diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.7506430933812421,
+  "epoch": 0.9998775060226206,
   "eval_steps": 766,
-  "global_step": 2298,
+  "global_step": 3061,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -16125,6 +16125,5347 @@
       "eval_samples_per_second": 5.119,
       "eval_steps_per_second": 2.559,
       "step": 2298
+    },
+    {
+      "epoch": 0.7509697439875873,
+      "grad_norm": 1.218799114227295,
+      "learning_rate": 1.4987523307921375e-05,
+      "loss": 1.515,
+      "step": 2299
+    },
+    {
+      "epoch": 0.7512963945939325,
+      "grad_norm": 1.8166571855545044,
+      "learning_rate": 1.4950299309775967e-05,
+      "loss": 2.6661,
+      "step": 2300
+    },
+    {
+      "epoch": 0.7516230452002777,
+      "grad_norm": 0.21096064150333405,
+      "learning_rate": 1.491311346760143e-05,
+      "loss": 0.7332,
+      "step": 2301
+    },
+    {
+      "epoch": 0.7519496958066229,
+      "grad_norm": 0.22826731204986572,
+      "learning_rate": 1.4875965821879212e-05,
+      "loss": 0.7841,
+      "step": 2302
+    },
+    {
+      "epoch": 0.752276346412968,
+      "grad_norm": 0.2562881410121918,
+      "learning_rate": 1.4838856413049107e-05,
+      "loss": 0.8088,
+      "step": 2303
+    },
+    {
+      "epoch": 0.7526029970193132,
+      "grad_norm": 0.2786596119403839,
+      "learning_rate": 1.4801785281509394e-05,
+      "loss": 0.8839,
+      "step": 2304
+    },
+    {
+      "epoch": 0.7529296476256584,
+      "grad_norm": 0.27765336632728577,
+      "learning_rate": 1.4764752467616588e-05,
+      "loss": 0.8375,
+      "step": 2305
+    },
+    {
+      "epoch": 0.7532562982320036,
+      "grad_norm": 0.2921684980392456,
+      "learning_rate": 1.4727758011685538e-05,
+      "loss": 0.7784,
+      "step": 2306
+    },
+    {
+      "epoch": 0.7535829488383488,
+      "grad_norm": 0.2950403094291687,
+      "learning_rate": 1.4690801953989313e-05,
+      "loss": 0.8052,
+      "step": 2307
+    },
+    {
+      "epoch": 0.7539095994446939,
+      "grad_norm": 0.3217078745365143,
+      "learning_rate": 1.4653884334759205e-05,
+      "loss": 0.7845,
+      "step": 2308
+    },
+    {
+      "epoch": 0.7542362500510391,
+      "grad_norm": 0.3093569874763489,
+      "learning_rate": 1.461700519418464e-05,
+      "loss": 0.811,
+      "step": 2309
+    },
+    {
+      "epoch": 0.7545629006573843,
+      "grad_norm": 0.3464542031288147,
+      "learning_rate": 1.4580164572413163e-05,
+      "loss": 0.8718,
+      "step": 2310
+    },
+    {
+      "epoch": 0.7548895512637296,
+      "grad_norm": 0.3515595495700836,
+      "learning_rate": 1.454336250955039e-05,
+      "loss": 0.9003,
+      "step": 2311
+    },
+    {
+      "epoch": 0.7552162018700748,
+      "grad_norm": 0.36814939975738525,
+      "learning_rate": 1.4506599045659951e-05,
+      "loss": 0.813,
+      "step": 2312
+    },
+    {
+      "epoch": 0.75554285247642,
+      "grad_norm": 0.42278334498405457,
+      "learning_rate": 1.4469874220763463e-05,
+      "loss": 1.0227,
+      "step": 2313
+    },
+    {
+      "epoch": 0.7558695030827651,
+      "grad_norm": 0.427903413772583,
+      "learning_rate": 1.4433188074840482e-05,
+      "loss": 0.9597,
+      "step": 2314
+    },
+    {
+      "epoch": 0.7561961536891103,
+      "grad_norm": 0.4434622824192047,
+      "learning_rate": 1.4396540647828467e-05,
+      "loss": 0.891,
+      "step": 2315
+    },
+    {
+      "epoch": 0.7565228042954555,
+      "grad_norm": 0.47153371572494507,
+      "learning_rate": 1.4359931979622665e-05,
+      "loss": 0.9432,
+      "step": 2316
+    },
+    {
+      "epoch": 0.7568494549018007,
+      "grad_norm": 0.4934622049331665,
+      "learning_rate": 1.4323362110076238e-05,
+      "loss": 1.0377,
+      "step": 2317
+    },
+    {
+      "epoch": 0.7571761055081458,
+      "grad_norm": 0.5599250197410583,
+      "learning_rate": 1.4286831078999997e-05,
+      "loss": 1.1322,
+      "step": 2318
+    },
+    {
+      "epoch": 0.757502756114491,
+      "grad_norm": 0.5557501316070557,
+      "learning_rate": 1.4250338926162583e-05,
+      "loss": 0.9105,
+      "step": 2319
+    },
+    {
+      "epoch": 0.7578294067208362,
+      "grad_norm": 0.6891188025474548,
+      "learning_rate": 1.4213885691290202e-05,
+      "loss": 1.2214,
+      "step": 2320
+    },
+    {
+      "epoch": 0.7581560573271814,
+      "grad_norm": 0.8066973090171814,
+      "learning_rate": 1.4177471414066773e-05,
+      "loss": 1.2513,
+      "step": 2321
+    },
+    {
+      "epoch": 0.7584827079335266,
+      "grad_norm": 0.9567353129386902,
+      "learning_rate": 1.4141096134133774e-05,
+      "loss": 1.245,
+      "step": 2322
+    },
+    {
+      "epoch": 0.7588093585398717,
+      "grad_norm": 1.3140424489974976,
+      "learning_rate": 1.4104759891090252e-05,
+      "loss": 1.274,
+      "step": 2323
+    },
+    {
+      "epoch": 0.7591360091462169,
+      "grad_norm": 1.6508132219314575,
+      "learning_rate": 1.4068462724492688e-05,
+      "loss": 1.4996,
+      "step": 2324
+    },
+    {
+      "epoch": 0.7594626597525622,
+      "grad_norm": 1.7418688535690308,
+      "learning_rate": 1.4032204673855142e-05,
+      "loss": 2.1463,
+      "step": 2325
+    },
+    {
+      "epoch": 0.7597893103589074,
+      "grad_norm": 0.2149413824081421,
+      "learning_rate": 1.3995985778648962e-05,
+      "loss": 0.7248,
+      "step": 2326
+    },
+    {
+      "epoch": 0.7601159609652526,
+      "grad_norm": 0.24548666179180145,
+      "learning_rate": 1.3959806078302995e-05,
+      "loss": 0.7657,
+      "step": 2327
+    },
+    {
+      "epoch": 0.7604426115715978,
+      "grad_norm": 0.26261597871780396,
+      "learning_rate": 1.3923665612203308e-05,
+      "loss": 0.7632,
+      "step": 2328
+    },
+    {
+      "epoch": 0.7607692621779429,
+      "grad_norm": 0.27283692359924316,
+      "learning_rate": 1.3887564419693328e-05,
+      "loss": 0.8575,
+      "step": 2329
+    },
+    {
+      "epoch": 0.7610959127842881,
+      "grad_norm": 0.28144407272338867,
+      "learning_rate": 1.3851502540073696e-05,
+      "loss": 0.8438,
+      "step": 2330
+    },
+    {
+      "epoch": 0.7614225633906333,
+      "grad_norm": 0.2961028218269348,
+      "learning_rate": 1.3815480012602272e-05,
+      "loss": 0.8569,
+      "step": 2331
+    },
+    {
+      "epoch": 0.7617492139969785,
+      "grad_norm": 0.30785036087036133,
+      "learning_rate": 1.377949687649407e-05,
+      "loss": 0.8506,
+      "step": 2332
+    },
+    {
+      "epoch": 0.7620758646033237,
+      "grad_norm": 0.29399818181991577,
+      "learning_rate": 1.3743553170921225e-05,
+      "loss": 0.7934,
+      "step": 2333
+    },
+    {
+      "epoch": 0.7624025152096688,
+      "grad_norm": 0.3192881643772125,
+      "learning_rate": 1.3707648935012935e-05,
+      "loss": 0.8179,
+      "step": 2334
+    },
+    {
+      "epoch": 0.762729165816014,
+      "grad_norm": 0.34171509742736816,
+      "learning_rate": 1.3671784207855442e-05,
+      "loss": 0.8689,
+      "step": 2335
+    },
+    {
+      "epoch": 0.7630558164223592,
+      "grad_norm": 0.3571321666240692,
+      "learning_rate": 1.3635959028492002e-05,
+      "loss": 0.8651,
+      "step": 2336
+    },
+    {
+      "epoch": 0.7633824670287044,
+      "grad_norm": 0.3970184922218323,
+      "learning_rate": 1.3600173435922725e-05,
+      "loss": 0.9319,
+      "step": 2337
+    },
+    {
+      "epoch": 0.7637091176350496,
+      "grad_norm": 0.39088132977485657,
+      "learning_rate": 1.3564427469104768e-05,
+      "loss": 0.9152,
+      "step": 2338
+    },
+    {
+      "epoch": 0.7640357682413949,
+      "grad_norm": 0.42941996455192566,
+      "learning_rate": 1.3528721166952007e-05,
+      "loss": 0.9292,
+      "step": 2339
+    },
+    {
+      "epoch": 0.76436241884774,
+      "grad_norm": 0.43511438369750977,
+      "learning_rate": 1.3493054568335273e-05,
+      "loss": 0.9781,
+      "step": 2340
+    },
+    {
+      "epoch": 0.7646890694540852,
+      "grad_norm": 0.5111512541770935,
+      "learning_rate": 1.3457427712082049e-05,
+      "loss": 0.9431,
+      "step": 2341
+    },
+    {
+      "epoch": 0.7650157200604304,
+      "grad_norm": 0.5455848574638367,
+      "learning_rate": 1.342184063697664e-05,
+      "loss": 1.2364,
+      "step": 2342
+    },
+    {
+      "epoch": 0.7653423706667756,
+      "grad_norm": 0.5164190530776978,
+      "learning_rate": 1.3386293381760013e-05,
+      "loss": 1.0224,
+      "step": 2343
+    },
+    {
+      "epoch": 0.7656690212731208,
+      "grad_norm": 0.6091725826263428,
+      "learning_rate": 1.3350785985129782e-05,
+      "loss": 1.1075,
+      "step": 2344
+    },
+    {
+      "epoch": 0.7659956718794659,
+      "grad_norm": 0.6855178475379944,
+      "learning_rate": 1.331531848574018e-05,
+      "loss": 1.1122,
+      "step": 2345
+    },
+    {
+      "epoch": 0.7663223224858111,
+      "grad_norm": 0.7483596205711365,
+      "learning_rate": 1.327989092220201e-05,
+      "loss": 1.2955,
+      "step": 2346
+    },
+    {
+      "epoch": 0.7666489730921563,
+      "grad_norm": 0.9962571263313293,
+      "learning_rate": 1.3244503333082586e-05,
+      "loss": 1.2699,
+      "step": 2347
+    },
+    {
+      "epoch": 0.7669756236985015,
+      "grad_norm": 1.1855285167694092,
+      "learning_rate": 1.3209155756905722e-05,
+      "loss": 1.257,
+      "step": 2348
+    },
+    {
+      "epoch": 0.7673022743048467,
+      "grad_norm": 1.2274976968765259,
+      "learning_rate": 1.3173848232151665e-05,
+      "loss": 1.2729,
+      "step": 2349
+    },
+    {
+      "epoch": 0.7676289249111918,
+      "grad_norm": 1.730171799659729,
+      "learning_rate": 1.3138580797257027e-05,
+      "loss": 1.3523,
+      "step": 2350
+    },
+    {
+      "epoch": 0.767955575517537,
+      "grad_norm": 0.20054863393306732,
+      "learning_rate": 1.3103353490614861e-05,
+      "loss": 0.7015,
+      "step": 2351
+    },
+    {
+      "epoch": 0.7682822261238822,
+      "grad_norm": 0.24544449150562286,
+      "learning_rate": 1.306816635057443e-05,
+      "loss": 0.8144,
+      "step": 2352
+    },
+    {
+      "epoch": 0.7686088767302274,
+      "grad_norm": 0.252315491437912,
+      "learning_rate": 1.3033019415441377e-05,
+      "loss": 0.7682,
+      "step": 2353
+    },
+    {
+      "epoch": 0.7689355273365727,
+      "grad_norm": 0.2725442051887512,
+      "learning_rate": 1.299791272347749e-05,
+      "loss": 0.7963,
+      "step": 2354
+    },
+    {
+      "epoch": 0.7692621779429178,
+      "grad_norm": 0.2909514307975769,
+      "learning_rate": 1.2962846312900789e-05,
+      "loss": 0.7745,
+      "step": 2355
+    },
+    {
+      "epoch": 0.769588828549263,
+      "grad_norm": 0.3024318814277649,
+      "learning_rate": 1.2927820221885446e-05,
+      "loss": 0.9283,
+      "step": 2356
+    },
+    {
+      "epoch": 0.7699154791556082,
+      "grad_norm": 0.32852616906166077,
+      "learning_rate": 1.2892834488561717e-05,
+      "loss": 0.9256,
+      "step": 2357
+    },
+    {
+      "epoch": 0.7702421297619534,
+      "grad_norm": 0.29283902049064636,
+      "learning_rate": 1.285788915101594e-05,
+      "loss": 0.8214,
+      "step": 2358
+    },
+    {
+      "epoch": 0.7705687803682986,
+      "grad_norm": 0.32372137904167175,
+      "learning_rate": 1.2822984247290493e-05,
+      "loss": 0.7905,
+      "step": 2359
+    },
+    {
+      "epoch": 0.7708954309746437,
+      "grad_norm": 0.3313743770122528,
+      "learning_rate": 1.278811981538367e-05,
+      "loss": 0.8942,
+      "step": 2360
+    },
+    {
+      "epoch": 0.7712220815809889,
+      "grad_norm": 0.370326429605484,
+      "learning_rate": 1.2753295893249811e-05,
+      "loss": 0.8639,
+      "step": 2361
+    },
+    {
+      "epoch": 0.7715487321873341,
+      "grad_norm": 0.371473491191864,
+      "learning_rate": 1.2718512518799059e-05,
+      "loss": 0.8924,
+      "step": 2362
+    },
+    {
+      "epoch": 0.7718753827936793,
+      "grad_norm": 0.3572691082954407,
+      "learning_rate": 1.2683769729897466e-05,
+      "loss": 0.9241,
+      "step": 2363
+    },
+    {
+      "epoch": 0.7722020334000245,
+      "grad_norm": 0.4208536446094513,
+      "learning_rate": 1.2649067564366896e-05,
+      "loss": 1.0033,
+      "step": 2364
+    },
+    {
+      "epoch": 0.7725286840063696,
+      "grad_norm": 0.4127427041530609,
+      "learning_rate": 1.2614406059984978e-05,
+      "loss": 0.881,
+      "step": 2365
+    },
+    {
+      "epoch": 0.7728553346127148,
+      "grad_norm": 0.4512838125228882,
+      "learning_rate": 1.2579785254485088e-05,
+      "loss": 1.03,
+      "step": 2366
+    },
+    {
+      "epoch": 0.77318198521906,
+      "grad_norm": 0.5090767741203308,
+      "learning_rate": 1.254520518555629e-05,
+      "loss": 1.1346,
+      "step": 2367
+    },
+    {
+      "epoch": 0.7735086358254053,
+      "grad_norm": 0.5246442556381226,
+      "learning_rate": 1.2510665890843298e-05,
+      "loss": 0.9972,
+      "step": 2368
+    },
+    {
+      "epoch": 0.7738352864317505,
+      "grad_norm": 0.5956628918647766,
+      "learning_rate": 1.2476167407946448e-05,
+      "loss": 1.188,
+      "step": 2369
+    },
+    {
+      "epoch": 0.7741619370380957,
+      "grad_norm": 0.6347416639328003,
+      "learning_rate": 1.2441709774421639e-05,
+      "loss": 1.2596,
+      "step": 2370
+    },
+    {
+      "epoch": 0.7744885876444408,
+      "grad_norm": 0.7028458714485168,
+      "learning_rate": 1.2407293027780304e-05,
+      "loss": 0.9761,
+      "step": 2371
+    },
+    {
+      "epoch": 0.774815238250786,
+      "grad_norm": 0.9047834277153015,
+      "learning_rate": 1.2372917205489376e-05,
+      "loss": 1.408,
+      "step": 2372
+    },
+    {
+      "epoch": 0.7751418888571312,
+      "grad_norm": 1.1155500411987305,
+      "learning_rate": 1.233858234497119e-05,
+      "loss": 1.2624,
+      "step": 2373
+    },
+    {
+      "epoch": 0.7754685394634764,
+      "grad_norm": 1.3037281036376953,
+      "learning_rate": 1.2304288483603565e-05,
+      "loss": 1.4873,
+      "step": 2374
+    },
+    {
+      "epoch": 0.7757951900698216,
+      "grad_norm": 1.3568501472473145,
+      "learning_rate": 1.227003565871962e-05,
+      "loss": 1.4208,
+      "step": 2375
+    },
+    {
+      "epoch": 0.7761218406761667,
+      "grad_norm": 0.21427606046199799,
+      "learning_rate": 1.2235823907607834e-05,
+      "loss": 0.7637,
+      "step": 2376
+    },
+    {
+      "epoch": 0.7764484912825119,
+      "grad_norm": 0.2435503751039505,
+      "learning_rate": 1.220165326751197e-05,
+      "loss": 0.7515,
+      "step": 2377
+    },
+    {
+      "epoch": 0.7767751418888571,
+      "grad_norm": 0.25543227791786194,
+      "learning_rate": 1.2167523775631029e-05,
+      "loss": 0.8145,
+      "step": 2378
+    },
+    {
+      "epoch": 0.7771017924952023,
+      "grad_norm": 0.2633327543735504,
+      "learning_rate": 1.2133435469119214e-05,
+      "loss": 0.8333,
+      "step": 2379
+    },
+    {
+      "epoch": 0.7774284431015475,
+      "grad_norm": 0.2795431613922119,
+      "learning_rate": 1.2099388385085908e-05,
+      "loss": 0.8558,
+      "step": 2380
+    },
+    {
+      "epoch": 0.7777550937078926,
+      "grad_norm": 0.28009021282196045,
+      "learning_rate": 1.20653825605956e-05,
+      "loss": 0.764,
+      "step": 2381
+    },
+    {
+      "epoch": 0.7780817443142379,
+      "grad_norm": 0.2952665686607361,
+      "learning_rate": 1.203141803266788e-05,
+      "loss": 0.7802,
+      "step": 2382
+    },
+    {
+      "epoch": 0.7784083949205831,
+      "grad_norm": 0.3152238130569458,
+      "learning_rate": 1.1997494838277368e-05,
+      "loss": 0.876,
+      "step": 2383
+    },
+    {
+      "epoch": 0.7787350455269283,
+      "grad_norm": 0.40310126543045044,
+      "learning_rate": 1.1963613014353692e-05,
+      "loss": 0.8793,
+      "step": 2384
+    },
+    {
+      "epoch": 0.7790616961332735,
+      "grad_norm": 0.35672134160995483,
+      "learning_rate": 1.1929772597781458e-05,
+      "loss": 0.9216,
+      "step": 2385
+    },
+    {
+      "epoch": 0.7793883467396187,
+      "grad_norm": 0.3469008207321167,
+      "learning_rate": 1.1895973625400136e-05,
+      "loss": 0.8372,
+      "step": 2386
+    },
+    {
+      "epoch": 0.7797149973459638,
+      "grad_norm": 0.4095902442932129,
+      "learning_rate": 1.1862216134004189e-05,
+      "loss": 0.9618,
+      "step": 2387
+    },
+    {
+      "epoch": 0.780041647952309,
+      "grad_norm": 0.4208645224571228,
+      "learning_rate": 1.1828500160342787e-05,
+      "loss": 0.886,
+      "step": 2388
+    },
+    {
+      "epoch": 0.7803682985586542,
+      "grad_norm": 0.4261435568332672,
+      "learning_rate": 1.1794825741120042e-05,
+      "loss": 1.0212,
+      "step": 2389
+    },
+    {
+      "epoch": 0.7806949491649994,
+      "grad_norm": 0.4427204430103302,
+      "learning_rate": 1.1761192912994706e-05,
+      "loss": 0.8466,
+      "step": 2390
+    },
+    {
+      "epoch": 0.7810215997713446,
+      "grad_norm": 0.5028002858161926,
+      "learning_rate": 1.1727601712580321e-05,
+      "loss": 1.0301,
+      "step": 2391
+    },
+    {
+      "epoch": 0.7813482503776897,
+      "grad_norm": 0.5116696953773499,
+      "learning_rate": 1.1694052176445114e-05,
+      "loss": 1.0023,
+      "step": 2392
+    },
+    {
+      "epoch": 0.7816749009840349,
+      "grad_norm": 0.5628235340118408,
+      "learning_rate": 1.166054434111194e-05,
+      "loss": 1.0638,
+      "step": 2393
+    },
+    {
+      "epoch": 0.7820015515903801,
+      "grad_norm": 0.628449022769928,
+      "learning_rate": 1.1627078243058214e-05,
+      "loss": 1.0702,
+      "step": 2394
+    },
+    {
+      "epoch": 0.7823282021967253,
+      "grad_norm": 0.7063364386558533,
+      "learning_rate": 1.159365391871602e-05,
+      "loss": 1.1558,
+      "step": 2395
+    },
+    {
+      "epoch": 0.7826548528030706,
+      "grad_norm": 0.847536027431488,
+      "learning_rate": 1.1560271404471845e-05,
+      "loss": 1.272,
+      "step": 2396
+    },
+    {
+      "epoch": 0.7829815034094157,
+      "grad_norm": 0.9608874320983887,
+      "learning_rate": 1.152693073666678e-05,
+      "loss": 1.2311,
+      "step": 2397
+    },
+    {
+      "epoch": 0.7833081540157609,
+      "grad_norm": 1.243880033493042,
+      "learning_rate": 1.149363195159625e-05,
+      "loss": 1.2427,
+      "step": 2398
+    },
+    {
+      "epoch": 0.7836348046221061,
+      "grad_norm": 1.3490464687347412,
+      "learning_rate": 1.1460375085510161e-05,
+      "loss": 1.6011,
+      "step": 2399
+    },
+    {
+      "epoch": 0.7839614552284513,
+      "grad_norm": 2.450794219970703,
+      "learning_rate": 1.142716017461275e-05,
+      "loss": 2.5833,
+      "step": 2400
+    },
+    {
+      "epoch": 0.7842881058347965,
+      "grad_norm": 0.20113646984100342,
+      "learning_rate": 1.1393987255062599e-05,
+      "loss": 0.7018,
+      "step": 2401
+    },
+    {
+      "epoch": 0.7846147564411416,
+      "grad_norm": 0.24163466691970825,
+      "learning_rate": 1.1360856362972555e-05,
+      "loss": 0.7365,
+      "step": 2402
+    },
+    {
+      "epoch": 0.7849414070474868,
+      "grad_norm": 0.2692144513130188,
+      "learning_rate": 1.1327767534409739e-05,
+      "loss": 0.8654,
+      "step": 2403
+    },
+    {
+      "epoch": 0.785268057653832,
+      "grad_norm": 0.271269291639328,
+      "learning_rate": 1.1294720805395464e-05,
+      "loss": 0.8093,
+      "step": 2404
+    },
+    {
+      "epoch": 0.7855947082601772,
+      "grad_norm": 0.2852034270763397,
+      "learning_rate": 1.126171621190522e-05,
+      "loss": 0.8286,
+      "step": 2405
+    },
+    {
+      "epoch": 0.7859213588665224,
+      "grad_norm": 0.2870404124259949,
+      "learning_rate": 1.122875378986863e-05,
+      "loss": 0.7744,
+      "step": 2406
+    },
+    {
+      "epoch": 0.7862480094728675,
+      "grad_norm": 0.3201829791069031,
+      "learning_rate": 1.1195833575169362e-05,
+      "loss": 0.8396,
+      "step": 2407
+    },
+    {
+      "epoch": 0.7865746600792127,
+      "grad_norm": 0.33272668719291687,
+      "learning_rate": 1.1162955603645236e-05,
+      "loss": 0.8156,
+      "step": 2408
+    },
+    {
+      "epoch": 0.7869013106855579,
+      "grad_norm": 0.34226053953170776,
+      "learning_rate": 1.1130119911087966e-05,
+      "loss": 0.8511,
+      "step": 2409
+    },
+    {
+      "epoch": 0.7872279612919032,
+      "grad_norm": 0.3556172847747803,
+      "learning_rate": 1.1097326533243352e-05,
+      "loss": 0.8328,
+      "step": 2410
+    },
+    {
+      "epoch": 0.7875546118982484,
+      "grad_norm": 0.3675041198730469,
+      "learning_rate": 1.1064575505811043e-05,
+      "loss": 0.9927,
+      "step": 2411
+    },
+    {
+      "epoch": 0.7878812625045936,
+      "grad_norm": 0.38928094506263733,
+      "learning_rate": 1.1031866864444618e-05,
+      "loss": 0.8938,
+      "step": 2412
+    },
+    {
+      "epoch": 0.7882079131109387,
+      "grad_norm": 0.3787207007408142,
+      "learning_rate": 1.0999200644751523e-05,
+      "loss": 0.8779,
+      "step": 2413
+    },
+    {
+      "epoch": 0.7885345637172839,
+      "grad_norm": 0.4117635190486908,
+      "learning_rate": 1.0966576882293012e-05,
+      "loss": 1.102,
+      "step": 2414
+    },
+    {
+      "epoch": 0.7888612143236291,
+      "grad_norm": 0.4420527517795563,
+      "learning_rate": 1.0933995612584114e-05,
+      "loss": 0.9014,
+      "step": 2415
+    },
+    {
+      "epoch": 0.7891878649299743,
+      "grad_norm": 0.4948444068431854,
+      "learning_rate": 1.0901456871093613e-05,
+      "loss": 1.0171,
+      "step": 2416
+    },
+    {
+      "epoch": 0.7895145155363195,
+      "grad_norm": 0.5133086442947388,
+      "learning_rate": 1.086896069324398e-05,
+      "loss": 0.9654,
+      "step": 2417
+    },
+    {
+      "epoch": 0.7898411661426646,
+      "grad_norm": 0.5456277132034302,
+      "learning_rate": 1.0836507114411359e-05,
+      "loss": 0.9479,
+      "step": 2418
+    },
+    {
+      "epoch": 0.7901678167490098,
+      "grad_norm": 0.6290357708930969,
+      "learning_rate": 1.0804096169925532e-05,
+      "loss": 1.1441,
+      "step": 2419
+    },
+    {
+      "epoch": 0.790494467355355,
+      "grad_norm": 0.7249221205711365,
+      "learning_rate": 1.0771727895069816e-05,
+      "loss": 1.2534,
+      "step": 2420
+    },
+    {
+      "epoch": 0.7908211179617002,
+      "grad_norm": 0.8690229654312134,
+      "learning_rate": 1.0739402325081172e-05,
+      "loss": 1.262,
+      "step": 2421
+    },
+    {
+      "epoch": 0.7911477685680454,
+      "grad_norm": 0.9221552014350891,
+      "learning_rate": 1.0707119495149958e-05,
+      "loss": 1.1823,
+      "step": 2422
+    },
+    {
+      "epoch": 0.7914744191743905,
+      "grad_norm": 1.201319694519043,
+      "learning_rate": 1.0674879440420117e-05,
+      "loss": 1.4645,
+      "step": 2423
+    },
+    {
+      "epoch": 0.7918010697807358,
+      "grad_norm": 1.345933198928833,
+      "learning_rate": 1.0642682195988917e-05,
+      "loss": 1.4429,
+      "step": 2424
+    },
+    {
+      "epoch": 0.792127720387081,
+      "grad_norm": 1.9995300769805908,
+      "learning_rate": 1.0610527796907104e-05,
+      "loss": 2.2084,
+      "step": 2425
+    },
+    {
+      "epoch": 0.7924543709934262,
+      "grad_norm": 0.2166060358285904,
+      "learning_rate": 1.057841627817875e-05,
+      "loss": 0.6935,
+      "step": 2426
+    },
+    {
+      "epoch": 0.7927810215997714,
+      "grad_norm": 0.24457815289497375,
+      "learning_rate": 1.0546347674761236e-05,
+      "loss": 0.7934,
+      "step": 2427
+    },
+    {
+      "epoch": 0.7931076722061166,
+      "grad_norm": 0.25961560010910034,
+      "learning_rate": 1.0514322021565248e-05,
+      "loss": 0.7806,
+      "step": 2428
+    },
+    {
+      "epoch": 0.7934343228124617,
+      "grad_norm": 0.2620426416397095,
+      "learning_rate": 1.0482339353454717e-05,
+      "loss": 0.8303,
+      "step": 2429
+    },
+    {
+      "epoch": 0.7937609734188069,
+      "grad_norm": 0.2988450229167938,
+      "learning_rate": 1.0450399705246721e-05,
+      "loss": 0.8409,
+      "step": 2430
+    },
+    {
+      "epoch": 0.7940876240251521,
+      "grad_norm": 0.27858319878578186,
+      "learning_rate": 1.0418503111711591e-05,
+      "loss": 0.7863,
+      "step": 2431
+    },
+    {
+      "epoch": 0.7944142746314973,
+      "grad_norm": 0.3067804276943207,
+      "learning_rate": 1.0386649607572758e-05,
+      "loss": 0.9039,
+      "step": 2432
+    },
+    {
+      "epoch": 0.7947409252378425,
+      "grad_norm": 0.31061863899230957,
+      "learning_rate": 1.035483922750668e-05,
+      "loss": 0.8666,
+      "step": 2433
+    },
+    {
+      "epoch": 0.7950675758441876,
+      "grad_norm": 0.324632853269577,
+      "learning_rate": 1.032307200614298e-05,
+      "loss": 0.9376,
+      "step": 2434
+    },
+    {
+      "epoch": 0.7953942264505328,
+      "grad_norm": 0.31783485412597656,
+      "learning_rate": 1.0291347978064197e-05,
+      "loss": 0.8443,
+      "step": 2435
+    },
+    {
+      "epoch": 0.795720877056878,
+      "grad_norm": 0.37628984451293945,
+      "learning_rate": 1.0259667177805937e-05,
+      "loss": 0.8687,
+      "step": 2436
+    },
+    {
+      "epoch": 0.7960475276632232,
+      "grad_norm": 0.35589486360549927,
+      "learning_rate": 1.0228029639856663e-05,
+      "loss": 0.8424,
+      "step": 2437
+    },
+    {
+      "epoch": 0.7963741782695685,
+      "grad_norm": 0.4087234437465668,
+      "learning_rate": 1.0196435398657795e-05,
+      "loss": 0.8694,
+      "step": 2438
+    },
+    {
+      "epoch": 0.7967008288759136,
+      "grad_norm": 0.4563935101032257,
+      "learning_rate": 1.016488448860361e-05,
+      "loss": 1.0653,
+      "step": 2439
+    },
+    {
+      "epoch": 0.7970274794822588,
+      "grad_norm": 0.4497915804386139,
+      "learning_rate": 1.0133376944041201e-05,
+      "loss": 0.9739,
+      "step": 2440
+    },
+    {
+      "epoch": 0.797354130088604,
+      "grad_norm": 0.5139485597610474,
+      "learning_rate": 1.0101912799270464e-05,
+      "loss": 1.0967,
+      "step": 2441
+    },
+    {
+      "epoch": 0.7976807806949492,
+      "grad_norm": 0.5172460079193115,
+      "learning_rate": 1.0070492088544058e-05,
+      "loss": 0.9596,
+      "step": 2442
+    },
+    {
+      "epoch": 0.7980074313012944,
+      "grad_norm": 0.5660231709480286,
+      "learning_rate": 1.003911484606731e-05,
+      "loss": 1.029,
+      "step": 2443
+    },
+    {
+      "epoch": 0.7983340819076395,
+      "grad_norm": 0.6900216341018677,
+      "learning_rate": 1.000778110599831e-05,
+      "loss": 1.3027,
+      "step": 2444
+    },
+    {
+      "epoch": 0.7986607325139847,
+      "grad_norm": 0.8324858546257019,
+      "learning_rate": 9.976490902447706e-06,
+      "loss": 1.3106,
+      "step": 2445
+    },
+    {
+      "epoch": 0.7989873831203299,
+      "grad_norm": 0.9467867016792297,
+      "learning_rate": 9.945244269478793e-06,
+      "loss": 1.2018,
+      "step": 2446
+    },
+    {
+      "epoch": 0.7993140337266751,
+      "grad_norm": 1.3282982110977173,
+      "learning_rate": 9.914041241107441e-06,
+      "loss": 1.225,
+      "step": 2447
+    },
+    {
+      "epoch": 0.7996406843330203,
+      "grad_norm": 1.7920281887054443,
+      "learning_rate": 9.882881851302023e-06,
+      "loss": 1.9809,
+      "step": 2448
+    },
+    {
+      "epoch": 0.7999673349393654,
+      "grad_norm": 1.387970209121704,
+      "learning_rate": 9.851766133983436e-06,
+      "loss": 1.6633,
+      "step": 2449
+    },
+    {
+      "epoch": 0.8002939855457106,
+      "grad_norm": 1.9271981716156006,
+      "learning_rate": 9.82069412302501e-06,
+      "loss": 1.7488,
+      "step": 2450
+    },
+    {
+      "epoch": 0.8006206361520558,
+      "grad_norm": 0.21858730912208557,
+      "learning_rate": 9.789665852252505e-06,
+      "loss": 0.6825,
+      "step": 2451
+    },
+    {
+      "epoch": 0.8009472867584011,
+      "grad_norm": 0.2585441470146179,
+      "learning_rate": 9.758681355444071e-06,
+      "loss": 0.8612,
+      "step": 2452
+    },
+    {
+      "epoch": 0.8012739373647463,
+      "grad_norm": 0.2628011405467987,
+      "learning_rate": 9.727740666330187e-06,
+      "loss": 0.8562,
+      "step": 2453
+    },
+    {
+      "epoch": 0.8016005879710915,
+      "grad_norm": 0.28666362166404724,
+      "learning_rate": 9.696843818593665e-06,
+      "loss": 0.8556,
+      "step": 2454
+    },
+    {
+      "epoch": 0.8019272385774366,
+      "grad_norm": 0.284396767616272,
+      "learning_rate": 9.665990845869578e-06,
+      "loss": 0.817,
+      "step": 2455
+    },
+    {
+      "epoch": 0.8022538891837818,
+      "grad_norm": 0.2911209464073181,
+      "learning_rate": 9.635181781745201e-06,
+      "loss": 0.8772,
+      "step": 2456
+    },
+    {
+      "epoch": 0.802580539790127,
+      "grad_norm": 0.31056690216064453,
+      "learning_rate": 9.604416659760091e-06,
+      "loss": 0.7981,
+      "step": 2457
+    },
+    {
+      "epoch": 0.8029071903964722,
+      "grad_norm": 0.3305649161338806,
+      "learning_rate": 9.573695513405894e-06,
+      "loss": 0.8759,
+      "step": 2458
+    },
+    {
+      "epoch": 0.8032338410028174,
+      "grad_norm": 0.344088077545166,
+      "learning_rate": 9.543018376126406e-06,
+      "loss": 0.8886,
+      "step": 2459
+    },
+    {
+      "epoch": 0.8035604916091625,
+      "grad_norm": 0.36468231678009033,
+      "learning_rate": 9.512385281317532e-06,
+      "loss": 0.9108,
+      "step": 2460
+    },
+    {
+      "epoch": 0.8038871422155077,
+      "grad_norm": 0.36691391468048096,
+      "learning_rate": 9.481796262327209e-06,
+      "loss": 0.8152,
+      "step": 2461
+    },
+    {
+      "epoch": 0.8042137928218529,
+      "grad_norm": 0.3990345299243927,
+      "learning_rate": 9.451251352455398e-06,
+      "loss": 0.9779,
+      "step": 2462
+    },
+    {
+      "epoch": 0.8045404434281981,
+      "grad_norm": 0.45102572441101074,
+      "learning_rate": 9.420750584954046e-06,
+      "loss": 0.9852,
+      "step": 2463
+    },
+    {
+      "epoch": 0.8048670940345433,
+      "grad_norm": 0.4353548288345337,
+      "learning_rate": 9.390293993027044e-06,
+      "loss": 0.9961,
+      "step": 2464
+    },
+    {
+      "epoch": 0.8051937446408884,
+      "grad_norm": 0.46060293912887573,
+      "learning_rate": 9.35988160983019e-06,
+      "loss": 0.9627,
+      "step": 2465
+    },
+    {
+      "epoch": 0.8055203952472336,
+      "grad_norm": 0.49444007873535156,
+      "learning_rate": 9.329513468471151e-06,
+      "loss": 1.0372,
+      "step": 2466
+    },
+    {
+      "epoch": 0.8058470458535789,
+      "grad_norm": 0.520687460899353,
+      "learning_rate": 9.299189602009445e-06,
+      "loss": 0.9868,
+      "step": 2467
+    },
+    {
+      "epoch": 0.8061736964599241,
+      "grad_norm": 0.5475303530693054,
+      "learning_rate": 9.268910043456391e-06,
+      "loss": 1.007,
+      "step": 2468
+    },
+    {
+      "epoch": 0.8065003470662693,
+      "grad_norm": 0.6077768802642822,
+      "learning_rate": 9.238674825775035e-06,
+      "loss": 1.2054,
+      "step": 2469
+    },
+    {
+      "epoch": 0.8068269976726145,
+      "grad_norm": 0.6752180457115173,
+      "learning_rate": 9.208483981880217e-06,
+      "loss": 1.1677,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8071536482789596,
+      "grad_norm": 0.7697159647941589,
+      "learning_rate": 9.178337544638398e-06,
+      "loss": 1.2908,
+      "step": 2471
+    },
+    {
+      "epoch": 0.8074802988853048,
+      "grad_norm": 0.9111776351928711,
+      "learning_rate": 9.148235546867783e-06,
+      "loss": 1.142,
+      "step": 2472
+    },
+    {
+      "epoch": 0.80780694949165,
+      "grad_norm": 1.0891627073287964,
+      "learning_rate": 9.118178021338114e-06,
+      "loss": 1.2352,
+      "step": 2473
+    },
+    {
+      "epoch": 0.8081336000979952,
+      "grad_norm": 1.2946839332580566,
+      "learning_rate": 9.088165000770766e-06,
+      "loss": 1.7181,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8084602507043404,
+      "grad_norm": 1.7693047523498535,
+      "learning_rate": 9.058196517838657e-06,
+      "loss": 1.4192,
+      "step": 2475
+    },
+    {
+      "epoch": 0.8087869013106855,
+      "grad_norm": 0.21156850457191467,
+      "learning_rate": 9.028272605166238e-06,
+      "loss": 0.7241,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8091135519170307,
+      "grad_norm": 0.2328900843858719,
+      "learning_rate": 8.998393295329371e-06,
+      "loss": 0.8189,
+      "step": 2477
+    },
+    {
+      "epoch": 0.8094402025233759,
+      "grad_norm": 0.25472328066825867,
+      "learning_rate": 8.968558620855466e-06,
+      "loss": 0.7547,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8097668531297211,
+      "grad_norm": 0.28889092803001404,
+      "learning_rate": 8.938768614223237e-06,
+      "loss": 0.8085,
+      "step": 2479
+    },
+    {
+      "epoch": 0.8100935037360663,
+      "grad_norm": 0.2841634750366211,
+      "learning_rate": 8.90902330786288e-06,
+      "loss": 0.7787,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8104201543424115,
+      "grad_norm": 0.2820068597793579,
+      "learning_rate": 8.87932273415582e-06,
+      "loss": 0.8728,
+      "step": 2481
+    },
+    {
+      "epoch": 0.8107468049487567,
+      "grad_norm": 0.2842544913291931,
+      "learning_rate": 8.849666925434857e-06,
+      "loss": 0.7728,
+      "step": 2482
+    },
+    {
+      "epoch": 0.8110734555551019,
+      "grad_norm": 0.30535271763801575,
+      "learning_rate": 8.820055913984033e-06,
+      "loss": 0.8933,
+      "step": 2483
+    },
+    {
+      "epoch": 0.8114001061614471,
+      "grad_norm": 0.3458235561847687,
+      "learning_rate": 8.790489732038638e-06,
+      "loss": 0.8484,
+      "step": 2484
+    },
+    {
+      "epoch": 0.8117267567677923,
+      "grad_norm": 0.3394298553466797,
+      "learning_rate": 8.76096841178513e-06,
+      "loss": 0.9129,
+      "step": 2485
+    },
+    {
+      "epoch": 0.8120534073741374,
+      "grad_norm": 0.3635712265968323,
+      "learning_rate": 8.731491985361167e-06,
+      "loss": 0.9531,
+      "step": 2486
+    },
+    {
+      "epoch": 0.8123800579804826,
+      "grad_norm": 0.38414081931114197,
+      "learning_rate": 8.702060484855506e-06,
+      "loss": 0.8569,
+      "step": 2487
+    },
+    {
+      "epoch": 0.8127067085868278,
+      "grad_norm": 0.3860231041908264,
+      "learning_rate": 8.672673942308008e-06,
+      "loss": 0.8907,
+      "step": 2488
+    },
+    {
+      "epoch": 0.813033359193173,
+      "grad_norm": 0.42519089579582214,
+      "learning_rate": 8.643332389709601e-06,
+      "loss": 1.015,
+      "step": 2489
+    },
+    {
+      "epoch": 0.8133600097995182,
+      "grad_norm": 0.4451555013656616,
+      "learning_rate": 8.61403585900219e-06,
+      "loss": 0.9342,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8136866604058633,
+      "grad_norm": 0.4713704586029053,
+      "learning_rate": 8.584784382078748e-06,
+      "loss": 0.9179,
+      "step": 2491
+    },
+    {
+      "epoch": 0.8140133110122085,
+      "grad_norm": 0.4954456388950348,
+      "learning_rate": 8.555577990783109e-06,
+      "loss": 1.0105,
+      "step": 2492
+    },
+    {
+      "epoch": 0.8143399616185537,
+      "grad_norm": 0.5571999549865723,
+      "learning_rate": 8.526416716910129e-06,
+      "loss": 1.0412,
+      "step": 2493
+    },
+    {
+      "epoch": 0.8146666122248989,
+      "grad_norm": 0.5991007685661316,
+      "learning_rate": 8.497300592205442e-06,
+      "loss": 1.0632,
+      "step": 2494
+    },
+    {
+      "epoch": 0.8149932628312442,
+      "grad_norm": 0.6661972999572754,
+      "learning_rate": 8.468229648365605e-06,
+      "loss": 1.0149,
+      "step": 2495
+    },
+    {
+      "epoch": 0.8153199134375894,
+      "grad_norm": 0.7918639183044434,
+      "learning_rate": 8.439203917037962e-06,
+      "loss": 1.289,
+      "step": 2496
+    },
+    {
+      "epoch": 0.8156465640439345,
+      "grad_norm": 1.0117945671081543,
+      "learning_rate": 8.41022342982064e-06,
+      "loss": 1.0847,
+      "step": 2497
+    },
+    {
+      "epoch": 0.8159732146502797,
+      "grad_norm": 1.2291086912155151,
+      "learning_rate": 8.381288218262517e-06,
+      "loss": 1.6722,
+      "step": 2498
+    },
+    {
+      "epoch": 0.8162998652566249,
+      "grad_norm": 1.2318812608718872,
+      "learning_rate": 8.352398313863174e-06,
+      "loss": 1.3529,
+      "step": 2499
+    },
+    {
+      "epoch": 0.8166265158629701,
+      "grad_norm": 1.7156823873519897,
+      "learning_rate": 8.323553748072883e-06,
+      "loss": 1.3938,
+      "step": 2500
+    },
+    {
+      "epoch": 0.8169531664693153,
+      "grad_norm": 0.2098822295665741,
+      "learning_rate": 8.29475455229255e-06,
+      "loss": 0.6383,
+      "step": 2501
+    },
+    {
+      "epoch": 0.8172798170756604,
+      "grad_norm": 0.24294321238994598,
+      "learning_rate": 8.266000757873699e-06,
+      "loss": 0.756,
+      "step": 2502
+    },
+    {
+      "epoch": 0.8176064676820056,
+      "grad_norm": 0.25103679299354553,
+      "learning_rate": 8.23729239611839e-06,
+      "loss": 0.7447,
+      "step": 2503
+    },
+    {
+      "epoch": 0.8179331182883508,
+      "grad_norm": 0.2764917016029358,
+      "learning_rate": 8.208629498279302e-06,
+      "loss": 0.8186,
+      "step": 2504
+    },
+    {
+      "epoch": 0.818259768894696,
+      "grad_norm": 0.281490296125412,
+      "learning_rate": 8.180012095559525e-06,
+      "loss": 0.8326,
+      "step": 2505
+    },
+    {
+      "epoch": 0.8185864195010412,
+      "grad_norm": 0.2983465790748596,
+      "learning_rate": 8.151440219112721e-06,
+      "loss": 0.8574,
+      "step": 2506
+    },
+    {
+      "epoch": 0.8189130701073863,
+      "grad_norm": 0.3088802397251129,
+      "learning_rate": 8.122913900042905e-06,
+      "loss": 0.8248,
+      "step": 2507
+    },
+    {
+      "epoch": 0.8192397207137315,
+      "grad_norm": 0.3155342638492584,
+      "learning_rate": 8.094433169404547e-06,
+      "loss": 0.8865,
+      "step": 2508
+    },
+    {
+      "epoch": 0.8195663713200768,
+      "grad_norm": 0.31594693660736084,
+      "learning_rate": 8.065998058202468e-06,
+      "loss": 0.8237,
+      "step": 2509
+    },
+    {
+      "epoch": 0.819893021926422,
+      "grad_norm": 0.3399094343185425,
+      "learning_rate": 8.037608597391844e-06,
+      "loss": 0.9411,
+      "step": 2510
+    },
+    {
+      "epoch": 0.8202196725327672,
+      "grad_norm": 0.3390636742115021,
+      "learning_rate": 8.009264817878132e-06,
+      "loss": 0.7855,
+      "step": 2511
+    },
+    {
+      "epoch": 0.8205463231391124,
+      "grad_norm": 0.36510950326919556,
+      "learning_rate": 7.980966750517094e-06,
+      "loss": 0.8056,
+      "step": 2512
+    },
+    {
+      "epoch": 0.8208729737454575,
+      "grad_norm": 0.3949703574180603,
+      "learning_rate": 7.952714426114665e-06,
+      "loss": 0.8022,
+      "step": 2513
+    },
+    {
+      "epoch": 0.8211996243518027,
+      "grad_norm": 0.44621437788009644,
+      "learning_rate": 7.924507875427068e-06,
+      "loss": 0.9611,
+      "step": 2514
+    },
+    {
+      "epoch": 0.8215262749581479,
+      "grad_norm": 0.4569244384765625,
+      "learning_rate": 7.896347129160625e-06,
+      "loss": 0.9466,
+      "step": 2515
+    },
+    {
+      "epoch": 0.8218529255644931,
+      "grad_norm": 0.5023897290229797,
+      "learning_rate": 7.868232217971821e-06,
+      "loss": 1.044,
+      "step": 2516
+    },
+    {
+      "epoch": 0.8221795761708383,
+      "grad_norm": 0.5417917966842651,
+      "learning_rate": 7.840163172467257e-06,
+      "loss": 1.0372,
+      "step": 2517
+    },
+    {
+      "epoch": 0.8225062267771834,
+      "grad_norm": 0.5474960207939148,
+      "learning_rate": 7.812140023203579e-06,
+      "loss": 1.061,
+      "step": 2518
+    },
+    {
+      "epoch": 0.8228328773835286,
+      "grad_norm": 0.6427183747291565,
+      "learning_rate": 7.784162800687484e-06,
+      "loss": 1.0506,
+      "step": 2519
+    },
+    {
+      "epoch": 0.8231595279898738,
+      "grad_norm": 0.6729391813278198,
+      "learning_rate": 7.756231535375674e-06,
+      "loss": 1.2564,
+      "step": 2520
+    },
+    {
+      "epoch": 0.823486178596219,
+      "grad_norm": 0.8698153495788574,
+      "learning_rate": 7.728346257674801e-06,
+      "loss": 1.2879,
+      "step": 2521
+    },
+    {
+      "epoch": 0.8238128292025642,
+      "grad_norm": 1.0003093481063843,
+      "learning_rate": 7.70050699794148e-06,
+      "loss": 1.3512,
+      "step": 2522
+    },
+    {
+      "epoch": 0.8241394798089094,
+      "grad_norm": 1.1123592853546143,
+      "learning_rate": 7.672713786482217e-06,
+      "loss": 1.3098,
+      "step": 2523
+    },
+    {
+      "epoch": 0.8244661304152546,
+      "grad_norm": 1.284902572631836,
+      "learning_rate": 7.644966653553388e-06,
+      "loss": 1.37,
+      "step": 2524
+    },
+    {
+      "epoch": 0.8247927810215998,
+      "grad_norm": 1.9688411951065063,
+      "learning_rate": 7.617265629361209e-06,
+      "loss": 1.9536,
+      "step": 2525
+    },
+    {
+      "epoch": 0.825119431627945,
+      "grad_norm": 0.21682727336883545,
+      "learning_rate": 7.589610744061681e-06,
+      "loss": 0.7734,
+      "step": 2526
+    },
+    {
+      "epoch": 0.8254460822342902,
+      "grad_norm": 0.24609944224357605,
+      "learning_rate": 7.562002027760634e-06,
+      "loss": 0.7812,
+      "step": 2527
+    },
+    {
+      "epoch": 0.8257727328406353,
+      "grad_norm": 0.2850510776042938,
+      "learning_rate": 7.5344395105135635e-06,
+      "loss": 0.861,
+      "step": 2528
+    },
+    {
+      "epoch": 0.8260993834469805,
+      "grad_norm": 0.2798521816730499,
+      "learning_rate": 7.506923222325724e-06,
+      "loss": 0.8429,
+      "step": 2529
+    },
+    {
+      "epoch": 0.8264260340533257,
+      "grad_norm": 0.28509312868118286,
+      "learning_rate": 7.479453193152025e-06,
+      "loss": 0.8388,
+      "step": 2530
+    },
+    {
+      "epoch": 0.8267526846596709,
+      "grad_norm": 0.29365110397338867,
+      "learning_rate": 7.452029452897019e-06,
+      "loss": 0.8429,
+      "step": 2531
+    },
+    {
+      "epoch": 0.8270793352660161,
+      "grad_norm": 0.2949029207229614,
+      "learning_rate": 7.424652031414864e-06,
+      "loss": 0.7964,
+      "step": 2532
+    },
+    {
+      "epoch": 0.8274059858723612,
+      "grad_norm": 0.31028300523757935,
+      "learning_rate": 7.397320958509308e-06,
+      "loss": 0.8371,
+      "step": 2533
+    },
+    {
+      "epoch": 0.8277326364787064,
+      "grad_norm": 0.3249179720878601,
+      "learning_rate": 7.370036263933621e-06,
+      "loss": 0.9011,
+      "step": 2534
+    },
+    {
+      "epoch": 0.8280592870850516,
+      "grad_norm": 0.32227975130081177,
+      "learning_rate": 7.342797977390603e-06,
+      "loss": 0.8263,
+      "step": 2535
+    },
+    {
+      "epoch": 0.8283859376913968,
+      "grad_norm": 0.34464603662490845,
+      "learning_rate": 7.315606128532526e-06,
+      "loss": 0.885,
+      "step": 2536
+    },
+    {
+      "epoch": 0.8287125882977421,
+      "grad_norm": 0.3631193935871124,
+      "learning_rate": 7.2884607469611066e-06,
+      "loss": 0.9489,
+      "step": 2537
+    },
+    {
+      "epoch": 0.8290392389040873,
+      "grad_norm": 0.3989101052284241,
+      "learning_rate": 7.261361862227495e-06,
+      "loss": 0.854,
+      "step": 2538
+    },
+    {
+      "epoch": 0.8293658895104324,
+      "grad_norm": 0.40562012791633606,
+      "learning_rate": 7.234309503832165e-06,
+      "loss": 0.9495,
+      "step": 2539
+    },
+    {
+      "epoch": 0.8296925401167776,
+      "grad_norm": 0.44497793912887573,
+      "learning_rate": 7.207303701225037e-06,
+      "loss": 0.9858,
+      "step": 2540
+    },
+    {
+      "epoch": 0.8300191907231228,
+      "grad_norm": 0.4510423243045807,
+      "learning_rate": 7.180344483805257e-06,
+      "loss": 1.0368,
+      "step": 2541
+    },
+    {
+      "epoch": 0.830345841329468,
+      "grad_norm": 0.5080751776695251,
+      "learning_rate": 7.153431880921302e-06,
+      "loss": 0.9558,
+      "step": 2542
+    },
+    {
+      "epoch": 0.8306724919358132,
+      "grad_norm": 0.5267191529273987,
+      "learning_rate": 7.126565921870909e-06,
+      "loss": 1.0393,
+      "step": 2543
+    },
+    {
+      "epoch": 0.8309991425421583,
+      "grad_norm": 0.6253758668899536,
+      "learning_rate": 7.099746635901017e-06,
+      "loss": 0.9967,
+      "step": 2544
+    },
+    {
+      "epoch": 0.8313257931485035,
+      "grad_norm": 0.6861200332641602,
+      "learning_rate": 7.072974052207765e-06,
+      "loss": 1.1487,
+      "step": 2545
+    },
+    {
+      "epoch": 0.8316524437548487,
+      "grad_norm": 0.842881977558136,
+      "learning_rate": 7.0462481999364614e-06,
+      "loss": 1.1706,
+      "step": 2546
+    },
+    {
+      "epoch": 0.8319790943611939,
+      "grad_norm": 0.9228523373603821,
+      "learning_rate": 7.019569108181495e-06,
+      "loss": 1.2162,
+      "step": 2547
+    },
+    {
+      "epoch": 0.8323057449675391,
+      "grad_norm": 1.1753313541412354,
+      "learning_rate": 6.992936805986433e-06,
+      "loss": 1.362,
+      "step": 2548
+    },
+    {
+      "epoch": 0.8326323955738842,
+      "grad_norm": 1.3156818151474,
+      "learning_rate": 6.966351322343806e-06,
+      "loss": 1.5403,
+      "step": 2549
+    },
+    {
+      "epoch": 0.8329590461802294,
+      "grad_norm": 1.4795371294021606,
+      "learning_rate": 6.939812686195285e-06,
+      "loss": 1.6863,
+      "step": 2550
+    },
+    {
+      "epoch": 0.8332856967865747,
+      "grad_norm": 0.21376070380210876,
+      "learning_rate": 6.913320926431443e-06,
+      "loss": 0.7305,
+      "step": 2551
+    },
+    {
+      "epoch": 0.8336123473929199,
+      "grad_norm": 0.2572622299194336,
+      "learning_rate": 6.886876071891879e-06,
+      "loss": 0.8207,
+      "step": 2552
+    },
+    {
+      "epoch": 0.8339389979992651,
+      "grad_norm": 0.26194846630096436,
+      "learning_rate": 6.8604781513651185e-06,
+      "loss": 0.7989,
+      "step": 2553
+    },
+    {
+      "epoch": 0.8342656486056103,
+      "grad_norm": 0.28000107407569885,
+      "learning_rate": 6.834127193588586e-06,
+      "loss": 0.8771,
+      "step": 2554
+    },
+    {
+      "epoch": 0.8345922992119554,
+      "grad_norm": 0.2950563430786133,
+      "learning_rate": 6.807823227248583e-06,
+      "loss": 0.8277,
+      "step": 2555
+    },
+    {
+      "epoch": 0.8349189498183006,
+      "grad_norm": 0.2996322214603424,
+      "learning_rate": 6.781566280980267e-06,
+      "loss": 0.8208,
+      "step": 2556
+    },
+    {
+      "epoch": 0.8352456004246458,
+      "grad_norm": 0.3321259915828705,
+      "learning_rate": 6.755356383367589e-06,
+      "loss": 0.8929,
+      "step": 2557
+    },
+    {
+      "epoch": 0.835572251030991,
+      "grad_norm": 0.3215591013431549,
+      "learning_rate": 6.7291935629432974e-06,
+      "loss": 0.846,
+      "step": 2558
+    },
+    {
+      "epoch": 0.8358989016373362,
+      "grad_norm": 0.34042298793792725,
+      "learning_rate": 6.703077848188899e-06,
+      "loss": 0.8229,
+      "step": 2559
+    },
+    {
+      "epoch": 0.8362255522436813,
+      "grad_norm": 0.367233544588089,
+      "learning_rate": 6.677009267534567e-06,
+      "loss": 0.9275,
+      "step": 2560
+    },
+    {
+      "epoch": 0.8365522028500265,
+      "grad_norm": 0.3613315522670746,
+      "learning_rate": 6.650987849359253e-06,
+      "loss": 0.8224,
+      "step": 2561
+    },
+    {
+      "epoch": 0.8368788534563717,
+      "grad_norm": 0.38285353779792786,
+      "learning_rate": 6.625013621990472e-06,
+      "loss": 0.8436,
+      "step": 2562
+    },
+    {
+      "epoch": 0.8372055040627169,
+      "grad_norm": 0.41062045097351074,
+      "learning_rate": 6.599086613704447e-06,
+      "loss": 0.9488,
+      "step": 2563
+    },
+    {
+      "epoch": 0.8375321546690621,
+      "grad_norm": 0.443733811378479,
+      "learning_rate": 6.5732068527259325e-06,
+      "loss": 0.94,
+      "step": 2564
+    },
+    {
+      "epoch": 0.8378588052754073,
+      "grad_norm": 0.4647623300552368,
+      "learning_rate": 6.547374367228287e-06,
+      "loss": 0.9088,
+      "step": 2565
+    },
+    {
+      "epoch": 0.8381854558817525,
+      "grad_norm": 0.5017650127410889,
+      "learning_rate": 6.5215891853333934e-06,
+      "loss": 1.1312,
+      "step": 2566
+    },
+    {
+      "epoch": 0.8385121064880977,
+      "grad_norm": 0.5342419743537903,
+      "learning_rate": 6.495851335111636e-06,
+      "loss": 1.0748,
+      "step": 2567
+    },
+    {
+      "epoch": 0.8388387570944429,
+      "grad_norm": 0.5602555871009827,
+      "learning_rate": 6.47016084458188e-06,
+      "loss": 1.0645,
+      "step": 2568
+    },
+    {
+      "epoch": 0.8391654077007881,
+      "grad_norm": 0.6325632929801941,
+      "learning_rate": 6.44451774171142e-06,
+      "loss": 1.0813,
+      "step": 2569
+    },
+    {
+      "epoch": 0.8394920583071332,
+      "grad_norm": 0.643717885017395,
+      "learning_rate": 6.418922054415982e-06,
+      "loss": 1.277,
+      "step": 2570
+    },
+    {
+      "epoch": 0.8398187089134784,
+      "grad_norm": 0.9045091271400452,
+      "learning_rate": 6.393373810559655e-06,
+      "loss": 1.2554,
+      "step": 2571
+    },
+    {
+      "epoch": 0.8401453595198236,
+      "grad_norm": 0.998929500579834,
+      "learning_rate": 6.367873037954908e-06,
+      "loss": 1.3071,
+      "step": 2572
+    },
+    {
+      "epoch": 0.8404720101261688,
+      "grad_norm": 1.309383511543274,
+      "learning_rate": 6.342419764362473e-06,
+      "loss": 1.4713,
+      "step": 2573
+    },
+    {
+      "epoch": 0.840798660732514,
+      "grad_norm": 1.271360993385315,
+      "learning_rate": 6.317014017491463e-06,
+      "loss": 1.7116,
+      "step": 2574
+    },
+    {
+      "epoch": 0.8411253113388591,
+      "grad_norm": 1.6010202169418335,
+      "learning_rate": 6.291655824999154e-06,
+      "loss": 1.7152,
+      "step": 2575
+    },
+    {
+      "epoch": 0.8414519619452043,
+      "grad_norm": 0.22518090903759003,
+      "learning_rate": 6.266345214491148e-06,
+      "loss": 0.761,
+      "step": 2576
+    },
+    {
+      "epoch": 0.8417786125515495,
+      "grad_norm": 0.23476476967334747,
+      "learning_rate": 6.241082213521166e-06,
+      "loss": 0.6972,
+      "step": 2577
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.2612394392490387,
+      "learning_rate": 6.2158668495911455e-06,
+      "loss": 0.7726,
+      "step": 2578
+    },
+    {
+      "epoch": 0.8424319137642399,
+      "grad_norm": 0.2693740129470825,
+      "learning_rate": 6.190699150151158e-06,
+      "loss": 0.7799,
+      "step": 2579
+    },
+    {
+      "epoch": 0.8427585643705852,
+      "grad_norm": 0.3026430606842041,
+      "learning_rate": 6.165579142599387e-06,
+      "loss": 0.8576,
+      "step": 2580
+    },
+    {
+      "epoch": 0.8430852149769303,
+      "grad_norm": 0.29252561926841736,
+      "learning_rate": 6.140506854282085e-06,
+      "loss": 0.8993,
+      "step": 2581
+    },
+    {
+      "epoch": 0.8434118655832755,
+      "grad_norm": 0.3165540397167206,
+      "learning_rate": 6.115482312493581e-06,
+      "loss": 0.8733,
+      "step": 2582
+    },
+    {
+      "epoch": 0.8437385161896207,
+      "grad_norm": 0.33594945073127747,
+      "learning_rate": 6.090505544476183e-06,
+      "loss": 0.8523,
+      "step": 2583
+    },
+    {
+      "epoch": 0.8440651667959659,
+      "grad_norm": 0.3217700719833374,
+      "learning_rate": 6.065576577420245e-06,
+      "loss": 0.8438,
+      "step": 2584
+    },
+    {
+      "epoch": 0.8443918174023111,
+      "grad_norm": 0.3341309130191803,
+      "learning_rate": 6.040695438464045e-06,
+      "loss": 0.8125,
+      "step": 2585
+    },
+    {
+      "epoch": 0.8447184680086562,
+      "grad_norm": 0.34477731585502625,
+      "learning_rate": 6.015862154693791e-06,
+      "loss": 0.9042,
+      "step": 2586
+    },
+    {
+      "epoch": 0.8450451186150014,
+      "grad_norm": 0.3631967306137085,
+      "learning_rate": 5.991076753143648e-06,
+      "loss": 0.8595,
+      "step": 2587
+    },
+    {
+      "epoch": 0.8453717692213466,
+      "grad_norm": 0.3938636779785156,
+      "learning_rate": 5.966339260795578e-06,
+      "loss": 0.9562,
+      "step": 2588
+    },
+    {
+      "epoch": 0.8456984198276918,
+      "grad_norm": 0.41510850191116333,
+      "learning_rate": 5.94164970457946e-06,
+      "loss": 0.8708,
+      "step": 2589
+    },
+    {
+      "epoch": 0.846025070434037,
+      "grad_norm": 0.45071718096733093,
+      "learning_rate": 5.917008111372929e-06,
+      "loss": 0.9527,
+      "step": 2590
+    },
+    {
+      "epoch": 0.8463517210403821,
+      "grad_norm": 0.48741504549980164,
+      "learning_rate": 5.892414508001443e-06,
+      "loss": 0.9604,
+      "step": 2591
+    },
+    {
+      "epoch": 0.8466783716467273,
+      "grad_norm": 0.4795851409435272,
+      "learning_rate": 5.867868921238212e-06,
+      "loss": 0.9465,
+      "step": 2592
+    },
+    {
+      "epoch": 0.8470050222530725,
+      "grad_norm": 0.5671098828315735,
+      "learning_rate": 5.8433713778041575e-06,
+      "loss": 1.069,
+      "step": 2593
+    },
+    {
+      "epoch": 0.8473316728594178,
+      "grad_norm": 0.5778964757919312,
+      "learning_rate": 5.818921904367919e-06,
+      "loss": 1.0361,
+      "step": 2594
+    },
+    {
+      "epoch": 0.847658323465763,
+      "grad_norm": 0.702641487121582,
+      "learning_rate": 5.794520527545805e-06,
+      "loss": 1.1656,
+      "step": 2595
+    },
+    {
+      "epoch": 0.8479849740721082,
+      "grad_norm": 0.7841630578041077,
+      "learning_rate": 5.770167273901728e-06,
+      "loss": 1.2315,
+      "step": 2596
+    },
+    {
+      "epoch": 0.8483116246784533,
+      "grad_norm": 1.0163825750350952,
+      "learning_rate": 5.745862169947286e-06,
+      "loss": 1.1117,
+      "step": 2597
+    },
+    {
+      "epoch": 0.8486382752847985,
+      "grad_norm": 1.122326135635376,
+      "learning_rate": 5.721605242141575e-06,
+      "loss": 1.3777,
+      "step": 2598
+    },
+    {
+      "epoch": 0.8489649258911437,
+      "grad_norm": 1.3500094413757324,
+      "learning_rate": 5.697396516891313e-06,
+      "loss": 1.7499,
+      "step": 2599
+    },
+    {
+      "epoch": 0.8492915764974889,
+      "grad_norm": 1.8633354902267456,
+      "learning_rate": 5.673236020550704e-06,
+      "loss": 1.8196,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8496182271038341,
+      "grad_norm": 0.20166555047035217,
+      "learning_rate": 5.6491237794214665e-06,
+      "loss": 0.6405,
+      "step": 2601
+    },
+    {
+      "epoch": 0.8499448777101792,
+      "grad_norm": 0.24433058500289917,
+      "learning_rate": 5.625059819752782e-06,
+      "loss": 0.8018,
+      "step": 2602
+    },
+    {
+      "epoch": 0.8502715283165244,
+      "grad_norm": 0.2631356120109558,
+      "learning_rate": 5.601044167741271e-06,
+      "loss": 0.8205,
+      "step": 2603
+    },
+    {
+      "epoch": 0.8505981789228696,
+      "grad_norm": 0.2602408230304718,
+      "learning_rate": 5.577076849530971e-06,
+      "loss": 0.8579,
+      "step": 2604
+    },
+    {
+      "epoch": 0.8509248295292148,
+      "grad_norm": 0.27190256118774414,
+      "learning_rate": 5.55315789121329e-06,
+      "loss": 0.821,
+      "step": 2605
+    },
+    {
+      "epoch": 0.85125148013556,
+      "grad_norm": 0.30052924156188965,
+      "learning_rate": 5.529287318826998e-06,
+      "loss": 0.8228,
+      "step": 2606
+    },
+    {
+      "epoch": 0.8515781307419051,
+      "grad_norm": 0.3002735674381256,
+      "learning_rate": 5.505465158358198e-06,
+      "loss": 0.754,
+      "step": 2607
+    },
+    {
+      "epoch": 0.8519047813482504,
+      "grad_norm": 0.3288699686527252,
+      "learning_rate": 5.481691435740283e-06,
+      "loss": 0.931,
+      "step": 2608
+    },
+    {
+      "epoch": 0.8522314319545956,
+      "grad_norm": 0.3346864581108093,
+      "learning_rate": 5.457966176853896e-06,
+      "loss": 0.9369,
+      "step": 2609
+    },
+    {
+      "epoch": 0.8525580825609408,
+      "grad_norm": 0.34095606207847595,
+      "learning_rate": 5.434289407526971e-06,
+      "loss": 0.8721,
+      "step": 2610
+    },
+    {
+      "epoch": 0.852884733167286,
+      "grad_norm": 0.36173370480537415,
+      "learning_rate": 5.410661153534602e-06,
+      "loss": 0.8597,
+      "step": 2611
+    },
+    {
+      "epoch": 0.8532113837736311,
+      "grad_norm": 0.38429251313209534,
+      "learning_rate": 5.3870814405990955e-06,
+      "loss": 0.955,
+      "step": 2612
+    },
+    {
+      "epoch": 0.8535380343799763,
+      "grad_norm": 0.4128096401691437,
+      "learning_rate": 5.363550294389907e-06,
+      "loss": 0.8608,
+      "step": 2613
+    },
+    {
+      "epoch": 0.8538646849863215,
+      "grad_norm": 0.4228519797325134,
+      "learning_rate": 5.340067740523635e-06,
+      "loss": 0.8909,
+      "step": 2614
+    },
+    {
+      "epoch": 0.8541913355926667,
+      "grad_norm": 0.4561985433101654,
+      "learning_rate": 5.316633804563958e-06,
+      "loss": 0.9841,
+      "step": 2615
+    },
+    {
+      "epoch": 0.8545179861990119,
+      "grad_norm": 0.5019810795783997,
+      "learning_rate": 5.293248512021648e-06,
+      "loss": 0.9614,
+      "step": 2616
+    },
+    {
+      "epoch": 0.854844636805357,
+      "grad_norm": 0.5194594264030457,
+      "learning_rate": 5.2699118883545e-06,
+      "loss": 0.9494,
+      "step": 2617
+    },
+    {
+      "epoch": 0.8551712874117022,
+      "grad_norm": 0.5753628015518188,
+      "learning_rate": 5.24662395896735e-06,
+      "loss": 1.0706,
+      "step": 2618
+    },
+    {
+      "epoch": 0.8554979380180474,
+      "grad_norm": 0.6136693358421326,
+      "learning_rate": 5.223384749212013e-06,
+      "loss": 1.0661,
+      "step": 2619
+    },
+    {
+      "epoch": 0.8558245886243926,
+      "grad_norm": 0.6912145018577576,
+      "learning_rate": 5.200194284387261e-06,
+      "loss": 1.1086,
+      "step": 2620
+    },
+    {
+      "epoch": 0.8561512392307378,
+      "grad_norm": 0.8217900395393372,
+      "learning_rate": 5.1770525897388234e-06,
+      "loss": 1.1226,
+      "step": 2621
+    },
+    {
+      "epoch": 0.8564778898370831,
+      "grad_norm": 0.910292387008667,
+      "learning_rate": 5.153959690459287e-06,
+      "loss": 1.4569,
+      "step": 2622
+    },
+    {
+      "epoch": 0.8568045404434282,
+      "grad_norm": 1.232712745666504,
+      "learning_rate": 5.130915611688192e-06,
+      "loss": 1.2706,
+      "step": 2623
+    },
+    {
+      "epoch": 0.8571311910497734,
+      "grad_norm": 1.3801976442337036,
+      "learning_rate": 5.107920378511871e-06,
+      "loss": 1.5923,
+      "step": 2624
+    },
+    {
+      "epoch": 0.8574578416561186,
+      "grad_norm": 1.6169432401657104,
+      "learning_rate": 5.084974015963506e-06,
+      "loss": 1.8258,
+      "step": 2625
+    },
+    {
+      "epoch": 0.8577844922624638,
+      "grad_norm": 0.20021358132362366,
+      "learning_rate": 5.062076549023076e-06,
+      "loss": 0.7602,
+      "step": 2626
+    },
+    {
+      "epoch": 0.858111142868809,
+      "grad_norm": 0.24645954370498657,
+      "learning_rate": 5.03922800261733e-06,
+      "loss": 0.7846,
+      "step": 2627
+    },
+    {
+      "epoch": 0.8584377934751541,
+      "grad_norm": 0.25612515211105347,
+      "learning_rate": 5.01642840161976e-06,
+      "loss": 0.8028,
+      "step": 2628
+    },
+    {
+      "epoch": 0.8587644440814993,
+      "grad_norm": 0.27695709466934204,
+      "learning_rate": 4.993677770850591e-06,
+      "loss": 0.799,
+      "step": 2629
+    },
+    {
+      "epoch": 0.8590910946878445,
+      "grad_norm": 0.29534533619880676,
+      "learning_rate": 4.970976135076683e-06,
+      "loss": 0.8947,
+      "step": 2630
+    },
+    {
+      "epoch": 0.8594177452941897,
+      "grad_norm": 0.3191416263580322,
+      "learning_rate": 4.948323519011644e-06,
+      "loss": 0.8109,
+      "step": 2631
+    },
+    {
+      "epoch": 0.8597443959005349,
+      "grad_norm": 0.32522448897361755,
+      "learning_rate": 4.925719947315632e-06,
+      "loss": 0.8372,
+      "step": 2632
+    },
+    {
+      "epoch": 0.86007104650688,
+      "grad_norm": 0.3430340886116028,
+      "learning_rate": 4.9031654445954925e-06,
+      "loss": 0.8355,
+      "step": 2633
+    },
+    {
+      "epoch": 0.8603976971132252,
+      "grad_norm": 0.33849358558654785,
+      "learning_rate": 4.8806600354045824e-06,
+      "loss": 0.8631,
+      "step": 2634
+    },
+    {
+      "epoch": 0.8607243477195704,
+      "grad_norm": 0.349334180355072,
+      "learning_rate": 4.858203744242857e-06,
+      "loss": 0.8974,
+      "step": 2635
+    },
+    {
+      "epoch": 0.8610509983259157,
+      "grad_norm": 0.38946714997291565,
+      "learning_rate": 4.835796595556796e-06,
+      "loss": 0.8642,
+      "step": 2636
+    },
+    {
+      "epoch": 0.8613776489322609,
+      "grad_norm": 0.37505602836608887,
+      "learning_rate": 4.813438613739374e-06,
+      "loss": 0.8739,
+      "step": 2637
+    },
+    {
+      "epoch": 0.8617042995386061,
+      "grad_norm": 0.4225607216358185,
+      "learning_rate": 4.791129823130036e-06,
+      "loss": 0.9068,
+      "step": 2638
+    },
+    {
+      "epoch": 0.8620309501449512,
+      "grad_norm": 0.4245143234729767,
+      "learning_rate": 4.768870248014695e-06,
+      "loss": 0.9315,
+      "step": 2639
+    },
+    {
+      "epoch": 0.8623576007512964,
+      "grad_norm": 0.5012677907943726,
+      "learning_rate": 4.746659912625662e-06,
+      "loss": 0.9509,
+      "step": 2640
+    },
+    {
+      "epoch": 0.8626842513576416,
+      "grad_norm": 0.4809205234050751,
+      "learning_rate": 4.724498841141667e-06,
+      "loss": 0.9483,
+      "step": 2641
+    },
+    {
+      "epoch": 0.8630109019639868,
+      "grad_norm": 0.5976522564888,
+      "learning_rate": 4.702387057687802e-06,
+      "loss": 1.1778,
+      "step": 2642
+    },
+    {
+      "epoch": 0.863337552570332,
+      "grad_norm": 0.6282837390899658,
+      "learning_rate": 4.6803245863354775e-06,
+      "loss": 1.1471,
+      "step": 2643
+    },
+    {
+      "epoch": 0.8636642031766771,
+      "grad_norm": 0.6567032933235168,
+      "learning_rate": 4.658311451102487e-06,
+      "loss": 1.1462,
+      "step": 2644
+    },
+    {
+      "epoch": 0.8639908537830223,
+      "grad_norm": 0.7036363482475281,
+      "learning_rate": 4.636347675952829e-06,
+      "loss": 1.1844,
+      "step": 2645
+    },
+    {
+      "epoch": 0.8643175043893675,
+      "grad_norm": 0.8441754579544067,
+      "learning_rate": 4.614433284796848e-06,
+      "loss": 1.229,
+      "step": 2646
+    },
+    {
+      "epoch": 0.8646441549957127,
+      "grad_norm": 1.0409406423568726,
+      "learning_rate": 4.592568301491073e-06,
+      "loss": 1.2641,
+      "step": 2647
+    },
+    {
+      "epoch": 0.8649708056020579,
+      "grad_norm": 1.464299201965332,
+      "learning_rate": 4.5707527498382695e-06,
+      "loss": 1.3463,
+      "step": 2648
+    },
+    {
+      "epoch": 0.865297456208403,
+      "grad_norm": 1.223347783088684,
+      "learning_rate": 4.548986653587389e-06,
+      "loss": 1.6237,
+      "step": 2649
+    },
+    {
+      "epoch": 0.8656241068147483,
+      "grad_norm": 1.9410229921340942,
+      "learning_rate": 4.527270036433539e-06,
+      "loss": 1.4677,
+      "step": 2650
+    },
+    {
+      "epoch": 0.8659507574210935,
+      "grad_norm": 0.21767649054527283,
+      "learning_rate": 4.5056029220179706e-06,
+      "loss": 0.7965,
+      "step": 2651
+    },
+    {
+      "epoch": 0.8662774080274387,
+      "grad_norm": 0.2621896266937256,
+      "learning_rate": 4.483985333928031e-06,
+      "loss": 0.8804,
+      "step": 2652
+    },
+    {
+      "epoch": 0.8666040586337839,
+      "grad_norm": 0.27004051208496094,
+      "learning_rate": 4.462417295697175e-06,
+      "loss": 0.7977,
+      "step": 2653
+    },
+    {
+      "epoch": 0.866930709240129,
+      "grad_norm": 0.28397756814956665,
+      "learning_rate": 4.440898830804891e-06,
+      "loss": 0.8856,
+      "step": 2654
+    },
+    {
+      "epoch": 0.8672573598464742,
+      "grad_norm": 0.28461506962776184,
+      "learning_rate": 4.419429962676725e-06,
+      "loss": 0.8844,
+      "step": 2655
+    },
+    {
+      "epoch": 0.8675840104528194,
+      "grad_norm": 0.29285040497779846,
+      "learning_rate": 4.398010714684186e-06,
+      "loss": 0.8644,
+      "step": 2656
+    },
+    {
+      "epoch": 0.8679106610591646,
+      "grad_norm": 0.2863633334636688,
+      "learning_rate": 4.376641110144841e-06,
+      "loss": 0.7856,
+      "step": 2657
+    },
+    {
+      "epoch": 0.8682373116655098,
+      "grad_norm": 0.305578351020813,
+      "learning_rate": 4.35532117232213e-06,
+      "loss": 0.8655,
+      "step": 2658
+    },
+    {
+      "epoch": 0.868563962271855,
+      "grad_norm": 0.3179665505886078,
+      "learning_rate": 4.334050924425504e-06,
+      "loss": 0.8276,
+      "step": 2659
+    },
+    {
+      "epoch": 0.8688906128782001,
+      "grad_norm": 0.33530572056770325,
+      "learning_rate": 4.312830389610245e-06,
+      "loss": 0.8437,
+      "step": 2660
+    },
+    {
+      "epoch": 0.8692172634845453,
+      "grad_norm": 0.3526410460472107,
+      "learning_rate": 4.29165959097757e-06,
+      "loss": 0.9232,
+      "step": 2661
+    },
+    {
+      "epoch": 0.8695439140908905,
+      "grad_norm": 0.373531699180603,
+      "learning_rate": 4.270538551574532e-06,
+      "loss": 0.8388,
+      "step": 2662
+    },
+    {
+      "epoch": 0.8698705646972357,
+      "grad_norm": 0.3668655753135681,
+      "learning_rate": 4.249467294394011e-06,
+      "loss": 0.8524,
+      "step": 2663
+    },
+    {
+      "epoch": 0.870197215303581,
+      "grad_norm": 0.4148674011230469,
+      "learning_rate": 4.228445842374706e-06,
+      "loss": 0.9739,
+      "step": 2664
+    },
+    {
+      "epoch": 0.8705238659099261,
+      "grad_norm": 0.4478146433830261,
+      "learning_rate": 4.2074742184010915e-06,
+      "loss": 1.0286,
+      "step": 2665
+    },
+    {
+      "epoch": 0.8708505165162713,
+      "grad_norm": 0.4490779936313629,
+      "learning_rate": 4.186552445303377e-06,
+      "loss": 0.9902,
+      "step": 2666
+    },
+    {
+      "epoch": 0.8711771671226165,
+      "grad_norm": 0.5150294303894043,
+      "learning_rate": 4.165680545857553e-06,
+      "loss": 0.9668,
+      "step": 2667
+    },
+    {
+      "epoch": 0.8715038177289617,
+      "grad_norm": 0.5332680344581604,
+      "learning_rate": 4.144858542785257e-06,
+      "loss": 1.0923,
+      "step": 2668
+    },
+    {
+      "epoch": 0.8718304683353069,
+      "grad_norm": 0.6344809532165527,
+      "learning_rate": 4.124086458753851e-06,
+      "loss": 1.0483,
+      "step": 2669
+    },
+    {
+      "epoch": 0.872157118941652,
+      "grad_norm": 0.661204993724823,
+      "learning_rate": 4.103364316376334e-06,
+      "loss": 1.1077,
+      "step": 2670
+    },
+    {
+      "epoch": 0.8724837695479972,
+      "grad_norm": 0.816225528717041,
+      "learning_rate": 4.0826921382113415e-06,
+      "loss": 1.1737,
+      "step": 2671
+    },
+    {
+      "epoch": 0.8728104201543424,
+      "grad_norm": 0.9259994029998779,
+      "learning_rate": 4.0620699467631255e-06,
+      "loss": 1.5668,
+      "step": 2672
+    },
+    {
+      "epoch": 0.8731370707606876,
+      "grad_norm": 1.154280662536621,
+      "learning_rate": 4.041497764481511e-06,
+      "loss": 1.2698,
+      "step": 2673
+    },
+    {
+      "epoch": 0.8734637213670328,
+      "grad_norm": 1.2868138551712036,
+      "learning_rate": 4.020975613761874e-06,
+      "loss": 1.3254,
+      "step": 2674
+    },
+    {
+      "epoch": 0.8737903719733779,
+      "grad_norm": 2.1193177700042725,
+      "learning_rate": 4.000503516945148e-06,
+      "loss": 1.5762,
+      "step": 2675
+    },
+    {
+      "epoch": 0.8741170225797231,
+      "grad_norm": 0.2401547133922577,
+      "learning_rate": 3.980081496317761e-06,
+      "loss": 0.795,
+      "step": 2676
+    },
+    {
+      "epoch": 0.8744436731860683,
+      "grad_norm": 0.24791757762432098,
+      "learning_rate": 3.959709574111625e-06,
+      "loss": 0.7803,
+      "step": 2677
+    },
+    {
+      "epoch": 0.8747703237924135,
+      "grad_norm": 0.2674658000469208,
+      "learning_rate": 3.939387772504133e-06,
+      "loss": 0.8197,
+      "step": 2678
+    },
+    {
+      "epoch": 0.8750969743987588,
+      "grad_norm": 0.27707958221435547,
+      "learning_rate": 3.919116113618071e-06,
+      "loss": 0.8578,
+      "step": 2679
+    },
+    {
+      "epoch": 0.875423625005104,
+      "grad_norm": 0.28398212790489197,
+      "learning_rate": 3.898894619521704e-06,
+      "loss": 0.8165,
+      "step": 2680
+    },
+    {
+      "epoch": 0.8757502756114491,
+      "grad_norm": 0.30292728543281555,
+      "learning_rate": 3.8787233122286215e-06,
+      "loss": 0.8953,
+      "step": 2681
+    },
+    {
+      "epoch": 0.8760769262177943,
+      "grad_norm": 0.30174580216407776,
+      "learning_rate": 3.8586022136978206e-06,
+      "loss": 0.8116,
+      "step": 2682
+    },
+    {
+      "epoch": 0.8764035768241395,
+      "grad_norm": 0.3020266890525818,
+      "learning_rate": 3.838531345833624e-06,
+      "loss": 0.7895,
+      "step": 2683
+    },
+    {
+      "epoch": 0.8767302274304847,
+      "grad_norm": 0.3383975028991699,
+      "learning_rate": 3.818510730485675e-06,
+      "loss": 0.9151,
+      "step": 2684
+    },
+    {
+      "epoch": 0.8770568780368299,
+      "grad_norm": 0.356997013092041,
+      "learning_rate": 3.7985403894489067e-06,
+      "loss": 0.8646,
+      "step": 2685
+    },
+    {
+      "epoch": 0.877383528643175,
+      "grad_norm": 0.3583001494407654,
+      "learning_rate": 3.7786203444635348e-06,
+      "loss": 0.9057,
+      "step": 2686
+    },
+    {
+      "epoch": 0.8777101792495202,
+      "grad_norm": 0.37976783514022827,
+      "learning_rate": 3.758750617215007e-06,
+      "loss": 0.9364,
+      "step": 2687
+    },
+    {
+      "epoch": 0.8780368298558654,
+      "grad_norm": 0.3958790898323059,
+      "learning_rate": 3.738931229334003e-06,
+      "loss": 0.8731,
+      "step": 2688
+    },
+    {
+      "epoch": 0.8783634804622106,
+      "grad_norm": 0.4313504993915558,
+      "learning_rate": 3.719162202396398e-06,
+      "loss": 0.9949,
+      "step": 2689
+    },
+    {
+      "epoch": 0.8786901310685558,
+      "grad_norm": 0.4511992037296295,
+      "learning_rate": 3.699443557923238e-06,
+      "loss": 0.8964,
+      "step": 2690
+    },
+    {
+      "epoch": 0.8790167816749009,
+      "grad_norm": 0.46949154138565063,
+      "learning_rate": 3.679775317380746e-06,
+      "loss": 1.008,
+      "step": 2691
+    },
+    {
+      "epoch": 0.8793434322812461,
+      "grad_norm": 0.5014150142669678,
+      "learning_rate": 3.660157502180217e-06,
+      "loss": 1.0773,
+      "step": 2692
+    },
+    {
+      "epoch": 0.8796700828875914,
+      "grad_norm": 0.5346405506134033,
+      "learning_rate": 3.6405901336781346e-06,
+      "loss": 1.1204,
+      "step": 2693
+    },
+    {
+      "epoch": 0.8799967334939366,
+      "grad_norm": 0.6744000911712646,
+      "learning_rate": 3.6210732331759865e-06,
+      "loss": 1.2427,
+      "step": 2694
+    },
+    {
+      "epoch": 0.8803233841002818,
+      "grad_norm": 0.6668764352798462,
+      "learning_rate": 3.601606821920367e-06,
+      "loss": 1.2329,
+      "step": 2695
+    },
+    {
+      "epoch": 0.880650034706627,
+      "grad_norm": 0.7776802182197571,
+      "learning_rate": 3.5821909211028906e-06,
+      "loss": 1.2254,
+      "step": 2696
+    },
+    {
+      "epoch": 0.8809766853129721,
+      "grad_norm": 0.8504863977432251,
+      "learning_rate": 3.5628255518601828e-06,
+      "loss": 1.1201,
+      "step": 2697
+    },
+    {
+      "epoch": 0.8813033359193173,
+      "grad_norm": 1.0933723449707031,
+      "learning_rate": 3.543510735273875e-06,
+      "loss": 1.432,
+      "step": 2698
+    },
+    {
+      "epoch": 0.8816299865256625,
+      "grad_norm": 1.16017746925354,
+      "learning_rate": 3.5242464923705534e-06,
+      "loss": 0.8979,
+      "step": 2699
+    },
+    {
+      "epoch": 0.8819566371320077,
+      "grad_norm": 1.525850534439087,
+      "learning_rate": 3.505032844121725e-06,
+      "loss": 1.4717,
+      "step": 2700
+    },
+    {
+      "epoch": 0.8822832877383529,
+      "grad_norm": 0.21816085278987885,
+      "learning_rate": 3.485869811443876e-06,
+      "loss": 0.7093,
+      "step": 2701
+    },
+    {
+      "epoch": 0.882609938344698,
+      "grad_norm": 0.25412845611572266,
+      "learning_rate": 3.466757415198324e-06,
+      "loss": 0.8078,
+      "step": 2702
+    },
+    {
+      "epoch": 0.8829365889510432,
+      "grad_norm": 0.25287115573883057,
+      "learning_rate": 3.447695676191337e-06,
+      "loss": 0.7362,
+      "step": 2703
+    },
+    {
+      "epoch": 0.8832632395573884,
+      "grad_norm": 0.27101483941078186,
+      "learning_rate": 3.4286846151739603e-06,
+      "loss": 0.8108,
+      "step": 2704
+    },
+    {
+      "epoch": 0.8835898901637336,
+      "grad_norm": 0.27851250767707825,
+      "learning_rate": 3.409724252842117e-06,
+      "loss": 0.7874,
+      "step": 2705
+    },
+    {
+      "epoch": 0.8839165407700788,
+      "grad_norm": 0.29681694507598877,
+      "learning_rate": 3.3908146098365235e-06,
+      "loss": 0.89,
+      "step": 2706
+    },
+    {
+      "epoch": 0.884243191376424,
+      "grad_norm": 0.30123892426490784,
+      "learning_rate": 3.371955706742691e-06,
+      "loss": 0.828,
+      "step": 2707
+    },
+    {
+      "epoch": 0.8845698419827692,
+      "grad_norm": 0.30781540274620056,
+      "learning_rate": 3.353147564090886e-06,
+      "loss": 0.9189,
+      "step": 2708
+    },
+    {
+      "epoch": 0.8848964925891144,
+      "grad_norm": 0.3321038782596588,
+      "learning_rate": 3.3343902023561134e-06,
+      "loss": 0.9027,
+      "step": 2709
+    },
+    {
+      "epoch": 0.8852231431954596,
+      "grad_norm": 0.3396836817264557,
+      "learning_rate": 3.315683641958106e-06,
+      "loss": 0.8452,
+      "step": 2710
+    },
+    {
+      "epoch": 0.8855497938018048,
+      "grad_norm": 0.370717316865921,
+      "learning_rate": 3.297027903261285e-06,
+      "loss": 0.8945,
+      "step": 2711
+    },
+    {
+      "epoch": 0.8858764444081499,
+      "grad_norm": 0.3481248617172241,
+      "learning_rate": 3.27842300657476e-06,
+      "loss": 0.8693,
+      "step": 2712
+    },
+    {
+      "epoch": 0.8862030950144951,
+      "grad_norm": 0.37252771854400635,
+      "learning_rate": 3.2598689721522525e-06,
+      "loss": 0.8701,
+      "step": 2713
+    },
+    {
+      "epoch": 0.8865297456208403,
+      "grad_norm": 0.4177666902542114,
+      "learning_rate": 3.241365820192177e-06,
+      "loss": 0.8574,
+      "step": 2714
+    },
+    {
+      "epoch": 0.8868563962271855,
+      "grad_norm": 0.43080759048461914,
+      "learning_rate": 3.2229135708374936e-06,
+      "loss": 0.9288,
+      "step": 2715
+    },
+    {
+      "epoch": 0.8871830468335307,
+      "grad_norm": 0.46755388379096985,
+      "learning_rate": 3.204512244175806e-06,
+      "loss": 0.894,
+      "step": 2716
+    },
+    {
+      "epoch": 0.8875096974398758,
+      "grad_norm": 0.5186474323272705,
+      "learning_rate": 3.186161860239234e-06,
+      "loss": 1.0247,
+      "step": 2717
+    },
+    {
+      "epoch": 0.887836348046221,
+      "grad_norm": 0.5322409868240356,
+      "learning_rate": 3.1678624390044596e-06,
+      "loss": 0.9993,
+      "step": 2718
+    },
+    {
+      "epoch": 0.8881629986525662,
+      "grad_norm": 0.6192087531089783,
+      "learning_rate": 3.1496140003926924e-06,
+      "loss": 1.0338,
+      "step": 2719
+    },
+    {
+      "epoch": 0.8884896492589114,
+      "grad_norm": 0.7641281485557556,
+      "learning_rate": 3.131416564269629e-06,
+      "loss": 1.188,
+      "step": 2720
+    },
+    {
+      "epoch": 0.8888162998652567,
+      "grad_norm": 0.8901844024658203,
+      "learning_rate": 3.1132701504454464e-06,
+      "loss": 1.2975,
+      "step": 2721
+    },
+    {
+      "epoch": 0.8891429504716019,
+      "grad_norm": 0.9301742911338806,
+      "learning_rate": 3.0951747786747865e-06,
+      "loss": 1.3024,
+      "step": 2722
+    },
+    {
+      "epoch": 0.889469601077947,
+      "grad_norm": 1.1372238397598267,
+      "learning_rate": 3.077130468656719e-06,
+      "loss": 1.4755,
+      "step": 2723
+    },
+    {
+      "epoch": 0.8897962516842922,
+      "grad_norm": 1.380213737487793,
+      "learning_rate": 3.0591372400347153e-06,
+      "loss": 1.8431,
+      "step": 2724
+    },
+    {
+      "epoch": 0.8901229022906374,
+      "grad_norm": 1.8983066082000732,
+      "learning_rate": 3.0411951123966666e-06,
+      "loss": 1.8139,
+      "step": 2725
+    },
+    {
+      "epoch": 0.8904495528969826,
+      "grad_norm": 0.21302877366542816,
+      "learning_rate": 3.0233041052747935e-06,
+      "loss": 0.7034,
+      "step": 2726
+    },
+    {
+      "epoch": 0.8907762035033278,
+      "grad_norm": 0.2391611784696579,
+      "learning_rate": 3.0054642381457108e-06,
+      "loss": 0.7476,
+      "step": 2727
+    },
+    {
+      "epoch": 0.8911028541096729,
+      "grad_norm": 0.2587495744228363,
+      "learning_rate": 2.987675530430317e-06,
+      "loss": 0.8294,
+      "step": 2728
+    },
+    {
+      "epoch": 0.8914295047160181,
+      "grad_norm": 0.2743024230003357,
+      "learning_rate": 2.9699380014938717e-06,
+      "loss": 0.823,
+      "step": 2729
+    },
+    {
+      "epoch": 0.8917561553223633,
+      "grad_norm": 0.2853001654148102,
+      "learning_rate": 2.9522516706458557e-06,
+      "loss": 0.8175,
+      "step": 2730
+    },
+    {
+      "epoch": 0.8920828059287085,
+      "grad_norm": 0.289151132106781,
+      "learning_rate": 2.9346165571400574e-06,
+      "loss": 0.8308,
+      "step": 2731
+    },
+    {
+      "epoch": 0.8924094565350537,
+      "grad_norm": 0.3165534734725952,
+      "learning_rate": 2.9170326801744974e-06,
+      "loss": 0.8892,
+      "step": 2732
+    },
+    {
+      "epoch": 0.8927361071413988,
+      "grad_norm": 0.33104461431503296,
+      "learning_rate": 2.899500058891419e-06,
+      "loss": 0.9116,
+      "step": 2733
+    },
+    {
+      "epoch": 0.893062757747744,
+      "grad_norm": 0.3390517234802246,
+      "learning_rate": 2.8820187123772712e-06,
+      "loss": 0.857,
+      "step": 2734
+    },
+    {
+      "epoch": 0.8933894083540893,
+      "grad_norm": 0.33683860301971436,
+      "learning_rate": 2.8645886596626814e-06,
+      "loss": 0.878,
+      "step": 2735
+    },
+    {
+      "epoch": 0.8937160589604345,
+      "grad_norm": 0.33593204617500305,
+      "learning_rate": 2.8472099197224155e-06,
+      "loss": 0.8384,
+      "step": 2736
+    },
+    {
+      "epoch": 0.8940427095667797,
+      "grad_norm": 0.38355734944343567,
+      "learning_rate": 2.829882511475429e-06,
+      "loss": 0.9405,
+      "step": 2737
+    },
+    {
+      "epoch": 0.8943693601731248,
+      "grad_norm": 0.4030013382434845,
+      "learning_rate": 2.8126064537847497e-06,
+      "loss": 0.843,
+      "step": 2738
+    },
+    {
+      "epoch": 0.89469601077947,
+      "grad_norm": 0.4197559356689453,
+      "learning_rate": 2.7953817654575166e-06,
+      "loss": 0.8953,
+      "step": 2739
+    },
+    {
+      "epoch": 0.8950226613858152,
+      "grad_norm": 0.5002544522285461,
+      "learning_rate": 2.7782084652449747e-06,
+      "loss": 1.0398,
+      "step": 2740
+    },
+    {
+      "epoch": 0.8953493119921604,
+      "grad_norm": 0.4735429286956787,
+      "learning_rate": 2.7610865718423694e-06,
+      "loss": 0.9453,
+      "step": 2741
+    },
+    {
+      "epoch": 0.8956759625985056,
+      "grad_norm": 0.49182236194610596,
+      "learning_rate": 2.744016103889052e-06,
+      "loss": 1.0104,
+      "step": 2742
+    },
+    {
+      "epoch": 0.8960026132048508,
+      "grad_norm": 0.5787506103515625,
+      "learning_rate": 2.7269970799683353e-06,
+      "loss": 1.0149,
+      "step": 2743
+    },
+    {
+      "epoch": 0.8963292638111959,
+      "grad_norm": 0.5907702445983887,
+      "learning_rate": 2.7100295186075543e-06,
+      "loss": 0.9488,
+      "step": 2744
+    },
+    {
+      "epoch": 0.8966559144175411,
+      "grad_norm": 0.7707468271255493,
+      "learning_rate": 2.693113438278011e-06,
+      "loss": 1.2416,
+      "step": 2745
+    },
+    {
+      "epoch": 0.8969825650238863,
+      "grad_norm": 0.7937350869178772,
+      "learning_rate": 2.6762488573949808e-06,
+      "loss": 1.0523,
+      "step": 2746
+    },
+    {
+      "epoch": 0.8973092156302315,
+      "grad_norm": 1.2601395845413208,
+      "learning_rate": 2.6594357943176496e-06,
+      "loss": 1.5764,
+      "step": 2747
+    },
+    {
+      "epoch": 0.8976358662365767,
+      "grad_norm": 1.1137841939926147,
+      "learning_rate": 2.6426742673491487e-06,
+      "loss": 1.2133,
+      "step": 2748
+    },
+    {
+      "epoch": 0.8979625168429219,
+      "grad_norm": 1.296323299407959,
+      "learning_rate": 2.6259642947364716e-06,
+      "loss": 1.1121,
+      "step": 2749
+    },
+    {
+      "epoch": 0.8982891674492671,
+      "grad_norm": 2.165759563446045,
+      "learning_rate": 2.6093058946705275e-06,
+      "loss": 2.157,
+      "step": 2750
+    },
+    {
+      "epoch": 0.8986158180556123,
+      "grad_norm": 0.22275981307029724,
+      "learning_rate": 2.5926990852860487e-06,
+      "loss": 0.7276,
+      "step": 2751
+    },
+    {
+      "epoch": 0.8989424686619575,
+      "grad_norm": 0.24508191645145416,
+      "learning_rate": 2.5761438846616248e-06,
+      "loss": 0.765,
+      "step": 2752
+    },
+    {
+      "epoch": 0.8992691192683027,
+      "grad_norm": 0.2617625594139099,
+      "learning_rate": 2.5596403108196555e-06,
+      "loss": 0.8318,
+      "step": 2753
+    },
+    {
+      "epoch": 0.8995957698746478,
+      "grad_norm": 0.2869446575641632,
+      "learning_rate": 2.5431883817263415e-06,
+      "loss": 0.851,
+      "step": 2754
+    },
+    {
+      "epoch": 0.899922420480993,
+      "grad_norm": 0.2984892725944519,
+      "learning_rate": 2.5267881152916507e-06,
+      "loss": 0.9036,
+      "step": 2755
+    },
+    {
+      "epoch": 0.9002490710873382,
+      "grad_norm": 0.29813170433044434,
+      "learning_rate": 2.5104395293693295e-06,
+      "loss": 0.8355,
+      "step": 2756
+    },
+    {
+      "epoch": 0.9005757216936834,
+      "grad_norm": 0.32043224573135376,
+      "learning_rate": 2.4941426417568468e-06,
+      "loss": 0.8489,
+      "step": 2757
+    },
+    {
+      "epoch": 0.9009023723000286,
+      "grad_norm": 0.33558252453804016,
+      "learning_rate": 2.4778974701954007e-06,
+      "loss": 0.9253,
+      "step": 2758
+    },
+    {
+      "epoch": 0.9012290229063737,
+      "grad_norm": 0.344395250082016,
+      "learning_rate": 2.4617040323698827e-06,
+      "loss": 0.8421,
+      "step": 2759
+    },
+    {
+      "epoch": 0.9015556735127189,
+      "grad_norm": 0.3580136299133301,
+      "learning_rate": 2.445562345908875e-06,
+      "loss": 0.9436,
+      "step": 2760
+    },
+    {
+      "epoch": 0.9018823241190641,
+      "grad_norm": 0.34943464398384094,
+      "learning_rate": 2.4294724283846158e-06,
+      "loss": 0.9067,
+      "step": 2761
+    },
+    {
+      "epoch": 0.9022089747254093,
+      "grad_norm": 0.3668978214263916,
+      "learning_rate": 2.4134342973129777e-06,
+      "loss": 0.918,
+      "step": 2762
+    },
+    {
+      "epoch": 0.9025356253317546,
+      "grad_norm": 0.38507401943206787,
+      "learning_rate": 2.3974479701534826e-06,
+      "loss": 0.8608,
+      "step": 2763
+    },
+    {
+      "epoch": 0.9028622759380998,
+      "grad_norm": 0.42256003618240356,
+      "learning_rate": 2.3815134643092264e-06,
+      "loss": 0.924,
+      "step": 2764
+    },
+    {
+      "epoch": 0.9031889265444449,
+      "grad_norm": 0.4842672348022461,
+      "learning_rate": 2.3656307971269165e-06,
+      "loss": 1.0541,
+      "step": 2765
+    },
+    {
+      "epoch": 0.9035155771507901,
+      "grad_norm": 0.5005545616149902,
+      "learning_rate": 2.349799985896811e-06,
+      "loss": 1.1173,
+      "step": 2766
+    },
+    {
+      "epoch": 0.9038422277571353,
+      "grad_norm": 0.5491520166397095,
+      "learning_rate": 2.334021047852725e-06,
+      "loss": 0.9751,
+      "step": 2767
+    },
+    {
+      "epoch": 0.9041688783634805,
+      "grad_norm": 0.6437044143676758,
+      "learning_rate": 2.318294000172e-06,
+      "loss": 1.1144,
+      "step": 2768
+    },
+    {
+      "epoch": 0.9044955289698257,
+      "grad_norm": 0.6922833919525146,
+      "learning_rate": 2.3026188599754915e-06,
+      "loss": 1.0302,
+      "step": 2769
+    },
+    {
+      "epoch": 0.9048221795761708,
+      "grad_norm": 1.6215999126434326,
+      "learning_rate": 2.2869956443275444e-06,
+      "loss": 1.3059,
+      "step": 2770
+    },
+    {
+      "epoch": 0.905148830182516,
+      "grad_norm": 0.9523685574531555,
+      "learning_rate": 2.2714243702359695e-06,
+      "loss": 1.0932,
+      "step": 2771
+    },
+    {
+      "epoch": 0.9054754807888612,
+      "grad_norm": 1.1872296333312988,
+      "learning_rate": 2.255905054652052e-06,
+      "loss": 1.2894,
+      "step": 2772
+    },
+    {
+      "epoch": 0.9058021313952064,
+      "grad_norm": 1.3475775718688965,
+      "learning_rate": 2.2404377144705e-06,
+      "loss": 1.5721,
+      "step": 2773
+    },
+    {
+      "epoch": 0.9061287820015516,
+      "grad_norm": 1.3719803094863892,
+      "learning_rate": 2.2250223665294446e-06,
+      "loss": 1.6983,
+      "step": 2774
+    },
+    {
+      "epoch": 0.9064554326078967,
+      "grad_norm": 2.0526633262634277,
+      "learning_rate": 2.2096590276103947e-06,
+      "loss": 2.0046,
+      "step": 2775
+    },
+    {
+      "epoch": 0.9067820832142419,
+      "grad_norm": 0.2175752967596054,
+      "learning_rate": 2.194347714438294e-06,
+      "loss": 0.7995,
+      "step": 2776
+    },
+    {
+      "epoch": 0.9071087338205872,
+      "grad_norm": 0.25334927439689636,
+      "learning_rate": 2.179088443681393e-06,
+      "loss": 0.8306,
+      "step": 2777
+    },
+    {
+      "epoch": 0.9074353844269324,
+      "grad_norm": 0.26047754287719727,
+      "learning_rate": 2.1638812319513145e-06,
+      "loss": 0.8277,
+      "step": 2778
+    },
+    {
+      "epoch": 0.9077620350332776,
+      "grad_norm": 0.27399611473083496,
+      "learning_rate": 2.1487260958030107e-06,
+      "loss": 0.8303,
+      "step": 2779
+    },
+    {
+      "epoch": 0.9080886856396228,
+      "grad_norm": 0.28062084317207336,
+      "learning_rate": 2.1336230517347342e-06,
+      "loss": 0.8613,
+      "step": 2780
+    },
+    {
+      "epoch": 0.9084153362459679,
+      "grad_norm": 0.29603925347328186,
+      "learning_rate": 2.1185721161880333e-06,
+      "loss": 0.8287,
+      "step": 2781
+    },
+    {
+      "epoch": 0.9087419868523131,
+      "grad_norm": 0.3076125383377075,
+      "learning_rate": 2.103573305547735e-06,
+      "loss": 0.9537,
+      "step": 2782
+    },
+    {
+      "epoch": 0.9090686374586583,
+      "grad_norm": 0.3091273307800293,
+      "learning_rate": 2.088626636141894e-06,
+      "loss": 0.8705,
+      "step": 2783
+    },
+    {
+      "epoch": 0.9093952880650035,
+      "grad_norm": 0.3315759301185608,
+      "learning_rate": 2.073732124241856e-06,
+      "loss": 0.8747,
+      "step": 2784
+    },
+    {
+      "epoch": 0.9097219386713487,
+      "grad_norm": 0.34617912769317627,
+      "learning_rate": 2.058889786062124e-06,
+      "loss": 0.9194,
+      "step": 2785
+    },
+    {
+      "epoch": 0.9100485892776938,
+      "grad_norm": 0.389011949300766,
+      "learning_rate": 2.0440996377604606e-06,
+      "loss": 0.8509,
+      "step": 2786
+    },
+    {
+      "epoch": 0.910375239884039,
+      "grad_norm": 0.35955238342285156,
+      "learning_rate": 2.0293616954377704e-06,
+      "loss": 0.8293,
+      "step": 2787
+    },
+    {
+      "epoch": 0.9107018904903842,
+      "grad_norm": 0.4118489623069763,
+      "learning_rate": 2.0146759751381474e-06,
+      "loss": 0.9932,
+      "step": 2788
+    },
+    {
+      "epoch": 0.9110285410967294,
+      "grad_norm": 0.4313391149044037,
+      "learning_rate": 2.0000424928488247e-06,
+      "loss": 0.9718,
+      "step": 2789
+    },
+    {
+      "epoch": 0.9113551917030746,
+      "grad_norm": 0.4242534637451172,
+      "learning_rate": 1.9854612645001767e-06,
+      "loss": 0.9077,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9116818423094197,
+      "grad_norm": 0.49969717860221863,
+      "learning_rate": 1.970932305965695e-06,
+      "loss": 0.9845,
+      "step": 2791
+    },
+    {
+      "epoch": 0.912008492915765,
+      "grad_norm": 0.49355196952819824,
+      "learning_rate": 1.9564556330619542e-06,
+      "loss": 0.8608,
+      "step": 2792
+    },
+    {
+      "epoch": 0.9123351435221102,
+      "grad_norm": 0.6009331941604614,
+      "learning_rate": 1.9420312615486215e-06,
+      "loss": 1.0791,
+      "step": 2793
+    },
+    {
+      "epoch": 0.9126617941284554,
+      "grad_norm": 0.655984103679657,
+      "learning_rate": 1.927659207128424e-06,
+      "loss": 1.0979,
+      "step": 2794
+    },
+    {
+      "epoch": 0.9129884447348006,
+      "grad_norm": 0.6680231094360352,
+      "learning_rate": 1.9133394854471465e-06,
+      "loss": 1.0447,
+      "step": 2795
+    },
+    {
+      "epoch": 0.9133150953411457,
+      "grad_norm": 0.8917760252952576,
+      "learning_rate": 1.8990721120935694e-06,
+      "loss": 1.2298,
+      "step": 2796
+    },
+    {
+      "epoch": 0.9136417459474909,
+      "grad_norm": 0.9837570190429688,
+      "learning_rate": 1.88485710259953e-06,
+      "loss": 1.358,
+      "step": 2797
+    },
+    {
+      "epoch": 0.9139683965538361,
+      "grad_norm": 1.2982633113861084,
+      "learning_rate": 1.8706944724398157e-06,
+      "loss": 1.2058,
+      "step": 2798
+    },
+    {
+      "epoch": 0.9142950471601813,
+      "grad_norm": 1.3844225406646729,
+      "learning_rate": 1.8565842370322495e-06,
+      "loss": 1.5844,
+      "step": 2799
+    },
+    {
+      "epoch": 0.9146216977665265,
+      "grad_norm": 1.6718311309814453,
+      "learning_rate": 1.8425264117375496e-06,
+      "loss": 2.01,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9149483483728716,
+      "grad_norm": 0.20186986029148102,
+      "learning_rate": 1.8285210118594297e-06,
+      "loss": 0.7628,
+      "step": 2801
+    },
+    {
+      "epoch": 0.9152749989792168,
+      "grad_norm": 0.24913327395915985,
+      "learning_rate": 1.8145680526445109e-06,
+      "loss": 0.8872,
+      "step": 2802
+    },
+    {
+      "epoch": 0.915601649585562,
+      "grad_norm": 0.25536009669303894,
+      "learning_rate": 1.8006675492823255e-06,
+      "loss": 0.7614,
+      "step": 2803
+    },
+    {
+      "epoch": 0.9159283001919072,
+      "grad_norm": 0.26635807752609253,
+      "learning_rate": 1.7868195169053137e-06,
+      "loss": 0.833,
+      "step": 2804
+    },
+    {
+      "epoch": 0.9162549507982524,
+      "grad_norm": 0.2926965653896332,
+      "learning_rate": 1.7730239705887774e-06,
+      "loss": 0.7751,
+      "step": 2805
+    },
+    {
+      "epoch": 0.9165816014045977,
+      "grad_norm": 0.28736618161201477,
+      "learning_rate": 1.7592809253508925e-06,
+      "loss": 0.8213,
+      "step": 2806
+    },
+    {
+      "epoch": 0.9169082520109428,
+      "grad_norm": 0.30105364322662354,
+      "learning_rate": 1.7455903961526798e-06,
+      "loss": 0.8296,
+      "step": 2807
+    },
+    {
+      "epoch": 0.917234902617288,
+      "grad_norm": 0.3077887296676636,
+      "learning_rate": 1.7319523978979956e-06,
+      "loss": 0.7668,
+      "step": 2808
+    },
+    {
+      "epoch": 0.9175615532236332,
+      "grad_norm": 0.3252915143966675,
+      "learning_rate": 1.7183669454334805e-06,
+      "loss": 0.8537,
+      "step": 2809
+    },
+    {
+      "epoch": 0.9178882038299784,
+      "grad_norm": 0.3521929979324341,
+      "learning_rate": 1.7048340535486151e-06,
+      "loss": 0.8731,
+      "step": 2810
+    },
+    {
+      "epoch": 0.9182148544363236,
+      "grad_norm": 0.349094033241272,
+      "learning_rate": 1.6913537369756205e-06,
+      "loss": 0.7871,
+      "step": 2811
+    },
+    {
+      "epoch": 0.9185415050426687,
+      "grad_norm": 0.3693489730358124,
+      "learning_rate": 1.67792601038953e-06,
+      "loss": 0.9557,
+      "step": 2812
+    },
+    {
+      "epoch": 0.9188681556490139,
+      "grad_norm": 0.37535011768341064,
+      "learning_rate": 1.664550888408073e-06,
+      "loss": 0.8216,
+      "step": 2813
+    },
+    {
+      "epoch": 0.9191948062553591,
+      "grad_norm": 0.3807601034641266,
+      "learning_rate": 1.6512283855917464e-06,
+      "loss": 0.9469,
+      "step": 2814
+    },
+    {
+      "epoch": 0.9195214568617043,
+      "grad_norm": 0.46269911527633667,
+      "learning_rate": 1.6379585164437605e-06,
+      "loss": 1.0594,
+      "step": 2815
+    },
+    {
+      "epoch": 0.9198481074680495,
+      "grad_norm": 0.44555824995040894,
+      "learning_rate": 1.6247412954100206e-06,
+      "loss": 1.012,
+      "step": 2816
+    },
+    {
+      "epoch": 0.9201747580743946,
+      "grad_norm": 0.5157310962677002,
+      "learning_rate": 1.611576736879117e-06,
+      "loss": 1.0669,
+      "step": 2817
+    },
+    {
+      "epoch": 0.9205014086807398,
+      "grad_norm": 0.5341678261756897,
+      "learning_rate": 1.5984648551823244e-06,
+      "loss": 1.1106,
+      "step": 2818
+    },
+    {
+      "epoch": 0.920828059287085,
+      "grad_norm": 0.5688596367835999,
+      "learning_rate": 1.5854056645935421e-06,
+      "loss": 1.0309,
+      "step": 2819
+    },
+    {
+      "epoch": 0.9211547098934303,
+      "grad_norm": 0.603225588798523,
+      "learning_rate": 1.572399179329348e-06,
+      "loss": 1.0216,
+      "step": 2820
+    },
+    {
+      "epoch": 0.9214813604997755,
+      "grad_norm": 0.7252687215805054,
+      "learning_rate": 1.5594454135489101e-06,
+      "loss": 1.0678,
+      "step": 2821
+    },
+    {
+      "epoch": 0.9218080111061207,
+      "grad_norm": 0.9123412370681763,
+      "learning_rate": 1.546544381354026e-06,
+      "loss": 1.3185,
+      "step": 2822
+    },
+    {
+      "epoch": 0.9221346617124658,
+      "grad_norm": 1.1781419515609741,
+      "learning_rate": 1.5336960967890722e-06,
+      "loss": 1.352,
+      "step": 2823
+    },
+    {
+      "epoch": 0.922461312318811,
+      "grad_norm": 1.3412940502166748,
+      "learning_rate": 1.520900573841011e-06,
+      "loss": 1.6097,
+      "step": 2824
+    },
+    {
+      "epoch": 0.9227879629251562,
+      "grad_norm": 1.9426753520965576,
+      "learning_rate": 1.5081578264393714e-06,
+      "loss": 1.9803,
+      "step": 2825
+    },
+    {
+      "epoch": 0.9231146135315014,
+      "grad_norm": 0.21749097108840942,
+      "learning_rate": 1.4954678684562128e-06,
+      "loss": 0.7693,
+      "step": 2826
+    },
+    {
+      "epoch": 0.9234412641378466,
+      "grad_norm": 0.23959995806217194,
+      "learning_rate": 1.4828307137061404e-06,
+      "loss": 0.7706,
+      "step": 2827
+    },
+    {
+      "epoch": 0.9237679147441917,
+      "grad_norm": 0.2661067247390747,
+      "learning_rate": 1.4702463759462827e-06,
+      "loss": 0.8191,
+      "step": 2828
+    },
+    {
+      "epoch": 0.9240945653505369,
+      "grad_norm": 0.27391666173934937,
+      "learning_rate": 1.4577148688762477e-06,
+      "loss": 0.7996,
+      "step": 2829
+    },
+    {
+      "epoch": 0.9244212159568821,
+      "grad_norm": 0.2843951880931854,
+      "learning_rate": 1.4452362061381509e-06,
+      "loss": 0.8425,
+      "step": 2830
+    },
+    {
+      "epoch": 0.9247478665632273,
+      "grad_norm": 0.30345508456230164,
+      "learning_rate": 1.4328104013165811e-06,
+      "loss": 0.9434,
+      "step": 2831
+    },
+    {
+      "epoch": 0.9250745171695725,
+      "grad_norm": 0.32872867584228516,
+      "learning_rate": 1.4204374679385513e-06,
+      "loss": 0.916,
+      "step": 2832
+    },
+    {
+      "epoch": 0.9254011677759176,
+      "grad_norm": 0.3325795829296112,
+      "learning_rate": 1.4081174194735758e-06,
+      "loss": 0.9412,
+      "step": 2833
+    },
+    {
+      "epoch": 0.9257278183822629,
+      "grad_norm": 0.3444099724292755,
+      "learning_rate": 1.3958502693335374e-06,
+      "loss": 0.9226,
+      "step": 2834
+    },
+    {
+      "epoch": 0.9260544689886081,
+      "grad_norm": 0.3495911955833435,
+      "learning_rate": 1.3836360308727703e-06,
+      "loss": 0.9122,
+      "step": 2835
+    },
+    {
+      "epoch": 0.9263811195949533,
+      "grad_norm": 0.3913719952106476,
+      "learning_rate": 1.3714747173879938e-06,
+      "loss": 1.0252,
+      "step": 2836
+    },
+    {
+      "epoch": 0.9267077702012985,
+      "grad_norm": 0.3872081935405731,
+      "learning_rate": 1.3593663421183177e-06,
+      "loss": 0.848,
+      "step": 2837
+    },
+    {
+      "epoch": 0.9270344208076436,
+      "grad_norm": 0.3887852132320404,
+      "learning_rate": 1.3473109182452203e-06,
+      "loss": 0.8442,
+      "step": 2838
+    },
+    {
+      "epoch": 0.9273610714139888,
+      "grad_norm": 0.4236868619918823,
+      "learning_rate": 1.3353084588925257e-06,
+      "loss": 1.0758,
+      "step": 2839
+    },
+    {
+      "epoch": 0.927687722020334,
+      "grad_norm": 0.4438207745552063,
+      "learning_rate": 1.3233589771264155e-06,
+      "loss": 0.9914,
+      "step": 2840
+    },
+    {
+      "epoch": 0.9280143726266792,
+      "grad_norm": 0.4978663921356201,
+      "learning_rate": 1.3114624859553892e-06,
+      "loss": 1.0171,
+      "step": 2841
+    },
+    {
+      "epoch": 0.9283410232330244,
+      "grad_norm": 0.5301693677902222,
+      "learning_rate": 1.2996189983302542e-06,
+      "loss": 1.0497,
+      "step": 2842
+    },
+    {
+      "epoch": 0.9286676738393695,
+      "grad_norm": 0.5215023159980774,
+      "learning_rate": 1.287828527144136e-06,
+      "loss": 1.0231,
+      "step": 2843
+    },
+    {
+      "epoch": 0.9289943244457147,
+      "grad_norm": 0.6315735578536987,
+      "learning_rate": 1.2760910852324338e-06,
+      "loss": 1.0299,
+      "step": 2844
+    },
+    {
+      "epoch": 0.9293209750520599,
+      "grad_norm": 0.6547741293907166,
+      "learning_rate": 1.2644066853727988e-06,
+      "loss": 1.3133,
+      "step": 2845
+    },
+    {
+      "epoch": 0.9296476256584051,
+      "grad_norm": 0.8934280276298523,
+      "learning_rate": 1.2527753402851782e-06,
+      "loss": 1.3854,
+      "step": 2846
+    },
+    {
+      "epoch": 0.9299742762647503,
+      "grad_norm": 1.3449379205703735,
+      "learning_rate": 1.2411970626317271e-06,
+      "loss": 1.6522,
+      "step": 2847
+    },
+    {
+      "epoch": 0.9303009268710956,
+      "grad_norm": 1.1736503839492798,
+      "learning_rate": 1.229671865016857e-06,
+      "loss": 1.2842,
+      "step": 2848
+    },
+    {
+      "epoch": 0.9306275774774407,
+      "grad_norm": 1.3014510869979858,
+      "learning_rate": 1.2181997599871709e-06,
+      "loss": 1.6985,
+      "step": 2849
+    },
+    {
+      "epoch": 0.9309542280837859,
+      "grad_norm": 1.8423781394958496,
+      "learning_rate": 1.2067807600314951e-06,
+      "loss": 2.0165,
+      "step": 2850
+    },
+    {
+      "epoch": 0.9312808786901311,
+      "grad_norm": 0.21177968382835388,
+      "learning_rate": 1.1954148775808305e-06,
+      "loss": 0.7456,
+      "step": 2851
+    },
+    {
+      "epoch": 0.9316075292964763,
+      "grad_norm": 0.23976528644561768,
+      "learning_rate": 1.1841021250083683e-06,
+      "loss": 0.775,
+      "step": 2852
+    },
+    {
+      "epoch": 0.9319341799028215,
+      "grad_norm": 0.2674348056316376,
+      "learning_rate": 1.1728425146294298e-06,
+      "loss": 0.8148,
+      "step": 2853
+    },
+    {
+      "epoch": 0.9322608305091666,
+      "grad_norm": 0.276597797870636,
+      "learning_rate": 1.1616360587015262e-06,
+      "loss": 0.8089,
+      "step": 2854
+    },
+    {
+      "epoch": 0.9325874811155118,
+      "grad_norm": 0.28054794669151306,
+      "learning_rate": 1.150482769424266e-06,
+      "loss": 0.749,
+      "step": 2855
+    },
+    {
+      "epoch": 0.932914131721857,
+      "grad_norm": 0.31427672505378723,
+      "learning_rate": 1.139382658939414e-06,
+      "loss": 0.8214,
+      "step": 2856
+    },
+    {
+      "epoch": 0.9332407823282022,
+      "grad_norm": 0.3161599338054657,
+      "learning_rate": 1.1283357393308102e-06,
+      "loss": 0.7862,
+      "step": 2857
+    },
+    {
+      "epoch": 0.9335674329345474,
+      "grad_norm": 0.35273951292037964,
+      "learning_rate": 1.117342022624407e-06,
+      "loss": 1.0511,
+      "step": 2858
+    },
+    {
+      "epoch": 0.9338940835408925,
+      "grad_norm": 0.3410721719264984,
+      "learning_rate": 1.106401520788236e-06,
+      "loss": 0.837,
+      "step": 2859
+    },
+    {
+      "epoch": 0.9342207341472377,
+      "grad_norm": 0.372831255197525,
+      "learning_rate": 1.0955142457324042e-06,
+      "loss": 0.867,
+      "step": 2860
+    },
+    {
+      "epoch": 0.9345473847535829,
+      "grad_norm": 0.37729281187057495,
+      "learning_rate": 1.084680209309058e-06,
+      "loss": 0.8085,
+      "step": 2861
+    },
+    {
+      "epoch": 0.9348740353599282,
+      "grad_norm": 0.4099488854408264,
+      "learning_rate": 1.0738994233124134e-06,
+      "loss": 0.9297,
+      "step": 2862
+    },
+    {
+      "epoch": 0.9352006859662734,
+      "grad_norm": 0.4232223629951477,
+      "learning_rate": 1.063171899478682e-06,
+      "loss": 0.9846,
+      "step": 2863
+    },
+    {
+      "epoch": 0.9355273365726186,
+      "grad_norm": 0.44969096779823303,
+      "learning_rate": 1.0524976494861227e-06,
+      "loss": 0.9578,
+      "step": 2864
+    },
+    {
+      "epoch": 0.9358539871789637,
+      "grad_norm": 0.49604538083076477,
+      "learning_rate": 1.0418766849549954e-06,
+      "loss": 1.0545,
+      "step": 2865
+    },
+    {
+      "epoch": 0.9361806377853089,
+      "grad_norm": 0.5002210140228271,
+      "learning_rate": 1.0313090174475182e-06,
+      "loss": 1.0935,
+      "step": 2866
+    },
+    {
+      "epoch": 0.9365072883916541,
+      "grad_norm": 0.5190601944923401,
+      "learning_rate": 1.0207946584679496e-06,
+      "loss": 1.0616,
+      "step": 2867
+    },
+    {
+      "epoch": 0.9368339389979993,
+      "grad_norm": 0.5428922772407532,
+      "learning_rate": 1.0103336194624446e-06,
+      "loss": 1.0405,
+      "step": 2868
+    },
+    {
+      "epoch": 0.9371605896043445,
+      "grad_norm": 0.6353383660316467,
+      "learning_rate": 9.999259118191827e-07,
+      "loss": 1.186,
+      "step": 2869
+    },
+    {
+      "epoch": 0.9374872402106896,
+      "grad_norm": 0.7398752570152283,
+      "learning_rate": 9.895715468682343e-07,
+      "loss": 1.214,
+      "step": 2870
+    },
+    {
+      "epoch": 0.9378138908170348,
+      "grad_norm": 0.7890881299972534,
+      "learning_rate": 9.792705358816156e-07,
+      "loss": 1.133,
+      "step": 2871
+    },
+    {
+      "epoch": 0.93814054142338,
+      "grad_norm": 0.9227431416511536,
+      "learning_rate": 9.69022890073268e-07,
+      "loss": 1.1887,
+      "step": 2872
+    },
+    {
+      "epoch": 0.9384671920297252,
+      "grad_norm": 1.1549128293991089,
+      "learning_rate": 9.588286205990281e-07,
+      "loss": 1.2085,
+      "step": 2873
+    },
+    {
+      "epoch": 0.9387938426360704,
+      "grad_norm": 1.3670769929885864,
+      "learning_rate": 9.486877385566307e-07,
+      "loss": 1.3928,
+      "step": 2874
+    },
+    {
+      "epoch": 0.9391204932424155,
+      "grad_norm": 1.7799900770187378,
+      "learning_rate": 9.386002549856942e-07,
+      "loss": 1.8047,
+      "step": 2875
+    },
+    {
+      "epoch": 0.9394471438487608,
+      "grad_norm": 0.2077762335538864,
+      "learning_rate": 9.28566180867696e-07,
+      "loss": 0.7245,
+      "step": 2876
+    },
+    {
+      "epoch": 0.939773794455106,
+      "grad_norm": 0.23600128293037415,
+      "learning_rate": 9.185855271259869e-07,
+      "loss": 0.8196,
+      "step": 2877
+    },
+    {
+      "epoch": 0.9401004450614512,
+      "grad_norm": 0.26550108194351196,
+      "learning_rate": 9.086583046257479e-07,
+      "loss": 0.8307,
+      "step": 2878
+    },
+    {
+      "epoch": 0.9404270956677964,
+      "grad_norm": 0.27199456095695496,
+      "learning_rate": 8.9878452417399e-07,
+      "loss": 0.8552,
+      "step": 2879
+    },
+    {
+      "epoch": 0.9407537462741415,
+      "grad_norm": 0.2838455140590668,
+      "learning_rate": 8.889641965195761e-07,
+      "loss": 0.835,
+      "step": 2880
+    },
+    {
+      "epoch": 0.9410803968804867,
+      "grad_norm": 0.3070827126502991,
+      "learning_rate": 8.791973323531322e-07,
+      "loss": 0.9071,
+      "step": 2881
+    },
+    {
+      "epoch": 0.9414070474868319,
+      "grad_norm": 0.3105681538581848,
+      "learning_rate": 8.694839423071311e-07,
+      "loss": 0.8274,
+      "step": 2882
+    },
+    {
+      "epoch": 0.9417336980931771,
+      "grad_norm": 0.32564812898635864,
+      "learning_rate": 8.598240369557975e-07,
+      "loss": 0.8845,
+      "step": 2883
+    },
+    {
+      "epoch": 0.9420603486995223,
+      "grad_norm": 0.3271031677722931,
+      "learning_rate": 8.502176268151474e-07,
+      "loss": 0.8739,
+      "step": 2884
+    },
+    {
+      "epoch": 0.9423869993058674,
+      "grad_norm": 0.3692990243434906,
+      "learning_rate": 8.406647223429487e-07,
+      "loss": 0.9222,
+      "step": 2885
+    },
+    {
+      "epoch": 0.9427136499122126,
+      "grad_norm": 0.3484625816345215,
+      "learning_rate": 8.311653339387381e-07,
+      "loss": 0.8741,
+      "step": 2886
+    },
+    {
+      "epoch": 0.9430403005185578,
+      "grad_norm": 0.3814089894294739,
+      "learning_rate": 8.21719471943777e-07,
+      "loss": 0.8628,
+      "step": 2887
+    },
+    {
+      "epoch": 0.943366951124903,
+      "grad_norm": 0.39709970355033875,
+      "learning_rate": 8.123271466410786e-07,
+      "loss": 0.9057,
+      "step": 2888
+    },
+    {
+      "epoch": 0.9436936017312482,
+      "grad_norm": 0.4413350522518158,
+      "learning_rate": 8.029883682553418e-07,
+      "loss": 0.8906,
+      "step": 2889
+    },
+    {
+      "epoch": 0.9440202523375933,
+      "grad_norm": 0.4726428687572479,
+      "learning_rate": 7.937031469530065e-07,
+      "loss": 1.084,
+      "step": 2890
+    },
+    {
+      "epoch": 0.9443469029439386,
+      "grad_norm": 0.4564618170261383,
+      "learning_rate": 7.844714928421871e-07,
+      "loss": 0.9873,
+      "step": 2891
+    },
+    {
+      "epoch": 0.9446735535502838,
+      "grad_norm": 0.5263504385948181,
+      "learning_rate": 7.752934159726888e-07,
+      "loss": 1.1024,
+      "step": 2892
+    },
+    {
+      "epoch": 0.945000204156629,
+      "grad_norm": 0.5723422765731812,
+      "learning_rate": 7.661689263359972e-07,
+      "loss": 0.9272,
+      "step": 2893
+    },
+    {
+      "epoch": 0.9453268547629742,
+      "grad_norm": 0.7069876790046692,
+      "learning_rate": 7.570980338652501e-07,
+      "loss": 1.2409,
+      "step": 2894
+    },
+    {
+      "epoch": 0.9456535053693194,
+      "grad_norm": 0.7442349195480347,
+      "learning_rate": 7.480807484352647e-07,
+      "loss": 1.2205,
+      "step": 2895
+    },
+    {
+      "epoch": 0.9459801559756645,
+      "grad_norm": 0.8722764849662781,
+      "learning_rate": 7.391170798624558e-07,
+      "loss": 1.0274,
+      "step": 2896
+    },
+    {
+      "epoch": 0.9463068065820097,
+      "grad_norm": 1.0321097373962402,
+      "learning_rate": 7.30207037904912e-07,
+      "loss": 1.3521,
+      "step": 2897
+    },
+    {
+      "epoch": 0.9466334571883549,
+      "grad_norm": 1.245355248451233,
+      "learning_rate": 7.213506322623187e-07,
+      "loss": 1.2544,
+      "step": 2898
+    },
+    {
+      "epoch": 0.9469601077947001,
+      "grad_norm": 1.5786315202713013,
+      "learning_rate": 7.125478725759805e-07,
+      "loss": 2.2946,
+      "step": 2899
+    },
+    {
+      "epoch": 0.9472867584010453,
+      "grad_norm": 1.5382407903671265,
+      "learning_rate": 7.037987684287984e-07,
+      "loss": 1.5537,
+      "step": 2900
+    },
+    {
+      "epoch": 0.9476134090073904,
+      "grad_norm": 0.21489055454730988,
+      "learning_rate": 6.951033293452646e-07,
+      "loss": 0.79,
+      "step": 2901
+    },
+    {
+      "epoch": 0.9479400596137356,
+      "grad_norm": 0.25648918747901917,
+      "learning_rate": 6.864615647914407e-07,
+      "loss": 0.7649,
+      "step": 2902
+    },
+    {
+      "epoch": 0.9482667102200808,
+      "grad_norm": 0.2571738362312317,
+      "learning_rate": 6.77873484174979e-07,
+      "loss": 0.8131,
+      "step": 2903
+    },
+    {
+      "epoch": 0.948593360826426,
+      "grad_norm": 0.28513866662979126,
+      "learning_rate": 6.693390968450674e-07,
+      "loss": 0.859,
+      "step": 2904
+    },
+    {
+      "epoch": 0.9489200114327713,
+      "grad_norm": 0.2826521694660187,
+      "learning_rate": 6.608584120924522e-07,
+      "loss": 0.8662,
+      "step": 2905
+    },
+    {
+      "epoch": 0.9492466620391165,
+      "grad_norm": 0.3028872609138489,
+      "learning_rate": 6.524314391494202e-07,
+      "loss": 0.8613,
+      "step": 2906
+    },
+    {
+      "epoch": 0.9495733126454616,
+      "grad_norm": 0.3100944459438324,
+      "learning_rate": 6.440581871897777e-07,
+      "loss": 0.8643,
+      "step": 2907
+    },
+    {
+      "epoch": 0.9498999632518068,
+      "grad_norm": 0.34333786368370056,
+      "learning_rate": 6.357386653288555e-07,
+      "loss": 0.8483,
+      "step": 2908
+    },
+    {
+      "epoch": 0.950226613858152,
+      "grad_norm": 0.3206077814102173,
+      "learning_rate": 6.274728826234922e-07,
+      "loss": 0.8385,
+      "step": 2909
+    },
+    {
+      "epoch": 0.9505532644644972,
+      "grad_norm": 0.3540595769882202,
+      "learning_rate": 6.192608480720175e-07,
+      "loss": 0.8793,
+      "step": 2910
+    },
+    {
+      "epoch": 0.9508799150708424,
+      "grad_norm": 0.368453711271286,
+      "learning_rate": 6.111025706142692e-07,
+      "loss": 0.9732,
+      "step": 2911
+    },
+    {
+      "epoch": 0.9512065656771875,
+      "grad_norm": 0.37942591309547424,
+      "learning_rate": 6.029980591315321e-07,
+      "loss": 0.9491,
+      "step": 2912
+    },
+    {
+      "epoch": 0.9515332162835327,
+      "grad_norm": 0.40083321928977966,
+      "learning_rate": 5.949473224465929e-07,
+      "loss": 0.9578,
+      "step": 2913
+    },
+    {
+      "epoch": 0.9518598668898779,
+      "grad_norm": 0.4563828706741333,
+      "learning_rate": 5.869503693236744e-07,
+      "loss": 0.868,
+      "step": 2914
+    },
+    {
+      "epoch": 0.9521865174962231,
+      "grad_norm": 0.4763037860393524,
+      "learning_rate": 5.790072084684573e-07,
+      "loss": 1.052,
+      "step": 2915
+    },
+    {
+      "epoch": 0.9525131681025683,
+      "grad_norm": 0.47979336977005005,
+      "learning_rate": 5.711178485280688e-07,
+      "loss": 1.0112,
+      "step": 2916
+    },
+    {
+      "epoch": 0.9528398187089134,
+      "grad_norm": 0.4837022125720978,
+      "learning_rate": 5.632822980910557e-07,
+      "loss": 1.0511,
+      "step": 2917
+    },
+    {
+      "epoch": 0.9531664693152586,
+      "grad_norm": 0.5643126964569092,
+      "learning_rate": 5.555005656873891e-07,
+      "loss": 1.0878,
+      "step": 2918
+    },
+    {
+      "epoch": 0.9534931199216039,
+      "grad_norm": 0.6676090955734253,
+      "learning_rate": 5.477726597884647e-07,
+      "loss": 1.1398,
+      "step": 2919
+    },
+    {
+      "epoch": 0.9538197705279491,
+      "grad_norm": 0.7085458040237427,
+      "learning_rate": 5.400985888070642e-07,
+      "loss": 1.0449,
+      "step": 2920
+    },
+    {
+      "epoch": 0.9541464211342943,
+      "grad_norm": 0.9497272968292236,
+      "learning_rate": 5.324783610973772e-07,
+      "loss": 1.4324,
+      "step": 2921
+    },
+    {
+      "epoch": 0.9544730717406394,
+      "grad_norm": 1.0609630346298218,
+      "learning_rate": 5.249119849549677e-07,
+      "loss": 1.3912,
+      "step": 2922
+    },
+    {
+      "epoch": 0.9547997223469846,
+      "grad_norm": 1.399525761604309,
+      "learning_rate": 5.173994686167805e-07,
+      "loss": 1.6172,
+      "step": 2923
+    },
+    {
+      "epoch": 0.9551263729533298,
+      "grad_norm": 1.3260670900344849,
+      "learning_rate": 5.099408202611289e-07,
+      "loss": 1.6486,
+      "step": 2924
+    },
+    {
+      "epoch": 0.955453023559675,
+      "grad_norm": 1.807623267173767,
+      "learning_rate": 5.02536048007679e-07,
+      "loss": 1.5547,
+      "step": 2925
+    },
+    {
+      "epoch": 0.9557796741660202,
+      "grad_norm": 0.21524813771247864,
+      "learning_rate": 4.951851599174495e-07,
+      "loss": 0.7755,
+      "step": 2926
+    },
+    {
+      "epoch": 0.9561063247723653,
+      "grad_norm": 0.25288552045822144,
+      "learning_rate": 4.878881639928001e-07,
+      "loss": 0.7779,
+      "step": 2927
+    },
+    {
+      "epoch": 0.9564329753787105,
+      "grad_norm": 0.27735069394111633,
+      "learning_rate": 4.806450681774155e-07,
+      "loss": 0.829,
+      "step": 2928
+    },
+    {
+      "epoch": 0.9567596259850557,
+      "grad_norm": 0.2676355540752411,
+      "learning_rate": 4.734558803563216e-07,
+      "loss": 0.7717,
+      "step": 2929
+    },
+    {
+      "epoch": 0.9570862765914009,
+      "grad_norm": 0.28219136595726013,
+      "learning_rate": 4.663206083558358e-07,
+      "loss": 0.7855,
+      "step": 2930
+    },
+    {
+      "epoch": 0.9574129271977461,
+      "grad_norm": 0.30700165033340454,
+      "learning_rate": 4.5923925994359464e-07,
+      "loss": 0.8179,
+      "step": 2931
+    },
+    {
+      "epoch": 0.9577395778040912,
+      "grad_norm": 0.31931400299072266,
+      "learning_rate": 4.5221184282852603e-07,
+      "loss": 0.8727,
+      "step": 2932
+    },
+    {
+      "epoch": 0.9580662284104365,
+      "grad_norm": 0.3536597788333893,
+      "learning_rate": 4.4523836466085487e-07,
+      "loss": 0.9074,
+      "step": 2933
+    },
+    {
+      "epoch": 0.9583928790167817,
+      "grad_norm": 0.3243158757686615,
+      "learning_rate": 4.383188330320753e-07,
+      "loss": 0.7781,
+      "step": 2934
+    },
+    {
+      "epoch": 0.9587195296231269,
+      "grad_norm": 0.37914979457855225,
+      "learning_rate": 4.3145325547497284e-07,
+      "loss": 0.8726,
+      "step": 2935
+    },
+    {
+      "epoch": 0.9590461802294721,
+      "grad_norm": 0.39078694581985474,
+      "learning_rate": 4.2464163946356884e-07,
+      "loss": 0.9549,
+      "step": 2936
+    },
+    {
+      "epoch": 0.9593728308358173,
+      "grad_norm": 0.4027955234050751,
+      "learning_rate": 4.1788399241317056e-07,
+      "loss": 0.8879,
+      "step": 2937
+    },
+    {
+      "epoch": 0.9596994814421624,
+      "grad_norm": 0.4388086497783661,
+      "learning_rate": 4.111803216803101e-07,
+      "loss": 0.9279,
+      "step": 2938
+    },
+    {
+      "epoch": 0.9600261320485076,
+      "grad_norm": 0.4330412745475769,
+      "learning_rate": 4.045306345627775e-07,
+      "loss": 0.8984,
+      "step": 2939
+    },
+    {
+      "epoch": 0.9603527826548528,
+      "grad_norm": 0.4897141456604004,
+      "learning_rate": 3.979349382995823e-07,
+      "loss": 0.9328,
+      "step": 2940
+    },
+    {
+      "epoch": 0.960679433261198,
+      "grad_norm": 0.5252649784088135,
+      "learning_rate": 3.9139324007095856e-07,
+      "loss": 1.015,
+      "step": 2941
+    },
+    {
+      "epoch": 0.9610060838675432,
+      "grad_norm": 0.5516462922096252,
+      "learning_rate": 3.849055469983709e-07,
+      "loss": 1.0131,
+      "step": 2942
+    },
+    {
+      "epoch": 0.9613327344738883,
+      "grad_norm": 0.627424955368042,
+      "learning_rate": 3.7847186614446975e-07,
+      "loss": 1.0651,
+      "step": 2943
+    },
+    {
+      "epoch": 0.9616593850802335,
+      "grad_norm": 0.6904886364936829,
+      "learning_rate": 3.7209220451313033e-07,
+      "loss": 1.1691,
+      "step": 2944
+    },
+    {
+      "epoch": 0.9619860356865787,
+      "grad_norm": 0.757783055305481,
+      "learning_rate": 3.6576656904940274e-07,
+      "loss": 1.0155,
+      "step": 2945
+    },
+    {
+      "epoch": 0.9623126862929239,
+      "grad_norm": 0.9803491830825806,
+      "learning_rate": 3.594949666395342e-07,
+      "loss": 1.3613,
+      "step": 2946
+    },
+    {
+      "epoch": 0.9626393368992692,
+      "grad_norm": 1.1308027505874634,
+      "learning_rate": 3.5327740411094655e-07,
+      "loss": 1.3419,
+      "step": 2947
+    },
+    {
+      "epoch": 0.9629659875056144,
+      "grad_norm": 1.2531534433364868,
+      "learning_rate": 3.4711388823223114e-07,
+      "loss": 1.4717,
+      "step": 2948
+    },
+    {
+      "epoch": 0.9632926381119595,
+      "grad_norm": 1.2751611471176147,
+      "learning_rate": 3.4100442571313727e-07,
+      "loss": 1.3097,
+      "step": 2949
+    },
+    {
+      "epoch": 0.9636192887183047,
+      "grad_norm": 1.4386547803878784,
+      "learning_rate": 3.3494902320458934e-07,
+      "loss": 1.4562,
+      "step": 2950
+    },
+    {
+      "epoch": 0.9639459393246499,
+      "grad_norm": 0.20834031701087952,
+      "learning_rate": 3.2894768729863636e-07,
+      "loss": 0.7,
+      "step": 2951
+    },
+    {
+      "epoch": 0.9642725899309951,
+      "grad_norm": 0.2570595145225525,
+      "learning_rate": 3.230004245284912e-07,
+      "loss": 0.8549,
+      "step": 2952
+    },
+    {
+      "epoch": 0.9645992405373403,
+      "grad_norm": 0.25834420323371887,
+      "learning_rate": 3.1710724136848613e-07,
+      "loss": 0.7985,
+      "step": 2953
+    },
+    {
+      "epoch": 0.9649258911436854,
+      "grad_norm": 0.2837374806404114,
+      "learning_rate": 3.112681442340837e-07,
+      "loss": 0.7945,
+      "step": 2954
+    },
+    {
+      "epoch": 0.9652525417500306,
+      "grad_norm": 0.2851658761501312,
+      "learning_rate": 3.0548313948187134e-07,
+      "loss": 0.8037,
+      "step": 2955
+    },
+    {
+      "epoch": 0.9655791923563758,
+      "grad_norm": 0.3020305037498474,
+      "learning_rate": 2.997522334095504e-07,
+      "loss": 0.8115,
+      "step": 2956
+    },
+    {
+      "epoch": 0.965905842962721,
+      "grad_norm": 0.30819737911224365,
+      "learning_rate": 2.940754322559247e-07,
+      "loss": 0.8555,
+      "step": 2957
+    },
+    {
+      "epoch": 0.9662324935690662,
+      "grad_norm": 0.31193971633911133,
+      "learning_rate": 2.8845274220090093e-07,
+      "loss": 0.8179,
+      "step": 2958
+    },
+    {
+      "epoch": 0.9665591441754113,
+      "grad_norm": 0.31887295842170715,
+      "learning_rate": 2.828841693654771e-07,
+      "loss": 0.8089,
+      "step": 2959
+    },
+    {
+      "epoch": 0.9668857947817565,
+      "grad_norm": 0.34238582849502563,
+      "learning_rate": 2.773697198117431e-07,
+      "loss": 0.9042,
+      "step": 2960
+    },
+    {
+      "epoch": 0.9672124453881018,
+      "grad_norm": 0.36260437965393066,
+      "learning_rate": 2.719093995428634e-07,
+      "loss": 0.9795,
+      "step": 2961
+    },
+    {
+      "epoch": 0.967539095994447,
+      "grad_norm": 0.36125776171684265,
+      "learning_rate": 2.6650321450307216e-07,
+      "loss": 0.8816,
+      "step": 2962
+    },
+    {
+      "epoch": 0.9678657466007922,
+      "grad_norm": 0.37313613295555115,
+      "learning_rate": 2.611511705776892e-07,
+      "loss": 0.8222,
+      "step": 2963
+    },
+    {
+      "epoch": 0.9681923972071373,
+      "grad_norm": 0.40587756037712097,
+      "learning_rate": 2.558532735930652e-07,
+      "loss": 0.9125,
+      "step": 2964
+    },
+    {
+      "epoch": 0.9685190478134825,
+      "grad_norm": 0.45703062415122986,
+      "learning_rate": 2.506095293166366e-07,
+      "loss": 0.9605,
+      "step": 2965
+    },
+    {
+      "epoch": 0.9688456984198277,
+      "grad_norm": 0.46383294463157654,
+      "learning_rate": 2.4541994345686494e-07,
+      "loss": 0.9941,
+      "step": 2966
+    },
+    {
+      "epoch": 0.9691723490261729,
+      "grad_norm": 0.5314494371414185,
+      "learning_rate": 2.402845216632643e-07,
+      "loss": 1.0825,
+      "step": 2967
+    },
+    {
+      "epoch": 0.9694989996325181,
+      "grad_norm": 0.5420945882797241,
+      "learning_rate": 2.3520326952638504e-07,
+      "loss": 1.0171,
+      "step": 2968
+    },
+    {
+      "epoch": 0.9698256502388632,
+      "grad_norm": 0.5639206767082214,
+      "learning_rate": 2.301761925777912e-07,
+      "loss": 1.0999,
+      "step": 2969
+    },
+    {
+      "epoch": 0.9701523008452084,
+      "grad_norm": 0.6383841037750244,
+      "learning_rate": 2.2520329629009962e-07,
+      "loss": 1.1892,
+      "step": 2970
+    },
+    {
+      "epoch": 0.9704789514515536,
+      "grad_norm": 0.7161938548088074,
+      "learning_rate": 2.2028458607691872e-07,
+      "loss": 1.1297,
+      "step": 2971
+    },
+    {
+      "epoch": 0.9708056020578988,
+      "grad_norm": 0.9672861695289612,
+      "learning_rate": 2.1542006729287634e-07,
+      "loss": 1.3065,
+      "step": 2972
+    },
+    {
+      "epoch": 0.971132252664244,
+      "grad_norm": 1.2224022150039673,
+      "learning_rate": 2.106097452336142e-07,
+      "loss": 1.2359,
+      "step": 2973
+    },
+    {
+      "epoch": 0.9714589032705891,
+      "grad_norm": 1.3247113227844238,
+      "learning_rate": 2.058536251357601e-07,
+      "loss": 1.5186,
+      "step": 2974
+    },
+    {
+      "epoch": 0.9717855538769344,
+      "grad_norm": 1.7058521509170532,
+      "learning_rate": 2.0115171217695573e-07,
+      "loss": 1.5777,
+      "step": 2975
+    },
+    {
+      "epoch": 0.9721122044832796,
+      "grad_norm": 0.22474831342697144,
+      "learning_rate": 1.9650401147580655e-07,
+      "loss": 0.7176,
+      "step": 2976
+    },
+    {
+      "epoch": 0.9724388550896248,
+      "grad_norm": 0.2482801228761673,
+      "learning_rate": 1.9191052809192643e-07,
+      "loss": 0.745,
+      "step": 2977
+    },
+    {
+      "epoch": 0.97276550569597,
+      "grad_norm": 0.2572484016418457,
+      "learning_rate": 1.8737126702588202e-07,
+      "loss": 0.8129,
+      "step": 2978
+    },
+    {
+      "epoch": 0.9730921563023152,
+      "grad_norm": 0.269260436296463,
+      "learning_rate": 1.8288623321924268e-07,
+      "loss": 0.801,
+      "step": 2979
+    },
+    {
+      "epoch": 0.9734188069086603,
+      "grad_norm": 0.2986462712287903,
+      "learning_rate": 1.784554315545084e-07,
+      "loss": 0.8864,
+      "step": 2980
+    },
+    {
+      "epoch": 0.9737454575150055,
+      "grad_norm": 0.302898108959198,
+      "learning_rate": 1.740788668551707e-07,
+      "loss": 0.8545,
+      "step": 2981
+    },
+    {
+      "epoch": 0.9740721081213507,
+      "grad_norm": 0.30456167459487915,
+      "learning_rate": 1.6975654388566298e-07,
+      "loss": 0.8325,
+      "step": 2982
+    },
+    {
+      "epoch": 0.9743987587276959,
+      "grad_norm": 0.3240984380245209,
+      "learning_rate": 1.654884673513768e-07,
+      "loss": 0.866,
+      "step": 2983
+    },
+    {
+      "epoch": 0.9747254093340411,
+      "grad_norm": 0.31887489557266235,
+      "learning_rate": 1.6127464189863995e-07,
+      "loss": 0.8524,
+      "step": 2984
+    },
+    {
+      "epoch": 0.9750520599403862,
+      "grad_norm": 0.3447200655937195,
+      "learning_rate": 1.5711507211472742e-07,
+      "loss": 0.8951,
+      "step": 2985
+    },
+    {
+      "epoch": 0.9753787105467314,
+      "grad_norm": 0.32973796129226685,
+      "learning_rate": 1.5300976252785593e-07,
+      "loss": 0.8277,
+      "step": 2986
+    },
+    {
+      "epoch": 0.9757053611530766,
+      "grad_norm": 0.38608047366142273,
+      "learning_rate": 1.4895871760716162e-07,
+      "loss": 0.9188,
+      "step": 2987
+    },
+    {
+      "epoch": 0.9760320117594218,
+      "grad_norm": 0.3714994788169861,
+      "learning_rate": 1.4496194176271126e-07,
+      "loss": 0.9459,
+      "step": 2988
+    },
+    {
+      "epoch": 0.9763586623657671,
+      "grad_norm": 0.4215068221092224,
+      "learning_rate": 1.4101943934549667e-07,
+      "loss": 0.871,
+      "step": 2989
+    },
+    {
+      "epoch": 0.9766853129721123,
+      "grad_norm": 0.4134935438632965,
+      "learning_rate": 1.371312146474235e-07,
+      "loss": 0.8685,
+      "step": 2990
+    },
+    {
+      "epoch": 0.9770119635784574,
+      "grad_norm": 0.47548481822013855,
+      "learning_rate": 1.3329727190130593e-07,
+      "loss": 0.9058,
+      "step": 2991
+    },
+    {
+      "epoch": 0.9773386141848026,
+      "grad_norm": 0.49817779660224915,
+      "learning_rate": 1.2951761528087746e-07,
+      "loss": 0.9534,
+      "step": 2992
+    },
+    {
+      "epoch": 0.9776652647911478,
+      "grad_norm": 0.5390109419822693,
+      "learning_rate": 1.2579224890075235e-07,
+      "loss": 1.0624,
+      "step": 2993
+    },
+    {
+      "epoch": 0.977991915397493,
+      "grad_norm": 0.5876456499099731,
+      "learning_rate": 1.2212117681646984e-07,
+      "loss": 1.0434,
+      "step": 2994
+    },
+    {
+      "epoch": 0.9783185660038382,
+      "grad_norm": 0.8520432114601135,
+      "learning_rate": 1.1850440302444421e-07,
+      "loss": 1.1705,
+      "step": 2995
+    },
+    {
+      "epoch": 0.9786452166101833,
+      "grad_norm": 0.8995637893676758,
+      "learning_rate": 1.1494193146198151e-07,
+      "loss": 1.3287,
+      "step": 2996
+    },
+    {
+      "epoch": 0.9789718672165285,
+      "grad_norm": 0.9267019033432007,
+      "learning_rate": 1.1143376600727951e-07,
+      "loss": 1.322,
+      "step": 2997
+    },
+    {
+      "epoch": 0.9792985178228737,
+      "grad_norm": 1.0994635820388794,
+      "learning_rate": 1.0797991047941102e-07,
+      "loss": 1.5212,
+      "step": 2998
+    },
+    {
+      "epoch": 0.9796251684292189,
+      "grad_norm": 1.2156322002410889,
+      "learning_rate": 1.0458036863834064e-07,
+      "loss": 1.5812,
+      "step": 2999
+    },
+    {
+      "epoch": 0.979951819035564,
+      "grad_norm": 1.8230586051940918,
+      "learning_rate": 1.0123514418487468e-07,
+      "loss": 1.9067,
+      "step": 3000
+    },
+    {
+      "epoch": 0.9802784696419092,
+      "grad_norm": 0.22708569467067719,
+      "learning_rate": 9.794424076072228e-08,
+      "loss": 0.813,
+      "step": 3001
+    },
+    {
+      "epoch": 0.9806051202482544,
+      "grad_norm": 0.24345462024211884,
+      "learning_rate": 9.470766194843439e-08,
+      "loss": 0.8086,
+      "step": 3002
+    },
+    {
+      "epoch": 0.9809317708545996,
+      "grad_norm": 0.2802831530570984,
+      "learning_rate": 9.152541127143699e-08,
+      "loss": 0.8417,
+      "step": 3003
+    },
+    {
+      "epoch": 0.9812584214609449,
+      "grad_norm": 0.27346834540367126,
+      "learning_rate": 8.839749219399784e-08,
+      "loss": 0.9053,
+      "step": 3004
+    },
+    {
+      "epoch": 0.9815850720672901,
+      "grad_norm": 0.2842161953449249,
+      "learning_rate": 8.532390812125424e-08,
+      "loss": 0.8295,
+      "step": 3005
+    },
+    {
+      "epoch": 0.9819117226736352,
+      "grad_norm": 0.29227709770202637,
+      "learning_rate": 8.230466239918521e-08,
+      "loss": 0.8382,
+      "step": 3006
+    },
+    {
+      "epoch": 0.9822383732799804,
+      "grad_norm": 0.3029593527317047,
+      "learning_rate": 7.933975831461716e-08,
+      "loss": 0.8816,
+      "step": 3007
+    },
+    {
+      "epoch": 0.9825650238863256,
+      "grad_norm": 0.31089669466018677,
+      "learning_rate": 7.642919909521263e-08,
+      "loss": 0.8376,
+      "step": 3008
+    },
+    {
+      "epoch": 0.9828916744926708,
+      "grad_norm": 0.31294307112693787,
+      "learning_rate": 7.357298790948152e-08,
+      "loss": 0.8148,
+      "step": 3009
+    },
+    {
+      "epoch": 0.983218325099016,
+      "grad_norm": 0.33494314551353455,
+      "learning_rate": 7.077112786676444e-08,
+      "loss": 0.7742,
+      "step": 3010
+    },
+    {
+      "epoch": 0.9835449757053611,
+      "grad_norm": 0.3963795006275177,
+      "learning_rate": 6.80236220172381e-08,
+      "loss": 0.88,
+      "step": 3011
+    },
+    {
+      "epoch": 0.9838716263117063,
+      "grad_norm": 0.40085482597351074,
+      "learning_rate": 6.533047335190446e-08,
+      "loss": 0.8962,
+      "step": 3012
+    },
+    {
+      "epoch": 0.9841982769180515,
+      "grad_norm": 0.4059949815273285,
+      "learning_rate": 6.269168480259602e-08,
+      "loss": 0.9012,
+      "step": 3013
+    },
+    {
+      "epoch": 0.9845249275243967,
+      "grad_norm": 0.4352096915245056,
+      "learning_rate": 6.010725924195382e-08,
+      "loss": 1.023,
+      "step": 3014
+    },
+    {
+      "epoch": 0.9848515781307419,
+      "grad_norm": 0.49773815274238586,
+      "learning_rate": 5.757719948344953e-08,
+      "loss": 0.9623,
+      "step": 3015
+    },
+    {
+      "epoch": 0.985178228737087,
+      "grad_norm": 0.4756074845790863,
+      "learning_rate": 5.5101508281379946e-08,
+      "loss": 0.9922,
+      "step": 3016
+    },
+    {
+      "epoch": 0.9855048793434322,
+      "grad_norm": 0.5341835021972656,
+      "learning_rate": 5.2680188330828106e-08,
+      "loss": 0.9406,
+      "step": 3017
+    },
+    {
+      "epoch": 0.9858315299497775,
+      "grad_norm": 0.5826855301856995,
+      "learning_rate": 5.031324226771328e-08,
+      "loss": 1.1138,
+      "step": 3018
+    },
+    {
+      "epoch": 0.9861581805561227,
+      "grad_norm": 0.6295057535171509,
+      "learning_rate": 4.800067266874653e-08,
+      "loss": 1.079,
+      "step": 3019
+    },
+    {
+      "epoch": 0.9864848311624679,
+      "grad_norm": 0.6822542548179626,
+      "learning_rate": 4.5742482051452925e-08,
+      "loss": 1.3472,
+      "step": 3020
+    },
+    {
+      "epoch": 0.9868114817688131,
+      "grad_norm": 0.7802051305770874,
+      "learning_rate": 4.353867287414937e-08,
+      "loss": 0.8963,
+      "step": 3021
+    },
+    {
+      "epoch": 0.9871381323751582,
+      "grad_norm": 0.9883480072021484,
+      "learning_rate": 4.1389247535961186e-08,
+      "loss": 1.3332,
+      "step": 3022
+    },
+    {
+      "epoch": 0.9874647829815034,
+      "grad_norm": 1.0861750841140747,
+      "learning_rate": 3.929420837679998e-08,
+      "loss": 1.2043,
+      "step": 3023
+    },
+    {
+      "epoch": 0.9877914335878486,
+      "grad_norm": 1.2433277368545532,
+      "learning_rate": 3.7253557677385805e-08,
+      "loss": 1.5074,
+      "step": 3024
+    },
+    {
+      "epoch": 0.9881180841941938,
+      "grad_norm": 1.6619473695755005,
+      "learning_rate": 3.5267297659213886e-08,
+      "loss": 1.6274,
+      "step": 3025
+    },
+    {
+      "epoch": 0.988444734800539,
+      "grad_norm": 0.20633414387702942,
+      "learning_rate": 3.333543048458232e-08,
+      "loss": 0.6526,
+      "step": 3026
+    },
+    {
+      "epoch": 0.9887713854068841,
+      "grad_norm": 0.2510916590690613,
+      "learning_rate": 3.145795825656439e-08,
+      "loss": 0.7986,
+      "step": 3027
+    },
+    {
+      "epoch": 0.9890980360132293,
+      "grad_norm": 0.2619028091430664,
+      "learning_rate": 2.9634883019025173e-08,
+      "loss": 0.8371,
+      "step": 3028
+    },
+    {
+      "epoch": 0.9894246866195745,
+      "grad_norm": 0.2645626664161682,
+      "learning_rate": 2.7866206756610445e-08,
+      "loss": 0.7909,
+      "step": 3029
+    },
+    {
+      "epoch": 0.9897513372259197,
+      "grad_norm": 0.28736555576324463,
+      "learning_rate": 2.6151931394741146e-08,
+      "loss": 0.843,
+      "step": 3030
+    },
+    {
+      "epoch": 0.9900779878322649,
+      "grad_norm": 0.28967446088790894,
+      "learning_rate": 2.449205879962446e-08,
+      "loss": 0.8462,
+      "step": 3031
+    },
+    {
+      "epoch": 0.9904046384386102,
+      "grad_norm": 0.3020203709602356,
+      "learning_rate": 2.288659077823163e-08,
+      "loss": 0.7806,
+      "step": 3032
+    },
+    {
+      "epoch": 0.9907312890449553,
+      "grad_norm": 0.33225804567337036,
+      "learning_rate": 2.13355290783257e-08,
+      "loss": 0.8409,
+      "step": 3033
+    },
+    {
+      "epoch": 0.9910579396513005,
+      "grad_norm": 0.3261914849281311,
+      "learning_rate": 1.983887538842266e-08,
+      "loss": 0.8474,
+      "step": 3034
+    },
+    {
+      "epoch": 0.9913845902576457,
+      "grad_norm": 0.32698002457618713,
+      "learning_rate": 1.8396631337813664e-08,
+      "loss": 0.8514,
+      "step": 3035
+    },
+    {
+      "epoch": 0.9917112408639909,
+      "grad_norm": 0.32505884766578674,
+      "learning_rate": 1.7008798496570556e-08,
+      "loss": 0.8421,
+      "step": 3036
+    },
+    {
+      "epoch": 0.992037891470336,
+      "grad_norm": 0.3715566098690033,
+      "learning_rate": 1.5675378375518135e-08,
+      "loss": 0.8502,
+      "step": 3037
+    },
+    {
+      "epoch": 0.9923645420766812,
+      "grad_norm": 0.38677921891212463,
+      "learning_rate": 1.4396372426250803e-08,
+      "loss": 0.8488,
+      "step": 3038
+    },
+    {
+      "epoch": 0.9926911926830264,
+      "grad_norm": 0.4004034996032715,
+      "learning_rate": 1.3171782041127012e-08,
+      "loss": 0.9237,
+      "step": 3039
+    },
+    {
+      "epoch": 0.9930178432893716,
+      "grad_norm": 0.4869138300418854,
+      "learning_rate": 1.200160855326371e-08,
+      "loss": 1.0046,
+      "step": 3040
+    },
+    {
+      "epoch": 0.9933444938957168,
+      "grad_norm": 0.4786950349807739,
+      "learning_rate": 1.0885853236541898e-08,
+      "loss": 0.9855,
+      "step": 3041
+    },
+    {
+      "epoch": 0.993671144502062,
+      "grad_norm": 0.5063715577125549,
+      "learning_rate": 9.824517305601077e-09,
+      "loss": 1.0504,
+      "step": 3042
+    },
+    {
+      "epoch": 0.9939977951084071,
+      "grad_norm": 0.573488712310791,
+      "learning_rate": 8.817601915839247e-09,
+      "loss": 1.0409,
+      "step": 3043
+    },
+    {
+      "epoch": 0.9943244457147523,
+      "grad_norm": 0.5906215310096741,
+      "learning_rate": 7.865108163407354e-09,
+      "loss": 1.032,
+      "step": 3044
+    },
+    {
+      "epoch": 0.9946510963210975,
+      "grad_norm": 0.5928130745887756,
+      "learning_rate": 6.967037085209293e-09,
+      "loss": 1.0514,
+      "step": 3045
+    },
+    {
+      "epoch": 0.9949777469274428,
+      "grad_norm": 0.7858253121376038,
+      "learning_rate": 6.123389658913015e-09,
+      "loss": 1.001,
+      "step": 3046
+    },
+    {
+      "epoch": 0.995304397533788,
+      "grad_norm": 0.9497416615486145,
+      "learning_rate": 5.33416680293386e-09,
+      "loss": 1.3928,
+      "step": 3047
+    },
+    {
+      "epoch": 0.9956310481401331,
+      "grad_norm": 1.2982540130615234,
+      "learning_rate": 4.599369376440122e-09,
+      "loss": 1.5056,
+      "step": 3048
+    },
+    {
+      "epoch": 0.9959576987464783,
+      "grad_norm": 1.4778263568878174,
+      "learning_rate": 3.9189981793474885e-09,
+      "loss": 1.7437,
+      "step": 3049
+    },
+    {
+      "epoch": 0.9962843493528235,
+      "grad_norm": 1.6468915939331055,
+      "learning_rate": 3.2930539523245983e-09,
+      "loss": 1.8305,
+      "step": 3050
+    },
+    {
+      "epoch": 0.9966109999591687,
+      "grad_norm": 0.2135321944952011,
+      "learning_rate": 2.7215373767930373e-09,
+      "loss": 0.7451,
+      "step": 3051
+    },
+    {
+      "epoch": 0.9969376505655139,
+      "grad_norm": 0.2706100046634674,
+      "learning_rate": 2.204449074916237e-09,
+      "loss": 0.7944,
+      "step": 3052
+    },
+    {
+      "epoch": 0.997264301171859,
+      "grad_norm": 0.28801220655441284,
+      "learning_rate": 1.741789609610578e-09,
+      "loss": 0.8088,
+      "step": 3053
+    },
+    {
+      "epoch": 0.9975909517782042,
+      "grad_norm": 0.3120700716972351,
+      "learning_rate": 1.3335594845453881e-09,
+      "loss": 0.8261,
+      "step": 3054
+    },
+    {
+      "epoch": 0.9979176023845494,
+      "grad_norm": 0.3604918122291565,
+      "learning_rate": 9.797591441151888e-10,
+      "loss": 0.8993,
+      "step": 3055
+    },
+    {
+      "epoch": 0.9982442529908946,
+      "grad_norm": 0.435921847820282,
+      "learning_rate": 6.803889734896541e-10,
+      "loss": 0.9562,
+      "step": 3056
+    },
+    {
+      "epoch": 0.9985709035972398,
+      "grad_norm": 0.528087317943573,
+      "learning_rate": 4.3544929856365046e-10,
+      "loss": 1.0983,
+      "step": 3057
+    },
+    {
+      "epoch": 0.998897554203585,
+      "grad_norm": 0.7063041925430298,
+      "learning_rate": 2.44940385984993e-10,
+      "loss": 1.25,
+      "step": 3058
+    },
+    {
+      "epoch": 0.9992242048099301,
+      "grad_norm": 0.9369118213653564,
+      "learning_rate": 1.0886244314889382e-10,
+      "loss": 1.2773,
+      "step": 3059
+    },
+    {
+      "epoch": 0.9995508554162754,
+      "grad_norm": 1.196264624595642,
+      "learning_rate": 2.721561819241103e-11,
+      "loss": 1.6142,
+      "step": 3060
+    },
+    {
+      "epoch": 0.9998775060226206,
+      "grad_norm": 1.3984349966049194,
+      "learning_rate": 0.0,
+      "loss": 1.7651,
+      "step": 3061
     }
   ],
   "logging_steps": 1,
@@ -16139,12 +21480,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": false
+        "should_training_stop": true
       },
       "attributes": {}
     }
   },
-  "total_flos": 1.5133624248230216e+18,
+  "total_flos": 2.0158272316658156e+18,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null