{ "best_metric": null, "best_model_checkpoint": null, "epoch": 9.994216310005784, "eval_steps": 500, "global_step": 4320, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02313475997686524, "grad_norm": 4.818889617919922, "learning_rate": 0.00019999735576321776, "loss": 2.7586, "step": 10 }, { "epoch": 0.04626951995373048, "grad_norm": 1.2888654470443726, "learning_rate": 0.00019998942319271077, "loss": 2.4225, "step": 20 }, { "epoch": 0.06940427993059572, "grad_norm": 0.9638609290122986, "learning_rate": 0.00019997620270799092, "loss": 2.2447, "step": 30 }, { "epoch": 0.09253903990746096, "grad_norm": 0.882834255695343, "learning_rate": 0.0001999576950082201, "loss": 2.2015, "step": 40 }, { "epoch": 0.1156737998843262, "grad_norm": 1.0840070247650146, "learning_rate": 0.00019993390107217302, "loss": 2.1489, "step": 50 }, { "epoch": 0.13880855986119145, "grad_norm": 13.946427345275879, "learning_rate": 0.0001999048221581858, "loss": 2.1928, "step": 60 }, { "epoch": 0.16194331983805668, "grad_norm": 0.9393675327301025, "learning_rate": 0.00019987045980408906, "loss": 2.172, "step": 70 }, { "epoch": 0.18507807981492191, "grad_norm": 0.882853627204895, "learning_rate": 0.00019983081582712685, "loss": 2.1491, "step": 80 }, { "epoch": 0.20821283979178715, "grad_norm": 0.8838891386985779, "learning_rate": 0.00019978589232386035, "loss": 2.1147, "step": 90 }, { "epoch": 0.2313475997686524, "grad_norm": 0.9649007320404053, "learning_rate": 0.00019973569167005723, "loss": 2.0984, "step": 100 }, { "epoch": 0.25448235974551764, "grad_norm": 0.9888186454772949, "learning_rate": 0.00019968021652056576, "loss": 2.0735, "step": 110 }, { "epoch": 0.2776171197223829, "grad_norm": 0.8648116588592529, "learning_rate": 0.00019961946980917456, "loss": 2.0005, "step": 120 }, { "epoch": 0.3007518796992481, "grad_norm": 0.932079553604126, "learning_rate": 0.0001995534547484574, "loss": 1.9938, "step": 130 }, { "epoch": 0.32388663967611336, "grad_norm": 0.9043927788734436, "learning_rate": 0.0001994821748296033, "loss": 1.9941, "step": 140 }, { "epoch": 0.3470213996529786, "grad_norm": 0.9096814393997192, "learning_rate": 0.00019940563382223197, "loss": 2.0175, "step": 150 }, { "epoch": 0.37015615962984383, "grad_norm": 0.9436717629432678, "learning_rate": 0.00019932383577419432, "loss": 2.0761, "step": 160 }, { "epoch": 0.3932909196067091, "grad_norm": 0.8598865270614624, "learning_rate": 0.00019923678501135848, "loss": 2.0289, "step": 170 }, { "epoch": 0.4164256795835743, "grad_norm": 0.8160977363586426, "learning_rate": 0.00019914448613738106, "loss": 2.0161, "step": 180 }, { "epoch": 0.43956043956043955, "grad_norm": 0.8758726119995117, "learning_rate": 0.0001990469440334636, "loss": 1.95, "step": 190 }, { "epoch": 0.4626951995373048, "grad_norm": 0.9540367126464844, "learning_rate": 0.00019894416385809444, "loss": 2.033, "step": 200 }, { "epoch": 0.48582995951417, "grad_norm": 0.9104657173156738, "learning_rate": 0.00019883615104677608, "loss": 1.965, "step": 210 }, { "epoch": 0.5089647194910353, "grad_norm": 0.9348485469818115, "learning_rate": 0.00019872291131173742, "loss": 1.9257, "step": 220 }, { "epoch": 0.5320994794679005, "grad_norm": 1.0083684921264648, "learning_rate": 0.00019860445064163193, "loss": 1.9851, "step": 230 }, { "epoch": 0.5552342394447658, "grad_norm": 0.9179242849349976, "learning_rate": 0.00019848077530122083, "loss": 1.9656, "step": 240 }, { "epoch": 0.578368999421631, "grad_norm": 0.881529688835144, "learning_rate": 0.00019835189183104178, "loss": 1.9819, "step": 250 }, { "epoch": 0.6015037593984962, "grad_norm": 0.931932270526886, "learning_rate": 0.00019821780704706307, "loss": 1.9791, "step": 260 }, { "epoch": 0.6246385193753615, "grad_norm": 0.9711050391197205, "learning_rate": 0.00019807852804032305, "loss": 1.9672, "step": 270 }, { "epoch": 0.6477732793522267, "grad_norm": 0.9100776314735413, "learning_rate": 0.00019793406217655517, "loss": 1.9403, "step": 280 }, { "epoch": 0.6709080393290919, "grad_norm": 1.0750184059143066, "learning_rate": 0.0001977844170957984, "loss": 1.9222, "step": 290 }, { "epoch": 0.6940427993059572, "grad_norm": 1.0510389804840088, "learning_rate": 0.00019762960071199333, "loss": 1.8928, "step": 300 }, { "epoch": 0.7171775592828225, "grad_norm": 0.8637176156044006, "learning_rate": 0.0001974696212125635, "loss": 1.9064, "step": 310 }, { "epoch": 0.7403123192596877, "grad_norm": 0.9364832043647766, "learning_rate": 0.00019730448705798239, "loss": 1.9649, "step": 320 }, { "epoch": 0.763447079236553, "grad_norm": 1.0437376499176025, "learning_rate": 0.00019713420698132614, "loss": 1.9472, "step": 330 }, { "epoch": 0.7865818392134182, "grad_norm": 0.8804395198822021, "learning_rate": 0.0001969587899878116, "loss": 1.8961, "step": 340 }, { "epoch": 0.8097165991902834, "grad_norm": 0.9558535814285278, "learning_rate": 0.0001967782453543201, "loss": 1.9841, "step": 350 }, { "epoch": 0.8328513591671486, "grad_norm": 1.0408087968826294, "learning_rate": 0.00019659258262890683, "loss": 1.9095, "step": 360 }, { "epoch": 0.8559861191440139, "grad_norm": 1.0767790079116821, "learning_rate": 0.00019640181163029595, "loss": 1.8774, "step": 370 }, { "epoch": 0.8791208791208791, "grad_norm": 0.9003483057022095, "learning_rate": 0.00019620594244736133, "loss": 1.8791, "step": 380 }, { "epoch": 0.9022556390977443, "grad_norm": 0.9258147478103638, "learning_rate": 0.0001960049854385929, "loss": 1.9243, "step": 390 }, { "epoch": 0.9253903990746096, "grad_norm": 0.9662225842475891, "learning_rate": 0.0001957989512315489, "loss": 1.8853, "step": 400 }, { "epoch": 0.9485251590514748, "grad_norm": 0.8478869795799255, "learning_rate": 0.00019558785072229396, "loss": 1.8573, "step": 410 }, { "epoch": 0.97165991902834, "grad_norm": 0.9442031979560852, "learning_rate": 0.0001953716950748227, "loss": 1.8542, "step": 420 }, { "epoch": 0.9947946790052054, "grad_norm": 0.970368504524231, "learning_rate": 0.00019515049572046937, "loss": 1.9055, "step": 430 }, { "epoch": 1.0179294389820706, "grad_norm": 0.9968038201332092, "learning_rate": 0.0001949242643573034, "loss": 1.7246, "step": 440 }, { "epoch": 1.0410641989589358, "grad_norm": 0.9842697978019714, "learning_rate": 0.0001946930129495106, "loss": 1.726, "step": 450 }, { "epoch": 1.064198958935801, "grad_norm": 1.029723882675171, "learning_rate": 0.0001944567537267605, "loss": 1.6401, "step": 460 }, { "epoch": 1.0873337189126664, "grad_norm": 0.95682692527771, "learning_rate": 0.00019421549918355966, "loss": 1.642, "step": 470 }, { "epoch": 1.1104684788895316, "grad_norm": 1.1155472993850708, "learning_rate": 0.00019396926207859084, "loss": 1.7003, "step": 480 }, { "epoch": 1.1336032388663968, "grad_norm": 1.0741841793060303, "learning_rate": 0.00019371805543403825, "loss": 1.6915, "step": 490 }, { "epoch": 1.156737998843262, "grad_norm": 1.0780987739562988, "learning_rate": 0.00019346189253489885, "loss": 1.5675, "step": 500 }, { "epoch": 1.1798727588201272, "grad_norm": 1.0540192127227783, "learning_rate": 0.00019320078692827987, "loss": 1.6459, "step": 510 }, { "epoch": 1.2030075187969924, "grad_norm": 1.0274903774261475, "learning_rate": 0.00019293475242268223, "loss": 1.6763, "step": 520 }, { "epoch": 1.2261422787738576, "grad_norm": 1.0954372882843018, "learning_rate": 0.00019266380308727055, "loss": 1.7276, "step": 530 }, { "epoch": 1.249277038750723, "grad_norm": 1.1079906225204468, "learning_rate": 0.0001923879532511287, "loss": 1.6581, "step": 540 }, { "epoch": 1.2724117987275883, "grad_norm": 1.053789734840393, "learning_rate": 0.00019210721750250235, "loss": 1.709, "step": 550 }, { "epoch": 1.2955465587044535, "grad_norm": 1.0135327577590942, "learning_rate": 0.00019182161068802741, "loss": 1.6606, "step": 560 }, { "epoch": 1.3186813186813187, "grad_norm": 1.0936933755874634, "learning_rate": 0.00019153114791194473, "loss": 1.6489, "step": 570 }, { "epoch": 1.3418160786581839, "grad_norm": 1.0684951543807983, "learning_rate": 0.00019123584453530144, "loss": 1.7086, "step": 580 }, { "epoch": 1.3649508386350493, "grad_norm": 1.1002228260040283, "learning_rate": 0.00019093571617513852, "loss": 1.6785, "step": 590 }, { "epoch": 1.3880855986119145, "grad_norm": 1.0296435356140137, "learning_rate": 0.000190630778703665, "loss": 1.6931, "step": 600 }, { "epoch": 1.4112203585887797, "grad_norm": 1.2503811120986938, "learning_rate": 0.00019032104824741843, "loss": 1.6651, "step": 610 }, { "epoch": 1.434355118565645, "grad_norm": 1.048912763595581, "learning_rate": 0.00019000654118641211, "loss": 1.7226, "step": 620 }, { "epoch": 1.45748987854251, "grad_norm": 1.1050567626953125, "learning_rate": 0.00018968727415326884, "loss": 1.6195, "step": 630 }, { "epoch": 1.4806246385193753, "grad_norm": 1.1447317600250244, "learning_rate": 0.00018936326403234125, "loss": 1.6719, "step": 640 }, { "epoch": 1.5037593984962405, "grad_norm": 1.2653794288635254, "learning_rate": 0.0001890345279588189, "loss": 1.6512, "step": 650 }, { "epoch": 1.5268941584731057, "grad_norm": 1.2562129497528076, "learning_rate": 0.00018870108331782217, "loss": 1.7135, "step": 660 }, { "epoch": 1.550028918449971, "grad_norm": 1.158982753753662, "learning_rate": 0.00018836294774348278, "loss": 1.6415, "step": 670 }, { "epoch": 1.5731636784268364, "grad_norm": 1.102187156677246, "learning_rate": 0.00018802013911801112, "loss": 1.589, "step": 680 }, { "epoch": 1.5962984384037016, "grad_norm": 1.126552939414978, "learning_rate": 0.0001876726755707508, "loss": 1.661, "step": 690 }, { "epoch": 1.6194331983805668, "grad_norm": 1.1704193353652954, "learning_rate": 0.0001873205754772196, "loss": 1.7446, "step": 700 }, { "epoch": 1.6425679583574322, "grad_norm": 1.1397557258605957, "learning_rate": 0.00018696385745813792, "loss": 1.6528, "step": 710 }, { "epoch": 1.6657027183342974, "grad_norm": 1.212633728981018, "learning_rate": 0.00018660254037844388, "loss": 1.6315, "step": 720 }, { "epoch": 1.6888374783111626, "grad_norm": 1.124883770942688, "learning_rate": 0.00018623664334629575, "loss": 1.6961, "step": 730 }, { "epoch": 1.7119722382880278, "grad_norm": 1.1361215114593506, "learning_rate": 0.00018586618571206134, "loss": 1.7002, "step": 740 }, { "epoch": 1.735106998264893, "grad_norm": 1.3424755334854126, "learning_rate": 0.00018549118706729468, "loss": 1.6754, "step": 750 }, { "epoch": 1.7582417582417582, "grad_norm": 1.1060627698898315, "learning_rate": 0.00018511166724369997, "loss": 1.6363, "step": 760 }, { "epoch": 1.7813765182186234, "grad_norm": 1.1465959548950195, "learning_rate": 0.0001847276463120828, "loss": 1.5913, "step": 770 }, { "epoch": 1.8045112781954886, "grad_norm": 1.1735104322433472, "learning_rate": 0.0001843391445812886, "loss": 1.6201, "step": 780 }, { "epoch": 1.8276460381723538, "grad_norm": 1.0827069282531738, "learning_rate": 0.00018394618259712865, "loss": 1.6184, "step": 790 }, { "epoch": 1.8507807981492193, "grad_norm": 1.0566381216049194, "learning_rate": 0.00018354878114129367, "loss": 1.64, "step": 800 }, { "epoch": 1.8739155581260845, "grad_norm": 1.223117470741272, "learning_rate": 0.00018314696123025454, "loss": 1.651, "step": 810 }, { "epoch": 1.8970503181029497, "grad_norm": 1.2156977653503418, "learning_rate": 0.00018274074411415105, "loss": 1.6914, "step": 820 }, { "epoch": 1.9201850780798149, "grad_norm": 1.179998517036438, "learning_rate": 0.00018233015127566807, "loss": 1.6974, "step": 830 }, { "epoch": 1.9433198380566803, "grad_norm": 1.1092870235443115, "learning_rate": 0.0001819152044288992, "loss": 1.5961, "step": 840 }, { "epoch": 1.9664545980335455, "grad_norm": 1.0674152374267578, "learning_rate": 0.0001814959255181988, "loss": 1.6687, "step": 850 }, { "epoch": 1.9895893580104107, "grad_norm": 1.3172643184661865, "learning_rate": 0.00018107233671702124, "loss": 1.7101, "step": 860 }, { "epoch": 2.012724117987276, "grad_norm": 1.3380361795425415, "learning_rate": 0.00018064446042674828, "loss": 1.5151, "step": 870 }, { "epoch": 2.035858877964141, "grad_norm": 1.431959867477417, "learning_rate": 0.0001802123192755044, "loss": 1.3543, "step": 880 }, { "epoch": 2.0589936379410063, "grad_norm": 1.2023745775222778, "learning_rate": 0.00017977593611696015, "loss": 1.3866, "step": 890 }, { "epoch": 2.0821283979178715, "grad_norm": 1.4215834140777588, "learning_rate": 0.00017933533402912354, "loss": 1.3933, "step": 900 }, { "epoch": 2.1052631578947367, "grad_norm": 1.2749346494674683, "learning_rate": 0.00017889053631311947, "loss": 1.3135, "step": 910 }, { "epoch": 2.128397917871602, "grad_norm": 1.347468376159668, "learning_rate": 0.00017844156649195759, "loss": 1.3276, "step": 920 }, { "epoch": 2.151532677848467, "grad_norm": 1.3235530853271484, "learning_rate": 0.00017798844830928817, "loss": 1.3517, "step": 930 }, { "epoch": 2.174667437825333, "grad_norm": 1.4696698188781738, "learning_rate": 0.0001775312057281466, "loss": 1.3442, "step": 940 }, { "epoch": 2.197802197802198, "grad_norm": 1.3937824964523315, "learning_rate": 0.00017706986292968582, "loss": 1.3082, "step": 950 }, { "epoch": 2.220936957779063, "grad_norm": 1.4411293268203735, "learning_rate": 0.0001766044443118978, "loss": 1.313, "step": 960 }, { "epoch": 2.2440717177559284, "grad_norm": 1.4743627309799194, "learning_rate": 0.00017613497448832312, "loss": 1.3783, "step": 970 }, { "epoch": 2.2672064777327936, "grad_norm": 1.3222951889038086, "learning_rate": 0.00017566147828674931, "loss": 1.3253, "step": 980 }, { "epoch": 2.290341237709659, "grad_norm": 1.3810490369796753, "learning_rate": 0.00017518398074789775, "loss": 1.3191, "step": 990 }, { "epoch": 2.313475997686524, "grad_norm": 1.3942046165466309, "learning_rate": 0.0001747025071240996, "loss": 1.4389, "step": 1000 }, { "epoch": 2.336610757663389, "grad_norm": 1.523859977722168, "learning_rate": 0.00017421708287796017, "loss": 1.4015, "step": 1010 }, { "epoch": 2.3597455176402544, "grad_norm": 1.435693383216858, "learning_rate": 0.0001737277336810124, "loss": 1.3434, "step": 1020 }, { "epoch": 2.3828802776171196, "grad_norm": 1.4890649318695068, "learning_rate": 0.00017323448541235924, "loss": 1.3337, "step": 1030 }, { "epoch": 2.406015037593985, "grad_norm": 1.3789806365966797, "learning_rate": 0.00017273736415730488, "loss": 1.3136, "step": 1040 }, { "epoch": 2.42914979757085, "grad_norm": 1.4059851169586182, "learning_rate": 0.00017223639620597556, "loss": 1.4041, "step": 1050 }, { "epoch": 2.4522845575477152, "grad_norm": 1.4800046682357788, "learning_rate": 0.00017173160805192893, "loss": 1.3156, "step": 1060 }, { "epoch": 2.4754193175245804, "grad_norm": 1.556971549987793, "learning_rate": 0.0001712230263907531, "loss": 1.394, "step": 1070 }, { "epoch": 2.498554077501446, "grad_norm": 1.4921575784683228, "learning_rate": 0.00017071067811865476, "loss": 1.3528, "step": 1080 }, { "epoch": 2.5216888374783113, "grad_norm": 1.53052818775177, "learning_rate": 0.00017019459033103682, "loss": 1.3099, "step": 1090 }, { "epoch": 2.5448235974551765, "grad_norm": 1.4016938209533691, "learning_rate": 0.0001696747903210655, "loss": 1.4067, "step": 1100 }, { "epoch": 2.5679583574320417, "grad_norm": 1.5849937200546265, "learning_rate": 0.00016915130557822695, "loss": 1.3557, "step": 1110 }, { "epoch": 2.591093117408907, "grad_norm": 1.5975929498672485, "learning_rate": 0.0001686241637868734, "loss": 1.304, "step": 1120 }, { "epoch": 2.614227877385772, "grad_norm": 1.3660000562667847, "learning_rate": 0.00016809339282475904, "loss": 1.3462, "step": 1130 }, { "epoch": 2.6373626373626373, "grad_norm": 1.4659340381622314, "learning_rate": 0.00016755902076156604, "loss": 1.3905, "step": 1140 }, { "epoch": 2.6604973973395025, "grad_norm": 1.398289680480957, "learning_rate": 0.0001670210758574196, "loss": 1.3498, "step": 1150 }, { "epoch": 2.6836321573163677, "grad_norm": 1.4304271936416626, "learning_rate": 0.00016647958656139378, "loss": 1.3641, "step": 1160 }, { "epoch": 2.706766917293233, "grad_norm": 1.455327033996582, "learning_rate": 0.00016593458151000688, "loss": 1.3586, "step": 1170 }, { "epoch": 2.7299016772700986, "grad_norm": 1.4968023300170898, "learning_rate": 0.00016538608952570698, "loss": 1.3731, "step": 1180 }, { "epoch": 2.753036437246964, "grad_norm": 1.5257724523544312, "learning_rate": 0.00016483413961534762, "loss": 1.3484, "step": 1190 }, { "epoch": 2.776171197223829, "grad_norm": 1.5006804466247559, "learning_rate": 0.00016427876096865394, "loss": 1.3702, "step": 1200 }, { "epoch": 2.799305957200694, "grad_norm": 1.376607060432434, "learning_rate": 0.00016371998295667886, "loss": 1.4002, "step": 1210 }, { "epoch": 2.8224407171775594, "grad_norm": 1.5122621059417725, "learning_rate": 0.00016315783513024977, "loss": 1.368, "step": 1220 }, { "epoch": 2.8455754771544246, "grad_norm": 1.5241082906723022, "learning_rate": 0.00016259234721840591, "loss": 1.3907, "step": 1230 }, { "epoch": 2.86871023713129, "grad_norm": 1.5061618089675903, "learning_rate": 0.000162023549126826, "loss": 1.3166, "step": 1240 }, { "epoch": 2.891844997108155, "grad_norm": 1.3175408840179443, "learning_rate": 0.00016145147093624677, "loss": 1.3919, "step": 1250 }, { "epoch": 2.91497975708502, "grad_norm": 1.614513635635376, "learning_rate": 0.00016087614290087208, "loss": 1.3702, "step": 1260 }, { "epoch": 2.9381145170618854, "grad_norm": 1.460629940032959, "learning_rate": 0.00016029759544677297, "loss": 1.3504, "step": 1270 }, { "epoch": 2.9612492770387506, "grad_norm": 1.5735194683074951, "learning_rate": 0.00015971585917027862, "loss": 1.3726, "step": 1280 }, { "epoch": 2.984384037015616, "grad_norm": 1.3974887132644653, "learning_rate": 0.00015913096483635824, "loss": 1.3811, "step": 1290 }, { "epoch": 3.007518796992481, "grad_norm": 1.3209352493286133, "learning_rate": 0.00015854294337699408, "loss": 1.2467, "step": 1300 }, { "epoch": 3.0306535569693462, "grad_norm": 1.64750337600708, "learning_rate": 0.0001579518258895455, "loss": 1.0457, "step": 1310 }, { "epoch": 3.053788316946212, "grad_norm": 1.7243373394012451, "learning_rate": 0.0001573576436351046, "loss": 1.0609, "step": 1320 }, { "epoch": 3.076923076923077, "grad_norm": 1.5499110221862793, "learning_rate": 0.00015676042803684288, "loss": 1.0522, "step": 1330 }, { "epoch": 3.1000578368999423, "grad_norm": 1.7944879531860352, "learning_rate": 0.0001561602106783493, "loss": 1.0416, "step": 1340 }, { "epoch": 3.1231925968768075, "grad_norm": 1.7069816589355469, "learning_rate": 0.00015555702330196023, "loss": 0.9983, "step": 1350 }, { "epoch": 3.1463273568536727, "grad_norm": 1.7057902812957764, "learning_rate": 0.0001549508978070806, "loss": 1.0115, "step": 1360 }, { "epoch": 3.169462116830538, "grad_norm": 1.6382006406784058, "learning_rate": 0.000154341866248497, "loss": 1.0349, "step": 1370 }, { "epoch": 3.192596876807403, "grad_norm": 1.6651489734649658, "learning_rate": 0.0001537299608346824, "loss": 1.092, "step": 1380 }, { "epoch": 3.2157316367842683, "grad_norm": 1.7190675735473633, "learning_rate": 0.00015311521392609282, "loss": 1.0611, "step": 1390 }, { "epoch": 3.2388663967611335, "grad_norm": 1.6004880666732788, "learning_rate": 0.000152497658033456, "loss": 1.0455, "step": 1400 }, { "epoch": 3.2620011567379987, "grad_norm": 1.870151162147522, "learning_rate": 0.00015187732581605217, "loss": 1.0883, "step": 1410 }, { "epoch": 3.285135916714864, "grad_norm": 1.7651612758636475, "learning_rate": 0.00015125425007998653, "loss": 1.0853, "step": 1420 }, { "epoch": 3.308270676691729, "grad_norm": 1.578336477279663, "learning_rate": 0.00015062846377645475, "loss": 1.0612, "step": 1430 }, { "epoch": 3.331405436668595, "grad_norm": 1.8268855810165405, "learning_rate": 0.00015000000000000001, "loss": 1.112, "step": 1440 }, { "epoch": 3.35454019664546, "grad_norm": 1.663640022277832, "learning_rate": 0.00014936889198676303, "loss": 1.0387, "step": 1450 }, { "epoch": 3.377674956622325, "grad_norm": 1.7626187801361084, "learning_rate": 0.00014873517311272423, "loss": 1.0766, "step": 1460 }, { "epoch": 3.4008097165991904, "grad_norm": 1.683388113975525, "learning_rate": 0.00014809887689193877, "loss": 1.0982, "step": 1470 }, { "epoch": 3.4239444765760556, "grad_norm": 1.6581867933273315, "learning_rate": 0.00014746003697476404, "loss": 1.1141, "step": 1480 }, { "epoch": 3.447079236552921, "grad_norm": 1.9846975803375244, "learning_rate": 0.0001468186871460802, "loss": 1.0853, "step": 1490 }, { "epoch": 3.470213996529786, "grad_norm": 1.7938522100448608, "learning_rate": 0.00014617486132350343, "loss": 1.1085, "step": 1500 }, { "epoch": 3.493348756506651, "grad_norm": 1.7541499137878418, "learning_rate": 0.00014552859355559204, "loss": 1.0778, "step": 1510 }, { "epoch": 3.5164835164835164, "grad_norm": 1.7515652179718018, "learning_rate": 0.00014487991802004623, "loss": 1.1077, "step": 1520 }, { "epoch": 3.5396182764603816, "grad_norm": 1.7582262754440308, "learning_rate": 0.00014422886902190014, "loss": 1.0703, "step": 1530 }, { "epoch": 3.562753036437247, "grad_norm": 1.6952829360961914, "learning_rate": 0.00014357548099170795, "loss": 1.0588, "step": 1540 }, { "epoch": 3.585887796414112, "grad_norm": 1.8166347742080688, "learning_rate": 0.00014291978848372293, "loss": 1.053, "step": 1550 }, { "epoch": 3.6090225563909772, "grad_norm": 1.8070217370986938, "learning_rate": 0.00014226182617406996, "loss": 1.1416, "step": 1560 }, { "epoch": 3.6321573163678424, "grad_norm": 1.9227118492126465, "learning_rate": 0.00014160162885891193, "loss": 1.0947, "step": 1570 }, { "epoch": 3.6552920763447077, "grad_norm": 1.8285157680511475, "learning_rate": 0.00014093923145260925, "loss": 1.062, "step": 1580 }, { "epoch": 3.6784268363215733, "grad_norm": 1.8435360193252563, "learning_rate": 0.00014027466898587374, "loss": 1.0761, "step": 1590 }, { "epoch": 3.7015615962984385, "grad_norm": 1.8980731964111328, "learning_rate": 0.0001396079766039157, "loss": 1.0855, "step": 1600 }, { "epoch": 3.7246963562753037, "grad_norm": 1.8872120380401611, "learning_rate": 0.00013893918956458552, "loss": 1.0775, "step": 1610 }, { "epoch": 3.747831116252169, "grad_norm": 1.8531919717788696, "learning_rate": 0.000138268343236509, "loss": 1.0411, "step": 1620 }, { "epoch": 3.770965876229034, "grad_norm": 1.6847087144851685, "learning_rate": 0.0001375954730972168, "loss": 1.0747, "step": 1630 }, { "epoch": 3.7941006362058993, "grad_norm": 1.818551778793335, "learning_rate": 0.00013692061473126845, "loss": 1.0527, "step": 1640 }, { "epoch": 3.8172353961827645, "grad_norm": 1.7148271799087524, "learning_rate": 0.00013624380382837016, "loss": 1.0999, "step": 1650 }, { "epoch": 3.8403701561596297, "grad_norm": 1.6660877466201782, "learning_rate": 0.0001355650761814877, "loss": 1.0868, "step": 1660 }, { "epoch": 3.863504916136495, "grad_norm": 1.787760615348816, "learning_rate": 0.0001348844676849531, "loss": 1.1218, "step": 1670 }, { "epoch": 3.8866396761133606, "grad_norm": 1.7259087562561035, "learning_rate": 0.00013420201433256689, "loss": 1.085, "step": 1680 }, { "epoch": 3.909774436090226, "grad_norm": 1.666365146636963, "learning_rate": 0.00013351775221569415, "loss": 1.1676, "step": 1690 }, { "epoch": 3.932909196067091, "grad_norm": 1.7131948471069336, "learning_rate": 0.00013283171752135614, "loss": 1.0713, "step": 1700 }, { "epoch": 3.956043956043956, "grad_norm": 1.7973147630691528, "learning_rate": 0.00013214394653031616, "loss": 1.0864, "step": 1710 }, { "epoch": 3.9791787160208214, "grad_norm": 1.7572144269943237, "learning_rate": 0.00013145447561516138, "loss": 1.0458, "step": 1720 }, { "epoch": 4.002313475997687, "grad_norm": 1.5910913944244385, "learning_rate": 0.00013076334123837883, "loss": 1.0368, "step": 1730 }, { "epoch": 4.025448235974552, "grad_norm": 1.9898087978363037, "learning_rate": 0.00013007057995042732, "loss": 0.7595, "step": 1740 }, { "epoch": 4.048582995951417, "grad_norm": 1.8028409481048584, "learning_rate": 0.00012937622838780444, "loss": 0.7896, "step": 1750 }, { "epoch": 4.071717755928282, "grad_norm": 2.1292295455932617, "learning_rate": 0.00012868032327110904, "loss": 0.748, "step": 1760 }, { "epoch": 4.094852515905147, "grad_norm": 2.0062098503112793, "learning_rate": 0.00012798290140309923, "loss": 0.7741, "step": 1770 }, { "epoch": 4.117987275882013, "grad_norm": 2.028233528137207, "learning_rate": 0.00012728399966674612, "loss": 0.8249, "step": 1780 }, { "epoch": 4.141122035858878, "grad_norm": 2.332066297531128, "learning_rate": 0.0001265836550232833, "loss": 0.8377, "step": 1790 }, { "epoch": 4.164256795835743, "grad_norm": 2.0353000164031982, "learning_rate": 0.00012588190451025207, "loss": 0.7875, "step": 1800 }, { "epoch": 4.187391555812608, "grad_norm": 2.017143964767456, "learning_rate": 0.00012517878523954286, "loss": 0.7837, "step": 1810 }, { "epoch": 4.2105263157894735, "grad_norm": 2.2068400382995605, "learning_rate": 0.0001244743343954324, "loss": 0.844, "step": 1820 }, { "epoch": 4.233661075766339, "grad_norm": 2.1417810916900635, "learning_rate": 0.00012376858923261733, "loss": 0.7947, "step": 1830 }, { "epoch": 4.256795835743204, "grad_norm": 2.1283600330352783, "learning_rate": 0.00012306158707424403, "loss": 0.846, "step": 1840 }, { "epoch": 4.279930595720069, "grad_norm": 1.9856690168380737, "learning_rate": 0.00012235336530993474, "loss": 0.8118, "step": 1850 }, { "epoch": 4.303065355696934, "grad_norm": 1.9329304695129395, "learning_rate": 0.00012164396139381029, "loss": 0.78, "step": 1860 }, { "epoch": 4.3262001156737995, "grad_norm": 1.9364598989486694, "learning_rate": 0.0001209334128425092, "loss": 0.8625, "step": 1870 }, { "epoch": 4.349334875650666, "grad_norm": 1.803899884223938, "learning_rate": 0.00012022175723320381, "loss": 0.7904, "step": 1880 }, { "epoch": 4.372469635627531, "grad_norm": 1.9305102825164795, "learning_rate": 0.00011950903220161285, "loss": 0.8376, "step": 1890 }, { "epoch": 4.395604395604396, "grad_norm": 2.184109687805176, "learning_rate": 0.00011879527544001119, "loss": 0.755, "step": 1900 }, { "epoch": 4.418739155581261, "grad_norm": 1.9741244316101074, "learning_rate": 0.00011808052469523654, "loss": 0.8244, "step": 1910 }, { "epoch": 4.441873915558126, "grad_norm": 1.954925537109375, "learning_rate": 0.00011736481776669306, "loss": 0.8061, "step": 1920 }, { "epoch": 4.465008675534992, "grad_norm": 1.891669750213623, "learning_rate": 0.00011664819250435246, "loss": 0.8089, "step": 1930 }, { "epoch": 4.488143435511857, "grad_norm": 2.0183587074279785, "learning_rate": 0.00011593068680675228, "loss": 0.8304, "step": 1940 }, { "epoch": 4.511278195488722, "grad_norm": 1.9433139562606812, "learning_rate": 0.00011521233861899167, "loss": 0.8356, "step": 1950 }, { "epoch": 4.534412955465587, "grad_norm": 2.0215744972229004, "learning_rate": 0.00011449318593072466, "loss": 0.8106, "step": 1960 }, { "epoch": 4.557547715442452, "grad_norm": 2.0268826484680176, "learning_rate": 0.00011377326677415108, "loss": 0.8186, "step": 1970 }, { "epoch": 4.580682475419318, "grad_norm": 2.270048141479492, "learning_rate": 0.00011305261922200519, "loss": 0.7984, "step": 1980 }, { "epoch": 4.603817235396183, "grad_norm": 1.9835270643234253, "learning_rate": 0.0001123312813855422, "loss": 0.8106, "step": 1990 }, { "epoch": 4.626951995373048, "grad_norm": 2.036339044570923, "learning_rate": 0.00011160929141252303, "loss": 0.8778, "step": 2000 }, { "epoch": 4.650086755349913, "grad_norm": 2.1386802196502686, "learning_rate": 0.00011088668748519647, "loss": 0.8408, "step": 2010 }, { "epoch": 4.673221515326778, "grad_norm": 2.199148416519165, "learning_rate": 0.00011016350781828019, "loss": 0.8356, "step": 2020 }, { "epoch": 4.696356275303644, "grad_norm": 1.943888783454895, "learning_rate": 0.00010943979065693975, "loss": 0.8521, "step": 2030 }, { "epoch": 4.719491035280509, "grad_norm": 2.2225332260131836, "learning_rate": 0.00010871557427476583, "loss": 0.8487, "step": 2040 }, { "epoch": 4.742625795257374, "grad_norm": 2.0183157920837402, "learning_rate": 0.0001079908969717504, "loss": 0.8258, "step": 2050 }, { "epoch": 4.765760555234239, "grad_norm": 2.1665942668914795, "learning_rate": 0.00010726579707226108, "loss": 0.8786, "step": 2060 }, { "epoch": 4.7888953152111045, "grad_norm": 2.0331172943115234, "learning_rate": 0.00010654031292301432, "loss": 0.8344, "step": 2070 }, { "epoch": 4.81203007518797, "grad_norm": 2.299880266189575, "learning_rate": 0.00010581448289104758, "loss": 0.8267, "step": 2080 }, { "epoch": 4.835164835164835, "grad_norm": 2.0382227897644043, "learning_rate": 0.00010508834536169028, "loss": 0.8533, "step": 2090 }, { "epoch": 4.8582995951417, "grad_norm": 1.9726015329360962, "learning_rate": 0.00010436193873653361, "loss": 0.8654, "step": 2100 }, { "epoch": 4.881434355118565, "grad_norm": 2.1659207344055176, "learning_rate": 0.00010363530143139999, "loss": 0.8723, "step": 2110 }, { "epoch": 4.9045691150954305, "grad_norm": 2.0480141639709473, "learning_rate": 0.00010290847187431113, "loss": 0.8272, "step": 2120 }, { "epoch": 4.927703875072297, "grad_norm": 2.1771929264068604, "learning_rate": 0.00010218148850345613, "loss": 0.8352, "step": 2130 }, { "epoch": 4.950838635049161, "grad_norm": 2.022383689880371, "learning_rate": 0.00010145438976515828, "loss": 0.9057, "step": 2140 }, { "epoch": 4.973973395026027, "grad_norm": 2.1294972896575928, "learning_rate": 0.00010072721411184219, "loss": 0.8775, "step": 2150 }, { "epoch": 4.997108155002892, "grad_norm": 2.14635968208313, "learning_rate": 0.0001, "loss": 0.8175, "step": 2160 }, { "epoch": 5.020242914979757, "grad_norm": 2.4558303356170654, "learning_rate": 9.927278588815786e-05, "loss": 0.6267, "step": 2170 }, { "epoch": 5.043377674956623, "grad_norm": 2.078815221786499, "learning_rate": 9.854561023484173e-05, "loss": 0.6033, "step": 2180 }, { "epoch": 5.066512434933488, "grad_norm": 2.428105354309082, "learning_rate": 9.78185114965439e-05, "loss": 0.6157, "step": 2190 }, { "epoch": 5.089647194910353, "grad_norm": 2.1209535598754883, "learning_rate": 9.709152812568886e-05, "loss": 0.5899, "step": 2200 }, { "epoch": 5.112781954887218, "grad_norm": 2.1761510372161865, "learning_rate": 9.636469856860005e-05, "loss": 0.5986, "step": 2210 }, { "epoch": 5.135916714864083, "grad_norm": 2.0283255577087402, "learning_rate": 9.563806126346642e-05, "loss": 0.5719, "step": 2220 }, { "epoch": 5.159051474840949, "grad_norm": 2.2018613815307617, "learning_rate": 9.491165463830975e-05, "loss": 0.5884, "step": 2230 }, { "epoch": 5.182186234817814, "grad_norm": 2.153411626815796, "learning_rate": 9.418551710895243e-05, "loss": 0.5664, "step": 2240 }, { "epoch": 5.205320994794679, "grad_norm": 2.133164405822754, "learning_rate": 9.345968707698569e-05, "loss": 0.5873, "step": 2250 }, { "epoch": 5.228455754771544, "grad_norm": 2.3184361457824707, "learning_rate": 9.273420292773894e-05, "loss": 0.6031, "step": 2260 }, { "epoch": 5.251590514748409, "grad_norm": 1.985695481300354, "learning_rate": 9.200910302824963e-05, "loss": 0.603, "step": 2270 }, { "epoch": 5.274725274725275, "grad_norm": 2.204653024673462, "learning_rate": 9.128442572523417e-05, "loss": 0.5943, "step": 2280 }, { "epoch": 5.29786003470214, "grad_norm": 2.212740659713745, "learning_rate": 9.056020934306032e-05, "loss": 0.6301, "step": 2290 }, { "epoch": 5.320994794679005, "grad_norm": 2.320549964904785, "learning_rate": 8.983649218171982e-05, "loss": 0.6231, "step": 2300 }, { "epoch": 5.34412955465587, "grad_norm": 2.21589732170105, "learning_rate": 8.911331251480357e-05, "loss": 0.5959, "step": 2310 }, { "epoch": 5.3672643146327355, "grad_norm": 2.234104633331299, "learning_rate": 8.839070858747697e-05, "loss": 0.6234, "step": 2320 }, { "epoch": 5.390399074609601, "grad_norm": 2.3640241622924805, "learning_rate": 8.76687186144578e-05, "loss": 0.6501, "step": 2330 }, { "epoch": 5.413533834586466, "grad_norm": 2.0853078365325928, "learning_rate": 8.694738077799488e-05, "loss": 0.5925, "step": 2340 }, { "epoch": 5.436668594563331, "grad_norm": 2.4165499210357666, "learning_rate": 8.622673322584893e-05, "loss": 0.6399, "step": 2350 }, { "epoch": 5.459803354540196, "grad_norm": 2.2116127014160156, "learning_rate": 8.550681406927535e-05, "loss": 0.6438, "step": 2360 }, { "epoch": 5.4829381145170615, "grad_norm": 2.056300163269043, "learning_rate": 8.478766138100834e-05, "loss": 0.5918, "step": 2370 }, { "epoch": 5.506072874493928, "grad_norm": 2.453059196472168, "learning_rate": 8.406931319324776e-05, "loss": 0.6414, "step": 2380 }, { "epoch": 5.529207634470792, "grad_norm": 2.425656318664551, "learning_rate": 8.33518074956476e-05, "loss": 0.6227, "step": 2390 }, { "epoch": 5.552342394447658, "grad_norm": 2.2804975509643555, "learning_rate": 8.263518223330697e-05, "loss": 0.6061, "step": 2400 }, { "epoch": 5.575477154424523, "grad_norm": 2.6980812549591064, "learning_rate": 8.19194753047635e-05, "loss": 0.5937, "step": 2410 }, { "epoch": 5.598611914401388, "grad_norm": 2.084348201751709, "learning_rate": 8.120472455998882e-05, "loss": 0.6146, "step": 2420 }, { "epoch": 5.621746674378254, "grad_norm": 2.330526113510132, "learning_rate": 8.049096779838719e-05, "loss": 0.6077, "step": 2430 }, { "epoch": 5.644881434355119, "grad_norm": 2.280693769454956, "learning_rate": 7.977824276679623e-05, "loss": 0.6267, "step": 2440 }, { "epoch": 5.668016194331984, "grad_norm": 2.3855202198028564, "learning_rate": 7.90665871574908e-05, "loss": 0.6045, "step": 2450 }, { "epoch": 5.691150954308849, "grad_norm": 2.3103318214416504, "learning_rate": 7.835603860618972e-05, "loss": 0.6166, "step": 2460 }, { "epoch": 5.714285714285714, "grad_norm": 2.136725425720215, "learning_rate": 7.764663469006526e-05, "loss": 0.6116, "step": 2470 }, { "epoch": 5.73742047426258, "grad_norm": 2.1863021850585938, "learning_rate": 7.693841292575598e-05, "loss": 0.6302, "step": 2480 }, { "epoch": 5.760555234239445, "grad_norm": 2.1635851860046387, "learning_rate": 7.623141076738271e-05, "loss": 0.6232, "step": 2490 }, { "epoch": 5.78368999421631, "grad_norm": 2.2908101081848145, "learning_rate": 7.552566560456762e-05, "loss": 0.6573, "step": 2500 }, { "epoch": 5.806824754193175, "grad_norm": 2.1210217475891113, "learning_rate": 7.482121476045716e-05, "loss": 0.6391, "step": 2510 }, { "epoch": 5.82995951417004, "grad_norm": 2.4276537895202637, "learning_rate": 7.411809548974792e-05, "loss": 0.6313, "step": 2520 }, { "epoch": 5.853094274146906, "grad_norm": 2.296722412109375, "learning_rate": 7.341634497671672e-05, "loss": 0.6218, "step": 2530 }, { "epoch": 5.876229034123771, "grad_norm": 2.1190712451934814, "learning_rate": 7.271600033325393e-05, "loss": 0.6521, "step": 2540 }, { "epoch": 5.899363794100636, "grad_norm": 2.1762571334838867, "learning_rate": 7.20170985969008e-05, "loss": 0.6107, "step": 2550 }, { "epoch": 5.922498554077501, "grad_norm": 2.5130181312561035, "learning_rate": 7.131967672889101e-05, "loss": 0.6325, "step": 2560 }, { "epoch": 5.9456333140543665, "grad_norm": 2.1171939373016357, "learning_rate": 7.062377161219556e-05, "loss": 0.6242, "step": 2570 }, { "epoch": 5.968768074031232, "grad_norm": 2.013845682144165, "learning_rate": 6.992942004957271e-05, "loss": 0.6266, "step": 2580 }, { "epoch": 5.991902834008097, "grad_norm": 2.2959377765655518, "learning_rate": 6.923665876162118e-05, "loss": 0.6191, "step": 2590 }, { "epoch": 6.015037593984962, "grad_norm": 2.0503013134002686, "learning_rate": 6.854552438483865e-05, "loss": 0.5465, "step": 2600 }, { "epoch": 6.038172353961827, "grad_norm": 2.3042502403259277, "learning_rate": 6.785605346968386e-05, "loss": 0.4176, "step": 2610 }, { "epoch": 6.0613071139386925, "grad_norm": 2.150449275970459, "learning_rate": 6.71682824786439e-05, "loss": 0.4201, "step": 2620 }, { "epoch": 6.084441873915559, "grad_norm": 2.2146551609039307, "learning_rate": 6.648224778430586e-05, "loss": 0.4305, "step": 2630 }, { "epoch": 6.107576633892424, "grad_norm": 2.163520097732544, "learning_rate": 6.579798566743314e-05, "loss": 0.4445, "step": 2640 }, { "epoch": 6.130711393869289, "grad_norm": 2.339061975479126, "learning_rate": 6.511553231504693e-05, "loss": 0.4453, "step": 2650 }, { "epoch": 6.153846153846154, "grad_norm": 2.3959970474243164, "learning_rate": 6.443492381851237e-05, "loss": 0.4061, "step": 2660 }, { "epoch": 6.176980913823019, "grad_norm": 2.2613327503204346, "learning_rate": 6.375619617162985e-05, "loss": 0.4074, "step": 2670 }, { "epoch": 6.200115673799885, "grad_norm": 2.2177650928497314, "learning_rate": 6.307938526873157e-05, "loss": 0.441, "step": 2680 }, { "epoch": 6.22325043377675, "grad_norm": 2.184535503387451, "learning_rate": 6.24045269027832e-05, "loss": 0.4518, "step": 2690 }, { "epoch": 6.246385193753615, "grad_norm": 2.2392778396606445, "learning_rate": 6.173165676349103e-05, "loss": 0.4613, "step": 2700 }, { "epoch": 6.26951995373048, "grad_norm": 2.2998814582824707, "learning_rate": 6.106081043541451e-05, "loss": 0.4466, "step": 2710 }, { "epoch": 6.292654713707345, "grad_norm": 2.1889827251434326, "learning_rate": 6.039202339608432e-05, "loss": 0.4625, "step": 2720 }, { "epoch": 6.315789473684211, "grad_norm": 2.439364194869995, "learning_rate": 5.9725331014126294e-05, "loss": 0.4722, "step": 2730 }, { "epoch": 6.338924233661076, "grad_norm": 2.3686563968658447, "learning_rate": 5.906076854739074e-05, "loss": 0.4256, "step": 2740 }, { "epoch": 6.362058993637941, "grad_norm": 2.355642795562744, "learning_rate": 5.83983711410881e-05, "loss": 0.4405, "step": 2750 }, { "epoch": 6.385193753614806, "grad_norm": 2.5149965286254883, "learning_rate": 5.773817382593008e-05, "loss": 0.4775, "step": 2760 }, { "epoch": 6.408328513591671, "grad_norm": 2.1884281635284424, "learning_rate": 5.708021151627712e-05, "loss": 0.4219, "step": 2770 }, { "epoch": 6.431463273568537, "grad_norm": 2.5396625995635986, "learning_rate": 5.642451900829209e-05, "loss": 0.4617, "step": 2780 }, { "epoch": 6.454598033545402, "grad_norm": 2.1632354259490967, "learning_rate": 5.577113097809989e-05, "loss": 0.4407, "step": 2790 }, { "epoch": 6.477732793522267, "grad_norm": 2.3358309268951416, "learning_rate": 5.5120081979953785e-05, "loss": 0.4526, "step": 2800 }, { "epoch": 6.500867553499132, "grad_norm": 2.1623499393463135, "learning_rate": 5.447140644440798e-05, "loss": 0.4408, "step": 2810 }, { "epoch": 6.5240023134759975, "grad_norm": 2.002521514892578, "learning_rate": 5.382513867649663e-05, "loss": 0.4536, "step": 2820 }, { "epoch": 6.547137073452863, "grad_norm": 2.0050201416015625, "learning_rate": 5.3181312853919804e-05, "loss": 0.4468, "step": 2830 }, { "epoch": 6.570271833429728, "grad_norm": 1.9836812019348145, "learning_rate": 5.253996302523596e-05, "loss": 0.4524, "step": 2840 }, { "epoch": 6.593406593406593, "grad_norm": 2.3806679248809814, "learning_rate": 5.190112310806126e-05, "loss": 0.4598, "step": 2850 }, { "epoch": 6.616541353383458, "grad_norm": 2.2022571563720703, "learning_rate": 5.1264826887275776e-05, "loss": 0.4798, "step": 2860 }, { "epoch": 6.6396761133603235, "grad_norm": 2.4624128341674805, "learning_rate": 5.063110801323697e-05, "loss": 0.4561, "step": 2870 }, { "epoch": 6.66281087333719, "grad_norm": 2.4618606567382812, "learning_rate": 5.000000000000002e-05, "loss": 0.4389, "step": 2880 }, { "epoch": 6.685945633314054, "grad_norm": 2.107267379760742, "learning_rate": 4.937153622354529e-05, "loss": 0.4696, "step": 2890 }, { "epoch": 6.70908039329092, "grad_norm": 2.5566253662109375, "learning_rate": 4.8745749920013486e-05, "loss": 0.4778, "step": 2900 }, { "epoch": 6.732215153267785, "grad_norm": 2.2925281524658203, "learning_rate": 4.8122674183947836e-05, "loss": 0.4536, "step": 2910 }, { "epoch": 6.75534991324465, "grad_norm": 2.449045419692993, "learning_rate": 4.7502341966544e-05, "loss": 0.4679, "step": 2920 }, { "epoch": 6.778484673221516, "grad_norm": 2.3320090770721436, "learning_rate": 4.688478607390723e-05, "loss": 0.4517, "step": 2930 }, { "epoch": 6.801619433198381, "grad_norm": 2.4661507606506348, "learning_rate": 4.6270039165317605e-05, "loss": 0.4699, "step": 2940 }, { "epoch": 6.824754193175246, "grad_norm": 2.398205280303955, "learning_rate": 4.565813375150302e-05, "loss": 0.4614, "step": 2950 }, { "epoch": 6.847888953152111, "grad_norm": 2.3216257095336914, "learning_rate": 4.50491021929194e-05, "loss": 0.465, "step": 2960 }, { "epoch": 6.871023713128976, "grad_norm": 2.2298338413238525, "learning_rate": 4.444297669803981e-05, "loss": 0.4679, "step": 2970 }, { "epoch": 6.894158473105842, "grad_norm": 2.1584606170654297, "learning_rate": 4.383978932165073e-05, "loss": 0.4532, "step": 2980 }, { "epoch": 6.917293233082707, "grad_norm": 2.500182628631592, "learning_rate": 4.3239571963157145e-05, "loss": 0.4784, "step": 2990 }, { "epoch": 6.940427993059572, "grad_norm": 2.2554585933685303, "learning_rate": 4.264235636489542e-05, "loss": 0.471, "step": 3000 }, { "epoch": 6.963562753036437, "grad_norm": 2.2849414348602295, "learning_rate": 4.2048174110454496e-05, "loss": 0.4595, "step": 3010 }, { "epoch": 6.986697513013302, "grad_norm": 2.273737668991089, "learning_rate": 4.145705662300595e-05, "loss": 0.4557, "step": 3020 }, { "epoch": 7.009832272990168, "grad_norm": 1.8555997610092163, "learning_rate": 4.086903516364179e-05, "loss": 0.4155, "step": 3030 }, { "epoch": 7.032967032967033, "grad_norm": 2.351282835006714, "learning_rate": 4.028414082972141e-05, "loss": 0.3444, "step": 3040 }, { "epoch": 7.056101792943898, "grad_norm": 1.9804246425628662, "learning_rate": 3.9702404553227046e-05, "loss": 0.3247, "step": 3050 }, { "epoch": 7.079236552920763, "grad_norm": 2.0093185901641846, "learning_rate": 3.9123857099127936e-05, "loss": 0.3455, "step": 3060 }, { "epoch": 7.1023713128976285, "grad_norm": 2.369689702987671, "learning_rate": 3.854852906375326e-05, "loss": 0.3474, "step": 3070 }, { "epoch": 7.125506072874494, "grad_norm": 2.10414981842041, "learning_rate": 3.7976450873174005e-05, "loss": 0.3179, "step": 3080 }, { "epoch": 7.148640832851359, "grad_norm": 2.3380303382873535, "learning_rate": 3.7407652781594095e-05, "loss": 0.3298, "step": 3090 }, { "epoch": 7.171775592828224, "grad_norm": 1.9055235385894775, "learning_rate": 3.684216486975026e-05, "loss": 0.3446, "step": 3100 }, { "epoch": 7.194910352805089, "grad_norm": 2.2698934078216553, "learning_rate": 3.628001704332118e-05, "loss": 0.3202, "step": 3110 }, { "epoch": 7.2180451127819545, "grad_norm": 2.1574339866638184, "learning_rate": 3.5721239031346066e-05, "loss": 0.3299, "step": 3120 }, { "epoch": 7.241179872758821, "grad_norm": 2.068159818649292, "learning_rate": 3.516586038465237e-05, "loss": 0.3158, "step": 3130 }, { "epoch": 7.264314632735685, "grad_norm": 2.033571720123291, "learning_rate": 3.461391047429304e-05, "loss": 0.3511, "step": 3140 }, { "epoch": 7.287449392712551, "grad_norm": 2.3351962566375732, "learning_rate": 3.406541848999312e-05, "loss": 0.3495, "step": 3150 }, { "epoch": 7.310584152689416, "grad_norm": 1.9200048446655273, "learning_rate": 3.352041343860621e-05, "loss": 0.3329, "step": 3160 }, { "epoch": 7.333718912666281, "grad_norm": 2.1292202472686768, "learning_rate": 3.297892414258043e-05, "loss": 0.3233, "step": 3170 }, { "epoch": 7.356853672643147, "grad_norm": 2.009242057800293, "learning_rate": 3.244097923843398e-05, "loss": 0.3412, "step": 3180 }, { "epoch": 7.379988432620012, "grad_norm": 2.1473593711853027, "learning_rate": 3.190660717524094e-05, "loss": 0.3379, "step": 3190 }, { "epoch": 7.403123192596877, "grad_norm": 2.2157785892486572, "learning_rate": 3.137583621312665e-05, "loss": 0.3417, "step": 3200 }, { "epoch": 7.426257952573742, "grad_norm": 2.3066554069519043, "learning_rate": 3.0848694421773075e-05, "loss": 0.3319, "step": 3210 }, { "epoch": 7.449392712550607, "grad_norm": 2.0240354537963867, "learning_rate": 3.032520967893453e-05, "loss": 0.3343, "step": 3220 }, { "epoch": 7.472527472527473, "grad_norm": 2.417139768600464, "learning_rate": 2.9805409668963168e-05, "loss": 0.3236, "step": 3230 }, { "epoch": 7.495662232504338, "grad_norm": 2.220024347305298, "learning_rate": 2.9289321881345254e-05, "loss": 0.3374, "step": 3240 }, { "epoch": 7.518796992481203, "grad_norm": 2.071851968765259, "learning_rate": 2.8776973609246928e-05, "loss": 0.3625, "step": 3250 }, { "epoch": 7.541931752458068, "grad_norm": 2.0256235599517822, "learning_rate": 2.826839194807105e-05, "loss": 0.346, "step": 3260 }, { "epoch": 7.565066512434933, "grad_norm": 2.0721919536590576, "learning_rate": 2.776360379402445e-05, "loss": 0.3444, "step": 3270 }, { "epoch": 7.588201272411799, "grad_norm": 2.2439305782318115, "learning_rate": 2.7262635842695127e-05, "loss": 0.3413, "step": 3280 }, { "epoch": 7.611336032388664, "grad_norm": 2.2420361042022705, "learning_rate": 2.6765514587640818e-05, "loss": 0.3413, "step": 3290 }, { "epoch": 7.634470792365529, "grad_norm": 2.33591628074646, "learning_rate": 2.6272266318987603e-05, "loss": 0.3309, "step": 3300 }, { "epoch": 7.657605552342394, "grad_norm": 2.183473825454712, "learning_rate": 2.578291712203983e-05, "loss": 0.3341, "step": 3310 }, { "epoch": 7.6807403123192595, "grad_norm": 2.333395481109619, "learning_rate": 2.529749287590042e-05, "loss": 0.3665, "step": 3320 }, { "epoch": 7.703875072296125, "grad_norm": 2.368133783340454, "learning_rate": 2.4816019252102273e-05, "loss": 0.3341, "step": 3330 }, { "epoch": 7.72700983227299, "grad_norm": 2.055673360824585, "learning_rate": 2.433852171325072e-05, "loss": 0.331, "step": 3340 }, { "epoch": 7.750144592249855, "grad_norm": 2.531468391418457, "learning_rate": 2.3865025511676897e-05, "loss": 0.3553, "step": 3350 }, { "epoch": 7.77327935222672, "grad_norm": 2.126619338989258, "learning_rate": 2.339555568810221e-05, "loss": 0.3332, "step": 3360 }, { "epoch": 7.7964141122035855, "grad_norm": 2.1222150325775146, "learning_rate": 2.2930137070314194e-05, "loss": 0.3428, "step": 3370 }, { "epoch": 7.819548872180452, "grad_norm": 2.4255363941192627, "learning_rate": 2.246879427185341e-05, "loss": 0.3372, "step": 3380 }, { "epoch": 7.842683632157316, "grad_norm": 2.0324771404266357, "learning_rate": 2.201155169071184e-05, "loss": 0.3334, "step": 3390 }, { "epoch": 7.865818392134182, "grad_norm": 2.5304958820343018, "learning_rate": 2.155843350804243e-05, "loss": 0.3539, "step": 3400 }, { "epoch": 7.888953152111047, "grad_norm": 2.394742965698242, "learning_rate": 2.110946368688055e-05, "loss": 0.3522, "step": 3410 }, { "epoch": 7.912087912087912, "grad_norm": 2.2429323196411133, "learning_rate": 2.0664665970876496e-05, "loss": 0.3564, "step": 3420 }, { "epoch": 7.935222672064778, "grad_norm": 2.269540309906006, "learning_rate": 2.0224063883039868e-05, "loss": 0.3414, "step": 3430 }, { "epoch": 7.958357432041643, "grad_norm": 2.2664902210235596, "learning_rate": 1.9787680724495617e-05, "loss": 0.3422, "step": 3440 }, { "epoch": 7.981492192018508, "grad_norm": 2.529334545135498, "learning_rate": 1.9355539573251734e-05, "loss": 0.3384, "step": 3450 }, { "epoch": 8.004626951995373, "grad_norm": 1.9292296171188354, "learning_rate": 1.892766328297878e-05, "loss": 0.3313, "step": 3460 }, { "epoch": 8.027761711972238, "grad_norm": 1.851493000984192, "learning_rate": 1.8504074481801238e-05, "loss": 0.2768, "step": 3470 }, { "epoch": 8.050896471949104, "grad_norm": 2.0147864818573, "learning_rate": 1.808479557110081e-05, "loss": 0.2539, "step": 3480 }, { "epoch": 8.074031231925968, "grad_norm": 2.2376036643981934, "learning_rate": 1.7669848724331984e-05, "loss": 0.2815, "step": 3490 }, { "epoch": 8.097165991902834, "grad_norm": 2.2430434226989746, "learning_rate": 1.7259255885848946e-05, "loss": 0.2603, "step": 3500 }, { "epoch": 8.1203007518797, "grad_norm": 1.880216121673584, "learning_rate": 1.6853038769745467e-05, "loss": 0.2647, "step": 3510 }, { "epoch": 8.143435511856564, "grad_norm": 2.154496192932129, "learning_rate": 1.6451218858706374e-05, "loss": 0.2648, "step": 3520 }, { "epoch": 8.16657027183343, "grad_norm": 1.9910736083984375, "learning_rate": 1.6053817402871362e-05, "loss": 0.2737, "step": 3530 }, { "epoch": 8.189705031810295, "grad_norm": 1.7439430952072144, "learning_rate": 1.566085541871145e-05, "loss": 0.2728, "step": 3540 }, { "epoch": 8.212839791787161, "grad_norm": 1.8878188133239746, "learning_rate": 1.5272353687917196e-05, "loss": 0.2678, "step": 3550 }, { "epoch": 8.235974551764025, "grad_norm": 1.8868342638015747, "learning_rate": 1.4888332756300027e-05, "loss": 0.2629, "step": 3560 }, { "epoch": 8.259109311740891, "grad_norm": 2.0473947525024414, "learning_rate": 1.4508812932705363e-05, "loss": 0.2706, "step": 3570 }, { "epoch": 8.282244071717756, "grad_norm": 2.090017795562744, "learning_rate": 1.4133814287938706e-05, "loss": 0.2754, "step": 3580 }, { "epoch": 8.305378831694622, "grad_norm": 2.0764472484588623, "learning_rate": 1.3763356653704273e-05, "loss": 0.2801, "step": 3590 }, { "epoch": 8.328513591671486, "grad_norm": 1.978501796722412, "learning_rate": 1.339745962155613e-05, "loss": 0.2754, "step": 3600 }, { "epoch": 8.351648351648352, "grad_norm": 2.1843178272247314, "learning_rate": 1.3036142541862118e-05, "loss": 0.2734, "step": 3610 }, { "epoch": 8.374783111625216, "grad_norm": 2.240216016769409, "learning_rate": 1.2679424522780426e-05, "loss": 0.2819, "step": 3620 }, { "epoch": 8.397917871602083, "grad_norm": 2.18241548538208, "learning_rate": 1.2327324429249232e-05, "loss": 0.2744, "step": 3630 }, { "epoch": 8.421052631578947, "grad_norm": 2.1984431743621826, "learning_rate": 1.1979860881988902e-05, "loss": 0.2867, "step": 3640 }, { "epoch": 8.444187391555813, "grad_norm": 2.063441753387451, "learning_rate": 1.1637052256517244e-05, "loss": 0.2687, "step": 3650 }, { "epoch": 8.467322151532677, "grad_norm": 2.132188558578491, "learning_rate": 1.129891668217783e-05, "loss": 0.2542, "step": 3660 }, { "epoch": 8.490456911509543, "grad_norm": 2.227787971496582, "learning_rate": 1.0965472041181102e-05, "loss": 0.2849, "step": 3670 }, { "epoch": 8.513591671486408, "grad_norm": 1.9620667695999146, "learning_rate": 1.0636735967658784e-05, "loss": 0.2746, "step": 3680 }, { "epoch": 8.536726431463274, "grad_norm": 1.9825657606124878, "learning_rate": 1.0312725846731175e-05, "loss": 0.2665, "step": 3690 }, { "epoch": 8.559861191440138, "grad_norm": 1.9591861963272095, "learning_rate": 9.993458813587885e-06, "loss": 0.2598, "step": 3700 }, { "epoch": 8.582995951417004, "grad_norm": 2.2019407749176025, "learning_rate": 9.678951752581578e-06, "loss": 0.2801, "step": 3710 }, { "epoch": 8.606130711393869, "grad_norm": 2.131798267364502, "learning_rate": 9.369221296335006e-06, "loss": 0.2778, "step": 3720 }, { "epoch": 8.629265471370735, "grad_norm": 2.081733465194702, "learning_rate": 9.064283824861486e-06, "loss": 0.2774, "step": 3730 }, { "epoch": 8.652400231347599, "grad_norm": 2.1532142162323, "learning_rate": 8.764155464698597e-06, "loss": 0.2754, "step": 3740 }, { "epoch": 8.675534991324465, "grad_norm": 2.2483901977539062, "learning_rate": 8.46885208805529e-06, "loss": 0.2618, "step": 3750 }, { "epoch": 8.698669751301331, "grad_norm": 1.9993518590927124, "learning_rate": 8.178389311972612e-06, "loss": 0.2892, "step": 3760 }, { "epoch": 8.721804511278195, "grad_norm": 2.1293857097625732, "learning_rate": 7.892782497497642e-06, "loss": 0.2909, "step": 3770 }, { "epoch": 8.744939271255062, "grad_norm": 1.9576681852340698, "learning_rate": 7.612046748871327e-06, "loss": 0.2783, "step": 3780 }, { "epoch": 8.768074031231926, "grad_norm": 1.7659740447998047, "learning_rate": 7.336196912729487e-06, "loss": 0.2757, "step": 3790 }, { "epoch": 8.791208791208792, "grad_norm": 1.8570443391799927, "learning_rate": 7.0652475773177464e-06, "loss": 0.262, "step": 3800 }, { "epoch": 8.814343551185656, "grad_norm": 1.9978190660476685, "learning_rate": 6.7992130717201564e-06, "loss": 0.2686, "step": 3810 }, { "epoch": 8.837478311162522, "grad_norm": 2.242790460586548, "learning_rate": 6.538107465101162e-06, "loss": 0.2857, "step": 3820 }, { "epoch": 8.860613071139387, "grad_norm": 2.2894692420959473, "learning_rate": 6.281944565961773e-06, "loss": 0.2762, "step": 3830 }, { "epoch": 8.883747831116253, "grad_norm": 1.879543662071228, "learning_rate": 6.030737921409169e-06, "loss": 0.275, "step": 3840 }, { "epoch": 8.906882591093117, "grad_norm": 1.8957421779632568, "learning_rate": 5.784500816440353e-06, "loss": 0.2736, "step": 3850 }, { "epoch": 8.930017351069983, "grad_norm": 2.0215442180633545, "learning_rate": 5.543246273239533e-06, "loss": 0.2698, "step": 3860 }, { "epoch": 8.953152111046847, "grad_norm": 2.278769016265869, "learning_rate": 5.306987050489442e-06, "loss": 0.2832, "step": 3870 }, { "epoch": 8.976286871023714, "grad_norm": 2.370384931564331, "learning_rate": 5.075735642696611e-06, "loss": 0.2739, "step": 3880 }, { "epoch": 8.999421631000578, "grad_norm": 2.1495909690856934, "learning_rate": 4.849504279530636e-06, "loss": 0.2664, "step": 3890 }, { "epoch": 9.022556390977444, "grad_norm": 1.8931546211242676, "learning_rate": 4.628304925177318e-06, "loss": 0.251, "step": 3900 }, { "epoch": 9.045691150954308, "grad_norm": 1.964045763015747, "learning_rate": 4.412149277706046e-06, "loss": 0.2394, "step": 3910 }, { "epoch": 9.068825910931174, "grad_norm": 1.8425894975662231, "learning_rate": 4.20104876845111e-06, "loss": 0.2368, "step": 3920 }, { "epoch": 9.091960670908039, "grad_norm": 1.8906612396240234, "learning_rate": 3.99501456140714e-06, "loss": 0.2517, "step": 3930 }, { "epoch": 9.115095430884905, "grad_norm": 1.9606223106384277, "learning_rate": 3.7940575526386857e-06, "loss": 0.2638, "step": 3940 }, { "epoch": 9.13823019086177, "grad_norm": 2.0585832595825195, "learning_rate": 3.598188369704036e-06, "loss": 0.2465, "step": 3950 }, { "epoch": 9.161364950838635, "grad_norm": 2.0932703018188477, "learning_rate": 3.40741737109318e-06, "loss": 0.2454, "step": 3960 }, { "epoch": 9.1844997108155, "grad_norm": 1.9221563339233398, "learning_rate": 3.2217546456799086e-06, "loss": 0.2387, "step": 3970 }, { "epoch": 9.207634470792366, "grad_norm": 1.973889946937561, "learning_rate": 3.0412100121884e-06, "loss": 0.2475, "step": 3980 }, { "epoch": 9.23076923076923, "grad_norm": 1.8199151754379272, "learning_rate": 2.8657930186738567e-06, "loss": 0.2364, "step": 3990 }, { "epoch": 9.253903990746096, "grad_norm": 1.847708821296692, "learning_rate": 2.6955129420176196e-06, "loss": 0.2431, "step": 4000 }, { "epoch": 9.277038750722962, "grad_norm": 1.6919292211532593, "learning_rate": 2.530378787436527e-06, "loss": 0.2546, "step": 4010 }, { "epoch": 9.300173510699826, "grad_norm": 1.7380826473236084, "learning_rate": 2.3703992880066638e-06, "loss": 0.2471, "step": 4020 }, { "epoch": 9.323308270676693, "grad_norm": 2.1914947032928467, "learning_rate": 2.2155829042015962e-06, "loss": 0.2406, "step": 4030 }, { "epoch": 9.346443030653557, "grad_norm": 1.788342833518982, "learning_rate": 2.0659378234448525e-06, "loss": 0.2292, "step": 4040 }, { "epoch": 9.369577790630423, "grad_norm": 1.953272819519043, "learning_rate": 1.921471959676957e-06, "loss": 0.2393, "step": 4050 }, { "epoch": 9.392712550607287, "grad_norm": 2.092925548553467, "learning_rate": 1.7821929529369342e-06, "loss": 0.2455, "step": 4060 }, { "epoch": 9.415847310584153, "grad_norm": 1.312064528465271, "learning_rate": 1.648108168958229e-06, "loss": 0.2464, "step": 4070 }, { "epoch": 9.438982070561018, "grad_norm": 1.8902608156204224, "learning_rate": 1.5192246987791981e-06, "loss": 0.2503, "step": 4080 }, { "epoch": 9.462116830537884, "grad_norm": 2.0683326721191406, "learning_rate": 1.3955493583680868e-06, "loss": 0.2532, "step": 4090 }, { "epoch": 9.485251590514748, "grad_norm": 2.037275552749634, "learning_rate": 1.2770886882625955e-06, "loss": 0.2494, "step": 4100 }, { "epoch": 9.508386350491614, "grad_norm": 1.884734034538269, "learning_rate": 1.163848953223934e-06, "loss": 0.2427, "step": 4110 }, { "epoch": 9.531521110468478, "grad_norm": 2.0720531940460205, "learning_rate": 1.055836141905553e-06, "loss": 0.2361, "step": 4120 }, { "epoch": 9.554655870445345, "grad_norm": 1.7988377809524536, "learning_rate": 9.530559665364202e-07, "loss": 0.2524, "step": 4130 }, { "epoch": 9.577790630422209, "grad_norm": 2.159759044647217, "learning_rate": 8.555138626189618e-07, "loss": 0.2562, "step": 4140 }, { "epoch": 9.600925390399075, "grad_norm": 2.3228530883789062, "learning_rate": 7.632149886415363e-07, "loss": 0.2326, "step": 4150 }, { "epoch": 9.62406015037594, "grad_norm": 2.210181713104248, "learning_rate": 6.761642258056978e-07, "loss": 0.2628, "step": 4160 }, { "epoch": 9.647194910352805, "grad_norm": 1.6865817308425903, "learning_rate": 5.943661777680354e-07, "loss": 0.2409, "step": 4170 }, { "epoch": 9.67032967032967, "grad_norm": 1.9101097583770752, "learning_rate": 5.178251703967018e-07, "loss": 0.2453, "step": 4180 }, { "epoch": 9.693464430306536, "grad_norm": 2.2354347705841064, "learning_rate": 4.465452515426072e-07, "loss": 0.2339, "step": 4190 }, { "epoch": 9.7165991902834, "grad_norm": 2.173842668533325, "learning_rate": 3.805301908254455e-07, "loss": 0.2369, "step": 4200 }, { "epoch": 9.739733950260266, "grad_norm": 1.74326491355896, "learning_rate": 3.197834794342436e-07, "loss": 0.2431, "step": 4210 }, { "epoch": 9.76286871023713, "grad_norm": 1.731026291847229, "learning_rate": 2.643083299427751e-07, "loss": 0.2587, "step": 4220 }, { "epoch": 9.786003470213997, "grad_norm": 2.085442066192627, "learning_rate": 2.141076761396521e-07, "loss": 0.2525, "step": 4230 }, { "epoch": 9.809138230190861, "grad_norm": 1.921111822128296, "learning_rate": 1.6918417287318245e-07, "loss": 0.2561, "step": 4240 }, { "epoch": 9.832272990167727, "grad_norm": 1.7905348539352417, "learning_rate": 1.295401959109599e-07, "loss": 0.2425, "step": 4250 }, { "epoch": 9.855407750144593, "grad_norm": 2.0938756465911865, "learning_rate": 9.517784181422019e-08, "loss": 0.2486, "step": 4260 }, { "epoch": 9.878542510121457, "grad_norm": 2.0553646087646484, "learning_rate": 6.609892782699633e-08, "loss": 0.2397, "step": 4270 }, { "epoch": 9.901677270098324, "grad_norm": 1.8413350582122803, "learning_rate": 4.230499177994007e-08, "loss": 0.2504, "step": 4280 }, { "epoch": 9.924812030075188, "grad_norm": 1.7618731260299683, "learning_rate": 2.379729200908676e-08, "loss": 0.2491, "step": 4290 }, { "epoch": 9.947946790052054, "grad_norm": 1.7408370971679688, "learning_rate": 1.0576807289253143e-08, "loss": 0.2448, "step": 4300 }, { "epoch": 9.971081550028918, "grad_norm": 1.9657200574874878, "learning_rate": 2.6442367822565596e-09, "loss": 0.2623, "step": 4310 }, { "epoch": 9.994216310005784, "grad_norm": 1.993733525276184, "learning_rate": 0.0, "loss": 0.2453, "step": 4320 }, { "epoch": 9.994216310005784, "step": 4320, "total_flos": 2.1249547460149248e+17, "train_loss": 0.887424631913503, "train_runtime": 17181.5838, "train_samples_per_second": 1.006, "train_steps_per_second": 0.251 } ], "logging_steps": 10, "max_steps": 4320, "num_input_tokens_seen": 0, "num_train_epochs": 10, "save_steps": 500, "total_flos": 2.1249547460149248e+17, "train_batch_size": 1, "trial_name": null, "trial_params": null }