{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 2554, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.007832770352964214, "grad_norm": 0.7764872908592224, "learning_rate": 7.031250000000001e-06, "loss": 2.2154, "step": 10 }, { "epoch": 0.01566554070592843, "grad_norm": 0.6605860590934753, "learning_rate": 1.484375e-05, "loss": 2.1893, "step": 20 }, { "epoch": 0.02349831105889264, "grad_norm": 0.283372163772583, "learning_rate": 2.2656250000000002e-05, "loss": 2.0587, "step": 30 }, { "epoch": 0.03133108141185686, "grad_norm": 0.26019203662872314, "learning_rate": 3.0468750000000002e-05, "loss": 1.9259, "step": 40 }, { "epoch": 0.03916385176482107, "grad_norm": 0.25377678871154785, "learning_rate": 3.828125e-05, "loss": 1.8273, "step": 50 }, { "epoch": 0.04699662211778528, "grad_norm": 0.2096589356660843, "learning_rate": 4.609375e-05, "loss": 1.7443, "step": 60 }, { "epoch": 0.0548293924707495, "grad_norm": 0.2725999057292938, "learning_rate": 5.3906250000000006e-05, "loss": 1.6288, "step": 70 }, { "epoch": 0.06266216282371372, "grad_norm": 0.2633262276649475, "learning_rate": 6.171875e-05, "loss": 1.6316, "step": 80 }, { "epoch": 0.07049493317667793, "grad_norm": 0.29914507269859314, "learning_rate": 6.953125e-05, "loss": 1.605, "step": 90 }, { "epoch": 0.07832770352964213, "grad_norm": 0.3007325828075409, "learning_rate": 7.734375e-05, "loss": 1.5412, "step": 100 }, { "epoch": 0.08616047388260635, "grad_norm": 0.2865763008594513, "learning_rate": 8.515625e-05, "loss": 1.4804, "step": 110 }, { "epoch": 0.09399324423557057, "grad_norm": 0.3008711636066437, "learning_rate": 9.296875e-05, "loss": 1.5436, "step": 120 }, { "epoch": 0.10182601458853478, "grad_norm": 0.29449915885925293, "learning_rate": 0.00010078125, "loss": 1.4846, "step": 130 }, { "epoch": 0.109658784941499, "grad_norm": 0.2944643497467041, "learning_rate": 0.00010859375000000001, "loss": 1.4749, "step": 140 }, { "epoch": 0.11749155529446322, "grad_norm": 0.30779531598091125, "learning_rate": 0.00011640625, "loss": 1.4789, "step": 150 }, { "epoch": 0.12532432564742743, "grad_norm": 0.2892812192440033, "learning_rate": 0.00012421875, "loss": 1.4714, "step": 160 }, { "epoch": 0.13315709600039163, "grad_norm": 0.3098798990249634, "learning_rate": 0.00013203125000000001, "loss": 1.4716, "step": 170 }, { "epoch": 0.14098986635335586, "grad_norm": 0.3093400001525879, "learning_rate": 0.00013984375, "loss": 1.4419, "step": 180 }, { "epoch": 0.14882263670632007, "grad_norm": 0.2720882296562195, "learning_rate": 0.00014765625, "loss": 1.4468, "step": 190 }, { "epoch": 0.15665540705928427, "grad_norm": 0.2835504710674286, "learning_rate": 0.00015546875, "loss": 1.4678, "step": 200 }, { "epoch": 0.1644881774122485, "grad_norm": 0.2735293209552765, "learning_rate": 0.00016328125000000001, "loss": 1.4684, "step": 210 }, { "epoch": 0.1723209477652127, "grad_norm": 0.2720692455768585, "learning_rate": 0.00017109375, "loss": 1.4794, "step": 220 }, { "epoch": 0.18015371811817693, "grad_norm": 0.2576466500759125, "learning_rate": 0.00017890625, "loss": 1.478, "step": 230 }, { "epoch": 0.18798648847114113, "grad_norm": 0.2633955776691437, "learning_rate": 0.00018671875, "loss": 1.4581, "step": 240 }, { "epoch": 0.19581925882410536, "grad_norm": 0.2795034646987915, "learning_rate": 0.00019453125000000002, "loss": 1.4526, "step": 250 }, { "epoch": 0.20365202917706957, "grad_norm": 0.2619614601135254, "learning_rate": 0.00019999915896977904, "loss": 1.4462, "step": 260 }, { "epoch": 0.21148479953003377, "grad_norm": 0.23947100341320038, "learning_rate": 0.00019998420771494732, "loss": 1.4216, "step": 270 }, { "epoch": 0.219317569882998, "grad_norm": 0.289479523897171, "learning_rate": 0.00019995057011600097, "loss": 1.4077, "step": 280 }, { "epoch": 0.2271503402359622, "grad_norm": 0.22916744649410248, "learning_rate": 0.00019989825245957037, "loss": 1.4487, "step": 290 }, { "epoch": 0.23498311058892643, "grad_norm": 0.2572559118270874, "learning_rate": 0.0001998272645234569, "loss": 1.4164, "step": 300 }, { "epoch": 0.24281588094189063, "grad_norm": 0.265937864780426, "learning_rate": 0.00019973761957480578, "loss": 1.3709, "step": 310 }, { "epoch": 0.25064865129485486, "grad_norm": 0.25126948952674866, "learning_rate": 0.00019962933436762646, "loss": 1.3979, "step": 320 }, { "epoch": 0.25848142164781907, "grad_norm": 0.2443549633026123, "learning_rate": 0.00019950242913966136, "loss": 1.4048, "step": 330 }, { "epoch": 0.26631419200078327, "grad_norm": 0.23647847771644592, "learning_rate": 0.00019935692760860362, "loss": 1.4457, "step": 340 }, { "epoch": 0.27414696235374747, "grad_norm": 0.2688600420951843, "learning_rate": 0.00019919285696766453, "loss": 1.4205, "step": 350 }, { "epoch": 0.2819797327067117, "grad_norm": 0.3046218156814575, "learning_rate": 0.00019901024788049107, "loss": 1.388, "step": 360 }, { "epoch": 0.28981250305967593, "grad_norm": 0.25633004307746887, "learning_rate": 0.00019880913447543533, "loss": 1.4019, "step": 370 }, { "epoch": 0.29764527341264013, "grad_norm": 0.27422916889190674, "learning_rate": 0.00019858955433917604, "loss": 1.4325, "step": 380 }, { "epoch": 0.30547804376560433, "grad_norm": 0.23750437796115875, "learning_rate": 0.00019835154850969384, "loss": 1.4336, "step": 390 }, { "epoch": 0.31331081411856854, "grad_norm": 0.26217496395111084, "learning_rate": 0.00019809516146860177, "loss": 1.438, "step": 400 }, { "epoch": 0.3211435844715328, "grad_norm": 0.24527603387832642, "learning_rate": 0.00019782044113283181, "loss": 1.3965, "step": 410 }, { "epoch": 0.328976354824497, "grad_norm": 0.21317604184150696, "learning_rate": 0.00019752743884567944, "loss": 1.3964, "step": 420 }, { "epoch": 0.3368091251774612, "grad_norm": 0.3109080195426941, "learning_rate": 0.0001972162093672082, "loss": 1.4323, "step": 430 }, { "epoch": 0.3446418955304254, "grad_norm": 0.2410540133714676, "learning_rate": 0.00019688681086401527, "loss": 1.42, "step": 440 }, { "epoch": 0.3524746658833896, "grad_norm": 0.22951515018939972, "learning_rate": 0.00019653930489836054, "loss": 1.3876, "step": 450 }, { "epoch": 0.36030743623635386, "grad_norm": 0.274689644575119, "learning_rate": 0.0001961737564166611, "loss": 1.3841, "step": 460 }, { "epoch": 0.36814020658931806, "grad_norm": 0.1923462450504303, "learning_rate": 0.00019579023373735322, "loss": 1.3814, "step": 470 }, { "epoch": 0.37597297694228227, "grad_norm": 0.22620543837547302, "learning_rate": 0.000195388808538124, "loss": 1.3986, "step": 480 }, { "epoch": 0.38380574729524647, "grad_norm": 0.2504560947418213, "learning_rate": 0.00019496955584251548, "loss": 1.4099, "step": 490 }, { "epoch": 0.3916385176482107, "grad_norm": 0.2991662919521332, "learning_rate": 0.0001945325540059032, "loss": 1.4171, "step": 500 }, { "epoch": 0.3916385176482107, "eval_loss": 1.401666522026062, "eval_runtime": 132.1967, "eval_samples_per_second": 8.139, "eval_steps_per_second": 8.139, "step": 500 }, { "epoch": 0.39947128800117493, "grad_norm": 0.22948013246059418, "learning_rate": 0.00019407788470085207, "loss": 1.3921, "step": 510 }, { "epoch": 0.40730405835413913, "grad_norm": 0.2110731303691864, "learning_rate": 0.00019360563290185247, "loss": 1.4063, "step": 520 }, { "epoch": 0.41513682870710333, "grad_norm": 0.23307982087135315, "learning_rate": 0.00019311588686943897, "loss": 1.3902, "step": 530 }, { "epoch": 0.42296959906006754, "grad_norm": 0.228043794631958, "learning_rate": 0.00019260873813369515, "loss": 1.391, "step": 540 }, { "epoch": 0.4308023694130318, "grad_norm": 0.22130551934242249, "learning_rate": 0.0001920842814771473, "loss": 1.3923, "step": 550 }, { "epoch": 0.438635139765996, "grad_norm": 0.2160315215587616, "learning_rate": 0.0001915426149170502, "loss": 1.3863, "step": 560 }, { "epoch": 0.4464679101189602, "grad_norm": 0.2282577008008957, "learning_rate": 0.00019098383968706823, "loss": 1.3799, "step": 570 }, { "epoch": 0.4543006804719244, "grad_norm": 0.25913530588150024, "learning_rate": 0.0001904080602183558, "loss": 1.3897, "step": 580 }, { "epoch": 0.4621334508248886, "grad_norm": 0.22399315237998962, "learning_rate": 0.00018981538412003978, "loss": 1.4013, "step": 590 }, { "epoch": 0.46996622117785286, "grad_norm": 0.19405752420425415, "learning_rate": 0.00018920592215910796, "loss": 1.4106, "step": 600 }, { "epoch": 0.47779899153081706, "grad_norm": 0.25263646245002747, "learning_rate": 0.0001885797882397077, "loss": 1.3589, "step": 610 }, { "epoch": 0.48563176188378127, "grad_norm": 0.18886186182498932, "learning_rate": 0.00018793709938185789, "loss": 1.3609, "step": 620 }, { "epoch": 0.49346453223674547, "grad_norm": 0.19153770804405212, "learning_rate": 0.00018727797569957887, "loss": 1.403, "step": 630 }, { "epoch": 0.5012973025897097, "grad_norm": 0.22722338140010834, "learning_rate": 0.00018660254037844388, "loss": 1.3901, "step": 640 }, { "epoch": 0.5091300729426739, "grad_norm": 0.2029528021812439, "learning_rate": 0.00018591091965255654, "loss": 1.3842, "step": 650 }, { "epoch": 0.5169628432956381, "grad_norm": 0.17809070646762848, "learning_rate": 0.0001852032427809586, "loss": 1.3736, "step": 660 }, { "epoch": 0.5247956136486024, "grad_norm": 0.20622004568576813, "learning_rate": 0.00018447964202347237, "loss": 1.3794, "step": 670 }, { "epoch": 0.5326283840015665, "grad_norm": 0.22416748106479645, "learning_rate": 0.00018374025261598225, "loss": 1.3754, "step": 680 }, { "epoch": 0.5404611543545308, "grad_norm": 0.20559127628803253, "learning_rate": 0.0001829852127451601, "loss": 1.3945, "step": 690 }, { "epoch": 0.5482939247074949, "grad_norm": 0.22158539295196533, "learning_rate": 0.0001822146635226391, "loss": 1.3768, "step": 700 }, { "epoch": 0.5561266950604592, "grad_norm": 0.22523440420627594, "learning_rate": 0.00018142874895864112, "loss": 1.3788, "step": 710 }, { "epoch": 0.5639594654134235, "grad_norm": 0.22926659882068634, "learning_rate": 0.00018062761593506198, "loss": 1.387, "step": 720 }, { "epoch": 0.5717922357663876, "grad_norm": 0.19991280138492584, "learning_rate": 0.0001798114141780205, "loss": 1.3554, "step": 730 }, { "epoch": 0.5796250061193519, "grad_norm": 0.20266345143318176, "learning_rate": 0.00017898029622987556, "loss": 1.3558, "step": 740 }, { "epoch": 0.587457776472316, "grad_norm": 0.24142096936702728, "learning_rate": 0.00017813441742071712, "loss": 1.382, "step": 750 }, { "epoch": 0.5952905468252803, "grad_norm": 0.2066735178232193, "learning_rate": 0.0001772739358393361, "loss": 1.3937, "step": 760 }, { "epoch": 0.6031233171782445, "grad_norm": 0.1943921595811844, "learning_rate": 0.00017639901230367868, "loss": 1.3681, "step": 770 }, { "epoch": 0.6109560875312087, "grad_norm": 0.2078547179698944, "learning_rate": 0.00017550981033079052, "loss": 1.3838, "step": 780 }, { "epoch": 0.6187888578841729, "grad_norm": 0.21833692491054535, "learning_rate": 0.00017460649610625666, "loss": 1.356, "step": 790 }, { "epoch": 0.6266216282371371, "grad_norm": 0.23712702095508575, "learning_rate": 0.0001736892384531426, "loss": 1.3575, "step": 800 }, { "epoch": 0.6344543985901013, "grad_norm": 0.24740666151046753, "learning_rate": 0.00017275820880044228, "loss": 1.4028, "step": 810 }, { "epoch": 0.6422871689430656, "grad_norm": 0.20247240364551544, "learning_rate": 0.00017181358115103953, "loss": 1.3824, "step": 820 }, { "epoch": 0.6501199392960297, "grad_norm": 0.23941560089588165, "learning_rate": 0.0001708555320491878, "loss": 1.3772, "step": 830 }, { "epoch": 0.657952709648994, "grad_norm": 0.20534786581993103, "learning_rate": 0.00016988424054751564, "loss": 1.3617, "step": 840 }, { "epoch": 0.6657854800019581, "grad_norm": 0.22657494246959686, "learning_rate": 0.00016889988817356274, "loss": 1.3929, "step": 850 }, { "epoch": 0.6736182503549224, "grad_norm": 0.20961667597293854, "learning_rate": 0.0001679026588958538, "loss": 1.3624, "step": 860 }, { "epoch": 0.6814510207078867, "grad_norm": 0.21676991879940033, "learning_rate": 0.00016689273908951612, "loss": 1.3423, "step": 870 }, { "epoch": 0.6892837910608508, "grad_norm": 0.22844044864177704, "learning_rate": 0.00016587031750144756, "loss": 1.3799, "step": 880 }, { "epoch": 0.6971165614138151, "grad_norm": 0.22429795563220978, "learning_rate": 0.00016483558521504068, "loss": 1.377, "step": 890 }, { "epoch": 0.7049493317667792, "grad_norm": 0.1610354334115982, "learning_rate": 0.000163788735614471, "loss": 1.3587, "step": 900 }, { "epoch": 0.7127821021197435, "grad_norm": 0.22684766352176666, "learning_rate": 0.0001627299643485547, "loss": 1.3739, "step": 910 }, { "epoch": 0.7206148724727077, "grad_norm": 0.20572352409362793, "learning_rate": 0.00016165946929418322, "loss": 1.3475, "step": 920 }, { "epoch": 0.7284476428256719, "grad_norm": 0.20187753438949585, "learning_rate": 0.00016057745051934133, "loss": 1.3645, "step": 930 }, { "epoch": 0.7362804131786361, "grad_norm": 0.2022079974412918, "learning_rate": 0.00015948411024571622, "loss": 1.3362, "step": 940 }, { "epoch": 0.7441131835316004, "grad_norm": 0.19844278693199158, "learning_rate": 0.0001583796528109033, "loss": 1.3852, "step": 950 }, { "epoch": 0.7519459538845645, "grad_norm": 0.23300006985664368, "learning_rate": 0.00015726428463021716, "loss": 1.3492, "step": 960 }, { "epoch": 0.7597787242375288, "grad_norm": 0.20079776644706726, "learning_rate": 0.0001561382141581139, "loss": 1.3599, "step": 970 }, { "epoch": 0.7676114945904929, "grad_norm": 0.22002668678760529, "learning_rate": 0.00015500165184923248, "loss": 1.3934, "step": 980 }, { "epoch": 0.7754442649434572, "grad_norm": 0.21567285060882568, "learning_rate": 0.00015385481011906228, "loss": 1.373, "step": 990 }, { "epoch": 0.7832770352964215, "grad_norm": 0.19532336294651031, "learning_rate": 0.00015269790330424413, "loss": 1.3642, "step": 1000 }, { "epoch": 0.7832770352964215, "eval_loss": 1.3580660820007324, "eval_runtime": 132.097, "eval_samples_per_second": 8.146, "eval_steps_per_second": 8.146, "step": 1000 }, { "epoch": 0.7911098056493856, "grad_norm": 0.19254837930202484, "learning_rate": 0.00015153114762251222, "loss": 1.3886, "step": 1010 }, { "epoch": 0.7989425760023499, "grad_norm": 0.23445847630500793, "learning_rate": 0.00015035476113228467, "loss": 1.3392, "step": 1020 }, { "epoch": 0.806775346355314, "grad_norm": 0.21545809507369995, "learning_rate": 0.00014916896369190975, "loss": 1.3544, "step": 1030 }, { "epoch": 0.8146081167082783, "grad_norm": 0.2504936754703522, "learning_rate": 0.00014797397691857612, "loss": 1.3757, "step": 1040 }, { "epoch": 0.8224408870612425, "grad_norm": 0.21092313528060913, "learning_rate": 0.00014677002414689402, "loss": 1.3721, "step": 1050 }, { "epoch": 0.8302736574142067, "grad_norm": 0.22991657257080078, "learning_rate": 0.0001455573303871556, "loss": 1.377, "step": 1060 }, { "epoch": 0.8381064277671709, "grad_norm": 0.20714305341243744, "learning_rate": 0.00014433612228328214, "loss": 1.3704, "step": 1070 }, { "epoch": 0.8459391981201351, "grad_norm": 0.22241558134555817, "learning_rate": 0.0001431066280704658, "loss": 1.3668, "step": 1080 }, { "epoch": 0.8537719684730993, "grad_norm": 0.19797885417938232, "learning_rate": 0.0001418690775325142, "loss": 1.3762, "step": 1090 }, { "epoch": 0.8616047388260636, "grad_norm": 0.22674763202667236, "learning_rate": 0.0001406237019589053, "loss": 1.3402, "step": 1100 }, { "epoch": 0.8694375091790277, "grad_norm": 0.18697579205036163, "learning_rate": 0.00013937073410156119, "loss": 1.3703, "step": 1110 }, { "epoch": 0.877270279531992, "grad_norm": 0.19405877590179443, "learning_rate": 0.0001381104081313483, "loss": 1.3731, "step": 1120 }, { "epoch": 0.8851030498849561, "grad_norm": 0.252853125333786, "learning_rate": 0.0001368429595943124, "loss": 1.3417, "step": 1130 }, { "epoch": 0.8929358202379204, "grad_norm": 0.2199040800333023, "learning_rate": 0.000135568625367657, "loss": 1.3142, "step": 1140 }, { "epoch": 0.9007685905908847, "grad_norm": 0.22553178668022156, "learning_rate": 0.00013428764361547238, "loss": 1.3341, "step": 1150 }, { "epoch": 0.9086013609438488, "grad_norm": 0.20995256304740906, "learning_rate": 0.0001330002537442246, "loss": 1.2806, "step": 1160 }, { "epoch": 0.9164341312968131, "grad_norm": 0.21296226978302002, "learning_rate": 0.0001317066963580119, "loss": 1.3213, "step": 1170 }, { "epoch": 0.9242669016497772, "grad_norm": 0.1688837707042694, "learning_rate": 0.0001304072132135977, "loss": 1.347, "step": 1180 }, { "epoch": 0.9320996720027415, "grad_norm": 0.21325655281543732, "learning_rate": 0.00012910204717522805, "loss": 1.3446, "step": 1190 }, { "epoch": 0.9399324423557057, "grad_norm": 0.22042647004127502, "learning_rate": 0.0001277914421692415, "loss": 1.2947, "step": 1200 }, { "epoch": 0.9477652127086699, "grad_norm": 0.2434263974428177, "learning_rate": 0.0001264756431384816, "loss": 1.377, "step": 1210 }, { "epoch": 0.9555979830616341, "grad_norm": 0.18997617065906525, "learning_rate": 0.00012515489599651848, "loss": 1.3425, "step": 1220 }, { "epoch": 0.9634307534145983, "grad_norm": 0.20458567142486572, "learning_rate": 0.00012382944758168937, "loss": 1.3514, "step": 1230 }, { "epoch": 0.9712635237675625, "grad_norm": 0.2017565369606018, "learning_rate": 0.00012249954561096627, "loss": 1.3151, "step": 1240 }, { "epoch": 0.9790962941205268, "grad_norm": 0.2048238217830658, "learning_rate": 0.0001211654386336595, "loss": 1.3733, "step": 1250 }, { "epoch": 0.9869290644734909, "grad_norm": 0.25371628999710083, "learning_rate": 0.00011982737598496536, "loss": 1.3546, "step": 1260 }, { "epoch": 0.9947618348264552, "grad_norm": 0.23323997855186462, "learning_rate": 0.00011848560773936736, "loss": 1.3052, "step": 1270 }, { "epoch": 1.0023498311058892, "grad_norm": 0.18340709805488586, "learning_rate": 0.00011714038466389891, "loss": 1.3483, "step": 1280 }, { "epoch": 1.0101826014588535, "grad_norm": 0.19518490135669708, "learning_rate": 0.00011579195817127684, "loss": 1.3034, "step": 1290 }, { "epoch": 1.0180153718118177, "grad_norm": 0.21178381145000458, "learning_rate": 0.00011444058027291404, "loss": 1.2866, "step": 1300 }, { "epoch": 1.0258481421647818, "grad_norm": 0.20926272869110107, "learning_rate": 0.00011308650353182036, "loss": 1.3646, "step": 1310 }, { "epoch": 1.0336809125177462, "grad_norm": 0.21566914021968842, "learning_rate": 0.00011172998101540036, "loss": 1.3352, "step": 1320 }, { "epoch": 1.0415136828707103, "grad_norm": 0.20395275950431824, "learning_rate": 0.00011037126624815675, "loss": 1.28, "step": 1330 }, { "epoch": 1.0493464532236745, "grad_norm": 0.2022181898355484, "learning_rate": 0.00010901061316430847, "loss": 1.328, "step": 1340 }, { "epoch": 1.0571792235766388, "grad_norm": 0.23947516083717346, "learning_rate": 0.00010764827606033221, "loss": 1.2655, "step": 1350 }, { "epoch": 1.065011993929603, "grad_norm": 0.23035083711147308, "learning_rate": 0.00010628450954743616, "loss": 1.3069, "step": 1360 }, { "epoch": 1.0728447642825671, "grad_norm": 0.22558419406414032, "learning_rate": 0.00010491956850397495, "loss": 1.3163, "step": 1370 }, { "epoch": 1.0806775346355315, "grad_norm": 0.20729310810565948, "learning_rate": 0.0001035537080278147, "loss": 1.3001, "step": 1380 }, { "epoch": 1.0885103049884957, "grad_norm": 0.21035364270210266, "learning_rate": 0.00010218718338865703, "loss": 1.2984, "step": 1390 }, { "epoch": 1.0963430753414598, "grad_norm": 0.25751104950904846, "learning_rate": 0.00010082024998033092, "loss": 1.2686, "step": 1400 }, { "epoch": 1.104175845694424, "grad_norm": 0.22621549665927887, "learning_rate": 9.94531632730615e-05, "loss": 1.2769, "step": 1410 }, { "epoch": 1.1120086160473883, "grad_norm": 0.2120325118303299, "learning_rate": 9.808617876572433e-05, "loss": 1.3278, "step": 1420 }, { "epoch": 1.1198413864003525, "grad_norm": 0.22484582662582397, "learning_rate": 9.671955193809452e-05, "loss": 1.254, "step": 1430 }, { "epoch": 1.1276741567533166, "grad_norm": 0.2620067298412323, "learning_rate": 9.535353820309943e-05, "loss": 1.2603, "step": 1440 }, { "epoch": 1.135506927106281, "grad_norm": 0.22046127915382385, "learning_rate": 9.398839285908363e-05, "loss": 1.2938, "step": 1450 }, { "epoch": 1.1433396974592451, "grad_norm": 0.23320536315441132, "learning_rate": 9.262437104209555e-05, "loss": 1.2498, "step": 1460 }, { "epoch": 1.1511724678122093, "grad_norm": 0.20881006121635437, "learning_rate": 9.126172767820446e-05, "loss": 1.2726, "step": 1470 }, { "epoch": 1.1590052381651734, "grad_norm": 0.21844594180583954, "learning_rate": 8.990071743585622e-05, "loss": 1.301, "step": 1480 }, { "epoch": 1.1668380085181378, "grad_norm": 0.2437780499458313, "learning_rate": 8.854159467827807e-05, "loss": 1.2808, "step": 1490 }, { "epoch": 1.174670778871102, "grad_norm": 0.2370881587266922, "learning_rate": 8.718461341593938e-05, "loss": 1.238, "step": 1500 }, { "epoch": 1.174670778871102, "eval_loss": 1.338244080543518, "eval_runtime": 132.0519, "eval_samples_per_second": 8.148, "eval_steps_per_second": 8.148, "step": 1500 }, { "epoch": 1.1825035492240663, "grad_norm": 0.22426699101924896, "learning_rate": 8.58300272590793e-05, "loss": 1.2714, "step": 1510 }, { "epoch": 1.1903363195770305, "grad_norm": 0.25031834840774536, "learning_rate": 8.447808937030856e-05, "loss": 1.3078, "step": 1520 }, { "epoch": 1.1981690899299946, "grad_norm": 0.22611552476882935, "learning_rate": 8.312905241729533e-05, "loss": 1.3119, "step": 1530 }, { "epoch": 1.2060018602829587, "grad_norm": 0.23634226620197296, "learning_rate": 8.178316852554317e-05, "loss": 1.321, "step": 1540 }, { "epoch": 1.2138346306359231, "grad_norm": 0.23724697530269623, "learning_rate": 8.044068923127087e-05, "loss": 1.2899, "step": 1550 }, { "epoch": 1.2216674009888873, "grad_norm": 0.2446095198392868, "learning_rate": 7.910186543440181e-05, "loss": 1.3279, "step": 1560 }, { "epoch": 1.2295001713418514, "grad_norm": 0.24685195088386536, "learning_rate": 7.776694735167276e-05, "loss": 1.2999, "step": 1570 }, { "epoch": 1.2373329416948158, "grad_norm": 0.19881027936935425, "learning_rate": 7.643618446986988e-05, "loss": 1.2848, "step": 1580 }, { "epoch": 1.24516571204778, "grad_norm": 0.2853754758834839, "learning_rate": 7.510982549920171e-05, "loss": 1.277, "step": 1590 }, { "epoch": 1.252998482400744, "grad_norm": 0.24037949740886688, "learning_rate": 7.378811832681671e-05, "loss": 1.2568, "step": 1600 }, { "epoch": 1.2608312527537082, "grad_norm": 0.2148653119802475, "learning_rate": 7.247130997047511e-05, "loss": 1.2616, "step": 1610 }, { "epoch": 1.2686640231066726, "grad_norm": 0.22323636710643768, "learning_rate": 7.11596465323828e-05, "loss": 1.2845, "step": 1620 }, { "epoch": 1.2764967934596367, "grad_norm": 0.2125261425971985, "learning_rate": 6.985337315319676e-05, "loss": 1.2479, "step": 1630 }, { "epoch": 1.284329563812601, "grad_norm": 0.24080610275268555, "learning_rate": 6.855273396620998e-05, "loss": 1.2426, "step": 1640 }, { "epoch": 1.2921623341655653, "grad_norm": 0.24927949905395508, "learning_rate": 6.725797205172451e-05, "loss": 1.265, "step": 1650 }, { "epoch": 1.2999951045185294, "grad_norm": 0.26761099696159363, "learning_rate": 6.596932939162165e-05, "loss": 1.2606, "step": 1660 }, { "epoch": 1.3078278748714935, "grad_norm": 0.24653536081314087, "learning_rate": 6.468704682413711e-05, "loss": 1.2561, "step": 1670 }, { "epoch": 1.3156606452244577, "grad_norm": 0.2502845823764801, "learning_rate": 6.341136399885017e-05, "loss": 1.2778, "step": 1680 }, { "epoch": 1.323493415577422, "grad_norm": 0.2217158079147339, "learning_rate": 6.214251933189468e-05, "loss": 1.2609, "step": 1690 }, { "epoch": 1.3313261859303862, "grad_norm": 0.21680890023708344, "learning_rate": 6.0880749961400964e-05, "loss": 1.271, "step": 1700 }, { "epoch": 1.3391589562833506, "grad_norm": 0.20667967200279236, "learning_rate": 5.962629170317616e-05, "loss": 1.2731, "step": 1710 }, { "epoch": 1.3469917266363147, "grad_norm": 0.23594318330287933, "learning_rate": 5.8379379006632154e-05, "loss": 1.283, "step": 1720 }, { "epoch": 1.3548244969892789, "grad_norm": 0.3072715699672699, "learning_rate": 5.714024491096841e-05, "loss": 1.2697, "step": 1730 }, { "epoch": 1.362657267342243, "grad_norm": 0.36377736926078796, "learning_rate": 5.5909121001618755e-05, "loss": 1.2634, "step": 1740 }, { "epoch": 1.3704900376952074, "grad_norm": 0.2084302008152008, "learning_rate": 5.4686237366969536e-05, "loss": 1.2863, "step": 1750 }, { "epoch": 1.3783228080481715, "grad_norm": 0.2367020696401596, "learning_rate": 5.3471822555357896e-05, "loss": 1.2573, "step": 1760 }, { "epoch": 1.3861555784011357, "grad_norm": 0.22081467509269714, "learning_rate": 5.22661035323575e-05, "loss": 1.2844, "step": 1770 }, { "epoch": 1.3939883487541, "grad_norm": 0.28228339552879333, "learning_rate": 5.1069305638360475e-05, "loss": 1.2759, "step": 1780 }, { "epoch": 1.4018211191070642, "grad_norm": 0.26325392723083496, "learning_rate": 4.988165254646277e-05, "loss": 1.2563, "step": 1790 }, { "epoch": 1.4096538894600283, "grad_norm": 0.20595784485340118, "learning_rate": 4.870336622066116e-05, "loss": 1.2765, "step": 1800 }, { "epoch": 1.4174866598129925, "grad_norm": 0.24151615798473358, "learning_rate": 4.7534666874369924e-05, "loss": 1.2668, "step": 1810 }, { "epoch": 1.4253194301659569, "grad_norm": 0.22979967296123505, "learning_rate": 4.637577292926428e-05, "loss": 1.2224, "step": 1820 }, { "epoch": 1.433152200518921, "grad_norm": 0.24872545897960663, "learning_rate": 4.522690097445919e-05, "loss": 1.3118, "step": 1830 }, { "epoch": 1.4409849708718854, "grad_norm": 0.2359861135482788, "learning_rate": 4.4088265726029995e-05, "loss": 1.2809, "step": 1840 }, { "epoch": 1.4488177412248495, "grad_norm": 0.24434255063533783, "learning_rate": 4.296007998688405e-05, "loss": 1.2475, "step": 1850 }, { "epoch": 1.4566505115778137, "grad_norm": 0.2121056616306305, "learning_rate": 4.184255460698873e-05, "loss": 1.2603, "step": 1860 }, { "epoch": 1.4644832819307778, "grad_norm": 0.2658556401729584, "learning_rate": 4.073589844396555e-05, "loss": 1.2728, "step": 1870 }, { "epoch": 1.4723160522837422, "grad_norm": 0.229701966047287, "learning_rate": 3.9640318324055815e-05, "loss": 1.2413, "step": 1880 }, { "epoch": 1.4801488226367063, "grad_norm": 0.2595929801464081, "learning_rate": 3.855601900346655e-05, "loss": 1.2961, "step": 1890 }, { "epoch": 1.4879815929896705, "grad_norm": 0.24406304955482483, "learning_rate": 3.748320313010267e-05, "loss": 1.2476, "step": 1900 }, { "epoch": 1.4958143633426348, "grad_norm": 0.22859641909599304, "learning_rate": 3.6422071205693866e-05, "loss": 1.2773, "step": 1910 }, { "epoch": 1.503647133695599, "grad_norm": 0.24212431907653809, "learning_rate": 3.5372821548321966e-05, "loss": 1.2716, "step": 1920 }, { "epoch": 1.5114799040485631, "grad_norm": 0.24502380192279816, "learning_rate": 3.433565025535688e-05, "loss": 1.2548, "step": 1930 }, { "epoch": 1.5193126744015273, "grad_norm": 0.23860076069831848, "learning_rate": 3.331075116680715e-05, "loss": 1.24, "step": 1940 }, { "epoch": 1.5271454447544917, "grad_norm": 0.25744864344596863, "learning_rate": 3.2298315829092915e-05, "loss": 1.2595, "step": 1950 }, { "epoch": 1.5349782151074558, "grad_norm": 0.29559800028800964, "learning_rate": 3.1298533459247e-05, "loss": 1.2669, "step": 1960 }, { "epoch": 1.5428109854604202, "grad_norm": 0.23177428543567657, "learning_rate": 3.0311590909551625e-05, "loss": 1.256, "step": 1970 }, { "epoch": 1.5506437558133843, "grad_norm": 0.23339729011058807, "learning_rate": 2.9337672632617153e-05, "loss": 1.269, "step": 1980 }, { "epoch": 1.5584765261663485, "grad_norm": 0.2607370913028717, "learning_rate": 2.837696064690909e-05, "loss": 1.2227, "step": 1990 }, { "epoch": 1.5663092965193126, "grad_norm": 0.2623726427555084, "learning_rate": 2.742963450273016e-05, "loss": 1.3017, "step": 2000 }, { "epoch": 1.5663092965193126, "eval_loss": 1.3199408054351807, "eval_runtime": 132.0171, "eval_samples_per_second": 8.15, "eval_steps_per_second": 8.15, "step": 2000 }, { "epoch": 1.5741420668722768, "grad_norm": 0.2687166631221771, "learning_rate": 2.6495871248663495e-05, "loss": 1.3054, "step": 2010 }, { "epoch": 1.5819748372252411, "grad_norm": 0.24729202687740326, "learning_rate": 2.5575845398483766e-05, "loss": 1.2441, "step": 2020 }, { "epoch": 1.5898076075782053, "grad_norm": 0.26290377974510193, "learning_rate": 2.4669728898541456e-05, "loss": 1.2684, "step": 2030 }, { "epoch": 1.5976403779311696, "grad_norm": 0.2935054302215576, "learning_rate": 2.3777691095627587e-05, "loss": 1.2549, "step": 2040 }, { "epoch": 1.6054731482841338, "grad_norm": 0.25191736221313477, "learning_rate": 2.2899898705323842e-05, "loss": 1.318, "step": 2050 }, { "epoch": 1.613305918637098, "grad_norm": 0.23455071449279785, "learning_rate": 2.2036515780844667e-05, "loss": 1.2881, "step": 2060 }, { "epoch": 1.621138688990062, "grad_norm": 0.23078800737857819, "learning_rate": 2.1187703682376758e-05, "loss": 1.2852, "step": 2070 }, { "epoch": 1.6289714593430262, "grad_norm": 0.2540266513824463, "learning_rate": 2.0353621046922188e-05, "loss": 1.253, "step": 2080 }, { "epoch": 1.6368042296959906, "grad_norm": 0.24285629391670227, "learning_rate": 1.9534423758650043e-05, "loss": 1.307, "step": 2090 }, { "epoch": 1.644637000048955, "grad_norm": 0.26439082622528076, "learning_rate": 1.8730264919762964e-05, "loss": 1.2913, "step": 2100 }, { "epoch": 1.6524697704019191, "grad_norm": 0.31153881549835205, "learning_rate": 1.794129482188325e-05, "loss": 1.2674, "step": 2110 }, { "epoch": 1.6603025407548833, "grad_norm": 0.2785818576812744, "learning_rate": 1.7167660917964556e-05, "loss": 1.2461, "step": 2120 }, { "epoch": 1.6681353111078474, "grad_norm": 0.2300751656293869, "learning_rate": 1.6409507794733947e-05, "loss": 1.2423, "step": 2130 }, { "epoch": 1.6759680814608116, "grad_norm": 0.2691507339477539, "learning_rate": 1.5666977145669516e-05, "loss": 1.2388, "step": 2140 }, { "epoch": 1.683800851813776, "grad_norm": 0.25373420119285583, "learning_rate": 1.4940207744519118e-05, "loss": 1.2638, "step": 2150 }, { "epoch": 1.69163362216674, "grad_norm": 0.27554184198379517, "learning_rate": 1.4229335419364264e-05, "loss": 1.2777, "step": 2160 }, { "epoch": 1.6994663925197044, "grad_norm": 0.2220025211572647, "learning_rate": 1.3534493027235062e-05, "loss": 1.2546, "step": 2170 }, { "epoch": 1.7072991628726686, "grad_norm": 0.24049687385559082, "learning_rate": 1.2855810429279957e-05, "loss": 1.2853, "step": 2180 }, { "epoch": 1.7151319332256327, "grad_norm": 0.2137405276298523, "learning_rate": 1.2193414466495766e-05, "loss": 1.2818, "step": 2190 }, { "epoch": 1.7229647035785969, "grad_norm": 0.24141110479831696, "learning_rate": 1.1547428936021887e-05, "loss": 1.2616, "step": 2200 }, { "epoch": 1.730797473931561, "grad_norm": 0.22827230393886566, "learning_rate": 1.0917974568003552e-05, "loss": 1.2494, "step": 2210 }, { "epoch": 1.7386302442845254, "grad_norm": 0.28841832280158997, "learning_rate": 1.0305169003028093e-05, "loss": 1.269, "step": 2220 }, { "epoch": 1.7464630146374898, "grad_norm": 0.29105958342552185, "learning_rate": 9.709126770138943e-06, "loss": 1.2773, "step": 2230 }, { "epoch": 1.754295784990454, "grad_norm": 0.2167786806821823, "learning_rate": 9.129959265430766e-06, "loss": 1.2601, "step": 2240 }, { "epoch": 1.762128555343418, "grad_norm": 0.2262933999300003, "learning_rate": 8.567774731230416e-06, "loss": 1.2322, "step": 2250 }, { "epoch": 1.7699613256963822, "grad_norm": 0.23295988142490387, "learning_rate": 8.022678235867243e-06, "loss": 1.2682, "step": 2260 }, { "epoch": 1.7777940960493464, "grad_norm": 0.24120023846626282, "learning_rate": 7.4947716540364475e-06, "loss": 1.2729, "step": 2270 }, { "epoch": 1.7856268664023107, "grad_norm": 0.3200109601020813, "learning_rate": 6.984153647759495e-06, "loss": 1.2913, "step": 2280 }, { "epoch": 1.7934596367552749, "grad_norm": 0.2458198368549347, "learning_rate": 6.490919647944893e-06, "loss": 1.255, "step": 2290 }, { "epoch": 1.8012924071082392, "grad_norm": 0.2539612948894501, "learning_rate": 6.015161836552763e-06, "loss": 1.2843, "step": 2300 }, { "epoch": 1.8091251774612034, "grad_norm": 0.22716960310935974, "learning_rate": 5.556969129366707e-06, "loss": 1.2288, "step": 2310 }, { "epoch": 1.8169579478141675, "grad_norm": 0.23034286499023438, "learning_rate": 5.116427159376125e-06, "loss": 1.2519, "step": 2320 }, { "epoch": 1.8247907181671317, "grad_norm": 0.23222385346889496, "learning_rate": 4.693618260771937e-06, "loss": 1.2324, "step": 2330 }, { "epoch": 1.8326234885200958, "grad_norm": 0.27542534470558167, "learning_rate": 4.288621453558972e-06, "loss": 1.2678, "step": 2340 }, { "epoch": 1.8404562588730602, "grad_norm": 0.22792376577854156, "learning_rate": 3.901512428787601e-06, "loss": 1.2561, "step": 2350 }, { "epoch": 1.8482890292260243, "grad_norm": 0.22720928490161896, "learning_rate": 3.532363534407712e-06, "loss": 1.2571, "step": 2360 }, { "epoch": 1.8561217995789887, "grad_norm": 0.2314911037683487, "learning_rate": 3.1812437617472656e-06, "loss": 1.2679, "step": 2370 }, { "epoch": 1.8639545699319529, "grad_norm": 0.20459794998168945, "learning_rate": 2.848218732618468e-06, "loss": 1.2507, "step": 2380 }, { "epoch": 1.871787340284917, "grad_norm": 0.2685185670852661, "learning_rate": 2.5333506870533374e-06, "loss": 1.262, "step": 2390 }, { "epoch": 1.8796201106378811, "grad_norm": 0.310500830411911, "learning_rate": 2.236698471671628e-06, "loss": 1.2515, "step": 2400 }, { "epoch": 1.8874528809908453, "grad_norm": 0.2637697756290436, "learning_rate": 1.958317528682696e-06, "loss": 1.2755, "step": 2410 }, { "epoch": 1.8952856513438097, "grad_norm": 0.2551523745059967, "learning_rate": 1.6982598855238564e-06, "loss": 1.2431, "step": 2420 }, { "epoch": 1.903118421696774, "grad_norm": 0.29257139563560486, "learning_rate": 1.4565741451366688e-06, "loss": 1.2684, "step": 2430 }, { "epoch": 1.9109511920497382, "grad_norm": 0.25060567259788513, "learning_rate": 1.233305476883584e-06, "loss": 1.2715, "step": 2440 }, { "epoch": 1.9187839624027023, "grad_norm": 0.2520369589328766, "learning_rate": 1.0284956081059172e-06, "loss": 1.2632, "step": 2450 }, { "epoch": 1.9266167327556665, "grad_norm": 0.25969362258911133, "learning_rate": 8.42182816325432e-07, "loss": 1.2785, "step": 2460 }, { "epoch": 1.9344495031086306, "grad_norm": 0.25107839703559875, "learning_rate": 6.744019220905063e-07, "loss": 1.2473, "step": 2470 }, { "epoch": 1.942282273461595, "grad_norm": 0.29540321230888367, "learning_rate": 5.251842824683716e-07, "loss": 1.2522, "step": 2480 }, { "epoch": 1.9501150438145591, "grad_norm": 0.2548975944519043, "learning_rate": 3.9455778518476813e-07, "loss": 1.2454, "step": 2490 }, { "epoch": 1.9579478141675235, "grad_norm": 0.24709321558475494, "learning_rate": 2.8254684341187944e-07, "loss": 1.2827, "step": 2500 }, { "epoch": 1.9579478141675235, "eval_loss": 1.3144887685775757, "eval_runtime": 132.0892, "eval_samples_per_second": 8.146, "eval_steps_per_second": 8.146, "step": 2500 }, { "epoch": 1.9657805845204877, "grad_norm": 0.25036099553108215, "learning_rate": 1.8917239120569418e-07, "loss": 1.2941, "step": 2510 }, { "epoch": 1.9736133548734518, "grad_norm": 0.24968591332435608, "learning_rate": 1.1445187959359116e-07, "loss": 1.2658, "step": 2520 }, { "epoch": 1.981446125226416, "grad_norm": 0.2779126465320587, "learning_rate": 5.8399273312825795e-08, "loss": 1.2604, "step": 2530 }, { "epoch": 1.98927889557938, "grad_norm": 0.2542530298233032, "learning_rate": 2.102504820069573e-08, "loss": 1.2684, "step": 2540 }, { "epoch": 1.9971116659323445, "grad_norm": 0.24978196620941162, "learning_rate": 2.3361892365847403e-09, "loss": 1.2584, "step": 2550 }, { "epoch": 2.0, "step": 2554, "total_flos": 1.9453038489124209e+18, "train_loss": 1.3485698382557605, "train_runtime": 17547.92, "train_samples_per_second": 2.328, "train_steps_per_second": 0.146 } ], "logging_steps": 10, "max_steps": 2554, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 1.9453038489124209e+18, "train_batch_size": 1, "trial_name": null, "trial_params": null }