{ "best_global_step": 4210, "best_metric": 1.6282436847686768, "best_model_checkpoint": "mistralai/Mistral-7B-Instruct-v0.3/Prefix-Legal/checkpoint-4210", "epoch": 5.0, "eval_steps": 500, "global_step": 4210, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.011876484560570071, "grad_norm": 3.0710983276367188, "learning_rate": 0.00019957244655581947, "loss": 8.9759, "step": 10 }, { "epoch": 0.023752969121140142, "grad_norm": 2.2068538665771484, "learning_rate": 0.00019909738717339666, "loss": 6.7509, "step": 20 }, { "epoch": 0.035629453681710214, "grad_norm": 1.0512908697128296, "learning_rate": 0.00019862232779097389, "loss": 4.8051, "step": 30 }, { "epoch": 0.047505938242280284, "grad_norm": 0.5420308709144592, "learning_rate": 0.00019814726840855108, "loss": 3.5751, "step": 40 }, { "epoch": 0.05938242280285035, "grad_norm": 0.601415753364563, "learning_rate": 0.00019767220902612828, "loss": 3.1808, "step": 50 }, { "epoch": 0.07125890736342043, "grad_norm": 0.4707927405834198, "learning_rate": 0.00019719714964370547, "loss": 2.7878, "step": 60 }, { "epoch": 0.0831353919239905, "grad_norm": 0.19448263943195343, "learning_rate": 0.00019672209026128267, "loss": 2.4629, "step": 70 }, { "epoch": 0.09501187648456057, "grad_norm": 0.32429736852645874, "learning_rate": 0.00019624703087885986, "loss": 2.5181, "step": 80 }, { "epoch": 0.10688836104513064, "grad_norm": 0.21264347434043884, "learning_rate": 0.00019577197149643705, "loss": 2.189, "step": 90 }, { "epoch": 0.1187648456057007, "grad_norm": 0.28584185242652893, "learning_rate": 0.00019529691211401428, "loss": 2.4082, "step": 100 }, { "epoch": 0.13064133016627077, "grad_norm": 0.21001407504081726, "learning_rate": 0.00019482185273159147, "loss": 2.1603, "step": 110 }, { "epoch": 0.14251781472684086, "grad_norm": 0.3775597810745239, "learning_rate": 0.00019434679334916867, "loss": 2.3354, "step": 120 }, { "epoch": 0.1543942992874109, "grad_norm": 0.3545965552330017, "learning_rate": 0.00019387173396674586, "loss": 2.2351, "step": 130 }, { "epoch": 0.166270783847981, "grad_norm": 0.23063507676124573, "learning_rate": 0.00019339667458432306, "loss": 2.0247, "step": 140 }, { "epoch": 0.17814726840855108, "grad_norm": 0.3464234471321106, "learning_rate": 0.00019292161520190025, "loss": 2.0406, "step": 150 }, { "epoch": 0.19002375296912113, "grad_norm": 0.4495205879211426, "learning_rate": 0.00019244655581947745, "loss": 2.2667, "step": 160 }, { "epoch": 0.20190023752969122, "grad_norm": 0.33238738775253296, "learning_rate": 0.00019197149643705464, "loss": 2.383, "step": 170 }, { "epoch": 0.21377672209026127, "grad_norm": 0.3044181168079376, "learning_rate": 0.00019149643705463184, "loss": 2.2237, "step": 180 }, { "epoch": 0.22565320665083136, "grad_norm": 0.3507155776023865, "learning_rate": 0.00019102137767220906, "loss": 2.4906, "step": 190 }, { "epoch": 0.2375296912114014, "grad_norm": 0.15554989874362946, "learning_rate": 0.00019054631828978625, "loss": 2.3929, "step": 200 }, { "epoch": 0.2494061757719715, "grad_norm": 0.1731153279542923, "learning_rate": 0.00019007125890736345, "loss": 2.1798, "step": 210 }, { "epoch": 0.26128266033254155, "grad_norm": 0.3168913424015045, "learning_rate": 0.00018959619952494064, "loss": 2.1145, "step": 220 }, { "epoch": 0.27315914489311166, "grad_norm": 0.1382881999015808, "learning_rate": 0.00018912114014251784, "loss": 2.2421, "step": 230 }, { "epoch": 0.2850356294536817, "grad_norm": 0.4627864360809326, "learning_rate": 0.00018864608076009503, "loss": 2.0737, "step": 240 }, { "epoch": 0.29691211401425177, "grad_norm": 0.1689852476119995, "learning_rate": 0.00018817102137767223, "loss": 2.1153, "step": 250 }, { "epoch": 0.3087885985748218, "grad_norm": 0.2260257601737976, "learning_rate": 0.00018769596199524942, "loss": 2.3561, "step": 260 }, { "epoch": 0.32066508313539194, "grad_norm": NaN, "learning_rate": 0.00018722090261282662, "loss": 1.9748, "step": 270 }, { "epoch": 0.332541567695962, "grad_norm": 0.1930597722530365, "learning_rate": 0.0001867458432304038, "loss": 2.5756, "step": 280 }, { "epoch": 0.34441805225653205, "grad_norm": 0.25983959436416626, "learning_rate": 0.000186270783847981, "loss": 2.1881, "step": 290 }, { "epoch": 0.35629453681710216, "grad_norm": 0.3136606514453888, "learning_rate": 0.0001857957244655582, "loss": 2.4223, "step": 300 }, { "epoch": 0.3681710213776722, "grad_norm": 0.20531731843948364, "learning_rate": 0.0001853206650831354, "loss": 2.131, "step": 310 }, { "epoch": 0.38004750593824227, "grad_norm": 0.30765125155448914, "learning_rate": 0.0001848456057007126, "loss": 2.0688, "step": 320 }, { "epoch": 0.3919239904988123, "grad_norm": 0.2848324179649353, "learning_rate": 0.0001843705463182898, "loss": 1.9626, "step": 330 }, { "epoch": 0.40380047505938244, "grad_norm": 0.3842903673648834, "learning_rate": 0.000183895486935867, "loss": 2.2144, "step": 340 }, { "epoch": 0.4156769596199525, "grad_norm": 0.24172411859035492, "learning_rate": 0.0001834204275534442, "loss": 2.3502, "step": 350 }, { "epoch": 0.42755344418052255, "grad_norm": 0.4283181428909302, "learning_rate": 0.0001829453681710214, "loss": 1.9953, "step": 360 }, { "epoch": 0.43942992874109266, "grad_norm": 0.2430773675441742, "learning_rate": 0.0001824703087885986, "loss": 2.0159, "step": 370 }, { "epoch": 0.4513064133016627, "grad_norm": 0.2148657590150833, "learning_rate": 0.0001819952494061758, "loss": 1.6797, "step": 380 }, { "epoch": 0.46318289786223277, "grad_norm": 0.15379592776298523, "learning_rate": 0.00018152019002375298, "loss": 1.9509, "step": 390 }, { "epoch": 0.4750593824228028, "grad_norm": 0.12941691279411316, "learning_rate": 0.00018104513064133018, "loss": 1.9087, "step": 400 }, { "epoch": 0.48693586698337293, "grad_norm": 0.2712809145450592, "learning_rate": 0.00018057007125890737, "loss": 2.2381, "step": 410 }, { "epoch": 0.498812351543943, "grad_norm": 0.2060050666332245, "learning_rate": 0.00018009501187648457, "loss": 1.7771, "step": 420 }, { "epoch": 0.5106888361045131, "grad_norm": 0.35927721858024597, "learning_rate": 0.00017961995249406176, "loss": 1.8161, "step": 430 }, { "epoch": 0.5225653206650831, "grad_norm": 0.5299475789070129, "learning_rate": 0.00017914489311163896, "loss": 1.6709, "step": 440 }, { "epoch": 0.5344418052256532, "grad_norm": 0.1508459895849228, "learning_rate": 0.00017866983372921615, "loss": 1.9625, "step": 450 }, { "epoch": 0.5463182897862233, "grad_norm": 0.15887700021266937, "learning_rate": 0.00017819477434679335, "loss": 1.7946, "step": 460 }, { "epoch": 0.5581947743467933, "grad_norm": 0.19263148307800293, "learning_rate": 0.00017771971496437057, "loss": 2.1326, "step": 470 }, { "epoch": 0.5700712589073634, "grad_norm": 0.13795185089111328, "learning_rate": 0.00017724465558194776, "loss": 1.7455, "step": 480 }, { "epoch": 0.5819477434679335, "grad_norm": 0.15855297446250916, "learning_rate": 0.00017676959619952496, "loss": 1.9084, "step": 490 }, { "epoch": 0.5938242280285035, "grad_norm": 0.2710108458995819, "learning_rate": 0.00017629453681710215, "loss": 1.6366, "step": 500 }, { "epoch": 0.6057007125890737, "grad_norm": 0.26268941164016724, "learning_rate": 0.00017581947743467935, "loss": 1.4961, "step": 510 }, { "epoch": 0.6175771971496437, "grad_norm": 0.3674164116382599, "learning_rate": 0.00017534441805225654, "loss": 1.7896, "step": 520 }, { "epoch": 0.6294536817102138, "grad_norm": 0.19550037384033203, "learning_rate": 0.00017486935866983374, "loss": 1.743, "step": 530 }, { "epoch": 0.6413301662707839, "grad_norm": 0.2876305878162384, "learning_rate": 0.00017439429928741093, "loss": 1.7123, "step": 540 }, { "epoch": 0.6532066508313539, "grad_norm": 0.14980630576610565, "learning_rate": 0.00017391923990498813, "loss": 1.8916, "step": 550 }, { "epoch": 0.665083135391924, "grad_norm": 0.6001176834106445, "learning_rate": 0.00017344418052256532, "loss": 1.7289, "step": 560 }, { "epoch": 0.6769596199524941, "grad_norm": 0.23576416075229645, "learning_rate": 0.00017296912114014252, "loss": 1.682, "step": 570 }, { "epoch": 0.6888361045130641, "grad_norm": 0.33259084820747375, "learning_rate": 0.0001724940617577197, "loss": 1.8476, "step": 580 }, { "epoch": 0.7007125890736342, "grad_norm": 0.28322118520736694, "learning_rate": 0.0001720190023752969, "loss": 1.9753, "step": 590 }, { "epoch": 0.7125890736342043, "grad_norm": 0.29700034856796265, "learning_rate": 0.0001715439429928741, "loss": 1.6133, "step": 600 }, { "epoch": 0.7244655581947743, "grad_norm": 0.3623453676700592, "learning_rate": 0.00017106888361045132, "loss": 1.4567, "step": 610 }, { "epoch": 0.7363420427553444, "grad_norm": 0.25531700253486633, "learning_rate": 0.00017059382422802852, "loss": 1.4847, "step": 620 }, { "epoch": 0.7482185273159145, "grad_norm": 0.19563037157058716, "learning_rate": 0.0001701187648456057, "loss": 1.668, "step": 630 }, { "epoch": 0.7600950118764845, "grad_norm": 0.35908615589141846, "learning_rate": 0.0001696437054631829, "loss": 1.6318, "step": 640 }, { "epoch": 0.7719714964370546, "grad_norm": 0.28938376903533936, "learning_rate": 0.0001691686460807601, "loss": 1.9201, "step": 650 }, { "epoch": 0.7838479809976246, "grad_norm": 0.09309093654155731, "learning_rate": 0.0001686935866983373, "loss": 1.7287, "step": 660 }, { "epoch": 0.7957244655581948, "grad_norm": 0.25619930028915405, "learning_rate": 0.0001682185273159145, "loss": 2.0456, "step": 670 }, { "epoch": 0.8076009501187649, "grad_norm": 0.2682470381259918, "learning_rate": 0.0001677434679334917, "loss": 1.6818, "step": 680 }, { "epoch": 0.8194774346793349, "grad_norm": 0.2695891857147217, "learning_rate": 0.00016726840855106888, "loss": 1.7435, "step": 690 }, { "epoch": 0.831353919239905, "grad_norm": 0.17132176458835602, "learning_rate": 0.00016679334916864608, "loss": 1.5322, "step": 700 }, { "epoch": 0.8432304038004751, "grad_norm": 0.2175643891096115, "learning_rate": 0.00016631828978622327, "loss": 1.8344, "step": 710 }, { "epoch": 0.8551068883610451, "grad_norm": 0.2726324498653412, "learning_rate": 0.00016584323040380047, "loss": 1.5398, "step": 720 }, { "epoch": 0.8669833729216152, "grad_norm": 0.25916969776153564, "learning_rate": 0.00016536817102137766, "loss": 1.2104, "step": 730 }, { "epoch": 0.8788598574821853, "grad_norm": 0.2567419409751892, "learning_rate": 0.00016489311163895486, "loss": 1.991, "step": 740 }, { "epoch": 0.8907363420427553, "grad_norm": 0.2414155900478363, "learning_rate": 0.00016441805225653208, "loss": 1.2854, "step": 750 }, { "epoch": 0.9026128266033254, "grad_norm": 0.33677032589912415, "learning_rate": 0.00016394299287410927, "loss": 1.6606, "step": 760 }, { "epoch": 0.9144893111638955, "grad_norm": 0.24837538599967957, "learning_rate": 0.00016346793349168647, "loss": 1.9541, "step": 770 }, { "epoch": 0.9263657957244655, "grad_norm": 0.20428167283535004, "learning_rate": 0.00016299287410926366, "loss": 1.3513, "step": 780 }, { "epoch": 0.9382422802850356, "grad_norm": 0.2312709093093872, "learning_rate": 0.00016251781472684086, "loss": 1.3695, "step": 790 }, { "epoch": 0.9501187648456056, "grad_norm": 0.13046520948410034, "learning_rate": 0.00016204275534441805, "loss": 1.3863, "step": 800 }, { "epoch": 0.9619952494061758, "grad_norm": 0.1456942856311798, "learning_rate": 0.00016156769596199525, "loss": 1.5607, "step": 810 }, { "epoch": 0.9738717339667459, "grad_norm": 0.14096485078334808, "learning_rate": 0.00016109263657957244, "loss": 1.3027, "step": 820 }, { "epoch": 0.9857482185273159, "grad_norm": 0.2699258029460907, "learning_rate": 0.00016061757719714964, "loss": 1.7399, "step": 830 }, { "epoch": 0.997624703087886, "grad_norm": 0.32892510294914246, "learning_rate": 0.00016014251781472683, "loss": 1.3922, "step": 840 }, { "epoch": 1.0, "eval_loss": 2.8184680938720703, "eval_runtime": 47.3518, "eval_samples_per_second": 7.919, "eval_steps_per_second": 1.985, "step": 842 }, { "epoch": 1.009501187648456, "grad_norm": 0.12905089557170868, "learning_rate": 0.00015966745843230403, "loss": 1.2629, "step": 850 }, { "epoch": 1.0213776722090262, "grad_norm": 0.2982773184776306, "learning_rate": 0.00015919239904988122, "loss": 1.5383, "step": 860 }, { "epoch": 1.0332541567695963, "grad_norm": 0.21285922825336456, "learning_rate": 0.00015871733966745844, "loss": 1.4079, "step": 870 }, { "epoch": 1.0451306413301662, "grad_norm": 0.16437295079231262, "learning_rate": 0.00015824228028503564, "loss": 1.4678, "step": 880 }, { "epoch": 1.0570071258907363, "grad_norm": 0.38892456889152527, "learning_rate": 0.00015776722090261283, "loss": 1.5817, "step": 890 }, { "epoch": 1.0688836104513064, "grad_norm": 0.3265591561794281, "learning_rate": 0.00015729216152019003, "loss": 1.0689, "step": 900 }, { "epoch": 1.0807600950118765, "grad_norm": 0.09871666133403778, "learning_rate": 0.00015681710213776722, "loss": 1.3273, "step": 910 }, { "epoch": 1.0926365795724466, "grad_norm": 0.20414851605892181, "learning_rate": 0.00015634204275534442, "loss": 1.4322, "step": 920 }, { "epoch": 1.1045130641330165, "grad_norm": 0.10297483205795288, "learning_rate": 0.00015586698337292161, "loss": 1.3472, "step": 930 }, { "epoch": 1.1163895486935866, "grad_norm": 0.22927910089492798, "learning_rate": 0.0001553919239904988, "loss": 1.5666, "step": 940 }, { "epoch": 1.1282660332541568, "grad_norm": 0.14535310864448547, "learning_rate": 0.00015491686460807603, "loss": 1.2523, "step": 950 }, { "epoch": 1.1401425178147269, "grad_norm": 0.2174110859632492, "learning_rate": 0.00015444180522565323, "loss": 1.1953, "step": 960 }, { "epoch": 1.152019002375297, "grad_norm": 0.1418980062007904, "learning_rate": 0.00015396674584323042, "loss": 1.2709, "step": 970 }, { "epoch": 1.1638954869358669, "grad_norm": 0.34407395124435425, "learning_rate": 0.00015349168646080762, "loss": 1.2696, "step": 980 }, { "epoch": 1.175771971496437, "grad_norm": 0.23791618645191193, "learning_rate": 0.0001530166270783848, "loss": 1.5138, "step": 990 }, { "epoch": 1.187648456057007, "grad_norm": 0.11311549693346024, "learning_rate": 0.000152541567695962, "loss": 1.3508, "step": 1000 }, { "epoch": 1.1995249406175772, "grad_norm": 0.2730580270290375, "learning_rate": 0.0001520665083135392, "loss": 1.3165, "step": 1010 }, { "epoch": 1.2114014251781473, "grad_norm": 0.23997551202774048, "learning_rate": 0.0001515914489311164, "loss": 1.1388, "step": 1020 }, { "epoch": 1.2232779097387174, "grad_norm": 0.3308527171611786, "learning_rate": 0.00015111638954869362, "loss": 1.127, "step": 1030 }, { "epoch": 1.2351543942992875, "grad_norm": 0.15757586061954498, "learning_rate": 0.0001506413301662708, "loss": 1.118, "step": 1040 }, { "epoch": 1.2470308788598574, "grad_norm": 0.2413954883813858, "learning_rate": 0.000150166270783848, "loss": 1.341, "step": 1050 }, { "epoch": 1.2589073634204275, "grad_norm": 0.19545750319957733, "learning_rate": 0.0001496912114014252, "loss": 1.1667, "step": 1060 }, { "epoch": 1.2707838479809976, "grad_norm": 0.33894041180610657, "learning_rate": 0.0001492161520190024, "loss": 1.0079, "step": 1070 }, { "epoch": 1.2826603325415677, "grad_norm": 0.26464515924453735, "learning_rate": 0.0001487410926365796, "loss": 1.7574, "step": 1080 }, { "epoch": 1.2945368171021379, "grad_norm": 0.15955770015716553, "learning_rate": 0.00014826603325415679, "loss": 1.016, "step": 1090 }, { "epoch": 1.3064133016627077, "grad_norm": 0.2969847321510315, "learning_rate": 0.00014779097387173398, "loss": 1.6192, "step": 1100 }, { "epoch": 1.3182897862232779, "grad_norm": 0.22145484387874603, "learning_rate": 0.00014731591448931118, "loss": 1.1592, "step": 1110 }, { "epoch": 1.330166270783848, "grad_norm": 0.1680860072374344, "learning_rate": 0.00014684085510688837, "loss": 1.3513, "step": 1120 }, { "epoch": 1.342042755344418, "grad_norm": 0.15734612941741943, "learning_rate": 0.00014636579572446557, "loss": 1.1277, "step": 1130 }, { "epoch": 1.3539192399049882, "grad_norm": 0.12315751612186432, "learning_rate": 0.00014589073634204276, "loss": 1.1505, "step": 1140 }, { "epoch": 1.365795724465558, "grad_norm": 0.06593690812587738, "learning_rate": 0.00014541567695961996, "loss": 1.3138, "step": 1150 }, { "epoch": 1.3776722090261282, "grad_norm": 0.07469449192285538, "learning_rate": 0.00014494061757719715, "loss": 1.1481, "step": 1160 }, { "epoch": 1.3895486935866983, "grad_norm": 0.06956204771995544, "learning_rate": 0.00014446555819477437, "loss": 1.0823, "step": 1170 }, { "epoch": 1.4014251781472684, "grad_norm": 0.20060239732265472, "learning_rate": 0.00014399049881235157, "loss": 1.3006, "step": 1180 }, { "epoch": 1.4133016627078385, "grad_norm": 0.200018048286438, "learning_rate": 0.00014351543942992876, "loss": 1.2379, "step": 1190 }, { "epoch": 1.4251781472684084, "grad_norm": 0.06538112461566925, "learning_rate": 0.00014304038004750596, "loss": 1.0656, "step": 1200 }, { "epoch": 1.4370546318289787, "grad_norm": 0.17270460724830627, "learning_rate": 0.00014256532066508315, "loss": 1.2951, "step": 1210 }, { "epoch": 1.4489311163895486, "grad_norm": 0.38526710867881775, "learning_rate": 0.00014209026128266035, "loss": 1.4878, "step": 1220 }, { "epoch": 1.4608076009501187, "grad_norm": 0.126207172870636, "learning_rate": 0.00014161520190023754, "loss": 1.1548, "step": 1230 }, { "epoch": 1.4726840855106889, "grad_norm": 0.08622638136148453, "learning_rate": 0.00014114014251781474, "loss": 1.3664, "step": 1240 }, { "epoch": 1.484560570071259, "grad_norm": 0.09776638448238373, "learning_rate": 0.00014066508313539193, "loss": 1.0445, "step": 1250 }, { "epoch": 1.496437054631829, "grad_norm": 0.15026284754276276, "learning_rate": 0.00014019002375296913, "loss": 1.112, "step": 1260 }, { "epoch": 1.508313539192399, "grad_norm": 0.3850550353527069, "learning_rate": 0.00013971496437054632, "loss": 1.4271, "step": 1270 }, { "epoch": 1.520190023752969, "grad_norm": 0.11536882817745209, "learning_rate": 0.00013923990498812352, "loss": 0.9066, "step": 1280 }, { "epoch": 1.5320665083135392, "grad_norm": 0.21992270648479462, "learning_rate": 0.0001387648456057007, "loss": 1.2101, "step": 1290 }, { "epoch": 1.5439429928741093, "grad_norm": 0.18977196514606476, "learning_rate": 0.0001382897862232779, "loss": 1.181, "step": 1300 }, { "epoch": 1.5558194774346794, "grad_norm": 0.07094086706638336, "learning_rate": 0.00013781472684085513, "loss": 1.129, "step": 1310 }, { "epoch": 1.5676959619952493, "grad_norm": 0.07848837226629257, "learning_rate": 0.00013733966745843232, "loss": 0.925, "step": 1320 }, { "epoch": 1.5795724465558196, "grad_norm": 0.07464569807052612, "learning_rate": 0.00013686460807600952, "loss": 1.1885, "step": 1330 }, { "epoch": 1.5914489311163895, "grad_norm": 0.0646548941731453, "learning_rate": 0.0001363895486935867, "loss": 1.0905, "step": 1340 }, { "epoch": 1.6033254156769596, "grad_norm": 0.22282616794109344, "learning_rate": 0.0001359144893111639, "loss": 1.2456, "step": 1350 }, { "epoch": 1.6152019002375297, "grad_norm": 0.24411749839782715, "learning_rate": 0.0001354394299287411, "loss": 1.1931, "step": 1360 }, { "epoch": 1.6270783847980996, "grad_norm": 0.07834232598543167, "learning_rate": 0.0001349643705463183, "loss": 1.1742, "step": 1370 }, { "epoch": 1.63895486935867, "grad_norm": 0.11734293401241302, "learning_rate": 0.0001344893111638955, "loss": 1.3415, "step": 1380 }, { "epoch": 1.6508313539192399, "grad_norm": 0.09744986891746521, "learning_rate": 0.0001340142517814727, "loss": 1.3484, "step": 1390 }, { "epoch": 1.66270783847981, "grad_norm": 0.11171699315309525, "learning_rate": 0.00013353919239904988, "loss": 1.2435, "step": 1400 }, { "epoch": 1.67458432304038, "grad_norm": 0.07334356755018234, "learning_rate": 0.00013306413301662708, "loss": 1.057, "step": 1410 }, { "epoch": 1.68646080760095, "grad_norm": 0.09792204946279526, "learning_rate": 0.00013258907363420427, "loss": 0.9427, "step": 1420 }, { "epoch": 1.6983372921615203, "grad_norm": 0.07926569133996964, "learning_rate": 0.00013211401425178147, "loss": 1.0408, "step": 1430 }, { "epoch": 1.7102137767220902, "grad_norm": 0.26732683181762695, "learning_rate": 0.00013163895486935866, "loss": 1.4231, "step": 1440 }, { "epoch": 1.7220902612826603, "grad_norm": 0.09599801898002625, "learning_rate": 0.00013116389548693588, "loss": 0.9782, "step": 1450 }, { "epoch": 1.7339667458432304, "grad_norm": 0.140996053814888, "learning_rate": 0.00013068883610451308, "loss": 1.2681, "step": 1460 }, { "epoch": 1.7458432304038005, "grad_norm": 0.3086858093738556, "learning_rate": 0.00013021377672209027, "loss": 1.2463, "step": 1470 }, { "epoch": 1.7577197149643706, "grad_norm": 0.29784321784973145, "learning_rate": 0.00012973871733966747, "loss": 1.602, "step": 1480 }, { "epoch": 1.7695961995249405, "grad_norm": 0.24647115170955658, "learning_rate": 0.00012926365795724466, "loss": 1.2578, "step": 1490 }, { "epoch": 1.7814726840855108, "grad_norm": 0.30543509125709534, "learning_rate": 0.00012878859857482186, "loss": 1.0783, "step": 1500 }, { "epoch": 1.7933491686460807, "grad_norm": 0.17075712978839874, "learning_rate": 0.00012831353919239905, "loss": 1.3717, "step": 1510 }, { "epoch": 1.8052256532066508, "grad_norm": 0.21059133112430573, "learning_rate": 0.00012783847980997625, "loss": 1.0017, "step": 1520 }, { "epoch": 1.817102137767221, "grad_norm": 0.256629079580307, "learning_rate": 0.00012736342042755344, "loss": 1.1987, "step": 1530 }, { "epoch": 1.8289786223277908, "grad_norm": 0.06935455650091171, "learning_rate": 0.00012688836104513064, "loss": 0.9272, "step": 1540 }, { "epoch": 1.8408551068883612, "grad_norm": 0.21289746463298798, "learning_rate": 0.00012641330166270783, "loss": 1.1575, "step": 1550 }, { "epoch": 1.852731591448931, "grad_norm": 0.10485855489969254, "learning_rate": 0.00012593824228028503, "loss": 1.0231, "step": 1560 }, { "epoch": 1.8646080760095012, "grad_norm": 0.06885767728090286, "learning_rate": 0.00012546318289786222, "loss": 0.9897, "step": 1570 }, { "epoch": 1.8764845605700713, "grad_norm": 0.19760191440582275, "learning_rate": 0.00012498812351543942, "loss": 1.0087, "step": 1580 }, { "epoch": 1.8883610451306412, "grad_norm": 0.12420526146888733, "learning_rate": 0.00012451306413301664, "loss": 1.04, "step": 1590 }, { "epoch": 1.9002375296912115, "grad_norm": 0.11657869070768356, "learning_rate": 0.00012403800475059383, "loss": 1.0468, "step": 1600 }, { "epoch": 1.9121140142517814, "grad_norm": 0.09553329646587372, "learning_rate": 0.00012356294536817103, "loss": 1.201, "step": 1610 }, { "epoch": 1.9239904988123515, "grad_norm": 0.16117292642593384, "learning_rate": 0.00012308788598574822, "loss": 1.2071, "step": 1620 }, { "epoch": 1.9358669833729216, "grad_norm": 0.09189121425151825, "learning_rate": 0.00012261282660332542, "loss": 1.4344, "step": 1630 }, { "epoch": 1.9477434679334917, "grad_norm": 0.18206843733787537, "learning_rate": 0.0001221377672209026, "loss": 1.0871, "step": 1640 }, { "epoch": 1.9596199524940618, "grad_norm": 0.0756635069847107, "learning_rate": 0.00012166270783847982, "loss": 1.1763, "step": 1650 }, { "epoch": 1.9714964370546317, "grad_norm": 0.13456282019615173, "learning_rate": 0.00012118764845605702, "loss": 1.0591, "step": 1660 }, { "epoch": 1.9833729216152018, "grad_norm": 0.2958761751651764, "learning_rate": 0.00012071258907363421, "loss": 1.1351, "step": 1670 }, { "epoch": 1.995249406175772, "grad_norm": 0.09702125936746597, "learning_rate": 0.0001202375296912114, "loss": 0.8416, "step": 1680 }, { "epoch": 2.0, "eval_loss": 2.022632360458374, "eval_runtime": 47.3643, "eval_samples_per_second": 7.917, "eval_steps_per_second": 1.985, "step": 1684 }, { "epoch": 2.007125890736342, "grad_norm": 0.09088994562625885, "learning_rate": 0.0001197624703087886, "loss": 0.8812, "step": 1690 }, { "epoch": 2.019002375296912, "grad_norm": 0.14102257788181305, "learning_rate": 0.0001192874109263658, "loss": 0.9463, "step": 1700 }, { "epoch": 2.030878859857482, "grad_norm": 0.05889728665351868, "learning_rate": 0.00011881235154394299, "loss": 0.8084, "step": 1710 }, { "epoch": 2.0427553444180524, "grad_norm": 0.26113539934158325, "learning_rate": 0.00011833729216152019, "loss": 0.937, "step": 1720 }, { "epoch": 2.0546318289786223, "grad_norm": 0.19298502802848816, "learning_rate": 0.00011786223277909741, "loss": 1.0235, "step": 1730 }, { "epoch": 2.0665083135391926, "grad_norm": 0.2326146960258484, "learning_rate": 0.0001173871733966746, "loss": 1.3342, "step": 1740 }, { "epoch": 2.0783847980997625, "grad_norm": 0.10600478947162628, "learning_rate": 0.0001169121140142518, "loss": 0.8438, "step": 1750 }, { "epoch": 2.0902612826603324, "grad_norm": 0.13446630537509918, "learning_rate": 0.00011643705463182899, "loss": 1.0196, "step": 1760 }, { "epoch": 2.1021377672209027, "grad_norm": 0.22078374028205872, "learning_rate": 0.00011596199524940619, "loss": 0.9152, "step": 1770 }, { "epoch": 2.1140142517814726, "grad_norm": 0.2300899475812912, "learning_rate": 0.00011548693586698338, "loss": 1.4744, "step": 1780 }, { "epoch": 2.125890736342043, "grad_norm": 0.22019624710083008, "learning_rate": 0.00011501187648456058, "loss": 1.0989, "step": 1790 }, { "epoch": 2.137767220902613, "grad_norm": 0.12800709903240204, "learning_rate": 0.00011453681710213777, "loss": 1.1071, "step": 1800 }, { "epoch": 2.1496437054631827, "grad_norm": 0.0478794202208519, "learning_rate": 0.00011406175771971497, "loss": 0.7498, "step": 1810 }, { "epoch": 2.161520190023753, "grad_norm": 0.13201993703842163, "learning_rate": 0.00011358669833729216, "loss": 1.3406, "step": 1820 }, { "epoch": 2.173396674584323, "grad_norm": 0.24289534986019135, "learning_rate": 0.00011311163895486936, "loss": 1.0622, "step": 1830 }, { "epoch": 2.1852731591448933, "grad_norm": 0.19802944362163544, "learning_rate": 0.00011263657957244655, "loss": 1.235, "step": 1840 }, { "epoch": 2.197149643705463, "grad_norm": 0.29155072569847107, "learning_rate": 0.00011216152019002375, "loss": 1.0171, "step": 1850 }, { "epoch": 2.209026128266033, "grad_norm": 0.24313387274742126, "learning_rate": 0.00011168646080760094, "loss": 0.97, "step": 1860 }, { "epoch": 2.2209026128266034, "grad_norm": 0.16295507550239563, "learning_rate": 0.00011121140142517816, "loss": 1.0464, "step": 1870 }, { "epoch": 2.2327790973871733, "grad_norm": 0.26696816086769104, "learning_rate": 0.00011073634204275536, "loss": 1.1182, "step": 1880 }, { "epoch": 2.2446555819477436, "grad_norm": 0.047891970723867416, "learning_rate": 0.00011026128266033255, "loss": 1.0058, "step": 1890 }, { "epoch": 2.2565320665083135, "grad_norm": 0.4074099659919739, "learning_rate": 0.00010978622327790975, "loss": 1.2803, "step": 1900 }, { "epoch": 2.268408551068884, "grad_norm": 0.14939150214195251, "learning_rate": 0.00010931116389548694, "loss": 1.2936, "step": 1910 }, { "epoch": 2.2802850356294537, "grad_norm": 0.05654413253068924, "learning_rate": 0.00010883610451306414, "loss": 0.9922, "step": 1920 }, { "epoch": 2.2921615201900236, "grad_norm": 0.18209122121334076, "learning_rate": 0.00010836104513064133, "loss": 1.0768, "step": 1930 }, { "epoch": 2.304038004750594, "grad_norm": 0.09877350926399231, "learning_rate": 0.00010788598574821853, "loss": 1.0292, "step": 1940 }, { "epoch": 2.315914489311164, "grad_norm": 0.21846416592597961, "learning_rate": 0.00010741092636579572, "loss": 1.187, "step": 1950 }, { "epoch": 2.3277909738717337, "grad_norm": 0.10720224678516388, "learning_rate": 0.00010693586698337293, "loss": 0.9711, "step": 1960 }, { "epoch": 2.339667458432304, "grad_norm": 0.21548321843147278, "learning_rate": 0.00010646080760095012, "loss": 0.9764, "step": 1970 }, { "epoch": 2.351543942992874, "grad_norm": 0.0873328223824501, "learning_rate": 0.00010598574821852732, "loss": 0.9905, "step": 1980 }, { "epoch": 2.3634204275534443, "grad_norm": 0.13550794124603271, "learning_rate": 0.00010551068883610451, "loss": 1.2503, "step": 1990 }, { "epoch": 2.375296912114014, "grad_norm": 0.09414742141962051, "learning_rate": 0.00010503562945368171, "loss": 1.2675, "step": 2000 }, { "epoch": 2.3871733966745845, "grad_norm": 0.07908231019973755, "learning_rate": 0.00010456057007125892, "loss": 1.0461, "step": 2010 }, { "epoch": 2.3990498812351544, "grad_norm": 0.14820969104766846, "learning_rate": 0.00010408551068883611, "loss": 1.0111, "step": 2020 }, { "epoch": 2.4109263657957243, "grad_norm": 0.09235847741365433, "learning_rate": 0.00010361045130641331, "loss": 0.9179, "step": 2030 }, { "epoch": 2.4228028503562946, "grad_norm": 0.05410043150186539, "learning_rate": 0.00010313539192399052, "loss": 1.1183, "step": 2040 }, { "epoch": 2.4346793349168645, "grad_norm": 0.15485779941082, "learning_rate": 0.00010266033254156771, "loss": 1.3183, "step": 2050 }, { "epoch": 2.446555819477435, "grad_norm": 0.26478269696235657, "learning_rate": 0.0001021852731591449, "loss": 1.0667, "step": 2060 }, { "epoch": 2.4584323040380047, "grad_norm": 0.13515473902225494, "learning_rate": 0.0001017102137767221, "loss": 0.9078, "step": 2070 }, { "epoch": 2.470308788598575, "grad_norm": 0.07300888746976852, "learning_rate": 0.0001012351543942993, "loss": 1.0377, "step": 2080 }, { "epoch": 2.482185273159145, "grad_norm": 0.060126177966594696, "learning_rate": 0.00010076009501187649, "loss": 0.9788, "step": 2090 }, { "epoch": 2.494061757719715, "grad_norm": 0.17379595339298248, "learning_rate": 0.00010028503562945369, "loss": 1.0362, "step": 2100 }, { "epoch": 2.505938242280285, "grad_norm": 0.057881057262420654, "learning_rate": 9.980997624703088e-05, "loss": 0.8976, "step": 2110 }, { "epoch": 2.517814726840855, "grad_norm": 0.2135135978460312, "learning_rate": 9.933491686460809e-05, "loss": 0.9377, "step": 2120 }, { "epoch": 2.529691211401425, "grad_norm": 0.042427126318216324, "learning_rate": 9.885985748218528e-05, "loss": 0.9857, "step": 2130 }, { "epoch": 2.5415676959619953, "grad_norm": 0.21827517449855804, "learning_rate": 9.838479809976248e-05, "loss": 0.9525, "step": 2140 }, { "epoch": 2.553444180522565, "grad_norm": 0.0543668232858181, "learning_rate": 9.790973871733967e-05, "loss": 1.0002, "step": 2150 }, { "epoch": 2.5653206650831355, "grad_norm": 0.08075886219739914, "learning_rate": 9.743467933491687e-05, "loss": 0.8931, "step": 2160 }, { "epoch": 2.5771971496437054, "grad_norm": 0.2579517960548401, "learning_rate": 9.695961995249406e-05, "loss": 0.9135, "step": 2170 }, { "epoch": 2.5890736342042757, "grad_norm": 0.050557930022478104, "learning_rate": 9.648456057007126e-05, "loss": 0.827, "step": 2180 }, { "epoch": 2.6009501187648456, "grad_norm": 0.20607951283454895, "learning_rate": 9.600950118764847e-05, "loss": 0.9406, "step": 2190 }, { "epoch": 2.6128266033254155, "grad_norm": 0.051520612090826035, "learning_rate": 9.553444180522566e-05, "loss": 0.8994, "step": 2200 }, { "epoch": 2.624703087885986, "grad_norm": 0.06364279240369797, "learning_rate": 9.505938242280286e-05, "loss": 1.2281, "step": 2210 }, { "epoch": 2.6365795724465557, "grad_norm": 0.0908183827996254, "learning_rate": 9.458432304038005e-05, "loss": 1.0889, "step": 2220 }, { "epoch": 2.648456057007126, "grad_norm": 0.15008820593357086, "learning_rate": 9.410926365795725e-05, "loss": 0.8651, "step": 2230 }, { "epoch": 2.660332541567696, "grad_norm": 0.06086505949497223, "learning_rate": 9.363420427553444e-05, "loss": 0.8643, "step": 2240 }, { "epoch": 2.6722090261282663, "grad_norm": 0.12590943276882172, "learning_rate": 9.315914489311164e-05, "loss": 0.8751, "step": 2250 }, { "epoch": 2.684085510688836, "grad_norm": 0.07025561481714249, "learning_rate": 9.268408551068884e-05, "loss": 0.8319, "step": 2260 }, { "epoch": 2.695961995249406, "grad_norm": 0.05536743625998497, "learning_rate": 9.220902612826604e-05, "loss": 0.7906, "step": 2270 }, { "epoch": 2.7078384798099764, "grad_norm": 0.09530112147331238, "learning_rate": 9.173396674584323e-05, "loss": 0.8542, "step": 2280 }, { "epoch": 2.7197149643705463, "grad_norm": 0.0815366581082344, "learning_rate": 9.125890736342043e-05, "loss": 0.7518, "step": 2290 }, { "epoch": 2.731591448931116, "grad_norm": 0.062431272119283676, "learning_rate": 9.078384798099762e-05, "loss": 1.1814, "step": 2300 }, { "epoch": 2.7434679334916865, "grad_norm": 0.06672868132591248, "learning_rate": 9.030878859857482e-05, "loss": 1.0492, "step": 2310 }, { "epoch": 2.7553444180522564, "grad_norm": 0.3035074770450592, "learning_rate": 8.983372921615201e-05, "loss": 0.9671, "step": 2320 }, { "epoch": 2.7672209026128267, "grad_norm": 0.16409927606582642, "learning_rate": 8.935866983372922e-05, "loss": 1.011, "step": 2330 }, { "epoch": 2.7790973871733966, "grad_norm": 0.07125268131494522, "learning_rate": 8.888361045130642e-05, "loss": 0.9033, "step": 2340 }, { "epoch": 2.790973871733967, "grad_norm": 0.22371023893356323, "learning_rate": 8.840855106888361e-05, "loss": 1.3109, "step": 2350 }, { "epoch": 2.802850356294537, "grad_norm": 0.05274183303117752, "learning_rate": 8.79334916864608e-05, "loss": 0.8897, "step": 2360 }, { "epoch": 2.8147268408551067, "grad_norm": 0.0982283353805542, "learning_rate": 8.7458432304038e-05, "loss": 1.0437, "step": 2370 }, { "epoch": 2.826603325415677, "grad_norm": 0.06932730972766876, "learning_rate": 8.698337292161521e-05, "loss": 0.8968, "step": 2380 }, { "epoch": 2.838479809976247, "grad_norm": 0.06926975399255753, "learning_rate": 8.65083135391924e-05, "loss": 0.9674, "step": 2390 }, { "epoch": 2.850356294536817, "grad_norm": 0.2094772607088089, "learning_rate": 8.60332541567696e-05, "loss": 0.8637, "step": 2400 }, { "epoch": 2.862232779097387, "grad_norm": 0.04431522637605667, "learning_rate": 8.55581947743468e-05, "loss": 1.0707, "step": 2410 }, { "epoch": 2.8741092636579575, "grad_norm": 0.0706724151968956, "learning_rate": 8.5083135391924e-05, "loss": 1.1387, "step": 2420 }, { "epoch": 2.8859857482185274, "grad_norm": 0.13831046223640442, "learning_rate": 8.46080760095012e-05, "loss": 0.8456, "step": 2430 }, { "epoch": 2.8978622327790973, "grad_norm": 0.22409480810165405, "learning_rate": 8.413301662707839e-05, "loss": 1.0635, "step": 2440 }, { "epoch": 2.9097387173396676, "grad_norm": 0.10079943388700485, "learning_rate": 8.365795724465559e-05, "loss": 0.9804, "step": 2450 }, { "epoch": 2.9216152019002375, "grad_norm": 0.042890481650829315, "learning_rate": 8.318289786223278e-05, "loss": 0.8046, "step": 2460 }, { "epoch": 2.9334916864608074, "grad_norm": 0.3232579827308655, "learning_rate": 8.270783847980999e-05, "loss": 0.9013, "step": 2470 }, { "epoch": 2.9453681710213777, "grad_norm": 0.06542018055915833, "learning_rate": 8.223277909738719e-05, "loss": 1.1108, "step": 2480 }, { "epoch": 2.9572446555819476, "grad_norm": 0.06973831355571747, "learning_rate": 8.175771971496438e-05, "loss": 0.9036, "step": 2490 }, { "epoch": 2.969121140142518, "grad_norm": 0.10869220644235611, "learning_rate": 8.128266033254157e-05, "loss": 1.184, "step": 2500 }, { "epoch": 2.980997624703088, "grad_norm": 0.2329593151807785, "learning_rate": 8.080760095011877e-05, "loss": 0.8057, "step": 2510 }, { "epoch": 2.992874109263658, "grad_norm": 0.10450086742639542, "learning_rate": 8.033254156769596e-05, "loss": 0.9202, "step": 2520 }, { "epoch": 3.0, "eval_loss": 1.7657307386398315, "eval_runtime": 47.3747, "eval_samples_per_second": 7.916, "eval_steps_per_second": 1.984, "step": 2526 }, { "epoch": 3.004750593824228, "grad_norm": 0.0954449325799942, "learning_rate": 7.985748218527316e-05, "loss": 0.9937, "step": 2530 }, { "epoch": 3.016627078384798, "grad_norm": 0.1135089173913002, "learning_rate": 7.938242280285037e-05, "loss": 0.9659, "step": 2540 }, { "epoch": 3.0285035629453683, "grad_norm": 0.2668828070163727, "learning_rate": 7.890736342042756e-05, "loss": 1.1051, "step": 2550 }, { "epoch": 3.040380047505938, "grad_norm": 0.04385481774806976, "learning_rate": 7.843230403800476e-05, "loss": 0.905, "step": 2560 }, { "epoch": 3.0522565320665085, "grad_norm": 0.16458982229232788, "learning_rate": 7.795724465558195e-05, "loss": 0.95, "step": 2570 }, { "epoch": 3.0641330166270784, "grad_norm": 0.06375309824943542, "learning_rate": 7.748218527315915e-05, "loss": 0.7936, "step": 2580 }, { "epoch": 3.0760095011876483, "grad_norm": 0.2245360165834427, "learning_rate": 7.700712589073634e-05, "loss": 0.8329, "step": 2590 }, { "epoch": 3.0878859857482186, "grad_norm": 0.1618458777666092, "learning_rate": 7.653206650831354e-05, "loss": 1.1232, "step": 2600 }, { "epoch": 3.0997624703087885, "grad_norm": 0.17201924324035645, "learning_rate": 7.605700712589075e-05, "loss": 1.0214, "step": 2610 }, { "epoch": 3.111638954869359, "grad_norm": 0.0636308342218399, "learning_rate": 7.558194774346794e-05, "loss": 0.8034, "step": 2620 }, { "epoch": 3.1235154394299287, "grad_norm": 0.31130924820899963, "learning_rate": 7.510688836104514e-05, "loss": 0.9235, "step": 2630 }, { "epoch": 3.1353919239904986, "grad_norm": 0.05764823779463768, "learning_rate": 7.463182897862233e-05, "loss": 1.1852, "step": 2640 }, { "epoch": 3.147268408551069, "grad_norm": 0.0911896675825119, "learning_rate": 7.415676959619953e-05, "loss": 0.8264, "step": 2650 }, { "epoch": 3.159144893111639, "grad_norm": 0.05941038951277733, "learning_rate": 7.368171021377672e-05, "loss": 1.0462, "step": 2660 }, { "epoch": 3.171021377672209, "grad_norm": 0.10101119428873062, "learning_rate": 7.320665083135391e-05, "loss": 0.7916, "step": 2670 }, { "epoch": 3.182897862232779, "grad_norm": 0.049820832908153534, "learning_rate": 7.273159144893112e-05, "loss": 1.0767, "step": 2680 }, { "epoch": 3.1947743467933494, "grad_norm": 0.0488307848572731, "learning_rate": 7.225653206650832e-05, "loss": 1.1087, "step": 2690 }, { "epoch": 3.2066508313539193, "grad_norm": 0.044484905898571014, "learning_rate": 7.178147268408551e-05, "loss": 0.8214, "step": 2700 }, { "epoch": 3.218527315914489, "grad_norm": 0.21011117100715637, "learning_rate": 7.130641330166271e-05, "loss": 1.016, "step": 2710 }, { "epoch": 3.2304038004750595, "grad_norm": 0.11267800629138947, "learning_rate": 7.08313539192399e-05, "loss": 0.8472, "step": 2720 }, { "epoch": 3.2422802850356294, "grad_norm": 0.041125789284706116, "learning_rate": 7.03562945368171e-05, "loss": 0.8789, "step": 2730 }, { "epoch": 3.2541567695961997, "grad_norm": 0.17443105578422546, "learning_rate": 6.988123515439429e-05, "loss": 0.9556, "step": 2740 }, { "epoch": 3.2660332541567696, "grad_norm": 0.062060561031103134, "learning_rate": 6.94061757719715e-05, "loss": 0.994, "step": 2750 }, { "epoch": 3.2779097387173395, "grad_norm": 0.06218833476305008, "learning_rate": 6.89311163895487e-05, "loss": 0.7544, "step": 2760 }, { "epoch": 3.28978622327791, "grad_norm": 0.04432019218802452, "learning_rate": 6.845605700712589e-05, "loss": 0.9177, "step": 2770 }, { "epoch": 3.3016627078384797, "grad_norm": 0.10758528113365173, "learning_rate": 6.798099762470309e-05, "loss": 0.9384, "step": 2780 }, { "epoch": 3.31353919239905, "grad_norm": 0.08597265183925629, "learning_rate": 6.750593824228028e-05, "loss": 1.105, "step": 2790 }, { "epoch": 3.32541567695962, "grad_norm": 0.054796867072582245, "learning_rate": 6.703087885985749e-05, "loss": 1.0362, "step": 2800 }, { "epoch": 3.33729216152019, "grad_norm": 0.04514233022928238, "learning_rate": 6.655581947743468e-05, "loss": 1.0121, "step": 2810 }, { "epoch": 3.34916864608076, "grad_norm": 0.07130993902683258, "learning_rate": 6.608076009501188e-05, "loss": 0.7462, "step": 2820 }, { "epoch": 3.36104513064133, "grad_norm": 0.04843717813491821, "learning_rate": 6.560570071258907e-05, "loss": 0.8745, "step": 2830 }, { "epoch": 3.3729216152019004, "grad_norm": 0.19065134227275848, "learning_rate": 6.513064133016628e-05, "loss": 0.8703, "step": 2840 }, { "epoch": 3.3847980997624703, "grad_norm": 0.10798069834709167, "learning_rate": 6.465558194774348e-05, "loss": 1.007, "step": 2850 }, { "epoch": 3.3966745843230406, "grad_norm": 0.0635639876127243, "learning_rate": 6.418052256532067e-05, "loss": 1.0906, "step": 2860 }, { "epoch": 3.4085510688836105, "grad_norm": 0.04843032360076904, "learning_rate": 6.370546318289787e-05, "loss": 1.0584, "step": 2870 }, { "epoch": 3.4204275534441804, "grad_norm": 0.05005785822868347, "learning_rate": 6.323040380047506e-05, "loss": 0.8419, "step": 2880 }, { "epoch": 3.4323040380047507, "grad_norm": 0.060981009155511856, "learning_rate": 6.275534441805227e-05, "loss": 0.8225, "step": 2890 }, { "epoch": 3.4441805225653206, "grad_norm": 0.4528219997882843, "learning_rate": 6.228028503562946e-05, "loss": 0.9944, "step": 2900 }, { "epoch": 3.456057007125891, "grad_norm": 0.07064502686262131, "learning_rate": 6.180522565320666e-05, "loss": 0.7643, "step": 2910 }, { "epoch": 3.467933491686461, "grad_norm": 0.07325369119644165, "learning_rate": 6.133016627078385e-05, "loss": 1.1436, "step": 2920 }, { "epoch": 3.4798099762470307, "grad_norm": 0.16508881747722626, "learning_rate": 6.085510688836105e-05, "loss": 0.9703, "step": 2930 }, { "epoch": 3.491686460807601, "grad_norm": 0.05933353677392006, "learning_rate": 6.0380047505938244e-05, "loss": 0.8623, "step": 2940 }, { "epoch": 3.503562945368171, "grad_norm": 0.04389052465558052, "learning_rate": 5.990498812351544e-05, "loss": 0.8283, "step": 2950 }, { "epoch": 3.5154394299287413, "grad_norm": 0.25019779801368713, "learning_rate": 5.942992874109265e-05, "loss": 0.9623, "step": 2960 }, { "epoch": 3.527315914489311, "grad_norm": 0.10527379810810089, "learning_rate": 5.895486935866984e-05, "loss": 0.8582, "step": 2970 }, { "epoch": 3.539192399049881, "grad_norm": 0.0807216614484787, "learning_rate": 5.847980997624704e-05, "loss": 1.013, "step": 2980 }, { "epoch": 3.5510688836104514, "grad_norm": 0.06806305795907974, "learning_rate": 5.800475059382423e-05, "loss": 1.0019, "step": 2990 }, { "epoch": 3.5629453681710213, "grad_norm": 0.06385791301727295, "learning_rate": 5.752969121140143e-05, "loss": 1.0457, "step": 3000 }, { "epoch": 3.5748218527315916, "grad_norm": 0.15625765919685364, "learning_rate": 5.705463182897862e-05, "loss": 1.1656, "step": 3010 }, { "epoch": 3.5866983372921615, "grad_norm": 0.037077538669109344, "learning_rate": 5.657957244655582e-05, "loss": 0.8982, "step": 3020 }, { "epoch": 3.598574821852732, "grad_norm": 0.05139300227165222, "learning_rate": 5.6104513064133025e-05, "loss": 0.8765, "step": 3030 }, { "epoch": 3.6104513064133017, "grad_norm": 0.04318641126155853, "learning_rate": 5.562945368171022e-05, "loss": 0.7967, "step": 3040 }, { "epoch": 3.6223277909738716, "grad_norm": 0.040078990161418915, "learning_rate": 5.5154394299287415e-05, "loss": 0.73, "step": 3050 }, { "epoch": 3.634204275534442, "grad_norm": 0.06352245062589645, "learning_rate": 5.467933491686461e-05, "loss": 1.1627, "step": 3060 }, { "epoch": 3.646080760095012, "grad_norm": 0.03417419269680977, "learning_rate": 5.4204275534441805e-05, "loss": 0.8954, "step": 3070 }, { "epoch": 3.6579572446555817, "grad_norm": 0.06765933334827423, "learning_rate": 5.3729216152019e-05, "loss": 0.8146, "step": 3080 }, { "epoch": 3.669833729216152, "grad_norm": 0.0948178619146347, "learning_rate": 5.3254156769596194e-05, "loss": 0.7459, "step": 3090 }, { "epoch": 3.6817102137767224, "grad_norm": 0.41120731830596924, "learning_rate": 5.27790973871734e-05, "loss": 1.0802, "step": 3100 }, { "epoch": 3.6935866983372923, "grad_norm": 0.04571249336004257, "learning_rate": 5.23040380047506e-05, "loss": 1.0113, "step": 3110 }, { "epoch": 3.705463182897862, "grad_norm": 0.06355877220630646, "learning_rate": 5.182897862232779e-05, "loss": 0.8512, "step": 3120 }, { "epoch": 3.7173396674584325, "grad_norm": 0.20209620893001556, "learning_rate": 5.135391923990499e-05, "loss": 1.1125, "step": 3130 }, { "epoch": 3.7292161520190024, "grad_norm": 0.0654263123869896, "learning_rate": 5.087885985748219e-05, "loss": 0.9359, "step": 3140 }, { "epoch": 3.7410926365795723, "grad_norm": 0.1410745084285736, "learning_rate": 5.0403800475059384e-05, "loss": 0.9695, "step": 3150 }, { "epoch": 3.7529691211401426, "grad_norm": 0.0441279262304306, "learning_rate": 4.9928741092636586e-05, "loss": 0.9279, "step": 3160 }, { "epoch": 3.7648456057007125, "grad_norm": 0.042252734303474426, "learning_rate": 4.945368171021378e-05, "loss": 0.9513, "step": 3170 }, { "epoch": 3.776722090261283, "grad_norm": 0.10475219786167145, "learning_rate": 4.8978622327790975e-05, "loss": 0.9602, "step": 3180 }, { "epoch": 3.7885985748218527, "grad_norm": 0.561947226524353, "learning_rate": 4.850356294536818e-05, "loss": 0.9196, "step": 3190 }, { "epoch": 3.800475059382423, "grad_norm": 0.07738777995109558, "learning_rate": 4.802850356294537e-05, "loss": 0.8386, "step": 3200 }, { "epoch": 3.812351543942993, "grad_norm": 0.0962240993976593, "learning_rate": 4.755344418052257e-05, "loss": 0.8713, "step": 3210 }, { "epoch": 3.824228028503563, "grad_norm": 0.19785310328006744, "learning_rate": 4.707838479809977e-05, "loss": 0.9994, "step": 3220 }, { "epoch": 3.836104513064133, "grad_norm": 0.1203259751200676, "learning_rate": 4.660332541567696e-05, "loss": 0.8608, "step": 3230 }, { "epoch": 3.847980997624703, "grad_norm": 0.041854988783597946, "learning_rate": 4.612826603325416e-05, "loss": 0.7506, "step": 3240 }, { "epoch": 3.859857482185273, "grad_norm": 0.03335632383823395, "learning_rate": 4.565320665083135e-05, "loss": 0.721, "step": 3250 }, { "epoch": 3.8717339667458432, "grad_norm": 0.110906220972538, "learning_rate": 4.5178147268408555e-05, "loss": 0.8615, "step": 3260 }, { "epoch": 3.883610451306413, "grad_norm": 0.23366965353488922, "learning_rate": 4.470308788598575e-05, "loss": 1.2421, "step": 3270 }, { "epoch": 3.8954869358669835, "grad_norm": 0.07156733423471451, "learning_rate": 4.4228028503562945e-05, "loss": 0.8366, "step": 3280 }, { "epoch": 3.9073634204275534, "grad_norm": 0.169892355799675, "learning_rate": 4.3752969121140146e-05, "loss": 0.8799, "step": 3290 }, { "epoch": 3.9192399049881237, "grad_norm": 0.04358002915978432, "learning_rate": 4.327790973871734e-05, "loss": 0.9512, "step": 3300 }, { "epoch": 3.9311163895486936, "grad_norm": 0.21528488397598267, "learning_rate": 4.2802850356294536e-05, "loss": 0.8054, "step": 3310 }, { "epoch": 3.9429928741092635, "grad_norm": 0.07016022503376007, "learning_rate": 4.232779097387173e-05, "loss": 0.9903, "step": 3320 }, { "epoch": 3.954869358669834, "grad_norm": 0.06365745514631271, "learning_rate": 4.185273159144893e-05, "loss": 0.7909, "step": 3330 }, { "epoch": 3.9667458432304037, "grad_norm": 0.20040567219257355, "learning_rate": 4.137767220902613e-05, "loss": 0.9326, "step": 3340 }, { "epoch": 3.978622327790974, "grad_norm": 0.0482853502035141, "learning_rate": 4.090261282660333e-05, "loss": 0.811, "step": 3350 }, { "epoch": 3.990498812351544, "grad_norm": 0.09436655789613724, "learning_rate": 4.0427553444180524e-05, "loss": 0.8454, "step": 3360 }, { "epoch": 4.0, "eval_loss": 1.6603933572769165, "eval_runtime": 47.38, "eval_samples_per_second": 7.915, "eval_steps_per_second": 1.984, "step": 3368 }, { "epoch": 4.002375296912114, "grad_norm": 0.0664132684469223, "learning_rate": 3.9952494061757725e-05, "loss": 1.0412, "step": 3370 }, { "epoch": 4.014251781472684, "grad_norm": 0.04716278240084648, "learning_rate": 3.947743467933492e-05, "loss": 0.9108, "step": 3380 }, { "epoch": 4.026128266033254, "grad_norm": 0.07922456413507462, "learning_rate": 3.9002375296912115e-05, "loss": 0.981, "step": 3390 }, { "epoch": 4.038004750593824, "grad_norm": 0.12802214920520782, "learning_rate": 3.852731591448932e-05, "loss": 1.1224, "step": 3400 }, { "epoch": 4.049881235154395, "grad_norm": 0.2722633481025696, "learning_rate": 3.805225653206651e-05, "loss": 0.9692, "step": 3410 }, { "epoch": 4.061757719714964, "grad_norm": 0.056470565497875214, "learning_rate": 3.757719714964371e-05, "loss": 0.8099, "step": 3420 }, { "epoch": 4.0736342042755345, "grad_norm": 0.039738211780786514, "learning_rate": 3.710213776722091e-05, "loss": 1.0946, "step": 3430 }, { "epoch": 4.085510688836105, "grad_norm": 0.06954427808523178, "learning_rate": 3.66270783847981e-05, "loss": 0.9666, "step": 3440 }, { "epoch": 4.097387173396674, "grad_norm": 0.13244707882404327, "learning_rate": 3.61520190023753e-05, "loss": 1.1186, "step": 3450 }, { "epoch": 4.109263657957245, "grad_norm": 0.050309810787439346, "learning_rate": 3.567695961995249e-05, "loss": 0.8773, "step": 3460 }, { "epoch": 4.121140142517815, "grad_norm": 0.041740868240594864, "learning_rate": 3.5201900237529695e-05, "loss": 0.9862, "step": 3470 }, { "epoch": 4.133016627078385, "grad_norm": 0.13563305139541626, "learning_rate": 3.472684085510689e-05, "loss": 0.8709, "step": 3480 }, { "epoch": 4.144893111638955, "grad_norm": 0.0343937985599041, "learning_rate": 3.4251781472684084e-05, "loss": 0.7728, "step": 3490 }, { "epoch": 4.156769596199525, "grad_norm": 0.08868346363306046, "learning_rate": 3.3776722090261286e-05, "loss": 1.0534, "step": 3500 }, { "epoch": 4.168646080760095, "grad_norm": 0.2531236410140991, "learning_rate": 3.330166270783848e-05, "loss": 1.2138, "step": 3510 }, { "epoch": 4.180522565320665, "grad_norm": 0.07945922762155533, "learning_rate": 3.2826603325415676e-05, "loss": 0.8253, "step": 3520 }, { "epoch": 4.192399049881235, "grad_norm": 0.07955809682607651, "learning_rate": 3.235154394299287e-05, "loss": 0.9263, "step": 3530 }, { "epoch": 4.2042755344418055, "grad_norm": 0.04066227748990059, "learning_rate": 3.187648456057007e-05, "loss": 0.9605, "step": 3540 }, { "epoch": 4.216152019002375, "grad_norm": 0.04212834686040878, "learning_rate": 3.140142517814727e-05, "loss": 0.7031, "step": 3550 }, { "epoch": 4.228028503562945, "grad_norm": 0.07302612066268921, "learning_rate": 3.092636579572447e-05, "loss": 0.9753, "step": 3560 }, { "epoch": 4.239904988123516, "grad_norm": 0.07105498760938644, "learning_rate": 3.0451306413301667e-05, "loss": 0.9347, "step": 3570 }, { "epoch": 4.251781472684086, "grad_norm": 0.06369230151176453, "learning_rate": 2.9976247030878862e-05, "loss": 0.7127, "step": 3580 }, { "epoch": 4.263657957244655, "grad_norm": 0.20201636850833893, "learning_rate": 2.9501187648456057e-05, "loss": 1.0062, "step": 3590 }, { "epoch": 4.275534441805226, "grad_norm": 0.03355458378791809, "learning_rate": 2.9026128266033255e-05, "loss": 0.7946, "step": 3600 }, { "epoch": 4.287410926365796, "grad_norm": 0.0544753298163414, "learning_rate": 2.8551068883610453e-05, "loss": 0.8509, "step": 3610 }, { "epoch": 4.2992874109263655, "grad_norm": 0.09867158532142639, "learning_rate": 2.807600950118765e-05, "loss": 1.127, "step": 3620 }, { "epoch": 4.311163895486936, "grad_norm": 0.03889361396431923, "learning_rate": 2.7600950118764846e-05, "loss": 0.8523, "step": 3630 }, { "epoch": 4.323040380047506, "grad_norm": 0.19011080265045166, "learning_rate": 2.7125890736342048e-05, "loss": 0.9723, "step": 3640 }, { "epoch": 4.334916864608076, "grad_norm": 0.08830198645591736, "learning_rate": 2.6650831353919243e-05, "loss": 0.7572, "step": 3650 }, { "epoch": 4.346793349168646, "grad_norm": 0.0878998190164566, "learning_rate": 2.6175771971496438e-05, "loss": 0.9713, "step": 3660 }, { "epoch": 4.358669833729216, "grad_norm": 0.057567398995161057, "learning_rate": 2.5700712589073633e-05, "loss": 0.9692, "step": 3670 }, { "epoch": 4.370546318289787, "grad_norm": 0.06174805015325546, "learning_rate": 2.5225653206650834e-05, "loss": 0.8154, "step": 3680 }, { "epoch": 4.382422802850356, "grad_norm": 0.028641091659665108, "learning_rate": 2.475059382422803e-05, "loss": 0.754, "step": 3690 }, { "epoch": 4.394299287410926, "grad_norm": 0.03536592051386833, "learning_rate": 2.4275534441805224e-05, "loss": 0.8561, "step": 3700 }, { "epoch": 4.406175771971497, "grad_norm": 0.07805202156305313, "learning_rate": 2.3800475059382422e-05, "loss": 0.8198, "step": 3710 }, { "epoch": 4.418052256532066, "grad_norm": 0.06069310009479523, "learning_rate": 2.332541567695962e-05, "loss": 0.836, "step": 3720 }, { "epoch": 4.4299287410926365, "grad_norm": 0.05068911612033844, "learning_rate": 2.285035629453682e-05, "loss": 0.8898, "step": 3730 }, { "epoch": 4.441805225653207, "grad_norm": 0.06551819294691086, "learning_rate": 2.2375296912114017e-05, "loss": 0.6919, "step": 3740 }, { "epoch": 4.453681710213777, "grad_norm": 0.11634793132543564, "learning_rate": 2.1900237529691212e-05, "loss": 0.8043, "step": 3750 }, { "epoch": 4.465558194774347, "grad_norm": 0.13671091198921204, "learning_rate": 2.142517814726841e-05, "loss": 0.895, "step": 3760 }, { "epoch": 4.477434679334917, "grad_norm": 0.05117421597242355, "learning_rate": 2.0950118764845605e-05, "loss": 0.9661, "step": 3770 }, { "epoch": 4.489311163895487, "grad_norm": 0.07194402813911438, "learning_rate": 2.0475059382422803e-05, "loss": 0.8791, "step": 3780 }, { "epoch": 4.501187648456057, "grad_norm": 0.04208892583847046, "learning_rate": 2e-05, "loss": 0.9381, "step": 3790 }, { "epoch": 4.513064133016627, "grad_norm": 0.06842586398124695, "learning_rate": 1.9524940617577197e-05, "loss": 0.9378, "step": 3800 }, { "epoch": 4.524940617577197, "grad_norm": 0.06581088900566101, "learning_rate": 1.9049881235154395e-05, "loss": 0.7979, "step": 3810 }, { "epoch": 4.536817102137768, "grad_norm": 0.1313990205526352, "learning_rate": 1.8574821852731593e-05, "loss": 1.0281, "step": 3820 }, { "epoch": 4.548693586698337, "grad_norm": 0.04237492382526398, "learning_rate": 1.809976247030879e-05, "loss": 0.8999, "step": 3830 }, { "epoch": 4.5605700712589075, "grad_norm": 0.16546116769313812, "learning_rate": 1.7624703087885986e-05, "loss": 0.887, "step": 3840 }, { "epoch": 4.572446555819478, "grad_norm": 0.04888550564646721, "learning_rate": 1.7149643705463184e-05, "loss": 0.8049, "step": 3850 }, { "epoch": 4.584323040380047, "grad_norm": 0.08607861399650574, "learning_rate": 1.6674584323040383e-05, "loss": 0.8848, "step": 3860 }, { "epoch": 4.596199524940618, "grad_norm": 0.203558549284935, "learning_rate": 1.6199524940617578e-05, "loss": 0.9295, "step": 3870 }, { "epoch": 4.608076009501188, "grad_norm": 0.0457516685128212, "learning_rate": 1.5724465558194776e-05, "loss": 1.0568, "step": 3880 }, { "epoch": 4.619952494061757, "grad_norm": 0.04271242395043373, "learning_rate": 1.524940617577197e-05, "loss": 1.1772, "step": 3890 }, { "epoch": 4.631828978622328, "grad_norm": 0.05776529759168625, "learning_rate": 1.4774346793349169e-05, "loss": 0.9445, "step": 3900 }, { "epoch": 4.643705463182898, "grad_norm": 0.07290226221084595, "learning_rate": 1.4299287410926366e-05, "loss": 0.819, "step": 3910 }, { "epoch": 4.6555819477434675, "grad_norm": 0.23439574241638184, "learning_rate": 1.3824228028503564e-05, "loss": 1.0728, "step": 3920 }, { "epoch": 4.667458432304038, "grad_norm": 0.2876322567462921, "learning_rate": 1.3349168646080762e-05, "loss": 1.0477, "step": 3930 }, { "epoch": 4.679334916864608, "grad_norm": 0.05245677009224892, "learning_rate": 1.2874109263657957e-05, "loss": 0.783, "step": 3940 }, { "epoch": 4.6912114014251785, "grad_norm": 0.05827531963586807, "learning_rate": 1.2399049881235155e-05, "loss": 0.7584, "step": 3950 }, { "epoch": 4.703087885985748, "grad_norm": 0.2577458322048187, "learning_rate": 1.1923990498812353e-05, "loss": 0.7678, "step": 3960 }, { "epoch": 4.714964370546318, "grad_norm": 0.10721047222614288, "learning_rate": 1.144893111638955e-05, "loss": 0.8898, "step": 3970 }, { "epoch": 4.726840855106889, "grad_norm": 0.08441893756389618, "learning_rate": 1.0973871733966747e-05, "loss": 0.8402, "step": 3980 }, { "epoch": 4.738717339667458, "grad_norm": 0.09087421745061874, "learning_rate": 1.0498812351543943e-05, "loss": 1.1081, "step": 3990 }, { "epoch": 4.750593824228028, "grad_norm": 0.04736216366291046, "learning_rate": 1.002375296912114e-05, "loss": 0.6902, "step": 4000 }, { "epoch": 4.762470308788599, "grad_norm": 0.48896902799606323, "learning_rate": 9.548693586698338e-06, "loss": 0.8916, "step": 4010 }, { "epoch": 4.774346793349169, "grad_norm": 0.08265817910432816, "learning_rate": 9.073634204275535e-06, "loss": 0.9575, "step": 4020 }, { "epoch": 4.7862232779097384, "grad_norm": 0.08426305651664734, "learning_rate": 8.598574821852733e-06, "loss": 0.9751, "step": 4030 }, { "epoch": 4.798099762470309, "grad_norm": 0.09328274428844452, "learning_rate": 8.12351543942993e-06, "loss": 0.8666, "step": 4040 }, { "epoch": 4.809976247030879, "grad_norm": 0.040312737226486206, "learning_rate": 7.648456057007126e-06, "loss": 0.8568, "step": 4050 }, { "epoch": 4.821852731591449, "grad_norm": 0.0362665094435215, "learning_rate": 7.173396674584323e-06, "loss": 0.736, "step": 4060 }, { "epoch": 4.833729216152019, "grad_norm": 0.05081469193100929, "learning_rate": 6.69833729216152e-06, "loss": 0.8109, "step": 4070 }, { "epoch": 4.845605700712589, "grad_norm": 0.05075117200613022, "learning_rate": 6.223277909738718e-06, "loss": 0.7773, "step": 4080 }, { "epoch": 4.85748218527316, "grad_norm": 0.047186557203531265, "learning_rate": 5.748218527315915e-06, "loss": 0.8404, "step": 4090 }, { "epoch": 4.869358669833729, "grad_norm": 0.20718060433864594, "learning_rate": 5.273159144893112e-06, "loss": 0.955, "step": 4100 }, { "epoch": 4.881235154394299, "grad_norm": 0.17017701268196106, "learning_rate": 4.798099762470309e-06, "loss": 0.8501, "step": 4110 }, { "epoch": 4.89311163895487, "grad_norm": 0.05619393661618233, "learning_rate": 4.323040380047506e-06, "loss": 0.9074, "step": 4120 }, { "epoch": 4.904988123515439, "grad_norm": 0.06167184188961983, "learning_rate": 3.8479809976247036e-06, "loss": 0.8147, "step": 4130 }, { "epoch": 4.9168646080760094, "grad_norm": 0.06882195919752121, "learning_rate": 3.3729216152019e-06, "loss": 0.7927, "step": 4140 }, { "epoch": 4.92874109263658, "grad_norm": 0.08290237188339233, "learning_rate": 2.8978622327790975e-06, "loss": 0.8274, "step": 4150 }, { "epoch": 4.94061757719715, "grad_norm": 0.08587727695703506, "learning_rate": 2.422802850356295e-06, "loss": 0.808, "step": 4160 }, { "epoch": 4.95249406175772, "grad_norm": 0.05585173889994621, "learning_rate": 1.947743467933492e-06, "loss": 1.0701, "step": 4170 }, { "epoch": 4.96437054631829, "grad_norm": 0.039784397929906845, "learning_rate": 1.472684085510689e-06, "loss": 1.0414, "step": 4180 }, { "epoch": 4.97624703087886, "grad_norm": 0.03638745844364166, "learning_rate": 9.97624703087886e-07, "loss": 0.9453, "step": 4190 }, { "epoch": 4.98812351543943, "grad_norm": 0.030311966314911842, "learning_rate": 5.225653206650831e-07, "loss": 0.7845, "step": 4200 }, { "epoch": 5.0, "grad_norm": 0.10933826118707657, "learning_rate": 4.750593824228029e-08, "loss": 0.7601, "step": 4210 }, { "epoch": 5.0, "eval_loss": 1.6282436847686768, "eval_runtime": 47.3815, "eval_samples_per_second": 7.914, "eval_steps_per_second": 1.984, "step": 4210 }, { "epoch": 5.0, "step": 4210, "total_flos": 7.358110989130138e+17, "train_loss": 1.2386485581160158, "train_runtime": 4897.8019, "train_samples_per_second": 3.437, "train_steps_per_second": 0.86 }, { "epoch": 5.0, "eval_loss": 1.6282436847686768, "eval_runtime": 47.1204, "eval_samples_per_second": 7.958, "eval_steps_per_second": 1.995, "step": 4210 } ], "logging_steps": 10, "max_steps": 4210, "num_input_tokens_seen": 0, "num_train_epochs": 5, "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": 7.358110989130138e+17, "train_batch_size": 4, "trial_name": null, "trial_params": null }