{"step": 50, "train_loss": 10.735717639923095, "lr": 6.25e-05, "grad_norm": 1.2113155126571655, "grad_norm_exchange": 0.18740491569042206, "grad_norm_vphi": 0.00341395428404212, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.0002210046877735e-06, 0.7843382954597473, 0.9536454677581787, 0.9900128841400146], "elapsed_s": 38.69459009170532, "exchange_scale": -0.001701180823147297} {"step": 100, "train_loss": 9.366235313415528, "lr": 0.000125, "grad_norm": 1.8253322839736938, "grad_norm_exchange": 0.04131782427430153, "grad_norm_vphi": 0.00026462136884219944, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.0010169262386626e-06, 0.7847276329994202, 0.9537814259529114, 0.9900409579277039], "elapsed_s": 76.26778650283813, "exchange_scale": -0.004905435722321272} {"step": 150, "train_loss": 7.095107135772705, "lr": 0.0001875, "grad_norm": 1.0719374418258667, "grad_norm_exchange": 0.059556424617767334, "grad_norm_vphi": 0.0019178729271516204, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.0014265399149735e-06, 0.7840145826339722, 0.9539505243301392, 0.9900752902030945], "elapsed_s": 113.92081665992737, "exchange_scale": -0.005602681078016758} {"step": 200, "train_loss": 5.915346460342407, "lr": 0.00025, "grad_norm": 2.5012388229370117, "grad_norm_exchange": 1.947786808013916, "grad_norm_vphi": 0.0017839892534539104, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [9.996451808547135e-07, 0.7819705605506897, 0.9535588622093201, 0.990121603012085], "elapsed_s": 151.89357590675354, "exchange_scale": -0.006395751144737005} {"step": 250, "train_loss": 5.654758243560791, "lr": 0.0003125, "grad_norm": 1.2386107444763184, "grad_norm_exchange": 0.25130078196525574, "grad_norm_vphi": 0.00108203268609941, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [9.999301937568816e-07, 0.7795256972312927, 0.952989399433136, 0.9900465607643127], "elapsed_s": 189.47353887557983, "exchange_scale": -0.006258411332964897} {"step": 300, "train_loss": 5.4462109375, "lr": 0.000375, "grad_norm": 1.2733908891677856, "grad_norm_exchange": 0.8311342000961304, "grad_norm_vphi": 0.0016935118474066257, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [9.999539543059655e-07, 0.7768216729164124, 0.9528874158859253, 0.9901385307312012], "elapsed_s": 227.1637728214264, "exchange_scale": -0.006147380452603102} {"step": 350, "train_loss": 5.064507522583008, "lr": 0.00043750000000000006, "grad_norm": 0.8232427835464478, "grad_norm_exchange": 0.3032451272010803, "grad_norm_vphi": 0.0016481023048982024, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.0001036798712448e-06, 0.7742690443992615, 0.9534334540367126, 0.9903045296669006], "elapsed_s": 264.720196723938, "exchange_scale": -0.005121992435306311} {"step": 400, "train_loss": 4.491395606994629, "lr": 0.0005, "grad_norm": 0.5234787464141846, "grad_norm_exchange": 0.1224956065416336, "grad_norm_vphi": 0.0010677026584744453, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.0058532780021778e-06, 0.7713053226470947, 0.9528453946113586, 0.9904308915138245], "elapsed_s": 302.4109089374542, "exchange_scale": -0.006380455568432808} {"step": 400, "val_loss": 4.308983325958252, "val_ppl": 74.36484568636672} {"step": 450, "train_loss": 4.102670602798462, "lr": 0.0004999878283572284, "grad_norm": 0.732074499130249, "grad_norm_exchange": 0.07761507481336594, "grad_norm_vphi": 0.0010823225602507591, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.020903255266603e-06, 0.7686778903007507, 0.9516043066978455, 0.9904471635818481], "elapsed_s": 346.0465314388275, "exchange_scale": -0.014696536585688591} {"step": 500, "train_loss": 3.887433099746704, "lr": 0.0004999503159985978, "grad_norm": 0.964175820350647, "grad_norm_exchange": 0.020789802074432373, "grad_norm_vphi": 0.0011327589163556695, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.0403639407741139e-06, 0.7661571502685547, 0.9504058361053467, 0.9903075695037842], "elapsed_s": 383.6526975631714, "exchange_scale": -0.019501708447933197} {"step": 550, "train_loss": 3.744117431640625, "lr": 0.0004998874616579676, "grad_norm": 0.9214540123939514, "grad_norm_exchange": 0.11109906435012817, "grad_norm_vphi": 0.0011399940121918917, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.0589303656161064e-06, 0.7635288238525391, 0.9492416381835938, 0.9901236891746521], "elapsed_s": 421.39558243751526, "exchange_scale": -0.022922443225979805} {"step": 600, "train_loss": 3.6330426692962647, "lr": 0.0004997992717080183, "grad_norm": 0.7373350858688354, "grad_norm_exchange": 0.12457813322544098, "grad_norm_vphi": 0.0008125093299895525, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.0815059567903518e-06, 0.7604596018791199, 0.9480984210968018, 0.9899068474769592], "elapsed_s": 459.1642892360687, "exchange_scale": -0.02464558742940426} {"step": 650, "train_loss": 3.5386134004592895, "lr": 0.0004996857550901596, "grad_norm": 0.7778695821762085, "grad_norm_exchange": 0.11103705316781998, "grad_norm_vphi": 0.0009792671771720052, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.1033354212486302e-06, 0.75715172290802, 0.9470006227493286, 0.9896893501281738], "elapsed_s": 497.02129197120667, "exchange_scale": -0.02611404098570347} {"step": 700, "train_loss": 3.453303828239441, "lr": 0.0004995469233136228, "grad_norm": 0.9283523559570312, "grad_norm_exchange": 0.042273178696632385, "grad_norm_vphi": 0.0007983493851497769, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.1250688203290338e-06, 0.7536560893058777, 0.9458930492401123, 0.9894535541534424], "elapsed_s": 534.6212291717529, "exchange_scale": -0.026468530297279358} {"step": 750, "train_loss": 3.3925917243957517, "lr": 0.0004993827904542953, "grad_norm": 0.6577854752540588, "grad_norm_exchange": 0.04238423705101013, "grad_norm_vphi": 0.0008633147226646543, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.146971044363454e-06, 0.7499949336051941, 0.9446729421615601, 0.9892065525054932], "elapsed_s": 572.3426048755646, "exchange_scale": -0.027982721105217934} {"step": 800, "train_loss": 3.34983482837677, "lr": 0.0004991933731532918, "grad_norm": 0.6903600096702576, "grad_norm_exchange": 0.09878619015216827, "grad_norm_vphi": 0.000752938911318779, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.1706855502779945e-06, 0.7462835311889648, 0.9435026049613953, 0.98896324634552], "elapsed_s": 609.8922860622406, "exchange_scale": -0.02913730777800083} {"step": 800, "val_loss": 3.3574038445949554, "val_ppl": 28.71454660173227} {"step": 850, "train_loss": 3.309083104133606, "lr": 0.0004989786906152684, "grad_norm": 0.7526919841766357, "grad_norm_exchange": 0.07577117532491684, "grad_norm_vphi": 0.0007505117100663483, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.193364369100891e-06, 0.74245285987854, 0.9422762989997864, 0.9887070655822754], "elapsed_s": 653.6671171188354, "exchange_scale": -0.029393615201115608} {"step": 900, "train_loss": 3.251116180419922, "lr": 0.0004987387646064748, "grad_norm": 0.9586079716682434, "grad_norm_exchange": 0.1480468511581421, "grad_norm_vphi": 0.0008824962424114347, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.21605205549713e-06, 0.7387290596961975, 0.9410456418991089, 0.9884338974952698], "elapsed_s": 691.1790435314178, "exchange_scale": -0.029249388724565506} {"step": 950, "train_loss": 3.222291836738586, "lr": 0.000498473619452548, "grad_norm": 0.673048734664917, "grad_norm_exchange": 0.03592485561966896, "grad_norm_vphi": 0.0007530678412877023, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.235484887729399e-06, 0.7347843050956726, 0.9397901296615601, 0.9881677031517029], "elapsed_s": 728.7981328964233, "exchange_scale": -0.02926764450967312} {"step": 1000, "train_loss": 3.1657316970825193, "lr": 0.0004981832820360452, "grad_norm": 0.707302451133728, "grad_norm_exchange": 0.05821112170815468, "grad_norm_vphi": 0.0007228281465359032, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.25630128877674e-06, 0.7308279275894165, 0.9385043382644653, 0.9878935217857361], "elapsed_s": 766.3744957447052, "exchange_scale": -0.03035726584494114} {"step": 1050, "train_loss": 3.1370191049575804, "lr": 0.0004978677817937189, "grad_norm": 0.6412765979766846, "grad_norm_exchange": 0.21650472283363342, "grad_norm_vphi": 0.000855078746099025, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.277157707590959e-06, 0.7273361086845398, 0.937260627746582, 0.9876093864440918], "elapsed_s": 804.0408840179443, "exchange_scale": -0.0318235419690609} {"step": 1100, "train_loss": 3.125686626434326, "lr": 0.000497527150713532, "grad_norm": 0.7094362378120422, "grad_norm_exchange": 0.14535947144031525, "grad_norm_vphi": 0.0007750660879537463, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.2978749737158068e-06, 0.7234654426574707, 0.9359796643257141, 0.9873178601264954], "elapsed_s": 841.6021361351013, "exchange_scale": -0.03324737772345543} {"step": 1150, "train_loss": 3.0784267807006835, "lr": 0.0004971614233314151, "grad_norm": 0.8729233145713806, "grad_norm_exchange": 0.06205297261476517, "grad_norm_vphi": 0.001160109299235046, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.317803139500029e-06, 0.7193840146064758, 0.934656023979187, 0.9870166182518005], "elapsed_s": 879.2999241352081, "exchange_scale": -0.034495942294597626} {"step": 1200, "train_loss": 3.0718584775924684, "lr": 0.0004967706367277644, "grad_norm": 0.7081983089447021, "grad_norm_exchange": 0.07027925550937653, "grad_norm_vphi": 0.0008247635560110211, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.3371073919188348e-06, 0.7154524326324463, 0.9334602952003479, 0.9867363572120667], "elapsed_s": 916.7409288883209, "exchange_scale": -0.03588217496871948} {"step": 1200, "val_loss": 3.0893099188804625, "val_ppl": 21.961917240859844} {"step": 1250, "train_loss": 3.045678653717041, "lr": 0.0004963548305236819, "grad_norm": 0.7734878659248352, "grad_norm_exchange": 0.04436572268605232, "grad_norm_vphi": 0.0008471467299386859, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.3534236131818034e-06, 0.71182781457901, 0.9321885108947754, 0.9864534735679626], "elapsed_s": 960.3070130348206, "exchange_scale": -0.03678915277123451} {"step": 1300, "train_loss": 3.0099577713012695, "lr": 0.0004959140468769596, "grad_norm": 0.6722947359085083, "grad_norm_exchange": 0.11096058785915375, "grad_norm_vphi": 0.000812734360806644, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.3731070112044108e-06, 0.7080655694007874, 0.9310033321380615, 0.9861714243888855], "elapsed_s": 998.0294544696808, "exchange_scale": -0.03828781098127365} {"step": 1350, "train_loss": 2.988178515434265, "lr": 0.0004954483304778039, "grad_norm": 0.7213640809059143, "grad_norm_exchange": 0.03254697471857071, "grad_norm_vphi": 0.0008542901487089694, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.3932545925854356e-06, 0.7047551870346069, 0.9297344088554382, 0.9858818054199219], "elapsed_s": 1035.9277029037476, "exchange_scale": -0.03968434780836105} {"step": 1400, "train_loss": 2.962182335853577, "lr": 0.0004949577285443053, "grad_norm": 0.7111856937408447, "grad_norm_exchange": 0.035597000271081924, "grad_norm_vphi": 0.0011430532904341817, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.4128190741757862e-06, 0.7011443972587585, 0.9284769296646118, 0.9855921268463135], "elapsed_s": 1073.7133829593658, "exchange_scale": -0.04174920544028282} {"step": 1450, "train_loss": 2.917003002166748, "lr": 0.0004944422908176503, "grad_norm": 0.6959089636802673, "grad_norm_exchange": 0.11447823792695999, "grad_norm_vphi": 0.0010642202105373144, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.4353195183502976e-06, 0.6971988081932068, 0.927270770072937, 0.9853019118309021], "elapsed_s": 1111.2885715961456, "exchange_scale": -0.04255751147866249} {"step": 1500, "train_loss": 2.9447687578201296, "lr": 0.0004939020695570793, "grad_norm": 0.7619471549987793, "grad_norm_exchange": 0.19032804667949677, "grad_norm_vphi": 0.0008267447701655328, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.4564286630047718e-06, 0.6932862997055054, 0.9259931445121765, 0.9850172400474548], "elapsed_s": 1148.9044697284698, "exchange_scale": -0.045273710042238235} {"step": 1550, "train_loss": 2.9331689977645876, "lr": 0.0004933371195345874, "grad_norm": 0.7492120265960693, "grad_norm_exchange": 0.2535122334957123, "grad_norm_vphi": 0.001009089988656342, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.473707584409567e-06, 0.6900461912155151, 0.9247824549674988, 0.9847356081008911], "elapsed_s": 1186.2878894805908, "exchange_scale": -0.04839552193880081} {"step": 1600, "train_loss": 2.8971718978881835, "lr": 0.000492747498029371, "grad_norm": 0.6624844074249268, "grad_norm_exchange": 0.18440309166908264, "grad_norm_vphi": 0.0009629753767512739, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.4917048929419252e-06, 0.6863934397697449, 0.923500657081604, 0.9844947457313538], "elapsed_s": 1223.8574879169464, "exchange_scale": -0.051609642803668976} {"step": 1600, "val_loss": 2.9417566299438476, "val_ppl": 18.94910365617147} {"step": 1650, "train_loss": 2.8935552263259887, "lr": 0.000492133264822021, "grad_norm": 0.5784295201301575, "grad_norm_exchange": 0.05920685827732086, "grad_norm_vphi": 0.000787894066888839, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.5145701581786852e-06, 0.6830248832702637, 0.9225513339042664, 0.9842039942741394], "elapsed_s": 1267.2567574977875, "exchange_scale": -0.05515521764755249} {"step": 1700, "train_loss": 2.8421229696273804, "lr": 0.0004914944821884616, "grad_norm": 0.6779302358627319, "grad_norm_exchange": 0.15924449265003204, "grad_norm_vphi": 0.001040286268107593, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.536160539217235e-06, 0.6795663237571716, 0.9214335680007935, 0.9839591383934021], "elapsed_s": 1304.7862277030945, "exchange_scale": -0.05960502848029137} {"step": 1750, "train_loss": 2.839250283241272, "lr": 0.0004908312148936357, "grad_norm": 0.6606699228286743, "grad_norm_exchange": 0.07154949754476547, "grad_norm_vphi": 0.0010128134163096547, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.5552038803434698e-06, 0.6760760545730591, 0.9203491806983948, 0.9836551547050476], "elapsed_s": 1342.1685509681702, "exchange_scale": -0.06288180500268936} {"step": 1800, "train_loss": 2.8299870395660403, "lr": 0.0004901435301849393, "grad_norm": 0.7442467212677002, "grad_norm_exchange": 0.2126190960407257, "grad_norm_vphi": 0.001240569748915732, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.572778501213179e-06, 0.6731129288673401, 0.9194254279136658, 0.9833868145942688], "elapsed_s": 1379.7401745319366, "exchange_scale": -0.06861528754234314} {"step": 1850, "train_loss": 2.799956841468811, "lr": 0.0004894314977854029, "grad_norm": 0.6611544489860535, "grad_norm_exchange": 0.08260274678468704, "grad_norm_vphi": 0.0011125464225187898, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.5890037730059703e-06, 0.6699761152267456, 0.918336808681488, 0.9831019043922424], "elapsed_s": 1417.1236696243286, "exchange_scale": -0.07416015863418579} {"step": 1900, "train_loss": 2.7944251489639282, "lr": 0.0004886951898866224, "grad_norm": 0.6518781781196594, "grad_norm_exchange": 0.1747307926416397, "grad_norm_vphi": 0.0011158282868564129, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.6078309954536962e-06, 0.6669016480445862, 0.9173358678817749, 0.9828365445137024], "elapsed_s": 1454.6670815944672, "exchange_scale": -0.0798894539475441} {"step": 1950, "train_loss": 2.778619065284729, "lr": 0.0004879346811414404, "grad_norm": 0.6383560299873352, "grad_norm_exchange": 0.056751832365989685, "grad_norm_vphi": 0.0011967932805418968, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.6292630107272998e-06, 0.6639911532402039, 0.9164338111877441, 0.9825892448425293], "elapsed_s": 1492.1338889598846, "exchange_scale": -0.08414430916309357} {"step": 2000, "train_loss": 2.7431361198425295, "lr": 0.0004871500486563761, "grad_norm": 0.6791995763778687, "grad_norm_exchange": 0.10102373361587524, "grad_norm_vphi": 0.001205992535687983, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.6476980135848862e-06, 0.6610056161880493, 0.915619969367981, 0.9823352694511414], "elapsed_s": 1529.7253241539001, "exchange_scale": -0.08759690821170807} {"step": 2000, "val_loss": 2.8375774919986725, "val_ppl": 17.074352639973462} {"step": 2050, "train_loss": 2.7452053928375246, "lr": 0.00048634137198380904, "grad_norm": 0.6533215045928955, "grad_norm_exchange": 0.06687762588262558, "grad_norm_vphi": 0.0011139661073684692, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.6687909010215662e-06, 0.6585569977760315, 0.9145583510398865, 0.982060432434082], "elapsed_s": 1573.241644859314, "exchange_scale": -0.09187762439250946} {"step": 2100, "train_loss": 2.719812469482422, "lr": 0.0004855087331139121, "grad_norm": 0.6620327830314636, "grad_norm_exchange": 0.11617132276296616, "grad_norm_vphi": 0.0011104476870968938, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.6857493392308243e-06, 0.6553741097450256, 0.9136936664581299, 0.9817901253700256], "elapsed_s": 1610.768654346466, "exchange_scale": -0.0962505042552948} {"step": 2150, "train_loss": 2.74181351184845, "lr": 0.0004846522164663396, "grad_norm": 0.6393585205078125, "grad_norm_exchange": 0.09186982363462448, "grad_norm_vphi": 0.001173448166809976, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.7043959132934106e-06, 0.6531606912612915, 0.9128919839859009, 0.9815829396247864], "elapsed_s": 1648.160712480545, "exchange_scale": -0.10094237327575684} {"step": 2200, "train_loss": 2.7146255731582642, "lr": 0.0004837719088816677, "grad_norm": 0.5988771915435791, "grad_norm_exchange": 0.14998675882816315, "grad_norm_vphi": 0.001070843543857336, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.7174000959130353e-06, 0.6505475044250488, 0.9120349884033203, 0.9813527464866638], "elapsed_s": 1685.6974611282349, "exchange_scale": -0.10636617988348007} {"step": 2250, "train_loss": 2.7056655645370484, "lr": 0.00048286789961259004, "grad_norm": 0.6509333848953247, "grad_norm_exchange": 0.08552737534046173, "grad_norm_vphi": 0.0011366631370037794, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.7368638509651646e-06, 0.648357629776001, 0.9110373854637146, 0.9811171889305115], "elapsed_s": 1723.0705680847168, "exchange_scale": -0.11083689332008362} {"step": 2300, "train_loss": 2.686937565803528, "lr": 0.0004819402803148684, "grad_norm": 0.7234294414520264, "grad_norm_exchange": 0.1738182008266449, "grad_norm_vphi": 0.001658255816437304, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.7529536080473918e-06, 0.6459119915962219, 0.9101336598396301, 0.9808793663978577], "elapsed_s": 1760.5965921878815, "exchange_scale": -0.11368405818939209} {"step": 2350, "train_loss": 2.67051233291626, "lr": 0.00048098914503804, "grad_norm": 0.5677809119224548, "grad_norm_exchange": 0.08104540407657623, "grad_norm_vphi": 0.0012302035465836525, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.7703756611808785e-06, 0.6435154676437378, 0.9092651009559631, 0.9806495904922485], "elapsed_s": 1797.9421496391296, "exchange_scale": -0.1187753900885582} {"step": 2400, "train_loss": 2.673477416038513, "lr": 0.0004800145902158819, "grad_norm": 0.6731405854225159, "grad_norm_exchange": 0.1503412425518036, "grad_norm_vphi": 0.0013517714105546474, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.7876450328913052e-06, 0.6415742039680481, 0.9083136320114136, 0.9804095029830933], "elapsed_s": 1835.3345756530762, "exchange_scale": -0.12278979271650314} {"step": 2400, "val_loss": 2.761968266963959, "val_ppl": 15.830971882147644} {"step": 2450, "train_loss": 2.64260751247406, "lr": 0.00047901671465663397, "grad_norm": 0.6506960988044739, "grad_norm_exchange": 0.1864033043384552, "grad_norm_vphi": 0.0012471192749217153, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.8032884554486373e-06, 0.6401296854019165, 0.9076169729232788, 0.9801514744758606], "elapsed_s": 1878.8888816833496, "exchange_scale": -0.12704768776893616} {"step": 2500, "train_loss": 2.654193835258484, "lr": 0.00047799561953298063, "grad_norm": 0.6498052477836609, "grad_norm_exchange": 0.10507525503635406, "grad_norm_vphi": 0.0013845275389030576, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.8193029518442927e-06, 0.6382409334182739, 0.9063790440559387, 0.9799324870109558], "elapsed_s": 1916.2252538204193, "exchange_scale": -0.1307138055562973} {"step": 2550, "train_loss": 2.623216333389282, "lr": 0.0004769514083717935, "grad_norm": 0.6305498480796814, "grad_norm_exchange": 0.18734121322631836, "grad_norm_vphi": 0.001425879425369203, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.8338201925871545e-06, 0.6365327835083008, 0.9056119918823242, 0.9797660708427429], "elapsed_s": 1953.7049310207367, "exchange_scale": -0.13437627255916595} {"step": 2600, "train_loss": 2.6200959634780885, "lr": 0.0004758841870436346, "grad_norm": 0.6666208505630493, "grad_norm_exchange": 0.12204091995954514, "grad_norm_vphi": 0.0015332679031416774, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.85285500720056e-06, 0.6348372101783752, 0.9048495888710022, 0.9795475602149963], "elapsed_s": 1991.087396621704, "exchange_scale": -0.13884440064430237} {"step": 2650, "train_loss": 2.5968609476089477, "lr": 0.0004747940637520226, "grad_norm": 0.6685252785682678, "grad_norm_exchange": 0.12699241936206818, "grad_norm_vphi": 0.0017124898731708527, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.8671127008929034e-06, 0.6328737139701843, 0.9039490222930908, 0.9793506264686584], "elapsed_s": 2028.5507423877716, "exchange_scale": -0.14238475263118744} {"step": 2700, "train_loss": 2.5967616033554077, "lr": 0.0004736811490224622, "grad_norm": 0.6712045073509216, "grad_norm_exchange": 0.12064269185066223, "grad_norm_vphi": 0.0018820108380168676, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.881553998828167e-06, 0.631506621837616, 0.9032604694366455, 0.9792976379394531], "elapsed_s": 2065.8683178424835, "exchange_scale": -0.14527326822280884} {"step": 2750, "train_loss": 2.575402607917786, "lr": 0.00047254555569123835, "grad_norm": 0.6417486071586609, "grad_norm_exchange": 0.11778359860181808, "grad_norm_vphi": 0.001463401480577886, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.8984443386216299e-06, 0.6305781602859497, 0.902479350566864, 0.9792214035987854], "elapsed_s": 2103.33274269104, "exchange_scale": -0.14862236380577087} {"step": 2800, "train_loss": 2.564192223548889, "lr": 0.00047138739889397555, "grad_norm": 0.6214755177497864, "grad_norm_exchange": 0.10082126408815384, "grad_norm_vphi": 0.001638874993659556, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.9084811810898827e-06, 0.6292877197265625, 0.9018064141273499, 0.9790380001068115], "elapsed_s": 2140.6284096240997, "exchange_scale": -0.1523287147283554} {"step": 2800, "val_loss": 2.671449691057205, "val_ppl": 14.460917867892658} {"step": 2850, "train_loss": 2.5696367406845093, "lr": 0.00047020679605396486, "grad_norm": 0.6158849000930786, "grad_norm_exchange": 0.1325351744890213, "grad_norm_vphi": 0.0016857737209647894, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.926967343024444e-06, 0.6276715397834778, 0.9009198546409607, 0.9789542555809021], "elapsed_s": 2184.1964209079742, "exchange_scale": -0.15563108026981354} {"step": 2900, "train_loss": 2.5509891605377195, "lr": 0.00046900386687025853, "grad_norm": 0.6246640682220459, "grad_norm_exchange": 0.1093202605843544, "grad_norm_vphi": 0.0017915595090016723, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.9415751921769697e-06, 0.6259649991989136, 0.9001258611679077, 0.9787893891334534], "elapsed_s": 2221.486897468567, "exchange_scale": -0.15923015773296356} {"step": 2950, "train_loss": 2.534701690673828, "lr": 0.00046777873330553383, "grad_norm": 0.6794106364250183, "grad_norm_exchange": 0.14118996262550354, "grad_norm_vphi": 0.0020457643549889326, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.9589224393712357e-06, 0.6243980526924133, 0.8991411328315735, 0.9786375164985657], "elapsed_s": 2258.930177927017, "exchange_scale": -0.1639416366815567} {"step": 3000, "train_loss": 2.540053267478943, "lr": 0.00046653151957372753, "grad_norm": 0.6992831230163574, "grad_norm_exchange": 0.12218647450208664, "grad_norm_vphi": 0.0020271223038434982, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.9711972072400386e-06, 0.6232291460037231, 0.8983547687530518, 0.9784992933273315], "elapsed_s": 2296.229641675949, "exchange_scale": -0.1674705296754837} {"step": 3050, "train_loss": 2.5064321756362915, "lr": 0.000465262352127442, "grad_norm": 0.651799738407135, "grad_norm_exchange": 0.13338719308376312, "grad_norm_vphi": 0.0024933950044214725, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.9817241536657093e-06, 0.6225718855857849, 0.8977352380752563, 0.9783455729484558], "elapsed_s": 2333.670962333679, "exchange_scale": -0.16931182146072388} {"step": 3100, "train_loss": 2.4979369974136354, "lr": 0.00046397135964512464, "grad_norm": 0.6041609644889832, "grad_norm_exchange": 0.11656191945075989, "grad_norm_vphi": 0.0017424436518922448, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [1.9942483504564734e-06, 0.621833086013794, 0.8969284892082214, 0.9782004952430725], "elapsed_s": 2370.9615001678467, "exchange_scale": -0.1734558492898941} {"step": 3150, "train_loss": 2.499946665763855, "lr": 0.00046265867301802126, "grad_norm": 0.6208639740943909, "grad_norm_exchange": 0.126822367310524, "grad_norm_vphi": 0.0025481784250587225, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [2.0086899894522503e-06, 0.6212577819824219, 0.8962067365646362, 0.9781453609466553], "elapsed_s": 2408.273099899292, "exchange_scale": -0.1761959344148636} {"step": 3200, "train_loss": 2.4874227333068846, "lr": 0.00046132442533690533, "grad_norm": 0.683434784412384, "grad_norm_exchange": 0.14430686831474304, "grad_norm_vphi": 0.002695747185498476, "gamma": 0.30000001192092896, "gumbel_tau": 1.0, "xi_alphas": [2.0200375274725957e-06, 0.6199281811714172, 0.895392656326294, 0.9780876636505127], "elapsed_s": 2445.8096928596497, "exchange_scale": -0.17886468768119812} {"step": 3200, "val_loss": 2.587310367822647, "val_ppl": 13.293967591969599} {"step": 3250, "train_loss": 2.4820151329040527, "lr": 0.000459968751878584, "grad_norm": 0.633621871471405, "grad_norm_exchange": 0.14650849997997284, "grad_norm_vphi": 0.0026574027724564075, "gamma": 0.30000001192092896, "gumbel_tau": 0.9972658157348633, "xi_alphas": [2.0321701867942465e-06, 0.6185626983642578, 0.8947170972824097, 0.9779515266418457], "elapsed_s": 2489.272466659546, "exchange_scale": -0.18373000621795654} {"step": 3300, "train_loss": 2.4742252254486083, "lr": 0.0004585917900921829, "grad_norm": 0.6160136461257935, "grad_norm_exchange": 0.14786556363105774, "grad_norm_vphi": 0.0025972023140639067, "gamma": 0.30000001192092896, "gumbel_tau": 0.9945316910743713, "xi_alphas": [2.0449604107852792e-06, 0.6177421808242798, 0.894099235534668, 0.977828860282898], "elapsed_s": 2526.6847746372223, "exchange_scale": -0.18507477641105652} {"step": 3350, "train_loss": 2.476265215873718, "lr": 0.00045719367958521037, "grad_norm": 0.6352079510688782, "grad_norm_exchange": 0.1303633153438568, "grad_norm_vphi": 0.003949078265577555, "gamma": 0.30000001192092896, "gumbel_tau": 0.9917975068092346, "xi_alphas": [2.0585321180988103e-06, 0.616793692111969, 0.8935223817825317, 0.9777886271476746], "elapsed_s": 2563.973461866379, "exchange_scale": -0.18712012469768524} {"step": 3400, "train_loss": 2.4403582239151, "lr": 0.0004557745621094027, "grad_norm": 0.6407897472381592, "grad_norm_exchange": 0.18497906625270844, "grad_norm_vphi": 0.0031208840664476156, "gamma": 0.30000001192092896, "gumbel_tau": 0.9890633821487427, "xi_alphas": [2.072766847049934e-06, 0.615760326385498, 0.8926767110824585, 0.9777352809906006], "elapsed_s": 2601.4056951999664, "exchange_scale": -0.18860910832881927} {"step": 3450, "train_loss": 2.450519313812256, "lr": 0.00045433458154635245, "grad_norm": 0.6342135667800903, "grad_norm_exchange": 0.14637577533721924, "grad_norm_vphi": 0.0026136189699172974, "gamma": 0.30000001192092896, "gumbel_tau": 0.986329197883606, "xi_alphas": [2.085627556880354e-06, 0.6149656176567078, 0.8918778300285339, 0.9775673151016235], "elapsed_s": 2638.71914935112, "exchange_scale": -0.19194138050079346} {"step": 3500, "train_loss": 2.4340340662002564, "lr": 0.0004528738838929205, "grad_norm": 0.6359691619873047, "grad_norm_exchange": 0.1602782905101776, "grad_norm_vphi": 0.0028770766220986843, "gamma": 0.30000001192092896, "gumbel_tau": 0.9835950136184692, "xi_alphas": [2.1014284357079305e-06, 0.6144237518310547, 0.8913914561271667, 0.9774078726768494], "elapsed_s": 2676.1352977752686, "exchange_scale": -0.1961699277162552} {"step": 3550, "train_loss": 2.4472886323928833, "lr": 0.00045139261724643377, "grad_norm": 0.6746929287910461, "grad_norm_exchange": 0.14278990030288696, "grad_norm_vphi": 0.0027462139260023832, "gamma": 0.30000001192092896, "gumbel_tau": 0.9808608889579773, "xi_alphas": [2.1085286334709963e-06, 0.6132400631904602, 0.8907248377799988, 0.9773338437080383], "elapsed_s": 2713.4030594825745, "exchange_scale": -0.19848206639289856} {"step": 3600, "train_loss": 2.4170549488067627, "lr": 0.00044989093178966966, "grad_norm": 0.6305940747261047, "grad_norm_exchange": 0.16684137284755707, "grad_norm_vphi": 0.002504212548956275, "gamma": 0.30000001192092896, "gumbel_tau": 0.9781267046928406, "xi_alphas": [2.1191465293668443e-06, 0.6129547953605652, 0.89009028673172, 0.9772695302963257], "elapsed_s": 2750.858076810837, "exchange_scale": -0.2002360224723816} {"step": 3600, "val_loss": 2.5386301457881926, "val_ppl": 12.662313561424606} {"step": 3650, "train_loss": 2.399977960586548, "lr": 0.00044836897977562965, "grad_norm": 0.6161808967590332, "grad_norm_exchange": 0.2033817619085312, "grad_norm_vphi": 0.002899462589994073, "gamma": 0.30000001192092896, "gumbel_tau": 0.9753925204277039, "xi_alphas": [2.1326711703295587e-06, 0.611973762512207, 0.8894480466842651, 0.9771585464477539], "elapsed_s": 2794.271610736847, "exchange_scale": -0.20385220646858215} {"step": 3700, "train_loss": 2.4288591527938843, "lr": 0.0004468269155121023, "grad_norm": 0.642180860042572, "grad_norm_exchange": 0.15998125076293945, "grad_norm_vphi": 0.0026400131173431873, "gamma": 0.30000001192092896, "gumbel_tau": 0.9726583957672119, "xi_alphas": [2.141782033504569e-06, 0.6110957860946655, 0.8886509537696838, 0.9769929647445679], "elapsed_s": 2831.733435869217, "exchange_scale": -0.20534132421016693} {"step": 3750, "train_loss": 2.4097083473205565, "lr": 0.0004452648953460187, "grad_norm": 0.6354423761367798, "grad_norm_exchange": 0.17023146152496338, "grad_norm_vphi": 0.0026914235204458237, "gamma": 0.30000001192092896, "gumbel_tau": 0.9699242115020752, "xi_alphas": [2.149207148249843e-06, 0.6102091670036316, 0.8879870772361755, 0.9769464135169983], "elapsed_s": 2869.1242547035217, "exchange_scale": -0.20785769820213318} {"step": 3800, "train_loss": 2.3834946870803835, "lr": 0.00044368307764760054, "grad_norm": 0.6435165405273438, "grad_norm_exchange": 0.18273229897022247, "grad_norm_vphi": 0.002353107091039419, "gamma": 0.30000001192092896, "gumbel_tau": 0.9671900868415833, "xi_alphas": [2.1528628622036194e-06, 0.6097157597541809, 0.887310802936554, 0.9768857955932617], "elapsed_s": 2906.465768814087, "exchange_scale": -0.21015022695064545} {"step": 3850, "train_loss": 2.3925112819671632, "lr": 0.000442081622794303, "grad_norm": 0.6547613739967346, "grad_norm_exchange": 0.1440497189760208, "grad_norm_vphi": 0.0028849344234913588, "gamma": 0.30000001192092896, "gumbel_tau": 0.9644559025764465, "xi_alphas": [2.1601597381959436e-06, 0.6090810894966125, 0.8868290185928345, 0.9768631458282471], "elapsed_s": 2943.9541664123535, "exchange_scale": -0.2125634253025055} {"step": 3900, "train_loss": 2.3819836521148683, "lr": 0.00044046069315455485, "grad_norm": 0.6614635586738586, "grad_norm_exchange": 0.21547667682170868, "grad_norm_vphi": 0.003317322349175811, "gamma": 0.30000001192092896, "gumbel_tau": 0.9617217183113098, "xi_alphas": [2.16870944313996e-06, 0.6079652309417725, 0.8862969875335693, 0.9767400622367859], "elapsed_s": 2981.3906762599945, "exchange_scale": -0.21416990458965302} {"step": 3950, "train_loss": 2.376177110671997, "lr": 0.0004388204530712959, "grad_norm": 0.623752236366272, "grad_norm_exchange": 0.13916562497615814, "grad_norm_vphi": 0.0028992299921810627, "gamma": 0.30000001192092896, "gumbel_tau": 0.9589875936508179, "xi_alphas": [2.178566091970424e-06, 0.6074586510658264, 0.8857297897338867, 0.9766716957092285], "elapsed_s": 3018.7118809223175, "exchange_scale": -0.21611350774765015} {"step": 4000, "train_loss": 2.357842082977295, "lr": 0.0004371610688453149, "grad_norm": 0.6413677334785461, "grad_norm_exchange": 0.15800204873085022, "grad_norm_vphi": 0.0027468998450785875, "gamma": 0.30000001192092896, "gumbel_tau": 0.9562534093856812, "xi_alphas": [2.1860128072148655e-06, 0.6069291234016418, 0.8852424621582031, 0.976571798324585], "elapsed_s": 3055.9950878620148, "exchange_scale": -0.2189801186323166} {"step": 4000, "val_loss": 2.519484043121338, "val_ppl": 12.422185697670056} {"step": 4050, "train_loss": 2.3554503440856935, "lr": 0.0004354827087183882, "grad_norm": 0.6304952502250671, "grad_norm_exchange": 0.16645605862140656, "grad_norm_vphi": 0.003709136275574565, "gamma": 0.30000001192092896, "gumbel_tau": 0.9535192847251892, "xi_alphas": [2.196197556258994e-06, 0.6065400838851929, 0.8848246335983276, 0.9764508008956909], "elapsed_s": 3099.4479336738586, "exchange_scale": -0.22070743143558502} {"step": 4100, "train_loss": 2.3574172353744505, "lr": 0.0004337855428562224, "grad_norm": 0.640358567237854, "grad_norm_exchange": 0.16209928691387177, "grad_norm_vphi": 0.0032236191909760237, "gamma": 0.30000001192092896, "gumbel_tau": 0.9507851004600525, "xi_alphas": [2.2041390366212e-06, 0.605715811252594, 0.8842684030532837, 0.9765111804008484], "elapsed_s": 3136.728583097458, "exchange_scale": -0.2224615067243576} {"step": 4150, "train_loss": 2.3415096044540404, "lr": 0.0004320697433312014, "grad_norm": 0.6415835022926331, "grad_norm_exchange": 0.14713814854621887, "grad_norm_vphi": 0.0029310081154108047, "gamma": 0.30000001192092896, "gumbel_tau": 0.9480509161949158, "xi_alphas": [2.205737473559566e-06, 0.6056197881698608, 0.8835699558258057, 0.9763479828834534], "elapsed_s": 3174.1780455112457, "exchange_scale": -0.2250729650259018} {"step": 4200, "train_loss": 2.340989909172058, "lr": 0.0004303354841049402, "grad_norm": 0.6153716444969177, "grad_norm_exchange": 0.15430498123168945, "grad_norm_vphi": 0.0032740526366978884, "gamma": 0.30000001192092896, "gumbel_tau": 0.9453167915344238, "xi_alphas": [2.212768094977946e-06, 0.6051507592201233, 0.8829449415206909, 0.9762794375419617], "elapsed_s": 3211.4896392822266, "exchange_scale": -0.2266118973493576} {"step": 4250, "train_loss": 2.3427305173873902, "lr": 0.0004285829410106478, "grad_norm": 0.6389195322990417, "grad_norm_exchange": 0.16694845259189606, "grad_norm_vphi": 0.003562244586646557, "gamma": 0.30000001192092896, "gumbel_tau": 0.9425826072692871, "xi_alphas": [2.2211340819922043e-06, 0.6046202778816223, 0.8823578953742981, 0.9762261509895325], "elapsed_s": 3248.952943086624, "exchange_scale": -0.22784335911273956} {"step": 4300, "train_loss": 2.341027102470398, "lr": 0.00042681229173529905, "grad_norm": 0.6415448188781738, "grad_norm_exchange": 0.17978931963443756, "grad_norm_vphi": 0.0033571727108210325, "gamma": 0.30000001192092896, "gumbel_tau": 0.9398484230041504, "xi_alphas": [2.2295446342468495e-06, 0.6044477224349976, 0.8820528388023376, 0.9761605262756348], "elapsed_s": 3286.255385875702, "exchange_scale": -0.22961905598640442} {"step": 4350, "train_loss": 2.336559467315674, "lr": 0.0004250237158016199, "grad_norm": 0.6049155592918396, "grad_norm_exchange": 0.14928479492664337, "grad_norm_vphi": 0.0037606179248541594, "gamma": 0.30000001192092896, "gumbel_tau": 0.9371142983436584, "xi_alphas": [2.2376093511411455e-06, 0.6034195423126221, 0.8814461827278137, 0.9760376214981079], "elapsed_s": 3323.6821036338806, "exchange_scale": -0.23098404705524445} {"step": 4400, "train_loss": 2.314175624847412, "lr": 0.00042321739454988564, "grad_norm": 0.6566189527511597, "grad_norm_exchange": 0.16686564683914185, "grad_norm_vphi": 0.004519280046224594, "gamma": 0.30000001192092896, "gumbel_tau": 0.9343801140785217, "xi_alphas": [2.2418107619159855e-06, 0.6030527353286743, 0.8807815909385681, 0.9759687781333923], "elapsed_s": 3360.938243627548, "exchange_scale": -0.23241358995437622} {"step": 4400, "val_loss": 2.4785025656223296, "val_ppl": 11.923396539323521} {"step": 4450, "train_loss": 2.3426364994049074, "lr": 0.00042139351111953515, "grad_norm": 0.6307303309440613, "grad_norm_exchange": 0.15287712216377258, "grad_norm_vphi": 0.0032662858720868826, "gamma": 0.30000001192092896, "gumbel_tau": 0.9316459894180298, "xi_alphas": [2.2392828213924076e-06, 0.6020554900169373, 0.8804094195365906, 0.9759105443954468], "elapsed_s": 3404.470363140106, "exchange_scale": -0.2340468019247055} {"step": 4500, "train_loss": 2.3104604959487913, "lr": 0.00041955225043060313, "grad_norm": 0.6799548268318176, "grad_norm_exchange": 0.18951427936553955, "grad_norm_vphi": 0.003970862366259098, "gamma": 0.30000001192092896, "gumbel_tau": 0.9289118051528931, "xi_alphas": [2.2438616724684834e-06, 0.601833164691925, 0.8800464272499084, 0.9758448004722595], "elapsed_s": 3441.727232694626, "exchange_scale": -0.23658810555934906} {"step": 4550, "train_loss": 2.295965566635132, "lr": 0.00041769379916497087, "grad_norm": 0.6397117972373962, "grad_norm_exchange": 0.20240886509418488, "grad_norm_vphi": 0.0036297773476690054, "gamma": 0.30000001192092896, "gumbel_tau": 0.9261776208877563, "xi_alphas": [2.2465189886133885e-06, 0.601377010345459, 0.8796066641807556, 0.9757196307182312], "elapsed_s": 3479.2045452594757, "exchange_scale": -0.23749084770679474} {"step": 4600, "train_loss": 2.294536943435669, "lr": 0.0004158183457474392, "grad_norm": 0.7084974050521851, "grad_norm_exchange": 0.1955181509256363, "grad_norm_vphi": 0.0056784069165587425, "gamma": 0.30000001192092896, "gumbel_tau": 0.9234434962272644, "xi_alphas": [2.255649405924487e-06, 0.6009387969970703, 0.8791852593421936, 0.9755949378013611], "elapsed_s": 3516.519773244858, "exchange_scale": -0.2389533817768097} {"step": 4650, "train_loss": 2.3005834579467774, "lr": 0.0004139260803266245, "grad_norm": 0.7159048318862915, "grad_norm_exchange": 0.24049358069896698, "grad_norm_vphi": 0.004477233625948429, "gamma": 0.30000001192092896, "gumbel_tau": 0.9207093119621277, "xi_alphas": [2.256004336231854e-06, 0.6011073589324951, 0.8786956667900085, 0.975496232509613], "elapsed_s": 3553.9675574302673, "exchange_scale": -0.23985619843006134} {"step": 4700, "train_loss": 2.290555934906006, "lr": 0.0004120171947556799, "grad_norm": 0.6200132966041565, "grad_norm_exchange": 0.16727156937122345, "grad_norm_vphi": 0.004086608998477459, "gamma": 0.30000001192092896, "gumbel_tau": 0.9179751873016357, "xi_alphas": [2.258981794511783e-06, 0.6005191206932068, 0.8781521320343018, 0.9754085540771484], "elapsed_s": 3591.283405303955, "exchange_scale": -0.24191057682037354} {"step": 4750, "train_loss": 2.267336993217468, "lr": 0.0004100918825728434, "grad_norm": 0.6636335253715515, "grad_norm_exchange": 0.19188319146633148, "grad_norm_vphi": 0.00427925493568182, "gamma": 0.30000001192092896, "gumbel_tau": 0.915241003036499, "xi_alphas": [2.265650664412533e-06, 0.6003021001815796, 0.8774450421333313, 0.9753161072731018], "elapsed_s": 3628.5402755737305, "exchange_scale": -0.24417230486869812} {"step": 4800, "train_loss": 2.2970441341400147, "lr": 0.0004081503389818158, "grad_norm": 0.61969393491745, "grad_norm_exchange": 0.15777745842933655, "grad_norm_vphi": 0.004248629789799452, "gamma": 0.30000001192092896, "gumbel_tau": 0.9125068187713623, "xi_alphas": [2.2688095668854658e-06, 0.5996894240379333, 0.8769859075546265, 0.9752630591392517], "elapsed_s": 3666.245157957077, "exchange_scale": -0.24481596052646637} {"step": 4800, "val_loss": 2.412916737794876, "val_ppl": 11.166483396727559} {"step": 4850, "train_loss": 2.255691237449646, "lr": 0.00040619276083196914, "grad_norm": 0.6836687326431274, "grad_norm_exchange": 0.18858444690704346, "grad_norm_vphi": 0.004420339595526457, "gamma": 0.30000001192092896, "gumbel_tau": 0.9097726941108704, "xi_alphas": [2.270625600431231e-06, 0.599392294883728, 0.876815140247345, 0.9752389788627625], "elapsed_s": 3710.0457525253296, "exchange_scale": -0.24608385562896729} {"step": 4900, "train_loss": 2.2575559759140016, "lr": 0.0004042193465983885, "grad_norm": 0.6424733996391296, "grad_norm_exchange": 0.17995096743106842, "grad_norm_vphi": 0.004698711913079023, "gamma": 0.30000001192092896, "gumbel_tau": 0.9070385098457336, "xi_alphas": [2.2758285922463983e-06, 0.5987332463264465, 0.8765550255775452, 0.9752011299133301], "elapsed_s": 3747.661484003067, "exchange_scale": -0.24783551692962646} {"step": 4950, "train_loss": 2.251234438419342, "lr": 0.000402230296361749, "grad_norm": 0.6461801528930664, "grad_norm_exchange": 0.1956123560667038, "grad_norm_vphi": 0.004067959263920784, "gamma": 0.30000001192092896, "gumbel_tau": 0.9043043255805969, "xi_alphas": [2.274303369631525e-06, 0.5984660983085632, 0.8760705590248108, 0.9751446843147278], "elapsed_s": 3785.0558638572693, "exchange_scale": -0.24830272793769836} {"step": 5000, "train_loss": 2.2629031896591187, "lr": 0.0004002258117880302, "grad_norm": 0.6868823170661926, "grad_norm_exchange": 0.2354251742362976, "grad_norm_vphi": 0.004520829766988754, "gamma": 0.30000001192092896, "gumbel_tau": 0.901570200920105, "xi_alphas": [2.282410150655778e-06, 0.5985332131385803, 0.8757113218307495, 0.9750789403915405], "elapsed_s": 3822.596029520035, "exchange_scale": -0.24939075112342834} {"step": 5050, "train_loss": 2.2411500358581544, "lr": 0.00039820609610806916, "grad_norm": 0.6348738670349121, "grad_norm_exchange": 0.17677144706249237, "grad_norm_vphi": 0.005630052648484707, "gamma": 0.30000001192092896, "gumbel_tau": 0.8988360166549683, "xi_alphas": [2.2849080778541975e-06, 0.5977869629859924, 0.8753580451011658, 0.9750030040740967], "elapsed_s": 3859.933066368103, "exchange_scale": -0.25008949637413025} {"step": 5100, "train_loss": 2.242102279663086, "lr": 0.0003961713540969559, "grad_norm": 0.6484432816505432, "grad_norm_exchange": 0.20939990878105164, "grad_norm_vphi": 0.004317075479775667, "gamma": 0.30000001192092896, "gumbel_tau": 0.8961018919944763, "xi_alphas": [2.2837405140307965e-06, 0.5971237421035767, 0.8748264908790588, 0.9748942255973816], "elapsed_s": 3897.4336817264557, "exchange_scale": -0.2516474723815918} {"step": 5150, "train_loss": 2.221005940437317, "lr": 0.00039412179205327094, "grad_norm": 0.6479594111442566, "grad_norm_exchange": 0.18873488903045654, "grad_norm_vphi": 0.008046077564358711, "gamma": 0.30000001192092896, "gumbel_tau": 0.8933677077293396, "xi_alphas": [2.29121656047937e-06, 0.5970373749732971, 0.8745507597923279, 0.9748647809028625], "elapsed_s": 3934.823454141617, "exchange_scale": -0.2525864243507385} {"step": 5200, "train_loss": 2.241099967956543, "lr": 0.0003920576177781695, "grad_norm": 0.6522452235221863, "grad_norm_exchange": 0.16459056735038757, "grad_norm_vphi": 0.005107297096401453, "gamma": 0.30000001192092896, "gumbel_tau": 0.8906335234642029, "xi_alphas": [2.2929584702069405e-06, 0.5972076654434204, 0.8743889927864075, 0.9748108386993408], "elapsed_s": 3972.300128698349, "exchange_scale": -0.25406649708747864} {"step": 5200, "val_loss": 2.425601089000702, "val_ppl": 11.309025107268958} {"step": 5250, "train_loss": 2.2180765151977537, "lr": 0.0003899790405543129, "grad_norm": 0.6682084202766418, "grad_norm_exchange": 0.1916840672492981, "grad_norm_vphi": 0.006433113012462854, "gamma": 0.30000001192092896, "gumbel_tau": 0.8878993988037109, "xi_alphas": [2.2938772872294066e-06, 0.5969717502593994, 0.8738095164299011, 0.9747034907341003], "elapsed_s": 4015.785270690918, "exchange_scale": -0.25448259711265564} {"step": 5300, "train_loss": 2.236254301071167, "lr": 0.00038788627112464974, "grad_norm": 0.6416710615158081, "grad_norm_exchange": 0.16928473114967346, "grad_norm_vphi": 0.006014837883412838, "gamma": 0.30000001192092896, "gumbel_tau": 0.8851652145385742, "xi_alphas": [2.3022753339319024e-06, 0.5963794589042664, 0.8736115097999573, 0.9746330380439758], "elapsed_s": 4053.301927804947, "exchange_scale": -0.25623035430908203} {"step": 5350, "train_loss": 2.212815113067627, "lr": 0.00038577952167104907, "grad_norm": 0.729274570941925, "grad_norm_exchange": 0.2136003077030182, "grad_norm_vphi": 0.006588790100067854, "gamma": 0.30000001192092896, "gumbel_tau": 0.8824310302734375, "xi_alphas": [2.3062091258907458e-06, 0.5960841178894043, 0.8734070062637329, 0.9745666980743408], "elapsed_s": 4091.0075783729553, "exchange_scale": -0.25721850991249084} {"step": 5400, "train_loss": 2.2381096982955935, "lr": 0.0003836590057927877, "grad_norm": 0.6851189136505127, "grad_norm_exchange": 0.20249930024147034, "grad_norm_vphi": 0.005855890456587076, "gamma": 0.30000001192092896, "gumbel_tau": 0.8796969056129456, "xi_alphas": [2.3092945866665104e-06, 0.5960162878036499, 0.8729841709136963, 0.9745252132415771], "elapsed_s": 4128.604075908661, "exchange_scale": -0.257936030626297} {"step": 5450, "train_loss": 2.236113328933716, "lr": 0.0003815249384848936, "grad_norm": 0.6375213861465454, "grad_norm_exchange": 0.1846950203180313, "grad_norm_vphi": 0.006214214954525232, "gamma": 0.30000001192092896, "gumbel_tau": 0.8769627213478088, "xi_alphas": [2.316365907972795e-06, 0.5957891941070557, 0.8728134036064148, 0.9745275974273682], "elapsed_s": 4166.186811923981, "exchange_scale": -0.25858771800994873} {"step": 5500, "train_loss": 2.2199003744125365, "lr": 0.0003793775361163485, "grad_norm": 0.6323482990264893, "grad_norm_exchange": 0.179234579205513, "grad_norm_vphi": 0.006777302827686071, "gamma": 0.30000001192092896, "gumbel_tau": 0.8742285966873169, "xi_alphas": [2.320763769603218e-06, 0.5954223275184631, 0.8725306987762451, 0.9745157361030579], "elapsed_s": 4203.680114507675, "exchange_scale": -0.25967055559158325} {"step": 5550, "train_loss": 2.1992787504196167, "lr": 0.00037721701640814996, "grad_norm": 0.665774405002594, "grad_norm_exchange": 0.17655231058597565, "grad_norm_vphi": 0.006752721965312958, "gamma": 0.30000001192092896, "gumbel_tau": 0.8714944124221802, "xi_alphas": [2.32629486163205e-06, 0.5950061082839966, 0.8721016645431519, 0.9744641184806824], "elapsed_s": 4241.286914587021, "exchange_scale": -0.26091983914375305} {"step": 5600, "train_loss": 2.1935381841659547, "lr": 0.0003750435984112376, "grad_norm": 0.6555315256118774, "grad_norm_exchange": 0.17988446354866028, "grad_norm_vphi": 0.005479964893311262, "gamma": 0.30000001192092896, "gumbel_tau": 0.8687602281570435, "xi_alphas": [2.3240374957822496e-06, 0.5952355265617371, 0.8717759251594543, 0.9744018316268921], "elapsed_s": 4279.126819849014, "exchange_scale": -0.26084136962890625} {"step": 5600, "val_loss": 2.3992829382419587, "val_ppl": 11.015274915664646} {"step": 5650, "train_loss": 2.2051538944244387, "lr": 0.0003728575024842835, "grad_norm": 0.6583641767501831, "grad_norm_exchange": 0.18404577672481537, "grad_norm_vphi": 0.005951684433966875, "gamma": 0.30000001192092896, "gumbel_tau": 0.8660261034965515, "xi_alphas": [2.3265208710654406e-06, 0.5949772596359253, 0.8711452484130859, 0.9743605256080627], "elapsed_s": 4323.602722406387, "exchange_scale": -0.26127344369888306} {"step": 5700, "train_loss": 2.2032509994506837, "lr": 0.0003706589502713509, "grad_norm": 0.7054215669631958, "grad_norm_exchange": 0.1855049580335617, "grad_norm_vphi": 0.00753336725756526, "gamma": 0.30000001192092896, "gumbel_tau": 0.8632919192314148, "xi_alphas": [2.32663865062932e-06, 0.5942375063896179, 0.8708745241165161, 0.9743066430091858], "elapsed_s": 4361.370906591415, "exchange_scale": -0.2621898353099823} {"step": 5750, "train_loss": 2.1979302263259886, "lr": 0.000368448164679422, "grad_norm": 0.65067458152771, "grad_norm_exchange": 0.18563251197338104, "grad_norm_vphi": 0.011235659010708332, "gamma": 0.30000001192092896, "gumbel_tau": 0.8605577945709229, "xi_alphas": [2.3221541596285533e-06, 0.5940338969230652, 0.8706340789794922, 0.9742172956466675], "elapsed_s": 4399.440932512283, "exchange_scale": -0.2635408341884613} {"step": 5800, "train_loss": 2.1689810514450074, "lr": 0.0003662253698557975, "grad_norm": 0.6406531929969788, "grad_norm_exchange": 0.1778310388326645, "grad_norm_vphi": 0.005246561020612717, "gamma": 0.30000001192092896, "gumbel_tau": 0.8578236103057861, "xi_alphas": [2.3255117866938235e-06, 0.5937809348106384, 0.8703000545501709, 0.9740709066390991], "elapsed_s": 4437.5906021595, "exchange_scale": -0.263945996761322} {"step": 5850, "train_loss": 2.161520788669586, "lr": 0.0003639907911653711, "grad_norm": 0.6432987451553345, "grad_norm_exchange": 0.18931324779987335, "grad_norm_vphi": 0.006778923328965902, "gamma": 0.30000001192092896, "gumbel_tau": 0.8550894260406494, "xi_alphas": [2.3266829884960316e-06, 0.5934986472129822, 0.8700626492500305, 0.9740556478500366], "elapsed_s": 4475.417424917221, "exchange_scale": -0.2645140290260315} {"step": 5900, "train_loss": 2.165541954040527, "lr": 0.0003617446551677803, "grad_norm": 0.6093708276748657, "grad_norm_exchange": 0.17276938259601593, "grad_norm_vphi": 0.005990162957459688, "gamma": 0.30000001192092896, "gumbel_tau": 0.8523553013801575, "xi_alphas": [2.3264879018825013e-06, 0.5933033227920532, 0.8698403239250183, 0.9740083813667297], "elapsed_s": 4513.018283367157, "exchange_scale": -0.26597124338150024} {"step": 5950, "train_loss": 2.174322347640991, "lr": 0.0003594871895944355, "grad_norm": 0.6813655495643616, "grad_norm_exchange": 0.1971380114555359, "grad_norm_vphi": 0.006967451423406601, "gamma": 0.30000001192092896, "gumbel_tau": 0.8496211171150208, "xi_alphas": [2.3291918296308722e-06, 0.5930265784263611, 0.869583249092102, 0.9739383459091187], "elapsed_s": 4550.847613811493, "exchange_scale": -0.26610854268074036} {"step": 6000, "train_loss": 2.165416131019592, "lr": 0.0003572186233254312, "grad_norm": 0.6810392737388611, "grad_norm_exchange": 0.18573501706123352, "grad_norm_vphi": 0.00672509428113699, "gamma": 0.30000001192092896, "gumbel_tau": 0.846886932849884, "xi_alphas": [2.330433744646143e-06, 0.5925760865211487, 0.8694472312927246, 0.9739056825637817], "elapsed_s": 4588.411830425262, "exchange_scale": -0.26655563712120056} {"step": 6000, "val_loss": 2.3699859380722046, "val_ppl": 10.697241859210788} {"step": 6050, "train_loss": 2.1705628538131716, "lr": 0.00035493918636634003, "grad_norm": 0.6321671605110168, "grad_norm_exchange": 0.18672305345535278, "grad_norm_vphi": 0.007929923944175243, "gamma": 0.30000001192092896, "gumbel_tau": 0.8441528081893921, "xi_alphas": [2.3326172140514245e-06, 0.5921626687049866, 0.8692097067832947, 0.9738780856132507], "elapsed_s": 4632.320437431335, "exchange_scale": -0.2672691345214844} {"step": 6100, "train_loss": 2.1568753838539125, "lr": 0.00035264910982489285, "grad_norm": 0.6522237658500671, "grad_norm_exchange": 0.19472479820251465, "grad_norm_vphi": 0.00815716665238142, "gamma": 0.30000001192092896, "gumbel_tau": 0.8414186239242554, "xi_alphas": [2.3385823624266777e-06, 0.5921240448951721, 0.8689383864402771, 0.9737798571586609], "elapsed_s": 4669.898848295212, "exchange_scale": -0.2683521509170532} {"step": 6150, "train_loss": 2.1515807723999023, "lr": 0.0003503486258875475, "grad_norm": 0.6653701663017273, "grad_norm_exchange": 0.20075564086437225, "grad_norm_vphi": 0.007124514319002628, "gamma": 0.30000001192092896, "gumbel_tau": 0.8386844992637634, "xi_alphas": [2.340038918191567e-06, 0.5922013521194458, 0.8685249090194702, 0.9737701416015625], "elapsed_s": 4707.697709083557, "exchange_scale": -0.2690650522708893} {"step": 6200, "train_loss": 2.145143747329712, "lr": 0.00034803796779594754, "grad_norm": 0.6704351305961609, "grad_norm_exchange": 0.18717843294143677, "grad_norm_vphi": 0.006703866180032492, "gamma": 0.30000001192092896, "gumbel_tau": 0.8359503149986267, "xi_alphas": [2.3418003820552258e-06, 0.5918756723403931, 0.8683827519416809, 0.9737140536308289], "elapsed_s": 4745.217702388763, "exchange_scale": -0.2697013020515442} {"step": 6250, "train_loss": 2.126769676208496, "lr": 0.00034571736982327476, "grad_norm": 0.6420971155166626, "grad_norm_exchange": 0.17653122544288635, "grad_norm_vphi": 0.0067277043126523495, "gamma": 0.30000001192092896, "gumbel_tau": 0.83321613073349, "xi_alphas": [2.3471575332223438e-06, 0.5920233726501465, 0.8680567741394043, 0.9736284017562866], "elapsed_s": 4782.862002372742, "exchange_scale": -0.27098581194877625} {"step": 6300, "train_loss": 2.142837116718292, "lr": 0.0003433870672504959, "grad_norm": 0.6804458498954773, "grad_norm_exchange": 0.2161620855331421, "grad_norm_vphi": 0.006368003785610199, "gamma": 0.30000001192092896, "gumbel_tau": 0.830482006072998, "xi_alphas": [2.3427990072377725e-06, 0.591507077217102, 0.8676604628562927, 0.9735735058784485], "elapsed_s": 4820.538534641266, "exchange_scale": -0.27073463797569275} {"step": 6350, "train_loss": 2.1391812801361083, "lr": 0.0003410472963425089, "grad_norm": 0.6449630260467529, "grad_norm_exchange": 0.18463100492954254, "grad_norm_vphi": 0.007637595292180777, "gamma": 0.30000001192092896, "gumbel_tau": 0.8277478218078613, "xi_alphas": [2.3482791675633052e-06, 0.5912232995033264, 0.867558479309082, 0.9734717011451721], "elapsed_s": 4858.010662317276, "exchange_scale": -0.2714729607105255} {"step": 6400, "train_loss": 2.150410828590393, "lr": 0.00033869829432418776, "grad_norm": 0.6549612283706665, "grad_norm_exchange": 0.19480010867118835, "grad_norm_vphi": 0.007359141018241644, "gamma": 0.30000001192092896, "gumbel_tau": 0.8250136971473694, "xi_alphas": [2.349152737224358e-06, 0.5914024710655212, 0.8676968216896057, 0.9734961986541748], "elapsed_s": 4895.742268562317, "exchange_scale": -0.27296435832977295} {"step": 6400, "val_loss": 2.346145761013031, "val_ppl": 10.445233615529537} {"step": 6450, "train_loss": 2.1365113043785096, "lr": 0.000336340299356331, "grad_norm": 0.6629350185394287, "grad_norm_exchange": 0.23043414950370789, "grad_norm_vphi": 0.009197666309773922, "gamma": 0.30000001192092896, "gumbel_tau": 0.8222795128822327, "xi_alphas": [2.349296210013563e-06, 0.5912691354751587, 0.8674625754356384, 0.9734731316566467], "elapsed_s": 4939.430516242981, "exchange_scale": -0.27345046401023865} {"step": 6500, "train_loss": 2.124113919734955, "lr": 0.00033397355051151517, "grad_norm": 0.650231122970581, "grad_norm_exchange": 0.19385774433612823, "grad_norm_vphi": 0.0066041117534041405, "gamma": 0.30000001192092896, "gumbel_tau": 0.819545328617096, "xi_alphas": [2.3572406462335493e-06, 0.590772807598114, 0.8670341968536377, 0.9734676480293274], "elapsed_s": 4977.166532039642, "exchange_scale": -0.2738844156265259} {"step": 6550, "train_loss": 2.1412638521194456, "lr": 0.00033159828774985547, "grad_norm": 0.7502943277359009, "grad_norm_exchange": 0.2190323919057846, "grad_norm_vphi": 0.006903390865772963, "gamma": 0.30000001192092896, "gumbel_tau": 0.816811203956604, "xi_alphas": [2.3545289877802134e-06, 0.5905030369758606, 0.866986870765686, 0.9734312295913696], "elapsed_s": 5014.611162662506, "exchange_scale": -0.2742122709751129} {"step": 6600, "train_loss": 2.1288144254684447, "lr": 0.0003292147518946767, "grad_norm": 0.6855027079582214, "grad_norm_exchange": 0.21006357669830322, "grad_norm_vphi": 0.006346848327666521, "gamma": 0.30000001192092896, "gumbel_tau": 0.8140770196914673, "xi_alphas": [2.358452775297337e-06, 0.5906682014465332, 0.86662757396698, 0.9733773469924927], "elapsed_s": 5052.240911245346, "exchange_scale": -0.2742488980293274} {"step": 6650, "train_loss": 2.10924211025238, "lr": 0.0003268231846080969, "grad_norm": 0.6690341234207153, "grad_norm_exchange": 0.193440243601799, "grad_norm_vphi": 0.006413344759494066, "gamma": 0.30000001192092896, "gumbel_tau": 0.8113428354263306, "xi_alphas": [2.3596026039740536e-06, 0.5907105207443237, 0.8666331768035889, 0.9733047485351562], "elapsed_s": 5089.681811571121, "exchange_scale": -0.2749035954475403} {"step": 6700, "train_loss": 2.1105611753463744, "lr": 0.0003244238283665252, "grad_norm": 0.6515729427337646, "grad_norm_exchange": 0.20384755730628967, "grad_norm_vphi": 0.006714850198477507, "gamma": 0.30000001192092896, "gumbel_tau": 0.8086087107658386, "xi_alphas": [2.360061444051098e-06, 0.5903661847114563, 0.8665302395820618, 0.9733245968818665], "elapsed_s": 5127.332571983337, "exchange_scale": -0.2756611108779907} {"step": 6750, "train_loss": 2.1302819561958315, "lr": 0.00032201692643607814, "grad_norm": 0.6717351078987122, "grad_norm_exchange": 0.18951229751110077, "grad_norm_vphi": 0.007435966748744249, "gamma": 0.30000001192092896, "gumbel_tau": 0.8058745265007019, "xi_alphas": [2.362356099183671e-06, 0.5901247262954712, 0.8661995530128479, 0.9732940196990967], "elapsed_s": 5164.870180845261, "exchange_scale": -0.2758517563343048} {"step": 6800, "train_loss": 2.1106344079971313, "lr": 0.00031960272284791525, "grad_norm": 0.6660448312759399, "grad_norm_exchange": 0.17841924726963043, "grad_norm_vphi": 0.0066132270731031895, "gamma": 0.30000001192092896, "gumbel_tau": 0.80314040184021, "xi_alphas": [2.3656862140342128e-06, 0.5895753502845764, 0.8659015893936157, 0.9732084274291992], "elapsed_s": 5202.7504777908325, "exchange_scale": -0.27658548951148987} {"step": 6800, "val_loss": 2.3429783642292024, "val_ppl": 10.41220175628747} {"step": 6850, "train_loss": 2.1299562501907348, "lr": 0.00031718146237349673, "grad_norm": 0.6604442000389099, "grad_norm_exchange": 0.18793603777885437, "grad_norm_vphi": 0.00673700263723731, "gamma": 0.30000001192092896, "gumbel_tau": 0.8004062175750732, "xi_alphas": [2.3701820737187518e-06, 0.5896011590957642, 0.8658062219619751, 0.9731870889663696], "elapsed_s": 5246.597423315048, "exchange_scale": -0.2773466408252716} {"step": 6900, "train_loss": 2.139301462173462, "lr": 0.00031475339049976715, "grad_norm": 0.6956095695495605, "grad_norm_exchange": 0.22272032499313354, "grad_norm_vphi": 0.006890607997775078, "gamma": 0.30000001192092896, "gumbel_tau": 0.7976720333099365, "xi_alphas": [2.3687111934123095e-06, 0.5895618796348572, 0.8657318949699402, 0.9731388688087463], "elapsed_s": 5284.4603180885315, "exchange_scale": -0.27740418910980225} {"step": 6950, "train_loss": 2.123184869289398, "lr": 0.0003123187534042657, "grad_norm": 0.6777981519699097, "grad_norm_exchange": 0.1986270397901535, "grad_norm_vphi": 0.0071557313203811646, "gamma": 0.30000001192092896, "gumbel_tau": 0.7949379086494446, "xi_alphas": [2.3685327050770866e-06, 0.5891907811164856, 0.8655668497085571, 0.97307950258255], "elapsed_s": 5322.1885368824005, "exchange_scale": -0.2782149314880371} {"step": 7000, "train_loss": 2.1145767402648925, "lr": 0.00030987779793016694, "grad_norm": 0.681355357170105, "grad_norm_exchange": 0.20024794340133667, "grad_norm_vphi": 0.007430665194988251, "gamma": 0.30000001192092896, "gumbel_tau": 0.7922037243843079, "xi_alphas": [2.366205080761574e-06, 0.5887132287025452, 0.8654649257659912, 0.973055899143219], "elapsed_s": 5360.008960723877, "exchange_scale": -0.2785715162754059} {"step": 7050, "train_loss": 2.120422501564026, "lr": 0.00030743077156125345, "grad_norm": 0.6638234853744507, "grad_norm_exchange": 0.19148561358451843, "grad_norm_vphi": 0.0068492102436721325, "gamma": 0.30000001192092896, "gumbel_tau": 0.7894695997238159, "xi_alphas": [2.360640110055101e-06, 0.5882075428962708, 0.8651730418205261, 0.9730545878410339], "elapsed_s": 5397.708078384399, "exchange_scale": -0.2786548137664795} {"step": 7100, "train_loss": 2.1105499625205995, "lr": 0.0003049779223968243, "grad_norm": 0.6706815958023071, "grad_norm_exchange": 0.19424700736999512, "grad_norm_vphi": 0.006807568483054638, "gamma": 0.30000001192092896, "gumbel_tau": 0.7867354154586792, "xi_alphas": [2.3631109797861427e-06, 0.5885413289070129, 0.8649470210075378, 0.9730703830718994], "elapsed_s": 5435.5108008384705, "exchange_scale": -0.2794099450111389} {"step": 7150, "train_loss": 2.0891085982322695, "lr": 0.00030251949912654056, "grad_norm": 0.6739246845245361, "grad_norm_exchange": 0.22935311496257782, "grad_norm_vphi": 0.006848398596048355, "gamma": 0.30000001192092896, "gumbel_tau": 0.7840012311935425, "xi_alphas": [2.3650884486414725e-06, 0.5887273550033569, 0.8649982810020447, 0.9730481505393982], "elapsed_s": 5473.212294101715, "exchange_scale": -0.27997761964797974} {"step": 7200, "train_loss": 2.116494779586792, "lr": 0.00030005575100521117, "grad_norm": 0.6675117015838623, "grad_norm_exchange": 0.20062100887298584, "grad_norm_vphi": 0.008989613503217697, "gamma": 0.30000001192092896, "gumbel_tau": 0.7812671065330505, "xi_alphas": [2.3658169538975926e-06, 0.5888388752937317, 0.8649299740791321, 0.9729328155517578], "elapsed_s": 5510.938129901886, "exchange_scale": -0.2800743877887726} {"step": 7200, "val_loss": 2.329157418012619, "val_ppl": 10.269285171999424} {"step": 7250, "train_loss": 2.1035180139541625, "lr": 0.00029758692782752154, "grad_norm": 0.6853606104850769, "grad_norm_exchange": 0.20747917890548706, "grad_norm_vphi": 0.012543397024273872, "gamma": 0.30000001192092896, "gumbel_tau": 0.7785329222679138, "xi_alphas": [2.3659183625568403e-06, 0.5879847407341003, 0.86497962474823, 0.9728735685348511], "elapsed_s": 5554.916660070419, "exchange_scale": -0.2811262011528015} {"step": 7300, "train_loss": 2.0770190811157225, "lr": 0.00029511327990270725, "grad_norm": 0.6478946208953857, "grad_norm_exchange": 0.19171907007694244, "grad_norm_vphi": 0.008099878206849098, "gamma": 0.30000001192092896, "gumbel_tau": 0.7757987380027771, "xi_alphas": [2.3638572201889474e-06, 0.5880988836288452, 0.8650956749916077, 0.972800076007843], "elapsed_s": 5592.5702958106995, "exchange_scale": -0.28131428360939026} {"step": 7350, "train_loss": 2.0628980946540834, "lr": 0.00029263505802917567, "grad_norm": 0.6469804644584656, "grad_norm_exchange": 0.19810451567173004, "grad_norm_vphi": 0.006910902447998524, "gamma": 0.30000001192092896, "gumbel_tau": 0.7730646133422852, "xi_alphas": [2.3688398869126104e-06, 0.5876578092575073, 0.8647469878196716, 0.9727658033370972], "elapsed_s": 5630.26127743721, "exchange_scale": -0.28265413641929626} {"step": 7400, "train_loss": 2.0890888667106626, "lr": 0.00029015251346907803, "grad_norm": 0.689546525478363, "grad_norm_exchange": 0.21396762132644653, "grad_norm_vphi": 0.007546978536993265, "gamma": 0.30000001192092896, "gumbel_tau": 0.7703304290771484, "xi_alphas": [2.3660968508920632e-06, 0.587315559387207, 0.8645201921463013, 0.9727293848991394], "elapsed_s": 5667.7669496536255, "exchange_scale": -0.2833326756954193} {"step": 7450, "train_loss": 2.0641389894485473, "lr": 0.0002876658979228346, "grad_norm": 0.6781651973724365, "grad_norm_exchange": 0.19404183328151703, "grad_norm_vphi": 0.007543710991740227, "gamma": 0.30000001192092896, "gumbel_tau": 0.7675963044166565, "xi_alphas": [2.364017063882784e-06, 0.5870693325996399, 0.8644504547119141, 0.9726928472518921], "elapsed_s": 5705.439056634903, "exchange_scale": -0.2833763062953949} {"step": 7500, "train_loss": 2.059312598705292, "lr": 0.00028517546350361497, "grad_norm": 0.698283314704895, "grad_norm_exchange": 0.20529630780220032, "grad_norm_vphi": 0.00786793977022171, "gamma": 0.30000001192092896, "gumbel_tau": 0.7648621201515198, "xi_alphas": [2.3648515252716606e-06, 0.5866138935089111, 0.8642129898071289, 0.9727211594581604], "elapsed_s": 5742.909404039383, "exchange_scale": -0.283267080783844} {"step": 7550, "train_loss": 2.0768220591545106, "lr": 0.0002826814627117769, "grad_norm": 0.7351411581039429, "grad_norm_exchange": 0.23381531238555908, "grad_norm_vphi": 0.00862140767276287, "gamma": 0.30000001192092896, "gumbel_tau": 0.7621279358863831, "xi_alphas": [2.366191665714723e-06, 0.5865315794944763, 0.8640816807746887, 0.9727025032043457], "elapsed_s": 5780.6671051979065, "exchange_scale": -0.284047931432724} {"step": 7600, "train_loss": 2.076631734371185, "lr": 0.00028018414840926594, "grad_norm": 0.6954320073127747, "grad_norm_exchange": 0.20797352492809296, "grad_norm_vphi": 0.008215363137423992, "gamma": 0.30000001192092896, "gumbel_tau": 0.7593938112258911, "xi_alphas": [2.3616196358489105e-06, 0.5866837501525879, 0.8640627264976501, 0.9726777076721191], "elapsed_s": 5818.274399995804, "exchange_scale": -0.2851586639881134} {"step": 7600, "val_loss": 2.2903176724910734, "val_ppl": 9.878075175546998} {"step": 7650, "train_loss": 2.0507375836372375, "lr": 0.0002776837737939779, "grad_norm": 0.6844124794006348, "grad_norm_exchange": 0.19330677390098572, "grad_norm_vphi": 0.007359675131738186, "gamma": 0.30000001192092896, "gumbel_tau": 0.7566596269607544, "xi_alphas": [2.361991164434585e-06, 0.5862722396850586, 0.8638953566551208, 0.9726606607437134], "elapsed_s": 5862.1207411289215, "exchange_scale": -0.2845436632633209} {"step": 7700, "train_loss": 2.077149798870087, "lr": 0.00027518059237408813, "grad_norm": 0.6888574957847595, "grad_norm_exchange": 0.20178934931755066, "grad_norm_vphi": 0.009097537025809288, "gamma": 0.30000001192092896, "gumbel_tau": 0.7539255023002625, "xi_alphas": [2.3613606572325807e-06, 0.5856891870498657, 0.8637831807136536, 0.9726423025131226], "elapsed_s": 5899.701120138168, "exchange_scale": -0.2848764955997467} {"step": 7750, "train_loss": 2.0583940720558167, "lr": 0.0002726748579423481, "grad_norm": 0.6899640560150146, "grad_norm_exchange": 0.19740407168865204, "grad_norm_vphi": 0.008233906701207161, "gamma": 0.30000001192092896, "gumbel_tau": 0.7511913180351257, "xi_alphas": [2.3591614990436938e-06, 0.5858553647994995, 0.863776683807373, 0.9726013541221619], "elapsed_s": 5937.468163013458, "exchange_scale": -0.2855445146560669} {"step": 7800, "train_loss": 2.069489395618439, "lr": 0.0002701668245503547, "grad_norm": 0.685694694519043, "grad_norm_exchange": 0.20650643110275269, "grad_norm_vphi": 0.007602003403007984, "gamma": 0.30000001192092896, "gumbel_tau": 0.748457133769989, "xi_alphas": [2.3623338165634777e-06, 0.5863195657730103, 0.8637571930885315, 0.9725944995880127], "elapsed_s": 5975.074733257294, "exchange_scale": -0.28617218136787415} {"step": 7850, "train_loss": 2.084970588684082, "lr": 0.0002676567464827917, "grad_norm": 0.7231524586677551, "grad_norm_exchange": 0.22798602283000946, "grad_norm_vphi": 0.007600743323564529, "gamma": 0.30000001192092896, "gumbel_tau": 0.7457230091094971, "xi_alphas": [2.3619463718205225e-06, 0.5859630107879639, 0.8637694120407104, 0.972574770450592], "elapsed_s": 6012.736558198929, "exchange_scale": -0.2866959273815155} {"step": 7900, "train_loss": 2.0436726212501526, "lr": 0.0002651448782316485, "grad_norm": 0.6979507803916931, "grad_norm_exchange": 0.1995808333158493, "grad_norm_vphi": 0.007712284103035927, "gamma": 0.30000001192092896, "gumbel_tau": 0.7429888248443604, "xi_alphas": [2.363751264056191e-06, 0.5859810709953308, 0.863519549369812, 0.9725030660629272], "elapsed_s": 6050.240439176559, "exchange_scale": -0.2871428430080414} {"step": 7950, "train_loss": 2.0676570463180544, "lr": 0.0002626314744704181, "grad_norm": 0.7179150581359863, "grad_norm_exchange": 0.20498022437095642, "grad_norm_vphi": 0.008517836220562458, "gamma": 0.30000001192092896, "gumbel_tau": 0.7402546405792236, "xi_alphas": [2.3648426576983184e-06, 0.5859748125076294, 0.8635762929916382, 0.9724652767181396], "elapsed_s": 6087.895434856415, "exchange_scale": -0.2866741120815277} {"step": 8000, "train_loss": 2.046935651302338, "lr": 0.0002601167900282756, "grad_norm": 0.69687819480896, "grad_norm_exchange": 0.21790006756782532, "grad_norm_vphi": 0.009695321321487427, "gamma": 0.30000001192092896, "gumbel_tau": 0.7375205159187317, "xi_alphas": [2.361761517022387e-06, 0.5854406356811523, 0.8635093569755554, 0.9724999070167542], "elapsed_s": 6125.383004426956, "exchange_scale": -0.28653454780578613} {"step": 8000, "val_loss": 2.3049599647521974, "val_ppl": 10.023776939997974} {"step": 8050, "train_loss": 2.0638915634155275, "lr": 0.0002576010798642421, "grad_norm": 0.6629579663276672, "grad_norm_exchange": 0.20308199524879456, "grad_norm_vphi": 0.00757538340985775, "gamma": 0.30000001192092896, "gumbel_tau": 0.734786331653595, "xi_alphas": [2.361642373216455e-06, 0.5854941606521606, 0.8633810877799988, 0.972499430179596], "elapsed_s": 6168.94889831543, "exchange_scale": -0.28733059763908386} {"step": 8100, "train_loss": 2.0353920817375184, "lr": 0.00025508459904133453, "grad_norm": 0.6711968183517456, "grad_norm_exchange": 0.1841759830713272, "grad_norm_vphi": 0.007655059453099966, "gamma": 0.30000001192092896, "gumbel_tau": 0.732052206993103, "xi_alphas": [2.3576747025799705e-06, 0.5853539705276489, 0.8631907105445862, 0.9724506139755249], "elapsed_s": 6206.548418521881, "exchange_scale": -0.28758564591407776} {"step": 8150, "train_loss": 2.0406621408462526, "lr": 0.00025256760270070546, "grad_norm": 0.6811435222625732, "grad_norm_exchange": 0.1936260312795639, "grad_norm_vphi": 0.007811093702912331, "gamma": 0.30000001192092896, "gumbel_tau": 0.7293180227279663, "xi_alphas": [2.3591546778334305e-06, 0.5847454071044922, 0.8629579544067383, 0.9724368453025818], "elapsed_s": 6243.9785697460175, "exchange_scale": -0.2880033552646637} {"step": 8200, "train_loss": 2.043025162220001, "lr": 0.00025005034603577494, "grad_norm": 0.722499430179596, "grad_norm_exchange": 0.20006249845027924, "grad_norm_vphi": 0.00931552704423666, "gamma": 0.30000001192092896, "gumbel_tau": 0.7265838384628296, "xi_alphas": [2.3582661015097983e-06, 0.5847421288490295, 0.8629055619239807, 0.972387969493866], "elapsed_s": 6281.621560335159, "exchange_scale": -0.28775009512901306} {"step": 8250, "train_loss": 2.033749520778656, "lr": 0.00024753308426635656, "grad_norm": 0.6822290420532227, "grad_norm_exchange": 0.19960442185401917, "grad_norm_vphi": 0.010221024043858051, "gamma": 0.30000001192092896, "gumbel_tau": 0.7238497138023376, "xi_alphas": [2.3564700768474722e-06, 0.584711492061615, 0.8626986145973206, 0.9723415970802307], "elapsed_s": 6319.122237443924, "exchange_scale": -0.2878040075302124} {"step": 8300, "train_loss": 2.0238655638694762, "lr": 0.00024501607261278175, "grad_norm": 0.6781486868858337, "grad_norm_exchange": 0.19965489208698273, "grad_norm_vphi": 0.007720953319221735, "gamma": 0.30000001192092896, "gumbel_tau": 0.7211155295372009, "xi_alphas": [2.354760226808139e-06, 0.5847037434577942, 0.8625202775001526, 0.9723325967788696], "elapsed_s": 6356.813641786575, "exchange_scale": -0.2890428304672241} {"step": 8350, "train_loss": 2.0367105293273924, "lr": 0.00024249956627002292, "grad_norm": 0.6860981583595276, "grad_norm_exchange": 0.219077005982399, "grad_norm_vphi": 0.008519696071743965, "gamma": 0.30000001192092896, "gumbel_tau": 0.718381404876709, "xi_alphas": [2.3549960133095738e-06, 0.5848673582077026, 0.8624910116195679, 0.972322940826416], "elapsed_s": 6394.344164848328, "exchange_scale": -0.2887413501739502} {"step": 8400, "train_loss": 2.057199251651764, "lr": 0.0002399838203818201, "grad_norm": 0.6771015524864197, "grad_norm_exchange": 0.1910424530506134, "grad_norm_vphi": 0.008197457529604435, "gamma": 0.30000001192092896, "gumbel_tau": 0.7156472206115723, "xi_alphas": [2.354300022489042e-06, 0.5844197273254395, 0.8623120784759521, 0.9722716212272644], "elapsed_s": 6432.123201847076, "exchange_scale": -0.2896411716938019} {"step": 8400, "val_loss": 2.2777896344661714, "val_ppl": 9.755094237570335} {"step": 8450, "train_loss": 2.015602569580078, "lr": 0.00023746909001481225, "grad_norm": 0.6817330718040466, "grad_norm_exchange": 0.20476126670837402, "grad_norm_vphi": 0.007827460765838623, "gamma": 0.30000001192092896, "gumbel_tau": 0.7129130363464355, "xi_alphas": [2.349558371861349e-06, 0.5842601656913757, 0.8621406555175781, 0.9722336530685425], "elapsed_s": 6476.173464536667, "exchange_scale": -0.29024234414100647} {"step": 8500, "train_loss": 2.033328266143799, "lr": 0.00023495563013267666, "grad_norm": 0.7036839127540588, "grad_norm_exchange": 0.22499558329582214, "grad_norm_vphi": 0.010613595135509968, "gamma": 0.30000001192092896, "gumbel_tau": 0.7101789116859436, "xi_alphas": [2.347305326111382e-06, 0.5841830372810364, 0.86199951171875, 0.9721818566322327], "elapsed_s": 6513.832312345505, "exchange_scale": -0.2910557687282562} {"step": 8550, "train_loss": 2.018653075695038, "lr": 0.0002324436955702785, "grad_norm": 0.6709666848182678, "grad_norm_exchange": 0.19625425338745117, "grad_norm_vphi": 0.009138090535998344, "gamma": 0.30000001192092896, "gumbel_tau": 0.7074447274208069, "xi_alphas": [2.345152779525961e-06, 0.5837497711181641, 0.8619978427886963, 0.9721890687942505], "elapsed_s": 6551.341810703278, "exchange_scale": -0.29116353392601013} {"step": 8600, "train_loss": 2.011982033252716, "lr": 0.00022993354100783378, "grad_norm": 0.6991732120513916, "grad_norm_exchange": 0.20397457480430603, "grad_norm_vphi": 0.009880993515253067, "gamma": 0.30000001192092896, "gumbel_tau": 0.7047105431556702, "xi_alphas": [2.34566277867998e-06, 0.5840273499488831, 0.8620229363441467, 0.9721388220787048], "elapsed_s": 6588.893607139587, "exchange_scale": -0.2921959161758423} {"step": 8650, "train_loss": 2.011818253993988, "lr": 0.00022742542094508772, "grad_norm": 0.7487470507621765, "grad_norm_exchange": 0.25323012471199036, "grad_norm_vphi": 0.008258476853370667, "gamma": 0.30000001192092896, "gumbel_tau": 0.7019764184951782, "xi_alphas": [2.344973836443387e-06, 0.5839269161224365, 0.8620256781578064, 0.9721043109893799], "elapsed_s": 6626.333908557892, "exchange_scale": -0.2922898828983307} {"step": 8700, "train_loss": 2.009059898853302, "lr": 0.0002249195896755115, "grad_norm": 0.6844146847724915, "grad_norm_exchange": 0.1974182277917862, "grad_norm_vphi": 0.008929332718253136, "gamma": 0.30000001192092896, "gumbel_tau": 0.6992422342300415, "xi_alphas": [2.350652039240231e-06, 0.5839521288871765, 0.8620216846466064, 0.9721349477767944], "elapsed_s": 6664.082893371582, "exchange_scale": -0.2925584316253662} {"step": 8750, "train_loss": 2.0356809258461, "lr": 0.00022241630126051991, "grad_norm": 0.6789827346801758, "grad_norm_exchange": 0.19397707283496857, "grad_norm_vphi": 0.008808151818811893, "gamma": 0.30000001192092896, "gumbel_tau": 0.6965081095695496, "xi_alphas": [2.349596343265148e-06, 0.5836429595947266, 0.8618916869163513, 0.9721170663833618], "elapsed_s": 6701.639944553375, "exchange_scale": -0.2936629354953766} {"step": 8800, "train_loss": 1.999792022705078, "lr": 0.0002199158095037129, "grad_norm": 0.6871477961540222, "grad_norm_exchange": 0.19451496005058289, "grad_norm_vphi": 0.00750472629442811, "gamma": 0.30000001192092896, "gumbel_tau": 0.6937739253044128, "xi_alphas": [2.348574980715057e-06, 0.583530068397522, 0.8616505265235901, 0.9721107482910156], "elapsed_s": 6739.141548871994, "exchange_scale": -0.29312795400619507} {"step": 8800, "val_loss": 2.2894356489181518, "val_ppl": 9.869366321658884} {"step": 8850, "train_loss": 2.027001919746399, "lr": 0.00021741836792514224, "grad_norm": 0.6910943388938904, "grad_norm_exchange": 0.2071821093559265, "grad_norm_vphi": 0.009052886627614498, "gamma": 0.30000001192092896, "gumbel_tau": 0.6910397410392761, "xi_alphas": [2.3491281808674103e-06, 0.5834749937057495, 0.8616299629211426, 0.9721187353134155], "elapsed_s": 6782.84765791893, "exchange_scale": -0.29398852586746216} {"step": 8900, "train_loss": 1.9984537982940673, "lr": 0.00021492422973560819, "grad_norm": 0.6910784244537354, "grad_norm_exchange": 0.2080104649066925, "grad_norm_vphi": 0.009504054673016071, "gamma": 0.30000001192092896, "gumbel_tau": 0.6883056163787842, "xi_alphas": [2.3502373096562224e-06, 0.5831965208053589, 0.8615476489067078, 0.9720994830131531], "elapsed_s": 6820.410843133926, "exchange_scale": -0.2942936420440674} {"step": 8950, "train_loss": 2.0140367674827577, "lr": 0.00021243364781098667, "grad_norm": 0.6927916407585144, "grad_norm_exchange": 0.19921447336673737, "grad_norm_vphi": 0.009138676337897778, "gamma": 0.30000001192092896, "gumbel_tau": 0.6855714321136475, "xi_alphas": [2.350513113924535e-06, 0.5826552510261536, 0.8616246581077576, 0.9720297455787659], "elapsed_s": 6858.093241453171, "exchange_scale": -0.29435116052627563} {"step": 9000, "train_loss": 2.019037699699402, "lr": 0.00020994687466659078, "grad_norm": 0.726657509803772, "grad_norm_exchange": 0.23434622585773468, "grad_norm_vphi": 0.009210821241140366, "gamma": 0.30000001192092896, "gumbel_tau": 0.6828373074531555, "xi_alphas": [2.3511340714321705e-06, 0.5827865600585938, 0.8614493608474731, 0.9719988703727722], "elapsed_s": 6895.64342713356, "exchange_scale": -0.2945821285247803} {"step": 9050, "train_loss": 2.000070707798004, "lr": 0.0002074641624315688, "grad_norm": 0.746423602104187, "grad_norm_exchange": 0.24090732634067535, "grad_norm_vphi": 0.009359815157949924, "gamma": 0.30000001192092896, "gumbel_tau": 0.6801031231880188, "xi_alphas": [2.3491281808674103e-06, 0.5825295448303223, 0.8614017963409424, 0.9719852805137634], "elapsed_s": 6933.280478477478, "exchange_scale": -0.2949340343475342} {"step": 9100, "train_loss": 1.9805985236167907, "lr": 0.000204985762823341, "grad_norm": 0.700850784778595, "grad_norm_exchange": 0.2000301331281662, "grad_norm_vphi": 0.009885111823678017, "gamma": 0.30000001192092896, "gumbel_tau": 0.6773689389228821, "xi_alphas": [2.346790324736503e-06, 0.5821589827537537, 0.8612334132194519, 0.9719459414482117], "elapsed_s": 6970.767688512802, "exchange_scale": -0.29466795921325684} {"step": 9150, "train_loss": 1.9694611978530885, "lr": 0.0002025119271220789, "grad_norm": 0.7187510132789612, "grad_norm_exchange": 0.22142325341701508, "grad_norm_vphi": 0.00963281374424696, "gamma": 0.30000001192092896, "gumbel_tau": 0.6746348142623901, "xi_alphas": [2.350235035919468e-06, 0.5822198390960693, 0.8612868189811707, 0.9719504714012146], "elapsed_s": 7008.382661581039, "exchange_scale": -0.2959262728691101} {"step": 9200, "train_loss": 1.9985794973373414, "lr": 0.00020004290614522818, "grad_norm": 0.7003361582756042, "grad_norm_exchange": 0.1913573294878006, "grad_norm_vphi": 0.008813654072582722, "gamma": 0.30000001192092896, "gumbel_tau": 0.6719006299972534, "xi_alphas": [2.34953586186748e-06, 0.5821545124053955, 0.8613877892494202, 0.9719468355178833], "elapsed_s": 7045.852157831192, "exchange_scale": -0.29575031995773315} {"step": 9200, "val_loss": 2.2828803837299345, "val_ppl": 9.80488159633525} {"step": 9250, "train_loss": 1.9995021152496337, "lr": 0.00019757895022207882, "grad_norm": 0.6930111646652222, "grad_norm_exchange": 0.20887252688407898, "grad_norm_vphi": 0.010288972407579422, "gamma": 0.30000001192092896, "gumbel_tau": 0.6691664457321167, "xi_alphas": [2.3514255644840887e-06, 0.5818066596984863, 0.8611264228820801, 0.9718815684318542], "elapsed_s": 7089.579448461533, "exchange_scale": -0.2960110902786255} {"step": 9300, "train_loss": 1.992476029396057, "lr": 0.00019512030916838503, "grad_norm": 0.7193723320960999, "grad_norm_exchange": 0.20532958209514618, "grad_norm_vphi": 0.008591104298830032, "gamma": 0.30000001192092896, "gumbel_tau": 0.6664323210716248, "xi_alphas": [2.352275714656571e-06, 0.5820820331573486, 0.8611350059509277, 0.9718705415725708], "elapsed_s": 7127.084730148315, "exchange_scale": -0.29654163122177124} {"step": 9350, "train_loss": 1.9701823234558105, "lr": 0.00019266723226103627, "grad_norm": 0.7285401821136475, "grad_norm_exchange": 0.22021645307540894, "grad_norm_vphi": 0.00952918455004692, "gamma": 0.30000001192092896, "gumbel_tau": 0.663698136806488, "xi_alphas": [2.3513807718700264e-06, 0.5814782977104187, 0.8610974550247192, 0.9718191623687744], "elapsed_s": 7164.8040606975555, "exchange_scale": -0.2968868911266327} {"step": 9400, "train_loss": 1.9946579980850219, "lr": 0.00019021996821278405, "grad_norm": 0.7639994025230408, "grad_norm_exchange": 0.22514207661151886, "grad_norm_vphi": 0.010070769116282463, "gamma": 0.30000001192092896, "gumbel_tau": 0.6609640121459961, "xi_alphas": [2.3486172722186893e-06, 0.5814656615257263, 0.8611426949501038, 0.9718116521835327], "elapsed_s": 7202.279188156128, "exchange_scale": -0.29664304852485657} {"step": 9450, "train_loss": 1.9868729281425477, "lr": 0.00018777876514702526, "grad_norm": 0.7164201140403748, "grad_norm_exchange": 0.2080976814031601, "grad_norm_vphi": 0.007756154052913189, "gamma": 0.30000001192092896, "gumbel_tau": 0.6582298278808594, "xi_alphas": [2.3470322503271746e-06, 0.5814769864082336, 0.86101233959198, 0.9718303084373474], "elapsed_s": 7239.913954734802, "exchange_scale": -0.2970592975616455} {"step": 9500, "train_loss": 1.9728069591522217, "lr": 0.00018534387057264544, "grad_norm": 0.748760998249054, "grad_norm_exchange": 0.25170576572418213, "grad_norm_vphi": 0.010533014312386513, "gamma": 0.30000001192092896, "gumbel_tau": 0.6554956436157227, "xi_alphas": [2.3462623630621238e-06, 0.5817421674728394, 0.8608551025390625, 0.9718106389045715], "elapsed_s": 7277.468231678009, "exchange_scale": -0.29807233810424805} {"step": 9550, "train_loss": 1.960328495502472, "lr": 0.0001829155313589243, "grad_norm": 0.7122952342033386, "grad_norm_exchange": 0.20605060458183289, "grad_norm_vphi": 0.008769186213612556, "gamma": 0.30000001192092896, "gumbel_tau": 0.6527615189552307, "xi_alphas": [2.346127985219937e-06, 0.5815210342407227, 0.8608766198158264, 0.9717679619789124], "elapsed_s": 7314.98069524765, "exchange_scale": -0.2980574369430542} {"step": 9600, "train_loss": 1.9696760773658752, "lr": 0.00018049399371050613, "grad_norm": 0.7279921770095825, "grad_norm_exchange": 0.21688275039196014, "grad_norm_vphi": 0.010229318402707577, "gamma": 0.30000001192092896, "gumbel_tau": 0.650027334690094, "xi_alphas": [2.3462848730559926e-06, 0.581277072429657, 0.8607532978057861, 0.9717515110969543], "elapsed_s": 7352.652396917343, "exchange_scale": -0.2988136112689972} {"step": 9600, "val_loss": 2.2719862937927244, "val_ppl": 9.698646054850819} {"step": 9650, "train_loss": 1.9954991960525512, "lr": 0.0001780795031424376, "grad_norm": 0.7078324556350708, "grad_norm_exchange": 0.20753026008605957, "grad_norm_vphi": 0.009869877249002457, "gamma": 0.30000001192092896, "gumbel_tau": 0.6472931504249573, "xi_alphas": [2.347658892176696e-06, 0.5813990235328674, 0.8608579635620117, 0.9717200994491577], "elapsed_s": 7396.29561829567, "exchange_scale": -0.2993125915527344} {"step": 9700, "train_loss": 1.996846580505371, "lr": 0.00017567230445527548, "grad_norm": 0.7317695021629333, "grad_norm_exchange": 0.21642759442329407, "grad_norm_vphi": 0.01115950383245945, "gamma": 0.30000001192092896, "gumbel_tau": 0.6445590257644653, "xi_alphas": [2.3482994038204197e-06, 0.5812316536903381, 0.8607001900672913, 0.9717166423797607], "elapsed_s": 7433.9075582027435, "exchange_scale": -0.2991675138473511} {"step": 9750, "train_loss": 1.9617054200172424, "lr": 0.0001732726417102666, "grad_norm": 0.7567629814147949, "grad_norm_exchange": 0.2372010499238968, "grad_norm_vphi": 0.009719368070363998, "gamma": 0.30000001192092896, "gumbel_tau": 0.6418248414993286, "xi_alphas": [2.3482189135393128e-06, 0.5810006856918335, 0.8605148196220398, 0.9716954827308655], "elapsed_s": 7471.397312402725, "exchange_scale": -0.2998640537261963} {"step": 9800, "train_loss": 1.9598763465881348, "lr": 0.00017088075820460348, "grad_norm": 0.7052491903305054, "grad_norm_exchange": 0.22080166637897491, "grad_norm_vphi": 0.01352689042687416, "gamma": 0.30000001192092896, "gumbel_tau": 0.6390907168388367, "xi_alphas": [2.3481695734517416e-06, 0.5808036923408508, 0.8604972958564758, 0.9716356992721558], "elapsed_s": 7509.05957531929, "exchange_scale": -0.3001187741756439} {"step": 9850, "train_loss": 1.9697315669059754, "lr": 0.00016849689644675607, "grad_norm": 0.7505849003791809, "grad_norm_exchange": 0.21840326488018036, "grad_norm_vphi": 0.01131909154355526, "gamma": 0.30000001192092896, "gumbel_tau": 0.6363565325737, "xi_alphas": [2.347793270018883e-06, 0.5806993246078491, 0.8605379462242126, 0.9716278314590454], "elapsed_s": 7546.5698783397675, "exchange_scale": -0.3009811043739319} {"step": 9900, "train_loss": 1.974869728088379, "lr": 0.00016612129813188501, "grad_norm": 0.7388052344322205, "grad_norm_exchange": 0.2207341492176056, "grad_norm_vphi": 0.00909148808568716, "gamma": 0.30000001192092896, "gumbel_tau": 0.6336223483085632, "xi_alphas": [2.3482994038204197e-06, 0.5806638598442078, 0.8605413436889648, 0.9716501832008362], "elapsed_s": 7584.287779092789, "exchange_scale": -0.30109190940856934} {"step": 9950, "train_loss": 1.963337857723236, "lr": 0.0001637542041173362, "grad_norm": 0.712104082107544, "grad_norm_exchange": 0.20450133085250854, "grad_norm_vphi": 0.00870495568960905, "gamma": 0.30000001192092896, "gumbel_tau": 0.6308882236480713, "xi_alphas": [2.3474149202229455e-06, 0.580542802810669, 0.8605563640594482, 0.9716265797615051], "elapsed_s": 7621.864525079727, "exchange_scale": -0.30129721760749817} {"step": 10000, "train_loss": 1.9600527477264404, "lr": 0.0001613958543982211, "grad_norm": 0.7177640795707703, "grad_norm_exchange": 0.21785207092761993, "grad_norm_vphi": 0.009911024942994118, "gamma": 0.30000001192092896, "gumbel_tau": 0.6281540393829346, "xi_alphas": [2.34602953241847e-06, 0.5803585052490234, 0.8603602647781372, 0.9716213941574097], "elapsed_s": 7659.523653745651, "exchange_scale": -0.3015657067298889} {"step": 10000, "val_loss": 2.278463876247406, "val_ppl": 9.761673747526189} {"step": 10050, "train_loss": 1.9691708016395568, "lr": 0.00015904648808308348, "grad_norm": 0.7388735413551331, "grad_norm_exchange": 0.21632133424282074, "grad_norm_vphi": 0.009769349358975887, "gamma": 0.30000001192092896, "gumbel_tau": 0.6254199147224426, "xi_alphas": [2.3467009668820538e-06, 0.5798870921134949, 0.8603288531303406, 0.9715483784675598], "elapsed_s": 7703.217350721359, "exchange_scale": -0.3015611171722412} {"step": 10100, "train_loss": 1.949467706680298, "lr": 0.00015670634336965744, "grad_norm": 0.741375207901001, "grad_norm_exchange": 0.21689967811107635, "grad_norm_vphi": 0.009249687194824219, "gamma": 0.30000001192092896, "gumbel_tau": 0.6226857304573059, "xi_alphas": [2.346040901102242e-06, 0.57976233959198, 0.8603526949882507, 0.9715206623077393], "elapsed_s": 7740.9227504730225, "exchange_scale": -0.3020678460597992} {"step": 10150, "train_loss": 1.9554670095443725, "lr": 0.00015437565752071642, "grad_norm": 0.6917828917503357, "grad_norm_exchange": 0.20859141647815704, "grad_norm_vphi": 0.008433403447270393, "gamma": 0.30000001192092896, "gumbel_tau": 0.6199515461921692, "xi_alphas": [2.3459938347514253e-06, 0.579677164554596, 0.8603083491325378, 0.971498966217041], "elapsed_s": 7778.577641248703, "exchange_scale": -0.3021164536476135} {"step": 10200, "train_loss": 1.952449676990509, "lr": 0.00015205466684001768, "grad_norm": 0.7131646275520325, "grad_norm_exchange": 0.2013605535030365, "grad_norm_vphi": 0.009928001090884209, "gamma": 0.30000001192092896, "gumbel_tau": 0.6172174215316772, "xi_alphas": [2.3440773020411143e-06, 0.5799010396003723, 0.8602440357208252, 0.9714985489845276], "elapsed_s": 7816.1151604652405, "exchange_scale": -0.30264949798583984} {"step": 10250, "train_loss": 1.9593059730529785, "lr": 0.00014974360664834436, "grad_norm": 0.7286937832832336, "grad_norm_exchange": 0.2175651490688324, "grad_norm_vphi": 0.009997368790209293, "gamma": 0.30000001192092896, "gumbel_tau": 0.6144832372665405, "xi_alphas": [2.342474999750266e-06, 0.5796459913253784, 0.8602641820907593, 0.9714887142181396], "elapsed_s": 7853.733071565628, "exchange_scale": -0.30270078778266907} {"step": 10300, "train_loss": 1.95177197933197, "lr": 0.00014744271125964609, "grad_norm": 0.7473929524421692, "grad_norm_exchange": 0.23841799795627594, "grad_norm_vphi": 0.011009376496076584, "gamma": 0.30000001192092896, "gumbel_tau": 0.6117490530014038, "xi_alphas": [2.3433217393176164e-06, 0.5793759822845459, 0.8602656722068787, 0.971467912197113], "elapsed_s": 7891.360934019089, "exchange_scale": -0.30306529998779297} {"step": 10350, "train_loss": 1.954751226902008, "lr": 0.0001451522139572828, "grad_norm": 0.7144908905029297, "grad_norm_exchange": 0.21381798386573792, "grad_norm_vphi": 0.008426161482930183, "gamma": 0.30000001192092896, "gumbel_tau": 0.6090149283409119, "xi_alphas": [2.3412401333189337e-06, 0.579170823097229, 0.8602195382118225, 0.9715052843093872], "elapsed_s": 7928.840879917145, "exchange_scale": -0.3037211000919342} {"step": 10400, "train_loss": 1.937452974319458, "lr": 0.00014287234697037235, "grad_norm": 0.7686460018157959, "grad_norm_exchange": 0.2474180907011032, "grad_norm_vphi": 0.009360440075397491, "gamma": 0.30000001192092896, "gumbel_tau": 0.6062807440757751, "xi_alphas": [2.3399206838803366e-06, 0.5791313052177429, 0.8602054119110107, 0.9714785218238831], "elapsed_s": 7966.3457136154175, "exchange_scale": -0.3040940761566162} {"step": 10400, "val_loss": 2.2550561845302584, "val_ppl": 9.535829061172585} {"step": 10450, "train_loss": 1.9666728115081786, "lr": 0.00014060334145024543, "grad_norm": 0.7238714098930359, "grad_norm_exchange": 0.22253067791461945, "grad_norm_vphi": 0.009217293001711369, "gamma": 0.30000001192092896, "gumbel_tau": 0.6035466194152832, "xi_alphas": [2.339595084777102e-06, 0.5789041519165039, 0.860226035118103, 0.9714491367340088], "elapsed_s": 8010.050325870514, "exchange_scale": -0.30394741892814636} {"step": 10500, "train_loss": 1.943447241783142, "lr": 0.00013834542744700946, "grad_norm": 0.7343020439147949, "grad_norm_exchange": 0.21067774295806885, "grad_norm_vphi": 0.008475088514387608, "gamma": 0.30000001192092896, "gumbel_tau": 0.6008124351501465, "xi_alphas": [2.3386223801935557e-06, 0.5791381001472473, 0.860149085521698, 0.9714387655258179], "elapsed_s": 8047.6118268966675, "exchange_scale": -0.30439525842666626} {"step": 10550, "train_loss": 1.930708429813385, "lr": 0.0001360988338862243, "grad_norm": 0.7166001200675964, "grad_norm_exchange": 0.21182343363761902, "grad_norm_vphi": 0.008995573036372662, "gamma": 0.30000001192092896, "gumbel_tau": 0.5980782508850098, "xi_alphas": [2.3391444301523734e-06, 0.5792127847671509, 0.8600566983222961, 0.9714211225509644], "elapsed_s": 8085.391915798187, "exchange_scale": -0.30448901653289795} {"step": 10600, "train_loss": 1.9650618648529052, "lr": 0.00013386378854569195, "grad_norm": 0.7197250127792358, "grad_norm_exchange": 0.20995080471038818, "grad_norm_vphi": 0.011460576206445694, "gamma": 0.30000001192092896, "gumbel_tau": 0.5953441262245178, "xi_alphas": [2.335116732865572e-06, 0.5789510607719421, 0.8599470257759094, 0.9714134931564331], "elapsed_s": 8123.009086370468, "exchange_scale": -0.3050985038280487} {"step": 10650, "train_loss": 1.9224012231826781, "lr": 0.0001316405180323624, "grad_norm": 0.7112909555435181, "grad_norm_exchange": 0.2125304937362671, "grad_norm_vphi": 0.00837060995399952, "gamma": 0.30000001192092896, "gumbel_tau": 0.5926099419593811, "xi_alphas": [2.334992132091429e-06, 0.5789490342140198, 0.8599036931991577, 0.9714169502258301], "elapsed_s": 8160.738867998123, "exchange_scale": -0.30570754408836365} {"step": 10700, "train_loss": 1.9418506956100463, "lr": 0.00012942924775935884, "grad_norm": 0.7351611256599426, "grad_norm_exchange": 0.22134250402450562, "grad_norm_vphi": 0.010698288679122925, "gamma": 0.30000001192092896, "gumbel_tau": 0.5898758172988892, "xi_alphas": [2.33610580835375e-06, 0.5788106918334961, 0.8598169684410095, 0.9714212417602539], "elapsed_s": 8198.267460107803, "exchange_scale": -0.3058582544326782} {"step": 10750, "train_loss": 1.9618505573272704, "lr": 0.00012723020192312296, "grad_norm": 0.7358535528182983, "grad_norm_exchange": 0.22814442217350006, "grad_norm_vphi": 0.00991804338991642, "gamma": 0.30000001192092896, "gumbel_tau": 0.5871416330337524, "xi_alphas": [2.3364600565400906e-06, 0.5788372755050659, 0.8598771691322327, 0.9714216589927673], "elapsed_s": 8235.91930437088, "exchange_scale": -0.305960088968277} {"step": 10800, "train_loss": 1.9436074447631837, "lr": 0.0001250436034806843, "grad_norm": 0.7310090661048889, "grad_norm_exchange": 0.21201951801776886, "grad_norm_vphi": 0.008571653626859188, "gamma": 0.30000001192092896, "gumbel_tau": 0.5844074487686157, "xi_alphas": [2.336515763090574e-06, 0.5786827802658081, 0.859867513179779, 0.9714025855064392], "elapsed_s": 8273.424757003784, "exchange_scale": -0.30623888969421387} {"step": 10800, "val_loss": 2.2586009383201597, "val_ppl": 9.569691208407129} {"step": 10850, "train_loss": 1.960360391139984, "lr": 0.00012286967412705513, "grad_norm": 0.737899899482727, "grad_norm_exchange": 0.21254998445510864, "grad_norm_vphi": 0.008894022554159164, "gamma": 0.30000001192092896, "gumbel_tau": 0.5816733241081238, "xi_alphas": [2.3364532353298273e-06, 0.5785892605781555, 0.8599036931991577, 0.9713915586471558], "elapsed_s": 8317.207721710205, "exchange_scale": -0.30650410056114197} {"step": 10900, "train_loss": 1.9405962562561034, "lr": 0.00012070863427275313, "grad_norm": 0.7598791718482971, "grad_norm_exchange": 0.21936039626598358, "grad_norm_vphi": 0.012010744772851467, "gamma": 0.30000001192092896, "gumbel_tau": 0.5789391398429871, "xi_alphas": [2.338118292755098e-06, 0.5783443450927734, 0.8598052859306335, 0.9713835716247559], "elapsed_s": 8354.655797719955, "exchange_scale": -0.30633658170700073} {"step": 10950, "train_loss": 1.9336509895324707, "lr": 0.00011856070302145446, "grad_norm": 0.7400899529457092, "grad_norm_exchange": 0.2067195177078247, "grad_norm_vphi": 0.01017720252275467, "gamma": 0.30000001192092896, "gumbel_tau": 0.5762049555778503, "xi_alphas": [2.33754985856649e-06, 0.5780614614486694, 0.8597326278686523, 0.9713626503944397], "elapsed_s": 8392.295881986618, "exchange_scale": -0.30671441555023193} {"step": 11000, "train_loss": 1.932325885295868, "lr": 0.00011642609814777921, "grad_norm": 0.7737302780151367, "grad_norm_exchange": 0.21693451702594757, "grad_norm_vphi": 0.012750490568578243, "gamma": 0.30000001192092896, "gumbel_tau": 0.5734708309173584, "xi_alphas": [2.337937758056796e-06, 0.577860414981842, 0.8596416711807251, 0.971367597579956], "elapsed_s": 8429.786516189575, "exchange_scale": -0.30699655413627625} {"step": 11050, "train_loss": 1.9518938231468201, "lr": 0.00011430503607521165, "grad_norm": 0.7614089846611023, "grad_norm_exchange": 0.23046134412288666, "grad_norm_vphi": 0.010585899464786053, "gamma": 0.30000001192092896, "gumbel_tau": 0.5707366466522217, "xi_alphas": [2.3394522941089235e-06, 0.578011155128479, 0.8597427606582642, 0.9713699221611023], "elapsed_s": 8467.435214281082, "exchange_scale": -0.3072666525840759} {"step": 11100, "train_loss": 1.935980703830719, "lr": 0.00011219773185415749, "grad_norm": 0.7668430805206299, "grad_norm_exchange": 0.22720463573932648, "grad_norm_vphi": 0.011012551374733448, "gamma": 0.30000001192092896, "gumbel_tau": 0.5680025219917297, "xi_alphas": [2.3382185645459685e-06, 0.5778048634529114, 0.8596937656402588, 0.9713539481163025], "elapsed_s": 8504.922073841095, "exchange_scale": -0.3072502613067627} {"step": 11150, "train_loss": 1.9239383769035339, "lr": 0.00011010439914014025, "grad_norm": 0.7389852404594421, "grad_norm_exchange": 0.21533571183681488, "grad_norm_vphi": 0.01057231705635786, "gamma": 0.30000001192092896, "gumbel_tau": 0.565268337726593, "xi_alphas": [2.3379466256301384e-06, 0.5778921842575073, 0.8596826791763306, 0.9713181853294373], "elapsed_s": 8542.518859386444, "exchange_scale": -0.30751779675483704} {"step": 11200, "train_loss": 1.9382430791854859, "lr": 0.00010802525017213948, "grad_norm": 0.7580653429031372, "grad_norm_exchange": 0.2255992442369461, "grad_norm_vphi": 0.012586411088705063, "gamma": 0.30000001192092896, "gumbel_tau": 0.5625341534614563, "xi_alphas": [2.337761770832003e-06, 0.577782928943634, 0.8596951961517334, 0.971324622631073], "elapsed_s": 8580.282788276672, "exchange_scale": -0.3078555166721344} {"step": 11200, "val_loss": 2.265278923511505, "val_ppl": 9.633811322881284} {"step": 11250, "train_loss": 1.9236542654037476, "lr": 0.00010596049575107167, "grad_norm": 0.7423180937767029, "grad_norm_exchange": 0.21894145011901855, "grad_norm_vphi": 0.009359038434922695, "gamma": 0.30000001192092896, "gumbel_tau": 0.5598000288009644, "xi_alphas": [2.3389145553665003e-06, 0.577724039554596, 0.8597070574760437, 0.9713293313980103], "elapsed_s": 8623.961544513702, "exchange_scale": -0.3084457814693451} {"step": 11300, "train_loss": 1.946609823703766, "lr": 0.00010391034521841805, "grad_norm": 0.7545937895774841, "grad_norm_exchange": 0.22174423933029175, "grad_norm_vphi": 0.00942992139607668, "gamma": 0.30000001192092896, "gumbel_tau": 0.5570658445358276, "xi_alphas": [2.3371485440293327e-06, 0.5775259137153625, 0.8596530556678772, 0.9713206887245178], "elapsed_s": 8661.611247062683, "exchange_scale": -0.308651864528656} {"step": 11350, "train_loss": 1.9321112370491027, "lr": 0.00010187500643499958, "grad_norm": 0.7429212331771851, "grad_norm_exchange": 0.2111595869064331, "grad_norm_vphi": 0.015302258543670177, "gamma": 0.30000001192092896, "gumbel_tau": 0.5543317198753357, "xi_alphas": [2.337846353839268e-06, 0.5775271058082581, 0.8596366047859192, 0.971316397190094], "elapsed_s": 8699.103975534439, "exchange_scale": -0.30861175060272217} {"step": 11400, "train_loss": 1.9264032888412475, "lr": 9.985468575990253e-05, "grad_norm": 0.7334105372428894, "grad_norm_exchange": 0.21838362514972687, "grad_norm_vphi": 0.009217179380357265, "gamma": 0.30000001192092896, "gumbel_tau": 0.551597535610199, "xi_alphas": [2.3391689865093213e-06, 0.5775173902511597, 0.8596147298812866, 0.9713191986083984], "elapsed_s": 8736.782582044601, "exchange_scale": -0.30878961086273193} {"step": 11450, "train_loss": 1.9195692467689514, "lr": 9.784958802955596e-05, "grad_norm": 0.7411181926727295, "grad_norm_exchange": 0.22872775793075562, "grad_norm_vphi": 0.010440163314342499, "gamma": 0.30000001192092896, "gumbel_tau": 0.5488633513450623, "xi_alphas": [2.339605998713523e-06, 0.577328622341156, 0.8596364259719849, 0.9712880849838257], "elapsed_s": 8774.253230810165, "exchange_scale": -0.30874305963516235} {"step": 11500, "train_loss": 1.95079003572464, "lr": 9.585991653696391e-05, "grad_norm": 0.7423210740089417, "grad_norm_exchange": 0.22242137789726257, "grad_norm_vphi": 0.009898561984300613, "gamma": 0.30000001192092896, "gumbel_tau": 0.5461292266845703, "xi_alphas": [2.3390527985611698e-06, 0.577226459980011, 0.8595935106277466, 0.9713003039360046], "elapsed_s": 8811.895185470581, "exchange_scale": -0.30890417098999023} {"step": 11550, "train_loss": 1.9089111113548278, "lr": 9.388587301109381e-05, "grad_norm": 0.7940544486045837, "grad_norm_exchange": 0.23908820748329163, "grad_norm_vphi": 0.01071004569530487, "gamma": 0.30000001192092896, "gumbel_tau": 0.5433950424194336, "xi_alphas": [2.338595777473529e-06, 0.5771317481994629, 0.8596193194389343, 0.9712908864021301], "elapsed_s": 8849.438906431198, "exchange_scale": -0.3092745244503021} {"step": 11600, "train_loss": 1.9191077995300292, "lr": 9.19276575964235e-05, "grad_norm": 0.758101761341095, "grad_norm_exchange": 0.2216409295797348, "grad_norm_vphi": 0.00988239236176014, "gamma": 0.30000001192092896, "gumbel_tau": 0.5406608581542969, "xi_alphas": [2.3380271159112453e-06, 0.5771769881248474, 0.8596481084823608, 0.9712944626808167], "elapsed_s": 8887.077696561813, "exchange_scale": -0.30976611375808716} {"step": 11600, "val_loss": 2.2318554908037185, "val_ppl": 9.317137914379932} {"step": 11650, "train_loss": 1.9326915049552917, "lr": 8.998546883264911e-05, "grad_norm": 0.7369346618652344, "grad_norm_exchange": 0.23335716128349304, "grad_norm_vphi": 0.010198927484452724, "gamma": 0.30000001192092896, "gumbel_tau": 0.5379267334938049, "xi_alphas": [2.337768364668591e-06, 0.5771420001983643, 0.8597098588943481, 0.971290647983551], "elapsed_s": 8930.706982374191, "exchange_scale": -0.3102961480617523} {"step": 11700, "train_loss": 1.9075058054924012, "lr": 8.805950363455559e-05, "grad_norm": 0.7574648857116699, "grad_norm_exchange": 0.22010383009910583, "grad_norm_vphi": 0.011061605997383595, "gamma": 0.30000001192092896, "gumbel_tau": 0.5351925492286682, "xi_alphas": [2.3373715976049425e-06, 0.5770808458328247, 0.8596205711364746, 0.971278727054596], "elapsed_s": 8968.340255260468, "exchange_scale": -0.3104842007160187} {"step": 11750, "train_loss": 1.9134907579421998, "lr": 8.614995727205155e-05, "grad_norm": 0.7321592569351196, "grad_norm_exchange": 0.22265873849391937, "grad_norm_vphi": 0.011136888526380062, "gamma": 0.30000001192092896, "gumbel_tau": 0.5324584245681763, "xi_alphas": [2.33705736718548e-06, 0.5769887566566467, 0.8596280217170715, 0.971274197101593], "elapsed_s": 9005.92540383339, "exchange_scale": -0.3106803596019745} {"step": 11800, "train_loss": 1.8967229413986206, "lr": 8.425702335037153e-05, "grad_norm": 0.7498713731765747, "grad_norm_exchange": 0.2150365263223648, "grad_norm_vphi": 0.01128822099417448, "gamma": 0.30000001192092896, "gumbel_tau": 0.5297242403030396, "xi_alphas": [2.336208353881375e-06, 0.5767865180969238, 0.8596426844596863, 0.9712480306625366], "elapsed_s": 9043.472266674042, "exchange_scale": -0.3107185661792755} {"step": 11850, "train_loss": 1.8927727961540222, "lr": 8.238089379044661e-05, "grad_norm": 0.7440163493156433, "grad_norm_exchange": 0.21950876712799072, "grad_norm_vphi": 0.009578959085047245, "gamma": 0.30000001192092896, "gumbel_tau": 0.5269900560379028, "xi_alphas": [2.33705736718548e-06, 0.5767737030982971, 0.8596038818359375, 0.9712523221969604], "elapsed_s": 9081.162553071976, "exchange_scale": -0.31070154905319214} {"step": 11900, "train_loss": 1.8983879089355469, "lr": 8.05217588094459e-05, "grad_norm": 0.7707085013389587, "grad_norm_exchange": 0.22183769941329956, "grad_norm_vphi": 0.009956668131053448, "gamma": 0.30000001192092896, "gumbel_tau": 0.5242559313774109, "xi_alphas": [2.33705736718548e-06, 0.5766333341598511, 0.8595903515815735, 0.9712386727333069], "elapsed_s": 9118.672793626785, "exchange_scale": -0.31103649735450745} {"step": 11950, "train_loss": 1.9109831905364991, "lr": 7.86798069014909e-05, "grad_norm": 0.737337052822113, "grad_norm_exchange": 0.2135179340839386, "grad_norm_vphi": 0.011378408409655094, "gamma": 0.30000001192092896, "gumbel_tau": 0.5215217471122742, "xi_alphas": [2.336234956601402e-06, 0.5764716267585754, 0.8596106171607971, 0.9712329506874084], "elapsed_s": 9156.403069972992, "exchange_scale": -0.31151652336120605} {"step": 12000, "train_loss": 1.9036629033088683, "lr": 7.685522481854432e-05, "grad_norm": 0.7584449052810669, "grad_norm_exchange": 0.2275920808315277, "grad_norm_vphi": 0.013177013956010342, "gamma": 0.30000001192092896, "gumbel_tau": 0.5187876224517822, "xi_alphas": [2.3347738533630036e-06, 0.5765308737754822, 0.8595544099807739, 0.971216082572937], "elapsed_s": 9194.017760276794, "exchange_scale": -0.3115788400173187} {"step": 12000, "val_loss": 2.2660814940929415, "val_ppl": 9.641546239929392} {"step": 12050, "train_loss": 1.9110637855529786, "lr": 7.504819755147574e-05, "grad_norm": 0.7486110925674438, "grad_norm_exchange": 0.2185380756855011, "grad_norm_vphi": 0.010272989049553871, "gamma": 0.30000001192092896, "gumbel_tau": 0.5160534381866455, "xi_alphas": [2.3344532564806286e-06, 0.5763482451438904, 0.8595536351203918, 0.9711967706680298], "elapsed_s": 9237.803401470184, "exchange_scale": -0.3121049404144287} {"step": 12100, "train_loss": 1.911932668685913, "lr": 7.325890831130555e-05, "grad_norm": 0.7357195019721985, "grad_norm_exchange": 0.20650731027126312, "grad_norm_vphi": 0.009005294181406498, "gamma": 0.30000001192092896, "gumbel_tau": 0.5133192539215088, "xi_alphas": [2.335479848625255e-06, 0.5761674046516418, 0.8595049977302551, 0.9711709022521973], "elapsed_s": 9275.371569395065, "exchange_scale": -0.31217876076698303} {"step": 12150, "train_loss": 1.9118061876296997, "lr": 7.148753851063005e-05, "grad_norm": 0.7714627981185913, "grad_norm_exchange": 0.22436630725860596, "grad_norm_vphi": 0.011003135703504086, "gamma": 0.30000001192092896, "gumbel_tau": 0.5105851292610168, "xi_alphas": [2.335513272555545e-06, 0.576148271560669, 0.8595118522644043, 0.9711662530899048], "elapsed_s": 9313.228547096252, "exchange_scale": -0.3124440908432007} {"step": 12200, "train_loss": 1.9031609773635865, "lr": 6.97342677452277e-05, "grad_norm": 0.7351601123809814, "grad_norm_exchange": 0.21133507788181305, "grad_norm_vphi": 0.00801935326308012, "gamma": 0.30000001192092896, "gumbel_tau": 0.5078509449958801, "xi_alphas": [2.3344111923506716e-06, 0.5761792659759521, 0.8595903515815735, 0.9711659550666809], "elapsed_s": 9350.755737304688, "exchange_scale": -0.3122459352016449} {"step": 12250, "train_loss": 1.8870721530914307, "lr": 6.799927377585066e-05, "grad_norm": 0.7382734417915344, "grad_norm_exchange": 0.21107833087444305, "grad_norm_vphi": 0.016320763155817986, "gamma": 0.30000001192092896, "gumbel_tau": 0.5051167607307434, "xi_alphas": [2.3352749849436805e-06, 0.576185941696167, 0.8595271706581116, 0.9711365699768066], "elapsed_s": 9388.403141736984, "exchange_scale": -0.3125786781311035} {"step": 12300, "train_loss": 1.9008273911476135, "lr": 6.628273251020192e-05, "grad_norm": 0.7663930654525757, "grad_norm_exchange": 0.2218717783689499, "grad_norm_vphi": 0.010108746588230133, "gamma": 0.30000001192092896, "gumbel_tau": 0.5023826360702515, "xi_alphas": [2.3334025627264054e-06, 0.576018750667572, 0.8594509959220886, 0.9711347818374634], "elapsed_s": 9425.900233507156, "exchange_scale": -0.3128425180912018} {"step": 12350, "train_loss": 1.9006095433235168, "lr": 6.458481798510024e-05, "grad_norm": 0.750572144985199, "grad_norm_exchange": 0.2268831729888916, "grad_norm_vphi": 0.009679722599685192, "gamma": 0.30000001192092896, "gumbel_tau": 0.49964845180511475, "xi_alphas": [2.3329598661803175e-06, 0.5759119391441345, 0.8593895435333252, 0.9711222052574158], "elapsed_s": 9463.590856313705, "exchange_scale": -0.313070148229599} {"step": 12400, "train_loss": 1.8939482736587525, "lr": 6.290570234883505e-05, "grad_norm": 0.7848731279373169, "grad_norm_exchange": 0.2453247606754303, "grad_norm_vphi": 0.00957657489925623, "gamma": 0.30000001192092896, "gumbel_tau": 0.4969142973423004, "xi_alphas": [2.3346892703557387e-06, 0.5757834911346436, 0.8593705296516418, 0.9711082577705383], "elapsed_s": 9501.082515954971, "exchange_scale": -0.3132226765155792} {"step": 12400, "val_loss": 2.2411867678165436, "val_ppl": 9.404485608190129} {"step": 12450, "train_loss": 1.882430067062378, "lr": 6.124555584371256e-05, "grad_norm": 0.7846888303756714, "grad_norm_exchange": 0.22689847648143768, "grad_norm_vphi": 0.010816576890647411, "gamma": 0.30000001192092896, "gumbel_tau": 0.4941801428794861, "xi_alphas": [2.334796135983197e-06, 0.5758523941040039, 0.8593719601631165, 0.9710896015167236], "elapsed_s": 9544.88162946701, "exchange_scale": -0.3134877681732178} {"step": 12500, "train_loss": 1.9090777134895325, "lr": 5.960454678879518e-05, "grad_norm": 0.7850832343101501, "grad_norm_exchange": 0.23480798304080963, "grad_norm_vphi": 0.013412690721452236, "gamma": 0.30000001192092896, "gumbel_tau": 0.49144598841667175, "xi_alphas": [2.3331333522946807e-06, 0.5757837891578674, 0.8593654036521912, 0.9710761308670044], "elapsed_s": 9582.385235071182, "exchange_scale": -0.31376320123672485} {"step": 12550, "train_loss": 1.889548087120056, "lr": 5.798284156283609e-05, "grad_norm": 0.8152822256088257, "grad_norm_exchange": 0.23493286967277527, "grad_norm_vphi": 0.01047549583017826, "gamma": 0.30000001192092896, "gumbel_tau": 0.4887118339538574, "xi_alphas": [2.332426220164052e-06, 0.5756568312644958, 0.8593147397041321, 0.9710602760314941], "elapsed_s": 9620.080744743347, "exchange_scale": -0.31377100944519043} {"step": 12600, "train_loss": 1.9214084029197693, "lr": 5.638060458741032e-05, "grad_norm": 0.7746152877807617, "grad_norm_exchange": 0.22363346815109253, "grad_norm_vphi": 0.009973883628845215, "gamma": 0.30000001192092896, "gumbel_tau": 0.4859776496887207, "xi_alphas": [2.3330487692874158e-06, 0.5755903124809265, 0.8593153357505798, 0.9710690379142761], "elapsed_s": 9657.575790643692, "exchange_scale": -0.31376931071281433} {"step": 12650, "train_loss": 1.916095666885376, "lr": 5.479799831024457e-05, "grad_norm": 0.7635933756828308, "grad_norm_exchange": 0.22164086997509003, "grad_norm_vphi": 0.010805233381688595, "gamma": 0.30000001192092896, "gumbel_tau": 0.48324349522590637, "xi_alphas": [2.3334559955401346e-06, 0.5756043791770935, 0.8593344688415527, 0.9710666537284851], "elapsed_s": 9695.132489204407, "exchange_scale": -0.31424441933631897} {"step": 12700, "train_loss": 1.8947441744804383, "lr": 5.3235183188746596e-05, "grad_norm": 0.7583755850791931, "grad_norm_exchange": 0.21701325476169586, "grad_norm_vphi": 0.013788873329758644, "gamma": 0.30000001192092896, "gumbel_tau": 0.48050934076309204, "xi_alphas": [2.332697476958856e-06, 0.5755080580711365, 0.8593388199806213, 0.9710593819618225], "elapsed_s": 9732.843006849289, "exchange_scale": -0.31453099846839905} {"step": 12750, "train_loss": 1.9032840299606324, "lr": 5.169231767373697e-05, "grad_norm": 0.7925540804862976, "grad_norm_exchange": 0.23672059178352356, "grad_norm_vphi": 0.011170880869030952, "gamma": 0.30000001192092896, "gumbel_tau": 0.4777751863002777, "xi_alphas": [2.3324125777435256e-06, 0.5754852890968323, 0.8593335747718811, 0.971047580242157], "elapsed_s": 9770.274359226227, "exchange_scale": -0.3144637644290924} {"step": 12800, "train_loss": 1.8891470694541932, "lr": 5.016955819338403e-05, "grad_norm": 0.7688220143318176, "grad_norm_exchange": 0.2246275246143341, "grad_norm_vphi": 0.012032379396259785, "gamma": 0.30000001192092896, "gumbel_tau": 0.475041002035141, "xi_alphas": [2.332744088562322e-06, 0.57537841796875, 0.8593372702598572, 0.9710379242897034], "elapsed_s": 9807.918674707413, "exchange_scale": -0.314662367105484} {"step": 12800, "val_loss": 2.243387061357498, "val_ppl": 9.425201018764426} {"step": 12850, "train_loss": 1.8995095109939575, "lr": 4.866705913734398e-05, "grad_norm": 0.7797132730484009, "grad_norm_exchange": 0.22824601829051971, "grad_norm_vphi": 0.012179655954241753, "gamma": 0.30000001192092896, "gumbel_tau": 0.47230684757232666, "xi_alphas": [2.332221356482478e-06, 0.575222909450531, 0.8593277335166931, 0.9710366725921631], "elapsed_s": 9851.552411317825, "exchange_scale": -0.3148028254508972} {"step": 12900, "train_loss": 1.8974608755111695, "lr": 4.718497284110746e-05, "grad_norm": 0.7752254605293274, "grad_norm_exchange": 0.21886815130710602, "grad_norm_vphi": 0.010878796689212322, "gamma": 0.30000001192092896, "gumbel_tau": 0.46957269310951233, "xi_alphas": [2.3312184112000978e-06, 0.5751054883003235, 0.8593323230743408, 0.9710261225700378], "elapsed_s": 9889.235200166702, "exchange_scale": -0.3151853680610657} {"step": 12950, "train_loss": 1.8873084449768067, "lr": 4.572344957055466e-05, "grad_norm": 0.7617862820625305, "grad_norm_exchange": 0.2359786480665207, "grad_norm_vphi": 0.011333232745528221, "gamma": 0.30000001192092896, "gumbel_tau": 0.466838538646698, "xi_alphas": [2.3320678792515537e-06, 0.5750601887702942, 0.8593044877052307, 0.9710314869880676], "elapsed_s": 9926.731809139252, "exchange_scale": -0.3151940703392029} {"step": 13000, "train_loss": 1.901939618587494, "lr": 4.428263750672018e-05, "grad_norm": 0.7911465764045715, "grad_norm_exchange": 0.23484574258327484, "grad_norm_vphi": 0.010920871049165726, "gamma": 0.30000001192092896, "gumbel_tau": 0.4641043543815613, "xi_alphas": [2.3313650672207586e-06, 0.5751597881317139, 0.8593586683273315, 0.9710248708724976], "elapsed_s": 9964.424211740494, "exchange_scale": -0.3153129816055298} {"step": 13050, "train_loss": 1.9060602736473085, "lr": 4.2862682730769154e-05, "grad_norm": 0.7892533540725708, "grad_norm_exchange": 0.23355866968631744, "grad_norm_vphi": 0.010067534632980824, "gamma": 0.30000001192092896, "gumbel_tau": 0.46137019991874695, "xi_alphas": [2.3308427898882655e-06, 0.5751941800117493, 0.8593393564224243, 0.9710110425949097], "elapsed_s": 10001.92383813858, "exchange_scale": -0.315500408411026} {"step": 13100, "train_loss": 1.9178500914573668, "lr": 4.1463729209186286e-05, "grad_norm": 0.7900871634483337, "grad_norm_exchange": 0.22434498369693756, "grad_norm_vphi": 0.009610023349523544, "gamma": 0.30000001192092896, "gumbel_tau": 0.4586360454559326, "xi_alphas": [2.33199671129114e-06, 0.5751445293426514, 0.8593428134918213, 0.9710054993629456], "elapsed_s": 10039.96907711029, "exchange_scale": -0.3156341314315796} {"step": 13150, "train_loss": 1.895253758430481, "lr": 4.008591877917978e-05, "grad_norm": 0.7638344764709473, "grad_norm_exchange": 0.22171048820018768, "grad_norm_vphi": 0.010881759226322174, "gamma": 0.30000001192092896, "gumbel_tau": 0.4559018909931183, "xi_alphas": [2.331080622752779e-06, 0.5750739574432373, 0.859338641166687, 0.970995306968689], "elapsed_s": 10077.788608789444, "exchange_scale": -0.3157362639904022} {"step": 13200, "train_loss": 1.8989457750320435, "lr": 3.87293911343001e-05, "grad_norm": 0.78743976354599, "grad_norm_exchange": 0.21082019805908203, "grad_norm_vphi": 0.008784106001257896, "gamma": 0.30000001192092896, "gumbel_tau": 0.45316770672798157, "xi_alphas": [2.3318477815337246e-06, 0.5750808119773865, 0.8593219518661499, 0.9709831476211548], "elapsed_s": 10115.80887913704, "exchange_scale": -0.31582558155059814} {"step": 13200, "val_loss": 2.2413078010082246, "val_ppl": 9.405623931985497} {"step": 13250, "train_loss": 1.8754441595077516, "lr": 3.739428381027726e-05, "grad_norm": 0.7534888982772827, "grad_norm_exchange": 0.2187434881925583, "grad_norm_vphi": 0.0091418931260705, "gamma": 0.30000001192092896, "gumbel_tau": 0.45043355226516724, "xi_alphas": [2.331100631636218e-06, 0.5749892592430115, 0.8592900633811951, 0.9709864258766174], "elapsed_s": 10160.208707332611, "exchange_scale": -0.315994530916214} {"step": 13300, "train_loss": 1.8978002858161926, "lr": 3.608073217107599e-05, "grad_norm": 0.7792009711265564, "grad_norm_exchange": 0.2275429666042328, "grad_norm_vphi": 0.012402663938701153, "gamma": 0.30000001192092896, "gumbel_tau": 0.4476993978023529, "xi_alphas": [2.33106061386934e-06, 0.5749520063400269, 0.8593192100524902, 0.9709824919700623], "elapsed_s": 10198.609374284744, "exchange_scale": -0.3161337375640869} {"step": 13350, "train_loss": 1.8947645664215087, "lr": 3.478886939517167e-05, "grad_norm": 0.7721938490867615, "grad_norm_exchange": 0.22707512974739075, "grad_norm_vphi": 0.010615166276693344, "gamma": 0.30000001192092896, "gumbel_tau": 0.4449652433395386, "xi_alphas": [2.3308828076551436e-06, 0.5749532580375671, 0.8593075275421143, 0.970971941947937], "elapsed_s": 10236.961880683899, "exchange_scale": -0.31641456484794617} {"step": 13400, "train_loss": 1.9018290376663207, "lr": 3.351882646204746e-05, "grad_norm": 0.7719407677650452, "grad_norm_exchange": 0.21499894559383392, "grad_norm_vphi": 0.010232721455395222, "gamma": 0.30000001192092896, "gumbel_tau": 0.44223108887672424, "xi_alphas": [2.330144752704655e-06, 0.5748491287231445, 0.8592907786369324, 0.9709623456001282], "elapsed_s": 10275.378651857376, "exchange_scale": -0.3162859082221985} {"step": 13450, "train_loss": 1.8927726936340332, "lr": 3.2270732138914654e-05, "grad_norm": 0.8012478947639465, "grad_norm_exchange": 0.2391803115606308, "grad_norm_vphi": 0.011309854686260223, "gamma": 0.30000001192092896, "gumbel_tau": 0.4394969046115875, "xi_alphas": [2.3302538920688676e-06, 0.5747353434562683, 0.8592984676361084, 0.9709591269493103], "elapsed_s": 10313.678936243057, "exchange_scale": -0.3165181577205658} {"step": 13500, "train_loss": 1.8780881786346435, "lr": 3.104471296765718e-05, "grad_norm": 0.7982211709022522, "grad_norm_exchange": 0.2365650236606598, "grad_norm_vphi": 0.012398429214954376, "gamma": 0.30000001192092896, "gumbel_tau": 0.4367627501487732, "xi_alphas": [2.329784820176428e-06, 0.5747663378715515, 0.8592956066131592, 0.9709616899490356], "elapsed_s": 10351.930108070374, "exchange_scale": -0.3165301978588104} {"step": 13550, "train_loss": 1.8907312989234923, "lr": 2.9840893252001706e-05, "grad_norm": 0.7710391283035278, "grad_norm_exchange": 0.23303601145744324, "grad_norm_vphi": 0.009586144238710403, "gamma": 0.30000001192092896, "gumbel_tau": 0.43402859568595886, "xi_alphas": [2.329904873477062e-06, 0.5747901201248169, 0.859296441078186, 0.9709605574607849], "elapsed_s": 10389.415353298187, "exchange_scale": -0.3166668713092804} {"step": 13600, "train_loss": 1.8853900623321533, "lr": 2.8659395044914798e-05, "grad_norm": 0.7646403908729553, "grad_norm_exchange": 0.220048725605011, "grad_norm_vphi": 0.011117377318441868, "gamma": 0.30000001192092896, "gumbel_tau": 0.43129444122314453, "xi_alphas": [2.3293694084713934e-06, 0.574702799320221, 0.8592575192451477, 0.970960259437561], "elapsed_s": 10426.90375828743, "exchange_scale": -0.31674399971961975} {"step": 13600, "val_loss": 2.259822744131088, "val_ppl": 9.581390658506988} {"step": 13650, "train_loss": 1.89985267162323, "lr": 2.750033813622829e-05, "grad_norm": 0.7818869948387146, "grad_norm_exchange": 0.23072780668735504, "grad_norm_vphi": 0.010804619640111923, "gamma": 0.30000001192092896, "gumbel_tau": 0.4285602569580078, "xi_alphas": [2.329587232452468e-06, 0.5746707320213318, 0.8592653274536133, 0.9709567427635193], "elapsed_s": 10470.655120372772, "exchange_scale": -0.3167136609554291} {"step": 13700, "train_loss": 1.8805175876617433, "lr": 2.6363840040493748e-05, "grad_norm": 0.7812139391899109, "grad_norm_exchange": 0.22497227787971497, "grad_norm_vphi": 0.016969889402389526, "gamma": 0.30000001192092896, "gumbel_tau": 0.4258261024951935, "xi_alphas": [2.32991396842408e-06, 0.5746397972106934, 0.8592659831047058, 0.9709522724151611], "elapsed_s": 10508.136870861053, "exchange_scale": -0.316729336977005} {"step": 13750, "train_loss": 1.8956913590431212, "lr": 2.5250015985068152e-05, "grad_norm": 0.7634011507034302, "grad_norm_exchange": 0.1999131441116333, "grad_norm_vphi": 0.008276876993477345, "gamma": 0.30000001192092896, "gumbel_tau": 0.42309194803237915, "xi_alphas": [2.3298225642065518e-06, 0.574611485004425, 0.859286367893219, 0.9709556102752686], "elapsed_s": 10545.826165914536, "exchange_scale": -0.3168658912181854} {"step": 13800, "train_loss": 1.8865007376670837, "lr": 2.4158978898431106e-05, "grad_norm": 0.7945230603218079, "grad_norm_exchange": 0.22574257850646973, "grad_norm_vphi": 0.009670931845903397, "gamma": 0.30000001192092896, "gumbel_tau": 0.4203577935695648, "xi_alphas": [2.3298537144000875e-06, 0.5745905041694641, 0.8592752814292908, 0.9709510207176208], "elapsed_s": 10583.391526460648, "exchange_scale": -0.3168248236179352} {"step": 13850, "train_loss": 1.8850503277778625, "lr": 2.309083939873527e-05, "grad_norm": 0.7662644386291504, "grad_norm_exchange": 0.2298402041196823, "grad_norm_vphi": 0.01137507800012827, "gamma": 0.30000001192092896, "gumbel_tau": 0.4176236093044281, "xi_alphas": [2.3299403437704314e-06, 0.5745428204536438, 0.8592771291732788, 0.9709436893463135], "elapsed_s": 10621.077138185501, "exchange_scale": -0.3169685900211334} {"step": 13900, "train_loss": 1.890521855354309, "lr": 2.204570578259099e-05, "grad_norm": 0.7529791593551636, "grad_norm_exchange": 0.2120165377855301, "grad_norm_vphi": 0.009537867270410061, "gamma": 0.30000001192092896, "gumbel_tau": 0.41488945484161377, "xi_alphas": [2.3301292912947247e-06, 0.5745423436164856, 0.8592398166656494, 0.9709375500679016], "elapsed_s": 10658.635739326477, "exchange_scale": -0.317250519990921} {"step": 13950, "train_loss": 1.9013467645645141, "lr": 2.102368401408625e-05, "grad_norm": 0.7814764976501465, "grad_norm_exchange": 0.24020054936408997, "grad_norm_vphi": 0.00952623039484024, "gamma": 0.30000001192092896, "gumbel_tau": 0.41215530037879944, "xi_alphas": [2.330158167751506e-06, 0.5745142698287964, 0.8592357635498047, 0.970935046672821], "elapsed_s": 10696.314475774765, "exchange_scale": -0.3173271715641022} {"step": 14000, "train_loss": 1.887969377040863, "lr": 2.0024877714043322e-05, "grad_norm": 0.7760224342346191, "grad_norm_exchange": 0.22447419166564941, "grad_norm_vphi": 0.011345947161316872, "gamma": 0.30000001192092896, "gumbel_tau": 0.4094211459159851, "xi_alphas": [2.330104734937777e-06, 0.5745547413825989, 0.8592441082000732, 0.97092604637146], "elapsed_s": 10734.058781862259, "exchange_scale": -0.3173128366470337} {"step": 14000, "val_loss": 2.25281885266304, "val_ppl": 9.514518095674402} {"step": 14050, "train_loss": 1.8714897203445435, "lr": 1.904938814951268e-05, "grad_norm": 0.7710053324699402, "grad_norm_exchange": 0.21985174715518951, "grad_norm_vphi": 0.009578453376889229, "gamma": 0.30000001192092896, "gumbel_tau": 0.4066869914531708, "xi_alphas": [2.3298182441067183e-06, 0.574529767036438, 0.8592237234115601, 0.9709219932556152], "elapsed_s": 10778.079517126083, "exchange_scale": -0.3174569308757782} {"step": 14100, "train_loss": 1.891333177089691, "lr": 1.809731422350591e-05, "grad_norm": 0.7764813899993896, "grad_norm_exchange": 0.2297854870557785, "grad_norm_vphi": 0.011290659196674824, "gamma": 0.30000001192092896, "gumbel_tau": 0.40395280718803406, "xi_alphas": [2.3297691313928226e-06, 0.5744850039482117, 0.8592163324356079, 0.9709179997444153], "elapsed_s": 10815.810813903809, "exchange_scale": -0.31754618883132935} {"step": 14150, "train_loss": 1.888555235862732, "lr": 1.716875246496796e-05, "grad_norm": 0.7851182818412781, "grad_norm_exchange": 0.21547193825244904, "grad_norm_vphi": 0.010087196715176105, "gamma": 0.30000001192092896, "gumbel_tau": 0.4012186527252197, "xi_alphas": [2.3297784537135158e-06, 0.5744088888168335, 0.8592163324356079, 0.970920979976654], "elapsed_s": 10853.89396739006, "exchange_scale": -0.3175753653049469} {"step": 14200, "train_loss": 1.8824157619476318, "lr": 1.6263797018990306e-05, "grad_norm": 0.749580442905426, "grad_norm_exchange": 0.22305022180080414, "grad_norm_vphi": 0.009506761096417904, "gamma": 0.30000001192092896, "gumbel_tau": 0.3984844982624054, "xi_alphas": [2.3299248823605012e-06, 0.5744484663009644, 0.8592458963394165, 0.9709184169769287], "elapsed_s": 10892.148873329163, "exchange_scale": -0.31760695576667786} {"step": 14250, "train_loss": 1.8749587798118592, "lr": 1.5382539637265825e-05, "grad_norm": 0.7664158344268799, "grad_norm_exchange": 0.21934354305267334, "grad_norm_vphi": 0.010169901885092258, "gamma": 0.30000001192092896, "gumbel_tau": 0.39575034379959106, "xi_alphas": [2.329864855710184e-06, 0.5744010210037231, 0.8592444658279419, 0.9709154963493347], "elapsed_s": 10930.577009916306, "exchange_scale": -0.31753087043762207} {"step": 14300, "train_loss": 1.851404995918274, "lr": 1.4525069668786195e-05, "grad_norm": 0.7797439098358154, "grad_norm_exchange": 0.21682119369506836, "grad_norm_vphi": 0.01039162278175354, "gamma": 0.30000001192092896, "gumbel_tau": 0.39301615953445435, "xi_alphas": [2.3298805444937898e-06, 0.5743927359580994, 0.859238862991333, 0.9709161520004272], "elapsed_s": 10968.834483385086, "exchange_scale": -0.3175695836544037} {"step": 14350, "train_loss": 1.8801818513870239, "lr": 1.369147405078297e-05, "grad_norm": 0.7425593733787537, "grad_norm_exchange": 0.20758478343486786, "grad_norm_vphi": 0.010072180069983006, "gamma": 0.30000001192092896, "gumbel_tau": 0.39028200507164, "xi_alphas": [2.329627250219346e-06, 0.5743739604949951, 0.8592426180839539, 0.9709156155586243], "elapsed_s": 11007.31464767456, "exchange_scale": -0.31767889857292175} {"step": 14400, "train_loss": 1.8919232821464538, "lr": 1.2881837299913267e-05, "grad_norm": 0.788256824016571, "grad_norm_exchange": 0.2194463312625885, "grad_norm_vphi": 0.012772302143275738, "gamma": 0.30000001192092896, "gumbel_tau": 0.3875478506088257, "xi_alphas": [2.329629296582425e-06, 0.5743584036827087, 0.8592377305030823, 0.9709166288375854], "elapsed_s": 11045.51477599144, "exchange_scale": -0.3176770508289337} {"step": 14400, "val_loss": 2.207102435827255, "val_ppl": 9.089341250099086} {"step": 14450, "train_loss": 1.876535003185272, "lr": 1.2096241503690709e-05, "grad_norm": 0.7480401992797852, "grad_norm_exchange": 0.21635009348392487, "grad_norm_vphi": 0.010638589970767498, "gamma": 0.30000001192092896, "gumbel_tau": 0.38481369614601135, "xi_alphas": [2.3298116502701305e-06, 0.574329137802124, 0.8592460751533508, 0.9709193110466003], "elapsed_s": 11089.973759174347, "exchange_scale": -0.31785643100738525} {"step": 14500, "train_loss": 1.8722727394104004, "lr": 1.133476631216271e-05, "grad_norm": 0.7632542848587036, "grad_norm_exchange": 0.2189166396856308, "grad_norm_vphi": 0.009656308218836784, "gamma": 0.30000001192092896, "gumbel_tau": 0.38207951188087463, "xi_alphas": [2.329702738279593e-06, 0.5743270516395569, 0.8592337369918823, 0.9709152579307556], "elapsed_s": 11128.439857006073, "exchange_scale": -0.3178633749485016} {"step": 14550, "train_loss": 1.877097852230072, "lr": 1.0597488929834948e-05, "grad_norm": 0.7721677422523499, "grad_norm_exchange": 0.21922913193702698, "grad_norm_vphi": 0.011461593210697174, "gamma": 0.30000001192092896, "gumbel_tau": 0.3793453574180603, "xi_alphas": [2.3294560378417373e-06, 0.5743036866188049, 0.8592250347137451, 0.9709139466285706], "elapsed_s": 11166.79330945015, "exchange_scale": -0.317973256111145} {"step": 14600, "train_loss": 1.8924759030342102, "lr": 9.884484107843845e-06, "grad_norm": 0.7620969414710999, "grad_norm_exchange": 0.21808677911758423, "grad_norm_vphi": 0.010306835174560547, "gamma": 0.30000001192092896, "gumbel_tau": 0.37661120295524597, "xi_alphas": [2.3293293907045154e-06, 0.5742747783660889, 0.8592097163200378, 0.9709107875823975], "elapsed_s": 11205.286284685135, "exchange_scale": -0.31803667545318604} {"step": 14650, "train_loss": 1.8657232904434204, "lr": 9.195824136377485e-06, "grad_norm": 0.7590709328651428, "grad_norm_exchange": 0.21548394858837128, "grad_norm_vphi": 0.011732647195458412, "gamma": 0.30000001192092896, "gumbel_tau": 0.37387704849243164, "xi_alphas": [2.3295028768188786e-06, 0.5742800831794739, 0.8592092990875244, 0.9709091186523438], "elapsed_s": 11243.493426322937, "exchange_scale": -0.3180314600467682} {"step": 14700, "train_loss": 1.884633550643921, "lr": 8.531578837346371e-06, "grad_norm": 0.7384546399116516, "grad_norm_exchange": 0.21218381822109222, "grad_norm_vphi": 0.008959829807281494, "gamma": 0.30000001192092896, "gumbel_tau": 0.3711428940296173, "xi_alphas": [2.329533799638739e-06, 0.5742578506469727, 0.8592151999473572, 0.9709081053733826], "elapsed_s": 11281.882601261139, "exchange_scale": -0.31811320781707764} {"step": 14750, "train_loss": 1.8988505911827087, "lr": 7.891815557304393e-06, "grad_norm": 0.8336011171340942, "grad_norm_exchange": 0.24383415281772614, "grad_norm_vphi": 0.012667479924857616, "gamma": 0.30000001192092896, "gumbel_tau": 0.3684087097644806, "xi_alphas": [2.329471726625343e-06, 0.5742353796958923, 0.8592135310173035, 0.9709063768386841], "elapsed_s": 11320.2397646904, "exchange_scale": -0.3181079030036926} {"step": 14800, "train_loss": 1.8860522174835206, "lr": 7.276599160620512e-06, "grad_norm": 0.7943414449691772, "grad_norm_exchange": 0.21029730141162872, "grad_norm_vphi": 0.011987827718257904, "gamma": 0.30000001192092896, "gumbel_tau": 0.36567455530166626, "xi_alphas": [2.329405106138438e-06, 0.5742195844650269, 0.8592037558555603, 0.9709069728851318], "elapsed_s": 11358.698645353317, "exchange_scale": -0.3181387186050415} {"step": 14800, "val_loss": 2.21867493391037, "val_ppl": 9.195138623523247} {"step": 14850, "train_loss": 1.880811996459961, "lr": 6.6859920229024044e-06, "grad_norm": 0.7938722372055054, "grad_norm_exchange": 0.22679047286510468, "grad_norm_vphi": 0.011716622859239578, "gamma": 0.30000001192092896, "gumbel_tau": 0.36294040083885193, "xi_alphas": [2.3292848254641285e-06, 0.574215829372406, 0.8592092990875244, 0.9709060788154602], "elapsed_s": 11403.2220993042, "exchange_scale": -0.3181796967983246} {"step": 14900, "train_loss": 1.88514080286026, "lr": 6.1200540246723055e-06, "grad_norm": 0.7853375673294067, "grad_norm_exchange": 0.23594367504119873, "grad_norm_vphi": 0.01340485643595457, "gamma": 0.30000001192092896, "gumbel_tau": 0.3602062463760376, "xi_alphas": [2.3293939648283413e-06, 0.5742073059082031, 0.8592156171798706, 0.9709058403968811], "elapsed_s": 11441.63601231575, "exchange_scale": -0.3182125389575958} {"step": 14950, "train_loss": 1.8523664283752441, "lr": 5.578842545295809e-06, "grad_norm": 0.7820265889167786, "grad_norm_exchange": 0.2208576649427414, "grad_norm_vphi": 0.011329056695103645, "gamma": 0.30000001192092896, "gumbel_tau": 0.3574720621109009, "xi_alphas": [2.3293493995879544e-06, 0.5742105841636658, 0.8592177033424377, 0.9709057211875916], "elapsed_s": 11479.872886896133, "exchange_scale": -0.31822916865348816} {"step": 15000, "train_loss": 1.8819332432746887, "lr": 5.0624124571643225e-06, "grad_norm": 0.7454407215118408, "grad_norm_exchange": 0.2102358341217041, "grad_norm_vphi": 0.0114659508690238, "gamma": 0.30000001192092896, "gumbel_tau": 0.35473790764808655, "xi_alphas": [2.32933166444127e-06, 0.5741971135139465, 0.8592203855514526, 0.9709053635597229], "elapsed_s": 11518.238907814026, "exchange_scale": -0.3182472586631775} {"step": 15050, "train_loss": 1.8657089138031007, "lr": 4.570816120131605e-06, "grad_norm": 0.7896620035171509, "grad_norm_exchange": 0.23555481433868408, "grad_norm_vphi": 0.012024596333503723, "gamma": 0.30000001192092896, "gumbel_tau": 0.3520037531852722, "xi_alphas": [2.3292827791010495e-06, 0.5741836428642273, 0.859222412109375, 0.9709061980247498], "elapsed_s": 11556.396130561829, "exchange_scale": -0.3182663023471832} {"step": 15100, "train_loss": 1.8926805019378663, "lr": 4.104103376205176e-06, "grad_norm": 0.7531525492668152, "grad_norm_exchange": 0.21780133247375488, "grad_norm_vphi": 0.00996685866266489, "gamma": 0.30000001192092896, "gumbel_tau": 0.3492695987224579, "xi_alphas": [2.329407152501517e-06, 0.5741705894470215, 0.8592265248298645, 0.9709064960479736], "elapsed_s": 11593.891069173813, "exchange_scale": -0.31829962134361267} {"step": 15150, "train_loss": 1.8657558345794678, "lr": 3.6623215444928638e-06, "grad_norm": 0.8028180003166199, "grad_norm_exchange": 0.21545548737049103, "grad_norm_vphi": 0.01769321784377098, "gamma": 0.30000001192092896, "gumbel_tau": 0.34653541445732117, "xi_alphas": [2.3294583115784917e-06, 0.5741689205169678, 0.8592231869697571, 0.970905065536499], "elapsed_s": 11631.23337340355, "exchange_scale": -0.3183188736438751} {"step": 15200, "train_loss": 1.8726237630844116, "lr": 3.245515416405248e-06, "grad_norm": 0.7440043091773987, "grad_norm_exchange": 0.22054679691791534, "grad_norm_vphi": 0.008861113339662552, "gamma": 0.30000001192092896, "gumbel_tau": 0.34380125999450684, "xi_alphas": [2.3294428501685616e-06, 0.5741693377494812, 0.8592209219932556, 0.9709040522575378], "elapsed_s": 11668.651556253433, "exchange_scale": -0.31833624839782715} {"step": 15200, "val_loss": 2.239406555891037, "val_ppl": 9.387758524052064} {"step": 15250, "train_loss": 1.8958117151260376, "lr": 2.8537272511142962e-06, "grad_norm": 0.7539320588111877, "grad_norm_exchange": 0.2085154950618744, "grad_norm_vphi": 0.010490478947758675, "gamma": 0.30000001192092896, "gumbel_tau": 0.3410671055316925, "xi_alphas": [2.3294649054150796e-06, 0.5741718411445618, 0.8592191338539124, 0.9709041714668274], "elapsed_s": 11712.089228153229, "exchange_scale": -0.31835123896598816} {"step": 15300, "train_loss": 1.8544205498695374, "lr": 2.486996771268901e-06, "grad_norm": 0.7596100568771362, "grad_norm_exchange": 0.2200525850057602, "grad_norm_vphi": 0.009900560602545738, "gamma": 0.30000001192092896, "gumbel_tau": 0.3383329510688782, "xi_alphas": [2.3294428501685616e-06, 0.5741689801216125, 0.859220027923584, 0.9709042906761169], "elapsed_s": 11749.347510814667, "exchange_scale": -0.31835058331489563} {"step": 15350, "train_loss": 1.8821646904945373, "lr": 2.1453611589673516e-06, "grad_norm": 0.7535692453384399, "grad_norm_exchange": 0.20952856540679932, "grad_norm_vphi": 0.010831538587808609, "gamma": 0.30000001192092896, "gumbel_tau": 0.33559876680374146, "xi_alphas": [2.3294360289582983e-06, 0.5741629600524902, 0.8592196702957153, 0.9709044694900513], "elapsed_s": 11786.757153272629, "exchange_scale": -0.3183608949184418} {"step": 15400, "train_loss": 1.876795084476471, "lr": 1.8288550519875435e-06, "grad_norm": 0.7592943906784058, "grad_norm_exchange": 0.2204495370388031, "grad_norm_vphi": 0.009793960489332676, "gamma": 0.30000001192092896, "gumbel_tau": 0.3328646123409271, "xi_alphas": [2.329387143618078e-06, 0.574158787727356, 0.859220564365387, 0.9709045886993408], "elapsed_s": 11824.055218458176, "exchange_scale": -0.318376749753952} {"step": 15450, "train_loss": 1.8693657636642456, "lr": 1.5375105402752055e-06, "grad_norm": 0.7644607424736023, "grad_norm_exchange": 0.21847327053546906, "grad_norm_vphi": 0.016851741820573807, "gamma": 0.30000001192092896, "gumbel_tau": 0.3301304578781128, "xi_alphas": [2.3293807771551656e-06, 0.574157178401947, 0.8592200875282288, 0.9709041714668274], "elapsed_s": 11861.408723831177, "exchange_scale": -0.3183763325214386} {"step": 15500, "train_loss": 1.878402750492096, "lr": 1.2713571626902233e-06, "grad_norm": 0.7815466523170471, "grad_norm_exchange": 0.2237754911184311, "grad_norm_vphi": 0.011043339036405087, "gamma": 0.30000001192092896, "gumbel_tau": 0.32739630341529846, "xi_alphas": [2.3293539470614633e-06, 0.5741553902626038, 0.859220564365387, 0.9709036946296692], "elapsed_s": 11898.62748670578, "exchange_scale": -0.31839144229888916} {"step": 15550, "train_loss": 1.9036594700813294, "lr": 1.030421904011869e-06, "grad_norm": 0.7827537059783936, "grad_norm_exchange": 0.220911905169487, "grad_norm_vphi": 0.010878103785216808, "gamma": 0.30000001192092896, "gumbel_tau": 0.32466214895248413, "xi_alphas": [2.329345079488121e-06, 0.5741552710533142, 0.8592202067375183, 0.9709038138389587], "elapsed_s": 11936.051411628723, "exchange_scale": -0.3184112012386322} {"step": 15600, "train_loss": 1.899778573513031, "lr": 8.147291922028511e-07, "grad_norm": 0.8195260763168335, "grad_norm_exchange": 0.21538175642490387, "grad_norm_vphi": 0.022201964631676674, "gamma": 0.30000001192092896, "gumbel_tau": 0.3219279646873474, "xi_alphas": [2.3293428057513665e-06, 0.574155867099762, 0.8592202067375183, 0.9709038138389587], "elapsed_s": 11973.303415298462, "exchange_scale": -0.3184111714363098} {"step": 15600, "val_loss": 2.2301635682582854, "val_ppl": 9.30138736679426} {"step": 15650, "train_loss": 1.8779011154174805, "lr": 6.243008959324892e-07, "grad_norm": 0.7575137615203857, "grad_norm_exchange": 0.21488763391971588, "grad_norm_vphi": 0.0099678048864007, "gamma": 0.30000001192092896, "gumbel_tau": 0.3191938102245331, "xi_alphas": [2.3293428057513665e-06, 0.5741565227508545, 0.8592207431793213, 0.9709039330482483], "elapsed_s": 12016.821104049683, "exchange_scale": -0.31842002272605896} {"step": 15700, "train_loss": 1.8813371872901916, "lr": 4.591563223596551e-07, "grad_norm": 0.7457916140556335, "grad_norm_exchange": 0.22173839807510376, "grad_norm_vphi": 0.010596065782010555, "gamma": 0.30000001192092896, "gumbel_tau": 0.31645965576171875, "xi_alphas": [2.3293428057513665e-06, 0.5741565227508545, 0.859220564365387, 0.9709038138389587], "elapsed_s": 12054.060637950897, "exchange_scale": -0.3184206485748291} {"step": 15750, "train_loss": 1.8578398513793946, "lr": 3.193122151751715e-07, "grad_norm": 0.7630135416984558, "grad_norm_exchange": 0.2191162407398224, "grad_norm_vphi": 0.01355243194848299, "gamma": 0.30000001192092896, "gumbel_tau": 0.3137255012989044, "xi_alphas": [2.3293428057513665e-06, 0.5741562843322754, 0.8592197299003601, 0.9709038138389587], "elapsed_s": 12091.505070209503, "exchange_scale": -0.3184228539466858} {"step": 15800, "train_loss": 1.8923493432998657, "lr": 2.047827529041979e-07, "grad_norm": 0.7861016988754272, "grad_norm_exchange": 0.2192036360502243, "grad_norm_vphi": 0.01179142389446497, "gamma": 0.30000001192092896, "gumbel_tau": 0.3109913170337677, "xi_alphas": [2.3293428057513665e-06, 0.5741553902626038, 0.8592198491096497, 0.9709038138389587], "elapsed_s": 12128.714116811752, "exchange_scale": -0.31842318177223206} {"step": 15850, "train_loss": 1.858220829963684, "lr": 1.155795474687693e-07, "grad_norm": 0.7661407589912415, "grad_norm_exchange": 0.2103235423564911, "grad_norm_vphi": 0.013820115476846695, "gamma": 0.30000001192092896, "gumbel_tau": 0.30825716257095337, "xi_alphas": [2.3293428057513665e-06, 0.5741552710533142, 0.8592199087142944, 0.9709038138389587], "elapsed_s": 12166.167088031769, "exchange_scale": -0.31842416524887085} {"step": 15900, "train_loss": 1.8935984325408937, "lr": 5.1711643010321496e-08, "grad_norm": 0.794019877910614, "grad_norm_exchange": 0.22893746197223663, "grad_norm_vphi": 0.010995371267199516, "gamma": 0.30000001192092896, "gumbel_tau": 0.30552300810813904, "xi_alphas": [2.3293428057513665e-06, 0.574155330657959, 0.8592199087142944, 0.9709038138389587], "elapsed_s": 12203.434599637985, "exchange_scale": -0.31842491030693054} {"step": 15950, "train_loss": 1.8489753270149232, "lr": 1.3185514972952062e-08, "grad_norm": 0.7908787131309509, "grad_norm_exchange": 0.22014492750167847, "grad_norm_vphi": 0.009293565526604652, "gamma": 0.30000001192092896, "gumbel_tau": 0.3027888536453247, "xi_alphas": [2.3293428057513665e-06, 0.5741551518440247, 0.8592199087142944, 0.9709038138389587], "elapsed_s": 12240.717810153961, "exchange_scale": -0.31842508912086487} {"step": 16000, "train_loss": 1.88704097032547, "lr": 5.069446695760149e-12, "grad_norm": 0.7663404941558838, "grad_norm_exchange": 0.21298648416996002, "grad_norm_vphi": 0.013847866095602512, "gamma": 0.30000001192092896, "gumbel_tau": 0.300054669380188, "xi_alphas": [2.3293428057513665e-06, 0.5741551518440247, 0.8592199087142944, 0.9709038138389587], "elapsed_s": 12278.131425142288, "exchange_scale": -0.31842508912086487} {"step": 16000, "val_loss": 2.2371300935745237, "val_ppl": 9.366411952087525}