diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4086 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.468488566648076,
+  "eval_steps": 500,
+  "global_step": 5799,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.011154489682097044,
+      "grad_norm": 0.4311506450176239,
+      "learning_rate": 1e-05,
+      "loss": 3.2953,
+      "step": 10
+    },
+    {
+      "epoch": 0.022308979364194088,
+      "grad_norm": 0.6115325689315796,
+      "learning_rate": 2e-05,
+      "loss": 3.3283,
+      "step": 20
+    },
+    {
+      "epoch": 0.03346346904629113,
+      "grad_norm": 0.831686794757843,
+      "learning_rate": 1.9999852237819516e-05,
+      "loss": 2.9624,
+      "step": 30
+    },
+    {
+      "epoch": 0.044617958728388175,
+      "grad_norm": 1.2678371667861938,
+      "learning_rate": 1.9999408955644783e-05,
+      "loss": 2.891,
+      "step": 40
+    },
+    {
+      "epoch": 0.05577244841048522,
+      "grad_norm": 0.7378564476966858,
+      "learning_rate": 1.9998670166575877e-05,
+      "loss": 2.6735,
+      "step": 50
+    },
+    {
+      "epoch": 0.06692693809258227,
+      "grad_norm": 1.0785635709762573,
+      "learning_rate": 1.9997635892445808e-05,
+      "loss": 2.6967,
+      "step": 60
+    },
+    {
+      "epoch": 0.0780814277746793,
+      "grad_norm": 0.9325912594795227,
+      "learning_rate": 1.9996306163819902e-05,
+      "loss": 2.1375,
+      "step": 70
+    },
+    {
+      "epoch": 0.08923591745677635,
+      "grad_norm": 0.4600697159767151,
+      "learning_rate": 1.999468101999488e-05,
+      "loss": 2.1378,
+      "step": 80
+    },
+    {
+      "epoch": 0.1003904071388734,
+      "grad_norm": 7.8976664543151855,
+      "learning_rate": 1.9992760508997694e-05,
+      "loss": 2.0217,
+      "step": 90
+    },
+    {
+      "epoch": 0.11154489682097044,
+      "grad_norm": 1.2072306871414185,
+      "learning_rate": 1.999054468758413e-05,
+      "loss": 2.0504,
+      "step": 100
+    },
+    {
+      "epoch": 0.12269938650306748,
+      "grad_norm": 0.561038613319397,
+      "learning_rate": 1.9988033621237107e-05,
+      "loss": 1.9364,
+      "step": 110
+    },
+    {
+      "epoch": 0.13385387618516453,
+      "grad_norm": 0.6420221924781799,
+      "learning_rate": 1.998522738416475e-05,
+      "loss": 1.8163,
+      "step": 120
+    },
+    {
+      "epoch": 0.14500836586726157,
+      "grad_norm": 1.0922359228134155,
+      "learning_rate": 1.9982126059298202e-05,
+      "loss": 1.7226,
+      "step": 130
+    },
+    {
+      "epoch": 0.1561628555493586,
+      "grad_norm": 4.673393726348877,
+      "learning_rate": 1.997872973828917e-05,
+      "loss": 1.6478,
+      "step": 140
+    },
+    {
+      "epoch": 0.16731734523145567,
+      "grad_norm": 0.7612175345420837,
+      "learning_rate": 1.997503852150721e-05,
+      "loss": 1.5993,
+      "step": 150
+    },
+    {
+      "epoch": 0.1784718349135527,
+      "grad_norm": 0.4687439799308777,
+      "learning_rate": 1.997105251803677e-05,
+      "loss": 1.5498,
+      "step": 160
+    },
+    {
+      "epoch": 0.18962632459564974,
+      "grad_norm": 0.5883293747901917,
+      "learning_rate": 1.9966771845673968e-05,
+      "loss": 1.4306,
+      "step": 170
+    },
+    {
+      "epoch": 0.2007808142777468,
+      "grad_norm": 0.6468557715415955,
+      "learning_rate": 1.9962196630923095e-05,
+      "loss": 1.5006,
+      "step": 180
+    },
+    {
+      "epoch": 0.21193530395984383,
+      "grad_norm": 0.487519234418869,
+      "learning_rate": 1.9957327008992895e-05,
+      "loss": 1.4175,
+      "step": 190
+    },
+    {
+      "epoch": 0.22308979364194087,
+      "grad_norm": 0.5795869827270508,
+      "learning_rate": 1.995216312379256e-05,
+      "loss": 1.3625,
+      "step": 200
+    },
+    {
+      "epoch": 0.23424428332403793,
+      "grad_norm": 0.7335501313209534,
+      "learning_rate": 1.9946705127927474e-05,
+      "loss": 1.4419,
+      "step": 210
+    },
+    {
+      "epoch": 0.24539877300613497,
+      "grad_norm": 0.6072186231613159,
+      "learning_rate": 1.9940953182694716e-05,
+      "loss": 1.3177,
+      "step": 220
+    },
+    {
+      "epoch": 0.25655326268823203,
+      "grad_norm": 0.5896971225738525,
+      "learning_rate": 1.9934907458078274e-05,
+      "loss": 1.3797,
+      "step": 230
+    },
+    {
+      "epoch": 0.26770775237032907,
+      "grad_norm": 0.4922662675380707,
+      "learning_rate": 1.9928568132744042e-05,
+      "loss": 1.3181,
+      "step": 240
+    },
+    {
+      "epoch": 0.2788622420524261,
+      "grad_norm": 0.6160895228385925,
+      "learning_rate": 1.9921935394034525e-05,
+      "loss": 1.2891,
+      "step": 250
+    },
+    {
+      "epoch": 0.29001673173452314,
+      "grad_norm": 0.690238356590271,
+      "learning_rate": 1.9915009437963315e-05,
+      "loss": 1.2798,
+      "step": 260
+    },
+    {
+      "epoch": 0.30117122141662017,
+      "grad_norm": 1.5020575523376465,
+      "learning_rate": 1.9907790469209282e-05,
+      "loss": 1.2952,
+      "step": 270
+    },
+    {
+      "epoch": 0.3123257110987172,
+      "grad_norm": 0.5224294066429138,
+      "learning_rate": 1.9900278701110536e-05,
+      "loss": 1.2377,
+      "step": 280
+    },
+    {
+      "epoch": 0.3234802007808143,
+      "grad_norm": 0.5903538465499878,
+      "learning_rate": 1.989247435565813e-05,
+      "loss": 1.177,
+      "step": 290
+    },
+    {
+      "epoch": 0.33463469046291133,
+      "grad_norm": 0.5945698618888855,
+      "learning_rate": 1.988437766348948e-05,
+      "loss": 1.3077,
+      "step": 300
+    },
+    {
+      "epoch": 0.34578918014500837,
+      "grad_norm": 3.4969353675842285,
+      "learning_rate": 1.9875988863881562e-05,
+      "loss": 1.2398,
+      "step": 310
+    },
+    {
+      "epoch": 0.3569436698271054,
+      "grad_norm": 0.7088640332221985,
+      "learning_rate": 1.9867308204743846e-05,
+      "loss": 1.2185,
+      "step": 320
+    },
+    {
+      "epoch": 0.36809815950920244,
+      "grad_norm": 0.6107991933822632,
+      "learning_rate": 1.985833594261095e-05,
+      "loss": 1.2245,
+      "step": 330
+    },
+    {
+      "epoch": 0.3792526491912995,
+      "grad_norm": 0.513905942440033,
+      "learning_rate": 1.9849072342635086e-05,
+      "loss": 1.17,
+      "step": 340
+    },
+    {
+      "epoch": 0.39040713887339656,
+      "grad_norm": 0.7432478666305542,
+      "learning_rate": 1.9839517678578194e-05,
+      "loss": 1.1798,
+      "step": 350
+    },
+    {
+      "epoch": 0.4015616285554936,
+      "grad_norm": 0.4735497832298279,
+      "learning_rate": 1.9829672232803873e-05,
+      "loss": 1.1861,
+      "step": 360
+    },
+    {
+      "epoch": 0.41271611823759063,
+      "grad_norm": 0.4925360083580017,
+      "learning_rate": 1.9819536296269036e-05,
+      "loss": 1.1734,
+      "step": 370
+    },
+    {
+      "epoch": 0.42387060791968767,
+      "grad_norm": 0.48574864864349365,
+      "learning_rate": 1.980911016851529e-05,
+      "loss": 1.1775,
+      "step": 380
+    },
+    {
+      "epoch": 0.4350250976017847,
+      "grad_norm": 0.5457922220230103,
+      "learning_rate": 1.9798394157660116e-05,
+      "loss": 1.1733,
+      "step": 390
+    },
+    {
+      "epoch": 0.44617958728388174,
+      "grad_norm": 0.5231193900108337,
+      "learning_rate": 1.9787388580387738e-05,
+      "loss": 1.1468,
+      "step": 400
+    },
+    {
+      "epoch": 0.45733407696597883,
+      "grad_norm": 0.7397568821907043,
+      "learning_rate": 1.9776093761939776e-05,
+      "loss": 1.139,
+      "step": 410
+    },
+    {
+      "epoch": 0.46848856664807587,
+      "grad_norm": 0.5115352869033813,
+      "learning_rate": 1.976451003610563e-05,
+      "loss": 1.2086,
+      "step": 420
+    },
+    {
+      "epoch": 0.4796430563301729,
+      "grad_norm": 0.6861724853515625,
+      "learning_rate": 1.975263774521262e-05,
+      "loss": 1.1599,
+      "step": 430
+    },
+    {
+      "epoch": 0.49079754601226994,
+      "grad_norm": 0.5191764235496521,
+      "learning_rate": 1.974047724011586e-05,
+      "loss": 1.1327,
+      "step": 440
+    },
+    {
+      "epoch": 0.501952035694367,
+      "grad_norm": 0.5202570557594299,
+      "learning_rate": 1.97280288801879e-05,
+      "loss": 1.1402,
+      "step": 450
+    },
+    {
+      "epoch": 0.5131065253764641,
+      "grad_norm": 0.5279308557510376,
+      "learning_rate": 1.9715293033308102e-05,
+      "loss": 1.1664,
+      "step": 460
+    },
+    {
+      "epoch": 0.524261015058561,
+      "grad_norm": 0.6293380856513977,
+      "learning_rate": 1.9702270075851767e-05,
+      "loss": 1.1578,
+      "step": 470
+    },
+    {
+      "epoch": 0.5354155047406581,
+      "grad_norm": 6.1000518798828125,
+      "learning_rate": 1.9688960392679014e-05,
+      "loss": 1.1117,
+      "step": 480
+    },
+    {
+      "epoch": 0.5465699944227551,
+      "grad_norm": 0.5824970602989197,
+      "learning_rate": 1.9675364377123405e-05,
+      "loss": 1.1407,
+      "step": 490
+    },
+    {
+      "epoch": 0.5577244841048522,
+      "grad_norm": 0.5770475268363953,
+      "learning_rate": 1.966148243098032e-05,
+      "loss": 1.1315,
+      "step": 500
+    },
+    {
+      "epoch": 0.5688789737869493,
+      "grad_norm": 103.2529296875,
+      "learning_rate": 1.9647314964495084e-05,
+      "loss": 1.1193,
+      "step": 510
+    },
+    {
+      "epoch": 0.5800334634690463,
+      "grad_norm": 0.5626741051673889,
+      "learning_rate": 1.9632862396350845e-05,
+      "loss": 1.1103,
+      "step": 520
+    },
+    {
+      "epoch": 0.5911879531511434,
+      "grad_norm": 0.5797271132469177,
+      "learning_rate": 1.9618125153656204e-05,
+      "loss": 1.0689,
+      "step": 530
+    },
+    {
+      "epoch": 0.6023424428332403,
+      "grad_norm": 0.6117788553237915,
+      "learning_rate": 1.960310367193258e-05,
+      "loss": 1.0919,
+      "step": 540
+    },
+    {
+      "epoch": 0.6134969325153374,
+      "grad_norm": 0.9841444492340088,
+      "learning_rate": 1.958779839510135e-05,
+      "loss": 1.1139,
+      "step": 550
+    },
+    {
+      "epoch": 0.6246514221974344,
+      "grad_norm": 0.5128628015518188,
+      "learning_rate": 1.9572209775470735e-05,
+      "loss": 1.1207,
+      "step": 560
+    },
+    {
+      "epoch": 0.6358059118795315,
+      "grad_norm": 0.6127890348434448,
+      "learning_rate": 1.9556338273722415e-05,
+      "loss": 1.0485,
+      "step": 570
+    },
+    {
+      "epoch": 0.6469604015616286,
+      "grad_norm": 0.6080933809280396,
+      "learning_rate": 1.9540184358897932e-05,
+      "loss": 1.0677,
+      "step": 580
+    },
+    {
+      "epoch": 0.6581148912437256,
+      "grad_norm": 136.31642150878906,
+      "learning_rate": 1.9523748508384827e-05,
+      "loss": 1.0736,
+      "step": 590
+    },
+    {
+      "epoch": 0.6692693809258227,
+      "grad_norm": 0.5842283368110657,
+      "learning_rate": 1.9507031207902515e-05,
+      "loss": 1.1008,
+      "step": 600
+    },
+    {
+      "epoch": 0.6804238706079196,
+      "grad_norm": 0.8367534279823303,
+      "learning_rate": 1.9490032951487955e-05,
+      "loss": 1.1138,
+      "step": 610
+    },
+    {
+      "epoch": 0.6915783602900167,
+      "grad_norm": 0.5327929258346558,
+      "learning_rate": 1.9472754241481035e-05,
+      "loss": 1.0249,
+      "step": 620
+    },
+    {
+      "epoch": 0.7027328499721138,
+      "grad_norm": 0.5699859857559204,
+      "learning_rate": 1.9455195588509723e-05,
+      "loss": 1.0513,
+      "step": 630
+    },
+    {
+      "epoch": 0.7138873396542108,
+      "grad_norm": 0.7566335797309875,
+      "learning_rate": 1.9437357511474987e-05,
+      "loss": 1.1048,
+      "step": 640
+    },
+    {
+      "epoch": 0.7250418293363079,
+      "grad_norm": 0.5633962154388428,
+      "learning_rate": 1.9419240537535468e-05,
+      "loss": 1.0867,
+      "step": 650
+    },
+    {
+      "epoch": 0.7361963190184049,
+      "grad_norm": 0.6204596161842346,
+      "learning_rate": 1.9400845202091872e-05,
+      "loss": 1.0773,
+      "step": 660
+    },
+    {
+      "epoch": 0.747350808700502,
+      "grad_norm": 0.6118286848068237,
+      "learning_rate": 1.938217204877118e-05,
+      "loss": 1.0388,
+      "step": 670
+    },
+    {
+      "epoch": 0.758505298382599,
+      "grad_norm": 0.5015501379966736,
+      "learning_rate": 1.9363221629410558e-05,
+      "loss": 1.0807,
+      "step": 680
+    },
+    {
+      "epoch": 0.769659788064696,
+      "grad_norm": 0.5233261585235596,
+      "learning_rate": 1.9343994504041067e-05,
+      "loss": 0.9913,
+      "step": 690
+    },
+    {
+      "epoch": 0.7808142777467931,
+      "grad_norm": 0.5710963010787964,
+      "learning_rate": 1.9324491240871098e-05,
+      "loss": 1.1108,
+      "step": 700
+    },
+    {
+      "epoch": 0.7919687674288901,
+      "grad_norm": 1.604062795639038,
+      "learning_rate": 1.930471241626959e-05,
+      "loss": 1.0384,
+      "step": 710
+    },
+    {
+      "epoch": 0.8031232571109872,
+      "grad_norm": 0.5526170134544373,
+      "learning_rate": 1.928465861474899e-05,
+      "loss": 1.0397,
+      "step": 720
+    },
+    {
+      "epoch": 0.8142777467930842,
+      "grad_norm": 0.7330529689788818,
+      "learning_rate": 1.926433042894799e-05,
+      "loss": 1.0294,
+      "step": 730
+    },
+    {
+      "epoch": 0.8254322364751813,
+      "grad_norm": 1.0434292554855347,
+      "learning_rate": 1.9243728459614006e-05,
+      "loss": 1.0274,
+      "step": 740
+    },
+    {
+      "epoch": 0.8365867261572784,
+      "grad_norm": 0.5535615682601929,
+      "learning_rate": 1.922285331558541e-05,
+      "loss": 1.0717,
+      "step": 750
+    },
+    {
+      "epoch": 0.8477412158393753,
+      "grad_norm": 1.111730694770813,
+      "learning_rate": 1.9201705613773575e-05,
+      "loss": 1.0248,
+      "step": 760
+    },
+    {
+      "epoch": 0.8588957055214724,
+      "grad_norm": 0.8851047158241272,
+      "learning_rate": 1.9180285979144594e-05,
+      "loss": 1.1022,
+      "step": 770
+    },
+    {
+      "epoch": 0.8700501952035694,
+      "grad_norm": 0.5645832419395447,
+      "learning_rate": 1.9158595044700862e-05,
+      "loss": 1.0598,
+      "step": 780
+    },
+    {
+      "epoch": 0.8812046848856665,
+      "grad_norm": 0.536708414554596,
+      "learning_rate": 1.913663345146232e-05,
+      "loss": 1.0778,
+      "step": 790
+    },
+    {
+      "epoch": 0.8923591745677635,
+      "grad_norm": 0.7798624038696289,
+      "learning_rate": 1.911440184844756e-05,
+      "loss": 1.0878,
+      "step": 800
+    },
+    {
+      "epoch": 0.9035136642498606,
+      "grad_norm": 0.5899933576583862,
+      "learning_rate": 1.9091900892654605e-05,
+      "loss": 1.0836,
+      "step": 810
+    },
+    {
+      "epoch": 0.9146681539319577,
+      "grad_norm": 1.4796583652496338,
+      "learning_rate": 1.9069131249041515e-05,
+      "loss": 1.0304,
+      "step": 820
+    },
+    {
+      "epoch": 0.9258226436140546,
+      "grad_norm": 0.5310823321342468,
+      "learning_rate": 1.9046093590506727e-05,
+      "loss": 1.0294,
+      "step": 830
+    },
+    {
+      "epoch": 0.9369771332961517,
+      "grad_norm": 3.327803373336792,
+      "learning_rate": 1.9022788597869174e-05,
+      "loss": 1.0093,
+      "step": 840
+    },
+    {
+      "epoch": 0.9481316229782487,
+      "grad_norm": 0.5713508129119873,
+      "learning_rate": 1.8999216959848154e-05,
+      "loss": 1.1078,
+      "step": 850
+    },
+    {
+      "epoch": 0.9592861126603458,
+      "grad_norm": 1.0793242454528809,
+      "learning_rate": 1.8975379373043004e-05,
+      "loss": 1.0149,
+      "step": 860
+    },
+    {
+      "epoch": 0.9704406023424428,
+      "grad_norm": 0.6370890140533447,
+      "learning_rate": 1.895127654191248e-05,
+      "loss": 1.0509,
+      "step": 870
+    },
+    {
+      "epoch": 0.9815950920245399,
+      "grad_norm": 0.4927578270435333,
+      "learning_rate": 1.892690917875396e-05,
+      "loss": 1.0758,
+      "step": 880
+    },
+    {
+      "epoch": 0.992749581706637,
+      "grad_norm": 0.6699560880661011,
+      "learning_rate": 1.8902278003682384e-05,
+      "loss": 1.0646,
+      "step": 890
+    },
+    {
+      "epoch": 1.003904071388734,
+      "grad_norm": 0.6143092513084412,
+      "learning_rate": 1.8877383744608978e-05,
+      "loss": 1.043,
+      "step": 900
+    },
+    {
+      "epoch": 1.015058561070831,
+      "grad_norm": 0.9502522945404053,
+      "learning_rate": 1.885222713721975e-05,
+      "loss": 1.0772,
+      "step": 910
+    },
+    {
+      "epoch": 1.0262130507529281,
+      "grad_norm": 1.8059970140457153,
+      "learning_rate": 1.8826808924953727e-05,
+      "loss": 1.0019,
+      "step": 920
+    },
+    {
+      "epoch": 1.037367540435025,
+      "grad_norm": 0.6168321967124939,
+      "learning_rate": 1.8801129858981002e-05,
+      "loss": 1.0157,
+      "step": 930
+    },
+    {
+      "epoch": 1.048522030117122,
+      "grad_norm": 0.572365939617157,
+      "learning_rate": 1.8775190698180533e-05,
+      "loss": 1.0131,
+      "step": 940
+    },
+    {
+      "epoch": 1.0596765197992193,
+      "grad_norm": 0.832007884979248,
+      "learning_rate": 1.8748992209117707e-05,
+      "loss": 1.1186,
+      "step": 950
+    },
+    {
+      "epoch": 1.0708310094813163,
+      "grad_norm": 0.5764344334602356,
+      "learning_rate": 1.8722535166021712e-05,
+      "loss": 1.0064,
+      "step": 960
+    },
+    {
+      "epoch": 1.0819854991634132,
+      "grad_norm": 0.6819553971290588,
+      "learning_rate": 1.8695820350762608e-05,
+      "loss": 1.0168,
+      "step": 970
+    },
+    {
+      "epoch": 1.0931399888455102,
+      "grad_norm": 0.5008575916290283,
+      "learning_rate": 1.8668848552828272e-05,
+      "loss": 0.9948,
+      "step": 980
+    },
+    {
+      "epoch": 1.1042944785276074,
+      "grad_norm": 0.5437910556793213,
+      "learning_rate": 1.864162056930104e-05,
+      "loss": 1.0749,
+      "step": 990
+    },
+    {
+      "epoch": 1.1154489682097044,
+      "grad_norm": 0.5878276824951172,
+      "learning_rate": 1.8614137204834152e-05,
+      "loss": 1.0363,
+      "step": 1000
+    },
+    {
+      "epoch": 1.1266034578918014,
+      "grad_norm": 0.6696128845214844,
+      "learning_rate": 1.8586399271627985e-05,
+      "loss": 0.9949,
+      "step": 1010
+    },
+    {
+      "epoch": 1.1377579475738986,
+      "grad_norm": 0.7180065512657166,
+      "learning_rate": 1.855840758940603e-05,
+      "loss": 1.0338,
+      "step": 1020
+    },
+    {
+      "epoch": 1.1489124372559956,
+      "grad_norm": 0.5862113833427429,
+      "learning_rate": 1.8530162985390697e-05,
+      "loss": 0.9799,
+      "step": 1030
+    },
+    {
+      "epoch": 1.1600669269380925,
+      "grad_norm": 0.5569765567779541,
+      "learning_rate": 1.8501666294278832e-05,
+      "loss": 1.0163,
+      "step": 1040
+    },
+    {
+      "epoch": 1.1712214166201895,
+      "grad_norm": 0.5795998573303223,
+      "learning_rate": 1.8472918358217084e-05,
+      "loss": 1.0079,
+      "step": 1050
+    },
+    {
+      "epoch": 1.1823759063022867,
+      "grad_norm": 1.1413137912750244,
+      "learning_rate": 1.8443920026776994e-05,
+      "loss": 1.0019,
+      "step": 1060
+    },
+    {
+      "epoch": 1.1935303959843837,
+      "grad_norm": 0.6494612097740173,
+      "learning_rate": 1.84146721569299e-05,
+      "loss": 1.0275,
+      "step": 1070
+    },
+    {
+      "epoch": 1.2046848856664807,
+      "grad_norm": 0.5971665978431702,
+      "learning_rate": 1.8385175613021603e-05,
+      "loss": 1.0442,
+      "step": 1080
+    },
+    {
+      "epoch": 1.2158393753485779,
+      "grad_norm": 0.5472281575202942,
+      "learning_rate": 1.835543126674684e-05,
+      "loss": 0.9969,
+      "step": 1090
+    },
+    {
+      "epoch": 1.2269938650306749,
+      "grad_norm": 0.9368584156036377,
+      "learning_rate": 1.832543999712349e-05,
+      "loss": 1.0235,
+      "step": 1100
+    },
+    {
+      "epoch": 1.2381483547127718,
+      "grad_norm": 0.6323676705360413,
+      "learning_rate": 1.8295202690466648e-05,
+      "loss": 1.0082,
+      "step": 1110
+    },
+    {
+      "epoch": 1.2493028443948688,
+      "grad_norm": 0.5201751589775085,
+      "learning_rate": 1.8264720240362376e-05,
+      "loss": 1.0199,
+      "step": 1120
+    },
+    {
+      "epoch": 1.260457334076966,
+      "grad_norm": 0.6257196068763733,
+      "learning_rate": 1.823399354764134e-05,
+      "loss": 1.0356,
+      "step": 1130
+    },
+    {
+      "epoch": 1.271611823759063,
+      "grad_norm": 0.517178475856781,
+      "learning_rate": 1.8203023520352155e-05,
+      "loss": 1.0162,
+      "step": 1140
+    },
+    {
+      "epoch": 1.28276631344116,
+      "grad_norm": 0.5029575228691101,
+      "learning_rate": 1.817181107373458e-05,
+      "loss": 1.0381,
+      "step": 1150
+    },
+    {
+      "epoch": 1.2939208031232572,
+      "grad_norm": 0.49766603112220764,
+      "learning_rate": 1.8140357130192443e-05,
+      "loss": 0.9934,
+      "step": 1160
+    },
+    {
+      "epoch": 1.3050752928053542,
+      "grad_norm": 0.5907878875732422,
+      "learning_rate": 1.8108662619266405e-05,
+      "loss": 0.9735,
+      "step": 1170
+    },
+    {
+      "epoch": 1.3162297824874511,
+      "grad_norm": 0.5396254062652588,
+      "learning_rate": 1.8076728477606476e-05,
+      "loss": 1.0221,
+      "step": 1180
+    },
+    {
+      "epoch": 1.3273842721695481,
+      "grad_norm": 0.5513020753860474,
+      "learning_rate": 1.8044555648944335e-05,
+      "loss": 1.0379,
+      "step": 1190
+    },
+    {
+      "epoch": 1.3385387618516453,
+      "grad_norm": 0.6715849041938782,
+      "learning_rate": 1.8012145084065445e-05,
+      "loss": 1.0168,
+      "step": 1200
+    },
+    {
+      "epoch": 1.3496932515337423,
+      "grad_norm": 0.949293315410614,
+      "learning_rate": 1.7979497740780953e-05,
+      "loss": 0.9649,
+      "step": 1210
+    },
+    {
+      "epoch": 1.3608477412158395,
+      "grad_norm": 0.6600781679153442,
+      "learning_rate": 1.794661458389938e-05,
+      "loss": 1.0238,
+      "step": 1220
+    },
+    {
+      "epoch": 1.3720022308979365,
+      "grad_norm": 0.811071515083313,
+      "learning_rate": 1.791349658519813e-05,
+      "loss": 1.0417,
+      "step": 1230
+    },
+    {
+      "epoch": 1.3831567205800335,
+      "grad_norm": 0.47761473059654236,
+      "learning_rate": 1.7880144723394735e-05,
+      "loss": 1.0288,
+      "step": 1240
+    },
+    {
+      "epoch": 1.3943112102621305,
+      "grad_norm": 0.627876341342926,
+      "learning_rate": 1.7846559984117963e-05,
+      "loss": 0.989,
+      "step": 1250
+    },
+    {
+      "epoch": 1.4054656999442274,
+      "grad_norm": 0.7207409739494324,
+      "learning_rate": 1.7812743359878672e-05,
+      "loss": 0.9831,
+      "step": 1260
+    },
+    {
+      "epoch": 1.4166201896263246,
+      "grad_norm": 0.7529335618019104,
+      "learning_rate": 1.7778695850040493e-05,
+      "loss": 0.9723,
+      "step": 1270
+    },
+    {
+      "epoch": 1.4277746793084216,
+      "grad_norm": 0.5672315359115601,
+      "learning_rate": 1.7744418460790282e-05,
+      "loss": 1.0139,
+      "step": 1280
+    },
+    {
+      "epoch": 1.4389291689905188,
+      "grad_norm": 1.991326928138733,
+      "learning_rate": 1.77099122051084e-05,
+      "loss": 0.9925,
+      "step": 1290
+    },
+    {
+      "epoch": 1.4500836586726158,
+      "grad_norm": 0.7485338449478149,
+      "learning_rate": 1.767517810273875e-05,
+      "loss": 1.0279,
+      "step": 1300
+    },
+    {
+      "epoch": 1.4612381483547128,
+      "grad_norm": 0.5592474937438965,
+      "learning_rate": 1.7640217180158688e-05,
+      "loss": 1.0485,
+      "step": 1310
+    },
+    {
+      "epoch": 1.4723926380368098,
+      "grad_norm": 0.5996952652931213,
+      "learning_rate": 1.7605030470548632e-05,
+      "loss": 1.0164,
+      "step": 1320
+    },
+    {
+      "epoch": 1.4835471277189067,
+      "grad_norm": 0.5637550354003906,
+      "learning_rate": 1.756961901376158e-05,
+      "loss": 1.0138,
+      "step": 1330
+    },
+    {
+      "epoch": 1.494701617401004,
+      "grad_norm": 0.6679148077964783,
+      "learning_rate": 1.7533983856292337e-05,
+      "loss": 1.0139,
+      "step": 1340
+    },
+    {
+      "epoch": 1.505856107083101,
+      "grad_norm": 0.5069565773010254,
+      "learning_rate": 1.749812605124662e-05,
+      "loss": 1.0234,
+      "step": 1350
+    },
+    {
+      "epoch": 1.5170105967651981,
+      "grad_norm": 0.6209201216697693,
+      "learning_rate": 1.7462046658309923e-05,
+      "loss": 0.9937,
+      "step": 1360
+    },
+    {
+      "epoch": 1.528165086447295,
+      "grad_norm": 0.5736936926841736,
+      "learning_rate": 1.74257467437162e-05,
+      "loss": 0.9878,
+      "step": 1370
+    },
+    {
+      "epoch": 1.539319576129392,
+      "grad_norm": 0.5310850739479065,
+      "learning_rate": 1.7389227380216353e-05,
+      "loss": 0.9825,
+      "step": 1380
+    },
+    {
+      "epoch": 1.550474065811489,
+      "grad_norm": 0.48725244402885437,
+      "learning_rate": 1.735248964704654e-05,
+      "loss": 1.041,
+      "step": 1390
+    },
+    {
+      "epoch": 1.561628555493586,
+      "grad_norm": 0.5187913775444031,
+      "learning_rate": 1.7315534629896282e-05,
+      "loss": 1.0171,
+      "step": 1400
+    },
+    {
+      "epoch": 1.5727830451756832,
+      "grad_norm": 0.5149972438812256,
+      "learning_rate": 1.7278363420876346e-05,
+      "loss": 1.0164,
+      "step": 1410
+    },
+    {
+      "epoch": 1.5839375348577802,
+      "grad_norm": 0.5611596703529358,
+      "learning_rate": 1.7240977118486523e-05,
+      "loss": 1.006,
+      "step": 1420
+    },
+    {
+      "epoch": 1.5950920245398774,
+      "grad_norm": 0.5297011137008667,
+      "learning_rate": 1.720337682758312e-05,
+      "loss": 0.9757,
+      "step": 1430
+    },
+    {
+      "epoch": 1.6062465142219744,
+      "grad_norm": 0.7282742857933044,
+      "learning_rate": 1.7165563659346336e-05,
+      "loss": 1.0052,
+      "step": 1440
+    },
+    {
+      "epoch": 1.6174010039040714,
+      "grad_norm": 0.6049071550369263,
+      "learning_rate": 1.7127538731247405e-05,
+      "loss": 0.9891,
+      "step": 1450
+    },
+    {
+      "epoch": 1.6285554935861684,
+      "grad_norm": 0.7181625366210938,
+      "learning_rate": 1.7089303167015582e-05,
+      "loss": 0.9877,
+      "step": 1460
+    },
+    {
+      "epoch": 1.6397099832682653,
+      "grad_norm": 0.7313211560249329,
+      "learning_rate": 1.7050858096604943e-05,
+      "loss": 0.9828,
+      "step": 1470
+    },
+    {
+      "epoch": 1.6508644729503625,
+      "grad_norm": 0.49667495489120483,
+      "learning_rate": 1.7012204656160968e-05,
+      "loss": 1.0058,
+      "step": 1480
+    },
+    {
+      "epoch": 1.6620189626324595,
+      "grad_norm": 2.2346487045288086,
+      "learning_rate": 1.697334398798699e-05,
+      "loss": 0.9777,
+      "step": 1490
+    },
+    {
+      "epoch": 1.6731734523145567,
+      "grad_norm": 0.6199758052825928,
+      "learning_rate": 1.693427724051042e-05,
+      "loss": 0.9339,
+      "step": 1500
+    },
+    {
+      "epoch": 1.6843279419966537,
+      "grad_norm": 0.5861572027206421,
+      "learning_rate": 1.6895005568248818e-05,
+      "loss": 1.0539,
+      "step": 1510
+    },
+    {
+      "epoch": 1.6954824316787507,
+      "grad_norm": 0.5904344320297241,
+      "learning_rate": 1.6855530131775765e-05,
+      "loss": 1.0419,
+      "step": 1520
+    },
+    {
+      "epoch": 1.7066369213608477,
+      "grad_norm": 0.5856732130050659,
+      "learning_rate": 1.6815852097686577e-05,
+      "loss": 0.9253,
+      "step": 1530
+    },
+    {
+      "epoch": 1.7177914110429446,
+      "grad_norm": 0.5267894864082336,
+      "learning_rate": 1.677597263856382e-05,
+      "loss": 0.958,
+      "step": 1540
+    },
+    {
+      "epoch": 1.7289459007250418,
+      "grad_norm": 3.836613178253174,
+      "learning_rate": 1.673589293294267e-05,
+      "loss": 0.9697,
+      "step": 1550
+    },
+    {
+      "epoch": 1.7401003904071388,
+      "grad_norm": 0.5841746926307678,
+      "learning_rate": 1.6695614165276052e-05,
+      "loss": 1.011,
+      "step": 1560
+    },
+    {
+      "epoch": 1.751254880089236,
+      "grad_norm": 0.6810708045959473,
+      "learning_rate": 1.665513752589968e-05,
+      "loss": 0.9918,
+      "step": 1570
+    },
+    {
+      "epoch": 1.762409369771333,
+      "grad_norm": 0.537075936794281,
+      "learning_rate": 1.6614464210996856e-05,
+      "loss": 0.9936,
+      "step": 1580
+    },
+    {
+      "epoch": 1.77356385945343,
+      "grad_norm": 0.53858482837677,
+      "learning_rate": 1.6573595422563117e-05,
+      "loss": 1.0108,
+      "step": 1590
+    },
+    {
+      "epoch": 1.784718349135527,
+      "grad_norm": 0.8306734561920166,
+      "learning_rate": 1.653253236837072e-05,
+      "loss": 0.9351,
+      "step": 1600
+    },
+    {
+      "epoch": 1.795872838817624,
+      "grad_norm": 1.4435315132141113,
+      "learning_rate": 1.6491276261932952e-05,
+      "loss": 0.9706,
+      "step": 1610
+    },
+    {
+      "epoch": 1.8070273284997211,
+      "grad_norm": 0.6465569138526917,
+      "learning_rate": 1.6449828322468258e-05,
+      "loss": 1.0211,
+      "step": 1620
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 0.5587633848190308,
+      "learning_rate": 1.640818977486423e-05,
+      "loss": 0.9857,
+      "step": 1630
+    },
+    {
+      "epoch": 1.8293363078639153,
+      "grad_norm": 0.5517516732215881,
+      "learning_rate": 1.636636184964137e-05,
+      "loss": 0.9574,
+      "step": 1640
+    },
+    {
+      "epoch": 1.8404907975460123,
+      "grad_norm": 0.5436330437660217,
+      "learning_rate": 1.632434578291678e-05,
+      "loss": 0.9546,
+      "step": 1650
+    },
+    {
+      "epoch": 1.8516452872281093,
+      "grad_norm": 0.5434009432792664,
+      "learning_rate": 1.6282142816367582e-05,
+      "loss": 0.99,
+      "step": 1660
+    },
+    {
+      "epoch": 1.8627997769102063,
+      "grad_norm": 0.6184738874435425,
+      "learning_rate": 1.6239754197194245e-05,
+      "loss": 0.9811,
+      "step": 1670
+    },
+    {
+      "epoch": 1.8739542665923032,
+      "grad_norm": 0.5969653725624084,
+      "learning_rate": 1.6197181178083726e-05,
+      "loss": 0.9942,
+      "step": 1680
+    },
+    {
+      "epoch": 1.8851087562744004,
+      "grad_norm": 0.5081932544708252,
+      "learning_rate": 1.6154425017172462e-05,
+      "loss": 1.0081,
+      "step": 1690
+    },
+    {
+      "epoch": 1.8962632459564976,
+      "grad_norm": 0.5772440433502197,
+      "learning_rate": 1.6111486978009162e-05,
+      "loss": 0.9536,
+      "step": 1700
+    },
+    {
+      "epoch": 1.9074177356385946,
+      "grad_norm": 0.7058711647987366,
+      "learning_rate": 1.606836832951748e-05,
+      "loss": 0.9602,
+      "step": 1710
+    },
+    {
+      "epoch": 1.9185722253206916,
+      "grad_norm": 0.5237627029418945,
+      "learning_rate": 1.6025070345958523e-05,
+      "loss": 0.9813,
+      "step": 1720
+    },
+    {
+      "epoch": 1.9297267150027886,
+      "grad_norm": 0.5791120529174805,
+      "learning_rate": 1.5981594306893186e-05,
+      "loss": 0.9906,
+      "step": 1730
+    },
+    {
+      "epoch": 1.9408812046848856,
+      "grad_norm": 0.5721683502197266,
+      "learning_rate": 1.593794149714433e-05,
+      "loss": 0.9907,
+      "step": 1740
+    },
+    {
+      "epoch": 1.9520356943669828,
+      "grad_norm": 0.5660907030105591,
+      "learning_rate": 1.589411320675883e-05,
+      "loss": 0.9786,
+      "step": 1750
+    },
+    {
+      "epoch": 1.9631901840490797,
+      "grad_norm": 0.6208905577659607,
+      "learning_rate": 1.5850110730969433e-05,
+      "loss": 0.9764,
+      "step": 1760
+    },
+    {
+      "epoch": 1.974344673731177,
+      "grad_norm": 0.5601891279220581,
+      "learning_rate": 1.5805935370156494e-05,
+      "loss": 0.9596,
+      "step": 1770
+    },
+    {
+      "epoch": 1.985499163413274,
+      "grad_norm": 0.4851455092430115,
+      "learning_rate": 1.5761588429809544e-05,
+      "loss": 0.9706,
+      "step": 1780
+    },
+    {
+      "epoch": 1.996653653095371,
+      "grad_norm": 0.5879290103912354,
+      "learning_rate": 1.57170712204887e-05,
+      "loss": 0.9856,
+      "step": 1790
+    },
+    {
+      "epoch": 2.007808142777468,
+      "grad_norm": 0.5131300687789917,
+      "learning_rate": 1.5672385057785948e-05,
+      "loss": 0.9965,
+      "step": 1800
+    },
+    {
+      "epoch": 2.018962632459565,
+      "grad_norm": 0.6118858456611633,
+      "learning_rate": 1.5627531262286253e-05,
+      "loss": 0.976,
+      "step": 1810
+    },
+    {
+      "epoch": 2.030117122141662,
+      "grad_norm": 0.6357288956642151,
+      "learning_rate": 1.5582511159528544e-05,
+      "loss": 0.968,
+      "step": 1820
+    },
+    {
+      "epoch": 2.0412716118237593,
+      "grad_norm": 0.5152722597122192,
+      "learning_rate": 1.553732607996653e-05,
+      "loss": 0.9813,
+      "step": 1830
+    },
+    {
+      "epoch": 2.0524261015058562,
+      "grad_norm": 0.5387617945671082,
+      "learning_rate": 1.5491977358929382e-05,
+      "loss": 0.9899,
+      "step": 1840
+    },
+    {
+      "epoch": 2.0635805911879532,
+      "grad_norm": 0.5080958604812622,
+      "learning_rate": 1.544646633658229e-05,
+      "loss": 0.9904,
+      "step": 1850
+    },
+    {
+      "epoch": 2.07473508087005,
+      "grad_norm": 0.6057727932929993,
+      "learning_rate": 1.5400794357886823e-05,
+      "loss": 0.9732,
+      "step": 1860
+    },
+    {
+      "epoch": 2.085889570552147,
+      "grad_norm": 0.7106322646141052,
+      "learning_rate": 1.5354962772561218e-05,
+      "loss": 0.9706,
+      "step": 1870
+    },
+    {
+      "epoch": 2.097044060234244,
+      "grad_norm": 1.0154658555984497,
+      "learning_rate": 1.5308972935040472e-05,
+      "loss": 0.9527,
+      "step": 1880
+    },
+    {
+      "epoch": 2.108198549916341,
+      "grad_norm": 0.6521228551864624,
+      "learning_rate": 1.5262826204436325e-05,
+      "loss": 0.9869,
+      "step": 1890
+    },
+    {
+      "epoch": 2.1193530395984386,
+      "grad_norm": 0.5595114827156067,
+      "learning_rate": 1.5216523944497077e-05,
+      "loss": 0.9844,
+      "step": 1900
+    },
+    {
+      "epoch": 2.1305075292805356,
+      "grad_norm": 0.7357882857322693,
+      "learning_rate": 1.5170067523567303e-05,
+      "loss": 0.951,
+      "step": 1910
+    },
+    {
+      "epoch": 2.1416620189626325,
+      "grad_norm": 0.7726870179176331,
+      "learning_rate": 1.5123458314547422e-05,
+      "loss": 0.9568,
+      "step": 1920
+    },
+    {
+      "epoch": 2.1528165086447295,
+      "grad_norm": 0.5238220691680908,
+      "learning_rate": 1.50766976948531e-05,
+      "loss": 0.9825,
+      "step": 1930
+    },
+    {
+      "epoch": 2.1639709983268265,
+      "grad_norm": 0.5988953113555908,
+      "learning_rate": 1.5029787046374562e-05,
+      "loss": 0.9701,
+      "step": 1940
+    },
+    {
+      "epoch": 2.1751254880089235,
+      "grad_norm": 0.5412775278091431,
+      "learning_rate": 1.4982727755435755e-05,
+      "loss": 0.9319,
+      "step": 1950
+    },
+    {
+      "epoch": 2.1862799776910204,
+      "grad_norm": 0.5029479265213013,
+      "learning_rate": 1.4935521212753364e-05,
+      "loss": 0.9665,
+      "step": 1960
+    },
+    {
+      "epoch": 2.197434467373118,
+      "grad_norm": 0.5240495800971985,
+      "learning_rate": 1.488816881339572e-05,
+      "loss": 0.9788,
+      "step": 1970
+    },
+    {
+      "epoch": 2.208588957055215,
+      "grad_norm": 0.5389825701713562,
+      "learning_rate": 1.4840671956741589e-05,
+      "loss": 0.9754,
+      "step": 1980
+    },
+    {
+      "epoch": 2.219743446737312,
+      "grad_norm": 0.5244821310043335,
+      "learning_rate": 1.4793032046438783e-05,
+      "loss": 0.9618,
+      "step": 1990
+    },
+    {
+      "epoch": 2.230897936419409,
+      "grad_norm": 0.5422099828720093,
+      "learning_rate": 1.474525049036271e-05,
+      "loss": 0.9568,
+      "step": 2000
+    },
+    {
+      "epoch": 2.242052426101506,
+      "grad_norm": 0.6588592529296875,
+      "learning_rate": 1.4697328700574756e-05,
+      "loss": 0.9553,
+      "step": 2010
+    },
+    {
+      "epoch": 2.2532069157836028,
+      "grad_norm": 0.5385764241218567,
+      "learning_rate": 1.4649268093280552e-05,
+      "loss": 0.9679,
+      "step": 2020
+    },
+    {
+      "epoch": 2.2643614054656998,
+      "grad_norm": 0.5763164162635803,
+      "learning_rate": 1.4601070088788114e-05,
+      "loss": 0.9802,
+      "step": 2030
+    },
+    {
+      "epoch": 2.275515895147797,
+      "grad_norm": 0.5320225358009338,
+      "learning_rate": 1.45527361114659e-05,
+      "loss": 0.952,
+      "step": 2040
+    },
+    {
+      "epoch": 2.286670384829894,
+      "grad_norm": 0.5828911662101746,
+      "learning_rate": 1.4504267589700683e-05,
+      "loss": 0.9588,
+      "step": 2050
+    },
+    {
+      "epoch": 2.297824874511991,
+      "grad_norm": 0.5583097338676453,
+      "learning_rate": 1.4455665955855351e-05,
+      "loss": 0.9397,
+      "step": 2060
+    },
+    {
+      "epoch": 2.308979364194088,
+      "grad_norm": 0.5835484862327576,
+      "learning_rate": 1.440693264622659e-05,
+      "loss": 0.9765,
+      "step": 2070
+    },
+    {
+      "epoch": 2.320133853876185,
+      "grad_norm": 0.5030141472816467,
+      "learning_rate": 1.4358069101002413e-05,
+      "loss": 0.96,
+      "step": 2080
+    },
+    {
+      "epoch": 2.331288343558282,
+      "grad_norm": 1.1575989723205566,
+      "learning_rate": 1.4309076764219621e-05,
+      "loss": 0.9961,
+      "step": 2090
+    },
+    {
+      "epoch": 2.342442833240379,
+      "grad_norm": 0.5824646353721619,
+      "learning_rate": 1.4259957083721112e-05,
+      "loss": 0.9633,
+      "step": 2100
+    },
+    {
+      "epoch": 2.3535973229224765,
+      "grad_norm": 0.5903834104537964,
+      "learning_rate": 1.421071151111311e-05,
+      "loss": 0.9618,
+      "step": 2110
+    },
+    {
+      "epoch": 2.3647518126045735,
+      "grad_norm": 0.5352372527122498,
+      "learning_rate": 1.4161341501722251e-05,
+      "loss": 0.9595,
+      "step": 2120
+    },
+    {
+      "epoch": 2.3759063022866704,
+      "grad_norm": 0.5336405038833618,
+      "learning_rate": 1.4111848514552582e-05,
+      "loss": 0.9593,
+      "step": 2130
+    },
+    {
+      "epoch": 2.3870607919687674,
+      "grad_norm": 0.5585698485374451,
+      "learning_rate": 1.4062234012242444e-05,
+      "loss": 0.9849,
+      "step": 2140
+    },
+    {
+      "epoch": 2.3982152816508644,
+      "grad_norm": 161.10671997070312,
+      "learning_rate": 1.4012499461021243e-05,
+      "loss": 1.0208,
+      "step": 2150
+    },
+    {
+      "epoch": 2.4093697713329614,
+      "grad_norm": 0.5689246654510498,
+      "learning_rate": 1.396264633066613e-05,
+      "loss": 0.9939,
+      "step": 2160
+    },
+    {
+      "epoch": 2.420524261015059,
+      "grad_norm": 0.5657776594161987,
+      "learning_rate": 1.3912676094458552e-05,
+      "loss": 0.966,
+      "step": 2170
+    },
+    {
+      "epoch": 2.4316787506971558,
+      "grad_norm": 0.5576151013374329,
+      "learning_rate": 1.386259022914072e-05,
+      "loss": 0.9287,
+      "step": 2180
+    },
+    {
+      "epoch": 2.4428332403792528,
+      "grad_norm": 0.5689704418182373,
+      "learning_rate": 1.3812390214871973e-05,
+      "loss": 0.9863,
+      "step": 2190
+    },
+    {
+      "epoch": 2.4539877300613497,
+      "grad_norm": 0.5151479840278625,
+      "learning_rate": 1.3762077535185022e-05,
+      "loss": 0.9665,
+      "step": 2200
+    },
+    {
+      "epoch": 2.4651422197434467,
+      "grad_norm": 0.6008809804916382,
+      "learning_rate": 1.3711653676942115e-05,
+      "loss": 0.933,
+      "step": 2210
+    },
+    {
+      "epoch": 2.4762967094255437,
+      "grad_norm": 0.6128340363502502,
+      "learning_rate": 1.3661120130291106e-05,
+      "loss": 0.9411,
+      "step": 2220
+    },
+    {
+      "epoch": 2.4874511991076407,
+      "grad_norm": 0.5954192280769348,
+      "learning_rate": 1.3610478388621402e-05,
+      "loss": 0.9547,
+      "step": 2230
+    },
+    {
+      "epoch": 2.4986056887897377,
+      "grad_norm": 0.5410999059677124,
+      "learning_rate": 1.3559729948519835e-05,
+      "loss": 1.002,
+      "step": 2240
+    },
+    {
+      "epoch": 2.509760178471835,
+      "grad_norm": 0.5633314251899719,
+      "learning_rate": 1.3508876309726438e-05,
+      "loss": 0.9347,
+      "step": 2250
+    },
+    {
+      "epoch": 2.520914668153932,
+      "grad_norm": 0.701754629611969,
+      "learning_rate": 1.3457918975090127e-05,
+      "loss": 0.9471,
+      "step": 2260
+    },
+    {
+      "epoch": 2.532069157836029,
+      "grad_norm": 0.5823484659194946,
+      "learning_rate": 1.3406859450524271e-05,
+      "loss": 0.9618,
+      "step": 2270
+    },
+    {
+      "epoch": 2.543223647518126,
+      "grad_norm": 0.6154325008392334,
+      "learning_rate": 1.335569924496221e-05,
+      "loss": 0.9747,
+      "step": 2280
+    },
+    {
+      "epoch": 2.554378137200223,
+      "grad_norm": 0.5738359093666077,
+      "learning_rate": 1.3304439870312651e-05,
+      "loss": 0.9645,
+      "step": 2290
+    },
+    {
+      "epoch": 2.56553262688232,
+      "grad_norm": 0.7494403719902039,
+      "learning_rate": 1.3253082841414984e-05,
+      "loss": 0.9645,
+      "step": 2300
+    },
+    {
+      "epoch": 2.5766871165644174,
+      "grad_norm": 0.6298063397407532,
+      "learning_rate": 1.3201629675994525e-05,
+      "loss": 0.9645,
+      "step": 2310
+    },
+    {
+      "epoch": 2.5878416062465144,
+      "grad_norm": 0.561741292476654,
+      "learning_rate": 1.3150081894617662e-05,
+      "loss": 0.9951,
+      "step": 2320
+    },
+    {
+      "epoch": 2.5989960959286114,
+      "grad_norm": 0.655669093132019,
+      "learning_rate": 1.3098441020646904e-05,
+      "loss": 0.9809,
+      "step": 2330
+    },
+    {
+      "epoch": 2.6101505856107083,
+      "grad_norm": 0.5869964957237244,
+      "learning_rate": 1.3046708580195882e-05,
+      "loss": 0.968,
+      "step": 2340
+    },
+    {
+      "epoch": 2.6213050752928053,
+      "grad_norm": 0.5096555948257446,
+      "learning_rate": 1.2994886102084235e-05,
+      "loss": 0.982,
+      "step": 2350
+    },
+    {
+      "epoch": 2.6324595649749023,
+      "grad_norm": 0.4704292118549347,
+      "learning_rate": 1.2942975117792436e-05,
+      "loss": 0.9422,
+      "step": 2360
+    },
+    {
+      "epoch": 2.6436140546569993,
+      "grad_norm": 0.6242173314094543,
+      "learning_rate": 1.289097716141653e-05,
+      "loss": 0.984,
+      "step": 2370
+    },
+    {
+      "epoch": 2.6547685443390963,
+      "grad_norm": 0.5799645185470581,
+      "learning_rate": 1.2838893769622804e-05,
+      "loss": 0.9589,
+      "step": 2380
+    },
+    {
+      "epoch": 2.6659230340211937,
+      "grad_norm": 0.49811428785324097,
+      "learning_rate": 1.2786726481602365e-05,
+      "loss": 0.946,
+      "step": 2390
+    },
+    {
+      "epoch": 2.6770775237032907,
+      "grad_norm": 0.5958947539329529,
+      "learning_rate": 1.2734476839025654e-05,
+      "loss": 1.0,
+      "step": 2400
+    },
+    {
+      "epoch": 2.6882320133853876,
+      "grad_norm": 0.5652265548706055,
+      "learning_rate": 1.2682146385996896e-05,
+      "loss": 0.9196,
+      "step": 2410
+    },
+    {
+      "epoch": 2.6993865030674846,
+      "grad_norm": 0.5902454257011414,
+      "learning_rate": 1.2629736669008464e-05,
+      "loss": 0.9854,
+      "step": 2420
+    },
+    {
+      "epoch": 2.7105409927495816,
+      "grad_norm": 1.0020962953567505,
+      "learning_rate": 1.2577249236895166e-05,
+      "loss": 0.9863,
+      "step": 2430
+    },
+    {
+      "epoch": 2.721695482431679,
+      "grad_norm": 0.6766297221183777,
+      "learning_rate": 1.2524685640788488e-05,
+      "loss": 0.9449,
+      "step": 2440
+    },
+    {
+      "epoch": 2.732849972113776,
+      "grad_norm": 0.5196471214294434,
+      "learning_rate": 1.2472047434070743e-05,
+      "loss": 0.9362,
+      "step": 2450
+    },
+    {
+      "epoch": 2.744004461795873,
+      "grad_norm": 0.6119663119316101,
+      "learning_rate": 1.2419336172329175e-05,
+      "loss": 0.9933,
+      "step": 2460
+    },
+    {
+      "epoch": 2.75515895147797,
+      "grad_norm": 0.6461163759231567,
+      "learning_rate": 1.2366553413309971e-05,
+      "loss": 0.9695,
+      "step": 2470
+    },
+    {
+      "epoch": 2.766313441160067,
+      "grad_norm": 0.6933746337890625,
+      "learning_rate": 1.2313700716872253e-05,
+      "loss": 0.8987,
+      "step": 2480
+    },
+    {
+      "epoch": 2.777467930842164,
+      "grad_norm": 0.602975070476532,
+      "learning_rate": 1.2260779644941949e-05,
+      "loss": 0.9538,
+      "step": 2490
+    },
+    {
+      "epoch": 2.788622420524261,
+      "grad_norm": 0.6569437980651855,
+      "learning_rate": 1.2207791761465658e-05,
+      "loss": 0.9537,
+      "step": 2500
+    },
+    {
+      "epoch": 2.799776910206358,
+      "grad_norm": 0.6225653886795044,
+      "learning_rate": 1.2154738632364418e-05,
+      "loss": 0.956,
+      "step": 2510
+    },
+    {
+      "epoch": 2.810931399888455,
+      "grad_norm": 0.6410306692123413,
+      "learning_rate": 1.2101621825487438e-05,
+      "loss": 0.9102,
+      "step": 2520
+    },
+    {
+      "epoch": 2.8220858895705523,
+      "grad_norm": 0.4937409460544586,
+      "learning_rate": 1.2048442910565756e-05,
+      "loss": 0.9149,
+      "step": 2530
+    },
+    {
+      "epoch": 2.8332403792526493,
+      "grad_norm": 0.5088676810264587,
+      "learning_rate": 1.1995203459165866e-05,
+      "loss": 0.9319,
+      "step": 2540
+    },
+    {
+      "epoch": 2.8443948689347462,
+      "grad_norm": 0.5606518983840942,
+      "learning_rate": 1.1941905044643244e-05,
+      "loss": 0.925,
+      "step": 2550
+    },
+    {
+      "epoch": 2.8555493586168432,
+      "grad_norm": 0.541970431804657,
+      "learning_rate": 1.1888549242095881e-05,
+      "loss": 0.9692,
+      "step": 2560
+    },
+    {
+      "epoch": 2.86670384829894,
+      "grad_norm": 0.5640259385108948,
+      "learning_rate": 1.1835137628317728e-05,
+      "loss": 0.9588,
+      "step": 2570
+    },
+    {
+      "epoch": 2.8778583379810376,
+      "grad_norm": 0.7504070401191711,
+      "learning_rate": 1.1781671781752082e-05,
+      "loss": 0.9387,
+      "step": 2580
+    },
+    {
+      "epoch": 2.8890128276631346,
+      "grad_norm": 0.5966542363166809,
+      "learning_rate": 1.1728153282444956e-05,
+      "loss": 0.9601,
+      "step": 2590
+    },
+    {
+      "epoch": 2.9001673173452316,
+      "grad_norm": 0.5044750571250916,
+      "learning_rate": 1.1674583711998386e-05,
+      "loss": 0.9204,
+      "step": 2600
+    },
+    {
+      "epoch": 2.9113218070273286,
+      "grad_norm": 0.5839060544967651,
+      "learning_rate": 1.1620964653523679e-05,
+      "loss": 0.9646,
+      "step": 2610
+    },
+    {
+      "epoch": 2.9224762967094255,
+      "grad_norm": 0.6584033370018005,
+      "learning_rate": 1.1567297691594628e-05,
+      "loss": 0.9856,
+      "step": 2620
+    },
+    {
+      "epoch": 2.9336307863915225,
+      "grad_norm": 0.5905307531356812,
+      "learning_rate": 1.1513584412200702e-05,
+      "loss": 0.972,
+      "step": 2630
+    },
+    {
+      "epoch": 2.9447852760736195,
+      "grad_norm": 0.5503203868865967,
+      "learning_rate": 1.1459826402700159e-05,
+      "loss": 0.9639,
+      "step": 2640
+    },
+    {
+      "epoch": 2.9559397657557165,
+      "grad_norm": 0.6256836652755737,
+      "learning_rate": 1.1406025251773131e-05,
+      "loss": 0.965,
+      "step": 2650
+    },
+    {
+      "epoch": 2.9670942554378135,
+      "grad_norm": 0.4728027284145355,
+      "learning_rate": 1.1352182549374702e-05,
+      "loss": 0.9385,
+      "step": 2660
+    },
+    {
+      "epoch": 2.978248745119911,
+      "grad_norm": 0.5927023887634277,
+      "learning_rate": 1.1298299886687891e-05,
+      "loss": 0.9852,
+      "step": 2670
+    },
+    {
+      "epoch": 2.989403234802008,
+      "grad_norm": 0.4461129307746887,
+      "learning_rate": 1.124437885607664e-05,
+      "loss": 0.9615,
+      "step": 2680
+    },
+    {
+      "epoch": 3.000557724484105,
+      "grad_norm": 0.6086824536323547,
+      "learning_rate": 1.1190421051038766e-05,
+      "loss": 0.9557,
+      "step": 2690
+    },
+    {
+      "epoch": 3.011712214166202,
+      "grad_norm": 0.6160380244255066,
+      "learning_rate": 1.1136428066158852e-05,
+      "loss": 0.9859,
+      "step": 2700
+    },
+    {
+      "epoch": 3.022866703848299,
+      "grad_norm": 0.579237163066864,
+      "learning_rate": 1.1082401497061133e-05,
+      "loss": 0.9641,
+      "step": 2710
+    },
+    {
+      "epoch": 3.034021193530396,
+      "grad_norm": 0.5745778679847717,
+      "learning_rate": 1.102834294036234e-05,
+      "loss": 0.96,
+      "step": 2720
+    },
+    {
+      "epoch": 3.045175683212493,
+      "grad_norm": 0.6007333993911743,
+      "learning_rate": 1.0974253993624514e-05,
+      "loss": 0.9637,
+      "step": 2730
+    },
+    {
+      "epoch": 3.05633017289459,
+      "grad_norm": 0.5479885339736938,
+      "learning_rate": 1.0920136255307801e-05,
+      "loss": 0.9359,
+      "step": 2740
+    },
+    {
+      "epoch": 3.067484662576687,
+      "grad_norm": 0.6558302044868469,
+      "learning_rate": 1.08659913247232e-05,
+      "loss": 0.9709,
+      "step": 2750
+    },
+    {
+      "epoch": 3.078639152258784,
+      "grad_norm": 0.5437296628952026,
+      "learning_rate": 1.0811820801985316e-05,
+      "loss": 0.9479,
+      "step": 2760
+    },
+    {
+      "epoch": 3.089793641940881,
+      "grad_norm": 0.6434791684150696,
+      "learning_rate": 1.0757626287965057e-05,
+      "loss": 0.9421,
+      "step": 2770
+    },
+    {
+      "epoch": 3.100948131622978,
+      "grad_norm": 0.5511924624443054,
+      "learning_rate": 1.0703409384242335e-05,
+      "loss": 0.989,
+      "step": 2780
+    },
+    {
+      "epoch": 3.112102621305075,
+      "grad_norm": 0.6061602234840393,
+      "learning_rate": 1.0649171693058738e-05,
+      "loss": 1.0097,
+      "step": 2790
+    },
+    {
+      "epoch": 3.1232571109871725,
+      "grad_norm": 0.7485038042068481,
+      "learning_rate": 1.0594914817270165e-05,
+      "loss": 0.9886,
+      "step": 2800
+    },
+    {
+      "epoch": 3.1344116006692695,
+      "grad_norm": 0.7252965569496155,
+      "learning_rate": 1.054064036029947e-05,
+      "loss": 0.9489,
+      "step": 2810
+    },
+    {
+      "epoch": 3.1455660903513665,
+      "grad_norm": 0.5706859827041626,
+      "learning_rate": 1.0486349926089077e-05,
+      "loss": 0.98,
+      "step": 2820
+    },
+    {
+      "epoch": 3.1567205800334635,
+      "grad_norm": 0.5560559034347534,
+      "learning_rate": 1.0432045119053572e-05,
+      "loss": 0.9621,
+      "step": 2830
+    },
+    {
+      "epoch": 3.1678750697155604,
+      "grad_norm": 0.5444208383560181,
+      "learning_rate": 1.0377727544032292e-05,
+      "loss": 0.8854,
+      "step": 2840
+    },
+    {
+      "epoch": 3.1790295593976574,
+      "grad_norm": 0.5219970941543579,
+      "learning_rate": 1.0323398806241907e-05,
+      "loss": 0.9418,
+      "step": 2850
+    },
+    {
+      "epoch": 3.190184049079755,
+      "grad_norm": 0.5396592617034912,
+      "learning_rate": 1.0269060511228968e-05,
+      "loss": 0.9515,
+      "step": 2860
+    },
+    {
+      "epoch": 3.201338538761852,
+      "grad_norm": 0.6761412620544434,
+      "learning_rate": 1.0214714264822461e-05,
+      "loss": 0.9301,
+      "step": 2870
+    },
+    {
+      "epoch": 3.212493028443949,
+      "grad_norm": 0.5131610035896301,
+      "learning_rate": 1.0160361673086365e-05,
+      "loss": 0.9679,
+      "step": 2880
+    },
+    {
+      "epoch": 3.2236475181260458,
+      "grad_norm": 0.5312976837158203,
+      "learning_rate": 1.0106004342272176e-05,
+      "loss": 0.975,
+      "step": 2890
+    },
+    {
+      "epoch": 3.2348020078081428,
+      "grad_norm": 0.5575296878814697,
+      "learning_rate": 1.005164387877143e-05,
+      "loss": 0.9748,
+      "step": 2900
+    },
+    {
+      "epoch": 3.2459564974902397,
+      "grad_norm": 0.5677649974822998,
+      "learning_rate": 9.997281889068261e-06,
+      "loss": 0.954,
+      "step": 2910
+    },
+    {
+      "epoch": 3.2571109871723367,
+      "grad_norm": 0.5441506505012512,
+      "learning_rate": 9.94291997969189e-06,
+      "loss": 0.9781,
+      "step": 2920
+    },
+    {
+      "epoch": 3.2682654768544337,
+      "grad_norm": 0.6877113580703735,
+      "learning_rate": 9.888559757169172e-06,
+      "loss": 0.9531,
+      "step": 2930
+    },
+    {
+      "epoch": 3.279419966536531,
+      "grad_norm": 0.5976460576057434,
+      "learning_rate": 9.834202827977107e-06,
+      "loss": 0.9579,
+      "step": 2940
+    },
+    {
+      "epoch": 3.290574456218628,
+      "grad_norm": 0.5610713362693787,
+      "learning_rate": 9.77985079849538e-06,
+      "loss": 0.9615,
+      "step": 2950
+    },
+    {
+      "epoch": 3.301728945900725,
+      "grad_norm": 0.5829696655273438,
+      "learning_rate": 9.725505274958855e-06,
+      "loss": 0.9353,
+      "step": 2960
+    },
+    {
+      "epoch": 3.312883435582822,
+      "grad_norm": 0.5707248449325562,
+      "learning_rate": 9.671167863410156e-06,
+      "loss": 0.9578,
+      "step": 2970
+    },
+    {
+      "epoch": 3.324037925264919,
+      "grad_norm": 0.5084194540977478,
+      "learning_rate": 9.616840169652156e-06,
+      "loss": 0.9283,
+      "step": 2980
+    },
+    {
+      "epoch": 3.335192414947016,
+      "grad_norm": 0.5233070850372314,
+      "learning_rate": 9.562523799200558e-06,
+      "loss": 0.9249,
+      "step": 2990
+    },
+    {
+      "epoch": 3.3463469046291134,
+      "grad_norm": 0.5390496850013733,
+      "learning_rate": 9.508220357236431e-06,
+      "loss": 0.9806,
+      "step": 3000
+    },
+    {
+      "epoch": 3.3575013943112104,
+      "grad_norm": 0.5927379131317139,
+      "learning_rate": 9.453931448558768e-06,
+      "loss": 1.0214,
+      "step": 3010
+    },
+    {
+      "epoch": 3.3686558839933074,
+      "grad_norm": 0.5973530411720276,
+      "learning_rate": 9.39965867753708e-06,
+      "loss": 0.9858,
+      "step": 3020
+    },
+    {
+      "epoch": 3.3798103736754044,
+      "grad_norm": 0.5969598293304443,
+      "learning_rate": 9.345403648063958e-06,
+      "loss": 0.944,
+      "step": 3030
+    },
+    {
+      "epoch": 3.3909648633575014,
+      "grad_norm": 0.5890622735023499,
+      "learning_rate": 9.2911679635077e-06,
+      "loss": 0.9206,
+      "step": 3040
+    },
+    {
+      "epoch": 3.4021193530395983,
+      "grad_norm": 0.552482008934021,
+      "learning_rate": 9.2369532266649e-06,
+      "loss": 0.8879,
+      "step": 3050
+    },
+    {
+      "epoch": 3.4132738427216953,
+      "grad_norm": 0.7639123201370239,
+      "learning_rate": 9.182761039713112e-06,
+      "loss": 0.9069,
+      "step": 3060
+    },
+    {
+      "epoch": 3.4244283324037923,
+      "grad_norm": 0.6415414810180664,
+      "learning_rate": 9.12859300416347e-06,
+      "loss": 0.9366,
+      "step": 3070
+    },
+    {
+      "epoch": 3.4355828220858897,
+      "grad_norm": 0.5764932632446289,
+      "learning_rate": 9.07445072081339e-06,
+      "loss": 0.9503,
+      "step": 3080
+    },
+    {
+      "epoch": 3.4467373117679867,
+      "grad_norm": 0.5804705023765564,
+      "learning_rate": 9.020335789699238e-06,
+      "loss": 0.9067,
+      "step": 3090
+    },
+    {
+      "epoch": 3.4578918014500837,
+      "grad_norm": 0.48473402857780457,
+      "learning_rate": 8.966249810049057e-06,
+      "loss": 0.963,
+      "step": 3100
+    },
+    {
+      "epoch": 3.4690462911321807,
+      "grad_norm": 0.5405654311180115,
+      "learning_rate": 8.91219438023531e-06,
+      "loss": 0.965,
+      "step": 3110
+    },
+    {
+      "epoch": 3.4802007808142776,
+      "grad_norm": 0.5696051716804504,
+      "learning_rate": 8.85817109772762e-06,
+      "loss": 0.9686,
+      "step": 3120
+    },
+    {
+      "epoch": 3.4913552704963746,
+      "grad_norm": 0.45706358551979065,
+      "learning_rate": 8.804181559045609e-06,
+      "loss": 0.9337,
+      "step": 3130
+    },
+    {
+      "epoch": 3.502509760178472,
+      "grad_norm": 0.615371823310852,
+      "learning_rate": 8.750227359711652e-06,
+      "loss": 0.9665,
+      "step": 3140
+    },
+    {
+      "epoch": 3.513664249860569,
+      "grad_norm": 0.6421345472335815,
+      "learning_rate": 8.696310094203785e-06,
+      "loss": 0.9522,
+      "step": 3150
+    },
+    {
+      "epoch": 3.524818739542666,
+      "grad_norm": 0.5607454776763916,
+      "learning_rate": 8.642431355908549e-06,
+      "loss": 0.8975,
+      "step": 3160
+    },
+    {
+      "epoch": 3.535973229224763,
+      "grad_norm": 0.5480626225471497,
+      "learning_rate": 8.58859273707392e-06,
+      "loss": 0.9499,
+      "step": 3170
+    },
+    {
+      "epoch": 3.54712771890686,
+      "grad_norm": 0.6433860659599304,
+      "learning_rate": 8.53479582876223e-06,
+      "loss": 0.9716,
+      "step": 3180
+    },
+    {
+      "epoch": 3.558282208588957,
+      "grad_norm": 0.6692389845848083,
+      "learning_rate": 8.481042220803185e-06,
+      "loss": 0.9205,
+      "step": 3190
+    },
+    {
+      "epoch": 3.569436698271054,
+      "grad_norm": 0.529801070690155,
+      "learning_rate": 8.427333501746841e-06,
+      "loss": 0.9583,
+      "step": 3200
+    },
+    {
+      "epoch": 3.580591187953151,
+      "grad_norm": 0.47987642884254456,
+      "learning_rate": 8.373671258816692e-06,
+      "loss": 0.9565,
+      "step": 3210
+    },
+    {
+      "epoch": 3.5917456776352483,
+      "grad_norm": 0.5379712581634521,
+      "learning_rate": 8.32005707786274e-06,
+      "loss": 0.9489,
+      "step": 3220
+    },
+    {
+      "epoch": 3.6029001673173453,
+      "grad_norm": 0.5486000776290894,
+      "learning_rate": 8.266492543314642e-06,
+      "loss": 0.9547,
+      "step": 3230
+    },
+    {
+      "epoch": 3.6140546569994423,
+      "grad_norm": 0.5590048432350159,
+      "learning_rate": 8.212979238134883e-06,
+      "loss": 0.9159,
+      "step": 3240
+    },
+    {
+      "epoch": 3.6252091466815393,
+      "grad_norm": 0.5172351598739624,
+      "learning_rate": 8.159518743771992e-06,
+      "loss": 0.9404,
+      "step": 3250
+    },
+    {
+      "epoch": 3.6363636363636362,
+      "grad_norm": 0.5739352703094482,
+      "learning_rate": 8.10611264011382e-06,
+      "loss": 0.9201,
+      "step": 3260
+    },
+    {
+      "epoch": 3.6475181260457337,
+      "grad_norm": 0.6838170886039734,
+      "learning_rate": 8.052762505440822e-06,
+      "loss": 0.9567,
+      "step": 3270
+    },
+    {
+      "epoch": 3.6586726157278306,
+      "grad_norm": 0.4739402234554291,
+      "learning_rate": 7.999469916379452e-06,
+      "loss": 0.9166,
+      "step": 3280
+    },
+    {
+      "epoch": 3.6698271054099276,
+      "grad_norm": 0.5572711229324341,
+      "learning_rate": 7.946236447855535e-06,
+      "loss": 0.9301,
+      "step": 3290
+    },
+    {
+      "epoch": 3.6809815950920246,
+      "grad_norm": 0.47544723749160767,
+      "learning_rate": 7.893063673047758e-06,
+      "loss": 0.911,
+      "step": 3300
+    },
+    {
+      "epoch": 3.6921360847741216,
+      "grad_norm": 0.5537176132202148,
+      "learning_rate": 7.839953163341142e-06,
+      "loss": 0.9685,
+      "step": 3310
+    },
+    {
+      "epoch": 3.7032905744562186,
+      "grad_norm": 0.5580974221229553,
+      "learning_rate": 7.786906488280631e-06,
+      "loss": 0.9383,
+      "step": 3320
+    },
+    {
+      "epoch": 3.7144450641383155,
+      "grad_norm": 0.5065183639526367,
+      "learning_rate": 7.73392521552471e-06,
+      "loss": 0.9584,
+      "step": 3330
+    },
+    {
+      "epoch": 3.7255995538204125,
+      "grad_norm": 0.5675489902496338,
+      "learning_rate": 7.681010910799043e-06,
+      "loss": 0.9382,
+      "step": 3340
+    },
+    {
+      "epoch": 3.7367540435025095,
+      "grad_norm": 0.56146240234375,
+      "learning_rate": 7.628165137850252e-06,
+      "loss": 0.8888,
+      "step": 3350
+    },
+    {
+      "epoch": 3.747908533184607,
+      "grad_norm": 0.5241889953613281,
+      "learning_rate": 7.575389458399655e-06,
+      "loss": 0.8833,
+      "step": 3360
+    },
+    {
+      "epoch": 3.759063022866704,
+      "grad_norm": 0.6032584309577942,
+      "learning_rate": 7.522685432097152e-06,
+      "loss": 0.9508,
+      "step": 3370
+    },
+    {
+      "epoch": 3.770217512548801,
+      "grad_norm": 0.5508081912994385,
+      "learning_rate": 7.470054616475109e-06,
+      "loss": 0.9362,
+      "step": 3380
+    },
+    {
+      "epoch": 3.781372002230898,
+      "grad_norm": 0.5076966881752014,
+      "learning_rate": 7.417498566902344e-06,
+      "loss": 0.9168,
+      "step": 3390
+    },
+    {
+      "epoch": 3.792526491912995,
+      "grad_norm": 0.6123913526535034,
+      "learning_rate": 7.365018836538151e-06,
+      "loss": 0.9301,
+      "step": 3400
+    },
+    {
+      "epoch": 3.8036809815950923,
+      "grad_norm": 0.4558752477169037,
+      "learning_rate": 7.312616976286414e-06,
+      "loss": 0.9483,
+      "step": 3410
+    },
+    {
+      "epoch": 3.8148354712771892,
+      "grad_norm": 0.5639483332633972,
+      "learning_rate": 7.260294534749751e-06,
+      "loss": 0.9473,
+      "step": 3420
+    },
+    {
+      "epoch": 3.8259899609592862,
+      "grad_norm": 0.4958498477935791,
+      "learning_rate": 7.208053058183776e-06,
+      "loss": 0.9694,
+      "step": 3430
+    },
+    {
+      "epoch": 3.837144450641383,
+      "grad_norm": 0.5522082448005676,
+      "learning_rate": 7.15589409045139e-06,
+      "loss": 0.925,
+      "step": 3440
+    },
+    {
+      "epoch": 3.84829894032348,
+      "grad_norm": 0.512566328048706,
+      "learning_rate": 7.103819172977147e-06,
+      "loss": 0.8997,
+      "step": 3450
+    },
+    {
+      "epoch": 3.859453430005577,
+      "grad_norm": 0.6051592826843262,
+      "learning_rate": 7.0518298447017255e-06,
+      "loss": 0.966,
+      "step": 3460
+    },
+    {
+      "epoch": 3.870607919687674,
+      "grad_norm": 0.5349409580230713,
+      "learning_rate": 6.99992764203642e-06,
+      "loss": 0.9489,
+      "step": 3470
+    },
+    {
+      "epoch": 3.881762409369771,
+      "grad_norm": 0.568419873714447,
+      "learning_rate": 6.948114098817767e-06,
+      "loss": 0.9379,
+      "step": 3480
+    },
+    {
+      "epoch": 3.892916899051868,
+      "grad_norm": 0.5198797583580017,
+      "learning_rate": 6.8963907462621786e-06,
+      "loss": 0.9258,
+      "step": 3490
+    },
+    {
+      "epoch": 3.9040713887339655,
+      "grad_norm": 0.5569652915000916,
+      "learning_rate": 6.844759112920738e-06,
+      "loss": 0.9267,
+      "step": 3500
+    },
+    {
+      "epoch": 3.9152258784160625,
+      "grad_norm": 0.6207433938980103,
+      "learning_rate": 6.7932207246339845e-06,
+      "loss": 0.9034,
+      "step": 3510
+    },
+    {
+      "epoch": 3.9263803680981595,
+      "grad_norm": 0.8496944904327393,
+      "learning_rate": 6.741777104486843e-06,
+      "loss": 0.9285,
+      "step": 3520
+    },
+    {
+      "epoch": 3.9375348577802565,
+      "grad_norm": 0.5930597186088562,
+      "learning_rate": 6.690429772763616e-06,
+      "loss": 1.0073,
+      "step": 3530
+    },
+    {
+      "epoch": 3.9486893474623534,
+      "grad_norm": 0.5414257049560547,
+      "learning_rate": 6.6391802469030375e-06,
+      "loss": 0.9238,
+      "step": 3540
+    },
+    {
+      "epoch": 3.959843837144451,
+      "grad_norm": 0.5691171288490295,
+      "learning_rate": 6.58803004145345e-06,
+      "loss": 0.9394,
+      "step": 3550
+    },
+    {
+      "epoch": 3.970998326826548,
+      "grad_norm": 0.5480623841285706,
+      "learning_rate": 6.5369806680280275e-06,
+      "loss": 0.9349,
+      "step": 3560
+    },
+    {
+      "epoch": 3.982152816508645,
+      "grad_norm": 0.6629100441932678,
+      "learning_rate": 6.4860336352601225e-06,
+      "loss": 0.981,
+      "step": 3570
+    },
+    {
+      "epoch": 3.993307306190742,
+      "grad_norm": 0.5570813417434692,
+      "learning_rate": 6.435190448758657e-06,
+      "loss": 0.8973,
+      "step": 3580
+    },
+    {
+      "epoch": 4.004461795872839,
+      "grad_norm": 0.6893118619918823,
+      "learning_rate": 6.384452611063656e-06,
+      "loss": 0.9315,
+      "step": 3590
+    },
+    {
+      "epoch": 4.015616285554936,
+      "grad_norm": 0.5445614457130432,
+      "learning_rate": 6.3338216216018235e-06,
+      "loss": 0.9352,
+      "step": 3600
+    },
+    {
+      "epoch": 4.026770775237033,
+      "grad_norm": 0.5540854334831238,
+      "learning_rate": 6.283298976642242e-06,
+      "loss": 0.9668,
+      "step": 3610
+    },
+    {
+      "epoch": 4.03792526491913,
+      "grad_norm": 0.8352875113487244,
+      "learning_rate": 6.232886169252145e-06,
+      "loss": 0.9201,
+      "step": 3620
+    },
+    {
+      "epoch": 4.049079754601227,
+      "grad_norm": 0.5931369662284851,
+      "learning_rate": 6.182584689252804e-06,
+      "loss": 0.9525,
+      "step": 3630
+    },
+    {
+      "epoch": 4.060234244283324,
+      "grad_norm": 0.5365961790084839,
+      "learning_rate": 6.1323960231754955e-06,
+      "loss": 0.9167,
+      "step": 3640
+    },
+    {
+      "epoch": 4.071388733965421,
+      "grad_norm": 0.4611726999282837,
+      "learning_rate": 6.082321654217558e-06,
+      "loss": 0.9409,
+      "step": 3650
+    },
+    {
+      "epoch": 4.0825432236475185,
+      "grad_norm": 0.42769667506217957,
+      "learning_rate": 6.03236306219859e-06,
+      "loss": 0.9358,
+      "step": 3660
+    },
+    {
+      "epoch": 4.0936977133296155,
+      "grad_norm": 0.571824848651886,
+      "learning_rate": 5.982521723516683e-06,
+      "loss": 0.9714,
+      "step": 3670
+    },
+    {
+      "epoch": 4.1048522030117125,
+      "grad_norm": 0.5529113411903381,
+      "learning_rate": 5.932799111104815e-06,
+      "loss": 0.8918,
+      "step": 3680
+    },
+    {
+      "epoch": 4.1160066926938095,
+      "grad_norm": 0.57806795835495,
+      "learning_rate": 5.883196694387312e-06,
+      "loss": 0.9456,
+      "step": 3690
+    },
+    {
+      "epoch": 4.1271611823759065,
+      "grad_norm": 0.5577197670936584,
+      "learning_rate": 5.833715939236428e-06,
+      "loss": 0.9498,
+      "step": 3700
+    },
+    {
+      "epoch": 4.138315672058003,
+      "grad_norm": 0.7209545969963074,
+      "learning_rate": 5.78435830792901e-06,
+      "loss": 0.9183,
+      "step": 3710
+    },
+    {
+      "epoch": 4.1494701617401,
+      "grad_norm": 1.1457955837249756,
+      "learning_rate": 5.7351252591033065e-06,
+      "loss": 0.9535,
+      "step": 3720
+    },
+    {
+      "epoch": 4.160624651422197,
+      "grad_norm": 0.5285237431526184,
+      "learning_rate": 5.68601824771585e-06,
+      "loss": 0.94,
+      "step": 3730
+    },
+    {
+      "epoch": 4.171779141104294,
+      "grad_norm": 0.5437790751457214,
+      "learning_rate": 5.637038724998453e-06,
+      "loss": 0.9005,
+      "step": 3740
+    },
+    {
+      "epoch": 4.182933630786391,
+      "grad_norm": 0.5755524039268494,
+      "learning_rate": 5.588188138415336e-06,
+      "loss": 0.9191,
+      "step": 3750
+    },
+    {
+      "epoch": 4.194088120468488,
+      "grad_norm": 0.5454072952270508,
+      "learning_rate": 5.539467931620328e-06,
+      "loss": 0.9178,
+      "step": 3760
+    },
+    {
+      "epoch": 4.205242610150585,
+      "grad_norm": 0.5978593230247498,
+      "learning_rate": 5.490879544414238e-06,
+      "loss": 0.9391,
+      "step": 3770
+    },
+    {
+      "epoch": 4.216397099832682,
+      "grad_norm": 0.6521888971328735,
+      "learning_rate": 5.442424412702263e-06,
+      "loss": 0.9057,
+      "step": 3780
+    },
+    {
+      "epoch": 4.227551589514779,
+      "grad_norm": 0.5424159169197083,
+      "learning_rate": 5.394103968451592e-06,
+      "loss": 0.9175,
+      "step": 3790
+    },
+    {
+      "epoch": 4.238706079196877,
+      "grad_norm": 0.6317155361175537,
+      "learning_rate": 5.345919639649067e-06,
+      "loss": 0.9336,
+      "step": 3800
+    },
+    {
+      "epoch": 4.249860568878974,
+      "grad_norm": 0.5228768587112427,
+      "learning_rate": 5.297872850258987e-06,
+      "loss": 0.8923,
+      "step": 3810
+    },
+    {
+      "epoch": 4.261015058561071,
+      "grad_norm": 0.5018953084945679,
+      "learning_rate": 5.249965020181018e-06,
+      "loss": 0.9049,
+      "step": 3820
+    },
+    {
+      "epoch": 4.272169548243168,
+      "grad_norm": 0.5086571574211121,
+      "learning_rate": 5.202197565208256e-06,
+      "loss": 0.9354,
+      "step": 3830
+    },
+    {
+      "epoch": 4.283324037925265,
+      "grad_norm": 0.5657381415367126,
+      "learning_rate": 5.1545718969853546e-06,
+      "loss": 0.9411,
+      "step": 3840
+    },
+    {
+      "epoch": 4.294478527607362,
+      "grad_norm": 0.5901178121566772,
+      "learning_rate": 5.107089422966831e-06,
+      "loss": 0.9259,
+      "step": 3850
+    },
+    {
+      "epoch": 4.305633017289459,
+      "grad_norm": 0.5313717126846313,
+      "learning_rate": 5.059751546375469e-06,
+      "loss": 0.9341,
+      "step": 3860
+    },
+    {
+      "epoch": 4.316787506971556,
+      "grad_norm": 0.5543375015258789,
+      "learning_rate": 5.012559666160836e-06,
+      "loss": 0.9601,
+      "step": 3870
+    },
+    {
+      "epoch": 4.327941996653653,
+      "grad_norm": 0.6120020151138306,
+      "learning_rate": 4.965515176957966e-06,
+      "loss": 0.9203,
+      "step": 3880
+    },
+    {
+      "epoch": 4.33909648633575,
+      "grad_norm": 0.6483854651451111,
+      "learning_rate": 4.918619469046108e-06,
+      "loss": 0.9579,
+      "step": 3890
+    },
+    {
+      "epoch": 4.350250976017847,
+      "grad_norm": 0.5992943048477173,
+      "learning_rate": 4.871873928307684e-06,
+      "loss": 0.959,
+      "step": 3900
+    },
+    {
+      "epoch": 4.361405465699944,
+      "grad_norm": 0.547232985496521,
+      "learning_rate": 4.825279936187289e-06,
+      "loss": 0.9164,
+      "step": 3910
+    },
+    {
+      "epoch": 4.372559955382041,
+      "grad_norm": 0.5406577587127686,
+      "learning_rate": 4.7788388696509035e-06,
+      "loss": 0.9526,
+      "step": 3920
+    },
+    {
+      "epoch": 4.383714445064138,
+      "grad_norm": 0.5376970171928406,
+      "learning_rate": 4.732552101145179e-06,
+      "loss": 0.928,
+      "step": 3930
+    },
+    {
+      "epoch": 4.394868934746236,
+      "grad_norm": 0.5624706745147705,
+      "learning_rate": 4.686420998556885e-06,
+      "loss": 0.8763,
+      "step": 3940
+    },
+    {
+      "epoch": 4.406023424428333,
+      "grad_norm": 0.6041187644004822,
+      "learning_rate": 4.640446925172478e-06,
+      "loss": 0.9731,
+      "step": 3950
+    },
+    {
+      "epoch": 4.41717791411043,
+      "grad_norm": 0.5117301940917969,
+      "learning_rate": 4.594631239637826e-06,
+      "loss": 0.9687,
+      "step": 3960
+    },
+    {
+      "epoch": 4.428332403792527,
+      "grad_norm": 0.5197976231575012,
+      "learning_rate": 4.54897529591805e-06,
+      "loss": 0.9019,
+      "step": 3970
+    },
+    {
+      "epoch": 4.439486893474624,
+      "grad_norm": 0.5039201378822327,
+      "learning_rate": 4.503480443257508e-06,
+      "loss": 0.8707,
+      "step": 3980
+    },
+    {
+      "epoch": 4.450641383156721,
+      "grad_norm": 0.5956012606620789,
+      "learning_rate": 4.458148026139928e-06,
+      "loss": 0.9236,
+      "step": 3990
+    },
+    {
+      "epoch": 4.461795872838818,
+      "grad_norm": 0.47555774450302124,
+      "learning_rate": 4.412979384248665e-06,
+      "loss": 0.9178,
+      "step": 4000
+    },
+    {
+      "epoch": 4.472950362520915,
+      "grad_norm": 0.49644574522972107,
+      "learning_rate": 4.367975852427129e-06,
+      "loss": 0.9034,
+      "step": 4010
+    },
+    {
+      "epoch": 4.484104852203012,
+      "grad_norm": 0.6341298818588257,
+      "learning_rate": 4.32313876063931e-06,
+      "loss": 0.9602,
+      "step": 4020
+    },
+    {
+      "epoch": 4.495259341885109,
+      "grad_norm": 0.5583815574645996,
+      "learning_rate": 4.278469433930503e-06,
+      "loss": 0.9558,
+      "step": 4030
+    },
+    {
+      "epoch": 4.5064138315672055,
+      "grad_norm": 0.5553238391876221,
+      "learning_rate": 4.23396919238813e-06,
+      "loss": 0.934,
+      "step": 4040
+    },
+    {
+      "epoch": 4.5175683212493025,
+      "grad_norm": 0.47058847546577454,
+      "learning_rate": 4.1896393511027356e-06,
+      "loss": 0.9619,
+      "step": 4050
+    },
+    {
+      "epoch": 4.5287228109313995,
+      "grad_norm": 0.5800861120223999,
+      "learning_rate": 4.145481220129126e-06,
+      "loss": 0.9468,
+      "step": 4060
+    },
+    {
+      "epoch": 4.539877300613497,
+      "grad_norm": 0.5830948352813721,
+      "learning_rate": 4.101496104447637e-06,
+      "loss": 0.9265,
+      "step": 4070
+    },
+    {
+      "epoch": 4.551031790295594,
+      "grad_norm": 0.5482431054115295,
+      "learning_rate": 4.057685303925597e-06,
+      "loss": 0.8726,
+      "step": 4080
+    },
+    {
+      "epoch": 4.562186279977691,
+      "grad_norm": 0.5066527724266052,
+      "learning_rate": 4.014050113278881e-06,
+      "loss": 0.9462,
+      "step": 4090
+    },
+    {
+      "epoch": 4.573340769659788,
+      "grad_norm": 0.6379840970039368,
+      "learning_rate": 3.970591822033676e-06,
+      "loss": 0.946,
+      "step": 4100
+    },
+    {
+      "epoch": 4.584495259341885,
+      "grad_norm": 0.5422645807266235,
+      "learning_rate": 3.927311714488356e-06,
+      "loss": 0.9416,
+      "step": 4110
+    },
+    {
+      "epoch": 4.595649749023982,
+      "grad_norm": 0.5916184782981873,
+      "learning_rate": 3.884211069675539e-06,
+      "loss": 0.9111,
+      "step": 4120
+    },
+    {
+      "epoch": 4.606804238706079,
+      "grad_norm": 0.5947620272636414,
+      "learning_rate": 3.841291161324267e-06,
+      "loss": 0.9389,
+      "step": 4130
+    },
+    {
+      "epoch": 4.617958728388176,
+      "grad_norm": 0.8486486077308655,
+      "learning_rate": 3.7985532578223984e-06,
+      "loss": 0.9219,
+      "step": 4140
+    },
+    {
+      "epoch": 4.629113218070273,
+      "grad_norm": 0.5247789621353149,
+      "learning_rate": 3.755998622179087e-06,
+      "loss": 0.9282,
+      "step": 4150
+    },
+    {
+      "epoch": 4.64026770775237,
+      "grad_norm": 0.6193472743034363,
+      "learning_rate": 3.713628511987486e-06,
+      "loss": 0.9463,
+      "step": 4160
+    },
+    {
+      "epoch": 4.651422197434467,
+      "grad_norm": 0.5674899220466614,
+      "learning_rate": 3.671444179387572e-06,
+      "loss": 0.9793,
+      "step": 4170
+    },
+    {
+      "epoch": 4.662576687116564,
+      "grad_norm": 0.5814547538757324,
+      "learning_rate": 3.6294468710291365e-06,
+      "loss": 0.9772,
+      "step": 4180
+    },
+    {
+      "epoch": 4.673731176798661,
+      "grad_norm": 0.5413376092910767,
+      "learning_rate": 3.5876378280349543e-06,
+      "loss": 0.9171,
+      "step": 4190
+    },
+    {
+      "epoch": 4.684885666480758,
+      "grad_norm": 0.5505338311195374,
+      "learning_rate": 3.5460182859640914e-06,
+      "loss": 0.9548,
+      "step": 4200
+    },
+    {
+      "epoch": 4.696040156162855,
+      "grad_norm": 0.5410326719284058,
+      "learning_rate": 3.5045894747754104e-06,
+      "loss": 0.9124,
+      "step": 4210
+    },
+    {
+      "epoch": 4.707194645844953,
+      "grad_norm": 0.5259800553321838,
+      "learning_rate": 3.4633526187912004e-06,
+      "loss": 0.933,
+      "step": 4220
+    },
+    {
+      "epoch": 4.71834913552705,
+      "grad_norm": 0.583570122718811,
+      "learning_rate": 3.4223089366610174e-06,
+      "loss": 0.9591,
+      "step": 4230
+    },
+    {
+      "epoch": 4.729503625209147,
+      "grad_norm": 0.5068359375,
+      "learning_rate": 3.381459641325653e-06,
+      "loss": 0.9328,
+      "step": 4240
+    },
+    {
+      "epoch": 4.740658114891244,
+      "grad_norm": 0.5359495282173157,
+      "learning_rate": 3.3408059399813007e-06,
+      "loss": 0.979,
+      "step": 4250
+    },
+    {
+      "epoch": 4.751812604573341,
+      "grad_norm": 0.5905212163925171,
+      "learning_rate": 3.300349034043865e-06,
+      "loss": 0.9475,
+      "step": 4260
+    },
+    {
+      "epoch": 4.762967094255438,
+      "grad_norm": 0.529193103313446,
+      "learning_rate": 3.260090119113478e-06,
+      "loss": 0.9545,
+      "step": 4270
+    },
+    {
+      "epoch": 4.774121583937535,
+      "grad_norm": 0.6047744154930115,
+      "learning_rate": 3.22003038493915e-06,
+      "loss": 0.9565,
+      "step": 4280
+    },
+    {
+      "epoch": 4.785276073619632,
+      "grad_norm": 0.5448325276374817,
+      "learning_rate": 3.180171015383614e-06,
+      "loss": 0.9583,
+      "step": 4290
+    },
+    {
+      "epoch": 4.796430563301729,
+      "grad_norm": 0.6160260438919067,
+      "learning_rate": 3.1405131883883466e-06,
+      "loss": 0.9427,
+      "step": 4300
+    },
+    {
+      "epoch": 4.807585052983826,
+      "grad_norm": 0.5397651195526123,
+      "learning_rate": 3.1010580759387377e-06,
+      "loss": 0.934,
+      "step": 4310
+    },
+    {
+      "epoch": 4.818739542665923,
+      "grad_norm": 0.5519613027572632,
+      "learning_rate": 3.0618068440294847e-06,
+      "loss": 0.9485,
+      "step": 4320
+    },
+    {
+      "epoch": 4.82989403234802,
+      "grad_norm": 0.5627745985984802,
+      "learning_rate": 3.0227606526301032e-06,
+      "loss": 0.9403,
+      "step": 4330
+    },
+    {
+      "epoch": 4.841048522030118,
+      "grad_norm": 0.5761961340904236,
+      "learning_rate": 2.983920655650673e-06,
+      "loss": 0.9949,
+      "step": 4340
+    },
+    {
+      "epoch": 4.852203011712215,
+      "grad_norm": 0.4996737241744995,
+      "learning_rate": 2.9452880009077212e-06,
+      "loss": 0.8788,
+      "step": 4350
+    },
+    {
+      "epoch": 4.8633575013943116,
+      "grad_norm": 0.5728593468666077,
+      "learning_rate": 2.906863830090314e-06,
+      "loss": 0.9781,
+      "step": 4360
+    },
+    {
+      "epoch": 4.8745119910764085,
+      "grad_norm": 54.11509323120117,
+      "learning_rate": 2.8686492787262966e-06,
+      "loss": 0.9177,
+      "step": 4370
+    },
+    {
+      "epoch": 4.8856664807585055,
+      "grad_norm": 0.5546409487724304,
+      "learning_rate": 2.830645476148759e-06,
+      "loss": 0.9255,
+      "step": 4380
+    },
+    {
+      "epoch": 4.8968209704406025,
+      "grad_norm": 0.6528788805007935,
+      "learning_rate": 2.7928535454626514e-06,
+      "loss": 0.9241,
+      "step": 4390
+    },
+    {
+      "epoch": 4.9079754601226995,
+      "grad_norm": 0.7199203372001648,
+      "learning_rate": 2.755274603511585e-06,
+      "loss": 0.9972,
+      "step": 4400
+    },
+    {
+      "epoch": 4.9191299498047965,
+      "grad_norm": 0.5689542889595032,
+      "learning_rate": 2.7179097608448435e-06,
+      "loss": 0.9259,
+      "step": 4410
+    },
+    {
+      "epoch": 4.930284439486893,
+      "grad_norm": 0.5618848204612732,
+      "learning_rate": 2.680760121684551e-06,
+      "loss": 0.9542,
+      "step": 4420
+    },
+    {
+      "epoch": 4.94143892916899,
+      "grad_norm": 0.5599839091300964,
+      "learning_rate": 2.64382678389305e-06,
+      "loss": 0.9011,
+      "step": 4430
+    },
+    {
+      "epoch": 4.952593418851087,
+      "grad_norm": 0.5971745252609253,
+      "learning_rate": 2.6071108389404367e-06,
+      "loss": 0.9091,
+      "step": 4440
+    },
+    {
+      "epoch": 4.963747908533184,
+      "grad_norm": 0.5364637970924377,
+      "learning_rate": 2.570613371872336e-06,
+      "loss": 0.9159,
+      "step": 4450
+    },
+    {
+      "epoch": 4.974902398215281,
+      "grad_norm": 0.5795035362243652,
+      "learning_rate": 2.534335461277805e-06,
+      "loss": 0.92,
+      "step": 4460
+    },
+    {
+      "epoch": 4.986056887897378,
+      "grad_norm": 0.5911030769348145,
+      "learning_rate": 2.4982781792574794e-06,
+      "loss": 0.9499,
+      "step": 4470
+    },
+    {
+      "epoch": 4.997211377579475,
+      "grad_norm": 0.527425229549408,
+      "learning_rate": 2.462442591391885e-06,
+      "loss": 0.9368,
+      "step": 4480
+    },
+    {
+      "epoch": 5.008365867261573,
+      "grad_norm": 0.6527435183525085,
+      "learning_rate": 2.4268297567099395e-06,
+      "loss": 0.9342,
+      "step": 4490
+    },
+    {
+      "epoch": 5.01952035694367,
+      "grad_norm": 0.5617554783821106,
+      "learning_rate": 2.391440727657669e-06,
+      "loss": 0.9302,
+      "step": 4500
+    },
+    {
+      "epoch": 5.030674846625767,
+      "grad_norm": 0.5033459067344666,
+      "learning_rate": 2.356276550067086e-06,
+      "loss": 0.9211,
+      "step": 4510
+    },
+    {
+      "epoch": 5.041829336307864,
+      "grad_norm": 0.545575737953186,
+      "learning_rate": 2.321338263125308e-06,
+      "loss": 0.9204,
+      "step": 4520
+    },
+    {
+      "epoch": 5.052983825989961,
+      "grad_norm": 0.5373018383979797,
+      "learning_rate": 2.2866268993438214e-06,
+      "loss": 0.9736,
+      "step": 4530
+    },
+    {
+      "epoch": 5.064138315672058,
+      "grad_norm": 0.6192358732223511,
+      "learning_rate": 2.252143484527989e-06,
+      "loss": 0.9352,
+      "step": 4540
+    },
+    {
+      "epoch": 5.075292805354155,
+      "grad_norm": 0.514544665813446,
+      "learning_rate": 2.2178890377467234e-06,
+      "loss": 0.9198,
+      "step": 4550
+    },
+    {
+      "epoch": 5.086447295036252,
+      "grad_norm": 0.5495268702507019,
+      "learning_rate": 2.1838645713023767e-06,
+      "loss": 0.9173,
+      "step": 4560
+    },
+    {
+      "epoch": 5.097601784718349,
+      "grad_norm": 0.557684600353241,
+      "learning_rate": 2.150071090700815e-06,
+      "loss": 0.8829,
+      "step": 4570
+    },
+    {
+      "epoch": 5.108756274400446,
+      "grad_norm": 0.6613608002662659,
+      "learning_rate": 2.116509594621716e-06,
+      "loss": 0.9358,
+      "step": 4580
+    },
+    {
+      "epoch": 5.119910764082543,
+      "grad_norm": 0.537002444267273,
+      "learning_rate": 2.0831810748890513e-06,
+      "loss": 0.9361,
+      "step": 4590
+    },
+    {
+      "epoch": 5.13106525376464,
+      "grad_norm": 0.5655337572097778,
+      "learning_rate": 2.0500865164417614e-06,
+      "loss": 0.8873,
+      "step": 4600
+    },
+    {
+      "epoch": 5.142219743446737,
+      "grad_norm": 0.5133952498435974,
+      "learning_rate": 2.0172268973046815e-06,
+      "loss": 0.9147,
+      "step": 4610
+    },
+    {
+      "epoch": 5.153374233128835,
+      "grad_norm": 0.46780067682266235,
+      "learning_rate": 1.9846031885595984e-06,
+      "loss": 0.9471,
+      "step": 4620
+    },
+    {
+      "epoch": 5.164528722810932,
+      "grad_norm": 0.5493624210357666,
+      "learning_rate": 1.952216354316585e-06,
+      "loss": 0.9283,
+      "step": 4630
+    },
+    {
+      "epoch": 5.175683212493029,
+      "grad_norm": 0.53765469789505,
+      "learning_rate": 1.920067351685485e-06,
+      "loss": 0.9098,
+      "step": 4640
+    },
+    {
+      "epoch": 5.186837702175126,
+      "grad_norm": 0.6295993328094482,
+      "learning_rate": 1.8881571307476477e-06,
+      "loss": 0.9579,
+      "step": 4650
+    },
+    {
+      "epoch": 5.197992191857223,
+      "grad_norm": 0.5361995697021484,
+      "learning_rate": 1.8564866345278376e-06,
+      "loss": 0.9299,
+      "step": 4660
+    },
+    {
+      "epoch": 5.20914668153932,
+      "grad_norm": 0.5145257711410522,
+      "learning_rate": 1.8250567989663736e-06,
+      "loss": 0.8962,
+      "step": 4670
+    },
+    {
+      "epoch": 5.220301171221417,
+      "grad_norm": 0.5769006013870239,
+      "learning_rate": 1.7938685528914579e-06,
+      "loss": 0.9527,
+      "step": 4680
+    },
+    {
+      "epoch": 5.231455660903514,
+      "grad_norm": 0.5536583662033081,
+      "learning_rate": 1.7629228179917413e-06,
+      "loss": 0.9427,
+      "step": 4690
+    },
+    {
+      "epoch": 5.242610150585611,
+      "grad_norm": 0.6008259654045105,
+      "learning_rate": 1.7322205087890798e-06,
+      "loss": 0.959,
+      "step": 4700
+    },
+    {
+      "epoch": 5.253764640267708,
+      "grad_norm": 0.5460879802703857,
+      "learning_rate": 1.7017625326114983e-06,
+      "loss": 0.9517,
+      "step": 4710
+    },
+    {
+      "epoch": 5.264919129949805,
+      "grad_norm": 0.5374237298965454,
+      "learning_rate": 1.6715497895663945e-06,
+      "loss": 0.8733,
+      "step": 4720
+    },
+    {
+      "epoch": 5.276073619631902,
+      "grad_norm": 0.5077516436576843,
+      "learning_rate": 1.6415831725139275e-06,
+      "loss": 0.8602,
+      "step": 4730
+    },
+    {
+      "epoch": 5.287228109313999,
+      "grad_norm": 0.5702505707740784,
+      "learning_rate": 1.6118635670406346e-06,
+      "loss": 0.9652,
+      "step": 4740
+    },
+    {
+      "epoch": 5.2983825989960955,
+      "grad_norm": 0.5563054084777832,
+      "learning_rate": 1.5823918514332515e-06,
+      "loss": 0.9693,
+      "step": 4750
+    },
+    {
+      "epoch": 5.3095370886781925,
+      "grad_norm": 0.5390822887420654,
+      "learning_rate": 1.553168896652777e-06,
+      "loss": 0.9936,
+      "step": 4760
+    },
+    {
+      "epoch": 5.32069157836029,
+      "grad_norm": 0.6257708668708801,
+      "learning_rate": 1.5241955663087094e-06,
+      "loss": 0.9668,
+      "step": 4770
+    },
+    {
+      "epoch": 5.331846068042387,
+      "grad_norm": 0.5015426874160767,
+      "learning_rate": 1.4954727166335436e-06,
+      "loss": 0.8796,
+      "step": 4780
+    },
+    {
+      "epoch": 5.343000557724484,
+      "grad_norm": 0.5086983442306519,
+      "learning_rate": 1.4670011964574605e-06,
+      "loss": 0.9337,
+      "step": 4790
+    },
+    {
+      "epoch": 5.354155047406581,
+      "grad_norm": 0.5448817610740662,
+      "learning_rate": 1.4387818471832403e-06,
+      "loss": 0.9192,
+      "step": 4800
+    },
+    {
+      "epoch": 5.365309537088678,
+      "grad_norm": 0.65288907289505,
+      "learning_rate": 1.4108155027614e-06,
+      "loss": 0.878,
+      "step": 4810
+    },
+    {
+      "epoch": 5.376464026770775,
+      "grad_norm": 0.5269010663032532,
+      "learning_rate": 1.3831029896655446e-06,
+      "loss": 0.9213,
+      "step": 4820
+    },
+    {
+      "epoch": 5.387618516452872,
+      "grad_norm": 0.5757723450660706,
+      "learning_rate": 1.3556451268679483e-06,
+      "loss": 0.9588,
+      "step": 4830
+    },
+    {
+      "epoch": 5.398773006134969,
+      "grad_norm": 0.49955371022224426,
+      "learning_rate": 1.3284427258153433e-06,
+      "loss": 0.9057,
+      "step": 4840
+    },
+    {
+      "epoch": 5.409927495817066,
+      "grad_norm": 0.5367484092712402,
+      "learning_rate": 1.3014965904049492e-06,
+      "loss": 0.8871,
+      "step": 4850
+    },
+    {
+      "epoch": 5.421081985499163,
+      "grad_norm": 0.5186311602592468,
+      "learning_rate": 1.2748075169607132e-06,
+      "loss": 0.924,
+      "step": 4860
+    },
+    {
+      "epoch": 5.43223647518126,
+      "grad_norm": 0.4818533957004547,
+      "learning_rate": 1.2483762942097744e-06,
+      "loss": 0.9807,
+      "step": 4870
+    },
+    {
+      "epoch": 5.443390964863357,
+      "grad_norm": 0.501106858253479,
+      "learning_rate": 1.2222037032591484e-06,
+      "loss": 0.9318,
+      "step": 4880
+    },
+    {
+      "epoch": 5.454545454545454,
+      "grad_norm": 0.6269428133964539,
+      "learning_rate": 1.1962905175726626e-06,
+      "loss": 0.9632,
+      "step": 4890
+    },
+    {
+      "epoch": 5.465699944227552,
+      "grad_norm": 0.5876699090003967,
+      "learning_rate": 1.1706375029480755e-06,
+      "loss": 0.9098,
+      "step": 4900
+    },
+    {
+      "epoch": 5.476854433909649,
+      "grad_norm": 0.6998534202575684,
+      "learning_rate": 1.145245417494464e-06,
+      "loss": 0.9313,
+      "step": 4910
+    },
+    {
+      "epoch": 5.488008923591746,
+      "grad_norm": 0.5238382816314697,
+      "learning_rate": 1.1201150116098148e-06,
+      "loss": 0.9091,
+      "step": 4920
+    },
+    {
+      "epoch": 5.499163413273843,
+      "grad_norm": 0.562708854675293,
+      "learning_rate": 1.0952470279588378e-06,
+      "loss": 0.922,
+      "step": 4930
+    },
+    {
+      "epoch": 5.51031790295594,
+      "grad_norm": 0.5080922245979309,
+      "learning_rate": 1.0706422014510343e-06,
+      "loss": 0.9167,
+      "step": 4940
+    },
+    {
+      "epoch": 5.521472392638037,
+      "grad_norm": 0.524152398109436,
+      "learning_rate": 1.0463012592189636e-06,
+      "loss": 0.9885,
+      "step": 4950
+    },
+    {
+      "epoch": 5.532626882320134,
+      "grad_norm": 0.5330775380134583,
+      "learning_rate": 1.0222249205967693e-06,
+      "loss": 0.9665,
+      "step": 4960
+    },
+    {
+      "epoch": 5.543781372002231,
+      "grad_norm": 0.5838893055915833,
+      "learning_rate": 9.984138970989033e-07,
+      "loss": 0.9439,
+      "step": 4970
+    },
+    {
+      "epoch": 5.554935861684328,
+      "grad_norm": 0.5749791264533997,
+      "learning_rate": 9.748688923991245e-07,
+      "loss": 0.9415,
+      "step": 4980
+    },
+    {
+      "epoch": 5.566090351366425,
+      "grad_norm": 0.5631915926933289,
+      "learning_rate": 9.515906023096733e-07,
+      "loss": 0.9282,
+      "step": 4990
+    },
+    {
+      "epoch": 5.577244841048522,
+      "grad_norm": 0.5311354398727417,
+      "learning_rate": 9.285797147607356e-07,
+      "loss": 0.9404,
+      "step": 5000
+    },
+    {
+      "epoch": 5.588399330730619,
+      "grad_norm": 0.47747719287872314,
+      "learning_rate": 9.05836909780089e-07,
+      "loss": 0.9537,
+      "step": 5010
+    },
+    {
+      "epoch": 5.599553820412716,
+      "grad_norm": 0.484561562538147,
+      "learning_rate": 8.833628594730281e-07,
+      "loss": 0.9638,
+      "step": 5020
+    },
+    {
+      "epoch": 5.610708310094813,
+      "grad_norm": 0.5075972676277161,
+      "learning_rate": 8.611582280024877e-07,
+      "loss": 0.8905,
+      "step": 5030
+    },
+    {
+      "epoch": 5.62186279977691,
+      "grad_norm": 0.571488082408905,
+      "learning_rate": 8.392236715694213e-07,
+      "loss": 0.9218,
+      "step": 5040
+    },
+    {
+      "epoch": 5.633017289459008,
+      "grad_norm": 0.4815698266029358,
+      "learning_rate": 8.175598383934058e-07,
+      "loss": 0.9161,
+      "step": 5050
+    },
+    {
+      "epoch": 5.644171779141105,
+      "grad_norm": 0.4516790807247162,
+      "learning_rate": 7.961673686934857e-07,
+      "loss": 0.9217,
+      "step": 5060
+    },
+    {
+      "epoch": 5.6553262688232016,
+      "grad_norm": 0.582046389579773,
+      "learning_rate": 7.750468946692569e-07,
+      "loss": 0.8974,
+      "step": 5070
+    },
+    {
+      "epoch": 5.6664807585052985,
+      "grad_norm": 0.6478422284126282,
+      "learning_rate": 7.541990404821753e-07,
+      "loss": 0.9315,
+      "step": 5080
+    },
+    {
+      "epoch": 5.6776352481873955,
+      "grad_norm": 0.47425535321235657,
+      "learning_rate": 7.336244222371202e-07,
+      "loss": 0.9353,
+      "step": 5090
+    },
+    {
+      "epoch": 5.6887897378694925,
+      "grad_norm": 0.541813850402832,
+      "learning_rate": 7.133236479641847e-07,
+      "loss": 0.8781,
+      "step": 5100
+    },
+    {
+      "epoch": 5.6999442275515895,
+      "grad_norm": 0.6376804113388062,
+      "learning_rate": 6.932973176007019e-07,
+      "loss": 0.9331,
+      "step": 5110
+    },
+    {
+      "epoch": 5.7110987172336865,
+      "grad_norm": 0.5788676738739014,
+      "learning_rate": 6.735460229735213e-07,
+      "loss": 0.9374,
+      "step": 5120
+    },
+    {
+      "epoch": 5.722253206915783,
+      "grad_norm": 0.6712315678596497,
+      "learning_rate": 6.540703477815136e-07,
+      "loss": 0.9127,
+      "step": 5130
+    },
+    {
+      "epoch": 5.73340769659788,
+      "grad_norm": 0.5419293642044067,
+      "learning_rate": 6.348708675783266e-07,
+      "loss": 0.9239,
+      "step": 5140
+    },
+    {
+      "epoch": 5.744562186279977,
+      "grad_norm": 0.5727126002311707,
+      "learning_rate": 6.159481497553699e-07,
+      "loss": 0.9251,
+      "step": 5150
+    },
+    {
+      "epoch": 5.755716675962074,
+      "grad_norm": 0.5651090741157532,
+      "learning_rate": 5.973027535250541e-07,
+      "loss": 0.9339,
+      "step": 5160
+    },
+    {
+      "epoch": 5.766871165644172,
+      "grad_norm": 0.5012417435646057,
+      "learning_rate": 5.789352299042606e-07,
+      "loss": 0.9172,
+      "step": 5170
+    },
+    {
+      "epoch": 5.778025655326269,
+      "grad_norm": 0.4940890967845917,
+      "learning_rate": 5.608461216980587e-07,
+      "loss": 0.9517,
+      "step": 5180
+    },
+    {
+      "epoch": 5.789180145008366,
+      "grad_norm": 0.551123321056366,
+      "learning_rate": 5.43035963483659e-07,
+      "loss": 0.898,
+      "step": 5190
+    },
+    {
+      "epoch": 5.800334634690463,
+      "grad_norm": 0.5738024115562439,
+      "learning_rate": 5.255052815946271e-07,
+      "loss": 0.9385,
+      "step": 5200
+    },
+    {
+      "epoch": 5.81148912437256,
+      "grad_norm": 0.6367058753967285,
+      "learning_rate": 5.082545941053174e-07,
+      "loss": 0.9842,
+      "step": 5210
+    },
+    {
+      "epoch": 5.822643614054657,
+      "grad_norm": 0.5020179748535156,
+      "learning_rate": 4.912844108155701e-07,
+      "loss": 0.8957,
+      "step": 5220
+    },
+    {
+      "epoch": 5.833798103736754,
+      "grad_norm": 0.5717551112174988,
+      "learning_rate": 4.745952332356418e-07,
+      "loss": 0.929,
+      "step": 5230
+    },
+    {
+      "epoch": 5.844952593418851,
+      "grad_norm": 0.5765886902809143,
+      "learning_rate": 4.5818755457138876e-07,
+      "loss": 0.9565,
+      "step": 5240
+    },
+    {
+      "epoch": 5.856107083100948,
+      "grad_norm": 0.5317378044128418,
+      "learning_rate": 4.4206185970968486e-07,
+      "loss": 0.9791,
+      "step": 5250
+    },
+    {
+      "epoch": 5.867261572783045,
+      "grad_norm": 0.47482070326805115,
+      "learning_rate": 4.262186252040956e-07,
+      "loss": 0.9215,
+      "step": 5260
+    },
+    {
+      "epoch": 5.878416062465142,
+      "grad_norm": 0.5285121202468872,
+      "learning_rate": 4.1065831926079937e-07,
+      "loss": 0.9223,
+      "step": 5270
+    },
+    {
+      "epoch": 5.889570552147239,
+      "grad_norm": 0.5711480975151062,
+      "learning_rate": 3.9538140172473927e-07,
+      "loss": 0.9168,
+      "step": 5280
+    },
+    {
+      "epoch": 5.900725041829336,
+      "grad_norm": 0.594833254814148,
+      "learning_rate": 3.8038832406604775e-07,
+      "loss": 0.9215,
+      "step": 5290
+    },
+    {
+      "epoch": 5.911879531511433,
+      "grad_norm": 0.5133140683174133,
+      "learning_rate": 3.656795293666937e-07,
+      "loss": 0.934,
+      "step": 5300
+    },
+    {
+      "epoch": 5.92303402119353,
+      "grad_norm": 0.6132499575614929,
+      "learning_rate": 3.512554523073919e-07,
+      "loss": 0.9317,
+      "step": 5310
+    },
+    {
+      "epoch": 5.934188510875628,
+      "grad_norm": 0.6130938529968262,
+      "learning_rate": 3.3711651915475765e-07,
+      "loss": 0.9722,
+      "step": 5320
+    },
+    {
+      "epoch": 5.945343000557725,
+      "grad_norm": 0.5470441579818726,
+      "learning_rate": 3.2326314774870913e-07,
+      "loss": 0.9487,
+      "step": 5330
+    },
+    {
+      "epoch": 5.956497490239822,
+      "grad_norm": 0.6502159237861633,
+      "learning_rate": 3.0969574749012074e-07,
+      "loss": 0.9164,
+      "step": 5340
+    },
+    {
+      "epoch": 5.967651979921919,
+      "grad_norm": 0.4922711253166199,
+      "learning_rate": 2.964147193287215e-07,
+      "loss": 0.9312,
+      "step": 5350
+    },
+    {
+      "epoch": 5.978806469604016,
+      "grad_norm": 0.5844548344612122,
+      "learning_rate": 2.8342045575124764e-07,
+      "loss": 0.9549,
+      "step": 5360
+    },
+    {
+      "epoch": 5.989960959286113,
+      "grad_norm": 0.4878758192062378,
+      "learning_rate": 2.707133407698426e-07,
+      "loss": 0.9225,
+      "step": 5370
+    },
+    {
+      "epoch": 6.00111544896821,
+      "grad_norm": 0.5165712237358093,
+      "learning_rate": 2.5829374991070967e-07,
+      "loss": 0.9135,
+      "step": 5380
+    },
+    {
+      "epoch": 6.012269938650307,
+      "grad_norm": 0.5230659246444702,
+      "learning_rate": 2.4616205020301373e-07,
+      "loss": 0.9555,
+      "step": 5390
+    },
+    {
+      "epoch": 6.023424428332404,
+      "grad_norm": 0.6041128039360046,
+      "learning_rate": 2.3431860016803532e-07,
+      "loss": 0.9014,
+      "step": 5400
+    },
+    {
+      "epoch": 6.034578918014501,
+      "grad_norm": 0.525059700012207,
+      "learning_rate": 2.2276374980857552e-07,
+      "loss": 0.9216,
+      "step": 5410
+    },
+    {
+      "epoch": 6.045733407696598,
+      "grad_norm": 0.5333075523376465,
+      "learning_rate": 2.1149784059861233e-07,
+      "loss": 0.9157,
+      "step": 5420
+    },
+    {
+      "epoch": 6.056887897378695,
+      "grad_norm": 0.5331766605377197,
+      "learning_rate": 2.005212054732042e-07,
+      "loss": 0.9157,
+      "step": 5430
+    },
+    {
+      "epoch": 6.068042387060792,
+      "grad_norm": 0.6051472425460815,
+      "learning_rate": 1.8983416881866225e-07,
+      "loss": 0.952,
+      "step": 5440
+    },
+    {
+      "epoch": 6.079196876742889,
+      "grad_norm": 0.5551755428314209,
+      "learning_rate": 1.7943704646295356e-07,
+      "loss": 0.9298,
+      "step": 5450
+    },
+    {
+      "epoch": 6.090351366424986,
+      "grad_norm": 0.6593601703643799,
+      "learning_rate": 1.6933014566637206e-07,
+      "loss": 0.9707,
+      "step": 5460
+    },
+    {
+      "epoch": 6.101505856107083,
+      "grad_norm": 0.44987979531288147,
+      "learning_rate": 1.5951376511245676e-07,
+      "loss": 0.9041,
+      "step": 5470
+    },
+    {
+      "epoch": 6.11266034578918,
+      "grad_norm": 0.5908699631690979,
+      "learning_rate": 1.4998819489916771e-07,
+      "loss": 0.95,
+      "step": 5480
+    },
+    {
+      "epoch": 6.123814835471277,
+      "grad_norm": 0.5343087315559387,
+      "learning_rate": 1.4075371653031078e-07,
+      "loss": 0.9499,
+      "step": 5490
+    },
+    {
+      "epoch": 6.134969325153374,
+      "grad_norm": 0.47075924277305603,
+      "learning_rate": 1.3181060290721525e-07,
+      "loss": 0.9594,
+      "step": 5500
+    },
+    {
+      "epoch": 6.146123814835471,
+      "grad_norm": 0.5868487358093262,
+      "learning_rate": 1.2315911832067818e-07,
+      "loss": 0.8791,
+      "step": 5510
+    },
+    {
+      "epoch": 6.157278304517568,
+      "grad_norm": 0.5475588440895081,
+      "learning_rate": 1.1479951844314164e-07,
+      "loss": 0.8894,
+      "step": 5520
+    },
+    {
+      "epoch": 6.168432794199665,
+      "grad_norm": 0.4885677099227905,
+      "learning_rate": 1.0673205032114886e-07,
+      "loss": 0.9173,
+      "step": 5530
+    },
+    {
+      "epoch": 6.179587283881762,
+      "grad_norm": 0.5436893105506897,
+      "learning_rate": 9.895695236803559e-08,
+      "loss": 0.8971,
+      "step": 5540
+    },
+    {
+      "epoch": 6.190741773563859,
+      "grad_norm": 0.6072864532470703,
+      "learning_rate": 9.147445435688796e-08,
+      "loss": 0.9459,
+      "step": 5550
+    },
+    {
+      "epoch": 6.201896263245956,
+      "grad_norm": 0.7242813110351562,
+      "learning_rate": 8.428477741375008e-08,
+      "loss": 0.9318,
+      "step": 5560
+    },
+    {
+      "epoch": 6.213050752928053,
+      "grad_norm": 0.5594833493232727,
+      "learning_rate": 7.738813401108824e-08,
+      "loss": 0.9021,
+      "step": 5570
+    },
+    {
+      "epoch": 6.22420524261015,
+      "grad_norm": 0.5553674101829529,
+      "learning_rate": 7.078472796151925e-08,
+      "loss": 0.9286,
+      "step": 5580
+    },
+    {
+      "epoch": 6.235359732292248,
+      "grad_norm": 0.5133783221244812,
+      "learning_rate": 6.44747544117752e-08,
+      "loss": 0.9517,
+      "step": 5590
+    },
+    {
+      "epoch": 6.246514221974345,
+      "grad_norm": 1.0486400127410889,
+      "learning_rate": 5.845839983694812e-08,
+      "loss": 0.9505,
+      "step": 5600
+    },
+    {
+      "epoch": 6.257668711656442,
+      "grad_norm": 0.5520800352096558,
+      "learning_rate": 5.2735842034971065e-08,
+      "loss": 0.95,
+      "step": 5610
+    },
+    {
+      "epoch": 6.268823201338539,
+      "grad_norm": 0.5525377988815308,
+      "learning_rate": 4.730725012137005e-08,
+      "loss": 0.9417,
+      "step": 5620
+    },
+    {
+      "epoch": 6.279977691020636,
+      "grad_norm": 0.5942236185073853,
+      "learning_rate": 4.2172784524258106e-08,
+      "loss": 0.9356,
+      "step": 5630
+    },
+    {
+      "epoch": 6.291132180702733,
+      "grad_norm": 0.5851506590843201,
+      "learning_rate": 3.733259697960234e-08,
+      "loss": 0.9426,
+      "step": 5640
+    },
+    {
+      "epoch": 6.30228667038483,
+      "grad_norm": 0.5317173600196838,
+      "learning_rate": 3.278683052673648e-08,
+      "loss": 0.9644,
+      "step": 5650
+    },
+    {
+      "epoch": 6.313441160066927,
+      "grad_norm": 0.5416374802589417,
+      "learning_rate": 2.8535619504133082e-08,
+      "loss": 0.896,
+      "step": 5660
+    },
+    {
+      "epoch": 6.324595649749024,
+      "grad_norm": 0.5867795348167419,
+      "learning_rate": 2.457908954543453e-08,
+      "loss": 0.9476,
+      "step": 5670
+    },
+    {
+      "epoch": 6.335750139431121,
+      "grad_norm": 0.5497348308563232,
+      "learning_rate": 2.0917357575738206e-08,
+      "loss": 0.9689,
+      "step": 5680
+    },
+    {
+      "epoch": 6.346904629113218,
+      "grad_norm": 0.5706804990768433,
+      "learning_rate": 1.755053180814481e-08,
+      "loss": 0.9638,
+      "step": 5690
+    },
+    {
+      "epoch": 6.358059118795315,
+      "grad_norm": 0.5580919981002808,
+      "learning_rate": 1.4478711740558704e-08,
+      "loss": 0.933,
+      "step": 5700
+    },
+    {
+      "epoch": 6.369213608477412,
+      "grad_norm": 0.5722364783287048,
+      "learning_rate": 1.1701988152744703e-08,
+      "loss": 0.9465,
+      "step": 5710
+    },
+    {
+      "epoch": 6.38036809815951,
+      "grad_norm": 0.5530805587768555,
+      "learning_rate": 9.220443103650222e-09,
+      "loss": 0.9202,
+      "step": 5720
+    },
+    {
+      "epoch": 6.391522587841607,
+      "grad_norm": 0.6193326115608215,
+      "learning_rate": 7.034149928976108e-09,
+      "loss": 0.9017,
+      "step": 5730
+    },
+    {
+      "epoch": 6.402677077523704,
+      "grad_norm": 0.6208977699279785,
+      "learning_rate": 5.1431732390117004e-09,
+      "loss": 0.9061,
+      "step": 5740
+    },
+    {
+      "epoch": 6.413831567205801,
+      "grad_norm": 0.5320120453834534,
+      "learning_rate": 3.5475689167252525e-09,
+      "loss": 0.9409,
+      "step": 5750
+    },
+    {
+      "epoch": 6.424986056887898,
+      "grad_norm": 0.5276364088058472,
+      "learning_rate": 2.2473841161108068e-09,
+      "loss": 0.9083,
+      "step": 5760
+    },
+    {
+      "epoch": 6.436140546569995,
+      "grad_norm": 0.44757771492004395,
+      "learning_rate": 1.2426572607981967e-09,
+      "loss": 0.953,
+      "step": 5770
+    },
+    {
+      "epoch": 6.4472950362520915,
+      "grad_norm": 0.583735466003418,
+      "learning_rate": 5.334180429117374e-10,
+      "loss": 0.9369,
+      "step": 5780
+    },
+    {
+      "epoch": 6.4584495259341885,
+      "grad_norm": 0.5257657170295715,
+      "learning_rate": 1.196874221986999e-10,
+      "loss": 0.9185,
+      "step": 5790
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5799,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 7,
+  "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": 8.275105081636946e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}