diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4879 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.997832369942197,
+  "eval_steps": 500,
+  "global_step": 6917,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0072254335260115606,
+      "grad_norm": 10.139894485473633,
+      "learning_rate": 5.780346820809249e-06,
+      "loss": 1.2772,
+      "step": 10
+    },
+    {
+      "epoch": 0.014450867052023121,
+      "grad_norm": 6.952826499938965,
+      "learning_rate": 1.1560693641618498e-05,
+      "loss": 1.0363,
+      "step": 20
+    },
+    {
+      "epoch": 0.02167630057803468,
+      "grad_norm": 2.2088680267333984,
+      "learning_rate": 1.7341040462427746e-05,
+      "loss": 0.5553,
+      "step": 30
+    },
+    {
+      "epoch": 0.028901734104046242,
+      "grad_norm": 2.3760039806365967,
+      "learning_rate": 2.3121387283236996e-05,
+      "loss": 0.402,
+      "step": 40
+    },
+    {
+      "epoch": 0.036127167630057806,
+      "grad_norm": 3.3085403442382812,
+      "learning_rate": 2.8901734104046245e-05,
+      "loss": 0.2577,
+      "step": 50
+    },
+    {
+      "epoch": 0.04335260115606936,
+      "grad_norm": 1.8953660726547241,
+      "learning_rate": 3.468208092485549e-05,
+      "loss": 0.1902,
+      "step": 60
+    },
+    {
+      "epoch": 0.05057803468208093,
+      "grad_norm": 2.4242184162139893,
+      "learning_rate": 4.046242774566474e-05,
+      "loss": 0.1419,
+      "step": 70
+    },
+    {
+      "epoch": 0.057803468208092484,
+      "grad_norm": 1.6606438159942627,
+      "learning_rate": 4.624277456647399e-05,
+      "loss": 0.1466,
+      "step": 80
+    },
+    {
+      "epoch": 0.06502890173410404,
+      "grad_norm": 1.1051759719848633,
+      "learning_rate": 5.2023121387283234e-05,
+      "loss": 0.1341,
+      "step": 90
+    },
+    {
+      "epoch": 0.07225433526011561,
+      "grad_norm": 1.275235891342163,
+      "learning_rate": 5.780346820809249e-05,
+      "loss": 0.1137,
+      "step": 100
+    },
+    {
+      "epoch": 0.07947976878612717,
+      "grad_norm": 1.6539286375045776,
+      "learning_rate": 6.358381502890174e-05,
+      "loss": 0.0994,
+      "step": 110
+    },
+    {
+      "epoch": 0.08670520231213873,
+      "grad_norm": 1.2315586805343628,
+      "learning_rate": 6.936416184971098e-05,
+      "loss": 0.1045,
+      "step": 120
+    },
+    {
+      "epoch": 0.09393063583815028,
+      "grad_norm": 0.958812952041626,
+      "learning_rate": 7.514450867052023e-05,
+      "loss": 0.0816,
+      "step": 130
+    },
+    {
+      "epoch": 0.10115606936416185,
+      "grad_norm": 1.3614000082015991,
+      "learning_rate": 8.092485549132948e-05,
+      "loss": 0.0909,
+      "step": 140
+    },
+    {
+      "epoch": 0.10838150289017341,
+      "grad_norm": 1.5399832725524902,
+      "learning_rate": 8.670520231213874e-05,
+      "loss": 0.0911,
+      "step": 150
+    },
+    {
+      "epoch": 0.11560693641618497,
+      "grad_norm": 1.1797833442687988,
+      "learning_rate": 9.248554913294798e-05,
+      "loss": 0.0854,
+      "step": 160
+    },
+    {
+      "epoch": 0.12283236994219653,
+      "grad_norm": 1.557212233543396,
+      "learning_rate": 9.826589595375723e-05,
+      "loss": 0.0963,
+      "step": 170
+    },
+    {
+      "epoch": 0.13005780346820808,
+      "grad_norm": 0.5504224896430969,
+      "learning_rate": 0.00010404624277456647,
+      "loss": 0.0667,
+      "step": 180
+    },
+    {
+      "epoch": 0.13728323699421965,
+      "grad_norm": 0.6468238234519958,
+      "learning_rate": 0.00010982658959537572,
+      "loss": 0.0917,
+      "step": 190
+    },
+    {
+      "epoch": 0.14450867052023122,
+      "grad_norm": 1.564103364944458,
+      "learning_rate": 0.00011560693641618498,
+      "loss": 0.1012,
+      "step": 200
+    },
+    {
+      "epoch": 0.15173410404624277,
+      "grad_norm": 0.7247146368026733,
+      "learning_rate": 0.00012138728323699422,
+      "loss": 0.0768,
+      "step": 210
+    },
+    {
+      "epoch": 0.15895953757225434,
+      "grad_norm": 0.9195826053619385,
+      "learning_rate": 0.00012716763005780348,
+      "loss": 0.0842,
+      "step": 220
+    },
+    {
+      "epoch": 0.16618497109826588,
+      "grad_norm": 1.3616660833358765,
+      "learning_rate": 0.0001329479768786127,
+      "loss": 0.0722,
+      "step": 230
+    },
+    {
+      "epoch": 0.17341040462427745,
+      "grad_norm": 0.9556477069854736,
+      "learning_rate": 0.00013872832369942197,
+      "loss": 0.0767,
+      "step": 240
+    },
+    {
+      "epoch": 0.18063583815028902,
+      "grad_norm": 0.6739767789840698,
+      "learning_rate": 0.00014450867052023122,
+      "loss": 0.0768,
+      "step": 250
+    },
+    {
+      "epoch": 0.18786127167630057,
+      "grad_norm": 0.9936643242835999,
+      "learning_rate": 0.00015028901734104045,
+      "loss": 0.0686,
+      "step": 260
+    },
+    {
+      "epoch": 0.19508670520231214,
+      "grad_norm": 0.8351348638534546,
+      "learning_rate": 0.0001560693641618497,
+      "loss": 0.0677,
+      "step": 270
+    },
+    {
+      "epoch": 0.2023121387283237,
+      "grad_norm": 1.2121962308883667,
+      "learning_rate": 0.00016184971098265897,
+      "loss": 0.0858,
+      "step": 280
+    },
+    {
+      "epoch": 0.20953757225433525,
+      "grad_norm": 0.4404231309890747,
+      "learning_rate": 0.00016763005780346822,
+      "loss": 0.0653,
+      "step": 290
+    },
+    {
+      "epoch": 0.21676300578034682,
+      "grad_norm": 0.5475902557373047,
+      "learning_rate": 0.00017341040462427748,
+      "loss": 0.0733,
+      "step": 300
+    },
+    {
+      "epoch": 0.2239884393063584,
+      "grad_norm": 0.6645906567573547,
+      "learning_rate": 0.00017919075144508673,
+      "loss": 0.0603,
+      "step": 310
+    },
+    {
+      "epoch": 0.23121387283236994,
+      "grad_norm": 0.9024297595024109,
+      "learning_rate": 0.00018497109826589596,
+      "loss": 0.0638,
+      "step": 320
+    },
+    {
+      "epoch": 0.2384393063583815,
+      "grad_norm": 0.8652449250221252,
+      "learning_rate": 0.00019075144508670522,
+      "loss": 0.0674,
+      "step": 330
+    },
+    {
+      "epoch": 0.24566473988439305,
+      "grad_norm": 1.0030561685562134,
+      "learning_rate": 0.00019653179190751445,
+      "loss": 0.0529,
+      "step": 340
+    },
+    {
+      "epoch": 0.25289017341040465,
+      "grad_norm": 0.8162288665771484,
+      "learning_rate": 0.00019999981713665253,
+      "loss": 0.0651,
+      "step": 350
+    },
+    {
+      "epoch": 0.26011560693641617,
+      "grad_norm": 1.0239158868789673,
+      "learning_rate": 0.00019999775993167403,
+      "loss": 0.0622,
+      "step": 360
+    },
+    {
+      "epoch": 0.26734104046242774,
+      "grad_norm": 0.6754204034805298,
+      "learning_rate": 0.0001999934169897127,
+      "loss": 0.0594,
+      "step": 370
+    },
+    {
+      "epoch": 0.2745664739884393,
+      "grad_norm": 0.7634507417678833,
+      "learning_rate": 0.00019998678841003908,
+      "loss": 0.0621,
+      "step": 380
+    },
+    {
+      "epoch": 0.2817919075144509,
+      "grad_norm": 0.5819258093833923,
+      "learning_rate": 0.0001999778743441684,
+      "loss": 0.0666,
+      "step": 390
+    },
+    {
+      "epoch": 0.28901734104046245,
+      "grad_norm": 0.7896689176559448,
+      "learning_rate": 0.0001999666749958574,
+      "loss": 0.0468,
+      "step": 400
+    },
+    {
+      "epoch": 0.29624277456647397,
+      "grad_norm": 0.5839788913726807,
+      "learning_rate": 0.00019995319062109936,
+      "loss": 0.0626,
+      "step": 410
+    },
+    {
+      "epoch": 0.30346820809248554,
+      "grad_norm": 0.6790082454681396,
+      "learning_rate": 0.00019993742152811864,
+      "loss": 0.0646,
+      "step": 420
+    },
+    {
+      "epoch": 0.3106936416184971,
+      "grad_norm": 0.72486811876297,
+      "learning_rate": 0.00019991936807736317,
+      "loss": 0.0595,
+      "step": 430
+    },
+    {
+      "epoch": 0.3179190751445087,
+      "grad_norm": 0.5554680228233337,
+      "learning_rate": 0.00019989903068149672,
+      "loss": 0.0627,
+      "step": 440
+    },
+    {
+      "epoch": 0.32514450867052025,
+      "grad_norm": 0.691638708114624,
+      "learning_rate": 0.00019987640980538904,
+      "loss": 0.0639,
+      "step": 450
+    },
+    {
+      "epoch": 0.33236994219653176,
+      "grad_norm": 0.9424274563789368,
+      "learning_rate": 0.0001998515059661054,
+      "loss": 0.0538,
+      "step": 460
+    },
+    {
+      "epoch": 0.33959537572254334,
+      "grad_norm": 0.7785006165504456,
+      "learning_rate": 0.0001998243197328949,
+      "loss": 0.0577,
+      "step": 470
+    },
+    {
+      "epoch": 0.3468208092485549,
+      "grad_norm": 0.7871196269989014,
+      "learning_rate": 0.00019979485172717714,
+      "loss": 0.0521,
+      "step": 480
+    },
+    {
+      "epoch": 0.3540462427745665,
+      "grad_norm": 0.6548967361450195,
+      "learning_rate": 0.00019976310262252839,
+      "loss": 0.0592,
+      "step": 490
+    },
+    {
+      "epoch": 0.36127167630057805,
+      "grad_norm": 0.7058358192443848,
+      "learning_rate": 0.0001997290731446659,
+      "loss": 0.0492,
+      "step": 500
+    },
+    {
+      "epoch": 0.3684971098265896,
+      "grad_norm": 0.49141350388526917,
+      "learning_rate": 0.0001996927640714315,
+      "loss": 0.051,
+      "step": 510
+    },
+    {
+      "epoch": 0.37572254335260113,
+      "grad_norm": 0.5175216197967529,
+      "learning_rate": 0.00019965417623277364,
+      "loss": 0.042,
+      "step": 520
+    },
+    {
+      "epoch": 0.3829479768786127,
+      "grad_norm": 0.6683163046836853,
+      "learning_rate": 0.0001996133105107286,
+      "loss": 0.0573,
+      "step": 530
+    },
+    {
+      "epoch": 0.3901734104046243,
+      "grad_norm": 0.9746570587158203,
+      "learning_rate": 0.00019957016783940024,
+      "loss": 0.0587,
+      "step": 540
+    },
+    {
+      "epoch": 0.39739884393063585,
+      "grad_norm": 0.6539334058761597,
+      "learning_rate": 0.00019952474920493864,
+      "loss": 0.0466,
+      "step": 550
+    },
+    {
+      "epoch": 0.4046242774566474,
+      "grad_norm": 0.36408939957618713,
+      "learning_rate": 0.0001994770556455176,
+      "loss": 0.0485,
+      "step": 560
+    },
+    {
+      "epoch": 0.41184971098265893,
+      "grad_norm": 0.9217486381530762,
+      "learning_rate": 0.0001994270882513108,
+      "loss": 0.0576,
+      "step": 570
+    },
+    {
+      "epoch": 0.4190751445086705,
+      "grad_norm": 0.4366924464702606,
+      "learning_rate": 0.00019937484816446706,
+      "loss": 0.0518,
+      "step": 580
+    },
+    {
+      "epoch": 0.4263005780346821,
+      "grad_norm": 0.7538712024688721,
+      "learning_rate": 0.00019932033657908417,
+      "loss": 0.0442,
+      "step": 590
+    },
+    {
+      "epoch": 0.43352601156069365,
+      "grad_norm": 0.5748066902160645,
+      "learning_rate": 0.00019926355474118144,
+      "loss": 0.0526,
+      "step": 600
+    },
+    {
+      "epoch": 0.4407514450867052,
+      "grad_norm": 0.8760377168655396,
+      "learning_rate": 0.00019920450394867145,
+      "loss": 0.0612,
+      "step": 610
+    },
+    {
+      "epoch": 0.4479768786127168,
+      "grad_norm": 0.5834543108940125,
+      "learning_rate": 0.0001991431855513302,
+      "loss": 0.0539,
+      "step": 620
+    },
+    {
+      "epoch": 0.4552023121387283,
+      "grad_norm": 0.5197954773902893,
+      "learning_rate": 0.00019907960095076645,
+      "loss": 0.0461,
+      "step": 630
+    },
+    {
+      "epoch": 0.4624277456647399,
+      "grad_norm": 0.4933251440525055,
+      "learning_rate": 0.00019901375160038943,
+      "loss": 0.0482,
+      "step": 640
+    },
+    {
+      "epoch": 0.46965317919075145,
+      "grad_norm": 0.4894423186779022,
+      "learning_rate": 0.00019894563900537578,
+      "loss": 0.0451,
+      "step": 650
+    },
+    {
+      "epoch": 0.476878612716763,
+      "grad_norm": 0.6114597320556641,
+      "learning_rate": 0.0001988752647226352,
+      "loss": 0.048,
+      "step": 660
+    },
+    {
+      "epoch": 0.4841040462427746,
+      "grad_norm": 0.5920137763023376,
+      "learning_rate": 0.00019880263036077477,
+      "loss": 0.0382,
+      "step": 670
+    },
+    {
+      "epoch": 0.4913294797687861,
+      "grad_norm": 0.43964025378227234,
+      "learning_rate": 0.00019872773758006202,
+      "loss": 0.0566,
+      "step": 680
+    },
+    {
+      "epoch": 0.4985549132947977,
+      "grad_norm": 0.5705758929252625,
+      "learning_rate": 0.0001986505880923874,
+      "loss": 0.0358,
+      "step": 690
+    },
+    {
+      "epoch": 0.5057803468208093,
+      "grad_norm": 0.49037960171699524,
+      "learning_rate": 0.00019857118366122477,
+      "loss": 0.041,
+      "step": 700
+    },
+    {
+      "epoch": 0.5130057803468208,
+      "grad_norm": 0.4990547299385071,
+      "learning_rate": 0.00019848952610159124,
+      "loss": 0.0391,
+      "step": 710
+    },
+    {
+      "epoch": 0.5202312138728323,
+      "grad_norm": 0.5737881660461426,
+      "learning_rate": 0.0001984056172800056,
+      "loss": 0.0381,
+      "step": 720
+    },
+    {
+      "epoch": 0.5274566473988439,
+      "grad_norm": 0.7038660049438477,
+      "learning_rate": 0.00019831945911444584,
+      "loss": 0.0443,
+      "step": 730
+    },
+    {
+      "epoch": 0.5346820809248555,
+      "grad_norm": 0.5946897864341736,
+      "learning_rate": 0.0001982310535743051,
+      "loss": 0.0485,
+      "step": 740
+    },
+    {
+      "epoch": 0.541907514450867,
+      "grad_norm": 0.44982287287712097,
+      "learning_rate": 0.00019814040268034677,
+      "loss": 0.043,
+      "step": 750
+    },
+    {
+      "epoch": 0.5491329479768786,
+      "grad_norm": 0.4566514790058136,
+      "learning_rate": 0.0001980475085046583,
+      "loss": 0.0424,
+      "step": 760
+    },
+    {
+      "epoch": 0.5563583815028902,
+      "grad_norm": 0.4216271936893463,
+      "learning_rate": 0.00019795237317060376,
+      "loss": 0.0376,
+      "step": 770
+    },
+    {
+      "epoch": 0.5635838150289018,
+      "grad_norm": 0.5745754241943359,
+      "learning_rate": 0.00019785499885277541,
+      "loss": 0.0451,
+      "step": 780
+    },
+    {
+      "epoch": 0.5708092485549133,
+      "grad_norm": 0.7188005447387695,
+      "learning_rate": 0.00019775538777694384,
+      "loss": 0.0367,
+      "step": 790
+    },
+    {
+      "epoch": 0.5780346820809249,
+      "grad_norm": 0.24551555514335632,
+      "learning_rate": 0.00019765354222000734,
+      "loss": 0.0395,
+      "step": 800
+    },
+    {
+      "epoch": 0.5852601156069365,
+      "grad_norm": 0.39921292662620544,
+      "learning_rate": 0.0001975494645099395,
+      "loss": 0.0335,
+      "step": 810
+    },
+    {
+      "epoch": 0.5924855491329479,
+      "grad_norm": 0.6041325926780701,
+      "learning_rate": 0.00019744315702573646,
+      "loss": 0.0512,
+      "step": 820
+    },
+    {
+      "epoch": 0.5997109826589595,
+      "grad_norm": 0.3695695698261261,
+      "learning_rate": 0.00019733462219736206,
+      "loss": 0.0386,
+      "step": 830
+    },
+    {
+      "epoch": 0.6069364161849711,
+      "grad_norm": 0.46152517199516296,
+      "learning_rate": 0.00019722386250569258,
+      "loss": 0.0313,
+      "step": 840
+    },
+    {
+      "epoch": 0.6141618497109826,
+      "grad_norm": 0.39752018451690674,
+      "learning_rate": 0.00019711088048245996,
+      "loss": 0.0382,
+      "step": 850
+    },
+    {
+      "epoch": 0.6213872832369942,
+      "grad_norm": 0.5013191103935242,
+      "learning_rate": 0.00019699567871019397,
+      "loss": 0.0452,
+      "step": 860
+    },
+    {
+      "epoch": 0.6286127167630058,
+      "grad_norm": 0.6132786273956299,
+      "learning_rate": 0.0001968782598221631,
+      "loss": 0.0376,
+      "step": 870
+    },
+    {
+      "epoch": 0.6358381502890174,
+      "grad_norm": 0.6125185489654541,
+      "learning_rate": 0.0001967586265023144,
+      "loss": 0.0447,
+      "step": 880
+    },
+    {
+      "epoch": 0.6430635838150289,
+      "grad_norm": 0.5613898634910583,
+      "learning_rate": 0.00019663678148521224,
+      "loss": 0.0442,
+      "step": 890
+    },
+    {
+      "epoch": 0.6502890173410405,
+      "grad_norm": 0.5539945363998413,
+      "learning_rate": 0.0001965127275559755,
+      "loss": 0.041,
+      "step": 900
+    },
+    {
+      "epoch": 0.6575144508670521,
+      "grad_norm": 0.5628465414047241,
+      "learning_rate": 0.00019638646755021434,
+      "loss": 0.0432,
+      "step": 910
+    },
+    {
+      "epoch": 0.6647398843930635,
+      "grad_norm": 0.4599197804927826,
+      "learning_rate": 0.00019625800435396501,
+      "loss": 0.0437,
+      "step": 920
+    },
+    {
+      "epoch": 0.6719653179190751,
+      "grad_norm": 0.6037493944168091,
+      "learning_rate": 0.0001961273409036241,
+      "loss": 0.0504,
+      "step": 930
+    },
+    {
+      "epoch": 0.6791907514450867,
+      "grad_norm": 0.4240657091140747,
+      "learning_rate": 0.0001959944801858813,
+      "loss": 0.0348,
+      "step": 940
+    },
+    {
+      "epoch": 0.6864161849710982,
+      "grad_norm": 0.29284748435020447,
+      "learning_rate": 0.00019585942523765119,
+      "loss": 0.0337,
+      "step": 950
+    },
+    {
+      "epoch": 0.6936416184971098,
+      "grad_norm": 0.4377879500389099,
+      "learning_rate": 0.00019572217914600387,
+      "loss": 0.0446,
+      "step": 960
+    },
+    {
+      "epoch": 0.7008670520231214,
+      "grad_norm": 0.5444635152816772,
+      "learning_rate": 0.0001955827450480942,
+      "loss": 0.0414,
+      "step": 970
+    },
+    {
+      "epoch": 0.708092485549133,
+      "grad_norm": 0.5109856724739075,
+      "learning_rate": 0.0001954411261310904,
+      "loss": 0.038,
+      "step": 980
+    },
+    {
+      "epoch": 0.7153179190751445,
+      "grad_norm": 0.6839922666549683,
+      "learning_rate": 0.00019529732563210094,
+      "loss": 0.0394,
+      "step": 990
+    },
+    {
+      "epoch": 0.7225433526011561,
+      "grad_norm": 0.5683702230453491,
+      "learning_rate": 0.0001951513468381006,
+      "loss": 0.0344,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7297687861271677,
+      "grad_norm": 0.41708916425704956,
+      "learning_rate": 0.0001950031930858554,
+      "loss": 0.0478,
+      "step": 1010
+    },
+    {
+      "epoch": 0.7369942196531792,
+      "grad_norm": 0.3592592477798462,
+      "learning_rate": 0.00019485286776184634,
+      "loss": 0.0366,
+      "step": 1020
+    },
+    {
+      "epoch": 0.7442196531791907,
+      "grad_norm": 0.42531320452690125,
+      "learning_rate": 0.00019470037430219193,
+      "loss": 0.0392,
+      "step": 1030
+    },
+    {
+      "epoch": 0.7514450867052023,
+      "grad_norm": 0.5057926177978516,
+      "learning_rate": 0.0001945457161925696,
+      "loss": 0.0357,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7586705202312138,
+      "grad_norm": 0.26346009969711304,
+      "learning_rate": 0.00019438889696813617,
+      "loss": 0.0331,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7658959537572254,
+      "grad_norm": 0.5742522478103638,
+      "learning_rate": 0.00019422992021344695,
+      "loss": 0.0426,
+      "step": 1060
+    },
+    {
+      "epoch": 0.773121387283237,
+      "grad_norm": 0.3441261351108551,
+      "learning_rate": 0.00019406878956237386,
+      "loss": 0.0305,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7803468208092486,
+      "grad_norm": 0.2888159453868866,
+      "learning_rate": 0.0001939055086980222,
+      "loss": 0.0383,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7875722543352601,
+      "grad_norm": 0.2590026557445526,
+      "learning_rate": 0.00019374008135264673,
+      "loss": 0.0329,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7947976878612717,
+      "grad_norm": 0.33079084753990173,
+      "learning_rate": 0.0001935725113075662,
+      "loss": 0.0295,
+      "step": 1100
+    },
+    {
+      "epoch": 0.8020231213872833,
+      "grad_norm": 0.46009305119514465,
+      "learning_rate": 0.00019340280239307683,
+      "loss": 0.0375,
+      "step": 1110
+    },
+    {
+      "epoch": 0.8092485549132948,
+      "grad_norm": 0.6054324507713318,
+      "learning_rate": 0.00019323095848836498,
+      "loss": 0.0316,
+      "step": 1120
+    },
+    {
+      "epoch": 0.8164739884393064,
+      "grad_norm": 0.6735290884971619,
+      "learning_rate": 0.00019305698352141835,
+      "loss": 0.0392,
+      "step": 1130
+    },
+    {
+      "epoch": 0.8236994219653179,
+      "grad_norm": 0.396467387676239,
+      "learning_rate": 0.0001928808814689361,
+      "loss": 0.0307,
+      "step": 1140
+    },
+    {
+      "epoch": 0.8309248554913294,
+      "grad_norm": 0.6033023595809937,
+      "learning_rate": 0.00019270265635623819,
+      "loss": 0.0365,
+      "step": 1150
+    },
+    {
+      "epoch": 0.838150289017341,
+      "grad_norm": 0.3280772566795349,
+      "learning_rate": 0.0001925223122571732,
+      "loss": 0.0318,
+      "step": 1160
+    },
+    {
+      "epoch": 0.8453757225433526,
+      "grad_norm": 0.5830028653144836,
+      "learning_rate": 0.00019233985329402523,
+      "loss": 0.0408,
+      "step": 1170
+    },
+    {
+      "epoch": 0.8526011560693642,
+      "grad_norm": 0.4894756078720093,
+      "learning_rate": 0.00019215528363741968,
+      "loss": 0.0333,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8598265895953757,
+      "grad_norm": 0.41378799080848694,
+      "learning_rate": 0.000191968607506228,
+      "loss": 0.0365,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8670520231213873,
+      "grad_norm": 0.3788554072380066,
+      "learning_rate": 0.00019177982916747111,
+      "loss": 0.0394,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8742774566473989,
+      "grad_norm": 0.37194758653640747,
+      "learning_rate": 0.00019158895293622198,
+      "loss": 0.0394,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8815028901734104,
+      "grad_norm": 0.602898895740509,
+      "learning_rate": 0.00019139598317550698,
+      "loss": 0.0419,
+      "step": 1220
+    },
+    {
+      "epoch": 0.888728323699422,
+      "grad_norm": 0.38432008028030396,
+      "learning_rate": 0.00019120092429620607,
+      "loss": 0.0283,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8959537572254336,
+      "grad_norm": 0.43441343307495117,
+      "learning_rate": 0.00019100378075695207,
+      "loss": 0.035,
+      "step": 1240
+    },
+    {
+      "epoch": 0.903179190751445,
+      "grad_norm": 0.32012882828712463,
+      "learning_rate": 0.00019080455706402873,
+      "loss": 0.0313,
+      "step": 1250
+    },
+    {
+      "epoch": 0.9104046242774566,
+      "grad_norm": 0.3490202724933624,
+      "learning_rate": 0.00019060325777126767,
+      "loss": 0.0306,
+      "step": 1260
+    },
+    {
+      "epoch": 0.9176300578034682,
+      "grad_norm": 0.44709476828575134,
+      "learning_rate": 0.00019039988747994438,
+      "loss": 0.0296,
+      "step": 1270
+    },
+    {
+      "epoch": 0.9248554913294798,
+      "grad_norm": 0.47073352336883545,
+      "learning_rate": 0.0001901944508386729,
+      "loss": 0.0425,
+      "step": 1280
+    },
+    {
+      "epoch": 0.9320809248554913,
+      "grad_norm": 0.3281887173652649,
+      "learning_rate": 0.0001899869525432997,
+      "loss": 0.0356,
+      "step": 1290
+    },
+    {
+      "epoch": 0.9393063583815029,
+      "grad_norm": 0.3667585551738739,
+      "learning_rate": 0.00018977739733679635,
+      "loss": 0.039,
+      "step": 1300
+    },
+    {
+      "epoch": 0.9465317919075145,
+      "grad_norm": 0.2563478648662567,
+      "learning_rate": 0.00018956579000915103,
+      "loss": 0.0341,
+      "step": 1310
+    },
+    {
+      "epoch": 0.953757225433526,
+      "grad_norm": 0.29779765009880066,
+      "learning_rate": 0.00018935213539725894,
+      "loss": 0.033,
+      "step": 1320
+    },
+    {
+      "epoch": 0.9609826589595376,
+      "grad_norm": 0.35488152503967285,
+      "learning_rate": 0.000189136438384812,
+      "loss": 0.0356,
+      "step": 1330
+    },
+    {
+      "epoch": 0.9682080924855492,
+      "grad_norm": 0.4382874667644501,
+      "learning_rate": 0.00018891870390218705,
+      "loss": 0.0413,
+      "step": 1340
+    },
+    {
+      "epoch": 0.9754335260115607,
+      "grad_norm": 0.36856570839881897,
+      "learning_rate": 0.00018869893692633318,
+      "loss": 0.0288,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9826589595375722,
+      "grad_norm": 0.3422045111656189,
+      "learning_rate": 0.00018847714248065796,
+      "loss": 0.0396,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9898843930635838,
+      "grad_norm": 0.27691277861595154,
+      "learning_rate": 0.00018825332563491258,
+      "loss": 0.0317,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9971098265895953,
+      "grad_norm": 0.3529070317745209,
+      "learning_rate": 0.0001880274915050761,
+      "loss": 0.0362,
+      "step": 1380
+    },
+    {
+      "epoch": 1.004335260115607,
+      "grad_norm": 0.447391152381897,
+      "learning_rate": 0.0001877996452532384,
+      "loss": 0.0461,
+      "step": 1390
+    },
+    {
+      "epoch": 1.0115606936416186,
+      "grad_norm": 0.41350966691970825,
+      "learning_rate": 0.00018756979208748212,
+      "loss": 0.0346,
+      "step": 1400
+    },
+    {
+      "epoch": 1.01878612716763,
+      "grad_norm": 0.4052940011024475,
+      "learning_rate": 0.00018733793726176382,
+      "loss": 0.0467,
+      "step": 1410
+    },
+    {
+      "epoch": 1.0260115606936415,
+      "grad_norm": 0.4326450824737549,
+      "learning_rate": 0.00018710408607579369,
+      "loss": 0.0337,
+      "step": 1420
+    },
+    {
+      "epoch": 1.033236994219653,
+      "grad_norm": 0.6959808468818665,
+      "learning_rate": 0.00018686824387491454,
+      "loss": 0.0414,
+      "step": 1430
+    },
+    {
+      "epoch": 1.0404624277456647,
+      "grad_norm": 0.46198827028274536,
+      "learning_rate": 0.00018663041604997948,
+      "loss": 0.0304,
+      "step": 1440
+    },
+    {
+      "epoch": 1.0476878612716762,
+      "grad_norm": 0.3633591830730438,
+      "learning_rate": 0.00018639060803722895,
+      "loss": 0.0427,
+      "step": 1450
+    },
+    {
+      "epoch": 1.0549132947976878,
+      "grad_norm": 0.417981892824173,
+      "learning_rate": 0.00018614882531816602,
+      "loss": 0.0335,
+      "step": 1460
+    },
+    {
+      "epoch": 1.0621387283236994,
+      "grad_norm": 0.30708634853363037,
+      "learning_rate": 0.00018590507341943163,
+      "loss": 0.027,
+      "step": 1470
+    },
+    {
+      "epoch": 1.069364161849711,
+      "grad_norm": 0.3463022708892822,
+      "learning_rate": 0.00018565935791267784,
+      "loss": 0.0324,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0765895953757225,
+      "grad_norm": 0.294916570186615,
+      "learning_rate": 0.00018541168441444067,
+      "loss": 0.027,
+      "step": 1490
+    },
+    {
+      "epoch": 1.083815028901734,
+      "grad_norm": 0.4176677167415619,
+      "learning_rate": 0.00018516205858601175,
+      "loss": 0.0331,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0910404624277457,
+      "grad_norm": 0.2961694896221161,
+      "learning_rate": 0.00018491048613330867,
+      "loss": 0.0284,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0982658959537572,
+      "grad_norm": 0.33187049627304077,
+      "learning_rate": 0.0001846569728067449,
+      "loss": 0.0287,
+      "step": 1520
+    },
+    {
+      "epoch": 1.1054913294797688,
+      "grad_norm": 0.348763108253479,
+      "learning_rate": 0.0001844015244010981,
+      "loss": 0.0356,
+      "step": 1530
+    },
+    {
+      "epoch": 1.1127167630057804,
+      "grad_norm": 0.5439162850379944,
+      "learning_rate": 0.00018414414675537772,
+      "loss": 0.0257,
+      "step": 1540
+    },
+    {
+      "epoch": 1.119942196531792,
+      "grad_norm": 0.41269758343696594,
+      "learning_rate": 0.00018388484575269164,
+      "loss": 0.0376,
+      "step": 1550
+    },
+    {
+      "epoch": 1.1271676300578035,
+      "grad_norm": 0.38240736722946167,
+      "learning_rate": 0.00018362362732011147,
+      "loss": 0.0317,
+      "step": 1560
+    },
+    {
+      "epoch": 1.134393063583815,
+      "grad_norm": 0.26935675740242004,
+      "learning_rate": 0.00018336049742853733,
+      "loss": 0.0289,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1416184971098267,
+      "grad_norm": 0.32840201258659363,
+      "learning_rate": 0.0001830954620925612,
+      "loss": 0.0309,
+      "step": 1580
+    },
+    {
+      "epoch": 1.1488439306358382,
+      "grad_norm": 0.3386699855327606,
+      "learning_rate": 0.00018282852737032946,
+      "loss": 0.0335,
+      "step": 1590
+    },
+    {
+      "epoch": 1.1560693641618498,
+      "grad_norm": 0.41219276189804077,
+      "learning_rate": 0.00018255969936340443,
+      "loss": 0.0285,
+      "step": 1600
+    },
+    {
+      "epoch": 1.1632947976878611,
+      "grad_norm": 0.5242831110954285,
+      "learning_rate": 0.00018228898421662496,
+      "loss": 0.031,
+      "step": 1610
+    },
+    {
+      "epoch": 1.1705202312138727,
+      "grad_norm": 0.4546612501144409,
+      "learning_rate": 0.00018201638811796592,
+      "loss": 0.0352,
+      "step": 1620
+    },
+    {
+      "epoch": 1.1777456647398843,
+      "grad_norm": 0.3911999762058258,
+      "learning_rate": 0.00018174191729839669,
+      "loss": 0.0371,
+      "step": 1630
+    },
+    {
+      "epoch": 1.1849710982658959,
+      "grad_norm": 0.3058600127696991,
+      "learning_rate": 0.00018146557803173883,
+      "loss": 0.0285,
+      "step": 1640
+    },
+    {
+      "epoch": 1.1921965317919074,
+      "grad_norm": 0.37400105595588684,
+      "learning_rate": 0.00018118737663452267,
+      "loss": 0.0286,
+      "step": 1650
+    },
+    {
+      "epoch": 1.199421965317919,
+      "grad_norm": 0.4513632655143738,
+      "learning_rate": 0.00018090731946584292,
+      "loss": 0.0463,
+      "step": 1660
+    },
+    {
+      "epoch": 1.2066473988439306,
+      "grad_norm": 0.4379500448703766,
+      "learning_rate": 0.0001806254129272132,
+      "loss": 0.0261,
+      "step": 1670
+    },
+    {
+      "epoch": 1.2138728323699421,
+      "grad_norm": 0.4692801237106323,
+      "learning_rate": 0.00018034166346241986,
+      "loss": 0.0323,
+      "step": 1680
+    },
+    {
+      "epoch": 1.2210982658959537,
+      "grad_norm": 0.40697070956230164,
+      "learning_rate": 0.00018005607755737467,
+      "loss": 0.0471,
+      "step": 1690
+    },
+    {
+      "epoch": 1.2283236994219653,
+      "grad_norm": 0.3501347005367279,
+      "learning_rate": 0.00017976866173996652,
+      "loss": 0.0339,
+      "step": 1700
+    },
+    {
+      "epoch": 1.2355491329479769,
+      "grad_norm": 0.2515588104724884,
+      "learning_rate": 0.00017947942257991216,
+      "loss": 0.0327,
+      "step": 1710
+    },
+    {
+      "epoch": 1.2427745664739884,
+      "grad_norm": 0.276937872171402,
+      "learning_rate": 0.0001791883666886062,
+      "loss": 0.03,
+      "step": 1720
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4615573287010193,
+      "learning_rate": 0.0001788955007189698,
+      "loss": 0.0286,
+      "step": 1730
+    },
+    {
+      "epoch": 1.2572254335260116,
+      "grad_norm": 0.5667081475257874,
+      "learning_rate": 0.00017860083136529868,
+      "loss": 0.0326,
+      "step": 1740
+    },
+    {
+      "epoch": 1.2644508670520231,
+      "grad_norm": 0.45345941185951233,
+      "learning_rate": 0.0001783043653631101,
+      "loss": 0.0289,
+      "step": 1750
+    },
+    {
+      "epoch": 1.2716763005780347,
+      "grad_norm": 0.47316843271255493,
+      "learning_rate": 0.00017800610948898895,
+      "loss": 0.0313,
+      "step": 1760
+    },
+    {
+      "epoch": 1.2789017341040463,
+      "grad_norm": 0.25375911593437195,
+      "learning_rate": 0.00017770607056043276,
+      "loss": 0.027,
+      "step": 1770
+    },
+    {
+      "epoch": 1.2861271676300579,
+      "grad_norm": 0.2597872316837311,
+      "learning_rate": 0.00017740425543569588,
+      "loss": 0.0288,
+      "step": 1780
+    },
+    {
+      "epoch": 1.2933526011560694,
+      "grad_norm": 0.38263481855392456,
+      "learning_rate": 0.00017710067101363278,
+      "loss": 0.0337,
+      "step": 1790
+    },
+    {
+      "epoch": 1.300578034682081,
+      "grad_norm": 0.3304251432418823,
+      "learning_rate": 0.00017679532423354034,
+      "loss": 0.0326,
+      "step": 1800
+    },
+    {
+      "epoch": 1.3078034682080926,
+      "grad_norm": 0.30174291133880615,
+      "learning_rate": 0.00017648822207499919,
+      "loss": 0.0294,
+      "step": 1810
+    },
+    {
+      "epoch": 1.3150289017341041,
+      "grad_norm": 0.3475726544857025,
+      "learning_rate": 0.00017617937155771417,
+      "loss": 0.0253,
+      "step": 1820
+    },
+    {
+      "epoch": 1.3222543352601157,
+      "grad_norm": 0.3054395318031311,
+      "learning_rate": 0.00017586877974135393,
+      "loss": 0.0276,
+      "step": 1830
+    },
+    {
+      "epoch": 1.3294797687861273,
+      "grad_norm": 0.28724974393844604,
+      "learning_rate": 0.00017555645372538948,
+      "loss": 0.0297,
+      "step": 1840
+    },
+    {
+      "epoch": 1.3367052023121389,
+      "grad_norm": 0.42262300848960876,
+      "learning_rate": 0.000175242400648932,
+      "loss": 0.0302,
+      "step": 1850
+    },
+    {
+      "epoch": 1.3439306358381504,
+      "grad_norm": 0.29184046387672424,
+      "learning_rate": 0.00017492662769056966,
+      "loss": 0.0299,
+      "step": 1860
+    },
+    {
+      "epoch": 1.3511560693641618,
+      "grad_norm": 0.45122018456459045,
+      "learning_rate": 0.00017460914206820341,
+      "loss": 0.0269,
+      "step": 1870
+    },
+    {
+      "epoch": 1.3583815028901733,
+      "grad_norm": 0.36067941784858704,
+      "learning_rate": 0.0001742899510388821,
+      "loss": 0.0292,
+      "step": 1880
+    },
+    {
+      "epoch": 1.365606936416185,
+      "grad_norm": 0.4859599471092224,
+      "learning_rate": 0.00017396906189863655,
+      "loss": 0.032,
+      "step": 1890
+    },
+    {
+      "epoch": 1.3728323699421965,
+      "grad_norm": 0.36321279406547546,
+      "learning_rate": 0.00017364648198231282,
+      "loss": 0.0439,
+      "step": 1900
+    },
+    {
+      "epoch": 1.380057803468208,
+      "grad_norm": 0.3595023453235626,
+      "learning_rate": 0.00017332221866340455,
+      "loss": 0.0301,
+      "step": 1910
+    },
+    {
+      "epoch": 1.3872832369942196,
+      "grad_norm": 0.26867198944091797,
+      "learning_rate": 0.0001729962793538844,
+      "loss": 0.0284,
+      "step": 1920
+    },
+    {
+      "epoch": 1.3945086705202312,
+      "grad_norm": 0.5771405696868896,
+      "learning_rate": 0.00017266867150403451,
+      "loss": 0.0267,
+      "step": 1930
+    },
+    {
+      "epoch": 1.4017341040462428,
+      "grad_norm": 0.5397071838378906,
+      "learning_rate": 0.00017233940260227647,
+      "loss": 0.0252,
+      "step": 1940
+    },
+    {
+      "epoch": 1.4089595375722543,
+      "grad_norm": 0.4514451324939728,
+      "learning_rate": 0.00017200848017499994,
+      "loss": 0.0333,
+      "step": 1950
+    },
+    {
+      "epoch": 1.416184971098266,
+      "grad_norm": 0.6613513827323914,
+      "learning_rate": 0.00017167591178639065,
+      "loss": 0.0261,
+      "step": 1960
+    },
+    {
+      "epoch": 1.4234104046242775,
+      "grad_norm": 0.47511669993400574,
+      "learning_rate": 0.00017134170503825758,
+      "loss": 0.0366,
+      "step": 1970
+    },
+    {
+      "epoch": 1.430635838150289,
+      "grad_norm": 0.4412175118923187,
+      "learning_rate": 0.000171005867569859,
+      "loss": 0.0297,
+      "step": 1980
+    },
+    {
+      "epoch": 1.4378612716763006,
+      "grad_norm": 0.3008122146129608,
+      "learning_rate": 0.0001706684070577282,
+      "loss": 0.0264,
+      "step": 1990
+    },
+    {
+      "epoch": 1.4450867052023122,
+      "grad_norm": 0.31531664729118347,
+      "learning_rate": 0.00017032933121549762,
+      "loss": 0.0289,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4523121387283238,
+      "grad_norm": 0.4209290146827698,
+      "learning_rate": 0.0001699886477937228,
+      "loss": 0.0247,
+      "step": 2010
+    },
+    {
+      "epoch": 1.4595375722543353,
+      "grad_norm": 0.18282712996006012,
+      "learning_rate": 0.0001696463645797052,
+      "loss": 0.0206,
+      "step": 2020
+    },
+    {
+      "epoch": 1.4667630057803467,
+      "grad_norm": 0.4704382121562958,
+      "learning_rate": 0.00016930248939731407,
+      "loss": 0.0272,
+      "step": 2030
+    },
+    {
+      "epoch": 1.4739884393063583,
+      "grad_norm": 0.2840489447116852,
+      "learning_rate": 0.00016895703010680771,
+      "loss": 0.0263,
+      "step": 2040
+    },
+    {
+      "epoch": 1.4812138728323698,
+      "grad_norm": 0.3051230013370514,
+      "learning_rate": 0.00016860999460465372,
+      "loss": 0.0242,
+      "step": 2050
+    },
+    {
+      "epoch": 1.4884393063583814,
+      "grad_norm": 0.3468949794769287,
+      "learning_rate": 0.00016826139082334868,
+      "loss": 0.0277,
+      "step": 2060
+    },
+    {
+      "epoch": 1.495664739884393,
+      "grad_norm": 0.30354368686676025,
+      "learning_rate": 0.0001679112267312366,
+      "loss": 0.0398,
+      "step": 2070
+    },
+    {
+      "epoch": 1.5028901734104045,
+      "grad_norm": 0.4176887273788452,
+      "learning_rate": 0.00016755951033232692,
+      "loss": 0.0274,
+      "step": 2080
+    },
+    {
+      "epoch": 1.510115606936416,
+      "grad_norm": 0.3101764917373657,
+      "learning_rate": 0.00016720624966611153,
+      "loss": 0.0272,
+      "step": 2090
+    },
+    {
+      "epoch": 1.5173410404624277,
+      "grad_norm": 0.32918646931648254,
+      "learning_rate": 0.00016685145280738106,
+      "loss": 0.0244,
+      "step": 2100
+    },
+    {
+      "epoch": 1.5245664739884393,
+      "grad_norm": 0.3785788416862488,
+      "learning_rate": 0.00016649512786604014,
+      "loss": 0.0284,
+      "step": 2110
+    },
+    {
+      "epoch": 1.5317919075144508,
+      "grad_norm": 0.8121911287307739,
+      "learning_rate": 0.00016613728298692218,
+      "loss": 0.035,
+      "step": 2120
+    },
+    {
+      "epoch": 1.5390173410404624,
+      "grad_norm": 0.2641575038433075,
+      "learning_rate": 0.0001657779263496032,
+      "loss": 0.0316,
+      "step": 2130
+    },
+    {
+      "epoch": 1.546242774566474,
+      "grad_norm": 0.32332223653793335,
+      "learning_rate": 0.00016541706616821474,
+      "loss": 0.0275,
+      "step": 2140
+    },
+    {
+      "epoch": 1.5534682080924855,
+      "grad_norm": 0.39668551087379456,
+      "learning_rate": 0.00016505471069125615,
+      "loss": 0.0281,
+      "step": 2150
+    },
+    {
+      "epoch": 1.560693641618497,
+      "grad_norm": 0.5112664103507996,
+      "learning_rate": 0.00016469086820140616,
+      "loss": 0.0358,
+      "step": 2160
+    },
+    {
+      "epoch": 1.5679190751445087,
+      "grad_norm": 0.27202361822128296,
+      "learning_rate": 0.0001643255470153334,
+      "loss": 0.0242,
+      "step": 2170
+    },
+    {
+      "epoch": 1.5751445086705202,
+      "grad_norm": 0.2680519223213196,
+      "learning_rate": 0.0001639587554835064,
+      "loss": 0.0187,
+      "step": 2180
+    },
+    {
+      "epoch": 1.5823699421965318,
+      "grad_norm": 0.3848129212856293,
+      "learning_rate": 0.0001635905019900026,
+      "loss": 0.0252,
+      "step": 2190
+    },
+    {
+      "epoch": 1.5895953757225434,
+      "grad_norm": 0.29368847608566284,
+      "learning_rate": 0.00016322079495231697,
+      "loss": 0.0267,
+      "step": 2200
+    },
+    {
+      "epoch": 1.596820809248555,
+      "grad_norm": 0.232693612575531,
+      "learning_rate": 0.00016284964282116916,
+      "loss": 0.0256,
+      "step": 2210
+    },
+    {
+      "epoch": 1.6040462427745665,
+      "grad_norm": 0.14202171564102173,
+      "learning_rate": 0.00016247705408031085,
+      "loss": 0.021,
+      "step": 2220
+    },
+    {
+      "epoch": 1.611271676300578,
+      "grad_norm": 0.37467068433761597,
+      "learning_rate": 0.00016210303724633143,
+      "loss": 0.0267,
+      "step": 2230
+    },
+    {
+      "epoch": 1.6184971098265897,
+      "grad_norm": 0.3563550114631653,
+      "learning_rate": 0.00016172760086846343,
+      "loss": 0.0259,
+      "step": 2240
+    },
+    {
+      "epoch": 1.6257225433526012,
+      "grad_norm": 0.31626465916633606,
+      "learning_rate": 0.00016135075352838736,
+      "loss": 0.0267,
+      "step": 2250
+    },
+    {
+      "epoch": 1.6329479768786128,
+      "grad_norm": 0.471219003200531,
+      "learning_rate": 0.00016097250384003507,
+      "loss": 0.0411,
+      "step": 2260
+    },
+    {
+      "epoch": 1.6401734104046244,
+      "grad_norm": 0.4073711335659027,
+      "learning_rate": 0.00016059286044939328,
+      "loss": 0.0295,
+      "step": 2270
+    },
+    {
+      "epoch": 1.647398843930636,
+      "grad_norm": 0.3200165331363678,
+      "learning_rate": 0.00016021183203430582,
+      "loss": 0.0285,
+      "step": 2280
+    },
+    {
+      "epoch": 1.6546242774566475,
+      "grad_norm": 0.25765568017959595,
+      "learning_rate": 0.00015982942730427514,
+      "loss": 0.0293,
+      "step": 2290
+    },
+    {
+      "epoch": 1.661849710982659,
+      "grad_norm": 0.2762660086154938,
+      "learning_rate": 0.0001594456550002634,
+      "loss": 0.031,
+      "step": 2300
+    },
+    {
+      "epoch": 1.6690751445086707,
+      "grad_norm": 0.39115095138549805,
+      "learning_rate": 0.00015906052389449258,
+      "loss": 0.0231,
+      "step": 2310
+    },
+    {
+      "epoch": 1.6763005780346822,
+      "grad_norm": 0.4150591790676117,
+      "learning_rate": 0.00015867404279024396,
+      "loss": 0.0321,
+      "step": 2320
+    },
+    {
+      "epoch": 1.6835260115606936,
+      "grad_norm": 0.2964015007019043,
+      "learning_rate": 0.00015828622052165706,
+      "loss": 0.0242,
+      "step": 2330
+    },
+    {
+      "epoch": 1.6907514450867052,
+      "grad_norm": 0.3821788430213928,
+      "learning_rate": 0.00015789706595352733,
+      "loss": 0.0258,
+      "step": 2340
+    },
+    {
+      "epoch": 1.6979768786127167,
+      "grad_norm": 0.39060547947883606,
+      "learning_rate": 0.00015750658798110404,
+      "loss": 0.0288,
+      "step": 2350
+    },
+    {
+      "epoch": 1.7052023121387283,
+      "grad_norm": 0.4912872016429901,
+      "learning_rate": 0.00015711479552988651,
+      "loss": 0.0279,
+      "step": 2360
+    },
+    {
+      "epoch": 1.7124277456647399,
+      "grad_norm": 0.306397944688797,
+      "learning_rate": 0.00015672169755542032,
+      "loss": 0.0332,
+      "step": 2370
+    },
+    {
+      "epoch": 1.7196531791907514,
+      "grad_norm": 0.3365071415901184,
+      "learning_rate": 0.0001563273030430925,
+      "loss": 0.0237,
+      "step": 2380
+    },
+    {
+      "epoch": 1.726878612716763,
+      "grad_norm": 0.2674650251865387,
+      "learning_rate": 0.00015593162100792623,
+      "loss": 0.0277,
+      "step": 2390
+    },
+    {
+      "epoch": 1.7341040462427746,
+      "grad_norm": 0.21112091839313507,
+      "learning_rate": 0.00015553466049437472,
+      "loss": 0.026,
+      "step": 2400
+    },
+    {
+      "epoch": 1.7413294797687862,
+      "grad_norm": 0.3916032612323761,
+      "learning_rate": 0.0001551364305761145,
+      "loss": 0.028,
+      "step": 2410
+    },
+    {
+      "epoch": 1.7485549132947977,
+      "grad_norm": 0.6372588872909546,
+      "learning_rate": 0.000154736940355838,
+      "loss": 0.0265,
+      "step": 2420
+    },
+    {
+      "epoch": 1.7557803468208093,
+      "grad_norm": 0.2447371780872345,
+      "learning_rate": 0.00015433619896504548,
+      "loss": 0.0233,
+      "step": 2430
+    },
+    {
+      "epoch": 1.7630057803468207,
+      "grad_norm": 0.3194689452648163,
+      "learning_rate": 0.00015393421556383628,
+      "loss": 0.026,
+      "step": 2440
+    },
+    {
+      "epoch": 1.7702312138728322,
+      "grad_norm": 0.2731267511844635,
+      "learning_rate": 0.00015353099934069952,
+      "loss": 0.0324,
+      "step": 2450
+    },
+    {
+      "epoch": 1.7774566473988438,
+      "grad_norm": 0.2314785122871399,
+      "learning_rate": 0.00015312655951230398,
+      "loss": 0.0249,
+      "step": 2460
+    },
+    {
+      "epoch": 1.7846820809248554,
+      "grad_norm": 0.39271441102027893,
+      "learning_rate": 0.00015272090532328752,
+      "loss": 0.026,
+      "step": 2470
+    },
+    {
+      "epoch": 1.791907514450867,
+      "grad_norm": 0.26557862758636475,
+      "learning_rate": 0.00015231404604604558,
+      "loss": 0.0257,
+      "step": 2480
+    },
+    {
+      "epoch": 1.7991329479768785,
+      "grad_norm": 0.33000999689102173,
+      "learning_rate": 0.00015190599098051952,
+      "loss": 0.0309,
+      "step": 2490
+    },
+    {
+      "epoch": 1.80635838150289,
+      "grad_norm": 0.2772347331047058,
+      "learning_rate": 0.00015149674945398383,
+      "loss": 0.032,
+      "step": 2500
+    },
+    {
+      "epoch": 1.8135838150289016,
+      "grad_norm": 0.24811744689941406,
+      "learning_rate": 0.00015108633082083295,
+      "loss": 0.0312,
+      "step": 2510
+    },
+    {
+      "epoch": 1.8208092485549132,
+      "grad_norm": 0.25412219762802124,
+      "learning_rate": 0.00015067474446236746,
+      "loss": 0.0234,
+      "step": 2520
+    },
+    {
+      "epoch": 1.8280346820809248,
+      "grad_norm": 0.22590336203575134,
+      "learning_rate": 0.0001502619997865798,
+      "loss": 0.0274,
+      "step": 2530
+    },
+    {
+      "epoch": 1.8352601156069364,
+      "grad_norm": 0.5687700510025024,
+      "learning_rate": 0.000149848106227939,
+      "loss": 0.036,
+      "step": 2540
+    },
+    {
+      "epoch": 1.842485549132948,
+      "grad_norm": 0.3410959541797638,
+      "learning_rate": 0.00014943307324717507,
+      "loss": 0.0251,
+      "step": 2550
+    },
+    {
+      "epoch": 1.8497109826589595,
+      "grad_norm": 0.4201713800430298,
+      "learning_rate": 0.0001490169103310629,
+      "loss": 0.0341,
+      "step": 2560
+    },
+    {
+      "epoch": 1.856936416184971,
+      "grad_norm": 0.31435415148735046,
+      "learning_rate": 0.00014859962699220522,
+      "loss": 0.0285,
+      "step": 2570
+    },
+    {
+      "epoch": 1.8641618497109826,
+      "grad_norm": 0.6726494431495667,
+      "learning_rate": 0.00014818123276881531,
+      "loss": 0.0331,
+      "step": 2580
+    },
+    {
+      "epoch": 1.8713872832369942,
+      "grad_norm": 0.25882837176322937,
+      "learning_rate": 0.00014776173722449892,
+      "loss": 0.0241,
+      "step": 2590
+    },
+    {
+      "epoch": 1.8786127167630058,
+      "grad_norm": 0.34781867265701294,
+      "learning_rate": 0.00014734114994803562,
+      "loss": 0.0358,
+      "step": 2600
+    },
+    {
+      "epoch": 1.8858381502890174,
+      "grad_norm": 0.3704645335674286,
+      "learning_rate": 0.0001469194805531597,
+      "loss": 0.0276,
+      "step": 2610
+    },
+    {
+      "epoch": 1.893063583815029,
+      "grad_norm": 0.3247392177581787,
+      "learning_rate": 0.00014649673867834037,
+      "loss": 0.0247,
+      "step": 2620
+    },
+    {
+      "epoch": 1.9002890173410405,
+      "grad_norm": 0.38688188791275024,
+      "learning_rate": 0.00014607293398656148,
+      "loss": 0.0319,
+      "step": 2630
+    },
+    {
+      "epoch": 1.907514450867052,
+      "grad_norm": 0.41135814785957336,
+      "learning_rate": 0.00014564807616510055,
+      "loss": 0.034,
+      "step": 2640
+    },
+    {
+      "epoch": 1.9147398843930636,
+      "grad_norm": 0.3988231122493744,
+      "learning_rate": 0.0001452221749253075,
+      "loss": 0.0241,
+      "step": 2650
+    },
+    {
+      "epoch": 1.9219653179190752,
+      "grad_norm": 0.22677220404148102,
+      "learning_rate": 0.00014479524000238258,
+      "loss": 0.0232,
+      "step": 2660
+    },
+    {
+      "epoch": 1.9291907514450868,
+      "grad_norm": 0.3277333080768585,
+      "learning_rate": 0.0001443672811551538,
+      "loss": 0.0317,
+      "step": 2670
+    },
+    {
+      "epoch": 1.9364161849710984,
+      "grad_norm": 0.5304914712905884,
+      "learning_rate": 0.00014393830816585392,
+      "loss": 0.029,
+      "step": 2680
+    },
+    {
+      "epoch": 1.94364161849711,
+      "grad_norm": 0.23999623954296112,
+      "learning_rate": 0.00014350833083989686,
+      "loss": 0.0207,
+      "step": 2690
+    },
+    {
+      "epoch": 1.9508670520231215,
+      "grad_norm": 0.250896692276001,
+      "learning_rate": 0.00014307735900565355,
+      "loss": 0.0283,
+      "step": 2700
+    },
+    {
+      "epoch": 1.958092485549133,
+      "grad_norm": 0.5534538626670837,
+      "learning_rate": 0.00014264540251422722,
+      "loss": 0.023,
+      "step": 2710
+    },
+    {
+      "epoch": 1.9653179190751446,
+      "grad_norm": 0.309709370136261,
+      "learning_rate": 0.0001422124712392284,
+      "loss": 0.0278,
+      "step": 2720
+    },
+    {
+      "epoch": 1.9725433526011562,
+      "grad_norm": 0.37922394275665283,
+      "learning_rate": 0.00014177857507654892,
+      "loss": 0.0174,
+      "step": 2730
+    },
+    {
+      "epoch": 1.9797687861271678,
+      "grad_norm": 0.29452425241470337,
+      "learning_rate": 0.0001413437239441361,
+      "loss": 0.023,
+      "step": 2740
+    },
+    {
+      "epoch": 1.9869942196531793,
+      "grad_norm": 0.2415676862001419,
+      "learning_rate": 0.0001409079277817657,
+      "loss": 0.0197,
+      "step": 2750
+    },
+    {
+      "epoch": 1.9942196531791907,
+      "grad_norm": 0.31632542610168457,
+      "learning_rate": 0.00014047119655081493,
+      "loss": 0.0232,
+      "step": 2760
+    },
+    {
+      "epoch": 2.0014450867052025,
+      "grad_norm": 0.32267820835113525,
+      "learning_rate": 0.0001400335402340347,
+      "loss": 0.0205,
+      "step": 2770
+    },
+    {
+      "epoch": 2.008670520231214,
+      "grad_norm": 0.2673828601837158,
+      "learning_rate": 0.00013959496883532134,
+      "loss": 0.0241,
+      "step": 2780
+    },
+    {
+      "epoch": 2.0158959537572256,
+      "grad_norm": 0.2748425602912903,
+      "learning_rate": 0.00013915549237948808,
+      "loss": 0.0239,
+      "step": 2790
+    },
+    {
+      "epoch": 2.023121387283237,
+      "grad_norm": 0.2656760513782501,
+      "learning_rate": 0.00013871512091203587,
+      "loss": 0.0268,
+      "step": 2800
+    },
+    {
+      "epoch": 2.0303468208092488,
+      "grad_norm": 0.38731709122657776,
+      "learning_rate": 0.00013827386449892366,
+      "loss": 0.0233,
+      "step": 2810
+    },
+    {
+      "epoch": 2.03757225433526,
+      "grad_norm": 0.393639475107193,
+      "learning_rate": 0.0001378317332263384,
+      "loss": 0.0205,
+      "step": 2820
+    },
+    {
+      "epoch": 2.0447976878612715,
+      "grad_norm": 0.32896026968955994,
+      "learning_rate": 0.00013738873720046447,
+      "loss": 0.0284,
+      "step": 2830
+    },
+    {
+      "epoch": 2.052023121387283,
+      "grad_norm": 0.2494177371263504,
+      "learning_rate": 0.00013694488654725277,
+      "loss": 0.0201,
+      "step": 2840
+    },
+    {
+      "epoch": 2.0592485549132946,
+      "grad_norm": 0.3207587003707886,
+      "learning_rate": 0.000136500191412189,
+      "loss": 0.0235,
+      "step": 2850
+    },
+    {
+      "epoch": 2.066473988439306,
+      "grad_norm": 0.38535335659980774,
+      "learning_rate": 0.00013605466196006208,
+      "loss": 0.0234,
+      "step": 2860
+    },
+    {
+      "epoch": 2.0736994219653178,
+      "grad_norm": 0.3191238045692444,
+      "learning_rate": 0.00013560830837473163,
+      "loss": 0.0248,
+      "step": 2870
+    },
+    {
+      "epoch": 2.0809248554913293,
+      "grad_norm": 0.4017016589641571,
+      "learning_rate": 0.00013516114085889507,
+      "loss": 0.0223,
+      "step": 2880
+    },
+    {
+      "epoch": 2.088150289017341,
+      "grad_norm": 0.1604500561952591,
+      "learning_rate": 0.00013471316963385466,
+      "loss": 0.023,
+      "step": 2890
+    },
+    {
+      "epoch": 2.0953757225433525,
+      "grad_norm": 0.4814695119857788,
+      "learning_rate": 0.00013426440493928383,
+      "loss": 0.0263,
+      "step": 2900
+    },
+    {
+      "epoch": 2.102601156069364,
+      "grad_norm": 0.2005733698606491,
+      "learning_rate": 0.0001338148570329929,
+      "loss": 0.0225,
+      "step": 2910
+    },
+    {
+      "epoch": 2.1098265895953756,
+      "grad_norm": 0.25928524136543274,
+      "learning_rate": 0.00013336453619069476,
+      "loss": 0.0263,
+      "step": 2920
+    },
+    {
+      "epoch": 2.117052023121387,
+      "grad_norm": 0.44216403365135193,
+      "learning_rate": 0.00013291345270577008,
+      "loss": 0.0193,
+      "step": 2930
+    },
+    {
+      "epoch": 2.1242774566473988,
+      "grad_norm": 0.31755897402763367,
+      "learning_rate": 0.00013246161688903186,
+      "loss": 0.0223,
+      "step": 2940
+    },
+    {
+      "epoch": 2.1315028901734103,
+      "grad_norm": 0.3917829990386963,
+      "learning_rate": 0.00013200903906848978,
+      "loss": 0.0317,
+      "step": 2950
+    },
+    {
+      "epoch": 2.138728323699422,
+      "grad_norm": 0.3507084250450134,
+      "learning_rate": 0.00013155572958911424,
+      "loss": 0.0231,
+      "step": 2960
+    },
+    {
+      "epoch": 2.1459537572254335,
+      "grad_norm": 0.39264169335365295,
+      "learning_rate": 0.00013110169881259978,
+      "loss": 0.0244,
+      "step": 2970
+    },
+    {
+      "epoch": 2.153179190751445,
+      "grad_norm": 0.47414886951446533,
+      "learning_rate": 0.00013064695711712817,
+      "loss": 0.0218,
+      "step": 2980
+    },
+    {
+      "epoch": 2.1604046242774566,
+      "grad_norm": 0.4622926115989685,
+      "learning_rate": 0.00013019151489713143,
+      "loss": 0.0243,
+      "step": 2990
+    },
+    {
+      "epoch": 2.167630057803468,
+      "grad_norm": 0.35504287481307983,
+      "learning_rate": 0.00012973538256305398,
+      "loss": 0.0355,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1748554913294798,
+      "grad_norm": 0.2666931748390198,
+      "learning_rate": 0.00012927857054111487,
+      "loss": 0.0245,
+      "step": 3010
+    },
+    {
+      "epoch": 2.1820809248554913,
+      "grad_norm": 0.30214881896972656,
+      "learning_rate": 0.0001288210892730693,
+      "loss": 0.0233,
+      "step": 3020
+    },
+    {
+      "epoch": 2.189306358381503,
+      "grad_norm": 0.28385064005851746,
+      "learning_rate": 0.00012836294921597004,
+      "loss": 0.0215,
+      "step": 3030
+    },
+    {
+      "epoch": 2.1965317919075145,
+      "grad_norm": 0.2620534896850586,
+      "learning_rate": 0.00012790416084192843,
+      "loss": 0.0189,
+      "step": 3040
+    },
+    {
+      "epoch": 2.203757225433526,
+      "grad_norm": 0.25245940685272217,
+      "learning_rate": 0.00012744473463787486,
+      "loss": 0.0232,
+      "step": 3050
+    },
+    {
+      "epoch": 2.2109826589595376,
+      "grad_norm": 0.37120312452316284,
+      "learning_rate": 0.00012698468110531927,
+      "loss": 0.0212,
+      "step": 3060
+    },
+    {
+      "epoch": 2.218208092485549,
+      "grad_norm": 0.287727415561676,
+      "learning_rate": 0.00012652401076011097,
+      "loss": 0.0183,
+      "step": 3070
+    },
+    {
+      "epoch": 2.2254335260115607,
+      "grad_norm": 0.32101157307624817,
+      "learning_rate": 0.00012606273413219823,
+      "loss": 0.0296,
+      "step": 3080
+    },
+    {
+      "epoch": 2.2326589595375723,
+      "grad_norm": 0.15693742036819458,
+      "learning_rate": 0.00012560086176538775,
+      "loss": 0.0256,
+      "step": 3090
+    },
+    {
+      "epoch": 2.239884393063584,
+      "grad_norm": 0.21965362131595612,
+      "learning_rate": 0.00012513840421710354,
+      "loss": 0.0292,
+      "step": 3100
+    },
+    {
+      "epoch": 2.2471098265895955,
+      "grad_norm": 0.2884056568145752,
+      "learning_rate": 0.00012467537205814559,
+      "loss": 0.0196,
+      "step": 3110
+    },
+    {
+      "epoch": 2.254335260115607,
+      "grad_norm": 0.2577861249446869,
+      "learning_rate": 0.0001242117758724483,
+      "loss": 0.0204,
+      "step": 3120
+    },
+    {
+      "epoch": 2.2615606936416186,
+      "grad_norm": 0.3044709861278534,
+      "learning_rate": 0.00012374762625683844,
+      "loss": 0.0225,
+      "step": 3130
+    },
+    {
+      "epoch": 2.26878612716763,
+      "grad_norm": 0.30571049451828003,
+      "learning_rate": 0.00012328293382079318,
+      "loss": 0.0198,
+      "step": 3140
+    },
+    {
+      "epoch": 2.2760115606936417,
+      "grad_norm": 0.20210671424865723,
+      "learning_rate": 0.00012281770918619732,
+      "loss": 0.0209,
+      "step": 3150
+    },
+    {
+      "epoch": 2.2832369942196533,
+      "grad_norm": 0.4498327374458313,
+      "learning_rate": 0.00012235196298710055,
+      "loss": 0.0203,
+      "step": 3160
+    },
+    {
+      "epoch": 2.290462427745665,
+      "grad_norm": 0.33246350288391113,
+      "learning_rate": 0.00012188570586947446,
+      "loss": 0.0205,
+      "step": 3170
+    },
+    {
+      "epoch": 2.2976878612716765,
+      "grad_norm": 0.24479849636554718,
+      "learning_rate": 0.00012141894849096919,
+      "loss": 0.0232,
+      "step": 3180
+    },
+    {
+      "epoch": 2.304913294797688,
+      "grad_norm": 0.3892286717891693,
+      "learning_rate": 0.00012095170152066979,
+      "loss": 0.0179,
+      "step": 3190
+    },
+    {
+      "epoch": 2.3121387283236996,
+      "grad_norm": 0.3496164381504059,
+      "learning_rate": 0.00012048397563885224,
+      "loss": 0.0246,
+      "step": 3200
+    },
+    {
+      "epoch": 2.319364161849711,
+      "grad_norm": 0.38923704624176025,
+      "learning_rate": 0.00012001578153673951,
+      "loss": 0.018,
+      "step": 3210
+    },
+    {
+      "epoch": 2.3265895953757223,
+      "grad_norm": 0.2374027520418167,
+      "learning_rate": 0.00011954712991625707,
+      "loss": 0.0201,
+      "step": 3220
+    },
+    {
+      "epoch": 2.333815028901734,
+      "grad_norm": 0.36562487483024597,
+      "learning_rate": 0.00011907803148978832,
+      "loss": 0.0215,
+      "step": 3230
+    },
+    {
+      "epoch": 2.3410404624277454,
+      "grad_norm": 0.19129693508148193,
+      "learning_rate": 0.00011860849697992966,
+      "loss": 0.0186,
+      "step": 3240
+    },
+    {
+      "epoch": 2.348265895953757,
+      "grad_norm": 0.24849437177181244,
+      "learning_rate": 0.00011813853711924543,
+      "loss": 0.0211,
+      "step": 3250
+    },
+    {
+      "epoch": 2.3554913294797686,
+      "grad_norm": 0.24717555940151215,
+      "learning_rate": 0.0001176681626500226,
+      "loss": 0.0183,
+      "step": 3260
+    },
+    {
+      "epoch": 2.36271676300578,
+      "grad_norm": 0.32142990827560425,
+      "learning_rate": 0.00011719738432402525,
+      "loss": 0.0228,
+      "step": 3270
+    },
+    {
+      "epoch": 2.3699421965317917,
+      "grad_norm": 0.22533449530601501,
+      "learning_rate": 0.00011672621290224875,
+      "loss": 0.017,
+      "step": 3280
+    },
+    {
+      "epoch": 2.3771676300578033,
+      "grad_norm": 0.23009657859802246,
+      "learning_rate": 0.00011625465915467376,
+      "loss": 0.0213,
+      "step": 3290
+    },
+    {
+      "epoch": 2.384393063583815,
+      "grad_norm": 0.3865809142589569,
+      "learning_rate": 0.0001157827338600201,
+      "loss": 0.031,
+      "step": 3300
+    },
+    {
+      "epoch": 2.3916184971098264,
+      "grad_norm": 0.20559871196746826,
+      "learning_rate": 0.00011531044780550044,
+      "loss": 0.0162,
+      "step": 3310
+    },
+    {
+      "epoch": 2.398843930635838,
+      "grad_norm": 0.17667371034622192,
+      "learning_rate": 0.00011483781178657367,
+      "loss": 0.0174,
+      "step": 3320
+    },
+    {
+      "epoch": 2.4060693641618496,
+      "grad_norm": 0.19531908631324768,
+      "learning_rate": 0.00011436483660669797,
+      "loss": 0.0223,
+      "step": 3330
+    },
+    {
+      "epoch": 2.413294797687861,
+      "grad_norm": 0.24337828159332275,
+      "learning_rate": 0.0001138915330770842,
+      "loss": 0.02,
+      "step": 3340
+    },
+    {
+      "epoch": 2.4205202312138727,
+      "grad_norm": 0.224623441696167,
+      "learning_rate": 0.00011341791201644853,
+      "loss": 0.019,
+      "step": 3350
+    },
+    {
+      "epoch": 2.4277456647398843,
+      "grad_norm": 0.43612486124038696,
+      "learning_rate": 0.00011294398425076514,
+      "loss": 0.0252,
+      "step": 3360
+    },
+    {
+      "epoch": 2.434971098265896,
+      "grad_norm": 0.23925966024398804,
+      "learning_rate": 0.00011246976061301895,
+      "loss": 0.0239,
+      "step": 3370
+    },
+    {
+      "epoch": 2.4421965317919074,
+      "grad_norm": 0.20666956901550293,
+      "learning_rate": 0.00011199525194295786,
+      "loss": 0.0192,
+      "step": 3380
+    },
+    {
+      "epoch": 2.449421965317919,
+      "grad_norm": 0.18628180027008057,
+      "learning_rate": 0.00011152046908684497,
+      "loss": 0.0165,
+      "step": 3390
+    },
+    {
+      "epoch": 2.4566473988439306,
+      "grad_norm": 0.46877655386924744,
+      "learning_rate": 0.00011104542289721074,
+      "loss": 0.0263,
+      "step": 3400
+    },
+    {
+      "epoch": 2.463872832369942,
+      "grad_norm": 0.28481748700141907,
+      "learning_rate": 0.0001105701242326048,
+      "loss": 0.0205,
+      "step": 3410
+    },
+    {
+      "epoch": 2.4710982658959537,
+      "grad_norm": 0.30410829186439514,
+      "learning_rate": 0.00011009458395734795,
+      "loss": 0.0232,
+      "step": 3420
+    },
+    {
+      "epoch": 2.4783236994219653,
+      "grad_norm": 0.26160645484924316,
+      "learning_rate": 0.00010961881294128354,
+      "loss": 0.0192,
+      "step": 3430
+    },
+    {
+      "epoch": 2.485549132947977,
+      "grad_norm": 0.5035533308982849,
+      "learning_rate": 0.00010914282205952929,
+      "loss": 0.0319,
+      "step": 3440
+    },
+    {
+      "epoch": 2.4927745664739884,
+      "grad_norm": 0.3156908452510834,
+      "learning_rate": 0.00010866662219222854,
+      "loss": 0.0217,
+      "step": 3450
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.3060540556907654,
+      "learning_rate": 0.00010819022422430163,
+      "loss": 0.0207,
+      "step": 3460
+    },
+    {
+      "epoch": 2.5072254335260116,
+      "grad_norm": 0.2347673624753952,
+      "learning_rate": 0.00010771363904519693,
+      "loss": 0.0228,
+      "step": 3470
+    },
+    {
+      "epoch": 2.514450867052023,
+      "grad_norm": 0.35762542486190796,
+      "learning_rate": 0.00010723687754864226,
+      "loss": 0.0161,
+      "step": 3480
+    },
+    {
+      "epoch": 2.5216763005780347,
+      "grad_norm": 0.29743993282318115,
+      "learning_rate": 0.00010675995063239557,
+      "loss": 0.0198,
+      "step": 3490
+    },
+    {
+      "epoch": 2.5289017341040463,
+      "grad_norm": 0.2535516321659088,
+      "learning_rate": 0.00010628286919799592,
+      "loss": 0.0213,
+      "step": 3500
+    },
+    {
+      "epoch": 2.536127167630058,
+      "grad_norm": 0.21419911086559296,
+      "learning_rate": 0.00010580564415051439,
+      "loss": 0.026,
+      "step": 3510
+    },
+    {
+      "epoch": 2.5433526011560694,
+      "grad_norm": 0.24240179359912872,
+      "learning_rate": 0.00010532828639830479,
+      "loss": 0.0223,
+      "step": 3520
+    },
+    {
+      "epoch": 2.550578034682081,
+      "grad_norm": 0.29356133937835693,
+      "learning_rate": 0.00010485080685275415,
+      "loss": 0.0188,
+      "step": 3530
+    },
+    {
+      "epoch": 2.5578034682080926,
+      "grad_norm": 0.25931376218795776,
+      "learning_rate": 0.00010437321642803352,
+      "loss": 0.0239,
+      "step": 3540
+    },
+    {
+      "epoch": 2.565028901734104,
+      "grad_norm": 0.276324063539505,
+      "learning_rate": 0.00010389552604084846,
+      "loss": 0.0249,
+      "step": 3550
+    },
+    {
+      "epoch": 2.5722543352601157,
+      "grad_norm": 0.463504433631897,
+      "learning_rate": 0.00010341774661018939,
+      "loss": 0.0231,
+      "step": 3560
+    },
+    {
+      "epoch": 2.5794797687861273,
+      "grad_norm": 0.3001970946788788,
+      "learning_rate": 0.00010293988905708205,
+      "loss": 0.0243,
+      "step": 3570
+    },
+    {
+      "epoch": 2.586705202312139,
+      "grad_norm": 0.3376437723636627,
+      "learning_rate": 0.00010246196430433797,
+      "loss": 0.0234,
+      "step": 3580
+    },
+    {
+      "epoch": 2.5939306358381504,
+      "grad_norm": 0.3224913775920868,
+      "learning_rate": 0.0001019839832763047,
+      "loss": 0.0235,
+      "step": 3590
+    },
+    {
+      "epoch": 2.601156069364162,
+      "grad_norm": 0.19860439002513885,
+      "learning_rate": 0.00010150595689861606,
+      "loss": 0.0231,
+      "step": 3600
+    },
+    {
+      "epoch": 2.6083815028901736,
+      "grad_norm": 0.3264404237270355,
+      "learning_rate": 0.00010102789609794256,
+      "loss": 0.0186,
+      "step": 3610
+    },
+    {
+      "epoch": 2.615606936416185,
+      "grad_norm": 0.31927433609962463,
+      "learning_rate": 0.0001005498118017415,
+      "loss": 0.0212,
+      "step": 3620
+    },
+    {
+      "epoch": 2.6228323699421967,
+      "grad_norm": 0.20473706722259521,
+      "learning_rate": 0.0001000717149380072,
+      "loss": 0.0166,
+      "step": 3630
+    },
+    {
+      "epoch": 2.6300578034682083,
+      "grad_norm": 0.3435823321342468,
+      "learning_rate": 9.959361643502135e-05,
+      "loss": 0.0134,
+      "step": 3640
+    },
+    {
+      "epoch": 2.63728323699422,
+      "grad_norm": 0.3076665997505188,
+      "learning_rate": 9.911552722110297e-05,
+      "loss": 0.0192,
+      "step": 3650
+    },
+    {
+      "epoch": 2.6445086705202314,
+      "grad_norm": 0.3391270339488983,
+      "learning_rate": 9.863745822435893e-05,
+      "loss": 0.0171,
+      "step": 3660
+    },
+    {
+      "epoch": 2.651734104046243,
+      "grad_norm": 0.23569989204406738,
+      "learning_rate": 9.815942037243377e-05,
+      "loss": 0.0176,
+      "step": 3670
+    },
+    {
+      "epoch": 2.6589595375722546,
+      "grad_norm": 0.20910443365573883,
+      "learning_rate": 9.768142459226035e-05,
+      "loss": 0.0189,
+      "step": 3680
+    },
+    {
+      "epoch": 2.666184971098266,
+      "grad_norm": 0.23779253661632538,
+      "learning_rate": 9.720348180980966e-05,
+      "loss": 0.0157,
+      "step": 3690
+    },
+    {
+      "epoch": 2.6734104046242777,
+      "grad_norm": 0.20809420943260193,
+      "learning_rate": 9.672560294984137e-05,
+      "loss": 0.0171,
+      "step": 3700
+    },
+    {
+      "epoch": 2.6806358381502893,
+      "grad_norm": 0.44309884309768677,
+      "learning_rate": 9.624779893565405e-05,
+      "loss": 0.0245,
+      "step": 3710
+    },
+    {
+      "epoch": 2.687861271676301,
+      "grad_norm": 0.40095505118370056,
+      "learning_rate": 9.577008068883533e-05,
+      "loss": 0.0242,
+      "step": 3720
+    },
+    {
+      "epoch": 2.6950867052023124,
+      "grad_norm": 0.3243694007396698,
+      "learning_rate": 9.529245912901254e-05,
+      "loss": 0.0205,
+      "step": 3730
+    },
+    {
+      "epoch": 2.7023121387283235,
+      "grad_norm": 0.2903628945350647,
+      "learning_rate": 9.481494517360283e-05,
+      "loss": 0.023,
+      "step": 3740
+    },
+    {
+      "epoch": 2.709537572254335,
+      "grad_norm": 0.25138890743255615,
+      "learning_rate": 9.433754973756371e-05,
+      "loss": 0.0226,
+      "step": 3750
+    },
+    {
+      "epoch": 2.7167630057803467,
+      "grad_norm": 0.2902311682701111,
+      "learning_rate": 9.386028373314382e-05,
+      "loss": 0.0158,
+      "step": 3760
+    },
+    {
+      "epoch": 2.7239884393063583,
+      "grad_norm": 0.18118108808994293,
+      "learning_rate": 9.338315806963293e-05,
+      "loss": 0.0141,
+      "step": 3770
+    },
+    {
+      "epoch": 2.73121387283237,
+      "grad_norm": 0.24059826135635376,
+      "learning_rate": 9.290618365311322e-05,
+      "loss": 0.017,
+      "step": 3780
+    },
+    {
+      "epoch": 2.7384393063583814,
+      "grad_norm": 0.28333336114883423,
+      "learning_rate": 9.242937138620951e-05,
+      "loss": 0.0206,
+      "step": 3790
+    },
+    {
+      "epoch": 2.745664739884393,
+      "grad_norm": 0.31268417835235596,
+      "learning_rate": 9.195273216784022e-05,
+      "loss": 0.0233,
+      "step": 3800
+    },
+    {
+      "epoch": 2.7528901734104045,
+      "grad_norm": 0.28576889634132385,
+      "learning_rate": 9.147627689296836e-05,
+      "loss": 0.0221,
+      "step": 3810
+    },
+    {
+      "epoch": 2.760115606936416,
+      "grad_norm": 0.2438884973526001,
+      "learning_rate": 9.100001645235227e-05,
+      "loss": 0.0178,
+      "step": 3820
+    },
+    {
+      "epoch": 2.7673410404624277,
+      "grad_norm": 0.27602580189704895,
+      "learning_rate": 9.052396173229688e-05,
+      "loss": 0.0184,
+      "step": 3830
+    },
+    {
+      "epoch": 2.7745664739884393,
+      "grad_norm": 0.20872294902801514,
+      "learning_rate": 9.004812361440471e-05,
+      "loss": 0.0195,
+      "step": 3840
+    },
+    {
+      "epoch": 2.781791907514451,
+      "grad_norm": 0.29158633947372437,
+      "learning_rate": 8.95725129753272e-05,
+      "loss": 0.018,
+      "step": 3850
+    },
+    {
+      "epoch": 2.7890173410404624,
+      "grad_norm": 0.18671691417694092,
+      "learning_rate": 8.909714068651625e-05,
+      "loss": 0.018,
+      "step": 3860
+    },
+    {
+      "epoch": 2.796242774566474,
+      "grad_norm": 0.31880393624305725,
+      "learning_rate": 8.862201761397539e-05,
+      "loss": 0.0204,
+      "step": 3870
+    },
+    {
+      "epoch": 2.8034682080924855,
+      "grad_norm": 0.4597291648387909,
+      "learning_rate": 8.814715461801173e-05,
+      "loss": 0.0182,
+      "step": 3880
+    },
+    {
+      "epoch": 2.810693641618497,
+      "grad_norm": 0.2754676938056946,
+      "learning_rate": 8.767256255298751e-05,
+      "loss": 0.0181,
+      "step": 3890
+    },
+    {
+      "epoch": 2.8179190751445087,
+      "grad_norm": 0.19947566092014313,
+      "learning_rate": 8.71982522670721e-05,
+      "loss": 0.0195,
+      "step": 3900
+    },
+    {
+      "epoch": 2.8251445086705202,
+      "grad_norm": 0.22111451625823975,
+      "learning_rate": 8.672423460199396e-05,
+      "loss": 0.0155,
+      "step": 3910
+    },
+    {
+      "epoch": 2.832369942196532,
+      "grad_norm": 0.2328019142150879,
+      "learning_rate": 8.625052039279287e-05,
+      "loss": 0.014,
+      "step": 3920
+    },
+    {
+      "epoch": 2.8395953757225434,
+      "grad_norm": 0.3813166618347168,
+      "learning_rate": 8.577712046757233e-05,
+      "loss": 0.0212,
+      "step": 3930
+    },
+    {
+      "epoch": 2.846820809248555,
+      "grad_norm": 0.3128349184989929,
+      "learning_rate": 8.530404564725179e-05,
+      "loss": 0.0235,
+      "step": 3940
+    },
+    {
+      "epoch": 2.8540462427745665,
+      "grad_norm": 0.1525730937719345,
+      "learning_rate": 8.483130674531971e-05,
+      "loss": 0.0167,
+      "step": 3950
+    },
+    {
+      "epoch": 2.861271676300578,
+      "grad_norm": 0.37467721104621887,
+      "learning_rate": 8.4358914567586e-05,
+      "loss": 0.0292,
+      "step": 3960
+    },
+    {
+      "epoch": 2.8684971098265897,
+      "grad_norm": 0.2686941921710968,
+      "learning_rate": 8.388687991193526e-05,
+      "loss": 0.016,
+      "step": 3970
+    },
+    {
+      "epoch": 2.8757225433526012,
+      "grad_norm": 0.42237579822540283,
+      "learning_rate": 8.34152135680799e-05,
+      "loss": 0.0165,
+      "step": 3980
+    },
+    {
+      "epoch": 2.882947976878613,
+      "grad_norm": 0.29524099826812744,
+      "learning_rate": 8.294392631731344e-05,
+      "loss": 0.0165,
+      "step": 3990
+    },
+    {
+      "epoch": 2.8901734104046244,
+      "grad_norm": 0.11127469688653946,
+      "learning_rate": 8.247302893226425e-05,
+      "loss": 0.0132,
+      "step": 4000
+    },
+    {
+      "epoch": 2.897398843930636,
+      "grad_norm": 0.21987956762313843,
+      "learning_rate": 8.20025321766491e-05,
+      "loss": 0.0168,
+      "step": 4010
+    },
+    {
+      "epoch": 2.9046242774566475,
+      "grad_norm": 0.24067598581314087,
+      "learning_rate": 8.153244680502716e-05,
+      "loss": 0.014,
+      "step": 4020
+    },
+    {
+      "epoch": 2.911849710982659,
+      "grad_norm": 0.3013963997364044,
+      "learning_rate": 8.106278356255449e-05,
+      "loss": 0.0173,
+      "step": 4030
+    },
+    {
+      "epoch": 2.9190751445086707,
+      "grad_norm": 0.250764936208725,
+      "learning_rate": 8.059355318473781e-05,
+      "loss": 0.0161,
+      "step": 4040
+    },
+    {
+      "epoch": 2.9263005780346822,
+      "grad_norm": 0.28107234835624695,
+      "learning_rate": 8.012476639718982e-05,
+      "loss": 0.0252,
+      "step": 4050
+    },
+    {
+      "epoch": 2.9335260115606934,
+      "grad_norm": 0.17669588327407837,
+      "learning_rate": 7.965643391538344e-05,
+      "loss": 0.0233,
+      "step": 4060
+    },
+    {
+      "epoch": 2.940751445086705,
+      "grad_norm": 0.21567624807357788,
+      "learning_rate": 7.918856644440724e-05,
+      "loss": 0.0228,
+      "step": 4070
+    },
+    {
+      "epoch": 2.9479768786127165,
+      "grad_norm": 0.3180129826068878,
+      "learning_rate": 7.872117467872059e-05,
+      "loss": 0.0211,
+      "step": 4080
+    },
+    {
+      "epoch": 2.955202312138728,
+      "grad_norm": 0.2351105809211731,
+      "learning_rate": 7.825426930190922e-05,
+      "loss": 0.0159,
+      "step": 4090
+    },
+    {
+      "epoch": 2.9624277456647397,
+      "grad_norm": 0.32194286584854126,
+      "learning_rate": 7.778786098644115e-05,
+      "loss": 0.0176,
+      "step": 4100
+    },
+    {
+      "epoch": 2.9696531791907512,
+      "grad_norm": 0.23887833952903748,
+      "learning_rate": 7.73219603934225e-05,
+      "loss": 0.0141,
+      "step": 4110
+    },
+    {
+      "epoch": 2.976878612716763,
+      "grad_norm": 0.2731233239173889,
+      "learning_rate": 7.685657817235394e-05,
+      "loss": 0.0165,
+      "step": 4120
+    },
+    {
+      "epoch": 2.9841040462427744,
+      "grad_norm": 0.3001903295516968,
+      "learning_rate": 7.63917249608874e-05,
+      "loss": 0.0119,
+      "step": 4130
+    },
+    {
+      "epoch": 2.991329479768786,
+      "grad_norm": 0.23992595076560974,
+      "learning_rate": 7.592741138458262e-05,
+      "loss": 0.0137,
+      "step": 4140
+    },
+    {
+      "epoch": 2.9985549132947975,
+      "grad_norm": 0.2830798923969269,
+      "learning_rate": 7.546364805666446e-05,
+      "loss": 0.0199,
+      "step": 4150
+    },
+    {
+      "epoch": 3.005780346820809,
+      "grad_norm": 0.34715065360069275,
+      "learning_rate": 7.500044557778033e-05,
+      "loss": 0.0142,
+      "step": 4160
+    },
+    {
+      "epoch": 3.0130057803468207,
+      "grad_norm": 0.2551473379135132,
+      "learning_rate": 7.453781453575767e-05,
+      "loss": 0.012,
+      "step": 4170
+    },
+    {
+      "epoch": 3.020231213872832,
+      "grad_norm": 0.2591426372528076,
+      "learning_rate": 7.40757655053623e-05,
+      "loss": 0.0194,
+      "step": 4180
+    },
+    {
+      "epoch": 3.027456647398844,
+      "grad_norm": 0.24271638691425323,
+      "learning_rate": 7.361430904805626e-05,
+      "loss": 0.0144,
+      "step": 4190
+    },
+    {
+      "epoch": 3.0346820809248554,
+      "grad_norm": 0.3379476070404053,
+      "learning_rate": 7.315345571175687e-05,
+      "loss": 0.0134,
+      "step": 4200
+    },
+    {
+      "epoch": 3.041907514450867,
+      "grad_norm": 0.16557613015174866,
+      "learning_rate": 7.269321603059515e-05,
+      "loss": 0.012,
+      "step": 4210
+    },
+    {
+      "epoch": 3.0491329479768785,
+      "grad_norm": 0.2421591579914093,
+      "learning_rate": 7.223360052467531e-05,
+      "loss": 0.015,
+      "step": 4220
+    },
+    {
+      "epoch": 3.05635838150289,
+      "grad_norm": 0.18372207880020142,
+      "learning_rate": 7.177461969983441e-05,
+      "loss": 0.0129,
+      "step": 4230
+    },
+    {
+      "epoch": 3.0635838150289016,
+      "grad_norm": 0.1988380253314972,
+      "learning_rate": 7.131628404740184e-05,
+      "loss": 0.0119,
+      "step": 4240
+    },
+    {
+      "epoch": 3.070809248554913,
+      "grad_norm": 0.1848914474248886,
+      "learning_rate": 7.085860404395985e-05,
+      "loss": 0.0157,
+      "step": 4250
+    },
+    {
+      "epoch": 3.078034682080925,
+      "grad_norm": 0.188431978225708,
+      "learning_rate": 7.040159015110385e-05,
+      "loss": 0.0157,
+      "step": 4260
+    },
+    {
+      "epoch": 3.0852601156069364,
+      "grad_norm": 0.12507693469524384,
+      "learning_rate": 6.994525281520339e-05,
+      "loss": 0.0149,
+      "step": 4270
+    },
+    {
+      "epoch": 3.092485549132948,
+      "grad_norm": 0.192454531788826,
+      "learning_rate": 6.948960246716345e-05,
+      "loss": 0.0152,
+      "step": 4280
+    },
+    {
+      "epoch": 3.0997109826589595,
+      "grad_norm": 0.27219805121421814,
+      "learning_rate": 6.903464952218574e-05,
+      "loss": 0.0251,
+      "step": 4290
+    },
+    {
+      "epoch": 3.106936416184971,
+      "grad_norm": 0.26388370990753174,
+      "learning_rate": 6.858040437953108e-05,
+      "loss": 0.0152,
+      "step": 4300
+    },
+    {
+      "epoch": 3.1141618497109826,
+      "grad_norm": 0.2379855066537857,
+      "learning_rate": 6.812687742228121e-05,
+      "loss": 0.0133,
+      "step": 4310
+    },
+    {
+      "epoch": 3.121387283236994,
+      "grad_norm": 0.14189544320106506,
+      "learning_rate": 6.767407901710165e-05,
+      "loss": 0.0165,
+      "step": 4320
+    },
+    {
+      "epoch": 3.128612716763006,
+      "grad_norm": 0.29239585995674133,
+      "learning_rate": 6.722201951400498e-05,
+      "loss": 0.0148,
+      "step": 4330
+    },
+    {
+      "epoch": 3.1358381502890174,
+      "grad_norm": 0.29002389311790466,
+      "learning_rate": 6.677070924611386e-05,
+      "loss": 0.0131,
+      "step": 4340
+    },
+    {
+      "epoch": 3.143063583815029,
+      "grad_norm": 0.19561998546123505,
+      "learning_rate": 6.632015852942508e-05,
+      "loss": 0.0124,
+      "step": 4350
+    },
+    {
+      "epoch": 3.1502890173410405,
+      "grad_norm": 0.30721139907836914,
+      "learning_rate": 6.587037766257366e-05,
+      "loss": 0.0182,
+      "step": 4360
+    },
+    {
+      "epoch": 3.157514450867052,
+      "grad_norm": 0.24672167003154755,
+      "learning_rate": 6.542137692659764e-05,
+      "loss": 0.0149,
+      "step": 4370
+    },
+    {
+      "epoch": 3.1647398843930636,
+      "grad_norm": 0.26196083426475525,
+      "learning_rate": 6.497316658470267e-05,
+      "loss": 0.0147,
+      "step": 4380
+    },
+    {
+      "epoch": 3.171965317919075,
+      "grad_norm": 0.22746267914772034,
+      "learning_rate": 6.452575688202779e-05,
+      "loss": 0.0144,
+      "step": 4390
+    },
+    {
+      "epoch": 3.179190751445087,
+      "grad_norm": 0.22506026923656464,
+      "learning_rate": 6.407915804541123e-05,
+      "loss": 0.0168,
+      "step": 4400
+    },
+    {
+      "epoch": 3.1864161849710984,
+      "grad_norm": 0.2553040385246277,
+      "learning_rate": 6.363338028315633e-05,
+      "loss": 0.0119,
+      "step": 4410
+    },
+    {
+      "epoch": 3.19364161849711,
+      "grad_norm": 0.20052765309810638,
+      "learning_rate": 6.31884337847986e-05,
+      "loss": 0.0128,
+      "step": 4420
+    },
+    {
+      "epoch": 3.2008670520231215,
+      "grad_norm": 0.1490064561367035,
+      "learning_rate": 6.27443287208725e-05,
+      "loss": 0.0104,
+      "step": 4430
+    },
+    {
+      "epoch": 3.208092485549133,
+      "grad_norm": 0.14192526042461395,
+      "learning_rate": 6.230107524267907e-05,
+      "loss": 0.0135,
+      "step": 4440
+    },
+    {
+      "epoch": 3.2153179190751446,
+      "grad_norm": 0.25964197516441345,
+      "learning_rate": 6.185868348205402e-05,
+      "loss": 0.0145,
+      "step": 4450
+    },
+    {
+      "epoch": 3.222543352601156,
+      "grad_norm": 0.24748851358890533,
+      "learning_rate": 6.141716355113589e-05,
+      "loss": 0.012,
+      "step": 4460
+    },
+    {
+      "epoch": 3.229768786127168,
+      "grad_norm": 0.23189860582351685,
+      "learning_rate": 6.0976525542135154e-05,
+      "loss": 0.0126,
+      "step": 4470
+    },
+    {
+      "epoch": 3.2369942196531793,
+      "grad_norm": 0.1691814512014389,
+      "learning_rate": 6.0536779527103374e-05,
+      "loss": 0.0113,
+      "step": 4480
+    },
+    {
+      "epoch": 3.244219653179191,
+      "grad_norm": 0.17932894825935364,
+      "learning_rate": 6.009793555770291e-05,
+      "loss": 0.011,
+      "step": 4490
+    },
+    {
+      "epoch": 3.2514450867052025,
+      "grad_norm": 0.2402486354112625,
+      "learning_rate": 5.966000366497748e-05,
+      "loss": 0.0178,
+      "step": 4500
+    },
+    {
+      "epoch": 3.258670520231214,
+      "grad_norm": 0.16940508782863617,
+      "learning_rate": 5.922299385912248e-05,
+      "loss": 0.0102,
+      "step": 4510
+    },
+    {
+      "epoch": 3.2658959537572256,
+      "grad_norm": 0.33355531096458435,
+      "learning_rate": 5.8786916129256464e-05,
+      "loss": 0.0173,
+      "step": 4520
+    },
+    {
+      "epoch": 3.273121387283237,
+      "grad_norm": 0.2767535150051117,
+      "learning_rate": 5.835178044319266e-05,
+      "loss": 0.0146,
+      "step": 4530
+    },
+    {
+      "epoch": 3.2803468208092488,
+      "grad_norm": 0.15376369655132294,
+      "learning_rate": 5.791759674721111e-05,
+      "loss": 0.0091,
+      "step": 4540
+    },
+    {
+      "epoch": 3.2875722543352603,
+      "grad_norm": 0.4249739348888397,
+      "learning_rate": 5.7484374965831544e-05,
+      "loss": 0.0181,
+      "step": 4550
+    },
+    {
+      "epoch": 3.294797687861272,
+      "grad_norm": 0.26152127981185913,
+      "learning_rate": 5.705212500158621e-05,
+      "loss": 0.0123,
+      "step": 4560
+    },
+    {
+      "epoch": 3.302023121387283,
+      "grad_norm": 0.11546692997217178,
+      "learning_rate": 5.662085673479373e-05,
+      "loss": 0.0121,
+      "step": 4570
+    },
+    {
+      "epoch": 3.3092485549132946,
+      "grad_norm": 0.21385464072227478,
+      "learning_rate": 5.6190580023333185e-05,
+      "loss": 0.0135,
+      "step": 4580
+    },
+    {
+      "epoch": 3.316473988439306,
+      "grad_norm": 0.13952258229255676,
+      "learning_rate": 5.5761304702418784e-05,
+      "loss": 0.0118,
+      "step": 4590
+    },
+    {
+      "epoch": 3.3236994219653178,
+      "grad_norm": 0.427616149187088,
+      "learning_rate": 5.533304058437516e-05,
+      "loss": 0.0124,
+      "step": 4600
+    },
+    {
+      "epoch": 3.3309248554913293,
+      "grad_norm": 0.23015211522579193,
+      "learning_rate": 5.4905797458412866e-05,
+      "loss": 0.0163,
+      "step": 4610
+    },
+    {
+      "epoch": 3.338150289017341,
+      "grad_norm": 0.2182857245206833,
+      "learning_rate": 5.447958509040488e-05,
+      "loss": 0.0128,
+      "step": 4620
+    },
+    {
+      "epoch": 3.3453757225433525,
+      "grad_norm": 0.16086873412132263,
+      "learning_rate": 5.405441322266311e-05,
+      "loss": 0.0152,
+      "step": 4630
+    },
+    {
+      "epoch": 3.352601156069364,
+      "grad_norm": 0.2987256348133087,
+      "learning_rate": 5.363029157371588e-05,
+      "loss": 0.0098,
+      "step": 4640
+    },
+    {
+      "epoch": 3.3598265895953756,
+      "grad_norm": 0.2915881276130676,
+      "learning_rate": 5.3207229838085745e-05,
+      "loss": 0.0179,
+      "step": 4650
+    },
+    {
+      "epoch": 3.367052023121387,
+      "grad_norm": 0.25969257950782776,
+      "learning_rate": 5.2785237686067826e-05,
+      "loss": 0.0164,
+      "step": 4660
+    },
+    {
+      "epoch": 3.3742774566473988,
+      "grad_norm": 0.3865472376346588,
+      "learning_rate": 5.236432476350897e-05,
+      "loss": 0.0203,
+      "step": 4670
+    },
+    {
+      "epoch": 3.3815028901734103,
+      "grad_norm": 0.22970513999462128,
+      "learning_rate": 5.194450069158696e-05,
+      "loss": 0.012,
+      "step": 4680
+    },
+    {
+      "epoch": 3.388728323699422,
+      "grad_norm": 0.20756426453590393,
+      "learning_rate": 5.1525775066590856e-05,
+      "loss": 0.0174,
+      "step": 4690
+    },
+    {
+      "epoch": 3.3959537572254335,
+      "grad_norm": 0.20011624693870544,
+      "learning_rate": 5.110815745970159e-05,
+      "loss": 0.0132,
+      "step": 4700
+    },
+    {
+      "epoch": 3.403179190751445,
+      "grad_norm": 0.15530593693256378,
+      "learning_rate": 5.069165741677295e-05,
+      "loss": 0.0154,
+      "step": 4710
+    },
+    {
+      "epoch": 3.4104046242774566,
+      "grad_norm": 0.33420002460479736,
+      "learning_rate": 5.0276284458113834e-05,
+      "loss": 0.0141,
+      "step": 4720
+    },
+    {
+      "epoch": 3.417630057803468,
+      "grad_norm": 0.1256830096244812,
+      "learning_rate": 4.9862048078270204e-05,
+      "loss": 0.0106,
+      "step": 4730
+    },
+    {
+      "epoch": 3.4248554913294798,
+      "grad_norm": 0.2905787527561188,
+      "learning_rate": 4.9448957745808264e-05,
+      "loss": 0.0162,
+      "step": 4740
+    },
+    {
+      "epoch": 3.4320809248554913,
+      "grad_norm": 0.17276261746883392,
+      "learning_rate": 4.9037022903098104e-05,
+      "loss": 0.0113,
+      "step": 4750
+    },
+    {
+      "epoch": 3.439306358381503,
+      "grad_norm": 0.21461628377437592,
+      "learning_rate": 4.8626252966097666e-05,
+      "loss": 0.0121,
+      "step": 4760
+    },
+    {
+      "epoch": 3.4465317919075145,
+      "grad_norm": 0.22955593466758728,
+      "learning_rate": 4.821665732413761e-05,
+      "loss": 0.0129,
+      "step": 4770
+    },
+    {
+      "epoch": 3.453757225433526,
+      "grad_norm": 0.2228996753692627,
+      "learning_rate": 4.780824533970677e-05,
+      "loss": 0.0217,
+      "step": 4780
+    },
+    {
+      "epoch": 3.4609826589595376,
+      "grad_norm": 0.18699540197849274,
+      "learning_rate": 4.7401026348237966e-05,
+      "loss": 0.0203,
+      "step": 4790
+    },
+    {
+      "epoch": 3.468208092485549,
+      "grad_norm": 0.2253578007221222,
+      "learning_rate": 4.699500965789489e-05,
+      "loss": 0.0136,
+      "step": 4800
+    },
+    {
+      "epoch": 3.4754335260115607,
+      "grad_norm": 0.2650775909423828,
+      "learning_rate": 4.6590204549359004e-05,
+      "loss": 0.0125,
+      "step": 4810
+    },
+    {
+      "epoch": 3.4826589595375723,
+      "grad_norm": 0.15498478710651398,
+      "learning_rate": 4.618662027561773e-05,
+      "loss": 0.0119,
+      "step": 4820
+    },
+    {
+      "epoch": 3.489884393063584,
+      "grad_norm": 0.21683159470558167,
+      "learning_rate": 4.578426606175268e-05,
+      "loss": 0.0132,
+      "step": 4830
+    },
+    {
+      "epoch": 3.4971098265895955,
+      "grad_norm": 0.14817044138908386,
+      "learning_rate": 4.538315110472894e-05,
+      "loss": 0.0138,
+      "step": 4840
+    },
+    {
+      "epoch": 3.504335260115607,
+      "grad_norm": 0.14653435349464417,
+      "learning_rate": 4.498328457318481e-05,
+      "loss": 0.0101,
+      "step": 4850
+    },
+    {
+      "epoch": 3.5115606936416186,
+      "grad_norm": 0.20104296505451202,
+      "learning_rate": 4.458467560722219e-05,
+      "loss": 0.0129,
+      "step": 4860
+    },
+    {
+      "epoch": 3.51878612716763,
+      "grad_norm": 0.19111691415309906,
+      "learning_rate": 4.418733331819781e-05,
+      "loss": 0.0137,
+      "step": 4870
+    },
+    {
+      "epoch": 3.5260115606936417,
+      "grad_norm": 0.22066420316696167,
+      "learning_rate": 4.3791266788514696e-05,
+      "loss": 0.0124,
+      "step": 4880
+    },
+    {
+      "epoch": 3.5332369942196533,
+      "grad_norm": 0.1991186887025833,
+      "learning_rate": 4.339648507141485e-05,
+      "loss": 0.0135,
+      "step": 4890
+    },
+    {
+      "epoch": 3.540462427745665,
+      "grad_norm": 0.22687891125679016,
+      "learning_rate": 4.300299719077211e-05,
+      "loss": 0.0113,
+      "step": 4900
+    },
+    {
+      "epoch": 3.5476878612716765,
+      "grad_norm": 0.2814226746559143,
+      "learning_rate": 4.261081214088595e-05,
+      "loss": 0.0135,
+      "step": 4910
+    },
+    {
+      "epoch": 3.5549132947976876,
+      "grad_norm": 0.23562021553516388,
+      "learning_rate": 4.221993888627591e-05,
+      "loss": 0.0161,
+      "step": 4920
+    },
+    {
+      "epoch": 3.562138728323699,
+      "grad_norm": 0.16486051678657532,
+      "learning_rate": 4.1830386361476637e-05,
+      "loss": 0.0134,
+      "step": 4930
+    },
+    {
+      "epoch": 3.5693641618497107,
+      "grad_norm": 0.24152740836143494,
+      "learning_rate": 4.144216347083378e-05,
+      "loss": 0.0107,
+      "step": 4940
+    },
+    {
+      "epoch": 3.5765895953757223,
+      "grad_norm": 0.19932973384857178,
+      "learning_rate": 4.1055279088300294e-05,
+      "loss": 0.011,
+      "step": 4950
+    },
+    {
+      "epoch": 3.583815028901734,
+      "grad_norm": 0.2200741022825241,
+      "learning_rate": 4.066974205723362e-05,
+      "loss": 0.015,
+      "step": 4960
+    },
+    {
+      "epoch": 3.5910404624277454,
+      "grad_norm": 0.18681202828884125,
+      "learning_rate": 4.02855611901938e-05,
+      "loss": 0.0136,
+      "step": 4970
+    },
+    {
+      "epoch": 3.598265895953757,
+      "grad_norm": 0.14601317048072815,
+      "learning_rate": 3.990274526874156e-05,
+      "loss": 0.0091,
+      "step": 4980
+    },
+    {
+      "epoch": 3.6054913294797686,
+      "grad_norm": 0.22250248491764069,
+      "learning_rate": 3.95213030432381e-05,
+      "loss": 0.0153,
+      "step": 4990
+    },
+    {
+      "epoch": 3.61271676300578,
+      "grad_norm": 0.17519167065620422,
+      "learning_rate": 3.914124323264473e-05,
+      "loss": 0.0114,
+      "step": 5000
+    },
+    {
+      "epoch": 3.6199421965317917,
+      "grad_norm": 0.17757725715637207,
+      "learning_rate": 3.876257452432365e-05,
+      "loss": 0.0144,
+      "step": 5010
+    },
+    {
+      "epoch": 3.6271676300578033,
+      "grad_norm": 0.4040830731391907,
+      "learning_rate": 3.838530557383953e-05,
+      "loss": 0.0119,
+      "step": 5020
+    },
+    {
+      "epoch": 3.634393063583815,
+      "grad_norm": 0.16794376075267792,
+      "learning_rate": 3.800944500476145e-05,
+      "loss": 0.0118,
+      "step": 5030
+    },
+    {
+      "epoch": 3.6416184971098264,
+      "grad_norm": 0.23436979949474335,
+      "learning_rate": 3.7635001408465875e-05,
+      "loss": 0.0096,
+      "step": 5040
+    },
+    {
+      "epoch": 3.648843930635838,
+      "grad_norm": 0.3052511215209961,
+      "learning_rate": 3.72619833439403e-05,
+      "loss": 0.0131,
+      "step": 5050
+    },
+    {
+      "epoch": 3.6560693641618496,
+      "grad_norm": 0.19484452903270721,
+      "learning_rate": 3.689039933758754e-05,
+      "loss": 0.0107,
+      "step": 5060
+    },
+    {
+      "epoch": 3.663294797687861,
+      "grad_norm": 0.11929387599229813,
+      "learning_rate": 3.652025788303099e-05,
+      "loss": 0.0142,
+      "step": 5070
+    },
+    {
+      "epoch": 3.6705202312138727,
+      "grad_norm": 0.13904815912246704,
+      "learning_rate": 3.6151567440920206e-05,
+      "loss": 0.011,
+      "step": 5080
+    },
+    {
+      "epoch": 3.6777456647398843,
+      "grad_norm": 0.20277534425258636,
+      "learning_rate": 3.578433643873784e-05,
+      "loss": 0.0108,
+      "step": 5090
+    },
+    {
+      "epoch": 3.684971098265896,
+      "grad_norm": 0.14958781003952026,
+      "learning_rate": 3.541857327060671e-05,
+      "loss": 0.0174,
+      "step": 5100
+    },
+    {
+      "epoch": 3.6921965317919074,
+      "grad_norm": 0.18210665881633759,
+      "learning_rate": 3.5054286297098117e-05,
+      "loss": 0.0143,
+      "step": 5110
+    },
+    {
+      "epoch": 3.699421965317919,
+      "grad_norm": 0.17495621740818024,
+      "learning_rate": 3.469148384504065e-05,
+      "loss": 0.0105,
+      "step": 5120
+    },
+    {
+      "epoch": 3.7066473988439306,
+      "grad_norm": 0.15843825042247772,
+      "learning_rate": 3.4330174207329855e-05,
+      "loss": 0.0089,
+      "step": 5130
+    },
+    {
+      "epoch": 3.713872832369942,
+      "grad_norm": 0.14327116310596466,
+      "learning_rate": 3.3970365642738846e-05,
+      "loss": 0.0105,
+      "step": 5140
+    },
+    {
+      "epoch": 3.7210982658959537,
+      "grad_norm": 0.14071951806545258,
+      "learning_rate": 3.361206637572921e-05,
+      "loss": 0.0145,
+      "step": 5150
+    },
+    {
+      "epoch": 3.7283236994219653,
+      "grad_norm": 0.17072631418704987,
+      "learning_rate": 3.3255284596263256e-05,
+      "loss": 0.0121,
+      "step": 5160
+    },
+    {
+      "epoch": 3.735549132947977,
+      "grad_norm": 0.14465495944023132,
+      "learning_rate": 3.290002845961682e-05,
+      "loss": 0.0109,
+      "step": 5170
+    },
+    {
+      "epoch": 3.7427745664739884,
+      "grad_norm": 0.11986351013183594,
+      "learning_rate": 3.2546306086192636e-05,
+      "loss": 0.0113,
+      "step": 5180
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.19396544992923737,
+      "learning_rate": 3.219412556133497e-05,
+      "loss": 0.0089,
+      "step": 5190
+    },
+    {
+      "epoch": 3.7572254335260116,
+      "grad_norm": 0.15157504379749298,
+      "learning_rate": 3.1843494935144626e-05,
+      "loss": 0.013,
+      "step": 5200
+    },
+    {
+      "epoch": 3.764450867052023,
+      "grad_norm": 0.26728197932243347,
+      "learning_rate": 3.1494422222295e-05,
+      "loss": 0.0113,
+      "step": 5210
+    },
+    {
+      "epoch": 3.7716763005780347,
+      "grad_norm": 0.3146849274635315,
+      "learning_rate": 3.114691540184901e-05,
+      "loss": 0.0078,
+      "step": 5220
+    },
+    {
+      "epoch": 3.7789017341040463,
+      "grad_norm": 0.19057181477546692,
+      "learning_rate": 3.080098241707641e-05,
+      "loss": 0.0112,
+      "step": 5230
+    },
+    {
+      "epoch": 3.786127167630058,
+      "grad_norm": 0.10792943090200424,
+      "learning_rate": 3.0456631175272608e-05,
+      "loss": 0.0098,
+      "step": 5240
+    },
+    {
+      "epoch": 3.7933526011560694,
+      "grad_norm": 0.15339750051498413,
+      "learning_rate": 3.0113869547577522e-05,
+      "loss": 0.0089,
+      "step": 5250
+    },
+    {
+      "epoch": 3.800578034682081,
+      "grad_norm": 0.14971426129341125,
+      "learning_rate": 2.9772705368795927e-05,
+      "loss": 0.014,
+      "step": 5260
+    },
+    {
+      "epoch": 3.8078034682080926,
+      "grad_norm": 0.12296771258115768,
+      "learning_rate": 2.9433146437218394e-05,
+      "loss": 0.0093,
+      "step": 5270
+    },
+    {
+      "epoch": 3.815028901734104,
+      "grad_norm": 0.21407096087932587,
+      "learning_rate": 2.909520051444279e-05,
+      "loss": 0.0084,
+      "step": 5280
+    },
+    {
+      "epoch": 3.8222543352601157,
+      "grad_norm": 0.1819303184747696,
+      "learning_rate": 2.8758875325197167e-05,
+      "loss": 0.0096,
+      "step": 5290
+    },
+    {
+      "epoch": 3.8294797687861273,
+      "grad_norm": 0.30501365661621094,
+      "learning_rate": 2.8424178557162915e-05,
+      "loss": 0.0138,
+      "step": 5300
+    },
+    {
+      "epoch": 3.836705202312139,
+      "grad_norm": 0.13662755489349365,
+      "learning_rate": 2.8091117860799244e-05,
+      "loss": 0.0214,
+      "step": 5310
+    },
+    {
+      "epoch": 3.8439306358381504,
+      "grad_norm": 0.17908525466918945,
+      "learning_rate": 2.775970084916818e-05,
+      "loss": 0.0123,
+      "step": 5320
+    },
+    {
+      "epoch": 3.851156069364162,
+      "grad_norm": 0.22172127664089203,
+      "learning_rate": 2.7429935097760605e-05,
+      "loss": 0.0136,
+      "step": 5330
+    },
+    {
+      "epoch": 3.8583815028901736,
+      "grad_norm": 0.23410490155220032,
+      "learning_rate": 2.710182814432317e-05,
+      "loss": 0.009,
+      "step": 5340
+    },
+    {
+      "epoch": 3.865606936416185,
+      "grad_norm": 0.11713831126689911,
+      "learning_rate": 2.6775387488685778e-05,
+      "loss": 0.0068,
+      "step": 5350
+    },
+    {
+      "epoch": 3.8728323699421967,
+      "grad_norm": 0.20926479995250702,
+      "learning_rate": 2.645062059259045e-05,
+      "loss": 0.021,
+      "step": 5360
+    },
+    {
+      "epoch": 3.8800578034682083,
+      "grad_norm": 0.2228723168373108,
+      "learning_rate": 2.6127534879520476e-05,
+      "loss": 0.0082,
+      "step": 5370
+    },
+    {
+      "epoch": 3.88728323699422,
+      "grad_norm": 0.1596720665693283,
+      "learning_rate": 2.5806137734530934e-05,
+      "loss": 0.0127,
+      "step": 5380
+    },
+    {
+      "epoch": 3.8945086705202314,
+      "grad_norm": 0.2996460199356079,
+      "learning_rate": 2.5486436504079792e-05,
+      "loss": 0.0107,
+      "step": 5390
+    },
+    {
+      "epoch": 3.901734104046243,
+      "grad_norm": 0.215359628200531,
+      "learning_rate": 2.516843849585997e-05,
+      "loss": 0.0142,
+      "step": 5400
+    },
+    {
+      "epoch": 3.9089595375722546,
+      "grad_norm": 0.17319220304489136,
+      "learning_rate": 2.4852150978632427e-05,
+      "loss": 0.0107,
+      "step": 5410
+    },
+    {
+      "epoch": 3.916184971098266,
+      "grad_norm": 0.14736515283584595,
+      "learning_rate": 2.4537581182059844e-05,
+      "loss": 0.0146,
+      "step": 5420
+    },
+    {
+      "epoch": 3.9234104046242777,
+      "grad_norm": 0.08320306241512299,
+      "learning_rate": 2.422473629654145e-05,
+      "loss": 0.0073,
+      "step": 5430
+    },
+    {
+      "epoch": 3.9306358381502893,
+      "grad_norm": 0.15274234116077423,
+      "learning_rate": 2.3913623473048708e-05,
+      "loss": 0.0124,
+      "step": 5440
+    },
+    {
+      "epoch": 3.937861271676301,
+      "grad_norm": 0.2870109975337982,
+      "learning_rate": 2.3604249822961776e-05,
+      "loss": 0.0128,
+      "step": 5450
+    },
+    {
+      "epoch": 3.9450867052023124,
+      "grad_norm": 0.20727121829986572,
+      "learning_rate": 2.3296622417906988e-05,
+      "loss": 0.0066,
+      "step": 5460
+    },
+    {
+      "epoch": 3.9523121387283235,
+      "grad_norm": 0.19767476618289948,
+      "learning_rate": 2.2990748289595243e-05,
+      "loss": 0.0112,
+      "step": 5470
+    },
+    {
+      "epoch": 3.959537572254335,
+      "grad_norm": 0.20713435113430023,
+      "learning_rate": 2.268663442966117e-05,
+      "loss": 0.0092,
+      "step": 5480
+    },
+    {
+      "epoch": 3.9667630057803467,
+      "grad_norm": 0.1702476590871811,
+      "learning_rate": 2.2384287789503533e-05,
+      "loss": 0.0084,
+      "step": 5490
+    },
+    {
+      "epoch": 3.9739884393063583,
+      "grad_norm": 0.14406177401542664,
+      "learning_rate": 2.2083715280126062e-05,
+      "loss": 0.0077,
+      "step": 5500
+    },
+    {
+      "epoch": 3.98121387283237,
+      "grad_norm": 0.14436282217502594,
+      "learning_rate": 2.1784923771979694e-05,
+      "loss": 0.0104,
+      "step": 5510
+    },
+    {
+      "epoch": 3.9884393063583814,
+      "grad_norm": 0.17764638364315033,
+      "learning_rate": 2.1487920094805446e-05,
+      "loss": 0.0085,
+      "step": 5520
+    },
+    {
+      "epoch": 3.995664739884393,
+      "grad_norm": 0.1937018483877182,
+      "learning_rate": 2.1192711037478187e-05,
+      "loss": 0.0083,
+      "step": 5530
+    },
+    {
+      "epoch": 4.002890173410405,
+      "grad_norm": 0.1566731184720993,
+      "learning_rate": 2.0899303347851772e-05,
+      "loss": 0.0092,
+      "step": 5540
+    },
+    {
+      "epoch": 4.0101156069364166,
+      "grad_norm": 0.21031144261360168,
+      "learning_rate": 2.0607703732604435e-05,
+      "loss": 0.0089,
+      "step": 5550
+    },
+    {
+      "epoch": 4.017341040462428,
+      "grad_norm": 0.24341128766536713,
+      "learning_rate": 2.0317918857085805e-05,
+      "loss": 0.0108,
+      "step": 5560
+    },
+    {
+      "epoch": 4.02456647398844,
+      "grad_norm": 0.18322938680648804,
+      "learning_rate": 2.0029955345164286e-05,
+      "loss": 0.0107,
+      "step": 5570
+    },
+    {
+      "epoch": 4.031791907514451,
+      "grad_norm": 0.1266375333070755,
+      "learning_rate": 1.9743819779075823e-05,
+      "loss": 0.0049,
+      "step": 5580
+    },
+    {
+      "epoch": 4.039017341040463,
+      "grad_norm": 0.2991048991680145,
+      "learning_rate": 1.945951869927337e-05,
+      "loss": 0.0082,
+      "step": 5590
+    },
+    {
+      "epoch": 4.046242774566474,
+      "grad_norm": 0.11648675799369812,
+      "learning_rate": 1.917705860427741e-05,
+      "loss": 0.0101,
+      "step": 5600
+    },
+    {
+      "epoch": 4.053468208092486,
+      "grad_norm": 0.15041537582874298,
+      "learning_rate": 1.8896445950527475e-05,
+      "loss": 0.0083,
+      "step": 5610
+    },
+    {
+      "epoch": 4.0606936416184976,
+      "grad_norm": 0.28212547302246094,
+      "learning_rate": 1.861768715223443e-05,
+      "loss": 0.0117,
+      "step": 5620
+    },
+    {
+      "epoch": 4.067919075144508,
+      "grad_norm": 0.10794773697853088,
+      "learning_rate": 1.8340788581233913e-05,
+      "loss": 0.0049,
+      "step": 5630
+    },
+    {
+      "epoch": 4.07514450867052,
+      "grad_norm": 0.1072087287902832,
+      "learning_rate": 1.8065756566840807e-05,
+      "loss": 0.0073,
+      "step": 5640
+    },
+    {
+      "epoch": 4.082369942196531,
+      "grad_norm": 0.21496444940567017,
+      "learning_rate": 1.779259739570438e-05,
+      "loss": 0.0121,
+      "step": 5650
+    },
+    {
+      "epoch": 4.089595375722543,
+      "grad_norm": 0.1343175172805786,
+      "learning_rate": 1.7521317311664696e-05,
+      "loss": 0.0077,
+      "step": 5660
+    },
+    {
+      "epoch": 4.0968208092485545,
+      "grad_norm": 0.13832974433898926,
+      "learning_rate": 1.7251922515609863e-05,
+      "loss": 0.0096,
+      "step": 5670
+    },
+    {
+      "epoch": 4.104046242774566,
+      "grad_norm": 0.18991577625274658,
+      "learning_rate": 1.6984419165334277e-05,
+      "loss": 0.0161,
+      "step": 5680
+    },
+    {
+      "epoch": 4.111271676300578,
+      "grad_norm": 0.2303207516670227,
+      "learning_rate": 1.671881337539798e-05,
+      "loss": 0.0082,
+      "step": 5690
+    },
+    {
+      "epoch": 4.118497109826589,
+      "grad_norm": 0.22146502137184143,
+      "learning_rate": 1.6455111216986664e-05,
+      "loss": 0.01,
+      "step": 5700
+    },
+    {
+      "epoch": 4.125722543352601,
+      "grad_norm": 0.38750120997428894,
+      "learning_rate": 1.6193318717773187e-05,
+      "loss": 0.0117,
+      "step": 5710
+    },
+    {
+      "epoch": 4.132947976878612,
+      "grad_norm": 0.09928886592388153,
+      "learning_rate": 1.5933441861779518e-05,
+      "loss": 0.0067,
+      "step": 5720
+    },
+    {
+      "epoch": 4.140173410404624,
+      "grad_norm": 0.14484621584415436,
+      "learning_rate": 1.567548658924016e-05,
+      "loss": 0.0096,
+      "step": 5730
+    },
+    {
+      "epoch": 4.1473988439306355,
+      "grad_norm": 0.17204761505126953,
+      "learning_rate": 1.5419458796466234e-05,
+      "loss": 0.0065,
+      "step": 5740
+    },
+    {
+      "epoch": 4.154624277456647,
+      "grad_norm": 0.06153562292456627,
+      "learning_rate": 1.5165364335710764e-05,
+      "loss": 0.0088,
+      "step": 5750
+    },
+    {
+      "epoch": 4.161849710982659,
+      "grad_norm": 0.15422751009464264,
+      "learning_rate": 1.4913209015034956e-05,
+      "loss": 0.0098,
+      "step": 5760
+    },
+    {
+      "epoch": 4.16907514450867,
+      "grad_norm": 0.2040354162454605,
+      "learning_rate": 1.4662998598175314e-05,
+      "loss": 0.0072,
+      "step": 5770
+    },
+    {
+      "epoch": 4.176300578034682,
+      "grad_norm": 0.22488704323768616,
+      "learning_rate": 1.441473880441201e-05,
+      "loss": 0.0104,
+      "step": 5780
+    },
+    {
+      "epoch": 4.183526011560693,
+      "grad_norm": 0.17628708481788635,
+      "learning_rate": 1.4168435308438088e-05,
+      "loss": 0.0074,
+      "step": 5790
+    },
+    {
+      "epoch": 4.190751445086705,
+      "grad_norm": 0.12489684671163559,
+      "learning_rate": 1.3924093740229704e-05,
+      "loss": 0.0092,
+      "step": 5800
+    },
+    {
+      "epoch": 4.1979768786127165,
+      "grad_norm": 0.13723787665367126,
+      "learning_rate": 1.3681719684917604e-05,
+      "loss": 0.0132,
+      "step": 5810
+    },
+    {
+      "epoch": 4.205202312138728,
+      "grad_norm": 0.09958511590957642,
+      "learning_rate": 1.3441318682659276e-05,
+      "loss": 0.0145,
+      "step": 5820
+    },
+    {
+      "epoch": 4.21242774566474,
+      "grad_norm": 0.2590515613555908,
+      "learning_rate": 1.3202896228512485e-05,
+      "loss": 0.0095,
+      "step": 5830
+    },
+    {
+      "epoch": 4.219653179190751,
+      "grad_norm": 0.0941896140575409,
+      "learning_rate": 1.296645777230947e-05,
+      "loss": 0.0065,
+      "step": 5840
+    },
+    {
+      "epoch": 4.226878612716763,
+      "grad_norm": 0.23163746297359467,
+      "learning_rate": 1.2732008718532562e-05,
+      "loss": 0.008,
+      "step": 5850
+    },
+    {
+      "epoch": 4.234104046242774,
+      "grad_norm": 0.08037468045949936,
+      "learning_rate": 1.2499554426190518e-05,
+      "loss": 0.0077,
+      "step": 5860
+    },
+    {
+      "epoch": 4.241329479768786,
+      "grad_norm": 0.14902599155902863,
+      "learning_rate": 1.2269100208696061e-05,
+      "loss": 0.0059,
+      "step": 5870
+    },
+    {
+      "epoch": 4.2485549132947975,
+      "grad_norm": 0.2328324019908905,
+      "learning_rate": 1.2040651333744524e-05,
+      "loss": 0.012,
+      "step": 5880
+    },
+    {
+      "epoch": 4.255780346820809,
+      "grad_norm": 0.19932933151721954,
+      "learning_rate": 1.1814213023193277e-05,
+      "loss": 0.0074,
+      "step": 5890
+    },
+    {
+      "epoch": 4.263005780346821,
+      "grad_norm": 0.21653766930103302,
+      "learning_rate": 1.158979045294245e-05,
+      "loss": 0.0083,
+      "step": 5900
+    },
+    {
+      "epoch": 4.270231213872832,
+      "grad_norm": 0.23117244243621826,
+      "learning_rate": 1.1367388752816688e-05,
+      "loss": 0.0108,
+      "step": 5910
+    },
+    {
+      "epoch": 4.277456647398844,
+      "grad_norm": 0.09811783581972122,
+      "learning_rate": 1.1147013006447748e-05,
+      "loss": 0.0095,
+      "step": 5920
+    },
+    {
+      "epoch": 4.284682080924855,
+      "grad_norm": 0.39156588912010193,
+      "learning_rate": 1.0928668251158413e-05,
+      "loss": 0.0071,
+      "step": 5930
+    },
+    {
+      "epoch": 4.291907514450867,
+      "grad_norm": 0.1781422346830368,
+      "learning_rate": 1.0712359477847288e-05,
+      "loss": 0.0159,
+      "step": 5940
+    },
+    {
+      "epoch": 4.2991329479768785,
+      "grad_norm": 0.2437022626399994,
+      "learning_rate": 1.0498091630874763e-05,
+      "loss": 0.0065,
+      "step": 5950
+    },
+    {
+      "epoch": 4.30635838150289,
+      "grad_norm": 0.15582063794136047,
+      "learning_rate": 1.0285869607949994e-05,
+      "loss": 0.008,
+      "step": 5960
+    },
+    {
+      "epoch": 4.313583815028902,
+      "grad_norm": 0.1349640190601349,
+      "learning_rate": 1.0075698260018873e-05,
+      "loss": 0.0069,
+      "step": 5970
+    },
+    {
+      "epoch": 4.320809248554913,
+      "grad_norm": 0.12121254205703735,
+      "learning_rate": 9.867582391153274e-06,
+      "loss": 0.0049,
+      "step": 5980
+    },
+    {
+      "epoch": 4.328034682080925,
+      "grad_norm": 0.1582227647304535,
+      "learning_rate": 9.661526758441141e-06,
+      "loss": 0.008,
+      "step": 5990
+    },
+    {
+      "epoch": 4.335260115606936,
+      "grad_norm": 0.10006178170442581,
+      "learning_rate": 9.45753607187776e-06,
+      "loss": 0.0124,
+      "step": 6000
+    },
+    {
+      "epoch": 4.342485549132948,
+      "grad_norm": 0.21611645817756653,
+      "learning_rate": 9.255614994258154e-06,
+      "loss": 0.0085,
+      "step": 6010
+    },
+    {
+      "epoch": 4.3497109826589595,
+      "grad_norm": 0.1608915776014328,
+      "learning_rate": 9.055768141070454e-06,
+      "loss": 0.0099,
+      "step": 6020
+    },
+    {
+      "epoch": 4.356936416184971,
+      "grad_norm": 0.09145373106002808,
+      "learning_rate": 8.858000080390439e-06,
+      "loss": 0.0111,
+      "step": 6030
+    },
+    {
+      "epoch": 4.364161849710983,
+      "grad_norm": 0.07861532270908356,
+      "learning_rate": 8.66231533277707e-06,
+      "loss": 0.0112,
+      "step": 6040
+    },
+    {
+      "epoch": 4.371387283236994,
+      "grad_norm": 0.23200583457946777,
+      "learning_rate": 8.46871837116916e-06,
+      "loss": 0.0088,
+      "step": 6050
+    },
+    {
+      "epoch": 4.378612716763006,
+      "grad_norm": 0.21334034204483032,
+      "learning_rate": 8.27721362078323e-06,
+      "loss": 0.0094,
+      "step": 6060
+    },
+    {
+      "epoch": 4.385838150289017,
+      "grad_norm": 0.08083529025316238,
+      "learning_rate": 8.08780545901222e-06,
+      "loss": 0.008,
+      "step": 6070
+    },
+    {
+      "epoch": 4.393063583815029,
+      "grad_norm": 0.19791102409362793,
+      "learning_rate": 7.900498215325514e-06,
+      "loss": 0.0072,
+      "step": 6080
+    },
+    {
+      "epoch": 4.4002890173410405,
+      "grad_norm": 0.1434745341539383,
+      "learning_rate": 7.71529617116996e-06,
+      "loss": 0.0069,
+      "step": 6090
+    },
+    {
+      "epoch": 4.407514450867052,
+      "grad_norm": 0.2680268883705139,
+      "learning_rate": 7.532203559871986e-06,
+      "loss": 0.0129,
+      "step": 6100
+    },
+    {
+      "epoch": 4.414739884393064,
+      "grad_norm": 0.18372227251529694,
+      "learning_rate": 7.351224566540915e-06,
+      "loss": 0.0125,
+      "step": 6110
+    },
+    {
+      "epoch": 4.421965317919075,
+      "grad_norm": 0.11005425453186035,
+      "learning_rate": 7.1723633279731415e-06,
+      "loss": 0.0061,
+      "step": 6120
+    },
+    {
+      "epoch": 4.429190751445087,
+      "grad_norm": 0.18487311899662018,
+      "learning_rate": 6.995623932557794e-06,
+      "loss": 0.0104,
+      "step": 6130
+    },
+    {
+      "epoch": 4.436416184971098,
+      "grad_norm": 0.23542456328868866,
+      "learning_rate": 6.821010420183038e-06,
+      "loss": 0.0083,
+      "step": 6140
+    },
+    {
+      "epoch": 4.44364161849711,
+      "grad_norm": 0.10208717733621597,
+      "learning_rate": 6.648526782143905e-06,
+      "loss": 0.0079,
+      "step": 6150
+    },
+    {
+      "epoch": 4.4508670520231215,
+      "grad_norm": 0.17814548313617706,
+      "learning_rate": 6.478176961051052e-06,
+      "loss": 0.0087,
+      "step": 6160
+    },
+    {
+      "epoch": 4.458092485549133,
+      "grad_norm": 0.12302400916814804,
+      "learning_rate": 6.309964850740535e-06,
+      "loss": 0.0093,
+      "step": 6170
+    },
+    {
+      "epoch": 4.465317919075145,
+      "grad_norm": 0.2888944745063782,
+      "learning_rate": 6.143894296184927e-06,
+      "loss": 0.0116,
+      "step": 6180
+    },
+    {
+      "epoch": 4.472543352601156,
+      "grad_norm": 0.2792350947856903,
+      "learning_rate": 5.979969093405336e-06,
+      "loss": 0.0059,
+      "step": 6190
+    },
+    {
+      "epoch": 4.479768786127168,
+      "grad_norm": 0.18412691354751587,
+      "learning_rate": 5.818192989384663e-06,
+      "loss": 0.0094,
+      "step": 6200
+    },
+    {
+      "epoch": 4.486994219653179,
+      "grad_norm": 0.060731563717126846,
+      "learning_rate": 5.658569681982007e-06,
+      "loss": 0.009,
+      "step": 6210
+    },
+    {
+      "epoch": 4.494219653179191,
+      "grad_norm": 0.23368248343467712,
+      "learning_rate": 5.501102819848037e-06,
+      "loss": 0.0136,
+      "step": 6220
+    },
+    {
+      "epoch": 4.5014450867052025,
+      "grad_norm": 0.16230285167694092,
+      "learning_rate": 5.345796002341719e-06,
+      "loss": 0.0073,
+      "step": 6230
+    },
+    {
+      "epoch": 4.508670520231214,
+      "grad_norm": 0.13031484186649323,
+      "learning_rate": 5.1926527794479154e-06,
+      "loss": 0.0102,
+      "step": 6240
+    },
+    {
+      "epoch": 4.515895953757226,
+      "grad_norm": 0.07982811331748962,
+      "learning_rate": 5.041676651696336e-06,
+      "loss": 0.007,
+      "step": 6250
+    },
+    {
+      "epoch": 4.523121387283237,
+      "grad_norm": 0.2615806460380554,
+      "learning_rate": 4.892871070081473e-06,
+      "loss": 0.0105,
+      "step": 6260
+    },
+    {
+      "epoch": 4.530346820809249,
+      "grad_norm": 0.14185559749603271,
+      "learning_rate": 4.746239435983735e-06,
+      "loss": 0.0084,
+      "step": 6270
+    },
+    {
+      "epoch": 4.53757225433526,
+      "grad_norm": 0.1292169988155365,
+      "learning_rate": 4.601785101091683e-06,
+      "loss": 0.0107,
+      "step": 6280
+    },
+    {
+      "epoch": 4.544797687861272,
+      "grad_norm": 0.20721952617168427,
+      "learning_rate": 4.4595113673254196e-06,
+      "loss": 0.0085,
+      "step": 6290
+    },
+    {
+      "epoch": 4.5520231213872835,
+      "grad_norm": 0.19826039671897888,
+      "learning_rate": 4.319421486761177e-06,
+      "loss": 0.0105,
+      "step": 6300
+    },
+    {
+      "epoch": 4.559248554913295,
+      "grad_norm": 0.1031193882226944,
+      "learning_rate": 4.181518661556882e-06,
+      "loss": 0.0054,
+      "step": 6310
+    },
+    {
+      "epoch": 4.566473988439307,
+      "grad_norm": 0.2350577861070633,
+      "learning_rate": 4.045806043879008e-06,
+      "loss": 0.0074,
+      "step": 6320
+    },
+    {
+      "epoch": 4.573699421965318,
+      "grad_norm": 0.10528607666492462,
+      "learning_rate": 3.91228673583055e-06,
+      "loss": 0.0079,
+      "step": 6330
+    },
+    {
+      "epoch": 4.58092485549133,
+      "grad_norm": 0.15961109101772308,
+      "learning_rate": 3.7809637893800875e-06,
+      "loss": 0.0131,
+      "step": 6340
+    },
+    {
+      "epoch": 4.588150289017341,
+      "grad_norm": 0.211639404296875,
+      "learning_rate": 3.651840206292001e-06,
+      "loss": 0.0116,
+      "step": 6350
+    },
+    {
+      "epoch": 4.595375722543353,
+      "grad_norm": 0.0725213959813118,
+      "learning_rate": 3.5249189380578975e-06,
+      "loss": 0.0067,
+      "step": 6360
+    },
+    {
+      "epoch": 4.6026011560693645,
+      "grad_norm": 0.17431780695915222,
+      "learning_rate": 3.400202885829129e-06,
+      "loss": 0.0082,
+      "step": 6370
+    },
+    {
+      "epoch": 4.609826589595376,
+      "grad_norm": 0.0443602092564106,
+      "learning_rate": 3.277694900350503e-06,
+      "loss": 0.0077,
+      "step": 6380
+    },
+    {
+      "epoch": 4.617052023121388,
+      "grad_norm": 0.1488499492406845,
+      "learning_rate": 3.157397781895044e-06,
+      "loss": 0.0104,
+      "step": 6390
+    },
+    {
+      "epoch": 4.624277456647399,
+      "grad_norm": 0.08466222137212753,
+      "learning_rate": 3.039314280200123e-06,
+      "loss": 0.0085,
+      "step": 6400
+    },
+    {
+      "epoch": 4.631502890173411,
+      "grad_norm": 0.19930359721183777,
+      "learning_rate": 2.92344709440443e-06,
+      "loss": 0.0092,
+      "step": 6410
+    },
+    {
+      "epoch": 4.638728323699422,
+      "grad_norm": 0.15144643187522888,
+      "learning_rate": 2.8097988729864134e-06,
+      "loss": 0.0083,
+      "step": 6420
+    },
+    {
+      "epoch": 4.645953757225434,
+      "grad_norm": 0.24370171129703522,
+      "learning_rate": 2.6983722137037147e-06,
+      "loss": 0.0077,
+      "step": 6430
+    },
+    {
+      "epoch": 4.653179190751445,
+      "grad_norm": 0.15722382068634033,
+      "learning_rate": 2.589169663533719e-06,
+      "loss": 0.0062,
+      "step": 6440
+    },
+    {
+      "epoch": 4.660404624277456,
+      "grad_norm": 0.23215506970882416,
+      "learning_rate": 2.4821937186154554e-06,
+      "loss": 0.0093,
+      "step": 6450
+    },
+    {
+      "epoch": 4.667630057803468,
+      "grad_norm": 0.08635558933019638,
+      "learning_rate": 2.3774468241923975e-06,
+      "loss": 0.0071,
+      "step": 6460
+    },
+    {
+      "epoch": 4.674855491329479,
+      "grad_norm": 0.09789091348648071,
+      "learning_rate": 2.274931374556699e-06,
+      "loss": 0.0065,
+      "step": 6470
+    },
+    {
+      "epoch": 4.682080924855491,
+      "grad_norm": 0.13124142587184906,
+      "learning_rate": 2.1746497129944034e-06,
+      "loss": 0.0054,
+      "step": 6480
+    },
+    {
+      "epoch": 4.6893063583815024,
+      "grad_norm": 0.15098749101161957,
+      "learning_rate": 2.0766041317318517e-06,
+      "loss": 0.0055,
+      "step": 6490
+    },
+    {
+      "epoch": 4.696531791907514,
+      "grad_norm": 0.25133273005485535,
+      "learning_rate": 1.9807968718833947e-06,
+      "loss": 0.0072,
+      "step": 6500
+    },
+    {
+      "epoch": 4.703757225433526,
+      "grad_norm": 0.09292857348918915,
+      "learning_rate": 1.8872301234000522e-06,
+      "loss": 0.009,
+      "step": 6510
+    },
+    {
+      "epoch": 4.710982658959537,
+      "grad_norm": 0.08246719837188721,
+      "learning_rate": 1.7959060250194891e-06,
+      "loss": 0.0114,
+      "step": 6520
+    },
+    {
+      "epoch": 4.718208092485549,
+      "grad_norm": 0.16808311641216278,
+      "learning_rate": 1.706826664217165e-06,
+      "loss": 0.0051,
+      "step": 6530
+    },
+    {
+      "epoch": 4.72543352601156,
+      "grad_norm": 0.21554207801818848,
+      "learning_rate": 1.6199940771586041e-06,
+      "loss": 0.0104,
+      "step": 6540
+    },
+    {
+      "epoch": 4.732658959537572,
+      "grad_norm": 0.2452739179134369,
+      "learning_rate": 1.535410248652791e-06,
+      "loss": 0.0081,
+      "step": 6550
+    },
+    {
+      "epoch": 4.7398843930635834,
+      "grad_norm": 0.13148806989192963,
+      "learning_rate": 1.4530771121069153e-06,
+      "loss": 0.0057,
+      "step": 6560
+    },
+    {
+      "epoch": 4.747109826589595,
+      "grad_norm": 0.1118473932147026,
+      "learning_rate": 1.372996549482053e-06,
+      "loss": 0.0091,
+      "step": 6570
+    },
+    {
+      "epoch": 4.754335260115607,
+      "grad_norm": 0.0854889303445816,
+      "learning_rate": 1.2951703912502777e-06,
+      "loss": 0.0067,
+      "step": 6580
+    },
+    {
+      "epoch": 4.761560693641618,
+      "grad_norm": 0.10423944145441055,
+      "learning_rate": 1.2196004163526954e-06,
+      "loss": 0.0074,
+      "step": 6590
+    },
+    {
+      "epoch": 4.76878612716763,
+      "grad_norm": 0.08598557114601135,
+      "learning_rate": 1.1462883521588753e-06,
+      "loss": 0.0053,
+      "step": 6600
+    },
+    {
+      "epoch": 4.776011560693641,
+      "grad_norm": 0.2087874561548233,
+      "learning_rate": 1.075235874427316e-06,
+      "loss": 0.0056,
+      "step": 6610
+    },
+    {
+      "epoch": 4.783236994219653,
+      "grad_norm": 0.26203596591949463,
+      "learning_rate": 1.0064446072671651e-06,
+      "loss": 0.01,
+      "step": 6620
+    },
+    {
+      "epoch": 4.790462427745664,
+      "grad_norm": 0.1384071260690689,
+      "learning_rate": 9.399161231010811e-07,
+      "loss": 0.0059,
+      "step": 6630
+    },
+    {
+      "epoch": 4.797687861271676,
+      "grad_norm": 0.11224517226219177,
+      "learning_rate": 8.756519426292742e-07,
+      "loss": 0.0065,
+      "step": 6640
+    },
+    {
+      "epoch": 4.804913294797688,
+      "grad_norm": 0.05504254996776581,
+      "learning_rate": 8.136535347948226e-07,
+      "loss": 0.0073,
+      "step": 6650
+    },
+    {
+      "epoch": 4.812138728323699,
+      "grad_norm": 0.10972186177968979,
+      "learning_rate": 7.539223167499776e-07,
+      "loss": 0.01,
+      "step": 6660
+    },
+    {
+      "epoch": 4.819364161849711,
+      "grad_norm": 0.09454052150249481,
+      "learning_rate": 6.964596538238777e-07,
+      "loss": 0.01,
+      "step": 6670
+    },
+    {
+      "epoch": 4.826589595375722,
+      "grad_norm": 0.10990437120199203,
+      "learning_rate": 6.412668594912963e-07,
+      "loss": 0.0072,
+      "step": 6680
+    },
+    {
+      "epoch": 4.833815028901734,
+      "grad_norm": 0.10330501943826675,
+      "learning_rate": 5.883451953425879e-07,
+      "loss": 0.0069,
+      "step": 6690
+    },
+    {
+      "epoch": 4.841040462427745,
+      "grad_norm": 0.28672897815704346,
+      "learning_rate": 5.376958710548996e-07,
+      "loss": 0.0117,
+      "step": 6700
+    },
+    {
+      "epoch": 4.848265895953757,
+      "grad_norm": 0.31269845366477966,
+      "learning_rate": 4.893200443645163e-07,
+      "loss": 0.0077,
+      "step": 6710
+    },
+    {
+      "epoch": 4.855491329479769,
+      "grad_norm": 0.1288710981607437,
+      "learning_rate": 4.432188210403476e-07,
+      "loss": 0.0063,
+      "step": 6720
+    },
+    {
+      "epoch": 4.86271676300578,
+      "grad_norm": 0.2598627209663391,
+      "learning_rate": 3.993932548587376e-07,
+      "loss": 0.0124,
+      "step": 6730
+    },
+    {
+      "epoch": 4.869942196531792,
+      "grad_norm": 0.1576113998889923,
+      "learning_rate": 3.5784434757928364e-07,
+      "loss": 0.0114,
+      "step": 6740
+    },
+    {
+      "epoch": 4.877167630057803,
+      "grad_norm": 0.0818244218826294,
+      "learning_rate": 3.1857304892198845e-07,
+      "loss": 0.011,
+      "step": 6750
+    },
+    {
+      "epoch": 4.884393063583815,
+      "grad_norm": 0.17850443720817566,
+      "learning_rate": 2.815802565455772e-07,
+      "loss": 0.0099,
+      "step": 6760
+    },
+    {
+      "epoch": 4.891618497109826,
+      "grad_norm": 0.15844108164310455,
+      "learning_rate": 2.46866816026925e-07,
+      "loss": 0.011,
+      "step": 6770
+    },
+    {
+      "epoch": 4.898843930635838,
+      "grad_norm": 0.26582100987434387,
+      "learning_rate": 2.1443352084175027e-07,
+      "loss": 0.0065,
+      "step": 6780
+    },
+    {
+      "epoch": 4.90606936416185,
+      "grad_norm": 0.1282494217157364,
+      "learning_rate": 1.8428111234648493e-07,
+      "loss": 0.006,
+      "step": 6790
+    },
+    {
+      "epoch": 4.913294797687861,
+      "grad_norm": 0.15636210143566132,
+      "learning_rate": 1.5641027976130984e-07,
+      "loss": 0.0141,
+      "step": 6800
+    },
+    {
+      "epoch": 4.920520231213873,
+      "grad_norm": 0.07667110115289688,
+      "learning_rate": 1.3082166015443432e-07,
+      "loss": 0.0066,
+      "step": 6810
+    },
+    {
+      "epoch": 4.927745664739884,
+      "grad_norm": 0.06386137008666992,
+      "learning_rate": 1.0751583842750768e-07,
+      "loss": 0.012,
+      "step": 6820
+    },
+    {
+      "epoch": 4.934971098265896,
+      "grad_norm": 0.18531806766986847,
+      "learning_rate": 8.649334730223002e-08,
+      "loss": 0.0105,
+      "step": 6830
+    },
+    {
+      "epoch": 4.942196531791907,
+      "grad_norm": 0.23768620193004608,
+      "learning_rate": 6.775466730823965e-08,
+      "loss": 0.0091,
+      "step": 6840
+    },
+    {
+      "epoch": 4.949421965317919,
+      "grad_norm": 0.16282473504543304,
+      "learning_rate": 5.130022677205526e-08,
+      "loss": 0.0066,
+      "step": 6850
+    },
+    {
+      "epoch": 4.956647398843931,
+      "grad_norm": 0.11510784178972244,
+      "learning_rate": 3.713040180733929e-08,
+      "loss": 0.0057,
+      "step": 6860
+    },
+    {
+      "epoch": 4.963872832369942,
+      "grad_norm": 0.10748834908008575,
+      "learning_rate": 2.5245516306293683e-08,
+      "loss": 0.0073,
+      "step": 6870
+    },
+    {
+      "epoch": 4.971098265895954,
+      "grad_norm": 0.22777627408504486,
+      "learning_rate": 1.56458419321992e-08,
+      "loss": 0.0088,
+      "step": 6880
+    },
+    {
+      "epoch": 4.978323699421965,
+      "grad_norm": 0.1276501715183258,
+      "learning_rate": 8.331598113298089e-09,
+      "loss": 0.0141,
+      "step": 6890
+    },
+    {
+      "epoch": 4.985549132947977,
+      "grad_norm": 0.22229315340518951,
+      "learning_rate": 3.3029520376981483e-09,
+      "loss": 0.0077,
+      "step": 6900
+    },
+    {
+      "epoch": 4.992774566473988,
+      "grad_norm": 0.1325089931488037,
+      "learning_rate": 5.600186496090842e-10,
+      "loss": 0.0102,
+      "step": 6910
+    },
+    {
+      "epoch": 4.997832369942197,
+      "step": 6917,
+      "total_flos": 2.4387263668176077e+17,
+      "train_loss": 0.02989626239009877,
+      "train_runtime": 3134.2246,
+      "train_samples_per_second": 35.311,
+      "train_steps_per_second": 2.207
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6917,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.4387263668176077e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}