{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.615222974550549, "eval_steps": 500, "global_step": 2800, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0009339248190520663, "grad_norm": 6.638877692627699, "learning_rate": 9.345794392523364e-07, "loss": 9.2917, "step": 1 }, { "epoch": 0.009339248190520663, "grad_norm": 1.1560921335705272, "learning_rate": 9.345794392523365e-06, "loss": 9.0876, "step": 10 }, { "epoch": 0.018678496381041326, "grad_norm": 0.8415132296956432, "learning_rate": 1.869158878504673e-05, "loss": 8.2164, "step": 20 }, { "epoch": 0.02801774457156199, "grad_norm": 0.45381630992958155, "learning_rate": 2.8037383177570094e-05, "loss": 7.5184, "step": 30 }, { "epoch": 0.03735699276208265, "grad_norm": 0.8400636107958425, "learning_rate": 3.738317757009346e-05, "loss": 6.6507, "step": 40 }, { "epoch": 0.046696240952603316, "grad_norm": 0.557696240829066, "learning_rate": 4.672897196261683e-05, "loss": 5.8909, "step": 50 }, { "epoch": 0.05603548914312398, "grad_norm": 0.3971996057467842, "learning_rate": 5.607476635514019e-05, "loss": 5.4127, "step": 60 }, { "epoch": 0.06537473733364464, "grad_norm": 0.2932710540265688, "learning_rate": 6.542056074766355e-05, "loss": 5.0106, "step": 70 }, { "epoch": 0.0747139855241653, "grad_norm": 0.3682690443551033, "learning_rate": 7.476635514018692e-05, "loss": 4.6042, "step": 80 }, { "epoch": 0.08405323371468597, "grad_norm": 0.3132971920011515, "learning_rate": 8.411214953271028e-05, "loss": 4.2031, "step": 90 }, { "epoch": 0.09339248190520663, "grad_norm": 0.6731868159213446, "learning_rate": 9.345794392523365e-05, "loss": 3.9423, "step": 100 }, { "epoch": 0.1027317300957273, "grad_norm": 0.27848867836763197, "learning_rate": 0.000102803738317757, "loss": 3.7157, "step": 110 }, { "epoch": 0.11207097828624796, "grad_norm": 0.24642109032991807, "learning_rate": 0.00011214953271028037, "loss": 3.4516, "step": 120 }, { "epoch": 0.12141022647676862, "grad_norm": 0.25717384664029797, "learning_rate": 0.00012149532710280373, "loss": 3.2167, "step": 130 }, { "epoch": 0.13074947466728928, "grad_norm": 0.20912922668565637, "learning_rate": 0.0001308411214953271, "loss": 3.0237, "step": 140 }, { "epoch": 0.14008872285780993, "grad_norm": 0.15805888388706113, "learning_rate": 0.00014018691588785047, "loss": 2.8529, "step": 150 }, { "epoch": 0.1494279710483306, "grad_norm": 0.23370349497479534, "learning_rate": 0.00014953271028037384, "loss": 2.7078, "step": 160 }, { "epoch": 0.15876721923885126, "grad_norm": 0.1802138633012483, "learning_rate": 0.0001588785046728972, "loss": 2.6115, "step": 170 }, { "epoch": 0.16810646742937194, "grad_norm": 0.13354347610039718, "learning_rate": 0.00016822429906542056, "loss": 2.5309, "step": 180 }, { "epoch": 0.17744571561989259, "grad_norm": 0.09414865188086892, "learning_rate": 0.00017757009345794393, "loss": 2.4452, "step": 190 }, { "epoch": 0.18678496381041326, "grad_norm": 0.08333601554768896, "learning_rate": 0.0001869158878504673, "loss": 2.3832, "step": 200 }, { "epoch": 0.1961242120009339, "grad_norm": 0.15926414699806835, "learning_rate": 0.00019626168224299065, "loss": 2.3492, "step": 210 }, { "epoch": 0.2054634601914546, "grad_norm": 0.09492820761057012, "learning_rate": 0.0001999989254250208, "loss": 2.323, "step": 220 }, { "epoch": 0.21480270838197524, "grad_norm": 0.0801349259356147, "learning_rate": 0.00019999235866155886, "loss": 2.2731, "step": 230 }, { "epoch": 0.22414195657249592, "grad_norm": 0.12210960524693895, "learning_rate": 0.00019997982251228469, "loss": 2.2433, "step": 240 }, { "epoch": 0.23348120476301656, "grad_norm": 3.14289498732125, "learning_rate": 0.00019996131772558666, "loss": 3.2769, "step": 250 }, { "epoch": 0.24282045295353724, "grad_norm": 1.632940983166179, "learning_rate": 0.00019993684540617132, "loss": 4.9343, "step": 260 }, { "epoch": 0.2521597011440579, "grad_norm": 3.4831252230225416, "learning_rate": 0.00019990640701499736, "loss": 4.2768, "step": 270 }, { "epoch": 0.26149894933457857, "grad_norm": 1.6069045920523788, "learning_rate": 0.00019987000436918874, "loss": 5.9581, "step": 280 }, { "epoch": 0.27083819752509924, "grad_norm": 0.2220907936615993, "learning_rate": 0.00019982763964192585, "loss": 3.8228, "step": 290 }, { "epoch": 0.28017744571561987, "grad_norm": 0.24737284913291765, "learning_rate": 0.00019977931536231596, "loss": 3.1413, "step": 300 }, { "epoch": 0.28951669390614054, "grad_norm": 4.010404518241152, "learning_rate": 0.00019972503441524224, "loss": 2.8432, "step": 310 }, { "epoch": 0.2988559420966612, "grad_norm": 0.1515583580811596, "learning_rate": 0.00019966480004119142, "loss": 2.7859, "step": 320 }, { "epoch": 0.3081951902871819, "grad_norm": 0.11259395750650594, "learning_rate": 0.00019959861583606045, "loss": 2.5821, "step": 330 }, { "epoch": 0.3175344384777025, "grad_norm": 0.22514797814956813, "learning_rate": 0.00019952648575094183, "loss": 2.4517, "step": 340 }, { "epoch": 0.3268736866682232, "grad_norm": 0.08040136172033542, "learning_rate": 0.00019944841409188767, "loss": 2.3794, "step": 350 }, { "epoch": 0.3362129348587439, "grad_norm": 0.054758073593565354, "learning_rate": 0.00019936440551965263, "loss": 2.3232, "step": 360 }, { "epoch": 0.34555218304926455, "grad_norm": 0.06742998909645591, "learning_rate": 0.00019927446504941577, "loss": 2.2776, "step": 370 }, { "epoch": 0.35489143123978517, "grad_norm": 0.048780907584876736, "learning_rate": 0.00019917859805048096, "loss": 2.2376, "step": 380 }, { "epoch": 0.36423067943030585, "grad_norm": 0.0475325963052214, "learning_rate": 0.00019907681024595663, "loss": 2.2191, "step": 390 }, { "epoch": 0.3735699276208265, "grad_norm": 0.054089563211590065, "learning_rate": 0.00019896910771241387, "loss": 2.1961, "step": 400 }, { "epoch": 0.3829091758113472, "grad_norm": 0.21798406131864823, "learning_rate": 0.00019885549687952372, "loss": 2.2078, "step": 410 }, { "epoch": 0.3922484240018678, "grad_norm": 0.8673185709111124, "learning_rate": 0.00019873598452967338, "loss": 2.3731, "step": 420 }, { "epoch": 0.4015876721923885, "grad_norm": 0.22424350669971718, "learning_rate": 0.0001986105777975613, "loss": 2.6195, "step": 430 }, { "epoch": 0.4109269203829092, "grad_norm": 0.307418135168262, "learning_rate": 0.00019847928416977126, "loss": 2.3624, "step": 440 }, { "epoch": 0.42026616857342985, "grad_norm": 0.07944722668080402, "learning_rate": 0.00019834211148432536, "loss": 2.2799, "step": 450 }, { "epoch": 0.4296054167639505, "grad_norm": 0.18146933758664588, "learning_rate": 0.00019819906793021614, "loss": 2.2177, "step": 460 }, { "epoch": 0.43894466495447115, "grad_norm": 0.07035825837333018, "learning_rate": 0.0001980501620469178, "loss": 2.1767, "step": 470 }, { "epoch": 0.44828391314499183, "grad_norm": 0.04596186944454228, "learning_rate": 0.0001978954027238763, "loss": 2.1598, "step": 480 }, { "epoch": 0.4576231613355125, "grad_norm": 0.041342347745088055, "learning_rate": 0.0001977347991999786, "loss": 2.131, "step": 490 }, { "epoch": 0.46696240952603313, "grad_norm": 0.04172063219841485, "learning_rate": 0.00019756836106300137, "loss": 2.1231, "step": 500 }, { "epoch": 0.4763016577165538, "grad_norm": 0.03373646457711144, "learning_rate": 0.00019739609824903843, "loss": 2.1146, "step": 510 }, { "epoch": 0.4856409059070745, "grad_norm": 0.03736871030676605, "learning_rate": 0.00019721802104190748, "loss": 2.1003, "step": 520 }, { "epoch": 0.49498015409759516, "grad_norm": 0.033931028038211034, "learning_rate": 0.00019703414007253645, "loss": 2.0983, "step": 530 }, { "epoch": 0.5043194022881158, "grad_norm": 0.03790055446070549, "learning_rate": 0.00019684446631832868, "loss": 2.092, "step": 540 }, { "epoch": 0.5136586504786365, "grad_norm": 0.030956192803893078, "learning_rate": 0.00019664901110250758, "loss": 2.0807, "step": 550 }, { "epoch": 0.5229978986691571, "grad_norm": 0.03542530209935129, "learning_rate": 0.00019644778609344068, "loss": 2.0773, "step": 560 }, { "epoch": 0.5323371468596778, "grad_norm": 0.040947757568902336, "learning_rate": 0.00019624080330394306, "loss": 2.0649, "step": 570 }, { "epoch": 0.5416763950501985, "grad_norm": 0.034273415973688146, "learning_rate": 0.00019602807509056018, "loss": 2.0479, "step": 580 }, { "epoch": 0.5510156432407192, "grad_norm": 0.031427481498873144, "learning_rate": 0.00019580961415283028, "loss": 2.0563, "step": 590 }, { "epoch": 0.5603548914312397, "grad_norm": 0.03141549752041532, "learning_rate": 0.00019558543353252611, "loss": 2.0503, "step": 600 }, { "epoch": 0.5696941396217604, "grad_norm": 0.033012392726428204, "learning_rate": 0.00019535554661287652, "loss": 2.0389, "step": 610 }, { "epoch": 0.5790333878122811, "grad_norm": 0.02913261992661444, "learning_rate": 0.0001951199671177673, "loss": 2.036, "step": 620 }, { "epoch": 0.5883726360028018, "grad_norm": 0.030543903708435332, "learning_rate": 0.00019487870911092214, "loss": 2.0326, "step": 630 }, { "epoch": 0.5977118841933224, "grad_norm": 0.03215005545393897, "learning_rate": 0.00019463178699506277, "loss": 2.0231, "step": 640 }, { "epoch": 0.6070511323838431, "grad_norm": 0.03823630791937631, "learning_rate": 0.00019437921551104933, "loss": 2.0293, "step": 650 }, { "epoch": 0.6163903805743638, "grad_norm": 0.03200103149471209, "learning_rate": 0.00019412100973700038, "loss": 2.017, "step": 660 }, { "epoch": 0.6257296287648845, "grad_norm": 0.03841804153577787, "learning_rate": 0.00019385718508739262, "loss": 2.0135, "step": 670 }, { "epoch": 0.635068876955405, "grad_norm": 0.03052396655271533, "learning_rate": 0.0001935877573121407, "loss": 2.0237, "step": 680 }, { "epoch": 0.6444081251459257, "grad_norm": 0.033817837533771815, "learning_rate": 0.00019331274249565717, "loss": 2.0069, "step": 690 }, { "epoch": 0.6537473733364464, "grad_norm": 0.028286999650643876, "learning_rate": 0.00019303215705589194, "loss": 2.0112, "step": 700 }, { "epoch": 0.6630866215269671, "grad_norm": 0.030693948626257357, "learning_rate": 0.00019274601774335243, "loss": 2.0, "step": 710 }, { "epoch": 0.6724258697174877, "grad_norm": 0.029984615635953022, "learning_rate": 0.0001924543416401035, "loss": 2.0028, "step": 720 }, { "epoch": 0.6817651179080084, "grad_norm": 0.02915985673921391, "learning_rate": 0.00019215714615874755, "loss": 2.0031, "step": 730 }, { "epoch": 0.6911043660985291, "grad_norm": 0.028305399777245336, "learning_rate": 0.00019185444904138528, "loss": 1.9924, "step": 740 }, { "epoch": 0.7004436142890498, "grad_norm": 0.036720505429756495, "learning_rate": 0.00019154626835855628, "loss": 1.9981, "step": 750 }, { "epoch": 0.7097828624795703, "grad_norm": 0.0287124048917296, "learning_rate": 0.00019123262250816034, "loss": 1.9868, "step": 760 }, { "epoch": 0.719122110670091, "grad_norm": 0.03318092492837997, "learning_rate": 0.00019091353021435915, "loss": 1.9943, "step": 770 }, { "epoch": 0.7284613588606117, "grad_norm": 0.054035272137015325, "learning_rate": 0.00019058901052645844, "loss": 1.9838, "step": 780 }, { "epoch": 0.7378006070511324, "grad_norm": 0.03184392761983255, "learning_rate": 0.00019025908281777078, "loss": 1.982, "step": 790 }, { "epoch": 0.747139855241653, "grad_norm": 0.029409948164434735, "learning_rate": 0.00018992376678445908, "loss": 1.9693, "step": 800 }, { "epoch": 0.7564791034321737, "grad_norm": 0.029656963043919016, "learning_rate": 0.00018958308244436064, "loss": 1.9914, "step": 810 }, { "epoch": 0.7658183516226944, "grad_norm": 0.030843610865326686, "learning_rate": 0.00018923705013579233, "loss": 1.9749, "step": 820 }, { "epoch": 0.7751575998132151, "grad_norm": 0.0377584286045999, "learning_rate": 0.00018888569051633613, "loss": 1.9606, "step": 830 }, { "epoch": 0.7844968480037356, "grad_norm": 0.03147699916274391, "learning_rate": 0.00018852902456160616, "loss": 1.9696, "step": 840 }, { "epoch": 0.7938360961942563, "grad_norm": 0.029193153251471263, "learning_rate": 0.0001881670735639963, "loss": 1.9687, "step": 850 }, { "epoch": 0.803175344384777, "grad_norm": 0.030201895228156087, "learning_rate": 0.00018779985913140924, "loss": 1.9678, "step": 860 }, { "epoch": 0.8125145925752977, "grad_norm": 0.029272319485493213, "learning_rate": 0.00018742740318596632, "loss": 1.9697, "step": 870 }, { "epoch": 0.8218538407658184, "grad_norm": 0.033740176465285654, "learning_rate": 0.000187049727962699, "loss": 1.9647, "step": 880 }, { "epoch": 0.831193088956339, "grad_norm": 0.029539399251208593, "learning_rate": 0.0001866668560082213, "loss": 1.9627, "step": 890 }, { "epoch": 0.8405323371468597, "grad_norm": 0.028666077337951026, "learning_rate": 0.0001862788101793839, "loss": 1.9529, "step": 900 }, { "epoch": 0.8498715853373804, "grad_norm": 0.030280792492665805, "learning_rate": 0.0001858856136419097, "loss": 1.9536, "step": 910 }, { "epoch": 0.859210833527901, "grad_norm": 0.03921536105057096, "learning_rate": 0.0001854872898690106, "loss": 1.9474, "step": 920 }, { "epoch": 0.8685500817184216, "grad_norm": 0.030632523637038354, "learning_rate": 0.0001850838626399865, "loss": 1.9423, "step": 930 }, { "epoch": 0.8778893299089423, "grad_norm": 0.04615147601979514, "learning_rate": 0.00018467535603880548, "loss": 1.946, "step": 940 }, { "epoch": 0.887228578099463, "grad_norm": 0.028216236017006333, "learning_rate": 0.00018426179445266616, "loss": 1.9408, "step": 950 }, { "epoch": 0.8965678262899837, "grad_norm": 0.0282407563402959, "learning_rate": 0.00018384320257054177, "loss": 1.9447, "step": 960 }, { "epoch": 0.9059070744805043, "grad_norm": 0.029365880854687894, "learning_rate": 0.0001834196053817062, "loss": 1.9389, "step": 970 }, { "epoch": 0.915246322671025, "grad_norm": 0.02855404439130719, "learning_rate": 0.00018299102817424234, "loss": 1.9425, "step": 980 }, { "epoch": 0.9245855708615457, "grad_norm": 0.03238310984070135, "learning_rate": 0.00018255749653353225, "loss": 1.9392, "step": 990 }, { "epoch": 0.9339248190520663, "grad_norm": 0.030115350805430388, "learning_rate": 0.00018211903634072983, "loss": 1.956, "step": 1000 }, { "epoch": 0.9432640672425869, "grad_norm": 0.039090564032501135, "learning_rate": 0.0001816756737712158, "loss": 1.9358, "step": 1010 }, { "epoch": 0.9526033154331076, "grad_norm": 0.028567954476327023, "learning_rate": 0.000181227435293035, "loss": 1.9342, "step": 1020 }, { "epoch": 0.9619425636236283, "grad_norm": 0.028158653241284505, "learning_rate": 0.00018077434766531624, "loss": 1.9287, "step": 1030 }, { "epoch": 0.971281811814149, "grad_norm": 0.02927795717651538, "learning_rate": 0.00018031643793667504, "loss": 1.9298, "step": 1040 }, { "epoch": 0.9806210600046696, "grad_norm": 0.06298340584032344, "learning_rate": 0.0001798537334435986, "loss": 1.9303, "step": 1050 }, { "epoch": 0.9899603081951903, "grad_norm": 0.03627278406983255, "learning_rate": 0.00017938626180881407, "loss": 1.9285, "step": 1060 }, { "epoch": 0.999299556385711, "grad_norm": 0.03916823421329747, "learning_rate": 0.00017891405093963938, "loss": 1.9239, "step": 1070 }, { "epoch": 1.0087555451786132, "grad_norm": 0.03383539251115568, "learning_rate": 0.00017843712902631723, "loss": 1.8855, "step": 1080 }, { "epoch": 1.0180947933691338, "grad_norm": 0.02931021936560147, "learning_rate": 0.00017795552454033224, "loss": 1.8004, "step": 1090 }, { "epoch": 1.0274340415596543, "grad_norm": 0.029173295095350292, "learning_rate": 0.0001774692662327113, "loss": 1.7912, "step": 1100 }, { "epoch": 1.0367732897501751, "grad_norm": 0.035210065642974735, "learning_rate": 0.000176978383132307, "loss": 1.7902, "step": 1110 }, { "epoch": 1.0461125379406957, "grad_norm": 0.029794447659573477, "learning_rate": 0.00017648290454406475, "loss": 1.8072, "step": 1120 }, { "epoch": 1.0554517861312165, "grad_norm": 0.03194584938279939, "learning_rate": 0.0001759828600472734, "loss": 1.803, "step": 1130 }, { "epoch": 1.064791034321737, "grad_norm": 0.031028415312581603, "learning_rate": 0.00017547827949379924, "loss": 1.7945, "step": 1140 }, { "epoch": 1.0741302825122578, "grad_norm": 0.03313245828751237, "learning_rate": 0.00017496919300630403, "loss": 1.8139, "step": 1150 }, { "epoch": 1.0834695307027784, "grad_norm": 0.03049431724979126, "learning_rate": 0.00017445563097644664, "loss": 1.8031, "step": 1160 }, { "epoch": 1.0928087788932992, "grad_norm": 0.02880548705343715, "learning_rate": 0.00017393762406306878, "loss": 1.7974, "step": 1170 }, { "epoch": 1.1021480270838198, "grad_norm": 0.03200427895977668, "learning_rate": 0.00017341520319036469, "loss": 1.7994, "step": 1180 }, { "epoch": 1.1114872752743403, "grad_norm": 0.031293532815600045, "learning_rate": 0.00017288839954603496, "loss": 1.8073, "step": 1190 }, { "epoch": 1.1208265234648611, "grad_norm": 0.032367211109345505, "learning_rate": 0.00017235724457942468, "loss": 1.7944, "step": 1200 }, { "epoch": 1.1301657716553817, "grad_norm": 0.037783793923191374, "learning_rate": 0.0001718217699996462, "loss": 1.7948, "step": 1210 }, { "epoch": 1.1395050198459025, "grad_norm": 0.02881083117349317, "learning_rate": 0.00017128200777368567, "loss": 1.8029, "step": 1220 }, { "epoch": 1.148844268036423, "grad_norm": 0.032997408408175985, "learning_rate": 0.00017073799012449524, "loss": 1.7914, "step": 1230 }, { "epoch": 1.1581835162269438, "grad_norm": 0.030263615801190885, "learning_rate": 0.00017018974952906884, "loss": 1.792, "step": 1240 }, { "epoch": 1.1675227644174644, "grad_norm": 0.03286310429098484, "learning_rate": 0.00016963731871650378, "loss": 1.8149, "step": 1250 }, { "epoch": 1.1768620126079852, "grad_norm": 0.031111621587597126, "learning_rate": 0.00016908073066604663, "loss": 1.8092, "step": 1260 }, { "epoch": 1.1862012607985057, "grad_norm": 0.03372950172075241, "learning_rate": 0.0001685200186051246, "loss": 1.818, "step": 1270 }, { "epoch": 1.1955405089890263, "grad_norm": 0.040782040436567434, "learning_rate": 0.00016795521600736164, "loss": 1.7999, "step": 1280 }, { "epoch": 1.204879757179547, "grad_norm": 0.03451605168178924, "learning_rate": 0.00016738635659058044, "loss": 1.7945, "step": 1290 }, { "epoch": 1.2142190053700677, "grad_norm": 0.03235681588882673, "learning_rate": 0.00016681347431478933, "loss": 1.8087, "step": 1300 }, { "epoch": 1.2235582535605884, "grad_norm": 0.030750745605971932, "learning_rate": 0.00016623660338015487, "loss": 1.7995, "step": 1310 }, { "epoch": 1.232897501751109, "grad_norm": 0.029444668665577274, "learning_rate": 0.00016565577822496042, "loss": 1.8025, "step": 1320 }, { "epoch": 1.2422367499416298, "grad_norm": 0.038528856709584745, "learning_rate": 0.00016507103352354996, "loss": 1.7954, "step": 1330 }, { "epoch": 1.2515759981321504, "grad_norm": 0.034217088004383035, "learning_rate": 0.00016448240418425814, "loss": 1.7962, "step": 1340 }, { "epoch": 1.260915246322671, "grad_norm": 0.030205405393195585, "learning_rate": 0.00016388992534732645, "loss": 1.7973, "step": 1350 }, { "epoch": 1.2702544945131917, "grad_norm": 0.029082218516562994, "learning_rate": 0.00016329363238280528, "loss": 1.796, "step": 1360 }, { "epoch": 1.2795937427037123, "grad_norm": 0.029003887688766505, "learning_rate": 0.00016269356088844238, "loss": 1.7946, "step": 1370 }, { "epoch": 1.288932990894233, "grad_norm": 0.03341157363649238, "learning_rate": 0.00016208974668755779, "loss": 1.7972, "step": 1380 }, { "epoch": 1.2982722390847536, "grad_norm": 0.030614480844663026, "learning_rate": 0.00016148222582690517, "loss": 1.7973, "step": 1390 }, { "epoch": 1.3076114872752744, "grad_norm": 0.029741346740467405, "learning_rate": 0.00016087103457452, "loss": 1.8076, "step": 1400 }, { "epoch": 1.316950735465795, "grad_norm": 0.029569313554185597, "learning_rate": 0.00016025620941755424, "loss": 1.8043, "step": 1410 }, { "epoch": 1.3262899836563156, "grad_norm": 0.02947637404374054, "learning_rate": 0.0001596377870600983, "loss": 1.797, "step": 1420 }, { "epoch": 1.3356292318468364, "grad_norm": 0.031005062093959545, "learning_rate": 0.00015901580442098968, "loss": 1.8086, "step": 1430 }, { "epoch": 1.344968480037357, "grad_norm": 0.029493792984873927, "learning_rate": 0.00015839029863160922, "loss": 1.8026, "step": 1440 }, { "epoch": 1.3543077282278777, "grad_norm": 0.0288068155951218, "learning_rate": 0.0001577613070336641, "loss": 1.7951, "step": 1450 }, { "epoch": 1.3636469764183983, "grad_norm": 0.03380404824627639, "learning_rate": 0.00015712886717695885, "loss": 1.7938, "step": 1460 }, { "epoch": 1.372986224608919, "grad_norm": 0.039744706189693335, "learning_rate": 0.0001564930168171536, "loss": 1.8016, "step": 1470 }, { "epoch": 1.3823254727994396, "grad_norm": 0.030565530594285437, "learning_rate": 0.00015585379391351012, "loss": 1.7984, "step": 1480 }, { "epoch": 1.3916647209899602, "grad_norm": 0.04009392805554255, "learning_rate": 0.00015521123662662567, "loss": 1.7999, "step": 1490 }, { "epoch": 1.401003969180481, "grad_norm": 0.03516196009586836, "learning_rate": 0.000154565383316155, "loss": 1.7979, "step": 1500 }, { "epoch": 1.4103432173710018, "grad_norm": 0.03534161399054556, "learning_rate": 0.0001539162725385202, "loss": 1.8057, "step": 1510 }, { "epoch": 1.4196824655615223, "grad_norm": 0.028488879438601067, "learning_rate": 0.000153263943044609, "loss": 1.792, "step": 1520 }, { "epoch": 1.429021713752043, "grad_norm": 0.03125154490954804, "learning_rate": 0.00015260843377746147, "loss": 1.8008, "step": 1530 }, { "epoch": 1.4383609619425637, "grad_norm": 0.030194357488801882, "learning_rate": 0.00015194978386994507, "loss": 1.7948, "step": 1540 }, { "epoch": 1.4477002101330843, "grad_norm": 0.03049246845786265, "learning_rate": 0.00015128803264241852, "loss": 1.7967, "step": 1550 }, { "epoch": 1.4570394583236048, "grad_norm": 0.030497211097258083, "learning_rate": 0.0001506232196003844, "loss": 1.7894, "step": 1560 }, { "epoch": 1.4663787065141256, "grad_norm": 0.028748806119737205, "learning_rate": 0.00014995538443213094, "loss": 1.806, "step": 1570 }, { "epoch": 1.4757179547046464, "grad_norm": 0.036423750322912396, "learning_rate": 0.00014928456700636237, "loss": 1.7995, "step": 1580 }, { "epoch": 1.485057202895167, "grad_norm": 0.039101516109204065, "learning_rate": 0.00014861080736981906, "loss": 1.8028, "step": 1590 }, { "epoch": 1.4943964510856875, "grad_norm": 0.031368399541673815, "learning_rate": 0.00014793414574488663, "loss": 1.8005, "step": 1600 }, { "epoch": 1.5037356992762083, "grad_norm": 0.029788484702512056, "learning_rate": 0.00014725462252719495, "loss": 1.7963, "step": 1610 }, { "epoch": 1.5130749474667289, "grad_norm": 0.029719041811636312, "learning_rate": 0.00014657227828320635, "loss": 1.7957, "step": 1620 }, { "epoch": 1.5224141956572494, "grad_norm": 0.02820041575417432, "learning_rate": 0.00014588715374779407, "loss": 1.7986, "step": 1630 }, { "epoch": 1.5317534438477702, "grad_norm": 1.8322544897261024, "learning_rate": 0.0001451992898218102, "loss": 2.016, "step": 1640 }, { "epoch": 1.541092692038291, "grad_norm": 0.8676995793107466, "learning_rate": 0.0001445087275696443, "loss": 2.1095, "step": 1650 }, { "epoch": 1.5504319402288116, "grad_norm": 0.08688193434631736, "learning_rate": 0.00014381550821677155, "loss": 2.0497, "step": 1660 }, { "epoch": 1.5597711884193322, "grad_norm": 0.18321381224589608, "learning_rate": 0.0001431196731472921, "loss": 1.9023, "step": 1670 }, { "epoch": 1.569110436609853, "grad_norm": 0.058214343698110564, "learning_rate": 0.00014242126390145998, "loss": 1.8566, "step": 1680 }, { "epoch": 1.5784496848003737, "grad_norm": 0.03965578147557666, "learning_rate": 0.0001417203221732036, "loss": 1.8206, "step": 1690 }, { "epoch": 1.587788932990894, "grad_norm": 0.03131802880017099, "learning_rate": 0.00014101688980763658, "loss": 1.8272, "step": 1700 }, { "epoch": 1.5971281811814149, "grad_norm": 0.02788722706683908, "learning_rate": 0.00014031100879855968, "loss": 1.8145, "step": 1710 }, { "epoch": 1.6064674293719357, "grad_norm": 0.02781057130092059, "learning_rate": 0.00013960272128595372, "loss": 1.8122, "step": 1720 }, { "epoch": 1.6158066775624562, "grad_norm": 0.032067383734154756, "learning_rate": 0.00013889206955346403, "loss": 1.8064, "step": 1730 }, { "epoch": 1.6251459257529768, "grad_norm": 0.03448807133884858, "learning_rate": 0.00013817909602587613, "loss": 1.8128, "step": 1740 }, { "epoch": 1.6344851739434976, "grad_norm": 0.04127395288271696, "learning_rate": 0.00013746384326658305, "loss": 1.8041, "step": 1750 }, { "epoch": 1.6438244221340184, "grad_norm": 0.030243192787820814, "learning_rate": 0.00013674635397504427, "loss": 1.803, "step": 1760 }, { "epoch": 1.653163670324539, "grad_norm": 0.032183525625428915, "learning_rate": 0.00013602667098423687, "loss": 1.8092, "step": 1770 }, { "epoch": 1.6625029185150595, "grad_norm": 0.030391044676815377, "learning_rate": 0.00013530483725809818, "loss": 1.8039, "step": 1780 }, { "epoch": 1.6718421667055803, "grad_norm": 0.03204756818238517, "learning_rate": 0.000134580895888961, "loss": 1.8017, "step": 1790 }, { "epoch": 1.6811814148961008, "grad_norm": 0.031117112662107062, "learning_rate": 0.00013385489009498124, "loss": 1.8017, "step": 1800 }, { "epoch": 1.6905206630866214, "grad_norm": 0.028389293445401805, "learning_rate": 0.00013312686321755761, "loss": 1.811, "step": 1810 }, { "epoch": 1.6998599112771422, "grad_norm": 4.908185097372493, "learning_rate": 0.0001323968587187443, "loss": 2.194, "step": 1820 }, { "epoch": 1.709199159467663, "grad_norm": 0.2849452041194025, "learning_rate": 0.00013166492017865637, "loss": 2.0785, "step": 1830 }, { "epoch": 1.7185384076581836, "grad_norm": 0.10989252058989733, "learning_rate": 0.0001309310912928682, "loss": 1.986, "step": 1840 }, { "epoch": 1.7278776558487041, "grad_norm": 0.057475612656740484, "learning_rate": 0.00013019541586980463, "loss": 1.8614, "step": 1850 }, { "epoch": 1.737216904039225, "grad_norm": 0.034908181734207726, "learning_rate": 0.000129457937828126, "loss": 1.8326, "step": 1860 }, { "epoch": 1.7465561522297455, "grad_norm": 0.02892836681897248, "learning_rate": 0.00012871870119410614, "loss": 1.8243, "step": 1870 }, { "epoch": 1.755895400420266, "grad_norm": 0.03311636981729384, "learning_rate": 0.00012797775009900397, "loss": 1.8183, "step": 1880 }, { "epoch": 1.7652346486107868, "grad_norm": 0.03151917952358458, "learning_rate": 0.00012723512877642904, "loss": 1.8034, "step": 1890 }, { "epoch": 1.7745738968013076, "grad_norm": 0.028109921832296925, "learning_rate": 0.000126490881559701, "loss": 1.8129, "step": 1900 }, { "epoch": 1.7839131449918282, "grad_norm": 0.030350462454962698, "learning_rate": 0.00012574505287920259, "loss": 1.8003, "step": 1910 }, { "epoch": 1.7932523931823487, "grad_norm": 0.03131380630103849, "learning_rate": 0.00012499768725972754, "loss": 1.814, "step": 1920 }, { "epoch": 1.8025916413728695, "grad_norm": 0.029450198273050322, "learning_rate": 0.00012424882931782243, "loss": 1.7998, "step": 1930 }, { "epoch": 1.81193088956339, "grad_norm": 0.0310524261453681, "learning_rate": 0.0001234985237591231, "loss": 1.8078, "step": 1940 }, { "epoch": 1.8212701377539107, "grad_norm": 0.029362038478982822, "learning_rate": 0.00012274681537568585, "loss": 1.8014, "step": 1950 }, { "epoch": 1.8306093859444315, "grad_norm": 0.027166816460226118, "learning_rate": 0.00012199374904331337, "loss": 1.8021, "step": 1960 }, { "epoch": 1.8399486341349522, "grad_norm": 0.03680509283276228, "learning_rate": 0.00012123936971887578, "loss": 1.7973, "step": 1970 }, { "epoch": 1.8492878823254728, "grad_norm": 0.03135174840346185, "learning_rate": 0.0001204837224376267, "loss": 1.7874, "step": 1980 }, { "epoch": 1.8586271305159934, "grad_norm": 0.02644533256389969, "learning_rate": 0.0001197268523105148, "loss": 1.798, "step": 1990 }, { "epoch": 1.8679663787065142, "grad_norm": 0.02999453651649614, "learning_rate": 0.00011896880452149077, "loss": 1.7957, "step": 2000 }, { "epoch": 1.877305626897035, "grad_norm": 0.026905209700322272, "learning_rate": 0.00011820962432480985, "loss": 1.793, "step": 2010 }, { "epoch": 1.8866448750875553, "grad_norm": 0.027263640323285022, "learning_rate": 0.00011744935704233005, "loss": 1.7974, "step": 2020 }, { "epoch": 1.895984123278076, "grad_norm": 0.030479226063932337, "learning_rate": 0.00011668804806080693, "loss": 1.7898, "step": 2030 }, { "epoch": 1.9053233714685969, "grad_norm": 0.030129902025534238, "learning_rate": 0.00011592574282918369, "loss": 1.7856, "step": 2040 }, { "epoch": 1.9146626196591174, "grad_norm": 0.027884976674153635, "learning_rate": 0.00011516248685587814, "loss": 1.7858, "step": 2050 }, { "epoch": 1.924001867849638, "grad_norm": 0.02925266011156687, "learning_rate": 0.00011439832570606586, "loss": 1.7876, "step": 2060 }, { "epoch": 1.9333411160401588, "grad_norm": 0.028472914828616754, "learning_rate": 0.00011363330499895997, "loss": 1.7834, "step": 2070 }, { "epoch": 1.9426803642306796, "grad_norm": 0.025877740032137875, "learning_rate": 0.00011286747040508789, "loss": 1.7955, "step": 2080 }, { "epoch": 1.9520196124212001, "grad_norm": 0.02605295620697312, "learning_rate": 0.0001121008676435648, "loss": 1.7877, "step": 2090 }, { "epoch": 1.9613588606117207, "grad_norm": 0.026887649929567867, "learning_rate": 0.00011133354247936423, "loss": 1.773, "step": 2100 }, { "epoch": 1.9706981088022415, "grad_norm": 0.027982045915154026, "learning_rate": 0.00011056554072058596, "loss": 1.7762, "step": 2110 }, { "epoch": 1.980037356992762, "grad_norm": 0.028077027059053006, "learning_rate": 0.0001097969082157215, "loss": 1.7963, "step": 2120 }, { "epoch": 1.9893766051832826, "grad_norm": 0.027877078975954036, "learning_rate": 0.00010902769085091686, "loss": 1.7787, "step": 2130 }, { "epoch": 1.9987158533738034, "grad_norm": 0.026120077388738373, "learning_rate": 0.00010825793454723325, "loss": 1.7842, "step": 2140 }, { "epoch": 2.0081718421667056, "grad_norm": 0.044810283079268924, "learning_rate": 0.00010748768525790569, "loss": 1.6591, "step": 2150 }, { "epoch": 2.0175110903572264, "grad_norm": 0.03431848280739808, "learning_rate": 0.00010671698896559968, "loss": 1.5599, "step": 2160 }, { "epoch": 2.0268503385477468, "grad_norm": 0.04234332973849956, "learning_rate": 0.00010594589167966606, "loss": 1.5494, "step": 2170 }, { "epoch": 2.0361895867382676, "grad_norm": 0.03260321438171042, "learning_rate": 0.00010517443943339438, "loss": 1.5473, "step": 2180 }, { "epoch": 2.0455288349287883, "grad_norm": 0.034116901745609114, "learning_rate": 0.00010440267828126478, "loss": 1.5464, "step": 2190 }, { "epoch": 2.0548680831193087, "grad_norm": 0.030992757239375807, "learning_rate": 0.00010363065429619858, "loss": 1.5514, "step": 2200 }, { "epoch": 2.0642073313098295, "grad_norm": 0.03365516197786113, "learning_rate": 0.0001028584135668077, "loss": 1.5493, "step": 2210 }, { "epoch": 2.0735465795003503, "grad_norm": 0.033293307482261586, "learning_rate": 0.00010208600219464355, "loss": 1.5426, "step": 2220 }, { "epoch": 2.082885827690871, "grad_norm": 0.03653097737467338, "learning_rate": 0.00010131346629144451, "loss": 1.5471, "step": 2230 }, { "epoch": 2.0922250758813914, "grad_norm": 0.03390291511697895, "learning_rate": 0.0001005408519763833, "loss": 1.5568, "step": 2240 }, { "epoch": 2.101564324071912, "grad_norm": 0.03192694661852283, "learning_rate": 9.976820537331374e-05, "loss": 1.5452, "step": 2250 }, { "epoch": 2.110903572262433, "grad_norm": 0.03561740193515691, "learning_rate": 9.899557260801707e-05, "loss": 1.546, "step": 2260 }, { "epoch": 2.1202428204529538, "grad_norm": 0.029803732953068658, "learning_rate": 9.822299980544862e-05, "loss": 1.5533, "step": 2270 }, { "epoch": 2.129582068643474, "grad_norm": 0.031232417271289125, "learning_rate": 9.745053308698392e-05, "loss": 1.5469, "step": 2280 }, { "epoch": 2.138921316833995, "grad_norm": 0.032434793780181034, "learning_rate": 9.667821856766548e-05, "loss": 1.5514, "step": 2290 }, { "epoch": 2.1482605650245157, "grad_norm": 0.03579370906405582, "learning_rate": 9.590610235344972e-05, "loss": 1.5577, "step": 2300 }, { "epoch": 2.157599813215036, "grad_norm": 0.029662202478648328, "learning_rate": 9.51342305384546e-05, "loss": 1.5543, "step": 2310 }, { "epoch": 2.166939061405557, "grad_norm": 0.03178715913934592, "learning_rate": 9.436264920220781e-05, "loss": 1.5579, "step": 2320 }, { "epoch": 2.1762783095960776, "grad_norm": 0.03384008887051677, "learning_rate": 9.359140440689601e-05, "loss": 1.5595, "step": 2330 }, { "epoch": 2.1856175577865984, "grad_norm": 0.03316450664408166, "learning_rate": 9.282054219461475e-05, "loss": 1.5556, "step": 2340 }, { "epoch": 2.1949568059771187, "grad_norm": 0.032176305552558876, "learning_rate": 9.205010858462007e-05, "loss": 1.5638, "step": 2350 }, { "epoch": 2.2042960541676395, "grad_norm": 0.031134335256756362, "learning_rate": 9.128014957058109e-05, "loss": 1.5629, "step": 2360 }, { "epoch": 2.2136353023581603, "grad_norm": 0.032205851810441756, "learning_rate": 9.051071111783436e-05, "loss": 1.5613, "step": 2370 }, { "epoch": 2.2229745505486807, "grad_norm": 0.029048245476020442, "learning_rate": 8.974183916063968e-05, "loss": 1.5594, "step": 2380 }, { "epoch": 2.2323137987392014, "grad_norm": 0.03183166054573621, "learning_rate": 8.897357959943795e-05, "loss": 1.5606, "step": 2390 }, { "epoch": 2.2416530469297222, "grad_norm": 0.032309922117136916, "learning_rate": 8.820597829811109e-05, "loss": 1.5524, "step": 2400 }, { "epoch": 2.250992295120243, "grad_norm": 0.03598922231958808, "learning_rate": 8.743908108124388e-05, "loss": 1.5604, "step": 2410 }, { "epoch": 2.2603315433107634, "grad_norm": 0.03046424735786346, "learning_rate": 8.667293373138835e-05, "loss": 1.5598, "step": 2420 }, { "epoch": 2.269670791501284, "grad_norm": 0.030995453538377543, "learning_rate": 8.59075819863307e-05, "loss": 1.5652, "step": 2430 }, { "epoch": 2.279010039691805, "grad_norm": 0.029309020623010097, "learning_rate": 8.514307153636077e-05, "loss": 1.5651, "step": 2440 }, { "epoch": 2.2883492878823253, "grad_norm": 0.03158721106736763, "learning_rate": 8.437944802154434e-05, "loss": 1.5581, "step": 2450 }, { "epoch": 2.297688536072846, "grad_norm": 0.03168229084049938, "learning_rate": 8.361675702899871e-05, "loss": 1.5671, "step": 2460 }, { "epoch": 2.307027784263367, "grad_norm": 0.031335222148495136, "learning_rate": 8.2855044090171e-05, "loss": 1.5675, "step": 2470 }, { "epoch": 2.3163670324538876, "grad_norm": 0.031073847325941303, "learning_rate": 8.209435467811998e-05, "loss": 1.5624, "step": 2480 }, { "epoch": 2.325706280644408, "grad_norm": 0.030099100631045844, "learning_rate": 8.133473420480161e-05, "loss": 1.5606, "step": 2490 }, { "epoch": 2.3350455288349288, "grad_norm": 0.034147632950361176, "learning_rate": 8.057622801835788e-05, "loss": 1.5703, "step": 2500 }, { "epoch": 2.3443847770254496, "grad_norm": 0.03051580784550685, "learning_rate": 7.981888140040955e-05, "loss": 1.5731, "step": 2510 }, { "epoch": 2.3537240252159704, "grad_norm": 0.03068917065832597, "learning_rate": 7.9062739563353e-05, "loss": 1.5723, "step": 2520 }, { "epoch": 2.3630632734064907, "grad_norm": 0.02899547641705554, "learning_rate": 7.830784764766118e-05, "loss": 1.5691, "step": 2530 }, { "epoch": 2.3724025215970115, "grad_norm": 0.030965383166701443, "learning_rate": 7.755425071918858e-05, "loss": 1.5627, "step": 2540 }, { "epoch": 2.3817417697875323, "grad_norm": 0.03252440018336625, "learning_rate": 7.680199376648108e-05, "loss": 1.5536, "step": 2550 }, { "epoch": 2.3910810179780526, "grad_norm": 0.031720485449340044, "learning_rate": 7.605112169809008e-05, "loss": 1.5617, "step": 2560 }, { "epoch": 2.4004202661685734, "grad_norm": 0.031796658969132544, "learning_rate": 7.530167933989161e-05, "loss": 1.5595, "step": 2570 }, { "epoch": 2.409759514359094, "grad_norm": 0.03218288097429844, "learning_rate": 7.45537114324102e-05, "loss": 1.5628, "step": 2580 }, { "epoch": 2.4190987625496145, "grad_norm": 0.0305713183075559, "learning_rate": 7.380726262814814e-05, "loss": 1.5717, "step": 2590 }, { "epoch": 2.4284380107401353, "grad_norm": 0.06879342166341705, "learning_rate": 7.30623774889195e-05, "loss": 1.5726, "step": 2600 }, { "epoch": 2.437777258930656, "grad_norm": 0.04101428600237338, "learning_rate": 7.231910048319011e-05, "loss": 1.5679, "step": 2610 }, { "epoch": 2.447116507121177, "grad_norm": 0.031060002638443395, "learning_rate": 7.157747598342274e-05, "loss": 1.562, "step": 2620 }, { "epoch": 2.4564557553116972, "grad_norm": 0.032302829437386466, "learning_rate": 7.083754826342816e-05, "loss": 1.5767, "step": 2630 }, { "epoch": 2.465795003502218, "grad_norm": 0.03111462744196413, "learning_rate": 7.009936149572205e-05, "loss": 1.5672, "step": 2640 }, { "epoch": 2.475134251692739, "grad_norm": 0.031134083521743777, "learning_rate": 6.936295974888807e-05, "loss": 1.5665, "step": 2650 }, { "epoch": 2.4844734998832596, "grad_norm": 0.030961556373721985, "learning_rate": 6.862838698494693e-05, "loss": 1.5608, "step": 2660 }, { "epoch": 2.49381274807378, "grad_norm": 0.03168121700432082, "learning_rate": 6.789568705673183e-05, "loss": 1.566, "step": 2670 }, { "epoch": 2.5031519962643007, "grad_norm": 0.030850372541726772, "learning_rate": 6.716490370527081e-05, "loss": 1.5651, "step": 2680 }, { "epoch": 2.5124912444548215, "grad_norm": 0.03076635908430861, "learning_rate": 6.643608055717519e-05, "loss": 1.5596, "step": 2690 }, { "epoch": 2.521830492645342, "grad_norm": 0.031897253741779714, "learning_rate": 6.570926112203528e-05, "loss": 1.5716, "step": 2700 }, { "epoch": 2.5311697408358627, "grad_norm": 0.03085546721246857, "learning_rate": 6.498448878982291e-05, "loss": 1.5647, "step": 2710 }, { "epoch": 2.5405089890263834, "grad_norm": 0.03127518794548787, "learning_rate": 6.426180682830107e-05, "loss": 1.5573, "step": 2720 }, { "epoch": 2.549848237216904, "grad_norm": 0.03196649247686066, "learning_rate": 6.354125838044098e-05, "loss": 1.5597, "step": 2730 }, { "epoch": 2.5591874854074246, "grad_norm": 0.030359755432035333, "learning_rate": 6.282288646184638e-05, "loss": 1.5625, "step": 2740 }, { "epoch": 2.5685267335979454, "grad_norm": 0.03030640438940187, "learning_rate": 6.210673395818571e-05, "loss": 1.5717, "step": 2750 }, { "epoch": 2.577865981788466, "grad_norm": 0.032197470232298186, "learning_rate": 6.139284362263185e-05, "loss": 1.5663, "step": 2760 }, { "epoch": 2.587205229978987, "grad_norm": 0.030983733397891462, "learning_rate": 6.0681258073309756e-05, "loss": 1.5657, "step": 2770 }, { "epoch": 2.5965444781695073, "grad_norm": 0.030427577702286164, "learning_rate": 5.9972019790752385e-05, "loss": 1.5708, "step": 2780 }, { "epoch": 2.605883726360028, "grad_norm": 0.032761226318855745, "learning_rate": 5.9265171115364495e-05, "loss": 1.5641, "step": 2790 }, { "epoch": 2.615222974550549, "grad_norm": 0.0317533648622182, "learning_rate": 5.856075424489511e-05, "loss": 1.5613, "step": 2800 } ], "logging_steps": 10, "max_steps": 4280, "num_input_tokens_seen": 0, "num_train_epochs": 4, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 2.267163894821054e+20, "train_batch_size": 2, "trial_name": null, "trial_params": null }