{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 3895,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.012836970474967908,
      "grad_norm": 76.69837951660156,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 14.1357,
      "step": 10
    },
    {
      "epoch": 0.025673940949935817,
      "grad_norm": 78.64288330078125,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 13.9361,
      "step": 20
    },
    {
      "epoch": 0.038510911424903725,
      "grad_norm": NaN,
      "learning_rate": 5.2e-06,
      "loss": 13.2456,
      "step": 30
    },
    {
      "epoch": 0.051347881899871634,
      "grad_norm": 84.45552825927734,
      "learning_rate": 7.2e-06,
      "loss": 12.3189,
      "step": 40
    },
    {
      "epoch": 0.06418485237483953,
      "grad_norm": 90.85704040527344,
      "learning_rate": 9e-06,
      "loss": 10.2159,
      "step": 50
    },
    {
      "epoch": 0.07702182284980745,
      "grad_norm": 90.75808715820312,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 8.5353,
      "step": 60
    },
    {
      "epoch": 0.08985879332477535,
      "grad_norm": 77.908447265625,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 5.7586,
      "step": 70
    },
    {
      "epoch": 0.10269576379974327,
      "grad_norm": 51.35749435424805,
      "learning_rate": 1.5e-05,
      "loss": 3.5377,
      "step": 80
    },
    {
      "epoch": 0.11553273427471117,
      "grad_norm": 22.19648551940918,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 2.0392,
      "step": 90
    },
    {
      "epoch": 0.12836970474967907,
      "grad_norm": 4.810085773468018,
      "learning_rate": 1.9e-05,
      "loss": 1.3851,
      "step": 100
    },
    {
      "epoch": 0.14120667522464697,
      "grad_norm": 3.834592342376709,
      "learning_rate": 2.1e-05,
      "loss": 0.9423,
      "step": 110
    },
    {
      "epoch": 0.1540436456996149,
      "grad_norm": 2.0345306396484375,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 0.5697,
      "step": 120
    },
    {
      "epoch": 0.1668806161745828,
      "grad_norm": 0.8915618658065796,
      "learning_rate": 2.5e-05,
      "loss": 0.3697,
      "step": 130
    },
    {
      "epoch": 0.1797175866495507,
      "grad_norm": 0.4180503189563751,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.2708,
      "step": 140
    },
    {
      "epoch": 0.1925545571245186,
      "grad_norm": 0.3554935157299042,
      "learning_rate": 2.9e-05,
      "loss": 0.2456,
      "step": 150
    },
    {
      "epoch": 0.20539152759948653,
      "grad_norm": 0.30976223945617676,
      "learning_rate": 3.1e-05,
      "loss": 0.2336,
      "step": 160
    },
    {
      "epoch": 0.21822849807445444,
      "grad_norm": 0.40389877557754517,
      "learning_rate": 3.3e-05,
      "loss": 0.2096,
      "step": 170
    },
    {
      "epoch": 0.23106546854942234,
      "grad_norm": 0.2854978144168854,
      "learning_rate": 3.5e-05,
      "loss": 0.198,
      "step": 180
    },
    {
      "epoch": 0.24390243902439024,
      "grad_norm": 0.22607630491256714,
      "learning_rate": 3.7e-05,
      "loss": 0.1828,
      "step": 190
    },
    {
      "epoch": 0.25673940949935814,
      "grad_norm": 0.3487206697463989,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 0.1848,
      "step": 200
    },
    {
      "epoch": 0.26957637997432604,
      "grad_norm": 0.27090469002723694,
      "learning_rate": 4.1e-05,
      "loss": 0.171,
      "step": 210
    },
    {
      "epoch": 0.28241335044929394,
      "grad_norm": 0.2992759644985199,
      "learning_rate": 4.3e-05,
      "loss": 0.1575,
      "step": 220
    },
    {
      "epoch": 0.2952503209242619,
      "grad_norm": 0.25752368569374084,
      "learning_rate": 4.5e-05,
      "loss": 0.1585,
      "step": 230
    },
    {
      "epoch": 0.3080872913992298,
      "grad_norm": 0.1718735545873642,
      "learning_rate": 4.7e-05,
      "loss": 0.1648,
      "step": 240
    },
    {
      "epoch": 0.3209242618741977,
      "grad_norm": 0.23267413675785065,
      "learning_rate": 4.9e-05,
      "loss": 0.1536,
      "step": 250
    },
    {
      "epoch": 0.3337612323491656,
      "grad_norm": 0.2425757646560669,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 0.1499,
      "step": 260
    },
    {
      "epoch": 0.3465982028241335,
      "grad_norm": 0.30258190631866455,
      "learning_rate": 5.300000000000001e-05,
      "loss": 0.1425,
      "step": 270
    },
    {
      "epoch": 0.3594351732991014,
      "grad_norm": 0.21171259880065918,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.1561,
      "step": 280
    },
    {
      "epoch": 0.3722721437740693,
      "grad_norm": 0.15556274354457855,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 0.1482,
      "step": 290
    },
    {
      "epoch": 0.3851091142490372,
      "grad_norm": 0.13324399292469025,
      "learning_rate": 5.9e-05,
      "loss": 0.1384,
      "step": 300
    },
    {
      "epoch": 0.3979460847240051,
      "grad_norm": 0.2825668156147003,
      "learning_rate": 6.1e-05,
      "loss": 0.1484,
      "step": 310
    },
    {
      "epoch": 0.41078305519897307,
      "grad_norm": 0.20607219636440277,
      "learning_rate": 6.3e-05,
      "loss": 0.1528,
      "step": 320
    },
    {
      "epoch": 0.42362002567394097,
      "grad_norm": 0.17047886550426483,
      "learning_rate": 6.500000000000001e-05,
      "loss": 0.1354,
      "step": 330
    },
    {
      "epoch": 0.43645699614890887,
      "grad_norm": 0.15508385002613068,
      "learning_rate": 6.7e-05,
      "loss": 0.1432,
      "step": 340
    },
    {
      "epoch": 0.4492939666238768,
      "grad_norm": 0.14523576200008392,
      "learning_rate": 6.9e-05,
      "loss": 0.1343,
      "step": 350
    },
    {
      "epoch": 0.4621309370988447,
      "grad_norm": 0.14676764607429504,
      "learning_rate": 7.1e-05,
      "loss": 0.1414,
      "step": 360
    },
    {
      "epoch": 0.4749679075738126,
      "grad_norm": 0.166632741689682,
      "learning_rate": 7.3e-05,
      "loss": 0.1315,
      "step": 370
    },
    {
      "epoch": 0.4878048780487805,
      "grad_norm": 0.13161858916282654,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.1305,
      "step": 380
    },
    {
      "epoch": 0.5006418485237484,
      "grad_norm": 0.19676858186721802,
      "learning_rate": 7.7e-05,
      "loss": 0.1356,
      "step": 390
    },
    {
      "epoch": 0.5134788189987163,
      "grad_norm": 0.12109082192182541,
      "learning_rate": 7.900000000000001e-05,
      "loss": 0.133,
      "step": 400
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 0.16240255534648895,
      "learning_rate": 8.1e-05,
      "loss": 0.141,
      "step": 410
    },
    {
      "epoch": 0.5391527599486521,
      "grad_norm": 0.16620230674743652,
      "learning_rate": 8.3e-05,
      "loss": 0.1308,
      "step": 420
    },
    {
      "epoch": 0.55198973042362,
      "grad_norm": 0.14271290600299835,
      "learning_rate": 8.5e-05,
      "loss": 0.1258,
      "step": 430
    },
    {
      "epoch": 0.5648267008985879,
      "grad_norm": 0.1443270742893219,
      "learning_rate": 8.7e-05,
      "loss": 0.1312,
      "step": 440
    },
    {
      "epoch": 0.5776636713735558,
      "grad_norm": 0.15797071158885956,
      "learning_rate": 8.900000000000001e-05,
      "loss": 0.1286,
      "step": 450
    },
    {
      "epoch": 0.5905006418485238,
      "grad_norm": 0.19477838277816772,
      "learning_rate": 9.1e-05,
      "loss": 0.1419,
      "step": 460
    },
    {
      "epoch": 0.6033376123234917,
      "grad_norm": 0.1290614753961563,
      "learning_rate": 9.300000000000001e-05,
      "loss": 0.129,
      "step": 470
    },
    {
      "epoch": 0.6161745827984596,
      "grad_norm": 0.12268809229135513,
      "learning_rate": 9.5e-05,
      "loss": 0.1232,
      "step": 480
    },
    {
      "epoch": 0.6290115532734275,
      "grad_norm": 0.1401013731956482,
      "learning_rate": 9.7e-05,
      "loss": 0.1229,
      "step": 490
    },
    {
      "epoch": 0.6418485237483954,
      "grad_norm": 0.11726848036050797,
      "learning_rate": 9.900000000000001e-05,
      "loss": 0.1178,
      "step": 500
    },
    {
      "epoch": 0.6418485237483954,
      "eval_gen_len": 38.946,
      "eval_loss": 0.10930793732404709,
      "eval_rouge1": 0.5574,
      "eval_rouge2": 0.2575,
      "eval_rougeL": 0.5149,
      "eval_runtime": 45.7535,
      "eval_samples_per_second": 10.928,
      "eval_steps_per_second": 1.377,
      "step": 500
    },
    {
      "epoch": 0.6546854942233633,
      "grad_norm": 0.13438373804092407,
      "learning_rate": 9.985272459499264e-05,
      "loss": 0.1331,
      "step": 510
    },
    {
      "epoch": 0.6675224646983312,
      "grad_norm": 0.12564225494861603,
      "learning_rate": 9.955817378497792e-05,
      "loss": 0.1257,
      "step": 520
    },
    {
      "epoch": 0.6803594351732991,
      "grad_norm": 0.11165695637464523,
      "learning_rate": 9.926362297496319e-05,
      "loss": 0.1255,
      "step": 530
    },
    {
      "epoch": 0.693196405648267,
      "grad_norm": 0.16142414510250092,
      "learning_rate": 9.896907216494846e-05,
      "loss": 0.1384,
      "step": 540
    },
    {
      "epoch": 0.7060333761232349,
      "grad_norm": 0.15439870953559875,
      "learning_rate": 9.867452135493374e-05,
      "loss": 0.1188,
      "step": 550
    },
    {
      "epoch": 0.7188703465982028,
      "grad_norm": 0.1158837303519249,
      "learning_rate": 9.8379970544919e-05,
      "loss": 0.1256,
      "step": 560
    },
    {
      "epoch": 0.7317073170731707,
      "grad_norm": 0.1260393112897873,
      "learning_rate": 9.808541973490427e-05,
      "loss": 0.1327,
      "step": 570
    },
    {
      "epoch": 0.7445442875481386,
      "grad_norm": 0.12582792341709137,
      "learning_rate": 9.779086892488954e-05,
      "loss": 0.1244,
      "step": 580
    },
    {
      "epoch": 0.7573812580231065,
      "grad_norm": 0.967321515083313,
      "learning_rate": 9.749631811487482e-05,
      "loss": 0.1283,
      "step": 590
    },
    {
      "epoch": 0.7702182284980744,
      "grad_norm": 0.14996884763240814,
      "learning_rate": 9.720176730486009e-05,
      "loss": 0.1378,
      "step": 600
    },
    {
      "epoch": 0.7830551989730423,
      "grad_norm": 0.13016603887081146,
      "learning_rate": 9.690721649484537e-05,
      "loss": 0.1249,
      "step": 610
    },
    {
      "epoch": 0.7958921694480102,
      "grad_norm": 0.11623676121234894,
      "learning_rate": 9.661266568483064e-05,
      "loss": 0.135,
      "step": 620
    },
    {
      "epoch": 0.8087291399229781,
      "grad_norm": 0.14658813178539276,
      "learning_rate": 9.631811487481591e-05,
      "loss": 0.1176,
      "step": 630
    },
    {
      "epoch": 0.8215661103979461,
      "grad_norm": 0.13376665115356445,
      "learning_rate": 9.602356406480119e-05,
      "loss": 0.1208,
      "step": 640
    },
    {
      "epoch": 0.834403080872914,
      "grad_norm": 0.13538646697998047,
      "learning_rate": 9.572901325478646e-05,
      "loss": 0.1192,
      "step": 650
    },
    {
      "epoch": 0.8472400513478819,
      "grad_norm": 0.13593578338623047,
      "learning_rate": 9.543446244477173e-05,
      "loss": 0.1227,
      "step": 660
    },
    {
      "epoch": 0.8600770218228498,
      "grad_norm": 0.13085196912288666,
      "learning_rate": 9.5139911634757e-05,
      "loss": 0.1275,
      "step": 670
    },
    {
      "epoch": 0.8729139922978177,
      "grad_norm": 0.1571151316165924,
      "learning_rate": 9.484536082474227e-05,
      "loss": 0.1274,
      "step": 680
    },
    {
      "epoch": 0.8857509627727856,
      "grad_norm": 0.14053873717784882,
      "learning_rate": 9.455081001472754e-05,
      "loss": 0.1205,
      "step": 690
    },
    {
      "epoch": 0.8985879332477535,
      "grad_norm": 0.11424735933542252,
      "learning_rate": 9.425625920471282e-05,
      "loss": 0.1182,
      "step": 700
    },
    {
      "epoch": 0.9114249037227214,
      "grad_norm": 0.1393001526594162,
      "learning_rate": 9.396170839469809e-05,
      "loss": 0.1229,
      "step": 710
    },
    {
      "epoch": 0.9242618741976893,
      "grad_norm": 0.1388111263513565,
      "learning_rate": 9.366715758468336e-05,
      "loss": 0.1274,
      "step": 720
    },
    {
      "epoch": 0.9370988446726572,
      "grad_norm": 0.11912015080451965,
      "learning_rate": 9.337260677466864e-05,
      "loss": 0.118,
      "step": 730
    },
    {
      "epoch": 0.9499358151476252,
      "grad_norm": 0.15506206452846527,
      "learning_rate": 9.307805596465391e-05,
      "loss": 0.1258,
      "step": 740
    },
    {
      "epoch": 0.962772785622593,
      "grad_norm": 0.158128559589386,
      "learning_rate": 9.278350515463918e-05,
      "loss": 0.1229,
      "step": 750
    },
    {
      "epoch": 0.975609756097561,
      "grad_norm": 0.16017945110797882,
      "learning_rate": 9.248895434462446e-05,
      "loss": 0.1239,
      "step": 760
    },
    {
      "epoch": 0.9884467265725289,
      "grad_norm": 0.1304791122674942,
      "learning_rate": 9.219440353460973e-05,
      "loss": 0.1122,
      "step": 770
    },
    {
      "epoch": 1.0012836970474968,
      "grad_norm": 0.12392584979534149,
      "learning_rate": 9.189985272459499e-05,
      "loss": 0.127,
      "step": 780
    },
    {
      "epoch": 1.0141206675224648,
      "grad_norm": 0.10105091333389282,
      "learning_rate": 9.160530191458027e-05,
      "loss": 0.1155,
      "step": 790
    },
    {
      "epoch": 1.0269576379974326,
      "grad_norm": 0.10798871517181396,
      "learning_rate": 9.131075110456554e-05,
      "loss": 0.1263,
      "step": 800
    },
    {
      "epoch": 1.0397946084724006,
      "grad_norm": 0.1646156907081604,
      "learning_rate": 9.101620029455081e-05,
      "loss": 0.1135,
      "step": 810
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 0.13659092783927917,
      "learning_rate": 9.072164948453609e-05,
      "loss": 0.118,
      "step": 820
    },
    {
      "epoch": 1.0654685494223364,
      "grad_norm": 0.1449888050556183,
      "learning_rate": 9.042709867452136e-05,
      "loss": 0.1252,
      "step": 830
    },
    {
      "epoch": 1.0783055198973042,
      "grad_norm": 0.14084464311599731,
      "learning_rate": 9.013254786450663e-05,
      "loss": 0.1254,
      "step": 840
    },
    {
      "epoch": 1.0911424903722722,
      "grad_norm": 0.14235644042491913,
      "learning_rate": 8.983799705449191e-05,
      "loss": 0.1169,
      "step": 850
    },
    {
      "epoch": 1.10397946084724,
      "grad_norm": 0.13668040931224823,
      "learning_rate": 8.954344624447718e-05,
      "loss": 0.1199,
      "step": 860
    },
    {
      "epoch": 1.116816431322208,
      "grad_norm": 0.20514467358589172,
      "learning_rate": 8.924889543446246e-05,
      "loss": 0.118,
      "step": 870
    },
    {
      "epoch": 1.1296534017971758,
      "grad_norm": 0.13541142642498016,
      "learning_rate": 8.895434462444772e-05,
      "loss": 0.129,
      "step": 880
    },
    {
      "epoch": 1.1424903722721438,
      "grad_norm": 0.11607542634010315,
      "learning_rate": 8.865979381443299e-05,
      "loss": 0.1226,
      "step": 890
    },
    {
      "epoch": 1.1553273427471118,
      "grad_norm": 0.134485125541687,
      "learning_rate": 8.836524300441826e-05,
      "loss": 0.1166,
      "step": 900
    },
    {
      "epoch": 1.1681643132220796,
      "grad_norm": 0.12031760811805725,
      "learning_rate": 8.807069219440354e-05,
      "loss": 0.1118,
      "step": 910
    },
    {
      "epoch": 1.1810012836970474,
      "grad_norm": 0.14312677085399628,
      "learning_rate": 8.777614138438881e-05,
      "loss": 0.1271,
      "step": 920
    },
    {
      "epoch": 1.1938382541720154,
      "grad_norm": 0.1250048577785492,
      "learning_rate": 8.748159057437408e-05,
      "loss": 0.1116,
      "step": 930
    },
    {
      "epoch": 1.2066752246469834,
      "grad_norm": 0.12448818981647491,
      "learning_rate": 8.718703976435936e-05,
      "loss": 0.12,
      "step": 940
    },
    {
      "epoch": 1.2195121951219512,
      "grad_norm": 0.1207614317536354,
      "learning_rate": 8.689248895434463e-05,
      "loss": 0.1142,
      "step": 950
    },
    {
      "epoch": 1.2323491655969192,
      "grad_norm": 0.12647990882396698,
      "learning_rate": 8.65979381443299e-05,
      "loss": 0.1151,
      "step": 960
    },
    {
      "epoch": 1.245186136071887,
      "grad_norm": 0.12347666174173355,
      "learning_rate": 8.630338733431518e-05,
      "loss": 0.1217,
      "step": 970
    },
    {
      "epoch": 1.258023106546855,
      "grad_norm": 0.16950395703315735,
      "learning_rate": 8.600883652430045e-05,
      "loss": 0.1218,
      "step": 980
    },
    {
      "epoch": 1.2708600770218228,
      "grad_norm": 0.12122233211994171,
      "learning_rate": 8.571428571428571e-05,
      "loss": 0.1139,
      "step": 990
    },
    {
      "epoch": 1.2836970474967908,
      "grad_norm": 0.1224595233798027,
      "learning_rate": 8.541973490427099e-05,
      "loss": 0.1162,
      "step": 1000
    },
    {
      "epoch": 1.2836970474967908,
      "eval_gen_len": 38.946,
      "eval_loss": 0.10218328982591629,
      "eval_rouge1": 0.5746,
      "eval_rouge2": 0.2819,
      "eval_rougeL": 0.5376,
      "eval_runtime": 45.0734,
      "eval_samples_per_second": 11.093,
      "eval_steps_per_second": 1.398,
      "step": 1000
    },
    {
      "epoch": 1.2965340179717586,
      "grad_norm": 0.0963892713189125,
      "learning_rate": 8.512518409425626e-05,
      "loss": 0.1222,
      "step": 1010
    },
    {
      "epoch": 1.3093709884467266,
      "grad_norm": 0.1362753063440323,
      "learning_rate": 8.483063328424153e-05,
      "loss": 0.1145,
      "step": 1020
    },
    {
      "epoch": 1.3222079589216944,
      "grad_norm": 0.09700595587491989,
      "learning_rate": 8.453608247422681e-05,
      "loss": 0.1185,
      "step": 1030
    },
    {
      "epoch": 1.3350449293966624,
      "grad_norm": 0.13865801692008972,
      "learning_rate": 8.424153166421208e-05,
      "loss": 0.1228,
      "step": 1040
    },
    {
      "epoch": 1.3478818998716302,
      "grad_norm": 0.11458808183670044,
      "learning_rate": 8.394698085419735e-05,
      "loss": 0.1092,
      "step": 1050
    },
    {
      "epoch": 1.3607188703465982,
      "grad_norm": 0.12909230589866638,
      "learning_rate": 8.365243004418263e-05,
      "loss": 0.1153,
      "step": 1060
    },
    {
      "epoch": 1.3735558408215662,
      "grad_norm": 0.13270749151706696,
      "learning_rate": 8.33578792341679e-05,
      "loss": 0.1207,
      "step": 1070
    },
    {
      "epoch": 1.386392811296534,
      "grad_norm": 0.1388499140739441,
      "learning_rate": 8.306332842415318e-05,
      "loss": 0.1271,
      "step": 1080
    },
    {
      "epoch": 1.3992297817715018,
      "grad_norm": 0.11819977313280106,
      "learning_rate": 8.276877761413844e-05,
      "loss": 0.1221,
      "step": 1090
    },
    {
      "epoch": 1.4120667522464698,
      "grad_norm": 0.13087821006774902,
      "learning_rate": 8.247422680412371e-05,
      "loss": 0.1092,
      "step": 1100
    },
    {
      "epoch": 1.4249037227214378,
      "grad_norm": 0.14269869029521942,
      "learning_rate": 8.217967599410898e-05,
      "loss": 0.1323,
      "step": 1110
    },
    {
      "epoch": 1.4377406931964056,
      "grad_norm": 0.11728661507368088,
      "learning_rate": 8.188512518409426e-05,
      "loss": 0.116,
      "step": 1120
    },
    {
      "epoch": 1.4505776636713734,
      "grad_norm": 0.12322080880403519,
      "learning_rate": 8.159057437407953e-05,
      "loss": 0.1162,
      "step": 1130
    },
    {
      "epoch": 1.4634146341463414,
      "grad_norm": 0.14137819409370422,
      "learning_rate": 8.12960235640648e-05,
      "loss": 0.1166,
      "step": 1140
    },
    {
      "epoch": 1.4762516046213094,
      "grad_norm": 0.11483679711818695,
      "learning_rate": 8.100147275405008e-05,
      "loss": 0.113,
      "step": 1150
    },
    {
      "epoch": 1.4890885750962772,
      "grad_norm": 0.107373908162117,
      "learning_rate": 8.070692194403535e-05,
      "loss": 0.1086,
      "step": 1160
    },
    {
      "epoch": 1.501925545571245,
      "grad_norm": 0.1413784623146057,
      "learning_rate": 8.041237113402063e-05,
      "loss": 0.1127,
      "step": 1170
    },
    {
      "epoch": 1.514762516046213,
      "grad_norm": 0.15129907429218292,
      "learning_rate": 8.01178203240059e-05,
      "loss": 0.1094,
      "step": 1180
    },
    {
      "epoch": 1.527599486521181,
      "grad_norm": 0.11779375374317169,
      "learning_rate": 7.982326951399117e-05,
      "loss": 0.1113,
      "step": 1190
    },
    {
      "epoch": 1.540436456996149,
      "grad_norm": 0.11644338816404343,
      "learning_rate": 7.952871870397643e-05,
      "loss": 0.1242,
      "step": 1200
    },
    {
      "epoch": 1.5532734274711169,
      "grad_norm": 0.14850880205631256,
      "learning_rate": 7.923416789396171e-05,
      "loss": 0.1192,
      "step": 1210
    },
    {
      "epoch": 1.5661103979460846,
      "grad_norm": 0.11069446802139282,
      "learning_rate": 7.893961708394698e-05,
      "loss": 0.1043,
      "step": 1220
    },
    {
      "epoch": 1.5789473684210527,
      "grad_norm": 0.0926310271024704,
      "learning_rate": 7.864506627393225e-05,
      "loss": 0.1218,
      "step": 1230
    },
    {
      "epoch": 1.5917843388960207,
      "grad_norm": 0.1573435515165329,
      "learning_rate": 7.835051546391753e-05,
      "loss": 0.1181,
      "step": 1240
    },
    {
      "epoch": 1.6046213093709885,
      "grad_norm": 0.11641980707645416,
      "learning_rate": 7.80559646539028e-05,
      "loss": 0.1206,
      "step": 1250
    },
    {
      "epoch": 1.6174582798459562,
      "grad_norm": 0.119265116751194,
      "learning_rate": 7.776141384388808e-05,
      "loss": 0.1148,
      "step": 1260
    },
    {
      "epoch": 1.6302952503209243,
      "grad_norm": 0.12308722734451294,
      "learning_rate": 7.746686303387335e-05,
      "loss": 0.1132,
      "step": 1270
    },
    {
      "epoch": 1.6431322207958923,
      "grad_norm": 0.12015580385923386,
      "learning_rate": 7.717231222385862e-05,
      "loss": 0.113,
      "step": 1280
    },
    {
      "epoch": 1.65596919127086,
      "grad_norm": 0.09488837420940399,
      "learning_rate": 7.68777614138439e-05,
      "loss": 0.112,
      "step": 1290
    },
    {
      "epoch": 1.6688061617458279,
      "grad_norm": 0.11132404953241348,
      "learning_rate": 7.658321060382917e-05,
      "loss": 0.1085,
      "step": 1300
    },
    {
      "epoch": 1.6816431322207959,
      "grad_norm": 0.15267649292945862,
      "learning_rate": 7.628865979381443e-05,
      "loss": 0.1136,
      "step": 1310
    },
    {
      "epoch": 1.6944801026957639,
      "grad_norm": 0.13199123740196228,
      "learning_rate": 7.59941089837997e-05,
      "loss": 0.1145,
      "step": 1320
    },
    {
      "epoch": 1.7073170731707317,
      "grad_norm": 0.11657251417636871,
      "learning_rate": 7.569955817378498e-05,
      "loss": 0.111,
      "step": 1330
    },
    {
      "epoch": 1.7201540436456995,
      "grad_norm": 0.15839266777038574,
      "learning_rate": 7.540500736377025e-05,
      "loss": 0.1166,
      "step": 1340
    },
    {
      "epoch": 1.7329910141206675,
      "grad_norm": 0.10701854526996613,
      "learning_rate": 7.511045655375553e-05,
      "loss": 0.1127,
      "step": 1350
    },
    {
      "epoch": 1.7458279845956355,
      "grad_norm": 0.16936787962913513,
      "learning_rate": 7.48159057437408e-05,
      "loss": 0.113,
      "step": 1360
    },
    {
      "epoch": 1.7586649550706035,
      "grad_norm": 0.10048405826091766,
      "learning_rate": 7.452135493372607e-05,
      "loss": 0.1139,
      "step": 1370
    },
    {
      "epoch": 1.7715019255455713,
      "grad_norm": 0.11125636100769043,
      "learning_rate": 7.422680412371135e-05,
      "loss": 0.1156,
      "step": 1380
    },
    {
      "epoch": 1.784338896020539,
      "grad_norm": 0.10491804033517838,
      "learning_rate": 7.393225331369662e-05,
      "loss": 0.1227,
      "step": 1390
    },
    {
      "epoch": 1.797175866495507,
      "grad_norm": 0.1252523511648178,
      "learning_rate": 7.36377025036819e-05,
      "loss": 0.1258,
      "step": 1400
    },
    {
      "epoch": 1.810012836970475,
      "grad_norm": 0.10493435710668564,
      "learning_rate": 7.334315169366715e-05,
      "loss": 0.1153,
      "step": 1410
    },
    {
      "epoch": 1.822849807445443,
      "grad_norm": 0.11997751146554947,
      "learning_rate": 7.304860088365243e-05,
      "loss": 0.1171,
      "step": 1420
    },
    {
      "epoch": 1.8356867779204107,
      "grad_norm": 0.10238035768270493,
      "learning_rate": 7.27540500736377e-05,
      "loss": 0.1144,
      "step": 1430
    },
    {
      "epoch": 1.8485237483953787,
      "grad_norm": 0.12434195727109909,
      "learning_rate": 7.245949926362298e-05,
      "loss": 0.1212,
      "step": 1440
    },
    {
      "epoch": 1.8613607188703467,
      "grad_norm": 0.09716463088989258,
      "learning_rate": 7.216494845360825e-05,
      "loss": 0.1161,
      "step": 1450
    },
    {
      "epoch": 1.8741976893453145,
      "grad_norm": 0.18025720119476318,
      "learning_rate": 7.187039764359352e-05,
      "loss": 0.1191,
      "step": 1460
    },
    {
      "epoch": 1.8870346598202823,
      "grad_norm": 0.10300459712743759,
      "learning_rate": 7.15758468335788e-05,
      "loss": 0.1161,
      "step": 1470
    },
    {
      "epoch": 1.8998716302952503,
      "grad_norm": 0.11518652737140656,
      "learning_rate": 7.128129602356407e-05,
      "loss": 0.1062,
      "step": 1480
    },
    {
      "epoch": 1.9127086007702183,
      "grad_norm": 0.10772902518510818,
      "learning_rate": 7.098674521354934e-05,
      "loss": 0.114,
      "step": 1490
    },
    {
      "epoch": 1.925545571245186,
      "grad_norm": 0.1605733186006546,
      "learning_rate": 7.069219440353462e-05,
      "loss": 0.1106,
      "step": 1500
    },
    {
      "epoch": 1.925545571245186,
      "eval_gen_len": 38.946,
      "eval_loss": 0.09927912056446075,
      "eval_rouge1": 0.5823,
      "eval_rouge2": 0.2956,
      "eval_rougeL": 0.5489,
      "eval_runtime": 45.4027,
      "eval_samples_per_second": 11.013,
      "eval_steps_per_second": 1.388,
      "step": 1500
    },
    {
      "epoch": 1.938382541720154,
      "grad_norm": 0.12198202311992645,
      "learning_rate": 7.039764359351989e-05,
      "loss": 0.1183,
      "step": 1510
    },
    {
      "epoch": 1.951219512195122,
      "grad_norm": 0.1133396103978157,
      "learning_rate": 7.010309278350515e-05,
      "loss": 0.1188,
      "step": 1520
    },
    {
      "epoch": 1.96405648267009,
      "grad_norm": 0.10706740617752075,
      "learning_rate": 6.980854197349043e-05,
      "loss": 0.1161,
      "step": 1530
    },
    {
      "epoch": 1.976893453145058,
      "grad_norm": 0.10656236857175827,
      "learning_rate": 6.95139911634757e-05,
      "loss": 0.1113,
      "step": 1540
    },
    {
      "epoch": 1.9897304236200257,
      "grad_norm": 0.1278591752052307,
      "learning_rate": 6.921944035346097e-05,
      "loss": 0.1262,
      "step": 1550
    },
    {
      "epoch": 2.0025673940949935,
      "grad_norm": 0.11729343235492706,
      "learning_rate": 6.892488954344625e-05,
      "loss": 0.1267,
      "step": 1560
    },
    {
      "epoch": 2.0154043645699615,
      "grad_norm": 0.10659424960613251,
      "learning_rate": 6.863033873343152e-05,
      "loss": 0.0972,
      "step": 1570
    },
    {
      "epoch": 2.0282413350449295,
      "grad_norm": 0.13483816385269165,
      "learning_rate": 6.83357879234168e-05,
      "loss": 0.1075,
      "step": 1580
    },
    {
      "epoch": 2.041078305519897,
      "grad_norm": 0.11035773903131485,
      "learning_rate": 6.804123711340207e-05,
      "loss": 0.1041,
      "step": 1590
    },
    {
      "epoch": 2.053915275994865,
      "grad_norm": 0.11331629753112793,
      "learning_rate": 6.774668630338734e-05,
      "loss": 0.1107,
      "step": 1600
    },
    {
      "epoch": 2.066752246469833,
      "grad_norm": 0.10604571551084518,
      "learning_rate": 6.745213549337262e-05,
      "loss": 0.113,
      "step": 1610
    },
    {
      "epoch": 2.079589216944801,
      "grad_norm": 0.1317732185125351,
      "learning_rate": 6.715758468335788e-05,
      "loss": 0.1205,
      "step": 1620
    },
    {
      "epoch": 2.092426187419769,
      "grad_norm": 0.1633516103029251,
      "learning_rate": 6.686303387334315e-05,
      "loss": 0.1213,
      "step": 1630
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 0.1562272608280182,
      "learning_rate": 6.656848306332842e-05,
      "loss": 0.1082,
      "step": 1640
    },
    {
      "epoch": 2.1181001283697047,
      "grad_norm": 0.17369812726974487,
      "learning_rate": 6.62739322533137e-05,
      "loss": 0.1042,
      "step": 1650
    },
    {
      "epoch": 2.1309370988446728,
      "grad_norm": 0.12046299129724503,
      "learning_rate": 6.597938144329897e-05,
      "loss": 0.1097,
      "step": 1660
    },
    {
      "epoch": 2.1437740693196408,
      "grad_norm": 0.11788310110569,
      "learning_rate": 6.568483063328424e-05,
      "loss": 0.1161,
      "step": 1670
    },
    {
      "epoch": 2.1566110397946083,
      "grad_norm": 0.11287978291511536,
      "learning_rate": 6.539027982326952e-05,
      "loss": 0.1196,
      "step": 1680
    },
    {
      "epoch": 2.1694480102695763,
      "grad_norm": 0.1154395118355751,
      "learning_rate": 6.509572901325479e-05,
      "loss": 0.1145,
      "step": 1690
    },
    {
      "epoch": 2.1822849807445444,
      "grad_norm": 0.10837022215127945,
      "learning_rate": 6.480117820324007e-05,
      "loss": 0.1167,
      "step": 1700
    },
    {
      "epoch": 2.1951219512195124,
      "grad_norm": 0.11331544071435928,
      "learning_rate": 6.450662739322534e-05,
      "loss": 0.1115,
      "step": 1710
    },
    {
      "epoch": 2.20795892169448,
      "grad_norm": 0.11895423382520676,
      "learning_rate": 6.421207658321061e-05,
      "loss": 0.1102,
      "step": 1720
    },
    {
      "epoch": 2.220795892169448,
      "grad_norm": 0.14609269797801971,
      "learning_rate": 6.391752577319587e-05,
      "loss": 0.1133,
      "step": 1730
    },
    {
      "epoch": 2.233632862644416,
      "grad_norm": 0.13491936028003693,
      "learning_rate": 6.362297496318115e-05,
      "loss": 0.1072,
      "step": 1740
    },
    {
      "epoch": 2.246469833119384,
      "grad_norm": 0.13540756702423096,
      "learning_rate": 6.332842415316642e-05,
      "loss": 0.114,
      "step": 1750
    },
    {
      "epoch": 2.2593068035943515,
      "grad_norm": 0.15043611824512482,
      "learning_rate": 6.30338733431517e-05,
      "loss": 0.1132,
      "step": 1760
    },
    {
      "epoch": 2.2721437740693196,
      "grad_norm": 0.14203940331935883,
      "learning_rate": 6.273932253313697e-05,
      "loss": 0.1142,
      "step": 1770
    },
    {
      "epoch": 2.2849807445442876,
      "grad_norm": 0.11540842801332474,
      "learning_rate": 6.244477172312224e-05,
      "loss": 0.1055,
      "step": 1780
    },
    {
      "epoch": 2.2978177150192556,
      "grad_norm": 0.1832491159439087,
      "learning_rate": 6.215022091310752e-05,
      "loss": 0.1177,
      "step": 1790
    },
    {
      "epoch": 2.3106546854942236,
      "grad_norm": 0.11431492120027542,
      "learning_rate": 6.185567010309279e-05,
      "loss": 0.1058,
      "step": 1800
    },
    {
      "epoch": 2.323491655969191,
      "grad_norm": 0.12939855456352234,
      "learning_rate": 6.156111929307806e-05,
      "loss": 0.1122,
      "step": 1810
    },
    {
      "epoch": 2.336328626444159,
      "grad_norm": 0.12890468537807465,
      "learning_rate": 6.126656848306334e-05,
      "loss": 0.1127,
      "step": 1820
    },
    {
      "epoch": 2.349165596919127,
      "grad_norm": 0.11087578535079956,
      "learning_rate": 6.097201767304861e-05,
      "loss": 0.1106,
      "step": 1830
    },
    {
      "epoch": 2.3620025673940948,
      "grad_norm": 0.12481900304555893,
      "learning_rate": 6.067746686303387e-05,
      "loss": 0.1155,
      "step": 1840
    },
    {
      "epoch": 2.3748395378690628,
      "grad_norm": 0.11346503347158432,
      "learning_rate": 6.0382916053019144e-05,
      "loss": 0.1095,
      "step": 1850
    },
    {
      "epoch": 2.387676508344031,
      "grad_norm": 0.13271845877170563,
      "learning_rate": 6.008836524300442e-05,
      "loss": 0.1035,
      "step": 1860
    },
    {
      "epoch": 2.400513478818999,
      "grad_norm": 0.12727977335453033,
      "learning_rate": 5.979381443298969e-05,
      "loss": 0.1163,
      "step": 1870
    },
    {
      "epoch": 2.413350449293967,
      "grad_norm": 0.11618130654096603,
      "learning_rate": 5.9499263622974965e-05,
      "loss": 0.1127,
      "step": 1880
    },
    {
      "epoch": 2.4261874197689344,
      "grad_norm": 0.12864802777767181,
      "learning_rate": 5.920471281296024e-05,
      "loss": 0.1137,
      "step": 1890
    },
    {
      "epoch": 2.4390243902439024,
      "grad_norm": 0.10020538419485092,
      "learning_rate": 5.891016200294551e-05,
      "loss": 0.1054,
      "step": 1900
    },
    {
      "epoch": 2.4518613607188704,
      "grad_norm": 0.25161072611808777,
      "learning_rate": 5.861561119293079e-05,
      "loss": 0.107,
      "step": 1910
    },
    {
      "epoch": 2.4646983311938384,
      "grad_norm": 0.11326297372579575,
      "learning_rate": 5.832106038291606e-05,
      "loss": 0.1132,
      "step": 1920
    },
    {
      "epoch": 2.477535301668806,
      "grad_norm": 0.28271356225013733,
      "learning_rate": 5.8026509572901334e-05,
      "loss": 0.1121,
      "step": 1930
    },
    {
      "epoch": 2.490372272143774,
      "grad_norm": 0.10053003579378128,
      "learning_rate": 5.7731958762886594e-05,
      "loss": 0.1092,
      "step": 1940
    },
    {
      "epoch": 2.503209242618742,
      "grad_norm": 0.1034500002861023,
      "learning_rate": 5.743740795287187e-05,
      "loss": 0.1111,
      "step": 1950
    },
    {
      "epoch": 2.51604621309371,
      "grad_norm": 0.12638093531131744,
      "learning_rate": 5.714285714285714e-05,
      "loss": 0.1122,
      "step": 1960
    },
    {
      "epoch": 2.528883183568678,
      "grad_norm": 0.12756876647472382,
      "learning_rate": 5.6848306332842415e-05,
      "loss": 0.1053,
      "step": 1970
    },
    {
      "epoch": 2.5417201540436456,
      "grad_norm": 0.14028741419315338,
      "learning_rate": 5.655375552282769e-05,
      "loss": 0.1143,
      "step": 1980
    },
    {
      "epoch": 2.5545571245186136,
      "grad_norm": 0.1252150535583496,
      "learning_rate": 5.625920471281296e-05,
      "loss": 0.1172,
      "step": 1990
    },
    {
      "epoch": 2.5673940949935816,
      "grad_norm": 0.09696774929761887,
      "learning_rate": 5.596465390279824e-05,
      "loss": 0.1047,
      "step": 2000
    },
    {
      "epoch": 2.5673940949935816,
      "eval_gen_len": 38.946,
      "eval_loss": 0.09703746438026428,
      "eval_rouge1": 0.5894,
      "eval_rouge2": 0.3049,
      "eval_rougeL": 0.5547,
      "eval_runtime": 45.1347,
      "eval_samples_per_second": 11.078,
      "eval_steps_per_second": 1.396,
      "step": 2000
    },
    {
      "epoch": 2.580231065468549,
      "grad_norm": 0.13524310290813446,
      "learning_rate": 5.567010309278351e-05,
      "loss": 0.1112,
      "step": 2010
    },
    {
      "epoch": 2.593068035943517,
      "grad_norm": 0.10453896969556808,
      "learning_rate": 5.5375552282768784e-05,
      "loss": 0.1073,
      "step": 2020
    },
    {
      "epoch": 2.605905006418485,
      "grad_norm": 0.1299864947795868,
      "learning_rate": 5.508100147275406e-05,
      "loss": 0.1067,
      "step": 2030
    },
    {
      "epoch": 2.6187419768934532,
      "grad_norm": 0.11628852039575577,
      "learning_rate": 5.478645066273933e-05,
      "loss": 0.1072,
      "step": 2040
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 0.12102477252483368,
      "learning_rate": 5.449189985272459e-05,
      "loss": 0.1133,
      "step": 2050
    },
    {
      "epoch": 2.644415917843389,
      "grad_norm": 0.08944839239120483,
      "learning_rate": 5.4197349042709865e-05,
      "loss": 0.1058,
      "step": 2060
    },
    {
      "epoch": 2.657252888318357,
      "grad_norm": 0.11976416409015656,
      "learning_rate": 5.390279823269514e-05,
      "loss": 0.115,
      "step": 2070
    },
    {
      "epoch": 2.670089858793325,
      "grad_norm": 0.11056056618690491,
      "learning_rate": 5.360824742268041e-05,
      "loss": 0.1107,
      "step": 2080
    },
    {
      "epoch": 2.682926829268293,
      "grad_norm": 0.12344369292259216,
      "learning_rate": 5.3313696612665687e-05,
      "loss": 0.1255,
      "step": 2090
    },
    {
      "epoch": 2.6957637997432604,
      "grad_norm": 0.10227925330400467,
      "learning_rate": 5.301914580265096e-05,
      "loss": 0.1179,
      "step": 2100
    },
    {
      "epoch": 2.7086007702182284,
      "grad_norm": 0.09863247722387314,
      "learning_rate": 5.2724594992636234e-05,
      "loss": 0.1034,
      "step": 2110
    },
    {
      "epoch": 2.7214377406931964,
      "grad_norm": 0.12329726666212082,
      "learning_rate": 5.243004418262151e-05,
      "loss": 0.109,
      "step": 2120
    },
    {
      "epoch": 2.7342747111681645,
      "grad_norm": 0.128681480884552,
      "learning_rate": 5.213549337260678e-05,
      "loss": 0.111,
      "step": 2130
    },
    {
      "epoch": 2.7471116816431325,
      "grad_norm": 0.1252344846725464,
      "learning_rate": 5.1840942562592055e-05,
      "loss": 0.1052,
      "step": 2140
    },
    {
      "epoch": 2.7599486521181,
      "grad_norm": 0.11999202519655228,
      "learning_rate": 5.1546391752577315e-05,
      "loss": 0.1056,
      "step": 2150
    },
    {
      "epoch": 2.772785622593068,
      "grad_norm": 0.12448093295097351,
      "learning_rate": 5.125184094256259e-05,
      "loss": 0.1279,
      "step": 2160
    },
    {
      "epoch": 2.785622593068036,
      "grad_norm": 0.32093188166618347,
      "learning_rate": 5.095729013254786e-05,
      "loss": 0.1018,
      "step": 2170
    },
    {
      "epoch": 2.7984595635430036,
      "grad_norm": 0.10641879588365555,
      "learning_rate": 5.0662739322533137e-05,
      "loss": 0.11,
      "step": 2180
    },
    {
      "epoch": 2.8112965340179716,
      "grad_norm": 0.12216117233037949,
      "learning_rate": 5.036818851251841e-05,
      "loss": 0.1099,
      "step": 2190
    },
    {
      "epoch": 2.8241335044929397,
      "grad_norm": 0.1177031397819519,
      "learning_rate": 5.0073637702503684e-05,
      "loss": 0.1156,
      "step": 2200
    },
    {
      "epoch": 2.8369704749679077,
      "grad_norm": 0.21809132397174835,
      "learning_rate": 4.977908689248896e-05,
      "loss": 0.1033,
      "step": 2210
    },
    {
      "epoch": 2.8498074454428757,
      "grad_norm": 0.16358229517936707,
      "learning_rate": 4.948453608247423e-05,
      "loss": 0.1046,
      "step": 2220
    },
    {
      "epoch": 2.8626444159178432,
      "grad_norm": 0.20664657652378082,
      "learning_rate": 4.91899852724595e-05,
      "loss": 0.1111,
      "step": 2230
    },
    {
      "epoch": 2.8754813863928113,
      "grad_norm": 0.1042567566037178,
      "learning_rate": 4.889543446244477e-05,
      "loss": 0.1122,
      "step": 2240
    },
    {
      "epoch": 2.8883183568677793,
      "grad_norm": 0.12193156033754349,
      "learning_rate": 4.8600883652430046e-05,
      "loss": 0.1024,
      "step": 2250
    },
    {
      "epoch": 2.901155327342747,
      "grad_norm": 0.12314718961715698,
      "learning_rate": 4.830633284241532e-05,
      "loss": 0.1114,
      "step": 2260
    },
    {
      "epoch": 2.913992297817715,
      "grad_norm": 0.12970231473445892,
      "learning_rate": 4.801178203240059e-05,
      "loss": 0.1062,
      "step": 2270
    },
    {
      "epoch": 2.926829268292683,
      "grad_norm": 0.11207956075668335,
      "learning_rate": 4.771723122238587e-05,
      "loss": 0.1135,
      "step": 2280
    },
    {
      "epoch": 2.939666238767651,
      "grad_norm": 0.10265418887138367,
      "learning_rate": 4.7422680412371134e-05,
      "loss": 0.1095,
      "step": 2290
    },
    {
      "epoch": 2.952503209242619,
      "grad_norm": 0.17097508907318115,
      "learning_rate": 4.712812960235641e-05,
      "loss": 0.1063,
      "step": 2300
    },
    {
      "epoch": 2.965340179717587,
      "grad_norm": 0.1322011947631836,
      "learning_rate": 4.683357879234168e-05,
      "loss": 0.111,
      "step": 2310
    },
    {
      "epoch": 2.9781771501925545,
      "grad_norm": 0.10939087718725204,
      "learning_rate": 4.6539027982326955e-05,
      "loss": 0.1177,
      "step": 2320
    },
    {
      "epoch": 2.9910141206675225,
      "grad_norm": 0.18089720606803894,
      "learning_rate": 4.624447717231223e-05,
      "loss": 0.1042,
      "step": 2330
    },
    {
      "epoch": 3.0038510911424905,
      "grad_norm": 0.1323990374803543,
      "learning_rate": 4.5949926362297496e-05,
      "loss": 0.1098,
      "step": 2340
    },
    {
      "epoch": 3.016688061617458,
      "grad_norm": 0.15342199802398682,
      "learning_rate": 4.565537555228277e-05,
      "loss": 0.0985,
      "step": 2350
    },
    {
      "epoch": 3.029525032092426,
      "grad_norm": 0.11682497709989548,
      "learning_rate": 4.536082474226804e-05,
      "loss": 0.1109,
      "step": 2360
    },
    {
      "epoch": 3.042362002567394,
      "grad_norm": 0.17593421041965485,
      "learning_rate": 4.506627393225332e-05,
      "loss": 0.1053,
      "step": 2370
    },
    {
      "epoch": 3.055198973042362,
      "grad_norm": 0.13586147129535675,
      "learning_rate": 4.477172312223859e-05,
      "loss": 0.1029,
      "step": 2380
    },
    {
      "epoch": 3.06803594351733,
      "grad_norm": 0.14202643930912018,
      "learning_rate": 4.447717231222386e-05,
      "loss": 0.1087,
      "step": 2390
    },
    {
      "epoch": 3.0808729139922977,
      "grad_norm": 0.10535432398319244,
      "learning_rate": 4.418262150220913e-05,
      "loss": 0.1199,
      "step": 2400
    },
    {
      "epoch": 3.0937098844672657,
      "grad_norm": 0.13306142389774323,
      "learning_rate": 4.3888070692194405e-05,
      "loss": 0.1031,
      "step": 2410
    },
    {
      "epoch": 3.1065468549422337,
      "grad_norm": 0.14567367732524872,
      "learning_rate": 4.359351988217968e-05,
      "loss": 0.1091,
      "step": 2420
    },
    {
      "epoch": 3.1193838254172017,
      "grad_norm": 0.12033534795045853,
      "learning_rate": 4.329896907216495e-05,
      "loss": 0.1014,
      "step": 2430
    },
    {
      "epoch": 3.1322207958921693,
      "grad_norm": 0.11646905541419983,
      "learning_rate": 4.3004418262150226e-05,
      "loss": 0.1059,
      "step": 2440
    },
    {
      "epoch": 3.1450577663671373,
      "grad_norm": 0.11194875836372375,
      "learning_rate": 4.270986745213549e-05,
      "loss": 0.1051,
      "step": 2450
    },
    {
      "epoch": 3.1578947368421053,
      "grad_norm": 0.09517650306224823,
      "learning_rate": 4.241531664212077e-05,
      "loss": 0.106,
      "step": 2460
    },
    {
      "epoch": 3.1707317073170733,
      "grad_norm": 0.10933737456798553,
      "learning_rate": 4.212076583210604e-05,
      "loss": 0.1038,
      "step": 2470
    },
    {
      "epoch": 3.183568677792041,
      "grad_norm": 0.11497558653354645,
      "learning_rate": 4.1826215022091314e-05,
      "loss": 0.121,
      "step": 2480
    },
    {
      "epoch": 3.196405648267009,
      "grad_norm": 0.10219521820545197,
      "learning_rate": 4.153166421207659e-05,
      "loss": 0.1137,
      "step": 2490
    },
    {
      "epoch": 3.209242618741977,
      "grad_norm": 0.13737627863883972,
      "learning_rate": 4.1237113402061855e-05,
      "loss": 0.1052,
      "step": 2500
    },
    {
      "epoch": 3.209242618741977,
      "eval_gen_len": 38.946,
      "eval_loss": 0.09647729247808456,
      "eval_rouge1": 0.5912,
      "eval_rouge2": 0.3082,
      "eval_rougeL": 0.5595,
      "eval_runtime": 47.0681,
      "eval_samples_per_second": 10.623,
      "eval_steps_per_second": 1.338,
      "step": 2500
    },
    {
      "epoch": 3.222079589216945,
      "grad_norm": 0.11823912709951401,
      "learning_rate": 4.094256259204713e-05,
      "loss": 0.1061,
      "step": 2510
    },
    {
      "epoch": 3.2349165596919125,
      "grad_norm": 0.1412975937128067,
      "learning_rate": 4.06480117820324e-05,
      "loss": 0.1071,
      "step": 2520
    },
    {
      "epoch": 3.2477535301668805,
      "grad_norm": 0.09195394068956375,
      "learning_rate": 4.0353460972017676e-05,
      "loss": 0.11,
      "step": 2530
    },
    {
      "epoch": 3.2605905006418485,
      "grad_norm": 0.11850762367248535,
      "learning_rate": 4.005891016200295e-05,
      "loss": 0.1033,
      "step": 2540
    },
    {
      "epoch": 3.2734274711168165,
      "grad_norm": 0.09788330644369125,
      "learning_rate": 3.976435935198822e-05,
      "loss": 0.1102,
      "step": 2550
    },
    {
      "epoch": 3.2862644415917845,
      "grad_norm": 0.3657642602920532,
      "learning_rate": 3.946980854197349e-05,
      "loss": 0.1062,
      "step": 2560
    },
    {
      "epoch": 3.299101412066752,
      "grad_norm": 0.1347455084323883,
      "learning_rate": 3.9175257731958764e-05,
      "loss": 0.1036,
      "step": 2570
    },
    {
      "epoch": 3.31193838254172,
      "grad_norm": 0.11970392614603043,
      "learning_rate": 3.888070692194404e-05,
      "loss": 0.1059,
      "step": 2580
    },
    {
      "epoch": 3.324775353016688,
      "grad_norm": 0.13223974406719208,
      "learning_rate": 3.858615611192931e-05,
      "loss": 0.1104,
      "step": 2590
    },
    {
      "epoch": 3.337612323491656,
      "grad_norm": 0.13846062123775482,
      "learning_rate": 3.8291605301914585e-05,
      "loss": 0.1051,
      "step": 2600
    },
    {
      "epoch": 3.3504492939666237,
      "grad_norm": 0.12654447555541992,
      "learning_rate": 3.799705449189985e-05,
      "loss": 0.1047,
      "step": 2610
    },
    {
      "epoch": 3.3632862644415917,
      "grad_norm": 0.18010291457176208,
      "learning_rate": 3.7702503681885126e-05,
      "loss": 0.1049,
      "step": 2620
    },
    {
      "epoch": 3.3761232349165597,
      "grad_norm": 0.09747201949357986,
      "learning_rate": 3.74079528718704e-05,
      "loss": 0.1117,
      "step": 2630
    },
    {
      "epoch": 3.3889602053915278,
      "grad_norm": 0.105229951441288,
      "learning_rate": 3.7113402061855674e-05,
      "loss": 0.1036,
      "step": 2640
    },
    {
      "epoch": 3.4017971758664953,
      "grad_norm": 0.11923747509717941,
      "learning_rate": 3.681885125184095e-05,
      "loss": 0.1161,
      "step": 2650
    },
    {
      "epoch": 3.4146341463414633,
      "grad_norm": 0.11879535019397736,
      "learning_rate": 3.6524300441826214e-05,
      "loss": 0.1048,
      "step": 2660
    },
    {
      "epoch": 3.4274711168164314,
      "grad_norm": 0.12212618440389633,
      "learning_rate": 3.622974963181149e-05,
      "loss": 0.1156,
      "step": 2670
    },
    {
      "epoch": 3.4403080872913994,
      "grad_norm": 0.11665867269039154,
      "learning_rate": 3.593519882179676e-05,
      "loss": 0.1062,
      "step": 2680
    },
    {
      "epoch": 3.453145057766367,
      "grad_norm": 0.162534698843956,
      "learning_rate": 3.5640648011782035e-05,
      "loss": 0.1104,
      "step": 2690
    },
    {
      "epoch": 3.465982028241335,
      "grad_norm": 0.1339045912027359,
      "learning_rate": 3.534609720176731e-05,
      "loss": 0.1218,
      "step": 2700
    },
    {
      "epoch": 3.478818998716303,
      "grad_norm": 0.13339383900165558,
      "learning_rate": 3.5051546391752576e-05,
      "loss": 0.1099,
      "step": 2710
    },
    {
      "epoch": 3.491655969191271,
      "grad_norm": 0.22776636481285095,
      "learning_rate": 3.475699558173785e-05,
      "loss": 0.1053,
      "step": 2720
    },
    {
      "epoch": 3.504492939666239,
      "grad_norm": 0.10782419145107269,
      "learning_rate": 3.4462444771723124e-05,
      "loss": 0.1109,
      "step": 2730
    },
    {
      "epoch": 3.5173299101412066,
      "grad_norm": 0.17186006903648376,
      "learning_rate": 3.41678939617084e-05,
      "loss": 0.1116,
      "step": 2740
    },
    {
      "epoch": 3.5301668806161746,
      "grad_norm": 0.11577029526233673,
      "learning_rate": 3.387334315169367e-05,
      "loss": 0.1112,
      "step": 2750
    },
    {
      "epoch": 3.5430038510911426,
      "grad_norm": 0.12498428672552109,
      "learning_rate": 3.357879234167894e-05,
      "loss": 0.1043,
      "step": 2760
    },
    {
      "epoch": 3.55584082156611,
      "grad_norm": 0.11685289442539215,
      "learning_rate": 3.328424153166421e-05,
      "loss": 0.1029,
      "step": 2770
    },
    {
      "epoch": 3.568677792041078,
      "grad_norm": 0.10624662786722183,
      "learning_rate": 3.2989690721649485e-05,
      "loss": 0.1104,
      "step": 2780
    },
    {
      "epoch": 3.581514762516046,
      "grad_norm": 0.3390846252441406,
      "learning_rate": 3.269513991163476e-05,
      "loss": 0.1019,
      "step": 2790
    },
    {
      "epoch": 3.594351732991014,
      "grad_norm": 0.10731125622987747,
      "learning_rate": 3.240058910162003e-05,
      "loss": 0.103,
      "step": 2800
    },
    {
      "epoch": 3.607188703465982,
      "grad_norm": 0.1548955887556076,
      "learning_rate": 3.2106038291605307e-05,
      "loss": 0.1069,
      "step": 2810
    },
    {
      "epoch": 3.62002567394095,
      "grad_norm": 0.11366964876651764,
      "learning_rate": 3.1811487481590573e-05,
      "loss": 0.1119,
      "step": 2820
    },
    {
      "epoch": 3.6328626444159178,
      "grad_norm": 0.13698551058769226,
      "learning_rate": 3.151693667157585e-05,
      "loss": 0.1035,
      "step": 2830
    },
    {
      "epoch": 3.645699614890886,
      "grad_norm": 0.20564177632331848,
      "learning_rate": 3.122238586156112e-05,
      "loss": 0.1073,
      "step": 2840
    },
    {
      "epoch": 3.658536585365854,
      "grad_norm": 0.13262344896793365,
      "learning_rate": 3.0927835051546395e-05,
      "loss": 0.1037,
      "step": 2850
    },
    {
      "epoch": 3.6713735558408214,
      "grad_norm": 0.10157434642314911,
      "learning_rate": 3.063328424153167e-05,
      "loss": 0.0991,
      "step": 2860
    },
    {
      "epoch": 3.6842105263157894,
      "grad_norm": 0.14752894639968872,
      "learning_rate": 3.0338733431516935e-05,
      "loss": 0.1154,
      "step": 2870
    },
    {
      "epoch": 3.6970474967907574,
      "grad_norm": 0.1402595192193985,
      "learning_rate": 3.004418262150221e-05,
      "loss": 0.1068,
      "step": 2880
    },
    {
      "epoch": 3.7098844672657254,
      "grad_norm": 0.1572103500366211,
      "learning_rate": 2.9749631811487483e-05,
      "loss": 0.1017,
      "step": 2890
    },
    {
      "epoch": 3.7227214377406934,
      "grad_norm": 0.10825837403535843,
      "learning_rate": 2.9455081001472756e-05,
      "loss": 0.108,
      "step": 2900
    },
    {
      "epoch": 3.735558408215661,
      "grad_norm": 0.09703366458415985,
      "learning_rate": 2.916053019145803e-05,
      "loss": 0.103,
      "step": 2910
    },
    {
      "epoch": 3.748395378690629,
      "grad_norm": 0.11436077952384949,
      "learning_rate": 2.8865979381443297e-05,
      "loss": 0.1163,
      "step": 2920
    },
    {
      "epoch": 3.761232349165597,
      "grad_norm": 0.13263103365898132,
      "learning_rate": 2.857142857142857e-05,
      "loss": 0.1078,
      "step": 2930
    },
    {
      "epoch": 3.7740693196405646,
      "grad_norm": 0.12120107561349869,
      "learning_rate": 2.8276877761413845e-05,
      "loss": 0.1061,
      "step": 2940
    },
    {
      "epoch": 3.7869062901155326,
      "grad_norm": 0.10355032235383987,
      "learning_rate": 2.798232695139912e-05,
      "loss": 0.1084,
      "step": 2950
    },
    {
      "epoch": 3.7997432605905006,
      "grad_norm": 0.12106519192457199,
      "learning_rate": 2.7687776141384392e-05,
      "loss": 0.099,
      "step": 2960
    },
    {
      "epoch": 3.8125802310654686,
      "grad_norm": 0.11807779967784882,
      "learning_rate": 2.7393225331369666e-05,
      "loss": 0.1076,
      "step": 2970
    },
    {
      "epoch": 3.8254172015404366,
      "grad_norm": 0.22053521871566772,
      "learning_rate": 2.7098674521354933e-05,
      "loss": 0.0998,
      "step": 2980
    },
    {
      "epoch": 3.8382541720154046,
      "grad_norm": 0.12982416152954102,
      "learning_rate": 2.6804123711340206e-05,
      "loss": 0.1112,
      "step": 2990
    },
    {
      "epoch": 3.851091142490372,
      "grad_norm": 0.10256698727607727,
      "learning_rate": 2.650957290132548e-05,
      "loss": 0.1081,
      "step": 3000
    },
    {
      "epoch": 3.851091142490372,
      "eval_gen_len": 38.946,
      "eval_loss": 0.09579482674598694,
      "eval_rouge1": 0.592,
      "eval_rouge2": 0.3111,
      "eval_rougeL": 0.5586,
      "eval_runtime": 46.2091,
      "eval_samples_per_second": 10.82,
      "eval_steps_per_second": 1.363,
      "step": 3000
    },
    {
      "epoch": 3.8639281129653402,
      "grad_norm": 0.09973964840173721,
      "learning_rate": 2.6215022091310754e-05,
      "loss": 0.1113,
      "step": 3010
    },
    {
      "epoch": 3.8767650834403082,
      "grad_norm": 0.1375608742237091,
      "learning_rate": 2.5920471281296028e-05,
      "loss": 0.0979,
      "step": 3020
    },
    {
      "epoch": 3.889602053915276,
      "grad_norm": 0.129994735121727,
      "learning_rate": 2.5625920471281295e-05,
      "loss": 0.1094,
      "step": 3030
    },
    {
      "epoch": 3.902439024390244,
      "grad_norm": 0.11172019690275192,
      "learning_rate": 2.5331369661266568e-05,
      "loss": 0.1061,
      "step": 3040
    },
    {
      "epoch": 3.915275994865212,
      "grad_norm": 0.13806381821632385,
      "learning_rate": 2.5036818851251842e-05,
      "loss": 0.1001,
      "step": 3050
    },
    {
      "epoch": 3.92811296534018,
      "grad_norm": 0.10557816922664642,
      "learning_rate": 2.4742268041237116e-05,
      "loss": 0.1035,
      "step": 3060
    },
    {
      "epoch": 3.940949935815148,
      "grad_norm": 0.13717150688171387,
      "learning_rate": 2.4447717231222386e-05,
      "loss": 0.1048,
      "step": 3070
    },
    {
      "epoch": 3.9537869062901154,
      "grad_norm": 0.31350722908973694,
      "learning_rate": 2.415316642120766e-05,
      "loss": 0.1171,
      "step": 3080
    },
    {
      "epoch": 3.9666238767650834,
      "grad_norm": 0.12666529417037964,
      "learning_rate": 2.3858615611192933e-05,
      "loss": 0.1111,
      "step": 3090
    },
    {
      "epoch": 3.9794608472400514,
      "grad_norm": 0.1276416778564453,
      "learning_rate": 2.3564064801178204e-05,
      "loss": 0.1151,
      "step": 3100
    },
    {
      "epoch": 3.992297817715019,
      "grad_norm": 0.11248279362916946,
      "learning_rate": 2.3269513991163478e-05,
      "loss": 0.1013,
      "step": 3110
    },
    {
      "epoch": 4.005134788189987,
      "grad_norm": 0.14889618754386902,
      "learning_rate": 2.2974963181148748e-05,
      "loss": 0.1072,
      "step": 3120
    },
    {
      "epoch": 4.017971758664955,
      "grad_norm": 0.10598665475845337,
      "learning_rate": 2.268041237113402e-05,
      "loss": 0.1017,
      "step": 3130
    },
    {
      "epoch": 4.030808729139923,
      "grad_norm": 0.11106213927268982,
      "learning_rate": 2.2385861561119295e-05,
      "loss": 0.1028,
      "step": 3140
    },
    {
      "epoch": 4.043645699614891,
      "grad_norm": 0.14373856782913208,
      "learning_rate": 2.2091310751104566e-05,
      "loss": 0.1086,
      "step": 3150
    },
    {
      "epoch": 4.056482670089859,
      "grad_norm": 0.13207481801509857,
      "learning_rate": 2.179675994108984e-05,
      "loss": 0.0997,
      "step": 3160
    },
    {
      "epoch": 4.069319640564827,
      "grad_norm": 0.11155064404010773,
      "learning_rate": 2.1502209131075113e-05,
      "loss": 0.1083,
      "step": 3170
    },
    {
      "epoch": 4.082156611039794,
      "grad_norm": 0.10416271537542343,
      "learning_rate": 2.1207658321060383e-05,
      "loss": 0.1164,
      "step": 3180
    },
    {
      "epoch": 4.094993581514762,
      "grad_norm": 0.11257210373878479,
      "learning_rate": 2.0913107511045657e-05,
      "loss": 0.1121,
      "step": 3190
    },
    {
      "epoch": 4.10783055198973,
      "grad_norm": 0.2550837993621826,
      "learning_rate": 2.0618556701030927e-05,
      "loss": 0.1112,
      "step": 3200
    },
    {
      "epoch": 4.120667522464698,
      "grad_norm": 0.1105370745062828,
      "learning_rate": 2.03240058910162e-05,
      "loss": 0.1064,
      "step": 3210
    },
    {
      "epoch": 4.133504492939666,
      "grad_norm": 0.11678767949342728,
      "learning_rate": 2.0029455081001475e-05,
      "loss": 0.1051,
      "step": 3220
    },
    {
      "epoch": 4.146341463414634,
      "grad_norm": 0.09064505994319916,
      "learning_rate": 1.9734904270986745e-05,
      "loss": 0.1062,
      "step": 3230
    },
    {
      "epoch": 4.159178433889602,
      "grad_norm": 0.11429174244403839,
      "learning_rate": 1.944035346097202e-05,
      "loss": 0.1082,
      "step": 3240
    },
    {
      "epoch": 4.17201540436457,
      "grad_norm": 0.11602449417114258,
      "learning_rate": 1.9145802650957293e-05,
      "loss": 0.1008,
      "step": 3250
    },
    {
      "epoch": 4.184852374839538,
      "grad_norm": 0.09767124801874161,
      "learning_rate": 1.8851251840942563e-05,
      "loss": 0.1052,
      "step": 3260
    },
    {
      "epoch": 4.197689345314505,
      "grad_norm": 0.08755617588758469,
      "learning_rate": 1.8556701030927837e-05,
      "loss": 0.1021,
      "step": 3270
    },
    {
      "epoch": 4.2105263157894735,
      "grad_norm": 0.08597078919410706,
      "learning_rate": 1.8262150220913107e-05,
      "loss": 0.1034,
      "step": 3280
    },
    {
      "epoch": 4.2233632862644415,
      "grad_norm": 0.12039302289485931,
      "learning_rate": 1.796759941089838e-05,
      "loss": 0.1171,
      "step": 3290
    },
    {
      "epoch": 4.2362002567394095,
      "grad_norm": 0.11615891754627228,
      "learning_rate": 1.7673048600883655e-05,
      "loss": 0.1044,
      "step": 3300
    },
    {
      "epoch": 4.2490372272143775,
      "grad_norm": 0.1104285717010498,
      "learning_rate": 1.7378497790868925e-05,
      "loss": 0.1025,
      "step": 3310
    },
    {
      "epoch": 4.2618741976893455,
      "grad_norm": 0.10667522996664047,
      "learning_rate": 1.70839469808542e-05,
      "loss": 0.099,
      "step": 3320
    },
    {
      "epoch": 4.2747111681643135,
      "grad_norm": 0.11607446521520615,
      "learning_rate": 1.678939617083947e-05,
      "loss": 0.1085,
      "step": 3330
    },
    {
      "epoch": 4.2875481386392815,
      "grad_norm": 0.09650270640850067,
      "learning_rate": 1.6494845360824743e-05,
      "loss": 0.0944,
      "step": 3340
    },
    {
      "epoch": 4.300385109114249,
      "grad_norm": 0.11899662762880325,
      "learning_rate": 1.6200294550810016e-05,
      "loss": 0.1002,
      "step": 3350
    },
    {
      "epoch": 4.313222079589217,
      "grad_norm": 0.12044407427310944,
      "learning_rate": 1.5905743740795287e-05,
      "loss": 0.1083,
      "step": 3360
    },
    {
      "epoch": 4.326059050064185,
      "grad_norm": 0.09684235602617264,
      "learning_rate": 1.561119293078056e-05,
      "loss": 0.1012,
      "step": 3370
    },
    {
      "epoch": 4.338896020539153,
      "grad_norm": 0.12105019390583038,
      "learning_rate": 1.5316642120765834e-05,
      "loss": 0.1034,
      "step": 3380
    },
    {
      "epoch": 4.351732991014121,
      "grad_norm": 0.14494775235652924,
      "learning_rate": 1.5022091310751105e-05,
      "loss": 0.0926,
      "step": 3390
    },
    {
      "epoch": 4.364569961489089,
      "grad_norm": 0.1320798546075821,
      "learning_rate": 1.4727540500736378e-05,
      "loss": 0.1137,
      "step": 3400
    },
    {
      "epoch": 4.377406931964057,
      "grad_norm": 0.1228160411119461,
      "learning_rate": 1.4432989690721649e-05,
      "loss": 0.1071,
      "step": 3410
    },
    {
      "epoch": 4.390243902439025,
      "grad_norm": 0.12092225253582001,
      "learning_rate": 1.4138438880706922e-05,
      "loss": 0.1003,
      "step": 3420
    },
    {
      "epoch": 4.403080872913993,
      "grad_norm": 0.10531840473413467,
      "learning_rate": 1.3843888070692196e-05,
      "loss": 0.1083,
      "step": 3430
    },
    {
      "epoch": 4.41591784338896,
      "grad_norm": 0.14029191434383392,
      "learning_rate": 1.3549337260677466e-05,
      "loss": 0.1074,
      "step": 3440
    },
    {
      "epoch": 4.428754813863928,
      "grad_norm": 0.10859360545873642,
      "learning_rate": 1.325478645066274e-05,
      "loss": 0.1004,
      "step": 3450
    },
    {
      "epoch": 4.441591784338896,
      "grad_norm": 0.10318469256162643,
      "learning_rate": 1.2960235640648014e-05,
      "loss": 0.104,
      "step": 3460
    },
    {
      "epoch": 4.454428754813864,
      "grad_norm": 0.12979723513126373,
      "learning_rate": 1.2665684830633284e-05,
      "loss": 0.0968,
      "step": 3470
    },
    {
      "epoch": 4.467265725288832,
      "grad_norm": 0.14862096309661865,
      "learning_rate": 1.2371134020618558e-05,
      "loss": 0.1124,
      "step": 3480
    },
    {
      "epoch": 4.4801026957638,
      "grad_norm": 0.1266416311264038,
      "learning_rate": 1.207658321060383e-05,
      "loss": 0.1135,
      "step": 3490
    },
    {
      "epoch": 4.492939666238768,
      "grad_norm": 0.12597477436065674,
      "learning_rate": 1.1782032400589102e-05,
      "loss": 0.1071,
      "step": 3500
    },
    {
      "epoch": 4.492939666238768,
      "eval_gen_len": 38.946,
      "eval_loss": 0.09506849944591522,
      "eval_rouge1": 0.5959,
      "eval_rouge2": 0.3135,
      "eval_rougeL": 0.5625,
      "eval_runtime": 45.8514,
      "eval_samples_per_second": 10.905,
      "eval_steps_per_second": 1.374,
      "step": 3500
    },
    {
      "epoch": 4.505776636713735,
      "grad_norm": 0.11168000102043152,
      "learning_rate": 1.1487481590574374e-05,
      "loss": 0.107,
      "step": 3510
    },
    {
      "epoch": 4.518613607188703,
      "grad_norm": 0.1055913120508194,
      "learning_rate": 1.1192930780559648e-05,
      "loss": 0.1001,
      "step": 3520
    },
    {
      "epoch": 4.531450577663671,
      "grad_norm": 0.10443459451198578,
      "learning_rate": 1.089837997054492e-05,
      "loss": 0.1117,
      "step": 3530
    },
    {
      "epoch": 4.544287548138639,
      "grad_norm": 0.1707678884267807,
      "learning_rate": 1.0603829160530192e-05,
      "loss": 0.095,
      "step": 3540
    },
    {
      "epoch": 4.557124518613607,
      "grad_norm": 0.10272827744483948,
      "learning_rate": 1.0309278350515464e-05,
      "loss": 0.1027,
      "step": 3550
    },
    {
      "epoch": 4.569961489088575,
      "grad_norm": 0.14477787911891937,
      "learning_rate": 1.0014727540500737e-05,
      "loss": 0.114,
      "step": 3560
    },
    {
      "epoch": 4.582798459563543,
      "grad_norm": 0.12023382633924484,
      "learning_rate": 9.72017673048601e-06,
      "loss": 0.111,
      "step": 3570
    },
    {
      "epoch": 4.595635430038511,
      "grad_norm": 0.11823578923940659,
      "learning_rate": 9.425625920471282e-06,
      "loss": 0.1149,
      "step": 3580
    },
    {
      "epoch": 4.608472400513479,
      "grad_norm": 0.13746647536754608,
      "learning_rate": 9.131075110456554e-06,
      "loss": 0.1068,
      "step": 3590
    },
    {
      "epoch": 4.621309370988447,
      "grad_norm": 0.11988291144371033,
      "learning_rate": 8.836524300441827e-06,
      "loss": 0.0955,
      "step": 3600
    },
    {
      "epoch": 4.634146341463414,
      "grad_norm": 0.09879586100578308,
      "learning_rate": 8.5419734904271e-06,
      "loss": 0.1031,
      "step": 3610
    },
    {
      "epoch": 4.646983311938382,
      "grad_norm": 0.1175203025341034,
      "learning_rate": 8.247422680412371e-06,
      "loss": 0.1135,
      "step": 3620
    },
    {
      "epoch": 4.65982028241335,
      "grad_norm": 0.10659980028867722,
      "learning_rate": 7.952871870397643e-06,
      "loss": 0.1005,
      "step": 3630
    },
    {
      "epoch": 4.672657252888318,
      "grad_norm": 0.12701040506362915,
      "learning_rate": 7.658321060382917e-06,
      "loss": 0.1088,
      "step": 3640
    },
    {
      "epoch": 4.685494223363286,
      "grad_norm": 0.10160648077726364,
      "learning_rate": 7.363770250368189e-06,
      "loss": 0.1076,
      "step": 3650
    },
    {
      "epoch": 4.698331193838254,
      "grad_norm": 0.11782427132129669,
      "learning_rate": 7.069219440353461e-06,
      "loss": 0.1054,
      "step": 3660
    },
    {
      "epoch": 4.711168164313222,
      "grad_norm": 0.11307085305452347,
      "learning_rate": 6.774668630338733e-06,
      "loss": 0.1093,
      "step": 3670
    },
    {
      "epoch": 4.7240051347881895,
      "grad_norm": 0.13703173398971558,
      "learning_rate": 6.480117820324007e-06,
      "loss": 0.1043,
      "step": 3680
    },
    {
      "epoch": 4.7368421052631575,
      "grad_norm": 0.11539970338344574,
      "learning_rate": 6.185567010309279e-06,
      "loss": 0.1016,
      "step": 3690
    },
    {
      "epoch": 4.7496790757381255,
      "grad_norm": 0.10878204554319382,
      "learning_rate": 5.891016200294551e-06,
      "loss": 0.1065,
      "step": 3700
    },
    {
      "epoch": 4.7625160462130935,
      "grad_norm": 0.11433741450309753,
      "learning_rate": 5.596465390279824e-06,
      "loss": 0.1097,
      "step": 3710
    },
    {
      "epoch": 4.775353016688062,
      "grad_norm": 0.12714996933937073,
      "learning_rate": 5.301914580265096e-06,
      "loss": 0.1102,
      "step": 3720
    },
    {
      "epoch": 4.78818998716303,
      "grad_norm": 0.1310754418373108,
      "learning_rate": 5.007363770250369e-06,
      "loss": 0.1029,
      "step": 3730
    },
    {
      "epoch": 4.801026957637998,
      "grad_norm": 0.10799030214548111,
      "learning_rate": 4.712812960235641e-06,
      "loss": 0.1072,
      "step": 3740
    },
    {
      "epoch": 4.813863928112966,
      "grad_norm": 0.12069649994373322,
      "learning_rate": 4.418262150220914e-06,
      "loss": 0.1092,
      "step": 3750
    },
    {
      "epoch": 4.826700898587934,
      "grad_norm": 0.09965723752975464,
      "learning_rate": 4.123711340206186e-06,
      "loss": 0.0987,
      "step": 3760
    },
    {
      "epoch": 4.839537869062902,
      "grad_norm": 0.1061498150229454,
      "learning_rate": 3.8291605301914585e-06,
      "loss": 0.1026,
      "step": 3770
    },
    {
      "epoch": 4.852374839537869,
      "grad_norm": 0.10951300710439682,
      "learning_rate": 3.5346097201767306e-06,
      "loss": 0.101,
      "step": 3780
    },
    {
      "epoch": 4.865211810012837,
      "grad_norm": 0.11674579232931137,
      "learning_rate": 3.2400589101620034e-06,
      "loss": 0.1026,
      "step": 3790
    },
    {
      "epoch": 4.878048780487805,
      "grad_norm": 0.12058935314416885,
      "learning_rate": 2.9455081001472755e-06,
      "loss": 0.1021,
      "step": 3800
    },
    {
      "epoch": 4.890885750962773,
      "grad_norm": 0.11995814740657806,
      "learning_rate": 2.650957290132548e-06,
      "loss": 0.1066,
      "step": 3810
    },
    {
      "epoch": 4.903722721437741,
      "grad_norm": 0.14141815900802612,
      "learning_rate": 2.3564064801178204e-06,
      "loss": 0.1064,
      "step": 3820
    },
    {
      "epoch": 4.916559691912709,
      "grad_norm": 0.10517096519470215,
      "learning_rate": 2.061855670103093e-06,
      "loss": 0.1008,
      "step": 3830
    },
    {
      "epoch": 4.929396662387677,
      "grad_norm": 0.10266517102718353,
      "learning_rate": 1.7673048600883653e-06,
      "loss": 0.1037,
      "step": 3840
    },
    {
      "epoch": 4.942233632862644,
      "grad_norm": 0.114813894033432,
      "learning_rate": 1.4727540500736377e-06,
      "loss": 0.1026,
      "step": 3850
    },
    {
      "epoch": 4.955070603337612,
      "grad_norm": 0.09497864544391632,
      "learning_rate": 1.1782032400589102e-06,
      "loss": 0.112,
      "step": 3860
    },
    {
      "epoch": 4.96790757381258,
      "grad_norm": 0.11201301217079163,
      "learning_rate": 8.836524300441826e-07,
      "loss": 0.1083,
      "step": 3870
    },
    {
      "epoch": 4.980744544287548,
      "grad_norm": 0.12003788352012634,
      "learning_rate": 5.891016200294551e-07,
      "loss": 0.1117,
      "step": 3880
    },
    {
      "epoch": 4.993581514762516,
      "grad_norm": 0.1579122096300125,
      "learning_rate": 2.9455081001472755e-07,
      "loss": 0.1106,
      "step": 3890
    }
  ],
  "logging_steps": 10,
  "max_steps": 3895,
  "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": 8431794231705600.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}