 13552 }, { "epoch": 1.5770304863858506, "grad_norm": 1.1600900888442993, "learning_rate": 0.0005, "loss": 1.4245173931121826, "step": 13553 }, { "epoch": 1.5771468466371887, "grad_norm": 0.9004865884780884, "learning_rate": 0.0005, "loss": 1.6160824298858643, "step": 13554 }, { "epoch": 1.5772632068885268, "grad_norm": 1.0787166357040405, "learning_rate": 0.0005, "loss": 1.6311805248260498, "step": 13555 }, { "epoch": 1.577379567139865, "grad_norm": 1.0408318042755127, "learning_rate": 0.0005, "loss": 1.4471006393432617, "step": 13556 }, { "epoch": 1.5774959273912033, "grad_norm": 1.166374683380127, "learning_rate": 0.0005, "loss": 1.4819908142089844, "step": 13557 }, { "epoch": 1.5776122876425414, "grad_norm": 0.9537472724914551, "learning_rate": 0.0005, "loss": 1.5684657096862793, "step": 13558 }, { "epoch": 1.5777286478938795, "grad_norm": 1.885277271270752, "learning_rate": 0.0005, "loss": 3.2112464904785156, "step": 13559 }, { "epoch": 1.5778450081452176, "grad_norm": 1.2678967714309692, "learning_rate": 0.0005, "loss": 1.9639883041381836, "step": 13560 }, { "epoch": 1.5779613683965557, "grad_norm": 0.9958217740058899, "learning_rate": 0.0005, "loss": 1.5335161685943604, "step": 13561 }, { "epoch": 1.578077728647894, "grad_norm": 1.1859853267669678, "learning_rate": 0.0005, "loss": 1.939717411994934, "step": 13562 }, { "epoch": 1.578194088899232, "grad_norm": 0.9932119250297546, "learning_rate": 0.0005, "loss": 1.5159621238708496, "step": 13563 }, { "epoch": 1.5783104491505702, "grad_norm": 1.2851637601852417, "learning_rate": 0.0005, "loss": 3.160140037536621, "step": 13564 }, { "epoch": 1.5784268094019083, "grad_norm": 0.7169296741485596, "learning_rate": 0.0005, "loss": 2.090181350708008, "step": 13565 }, { "epoch": 1.5785431696532464, "grad_norm": 1.259899616241455, "learning_rate": 0.0005, "loss": 2.0561866760253906, "step": 13566 }, { "epoch": 1.5786595299045847, "grad_norm": 1.243528127670288, "learning_rate": 0.0005, "loss": 1.8994499444961548, "step": 13567 }, { "epoch": 1.5787758901559228, "grad_norm": 0.9980807900428772, "learning_rate": 0.0005, "loss": 1.1961262226104736, "step": 13568 }, { "epoch": 1.5788922504072609, "grad_norm": 1.5995317697525024, "learning_rate": 0.0005, "loss": 1.8069816827774048, "step": 13569 }, { "epoch": 1.579008610658599, "grad_norm": 1.0061780214309692, "learning_rate": 0.0005, "loss": 1.922109603881836, "step": 13570 }, { "epoch": 1.579124970909937, "grad_norm": 1.645803451538086, "learning_rate": 0.0005, "loss": 2.3173131942749023, "step": 13571 }, { "epoch": 1.5792413311612754, "grad_norm": 1.0100510120391846, "learning_rate": 0.0005, "loss": 2.19840669631958, "step": 13572 }, { "epoch": 1.5793576914126133, "grad_norm": 0.9180929064750671, "learning_rate": 0.0005, "loss": 1.444033145904541, "step": 13573 }, { "epoch": 1.5794740516639516, "grad_norm": 0.9554294347763062, "learning_rate": 0.0005, "loss": 1.5517888069152832, "step": 13574 }, { "epoch": 1.5795904119152897, "grad_norm": 2.636531352996826, "learning_rate": 0.0005, "loss": 1.6338965892791748, "step": 13575 }, { "epoch": 1.5797067721666278, "grad_norm": 0.9060589075088501, "learning_rate": 0.0005, "loss": 2.3159329891204834, "step": 13576 }, { "epoch": 1.5798231324179661, "grad_norm": 1.5647810697555542, "learning_rate": 0.0005, "loss": 1.6901885271072388, "step": 13577 }, { "epoch": 1.579939492669304, "grad_norm": 1.6012303829193115, "learning_rate": 0.0005, "loss": 2.877936840057373, "step": 13578 }, { "epoch": 1.5800558529206423, "grad_norm": 1.1203572750091553, "learning_rate": 0.0005, "loss": 1.4730157852172852, "step": 13579 }, { "epoch": 1.5801722131719804, "grad_norm": 0.8464881777763367, "learning_rate": 0.0005, "loss": 1.9527581930160522, "step": 13580 }, { "epoch": 1.5802885734233185, "grad_norm": 1.0529893636703491, "learning_rate": 0.0005, "loss": 1.6621028184890747, "step": 13581 }, { "epoch": 1.5804049336746568, "grad_norm": 1.0772483348846436, "learning_rate": 0.0005, "loss": 2.0223875045776367, "step": 13582 }, { "epoch": 1.5805212939259947, "grad_norm": 1.0430625677108765, "learning_rate": 0.0005, "loss": 1.7581206560134888, "step": 13583 }, { "epoch": 1.580637654177333, "grad_norm": 1.776906132698059, "learning_rate": 0.0005, "loss": 2.431032657623291, "step": 13584 }, { "epoch": 1.5807540144286711, "grad_norm": 1.821341872215271, "learning_rate": 0.0005, "loss": 1.5285946130752563, "step": 13585 }, { "epoch": 1.5808703746800092, "grad_norm": 1.2131108045578003, "learning_rate": 0.0005, "loss": 2.2751617431640625, "step": 13586 }, { "epoch": 1.5809867349313476, "grad_norm": 0.884438693523407, "learning_rate": 0.0005, "loss": 1.5791471004486084, "step": 13587 }, { "epoch": 1.5811030951826854, "grad_norm": 1.9183119535446167, "learning_rate": 0.0005, "loss": 2.1767852306365967, "step": 13588 }, { "epoch": 1.5812194554340238, "grad_norm": 1.364047646522522, "learning_rate": 0.0005, "loss": 2.118443012237549, "step": 13589 }, { "epoch": 1.5813358156853619, "grad_norm": 1.252060890197754, "learning_rate": 0.0005, "loss": 1.9208519458770752, "step": 13590 }, { "epoch": 1.5814521759367, "grad_norm": 1.0716043710708618, "learning_rate": 0.0005, "loss": 1.6004490852355957, "step": 13591 }, { "epoch": 1.5815685361880383, "grad_norm": 1.4924367666244507, "learning_rate": 0.0005, "loss": 1.7952823638916016, "step": 13592 }, { "epoch": 1.5816848964393762, "grad_norm": 1.306454062461853, "learning_rate": 0.0005, "loss": 2.592679500579834, "step": 13593 }, { "epoch": 1.5818012566907145, "grad_norm": 1.4013986587524414, "learning_rate": 0.0005, "loss": 2.3874340057373047, "step": 13594 }, { "epoch": 1.5819176169420526, "grad_norm": 1.2447282075881958, "learning_rate": 0.0005, "loss": 2.1593868732452393, "step": 13595 }, { "epoch": 1.5820339771933907, "grad_norm": 0.9933803081512451, "learning_rate": 0.0005, "loss": 1.2204984426498413, "step": 13596 }, { "epoch": 1.582150337444729, "grad_norm": 1.2389274835586548, "learning_rate": 0.0005, "loss": 1.578331708908081, "step": 13597 }, { "epoch": 1.5822666976960669, "grad_norm": 0.84710294008255, "learning_rate": 0.0005, "loss": 1.7724814414978027, "step": 13598 }, { "epoch": 1.5823830579474052, "grad_norm": 1.1036490201950073, "learning_rate": 0.0005, "loss": 2.809032678604126, "step": 13599 }, { "epoch": 1.5824994181987433, "grad_norm": 1.1702508926391602, "learning_rate": 0.0005, "loss": 1.7801569700241089, "step": 13600 }, { "epoch": 1.5826157784500814, "grad_norm": 0.8545762300491333, "learning_rate": 0.0005, "loss": 1.7994756698608398, "step": 13601 }, { "epoch": 1.5827321387014197, "grad_norm": 0.9993621706962585, "learning_rate": 0.0005, "loss": 2.2470383644104004, "step": 13602 }, { "epoch": 1.5828484989527576, "grad_norm": 0.6427273154258728, "learning_rate": 0.0005, "loss": 2.0997674465179443, "step": 13603 }, { "epoch": 1.582964859204096, "grad_norm": 1.0197268724441528, "learning_rate": 0.0005, "loss": 1.5975894927978516, "step": 13604 }, { "epoch": 1.583081219455434, "grad_norm": 1.1912099123001099, "learning_rate": 0.0005, "loss": 2.134751081466675, "step": 13605 }, { "epoch": 1.583197579706772, "grad_norm": 0.9066997170448303, "learning_rate": 0.0005, "loss": 1.7810873985290527, "step": 13606 }, { "epoch": 1.5833139399581104, "grad_norm": 1.6991127729415894, "learning_rate": 0.0005, "loss": 2.34578800201416, "step": 13607 }, { "epoch": 1.5834303002094483, "grad_norm": 1.1109585762023926, "learning_rate": 0.0005, "loss": 1.5972859859466553, "step": 13608 }, { "epoch": 1.5835466604607866, "grad_norm": 1.5474166870117188, "learning_rate": 0.0005, "loss": 1.898306131362915, "step": 13609 }, { "epoch": 1.5836630207121247, "grad_norm": 1.4324848651885986, "learning_rate": 0.0005, "loss": 1.818199634552002, "step": 13610 }, { "epoch": 1.5837793809634628, "grad_norm": 1.2064225673675537, "learning_rate": 0.0005, "loss": 1.48910653591156, "step": 13611 }, { "epoch": 1.5838957412148011, "grad_norm": 1.0898135900497437, "learning_rate": 0.0005, "loss": 2.055833339691162, "step": 13612 }, { "epoch": 1.584012101466139, "grad_norm": 0.9516192674636841, "learning_rate": 0.0005, "loss": 2.2141175270080566, "step": 13613 }, { "epoch": 1.5841284617174773, "grad_norm": 0.8168588280677795, "learning_rate": 0.0005, "loss": 2.2326276302337646, "step": 13614 }, { "epoch": 1.5842448219688154, "grad_norm": 0.5520424246788025, "learning_rate": 0.0005, "loss": 1.074927806854248, "step": 13615 }, { "epoch": 1.5843611822201535, "grad_norm": 1.3798847198486328, "learning_rate": 0.0005, "loss": 1.9943108558654785, "step": 13616 }, { "epoch": 1.5844775424714919, "grad_norm": 1.086449146270752, "learning_rate": 0.0005, "loss": 2.1984145641326904, "step": 13617 }, { "epoch": 1.5845939027228297, "grad_norm": 1.3127986192703247, "learning_rate": 0.0005, "loss": 1.8116904497146606, "step": 13618 }, { "epoch": 1.584710262974168, "grad_norm": 1.1813658475875854, "learning_rate": 0.0005, "loss": 1.8592768907546997, "step": 13619 }, { "epoch": 1.5848266232255062, "grad_norm": 1.5915888547897339, "learning_rate": 0.0005, "loss": 2.1983160972595215, "step": 13620 }, { "epoch": 1.5849429834768443, "grad_norm": 1.267520546913147, "learning_rate": 0.0005, "loss": 1.9830362796783447, "step": 13621 }, { "epoch": 1.5850593437281826, "grad_norm": 1.208980679512024, "learning_rate": 0.0005, "loss": 2.000699043273926, "step": 13622 }, { "epoch": 1.5851757039795205, "grad_norm": 1.0145792961120605, "learning_rate": 0.0005, "loss": 2.1134159564971924, "step": 13623 }, { "epoch": 1.5852920642308588, "grad_norm": 1.8414833545684814, "learning_rate": 0.0005, "loss": 1.8099124431610107, "step": 13624 }, { "epoch": 1.5854084244821969, "grad_norm": 0.8128982782363892, "learning_rate": 0.0005, "loss": 1.3841851949691772, "step": 13625 }, { "epoch": 1.585524784733535, "grad_norm": 1.3407424688339233, "learning_rate": 0.0005, "loss": 2.518258810043335, "step": 13626 }, { "epoch": 1.5856411449848733, "grad_norm": 0.8801430463790894, "learning_rate": 0.0005, "loss": 1.8194386959075928, "step": 13627 }, { "epoch": 1.5857575052362112, "grad_norm": 11.833470344543457, "learning_rate": 0.0005, "loss": 1.5310993194580078, "step": 13628 }, { "epoch": 1.5858738654875495, "grad_norm": 0.7772406339645386, "learning_rate": 0.0005, "loss": 1.7296417951583862, "step": 13629 }, { "epoch": 1.5859902257388876, "grad_norm": 1.8052953481674194, "learning_rate": 0.0005, "loss": 2.0549206733703613, "step": 13630 }, { "epoch": 1.5861065859902257, "grad_norm": 1.2265077829360962, "learning_rate": 0.0005, "loss": 2.5094847679138184, "step": 13631 }, { "epoch": 1.586222946241564, "grad_norm": 1.634034276008606, "learning_rate": 0.0005, "loss": 2.573728084564209, "step": 13632 }, { "epoch": 1.586339306492902, "grad_norm": 1.3355135917663574, "learning_rate": 0.0005, "loss": 1.984614372253418, "step": 13633 }, { "epoch": 1.5864556667442402, "grad_norm": 2.2939186096191406, "learning_rate": 0.0005, "loss": 1.7088429927825928, "step": 13634 }, { "epoch": 1.5865720269955783, "grad_norm": 1.0057525634765625, "learning_rate": 0.0005, "loss": 1.4427964687347412, "step": 13635 }, { "epoch": 1.5866883872469164, "grad_norm": 1.352016568183899, "learning_rate": 0.0005, "loss": 2.2911570072174072, "step": 13636 }, { "epoch": 1.5868047474982547, "grad_norm": 0.9578656554222107, "learning_rate": 0.0005, "loss": 1.576540231704712, "step": 13637 }, { "epoch": 1.5869211077495926, "grad_norm": 1.155719518661499, "learning_rate": 0.0005, "loss": 2.299450159072876, "step": 13638 }, { "epoch": 1.587037468000931, "grad_norm": 0.9150202870368958, "learning_rate": 0.0005, "loss": 2.136373519897461, "step": 13639 }, { "epoch": 1.587153828252269, "grad_norm": 1.0122865438461304, "learning_rate": 0.0005, "loss": 1.6995220184326172, "step": 13640 }, { "epoch": 1.5872701885036071, "grad_norm": 1.081810712814331, "learning_rate": 0.0005, "loss": 2.142754554748535, "step": 13641 }, { "epoch": 1.5873865487549454, "grad_norm": 0.8908980488777161, "learning_rate": 0.0005, "loss": 1.859694242477417, "step": 13642 }, { "epoch": 1.5875029090062833, "grad_norm": 0.6803258657455444, "learning_rate": 0.0005, "loss": 1.1995265483856201, "step": 13643 }, { "epoch": 1.5876192692576216, "grad_norm": 0.7576773762702942, "learning_rate": 0.0005, "loss": 1.5308666229248047, "step": 13644 }, { "epoch": 1.5877356295089597, "grad_norm": 1.135918140411377, "learning_rate": 0.0005, "loss": 2.545844793319702, "step": 13645 }, { "epoch": 1.5878519897602978, "grad_norm": 1.0039489269256592, "learning_rate": 0.0005, "loss": 1.8207699060440063, "step": 13646 }, { "epoch": 1.5879683500116362, "grad_norm": 1.3061704635620117, "learning_rate": 0.0005, "loss": 1.9272620677947998, "step": 13647 }, { "epoch": 1.588084710262974, "grad_norm": 0.8807373046875, "learning_rate": 0.0005, "loss": 2.015143871307373, "step": 13648 }, { "epoch": 1.5882010705143124, "grad_norm": 1.9150480031967163, "learning_rate": 0.0005, "loss": 1.5168102979660034, "step": 13649 }, { "epoch": 1.5883174307656505, "grad_norm": 1.339318037033081, "learning_rate": 0.0005, "loss": 1.9729944467544556, "step": 13650 }, { "epoch": 1.5884337910169886, "grad_norm": 0.9930334091186523, "learning_rate": 0.0005, "loss": 1.4810888767242432, "step": 13651 }, { "epoch": 1.5885501512683269, "grad_norm": 1.5951118469238281, "learning_rate": 0.0005, "loss": 1.8556005954742432, "step": 13652 }, { "epoch": 1.5886665115196648, "grad_norm": 1.0674678087234497, "learning_rate": 0.0005, "loss": 1.9072648286819458, "step": 13653 }, { "epoch": 1.588782871771003, "grad_norm": 1.5152885913848877, "learning_rate": 0.0005, "loss": 1.7820398807525635, "step": 13654 }, { "epoch": 1.5888992320223412, "grad_norm": 1.0423150062561035, "learning_rate": 0.0005, "loss": 1.9942948818206787, "step": 13655 }, { "epoch": 1.5890155922736793, "grad_norm": 0.7510014176368713, "learning_rate": 0.0005, "loss": 1.265350103378296, "step": 13656 }, { "epoch": 1.5891319525250176, "grad_norm": 1.3093154430389404, "learning_rate": 0.0005, "loss": 1.5291144847869873, "step": 13657 }, { "epoch": 1.5892483127763555, "grad_norm": 1.345259189605713, "learning_rate": 0.0005, "loss": 2.059067726135254, "step": 13658 }, { "epoch": 1.5893646730276938, "grad_norm": 1.5479379892349243, "learning_rate": 0.0005, "loss": 2.5734939575195312, "step": 13659 }, { "epoch": 1.589481033279032, "grad_norm": 1.6867715120315552, "learning_rate": 0.0005, "loss": 1.9178791046142578, "step": 13660 }, { "epoch": 1.58959739353037, "grad_norm": 1.2108269929885864, "learning_rate": 0.0005, "loss": 2.2816414833068848, "step": 13661 }, { "epoch": 1.5897137537817083, "grad_norm": 0.7574223279953003, "learning_rate": 0.0005, "loss": 1.4767409563064575, "step": 13662 }, { "epoch": 1.5898301140330462, "grad_norm": 1.1928337812423706, "learning_rate": 0.0005, "loss": 2.1002719402313232, "step": 13663 }, { "epoch": 1.5899464742843845, "grad_norm": 1.4887076616287231, "learning_rate": 0.0005, "loss": 2.3101963996887207, "step": 13664 }, { "epoch": 1.5900628345357226, "grad_norm": 0.9076827168464661, "learning_rate": 0.0005, "loss": 1.6571509838104248, "step": 13665 }, { "epoch": 1.5901791947870607, "grad_norm": 1.2041295766830444, "learning_rate": 0.0005, "loss": 1.8717124462127686, "step": 13666 }, { "epoch": 1.590295555038399, "grad_norm": 1.4091386795043945, "learning_rate": 0.0005, "loss": 2.1527419090270996, "step": 13667 }, { "epoch": 1.590411915289737, "grad_norm": 0.8802719116210938, "learning_rate": 0.0005, "loss": 2.080130100250244, "step": 13668 }, { "epoch": 1.5905282755410752, "grad_norm": 3.9222517013549805, "learning_rate": 0.0005, "loss": 1.6942416429519653, "step": 13669 }, { "epoch": 1.5906446357924133, "grad_norm": 1.8787496089935303, "learning_rate": 0.0005, "loss": 2.2048726081848145, "step": 13670 }, { "epoch": 1.5907609960437514, "grad_norm": 0.8119310140609741, "learning_rate": 0.0005, "loss": 1.519181251525879, "step": 13671 }, { "epoch": 1.5908773562950898, "grad_norm": 1.308310627937317, "learning_rate": 0.0005, "loss": 2.212832450866699, "step": 13672 }, { "epoch": 1.5909937165464276, "grad_norm": 0.8698952794075012, "learning_rate": 0.0005, "loss": 1.384289026260376, "step": 13673 }, { "epoch": 1.591110076797766, "grad_norm": 1.0475504398345947, "learning_rate": 0.0005, "loss": 1.4735013246536255, "step": 13674 }, { "epoch": 1.591226437049104, "grad_norm": 3.827080011367798, "learning_rate": 0.0005, "loss": 2.5353429317474365, "step": 13675 }, { "epoch": 1.5913427973004421, "grad_norm": 1.0379321575164795, "learning_rate": 0.0005, "loss": 1.9746794700622559, "step": 13676 }, { "epoch": 1.5914591575517805, "grad_norm": 1.1309641599655151, "learning_rate": 0.0005, "loss": 2.0007529258728027, "step": 13677 }, { "epoch": 1.5915755178031183, "grad_norm": 0.9778162837028503, "learning_rate": 0.0005, "loss": 1.3965827226638794, "step": 13678 }, { "epoch": 1.5916918780544567, "grad_norm": 1.1616963148117065, "learning_rate": 0.0005, "loss": 1.9029849767684937, "step": 13679 }, { "epoch": 1.5918082383057948, "grad_norm": 1.1734188795089722, "learning_rate": 0.0005, "loss": 2.820044994354248, "step": 13680 }, { "epoch": 1.5919245985571329, "grad_norm": 1.6708780527114868, "learning_rate": 0.0005, "loss": 2.569861888885498, "step": 13681 }, { "epoch": 1.5920409588084712, "grad_norm": 1.424141764640808, "learning_rate": 0.0005, "loss": 1.9142515659332275, "step": 13682 }, { "epoch": 1.592157319059809, "grad_norm": 2.9091358184814453, "learning_rate": 0.0005, "loss": 1.7886073589324951, "step": 13683 }, { "epoch": 1.5922736793111474, "grad_norm": 1.094732403755188, "learning_rate": 0.0005, "loss": 1.789285659790039, "step": 13684 }, { "epoch": 1.5923900395624855, "grad_norm": 1.52585768699646, "learning_rate": 0.0005, "loss": 1.6226112842559814, "step": 13685 }, { "epoch": 1.5925063998138236, "grad_norm": 0.8151445388793945, "learning_rate": 0.0005, "loss": 1.9548933506011963, "step": 13686 }, { "epoch": 1.592622760065162, "grad_norm": 1.2990562915802002, "learning_rate": 0.0005, "loss": 1.7987678050994873, "step": 13687 }, { "epoch": 1.5927391203164998, "grad_norm": 1.0915263891220093, "learning_rate": 0.0005, "loss": 2.841392993927002, "step": 13688 }, { "epoch": 1.592855480567838, "grad_norm": 1.0747816562652588, "learning_rate": 0.0005, "loss": 1.304721713066101, "step": 13689 }, { "epoch": 1.5929718408191762, "grad_norm": 1.0821478366851807, "learning_rate": 0.0005, "loss": 1.6577690839767456, "step": 13690 }, { "epoch": 1.5930882010705143, "grad_norm": 1.1717536449432373, "learning_rate": 0.0005, "loss": 2.1134116649627686, "step": 13691 }, { "epoch": 1.5932045613218526, "grad_norm": 1.3162107467651367, "learning_rate": 0.0005, "loss": 1.9248113632202148, "step": 13692 }, { "epoch": 1.5933209215731905, "grad_norm": 1.5580312013626099, "learning_rate": 0.0005, "loss": 3.1002423763275146, "step": 13693 }, { "epoch": 1.5934372818245288, "grad_norm": 1.3681564331054688, "learning_rate": 0.0005, "loss": 1.2778180837631226, "step": 13694 }, { "epoch": 1.593553642075867, "grad_norm": 0.9700294137001038, "learning_rate": 0.0005, "loss": 2.132828712463379, "step": 13695 }, { "epoch": 1.593670002327205, "grad_norm": 2.939945936203003, "learning_rate": 0.0005, "loss": 2.7287285327911377, "step": 13696 }, { "epoch": 1.5937863625785431, "grad_norm": 0.2777889668941498, "learning_rate": 0.0005, "loss": 0.8767991662025452, "step": 13697 }, { "epoch": 1.5939027228298812, "grad_norm": 0.9206069111824036, "learning_rate": 0.0005, "loss": 2.4146616458892822, "step": 13698 }, { "epoch": 1.5940190830812195, "grad_norm": 1.2698626518249512, "learning_rate": 0.0005, "loss": 1.9262869358062744, "step": 13699 }, { "epoch": 1.5941354433325576, "grad_norm": 0.9174193143844604, "learning_rate": 0.0005, "loss": 1.4677488803863525, "step": 13700 }, { "epoch": 1.5942518035838957, "grad_norm": 0.6056800484657288, "learning_rate": 0.0005, "loss": 1.5696007013320923, "step": 13701 }, { "epoch": 1.5943681638352338, "grad_norm": 1.058434009552002, "learning_rate": 0.0005, "loss": 1.4872663021087646, "step": 13702 }, { "epoch": 1.594484524086572, "grad_norm": 0.9541731476783752, "learning_rate": 0.0005, "loss": 2.029743194580078, "step": 13703 }, { "epoch": 1.5946008843379103, "grad_norm": 0.817230224609375, "learning_rate": 0.0005, "loss": 2.425393581390381, "step": 13704 }, { "epoch": 1.5947172445892484, "grad_norm": 1.124206781387329, "learning_rate": 0.0005, "loss": 1.9621973037719727, "step": 13705 }, { "epoch": 1.5948336048405864, "grad_norm": 1.492575764656067, "learning_rate": 0.0005, "loss": 2.5862460136413574, "step": 13706 }, { "epoch": 1.5949499650919245, "grad_norm": 0.852711021900177, "learning_rate": 0.0005, "loss": 1.5264294147491455, "step": 13707 }, { "epoch": 1.5950663253432626, "grad_norm": 1.5263854265213013, "learning_rate": 0.0005, "loss": 1.6679205894470215, "step": 13708 }, { "epoch": 1.595182685594601, "grad_norm": 0.5904922485351562, "learning_rate": 0.0005, "loss": 1.435412883758545, "step": 13709 }, { "epoch": 1.595299045845939, "grad_norm": 1.2487438917160034, "learning_rate": 0.0005, "loss": 1.6251723766326904, "step": 13710 }, { "epoch": 1.5954154060972772, "grad_norm": 1.0179942846298218, "learning_rate": 0.0005, "loss": 1.9216443300247192, "step": 13711 }, { "epoch": 1.5955317663486153, "grad_norm": 2.1658928394317627, "learning_rate": 0.0005, "loss": 1.9984302520751953, "step": 13712 }, { "epoch": 1.5956481265999534, "grad_norm": 0.9349863529205322, "learning_rate": 0.0005, "loss": 2.1007885932922363, "step": 13713 }, { "epoch": 1.5957644868512917, "grad_norm": 0.8400223851203918, "learning_rate": 0.0005, "loss": 1.2336723804473877, "step": 13714 }, { "epoch": 1.5958808471026298, "grad_norm": 1.160927653312683, "learning_rate": 0.0005, "loss": 1.7200398445129395, "step": 13715 }, { "epoch": 1.5959972073539679, "grad_norm": 1.2097601890563965, "learning_rate": 0.0005, "loss": 2.454481840133667, "step": 13716 }, { "epoch": 1.596113567605306, "grad_norm": 1.0791404247283936, "learning_rate": 0.0005, "loss": 2.2248053550720215, "step": 13717 }, { "epoch": 1.596229927856644, "grad_norm": 1.2529176473617554, "learning_rate": 0.0005, "loss": 1.8321139812469482, "step": 13718 }, { "epoch": 1.5963462881079824, "grad_norm": 1.0447977781295776, "learning_rate": 0.0005, "loss": 2.193013906478882, "step": 13719 }, { "epoch": 1.5964626483593205, "grad_norm": 1.0957187414169312, "learning_rate": 0.0005, "loss": 1.4198957681655884, "step": 13720 }, { "epoch": 1.5965790086106586, "grad_norm": 1.3993005752563477, "learning_rate": 0.0005, "loss": 2.0919132232666016, "step": 13721 }, { "epoch": 1.5966953688619967, "grad_norm": 0.9770808219909668, "learning_rate": 0.0005, "loss": 2.378150224685669, "step": 13722 }, { "epoch": 1.5968117291133348, "grad_norm": 1.0633642673492432, "learning_rate": 0.0005, "loss": 1.8119432926177979, "step": 13723 }, { "epoch": 1.5969280893646731, "grad_norm": 0.8544294834136963, "learning_rate": 0.0005, "loss": 2.0936269760131836, "step": 13724 }, { "epoch": 1.5970444496160112, "grad_norm": 1.0431303977966309, "learning_rate": 0.0005, "loss": 1.654165267944336, "step": 13725 }, { "epoch": 1.5971608098673493, "grad_norm": 0.8987507224082947, "learning_rate": 0.0005, "loss": 2.0318872928619385, "step": 13726 }, { "epoch": 1.5972771701186874, "grad_norm": 1.0452146530151367, "learning_rate": 0.0005, "loss": 2.4020204544067383, "step": 13727 }, { "epoch": 1.5973935303700255, "grad_norm": 1.5796923637390137, "learning_rate": 0.0005, "loss": 1.9092668294906616, "step": 13728 }, { "epoch": 1.5975098906213638, "grad_norm": 1.015109658241272, "learning_rate": 0.0005, "loss": 1.662386178970337, "step": 13729 }, { "epoch": 1.597626250872702, "grad_norm": 0.7731889486312866, "learning_rate": 0.0005, "loss": 2.3449666500091553, "step": 13730 }, { "epoch": 1.59774261112404, "grad_norm": 1.229602575302124, "learning_rate": 0.0005, "loss": 1.7786632776260376, "step": 13731 }, { "epoch": 1.5978589713753781, "grad_norm": 0.7402636408805847, "learning_rate": 0.0005, "loss": 1.5687274932861328, "step": 13732 }, { "epoch": 1.5979753316267162, "grad_norm": 1.5147273540496826, "learning_rate": 0.0005, "loss": 2.345376968383789, "step": 13733 }, { "epoch": 1.5980916918780546, "grad_norm": 0.9290920495986938, "learning_rate": 0.0005, "loss": 1.6624902486801147, "step": 13734 }, { "epoch": 1.5982080521293927, "grad_norm": 1.1954474449157715, "learning_rate": 0.0005, "loss": 1.7163258790969849, "step": 13735 }, { "epoch": 1.5983244123807308, "grad_norm": 0.8425235748291016, "learning_rate": 0.0005, "loss": 1.7377973794937134, "step": 13736 }, { "epoch": 1.5984407726320689, "grad_norm": 0.9694026112556458, "learning_rate": 0.0005, "loss": 1.687091588973999, "step": 13737 }, { "epoch": 1.598557132883407, "grad_norm": 1.496996283531189, "learning_rate": 0.0005, "loss": 1.840097188949585, "step": 13738 }, { "epoch": 1.5986734931347453, "grad_norm": 1.4550142288208008, "learning_rate": 0.0005, "loss": 2.824967384338379, "step": 13739 }, { "epoch": 1.5987898533860831, "grad_norm": 1.0009722709655762, "learning_rate": 0.0005, "loss": 1.678128957748413, "step": 13740 }, { "epoch": 1.5989062136374215, "grad_norm": 1.0680501461029053, "learning_rate": 0.0005, "loss": 1.1603331565856934, "step": 13741 }, { "epoch": 1.5990225738887596, "grad_norm": 1.338678002357483, "learning_rate": 0.0005, "loss": 2.491237163543701, "step": 13742 }, { "epoch": 1.5991389341400977, "grad_norm": 0.7676107883453369, "learning_rate": 0.0005, "loss": 1.7055304050445557, "step": 13743 }, { "epoch": 1.599255294391436, "grad_norm": 1.5051746368408203, "learning_rate": 0.0005, "loss": 2.024589776992798, "step": 13744 }, { "epoch": 1.5993716546427739, "grad_norm": 0.9062339663505554, "learning_rate": 0.0005, "loss": 2.4476563930511475, "step": 13745 }, { "epoch": 1.5994880148941122, "grad_norm": 1.3239541053771973, "learning_rate": 0.0005, "loss": 2.0415148735046387, "step": 13746 }, { "epoch": 1.5996043751454503, "grad_norm": 0.7101460099220276, "learning_rate": 0.0005, "loss": 1.3892384767532349, "step": 13747 }, { "epoch": 1.5997207353967884, "grad_norm": 1.1826491355895996, "learning_rate": 0.0005, "loss": 2.3704121112823486, "step": 13748 }, { "epoch": 1.5998370956481267, "grad_norm": 1.3330717086791992, "learning_rate": 0.0005, "loss": 2.0409092903137207, "step": 13749 }, { "epoch": 1.5999534558994646, "grad_norm": 1.455106496810913, "learning_rate": 0.0005, "loss": 2.2342867851257324, "step": 13750 }, { "epoch": 1.600069816150803, "grad_norm": 0.8820761442184448, "learning_rate": 0.0005, "loss": 1.5776278972625732, "step": 13751 }, { "epoch": 1.600186176402141, "grad_norm": 1.5457113981246948, "learning_rate": 0.0005, "loss": 1.9203499555587769, "step": 13752 }, { "epoch": 1.600302536653479, "grad_norm": 1.9009453058242798, "learning_rate": 0.0004999998955638956, "loss": 2.170713424682617, "step": 13753 }, { "epoch": 1.6004188969048174, "grad_norm": 1.1274635791778564, "learning_rate": 0.0004999995822556695, "loss": 3.116755485534668, "step": 13754 }, { "epoch": 1.6005352571561553, "grad_norm": 0.8843479156494141, "learning_rate": 0.0004999990600755836, "loss": 1.8797049522399902, "step": 13755 }, { "epoch": 1.6006516174074936, "grad_norm": 1.0305612087249756, "learning_rate": 0.000499998329024074, "loss": 2.2457923889160156, "step": 13756 }, { "epoch": 1.6007679776588317, "grad_norm": 1.4718191623687744, "learning_rate": 0.0004999973891017518, "loss": 1.5776312351226807, "step": 13757 }, { "epoch": 1.6008843379101698, "grad_norm": 1.5485492944717407, "learning_rate": 0.000499996240309402, "loss": 1.7356200218200684, "step": 13758 }, { "epoch": 1.6010006981615081, "grad_norm": 1.0752686262130737, "learning_rate": 0.0004999948826479845, "loss": 2.0230095386505127, "step": 13759 }, { "epoch": 1.601117058412846, "grad_norm": 1.4176987409591675, "learning_rate": 0.0004999933161186336, "loss": 1.6260510683059692, "step": 13760 }, { "epoch": 1.6012334186641843, "grad_norm": 0.8368831872940063, "learning_rate": 0.000499991540722658, "loss": 1.327885389328003, "step": 13761 }, { "epoch": 1.6013497789155224, "grad_norm": 0.6666719317436218, "learning_rate": 0.0004999895564615413, "loss": 1.3477551937103271, "step": 13762 }, { "epoch": 1.6014661391668605, "grad_norm": 2.6038215160369873, "learning_rate": 0.0004999873633369411, "loss": 2.028392791748047, "step": 13763 }, { "epoch": 1.6015824994181989, "grad_norm": 1.2957332134246826, "learning_rate": 0.0004999849613506898, "loss": 1.8295645713806152, "step": 13764 }, { "epoch": 1.6016988596695367, "grad_norm": 1.0514419078826904, "learning_rate": 0.0004999823505047943, "loss": 2.3821702003479004, "step": 13765 }, { "epoch": 1.601815219920875, "grad_norm": 1.2592021226882935, "learning_rate": 0.0004999795308014359, "loss": 1.975005030632019, "step": 13766 }, { "epoch": 1.6019315801722132, "grad_norm": 1.1983554363250732, "learning_rate": 0.0004999765022429704, "loss": 2.097935199737549, "step": 13767 }, { "epoch": 1.6020479404235513, "grad_norm": 1.0052602291107178, "learning_rate": 0.0004999732648319281, "loss": 1.7703146934509277, "step": 13768 }, { "epoch": 1.6021643006748896, "grad_norm": 0.8837888836860657, "learning_rate": 0.0004999698185710139, "loss": 1.34297776222229, "step": 13769 }, { "epoch": 1.6022806609262275, "grad_norm": 1.203033685684204, "learning_rate": 0.000499966163463107, "loss": 2.696378231048584, "step": 13770 }, { "epoch": 1.6023970211775658, "grad_norm": 0.908443808555603, "learning_rate": 0.0004999622995112612, "loss": 1.4622117280960083, "step": 13771 }, { "epoch": 1.6025133814289039, "grad_norm": 1.1068843603134155, "learning_rate": 0.000499958226718705, "loss": 1.8841848373413086, "step": 13772 }, { "epoch": 1.602629741680242, "grad_norm": 1.2074716091156006, "learning_rate": 0.000499953945088841, "loss": 2.155503749847412, "step": 13773 }, { "epoch": 1.6027461019315803, "grad_norm": 1.476547122001648, "learning_rate": 0.0004999494546252465, "loss": 2.8238182067871094, "step": 13774 }, { "epoch": 1.6028624621829182, "grad_norm": 1.0080009698867798, "learning_rate": 0.0004999447553316731, "loss": 1.7569546699523926, "step": 13775 }, { "epoch": 1.6029788224342565, "grad_norm": 1.2830058336257935, "learning_rate": 0.0004999398472120472, "loss": 1.9442417621612549, "step": 13776 }, { "epoch": 1.6030951826855946, "grad_norm": 1.2628271579742432, "learning_rate": 0.0004999347302704695, "loss": 1.414507269859314, "step": 13777 }, { "epoch": 1.6032115429369327, "grad_norm": 1.119333028793335, "learning_rate": 0.0004999294045112149, "loss": 2.391479015350342, "step": 13778 }, { "epoch": 1.603327903188271, "grad_norm": 1.5995042324066162, "learning_rate": 0.0004999238699387331, "loss": 2.7439393997192383, "step": 13779 }, { "epoch": 1.6034442634396089, "grad_norm": 1.121342420578003, "learning_rate": 0.0004999181265576484, "loss": 2.0569522380828857, "step": 13780 }, { "epoch": 1.6035606236909472, "grad_norm": 1.1909499168395996, "learning_rate": 0.000499912174372759, "loss": 1.6910480260849, "step": 13781 }, { "epoch": 1.6036769839422853, "grad_norm": 1.425291657447815, "learning_rate": 0.0004999060133890382, "loss": 2.3401472568511963, "step": 13782 }, { "epoch": 1.6037933441936234, "grad_norm": 1.3736814260482788, "learning_rate": 0.0004998996436116332, "loss": 1.9501562118530273, "step": 13783 }, { "epoch": 1.6039097044449617, "grad_norm": 0.9801567792892456, "learning_rate": 0.0004998930650458661, "loss": 1.914605975151062, "step": 13784 }, { "epoch": 1.6040260646962996, "grad_norm": 1.9068517684936523, "learning_rate": 0.0004998862776972329, "loss": 2.452497720718384, "step": 13785 }, { "epoch": 1.604142424947638, "grad_norm": 1.5373865365982056, "learning_rate": 0.0004998792815714046, "loss": 2.4497506618499756, "step": 13786 }, { "epoch": 1.604258785198976, "grad_norm": 1.5813676118850708, "learning_rate": 0.0004998720766742264, "loss": 2.2214646339416504, "step": 13787 }, { "epoch": 1.6043751454503141, "grad_norm": 1.8033570051193237, "learning_rate": 0.0004998646630117177, "loss": 1.9340437650680542, "step": 13788 }, { "epoch": 1.6044915057016524, "grad_norm": 1.2155004739761353, "learning_rate": 0.0004998570405900728, "loss": 1.9867026805877686, "step": 13789 }, { "epoch": 1.6046078659529903, "grad_norm": 1.3161414861679077, "learning_rate": 0.0004998492094156599, "loss": 2.2802934646606445, "step": 13790 }, { "epoch": 1.6047242262043286, "grad_norm": 1.2181601524353027, "learning_rate": 0.0004998411694950219, "loss": 2.6662425994873047, "step": 13791 }, { "epoch": 1.6048405864556667, "grad_norm": 0.9522082805633545, "learning_rate": 0.0004998329208348762, "loss": 1.7054729461669922, "step": 13792 }, { "epoch": 1.6049569467070048, "grad_norm": 1.2250511646270752, "learning_rate": 0.0004998244634421145, "loss": 2.2015576362609863, "step": 13793 }, { "epoch": 1.6050733069583432, "grad_norm": 1.2327276468276978, "learning_rate": 0.0004998157973238027, "loss": 1.6234068870544434, "step": 13794 }, { "epoch": 1.605189667209681, "grad_norm": 1.568909764289856, "learning_rate": 0.0004998069224871811, "loss": 2.367802381515503, "step": 13795 }, { "epoch": 1.6053060274610194, "grad_norm": 1.009468913078308, "learning_rate": 0.000499797838939665, "loss": 1.8305785655975342, "step": 13796 }, { "epoch": 1.6054223877123575, "grad_norm": 1.1408958435058594, "learning_rate": 0.0004997885466888432, "loss": 1.843991994857788, "step": 13797 }, { "epoch": 1.6055387479636956, "grad_norm": 0.8597704172134399, "learning_rate": 0.0004997790457424795, "loss": 1.5269880294799805, "step": 13798 }, { "epoch": 1.6056551082150339, "grad_norm": 0.9878445863723755, "learning_rate": 0.0004997693361085116, "loss": 1.5740238428115845, "step": 13799 }, { "epoch": 1.6057714684663718, "grad_norm": 0.9971354007720947, "learning_rate": 0.0004997594177950519, "loss": 1.5456706285476685, "step": 13800 }, { "epoch": 1.60588782871771, "grad_norm": 1.3633499145507812, "learning_rate": 0.0004997492908103872, "loss": 1.881948471069336, "step": 13801 }, { "epoch": 1.6060041889690482, "grad_norm": 0.7964608073234558, "learning_rate": 0.0004997389551629783, "loss": 1.497289776802063, "step": 13802 }, { "epoch": 1.6061205492203863, "grad_norm": 0.7225955128669739, "learning_rate": 0.0004997284108614606, "loss": 1.6010310649871826, "step": 13803 }, { "epoch": 1.6062369094717246, "grad_norm": 0.967455267906189, "learning_rate": 0.0004997176579146437, "loss": 2.678199291229248, "step": 13804 }, { "epoch": 1.6063532697230625, "grad_norm": 1.1184977293014526, "learning_rate": 0.0004997066963315115, "loss": 1.8439819812774658, "step": 13805 }, { "epoch": 1.6064696299744008, "grad_norm": 0.9778763055801392, "learning_rate": 0.0004996955261212224, "loss": 1.3394054174423218, "step": 13806 }, { "epoch": 1.606585990225739, "grad_norm": 1.0901861190795898, "learning_rate": 0.0004996841472931088, "loss": 2.1293678283691406, "step": 13807 }, { "epoch": 1.606702350477077, "grad_norm": 1.194700837135315, "learning_rate": 0.0004996725598566779, "loss": 2.357316493988037, "step": 13808 }, { "epoch": 1.6068187107284153, "grad_norm": 1.0532327890396118, "learning_rate": 0.0004996607638216107, "loss": 2.14741849899292, "step": 13809 }, { "epoch": 1.6069350709797532, "grad_norm": 1.1005990505218506, "learning_rate": 0.0004996487591977626, "loss": 1.9792983531951904, "step": 13810 }, { "epoch": 1.6070514312310915, "grad_norm": 1.0032579898834229, "learning_rate": 0.0004996365459951635, "loss": 1.6337319612503052, "step": 13811 }, { "epoch": 1.6071677914824296, "grad_norm": 1.0751641988754272, "learning_rate": 0.0004996241242240172, "loss": 2.105682373046875, "step": 13812 }, { "epoch": 1.6072841517337677, "grad_norm": 1.0008538961410522, "learning_rate": 0.0004996114938947021, "loss": 2.3156960010528564, "step": 13813 }, { "epoch": 1.607400511985106, "grad_norm": 0.7026248574256897, "learning_rate": 0.0004995986550177705, "loss": 1.5728180408477783, "step": 13814 }, { "epoch": 1.607516872236444, "grad_norm": 1.1633886098861694, "learning_rate": 0.0004995856076039493, "loss": 1.647719144821167, "step": 13815 }, { "epoch": 1.6076332324877822, "grad_norm": 1.0637205839157104, "learning_rate": 0.0004995723516641395, "loss": 2.6012558937072754, "step": 13816 }, { "epoch": 1.6077495927391203, "grad_norm": 0.9985119700431824, "learning_rate": 0.0004995588872094163, "loss": 1.315373182296753, "step": 13817 }, { "epoch": 1.6078659529904584, "grad_norm": 1.1301532983779907, "learning_rate": 0.0004995452142510288, "loss": 1.6234893798828125, "step": 13818 }, { "epoch": 1.6079823132417967, "grad_norm": 0.5561713576316833, "learning_rate": 0.000499531332800401, "loss": 1.2828434705734253, "step": 13819 }, { "epoch": 1.6080986734931346, "grad_norm": 0.8728594779968262, "learning_rate": 0.0004995172428691305, "loss": 1.868134617805481, "step": 13820 }, { "epoch": 1.608215033744473, "grad_norm": 3.517822265625, "learning_rate": 0.0004995029444689893, "loss": 1.57025146484375, "step": 13821 }, { "epoch": 1.608331393995811, "grad_norm": 0.9963717460632324, "learning_rate": 0.0004994884376119236, "loss": 1.578580379486084, "step": 13822 }, { "epoch": 1.6084477542471491, "grad_norm": 1.4564836025238037, "learning_rate": 0.0004994737223100536, "loss": 2.133336067199707, "step": 13823 }, { "epoch": 1.6085641144984875, "grad_norm": 1.0052179098129272, "learning_rate": 0.0004994587985756741, "loss": 1.718192458152771, "step": 13824 }, { "epoch": 1.6086804747498253, "grad_norm": 1.260582447052002, "learning_rate": 0.0004994436664212532, "loss": 1.2767902612686157, "step": 13825 }, { "epoch": 1.6087968350011637, "grad_norm": 1.3644686937332153, "learning_rate": 0.000499428325859434, "loss": 1.9657282829284668, "step": 13826 }, { "epoch": 1.6089131952525018, "grad_norm": 1.4954109191894531, "learning_rate": 0.0004994127769030332, "loss": 1.7424814701080322, "step": 13827 }, { "epoch": 1.6090295555038399, "grad_norm": 1.6553936004638672, "learning_rate": 0.0004993970195650419, "loss": 1.5866260528564453, "step": 13828 }, { "epoch": 1.6091459157551782, "grad_norm": 0.9524872899055481, "learning_rate": 0.0004993810538586252, "loss": 2.6143975257873535, "step": 13829 }, { "epoch": 1.609262276006516, "grad_norm": 1.4699722528457642, "learning_rate": 0.0004993648797971222, "loss": 2.470602512359619, "step": 13830 }, { "epoch": 1.6093786362578544, "grad_norm": 0.89785236120224, "learning_rate": 0.0004993484973940461, "loss": 1.9171255826950073, "step": 13831 }, { "epoch": 1.6094949965091925, "grad_norm": 1.070615291595459, "learning_rate": 0.0004993319066630843, "loss": 2.3244056701660156, "step": 13832 }, { "epoch": 1.6096113567605306, "grad_norm": 1.6762502193450928, "learning_rate": 0.0004993151076180982, "loss": 1.9249067306518555, "step": 13833 }, { "epoch": 1.609727717011869, "grad_norm": 0.8752719759941101, "learning_rate": 0.000499298100273123, "loss": 1.4676762819290161, "step": 13834 }, { "epoch": 1.6098440772632068, "grad_norm": 0.8267964720726013, "learning_rate": 0.0004992808846423684, "loss": 1.3482353687286377, "step": 13835 }, { "epoch": 1.609960437514545, "grad_norm": 2.7343974113464355, "learning_rate": 0.0004992634607402178, "loss": 1.785365343093872, "step": 13836 }, { "epoch": 1.6100767977658832, "grad_norm": 1.8847246170043945, "learning_rate": 0.0004992458285812287, "loss": 3.470135450363159, "step": 13837 }, { "epoch": 1.6101931580172213, "grad_norm": 1.023834228515625, "learning_rate": 0.0004992279881801324, "loss": 1.636373519897461, "step": 13838 }, { "epoch": 1.6103095182685596, "grad_norm": 1.031051516532898, "learning_rate": 0.0004992099395518346, "loss": 1.2979750633239746, "step": 13839 }, { "epoch": 1.6104258785198975, "grad_norm": 1.1309754848480225, "learning_rate": 0.0004991916827114145, "loss": 1.8656572103500366, "step": 13840 }, { "epoch": 1.6105422387712358, "grad_norm": 1.3273065090179443, "learning_rate": 0.0004991732176741256, "loss": 2.003162384033203, "step": 13841 }, { "epoch": 1.610658599022574, "grad_norm": 1.2953808307647705, "learning_rate": 0.0004991545444553952, "loss": 1.9642417430877686, "step": 13842 }, { "epoch": 1.610774959273912, "grad_norm": 1.0087802410125732, "learning_rate": 0.0004991356630708245, "loss": 1.734212875366211, "step": 13843 }, { "epoch": 1.6108913195252503, "grad_norm": 1.3990557193756104, "learning_rate": 0.000499116573536189, "loss": 2.2029361724853516, "step": 13844 }, { "epoch": 1.6110076797765882, "grad_norm": 1.111480712890625, "learning_rate": 0.0004990972758674375, "loss": 1.9733648300170898, "step": 13845 }, { "epoch": 1.6111240400279265, "grad_norm": 0.983731210231781, "learning_rate": 0.000499077770080693, "loss": 1.9551591873168945, "step": 13846 }, { "epoch": 1.6112404002792646, "grad_norm": 1.171180248260498, "learning_rate": 0.0004990580561922524, "loss": 1.6321388483047485, "step": 13847 }, { "epoch": 1.6113567605306027, "grad_norm": 0.8465421199798584, "learning_rate": 0.0004990381342185865, "loss": 1.3996169567108154, "step": 13848 }, { "epoch": 1.611473120781941, "grad_norm": 0.8876653909683228, "learning_rate": 0.0004990180041763398, "loss": 1.453972578048706, "step": 13849 }, { "epoch": 1.611589481033279, "grad_norm": 1.3733960390090942, "learning_rate": 0.0004989976660823307, "loss": 2.3646864891052246, "step": 13850 }, { "epoch": 1.6117058412846172, "grad_norm": 1.2102916240692139, "learning_rate": 0.0004989771199535517, "loss": 2.072526216506958, "step": 13851 }, { "epoch": 1.6118222015359553, "grad_norm": 3.427124500274658, "learning_rate": 0.0004989563658071686, "loss": 2.092802047729492, "step": 13852 }, { "epoch": 1.6119385617872934, "grad_norm": 0.9479905962944031, "learning_rate": 0.0004989354036605213, "loss": 1.9323903322219849, "step": 13853 }, { "epoch": 1.6120549220386318, "grad_norm": 1.1118500232696533, "learning_rate": 0.0004989142335311235, "loss": 2.191297769546509, "step": 13854 }, { "epoch": 1.6121712822899696, "grad_norm": 1.8425500392913818, "learning_rate": 0.0004988928554366625, "loss": 1.7336792945861816, "step": 13855 }, { "epoch": 1.612287642541308, "grad_norm": 4.649609088897705, "learning_rate": 0.0004988712693949997, "loss": 2.077670097351074, "step": 13856 }, { "epoch": 1.612404002792646, "grad_norm": 1.495671033859253, "learning_rate": 0.0004988494754241696, "loss": 1.896170973777771, "step": 13857 }, { "epoch": 1.6125203630439842, "grad_norm": 1.2934767007827759, "learning_rate": 0.0004988274735423814, "loss": 2.0646629333496094, "step": 13858 }, { "epoch": 1.6126367232953225, "grad_norm": 0.8915260434150696, "learning_rate": 0.0004988052637680168, "loss": 1.4304111003875732, "step": 13859 }, { "epoch": 1.6127530835466604, "grad_norm": 1.2687339782714844, "learning_rate": 0.0004987828461196322, "loss": 2.0028460025787354, "step": 13860 }, { "epoch": 1.6128694437979987, "grad_norm": 1.0652285814285278, "learning_rate": 0.0004987602206159571, "loss": 2.0601696968078613, "step": 13861 }, { "epoch": 1.6129858040493368, "grad_norm": 1.6638331413269043, "learning_rate": 0.0004987373872758951, "loss": 2.5497467517852783, "step": 13862 }, { "epoch": 1.6131021643006749, "grad_norm": 1.0010865926742554, "learning_rate": 0.000498714346118523, "loss": 1.9158538579940796, "step": 13863 }, { "epoch": 1.613218524552013, "grad_norm": 1.8055354356765747, "learning_rate": 0.0004986910971630915, "loss": 2.3092823028564453, "step": 13864 }, { "epoch": 1.613334884803351, "grad_norm": 0.8627516031265259, "learning_rate": 0.0004986676404290248, "loss": 1.3642916679382324, "step": 13865 }, { "epoch": 1.6134512450546894, "grad_norm": 1.6586923599243164, "learning_rate": 0.0004986439759359207, "loss": 1.6753742694854736, "step": 13866 }, { "epoch": 1.6135676053060275, "grad_norm": 0.5126972794532776, "learning_rate": 0.0004986201037035508, "loss": 1.3140158653259277, "step": 13867 }, { "epoch": 1.6136839655573656, "grad_norm": 1.2386075258255005, "learning_rate": 0.0004985960237518599, "loss": 2.1960525512695312, "step": 13868 }, { "epoch": 1.6138003258087037, "grad_norm": 1.2829378843307495, "learning_rate": 0.0004985717361009666, "loss": 2.026902675628662, "step": 13869 }, { "epoch": 1.6139166860600418, "grad_norm": 1.0050867795944214, "learning_rate": 0.000498547240771163, "loss": 1.7409652471542358, "step": 13870 }, { "epoch": 1.6140330463113801, "grad_norm": 0.7082884311676025, "learning_rate": 0.0004985225377829146, "loss": 1.4520334005355835, "step": 13871 }, { "epoch": 1.6141494065627182, "grad_norm": 1.3979347944259644, "learning_rate": 0.0004984976271568605, "loss": 2.036203384399414, "step": 13872 }, { "epoch": 1.6142657668140563, "grad_norm": 1.297316312789917, "learning_rate": 0.0004984725089138132, "loss": 2.382453203201294, "step": 13873 }, { "epoch": 1.6143821270653944, "grad_norm": 1.115417242050171, "learning_rate": 0.0004984471830747589, "loss": 1.9977864027023315, "step": 13874 }, { "epoch": 1.6144984873167325, "grad_norm": 1.422011375427246, "learning_rate": 0.0004984216496608569, "loss": 1.0143049955368042, "step": 13875 }, { "epoch": 1.6146148475680708, "grad_norm": 1.8082146644592285, "learning_rate": 0.0004983959086934399, "loss": 2.3701677322387695, "step": 13876 }, { "epoch": 1.614731207819409, "grad_norm": 1.7867460250854492, "learning_rate": 0.0004983699601940147, "loss": 1.6029253005981445, "step": 13877 }, { "epoch": 1.614847568070747, "grad_norm": 1.274900197982788, "learning_rate": 0.0004983438041842605, "loss": 1.642620325088501, "step": 13878 }, { "epoch": 1.6149639283220851, "grad_norm": 1.1483368873596191, "learning_rate": 0.0004983174406860306, "loss": 1.485243558883667, "step": 13879 }, { "epoch": 1.6150802885734232, "grad_norm": 1.0251237154006958, "learning_rate": 0.0004982908697213513, "loss": 2.012554883956909, "step": 13880 }, { "epoch": 1.6151966488247615, "grad_norm": 1.4550855159759521, "learning_rate": 0.0004982640913124224, "loss": 1.9992486238479614, "step": 13881 }, { "epoch": 1.6153130090760996, "grad_norm": 0.9776541590690613, "learning_rate": 0.0004982371054816169, "loss": 2.4775750637054443, "step": 13882 }, { "epoch": 1.6154293693274377, "grad_norm": 0.9606998562812805, "learning_rate": 0.0004982099122514813, "loss": 1.8119524717330933, "step": 13883 }, { "epoch": 1.6155457295787758, "grad_norm": 1.0420469045639038, "learning_rate": 0.000498182511644735, "loss": 1.6175137758255005, "step": 13884 }, { "epoch": 1.615662089830114, "grad_norm": 0.9484551548957825, "learning_rate": 0.0004981549036842712, "loss": 2.6158578395843506, "step": 13885 }, { "epoch": 1.6157784500814523, "grad_norm": 1.5558544397354126, "learning_rate": 0.0004981270883931559, "loss": 2.4882941246032715, "step": 13886 }, { "epoch": 1.6158948103327904, "grad_norm": 0.8440943360328674, "learning_rate": 0.0004980990657946283, "loss": 1.7475814819335938, "step": 13887 }, { "epoch": 1.6160111705841285, "grad_norm": 1.2704384326934814, "learning_rate": 0.0004980708359121012, "loss": 2.1444597244262695, "step": 13888 }, { "epoch": 1.6161275308354666, "grad_norm": 1.2150341272354126, "learning_rate": 0.0004980423987691604, "loss": 1.6493316888809204, "step": 13889 }, { "epoch": 1.6162438910868047, "grad_norm": 2.7335777282714844, "learning_rate": 0.0004980137543895645, "loss": 1.5085110664367676, "step": 13890 }, { "epoch": 1.616360251338143, "grad_norm": 1.278225302696228, "learning_rate": 0.0004979849027972459, "loss": 2.2156522274017334, "step": 13891 }, { "epoch": 1.616476611589481, "grad_norm": 1.039495587348938, "learning_rate": 0.0004979558440163095, "loss": 2.3253111839294434, "step": 13892 }, { "epoch": 1.6165929718408192, "grad_norm": 1.3115475177764893, "learning_rate": 0.0004979265780710338, "loss": 2.2596042156219482, "step": 13893 }, { "epoch": 1.6167093320921573, "grad_norm": 1.1940052509307861, "learning_rate": 0.00049789710498587, "loss": 1.8876338005065918, "step": 13894 }, { "epoch": 1.6168256923434954, "grad_norm": 1.3683041334152222, "learning_rate": 0.0004978674247854428, "loss": 2.0397822856903076, "step": 13895 }, { "epoch": 1.6169420525948337, "grad_norm": 1.008244514465332, "learning_rate": 0.0004978375374945493, "loss": 1.9825503826141357, "step": 13896 }, { "epoch": 1.6170584128461718, "grad_norm": 1.1827406883239746, "learning_rate": 0.0004978074431381603, "loss": 2.321448802947998, "step": 13897 }, { "epoch": 1.61717477309751, "grad_norm": 0.8065472841262817, "learning_rate": 0.0004977771417414192, "loss": 1.829028606414795, "step": 13898 }, { "epoch": 1.617291133348848, "grad_norm": 1.3446319103240967, "learning_rate": 0.0004977466333296425, "loss": 2.3363611698150635, "step": 13899 }, { "epoch": 1.617407493600186, "grad_norm": 1.2604281902313232, "learning_rate": 0.0004977159179283196, "loss": 1.348332405090332, "step": 13900 }, { "epoch": 1.6175238538515244, "grad_norm": 0.9858368039131165, "learning_rate": 0.0004976849955631128, "loss": 1.2516744136810303, "step": 13901 }, { "epoch": 1.6176402141028625, "grad_norm": 0.940731406211853, "learning_rate": 0.0004976538662598576, "loss": 1.519946813583374, "step": 13902 }, { "epoch": 1.6177565743542006, "grad_norm": 0.6002929210662842, "learning_rate": 0.000497622530044562, "loss": 1.266740083694458, "step": 13903 }, { "epoch": 1.6178729346055387, "grad_norm": 1.110445261001587, "learning_rate": 0.0004975909869434072, "loss": 1.9885475635528564, "step": 13904 }, { "epoch": 1.6179892948568768, "grad_norm": 2.752169609069824, "learning_rate": 0.000497559236982747, "loss": 1.559030532836914, "step": 13905 }, { "epoch": 1.6181056551082151, "grad_norm": 0.8474787473678589, "learning_rate": 0.0004975272801891081, "loss": 1.83378005027771, "step": 13906 }, { "epoch": 1.618222015359553, "grad_norm": 1.0196640491485596, "learning_rate": 0.0004974951165891902, "loss": 2.366976499557495, "step": 13907 }, { "epoch": 1.6183383756108913, "grad_norm": 1.1420928239822388, "learning_rate": 0.0004974627462098654, "loss": 2.1234302520751953, "step": 13908 }, { "epoch": 1.6184547358622294, "grad_norm": 0.9638470411300659, "learning_rate": 0.0004974301690781791, "loss": 1.5680662393569946, "step": 13909 }, { "epoch": 1.6185710961135675, "grad_norm": 1.1153075695037842, "learning_rate": 0.000497397385221349, "loss": 2.0010764598846436, "step": 13910 }, { "epoch": 1.6186874563649059, "grad_norm": 0.9059368968009949, "learning_rate": 0.0004973643946667654, "loss": 2.3277535438537598, "step": 13911 }, { "epoch": 1.6188038166162437, "grad_norm": 1.2625303268432617, "learning_rate": 0.0004973311974419918, "loss": 2.0879616737365723, "step": 13912 }, { "epoch": 1.618920176867582, "grad_norm": 2.301234483718872, "learning_rate": 0.0004972977935747641, "loss": 2.064941883087158, "step": 13913 }, { "epoch": 1.6190365371189201, "grad_norm": 0.8024870157241821, "learning_rate": 0.0004972641830929908, "loss": 1.4870154857635498, "step": 13914 }, { "epoch": 1.6191528973702582, "grad_norm": 1.5223597288131714, "learning_rate": 0.0004972303660247532, "loss": 2.0748534202575684, "step": 13915 }, { "epoch": 1.6192692576215966, "grad_norm": 1.4013131856918335, "learning_rate": 0.0004971963423983049, "loss": 1.6721547842025757, "step": 13916 }, { "epoch": 1.6193856178729344, "grad_norm": 0.37785154581069946, "learning_rate": 0.0004971621122420724, "loss": 1.1546597480773926, "step": 13917 }, { "epoch": 1.6195019781242728, "grad_norm": 1.1295942068099976, "learning_rate": 0.0004971276755846545, "loss": 1.8323965072631836, "step": 13918 }, { "epoch": 1.6196183383756109, "grad_norm": 0.8218413591384888, "learning_rate": 0.0004970930324548228, "loss": 1.8130087852478027, "step": 13919 }, { "epoch": 1.619734698626949, "grad_norm": 0.6473281383514404, "learning_rate": 0.0004970581828815209, "loss": 1.6775987148284912, "step": 13920 }, { "epoch": 1.6198510588782873, "grad_norm": 0.9508517384529114, "learning_rate": 0.0004970231268938657, "loss": 1.4574196338653564, "step": 13921 }, { "epoch": 1.6199674191296252, "grad_norm": 1.1214393377304077, "learning_rate": 0.0004969878645211458, "loss": 2.5707578659057617, "step": 13922 }, { "epoch": 1.6200837793809635, "grad_norm": 0.9600701928138733, "learning_rate": 0.0004969523957928226, "loss": 1.1550209522247314, "step": 13923 }, { "epoch": 1.6202001396323016, "grad_norm": 0.9457128643989563, "learning_rate": 0.0004969167207385298, "loss": 1.3972349166870117, "step": 13924 }, { "epoch": 1.6203164998836397, "grad_norm": 0.4931940734386444, "learning_rate": 0.0004968808393880735, "loss": 1.0951216220855713, "step": 13925 }, { "epoch": 1.620432860134978, "grad_norm": 1.0546890497207642, "learning_rate": 0.0004968447517714322, "loss": 2.1148228645324707, "step": 13926 }, { "epoch": 1.6205492203863159, "grad_norm": 1.079877495765686, "learning_rate": 0.0004968084579187566, "loss": 2.2043607234954834, "step": 13927 }, { "epoch": 1.6206655806376542, "grad_norm": 1.307610273361206, "learning_rate": 0.00049677195786037, "loss": 1.8763635158538818, "step": 13928 }, { "epoch": 1.6207819408889923, "grad_norm": 1.2140357494354248, "learning_rate": 0.0004967352516267677, "loss": 1.9407377243041992, "step": 13929 }, { "epoch": 1.6208983011403304, "grad_norm": 0.7863349318504333, "learning_rate": 0.0004966983392486173, "loss": 1.2894684076309204, "step": 13930 }, { "epoch": 1.6210146613916687, "grad_norm": 1.1123641729354858, "learning_rate": 0.0004966612207567587, "loss": 1.7405970096588135, "step": 13931 }, { "epoch": 1.6211310216430066, "grad_norm": 1.225904107093811, "learning_rate": 0.000496623896182204, "loss": 1.4009602069854736, "step": 13932 }, { "epoch": 1.621247381894345, "grad_norm": 0.9314991235733032, "learning_rate": 0.0004965863655561375, "loss": 1.730851650238037, "step": 13933 }, { "epoch": 1.621363742145683, "grad_norm": 0.708278477191925, "learning_rate": 0.0004965486289099155, "loss": 1.1203769445419312, "step": 13934 }, { "epoch": 1.6214801023970211, "grad_norm": 1.32573401927948, "learning_rate": 0.0004965106862750668, "loss": 2.102515697479248, "step": 13935 }, { "epoch": 1.6215964626483594, "grad_norm": 1.3958741426467896, "learning_rate": 0.0004964725376832917, "loss": 2.1672873497009277, "step": 13936 }, { "epoch": 1.6217128228996973, "grad_norm": 1.7167991399765015, "learning_rate": 0.0004964341831664631, "loss": 1.8990979194641113, "step": 13937 }, { "epoch": 1.6218291831510356, "grad_norm": 1.1590396165847778, "learning_rate": 0.000496395622756626, "loss": 2.1459763050079346, "step": 13938 }, { "epoch": 1.6219455434023737, "grad_norm": 0.9205518960952759, "learning_rate": 0.0004963568564859967, "loss": 2.044200897216797, "step": 13939 }, { "epoch": 1.6220619036537118, "grad_norm": 1.1388373374938965, "learning_rate": 0.0004963178843869643, "loss": 1.7630952596664429, "step": 13940 }, { "epoch": 1.6221782639050502, "grad_norm": 1.123687982559204, "learning_rate": 0.0004962787064920896, "loss": 1.486252784729004, "step": 13941 }, { "epoch": 1.622294624156388, "grad_norm": 1.2478346824645996, "learning_rate": 0.0004962393228341053, "loss": 1.5874313116073608, "step": 13942 }, { "epoch": 1.6224109844077264, "grad_norm": 2.54662823677063, "learning_rate": 0.0004961997334459157, "loss": 1.9545669555664062, "step": 13943 }, { "epoch": 1.6225273446590645, "grad_norm": 0.953446626663208, "learning_rate": 0.0004961599383605975, "loss": 2.3662450313568115, "step": 13944 }, { "epoch": 1.6226437049104026, "grad_norm": 1.586772084236145, "learning_rate": 0.0004961199376113992, "loss": 2.0901577472686768, "step": 13945 }, { "epoch": 1.6227600651617409, "grad_norm": 1.2180758714675903, "learning_rate": 0.0004960797312317407, "loss": 1.5480709075927734, "step": 13946 }, { "epoch": 1.6228764254130787, "grad_norm": 0.8800512552261353, "learning_rate": 0.0004960393192552142, "loss": 1.9190187454223633, "step": 13947 }, { "epoch": 1.622992785664417, "grad_norm": 0.9567000269889832, "learning_rate": 0.0004959987017155833, "loss": 1.8017816543579102, "step": 13948 }, { "epoch": 1.6231091459157552, "grad_norm": 0.7383018136024475, "learning_rate": 0.0004959578786467835, "loss": 1.6975088119506836, "step": 13949 }, { "epoch": 1.6232255061670933, "grad_norm": 0.6656829714775085, "learning_rate": 0.0004959168500829221, "loss": 1.484917163848877, "step": 13950 }, { "epoch": 1.6233418664184316, "grad_norm": 0.8419094085693359, "learning_rate": 0.0004958756160582781, "loss": 1.357529878616333, "step": 13951 }, { "epoch": 1.6234582266697695, "grad_norm": 1.1758167743682861, "learning_rate": 0.0004958341766073018, "loss": 1.8990284204483032, "step": 13952 }, { "epoch": 1.6235745869211078, "grad_norm": 1.4558945894241333, "learning_rate": 0.0004957925317646157, "loss": 2.159247398376465, "step": 13953 }, { "epoch": 1.6236909471724459, "grad_norm": 0.9156855940818787, "learning_rate": 0.0004957506815650134, "loss": 2.004985809326172, "step": 13954 }, { "epoch": 1.623807307423784, "grad_norm": 1.4306840896606445, "learning_rate": 0.0004957086260434603, "loss": 2.4167323112487793, "step": 13955 }, { "epoch": 1.6239236676751223, "grad_norm": 1.8285799026489258, "learning_rate": 0.0004956663652350933, "loss": 1.6519265174865723, "step": 13956 }, { "epoch": 1.6240400279264602, "grad_norm": 0.5280382037162781, "learning_rate": 0.000495623899175221, "loss": 1.3949863910675049, "step": 13957 }, { "epoch": 1.6241563881777985, "grad_norm": 0.9992363452911377, "learning_rate": 0.0004955812278993231, "loss": 1.71132230758667, "step": 13958 }, { "epoch": 1.6242727484291366, "grad_norm": 1.8207011222839355, "learning_rate": 0.0004955383514430512, "loss": 1.655134916305542, "step": 13959 }, { "epoch": 1.6243891086804747, "grad_norm": 1.3991835117340088, "learning_rate": 0.000495495269842228, "loss": 1.9941545724868774, "step": 13960 }, { "epoch": 1.624505468931813, "grad_norm": 1.630895733833313, "learning_rate": 0.0004954519831328475, "loss": 2.1040897369384766, "step": 13961 }, { "epoch": 1.624621829183151, "grad_norm": 1.0229732990264893, "learning_rate": 0.0004954084913510755, "loss": 2.748098373413086, "step": 13962 }, { "epoch": 1.6247381894344892, "grad_norm": 1.4039764404296875, "learning_rate": 0.0004953647945332489, "loss": 2.3782572746276855, "step": 13963 }, { "epoch": 1.6248545496858273, "grad_norm": 0.9854406714439392, "learning_rate": 0.0004953208927158758, "loss": 2.0860161781311035, "step": 13964 }, { "epoch": 1.6249709099371654, "grad_norm": 0.7662069201469421, "learning_rate": 0.0004952767859356356, "loss": 1.493422508239746, "step": 13965 }, { "epoch": 1.6250872701885037, "grad_norm": 1.0866057872772217, "learning_rate": 0.0004952324742293793, "loss": 1.6851983070373535, "step": 13966 }, { "epoch": 1.6252036304398416, "grad_norm": 1.0347638130187988, "learning_rate": 0.0004951879576341286, "loss": 1.511987566947937, "step": 13967 }, { "epoch": 1.62531999069118, "grad_norm": 2.268639326095581, "learning_rate": 0.0004951432361870767, "loss": 1.6621291637420654, "step": 13968 }, { "epoch": 1.625436350942518, "grad_norm": 1.2126411199569702, "learning_rate": 0.0004950983099255878, "loss": 2.243774890899658, "step": 13969 }, { "epoch": 1.6255527111938561, "grad_norm": 1.4414196014404297, "learning_rate": 0.0004950531788871974, "loss": 2.661114454269409, "step": 13970 }, { "epoch": 1.6256690714451945, "grad_norm": 1.0498384237289429, "learning_rate": 0.000495007843109612, "loss": 2.0943992137908936, "step": 13971 }, { "epoch": 1.6257854316965323, "grad_norm": 1.4345440864562988, "learning_rate": 0.000494962302630709, "loss": 2.2711596488952637, "step": 13972 }, { "epoch": 1.6259017919478707, "grad_norm": 0.9977004528045654, "learning_rate": 0.000494916557488537, "loss": 2.2248783111572266, "step": 13973 }, { "epoch": 1.6260181521992088, "grad_norm": 1.3626019954681396, "learning_rate": 0.0004948706077213156, "loss": 1.699671745300293, "step": 13974 }, { "epoch": 1.6261345124505469, "grad_norm": 1.1334691047668457, "learning_rate": 0.0004948244533674353, "loss": 1.576979398727417, "step": 13975 }, { "epoch": 1.6262508727018852, "grad_norm": 1.1287366151809692, "learning_rate": 0.0004947780944654576, "loss": 1.9449975490570068, "step": 13976 }, { "epoch": 1.626367232953223, "grad_norm": 0.8679631352424622, "learning_rate": 0.0004947315310541147, "loss": 2.2582507133483887, "step": 13977 }, { "epoch": 1.6264835932045614, "grad_norm": 0.9159374833106995, "learning_rate": 0.00049468476317231, "loss": 1.6577293872833252, "step": 13978 }, { "epoch": 1.6265999534558995, "grad_norm": 1.6414077281951904, "learning_rate": 0.0004946377908591174, "loss": 1.5271719694137573, "step": 13979 }, { "epoch": 1.6267163137072376, "grad_norm": 1.5289615392684937, "learning_rate": 0.0004945906141537818, "loss": 1.6644477844238281, "step": 13980 }, { "epoch": 1.626832673958576, "grad_norm": 1.0395100116729736, "learning_rate": 0.0004945432330957188, "loss": 1.746694803237915, "step": 13981 }, { "epoch": 1.6269490342099138, "grad_norm": 1.7080378532409668, "learning_rate": 0.0004944956477245147, "loss": 2.390779495239258, "step": 13982 }, { "epoch": 1.627065394461252, "grad_norm": 0.8740108609199524, "learning_rate": 0.0004944478580799266, "loss": 1.179348349571228, "step": 13983 }, { "epoch": 1.6271817547125902, "grad_norm": 1.1034725904464722, "learning_rate": 0.0004943998642018822, "loss": 1.5534460544586182, "step": 13984 }, { "epoch": 1.6272981149639283, "grad_norm": 1.1470625400543213, "learning_rate": 0.0004943516661304798, "loss": 2.0100955963134766, "step": 13985 }, { "epoch": 1.6274144752152666, "grad_norm": 1.2524383068084717, "learning_rate": 0.0004943032639059886, "loss": 2.5051002502441406, "step": 13986 }, { "epoch": 1.6275308354666045, "grad_norm": 1.2393420934677124, "learning_rate": 0.0004942546575688477, "loss": 1.821035623550415, "step": 13987 }, { "epoch": 1.6276471957179428, "grad_norm": 2.7343266010284424, "learning_rate": 0.0004942058471596677, "loss": 1.9683135747909546, "step": 13988 }, { "epoch": 1.627763555969281, "grad_norm": 1.4117743968963623, "learning_rate": 0.0004941568327192285, "loss": 1.803572177886963, "step": 13989 }, { "epoch": 1.627879916220619, "grad_norm": 1.302253007888794, "learning_rate": 0.0004941076142884817, "loss": 2.1894750595092773, "step": 13990 }, { "epoch": 1.6279962764719573, "grad_norm": 2.610032320022583, "learning_rate": 0.0004940581919085483, "loss": 2.0427145957946777, "step": 13991 }, { "epoch": 1.6281126367232952, "grad_norm": 0.7454812526702881, "learning_rate": 0.0004940085656207204, "loss": 1.2441593408584595, "step": 13992 }, { "epoch": 1.6282289969746335, "grad_norm": 1.2500391006469727, "learning_rate": 0.00049395873546646, "loss": 1.6173665523529053, "step": 13993 }, { "epoch": 1.6283453572259716, "grad_norm": 1.6676105260849, "learning_rate": 0.0004939087014874001, "loss": 2.373695135116577, "step": 13994 }, { "epoch": 1.6284617174773097, "grad_norm": 1.9423359632492065, "learning_rate": 0.0004938584637253431, "loss": 1.741048812866211, "step": 13995 }, { "epoch": 1.628578077728648, "grad_norm": 1.9772881269454956, "learning_rate": 0.0004938080222222621, "loss": 2.976816177368164, "step": 13996 }, { "epoch": 1.628694437979986, "grad_norm": 1.562342882156372, "learning_rate": 0.0004937573770203006, "loss": 1.9450910091400146, "step": 13997 }, { "epoch": 1.6288107982313242, "grad_norm": 1.2237471342086792, "learning_rate": 0.0004937065281617719, "loss": 2.094447135925293, "step": 13998 }, { "epoch": 1.6289271584826623, "grad_norm": 1.3304411172866821, "learning_rate": 0.00049365547568916, "loss": 2.6940665245056152, "step": 13999 }, { "epoch": 1.6290435187340004, "grad_norm": 1.5956430435180664, "learning_rate": 0.0004936042196451183, "loss": 2.137430191040039, "step": 14000 }, { "epoch": 1.6291598789853388, "grad_norm": 1.5227841138839722, "learning_rate": 0.0004935527600724709, "loss": 1.566148042678833, "step": 14001 }, { "epoch": 1.6292762392366766, "grad_norm": 0.9667123556137085, "learning_rate": 0.0004935010970142115, "loss": 2.4133973121643066, "step": 14002 }, { "epoch": 1.629392599488015, "grad_norm": 0.9802152514457703, "learning_rate": 0.0004934492305135041, "loss": 1.2284820079803467, "step": 14003 }, { "epoch": 1.629508959739353, "grad_norm": 1.0570569038391113, "learning_rate": 0.0004933971606136826, "loss": 1.8400110006332397, "step": 14004 }, { "epoch": 1.6296253199906912, "grad_norm": 0.46214449405670166, "learning_rate": 0.0004933448873582508, "loss": 1.001438856124878, "step": 14005 }, { "epoch": 1.6297416802420295, "grad_norm": 0.9600604176521301, "learning_rate": 0.0004932924107908825, "loss": 2.078709363937378, "step": 14006 }, { "epoch": 1.6298580404933674, "grad_norm": 1.338693618774414, "learning_rate": 0.0004932397309554212, "loss": 2.192668914794922, "step": 14007 }, { "epoch": 1.6299744007447057, "grad_norm": 1.15443754196167, "learning_rate": 0.0004931868478958803, "loss": 1.5607216358184814, "step": 14008 }, { "epoch": 1.6300907609960438, "grad_norm": 1.0297539234161377, "learning_rate": 0.0004931337616564431, "loss": 2.653933525085449, "step": 14009 }, { "epoch": 1.6302071212473819, "grad_norm": 1.325615406036377, "learning_rate": 0.0004930804722814623, "loss": 2.846874237060547, "step": 14010 }, { "epoch": 1.6303234814987202, "grad_norm": 1.244092583656311, "learning_rate": 0.000493026979815461, "loss": 1.8349025249481201, "step": 14011 }, { "epoch": 1.630439841750058, "grad_norm": 0.6879226565361023, "learning_rate": 0.0004929732843031312, "loss": 1.3345171213150024, "step": 14012 }, { "epoch": 1.6305562020013964, "grad_norm": 0.8669719099998474, "learning_rate": 0.0004929193857893351, "loss": 1.8064546585083008, "step": 14013 }, { "epoch": 1.6306725622527345, "grad_norm": 1.1331368684768677, "learning_rate": 0.0004928652843191043, "loss": 2.0145604610443115, "step": 14014 }, { "epoch": 1.6307889225040726, "grad_norm": 0.7275893688201904, "learning_rate": 0.0004928109799376398, "loss": 1.5504543781280518, "step": 14015 }, { "epoch": 1.630905282755411, "grad_norm": 0.7928239107131958, "learning_rate": 0.0004927564726903125, "loss": 1.560194492340088, "step": 14016 }, { "epoch": 1.6310216430067488, "grad_norm": 1.1621334552764893, "learning_rate": 0.0004927017626226624, "loss": 1.958379864692688, "step": 14017 }, { "epoch": 1.631138003258087, "grad_norm": 1.033800721168518, "learning_rate": 0.0004926468497803993, "loss": 2.0124921798706055, "step": 14018 }, { "epoch": 1.6312543635094252, "grad_norm": 1.0736384391784668, "learning_rate": 0.0004925917342094023, "loss": 1.8897643089294434, "step": 14019 }, { "epoch": 1.6313707237607633, "grad_norm": 1.4593814611434937, "learning_rate": 0.0004925364159557198, "loss": 1.6385776996612549, "step": 14020 }, { "epoch": 1.6314870840121016, "grad_norm": 0.7913676500320435, "learning_rate": 0.0004924808950655695, "loss": 1.222063422203064, "step": 14021 }, { "epoch": 1.6316034442634395, "grad_norm": 0.9747443795204163, "learning_rate": 0.0004924251715853385, "loss": 1.5206341743469238, "step": 14022 }, { "epoch": 1.6317198045147778, "grad_norm": 1.2945197820663452, "learning_rate": 0.0004923692455615832, "loss": 2.0954623222351074, "step": 14023 }, { "epoch": 1.631836164766116, "grad_norm": 1.104199767112732, "learning_rate": 0.0004923131170410292, "loss": 2.1190409660339355, "step": 14024 }, { "epoch": 1.631952525017454, "grad_norm": 0.6899248957633972, "learning_rate": 0.0004922567860705712, "loss": 1.6359220743179321, "step": 14025 }, { "epoch": 1.6320688852687923, "grad_norm": 1.6814452409744263, "learning_rate": 0.0004922002526972732, "loss": 1.856919527053833, "step": 14026 }, { "epoch": 1.6321852455201302, "grad_norm": 0.8626749515533447, "learning_rate": 0.000492143516968368, "loss": 1.3120744228363037, "step": 14027 }, { "epoch": 1.6323016057714685, "grad_norm": 0.9737557768821716, "learning_rate": 0.0004920865789312578, "loss": 2.2306365966796875, "step": 14028 }, { "epoch": 1.6324179660228066, "grad_norm": 1.4901390075683594, "learning_rate": 0.0004920294386335137, "loss": 1.902206540107727, "step": 14029 }, { "epoch": 1.6325343262741447, "grad_norm": 0.9358533024787903, "learning_rate": 0.0004919720961228757, "loss": 1.5626449584960938, "step": 14030 }, { "epoch": 1.6326506865254828, "grad_norm": 0.35893648862838745, "learning_rate": 0.0004919145514472529, "loss": 1.0453190803527832, "step": 14031 }, { "epoch": 1.632767046776821, "grad_norm": 0.9666571021080017, "learning_rate": 0.0004918568046547232, "loss": 1.7260605096817017, "step": 14032 }, { "epoch": 1.6328834070281593, "grad_norm": 0.9171991348266602, "learning_rate": 0.0004917988557935335, "loss": 2.185264825820923, "step": 14033 }, { "epoch": 1.6329997672794974, "grad_norm": 0.9305816888809204, "learning_rate": 0.0004917407049120992, "loss": 1.8376173973083496, "step": 14034 }, { "epoch": 1.6331161275308355, "grad_norm": 0.738308310508728, "learning_rate": 0.000491682352059005, "loss": 1.5947173833847046, "step": 14035 }, { "epoch": 1.6332324877821736, "grad_norm": 0.9588669538497925, "learning_rate": 0.0004916237972830038, "loss": 1.7126836776733398, "step": 14036 }, { "epoch": 1.6333488480335117, "grad_norm": 1.3587498664855957, "learning_rate": 0.0004915650406330177, "loss": 2.2491488456726074, "step": 14037 }, { "epoch": 1.63346520828485, "grad_norm": 1.3009045124053955, "learning_rate": 0.0004915060821581369, "loss": 1.6790516376495361, "step": 14038 }, { "epoch": 1.633581568536188, "grad_norm": 1.3384910821914673, "learning_rate": 0.0004914469219076209, "loss": 2.3846216201782227, "step": 14039 }, { "epoch": 1.6336979287875262, "grad_norm": 1.0670520067214966, "learning_rate": 0.0004913875599308973, "loss": 2.054048538208008, "step": 14040 }, { "epoch": 1.6338142890388643, "grad_norm": 1.173587679862976, "learning_rate": 0.0004913279962775622, "loss": 1.8786745071411133, "step": 14041 }, { "epoch": 1.6339306492902024, "grad_norm": 1.0295363664627075, "learning_rate": 0.0004912682309973806, "loss": 1.7134684324264526, "step": 14042 }, { "epoch": 1.6340470095415407, "grad_norm": 1.017762303352356, "learning_rate": 0.0004912082641402856, "loss": 1.3014936447143555, "step": 14043 }, { "epoch": 1.6341633697928788, "grad_norm": 1.2427622079849243, "learning_rate": 0.0004911480957563789, "loss": 2.1107616424560547, "step": 14044 }, { "epoch": 1.634279730044217, "grad_norm": 1.3788639307022095, "learning_rate": 0.0004910877258959305, "loss": 2.406411647796631, "step": 14045 }, { "epoch": 1.634396090295555, "grad_norm": 1.590275764465332, "learning_rate": 0.0004910271546093787, "loss": 2.160285234451294, "step": 14046 }, { "epoch": 1.634512450546893, "grad_norm": 0.9512869119644165, "learning_rate": 0.0004909663819473301, "loss": 1.725653886795044, "step": 14047 }, { "epoch": 1.6346288107982314, "grad_norm": 0.8351947665214539, "learning_rate": 0.0004909054079605598, "loss": 1.4351996183395386, "step": 14048 }, { "epoch": 1.6347451710495695, "grad_norm": 0.7934029698371887, "learning_rate": 0.0004908442327000106, "loss": 1.3992152214050293, "step": 14049 }, { "epoch": 1.6348615313009076, "grad_norm": 0.9889302849769592, "learning_rate": 0.000490782856216794, "loss": 1.864994764328003, "step": 14050 }, { "epoch": 1.6349778915522457, "grad_norm": 0.8800854682922363, "learning_rate": 0.0004907212785621891, "loss": 1.7100564241409302, "step": 14051 }, { "epoch": 1.6350942518035838, "grad_norm": 0.7431103587150574, "learning_rate": 0.0004906594997876436, "loss": 1.5051521062850952, "step": 14052 }, { "epoch": 1.6352106120549221, "grad_norm": 1.2399674654006958, "learning_rate": 0.0004905975199447728, "loss": 1.603585958480835, "step": 14053 }, { "epoch": 1.6353269723062602, "grad_norm": 0.6967872381210327, "learning_rate": 0.0004905353390853602, "loss": 1.1027281284332275, "step": 14054 }, { "epoch": 1.6354433325575983, "grad_norm": 0.8410109877586365, "learning_rate": 0.0004904729572613574, "loss": 1.17825448513031, "step": 14055 }, { "epoch": 1.6355596928089364, "grad_norm": 1.3390291929244995, "learning_rate": 0.0004904103745248834, "loss": 2.0713605880737305, "step": 14056 }, { "epoch": 1.6356760530602745, "grad_norm": 1.0130528211593628, "learning_rate": 0.0004903475909282256, "loss": 1.8802703619003296, "step": 14057 }, { "epoch": 1.6357924133116128, "grad_norm": 0.9218980669975281, "learning_rate": 0.000490284606523839, "loss": 1.7439265251159668, "step": 14058 }, { "epoch": 1.635908773562951, "grad_norm": 1.5940598249435425, "learning_rate": 0.0004902214213643462, "loss": 2.5078659057617188, "step": 14059 }, { "epoch": 1.636025133814289, "grad_norm": 1.8433772325515747, "learning_rate": 0.0004901580355025378, "loss": 1.2815942764282227, "step": 14060 }, { "epoch": 1.6361414940656271, "grad_norm": 1.5063639879226685, "learning_rate": 0.000490094448991372, "loss": 2.4748940467834473, "step": 14061 }, { "epoch": 1.6362578543169652, "grad_norm": 0.6492930054664612, "learning_rate": 0.0004900306618839746, "loss": 1.7800254821777344, "step": 14062 }, { "epoch": 1.6363742145683036, "grad_norm": 0.8186988234519958, "learning_rate": 0.0004899666742336391, "loss": 1.8110737800598145, "step": 14063 }, { "epoch": 1.6364905748196417, "grad_norm": 1.3923741579055786, "learning_rate": 0.0004899024860938263, "loss": 2.1540582180023193, "step": 14064 }, { "epoch": 1.6366069350709798, "grad_norm": 1.3246077299118042, "learning_rate": 0.0004898380975181647, "loss": 1.4703874588012695, "step": 14065 }, { "epoch": 1.6367232953223179, "grad_norm": 1.160491704940796, "learning_rate": 0.0004897735085604504, "loss": 1.7792528867721558, "step": 14066 }, { "epoch": 1.636839655573656, "grad_norm": 0.8823750019073486, "learning_rate": 0.0004897087192746465, "loss": 1.7259602546691895, "step": 14067 }, { "epoch": 1.6369560158249943, "grad_norm": 1.6623036861419678, "learning_rate": 0.0004896437297148839, "loss": 1.965462327003479, "step": 14068 }, { "epoch": 1.6370723760763322, "grad_norm": 1.152115821838379, "learning_rate": 0.0004895785399354607, "loss": 1.8632121086120605, "step": 14069 }, { "epoch": 1.6371887363276705, "grad_norm": 1.315213680267334, "learning_rate": 0.0004895131499908421, "loss": 2.279393434524536, "step": 14070 }, { "epoch": 1.6373050965790086, "grad_norm": 0.8579970598220825, "learning_rate": 0.0004894475599356607, "loss": 1.6123011112213135, "step": 14071 }, { "epoch": 1.6374214568303467, "grad_norm": 1.6990336179733276, "learning_rate": 0.0004893817698247163, "loss": 2.2088229656219482, "step": 14072 }, { "epoch": 1.637537817081685, "grad_norm": 1.110506534576416, "learning_rate": 0.0004893157797129759, "loss": 2.2586493492126465, "step": 14073 }, { "epoch": 1.6376541773330229, "grad_norm": 0.9937312006950378, "learning_rate": 0.0004892495896555733, "loss": 2.0752291679382324, "step": 14074 }, { "epoch": 1.6377705375843612, "grad_norm": 0.8383351564407349, "learning_rate": 0.0004891831997078096, "loss": 1.639857530593872, "step": 14075 }, { "epoch": 1.6378868978356993, "grad_norm": 0.9419999122619629, "learning_rate": 0.000489116609925153, "loss": 1.5856517553329468, "step": 14076 }, { "epoch": 1.6380032580870374, "grad_norm": 1.6022920608520508, "learning_rate": 0.0004890498203632383, "loss": 2.6406822204589844, "step": 14077 }, { "epoch": 1.6381196183383757, "grad_norm": 0.9761084914207458, "learning_rate": 0.0004889828310778676, "loss": 1.8104033470153809, "step": 14078 }, { "epoch": 1.6382359785897136, "grad_norm": 1.2920767068862915, "learning_rate": 0.0004889156421250096, "loss": 1.9605172872543335, "step": 14079 }, { "epoch": 1.638352338841052, "grad_norm": 0.9255894422531128, "learning_rate": 0.0004888482535608, "loss": 1.3951689004898071, "step": 14080 }, { "epoch": 1.63846869909239, "grad_norm": 1.1743214130401611, "learning_rate": 0.0004887806654415412, "loss": 1.832796573638916, "step": 14081 }, { "epoch": 1.6385850593437281, "grad_norm": 1.3101806640625, "learning_rate": 0.0004887128778237023, "loss": 1.9235955476760864, "step": 14082 }, { "epoch": 1.6387014195950664, "grad_norm": 1.1191219091415405, "learning_rate": 0.000488644890763919, "loss": 2.2421393394470215, "step": 14083 }, { "epoch": 1.6388177798464043, "grad_norm": 1.361079216003418, "learning_rate": 0.0004885767043189939, "loss": 1.9953742027282715, "step": 14084 }, { "epoch": 1.6389341400977426, "grad_norm": 1.137132167816162, "learning_rate": 0.0004885083185458958, "loss": 1.3334462642669678, "step": 14085 }, { "epoch": 1.6390505003490807, "grad_norm": 2.0497729778289795, "learning_rate": 0.0004884397335017605, "loss": 2.5558016300201416, "step": 14086 }, { "epoch": 1.6391668606004188, "grad_norm": 1.099760890007019, "learning_rate": 0.0004883709492438899, "loss": 2.2685930728912354, "step": 14087 }, { "epoch": 1.6392832208517572, "grad_norm": 1.1276710033416748, "learning_rate": 0.0004883019658297523, "loss": 1.3446325063705444, "step": 14088 }, { "epoch": 1.639399581103095, "grad_norm": 1.1518274545669556, "learning_rate": 0.00048823278331698307, "loss": 2.1180989742279053, "step": 14089 }, { "epoch": 1.6395159413544333, "grad_norm": 1.8522460460662842, "learning_rate": 0.00048816340176338293, "loss": 2.3293139934539795, "step": 14090 }, { "epoch": 1.6396323016057714, "grad_norm": 1.1845948696136475, "learning_rate": 0.0004880938212269197, "loss": 1.4100151062011719, "step": 14091 }, { "epoch": 1.6397486618571095, "grad_norm": 1.5347414016723633, "learning_rate": 0.0004880240417657269, "loss": 1.6369751691818237, "step": 14092 }, { "epoch": 1.6398650221084479, "grad_norm": 1.321790337562561, "learning_rate": 0.0004879540634381047, "loss": 1.9041519165039062, "step": 14093 }, { "epoch": 1.6399813823597857, "grad_norm": 1.003399133682251, "learning_rate": 0.0004878838863025191, "loss": 1.6195135116577148, "step": 14094 }, { "epoch": 1.640097742611124, "grad_norm": 1.4058055877685547, "learning_rate": 0.00048781351041760236, "loss": 2.0030593872070312, "step": 14095 }, { "epoch": 1.6402141028624622, "grad_norm": 1.100974440574646, "learning_rate": 0.0004877429358421527, "loss": 1.9203534126281738, "step": 14096 }, { "epoch": 1.6403304631138003, "grad_norm": 0.7954829335212708, "learning_rate": 0.0004876721626351345, "loss": 1.433027744293213, "step": 14097 }, { "epoch": 1.6404468233651386, "grad_norm": 1.4703117609024048, "learning_rate": 0.0004876011908556778, "loss": 1.8832807540893555, "step": 14098 }, { "epoch": 1.6405631836164765, "grad_norm": 1.0372318029403687, "learning_rate": 0.00048753002056307887, "loss": 1.8638272285461426, "step": 14099 }, { "epoch": 1.6406795438678148, "grad_norm": 1.174626350402832, "learning_rate": 0.00048745865181679967, "loss": 2.0437216758728027, "step": 14100 }, { "epoch": 1.6407959041191529, "grad_norm": 0.9691397547721863, "learning_rate": 0.000487387084676468, "loss": 1.677007794380188, "step": 14101 }, { "epoch": 1.640912264370491, "grad_norm": 0.9237362146377563, "learning_rate": 0.0004873153192018773, "loss": 2.102181911468506, "step": 14102 }, { "epoch": 1.6410286246218293, "grad_norm": 0.679270327091217, "learning_rate": 0.000487243355452987, "loss": 1.3722496032714844, "step": 14103 }, { "epoch": 1.6411449848731672, "grad_norm": 1.2253996133804321, "learning_rate": 0.0004871711934899219, "loss": 2.025102138519287, "step": 14104 }, { "epoch": 1.6412613451245055, "grad_norm": 1.0804400444030762, "learning_rate": 0.00048709883337297246, "loss": 1.9962406158447266, "step": 14105 }, { "epoch": 1.6413777053758436, "grad_norm": 0.8687331676483154, "learning_rate": 0.0004870262751625949, "loss": 2.079232692718506, "step": 14106 }, { "epoch": 1.6414940656271817, "grad_norm": 0.8105237483978271, "learning_rate": 0.0004869535189194106, "loss": 1.7906174659729004, "step": 14107 }, { "epoch": 1.64161042587852, "grad_norm": 0.9127864241600037, "learning_rate": 0.0004868805647042068, "loss": 1.3506314754486084, "step": 14108 }, { "epoch": 1.641726786129858, "grad_norm": 2.402621269226074, "learning_rate": 0.00048680741257793586, "loss": 2.1041507720947266, "step": 14109 }, { "epoch": 1.6418431463811962, "grad_norm": 1.2782224416732788, "learning_rate": 0.0004867340626017155, "loss": 1.6872272491455078, "step": 14110 }, { "epoch": 1.6419595066325343, "grad_norm": 1.3974335193634033, "learning_rate": 0.0004866605148368288, "loss": 2.624572277069092, "step": 14111 }, { "epoch": 1.6420758668838724, "grad_norm": 0.9476233124732971, "learning_rate": 0.0004865867693447242, "loss": 1.7828056812286377, "step": 14112 }, { "epoch": 1.6421922271352107, "grad_norm": 0.49278658628463745, "learning_rate": 0.0004865128261870152, "loss": 1.0245168209075928, "step": 14113 }, { "epoch": 1.6423085873865486, "grad_norm": 0.9794467687606812, "learning_rate": 0.00048643868542548043, "loss": 2.6387643814086914, "step": 14114 }, { "epoch": 1.642424947637887, "grad_norm": 1.5215705633163452, "learning_rate": 0.0004863643471220637, "loss": 1.5636451244354248, "step": 14115 }, { "epoch": 1.642541307889225, "grad_norm": 0.8345021605491638, "learning_rate": 0.00048628981133887384, "loss": 1.948599934577942, "step": 14116 }, { "epoch": 1.6426576681405631, "grad_norm": 0.7187498211860657, "learning_rate": 0.0004862150781381847, "loss": 1.4183766841888428, "step": 14117 }, { "epoch": 1.6427740283919015, "grad_norm": 1.1201186180114746, "learning_rate": 0.00048614014758243497, "loss": 1.572871208190918, "step": 14118 }, { "epoch": 1.6428903886432393, "grad_norm": 0.9990332126617432, "learning_rate": 0.00048606501973422824, "loss": 2.159233570098877, "step": 14119 }, { "epoch": 1.6430067488945777, "grad_norm": 0.7895102500915527, "learning_rate": 0.00048598969465633315, "loss": 1.8425320386886597, "step": 14120 }, { "epoch": 1.6431231091459158, "grad_norm": 1.0467311143875122, "learning_rate": 0.0004859141724116828, "loss": 1.891430139541626, "step": 14121 }, { "epoch": 1.6432394693972538, "grad_norm": 1.1351226568222046, "learning_rate": 0.00048583845306337533, "loss": 1.9888222217559814, "step": 14122 }, { "epoch": 1.6433558296485922, "grad_norm": 0.9176718592643738, "learning_rate": 0.0004857625366746733, "loss": 1.8373236656188965, "step": 14123 }, { "epoch": 1.64347218989993, "grad_norm": 1.0739762783050537, "learning_rate": 0.00048568642330900404, "loss": 2.266803741455078, "step": 14124 }, { "epoch": 1.6435885501512684, "grad_norm": 1.2332552671432495, "learning_rate": 0.00048561011302995945, "loss": 1.6799917221069336, "step": 14125 }, { "epoch": 1.6437049104026065, "grad_norm": 0.9615172743797302, "learning_rate": 0.00048553360590129584, "loss": 1.6931066513061523, "step": 14126 }, { "epoch": 1.6438212706539446, "grad_norm": 1.0989658832550049, "learning_rate": 0.00048545690198693414, "loss": 1.975261926651001, "step": 14127 }, { "epoch": 1.6439376309052829, "grad_norm": 1.3078515529632568, "learning_rate": 0.0004853800013509596, "loss": 2.0242252349853516, "step": 14128 }, { "epoch": 1.6440539911566208, "grad_norm": 1.8293848037719727, "learning_rate": 0.0004853029040576218, "loss": 2.805037498474121, "step": 14129 }, { "epoch": 1.644170351407959, "grad_norm": 1.0964457988739014, "learning_rate": 0.0004852256101713347, "loss": 1.5378915071487427, "step": 14130 }, { "epoch": 1.6442867116592972, "grad_norm": 2.368969440460205, "learning_rate": 0.00048514811975667644, "loss": 2.1830191612243652, "step": 14131 }, { "epoch": 1.6444030719106353, "grad_norm": 1.107041597366333, "learning_rate": 0.0004850704328783895, "loss": 2.7566230297088623, "step": 14132 }, { "epoch": 1.6445194321619736, "grad_norm": 1.1571794748306274, "learning_rate": 0.0004849925496013803, "loss": 2.3854939937591553, "step": 14133 }, { "epoch": 1.6446357924133115, "grad_norm": 1.265122890472412, "learning_rate": 0.00048491446999071944, "loss": 1.919357419013977, "step": 14134 }, { "epoch": 1.6447521526646498, "grad_norm": 1.0004674196243286, "learning_rate": 0.0004848361941116416, "loss": 1.5177905559539795, "step": 14135 }, { "epoch": 1.644868512915988, "grad_norm": 1.1491793394088745, "learning_rate": 0.00048475772202954536, "loss": 2.0415258407592773, "step": 14136 }, { "epoch": 1.644984873167326, "grad_norm": 0.8340983390808105, "learning_rate": 0.0004846790538099934, "loss": 1.6909430027008057, "step": 14137 }, { "epoch": 1.6451012334186643, "grad_norm": 0.8092512488365173, "learning_rate": 0.000484600189518712, "loss": 1.2351083755493164, "step": 14138 }, { "epoch": 1.6452175936700022, "grad_norm": 1.163865327835083, "learning_rate": 0.0004845211292215915, "loss": 1.5223674774169922, "step": 14139 }, { "epoch": 1.6453339539213405, "grad_norm": 1.1138792037963867, "learning_rate": 0.0004844418729846858, "loss": 1.6223446130752563, "step": 14140 }, { "epoch": 1.6454503141726786, "grad_norm": 0.9544771909713745, "learning_rate": 0.00048436242087421266, "loss": 1.876983642578125, "step": 14141 }, { "epoch": 1.6455666744240167, "grad_norm": 1.0889334678649902, "learning_rate": 0.00048428277295655336, "loss": 1.4874155521392822, "step": 14142 }, { "epoch": 1.645683034675355, "grad_norm": 1.087013840675354, "learning_rate": 0.0004842029292982529, "loss": 2.0961031913757324, "step": 14143 }, { "epoch": 1.645799394926693, "grad_norm": 0.6102704405784607, "learning_rate": 0.00048412288996601985, "loss": 1.479455590248108, "step": 14144 }, { "epoch": 1.6459157551780312, "grad_norm": 0.5991660952568054, "learning_rate": 0.00048404265502672607, "loss": 1.0974931716918945, "step": 14145 }, { "epoch": 1.6460321154293693, "grad_norm": 1.4052889347076416, "learning_rate": 0.0004839622245474069, "loss": 2.3815019130706787, "step": 14146 }, { "epoch": 1.6461484756807074, "grad_norm": 1.5091580152511597, "learning_rate": 0.0004838815985952613, "loss": 1.5737088918685913, "step": 14147 }, { "epoch": 1.6462648359320458, "grad_norm": 1.4389798641204834, "learning_rate": 0.00048380077723765113, "loss": 2.212367534637451, "step": 14148 }, { "epoch": 1.6463811961833836, "grad_norm": 1.6138008832931519, "learning_rate": 0.00048371976054210187, "loss": 1.804025411605835, "step": 14149 }, { "epoch": 1.646497556434722, "grad_norm": 1.0764626264572144, "learning_rate": 0.00048363854857630205, "loss": 2.0689640045166016, "step": 14150 }, { "epoch": 1.64661391668606, "grad_norm": 1.343229055404663, "learning_rate": 0.00048355714140810334, "loss": 1.5263581275939941, "step": 14151 }, { "epoch": 1.6467302769373982, "grad_norm": 1.0689806938171387, "learning_rate": 0.00048347553910552056, "loss": 1.986612319946289, "step": 14152 }, { "epoch": 1.6468466371887365, "grad_norm": 1.0290956497192383, "learning_rate": 0.00048339374173673143, "loss": 2.64162015914917, "step": 14153 }, { "epoch": 1.6469629974400743, "grad_norm": 0.9570325016975403, "learning_rate": 0.00048331174937007684, "loss": 1.5931954383850098, "step": 14154 }, { "epoch": 1.6470793576914127, "grad_norm": 0.8872421383857727, "learning_rate": 0.0004832295620740604, "loss": 1.2707945108413696, "step": 14155 }, { "epoch": 1.6471957179427508, "grad_norm": 1.144132375717163, "learning_rate": 0.0004831471799173488, "loss": 2.1796107292175293, "step": 14156 }, { "epoch": 1.6473120781940889, "grad_norm": 0.782243013381958, "learning_rate": 0.00048306460296877133, "loss": 1.3470633029937744, "step": 14157 }, { "epoch": 1.6474284384454272, "grad_norm": 1.2367509603500366, "learning_rate": 0.0004829818312973201, "loss": 1.5118629932403564, "step": 14158 }, { "epoch": 1.647544798696765, "grad_norm": 1.1999766826629639, "learning_rate": 0.00048289886497214995, "loss": 2.1874499320983887, "step": 14159 }, { "epoch": 1.6476611589481034, "grad_norm": 1.3481342792510986, "learning_rate": 0.00048281570406257836, "loss": 1.8144209384918213, "step": 14160 }, { "epoch": 1.6477775191994415, "grad_norm": 1.6294867992401123, "learning_rate": 0.00048273234863808524, "loss": 2.189547300338745, "step": 14161 }, { "epoch": 1.6478938794507796, "grad_norm": 1.1495845317840576, "learning_rate": 0.0004826487987683131, "loss": 1.6757874488830566, "step": 14162 }, { "epoch": 1.648010239702118, "grad_norm": 1.2277486324310303, "learning_rate": 0.0004825650545230671, "loss": 2.263546943664551, "step": 14163 }, { "epoch": 1.6481265999534558, "grad_norm": 0.8358234167098999, "learning_rate": 0.0004824811159723145, "loss": 2.177323579788208, "step": 14164 }, { "epoch": 1.648242960204794, "grad_norm": 1.360769271850586, "learning_rate": 0.000482396983186185, "loss": 2.010547161102295, "step": 14165 }, { "epoch": 1.6483593204561322, "grad_norm": 1.341064453125, "learning_rate": 0.0004823126562349706, "loss": 2.24035382270813, "step": 14166 }, { "epoch": 1.6484756807074703, "grad_norm": 1.1099003553390503, "learning_rate": 0.0004822281351891256, "loss": 1.482375144958496, "step": 14167 }, { "epoch": 1.6485920409588086, "grad_norm": 1.0768440961837769, "learning_rate": 0.0004821434201192664, "loss": 2.130270004272461, "step": 14168 }, { "epoch": 1.6487084012101465, "grad_norm": 1.2306501865386963, "learning_rate": 0.0004820585110961714, "loss": 2.502366542816162, "step": 14169 }, { "epoch": 1.6488247614614848, "grad_norm": 1.4565768241882324, "learning_rate": 0.00048197340819078116, "loss": 2.392578601837158, "step": 14170 }, { "epoch": 1.648941121712823, "grad_norm": 1.112788438796997, "learning_rate": 0.0004818881114741983, "loss": 2.0820322036743164, "step": 14171 }, { "epoch": 1.649057481964161, "grad_norm": 1.070878505706787, "learning_rate": 0.0004818026210176872, "loss": 1.877503514289856, "step": 14172 }, { "epoch": 1.6491738422154993, "grad_norm": 0.8934618234634399, "learning_rate": 0.00048171693689267415, "loss": 1.7517765760421753, "step": 14173 }, { "epoch": 1.6492902024668372, "grad_norm": 1.2201530933380127, "learning_rate": 0.00048163105917074735, "loss": 2.2102866172790527, "step": 14174 }, { "epoch": 1.6494065627181755, "grad_norm": 0.7433445453643799, "learning_rate": 0.00048154498792365665, "loss": 1.5562011003494263, "step": 14175 }, { "epoch": 1.6495229229695136, "grad_norm": 1.0023365020751953, "learning_rate": 0.00048145872322331363, "loss": 2.1339643001556396, "step": 14176 }, { "epoch": 1.6496392832208517, "grad_norm": 1.4662096500396729, "learning_rate": 0.00048137226514179143, "loss": 2.6208736896514893, "step": 14177 }, { "epoch": 1.64975564347219, "grad_norm": 0.9942989349365234, "learning_rate": 0.00048128561375132496, "loss": 1.7908580303192139, "step": 14178 }, { "epoch": 1.649872003723528, "grad_norm": 1.5278468132019043, "learning_rate": 0.00048119876912431026, "loss": 2.583188056945801, "step": 14179 }, { "epoch": 1.6499883639748663, "grad_norm": 1.1003944873809814, "learning_rate": 0.0004811117313333052, "loss": 1.373710036277771, "step": 14180 }, { "epoch": 1.6501047242262044, "grad_norm": 1.3507736921310425, "learning_rate": 0.00048102450045102896, "loss": 0.9679909944534302, "step": 14181 }, { "epoch": 1.6502210844775425, "grad_norm": 1.1350854635238647, "learning_rate": 0.00048093707655036177, "loss": 1.7710940837860107, "step": 14182 }, { "epoch": 1.6503374447288808, "grad_norm": 0.8994205594062805, "learning_rate": 0.0004808494597043454, "loss": 1.210651159286499, "step": 14183 }, { "epoch": 1.6504538049802187, "grad_norm": 1.1270650625228882, "learning_rate": 0.0004807616499861828, "loss": 1.5139446258544922, "step": 14184 }, { "epoch": 1.650570165231557, "grad_norm": 1.6444467306137085, "learning_rate": 0.00048067364746923793, "loss": 1.6193915605545044, "step": 14185 }, { "epoch": 1.650686525482895, "grad_norm": 0.7672353386878967, "learning_rate": 0.000480585452227036, "loss": 1.319692611694336, "step": 14186 }, { "epoch": 1.6508028857342332, "grad_norm": 1.1821773052215576, "learning_rate": 0.00048049706433326303, "loss": 1.4815266132354736, "step": 14187 }, { "epoch": 1.6509192459855715, "grad_norm": 0.8602755069732666, "learning_rate": 0.0004804084838617662, "loss": 1.534318447113037, "step": 14188 }, { "epoch": 1.6510356062369094, "grad_norm": 0.9993393421173096, "learning_rate": 0.0004803197108865535, "loss": 1.8783191442489624, "step": 14189 }, { "epoch": 1.6511519664882477, "grad_norm": 0.906825065612793, "learning_rate": 0.0004802307454817938, "loss": 1.1186741590499878, "step": 14190 }, { "epoch": 1.6512683267395858, "grad_norm": 0.5796145796775818, "learning_rate": 0.0004801415877218166, "loss": 1.7305254936218262, "step": 14191 }, { "epoch": 1.651384686990924, "grad_norm": 1.1674127578735352, "learning_rate": 0.0004800522376811123, "loss": 2.3146281242370605, "step": 14192 }, { "epoch": 1.651501047242262, "grad_norm": 1.3274123668670654, "learning_rate": 0.0004799626954343318, "loss": 1.3951375484466553, "step": 14193 }, { "epoch": 1.6516174074936, "grad_norm": 0.9610176086425781, "learning_rate": 0.0004798729610562867, "loss": 1.6587696075439453, "step": 14194 }, { "epoch": 1.6517337677449384, "grad_norm": 1.364863634109497, "learning_rate": 0.00047978303462194904, "loss": 2.332315444946289, "step": 14195 }, { "epoch": 1.6518501279962765, "grad_norm": 0.8780062198638916, "learning_rate": 0.0004796929162064514, "loss": 1.2712903022766113, "step": 14196 }, { "epoch": 1.6519664882476146, "grad_norm": 1.1605415344238281, "learning_rate": 0.0004796026058850866, "loss": 1.9199905395507812, "step": 14197 }, { "epoch": 1.6520828484989527, "grad_norm": 1.569667100906372, "learning_rate": 0.000479512103733308, "loss": 1.7229700088500977, "step": 14198 }, { "epoch": 1.6521992087502908, "grad_norm": 1.0388041734695435, "learning_rate": 0.0004794214098267291, "loss": 1.9564130306243896, "step": 14199 }, { "epoch": 1.6523155690016291, "grad_norm": 1.041995644569397, "learning_rate": 0.0004793305242411237, "loss": 2.196078300476074, "step": 14200 }, { "epoch": 1.6524319292529672, "grad_norm": 3.9750823974609375, "learning_rate": 0.00047923944705242563, "loss": 2.136448860168457, "step": 14201 }, { "epoch": 1.6525482895043053, "grad_norm": 0.8396730422973633, "learning_rate": 0.00047914817833672884, "loss": 1.6332894563674927, "step": 14202 }, { "epoch": 1.6526646497556434, "grad_norm": 1.1236218214035034, "learning_rate": 0.00047905671817028746, "loss": 2.7260119915008545, "step": 14203 }, { "epoch": 1.6527810100069815, "grad_norm": 0.7120646834373474, "learning_rate": 0.00047896506662951536, "loss": 1.6474175453186035, "step": 14204 }, { "epoch": 1.6528973702583198, "grad_norm": 1.8811999559402466, "learning_rate": 0.0004788732237909862, "loss": 2.1985373497009277, "step": 14205 }, { "epoch": 1.653013730509658, "grad_norm": 1.0827215909957886, "learning_rate": 0.00047878118973143397, "loss": 1.6998341083526611, "step": 14206 }, { "epoch": 1.653130090760996, "grad_norm": 1.2493118047714233, "learning_rate": 0.0004786889645277519, "loss": 2.3996381759643555, "step": 14207 }, { "epoch": 1.6532464510123341, "grad_norm": 1.5979533195495605, "learning_rate": 0.00047859654825699316, "loss": 1.9072571992874146, "step": 14208 }, { "epoch": 1.6533628112636722, "grad_norm": 1.0841503143310547, "learning_rate": 0.0004785039409963705, "loss": 2.302497386932373, "step": 14209 }, { "epoch": 1.6534791715150106, "grad_norm": 0.9007861614227295, "learning_rate": 0.00047841114282325624, "loss": 1.9611445665359497, "step": 14210 }, { "epoch": 1.6535955317663487, "grad_norm": 0.9926010966300964, "learning_rate": 0.00047831815381518225, "loss": 1.600298523902893, "step": 14211 }, { "epoch": 1.6537118920176868, "grad_norm": 0.9873577356338501, "learning_rate": 0.0004782249740498398, "loss": 1.448281168937683, "step": 14212 }, { "epoch": 1.6538282522690249, "grad_norm": 1.5048167705535889, "learning_rate": 0.00047813160360507957, "loss": 1.8125252723693848, "step": 14213 }, { "epoch": 1.653944612520363, "grad_norm": 1.2452863454818726, "learning_rate": 0.0004780380425589115, "loss": 1.9791160821914673, "step": 14214 }, { "epoch": 1.6540609727717013, "grad_norm": 1.118888258934021, "learning_rate": 0.00047794429098950474, "loss": 1.876990795135498, "step": 14215 }, { "epoch": 1.6541773330230394, "grad_norm": 0.9349873661994934, "learning_rate": 0.00047785034897518776, "loss": 2.005978584289551, "step": 14216 }, { "epoch": 1.6542936932743775, "grad_norm": 1.2731385231018066, "learning_rate": 0.0004777562165944481, "loss": 2.483834743499756, "step": 14217 }, { "epoch": 1.6544100535257156, "grad_norm": 1.1294983625411987, "learning_rate": 0.0004776618939259322, "loss": 2.026757001876831, "step": 14218 }, { "epoch": 1.6545264137770537, "grad_norm": 1.3343335390090942, "learning_rate": 0.00047756738104844565, "loss": 2.2576451301574707, "step": 14219 }, { "epoch": 1.654642774028392, "grad_norm": 1.0995769500732422, "learning_rate": 0.0004774726780409529, "loss": 1.9819194078445435, "step": 14220 }, { "epoch": 1.65475913427973, "grad_norm": 0.8778105974197388, "learning_rate": 0.0004773777849825773, "loss": 1.3957264423370361, "step": 14221 }, { "epoch": 1.6548754945310682, "grad_norm": 1.6650924682617188, "learning_rate": 0.0004772827019526008, "loss": 2.136199951171875, "step": 14222 }, { "epoch": 1.6549918547824063, "grad_norm": 0.6749749183654785, "learning_rate": 0.00047718742903046443, "loss": 1.6500458717346191, "step": 14223 }, { "epoch": 1.6551082150337444, "grad_norm": 1.0516023635864258, "learning_rate": 0.0004770919662957675, "loss": 1.6450974941253662, "step": 14224 }, { "epoch": 1.6552245752850827, "grad_norm": 1.6402326822280884, "learning_rate": 0.0004769963138282681, "loss": 1.7517471313476562, "step": 14225 }, { "epoch": 1.6553409355364208, "grad_norm": 1.443645715713501, "learning_rate": 0.00047690047170788276, "loss": 2.2651145458221436, "step": 14226 }, { "epoch": 1.655457295787759, "grad_norm": 2.0453975200653076, "learning_rate": 0.00047680444001468647, "loss": 1.7266948223114014, "step": 14227 }, { "epoch": 1.655573656039097, "grad_norm": 1.1002216339111328, "learning_rate": 0.0004767082188289128, "loss": 1.6266103982925415, "step": 14228 }, { "epoch": 1.655690016290435, "grad_norm": 1.3201693296432495, "learning_rate": 0.00047661180823095335, "loss": 1.6912239789962769, "step": 14229 }, { "epoch": 1.6558063765417734, "grad_norm": 0.968723714351654, "learning_rate": 0.00047651520830135815, "loss": 1.9549773931503296, "step": 14230 }, { "epoch": 1.6559227367931115, "grad_norm": 1.267517328262329, "learning_rate": 0.00047641841912083525, "loss": 2.681751251220703, "step": 14231 }, { "epoch": 1.6560390970444496, "grad_norm": 1.510188341140747, "learning_rate": 0.00047632144077025106, "loss": 1.7266165018081665, "step": 14232 }, { "epoch": 1.6561554572957877, "grad_norm": 1.3761733770370483, "learning_rate": 0.0004762242733306299, "loss": 2.2616734504699707, "step": 14233 }, { "epoch": 1.6562718175471258, "grad_norm": 1.143988013267517, "learning_rate": 0.00047612691688315404, "loss": 2.970945358276367, "step": 14234 }, { "epoch": 1.6563881777984641, "grad_norm": 0.8315811157226562, "learning_rate": 0.0004760293715091637, "loss": 2.233935832977295, "step": 14235 }, { "epoch": 1.656504538049802, "grad_norm": 1.028910756111145, "learning_rate": 0.000475931637290157, "loss": 2.3555970191955566, "step": 14236 }, { "epoch": 1.6566208983011403, "grad_norm": 1.2535364627838135, "learning_rate": 0.00047583371430778967, "loss": 1.5764787197113037, "step": 14237 }, { "epoch": 1.6567372585524784, "grad_norm": 1.3060314655303955, "learning_rate": 0.00047573560264387537, "loss": 1.916617751121521, "step": 14238 }, { "epoch": 1.6568536188038165, "grad_norm": 0.9079203009605408, "learning_rate": 0.0004756373023803854, "loss": 1.4832611083984375, "step": 14239 }, { "epoch": 1.6569699790551549, "grad_norm": 1.0668866634368896, "learning_rate": 0.0004755388135994483, "loss": 2.0019497871398926, "step": 14240 }, { "epoch": 1.6570863393064927, "grad_norm": 1.1924841403961182, "learning_rate": 0.00047544013638335046, "loss": 2.5199055671691895, "step": 14241 }, { "epoch": 1.657202699557831, "grad_norm": 1.160707950592041, "learning_rate": 0.0004753412708145356, "loss": 1.422271728515625, "step": 14242 }, { "epoch": 1.6573190598091692, "grad_norm": 1.6369166374206543, "learning_rate": 0.00047524221697560476, "loss": 2.404087781906128, "step": 14243 }, { "epoch": 1.6574354200605073, "grad_norm": 0.8143041133880615, "learning_rate": 0.0004751429749493164, "loss": 1.3925371170043945, "step": 14244 }, { "epoch": 1.6575517803118456, "grad_norm": 1.0984293222427368, "learning_rate": 0.0004750435448185861, "loss": 2.005967855453491, "step": 14245 }, { "epoch": 1.6576681405631835, "grad_norm": 1.7575923204421997, "learning_rate": 0.00047494392666648657, "loss": 1.47640061378479, "step": 14246 }, { "epoch": 1.6577845008145218, "grad_norm": 1.0641449689865112, "learning_rate": 0.0004748441205762476, "loss": 1.98215651512146, "step": 14247 }, { "epoch": 1.6579008610658599, "grad_norm": 1.093598484992981, "learning_rate": 0.00047474412663125624, "loss": 1.3865058422088623, "step": 14248 }, { "epoch": 1.658017221317198, "grad_norm": 1.2436755895614624, "learning_rate": 0.00047464394491505626, "loss": 1.5218005180358887, "step": 14249 }, { "epoch": 1.6581335815685363, "grad_norm": 1.4325696229934692, "learning_rate": 0.0004745435755113483, "loss": 2.343247413635254, "step": 14250 }, { "epoch": 1.6582499418198742, "grad_norm": 1.674781322479248, "learning_rate": 0.00047444301850398986, "loss": 2.576690673828125, "step": 14251 }, { "epoch": 1.6583663020712125, "grad_norm": 0.9804346561431885, "learning_rate": 0.0004743422739769954, "loss": 1.368147850036621, "step": 14252 }, { "epoch": 1.6584826623225506, "grad_norm": 0.7971183061599731, "learning_rate": 0.0004742413420145355, "loss": 1.1598544120788574, "step": 14253 }, { "epoch": 1.6585990225738887, "grad_norm": 1.683907151222229, "learning_rate": 0.000474140222700938, "loss": 1.5957229137420654, "step": 14254 }, { "epoch": 1.658715382825227, "grad_norm": 1.2693660259246826, "learning_rate": 0.00047403891612068683, "loss": 1.9262371063232422, "step": 14255 }, { "epoch": 1.658831743076565, "grad_norm": 0.8299717903137207, "learning_rate": 0.00047393742235842245, "loss": 1.5051885843276978, "step": 14256 }, { "epoch": 1.6589481033279032, "grad_norm": 0.7865800261497498, "learning_rate": 0.0004738357414989419, "loss": 1.743201494216919, "step": 14257 }, { "epoch": 1.6590644635792413, "grad_norm": 0.8689845204353333, "learning_rate": 0.0004737338736271983, "loss": 1.7873914241790771, "step": 14258 }, { "epoch": 1.6591808238305794, "grad_norm": 1.140257716178894, "learning_rate": 0.0004736318188283012, "loss": 1.6520142555236816, "step": 14259 }, { "epoch": 1.6592971840819177, "grad_norm": 0.9227951169013977, "learning_rate": 0.0004735295771875162, "loss": 1.5099000930786133, "step": 14260 }, { "epoch": 1.6594135443332556, "grad_norm": 0.8595354557037354, "learning_rate": 0.00047342714879026505, "loss": 1.5750986337661743, "step": 14261 }, { "epoch": 1.659529904584594, "grad_norm": 1.0651295185089111, "learning_rate": 0.0004733245337221255, "loss": 1.9149044752120972, "step": 14262 }, { "epoch": 1.659646264835932, "grad_norm": 1.2976739406585693, "learning_rate": 0.0004732217320688313, "loss": 1.9814810752868652, "step": 14263 }, { "epoch": 1.6597626250872701, "grad_norm": 0.9417767524719238, "learning_rate": 0.00047311874391627214, "loss": 1.479232907295227, "step": 14264 }, { "epoch": 1.6598789853386084, "grad_norm": 1.0666115283966064, "learning_rate": 0.0004730155693504934, "loss": 1.6606578826904297, "step": 14265 }, { "epoch": 1.6599953455899463, "grad_norm": 1.2891716957092285, "learning_rate": 0.00047291220845769645, "loss": 2.548968553543091, "step": 14266 }, { "epoch": 1.6601117058412846, "grad_norm": 2.0137436389923096, "learning_rate": 0.0004728086613242379, "loss": 1.984356164932251, "step": 14267 }, { "epoch": 1.6602280660926227, "grad_norm": 0.7760180830955505, "learning_rate": 0.0004727049280366304, "loss": 1.3208069801330566, "step": 14268 }, { "epoch": 1.6603444263439608, "grad_norm": 1.0093313455581665, "learning_rate": 0.0004726010086815419, "loss": 1.6474974155426025, "step": 14269 }, { "epoch": 1.6604607865952992, "grad_norm": 1.1431550979614258, "learning_rate": 0.0004724969033457959, "loss": 1.7281110286712646, "step": 14270 }, { "epoch": 1.660577146846637, "grad_norm": 1.127326488494873, "learning_rate": 0.0004723926121163712, "loss": 2.0193259716033936, "step": 14271 }, { "epoch": 1.6606935070979754, "grad_norm": 0.9681013226509094, "learning_rate": 0.0004722881350804019, "loss": 1.2362332344055176, "step": 14272 }, { "epoch": 1.6608098673493135, "grad_norm": 1.7575184106826782, "learning_rate": 0.00047218347232517755, "loss": 2.3655197620391846, "step": 14273 }, { "epoch": 1.6609262276006516, "grad_norm": 1.0604307651519775, "learning_rate": 0.0004720786239381426, "loss": 1.8241124153137207, "step": 14274 }, { "epoch": 1.6610425878519899, "grad_norm": 1.0145988464355469, "learning_rate": 0.0004719735900068968, "loss": 1.9274102449417114, "step": 14275 }, { "epoch": 1.6611589481033278, "grad_norm": 1.1004916429519653, "learning_rate": 0.00047186837061919475, "loss": 2.0624094009399414, "step": 14276 }, { "epoch": 1.661275308354666, "grad_norm": 0.45077207684516907, "learning_rate": 0.000471762965862946, "loss": 1.02278470993042, "step": 14277 }, { "epoch": 1.6613916686060042, "grad_norm": 0.7232096195220947, "learning_rate": 0.00047165737582621526, "loss": 1.3743584156036377, "step": 14278 }, { "epoch": 1.6615080288573423, "grad_norm": 1.1117722988128662, "learning_rate": 0.00047155160059722165, "loss": 1.7751781940460205, "step": 14279 }, { "epoch": 1.6616243891086806, "grad_norm": 1.7357792854309082, "learning_rate": 0.00047144564026433924, "loss": 2.630178451538086, "step": 14280 }, { "epoch": 1.6617407493600185, "grad_norm": 0.8508338332176208, "learning_rate": 0.00047133949491609675, "loss": 1.3651784658432007, "step": 14281 }, { "epoch": 1.6618571096113568, "grad_norm": 1.9350519180297852, "learning_rate": 0.0004712331646411773, "loss": 2.1036531925201416, "step": 14282 }, { "epoch": 1.661973469862695, "grad_norm": 1.3172460794448853, "learning_rate": 0.0004711266495284189, "loss": 2.4487452507019043, "step": 14283 }, { "epoch": 1.662089830114033, "grad_norm": 0.9349721074104309, "learning_rate": 0.00047101994966681343, "loss": 1.6099276542663574, "step": 14284 }, { "epoch": 1.6622061903653713, "grad_norm": 1.2098933458328247, "learning_rate": 0.0004709130651455077, "loss": 1.9910229444503784, "step": 14285 }, { "epoch": 1.6623225506167092, "grad_norm": 1.5600370168685913, "learning_rate": 0.00047080599605380235, "loss": 2.838939905166626, "step": 14286 }, { "epoch": 1.6624389108680475, "grad_norm": 0.8417020440101624, "learning_rate": 0.0004706987424811525, "loss": 1.5663514137268066, "step": 14287 }, { "epoch": 1.6625552711193856, "grad_norm": 0.6291742324829102, "learning_rate": 0.00047059130451716725, "loss": 1.5830092430114746, "step": 14288 }, { "epoch": 1.6626716313707237, "grad_norm": 0.8462441563606262, "learning_rate": 0.00047048368225160986, "loss": 1.4645887613296509, "step": 14289 }, { "epoch": 1.662787991622062, "grad_norm": 48.41657257080078, "learning_rate": 0.00047037587577439764, "loss": 1.5720722675323486, "step": 14290 }, { "epoch": 1.6629043518734, "grad_norm": 1.2174519300460815, "learning_rate": 0.0004702678851756015, "loss": 1.5420286655426025, "step": 14291 }, { "epoch": 1.6630207121247382, "grad_norm": 1.4112145900726318, "learning_rate": 0.0004701597105454465, "loss": 1.9822206497192383, "step": 14292 }, { "epoch": 1.6631370723760763, "grad_norm": 1.005281925201416, "learning_rate": 0.00047005135197431127, "loss": 1.9730737209320068, "step": 14293 }, { "epoch": 1.6632534326274144, "grad_norm": 1.2646900415420532, "learning_rate": 0.0004699428095527283, "loss": 1.694250226020813, "step": 14294 }, { "epoch": 1.6633697928787528, "grad_norm": 1.4682841300964355, "learning_rate": 0.00046983408337138345, "loss": 1.8350780010223389, "step": 14295 }, { "epoch": 1.6634861531300906, "grad_norm": 0.8387638330459595, "learning_rate": 0.00046972517352111625, "loss": 1.6654188632965088, "step": 14296 }, { "epoch": 1.663602513381429, "grad_norm": 0.8523831367492676, "learning_rate": 0.0004696160800929197, "loss": 1.6202988624572754, "step": 14297 }, { "epoch": 1.663718873632767, "grad_norm": 1.0894513130187988, "learning_rate": 0.00046950680317794013, "loss": 2.352970838546753, "step": 14298 }, { "epoch": 1.6638352338841051, "grad_norm": 1.1414192914962769, "learning_rate": 0.0004693973428674773, "loss": 2.4142913818359375, "step": 14299 }, { "epoch": 1.6639515941354435, "grad_norm": 0.6028069853782654, "learning_rate": 0.0004692876992529838, "loss": 1.34269118309021, "step": 14300 }, { "epoch": 1.6640679543867813, "grad_norm": 0.8058368563652039, "learning_rate": 0.00046917787242606584, "loss": 1.8290642499923706, "step": 14301 }, { "epoch": 1.6641843146381197, "grad_norm": 0.9591583013534546, "learning_rate": 0.0004690678624784825, "loss": 1.672274112701416, "step": 14302 }, { "epoch": 1.6643006748894578, "grad_norm": 0.7298346161842346, "learning_rate": 0.0004689576695021458, "loss": 1.7109673023223877, "step": 14303 }, { "epoch": 1.6644170351407959, "grad_norm": 1.651451587677002, "learning_rate": 0.00046884729358912083, "loss": 2.374711036682129, "step": 14304 }, { "epoch": 1.6645333953921342, "grad_norm": 1.3717714548110962, "learning_rate": 0.0004687367348316254, "loss": 1.7578697204589844, "step": 14305 }, { "epoch": 1.664649755643472, "grad_norm": 1.061063528060913, "learning_rate": 0.00046862599332203006, "loss": 1.959167718887329, "step": 14306 }, { "epoch": 1.6647661158948104, "grad_norm": 1.8193966150283813, "learning_rate": 0.0004685150691528581, "loss": 1.6678777933120728, "step": 14307 }, { "epoch": 1.6648824761461485, "grad_norm": 1.0675698518753052, "learning_rate": 0.00046840396241678553, "loss": 1.6032638549804688, "step": 14308 }, { "epoch": 1.6649988363974866, "grad_norm": 1.1324331760406494, "learning_rate": 0.0004682926732066407, "loss": 1.5318046808242798, "step": 14309 }, { "epoch": 1.665115196648825, "grad_norm": 1.961588740348816, "learning_rate": 0.00046818120161540457, "loss": 1.7631330490112305, "step": 14310 }, { "epoch": 1.6652315569001628, "grad_norm": 0.7621225118637085, "learning_rate": 0.0004680695477362102, "loss": 2.508256673812866, "step": 14311 }, { "epoch": 1.665347917151501, "grad_norm": 1.4342883825302124, "learning_rate": 0.0004679577116623436, "loss": 2.433901786804199, "step": 14312 }, { "epoch": 1.6654642774028392, "grad_norm": 0.7669912576675415, "learning_rate": 0.0004678456934872421, "loss": 1.5594710111618042, "step": 14313 }, { "epoch": 1.6655806376541773, "grad_norm": 0.8267554044723511, "learning_rate": 0.0004677334933044959, "loss": 1.5055553913116455, "step": 14314 }, { "epoch": 1.6656969979055156, "grad_norm": 1.2286995649337769, "learning_rate": 0.0004676211112078469, "loss": 2.750880241394043, "step": 14315 }, { "epoch": 1.6658133581568535, "grad_norm": 1.4694632291793823, "learning_rate": 0.0004675085472911891, "loss": 2.0947988033294678, "step": 14316 }, { "epoch": 1.6659297184081918, "grad_norm": 1.1421211957931519, "learning_rate": 0.0004673958016485684, "loss": 2.4137752056121826, "step": 14317 }, { "epoch": 1.66604607865953, "grad_norm": 0.7818774580955505, "learning_rate": 0.0004672828743741825, "loss": 1.6920983791351318, "step": 14318 }, { "epoch": 1.666162438910868, "grad_norm": 5.560137748718262, "learning_rate": 0.000467169765562381, "loss": 1.6015398502349854, "step": 14319 }, { "epoch": 1.6662787991622063, "grad_norm": 1.1078295707702637, "learning_rate": 0.00046705647530766495, "loss": 1.9322550296783447, "step": 14320 }, { "epoch": 1.6663951594135442, "grad_norm": 1.251259684562683, "learning_rate": 0.000466943003704687, "loss": 2.094618320465088, "step": 14321 }, { "epoch": 1.6665115196648825, "grad_norm": 0.6536023020744324, "learning_rate": 0.00046682935084825164, "loss": 1.2897065877914429, "step": 14322 }, { "epoch": 1.6666278799162206, "grad_norm": 0.42729541659355164, "learning_rate": 0.0004667155168333143, "loss": 1.0147520303726196, "step": 14323 }, { "epoch": 1.6667442401675587, "grad_norm": 0.8592547178268433, "learning_rate": 0.0004666015017549822, "loss": 1.848705768585205, "step": 14324 }, { "epoch": 1.666860600418897, "grad_norm": 0.9707463979721069, "learning_rate": 0.0004664873057085137, "loss": 1.5923058986663818, "step": 14325 }, { "epoch": 1.666976960670235, "grad_norm": 1.0842692852020264, "learning_rate": 0.0004663729287893182, "loss": 1.7838268280029297, "step": 14326 }, { "epoch": 1.6670933209215733, "grad_norm": 1.1768128871917725, "learning_rate": 0.0004662583710929564, "loss": 1.8668155670166016, "step": 14327 }, { "epoch": 1.6672096811729114, "grad_norm": 0.7134014368057251, "learning_rate": 0.0004661436327151399, "loss": 1.3364449739456177, "step": 14328 }, { "epoch": 1.6673260414242494, "grad_norm": 0.9417515397071838, "learning_rate": 0.0004660287137517315, "loss": 1.8775732517242432, "step": 14329 }, { "epoch": 1.6674424016755878, "grad_norm": 1.302489161491394, "learning_rate": 0.0004659136142987445, "loss": 2.0878586769104004, "step": 14330 }, { "epoch": 1.6675587619269256, "grad_norm": 0.7367764711380005, "learning_rate": 0.0004657983344523434, "loss": 1.9740896224975586, "step": 14331 }, { "epoch": 1.667675122178264, "grad_norm": 1.2762404680252075, "learning_rate": 0.0004656828743088431, "loss": 2.0586435794830322, "step": 14332 }, { "epoch": 1.667791482429602, "grad_norm": 1.4400054216384888, "learning_rate": 0.00046556723396470927, "loss": 1.8016555309295654, "step": 14333 }, { "epoch": 1.6679078426809402, "grad_norm": 1.2221183776855469, "learning_rate": 0.0004654514135165582, "loss": 2.511378765106201, "step": 14334 }, { "epoch": 1.6680242029322785, "grad_norm": 1.0078662633895874, "learning_rate": 0.00046533541306115656, "loss": 1.718150019645691, "step": 14335 }, { "epoch": 1.6681405631836164, "grad_norm": 1.0554745197296143, "learning_rate": 0.00046521923269542137, "loss": 1.735384225845337, "step": 14336 }, { "epoch": 1.6682569234349547, "grad_norm": 0.9243166446685791, "learning_rate": 0.00046510287251642007, "loss": 1.7620271444320679, "step": 14337 }, { "epoch": 1.6683732836862928, "grad_norm": 1.3979120254516602, "learning_rate": 0.00046498633262137036, "loss": 1.952048659324646, "step": 14338 }, { "epoch": 1.6684896439376309, "grad_norm": 0.8229039311408997, "learning_rate": 0.0004648696131076399, "loss": 1.834442377090454, "step": 14339 }, { "epoch": 1.6686060041889692, "grad_norm": 1.314212679862976, "learning_rate": 0.00046475271407274654, "loss": 2.069812059402466, "step": 14340 }, { "epoch": 1.668722364440307, "grad_norm": 1.1720050573349, "learning_rate": 0.00046463563561435826, "loss": 1.9843758344650269, "step": 14341 }, { "epoch": 1.6688387246916454, "grad_norm": 1.2606805562973022, "learning_rate": 0.00046451837783029274, "loss": 1.984510898590088, "step": 14342 }, { "epoch": 1.6689550849429835, "grad_norm": 0.6410674452781677, "learning_rate": 0.00046440094081851745, "loss": 1.7194554805755615, "step": 14343 }, { "epoch": 1.6690714451943216, "grad_norm": 1.3836008310317993, "learning_rate": 0.0004642833246771498, "loss": 1.8801414966583252, "step": 14344 }, { "epoch": 1.66918780544566, "grad_norm": 1.0903148651123047, "learning_rate": 0.00046416552950445677, "loss": 1.4949042797088623, "step": 14345 }, { "epoch": 1.6693041656969978, "grad_norm": 0.8502088189125061, "learning_rate": 0.0004640475553988549, "loss": 1.303837537765503, "step": 14346 }, { "epoch": 1.6694205259483361, "grad_norm": 1.3265714645385742, "learning_rate": 0.00046392940245891013, "loss": 1.767092227935791, "step": 14347 }, { "epoch": 1.6695368861996742, "grad_norm": 0.7861955165863037, "learning_rate": 0.0004638110707833381, "loss": 1.308279275894165, "step": 14348 }, { "epoch": 1.6696532464510123, "grad_norm": 1.571479320526123, "learning_rate": 0.0004636925604710035, "loss": 2.4545555114746094, "step": 14349 }, { "epoch": 1.6697696067023506, "grad_norm": 1.646997332572937, "learning_rate": 0.00046357387162092035, "loss": 1.690112829208374, "step": 14350 }, { "epoch": 1.6698859669536885, "grad_norm": 1.3462907075881958, "learning_rate": 0.00046345500433225194, "loss": 2.095677375793457, "step": 14351 }, { "epoch": 1.6700023272050268, "grad_norm": 1.5951905250549316, "learning_rate": 0.0004633359587043105, "loss": 2.2346487045288086, "step": 14352 }, { "epoch": 1.670118687456365, "grad_norm": 1.334489345550537, "learning_rate": 0.00046321673483655734, "loss": 2.6195271015167236, "step": 14353 }, { "epoch": 1.670235047707703, "grad_norm": 0.9041761755943298, "learning_rate": 0.0004630973328286027, "loss": 1.4035826921463013, "step": 14354 }, { "epoch": 1.6703514079590414, "grad_norm": 1.069954514503479, "learning_rate": 0.00046297775278020557, "loss": 1.7599886655807495, "step": 14355 }, { "epoch": 1.6704677682103792, "grad_norm": 1.4306044578552246, "learning_rate": 0.0004628579947912739, "loss": 1.8936803340911865, "step": 14356 }, { "epoch": 1.6705841284617176, "grad_norm": 1.0083954334259033, "learning_rate": 0.00046273805896186396, "loss": 1.7011358737945557, "step": 14357 }, { "epoch": 1.6707004887130557, "grad_norm": 0.8572927117347717, "learning_rate": 0.0004626179453921808, "loss": 1.552855372428894, "step": 14358 }, { "epoch": 1.6708168489643938, "grad_norm": 1.31259286403656, "learning_rate": 0.00046249765418257804, "loss": 1.8875325918197632, "step": 14359 }, { "epoch": 1.6709332092157319, "grad_norm": 0.9751158952713013, "learning_rate": 0.00046237718543355765, "loss": 1.6796168088912964, "step": 14360 }, { "epoch": 1.67104956946707, "grad_norm": 1.2467668056488037, "learning_rate": 0.0004622565392457699, "loss": 1.814902901649475, "step": 14361 }, { "epoch": 1.6711659297184083, "grad_norm": 1.2758101224899292, "learning_rate": 0.00046213571572001344, "loss": 2.595585346221924, "step": 14362 }, { "epoch": 1.6712822899697464, "grad_norm": 0.8918552994728088, "learning_rate": 0.0004620147149572348, "loss": 1.3922717571258545, "step": 14363 }, { "epoch": 1.6713986502210845, "grad_norm": 0.6589128971099854, "learning_rate": 0.00046189353705852884, "loss": 1.258653163909912, "step": 14364 }, { "epoch": 1.6715150104724226, "grad_norm": 0.8854175806045532, "learning_rate": 0.00046177218212513837, "loss": 2.0180282592773438, "step": 14365 }, { "epoch": 1.6716313707237607, "grad_norm": 0.9107860326766968, "learning_rate": 0.00046165065025845404, "loss": 1.6820154190063477, "step": 14366 }, { "epoch": 1.671747730975099, "grad_norm": 1.1292517185211182, "learning_rate": 0.0004615289415600144, "loss": 2.3366305828094482, "step": 14367 }, { "epoch": 1.671864091226437, "grad_norm": 1.4776884317398071, "learning_rate": 0.0004614070561315057, "loss": 2.850090742111206, "step": 14368 }, { "epoch": 1.6719804514777752, "grad_norm": 0.6735110878944397, "learning_rate": 0.00046128499407476186, "loss": 1.4742437601089478, "step": 14369 }, { "epoch": 1.6720968117291133, "grad_norm": 1.2408794164657593, "learning_rate": 0.00046116275549176435, "loss": 2.4393820762634277, "step": 14370 }, { "epoch": 1.6722131719804514, "grad_norm": 1.1919015645980835, "learning_rate": 0.0004610403404846421, "loss": 1.704784631729126, "step": 14371 }, { "epoch": 1.6723295322317897, "grad_norm": 0.9146542549133301, "learning_rate": 0.0004609177491556715, "loss": 1.6095836162567139, "step": 14372 }, { "epoch": 1.6724458924831278, "grad_norm": 1.4000608921051025, "learning_rate": 0.0004607949816072764, "loss": 2.2160046100616455, "step": 14373 }, { "epoch": 1.672562252734466, "grad_norm": 1.0567482709884644, "learning_rate": 0.0004606720379420275, "loss": 1.6064913272857666, "step": 14374 }, { "epoch": 1.672678612985804, "grad_norm": 0.9511155486106873, "learning_rate": 0.000460548918262643, "loss": 1.8880410194396973, "step": 14375 }, { "epoch": 1.672794973237142, "grad_norm": 0.7647130489349365, "learning_rate": 0.0004604256226719879, "loss": 1.1983760595321655, "step": 14376 }, { "epoch": 1.6729113334884804, "grad_norm": 0.7793443202972412, "learning_rate": 0.00046030215127307436, "loss": 2.0559675693511963, "step": 14377 }, { "epoch": 1.6730276937398185, "grad_norm": 1.78360116481781, "learning_rate": 0.00046017850416906137, "loss": 1.6894943714141846, "step": 14378 }, { "epoch": 1.6731440539911566, "grad_norm": 1.1542036533355713, "learning_rate": 0.00046005468146325464, "loss": 1.9886274337768555, "step": 14379 }, { "epoch": 1.6732604142424947, "grad_norm": 1.1473650932312012, "learning_rate": 0.0004599306832591067, "loss": 2.593876838684082, "step": 14380 }, { "epoch": 1.6733767744938328, "grad_norm": 0.8159880042076111, "learning_rate": 0.00045980650966021664, "loss": 1.5721886157989502, "step": 14381 }, { "epoch": 1.6734931347451711, "grad_norm": 1.2116469144821167, "learning_rate": 0.0004596821607703302, "loss": 2.468682289123535, "step": 14382 }, { "epoch": 1.6736094949965092, "grad_norm": 0.931789755821228, "learning_rate": 0.0004595576366933394, "loss": 2.157589912414551, "step": 14383 }, { "epoch": 1.6737258552478473, "grad_norm": 1.3604382276535034, "learning_rate": 0.00045943293753328274, "loss": 1.8959836959838867, "step": 14384 }, { "epoch": 1.6738422154991854, "grad_norm": 1.2877730131149292, "learning_rate": 0.000459308063394345, "loss": 2.3748974800109863, "step": 14385 }, { "epoch": 1.6739585757505235, "grad_norm": 0.8822826743125916, "learning_rate": 0.00045918301438085707, "loss": 1.8139601945877075, "step": 14386 }, { "epoch": 1.6740749360018619, "grad_norm": 0.9903379082679749, "learning_rate": 0.00045905779059729615, "loss": 1.3391755819320679, "step": 14387 }, { "epoch": 1.6741912962532, "grad_norm": 0.8071383833885193, "learning_rate": 0.0004589323921482851, "loss": 1.3598999977111816, "step": 14388 }, { "epoch": 1.674307656504538, "grad_norm": 0.8950072526931763, "learning_rate": 0.000458806819138593, "loss": 1.6083879470825195, "step": 14389 }, { "epoch": 1.6744240167558762, "grad_norm": 1.101189136505127, "learning_rate": 0.0004586810716731348, "loss": 1.7182480096817017, "step": 14390 }, { "epoch": 1.6745403770072143, "grad_norm": 1.1245791912078857, "learning_rate": 0.000458555149856971, "loss": 1.7100272178649902, "step": 14391 }, { "epoch": 1.6746567372585526, "grad_norm": 1.3054533004760742, "learning_rate": 0.00045842905379530794, "loss": 2.424476146697998, "step": 14392 }, { "epoch": 1.6747730975098907, "grad_norm": 0.6612729430198669, "learning_rate": 0.00045830278359349744, "loss": 1.1619675159454346, "step": 14393 }, { "epoch": 1.6748894577612288, "grad_norm": 0.901311993598938, "learning_rate": 0.00045817633935703675, "loss": 1.3148269653320312, "step": 14394 }, { "epoch": 1.6750058180125669, "grad_norm": 0.8457803130149841, "learning_rate": 0.00045804972119156874, "loss": 1.5443520545959473, "step": 14395 }, { "epoch": 1.675122178263905, "grad_norm": 0.7049079537391663, "learning_rate": 0.00045792292920288145, "loss": 1.8837858438491821, "step": 14396 }, { "epoch": 1.6752385385152433, "grad_norm": 0.7531070709228516, "learning_rate": 0.0004577959634969082, "loss": 1.541763186454773, "step": 14397 }, { "epoch": 1.6753548987665814, "grad_norm": 0.8228740692138672, "learning_rate": 0.0004576688241797274, "loss": 1.7747610807418823, "step": 14398 }, { "epoch": 1.6754712590179195, "grad_norm": 1.1977919340133667, "learning_rate": 0.00045754151135756247, "loss": 2.2812228202819824, "step": 14399 }, { "epoch": 1.6755876192692576, "grad_norm": 1.506848931312561, "learning_rate": 0.0004574140251367819, "loss": 2.7902512550354004, "step": 14400 }, { "epoch": 1.6757039795205957, "grad_norm": 1.048891544342041, "learning_rate": 0.00045728636562389894, "loss": 1.900012731552124, "step": 14401 }, { "epoch": 1.675820339771934, "grad_norm": 0.8851274251937866, "learning_rate": 0.00045715853292557177, "loss": 1.651383876800537, "step": 14402 }, { "epoch": 1.6759367000232719, "grad_norm": 1.1442770957946777, "learning_rate": 0.0004570305271486032, "loss": 1.6401026248931885, "step": 14403 }, { "epoch": 1.6760530602746102, "grad_norm": 1.145787000656128, "learning_rate": 0.00045690234839994054, "loss": 1.8705496788024902, "step": 14404 }, { "epoch": 1.6761694205259483, "grad_norm": 1.1127982139587402, "learning_rate": 0.00045677399678667577, "loss": 1.7980060577392578, "step": 14405 }, { "epoch": 1.6762857807772864, "grad_norm": 1.3235507011413574, "learning_rate": 0.0004566454724160452, "loss": 1.8992547988891602, "step": 14406 }, { "epoch": 1.6764021410286247, "grad_norm": 1.343353509902954, "learning_rate": 0.0004565167753954296, "loss": 1.6698180437088013, "step": 14407 }, { "epoch": 1.6765185012799626, "grad_norm": 0.715912401676178, "learning_rate": 0.00045638790583235367, "loss": 1.8724232912063599, "step": 14408 }, { "epoch": 1.676634861531301, "grad_norm": 1.193069577217102, "learning_rate": 0.00045625886383448666, "loss": 1.5107643604278564, "step": 14409 }, { "epoch": 1.676751221782639, "grad_norm": 2.222001075744629, "learning_rate": 0.00045612964950964176, "loss": 1.6284383535385132, "step": 14410 }, { "epoch": 1.6768675820339771, "grad_norm": 1.4063193798065186, "learning_rate": 0.0004560002629657759, "loss": 1.617874264717102, "step": 14411 }, { "epoch": 1.6769839422853154, "grad_norm": 1.053439736366272, "learning_rate": 0.0004558707043109903, "loss": 2.183763265609741, "step": 14412 }, { "epoch": 1.6771003025366533, "grad_norm": 1.0939931869506836, "learning_rate": 0.00045574097365352976, "loss": 2.0605502128601074, "step": 14413 }, { "epoch": 1.6772166627879916, "grad_norm": 1.1630011796951294, "learning_rate": 0.0004556110711017826, "loss": 2.7790188789367676, "step": 14414 }, { "epoch": 1.6773330230393297, "grad_norm": 0.791286826133728, "learning_rate": 0.0004554809967642811, "loss": 1.4657642841339111, "step": 14415 }, { "epoch": 1.6774493832906678, "grad_norm": 0.9327648282051086, "learning_rate": 0.0004553507507497009, "loss": 1.8971151113510132, "step": 14416 }, { "epoch": 1.6775657435420062, "grad_norm": 1.0031310319900513, "learning_rate": 0.00045522033316686107, "loss": 1.5791053771972656, "step": 14417 }, { "epoch": 1.677682103793344, "grad_norm": 1.1131250858306885, "learning_rate": 0.0004550897441247241, "loss": 1.9023990631103516, "step": 14418 }, { "epoch": 1.6777984640446824, "grad_norm": 1.1773664951324463, "learning_rate": 0.00045495898373239566, "loss": 2.3163657188415527, "step": 14419 }, { "epoch": 1.6779148242960205, "grad_norm": 1.2410277128219604, "learning_rate": 0.0004548280520991245, "loss": 2.3610925674438477, "step": 14420 }, { "epoch": 1.6780311845473586, "grad_norm": 0.8803532719612122, "learning_rate": 0.0004546969493343026, "loss": 1.7080235481262207, "step": 14421 }, { "epoch": 1.6781475447986969, "grad_norm": 0.8651308417320251, "learning_rate": 0.00045456567554746485, "loss": 1.6604299545288086, "step": 14422 }, { "epoch": 1.6782639050500348, "grad_norm": 1.0024644136428833, "learning_rate": 0.0004544342308482891, "loss": 1.6175518035888672, "step": 14423 }, { "epoch": 1.678380265301373, "grad_norm": 1.223852276802063, "learning_rate": 0.0004543026153465959, "loss": 2.088239908218384, "step": 14424 }, { "epoch": 1.6784966255527112, "grad_norm": 1.037798285484314, "learning_rate": 0.00045417082915234845, "loss": 1.6645467281341553, "step": 14425 }, { "epoch": 1.6786129858040493, "grad_norm": 1.0953747034072876, "learning_rate": 0.00045403887237565276, "loss": 1.6974061727523804, "step": 14426 }, { "epoch": 1.6787293460553876, "grad_norm": 1.7123379707336426, "learning_rate": 0.00045390674512675713, "loss": 1.9605151414871216, "step": 14427 }, { "epoch": 1.6788457063067255, "grad_norm": 1.2537187337875366, "learning_rate": 0.0004537744475160525, "loss": 2.100750207901001, "step": 14428 }, { "epoch": 1.6789620665580638, "grad_norm": 1.5187180042266846, "learning_rate": 0.00045364197965407196, "loss": 1.9772367477416992, "step": 14429 }, { "epoch": 1.679078426809402, "grad_norm": 1.0303916931152344, "learning_rate": 0.000453509341651491, "loss": 2.1308062076568604, "step": 14430 }, { "epoch": 1.67919478706074, "grad_norm": 1.3351987600326538, "learning_rate": 0.00045337653361912715, "loss": 1.859034538269043, "step": 14431 }, { "epoch": 1.6793111473120783, "grad_norm": 1.2137205600738525, "learning_rate": 0.0004532435556679401, "loss": 1.9467493295669556, "step": 14432 }, { "epoch": 1.6794275075634162, "grad_norm": 0.8170499205589294, "learning_rate": 0.00045311040790903137, "loss": 1.464565396308899, "step": 14433 }, { "epoch": 1.6795438678147545, "grad_norm": 1.3854081630706787, "learning_rate": 0.00045297709045364444, "loss": 1.9544265270233154, "step": 14434 }, { "epoch": 1.6796602280660926, "grad_norm": 0.9672698378562927, "learning_rate": 0.00045284360341316457, "loss": 1.6992816925048828, "step": 14435 }, { "epoch": 1.6797765883174307, "grad_norm": 1.0954002141952515, "learning_rate": 0.0004527099468991187, "loss": 1.8034058809280396, "step": 14436 }, { "epoch": 1.679892948568769, "grad_norm": 1.395764708518982, "learning_rate": 0.00045257612102317537, "loss": 1.8621885776519775, "step": 14437 }, { "epoch": 1.680009308820107, "grad_norm": 0.7315537333488464, "learning_rate": 0.0004524421258971445, "loss": 1.6300060749053955, "step": 14438 }, { "epoch": 1.6801256690714452, "grad_norm": 0.9416141510009766, "learning_rate": 0.00045230796163297767, "loss": 2.081789016723633, "step": 14439 }, { "epoch": 1.6802420293227833, "grad_norm": 0.956331193447113, "learning_rate": 0.0004521736283427675, "loss": 1.520169734954834, "step": 14440 }, { "epoch": 1.6803583895741214, "grad_norm": 0.8726702928543091, "learning_rate": 0.0004520391261387481, "loss": 2.1426806449890137, "step": 14441 }, { "epoch": 1.6804747498254597, "grad_norm": 1.198778748512268, "learning_rate": 0.00045190445513329443, "loss": 2.0272064208984375, "step": 14442 }, { "epoch": 1.6805911100767976, "grad_norm": 1.6306049823760986, "learning_rate": 0.0004517696154389226, "loss": 1.9351035356521606, "step": 14443 }, { "epoch": 1.680707470328136, "grad_norm": 1.074625849723816, "learning_rate": 0.0004516346071682898, "loss": 2.363513946533203, "step": 14444 }, { "epoch": 1.680823830579474, "grad_norm": 1.5240715742111206, "learning_rate": 0.00045149943043419383, "loss": 2.538891077041626, "step": 14445 }, { "epoch": 1.6809401908308121, "grad_norm": 1.312880516052246, "learning_rate": 0.0004513640853495733, "loss": 2.428663969039917, "step": 14446 }, { "epoch": 1.6810565510821505, "grad_norm": 1.1988297700881958, "learning_rate": 0.00045122857202750774, "loss": 1.7411894798278809, "step": 14447 }, { "epoch": 1.6811729113334883, "grad_norm": 1.112514853477478, "learning_rate": 0.00045109289058121673, "loss": 2.4528229236602783, "step": 14448 }, { "epoch": 1.6812892715848267, "grad_norm": 0.8453183174133301, "learning_rate": 0.0004509570411240608, "loss": 2.334167003631592, "step": 14449 }, { "epoch": 1.6814056318361648, "grad_norm": 0.9951156377792358, "learning_rate": 0.00045082102376954055, "loss": 2.251765251159668, "step": 14450 }, { "epoch": 1.6815219920875029, "grad_norm": 1.149295687675476, "learning_rate": 0.000450684838631297, "loss": 2.3570549488067627, "step": 14451 }, { "epoch": 1.6816383523388412, "grad_norm": 1.0078777074813843, "learning_rate": 0.0004505484858231114, "loss": 2.0023021697998047, "step": 14452 }, { "epoch": 1.681754712590179, "grad_norm": 0.9176937937736511, "learning_rate": 0.0004504119654589048, "loss": 1.7231862545013428, "step": 14453 }, { "epoch": 1.6818710728415174, "grad_norm": 1.1789554357528687, "learning_rate": 0.00045027527765273856, "loss": 2.411233425140381, "step": 14454 }, { "epoch": 1.6819874330928555, "grad_norm": 1.5170482397079468, "learning_rate": 0.0004501384225188139, "loss": 1.4816393852233887, "step": 14455 }, { "epoch": 1.6821037933441936, "grad_norm": 0.9536398649215698, "learning_rate": 0.00045000140017147154, "loss": 1.9885942935943604, "step": 14456 }, { "epoch": 1.682220153595532, "grad_norm": 1.3799848556518555, "learning_rate": 0.00044986421072519236, "loss": 1.923412561416626, "step": 14457 }, { "epoch": 1.6823365138468698, "grad_norm": 1.2973558902740479, "learning_rate": 0.00044972685429459645, "loss": 1.881055474281311, "step": 14458 }, { "epoch": 1.682452874098208, "grad_norm": 1.42600679397583, "learning_rate": 0.0004495893309944437, "loss": 1.2718610763549805, "step": 14459 }, { "epoch": 1.6825692343495462, "grad_norm": 1.590488314628601, "learning_rate": 0.0004494516409396331, "loss": 2.2266340255737305, "step": 14460 }, { "epoch": 1.6826855946008843, "grad_norm": 1.2115216255187988, "learning_rate": 0.00044931378424520324, "loss": 1.475005865097046, "step": 14461 }, { "epoch": 1.6828019548522226, "grad_norm": 0.8352301120758057, "learning_rate": 0.00044917576102633197, "loss": 1.750140905380249, "step": 14462 }, { "epoch": 1.6829183151035605, "grad_norm": 0.8688977360725403, "learning_rate": 0.000449037571398336, "loss": 1.4904452562332153, "step": 14463 }, { "epoch": 1.6830346753548988, "grad_norm": 0.8718092441558838, "learning_rate": 0.00044889921547667134, "loss": 1.396476149559021, "step": 14464 }, { "epoch": 1.683151035606237, "grad_norm": 1.2647693157196045, "learning_rate": 0.0004487606933769327, "loss": 2.1380326747894287, "step": 14465 }, { "epoch": 1.683267395857575, "grad_norm": 1.7000569105148315, "learning_rate": 0.00044862200521485386, "loss": 1.6403870582580566, "step": 14466 }, { "epoch": 1.6833837561089133, "grad_norm": 6.004940032958984, "learning_rate": 0.0004484831511063071, "loss": 2.0329575538635254, "step": 14467 }, { "epoch": 1.6835001163602512, "grad_norm": 1.5540523529052734, "learning_rate": 0.0004483441311673036, "loss": 2.1464433670043945, "step": 14468 }, { "epoch": 1.6836164766115895, "grad_norm": 1.3436899185180664, "learning_rate": 0.00044820494551399295, "loss": 1.99034583568573, "step": 14469 }, { "epoch": 1.6837328368629276, "grad_norm": 1.016822099685669, "learning_rate": 0.00044806559426266305, "loss": 1.7514162063598633, "step": 14470 }, { "epoch": 1.6838491971142657, "grad_norm": 1.2096748352050781, "learning_rate": 0.00044792607752974057, "loss": 1.8020585775375366, "step": 14471 }, { "epoch": 1.683965557365604, "grad_norm": 1.0932425260543823, "learning_rate": 0.00044778639543179, "loss": 1.5374300479888916, "step": 14472 }, { "epoch": 1.684081917616942, "grad_norm": 1.2517632246017456, "learning_rate": 0.0004476465480855142, "loss": 1.6224998235702515, "step": 14473 }, { "epoch": 1.6841982778682802, "grad_norm": 1.0221889019012451, "learning_rate": 0.00044750653560775413, "loss": 1.3934967517852783, "step": 14474 }, { "epoch": 1.6843146381196183, "grad_norm": 0.545113742351532, "learning_rate": 0.00044736635811548864, "loss": 1.196605920791626, "step": 14475 }, { "epoch": 1.6844309983709564, "grad_norm": 0.8098474740982056, "learning_rate": 0.0004472260157258344, "loss": 1.4238163232803345, "step": 14476 }, { "epoch": 1.6845473586222948, "grad_norm": 1.039687991142273, "learning_rate": 0.00044708550855604597, "loss": 2.5355093479156494, "step": 14477 }, { "epoch": 1.6846637188736326, "grad_norm": 0.8199695944786072, "learning_rate": 0.0004469448367235155, "loss": 1.8399488925933838, "step": 14478 }, { "epoch": 1.684780079124971, "grad_norm": 1.327901005744934, "learning_rate": 0.00044680400034577265, "loss": 1.8514904975891113, "step": 14479 }, { "epoch": 1.684896439376309, "grad_norm": 0.9487435817718506, "learning_rate": 0.00044666299954048485, "loss": 1.76180899143219, "step": 14480 }, { "epoch": 1.6850127996276472, "grad_norm": 0.8265935778617859, "learning_rate": 0.0004465218344254565, "loss": 1.6934621334075928, "step": 14481 }, { "epoch": 1.6851291598789855, "grad_norm": 0.752443790435791, "learning_rate": 0.00044638050511862956, "loss": 1.6484127044677734, "step": 14482 }, { "epoch": 1.6852455201303234, "grad_norm": 1.4416611194610596, "learning_rate": 0.000446239011738083, "loss": 2.4147300720214844, "step": 14483 }, { "epoch": 1.6853618803816617, "grad_norm": 0.8657114505767822, "learning_rate": 0.0004460973544020331, "loss": 1.5732800960540771, "step": 14484 }, { "epoch": 1.6854782406329998, "grad_norm": 1.0954419374465942, "learning_rate": 0.000445955533228833, "loss": 2.100322723388672, "step": 14485 }, { "epoch": 1.6855946008843379, "grad_norm": 1.2903807163238525, "learning_rate": 0.0004458135483369726, "loss": 2.448147773742676, "step": 14486 }, { "epoch": 1.6857109611356762, "grad_norm": 1.321593165397644, "learning_rate": 0.0004456713998450787, "loss": 2.2815048694610596, "step": 14487 }, { "epoch": 1.685827321387014, "grad_norm": 1.2028167247772217, "learning_rate": 0.0004455290878719147, "loss": 2.000664710998535, "step": 14488 }, { "epoch": 1.6859436816383524, "grad_norm": 0.7407641410827637, "learning_rate": 0.0004453866125363809, "loss": 1.3918389081954956, "step": 14489 }, { "epoch": 1.6860600418896905, "grad_norm": 0.8911927938461304, "learning_rate": 0.00044524397395751366, "loss": 1.6118841171264648, "step": 14490 }, { "epoch": 1.6861764021410286, "grad_norm": 1.1977550983428955, "learning_rate": 0.000445101172254486, "loss": 2.1932730674743652, "step": 14491 }, { "epoch": 1.686292762392367, "grad_norm": 0.8005632162094116, "learning_rate": 0.00044495820754660703, "loss": 1.3579288721084595, "step": 14492 }, { "epoch": 1.6864091226437048, "grad_norm": 1.1247920989990234, "learning_rate": 0.0004448150799533224, "loss": 1.6297773122787476, "step": 14493 }, { "epoch": 1.6865254828950431, "grad_norm": 0.9863988161087036, "learning_rate": 0.00044467178959421336, "loss": 2.0772006511688232, "step": 14494 }, { "epoch": 1.6866418431463812, "grad_norm": 1.0843658447265625, "learning_rate": 0.00044452833658899753, "loss": 2.535749912261963, "step": 14495 }, { "epoch": 1.6867582033977193, "grad_norm": 1.6167622804641724, "learning_rate": 0.0004443847210575284, "loss": 2.0427114963531494, "step": 14496 }, { "epoch": 1.6868745636490576, "grad_norm": 1.0417568683624268, "learning_rate": 0.0004442409431197949, "loss": 1.7790547609329224, "step": 14497 }, { "epoch": 1.6869909239003955, "grad_norm": 1.0708987712860107, "learning_rate": 0.000444097002895922, "loss": 1.5376204252243042, "step": 14498 }, { "epoch": 1.6871072841517338, "grad_norm": 0.9515373110771179, "learning_rate": 0.00044395290050617034, "loss": 1.9542659521102905, "step": 14499 }, { "epoch": 1.687223644403072, "grad_norm": 1.212882399559021, "learning_rate": 0.0004438086360709356, "loss": 2.1823925971984863, "step": 14500 }, { "epoch": 1.68734000465441, "grad_norm": 1.852734923362732, "learning_rate": 0.00044366420971074917, "loss": 2.1332507133483887, "step": 14501 }, { "epoch": 1.6874563649057484, "grad_norm": 1.1336870193481445, "learning_rate": 0.0004435196215462777, "loss": 1.9620170593261719, "step": 14502 }, { "epoch": 1.6875727251570862, "grad_norm": 1.3269160985946655, "learning_rate": 0.00044337487169832297, "loss": 1.7044566869735718, "step": 14503 }, { "epoch": 1.6876890854084245, "grad_norm": 1.2652443647384644, "learning_rate": 0.00044322996028782187, "loss": 2.418585777282715, "step": 14504 }, { "epoch": 1.6878054456597626, "grad_norm": 1.101853847503662, "learning_rate": 0.0004430848874358462, "loss": 1.835364580154419, "step": 14505 }, { "epoch": 1.6879218059111007, "grad_norm": 1.1771087646484375, "learning_rate": 0.0004429396532636029, "loss": 2.280453681945801, "step": 14506 }, { "epoch": 1.688038166162439, "grad_norm": 1.0778816938400269, "learning_rate": 0.00044279425789243326, "loss": 1.8918564319610596, "step": 14507 }, { "epoch": 1.688154526413777, "grad_norm": 2.2638089656829834, "learning_rate": 0.00044264870144381365, "loss": 2.120936632156372, "step": 14508 }, { "epoch": 1.6882708866651153, "grad_norm": 1.5840755701065063, "learning_rate": 0.0004425029840393548, "loss": 2.474555492401123, "step": 14509 }, { "epoch": 1.6883872469164534, "grad_norm": 1.2547578811645508, "learning_rate": 0.000442357105800802, "loss": 2.1182444095611572, "step": 14510 }, { "epoch": 1.6885036071677915, "grad_norm": 1.333574891090393, "learning_rate": 0.00044221106685003476, "loss": 2.2486119270324707, "step": 14511 }, { "epoch": 1.6886199674191298, "grad_norm": 0.8789123296737671, "learning_rate": 0.0004420648673090672, "loss": 1.4712690114974976, "step": 14512 }, { "epoch": 1.6887363276704677, "grad_norm": 0.8406370878219604, "learning_rate": 0.0004419185073000473, "loss": 1.8239712715148926, "step": 14513 }, { "epoch": 1.688852687921806, "grad_norm": 0.9482595920562744, "learning_rate": 0.0004417719869452572, "loss": 2.04557466506958, "step": 14514 }, { "epoch": 1.688969048173144, "grad_norm": 3.314279317855835, "learning_rate": 0.00044162530636711306, "loss": 2.078979015350342, "step": 14515 }, { "epoch": 1.6890854084244822, "grad_norm": 1.0393543243408203, "learning_rate": 0.0004414784656881649, "loss": 1.8642830848693848, "step": 14516 }, { "epoch": 1.6892017686758205, "grad_norm": 1.2247556447982788, "learning_rate": 0.0004413314650310962, "loss": 2.293726682662964, "step": 14517 }, { "epoch": 1.6893181289271584, "grad_norm": 1.375605821609497, "learning_rate": 0.0004411843045187248, "loss": 1.4714590311050415, "step": 14518 }, { "epoch": 1.6894344891784967, "grad_norm": 1.5504939556121826, "learning_rate": 0.0004410369842740014, "loss": 1.8845738172531128, "step": 14519 }, { "epoch": 1.6895508494298348, "grad_norm": 0.6102472543716431, "learning_rate": 0.0004408895044200104, "loss": 1.2990126609802246, "step": 14520 }, { "epoch": 1.689667209681173, "grad_norm": 1.0309783220291138, "learning_rate": 0.0004407418650799697, "loss": 1.7368924617767334, "step": 14521 }, { "epoch": 1.6897835699325112, "grad_norm": 1.0985028743743896, "learning_rate": 0.0004405940663772302, "loss": 2.617130756378174, "step": 14522 }, { "epoch": 1.689899930183849, "grad_norm": 0.7901005744934082, "learning_rate": 0.00044044610843527616, "loss": 1.8943946361541748, "step": 14523 }, { "epoch": 1.6900162904351874, "grad_norm": 0.7783165574073792, "learning_rate": 0.00044029799137772473, "loss": 1.5206670761108398, "step": 14524 }, { "epoch": 1.6901326506865255, "grad_norm": 1.3290563821792603, "learning_rate": 0.00044014971532832617, "loss": 2.504664421081543, "step": 14525 }, { "epoch": 1.6902490109378636, "grad_norm": 0.6549190282821655, "learning_rate": 0.00044000128041096325, "loss": 1.821958303451538, "step": 14526 }, { "epoch": 1.6903653711892017, "grad_norm": 1.3502954244613647, "learning_rate": 0.0004398526867496519, "loss": 2.0521302223205566, "step": 14527 }, { "epoch": 1.6904817314405398, "grad_norm": 0.9921606779098511, "learning_rate": 0.00043970393446854036, "loss": 1.8669599294662476, "step": 14528 }, { "epoch": 1.6905980916918781, "grad_norm": 0.9523692727088928, "learning_rate": 0.0004395550236919095, "loss": 1.6832225322723389, "step": 14529 }, { "epoch": 1.6907144519432162, "grad_norm": 0.6696590185165405, "learning_rate": 0.00043940595454417257, "loss": 1.4521636962890625, "step": 14530 }, { "epoch": 1.6908308121945543, "grad_norm": 1.1831012964248657, "learning_rate": 0.0004392567271498753, "loss": 1.7125447988510132, "step": 14531 }, { "epoch": 1.6909471724458924, "grad_norm": 0.9138147234916687, "learning_rate": 0.0004391073416336955, "loss": 2.247255325317383, "step": 14532 }, { "epoch": 1.6910635326972305, "grad_norm": 0.76544189453125, "learning_rate": 0.00043895779812044296, "loss": 2.085869789123535, "step": 14533 }, { "epoch": 1.6911798929485689, "grad_norm": 1.44756281375885, "learning_rate": 0.00043880809673505975, "loss": 1.7660564184188843, "step": 14534 }, { "epoch": 1.691296253199907, "grad_norm": 1.0784941911697388, "learning_rate": 0.00043865823760261963, "loss": 1.581678032875061, "step": 14535 }, { "epoch": 1.691412613451245, "grad_norm": 1.3378355503082275, "learning_rate": 0.00043850822084832823, "loss": 1.7822582721710205, "step": 14536 }, { "epoch": 1.6915289737025831, "grad_norm": 1.1319507360458374, "learning_rate": 0.000438358046597523, "loss": 2.2034366130828857, "step": 14537 }, { "epoch": 1.6916453339539212, "grad_norm": 0.9991804957389832, "learning_rate": 0.0004382077149756727, "loss": 1.1289747953414917, "step": 14538 }, { "epoch": 1.6917616942052596, "grad_norm": 2.4923858642578125, "learning_rate": 0.0004380572261083778, "loss": 1.899792194366455, "step": 14539 }, { "epoch": 1.6918780544565977, "grad_norm": 1.5185842514038086, "learning_rate": 0.00043790658012137, "loss": 1.9044957160949707, "step": 14540 }, { "epoch": 1.6919944147079358, "grad_norm": 1.155356764793396, "learning_rate": 0.00043775577714051244, "loss": 1.9498586654663086, "step": 14541 }, { "epoch": 1.6921107749592739, "grad_norm": 0.9382893443107605, "learning_rate": 0.00043760481729179925, "loss": 1.772183895111084, "step": 14542 }, { "epoch": 1.692227135210612, "grad_norm": 1.1253725290298462, "learning_rate": 0.00043745370070135574, "loss": 2.320005178451538, "step": 14543 }, { "epoch": 1.6923434954619503, "grad_norm": 1.011837124824524, "learning_rate": 0.0004373024274954381, "loss": 2.065251350402832, "step": 14544 }, { "epoch": 1.6924598557132884, "grad_norm": 0.9533676505088806, "learning_rate": 0.00043715099780043345, "loss": 1.5838638544082642, "step": 14545 }, { "epoch": 1.6925762159646265, "grad_norm": 0.7320411205291748, "learning_rate": 0.0004369994117428595, "loss": 1.800782561302185, "step": 14546 }, { "epoch": 1.6926925762159646, "grad_norm": 0.7918957471847534, "learning_rate": 0.0004368476694493649, "loss": 1.5734047889709473, "step": 14547 }, { "epoch": 1.6928089364673027, "grad_norm": 1.2249506711959839, "learning_rate": 0.00043669577104672844, "loss": 2.0875954627990723, "step": 14548 }, { "epoch": 1.692925296718641, "grad_norm": 2.3378403186798096, "learning_rate": 0.00043654371666185964, "loss": 1.6451853513717651, "step": 14549 }, { "epoch": 1.693041656969979, "grad_norm": 2.3207714557647705, "learning_rate": 0.00043639150642179827, "loss": 2.6227409839630127, "step": 14550 }, { "epoch": 1.6931580172213172, "grad_norm": 1.552559494972229, "learning_rate": 0.00043623914045371425, "loss": 1.7543151378631592, "step": 14551 }, { "epoch": 1.6932743774726553, "grad_norm": 1.4205479621887207, "learning_rate": 0.00043608661888490765, "loss": 2.3277883529663086, "step": 14552 }, { "epoch": 1.6933907377239934, "grad_norm": 1.349165678024292, "learning_rate": 0.00043593394184280844, "loss": 1.6704440116882324, "step": 14553 }, { "epoch": 1.6935070979753317, "grad_norm": 1.0765706300735474, "learning_rate": 0.0004357811094549767, "loss": 1.892680287361145, "step": 14554 }, { "epoch": 1.6936234582266698, "grad_norm": 1.3510222434997559, "learning_rate": 0.0004356281218491023, "loss": 1.754220724105835, "step": 14555 }, { "epoch": 1.693739818478008, "grad_norm": 1.5255622863769531, "learning_rate": 0.0004354749791530044, "loss": 1.6947778463363647, "step": 14556 }, { "epoch": 1.693856178729346, "grad_norm": 1.0699067115783691, "learning_rate": 0.0004353216814946321, "loss": 1.74424147605896, "step": 14557 }, { "epoch": 1.6939725389806841, "grad_norm": 0.8754026293754578, "learning_rate": 0.0004351682290020641, "loss": 1.2306478023529053, "step": 14558 }, { "epoch": 1.6940888992320224, "grad_norm": 0.9939811825752258, "learning_rate": 0.00043501462180350794, "loss": 1.9480644464492798, "step": 14559 }, { "epoch": 1.6942052594833605, "grad_norm": 1.0039832592010498, "learning_rate": 0.00043486086002730097, "loss": 1.8395882844924927, "step": 14560 }, { "epoch": 1.6943216197346986, "grad_norm": 4.848010063171387, "learning_rate": 0.0004347069438019092, "loss": 1.9747562408447266, "step": 14561 }, { "epoch": 1.6944379799860367, "grad_norm": 1.0013227462768555, "learning_rate": 0.00043455287325592807, "loss": 1.878964900970459, "step": 14562 }, { "epoch": 1.6945543402373748, "grad_norm": 1.2746925354003906, "learning_rate": 0.0004343986485180818, "loss": 1.869282841682434, "step": 14563 }, { "epoch": 1.6946707004887132, "grad_norm": 0.7376132607460022, "learning_rate": 0.0004342442697172234, "loss": 1.2661948204040527, "step": 14564 }, { "epoch": 1.694787060740051, "grad_norm": 1.272990107536316, "learning_rate": 0.00043408973698233467, "loss": 1.5460591316223145, "step": 14565 }, { "epoch": 1.6949034209913894, "grad_norm": 1.6754803657531738, "learning_rate": 0.0004339350504425259, "loss": 1.882988452911377, "step": 14566 }, { "epoch": 1.6950197812427275, "grad_norm": 1.5118408203125, "learning_rate": 0.00043378021022703607, "loss": 1.8399648666381836, "step": 14567 }, { "epoch": 1.6951361414940656, "grad_norm": 1.1206122636795044, "learning_rate": 0.00043362521646523233, "loss": 1.7323620319366455, "step": 14568 }, { "epoch": 1.6952525017454039, "grad_norm": 0.7544728517532349, "learning_rate": 0.0004334700692866104, "loss": 1.2840405702590942, "step": 14569 }, { "epoch": 1.6953688619967417, "grad_norm": 1.3282467126846313, "learning_rate": 0.0004333147688207939, "loss": 1.832707166671753, "step": 14570 }, { "epoch": 1.69548522224808, "grad_norm": 1.2865849733352661, "learning_rate": 0.00043315931519753465, "loss": 1.6911108493804932, "step": 14571 }, { "epoch": 1.6956015824994182, "grad_norm": 1.109565019607544, "learning_rate": 0.00043300370854671246, "loss": 1.914806604385376, "step": 14572 }, { "epoch": 1.6957179427507563, "grad_norm": 1.557887077331543, "learning_rate": 0.0004328479489983349, "loss": 2.3229174613952637, "step": 14573 }, { "epoch": 1.6958343030020946, "grad_norm": 0.716713547706604, "learning_rate": 0.0004326920366825373, "loss": 2.0247347354888916, "step": 14574 }, { "epoch": 1.6959506632534325, "grad_norm": 0.6315485835075378, "learning_rate": 0.0004325359717295828, "loss": 1.2240338325500488, "step": 14575 }, { "epoch": 1.6960670235047708, "grad_norm": 1.0973149538040161, "learning_rate": 0.0004323797542698618, "loss": 1.9733225107192993, "step": 14576 }, { "epoch": 1.6961833837561089, "grad_norm": 0.6852884888648987, "learning_rate": 0.0004322233844338922, "loss": 1.3466084003448486, "step": 14577 }, { "epoch": 1.696299744007447, "grad_norm": 0.9414016008377075, "learning_rate": 0.0004320668623523195, "loss": 2.2561464309692383, "step": 14578 }, { "epoch": 1.6964161042587853, "grad_norm": 0.8195067048072815, "learning_rate": 0.0004319101881559159, "loss": 1.7620824575424194, "step": 14579 }, { "epoch": 1.6965324645101232, "grad_norm": 0.6653832197189331, "learning_rate": 0.000431753361975581, "loss": 1.1265226602554321, "step": 14580 }, { "epoch": 1.6966488247614615, "grad_norm": 1.1037620306015015, "learning_rate": 0.00043159638394234137, "loss": 2.124387741088867, "step": 14581 }, { "epoch": 1.6967651850127996, "grad_norm": 1.0836334228515625, "learning_rate": 0.00043143925418735035, "loss": 2.0313186645507812, "step": 14582 }, { "epoch": 1.6968815452641377, "grad_norm": 0.6734718084335327, "learning_rate": 0.0004312819728418882, "loss": 1.7278470993041992, "step": 14583 }, { "epoch": 1.696997905515476, "grad_norm": 1.3552578687667847, "learning_rate": 0.0004311245400373616, "loss": 1.809646725654602, "step": 14584 }, { "epoch": 1.697114265766814, "grad_norm": 1.3110765218734741, "learning_rate": 0.000430966955905304, "loss": 2.5262880325317383, "step": 14585 }, { "epoch": 1.6972306260181522, "grad_norm": 0.8500747084617615, "learning_rate": 0.00043080922057737515, "loss": 1.174079179763794, "step": 14586 }, { "epoch": 1.6973469862694903, "grad_norm": 0.839529275894165, "learning_rate": 0.00043065133418536107, "loss": 2.0890965461730957, "step": 14587 }, { "epoch": 1.6974633465208284, "grad_norm": 0.9229481816291809, "learning_rate": 0.0004304932968611742, "loss": 1.9294108152389526, "step": 14588 }, { "epoch": 1.6975797067721667, "grad_norm": 1.2121057510375977, "learning_rate": 0.00043033510873685297, "loss": 1.7308781147003174, "step": 14589 }, { "epoch": 1.6976960670235046, "grad_norm": 0.7531486749649048, "learning_rate": 0.00043017676994456166, "loss": 1.4350647926330566, "step": 14590 }, { "epoch": 1.697812427274843, "grad_norm": 1.8937543630599976, "learning_rate": 0.0004300182806165907, "loss": 1.7673487663269043, "step": 14591 }, { "epoch": 1.697928787526181, "grad_norm": 1.1828657388687134, "learning_rate": 0.00042985964088535606, "loss": 2.354156017303467, "step": 14592 }, { "epoch": 1.6980451477775191, "grad_norm": 1.4349292516708374, "learning_rate": 0.0004297008508833995, "loss": 1.8952229022979736, "step": 14593 }, { "epoch": 1.6981615080288575, "grad_norm": 1.3314539194107056, "learning_rate": 0.00042954191074338834, "loss": 1.8778071403503418, "step": 14594 }, { "epoch": 1.6982778682801953, "grad_norm": 0.9788405299186707, "learning_rate": 0.0004293828205981152, "loss": 2.2159922122955322, "step": 14595 }, { "epoch": 1.6983942285315337, "grad_norm": 1.0975450277328491, "learning_rate": 0.00042922358058049807, "loss": 1.5484154224395752, "step": 14596 }, { "epoch": 1.6985105887828718, "grad_norm": 1.0851130485534668, "learning_rate": 0.00042906419082358046, "loss": 1.8386330604553223, "step": 14597 }, { "epoch": 1.6986269490342099, "grad_norm": 0.6128645539283752, "learning_rate": 0.00042890465146053046, "loss": 1.3859593868255615, "step": 14598 }, { "epoch": 1.6987433092855482, "grad_norm": 1.335679292678833, "learning_rate": 0.0004287449626246416, "loss": 1.9265902042388916, "step": 14599 }, { "epoch": 1.698859669536886, "grad_norm": 1.1502430438995361, "learning_rate": 0.00042858512444933207, "loss": 2.448038101196289, "step": 14600 }, { "epoch": 1.6989760297882244, "grad_norm": 0.7704970836639404, "learning_rate": 0.00042842513706814483, "loss": 1.9495893716812134, "step": 14601 }, { "epoch": 1.6990923900395625, "grad_norm": 2.2828638553619385, "learning_rate": 0.0004282650006147476, "loss": 2.3093647956848145, "step": 14602 }, { "epoch": 1.6992087502909006, "grad_norm": 1.2932008504867554, "learning_rate": 0.00042810471522293264, "loss": 2.000164270401001, "step": 14603 }, { "epoch": 1.699325110542239, "grad_norm": 0.9433635473251343, "learning_rate": 0.0004279442810266165, "loss": 1.569784164428711, "step": 14604 }, { "epoch": 1.6994414707935768, "grad_norm": 1.9569118022918701, "learning_rate": 0.0004277836981598403, "loss": 2.6276540756225586, "step": 14605 }, { "epoch": 1.699557831044915, "grad_norm": 0.8734184503555298, "learning_rate": 0.00042762296675676905, "loss": 1.5272343158721924, "step": 14606 }, { "epoch": 1.6996741912962532, "grad_norm": 1.1520527601242065, "learning_rate": 0.0004274620869516922, "loss": 1.7545180320739746, "step": 14607 }, { "epoch": 1.6997905515475913, "grad_norm": 1.1101638078689575, "learning_rate": 0.0004273010588790229, "loss": 2.176844358444214, "step": 14608 }, { "epoch": 1.6999069117989296, "grad_norm": 1.3922712802886963, "learning_rate": 0.0004271398826732985, "loss": 1.6611213684082031, "step": 14609 }, { "epoch": 1.7000232720502675, "grad_norm": 0.9761262536048889, "learning_rate": 0.0004269785584691797, "loss": 1.7132017612457275, "step": 14610 }, { "epoch": 1.7001396323016058, "grad_norm": 1.1928280591964722, "learning_rate": 0.0004268170864014512, "loss": 1.5999469757080078, "step": 14611 }, { "epoch": 1.700255992552944, "grad_norm": 1.0002204179763794, "learning_rate": 0.0004266554666050211, "loss": 1.2524789571762085, "step": 14612 }, { "epoch": 1.700372352804282, "grad_norm": 1.376695156097412, "learning_rate": 0.0004264936992149209, "loss": 2.9503846168518066, "step": 14613 }, { "epoch": 1.7004887130556203, "grad_norm": 0.8006049990653992, "learning_rate": 0.0004263317843663055, "loss": 1.7612446546554565, "step": 14614 }, { "epoch": 1.7006050733069582, "grad_norm": 0.7518108487129211, "learning_rate": 0.00042616972219445285, "loss": 1.4516184329986572, "step": 14615 }, { "epoch": 1.7007214335582965, "grad_norm": 0.9924318790435791, "learning_rate": 0.00042600751283476413, "loss": 2.1118710041046143, "step": 14616 }, { "epoch": 1.7008377938096346, "grad_norm": 1.2881461381912231, "learning_rate": 0.0004258451564227636, "loss": 1.7692677974700928, "step": 14617 }, { "epoch": 1.7009541540609727, "grad_norm": 0.8310421109199524, "learning_rate": 0.00042568265309409793, "loss": 1.6592791080474854, "step": 14618 }, { "epoch": 1.701070514312311, "grad_norm": 1.7135266065597534, "learning_rate": 0.00042552000298453713, "loss": 2.1014022827148438, "step": 14619 }, { "epoch": 1.701186874563649, "grad_norm": 1.2397569417953491, "learning_rate": 0.0004253572062299734, "loss": 2.7482810020446777, "step": 14620 }, { "epoch": 1.7013032348149872, "grad_norm": 1.1769832372665405, "learning_rate": 0.0004251942629664216, "loss": 1.5594537258148193, "step": 14621 }, { "epoch": 1.7014195950663253, "grad_norm": 1.9994620084762573, "learning_rate": 0.00042503117333001904, "loss": 2.3039326667785645, "step": 14622 }, { "epoch": 1.7015359553176634, "grad_norm": 0.8801234364509583, "learning_rate": 0.00042486793745702536, "loss": 1.7032285928726196, "step": 14623 }, { "epoch": 1.7016523155690018, "grad_norm": 0.9956095814704895, "learning_rate": 0.0004247045554838222, "loss": 2.214597702026367, "step": 14624 }, { "epoch": 1.7017686758203396, "grad_norm": 1.2286370992660522, "learning_rate": 0.0004245410275469134, "loss": 2.210789442062378, "step": 14625 }, { "epoch": 1.701885036071678, "grad_norm": 0.9692370891571045, "learning_rate": 0.00042437735378292476, "loss": 1.7643091678619385, "step": 14626 }, { "epoch": 1.702001396323016, "grad_norm": 1.2294014692306519, "learning_rate": 0.0004242135343286039, "loss": 1.8353970050811768, "step": 14627 }, { "epoch": 1.7021177565743542, "grad_norm": 1.6205884218215942, "learning_rate": 0.00042404956932082005, "loss": 1.7329169511795044, "step": 14628 }, { "epoch": 1.7022341168256925, "grad_norm": 0.6351675987243652, "learning_rate": 0.0004238854588965642, "loss": 1.3604731559753418, "step": 14629 }, { "epoch": 1.7023504770770304, "grad_norm": 1.0511800050735474, "learning_rate": 0.0004237212031929488, "loss": 2.3340487480163574, "step": 14630 }, { "epoch": 1.7024668373283687, "grad_norm": 1.961525321006775, "learning_rate": 0.0004235568023472076, "loss": 1.7784087657928467, "step": 14631 }, { "epoch": 1.7025831975797068, "grad_norm": 1.3321471214294434, "learning_rate": 0.00042339225649669576, "loss": 2.283883810043335, "step": 14632 }, { "epoch": 1.7026995578310449, "grad_norm": 1.0542742013931274, "learning_rate": 0.0004232275657788895, "loss": 2.095870018005371, "step": 14633 }, { "epoch": 1.7028159180823832, "grad_norm": 0.5812451243400574, "learning_rate": 0.000423062730331386, "loss": 1.0516620874404907, "step": 14634 }, { "epoch": 1.702932278333721, "grad_norm": 1.1893703937530518, "learning_rate": 0.00042289775029190335, "loss": 1.5762454271316528, "step": 14635 }, { "epoch": 1.7030486385850594, "grad_norm": 1.1317404508590698, "learning_rate": 0.0004227326257982808, "loss": 2.2685012817382812, "step": 14636 }, { "epoch": 1.7031649988363975, "grad_norm": 1.2318079471588135, "learning_rate": 0.00042256735698847784, "loss": 1.4105443954467773, "step": 14637 }, { "epoch": 1.7032813590877356, "grad_norm": 0.7560974359512329, "learning_rate": 0.00042240194400057463, "loss": 1.6953961849212646, "step": 14638 }, { "epoch": 1.703397719339074, "grad_norm": 0.7451494932174683, "learning_rate": 0.00042223638697277206, "loss": 1.4088748693466187, "step": 14639 }, { "epoch": 1.7035140795904118, "grad_norm": 1.2538704872131348, "learning_rate": 0.00042207068604339107, "loss": 1.6913492679595947, "step": 14640 }, { "epoch": 1.70363043984175, "grad_norm": 1.0929869413375854, "learning_rate": 0.000421904841350873, "loss": 1.8423213958740234, "step": 14641 }, { "epoch": 1.7037468000930882, "grad_norm": 1.3229414224624634, "learning_rate": 0.0004217388530337792, "loss": 2.2200284004211426, "step": 14642 }, { "epoch": 1.7038631603444263, "grad_norm": 1.1594977378845215, "learning_rate": 0.00042157272123079107, "loss": 1.5634517669677734, "step": 14643 }, { "epoch": 1.7039795205957646, "grad_norm": 1.6058721542358398, "learning_rate": 0.0004214064460807099, "loss": 2.1293983459472656, "step": 14644 }, { "epoch": 1.7040958808471025, "grad_norm": 0.8665992617607117, "learning_rate": 0.0004212400277224566, "loss": 1.9950449466705322, "step": 14645 }, { "epoch": 1.7042122410984408, "grad_norm": 1.2168861627578735, "learning_rate": 0.000421073466295072, "loss": 1.9961315393447876, "step": 14646 }, { "epoch": 1.704328601349779, "grad_norm": 1.504387378692627, "learning_rate": 0.0004209067619377163, "loss": 2.085339069366455, "step": 14647 }, { "epoch": 1.704444961601117, "grad_norm": 1.083018183708191, "learning_rate": 0.000420739914789669, "loss": 1.9068946838378906, "step": 14648 }, { "epoch": 1.7045613218524553, "grad_norm": 1.0337069034576416, "learning_rate": 0.00042057292499032915, "loss": 2.162019729614258, "step": 14649 }, { "epoch": 1.7046776821037932, "grad_norm": 1.4843413829803467, "learning_rate": 0.00042040579267921485, "loss": 1.7139183282852173, "step": 14650 }, { "epoch": 1.7047940423551315, "grad_norm": 1.278223991394043, "learning_rate": 0.00042023851799596326, "loss": 1.9603185653686523, "step": 14651 }, { "epoch": 1.7049104026064696, "grad_norm": 0.7414475679397583, "learning_rate": 0.0004200711010803304, "loss": 1.8468818664550781, "step": 14652 }, { "epoch": 1.7050267628578077, "grad_norm": 1.300520420074463, "learning_rate": 0.0004199035420721915, "loss": 2.007312536239624, "step": 14653 }, { "epoch": 1.705143123109146, "grad_norm": 1.0190026760101318, "learning_rate": 0.00041973584111154, "loss": 2.212982654571533, "step": 14654 }, { "epoch": 1.705259483360484, "grad_norm": 1.2296046018600464, "learning_rate": 0.00041956799833848824, "loss": 1.9401006698608398, "step": 14655 }, { "epoch": 1.7053758436118223, "grad_norm": 0.8230233192443848, "learning_rate": 0.000419400013893267, "loss": 1.9429200887680054, "step": 14656 }, { "epoch": 1.7054922038631604, "grad_norm": 0.6199017763137817, "learning_rate": 0.00041923188791622546, "loss": 1.5547974109649658, "step": 14657 }, { "epoch": 1.7056085641144985, "grad_norm": 1.581382393836975, "learning_rate": 0.00041906362054783096, "loss": 2.568617105484009, "step": 14658 }, { "epoch": 1.7057249243658368, "grad_norm": 1.2584140300750732, "learning_rate": 0.00041889521192866897, "loss": 1.805980920791626, "step": 14659 }, { "epoch": 1.7058412846171747, "grad_norm": 0.36531534790992737, "learning_rate": 0.0004187266621994431, "loss": 0.9085031747817993, "step": 14660 }, { "epoch": 1.705957644868513, "grad_norm": 1.2681841850280762, "learning_rate": 0.0004185579715009746, "loss": 1.7858806848526, "step": 14661 }, { "epoch": 1.706074005119851, "grad_norm": 0.8221336603164673, "learning_rate": 0.0004183891399742029, "loss": 1.9047982692718506, "step": 14662 }, { "epoch": 1.7061903653711892, "grad_norm": 0.8197426199913025, "learning_rate": 0.00041822016776018456, "loss": 1.1895654201507568, "step": 14663 }, { "epoch": 1.7063067256225275, "grad_norm": 0.724694550037384, "learning_rate": 0.0004180510550000943, "loss": 1.1562633514404297, "step": 14664 }, { "epoch": 1.7064230858738654, "grad_norm": 0.891741156578064, "learning_rate": 0.00041788180183522375, "loss": 2.0941450595855713, "step": 14665 }, { "epoch": 1.7065394461252037, "grad_norm": 1.0413131713867188, "learning_rate": 0.000417712408406982, "loss": 1.4927369356155396, "step": 14666 }, { "epoch": 1.7066558063765418, "grad_norm": 1.386926531791687, "learning_rate": 0.0004175428748568955, "loss": 2.2632482051849365, "step": 14667 }, { "epoch": 1.70677216662788, "grad_norm": 1.0120553970336914, "learning_rate": 0.0004173732013266076, "loss": 1.7952699661254883, "step": 14668 }, { "epoch": 1.7068885268792182, "grad_norm": 3.475724458694458, "learning_rate": 0.0004172033879578786, "loss": 1.6850497722625732, "step": 14669 }, { "epoch": 1.707004887130556, "grad_norm": 0.989127516746521, "learning_rate": 0.00041703343489258573, "loss": 1.4070658683776855, "step": 14670 }, { "epoch": 1.7071212473818944, "grad_norm": 9.321874618530273, "learning_rate": 0.0004168633422727228, "loss": 1.408250093460083, "step": 14671 }, { "epoch": 1.7072376076332325, "grad_norm": 0.8961087465286255, "learning_rate": 0.00041669311024040046, "loss": 1.1956285238265991, "step": 14672 }, { "epoch": 1.7073539678845706, "grad_norm": 0.7685534954071045, "learning_rate": 0.00041652273893784544, "loss": 1.9482593536376953, "step": 14673 }, { "epoch": 1.707470328135909, "grad_norm": 1.0589853525161743, "learning_rate": 0.0004163522285074012, "loss": 1.7727559804916382, "step": 14674 }, { "epoch": 1.7075866883872468, "grad_norm": 1.2788804769515991, "learning_rate": 0.0004161815790915273, "loss": 1.5893815755844116, "step": 14675 }, { "epoch": 1.7077030486385851, "grad_norm": 1.44668447971344, "learning_rate": 0.00041601079083279934, "loss": 2.902863025665283, "step": 14676 }, { "epoch": 1.7078194088899232, "grad_norm": 0.8696115016937256, "learning_rate": 0.00041583986387390913, "loss": 1.7075397968292236, "step": 14677 }, { "epoch": 1.7079357691412613, "grad_norm": 1.436594009399414, "learning_rate": 0.0004156687983576641, "loss": 1.8088082075119019, "step": 14678 }, { "epoch": 1.7080521293925996, "grad_norm": 1.1929309368133545, "learning_rate": 0.0004154975944269877, "loss": 2.2973313331604004, "step": 14679 }, { "epoch": 1.7081684896439375, "grad_norm": 1.4137611389160156, "learning_rate": 0.00041532625222491884, "loss": 2.3847708702087402, "step": 14680 }, { "epoch": 1.7082848498952758, "grad_norm": 1.3220927715301514, "learning_rate": 0.000415154771894612, "loss": 1.7591564655303955, "step": 14681 }, { "epoch": 1.708401210146614, "grad_norm": 1.064487338066101, "learning_rate": 0.00041498315357933714, "loss": 1.447339415550232, "step": 14682 }, { "epoch": 1.708517570397952, "grad_norm": 1.6682132482528687, "learning_rate": 0.00041481139742247943, "loss": 1.2489749193191528, "step": 14683 }, { "epoch": 1.7086339306492904, "grad_norm": 1.3068302869796753, "learning_rate": 0.0004146395035675392, "loss": 2.4671630859375, "step": 14684 }, { "epoch": 1.7087502909006282, "grad_norm": 1.435190200805664, "learning_rate": 0.00041446747215813183, "loss": 2.2522315979003906, "step": 14685 }, { "epoch": 1.7088666511519666, "grad_norm": 1.8726741075515747, "learning_rate": 0.0004142953033379877, "loss": 1.8356590270996094, "step": 14686 }, { "epoch": 1.7089830114033047, "grad_norm": 0.9298648238182068, "learning_rate": 0.0004141229972509518, "loss": 2.3958373069763184, "step": 14687 }, { "epoch": 1.7090993716546428, "grad_norm": 1.612711787223816, "learning_rate": 0.00041395055404098416, "loss": 1.8798611164093018, "step": 14688 }, { "epoch": 1.7092157319059809, "grad_norm": 0.7287834882736206, "learning_rate": 0.000413777973852159, "loss": 1.3831707239151, "step": 14689 }, { "epoch": 1.709332092157319, "grad_norm": 2.7808711528778076, "learning_rate": 0.00041360525682866515, "loss": 2.134728193283081, "step": 14690 }, { "epoch": 1.7094484524086573, "grad_norm": 1.8304400444030762, "learning_rate": 0.0004134324031148058, "loss": 2.4415934085845947, "step": 14691 }, { "epoch": 1.7095648126599954, "grad_norm": 1.1438027620315552, "learning_rate": 0.00041325941285499837, "loss": 2.2233810424804688, "step": 14692 }, { "epoch": 1.7096811729113335, "grad_norm": 0.8900474309921265, "learning_rate": 0.0004130862861937741, "loss": 1.322532296180725, "step": 14693 }, { "epoch": 1.7097975331626716, "grad_norm": 1.1505210399627686, "learning_rate": 0.0004129130232757785, "loss": 2.1700780391693115, "step": 14694 }, { "epoch": 1.7099138934140097, "grad_norm": 0.5683495998382568, "learning_rate": 0.0004127396242457708, "loss": 1.013267993927002, "step": 14695 }, { "epoch": 1.710030253665348, "grad_norm": 1.2779948711395264, "learning_rate": 0.000412566089248624, "loss": 1.7213146686553955, "step": 14696 }, { "epoch": 1.710146613916686, "grad_norm": 1.0495837926864624, "learning_rate": 0.00041239241842932453, "loss": 1.5895442962646484, "step": 14697 }, { "epoch": 1.7102629741680242, "grad_norm": 1.2296046018600464, "learning_rate": 0.0004122186119329725, "loss": 2.0605602264404297, "step": 14698 }, { "epoch": 1.7103793344193623, "grad_norm": 3.0869300365448, "learning_rate": 0.0004120446699047813, "loss": 2.074355125427246, "step": 14699 }, { "epoch": 1.7104956946707004, "grad_norm": 0.9679393172264099, "learning_rate": 0.0004118705924900775, "loss": 2.0191385746002197, "step": 14700 }, { "epoch": 1.7106120549220387, "grad_norm": 1.2082936763763428, "learning_rate": 0.00041169637983430097, "loss": 1.7158368825912476, "step": 14701 }, { "epoch": 1.7107284151733768, "grad_norm": 0.8146830797195435, "learning_rate": 0.00041152203208300427, "loss": 2.060608386993408, "step": 14702 }, { "epoch": 1.710844775424715, "grad_norm": 1.1039780378341675, "learning_rate": 0.00041134754938185306, "loss": 2.375215530395508, "step": 14703 }, { "epoch": 1.710961135676053, "grad_norm": 1.0346068143844604, "learning_rate": 0.00041117293187662574, "loss": 1.5079396963119507, "step": 14704 }, { "epoch": 1.7110774959273911, "grad_norm": 0.9559508562088013, "learning_rate": 0.0004109981797132132, "loss": 1.6970546245574951, "step": 14705 }, { "epoch": 1.7111938561787294, "grad_norm": 0.7662757635116577, "learning_rate": 0.000410823293037619, "loss": 1.5540146827697754, "step": 14706 }, { "epoch": 1.7113102164300675, "grad_norm": 0.9776404500007629, "learning_rate": 0.000410648271995959, "loss": 2.091559410095215, "step": 14707 }, { "epoch": 1.7114265766814056, "grad_norm": 1.6802173852920532, "learning_rate": 0.0004104731167344613, "loss": 1.7303214073181152, "step": 14708 }, { "epoch": 1.7115429369327437, "grad_norm": 0.6966922879219055, "learning_rate": 0.0004102978273994661, "loss": 1.1616817712783813, "step": 14709 }, { "epoch": 1.7116592971840818, "grad_norm": 1.1015963554382324, "learning_rate": 0.00041012240413742585, "loss": 1.834468126296997, "step": 14710 }, { "epoch": 1.7117756574354202, "grad_norm": 1.1599376201629639, "learning_rate": 0.00040994684709490454, "loss": 2.0422446727752686, "step": 14711 }, { "epoch": 1.7118920176867582, "grad_norm": 3.420616388320923, "learning_rate": 0.00040977115641857825, "loss": 2.008352756500244, "step": 14712 }, { "epoch": 1.7120083779380963, "grad_norm": 1.2850017547607422, "learning_rate": 0.0004095953322552345, "loss": 1.7987573146820068, "step": 14713 }, { "epoch": 1.7121247381894344, "grad_norm": 1.3634120225906372, "learning_rate": 0.0004094193747517725, "loss": 2.137930393218994, "step": 14714 }, { "epoch": 1.7122410984407725, "grad_norm": 1.447623610496521, "learning_rate": 0.00040924328405520264, "loss": 2.189568042755127, "step": 14715 }, { "epoch": 1.7123574586921109, "grad_norm": 1.5146819353103638, "learning_rate": 0.00040906706031264685, "loss": 2.0761289596557617, "step": 14716 }, { "epoch": 1.712473818943449, "grad_norm": 1.5760585069656372, "learning_rate": 0.00040889070367133813, "loss": 1.9696094989776611, "step": 14717 }, { "epoch": 1.712590179194787, "grad_norm": 1.4752227067947388, "learning_rate": 0.00040871421427862036, "loss": 2.3901844024658203, "step": 14718 }, { "epoch": 1.7127065394461252, "grad_norm": 0.9381336569786072, "learning_rate": 0.0004085375922819485, "loss": 2.448617458343506, "step": 14719 }, { "epoch": 1.7128228996974633, "grad_norm": 1.0760419368743896, "learning_rate": 0.00040836083782888843, "loss": 1.881824254989624, "step": 14720 }, { "epoch": 1.7129392599488016, "grad_norm": 1.1539405584335327, "learning_rate": 0.00040818395106711623, "loss": 1.8901294469833374, "step": 14721 }, { "epoch": 1.7130556202001397, "grad_norm": 1.405319333076477, "learning_rate": 0.000408006932144419, "loss": 2.8659720420837402, "step": 14722 }, { "epoch": 1.7131719804514778, "grad_norm": 1.1275275945663452, "learning_rate": 0.00040782978120869394, "loss": 1.8242242336273193, "step": 14723 }, { "epoch": 1.7132883407028159, "grad_norm": 1.4146982431411743, "learning_rate": 0.0004076524984079489, "loss": 1.7643520832061768, "step": 14724 }, { "epoch": 1.713404700954154, "grad_norm": 1.1079590320587158, "learning_rate": 0.00040747508389030147, "loss": 2.1570229530334473, "step": 14725 }, { "epoch": 1.7135210612054923, "grad_norm": 1.1739176511764526, "learning_rate": 0.0004072975378039796, "loss": 1.874774694442749, "step": 14726 }, { "epoch": 1.7136374214568304, "grad_norm": 1.2366993427276611, "learning_rate": 0.00040711986029732107, "loss": 2.0660412311553955, "step": 14727 }, { "epoch": 1.7137537817081685, "grad_norm": 0.9108996987342834, "learning_rate": 0.00040694205151877333, "loss": 1.1277910470962524, "step": 14728 }, { "epoch": 1.7138701419595066, "grad_norm": 2.146245241165161, "learning_rate": 0.00040676411161689373, "loss": 2.3219239711761475, "step": 14729 }, { "epoch": 1.7139865022108447, "grad_norm": 0.6271599531173706, "learning_rate": 0.00040658604074034915, "loss": 1.3899731636047363, "step": 14730 }, { "epoch": 1.714102862462183, "grad_norm": 1.3875759840011597, "learning_rate": 0.00040640783903791566, "loss": 2.3351197242736816, "step": 14731 }, { "epoch": 1.714219222713521, "grad_norm": 1.1909314393997192, "learning_rate": 0.000406229506658479, "loss": 1.3511052131652832, "step": 14732 }, { "epoch": 1.7143355829648592, "grad_norm": 0.9778677821159363, "learning_rate": 0.00040605104375103366, "loss": 2.121293544769287, "step": 14733 }, { "epoch": 1.7144519432161973, "grad_norm": 0.8023042678833008, "learning_rate": 0.0004058724504646834, "loss": 1.3652358055114746, "step": 14734 }, { "epoch": 1.7145683034675354, "grad_norm": 1.1958948373794556, "learning_rate": 0.0004056937269486411, "loss": 2.2182106971740723, "step": 14735 }, { "epoch": 1.7146846637188737, "grad_norm": 0.9531533718109131, "learning_rate": 0.00040551487335222815, "loss": 1.2367280721664429, "step": 14736 }, { "epoch": 1.7148010239702116, "grad_norm": 1.1435346603393555, "learning_rate": 0.0004053358898248748, "loss": 1.9332882165908813, "step": 14737 }, { "epoch": 1.71491738422155, "grad_norm": 1.4328418970108032, "learning_rate": 0.00040515677651611963, "loss": 2.779569149017334, "step": 14738 }, { "epoch": 1.715033744472888, "grad_norm": 0.8363649845123291, "learning_rate": 0.00040497753357561007, "loss": 1.5589593648910522, "step": 14739 }, { "epoch": 1.7151501047242261, "grad_norm": 0.7137078046798706, "learning_rate": 0.0004047981611531014, "loss": 1.457026481628418, "step": 14740 }, { "epoch": 1.7152664649755645, "grad_norm": 0.9671376943588257, "learning_rate": 0.0004046186593984572, "loss": 2.540475845336914, "step": 14741 }, { "epoch": 1.7153828252269023, "grad_norm": 1.2497438192367554, "learning_rate": 0.00040443902846164947, "loss": 1.6522618532180786, "step": 14742 }, { "epoch": 1.7154991854782407, "grad_norm": 0.8553128242492676, "learning_rate": 0.0004042592684927576, "loss": 1.2144789695739746, "step": 14743 }, { "epoch": 1.7156155457295788, "grad_norm": 0.5736337304115295, "learning_rate": 0.00040407937964196923, "loss": 1.2857916355133057, "step": 14744 }, { "epoch": 1.7157319059809168, "grad_norm": 1.2320704460144043, "learning_rate": 0.00040389936205957936, "loss": 1.3981602191925049, "step": 14745 }, { "epoch": 1.7158482662322552, "grad_norm": 0.6261950731277466, "learning_rate": 0.00040371921589599057, "loss": 1.176263689994812, "step": 14746 }, { "epoch": 1.715964626483593, "grad_norm": 0.9395867586135864, "learning_rate": 0.0004035389413017132, "loss": 1.8759605884552002, "step": 14747 }, { "epoch": 1.7160809867349314, "grad_norm": 1.5992053747177124, "learning_rate": 0.0004033585384273646, "loss": 1.859655737876892, "step": 14748 }, { "epoch": 1.7161973469862695, "grad_norm": 0.7685105800628662, "learning_rate": 0.0004031780074236692, "loss": 2.0065271854400635, "step": 14749 }, { "epoch": 1.7163137072376076, "grad_norm": 1.0425665378570557, "learning_rate": 0.00040299734844145873, "loss": 1.0996158123016357, "step": 14750 }, { "epoch": 1.7164300674889459, "grad_norm": 1.230085849761963, "learning_rate": 0.0004028165616316719, "loss": 2.1255359649658203, "step": 14751 }, { "epoch": 1.7165464277402838, "grad_norm": 1.2661776542663574, "learning_rate": 0.00040263564714535384, "loss": 2.0254547595977783, "step": 14752 }, { "epoch": 1.716662787991622, "grad_norm": 1.3631646633148193, "learning_rate": 0.00040245460513365673, "loss": 2.083477020263672, "step": 14753 }, { "epoch": 1.7167791482429602, "grad_norm": 1.2024345397949219, "learning_rate": 0.0004022734357478391, "loss": 2.1980292797088623, "step": 14754 }, { "epoch": 1.7168955084942983, "grad_norm": 0.43529844284057617, "learning_rate": 0.00040209213913926597, "loss": 1.0242942571640015, "step": 14755 }, { "epoch": 1.7170118687456366, "grad_norm": 1.5844449996948242, "learning_rate": 0.00040191071545940863, "loss": 2.1462135314941406, "step": 14756 }, { "epoch": 1.7171282289969745, "grad_norm": 1.2914867401123047, "learning_rate": 0.00040172916485984445, "loss": 1.998215675354004, "step": 14757 }, { "epoch": 1.7172445892483128, "grad_norm": 1.121661901473999, "learning_rate": 0.0004015474874922571, "loss": 1.960606575012207, "step": 14758 }, { "epoch": 1.717360949499651, "grad_norm": 1.5559828281402588, "learning_rate": 0.0004013656835084358, "loss": 1.4968477487564087, "step": 14759 }, { "epoch": 1.717477309750989, "grad_norm": 1.0642927885055542, "learning_rate": 0.0004011837530602759, "loss": 2.3330225944519043, "step": 14760 }, { "epoch": 1.7175936700023273, "grad_norm": 1.1577703952789307, "learning_rate": 0.0004010016962997782, "loss": 1.764404296875, "step": 14761 }, { "epoch": 1.7177100302536652, "grad_norm": 0.8717832565307617, "learning_rate": 0.00040081951337904914, "loss": 1.5303306579589844, "step": 14762 }, { "epoch": 1.7178263905050035, "grad_norm": 0.5671027302742004, "learning_rate": 0.0004006372044503004, "loss": 1.4388856887817383, "step": 14763 }, { "epoch": 1.7179427507563416, "grad_norm": 1.2191466093063354, "learning_rate": 0.00040045476966584913, "loss": 1.7006332874298096, "step": 14764 }, { "epoch": 1.7180591110076797, "grad_norm": 1.2199292182922363, "learning_rate": 0.0004002722091781176, "loss": 1.935092568397522, "step": 14765 }, { "epoch": 1.718175471259018, "grad_norm": 1.1444660425186157, "learning_rate": 0.00040008952313963297, "loss": 1.5022053718566895, "step": 14766 }, { "epoch": 1.718291831510356, "grad_norm": 1.0324839353561401, "learning_rate": 0.0003999067117030275, "loss": 1.850529670715332, "step": 14767 }, { "epoch": 1.7184081917616942, "grad_norm": 0.417407363653183, "learning_rate": 0.000399723775021038, "loss": 1.169252634048462, "step": 14768 }, { "epoch": 1.7185245520130323, "grad_norm": 0.663131058216095, "learning_rate": 0.0003995407132465061, "loss": 1.4804465770721436, "step": 14769 }, { "epoch": 1.7186409122643704, "grad_norm": 1.0894019603729248, "learning_rate": 0.00039935752653237773, "loss": 1.5195674896240234, "step": 14770 }, { "epoch": 1.7187572725157088, "grad_norm": 0.8843579292297363, "learning_rate": 0.00039917421503170346, "loss": 1.397656798362732, "step": 14771 }, { "epoch": 1.7188736327670466, "grad_norm": 1.4411780834197998, "learning_rate": 0.000398990778897638, "loss": 2.194256544113159, "step": 14772 }, { "epoch": 1.718989993018385, "grad_norm": 0.8747337460517883, "learning_rate": 0.00039880721828344014, "loss": 1.6547260284423828, "step": 14773 }, { "epoch": 1.719106353269723, "grad_norm": 0.6138558387756348, "learning_rate": 0.0003986235333424728, "loss": 1.7971525192260742, "step": 14774 }, { "epoch": 1.7192227135210612, "grad_norm": 1.0394666194915771, "learning_rate": 0.00039843972422820263, "loss": 2.0465030670166016, "step": 14775 }, { "epoch": 1.7193390737723995, "grad_norm": 0.7298222780227661, "learning_rate": 0.00039825579109420016, "loss": 1.7197856903076172, "step": 14776 }, { "epoch": 1.7194554340237374, "grad_norm": 1.22183096408844, "learning_rate": 0.00039807173409413943, "loss": 2.186349868774414, "step": 14777 }, { "epoch": 1.7195717942750757, "grad_norm": 1.0158202648162842, "learning_rate": 0.00039788755338179794, "loss": 1.5982732772827148, "step": 14778 }, { "epoch": 1.7196881545264138, "grad_norm": 0.6925305724143982, "learning_rate": 0.0003977032491110567, "loss": 1.34335458278656, "step": 14779 }, { "epoch": 1.7198045147777519, "grad_norm": 1.096200704574585, "learning_rate": 0.00039751882143589996, "loss": 2.445510149002075, "step": 14780 }, { "epoch": 1.7199208750290902, "grad_norm": 1.0568273067474365, "learning_rate": 0.0003973342705104148, "loss": 2.0641703605651855, "step": 14781 }, { "epoch": 1.720037235280428, "grad_norm": 0.9444316625595093, "learning_rate": 0.0003971495964887916, "loss": 1.3964717388153076, "step": 14782 }, { "epoch": 1.7201535955317664, "grad_norm": 0.8415089845657349, "learning_rate": 0.00039696479952532326, "loss": 1.946249008178711, "step": 14783 }, { "epoch": 1.7202699557831045, "grad_norm": 1.1389397382736206, "learning_rate": 0.00039677987977440577, "loss": 1.6126431226730347, "step": 14784 }, { "epoch": 1.7203863160344426, "grad_norm": 1.7096153497695923, "learning_rate": 0.0003965948373905374, "loss": 1.9422953128814697, "step": 14785 }, { "epoch": 1.720502676285781, "grad_norm": 1.2186249494552612, "learning_rate": 0.00039640967252831913, "loss": 2.3064324855804443, "step": 14786 }, { "epoch": 1.7206190365371188, "grad_norm": 1.215849757194519, "learning_rate": 0.000396224385342454, "loss": 1.4717390537261963, "step": 14787 }, { "epoch": 1.720735396788457, "grad_norm": 1.1850402355194092, "learning_rate": 0.0003960389759877474, "loss": 1.414838433265686, "step": 14788 }, { "epoch": 1.7208517570397952, "grad_norm": 0.9520288705825806, "learning_rate": 0.0003958534446191069, "loss": 1.8940846920013428, "step": 14789 }, { "epoch": 1.7209681172911333, "grad_norm": 0.7384576201438904, "learning_rate": 0.00039566779139154175, "loss": 1.3673334121704102, "step": 14790 }, { "epoch": 1.7210844775424716, "grad_norm": 0.7412254810333252, "learning_rate": 0.00039548201646016315, "loss": 1.2215710878372192, "step": 14791 }, { "epoch": 1.7212008377938095, "grad_norm": 1.4600979089736938, "learning_rate": 0.0003952961199801841, "loss": 2.0373072624206543, "step": 14792 }, { "epoch": 1.7213171980451478, "grad_norm": 1.0588244199752808, "learning_rate": 0.00039511010210691895, "loss": 1.7999433279037476, "step": 14793 }, { "epoch": 1.721433558296486, "grad_norm": 1.0413093566894531, "learning_rate": 0.0003949239629957836, "loss": 1.463959813117981, "step": 14794 }, { "epoch": 1.721549918547824, "grad_norm": 0.6836608648300171, "learning_rate": 0.000394737702802295, "loss": 1.458444356918335, "step": 14795 }, { "epoch": 1.7216662787991623, "grad_norm": 0.9118222594261169, "learning_rate": 0.00039455132168207186, "loss": 2.0904273986816406, "step": 14796 }, { "epoch": 1.7217826390505002, "grad_norm": 1.234291672706604, "learning_rate": 0.00039436481979083317, "loss": 1.6929750442504883, "step": 14797 }, { "epoch": 1.7218989993018385, "grad_norm": 1.7681775093078613, "learning_rate": 0.0003941781972843993, "loss": 2.5307414531707764, "step": 14798 }, { "epoch": 1.7220153595531766, "grad_norm": 1.6168806552886963, "learning_rate": 0.0003939914543186913, "loss": 2.380835771560669, "step": 14799 }, { "epoch": 1.7221317198045147, "grad_norm": 1.484894871711731, "learning_rate": 0.0003938045910497308, "loss": 2.256654977798462, "step": 14800 }, { "epoch": 1.722248080055853, "grad_norm": 1.6171120405197144, "learning_rate": 0.00039361760763363993, "loss": 1.7876288890838623, "step": 14801 }, { "epoch": 1.722364440307191, "grad_norm": 1.4419790506362915, "learning_rate": 0.00039343050422664135, "loss": 2.8293628692626953, "step": 14802 }, { "epoch": 1.7224808005585293, "grad_norm": 1.444319486618042, "learning_rate": 0.0003932432809850578, "loss": 1.605300784111023, "step": 14803 }, { "epoch": 1.7225971608098674, "grad_norm": 1.036634087562561, "learning_rate": 0.00039305593806531223, "loss": 2.1306686401367188, "step": 14804 }, { "epoch": 1.7227135210612055, "grad_norm": 0.87519770860672, "learning_rate": 0.0003928684756239276, "loss": 2.050607204437256, "step": 14805 }, { "epoch": 1.7228298813125438, "grad_norm": 1.21290922164917, "learning_rate": 0.0003926808938175265, "loss": 1.704451560974121, "step": 14806 }, { "epoch": 1.7229462415638817, "grad_norm": 1.1631643772125244, "learning_rate": 0.0003924931928028317, "loss": 2.143301248550415, "step": 14807 }, { "epoch": 1.72306260181522, "grad_norm": 1.4042975902557373, "learning_rate": 0.0003923053727366651, "loss": 2.4908254146575928, "step": 14808 }, { "epoch": 1.723178962066558, "grad_norm": 0.8688745498657227, "learning_rate": 0.00039211743377594836, "loss": 1.5937347412109375, "step": 14809 }, { "epoch": 1.7232953223178962, "grad_norm": 0.8740183115005493, "learning_rate": 0.0003919293760777024, "loss": 2.2571423053741455, "step": 14810 }, { "epoch": 1.7234116825692345, "grad_norm": 0.48201683163642883, "learning_rate": 0.00039174119979904734, "loss": 1.0064237117767334, "step": 14811 }, { "epoch": 1.7235280428205724, "grad_norm": 0.9933179020881653, "learning_rate": 0.0003915529050972022, "loss": 1.3097938299179077, "step": 14812 }, { "epoch": 1.7236444030719107, "grad_norm": 1.2752712965011597, "learning_rate": 0.0003913644921294852, "loss": 2.3136091232299805, "step": 14813 }, { "epoch": 1.7237607633232488, "grad_norm": 0.7444095611572266, "learning_rate": 0.0003911759610533134, "loss": 1.617703914642334, "step": 14814 }, { "epoch": 1.723877123574587, "grad_norm": 0.784424901008606, "learning_rate": 0.00039098731202620227, "loss": 1.4066519737243652, "step": 14815 }, { "epoch": 1.7239934838259252, "grad_norm": 1.1797854900360107, "learning_rate": 0.000390798545205766, "loss": 1.5664024353027344, "step": 14816 }, { "epoch": 1.724109844077263, "grad_norm": 1.4699395895004272, "learning_rate": 0.000390609660749717, "loss": 1.6932224035263062, "step": 14817 }, { "epoch": 1.7242262043286014, "grad_norm": 1.2595547437667847, "learning_rate": 0.0003904206588158665, "loss": 2.3027029037475586, "step": 14818 }, { "epoch": 1.7243425645799395, "grad_norm": 1.4007114171981812, "learning_rate": 0.00039023153956212317, "loss": 1.8602608442306519, "step": 14819 }, { "epoch": 1.7244589248312776, "grad_norm": 1.2311211824417114, "learning_rate": 0.00039004230314649413, "loss": 1.930877685546875, "step": 14820 }, { "epoch": 1.724575285082616, "grad_norm": 1.1472185850143433, "learning_rate": 0.0003898529497270843, "loss": 2.07395601272583, "step": 14821 }, { "epoch": 1.7246916453339538, "grad_norm": 1.769943118095398, "learning_rate": 0.0003896634794620964, "loss": 2.8451008796691895, "step": 14822 }, { "epoch": 1.7248080055852921, "grad_norm": 1.950082540512085, "learning_rate": 0.0003894738925098307, "loss": 1.4382638931274414, "step": 14823 }, { "epoch": 1.7249243658366302, "grad_norm": 1.3995933532714844, "learning_rate": 0.0003892841890286849, "loss": 2.015777111053467, "step": 14824 }, { "epoch": 1.7250407260879683, "grad_norm": 0.8309323191642761, "learning_rate": 0.0003890943691771542, "loss": 1.2553162574768066, "step": 14825 }, { "epoch": 1.7251570863393066, "grad_norm": 1.1952126026153564, "learning_rate": 0.00038890443311383107, "loss": 2.1071810722351074, "step": 14826 }, { "epoch": 1.7252734465906445, "grad_norm": 1.14058518409729, "learning_rate": 0.00038871438099740476, "loss": 2.464503049850464, "step": 14827 }, { "epoch": 1.7253898068419828, "grad_norm": 1.9663587808609009, "learning_rate": 0.00038852421298666184, "loss": 2.490957736968994, "step": 14828 }, { "epoch": 1.725506167093321, "grad_norm": 1.3363666534423828, "learning_rate": 0.0003883339292404855, "loss": 2.0915417671203613, "step": 14829 }, { "epoch": 1.725622527344659, "grad_norm": 1.3079496622085571, "learning_rate": 0.0003881435299178557, "loss": 2.4997398853302, "step": 14830 }, { "epoch": 1.7257388875959974, "grad_norm": 1.6308091878890991, "learning_rate": 0.00038795301517784896, "loss": 2.489116907119751, "step": 14831 }, { "epoch": 1.7258552478473352, "grad_norm": 1.9667325019836426, "learning_rate": 0.00038776238517963823, "loss": 2.215913772583008, "step": 14832 }, { "epoch": 1.7259716080986736, "grad_norm": 0.6935341358184814, "learning_rate": 0.0003875716400824928, "loss": 1.1154792308807373, "step": 14833 }, { "epoch": 1.7260879683500117, "grad_norm": 1.0696368217468262, "learning_rate": 0.00038738078004577786, "loss": 2.7010250091552734, "step": 14834 }, { "epoch": 1.7262043286013498, "grad_norm": 1.2206674814224243, "learning_rate": 0.0003871898052289551, "loss": 2.8360280990600586, "step": 14835 }, { "epoch": 1.726320688852688, "grad_norm": 0.8511086702346802, "learning_rate": 0.00038699871579158166, "loss": 2.0817160606384277, "step": 14836 }, { "epoch": 1.726437049104026, "grad_norm": 1.376902461051941, "learning_rate": 0.00038680751189331064, "loss": 2.3760814666748047, "step": 14837 }, { "epoch": 1.7265534093553643, "grad_norm": 1.036044716835022, "learning_rate": 0.0003866161936938909, "loss": 2.1055750846862793, "step": 14838 }, { "epoch": 1.7266697696067024, "grad_norm": 0.8008891940116882, "learning_rate": 0.0003864247613531665, "loss": 2.1735825538635254, "step": 14839 }, { "epoch": 1.7267861298580405, "grad_norm": 1.424067735671997, "learning_rate": 0.0003862332150310772, "loss": 1.5374741554260254, "step": 14840 }, { "epoch": 1.7269024901093788, "grad_norm": 1.123732328414917, "learning_rate": 0.00038604155488765763, "loss": 1.473750352859497, "step": 14841 }, { "epoch": 1.7270188503607167, "grad_norm": 1.0632070302963257, "learning_rate": 0.00038584978108303783, "loss": 1.8133565187454224, "step": 14842 }, { "epoch": 1.727135210612055, "grad_norm": 1.615440845489502, "learning_rate": 0.0003856578937774426, "loss": 2.4435501098632812, "step": 14843 }, { "epoch": 1.727251570863393, "grad_norm": 1.197638750076294, "learning_rate": 0.0003854658931311916, "loss": 1.789677381515503, "step": 14844 }, { "epoch": 1.7273679311147312, "grad_norm": 1.7985901832580566, "learning_rate": 0.00038527377930469944, "loss": 1.6005134582519531, "step": 14845 }, { "epoch": 1.7274842913660695, "grad_norm": 0.9457374215126038, "learning_rate": 0.00038508155245847486, "loss": 1.8276652097702026, "step": 14846 }, { "epoch": 1.7276006516174074, "grad_norm": 0.8596615195274353, "learning_rate": 0.0003848892127531213, "loss": 1.5232508182525635, "step": 14847 }, { "epoch": 1.7277170118687457, "grad_norm": 1.4502516984939575, "learning_rate": 0.00038469676034933646, "loss": 2.405829429626465, "step": 14848 }, { "epoch": 1.7278333721200838, "grad_norm": 0.8771569132804871, "learning_rate": 0.0003845041954079122, "loss": 1.946448564529419, "step": 14849 }, { "epoch": 1.727949732371422, "grad_norm": 0.8888003826141357, "learning_rate": 0.0003843115180897343, "loss": 2.464541435241699, "step": 14850 }, { "epoch": 1.7280660926227602, "grad_norm": 1.7256025075912476, "learning_rate": 0.00038411872855578257, "loss": 1.9033411741256714, "step": 14851 }, { "epoch": 1.728182452874098, "grad_norm": 1.0354429483413696, "learning_rate": 0.00038392582696713054, "loss": 1.9470759630203247, "step": 14852 }, { "epoch": 1.7282988131254364, "grad_norm": 1.7340261936187744, "learning_rate": 0.0003837328134849453, "loss": 1.8472168445587158, "step": 14853 }, { "epoch": 1.7284151733767745, "grad_norm": 0.9000260829925537, "learning_rate": 0.00038353968827048733, "loss": 1.5218758583068848, "step": 14854 }, { "epoch": 1.7285315336281126, "grad_norm": 1.0964916944503784, "learning_rate": 0.0003833464514851108, "loss": 1.7114479541778564, "step": 14855 }, { "epoch": 1.7286478938794507, "grad_norm": 0.9061227440834045, "learning_rate": 0.00038315310329026275, "loss": 1.2289992570877075, "step": 14856 }, { "epoch": 1.7287642541307888, "grad_norm": 1.0060383081436157, "learning_rate": 0.00038295964384748356, "loss": 2.239920139312744, "step": 14857 }, { "epoch": 1.7288806143821271, "grad_norm": 1.243792176246643, "learning_rate": 0.0003827660733184063, "loss": 2.16182804107666, "step": 14858 }, { "epoch": 1.7289969746334652, "grad_norm": 1.4210197925567627, "learning_rate": 0.00038257239186475703, "loss": 2.172487735748291, "step": 14859 }, { "epoch": 1.7291133348848033, "grad_norm": 0.45799973607063293, "learning_rate": 0.00038237859964835447, "loss": 1.0019981861114502, "step": 14860 }, { "epoch": 1.7292296951361414, "grad_norm": 1.1912193298339844, "learning_rate": 0.0003821846968311099, "loss": 2.246370315551758, "step": 14861 }, { "epoch": 1.7293460553874795, "grad_norm": 0.7770628333091736, "learning_rate": 0.00038199068357502684, "loss": 1.950797438621521, "step": 14862 }, { "epoch": 1.7294624156388179, "grad_norm": 1.0218074321746826, "learning_rate": 0.0003817965600422013, "loss": 1.4398717880249023, "step": 14863 }, { "epoch": 1.729578775890156, "grad_norm": 1.676079511642456, "learning_rate": 0.00038160232639482126, "loss": 2.3865623474121094, "step": 14864 }, { "epoch": 1.729695136141494, "grad_norm": 2.017479419708252, "learning_rate": 0.00038140798279516676, "loss": 1.853191614151001, "step": 14865 }, { "epoch": 1.7298114963928322, "grad_norm": 1.408397912979126, "learning_rate": 0.0003812135294056098, "loss": 1.6645195484161377, "step": 14866 }, { "epoch": 1.7299278566441703, "grad_norm": 1.2318756580352783, "learning_rate": 0.0003810189663886139, "loss": 2.0970821380615234, "step": 14867 }, { "epoch": 1.7300442168955086, "grad_norm": 1.3531174659729004, "learning_rate": 0.00038082429390673433, "loss": 1.5885801315307617, "step": 14868 }, { "epoch": 1.7301605771468467, "grad_norm": 0.5139291882514954, "learning_rate": 0.00038062951212261786, "loss": 1.2262393236160278, "step": 14869 }, { "epoch": 1.7302769373981848, "grad_norm": 2.133767604827881, "learning_rate": 0.0003804346211990023, "loss": 1.7791674137115479, "step": 14870 }, { "epoch": 1.7303932976495229, "grad_norm": 1.1638344526290894, "learning_rate": 0.00038023962129871705, "loss": 1.9502332210540771, "step": 14871 }, { "epoch": 1.730509657900861, "grad_norm": 0.994395911693573, "learning_rate": 0.0003800445125846823, "loss": 1.742160677909851, "step": 14872 }, { "epoch": 1.7306260181521993, "grad_norm": 1.5085973739624023, "learning_rate": 0.0003798492952199091, "loss": 2.2863759994506836, "step": 14873 }, { "epoch": 1.7307423784035374, "grad_norm": 0.7834385633468628, "learning_rate": 0.0003796539693674994, "loss": 1.5708515644073486, "step": 14874 }, { "epoch": 1.7308587386548755, "grad_norm": 1.7656481266021729, "learning_rate": 0.00037945853519064587, "loss": 2.5051026344299316, "step": 14875 }, { "epoch": 1.7309750989062136, "grad_norm": 1.292968511581421, "learning_rate": 0.00037926299285263155, "loss": 1.406266689300537, "step": 14876 }, { "epoch": 1.7310914591575517, "grad_norm": 0.7362184524536133, "learning_rate": 0.00037906734251682986, "loss": 1.6652815341949463, "step": 14877 }, { "epoch": 1.73120781940889, "grad_norm": 0.932433545589447, "learning_rate": 0.0003788715843467044, "loss": 1.3781198263168335, "step": 14878 }, { "epoch": 1.7313241796602281, "grad_norm": 1.0665217638015747, "learning_rate": 0.000378675718505809, "loss": 2.108586549758911, "step": 14879 }, { "epoch": 1.7314405399115662, "grad_norm": 1.1831129789352417, "learning_rate": 0.00037847974515778745, "loss": 2.377653121948242, "step": 14880 }, { "epoch": 1.7315569001629043, "grad_norm": 0.5740700960159302, "learning_rate": 0.0003782836644663732, "loss": 1.4467265605926514, "step": 14881 }, { "epoch": 1.7316732604142424, "grad_norm": 1.0185198783874512, "learning_rate": 0.0003780874765953895, "loss": 2.2836620807647705, "step": 14882 }, { "epoch": 1.7317896206655807, "grad_norm": 0.7105375528335571, "learning_rate": 0.0003778911817087491, "loss": 1.4615895748138428, "step": 14883 }, { "epoch": 1.7319059809169188, "grad_norm": 0.920491635799408, "learning_rate": 0.00037769477997045425, "loss": 2.1482443809509277, "step": 14884 }, { "epoch": 1.732022341168257, "grad_norm": 1.4649455547332764, "learning_rate": 0.00037749827154459636, "loss": 2.3720455169677734, "step": 14885 }, { "epoch": 1.732138701419595, "grad_norm": 0.9616376757621765, "learning_rate": 0.0003773016565953561, "loss": 1.3239526748657227, "step": 14886 }, { "epoch": 1.7322550616709331, "grad_norm": 0.9854036569595337, "learning_rate": 0.00037710493528700297, "loss": 1.4621620178222656, "step": 14887 }, { "epoch": 1.7323714219222714, "grad_norm": 0.6180519461631775, "learning_rate": 0.0003769081077838955, "loss": 1.4626436233520508, "step": 14888 }, { "epoch": 1.7324877821736095, "grad_norm": 1.3609471321105957, "learning_rate": 0.00037671117425048086, "loss": 2.4734559059143066, "step": 14889 }, { "epoch": 1.7326041424249476, "grad_norm": 1.2258999347686768, "learning_rate": 0.00037651413485129473, "loss": 2.6341514587402344, "step": 14890 }, { "epoch": 1.7327205026762857, "grad_norm": 1.099613070487976, "learning_rate": 0.0003763169897509614, "loss": 1.784637451171875, "step": 14891 }, { "epoch": 1.7328368629276238, "grad_norm": 1.4504621028900146, "learning_rate": 0.00037611973911419347, "loss": 2.4577972888946533, "step": 14892 }, { "epoch": 1.7329532231789622, "grad_norm": 0.8478720784187317, "learning_rate": 0.0003759223831057915, "loss": 1.5807344913482666, "step": 14893 }, { "epoch": 1.7330695834303003, "grad_norm": 0.7304714322090149, "learning_rate": 0.00037572492189064436, "loss": 1.6165568828582764, "step": 14894 }, { "epoch": 1.7331859436816384, "grad_norm": 1.0961614847183228, "learning_rate": 0.00037552735563372853, "loss": 2.1841611862182617, "step": 14895 }, { "epoch": 1.7333023039329765, "grad_norm": 1.3849776983261108, "learning_rate": 0.0003753296845001085, "loss": 1.9836193323135376, "step": 14896 }, { "epoch": 1.7334186641843146, "grad_norm": 1.0769100189208984, "learning_rate": 0.0003751319086549364, "loss": 2.5443201065063477, "step": 14897 }, { "epoch": 1.7335350244356529, "grad_norm": 0.7980469465255737, "learning_rate": 0.0003749340282634517, "loss": 1.309158444404602, "step": 14898 }, { "epoch": 1.7336513846869908, "grad_norm": 1.1253737211227417, "learning_rate": 0.00037473604349098113, "loss": 2.645176887512207, "step": 14899 }, { "epoch": 1.733767744938329, "grad_norm": 1.229604959487915, "learning_rate": 0.0003745379545029389, "loss": 1.8152167797088623, "step": 14900 }, { "epoch": 1.7338841051896672, "grad_norm": 0.9177554845809937, "learning_rate": 0.000374339761464826, "loss": 1.7946276664733887, "step": 14901 }, { "epoch": 1.7340004654410053, "grad_norm": 0.817243754863739, "learning_rate": 0.0003741414645422306, "loss": 1.3544645309448242, "step": 14902 }, { "epoch": 1.7341168256923436, "grad_norm": 1.0751011371612549, "learning_rate": 0.0003739430639008276, "loss": 2.2218728065490723, "step": 14903 }, { "epoch": 1.7342331859436815, "grad_norm": 0.820819616317749, "learning_rate": 0.00037374455970637843, "loss": 2.0094614028930664, "step": 14904 }, { "epoch": 1.7343495461950198, "grad_norm": 1.5259042978286743, "learning_rate": 0.0003735459521247312, "loss": 2.0220236778259277, "step": 14905 }, { "epoch": 1.734465906446358, "grad_norm": 0.897402286529541, "learning_rate": 0.0003733472413218202, "loss": 1.76144278049469, "step": 14906 }, { "epoch": 1.734582266697696, "grad_norm": 0.7332766652107239, "learning_rate": 0.0003731484274636663, "loss": 1.5716216564178467, "step": 14907 }, { "epoch": 1.7346986269490343, "grad_norm": 1.2693133354187012, "learning_rate": 0.00037294951071637606, "loss": 1.6139180660247803, "step": 14908 }, { "epoch": 1.7348149872003722, "grad_norm": 1.0108290910720825, "learning_rate": 0.0003727504912461424, "loss": 1.5125560760498047, "step": 14909 }, { "epoch": 1.7349313474517105, "grad_norm": 1.139552354812622, "learning_rate": 0.0003725513692192436, "loss": 1.6669011116027832, "step": 14910 }, { "epoch": 1.7350477077030486, "grad_norm": 1.088335394859314, "learning_rate": 0.0003723521448020441, "loss": 1.6089961528778076, "step": 14911 }, { "epoch": 1.7351640679543867, "grad_norm": 2.112973928451538, "learning_rate": 0.00037215281816099365, "loss": 1.4518187046051025, "step": 14912 }, { "epoch": 1.735280428205725, "grad_norm": 1.1721417903900146, "learning_rate": 0.00037195338946262736, "loss": 1.720737338066101, "step": 14913 }, { "epoch": 1.735396788457063, "grad_norm": 0.851760745048523, "learning_rate": 0.0003717538588735657, "loss": 1.2208166122436523, "step": 14914 }, { "epoch": 1.7355131487084012, "grad_norm": 0.7932680249214172, "learning_rate": 0.0003715542265605143, "loss": 1.9812116622924805, "step": 14915 }, { "epoch": 1.7356295089597393, "grad_norm": 0.946251392364502, "learning_rate": 0.0003713544926902637, "loss": 1.8811429738998413, "step": 14916 }, { "epoch": 1.7357458692110774, "grad_norm": 1.1287119388580322, "learning_rate": 0.00037115465742968936, "loss": 1.8941346406936646, "step": 14917 }, { "epoch": 1.7358622294624158, "grad_norm": 1.5044678449630737, "learning_rate": 0.00037095472094575134, "loss": 1.516169548034668, "step": 14918 }, { "epoch": 1.7359785897137536, "grad_norm": 1.1450741291046143, "learning_rate": 0.0003707546834054943, "loss": 2.773831844329834, "step": 14919 }, { "epoch": 1.736094949965092, "grad_norm": 1.4142564535140991, "learning_rate": 0.00037055454497604746, "loss": 1.632807970046997, "step": 14920 }, { "epoch": 1.73621131021643, "grad_norm": 1.0120216608047485, "learning_rate": 0.0003703543058246242, "loss": 1.8388162851333618, "step": 14921 }, { "epoch": 1.7363276704677681, "grad_norm": 0.9963181018829346, "learning_rate": 0.0003701539661185222, "loss": 2.105228900909424, "step": 14922 }, { "epoch": 1.7364440307191065, "grad_norm": 1.453978419303894, "learning_rate": 0.0003699535260251229, "loss": 1.890036940574646, "step": 14923 }, { "epoch": 1.7365603909704443, "grad_norm": 0.7174695134162903, "learning_rate": 0.0003697529857118919, "loss": 1.468254804611206, "step": 14924 }, { "epoch": 1.7366767512217827, "grad_norm": 0.9648765921592712, "learning_rate": 0.00036955234534637827, "loss": 2.180377960205078, "step": 14925 }, { "epoch": 1.7367931114731208, "grad_norm": 0.8136347532272339, "learning_rate": 0.0003693516050962148, "loss": 1.5325210094451904, "step": 14926 }, { "epoch": 1.7369094717244589, "grad_norm": 1.1164132356643677, "learning_rate": 0.0003691507651291178, "loss": 1.8912783861160278, "step": 14927 }, { "epoch": 1.7370258319757972, "grad_norm": 1.45402991771698, "learning_rate": 0.00036894982561288674, "loss": 2.481964349746704, "step": 14928 }, { "epoch": 1.737142192227135, "grad_norm": 1.0667492151260376, "learning_rate": 0.0003687487867154045, "loss": 2.0140442848205566, "step": 14929 }, { "epoch": 1.7372585524784734, "grad_norm": 0.9508426189422607, "learning_rate": 0.0003685476486046366, "loss": 1.6109302043914795, "step": 14930 }, { "epoch": 1.7373749127298115, "grad_norm": 1.0564720630645752, "learning_rate": 0.000368346411448632, "loss": 1.831624984741211, "step": 14931 }, { "epoch": 1.7374912729811496, "grad_norm": 1.4109560251235962, "learning_rate": 0.0003681450754155217, "loss": 2.1612937450408936, "step": 14932 }, { "epoch": 1.737607633232488, "grad_norm": 1.2557058334350586, "learning_rate": 0.00036794364067352007, "loss": 2.2597155570983887, "step": 14933 }, { "epoch": 1.7377239934838258, "grad_norm": 1.4288156032562256, "learning_rate": 0.0003677421073909233, "loss": 2.0328996181488037, "step": 14934 }, { "epoch": 1.737840353735164, "grad_norm": 1.2148336172103882, "learning_rate": 0.0003675404757361104, "loss": 2.2355451583862305, "step": 14935 }, { "epoch": 1.7379567139865022, "grad_norm": 0.7998300194740295, "learning_rate": 0.0003673387458775424, "loss": 1.6773673295974731, "step": 14936 }, { "epoch": 1.7380730742378403, "grad_norm": 1.1325284242630005, "learning_rate": 0.00036713691798376214, "loss": 1.801598310470581, "step": 14937 }, { "epoch": 1.7381894344891786, "grad_norm": 1.4477404356002808, "learning_rate": 0.00036693499222339467, "loss": 1.8412147760391235, "step": 14938 }, { "epoch": 1.7383057947405165, "grad_norm": 0.4111109972000122, "learning_rate": 0.0003667329687651467, "loss": 0.9215319156646729, "step": 14939 }, { "epoch": 1.7384221549918548, "grad_norm": 1.4503782987594604, "learning_rate": 0.00036653084777780667, "loss": 2.3526411056518555, "step": 14940 }, { "epoch": 1.738538515243193, "grad_norm": 11.276339530944824, "learning_rate": 0.0003663286294302443, "loss": 1.9537904262542725, "step": 14941 }, { "epoch": 1.738654875494531, "grad_norm": 0.9701865911483765, "learning_rate": 0.0003661263138914108, "loss": 2.177485942840576, "step": 14942 }, { "epoch": 1.7387712357458693, "grad_norm": 1.1814967393875122, "learning_rate": 0.0003659239013303385, "loss": 1.3510901927947998, "step": 14943 }, { "epoch": 1.7388875959972072, "grad_norm": 1.1606237888336182, "learning_rate": 0.0003657213919161408, "loss": 1.3020237684249878, "step": 14944 }, { "epoch": 1.7390039562485455, "grad_norm": 1.1815000772476196, "learning_rate": 0.0003655187858180122, "loss": 2.467327356338501, "step": 14945 }, { "epoch": 1.7391203164998836, "grad_norm": 0.9926049113273621, "learning_rate": 0.0003653160832052277, "loss": 2.3311941623687744, "step": 14946 }, { "epoch": 1.7392366767512217, "grad_norm": 1.6402220726013184, "learning_rate": 0.00036511328424714326, "loss": 2.3025879859924316, "step": 14947 }, { "epoch": 1.73935303700256, "grad_norm": 0.9625191688537598, "learning_rate": 0.000364910389113195, "loss": 1.9014832973480225, "step": 14948 }, { "epoch": 1.739469397253898, "grad_norm": 1.2158602476119995, "learning_rate": 0.0003647073979728994, "loss": 1.4087860584259033, "step": 14949 }, { "epoch": 1.7395857575052363, "grad_norm": 1.087723970413208, "learning_rate": 0.00036450431099585356, "loss": 1.8575108051300049, "step": 14950 }, { "epoch": 1.7397021177565744, "grad_norm": 0.7581561207771301, "learning_rate": 0.0003643011283517341, "loss": 1.6419355869293213, "step": 14951 }, { "epoch": 1.7398184780079125, "grad_norm": 0.2560487389564514, "learning_rate": 0.0003640978502102982, "loss": 0.8655884265899658, "step": 14952 }, { "epoch": 1.7399348382592508, "grad_norm": 0.9507570862770081, "learning_rate": 0.00036389447674138223, "loss": 2.3384780883789062, "step": 14953 }, { "epoch": 1.7400511985105886, "grad_norm": 1.4940205812454224, "learning_rate": 0.00036369100811490255, "loss": 1.5049848556518555, "step": 14954 }, { "epoch": 1.740167558761927, "grad_norm": 1.0985006093978882, "learning_rate": 0.0003634874445008548, "loss": 1.823809027671814, "step": 14955 }, { "epoch": 1.740283919013265, "grad_norm": 1.2859268188476562, "learning_rate": 0.00036328378606931424, "loss": 1.91239595413208, "step": 14956 }, { "epoch": 1.7404002792646032, "grad_norm": 1.1226552724838257, "learning_rate": 0.00036308003299043525, "loss": 1.4627182483673096, "step": 14957 }, { "epoch": 1.7405166395159415, "grad_norm": 1.276896357536316, "learning_rate": 0.000362876185434451, "loss": 1.3119149208068848, "step": 14958 }, { "epoch": 1.7406329997672794, "grad_norm": 1.2583141326904297, "learning_rate": 0.00036267224357167414, "loss": 1.825251579284668, "step": 14959 }, { "epoch": 1.7407493600186177, "grad_norm": 1.1604336500167847, "learning_rate": 0.00036246820757249577, "loss": 1.8828173875808716, "step": 14960 }, { "epoch": 1.7408657202699558, "grad_norm": 1.0781997442245483, "learning_rate": 0.0003622640776073856, "loss": 2.023348331451416, "step": 14961 }, { "epoch": 1.7409820805212939, "grad_norm": 0.902349591255188, "learning_rate": 0.0003620598538468918, "loss": 1.6048142910003662, "step": 14962 }, { "epoch": 1.7410984407726322, "grad_norm": 1.0809773206710815, "learning_rate": 0.0003618555364616413, "loss": 2.2182490825653076, "step": 14963 }, { "epoch": 1.74121480102397, "grad_norm": 0.41504016518592834, "learning_rate": 0.00036165112562233886, "loss": 1.0719863176345825, "step": 14964 }, { "epoch": 1.7413311612753084, "grad_norm": 2.069174289703369, "learning_rate": 0.0003614466214997675, "loss": 2.599529266357422, "step": 14965 }, { "epoch": 1.7414475215266465, "grad_norm": 1.291735291481018, "learning_rate": 0.0003612420242647882, "loss": 1.7095263004302979, "step": 14966 }, { "epoch": 1.7415638817779846, "grad_norm": 0.7831935882568359, "learning_rate": 0.0003610373340883395, "loss": 1.7175970077514648, "step": 14967 }, { "epoch": 1.741680242029323, "grad_norm": 0.9140563011169434, "learning_rate": 0.0003608325511414379, "loss": 1.6171743869781494, "step": 14968 }, { "epoch": 1.7417966022806608, "grad_norm": 0.9633253216743469, "learning_rate": 0.00036062767559517716, "loss": 1.7099241018295288, "step": 14969 }, { "epoch": 1.7419129625319991, "grad_norm": 1.1814075708389282, "learning_rate": 0.00036042270762072856, "loss": 1.706937313079834, "step": 14970 }, { "epoch": 1.7420293227833372, "grad_norm": 0.9753248691558838, "learning_rate": 0.0003602176473893406, "loss": 2.0155715942382812, "step": 14971 }, { "epoch": 1.7421456830346753, "grad_norm": 1.212372064590454, "learning_rate": 0.0003600124950723388, "loss": 2.1354923248291016, "step": 14972 }, { "epoch": 1.7422620432860136, "grad_norm": 1.3793412446975708, "learning_rate": 0.00035980725084112556, "loss": 2.4629857540130615, "step": 14973 }, { "epoch": 1.7423784035373515, "grad_norm": 1.3644884824752808, "learning_rate": 0.00035960191486718014, "loss": 2.5599889755249023, "step": 14974 }, { "epoch": 1.7424947637886898, "grad_norm": 1.1634786128997803, "learning_rate": 0.00035939648732205847, "loss": 1.792891502380371, "step": 14975 }, { "epoch": 1.742611124040028, "grad_norm": 1.3529258966445923, "learning_rate": 0.00035919096837739304, "loss": 1.6381704807281494, "step": 14976 }, { "epoch": 1.742727484291366, "grad_norm": 0.7108595371246338, "learning_rate": 0.0003589853582048926, "loss": 1.9148223400115967, "step": 14977 }, { "epoch": 1.7428438445427044, "grad_norm": 1.035975456237793, "learning_rate": 0.00035877965697634213, "loss": 2.598384380340576, "step": 14978 }, { "epoch": 1.7429602047940422, "grad_norm": 0.998360276222229, "learning_rate": 0.00035857386486360277, "loss": 1.8661106824874878, "step": 14979 }, { "epoch": 1.7430765650453806, "grad_norm": 1.3453501462936401, "learning_rate": 0.0003583679820386114, "loss": 1.84804105758667, "step": 14980 }, { "epoch": 1.7431929252967187, "grad_norm": 1.1110155582427979, "learning_rate": 0.000358162008673381, "loss": 2.334111213684082, "step": 14981 }, { "epoch": 1.7433092855480568, "grad_norm": 1.0469874143600464, "learning_rate": 0.0003579559449399999, "loss": 1.2645872831344604, "step": 14982 }, { "epoch": 1.743425645799395, "grad_norm": 0.8466288447380066, "learning_rate": 0.000357749791010632, "loss": 2.0275490283966064, "step": 14983 }, { "epoch": 1.743542006050733, "grad_norm": 0.718880295753479, "learning_rate": 0.0003575435470575168, "loss": 1.4600067138671875, "step": 14984 }, { "epoch": 1.7436583663020713, "grad_norm": 0.4405234158039093, "learning_rate": 0.00035733721325296867, "loss": 1.0021724700927734, "step": 14985 }, { "epoch": 1.7437747265534094, "grad_norm": 1.4780097007751465, "learning_rate": 0.0003571307897693771, "loss": 2.09262752532959, "step": 14986 }, { "epoch": 1.7438910868047475, "grad_norm": 0.7679460048675537, "learning_rate": 0.0003569242767792069, "loss": 1.4331507682800293, "step": 14987 }, { "epoch": 1.7440074470560858, "grad_norm": 1.1991502046585083, "learning_rate": 0.0003567176744549971, "loss": 2.5372486114501953, "step": 14988 }, { "epoch": 1.7441238073074237, "grad_norm": 1.33937406539917, "learning_rate": 0.00035651098296936176, "loss": 1.8803503513336182, "step": 14989 }, { "epoch": 1.744240167558762, "grad_norm": 1.0558149814605713, "learning_rate": 0.0003563042024949893, "loss": 2.1335151195526123, "step": 14990 }, { "epoch": 1.7443565278101, "grad_norm": 1.0812495946884155, "learning_rate": 0.00035609733320464247, "loss": 1.5467345714569092, "step": 14991 }, { "epoch": 1.7444728880614382, "grad_norm": 1.1096141338348389, "learning_rate": 0.0003558903752711583, "loss": 2.142235517501831, "step": 14992 }, { "epoch": 1.7445892483127765, "grad_norm": 0.8072879910469055, "learning_rate": 0.0003556833288674478, "loss": 1.3566744327545166, "step": 14993 }, { "epoch": 1.7447056085641144, "grad_norm": 1.495894432067871, "learning_rate": 0.00035547619416649595, "loss": 2.529674530029297, "step": 14994 }, { "epoch": 1.7448219688154527, "grad_norm": 0.9896625876426697, "learning_rate": 0.0003552689713413614, "loss": 2.3759591579437256, "step": 14995 }, { "epoch": 1.7449383290667908, "grad_norm": 1.2265757322311401, "learning_rate": 0.00035506166056517666, "loss": 2.331953287124634, "step": 14996 }, { "epoch": 1.745054689318129, "grad_norm": 1.067365288734436, "learning_rate": 0.0003548542620111474, "loss": 1.9022177457809448, "step": 14997 }, { "epoch": 1.7451710495694672, "grad_norm": 1.2939574718475342, "learning_rate": 0.0003546467758525529, "loss": 2.270352840423584, "step": 14998 }, { "epoch": 1.745287409820805, "grad_norm": 0.930367648601532, "learning_rate": 0.0003544392022627455, "loss": 1.931063175201416, "step": 14999 }, { "epoch": 1.7454037700721434, "grad_norm": 0.8172273635864258, "learning_rate": 0.00035423154141515055, "loss": 1.874072790145874, "step": 15000 }, { "epoch": 1.7455201303234815, "grad_norm": 1.0492860078811646, "learning_rate": 0.0003540237934832665, "loss": 2.0520834922790527, "step": 15001 }, { "epoch": 1.7456364905748196, "grad_norm": 1.2726168632507324, "learning_rate": 0.00035381595864066435, "loss": 1.9778735637664795, "step": 15002 }, { "epoch": 1.745752850826158, "grad_norm": 2.7910494804382324, "learning_rate": 0.00035360803706098784, "loss": 1.6777024269104004, "step": 15003 }, { "epoch": 1.7458692110774958, "grad_norm": 0.887399435043335, "learning_rate": 0.000353400028917953, "loss": 1.7815744876861572, "step": 15004 }, { "epoch": 1.7459855713288341, "grad_norm": 0.7809194922447205, "learning_rate": 0.00035319193438534854, "loss": 1.6972944736480713, "step": 15005 }, { "epoch": 1.7461019315801722, "grad_norm": 1.6798285245895386, "learning_rate": 0.00035298375363703496, "loss": 2.742272138595581, "step": 15006 }, { "epoch": 1.7462182918315103, "grad_norm": 0.9728901982307434, "learning_rate": 0.00035277548684694486, "loss": 2.0761325359344482, "step": 15007 }, { "epoch": 1.7463346520828487, "grad_norm": 1.1314091682434082, "learning_rate": 0.00035256713418908305, "loss": 1.8089878559112549, "step": 15008 }, { "epoch": 1.7464510123341865, "grad_norm": 0.6810213327407837, "learning_rate": 0.00035235869583752566, "loss": 1.5933889150619507, "step": 15009 }, { "epoch": 1.7465673725855249, "grad_norm": 1.2489937543869019, "learning_rate": 0.0003521501719664207, "loss": 2.231372356414795, "step": 15010 }, { "epoch": 1.746683732836863, "grad_norm": 2.066723346710205, "learning_rate": 0.00035194156274998754, "loss": 1.7570397853851318, "step": 15011 }, { "epoch": 1.746800093088201, "grad_norm": 1.1100176572799683, "learning_rate": 0.0003517328683625168, "loss": 1.5550429821014404, "step": 15012 }, { "epoch": 1.7469164533395394, "grad_norm": 0.7679476141929626, "learning_rate": 0.00035152408897837033, "loss": 1.6559865474700928, "step": 15013 }, { "epoch": 1.7470328135908773, "grad_norm": 0.8858062028884888, "learning_rate": 0.000351315224771981, "loss": 1.542320966720581, "step": 15014 }, { "epoch": 1.7471491738422156, "grad_norm": 1.212936282157898, "learning_rate": 0.00035110627591785254, "loss": 1.6042487621307373, "step": 15015 }, { "epoch": 1.7472655340935537, "grad_norm": 0.8385387063026428, "learning_rate": 0.0003508972425905593, "loss": 1.902368187904358, "step": 15016 }, { "epoch": 1.7473818943448918, "grad_norm": 1.3502503633499146, "learning_rate": 0.0003506881249647464, "loss": 1.8486740589141846, "step": 15017 }, { "epoch": 1.74749825459623, "grad_norm": 0.7297903299331665, "learning_rate": 0.0003504789232151291, "loss": 1.5251437425613403, "step": 15018 }, { "epoch": 1.747614614847568, "grad_norm": 1.0530099868774414, "learning_rate": 0.0003502696375164934, "loss": 1.6839125156402588, "step": 15019 }, { "epoch": 1.7477309750989063, "grad_norm": 0.7432464361190796, "learning_rate": 0.00035006026804369495, "loss": 1.3656489849090576, "step": 15020 }, { "epoch": 1.7478473353502444, "grad_norm": 1.0940167903900146, "learning_rate": 0.0003498508149716596, "loss": 2.539799690246582, "step": 15021 }, { "epoch": 1.7479636956015825, "grad_norm": 1.0830714702606201, "learning_rate": 0.0003496412784753832, "loss": 2.384644031524658, "step": 15022 }, { "epoch": 1.7480800558529206, "grad_norm": 1.2556945085525513, "learning_rate": 0.0003494316587299311, "loss": 1.7585967779159546, "step": 15023 }, { "epoch": 1.7481964161042587, "grad_norm": 0.7391811609268188, "learning_rate": 0.00034922195591043814, "loss": 1.6665363311767578, "step": 15024 }, { "epoch": 1.748312776355597, "grad_norm": 1.2092311382293701, "learning_rate": 0.00034901217019210883, "loss": 1.8821499347686768, "step": 15025 }, { "epoch": 1.748429136606935, "grad_norm": 0.8826490044593811, "learning_rate": 0.0003488023017502167, "loss": 1.9331250190734863, "step": 15026 }, { "epoch": 1.7485454968582732, "grad_norm": 1.441429615020752, "learning_rate": 0.0003485923507601046, "loss": 2.23905873298645, "step": 15027 }, { "epoch": 1.7486618571096113, "grad_norm": 0.527307391166687, "learning_rate": 0.0003483823173971841, "loss": 1.150529146194458, "step": 15028 }, { "epoch": 1.7487782173609494, "grad_norm": 0.8159105181694031, "learning_rate": 0.0003481722018369359, "loss": 1.6187487840652466, "step": 15029 }, { "epoch": 1.7488945776122877, "grad_norm": 0.633083164691925, "learning_rate": 0.00034796200425490903, "loss": 1.1174945831298828, "step": 15030 }, { "epoch": 1.7490109378636258, "grad_norm": 1.3135112524032593, "learning_rate": 0.00034775172482672136, "loss": 2.4066505432128906, "step": 15031 }, { "epoch": 1.749127298114964, "grad_norm": 1.4475455284118652, "learning_rate": 0.0003475413637280589, "loss": 2.006993055343628, "step": 15032 }, { "epoch": 1.749243658366302, "grad_norm": 1.1602720022201538, "learning_rate": 0.000347330921134676, "loss": 2.9015536308288574, "step": 15033 }, { "epoch": 1.7493600186176401, "grad_norm": 1.0287740230560303, "learning_rate": 0.00034712039722239525, "loss": 2.009324789047241, "step": 15034 }, { "epoch": 1.7494763788689784, "grad_norm": 1.032100796699524, "learning_rate": 0.0003469097921671068, "loss": 1.6203083992004395, "step": 15035 }, { "epoch": 1.7495927391203165, "grad_norm": 1.2200778722763062, "learning_rate": 0.0003466991061447691, "loss": 2.2563018798828125, "step": 15036 }, { "epoch": 1.7497090993716546, "grad_norm": 1.6818886995315552, "learning_rate": 0.00034648833933140764, "loss": 1.7777540683746338, "step": 15037 }, { "epoch": 1.7498254596229927, "grad_norm": 0.912788987159729, "learning_rate": 0.000346277491903116, "loss": 2.384551763534546, "step": 15038 }, { "epoch": 1.7499418198743308, "grad_norm": 0.7620605230331421, "learning_rate": 0.0003460665640360547, "loss": 1.9786648750305176, "step": 15039 }, { "epoch": 1.7500581801256692, "grad_norm": 0.996859610080719, "learning_rate": 0.00034585555590645173, "loss": 1.7781076431274414, "step": 15040 }, { "epoch": 1.7501745403770073, "grad_norm": 0.9528031945228577, "learning_rate": 0.000345644467690602, "loss": 1.4932079315185547, "step": 15041 }, { "epoch": 1.7502909006283454, "grad_norm": 1.0060718059539795, "learning_rate": 0.00034543329956486724, "loss": 2.1120755672454834, "step": 15042 }, { "epoch": 1.7504072608796835, "grad_norm": 1.0651592016220093, "learning_rate": 0.0003452220517056762, "loss": 2.11981201171875, "step": 15043 }, { "epoch": 1.7505236211310216, "grad_norm": 0.7752423286437988, "learning_rate": 0.0003450107242895241, "loss": 1.4549686908721924, "step": 15044 }, { "epoch": 1.7506399813823599, "grad_norm": 0.9000661373138428, "learning_rate": 0.00034479931749297263, "loss": 1.6208906173706055, "step": 15045 }, { "epoch": 1.750756341633698, "grad_norm": 1.304671287536621, "learning_rate": 0.00034458783149264973, "loss": 2.344817638397217, "step": 15046 }, { "epoch": 1.750872701885036, "grad_norm": 0.7451351284980774, "learning_rate": 0.0003443762664652497, "loss": 1.260498285293579, "step": 15047 }, { "epoch": 1.7509890621363742, "grad_norm": 0.8558517098426819, "learning_rate": 0.0003441646225875326, "loss": 1.9148950576782227, "step": 15048 }, { "epoch": 1.7511054223877123, "grad_norm": 1.583593726158142, "learning_rate": 0.0003439529000363247, "loss": 2.683046579360962, "step": 15049 }, { "epoch": 1.7512217826390506, "grad_norm": 1.587469458580017, "learning_rate": 0.0003437410989885177, "loss": 1.707097053527832, "step": 15050 }, { "epoch": 1.7513381428903887, "grad_norm": 0.8116244673728943, "learning_rate": 0.000343529219621069, "loss": 2.2170774936676025, "step": 15051 }, { "epoch": 1.7514545031417268, "grad_norm": 1.500540852546692, "learning_rate": 0.0003433172621110017, "loss": 2.24788498878479, "step": 15052 }, { "epoch": 1.751570863393065, "grad_norm": 0.4348095953464508, "learning_rate": 0.00034310522663540355, "loss": 0.9228298664093018, "step": 15053 }, { "epoch": 1.751687223644403, "grad_norm": 1.40663743019104, "learning_rate": 0.00034289311337142817, "loss": 1.8929380178451538, "step": 15054 }, { "epoch": 1.7518035838957413, "grad_norm": 1.250631332397461, "learning_rate": 0.0003426809224962936, "loss": 1.705726981163025, "step": 15055 }, { "epoch": 1.7519199441470794, "grad_norm": 0.7017019391059875, "learning_rate": 0.00034246865418728303, "loss": 1.6879143714904785, "step": 15056 }, { "epoch": 1.7520363043984175, "grad_norm": 0.7501636743545532, "learning_rate": 0.0003422563086217443, "loss": 2.2627604007720947, "step": 15057 }, { "epoch": 1.7521526646497556, "grad_norm": 1.2696667909622192, "learning_rate": 0.00034204388597708964, "loss": 2.102412700653076, "step": 15058 }, { "epoch": 1.7522690249010937, "grad_norm": 1.0463693141937256, "learning_rate": 0.00034183138643079585, "loss": 1.7665958404541016, "step": 15059 }, { "epoch": 1.752385385152432, "grad_norm": 1.0324701070785522, "learning_rate": 0.000341618810160404, "loss": 2.288721799850464, "step": 15060 }, { "epoch": 1.75250174540377, "grad_norm": 1.0015538930892944, "learning_rate": 0.00034140615734351917, "loss": 1.2651867866516113, "step": 15061 }, { "epoch": 1.7526181056551082, "grad_norm": 0.8638918399810791, "learning_rate": 0.0003411934281578103, "loss": 1.5021213293075562, "step": 15062 }, { "epoch": 1.7527344659064463, "grad_norm": 0.7232128381729126, "learning_rate": 0.0003409806227810103, "loss": 1.531151533126831, "step": 15063 }, { "epoch": 1.7528508261577844, "grad_norm": 1.373441457748413, "learning_rate": 0.0003407677413909158, "loss": 1.9876526594161987, "step": 15064 }, { "epoch": 1.7529671864091227, "grad_norm": 0.8572701215744019, "learning_rate": 0.0003405547841653867, "loss": 1.5695114135742188, "step": 15065 }, { "epoch": 1.7530835466604606, "grad_norm": 1.2127686738967896, "learning_rate": 0.0003403417512823465, "loss": 1.1847589015960693, "step": 15066 }, { "epoch": 1.753199906911799, "grad_norm": 1.281949758529663, "learning_rate": 0.00034012864291978163, "loss": 2.159114360809326, "step": 15067 }, { "epoch": 1.753316267163137, "grad_norm": 0.9960168600082397, "learning_rate": 0.000339915459255742, "loss": 1.3700051307678223, "step": 15068 }, { "epoch": 1.7534326274144751, "grad_norm": 1.1751799583435059, "learning_rate": 0.0003397022004683398, "loss": 2.1364293098449707, "step": 15069 }, { "epoch": 1.7535489876658135, "grad_norm": 1.0184632539749146, "learning_rate": 0.0003394888667357507, "loss": 2.540383815765381, "step": 15070 }, { "epoch": 1.7536653479171513, "grad_norm": 0.6885095238685608, "learning_rate": 0.00033927545823621253, "loss": 1.625402808189392, "step": 15071 }, { "epoch": 1.7537817081684897, "grad_norm": 0.7599919438362122, "learning_rate": 0.00033906197514802574, "loss": 2.1657729148864746, "step": 15072 }, { "epoch": 1.7538980684198278, "grad_norm": 1.1643321514129639, "learning_rate": 0.000338848417649553, "loss": 2.0339601039886475, "step": 15073 }, { "epoch": 1.7540144286711659, "grad_norm": 1.1999423503875732, "learning_rate": 0.0003386347859192193, "loss": 1.4440858364105225, "step": 15074 }, { "epoch": 1.7541307889225042, "grad_norm": 0.9829688668251038, "learning_rate": 0.0003384210801355115, "loss": 1.7344541549682617, "step": 15075 }, { "epoch": 1.754247149173842, "grad_norm": 0.8426505327224731, "learning_rate": 0.00033820730047697835, "loss": 2.116224527359009, "step": 15076 }, { "epoch": 1.7543635094251804, "grad_norm": 0.8468539118766785, "learning_rate": 0.0003379934471222305, "loss": 1.7308175563812256, "step": 15077 }, { "epoch": 1.7544798696765185, "grad_norm": 0.8059658408164978, "learning_rate": 0.00033777952024994004, "loss": 1.2106783390045166, "step": 15078 }, { "epoch": 1.7545962299278566, "grad_norm": 0.5578182935714722, "learning_rate": 0.00033756552003884034, "loss": 1.2605704069137573, "step": 15079 }, { "epoch": 1.754712590179195, "grad_norm": 1.5496063232421875, "learning_rate": 0.0003373514466677263, "loss": 2.4583373069763184, "step": 15080 }, { "epoch": 1.7548289504305328, "grad_norm": 0.9047459363937378, "learning_rate": 0.0003371373003154538, "loss": 1.2700366973876953, "step": 15081 }, { "epoch": 1.754945310681871, "grad_norm": 1.4789026975631714, "learning_rate": 0.0003369230811609397, "loss": 2.0799033641815186, "step": 15082 }, { "epoch": 1.7550616709332092, "grad_norm": 0.9082720279693604, "learning_rate": 0.0003367087893831618, "loss": 2.176300287246704, "step": 15083 }, { "epoch": 1.7551780311845473, "grad_norm": 1.2475613355636597, "learning_rate": 0.00033649442516115846, "loss": 1.5063867568969727, "step": 15084 }, { "epoch": 1.7552943914358856, "grad_norm": 1.1167473793029785, "learning_rate": 0.0003362799886740285, "loss": 1.9028279781341553, "step": 15085 }, { "epoch": 1.7554107516872235, "grad_norm": 1.1402862071990967, "learning_rate": 0.0003360654801009313, "loss": 1.5085813999176025, "step": 15086 }, { "epoch": 1.7555271119385618, "grad_norm": 1.1123948097229004, "learning_rate": 0.0003358508996210863, "loss": 1.5760221481323242, "step": 15087 }, { "epoch": 1.7556434721899, "grad_norm": 1.3165700435638428, "learning_rate": 0.0003356362474137731, "loss": 1.4766240119934082, "step": 15088 }, { "epoch": 1.755759832441238, "grad_norm": 1.2731561660766602, "learning_rate": 0.0003354215236583314, "loss": 2.0232789516448975, "step": 15089 }, { "epoch": 1.7558761926925763, "grad_norm": 0.8909504413604736, "learning_rate": 0.0003352067285341603, "loss": 1.5316650867462158, "step": 15090 }, { "epoch": 1.7559925529439142, "grad_norm": 0.9515268206596375, "learning_rate": 0.0003349918622207188, "loss": 1.9763495922088623, "step": 15091 }, { "epoch": 1.7561089131952525, "grad_norm": 1.0708417892456055, "learning_rate": 0.00033477692489752544, "loss": 1.6828724145889282, "step": 15092 }, { "epoch": 1.7562252734465906, "grad_norm": 0.8920538425445557, "learning_rate": 0.0003345619167441576, "loss": 2.0234334468841553, "step": 15093 }, { "epoch": 1.7563416336979287, "grad_norm": 0.7483212351799011, "learning_rate": 0.0003343468379402525, "loss": 1.7963151931762695, "step": 15094 }, { "epoch": 1.756457993949267, "grad_norm": 1.4008166790008545, "learning_rate": 0.000334131688665506, "loss": 1.7423814535140991, "step": 15095 }, { "epoch": 1.756574354200605, "grad_norm": 1.2669259309768677, "learning_rate": 0.0003339164690996729, "loss": 2.016378879547119, "step": 15096 }, { "epoch": 1.7566907144519432, "grad_norm": 1.4087144136428833, "learning_rate": 0.0003337011794225668, "loss": 2.0602943897247314, "step": 15097 }, { "epoch": 1.7568070747032813, "grad_norm": 0.6436111330986023, "learning_rate": 0.0003334858198140597, "loss": 1.8039908409118652, "step": 15098 }, { "epoch": 1.7569234349546194, "grad_norm": 1.0434879064559937, "learning_rate": 0.00033327039045408225, "loss": 1.7830572128295898, "step": 15099 }, { "epoch": 1.7570397952059578, "grad_norm": 1.1227161884307861, "learning_rate": 0.00033305489152262325, "loss": 1.8866004943847656, "step": 15100 }, { "epoch": 1.7571561554572956, "grad_norm": 0.9668207168579102, "learning_rate": 0.0003328393231997296, "loss": 2.0223886966705322, "step": 15101 }, { "epoch": 1.757272515708634, "grad_norm": 1.1122180223464966, "learning_rate": 0.0003326236856655063, "loss": 1.298290491104126, "step": 15102 }, { "epoch": 1.757388875959972, "grad_norm": 0.9379985928535461, "learning_rate": 0.0003324079791001161, "loss": 1.540099859237671, "step": 15103 }, { "epoch": 1.7575052362113102, "grad_norm": 0.9181327223777771, "learning_rate": 0.00033219220368377946, "loss": 1.3853003978729248, "step": 15104 }, { "epoch": 1.7576215964626485, "grad_norm": 3.111300230026245, "learning_rate": 0.00033197635959677417, "loss": 1.8266984224319458, "step": 15105 }, { "epoch": 1.7577379567139864, "grad_norm": 0.9278640151023865, "learning_rate": 0.00033176044701943564, "loss": 1.2644988298416138, "step": 15106 }, { "epoch": 1.7578543169653247, "grad_norm": 1.1067901849746704, "learning_rate": 0.0003315444661321565, "loss": 2.4197592735290527, "step": 15107 }, { "epoch": 1.7579706772166628, "grad_norm": 0.8848555088043213, "learning_rate": 0.0003313284171153863, "loss": 1.3190646171569824, "step": 15108 }, { "epoch": 1.7580870374680009, "grad_norm": 1.219762921333313, "learning_rate": 0.0003311123001496316, "loss": 2.083745002746582, "step": 15109 }, { "epoch": 1.7582033977193392, "grad_norm": 0.5757243037223816, "learning_rate": 0.00033089611541545574, "loss": 1.1726453304290771, "step": 15110 }, { "epoch": 1.758319757970677, "grad_norm": 1.0103306770324707, "learning_rate": 0.00033067986309347853, "loss": 2.1560230255126953, "step": 15111 }, { "epoch": 1.7584361182220154, "grad_norm": 1.2281427383422852, "learning_rate": 0.00033046354336437647, "loss": 1.677394151687622, "step": 15112 }, { "epoch": 1.7585524784733535, "grad_norm": 0.8389877676963806, "learning_rate": 0.00033024715640888225, "loss": 1.8753430843353271, "step": 15113 }, { "epoch": 1.7586688387246916, "grad_norm": 1.1989669799804688, "learning_rate": 0.00033003070240778475, "loss": 2.464109420776367, "step": 15114 }, { "epoch": 1.75878519897603, "grad_norm": 0.9593507051467896, "learning_rate": 0.000329814181541929, "loss": 2.0883255004882812, "step": 15115 }, { "epoch": 1.7589015592273678, "grad_norm": 1.1602433919906616, "learning_rate": 0.0003295975939922156, "loss": 1.9157493114471436, "step": 15116 }, { "epoch": 1.7590179194787061, "grad_norm": 1.388975977897644, "learning_rate": 0.00032938093993960114, "loss": 2.705333709716797, "step": 15117 }, { "epoch": 1.7591342797300442, "grad_norm": 0.9239094853401184, "learning_rate": 0.0003291642195650975, "loss": 2.478393077850342, "step": 15118 }, { "epoch": 1.7592506399813823, "grad_norm": 0.8406659960746765, "learning_rate": 0.00032894743304977226, "loss": 2.2708017826080322, "step": 15119 }, { "epoch": 1.7593670002327206, "grad_norm": 1.9210270643234253, "learning_rate": 0.0003287305805747481, "loss": 2.0796408653259277, "step": 15120 }, { "epoch": 1.7594833604840585, "grad_norm": 1.0066949129104614, "learning_rate": 0.00032851366232120293, "loss": 1.7469065189361572, "step": 15121 }, { "epoch": 1.7595997207353968, "grad_norm": 1.478852391242981, "learning_rate": 0.00032829667847036935, "loss": 2.343885660171509, "step": 15122 }, { "epoch": 1.759716080986735, "grad_norm": 0.8250636458396912, "learning_rate": 0.0003280796292035352, "loss": 1.6626371145248413, "step": 15123 }, { "epoch": 1.759832441238073, "grad_norm": 1.2372080087661743, "learning_rate": 0.00032786251470204243, "loss": 1.7608311176300049, "step": 15124 }, { "epoch": 1.7599488014894114, "grad_norm": 0.8371021151542664, "learning_rate": 0.000327645335147288, "loss": 1.5191688537597656, "step": 15125 }, { "epoch": 1.7600651617407492, "grad_norm": 0.7738123536109924, "learning_rate": 0.00032742809072072287, "loss": 1.607990026473999, "step": 15126 }, { "epoch": 1.7601815219920875, "grad_norm": 0.995027482509613, "learning_rate": 0.0003272107816038524, "loss": 1.6917637586593628, "step": 15127 }, { "epoch": 1.7602978822434256, "grad_norm": 0.6591373085975647, "learning_rate": 0.00032699340797823597, "loss": 1.5124868154525757, "step": 15128 }, { "epoch": 1.7604142424947637, "grad_norm": 1.3041502237319946, "learning_rate": 0.00032677597002548676, "loss": 1.4051201343536377, "step": 15129 }, { "epoch": 1.760530602746102, "grad_norm": 1.1162269115447998, "learning_rate": 0.0003265584679272717, "loss": 1.819180965423584, "step": 15130 }, { "epoch": 1.76064696299744, "grad_norm": 1.2816860675811768, "learning_rate": 0.0003263409018653114, "loss": 1.5597996711730957, "step": 15131 }, { "epoch": 1.7607633232487783, "grad_norm": 0.848837673664093, "learning_rate": 0.0003261232720213799, "loss": 1.9311861991882324, "step": 15132 }, { "epoch": 1.7608796835001164, "grad_norm": 1.4756684303283691, "learning_rate": 0.0003259055785773046, "loss": 2.2141971588134766, "step": 15133 }, { "epoch": 1.7609960437514545, "grad_norm": 1.6405442953109741, "learning_rate": 0.0003256878217149658, "loss": 1.9283077716827393, "step": 15134 }, { "epoch": 1.7611124040027928, "grad_norm": 1.042883038520813, "learning_rate": 0.0003254700016162969, "loss": 2.4238171577453613, "step": 15135 }, { "epoch": 1.7612287642541307, "grad_norm": 0.838479220867157, "learning_rate": 0.0003252521184632843, "loss": 2.0435359477996826, "step": 15136 }, { "epoch": 1.761345124505469, "grad_norm": 0.6281134486198425, "learning_rate": 0.0003250341724379669, "loss": 1.2375297546386719, "step": 15137 }, { "epoch": 1.761461484756807, "grad_norm": 1.1770048141479492, "learning_rate": 0.00032481616372243607, "loss": 1.7187083959579468, "step": 15138 }, { "epoch": 1.7615778450081452, "grad_norm": 0.9399323463439941, "learning_rate": 0.00032459809249883574, "loss": 1.6810734272003174, "step": 15139 }, { "epoch": 1.7616942052594835, "grad_norm": 0.7084921598434448, "learning_rate": 0.00032437995894936205, "loss": 1.6915653944015503, "step": 15140 }, { "epoch": 1.7618105655108214, "grad_norm": 1.013045072555542, "learning_rate": 0.0003241617632562631, "loss": 1.4191675186157227, "step": 15141 }, { "epoch": 1.7619269257621597, "grad_norm": 0.8528897762298584, "learning_rate": 0.0003239435056018388, "loss": 1.7128894329071045, "step": 15142 }, { "epoch": 1.7620432860134978, "grad_norm": 1.8837343454360962, "learning_rate": 0.0003237251861684412, "loss": 2.1616337299346924, "step": 15143 }, { "epoch": 1.762159646264836, "grad_norm": 1.4374114274978638, "learning_rate": 0.0003235068051384736, "loss": 1.959468126296997, "step": 15144 }, { "epoch": 1.7622760065161742, "grad_norm": 1.8504444360733032, "learning_rate": 0.000323288362694391, "loss": 2.1983892917633057, "step": 15145 }, { "epoch": 1.762392366767512, "grad_norm": 2.2512457370758057, "learning_rate": 0.00032306985901869964, "loss": 2.947531223297119, "step": 15146 }, { "epoch": 1.7625087270188504, "grad_norm": 0.754953145980835, "learning_rate": 0.00032285129429395675, "loss": 1.7793388366699219, "step": 15147 }, { "epoch": 1.7626250872701885, "grad_norm": 1.440983533859253, "learning_rate": 0.00032263266870277094, "loss": 1.8352210521697998, "step": 15148 }, { "epoch": 1.7627414475215266, "grad_norm": 0.4830252230167389, "learning_rate": 0.0003224139824278013, "loss": 1.1180076599121094, "step": 15149 }, { "epoch": 1.762857807772865, "grad_norm": 1.1179561614990234, "learning_rate": 0.0003221952356517577, "loss": 2.300178050994873, "step": 15150 }, { "epoch": 1.7629741680242028, "grad_norm": 1.503373384475708, "learning_rate": 0.00032197642855740085, "loss": 1.7252013683319092, "step": 15151 }, { "epoch": 1.7630905282755411, "grad_norm": 0.7977147698402405, "learning_rate": 0.0003217575613275414, "loss": 1.6225024461746216, "step": 15152 }, { "epoch": 1.7632068885268792, "grad_norm": 1.0620150566101074, "learning_rate": 0.0003215386341450408, "loss": 1.9727504253387451, "step": 15153 }, { "epoch": 1.7633232487782173, "grad_norm": 0.9247502684593201, "learning_rate": 0.0003213196471928099, "loss": 2.2373974323272705, "step": 15154 }, { "epoch": 1.7634396090295557, "grad_norm": 0.7317755818367004, "learning_rate": 0.0003211006006538101, "loss": 1.3014050722122192, "step": 15155 }, { "epoch": 1.7635559692808935, "grad_norm": 0.8377525806427002, "learning_rate": 0.00032088149471105215, "loss": 1.6308929920196533, "step": 15156 }, { "epoch": 1.7636723295322319, "grad_norm": 1.4518892765045166, "learning_rate": 0.0003206623295475968, "loss": 2.1799004077911377, "step": 15157 }, { "epoch": 1.76378868978357, "grad_norm": 0.9621354937553406, "learning_rate": 0.000320443105346554, "loss": 2.2486572265625, "step": 15158 }, { "epoch": 1.763905050034908, "grad_norm": 1.5740433931350708, "learning_rate": 0.00032022382229108317, "loss": 1.5170352458953857, "step": 15159 }, { "epoch": 1.7640214102862464, "grad_norm": 0.7960808873176575, "learning_rate": 0.00032000448056439284, "loss": 1.7156786918640137, "step": 15160 }, { "epoch": 1.7641377705375842, "grad_norm": 0.8589920997619629, "learning_rate": 0.00031978508034974053, "loss": 2.350482940673828, "step": 15161 }, { "epoch": 1.7642541307889226, "grad_norm": 0.622676432132721, "learning_rate": 0.0003195656218304327, "loss": 1.172335147857666, "step": 15162 }, { "epoch": 1.7643704910402607, "grad_norm": 1.3239142894744873, "learning_rate": 0.00031934610518982444, "loss": 2.1364340782165527, "step": 15163 }, { "epoch": 1.7644868512915988, "grad_norm": 0.7597663998603821, "learning_rate": 0.00031912653061131957, "loss": 1.41628897190094, "step": 15164 }, { "epoch": 1.764603211542937, "grad_norm": 0.8863285779953003, "learning_rate": 0.00031890689827837024, "loss": 2.0214576721191406, "step": 15165 }, { "epoch": 1.764719571794275, "grad_norm": 0.6227912306785583, "learning_rate": 0.0003186872083744766, "loss": 1.5729749202728271, "step": 15166 }, { "epoch": 1.7648359320456133, "grad_norm": 1.5803602933883667, "learning_rate": 0.0003184674610831872, "loss": 1.3085992336273193, "step": 15167 }, { "epoch": 1.7649522922969514, "grad_norm": 1.2849783897399902, "learning_rate": 0.0003182476565880985, "loss": 2.529209613800049, "step": 15168 }, { "epoch": 1.7650686525482895, "grad_norm": 1.714684247970581, "learning_rate": 0.00031802779507285464, "loss": 2.4593515396118164, "step": 15169 }, { "epoch": 1.7651850127996278, "grad_norm": 0.8969630002975464, "learning_rate": 0.0003178078767211475, "loss": 2.905360221862793, "step": 15170 }, { "epoch": 1.7653013730509657, "grad_norm": 0.7539158463478088, "learning_rate": 0.00031758790171671636, "loss": 1.3353582620620728, "step": 15171 }, { "epoch": 1.765417733302304, "grad_norm": 1.0432237386703491, "learning_rate": 0.00031736787024334796, "loss": 1.9314942359924316, "step": 15172 }, { "epoch": 1.765534093553642, "grad_norm": 1.0257576704025269, "learning_rate": 0.00031714778248487615, "loss": 2.092928409576416, "step": 15173 }, { "epoch": 1.7656504538049802, "grad_norm": 1.0870082378387451, "learning_rate": 0.0003169276386251817, "loss": 2.5574212074279785, "step": 15174 }, { "epoch": 1.7657668140563185, "grad_norm": 1.0800538063049316, "learning_rate": 0.0003167074388481924, "loss": 2.7549266815185547, "step": 15175 }, { "epoch": 1.7658831743076564, "grad_norm": 1.1931898593902588, "learning_rate": 0.00031648718333788264, "loss": 1.5490875244140625, "step": 15176 }, { "epoch": 1.7659995345589947, "grad_norm": 1.0068142414093018, "learning_rate": 0.00031626687227827353, "loss": 1.8124198913574219, "step": 15177 }, { "epoch": 1.7661158948103328, "grad_norm": 1.798052430152893, "learning_rate": 0.00031604650585343255, "loss": 1.584320068359375, "step": 15178 }, { "epoch": 1.766232255061671, "grad_norm": 0.973971426486969, "learning_rate": 0.00031582608424747316, "loss": 1.1775596141815186, "step": 15179 }, { "epoch": 1.7663486153130092, "grad_norm": 1.410889983177185, "learning_rate": 0.0003156056076445554, "loss": 1.6759326457977295, "step": 15180 }, { "epoch": 1.7664649755643471, "grad_norm": 0.727520763874054, "learning_rate": 0.0003153850762288848, "loss": 1.727464199066162, "step": 15181 }, { "epoch": 1.7665813358156854, "grad_norm": 0.944648802280426, "learning_rate": 0.0003151644901847131, "loss": 2.151249408721924, "step": 15182 }, { "epoch": 1.7666976960670235, "grad_norm": 0.942806601524353, "learning_rate": 0.0003149438496963373, "loss": 1.8046174049377441, "step": 15183 }, { "epoch": 1.7668140563183616, "grad_norm": 1.5421457290649414, "learning_rate": 0.00031472315494810024, "loss": 1.8944016695022583, "step": 15184 }, { "epoch": 1.7669304165696997, "grad_norm": 0.9541491270065308, "learning_rate": 0.00031450240612438974, "loss": 1.5235528945922852, "step": 15185 }, { "epoch": 1.7670467768210378, "grad_norm": 2.127525568008423, "learning_rate": 0.0003142816034096391, "loss": 2.4026451110839844, "step": 15186 }, { "epoch": 1.7671631370723762, "grad_norm": 1.1434357166290283, "learning_rate": 0.00031406074698832646, "loss": 1.802704095840454, "step": 15187 }, { "epoch": 1.7672794973237143, "grad_norm": 0.601876974105835, "learning_rate": 0.0003138398370449749, "loss": 1.2706924676895142, "step": 15188 }, { "epoch": 1.7673958575750524, "grad_norm": 2.5357742309570312, "learning_rate": 0.00031361887376415233, "loss": 2.5257744789123535, "step": 15189 }, { "epoch": 1.7675122178263905, "grad_norm": 1.3170371055603027, "learning_rate": 0.00031339785733047087, "loss": 1.8151664733886719, "step": 15190 }, { "epoch": 1.7676285780777286, "grad_norm": 0.6655682921409607, "learning_rate": 0.0003131767879285875, "loss": 0.9576240181922913, "step": 15191 }, { "epoch": 1.7677449383290669, "grad_norm": 1.0541692972183228, "learning_rate": 0.0003129556657432031, "loss": 1.685377836227417, "step": 15192 }, { "epoch": 1.767861298580405, "grad_norm": 1.4074323177337646, "learning_rate": 0.0003127344909590629, "loss": 2.615676164627075, "step": 15193 }, { "epoch": 1.767977658831743, "grad_norm": 0.8345040678977966, "learning_rate": 0.0003125132637609558, "loss": 1.9229350090026855, "step": 15194 }, { "epoch": 1.7680940190830812, "grad_norm": 1.0711073875427246, "learning_rate": 0.0003122919843337148, "loss": 1.737837791442871, "step": 15195 }, { "epoch": 1.7682103793344193, "grad_norm": 2.564289093017578, "learning_rate": 0.00031207065286221634, "loss": 2.3580946922302246, "step": 15196 }, { "epoch": 1.7683267395857576, "grad_norm": 0.972680926322937, "learning_rate": 0.0003118492695313805, "loss": 1.8884069919586182, "step": 15197 }, { "epoch": 1.7684430998370957, "grad_norm": 1.0184813737869263, "learning_rate": 0.00031162783452617037, "loss": 1.6802291870117188, "step": 15198 }, { "epoch": 1.7685594600884338, "grad_norm": 1.2768391370773315, "learning_rate": 0.00031140634803159255, "loss": 2.3018417358398438, "step": 15199 }, { "epoch": 1.7686758203397719, "grad_norm": 1.2389808893203735, "learning_rate": 0.0003111848102326966, "loss": 1.6903585195541382, "step": 15200 }, { "epoch": 1.76879218059111, "grad_norm": 1.4892889261245728, "learning_rate": 0.0003109632213145747, "loss": 2.066865921020508, "step": 15201 }, { "epoch": 1.7689085408424483, "grad_norm": 2.22945237159729, "learning_rate": 0.0003107415814623621, "loss": 1.9278509616851807, "step": 15202 }, { "epoch": 1.7690249010937864, "grad_norm": 0.6695588827133179, "learning_rate": 0.0003105198908612362, "loss": 1.3614531755447388, "step": 15203 }, { "epoch": 1.7691412613451245, "grad_norm": 1.0149762630462646, "learning_rate": 0.0003102981496964173, "loss": 1.6109246015548706, "step": 15204 }, { "epoch": 1.7692576215964626, "grad_norm": 0.9527615904808044, "learning_rate": 0.0003100763581531675, "loss": 1.5513131618499756, "step": 15205 }, { "epoch": 1.7693739818478007, "grad_norm": 0.9532979726791382, "learning_rate": 0.0003098545164167911, "loss": 1.5349540710449219, "step": 15206 }, { "epoch": 1.769490342099139, "grad_norm": 0.7851508855819702, "learning_rate": 0.0003096326246726345, "loss": 1.4967025518417358, "step": 15207 }, { "epoch": 1.7696067023504771, "grad_norm": 0.5961385369300842, "learning_rate": 0.0003094106831060858, "loss": 0.9350380897521973, "step": 15208 }, { "epoch": 1.7697230626018152, "grad_norm": 0.9962628483772278, "learning_rate": 0.0003091886919025746, "loss": 1.4904053211212158, "step": 15209 }, { "epoch": 1.7698394228531533, "grad_norm": 1.1647082567214966, "learning_rate": 0.0003089666512475722, "loss": 2.308012008666992, "step": 15210 }, { "epoch": 1.7699557831044914, "grad_norm": 0.9963620901107788, "learning_rate": 0.00030874456132659096, "loss": 1.6554864645004272, "step": 15211 }, { "epoch": 1.7700721433558297, "grad_norm": 0.2533201277256012, "learning_rate": 0.00030852242232518463, "loss": 0.8770856857299805, "step": 15212 }, { "epoch": 1.7701885036071678, "grad_norm": 1.0149366855621338, "learning_rate": 0.0003083002344289478, "loss": 1.8042778968811035, "step": 15213 }, { "epoch": 1.770304863858506, "grad_norm": 0.8190166354179382, "learning_rate": 0.00030807799782351595, "loss": 1.6696527004241943, "step": 15214 }, { "epoch": 1.770421224109844, "grad_norm": 0.8895972371101379, "learning_rate": 0.00030785571269456536, "loss": 1.744138479232788, "step": 15215 }, { "epoch": 1.7705375843611821, "grad_norm": 1.137963891029358, "learning_rate": 0.00030763337922781275, "loss": 1.9912998676300049, "step": 15216 }, { "epoch": 1.7706539446125205, "grad_norm": 1.4146538972854614, "learning_rate": 0.00030741099760901524, "loss": 2.5820064544677734, "step": 15217 }, { "epoch": 1.7707703048638586, "grad_norm": 0.7592745423316956, "learning_rate": 0.0003071885680239702, "loss": 1.663602590560913, "step": 15218 }, { "epoch": 1.7708866651151967, "grad_norm": 1.0791821479797363, "learning_rate": 0.00030696609065851497, "loss": 1.947438359260559, "step": 15219 }, { "epoch": 1.7710030253665348, "grad_norm": 0.7331252098083496, "learning_rate": 0.00030674356569852714, "loss": 1.758797526359558, "step": 15220 }, { "epoch": 1.7711193856178729, "grad_norm": 1.1096068620681763, "learning_rate": 0.00030652099332992354, "loss": 1.5945751667022705, "step": 15221 }, { "epoch": 1.7712357458692112, "grad_norm": 1.077453851699829, "learning_rate": 0.00030629837373866115, "loss": 2.8364295959472656, "step": 15222 }, { "epoch": 1.7713521061205493, "grad_norm": 0.8884977698326111, "learning_rate": 0.000306075707110736, "loss": 1.13718581199646, "step": 15223 }, { "epoch": 1.7714684663718874, "grad_norm": 1.017416000366211, "learning_rate": 0.0003058529936321836, "loss": 1.2468867301940918, "step": 15224 }, { "epoch": 1.7715848266232255, "grad_norm": 1.1615177392959595, "learning_rate": 0.0003056302334890786, "loss": 2.6382639408111572, "step": 15225 }, { "epoch": 1.7717011868745636, "grad_norm": 0.9068187475204468, "learning_rate": 0.00030540742686753464, "loss": 1.90407395362854, "step": 15226 }, { "epoch": 1.771817547125902, "grad_norm": 0.7910125255584717, "learning_rate": 0.0003051845739537041, "loss": 1.6232600212097168, "step": 15227 }, { "epoch": 1.7719339073772398, "grad_norm": 0.8088010549545288, "learning_rate": 0.0003049616749337781, "loss": 1.9427108764648438, "step": 15228 }, { "epoch": 1.772050267628578, "grad_norm": 1.030890703201294, "learning_rate": 0.00030473872999398636, "loss": 1.9814784526824951, "step": 15229 }, { "epoch": 1.7721666278799162, "grad_norm": 0.6146182417869568, "learning_rate": 0.00030451573932059676, "loss": 1.330041527748108, "step": 15230 }, { "epoch": 1.7722829881312543, "grad_norm": 1.1048595905303955, "learning_rate": 0.00030429270309991564, "loss": 2.152658462524414, "step": 15231 }, { "epoch": 1.7723993483825926, "grad_norm": 1.2012147903442383, "learning_rate": 0.0003040696215182872, "loss": 1.9298207759857178, "step": 15232 }, { "epoch": 1.7725157086339305, "grad_norm": 1.0613692998886108, "learning_rate": 0.0003038464947620937, "loss": 2.6677744388580322, "step": 15233 }, { "epoch": 1.7726320688852688, "grad_norm": 1.667556643486023, "learning_rate": 0.00030362332301775495, "loss": 2.4303958415985107, "step": 15234 }, { "epoch": 1.772748429136607, "grad_norm": 0.9133595824241638, "learning_rate": 0.00030340010647172857, "loss": 1.788609266281128, "step": 15235 }, { "epoch": 1.772864789387945, "grad_norm": 1.1496522426605225, "learning_rate": 0.00030317684531050936, "loss": 1.705109715461731, "step": 15236 }, { "epoch": 1.7729811496392833, "grad_norm": 1.5746690034866333, "learning_rate": 0.00030295353972062964, "loss": 2.139272689819336, "step": 15237 }, { "epoch": 1.7730975098906212, "grad_norm": 1.266884684562683, "learning_rate": 0.0003027301898886586, "loss": 2.3498902320861816, "step": 15238 }, { "epoch": 1.7732138701419595, "grad_norm": 1.0620825290679932, "learning_rate": 0.00030250679600120264, "loss": 1.842814564704895, "step": 15239 }, { "epoch": 1.7733302303932976, "grad_norm": 1.184246301651001, "learning_rate": 0.0003022833582449048, "loss": 2.632413387298584, "step": 15240 }, { "epoch": 1.7734465906446357, "grad_norm": 1.525219440460205, "learning_rate": 0.0003020598768064449, "loss": 2.462350368499756, "step": 15241 }, { "epoch": 1.773562950895974, "grad_norm": 1.3080599308013916, "learning_rate": 0.00030183635187253923, "loss": 1.9126006364822388, "step": 15242 }, { "epoch": 1.773679311147312, "grad_norm": 1.3311940431594849, "learning_rate": 0.00030161278362994016, "loss": 2.066474437713623, "step": 15243 }, { "epoch": 1.7737956713986502, "grad_norm": 1.054658055305481, "learning_rate": 0.0003013891722654366, "loss": 1.8128063678741455, "step": 15244 }, { "epoch": 1.7739120316499883, "grad_norm": 0.8110102415084839, "learning_rate": 0.0003011655179658534, "loss": 1.592420220375061, "step": 15245 }, { "epoch": 1.7740283919013264, "grad_norm": 0.9404286742210388, "learning_rate": 0.0003009418209180512, "loss": 2.1415669918060303, "step": 15246 }, { "epoch": 1.7741447521526648, "grad_norm": 1.4407650232315063, "learning_rate": 0.0003007180813089263, "loss": 2.218907117843628, "step": 15247 }, { "epoch": 1.7742611124040026, "grad_norm": 1.1252026557922363, "learning_rate": 0.00030049429932541076, "loss": 1.5895092487335205, "step": 15248 }, { "epoch": 1.774377472655341, "grad_norm": 0.7263443470001221, "learning_rate": 0.00030027047515447186, "loss": 2.166172981262207, "step": 15249 }, { "epoch": 1.774493832906679, "grad_norm": 0.8786507248878479, "learning_rate": 0.0003000466089831122, "loss": 2.473581552505493, "step": 15250 }, { "epoch": 1.7746101931580172, "grad_norm": 0.7667571306228638, "learning_rate": 0.0002998227009983695, "loss": 1.4450974464416504, "step": 15251 }, { "epoch": 1.7747265534093555, "grad_norm": 1.144102931022644, "learning_rate": 0.00029959875138731634, "loss": 1.876476526260376, "step": 15252 }, { "epoch": 1.7748429136606934, "grad_norm": 1.0413720607757568, "learning_rate": 0.0002993747603370601, "loss": 1.8975892066955566, "step": 15253 }, { "epoch": 1.7749592739120317, "grad_norm": 1.2438793182373047, "learning_rate": 0.000299150728034743, "loss": 2.551159143447876, "step": 15254 }, { "epoch": 1.7750756341633698, "grad_norm": 1.0485937595367432, "learning_rate": 0.0002989266546675412, "loss": 2.288975715637207, "step": 15255 }, { "epoch": 1.7751919944147079, "grad_norm": 0.8855673670768738, "learning_rate": 0.00029870254042266563, "loss": 1.2170498371124268, "step": 15256 }, { "epoch": 1.7753083546660462, "grad_norm": 1.2514361143112183, "learning_rate": 0.0002984783854873614, "loss": 1.9444968700408936, "step": 15257 }, { "epoch": 1.775424714917384, "grad_norm": 1.1549720764160156, "learning_rate": 0.0002982541900489073, "loss": 1.9951618909835815, "step": 15258 }, { "epoch": 1.7755410751687224, "grad_norm": 0.7922121286392212, "learning_rate": 0.00029802995429461615, "loss": 1.6644704341888428, "step": 15259 }, { "epoch": 1.7756574354200605, "grad_norm": 0.7465473413467407, "learning_rate": 0.0002978056784118344, "loss": 1.4876304864883423, "step": 15260 }, { "epoch": 1.7757737956713986, "grad_norm": 0.8557562232017517, "learning_rate": 0.00029758136258794214, "loss": 1.8278778791427612, "step": 15261 }, { "epoch": 1.775890155922737, "grad_norm": 0.7884441018104553, "learning_rate": 0.0002973570070103526, "loss": 2.1961448192596436, "step": 15262 }, { "epoch": 1.7760065161740748, "grad_norm": 1.0454469919204712, "learning_rate": 0.00029713261186651235, "loss": 1.6680210828781128, "step": 15263 }, { "epoch": 1.776122876425413, "grad_norm": 1.1014251708984375, "learning_rate": 0.0002969081773439012, "loss": 1.7885053157806396, "step": 15264 }, { "epoch": 1.7762392366767512, "grad_norm": 1.237493634223938, "learning_rate": 0.0002966837036300315, "loss": 2.048703908920288, "step": 15265 }, { "epoch": 1.7763555969280893, "grad_norm": 0.9276863932609558, "learning_rate": 0.00029645919091244867, "loss": 1.8591976165771484, "step": 15266 }, { "epoch": 1.7764719571794276, "grad_norm": 0.7128316760063171, "learning_rate": 0.0002962346393787305, "loss": 1.620678186416626, "step": 15267 }, { "epoch": 1.7765883174307655, "grad_norm": 0.925899863243103, "learning_rate": 0.0002960100492164873, "loss": 1.2477927207946777, "step": 15268 }, { "epoch": 1.7767046776821038, "grad_norm": 1.168434500694275, "learning_rate": 0.00029578542061336165, "loss": 1.5372846126556396, "step": 15269 }, { "epoch": 1.776821037933442, "grad_norm": 1.0634974241256714, "learning_rate": 0.0002955607537570283, "loss": 2.254733085632324, "step": 15270 }, { "epoch": 1.77693739818478, "grad_norm": 0.9304736852645874, "learning_rate": 0.0002953360488351938, "loss": 2.057082176208496, "step": 15271 }, { "epoch": 1.7770537584361183, "grad_norm": 0.7178184986114502, "learning_rate": 0.00029511130603559674, "loss": 1.735202670097351, "step": 15272 }, { "epoch": 1.7771701186874562, "grad_norm": 1.0205392837524414, "learning_rate": 0.00029488652554600717, "loss": 1.3405499458312988, "step": 15273 }, { "epoch": 1.7772864789387945, "grad_norm": 0.8042083382606506, "learning_rate": 0.00029466170755422646, "loss": 1.6021373271942139, "step": 15274 }, { "epoch": 1.7774028391901326, "grad_norm": 1.3128007650375366, "learning_rate": 0.00029443685224808783, "loss": 2.155064821243286, "step": 15275 }, { "epoch": 1.7775191994414707, "grad_norm": 0.3346087634563446, "learning_rate": 0.0002942119598154552, "loss": 0.9535661935806274, "step": 15276 }, { "epoch": 1.777635559692809, "grad_norm": 1.454776406288147, "learning_rate": 0.00029398703044422375, "loss": 1.5472755432128906, "step": 15277 }, { "epoch": 1.777751919944147, "grad_norm": 0.8711543679237366, "learning_rate": 0.0002937620643223195, "loss": 1.3162070512771606, "step": 15278 }, { "epoch": 1.7778682801954853, "grad_norm": 1.099839687347412, "learning_rate": 0.00029353706163769903, "loss": 1.6065735816955566, "step": 15279 }, { "epoch": 1.7779846404468234, "grad_norm": 1.190102219581604, "learning_rate": 0.0002933120225783497, "loss": 1.693465232849121, "step": 15280 }, { "epoch": 1.7781010006981615, "grad_norm": 0.8581544756889343, "learning_rate": 0.00029308694733228897, "loss": 2.5277538299560547, "step": 15281 }, { "epoch": 1.7782173609494998, "grad_norm": 1.0928412675857544, "learning_rate": 0.00029286183608756485, "loss": 2.022608995437622, "step": 15282 }, { "epoch": 1.7783337212008377, "grad_norm": 1.019352674484253, "learning_rate": 0.0002926366890322552, "loss": 1.9487357139587402, "step": 15283 }, { "epoch": 1.778450081452176, "grad_norm": 0.37316250801086426, "learning_rate": 0.0002924115063544679, "loss": 1.0379770994186401, "step": 15284 }, { "epoch": 1.778566441703514, "grad_norm": 0.7834427356719971, "learning_rate": 0.00029218628824234064, "loss": 1.4397536516189575, "step": 15285 }, { "epoch": 1.7786828019548522, "grad_norm": 0.7310730218887329, "learning_rate": 0.00029196103488404035, "loss": 1.2554218769073486, "step": 15286 }, { "epoch": 1.7787991622061905, "grad_norm": 0.9441866874694824, "learning_rate": 0.00029173574646776396, "loss": 1.7130786180496216, "step": 15287 }, { "epoch": 1.7789155224575284, "grad_norm": 1.1244133710861206, "learning_rate": 0.00029151042318173735, "loss": 2.694350242614746, "step": 15288 }, { "epoch": 1.7790318827088667, "grad_norm": 1.2765915393829346, "learning_rate": 0.0002912850652142156, "loss": 1.889662504196167, "step": 15289 }, { "epoch": 1.7791482429602048, "grad_norm": 1.1693216562271118, "learning_rate": 0.00029105967275348284, "loss": 1.9159587621688843, "step": 15290 }, { "epoch": 1.779264603211543, "grad_norm": 0.7538836002349854, "learning_rate": 0.0002908342459878519, "loss": 1.5604238510131836, "step": 15291 }, { "epoch": 1.7793809634628812, "grad_norm": 0.9180962443351746, "learning_rate": 0.00029060878510566427, "loss": 1.6216673851013184, "step": 15292 }, { "epoch": 1.779497323714219, "grad_norm": 2.032989025115967, "learning_rate": 0.0002903832902952901, "loss": 1.387790322303772, "step": 15293 }, { "epoch": 1.7796136839655574, "grad_norm": 0.9997527599334717, "learning_rate": 0.0002901577617451277, "loss": 1.7763984203338623, "step": 15294 }, { "epoch": 1.7797300442168955, "grad_norm": 1.6090563535690308, "learning_rate": 0.00028993219964360375, "loss": 2.2530407905578613, "step": 15295 }, { "epoch": 1.7798464044682336, "grad_norm": 0.9761672019958496, "learning_rate": 0.0002897066041791728, "loss": 1.698654055595398, "step": 15296 }, { "epoch": 1.779962764719572, "grad_norm": 1.064993143081665, "learning_rate": 0.00028948097554031747, "loss": 1.884228229522705, "step": 15297 }, { "epoch": 1.7800791249709098, "grad_norm": 1.3382861614227295, "learning_rate": 0.0002892553139155478, "loss": 2.310004711151123, "step": 15298 }, { "epoch": 1.7801954852222481, "grad_norm": 1.411401391029358, "learning_rate": 0.0002890296194934015, "loss": 2.6717758178710938, "step": 15299 }, { "epoch": 1.7803118454735862, "grad_norm": 0.9805833101272583, "learning_rate": 0.0002888038924624439, "loss": 2.0613441467285156, "step": 15300 }, { "epoch": 1.7804282057249243, "grad_norm": 1.4431040287017822, "learning_rate": 0.0002885781330112675, "loss": 2.0317904949188232, "step": 15301 }, { "epoch": 1.7805445659762626, "grad_norm": 0.6786060929298401, "learning_rate": 0.0002883523413284916, "loss": 1.79506516456604, "step": 15302 }, { "epoch": 1.7806609262276005, "grad_norm": 1.1216250658035278, "learning_rate": 0.00028812651760276267, "loss": 2.1391279697418213, "step": 15303 }, { "epoch": 1.7807772864789388, "grad_norm": 1.32095205783844, "learning_rate": 0.00028790066202275395, "loss": 2.197237014770508, "step": 15304 }, { "epoch": 1.780893646730277, "grad_norm": 0.3771529793739319, "learning_rate": 0.0002876747747771653, "loss": 1.0972059965133667, "step": 15305 }, { "epoch": 1.781010006981615, "grad_norm": 1.1930919885635376, "learning_rate": 0.0002874488560547228, "loss": 2.9273223876953125, "step": 15306 }, { "epoch": 1.7811263672329534, "grad_norm": 2.0540788173675537, "learning_rate": 0.0002872229060441793, "loss": 2.1203103065490723, "step": 15307 }, { "epoch": 1.7812427274842912, "grad_norm": 1.1403112411499023, "learning_rate": 0.0002869969249343134, "loss": 1.7525479793548584, "step": 15308 }, { "epoch": 1.7813590877356296, "grad_norm": 0.6961652040481567, "learning_rate": 0.0002867709129139298, "loss": 1.4299262762069702, "step": 15309 }, { "epoch": 1.7814754479869677, "grad_norm": 1.095672607421875, "learning_rate": 0.0002865448701718589, "loss": 1.693528652191162, "step": 15310 }, { "epoch": 1.7815918082383058, "grad_norm": 0.6846993565559387, "learning_rate": 0.00028631879689695693, "loss": 1.6737208366394043, "step": 15311 }, { "epoch": 1.781708168489644, "grad_norm": 1.216225028038025, "learning_rate": 0.00028609269327810566, "loss": 1.6644717454910278, "step": 15312 }, { "epoch": 1.781824528740982, "grad_norm": 1.7463710308074951, "learning_rate": 0.00028586655950421213, "loss": 2.3410611152648926, "step": 15313 }, { "epoch": 1.7819408889923203, "grad_norm": 1.1331506967544556, "learning_rate": 0.0002856403957642086, "loss": 2.254939079284668, "step": 15314 }, { "epoch": 1.7820572492436584, "grad_norm": 1.4063897132873535, "learning_rate": 0.0002854142022470523, "loss": 1.9580061435699463, "step": 15315 }, { "epoch": 1.7821736094949965, "grad_norm": 0.8698957562446594, "learning_rate": 0.0002851879791417254, "loss": 1.9827114343643188, "step": 15316 }, { "epoch": 1.7822899697463348, "grad_norm": 1.7047019004821777, "learning_rate": 0.00028496172663723473, "loss": 1.942833662033081, "step": 15317 }, { "epoch": 1.7824063299976727, "grad_norm": 0.6383635401725769, "learning_rate": 0.0002847354449226118, "loss": 1.7006428241729736, "step": 15318 }, { "epoch": 1.782522690249011, "grad_norm": 0.9892846941947937, "learning_rate": 0.0002845091341869125, "loss": 2.030876636505127, "step": 15319 }, { "epoch": 1.782639050500349, "grad_norm": 0.8013651371002197, "learning_rate": 0.0002842827946192168, "loss": 2.1399078369140625, "step": 15320 }, { "epoch": 1.7827554107516872, "grad_norm": 0.7781171202659607, "learning_rate": 0.00028405642640862903, "loss": 1.5011197328567505, "step": 15321 }, { "epoch": 1.7828717710030255, "grad_norm": 0.6478556990623474, "learning_rate": 0.0002838300297442772, "loss": 1.8275458812713623, "step": 15322 }, { "epoch": 1.7829881312543634, "grad_norm": 1.1670072078704834, "learning_rate": 0.0002836036048153132, "loss": 2.0116126537323, "step": 15323 }, { "epoch": 1.7831044915057017, "grad_norm": 1.0372971296310425, "learning_rate": 0.0002833771518109124, "loss": 1.9570116996765137, "step": 15324 }, { "epoch": 1.7832208517570398, "grad_norm": 0.7186075448989868, "learning_rate": 0.000283150670920274, "loss": 1.3665075302124023, "step": 15325 }, { "epoch": 1.783337212008378, "grad_norm": 1.031633973121643, "learning_rate": 0.00028292416233262026, "loss": 1.2804441452026367, "step": 15326 }, { "epoch": 1.7834535722597162, "grad_norm": 0.9654483199119568, "learning_rate": 0.00028269762623719637, "loss": 1.6082806587219238, "step": 15327 }, { "epoch": 1.7835699325110541, "grad_norm": 1.1982797384262085, "learning_rate": 0.00028247106282327085, "loss": 2.412087917327881, "step": 15328 }, { "epoch": 1.7836862927623924, "grad_norm": 0.9570786356925964, "learning_rate": 0.00028224447228013483, "loss": 2.357041835784912, "step": 15329 }, { "epoch": 1.7838026530137305, "grad_norm": 1.1400645971298218, "learning_rate": 0.00028201785479710226, "loss": 1.2905765771865845, "step": 15330 }, { "epoch": 1.7839190132650686, "grad_norm": 0.9819655418395996, "learning_rate": 0.0002817912105635093, "loss": 2.403634548187256, "step": 15331 }, { "epoch": 1.784035373516407, "grad_norm": 0.7280257344245911, "learning_rate": 0.00028156453976871507, "loss": 1.5184401273727417, "step": 15332 }, { "epoch": 1.7841517337677448, "grad_norm": 1.0683499574661255, "learning_rate": 0.0002813378426021002, "loss": 2.1444735527038574, "step": 15333 }, { "epoch": 1.7842680940190832, "grad_norm": 0.8736588358879089, "learning_rate": 0.00028111111925306767, "loss": 1.7065638303756714, "step": 15334 }, { "epoch": 1.7843844542704212, "grad_norm": 1.2353569269180298, "learning_rate": 0.0002808843699110424, "loss": 2.040942668914795, "step": 15335 }, { "epoch": 1.7845008145217593, "grad_norm": 1.1972006559371948, "learning_rate": 0.00028065759476547083, "loss": 2.7138376235961914, "step": 15336 }, { "epoch": 1.7846171747730977, "grad_norm": 0.9679571390151978, "learning_rate": 0.0002804307940058211, "loss": 1.6441583633422852, "step": 15337 }, { "epoch": 1.7847335350244355, "grad_norm": 0.7877256274223328, "learning_rate": 0.0002802039678215826, "loss": 1.9175983667373657, "step": 15338 }, { "epoch": 1.7848498952757739, "grad_norm": 1.5470342636108398, "learning_rate": 0.0002799771164022664, "loss": 1.9601054191589355, "step": 15339 }, { "epoch": 1.784966255527112, "grad_norm": 1.1237069368362427, "learning_rate": 0.00027975023993740403, "loss": 1.5533303022384644, "step": 15340 }, { "epoch": 1.78508261577845, "grad_norm": 0.8529995083808899, "learning_rate": 0.0002795233386165484, "loss": 2.1408486366271973, "step": 15341 }, { "epoch": 1.7851989760297884, "grad_norm": 1.395214557647705, "learning_rate": 0.00027929641262927293, "loss": 2.200258731842041, "step": 15342 }, { "epoch": 1.7853153362811263, "grad_norm": 0.9892062544822693, "learning_rate": 0.00027906946216517175, "loss": 1.8650689125061035, "step": 15343 }, { "epoch": 1.7854316965324646, "grad_norm": 1.0276774168014526, "learning_rate": 0.0002788424874138595, "loss": 1.6965107917785645, "step": 15344 }, { "epoch": 1.7855480567838027, "grad_norm": 1.4554706811904907, "learning_rate": 0.0002786154885649711, "loss": 2.261678695678711, "step": 15345 }, { "epoch": 1.7856644170351408, "grad_norm": 0.8893299698829651, "learning_rate": 0.00027838846580816143, "loss": 2.052480459213257, "step": 15346 }, { "epoch": 1.785780777286479, "grad_norm": 0.7515279650688171, "learning_rate": 0.00027816141933310553, "loss": 1.8617960214614868, "step": 15347 }, { "epoch": 1.785897137537817, "grad_norm": 1.0384000539779663, "learning_rate": 0.00027793434932949813, "loss": 1.4067513942718506, "step": 15348 }, { "epoch": 1.7860134977891553, "grad_norm": 0.7933790683746338, "learning_rate": 0.00027770725598705374, "loss": 1.7360643148422241, "step": 15349 }, { "epoch": 1.7861298580404934, "grad_norm": 1.0240371227264404, "learning_rate": 0.0002774801394955063, "loss": 2.0843026638031006, "step": 15350 }, { "epoch": 1.7862462182918315, "grad_norm": 0.6370763182640076, "learning_rate": 0.00027725300004460916, "loss": 1.7128537893295288, "step": 15351 }, { "epoch": 1.7863625785431696, "grad_norm": 1.3465068340301514, "learning_rate": 0.0002770258378241348, "loss": 2.0663630962371826, "step": 15352 }, { "epoch": 1.7864789387945077, "grad_norm": 0.8348544836044312, "learning_rate": 0.00027679865302387464, "loss": 1.6311265230178833, "step": 15353 }, { "epoch": 1.786595299045846, "grad_norm": 1.345571517944336, "learning_rate": 0.00027657144583363903, "loss": 1.9752445220947266, "step": 15354 }, { "epoch": 1.7867116592971841, "grad_norm": 0.7662490606307983, "learning_rate": 0.00027634421644325707, "loss": 1.6013292074203491, "step": 15355 }, { "epoch": 1.7868280195485222, "grad_norm": 1.0194600820541382, "learning_rate": 0.00027611696504257637, "loss": 1.771392583847046, "step": 15356 }, { "epoch": 1.7869443797998603, "grad_norm": 1.530894160270691, "learning_rate": 0.00027588969182146305, "loss": 1.8633311986923218, "step": 15357 }, { "epoch": 1.7870607400511984, "grad_norm": 1.222036600112915, "learning_rate": 0.00027566239696980125, "loss": 1.9931585788726807, "step": 15358 }, { "epoch": 1.7871771003025367, "grad_norm": 1.4251099824905396, "learning_rate": 0.00027543508067749326, "loss": 1.671196460723877, "step": 15359 }, { "epoch": 1.7872934605538748, "grad_norm": 0.9620079398155212, "learning_rate": 0.00027520774313445927, "loss": 1.639323353767395, "step": 15360 }, { "epoch": 1.787409820805213, "grad_norm": 0.66109699010849, "learning_rate": 0.00027498038453063745, "loss": 1.2669481039047241, "step": 15361 }, { "epoch": 1.787526181056551, "grad_norm": 0.9755867719650269, "learning_rate": 0.0002747530050559831, "loss": 2.113193988800049, "step": 15362 }, { "epoch": 1.7876425413078891, "grad_norm": 1.1495286226272583, "learning_rate": 0.00027452560490046947, "loss": 2.155838966369629, "step": 15363 }, { "epoch": 1.7877589015592275, "grad_norm": 0.9366558194160461, "learning_rate": 0.0002742981842540868, "loss": 1.4720648527145386, "step": 15364 }, { "epoch": 1.7878752618105656, "grad_norm": 1.20638906955719, "learning_rate": 0.00027407074330684236, "loss": 2.022611141204834, "step": 15365 }, { "epoch": 1.7879916220619037, "grad_norm": 1.3761698007583618, "learning_rate": 0.0002738432822487607, "loss": 2.216193199157715, "step": 15366 }, { "epoch": 1.7881079823132418, "grad_norm": 1.5189622640609741, "learning_rate": 0.00027361580126988286, "loss": 1.685236930847168, "step": 15367 }, { "epoch": 1.7882243425645798, "grad_norm": 0.9805511236190796, "learning_rate": 0.00027338830056026673, "loss": 1.8739917278289795, "step": 15368 }, { "epoch": 1.7883407028159182, "grad_norm": 2.145040512084961, "learning_rate": 0.00027316078030998665, "loss": 1.4521039724349976, "step": 15369 }, { "epoch": 1.7884570630672563, "grad_norm": 1.2813166379928589, "learning_rate": 0.0002729332407091332, "loss": 1.891343355178833, "step": 15370 }, { "epoch": 1.7885734233185944, "grad_norm": 1.110685110092163, "learning_rate": 0.0002727056819478131, "loss": 1.9208874702453613, "step": 15371 }, { "epoch": 1.7886897835699325, "grad_norm": 1.1455501317977905, "learning_rate": 0.00027247810421614926, "loss": 1.2042721509933472, "step": 15372 }, { "epoch": 1.7888061438212706, "grad_norm": 0.7986026406288147, "learning_rate": 0.00027225050770428034, "loss": 1.2959134578704834, "step": 15373 }, { "epoch": 1.7889225040726089, "grad_norm": 1.8057020902633667, "learning_rate": 0.0002720228926023606, "loss": 1.7590694427490234, "step": 15374 }, { "epoch": 1.789038864323947, "grad_norm": 1.6155873537063599, "learning_rate": 0.00027179525910056, "loss": 1.2678251266479492, "step": 15375 }, { "epoch": 1.789155224575285, "grad_norm": 0.8964939713478088, "learning_rate": 0.0002715676073890637, "loss": 1.430018424987793, "step": 15376 }, { "epoch": 1.7892715848266232, "grad_norm": 0.7506616115570068, "learning_rate": 0.0002713399376580723, "loss": 2.0277771949768066, "step": 15377 }, { "epoch": 1.7893879450779613, "grad_norm": 0.9201450347900391, "learning_rate": 0.00027111225009780114, "loss": 1.7913485765457153, "step": 15378 }, { "epoch": 1.7895043053292996, "grad_norm": 1.0135698318481445, "learning_rate": 0.0002708845448984808, "loss": 1.8855079412460327, "step": 15379 }, { "epoch": 1.7896206655806377, "grad_norm": 1.332822322845459, "learning_rate": 0.0002706568222503564, "loss": 1.3280243873596191, "step": 15380 }, { "epoch": 1.7897370258319758, "grad_norm": 0.947843074798584, "learning_rate": 0.00027042908234368757, "loss": 2.3613219261169434, "step": 15381 }, { "epoch": 1.789853386083314, "grad_norm": 0.91883385181427, "learning_rate": 0.0002702013253687485, "loss": 1.8562167882919312, "step": 15382 }, { "epoch": 1.789969746334652, "grad_norm": 1.1174644231796265, "learning_rate": 0.0002699735515158277, "loss": 2.494486093521118, "step": 15383 }, { "epoch": 1.7900861065859903, "grad_norm": 1.5763418674468994, "learning_rate": 0.0002697457609752275, "loss": 2.1995229721069336, "step": 15384 }, { "epoch": 1.7902024668373284, "grad_norm": 0.956982433795929, "learning_rate": 0.00026951795393726456, "loss": 2.482820510864258, "step": 15385 }, { "epoch": 1.7903188270886665, "grad_norm": 1.4055449962615967, "learning_rate": 0.000269290130592269, "loss": 2.143801212310791, "step": 15386 }, { "epoch": 1.7904351873400046, "grad_norm": 1.10942804813385, "learning_rate": 0.00026906229113058457, "loss": 2.35068416595459, "step": 15387 }, { "epoch": 1.7905515475913427, "grad_norm": 0.9428390264511108, "learning_rate": 0.00026883443574256884, "loss": 1.3737897872924805, "step": 15388 }, { "epoch": 1.790667907842681, "grad_norm": 1.1509281396865845, "learning_rate": 0.0002686065646185923, "loss": 1.9636256694793701, "step": 15389 }, { "epoch": 1.7907842680940191, "grad_norm": 1.0272711515426636, "learning_rate": 0.00026837867794903876, "loss": 1.347364068031311, "step": 15390 }, { "epoch": 1.7909006283453572, "grad_norm": 1.0444992780685425, "learning_rate": 0.00026815077592430494, "loss": 1.8467998504638672, "step": 15391 }, { "epoch": 1.7910169885966953, "grad_norm": 1.0037870407104492, "learning_rate": 0.0002679228587348005, "loss": 1.8011702299118042, "step": 15392 }, { "epoch": 1.7911333488480334, "grad_norm": 1.1833068132400513, "learning_rate": 0.00026769492657094775, "loss": 2.059936046600342, "step": 15393 }, { "epoch": 1.7912497090993718, "grad_norm": 0.9578254222869873, "learning_rate": 0.00026746697962318135, "loss": 1.4559292793273926, "step": 15394 }, { "epoch": 1.7913660693507096, "grad_norm": 0.6812796592712402, "learning_rate": 0.00026723901808194844, "loss": 1.5517405271530151, "step": 15395 }, { "epoch": 1.791482429602048, "grad_norm": 1.0071302652359009, "learning_rate": 0.0002670110421377084, "loss": 1.8430521488189697, "step": 15396 }, { "epoch": 1.791598789853386, "grad_norm": 1.0512230396270752, "learning_rate": 0.00026678305198093267, "loss": 1.8694431781768799, "step": 15397 }, { "epoch": 1.7917151501047242, "grad_norm": 1.1005873680114746, "learning_rate": 0.0002665550478021043, "loss": 1.730664610862732, "step": 15398 }, { "epoch": 1.7918315103560625, "grad_norm": 0.6689738035202026, "learning_rate": 0.0002663270297917183, "loss": 1.4528424739837646, "step": 15399 }, { "epoch": 1.7919478706074004, "grad_norm": 0.6587752103805542, "learning_rate": 0.0002660989981402812, "loss": 1.637170433998108, "step": 15400 }, { "epoch": 1.7920642308587387, "grad_norm": 1.1120997667312622, "learning_rate": 0.00026587095303831095, "loss": 2.764862060546875, "step": 15401 }, { "epoch": 1.7921805911100768, "grad_norm": 0.7718153595924377, "learning_rate": 0.0002656428946763366, "loss": 1.1138904094696045, "step": 15402 }, { "epoch": 1.7922969513614149, "grad_norm": 1.1629011631011963, "learning_rate": 0.0002654148232448984, "loss": 1.6442618370056152, "step": 15403 }, { "epoch": 1.7924133116127532, "grad_norm": 0.5972492098808289, "learning_rate": 0.0002651867389345475, "loss": 1.3777555227279663, "step": 15404 }, { "epoch": 1.792529671864091, "grad_norm": 1.4518479108810425, "learning_rate": 0.0002649586419358457, "loss": 1.8213870525360107, "step": 15405 }, { "epoch": 1.7926460321154294, "grad_norm": 0.7867499589920044, "learning_rate": 0.0002647305324393656, "loss": 1.501399278640747, "step": 15406 }, { "epoch": 1.7927623923667675, "grad_norm": 1.101355791091919, "learning_rate": 0.0002645024106356901, "loss": 1.8213284015655518, "step": 15407 }, { "epoch": 1.7928787526181056, "grad_norm": 0.7674575448036194, "learning_rate": 0.00026427427671541246, "loss": 1.4404957294464111, "step": 15408 }, { "epoch": 1.792995112869444, "grad_norm": 0.7225074172019958, "learning_rate": 0.000264046130869136, "loss": 1.7253817319869995, "step": 15409 }, { "epoch": 1.7931114731207818, "grad_norm": 0.9464331269264221, "learning_rate": 0.00026381797328747396, "loss": 1.9450132846832275, "step": 15410 }, { "epoch": 1.79322783337212, "grad_norm": 0.940423309803009, "learning_rate": 0.00026358980416104954, "loss": 1.6929242610931396, "step": 15411 }, { "epoch": 1.7933441936234582, "grad_norm": 1.0718938112258911, "learning_rate": 0.0002633616236804954, "loss": 1.5564638376235962, "step": 15412 }, { "epoch": 1.7934605538747963, "grad_norm": 1.1476916074752808, "learning_rate": 0.00026313343203645395, "loss": 1.8889532089233398, "step": 15413 }, { "epoch": 1.7935769141261346, "grad_norm": 0.9797337055206299, "learning_rate": 0.0002629052294195766, "loss": 1.7860018014907837, "step": 15414 }, { "epoch": 1.7936932743774725, "grad_norm": 1.4257687330245972, "learning_rate": 0.00026267701602052416, "loss": 2.3006234169006348, "step": 15415 }, { "epoch": 1.7938096346288108, "grad_norm": 1.3110928535461426, "learning_rate": 0.0002624487920299664, "loss": 2.121829032897949, "step": 15416 }, { "epoch": 1.793925994880149, "grad_norm": 0.8295475840568542, "learning_rate": 0.00026222055763858187, "loss": 1.7180899381637573, "step": 15417 }, { "epoch": 1.794042355131487, "grad_norm": 0.7191188931465149, "learning_rate": 0.00026199231303705787, "loss": 1.0089454650878906, "step": 15418 }, { "epoch": 1.7941587153828253, "grad_norm": 0.24639098346233368, "learning_rate": 0.00026176405841609027, "loss": 0.8731096386909485, "step": 15419 }, { "epoch": 1.7942750756341632, "grad_norm": 0.820279598236084, "learning_rate": 0.0002615357939663831, "loss": 1.511976718902588, "step": 15420 }, { "epoch": 1.7943914358855015, "grad_norm": 1.8790761232376099, "learning_rate": 0.00026130751987864893, "loss": 2.1254053115844727, "step": 15421 }, { "epoch": 1.7945077961368396, "grad_norm": 1.1058331727981567, "learning_rate": 0.0002610792363436081, "loss": 1.5710675716400146, "step": 15422 }, { "epoch": 1.7946241563881777, "grad_norm": 0.8526643514633179, "learning_rate": 0.000260850943551989, "loss": 1.8678096532821655, "step": 15423 }, { "epoch": 1.794740516639516, "grad_norm": 1.7737157344818115, "learning_rate": 0.0002606226416945277, "loss": 1.9305970668792725, "step": 15424 }, { "epoch": 1.794856876890854, "grad_norm": 0.6700982451438904, "learning_rate": 0.00026039433096196785, "loss": 1.3972933292388916, "step": 15425 }, { "epoch": 1.7949732371421923, "grad_norm": 2.047065258026123, "learning_rate": 0.0002601660115450606, "loss": 1.8460993766784668, "step": 15426 }, { "epoch": 1.7950895973935304, "grad_norm": 1.1071573495864868, "learning_rate": 0.0002599376836345642, "loss": 2.0776572227478027, "step": 15427 }, { "epoch": 1.7952059576448685, "grad_norm": 1.2024236917495728, "learning_rate": 0.00025970934742124396, "loss": 1.879361867904663, "step": 15428 }, { "epoch": 1.7953223178962068, "grad_norm": 0.7024199962615967, "learning_rate": 0.0002594810030958725, "loss": 2.0175623893737793, "step": 15429 }, { "epoch": 1.7954386781475447, "grad_norm": 1.2536365985870361, "learning_rate": 0.00025925265084922864, "loss": 1.60711669921875, "step": 15430 }, { "epoch": 1.795555038398883, "grad_norm": 0.9545138478279114, "learning_rate": 0.00025902429087209844, "loss": 1.911805510520935, "step": 15431 }, { "epoch": 1.795671398650221, "grad_norm": 1.2748503684997559, "learning_rate": 0.0002587959233552739, "loss": 2.174647808074951, "step": 15432 }, { "epoch": 1.7957877589015592, "grad_norm": 0.8890212774276733, "learning_rate": 0.0002585675484895536, "loss": 2.0816266536712646, "step": 15433 }, { "epoch": 1.7959041191528975, "grad_norm": 1.9314507246017456, "learning_rate": 0.00025833916646574224, "loss": 2.5868499279022217, "step": 15434 }, { "epoch": 1.7960204794042354, "grad_norm": 1.212550163269043, "learning_rate": 0.00025811077747465034, "loss": 1.7190968990325928, "step": 15435 }, { "epoch": 1.7961368396555737, "grad_norm": 0.9387694001197815, "learning_rate": 0.00025788238170709445, "loss": 1.9425393342971802, "step": 15436 }, { "epoch": 1.7962531999069118, "grad_norm": 0.45778509974479675, "learning_rate": 0.0002576539793538966, "loss": 0.9266271591186523, "step": 15437 }, { "epoch": 1.79636956015825, "grad_norm": 1.145443081855774, "learning_rate": 0.00025742557060588446, "loss": 1.8452776670455933, "step": 15438 }, { "epoch": 1.7964859204095882, "grad_norm": 1.8149152994155884, "learning_rate": 0.000257197155653891, "loss": 1.7964086532592773, "step": 15439 }, { "epoch": 1.796602280660926, "grad_norm": 1.3418545722961426, "learning_rate": 0.0002569687346887544, "loss": 1.6933209896087646, "step": 15440 }, { "epoch": 1.7967186409122644, "grad_norm": 0.967784583568573, "learning_rate": 0.00025674030790131776, "loss": 1.4228441715240479, "step": 15441 }, { "epoch": 1.7968350011636025, "grad_norm": 1.2405040264129639, "learning_rate": 0.00025651187548242897, "loss": 1.6945760250091553, "step": 15442 }, { "epoch": 1.7969513614149406, "grad_norm": 0.995795488357544, "learning_rate": 0.00025628343762294103, "loss": 1.4424505233764648, "step": 15443 }, { "epoch": 1.797067721666279, "grad_norm": 0.7064048647880554, "learning_rate": 0.0002560549945137111, "loss": 1.6288173198699951, "step": 15444 }, { "epoch": 1.7971840819176168, "grad_norm": 1.2946548461914062, "learning_rate": 0.0002558265463456009, "loss": 2.1281120777130127, "step": 15445 }, { "epoch": 1.7973004421689551, "grad_norm": 1.2040232419967651, "learning_rate": 0.0002555980933094762, "loss": 1.605994701385498, "step": 15446 }, { "epoch": 1.7974168024202932, "grad_norm": 1.6547136306762695, "learning_rate": 0.0002553696355962071, "loss": 1.5087608098983765, "step": 15447 }, { "epoch": 1.7975331626716313, "grad_norm": 1.189225435256958, "learning_rate": 0.0002551411733966674, "loss": 2.1170709133148193, "step": 15448 }, { "epoch": 1.7976495229229696, "grad_norm": 0.8871050477027893, "learning_rate": 0.0002549127069017346, "loss": 1.4289135932922363, "step": 15449 }, { "epoch": 1.7977658831743075, "grad_norm": 1.6385374069213867, "learning_rate": 0.0002546842363022902, "loss": 1.607601523399353, "step": 15450 }, { "epoch": 1.7978822434256458, "grad_norm": 0.7868629693984985, "learning_rate": 0.00025445576178921857, "loss": 1.4737218618392944, "step": 15451 }, { "epoch": 1.797998603676984, "grad_norm": 0.4993613362312317, "learning_rate": 0.0002542272835534078, "loss": 1.342895746231079, "step": 15452 }, { "epoch": 1.798114963928322, "grad_norm": 1.1068614721298218, "learning_rate": 0.0002539988017857488, "loss": 2.5983335971832275, "step": 15453 }, { "epoch": 1.7982313241796604, "grad_norm": 1.4203613996505737, "learning_rate": 0.00025377031667713553, "loss": 2.8060436248779297, "step": 15454 }, { "epoch": 1.7983476844309982, "grad_norm": 0.8623751401901245, "learning_rate": 0.00025354182841846466, "loss": 1.415804386138916, "step": 15455 }, { "epoch": 1.7984640446823366, "grad_norm": 0.7644100189208984, "learning_rate": 0.0002533133372006358, "loss": 2.0978684425354004, "step": 15456 }, { "epoch": 1.7985804049336747, "grad_norm": 0.825631320476532, "learning_rate": 0.00025308484321455075, "loss": 2.231565475463867, "step": 15457 }, { "epoch": 1.7986967651850128, "grad_norm": 0.928443193435669, "learning_rate": 0.0002528563466511135, "loss": 1.8991745710372925, "step": 15458 }, { "epoch": 1.798813125436351, "grad_norm": 1.1616264581680298, "learning_rate": 0.0002526278477012306, "loss": 2.029055595397949, "step": 15459 }, { "epoch": 1.798929485687689, "grad_norm": 0.9181994199752808, "learning_rate": 0.00025239934655581025, "loss": 1.4950684309005737, "step": 15460 }, { "epoch": 1.7990458459390273, "grad_norm": 2.1096692085266113, "learning_rate": 0.0002521708434057627, "loss": 2.4460067749023438, "step": 15461 }, { "epoch": 1.7991622061903654, "grad_norm": 0.7410248517990112, "learning_rate": 0.0002519423384419996, "loss": 1.2414783239364624, "step": 15462 }, { "epoch": 1.7992785664417035, "grad_norm": 0.9647077918052673, "learning_rate": 0.0002517138318554345, "loss": 1.4934028387069702, "step": 15463 }, { "epoch": 1.7993949266930418, "grad_norm": 0.9039479494094849, "learning_rate": 0.00025148532383698205, "loss": 1.5355567932128906, "step": 15464 }, { "epoch": 1.7995112869443797, "grad_norm": 0.8341068625450134, "learning_rate": 0.0002512568145775581, "loss": 1.8402280807495117, "step": 15465 }, { "epoch": 1.799627647195718, "grad_norm": 0.9642468094825745, "learning_rate": 0.00025102830426807957, "loss": 1.715658187866211, "step": 15466 }, { "epoch": 1.799744007447056, "grad_norm": 0.8107451796531677, "learning_rate": 0.0002507997930994643, "loss": 1.487099528312683, "step": 15467 }, { "epoch": 1.7998603676983942, "grad_norm": 1.19559645652771, "learning_rate": 0.00025057128126263086, "loss": 2.085203170776367, "step": 15468 }, { "epoch": 1.7999767279497325, "grad_norm": 0.9188997149467468, "learning_rate": 0.00025034276894849827, "loss": 2.334239959716797, "step": 15469 }, { "epoch": 1.8000930882010704, "grad_norm": 1.5532817840576172, "learning_rate": 0.0002501142563479861, "loss": 2.1379661560058594, "step": 15470 }, { "epoch": 1.8002094484524087, "grad_norm": 1.2684967517852783, "learning_rate": 0.00024988574365201403, "loss": 1.4364479780197144, "step": 15471 }, { "epoch": 1.8003258087037468, "grad_norm": 1.2425575256347656, "learning_rate": 0.0002496572310515018, "loss": 1.7030448913574219, "step": 15472 }, { "epoch": 1.800442168955085, "grad_norm": 1.1856788396835327, "learning_rate": 0.0002494287187373692, "loss": 1.7315235137939453, "step": 15473 }, { "epoch": 1.8005585292064232, "grad_norm": 1.6163277626037598, "learning_rate": 0.00024920020690053575, "loss": 2.14127779006958, "step": 15474 }, { "epoch": 1.800674889457761, "grad_norm": 0.8938845992088318, "learning_rate": 0.0002489716957319205, "loss": 2.387648105621338, "step": 15475 }, { "epoch": 1.8007912497090994, "grad_norm": 0.5355536937713623, "learning_rate": 0.000248743185422442, "loss": 1.2984874248504639, "step": 15476 }, { "epoch": 1.8009076099604375, "grad_norm": 1.6543046236038208, "learning_rate": 0.000248514676163018, "loss": 1.7292877435684204, "step": 15477 }, { "epoch": 1.8010239702117756, "grad_norm": 0.9674696922302246, "learning_rate": 0.00024828616814456556, "loss": 1.9141271114349365, "step": 15478 }, { "epoch": 1.801140330463114, "grad_norm": 1.2022693157196045, "learning_rate": 0.0002480576615580004, "loss": 2.695356845855713, "step": 15479 }, { "epoch": 1.8012566907144518, "grad_norm": 0.7084754109382629, "learning_rate": 0.00024782915659423733, "loss": 1.2620960474014282, "step": 15480 }, { "epoch": 1.8013730509657901, "grad_norm": 0.6834622621536255, "learning_rate": 0.00024760065344418976, "loss": 1.319406270980835, "step": 15481 }, { "epoch": 1.8014894112171282, "grad_norm": 0.7814759612083435, "learning_rate": 0.00024737215229876945, "loss": 1.4057183265686035, "step": 15482 }, { "epoch": 1.8016057714684663, "grad_norm": 1.2476147413253784, "learning_rate": 0.0002471436533488865, "loss": 1.7751960754394531, "step": 15483 }, { "epoch": 1.8017221317198047, "grad_norm": 1.3693439960479736, "learning_rate": 0.00024691515678544937, "loss": 2.284108877182007, "step": 15484 }, { "epoch": 1.8018384919711425, "grad_norm": 1.1633565425872803, "learning_rate": 0.0002466866627993642, "loss": 2.0874929428100586, "step": 15485 }, { "epoch": 1.8019548522224809, "grad_norm": 1.0139662027359009, "learning_rate": 0.00024645817158153535, "loss": 2.2007880210876465, "step": 15486 }, { "epoch": 1.802071212473819, "grad_norm": 0.9536485075950623, "learning_rate": 0.00024622968332286454, "loss": 1.6245975494384766, "step": 15487 }, { "epoch": 1.802187572725157, "grad_norm": 0.8562478423118591, "learning_rate": 0.00024600119821425125, "loss": 2.0610241889953613, "step": 15488 }, { "epoch": 1.8023039329764954, "grad_norm": 1.4244076013565063, "learning_rate": 0.0002457727164465922, "loss": 1.8572375774383545, "step": 15489 }, { "epoch": 1.8024202932278333, "grad_norm": 0.747927188873291, "learning_rate": 0.0002455442382107814, "loss": 1.3086553812026978, "step": 15490 }, { "epoch": 1.8025366534791716, "grad_norm": 1.1906275749206543, "learning_rate": 0.0002453157636977098, "loss": 1.8380558490753174, "step": 15491 }, { "epoch": 1.8026530137305097, "grad_norm": 1.655691146850586, "learning_rate": 0.00024508729309826534, "loss": 2.1762731075286865, "step": 15492 }, { "epoch": 1.8027693739818478, "grad_norm": 1.2359373569488525, "learning_rate": 0.0002448588266033327, "loss": 2.011803388595581, "step": 15493 }, { "epoch": 1.802885734233186, "grad_norm": 0.6215803623199463, "learning_rate": 0.00024463036440379286, "loss": 1.6580573320388794, "step": 15494 }, { "epoch": 1.803002094484524, "grad_norm": 0.9995742440223694, "learning_rate": 0.00024440190669052387, "loss": 2.0021190643310547, "step": 15495 }, { "epoch": 1.8031184547358623, "grad_norm": 1.027936577796936, "learning_rate": 0.00024417345365439924, "loss": 2.065877676010132, "step": 15496 }, { "epoch": 1.8032348149872004, "grad_norm": 1.2574445009231567, "learning_rate": 0.000243945005486289, "loss": 2.552900791168213, "step": 15497 }, { "epoch": 1.8033511752385385, "grad_norm": 1.0207055807113647, "learning_rate": 0.000243716562377059, "loss": 1.5940921306610107, "step": 15498 }, { "epoch": 1.8034675354898768, "grad_norm": 1.1449213027954102, "learning_rate": 0.00024348812451757104, "loss": 2.057760715484619, "step": 15499 }, { "epoch": 1.8035838957412147, "grad_norm": 0.6311033368110657, "learning_rate": 0.00024325969209868238, "loss": 1.3902676105499268, "step": 15500 }, { "epoch": 1.803700255992553, "grad_norm": 1.1212949752807617, "learning_rate": 0.0002430312653112457, "loss": 1.7339762449264526, "step": 15501 }, { "epoch": 1.8038166162438911, "grad_norm": 1.8981341123580933, "learning_rate": 0.00024280284434610903, "loss": 2.081798553466797, "step": 15502 }, { "epoch": 1.8039329764952292, "grad_norm": 1.2820656299591064, "learning_rate": 0.0002425744293941156, "loss": 1.6684551239013672, "step": 15503 }, { "epoch": 1.8040493367465675, "grad_norm": 0.7054462432861328, "learning_rate": 0.00024234602064610348, "loss": 1.7350025177001953, "step": 15504 }, { "epoch": 1.8041656969979054, "grad_norm": 1.018911361694336, "learning_rate": 0.0002421176182929056, "loss": 1.8115922212600708, "step": 15505 }, { "epoch": 1.8042820572492437, "grad_norm": 1.0023564100265503, "learning_rate": 0.0002418892225253497, "loss": 1.5615006685256958, "step": 15506 }, { "epoch": 1.8043984175005818, "grad_norm": 0.5783562064170837, "learning_rate": 0.0002416608335342578, "loss": 1.5287995338439941, "step": 15507 }, { "epoch": 1.80451477775192, "grad_norm": 1.181769847869873, "learning_rate": 0.0002414324515104464, "loss": 1.810622215270996, "step": 15508 }, { "epoch": 1.8046311380032583, "grad_norm": 0.9243332743644714, "learning_rate": 0.00024120407664472612, "loss": 1.6860747337341309, "step": 15509 }, { "epoch": 1.8047474982545961, "grad_norm": 1.3629474639892578, "learning_rate": 0.00024097570912790162, "loss": 1.3954740762710571, "step": 15510 }, { "epoch": 1.8048638585059344, "grad_norm": 1.2059653997421265, "learning_rate": 0.00024074734915077134, "loss": 2.0235114097595215, "step": 15511 }, { "epoch": 1.8049802187572725, "grad_norm": 0.8574126958847046, "learning_rate": 0.00024051899690412753, "loss": 1.4592318534851074, "step": 15512 }, { "epoch": 1.8050965790086106, "grad_norm": 0.8688458800315857, "learning_rate": 0.000240290652578756, "loss": 1.5420997142791748, "step": 15513 }, { "epoch": 1.805212939259949, "grad_norm": 1.0027517080307007, "learning_rate": 0.00024006231636543586, "loss": 1.6881844997406006, "step": 15514 }, { "epoch": 1.8053292995112868, "grad_norm": 1.288520336151123, "learning_rate": 0.0002398339884549394, "loss": 2.640627384185791, "step": 15515 }, { "epoch": 1.8054456597626252, "grad_norm": 1.12419855594635, "learning_rate": 0.0002396056690380321, "loss": 1.4445650577545166, "step": 15516 }, { "epoch": 1.8055620200139633, "grad_norm": 0.7868884205818176, "learning_rate": 0.0002393773583054723, "loss": 1.683053970336914, "step": 15517 }, { "epoch": 1.8056783802653014, "grad_norm": 1.1690839529037476, "learning_rate": 0.00023914905644801103, "loss": 2.3076188564300537, "step": 15518 }, { "epoch": 1.8057947405166395, "grad_norm": 0.9312322735786438, "learning_rate": 0.000238920763656392, "loss": 2.014594316482544, "step": 15519 }, { "epoch": 1.8059111007679776, "grad_norm": 0.6428157687187195, "learning_rate": 0.0002386924801213512, "loss": 1.7704968452453613, "step": 15520 }, { "epoch": 1.8060274610193159, "grad_norm": 0.780544638633728, "learning_rate": 0.00023846420603361698, "loss": 1.787508487701416, "step": 15521 }, { "epoch": 1.806143821270654, "grad_norm": 1.2091771364212036, "learning_rate": 0.00023823594158390988, "loss": 1.8836860656738281, "step": 15522 }, { "epoch": 1.806260181521992, "grad_norm": 1.1607639789581299, "learning_rate": 0.00023800768696294214, "loss": 1.715767502784729, "step": 15523 }, { "epoch": 1.8063765417733302, "grad_norm": 0.749732494354248, "learning_rate": 0.00023777944236141817, "loss": 1.1894768476486206, "step": 15524 }, { "epoch": 1.8064929020246683, "grad_norm": 1.3953750133514404, "learning_rate": 0.0002375512079700336, "loss": 3.0256505012512207, "step": 15525 }, { "epoch": 1.8066092622760066, "grad_norm": 1.1186023950576782, "learning_rate": 0.00023732298397947582, "loss": 1.5244510173797607, "step": 15526 }, { "epoch": 1.8067256225273447, "grad_norm": 0.5997509360313416, "learning_rate": 0.00023709477058042342, "loss": 1.1765339374542236, "step": 15527 }, { "epoch": 1.8068419827786828, "grad_norm": 1.255555272102356, "learning_rate": 0.00023686656796354609, "loss": 2.239672899246216, "step": 15528 }, { "epoch": 1.806958343030021, "grad_norm": 1.086472511291504, "learning_rate": 0.00023663837631950462, "loss": 2.5025439262390137, "step": 15529 }, { "epoch": 1.807074703281359, "grad_norm": 1.4832550287246704, "learning_rate": 0.00023641019583895047, "loss": 1.9774243831634521, "step": 15530 }, { "epoch": 1.8071910635326973, "grad_norm": 0.800323486328125, "learning_rate": 0.00023618202671252603, "loss": 1.1591521501541138, "step": 15531 }, { "epoch": 1.8073074237840354, "grad_norm": 1.1223623752593994, "learning_rate": 0.00023595386913086402, "loss": 2.0955777168273926, "step": 15532 }, { "epoch": 1.8074237840353735, "grad_norm": 0.685416579246521, "learning_rate": 0.00023572572328458755, "loss": 1.2983193397521973, "step": 15533 }, { "epoch": 1.8075401442867116, "grad_norm": 0.9340817928314209, "learning_rate": 0.0002354975893643099, "loss": 2.055776834487915, "step": 15534 }, { "epoch": 1.8076565045380497, "grad_norm": 1.9142788648605347, "learning_rate": 0.00023526946756063442, "loss": 1.6600453853607178, "step": 15535 }, { "epoch": 1.807772864789388, "grad_norm": 1.4959871768951416, "learning_rate": 0.00023504135806415432, "loss": 2.2270922660827637, "step": 15536 }, { "epoch": 1.8078892250407261, "grad_norm": 0.8574092388153076, "learning_rate": 0.00023481326106545252, "loss": 2.3694825172424316, "step": 15537 }, { "epoch": 1.8080055852920642, "grad_norm": 0.8303701281547546, "learning_rate": 0.00023458517675510157, "loss": 2.062969446182251, "step": 15538 }, { "epoch": 1.8081219455434023, "grad_norm": 0.9970717430114746, "learning_rate": 0.00023435710532366337, "loss": 1.979172945022583, "step": 15539 }, { "epoch": 1.8082383057947404, "grad_norm": 0.9282205700874329, "learning_rate": 0.000234129046961689, "loss": 1.4744793176651, "step": 15540 }, { "epoch": 1.8083546660460788, "grad_norm": 1.435476541519165, "learning_rate": 0.00023390100185971874, "loss": 1.9198503494262695, "step": 15541 }, { "epoch": 1.8084710262974169, "grad_norm": 0.6879584193229675, "learning_rate": 0.00023367297020828167, "loss": 1.4860508441925049, "step": 15542 }, { "epoch": 1.808587386548755, "grad_norm": 0.8632386326789856, "learning_rate": 0.00023344495219789578, "loss": 1.8812519311904907, "step": 15543 }, { "epoch": 1.808703746800093, "grad_norm": 0.9722815752029419, "learning_rate": 0.00023321694801906742, "loss": 2.1171863079071045, "step": 15544 }, { "epoch": 1.8088201070514311, "grad_norm": 1.2145614624023438, "learning_rate": 0.00023298895786229163, "loss": 1.5262881517410278, "step": 15545 }, { "epoch": 1.8089364673027695, "grad_norm": 1.8705332279205322, "learning_rate": 0.00023276098191805166, "loss": 1.5710431337356567, "step": 15546 }, { "epoch": 1.8090528275541076, "grad_norm": 1.0125380754470825, "learning_rate": 0.0002325330203768188, "loss": 1.91446852684021, "step": 15547 }, { "epoch": 1.8091691878054457, "grad_norm": 0.9336572885513306, "learning_rate": 0.00023230507342905231, "loss": 1.486367106437683, "step": 15548 }, { "epoch": 1.8092855480567838, "grad_norm": 0.8391199111938477, "learning_rate": 0.0002320771412651995, "loss": 1.798439621925354, "step": 15549 }, { "epoch": 1.8094019083081219, "grad_norm": 1.1887966394424438, "learning_rate": 0.00023184922407569507, "loss": 1.9362444877624512, "step": 15550 }, { "epoch": 1.8095182685594602, "grad_norm": 0.8693891167640686, "learning_rate": 0.0002316213220509613, "loss": 1.3840558528900146, "step": 15551 }, { "epoch": 1.8096346288107983, "grad_norm": 1.0058308839797974, "learning_rate": 0.00023139343538140775, "loss": 1.4172170162200928, "step": 15552 }, { "epoch": 1.8097509890621364, "grad_norm": 1.056077480316162, "learning_rate": 0.0002311655642574312, "loss": 1.9505559206008911, "step": 15553 }, { "epoch": 1.8098673493134745, "grad_norm": 1.1015517711639404, "learning_rate": 0.00023093770886941547, "loss": 2.025325298309326, "step": 15554 }, { "epoch": 1.8099837095648126, "grad_norm": 0.7002280950546265, "learning_rate": 0.00023070986940773106, "loss": 1.8299672603607178, "step": 15555 }, { "epoch": 1.810100069816151, "grad_norm": 1.1989027261734009, "learning_rate": 0.00023048204606273547, "loss": 1.7954001426696777, "step": 15556 }, { "epoch": 1.8102164300674888, "grad_norm": 0.9373524188995361, "learning_rate": 0.0002302542390247725, "loss": 1.6791472434997559, "step": 15557 }, { "epoch": 1.810332790318827, "grad_norm": 0.6039368510246277, "learning_rate": 0.00023002644848417236, "loss": 1.482286810874939, "step": 15558 }, { "epoch": 1.8104491505701652, "grad_norm": 0.9418113231658936, "learning_rate": 0.0002297986746312515, "loss": 1.4193205833435059, "step": 15559 }, { "epoch": 1.8105655108215033, "grad_norm": 1.817898154258728, "learning_rate": 0.0002295709176563125, "loss": 1.4453647136688232, "step": 15560 }, { "epoch": 1.8106818710728416, "grad_norm": 1.1937569379806519, "learning_rate": 0.00022934317774964366, "loss": 1.3707613945007324, "step": 15561 }, { "epoch": 1.8107982313241795, "grad_norm": 1.0213940143585205, "learning_rate": 0.00022911545510151916, "loss": 1.970970630645752, "step": 15562 }, { "epoch": 1.8109145915755178, "grad_norm": 1.475417137145996, "learning_rate": 0.00022888774990219882, "loss": 2.084949493408203, "step": 15563 }, { "epoch": 1.811030951826856, "grad_norm": 1.5654772520065308, "learning_rate": 0.0002286600623419277, "loss": 1.8178080320358276, "step": 15564 }, { "epoch": 1.811147312078194, "grad_norm": 1.4458450078964233, "learning_rate": 0.00022843239261093626, "loss": 1.6160283088684082, "step": 15565 }, { "epoch": 1.8112636723295323, "grad_norm": 0.5925198197364807, "learning_rate": 0.00022820474089944002, "loss": 1.5356082916259766, "step": 15566 }, { "epoch": 1.8113800325808702, "grad_norm": 0.9266213774681091, "learning_rate": 0.00022797710739763948, "loss": 2.0158193111419678, "step": 15567 }, { "epoch": 1.8114963928322085, "grad_norm": 0.7840945720672607, "learning_rate": 0.00022774949229571975, "loss": 1.448959469795227, "step": 15568 }, { "epoch": 1.8116127530835466, "grad_norm": 0.6733312010765076, "learning_rate": 0.00022752189578385083, "loss": 1.880797266960144, "step": 15569 }, { "epoch": 1.8117291133348847, "grad_norm": 0.9676988124847412, "learning_rate": 0.00022729431805218697, "loss": 1.7249761819839478, "step": 15570 }, { "epoch": 1.811845473586223, "grad_norm": 1.5666426420211792, "learning_rate": 0.0002270667592908669, "loss": 2.1991403102874756, "step": 15571 }, { "epoch": 1.811961833837561, "grad_norm": 1.0970686674118042, "learning_rate": 0.00022683921969001344, "loss": 1.7426602840423584, "step": 15572 }, { "epoch": 1.8120781940888993, "grad_norm": 1.2898091077804565, "learning_rate": 0.00022661169943973328, "loss": 1.7429149150848389, "step": 15573 }, { "epoch": 1.8121945543402374, "grad_norm": 1.60221266746521, "learning_rate": 0.00022638419873011718, "loss": 2.081789016723633, "step": 15574 }, { "epoch": 1.8123109145915755, "grad_norm": 1.0599172115325928, "learning_rate": 0.0002261567177512394, "loss": 1.9754663705825806, "step": 15575 }, { "epoch": 1.8124272748429138, "grad_norm": 1.4292237758636475, "learning_rate": 0.0002259292566931577, "loss": 2.5864224433898926, "step": 15576 }, { "epoch": 1.8125436350942516, "grad_norm": 0.8523709774017334, "learning_rate": 0.0002257018157459133, "loss": 2.2126877307891846, "step": 15577 }, { "epoch": 1.81265999534559, "grad_norm": 1.2096606492996216, "learning_rate": 0.00022547439509953057, "loss": 1.576369285583496, "step": 15578 }, { "epoch": 1.812776355596928, "grad_norm": 1.2047722339630127, "learning_rate": 0.00022524699494401695, "loss": 1.988490104675293, "step": 15579 }, { "epoch": 1.8128927158482662, "grad_norm": 1.5163960456848145, "learning_rate": 0.00022501961546936262, "loss": 1.9975574016571045, "step": 15580 }, { "epoch": 1.8130090760996045, "grad_norm": 0.6534649729728699, "learning_rate": 0.00022479225686554072, "loss": 1.7480807304382324, "step": 15581 }, { "epoch": 1.8131254363509424, "grad_norm": 0.9819580316543579, "learning_rate": 0.0002245649193225068, "loss": 1.5104224681854248, "step": 15582 }, { "epoch": 1.8132417966022807, "grad_norm": 1.3515762090682983, "learning_rate": 0.0002243376030301988, "loss": 2.1187193393707275, "step": 15583 }, { "epoch": 1.8133581568536188, "grad_norm": 0.9323120713233948, "learning_rate": 0.00022411030817853696, "loss": 1.668213129043579, "step": 15584 }, { "epoch": 1.8134745171049569, "grad_norm": 1.1004029512405396, "learning_rate": 0.0002238830349574236, "loss": 2.4536385536193848, "step": 15585 }, { "epoch": 1.8135908773562952, "grad_norm": 1.0333751440048218, "learning_rate": 0.00022365578355674297, "loss": 1.3304474353790283, "step": 15586 }, { "epoch": 1.813707237607633, "grad_norm": 1.0296789407730103, "learning_rate": 0.00022342855416636096, "loss": 1.9702236652374268, "step": 15587 }, { "epoch": 1.8138235978589714, "grad_norm": 1.458789587020874, "learning_rate": 0.00022320134697612537, "loss": 2.215766668319702, "step": 15588 }, { "epoch": 1.8139399581103095, "grad_norm": 0.9418751001358032, "learning_rate": 0.00022297416217586518, "loss": 1.8716498613357544, "step": 15589 }, { "epoch": 1.8140563183616476, "grad_norm": 0.9473221302032471, "learning_rate": 0.00022274699995539077, "loss": 2.1570870876312256, "step": 15590 }, { "epoch": 1.814172678612986, "grad_norm": 0.9071857333183289, "learning_rate": 0.00022251986050449368, "loss": 1.5403528213500977, "step": 15591 }, { "epoch": 1.8142890388643238, "grad_norm": 0.8755216002464294, "learning_rate": 0.0002222927440129463, "loss": 1.5433456897735596, "step": 15592 }, { "epoch": 1.8144053991156621, "grad_norm": 0.8992266058921814, "learning_rate": 0.00022206565067050193, "loss": 2.3134920597076416, "step": 15593 }, { "epoch": 1.8145217593670002, "grad_norm": 0.8272578716278076, "learning_rate": 0.00022183858066689459, "loss": 1.4229884147644043, "step": 15594 }, { "epoch": 1.8146381196183383, "grad_norm": 1.1928268671035767, "learning_rate": 0.00022161153419183866, "loss": 2.845399856567383, "step": 15595 }, { "epoch": 1.8147544798696766, "grad_norm": 0.8845775127410889, "learning_rate": 0.000221384511435029, "loss": 2.129117965698242, "step": 15596 }, { "epoch": 1.8148708401210145, "grad_norm": 0.9387876391410828, "learning_rate": 0.00022115751258614054, "loss": 1.7170896530151367, "step": 15597 }, { "epoch": 1.8149872003723528, "grad_norm": 1.3309797048568726, "learning_rate": 0.00022093053783482826, "loss": 2.6331710815429688, "step": 15598 }, { "epoch": 1.815103560623691, "grad_norm": 1.2730594873428345, "learning_rate": 0.00022070358737072713, "loss": 1.6024973392486572, "step": 15599 }, { "epoch": 1.815219920875029, "grad_norm": 2.6725668907165527, "learning_rate": 0.00022047666138345166, "loss": 2.6393051147460938, "step": 15600 }, { "epoch": 1.8153362811263674, "grad_norm": 0.8098715543746948, "learning_rate": 0.00022024976006259598, "loss": 2.068523645401001, "step": 15601 }, { "epoch": 1.8154526413777052, "grad_norm": 1.0724149942398071, "learning_rate": 0.00022002288359773364, "loss": 1.8084150552749634, "step": 15602 }, { "epoch": 1.8155690016290436, "grad_norm": 1.0389548540115356, "learning_rate": 0.00021979603217841737, "loss": 2.154738664627075, "step": 15603 }, { "epoch": 1.8156853618803817, "grad_norm": 1.3406859636306763, "learning_rate": 0.00021956920599417901, "loss": 1.487307071685791, "step": 15604 }, { "epoch": 1.8158017221317198, "grad_norm": 1.0250146389007568, "learning_rate": 0.00021934240523452926, "loss": 1.9646100997924805, "step": 15605 }, { "epoch": 1.815918082383058, "grad_norm": 1.4481441974639893, "learning_rate": 0.00021911563008895769, "loss": 1.788297414779663, "step": 15606 }, { "epoch": 1.816034442634396, "grad_norm": 0.49045073986053467, "learning_rate": 0.00021888888074693237, "loss": 1.521802306175232, "step": 15607 }, { "epoch": 1.8161508028857343, "grad_norm": 0.9230407476425171, "learning_rate": 0.00021866215739789986, "loss": 2.051454544067383, "step": 15608 }, { "epoch": 1.8162671631370724, "grad_norm": 0.9472509026527405, "learning_rate": 0.00021843546023128497, "loss": 1.698742151260376, "step": 15609 }, { "epoch": 1.8163835233884105, "grad_norm": 0.8776006102561951, "learning_rate": 0.00021820878943649063, "loss": 2.132662773132324, "step": 15610 }, { "epoch": 1.8164998836397488, "grad_norm": 1.141474723815918, "learning_rate": 0.00021798214520289783, "loss": 2.2257509231567383, "step": 15611 }, { "epoch": 1.8166162438910867, "grad_norm": 1.1090118885040283, "learning_rate": 0.00021775552771986518, "loss": 1.6750026941299438, "step": 15612 }, { "epoch": 1.816732604142425, "grad_norm": 1.3162628412246704, "learning_rate": 0.00021752893717672916, "loss": 2.185382127761841, "step": 15613 }, { "epoch": 1.816848964393763, "grad_norm": 1.010538101196289, "learning_rate": 0.0002173023737628036, "loss": 2.4400112628936768, "step": 15614 }, { "epoch": 1.8169653246451012, "grad_norm": 0.9983518123626709, "learning_rate": 0.00021707583766737988, "loss": 1.3570201396942139, "step": 15615 }, { "epoch": 1.8170816848964395, "grad_norm": 0.7342597246170044, "learning_rate": 0.00021684932907972598, "loss": 1.4141929149627686, "step": 15616 }, { "epoch": 1.8171980451477774, "grad_norm": 1.1568753719329834, "learning_rate": 0.00021662284818908762, "loss": 2.6294097900390625, "step": 15617 }, { "epoch": 1.8173144053991157, "grad_norm": 0.7942149043083191, "learning_rate": 0.00021639639518468697, "loss": 1.1431617736816406, "step": 15618 }, { "epoch": 1.8174307656504538, "grad_norm": 0.3358435332775116, "learning_rate": 0.00021616997025572293, "loss": 0.9935063123703003, "step": 15619 }, { "epoch": 1.817547125901792, "grad_norm": 1.2174060344696045, "learning_rate": 0.00021594357359137106, "loss": 1.6246974468231201, "step": 15620 }, { "epoch": 1.8176634861531302, "grad_norm": 0.9996810555458069, "learning_rate": 0.00021571720538078324, "loss": 1.8898110389709473, "step": 15621 }, { "epoch": 1.817779846404468, "grad_norm": 0.8739258050918579, "learning_rate": 0.00021549086581308758, "loss": 1.696482539176941, "step": 15622 }, { "epoch": 1.8178962066558064, "grad_norm": 1.1987859010696411, "learning_rate": 0.0002152645550773882, "loss": 2.440147876739502, "step": 15623 }, { "epoch": 1.8180125669071445, "grad_norm": 1.0963495969772339, "learning_rate": 0.0002150382733627653, "loss": 1.7021647691726685, "step": 15624 }, { "epoch": 1.8181289271584826, "grad_norm": 1.0596370697021484, "learning_rate": 0.00021481202085827468, "loss": 1.778193712234497, "step": 15625 }, { "epoch": 1.818245287409821, "grad_norm": 0.8728698492050171, "learning_rate": 0.00021458579775294776, "loss": 1.2290513515472412, "step": 15626 }, { "epoch": 1.8183616476611588, "grad_norm": 1.1836435794830322, "learning_rate": 0.00021435960423579146, "loss": 2.153259754180908, "step": 15627 }, { "epoch": 1.8184780079124971, "grad_norm": 1.1030985116958618, "learning_rate": 0.0002141334404957879, "loss": 1.6617707014083862, "step": 15628 }, { "epoch": 1.8185943681638352, "grad_norm": 0.690244197845459, "learning_rate": 0.00021390730672189438, "loss": 1.3604233264923096, "step": 15629 }, { "epoch": 1.8187107284151733, "grad_norm": 1.2614634037017822, "learning_rate": 0.00021368120310304308, "loss": 1.873069405555725, "step": 15630 }, { "epoch": 1.8188270886665117, "grad_norm": 0.6767799854278564, "learning_rate": 0.00021345512982814117, "loss": 1.707397699356079, "step": 15631 }, { "epoch": 1.8189434489178495, "grad_norm": 1.7040374279022217, "learning_rate": 0.0002132290870860703, "loss": 1.7761380672454834, "step": 15632 }, { "epoch": 1.8190598091691879, "grad_norm": 1.3267583847045898, "learning_rate": 0.00021300307506568662, "loss": 2.655547618865967, "step": 15633 }, { "epoch": 1.819176169420526, "grad_norm": 0.8565047979354858, "learning_rate": 0.00021277709395582065, "loss": 1.763076663017273, "step": 15634 }, { "epoch": 1.819292529671864, "grad_norm": 1.6754034757614136, "learning_rate": 0.00021255114394527714, "loss": 2.6314706802368164, "step": 15635 }, { "epoch": 1.8194088899232024, "grad_norm": 1.3756709098815918, "learning_rate": 0.00021232522522283476, "loss": 2.28094482421875, "step": 15636 }, { "epoch": 1.8195252501745403, "grad_norm": 1.0857727527618408, "learning_rate": 0.000212099337977246, "loss": 1.4236712455749512, "step": 15637 }, { "epoch": 1.8196416104258786, "grad_norm": 1.317086935043335, "learning_rate": 0.00021187348239723735, "loss": 1.6493360996246338, "step": 15638 }, { "epoch": 1.8197579706772167, "grad_norm": 1.1225714683532715, "learning_rate": 0.00021164765867150853, "loss": 1.6671710014343262, "step": 15639 }, { "epoch": 1.8198743309285548, "grad_norm": 1.0273725986480713, "learning_rate": 0.00021142186698873262, "loss": 3.0333266258239746, "step": 15640 }, { "epoch": 1.819990691179893, "grad_norm": 0.9751765131950378, "learning_rate": 0.0002111961075375561, "loss": 1.951418399810791, "step": 15641 }, { "epoch": 1.820107051431231, "grad_norm": 0.5348002910614014, "learning_rate": 0.00021097038050659853, "loss": 1.303789496421814, "step": 15642 }, { "epoch": 1.8202234116825693, "grad_norm": 1.0089106559753418, "learning_rate": 0.0002107446860844523, "loss": 1.7478052377700806, "step": 15643 }, { "epoch": 1.8203397719339074, "grad_norm": 1.1310420036315918, "learning_rate": 0.00021051902445968262, "loss": 2.298267126083374, "step": 15644 }, { "epoch": 1.8204561321852455, "grad_norm": 0.8255435228347778, "learning_rate": 0.00021029339582082722, "loss": 1.6628577709197998, "step": 15645 }, { "epoch": 1.8205724924365838, "grad_norm": 0.7326230406761169, "learning_rate": 0.0002100678003563963, "loss": 1.5246092081069946, "step": 15646 }, { "epoch": 1.8206888526879217, "grad_norm": 0.850886881351471, "learning_rate": 0.00020984223825487233, "loss": 1.373776912689209, "step": 15647 }, { "epoch": 1.82080521293926, "grad_norm": 0.3796163499355316, "learning_rate": 0.0002096167097047099, "loss": 1.0922529697418213, "step": 15648 }, { "epoch": 1.820921573190598, "grad_norm": 0.9353533387184143, "learning_rate": 0.00020939121489433574, "loss": 2.2804036140441895, "step": 15649 }, { "epoch": 1.8210379334419362, "grad_norm": 0.7378483414649963, "learning_rate": 0.00020916575401214816, "loss": 1.5541651248931885, "step": 15650 }, { "epoch": 1.8211542936932745, "grad_norm": 0.8032160401344299, "learning_rate": 0.0002089403272465172, "loss": 1.6754815578460693, "step": 15651 }, { "epoch": 1.8212706539446124, "grad_norm": 0.9288827180862427, "learning_rate": 0.0002087149347857844, "loss": 1.7122770547866821, "step": 15652 }, { "epoch": 1.8213870141959507, "grad_norm": 0.3475898504257202, "learning_rate": 0.00020848957681826269, "loss": 0.9953343868255615, "step": 15653 }, { "epoch": 1.8215033744472888, "grad_norm": 0.8768535852432251, "learning_rate": 0.0002082642535322361, "loss": 1.9627907276153564, "step": 15654 }, { "epoch": 1.821619734698627, "grad_norm": 1.1564984321594238, "learning_rate": 0.00020803896511595966, "loss": 1.4676967859268188, "step": 15655 }, { "epoch": 1.8217360949499652, "grad_norm": 1.16844642162323, "learning_rate": 0.00020781371175765945, "loss": 1.901845932006836, "step": 15656 }, { "epoch": 1.8218524552013031, "grad_norm": 0.8055492639541626, "learning_rate": 0.0002075884936455321, "loss": 1.2914377450942993, "step": 15657 }, { "epoch": 1.8219688154526414, "grad_norm": 1.0595554113388062, "learning_rate": 0.00020736331096774479, "loss": 1.2417805194854736, "step": 15658 }, { "epoch": 1.8220851757039795, "grad_norm": 0.5272083282470703, "learning_rate": 0.0002071381639124351, "loss": 1.2881693840026855, "step": 15659 }, { "epoch": 1.8222015359553176, "grad_norm": 1.4930890798568726, "learning_rate": 0.00020691305266771102, "loss": 1.834523320198059, "step": 15660 }, { "epoch": 1.822317896206656, "grad_norm": 0.920649528503418, "learning_rate": 0.00020668797742165037, "loss": 1.7869350910186768, "step": 15661 }, { "epoch": 1.8224342564579938, "grad_norm": 1.0753228664398193, "learning_rate": 0.00020646293836230092, "loss": 2.1295371055603027, "step": 15662 }, { "epoch": 1.8225506167093322, "grad_norm": 1.3780558109283447, "learning_rate": 0.0002062379356776806, "loss": 1.566754937171936, "step": 15663 }, { "epoch": 1.8226669769606703, "grad_norm": 1.1232820749282837, "learning_rate": 0.00020601296955577634, "loss": 1.7668304443359375, "step": 15664 }, { "epoch": 1.8227833372120084, "grad_norm": 0.834736168384552, "learning_rate": 0.00020578804018454492, "loss": 1.2680771350860596, "step": 15665 }, { "epoch": 1.8228996974633467, "grad_norm": 1.3923206329345703, "learning_rate": 0.00020556314775191226, "loss": 2.630291700363159, "step": 15666 }, { "epoch": 1.8230160577146846, "grad_norm": 0.8697594404220581, "learning_rate": 0.0002053382924457736, "loss": 1.244476079940796, "step": 15667 }, { "epoch": 1.8231324179660229, "grad_norm": 1.217079520225525, "learning_rate": 0.00020511347445399298, "loss": 2.0621395111083984, "step": 15668 }, { "epoch": 1.823248778217361, "grad_norm": 0.7620800733566284, "learning_rate": 0.0002048886939644033, "loss": 1.420647144317627, "step": 15669 }, { "epoch": 1.823365138468699, "grad_norm": 0.7731828093528748, "learning_rate": 0.00020466395116480618, "loss": 1.3343489170074463, "step": 15670 }, { "epoch": 1.8234814987200374, "grad_norm": 3.704683303833008, "learning_rate": 0.00020443924624297173, "loss": 1.762884497642517, "step": 15671 }, { "epoch": 1.8235978589713753, "grad_norm": 1.2447779178619385, "learning_rate": 0.00020421457938663834, "loss": 1.8429865837097168, "step": 15672 }, { "epoch": 1.8237142192227136, "grad_norm": 1.4640885591506958, "learning_rate": 0.00020398995078351267, "loss": 1.3024227619171143, "step": 15673 }, { "epoch": 1.8238305794740517, "grad_norm": 1.563245415687561, "learning_rate": 0.0002037653606212695, "loss": 2.0420618057250977, "step": 15674 }, { "epoch": 1.8239469397253898, "grad_norm": 0.8656738996505737, "learning_rate": 0.00020354080908755134, "loss": 1.3458054065704346, "step": 15675 }, { "epoch": 1.8240632999767281, "grad_norm": 0.8724096417427063, "learning_rate": 0.00020331629636996851, "loss": 1.369375467300415, "step": 15676 }, { "epoch": 1.824179660228066, "grad_norm": 0.8715832233428955, "learning_rate": 0.00020309182265609887, "loss": 1.6536195278167725, "step": 15677 }, { "epoch": 1.8242960204794043, "grad_norm": 0.586666464805603, "learning_rate": 0.0002028673881334877, "loss": 1.1828193664550781, "step": 15678 }, { "epoch": 1.8244123807307424, "grad_norm": 0.9780502319335938, "learning_rate": 0.00020264299298964751, "loss": 1.71730375289917, "step": 15679 }, { "epoch": 1.8245287409820805, "grad_norm": 0.5313581824302673, "learning_rate": 0.0002024186374120579, "loss": 1.3035341501235962, "step": 15680 }, { "epoch": 1.8246451012334186, "grad_norm": 0.9952511787414551, "learning_rate": 0.00020219432158816557, "loss": 1.5556325912475586, "step": 15681 }, { "epoch": 1.8247614614847567, "grad_norm": 0.7232285737991333, "learning_rate": 0.00020197004570538383, "loss": 1.8155455589294434, "step": 15682 }, { "epoch": 1.824877821736095, "grad_norm": 0.7945674061775208, "learning_rate": 0.0002017458099510927, "loss": 1.2591733932495117, "step": 15683 }, { "epoch": 1.8249941819874331, "grad_norm": 0.8285760879516602, "learning_rate": 0.00020152161451263855, "loss": 1.3933451175689697, "step": 15684 }, { "epoch": 1.8251105422387712, "grad_norm": 0.9498863816261292, "learning_rate": 0.0002012974595773343, "loss": 1.882197618484497, "step": 15685 }, { "epoch": 1.8252269024901093, "grad_norm": 1.863896131515503, "learning_rate": 0.00020107334533245883, "loss": 2.380340337753296, "step": 15686 }, { "epoch": 1.8253432627414474, "grad_norm": 0.6150678396224976, "learning_rate": 0.00020084927196525716, "loss": 1.0663481950759888, "step": 15687 }, { "epoch": 1.8254596229927857, "grad_norm": 1.1284514665603638, "learning_rate": 0.00020062523966293995, "loss": 1.9549674987792969, "step": 15688 }, { "epoch": 1.8255759832441238, "grad_norm": 1.248962163925171, "learning_rate": 0.00020040124861268375, "loss": 2.564577579498291, "step": 15689 }, { "epoch": 1.825692343495462, "grad_norm": 0.9014826416969299, "learning_rate": 0.0002001772990016306, "loss": 1.8464434146881104, "step": 15690 }, { "epoch": 1.8258087037468, "grad_norm": 1.046959400177002, "learning_rate": 0.00019995339101688784, "loss": 1.8731955289840698, "step": 15691 }, { "epoch": 1.8259250639981381, "grad_norm": 1.5803346633911133, "learning_rate": 0.0001997295248455282, "loss": 1.976105809211731, "step": 15692 }, { "epoch": 1.8260414242494765, "grad_norm": 1.2253831624984741, "learning_rate": 0.00019950570067458928, "loss": 1.3309823274612427, "step": 15693 }, { "epoch": 1.8261577845008146, "grad_norm": 1.5660274028778076, "learning_rate": 0.00019928191869107373, "loss": 2.0500636100769043, "step": 15694 }, { "epoch": 1.8262741447521527, "grad_norm": 0.9933498501777649, "learning_rate": 0.00019905817908194885, "loss": 1.8112547397613525, "step": 15695 }, { "epoch": 1.8263905050034908, "grad_norm": 1.002232551574707, "learning_rate": 0.00019883448203414665, "loss": 1.2940515279769897, "step": 15696 }, { "epoch": 1.8265068652548289, "grad_norm": 0.5198765397071838, "learning_rate": 0.00019861082773456342, "loss": 1.1046549081802368, "step": 15697 }, { "epoch": 1.8266232255061672, "grad_norm": 0.7441738843917847, "learning_rate": 0.00019838721637005985, "loss": 2.108430862426758, "step": 15698 }, { "epoch": 1.8267395857575053, "grad_norm": 1.0885133743286133, "learning_rate": 0.00019816364812746086, "loss": 1.9197897911071777, "step": 15699 }, { "epoch": 1.8268559460088434, "grad_norm": 0.7566596865653992, "learning_rate": 0.0001979401231935551, "loss": 1.4491910934448242, "step": 15700 }, { "epoch": 1.8269723062601815, "grad_norm": 1.2596139907836914, "learning_rate": 0.0001977166417550952, "loss": 1.5412213802337646, "step": 15701 }, { "epoch": 1.8270886665115196, "grad_norm": 0.8968963623046875, "learning_rate": 0.00019749320399879738, "loss": 1.3330602645874023, "step": 15702 }, { "epoch": 1.827205026762858, "grad_norm": 0.9613963961601257, "learning_rate": 0.00019726981011134145, "loss": 1.4942100048065186, "step": 15703 }, { "epoch": 1.827321387014196, "grad_norm": 1.1714653968811035, "learning_rate": 0.00019704646027937046, "loss": 1.6830394268035889, "step": 15704 }, { "epoch": 1.827437747265534, "grad_norm": 0.7826877236366272, "learning_rate": 0.00019682315468949062, "loss": 1.273191213607788, "step": 15705 }, { "epoch": 1.8275541075168722, "grad_norm": 1.0359671115875244, "learning_rate": 0.0001965998935282714, "loss": 1.668341875076294, "step": 15706 }, { "epoch": 1.8276704677682103, "grad_norm": 1.3036564588546753, "learning_rate": 0.00019637667698224498, "loss": 2.9809300899505615, "step": 15707 }, { "epoch": 1.8277868280195486, "grad_norm": 1.0458049774169922, "learning_rate": 0.00019615350523790627, "loss": 1.8128904104232788, "step": 15708 }, { "epoch": 1.8279031882708867, "grad_norm": 0.5644547939300537, "learning_rate": 0.00019593037848171274, "loss": 1.328629970550537, "step": 15709 }, { "epoch": 1.8280195485222248, "grad_norm": 0.986656904220581, "learning_rate": 0.00019570729690008434, "loss": 2.170064926147461, "step": 15710 }, { "epoch": 1.828135908773563, "grad_norm": 1.118298053741455, "learning_rate": 0.00019548426067940328, "loss": 2.587998867034912, "step": 15711 }, { "epoch": 1.828252269024901, "grad_norm": 1.5788437128067017, "learning_rate": 0.00019526127000601374, "loss": 2.0899016857147217, "step": 15712 }, { "epoch": 1.8283686292762393, "grad_norm": 0.7894521355628967, "learning_rate": 0.00019503832506622199, "loss": 1.7508121728897095, "step": 15713 }, { "epoch": 1.8284849895275774, "grad_norm": 0.6714400053024292, "learning_rate": 0.00019481542604629602, "loss": 1.695580244064331, "step": 15714 }, { "epoch": 1.8286013497789155, "grad_norm": 0.7376684546470642, "learning_rate": 0.00019459257313246547, "loss": 1.908225417137146, "step": 15715 }, { "epoch": 1.8287177100302536, "grad_norm": 0.9535755515098572, "learning_rate": 0.00019436976651092142, "loss": 2.3171865940093994, "step": 15716 }, { "epoch": 1.8288340702815917, "grad_norm": 1.309288501739502, "learning_rate": 0.00019414700636781644, "loss": 2.712559938430786, "step": 15717 }, { "epoch": 1.82895043053293, "grad_norm": 0.9552779793739319, "learning_rate": 0.00019392429288926405, "loss": 1.3931101560592651, "step": 15718 }, { "epoch": 1.8290667907842681, "grad_norm": 1.5896357297897339, "learning_rate": 0.0001937016262613389, "loss": 2.1827392578125, "step": 15719 }, { "epoch": 1.8291831510356062, "grad_norm": 0.9816489815711975, "learning_rate": 0.00019347900667007645, "loss": 1.7692763805389404, "step": 15720 }, { "epoch": 1.8292995112869443, "grad_norm": 0.4381556808948517, "learning_rate": 0.00019325643430147293, "loss": 0.9825310707092285, "step": 15721 }, { "epoch": 1.8294158715382824, "grad_norm": 1.2500895261764526, "learning_rate": 0.00019303390934148504, "loss": 2.12453031539917, "step": 15722 }, { "epoch": 1.8295322317896208, "grad_norm": 0.6672496199607849, "learning_rate": 0.00019281143197602982, "loss": 1.1515122652053833, "step": 15723 }, { "epoch": 1.8296485920409586, "grad_norm": 0.8668254613876343, "learning_rate": 0.0001925890023909848, "loss": 1.2541873455047607, "step": 15724 }, { "epoch": 1.829764952292297, "grad_norm": 0.631112277507782, "learning_rate": 0.00019236662077218728, "loss": 1.7712078094482422, "step": 15725 }, { "epoch": 1.829881312543635, "grad_norm": 1.1933683156967163, "learning_rate": 0.00019214428730543465, "loss": 2.289910316467285, "step": 15726 }, { "epoch": 1.8299976727949732, "grad_norm": 0.7593955397605896, "learning_rate": 0.0001919220021764841, "loss": 1.792138695716858, "step": 15727 }, { "epoch": 1.8301140330463115, "grad_norm": 0.7368651032447815, "learning_rate": 0.00019169976557105225, "loss": 1.8590269088745117, "step": 15728 }, { "epoch": 1.8302303932976494, "grad_norm": 1.183263897895813, "learning_rate": 0.00019147757767481544, "loss": 2.0587158203125, "step": 15729 }, { "epoch": 1.8303467535489877, "grad_norm": 0.8307294249534607, "learning_rate": 0.00019125543867340902, "loss": 1.8494577407836914, "step": 15730 }, { "epoch": 1.8304631138003258, "grad_norm": 0.9222984910011292, "learning_rate": 0.00019103334875242777, "loss": 2.1003060340881348, "step": 15731 }, { "epoch": 1.8305794740516639, "grad_norm": 0.8106768727302551, "learning_rate": 0.00019081130809742535, "loss": 1.380920648574829, "step": 15732 }, { "epoch": 1.8306958343030022, "grad_norm": 1.8706361055374146, "learning_rate": 0.0001905893168939142, "loss": 2.4587578773498535, "step": 15733 }, { "epoch": 1.83081219455434, "grad_norm": 0.7149333953857422, "learning_rate": 0.00019036737532736548, "loss": 1.7495551109313965, "step": 15734 }, { "epoch": 1.8309285548056784, "grad_norm": 1.4706201553344727, "learning_rate": 0.000190145483583209, "loss": 2.0094637870788574, "step": 15735 }, { "epoch": 1.8310449150570165, "grad_norm": 1.4788949489593506, "learning_rate": 0.00018992364184683265, "loss": 2.126901865005493, "step": 15736 }, { "epoch": 1.8311612753083546, "grad_norm": 0.9024854302406311, "learning_rate": 0.0001897018503035828, "loss": 1.3006088733673096, "step": 15737 }, { "epoch": 1.831277635559693, "grad_norm": 0.9064691066741943, "learning_rate": 0.00018948010913876384, "loss": 2.490736961364746, "step": 15738 }, { "epoch": 1.8313939958110308, "grad_norm": 1.054195761680603, "learning_rate": 0.00018925841853763804, "loss": 1.879334568977356, "step": 15739 }, { "epoch": 1.8315103560623691, "grad_norm": 1.0622807741165161, "learning_rate": 0.00018903677868542537, "loss": 1.9992120265960693, "step": 15740 }, { "epoch": 1.8316267163137072, "grad_norm": 1.0626274347305298, "learning_rate": 0.00018881518976730346, "loss": 1.9902535676956177, "step": 15741 }, { "epoch": 1.8317430765650453, "grad_norm": 1.1322438716888428, "learning_rate": 0.00018859365196840743, "loss": 2.2863755226135254, "step": 15742 }, { "epoch": 1.8318594368163836, "grad_norm": 0.9193758368492126, "learning_rate": 0.00018837216547382964, "loss": 1.9603028297424316, "step": 15743 }, { "epoch": 1.8319757970677215, "grad_norm": 0.5466508865356445, "learning_rate": 0.00018815073046861959, "loss": 1.1217948198318481, "step": 15744 }, { "epoch": 1.8320921573190598, "grad_norm": 1.167702555656433, "learning_rate": 0.00018792934713778364, "loss": 1.5160884857177734, "step": 15745 }, { "epoch": 1.832208517570398, "grad_norm": 1.13248610496521, "learning_rate": 0.0001877080156662852, "loss": 1.930403470993042, "step": 15746 }, { "epoch": 1.832324877821736, "grad_norm": 1.900089144706726, "learning_rate": 0.00018748673623904423, "loss": 1.677936315536499, "step": 15747 }, { "epoch": 1.8324412380730744, "grad_norm": 1.3347450494766235, "learning_rate": 0.00018726550904093713, "loss": 2.337747097015381, "step": 15748 }, { "epoch": 1.8325575983244122, "grad_norm": 0.7111122608184814, "learning_rate": 0.0001870443342567969, "loss": 1.812319278717041, "step": 15749 }, { "epoch": 1.8326739585757506, "grad_norm": 1.0797810554504395, "learning_rate": 0.0001868232120714125, "loss": 2.12395977973938, "step": 15750 }, { "epoch": 1.8327903188270886, "grad_norm": 0.8945922255516052, "learning_rate": 0.00018660214266952912, "loss": 2.120246171951294, "step": 15751 }, { "epoch": 1.8329066790784267, "grad_norm": 1.0334500074386597, "learning_rate": 0.00018638112623584773, "loss": 2.255311965942383, "step": 15752 }, { "epoch": 1.833023039329765, "grad_norm": 2.5453100204467773, "learning_rate": 0.0001861601629550251, "loss": 1.4094552993774414, "step": 15753 }, { "epoch": 1.833139399581103, "grad_norm": 0.9235179424285889, "learning_rate": 0.00018593925301167355, "loss": 1.6205530166625977, "step": 15754 }, { "epoch": 1.8332557598324413, "grad_norm": 1.2550914287567139, "learning_rate": 0.0001857183965903609, "loss": 2.0643198490142822, "step": 15755 }, { "epoch": 1.8333721200837794, "grad_norm": 0.6634029746055603, "learning_rate": 0.00018549759387561022, "loss": 1.4851529598236084, "step": 15756 }, { "epoch": 1.8334884803351175, "grad_norm": 1.0146828889846802, "learning_rate": 0.00018527684505189977, "loss": 2.2937817573547363, "step": 15757 }, { "epoch": 1.8336048405864558, "grad_norm": 0.5756692886352539, "learning_rate": 0.00018505615030366265, "loss": 1.3477922677993774, "step": 15758 }, { "epoch": 1.8337212008377937, "grad_norm": 0.30801403522491455, "learning_rate": 0.00018483550981528697, "loss": 0.981408953666687, "step": 15759 }, { "epoch": 1.833837561089132, "grad_norm": 1.1488953828811646, "learning_rate": 0.00018461492377111523, "loss": 1.9972741603851318, "step": 15760 }, { "epoch": 1.83395392134047, "grad_norm": 0.9554163813591003, "learning_rate": 0.0001843943923554447, "loss": 1.4161936044692993, "step": 15761 }, { "epoch": 1.8340702815918082, "grad_norm": 0.5889491438865662, "learning_rate": 0.0001841739157525269, "loss": 1.0880826711654663, "step": 15762 }, { "epoch": 1.8341866418431465, "grad_norm": 0.7494977116584778, "learning_rate": 0.0001839534941465676, "loss": 1.8102747201919556, "step": 15763 }, { "epoch": 1.8343030020944844, "grad_norm": 1.208962082862854, "learning_rate": 0.00018373312772172653, "loss": 2.1663684844970703, "step": 15764 }, { "epoch": 1.8344193623458227, "grad_norm": 1.1270971298217773, "learning_rate": 0.00018351281666211745, "loss": 1.8040854930877686, "step": 15765 }, { "epoch": 1.8345357225971608, "grad_norm": 2.1700918674468994, "learning_rate": 0.00018329256115180769, "loss": 2.7800791263580322, "step": 15766 }, { "epoch": 1.834652082848499, "grad_norm": 0.8582531213760376, "learning_rate": 0.00018307236137481836, "loss": 1.8561034202575684, "step": 15767 }, { "epoch": 1.8347684430998372, "grad_norm": 0.9519188404083252, "learning_rate": 0.00018285221751512392, "loss": 2.3669662475585938, "step": 15768 }, { "epoch": 1.834884803351175, "grad_norm": 1.253933072090149, "learning_rate": 0.00018263212975665205, "loss": 2.454073905944824, "step": 15769 }, { "epoch": 1.8350011636025134, "grad_norm": 0.7907527685165405, "learning_rate": 0.00018241209828328368, "loss": 1.4890856742858887, "step": 15770 }, { "epoch": 1.8351175238538515, "grad_norm": 1.0223594903945923, "learning_rate": 0.0001821921232788526, "loss": 1.7485321760177612, "step": 15771 }, { "epoch": 1.8352338841051896, "grad_norm": 1.3728232383728027, "learning_rate": 0.00018197220492714546, "loss": 2.3529391288757324, "step": 15772 }, { "epoch": 1.835350244356528, "grad_norm": 1.1847609281539917, "learning_rate": 0.00018175234341190157, "loss": 1.8781883716583252, "step": 15773 }, { "epoch": 1.8354666046078658, "grad_norm": 1.3824868202209473, "learning_rate": 0.00018153253891681285, "loss": 1.8880133628845215, "step": 15774 }, { "epoch": 1.8355829648592041, "grad_norm": 0.7334588170051575, "learning_rate": 0.00018131279162552348, "loss": 1.6326885223388672, "step": 15775 }, { "epoch": 1.8356993251105422, "grad_norm": 1.545265793800354, "learning_rate": 0.00018109310172162985, "loss": 3.0850672721862793, "step": 15776 }, { "epoch": 1.8358156853618803, "grad_norm": 1.2952308654785156, "learning_rate": 0.0001808734693886804, "loss": 2.396346092224121, "step": 15777 }, { "epoch": 1.8359320456132187, "grad_norm": 0.9375651478767395, "learning_rate": 0.00018065389481017552, "loss": 1.7496668100357056, "step": 15778 }, { "epoch": 1.8360484058645565, "grad_norm": 1.14775812625885, "learning_rate": 0.00018043437816956732, "loss": 1.5824211835861206, "step": 15779 }, { "epoch": 1.8361647661158949, "grad_norm": 0.9398106336593628, "learning_rate": 0.00018021491965025948, "loss": 1.6765114068984985, "step": 15780 }, { "epoch": 1.836281126367233, "grad_norm": 1.4477317333221436, "learning_rate": 0.00017999551943560717, "loss": 1.8644438982009888, "step": 15781 }, { "epoch": 1.836397486618571, "grad_norm": 0.8989177346229553, "learning_rate": 0.0001797761777089168, "loss": 1.8152408599853516, "step": 15782 }, { "epoch": 1.8365138468699094, "grad_norm": 1.0019181966781616, "learning_rate": 0.00017955689465344604, "loss": 1.4809787273406982, "step": 15783 }, { "epoch": 1.8366302071212472, "grad_norm": 1.0173767805099487, "learning_rate": 0.00017933767045240324, "loss": 1.7158946990966797, "step": 15784 }, { "epoch": 1.8367465673725856, "grad_norm": 1.231086015701294, "learning_rate": 0.00017911850528894786, "loss": 1.2703564167022705, "step": 15785 }, { "epoch": 1.8368629276239237, "grad_norm": 1.1609454154968262, "learning_rate": 0.00017889939934618998, "loss": 1.7088929414749146, "step": 15786 }, { "epoch": 1.8369792878752618, "grad_norm": 0.8138101100921631, "learning_rate": 0.00017868035280719014, "loss": 2.1656904220581055, "step": 15787 }, { "epoch": 1.8370956481266, "grad_norm": 1.4930880069732666, "learning_rate": 0.00017846136585495932, "loss": 2.2149510383605957, "step": 15788 }, { "epoch": 1.837212008377938, "grad_norm": 1.1659578084945679, "learning_rate": 0.0001782424386724586, "loss": 1.8670828342437744, "step": 15789 }, { "epoch": 1.8373283686292763, "grad_norm": 0.801990270614624, "learning_rate": 0.00017802357144259924, "loss": 1.7285335063934326, "step": 15790 }, { "epoch": 1.8374447288806144, "grad_norm": 0.9358230829238892, "learning_rate": 0.0001778047643482423, "loss": 2.0694894790649414, "step": 15791 }, { "epoch": 1.8375610891319525, "grad_norm": 1.1793049573898315, "learning_rate": 0.00017758601757219877, "loss": 1.3168878555297852, "step": 15792 }, { "epoch": 1.8376774493832908, "grad_norm": 1.9368304014205933, "learning_rate": 0.00017736733129722912, "loss": 2.184929370880127, "step": 15793 }, { "epoch": 1.8377938096346287, "grad_norm": 1.1724227666854858, "learning_rate": 0.00017714870570604324, "loss": 1.7596584558486938, "step": 15794 }, { "epoch": 1.837910169885967, "grad_norm": 1.1254912614822388, "learning_rate": 0.00017693014098130045, "loss": 2.2341861724853516, "step": 15795 }, { "epoch": 1.838026530137305, "grad_norm": 0.549747109413147, "learning_rate": 0.00017671163730560903, "loss": 0.9531397819519043, "step": 15796 }, { "epoch": 1.8381428903886432, "grad_norm": 1.0945425033569336, "learning_rate": 0.00017649319486152642, "loss": 1.7515103816986084, "step": 15797 }, { "epoch": 1.8382592506399815, "grad_norm": 0.7999181151390076, "learning_rate": 0.0001762748138315588, "loss": 1.1347086429595947, "step": 15798 }, { "epoch": 1.8383756108913194, "grad_norm": 0.7378674745559692, "learning_rate": 0.00017605649439816117, "loss": 1.6156491041183472, "step": 15799 }, { "epoch": 1.8384919711426577, "grad_norm": 0.8593227863311768, "learning_rate": 0.00017583823674373694, "loss": 2.1307458877563477, "step": 15800 }, { "epoch": 1.8386083313939958, "grad_norm": 1.4020401239395142, "learning_rate": 0.00017562004105063793, "loss": 2.3699355125427246, "step": 15801 }, { "epoch": 1.838724691645334, "grad_norm": 0.7883890271186829, "learning_rate": 0.0001754019075011642, "loss": 2.0895581245422363, "step": 15802 }, { "epoch": 1.8388410518966722, "grad_norm": 1.0387529134750366, "learning_rate": 0.00017518383627756391, "loss": 1.678938865661621, "step": 15803 }, { "epoch": 1.8389574121480101, "grad_norm": 1.0420176982879639, "learning_rate": 0.0001749658275620331, "loss": 1.5963175296783447, "step": 15804 }, { "epoch": 1.8390737723993484, "grad_norm": 1.1286979913711548, "learning_rate": 0.00017474788153671564, "loss": 1.4836959838867188, "step": 15805 }, { "epoch": 1.8391901326506865, "grad_norm": 1.2183794975280762, "learning_rate": 0.00017452999838370303, "loss": 2.1255412101745605, "step": 15806 }, { "epoch": 1.8393064929020246, "grad_norm": 1.0606098175048828, "learning_rate": 0.0001743121782850343, "loss": 1.9669626951217651, "step": 15807 }, { "epoch": 1.839422853153363, "grad_norm": 1.4615509510040283, "learning_rate": 0.0001740944214226955, "loss": 1.7538962364196777, "step": 15808 }, { "epoch": 1.8395392134047008, "grad_norm": 1.1194673776626587, "learning_rate": 0.0001738767279786201, "loss": 1.6628931760787964, "step": 15809 }, { "epoch": 1.8396555736560392, "grad_norm": 1.167276382446289, "learning_rate": 0.00017365909813468863, "loss": 1.9881073236465454, "step": 15810 }, { "epoch": 1.8397719339073773, "grad_norm": 0.8841530084609985, "learning_rate": 0.0001734415320727284, "loss": 1.581785798072815, "step": 15811 }, { "epoch": 1.8398882941587154, "grad_norm": 0.7945932149887085, "learning_rate": 0.00017322402997451336, "loss": 2.429759979248047, "step": 15812 }, { "epoch": 1.8400046544100537, "grad_norm": 0.7301977276802063, "learning_rate": 0.00017300659202176412, "loss": 1.6612348556518555, "step": 15813 }, { "epoch": 1.8401210146613916, "grad_norm": 1.0198917388916016, "learning_rate": 0.00017278921839614764, "loss": 1.1769139766693115, "step": 15814 }, { "epoch": 1.8402373749127299, "grad_norm": 0.9129747748374939, "learning_rate": 0.0001725719092792772, "loss": 1.4394323825836182, "step": 15815 }, { "epoch": 1.840353735164068, "grad_norm": 1.0777157545089722, "learning_rate": 0.00017235466485271203, "loss": 2.0050461292266846, "step": 15816 }, { "epoch": 1.840470095415406, "grad_norm": 0.8769164681434631, "learning_rate": 0.00017213748529795758, "loss": 1.386112928390503, "step": 15817 }, { "epoch": 1.8405864556667444, "grad_norm": 1.000267505645752, "learning_rate": 0.0001719203707964649, "loss": 1.5136723518371582, "step": 15818 }, { "epoch": 1.8407028159180823, "grad_norm": 1.152040719985962, "learning_rate": 0.00017170332152963063, "loss": 2.210404396057129, "step": 15819 }, { "epoch": 1.8408191761694206, "grad_norm": 1.1000412702560425, "learning_rate": 0.00017148633767879713, "loss": 2.09006667137146, "step": 15820 }, { "epoch": 1.8409355364207587, "grad_norm": 1.1149024963378906, "learning_rate": 0.00017126941942525192, "loss": 2.1387853622436523, "step": 15821 }, { "epoch": 1.8410518966720968, "grad_norm": 1.1815881729125977, "learning_rate": 0.0001710525669502278, "loss": 1.7070882320404053, "step": 15822 }, { "epoch": 1.841168256923435, "grad_norm": 0.9201821684837341, "learning_rate": 0.00017083578043490253, "loss": 2.1331636905670166, "step": 15823 }, { "epoch": 1.841284617174773, "grad_norm": 1.1085890531539917, "learning_rate": 0.00017061906006039893, "loss": 1.8945257663726807, "step": 15824 }, { "epoch": 1.8414009774261113, "grad_norm": 0.9831194281578064, "learning_rate": 0.00017040240600778439, "loss": 2.06282901763916, "step": 15825 }, { "epoch": 1.8415173376774494, "grad_norm": 1.014082908630371, "learning_rate": 0.000170185818458071, "loss": 1.921278715133667, "step": 15826 }, { "epoch": 1.8416336979287875, "grad_norm": 0.9351159334182739, "learning_rate": 0.00016996929759221518, "loss": 1.5431413650512695, "step": 15827 }, { "epoch": 1.8417500581801258, "grad_norm": 1.0466843843460083, "learning_rate": 0.00016975284359111776, "loss": 2.307802200317383, "step": 15828 }, { "epoch": 1.8418664184314637, "grad_norm": 1.488789439201355, "learning_rate": 0.00016953645663562357, "loss": 1.3883118629455566, "step": 15829 }, { "epoch": 1.841982778682802, "grad_norm": 1.3900848627090454, "learning_rate": 0.00016932013690652148, "loss": 2.250232696533203, "step": 15830 }, { "epoch": 1.8420991389341401, "grad_norm": 1.8739796876907349, "learning_rate": 0.0001691038845845443, "loss": 2.534195899963379, "step": 15831 }, { "epoch": 1.8422154991854782, "grad_norm": 0.38710206747055054, "learning_rate": 0.0001688876998503685, "loss": 1.0092030763626099, "step": 15832 }, { "epoch": 1.8423318594368165, "grad_norm": 1.4407153129577637, "learning_rate": 0.00016867158288461377, "loss": 2.121403932571411, "step": 15833 }, { "epoch": 1.8424482196881544, "grad_norm": 0.5405955910682678, "learning_rate": 0.00016845553386784351, "loss": 1.0567437410354614, "step": 15834 }, { "epoch": 1.8425645799394927, "grad_norm": 1.1722257137298584, "learning_rate": 0.00016823955298056437, "loss": 1.7045037746429443, "step": 15835 }, { "epoch": 1.8426809401908308, "grad_norm": 1.335649847984314, "learning_rate": 0.00016802364040322592, "loss": 2.4065332412719727, "step": 15836 }, { "epoch": 1.842797300442169, "grad_norm": 1.4870789051055908, "learning_rate": 0.00016780779631622066, "loss": 1.729196548461914, "step": 15837 }, { "epoch": 1.8429136606935073, "grad_norm": 0.6727168560028076, "learning_rate": 0.00016759202089988395, "loss": 1.4312047958374023, "step": 15838 }, { "epoch": 1.8430300209448451, "grad_norm": 1.0432144403457642, "learning_rate": 0.00016737631433449374, "loss": 1.9186948537826538, "step": 15839 }, { "epoch": 1.8431463811961835, "grad_norm": 1.3145318031311035, "learning_rate": 0.00016716067680027047, "loss": 1.7236745357513428, "step": 15840 }, { "epoch": 1.8432627414475216, "grad_norm": 0.7733906507492065, "learning_rate": 0.00016694510847737682, "loss": 1.425971269607544, "step": 15841 }, { "epoch": 1.8433791016988597, "grad_norm": 0.7985677719116211, "learning_rate": 0.0001667296095459178, "loss": 1.9278470277786255, "step": 15842 }, { "epoch": 1.843495461950198, "grad_norm": 1.0985071659088135, "learning_rate": 0.00016651418018594034, "loss": 1.6418137550354004, "step": 15843 }, { "epoch": 1.8436118222015359, "grad_norm": 1.0905588865280151, "learning_rate": 0.00016629882057743329, "loss": 2.278682231903076, "step": 15844 }, { "epoch": 1.8437281824528742, "grad_norm": 0.9258432388305664, "learning_rate": 0.00016608353090032712, "loss": 1.7991621494293213, "step": 15845 }, { "epoch": 1.8438445427042123, "grad_norm": 1.518871545791626, "learning_rate": 0.00016586831133449402, "loss": 1.8676083087921143, "step": 15846 }, { "epoch": 1.8439609029555504, "grad_norm": 1.3280060291290283, "learning_rate": 0.00016565316205974752, "loss": 2.1776559352874756, "step": 15847 }, { "epoch": 1.8440772632068885, "grad_norm": 0.7555177807807922, "learning_rate": 0.00016543808325584236, "loss": 1.561964750289917, "step": 15848 }, { "epoch": 1.8441936234582266, "grad_norm": 1.143213152885437, "learning_rate": 0.00016522307510247463, "loss": 2.102534294128418, "step": 15849 }, { "epoch": 1.844309983709565, "grad_norm": 1.2609435319900513, "learning_rate": 0.00016500813777928114, "loss": 2.1305665969848633, "step": 15850 }, { "epoch": 1.844426343960903, "grad_norm": 1.183366060256958, "learning_rate": 0.00016479327146583965, "loss": 2.011660099029541, "step": 15851 }, { "epoch": 1.844542704212241, "grad_norm": 0.979598879814148, "learning_rate": 0.0001645784763416686, "loss": 2.6683716773986816, "step": 15852 }, { "epoch": 1.8446590644635792, "grad_norm": 1.4223021268844604, "learning_rate": 0.00016436375258622684, "loss": 1.4956862926483154, "step": 15853 }, { "epoch": 1.8447754247149173, "grad_norm": 0.845576286315918, "learning_rate": 0.00016414910037891372, "loss": 1.9420554637908936, "step": 15854 }, { "epoch": 1.8448917849662556, "grad_norm": 1.0032440423965454, "learning_rate": 0.00016393451989906872, "loss": 1.904831886291504, "step": 15855 }, { "epoch": 1.8450081452175937, "grad_norm": 0.9707184433937073, "learning_rate": 0.00016372001132597158, "loss": 1.6183232069015503, "step": 15856 }, { "epoch": 1.8451245054689318, "grad_norm": 1.0127060413360596, "learning_rate": 0.00016350557483884166, "loss": 1.6637060642242432, "step": 15857 }, { "epoch": 1.84524086572027, "grad_norm": 1.0234469175338745, "learning_rate": 0.00016329121061683827, "loss": 2.422224521636963, "step": 15858 }, { "epoch": 1.845357225971608, "grad_norm": 1.3688467741012573, "learning_rate": 0.0001630769188390603, "loss": 2.18595552444458, "step": 15859 }, { "epoch": 1.8454735862229463, "grad_norm": 0.8898637890815735, "learning_rate": 0.00016286269968454622, "loss": 1.3193492889404297, "step": 15860 }, { "epoch": 1.8455899464742844, "grad_norm": 1.2424821853637695, "learning_rate": 0.00016264855333227374, "loss": 1.473091959953308, "step": 15861 }, { "epoch": 1.8457063067256225, "grad_norm": 1.2561347484588623, "learning_rate": 0.0001624344799611597, "loss": 2.0171937942504883, "step": 15862 }, { "epoch": 1.8458226669769606, "grad_norm": 0.8969172835350037, "learning_rate": 0.00016222047975006005, "loss": 2.1164379119873047, "step": 15863 }, { "epoch": 1.8459390272282987, "grad_norm": 0.6525982022285461, "learning_rate": 0.00016200655287776955, "loss": 1.447322130203247, "step": 15864 }, { "epoch": 1.846055387479637, "grad_norm": 0.9540758728981018, "learning_rate": 0.0001617926995230217, "loss": 2.5434017181396484, "step": 15865 }, { "epoch": 1.8461717477309751, "grad_norm": 1.3769760131835938, "learning_rate": 0.0001615789198644886, "loss": 1.6931097507476807, "step": 15866 }, { "epoch": 1.8462881079823132, "grad_norm": 1.3579820394515991, "learning_rate": 0.00016136521408078078, "loss": 1.8822435140609741, "step": 15867 }, { "epoch": 1.8464044682336513, "grad_norm": 1.03115713596344, "learning_rate": 0.00016115158235044706, "loss": 1.508441686630249, "step": 15868 }, { "epoch": 1.8465208284849894, "grad_norm": 1.3311430215835571, "learning_rate": 0.00016093802485197433, "loss": 2.2558279037475586, "step": 15869 }, { "epoch": 1.8466371887363278, "grad_norm": 1.3583964109420776, "learning_rate": 0.00016072454176378753, "loss": 1.3982000350952148, "step": 15870 }, { "epoch": 1.8467535489876659, "grad_norm": 1.1444675922393799, "learning_rate": 0.00016051113326424934, "loss": 1.6341917514801025, "step": 15871 }, { "epoch": 1.846869909239004, "grad_norm": 0.5113649964332581, "learning_rate": 0.00016029779953166023, "loss": 1.4094078540802002, "step": 15872 }, { "epoch": 1.846986269490342, "grad_norm": 1.3230986595153809, "learning_rate": 0.0001600845407442581, "loss": 1.8380130529403687, "step": 15873 }, { "epoch": 1.8471026297416802, "grad_norm": 0.6844573616981506, "learning_rate": 0.00015987135708021833, "loss": 1.648706316947937, "step": 15874 }, { "epoch": 1.8472189899930185, "grad_norm": 1.2434749603271484, "learning_rate": 0.0001596582487176535, "loss": 1.8596621751785278, "step": 15875 }, { "epoch": 1.8473353502443566, "grad_norm": 1.6581275463104248, "learning_rate": 0.00015944521583461323, "loss": 1.9999319314956665, "step": 15876 }, { "epoch": 1.8474517104956947, "grad_norm": 0.8206393718719482, "learning_rate": 0.0001592322586090842, "loss": 1.5470378398895264, "step": 15877 }, { "epoch": 1.8475680707470328, "grad_norm": 0.5032400488853455, "learning_rate": 0.00015901937721898968, "loss": 1.4282289743423462, "step": 15878 }, { "epoch": 1.8476844309983709, "grad_norm": 1.0585194826126099, "learning_rate": 0.00015880657184218973, "loss": 1.9583770036697388, "step": 15879 }, { "epoch": 1.8478007912497092, "grad_norm": 1.0243444442749023, "learning_rate": 0.00015859384265648097, "loss": 2.0502712726593018, "step": 15880 }, { "epoch": 1.8479171515010473, "grad_norm": 0.8458776473999023, "learning_rate": 0.00015838118983959608, "loss": 1.687575340270996, "step": 15881 }, { "epoch": 1.8480335117523854, "grad_norm": 1.0129362344741821, "learning_rate": 0.00015816861356920421, "loss": 2.475212812423706, "step": 15882 }, { "epoch": 1.8481498720037235, "grad_norm": 0.729197084903717, "learning_rate": 0.00015795611402291047, "loss": 1.4584541320800781, "step": 15883 }, { "epoch": 1.8482662322550616, "grad_norm": 0.4234781563282013, "learning_rate": 0.0001577436913782558, "loss": 1.0395610332489014, "step": 15884 }, { "epoch": 1.8483825925064, "grad_norm": 0.2714191675186157, "learning_rate": 0.000157531345812717, "loss": 0.8981874585151672, "step": 15885 }, { "epoch": 1.848498952757738, "grad_norm": 1.140747308731079, "learning_rate": 0.00015731907750370644, "loss": 1.6823415756225586, "step": 15886 }, { "epoch": 1.848615313009076, "grad_norm": 0.9774186015129089, "learning_rate": 0.00015710688662857187, "loss": 2.2698302268981934, "step": 15887 }, { "epoch": 1.8487316732604142, "grad_norm": 0.8665173649787903, "learning_rate": 0.00015689477336459646, "loss": 1.9048813581466675, "step": 15888 }, { "epoch": 1.8488480335117523, "grad_norm": 0.9342904686927795, "learning_rate": 0.00015668273788899839, "loss": 1.7651395797729492, "step": 15889 }, { "epoch": 1.8489643937630906, "grad_norm": 0.7765902876853943, "learning_rate": 0.00015647078037893102, "loss": 1.7807308435440063, "step": 15890 }, { "epoch": 1.8490807540144285, "grad_norm": 1.0569080114364624, "learning_rate": 0.00015625890101148232, "loss": 1.380908489227295, "step": 15891 }, { "epoch": 1.8491971142657668, "grad_norm": 0.9491564631462097, "learning_rate": 0.0001560470999636754, "loss": 1.462491750717163, "step": 15892 }, { "epoch": 1.849313474517105, "grad_norm": 0.5560059547424316, "learning_rate": 0.00015583537741246744, "loss": 1.1474697589874268, "step": 15893 }, { "epoch": 1.849429834768443, "grad_norm": 0.7296793460845947, "learning_rate": 0.0001556237335347504, "loss": 1.688453197479248, "step": 15894 }, { "epoch": 1.8495461950197813, "grad_norm": 1.39572012424469, "learning_rate": 0.00015541216850735028, "loss": 1.7686762809753418, "step": 15895 }, { "epoch": 1.8496625552711192, "grad_norm": 0.3726115822792053, "learning_rate": 0.0001552006825070274, "loss": 0.9314337372779846, "step": 15896 }, { "epoch": 1.8497789155224575, "grad_norm": 1.1202025413513184, "learning_rate": 0.0001549892757104759, "loss": 2.338862180709839, "step": 15897 }, { "epoch": 1.8498952757737956, "grad_norm": 1.2831406593322754, "learning_rate": 0.00015477794829432372, "loss": 1.7346254587173462, "step": 15898 }, { "epoch": 1.8500116360251337, "grad_norm": 1.1532539129257202, "learning_rate": 0.00015456670043513272, "loss": 1.650931477546692, "step": 15899 }, { "epoch": 1.850127996276472, "grad_norm": 0.7521829009056091, "learning_rate": 0.00015435553230939804, "loss": 1.4067175388336182, "step": 15900 }, { "epoch": 1.85024435652781, "grad_norm": 1.1097962856292725, "learning_rate": 0.00015414444409354828, "loss": 1.9708099365234375, "step": 15901 }, { "epoch": 1.8503607167791483, "grad_norm": 1.4624202251434326, "learning_rate": 0.0001539334359639453, "loss": 2.667356014251709, "step": 15902 }, { "epoch": 1.8504770770304864, "grad_norm": 0.635081946849823, "learning_rate": 0.00015372250809688403, "loss": 1.5196430683135986, "step": 15903 }, { "epoch": 1.8505934372818245, "grad_norm": 0.6830258965492249, "learning_rate": 0.00015351166066859243, "loss": 1.40812087059021, "step": 15904 }, { "epoch": 1.8507097975331628, "grad_norm": 0.8140493631362915, "learning_rate": 0.00015330089385523106, "loss": 1.199471354484558, "step": 15905 }, { "epoch": 1.8508261577845007, "grad_norm": 0.8537254333496094, "learning_rate": 0.00015309020783289324, "loss": 1.658390760421753, "step": 15906 }, { "epoch": 1.850942518035839, "grad_norm": 2.2425286769866943, "learning_rate": 0.00015287960277760487, "loss": 2.2809081077575684, "step": 15907 }, { "epoch": 1.851058878287177, "grad_norm": 0.8993375301361084, "learning_rate": 0.00015266907886532405, "loss": 1.7311294078826904, "step": 15908 }, { "epoch": 1.8511752385385152, "grad_norm": 0.8662264943122864, "learning_rate": 0.00015245863627194115, "loss": 2.084444522857666, "step": 15909 }, { "epoch": 1.8512915987898535, "grad_norm": 1.9810124635696411, "learning_rate": 0.0001522482751732787, "loss": 2.2719566822052, "step": 15910 }, { "epoch": 1.8514079590411914, "grad_norm": 0.7855046391487122, "learning_rate": 0.000152037995745091, "loss": 1.7541807889938354, "step": 15911 }, { "epoch": 1.8515243192925297, "grad_norm": 1.686707854270935, "learning_rate": 0.00015182779816306415, "loss": 1.581425428390503, "step": 15912 }, { "epoch": 1.8516406795438678, "grad_norm": 1.0392431020736694, "learning_rate": 0.0001516176826028159, "loss": 2.072873592376709, "step": 15913 }, { "epoch": 1.851757039795206, "grad_norm": 1.2965900897979736, "learning_rate": 0.00015140764923989546, "loss": 1.6152703762054443, "step": 15914 }, { "epoch": 1.8518734000465442, "grad_norm": 0.9279915690422058, "learning_rate": 0.00015119769824978332, "loss": 1.9138662815093994, "step": 15915 }, { "epoch": 1.851989760297882, "grad_norm": 0.7570846080780029, "learning_rate": 0.00015098782980789118, "loss": 2.249929904937744, "step": 15916 }, { "epoch": 1.8521061205492204, "grad_norm": 1.5202821493148804, "learning_rate": 0.00015077804408956185, "loss": 1.8373456001281738, "step": 15917 }, { "epoch": 1.8522224808005585, "grad_norm": 1.0892137289047241, "learning_rate": 0.00015056834127006897, "loss": 1.3599498271942139, "step": 15918 }, { "epoch": 1.8523388410518966, "grad_norm": 1.2240607738494873, "learning_rate": 0.0001503587215246168, "loss": 1.8534307479858398, "step": 15919 }, { "epoch": 1.852455201303235, "grad_norm": 0.7957481741905212, "learning_rate": 0.00015014918502834042, "loss": 1.8712191581726074, "step": 15920 }, { "epoch": 1.8525715615545728, "grad_norm": 0.8500834703445435, "learning_rate": 0.0001499397319563051, "loss": 1.6637827157974243, "step": 15921 }, { "epoch": 1.8526879218059111, "grad_norm": 0.2822958827018738, "learning_rate": 0.00014973036248350666, "loss": 0.8750965595245361, "step": 15922 }, { "epoch": 1.8528042820572492, "grad_norm": 1.058579921722412, "learning_rate": 0.00014952107678487084, "loss": 2.317436933517456, "step": 15923 }, { "epoch": 1.8529206423085873, "grad_norm": 1.3092349767684937, "learning_rate": 0.00014931187503525363, "loss": 2.0763368606567383, "step": 15924 }, { "epoch": 1.8530370025599257, "grad_norm": 0.6338416934013367, "learning_rate": 0.0001491027574094407, "loss": 1.7034097909927368, "step": 15925 }, { "epoch": 1.8531533628112635, "grad_norm": 0.9104360342025757, "learning_rate": 0.00014889372408214747, "loss": 1.456587791442871, "step": 15926 }, { "epoch": 1.8532697230626018, "grad_norm": 0.8244353532791138, "learning_rate": 0.00014868477522801896, "loss": 1.5807390213012695, "step": 15927 }, { "epoch": 1.85338608331394, "grad_norm": 0.9529459476470947, "learning_rate": 0.00014847591102162968, "loss": 1.7123680114746094, "step": 15928 }, { "epoch": 1.853502443565278, "grad_norm": 0.4327758550643921, "learning_rate": 0.00014826713163748324, "loss": 0.9836347103118896, "step": 15929 }, { "epoch": 1.8536188038166164, "grad_norm": 0.7874241471290588, "learning_rate": 0.0001480584372500125, "loss": 1.8228020668029785, "step": 15930 }, { "epoch": 1.8537351640679542, "grad_norm": 1.1717994213104248, "learning_rate": 0.00014784982803357934, "loss": 1.880483865737915, "step": 15931 }, { "epoch": 1.8538515243192926, "grad_norm": 1.1896417140960693, "learning_rate": 0.0001476413041624744, "loss": 2.3123388290405273, "step": 15932 }, { "epoch": 1.8539678845706307, "grad_norm": 0.9165055751800537, "learning_rate": 0.00014743286581091701, "loss": 1.585191011428833, "step": 15933 }, { "epoch": 1.8540842448219688, "grad_norm": 1.103488802909851, "learning_rate": 0.00014722451315305512, "loss": 2.2208328247070312, "step": 15934 }, { "epoch": 1.854200605073307, "grad_norm": 1.0128101110458374, "learning_rate": 0.00014701624636296513, "loss": 1.8083431720733643, "step": 15935 }, { "epoch": 1.854316965324645, "grad_norm": 2.085425853729248, "learning_rate": 0.00014680806561465152, "loss": 1.9889832735061646, "step": 15936 }, { "epoch": 1.8544333255759833, "grad_norm": 1.5671175718307495, "learning_rate": 0.000146599971082047, "loss": 1.9116270542144775, "step": 15937 }, { "epoch": 1.8545496858273214, "grad_norm": 1.3707737922668457, "learning_rate": 0.00014639196293901222, "loss": 1.7342798709869385, "step": 15938 }, { "epoch": 1.8546660460786595, "grad_norm": 1.552530288696289, "learning_rate": 0.00014618404135933568, "loss": 1.6254158020019531, "step": 15939 }, { "epoch": 1.8547824063299978, "grad_norm": 0.8172436952590942, "learning_rate": 0.00014597620651673354, "loss": 1.8080955743789673, "step": 15940 }, { "epoch": 1.8548987665813357, "grad_norm": 0.9197009205818176, "learning_rate": 0.00014576845858484943, "loss": 1.5574856996536255, "step": 15941 }, { "epoch": 1.855015126832674, "grad_norm": 0.4771321415901184, "learning_rate": 0.00014556079773725455, "loss": 1.0205445289611816, "step": 15942 }, { "epoch": 1.855131487084012, "grad_norm": 1.5355794429779053, "learning_rate": 0.00014535322414744716, "loss": 2.3827195167541504, "step": 15943 }, { "epoch": 1.8552478473353502, "grad_norm": 1.210343599319458, "learning_rate": 0.00014514573798885268, "loss": 1.818151593208313, "step": 15944 }, { "epoch": 1.8553642075866885, "grad_norm": 1.3696012496948242, "learning_rate": 0.0001449383394348234, "loss": 2.607464075088501, "step": 15945 }, { "epoch": 1.8554805678380264, "grad_norm": 0.9711229801177979, "learning_rate": 0.0001447310286586386, "loss": 1.2583240270614624, "step": 15946 }, { "epoch": 1.8555969280893647, "grad_norm": 1.2363077402114868, "learning_rate": 0.00014452380583350412, "loss": 2.1728367805480957, "step": 15947 }, { "epoch": 1.8557132883407028, "grad_norm": 1.3133273124694824, "learning_rate": 0.0001443166711325522, "loss": 1.635138750076294, "step": 15948 }, { "epoch": 1.855829648592041, "grad_norm": 1.0160086154937744, "learning_rate": 0.0001441096247288417, "loss": 1.8700488805770874, "step": 15949 }, { "epoch": 1.8559460088433792, "grad_norm": 1.3730071783065796, "learning_rate": 0.00014390266679535751, "loss": 2.146822452545166, "step": 15950 }, { "epoch": 1.8560623690947171, "grad_norm": 0.8822199702262878, "learning_rate": 0.0001436957975050107, "loss": 1.6424450874328613, "step": 15951 }, { "epoch": 1.8561787293460554, "grad_norm": 1.2818297147750854, "learning_rate": 0.00014348901703063827, "loss": 1.7516814470291138, "step": 15952 }, { "epoch": 1.8562950895973935, "grad_norm": 0.9439616203308105, "learning_rate": 0.00014328232554500298, "loss": 1.7744472026824951, "step": 15953 }, { "epoch": 1.8564114498487316, "grad_norm": 1.0195609331130981, "learning_rate": 0.0001430757232207932, "loss": 1.6437208652496338, "step": 15954 }, { "epoch": 1.85652781010007, "grad_norm": 0.49622994661331177, "learning_rate": 0.00014286921023062293, "loss": 1.1955220699310303, "step": 15955 }, { "epoch": 1.8566441703514078, "grad_norm": 1.378288984298706, "learning_rate": 0.00014266278674703148, "loss": 1.87178635597229, "step": 15956 }, { "epoch": 1.8567605306027462, "grad_norm": 0.7713316082954407, "learning_rate": 0.00014245645294248333, "loss": 1.7180534601211548, "step": 15957 }, { "epoch": 1.8568768908540842, "grad_norm": 1.422442078590393, "learning_rate": 0.00014225020898936806, "loss": 1.7399628162384033, "step": 15958 }, { "epoch": 1.8569932511054223, "grad_norm": 1.196339726448059, "learning_rate": 0.00014204405506000023, "loss": 2.4373416900634766, "step": 15959 }, { "epoch": 1.8571096113567607, "grad_norm": 1.3467001914978027, "learning_rate": 0.00014183799132661912, "loss": 2.1113076210021973, "step": 15960 }, { "epoch": 1.8572259716080985, "grad_norm": 0.7033156156539917, "learning_rate": 0.00014163201796138866, "loss": 1.5783402919769287, "step": 15961 }, { "epoch": 1.8573423318594369, "grad_norm": 1.1120198965072632, "learning_rate": 0.00014142613513639735, "loss": 2.029107093811035, "step": 15962 }, { "epoch": 1.857458692110775, "grad_norm": 1.660813331604004, "learning_rate": 0.00014122034302365785, "loss": 1.8258771896362305, "step": 15963 }, { "epoch": 1.857575052362113, "grad_norm": 1.2429025173187256, "learning_rate": 0.0001410146417951074, "loss": 2.0458433628082275, "step": 15964 }, { "epoch": 1.8576914126134514, "grad_norm": 1.0212582349777222, "learning_rate": 0.00014080903162260694, "loss": 2.1660962104797363, "step": 15965 }, { "epoch": 1.8578077728647893, "grad_norm": 1.2467360496520996, "learning_rate": 0.00014060351267794152, "loss": 1.5304768085479736, "step": 15966 }, { "epoch": 1.8579241331161276, "grad_norm": 0.6044280529022217, "learning_rate": 0.00014039808513281992, "loss": 1.306502103805542, "step": 15967 }, { "epoch": 1.8580404933674657, "grad_norm": 1.2874181270599365, "learning_rate": 0.00014019274915887453, "loss": 1.7381904125213623, "step": 15968 }, { "epoch": 1.8581568536188038, "grad_norm": 0.6686610579490662, "learning_rate": 0.00013998750492766126, "loss": 1.274807095527649, "step": 15969 }, { "epoch": 1.858273213870142, "grad_norm": 1.28663170337677, "learning_rate": 0.00013978235261065942, "loss": 1.5047035217285156, "step": 15970 }, { "epoch": 1.85838957412148, "grad_norm": 1.1284881830215454, "learning_rate": 0.00013957729237927142, "loss": 2.260755777359009, "step": 15971 }, { "epoch": 1.8585059343728183, "grad_norm": 0.9212063550949097, "learning_rate": 0.00013937232440482285, "loss": 1.432931900024414, "step": 15972 }, { "epoch": 1.8586222946241564, "grad_norm": 1.0810275077819824, "learning_rate": 0.00013916744885856215, "loss": 1.295366644859314, "step": 15973 }, { "epoch": 1.8587386548754945, "grad_norm": 0.7938243746757507, "learning_rate": 0.00013896266591166052, "loss": 1.483542561531067, "step": 15974 }, { "epoch": 1.8588550151268328, "grad_norm": 1.0210610628128052, "learning_rate": 0.00013875797573521185, "loss": 2.6244938373565674, "step": 15975 }, { "epoch": 1.8589713753781707, "grad_norm": 0.9077315926551819, "learning_rate": 0.00013855337850023249, "loss": 1.6653343439102173, "step": 15976 }, { "epoch": 1.859087735629509, "grad_norm": 0.5409520864486694, "learning_rate": 0.00013834887437766115, "loss": 1.5040454864501953, "step": 15977 }, { "epoch": 1.8592040958808471, "grad_norm": 0.7237740755081177, "learning_rate": 0.00013814446353835875, "loss": 1.4473292827606201, "step": 15978 }, { "epoch": 1.8593204561321852, "grad_norm": 0.947721004486084, "learning_rate": 0.00013794014615310825, "loss": 1.7667170763015747, "step": 15979 }, { "epoch": 1.8594368163835235, "grad_norm": 1.3322240114212036, "learning_rate": 0.00013773592239261457, "loss": 1.655660629272461, "step": 15980 }, { "epoch": 1.8595531766348614, "grad_norm": 1.0164761543273926, "learning_rate": 0.00013753179242750435, "loss": 1.609950304031372, "step": 15981 }, { "epoch": 1.8596695368861997, "grad_norm": 1.0472475290298462, "learning_rate": 0.0001373277564283259, "loss": 1.7171646356582642, "step": 15982 }, { "epoch": 1.8597858971375378, "grad_norm": 0.9767851233482361, "learning_rate": 0.000137123814565549, "loss": 1.6533186435699463, "step": 15983 }, { "epoch": 1.859902257388876, "grad_norm": 0.9199331402778625, "learning_rate": 0.0001369199670095649, "loss": 1.6474863290786743, "step": 15984 }, { "epoch": 1.8600186176402143, "grad_norm": 0.9918412566184998, "learning_rate": 0.00013671621393068585, "loss": 1.9354519844055176, "step": 15985 }, { "epoch": 1.8601349778915521, "grad_norm": 1.1562438011169434, "learning_rate": 0.00013651255549914526, "loss": 1.9746105670928955, "step": 15986 }, { "epoch": 1.8602513381428905, "grad_norm": 1.0819401741027832, "learning_rate": 0.0001363089918850976, "loss": 1.755056381225586, "step": 15987 }, { "epoch": 1.8603676983942286, "grad_norm": 1.5848695039749146, "learning_rate": 0.00013610552325861775, "loss": 2.1077332496643066, "step": 15988 }, { "epoch": 1.8604840586455667, "grad_norm": 0.7559387683868408, "learning_rate": 0.00013590214978970177, "loss": 1.9928648471832275, "step": 15989 }, { "epoch": 1.860600418896905, "grad_norm": 1.2414791584014893, "learning_rate": 0.00013569887164826584, "loss": 2.5698678493499756, "step": 15990 }, { "epoch": 1.8607167791482428, "grad_norm": 0.7677814364433289, "learning_rate": 0.0001354956890041465, "loss": 1.4878431558609009, "step": 15991 }, { "epoch": 1.8608331393995812, "grad_norm": 1.0334521532058716, "learning_rate": 0.0001352926020271006, "loss": 1.3726083040237427, "step": 15992 }, { "epoch": 1.8609494996509193, "grad_norm": 0.8960774540901184, "learning_rate": 0.0001350896108868051, "loss": 1.8736751079559326, "step": 15993 }, { "epoch": 1.8610658599022574, "grad_norm": 1.0778125524520874, "learning_rate": 0.00013488671575285677, "loss": 1.462033748626709, "step": 15994 }, { "epoch": 1.8611822201535957, "grad_norm": 1.29225492477417, "learning_rate": 0.00013468391679477226, "loss": 2.1777563095092773, "step": 15995 }, { "epoch": 1.8612985804049336, "grad_norm": 1.0426099300384521, "learning_rate": 0.0001344812141819878, "loss": 1.9702056646347046, "step": 15996 }, { "epoch": 1.8614149406562719, "grad_norm": 0.8408094644546509, "learning_rate": 0.0001342786080838592, "loss": 2.0548863410949707, "step": 15997 }, { "epoch": 1.86153130090761, "grad_norm": 1.2236093282699585, "learning_rate": 0.0001340760986696616, "loss": 2.004204273223877, "step": 15998 }, { "epoch": 1.861647661158948, "grad_norm": 1.8345801830291748, "learning_rate": 0.0001338736861085893, "loss": 1.943060278892517, "step": 15999 }, { "epoch": 1.8617640214102864, "grad_norm": 0.6748377680778503, "learning_rate": 0.00013367137056975576, "loss": 1.3790725469589233, "step": 16000 }, { "epoch": 1.8618803816616243, "grad_norm": 1.2187095880508423, "learning_rate": 0.00013346915222219337, "loss": 1.9646931886672974, "step": 16001 }, { "epoch": 1.8619967419129626, "grad_norm": 1.4167011976242065, "learning_rate": 0.00013326703123485328, "loss": 2.4193172454833984, "step": 16002 }, { "epoch": 1.8621131021643007, "grad_norm": 1.6756150722503662, "learning_rate": 0.00013306500777660537, "loss": 1.8665850162506104, "step": 16003 }, { "epoch": 1.8622294624156388, "grad_norm": 1.2800812721252441, "learning_rate": 0.00013286308201623792, "loss": 2.0889923572540283, "step": 16004 }, { "epoch": 1.8623458226669771, "grad_norm": 1.3273156881332397, "learning_rate": 0.0001326612541224577, "loss": 1.5342713594436646, "step": 16005 }, { "epoch": 1.862462182918315, "grad_norm": 1.3524231910705566, "learning_rate": 0.0001324595242638896, "loss": 2.6950833797454834, "step": 16006 }, { "epoch": 1.8625785431696533, "grad_norm": 0.7781963348388672, "learning_rate": 0.00013225789260907673, "loss": 1.1153013706207275, "step": 16007 }, { "epoch": 1.8626949034209914, "grad_norm": 1.9280873537063599, "learning_rate": 0.00013205635932648003, "loss": 1.4520883560180664, "step": 16008 }, { "epoch": 1.8628112636723295, "grad_norm": 1.0520403385162354, "learning_rate": 0.00013185492458447833, "loss": 1.9061164855957031, "step": 16009 }, { "epoch": 1.8629276239236678, "grad_norm": 1.36078941822052, "learning_rate": 0.00013165358855136814, "loss": 2.3069891929626465, "step": 16010 }, { "epoch": 1.8630439841750057, "grad_norm": 1.020318865776062, "learning_rate": 0.00013145235139536344, "loss": 2.6902990341186523, "step": 16011 }, { "epoch": 1.863160344426344, "grad_norm": 1.3501449823379517, "learning_rate": 0.0001312512132845956, "loss": 2.0396265983581543, "step": 16012 }, { "epoch": 1.8632767046776821, "grad_norm": 1.285258412361145, "learning_rate": 0.0001310501743871132, "loss": 1.907404899597168, "step": 16013 }, { "epoch": 1.8633930649290202, "grad_norm": 0.8520882725715637, "learning_rate": 0.0001308492348708822, "loss": 1.1208516359329224, "step": 16014 }, { "epoch": 1.8635094251803583, "grad_norm": 0.7595049142837524, "learning_rate": 0.0001306483949037852, "loss": 1.5753059387207031, "step": 16015 }, { "epoch": 1.8636257854316964, "grad_norm": 1.2457677125930786, "learning_rate": 0.0001304476546536218, "loss": 1.80134916305542, "step": 16016 }, { "epoch": 1.8637421456830348, "grad_norm": 0.9023498892784119, "learning_rate": 0.00013024701428810813, "loss": 1.4456937313079834, "step": 16017 }, { "epoch": 1.8638585059343729, "grad_norm": 0.6935563683509827, "learning_rate": 0.00013004647397487707, "loss": 1.526135802268982, "step": 16018 }, { "epoch": 1.863974866185711, "grad_norm": 1.8224263191223145, "learning_rate": 0.0001298460338814778, "loss": 1.6712803840637207, "step": 16019 }, { "epoch": 1.864091226437049, "grad_norm": 1.1487900018692017, "learning_rate": 0.00012964569417537575, "loss": 2.792011260986328, "step": 16020 }, { "epoch": 1.8642075866883872, "grad_norm": 0.8114936351776123, "learning_rate": 0.00012944545502395255, "loss": 1.6448299884796143, "step": 16021 }, { "epoch": 1.8643239469397255, "grad_norm": 0.9489263296127319, "learning_rate": 0.0001292453165945057, "loss": 1.7814040184020996, "step": 16022 }, { "epoch": 1.8644403071910636, "grad_norm": 0.6410452127456665, "learning_rate": 0.00012904527905424873, "loss": 1.2737081050872803, "step": 16023 }, { "epoch": 1.8645566674424017, "grad_norm": 1.2614833116531372, "learning_rate": 0.0001288453425703107, "loss": 2.2888383865356445, "step": 16024 }, { "epoch": 1.8646730276937398, "grad_norm": 1.3550595045089722, "learning_rate": 0.0001286455073097363, "loss": 1.9358649253845215, "step": 16025 }, { "epoch": 1.8647893879450779, "grad_norm": 0.29276663064956665, "learning_rate": 0.00012844577343948572, "loss": 0.9445778131484985, "step": 16026 }, { "epoch": 1.8649057481964162, "grad_norm": 1.014399528503418, "learning_rate": 0.0001282461411264343, "loss": 2.0484509468078613, "step": 16027 }, { "epoch": 1.8650221084477543, "grad_norm": 0.8365572690963745, "learning_rate": 0.00012804661053737268, "loss": 1.8668617010116577, "step": 16028 }, { "epoch": 1.8651384686990924, "grad_norm": 0.9223511815071106, "learning_rate": 0.0001278471818390064, "loss": 1.4058459997177124, "step": 16029 }, { "epoch": 1.8652548289504305, "grad_norm": 1.1314741373062134, "learning_rate": 0.0001276478551979559, "loss": 2.3136215209960938, "step": 16030 }, { "epoch": 1.8653711892017686, "grad_norm": 1.3769958019256592, "learning_rate": 0.00012744863078075643, "loss": 2.0572032928466797, "step": 16031 }, { "epoch": 1.865487549453107, "grad_norm": 0.647129237651825, "learning_rate": 0.00012724950875385771, "loss": 1.599357008934021, "step": 16032 }, { "epoch": 1.865603909704445, "grad_norm": 1.0456502437591553, "learning_rate": 0.00012705048928362395, "loss": 1.5753958225250244, "step": 16033 }, { "epoch": 1.865720269955783, "grad_norm": 0.6823003888130188, "learning_rate": 0.00012685157253633378, "loss": 1.4273699522018433, "step": 16034 }, { "epoch": 1.8658366302071212, "grad_norm": 0.8659537434577942, "learning_rate": 0.00012665275867817982, "loss": 2.249957323074341, "step": 16035 }, { "epoch": 1.8659529904584593, "grad_norm": 0.381732314825058, "learning_rate": 0.0001264540478752689, "loss": 1.3224949836730957, "step": 16036 }, { "epoch": 1.8660693507097976, "grad_norm": 0.9908025860786438, "learning_rate": 0.00012625544029362166, "loss": 1.5652625560760498, "step": 16037 }, { "epoch": 1.8661857109611357, "grad_norm": 1.035204529762268, "learning_rate": 0.0001260569360991724, "loss": 1.592188835144043, "step": 16038 }, { "epoch": 1.8663020712124738, "grad_norm": 0.9092565774917603, "learning_rate": 0.00012585853545776936, "loss": 1.7654976844787598, "step": 16039 }, { "epoch": 1.866418431463812, "grad_norm": 1.3273770809173584, "learning_rate": 0.000125660238535174, "loss": 2.1406667232513428, "step": 16040 }, { "epoch": 1.86653479171515, "grad_norm": 1.527514100074768, "learning_rate": 0.00012546204549706113, "loss": 1.9686579704284668, "step": 16041 }, { "epoch": 1.8666511519664883, "grad_norm": 1.338069200515747, "learning_rate": 0.00012526395650901885, "loss": 2.4120872020721436, "step": 16042 }, { "epoch": 1.8667675122178264, "grad_norm": 0.8275120854377747, "learning_rate": 0.00012506597173654832, "loss": 1.7143561840057373, "step": 16043 }, { "epoch": 1.8668838724691645, "grad_norm": 0.6307836174964905, "learning_rate": 0.00012486809134506355, "loss": 1.5293041467666626, "step": 16044 }, { "epoch": 1.8670002327205026, "grad_norm": 0.8778743147850037, "learning_rate": 0.00012467031549989144, "loss": 1.7775750160217285, "step": 16045 }, { "epoch": 1.8671165929718407, "grad_norm": 0.9291895031929016, "learning_rate": 0.00012447264436627148, "loss": 1.8353862762451172, "step": 16046 }, { "epoch": 1.867232953223179, "grad_norm": 0.7307614088058472, "learning_rate": 0.00012427507810935573, "loss": 1.663550853729248, "step": 16047 }, { "epoch": 1.8673493134745172, "grad_norm": 1.205014705657959, "learning_rate": 0.00012407761689420862, "loss": 1.5820200443267822, "step": 16048 }, { "epoch": 1.8674656737258553, "grad_norm": 1.1161991357803345, "learning_rate": 0.0001238802608858066, "loss": 2.13364315032959, "step": 16049 }, { "epoch": 1.8675820339771934, "grad_norm": 1.0299897193908691, "learning_rate": 0.0001236830102490386, "loss": 2.142416477203369, "step": 16050 }, { "epoch": 1.8676983942285315, "grad_norm": 1.0519685745239258, "learning_rate": 0.0001234858651487053, "loss": 1.8566899299621582, "step": 16051 }, { "epoch": 1.8678147544798698, "grad_norm": 1.3163186311721802, "learning_rate": 0.0001232888257495192, "loss": 1.9433943033218384, "step": 16052 }, { "epoch": 1.8679311147312077, "grad_norm": 0.6279142498970032, "learning_rate": 0.00012309189221610452, "loss": 1.3686805963516235, "step": 16053 }, { "epoch": 1.868047474982546, "grad_norm": 1.51559579372406, "learning_rate": 0.00012289506471299704, "loss": 1.811558485031128, "step": 16054 }, { "epoch": 1.868163835233884, "grad_norm": 1.0571634769439697, "learning_rate": 0.0001226983434046439, "loss": 1.8733025789260864, "step": 16055 }, { "epoch": 1.8682801954852222, "grad_norm": 1.4062203168869019, "learning_rate": 0.00012250172845540365, "loss": 1.7747371196746826, "step": 16056 }, { "epoch": 1.8683965557365605, "grad_norm": 0.6760832667350769, "learning_rate": 0.0001223052200295458, "loss": 1.5903925895690918, "step": 16057 }, { "epoch": 1.8685129159878984, "grad_norm": 0.7960960865020752, "learning_rate": 0.00012210881829125095, "loss": 1.191112756729126, "step": 16058 }, { "epoch": 1.8686292762392367, "grad_norm": 1.1150963306427002, "learning_rate": 0.0001219125234046106, "loss": 1.7007821798324585, "step": 16059 }, { "epoch": 1.8687456364905748, "grad_norm": 1.8573365211486816, "learning_rate": 0.00012171633553362691, "loss": 2.155261993408203, "step": 16060 }, { "epoch": 1.868861996741913, "grad_norm": 1.0836702585220337, "learning_rate": 0.00012152025484221266, "loss": 2.227177381515503, "step": 16061 }, { "epoch": 1.8689783569932512, "grad_norm": 0.7020211219787598, "learning_rate": 0.00012132428149419105, "loss": 1.1449549198150635, "step": 16062 }, { "epoch": 1.869094717244589, "grad_norm": 1.0944592952728271, "learning_rate": 0.0001211284156532956, "loss": 1.682725191116333, "step": 16063 }, { "epoch": 1.8692110774959274, "grad_norm": 0.9884363412857056, "learning_rate": 0.00012093265748317018, "loss": 2.23669695854187, "step": 16064 }, { "epoch": 1.8693274377472655, "grad_norm": 0.9797013401985168, "learning_rate": 0.00012073700714736846, "loss": 2.300784111022949, "step": 16065 }, { "epoch": 1.8694437979986036, "grad_norm": 1.3034448623657227, "learning_rate": 0.0001205414648093541, "loss": 1.863398551940918, "step": 16066 }, { "epoch": 1.869560158249942, "grad_norm": 1.0729820728302002, "learning_rate": 0.00012034603063250058, "loss": 1.4223911762237549, "step": 16067 }, { "epoch": 1.8696765185012798, "grad_norm": 1.0493521690368652, "learning_rate": 0.00012015070478009093, "loss": 1.9080191850662231, "step": 16068 }, { "epoch": 1.8697928787526181, "grad_norm": 1.0598112344741821, "learning_rate": 0.00011995548741531776, "loss": 1.9620227813720703, "step": 16069 }, { "epoch": 1.8699092390039562, "grad_norm": 0.9239459037780762, "learning_rate": 0.00011976037870128296, "loss": 1.8637917041778564, "step": 16070 }, { "epoch": 1.8700255992552943, "grad_norm": 1.2111845016479492, "learning_rate": 0.00011956537880099769, "loss": 2.010892629623413, "step": 16071 }, { "epoch": 1.8701419595066326, "grad_norm": 0.8425478935241699, "learning_rate": 0.00011937048787738228, "loss": 1.8929204940795898, "step": 16072 }, { "epoch": 1.8702583197579705, "grad_norm": 1.018025279045105, "learning_rate": 0.00011917570609326578, "loss": 1.6907681226730347, "step": 16073 }, { "epoch": 1.8703746800093088, "grad_norm": 1.139681339263916, "learning_rate": 0.00011898103361138615, "loss": 2.108503580093384, "step": 16074 }, { "epoch": 1.870491040260647, "grad_norm": 0.7690143585205078, "learning_rate": 0.00011878647059439026, "loss": 1.4509785175323486, "step": 16075 }, { "epoch": 1.870607400511985, "grad_norm": 1.0294057130813599, "learning_rate": 0.00011859201720483325, "loss": 1.8271546363830566, "step": 16076 }, { "epoch": 1.8707237607633234, "grad_norm": 1.6428014039993286, "learning_rate": 0.00011839767360517881, "loss": 1.6913050413131714, "step": 16077 }, { "epoch": 1.8708401210146612, "grad_norm": 0.8835406303405762, "learning_rate": 0.00011820343995779876, "loss": 1.3131377696990967, "step": 16078 }, { "epoch": 1.8709564812659996, "grad_norm": 1.088951587677002, "learning_rate": 0.0001180093164249732, "loss": 2.0162925720214844, "step": 16079 }, { "epoch": 1.8710728415173377, "grad_norm": 0.9204639196395874, "learning_rate": 0.00011781530316889016, "loss": 1.996328592300415, "step": 16080 }, { "epoch": 1.8711892017686758, "grad_norm": 1.0462679862976074, "learning_rate": 0.00011762140035164556, "loss": 1.7361955642700195, "step": 16081 }, { "epoch": 1.871305562020014, "grad_norm": 1.2790982723236084, "learning_rate": 0.00011742760813524301, "loss": 2.0927209854125977, "step": 16082 }, { "epoch": 1.871421922271352, "grad_norm": 0.8423670530319214, "learning_rate": 0.00011723392668159377, "loss": 1.9139914512634277, "step": 16083 }, { "epoch": 1.8715382825226903, "grad_norm": 1.738470435142517, "learning_rate": 0.00011704035615251654, "loss": 2.0279972553253174, "step": 16084 }, { "epoch": 1.8716546427740284, "grad_norm": 0.9392760992050171, "learning_rate": 0.0001168468967097373, "loss": 1.9275437593460083, "step": 16085 }, { "epoch": 1.8717710030253665, "grad_norm": 0.6057966947555542, "learning_rate": 0.00011665354851488927, "loss": 1.5969243049621582, "step": 16086 }, { "epoch": 1.8718873632767048, "grad_norm": 1.2909682989120483, "learning_rate": 0.00011646031172951275, "loss": 2.1162869930267334, "step": 16087 }, { "epoch": 1.8720037235280427, "grad_norm": 1.4271563291549683, "learning_rate": 0.00011626718651505475, "loss": 2.277499198913574, "step": 16088 }, { "epoch": 1.872120083779381, "grad_norm": 1.167273759841919, "learning_rate": 0.00011607417303286946, "loss": 1.4918510913848877, "step": 16089 }, { "epoch": 1.872236444030719, "grad_norm": 7.542725086212158, "learning_rate": 0.0001158812714442174, "loss": 2.1912519931793213, "step": 16090 }, { "epoch": 1.8723528042820572, "grad_norm": 1.218929409980774, "learning_rate": 0.00011568848191026567, "loss": 1.9583065509796143, "step": 16091 }, { "epoch": 1.8724691645333955, "grad_norm": 1.268776535987854, "learning_rate": 0.00011549580459208781, "loss": 1.9126684665679932, "step": 16092 }, { "epoch": 1.8725855247847334, "grad_norm": 0.946101725101471, "learning_rate": 0.00011530323965066352, "loss": 1.5711627006530762, "step": 16093 }, { "epoch": 1.8727018850360717, "grad_norm": 1.1100282669067383, "learning_rate": 0.00011511078724687868, "loss": 2.5319366455078125, "step": 16094 }, { "epoch": 1.8728182452874098, "grad_norm": 1.0219638347625732, "learning_rate": 0.00011491844754152513, "loss": 1.523339033126831, "step": 16095 }, { "epoch": 1.872934605538748, "grad_norm": 1.1125450134277344, "learning_rate": 0.00011472622069530067, "loss": 2.176792860031128, "step": 16096 }, { "epoch": 1.8730509657900862, "grad_norm": 1.0710452795028687, "learning_rate": 0.00011453410686880844, "loss": 1.9583384990692139, "step": 16097 }, { "epoch": 1.873167326041424, "grad_norm": 1.3649739027023315, "learning_rate": 0.00011434210622255753, "loss": 2.17933988571167, "step": 16098 }, { "epoch": 1.8732836862927624, "grad_norm": 1.064314365386963, "learning_rate": 0.0001141502189169622, "loss": 2.456406354904175, "step": 16099 }, { "epoch": 1.8734000465441005, "grad_norm": 0.7829712629318237, "learning_rate": 0.0001139584451123424, "loss": 1.5731031894683838, "step": 16100 }, { "epoch": 1.8735164067954386, "grad_norm": 1.232851505279541, "learning_rate": 0.00011376678496892286, "loss": 2.7394304275512695, "step": 16101 }, { "epoch": 1.873632767046777, "grad_norm": 1.231828212738037, "learning_rate": 0.0001135752386468335, "loss": 2.3687057495117188, "step": 16102 }, { "epoch": 1.8737491272981148, "grad_norm": 0.8125205636024475, "learning_rate": 0.00011338380630610914, "loss": 0.9983489513397217, "step": 16103 }, { "epoch": 1.8738654875494531, "grad_norm": 0.7280352711677551, "learning_rate": 0.0001131924881066894, "loss": 1.5686805248260498, "step": 16104 }, { "epoch": 1.8739818478007912, "grad_norm": 1.0360703468322754, "learning_rate": 0.00011300128420841843, "loss": 2.519036054611206, "step": 16105 }, { "epoch": 1.8740982080521293, "grad_norm": 0.7683697938919067, "learning_rate": 0.00011281019477104498, "loss": 1.8767809867858887, "step": 16106 }, { "epoch": 1.8742145683034677, "grad_norm": 0.746220052242279, "learning_rate": 0.00011261921995422217, "loss": 1.6268947124481201, "step": 16107 }, { "epoch": 1.8743309285548055, "grad_norm": 1.0037976503372192, "learning_rate": 0.00011242835991750727, "loss": 1.1926730871200562, "step": 16108 }, { "epoch": 1.8744472888061439, "grad_norm": 0.7630695700645447, "learning_rate": 0.00011223761482036177, "loss": 1.583160638809204, "step": 16109 }, { "epoch": 1.874563649057482, "grad_norm": 0.9155757427215576, "learning_rate": 0.00011204698482215107, "loss": 2.280942678451538, "step": 16110 }, { "epoch": 1.87468000930882, "grad_norm": 0.8019281029701233, "learning_rate": 0.00011185647008214433, "loss": 2.274653911590576, "step": 16111 }, { "epoch": 1.8747963695601584, "grad_norm": 0.5778816342353821, "learning_rate": 0.00011166607075951454, "loss": 1.8102784156799316, "step": 16112 }, { "epoch": 1.8749127298114963, "grad_norm": 1.8261078596115112, "learning_rate": 0.00011147578701333812, "loss": 1.8211169242858887, "step": 16113 }, { "epoch": 1.8750290900628346, "grad_norm": 1.0294456481933594, "learning_rate": 0.00011128561900259521, "loss": 1.7949918508529663, "step": 16114 }, { "epoch": 1.8751454503141727, "grad_norm": 1.1757395267486572, "learning_rate": 0.00011109556688616895, "loss": 1.3625380992889404, "step": 16115 }, { "epoch": 1.8752618105655108, "grad_norm": 0.805760383605957, "learning_rate": 0.00011090563082284574, "loss": 1.6965255737304688, "step": 16116 }, { "epoch": 1.875378170816849, "grad_norm": 0.6795887351036072, "learning_rate": 0.00011071581097131509, "loss": 1.6165497303009033, "step": 16117 }, { "epoch": 1.875494531068187, "grad_norm": 0.7464469075202942, "learning_rate": 0.0001105261074901693, "loss": 1.5694987773895264, "step": 16118 }, { "epoch": 1.8756108913195253, "grad_norm": 0.8719562292098999, "learning_rate": 0.00011033652053790357, "loss": 1.7799209356307983, "step": 16119 }, { "epoch": 1.8757272515708634, "grad_norm": 1.031821370124817, "learning_rate": 0.00011014705027291577, "loss": 2.5606794357299805, "step": 16120 }, { "epoch": 1.8758436118222015, "grad_norm": 1.3463501930236816, "learning_rate": 0.00010995769685350599, "loss": 1.7204537391662598, "step": 16121 }, { "epoch": 1.8759599720735398, "grad_norm": 1.157404899597168, "learning_rate": 0.00010976846043787698, "loss": 1.8491783142089844, "step": 16122 }, { "epoch": 1.8760763323248777, "grad_norm": 1.0818877220153809, "learning_rate": 0.00010957934118413365, "loss": 2.5975847244262695, "step": 16123 }, { "epoch": 1.876192692576216, "grad_norm": 1.2395142316818237, "learning_rate": 0.00010939033925028294, "loss": 2.1689810752868652, "step": 16124 }, { "epoch": 1.8763090528275541, "grad_norm": 1.385874629020691, "learning_rate": 0.00010920145479423408, "loss": 1.957027554512024, "step": 16125 }, { "epoch": 1.8764254130788922, "grad_norm": 1.3707412481307983, "learning_rate": 0.00010901268797379777, "loss": 1.7087771892547607, "step": 16126 }, { "epoch": 1.8765417733302305, "grad_norm": 0.9346516728401184, "learning_rate": 0.00010882403894668663, "loss": 1.4695287942886353, "step": 16127 }, { "epoch": 1.8766581335815684, "grad_norm": 0.9130263328552246, "learning_rate": 0.00010863550787051476, "loss": 1.8386614322662354, "step": 16128 }, { "epoch": 1.8767744938329067, "grad_norm": 0.882098913192749, "learning_rate": 0.00010844709490279783, "loss": 1.328862190246582, "step": 16129 }, { "epoch": 1.8768908540842448, "grad_norm": 0.8332406282424927, "learning_rate": 0.00010825880020095275, "loss": 1.9445587396621704, "step": 16130 }, { "epoch": 1.877007214335583, "grad_norm": 1.2613252401351929, "learning_rate": 0.00010807062392229763, "loss": 2.093425750732422, "step": 16131 }, { "epoch": 1.8771235745869213, "grad_norm": 0.9124023914337158, "learning_rate": 0.00010788256622405165, "loss": 1.197721242904663, "step": 16132 }, { "epoch": 1.8772399348382591, "grad_norm": 1.0523979663848877, "learning_rate": 0.00010769462726333493, "loss": 1.9944019317626953, "step": 16133 }, { "epoch": 1.8773562950895974, "grad_norm": 0.9421873688697815, "learning_rate": 0.00010750680719716838, "loss": 1.3369415998458862, "step": 16134 }, { "epoch": 1.8774726553409355, "grad_norm": 0.9156188368797302, "learning_rate": 0.00010731910618247353, "loss": 1.7380554676055908, "step": 16135 }, { "epoch": 1.8775890155922736, "grad_norm": 1.4958053827285767, "learning_rate": 0.00010713152437607251, "loss": 1.4657680988311768, "step": 16136 }, { "epoch": 1.877705375843612, "grad_norm": 1.3291170597076416, "learning_rate": 0.00010694406193468783, "loss": 1.8967082500457764, "step": 16137 }, { "epoch": 1.8778217360949498, "grad_norm": 1.1333459615707397, "learning_rate": 0.00010675671901494218, "loss": 1.9584383964538574, "step": 16138 }, { "epoch": 1.8779380963462882, "grad_norm": 0.9367607235908508, "learning_rate": 0.00010656949577335865, "loss": 1.5970381498336792, "step": 16139 }, { "epoch": 1.8780544565976263, "grad_norm": 0.903590977191925, "learning_rate": 0.00010638239236636005, "loss": 2.0234503746032715, "step": 16140 }, { "epoch": 1.8781708168489644, "grad_norm": 1.1021802425384521, "learning_rate": 0.00010619540895026922, "loss": 2.114316940307617, "step": 16141 }, { "epoch": 1.8782871771003027, "grad_norm": 1.1138566732406616, "learning_rate": 0.00010600854568130871, "loss": 2.3149986267089844, "step": 16142 }, { "epoch": 1.8784035373516406, "grad_norm": 0.826513946056366, "learning_rate": 0.00010582180271560071, "loss": 2.2697322368621826, "step": 16143 }, { "epoch": 1.8785198976029789, "grad_norm": 1.8735275268554688, "learning_rate": 0.00010563518020916695, "loss": 1.6151354312896729, "step": 16144 }, { "epoch": 1.878636257854317, "grad_norm": 0.8441319465637207, "learning_rate": 0.00010544867831792826, "loss": 2.015064001083374, "step": 16145 }, { "epoch": 1.878752618105655, "grad_norm": 0.8143008947372437, "learning_rate": 0.00010526229719770503, "loss": 2.378546714782715, "step": 16146 }, { "epoch": 1.8788689783569934, "grad_norm": 0.8469303846359253, "learning_rate": 0.00010507603700421655, "loss": 1.9471096992492676, "step": 16147 }, { "epoch": 1.8789853386083313, "grad_norm": 0.5542769432067871, "learning_rate": 0.00010488989789308115, "loss": 1.4663519859313965, "step": 16148 }, { "epoch": 1.8791016988596696, "grad_norm": 0.8060401082038879, "learning_rate": 0.0001047038800198159, "loss": 1.750814437866211, "step": 16149 }, { "epoch": 1.8792180591110077, "grad_norm": 0.9792070388793945, "learning_rate": 0.00010451798353983683, "loss": 2.4819016456604004, "step": 16150 }, { "epoch": 1.8793344193623458, "grad_norm": 1.1984636783599854, "learning_rate": 0.00010433220860845827, "loss": 1.7520012855529785, "step": 16151 }, { "epoch": 1.8794507796136841, "grad_norm": 1.2936866283416748, "learning_rate": 0.00010414655538089313, "loss": 1.4725300073623657, "step": 16152 }, { "epoch": 1.879567139865022, "grad_norm": 1.147218108177185, "learning_rate": 0.00010396102401225257, "loss": 1.808969497680664, "step": 16153 }, { "epoch": 1.8796835001163603, "grad_norm": 1.3180571794509888, "learning_rate": 0.00010377561465754604, "loss": 1.5194947719573975, "step": 16154 }, { "epoch": 1.8797998603676984, "grad_norm": 0.9614664316177368, "learning_rate": 0.00010359032747168093, "loss": 1.922347903251648, "step": 16155 }, { "epoch": 1.8799162206190365, "grad_norm": 0.5672977566719055, "learning_rate": 0.00010340516260946261, "loss": 1.460432767868042, "step": 16156 }, { "epoch": 1.8800325808703748, "grad_norm": 1.5763453245162964, "learning_rate": 0.0001032201202255943, "loss": 1.2853388786315918, "step": 16157 }, { "epoch": 1.8801489411217127, "grad_norm": 1.0724433660507202, "learning_rate": 0.0001030352004746768, "loss": 1.0337916612625122, "step": 16158 }, { "epoch": 1.880265301373051, "grad_norm": 0.9432407021522522, "learning_rate": 0.00010285040351120853, "loss": 1.419481873512268, "step": 16159 }, { "epoch": 1.8803816616243891, "grad_norm": 1.4748148918151855, "learning_rate": 0.00010266572948958527, "loss": 2.42610502243042, "step": 16160 }, { "epoch": 1.8804980218757272, "grad_norm": 1.2012112140655518, "learning_rate": 0.00010248117856410011, "loss": 2.46411395072937, "step": 16161 }, { "epoch": 1.8806143821270656, "grad_norm": 1.5265862941741943, "learning_rate": 0.0001022967508889433, "loss": 2.3206515312194824, "step": 16162 }, { "epoch": 1.8807307423784034, "grad_norm": 1.811795711517334, "learning_rate": 0.00010211244661820202, "loss": 1.9051611423492432, "step": 16163 }, { "epoch": 1.8808471026297418, "grad_norm": 1.0490144491195679, "learning_rate": 0.00010192826590586054, "loss": 1.5736095905303955, "step": 16164 }, { "epoch": 1.8809634628810799, "grad_norm": 1.150120496749878, "learning_rate": 0.0001017442089057998, "loss": 1.6044442653656006, "step": 16165 }, { "epoch": 1.881079823132418, "grad_norm": 1.1568487882614136, "learning_rate": 0.00010156027577179732, "loss": 1.7087271213531494, "step": 16166 }, { "epoch": 1.8811961833837563, "grad_norm": 1.553040862083435, "learning_rate": 0.00010137646665752717, "loss": 2.0079855918884277, "step": 16167 }, { "epoch": 1.8813125436350941, "grad_norm": 0.8152293562889099, "learning_rate": 0.00010119278171655991, "loss": 1.4865963459014893, "step": 16168 }, { "epoch": 1.8814289038864325, "grad_norm": 0.7761854529380798, "learning_rate": 0.00010100922110236208, "loss": 1.9114418029785156, "step": 16169 }, { "epoch": 1.8815452641377706, "grad_norm": 24.524389266967773, "learning_rate": 0.00010082578496829665, "loss": 1.8414233922958374, "step": 16170 }, { "epoch": 1.8816616243891087, "grad_norm": 1.124567985534668, "learning_rate": 0.00010064247346762237, "loss": 1.3634259700775146, "step": 16171 }, { "epoch": 1.881777984640447, "grad_norm": 1.0256569385528564, "learning_rate": 0.00010045928675349406, "loss": 1.7724952697753906, "step": 16172 }, { "epoch": 1.8818943448917849, "grad_norm": 1.3418329954147339, "learning_rate": 0.00010027622497896211, "loss": 1.6431283950805664, "step": 16173 }, { "epoch": 1.8820107051431232, "grad_norm": 1.056924819946289, "learning_rate": 0.00010009328829697248, "loss": 2.1809170246124268, "step": 16174 }, { "epoch": 1.8821270653944613, "grad_norm": 1.4728684425354004, "learning_rate": 9.991047686036697e-05, "loss": 2.314028739929199, "step": 16175 }, { "epoch": 1.8822434256457994, "grad_norm": 0.7973363995552063, "learning_rate": 9.972779082188237e-05, "loss": 1.6451361179351807, "step": 16176 }, { "epoch": 1.8823597858971375, "grad_norm": 0.5835299491882324, "learning_rate": 9.954523033415084e-05, "loss": 1.5925698280334473, "step": 16177 }, { "epoch": 1.8824761461484756, "grad_norm": 2.534841537475586, "learning_rate": 9.936279554969959e-05, "loss": 1.5993647575378418, "step": 16178 }, { "epoch": 1.882592506399814, "grad_norm": 0.2345229536294937, "learning_rate": 9.918048662095089e-05, "loss": 0.9564458727836609, "step": 16179 }, { "epoch": 1.882708866651152, "grad_norm": 0.9193663001060486, "learning_rate": 9.899830370022181e-05, "loss": 1.4359872341156006, "step": 16180 }, { "epoch": 1.88282522690249, "grad_norm": 0.9584746956825256, "learning_rate": 9.88162469397241e-05, "loss": 2.36683988571167, "step": 16181 }, { "epoch": 1.8829415871538282, "grad_norm": 1.2125513553619385, "learning_rate": 9.86343164915642e-05, "loss": 2.086683750152588, "step": 16182 }, { "epoch": 1.8830579474051663, "grad_norm": 1.4682807922363281, "learning_rate": 9.845251250774295e-05, "loss": 1.9292821884155273, "step": 16183 }, { "epoch": 1.8831743076565046, "grad_norm": 0.9161823987960815, "learning_rate": 9.827083514015559e-05, "loss": 1.668510913848877, "step": 16184 }, { "epoch": 1.8832906679078427, "grad_norm": 1.4348664283752441, "learning_rate": 9.808928454059146e-05, "loss": 2.1640024185180664, "step": 16185 }, { "epoch": 1.8834070281591808, "grad_norm": 1.0066558122634888, "learning_rate": 9.790786086073408e-05, "loss": 1.9262861013412476, "step": 16186 }, { "epoch": 1.883523388410519, "grad_norm": 0.903847873210907, "learning_rate": 9.772656425216095e-05, "loss": 2.0964126586914062, "step": 16187 }, { "epoch": 1.883639748661857, "grad_norm": 0.9219058156013489, "learning_rate": 9.754539486634322e-05, "loss": 1.9036967754364014, "step": 16188 }, { "epoch": 1.8837561089131953, "grad_norm": 1.1718056201934814, "learning_rate": 9.736435285464612e-05, "loss": 2.3907244205474854, "step": 16189 }, { "epoch": 1.8838724691645334, "grad_norm": 0.9658719897270203, "learning_rate": 9.718343836832811e-05, "loss": 1.739626169204712, "step": 16190 }, { "epoch": 1.8839888294158715, "grad_norm": 1.307012677192688, "learning_rate": 9.700265155854121e-05, "loss": 1.7276257276535034, "step": 16191 }, { "epoch": 1.8841051896672096, "grad_norm": 1.1966311931610107, "learning_rate": 9.68219925763309e-05, "loss": 2.055825710296631, "step": 16192 }, { "epoch": 1.8842215499185477, "grad_norm": 0.8867533206939697, "learning_rate": 9.664146157263554e-05, "loss": 1.3459599018096924, "step": 16193 }, { "epoch": 1.884337910169886, "grad_norm": 0.7474818825721741, "learning_rate": 9.646105869828687e-05, "loss": 2.0109996795654297, "step": 16194 }, { "epoch": 1.8844542704212242, "grad_norm": 1.018895149230957, "learning_rate": 9.628078410400948e-05, "loss": 2.1195287704467773, "step": 16195 }, { "epoch": 1.8845706306725623, "grad_norm": 1.25833261013031, "learning_rate": 9.610063794042079e-05, "loss": 1.364288330078125, "step": 16196 }, { "epoch": 1.8846869909239004, "grad_norm": 0.8278055787086487, "learning_rate": 9.592062035803087e-05, "loss": 1.967320203781128, "step": 16197 }, { "epoch": 1.8848033511752385, "grad_norm": 0.9561206102371216, "learning_rate": 9.574073150724244e-05, "loss": 2.295741081237793, "step": 16198 }, { "epoch": 1.8849197114265768, "grad_norm": 1.2340214252471924, "learning_rate": 9.556097153835053e-05, "loss": 1.9143950939178467, "step": 16199 }, { "epoch": 1.8850360716779149, "grad_norm": 1.0737028121948242, "learning_rate": 9.538134060154274e-05, "loss": 1.3274157047271729, "step": 16200 }, { "epoch": 1.885152431929253, "grad_norm": 0.5606240034103394, "learning_rate": 9.520183884689868e-05, "loss": 1.1618775129318237, "step": 16201 }, { "epoch": 1.885268792180591, "grad_norm": 1.1719988584518433, "learning_rate": 9.502246642438997e-05, "loss": 1.8119151592254639, "step": 16202 }, { "epoch": 1.8853851524319292, "grad_norm": 2.0992789268493652, "learning_rate": 9.484322348388035e-05, "loss": 2.352039337158203, "step": 16203 }, { "epoch": 1.8855015126832675, "grad_norm": 0.8996021747589111, "learning_rate": 9.466411017512528e-05, "loss": 1.827196717262268, "step": 16204 }, { "epoch": 1.8856178729346056, "grad_norm": 0.4740156829357147, "learning_rate": 9.44851266477719e-05, "loss": 1.1136455535888672, "step": 16205 }, { "epoch": 1.8857342331859437, "grad_norm": 0.9480624794960022, "learning_rate": 9.430627305135894e-05, "loss": 1.7286462783813477, "step": 16206 }, { "epoch": 1.8858505934372818, "grad_norm": 1.0333945751190186, "learning_rate": 9.412754953531663e-05, "loss": 1.3829712867736816, "step": 16207 }, { "epoch": 1.8859669536886199, "grad_norm": 0.6471548080444336, "learning_rate": 9.394895624896646e-05, "loss": 1.376216173171997, "step": 16208 }, { "epoch": 1.8860833139399582, "grad_norm": 1.2060209512710571, "learning_rate": 9.37704933415211e-05, "loss": 1.8565224409103394, "step": 16209 }, { "epoch": 1.8861996741912963, "grad_norm": 0.8587403893470764, "learning_rate": 9.359216096208434e-05, "loss": 1.0026297569274902, "step": 16210 }, { "epoch": 1.8863160344426344, "grad_norm": 1.1389566659927368, "learning_rate": 9.341395925965087e-05, "loss": 2.1081578731536865, "step": 16211 }, { "epoch": 1.8864323946939725, "grad_norm": 0.5807950496673584, "learning_rate": 9.323588838310627e-05, "loss": 1.8247710466384888, "step": 16212 }, { "epoch": 1.8865487549453106, "grad_norm": 1.1689929962158203, "learning_rate": 9.305794848122662e-05, "loss": 1.4822665452957153, "step": 16213 }, { "epoch": 1.886665115196649, "grad_norm": 1.122056484222412, "learning_rate": 9.288013970267894e-05, "loss": 1.3376777172088623, "step": 16214 }, { "epoch": 1.886781475447987, "grad_norm": 1.2850134372711182, "learning_rate": 9.270246219602038e-05, "loss": 2.1582839488983154, "step": 16215 }, { "epoch": 1.8868978356993251, "grad_norm": 0.9897527098655701, "learning_rate": 9.25249161096986e-05, "loss": 1.990699291229248, "step": 16216 }, { "epoch": 1.8870141959506632, "grad_norm": 0.5532604455947876, "learning_rate": 9.234750159205118e-05, "loss": 1.659524917602539, "step": 16217 }, { "epoch": 1.8871305562020013, "grad_norm": 1.0297554731369019, "learning_rate": 9.217021879130607e-05, "loss": 1.5116157531738281, "step": 16218 }, { "epoch": 1.8872469164533396, "grad_norm": 1.4090067148208618, "learning_rate": 9.19930678555811e-05, "loss": 1.6865956783294678, "step": 16219 }, { "epoch": 1.8873632767046775, "grad_norm": 1.06787109375, "learning_rate": 9.181604893288384e-05, "loss": 2.1024115085601807, "step": 16220 }, { "epoch": 1.8874796369560158, "grad_norm": 1.0125268697738647, "learning_rate": 9.163916217111168e-05, "loss": 1.6624691486358643, "step": 16221 }, { "epoch": 1.887595997207354, "grad_norm": 1.260245680809021, "learning_rate": 9.14624077180515e-05, "loss": 1.500730037689209, "step": 16222 }, { "epoch": 1.887712357458692, "grad_norm": 0.9712170958518982, "learning_rate": 9.128578572137969e-05, "loss": 1.7419615983963013, "step": 16223 }, { "epoch": 1.8878287177100304, "grad_norm": 0.9713394045829773, "learning_rate": 9.110929632866188e-05, "loss": 2.2583839893341064, "step": 16224 }, { "epoch": 1.8879450779613682, "grad_norm": 1.1444411277770996, "learning_rate": 9.09329396873531e-05, "loss": 2.3669674396514893, "step": 16225 }, { "epoch": 1.8880614382127066, "grad_norm": 1.192143440246582, "learning_rate": 9.075671594479734e-05, "loss": 1.3912358283996582, "step": 16226 }, { "epoch": 1.8881777984640447, "grad_norm": 1.1912567615509033, "learning_rate": 9.058062524822755e-05, "loss": 1.7657414674758911, "step": 16227 }, { "epoch": 1.8882941587153828, "grad_norm": 1.0428760051727295, "learning_rate": 9.040466774476552e-05, "loss": 2.049544334411621, "step": 16228 }, { "epoch": 1.888410518966721, "grad_norm": 0.7397036552429199, "learning_rate": 9.022884358142178e-05, "loss": 1.727495789527893, "step": 16229 }, { "epoch": 1.888526879218059, "grad_norm": 0.6857409477233887, "learning_rate": 9.005315290509547e-05, "loss": 1.3435392379760742, "step": 16230 }, { "epoch": 1.8886432394693973, "grad_norm": 1.0297907590866089, "learning_rate": 8.987759586257421e-05, "loss": 1.7545456886291504, "step": 16231 }, { "epoch": 1.8887595997207354, "grad_norm": 1.5843477249145508, "learning_rate": 8.97021726005339e-05, "loss": 1.389763355255127, "step": 16232 }, { "epoch": 1.8888759599720735, "grad_norm": 1.1772087812423706, "learning_rate": 8.952688326553876e-05, "loss": 2.050055980682373, "step": 16233 }, { "epoch": 1.8889923202234118, "grad_norm": 1.2839441299438477, "learning_rate": 8.935172800404104e-05, "loss": 2.338984489440918, "step": 16234 }, { "epoch": 1.8891086804747497, "grad_norm": 0.7286640405654907, "learning_rate": 8.9176706962381e-05, "loss": 2.4223954677581787, "step": 16235 }, { "epoch": 1.889225040726088, "grad_norm": 1.0120805501937866, "learning_rate": 8.900182028678683e-05, "loss": 1.94913911819458, "step": 16236 }, { "epoch": 1.889341400977426, "grad_norm": 1.0562204122543335, "learning_rate": 8.882706812337433e-05, "loss": 1.5773136615753174, "step": 16237 }, { "epoch": 1.8894577612287642, "grad_norm": 1.3762977123260498, "learning_rate": 8.865245061814689e-05, "loss": 2.406160593032837, "step": 16238 }, { "epoch": 1.8895741214801025, "grad_norm": 0.7059740424156189, "learning_rate": 8.847796791699573e-05, "loss": 1.2479023933410645, "step": 16239 }, { "epoch": 1.8896904817314404, "grad_norm": 1.1643730401992798, "learning_rate": 8.830362016569912e-05, "loss": 1.7074158191680908, "step": 16240 }, { "epoch": 1.8898068419827787, "grad_norm": 1.4563376903533936, "learning_rate": 8.812940750992254e-05, "loss": 1.6387001276016235, "step": 16241 }, { "epoch": 1.8899232022341168, "grad_norm": 1.0210987329483032, "learning_rate": 8.795533009521877e-05, "loss": 1.7367830276489258, "step": 16242 }, { "epoch": 1.890039562485455, "grad_norm": 0.9791836738586426, "learning_rate": 8.778138806702757e-05, "loss": 1.6930668354034424, "step": 16243 }, { "epoch": 1.8901559227367932, "grad_norm": 1.0152281522750854, "learning_rate": 8.760758157067556e-05, "loss": 1.5215587615966797, "step": 16244 }, { "epoch": 1.890272282988131, "grad_norm": 1.081297755241394, "learning_rate": 8.74339107513761e-05, "loss": 1.556739330291748, "step": 16245 }, { "epoch": 1.8903886432394694, "grad_norm": 0.8343310952186584, "learning_rate": 8.726037575422926e-05, "loss": 2.4400594234466553, "step": 16246 }, { "epoch": 1.8905050034908075, "grad_norm": 0.7448624968528748, "learning_rate": 8.708697672422159e-05, "loss": 1.8071262836456299, "step": 16247 }, { "epoch": 1.8906213637421456, "grad_norm": 0.8271570205688477, "learning_rate": 8.691371380622601e-05, "loss": 1.3217887878417969, "step": 16248 }, { "epoch": 1.890737723993484, "grad_norm": 0.8543450236320496, "learning_rate": 8.674058714500169e-05, "loss": 1.826119303703308, "step": 16249 }, { "epoch": 1.8908540842448218, "grad_norm": 1.9657652378082275, "learning_rate": 8.65675968851942e-05, "loss": 2.565707206726074, "step": 16250 }, { "epoch": 1.8909704444961601, "grad_norm": 1.2687933444976807, "learning_rate": 8.639474317133483e-05, "loss": 1.8250596523284912, "step": 16251 }, { "epoch": 1.8910868047474982, "grad_norm": 1.5603057146072388, "learning_rate": 8.622202614784103e-05, "loss": 2.4384846687316895, "step": 16252 }, { "epoch": 1.8912031649988363, "grad_norm": 1.5496408939361572, "learning_rate": 8.604944595901587e-05, "loss": 1.8669013977050781, "step": 16253 }, { "epoch": 1.8913195252501747, "grad_norm": 0.9830930829048157, "learning_rate": 8.58770027490482e-05, "loss": 1.7023594379425049, "step": 16254 }, { "epoch": 1.8914358855015125, "grad_norm": 1.0778586864471436, "learning_rate": 8.570469666201236e-05, "loss": 2.239645004272461, "step": 16255 }, { "epoch": 1.8915522457528509, "grad_norm": 0.8098641633987427, "learning_rate": 8.553252784186822e-05, "loss": 1.4170013666152954, "step": 16256 }, { "epoch": 1.891668606004189, "grad_norm": 2.568310260772705, "learning_rate": 8.536049643246082e-05, "loss": 1.8884035348892212, "step": 16257 }, { "epoch": 1.891784966255527, "grad_norm": 1.124136209487915, "learning_rate": 8.51886025775206e-05, "loss": 2.10139536857605, "step": 16258 }, { "epoch": 1.8919013265068654, "grad_norm": 0.7640135884284973, "learning_rate": 8.501684642066288e-05, "loss": 1.2908167839050293, "step": 16259 }, { "epoch": 1.8920176867582033, "grad_norm": 1.3950300216674805, "learning_rate": 8.4845228105388e-05, "loss": 2.1450839042663574, "step": 16260 }, { "epoch": 1.8921340470095416, "grad_norm": 0.8563112020492554, "learning_rate": 8.467374777508119e-05, "loss": 1.3217685222625732, "step": 16261 }, { "epoch": 1.8922504072608797, "grad_norm": 1.0687129497528076, "learning_rate": 8.450240557301233e-05, "loss": 2.08608341217041, "step": 16262 }, { "epoch": 1.8923667675122178, "grad_norm": 1.007157564163208, "learning_rate": 8.433120164233585e-05, "loss": 1.7680814266204834, "step": 16263 }, { "epoch": 1.892483127763556, "grad_norm": 1.339717149734497, "learning_rate": 8.416013612609094e-05, "loss": 2.039000988006592, "step": 16264 }, { "epoch": 1.892599488014894, "grad_norm": 1.2890156507492065, "learning_rate": 8.39892091672007e-05, "loss": 2.1605210304260254, "step": 16265 }, { "epoch": 1.8927158482662323, "grad_norm": 1.3027827739715576, "learning_rate": 8.381842090847278e-05, "loss": 2.862361431121826, "step": 16266 }, { "epoch": 1.8928322085175704, "grad_norm": 1.3008249998092651, "learning_rate": 8.364777149259886e-05, "loss": 1.6084035634994507, "step": 16267 }, { "epoch": 1.8929485687689085, "grad_norm": 0.8748474717140198, "learning_rate": 8.347726106215464e-05, "loss": 1.3300220966339111, "step": 16268 }, { "epoch": 1.8930649290202468, "grad_norm": 0.7128311991691589, "learning_rate": 8.330688975959966e-05, "loss": 1.0027985572814941, "step": 16269 }, { "epoch": 1.8931812892715847, "grad_norm": 0.9642407298088074, "learning_rate": 8.313665772727721e-05, "loss": 2.029189109802246, "step": 16270 }, { "epoch": 1.893297649522923, "grad_norm": 1.0401626825332642, "learning_rate": 8.296656510741432e-05, "loss": 2.1213836669921875, "step": 16271 }, { "epoch": 1.893414009774261, "grad_norm": 0.8063159584999084, "learning_rate": 8.279661204212144e-05, "loss": 1.8636810779571533, "step": 16272 }, { "epoch": 1.8935303700255992, "grad_norm": 0.6683843731880188, "learning_rate": 8.262679867339245e-05, "loss": 1.199871301651001, "step": 16273 }, { "epoch": 1.8936467302769375, "grad_norm": 0.845806360244751, "learning_rate": 8.245712514310447e-05, "loss": 1.3909891843795776, "step": 16274 }, { "epoch": 1.8937630905282754, "grad_norm": 1.1437747478485107, "learning_rate": 8.228759159301799e-05, "loss": 1.8385882377624512, "step": 16275 }, { "epoch": 1.8938794507796137, "grad_norm": 0.9153014421463013, "learning_rate": 8.211819816477631e-05, "loss": 1.283825397491455, "step": 16276 }, { "epoch": 1.8939958110309518, "grad_norm": 2.2927916049957275, "learning_rate": 8.194894499990574e-05, "loss": 2.1372129917144775, "step": 16277 }, { "epoch": 1.89411217128229, "grad_norm": 0.9135031700134277, "learning_rate": 8.177983223981541e-05, "loss": 2.0791141986846924, "step": 16278 }, { "epoch": 1.8942285315336282, "grad_norm": 0.680864155292511, "learning_rate": 8.161086002579717e-05, "loss": 1.3269150257110596, "step": 16279 }, { "epoch": 1.8943448917849661, "grad_norm": 1.090201735496521, "learning_rate": 8.14420284990254e-05, "loss": 1.2936031818389893, "step": 16280 }, { "epoch": 1.8944612520363044, "grad_norm": 1.3317725658416748, "learning_rate": 8.127333780055696e-05, "loss": 1.9727998971939087, "step": 16281 }, { "epoch": 1.8945776122876425, "grad_norm": 0.8864296674728394, "learning_rate": 8.110478807133104e-05, "loss": 1.6476645469665527, "step": 16282 }, { "epoch": 1.8946939725389806, "grad_norm": 1.130968689918518, "learning_rate": 8.093637945216905e-05, "loss": 2.244922161102295, "step": 16283 }, { "epoch": 1.894810332790319, "grad_norm": 1.6648634672164917, "learning_rate": 8.076811208377455e-05, "loss": 1.6781301498413086, "step": 16284 }, { "epoch": 1.8949266930416568, "grad_norm": 1.2336195707321167, "learning_rate": 8.059998610673299e-05, "loss": 1.7078912258148193, "step": 16285 }, { "epoch": 1.8950430532929952, "grad_norm": 0.5922573804855347, "learning_rate": 8.04320016615118e-05, "loss": 1.2655370235443115, "step": 16286 }, { "epoch": 1.8951594135443333, "grad_norm": 1.050115942955017, "learning_rate": 8.026415888846009e-05, "loss": 1.582214593887329, "step": 16287 }, { "epoch": 1.8952757737956714, "grad_norm": 0.7724305391311646, "learning_rate": 8.009645792780859e-05, "loss": 1.6138476133346558, "step": 16288 }, { "epoch": 1.8953921340470097, "grad_norm": 0.8345981240272522, "learning_rate": 7.992889891966961e-05, "loss": 1.4199179410934448, "step": 16289 }, { "epoch": 1.8955084942983476, "grad_norm": 1.4792113304138184, "learning_rate": 7.976148200403685e-05, "loss": 1.446336269378662, "step": 16290 }, { "epoch": 1.8956248545496859, "grad_norm": 1.0366941690444946, "learning_rate": 7.959420732078524e-05, "loss": 1.6550357341766357, "step": 16291 }, { "epoch": 1.895741214801024, "grad_norm": 0.6001859307289124, "learning_rate": 7.942707500967092e-05, "loss": 1.399186134338379, "step": 16292 }, { "epoch": 1.895857575052362, "grad_norm": 1.6307107210159302, "learning_rate": 7.926008521033107e-05, "loss": 1.8065345287322998, "step": 16293 }, { "epoch": 1.8959739353037004, "grad_norm": 1.1620583534240723, "learning_rate": 7.90932380622838e-05, "loss": 2.0116848945617676, "step": 16294 }, { "epoch": 1.8960902955550383, "grad_norm": 0.8356404900550842, "learning_rate": 7.892653370492805e-05, "loss": 1.527428150177002, "step": 16295 }, { "epoch": 1.8962066558063766, "grad_norm": 0.9457488656044006, "learning_rate": 7.875997227754348e-05, "loss": 1.7506659030914307, "step": 16296 }, { "epoch": 1.8963230160577147, "grad_norm": 1.2726695537567139, "learning_rate": 7.859355391929024e-05, "loss": 2.256241798400879, "step": 16297 }, { "epoch": 1.8964393763090528, "grad_norm": 1.3569231033325195, "learning_rate": 7.842727876920902e-05, "loss": 2.308781147003174, "step": 16298 }, { "epoch": 1.8965557365603911, "grad_norm": 0.9962016940116882, "learning_rate": 7.82611469662208e-05, "loss": 1.7547178268432617, "step": 16299 }, { "epoch": 1.896672096811729, "grad_norm": 1.484190583229065, "learning_rate": 7.8095158649127e-05, "loss": 1.861693263053894, "step": 16300 }, { "epoch": 1.8967884570630673, "grad_norm": 0.7088772058486938, "learning_rate": 7.792931395660894e-05, "loss": 1.5837678909301758, "step": 16301 }, { "epoch": 1.8969048173144054, "grad_norm": 1.2690849304199219, "learning_rate": 7.776361302722796e-05, "loss": 1.6428375244140625, "step": 16302 }, { "epoch": 1.8970211775657435, "grad_norm": 1.506834626197815, "learning_rate": 7.759805599942538e-05, "loss": 1.5535321235656738, "step": 16303 }, { "epoch": 1.8971375378170818, "grad_norm": 0.7893115878105164, "learning_rate": 7.743264301152223e-05, "loss": 2.0151662826538086, "step": 16304 }, { "epoch": 1.8972538980684197, "grad_norm": 0.8890986442565918, "learning_rate": 7.726737420171923e-05, "loss": 1.4025442600250244, "step": 16305 }, { "epoch": 1.897370258319758, "grad_norm": 1.0075098276138306, "learning_rate": 7.710224970809662e-05, "loss": 1.6325161457061768, "step": 16306 }, { "epoch": 1.8974866185710961, "grad_norm": 1.2084863185882568, "learning_rate": 7.693726966861406e-05, "loss": 1.5033519268035889, "step": 16307 }, { "epoch": 1.8976029788224342, "grad_norm": 0.8498715758323669, "learning_rate": 7.677243422111053e-05, "loss": 1.5318098068237305, "step": 16308 }, { "epoch": 1.8977193390737725, "grad_norm": 0.8309704661369324, "learning_rate": 7.66077435033042e-05, "loss": 1.436450719833374, "step": 16309 }, { "epoch": 1.8978356993251104, "grad_norm": 1.181960105895996, "learning_rate": 7.644319765279237e-05, "loss": 1.6312898397445679, "step": 16310 }, { "epoch": 1.8979520595764487, "grad_norm": 1.0389851331710815, "learning_rate": 7.627879680705124e-05, "loss": 1.6414629220962524, "step": 16311 }, { "epoch": 1.8980684198277868, "grad_norm": 0.8114297986030579, "learning_rate": 7.611454110343586e-05, "loss": 1.61482834815979, "step": 16312 }, { "epoch": 1.898184780079125, "grad_norm": 1.1988482475280762, "learning_rate": 7.595043067918004e-05, "loss": 1.3421345949172974, "step": 16313 }, { "epoch": 1.8983011403304633, "grad_norm": 1.1039608716964722, "learning_rate": 7.578646567139621e-05, "loss": 1.3481165170669556, "step": 16314 }, { "epoch": 1.8984175005818011, "grad_norm": 0.7334421873092651, "learning_rate": 7.562264621707532e-05, "loss": 1.7425097227096558, "step": 16315 }, { "epoch": 1.8985338608331395, "grad_norm": 1.1928461790084839, "learning_rate": 7.545897245308667e-05, "loss": 2.1203222274780273, "step": 16316 }, { "epoch": 1.8986502210844776, "grad_norm": 0.9867001175880432, "learning_rate": 7.52954445161779e-05, "loss": 1.5781549215316772, "step": 16317 }, { "epoch": 1.8987665813358157, "grad_norm": 0.9666743874549866, "learning_rate": 7.513206254297472e-05, "loss": 1.8477081060409546, "step": 16318 }, { "epoch": 1.898882941587154, "grad_norm": 0.9840586185455322, "learning_rate": 7.496882666998101e-05, "loss": 2.5957024097442627, "step": 16319 }, { "epoch": 1.8989993018384919, "grad_norm": 1.138791799545288, "learning_rate": 7.480573703357846e-05, "loss": 1.8909474611282349, "step": 16320 }, { "epoch": 1.8991156620898302, "grad_norm": 1.120780110359192, "learning_rate": 7.46427937700267e-05, "loss": 1.6690539121627808, "step": 16321 }, { "epoch": 1.8992320223411683, "grad_norm": 0.5161376595497131, "learning_rate": 7.447999701546296e-05, "loss": 1.2247580289840698, "step": 16322 }, { "epoch": 1.8993483825925064, "grad_norm": 1.2558963298797607, "learning_rate": 7.431734690590212e-05, "loss": 1.9693015813827515, "step": 16323 }, { "epoch": 1.8994647428438447, "grad_norm": 1.0459240674972534, "learning_rate": 7.415484357723645e-05, "loss": 1.2799731492996216, "step": 16324 }, { "epoch": 1.8995811030951826, "grad_norm": 0.9582260847091675, "learning_rate": 7.399248716523582e-05, "loss": 1.9895381927490234, "step": 16325 }, { "epoch": 1.899697463346521, "grad_norm": 1.334672212600708, "learning_rate": 7.383027780554716e-05, "loss": 1.425870656967163, "step": 16326 }, { "epoch": 1.899813823597859, "grad_norm": 1.6900430917739868, "learning_rate": 7.366821563369455e-05, "loss": 1.9843181371688843, "step": 16327 }, { "epoch": 1.899930183849197, "grad_norm": 0.762108564376831, "learning_rate": 7.35063007850791e-05, "loss": 1.5842549800872803, "step": 16328 }, { "epoch": 1.9000465441005354, "grad_norm": 1.263207197189331, "learning_rate": 7.334453339497892e-05, "loss": 2.319727659225464, "step": 16329 }, { "epoch": 1.9001629043518733, "grad_norm": 1.2410738468170166, "learning_rate": 7.318291359854881e-05, "loss": 2.3752925395965576, "step": 16330 }, { "epoch": 1.9002792646032116, "grad_norm": 0.6753607392311096, "learning_rate": 7.302144153082029e-05, "loss": 1.330385446548462, "step": 16331 }, { "epoch": 1.9003956248545497, "grad_norm": 1.1692018508911133, "learning_rate": 7.286011732670155e-05, "loss": 1.392053246498108, "step": 16332 }, { "epoch": 1.9005119851058878, "grad_norm": 1.166124701499939, "learning_rate": 7.269894112097708e-05, "loss": 2.4182722568511963, "step": 16333 }, { "epoch": 1.9006283453572261, "grad_norm": 1.2265136241912842, "learning_rate": 7.253791304830785e-05, "loss": 2.4388134479522705, "step": 16334 }, { "epoch": 1.900744705608564, "grad_norm": 0.9760536551475525, "learning_rate": 7.237703324323098e-05, "loss": 2.1340785026550293, "step": 16335 }, { "epoch": 1.9008610658599023, "grad_norm": 1.0354266166687012, "learning_rate": 7.221630184015976e-05, "loss": 1.7604591846466064, "step": 16336 }, { "epoch": 1.9009774261112404, "grad_norm": 0.9568356275558472, "learning_rate": 7.20557189733835e-05, "loss": 2.048614025115967, "step": 16337 }, { "epoch": 1.9010937863625785, "grad_norm": 0.955130934715271, "learning_rate": 7.18952847770674e-05, "loss": 1.8440213203430176, "step": 16338 }, { "epoch": 1.9012101466139169, "grad_norm": 0.9174306392669678, "learning_rate": 7.173499938525241e-05, "loss": 2.1677329540252686, "step": 16339 }, { "epoch": 1.9013265068652547, "grad_norm": 0.7760827541351318, "learning_rate": 7.15748629318552e-05, "loss": 2.047400951385498, "step": 16340 }, { "epoch": 1.901442867116593, "grad_norm": 0.9697023034095764, "learning_rate": 7.141487555066797e-05, "loss": 2.144028663635254, "step": 16341 }, { "epoch": 1.9015592273679311, "grad_norm": 1.176958680152893, "learning_rate": 7.125503737535843e-05, "loss": 2.087731122970581, "step": 16342 }, { "epoch": 1.9016755876192692, "grad_norm": 1.3529325723648071, "learning_rate": 7.109534853946955e-05, "loss": 2.2052438259124756, "step": 16343 }, { "epoch": 1.9017919478706073, "grad_norm": 0.5536536574363708, "learning_rate": 7.093580917641962e-05, "loss": 1.2932965755462646, "step": 16344 }, { "epoch": 1.9019083081219454, "grad_norm": 0.9108757376670837, "learning_rate": 7.077641941950194e-05, "loss": 2.0073280334472656, "step": 16345 }, { "epoch": 1.9020246683732838, "grad_norm": 1.0757147073745728, "learning_rate": 7.061717940188491e-05, "loss": 1.7485566139221191, "step": 16346 }, { "epoch": 1.9021410286246219, "grad_norm": 1.068900465965271, "learning_rate": 7.04580892566118e-05, "loss": 1.6266837120056152, "step": 16347 }, { "epoch": 1.90225738887596, "grad_norm": 0.9337220788002014, "learning_rate": 7.029914911660056e-05, "loss": 1.4194599390029907, "step": 16348 }, { "epoch": 1.902373749127298, "grad_norm": 0.9484491348266602, "learning_rate": 7.014035911464395e-05, "loss": 1.9494355916976929, "step": 16349 }, { "epoch": 1.9024901093786362, "grad_norm": 1.3483412265777588, "learning_rate": 6.998171938340931e-05, "loss": 2.7859268188476562, "step": 16350 }, { "epoch": 1.9026064696299745, "grad_norm": 0.8690640926361084, "learning_rate": 6.982323005543833e-05, "loss": 1.9680064916610718, "step": 16351 }, { "epoch": 1.9027228298813126, "grad_norm": 1.2544668912887573, "learning_rate": 6.966489126314704e-05, "loss": 1.8643628358840942, "step": 16352 }, { "epoch": 1.9028391901326507, "grad_norm": 0.8840720057487488, "learning_rate": 6.950670313882579e-05, "loss": 1.3590378761291504, "step": 16353 }, { "epoch": 1.9029555503839888, "grad_norm": 1.2626018524169922, "learning_rate": 6.934866581463891e-05, "loss": 1.4766559600830078, "step": 16354 }, { "epoch": 1.9030719106353269, "grad_norm": 1.0213390588760376, "learning_rate": 6.919077942262489e-05, "loss": 1.6375259160995483, "step": 16355 }, { "epoch": 1.9031882708866652, "grad_norm": 1.2004276514053345, "learning_rate": 6.903304409469602e-05, "loss": 1.6736581325531006, "step": 16356 }, { "epoch": 1.9033046311380033, "grad_norm": 0.9790814518928528, "learning_rate": 6.887545996263842e-05, "loss": 1.7307453155517578, "step": 16357 }, { "epoch": 1.9034209913893414, "grad_norm": 1.3012229204177856, "learning_rate": 6.871802715811185e-05, "loss": 1.3218069076538086, "step": 16358 }, { "epoch": 1.9035373516406795, "grad_norm": 0.8581468462944031, "learning_rate": 6.856074581264966e-05, "loss": 1.7288613319396973, "step": 16359 }, { "epoch": 1.9036537118920176, "grad_norm": 0.919111967086792, "learning_rate": 6.840361605765868e-05, "loss": 1.6904149055480957, "step": 16360 }, { "epoch": 1.903770072143356, "grad_norm": 1.1315279006958008, "learning_rate": 6.824663802441905e-05, "loss": 2.062333345413208, "step": 16361 }, { "epoch": 1.903886432394694, "grad_norm": 0.9089208841323853, "learning_rate": 6.808981184408419e-05, "loss": 2.2754642963409424, "step": 16362 }, { "epoch": 1.9040027926460321, "grad_norm": 1.4217852354049683, "learning_rate": 6.793313764768058e-05, "loss": 2.033623695373535, "step": 16363 }, { "epoch": 1.9041191528973702, "grad_norm": 0.9767346382141113, "learning_rate": 6.777661556610778e-05, "loss": 1.2323945760726929, "step": 16364 }, { "epoch": 1.9042355131487083, "grad_norm": 1.4523048400878906, "learning_rate": 6.762024573013828e-05, "loss": 2.5376970767974854, "step": 16365 }, { "epoch": 1.9043518734000466, "grad_norm": 0.9677520394325256, "learning_rate": 6.746402827041729e-05, "loss": 2.4101293087005615, "step": 16366 }, { "epoch": 1.9044682336513847, "grad_norm": 1.3063637018203735, "learning_rate": 6.730796331746272e-05, "loss": 2.326538562774658, "step": 16367 }, { "epoch": 1.9045845939027228, "grad_norm": 1.1873860359191895, "learning_rate": 6.715205100166518e-05, "loss": 2.1926772594451904, "step": 16368 }, { "epoch": 1.904700954154061, "grad_norm": 1.2786779403686523, "learning_rate": 6.699629145328762e-05, "loss": 2.335505247116089, "step": 16369 }, { "epoch": 1.904817314405399, "grad_norm": 1.4732167720794678, "learning_rate": 6.684068480246542e-05, "loss": 1.8489165306091309, "step": 16370 }, { "epoch": 1.9049336746567374, "grad_norm": 0.9046616554260254, "learning_rate": 6.668523117920617e-05, "loss": 1.8904669284820557, "step": 16371 }, { "epoch": 1.9050500349080755, "grad_norm": 1.2709325551986694, "learning_rate": 6.652993071338967e-05, "loss": 1.5077693462371826, "step": 16372 }, { "epoch": 1.9051663951594136, "grad_norm": 0.8592262268066406, "learning_rate": 6.637478353476769e-05, "loss": 1.6371302604675293, "step": 16373 }, { "epoch": 1.9052827554107516, "grad_norm": 1.0544164180755615, "learning_rate": 6.621978977296394e-05, "loss": 2.0946574211120605, "step": 16374 }, { "epoch": 1.9053991156620897, "grad_norm": 1.1912826299667358, "learning_rate": 6.606494955747408e-05, "loss": 1.7444179058074951, "step": 16375 }, { "epoch": 1.905515475913428, "grad_norm": 0.6417692303657532, "learning_rate": 6.591026301766534e-05, "loss": 1.620342493057251, "step": 16376 }, { "epoch": 1.9056318361647662, "grad_norm": 1.3216373920440674, "learning_rate": 6.575573028277659e-05, "loss": 1.8684334754943848, "step": 16377 }, { "epoch": 1.9057481964161043, "grad_norm": 0.6750581860542297, "learning_rate": 6.560135148191818e-05, "loss": 1.121532917022705, "step": 16378 }, { "epoch": 1.9058645566674424, "grad_norm": 1.34087073802948, "learning_rate": 6.544712674407191e-05, "loss": 1.5547717809677124, "step": 16379 }, { "epoch": 1.9059809169187805, "grad_norm": 1.0510631799697876, "learning_rate": 6.529305619809082e-05, "loss": 1.4155391454696655, "step": 16380 }, { "epoch": 1.9060972771701188, "grad_norm": 0.7549343109130859, "learning_rate": 6.513913997269907e-05, "loss": 1.2737541198730469, "step": 16381 }, { "epoch": 1.9062136374214569, "grad_norm": 1.2887383699417114, "learning_rate": 6.498537819649206e-05, "loss": 2.135197639465332, "step": 16382 }, { "epoch": 1.906329997672795, "grad_norm": 1.2045706510543823, "learning_rate": 6.483177099793594e-05, "loss": 2.100917100906372, "step": 16383 }, { "epoch": 1.906446357924133, "grad_norm": 1.1826046705245972, "learning_rate": 6.467831850536793e-05, "loss": 1.8924031257629395, "step": 16384 }, { "epoch": 1.9065627181754712, "grad_norm": 1.095035433769226, "learning_rate": 6.452502084699563e-05, "loss": 2.047661304473877, "step": 16385 }, { "epoch": 1.9066790784268095, "grad_norm": 1.135953664779663, "learning_rate": 6.437187815089777e-05, "loss": 1.67531418800354, "step": 16386 }, { "epoch": 1.9067954386781474, "grad_norm": 1.1293151378631592, "learning_rate": 6.421889054502328e-05, "loss": 1.5824120044708252, "step": 16387 }, { "epoch": 1.9069117989294857, "grad_norm": 1.2318544387817383, "learning_rate": 6.406605815719158e-05, "loss": 1.4199870824813843, "step": 16388 }, { "epoch": 1.9070281591808238, "grad_norm": 0.6512810587882996, "learning_rate": 6.391338111509243e-05, "loss": 1.4360272884368896, "step": 16389 }, { "epoch": 1.907144519432162, "grad_norm": 1.571312427520752, "learning_rate": 6.37608595462858e-05, "loss": 2.017878770828247, "step": 16390 }, { "epoch": 1.9072608796835002, "grad_norm": 0.6874610185623169, "learning_rate": 6.360849357820175e-05, "loss": 1.9021480083465576, "step": 16391 }, { "epoch": 1.907377239934838, "grad_norm": 0.8374168872833252, "learning_rate": 6.345628333814038e-05, "loss": 1.6660661697387695, "step": 16392 }, { "epoch": 1.9074936001861764, "grad_norm": 0.7186979055404663, "learning_rate": 6.330422895327159e-05, "loss": 1.078314185142517, "step": 16393 }, { "epoch": 1.9076099604375145, "grad_norm": 1.1389297246932983, "learning_rate": 6.315233055063518e-05, "loss": 1.7744616270065308, "step": 16394 }, { "epoch": 1.9077263206888526, "grad_norm": 1.578070044517517, "learning_rate": 6.300058825714052e-05, "loss": 1.412227988243103, "step": 16395 }, { "epoch": 1.907842680940191, "grad_norm": 1.4471417665481567, "learning_rate": 6.284900219956663e-05, "loss": 2.354257583618164, "step": 16396 }, { "epoch": 1.9079590411915288, "grad_norm": 0.8625450730323792, "learning_rate": 6.269757250456196e-05, "loss": 1.4454835653305054, "step": 16397 }, { "epoch": 1.9080754014428671, "grad_norm": 1.564307451248169, "learning_rate": 6.25462992986443e-05, "loss": 2.184518337249756, "step": 16398 }, { "epoch": 1.9081917616942052, "grad_norm": 0.8681061863899231, "learning_rate": 6.239518270820072e-05, "loss": 1.5308079719543457, "step": 16399 }, { "epoch": 1.9083081219455433, "grad_norm": 0.8916108012199402, "learning_rate": 6.224422285948752e-05, "loss": 1.3720643520355225, "step": 16400 }, { "epoch": 1.9084244821968817, "grad_norm": 0.7447324991226196, "learning_rate": 6.209341987862996e-05, "loss": 1.3864531517028809, "step": 16401 }, { "epoch": 1.9085408424482195, "grad_norm": 1.391153335571289, "learning_rate": 6.19427738916222e-05, "loss": 2.003573417663574, "step": 16402 }, { "epoch": 1.9086572026995579, "grad_norm": 0.6641868948936462, "learning_rate": 6.17922850243273e-05, "loss": 1.4685323238372803, "step": 16403 }, { "epoch": 1.908773562950896, "grad_norm": 0.8210703730583191, "learning_rate": 6.164195340247699e-05, "loss": 1.1867088079452515, "step": 16404 }, { "epoch": 1.908889923202234, "grad_norm": 0.9168957471847534, "learning_rate": 6.149177915167173e-05, "loss": 1.5793001651763916, "step": 16405 }, { "epoch": 1.9090062834535724, "grad_norm": 1.215641736984253, "learning_rate": 6.134176239738037e-05, "loss": 2.3595170974731445, "step": 16406 }, { "epoch": 1.9091226437049102, "grad_norm": 0.8933588266372681, "learning_rate": 6.119190326494028e-05, "loss": 2.259822368621826, "step": 16407 }, { "epoch": 1.9092390039562486, "grad_norm": 0.7037169933319092, "learning_rate": 6.104220187955714e-05, "loss": 1.2273280620574951, "step": 16408 }, { "epoch": 1.9093553642075867, "grad_norm": 1.072531819343567, "learning_rate": 6.0892658366304615e-05, "loss": 1.6544885635375977, "step": 16409 }, { "epoch": 1.9094717244589248, "grad_norm": 1.3001312017440796, "learning_rate": 6.074327285012468e-05, "loss": 2.08027982711792, "step": 16410 }, { "epoch": 1.909588084710263, "grad_norm": 0.973961591720581, "learning_rate": 6.059404545582742e-05, "loss": 1.7495431900024414, "step": 16411 }, { "epoch": 1.909704444961601, "grad_norm": 1.0184576511383057, "learning_rate": 6.044497630809054e-05, "loss": 1.4580631256103516, "step": 16412 }, { "epoch": 1.9098208052129393, "grad_norm": 0.8583409190177917, "learning_rate": 6.0296065531459694e-05, "loss": 1.8780853748321533, "step": 16413 }, { "epoch": 1.9099371654642774, "grad_norm": 0.8561868071556091, "learning_rate": 6.014731325034814e-05, "loss": 1.3279054164886475, "step": 16414 }, { "epoch": 1.9100535257156155, "grad_norm": 0.864230215549469, "learning_rate": 5.9998719589036744e-05, "loss": 2.1183228492736816, "step": 16415 }, { "epoch": 1.9101698859669538, "grad_norm": 0.8287439942359924, "learning_rate": 5.9850284671673886e-05, "loss": 1.4313688278198242, "step": 16416 }, { "epoch": 1.9102862462182917, "grad_norm": 1.1820107698440552, "learning_rate": 5.970200862227526e-05, "loss": 1.813730001449585, "step": 16417 }, { "epoch": 1.91040260646963, "grad_norm": 2.6786012649536133, "learning_rate": 5.955389156472388e-05, "loss": 1.898586392402649, "step": 16418 }, { "epoch": 1.910518966720968, "grad_norm": 3.796820640563965, "learning_rate": 5.940593362276983e-05, "loss": 1.6430543661117554, "step": 16419 }, { "epoch": 1.9106353269723062, "grad_norm": 1.0679597854614258, "learning_rate": 5.9258134920030376e-05, "loss": 1.9994847774505615, "step": 16420 }, { "epoch": 1.9107516872236445, "grad_norm": 0.7777121067047119, "learning_rate": 5.911049557998965e-05, "loss": 1.5847188234329224, "step": 16421 }, { "epoch": 1.9108680474749824, "grad_norm": 0.739382803440094, "learning_rate": 5.896301572599869e-05, "loss": 1.3477569818496704, "step": 16422 }, { "epoch": 1.9109844077263207, "grad_norm": 1.4366673231124878, "learning_rate": 5.881569548127527e-05, "loss": 2.25830078125, "step": 16423 }, { "epoch": 1.9111007679776588, "grad_norm": 0.8824511766433716, "learning_rate": 5.866853496890373e-05, "loss": 1.8614389896392822, "step": 16424 }, { "epoch": 1.911217128228997, "grad_norm": 1.3306435346603394, "learning_rate": 5.852153431183516e-05, "loss": 1.7210614681243896, "step": 16425 }, { "epoch": 1.9113334884803352, "grad_norm": 1.4185577630996704, "learning_rate": 5.837469363288694e-05, "loss": 1.6276018619537354, "step": 16426 }, { "epoch": 1.9114498487316731, "grad_norm": 0.8651497960090637, "learning_rate": 5.822801305474279e-05, "loss": 1.5118786096572876, "step": 16427 }, { "epoch": 1.9115662089830114, "grad_norm": 0.9517539143562317, "learning_rate": 5.808149269995269e-05, "loss": 2.229627847671509, "step": 16428 }, { "epoch": 1.9116825692343495, "grad_norm": 0.9543817639350891, "learning_rate": 5.7935132690932786e-05, "loss": 1.2439594268798828, "step": 16429 }, { "epoch": 1.9117989294856876, "grad_norm": 1.0076392889022827, "learning_rate": 5.778893314996519e-05, "loss": 1.5729000568389893, "step": 16430 }, { "epoch": 1.911915289737026, "grad_norm": 1.1692135334014893, "learning_rate": 5.7642894199198046e-05, "loss": 1.4905306100845337, "step": 16431 }, { "epoch": 1.9120316499883638, "grad_norm": 0.8727883100509644, "learning_rate": 5.7497015960645296e-05, "loss": 2.3435540199279785, "step": 16432 }, { "epoch": 1.9121480102397022, "grad_norm": 2.349355697631836, "learning_rate": 5.7351298556186426e-05, "loss": 1.5543742179870605, "step": 16433 }, { "epoch": 1.9122643704910403, "grad_norm": 1.2995827198028564, "learning_rate": 5.7205742107566796e-05, "loss": 2.265636444091797, "step": 16434 }, { "epoch": 1.9123807307423784, "grad_norm": 0.8002093434333801, "learning_rate": 5.706034673639712e-05, "loss": 1.6864062547683716, "step": 16435 }, { "epoch": 1.9124970909937167, "grad_norm": 0.9632132053375244, "learning_rate": 5.691511256415377e-05, "loss": 1.801491141319275, "step": 16436 }, { "epoch": 1.9126134512450546, "grad_norm": 0.8914538025856018, "learning_rate": 5.6770039712178147e-05, "loss": 1.2606220245361328, "step": 16437 }, { "epoch": 1.9127298114963929, "grad_norm": 1.0742332935333252, "learning_rate": 5.662512830167707e-05, "loss": 1.7320538759231567, "step": 16438 }, { "epoch": 1.912846171747731, "grad_norm": 0.8362167477607727, "learning_rate": 5.6480378453722335e-05, "loss": 1.9027948379516602, "step": 16439 }, { "epoch": 1.912962531999069, "grad_norm": 0.6504484415054321, "learning_rate": 5.633579028925087e-05, "loss": 1.313462734222412, "step": 16440 }, { "epoch": 1.9130788922504074, "grad_norm": 1.0732210874557495, "learning_rate": 5.619136392906446e-05, "loss": 2.144686698913574, "step": 16441 }, { "epoch": 1.9131952525017453, "grad_norm": 1.0272806882858276, "learning_rate": 5.604709949382972e-05, "loss": 1.6736465692520142, "step": 16442 }, { "epoch": 1.9133116127530836, "grad_norm": 1.1741939783096313, "learning_rate": 5.590299710407795e-05, "loss": 1.8115180730819702, "step": 16443 }, { "epoch": 1.9134279730044217, "grad_norm": 1.3948965072631836, "learning_rate": 5.5759056880205135e-05, "loss": 2.1369946002960205, "step": 16444 }, { "epoch": 1.9135443332557598, "grad_norm": 1.1129745244979858, "learning_rate": 5.56152789424717e-05, "loss": 2.029764175415039, "step": 16445 }, { "epoch": 1.913660693507098, "grad_norm": 0.45827844738960266, "learning_rate": 5.547166341100249e-05, "loss": 1.418026328086853, "step": 16446 }, { "epoch": 1.913777053758436, "grad_norm": 1.245302677154541, "learning_rate": 5.53282104057867e-05, "loss": 2.5104665756225586, "step": 16447 }, { "epoch": 1.9138934140097743, "grad_norm": 1.6440781354904175, "learning_rate": 5.518492004667769e-05, "loss": 1.8364161252975464, "step": 16448 }, { "epoch": 1.9140097742611124, "grad_norm": 0.5848422646522522, "learning_rate": 5.504179245339294e-05, "loss": 1.1078495979309082, "step": 16449 }, { "epoch": 1.9141261345124505, "grad_norm": 1.450358271598816, "learning_rate": 5.489882774551403e-05, "loss": 1.6554678678512573, "step": 16450 }, { "epoch": 1.9142424947637888, "grad_norm": 0.7608628273010254, "learning_rate": 5.475602604248636e-05, "loss": 1.5300705432891846, "step": 16451 }, { "epoch": 1.9143588550151267, "grad_norm": 0.5838325619697571, "learning_rate": 5.4613387463619105e-05, "loss": 1.0873926877975464, "step": 16452 }, { "epoch": 1.914475215266465, "grad_norm": 0.7782636284828186, "learning_rate": 5.4470912128085246e-05, "loss": 2.1554839611053467, "step": 16453 }, { "epoch": 1.9145915755178031, "grad_norm": 0.49295151233673096, "learning_rate": 5.432860015492133e-05, "loss": 1.29365873336792, "step": 16454 }, { "epoch": 1.9147079357691412, "grad_norm": 1.1600605249404907, "learning_rate": 5.418645166302743e-05, "loss": 2.057603359222412, "step": 16455 }, { "epoch": 1.9148242960204795, "grad_norm": 1.144240379333496, "learning_rate": 5.404446677116706e-05, "loss": 2.342132329940796, "step": 16456 }, { "epoch": 1.9149406562718174, "grad_norm": 1.3387556076049805, "learning_rate": 5.39026455979669e-05, "loss": 2.3608005046844482, "step": 16457 }, { "epoch": 1.9150570165231557, "grad_norm": 0.756686270236969, "learning_rate": 5.3760988261917024e-05, "loss": 2.074847459793091, "step": 16458 }, { "epoch": 1.9151733767744938, "grad_norm": 0.6539149880409241, "learning_rate": 5.361949488137055e-05, "loss": 1.3415963649749756, "step": 16459 }, { "epoch": 1.915289737025832, "grad_norm": 1.2601840496063232, "learning_rate": 5.347816557454355e-05, "loss": 1.7728406190872192, "step": 16460 }, { "epoch": 1.9154060972771703, "grad_norm": 1.4036270380020142, "learning_rate": 5.33370004595152e-05, "loss": 2.021463632583618, "step": 16461 }, { "epoch": 1.9155224575285081, "grad_norm": 0.8272901177406311, "learning_rate": 5.3195999654227365e-05, "loss": 1.2387917041778564, "step": 16462 }, { "epoch": 1.9156388177798465, "grad_norm": 1.4464945793151855, "learning_rate": 5.3055163276484577e-05, "loss": 2.0005369186401367, "step": 16463 }, { "epoch": 1.9157551780311846, "grad_norm": 1.385299563407898, "learning_rate": 5.2914491443954076e-05, "loss": 1.6337292194366455, "step": 16464 }, { "epoch": 1.9158715382825227, "grad_norm": 1.4982136487960815, "learning_rate": 5.2773984274165635e-05, "loss": 1.8025234937667847, "step": 16465 }, { "epoch": 1.915987898533861, "grad_norm": 0.7644548416137695, "learning_rate": 5.263364188451142e-05, "loss": 1.7733163833618164, "step": 16466 }, { "epoch": 1.9161042587851989, "grad_norm": 0.9208743572235107, "learning_rate": 5.249346439224589e-05, "loss": 2.1692686080932617, "step": 16467 }, { "epoch": 1.9162206190365372, "grad_norm": 0.8415719866752625, "learning_rate": 5.235345191448582e-05, "loss": 2.1292765140533447, "step": 16468 }, { "epoch": 1.9163369792878753, "grad_norm": 1.143101692199707, "learning_rate": 5.221360456821006e-05, "loss": 1.6395363807678223, "step": 16469 }, { "epoch": 1.9164533395392134, "grad_norm": 1.4543956518173218, "learning_rate": 5.2073922470259495e-05, "loss": 1.8434858322143555, "step": 16470 }, { "epoch": 1.9165696997905517, "grad_norm": 0.7483424544334412, "learning_rate": 5.193440573733696e-05, "loss": 1.465472936630249, "step": 16471 }, { "epoch": 1.9166860600418896, "grad_norm": 0.5531404614448547, "learning_rate": 5.179505448600716e-05, "loss": 1.169830083847046, "step": 16472 }, { "epoch": 1.916802420293228, "grad_norm": 1.2084472179412842, "learning_rate": 5.165586883269646e-05, "loss": 1.407548189163208, "step": 16473 }, { "epoch": 1.916918780544566, "grad_norm": 1.3583743572235107, "learning_rate": 5.1516848893692874e-05, "loss": 2.058291435241699, "step": 16474 }, { "epoch": 1.917035140795904, "grad_norm": 0.7014814615249634, "learning_rate": 5.137799478514615e-05, "loss": 1.546095609664917, "step": 16475 }, { "epoch": 1.9171515010472424, "grad_norm": 0.8226176500320435, "learning_rate": 5.123930662306728e-05, "loss": 2.4912424087524414, "step": 16476 }, { "epoch": 1.9172678612985803, "grad_norm": 1.3558259010314941, "learning_rate": 5.110078452332867e-05, "loss": 1.9021570682525635, "step": 16477 }, { "epoch": 1.9173842215499186, "grad_norm": 1.0417766571044922, "learning_rate": 5.096242860166397e-05, "loss": 1.8243365287780762, "step": 16478 }, { "epoch": 1.9175005818012567, "grad_norm": 1.5127085447311401, "learning_rate": 5.082423897366803e-05, "loss": 1.8487554788589478, "step": 16479 }, { "epoch": 1.9176169420525948, "grad_norm": 1.3549830913543701, "learning_rate": 5.0686215754796816e-05, "loss": 1.7880783081054688, "step": 16480 }, { "epoch": 1.9177333023039331, "grad_norm": 0.8069614171981812, "learning_rate": 5.054835906036701e-05, "loss": 1.314772129058838, "step": 16481 }, { "epoch": 1.917849662555271, "grad_norm": 1.095361590385437, "learning_rate": 5.0410669005556465e-05, "loss": 2.2284560203552246, "step": 16482 }, { "epoch": 1.9179660228066093, "grad_norm": 1.227738380432129, "learning_rate": 5.0273145705403646e-05, "loss": 1.7597424983978271, "step": 16483 }, { "epoch": 1.9180823830579474, "grad_norm": 0.9210007190704346, "learning_rate": 5.013578927480772e-05, "loss": 1.430715799331665, "step": 16484 }, { "epoch": 1.9181987433092855, "grad_norm": 0.5990397334098816, "learning_rate": 4.9998599828528445e-05, "loss": 1.536833643913269, "step": 16485 }, { "epoch": 1.9183151035606238, "grad_norm": 1.990678310394287, "learning_rate": 4.9861577481186164e-05, "loss": 1.9854629039764404, "step": 16486 }, { "epoch": 1.9184314638119617, "grad_norm": 1.02619206905365, "learning_rate": 4.972472234726144e-05, "loss": 1.3953449726104736, "step": 16487 }, { "epoch": 1.9185478240633, "grad_norm": 0.8289431929588318, "learning_rate": 4.958803454109526e-05, "loss": 1.710512638092041, "step": 16488 }, { "epoch": 1.9186641843146381, "grad_norm": 1.122894525527954, "learning_rate": 4.945151417688867e-05, "loss": 1.5498154163360596, "step": 16489 }, { "epoch": 1.9187805445659762, "grad_norm": 0.8630138039588928, "learning_rate": 4.931516136870301e-05, "loss": 2.0546061992645264, "step": 16490 }, { "epoch": 1.9188969048173146, "grad_norm": 0.8475006818771362, "learning_rate": 4.91789762304595e-05, "loss": 1.3804377317428589, "step": 16491 }, { "epoch": 1.9190132650686524, "grad_norm": 1.218714952468872, "learning_rate": 4.9042958875939264e-05, "loss": 2.0193986892700195, "step": 16492 }, { "epoch": 1.9191296253199908, "grad_norm": 1.3536279201507568, "learning_rate": 4.890710941878329e-05, "loss": 2.2348721027374268, "step": 16493 }, { "epoch": 1.9192459855713289, "grad_norm": 1.1358569860458374, "learning_rate": 4.877142797249234e-05, "loss": 2.0619499683380127, "step": 16494 }, { "epoch": 1.919362345822667, "grad_norm": 1.219594955444336, "learning_rate": 4.863591465042669e-05, "loss": 1.8868627548217773, "step": 16495 }, { "epoch": 1.9194787060740053, "grad_norm": 0.9876429438591003, "learning_rate": 4.85005695658062e-05, "loss": 1.9856727123260498, "step": 16496 }, { "epoch": 1.9195950663253432, "grad_norm": 0.7738869786262512, "learning_rate": 4.836539283171024e-05, "loss": 1.8620762825012207, "step": 16497 }, { "epoch": 1.9197114265766815, "grad_norm": 1.399328589439392, "learning_rate": 4.8230384561077424e-05, "loss": 1.1686265468597412, "step": 16498 }, { "epoch": 1.9198277868280196, "grad_norm": 1.1590896844863892, "learning_rate": 4.809554486670556e-05, "loss": 2.048292398452759, "step": 16499 }, { "epoch": 1.9199441470793577, "grad_norm": 0.8866741061210632, "learning_rate": 4.796087386125189e-05, "loss": 2.0037600994110107, "step": 16500 }, { "epoch": 1.920060507330696, "grad_norm": 1.329655647277832, "learning_rate": 4.782637165723244e-05, "loss": 1.8346666097640991, "step": 16501 }, { "epoch": 1.9201768675820339, "grad_norm": 0.8529329895973206, "learning_rate": 4.7692038367022324e-05, "loss": 1.629409909248352, "step": 16502 }, { "epoch": 1.9202932278333722, "grad_norm": 0.8211610317230225, "learning_rate": 4.755787410285548e-05, "loss": 1.398681879043579, "step": 16503 }, { "epoch": 1.9204095880847103, "grad_norm": 0.9622669219970703, "learning_rate": 4.742387897682465e-05, "loss": 1.9650976657867432, "step": 16504 }, { "epoch": 1.9205259483360484, "grad_norm": 1.1817647218704224, "learning_rate": 4.729005310088139e-05, "loss": 1.70014488697052, "step": 16505 }, { "epoch": 1.9206423085873867, "grad_norm": 1.329630732536316, "learning_rate": 4.715639658683549e-05, "loss": 2.0489888191223145, "step": 16506 }, { "epoch": 1.9207586688387246, "grad_norm": 0.88807213306427, "learning_rate": 4.702290954635563e-05, "loss": 1.3715629577636719, "step": 16507 }, { "epoch": 1.920875029090063, "grad_norm": 1.2590986490249634, "learning_rate": 4.688959209096871e-05, "loss": 1.6568796634674072, "step": 16508 }, { "epoch": 1.920991389341401, "grad_norm": 0.83356112241745, "learning_rate": 4.6756444332059974e-05, "loss": 1.8866667747497559, "step": 16509 }, { "epoch": 1.9211077495927391, "grad_norm": 0.8329506516456604, "learning_rate": 4.6623466380872836e-05, "loss": 1.6130380630493164, "step": 16510 }, { "epoch": 1.9212241098440772, "grad_norm": 2.2112183570861816, "learning_rate": 4.649065834850902e-05, "loss": 2.4320168495178223, "step": 16511 }, { "epoch": 1.9213404700954153, "grad_norm": 1.5288790464401245, "learning_rate": 4.635802034592804e-05, "loss": 1.7657132148742676, "step": 16512 }, { "epoch": 1.9214568303467536, "grad_norm": 0.914696991443634, "learning_rate": 4.6225552483947526e-05, "loss": 1.390716552734375, "step": 16513 }, { "epoch": 1.9215731905980917, "grad_norm": 1.0048755407333374, "learning_rate": 4.609325487324289e-05, "loss": 1.7078516483306885, "step": 16514 }, { "epoch": 1.9216895508494298, "grad_norm": 1.061572551727295, "learning_rate": 4.596112762434729e-05, "loss": 2.09377384185791, "step": 16515 }, { "epoch": 1.921805911100768, "grad_norm": 0.9211715459823608, "learning_rate": 4.582917084765156e-05, "loss": 2.010565757751465, "step": 16516 }, { "epoch": 1.921922271352106, "grad_norm": 1.0456411838531494, "learning_rate": 4.5697384653404146e-05, "loss": 1.627816915512085, "step": 16517 }, { "epoch": 1.9220386316034443, "grad_norm": 0.43706467747688293, "learning_rate": 4.55657691517109e-05, "loss": 1.074623703956604, "step": 16518 }, { "epoch": 1.9221549918547824, "grad_norm": 1.0326220989227295, "learning_rate": 4.543432445253515e-05, "loss": 1.9404761791229248, "step": 16519 }, { "epoch": 1.9222713521061205, "grad_norm": 0.9590874314308167, "learning_rate": 4.530305066569742e-05, "loss": 1.456479549407959, "step": 16520 }, { "epoch": 1.9223877123574586, "grad_norm": 1.0989826917648315, "learning_rate": 4.517194790087556e-05, "loss": 1.601418375968933, "step": 16521 }, { "epoch": 1.9225040726087967, "grad_norm": 0.9712986946105957, "learning_rate": 4.504101626760443e-05, "loss": 1.7037547826766968, "step": 16522 }, { "epoch": 1.922620432860135, "grad_norm": 1.479459285736084, "learning_rate": 4.4910255875275946e-05, "loss": 1.6188075542449951, "step": 16523 }, { "epoch": 1.9227367931114732, "grad_norm": 0.878883957862854, "learning_rate": 4.477966683313889e-05, "loss": 1.5363435745239258, "step": 16524 }, { "epoch": 1.9228531533628113, "grad_norm": 1.0071486234664917, "learning_rate": 4.464924925029906e-05, "loss": 2.040564775466919, "step": 16525 }, { "epoch": 1.9229695136141494, "grad_norm": 1.00980544090271, "learning_rate": 4.451900323571886e-05, "loss": 1.6450319290161133, "step": 16526 }, { "epoch": 1.9230858738654875, "grad_norm": 1.3858132362365723, "learning_rate": 4.438892889821739e-05, "loss": 1.8042564392089844, "step": 16527 }, { "epoch": 1.9232022341168258, "grad_norm": 1.2584623098373413, "learning_rate": 4.425902634647028e-05, "loss": 1.4293632507324219, "step": 16528 }, { "epoch": 1.9233185943681639, "grad_norm": 1.0206685066223145, "learning_rate": 4.4129295689009726e-05, "loss": 1.5340056419372559, "step": 16529 }, { "epoch": 1.923434954619502, "grad_norm": 0.5813087224960327, "learning_rate": 4.399973703422411e-05, "loss": 1.5803704261779785, "step": 16530 }, { "epoch": 1.92355131487084, "grad_norm": 0.6000463962554932, "learning_rate": 4.387035049035834e-05, "loss": 1.4075098037719727, "step": 16531 }, { "epoch": 1.9236676751221782, "grad_norm": 1.1259346008300781, "learning_rate": 4.3741136165513394e-05, "loss": 1.884935736656189, "step": 16532 }, { "epoch": 1.9237840353735165, "grad_norm": 1.1293262243270874, "learning_rate": 4.361209416764641e-05, "loss": 1.4189989566802979, "step": 16533 }, { "epoch": 1.9239003956248546, "grad_norm": 1.0184050798416138, "learning_rate": 4.348322460457055e-05, "loss": 1.8688907623291016, "step": 16534 }, { "epoch": 1.9240167558761927, "grad_norm": 0.782150387763977, "learning_rate": 4.335452758395481e-05, "loss": 1.2293004989624023, "step": 16535 }, { "epoch": 1.9241331161275308, "grad_norm": 0.6241796016693115, "learning_rate": 4.3226003213324235e-05, "loss": 1.6201667785644531, "step": 16536 }, { "epoch": 1.924249476378869, "grad_norm": 0.6313681602478027, "learning_rate": 4.309765160005946e-05, "loss": 1.1963858604431152, "step": 16537 }, { "epoch": 1.9243658366302072, "grad_norm": 3.3105971813201904, "learning_rate": 4.2969472851396826e-05, "loss": 1.5312819480895996, "step": 16538 }, { "epoch": 1.9244821968815453, "grad_norm": 0.8438129425048828, "learning_rate": 4.284146707442824e-05, "loss": 2.0671143531799316, "step": 16539 }, { "epoch": 1.9245985571328834, "grad_norm": 0.9995755553245544, "learning_rate": 4.271363437610107e-05, "loss": 1.3119444847106934, "step": 16540 }, { "epoch": 1.9247149173842215, "grad_norm": 0.685161292552948, "learning_rate": 4.2585974863218165e-05, "loss": 1.96376633644104, "step": 16541 }, { "epoch": 1.9248312776355596, "grad_norm": 0.6529319286346436, "learning_rate": 4.2458488642437594e-05, "loss": 1.6001431941986084, "step": 16542 }, { "epoch": 1.924947637886898, "grad_norm": 0.4440767168998718, "learning_rate": 4.233117582027265e-05, "loss": 1.2500438690185547, "step": 16543 }, { "epoch": 1.925063998138236, "grad_norm": 0.7582990527153015, "learning_rate": 4.220403650309182e-05, "loss": 1.5876314640045166, "step": 16544 }, { "epoch": 1.9251803583895741, "grad_norm": 1.6485919952392578, "learning_rate": 4.207707079711853e-05, "loss": 2.117727279663086, "step": 16545 }, { "epoch": 1.9252967186409122, "grad_norm": 1.7832202911376953, "learning_rate": 4.1950278808431274e-05, "loss": 2.6194660663604736, "step": 16546 }, { "epoch": 1.9254130788922503, "grad_norm": 0.41067036986351013, "learning_rate": 4.182366064296328e-05, "loss": 1.1886184215545654, "step": 16547 }, { "epoch": 1.9255294391435887, "grad_norm": 0.5970348715782166, "learning_rate": 4.1697216406502646e-05, "loss": 1.0565185546875, "step": 16548 }, { "epoch": 1.9256457993949265, "grad_norm": 1.5133970975875854, "learning_rate": 4.157094620469207e-05, "loss": 2.2543492317199707, "step": 16549 }, { "epoch": 1.9257621596462648, "grad_norm": 1.1559481620788574, "learning_rate": 4.1444850143028984e-05, "loss": 1.9213744401931763, "step": 16550 }, { "epoch": 1.925878519897603, "grad_norm": 0.5930035710334778, "learning_rate": 4.131892832686518e-05, "loss": 1.4333295822143555, "step": 16551 }, { "epoch": 1.925994880148941, "grad_norm": 1.8889455795288086, "learning_rate": 4.119318086140697e-05, "loss": 1.765930414199829, "step": 16552 }, { "epoch": 1.9261112404002794, "grad_norm": 0.9281548261642456, "learning_rate": 4.1067607851714985e-05, "loss": 2.1256585121154785, "step": 16553 }, { "epoch": 1.9262276006516172, "grad_norm": 1.3962258100509644, "learning_rate": 4.094220940270396e-05, "loss": 1.9750645160675049, "step": 16554 }, { "epoch": 1.9263439609029556, "grad_norm": 1.2255802154541016, "learning_rate": 4.081698561914296e-05, "loss": 2.186737060546875, "step": 16555 }, { "epoch": 1.9264603211542937, "grad_norm": 1.156559705734253, "learning_rate": 4.069193660565507e-05, "loss": 2.1118392944335938, "step": 16556 }, { "epoch": 1.9265766814056318, "grad_norm": 1.194810152053833, "learning_rate": 4.056706246671729e-05, "loss": 2.3928322792053223, "step": 16557 }, { "epoch": 1.92669304165697, "grad_norm": 1.3081938028335571, "learning_rate": 4.044236330666065e-05, "loss": 1.869224190711975, "step": 16558 }, { "epoch": 1.926809401908308, "grad_norm": 0.937804639339447, "learning_rate": 4.031783922966986e-05, "loss": 1.682616949081421, "step": 16559 }, { "epoch": 1.9269257621596463, "grad_norm": 1.2524125576019287, "learning_rate": 4.0193490339783337e-05, "loss": 2.5653343200683594, "step": 16560 }, { "epoch": 1.9270421224109844, "grad_norm": 0.6513866782188416, "learning_rate": 4.006931674089331e-05, "loss": 1.3438522815704346, "step": 16561 }, { "epoch": 1.9271584826623225, "grad_norm": 0.5805686116218567, "learning_rate": 3.994531853674538e-05, "loss": 1.5270782709121704, "step": 16562 }, { "epoch": 1.9272748429136608, "grad_norm": 0.9938287734985352, "learning_rate": 3.982149583093866e-05, "loss": 1.853136420249939, "step": 16563 }, { "epoch": 1.9273912031649987, "grad_norm": 0.8771350979804993, "learning_rate": 3.969784872692567e-05, "loss": 2.019496440887451, "step": 16564 }, { "epoch": 1.927507563416337, "grad_norm": 1.4115185737609863, "learning_rate": 3.957437732801214e-05, "loss": 2.3549556732177734, "step": 16565 }, { "epoch": 1.927623923667675, "grad_norm": 1.2734521627426147, "learning_rate": 3.9451081737357035e-05, "loss": 1.81206476688385, "step": 16566 }, { "epoch": 1.9277402839190132, "grad_norm": 1.0864373445510864, "learning_rate": 3.932796205797251e-05, "loss": 1.4085744619369507, "step": 16567 }, { "epoch": 1.9278566441703515, "grad_norm": 0.9868630766868591, "learning_rate": 3.920501839272361e-05, "loss": 2.2274560928344727, "step": 16568 }, { "epoch": 1.9279730044216894, "grad_norm": 1.244563102722168, "learning_rate": 3.908225084432845e-05, "loss": 2.121826171875, "step": 16569 }, { "epoch": 1.9280893646730277, "grad_norm": 1.1249834299087524, "learning_rate": 3.895965951535793e-05, "loss": 2.0413339138031006, "step": 16570 }, { "epoch": 1.9282057249243658, "grad_norm": 0.6356853246688843, "learning_rate": 3.883724450823573e-05, "loss": 1.541856050491333, "step": 16571 }, { "epoch": 1.928322085175704, "grad_norm": 0.7055017948150635, "learning_rate": 3.871500592523819e-05, "loss": 2.091256618499756, "step": 16572 }, { "epoch": 1.9284384454270422, "grad_norm": 0.9684703946113586, "learning_rate": 3.859294386849435e-05, "loss": 2.1909327507019043, "step": 16573 }, { "epoch": 1.9285548056783801, "grad_norm": 1.2450302839279175, "learning_rate": 3.8471058439985576e-05, "loss": 2.4783425331115723, "step": 16574 }, { "epoch": 1.9286711659297184, "grad_norm": 1.0265146493911743, "learning_rate": 3.834934974154594e-05, "loss": 1.7913705110549927, "step": 16575 }, { "epoch": 1.9287875261810565, "grad_norm": 1.320204496383667, "learning_rate": 3.8227817874861625e-05, "loss": 1.7575531005859375, "step": 16576 }, { "epoch": 1.9289038864323946, "grad_norm": 1.3077110052108765, "learning_rate": 3.810646294147121e-05, "loss": 1.7504006624221802, "step": 16577 }, { "epoch": 1.929020246683733, "grad_norm": 0.9080748558044434, "learning_rate": 3.798528504276527e-05, "loss": 1.5331718921661377, "step": 16578 }, { "epoch": 1.9291366069350708, "grad_norm": 1.0842938423156738, "learning_rate": 3.7864284279986645e-05, "loss": 1.355354905128479, "step": 16579 }, { "epoch": 1.9292529671864092, "grad_norm": 1.0156428813934326, "learning_rate": 3.77434607542301e-05, "loss": 1.5905838012695312, "step": 16580 }, { "epoch": 1.9293693274377473, "grad_norm": 0.951562762260437, "learning_rate": 3.7622814566442374e-05, "loss": 2.038670063018799, "step": 16581 }, { "epoch": 1.9294856876890853, "grad_norm": 0.693671464920044, "learning_rate": 3.7502345817422015e-05, "loss": 1.5691875219345093, "step": 16582 }, { "epoch": 1.9296020479404237, "grad_norm": 0.8501139879226685, "learning_rate": 3.738205460781929e-05, "loss": 1.578288197517395, "step": 16583 }, { "epoch": 1.9297184081917615, "grad_norm": 1.8730502128601074, "learning_rate": 3.726194103813618e-05, "loss": 1.6579099893569946, "step": 16584 }, { "epoch": 1.9298347684430999, "grad_norm": 0.8559530377388, "learning_rate": 3.7142005208726155e-05, "loss": 1.5647239685058594, "step": 16585 }, { "epoch": 1.929951128694438, "grad_norm": 1.4560409784317017, "learning_rate": 3.7022247219794385e-05, "loss": 2.073354482650757, "step": 16586 }, { "epoch": 1.930067488945776, "grad_norm": 1.2119792699813843, "learning_rate": 3.6902667171397284e-05, "loss": 2.2522647380828857, "step": 16587 }, { "epoch": 1.9301838491971144, "grad_norm": 1.0531467199325562, "learning_rate": 3.678326516344263e-05, "loss": 1.901342511177063, "step": 16588 }, { "epoch": 1.9303002094484523, "grad_norm": 0.8065428733825684, "learning_rate": 3.666404129568951e-05, "loss": 2.129093647003174, "step": 16589 }, { "epoch": 1.9304165696997906, "grad_norm": 1.095278024673462, "learning_rate": 3.654499566774808e-05, "loss": 2.0335607528686523, "step": 16590 }, { "epoch": 1.9305329299511287, "grad_norm": 2.031906843185425, "learning_rate": 3.642612837907966e-05, "loss": 1.9776101112365723, "step": 16591 }, { "epoch": 1.9306492902024668, "grad_norm": 1.4501404762268066, "learning_rate": 3.630743952899654e-05, "loss": 2.698683738708496, "step": 16592 }, { "epoch": 1.930765650453805, "grad_norm": 0.9189066290855408, "learning_rate": 3.618892921666195e-05, "loss": 1.4487712383270264, "step": 16593 }, { "epoch": 1.930882010705143, "grad_norm": 1.0685607194900513, "learning_rate": 3.6070597541089876e-05, "loss": 2.087812900543213, "step": 16594 }, { "epoch": 1.9309983709564813, "grad_norm": 2.8712592124938965, "learning_rate": 3.595244460114516e-05, "loss": 1.5640411376953125, "step": 16595 }, { "epoch": 1.9311147312078194, "grad_norm": 1.0695744752883911, "learning_rate": 3.583447049554328e-05, "loss": 1.9595401287078857, "step": 16596 }, { "epoch": 1.9312310914591575, "grad_norm": 1.425829291343689, "learning_rate": 3.571667532285022e-05, "loss": 2.080648422241211, "step": 16597 }, { "epoch": 1.9313474517104958, "grad_norm": 1.1561758518218994, "learning_rate": 3.559905918148259e-05, "loss": 2.328540802001953, "step": 16598 }, { "epoch": 1.9314638119618337, "grad_norm": 2.1046648025512695, "learning_rate": 3.548162216970727e-05, "loss": 2.09171199798584, "step": 16599 }, { "epoch": 1.931580172213172, "grad_norm": 1.3148678541183472, "learning_rate": 3.53643643856417e-05, "loss": 2.0806033611297607, "step": 16600 }, { "epoch": 1.9316965324645101, "grad_norm": 0.9242791533470154, "learning_rate": 3.524728592725346e-05, "loss": 1.716312050819397, "step": 16601 }, { "epoch": 1.9318128927158482, "grad_norm": 0.9470913410186768, "learning_rate": 3.513038689236015e-05, "loss": 1.8998955488204956, "step": 16602 }, { "epoch": 1.9319292529671865, "grad_norm": 0.7244907021522522, "learning_rate": 3.501366737862971e-05, "loss": 1.7445404529571533, "step": 16603 }, { "epoch": 1.9320456132185244, "grad_norm": 0.9233198165893555, "learning_rate": 3.4897127483579955e-05, "loss": 1.7064194679260254, "step": 16604 }, { "epoch": 1.9321619734698627, "grad_norm": 1.1861536502838135, "learning_rate": 3.4780767304578656e-05, "loss": 1.789407730102539, "step": 16605 }, { "epoch": 1.9322783337212008, "grad_norm": 1.0138962268829346, "learning_rate": 3.46645869388435e-05, "loss": 1.7419039011001587, "step": 16606 }, { "epoch": 1.932394693972539, "grad_norm": 0.814043402671814, "learning_rate": 3.454858648344184e-05, "loss": 1.8544113636016846, "step": 16607 }, { "epoch": 1.9325110542238773, "grad_norm": 0.8787824511528015, "learning_rate": 3.443276603529075e-05, "loss": 1.6029466390609741, "step": 16608 }, { "epoch": 1.9326274144752151, "grad_norm": 0.9381924271583557, "learning_rate": 3.431712569115697e-05, "loss": 1.9796205759048462, "step": 16609 }, { "epoch": 1.9327437747265535, "grad_norm": 1.2367459535598755, "learning_rate": 3.4201665547656605e-05, "loss": 1.865978479385376, "step": 16610 }, { "epoch": 1.9328601349778916, "grad_norm": 0.664634644985199, "learning_rate": 3.40863857012555e-05, "loss": 1.710362195968628, "step": 16611 }, { "epoch": 1.9329764952292297, "grad_norm": 0.8899794220924377, "learning_rate": 3.3971286248268534e-05, "loss": 1.4615058898925781, "step": 16612 }, { "epoch": 1.933092855480568, "grad_norm": 1.342569351196289, "learning_rate": 3.385636728486008e-05, "loss": 1.6558020114898682, "step": 16613 }, { "epoch": 1.9332092157319059, "grad_norm": 1.0537264347076416, "learning_rate": 3.374162890704363e-05, "loss": 1.9510289430618286, "step": 16614 }, { "epoch": 1.9333255759832442, "grad_norm": 1.7760306596755981, "learning_rate": 3.3627071210681816e-05, "loss": 2.3360347747802734, "step": 16615 }, { "epoch": 1.9334419362345823, "grad_norm": 1.1471461057662964, "learning_rate": 3.3512694291486316e-05, "loss": 1.8834309577941895, "step": 16616 }, { "epoch": 1.9335582964859204, "grad_norm": 1.542592167854309, "learning_rate": 3.339849824501778e-05, "loss": 2.403104782104492, "step": 16617 }, { "epoch": 1.9336746567372587, "grad_norm": 1.3308106660842896, "learning_rate": 3.32844831666857e-05, "loss": 1.9240059852600098, "step": 16618 }, { "epoch": 1.9337910169885966, "grad_norm": 0.9410421252250671, "learning_rate": 3.3170649151748444e-05, "loss": 1.7361748218536377, "step": 16619 }, { "epoch": 1.9339073772399349, "grad_norm": 0.9136552810668945, "learning_rate": 3.3056996295313003e-05, "loss": 1.362767219543457, "step": 16620 }, { "epoch": 1.934023737491273, "grad_norm": 1.0678095817565918, "learning_rate": 3.294352469233514e-05, "loss": 2.286954879760742, "step": 16621 }, { "epoch": 1.934140097742611, "grad_norm": 0.8175641298294067, "learning_rate": 3.283023443761904e-05, "loss": 1.3496894836425781, "step": 16622 }, { "epoch": 1.9342564579939494, "grad_norm": 1.6267184019088745, "learning_rate": 3.271712562581749e-05, "loss": 2.2979815006256104, "step": 16623 }, { "epoch": 1.9343728182452873, "grad_norm": 0.48612549901008606, "learning_rate": 3.260419835143161e-05, "loss": 1.2215348482131958, "step": 16624 }, { "epoch": 1.9344891784966256, "grad_norm": 0.9201804995536804, "learning_rate": 3.249145270881096e-05, "loss": 1.629508376121521, "step": 16625 }, { "epoch": 1.9346055387479637, "grad_norm": 0.9658105969429016, "learning_rate": 3.237888879215317e-05, "loss": 1.5796985626220703, "step": 16626 }, { "epoch": 1.9347218989993018, "grad_norm": 0.9690549969673157, "learning_rate": 3.2266506695504184e-05, "loss": 1.7782713174819946, "step": 16627 }, { "epoch": 1.9348382592506401, "grad_norm": 1.041969656944275, "learning_rate": 3.215430651275794e-05, "loss": 1.6153556108474731, "step": 16628 }, { "epoch": 1.934954619501978, "grad_norm": 0.9945124387741089, "learning_rate": 3.204228833765649e-05, "loss": 1.8785912990570068, "step": 16629 }, { "epoch": 1.9350709797533163, "grad_norm": 1.227165937423706, "learning_rate": 3.193045226378977e-05, "loss": 2.100318431854248, "step": 16630 }, { "epoch": 1.9351873400046544, "grad_norm": 1.228140115737915, "learning_rate": 3.181879838459553e-05, "loss": 1.9294037818908691, "step": 16631 }, { "epoch": 1.9353037002559925, "grad_norm": 1.0099380016326904, "learning_rate": 3.170732679335936e-05, "loss": 1.6052234172821045, "step": 16632 }, { "epoch": 1.9354200605073308, "grad_norm": 0.9710798263549805, "learning_rate": 3.1596037583214556e-05, "loss": 1.7699435949325562, "step": 16633 }, { "epoch": 1.9355364207586687, "grad_norm": 0.23280929028987885, "learning_rate": 3.148493084714196e-05, "loss": 0.8599789142608643, "step": 16634 }, { "epoch": 1.935652781010007, "grad_norm": 1.3670467138290405, "learning_rate": 3.1374006677969966e-05, "loss": 2.3413197994232178, "step": 16635 }, { "epoch": 1.9357691412613451, "grad_norm": 0.9679391980171204, "learning_rate": 3.126326516837463e-05, "loss": 1.6962077617645264, "step": 16636 }, { "epoch": 1.9358855015126832, "grad_norm": 0.9403186440467834, "learning_rate": 3.115270641087917e-05, "loss": 1.5972899198532104, "step": 16637 }, { "epoch": 1.9360018617640216, "grad_norm": 1.5413600206375122, "learning_rate": 3.104233049785418e-05, "loss": 2.0603187084198, "step": 16638 }, { "epoch": 1.9361182220153594, "grad_norm": 0.9793445467948914, "learning_rate": 3.0932137521517524e-05, "loss": 1.802337884902954, "step": 16639 }, { "epoch": 1.9362345822666978, "grad_norm": 0.7045362591743469, "learning_rate": 3.082212757393418e-05, "loss": 1.088858962059021, "step": 16640 }, { "epoch": 1.9363509425180359, "grad_norm": 0.862669825553894, "learning_rate": 3.0712300747016226e-05, "loss": 1.748358964920044, "step": 16641 }, { "epoch": 1.936467302769374, "grad_norm": 0.6030069589614868, "learning_rate": 3.060265713252278e-05, "loss": 1.1882457733154297, "step": 16642 }, { "epoch": 1.9365836630207123, "grad_norm": 1.404817819595337, "learning_rate": 3.0493196822059865e-05, "loss": 1.5152634382247925, "step": 16643 }, { "epoch": 1.9367000232720502, "grad_norm": 1.0689527988433838, "learning_rate": 3.038391990708031e-05, "loss": 1.9080135822296143, "step": 16644 }, { "epoch": 1.9368163835233885, "grad_norm": 0.8360065221786499, "learning_rate": 3.0274826478883773e-05, "loss": 1.80670166015625, "step": 16645 }, { "epoch": 1.9369327437747266, "grad_norm": 2.897264003753662, "learning_rate": 3.016591662861659e-05, "loss": 2.234548330307007, "step": 16646 }, { "epoch": 1.9370491040260647, "grad_norm": 0.9911261200904846, "learning_rate": 3.0057190447271733e-05, "loss": 2.3922266960144043, "step": 16647 }, { "epoch": 1.937165464277403, "grad_norm": 1.3607510328292847, "learning_rate": 2.9948648025688725e-05, "loss": 1.7933180332183838, "step": 16648 }, { "epoch": 1.9372818245287409, "grad_norm": 1.3071861267089844, "learning_rate": 2.9840289454553526e-05, "loss": 1.7529051303863525, "step": 16649 }, { "epoch": 1.9373981847800792, "grad_norm": 1.0292131900787354, "learning_rate": 2.9732114824398527e-05, "loss": 1.9732292890548706, "step": 16650 }, { "epoch": 1.9375145450314173, "grad_norm": 0.8584796786308289, "learning_rate": 2.96241242256024e-05, "loss": 2.3306291103363037, "step": 16651 }, { "epoch": 1.9376309052827554, "grad_norm": 0.8597685098648071, "learning_rate": 2.9516317748390113e-05, "loss": 1.7899582386016846, "step": 16652 }, { "epoch": 1.9377472655340937, "grad_norm": 1.4126582145690918, "learning_rate": 2.9408695482832786e-05, "loss": 1.7331721782684326, "step": 16653 }, { "epoch": 1.9378636257854316, "grad_norm": 0.4247773587703705, "learning_rate": 2.9301257518847575e-05, "loss": 0.9927431344985962, "step": 16654 }, { "epoch": 1.93797998603677, "grad_norm": 2.5938820838928223, "learning_rate": 2.9194003946197727e-05, "loss": 1.7560685873031616, "step": 16655 }, { "epoch": 1.938096346288108, "grad_norm": 1.1080883741378784, "learning_rate": 2.9086934854492374e-05, "loss": 1.8809515237808228, "step": 16656 }, { "epoch": 1.938212706539446, "grad_norm": 1.0239206552505493, "learning_rate": 2.8980050333186597e-05, "loss": 1.8654167652130127, "step": 16657 }, { "epoch": 1.9383290667907844, "grad_norm": 1.1165552139282227, "learning_rate": 2.8873350471581213e-05, "loss": 1.8075820207595825, "step": 16658 }, { "epoch": 1.9384454270421223, "grad_norm": 1.2124097347259521, "learning_rate": 2.8766835358822712e-05, "loss": 1.5910792350769043, "step": 16659 }, { "epoch": 1.9385617872934606, "grad_norm": 1.0624794960021973, "learning_rate": 2.8660505083903277e-05, "loss": 1.9944043159484863, "step": 16660 }, { "epoch": 1.9386781475447987, "grad_norm": 1.184279203414917, "learning_rate": 2.8554359735660756e-05, "loss": 2.160850763320923, "step": 16661 }, { "epoch": 1.9387945077961368, "grad_norm": 0.4592994451522827, "learning_rate": 2.844839940277838e-05, "loss": 1.237567663192749, "step": 16662 }, { "epoch": 1.9389108680474751, "grad_norm": 0.9769112467765808, "learning_rate": 2.8342624173784748e-05, "loss": 2.1267809867858887, "step": 16663 }, { "epoch": 1.939027228298813, "grad_norm": 0.5182191133499146, "learning_rate": 2.8237034137053973e-05, "loss": 1.0972322225570679, "step": 16664 }, { "epoch": 1.9391435885501513, "grad_norm": 0.8822106719017029, "learning_rate": 2.8131629380805298e-05, "loss": 1.8041530847549438, "step": 16665 }, { "epoch": 1.9392599488014894, "grad_norm": 0.7034659385681152, "learning_rate": 2.8026409993103214e-05, "loss": 1.458165168762207, "step": 16666 }, { "epoch": 1.9393763090528275, "grad_norm": 0.9771568775177002, "learning_rate": 2.7921376061857377e-05, "loss": 2.3296701908111572, "step": 16667 }, { "epoch": 1.9394926693041659, "grad_norm": 1.1039621829986572, "learning_rate": 2.7816527674822456e-05, "loss": 1.7412015199661255, "step": 16668 }, { "epoch": 1.9396090295555037, "grad_norm": 1.1609781980514526, "learning_rate": 2.7711864919598097e-05, "loss": 1.5747814178466797, "step": 16669 }, { "epoch": 1.939725389806842, "grad_norm": 1.3348194360733032, "learning_rate": 2.7607387883628847e-05, "loss": 1.9803133010864258, "step": 16670 }, { "epoch": 1.9398417500581802, "grad_norm": 1.0110106468200684, "learning_rate": 2.7503096654204142e-05, "loss": 2.169205665588379, "step": 16671 }, { "epoch": 1.9399581103095183, "grad_norm": 1.2422853708267212, "learning_rate": 2.7398991318458132e-05, "loss": 1.8833249807357788, "step": 16672 }, { "epoch": 1.9400744705608564, "grad_norm": 1.1356077194213867, "learning_rate": 2.729507196336964e-05, "loss": 2.153829336166382, "step": 16673 }, { "epoch": 1.9401908308121945, "grad_norm": 1.234243631362915, "learning_rate": 2.7191338675762122e-05, "loss": 2.276327610015869, "step": 16674 }, { "epoch": 1.9403071910635328, "grad_norm": 0.9525288343429565, "learning_rate": 2.708779154230362e-05, "loss": 2.2137837409973145, "step": 16675 }, { "epoch": 1.9404235513148709, "grad_norm": 1.0597350597381592, "learning_rate": 2.6984430649506593e-05, "loss": 1.6870927810668945, "step": 16676 }, { "epoch": 1.940539911566209, "grad_norm": 0.7223495841026306, "learning_rate": 2.6881256083727905e-05, "loss": 2.0236754417419434, "step": 16677 }, { "epoch": 1.940656271817547, "grad_norm": 0.8467711806297302, "learning_rate": 2.6778267931168747e-05, "loss": 1.4220364093780518, "step": 16678 }, { "epoch": 1.9407726320688852, "grad_norm": 0.8046419620513916, "learning_rate": 2.6675466277874565e-05, "loss": 1.8499438762664795, "step": 16679 }, { "epoch": 1.9408889923202235, "grad_norm": 1.1498022079467773, "learning_rate": 2.6572851209735045e-05, "loss": 2.1874771118164062, "step": 16680 }, { "epoch": 1.9410053525715616, "grad_norm": 0.816127598285675, "learning_rate": 2.6470422812483852e-05, "loss": 2.3453660011291504, "step": 16681 }, { "epoch": 1.9411217128228997, "grad_norm": 1.0218509435653687, "learning_rate": 2.636818117169884e-05, "loss": 1.8128478527069092, "step": 16682 }, { "epoch": 1.9412380730742378, "grad_norm": 0.27398473024368286, "learning_rate": 2.6266126372801708e-05, "loss": 1.0394032001495361, "step": 16683 }, { "epoch": 1.941354433325576, "grad_norm": 1.517002820968628, "learning_rate": 2.6164258501058137e-05, "loss": 1.630211353302002, "step": 16684 }, { "epoch": 1.9414707935769142, "grad_norm": 0.8348343372344971, "learning_rate": 2.6062577641577517e-05, "loss": 1.63959801197052, "step": 16685 }, { "epoch": 1.9415871538282523, "grad_norm": 1.0367456674575806, "learning_rate": 2.59610838793132e-05, "loss": 1.9609712362289429, "step": 16686 }, { "epoch": 1.9417035140795904, "grad_norm": 0.8929171562194824, "learning_rate": 2.5859777299061997e-05, "loss": 1.608047604560852, "step": 16687 }, { "epoch": 1.9418198743309285, "grad_norm": 0.93148273229599, "learning_rate": 2.5758657985464477e-05, "loss": 1.4957356452941895, "step": 16688 }, { "epoch": 1.9419362345822666, "grad_norm": 2.0324087142944336, "learning_rate": 2.5657726023004674e-05, "loss": 2.4497666358947754, "step": 16689 }, { "epoch": 1.942052594833605, "grad_norm": 1.1699047088623047, "learning_rate": 2.55569814960101e-05, "loss": 2.1795506477355957, "step": 16690 }, { "epoch": 1.942168955084943, "grad_norm": 1.0283795595169067, "learning_rate": 2.545642448865171e-05, "loss": 1.9338279962539673, "step": 16691 }, { "epoch": 1.9422853153362811, "grad_norm": 0.9821884036064148, "learning_rate": 2.5356055084943764e-05, "loss": 1.867023229598999, "step": 16692 }, { "epoch": 1.9424016755876192, "grad_norm": 1.0080255270004272, "learning_rate": 2.525587336874374e-05, "loss": 1.8340389728546143, "step": 16693 }, { "epoch": 1.9425180358389573, "grad_norm": 0.9787000417709351, "learning_rate": 2.5155879423752386e-05, "loss": 2.0138449668884277, "step": 16694 }, { "epoch": 1.9426343960902956, "grad_norm": 0.8194695711135864, "learning_rate": 2.505607333351348e-05, "loss": 2.4515557289123535, "step": 16695 }, { "epoch": 1.9427507563416337, "grad_norm": 0.47587308287620544, "learning_rate": 2.495645518141393e-05, "loss": 1.227562427520752, "step": 16696 }, { "epoch": 1.9428671165929718, "grad_norm": 0.9521583318710327, "learning_rate": 2.48570250506836e-05, "loss": 2.25716495513916, "step": 16697 }, { "epoch": 1.94298347684431, "grad_norm": 0.869739294052124, "learning_rate": 2.4757783024395242e-05, "loss": 1.6031900644302368, "step": 16698 }, { "epoch": 1.943099837095648, "grad_norm": 0.9483826756477356, "learning_rate": 2.465872918546444e-05, "loss": 1.7715349197387695, "step": 16699 }, { "epoch": 1.9432161973469864, "grad_norm": 0.9608874320983887, "learning_rate": 2.455986361664958e-05, "loss": 1.8878542184829712, "step": 16700 }, { "epoch": 1.9433325575983245, "grad_norm": 0.6194014549255371, "learning_rate": 2.4461186400551754e-05, "loss": 1.3054673671722412, "step": 16701 }, { "epoch": 1.9434489178496626, "grad_norm": 1.0084844827651978, "learning_rate": 2.4362697619614682e-05, "loss": 1.5854002237319946, "step": 16702 }, { "epoch": 1.9435652781010007, "grad_norm": 1.383565068244934, "learning_rate": 2.4264397356124624e-05, "loss": 2.131004810333252, "step": 16703 }, { "epoch": 1.9436816383523388, "grad_norm": 0.7640290856361389, "learning_rate": 2.4166285692210325e-05, "loss": 1.243073582649231, "step": 16704 }, { "epoch": 1.943797998603677, "grad_norm": 0.9874138832092285, "learning_rate": 2.4068362709843057e-05, "loss": 1.8698220252990723, "step": 16705 }, { "epoch": 1.9439143588550152, "grad_norm": 1.1601067781448364, "learning_rate": 2.397062849083631e-05, "loss": 1.9374072551727295, "step": 16706 }, { "epoch": 1.9440307191063533, "grad_norm": 1.3409688472747803, "learning_rate": 2.3873083116845966e-05, "loss": 2.2644529342651367, "step": 16707 }, { "epoch": 1.9441470793576914, "grad_norm": 0.8768302202224731, "learning_rate": 2.3775726669370112e-05, "loss": 1.5237832069396973, "step": 16708 }, { "epoch": 1.9442634396090295, "grad_norm": 1.0530248880386353, "learning_rate": 2.367855922974893e-05, "loss": 2.4816975593566895, "step": 16709 }, { "epoch": 1.9443797998603678, "grad_norm": 0.8887874484062195, "learning_rate": 2.3581580879164705e-05, "loss": 1.6162313222885132, "step": 16710 }, { "epoch": 1.944496160111706, "grad_norm": 1.2422692775726318, "learning_rate": 2.3484791698641873e-05, "loss": 1.645554780960083, "step": 16711 }, { "epoch": 1.944612520363044, "grad_norm": 0.6349306702613831, "learning_rate": 2.3388191769046633e-05, "loss": 1.8218488693237305, "step": 16712 }, { "epoch": 1.944728880614382, "grad_norm": 1.1793491840362549, "learning_rate": 2.3291781171087185e-05, "loss": 1.5124443769454956, "step": 16713 }, { "epoch": 1.9448452408657202, "grad_norm": 1.4112586975097656, "learning_rate": 2.3195559985313475e-05, "loss": 1.6866058111190796, "step": 16714 }, { "epoch": 1.9449616011170585, "grad_norm": 1.0207204818725586, "learning_rate": 2.3099528292117284e-05, "loss": 2.0859081745147705, "step": 16715 }, { "epoch": 1.9450779613683964, "grad_norm": 0.943208634853363, "learning_rate": 2.3003686171731932e-05, "loss": 1.3658536672592163, "step": 16716 }, { "epoch": 1.9451943216197347, "grad_norm": 0.6807466149330139, "learning_rate": 2.290803370423253e-05, "loss": 2.0187575817108154, "step": 16717 }, { "epoch": 1.9453106818710728, "grad_norm": 0.9858530759811401, "learning_rate": 2.281257096953557e-05, "loss": 1.462207555770874, "step": 16718 }, { "epoch": 1.945427042122411, "grad_norm": 1.4618546962738037, "learning_rate": 2.2717298047399166e-05, "loss": 2.0337886810302734, "step": 16719 }, { "epoch": 1.9455434023737492, "grad_norm": 1.0541316270828247, "learning_rate": 2.2622215017422732e-05, "loss": 1.6122817993164062, "step": 16720 }, { "epoch": 1.945659762625087, "grad_norm": 1.1813528537750244, "learning_rate": 2.252732195904711e-05, "loss": 2.2793874740600586, "step": 16721 }, { "epoch": 1.9457761228764254, "grad_norm": 1.3005746603012085, "learning_rate": 2.243261895155438e-05, "loss": 1.5055365562438965, "step": 16722 }, { "epoch": 1.9458924831277635, "grad_norm": 0.5701727867126465, "learning_rate": 2.2338106074067832e-05, "loss": 1.246889591217041, "step": 16723 }, { "epoch": 1.9460088433791016, "grad_norm": 0.9621323347091675, "learning_rate": 2.2243783405551944e-05, "loss": 2.052480936050415, "step": 16724 }, { "epoch": 1.94612520363044, "grad_norm": 1.2596828937530518, "learning_rate": 2.214965102481223e-05, "loss": 1.6579620838165283, "step": 16725 }, { "epoch": 1.9462415638817778, "grad_norm": 0.8614004850387573, "learning_rate": 2.2055709010495278e-05, "loss": 2.4805662631988525, "step": 16726 }, { "epoch": 1.9463579241331161, "grad_norm": 1.006489872932434, "learning_rate": 2.1961957441088558e-05, "loss": 1.492081880569458, "step": 16727 }, { "epoch": 1.9464742843844542, "grad_norm": 0.9138824939727783, "learning_rate": 2.186839639492047e-05, "loss": 2.0094077587127686, "step": 16728 }, { "epoch": 1.9465906446357923, "grad_norm": 2.106590986251831, "learning_rate": 2.177502595016023e-05, "loss": 2.013395071029663, "step": 16729 }, { "epoch": 1.9467070048871307, "grad_norm": 0.8130543231964111, "learning_rate": 2.1681846184817777e-05, "loss": 1.2983012199401855, "step": 16730 }, { "epoch": 1.9468233651384685, "grad_norm": 1.048598051071167, "learning_rate": 2.1588857176743804e-05, "loss": 1.4406113624572754, "step": 16731 }, { "epoch": 1.9469397253898069, "grad_norm": 1.2485896348953247, "learning_rate": 2.149605900362958e-05, "loss": 1.7890992164611816, "step": 16732 }, { "epoch": 1.947056085641145, "grad_norm": 0.5536385178565979, "learning_rate": 2.1403451743006895e-05, "loss": 1.070833444595337, "step": 16733 }, { "epoch": 1.947172445892483, "grad_norm": 1.3228814601898193, "learning_rate": 2.1311035472248165e-05, "loss": 1.6061899662017822, "step": 16734 }, { "epoch": 1.9472888061438214, "grad_norm": 1.0621007680892944, "learning_rate": 2.1218810268566034e-05, "loss": 1.719691276550293, "step": 16735 }, { "epoch": 1.9474051663951593, "grad_norm": 1.0719504356384277, "learning_rate": 2.112677620901374e-05, "loss": 1.8677252531051636, "step": 16736 }, { "epoch": 1.9475215266464976, "grad_norm": 1.003243088722229, "learning_rate": 2.1034933370484704e-05, "loss": 2.548034191131592, "step": 16737 }, { "epoch": 1.9476378868978357, "grad_norm": 1.0181598663330078, "learning_rate": 2.094328182971253e-05, "loss": 1.3428691625595093, "step": 16738 }, { "epoch": 1.9477542471491738, "grad_norm": 1.2653369903564453, "learning_rate": 2.0851821663271125e-05, "loss": 1.95807683467865, "step": 16739 }, { "epoch": 1.947870607400512, "grad_norm": 1.1493457555770874, "learning_rate": 2.0760552947574367e-05, "loss": 1.741516351699829, "step": 16740 }, { "epoch": 1.94798696765185, "grad_norm": 0.9646558165550232, "learning_rate": 2.0669475758876317e-05, "loss": 2.042931079864502, "step": 16741 }, { "epoch": 1.9481033279031883, "grad_norm": 1.1781799793243408, "learning_rate": 2.0578590173270878e-05, "loss": 1.9582469463348389, "step": 16742 }, { "epoch": 1.9482196881545264, "grad_norm": 1.2881461381912231, "learning_rate": 2.0487896266692006e-05, "loss": 2.167466640472412, "step": 16743 }, { "epoch": 1.9483360484058645, "grad_norm": 1.0805639028549194, "learning_rate": 2.0397394114913425e-05, "loss": 1.9026697874069214, "step": 16744 }, { "epoch": 1.9484524086572028, "grad_norm": 1.1138722896575928, "learning_rate": 2.0307083793548703e-05, "loss": 1.8791635036468506, "step": 16745 }, { "epoch": 1.9485687689085407, "grad_norm": 0.44044893980026245, "learning_rate": 2.0216965378050985e-05, "loss": 0.955511212348938, "step": 16746 }, { "epoch": 1.948685129159879, "grad_norm": 1.5181725025177002, "learning_rate": 2.0127038943713314e-05, "loss": 2.1455917358398438, "step": 16747 }, { "epoch": 1.9488014894112171, "grad_norm": 0.8873538374900818, "learning_rate": 2.0037304565668212e-05, "loss": 1.9761502742767334, "step": 16748 }, { "epoch": 1.9489178496625552, "grad_norm": 0.908360481262207, "learning_rate": 1.994776231888773e-05, "loss": 1.9039232730865479, "step": 16749 }, { "epoch": 1.9490342099138935, "grad_norm": 1.0232192277908325, "learning_rate": 1.9858412278183413e-05, "loss": 2.2243893146514893, "step": 16750 }, { "epoch": 1.9491505701652314, "grad_norm": 1.3137937784194946, "learning_rate": 1.976925451820624e-05, "loss": 2.093035936355591, "step": 16751 }, { "epoch": 1.9492669304165697, "grad_norm": 1.4419885873794556, "learning_rate": 1.968028911344652e-05, "loss": 1.975662350654602, "step": 16752 }, { "epoch": 1.9493832906679078, "grad_norm": 1.2253477573394775, "learning_rate": 1.9591516138233843e-05, "loss": 1.6648764610290527, "step": 16753 }, { "epoch": 1.949499650919246, "grad_norm": 0.5928701758384705, "learning_rate": 1.9502935666737026e-05, "loss": 1.1784478425979614, "step": 16754 }, { "epoch": 1.9496160111705843, "grad_norm": 1.2883803844451904, "learning_rate": 1.9414547772964087e-05, "loss": 1.7798151969909668, "step": 16755 }, { "epoch": 1.9497323714219221, "grad_norm": 1.3806580305099487, "learning_rate": 1.932635253076212e-05, "loss": 1.822072982788086, "step": 16756 }, { "epoch": 1.9498487316732604, "grad_norm": 1.1474943161010742, "learning_rate": 1.9238350013817268e-05, "loss": 1.4799904823303223, "step": 16757 }, { "epoch": 1.9499650919245985, "grad_norm": 1.4003857374191284, "learning_rate": 1.9150540295654633e-05, "loss": 1.5236338376998901, "step": 16758 }, { "epoch": 1.9500814521759366, "grad_norm": 1.1348143815994263, "learning_rate": 1.906292344963828e-05, "loss": 2.446072578430176, "step": 16759 }, { "epoch": 1.950197812427275, "grad_norm": 1.4790383577346802, "learning_rate": 1.897549954897107e-05, "loss": 2.235229969024658, "step": 16760 }, { "epoch": 1.9503141726786128, "grad_norm": 1.4387922286987305, "learning_rate": 1.8888268666694743e-05, "loss": 2.3456711769104004, "step": 16761 }, { "epoch": 1.9504305329299512, "grad_norm": 0.7420148253440857, "learning_rate": 1.880123087568969e-05, "loss": 1.412803053855896, "step": 16762 }, { "epoch": 1.9505468931812893, "grad_norm": 1.001121163368225, "learning_rate": 1.8714386248675053e-05, "loss": 2.997100353240967, "step": 16763 }, { "epoch": 1.9506632534326274, "grad_norm": 1.0134458541870117, "learning_rate": 1.862773485820851e-05, "loss": 1.4422669410705566, "step": 16764 }, { "epoch": 1.9507796136839657, "grad_norm": 0.9270461797714233, "learning_rate": 1.854127677668635e-05, "loss": 1.422227382659912, "step": 16765 }, { "epoch": 1.9508959739353036, "grad_norm": 1.1079366207122803, "learning_rate": 1.8455012076343342e-05, "loss": 1.5401611328125, "step": 16766 }, { "epoch": 1.9510123341866419, "grad_norm": 1.3562909364700317, "learning_rate": 1.836894082925264e-05, "loss": 1.7942837476730347, "step": 16767 }, { "epoch": 1.95112869443798, "grad_norm": 1.2361714839935303, "learning_rate": 1.8283063107325857e-05, "loss": 1.8421826362609863, "step": 16768 }, { "epoch": 1.951245054689318, "grad_norm": 1.406198501586914, "learning_rate": 1.8197378982312884e-05, "loss": 1.5107499361038208, "step": 16769 }, { "epoch": 1.9513614149406564, "grad_norm": 0.6582812666893005, "learning_rate": 1.8111888525801756e-05, "loss": 1.4552090167999268, "step": 16770 }, { "epoch": 1.9514777751919943, "grad_norm": 0.7849817276000977, "learning_rate": 1.802659180921884e-05, "loss": 1.634741187095642, "step": 16771 }, { "epoch": 1.9515941354433326, "grad_norm": 0.8736394643783569, "learning_rate": 1.794148890382863e-05, "loss": 1.6430134773254395, "step": 16772 }, { "epoch": 1.9517104956946707, "grad_norm": 0.7177172303199768, "learning_rate": 1.7856579880733664e-05, "loss": 1.5316541194915771, "step": 16773 }, { "epoch": 1.9518268559460088, "grad_norm": 0.8730233907699585, "learning_rate": 1.7771864810874422e-05, "loss": 1.7310477495193481, "step": 16774 }, { "epoch": 1.9519432161973471, "grad_norm": 0.7329384684562683, "learning_rate": 1.7687343765029425e-05, "loss": 1.1044921875, "step": 16775 }, { "epoch": 1.952059576448685, "grad_norm": 0.9001123309135437, "learning_rate": 1.760301681381507e-05, "loss": 1.448016881942749, "step": 16776 }, { "epoch": 1.9521759367000233, "grad_norm": 0.7089359164237976, "learning_rate": 1.7518884027685562e-05, "loss": 1.982997179031372, "step": 16777 }, { "epoch": 1.9522922969513614, "grad_norm": 0.7851971387863159, "learning_rate": 1.7434945476932944e-05, "loss": 1.61611008644104, "step": 16778 }, { "epoch": 1.9524086572026995, "grad_norm": 0.6744732856750488, "learning_rate": 1.7351201231686903e-05, "loss": 1.8815793991088867, "step": 16779 }, { "epoch": 1.9525250174540378, "grad_norm": 0.7531812787055969, "learning_rate": 1.7267651361914827e-05, "loss": 2.0437843799591064, "step": 16780 }, { "epoch": 1.9526413777053757, "grad_norm": 1.3054718971252441, "learning_rate": 1.7184295937421714e-05, "loss": 2.068779230117798, "step": 16781 }, { "epoch": 1.952757737956714, "grad_norm": 1.0915343761444092, "learning_rate": 1.7101135027850074e-05, "loss": 1.4627017974853516, "step": 16782 }, { "epoch": 1.9528740982080521, "grad_norm": 1.098943829536438, "learning_rate": 1.7018168702679913e-05, "loss": 1.8501442670822144, "step": 16783 }, { "epoch": 1.9529904584593902, "grad_norm": 0.9180223941802979, "learning_rate": 1.6935397031228704e-05, "loss": 1.8000271320343018, "step": 16784 }, { "epoch": 1.9531068187107286, "grad_norm": 0.43442657589912415, "learning_rate": 1.685282008265118e-05, "loss": 0.9138468503952026, "step": 16785 }, { "epoch": 1.9532231789620664, "grad_norm": 0.7797958850860596, "learning_rate": 1.6770437925939547e-05, "loss": 1.5558568239212036, "step": 16786 }, { "epoch": 1.9533395392134048, "grad_norm": 1.1084048748016357, "learning_rate": 1.6688250629923157e-05, "loss": 1.7085514068603516, "step": 16787 }, { "epoch": 1.9534558994647429, "grad_norm": 1.8335020542144775, "learning_rate": 1.660625826326856e-05, "loss": 1.8726218938827515, "step": 16788 }, { "epoch": 1.953572259716081, "grad_norm": 0.9781969785690308, "learning_rate": 1.6524460894479455e-05, "loss": 1.9107120037078857, "step": 16789 }, { "epoch": 1.9536886199674193, "grad_norm": 1.1700994968414307, "learning_rate": 1.644285859189665e-05, "loss": 1.365458607673645, "step": 16790 }, { "epoch": 1.9538049802187571, "grad_norm": 0.6091576814651489, "learning_rate": 1.6361451423697937e-05, "loss": 1.6793053150177002, "step": 16791 }, { "epoch": 1.9539213404700955, "grad_norm": 0.7239199876785278, "learning_rate": 1.628023945789811e-05, "loss": 1.4188774824142456, "step": 16792 }, { "epoch": 1.9540377007214336, "grad_norm": 1.2007877826690674, "learning_rate": 1.619922276234892e-05, "loss": 1.9417774677276611, "step": 16793 }, { "epoch": 1.9541540609727717, "grad_norm": 0.8981401920318604, "learning_rate": 1.6118401404738804e-05, "loss": 1.388846755027771, "step": 16794 }, { "epoch": 1.95427042122411, "grad_norm": 1.0029127597808838, "learning_rate": 1.6037775452593135e-05, "loss": 1.4746758937835693, "step": 16795 }, { "epoch": 1.9543867814754479, "grad_norm": 1.697204828262329, "learning_rate": 1.5957344973273973e-05, "loss": 1.645838737487793, "step": 16796 }, { "epoch": 1.9545031417267862, "grad_norm": 0.786925196647644, "learning_rate": 1.587711003398015e-05, "loss": 1.453967809677124, "step": 16797 }, { "epoch": 1.9546195019781243, "grad_norm": 0.9108633995056152, "learning_rate": 1.5797070701747074e-05, "loss": 1.897679328918457, "step": 16798 }, { "epoch": 1.9547358622294624, "grad_norm": 0.8617503643035889, "learning_rate": 1.5717227043446637e-05, "loss": 1.9974775314331055, "step": 16799 }, { "epoch": 1.9548522224808007, "grad_norm": 1.1476869583129883, "learning_rate": 1.5637579125787395e-05, "loss": 2.414229154586792, "step": 16800 }, { "epoch": 1.9549685827321386, "grad_norm": 1.229614019393921, "learning_rate": 1.5558127015314227e-05, "loss": 2.376828670501709, "step": 16801 }, { "epoch": 1.955084942983477, "grad_norm": 1.129241943359375, "learning_rate": 1.5478870778408537e-05, "loss": 1.756426453590393, "step": 16802 }, { "epoch": 1.955201303234815, "grad_norm": 1.09105384349823, "learning_rate": 1.5399810481287995e-05, "loss": 2.0150444507598877, "step": 16803 }, { "epoch": 1.955317663486153, "grad_norm": 1.0051548480987549, "learning_rate": 1.5320946190006624e-05, "loss": 2.108091354370117, "step": 16804 }, { "epoch": 1.9554340237374914, "grad_norm": 1.2278696298599243, "learning_rate": 1.5242277970454631e-05, "loss": 2.1524815559387207, "step": 16805 }, { "epoch": 1.9555503839888293, "grad_norm": 0.6167978644371033, "learning_rate": 1.5163805888358445e-05, "loss": 1.3550329208374023, "step": 16806 }, { "epoch": 1.9556667442401676, "grad_norm": 1.2968966960906982, "learning_rate": 1.5085530009280618e-05, "loss": 1.7528245449066162, "step": 16807 }, { "epoch": 1.9557831044915057, "grad_norm": 0.44338077306747437, "learning_rate": 1.5007450398619781e-05, "loss": 0.9954221248626709, "step": 16808 }, { "epoch": 1.9558994647428438, "grad_norm": 0.8988962769508362, "learning_rate": 1.492956712161056e-05, "loss": 1.8414206504821777, "step": 16809 }, { "epoch": 1.9560158249941821, "grad_norm": 1.6207698583602905, "learning_rate": 1.485188024332354e-05, "loss": 1.9531410932540894, "step": 16810 }, { "epoch": 1.95613218524552, "grad_norm": 1.0650510787963867, "learning_rate": 1.4774389828665302e-05, "loss": 1.809063196182251, "step": 16811 }, { "epoch": 1.9562485454968583, "grad_norm": 0.9119271039962769, "learning_rate": 1.4697095942378196e-05, "loss": 1.98964524269104, "step": 16812 }, { "epoch": 1.9563649057481964, "grad_norm": 0.7154960036277771, "learning_rate": 1.46199986490404e-05, "loss": 1.4540901184082031, "step": 16813 }, { "epoch": 1.9564812659995345, "grad_norm": 0.9379270672798157, "learning_rate": 1.4543098013065859e-05, "loss": 1.849640130996704, "step": 16814 }, { "epoch": 1.9565976262508729, "grad_norm": 1.1222120523452759, "learning_rate": 1.446639409870415e-05, "loss": 2.164364814758301, "step": 16815 }, { "epoch": 1.9567139865022107, "grad_norm": 1.1090368032455444, "learning_rate": 1.4389886970040567e-05, "loss": 1.8773621320724487, "step": 16816 }, { "epoch": 1.956830346753549, "grad_norm": 1.3404860496520996, "learning_rate": 1.4313576690996004e-05, "loss": 2.133680820465088, "step": 16817 }, { "epoch": 1.9569467070048872, "grad_norm": 0.7636700868606567, "learning_rate": 1.4237463325326743e-05, "loss": 1.8371012210845947, "step": 16818 }, { "epoch": 1.9570630672562253, "grad_norm": 1.0708186626434326, "learning_rate": 1.4161546936624719e-05, "loss": 1.7588492631912231, "step": 16819 }, { "epoch": 1.9571794275075636, "grad_norm": 0.7613438963890076, "learning_rate": 1.4085827588317224e-05, "loss": 1.1027977466583252, "step": 16820 }, { "epoch": 1.9572957877589015, "grad_norm": 0.8673760890960693, "learning_rate": 1.4010305343666873e-05, "loss": 2.2103428840637207, "step": 16821 }, { "epoch": 1.9574121480102398, "grad_norm": 1.4597913026809692, "learning_rate": 1.3934980265771752e-05, "loss": 2.518733501434326, "step": 16822 }, { "epoch": 1.9575285082615779, "grad_norm": 0.9019958972930908, "learning_rate": 1.3859852417565072e-05, "loss": 1.4408996105194092, "step": 16823 }, { "epoch": 1.957644868512916, "grad_norm": 1.7893344163894653, "learning_rate": 1.3784921861815347e-05, "loss": 2.131741762161255, "step": 16824 }, { "epoch": 1.9577612287642543, "grad_norm": 1.107621192932129, "learning_rate": 1.371018866112614e-05, "loss": 1.5899401903152466, "step": 16825 }, { "epoch": 1.9578775890155922, "grad_norm": 1.3963253498077393, "learning_rate": 1.3635652877936278e-05, "loss": 2.1540541648864746, "step": 16826 }, { "epoch": 1.9579939492669305, "grad_norm": 0.7821404337882996, "learning_rate": 1.3561314574519561e-05, "loss": 1.5096192359924316, "step": 16827 }, { "epoch": 1.9581103095182686, "grad_norm": 1.0368807315826416, "learning_rate": 1.348717381298481e-05, "loss": 1.9139398336410522, "step": 16828 }, { "epoch": 1.9582266697696067, "grad_norm": 0.7794911861419678, "learning_rate": 1.3413230655275777e-05, "loss": 1.8225802183151245, "step": 16829 }, { "epoch": 1.958343030020945, "grad_norm": 0.7756665945053101, "learning_rate": 1.3339485163171184e-05, "loss": 1.8031964302062988, "step": 16830 }, { "epoch": 1.9584593902722829, "grad_norm": 1.4630048274993896, "learning_rate": 1.3265937398284522e-05, "loss": 1.4805799722671509, "step": 16831 }, { "epoch": 1.9585757505236212, "grad_norm": 1.197139859199524, "learning_rate": 1.3192587422064167e-05, "loss": 1.8585959672927856, "step": 16832 }, { "epoch": 1.9586921107749593, "grad_norm": 0.7588847279548645, "learning_rate": 1.3119435295793208e-05, "loss": 1.4091581106185913, "step": 16833 }, { "epoch": 1.9588084710262974, "grad_norm": 1.1623977422714233, "learning_rate": 1.3046481080589396e-05, "loss": 1.322021722793579, "step": 16834 }, { "epoch": 1.9589248312776357, "grad_norm": 0.8687777519226074, "learning_rate": 1.2973724837405137e-05, "loss": 2.5619468688964844, "step": 16835 }, { "epoch": 1.9590411915289736, "grad_norm": 1.4178344011306763, "learning_rate": 1.2901166627027533e-05, "loss": 2.9598073959350586, "step": 16836 }, { "epoch": 1.959157551780312, "grad_norm": 1.3182942867279053, "learning_rate": 1.2828806510078145e-05, "loss": 2.1286187171936035, "step": 16837 }, { "epoch": 1.95927391203165, "grad_norm": 1.397394061088562, "learning_rate": 1.275664454701303e-05, "loss": 1.8327279090881348, "step": 16838 }, { "epoch": 1.9593902722829881, "grad_norm": 1.221923589706421, "learning_rate": 1.2684680798122682e-05, "loss": 1.2861026525497437, "step": 16839 }, { "epoch": 1.9595066325343262, "grad_norm": 1.23651123046875, "learning_rate": 1.2612915323532037e-05, "loss": 1.8501102924346924, "step": 16840 }, { "epoch": 1.9596229927856643, "grad_norm": 1.1169434785842896, "learning_rate": 1.254134818320038e-05, "loss": 2.4437952041625977, "step": 16841 }, { "epoch": 1.9597393530370026, "grad_norm": 1.455980896949768, "learning_rate": 1.2469979436921158e-05, "loss": 2.2610220909118652, "step": 16842 }, { "epoch": 1.9598557132883407, "grad_norm": 1.0618219375610352, "learning_rate": 1.2398809144322232e-05, "loss": 1.952211618423462, "step": 16843 }, { "epoch": 1.9599720735396788, "grad_norm": 1.1758906841278076, "learning_rate": 1.232783736486559e-05, "loss": 1.634946584701538, "step": 16844 }, { "epoch": 1.960088433791017, "grad_norm": 1.0804592370986938, "learning_rate": 1.225706415784733e-05, "loss": 1.6267707347869873, "step": 16845 }, { "epoch": 1.960204794042355, "grad_norm": 0.9501745700836182, "learning_rate": 1.2186489582397653e-05, "loss": 1.4060633182525635, "step": 16846 }, { "epoch": 1.9603211542936934, "grad_norm": 1.0095645189285278, "learning_rate": 1.2116113697480891e-05, "loss": 1.671341061592102, "step": 16847 }, { "epoch": 1.9604375145450315, "grad_norm": 0.9552636742591858, "learning_rate": 1.2045936561895316e-05, "loss": 2.1156439781188965, "step": 16848 }, { "epoch": 1.9605538747963696, "grad_norm": 0.7198314666748047, "learning_rate": 1.1975958234273083e-05, "loss": 1.4018938541412354, "step": 16849 }, { "epoch": 1.9606702350477077, "grad_norm": 1.8524361848831177, "learning_rate": 1.1906178773080367e-05, "loss": 1.9797134399414062, "step": 16850 }, { "epoch": 1.9607865952990458, "grad_norm": 1.3431355953216553, "learning_rate": 1.1836598236617064e-05, "loss": 1.596909761428833, "step": 16851 }, { "epoch": 1.960902955550384, "grad_norm": 1.3116064071655273, "learning_rate": 1.1767216683016974e-05, "loss": 2.168684720993042, "step": 16852 }, { "epoch": 1.9610193158017222, "grad_norm": 1.048867106437683, "learning_rate": 1.169803417024759e-05, "loss": 1.7851283550262451, "step": 16853 }, { "epoch": 1.9611356760530603, "grad_norm": 1.710553526878357, "learning_rate": 1.1629050756110176e-05, "loss": 2.4278178215026855, "step": 16854 }, { "epoch": 1.9612520363043984, "grad_norm": 1.1992971897125244, "learning_rate": 1.1560266498239547e-05, "loss": 1.8556458950042725, "step": 16855 }, { "epoch": 1.9613683965557365, "grad_norm": 0.9205023646354675, "learning_rate": 1.1491681454104202e-05, "loss": 1.4662232398986816, "step": 16856 }, { "epoch": 1.9614847568070748, "grad_norm": 0.6643799543380737, "learning_rate": 1.1423295681006168e-05, "loss": 1.25675368309021, "step": 16857 }, { "epoch": 1.961601117058413, "grad_norm": 1.0396838188171387, "learning_rate": 1.1355109236081046e-05, "loss": 1.2729623317718506, "step": 16858 }, { "epoch": 1.961717477309751, "grad_norm": 1.0760189294815063, "learning_rate": 1.1287122176297793e-05, "loss": 2.184138298034668, "step": 16859 }, { "epoch": 1.961833837561089, "grad_norm": 1.0697247982025146, "learning_rate": 1.1219334558458804e-05, "loss": 1.4308180809020996, "step": 16860 }, { "epoch": 1.9619501978124272, "grad_norm": 0.8842681646347046, "learning_rate": 1.11517464392e-05, "loss": 1.8548240661621094, "step": 16861 }, { "epoch": 1.9620665580637655, "grad_norm": 1.2213990688323975, "learning_rate": 1.1084357874990402e-05, "loss": 2.2760672569274902, "step": 16862 }, { "epoch": 1.9621829183151036, "grad_norm": 1.1561644077301025, "learning_rate": 1.1017168922132448e-05, "loss": 1.8467035293579102, "step": 16863 }, { "epoch": 1.9622992785664417, "grad_norm": 1.0397801399230957, "learning_rate": 1.095017963676176e-05, "loss": 1.8299397230148315, "step": 16864 }, { "epoch": 1.9624156388177798, "grad_norm": 0.7716365456581116, "learning_rate": 1.0883390074847122e-05, "loss": 1.5369727611541748, "step": 16865 }, { "epoch": 1.962531999069118, "grad_norm": 1.0259422063827515, "learning_rate": 1.0816800292190454e-05, "loss": 2.7754416465759277, "step": 16866 }, { "epoch": 1.9626483593204562, "grad_norm": 0.7092157006263733, "learning_rate": 1.0750410344426782e-05, "loss": 1.7053077220916748, "step": 16867 }, { "epoch": 1.9627647195717943, "grad_norm": 0.927607536315918, "learning_rate": 1.068422028702415e-05, "loss": 1.8493956327438354, "step": 16868 }, { "epoch": 1.9628810798231324, "grad_norm": 0.5440449118614197, "learning_rate": 1.0618230175283655e-05, "loss": 1.5676058530807495, "step": 16869 }, { "epoch": 1.9629974400744705, "grad_norm": 0.7847793102264404, "learning_rate": 1.0552440064339281e-05, "loss": 1.1449254751205444, "step": 16870 }, { "epoch": 1.9631138003258086, "grad_norm": 1.0217257738113403, "learning_rate": 1.048685000915789e-05, "loss": 1.7812379598617554, "step": 16871 }, { "epoch": 1.963230160577147, "grad_norm": 1.1326128244400024, "learning_rate": 1.0421460064539312e-05, "loss": 1.4071006774902344, "step": 16872 }, { "epoch": 1.963346520828485, "grad_norm": 0.7741479277610779, "learning_rate": 1.0356270285116071e-05, "loss": 1.2695996761322021, "step": 16873 }, { "epoch": 1.9634628810798231, "grad_norm": 0.9664552211761475, "learning_rate": 1.0291280725353513e-05, "loss": 1.8440014123916626, "step": 16874 }, { "epoch": 1.9635792413311612, "grad_norm": 0.9728609919548035, "learning_rate": 1.0226491439549706e-05, "loss": 2.6104602813720703, "step": 16875 }, { "epoch": 1.9636956015824993, "grad_norm": 1.5127747058868408, "learning_rate": 1.0161902481835323e-05, "loss": 1.9322179555892944, "step": 16876 }, { "epoch": 1.9638119618338377, "grad_norm": 0.9511109590530396, "learning_rate": 1.0097513906173777e-05, "loss": 1.6555752754211426, "step": 16877 }, { "epoch": 1.9639283220851758, "grad_norm": 1.0811140537261963, "learning_rate": 1.0033325766360985e-05, "loss": 1.8651484251022339, "step": 16878 }, { "epoch": 1.9640446823365139, "grad_norm": 1.1114667654037476, "learning_rate": 9.969338116025406e-06, "loss": 2.5356316566467285, "step": 16879 }, { "epoch": 1.964161042587852, "grad_norm": 0.7682817578315735, "learning_rate": 9.905551008628e-06, "loss": 2.3331656455993652, "step": 16880 }, { "epoch": 1.96427740283919, "grad_norm": 0.7619004845619202, "learning_rate": 9.84196449746219e-06, "loss": 1.4954222440719604, "step": 16881 }, { "epoch": 1.9643937630905284, "grad_norm": 1.124436378479004, "learning_rate": 9.778578635653845e-06, "loss": 1.5170423984527588, "step": 16882 }, { "epoch": 1.9645101233418663, "grad_norm": 0.7895695567131042, "learning_rate": 9.715393476161072e-06, "loss": 1.9620310068130493, "step": 16883 }, { "epoch": 1.9646264835932046, "grad_norm": 1.3642151355743408, "learning_rate": 9.65240907177442e-06, "loss": 1.8356695175170898, "step": 16884 }, { "epoch": 1.9647428438445427, "grad_norm": 1.013014554977417, "learning_rate": 9.589625475116576e-06, "loss": 2.0686750411987305, "step": 16885 }, { "epoch": 1.9648592040958808, "grad_norm": 0.924375057220459, "learning_rate": 9.527042738642632e-06, "loss": 1.6812870502471924, "step": 16886 }, { "epoch": 1.964975564347219, "grad_norm": 0.8325001001358032, "learning_rate": 9.464660914639739e-06, "loss": 1.890082597732544, "step": 16887 }, { "epoch": 1.965091924598557, "grad_norm": 2.233548641204834, "learning_rate": 9.402480055227203e-06, "loss": 2.030449628829956, "step": 16888 }, { "epoch": 1.9652082848498953, "grad_norm": 0.9907265901565552, "learning_rate": 9.340500212356467e-06, "loss": 1.6984190940856934, "step": 16889 }, { "epoch": 1.9653246451012334, "grad_norm": 0.5807708501815796, "learning_rate": 9.278721437810916e-06, "loss": 1.1964432001113892, "step": 16890 }, { "epoch": 1.9654410053525715, "grad_norm": 0.824781060218811, "learning_rate": 9.217143783206095e-06, "loss": 1.947819471359253, "step": 16891 }, { "epoch": 1.9655573656039098, "grad_norm": 1.3236424922943115, "learning_rate": 9.155767299989432e-06, "loss": 1.9424324035644531, "step": 16892 }, { "epoch": 1.9656737258552477, "grad_norm": 1.4353022575378418, "learning_rate": 9.094592039440269e-06, "loss": 1.8753770589828491, "step": 16893 }, { "epoch": 1.965790086106586, "grad_norm": 0.9402325749397278, "learning_rate": 9.03361805266989e-06, "loss": 2.374119758605957, "step": 16894 }, { "epoch": 1.965906446357924, "grad_norm": 1.0098947286605835, "learning_rate": 8.972845390621376e-06, "loss": 1.649557113647461, "step": 16895 }, { "epoch": 1.9660228066092622, "grad_norm": 0.9243122339248657, "learning_rate": 8.912274104069556e-06, "loss": 2.0236942768096924, "step": 16896 }, { "epoch": 1.9661391668606005, "grad_norm": 0.7576101422309875, "learning_rate": 8.851904243621145e-06, "loss": 1.3029265403747559, "step": 16897 }, { "epoch": 1.9662555271119384, "grad_norm": 1.0041042566299438, "learning_rate": 8.791735859714433e-06, "loss": 2.07712459564209, "step": 16898 }, { "epoch": 1.9663718873632767, "grad_norm": 1.355302095413208, "learning_rate": 8.731769002619433e-06, "loss": 1.7663195133209229, "step": 16899 }, { "epoch": 1.9664882476146148, "grad_norm": 1.0181275606155396, "learning_rate": 8.672003722437815e-06, "loss": 2.2267842292785645, "step": 16900 }, { "epoch": 1.966604607865953, "grad_norm": 0.925351083278656, "learning_rate": 8.612440069102806e-06, "loss": 1.7260150909423828, "step": 16901 }, { "epoch": 1.9667209681172912, "grad_norm": 0.8837306499481201, "learning_rate": 8.553078092379124e-06, "loss": 1.652748465538025, "step": 16902 }, { "epoch": 1.9668373283686291, "grad_norm": 1.2642430067062378, "learning_rate": 8.493917841863097e-06, "loss": 1.677506446838379, "step": 16903 }, { "epoch": 1.9669536886199674, "grad_norm": 0.8295931220054626, "learning_rate": 8.434959366982409e-06, "loss": 1.6158969402313232, "step": 16904 }, { "epoch": 1.9670700488713055, "grad_norm": 0.6267810463905334, "learning_rate": 8.376202716996212e-06, "loss": 1.5580230951309204, "step": 16905 }, { "epoch": 1.9671864091226436, "grad_norm": 0.9916048645973206, "learning_rate": 8.317647940995071e-06, "loss": 2.519688606262207, "step": 16906 }, { "epoch": 1.967302769373982, "grad_norm": 1.2587800025939941, "learning_rate": 8.259295087900797e-06, "loss": 1.1384527683258057, "step": 16907 }, { "epoch": 1.9674191296253198, "grad_norm": 1.1810866594314575, "learning_rate": 8.20114420646656e-06, "loss": 2.0515356063842773, "step": 16908 }, { "epoch": 1.9675354898766582, "grad_norm": 0.951458215713501, "learning_rate": 8.143195345276799e-06, "loss": 1.7769556045532227, "step": 16909 }, { "epoch": 1.9676518501279963, "grad_norm": 1.2293280363082886, "learning_rate": 8.085448552747093e-06, "loss": 2.187079429626465, "step": 16910 }, { "epoch": 1.9677682103793344, "grad_norm": 1.190621256828308, "learning_rate": 8.027903877124321e-06, "loss": 2.1570591926574707, "step": 16911 }, { "epoch": 1.9678845706306727, "grad_norm": 0.7549455165863037, "learning_rate": 7.97056136648633e-06, "loss": 1.2923657894134521, "step": 16912 }, { "epoch": 1.9680009308820106, "grad_norm": 0.8036062121391296, "learning_rate": 7.913421068742238e-06, "loss": 1.0490790605545044, "step": 16913 }, { "epoch": 1.9681172911333489, "grad_norm": 1.0361871719360352, "learning_rate": 7.856483031632027e-06, "loss": 1.689104676246643, "step": 16914 }, { "epoch": 1.968233651384687, "grad_norm": 0.39969849586486816, "learning_rate": 7.799747302726868e-06, "loss": 0.9631325006484985, "step": 16915 }, { "epoch": 1.968350011636025, "grad_norm": 0.956328272819519, "learning_rate": 7.743213929428783e-06, "loss": 1.5728952884674072, "step": 16916 }, { "epoch": 1.9684663718873634, "grad_norm": 1.1228762865066528, "learning_rate": 7.686882958970797e-06, "loss": 1.4928698539733887, "step": 16917 }, { "epoch": 1.9685827321387013, "grad_norm": 1.751799464225769, "learning_rate": 7.630754438416792e-06, "loss": 1.790462613105774, "step": 16918 }, { "epoch": 1.9686990923900396, "grad_norm": 0.88826984167099, "learning_rate": 7.574828414661533e-06, "loss": 1.5878442525863647, "step": 16919 }, { "epoch": 1.9688154526413777, "grad_norm": 1.4419620037078857, "learning_rate": 7.5191049344305875e-06, "loss": 1.5886051654815674, "step": 16920 }, { "epoch": 1.9689318128927158, "grad_norm": 0.667979896068573, "learning_rate": 7.463584044280242e-06, "loss": 1.8915067911148071, "step": 16921 }, { "epoch": 1.9690481731440541, "grad_norm": 1.561317801475525, "learning_rate": 7.408265790597696e-06, "loss": 2.4473042488098145, "step": 16922 }, { "epoch": 1.969164533395392, "grad_norm": 1.0534645318984985, "learning_rate": 7.3531502196006735e-06, "loss": 1.424534559249878, "step": 16923 }, { "epoch": 1.9692808936467303, "grad_norm": 1.3688204288482666, "learning_rate": 7.298237377337591e-06, "loss": 1.5890833139419556, "step": 16924 }, { "epoch": 1.9693972538980684, "grad_norm": 2.363226890563965, "learning_rate": 7.243527309687581e-06, "loss": 2.4897983074188232, "step": 16925 }, { "epoch": 1.9695136141494065, "grad_norm": 1.057842493057251, "learning_rate": 7.189020062360219e-06, "loss": 1.2431252002716064, "step": 16926 }, { "epoch": 1.9696299744007448, "grad_norm": 0.764369785785675, "learning_rate": 7.13471568089577e-06, "loss": 1.3652913570404053, "step": 16927 }, { "epoch": 1.9697463346520827, "grad_norm": 1.5520689487457275, "learning_rate": 7.080614210664887e-06, "loss": 1.6547231674194336, "step": 16928 }, { "epoch": 1.969862694903421, "grad_norm": 1.146031141281128, "learning_rate": 7.026715696868774e-06, "loss": 1.841039776802063, "step": 16929 }, { "epoch": 1.9699790551547591, "grad_norm": 0.787169337272644, "learning_rate": 6.97302018453902e-06, "loss": 1.2080632448196411, "step": 16930 }, { "epoch": 1.9700954154060972, "grad_norm": 1.065697431564331, "learning_rate": 6.9195277185376274e-06, "loss": 2.3794667720794678, "step": 16931 }, { "epoch": 1.9702117756574355, "grad_norm": 1.0383766889572144, "learning_rate": 6.8662383435569854e-06, "loss": 1.4730302095413208, "step": 16932 }, { "epoch": 1.9703281359087734, "grad_norm": 1.8986676931381226, "learning_rate": 6.81315210411973e-06, "loss": 2.298499345779419, "step": 16933 }, { "epoch": 1.9704444961601117, "grad_norm": 0.9269862174987793, "learning_rate": 6.760269044578826e-06, "loss": 2.269958019256592, "step": 16934 }, { "epoch": 1.9705608564114498, "grad_norm": 0.6556272506713867, "learning_rate": 6.707589209117487e-06, "loss": 1.459137201309204, "step": 16935 }, { "epoch": 1.970677216662788, "grad_norm": 0.7728450894355774, "learning_rate": 6.655112641749144e-06, "loss": 1.6273514032363892, "step": 16936 }, { "epoch": 1.9707935769141263, "grad_norm": 0.8645954728126526, "learning_rate": 6.602839386317394e-06, "loss": 1.5633412599563599, "step": 16937 }, { "epoch": 1.9709099371654641, "grad_norm": 1.28194260597229, "learning_rate": 6.550769486495939e-06, "loss": 1.6625885963439941, "step": 16938 }, { "epoch": 1.9710262974168025, "grad_norm": 1.0089613199234009, "learning_rate": 6.498902985788563e-06, "loss": 1.5076005458831787, "step": 16939 }, { "epoch": 1.9711426576681406, "grad_norm": 0.8306775093078613, "learning_rate": 6.447239927529186e-06, "loss": 2.242600917816162, "step": 16940 }, { "epoch": 1.9712590179194787, "grad_norm": 0.7803553938865662, "learning_rate": 6.395780354881725e-06, "loss": 1.7498326301574707, "step": 16941 }, { "epoch": 1.971375378170817, "grad_norm": 0.5404447913169861, "learning_rate": 6.344524310840066e-06, "loss": 0.945216178894043, "step": 16942 }, { "epoch": 1.9714917384221549, "grad_norm": 0.9840189814567566, "learning_rate": 6.293471838228065e-06, "loss": 1.6170405149459839, "step": 16943 }, { "epoch": 1.9716080986734932, "grad_norm": 0.8427772521972656, "learning_rate": 6.2426229796994635e-06, "loss": 1.4782785177230835, "step": 16944 }, { "epoch": 1.9717244589248313, "grad_norm": 1.0440045595169067, "learning_rate": 6.191977777737945e-06, "loss": 1.6983256340026855, "step": 16945 }, { "epoch": 1.9718408191761694, "grad_norm": 0.6699044704437256, "learning_rate": 6.141536274656995e-06, "loss": 1.3868677616119385, "step": 16946 }, { "epoch": 1.9719571794275077, "grad_norm": 1.137891411781311, "learning_rate": 6.091298512599958e-06, "loss": 1.8854151964187622, "step": 16947 }, { "epoch": 1.9720735396788456, "grad_norm": 0.8444801568984985, "learning_rate": 6.041264533539897e-06, "loss": 1.4118988513946533, "step": 16948 }, { "epoch": 1.972189899930184, "grad_norm": 2.8023624420166016, "learning_rate": 5.991434379279653e-06, "loss": 2.3442835807800293, "step": 16949 }, { "epoch": 1.972306260181522, "grad_norm": 1.0357604026794434, "learning_rate": 5.941808091451756e-06, "loss": 1.7781736850738525, "step": 16950 }, { "epoch": 1.97242262043286, "grad_norm": 0.7229450941085815, "learning_rate": 5.8923857115183995e-06, "loss": 1.8006117343902588, "step": 16951 }, { "epoch": 1.9725389806841984, "grad_norm": 1.2248305082321167, "learning_rate": 5.843167280771472e-06, "loss": 2.2302958965301514, "step": 16952 }, { "epoch": 1.9726553409355363, "grad_norm": 0.7252846956253052, "learning_rate": 5.794152840332384e-06, "loss": 1.7895476818084717, "step": 16953 }, { "epoch": 1.9727717011868746, "grad_norm": 1.8083330392837524, "learning_rate": 5.745342431152184e-06, "loss": 1.698411226272583, "step": 16954 }, { "epoch": 1.9728880614382127, "grad_norm": 1.3028287887573242, "learning_rate": 5.69673609401139e-06, "loss": 2.022770881652832, "step": 16955 }, { "epoch": 1.9730044216895508, "grad_norm": 1.2491031885147095, "learning_rate": 5.6483338695201e-06, "loss": 1.7956029176712036, "step": 16956 }, { "epoch": 1.9731207819408891, "grad_norm": 1.2098443508148193, "learning_rate": 5.600135798117773e-06, "loss": 2.056427478790283, "step": 16957 }, { "epoch": 1.973237142192227, "grad_norm": 1.3546490669250488, "learning_rate": 5.552141920073417e-06, "loss": 2.114048957824707, "step": 16958 }, { "epoch": 1.9733535024435653, "grad_norm": 1.4740896224975586, "learning_rate": 5.504352275485347e-06, "loss": 1.7538154125213623, "step": 16959 }, { "epoch": 1.9734698626949034, "grad_norm": 1.1243112087249756, "learning_rate": 5.456766904281263e-06, "loss": 1.6753482818603516, "step": 16960 }, { "epoch": 1.9735862229462415, "grad_norm": 1.4026107788085938, "learning_rate": 5.409385846218251e-06, "loss": 1.767415165901184, "step": 16961 }, { "epoch": 1.9737025831975799, "grad_norm": 1.3060158491134644, "learning_rate": 5.362209140882645e-06, "loss": 2.0998830795288086, "step": 16962 }, { "epoch": 1.9738189434489177, "grad_norm": 1.7036795616149902, "learning_rate": 5.315236827690057e-06, "loss": 2.380911350250244, "step": 16963 }, { "epoch": 1.973935303700256, "grad_norm": 1.011453628540039, "learning_rate": 5.268468945885313e-06, "loss": 1.74403715133667, "step": 16964 }, { "epoch": 1.9740516639515941, "grad_norm": 0.5515215396881104, "learning_rate": 5.221905534542465e-06, "loss": 1.6574753522872925, "step": 16965 }, { "epoch": 1.9741680242029322, "grad_norm": 0.8451024889945984, "learning_rate": 5.175546632564726e-06, "loss": 1.4793027639389038, "step": 16966 }, { "epoch": 1.9742843844542706, "grad_norm": 0.6350863575935364, "learning_rate": 5.129392278684419e-06, "loss": 1.6379423141479492, "step": 16967 }, { "epoch": 1.9744007447056084, "grad_norm": 1.2008541822433472, "learning_rate": 5.083442511463032e-06, "loss": 1.536489725112915, "step": 16968 }, { "epoch": 1.9745171049569468, "grad_norm": 0.8784599900245667, "learning_rate": 5.037697369291078e-06, "loss": 1.7738351821899414, "step": 16969 }, { "epoch": 1.9746334652082849, "grad_norm": 0.7776579856872559, "learning_rate": 4.992156890388067e-06, "loss": 1.911391258239746, "step": 16970 }, { "epoch": 1.974749825459623, "grad_norm": 1.110055923461914, "learning_rate": 4.946821112802591e-06, "loss": 2.3077197074890137, "step": 16971 }, { "epoch": 1.9748661857109613, "grad_norm": 0.7768155932426453, "learning_rate": 4.901690074412185e-06, "loss": 1.3713743686676025, "step": 16972 }, { "epoch": 1.9749825459622992, "grad_norm": 1.107674241065979, "learning_rate": 4.856763812923353e-06, "loss": 2.424485206604004, "step": 16973 }, { "epoch": 1.9750989062136375, "grad_norm": 1.5409138202667236, "learning_rate": 4.812042365871461e-06, "loss": 1.6604232788085938, "step": 16974 }, { "epoch": 1.9752152664649756, "grad_norm": 0.9161802530288696, "learning_rate": 4.767525770620756e-06, "loss": 1.7459895610809326, "step": 16975 }, { "epoch": 1.9753316267163137, "grad_norm": 1.2009235620498657, "learning_rate": 4.7232140643643785e-06, "loss": 1.9187321662902832, "step": 16976 }, { "epoch": 1.975447986967652, "grad_norm": 1.4167990684509277, "learning_rate": 4.679107284124296e-06, "loss": 1.8633610010147095, "step": 16977 }, { "epoch": 1.9755643472189899, "grad_norm": 1.6196585893630981, "learning_rate": 4.6352054667511685e-06, "loss": 2.056511878967285, "step": 16978 }, { "epoch": 1.9756807074703282, "grad_norm": 1.2913802862167358, "learning_rate": 4.5915086489245196e-06, "loss": 1.6594727039337158, "step": 16979 }, { "epoch": 1.9757970677216663, "grad_norm": 0.7925302386283875, "learning_rate": 4.548016867152533e-06, "loss": 2.0482165813446045, "step": 16980 }, { "epoch": 1.9759134279730044, "grad_norm": 1.0466828346252441, "learning_rate": 4.504730157772085e-06, "loss": 2.48512601852417, "step": 16981 }, { "epoch": 1.9760297882243427, "grad_norm": 1.2895174026489258, "learning_rate": 4.461648556948805e-06, "loss": 1.395603895187378, "step": 16982 }, { "epoch": 1.9761461484756806, "grad_norm": 1.2243224382400513, "learning_rate": 4.418772100676844e-06, "loss": 2.925797462463379, "step": 16983 }, { "epoch": 1.976262508727019, "grad_norm": 0.7158365249633789, "learning_rate": 4.376100824778989e-06, "loss": 1.3579363822937012, "step": 16984 }, { "epoch": 1.976378868978357, "grad_norm": 0.43158820271492004, "learning_rate": 4.333634764906641e-06, "loss": 0.9978821277618408, "step": 16985 }, { "epoch": 1.9764952292296951, "grad_norm": 0.7085555195808411, "learning_rate": 4.2913739565397235e-06, "loss": 1.3034625053405762, "step": 16986 }, { "epoch": 1.9766115894810334, "grad_norm": 1.0510592460632324, "learning_rate": 4.249318434986632e-06, "loss": 1.775303602218628, "step": 16987 }, { "epoch": 1.9767279497323713, "grad_norm": 0.8155607581138611, "learning_rate": 4.207468235384343e-06, "loss": 1.53696608543396, "step": 16988 }, { "epoch": 1.9768443099837096, "grad_norm": 0.8692724108695984, "learning_rate": 4.1658233926981925e-06, "loss": 1.4891133308410645, "step": 16989 }, { "epoch": 1.9769606702350477, "grad_norm": 0.7549371123313904, "learning_rate": 4.12438394172196e-06, "loss": 1.3105241060256958, "step": 16990 }, { "epoch": 1.9770770304863858, "grad_norm": 0.7489837408065796, "learning_rate": 4.0831499170778945e-06, "loss": 1.4668874740600586, "step": 16991 }, { "epoch": 1.9771933907377242, "grad_norm": 1.3944367170333862, "learning_rate": 4.042121353216549e-06, "loss": 2.465660810470581, "step": 16992 }, { "epoch": 1.977309750989062, "grad_norm": 0.9100819230079651, "learning_rate": 4.00129828441681e-06, "loss": 1.9763851165771484, "step": 16993 }, { "epoch": 1.9774261112404004, "grad_norm": 1.58423912525177, "learning_rate": 3.960680744785922e-06, "loss": 1.5520713329315186, "step": 16994 }, { "epoch": 1.9775424714917385, "grad_norm": 1.412947177886963, "learning_rate": 3.920268768259349e-06, "loss": 2.5532147884368896, "step": 16995 }, { "epoch": 1.9776588317430766, "grad_norm": 1.0919276475906372, "learning_rate": 3.880062388600864e-06, "loss": 1.8557660579681396, "step": 16996 }, { "epoch": 1.9777751919944149, "grad_norm": 0.8702998757362366, "learning_rate": 3.840061639402481e-06, "loss": 1.5348875522613525, "step": 16997 }, { "epoch": 1.9778915522457527, "grad_norm": 1.3290961980819702, "learning_rate": 3.8002665540843305e-06, "loss": 1.7086336612701416, "step": 16998 }, { "epoch": 1.978007912497091, "grad_norm": 0.8777669072151184, "learning_rate": 3.760677165894788e-06, "loss": 1.5191644430160522, "step": 16999 }, { "epoch": 1.9781242727484292, "grad_norm": 2.338679075241089, "learning_rate": 3.721293507910367e-06, "loss": 1.3439202308654785, "step": 17000 }, { "epoch": 1.9782406329997673, "grad_norm": 0.9618965983390808, "learning_rate": 3.682115613035636e-06, "loss": 1.4659011363983154, "step": 17001 }, { "epoch": 1.9783569932511056, "grad_norm": 0.7801740765571594, "learning_rate": 3.6431435140032996e-06, "loss": 1.5993276834487915, "step": 17002 }, { "epoch": 1.9784733535024435, "grad_norm": 1.4466133117675781, "learning_rate": 3.6043772433740907e-06, "loss": 1.58203125, "step": 17003 }, { "epoch": 1.9785897137537818, "grad_norm": 0.6494644284248352, "learning_rate": 3.56581683353685e-06, "loss": 1.7260713577270508, "step": 17004 }, { "epoch": 1.9787060740051199, "grad_norm": 1.0886439085006714, "learning_rate": 3.5274623167083074e-06, "loss": 2.4008469581604004, "step": 17005 }, { "epoch": 1.978822434256458, "grad_norm": 1.096432089805603, "learning_rate": 3.489313724933302e-06, "loss": 2.4750614166259766, "step": 17006 }, { "epoch": 1.978938794507796, "grad_norm": 0.9945959448814392, "learning_rate": 3.4513710900844763e-06, "loss": 1.4661445617675781, "step": 17007 }, { "epoch": 1.9790551547591342, "grad_norm": 2.0649149417877197, "learning_rate": 3.4136344438625564e-06, "loss": 1.4300646781921387, "step": 17008 }, { "epoch": 1.9791715150104725, "grad_norm": 1.0942714214324951, "learning_rate": 3.376103817796017e-06, "loss": 1.5452994108200073, "step": 17009 }, { "epoch": 1.9792878752618106, "grad_norm": 0.9335669279098511, "learning_rate": 3.33877924324133e-06, "loss": 1.5378016233444214, "step": 17010 }, { "epoch": 1.9794042355131487, "grad_norm": 0.9834478497505188, "learning_rate": 3.3016607513827455e-06, "loss": 1.4353351593017578, "step": 17011 }, { "epoch": 1.9795205957644868, "grad_norm": 0.959145188331604, "learning_rate": 3.2647483732323457e-06, "loss": 1.474111795425415, "step": 17012 }, { "epoch": 1.979636956015825, "grad_norm": 0.3922507166862488, "learning_rate": 3.2280421396299887e-06, "loss": 1.0087594985961914, "step": 17013 }, { "epoch": 1.9797533162671632, "grad_norm": 1.552812099456787, "learning_rate": 3.1915420812433648e-06, "loss": 1.8995544910430908, "step": 17014 }, { "epoch": 1.9798696765185013, "grad_norm": 0.4204505681991577, "learning_rate": 3.1552482285678584e-06, "loss": 1.1088159084320068, "step": 17015 }, { "epoch": 1.9799860367698394, "grad_norm": 1.6904703378677368, "learning_rate": 3.119160611926547e-06, "loss": 2.0072271823883057, "step": 17016 }, { "epoch": 1.9801023970211775, "grad_norm": 0.8656362891197205, "learning_rate": 3.0832792614702564e-06, "loss": 1.7679730653762817, "step": 17017 }, { "epoch": 1.9802187572725156, "grad_norm": 1.0531619787216187, "learning_rate": 3.0476042071774512e-06, "loss": 1.2943767309188843, "step": 17018 }, { "epoch": 1.980335117523854, "grad_norm": 0.6292510628700256, "learning_rate": 3.0121354788542332e-06, "loss": 1.4568930864334106, "step": 17019 }, { "epoch": 1.980451477775192, "grad_norm": 0.9769690632820129, "learning_rate": 2.9768731061343147e-06, "loss": 2.7038838863372803, "step": 17020 }, { "epoch": 1.9805678380265301, "grad_norm": 1.2307133674621582, "learning_rate": 2.9418171184790454e-06, "loss": 2.110098361968994, "step": 17021 }, { "epoch": 1.9806841982778682, "grad_norm": 0.5177388191223145, "learning_rate": 2.9069675451773024e-06, "loss": 1.1929912567138672, "step": 17022 }, { "epoch": 1.9808005585292063, "grad_norm": 1.5994449853897095, "learning_rate": 2.8723244153455176e-06, "loss": 1.957958459854126, "step": 17023 }, { "epoch": 1.9809169187805447, "grad_norm": 0.7845443487167358, "learning_rate": 2.837887757927621e-06, "loss": 1.7926571369171143, "step": 17024 }, { "epoch": 1.9810332790318828, "grad_norm": 1.4072760343551636, "learning_rate": 2.8036576016950975e-06, "loss": 2.208961248397827, "step": 17025 }, { "epoch": 1.9811496392832209, "grad_norm": 0.9190804958343506, "learning_rate": 2.7696339752467935e-06, "loss": 1.5756230354309082, "step": 17026 }, { "epoch": 1.981265999534559, "grad_norm": 0.8839023113250732, "learning_rate": 2.7358169070091367e-06, "loss": 1.3928210735321045, "step": 17027 }, { "epoch": 1.981382359785897, "grad_norm": 1.4208873510360718, "learning_rate": 2.7022064252358593e-06, "loss": 1.64729905128479, "step": 17028 }, { "epoch": 1.9814987200372354, "grad_norm": 0.9258085489273071, "learning_rate": 2.6688025580081653e-06, "loss": 1.7618459463119507, "step": 17029 }, { "epoch": 1.9816150802885735, "grad_norm": 0.754776120185852, "learning_rate": 2.6356053332346196e-06, "loss": 1.3408175706863403, "step": 17030 }, { "epoch": 1.9817314405399116, "grad_norm": 1.069387435913086, "learning_rate": 2.6026147786511188e-06, "loss": 1.9080579280853271, "step": 17031 }, { "epoch": 1.9818478007912497, "grad_norm": 0.8653770685195923, "learning_rate": 2.5698309218209204e-06, "loss": 2.287529706954956, "step": 17032 }, { "epoch": 1.9819641610425878, "grad_norm": 0.9353344440460205, "learning_rate": 2.537253790134558e-06, "loss": 2.271634340286255, "step": 17033 }, { "epoch": 1.982080521293926, "grad_norm": 1.0795652866363525, "learning_rate": 2.504883410809844e-06, "loss": 1.7369033098220825, "step": 17034 }, { "epoch": 1.9821968815452642, "grad_norm": 0.952073335647583, "learning_rate": 2.472719810891921e-06, "loss": 1.8257465362548828, "step": 17035 }, { "epoch": 1.9823132417966023, "grad_norm": 1.3554259538650513, "learning_rate": 2.4407630172530447e-06, "loss": 2.225780963897705, "step": 17036 }, { "epoch": 1.9824296020479404, "grad_norm": 0.7803105711936951, "learning_rate": 2.409013056592829e-06, "loss": 1.7855231761932373, "step": 17037 }, { "epoch": 1.9825459622992785, "grad_norm": 0.8781646490097046, "learning_rate": 2.377469955437972e-06, "loss": 1.6855632066726685, "step": 17038 }, { "epoch": 1.9826623225506168, "grad_norm": 0.8910143971443176, "learning_rate": 2.346133740142392e-06, "loss": 1.5770187377929688, "step": 17039 }, { "epoch": 1.982778682801955, "grad_norm": 2.3922598361968994, "learning_rate": 2.315004436887147e-06, "loss": 1.8378634452819824, "step": 17040 }, { "epoch": 1.982895043053293, "grad_norm": 1.4085289239883423, "learning_rate": 2.2840820716804044e-06, "loss": 1.2471873760223389, "step": 17041 }, { "epoch": 1.983011403304631, "grad_norm": 0.8502440452575684, "learning_rate": 2.2533666703574986e-06, "loss": 2.18731689453125, "step": 17042 }, { "epoch": 1.9831277635559692, "grad_norm": 0.8113730549812317, "learning_rate": 2.222858258580762e-06, "loss": 1.3104450702667236, "step": 17043 }, { "epoch": 1.9832441238073075, "grad_norm": 0.4267795979976654, "learning_rate": 2.1925568618396672e-06, "loss": 1.3382185697555542, "step": 17044 }, { "epoch": 1.9833604840586454, "grad_norm": 0.7478358745574951, "learning_rate": 2.1624625054506574e-06, "loss": 1.2295459508895874, "step": 17045 }, { "epoch": 1.9834768443099837, "grad_norm": 0.9214382767677307, "learning_rate": 2.1325752145572585e-06, "loss": 1.761581301689148, "step": 17046 }, { "epoch": 1.9835932045613218, "grad_norm": 1.164535403251648, "learning_rate": 2.102895014129941e-06, "loss": 1.642533779144287, "step": 17047 }, { "epoch": 1.98370956481266, "grad_norm": 0.7327388525009155, "learning_rate": 2.0734219289662294e-06, "loss": 1.6174966096878052, "step": 17048 }, { "epoch": 1.9838259250639982, "grad_norm": 0.6612952351570129, "learning_rate": 2.04415598369051e-06, "loss": 1.4771499633789062, "step": 17049 }, { "epoch": 1.9839422853153361, "grad_norm": 1.1035068035125732, "learning_rate": 2.0150972027541683e-06, "loss": 2.0283989906311035, "step": 17050 }, { "epoch": 1.9840586455666744, "grad_norm": 0.7870768308639526, "learning_rate": 1.986245610435505e-06, "loss": 1.891442060470581, "step": 17051 }, { "epoch": 1.9841750058180125, "grad_norm": 1.0184316635131836, "learning_rate": 1.957601230839684e-06, "loss": 2.105795383453369, "step": 17052 }, { "epoch": 1.9842913660693506, "grad_norm": 0.637081503868103, "learning_rate": 1.929164087898783e-06, "loss": 1.7684093713760376, "step": 17053 }, { "epoch": 1.984407726320689, "grad_norm": 0.9872689247131348, "learning_rate": 1.900934205371685e-06, "loss": 1.7166271209716797, "step": 17054 }, { "epoch": 1.9845240865720268, "grad_norm": 0.937605082988739, "learning_rate": 1.8729116068441633e-06, "loss": 1.8342485427856445, "step": 17055 }, { "epoch": 1.9846404468233652, "grad_norm": 1.154232382774353, "learning_rate": 1.8450963157287948e-06, "loss": 1.5869507789611816, "step": 17056 }, { "epoch": 1.9847568070747033, "grad_norm": 0.9490689635276794, "learning_rate": 1.8174883552649345e-06, "loss": 1.6696425676345825, "step": 17057 }, { "epoch": 1.9848731673260414, "grad_norm": 1.2102314233779907, "learning_rate": 1.7900877485187427e-06, "loss": 2.0247316360473633, "step": 17058 }, { "epoch": 1.9849895275773797, "grad_norm": 1.2821338176727295, "learning_rate": 1.7628945183830735e-06, "loss": 2.555588722229004, "step": 17059 }, { "epoch": 1.9851058878287176, "grad_norm": 0.729620635509491, "learning_rate": 1.7359086875776143e-06, "loss": 1.203964352607727, "step": 17060 }, { "epoch": 1.9852222480800559, "grad_norm": 1.8376089334487915, "learning_rate": 1.7091302786487184e-06, "loss": 2.5118050575256348, "step": 17061 }, { "epoch": 1.985338608331394, "grad_norm": 0.8044419288635254, "learning_rate": 1.682559313969434e-06, "loss": 1.8362829685211182, "step": 17062 }, { "epoch": 1.985454968582732, "grad_norm": 2.1611287593841553, "learning_rate": 1.656195815739503e-06, "loss": 2.3237905502319336, "step": 17063 }, { "epoch": 1.9855713288340704, "grad_norm": 0.860439658164978, "learning_rate": 1.6300398059853339e-06, "loss": 1.275280475616455, "step": 17064 }, { "epoch": 1.9856876890854083, "grad_norm": 1.1611543893814087, "learning_rate": 1.6040913065600015e-06, "loss": 1.749091625213623, "step": 17065 }, { "epoch": 1.9858040493367466, "grad_norm": 1.2050745487213135, "learning_rate": 1.5783503391431642e-06, "loss": 2.003420352935791, "step": 17066 }, { "epoch": 1.9859204095880847, "grad_norm": 1.261755108833313, "learning_rate": 1.5528169252411184e-06, "loss": 2.0345664024353027, "step": 17067 }, { "epoch": 1.9860367698394228, "grad_norm": 0.9574849009513855, "learning_rate": 1.5274910861867441e-06, "loss": 1.9824529886245728, "step": 17068 }, { "epoch": 1.986153130090761, "grad_norm": 1.4872297048568726, "learning_rate": 1.5023728431395322e-06, "loss": 1.6794966459274292, "step": 17069 }, { "epoch": 1.986269490342099, "grad_norm": 0.9513314366340637, "learning_rate": 1.4774622170854456e-06, "loss": 1.4083211421966553, "step": 17070 }, { "epoch": 1.9863858505934373, "grad_norm": 1.67152738571167, "learning_rate": 1.4527592288370305e-06, "loss": 2.126451015472412, "step": 17071 }, { "epoch": 1.9865022108447754, "grad_norm": 0.8929867744445801, "learning_rate": 1.4282638990334163e-06, "loss": 2.0719988346099854, "step": 17072 }, { "epoch": 1.9866185710961135, "grad_norm": 1.0790057182312012, "learning_rate": 1.403976248140121e-06, "loss": 1.4106254577636719, "step": 17073 }, { "epoch": 1.9867349313474518, "grad_norm": 1.2029014825820923, "learning_rate": 1.3798962964492455e-06, "loss": 2.03983473777771, "step": 17074 }, { "epoch": 1.9868512915987897, "grad_norm": 1.0104286670684814, "learning_rate": 1.3560240640792809e-06, "loss": 1.7800471782684326, "step": 17075 }, { "epoch": 1.986967651850128, "grad_norm": 2.287759304046631, "learning_rate": 1.3323595709752445e-06, "loss": 1.3877421617507935, "step": 17076 }, { "epoch": 1.9870840121014661, "grad_norm": 1.668088436126709, "learning_rate": 1.3089028369085432e-06, "loss": 1.422184705734253, "step": 17077 }, { "epoch": 1.9872003723528042, "grad_norm": 1.308578372001648, "learning_rate": 1.285653881477028e-06, "loss": 1.9754209518432617, "step": 17078 }, { "epoch": 1.9873167326041425, "grad_norm": 0.8762708306312561, "learning_rate": 1.2626127241049112e-06, "loss": 2.35005521774292, "step": 17079 }, { "epoch": 1.9874330928554804, "grad_norm": 0.9744309782981873, "learning_rate": 1.2397793840428494e-06, "loss": 1.9885170459747314, "step": 17080 }, { "epoch": 1.9875494531068187, "grad_norm": 1.2646349668502808, "learning_rate": 1.217153880367833e-06, "loss": 1.6472795009613037, "step": 17081 }, { "epoch": 1.9876658133581568, "grad_norm": 0.9006063342094421, "learning_rate": 1.1947362319832123e-06, "loss": 1.7071051597595215, "step": 17082 }, { "epoch": 1.987782173609495, "grad_norm": 1.1014286279678345, "learning_rate": 1.1725264576187001e-06, "loss": 2.057929039001465, "step": 17083 }, { "epoch": 1.9878985338608333, "grad_norm": 1.234232783317566, "learning_rate": 1.150524575830314e-06, "loss": 2.0174217224121094, "step": 17084 }, { "epoch": 1.9880148941121711, "grad_norm": 0.7968063950538635, "learning_rate": 1.1287306050003499e-06, "loss": 2.200965404510498, "step": 17085 }, { "epoch": 1.9881312543635095, "grad_norm": 0.9131824374198914, "learning_rate": 1.1071445633374644e-06, "loss": 1.932489037513733, "step": 17086 }, { "epoch": 1.9882476146148476, "grad_norm": 0.7891659736633301, "learning_rate": 1.0857664688765368e-06, "loss": 1.4736599922180176, "step": 17087 }, { "epoch": 1.9883639748661857, "grad_norm": 0.7317597270011902, "learning_rate": 1.0645963394787239e-06, "loss": 1.6888689994812012, "step": 17088 }, { "epoch": 1.988480335117524, "grad_norm": 1.3953593969345093, "learning_rate": 1.043634192831433e-06, "loss": 2.2865216732025146, "step": 17089 }, { "epoch": 1.9885966953688619, "grad_norm": 0.9448012113571167, "learning_rate": 1.0228800464483212e-06, "loss": 1.9871470928192139, "step": 17090 }, { "epoch": 1.9887130556202002, "grad_norm": 0.814773440361023, "learning_rate": 1.0023339176692126e-06, "loss": 1.994532823562622, "step": 17091 }, { "epoch": 1.9888294158715383, "grad_norm": 1.235611081123352, "learning_rate": 9.819958236602089e-07, "loss": 1.8405191898345947, "step": 17092 }, { "epoch": 1.9889457761228764, "grad_norm": 1.0195863246917725, "learning_rate": 9.618657814135235e-07, "loss": 1.7252178192138672, "step": 17093 }, { "epoch": 1.9890621363742147, "grad_norm": 1.1896766424179077, "learning_rate": 9.419438077476195e-07, "loss": 2.3157622814178467, "step": 17094 }, { "epoch": 1.9891784966255526, "grad_norm": 0.97632896900177, "learning_rate": 9.222299193070438e-07, "loss": 1.944188117980957, "step": 17095 }, { "epoch": 1.989294856876891, "grad_norm": 1.0044236183166504, "learning_rate": 9.027241325625379e-07, "loss": 1.6136019229888916, "step": 17096 }, { "epoch": 1.989411217128229, "grad_norm": 1.1235471963882446, "learning_rate": 8.834264638110101e-07, "loss": 1.6288533210754395, "step": 17097 }, { "epoch": 1.989527577379567, "grad_norm": 1.3080356121063232, "learning_rate": 8.643369291753966e-07, "loss": 1.6491222381591797, "step": 17098 }, { "epoch": 1.9896439376309054, "grad_norm": 0.7366129159927368, "learning_rate": 8.454555446048007e-07, "loss": 1.2358709573745728, "step": 17099 }, { "epoch": 1.9897602978822433, "grad_norm": 0.8377755284309387, "learning_rate": 8.267823258744367e-07, "loss": 1.4986495971679688, "step": 17100 }, { "epoch": 1.9898766581335816, "grad_norm": 0.7464997172355652, "learning_rate": 8.083172885855472e-07, "loss": 1.817049264907837, "step": 17101 }, { "epoch": 1.9899930183849197, "grad_norm": 1.4850369691848755, "learning_rate": 7.900604481654583e-07, "loss": 1.552201747894287, "step": 17102 }, { "epoch": 1.9901093786362578, "grad_norm": 1.1473275423049927, "learning_rate": 7.720118198675796e-07, "loss": 2.2488112449645996, "step": 17103 }, { "epoch": 1.9902257388875961, "grad_norm": 1.3727850914001465, "learning_rate": 7.541714187713489e-07, "loss": 1.5123775005340576, "step": 17104 }, { "epoch": 1.990342099138934, "grad_norm": 0.7945429086685181, "learning_rate": 7.365392597821763e-07, "loss": 2.1154298782348633, "step": 17105 }, { "epoch": 1.9904584593902723, "grad_norm": 0.9603518843650818, "learning_rate": 7.191153576315556e-07, "loss": 1.8839110136032104, "step": 17106 }, { "epoch": 1.9905748196416104, "grad_norm": 1.209609866142273, "learning_rate": 7.018997268769534e-07, "loss": 1.5833916664123535, "step": 17107 }, { "epoch": 1.9906911798929485, "grad_norm": 0.9655042886734009, "learning_rate": 6.848923819018637e-07, "loss": 1.934118390083313, "step": 17108 }, { "epoch": 1.9908075401442868, "grad_norm": 0.7623652815818787, "learning_rate": 6.680933369156983e-07, "loss": 1.6247750520706177, "step": 17109 }, { "epoch": 1.9909239003956247, "grad_norm": 1.689736008644104, "learning_rate": 6.515026059539242e-07, "loss": 1.7358629703521729, "step": 17110 }, { "epoch": 1.991040260646963, "grad_norm": 0.9729975461959839, "learning_rate": 6.351202028778425e-07, "loss": 2.3445634841918945, "step": 17111 }, { "epoch": 1.9911566208983011, "grad_norm": 0.8968340158462524, "learning_rate": 6.189461413748099e-07, "loss": 1.6754522323608398, "step": 17112 }, { "epoch": 1.9912729811496392, "grad_norm": 1.106221318244934, "learning_rate": 6.029804349581003e-07, "loss": 1.5839704275131226, "step": 17113 }, { "epoch": 1.9913893414009776, "grad_norm": 0.9727177619934082, "learning_rate": 5.872230969668213e-07, "loss": 2.070936441421509, "step": 17114 }, { "epoch": 1.9915057016523154, "grad_norm": 0.8810447454452515, "learning_rate": 5.716741405660808e-07, "loss": 1.2058494091033936, "step": 17115 }, { "epoch": 1.9916220619036538, "grad_norm": 1.0046097040176392, "learning_rate": 5.56333578746876e-07, "loss": 1.7156262397766113, "step": 17116 }, { "epoch": 1.9917384221549919, "grad_norm": 2.521847724914551, "learning_rate": 5.412014243260654e-07, "loss": 2.227259635925293, "step": 17117 }, { "epoch": 1.99185478240633, "grad_norm": 0.7664212584495544, "learning_rate": 5.262776899463695e-07, "loss": 1.7450242042541504, "step": 17118 }, { "epoch": 1.9919711426576683, "grad_norm": 1.031174898147583, "learning_rate": 5.115623880764253e-07, "loss": 1.2340130805969238, "step": 17119 }, { "epoch": 1.9920875029090062, "grad_norm": 0.8846153020858765, "learning_rate": 4.970555310107039e-07, "loss": 1.7012537717819214, "step": 17120 }, { "epoch": 1.9922038631603445, "grad_norm": 1.7239105701446533, "learning_rate": 4.8275713086951e-07, "loss": 1.9222041368484497, "step": 17121 }, { "epoch": 1.9923202234116826, "grad_norm": 1.1637847423553467, "learning_rate": 4.686671995990099e-07, "loss": 2.6950764656066895, "step": 17122 }, { "epoch": 1.9924365836630207, "grad_norm": 0.743194043636322, "learning_rate": 4.54785748971176e-07, "loss": 1.230454444885254, "step": 17123 }, { "epoch": 1.992552943914359, "grad_norm": 1.5698922872543335, "learning_rate": 4.4111279058381414e-07, "loss": 2.1309638023376465, "step": 17124 }, { "epoch": 1.9926693041656969, "grad_norm": 0.6496080756187439, "learning_rate": 4.2764833586050875e-07, "loss": 1.2166160345077515, "step": 17125 }, { "epoch": 1.9927856644170352, "grad_norm": 1.0821380615234375, "learning_rate": 4.1439239605067793e-07, "loss": 1.5822539329528809, "step": 17126 }, { "epoch": 1.9929020246683733, "grad_norm": 1.207963466644287, "learning_rate": 4.0134498222951786e-07, "loss": 1.8514885902404785, "step": 17127 }, { "epoch": 1.9930183849197114, "grad_norm": 1.3446404933929443, "learning_rate": 3.885061052979755e-07, "loss": 2.513298988342285, "step": 17128 }, { "epoch": 1.9931347451710497, "grad_norm": 0.7007224559783936, "learning_rate": 3.758757759828313e-07, "loss": 1.501521110534668, "step": 17129 }, { "epoch": 1.9932511054223876, "grad_norm": 0.8802503943443298, "learning_rate": 3.6345400483653313e-07, "loss": 2.326404333114624, "step": 17130 }, { "epoch": 1.993367465673726, "grad_norm": 1.6365853548049927, "learning_rate": 3.512408022373625e-07, "loss": 2.256287097930908, "step": 17131 }, { "epoch": 1.993483825925064, "grad_norm": 0.7874504327774048, "learning_rate": 3.3923617838926836e-07, "loss": 1.769803524017334, "step": 17132 }, { "epoch": 1.9936001861764021, "grad_norm": 0.8995595574378967, "learning_rate": 3.2744014332206084e-07, "loss": 1.6195662021636963, "step": 17133 }, { "epoch": 1.9937165464277404, "grad_norm": 1.1096200942993164, "learning_rate": 3.1585270689113433e-07, "loss": 1.8987998962402344, "step": 17134 }, { "epoch": 1.9938329066790783, "grad_norm": 0.9082577228546143, "learning_rate": 3.0447387877766133e-07, "loss": 1.6749801635742188, "step": 17135 }, { "epoch": 1.9939492669304166, "grad_norm": 0.932040810585022, "learning_rate": 2.933036684885648e-07, "loss": 1.6287260055541992, "step": 17136 }, { "epoch": 1.9940656271817547, "grad_norm": 1.1317051649093628, "learning_rate": 2.823420853564074e-07, "loss": 1.5569648742675781, "step": 17137 }, { "epoch": 1.9941819874330928, "grad_norm": 0.957603394985199, "learning_rate": 2.715891385394742e-07, "loss": 2.6278324127197266, "step": 17138 }, { "epoch": 1.9942983476844311, "grad_norm": 1.0144809484481812, "learning_rate": 2.6104483702174554e-07, "loss": 2.242182731628418, "step": 17139 }, { "epoch": 1.994414707935769, "grad_norm": 0.7901127338409424, "learning_rate": 2.507091896128133e-07, "loss": 1.806416392326355, "step": 17140 }, { "epoch": 1.9945310681871073, "grad_norm": 0.7798228859901428, "learning_rate": 2.405822049480477e-07, "loss": 1.8944227695465088, "step": 17141 }, { "epoch": 1.9946474284384454, "grad_norm": 1.2113401889801025, "learning_rate": 2.306638914884307e-07, "loss": 1.3868088722229004, "step": 17142 }, { "epoch": 1.9947637886897835, "grad_norm": 2.8605759143829346, "learning_rate": 2.2095425752058363e-07, "loss": 1.9680366516113281, "step": 17143 }, { "epoch": 1.9948801489411219, "grad_norm": 1.6627236604690552, "learning_rate": 2.1145331115679511e-07, "loss": 1.763300895690918, "step": 17144 }, { "epoch": 1.9949965091924597, "grad_norm": 1.0817039012908936, "learning_rate": 2.0216106033502102e-07, "loss": 2.214765787124634, "step": 17145 }, { "epoch": 1.995112869443798, "grad_norm": 0.9868322014808655, "learning_rate": 1.9307751281880116e-07, "loss": 2.1063389778137207, "step": 17146 }, { "epoch": 1.9952292296951362, "grad_norm": 0.8751339316368103, "learning_rate": 1.842026761973703e-07, "loss": 1.3458478450775146, "step": 17147 }, { "epoch": 1.9953455899464743, "grad_norm": 0.6728445887565613, "learning_rate": 1.7553655788554724e-07, "loss": 1.7638545036315918, "step": 17148 }, { "epoch": 1.9954619501978126, "grad_norm": 1.017808437347412, "learning_rate": 1.6707916512376242e-07, "loss": 2.194371223449707, "step": 17149 }, { "epoch": 1.9955783104491505, "grad_norm": 1.0246667861938477, "learning_rate": 1.5883050497808583e-07, "loss": 1.6430977582931519, "step": 17150 }, { "epoch": 1.9956946707004888, "grad_norm": 1.533618688583374, "learning_rate": 1.5079058434017133e-07, "loss": 1.8051791191101074, "step": 17151 }, { "epoch": 1.9958110309518269, "grad_norm": 0.816594123840332, "learning_rate": 1.429594099272846e-07, "loss": 2.230257272720337, "step": 17152 }, { "epoch": 1.995927391203165, "grad_norm": 0.7715560793876648, "learning_rate": 1.35336988282303e-07, "loss": 1.6743894815444946, "step": 17153 }, { "epoch": 1.9960437514545033, "grad_norm": 0.8511298298835754, "learning_rate": 1.2792332577366006e-07, "loss": 2.0256309509277344, "step": 17154 }, { "epoch": 1.9961601117058412, "grad_norm": 1.0280979871749878, "learning_rate": 1.207184285953733e-07, "loss": 1.8939990997314453, "step": 17155 }, { "epoch": 1.9962764719571795, "grad_norm": 1.224553108215332, "learning_rate": 1.1372230276709972e-07, "loss": 1.2587385177612305, "step": 17156 }, { "epoch": 1.9963928322085176, "grad_norm": 0.9542074799537659, "learning_rate": 1.0693495413396925e-07, "loss": 1.4899256229400635, "step": 17157 }, { "epoch": 1.9965091924598557, "grad_norm": 1.8256295919418335, "learning_rate": 1.0035638836675132e-07, "loss": 1.883965015411377, "step": 17158 }, { "epoch": 1.996625552711194, "grad_norm": 0.9252011179924011, "learning_rate": 9.398661096179928e-08, "loss": 1.4859720468521118, "step": 17159 }, { "epoch": 1.996741912962532, "grad_norm": 0.7970852851867676, "learning_rate": 8.782562724093945e-08, "loss": 1.6562072038650513, "step": 17160 }, { "epoch": 1.9968582732138702, "grad_norm": 0.8281762599945068, "learning_rate": 8.187344235163762e-08, "loss": 1.4302575588226318, "step": 17161 }, { "epoch": 1.9969746334652083, "grad_norm": 0.9282841682434082, "learning_rate": 7.613006126688804e-08, "loss": 1.917481541633606, "step": 17162 }, { "epoch": 1.9970909937165464, "grad_norm": 0.8383802771568298, "learning_rate": 7.059548878518562e-08, "loss": 1.7103568315505981, "step": 17163 }, { "epoch": 1.9972073539678847, "grad_norm": 1.754489779472351, "learning_rate": 6.526972953060928e-08, "loss": 2.132436752319336, "step": 17164 }, { "epoch": 1.9973237142192226, "grad_norm": 1.1300612688064575, "learning_rate": 6.015278795282186e-08, "loss": 1.8514671325683594, "step": 17165 }, { "epoch": 1.997440074470561, "grad_norm": 0.9883387684822083, "learning_rate": 5.524466832690367e-08, "loss": 1.4344916343688965, "step": 17166 }, { "epoch": 1.997556434721899, "grad_norm": 1.1826591491699219, "learning_rate": 5.0545374753574456e-08, "loss": 1.6922580003738403, "step": 17167 }, { "epoch": 1.9976727949732371, "grad_norm": 1.0597585439682007, "learning_rate": 4.6054911159026934e-08, "loss": 1.6028897762298584, "step": 17168 }, { "epoch": 1.9977891552245752, "grad_norm": 1.1846660375595093, "learning_rate": 4.1773281295010015e-08, "loss": 1.9821676015853882, "step": 17169 }, { "epoch": 1.9979055154759133, "grad_norm": 1.4696571826934814, "learning_rate": 3.770048873874554e-08, "loss": 2.309558391571045, "step": 17170 }, { "epoch": 1.9980218757272517, "grad_norm": 1.2387373447418213, "learning_rate": 3.3836536893039335e-08, "loss": 2.132235527038574, "step": 17171 }, { "epoch": 1.9981382359785897, "grad_norm": 0.968146026134491, "learning_rate": 3.01814289861424e-08, "loss": 2.1886959075927734, "step": 17172 }, { "epoch": 1.9982545962299278, "grad_norm": 0.7605814933776855, "learning_rate": 2.673516807191745e-08, "loss": 1.1366908550262451, "step": 17173 }, { "epoch": 1.998370956481266, "grad_norm": 1.0478962659835815, "learning_rate": 2.3497757029616873e-08, "loss": 2.0702710151672363, "step": 17174 }, { "epoch": 1.998487316732604, "grad_norm": 0.6062107086181641, "learning_rate": 2.0469198564077028e-08, "loss": 1.5873658657073975, "step": 17175 }, { "epoch": 1.9986036769839424, "grad_norm": 0.6222151517868042, "learning_rate": 1.7649495205662723e-08, "loss": 1.7892314195632935, "step": 17176 }, { "epoch": 1.9987200372352805, "grad_norm": 0.9936891198158264, "learning_rate": 1.5038649310156195e-08, "loss": 1.8643393516540527, "step": 17177 }, { "epoch": 1.9988363974866186, "grad_norm": 1.0670055150985718, "learning_rate": 1.2636663058895881e-08, "loss": 2.0375561714172363, "step": 17178 }, { "epoch": 1.9989527577379567, "grad_norm": 0.9570886492729187, "learning_rate": 1.044353845874868e-08, "loss": 1.8194963932037354, "step": 17179 }, { "epoch": 1.9990691179892948, "grad_norm": 0.850773811340332, "learning_rate": 8.459277341998917e-09, "loss": 1.1402162313461304, "step": 17180 }, { "epoch": 1.999185478240633, "grad_norm": 0.9247174859046936, "learning_rate": 6.683881366514877e-09, "loss": 1.187821626663208, "step": 17181 }, { "epoch": 1.9993018384919712, "grad_norm": 1.4913983345031738, "learning_rate": 5.117352015610033e-09, "loss": 2.100032329559326, "step": 17182 }, { "epoch": 1.9994181987433093, "grad_norm": 0.7933056950569153, "learning_rate": 3.759690598070797e-09, "loss": 1.473681926727295, "step": 17183 }, { "epoch": 1.9995345589946474, "grad_norm": 1.1919058561325073, "learning_rate": 2.610898248267546e-09, "loss": 1.9284765720367432, "step": 17184 }, { "epoch": 1.9996509192459855, "grad_norm": 1.0598902702331543, "learning_rate": 1.6709759259603274e-09, "loss": 2.3416285514831543, "step": 17185 }, { "epoch": 1.9997672794973238, "grad_norm": 0.8613791465759277, "learning_rate": 9.399244164376431e-10, "loss": 1.9608399868011475, "step": 17186 }, { "epoch": 1.999883639748662, "grad_norm": 0.8014645576477051, "learning_rate": 4.177443305164452e-10, "loss": 1.8705365657806396, "step": 17187 }, { "epoch": 2.0, "grad_norm": 0.969385027885437, "learning_rate": 1.0443610443111595e-10, "loss": 1.7658429145812988, "step": 17188 } ], "logging_steps": 1, "max_steps": 17188, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 859, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.1324175718781813e+19, "train_batch_size": 16, "trial_name": null, "trial_params": null }