diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7043 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 14.204545454545455,
+  "eval_steps": 500,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.014204545454545454,
+      "grad_norm": 12.840788841247559,
+      "learning_rate": 1.8e-06,
+      "loss": 1.3524,
+      "step": 10
+    },
+    {
+      "epoch": 0.028409090909090908,
+      "grad_norm": 5.040924549102783,
+      "learning_rate": 3.8e-06,
+      "loss": 1.0372,
+      "step": 20
+    },
+    {
+      "epoch": 0.04261363636363636,
+      "grad_norm": 4.226280689239502,
+      "learning_rate": 5.8e-06,
+      "loss": 0.538,
+      "step": 30
+    },
+    {
+      "epoch": 0.056818181818181816,
+      "grad_norm": 3.5317652225494385,
+      "learning_rate": 7.8e-06,
+      "loss": 0.3239,
+      "step": 40
+    },
+    {
+      "epoch": 0.07102272727272728,
+      "grad_norm": 1.819808840751648,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.2752,
+      "step": 50
+    },
+    {
+      "epoch": 0.08522727272727272,
+      "grad_norm": 2.0116279125213623,
+      "learning_rate": 1.18e-05,
+      "loss": 0.1845,
+      "step": 60
+    },
+    {
+      "epoch": 0.09943181818181818,
+      "grad_norm": 1.7987759113311768,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 0.1192,
+      "step": 70
+    },
+    {
+      "epoch": 0.11363636363636363,
+      "grad_norm": 1.3431462049484253,
+      "learning_rate": 1.58e-05,
+      "loss": 0.1265,
+      "step": 80
+    },
+    {
+      "epoch": 0.1278409090909091,
+      "grad_norm": 0.9252213835716248,
+      "learning_rate": 1.78e-05,
+      "loss": 0.085,
+      "step": 90
+    },
+    {
+      "epoch": 0.14204545454545456,
+      "grad_norm": 1.463362216949463,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 0.0845,
+      "step": 100
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 1.0895040035247803,
+      "learning_rate": 2.18e-05,
+      "loss": 0.0881,
+      "step": 110
+    },
+    {
+      "epoch": 0.17045454545454544,
+      "grad_norm": 1.1225346326828003,
+      "learning_rate": 2.38e-05,
+      "loss": 0.073,
+      "step": 120
+    },
+    {
+      "epoch": 0.1846590909090909,
+      "grad_norm": 1.0712026357650757,
+      "learning_rate": 2.58e-05,
+      "loss": 0.078,
+      "step": 130
+    },
+    {
+      "epoch": 0.19886363636363635,
+      "grad_norm": 0.6329981088638306,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 0.0756,
+      "step": 140
+    },
+    {
+      "epoch": 0.21306818181818182,
+      "grad_norm": 1.0224592685699463,
+      "learning_rate": 2.98e-05,
+      "loss": 0.0651,
+      "step": 150
+    },
+    {
+      "epoch": 0.22727272727272727,
+      "grad_norm": 0.5526351928710938,
+      "learning_rate": 3.18e-05,
+      "loss": 0.0629,
+      "step": 160
+    },
+    {
+      "epoch": 0.24147727272727273,
+      "grad_norm": 0.8906152844429016,
+      "learning_rate": 3.38e-05,
+      "loss": 0.0607,
+      "step": 170
+    },
+    {
+      "epoch": 0.2556818181818182,
+      "grad_norm": 1.0715312957763672,
+      "learning_rate": 3.58e-05,
+      "loss": 0.0642,
+      "step": 180
+    },
+    {
+      "epoch": 0.26988636363636365,
+      "grad_norm": 0.9697608351707458,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 0.0583,
+      "step": 190
+    },
+    {
+      "epoch": 0.2840909090909091,
+      "grad_norm": 0.5903347134590149,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 0.0622,
+      "step": 200
+    },
+    {
+      "epoch": 0.29829545454545453,
+      "grad_norm": 0.6113477945327759,
+      "learning_rate": 4.18e-05,
+      "loss": 0.0651,
+      "step": 210
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.7239237427711487,
+      "learning_rate": 4.38e-05,
+      "loss": 0.0631,
+      "step": 220
+    },
+    {
+      "epoch": 0.32670454545454547,
+      "grad_norm": 0.8973392844200134,
+      "learning_rate": 4.58e-05,
+      "loss": 0.0576,
+      "step": 230
+    },
+    {
+      "epoch": 0.3409090909090909,
+      "grad_norm": 0.8834447860717773,
+      "learning_rate": 4.78e-05,
+      "loss": 0.0553,
+      "step": 240
+    },
+    {
+      "epoch": 0.35511363636363635,
+      "grad_norm": 1.186426043510437,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 0.0532,
+      "step": 250
+    },
+    {
+      "epoch": 0.3693181818181818,
+      "grad_norm": 0.9951111674308777,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 0.0623,
+      "step": 260
+    },
+    {
+      "epoch": 0.3835227272727273,
+      "grad_norm": 1.2119064331054688,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 0.0582,
+      "step": 270
+    },
+    {
+      "epoch": 0.3977272727272727,
+      "grad_norm": 0.8901904821395874,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 0.0514,
+      "step": 280
+    },
+    {
+      "epoch": 0.4119318181818182,
+      "grad_norm": 0.5448264479637146,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.0582,
+      "step": 290
+    },
+    {
+      "epoch": 0.42613636363636365,
+      "grad_norm": 0.789374589920044,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.053,
+      "step": 300
+    },
+    {
+      "epoch": 0.4403409090909091,
+      "grad_norm": 0.8261776566505432,
+      "learning_rate": 6.18e-05,
+      "loss": 0.0532,
+      "step": 310
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 0.39446914196014404,
+      "learning_rate": 6.38e-05,
+      "loss": 0.0451,
+      "step": 320
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.5507211685180664,
+      "learning_rate": 6.58e-05,
+      "loss": 0.053,
+      "step": 330
+    },
+    {
+      "epoch": 0.48295454545454547,
+      "grad_norm": 0.9800366163253784,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 0.0501,
+      "step": 340
+    },
+    {
+      "epoch": 0.4971590909090909,
+      "grad_norm": 0.9592909216880798,
+      "learning_rate": 6.98e-05,
+      "loss": 0.0507,
+      "step": 350
+    },
+    {
+      "epoch": 0.5113636363636364,
+      "grad_norm": 0.7082611918449402,
+      "learning_rate": 7.18e-05,
+      "loss": 0.0535,
+      "step": 360
+    },
+    {
+      "epoch": 0.5255681818181818,
+      "grad_norm": 1.1562505960464478,
+      "learning_rate": 7.38e-05,
+      "loss": 0.0632,
+      "step": 370
+    },
+    {
+      "epoch": 0.5397727272727273,
+      "grad_norm": 0.5484411716461182,
+      "learning_rate": 7.58e-05,
+      "loss": 0.0486,
+      "step": 380
+    },
+    {
+      "epoch": 0.5539772727272727,
+      "grad_norm": 0.7503770589828491,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.0471,
+      "step": 390
+    },
+    {
+      "epoch": 0.5681818181818182,
+      "grad_norm": 0.7870602607727051,
+      "learning_rate": 7.98e-05,
+      "loss": 0.0476,
+      "step": 400
+    },
+    {
+      "epoch": 0.5823863636363636,
+      "grad_norm": 0.7637655735015869,
+      "learning_rate": 8.18e-05,
+      "loss": 0.0524,
+      "step": 410
+    },
+    {
+      "epoch": 0.5965909090909091,
+      "grad_norm": 0.49145546555519104,
+      "learning_rate": 8.38e-05,
+      "loss": 0.0397,
+      "step": 420
+    },
+    {
+      "epoch": 0.6107954545454546,
+      "grad_norm": 1.0262115001678467,
+      "learning_rate": 8.58e-05,
+      "loss": 0.0474,
+      "step": 430
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 1.046718716621399,
+      "learning_rate": 8.78e-05,
+      "loss": 0.0474,
+      "step": 440
+    },
+    {
+      "epoch": 0.6392045454545454,
+      "grad_norm": 0.8363006711006165,
+      "learning_rate": 8.98e-05,
+      "loss": 0.0444,
+      "step": 450
+    },
+    {
+      "epoch": 0.6534090909090909,
+      "grad_norm": 0.8325658440589905,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 0.0511,
+      "step": 460
+    },
+    {
+      "epoch": 0.6676136363636364,
+      "grad_norm": 0.711004376411438,
+      "learning_rate": 9.38e-05,
+      "loss": 0.0516,
+      "step": 470
+    },
+    {
+      "epoch": 0.6818181818181818,
+      "grad_norm": 0.8718818426132202,
+      "learning_rate": 9.58e-05,
+      "loss": 0.0557,
+      "step": 480
+    },
+    {
+      "epoch": 0.6960227272727273,
+      "grad_norm": 0.5283912420272827,
+      "learning_rate": 9.78e-05,
+      "loss": 0.0417,
+      "step": 490
+    },
+    {
+      "epoch": 0.7102272727272727,
+      "grad_norm": 0.6763261556625366,
+      "learning_rate": 9.98e-05,
+      "loss": 0.0526,
+      "step": 500
+    },
+    {
+      "epoch": 0.7244318181818182,
+      "grad_norm": 0.493693083524704,
+      "learning_rate": 9.9999778549206e-05,
+      "loss": 0.0478,
+      "step": 510
+    },
+    {
+      "epoch": 0.7386363636363636,
+      "grad_norm": 0.7801306843757629,
+      "learning_rate": 9.999901304280685e-05,
+      "loss": 0.0471,
+      "step": 520
+    },
+    {
+      "epoch": 0.7528409090909091,
+      "grad_norm": 0.5438108444213867,
+      "learning_rate": 9.999770075521164e-05,
+      "loss": 0.0505,
+      "step": 530
+    },
+    {
+      "epoch": 0.7670454545454546,
+      "grad_norm": 0.9757444858551025,
+      "learning_rate": 9.99958417007713e-05,
+      "loss": 0.0494,
+      "step": 540
+    },
+    {
+      "epoch": 0.78125,
+      "grad_norm": 0.5842338800430298,
+      "learning_rate": 9.999343589981615e-05,
+      "loss": 0.0422,
+      "step": 550
+    },
+    {
+      "epoch": 0.7954545454545454,
+      "grad_norm": 0.5453335642814636,
+      "learning_rate": 9.999048337865568e-05,
+      "loss": 0.0416,
+      "step": 560
+    },
+    {
+      "epoch": 0.8096590909090909,
+      "grad_norm": 0.454010933637619,
+      "learning_rate": 9.998698416957815e-05,
+      "loss": 0.0429,
+      "step": 570
+    },
+    {
+      "epoch": 0.8238636363636364,
+      "grad_norm": 0.9522406458854675,
+      "learning_rate": 9.998293831085037e-05,
+      "loss": 0.0395,
+      "step": 580
+    },
+    {
+      "epoch": 0.8380681818181818,
+      "grad_norm": 0.6580008864402771,
+      "learning_rate": 9.997834584671719e-05,
+      "loss": 0.0425,
+      "step": 590
+    },
+    {
+      "epoch": 0.8522727272727273,
+      "grad_norm": 0.6695672869682312,
+      "learning_rate": 9.997320682740107e-05,
+      "loss": 0.0482,
+      "step": 600
+    },
+    {
+      "epoch": 0.8664772727272727,
+      "grad_norm": 0.6290060877799988,
+      "learning_rate": 9.996752130910149e-05,
+      "loss": 0.0387,
+      "step": 610
+    },
+    {
+      "epoch": 0.8806818181818182,
+      "grad_norm": 0.7723461985588074,
+      "learning_rate": 9.99612893539944e-05,
+      "loss": 0.0352,
+      "step": 620
+    },
+    {
+      "epoch": 0.8948863636363636,
+      "grad_norm": 0.8723970055580139,
+      "learning_rate": 9.995451103023144e-05,
+      "loss": 0.0353,
+      "step": 630
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 0.6582695245742798,
+      "learning_rate": 9.994718641193928e-05,
+      "loss": 0.0324,
+      "step": 640
+    },
+    {
+      "epoch": 0.9232954545454546,
+      "grad_norm": 0.6307513117790222,
+      "learning_rate": 9.993931557921874e-05,
+      "loss": 0.042,
+      "step": 650
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 0.8004989624023438,
+      "learning_rate": 9.993089861814402e-05,
+      "loss": 0.0396,
+      "step": 660
+    },
+    {
+      "epoch": 0.9517045454545454,
+      "grad_norm": 0.697421669960022,
+      "learning_rate": 9.992193562076166e-05,
+      "loss": 0.0561,
+      "step": 670
+    },
+    {
+      "epoch": 0.9659090909090909,
+      "grad_norm": 0.9708011746406555,
+      "learning_rate": 9.991242668508954e-05,
+      "loss": 0.0501,
+      "step": 680
+    },
+    {
+      "epoch": 0.9801136363636364,
+      "grad_norm": 0.6217566132545471,
+      "learning_rate": 9.990237191511587e-05,
+      "loss": 0.046,
+      "step": 690
+    },
+    {
+      "epoch": 0.9943181818181818,
+      "grad_norm": 0.9400615692138672,
+      "learning_rate": 9.989177142079802e-05,
+      "loss": 0.0508,
+      "step": 700
+    },
+    {
+      "epoch": 1.0085227272727273,
+      "grad_norm": 0.8677662014961243,
+      "learning_rate": 9.988062531806126e-05,
+      "loss": 0.0526,
+      "step": 710
+    },
+    {
+      "epoch": 1.0227272727272727,
+      "grad_norm": 0.8316492438316345,
+      "learning_rate": 9.986893372879762e-05,
+      "loss": 0.0426,
+      "step": 720
+    },
+    {
+      "epoch": 1.0369318181818181,
+      "grad_norm": 0.3717138171195984,
+      "learning_rate": 9.985669678086443e-05,
+      "loss": 0.0389,
+      "step": 730
+    },
+    {
+      "epoch": 1.0511363636363635,
+      "grad_norm": 0.5807476043701172,
+      "learning_rate": 9.984391460808298e-05,
+      "loss": 0.0377,
+      "step": 740
+    },
+    {
+      "epoch": 1.0653409090909092,
+      "grad_norm": 0.5002987384796143,
+      "learning_rate": 9.983058735023709e-05,
+      "loss": 0.0373,
+      "step": 750
+    },
+    {
+      "epoch": 1.0795454545454546,
+      "grad_norm": 0.5314595699310303,
+      "learning_rate": 9.98167151530715e-05,
+      "loss": 0.0315,
+      "step": 760
+    },
+    {
+      "epoch": 1.09375,
+      "grad_norm": 0.6090950965881348,
+      "learning_rate": 9.980229816829034e-05,
+      "loss": 0.0386,
+      "step": 770
+    },
+    {
+      "epoch": 1.1079545454545454,
+      "grad_norm": 0.4911409616470337,
+      "learning_rate": 9.978733655355544e-05,
+      "loss": 0.0324,
+      "step": 780
+    },
+    {
+      "epoch": 1.1221590909090908,
+      "grad_norm": 0.49719247221946716,
+      "learning_rate": 9.977183047248464e-05,
+      "loss": 0.0319,
+      "step": 790
+    },
+    {
+      "epoch": 1.1363636363636362,
+      "grad_norm": 0.5944560170173645,
+      "learning_rate": 9.975578009464992e-05,
+      "loss": 0.0358,
+      "step": 800
+    },
+    {
+      "epoch": 1.1505681818181819,
+      "grad_norm": 0.6031031012535095,
+      "learning_rate": 9.97391855955757e-05,
+      "loss": 0.0408,
+      "step": 810
+    },
+    {
+      "epoch": 1.1647727272727273,
+      "grad_norm": 0.5565268993377686,
+      "learning_rate": 9.972204715673669e-05,
+      "loss": 0.0357,
+      "step": 820
+    },
+    {
+      "epoch": 1.1789772727272727,
+      "grad_norm": 0.6696913242340088,
+      "learning_rate": 9.970436496555617e-05,
+      "loss": 0.0325,
+      "step": 830
+    },
+    {
+      "epoch": 1.1931818181818181,
+      "grad_norm": 0.43231725692749023,
+      "learning_rate": 9.968613921540373e-05,
+      "loss": 0.029,
+      "step": 840
+    },
+    {
+      "epoch": 1.2073863636363638,
+      "grad_norm": 0.2592962086200714,
+      "learning_rate": 9.966737010559326e-05,
+      "loss": 0.0323,
+      "step": 850
+    },
+    {
+      "epoch": 1.2215909090909092,
+      "grad_norm": 0.4336261451244354,
+      "learning_rate": 9.964805784138072e-05,
+      "loss": 0.0325,
+      "step": 860
+    },
+    {
+      "epoch": 1.2357954545454546,
+      "grad_norm": 0.4406537711620331,
+      "learning_rate": 9.962820263396195e-05,
+      "loss": 0.0299,
+      "step": 870
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4915785789489746,
+      "learning_rate": 9.960780470047033e-05,
+      "loss": 0.0301,
+      "step": 880
+    },
+    {
+      "epoch": 1.2642045454545454,
+      "grad_norm": 0.44726672768592834,
+      "learning_rate": 9.958686426397437e-05,
+      "loss": 0.0285,
+      "step": 890
+    },
+    {
+      "epoch": 1.2784090909090908,
+      "grad_norm": 0.5937303900718689,
+      "learning_rate": 9.956538155347534e-05,
+      "loss": 0.0376,
+      "step": 900
+    },
+    {
+      "epoch": 1.2926136363636362,
+      "grad_norm": 0.7091760039329529,
+      "learning_rate": 9.95433568039047e-05,
+      "loss": 0.0275,
+      "step": 910
+    },
+    {
+      "epoch": 1.3068181818181819,
+      "grad_norm": 0.300912082195282,
+      "learning_rate": 9.952079025612162e-05,
+      "loss": 0.0391,
+      "step": 920
+    },
+    {
+      "epoch": 1.3210227272727273,
+      "grad_norm": 0.6073192358016968,
+      "learning_rate": 9.949768215691022e-05,
+      "loss": 0.04,
+      "step": 930
+    },
+    {
+      "epoch": 1.3352272727272727,
+      "grad_norm": 0.4603278934955597,
+      "learning_rate": 9.9474032758977e-05,
+      "loss": 0.0314,
+      "step": 940
+    },
+    {
+      "epoch": 1.3494318181818181,
+      "grad_norm": 0.40913715958595276,
+      "learning_rate": 9.944984232094794e-05,
+      "loss": 0.0411,
+      "step": 950
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 0.6388444900512695,
+      "learning_rate": 9.942511110736584e-05,
+      "loss": 0.0386,
+      "step": 960
+    },
+    {
+      "epoch": 1.3778409090909092,
+      "grad_norm": 0.4576031565666199,
+      "learning_rate": 9.939983938868726e-05,
+      "loss": 0.0285,
+      "step": 970
+    },
+    {
+      "epoch": 1.3920454545454546,
+      "grad_norm": 0.7075598239898682,
+      "learning_rate": 9.93740274412797e-05,
+      "loss": 0.0316,
+      "step": 980
+    },
+    {
+      "epoch": 1.40625,
+      "grad_norm": 0.5083079934120178,
+      "learning_rate": 9.934767554741846e-05,
+      "loss": 0.0378,
+      "step": 990
+    },
+    {
+      "epoch": 1.4204545454545454,
+      "grad_norm": 0.6349732279777527,
+      "learning_rate": 9.932078399528361e-05,
+      "loss": 0.0342,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4346590909090908,
+      "grad_norm": 0.47263529896736145,
+      "learning_rate": 9.929335307895689e-05,
+      "loss": 0.0305,
+      "step": 1010
+    },
+    {
+      "epoch": 1.4488636363636362,
+      "grad_norm": 0.40445226430892944,
+      "learning_rate": 9.926538309841839e-05,
+      "loss": 0.0312,
+      "step": 1020
+    },
+    {
+      "epoch": 1.4630681818181819,
+      "grad_norm": 0.4489366114139557,
+      "learning_rate": 9.923687435954334e-05,
+      "loss": 0.0327,
+      "step": 1030
+    },
+    {
+      "epoch": 1.4772727272727273,
+      "grad_norm": 0.634887158870697,
+      "learning_rate": 9.920782717409873e-05,
+      "loss": 0.0344,
+      "step": 1040
+    },
+    {
+      "epoch": 1.4914772727272727,
+      "grad_norm": 0.3030359745025635,
+      "learning_rate": 9.917824185973994e-05,
+      "loss": 0.0279,
+      "step": 1050
+    },
+    {
+      "epoch": 1.5056818181818183,
+      "grad_norm": 0.37087854743003845,
+      "learning_rate": 9.914811874000723e-05,
+      "loss": 0.0366,
+      "step": 1060
+    },
+    {
+      "epoch": 1.5198863636363638,
+      "grad_norm": 0.45555105805397034,
+      "learning_rate": 9.911745814432218e-05,
+      "loss": 0.0304,
+      "step": 1070
+    },
+    {
+      "epoch": 1.5340909090909092,
+      "grad_norm": 0.33378538489341736,
+      "learning_rate": 9.90862604079842e-05,
+      "loss": 0.0341,
+      "step": 1080
+    },
+    {
+      "epoch": 1.5482954545454546,
+      "grad_norm": 0.3594984710216522,
+      "learning_rate": 9.90545258721667e-05,
+      "loss": 0.0272,
+      "step": 1090
+    },
+    {
+      "epoch": 1.5625,
+      "grad_norm": 0.4058392643928528,
+      "learning_rate": 9.90222548839135e-05,
+      "loss": 0.0324,
+      "step": 1100
+    },
+    {
+      "epoch": 1.5767045454545454,
+      "grad_norm": 0.4372110664844513,
+      "learning_rate": 9.898944779613495e-05,
+      "loss": 0.0316,
+      "step": 1110
+    },
+    {
+      "epoch": 1.5909090909090908,
+      "grad_norm": 0.572181761264801,
+      "learning_rate": 9.89561049676041e-05,
+      "loss": 0.0269,
+      "step": 1120
+    },
+    {
+      "epoch": 1.6051136363636362,
+      "grad_norm": 0.3711259067058563,
+      "learning_rate": 9.89222267629528e-05,
+      "loss": 0.0335,
+      "step": 1130
+    },
+    {
+      "epoch": 1.6193181818181817,
+      "grad_norm": 0.34197402000427246,
+      "learning_rate": 9.888781355266763e-05,
+      "loss": 0.0254,
+      "step": 1140
+    },
+    {
+      "epoch": 1.6335227272727273,
+      "grad_norm": 0.5190102458000183,
+      "learning_rate": 9.885286571308598e-05,
+      "loss": 0.0314,
+      "step": 1150
+    },
+    {
+      "epoch": 1.6477272727272727,
+      "grad_norm": 0.5521939396858215,
+      "learning_rate": 9.881738362639182e-05,
+      "loss": 0.0315,
+      "step": 1160
+    },
+    {
+      "epoch": 1.6619318181818183,
+      "grad_norm": 0.5134572386741638,
+      "learning_rate": 9.878136768061154e-05,
+      "loss": 0.0283,
+      "step": 1170
+    },
+    {
+      "epoch": 1.6761363636363638,
+      "grad_norm": 0.40377435088157654,
+      "learning_rate": 9.874481826960979e-05,
+      "loss": 0.0267,
+      "step": 1180
+    },
+    {
+      "epoch": 1.6903409090909092,
+      "grad_norm": 0.5408985614776611,
+      "learning_rate": 9.870773579308503e-05,
+      "loss": 0.0267,
+      "step": 1190
+    },
+    {
+      "epoch": 1.7045454545454546,
+      "grad_norm": 0.5037901997566223,
+      "learning_rate": 9.867012065656533e-05,
+      "loss": 0.0322,
+      "step": 1200
+    },
+    {
+      "epoch": 1.71875,
+      "grad_norm": 0.6231687068939209,
+      "learning_rate": 9.863197327140376e-05,
+      "loss": 0.0342,
+      "step": 1210
+    },
+    {
+      "epoch": 1.7329545454545454,
+      "grad_norm": 0.3556486964225769,
+      "learning_rate": 9.859329405477403e-05,
+      "loss": 0.0256,
+      "step": 1220
+    },
+    {
+      "epoch": 1.7471590909090908,
+      "grad_norm": 0.5827201008796692,
+      "learning_rate": 9.855408342966585e-05,
+      "loss": 0.0299,
+      "step": 1230
+    },
+    {
+      "epoch": 1.7613636363636362,
+      "grad_norm": 0.45363086462020874,
+      "learning_rate": 9.851434182488033e-05,
+      "loss": 0.029,
+      "step": 1240
+    },
+    {
+      "epoch": 1.7755681818181817,
+      "grad_norm": 0.4468866288661957,
+      "learning_rate": 9.84740696750253e-05,
+      "loss": 0.0321,
+      "step": 1250
+    },
+    {
+      "epoch": 1.7897727272727273,
+      "grad_norm": 0.453463077545166,
+      "learning_rate": 9.843326742051055e-05,
+      "loss": 0.0268,
+      "step": 1260
+    },
+    {
+      "epoch": 1.8039772727272727,
+      "grad_norm": 0.5213397145271301,
+      "learning_rate": 9.839193550754297e-05,
+      "loss": 0.0341,
+      "step": 1270
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 0.38024085760116577,
+      "learning_rate": 9.835007438812177e-05,
+      "loss": 0.0325,
+      "step": 1280
+    },
+    {
+      "epoch": 1.8323863636363638,
+      "grad_norm": 0.4297795593738556,
+      "learning_rate": 9.830768452003341e-05,
+      "loss": 0.0344,
+      "step": 1290
+    },
+    {
+      "epoch": 1.8465909090909092,
+      "grad_norm": 0.5636400580406189,
+      "learning_rate": 9.826476636684671e-05,
+      "loss": 0.0332,
+      "step": 1300
+    },
+    {
+      "epoch": 1.8607954545454546,
+      "grad_norm": 0.30375558137893677,
+      "learning_rate": 9.822132039790773e-05,
+      "loss": 0.0256,
+      "step": 1310
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 0.5967488288879395,
+      "learning_rate": 9.817734708833461e-05,
+      "loss": 0.0313,
+      "step": 1320
+    },
+    {
+      "epoch": 1.8892045454545454,
+      "grad_norm": 0.7265820503234863,
+      "learning_rate": 9.813284691901243e-05,
+      "loss": 0.0324,
+      "step": 1330
+    },
+    {
+      "epoch": 1.9034090909090908,
+      "grad_norm": 0.5015584230422974,
+      "learning_rate": 9.808782037658792e-05,
+      "loss": 0.0298,
+      "step": 1340
+    },
+    {
+      "epoch": 1.9176136363636362,
+      "grad_norm": 0.46016088128089905,
+      "learning_rate": 9.804226795346411e-05,
+      "loss": 0.0263,
+      "step": 1350
+    },
+    {
+      "epoch": 1.9318181818181817,
+      "grad_norm": 0.4038504660129547,
+      "learning_rate": 9.799619014779503e-05,
+      "loss": 0.032,
+      "step": 1360
+    },
+    {
+      "epoch": 1.9460227272727273,
+      "grad_norm": 0.5546528697013855,
+      "learning_rate": 9.794958746348013e-05,
+      "loss": 0.0292,
+      "step": 1370
+    },
+    {
+      "epoch": 1.9602272727272727,
+      "grad_norm": 0.4329584538936615,
+      "learning_rate": 9.790246041015896e-05,
+      "loss": 0.0329,
+      "step": 1380
+    },
+    {
+      "epoch": 1.9744318181818183,
+      "grad_norm": 0.3155853748321533,
+      "learning_rate": 9.785480950320538e-05,
+      "loss": 0.0269,
+      "step": 1390
+    },
+    {
+      "epoch": 1.9886363636363638,
+      "grad_norm": 0.41486474871635437,
+      "learning_rate": 9.78066352637221e-05,
+      "loss": 0.0267,
+      "step": 1400
+    },
+    {
+      "epoch": 2.002840909090909,
+      "grad_norm": 0.3981183171272278,
+      "learning_rate": 9.775793821853488e-05,
+      "loss": 0.0224,
+      "step": 1410
+    },
+    {
+      "epoch": 2.0170454545454546,
+      "grad_norm": 0.5503771901130676,
+      "learning_rate": 9.77087189001868e-05,
+      "loss": 0.0279,
+      "step": 1420
+    },
+    {
+      "epoch": 2.03125,
+      "grad_norm": 0.38202399015426636,
+      "learning_rate": 9.765897784693243e-05,
+      "loss": 0.0269,
+      "step": 1430
+    },
+    {
+      "epoch": 2.0454545454545454,
+      "grad_norm": 0.2967958450317383,
+      "learning_rate": 9.760871560273197e-05,
+      "loss": 0.0239,
+      "step": 1440
+    },
+    {
+      "epoch": 2.059659090909091,
+      "grad_norm": 0.41418150067329407,
+      "learning_rate": 9.755793271724526e-05,
+      "loss": 0.026,
+      "step": 1450
+    },
+    {
+      "epoch": 2.0738636363636362,
+      "grad_norm": 0.3536577522754669,
+      "learning_rate": 9.750662974582584e-05,
+      "loss": 0.0271,
+      "step": 1460
+    },
+    {
+      "epoch": 2.0880681818181817,
+      "grad_norm": 0.3064652681350708,
+      "learning_rate": 9.745480724951473e-05,
+      "loss": 0.0282,
+      "step": 1470
+    },
+    {
+      "epoch": 2.102272727272727,
+      "grad_norm": 0.32034602761268616,
+      "learning_rate": 9.740246579503447e-05,
+      "loss": 0.0325,
+      "step": 1480
+    },
+    {
+      "epoch": 2.116477272727273,
+      "grad_norm": 0.19377802312374115,
+      "learning_rate": 9.734960595478284e-05,
+      "loss": 0.0246,
+      "step": 1490
+    },
+    {
+      "epoch": 2.1306818181818183,
+      "grad_norm": 0.4557360112667084,
+      "learning_rate": 9.729622830682657e-05,
+      "loss": 0.0297,
+      "step": 1500
+    },
+    {
+      "epoch": 2.1448863636363638,
+      "grad_norm": 0.5449606776237488,
+      "learning_rate": 9.724233343489504e-05,
+      "loss": 0.0327,
+      "step": 1510
+    },
+    {
+      "epoch": 2.159090909090909,
+      "grad_norm": 0.40295103192329407,
+      "learning_rate": 9.718792192837396e-05,
+      "loss": 0.0282,
+      "step": 1520
+    },
+    {
+      "epoch": 2.1732954545454546,
+      "grad_norm": 0.5138129591941833,
+      "learning_rate": 9.713299438229886e-05,
+      "loss": 0.0329,
+      "step": 1530
+    },
+    {
+      "epoch": 2.1875,
+      "grad_norm": 0.5106748938560486,
+      "learning_rate": 9.707755139734855e-05,
+      "loss": 0.031,
+      "step": 1540
+    },
+    {
+      "epoch": 2.2017045454545454,
+      "grad_norm": 0.3726430833339691,
+      "learning_rate": 9.702159357983866e-05,
+      "loss": 0.0371,
+      "step": 1550
+    },
+    {
+      "epoch": 2.215909090909091,
+      "grad_norm": 0.45045050978660583,
+      "learning_rate": 9.696512154171492e-05,
+      "loss": 0.0251,
+      "step": 1560
+    },
+    {
+      "epoch": 2.2301136363636362,
+      "grad_norm": 0.22920694947242737,
+      "learning_rate": 9.690813590054645e-05,
+      "loss": 0.0239,
+      "step": 1570
+    },
+    {
+      "epoch": 2.2443181818181817,
+      "grad_norm": 0.3205345869064331,
+      "learning_rate": 9.685063727951914e-05,
+      "loss": 0.0252,
+      "step": 1580
+    },
+    {
+      "epoch": 2.2585227272727275,
+      "grad_norm": 0.5788774490356445,
+      "learning_rate": 9.679262630742865e-05,
+      "loss": 0.0307,
+      "step": 1590
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 0.41571733355522156,
+      "learning_rate": 9.673410361867373e-05,
+      "loss": 0.0269,
+      "step": 1600
+    },
+    {
+      "epoch": 2.2869318181818183,
+      "grad_norm": 0.6232924461364746,
+      "learning_rate": 9.667506985324909e-05,
+      "loss": 0.026,
+      "step": 1610
+    },
+    {
+      "epoch": 2.3011363636363638,
+      "grad_norm": 0.3497864007949829,
+      "learning_rate": 9.661552565673855e-05,
+      "loss": 0.0239,
+      "step": 1620
+    },
+    {
+      "epoch": 2.315340909090909,
+      "grad_norm": 0.6076606512069702,
+      "learning_rate": 9.655547168030789e-05,
+      "loss": 0.0261,
+      "step": 1630
+    },
+    {
+      "epoch": 2.3295454545454546,
+      "grad_norm": 0.4465327262878418,
+      "learning_rate": 9.649490858069777e-05,
+      "loss": 0.0239,
+      "step": 1640
+    },
+    {
+      "epoch": 2.34375,
+      "grad_norm": 0.6688684225082397,
+      "learning_rate": 9.643383702021658e-05,
+      "loss": 0.033,
+      "step": 1650
+    },
+    {
+      "epoch": 2.3579545454545454,
+      "grad_norm": 0.35796913504600525,
+      "learning_rate": 9.637225766673307e-05,
+      "loss": 0.0261,
+      "step": 1660
+    },
+    {
+      "epoch": 2.372159090909091,
+      "grad_norm": 0.4622032642364502,
+      "learning_rate": 9.631017119366922e-05,
+      "loss": 0.0243,
+      "step": 1670
+    },
+    {
+      "epoch": 2.3863636363636362,
+      "grad_norm": 0.5926036238670349,
+      "learning_rate": 9.624757827999273e-05,
+      "loss": 0.0301,
+      "step": 1680
+    },
+    {
+      "epoch": 2.4005681818181817,
+      "grad_norm": 0.3882672190666199,
+      "learning_rate": 9.618447961020971e-05,
+      "loss": 0.0298,
+      "step": 1690
+    },
+    {
+      "epoch": 2.4147727272727275,
+      "grad_norm": 0.5415935516357422,
+      "learning_rate": 9.612087587435707e-05,
+      "loss": 0.0229,
+      "step": 1700
+    },
+    {
+      "epoch": 2.4289772727272725,
+      "grad_norm": 0.4419732391834259,
+      "learning_rate": 9.605676776799508e-05,
+      "loss": 0.0261,
+      "step": 1710
+    },
+    {
+      "epoch": 2.4431818181818183,
+      "grad_norm": 0.5006458759307861,
+      "learning_rate": 9.599215599219973e-05,
+      "loss": 0.0279,
+      "step": 1720
+    },
+    {
+      "epoch": 2.4573863636363638,
+      "grad_norm": 0.5435750484466553,
+      "learning_rate": 9.592704125355505e-05,
+      "loss": 0.0312,
+      "step": 1730
+    },
+    {
+      "epoch": 2.471590909090909,
+      "grad_norm": 0.541560173034668,
+      "learning_rate": 9.586142426414538e-05,
+      "loss": 0.0268,
+      "step": 1740
+    },
+    {
+      "epoch": 2.4857954545454546,
+      "grad_norm": 0.41980552673339844,
+      "learning_rate": 9.57953057415476e-05,
+      "loss": 0.034,
+      "step": 1750
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.43404412269592285,
+      "learning_rate": 9.572868640882328e-05,
+      "loss": 0.03,
+      "step": 1760
+    },
+    {
+      "epoch": 2.5142045454545454,
+      "grad_norm": 0.4226009249687195,
+      "learning_rate": 9.56615669945108e-05,
+      "loss": 0.0269,
+      "step": 1770
+    },
+    {
+      "epoch": 2.528409090909091,
+      "grad_norm": 0.39670586585998535,
+      "learning_rate": 9.55939482326173e-05,
+      "loss": 0.0278,
+      "step": 1780
+    },
+    {
+      "epoch": 2.5426136363636362,
+      "grad_norm": 0.31577837467193604,
+      "learning_rate": 9.552583086261069e-05,
+      "loss": 0.0317,
+      "step": 1790
+    },
+    {
+      "epoch": 2.5568181818181817,
+      "grad_norm": 0.4553263783454895,
+      "learning_rate": 9.545721562941168e-05,
+      "loss": 0.0281,
+      "step": 1800
+    },
+    {
+      "epoch": 2.5710227272727275,
+      "grad_norm": 0.4637368619441986,
+      "learning_rate": 9.538810328338543e-05,
+      "loss": 0.0284,
+      "step": 1810
+    },
+    {
+      "epoch": 2.5852272727272725,
+      "grad_norm": 0.5123273730278015,
+      "learning_rate": 9.531849458033349e-05,
+      "loss": 0.0248,
+      "step": 1820
+    },
+    {
+      "epoch": 2.5994318181818183,
+      "grad_norm": 0.31737199425697327,
+      "learning_rate": 9.524839028148547e-05,
+      "loss": 0.0244,
+      "step": 1830
+    },
+    {
+      "epoch": 2.6136363636363638,
+      "grad_norm": 0.4044305980205536,
+      "learning_rate": 9.517779115349077e-05,
+      "loss": 0.0285,
+      "step": 1840
+    },
+    {
+      "epoch": 2.627840909090909,
+      "grad_norm": 0.19932806491851807,
+      "learning_rate": 9.510669796841014e-05,
+      "loss": 0.0261,
+      "step": 1850
+    },
+    {
+      "epoch": 2.6420454545454546,
+      "grad_norm": 0.25742536783218384,
+      "learning_rate": 9.503511150370727e-05,
+      "loss": 0.0263,
+      "step": 1860
+    },
+    {
+      "epoch": 2.65625,
+      "grad_norm": 0.3203023672103882,
+      "learning_rate": 9.496303254224024e-05,
+      "loss": 0.0279,
+      "step": 1870
+    },
+    {
+      "epoch": 2.6704545454545454,
+      "grad_norm": 0.34098610281944275,
+      "learning_rate": 9.489046187225306e-05,
+      "loss": 0.0274,
+      "step": 1880
+    },
+    {
+      "epoch": 2.684659090909091,
+      "grad_norm": 0.2678031921386719,
+      "learning_rate": 9.481740028736692e-05,
+      "loss": 0.024,
+      "step": 1890
+    },
+    {
+      "epoch": 2.6988636363636362,
+      "grad_norm": 0.598612904548645,
+      "learning_rate": 9.474384858657164e-05,
+      "loss": 0.0302,
+      "step": 1900
+    },
+    {
+      "epoch": 2.7130681818181817,
+      "grad_norm": 0.417934387922287,
+      "learning_rate": 9.466980757421679e-05,
+      "loss": 0.0283,
+      "step": 1910
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 0.444865882396698,
+      "learning_rate": 9.459527806000305e-05,
+      "loss": 0.0287,
+      "step": 1920
+    },
+    {
+      "epoch": 2.7414772727272725,
+      "grad_norm": 0.5610676407814026,
+      "learning_rate": 9.452026085897325e-05,
+      "loss": 0.0271,
+      "step": 1930
+    },
+    {
+      "epoch": 2.7556818181818183,
+      "grad_norm": 0.3354893624782562,
+      "learning_rate": 9.444475679150348e-05,
+      "loss": 0.0243,
+      "step": 1940
+    },
+    {
+      "epoch": 2.7698863636363638,
+      "grad_norm": 0.2516446113586426,
+      "learning_rate": 9.436876668329411e-05,
+      "loss": 0.0218,
+      "step": 1950
+    },
+    {
+      "epoch": 2.784090909090909,
+      "grad_norm": 0.25111836194992065,
+      "learning_rate": 9.429229136536079e-05,
+      "loss": 0.0223,
+      "step": 1960
+    },
+    {
+      "epoch": 2.7982954545454546,
+      "grad_norm": 0.5295560359954834,
+      "learning_rate": 9.421533167402534e-05,
+      "loss": 0.0292,
+      "step": 1970
+    },
+    {
+      "epoch": 2.8125,
+      "grad_norm": 0.42748284339904785,
+      "learning_rate": 9.413788845090666e-05,
+      "loss": 0.0272,
+      "step": 1980
+    },
+    {
+      "epoch": 2.8267045454545454,
+      "grad_norm": 0.5062595009803772,
+      "learning_rate": 9.405996254291136e-05,
+      "loss": 0.0275,
+      "step": 1990
+    },
+    {
+      "epoch": 2.840909090909091,
+      "grad_norm": 0.4755995273590088,
+      "learning_rate": 9.398155480222474e-05,
+      "loss": 0.0257,
+      "step": 2000
+    },
+    {
+      "epoch": 2.8551136363636362,
+      "grad_norm": 0.28334909677505493,
+      "learning_rate": 9.390266608630128e-05,
+      "loss": 0.0213,
+      "step": 2010
+    },
+    {
+      "epoch": 2.8693181818181817,
+      "grad_norm": 0.41236427426338196,
+      "learning_rate": 9.38232972578553e-05,
+      "loss": 0.0257,
+      "step": 2020
+    },
+    {
+      "epoch": 2.8835227272727275,
+      "grad_norm": 0.2744649648666382,
+      "learning_rate": 9.374344918485164e-05,
+      "loss": 0.0289,
+      "step": 2030
+    },
+    {
+      "epoch": 2.8977272727272725,
+      "grad_norm": 0.4775767922401428,
+      "learning_rate": 9.366312274049602e-05,
+      "loss": 0.0304,
+      "step": 2040
+    },
+    {
+      "epoch": 2.9119318181818183,
+      "grad_norm": 0.3825248181819916,
+      "learning_rate": 9.358231880322554e-05,
+      "loss": 0.0246,
+      "step": 2050
+    },
+    {
+      "epoch": 2.9261363636363638,
+      "grad_norm": 0.4156688153743744,
+      "learning_rate": 9.350103825669916e-05,
+      "loss": 0.0225,
+      "step": 2060
+    },
+    {
+      "epoch": 2.940340909090909,
+      "grad_norm": 0.4629719853401184,
+      "learning_rate": 9.341928198978787e-05,
+      "loss": 0.0202,
+      "step": 2070
+    },
+    {
+      "epoch": 2.9545454545454546,
+      "grad_norm": 0.6713044047355652,
+      "learning_rate": 9.333705089656512e-05,
+      "loss": 0.0261,
+      "step": 2080
+    },
+    {
+      "epoch": 2.96875,
+      "grad_norm": 0.3837173283100128,
+      "learning_rate": 9.325434587629698e-05,
+      "loss": 0.0233,
+      "step": 2090
+    },
+    {
+      "epoch": 2.9829545454545454,
+      "grad_norm": 0.2235601395368576,
+      "learning_rate": 9.31711678334323e-05,
+      "loss": 0.026,
+      "step": 2100
+    },
+    {
+      "epoch": 2.997159090909091,
+      "grad_norm": 0.49260881543159485,
+      "learning_rate": 9.308751767759282e-05,
+      "loss": 0.0258,
+      "step": 2110
+    },
+    {
+      "epoch": 3.0113636363636362,
+      "grad_norm": 0.5787939429283142,
+      "learning_rate": 9.300339632356325e-05,
+      "loss": 0.0334,
+      "step": 2120
+    },
+    {
+      "epoch": 3.0255681818181817,
+      "grad_norm": 0.2638683617115021,
+      "learning_rate": 9.291880469128124e-05,
+      "loss": 0.0252,
+      "step": 2130
+    },
+    {
+      "epoch": 3.039772727272727,
+      "grad_norm": 0.44293075799942017,
+      "learning_rate": 9.283374370582732e-05,
+      "loss": 0.0271,
+      "step": 2140
+    },
+    {
+      "epoch": 3.053977272727273,
+      "grad_norm": 0.3912266194820404,
+      "learning_rate": 9.274821429741482e-05,
+      "loss": 0.0317,
+      "step": 2150
+    },
+    {
+      "epoch": 3.0681818181818183,
+      "grad_norm": 0.28338822722435,
+      "learning_rate": 9.266221740137961e-05,
+      "loss": 0.0197,
+      "step": 2160
+    },
+    {
+      "epoch": 3.0823863636363638,
+      "grad_norm": 0.2771352529525757,
+      "learning_rate": 9.257575395817001e-05,
+      "loss": 0.0236,
+      "step": 2170
+    },
+    {
+      "epoch": 3.096590909090909,
+      "grad_norm": 0.3769958019256592,
+      "learning_rate": 9.248882491333637e-05,
+      "loss": 0.022,
+      "step": 2180
+    },
+    {
+      "epoch": 3.1107954545454546,
+      "grad_norm": 0.3424307107925415,
+      "learning_rate": 9.240143121752076e-05,
+      "loss": 0.0234,
+      "step": 2190
+    },
+    {
+      "epoch": 3.125,
+      "grad_norm": 0.31134355068206787,
+      "learning_rate": 9.23135738264467e-05,
+      "loss": 0.0231,
+      "step": 2200
+    },
+    {
+      "epoch": 3.1392045454545454,
+      "grad_norm": 0.565000593662262,
+      "learning_rate": 9.222525370090849e-05,
+      "loss": 0.0217,
+      "step": 2210
+    },
+    {
+      "epoch": 3.153409090909091,
+      "grad_norm": 0.33660510182380676,
+      "learning_rate": 9.213647180676088e-05,
+      "loss": 0.0192,
+      "step": 2220
+    },
+    {
+      "epoch": 3.1676136363636362,
+      "grad_norm": 0.22033120691776276,
+      "learning_rate": 9.204722911490846e-05,
+      "loss": 0.0225,
+      "step": 2230
+    },
+    {
+      "epoch": 3.1818181818181817,
+      "grad_norm": 0.31112611293792725,
+      "learning_rate": 9.1957526601295e-05,
+      "loss": 0.0244,
+      "step": 2240
+    },
+    {
+      "epoch": 3.196022727272727,
+      "grad_norm": 0.3010839521884918,
+      "learning_rate": 9.186736524689281e-05,
+      "loss": 0.0276,
+      "step": 2250
+    },
+    {
+      "epoch": 3.210227272727273,
+      "grad_norm": 0.6507784128189087,
+      "learning_rate": 9.177674603769204e-05,
+      "loss": 0.0243,
+      "step": 2260
+    },
+    {
+      "epoch": 3.2244318181818183,
+      "grad_norm": 0.2832013964653015,
+      "learning_rate": 9.168566996468983e-05,
+      "loss": 0.0257,
+      "step": 2270
+    },
+    {
+      "epoch": 3.2386363636363638,
+      "grad_norm": 0.5976662039756775,
+      "learning_rate": 9.159413802387951e-05,
+      "loss": 0.0261,
+      "step": 2280
+    },
+    {
+      "epoch": 3.252840909090909,
+      "grad_norm": 0.3606194257736206,
+      "learning_rate": 9.150215121623974e-05,
+      "loss": 0.0247,
+      "step": 2290
+    },
+    {
+      "epoch": 3.2670454545454546,
+      "grad_norm": 0.5259012579917908,
+      "learning_rate": 9.140971054772349e-05,
+      "loss": 0.024,
+      "step": 2300
+    },
+    {
+      "epoch": 3.28125,
+      "grad_norm": 0.36408892273902893,
+      "learning_rate": 9.131681702924713e-05,
+      "loss": 0.0267,
+      "step": 2310
+    },
+    {
+      "epoch": 3.2954545454545454,
+      "grad_norm": 0.45214271545410156,
+      "learning_rate": 9.122347167667926e-05,
+      "loss": 0.0246,
+      "step": 2320
+    },
+    {
+      "epoch": 3.309659090909091,
+      "grad_norm": 0.24400173127651215,
+      "learning_rate": 9.112967551082973e-05,
+      "loss": 0.0309,
+      "step": 2330
+    },
+    {
+      "epoch": 3.3238636363636362,
+      "grad_norm": 0.5899666547775269,
+      "learning_rate": 9.103542955743835e-05,
+      "loss": 0.0331,
+      "step": 2340
+    },
+    {
+      "epoch": 3.3380681818181817,
+      "grad_norm": 0.26844707131385803,
+      "learning_rate": 9.094073484716381e-05,
+      "loss": 0.033,
+      "step": 2350
+    },
+    {
+      "epoch": 3.3522727272727275,
+      "grad_norm": 0.34470316767692566,
+      "learning_rate": 9.084559241557226e-05,
+      "loss": 0.0298,
+      "step": 2360
+    },
+    {
+      "epoch": 3.3664772727272725,
+      "grad_norm": 0.591109573841095,
+      "learning_rate": 9.075000330312608e-05,
+      "loss": 0.0257,
+      "step": 2370
+    },
+    {
+      "epoch": 3.3806818181818183,
+      "grad_norm": 0.42912691831588745,
+      "learning_rate": 9.065396855517253e-05,
+      "loss": 0.0237,
+      "step": 2380
+    },
+    {
+      "epoch": 3.3948863636363638,
+      "grad_norm": 0.19727517664432526,
+      "learning_rate": 9.055748922193219e-05,
+      "loss": 0.025,
+      "step": 2390
+    },
+    {
+      "epoch": 3.409090909090909,
+      "grad_norm": 0.4568416476249695,
+      "learning_rate": 9.046056635848761e-05,
+      "loss": 0.026,
+      "step": 2400
+    },
+    {
+      "epoch": 3.4232954545454546,
+      "grad_norm": 0.4267689287662506,
+      "learning_rate": 9.036320102477169e-05,
+      "loss": 0.0221,
+      "step": 2410
+    },
+    {
+      "epoch": 3.4375,
+      "grad_norm": 0.42088618874549866,
+      "learning_rate": 9.02653942855561e-05,
+      "loss": 0.0281,
+      "step": 2420
+    },
+    {
+      "epoch": 3.4517045454545454,
+      "grad_norm": 0.5751941800117493,
+      "learning_rate": 9.016714721043971e-05,
+      "loss": 0.0227,
+      "step": 2430
+    },
+    {
+      "epoch": 3.465909090909091,
+      "grad_norm": 0.5421398878097534,
+      "learning_rate": 9.006846087383675e-05,
+      "loss": 0.0306,
+      "step": 2440
+    },
+    {
+      "epoch": 3.4801136363636362,
+      "grad_norm": 0.5087980628013611,
+      "learning_rate": 8.996933635496523e-05,
+      "loss": 0.0299,
+      "step": 2450
+    },
+    {
+      "epoch": 3.4943181818181817,
+      "grad_norm": 0.5215504765510559,
+      "learning_rate": 8.986977473783498e-05,
+      "loss": 0.0264,
+      "step": 2460
+    },
+    {
+      "epoch": 3.5085227272727275,
+      "grad_norm": 0.2630001902580261,
+      "learning_rate": 8.97697771112359e-05,
+      "loss": 0.0254,
+      "step": 2470
+    },
+    {
+      "epoch": 3.5227272727272725,
+      "grad_norm": 0.48159345984458923,
+      "learning_rate": 8.966934456872602e-05,
+      "loss": 0.0246,
+      "step": 2480
+    },
+    {
+      "epoch": 3.5369318181818183,
+      "grad_norm": 0.405266135931015,
+      "learning_rate": 8.95684782086195e-05,
+      "loss": 0.0237,
+      "step": 2490
+    },
+    {
+      "epoch": 3.5511363636363638,
+      "grad_norm": 0.26516810059547424,
+      "learning_rate": 8.946717913397476e-05,
+      "loss": 0.0235,
+      "step": 2500
+    },
+    {
+      "epoch": 3.565340909090909,
+      "grad_norm": 0.4076722264289856,
+      "learning_rate": 8.93654484525822e-05,
+      "loss": 0.0256,
+      "step": 2510
+    },
+    {
+      "epoch": 3.5795454545454546,
+      "grad_norm": 0.19961880147457123,
+      "learning_rate": 8.926328727695226e-05,
+      "loss": 0.0231,
+      "step": 2520
+    },
+    {
+      "epoch": 3.59375,
+      "grad_norm": 0.40052559971809387,
+      "learning_rate": 8.916069672430319e-05,
+      "loss": 0.0185,
+      "step": 2530
+    },
+    {
+      "epoch": 3.6079545454545454,
+      "grad_norm": 0.3906126320362091,
+      "learning_rate": 8.905767791654884e-05,
+      "loss": 0.0234,
+      "step": 2540
+    },
+    {
+      "epoch": 3.622159090909091,
+      "grad_norm": 0.3080589473247528,
+      "learning_rate": 8.895423198028638e-05,
+      "loss": 0.0262,
+      "step": 2550
+    },
+    {
+      "epoch": 3.6363636363636362,
+      "grad_norm": 0.30532577633857727,
+      "learning_rate": 8.885036004678402e-05,
+      "loss": 0.0218,
+      "step": 2560
+    },
+    {
+      "epoch": 3.6505681818181817,
+      "grad_norm": 0.32045280933380127,
+      "learning_rate": 8.874606325196857e-05,
+      "loss": 0.0229,
+      "step": 2570
+    },
+    {
+      "epoch": 3.6647727272727275,
+      "grad_norm": 0.3186279833316803,
+      "learning_rate": 8.864134273641304e-05,
+      "loss": 0.0247,
+      "step": 2580
+    },
+    {
+      "epoch": 3.6789772727272725,
+      "grad_norm": 0.3219265043735504,
+      "learning_rate": 8.853619964532427e-05,
+      "loss": 0.0239,
+      "step": 2590
+    },
+    {
+      "epoch": 3.6931818181818183,
+      "grad_norm": 0.290261834859848,
+      "learning_rate": 8.843063512853019e-05,
+      "loss": 0.0211,
+      "step": 2600
+    },
+    {
+      "epoch": 3.7073863636363638,
+      "grad_norm": 0.38125187158584595,
+      "learning_rate": 8.832465034046749e-05,
+      "loss": 0.0223,
+      "step": 2610
+    },
+    {
+      "epoch": 3.721590909090909,
+      "grad_norm": 0.3656066060066223,
+      "learning_rate": 8.821824644016882e-05,
+      "loss": 0.0237,
+      "step": 2620
+    },
+    {
+      "epoch": 3.7357954545454546,
+      "grad_norm": 0.4070371687412262,
+      "learning_rate": 8.811142459125019e-05,
+      "loss": 0.0207,
+      "step": 2630
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.18587815761566162,
+      "learning_rate": 8.800418596189822e-05,
+      "loss": 0.0192,
+      "step": 2640
+    },
+    {
+      "epoch": 3.7642045454545454,
+      "grad_norm": 0.49753904342651367,
+      "learning_rate": 8.789653172485737e-05,
+      "loss": 0.0227,
+      "step": 2650
+    },
+    {
+      "epoch": 3.778409090909091,
+      "grad_norm": 0.42766085267066956,
+      "learning_rate": 8.778846305741715e-05,
+      "loss": 0.0336,
+      "step": 2660
+    },
+    {
+      "epoch": 3.7926136363636362,
+      "grad_norm": 0.519346296787262,
+      "learning_rate": 8.767998114139918e-05,
+      "loss": 0.0253,
+      "step": 2670
+    },
+    {
+      "epoch": 3.8068181818181817,
+      "grad_norm": 0.46870407462120056,
+      "learning_rate": 8.757108716314429e-05,
+      "loss": 0.0254,
+      "step": 2680
+    },
+    {
+      "epoch": 3.8210227272727275,
+      "grad_norm": 0.46235018968582153,
+      "learning_rate": 8.746178231349962e-05,
+      "loss": 0.0196,
+      "step": 2690
+    },
+    {
+      "epoch": 3.8352272727272725,
+      "grad_norm": 0.5213091373443604,
+      "learning_rate": 8.735206778780549e-05,
+      "loss": 0.0266,
+      "step": 2700
+    },
+    {
+      "epoch": 3.8494318181818183,
+      "grad_norm": 0.2617625296115875,
+      "learning_rate": 8.724194478588234e-05,
+      "loss": 0.0243,
+      "step": 2710
+    },
+    {
+      "epoch": 3.8636363636363638,
+      "grad_norm": 0.2595122754573822,
+      "learning_rate": 8.713141451201772e-05,
+      "loss": 0.0234,
+      "step": 2720
+    },
+    {
+      "epoch": 3.877840909090909,
+      "grad_norm": 0.39617201685905457,
+      "learning_rate": 8.702047817495295e-05,
+      "loss": 0.0239,
+      "step": 2730
+    },
+    {
+      "epoch": 3.8920454545454546,
+      "grad_norm": 0.3868929147720337,
+      "learning_rate": 8.69091369878701e-05,
+      "loss": 0.0239,
+      "step": 2740
+    },
+    {
+      "epoch": 3.90625,
+      "grad_norm": 0.2817317247390747,
+      "learning_rate": 8.679739216837849e-05,
+      "loss": 0.0181,
+      "step": 2750
+    },
+    {
+      "epoch": 3.9204545454545454,
+      "grad_norm": 0.4082745313644409,
+      "learning_rate": 8.66852449385016e-05,
+      "loss": 0.0211,
+      "step": 2760
+    },
+    {
+      "epoch": 3.934659090909091,
+      "grad_norm": 0.3983289301395416,
+      "learning_rate": 8.657269652466356e-05,
+      "loss": 0.0198,
+      "step": 2770
+    },
+    {
+      "epoch": 3.9488636363636362,
+      "grad_norm": 0.34506163001060486,
+      "learning_rate": 8.645974815767577e-05,
+      "loss": 0.0236,
+      "step": 2780
+    },
+    {
+      "epoch": 3.9630681818181817,
+      "grad_norm": 0.31240084767341614,
+      "learning_rate": 8.634640107272351e-05,
+      "loss": 0.0188,
+      "step": 2790
+    },
+    {
+      "epoch": 3.9772727272727275,
+      "grad_norm": 0.32371658086776733,
+      "learning_rate": 8.623265650935234e-05,
+      "loss": 0.0235,
+      "step": 2800
+    },
+    {
+      "epoch": 3.9914772727272725,
+      "grad_norm": 0.3867642283439636,
+      "learning_rate": 8.611851571145456e-05,
+      "loss": 0.0219,
+      "step": 2810
+    },
+    {
+      "epoch": 4.005681818181818,
+      "grad_norm": 0.32164689898490906,
+      "learning_rate": 8.600397992725566e-05,
+      "loss": 0.0156,
+      "step": 2820
+    },
+    {
+      "epoch": 4.019886363636363,
+      "grad_norm": 0.24819204211235046,
+      "learning_rate": 8.588905040930061e-05,
+      "loss": 0.0222,
+      "step": 2830
+    },
+    {
+      "epoch": 4.034090909090909,
+      "grad_norm": 0.4445822536945343,
+      "learning_rate": 8.577372841444022e-05,
+      "loss": 0.0255,
+      "step": 2840
+    },
+    {
+      "epoch": 4.048295454545454,
+      "grad_norm": 0.25916674733161926,
+      "learning_rate": 8.565801520381736e-05,
+      "loss": 0.0189,
+      "step": 2850
+    },
+    {
+      "epoch": 4.0625,
+      "grad_norm": 0.46001407504081726,
+      "learning_rate": 8.554191204285313e-05,
+      "loss": 0.0242,
+      "step": 2860
+    },
+    {
+      "epoch": 4.076704545454546,
+      "grad_norm": 0.3041122555732727,
+      "learning_rate": 8.542542020123315e-05,
+      "loss": 0.0208,
+      "step": 2870
+    },
+    {
+      "epoch": 4.090909090909091,
+      "grad_norm": 0.2924349904060364,
+      "learning_rate": 8.530854095289347e-05,
+      "loss": 0.0196,
+      "step": 2880
+    },
+    {
+      "epoch": 4.105113636363637,
+      "grad_norm": 0.49790266156196594,
+      "learning_rate": 8.519127557600688e-05,
+      "loss": 0.0222,
+      "step": 2890
+    },
+    {
+      "epoch": 4.119318181818182,
+      "grad_norm": 0.3667716681957245,
+      "learning_rate": 8.507362535296871e-05,
+      "loss": 0.0225,
+      "step": 2900
+    },
+    {
+      "epoch": 4.1335227272727275,
+      "grad_norm": 0.2710287272930145,
+      "learning_rate": 8.495559157038299e-05,
+      "loss": 0.0293,
+      "step": 2910
+    },
+    {
+      "epoch": 4.1477272727272725,
+      "grad_norm": 0.3627186119556427,
+      "learning_rate": 8.483717551904823e-05,
+      "loss": 0.0233,
+      "step": 2920
+    },
+    {
+      "epoch": 4.161931818181818,
+      "grad_norm": 0.32987794280052185,
+      "learning_rate": 8.47183784939434e-05,
+      "loss": 0.0238,
+      "step": 2930
+    },
+    {
+      "epoch": 4.176136363636363,
+      "grad_norm": 0.2976704239845276,
+      "learning_rate": 8.459920179421374e-05,
+      "loss": 0.0216,
+      "step": 2940
+    },
+    {
+      "epoch": 4.190340909090909,
+      "grad_norm": 0.4660015106201172,
+      "learning_rate": 8.447964672315656e-05,
+      "loss": 0.0253,
+      "step": 2950
+    },
+    {
+      "epoch": 4.204545454545454,
+      "grad_norm": 0.3315091133117676,
+      "learning_rate": 8.435971458820692e-05,
+      "loss": 0.0239,
+      "step": 2960
+    },
+    {
+      "epoch": 4.21875,
+      "grad_norm": 0.26883944869041443,
+      "learning_rate": 8.423940670092345e-05,
+      "loss": 0.0209,
+      "step": 2970
+    },
+    {
+      "epoch": 4.232954545454546,
+      "grad_norm": 0.3321470618247986,
+      "learning_rate": 8.411872437697394e-05,
+      "loss": 0.0208,
+      "step": 2980
+    },
+    {
+      "epoch": 4.247159090909091,
+      "grad_norm": 0.3498826324939728,
+      "learning_rate": 8.399766893612096e-05,
+      "loss": 0.0184,
+      "step": 2990
+    },
+    {
+      "epoch": 4.261363636363637,
+      "grad_norm": 0.28569039702415466,
+      "learning_rate": 8.38762417022074e-05,
+      "loss": 0.0204,
+      "step": 3000
+    },
+    {
+      "epoch": 4.275568181818182,
+      "grad_norm": 0.20908929407596588,
+      "learning_rate": 8.375444400314204e-05,
+      "loss": 0.0205,
+      "step": 3010
+    },
+    {
+      "epoch": 4.2897727272727275,
+      "grad_norm": 0.3200138211250305,
+      "learning_rate": 8.3632277170885e-05,
+      "loss": 0.0269,
+      "step": 3020
+    },
+    {
+      "epoch": 4.3039772727272725,
+      "grad_norm": 0.2988082766532898,
+      "learning_rate": 8.350974254143318e-05,
+      "loss": 0.0217,
+      "step": 3030
+    },
+    {
+      "epoch": 4.318181818181818,
+      "grad_norm": 0.3487236499786377,
+      "learning_rate": 8.338684145480566e-05,
+      "loss": 0.019,
+      "step": 3040
+    },
+    {
+      "epoch": 4.332386363636363,
+      "grad_norm": 0.24754469096660614,
+      "learning_rate": 8.326357525502904e-05,
+      "loss": 0.0185,
+      "step": 3050
+    },
+    {
+      "epoch": 4.346590909090909,
+      "grad_norm": 0.48998352885246277,
+      "learning_rate": 8.313994529012273e-05,
+      "loss": 0.0268,
+      "step": 3060
+    },
+    {
+      "epoch": 4.360795454545454,
+      "grad_norm": 0.2897888720035553,
+      "learning_rate": 8.301595291208422e-05,
+      "loss": 0.0246,
+      "step": 3070
+    },
+    {
+      "epoch": 4.375,
+      "grad_norm": 0.30872222781181335,
+      "learning_rate": 8.289159947687427e-05,
+      "loss": 0.0196,
+      "step": 3080
+    },
+    {
+      "epoch": 4.389204545454546,
+      "grad_norm": 0.34679895639419556,
+      "learning_rate": 8.276688634440216e-05,
+      "loss": 0.023,
+      "step": 3090
+    },
+    {
+      "epoch": 4.403409090909091,
+      "grad_norm": 0.42075955867767334,
+      "learning_rate": 8.26418148785107e-05,
+      "loss": 0.0235,
+      "step": 3100
+    },
+    {
+      "epoch": 4.417613636363637,
+      "grad_norm": 0.3550330102443695,
+      "learning_rate": 8.251638644696141e-05,
+      "loss": 0.0278,
+      "step": 3110
+    },
+    {
+      "epoch": 4.431818181818182,
+      "grad_norm": 0.35106098651885986,
+      "learning_rate": 8.23906024214195e-05,
+      "loss": 0.0288,
+      "step": 3120
+    },
+    {
+      "epoch": 4.4460227272727275,
+      "grad_norm": 0.4596957564353943,
+      "learning_rate": 8.226446417743897e-05,
+      "loss": 0.0232,
+      "step": 3130
+    },
+    {
+      "epoch": 4.4602272727272725,
+      "grad_norm": 0.1977410465478897,
+      "learning_rate": 8.213797309444742e-05,
+      "loss": 0.0242,
+      "step": 3140
+    },
+    {
+      "epoch": 4.474431818181818,
+      "grad_norm": 0.27501070499420166,
+      "learning_rate": 8.201113055573105e-05,
+      "loss": 0.0226,
+      "step": 3150
+    },
+    {
+      "epoch": 4.488636363636363,
+      "grad_norm": 0.3810543119907379,
+      "learning_rate": 8.188393794841958e-05,
+      "loss": 0.0213,
+      "step": 3160
+    },
+    {
+      "epoch": 4.502840909090909,
+      "grad_norm": 0.36581721901893616,
+      "learning_rate": 8.175639666347094e-05,
+      "loss": 0.027,
+      "step": 3170
+    },
+    {
+      "epoch": 4.517045454545455,
+      "grad_norm": 0.33507320284843445,
+      "learning_rate": 8.162850809565623e-05,
+      "loss": 0.0227,
+      "step": 3180
+    },
+    {
+      "epoch": 4.53125,
+      "grad_norm": 0.3294641077518463,
+      "learning_rate": 8.150027364354431e-05,
+      "loss": 0.0199,
+      "step": 3190
+    },
+    {
+      "epoch": 4.545454545454545,
+      "grad_norm": 0.48345720767974854,
+      "learning_rate": 8.137169470948662e-05,
+      "loss": 0.0272,
+      "step": 3200
+    },
+    {
+      "epoch": 4.559659090909091,
+      "grad_norm": 0.4137798845767975,
+      "learning_rate": 8.124277269960179e-05,
+      "loss": 0.0272,
+      "step": 3210
+    },
+    {
+      "epoch": 4.573863636363637,
+      "grad_norm": 0.4345813989639282,
+      "learning_rate": 8.111350902376023e-05,
+      "loss": 0.0229,
+      "step": 3220
+    },
+    {
+      "epoch": 4.588068181818182,
+      "grad_norm": 0.36524698138237,
+      "learning_rate": 8.098390509556883e-05,
+      "loss": 0.0243,
+      "step": 3230
+    },
+    {
+      "epoch": 4.6022727272727275,
+      "grad_norm": 0.3373654782772064,
+      "learning_rate": 8.085396233235536e-05,
+      "loss": 0.0186,
+      "step": 3240
+    },
+    {
+      "epoch": 4.6164772727272725,
+      "grad_norm": 0.3737955093383789,
+      "learning_rate": 8.072368215515306e-05,
+      "loss": 0.0175,
+      "step": 3250
+    },
+    {
+      "epoch": 4.630681818181818,
+      "grad_norm": 0.26559293270111084,
+      "learning_rate": 8.059306598868506e-05,
+      "loss": 0.0282,
+      "step": 3260
+    },
+    {
+      "epoch": 4.644886363636363,
+      "grad_norm": 0.19148071110248566,
+      "learning_rate": 8.046211526134888e-05,
+      "loss": 0.0173,
+      "step": 3270
+    },
+    {
+      "epoch": 4.659090909090909,
+      "grad_norm": 0.28995585441589355,
+      "learning_rate": 8.033083140520065e-05,
+      "loss": 0.0215,
+      "step": 3280
+    },
+    {
+      "epoch": 4.673295454545455,
+      "grad_norm": 0.32952094078063965,
+      "learning_rate": 8.019921585593962e-05,
+      "loss": 0.0201,
+      "step": 3290
+    },
+    {
+      "epoch": 4.6875,
+      "grad_norm": 0.24955284595489502,
+      "learning_rate": 8.006727005289232e-05,
+      "loss": 0.0174,
+      "step": 3300
+    },
+    {
+      "epoch": 4.701704545454545,
+      "grad_norm": 0.42458289861679077,
+      "learning_rate": 7.993499543899692e-05,
+      "loss": 0.0208,
+      "step": 3310
+    },
+    {
+      "epoch": 4.715909090909091,
+      "grad_norm": 0.30404138565063477,
+      "learning_rate": 7.980239346078742e-05,
+      "loss": 0.0207,
+      "step": 3320
+    },
+    {
+      "epoch": 4.730113636363637,
+      "grad_norm": 0.271896094083786,
+      "learning_rate": 7.966946556837778e-05,
+      "loss": 0.02,
+      "step": 3330
+    },
+    {
+      "epoch": 4.744318181818182,
+      "grad_norm": 0.3281709551811218,
+      "learning_rate": 7.953621321544616e-05,
+      "loss": 0.0232,
+      "step": 3340
+    },
+    {
+      "epoch": 4.7585227272727275,
+      "grad_norm": 0.2940441370010376,
+      "learning_rate": 7.940263785921896e-05,
+      "loss": 0.0216,
+      "step": 3350
+    },
+    {
+      "epoch": 4.7727272727272725,
+      "grad_norm": 0.27014073729515076,
+      "learning_rate": 7.926874096045482e-05,
+      "loss": 0.0239,
+      "step": 3360
+    },
+    {
+      "epoch": 4.786931818181818,
+      "grad_norm": 0.37260785698890686,
+      "learning_rate": 7.913452398342881e-05,
+      "loss": 0.0259,
+      "step": 3370
+    },
+    {
+      "epoch": 4.801136363636363,
+      "grad_norm": 0.2880043685436249,
+      "learning_rate": 7.89999883959163e-05,
+      "loss": 0.0242,
+      "step": 3380
+    },
+    {
+      "epoch": 4.815340909090909,
+      "grad_norm": 0.4994223117828369,
+      "learning_rate": 7.886513566917687e-05,
+      "loss": 0.0246,
+      "step": 3390
+    },
+    {
+      "epoch": 4.829545454545455,
+      "grad_norm": 0.30457958579063416,
+      "learning_rate": 7.872996727793838e-05,
+      "loss": 0.0246,
+      "step": 3400
+    },
+    {
+      "epoch": 4.84375,
+      "grad_norm": 0.8592410683631897,
+      "learning_rate": 7.859448470038069e-05,
+      "loss": 0.0274,
+      "step": 3410
+    },
+    {
+      "epoch": 4.857954545454545,
+      "grad_norm": 0.2518974840641022,
+      "learning_rate": 7.845868941811956e-05,
+      "loss": 0.0241,
+      "step": 3420
+    },
+    {
+      "epoch": 4.872159090909091,
+      "grad_norm": 0.3433271050453186,
+      "learning_rate": 7.832258291619043e-05,
+      "loss": 0.0215,
+      "step": 3430
+    },
+    {
+      "epoch": 4.886363636363637,
+      "grad_norm": 0.300483375787735,
+      "learning_rate": 7.81861666830322e-05,
+      "loss": 0.0211,
+      "step": 3440
+    },
+    {
+      "epoch": 4.900568181818182,
+      "grad_norm": 0.5095469951629639,
+      "learning_rate": 7.804944221047097e-05,
+      "loss": 0.0253,
+      "step": 3450
+    },
+    {
+      "epoch": 4.9147727272727275,
+      "grad_norm": 0.32967308163642883,
+      "learning_rate": 7.791241099370364e-05,
+      "loss": 0.0255,
+      "step": 3460
+    },
+    {
+      "epoch": 4.9289772727272725,
+      "grad_norm": 0.38610175251960754,
+      "learning_rate": 7.777507453128163e-05,
+      "loss": 0.0207,
+      "step": 3470
+    },
+    {
+      "epoch": 4.943181818181818,
+      "grad_norm": 0.29390275478363037,
+      "learning_rate": 7.763743432509451e-05,
+      "loss": 0.0192,
+      "step": 3480
+    },
+    {
+      "epoch": 4.957386363636363,
+      "grad_norm": 0.2913910150527954,
+      "learning_rate": 7.749949188035353e-05,
+      "loss": 0.0227,
+      "step": 3490
+    },
+    {
+      "epoch": 4.971590909090909,
+      "grad_norm": 0.27363091707229614,
+      "learning_rate": 7.736124870557516e-05,
+      "loss": 0.0242,
+      "step": 3500
+    },
+    {
+      "epoch": 4.985795454545455,
+      "grad_norm": 0.1807710975408554,
+      "learning_rate": 7.722270631256459e-05,
+      "loss": 0.0211,
+      "step": 3510
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.42457127571105957,
+      "learning_rate": 7.708386621639925e-05,
+      "loss": 0.0204,
+      "step": 3520
+    },
+    {
+      "epoch": 5.014204545454546,
+      "grad_norm": 0.27891087532043457,
+      "learning_rate": 7.694472993541219e-05,
+      "loss": 0.024,
+      "step": 3530
+    },
+    {
+      "epoch": 5.028409090909091,
+      "grad_norm": 0.29139384627342224,
+      "learning_rate": 7.680529899117547e-05,
+      "loss": 0.0228,
+      "step": 3540
+    },
+    {
+      "epoch": 5.042613636363637,
+      "grad_norm": 0.30129513144493103,
+      "learning_rate": 7.666557490848358e-05,
+      "loss": 0.0234,
+      "step": 3550
+    },
+    {
+      "epoch": 5.056818181818182,
+      "grad_norm": 0.29616716504096985,
+      "learning_rate": 7.65255592153367e-05,
+      "loss": 0.0213,
+      "step": 3560
+    },
+    {
+      "epoch": 5.0710227272727275,
+      "grad_norm": 0.39174526929855347,
+      "learning_rate": 7.638525344292402e-05,
+      "loss": 0.0232,
+      "step": 3570
+    },
+    {
+      "epoch": 5.0852272727272725,
+      "grad_norm": 0.22232000529766083,
+      "learning_rate": 7.624465912560697e-05,
+      "loss": 0.0199,
+      "step": 3580
+    },
+    {
+      "epoch": 5.099431818181818,
+      "grad_norm": 0.3431629538536072,
+      "learning_rate": 7.610377780090249e-05,
+      "loss": 0.0237,
+      "step": 3590
+    },
+    {
+      "epoch": 5.113636363636363,
+      "grad_norm": 0.24721530079841614,
+      "learning_rate": 7.596261100946618e-05,
+      "loss": 0.0201,
+      "step": 3600
+    },
+    {
+      "epoch": 5.127840909090909,
+      "grad_norm": 0.16461999714374542,
+      "learning_rate": 7.582116029507542e-05,
+      "loss": 0.0194,
+      "step": 3610
+    },
+    {
+      "epoch": 5.142045454545454,
+      "grad_norm": 0.42581430077552795,
+      "learning_rate": 7.56794272046126e-05,
+      "loss": 0.0169,
+      "step": 3620
+    },
+    {
+      "epoch": 5.15625,
+      "grad_norm": 0.39945194125175476,
+      "learning_rate": 7.55374132880481e-05,
+      "loss": 0.022,
+      "step": 3630
+    },
+    {
+      "epoch": 5.170454545454546,
+      "grad_norm": 0.21798691153526306,
+      "learning_rate": 7.539512009842333e-05,
+      "loss": 0.022,
+      "step": 3640
+    },
+    {
+      "epoch": 5.184659090909091,
+      "grad_norm": 0.3774815797805786,
+      "learning_rate": 7.525254919183382e-05,
+      "loss": 0.0214,
+      "step": 3650
+    },
+    {
+      "epoch": 5.198863636363637,
+      "grad_norm": 0.37779033184051514,
+      "learning_rate": 7.510970212741215e-05,
+      "loss": 0.0179,
+      "step": 3660
+    },
+    {
+      "epoch": 5.213068181818182,
+      "grad_norm": 0.444468230009079,
+      "learning_rate": 7.496658046731096e-05,
+      "loss": 0.0226,
+      "step": 3670
+    },
+    {
+      "epoch": 5.2272727272727275,
+      "grad_norm": 0.3628615140914917,
+      "learning_rate": 7.482318577668578e-05,
+      "loss": 0.0231,
+      "step": 3680
+    },
+    {
+      "epoch": 5.2414772727272725,
+      "grad_norm": 0.21392694115638733,
+      "learning_rate": 7.467951962367796e-05,
+      "loss": 0.0221,
+      "step": 3690
+    },
+    {
+      "epoch": 5.255681818181818,
+      "grad_norm": 0.2890244126319885,
+      "learning_rate": 7.453558357939755e-05,
+      "loss": 0.0201,
+      "step": 3700
+    },
+    {
+      "epoch": 5.269886363636363,
+      "grad_norm": 0.2643570303916931,
+      "learning_rate": 7.439137921790606e-05,
+      "loss": 0.0204,
+      "step": 3710
+    },
+    {
+      "epoch": 5.284090909090909,
+      "grad_norm": 0.27545976638793945,
+      "learning_rate": 7.42469081161993e-05,
+      "loss": 0.0214,
+      "step": 3720
+    },
+    {
+      "epoch": 5.298295454545454,
+      "grad_norm": 0.2684711515903473,
+      "learning_rate": 7.410217185419006e-05,
+      "loss": 0.015,
+      "step": 3730
+    },
+    {
+      "epoch": 5.3125,
+      "grad_norm": 0.30951064825057983,
+      "learning_rate": 7.395717201469095e-05,
+      "loss": 0.0171,
+      "step": 3740
+    },
+    {
+      "epoch": 5.326704545454546,
+      "grad_norm": 0.34822556376457214,
+      "learning_rate": 7.381191018339696e-05,
+      "loss": 0.0198,
+      "step": 3750
+    },
+    {
+      "epoch": 5.340909090909091,
+      "grad_norm": 0.2723725140094757,
+      "learning_rate": 7.36663879488682e-05,
+      "loss": 0.0172,
+      "step": 3760
+    },
+    {
+      "epoch": 5.355113636363637,
+      "grad_norm": 0.2919902205467224,
+      "learning_rate": 7.352060690251254e-05,
+      "loss": 0.0261,
+      "step": 3770
+    },
+    {
+      "epoch": 5.369318181818182,
+      "grad_norm": 0.37643224000930786,
+      "learning_rate": 7.337456863856811e-05,
+      "loss": 0.0213,
+      "step": 3780
+    },
+    {
+      "epoch": 5.3835227272727275,
+      "grad_norm": 0.47135400772094727,
+      "learning_rate": 7.3228274754086e-05,
+      "loss": 0.0194,
+      "step": 3790
+    },
+    {
+      "epoch": 5.3977272727272725,
+      "grad_norm": 0.1968531459569931,
+      "learning_rate": 7.308172684891267e-05,
+      "loss": 0.0219,
+      "step": 3800
+    },
+    {
+      "epoch": 5.411931818181818,
+      "grad_norm": 0.44277092814445496,
+      "learning_rate": 7.293492652567255e-05,
+      "loss": 0.0206,
+      "step": 3810
+    },
+    {
+      "epoch": 5.426136363636363,
+      "grad_norm": 0.26302438974380493,
+      "learning_rate": 7.278787538975043e-05,
+      "loss": 0.0198,
+      "step": 3820
+    },
+    {
+      "epoch": 5.440340909090909,
+      "grad_norm": 0.3885079026222229,
+      "learning_rate": 7.2640575049274e-05,
+      "loss": 0.0225,
+      "step": 3830
+    },
+    {
+      "epoch": 5.454545454545454,
+      "grad_norm": 0.3239285349845886,
+      "learning_rate": 7.249302711509616e-05,
+      "loss": 0.0156,
+      "step": 3840
+    },
+    {
+      "epoch": 5.46875,
+      "grad_norm": 0.34747716784477234,
+      "learning_rate": 7.23452332007775e-05,
+      "loss": 0.017,
+      "step": 3850
+    },
+    {
+      "epoch": 5.482954545454546,
+      "grad_norm": 0.480849027633667,
+      "learning_rate": 7.219719492256858e-05,
+      "loss": 0.0181,
+      "step": 3860
+    },
+    {
+      "epoch": 5.497159090909091,
+      "grad_norm": 0.33865511417388916,
+      "learning_rate": 7.20489138993923e-05,
+      "loss": 0.0198,
+      "step": 3870
+    },
+    {
+      "epoch": 5.511363636363637,
+      "grad_norm": 0.2188992202281952,
+      "learning_rate": 7.190039175282614e-05,
+      "loss": 0.0214,
+      "step": 3880
+    },
+    {
+      "epoch": 5.525568181818182,
+      "grad_norm": 0.3381461799144745,
+      "learning_rate": 7.175163010708455e-05,
+      "loss": 0.0172,
+      "step": 3890
+    },
+    {
+      "epoch": 5.5397727272727275,
+      "grad_norm": 0.37526050209999084,
+      "learning_rate": 7.1602630589001e-05,
+      "loss": 0.0247,
+      "step": 3900
+    },
+    {
+      "epoch": 5.5539772727272725,
+      "grad_norm": 0.21504007279872894,
+      "learning_rate": 7.14533948280104e-05,
+      "loss": 0.0184,
+      "step": 3910
+    },
+    {
+      "epoch": 5.568181818181818,
+      "grad_norm": 0.2110414057970047,
+      "learning_rate": 7.130392445613109e-05,
+      "loss": 0.0224,
+      "step": 3920
+    },
+    {
+      "epoch": 5.582386363636363,
+      "grad_norm": 0.2872629761695862,
+      "learning_rate": 7.115422110794711e-05,
+      "loss": 0.0189,
+      "step": 3930
+    },
+    {
+      "epoch": 5.596590909090909,
+      "grad_norm": 0.345014750957489,
+      "learning_rate": 7.100428642059033e-05,
+      "loss": 0.0201,
+      "step": 3940
+    },
+    {
+      "epoch": 5.610795454545455,
+      "grad_norm": 0.3859497308731079,
+      "learning_rate": 7.08541220337224e-05,
+      "loss": 0.0216,
+      "step": 3950
+    },
+    {
+      "epoch": 5.625,
+      "grad_norm": 0.45760706067085266,
+      "learning_rate": 7.070372958951706e-05,
+      "loss": 0.0221,
+      "step": 3960
+    },
+    {
+      "epoch": 5.639204545454545,
+      "grad_norm": 0.2422349601984024,
+      "learning_rate": 7.055311073264194e-05,
+      "loss": 0.0196,
+      "step": 3970
+    },
+    {
+      "epoch": 5.653409090909091,
+      "grad_norm": 0.21133822202682495,
+      "learning_rate": 7.040226711024077e-05,
+      "loss": 0.021,
+      "step": 3980
+    },
+    {
+      "epoch": 5.667613636363637,
+      "grad_norm": 0.23644445836544037,
+      "learning_rate": 7.02512003719152e-05,
+      "loss": 0.0196,
+      "step": 3990
+    },
+    {
+      "epoch": 5.681818181818182,
+      "grad_norm": 0.25289401412010193,
+      "learning_rate": 7.00999121697069e-05,
+      "loss": 0.0172,
+      "step": 4000
+    },
+    {
+      "epoch": 5.6960227272727275,
+      "grad_norm": 0.24326521158218384,
+      "learning_rate": 6.99484041580794e-05,
+      "loss": 0.0203,
+      "step": 4010
+    },
+    {
+      "epoch": 5.7102272727272725,
+      "grad_norm": 0.309621661901474,
+      "learning_rate": 6.979667799390004e-05,
+      "loss": 0.0188,
+      "step": 4020
+    },
+    {
+      "epoch": 5.724431818181818,
+      "grad_norm": 0.3210173547267914,
+      "learning_rate": 6.964473533642185e-05,
+      "loss": 0.0225,
+      "step": 4030
+    },
+    {
+      "epoch": 5.738636363636363,
+      "grad_norm": 0.4055038094520569,
+      "learning_rate": 6.949257784726539e-05,
+      "loss": 0.0219,
+      "step": 4040
+    },
+    {
+      "epoch": 5.752840909090909,
+      "grad_norm": 0.29915717244148254,
+      "learning_rate": 6.934020719040056e-05,
+      "loss": 0.0191,
+      "step": 4050
+    },
+    {
+      "epoch": 5.767045454545455,
+      "grad_norm": 0.2618870735168457,
+      "learning_rate": 6.918762503212848e-05,
+      "loss": 0.0228,
+      "step": 4060
+    },
+    {
+      "epoch": 5.78125,
+      "grad_norm": 0.5184383988380432,
+      "learning_rate": 6.903483304106319e-05,
+      "loss": 0.0223,
+      "step": 4070
+    },
+    {
+      "epoch": 5.795454545454545,
+      "grad_norm": 0.30066415667533875,
+      "learning_rate": 6.888183288811341e-05,
+      "loss": 0.0178,
+      "step": 4080
+    },
+    {
+      "epoch": 5.809659090909091,
+      "grad_norm": 0.2669849097728729,
+      "learning_rate": 6.87286262464643e-05,
+      "loss": 0.0216,
+      "step": 4090
+    },
+    {
+      "epoch": 5.823863636363637,
+      "grad_norm": 0.31216955184936523,
+      "learning_rate": 6.857521479155915e-05,
+      "loss": 0.0193,
+      "step": 4100
+    },
+    {
+      "epoch": 5.838068181818182,
+      "grad_norm": 0.20216317474842072,
+      "learning_rate": 6.842160020108104e-05,
+      "loss": 0.0182,
+      "step": 4110
+    },
+    {
+      "epoch": 5.8522727272727275,
+      "grad_norm": 0.22763049602508545,
+      "learning_rate": 6.826778415493455e-05,
+      "loss": 0.0224,
+      "step": 4120
+    },
+    {
+      "epoch": 5.8664772727272725,
+      "grad_norm": 0.28869321942329407,
+      "learning_rate": 6.811376833522729e-05,
+      "loss": 0.0198,
+      "step": 4130
+    },
+    {
+      "epoch": 5.880681818181818,
+      "grad_norm": 0.2887463867664337,
+      "learning_rate": 6.795955442625159e-05,
+      "loss": 0.0163,
+      "step": 4140
+    },
+    {
+      "epoch": 5.894886363636363,
+      "grad_norm": 0.21988768875598907,
+      "learning_rate": 6.780514411446608e-05,
+      "loss": 0.0177,
+      "step": 4150
+    },
+    {
+      "epoch": 5.909090909090909,
+      "grad_norm": 0.7256492972373962,
+      "learning_rate": 6.765053908847716e-05,
+      "loss": 0.0195,
+      "step": 4160
+    },
+    {
+      "epoch": 5.923295454545455,
+      "grad_norm": 0.34503793716430664,
+      "learning_rate": 6.749574103902064e-05,
+      "loss": 0.0184,
+      "step": 4170
+    },
+    {
+      "epoch": 5.9375,
+      "grad_norm": 0.28087738156318665,
+      "learning_rate": 6.734075165894317e-05,
+      "loss": 0.0224,
+      "step": 4180
+    },
+    {
+      "epoch": 5.951704545454545,
+      "grad_norm": 0.29598933458328247,
+      "learning_rate": 6.71855726431838e-05,
+      "loss": 0.0241,
+      "step": 4190
+    },
+    {
+      "epoch": 5.965909090909091,
+      "grad_norm": 0.2648463249206543,
+      "learning_rate": 6.703020568875538e-05,
+      "loss": 0.0215,
+      "step": 4200
+    },
+    {
+      "epoch": 5.980113636363637,
+      "grad_norm": 0.49615687131881714,
+      "learning_rate": 6.687465249472603e-05,
+      "loss": 0.019,
+      "step": 4210
+    },
+    {
+      "epoch": 5.994318181818182,
+      "grad_norm": 0.33503979444503784,
+      "learning_rate": 6.671891476220055e-05,
+      "loss": 0.0173,
+      "step": 4220
+    },
+    {
+      "epoch": 6.0085227272727275,
+      "grad_norm": 0.30252575874328613,
+      "learning_rate": 6.656299419430183e-05,
+      "loss": 0.0165,
+      "step": 4230
+    },
+    {
+      "epoch": 6.0227272727272725,
+      "grad_norm": 0.32734978199005127,
+      "learning_rate": 6.640689249615223e-05,
+      "loss": 0.0172,
+      "step": 4240
+    },
+    {
+      "epoch": 6.036931818181818,
+      "grad_norm": 0.2168329656124115,
+      "learning_rate": 6.625061137485491e-05,
+      "loss": 0.023,
+      "step": 4250
+    },
+    {
+      "epoch": 6.051136363636363,
+      "grad_norm": 0.3850686550140381,
+      "learning_rate": 6.609415253947517e-05,
+      "loss": 0.0214,
+      "step": 4260
+    },
+    {
+      "epoch": 6.065340909090909,
+      "grad_norm": 0.2830491364002228,
+      "learning_rate": 6.593751770102178e-05,
+      "loss": 0.0188,
+      "step": 4270
+    },
+    {
+      "epoch": 6.079545454545454,
+      "grad_norm": 0.3510623276233673,
+      "learning_rate": 6.578070857242823e-05,
+      "loss": 0.0166,
+      "step": 4280
+    },
+    {
+      "epoch": 6.09375,
+      "grad_norm": 0.19439226388931274,
+      "learning_rate": 6.562372686853402e-05,
+      "loss": 0.0163,
+      "step": 4290
+    },
+    {
+      "epoch": 6.107954545454546,
+      "grad_norm": 0.37282636761665344,
+      "learning_rate": 6.546657430606593e-05,
+      "loss": 0.0199,
+      "step": 4300
+    },
+    {
+      "epoch": 6.122159090909091,
+      "grad_norm": 0.2543478012084961,
+      "learning_rate": 6.530925260361918e-05,
+      "loss": 0.0167,
+      "step": 4310
+    },
+    {
+      "epoch": 6.136363636363637,
+      "grad_norm": 0.34718626737594604,
+      "learning_rate": 6.515176348163871e-05,
+      "loss": 0.0196,
+      "step": 4320
+    },
+    {
+      "epoch": 6.150568181818182,
+      "grad_norm": 0.3631095588207245,
+      "learning_rate": 6.499410866240032e-05,
+      "loss": 0.0162,
+      "step": 4330
+    },
+    {
+      "epoch": 6.1647727272727275,
+      "grad_norm": 0.19689616560935974,
+      "learning_rate": 6.48362898699919e-05,
+      "loss": 0.017,
+      "step": 4340
+    },
+    {
+      "epoch": 6.1789772727272725,
+      "grad_norm": 0.5665863752365112,
+      "learning_rate": 6.467830883029443e-05,
+      "loss": 0.0155,
+      "step": 4350
+    },
+    {
+      "epoch": 6.193181818181818,
+      "grad_norm": 0.23562046885490417,
+      "learning_rate": 6.452016727096326e-05,
+      "loss": 0.0166,
+      "step": 4360
+    },
+    {
+      "epoch": 6.207386363636363,
+      "grad_norm": 0.35843878984451294,
+      "learning_rate": 6.436186692140916e-05,
+      "loss": 0.0182,
+      "step": 4370
+    },
+    {
+      "epoch": 6.221590909090909,
+      "grad_norm": 0.19739370048046112,
+      "learning_rate": 6.420340951277938e-05,
+      "loss": 0.014,
+      "step": 4380
+    },
+    {
+      "epoch": 6.235795454545454,
+      "grad_norm": 0.37820932269096375,
+      "learning_rate": 6.404479677793874e-05,
+      "loss": 0.0164,
+      "step": 4390
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.25936558842658997,
+      "learning_rate": 6.388603045145075e-05,
+      "loss": 0.0187,
+      "step": 4400
+    },
+    {
+      "epoch": 6.264204545454546,
+      "grad_norm": 0.31805190443992615,
+      "learning_rate": 6.372711226955843e-05,
+      "loss": 0.0205,
+      "step": 4410
+    },
+    {
+      "epoch": 6.278409090909091,
+      "grad_norm": 0.28777316212654114,
+      "learning_rate": 6.356804397016564e-05,
+      "loss": 0.0176,
+      "step": 4420
+    },
+    {
+      "epoch": 6.292613636363637,
+      "grad_norm": 0.38546669483184814,
+      "learning_rate": 6.340882729281779e-05,
+      "loss": 0.0183,
+      "step": 4430
+    },
+    {
+      "epoch": 6.306818181818182,
+      "grad_norm": 0.34228515625,
+      "learning_rate": 6.324946397868294e-05,
+      "loss": 0.0211,
+      "step": 4440
+    },
+    {
+      "epoch": 6.3210227272727275,
+      "grad_norm": 0.24924390017986298,
+      "learning_rate": 6.308995577053276e-05,
+      "loss": 0.0234,
+      "step": 4450
+    },
+    {
+      "epoch": 6.3352272727272725,
+      "grad_norm": 0.22652466595172882,
+      "learning_rate": 6.293030441272347e-05,
+      "loss": 0.0157,
+      "step": 4460
+    },
+    {
+      "epoch": 6.349431818181818,
+      "grad_norm": 0.20441566407680511,
+      "learning_rate": 6.277051165117677e-05,
+      "loss": 0.0168,
+      "step": 4470
+    },
+    {
+      "epoch": 6.363636363636363,
+      "grad_norm": 0.22486749291419983,
+      "learning_rate": 6.261057923336064e-05,
+      "loss": 0.0158,
+      "step": 4480
+    },
+    {
+      "epoch": 6.377840909090909,
+      "grad_norm": 0.23320820927619934,
+      "learning_rate": 6.245050890827042e-05,
+      "loss": 0.0185,
+      "step": 4490
+    },
+    {
+      "epoch": 6.392045454545454,
+      "grad_norm": 0.24001918733119965,
+      "learning_rate": 6.229030242640952e-05,
+      "loss": 0.0184,
+      "step": 4500
+    },
+    {
+      "epoch": 6.40625,
+      "grad_norm": 0.2788299322128296,
+      "learning_rate": 6.212996153977037e-05,
+      "loss": 0.017,
+      "step": 4510
+    },
+    {
+      "epoch": 6.420454545454546,
+      "grad_norm": 0.32427749037742615,
+      "learning_rate": 6.196948800181523e-05,
+      "loss": 0.0174,
+      "step": 4520
+    },
+    {
+      "epoch": 6.434659090909091,
+      "grad_norm": 0.34911176562309265,
+      "learning_rate": 6.180888356745695e-05,
+      "loss": 0.016,
+      "step": 4530
+    },
+    {
+      "epoch": 6.448863636363637,
+      "grad_norm": 0.2421862781047821,
+      "learning_rate": 6.164814999303995e-05,
+      "loss": 0.0166,
+      "step": 4540
+    },
+    {
+      "epoch": 6.463068181818182,
+      "grad_norm": 0.3039713203907013,
+      "learning_rate": 6.148728903632081e-05,
+      "loss": 0.0174,
+      "step": 4550
+    },
+    {
+      "epoch": 6.4772727272727275,
+      "grad_norm": 0.2429555356502533,
+      "learning_rate": 6.132630245644921e-05,
+      "loss": 0.0165,
+      "step": 4560
+    },
+    {
+      "epoch": 6.4914772727272725,
+      "grad_norm": 0.3919421136379242,
+      "learning_rate": 6.116519201394857e-05,
+      "loss": 0.0201,
+      "step": 4570
+    },
+    {
+      "epoch": 6.505681818181818,
+      "grad_norm": 0.28709888458251953,
+      "learning_rate": 6.10039594706969e-05,
+      "loss": 0.0159,
+      "step": 4580
+    },
+    {
+      "epoch": 6.519886363636363,
+      "grad_norm": 0.2700004279613495,
+      "learning_rate": 6.084260658990744e-05,
+      "loss": 0.0167,
+      "step": 4590
+    },
+    {
+      "epoch": 6.534090909090909,
+      "grad_norm": 0.38184574246406555,
+      "learning_rate": 6.068113513610943e-05,
+      "loss": 0.0178,
+      "step": 4600
+    },
+    {
+      "epoch": 6.548295454545455,
+      "grad_norm": 0.2355833351612091,
+      "learning_rate": 6.0519546875128876e-05,
+      "loss": 0.02,
+      "step": 4610
+    },
+    {
+      "epoch": 6.5625,
+      "grad_norm": 0.40573781728744507,
+      "learning_rate": 6.035784357406906e-05,
+      "loss": 0.0183,
+      "step": 4620
+    },
+    {
+      "epoch": 6.576704545454545,
+      "grad_norm": 0.24055646359920502,
+      "learning_rate": 6.01960270012914e-05,
+      "loss": 0.018,
+      "step": 4630
+    },
+    {
+      "epoch": 6.590909090909091,
+      "grad_norm": 0.22576302289962769,
+      "learning_rate": 6.003409892639599e-05,
+      "loss": 0.018,
+      "step": 4640
+    },
+    {
+      "epoch": 6.605113636363637,
+      "grad_norm": 0.38820841908454895,
+      "learning_rate": 5.9872061120202336e-05,
+      "loss": 0.0213,
+      "step": 4650
+    },
+    {
+      "epoch": 6.619318181818182,
+      "grad_norm": 0.18722712993621826,
+      "learning_rate": 5.9709915354729914e-05,
+      "loss": 0.0232,
+      "step": 4660
+    },
+    {
+      "epoch": 6.6335227272727275,
+      "grad_norm": 0.2533348798751831,
+      "learning_rate": 5.9547663403178824e-05,
+      "loss": 0.0179,
+      "step": 4670
+    },
+    {
+      "epoch": 6.6477272727272725,
+      "grad_norm": 0.25274232029914856,
+      "learning_rate": 5.9385307039910445e-05,
+      "loss": 0.0149,
+      "step": 4680
+    },
+    {
+      "epoch": 6.661931818181818,
+      "grad_norm": 0.305418998003006,
+      "learning_rate": 5.922284804042792e-05,
+      "loss": 0.0181,
+      "step": 4690
+    },
+    {
+      "epoch": 6.676136363636363,
+      "grad_norm": 0.2761019468307495,
+      "learning_rate": 5.906028818135687e-05,
+      "loss": 0.0217,
+      "step": 4700
+    },
+    {
+      "epoch": 6.690340909090909,
+      "grad_norm": 0.2500939667224884,
+      "learning_rate": 5.889762924042585e-05,
+      "loss": 0.0159,
+      "step": 4710
+    },
+    {
+      "epoch": 6.704545454545455,
+      "grad_norm": 0.2895658612251282,
+      "learning_rate": 5.873487299644699e-05,
+      "loss": 0.0181,
+      "step": 4720
+    },
+    {
+      "epoch": 6.71875,
+      "grad_norm": 0.2787564992904663,
+      "learning_rate": 5.857202122929649e-05,
+      "loss": 0.0187,
+      "step": 4730
+    },
+    {
+      "epoch": 6.732954545454545,
+      "grad_norm": 0.3605051040649414,
+      "learning_rate": 5.840907571989518e-05,
+      "loss": 0.0154,
+      "step": 4740
+    },
+    {
+      "epoch": 6.747159090909091,
+      "grad_norm": 0.45073193311691284,
+      "learning_rate": 5.824603825018904e-05,
+      "loss": 0.0178,
+      "step": 4750
+    },
+    {
+      "epoch": 6.761363636363637,
+      "grad_norm": 0.33826345205307007,
+      "learning_rate": 5.808291060312975e-05,
+      "loss": 0.0171,
+      "step": 4760
+    },
+    {
+      "epoch": 6.775568181818182,
+      "grad_norm": 0.29454803466796875,
+      "learning_rate": 5.7919694562655083e-05,
+      "loss": 0.0203,
+      "step": 4770
+    },
+    {
+      "epoch": 6.7897727272727275,
+      "grad_norm": 0.28038325905799866,
+      "learning_rate": 5.775639191366954e-05,
+      "loss": 0.0206,
+      "step": 4780
+    },
+    {
+      "epoch": 6.8039772727272725,
+      "grad_norm": 0.20273201167583466,
+      "learning_rate": 5.75930044420247e-05,
+      "loss": 0.0156,
+      "step": 4790
+    },
+    {
+      "epoch": 6.818181818181818,
+      "grad_norm": 0.23498950898647308,
+      "learning_rate": 5.74295339344998e-05,
+      "loss": 0.0136,
+      "step": 4800
+    },
+    {
+      "epoch": 6.832386363636363,
+      "grad_norm": 0.30520352721214294,
+      "learning_rate": 5.726598217878211e-05,
+      "loss": 0.0184,
+      "step": 4810
+    },
+    {
+      "epoch": 6.846590909090909,
+      "grad_norm": 0.2773948013782501,
+      "learning_rate": 5.71023509634474e-05,
+      "loss": 0.0153,
+      "step": 4820
+    },
+    {
+      "epoch": 6.860795454545455,
+      "grad_norm": 0.2782022953033447,
+      "learning_rate": 5.693864207794049e-05,
+      "loss": 0.0179,
+      "step": 4830
+    },
+    {
+      "epoch": 6.875,
+      "grad_norm": 0.29613879323005676,
+      "learning_rate": 5.677485731255545e-05,
+      "loss": 0.0248,
+      "step": 4840
+    },
+    {
+      "epoch": 6.889204545454545,
+      "grad_norm": 0.2279467135667801,
+      "learning_rate": 5.6610998458416296e-05,
+      "loss": 0.0159,
+      "step": 4850
+    },
+    {
+      "epoch": 6.903409090909091,
+      "grad_norm": 0.3251686692237854,
+      "learning_rate": 5.644706730745716e-05,
+      "loss": 0.017,
+      "step": 4860
+    },
+    {
+      "epoch": 6.917613636363637,
+      "grad_norm": 0.19078370928764343,
+      "learning_rate": 5.628306565240287e-05,
+      "loss": 0.0179,
+      "step": 4870
+    },
+    {
+      "epoch": 6.931818181818182,
+      "grad_norm": 0.29418760538101196,
+      "learning_rate": 5.611899528674923e-05,
+      "loss": 0.019,
+      "step": 4880
+    },
+    {
+      "epoch": 6.9460227272727275,
+      "grad_norm": 0.2260211706161499,
+      "learning_rate": 5.595485800474349e-05,
+      "loss": 0.0171,
+      "step": 4890
+    },
+    {
+      "epoch": 6.9602272727272725,
+      "grad_norm": 0.25741514563560486,
+      "learning_rate": 5.579065560136467e-05,
+      "loss": 0.0179,
+      "step": 4900
+    },
+    {
+      "epoch": 6.974431818181818,
+      "grad_norm": 0.2897587716579437,
+      "learning_rate": 5.562638987230392e-05,
+      "loss": 0.0175,
+      "step": 4910
+    },
+    {
+      "epoch": 6.988636363636363,
+      "grad_norm": 0.3089926838874817,
+      "learning_rate": 5.546206261394498e-05,
+      "loss": 0.0165,
+      "step": 4920
+    },
+    {
+      "epoch": 7.002840909090909,
+      "grad_norm": 0.19878511130809784,
+      "learning_rate": 5.529767562334437e-05,
+      "loss": 0.0187,
+      "step": 4930
+    },
+    {
+      "epoch": 7.017045454545454,
+      "grad_norm": 1.4476524591445923,
+      "learning_rate": 5.5133230698211926e-05,
+      "loss": 0.0185,
+      "step": 4940
+    },
+    {
+      "epoch": 7.03125,
+      "grad_norm": 0.37928301095962524,
+      "learning_rate": 5.496872963689096e-05,
+      "loss": 0.0195,
+      "step": 4950
+    },
+    {
+      "epoch": 7.045454545454546,
+      "grad_norm": 0.21582043170928955,
+      "learning_rate": 5.4804174238338756e-05,
+      "loss": 0.0161,
+      "step": 4960
+    },
+    {
+      "epoch": 7.059659090909091,
+      "grad_norm": 0.2716629207134247,
+      "learning_rate": 5.463956630210678e-05,
+      "loss": 0.0151,
+      "step": 4970
+    },
+    {
+      "epoch": 7.073863636363637,
+      "grad_norm": 0.5411096215248108,
+      "learning_rate": 5.4474907628321046e-05,
+      "loss": 0.0206,
+      "step": 4980
+    },
+    {
+      "epoch": 7.088068181818182,
+      "grad_norm": 0.293901652097702,
+      "learning_rate": 5.431020001766244e-05,
+      "loss": 0.0192,
+      "step": 4990
+    },
+    {
+      "epoch": 7.1022727272727275,
+      "grad_norm": 0.19069164991378784,
+      "learning_rate": 5.4145445271346986e-05,
+      "loss": 0.0167,
+      "step": 5000
+    },
+    {
+      "epoch": 7.1164772727272725,
+      "grad_norm": 0.4058976173400879,
+      "learning_rate": 5.398064519110622e-05,
+      "loss": 0.0137,
+      "step": 5010
+    },
+    {
+      "epoch": 7.130681818181818,
+      "grad_norm": 0.3669425845146179,
+      "learning_rate": 5.3815801579167394e-05,
+      "loss": 0.018,
+      "step": 5020
+    },
+    {
+      "epoch": 7.144886363636363,
+      "grad_norm": 0.19557751715183258,
+      "learning_rate": 5.365091623823382e-05,
+      "loss": 0.0167,
+      "step": 5030
+    },
+    {
+      "epoch": 7.159090909090909,
+      "grad_norm": 0.2595018446445465,
+      "learning_rate": 5.348599097146521e-05,
+      "loss": 0.0154,
+      "step": 5040
+    },
+    {
+      "epoch": 7.173295454545454,
+      "grad_norm": 0.22913138568401337,
+      "learning_rate": 5.3321027582457836e-05,
+      "loss": 0.0194,
+      "step": 5050
+    },
+    {
+      "epoch": 7.1875,
+      "grad_norm": 0.2039172500371933,
+      "learning_rate": 5.315602787522491e-05,
+      "loss": 0.0171,
+      "step": 5060
+    },
+    {
+      "epoch": 7.201704545454546,
+      "grad_norm": 0.3027808964252472,
+      "learning_rate": 5.299099365417678e-05,
+      "loss": 0.017,
+      "step": 5070
+    },
+    {
+      "epoch": 7.215909090909091,
+      "grad_norm": 0.28882887959480286,
+      "learning_rate": 5.2825926724101236e-05,
+      "loss": 0.0147,
+      "step": 5080
+    },
+    {
+      "epoch": 7.230113636363637,
+      "grad_norm": 0.16582432389259338,
+      "learning_rate": 5.26608288901438e-05,
+      "loss": 0.0152,
+      "step": 5090
+    },
+    {
+      "epoch": 7.244318181818182,
+      "grad_norm": 0.2663433253765106,
+      "learning_rate": 5.24957019577879e-05,
+      "loss": 0.0216,
+      "step": 5100
+    },
+    {
+      "epoch": 7.2585227272727275,
+      "grad_norm": 0.18316423892974854,
+      "learning_rate": 5.2330547732835266e-05,
+      "loss": 0.0154,
+      "step": 5110
+    },
+    {
+      "epoch": 7.2727272727272725,
+      "grad_norm": 0.37813183665275574,
+      "learning_rate": 5.2165368021385996e-05,
+      "loss": 0.019,
+      "step": 5120
+    },
+    {
+      "epoch": 7.286931818181818,
+      "grad_norm": 0.2160104364156723,
+      "learning_rate": 5.200016462981897e-05,
+      "loss": 0.0157,
+      "step": 5130
+    },
+    {
+      "epoch": 7.301136363636363,
+      "grad_norm": 0.37289661169052124,
+      "learning_rate": 5.1834939364772015e-05,
+      "loss": 0.0172,
+      "step": 5140
+    },
+    {
+      "epoch": 7.315340909090909,
+      "grad_norm": 0.12553678452968597,
+      "learning_rate": 5.166969403312214e-05,
+      "loss": 0.015,
+      "step": 5150
+    },
+    {
+      "epoch": 7.329545454545454,
+      "grad_norm": 0.2512320280075073,
+      "learning_rate": 5.1504430441965844e-05,
+      "loss": 0.0165,
+      "step": 5160
+    },
+    {
+      "epoch": 7.34375,
+      "grad_norm": 0.25537386536598206,
+      "learning_rate": 5.133915039859923e-05,
+      "loss": 0.0129,
+      "step": 5170
+    },
+    {
+      "epoch": 7.357954545454546,
+      "grad_norm": 0.27626481652259827,
+      "learning_rate": 5.1173855710498444e-05,
+      "loss": 0.0157,
+      "step": 5180
+    },
+    {
+      "epoch": 7.372159090909091,
+      "grad_norm": 0.21502751111984253,
+      "learning_rate": 5.100854818529967e-05,
+      "loss": 0.0191,
+      "step": 5190
+    },
+    {
+      "epoch": 7.386363636363637,
+      "grad_norm": 0.25681155920028687,
+      "learning_rate": 5.084322963077951e-05,
+      "loss": 0.0193,
+      "step": 5200
+    },
+    {
+      "epoch": 7.400568181818182,
+      "grad_norm": 0.2623014450073242,
+      "learning_rate": 5.067790185483522e-05,
+      "loss": 0.0166,
+      "step": 5210
+    },
+    {
+      "epoch": 7.4147727272727275,
+      "grad_norm": 0.11685733497142792,
+      "learning_rate": 5.0512566665464844e-05,
+      "loss": 0.0169,
+      "step": 5220
+    },
+    {
+      "epoch": 7.4289772727272725,
+      "grad_norm": 0.23911871016025543,
+      "learning_rate": 5.034722587074755e-05,
+      "loss": 0.0156,
+      "step": 5230
+    },
+    {
+      "epoch": 7.443181818181818,
+      "grad_norm": 0.2551325857639313,
+      "learning_rate": 5.018188127882375e-05,
+      "loss": 0.0168,
+      "step": 5240
+    },
+    {
+      "epoch": 7.457386363636363,
+      "grad_norm": 0.32587486505508423,
+      "learning_rate": 5.0016534697875417e-05,
+      "loss": 0.0132,
+      "step": 5250
+    },
+    {
+      "epoch": 7.471590909090909,
+      "grad_norm": 0.3346705436706543,
+      "learning_rate": 4.9851187936106294e-05,
+      "loss": 0.0148,
+      "step": 5260
+    },
+    {
+      "epoch": 7.485795454545454,
+      "grad_norm": 0.22128836810588837,
+      "learning_rate": 4.968584280172206e-05,
+      "loss": 0.0188,
+      "step": 5270
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 0.21978847682476044,
+      "learning_rate": 4.95205011029106e-05,
+      "loss": 0.0174,
+      "step": 5280
+    },
+    {
+      "epoch": 7.514204545454545,
+      "grad_norm": 0.27243271470069885,
+      "learning_rate": 4.935516464782227e-05,
+      "loss": 0.0162,
+      "step": 5290
+    },
+    {
+      "epoch": 7.528409090909091,
+      "grad_norm": 0.19251912832260132,
+      "learning_rate": 4.918983524455003e-05,
+      "loss": 0.014,
+      "step": 5300
+    },
+    {
+      "epoch": 7.542613636363637,
+      "grad_norm": 0.2497345507144928,
+      "learning_rate": 4.9024514701109766e-05,
+      "loss": 0.0171,
+      "step": 5310
+    },
+    {
+      "epoch": 7.556818181818182,
+      "grad_norm": 0.19166937470436096,
+      "learning_rate": 4.885920482542043e-05,
+      "loss": 0.0136,
+      "step": 5320
+    },
+    {
+      "epoch": 7.5710227272727275,
+      "grad_norm": 0.2279549241065979,
+      "learning_rate": 4.869390742528438e-05,
+      "loss": 0.0164,
+      "step": 5330
+    },
+    {
+      "epoch": 7.5852272727272725,
+      "grad_norm": 0.19668598473072052,
+      "learning_rate": 4.852862430836744e-05,
+      "loss": 0.0174,
+      "step": 5340
+    },
+    {
+      "epoch": 7.599431818181818,
+      "grad_norm": 0.35120826959609985,
+      "learning_rate": 4.836335728217933e-05,
+      "loss": 0.0163,
+      "step": 5350
+    },
+    {
+      "epoch": 7.613636363636363,
+      "grad_norm": 0.33392730355262756,
+      "learning_rate": 4.819810815405379e-05,
+      "loss": 0.0181,
+      "step": 5360
+    },
+    {
+      "epoch": 7.627840909090909,
+      "grad_norm": 0.2028454840183258,
+      "learning_rate": 4.803287873112877e-05,
+      "loss": 0.0161,
+      "step": 5370
+    },
+    {
+      "epoch": 7.642045454545455,
+      "grad_norm": 0.6147955060005188,
+      "learning_rate": 4.786767082032681e-05,
+      "loss": 0.0184,
+      "step": 5380
+    },
+    {
+      "epoch": 7.65625,
+      "grad_norm": 0.2546097934246063,
+      "learning_rate": 4.77024862283351e-05,
+      "loss": 0.0172,
+      "step": 5390
+    },
+    {
+      "epoch": 7.670454545454545,
+      "grad_norm": 0.3042433559894562,
+      "learning_rate": 4.753732676158593e-05,
+      "loss": 0.0131,
+      "step": 5400
+    },
+    {
+      "epoch": 7.684659090909091,
+      "grad_norm": 0.285646915435791,
+      "learning_rate": 4.737219422623672e-05,
+      "loss": 0.0168,
+      "step": 5410
+    },
+    {
+      "epoch": 7.698863636363637,
+      "grad_norm": 0.32720091938972473,
+      "learning_rate": 4.720709042815044e-05,
+      "loss": 0.0142,
+      "step": 5420
+    },
+    {
+      "epoch": 7.713068181818182,
+      "grad_norm": 0.21543386578559875,
+      "learning_rate": 4.704201717287578e-05,
+      "loss": 0.0132,
+      "step": 5430
+    },
+    {
+      "epoch": 7.7272727272727275,
+      "grad_norm": 0.3292575776576996,
+      "learning_rate": 4.6876976265627404e-05,
+      "loss": 0.015,
+      "step": 5440
+    },
+    {
+      "epoch": 7.7414772727272725,
+      "grad_norm": 0.19490905106067657,
+      "learning_rate": 4.671196951126626e-05,
+      "loss": 0.0174,
+      "step": 5450
+    },
+    {
+      "epoch": 7.755681818181818,
+      "grad_norm": 0.20043650269508362,
+      "learning_rate": 4.654699871427971e-05,
+      "loss": 0.0175,
+      "step": 5460
+    },
+    {
+      "epoch": 7.769886363636363,
+      "grad_norm": 0.3100859522819519,
+      "learning_rate": 4.6382065678762034e-05,
+      "loss": 0.0164,
+      "step": 5470
+    },
+    {
+      "epoch": 7.784090909090909,
+      "grad_norm": 0.22486738860607147,
+      "learning_rate": 4.6217172208394424e-05,
+      "loss": 0.0127,
+      "step": 5480
+    },
+    {
+      "epoch": 7.798295454545455,
+      "grad_norm": 0.30427470803260803,
+      "learning_rate": 4.605232010642549e-05,
+      "loss": 0.0137,
+      "step": 5490
+    },
+    {
+      "epoch": 7.8125,
+      "grad_norm": 0.23857830464839935,
+      "learning_rate": 4.588751117565142e-05,
+      "loss": 0.0124,
+      "step": 5500
+    },
+    {
+      "epoch": 7.826704545454545,
+      "grad_norm": 0.1868191957473755,
+      "learning_rate": 4.5722747218396214e-05,
+      "loss": 0.0122,
+      "step": 5510
+    },
+    {
+      "epoch": 7.840909090909091,
+      "grad_norm": 0.14829157292842865,
+      "learning_rate": 4.5558030036492194e-05,
+      "loss": 0.0153,
+      "step": 5520
+    },
+    {
+      "epoch": 7.855113636363637,
+      "grad_norm": 0.22301068902015686,
+      "learning_rate": 4.539336143125999e-05,
+      "loss": 0.0152,
+      "step": 5530
+    },
+    {
+      "epoch": 7.869318181818182,
+      "grad_norm": 0.4403107762336731,
+      "learning_rate": 4.522874320348916e-05,
+      "loss": 0.0209,
+      "step": 5540
+    },
+    {
+      "epoch": 7.8835227272727275,
+      "grad_norm": 0.189331516623497,
+      "learning_rate": 4.506417715341821e-05,
+      "loss": 0.0132,
+      "step": 5550
+    },
+    {
+      "epoch": 7.8977272727272725,
+      "grad_norm": 0.3305831253528595,
+      "learning_rate": 4.489966508071511e-05,
+      "loss": 0.0151,
+      "step": 5560
+    },
+    {
+      "epoch": 7.911931818181818,
+      "grad_norm": 0.3293136656284332,
+      "learning_rate": 4.4735208784457575e-05,
+      "loss": 0.0128,
+      "step": 5570
+    },
+    {
+      "epoch": 7.926136363636363,
+      "grad_norm": 0.1408136934041977,
+      "learning_rate": 4.457081006311325e-05,
+      "loss": 0.0174,
+      "step": 5580
+    },
+    {
+      "epoch": 7.940340909090909,
+      "grad_norm": 0.2736789584159851,
+      "learning_rate": 4.440647071452027e-05,
+      "loss": 0.0195,
+      "step": 5590
+    },
+    {
+      "epoch": 7.954545454545455,
+      "grad_norm": 0.5062747597694397,
+      "learning_rate": 4.424219253586737e-05,
+      "loss": 0.0181,
+      "step": 5600
+    },
+    {
+      "epoch": 7.96875,
+      "grad_norm": 0.22360076010227203,
+      "learning_rate": 4.407797732367443e-05,
+      "loss": 0.0165,
+      "step": 5610
+    },
+    {
+      "epoch": 7.982954545454545,
+      "grad_norm": 0.16433657705783844,
+      "learning_rate": 4.391382687377268e-05,
+      "loss": 0.0122,
+      "step": 5620
+    },
+    {
+      "epoch": 7.997159090909091,
+      "grad_norm": 0.2987329959869385,
+      "learning_rate": 4.374974298128512e-05,
+      "loss": 0.013,
+      "step": 5630
+    },
+    {
+      "epoch": 8.011363636363637,
+      "grad_norm": 0.29294562339782715,
+      "learning_rate": 4.358572744060699e-05,
+      "loss": 0.0165,
+      "step": 5640
+    },
+    {
+      "epoch": 8.025568181818182,
+      "grad_norm": 0.2564385235309601,
+      "learning_rate": 4.342178204538588e-05,
+      "loss": 0.0169,
+      "step": 5650
+    },
+    {
+      "epoch": 8.039772727272727,
+      "grad_norm": 0.5954878330230713,
+      "learning_rate": 4.325790858850241e-05,
+      "loss": 0.0128,
+      "step": 5660
+    },
+    {
+      "epoch": 8.053977272727273,
+      "grad_norm": 0.2406962811946869,
+      "learning_rate": 4.309410886205043e-05,
+      "loss": 0.0171,
+      "step": 5670
+    },
+    {
+      "epoch": 8.068181818181818,
+      "grad_norm": 0.25110092759132385,
+      "learning_rate": 4.293038465731752e-05,
+      "loss": 0.0152,
+      "step": 5680
+    },
+    {
+      "epoch": 8.082386363636363,
+      "grad_norm": 0.25011420249938965,
+      "learning_rate": 4.276673776476533e-05,
+      "loss": 0.0153,
+      "step": 5690
+    },
+    {
+      "epoch": 8.096590909090908,
+      "grad_norm": 0.36977338790893555,
+      "learning_rate": 4.260316997401007e-05,
+      "loss": 0.0186,
+      "step": 5700
+    },
+    {
+      "epoch": 8.110795454545455,
+      "grad_norm": 0.3676074147224426,
+      "learning_rate": 4.243968307380293e-05,
+      "loss": 0.0147,
+      "step": 5710
+    },
+    {
+      "epoch": 8.125,
+      "grad_norm": 0.2725844979286194,
+      "learning_rate": 4.22762788520104e-05,
+      "loss": 0.0207,
+      "step": 5720
+    },
+    {
+      "epoch": 8.139204545454545,
+      "grad_norm": 0.2570739984512329,
+      "learning_rate": 4.211295909559491e-05,
+      "loss": 0.015,
+      "step": 5730
+    },
+    {
+      "epoch": 8.153409090909092,
+      "grad_norm": 0.3399117887020111,
+      "learning_rate": 4.194972559059511e-05,
+      "loss": 0.0169,
+      "step": 5740
+    },
+    {
+      "epoch": 8.167613636363637,
+      "grad_norm": 0.2465251088142395,
+      "learning_rate": 4.178658012210651e-05,
+      "loss": 0.0133,
+      "step": 5750
+    },
+    {
+      "epoch": 8.181818181818182,
+      "grad_norm": 0.24452093243598938,
+      "learning_rate": 4.162352447426177e-05,
+      "loss": 0.0158,
+      "step": 5760
+    },
+    {
+      "epoch": 8.196022727272727,
+      "grad_norm": 0.24202120304107666,
+      "learning_rate": 4.146056043021135e-05,
+      "loss": 0.0152,
+      "step": 5770
+    },
+    {
+      "epoch": 8.210227272727273,
+      "grad_norm": 0.28808751702308655,
+      "learning_rate": 4.1297689772103944e-05,
+      "loss": 0.014,
+      "step": 5780
+    },
+    {
+      "epoch": 8.224431818181818,
+      "grad_norm": 0.2995302677154541,
+      "learning_rate": 4.113491428106694e-05,
+      "loss": 0.0146,
+      "step": 5790
+    },
+    {
+      "epoch": 8.238636363636363,
+      "grad_norm": 0.2070378214120865,
+      "learning_rate": 4.0972235737187055e-05,
+      "loss": 0.0163,
+      "step": 5800
+    },
+    {
+      "epoch": 8.252840909090908,
+      "grad_norm": 0.27411264181137085,
+      "learning_rate": 4.080965591949076e-05,
+      "loss": 0.0132,
+      "step": 5810
+    },
+    {
+      "epoch": 8.267045454545455,
+      "grad_norm": 0.21780456602573395,
+      "learning_rate": 4.0647176605924924e-05,
+      "loss": 0.0134,
+      "step": 5820
+    },
+    {
+      "epoch": 8.28125,
+      "grad_norm": 0.18731209635734558,
+      "learning_rate": 4.0484799573337255e-05,
+      "loss": 0.0176,
+      "step": 5830
+    },
+    {
+      "epoch": 8.295454545454545,
+      "grad_norm": 0.2162862867116928,
+      "learning_rate": 4.032252659745699e-05,
+      "loss": 0.0155,
+      "step": 5840
+    },
+    {
+      "epoch": 8.309659090909092,
+      "grad_norm": 0.26912444829940796,
+      "learning_rate": 4.016035945287539e-05,
+      "loss": 0.0161,
+      "step": 5850
+    },
+    {
+      "epoch": 8.323863636363637,
+      "grad_norm": 0.27645617723464966,
+      "learning_rate": 3.999829991302635e-05,
+      "loss": 0.0191,
+      "step": 5860
+    },
+    {
+      "epoch": 8.338068181818182,
+      "grad_norm": 0.28506234288215637,
+      "learning_rate": 3.983634975016707e-05,
+      "loss": 0.0167,
+      "step": 5870
+    },
+    {
+      "epoch": 8.352272727272727,
+      "grad_norm": 0.2765267789363861,
+      "learning_rate": 3.967451073535854e-05,
+      "loss": 0.0132,
+      "step": 5880
+    },
+    {
+      "epoch": 8.366477272727273,
+      "grad_norm": 0.25387290120124817,
+      "learning_rate": 3.951278463844633e-05,
+      "loss": 0.019,
+      "step": 5890
+    },
+    {
+      "epoch": 8.380681818181818,
+      "grad_norm": 0.1997891217470169,
+      "learning_rate": 3.935117322804111e-05,
+      "loss": 0.0132,
+      "step": 5900
+    },
+    {
+      "epoch": 8.394886363636363,
+      "grad_norm": 0.22510719299316406,
+      "learning_rate": 3.918967827149938e-05,
+      "loss": 0.0153,
+      "step": 5910
+    },
+    {
+      "epoch": 8.409090909090908,
+      "grad_norm": 0.2639712989330292,
+      "learning_rate": 3.9028301534904094e-05,
+      "loss": 0.0156,
+      "step": 5920
+    },
+    {
+      "epoch": 8.423295454545455,
+      "grad_norm": 0.35471466183662415,
+      "learning_rate": 3.88670447830454e-05,
+      "loss": 0.0142,
+      "step": 5930
+    },
+    {
+      "epoch": 8.4375,
+      "grad_norm": 0.1877475380897522,
+      "learning_rate": 3.870590977940132e-05,
+      "loss": 0.014,
+      "step": 5940
+    },
+    {
+      "epoch": 8.451704545454545,
+      "grad_norm": 0.17209528386592865,
+      "learning_rate": 3.8544898286118404e-05,
+      "loss": 0.0114,
+      "step": 5950
+    },
+    {
+      "epoch": 8.465909090909092,
+      "grad_norm": 0.16585692763328552,
+      "learning_rate": 3.838401206399257e-05,
+      "loss": 0.0141,
+      "step": 5960
+    },
+    {
+      "epoch": 8.480113636363637,
+      "grad_norm": 0.32535460591316223,
+      "learning_rate": 3.822325287244975e-05,
+      "loss": 0.0141,
+      "step": 5970
+    },
+    {
+      "epoch": 8.494318181818182,
+      "grad_norm": 0.15518029034137726,
+      "learning_rate": 3.8062622469526725e-05,
+      "loss": 0.012,
+      "step": 5980
+    },
+    {
+      "epoch": 8.508522727272727,
+      "grad_norm": 0.19701546430587769,
+      "learning_rate": 3.790212261185183e-05,
+      "loss": 0.0141,
+      "step": 5990
+    },
+    {
+      "epoch": 8.522727272727273,
+      "grad_norm": 0.24397309124469757,
+      "learning_rate": 3.7741755054625794e-05,
+      "loss": 0.0135,
+      "step": 6000
+    },
+    {
+      "epoch": 8.536931818181818,
+      "grad_norm": 0.3112819790840149,
+      "learning_rate": 3.758152155160255e-05,
+      "loss": 0.0169,
+      "step": 6010
+    },
+    {
+      "epoch": 8.551136363636363,
+      "grad_norm": 0.26772013306617737,
+      "learning_rate": 3.742142385506999e-05,
+      "loss": 0.0128,
+      "step": 6020
+    },
+    {
+      "epoch": 8.565340909090908,
+      "grad_norm": 0.196287140250206,
+      "learning_rate": 3.72614637158309e-05,
+      "loss": 0.0143,
+      "step": 6030
+    },
+    {
+      "epoch": 8.579545454545455,
+      "grad_norm": 0.3775482475757599,
+      "learning_rate": 3.710164288318371e-05,
+      "loss": 0.014,
+      "step": 6040
+    },
+    {
+      "epoch": 8.59375,
+      "grad_norm": 0.2950417995452881,
+      "learning_rate": 3.694196310490345e-05,
+      "loss": 0.012,
+      "step": 6050
+    },
+    {
+      "epoch": 8.607954545454545,
+      "grad_norm": 0.23802444338798523,
+      "learning_rate": 3.678242612722259e-05,
+      "loss": 0.0104,
+      "step": 6060
+    },
+    {
+      "epoch": 8.622159090909092,
+      "grad_norm": 0.35057953000068665,
+      "learning_rate": 3.6623033694811953e-05,
+      "loss": 0.014,
+      "step": 6070
+    },
+    {
+      "epoch": 8.636363636363637,
+      "grad_norm": 0.21443971991539001,
+      "learning_rate": 3.6463787550761665e-05,
+      "loss": 0.0157,
+      "step": 6080
+    },
+    {
+      "epoch": 8.650568181818182,
+      "grad_norm": 0.19903965294361115,
+      "learning_rate": 3.630468943656202e-05,
+      "loss": 0.0101,
+      "step": 6090
+    },
+    {
+      "epoch": 8.664772727272727,
+      "grad_norm": 0.17095011472702026,
+      "learning_rate": 3.6145741092084523e-05,
+      "loss": 0.0167,
+      "step": 6100
+    },
+    {
+      "epoch": 8.678977272727273,
+      "grad_norm": 0.38296616077423096,
+      "learning_rate": 3.598694425556278e-05,
+      "loss": 0.0118,
+      "step": 6110
+    },
+    {
+      "epoch": 8.693181818181818,
+      "grad_norm": 0.24473357200622559,
+      "learning_rate": 3.58283006635736e-05,
+      "loss": 0.0146,
+      "step": 6120
+    },
+    {
+      "epoch": 8.707386363636363,
+      "grad_norm": 0.2163621187210083,
+      "learning_rate": 3.566981205101781e-05,
+      "loss": 0.0138,
+      "step": 6130
+    },
+    {
+      "epoch": 8.721590909090908,
+      "grad_norm": 0.1881503015756607,
+      "learning_rate": 3.5511480151101556e-05,
+      "loss": 0.0122,
+      "step": 6140
+    },
+    {
+      "epoch": 8.735795454545455,
+      "grad_norm": 0.18852360546588898,
+      "learning_rate": 3.5353306695317104e-05,
+      "loss": 0.0119,
+      "step": 6150
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.1903574913740158,
+      "learning_rate": 3.519529341342402e-05,
+      "loss": 0.0111,
+      "step": 6160
+    },
+    {
+      "epoch": 8.764204545454545,
+      "grad_norm": 0.33430373668670654,
+      "learning_rate": 3.503744203343026e-05,
+      "loss": 0.0126,
+      "step": 6170
+    },
+    {
+      "epoch": 8.778409090909092,
+      "grad_norm": 0.21746020019054413,
+      "learning_rate": 3.487975428157318e-05,
+      "loss": 0.0138,
+      "step": 6180
+    },
+    {
+      "epoch": 8.792613636363637,
+      "grad_norm": 0.2795451283454895,
+      "learning_rate": 3.472223188230083e-05,
+      "loss": 0.0148,
+      "step": 6190
+    },
+    {
+      "epoch": 8.806818181818182,
+      "grad_norm": 0.3572300970554352,
+      "learning_rate": 3.4564876558252866e-05,
+      "loss": 0.015,
+      "step": 6200
+    },
+    {
+      "epoch": 8.821022727272727,
+      "grad_norm": 0.2349061667919159,
+      "learning_rate": 3.440769003024195e-05,
+      "loss": 0.0119,
+      "step": 6210
+    },
+    {
+      "epoch": 8.835227272727273,
+      "grad_norm": 0.22528418898582458,
+      "learning_rate": 3.425067401723477e-05,
+      "loss": 0.0129,
+      "step": 6220
+    },
+    {
+      "epoch": 8.849431818181818,
+      "grad_norm": 0.23531129956245422,
+      "learning_rate": 3.409383023633325e-05,
+      "loss": 0.0138,
+      "step": 6230
+    },
+    {
+      "epoch": 8.863636363636363,
+      "grad_norm": 0.2939990162849426,
+      "learning_rate": 3.3937160402755894e-05,
+      "loss": 0.0121,
+      "step": 6240
+    },
+    {
+      "epoch": 8.877840909090908,
+      "grad_norm": 0.14647825062274933,
+      "learning_rate": 3.378066622981885e-05,
+      "loss": 0.0127,
+      "step": 6250
+    },
+    {
+      "epoch": 8.892045454545455,
+      "grad_norm": 0.17162439227104187,
+      "learning_rate": 3.362434942891738e-05,
+      "loss": 0.0152,
+      "step": 6260
+    },
+    {
+      "epoch": 8.90625,
+      "grad_norm": 0.2817434072494507,
+      "learning_rate": 3.346821170950693e-05,
+      "loss": 0.0171,
+      "step": 6270
+    },
+    {
+      "epoch": 8.920454545454545,
+      "grad_norm": 0.34158846735954285,
+      "learning_rate": 3.3312254779084585e-05,
+      "loss": 0.016,
+      "step": 6280
+    },
+    {
+      "epoch": 8.934659090909092,
+      "grad_norm": 0.257030189037323,
+      "learning_rate": 3.315648034317039e-05,
+      "loss": 0.0145,
+      "step": 6290
+    },
+    {
+      "epoch": 8.948863636363637,
+      "grad_norm": 0.17781952023506165,
+      "learning_rate": 3.3000890105288564e-05,
+      "loss": 0.0151,
+      "step": 6300
+    },
+    {
+      "epoch": 8.963068181818182,
+      "grad_norm": 0.2737427055835724,
+      "learning_rate": 3.284548576694908e-05,
+      "loss": 0.0113,
+      "step": 6310
+    },
+    {
+      "epoch": 8.977272727272727,
+      "grad_norm": 0.2249009609222412,
+      "learning_rate": 3.2690269027628815e-05,
+      "loss": 0.0108,
+      "step": 6320
+    },
+    {
+      "epoch": 8.991477272727273,
+      "grad_norm": 0.28954455256462097,
+      "learning_rate": 3.253524158475324e-05,
+      "loss": 0.0121,
+      "step": 6330
+    },
+    {
+      "epoch": 9.005681818181818,
+      "grad_norm": 0.36633729934692383,
+      "learning_rate": 3.238040513367757e-05,
+      "loss": 0.0148,
+      "step": 6340
+    },
+    {
+      "epoch": 9.019886363636363,
+      "grad_norm": 0.23659028112888336,
+      "learning_rate": 3.222576136766843e-05,
+      "loss": 0.0121,
+      "step": 6350
+    },
+    {
+      "epoch": 9.034090909090908,
+      "grad_norm": 0.15457887947559357,
+      "learning_rate": 3.2071311977885324e-05,
+      "loss": 0.0102,
+      "step": 6360
+    },
+    {
+      "epoch": 9.048295454545455,
+      "grad_norm": 0.3846118748188019,
+      "learning_rate": 3.191705865336197e-05,
+      "loss": 0.0144,
+      "step": 6370
+    },
+    {
+      "epoch": 9.0625,
+      "grad_norm": 0.26210084557533264,
+      "learning_rate": 3.1763003080988075e-05,
+      "loss": 0.0146,
+      "step": 6380
+    },
+    {
+      "epoch": 9.076704545454545,
+      "grad_norm": 0.22350722551345825,
+      "learning_rate": 3.160914694549063e-05,
+      "loss": 0.0137,
+      "step": 6390
+    },
+    {
+      "epoch": 9.090909090909092,
+      "grad_norm": 0.20034578442573547,
+      "learning_rate": 3.145549192941573e-05,
+      "loss": 0.0143,
+      "step": 6400
+    },
+    {
+      "epoch": 9.105113636363637,
+      "grad_norm": 0.23915934562683105,
+      "learning_rate": 3.130203971310999e-05,
+      "loss": 0.0138,
+      "step": 6410
+    },
+    {
+      "epoch": 9.119318181818182,
+      "grad_norm": 0.37302061915397644,
+      "learning_rate": 3.114879197470225e-05,
+      "loss": 0.0144,
+      "step": 6420
+    },
+    {
+      "epoch": 9.133522727272727,
+      "grad_norm": 0.23130542039871216,
+      "learning_rate": 3.0995750390085285e-05,
+      "loss": 0.0125,
+      "step": 6430
+    },
+    {
+      "epoch": 9.147727272727273,
+      "grad_norm": 0.19056399166584015,
+      "learning_rate": 3.084291663289728e-05,
+      "loss": 0.0109,
+      "step": 6440
+    },
+    {
+      "epoch": 9.161931818181818,
+      "grad_norm": 0.22148406505584717,
+      "learning_rate": 3.069029237450375e-05,
+      "loss": 0.011,
+      "step": 6450
+    },
+    {
+      "epoch": 9.176136363636363,
+      "grad_norm": 0.21201135218143463,
+      "learning_rate": 3.053787928397911e-05,
+      "loss": 0.0107,
+      "step": 6460
+    },
+    {
+      "epoch": 9.190340909090908,
+      "grad_norm": 0.24332214891910553,
+      "learning_rate": 3.0385679028088526e-05,
+      "loss": 0.0121,
+      "step": 6470
+    },
+    {
+      "epoch": 9.204545454545455,
+      "grad_norm": 0.24824318289756775,
+      "learning_rate": 3.023369327126959e-05,
+      "loss": 0.014,
+      "step": 6480
+    },
+    {
+      "epoch": 9.21875,
+      "grad_norm": 0.3004310131072998,
+      "learning_rate": 3.0081923675614198e-05,
+      "loss": 0.0102,
+      "step": 6490
+    },
+    {
+      "epoch": 9.232954545454545,
+      "grad_norm": 0.31485769152641296,
+      "learning_rate": 2.993037190085034e-05,
+      "loss": 0.012,
+      "step": 6500
+    },
+    {
+      "epoch": 9.247159090909092,
+      "grad_norm": 0.17692938446998596,
+      "learning_rate": 2.977903960432392e-05,
+      "loss": 0.0129,
+      "step": 6510
+    },
+    {
+      "epoch": 9.261363636363637,
+      "grad_norm": 0.21402569115161896,
+      "learning_rate": 2.9627928440980722e-05,
+      "loss": 0.0128,
+      "step": 6520
+    },
+    {
+      "epoch": 9.275568181818182,
+      "grad_norm": 0.19698747992515564,
+      "learning_rate": 2.9477040063348183e-05,
+      "loss": 0.0152,
+      "step": 6530
+    },
+    {
+      "epoch": 9.289772727272727,
+      "grad_norm": 0.19634680449962616,
+      "learning_rate": 2.9326376121517456e-05,
+      "loss": 0.0105,
+      "step": 6540
+    },
+    {
+      "epoch": 9.303977272727273,
+      "grad_norm": 0.18059450387954712,
+      "learning_rate": 2.9175938263125236e-05,
+      "loss": 0.0118,
+      "step": 6550
+    },
+    {
+      "epoch": 9.318181818181818,
+      "grad_norm": 0.2729738652706146,
+      "learning_rate": 2.9025728133335873e-05,
+      "loss": 0.0161,
+      "step": 6560
+    },
+    {
+      "epoch": 9.332386363636363,
+      "grad_norm": 0.19791188836097717,
+      "learning_rate": 2.8875747374823288e-05,
+      "loss": 0.013,
+      "step": 6570
+    },
+    {
+      "epoch": 9.346590909090908,
+      "grad_norm": 0.24051375687122345,
+      "learning_rate": 2.872599762775298e-05,
+      "loss": 0.0186,
+      "step": 6580
+    },
+    {
+      "epoch": 9.360795454545455,
+      "grad_norm": 0.21249376237392426,
+      "learning_rate": 2.857648052976425e-05,
+      "loss": 0.0141,
+      "step": 6590
+    },
+    {
+      "epoch": 9.375,
+      "grad_norm": 0.20016810297966003,
+      "learning_rate": 2.8427197715952047e-05,
+      "loss": 0.0098,
+      "step": 6600
+    },
+    {
+      "epoch": 9.389204545454545,
+      "grad_norm": 0.10853852331638336,
+      "learning_rate": 2.8278150818849393e-05,
+      "loss": 0.0124,
+      "step": 6610
+    },
+    {
+      "epoch": 9.403409090909092,
+      "grad_norm": 0.21107201278209686,
+      "learning_rate": 2.812934146840922e-05,
+      "loss": 0.0139,
+      "step": 6620
+    },
+    {
+      "epoch": 9.417613636363637,
+      "grad_norm": 0.22437851130962372,
+      "learning_rate": 2.7980771291986764e-05,
+      "loss": 0.0106,
+      "step": 6630
+    },
+    {
+      "epoch": 9.431818181818182,
+      "grad_norm": 0.22271908819675446,
+      "learning_rate": 2.783244191432167e-05,
+      "loss": 0.0121,
+      "step": 6640
+    },
+    {
+      "epoch": 9.446022727272727,
+      "grad_norm": 0.14856989681720734,
+      "learning_rate": 2.768435495752022e-05,
+      "loss": 0.0108,
+      "step": 6650
+    },
+    {
+      "epoch": 9.460227272727273,
+      "grad_norm": 0.16649934649467468,
+      "learning_rate": 2.753651204103771e-05,
+      "loss": 0.0122,
+      "step": 6660
+    },
+    {
+      "epoch": 9.474431818181818,
+      "grad_norm": 0.17911995947360992,
+      "learning_rate": 2.7388914781660523e-05,
+      "loss": 0.0099,
+      "step": 6670
+    },
+    {
+      "epoch": 9.488636363636363,
+      "grad_norm": 0.1906937211751938,
+      "learning_rate": 2.7241564793488693e-05,
+      "loss": 0.0111,
+      "step": 6680
+    },
+    {
+      "epoch": 9.502840909090908,
+      "grad_norm": 0.2392403483390808,
+      "learning_rate": 2.7094463687918037e-05,
+      "loss": 0.0121,
+      "step": 6690
+    },
+    {
+      "epoch": 9.517045454545455,
+      "grad_norm": 0.16135956346988678,
+      "learning_rate": 2.694761307362268e-05,
+      "loss": 0.0122,
+      "step": 6700
+    },
+    {
+      "epoch": 9.53125,
+      "grad_norm": 0.18655622005462646,
+      "learning_rate": 2.6801014556537467e-05,
+      "loss": 0.0142,
+      "step": 6710
+    },
+    {
+      "epoch": 9.545454545454545,
+      "grad_norm": 0.19424274563789368,
+      "learning_rate": 2.6654669739840243e-05,
+      "loss": 0.0136,
+      "step": 6720
+    },
+    {
+      "epoch": 9.559659090909092,
+      "grad_norm": 0.1531798243522644,
+      "learning_rate": 2.650858022393451e-05,
+      "loss": 0.0135,
+      "step": 6730
+    },
+    {
+      "epoch": 9.573863636363637,
+      "grad_norm": 0.2101268321275711,
+      "learning_rate": 2.6362747606431747e-05,
+      "loss": 0.0128,
+      "step": 6740
+    },
+    {
+      "epoch": 9.588068181818182,
+      "grad_norm": 0.15283378958702087,
+      "learning_rate": 2.6217173482134172e-05,
+      "loss": 0.0168,
+      "step": 6750
+    },
+    {
+      "epoch": 9.602272727272727,
+      "grad_norm": 0.2049722820520401,
+      "learning_rate": 2.6071859443017044e-05,
+      "loss": 0.0106,
+      "step": 6760
+    },
+    {
+      "epoch": 9.616477272727273,
+      "grad_norm": 0.17880743741989136,
+      "learning_rate": 2.5926807078211414e-05,
+      "loss": 0.0088,
+      "step": 6770
+    },
+    {
+      "epoch": 9.630681818181818,
+      "grad_norm": 0.22341349720954895,
+      "learning_rate": 2.5782017973986728e-05,
+      "loss": 0.0106,
+      "step": 6780
+    },
+    {
+      "epoch": 9.644886363636363,
+      "grad_norm": 0.2355606108903885,
+      "learning_rate": 2.5637493713733374e-05,
+      "loss": 0.0138,
+      "step": 6790
+    },
+    {
+      "epoch": 9.659090909090908,
+      "grad_norm": 0.1841660439968109,
+      "learning_rate": 2.549323587794559e-05,
+      "loss": 0.0117,
+      "step": 6800
+    },
+    {
+      "epoch": 9.673295454545455,
+      "grad_norm": 0.2315818965435028,
+      "learning_rate": 2.5349246044203895e-05,
+      "loss": 0.0125,
+      "step": 6810
+    },
+    {
+      "epoch": 9.6875,
+      "grad_norm": 0.2293039709329605,
+      "learning_rate": 2.520552578715808e-05,
+      "loss": 0.0104,
+      "step": 6820
+    },
+    {
+      "epoch": 9.701704545454545,
+      "grad_norm": 0.17732688784599304,
+      "learning_rate": 2.506207667850981e-05,
+      "loss": 0.0097,
+      "step": 6830
+    },
+    {
+      "epoch": 9.715909090909092,
+      "grad_norm": 0.3996809720993042,
+      "learning_rate": 2.4918900286995555e-05,
+      "loss": 0.0097,
+      "step": 6840
+    },
+    {
+      "epoch": 9.730113636363637,
+      "grad_norm": 0.16281253099441528,
+      "learning_rate": 2.4775998178369458e-05,
+      "loss": 0.0137,
+      "step": 6850
+    },
+    {
+      "epoch": 9.744318181818182,
+      "grad_norm": 0.18379844725131989,
+      "learning_rate": 2.4633371915386017e-05,
+      "loss": 0.0113,
+      "step": 6860
+    },
+    {
+      "epoch": 9.758522727272727,
+      "grad_norm": 0.1760406643152237,
+      "learning_rate": 2.4491023057783235e-05,
+      "loss": 0.0133,
+      "step": 6870
+    },
+    {
+      "epoch": 9.772727272727273,
+      "grad_norm": 0.29180634021759033,
+      "learning_rate": 2.4348953162265375e-05,
+      "loss": 0.0129,
+      "step": 6880
+    },
+    {
+      "epoch": 9.786931818181818,
+      "grad_norm": 0.27271148562431335,
+      "learning_rate": 2.420716378248607e-05,
+      "loss": 0.0146,
+      "step": 6890
+    },
+    {
+      "epoch": 9.801136363636363,
+      "grad_norm": 0.22666561603546143,
+      "learning_rate": 2.4065656469031266e-05,
+      "loss": 0.0116,
+      "step": 6900
+    },
+    {
+      "epoch": 9.815340909090908,
+      "grad_norm": 0.2339516133069992,
+      "learning_rate": 2.3924432769402268e-05,
+      "loss": 0.0124,
+      "step": 6910
+    },
+    {
+      "epoch": 9.829545454545455,
+      "grad_norm": 0.156479150056839,
+      "learning_rate": 2.3783494227998844e-05,
+      "loss": 0.0125,
+      "step": 6920
+    },
+    {
+      "epoch": 9.84375,
+      "grad_norm": 0.1421717256307602,
+      "learning_rate": 2.3642842386102264e-05,
+      "loss": 0.0139,
+      "step": 6930
+    },
+    {
+      "epoch": 9.857954545454545,
+      "grad_norm": 0.20596732199192047,
+      "learning_rate": 2.3502478781858567e-05,
+      "loss": 0.0081,
+      "step": 6940
+    },
+    {
+      "epoch": 9.872159090909092,
+      "grad_norm": 0.20319467782974243,
+      "learning_rate": 2.3362404950261628e-05,
+      "loss": 0.0118,
+      "step": 6950
+    },
+    {
+      "epoch": 9.886363636363637,
+      "grad_norm": 0.24336227774620056,
+      "learning_rate": 2.3222622423136458e-05,
+      "loss": 0.012,
+      "step": 6960
+    },
+    {
+      "epoch": 9.900568181818182,
+      "grad_norm": 0.1479557752609253,
+      "learning_rate": 2.3083132729122332e-05,
+      "loss": 0.0107,
+      "step": 6970
+    },
+    {
+      "epoch": 9.914772727272727,
+      "grad_norm": 0.2064417451620102,
+      "learning_rate": 2.294393739365621e-05,
+      "loss": 0.0153,
+      "step": 6980
+    },
+    {
+      "epoch": 9.928977272727273,
+      "grad_norm": 0.568552553653717,
+      "learning_rate": 2.2805037938956e-05,
+      "loss": 0.0119,
+      "step": 6990
+    },
+    {
+      "epoch": 9.943181818181818,
+      "grad_norm": 0.21720533072948456,
+      "learning_rate": 2.266643588400386e-05,
+      "loss": 0.0123,
+      "step": 7000
+    },
+    {
+      "epoch": 9.957386363636363,
+      "grad_norm": 0.18104858696460724,
+      "learning_rate": 2.252813274452969e-05,
+      "loss": 0.0112,
+      "step": 7010
+    },
+    {
+      "epoch": 9.971590909090908,
+      "grad_norm": 0.16788426041603088,
+      "learning_rate": 2.2390130032994427e-05,
+      "loss": 0.0111,
+      "step": 7020
+    },
+    {
+      "epoch": 9.985795454545455,
+      "grad_norm": 0.2534903585910797,
+      "learning_rate": 2.2252429258573633e-05,
+      "loss": 0.0117,
+      "step": 7030
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.3211720585823059,
+      "learning_rate": 2.2115031927140904e-05,
+      "loss": 0.0133,
+      "step": 7040
+    },
+    {
+      "epoch": 10.014204545454545,
+      "grad_norm": 0.30026450753211975,
+      "learning_rate": 2.1977939541251463e-05,
+      "loss": 0.0131,
+      "step": 7050
+    },
+    {
+      "epoch": 10.028409090909092,
+      "grad_norm": 0.25989219546318054,
+      "learning_rate": 2.1841153600125684e-05,
+      "loss": 0.0144,
+      "step": 7060
+    },
+    {
+      "epoch": 10.042613636363637,
+      "grad_norm": 0.24319857358932495,
+      "learning_rate": 2.170467559963267e-05,
+      "loss": 0.014,
+      "step": 7070
+    },
+    {
+      "epoch": 10.056818181818182,
+      "grad_norm": 0.16669923067092896,
+      "learning_rate": 2.1568507032273982e-05,
+      "loss": 0.0116,
+      "step": 7080
+    },
+    {
+      "epoch": 10.071022727272727,
+      "grad_norm": 0.1890438348054886,
+      "learning_rate": 2.1432649387167264e-05,
+      "loss": 0.0096,
+      "step": 7090
+    },
+    {
+      "epoch": 10.085227272727273,
+      "grad_norm": 0.1900719255208969,
+      "learning_rate": 2.1297104150029973e-05,
+      "loss": 0.0109,
+      "step": 7100
+    },
+    {
+      "epoch": 10.099431818181818,
+      "grad_norm": 0.277690052986145,
+      "learning_rate": 2.116187280316307e-05,
+      "loss": 0.0099,
+      "step": 7110
+    },
+    {
+      "epoch": 10.113636363636363,
+      "grad_norm": 0.32027512788772583,
+      "learning_rate": 2.1026956825434908e-05,
+      "loss": 0.0117,
+      "step": 7120
+    },
+    {
+      "epoch": 10.127840909090908,
+      "grad_norm": 0.1674855351448059,
+      "learning_rate": 2.0892357692265017e-05,
+      "loss": 0.011,
+      "step": 7130
+    },
+    {
+      "epoch": 10.142045454545455,
+      "grad_norm": 0.259202778339386,
+      "learning_rate": 2.0758076875607947e-05,
+      "loss": 0.0113,
+      "step": 7140
+    },
+    {
+      "epoch": 10.15625,
+      "grad_norm": 0.2248423546552658,
+      "learning_rate": 2.0624115843937207e-05,
+      "loss": 0.014,
+      "step": 7150
+    },
+    {
+      "epoch": 10.170454545454545,
+      "grad_norm": 0.28167814016342163,
+      "learning_rate": 2.0490476062229157e-05,
+      "loss": 0.0134,
+      "step": 7160
+    },
+    {
+      "epoch": 10.184659090909092,
+      "grad_norm": 0.2060270607471466,
+      "learning_rate": 2.035715899194704e-05,
+      "loss": 0.0108,
+      "step": 7170
+    },
+    {
+      "epoch": 10.198863636363637,
+      "grad_norm": 0.1836121380329132,
+      "learning_rate": 2.022416609102499e-05,
+      "loss": 0.0101,
+      "step": 7180
+    },
+    {
+      "epoch": 10.213068181818182,
+      "grad_norm": 0.5955054759979248,
+      "learning_rate": 2.009149881385205e-05,
+      "loss": 0.0141,
+      "step": 7190
+    },
+    {
+      "epoch": 10.227272727272727,
+      "grad_norm": 0.16109995543956757,
+      "learning_rate": 1.995915861125634e-05,
+      "loss": 0.0096,
+      "step": 7200
+    },
+    {
+      "epoch": 10.241477272727273,
+      "grad_norm": 0.21069282293319702,
+      "learning_rate": 1.9827146930489065e-05,
+      "loss": 0.0106,
+      "step": 7210
+    },
+    {
+      "epoch": 10.255681818181818,
+      "grad_norm": 0.21853871643543243,
+      "learning_rate": 1.9695465215208848e-05,
+      "loss": 0.0104,
+      "step": 7220
+    },
+    {
+      "epoch": 10.269886363636363,
+      "grad_norm": 0.1952890157699585,
+      "learning_rate": 1.9564114905465813e-05,
+      "loss": 0.0127,
+      "step": 7230
+    },
+    {
+      "epoch": 10.284090909090908,
+      "grad_norm": 0.18870890140533447,
+      "learning_rate": 1.9433097437685936e-05,
+      "loss": 0.0095,
+      "step": 7240
+    },
+    {
+      "epoch": 10.298295454545455,
+      "grad_norm": 0.18546724319458008,
+      "learning_rate": 1.930241424465521e-05,
+      "loss": 0.0106,
+      "step": 7250
+    },
+    {
+      "epoch": 10.3125,
+      "grad_norm": 0.23328813910484314,
+      "learning_rate": 1.9172066755504115e-05,
+      "loss": 0.0099,
+      "step": 7260
+    },
+    {
+      "epoch": 10.326704545454545,
+      "grad_norm": 0.22469045221805573,
+      "learning_rate": 1.9042056395691914e-05,
+      "loss": 0.0105,
+      "step": 7270
+    },
+    {
+      "epoch": 10.340909090909092,
+      "grad_norm": 0.09730111062526703,
+      "learning_rate": 1.8912384586991066e-05,
+      "loss": 0.0111,
+      "step": 7280
+    },
+    {
+      "epoch": 10.355113636363637,
+      "grad_norm": 0.15830957889556885,
+      "learning_rate": 1.8783052747471717e-05,
+      "loss": 0.0083,
+      "step": 7290
+    },
+    {
+      "epoch": 10.369318181818182,
+      "grad_norm": 0.21241037547588348,
+      "learning_rate": 1.865406229148611e-05,
+      "loss": 0.012,
+      "step": 7300
+    },
+    {
+      "epoch": 10.383522727272727,
+      "grad_norm": 0.17303265631198883,
+      "learning_rate": 1.8525414629653233e-05,
+      "loss": 0.0101,
+      "step": 7310
+    },
+    {
+      "epoch": 10.397727272727273,
+      "grad_norm": 0.23121459782123566,
+      "learning_rate": 1.8397111168843255e-05,
+      "loss": 0.0131,
+      "step": 7320
+    },
+    {
+      "epoch": 10.411931818181818,
+      "grad_norm": 0.145586758852005,
+      "learning_rate": 1.8269153312162323e-05,
+      "loss": 0.0094,
+      "step": 7330
+    },
+    {
+      "epoch": 10.426136363636363,
+      "grad_norm": 0.15283986926078796,
+      "learning_rate": 1.8141542458937054e-05,
+      "loss": 0.0116,
+      "step": 7340
+    },
+    {
+      "epoch": 10.440340909090908,
+      "grad_norm": 0.1770382821559906,
+      "learning_rate": 1.8014280004699268e-05,
+      "loss": 0.0103,
+      "step": 7350
+    },
+    {
+      "epoch": 10.454545454545455,
+      "grad_norm": 0.2646413743495941,
+      "learning_rate": 1.788736734117078e-05,
+      "loss": 0.0088,
+      "step": 7360
+    },
+    {
+      "epoch": 10.46875,
+      "grad_norm": 0.21698594093322754,
+      "learning_rate": 1.7760805856248152e-05,
+      "loss": 0.0103,
+      "step": 7370
+    },
+    {
+      "epoch": 10.482954545454545,
+      "grad_norm": 0.18520593643188477,
+      "learning_rate": 1.7634596933987518e-05,
+      "loss": 0.0117,
+      "step": 7380
+    },
+    {
+      "epoch": 10.497159090909092,
+      "grad_norm": 0.2715875506401062,
+      "learning_rate": 1.7508741954589404e-05,
+      "loss": 0.0112,
+      "step": 7390
+    },
+    {
+      "epoch": 10.511363636363637,
+      "grad_norm": 0.1851782649755478,
+      "learning_rate": 1.7383242294383717e-05,
+      "loss": 0.0102,
+      "step": 7400
+    },
+    {
+      "epoch": 10.525568181818182,
+      "grad_norm": 0.21845151484012604,
+      "learning_rate": 1.7258099325814632e-05,
+      "loss": 0.0101,
+      "step": 7410
+    },
+    {
+      "epoch": 10.539772727272727,
+      "grad_norm": 0.3004157841205597,
+      "learning_rate": 1.7133314417425594e-05,
+      "loss": 0.011,
+      "step": 7420
+    },
+    {
+      "epoch": 10.553977272727273,
+      "grad_norm": 0.1729803830385208,
+      "learning_rate": 1.7008888933844408e-05,
+      "loss": 0.0099,
+      "step": 7430
+    },
+    {
+      "epoch": 10.568181818181818,
+      "grad_norm": 0.3214139938354492,
+      "learning_rate": 1.6884824235768172e-05,
+      "loss": 0.0126,
+      "step": 7440
+    },
+    {
+      "epoch": 10.582386363636363,
+      "grad_norm": 0.2877606153488159,
+      "learning_rate": 1.6761121679948592e-05,
+      "loss": 0.0104,
+      "step": 7450
+    },
+    {
+      "epoch": 10.596590909090908,
+      "grad_norm": 0.22641102969646454,
+      "learning_rate": 1.663778261917695e-05,
+      "loss": 0.0079,
+      "step": 7460
+    },
+    {
+      "epoch": 10.610795454545455,
+      "grad_norm": 0.16428832709789276,
+      "learning_rate": 1.651480840226952e-05,
+      "loss": 0.0128,
+      "step": 7470
+    },
+    {
+      "epoch": 10.625,
+      "grad_norm": 0.09172098338603973,
+      "learning_rate": 1.639220037405258e-05,
+      "loss": 0.0094,
+      "step": 7480
+    },
+    {
+      "epoch": 10.639204545454545,
+      "grad_norm": 0.25063034892082214,
+      "learning_rate": 1.6269959875347906e-05,
+      "loss": 0.0128,
+      "step": 7490
+    },
+    {
+      "epoch": 10.653409090909092,
+      "grad_norm": 0.27451658248901367,
+      "learning_rate": 1.614808824295802e-05,
+      "loss": 0.0109,
+      "step": 7500
+    },
+    {
+      "epoch": 10.667613636363637,
+      "grad_norm": 0.2790447473526001,
+      "learning_rate": 1.602658680965152e-05,
+      "loss": 0.0109,
+      "step": 7510
+    },
+    {
+      "epoch": 10.681818181818182,
+      "grad_norm": 0.26930713653564453,
+      "learning_rate": 1.5905456904148686e-05,
+      "loss": 0.0123,
+      "step": 7520
+    },
+    {
+      "epoch": 10.696022727272727,
+      "grad_norm": 0.15025542676448822,
+      "learning_rate": 1.57846998511067e-05,
+      "loss": 0.0094,
+      "step": 7530
+    },
+    {
+      "epoch": 10.710227272727273,
+      "grad_norm": 0.179928719997406,
+      "learning_rate": 1.566431697110538e-05,
+      "loss": 0.0133,
+      "step": 7540
+    },
+    {
+      "epoch": 10.724431818181818,
+      "grad_norm": 0.24188250303268433,
+      "learning_rate": 1.554430958063259e-05,
+      "loss": 0.0103,
+      "step": 7550
+    },
+    {
+      "epoch": 10.738636363636363,
+      "grad_norm": 0.14740188419818878,
+      "learning_rate": 1.5424678992069912e-05,
+      "loss": 0.0089,
+      "step": 7560
+    },
+    {
+      "epoch": 10.752840909090908,
+      "grad_norm": 0.2390107810497284,
+      "learning_rate": 1.5305426513678362e-05,
+      "loss": 0.0108,
+      "step": 7570
+    },
+    {
+      "epoch": 10.767045454545455,
+      "grad_norm": 0.2551465332508087,
+      "learning_rate": 1.518655344958388e-05,
+      "loss": 0.0121,
+      "step": 7580
+    },
+    {
+      "epoch": 10.78125,
+      "grad_norm": 0.1420135349035263,
+      "learning_rate": 1.5068061099763275e-05,
+      "loss": 0.0113,
+      "step": 7590
+    },
+    {
+      "epoch": 10.795454545454545,
+      "grad_norm": 0.28341472148895264,
+      "learning_rate": 1.494995076002988e-05,
+      "loss": 0.0106,
+      "step": 7600
+    },
+    {
+      "epoch": 10.809659090909092,
+      "grad_norm": 0.11734076589345932,
+      "learning_rate": 1.4832223722019456e-05,
+      "loss": 0.0092,
+      "step": 7610
+    },
+    {
+      "epoch": 10.823863636363637,
+      "grad_norm": 0.16179542243480682,
+      "learning_rate": 1.4714881273176035e-05,
+      "loss": 0.0108,
+      "step": 7620
+    },
+    {
+      "epoch": 10.838068181818182,
+      "grad_norm": 0.18940120935440063,
+      "learning_rate": 1.4597924696737835e-05,
+      "loss": 0.0104,
+      "step": 7630
+    },
+    {
+      "epoch": 10.852272727272727,
+      "grad_norm": 0.23804594576358795,
+      "learning_rate": 1.4481355271723252e-05,
+      "loss": 0.0114,
+      "step": 7640
+    },
+    {
+      "epoch": 10.866477272727273,
+      "grad_norm": 0.3460429012775421,
+      "learning_rate": 1.4365174272916809e-05,
+      "loss": 0.0101,
+      "step": 7650
+    },
+    {
+      "epoch": 10.880681818181818,
+      "grad_norm": 0.17774467170238495,
+      "learning_rate": 1.4249382970855319e-05,
+      "loss": 0.0101,
+      "step": 7660
+    },
+    {
+      "epoch": 10.894886363636363,
+      "grad_norm": 0.25758641958236694,
+      "learning_rate": 1.4133982631813903e-05,
+      "loss": 0.0085,
+      "step": 7670
+    },
+    {
+      "epoch": 10.909090909090908,
+      "grad_norm": 0.15593396127223969,
+      "learning_rate": 1.4018974517792194e-05,
+      "loss": 0.01,
+      "step": 7680
+    },
+    {
+      "epoch": 10.923295454545455,
+      "grad_norm": 0.1858268827199936,
+      "learning_rate": 1.390435988650048e-05,
+      "loss": 0.011,
+      "step": 7690
+    },
+    {
+      "epoch": 10.9375,
+      "grad_norm": 0.18516603112220764,
+      "learning_rate": 1.3790139991346006e-05,
+      "loss": 0.0085,
+      "step": 7700
+    },
+    {
+      "epoch": 10.951704545454545,
+      "grad_norm": 0.1319231539964676,
+      "learning_rate": 1.367631608141926e-05,
+      "loss": 0.0095,
+      "step": 7710
+    },
+    {
+      "epoch": 10.965909090909092,
+      "grad_norm": 0.17596793174743652,
+      "learning_rate": 1.3562889401480278e-05,
+      "loss": 0.0086,
+      "step": 7720
+    },
+    {
+      "epoch": 10.980113636363637,
+      "grad_norm": 0.20282034575939178,
+      "learning_rate": 1.3449861191945074e-05,
+      "loss": 0.0115,
+      "step": 7730
+    },
+    {
+      "epoch": 10.994318181818182,
+      "grad_norm": 0.14738187193870544,
+      "learning_rate": 1.3337232688872009e-05,
+      "loss": 0.0108,
+      "step": 7740
+    },
+    {
+      "epoch": 11.008522727272727,
+      "grad_norm": 0.14532636106014252,
+      "learning_rate": 1.3225005123948364e-05,
+      "loss": 0.0098,
+      "step": 7750
+    },
+    {
+      "epoch": 11.022727272727273,
+      "grad_norm": 0.163893461227417,
+      "learning_rate": 1.311317972447681e-05,
+      "loss": 0.0111,
+      "step": 7760
+    },
+    {
+      "epoch": 11.036931818181818,
+      "grad_norm": 0.2067846655845642,
+      "learning_rate": 1.3001757713361996e-05,
+      "loss": 0.0092,
+      "step": 7770
+    },
+    {
+      "epoch": 11.051136363636363,
+      "grad_norm": 0.26071372628211975,
+      "learning_rate": 1.2890740309097204e-05,
+      "loss": 0.0112,
+      "step": 7780
+    },
+    {
+      "epoch": 11.065340909090908,
+      "grad_norm": 0.15502169728279114,
+      "learning_rate": 1.2780128725750944e-05,
+      "loss": 0.0098,
+      "step": 7790
+    },
+    {
+      "epoch": 11.079545454545455,
+      "grad_norm": 0.1900547295808792,
+      "learning_rate": 1.266992417295379e-05,
+      "loss": 0.0099,
+      "step": 7800
+    },
+    {
+      "epoch": 11.09375,
+      "grad_norm": 0.2872955799102783,
+      "learning_rate": 1.2560127855885073e-05,
+      "loss": 0.0107,
+      "step": 7810
+    },
+    {
+      "epoch": 11.107954545454545,
+      "grad_norm": 0.2135738581418991,
+      "learning_rate": 1.2450740975259745e-05,
+      "loss": 0.0085,
+      "step": 7820
+    },
+    {
+      "epoch": 11.122159090909092,
+      "grad_norm": 0.1478910744190216,
+      "learning_rate": 1.234176472731517e-05,
+      "loss": 0.0118,
+      "step": 7830
+    },
+    {
+      "epoch": 11.136363636363637,
+      "grad_norm": 0.16231344640254974,
+      "learning_rate": 1.2233200303798158e-05,
+      "loss": 0.0154,
+      "step": 7840
+    },
+    {
+      "epoch": 11.150568181818182,
+      "grad_norm": 0.20255939662456512,
+      "learning_rate": 1.2125048891951846e-05,
+      "loss": 0.0126,
+      "step": 7850
+    },
+    {
+      "epoch": 11.164772727272727,
+      "grad_norm": 0.1539156585931778,
+      "learning_rate": 1.2017311674502745e-05,
+      "loss": 0.0118,
+      "step": 7860
+    },
+    {
+      "epoch": 11.178977272727273,
+      "grad_norm": 0.32469236850738525,
+      "learning_rate": 1.1909989829647822e-05,
+      "loss": 0.0103,
+      "step": 7870
+    },
+    {
+      "epoch": 11.193181818181818,
+      "grad_norm": 0.23897293210029602,
+      "learning_rate": 1.1803084531041553e-05,
+      "loss": 0.0092,
+      "step": 7880
+    },
+    {
+      "epoch": 11.207386363636363,
+      "grad_norm": 0.24638468027114868,
+      "learning_rate": 1.1696596947783162e-05,
+      "loss": 0.0112,
+      "step": 7890
+    },
+    {
+      "epoch": 11.221590909090908,
+      "grad_norm": 0.2779979109764099,
+      "learning_rate": 1.1590528244403803e-05,
+      "loss": 0.0115,
+      "step": 7900
+    },
+    {
+      "epoch": 11.235795454545455,
+      "grad_norm": 0.2672639489173889,
+      "learning_rate": 1.148487958085382e-05,
+      "loss": 0.0087,
+      "step": 7910
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 0.24197432398796082,
+      "learning_rate": 1.1379652112490086e-05,
+      "loss": 0.0104,
+      "step": 7920
+    },
+    {
+      "epoch": 11.264204545454545,
+      "grad_norm": 0.22622835636138916,
+      "learning_rate": 1.1274846990063315e-05,
+      "loss": 0.0108,
+      "step": 7930
+    },
+    {
+      "epoch": 11.278409090909092,
+      "grad_norm": 0.16413919627666473,
+      "learning_rate": 1.117046535970554e-05,
+      "loss": 0.0116,
+      "step": 7940
+    },
+    {
+      "epoch": 11.292613636363637,
+      "grad_norm": 0.29245343804359436,
+      "learning_rate": 1.106650836291755e-05,
+      "loss": 0.011,
+      "step": 7950
+    },
+    {
+      "epoch": 11.306818181818182,
+      "grad_norm": 0.22967256605625153,
+      "learning_rate": 1.0962977136556418e-05,
+      "loss": 0.0151,
+      "step": 7960
+    },
+    {
+      "epoch": 11.321022727272727,
+      "grad_norm": 0.17389236390590668,
+      "learning_rate": 1.0859872812823024e-05,
+      "loss": 0.0151,
+      "step": 7970
+    },
+    {
+      "epoch": 11.335227272727273,
+      "grad_norm": 0.15689431130886078,
+      "learning_rate": 1.0757196519249747e-05,
+      "loss": 0.0101,
+      "step": 7980
+    },
+    {
+      "epoch": 11.349431818181818,
+      "grad_norm": 0.22899186611175537,
+      "learning_rate": 1.0654949378688077e-05,
+      "loss": 0.015,
+      "step": 7990
+    },
+    {
+      "epoch": 11.363636363636363,
+      "grad_norm": 0.5312137007713318,
+      "learning_rate": 1.0553132509296376e-05,
+      "loss": 0.0106,
+      "step": 8000
+    },
+    {
+      "epoch": 11.377840909090908,
+      "grad_norm": 0.18518638610839844,
+      "learning_rate": 1.0451747024527613e-05,
+      "loss": 0.0096,
+      "step": 8010
+    },
+    {
+      "epoch": 11.392045454545455,
+      "grad_norm": 0.1725063920021057,
+      "learning_rate": 1.0350794033117189e-05,
+      "loss": 0.0106,
+      "step": 8020
+    },
+    {
+      "epoch": 11.40625,
+      "grad_norm": 0.2230728715658188,
+      "learning_rate": 1.0250274639070856e-05,
+      "loss": 0.0113,
+      "step": 8030
+    },
+    {
+      "epoch": 11.420454545454545,
+      "grad_norm": 0.14705055952072144,
+      "learning_rate": 1.0150189941652599e-05,
+      "loss": 0.0099,
+      "step": 8040
+    },
+    {
+      "epoch": 11.434659090909092,
+      "grad_norm": 0.5630120038986206,
+      "learning_rate": 1.0050541035372635e-05,
+      "loss": 0.01,
+      "step": 8050
+    },
+    {
+      "epoch": 11.448863636363637,
+      "grad_norm": 0.32011404633522034,
+      "learning_rate": 9.951329009975458e-06,
+      "loss": 0.0088,
+      "step": 8060
+    },
+    {
+      "epoch": 11.463068181818182,
+      "grad_norm": 0.2796994745731354,
+      "learning_rate": 9.852554950427845e-06,
+      "loss": 0.0117,
+      "step": 8070
+    },
+    {
+      "epoch": 11.477272727272727,
+      "grad_norm": 0.2308991700410843,
+      "learning_rate": 9.754219936907105e-06,
+      "loss": 0.0079,
+      "step": 8080
+    },
+    {
+      "epoch": 11.491477272727273,
+      "grad_norm": 0.11351477354764938,
+      "learning_rate": 9.656325044789194e-06,
+      "loss": 0.0082,
+      "step": 8090
+    },
+    {
+      "epoch": 11.505681818181818,
+      "grad_norm": 0.17560327053070068,
+      "learning_rate": 9.55887134463697e-06,
+      "loss": 0.0081,
+      "step": 8100
+    },
+    {
+      "epoch": 11.519886363636363,
+      "grad_norm": 0.17340019345283508,
+      "learning_rate": 9.461859902188475e-06,
+      "loss": 0.0071,
+      "step": 8110
+    },
+    {
+      "epoch": 11.534090909090908,
+      "grad_norm": 0.2049873173236847,
+      "learning_rate": 9.365291778345303e-06,
+      "loss": 0.0092,
+      "step": 8120
+    },
+    {
+      "epoch": 11.548295454545455,
+      "grad_norm": 0.14400926232337952,
+      "learning_rate": 9.269168029160991e-06,
+      "loss": 0.0089,
+      "step": 8130
+    },
+    {
+      "epoch": 11.5625,
+      "grad_norm": 0.16723710298538208,
+      "learning_rate": 9.173489705829447e-06,
+      "loss": 0.0092,
+      "step": 8140
+    },
+    {
+      "epoch": 11.576704545454545,
+      "grad_norm": 0.16251735389232635,
+      "learning_rate": 9.078257854673516e-06,
+      "loss": 0.0101,
+      "step": 8150
+    },
+    {
+      "epoch": 11.590909090909092,
+      "grad_norm": 0.14746157824993134,
+      "learning_rate": 8.983473517133429e-06,
+      "loss": 0.0084,
+      "step": 8160
+    },
+    {
+      "epoch": 11.605113636363637,
+      "grad_norm": 0.14881163835525513,
+      "learning_rate": 8.889137729755537e-06,
+      "loss": 0.0085,
+      "step": 8170
+    },
+    {
+      "epoch": 11.619318181818182,
+      "grad_norm": 0.1976151317358017,
+      "learning_rate": 8.79525152418087e-06,
+      "loss": 0.0088,
+      "step": 8180
+    },
+    {
+      "epoch": 11.633522727272727,
+      "grad_norm": 0.3984404504299164,
+      "learning_rate": 8.701815927133961e-06,
+      "loss": 0.0105,
+      "step": 8190
+    },
+    {
+      "epoch": 11.647727272727273,
+      "grad_norm": 0.13391618430614471,
+      "learning_rate": 8.608831960411534e-06,
+      "loss": 0.0068,
+      "step": 8200
+    },
+    {
+      "epoch": 11.661931818181818,
+      "grad_norm": 0.12304878234863281,
+      "learning_rate": 8.516300640871321e-06,
+      "loss": 0.0081,
+      "step": 8210
+    },
+    {
+      "epoch": 11.676136363636363,
+      "grad_norm": 0.11924493312835693,
+      "learning_rate": 8.424222980421038e-06,
+      "loss": 0.014,
+      "step": 8220
+    },
+    {
+      "epoch": 11.690340909090908,
+      "grad_norm": 0.23459115624427795,
+      "learning_rate": 8.332599986007184e-06,
+      "loss": 0.0092,
+      "step": 8230
+    },
+    {
+      "epoch": 11.704545454545455,
+      "grad_norm": 0.5544418692588806,
+      "learning_rate": 8.241432659604203e-06,
+      "loss": 0.0091,
+      "step": 8240
+    },
+    {
+      "epoch": 11.71875,
+      "grad_norm": 0.34650495648384094,
+      "learning_rate": 8.150721998203331e-06,
+      "loss": 0.0094,
+      "step": 8250
+    },
+    {
+      "epoch": 11.732954545454545,
+      "grad_norm": 0.24444520473480225,
+      "learning_rate": 8.06046899380184e-06,
+      "loss": 0.0081,
+      "step": 8260
+    },
+    {
+      "epoch": 11.747159090909092,
+      "grad_norm": 0.1377067267894745,
+      "learning_rate": 7.970674633392133e-06,
+      "loss": 0.0086,
+      "step": 8270
+    },
+    {
+      "epoch": 11.761363636363637,
+      "grad_norm": 0.19450236856937408,
+      "learning_rate": 7.881339898950924e-06,
+      "loss": 0.0127,
+      "step": 8280
+    },
+    {
+      "epoch": 11.775568181818182,
+      "grad_norm": 0.18926279246807098,
+      "learning_rate": 7.792465767428597e-06,
+      "loss": 0.0099,
+      "step": 8290
+    },
+    {
+      "epoch": 11.789772727272727,
+      "grad_norm": 0.11233865469694138,
+      "learning_rate": 7.704053210738376e-06,
+      "loss": 0.0099,
+      "step": 8300
+    },
+    {
+      "epoch": 11.803977272727273,
+      "grad_norm": 0.24110765755176544,
+      "learning_rate": 7.6161031957458494e-06,
+      "loss": 0.0076,
+      "step": 8310
+    },
+    {
+      "epoch": 11.818181818181818,
+      "grad_norm": 0.20746886730194092,
+      "learning_rate": 7.5286166842582605e-06,
+      "loss": 0.0123,
+      "step": 8320
+    },
+    {
+      "epoch": 11.832386363636363,
+      "grad_norm": 0.28823456168174744,
+      "learning_rate": 7.4415946330140814e-06,
+      "loss": 0.0108,
+      "step": 8330
+    },
+    {
+      "epoch": 11.846590909090908,
+      "grad_norm": 0.15231630206108093,
+      "learning_rate": 7.3550379936725644e-06,
+      "loss": 0.0101,
+      "step": 8340
+    },
+    {
+      "epoch": 11.860795454545455,
+      "grad_norm": 0.1749686896800995,
+      "learning_rate": 7.2689477128032035e-06,
+      "loss": 0.0086,
+      "step": 8350
+    },
+    {
+      "epoch": 11.875,
+      "grad_norm": 0.22606955468654633,
+      "learning_rate": 7.183324731875551e-06,
+      "loss": 0.0088,
+      "step": 8360
+    },
+    {
+      "epoch": 11.889204545454545,
+      "grad_norm": 0.16468025743961334,
+      "learning_rate": 7.098169987248782e-06,
+      "loss": 0.0083,
+      "step": 8370
+    },
+    {
+      "epoch": 11.903409090909092,
+      "grad_norm": 0.18283675611019135,
+      "learning_rate": 7.013484410161553e-06,
+      "loss": 0.0079,
+      "step": 8380
+    },
+    {
+      "epoch": 11.917613636363637,
+      "grad_norm": 0.2217361181974411,
+      "learning_rate": 6.92926892672176e-06,
+      "loss": 0.01,
+      "step": 8390
+    },
+    {
+      "epoch": 11.931818181818182,
+      "grad_norm": 0.20115987956523895,
+      "learning_rate": 6.845524457896446e-06,
+      "loss": 0.0102,
+      "step": 8400
+    },
+    {
+      "epoch": 11.946022727272727,
+      "grad_norm": 0.1755692958831787,
+      "learning_rate": 6.7622519195017165e-06,
+      "loss": 0.0061,
+      "step": 8410
+    },
+    {
+      "epoch": 11.960227272727273,
+      "grad_norm": 0.2037981003522873,
+      "learning_rate": 6.679452222192684e-06,
+      "loss": 0.0086,
+      "step": 8420
+    },
+    {
+      "epoch": 11.974431818181818,
+      "grad_norm": 0.16050030291080475,
+      "learning_rate": 6.597126271453579e-06,
+      "loss": 0.0069,
+      "step": 8430
+    },
+    {
+      "epoch": 11.988636363636363,
+      "grad_norm": 0.2133057713508606,
+      "learning_rate": 6.51527496758782e-06,
+      "loss": 0.0089,
+      "step": 8440
+    },
+    {
+      "epoch": 12.002840909090908,
+      "grad_norm": 0.12774421274662018,
+      "learning_rate": 6.433899205708155e-06,
+      "loss": 0.007,
+      "step": 8450
+    },
+    {
+      "epoch": 12.017045454545455,
+      "grad_norm": 0.22030217945575714,
+      "learning_rate": 6.352999875726856e-06,
+      "loss": 0.0079,
+      "step": 8460
+    },
+    {
+      "epoch": 12.03125,
+      "grad_norm": 0.1830436885356903,
+      "learning_rate": 6.272577862346052e-06,
+      "loss": 0.0081,
+      "step": 8470
+    },
+    {
+      "epoch": 12.045454545454545,
+      "grad_norm": 0.23821671307086945,
+      "learning_rate": 6.192634045047996e-06,
+      "loss": 0.0072,
+      "step": 8480
+    },
+    {
+      "epoch": 12.059659090909092,
+      "grad_norm": 0.097575344145298,
+      "learning_rate": 6.113169298085458e-06,
+      "loss": 0.0124,
+      "step": 8490
+    },
+    {
+      "epoch": 12.073863636363637,
+      "grad_norm": 0.13037890195846558,
+      "learning_rate": 6.034184490472195e-06,
+      "loss": 0.0112,
+      "step": 8500
+    },
+    {
+      "epoch": 12.088068181818182,
+      "grad_norm": 0.17663180828094482,
+      "learning_rate": 5.955680485973386e-06,
+      "loss": 0.0068,
+      "step": 8510
+    },
+    {
+      "epoch": 12.102272727272727,
+      "grad_norm": 0.2542382478713989,
+      "learning_rate": 5.877658143096265e-06,
+      "loss": 0.0087,
+      "step": 8520
+    },
+    {
+      "epoch": 12.116477272727273,
+      "grad_norm": 0.23478275537490845,
+      "learning_rate": 5.800118315080661e-06,
+      "loss": 0.0088,
+      "step": 8530
+    },
+    {
+      "epoch": 12.130681818181818,
+      "grad_norm": 0.1812600940465927,
+      "learning_rate": 5.723061849889716e-06,
+      "loss": 0.008,
+      "step": 8540
+    },
+    {
+      "epoch": 12.144886363636363,
+      "grad_norm": 0.21002638339996338,
+      "learning_rate": 5.646489590200604e-06,
+      "loss": 0.0081,
+      "step": 8550
+    },
+    {
+      "epoch": 12.159090909090908,
+      "grad_norm": 0.1361236870288849,
+      "learning_rate": 5.570402373395256e-06,
+      "loss": 0.0078,
+      "step": 8560
+    },
+    {
+      "epoch": 12.173295454545455,
+      "grad_norm": 0.1938331425189972,
+      "learning_rate": 5.494801031551305e-06,
+      "loss": 0.0074,
+      "step": 8570
+    },
+    {
+      "epoch": 12.1875,
+      "grad_norm": 0.14236967265605927,
+      "learning_rate": 5.41968639143291e-06,
+      "loss": 0.0111,
+      "step": 8580
+    },
+    {
+      "epoch": 12.201704545454545,
+      "grad_norm": 0.16860653460025787,
+      "learning_rate": 5.345059274481751e-06,
+      "loss": 0.0058,
+      "step": 8590
+    },
+    {
+      "epoch": 12.215909090909092,
+      "grad_norm": 0.16406631469726562,
+      "learning_rate": 5.270920496808002e-06,
+      "loss": 0.0072,
+      "step": 8600
+    },
+    {
+      "epoch": 12.230113636363637,
+      "grad_norm": 0.21259747445583344,
+      "learning_rate": 5.1972708691814695e-06,
+      "loss": 0.0109,
+      "step": 8610
+    },
+    {
+      "epoch": 12.244318181818182,
+      "grad_norm": 0.2537532150745392,
+      "learning_rate": 5.124111197022674e-06,
+      "loss": 0.0101,
+      "step": 8620
+    },
+    {
+      "epoch": 12.258522727272727,
+      "grad_norm": 0.14652734994888306,
+      "learning_rate": 5.051442280394081e-06,
+      "loss": 0.0084,
+      "step": 8630
+    },
+    {
+      "epoch": 12.272727272727273,
+      "grad_norm": 0.2256622463464737,
+      "learning_rate": 4.979264913991322e-06,
+      "loss": 0.0092,
+      "step": 8640
+    },
+    {
+      "epoch": 12.286931818181818,
+      "grad_norm": 0.1760641634464264,
+      "learning_rate": 4.907579887134489e-06,
+      "loss": 0.0075,
+      "step": 8650
+    },
+    {
+      "epoch": 12.301136363636363,
+      "grad_norm": 0.14924417436122894,
+      "learning_rate": 4.836387983759572e-06,
+      "loss": 0.011,
+      "step": 8660
+    },
+    {
+      "epoch": 12.315340909090908,
+      "grad_norm": 0.15925931930541992,
+      "learning_rate": 4.765689982409816e-06,
+      "loss": 0.008,
+      "step": 8670
+    },
+    {
+      "epoch": 12.329545454545455,
+      "grad_norm": 0.14156000316143036,
+      "learning_rate": 4.695486656227233e-06,
+      "loss": 0.0083,
+      "step": 8680
+    },
+    {
+      "epoch": 12.34375,
+      "grad_norm": 0.13891640305519104,
+      "learning_rate": 4.625778772944156e-06,
+      "loss": 0.0092,
+      "step": 8690
+    },
+    {
+      "epoch": 12.357954545454545,
+      "grad_norm": 0.11400335282087326,
+      "learning_rate": 4.556567094874825e-06,
+      "loss": 0.0082,
+      "step": 8700
+    },
+    {
+      "epoch": 12.372159090909092,
+      "grad_norm": 0.16429784893989563,
+      "learning_rate": 4.487852378907059e-06,
+      "loss": 0.0093,
+      "step": 8710
+    },
+    {
+      "epoch": 12.386363636363637,
+      "grad_norm": 0.10111011564731598,
+      "learning_rate": 4.419635376493986e-06,
+      "loss": 0.0064,
+      "step": 8720
+    },
+    {
+      "epoch": 12.400568181818182,
+      "grad_norm": 0.16164110600948334,
+      "learning_rate": 4.351916833645825e-06,
+      "loss": 0.0084,
+      "step": 8730
+    },
+    {
+      "epoch": 12.414772727272727,
+      "grad_norm": 0.1821807324886322,
+      "learning_rate": 4.284697490921691e-06,
+      "loss": 0.0085,
+      "step": 8740
+    },
+    {
+      "epoch": 12.428977272727273,
+      "grad_norm": 0.13237673044204712,
+      "learning_rate": 4.2179780834215585e-06,
+      "loss": 0.008,
+      "step": 8750
+    },
+    {
+      "epoch": 12.443181818181818,
+      "grad_norm": 0.2113754153251648,
+      "learning_rate": 4.151759340778178e-06,
+      "loss": 0.0098,
+      "step": 8760
+    },
+    {
+      "epoch": 12.457386363636363,
+      "grad_norm": 0.10824495553970337,
+      "learning_rate": 4.086041987149109e-06,
+      "loss": 0.009,
+      "step": 8770
+    },
+    {
+      "epoch": 12.471590909090908,
+      "grad_norm": 0.15280917286872864,
+      "learning_rate": 4.020826741208811e-06,
+      "loss": 0.008,
+      "step": 8780
+    },
+    {
+      "epoch": 12.485795454545455,
+      "grad_norm": 0.28365927934646606,
+      "learning_rate": 3.956114316140746e-06,
+      "loss": 0.0095,
+      "step": 8790
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 0.2627256512641907,
+      "learning_rate": 3.891905419629643e-06,
+      "loss": 0.0093,
+      "step": 8800
+    },
+    {
+      "epoch": 12.514204545454545,
+      "grad_norm": 0.1927265226840973,
+      "learning_rate": 3.8282007538536946e-06,
+      "loss": 0.0114,
+      "step": 8810
+    },
+    {
+      "epoch": 12.528409090909092,
+      "grad_norm": 0.09738240391016006,
+      "learning_rate": 3.7650010154769265e-06,
+      "loss": 0.0076,
+      "step": 8820
+    },
+    {
+      "epoch": 12.542613636363637,
+      "grad_norm": 0.11914301663637161,
+      "learning_rate": 3.7023068956415608e-06,
+      "loss": 0.0103,
+      "step": 8830
+    },
+    {
+      "epoch": 12.556818181818182,
+      "grad_norm": 0.19162027537822723,
+      "learning_rate": 3.6401190799604303e-06,
+      "loss": 0.0095,
+      "step": 8840
+    },
+    {
+      "epoch": 12.571022727272727,
+      "grad_norm": 0.1131158098578453,
+      "learning_rate": 3.578438248509536e-06,
+      "loss": 0.0069,
+      "step": 8850
+    },
+    {
+      "epoch": 12.585227272727273,
+      "grad_norm": 0.10243741422891617,
+      "learning_rate": 3.5172650758205583e-06,
+      "loss": 0.0093,
+      "step": 8860
+    },
+    {
+      "epoch": 12.599431818181818,
+      "grad_norm": 0.2540527284145355,
+      "learning_rate": 3.45660023087353e-06,
+      "loss": 0.011,
+      "step": 8870
+    },
+    {
+      "epoch": 12.613636363636363,
+      "grad_norm": 0.19322006404399872,
+      "learning_rate": 3.3964443770894528e-06,
+      "loss": 0.0084,
+      "step": 8880
+    },
+    {
+      "epoch": 12.627840909090908,
+      "grad_norm": 0.14868277311325073,
+      "learning_rate": 3.3367981723231245e-06,
+      "loss": 0.0099,
+      "step": 8890
+    },
+    {
+      "epoch": 12.642045454545455,
+      "grad_norm": 0.12972506880760193,
+      "learning_rate": 3.2776622688558746e-06,
+      "loss": 0.0076,
+      "step": 8900
+    },
+    {
+      "epoch": 12.65625,
+      "grad_norm": 0.1375243216753006,
+      "learning_rate": 3.2190373133884677e-06,
+      "loss": 0.0119,
+      "step": 8910
+    },
+    {
+      "epoch": 12.670454545454545,
+      "grad_norm": 0.13518275320529938,
+      "learning_rate": 3.1609239470340446e-06,
+      "loss": 0.0066,
+      "step": 8920
+    },
+    {
+      "epoch": 12.684659090909092,
+      "grad_norm": 0.43873974680900574,
+      "learning_rate": 3.1033228053110373e-06,
+      "loss": 0.0076,
+      "step": 8930
+    },
+    {
+      "epoch": 12.698863636363637,
+      "grad_norm": 0.26058143377304077,
+      "learning_rate": 3.0462345181363314e-06,
+      "loss": 0.0065,
+      "step": 8940
+    },
+    {
+      "epoch": 12.713068181818182,
+      "grad_norm": 0.11189442873001099,
+      "learning_rate": 2.9896597098182654e-06,
+      "loss": 0.0071,
+      "step": 8950
+    },
+    {
+      "epoch": 12.727272727272727,
+      "grad_norm": 0.1724548488855362,
+      "learning_rate": 2.933598999049891e-06,
+      "loss": 0.0101,
+      "step": 8960
+    },
+    {
+      "epoch": 12.741477272727273,
+      "grad_norm": 0.19990795850753784,
+      "learning_rate": 2.8780529989021697e-06,
+      "loss": 0.0085,
+      "step": 8970
+    },
+    {
+      "epoch": 12.755681818181818,
+      "grad_norm": 0.15807050466537476,
+      "learning_rate": 2.823022316817242e-06,
+      "loss": 0.0076,
+      "step": 8980
+    },
+    {
+      "epoch": 12.769886363636363,
+      "grad_norm": 0.17358310520648956,
+      "learning_rate": 2.7685075546018456e-06,
+      "loss": 0.0113,
+      "step": 8990
+    },
+    {
+      "epoch": 12.784090909090908,
+      "grad_norm": 0.20097093284130096,
+      "learning_rate": 2.7145093084206598e-06,
+      "loss": 0.0083,
+      "step": 9000
+    },
+    {
+      "epoch": 12.798295454545455,
+      "grad_norm": 0.116364024579525,
+      "learning_rate": 2.661028168789892e-06,
+      "loss": 0.0102,
+      "step": 9010
+    },
+    {
+      "epoch": 12.8125,
+      "grad_norm": 0.2691571116447449,
+      "learning_rate": 2.6080647205706855e-06,
+      "loss": 0.0096,
+      "step": 9020
+    },
+    {
+      "epoch": 12.826704545454545,
+      "grad_norm": 0.17095306515693665,
+      "learning_rate": 2.555619542962834e-06,
+      "loss": 0.0071,
+      "step": 9030
+    },
+    {
+      "epoch": 12.840909090909092,
+      "grad_norm": 0.08522237837314606,
+      "learning_rate": 2.503693209498409e-06,
+      "loss": 0.0078,
+      "step": 9040
+    },
+    {
+      "epoch": 12.855113636363637,
+      "grad_norm": 0.16366495192050934,
+      "learning_rate": 2.452286288035449e-06,
+      "loss": 0.0069,
+      "step": 9050
+    },
+    {
+      "epoch": 12.869318181818182,
+      "grad_norm": 0.1985977739095688,
+      "learning_rate": 2.4013993407518363e-06,
+      "loss": 0.0078,
+      "step": 9060
+    },
+    {
+      "epoch": 12.883522727272727,
+      "grad_norm": 0.125263050198555,
+      "learning_rate": 2.351032924139063e-06,
+      "loss": 0.0089,
+      "step": 9070
+    },
+    {
+      "epoch": 12.897727272727273,
+      "grad_norm": 0.4779665470123291,
+      "learning_rate": 2.30118758899619e-06,
+      "loss": 0.0083,
+      "step": 9080
+    },
+    {
+      "epoch": 12.911931818181818,
+      "grad_norm": 0.118967205286026,
+      "learning_rate": 2.2518638804238157e-06,
+      "loss": 0.0088,
+      "step": 9090
+    },
+    {
+      "epoch": 12.926136363636363,
+      "grad_norm": 0.13806115090847015,
+      "learning_rate": 2.203062337818118e-06,
+      "loss": 0.0075,
+      "step": 9100
+    },
+    {
+      "epoch": 12.940340909090908,
+      "grad_norm": 0.15037260949611664,
+      "learning_rate": 2.1547834948649483e-06,
+      "loss": 0.0081,
+      "step": 9110
+    },
+    {
+      "epoch": 12.954545454545455,
+      "grad_norm": 0.23373590409755707,
+      "learning_rate": 2.1070278795340017e-06,
+      "loss": 0.0099,
+      "step": 9120
+    },
+    {
+      "epoch": 12.96875,
+      "grad_norm": 0.25523579120635986,
+      "learning_rate": 2.059796014073029e-06,
+      "loss": 0.0077,
+      "step": 9130
+    },
+    {
+      "epoch": 12.982954545454545,
+      "grad_norm": 0.10693769156932831,
+      "learning_rate": 2.01308841500214e-06,
+      "loss": 0.0065,
+      "step": 9140
+    },
+    {
+      "epoch": 12.997159090909092,
+      "grad_norm": 0.07628029584884644,
+      "learning_rate": 1.9669055931081704e-06,
+      "loss": 0.0088,
+      "step": 9150
+    },
+    {
+      "epoch": 13.011363636363637,
+      "grad_norm": 0.29461079835891724,
+      "learning_rate": 1.9212480534390507e-06,
+      "loss": 0.012,
+      "step": 9160
+    },
+    {
+      "epoch": 13.025568181818182,
+      "grad_norm": 0.35560137033462524,
+      "learning_rate": 1.8761162952983246e-06,
+      "loss": 0.0082,
+      "step": 9170
+    },
+    {
+      "epoch": 13.039772727272727,
+      "grad_norm": 0.19855539500713348,
+      "learning_rate": 1.8315108122396618e-06,
+      "loss": 0.0109,
+      "step": 9180
+    },
+    {
+      "epoch": 13.053977272727273,
+      "grad_norm": 0.15490369498729706,
+      "learning_rate": 1.787432092061475e-06,
+      "loss": 0.0079,
+      "step": 9190
+    },
+    {
+      "epoch": 13.068181818181818,
+      "grad_norm": 0.23204120993614197,
+      "learning_rate": 1.743880616801602e-06,
+      "loss": 0.0085,
+      "step": 9200
+    },
+    {
+      "epoch": 13.082386363636363,
+      "grad_norm": 0.13382819294929504,
+      "learning_rate": 1.7008568627319865e-06,
+      "loss": 0.0094,
+      "step": 9210
+    },
+    {
+      "epoch": 13.096590909090908,
+      "grad_norm": 0.18071945011615753,
+      "learning_rate": 1.6583613003535226e-06,
+      "loss": 0.0094,
+      "step": 9220
+    },
+    {
+      "epoch": 13.110795454545455,
+      "grad_norm": 0.16985177993774414,
+      "learning_rate": 1.6163943943908522e-06,
+      "loss": 0.0096,
+      "step": 9230
+    },
+    {
+      "epoch": 13.125,
+      "grad_norm": 0.20578616857528687,
+      "learning_rate": 1.5749566037873476e-06,
+      "loss": 0.009,
+      "step": 9240
+    },
+    {
+      "epoch": 13.139204545454545,
+      "grad_norm": 0.2541482150554657,
+      "learning_rate": 1.5340483817000428e-06,
+      "loss": 0.0101,
+      "step": 9250
+    },
+    {
+      "epoch": 13.153409090909092,
+      "grad_norm": 0.17913727462291718,
+      "learning_rate": 1.4936701754947101e-06,
+      "loss": 0.0084,
+      "step": 9260
+    },
+    {
+      "epoch": 13.167613636363637,
+      "grad_norm": 0.1537134051322937,
+      "learning_rate": 1.4538224267409361e-06,
+      "loss": 0.0075,
+      "step": 9270
+    },
+    {
+      "epoch": 13.181818181818182,
+      "grad_norm": 0.28880417346954346,
+      "learning_rate": 1.414505571207314e-06,
+      "loss": 0.0089,
+      "step": 9280
+    },
+    {
+      "epoch": 13.196022727272727,
+      "grad_norm": 0.09422419965267181,
+      "learning_rate": 1.3757200388566816e-06,
+      "loss": 0.0062,
+      "step": 9290
+    },
+    {
+      "epoch": 13.210227272727273,
+      "grad_norm": 0.26506584882736206,
+      "learning_rate": 1.3374662538414074e-06,
+      "loss": 0.0065,
+      "step": 9300
+    },
+    {
+      "epoch": 13.224431818181818,
+      "grad_norm": 0.17890435457229614,
+      "learning_rate": 1.2997446344987617e-06,
+      "loss": 0.0101,
+      "step": 9310
+    },
+    {
+      "epoch": 13.238636363636363,
+      "grad_norm": 0.18489724397659302,
+      "learning_rate": 1.262555593346315e-06,
+      "loss": 0.0078,
+      "step": 9320
+    },
+    {
+      "epoch": 13.252840909090908,
+      "grad_norm": 0.17822150886058807,
+      "learning_rate": 1.2258995370774685e-06,
+      "loss": 0.0079,
+      "step": 9330
+    },
+    {
+      "epoch": 13.267045454545455,
+      "grad_norm": 0.20881354808807373,
+      "learning_rate": 1.1897768665569798e-06,
+      "loss": 0.0105,
+      "step": 9340
+    },
+    {
+      "epoch": 13.28125,
+      "grad_norm": 0.1700887233018875,
+      "learning_rate": 1.1541879768165954e-06,
+      "loss": 0.0088,
+      "step": 9350
+    },
+    {
+      "epoch": 13.295454545454545,
+      "grad_norm": 0.1321496069431305,
+      "learning_rate": 1.1191332570507085e-06,
+      "loss": 0.0072,
+      "step": 9360
+    },
+    {
+      "epoch": 13.309659090909092,
+      "grad_norm": 0.3111177980899811,
+      "learning_rate": 1.0846130906121132e-06,
+      "loss": 0.0087,
+      "step": 9370
+    },
+    {
+      "epoch": 13.323863636363637,
+      "grad_norm": 0.1439957171678543,
+      "learning_rate": 1.0506278550078131e-06,
+      "loss": 0.01,
+      "step": 9380
+    },
+    {
+      "epoch": 13.338068181818182,
+      "grad_norm": 0.3624275326728821,
+      "learning_rate": 1.0171779218949185e-06,
+      "loss": 0.0101,
+      "step": 9390
+    },
+    {
+      "epoch": 13.352272727272727,
+      "grad_norm": 0.2051062434911728,
+      "learning_rate": 9.842636570765174e-07,
+      "loss": 0.0105,
+      "step": 9400
+    },
+    {
+      "epoch": 13.366477272727273,
+      "grad_norm": 0.3168241083621979,
+      "learning_rate": 9.518854204977612e-07,
+      "loss": 0.0078,
+      "step": 9410
+    },
+    {
+      "epoch": 13.380681818181818,
+      "grad_norm": 0.1734011024236679,
+      "learning_rate": 9.200435662418349e-07,
+      "loss": 0.0081,
+      "step": 9420
+    },
+    {
+      "epoch": 13.394886363636363,
+      "grad_norm": 0.18532846868038177,
+      "learning_rate": 8.887384425261658e-07,
+      "loss": 0.0073,
+      "step": 9430
+    },
+    {
+      "epoch": 13.409090909090908,
+      "grad_norm": 0.16800567507743835,
+      "learning_rate": 8.579703916985648e-07,
+      "loss": 0.0083,
+      "step": 9440
+    },
+    {
+      "epoch": 13.423295454545455,
+      "grad_norm": 0.1704150140285492,
+      "learning_rate": 8.277397502335194e-07,
+      "loss": 0.0103,
+      "step": 9450
+    },
+    {
+      "epoch": 13.4375,
+      "grad_norm": 0.10577347129583359,
+      "learning_rate": 7.980468487284675e-07,
+      "loss": 0.0058,
+      "step": 9460
+    },
+    {
+      "epoch": 13.451704545454545,
+      "grad_norm": 0.13024316728115082,
+      "learning_rate": 7.688920119002297e-07,
+      "loss": 0.0088,
+      "step": 9470
+    },
+    {
+      "epoch": 13.465909090909092,
+      "grad_norm": 0.2578657865524292,
+      "learning_rate": 7.402755585814269e-07,
+      "loss": 0.0079,
+      "step": 9480
+    },
+    {
+      "epoch": 13.480113636363637,
+      "grad_norm": 0.11337309330701828,
+      "learning_rate": 7.121978017170073e-07,
+      "loss": 0.0073,
+      "step": 9490
+    },
+    {
+      "epoch": 13.494318181818182,
+      "grad_norm": 0.18533821403980255,
+      "learning_rate": 6.846590483608306e-07,
+      "loss": 0.0112,
+      "step": 9500
+    },
+    {
+      "epoch": 13.508522727272727,
+      "grad_norm": 0.17261147499084473,
+      "learning_rate": 6.576595996722834e-07,
+      "loss": 0.0111,
+      "step": 9510
+    },
+    {
+      "epoch": 13.522727272727273,
+      "grad_norm": 0.1476374715566635,
+      "learning_rate": 6.311997509130141e-07,
+      "loss": 0.0078,
+      "step": 9520
+    },
+    {
+      "epoch": 13.536931818181818,
+      "grad_norm": 0.30872365832328796,
+      "learning_rate": 6.052797914436803e-07,
+      "loss": 0.0105,
+      "step": 9530
+    },
+    {
+      "epoch": 13.551136363636363,
+      "grad_norm": 0.1511140763759613,
+      "learning_rate": 5.799000047208181e-07,
+      "loss": 0.0089,
+      "step": 9540
+    },
+    {
+      "epoch": 13.565340909090908,
+      "grad_norm": 0.15941114723682404,
+      "learning_rate": 5.550606682937054e-07,
+      "loss": 0.0104,
+      "step": 9550
+    },
+    {
+      "epoch": 13.579545454545455,
+      "grad_norm": 0.15345117449760437,
+      "learning_rate": 5.307620538013481e-07,
+      "loss": 0.0084,
+      "step": 9560
+    },
+    {
+      "epoch": 13.59375,
+      "grad_norm": 0.1714678555727005,
+      "learning_rate": 5.070044269694874e-07,
+      "loss": 0.0088,
+      "step": 9570
+    },
+    {
+      "epoch": 13.607954545454545,
+      "grad_norm": 0.20954135060310364,
+      "learning_rate": 4.837880476077417e-07,
+      "loss": 0.0087,
+      "step": 9580
+    },
+    {
+      "epoch": 13.622159090909092,
+      "grad_norm": 0.10831267386674881,
+      "learning_rate": 4.6111316960670835e-07,
+      "loss": 0.0114,
+      "step": 9590
+    },
+    {
+      "epoch": 13.636363636363637,
+      "grad_norm": 0.08259033411741257,
+      "learning_rate": 4.389800409352218e-07,
+      "loss": 0.0063,
+      "step": 9600
+    },
+    {
+      "epoch": 13.650568181818182,
+      "grad_norm": 0.12250098586082458,
+      "learning_rate": 4.173889036376277e-07,
+      "loss": 0.007,
+      "step": 9610
+    },
+    {
+      "epoch": 13.664772727272727,
+      "grad_norm": 0.2607082426548004,
+      "learning_rate": 3.963399938311463e-07,
+      "loss": 0.0095,
+      "step": 9620
+    },
+    {
+      "epoch": 13.678977272727273,
+      "grad_norm": 0.14454485476016998,
+      "learning_rate": 3.7583354170328545e-07,
+      "loss": 0.0082,
+      "step": 9630
+    },
+    {
+      "epoch": 13.693181818181818,
+      "grad_norm": 0.13278765976428986,
+      "learning_rate": 3.558697715093207e-07,
+      "loss": 0.0098,
+      "step": 9640
+    },
+    {
+      "epoch": 13.707386363636363,
+      "grad_norm": 0.12161421775817871,
+      "learning_rate": 3.3644890156983576e-07,
+      "loss": 0.0062,
+      "step": 9650
+    },
+    {
+      "epoch": 13.721590909090908,
+      "grad_norm": 0.17322924733161926,
+      "learning_rate": 3.175711442683638e-07,
+      "loss": 0.0088,
+      "step": 9660
+    },
+    {
+      "epoch": 13.735795454545455,
+      "grad_norm": 0.19804055988788605,
+      "learning_rate": 2.9923670604902197e-07,
+      "loss": 0.0099,
+      "step": 9670
+    },
+    {
+      "epoch": 13.75,
+      "grad_norm": 0.12212982028722763,
+      "learning_rate": 2.814457874143028e-07,
+      "loss": 0.0087,
+      "step": 9680
+    },
+    {
+      "epoch": 13.764204545454545,
+      "grad_norm": 0.21638919413089752,
+      "learning_rate": 2.641985829228366e-07,
+      "loss": 0.0086,
+      "step": 9690
+    },
+    {
+      "epoch": 13.778409090909092,
+      "grad_norm": 0.10085052251815796,
+      "learning_rate": 2.474952811872877e-07,
+      "loss": 0.0063,
+      "step": 9700
+    },
+    {
+      "epoch": 13.792613636363637,
+      "grad_norm": 0.12627679109573364,
+      "learning_rate": 2.3133606487228397e-07,
+      "loss": 0.0078,
+      "step": 9710
+    },
+    {
+      "epoch": 13.806818181818182,
+      "grad_norm": 0.125399649143219,
+      "learning_rate": 2.157211106924295e-07,
+      "loss": 0.0093,
+      "step": 9720
+    },
+    {
+      "epoch": 13.821022727272727,
+      "grad_norm": 0.14021718502044678,
+      "learning_rate": 2.006505894103672e-07,
+      "loss": 0.0089,
+      "step": 9730
+    },
+    {
+      "epoch": 13.835227272727273,
+      "grad_norm": 0.24177104234695435,
+      "learning_rate": 1.8612466583489696e-07,
+      "loss": 0.011,
+      "step": 9740
+    },
+    {
+      "epoch": 13.849431818181818,
+      "grad_norm": 0.15933436155319214,
+      "learning_rate": 1.7214349881918834e-07,
+      "loss": 0.006,
+      "step": 9750
+    },
+    {
+      "epoch": 13.863636363636363,
+      "grad_norm": 0.10176098346710205,
+      "learning_rate": 1.5870724125904845e-07,
+      "loss": 0.0083,
+      "step": 9760
+    },
+    {
+      "epoch": 13.877840909090908,
+      "grad_norm": 0.16704827547073364,
+      "learning_rate": 1.4581604009124006e-07,
+      "loss": 0.0071,
+      "step": 9770
+    },
+    {
+      "epoch": 13.892045454545455,
+      "grad_norm": 0.11706065386533737,
+      "learning_rate": 1.334700362918717e-07,
+      "loss": 0.0078,
+      "step": 9780
+    },
+    {
+      "epoch": 13.90625,
+      "grad_norm": 0.24976439774036407,
+      "learning_rate": 1.2166936487486015e-07,
+      "loss": 0.0079,
+      "step": 9790
+    },
+    {
+      "epoch": 13.920454545454545,
+      "grad_norm": 0.10539773106575012,
+      "learning_rate": 1.1041415489045914e-07,
+      "loss": 0.008,
+      "step": 9800
+    },
+    {
+      "epoch": 13.934659090909092,
+      "grad_norm": 0.12890596687793732,
+      "learning_rate": 9.970452942384412e-08,
+      "loss": 0.0072,
+      "step": 9810
+    },
+    {
+      "epoch": 13.948863636363637,
+      "grad_norm": 0.22905589640140533,
+      "learning_rate": 8.954060559375754e-08,
+      "loss": 0.008,
+      "step": 9820
+    },
+    {
+      "epoch": 13.963068181818182,
+      "grad_norm": 0.20156610012054443,
+      "learning_rate": 7.99224945512489e-08,
+      "loss": 0.0072,
+      "step": 9830
+    },
+    {
+      "epoch": 13.977272727272727,
+      "grad_norm": 0.17252571880817413,
+      "learning_rate": 7.085030147843675e-08,
+      "loss": 0.0105,
+      "step": 9840
+    },
+    {
+      "epoch": 13.991477272727273,
+      "grad_norm": 0.14705359935760498,
+      "learning_rate": 6.232412558736523e-08,
+      "loss": 0.0084,
+      "step": 9850
+    },
+    {
+      "epoch": 14.005681818181818,
+      "grad_norm": 0.2474076747894287,
+      "learning_rate": 5.434406011893822e-08,
+      "loss": 0.0107,
+      "step": 9860
+    },
+    {
+      "epoch": 14.019886363636363,
+      "grad_norm": 0.18310357630252838,
+      "learning_rate": 4.6910192341864664e-08,
+      "loss": 0.01,
+      "step": 9870
+    },
+    {
+      "epoch": 14.034090909090908,
+      "grad_norm": 0.20719188451766968,
+      "learning_rate": 4.0022603551737035e-08,
+      "loss": 0.0068,
+      "step": 9880
+    },
+    {
+      "epoch": 14.048295454545455,
+      "grad_norm": 0.14706085622310638,
+      "learning_rate": 3.3681369070120985e-08,
+      "loss": 0.0085,
+      "step": 9890
+    },
+    {
+      "epoch": 14.0625,
+      "grad_norm": 0.20119525492191315,
+      "learning_rate": 2.7886558243744866e-08,
+      "loss": 0.011,
+      "step": 9900
+    },
+    {
+      "epoch": 14.076704545454545,
+      "grad_norm": 0.14875328540802002,
+      "learning_rate": 2.2638234443722596e-08,
+      "loss": 0.0101,
+      "step": 9910
+    },
+    {
+      "epoch": 14.090909090909092,
+      "grad_norm": 0.22535090148448944,
+      "learning_rate": 1.7936455064887504e-08,
+      "loss": 0.0094,
+      "step": 9920
+    },
+    {
+      "epoch": 14.105113636363637,
+      "grad_norm": 0.2326516956090927,
+      "learning_rate": 1.378127152514841e-08,
+      "loss": 0.0091,
+      "step": 9930
+    },
+    {
+      "epoch": 14.119318181818182,
+      "grad_norm": 0.15551239252090454,
+      "learning_rate": 1.0172729264917857e-08,
+      "loss": 0.0087,
+      "step": 9940
+    },
+    {
+      "epoch": 14.133522727272727,
+      "grad_norm": 0.16436028480529785,
+      "learning_rate": 7.1108677466458215e-09,
+      "loss": 0.0082,
+      "step": 9950
+    },
+    {
+      "epoch": 14.147727272727273,
+      "grad_norm": 0.19523940980434418,
+      "learning_rate": 4.595720454353414e-09,
+      "loss": 0.0101,
+      "step": 9960
+    },
+    {
+      "epoch": 14.161931818181818,
+      "grad_norm": 0.16174790263175964,
+      "learning_rate": 2.627314893294264e-09,
+      "loss": 0.0093,
+      "step": 9970
+    },
+    {
+      "epoch": 14.176136363636363,
+      "grad_norm": 0.20230384171009064,
+      "learning_rate": 1.2056725896270048e-09,
+      "loss": 0.0073,
+      "step": 9980
+    },
+    {
+      "epoch": 14.190340909090908,
+      "grad_norm": 0.2047746330499649,
+      "learning_rate": 3.308090902098826e-10,
+      "loss": 0.009,
+      "step": 9990
+    },
+    {
+      "epoch": 14.204545454545455,
+      "grad_norm": 0.07353850454092026,
+      "learning_rate": 2.7339624120159555e-12,
+      "loss": 0.008,
+      "step": 10000
+    },
+    {
+      "epoch": 14.204545454545455,
+      "step": 10000,
+      "total_flos": 0.0,
+      "train_loss": 0.009726445004343987,
+      "train_runtime": 4609.837,
+      "train_samples_per_second": 69.417,
+      "train_steps_per_second": 2.169
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 15,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}