diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10204 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500.0,
+  "global_step": 72699,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0006877673695649183,
+      "grad_norm": 0.06897997111082077,
+      "learning_rate": 0.0001,
+      "loss": 1.8351,
+      "step": 50
+    },
+    {
+      "epoch": 0.0013755347391298366,
+      "grad_norm": 0.0760108083486557,
+      "learning_rate": 0.0001,
+      "loss": 1.8141,
+      "step": 100
+    },
+    {
+      "epoch": 0.002063302108694755,
+      "grad_norm": 0.07610712945461273,
+      "learning_rate": 0.0001,
+      "loss": 1.8047,
+      "step": 150
+    },
+    {
+      "epoch": 0.0027510694782596733,
+      "grad_norm": 0.07757844030857086,
+      "learning_rate": 0.0001,
+      "loss": 1.7991,
+      "step": 200
+    },
+    {
+      "epoch": 0.003438836847824592,
+      "grad_norm": 0.07695277035236359,
+      "learning_rate": 0.0001,
+      "loss": 1.7939,
+      "step": 250
+    },
+    {
+      "epoch": 0.00412660421738951,
+      "grad_norm": 0.07816466689109802,
+      "learning_rate": 0.0001,
+      "loss": 1.7895,
+      "step": 300
+    },
+    {
+      "epoch": 0.004814371586954428,
+      "grad_norm": 0.07848047465085983,
+      "learning_rate": 0.0001,
+      "loss": 1.7861,
+      "step": 350
+    },
+    {
+      "epoch": 0.0055021389565193465,
+      "grad_norm": 0.08218562602996826,
+      "learning_rate": 0.0001,
+      "loss": 1.7777,
+      "step": 400
+    },
+    {
+      "epoch": 0.006189906326084265,
+      "grad_norm": 0.07656747847795486,
+      "learning_rate": 0.0001,
+      "loss": 1.7805,
+      "step": 450
+    },
+    {
+      "epoch": 0.006877673695649184,
+      "grad_norm": 0.07792618125677109,
+      "learning_rate": 0.0001,
+      "loss": 1.7791,
+      "step": 500
+    },
+    {
+      "epoch": 0.007565441065214102,
+      "grad_norm": 0.10222098976373672,
+      "learning_rate": 0.0001,
+      "loss": 1.7782,
+      "step": 550
+    },
+    {
+      "epoch": 0.00825320843477902,
+      "grad_norm": 0.07399246096611023,
+      "learning_rate": 0.0001,
+      "loss": 1.7734,
+      "step": 600
+    },
+    {
+      "epoch": 0.008940975804343939,
+      "grad_norm": 0.08359324187040329,
+      "learning_rate": 0.0001,
+      "loss": 1.7699,
+      "step": 650
+    },
+    {
+      "epoch": 0.009628743173908856,
+      "grad_norm": 0.08059624582529068,
+      "learning_rate": 0.0001,
+      "loss": 1.7735,
+      "step": 700
+    },
+    {
+      "epoch": 0.010316510543473776,
+      "grad_norm": 0.07659206539392471,
+      "learning_rate": 0.0001,
+      "loss": 1.767,
+      "step": 750
+    },
+    {
+      "epoch": 0.011004277913038693,
+      "grad_norm": 0.08097860962152481,
+      "learning_rate": 0.0001,
+      "loss": 1.7703,
+      "step": 800
+    },
+    {
+      "epoch": 0.011692045282603612,
+      "grad_norm": 0.08088383823633194,
+      "learning_rate": 0.0001,
+      "loss": 1.7648,
+      "step": 850
+    },
+    {
+      "epoch": 0.01237981265216853,
+      "grad_norm": 0.07697251439094543,
+      "learning_rate": 0.0001,
+      "loss": 1.7671,
+      "step": 900
+    },
+    {
+      "epoch": 0.013067580021733449,
+      "grad_norm": 0.08154206722974777,
+      "learning_rate": 0.0001,
+      "loss": 1.7629,
+      "step": 950
+    },
+    {
+      "epoch": 0.013755347391298368,
+      "grad_norm": 0.08085618168115616,
+      "learning_rate": 0.0001,
+      "loss": 1.7632,
+      "step": 1000
+    },
+    {
+      "epoch": 0.014443114760863286,
+      "grad_norm": 0.08026960492134094,
+      "learning_rate": 0.0001,
+      "loss": 1.7594,
+      "step": 1050
+    },
+    {
+      "epoch": 0.015130882130428205,
+      "grad_norm": 0.07709988951683044,
+      "learning_rate": 0.0001,
+      "loss": 1.7566,
+      "step": 1100
+    },
+    {
+      "epoch": 0.015818649499993124,
+      "grad_norm": 0.08290690928697586,
+      "learning_rate": 0.0001,
+      "loss": 1.7607,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01650641686955804,
+      "grad_norm": 0.07915584743022919,
+      "learning_rate": 0.0001,
+      "loss": 1.7527,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01719418423912296,
+      "grad_norm": 0.08169861882925034,
+      "learning_rate": 0.0001,
+      "loss": 1.7535,
+      "step": 1250
+    },
+    {
+      "epoch": 0.017881951608687878,
+      "grad_norm": 0.07745598256587982,
+      "learning_rate": 0.0001,
+      "loss": 1.7542,
+      "step": 1300
+    },
+    {
+      "epoch": 0.018569718978252797,
+      "grad_norm": 0.07793346047401428,
+      "learning_rate": 0.0001,
+      "loss": 1.7536,
+      "step": 1350
+    },
+    {
+      "epoch": 0.019257486347817713,
+      "grad_norm": 0.0771438255906105,
+      "learning_rate": 0.0001,
+      "loss": 1.7486,
+      "step": 1400
+    },
+    {
+      "epoch": 0.019945253717382632,
+      "grad_norm": 0.08003713190555573,
+      "learning_rate": 0.0001,
+      "loss": 1.7501,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02063302108694755,
+      "grad_norm": 0.08053753525018692,
+      "learning_rate": 0.0001,
+      "loss": 1.7493,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02132078845651247,
+      "grad_norm": 0.0768120214343071,
+      "learning_rate": 0.0001,
+      "loss": 1.7482,
+      "step": 1550
+    },
+    {
+      "epoch": 0.022008555826077386,
+      "grad_norm": 0.07853832095861435,
+      "learning_rate": 0.0001,
+      "loss": 1.7493,
+      "step": 1600
+    },
+    {
+      "epoch": 0.022696323195642305,
+      "grad_norm": 0.083896704018116,
+      "learning_rate": 0.0001,
+      "loss": 1.7467,
+      "step": 1650
+    },
+    {
+      "epoch": 0.023384090565207225,
+      "grad_norm": 0.07456424832344055,
+      "learning_rate": 0.0001,
+      "loss": 1.7456,
+      "step": 1700
+    },
+    {
+      "epoch": 0.024071857934772144,
+      "grad_norm": 0.0817745104432106,
+      "learning_rate": 0.0001,
+      "loss": 1.7438,
+      "step": 1750
+    },
+    {
+      "epoch": 0.02475962530433706,
+      "grad_norm": 0.0803341269493103,
+      "learning_rate": 0.0001,
+      "loss": 1.7434,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02544739267390198,
+      "grad_norm": 0.07569049298763275,
+      "learning_rate": 0.0001,
+      "loss": 1.7449,
+      "step": 1850
+    },
+    {
+      "epoch": 0.026135160043466898,
+      "grad_norm": 0.07928919047117233,
+      "learning_rate": 0.0001,
+      "loss": 1.7388,
+      "step": 1900
+    },
+    {
+      "epoch": 0.026822927413031817,
+      "grad_norm": 0.0802655890583992,
+      "learning_rate": 0.0001,
+      "loss": 1.7409,
+      "step": 1950
+    },
+    {
+      "epoch": 0.027510694782596736,
+      "grad_norm": 0.07988564670085907,
+      "learning_rate": 0.0001,
+      "loss": 1.7396,
+      "step": 2000
+    },
+    {
+      "epoch": 0.028198462152161652,
+      "grad_norm": 0.08523596078157425,
+      "learning_rate": 0.0001,
+      "loss": 1.7375,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02888622952172657,
+      "grad_norm": 0.08964947611093521,
+      "learning_rate": 0.0001,
+      "loss": 1.736,
+      "step": 2100
+    },
+    {
+      "epoch": 0.02957399689129149,
+      "grad_norm": 0.08145201951265335,
+      "learning_rate": 0.0001,
+      "loss": 1.7401,
+      "step": 2150
+    },
+    {
+      "epoch": 0.03026176426085641,
+      "grad_norm": 0.08227697014808655,
+      "learning_rate": 0.0001,
+      "loss": 1.7332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.030949531630421325,
+      "grad_norm": 0.07990765571594238,
+      "learning_rate": 0.0001,
+      "loss": 1.7359,
+      "step": 2250
+    },
+    {
+      "epoch": 0.03163729899998625,
+      "grad_norm": 0.0835530161857605,
+      "learning_rate": 0.0001,
+      "loss": 1.7353,
+      "step": 2300
+    },
+    {
+      "epoch": 0.03232506636955116,
+      "grad_norm": 0.09058014303445816,
+      "learning_rate": 0.0001,
+      "loss": 1.7337,
+      "step": 2350
+    },
+    {
+      "epoch": 0.03301283373911608,
+      "grad_norm": 0.07631311565637589,
+      "learning_rate": 0.0001,
+      "loss": 1.7319,
+      "step": 2400
+    },
+    {
+      "epoch": 0.033700601108681,
+      "grad_norm": 0.07975883036851883,
+      "learning_rate": 0.0001,
+      "loss": 1.7331,
+      "step": 2450
+    },
+    {
+      "epoch": 0.03438836847824592,
+      "grad_norm": 0.0856098085641861,
+      "learning_rate": 0.0001,
+      "loss": 1.729,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03507613584781084,
+      "grad_norm": 0.07782306522130966,
+      "learning_rate": 0.0001,
+      "loss": 1.7333,
+      "step": 2550
+    },
+    {
+      "epoch": 0.035763903217375756,
+      "grad_norm": 0.08808156102895737,
+      "learning_rate": 0.0001,
+      "loss": 1.7295,
+      "step": 2600
+    },
+    {
+      "epoch": 0.036451670586940675,
+      "grad_norm": 0.08983422815799713,
+      "learning_rate": 0.0001,
+      "loss": 1.7281,
+      "step": 2650
+    },
+    {
+      "epoch": 0.037139437956505594,
+      "grad_norm": 0.0850716382265091,
+      "learning_rate": 0.0001,
+      "loss": 1.7288,
+      "step": 2700
+    },
+    {
+      "epoch": 0.03782720532607051,
+      "grad_norm": 0.07847283780574799,
+      "learning_rate": 0.0001,
+      "loss": 1.7311,
+      "step": 2750
+    },
+    {
+      "epoch": 0.038514972695635426,
+      "grad_norm": 0.08320654928684235,
+      "learning_rate": 0.0001,
+      "loss": 1.7326,
+      "step": 2800
+    },
+    {
+      "epoch": 0.039202740065200345,
+      "grad_norm": 0.09107951819896698,
+      "learning_rate": 0.0001,
+      "loss": 1.7302,
+      "step": 2850
+    },
+    {
+      "epoch": 0.039890507434765264,
+      "grad_norm": 0.08038274943828583,
+      "learning_rate": 0.0001,
+      "loss": 1.7278,
+      "step": 2900
+    },
+    {
+      "epoch": 0.04057827480433018,
+      "grad_norm": 0.08225669711828232,
+      "learning_rate": 0.0001,
+      "loss": 1.7278,
+      "step": 2950
+    },
+    {
+      "epoch": 0.0412660421738951,
+      "grad_norm": 0.0852913185954094,
+      "learning_rate": 0.0001,
+      "loss": 1.7233,
+      "step": 3000
+    },
+    {
+      "epoch": 0.04195380954346002,
+      "grad_norm": 0.08353286981582642,
+      "learning_rate": 0.0001,
+      "loss": 1.7274,
+      "step": 3050
+    },
+    {
+      "epoch": 0.04264157691302494,
+      "grad_norm": 0.08236906677484512,
+      "learning_rate": 0.0001,
+      "loss": 1.7259,
+      "step": 3100
+    },
+    {
+      "epoch": 0.04332934428258986,
+      "grad_norm": 0.08094590902328491,
+      "learning_rate": 0.0001,
+      "loss": 1.7252,
+      "step": 3150
+    },
+    {
+      "epoch": 0.04401711165215477,
+      "grad_norm": 0.08384421467781067,
+      "learning_rate": 0.0001,
+      "loss": 1.7248,
+      "step": 3200
+    },
+    {
+      "epoch": 0.04470487902171969,
+      "grad_norm": 0.0810803771018982,
+      "learning_rate": 0.0001,
+      "loss": 1.7221,
+      "step": 3250
+    },
+    {
+      "epoch": 0.04539264639128461,
+      "grad_norm": 0.08863117545843124,
+      "learning_rate": 0.0001,
+      "loss": 1.7249,
+      "step": 3300
+    },
+    {
+      "epoch": 0.04608041376084953,
+      "grad_norm": 0.08272850513458252,
+      "learning_rate": 0.0001,
+      "loss": 1.7245,
+      "step": 3350
+    },
+    {
+      "epoch": 0.04676818113041445,
+      "grad_norm": 0.08872344344854355,
+      "learning_rate": 0.0001,
+      "loss": 1.7238,
+      "step": 3400
+    },
+    {
+      "epoch": 0.04745594849997937,
+      "grad_norm": 0.08702059090137482,
+      "learning_rate": 0.0001,
+      "loss": 1.723,
+      "step": 3450
+    },
+    {
+      "epoch": 0.04814371586954429,
+      "grad_norm": 0.07964511960744858,
+      "learning_rate": 0.0001,
+      "loss": 1.7208,
+      "step": 3500
+    },
+    {
+      "epoch": 0.04883148323910921,
+      "grad_norm": 0.08207014203071594,
+      "learning_rate": 0.0001,
+      "loss": 1.7189,
+      "step": 3550
+    },
+    {
+      "epoch": 0.04951925060867412,
+      "grad_norm": 0.08424731343984604,
+      "learning_rate": 0.0001,
+      "loss": 1.722,
+      "step": 3600
+    },
+    {
+      "epoch": 0.05020701797823904,
+      "grad_norm": 0.07758868485689163,
+      "learning_rate": 0.0001,
+      "loss": 1.7202,
+      "step": 3650
+    },
+    {
+      "epoch": 0.05089478534780396,
+      "grad_norm": 0.08630118519067764,
+      "learning_rate": 0.0001,
+      "loss": 1.7206,
+      "step": 3700
+    },
+    {
+      "epoch": 0.051582552717368876,
+      "grad_norm": 0.08104898035526276,
+      "learning_rate": 0.0001,
+      "loss": 1.7204,
+      "step": 3750
+    },
+    {
+      "epoch": 0.052270320086933796,
+      "grad_norm": 0.08586841821670532,
+      "learning_rate": 0.0001,
+      "loss": 1.7174,
+      "step": 3800
+    },
+    {
+      "epoch": 0.052958087456498715,
+      "grad_norm": 0.07795108109712601,
+      "learning_rate": 0.0001,
+      "loss": 1.7181,
+      "step": 3850
+    },
+    {
+      "epoch": 0.053645854826063634,
+      "grad_norm": 0.08409079909324646,
+      "learning_rate": 0.0001,
+      "loss": 1.7197,
+      "step": 3900
+    },
+    {
+      "epoch": 0.05433362219562855,
+      "grad_norm": 0.08350292593240738,
+      "learning_rate": 0.0001,
+      "loss": 1.7148,
+      "step": 3950
+    },
+    {
+      "epoch": 0.05502138956519347,
+      "grad_norm": 0.08318202942609787,
+      "learning_rate": 0.0001,
+      "loss": 1.7206,
+      "step": 4000
+    },
+    {
+      "epoch": 0.055709156934758385,
+      "grad_norm": 0.08408112823963165,
+      "learning_rate": 0.0001,
+      "loss": 1.7185,
+      "step": 4050
+    },
+    {
+      "epoch": 0.056396924304323304,
+      "grad_norm": 0.08753649145364761,
+      "learning_rate": 0.0001,
+      "loss": 1.7146,
+      "step": 4100
+    },
+    {
+      "epoch": 0.05708469167388822,
+      "grad_norm": 0.08772628009319305,
+      "learning_rate": 0.0001,
+      "loss": 1.719,
+      "step": 4150
+    },
+    {
+      "epoch": 0.05777245904345314,
+      "grad_norm": 0.08609924465417862,
+      "learning_rate": 0.0001,
+      "loss": 1.7189,
+      "step": 4200
+    },
+    {
+      "epoch": 0.05846022641301806,
+      "grad_norm": 0.08166562020778656,
+      "learning_rate": 0.0001,
+      "loss": 1.7156,
+      "step": 4250
+    },
+    {
+      "epoch": 0.05914799378258298,
+      "grad_norm": 0.10656413435935974,
+      "learning_rate": 0.0001,
+      "loss": 1.7198,
+      "step": 4300
+    },
+    {
+      "epoch": 0.0598357611521479,
+      "grad_norm": 0.08603313565254211,
+      "learning_rate": 0.0001,
+      "loss": 1.7152,
+      "step": 4350
+    },
+    {
+      "epoch": 0.06052352852171282,
+      "grad_norm": 0.08598031848669052,
+      "learning_rate": 0.0001,
+      "loss": 1.7163,
+      "step": 4400
+    },
+    {
+      "epoch": 0.06121129589127773,
+      "grad_norm": 0.08404461294412613,
+      "learning_rate": 0.0001,
+      "loss": 1.7125,
+      "step": 4450
+    },
+    {
+      "epoch": 0.06189906326084265,
+      "grad_norm": 0.08633497357368469,
+      "learning_rate": 0.0001,
+      "loss": 1.7118,
+      "step": 4500
+    },
+    {
+      "epoch": 0.06258683063040757,
+      "grad_norm": 0.087265245616436,
+      "learning_rate": 0.0001,
+      "loss": 1.7145,
+      "step": 4550
+    },
+    {
+      "epoch": 0.0632745979999725,
+      "grad_norm": 0.08514571934938431,
+      "learning_rate": 0.0001,
+      "loss": 1.7106,
+      "step": 4600
+    },
+    {
+      "epoch": 0.06396236536953741,
+      "grad_norm": 0.08637581020593643,
+      "learning_rate": 0.0001,
+      "loss": 1.7136,
+      "step": 4650
+    },
+    {
+      "epoch": 0.06465013273910232,
+      "grad_norm": 0.08497826009988785,
+      "learning_rate": 0.0001,
+      "loss": 1.7114,
+      "step": 4700
+    },
+    {
+      "epoch": 0.06533790010866725,
+      "grad_norm": 0.09578850120306015,
+      "learning_rate": 0.0001,
+      "loss": 1.7102,
+      "step": 4750
+    },
+    {
+      "epoch": 0.06602566747823216,
+      "grad_norm": 0.08667325973510742,
+      "learning_rate": 0.0001,
+      "loss": 1.7102,
+      "step": 4800
+    },
+    {
+      "epoch": 0.06671343484779708,
+      "grad_norm": 0.08680035918951035,
+      "learning_rate": 0.0001,
+      "loss": 1.7135,
+      "step": 4850
+    },
+    {
+      "epoch": 0.067401202217362,
+      "grad_norm": 0.0779147818684578,
+      "learning_rate": 0.0001,
+      "loss": 1.7144,
+      "step": 4900
+    },
+    {
+      "epoch": 0.06808896958692692,
+      "grad_norm": 0.08890321105718613,
+      "learning_rate": 0.0001,
+      "loss": 1.7105,
+      "step": 4950
+    },
+    {
+      "epoch": 0.06877673695649184,
+      "grad_norm": 0.08258546888828278,
+      "learning_rate": 0.0001,
+      "loss": 1.7126,
+      "step": 5000
+    },
+    {
+      "epoch": 0.06946450432605676,
+      "grad_norm": 0.0860370546579361,
+      "learning_rate": 0.0001,
+      "loss": 1.7084,
+      "step": 5050
+    },
+    {
+      "epoch": 0.07015227169562167,
+      "grad_norm": 0.0840732529759407,
+      "learning_rate": 0.0001,
+      "loss": 1.7153,
+      "step": 5100
+    },
+    {
+      "epoch": 0.07084003906518659,
+      "grad_norm": 0.08410490304231644,
+      "learning_rate": 0.0001,
+      "loss": 1.7105,
+      "step": 5150
+    },
+    {
+      "epoch": 0.07152780643475151,
+      "grad_norm": 0.07879506051540375,
+      "learning_rate": 0.0001,
+      "loss": 1.7067,
+      "step": 5200
+    },
+    {
+      "epoch": 0.07221557380431642,
+      "grad_norm": 0.0969550833106041,
+      "learning_rate": 0.0001,
+      "loss": 1.7033,
+      "step": 5250
+    },
+    {
+      "epoch": 0.07290334117388135,
+      "grad_norm": 0.08318573236465454,
+      "learning_rate": 0.0001,
+      "loss": 1.7101,
+      "step": 5300
+    },
+    {
+      "epoch": 0.07359110854344626,
+      "grad_norm": 0.08452852070331573,
+      "learning_rate": 0.0001,
+      "loss": 1.7081,
+      "step": 5350
+    },
+    {
+      "epoch": 0.07427887591301119,
+      "grad_norm": 0.08123467862606049,
+      "learning_rate": 0.0001,
+      "loss": 1.712,
+      "step": 5400
+    },
+    {
+      "epoch": 0.0749666432825761,
+      "grad_norm": 0.08749833703041077,
+      "learning_rate": 0.0001,
+      "loss": 1.7096,
+      "step": 5450
+    },
+    {
+      "epoch": 0.07565441065214101,
+      "grad_norm": 0.09270595759153366,
+      "learning_rate": 0.0001,
+      "loss": 1.7038,
+      "step": 5500
+    },
+    {
+      "epoch": 0.07634217802170594,
+      "grad_norm": 0.0823141559958458,
+      "learning_rate": 0.0001,
+      "loss": 1.7047,
+      "step": 5550
+    },
+    {
+      "epoch": 0.07702994539127085,
+      "grad_norm": 0.09130799025297165,
+      "learning_rate": 0.0001,
+      "loss": 1.707,
+      "step": 5600
+    },
+    {
+      "epoch": 0.07771771276083578,
+      "grad_norm": 0.08671557903289795,
+      "learning_rate": 0.0001,
+      "loss": 1.7083,
+      "step": 5650
+    },
+    {
+      "epoch": 0.07840548013040069,
+      "grad_norm": 0.08462578803300858,
+      "learning_rate": 0.0001,
+      "loss": 1.7018,
+      "step": 5700
+    },
+    {
+      "epoch": 0.07909324749996562,
+      "grad_norm": 0.08817154169082642,
+      "learning_rate": 0.0001,
+      "loss": 1.7024,
+      "step": 5750
+    },
+    {
+      "epoch": 0.07978101486953053,
+      "grad_norm": 0.09289363026618958,
+      "learning_rate": 0.0001,
+      "loss": 1.7077,
+      "step": 5800
+    },
+    {
+      "epoch": 0.08046878223909545,
+      "grad_norm": 0.08403845876455307,
+      "learning_rate": 0.0001,
+      "loss": 1.6997,
+      "step": 5850
+    },
+    {
+      "epoch": 0.08115654960866037,
+      "grad_norm": 0.08645929396152496,
+      "learning_rate": 0.0001,
+      "loss": 1.7051,
+      "step": 5900
+    },
+    {
+      "epoch": 0.08184431697822528,
+      "grad_norm": 0.08411643654108047,
+      "learning_rate": 0.0001,
+      "loss": 1.7059,
+      "step": 5950
+    },
+    {
+      "epoch": 0.0825320843477902,
+      "grad_norm": 0.09166127443313599,
+      "learning_rate": 0.0001,
+      "loss": 1.7036,
+      "step": 6000
+    },
+    {
+      "epoch": 0.08321985171735512,
+      "grad_norm": 0.08519306033849716,
+      "learning_rate": 0.0001,
+      "loss": 1.7056,
+      "step": 6050
+    },
+    {
+      "epoch": 0.08390761908692004,
+      "grad_norm": 0.08552943915128708,
+      "learning_rate": 0.0001,
+      "loss": 1.7034,
+      "step": 6100
+    },
+    {
+      "epoch": 0.08459538645648496,
+      "grad_norm": 0.08541347831487656,
+      "learning_rate": 0.0001,
+      "loss": 1.7046,
+      "step": 6150
+    },
+    {
+      "epoch": 0.08528315382604988,
+      "grad_norm": 0.08669677376747131,
+      "learning_rate": 0.0001,
+      "loss": 1.7042,
+      "step": 6200
+    },
+    {
+      "epoch": 0.0859709211956148,
+      "grad_norm": 0.08166626840829849,
+      "learning_rate": 0.0001,
+      "loss": 1.705,
+      "step": 6250
+    },
+    {
+      "epoch": 0.08665868856517972,
+      "grad_norm": 0.08560743182897568,
+      "learning_rate": 0.0001,
+      "loss": 1.7033,
+      "step": 6300
+    },
+    {
+      "epoch": 0.08734645593474463,
+      "grad_norm": 0.08644382655620575,
+      "learning_rate": 0.0001,
+      "loss": 1.7036,
+      "step": 6350
+    },
+    {
+      "epoch": 0.08803422330430954,
+      "grad_norm": 0.0843438133597374,
+      "learning_rate": 0.0001,
+      "loss": 1.6998,
+      "step": 6400
+    },
+    {
+      "epoch": 0.08872199067387447,
+      "grad_norm": 0.08665913343429565,
+      "learning_rate": 0.0001,
+      "loss": 1.6986,
+      "step": 6450
+    },
+    {
+      "epoch": 0.08940975804343938,
+      "grad_norm": 0.07717980444431305,
+      "learning_rate": 0.0001,
+      "loss": 1.6986,
+      "step": 6500
+    },
+    {
+      "epoch": 0.09009752541300431,
+      "grad_norm": 0.08198987692594528,
+      "learning_rate": 0.0001,
+      "loss": 1.7019,
+      "step": 6550
+    },
+    {
+      "epoch": 0.09078529278256922,
+      "grad_norm": 0.08437033742666245,
+      "learning_rate": 0.0001,
+      "loss": 1.7032,
+      "step": 6600
+    },
+    {
+      "epoch": 0.09147306015213415,
+      "grad_norm": 0.08697813004255295,
+      "learning_rate": 0.0001,
+      "loss": 1.701,
+      "step": 6650
+    },
+    {
+      "epoch": 0.09216082752169906,
+      "grad_norm": 0.08434218913316727,
+      "learning_rate": 0.0001,
+      "loss": 1.7,
+      "step": 6700
+    },
+    {
+      "epoch": 0.09284859489126399,
+      "grad_norm": 0.08920707553625107,
+      "learning_rate": 0.0001,
+      "loss": 1.6993,
+      "step": 6750
+    },
+    {
+      "epoch": 0.0935363622608289,
+      "grad_norm": 0.08233244717121124,
+      "learning_rate": 0.0001,
+      "loss": 1.6993,
+      "step": 6800
+    },
+    {
+      "epoch": 0.09422412963039381,
+      "grad_norm": 0.0831465944647789,
+      "learning_rate": 0.0001,
+      "loss": 1.7009,
+      "step": 6850
+    },
+    {
+      "epoch": 0.09491189699995874,
+      "grad_norm": 0.08565302193164825,
+      "learning_rate": 0.0001,
+      "loss": 1.699,
+      "step": 6900
+    },
+    {
+      "epoch": 0.09559966436952365,
+      "grad_norm": 0.08086910843849182,
+      "learning_rate": 0.0001,
+      "loss": 1.7014,
+      "step": 6950
+    },
+    {
+      "epoch": 0.09628743173908857,
+      "grad_norm": 0.08551020175218582,
+      "learning_rate": 0.0001,
+      "loss": 1.7005,
+      "step": 7000
+    },
+    {
+      "epoch": 0.09697519910865349,
+      "grad_norm": 0.08930126577615738,
+      "learning_rate": 0.0001,
+      "loss": 1.6996,
+      "step": 7050
+    },
+    {
+      "epoch": 0.09766296647821841,
+      "grad_norm": 0.0878065899014473,
+      "learning_rate": 0.0001,
+      "loss": 1.6976,
+      "step": 7100
+    },
+    {
+      "epoch": 0.09835073384778333,
+      "grad_norm": 0.09178236126899719,
+      "learning_rate": 0.0001,
+      "loss": 1.6957,
+      "step": 7150
+    },
+    {
+      "epoch": 0.09903850121734824,
+      "grad_norm": 0.08948659151792526,
+      "learning_rate": 0.0001,
+      "loss": 1.7008,
+      "step": 7200
+    },
+    {
+      "epoch": 0.09972626858691316,
+      "grad_norm": 0.0796937644481659,
+      "learning_rate": 0.0001,
+      "loss": 1.7023,
+      "step": 7250
+    },
+    {
+      "epoch": 0.10041403595647808,
+      "grad_norm": 0.08426263928413391,
+      "learning_rate": 0.0001,
+      "loss": 1.6992,
+      "step": 7300
+    },
+    {
+      "epoch": 0.101101803326043,
+      "grad_norm": 0.0868782177567482,
+      "learning_rate": 0.0001,
+      "loss": 1.6996,
+      "step": 7350
+    },
+    {
+      "epoch": 0.10178957069560791,
+      "grad_norm": 0.08032725006341934,
+      "learning_rate": 0.0001,
+      "loss": 1.7,
+      "step": 7400
+    },
+    {
+      "epoch": 0.10247733806517284,
+      "grad_norm": 0.08831656724214554,
+      "learning_rate": 0.0001,
+      "loss": 1.6943,
+      "step": 7450
+    },
+    {
+      "epoch": 0.10316510543473775,
+      "grad_norm": 0.0800151377916336,
+      "learning_rate": 0.0001,
+      "loss": 1.6986,
+      "step": 7500
+    },
+    {
+      "epoch": 0.10385287280430268,
+      "grad_norm": 0.09132002294063568,
+      "learning_rate": 0.0001,
+      "loss": 1.6953,
+      "step": 7550
+    },
+    {
+      "epoch": 0.10454064017386759,
+      "grad_norm": 0.0925612673163414,
+      "learning_rate": 0.0001,
+      "loss": 1.6975,
+      "step": 7600
+    },
+    {
+      "epoch": 0.1052284075434325,
+      "grad_norm": 0.08521568030118942,
+      "learning_rate": 0.0001,
+      "loss": 1.6918,
+      "step": 7650
+    },
+    {
+      "epoch": 0.10591617491299743,
+      "grad_norm": 0.08576962351799011,
+      "learning_rate": 0.0001,
+      "loss": 1.6951,
+      "step": 7700
+    },
+    {
+      "epoch": 0.10660394228256234,
+      "grad_norm": 0.08979818224906921,
+      "learning_rate": 0.0001,
+      "loss": 1.6994,
+      "step": 7750
+    },
+    {
+      "epoch": 0.10729170965212727,
+      "grad_norm": 0.08927462995052338,
+      "learning_rate": 0.0001,
+      "loss": 1.6946,
+      "step": 7800
+    },
+    {
+      "epoch": 0.10797947702169218,
+      "grad_norm": 0.0862877294421196,
+      "learning_rate": 0.0001,
+      "loss": 1.6994,
+      "step": 7850
+    },
+    {
+      "epoch": 0.1086672443912571,
+      "grad_norm": 0.0894964188337326,
+      "learning_rate": 0.0001,
+      "loss": 1.6953,
+      "step": 7900
+    },
+    {
+      "epoch": 0.10935501176082202,
+      "grad_norm": 0.09155282378196716,
+      "learning_rate": 0.0001,
+      "loss": 1.6929,
+      "step": 7950
+    },
+    {
+      "epoch": 0.11004277913038694,
+      "grad_norm": 0.08712241053581238,
+      "learning_rate": 0.0001,
+      "loss": 1.6958,
+      "step": 8000
+    },
+    {
+      "epoch": 0.11073054649995186,
+      "grad_norm": 0.08157862722873688,
+      "learning_rate": 0.0001,
+      "loss": 1.6942,
+      "step": 8050
+    },
+    {
+      "epoch": 0.11141831386951677,
+      "grad_norm": 0.08820342272520065,
+      "learning_rate": 0.0001,
+      "loss": 1.6928,
+      "step": 8100
+    },
+    {
+      "epoch": 0.1121060812390817,
+      "grad_norm": 0.08663163334131241,
+      "learning_rate": 0.0001,
+      "loss": 1.6936,
+      "step": 8150
+    },
+    {
+      "epoch": 0.11279384860864661,
+      "grad_norm": 0.08267082273960114,
+      "learning_rate": 0.0001,
+      "loss": 1.6939,
+      "step": 8200
+    },
+    {
+      "epoch": 0.11348161597821153,
+      "grad_norm": 0.0887032225728035,
+      "learning_rate": 0.0001,
+      "loss": 1.6965,
+      "step": 8250
+    },
+    {
+      "epoch": 0.11416938334777645,
+      "grad_norm": 0.0971740260720253,
+      "learning_rate": 0.0001,
+      "loss": 1.6919,
+      "step": 8300
+    },
+    {
+      "epoch": 0.11485715071734137,
+      "grad_norm": 0.08765862137079239,
+      "learning_rate": 0.0001,
+      "loss": 1.6878,
+      "step": 8350
+    },
+    {
+      "epoch": 0.11554491808690628,
+      "grad_norm": 0.08712899684906006,
+      "learning_rate": 0.0001,
+      "loss": 1.6927,
+      "step": 8400
+    },
+    {
+      "epoch": 0.1162326854564712,
+      "grad_norm": 0.08298210054636002,
+      "learning_rate": 0.0001,
+      "loss": 1.6909,
+      "step": 8450
+    },
+    {
+      "epoch": 0.11692045282603612,
+      "grad_norm": 0.08736353367567062,
+      "learning_rate": 0.0001,
+      "loss": 1.6906,
+      "step": 8500
+    },
+    {
+      "epoch": 0.11760822019560103,
+      "grad_norm": 0.0875655934214592,
+      "learning_rate": 0.0001,
+      "loss": 1.6945,
+      "step": 8550
+    },
+    {
+      "epoch": 0.11829598756516596,
+      "grad_norm": 0.08461877703666687,
+      "learning_rate": 0.0001,
+      "loss": 1.6938,
+      "step": 8600
+    },
+    {
+      "epoch": 0.11898375493473087,
+      "grad_norm": 0.08680044859647751,
+      "learning_rate": 0.0001,
+      "loss": 1.6914,
+      "step": 8650
+    },
+    {
+      "epoch": 0.1196715223042958,
+      "grad_norm": 0.0874258503317833,
+      "learning_rate": 0.0001,
+      "loss": 1.6886,
+      "step": 8700
+    },
+    {
+      "epoch": 0.12035928967386071,
+      "grad_norm": 0.08458847552537918,
+      "learning_rate": 0.0001,
+      "loss": 1.6889,
+      "step": 8750
+    },
+    {
+      "epoch": 0.12104705704342564,
+      "grad_norm": 0.08277745544910431,
+      "learning_rate": 0.0001,
+      "loss": 1.6955,
+      "step": 8800
+    },
+    {
+      "epoch": 0.12173482441299055,
+      "grad_norm": 0.08370788395404816,
+      "learning_rate": 0.0001,
+      "loss": 1.693,
+      "step": 8850
+    },
+    {
+      "epoch": 0.12242259178255546,
+      "grad_norm": 0.08819565922021866,
+      "learning_rate": 0.0001,
+      "loss": 1.6898,
+      "step": 8900
+    },
+    {
+      "epoch": 0.12311035915212039,
+      "grad_norm": 0.08511374890804291,
+      "learning_rate": 0.0001,
+      "loss": 1.6877,
+      "step": 8950
+    },
+    {
+      "epoch": 0.1237981265216853,
+      "grad_norm": 0.08699041604995728,
+      "learning_rate": 0.0001,
+      "loss": 1.6919,
+      "step": 9000
+    },
+    {
+      "epoch": 0.12448589389125023,
+      "grad_norm": 0.08358988165855408,
+      "learning_rate": 0.0001,
+      "loss": 1.692,
+      "step": 9050
+    },
+    {
+      "epoch": 0.12517366126081514,
+      "grad_norm": 0.08777803182601929,
+      "learning_rate": 0.0001,
+      "loss": 1.6906,
+      "step": 9100
+    },
+    {
+      "epoch": 0.12586142863038005,
+      "grad_norm": 0.08466688543558121,
+      "learning_rate": 0.0001,
+      "loss": 1.69,
+      "step": 9150
+    },
+    {
+      "epoch": 0.126549195999945,
+      "grad_norm": 0.09302639961242676,
+      "learning_rate": 0.0001,
+      "loss": 1.6905,
+      "step": 9200
+    },
+    {
+      "epoch": 0.1272369633695099,
+      "grad_norm": 0.085854172706604,
+      "learning_rate": 0.0001,
+      "loss": 1.6901,
+      "step": 9250
+    },
+    {
+      "epoch": 0.12792473073907482,
+      "grad_norm": 0.10086188465356827,
+      "learning_rate": 0.0001,
+      "loss": 1.6893,
+      "step": 9300
+    },
+    {
+      "epoch": 0.12861249810863973,
+      "grad_norm": 0.09083892405033112,
+      "learning_rate": 0.0001,
+      "loss": 1.6896,
+      "step": 9350
+    },
+    {
+      "epoch": 0.12930026547820464,
+      "grad_norm": 0.08710625022649765,
+      "learning_rate": 0.0001,
+      "loss": 1.6907,
+      "step": 9400
+    },
+    {
+      "epoch": 0.12998803284776958,
+      "grad_norm": 0.08261676877737045,
+      "learning_rate": 0.0001,
+      "loss": 1.6878,
+      "step": 9450
+    },
+    {
+      "epoch": 0.1306758002173345,
+      "grad_norm": 0.08663926273584366,
+      "learning_rate": 0.0001,
+      "loss": 1.686,
+      "step": 9500
+    },
+    {
+      "epoch": 0.1313635675868994,
+      "grad_norm": 0.09142150729894638,
+      "learning_rate": 0.0001,
+      "loss": 1.6848,
+      "step": 9550
+    },
+    {
+      "epoch": 0.13205133495646432,
+      "grad_norm": 0.0847783163189888,
+      "learning_rate": 0.0001,
+      "loss": 1.6907,
+      "step": 9600
+    },
+    {
+      "epoch": 0.13273910232602926,
+      "grad_norm": 0.08931740373373032,
+      "learning_rate": 0.0001,
+      "loss": 1.6863,
+      "step": 9650
+    },
+    {
+      "epoch": 0.13342686969559417,
+      "grad_norm": 0.10057684034109116,
+      "learning_rate": 0.0001,
+      "loss": 1.6874,
+      "step": 9700
+    },
+    {
+      "epoch": 0.13411463706515908,
+      "grad_norm": 0.08137702941894531,
+      "learning_rate": 0.0001,
+      "loss": 1.6899,
+      "step": 9750
+    },
+    {
+      "epoch": 0.134802404434724,
+      "grad_norm": 0.08957816660404205,
+      "learning_rate": 0.0001,
+      "loss": 1.6893,
+      "step": 9800
+    },
+    {
+      "epoch": 0.1354901718042889,
+      "grad_norm": 0.09797471016645432,
+      "learning_rate": 0.0001,
+      "loss": 1.6894,
+      "step": 9850
+    },
+    {
+      "epoch": 0.13617793917385385,
+      "grad_norm": 0.08011554181575775,
+      "learning_rate": 0.0001,
+      "loss": 1.6862,
+      "step": 9900
+    },
+    {
+      "epoch": 0.13686570654341876,
+      "grad_norm": 0.0851946547627449,
+      "learning_rate": 0.0001,
+      "loss": 1.6874,
+      "step": 9950
+    },
+    {
+      "epoch": 0.13755347391298367,
+      "grad_norm": 0.0880025252699852,
+      "learning_rate": 0.0001,
+      "loss": 1.6891,
+      "step": 10000
+    },
+    {
+      "epoch": 0.13824124128254858,
+      "grad_norm": 0.09258699417114258,
+      "learning_rate": 0.0001,
+      "loss": 1.6841,
+      "step": 10050
+    },
+    {
+      "epoch": 0.13892900865211352,
+      "grad_norm": 0.08737599104642868,
+      "learning_rate": 0.0001,
+      "loss": 1.6848,
+      "step": 10100
+    },
+    {
+      "epoch": 0.13961677602167843,
+      "grad_norm": 0.08749546855688095,
+      "learning_rate": 0.0001,
+      "loss": 1.686,
+      "step": 10150
+    },
+    {
+      "epoch": 0.14030454339124335,
+      "grad_norm": 0.09744163602590561,
+      "learning_rate": 0.0001,
+      "loss": 1.6864,
+      "step": 10200
+    },
+    {
+      "epoch": 0.14099231076080826,
+      "grad_norm": 0.08548065274953842,
+      "learning_rate": 0.0001,
+      "loss": 1.6868,
+      "step": 10250
+    },
+    {
+      "epoch": 0.14168007813037317,
+      "grad_norm": 0.0831003487110138,
+      "learning_rate": 0.0001,
+      "loss": 1.6886,
+      "step": 10300
+    },
+    {
+      "epoch": 0.1423678454999381,
+      "grad_norm": 0.08544298261404037,
+      "learning_rate": 0.0001,
+      "loss": 1.684,
+      "step": 10350
+    },
+    {
+      "epoch": 0.14305561286950302,
+      "grad_norm": 0.08695852756500244,
+      "learning_rate": 0.0001,
+      "loss": 1.6879,
+      "step": 10400
+    },
+    {
+      "epoch": 0.14374338023906794,
+      "grad_norm": 0.0854826420545578,
+      "learning_rate": 0.0001,
+      "loss": 1.685,
+      "step": 10450
+    },
+    {
+      "epoch": 0.14443114760863285,
+      "grad_norm": 0.08266725391149521,
+      "learning_rate": 0.0001,
+      "loss": 1.6854,
+      "step": 10500
+    },
+    {
+      "epoch": 0.14511891497819776,
+      "grad_norm": 0.08955837041139603,
+      "learning_rate": 0.0001,
+      "loss": 1.6856,
+      "step": 10550
+    },
+    {
+      "epoch": 0.1458066823477627,
+      "grad_norm": 0.11815349757671356,
+      "learning_rate": 0.0001,
+      "loss": 1.6823,
+      "step": 10600
+    },
+    {
+      "epoch": 0.1464944497173276,
+      "grad_norm": 0.08930766582489014,
+      "learning_rate": 0.0001,
+      "loss": 1.6848,
+      "step": 10650
+    },
+    {
+      "epoch": 0.14718221708689253,
+      "grad_norm": 0.08815452456474304,
+      "learning_rate": 0.0001,
+      "loss": 1.6819,
+      "step": 10700
+    },
+    {
+      "epoch": 0.14786998445645744,
+      "grad_norm": 0.09267221391201019,
+      "learning_rate": 0.0001,
+      "loss": 1.6829,
+      "step": 10750
+    },
+    {
+      "epoch": 0.14855775182602238,
+      "grad_norm": 0.08665202558040619,
+      "learning_rate": 0.0001,
+      "loss": 1.6802,
+      "step": 10800
+    },
+    {
+      "epoch": 0.1492455191955873,
+      "grad_norm": 0.083659827709198,
+      "learning_rate": 0.0001,
+      "loss": 1.6855,
+      "step": 10850
+    },
+    {
+      "epoch": 0.1499332865651522,
+      "grad_norm": 0.08788301050662994,
+      "learning_rate": 0.0001,
+      "loss": 1.6882,
+      "step": 10900
+    },
+    {
+      "epoch": 0.15062105393471711,
+      "grad_norm": 0.08252999931573868,
+      "learning_rate": 0.0001,
+      "loss": 1.6809,
+      "step": 10950
+    },
+    {
+      "epoch": 0.15130882130428203,
+      "grad_norm": 0.08625887334346771,
+      "learning_rate": 0.0001,
+      "loss": 1.6844,
+      "step": 11000
+    },
+    {
+      "epoch": 0.15199658867384697,
+      "grad_norm": 0.0943947359919548,
+      "learning_rate": 0.0001,
+      "loss": 1.684,
+      "step": 11050
+    },
+    {
+      "epoch": 0.15268435604341188,
+      "grad_norm": 0.08911854773759842,
+      "learning_rate": 0.0001,
+      "loss": 1.6873,
+      "step": 11100
+    },
+    {
+      "epoch": 0.1533721234129768,
+      "grad_norm": 0.08470957726240158,
+      "learning_rate": 0.0001,
+      "loss": 1.6807,
+      "step": 11150
+    },
+    {
+      "epoch": 0.1540598907825417,
+      "grad_norm": 0.08653486520051956,
+      "learning_rate": 0.0001,
+      "loss": 1.6822,
+      "step": 11200
+    },
+    {
+      "epoch": 0.15474765815210664,
+      "grad_norm": 0.08835257589817047,
+      "learning_rate": 0.0001,
+      "loss": 1.6827,
+      "step": 11250
+    },
+    {
+      "epoch": 0.15543542552167156,
+      "grad_norm": 0.08290986716747284,
+      "learning_rate": 0.0001,
+      "loss": 1.6795,
+      "step": 11300
+    },
+    {
+      "epoch": 0.15612319289123647,
+      "grad_norm": 0.36724504828453064,
+      "learning_rate": 0.0001,
+      "loss": 1.6793,
+      "step": 11350
+    },
+    {
+      "epoch": 0.15681096026080138,
+      "grad_norm": 0.0866614505648613,
+      "learning_rate": 0.0001,
+      "loss": 1.6815,
+      "step": 11400
+    },
+    {
+      "epoch": 0.1574987276303663,
+      "grad_norm": 0.09343897551298141,
+      "learning_rate": 0.0001,
+      "loss": 1.6824,
+      "step": 11450
+    },
+    {
+      "epoch": 0.15818649499993123,
+      "grad_norm": 0.09004870057106018,
+      "learning_rate": 0.0001,
+      "loss": 1.6821,
+      "step": 11500
+    },
+    {
+      "epoch": 0.15887426236949614,
+      "grad_norm": 0.08509238809347153,
+      "learning_rate": 0.0001,
+      "loss": 1.6816,
+      "step": 11550
+    },
+    {
+      "epoch": 0.15956202973906106,
+      "grad_norm": 0.09254981577396393,
+      "learning_rate": 0.0001,
+      "loss": 1.6821,
+      "step": 11600
+    },
+    {
+      "epoch": 0.16024979710862597,
+      "grad_norm": 0.08585833013057709,
+      "learning_rate": 0.0001,
+      "loss": 1.6812,
+      "step": 11650
+    },
+    {
+      "epoch": 0.1609375644781909,
+      "grad_norm": 0.08284648507833481,
+      "learning_rate": 0.0001,
+      "loss": 1.6843,
+      "step": 11700
+    },
+    {
+      "epoch": 0.16162533184775582,
+      "grad_norm": 0.08601938188076019,
+      "learning_rate": 0.0001,
+      "loss": 1.6829,
+      "step": 11750
+    },
+    {
+      "epoch": 0.16231309921732073,
+      "grad_norm": 0.08931539207696915,
+      "learning_rate": 0.0001,
+      "loss": 1.682,
+      "step": 11800
+    },
+    {
+      "epoch": 0.16300086658688565,
+      "grad_norm": 0.09171988070011139,
+      "learning_rate": 0.0001,
+      "loss": 1.6778,
+      "step": 11850
+    },
+    {
+      "epoch": 0.16368863395645056,
+      "grad_norm": 0.08799155801534653,
+      "learning_rate": 0.0001,
+      "loss": 1.6817,
+      "step": 11900
+    },
+    {
+      "epoch": 0.1643764013260155,
+      "grad_norm": 0.09240555018186569,
+      "learning_rate": 0.0001,
+      "loss": 1.681,
+      "step": 11950
+    },
+    {
+      "epoch": 0.1650641686955804,
+      "grad_norm": 0.08808182179927826,
+      "learning_rate": 0.0001,
+      "loss": 1.6796,
+      "step": 12000
+    },
+    {
+      "epoch": 0.16575193606514532,
+      "grad_norm": 0.09885094314813614,
+      "learning_rate": 0.0001,
+      "loss": 1.6763,
+      "step": 12050
+    },
+    {
+      "epoch": 0.16643970343471023,
+      "grad_norm": 0.0844956710934639,
+      "learning_rate": 0.0001,
+      "loss": 1.6812,
+      "step": 12100
+    },
+    {
+      "epoch": 0.16712747080427517,
+      "grad_norm": 0.08291836082935333,
+      "learning_rate": 0.0001,
+      "loss": 1.6803,
+      "step": 12150
+    },
+    {
+      "epoch": 0.1678152381738401,
+      "grad_norm": 0.085113026201725,
+      "learning_rate": 0.0001,
+      "loss": 1.6812,
+      "step": 12200
+    },
+    {
+      "epoch": 0.168503005543405,
+      "grad_norm": 0.09225833415985107,
+      "learning_rate": 0.0001,
+      "loss": 1.6814,
+      "step": 12250
+    },
+    {
+      "epoch": 0.1691907729129699,
+      "grad_norm": 0.09046541899442673,
+      "learning_rate": 0.0001,
+      "loss": 1.6763,
+      "step": 12300
+    },
+    {
+      "epoch": 0.16987854028253482,
+      "grad_norm": 0.08825410902500153,
+      "learning_rate": 0.0001,
+      "loss": 1.6804,
+      "step": 12350
+    },
+    {
+      "epoch": 0.17056630765209976,
+      "grad_norm": 0.08579885214567184,
+      "learning_rate": 0.0001,
+      "loss": 1.6768,
+      "step": 12400
+    },
+    {
+      "epoch": 0.17125407502166468,
+      "grad_norm": 0.08292757719755173,
+      "learning_rate": 0.0001,
+      "loss": 1.6787,
+      "step": 12450
+    },
+    {
+      "epoch": 0.1719418423912296,
+      "grad_norm": 0.08542559295892715,
+      "learning_rate": 0.0001,
+      "loss": 1.6805,
+      "step": 12500
+    },
+    {
+      "epoch": 0.1726296097607945,
+      "grad_norm": 0.08370521664619446,
+      "learning_rate": 0.0001,
+      "loss": 1.6802,
+      "step": 12550
+    },
+    {
+      "epoch": 0.17331737713035944,
+      "grad_norm": 0.08622439205646515,
+      "learning_rate": 0.0001,
+      "loss": 1.6768,
+      "step": 12600
+    },
+    {
+      "epoch": 0.17400514449992435,
+      "grad_norm": 0.08716084808111191,
+      "learning_rate": 0.0001,
+      "loss": 1.6809,
+      "step": 12650
+    },
+    {
+      "epoch": 0.17469291186948926,
+      "grad_norm": 0.09401235729455948,
+      "learning_rate": 0.0001,
+      "loss": 1.6774,
+      "step": 12700
+    },
+    {
+      "epoch": 0.17538067923905418,
+      "grad_norm": 0.08970025926828384,
+      "learning_rate": 0.0001,
+      "loss": 1.6756,
+      "step": 12750
+    },
+    {
+      "epoch": 0.1760684466086191,
+      "grad_norm": 0.09375713020563126,
+      "learning_rate": 0.0001,
+      "loss": 1.6758,
+      "step": 12800
+    },
+    {
+      "epoch": 0.17675621397818403,
+      "grad_norm": 0.09062221646308899,
+      "learning_rate": 0.0001,
+      "loss": 1.6775,
+      "step": 12850
+    },
+    {
+      "epoch": 0.17744398134774894,
+      "grad_norm": 0.08808109164237976,
+      "learning_rate": 0.0001,
+      "loss": 1.6783,
+      "step": 12900
+    },
+    {
+      "epoch": 0.17813174871731385,
+      "grad_norm": 0.08363673090934753,
+      "learning_rate": 0.0001,
+      "loss": 1.6796,
+      "step": 12950
+    },
+    {
+      "epoch": 0.17881951608687877,
+      "grad_norm": 0.09371782094240189,
+      "learning_rate": 0.0001,
+      "loss": 1.6757,
+      "step": 13000
+    },
+    {
+      "epoch": 0.1795072834564437,
+      "grad_norm": 0.09487029165029526,
+      "learning_rate": 0.0001,
+      "loss": 1.6801,
+      "step": 13050
+    },
+    {
+      "epoch": 0.18019505082600862,
+      "grad_norm": 0.09358393400907516,
+      "learning_rate": 0.0001,
+      "loss": 1.6772,
+      "step": 13100
+    },
+    {
+      "epoch": 0.18088281819557353,
+      "grad_norm": 0.09072808176279068,
+      "learning_rate": 0.0001,
+      "loss": 1.6764,
+      "step": 13150
+    },
+    {
+      "epoch": 0.18157058556513844,
+      "grad_norm": 0.10067198425531387,
+      "learning_rate": 0.0001,
+      "loss": 1.6779,
+      "step": 13200
+    },
+    {
+      "epoch": 0.18225835293470335,
+      "grad_norm": 0.0920378789305687,
+      "learning_rate": 0.0001,
+      "loss": 1.6737,
+      "step": 13250
+    },
+    {
+      "epoch": 0.1829461203042683,
+      "grad_norm": 0.09141296148300171,
+      "learning_rate": 0.0001,
+      "loss": 1.68,
+      "step": 13300
+    },
+    {
+      "epoch": 0.1836338876738332,
+      "grad_norm": 0.09030226618051529,
+      "learning_rate": 0.0001,
+      "loss": 1.6786,
+      "step": 13350
+    },
+    {
+      "epoch": 0.18432165504339812,
+      "grad_norm": 0.09276604652404785,
+      "learning_rate": 0.0001,
+      "loss": 1.681,
+      "step": 13400
+    },
+    {
+      "epoch": 0.18500942241296303,
+      "grad_norm": 0.08355134725570679,
+      "learning_rate": 0.0001,
+      "loss": 1.6777,
+      "step": 13450
+    },
+    {
+      "epoch": 0.18569718978252797,
+      "grad_norm": 0.0854455828666687,
+      "learning_rate": 0.0001,
+      "loss": 1.6763,
+      "step": 13500
+    },
+    {
+      "epoch": 0.18638495715209288,
+      "grad_norm": 0.09083615243434906,
+      "learning_rate": 0.0001,
+      "loss": 1.6734,
+      "step": 13550
+    },
+    {
+      "epoch": 0.1870727245216578,
+      "grad_norm": 0.08724341541528702,
+      "learning_rate": 0.0001,
+      "loss": 1.6751,
+      "step": 13600
+    },
+    {
+      "epoch": 0.1877604918912227,
+      "grad_norm": 0.08758584409952164,
+      "learning_rate": 0.0001,
+      "loss": 1.6751,
+      "step": 13650
+    },
+    {
+      "epoch": 0.18844825926078762,
+      "grad_norm": 0.09144899994134903,
+      "learning_rate": 0.0001,
+      "loss": 1.6779,
+      "step": 13700
+    },
+    {
+      "epoch": 0.18913602663035256,
+      "grad_norm": 0.09547972679138184,
+      "learning_rate": 0.0001,
+      "loss": 1.6739,
+      "step": 13750
+    },
+    {
+      "epoch": 0.18982379399991747,
+      "grad_norm": 0.09593083709478378,
+      "learning_rate": 0.0001,
+      "loss": 1.6762,
+      "step": 13800
+    },
+    {
+      "epoch": 0.19051156136948239,
+      "grad_norm": 0.09824243932962418,
+      "learning_rate": 0.0001,
+      "loss": 1.6744,
+      "step": 13850
+    },
+    {
+      "epoch": 0.1911993287390473,
+      "grad_norm": 0.08824923634529114,
+      "learning_rate": 0.0001,
+      "loss": 1.6771,
+      "step": 13900
+    },
+    {
+      "epoch": 0.1918870961086122,
+      "grad_norm": 0.09215164929628372,
+      "learning_rate": 0.0001,
+      "loss": 1.6786,
+      "step": 13950
+    },
+    {
+      "epoch": 0.19257486347817715,
+      "grad_norm": 0.09035980701446533,
+      "learning_rate": 0.0001,
+      "loss": 1.6734,
+      "step": 14000
+    },
+    {
+      "epoch": 0.19326263084774206,
+      "grad_norm": 0.09424874931573868,
+      "learning_rate": 0.0001,
+      "loss": 1.6753,
+      "step": 14050
+    },
+    {
+      "epoch": 0.19395039821730697,
+      "grad_norm": 0.08970798552036285,
+      "learning_rate": 0.0001,
+      "loss": 1.6771,
+      "step": 14100
+    },
+    {
+      "epoch": 0.1946381655868719,
+      "grad_norm": 0.0905926451086998,
+      "learning_rate": 0.0001,
+      "loss": 1.6734,
+      "step": 14150
+    },
+    {
+      "epoch": 0.19532593295643683,
+      "grad_norm": 0.0877937376499176,
+      "learning_rate": 0.0001,
+      "loss": 1.674,
+      "step": 14200
+    },
+    {
+      "epoch": 0.19601370032600174,
+      "grad_norm": 0.08889397233724594,
+      "learning_rate": 0.0001,
+      "loss": 1.6778,
+      "step": 14250
+    },
+    {
+      "epoch": 0.19670146769556665,
+      "grad_norm": 0.09094177931547165,
+      "learning_rate": 0.0001,
+      "loss": 1.6791,
+      "step": 14300
+    },
+    {
+      "epoch": 0.19738923506513156,
+      "grad_norm": 0.08906359225511551,
+      "learning_rate": 0.0001,
+      "loss": 1.6746,
+      "step": 14350
+    },
+    {
+      "epoch": 0.19807700243469648,
+      "grad_norm": 0.086732417345047,
+      "learning_rate": 0.0001,
+      "loss": 1.6729,
+      "step": 14400
+    },
+    {
+      "epoch": 0.19876476980426142,
+      "grad_norm": 0.08629673719406128,
+      "learning_rate": 0.0001,
+      "loss": 1.6768,
+      "step": 14450
+    },
+    {
+      "epoch": 0.19945253717382633,
+      "grad_norm": 0.08930768072605133,
+      "learning_rate": 0.0001,
+      "loss": 1.674,
+      "step": 14500
+    },
+    {
+      "epoch": 0.20014030454339124,
+      "grad_norm": 0.09292076528072357,
+      "learning_rate": 0.0001,
+      "loss": 1.6733,
+      "step": 14550
+    },
+    {
+      "epoch": 0.20082807191295615,
+      "grad_norm": 0.09250384569168091,
+      "learning_rate": 0.0001,
+      "loss": 1.6716,
+      "step": 14600
+    },
+    {
+      "epoch": 0.2015158392825211,
+      "grad_norm": 0.0884978249669075,
+      "learning_rate": 0.0001,
+      "loss": 1.6727,
+      "step": 14650
+    },
+    {
+      "epoch": 0.202203606652086,
+      "grad_norm": 0.09365909546613693,
+      "learning_rate": 0.0001,
+      "loss": 1.6732,
+      "step": 14700
+    },
+    {
+      "epoch": 0.20289137402165092,
+      "grad_norm": 0.09228827059268951,
+      "learning_rate": 0.0001,
+      "loss": 1.6774,
+      "step": 14750
+    },
+    {
+      "epoch": 0.20357914139121583,
+      "grad_norm": 0.08782710880041122,
+      "learning_rate": 0.0001,
+      "loss": 1.6724,
+      "step": 14800
+    },
+    {
+      "epoch": 0.20426690876078074,
+      "grad_norm": 0.09346231073141098,
+      "learning_rate": 0.0001,
+      "loss": 1.6758,
+      "step": 14850
+    },
+    {
+      "epoch": 0.20495467613034568,
+      "grad_norm": 0.08543895184993744,
+      "learning_rate": 0.0001,
+      "loss": 1.6719,
+      "step": 14900
+    },
+    {
+      "epoch": 0.2056424434999106,
+      "grad_norm": 0.08667740225791931,
+      "learning_rate": 0.0001,
+      "loss": 1.675,
+      "step": 14950
+    },
+    {
+      "epoch": 0.2063302108694755,
+      "grad_norm": 0.09252138435840607,
+      "learning_rate": 0.0001,
+      "loss": 1.6731,
+      "step": 15000
+    },
+    {
+      "epoch": 0.20701797823904042,
+      "grad_norm": 0.09669343382120132,
+      "learning_rate": 0.0001,
+      "loss": 1.6721,
+      "step": 15050
+    },
+    {
+      "epoch": 0.20770574560860536,
+      "grad_norm": 0.08828496187925339,
+      "learning_rate": 0.0001,
+      "loss": 1.6701,
+      "step": 15100
+    },
+    {
+      "epoch": 0.20839351297817027,
+      "grad_norm": 0.08292458951473236,
+      "learning_rate": 0.0001,
+      "loss": 1.6754,
+      "step": 15150
+    },
+    {
+      "epoch": 0.20908128034773518,
+      "grad_norm": 0.085512176156044,
+      "learning_rate": 0.0001,
+      "loss": 1.6758,
+      "step": 15200
+    },
+    {
+      "epoch": 0.2097690477173001,
+      "grad_norm": 0.08498524874448776,
+      "learning_rate": 0.0001,
+      "loss": 1.674,
+      "step": 15250
+    },
+    {
+      "epoch": 0.210456815086865,
+      "grad_norm": 0.08492223173379898,
+      "learning_rate": 0.0001,
+      "loss": 1.6758,
+      "step": 15300
+    },
+    {
+      "epoch": 0.21114458245642995,
+      "grad_norm": 0.08872613310813904,
+      "learning_rate": 0.0001,
+      "loss": 1.6722,
+      "step": 15350
+    },
+    {
+      "epoch": 0.21183234982599486,
+      "grad_norm": 0.08605389297008514,
+      "learning_rate": 0.0001,
+      "loss": 1.6783,
+      "step": 15400
+    },
+    {
+      "epoch": 0.21252011719555977,
+      "grad_norm": 0.09041046351194382,
+      "learning_rate": 0.0001,
+      "loss": 1.6735,
+      "step": 15450
+    },
+    {
+      "epoch": 0.21320788456512468,
+      "grad_norm": 0.09070464968681335,
+      "learning_rate": 0.0001,
+      "loss": 1.6758,
+      "step": 15500
+    },
+    {
+      "epoch": 0.21389565193468962,
+      "grad_norm": 0.09620167315006256,
+      "learning_rate": 0.0001,
+      "loss": 1.6707,
+      "step": 15550
+    },
+    {
+      "epoch": 0.21458341930425454,
+      "grad_norm": 0.08734453469514847,
+      "learning_rate": 0.0001,
+      "loss": 1.6693,
+      "step": 15600
+    },
+    {
+      "epoch": 0.21527118667381945,
+      "grad_norm": 0.09259238839149475,
+      "learning_rate": 0.0001,
+      "loss": 1.6723,
+      "step": 15650
+    },
+    {
+      "epoch": 0.21595895404338436,
+      "grad_norm": 0.08994511514902115,
+      "learning_rate": 0.0001,
+      "loss": 1.6734,
+      "step": 15700
+    },
+    {
+      "epoch": 0.21664672141294927,
+      "grad_norm": 0.08940455317497253,
+      "learning_rate": 0.0001,
+      "loss": 1.6694,
+      "step": 15750
+    },
+    {
+      "epoch": 0.2173344887825142,
+      "grad_norm": 0.08544927090406418,
+      "learning_rate": 0.0001,
+      "loss": 1.6726,
+      "step": 15800
+    },
+    {
+      "epoch": 0.21802225615207912,
+      "grad_norm": 0.08161517977714539,
+      "learning_rate": 0.0001,
+      "loss": 1.672,
+      "step": 15850
+    },
+    {
+      "epoch": 0.21871002352164404,
+      "grad_norm": 0.09207966923713684,
+      "learning_rate": 0.0001,
+      "loss": 1.6709,
+      "step": 15900
+    },
+    {
+      "epoch": 0.21939779089120895,
+      "grad_norm": 0.08921949565410614,
+      "learning_rate": 0.0001,
+      "loss": 1.6707,
+      "step": 15950
+    },
+    {
+      "epoch": 0.2200855582607739,
+      "grad_norm": 0.08339584618806839,
+      "learning_rate": 0.0001,
+      "loss": 1.6713,
+      "step": 16000
+    },
+    {
+      "epoch": 0.2207733256303388,
+      "grad_norm": 0.08629053831100464,
+      "learning_rate": 0.0001,
+      "loss": 1.6707,
+      "step": 16050
+    },
+    {
+      "epoch": 0.2214610929999037,
+      "grad_norm": 0.08945165574550629,
+      "learning_rate": 0.0001,
+      "loss": 1.6706,
+      "step": 16100
+    },
+    {
+      "epoch": 0.22214886036946863,
+      "grad_norm": 0.09606142342090607,
+      "learning_rate": 0.0001,
+      "loss": 1.6741,
+      "step": 16150
+    },
+    {
+      "epoch": 0.22283662773903354,
+      "grad_norm": 0.08945801854133606,
+      "learning_rate": 0.0001,
+      "loss": 1.6725,
+      "step": 16200
+    },
+    {
+      "epoch": 0.22352439510859848,
+      "grad_norm": 0.09277788549661636,
+      "learning_rate": 0.0001,
+      "loss": 1.6673,
+      "step": 16250
+    },
+    {
+      "epoch": 0.2242121624781634,
+      "grad_norm": 0.08870001137256622,
+      "learning_rate": 0.0001,
+      "loss": 1.6717,
+      "step": 16300
+    },
+    {
+      "epoch": 0.2248999298477283,
+      "grad_norm": 0.09212413430213928,
+      "learning_rate": 0.0001,
+      "loss": 1.6696,
+      "step": 16350
+    },
+    {
+      "epoch": 0.22558769721729321,
+      "grad_norm": 0.09183049201965332,
+      "learning_rate": 0.0001,
+      "loss": 1.6723,
+      "step": 16400
+    },
+    {
+      "epoch": 0.22627546458685815,
+      "grad_norm": 0.08794934302568436,
+      "learning_rate": 0.0001,
+      "loss": 1.6694,
+      "step": 16450
+    },
+    {
+      "epoch": 0.22696323195642307,
+      "grad_norm": 0.08112750947475433,
+      "learning_rate": 0.0001,
+      "loss": 1.6703,
+      "step": 16500
+    },
+    {
+      "epoch": 0.22765099932598798,
+      "grad_norm": 0.09124380350112915,
+      "learning_rate": 0.0001,
+      "loss": 1.6697,
+      "step": 16550
+    },
+    {
+      "epoch": 0.2283387666955529,
+      "grad_norm": 0.09015939384698868,
+      "learning_rate": 0.0001,
+      "loss": 1.6716,
+      "step": 16600
+    },
+    {
+      "epoch": 0.2290265340651178,
+      "grad_norm": 0.08522862941026688,
+      "learning_rate": 0.0001,
+      "loss": 1.6708,
+      "step": 16650
+    },
+    {
+      "epoch": 0.22971430143468274,
+      "grad_norm": 0.0881582573056221,
+      "learning_rate": 0.0001,
+      "loss": 1.671,
+      "step": 16700
+    },
+    {
+      "epoch": 0.23040206880424766,
+      "grad_norm": 0.09910543262958527,
+      "learning_rate": 0.0001,
+      "loss": 1.6674,
+      "step": 16750
+    },
+    {
+      "epoch": 0.23108983617381257,
+      "grad_norm": 0.08917541056871414,
+      "learning_rate": 0.0001,
+      "loss": 1.6682,
+      "step": 16800
+    },
+    {
+      "epoch": 0.23177760354337748,
+      "grad_norm": 0.1020062193274498,
+      "learning_rate": 0.0001,
+      "loss": 1.6699,
+      "step": 16850
+    },
+    {
+      "epoch": 0.2324653709129424,
+      "grad_norm": 0.09061840176582336,
+      "learning_rate": 0.0001,
+      "loss": 1.6651,
+      "step": 16900
+    },
+    {
+      "epoch": 0.23315313828250733,
+      "grad_norm": 0.08623366057872772,
+      "learning_rate": 0.0001,
+      "loss": 1.6663,
+      "step": 16950
+    },
+    {
+      "epoch": 0.23384090565207225,
+      "grad_norm": 0.09871737658977509,
+      "learning_rate": 0.0001,
+      "loss": 1.6686,
+      "step": 17000
+    },
+    {
+      "epoch": 0.23452867302163716,
+      "grad_norm": 0.08913227170705795,
+      "learning_rate": 0.0001,
+      "loss": 1.6701,
+      "step": 17050
+    },
+    {
+      "epoch": 0.23521644039120207,
+      "grad_norm": 0.08803007751703262,
+      "learning_rate": 0.0001,
+      "loss": 1.67,
+      "step": 17100
+    },
+    {
+      "epoch": 0.235904207760767,
+      "grad_norm": 0.08955027163028717,
+      "learning_rate": 0.0001,
+      "loss": 1.6726,
+      "step": 17150
+    },
+    {
+      "epoch": 0.23659197513033192,
+      "grad_norm": 0.09248829632997513,
+      "learning_rate": 0.0001,
+      "loss": 1.6669,
+      "step": 17200
+    },
+    {
+      "epoch": 0.23727974249989683,
+      "grad_norm": 0.0897279903292656,
+      "learning_rate": 0.0001,
+      "loss": 1.6715,
+      "step": 17250
+    },
+    {
+      "epoch": 0.23796750986946175,
+      "grad_norm": 0.09461032599210739,
+      "learning_rate": 0.0001,
+      "loss": 1.6683,
+      "step": 17300
+    },
+    {
+      "epoch": 0.23865527723902666,
+      "grad_norm": 0.09740615636110306,
+      "learning_rate": 0.0001,
+      "loss": 1.6717,
+      "step": 17350
+    },
+    {
+      "epoch": 0.2393430446085916,
+      "grad_norm": 0.08833985030651093,
+      "learning_rate": 0.0001,
+      "loss": 1.6629,
+      "step": 17400
+    },
+    {
+      "epoch": 0.2400308119781565,
+      "grad_norm": 0.08585455268621445,
+      "learning_rate": 0.0001,
+      "loss": 1.6701,
+      "step": 17450
+    },
+    {
+      "epoch": 0.24071857934772142,
+      "grad_norm": 0.08493249863386154,
+      "learning_rate": 0.0001,
+      "loss": 1.6678,
+      "step": 17500
+    },
+    {
+      "epoch": 0.24140634671728634,
+      "grad_norm": 0.09439011663198471,
+      "learning_rate": 0.0001,
+      "loss": 1.6693,
+      "step": 17550
+    },
+    {
+      "epoch": 0.24209411408685128,
+      "grad_norm": 0.08851542323827744,
+      "learning_rate": 0.0001,
+      "loss": 1.6664,
+      "step": 17600
+    },
+    {
+      "epoch": 0.2427818814564162,
+      "grad_norm": 0.09400813281536102,
+      "learning_rate": 0.0001,
+      "loss": 1.6686,
+      "step": 17650
+    },
+    {
+      "epoch": 0.2434696488259811,
+      "grad_norm": 0.09821904450654984,
+      "learning_rate": 0.0001,
+      "loss": 1.665,
+      "step": 17700
+    },
+    {
+      "epoch": 0.244157416195546,
+      "grad_norm": 0.09214925765991211,
+      "learning_rate": 0.0001,
+      "loss": 1.6662,
+      "step": 17750
+    },
+    {
+      "epoch": 0.24484518356511092,
+      "grad_norm": 0.09265078604221344,
+      "learning_rate": 0.0001,
+      "loss": 1.6653,
+      "step": 17800
+    },
+    {
+      "epoch": 0.24553295093467586,
+      "grad_norm": 0.08704175055027008,
+      "learning_rate": 0.0001,
+      "loss": 1.6695,
+      "step": 17850
+    },
+    {
+      "epoch": 0.24622071830424078,
+      "grad_norm": 0.09113669395446777,
+      "learning_rate": 0.0001,
+      "loss": 1.6695,
+      "step": 17900
+    },
+    {
+      "epoch": 0.2469084856738057,
+      "grad_norm": 0.09568019956350327,
+      "learning_rate": 0.0001,
+      "loss": 1.6646,
+      "step": 17950
+    },
+    {
+      "epoch": 0.2475962530433706,
+      "grad_norm": 0.08732044696807861,
+      "learning_rate": 0.0001,
+      "loss": 1.6682,
+      "step": 18000
+    },
+    {
+      "epoch": 0.24828402041293554,
+      "grad_norm": 0.10421797633171082,
+      "learning_rate": 0.0001,
+      "loss": 1.6665,
+      "step": 18050
+    },
+    {
+      "epoch": 0.24897178778250045,
+      "grad_norm": 0.0909595936536789,
+      "learning_rate": 0.0001,
+      "loss": 1.668,
+      "step": 18100
+    },
+    {
+      "epoch": 0.24965955515206537,
+      "grad_norm": 0.088596411049366,
+      "learning_rate": 0.0001,
+      "loss": 1.6643,
+      "step": 18150
+    },
+    {
+      "epoch": 0.2503473225216303,
+      "grad_norm": 0.0889756828546524,
+      "learning_rate": 0.0001,
+      "loss": 1.6658,
+      "step": 18200
+    },
+    {
+      "epoch": 0.2510350898911952,
+      "grad_norm": 0.09015469253063202,
+      "learning_rate": 0.0001,
+      "loss": 1.6648,
+      "step": 18250
+    },
+    {
+      "epoch": 0.2517228572607601,
+      "grad_norm": 0.08240851014852524,
+      "learning_rate": 0.0001,
+      "loss": 1.6655,
+      "step": 18300
+    },
+    {
+      "epoch": 0.25241062463032504,
+      "grad_norm": 0.08527081459760666,
+      "learning_rate": 0.0001,
+      "loss": 1.6738,
+      "step": 18350
+    },
+    {
+      "epoch": 0.25309839199989,
+      "grad_norm": 0.0891634002327919,
+      "learning_rate": 0.0001,
+      "loss": 1.6659,
+      "step": 18400
+    },
+    {
+      "epoch": 0.25378615936945487,
+      "grad_norm": 0.08512425422668457,
+      "learning_rate": 0.0001,
+      "loss": 1.6685,
+      "step": 18450
+    },
+    {
+      "epoch": 0.2544739267390198,
+      "grad_norm": 0.08884911984205246,
+      "learning_rate": 0.0001,
+      "loss": 1.6631,
+      "step": 18500
+    },
+    {
+      "epoch": 0.2551616941085847,
+      "grad_norm": 0.09091176837682724,
+      "learning_rate": 0.0001,
+      "loss": 1.6677,
+      "step": 18550
+    },
+    {
+      "epoch": 0.25584946147814963,
+      "grad_norm": 0.09358973056077957,
+      "learning_rate": 0.0001,
+      "loss": 1.6668,
+      "step": 18600
+    },
+    {
+      "epoch": 0.25653722884771457,
+      "grad_norm": 0.08879607170820236,
+      "learning_rate": 0.0001,
+      "loss": 1.6675,
+      "step": 18650
+    },
+    {
+      "epoch": 0.25722499621727946,
+      "grad_norm": 0.08596090972423553,
+      "learning_rate": 0.0001,
+      "loss": 1.6684,
+      "step": 18700
+    },
+    {
+      "epoch": 0.2579127635868444,
+      "grad_norm": 0.08910541981458664,
+      "learning_rate": 0.0001,
+      "loss": 1.6631,
+      "step": 18750
+    },
+    {
+      "epoch": 0.2586005309564093,
+      "grad_norm": 0.09641165286302567,
+      "learning_rate": 0.0001,
+      "loss": 1.6677,
+      "step": 18800
+    },
+    {
+      "epoch": 0.2592882983259742,
+      "grad_norm": 0.08356452733278275,
+      "learning_rate": 0.0001,
+      "loss": 1.6622,
+      "step": 18850
+    },
+    {
+      "epoch": 0.25997606569553916,
+      "grad_norm": 0.09701082855463028,
+      "learning_rate": 0.0001,
+      "loss": 1.6651,
+      "step": 18900
+    },
+    {
+      "epoch": 0.26066383306510404,
+      "grad_norm": 0.09678094834089279,
+      "learning_rate": 0.0001,
+      "loss": 1.6647,
+      "step": 18950
+    },
+    {
+      "epoch": 0.261351600434669,
+      "grad_norm": 0.0965198501944542,
+      "learning_rate": 0.0001,
+      "loss": 1.6637,
+      "step": 19000
+    },
+    {
+      "epoch": 0.26203936780423387,
+      "grad_norm": 0.08802270889282227,
+      "learning_rate": 0.0001,
+      "loss": 1.6668,
+      "step": 19050
+    },
+    {
+      "epoch": 0.2627271351737988,
+      "grad_norm": 0.09215667843818665,
+      "learning_rate": 0.0001,
+      "loss": 1.6654,
+      "step": 19100
+    },
+    {
+      "epoch": 0.26341490254336375,
+      "grad_norm": 0.08846131712198257,
+      "learning_rate": 0.0001,
+      "loss": 1.6646,
+      "step": 19150
+    },
+    {
+      "epoch": 0.26410266991292863,
+      "grad_norm": 0.09379055351018906,
+      "learning_rate": 0.0001,
+      "loss": 1.6637,
+      "step": 19200
+    },
+    {
+      "epoch": 0.2647904372824936,
+      "grad_norm": 0.08257373422384262,
+      "learning_rate": 0.0001,
+      "loss": 1.6599,
+      "step": 19250
+    },
+    {
+      "epoch": 0.2654782046520585,
+      "grad_norm": 0.10151083022356033,
+      "learning_rate": 0.0001,
+      "loss": 1.6632,
+      "step": 19300
+    },
+    {
+      "epoch": 0.2661659720216234,
+      "grad_norm": 0.08694400638341904,
+      "learning_rate": 0.0001,
+      "loss": 1.6627,
+      "step": 19350
+    },
+    {
+      "epoch": 0.26685373939118834,
+      "grad_norm": 0.10145068168640137,
+      "learning_rate": 0.0001,
+      "loss": 1.6624,
+      "step": 19400
+    },
+    {
+      "epoch": 0.2675415067607532,
+      "grad_norm": 0.09250553697347641,
+      "learning_rate": 0.0001,
+      "loss": 1.6654,
+      "step": 19450
+    },
+    {
+      "epoch": 0.26822927413031816,
+      "grad_norm": 0.09129736572504044,
+      "learning_rate": 0.0001,
+      "loss": 1.6639,
+      "step": 19500
+    },
+    {
+      "epoch": 0.2689170414998831,
+      "grad_norm": 0.08812187612056732,
+      "learning_rate": 0.0001,
+      "loss": 1.6626,
+      "step": 19550
+    },
+    {
+      "epoch": 0.269604808869448,
+      "grad_norm": 0.08967280387878418,
+      "learning_rate": 0.0001,
+      "loss": 1.6588,
+      "step": 19600
+    },
+    {
+      "epoch": 0.2702925762390129,
+      "grad_norm": 0.12475176900625229,
+      "learning_rate": 0.0001,
+      "loss": 1.6655,
+      "step": 19650
+    },
+    {
+      "epoch": 0.2709803436085778,
+      "grad_norm": 0.09744782745838165,
+      "learning_rate": 0.0001,
+      "loss": 1.6617,
+      "step": 19700
+    },
+    {
+      "epoch": 0.27166811097814275,
+      "grad_norm": 0.09547289460897446,
+      "learning_rate": 0.0001,
+      "loss": 1.669,
+      "step": 19750
+    },
+    {
+      "epoch": 0.2723558783477077,
+      "grad_norm": 0.10474015772342682,
+      "learning_rate": 0.0001,
+      "loss": 1.6626,
+      "step": 19800
+    },
+    {
+      "epoch": 0.2730436457172726,
+      "grad_norm": 0.09450657665729523,
+      "learning_rate": 0.0001,
+      "loss": 1.661,
+      "step": 19850
+    },
+    {
+      "epoch": 0.2737314130868375,
+      "grad_norm": 0.09040903300046921,
+      "learning_rate": 0.0001,
+      "loss": 1.6633,
+      "step": 19900
+    },
+    {
+      "epoch": 0.2744191804564024,
+      "grad_norm": 0.09171809256076813,
+      "learning_rate": 0.0001,
+      "loss": 1.6632,
+      "step": 19950
+    },
+    {
+      "epoch": 0.27510694782596734,
+      "grad_norm": 0.08326128870248795,
+      "learning_rate": 0.0001,
+      "loss": 1.6641,
+      "step": 20000
+    },
+    {
+      "epoch": 0.2757947151955323,
+      "grad_norm": 0.08934567868709564,
+      "learning_rate": 0.0001,
+      "loss": 1.6639,
+      "step": 20050
+    },
+    {
+      "epoch": 0.27648248256509717,
+      "grad_norm": 0.08973874151706696,
+      "learning_rate": 0.0001,
+      "loss": 1.6617,
+      "step": 20100
+    },
+    {
+      "epoch": 0.2771702499346621,
+      "grad_norm": 0.08561202138662338,
+      "learning_rate": 0.0001,
+      "loss": 1.6669,
+      "step": 20150
+    },
+    {
+      "epoch": 0.27785801730422705,
+      "grad_norm": 0.09638834744691849,
+      "learning_rate": 0.0001,
+      "loss": 1.6621,
+      "step": 20200
+    },
+    {
+      "epoch": 0.27854578467379193,
+      "grad_norm": 0.08998638391494751,
+      "learning_rate": 0.0001,
+      "loss": 1.6614,
+      "step": 20250
+    },
+    {
+      "epoch": 0.27923355204335687,
+      "grad_norm": 0.0963120236992836,
+      "learning_rate": 0.0001,
+      "loss": 1.6643,
+      "step": 20300
+    },
+    {
+      "epoch": 0.27992131941292175,
+      "grad_norm": 0.09031634032726288,
+      "learning_rate": 0.0001,
+      "loss": 1.6628,
+      "step": 20350
+    },
+    {
+      "epoch": 0.2806090867824867,
+      "grad_norm": 0.08368144184350967,
+      "learning_rate": 0.0001,
+      "loss": 1.6619,
+      "step": 20400
+    },
+    {
+      "epoch": 0.28129685415205163,
+      "grad_norm": 0.08716925233602524,
+      "learning_rate": 0.0001,
+      "loss": 1.6602,
+      "step": 20450
+    },
+    {
+      "epoch": 0.2819846215216165,
+      "grad_norm": 0.08537859469652176,
+      "learning_rate": 0.0001,
+      "loss": 1.6605,
+      "step": 20500
+    },
+    {
+      "epoch": 0.28267238889118146,
+      "grad_norm": 0.08953417092561722,
+      "learning_rate": 0.0001,
+      "loss": 1.6596,
+      "step": 20550
+    },
+    {
+      "epoch": 0.28336015626074634,
+      "grad_norm": 0.09007777273654938,
+      "learning_rate": 0.0001,
+      "loss": 1.6609,
+      "step": 20600
+    },
+    {
+      "epoch": 0.2840479236303113,
+      "grad_norm": 0.09223859757184982,
+      "learning_rate": 0.0001,
+      "loss": 1.6624,
+      "step": 20650
+    },
+    {
+      "epoch": 0.2847356909998762,
+      "grad_norm": 0.08682399988174438,
+      "learning_rate": 0.0001,
+      "loss": 1.6611,
+      "step": 20700
+    },
+    {
+      "epoch": 0.2854234583694411,
+      "grad_norm": 0.0930657759308815,
+      "learning_rate": 0.0001,
+      "loss": 1.6636,
+      "step": 20750
+    },
+    {
+      "epoch": 0.28611122573900605,
+      "grad_norm": 0.08895847946405411,
+      "learning_rate": 0.0001,
+      "loss": 1.6613,
+      "step": 20800
+    },
+    {
+      "epoch": 0.28679899310857093,
+      "grad_norm": 0.09470313787460327,
+      "learning_rate": 0.0001,
+      "loss": 1.6613,
+      "step": 20850
+    },
+    {
+      "epoch": 0.28748676047813587,
+      "grad_norm": 0.08398160338401794,
+      "learning_rate": 0.0001,
+      "loss": 1.6591,
+      "step": 20900
+    },
+    {
+      "epoch": 0.2881745278477008,
+      "grad_norm": 0.09906968474388123,
+      "learning_rate": 0.0001,
+      "loss": 1.6588,
+      "step": 20950
+    },
+    {
+      "epoch": 0.2888622952172657,
+      "grad_norm": 0.08456803858280182,
+      "learning_rate": 0.0001,
+      "loss": 1.6653,
+      "step": 21000
+    },
+    {
+      "epoch": 0.28955006258683064,
+      "grad_norm": 0.09115090221166611,
+      "learning_rate": 0.0001,
+      "loss": 1.6604,
+      "step": 21050
+    },
+    {
+      "epoch": 0.2902378299563955,
+      "grad_norm": 0.09190023690462112,
+      "learning_rate": 0.0001,
+      "loss": 1.6618,
+      "step": 21100
+    },
+    {
+      "epoch": 0.29092559732596046,
+      "grad_norm": 0.0813484638929367,
+      "learning_rate": 0.0001,
+      "loss": 1.6638,
+      "step": 21150
+    },
+    {
+      "epoch": 0.2916133646955254,
+      "grad_norm": 0.09336463361978531,
+      "learning_rate": 0.0001,
+      "loss": 1.6597,
+      "step": 21200
+    },
+    {
+      "epoch": 0.2923011320650903,
+      "grad_norm": 0.08862286061048508,
+      "learning_rate": 0.0001,
+      "loss": 1.6606,
+      "step": 21250
+    },
+    {
+      "epoch": 0.2929888994346552,
+      "grad_norm": 0.09896854311227798,
+      "learning_rate": 0.0001,
+      "loss": 1.6637,
+      "step": 21300
+    },
+    {
+      "epoch": 0.29367666680422017,
+      "grad_norm": 0.08877858519554138,
+      "learning_rate": 0.0001,
+      "loss": 1.6625,
+      "step": 21350
+    },
+    {
+      "epoch": 0.29436443417378505,
+      "grad_norm": 0.0959668979048729,
+      "learning_rate": 0.0001,
+      "loss": 1.6637,
+      "step": 21400
+    },
+    {
+      "epoch": 0.29505220154335,
+      "grad_norm": 0.09921707957983017,
+      "learning_rate": 0.0001,
+      "loss": 1.6636,
+      "step": 21450
+    },
+    {
+      "epoch": 0.2957399689129149,
+      "grad_norm": 0.08495631068944931,
+      "learning_rate": 0.0001,
+      "loss": 1.6628,
+      "step": 21500
+    },
+    {
+      "epoch": 0.2964277362824798,
+      "grad_norm": 0.09217420965433121,
+      "learning_rate": 0.0001,
+      "loss": 1.6599,
+      "step": 21550
+    },
+    {
+      "epoch": 0.29711550365204475,
+      "grad_norm": 0.08740544319152832,
+      "learning_rate": 0.0001,
+      "loss": 1.6563,
+      "step": 21600
+    },
+    {
+      "epoch": 0.29780327102160964,
+      "grad_norm": 0.0943366289138794,
+      "learning_rate": 0.0001,
+      "loss": 1.6638,
+      "step": 21650
+    },
+    {
+      "epoch": 0.2984910383911746,
+      "grad_norm": 0.08775294572114944,
+      "learning_rate": 0.0001,
+      "loss": 1.6624,
+      "step": 21700
+    },
+    {
+      "epoch": 0.29917880576073946,
+      "grad_norm": 0.09602498263120651,
+      "learning_rate": 0.0001,
+      "loss": 1.6616,
+      "step": 21750
+    },
+    {
+      "epoch": 0.2998665731303044,
+      "grad_norm": 0.09389972686767578,
+      "learning_rate": 0.0001,
+      "loss": 1.6592,
+      "step": 21800
+    },
+    {
+      "epoch": 0.30055434049986934,
+      "grad_norm": 0.08765329420566559,
+      "learning_rate": 0.0001,
+      "loss": 1.6638,
+      "step": 21850
+    },
+    {
+      "epoch": 0.30124210786943423,
+      "grad_norm": 0.08868211507797241,
+      "learning_rate": 0.0001,
+      "loss": 1.6577,
+      "step": 21900
+    },
+    {
+      "epoch": 0.30192987523899917,
+      "grad_norm": 0.08885643631219864,
+      "learning_rate": 0.0001,
+      "loss": 1.6607,
+      "step": 21950
+    },
+    {
+      "epoch": 0.30261764260856405,
+      "grad_norm": 0.090006984770298,
+      "learning_rate": 0.0001,
+      "loss": 1.6646,
+      "step": 22000
+    },
+    {
+      "epoch": 0.303305409978129,
+      "grad_norm": 0.09837059676647186,
+      "learning_rate": 0.0001,
+      "loss": 1.662,
+      "step": 22050
+    },
+    {
+      "epoch": 0.30399317734769393,
+      "grad_norm": 0.09263578057289124,
+      "learning_rate": 0.0001,
+      "loss": 1.659,
+      "step": 22100
+    },
+    {
+      "epoch": 0.3046809447172588,
+      "grad_norm": 0.10446369647979736,
+      "learning_rate": 0.0001,
+      "loss": 1.6585,
+      "step": 22150
+    },
+    {
+      "epoch": 0.30536871208682376,
+      "grad_norm": 0.0927567332983017,
+      "learning_rate": 0.0001,
+      "loss": 1.6621,
+      "step": 22200
+    },
+    {
+      "epoch": 0.3060564794563887,
+      "grad_norm": 0.08797762542963028,
+      "learning_rate": 0.0001,
+      "loss": 1.6603,
+      "step": 22250
+    },
+    {
+      "epoch": 0.3067442468259536,
+      "grad_norm": 0.09671605378389359,
+      "learning_rate": 0.0001,
+      "loss": 1.6595,
+      "step": 22300
+    },
+    {
+      "epoch": 0.3074320141955185,
+      "grad_norm": 0.08927121758460999,
+      "learning_rate": 0.0001,
+      "loss": 1.6605,
+      "step": 22350
+    },
+    {
+      "epoch": 0.3081197815650834,
+      "grad_norm": 0.09074302762746811,
+      "learning_rate": 0.0001,
+      "loss": 1.6578,
+      "step": 22400
+    },
+    {
+      "epoch": 0.30880754893464835,
+      "grad_norm": 0.07775942236185074,
+      "learning_rate": 0.0001,
+      "loss": 1.6588,
+      "step": 22450
+    },
+    {
+      "epoch": 0.3094953163042133,
+      "grad_norm": 0.08967916667461395,
+      "learning_rate": 0.0001,
+      "loss": 1.6594,
+      "step": 22500
+    },
+    {
+      "epoch": 0.31018308367377817,
+      "grad_norm": 0.09383999556303024,
+      "learning_rate": 0.0001,
+      "loss": 1.6567,
+      "step": 22550
+    },
+    {
+      "epoch": 0.3108708510433431,
+      "grad_norm": 0.08825831860303879,
+      "learning_rate": 0.0001,
+      "loss": 1.6604,
+      "step": 22600
+    },
+    {
+      "epoch": 0.311558618412908,
+      "grad_norm": 0.08609315752983093,
+      "learning_rate": 0.0001,
+      "loss": 1.6594,
+      "step": 22650
+    },
+    {
+      "epoch": 0.31224638578247293,
+      "grad_norm": 0.09169257432222366,
+      "learning_rate": 0.0001,
+      "loss": 1.6617,
+      "step": 22700
+    },
+    {
+      "epoch": 0.3129341531520379,
+      "grad_norm": 0.08700941503047943,
+      "learning_rate": 0.0001,
+      "loss": 1.6602,
+      "step": 22750
+    },
+    {
+      "epoch": 0.31362192052160276,
+      "grad_norm": 0.09563438594341278,
+      "learning_rate": 0.0001,
+      "loss": 1.6584,
+      "step": 22800
+    },
+    {
+      "epoch": 0.3143096878911677,
+      "grad_norm": 0.08614029735326767,
+      "learning_rate": 0.0001,
+      "loss": 1.6602,
+      "step": 22850
+    },
+    {
+      "epoch": 0.3149974552607326,
+      "grad_norm": 0.09383559972047806,
+      "learning_rate": 0.0001,
+      "loss": 1.6598,
+      "step": 22900
+    },
+    {
+      "epoch": 0.3156852226302975,
+      "grad_norm": 0.08948863297700882,
+      "learning_rate": 0.0001,
+      "loss": 1.6594,
+      "step": 22950
+    },
+    {
+      "epoch": 0.31637298999986246,
+      "grad_norm": 0.09441053867340088,
+      "learning_rate": 0.0001,
+      "loss": 1.6569,
+      "step": 23000
+    },
+    {
+      "epoch": 0.31706075736942735,
+      "grad_norm": 0.10128844529390335,
+      "learning_rate": 0.0001,
+      "loss": 1.6586,
+      "step": 23050
+    },
+    {
+      "epoch": 0.3177485247389923,
+      "grad_norm": 0.09148503839969635,
+      "learning_rate": 0.0001,
+      "loss": 1.6543,
+      "step": 23100
+    },
+    {
+      "epoch": 0.31843629210855723,
+      "grad_norm": 0.09208353608846664,
+      "learning_rate": 0.0001,
+      "loss": 1.658,
+      "step": 23150
+    },
+    {
+      "epoch": 0.3191240594781221,
+      "grad_norm": 0.09142562001943588,
+      "learning_rate": 0.0001,
+      "loss": 1.6577,
+      "step": 23200
+    },
+    {
+      "epoch": 0.31981182684768705,
+      "grad_norm": 0.0846257209777832,
+      "learning_rate": 0.0001,
+      "loss": 1.6552,
+      "step": 23250
+    },
+    {
+      "epoch": 0.32049959421725194,
+      "grad_norm": 0.09171921014785767,
+      "learning_rate": 0.0001,
+      "loss": 1.6551,
+      "step": 23300
+    },
+    {
+      "epoch": 0.3211873615868169,
+      "grad_norm": 0.0841546282172203,
+      "learning_rate": 0.0001,
+      "loss": 1.6578,
+      "step": 23350
+    },
+    {
+      "epoch": 0.3218751289563818,
+      "grad_norm": 0.09704132378101349,
+      "learning_rate": 0.0001,
+      "loss": 1.6595,
+      "step": 23400
+    },
+    {
+      "epoch": 0.3225628963259467,
+      "grad_norm": 0.08762117475271225,
+      "learning_rate": 0.0001,
+      "loss": 1.6589,
+      "step": 23450
+    },
+    {
+      "epoch": 0.32325066369551164,
+      "grad_norm": 0.0907997116446495,
+      "learning_rate": 0.0001,
+      "loss": 1.6586,
+      "step": 23500
+    },
+    {
+      "epoch": 0.3239384310650765,
+      "grad_norm": 0.09183737635612488,
+      "learning_rate": 0.0001,
+      "loss": 1.6564,
+      "step": 23550
+    },
+    {
+      "epoch": 0.32462619843464147,
+      "grad_norm": 0.08676784485578537,
+      "learning_rate": 0.0001,
+      "loss": 1.6568,
+      "step": 23600
+    },
+    {
+      "epoch": 0.3253139658042064,
+      "grad_norm": 0.09329142421483994,
+      "learning_rate": 0.0001,
+      "loss": 1.6558,
+      "step": 23650
+    },
+    {
+      "epoch": 0.3260017331737713,
+      "grad_norm": 0.09180528670549393,
+      "learning_rate": 0.0001,
+      "loss": 1.6562,
+      "step": 23700
+    },
+    {
+      "epoch": 0.32668950054333623,
+      "grad_norm": 0.09366358071565628,
+      "learning_rate": 0.0001,
+      "loss": 1.656,
+      "step": 23750
+    },
+    {
+      "epoch": 0.3273772679129011,
+      "grad_norm": 0.08830198645591736,
+      "learning_rate": 0.0001,
+      "loss": 1.6594,
+      "step": 23800
+    },
+    {
+      "epoch": 0.32806503528246606,
+      "grad_norm": 0.0980730950832367,
+      "learning_rate": 0.0001,
+      "loss": 1.6575,
+      "step": 23850
+    },
+    {
+      "epoch": 0.328752802652031,
+      "grad_norm": 0.08481217920780182,
+      "learning_rate": 0.0001,
+      "loss": 1.6572,
+      "step": 23900
+    },
+    {
+      "epoch": 0.3294405700215959,
+      "grad_norm": 0.10097907483577728,
+      "learning_rate": 0.0001,
+      "loss": 1.6559,
+      "step": 23950
+    },
+    {
+      "epoch": 0.3301283373911608,
+      "grad_norm": 0.09167522937059402,
+      "learning_rate": 0.0001,
+      "loss": 1.6561,
+      "step": 24000
+    },
+    {
+      "epoch": 0.33081610476072576,
+      "grad_norm": 0.0857388898730278,
+      "learning_rate": 0.0001,
+      "loss": 1.6554,
+      "step": 24050
+    },
+    {
+      "epoch": 0.33150387213029064,
+      "grad_norm": 0.08629937469959259,
+      "learning_rate": 0.0001,
+      "loss": 1.6598,
+      "step": 24100
+    },
+    {
+      "epoch": 0.3321916394998556,
+      "grad_norm": 0.09269680827856064,
+      "learning_rate": 0.0001,
+      "loss": 1.6553,
+      "step": 24150
+    },
+    {
+      "epoch": 0.33287940686942047,
+      "grad_norm": 0.08871278166770935,
+      "learning_rate": 0.0001,
+      "loss": 1.655,
+      "step": 24200
+    },
+    {
+      "epoch": 0.3335671742389854,
+      "grad_norm": 0.08813636004924774,
+      "learning_rate": 0.0001,
+      "loss": 1.6537,
+      "step": 24250
+    },
+    {
+      "epoch": 0.33425494160855035,
+      "grad_norm": 0.09300006926059723,
+      "learning_rate": 0.0001,
+      "loss": 1.6569,
+      "step": 24300
+    },
+    {
+      "epoch": 0.33494270897811523,
+      "grad_norm": 0.09674695879220963,
+      "learning_rate": 0.0001,
+      "loss": 1.658,
+      "step": 24350
+    },
+    {
+      "epoch": 0.3356304763476802,
+      "grad_norm": 0.08643494546413422,
+      "learning_rate": 0.0001,
+      "loss": 1.6557,
+      "step": 24400
+    },
+    {
+      "epoch": 0.33631824371724506,
+      "grad_norm": 0.09858324378728867,
+      "learning_rate": 0.0001,
+      "loss": 1.6592,
+      "step": 24450
+    },
+    {
+      "epoch": 0.33700601108681,
+      "grad_norm": 0.09983374178409576,
+      "learning_rate": 0.0001,
+      "loss": 1.6529,
+      "step": 24500
+    },
+    {
+      "epoch": 0.33769377845637494,
+      "grad_norm": 0.08795691281557083,
+      "learning_rate": 0.0001,
+      "loss": 1.6549,
+      "step": 24550
+    },
+    {
+      "epoch": 0.3383815458259398,
+      "grad_norm": 0.09592121094465256,
+      "learning_rate": 0.0001,
+      "loss": 1.6584,
+      "step": 24600
+    },
+    {
+      "epoch": 0.33906931319550476,
+      "grad_norm": 0.10493771731853485,
+      "learning_rate": 0.0001,
+      "loss": 1.654,
+      "step": 24650
+    },
+    {
+      "epoch": 0.33975708056506965,
+      "grad_norm": 0.09947793930768967,
+      "learning_rate": 0.0001,
+      "loss": 1.6553,
+      "step": 24700
+    },
+    {
+      "epoch": 0.3404448479346346,
+      "grad_norm": 0.08845814317464828,
+      "learning_rate": 0.0001,
+      "loss": 1.6544,
+      "step": 24750
+    },
+    {
+      "epoch": 0.3411326153041995,
+      "grad_norm": 0.09179400652647018,
+      "learning_rate": 0.0001,
+      "loss": 1.6559,
+      "step": 24800
+    },
+    {
+      "epoch": 0.3418203826737644,
+      "grad_norm": 0.09495829790830612,
+      "learning_rate": 0.0001,
+      "loss": 1.6566,
+      "step": 24850
+    },
+    {
+      "epoch": 0.34250815004332935,
+      "grad_norm": 0.08588258177042007,
+      "learning_rate": 0.0001,
+      "loss": 1.6566,
+      "step": 24900
+    },
+    {
+      "epoch": 0.34319591741289424,
+      "grad_norm": 0.094237320125103,
+      "learning_rate": 0.0001,
+      "loss": 1.6571,
+      "step": 24950
+    },
+    {
+      "epoch": 0.3438836847824592,
+      "grad_norm": 0.09663517773151398,
+      "learning_rate": 0.0001,
+      "loss": 1.6563,
+      "step": 25000
+    },
+    {
+      "epoch": 0.3445714521520241,
+      "grad_norm": 0.08958858996629715,
+      "learning_rate": 0.0001,
+      "loss": 1.6518,
+      "step": 25050
+    },
+    {
+      "epoch": 0.345259219521589,
+      "grad_norm": 0.09064103662967682,
+      "learning_rate": 0.0001,
+      "loss": 1.6542,
+      "step": 25100
+    },
+    {
+      "epoch": 0.34594698689115394,
+      "grad_norm": 0.09530662000179291,
+      "learning_rate": 0.0001,
+      "loss": 1.6561,
+      "step": 25150
+    },
+    {
+      "epoch": 0.3466347542607189,
+      "grad_norm": 0.10598750412464142,
+      "learning_rate": 0.0001,
+      "loss": 1.6536,
+      "step": 25200
+    },
+    {
+      "epoch": 0.34732252163028376,
+      "grad_norm": 0.09103056788444519,
+      "learning_rate": 0.0001,
+      "loss": 1.6571,
+      "step": 25250
+    },
+    {
+      "epoch": 0.3480102889998487,
+      "grad_norm": 0.08924443274736404,
+      "learning_rate": 0.0001,
+      "loss": 1.654,
+      "step": 25300
+    },
+    {
+      "epoch": 0.3486980563694136,
+      "grad_norm": 0.0930083692073822,
+      "learning_rate": 0.0001,
+      "loss": 1.6557,
+      "step": 25350
+    },
+    {
+      "epoch": 0.34938582373897853,
+      "grad_norm": 0.09939617663621902,
+      "learning_rate": 0.0001,
+      "loss": 1.6512,
+      "step": 25400
+    },
+    {
+      "epoch": 0.35007359110854347,
+      "grad_norm": 0.1029135212302208,
+      "learning_rate": 0.0001,
+      "loss": 1.6546,
+      "step": 25450
+    },
+    {
+      "epoch": 0.35076135847810835,
+      "grad_norm": 0.0853937491774559,
+      "learning_rate": 0.0001,
+      "loss": 1.6583,
+      "step": 25500
+    },
+    {
+      "epoch": 0.3514491258476733,
+      "grad_norm": 0.09008786827325821,
+      "learning_rate": 0.0001,
+      "loss": 1.6523,
+      "step": 25550
+    },
+    {
+      "epoch": 0.3521368932172382,
+      "grad_norm": 0.10317879915237427,
+      "learning_rate": 0.0001,
+      "loss": 1.6595,
+      "step": 25600
+    },
+    {
+      "epoch": 0.3528246605868031,
+      "grad_norm": 0.08662793040275574,
+      "learning_rate": 0.0001,
+      "loss": 1.6554,
+      "step": 25650
+    },
+    {
+      "epoch": 0.35351242795636806,
+      "grad_norm": 0.09117420762777328,
+      "learning_rate": 0.0001,
+      "loss": 1.6571,
+      "step": 25700
+    },
+    {
+      "epoch": 0.35420019532593294,
+      "grad_norm": 0.08435691148042679,
+      "learning_rate": 0.0001,
+      "loss": 1.6567,
+      "step": 25750
+    },
+    {
+      "epoch": 0.3548879626954979,
+      "grad_norm": 0.0903763696551323,
+      "learning_rate": 0.0001,
+      "loss": 1.6525,
+      "step": 25800
+    },
+    {
+      "epoch": 0.35557573006506277,
+      "grad_norm": 0.0919974073767662,
+      "learning_rate": 0.0001,
+      "loss": 1.6541,
+      "step": 25850
+    },
+    {
+      "epoch": 0.3562634974346277,
+      "grad_norm": 0.08953440189361572,
+      "learning_rate": 0.0001,
+      "loss": 1.6577,
+      "step": 25900
+    },
+    {
+      "epoch": 0.35695126480419265,
+      "grad_norm": 0.08938378095626831,
+      "learning_rate": 0.0001,
+      "loss": 1.6526,
+      "step": 25950
+    },
+    {
+      "epoch": 0.35763903217375753,
+      "grad_norm": 0.09824621677398682,
+      "learning_rate": 0.0001,
+      "loss": 1.6531,
+      "step": 26000
+    },
+    {
+      "epoch": 0.35832679954332247,
+      "grad_norm": 0.09804233908653259,
+      "learning_rate": 0.0001,
+      "loss": 1.6567,
+      "step": 26050
+    },
+    {
+      "epoch": 0.3590145669128874,
+      "grad_norm": 0.1049811840057373,
+      "learning_rate": 0.0001,
+      "loss": 1.6523,
+      "step": 26100
+    },
+    {
+      "epoch": 0.3597023342824523,
+      "grad_norm": 0.09391681849956512,
+      "learning_rate": 0.0001,
+      "loss": 1.6552,
+      "step": 26150
+    },
+    {
+      "epoch": 0.36039010165201724,
+      "grad_norm": 0.10802886635065079,
+      "learning_rate": 0.0001,
+      "loss": 1.6525,
+      "step": 26200
+    },
+    {
+      "epoch": 0.3610778690215821,
+      "grad_norm": 0.08910466730594635,
+      "learning_rate": 0.0001,
+      "loss": 1.6551,
+      "step": 26250
+    },
+    {
+      "epoch": 0.36176563639114706,
+      "grad_norm": 0.08862115442752838,
+      "learning_rate": 0.0001,
+      "loss": 1.6538,
+      "step": 26300
+    },
+    {
+      "epoch": 0.362453403760712,
+      "grad_norm": 0.08569826930761337,
+      "learning_rate": 0.0001,
+      "loss": 1.6521,
+      "step": 26350
+    },
+    {
+      "epoch": 0.3631411711302769,
+      "grad_norm": 0.09084028750658035,
+      "learning_rate": 0.0001,
+      "loss": 1.6531,
+      "step": 26400
+    },
+    {
+      "epoch": 0.3638289384998418,
+      "grad_norm": 0.08381815999746323,
+      "learning_rate": 0.0001,
+      "loss": 1.6569,
+      "step": 26450
+    },
+    {
+      "epoch": 0.3645167058694067,
+      "grad_norm": 0.08396822959184647,
+      "learning_rate": 0.0001,
+      "loss": 1.6583,
+      "step": 26500
+    },
+    {
+      "epoch": 0.36520447323897165,
+      "grad_norm": 0.0848502367734909,
+      "learning_rate": 0.0001,
+      "loss": 1.6572,
+      "step": 26550
+    },
+    {
+      "epoch": 0.3658922406085366,
+      "grad_norm": 0.09545731544494629,
+      "learning_rate": 0.0001,
+      "loss": 1.6577,
+      "step": 26600
+    },
+    {
+      "epoch": 0.3665800079781015,
+      "grad_norm": 0.09016140550374985,
+      "learning_rate": 0.0001,
+      "loss": 1.6554,
+      "step": 26650
+    },
+    {
+      "epoch": 0.3672677753476664,
+      "grad_norm": 0.08480773121118546,
+      "learning_rate": 0.0001,
+      "loss": 1.6552,
+      "step": 26700
+    },
+    {
+      "epoch": 0.3679555427172313,
+      "grad_norm": 0.08351889252662659,
+      "learning_rate": 0.0001,
+      "loss": 1.6534,
+      "step": 26750
+    },
+    {
+      "epoch": 0.36864331008679624,
+      "grad_norm": 0.08628764003515244,
+      "learning_rate": 0.0001,
+      "loss": 1.6527,
+      "step": 26800
+    },
+    {
+      "epoch": 0.3693310774563612,
+      "grad_norm": 0.09489446878433228,
+      "learning_rate": 0.0001,
+      "loss": 1.6475,
+      "step": 26850
+    },
+    {
+      "epoch": 0.37001884482592606,
+      "grad_norm": 0.09590397030115128,
+      "learning_rate": 0.0001,
+      "loss": 1.6544,
+      "step": 26900
+    },
+    {
+      "epoch": 0.370706612195491,
+      "grad_norm": 0.08726419508457184,
+      "learning_rate": 0.0001,
+      "loss": 1.6553,
+      "step": 26950
+    },
+    {
+      "epoch": 0.37139437956505594,
+      "grad_norm": 0.09084060788154602,
+      "learning_rate": 0.0001,
+      "loss": 1.6477,
+      "step": 27000
+    },
+    {
+      "epoch": 0.3720821469346208,
+      "grad_norm": 0.08947817236185074,
+      "learning_rate": 0.0001,
+      "loss": 1.6518,
+      "step": 27050
+    },
+    {
+      "epoch": 0.37276991430418577,
+      "grad_norm": 0.0947887971997261,
+      "learning_rate": 0.0001,
+      "loss": 1.6516,
+      "step": 27100
+    },
+    {
+      "epoch": 0.37345768167375065,
+      "grad_norm": 0.09059783816337585,
+      "learning_rate": 0.0001,
+      "loss": 1.6491,
+      "step": 27150
+    },
+    {
+      "epoch": 0.3741454490433156,
+      "grad_norm": 0.09635411202907562,
+      "learning_rate": 0.0001,
+      "loss": 1.6518,
+      "step": 27200
+    },
+    {
+      "epoch": 0.37483321641288053,
+      "grad_norm": 0.09230557829141617,
+      "learning_rate": 0.0001,
+      "loss": 1.6513,
+      "step": 27250
+    },
+    {
+      "epoch": 0.3755209837824454,
+      "grad_norm": 0.09676863998174667,
+      "learning_rate": 0.0001,
+      "loss": 1.653,
+      "step": 27300
+    },
+    {
+      "epoch": 0.37620875115201036,
+      "grad_norm": 0.08972521126270294,
+      "learning_rate": 0.0001,
+      "loss": 1.6556,
+      "step": 27350
+    },
+    {
+      "epoch": 0.37689651852157524,
+      "grad_norm": 0.09976805746555328,
+      "learning_rate": 0.0001,
+      "loss": 1.6509,
+      "step": 27400
+    },
+    {
+      "epoch": 0.3775842858911402,
+      "grad_norm": 0.08917619287967682,
+      "learning_rate": 0.0001,
+      "loss": 1.649,
+      "step": 27450
+    },
+    {
+      "epoch": 0.3782720532607051,
+      "grad_norm": 0.09673383086919785,
+      "learning_rate": 0.0001,
+      "loss": 1.6485,
+      "step": 27500
+    },
+    {
+      "epoch": 0.37895982063027,
+      "grad_norm": 0.0897962898015976,
+      "learning_rate": 0.0001,
+      "loss": 1.6533,
+      "step": 27550
+    },
+    {
+      "epoch": 0.37964758799983495,
+      "grad_norm": 0.10002953559160233,
+      "learning_rate": 0.0001,
+      "loss": 1.6526,
+      "step": 27600
+    },
+    {
+      "epoch": 0.38033535536939983,
+      "grad_norm": 0.10410989075899124,
+      "learning_rate": 0.0001,
+      "loss": 1.6523,
+      "step": 27650
+    },
+    {
+      "epoch": 0.38102312273896477,
+      "grad_norm": 0.09563115984201431,
+      "learning_rate": 0.0001,
+      "loss": 1.6519,
+      "step": 27700
+    },
+    {
+      "epoch": 0.3817108901085297,
+      "grad_norm": 0.08847146481275558,
+      "learning_rate": 0.0001,
+      "loss": 1.6531,
+      "step": 27750
+    },
+    {
+      "epoch": 0.3823986574780946,
+      "grad_norm": 0.088249571621418,
+      "learning_rate": 0.0001,
+      "loss": 1.6499,
+      "step": 27800
+    },
+    {
+      "epoch": 0.38308642484765953,
+      "grad_norm": 0.09289827942848206,
+      "learning_rate": 0.0001,
+      "loss": 1.6543,
+      "step": 27850
+    },
+    {
+      "epoch": 0.3837741922172244,
+      "grad_norm": 0.08918619155883789,
+      "learning_rate": 0.0001,
+      "loss": 1.6537,
+      "step": 27900
+    },
+    {
+      "epoch": 0.38446195958678936,
+      "grad_norm": 0.08992282301187515,
+      "learning_rate": 0.0001,
+      "loss": 1.6516,
+      "step": 27950
+    },
+    {
+      "epoch": 0.3851497269563543,
+      "grad_norm": 0.09168649464845657,
+      "learning_rate": 0.0001,
+      "loss": 1.6516,
+      "step": 28000
+    },
+    {
+      "epoch": 0.3858374943259192,
+      "grad_norm": 0.08888909220695496,
+      "learning_rate": 0.0001,
+      "loss": 1.6518,
+      "step": 28050
+    },
+    {
+      "epoch": 0.3865252616954841,
+      "grad_norm": 0.098940409719944,
+      "learning_rate": 0.0001,
+      "loss": 1.65,
+      "step": 28100
+    },
+    {
+      "epoch": 0.38721302906504906,
+      "grad_norm": 0.08919871598482132,
+      "learning_rate": 0.0001,
+      "loss": 1.6485,
+      "step": 28150
+    },
+    {
+      "epoch": 0.38790079643461395,
+      "grad_norm": 0.09120476245880127,
+      "learning_rate": 0.0001,
+      "loss": 1.6521,
+      "step": 28200
+    },
+    {
+      "epoch": 0.3885885638041789,
+      "grad_norm": 0.09309836477041245,
+      "learning_rate": 0.0001,
+      "loss": 1.6518,
+      "step": 28250
+    },
+    {
+      "epoch": 0.3892763311737438,
+      "grad_norm": 0.0869157537817955,
+      "learning_rate": 0.0001,
+      "loss": 1.6486,
+      "step": 28300
+    },
+    {
+      "epoch": 0.3899640985433087,
+      "grad_norm": 0.09697956591844559,
+      "learning_rate": 0.0001,
+      "loss": 1.6534,
+      "step": 28350
+    },
+    {
+      "epoch": 0.39065186591287365,
+      "grad_norm": 0.09053334593772888,
+      "learning_rate": 0.0001,
+      "loss": 1.6513,
+      "step": 28400
+    },
+    {
+      "epoch": 0.39133963328243854,
+      "grad_norm": 0.09129533171653748,
+      "learning_rate": 0.0001,
+      "loss": 1.6558,
+      "step": 28450
+    },
+    {
+      "epoch": 0.3920274006520035,
+      "grad_norm": 0.09009382128715515,
+      "learning_rate": 0.0001,
+      "loss": 1.6529,
+      "step": 28500
+    },
+    {
+      "epoch": 0.39271516802156836,
+      "grad_norm": 0.10035433620214462,
+      "learning_rate": 0.0001,
+      "loss": 1.6521,
+      "step": 28550
+    },
+    {
+      "epoch": 0.3934029353911333,
+      "grad_norm": 0.09056541323661804,
+      "learning_rate": 0.0001,
+      "loss": 1.6532,
+      "step": 28600
+    },
+    {
+      "epoch": 0.39409070276069824,
+      "grad_norm": 0.0904608890414238,
+      "learning_rate": 0.0001,
+      "loss": 1.6531,
+      "step": 28650
+    },
+    {
+      "epoch": 0.3947784701302631,
+      "grad_norm": 0.11022058129310608,
+      "learning_rate": 0.0001,
+      "loss": 1.6473,
+      "step": 28700
+    },
+    {
+      "epoch": 0.39546623749982807,
+      "grad_norm": 0.0866251140832901,
+      "learning_rate": 0.0001,
+      "loss": 1.6525,
+      "step": 28750
+    },
+    {
+      "epoch": 0.39615400486939295,
+      "grad_norm": 0.09494468569755554,
+      "learning_rate": 0.0001,
+      "loss": 1.6529,
+      "step": 28800
+    },
+    {
+      "epoch": 0.3968417722389579,
+      "grad_norm": 0.09128791838884354,
+      "learning_rate": 0.0001,
+      "loss": 1.6518,
+      "step": 28850
+    },
+    {
+      "epoch": 0.39752953960852283,
+      "grad_norm": 0.09931056201457977,
+      "learning_rate": 0.0001,
+      "loss": 1.6494,
+      "step": 28900
+    },
+    {
+      "epoch": 0.3982173069780877,
+      "grad_norm": 0.0899323970079422,
+      "learning_rate": 0.0001,
+      "loss": 1.648,
+      "step": 28950
+    },
+    {
+      "epoch": 0.39890507434765266,
+      "grad_norm": 0.09281651675701141,
+      "learning_rate": 0.0001,
+      "loss": 1.6461,
+      "step": 29000
+    },
+    {
+      "epoch": 0.3995928417172176,
+      "grad_norm": 0.09721958637237549,
+      "learning_rate": 0.0001,
+      "loss": 1.6495,
+      "step": 29050
+    },
+    {
+      "epoch": 0.4002806090867825,
+      "grad_norm": 0.09711379557847977,
+      "learning_rate": 0.0001,
+      "loss": 1.6473,
+      "step": 29100
+    },
+    {
+      "epoch": 0.4009683764563474,
+      "grad_norm": 0.09055937081575394,
+      "learning_rate": 0.0001,
+      "loss": 1.6505,
+      "step": 29150
+    },
+    {
+      "epoch": 0.4016561438259123,
+      "grad_norm": 0.09126099944114685,
+      "learning_rate": 0.0001,
+      "loss": 1.652,
+      "step": 29200
+    },
+    {
+      "epoch": 0.40234391119547724,
+      "grad_norm": 0.09342388063669205,
+      "learning_rate": 0.0001,
+      "loss": 1.6476,
+      "step": 29250
+    },
+    {
+      "epoch": 0.4030316785650422,
+      "grad_norm": 0.09094425290822983,
+      "learning_rate": 0.0001,
+      "loss": 1.6499,
+      "step": 29300
+    },
+    {
+      "epoch": 0.40371944593460707,
+      "grad_norm": 0.09739495813846588,
+      "learning_rate": 0.0001,
+      "loss": 1.6537,
+      "step": 29350
+    },
+    {
+      "epoch": 0.404407213304172,
+      "grad_norm": 0.08720117807388306,
+      "learning_rate": 0.0001,
+      "loss": 1.6501,
+      "step": 29400
+    },
+    {
+      "epoch": 0.4050949806737369,
+      "grad_norm": 0.10012072324752808,
+      "learning_rate": 0.0001,
+      "loss": 1.6486,
+      "step": 29450
+    },
+    {
+      "epoch": 0.40578274804330183,
+      "grad_norm": 0.09751909226179123,
+      "learning_rate": 0.0001,
+      "loss": 1.6503,
+      "step": 29500
+    },
+    {
+      "epoch": 0.4064705154128668,
+      "grad_norm": 0.0909474641084671,
+      "learning_rate": 0.0001,
+      "loss": 1.6487,
+      "step": 29550
+    },
+    {
+      "epoch": 0.40715828278243166,
+      "grad_norm": 0.09806191921234131,
+      "learning_rate": 0.0001,
+      "loss": 1.6476,
+      "step": 29600
+    },
+    {
+      "epoch": 0.4078460501519966,
+      "grad_norm": 0.08882980048656464,
+      "learning_rate": 0.0001,
+      "loss": 1.6461,
+      "step": 29650
+    },
+    {
+      "epoch": 0.4085338175215615,
+      "grad_norm": 0.08612953126430511,
+      "learning_rate": 0.0001,
+      "loss": 1.6517,
+      "step": 29700
+    },
+    {
+      "epoch": 0.4092215848911264,
+      "grad_norm": 0.09713184833526611,
+      "learning_rate": 0.0001,
+      "loss": 1.6505,
+      "step": 29750
+    },
+    {
+      "epoch": 0.40990935226069136,
+      "grad_norm": 0.08773653209209442,
+      "learning_rate": 0.0001,
+      "loss": 1.6513,
+      "step": 29800
+    },
+    {
+      "epoch": 0.41059711963025625,
+      "grad_norm": 0.09357753396034241,
+      "learning_rate": 0.0001,
+      "loss": 1.6512,
+      "step": 29850
+    },
+    {
+      "epoch": 0.4112848869998212,
+      "grad_norm": 0.10132279247045517,
+      "learning_rate": 0.0001,
+      "loss": 1.6499,
+      "step": 29900
+    },
+    {
+      "epoch": 0.4119726543693861,
+      "grad_norm": 0.09292474389076233,
+      "learning_rate": 0.0001,
+      "loss": 1.649,
+      "step": 29950
+    },
+    {
+      "epoch": 0.412660421738951,
+      "grad_norm": 0.09160569310188293,
+      "learning_rate": 0.0001,
+      "loss": 1.6472,
+      "step": 30000
+    },
+    {
+      "epoch": 0.41334818910851595,
+      "grad_norm": 0.09838960319757462,
+      "learning_rate": 0.0001,
+      "loss": 1.6498,
+      "step": 30050
+    },
+    {
+      "epoch": 0.41403595647808084,
+      "grad_norm": 0.0884481742978096,
+      "learning_rate": 0.0001,
+      "loss": 1.6498,
+      "step": 30100
+    },
+    {
+      "epoch": 0.4147237238476458,
+      "grad_norm": 0.09937591850757599,
+      "learning_rate": 0.0001,
+      "loss": 1.6516,
+      "step": 30150
+    },
+    {
+      "epoch": 0.4154114912172107,
+      "grad_norm": 0.08566869050264359,
+      "learning_rate": 0.0001,
+      "loss": 1.6495,
+      "step": 30200
+    },
+    {
+      "epoch": 0.4160992585867756,
+      "grad_norm": 0.09897395223379135,
+      "learning_rate": 0.0001,
+      "loss": 1.6427,
+      "step": 30250
+    },
+    {
+      "epoch": 0.41678702595634054,
+      "grad_norm": 0.10065190494060516,
+      "learning_rate": 0.0001,
+      "loss": 1.6478,
+      "step": 30300
+    },
+    {
+      "epoch": 0.4174747933259054,
+      "grad_norm": 0.09039245545864105,
+      "learning_rate": 0.0001,
+      "loss": 1.6511,
+      "step": 30350
+    },
+    {
+      "epoch": 0.41816256069547036,
+      "grad_norm": 0.08918923139572144,
+      "learning_rate": 0.0001,
+      "loss": 1.6454,
+      "step": 30400
+    },
+    {
+      "epoch": 0.4188503280650353,
+      "grad_norm": 0.08621367067098618,
+      "learning_rate": 0.0001,
+      "loss": 1.6505,
+      "step": 30450
+    },
+    {
+      "epoch": 0.4195380954346002,
+      "grad_norm": 0.107161745429039,
+      "learning_rate": 0.0001,
+      "loss": 1.6486,
+      "step": 30500
+    },
+    {
+      "epoch": 0.42022586280416513,
+      "grad_norm": 0.1019233912229538,
+      "learning_rate": 0.0001,
+      "loss": 1.6458,
+      "step": 30550
+    },
+    {
+      "epoch": 0.42091363017373,
+      "grad_norm": 0.10087515413761139,
+      "learning_rate": 0.0001,
+      "loss": 1.65,
+      "step": 30600
+    },
+    {
+      "epoch": 0.42160139754329495,
+      "grad_norm": 0.08938942849636078,
+      "learning_rate": 0.0001,
+      "loss": 1.6531,
+      "step": 30650
+    },
+    {
+      "epoch": 0.4222891649128599,
+      "grad_norm": 0.10269896686077118,
+      "learning_rate": 0.0001,
+      "loss": 1.6471,
+      "step": 30700
+    },
+    {
+      "epoch": 0.4229769322824248,
+      "grad_norm": 0.09027698636054993,
+      "learning_rate": 0.0001,
+      "loss": 1.6502,
+      "step": 30750
+    },
+    {
+      "epoch": 0.4236646996519897,
+      "grad_norm": 0.09330408275127411,
+      "learning_rate": 0.0001,
+      "loss": 1.6442,
+      "step": 30800
+    },
+    {
+      "epoch": 0.4243524670215546,
+      "grad_norm": 0.08705689758062363,
+      "learning_rate": 0.0001,
+      "loss": 1.648,
+      "step": 30850
+    },
+    {
+      "epoch": 0.42504023439111954,
+      "grad_norm": 0.09150363504886627,
+      "learning_rate": 0.0001,
+      "loss": 1.6446,
+      "step": 30900
+    },
+    {
+      "epoch": 0.4257280017606845,
+      "grad_norm": 0.09293975681066513,
+      "learning_rate": 0.0001,
+      "loss": 1.6506,
+      "step": 30950
+    },
+    {
+      "epoch": 0.42641576913024937,
+      "grad_norm": 0.10130377858877182,
+      "learning_rate": 0.0001,
+      "loss": 1.6489,
+      "step": 31000
+    },
+    {
+      "epoch": 0.4271035364998143,
+      "grad_norm": 0.08871527761220932,
+      "learning_rate": 0.0001,
+      "loss": 1.6485,
+      "step": 31050
+    },
+    {
+      "epoch": 0.42779130386937925,
+      "grad_norm": 0.0994982048869133,
+      "learning_rate": 0.0001,
+      "loss": 1.6463,
+      "step": 31100
+    },
+    {
+      "epoch": 0.42847907123894413,
+      "grad_norm": 0.09573712199926376,
+      "learning_rate": 0.0001,
+      "loss": 1.6508,
+      "step": 31150
+    },
+    {
+      "epoch": 0.42916683860850907,
+      "grad_norm": 0.10358277708292007,
+      "learning_rate": 0.0001,
+      "loss": 1.6502,
+      "step": 31200
+    },
+    {
+      "epoch": 0.42985460597807396,
+      "grad_norm": 0.0881354883313179,
+      "learning_rate": 0.0001,
+      "loss": 1.6444,
+      "step": 31250
+    },
+    {
+      "epoch": 0.4305423733476389,
+      "grad_norm": 0.08965957164764404,
+      "learning_rate": 0.0001,
+      "loss": 1.6464,
+      "step": 31300
+    },
+    {
+      "epoch": 0.43123014071720384,
+      "grad_norm": 0.09935072064399719,
+      "learning_rate": 0.0001,
+      "loss": 1.6504,
+      "step": 31350
+    },
+    {
+      "epoch": 0.4319179080867687,
+      "grad_norm": 0.09450331330299377,
+      "learning_rate": 0.0001,
+      "loss": 1.6434,
+      "step": 31400
+    },
+    {
+      "epoch": 0.43260567545633366,
+      "grad_norm": 0.09426354616880417,
+      "learning_rate": 0.0001,
+      "loss": 1.6446,
+      "step": 31450
+    },
+    {
+      "epoch": 0.43329344282589854,
+      "grad_norm": 0.09853871166706085,
+      "learning_rate": 0.0001,
+      "loss": 1.6479,
+      "step": 31500
+    },
+    {
+      "epoch": 0.4339812101954635,
+      "grad_norm": 0.09059667587280273,
+      "learning_rate": 0.0001,
+      "loss": 1.6465,
+      "step": 31550
+    },
+    {
+      "epoch": 0.4346689775650284,
+      "grad_norm": 0.10137219727039337,
+      "learning_rate": 0.0001,
+      "loss": 1.6497,
+      "step": 31600
+    },
+    {
+      "epoch": 0.4353567449345933,
+      "grad_norm": 0.09198787808418274,
+      "learning_rate": 0.0001,
+      "loss": 1.6493,
+      "step": 31650
+    },
+    {
+      "epoch": 0.43604451230415825,
+      "grad_norm": 0.1016339510679245,
+      "learning_rate": 0.0001,
+      "loss": 1.646,
+      "step": 31700
+    },
+    {
+      "epoch": 0.43673227967372313,
+      "grad_norm": 0.09335731714963913,
+      "learning_rate": 0.0001,
+      "loss": 1.6477,
+      "step": 31750
+    },
+    {
+      "epoch": 0.4374200470432881,
+      "grad_norm": 0.09104933589696884,
+      "learning_rate": 0.0001,
+      "loss": 1.6461,
+      "step": 31800
+    },
+    {
+      "epoch": 0.438107814412853,
+      "grad_norm": 0.08286421746015549,
+      "learning_rate": 0.0001,
+      "loss": 1.6425,
+      "step": 31850
+    },
+    {
+      "epoch": 0.4387955817824179,
+      "grad_norm": 0.0951671227812767,
+      "learning_rate": 0.0001,
+      "loss": 1.647,
+      "step": 31900
+    },
+    {
+      "epoch": 0.43948334915198284,
+      "grad_norm": 0.08751095086336136,
+      "learning_rate": 0.0001,
+      "loss": 1.6447,
+      "step": 31950
+    },
+    {
+      "epoch": 0.4401711165215478,
+      "grad_norm": 0.09626565128564835,
+      "learning_rate": 0.0001,
+      "loss": 1.6487,
+      "step": 32000
+    },
+    {
+      "epoch": 0.44085888389111266,
+      "grad_norm": 0.09458938241004944,
+      "learning_rate": 0.0001,
+      "loss": 1.6457,
+      "step": 32050
+    },
+    {
+      "epoch": 0.4415466512606776,
+      "grad_norm": 0.1030418872833252,
+      "learning_rate": 0.0001,
+      "loss": 1.6477,
+      "step": 32100
+    },
+    {
+      "epoch": 0.4422344186302425,
+      "grad_norm": 0.09569265693426132,
+      "learning_rate": 0.0001,
+      "loss": 1.6476,
+      "step": 32150
+    },
+    {
+      "epoch": 0.4429221859998074,
+      "grad_norm": 0.10030734539031982,
+      "learning_rate": 0.0001,
+      "loss": 1.6454,
+      "step": 32200
+    },
+    {
+      "epoch": 0.44360995336937237,
+      "grad_norm": 0.11456091701984406,
+      "learning_rate": 0.0001,
+      "loss": 1.6457,
+      "step": 32250
+    },
+    {
+      "epoch": 0.44429772073893725,
+      "grad_norm": 0.08938100934028625,
+      "learning_rate": 0.0001,
+      "loss": 1.6452,
+      "step": 32300
+    },
+    {
+      "epoch": 0.4449854881085022,
+      "grad_norm": 0.10480993241071701,
+      "learning_rate": 0.0001,
+      "loss": 1.6416,
+      "step": 32350
+    },
+    {
+      "epoch": 0.4456732554780671,
+      "grad_norm": 0.10453122854232788,
+      "learning_rate": 0.0001,
+      "loss": 1.6483,
+      "step": 32400
+    },
+    {
+      "epoch": 0.446361022847632,
+      "grad_norm": 0.10204358398914337,
+      "learning_rate": 0.0001,
+      "loss": 1.6457,
+      "step": 32450
+    },
+    {
+      "epoch": 0.44704879021719696,
+      "grad_norm": 0.0900953933596611,
+      "learning_rate": 0.0001,
+      "loss": 1.6479,
+      "step": 32500
+    },
+    {
+      "epoch": 0.44773655758676184,
+      "grad_norm": 0.10755446553230286,
+      "learning_rate": 0.0001,
+      "loss": 1.6431,
+      "step": 32550
+    },
+    {
+      "epoch": 0.4484243249563268,
+      "grad_norm": 0.08640281856060028,
+      "learning_rate": 0.0001,
+      "loss": 1.6458,
+      "step": 32600
+    },
+    {
+      "epoch": 0.44911209232589167,
+      "grad_norm": 0.09299297630786896,
+      "learning_rate": 0.0001,
+      "loss": 1.6465,
+      "step": 32650
+    },
+    {
+      "epoch": 0.4497998596954566,
+      "grad_norm": 0.09522339701652527,
+      "learning_rate": 0.0001,
+      "loss": 1.647,
+      "step": 32700
+    },
+    {
+      "epoch": 0.45048762706502155,
+      "grad_norm": 0.10342853516340256,
+      "learning_rate": 0.0001,
+      "loss": 1.6492,
+      "step": 32750
+    },
+    {
+      "epoch": 0.45117539443458643,
+      "grad_norm": 0.08631814271211624,
+      "learning_rate": 0.0001,
+      "loss": 1.647,
+      "step": 32800
+    },
+    {
+      "epoch": 0.45186316180415137,
+      "grad_norm": 0.09134256094694138,
+      "learning_rate": 0.0001,
+      "loss": 1.6444,
+      "step": 32850
+    },
+    {
+      "epoch": 0.4525509291737163,
+      "grad_norm": 0.09882289171218872,
+      "learning_rate": 0.0001,
+      "loss": 1.645,
+      "step": 32900
+    },
+    {
+      "epoch": 0.4532386965432812,
+      "grad_norm": 0.10555307567119598,
+      "learning_rate": 0.0001,
+      "loss": 1.6446,
+      "step": 32950
+    },
+    {
+      "epoch": 0.45392646391284613,
+      "grad_norm": 0.0928015485405922,
+      "learning_rate": 0.0001,
+      "loss": 1.6433,
+      "step": 33000
+    },
+    {
+      "epoch": 0.454614231282411,
+      "grad_norm": 0.09406717866659164,
+      "learning_rate": 0.0001,
+      "loss": 1.6418,
+      "step": 33050
+    },
+    {
+      "epoch": 0.45530199865197596,
+      "grad_norm": 0.09144891798496246,
+      "learning_rate": 0.0001,
+      "loss": 1.6487,
+      "step": 33100
+    },
+    {
+      "epoch": 0.4559897660215409,
+      "grad_norm": 0.09776079654693604,
+      "learning_rate": 0.0001,
+      "loss": 1.6453,
+      "step": 33150
+    },
+    {
+      "epoch": 0.4566775333911058,
+      "grad_norm": 0.09001190215349197,
+      "learning_rate": 0.0001,
+      "loss": 1.6447,
+      "step": 33200
+    },
+    {
+      "epoch": 0.4573653007606707,
+      "grad_norm": 0.09209096431732178,
+      "learning_rate": 0.0001,
+      "loss": 1.6437,
+      "step": 33250
+    },
+    {
+      "epoch": 0.4580530681302356,
+      "grad_norm": 0.09005513042211533,
+      "learning_rate": 0.0001,
+      "loss": 1.641,
+      "step": 33300
+    },
+    {
+      "epoch": 0.45874083549980055,
+      "grad_norm": 0.09380592405796051,
+      "learning_rate": 0.0001,
+      "loss": 1.6443,
+      "step": 33350
+    },
+    {
+      "epoch": 0.4594286028693655,
+      "grad_norm": 0.09379936754703522,
+      "learning_rate": 0.0001,
+      "loss": 1.6454,
+      "step": 33400
+    },
+    {
+      "epoch": 0.46011637023893037,
+      "grad_norm": 0.0955335795879364,
+      "learning_rate": 0.0001,
+      "loss": 1.645,
+      "step": 33450
+    },
+    {
+      "epoch": 0.4608041376084953,
+      "grad_norm": 0.10363741219043732,
+      "learning_rate": 0.0001,
+      "loss": 1.6439,
+      "step": 33500
+    },
+    {
+      "epoch": 0.4614919049780602,
+      "grad_norm": 0.08596354722976685,
+      "learning_rate": 0.0001,
+      "loss": 1.6446,
+      "step": 33550
+    },
+    {
+      "epoch": 0.46217967234762514,
+      "grad_norm": 0.08921484649181366,
+      "learning_rate": 0.0001,
+      "loss": 1.6443,
+      "step": 33600
+    },
+    {
+      "epoch": 0.4628674397171901,
+      "grad_norm": 0.09811948239803314,
+      "learning_rate": 0.0001,
+      "loss": 1.6462,
+      "step": 33650
+    },
+    {
+      "epoch": 0.46355520708675496,
+      "grad_norm": 0.0884086936712265,
+      "learning_rate": 0.0001,
+      "loss": 1.6458,
+      "step": 33700
+    },
+    {
+      "epoch": 0.4642429744563199,
+      "grad_norm": 0.08605129271745682,
+      "learning_rate": 0.0001,
+      "loss": 1.6428,
+      "step": 33750
+    },
+    {
+      "epoch": 0.4649307418258848,
+      "grad_norm": 0.10650883615016937,
+      "learning_rate": 0.0001,
+      "loss": 1.6468,
+      "step": 33800
+    },
+    {
+      "epoch": 0.4656185091954497,
+      "grad_norm": 0.09201018512248993,
+      "learning_rate": 0.0001,
+      "loss": 1.6463,
+      "step": 33850
+    },
+    {
+      "epoch": 0.46630627656501467,
+      "grad_norm": 0.09525769203901291,
+      "learning_rate": 0.0001,
+      "loss": 1.6444,
+      "step": 33900
+    },
+    {
+      "epoch": 0.46699404393457955,
+      "grad_norm": 0.10839025676250458,
+      "learning_rate": 0.0001,
+      "loss": 1.6393,
+      "step": 33950
+    },
+    {
+      "epoch": 0.4676818113041445,
+      "grad_norm": 0.09996642172336578,
+      "learning_rate": 0.0001,
+      "loss": 1.6469,
+      "step": 34000
+    },
+    {
+      "epoch": 0.46836957867370943,
+      "grad_norm": 0.09651756286621094,
+      "learning_rate": 0.0001,
+      "loss": 1.6418,
+      "step": 34050
+    },
+    {
+      "epoch": 0.4690573460432743,
+      "grad_norm": 0.10378842055797577,
+      "learning_rate": 0.0001,
+      "loss": 1.646,
+      "step": 34100
+    },
+    {
+      "epoch": 0.46974511341283925,
+      "grad_norm": 0.10965878516435623,
+      "learning_rate": 0.0001,
+      "loss": 1.6449,
+      "step": 34150
+    },
+    {
+      "epoch": 0.47043288078240414,
+      "grad_norm": 0.08958712965250015,
+      "learning_rate": 0.0001,
+      "loss": 1.6412,
+      "step": 34200
+    },
+    {
+      "epoch": 0.4711206481519691,
+      "grad_norm": 0.09303830564022064,
+      "learning_rate": 0.0001,
+      "loss": 1.6457,
+      "step": 34250
+    },
+    {
+      "epoch": 0.471808415521534,
+      "grad_norm": 0.08511397242546082,
+      "learning_rate": 0.0001,
+      "loss": 1.6408,
+      "step": 34300
+    },
+    {
+      "epoch": 0.4724961828910989,
+      "grad_norm": 0.09241296350955963,
+      "learning_rate": 0.0001,
+      "loss": 1.6452,
+      "step": 34350
+    },
+    {
+      "epoch": 0.47318395026066384,
+      "grad_norm": 0.10253586620092392,
+      "learning_rate": 0.0001,
+      "loss": 1.6424,
+      "step": 34400
+    },
+    {
+      "epoch": 0.47387171763022873,
+      "grad_norm": 0.09580568969249725,
+      "learning_rate": 0.0001,
+      "loss": 1.647,
+      "step": 34450
+    },
+    {
+      "epoch": 0.47455948499979367,
+      "grad_norm": 0.10204749554395676,
+      "learning_rate": 0.0001,
+      "loss": 1.6438,
+      "step": 34500
+    },
+    {
+      "epoch": 0.4752472523693586,
+      "grad_norm": 0.08885429054498672,
+      "learning_rate": 0.0001,
+      "loss": 1.6455,
+      "step": 34550
+    },
+    {
+      "epoch": 0.4759350197389235,
+      "grad_norm": 0.09475323557853699,
+      "learning_rate": 0.0001,
+      "loss": 1.6412,
+      "step": 34600
+    },
+    {
+      "epoch": 0.47662278710848843,
+      "grad_norm": 0.09055089205503464,
+      "learning_rate": 0.0001,
+      "loss": 1.6427,
+      "step": 34650
+    },
+    {
+      "epoch": 0.4773105544780533,
+      "grad_norm": 0.09198318421840668,
+      "learning_rate": 0.0001,
+      "loss": 1.6425,
+      "step": 34700
+    },
+    {
+      "epoch": 0.47799832184761826,
+      "grad_norm": 0.10331781953573227,
+      "learning_rate": 0.0001,
+      "loss": 1.649,
+      "step": 34750
+    },
+    {
+      "epoch": 0.4786860892171832,
+      "grad_norm": 0.0971902385354042,
+      "learning_rate": 0.0001,
+      "loss": 1.6438,
+      "step": 34800
+    },
+    {
+      "epoch": 0.4793738565867481,
+      "grad_norm": 0.13026170432567596,
+      "learning_rate": 0.0001,
+      "loss": 1.6443,
+      "step": 34850
+    },
+    {
+      "epoch": 0.480061623956313,
+      "grad_norm": 0.0918232873082161,
+      "learning_rate": 0.0001,
+      "loss": 1.6428,
+      "step": 34900
+    },
+    {
+      "epoch": 0.48074939132587796,
+      "grad_norm": 0.0905313566327095,
+      "learning_rate": 0.0001,
+      "loss": 1.645,
+      "step": 34950
+    },
+    {
+      "epoch": 0.48143715869544285,
+      "grad_norm": 0.08482769876718521,
+      "learning_rate": 0.0001,
+      "loss": 1.6442,
+      "step": 35000
+    },
+    {
+      "epoch": 0.4821249260650078,
+      "grad_norm": 0.09367966651916504,
+      "learning_rate": 0.0001,
+      "loss": 1.6408,
+      "step": 35050
+    },
+    {
+      "epoch": 0.48281269343457267,
+      "grad_norm": 0.08897031843662262,
+      "learning_rate": 0.0001,
+      "loss": 1.6465,
+      "step": 35100
+    },
+    {
+      "epoch": 0.4835004608041376,
+      "grad_norm": 0.09347031265497208,
+      "learning_rate": 0.0001,
+      "loss": 1.6466,
+      "step": 35150
+    },
+    {
+      "epoch": 0.48418822817370255,
+      "grad_norm": 0.09764907509088516,
+      "learning_rate": 0.0001,
+      "loss": 1.6449,
+      "step": 35200
+    },
+    {
+      "epoch": 0.48487599554326744,
+      "grad_norm": 0.09478213638067245,
+      "learning_rate": 0.0001,
+      "loss": 1.6415,
+      "step": 35250
+    },
+    {
+      "epoch": 0.4855637629128324,
+      "grad_norm": 0.09859916567802429,
+      "learning_rate": 0.0001,
+      "loss": 1.6416,
+      "step": 35300
+    },
+    {
+      "epoch": 0.48625153028239726,
+      "grad_norm": 0.10785259306430817,
+      "learning_rate": 0.0001,
+      "loss": 1.6452,
+      "step": 35350
+    },
+    {
+      "epoch": 0.4869392976519622,
+      "grad_norm": 0.09389789402484894,
+      "learning_rate": 0.0001,
+      "loss": 1.6422,
+      "step": 35400
+    },
+    {
+      "epoch": 0.48762706502152714,
+      "grad_norm": 0.0896189734339714,
+      "learning_rate": 0.0001,
+      "loss": 1.6444,
+      "step": 35450
+    },
+    {
+      "epoch": 0.488314832391092,
+      "grad_norm": 0.0927240401506424,
+      "learning_rate": 0.0001,
+      "loss": 1.642,
+      "step": 35500
+    },
+    {
+      "epoch": 0.48900259976065696,
+      "grad_norm": 0.09178788214921951,
+      "learning_rate": 0.0001,
+      "loss": 1.6427,
+      "step": 35550
+    },
+    {
+      "epoch": 0.48969036713022185,
+      "grad_norm": 0.09502946585416794,
+      "learning_rate": 0.0001,
+      "loss": 1.6416,
+      "step": 35600
+    },
+    {
+      "epoch": 0.4903781344997868,
+      "grad_norm": 0.11376187205314636,
+      "learning_rate": 0.0001,
+      "loss": 1.6445,
+      "step": 35650
+    },
+    {
+      "epoch": 0.49106590186935173,
+      "grad_norm": 0.09764853864908218,
+      "learning_rate": 0.0001,
+      "loss": 1.6415,
+      "step": 35700
+    },
+    {
+      "epoch": 0.4917536692389166,
+      "grad_norm": 0.09321732074022293,
+      "learning_rate": 0.0001,
+      "loss": 1.6413,
+      "step": 35750
+    },
+    {
+      "epoch": 0.49244143660848155,
+      "grad_norm": 0.09407221525907516,
+      "learning_rate": 0.0001,
+      "loss": 1.6441,
+      "step": 35800
+    },
+    {
+      "epoch": 0.4931292039780465,
+      "grad_norm": 0.09491898119449615,
+      "learning_rate": 0.0001,
+      "loss": 1.6455,
+      "step": 35850
+    },
+    {
+      "epoch": 0.4938169713476114,
+      "grad_norm": 0.0940353125333786,
+      "learning_rate": 0.0001,
+      "loss": 1.6444,
+      "step": 35900
+    },
+    {
+      "epoch": 0.4945047387171763,
+      "grad_norm": 0.10276436060667038,
+      "learning_rate": 0.0001,
+      "loss": 1.6444,
+      "step": 35950
+    },
+    {
+      "epoch": 0.4951925060867412,
+      "grad_norm": 0.10237590223550797,
+      "learning_rate": 0.0001,
+      "loss": 1.6423,
+      "step": 36000
+    },
+    {
+      "epoch": 0.49588027345630614,
+      "grad_norm": 0.08295126259326935,
+      "learning_rate": 0.0001,
+      "loss": 1.644,
+      "step": 36050
+    },
+    {
+      "epoch": 0.4965680408258711,
+      "grad_norm": 0.09782548248767853,
+      "learning_rate": 0.0001,
+      "loss": 1.6414,
+      "step": 36100
+    },
+    {
+      "epoch": 0.49725580819543597,
+      "grad_norm": 0.1010642945766449,
+      "learning_rate": 0.0001,
+      "loss": 1.6397,
+      "step": 36150
+    },
+    {
+      "epoch": 0.4979435755650009,
+      "grad_norm": 0.10704517364501953,
+      "learning_rate": 0.0001,
+      "loss": 1.6377,
+      "step": 36200
+    },
+    {
+      "epoch": 0.4986313429345658,
+      "grad_norm": 0.09292057156562805,
+      "learning_rate": 0.0001,
+      "loss": 1.6435,
+      "step": 36250
+    },
+    {
+      "epoch": 0.49931911030413073,
+      "grad_norm": 0.09314380586147308,
+      "learning_rate": 0.0001,
+      "loss": 1.6443,
+      "step": 36300
+    },
+    {
+      "epoch": 0.5000068776736957,
+      "grad_norm": 0.10215429961681366,
+      "learning_rate": 0.0001,
+      "loss": 1.6405,
+      "step": 36350
+    },
+    {
+      "epoch": 0.5006946450432606,
+      "grad_norm": 0.10266946256160736,
+      "learning_rate": 0.0001,
+      "loss": 1.6408,
+      "step": 36400
+    },
+    {
+      "epoch": 0.5013824124128254,
+      "grad_norm": 0.10435584932565689,
+      "learning_rate": 0.0001,
+      "loss": 1.6437,
+      "step": 36450
+    },
+    {
+      "epoch": 0.5020701797823904,
+      "grad_norm": 0.09806975722312927,
+      "learning_rate": 0.0001,
+      "loss": 1.6427,
+      "step": 36500
+    },
+    {
+      "epoch": 0.5027579471519553,
+      "grad_norm": 0.09484187513589859,
+      "learning_rate": 0.0001,
+      "loss": 1.6428,
+      "step": 36550
+    },
+    {
+      "epoch": 0.5034457145215202,
+      "grad_norm": 0.1000257357954979,
+      "learning_rate": 0.0001,
+      "loss": 1.6415,
+      "step": 36600
+    },
+    {
+      "epoch": 0.5041334818910852,
+      "grad_norm": 0.11959541589021683,
+      "learning_rate": 0.0001,
+      "loss": 1.6437,
+      "step": 36650
+    },
+    {
+      "epoch": 0.5048212492606501,
+      "grad_norm": 0.08579277247190475,
+      "learning_rate": 0.0001,
+      "loss": 1.6442,
+      "step": 36700
+    },
+    {
+      "epoch": 0.505509016630215,
+      "grad_norm": 0.09670394659042358,
+      "learning_rate": 0.0001,
+      "loss": 1.6422,
+      "step": 36750
+    },
+    {
+      "epoch": 0.50619678399978,
+      "grad_norm": 0.10155735909938812,
+      "learning_rate": 0.0001,
+      "loss": 1.6466,
+      "step": 36800
+    },
+    {
+      "epoch": 0.5068845513693448,
+      "grad_norm": 0.09957952052354813,
+      "learning_rate": 0.0001,
+      "loss": 1.6432,
+      "step": 36850
+    },
+    {
+      "epoch": 0.5075723187389097,
+      "grad_norm": 0.09222400933504105,
+      "learning_rate": 0.0001,
+      "loss": 1.6421,
+      "step": 36900
+    },
+    {
+      "epoch": 0.5082600861084746,
+      "grad_norm": 0.09004013985395432,
+      "learning_rate": 0.0001,
+      "loss": 1.6439,
+      "step": 36950
+    },
+    {
+      "epoch": 0.5089478534780396,
+      "grad_norm": 0.1122378334403038,
+      "learning_rate": 0.0001,
+      "loss": 1.643,
+      "step": 37000
+    },
+    {
+      "epoch": 0.5096356208476045,
+      "grad_norm": 0.10449572652578354,
+      "learning_rate": 0.0001,
+      "loss": 1.6461,
+      "step": 37050
+    },
+    {
+      "epoch": 0.5103233882171694,
+      "grad_norm": 0.09255963563919067,
+      "learning_rate": 0.0001,
+      "loss": 1.645,
+      "step": 37100
+    },
+    {
+      "epoch": 0.5110111555867344,
+      "grad_norm": 0.08528398722410202,
+      "learning_rate": 0.0001,
+      "loss": 1.6418,
+      "step": 37150
+    },
+    {
+      "epoch": 0.5116989229562993,
+      "grad_norm": 0.08895813673734665,
+      "learning_rate": 0.0001,
+      "loss": 1.6427,
+      "step": 37200
+    },
+    {
+      "epoch": 0.5123866903258641,
+      "grad_norm": 0.08886689692735672,
+      "learning_rate": 0.0001,
+      "loss": 1.6406,
+      "step": 37250
+    },
+    {
+      "epoch": 0.5130744576954291,
+      "grad_norm": 0.10976780205965042,
+      "learning_rate": 0.0001,
+      "loss": 1.6421,
+      "step": 37300
+    },
+    {
+      "epoch": 0.513762225064994,
+      "grad_norm": 0.09144289046525955,
+      "learning_rate": 0.0001,
+      "loss": 1.6379,
+      "step": 37350
+    },
+    {
+      "epoch": 0.5144499924345589,
+      "grad_norm": 0.0914168730378151,
+      "learning_rate": 0.0001,
+      "loss": 1.64,
+      "step": 37400
+    },
+    {
+      "epoch": 0.5151377598041239,
+      "grad_norm": 0.09350179880857468,
+      "learning_rate": 0.0001,
+      "loss": 1.6442,
+      "step": 37450
+    },
+    {
+      "epoch": 0.5158255271736888,
+      "grad_norm": 0.09165963530540466,
+      "learning_rate": 0.0001,
+      "loss": 1.6421,
+      "step": 37500
+    },
+    {
+      "epoch": 0.5165132945432537,
+      "grad_norm": 0.09791237860918045,
+      "learning_rate": 0.0001,
+      "loss": 1.6452,
+      "step": 37550
+    },
+    {
+      "epoch": 0.5172010619128186,
+      "grad_norm": 0.1264970451593399,
+      "learning_rate": 0.0001,
+      "loss": 1.6377,
+      "step": 37600
+    },
+    {
+      "epoch": 0.5178888292823836,
+      "grad_norm": 0.09959422051906586,
+      "learning_rate": 0.0001,
+      "loss": 1.6402,
+      "step": 37650
+    },
+    {
+      "epoch": 0.5185765966519484,
+      "grad_norm": 0.09144961088895798,
+      "learning_rate": 0.0001,
+      "loss": 1.639,
+      "step": 37700
+    },
+    {
+      "epoch": 0.5192643640215133,
+      "grad_norm": 0.09086447954177856,
+      "learning_rate": 0.0001,
+      "loss": 1.6479,
+      "step": 37750
+    },
+    {
+      "epoch": 0.5199521313910783,
+      "grad_norm": 0.10586626827716827,
+      "learning_rate": 0.0001,
+      "loss": 1.6416,
+      "step": 37800
+    },
+    {
+      "epoch": 0.5206398987606432,
+      "grad_norm": 0.10102024674415588,
+      "learning_rate": 0.0001,
+      "loss": 1.6378,
+      "step": 37850
+    },
+    {
+      "epoch": 0.5213276661302081,
+      "grad_norm": 0.09939133375883102,
+      "learning_rate": 0.0001,
+      "loss": 1.6402,
+      "step": 37900
+    },
+    {
+      "epoch": 0.5220154334997731,
+      "grad_norm": 0.0867060124874115,
+      "learning_rate": 0.0001,
+      "loss": 1.6412,
+      "step": 37950
+    },
+    {
+      "epoch": 0.522703200869338,
+      "grad_norm": 0.09193069487810135,
+      "learning_rate": 0.0001,
+      "loss": 1.6419,
+      "step": 38000
+    },
+    {
+      "epoch": 0.5233909682389029,
+      "grad_norm": 0.08879737555980682,
+      "learning_rate": 0.0001,
+      "loss": 1.6393,
+      "step": 38050
+    },
+    {
+      "epoch": 0.5240787356084677,
+      "grad_norm": 0.09856677055358887,
+      "learning_rate": 0.0001,
+      "loss": 1.6421,
+      "step": 38100
+    },
+    {
+      "epoch": 0.5247665029780327,
+      "grad_norm": 0.10344164073467255,
+      "learning_rate": 0.0001,
+      "loss": 1.6391,
+      "step": 38150
+    },
+    {
+      "epoch": 0.5254542703475976,
+      "grad_norm": 0.09736845642328262,
+      "learning_rate": 0.0001,
+      "loss": 1.6379,
+      "step": 38200
+    },
+    {
+      "epoch": 0.5261420377171625,
+      "grad_norm": 0.09295806288719177,
+      "learning_rate": 0.0001,
+      "loss": 1.6401,
+      "step": 38250
+    },
+    {
+      "epoch": 0.5268298050867275,
+      "grad_norm": 0.10306044667959213,
+      "learning_rate": 0.0001,
+      "loss": 1.6395,
+      "step": 38300
+    },
+    {
+      "epoch": 0.5275175724562924,
+      "grad_norm": 0.10452929884195328,
+      "learning_rate": 0.0001,
+      "loss": 1.642,
+      "step": 38350
+    },
+    {
+      "epoch": 0.5282053398258573,
+      "grad_norm": 0.1042618453502655,
+      "learning_rate": 0.0001,
+      "loss": 1.6434,
+      "step": 38400
+    },
+    {
+      "epoch": 0.5288931071954223,
+      "grad_norm": 0.09721370786428452,
+      "learning_rate": 0.0001,
+      "loss": 1.6395,
+      "step": 38450
+    },
+    {
+      "epoch": 0.5295808745649871,
+      "grad_norm": 0.09331890940666199,
+      "learning_rate": 0.0001,
+      "loss": 1.6418,
+      "step": 38500
+    },
+    {
+      "epoch": 0.530268641934552,
+      "grad_norm": 0.08918920904397964,
+      "learning_rate": 0.0001,
+      "loss": 1.6415,
+      "step": 38550
+    },
+    {
+      "epoch": 0.530956409304117,
+      "grad_norm": 0.10316510498523712,
+      "learning_rate": 0.0001,
+      "loss": 1.6388,
+      "step": 38600
+    },
+    {
+      "epoch": 0.5316441766736819,
+      "grad_norm": 0.0914444625377655,
+      "learning_rate": 0.0001,
+      "loss": 1.6436,
+      "step": 38650
+    },
+    {
+      "epoch": 0.5323319440432468,
+      "grad_norm": 0.09143774956464767,
+      "learning_rate": 0.0001,
+      "loss": 1.6434,
+      "step": 38700
+    },
+    {
+      "epoch": 0.5330197114128117,
+      "grad_norm": 0.10366670787334442,
+      "learning_rate": 0.0001,
+      "loss": 1.6444,
+      "step": 38750
+    },
+    {
+      "epoch": 0.5337074787823767,
+      "grad_norm": 0.09326903522014618,
+      "learning_rate": 0.0001,
+      "loss": 1.645,
+      "step": 38800
+    },
+    {
+      "epoch": 0.5343952461519416,
+      "grad_norm": 0.09438502043485641,
+      "learning_rate": 0.0001,
+      "loss": 1.6418,
+      "step": 38850
+    },
+    {
+      "epoch": 0.5350830135215064,
+      "grad_norm": 0.09484802931547165,
+      "learning_rate": 0.0001,
+      "loss": 1.6375,
+      "step": 38900
+    },
+    {
+      "epoch": 0.5357707808910714,
+      "grad_norm": 0.09375619888305664,
+      "learning_rate": 0.0001,
+      "loss": 1.6406,
+      "step": 38950
+    },
+    {
+      "epoch": 0.5364585482606363,
+      "grad_norm": 0.08983484655618668,
+      "learning_rate": 0.0001,
+      "loss": 1.6364,
+      "step": 39000
+    },
+    {
+      "epoch": 0.5371463156302012,
+      "grad_norm": 0.08946409076452255,
+      "learning_rate": 0.0001,
+      "loss": 1.6414,
+      "step": 39050
+    },
+    {
+      "epoch": 0.5378340829997662,
+      "grad_norm": 0.09255479276180267,
+      "learning_rate": 0.0001,
+      "loss": 1.6446,
+      "step": 39100
+    },
+    {
+      "epoch": 0.5385218503693311,
+      "grad_norm": 0.11457061767578125,
+      "learning_rate": 0.0001,
+      "loss": 1.6438,
+      "step": 39150
+    },
+    {
+      "epoch": 0.539209617738896,
+      "grad_norm": 0.09728623181581497,
+      "learning_rate": 0.0001,
+      "loss": 1.6433,
+      "step": 39200
+    },
+    {
+      "epoch": 0.5398973851084609,
+      "grad_norm": 0.09877431392669678,
+      "learning_rate": 0.0001,
+      "loss": 1.6404,
+      "step": 39250
+    },
+    {
+      "epoch": 0.5405851524780259,
+      "grad_norm": 0.09415557980537415,
+      "learning_rate": 0.0001,
+      "loss": 1.6403,
+      "step": 39300
+    },
+    {
+      "epoch": 0.5412729198475907,
+      "grad_norm": 0.1116420328617096,
+      "learning_rate": 0.0001,
+      "loss": 1.6395,
+      "step": 39350
+    },
+    {
+      "epoch": 0.5419606872171556,
+      "grad_norm": 0.0912848562002182,
+      "learning_rate": 0.0001,
+      "loss": 1.6419,
+      "step": 39400
+    },
+    {
+      "epoch": 0.5426484545867206,
+      "grad_norm": 0.0994088277220726,
+      "learning_rate": 0.0001,
+      "loss": 1.6401,
+      "step": 39450
+    },
+    {
+      "epoch": 0.5433362219562855,
+      "grad_norm": 0.12156657129526138,
+      "learning_rate": 0.0001,
+      "loss": 1.6411,
+      "step": 39500
+    },
+    {
+      "epoch": 0.5440239893258504,
+      "grad_norm": 0.09030243009328842,
+      "learning_rate": 0.0001,
+      "loss": 1.6396,
+      "step": 39550
+    },
+    {
+      "epoch": 0.5447117566954154,
+      "grad_norm": 0.09504235535860062,
+      "learning_rate": 0.0001,
+      "loss": 1.6417,
+      "step": 39600
+    },
+    {
+      "epoch": 0.5453995240649803,
+      "grad_norm": 0.09386838972568512,
+      "learning_rate": 0.0001,
+      "loss": 1.6399,
+      "step": 39650
+    },
+    {
+      "epoch": 0.5460872914345452,
+      "grad_norm": 0.09527257829904556,
+      "learning_rate": 0.0001,
+      "loss": 1.6379,
+      "step": 39700
+    },
+    {
+      "epoch": 0.5467750588041101,
+      "grad_norm": 0.09409776329994202,
+      "learning_rate": 0.0001,
+      "loss": 1.6377,
+      "step": 39750
+    },
+    {
+      "epoch": 0.547462826173675,
+      "grad_norm": 0.08981654047966003,
+      "learning_rate": 0.0001,
+      "loss": 1.6343,
+      "step": 39800
+    },
+    {
+      "epoch": 0.5481505935432399,
+      "grad_norm": 0.09664871543645859,
+      "learning_rate": 0.0001,
+      "loss": 1.6374,
+      "step": 39850
+    },
+    {
+      "epoch": 0.5488383609128048,
+      "grad_norm": 0.09075263887643814,
+      "learning_rate": 0.0001,
+      "loss": 1.6407,
+      "step": 39900
+    },
+    {
+      "epoch": 0.5495261282823698,
+      "grad_norm": 0.1084190309047699,
+      "learning_rate": 0.0001,
+      "loss": 1.6388,
+      "step": 39950
+    },
+    {
+      "epoch": 0.5502138956519347,
+      "grad_norm": 0.09443186968564987,
+      "learning_rate": 0.0001,
+      "loss": 1.6422,
+      "step": 40000
+    },
+    {
+      "epoch": 0.5509016630214996,
+      "grad_norm": 0.1035468801856041,
+      "learning_rate": 0.0001,
+      "loss": 1.6371,
+      "step": 40050
+    },
+    {
+      "epoch": 0.5515894303910646,
+      "grad_norm": 0.0966610461473465,
+      "learning_rate": 0.0001,
+      "loss": 1.6374,
+      "step": 40100
+    },
+    {
+      "epoch": 0.5522771977606294,
+      "grad_norm": 0.10241789370775223,
+      "learning_rate": 0.0001,
+      "loss": 1.6387,
+      "step": 40150
+    },
+    {
+      "epoch": 0.5529649651301943,
+      "grad_norm": 0.10644425451755524,
+      "learning_rate": 0.0001,
+      "loss": 1.6371,
+      "step": 40200
+    },
+    {
+      "epoch": 0.5536527324997593,
+      "grad_norm": 0.08675768971443176,
+      "learning_rate": 0.0001,
+      "loss": 1.6396,
+      "step": 40250
+    },
+    {
+      "epoch": 0.5543404998693242,
+      "grad_norm": 0.12207689881324768,
+      "learning_rate": 0.0001,
+      "loss": 1.6391,
+      "step": 40300
+    },
+    {
+      "epoch": 0.5550282672388891,
+      "grad_norm": 0.0907023698091507,
+      "learning_rate": 0.0001,
+      "loss": 1.6348,
+      "step": 40350
+    },
+    {
+      "epoch": 0.5557160346084541,
+      "grad_norm": 0.09063316136598587,
+      "learning_rate": 0.0001,
+      "loss": 1.6376,
+      "step": 40400
+    },
+    {
+      "epoch": 0.556403801978019,
+      "grad_norm": 0.10163768380880356,
+      "learning_rate": 0.0001,
+      "loss": 1.6361,
+      "step": 40450
+    },
+    {
+      "epoch": 0.5570915693475839,
+      "grad_norm": 0.11267077177762985,
+      "learning_rate": 0.0001,
+      "loss": 1.6377,
+      "step": 40500
+    },
+    {
+      "epoch": 0.5577793367171487,
+      "grad_norm": 0.11460031569004059,
+      "learning_rate": 0.0001,
+      "loss": 1.6365,
+      "step": 40550
+    },
+    {
+      "epoch": 0.5584671040867137,
+      "grad_norm": 0.09331660717725754,
+      "learning_rate": 0.0001,
+      "loss": 1.6411,
+      "step": 40600
+    },
+    {
+      "epoch": 0.5591548714562786,
+      "grad_norm": 0.09418249875307083,
+      "learning_rate": 0.0001,
+      "loss": 1.6372,
+      "step": 40650
+    },
+    {
+      "epoch": 0.5598426388258435,
+      "grad_norm": 0.10210473090410233,
+      "learning_rate": 0.0001,
+      "loss": 1.638,
+      "step": 40700
+    },
+    {
+      "epoch": 0.5605304061954085,
+      "grad_norm": 0.09525810927152634,
+      "learning_rate": 0.0001,
+      "loss": 1.6375,
+      "step": 40750
+    },
+    {
+      "epoch": 0.5612181735649734,
+      "grad_norm": 0.0911373421549797,
+      "learning_rate": 0.0001,
+      "loss": 1.6395,
+      "step": 40800
+    },
+    {
+      "epoch": 0.5619059409345383,
+      "grad_norm": 0.10007715970277786,
+      "learning_rate": 0.0001,
+      "loss": 1.6375,
+      "step": 40850
+    },
+    {
+      "epoch": 0.5625937083041033,
+      "grad_norm": 0.0913693979382515,
+      "learning_rate": 0.0001,
+      "loss": 1.6355,
+      "step": 40900
+    },
+    {
+      "epoch": 0.5632814756736682,
+      "grad_norm": 0.10602299869060516,
+      "learning_rate": 0.0001,
+      "loss": 1.6396,
+      "step": 40950
+    },
+    {
+      "epoch": 0.563969243043233,
+      "grad_norm": 0.10825448483228683,
+      "learning_rate": 0.0001,
+      "loss": 1.6365,
+      "step": 41000
+    },
+    {
+      "epoch": 0.5646570104127979,
+      "grad_norm": 0.09337472170591354,
+      "learning_rate": 0.0001,
+      "loss": 1.633,
+      "step": 41050
+    },
+    {
+      "epoch": 0.5653447777823629,
+      "grad_norm": 0.0868719145655632,
+      "learning_rate": 0.0001,
+      "loss": 1.6373,
+      "step": 41100
+    },
+    {
+      "epoch": 0.5660325451519278,
+      "grad_norm": 0.09493357688188553,
+      "learning_rate": 0.0001,
+      "loss": 1.6379,
+      "step": 41150
+    },
+    {
+      "epoch": 0.5667203125214927,
+      "grad_norm": 0.09529832005500793,
+      "learning_rate": 0.0001,
+      "loss": 1.6389,
+      "step": 41200
+    },
+    {
+      "epoch": 0.5674080798910577,
+      "grad_norm": 0.0934169664978981,
+      "learning_rate": 0.0001,
+      "loss": 1.6377,
+      "step": 41250
+    },
+    {
+      "epoch": 0.5680958472606226,
+      "grad_norm": 0.09936791658401489,
+      "learning_rate": 0.0001,
+      "loss": 1.6365,
+      "step": 41300
+    },
+    {
+      "epoch": 0.5687836146301875,
+      "grad_norm": 0.09473796933889389,
+      "learning_rate": 0.0001,
+      "loss": 1.6394,
+      "step": 41350
+    },
+    {
+      "epoch": 0.5694713819997524,
+      "grad_norm": 0.09517840296030045,
+      "learning_rate": 0.0001,
+      "loss": 1.6379,
+      "step": 41400
+    },
+    {
+      "epoch": 0.5701591493693173,
+      "grad_norm": 0.09487368911504745,
+      "learning_rate": 0.0001,
+      "loss": 1.6353,
+      "step": 41450
+    },
+    {
+      "epoch": 0.5708469167388822,
+      "grad_norm": 0.10064958035945892,
+      "learning_rate": 0.0001,
+      "loss": 1.6391,
+      "step": 41500
+    },
+    {
+      "epoch": 0.5715346841084472,
+      "grad_norm": 0.10654688626527786,
+      "learning_rate": 0.0001,
+      "loss": 1.6376,
+      "step": 41550
+    },
+    {
+      "epoch": 0.5722224514780121,
+      "grad_norm": 0.10369981825351715,
+      "learning_rate": 0.0001,
+      "loss": 1.6408,
+      "step": 41600
+    },
+    {
+      "epoch": 0.572910218847577,
+      "grad_norm": 0.09047607332468033,
+      "learning_rate": 0.0001,
+      "loss": 1.6369,
+      "step": 41650
+    },
+    {
+      "epoch": 0.5735979862171419,
+      "grad_norm": 0.0958494246006012,
+      "learning_rate": 0.0001,
+      "loss": 1.6422,
+      "step": 41700
+    },
+    {
+      "epoch": 0.5742857535867069,
+      "grad_norm": 0.09315364807844162,
+      "learning_rate": 0.0001,
+      "loss": 1.6366,
+      "step": 41750
+    },
+    {
+      "epoch": 0.5749735209562717,
+      "grad_norm": 0.10469076037406921,
+      "learning_rate": 0.0001,
+      "loss": 1.6376,
+      "step": 41800
+    },
+    {
+      "epoch": 0.5756612883258366,
+      "grad_norm": 0.1030019223690033,
+      "learning_rate": 0.0001,
+      "loss": 1.638,
+      "step": 41850
+    },
+    {
+      "epoch": 0.5763490556954016,
+      "grad_norm": 0.09740956872701645,
+      "learning_rate": 0.0001,
+      "loss": 1.6359,
+      "step": 41900
+    },
+    {
+      "epoch": 0.5770368230649665,
+      "grad_norm": 0.09355924278497696,
+      "learning_rate": 0.0001,
+      "loss": 1.6366,
+      "step": 41950
+    },
+    {
+      "epoch": 0.5777245904345314,
+      "grad_norm": 0.09326741844415665,
+      "learning_rate": 0.0001,
+      "loss": 1.6389,
+      "step": 42000
+    },
+    {
+      "epoch": 0.5784123578040964,
+      "grad_norm": 0.11062312871217728,
+      "learning_rate": 0.0001,
+      "loss": 1.6406,
+      "step": 42050
+    },
+    {
+      "epoch": 0.5791001251736613,
+      "grad_norm": 0.09104353934526443,
+      "learning_rate": 0.0001,
+      "loss": 1.64,
+      "step": 42100
+    },
+    {
+      "epoch": 0.5797878925432262,
+      "grad_norm": 0.1009470596909523,
+      "learning_rate": 0.0001,
+      "loss": 1.6396,
+      "step": 42150
+    },
+    {
+      "epoch": 0.580475659912791,
+      "grad_norm": 0.09980602562427521,
+      "learning_rate": 0.0001,
+      "loss": 1.6397,
+      "step": 42200
+    },
+    {
+      "epoch": 0.581163427282356,
+      "grad_norm": 0.08888303488492966,
+      "learning_rate": 0.0001,
+      "loss": 1.6393,
+      "step": 42250
+    },
+    {
+      "epoch": 0.5818511946519209,
+      "grad_norm": 0.09892967343330383,
+      "learning_rate": 0.0001,
+      "loss": 1.6405,
+      "step": 42300
+    },
+    {
+      "epoch": 0.5825389620214858,
+      "grad_norm": 0.09480975568294525,
+      "learning_rate": 0.0001,
+      "loss": 1.6401,
+      "step": 42350
+    },
+    {
+      "epoch": 0.5832267293910508,
+      "grad_norm": 0.08748164027929306,
+      "learning_rate": 0.0001,
+      "loss": 1.6385,
+      "step": 42400
+    },
+    {
+      "epoch": 0.5839144967606157,
+      "grad_norm": 0.0984395295381546,
+      "learning_rate": 0.0001,
+      "loss": 1.6399,
+      "step": 42450
+    },
+    {
+      "epoch": 0.5846022641301806,
+      "grad_norm": 0.09309285134077072,
+      "learning_rate": 0.0001,
+      "loss": 1.6395,
+      "step": 42500
+    },
+    {
+      "epoch": 0.5852900314997456,
+      "grad_norm": 0.09663578122854233,
+      "learning_rate": 0.0001,
+      "loss": 1.6366,
+      "step": 42550
+    },
+    {
+      "epoch": 0.5859777988693105,
+      "grad_norm": 0.1040002703666687,
+      "learning_rate": 0.0001,
+      "loss": 1.6373,
+      "step": 42600
+    },
+    {
+      "epoch": 0.5866655662388753,
+      "grad_norm": 0.0976615622639656,
+      "learning_rate": 0.0001,
+      "loss": 1.6372,
+      "step": 42650
+    },
+    {
+      "epoch": 0.5873533336084403,
+      "grad_norm": 0.0910048708319664,
+      "learning_rate": 0.0001,
+      "loss": 1.6404,
+      "step": 42700
+    },
+    {
+      "epoch": 0.5880411009780052,
+      "grad_norm": 0.10820795595645905,
+      "learning_rate": 0.0001,
+      "loss": 1.6364,
+      "step": 42750
+    },
+    {
+      "epoch": 0.5887288683475701,
+      "grad_norm": 0.08897601813077927,
+      "learning_rate": 0.0001,
+      "loss": 1.6323,
+      "step": 42800
+    },
+    {
+      "epoch": 0.589416635717135,
+      "grad_norm": 0.10053162276744843,
+      "learning_rate": 0.0001,
+      "loss": 1.6389,
+      "step": 42850
+    },
+    {
+      "epoch": 0.5901044030867,
+      "grad_norm": 0.10383192449808121,
+      "learning_rate": 0.0001,
+      "loss": 1.6396,
+      "step": 42900
+    },
+    {
+      "epoch": 0.5907921704562649,
+      "grad_norm": 0.11211832612752914,
+      "learning_rate": 0.0001,
+      "loss": 1.6399,
+      "step": 42950
+    },
+    {
+      "epoch": 0.5914799378258297,
+      "grad_norm": 0.10318583995103836,
+      "learning_rate": 0.0001,
+      "loss": 1.6385,
+      "step": 43000
+    },
+    {
+      "epoch": 0.5921677051953947,
+      "grad_norm": 0.09368006885051727,
+      "learning_rate": 0.0001,
+      "loss": 1.6385,
+      "step": 43050
+    },
+    {
+      "epoch": 0.5928554725649596,
+      "grad_norm": 0.10585025697946548,
+      "learning_rate": 0.0001,
+      "loss": 1.6404,
+      "step": 43100
+    },
+    {
+      "epoch": 0.5935432399345245,
+      "grad_norm": 0.11520183831453323,
+      "learning_rate": 0.0001,
+      "loss": 1.6371,
+      "step": 43150
+    },
+    {
+      "epoch": 0.5942310073040895,
+      "grad_norm": 0.10421840101480484,
+      "learning_rate": 0.0001,
+      "loss": 1.6384,
+      "step": 43200
+    },
+    {
+      "epoch": 0.5949187746736544,
+      "grad_norm": 0.11153177917003632,
+      "learning_rate": 0.0001,
+      "loss": 1.6395,
+      "step": 43250
+    },
+    {
+      "epoch": 0.5956065420432193,
+      "grad_norm": 0.09862000495195389,
+      "learning_rate": 0.0001,
+      "loss": 1.6343,
+      "step": 43300
+    },
+    {
+      "epoch": 0.5962943094127843,
+      "grad_norm": 0.11102604120969772,
+      "learning_rate": 0.0001,
+      "loss": 1.6375,
+      "step": 43350
+    },
+    {
+      "epoch": 0.5969820767823492,
+      "grad_norm": 0.11711497604846954,
+      "learning_rate": 0.0001,
+      "loss": 1.6384,
+      "step": 43400
+    },
+    {
+      "epoch": 0.597669844151914,
+      "grad_norm": 0.0938689261674881,
+      "learning_rate": 0.0001,
+      "loss": 1.6392,
+      "step": 43450
+    },
+    {
+      "epoch": 0.5983576115214789,
+      "grad_norm": 0.09853795170783997,
+      "learning_rate": 0.0001,
+      "loss": 1.6367,
+      "step": 43500
+    },
+    {
+      "epoch": 0.5990453788910439,
+      "grad_norm": 0.09836630523204803,
+      "learning_rate": 0.0001,
+      "loss": 1.6347,
+      "step": 43550
+    },
+    {
+      "epoch": 0.5997331462606088,
+      "grad_norm": 0.10666079819202423,
+      "learning_rate": 0.0001,
+      "loss": 1.6321,
+      "step": 43600
+    },
+    {
+      "epoch": 0.6004209136301737,
+      "grad_norm": 0.10402662307024002,
+      "learning_rate": 0.0001,
+      "loss": 1.6347,
+      "step": 43650
+    },
+    {
+      "epoch": 0.6011086809997387,
+      "grad_norm": 0.09178145229816437,
+      "learning_rate": 0.0001,
+      "loss": 1.6399,
+      "step": 43700
+    },
+    {
+      "epoch": 0.6017964483693036,
+      "grad_norm": 0.11560804396867752,
+      "learning_rate": 0.0001,
+      "loss": 1.6413,
+      "step": 43750
+    },
+    {
+      "epoch": 0.6024842157388685,
+      "grad_norm": 0.10294962674379349,
+      "learning_rate": 0.0001,
+      "loss": 1.6344,
+      "step": 43800
+    },
+    {
+      "epoch": 0.6031719831084335,
+      "grad_norm": 0.10806939750909805,
+      "learning_rate": 0.0001,
+      "loss": 1.637,
+      "step": 43850
+    },
+    {
+      "epoch": 0.6038597504779983,
+      "grad_norm": 0.09059430658817291,
+      "learning_rate": 0.0001,
+      "loss": 1.6391,
+      "step": 43900
+    },
+    {
+      "epoch": 0.6045475178475632,
+      "grad_norm": 0.09560652822256088,
+      "learning_rate": 0.0001,
+      "loss": 1.6384,
+      "step": 43950
+    },
+    {
+      "epoch": 0.6052352852171281,
+      "grad_norm": 0.1222764179110527,
+      "learning_rate": 0.0001,
+      "loss": 1.6378,
+      "step": 44000
+    },
+    {
+      "epoch": 0.6059230525866931,
+      "grad_norm": 0.09961801767349243,
+      "learning_rate": 0.0001,
+      "loss": 1.638,
+      "step": 44050
+    },
+    {
+      "epoch": 0.606610819956258,
+      "grad_norm": 0.09827202558517456,
+      "learning_rate": 0.0001,
+      "loss": 1.6351,
+      "step": 44100
+    },
+    {
+      "epoch": 0.6072985873258229,
+      "grad_norm": 0.09373018145561218,
+      "learning_rate": 0.0001,
+      "loss": 1.6367,
+      "step": 44150
+    },
+    {
+      "epoch": 0.6079863546953879,
+      "grad_norm": 0.1013326495885849,
+      "learning_rate": 0.0001,
+      "loss": 1.6359,
+      "step": 44200
+    },
+    {
+      "epoch": 0.6086741220649527,
+      "grad_norm": 0.09980370849370956,
+      "learning_rate": 0.0001,
+      "loss": 1.6361,
+      "step": 44250
+    },
+    {
+      "epoch": 0.6093618894345176,
+      "grad_norm": 0.0929165855050087,
+      "learning_rate": 0.0001,
+      "loss": 1.6415,
+      "step": 44300
+    },
+    {
+      "epoch": 0.6100496568040826,
+      "grad_norm": 0.11303722113370895,
+      "learning_rate": 0.0001,
+      "loss": 1.6324,
+      "step": 44350
+    },
+    {
+      "epoch": 0.6107374241736475,
+      "grad_norm": 0.09982936829328537,
+      "learning_rate": 0.0001,
+      "loss": 1.6383,
+      "step": 44400
+    },
+    {
+      "epoch": 0.6114251915432124,
+      "grad_norm": 0.10015662014484406,
+      "learning_rate": 0.0001,
+      "loss": 1.638,
+      "step": 44450
+    },
+    {
+      "epoch": 0.6121129589127774,
+      "grad_norm": 0.09394964575767517,
+      "learning_rate": 0.0001,
+      "loss": 1.6364,
+      "step": 44500
+    },
+    {
+      "epoch": 0.6128007262823423,
+      "grad_norm": 0.09288498759269714,
+      "learning_rate": 0.0001,
+      "loss": 1.6336,
+      "step": 44550
+    },
+    {
+      "epoch": 0.6134884936519072,
+      "grad_norm": 0.09083982557058334,
+      "learning_rate": 0.0001,
+      "loss": 1.6399,
+      "step": 44600
+    },
+    {
+      "epoch": 0.614176261021472,
+      "grad_norm": 0.11770026385784149,
+      "learning_rate": 0.0001,
+      "loss": 1.6363,
+      "step": 44650
+    },
+    {
+      "epoch": 0.614864028391037,
+      "grad_norm": 0.09519533067941666,
+      "learning_rate": 0.0001,
+      "loss": 1.6391,
+      "step": 44700
+    },
+    {
+      "epoch": 0.6155517957606019,
+      "grad_norm": 0.11158303171396255,
+      "learning_rate": 0.0001,
+      "loss": 1.6375,
+      "step": 44750
+    },
+    {
+      "epoch": 0.6162395631301668,
+      "grad_norm": 0.1033138781785965,
+      "learning_rate": 0.0001,
+      "loss": 1.6377,
+      "step": 44800
+    },
+    {
+      "epoch": 0.6169273304997318,
+      "grad_norm": 0.09701905399560928,
+      "learning_rate": 0.0001,
+      "loss": 1.6346,
+      "step": 44850
+    },
+    {
+      "epoch": 0.6176150978692967,
+      "grad_norm": 0.0948035940527916,
+      "learning_rate": 0.0001,
+      "loss": 1.6348,
+      "step": 44900
+    },
+    {
+      "epoch": 0.6183028652388616,
+      "grad_norm": 0.10962845385074615,
+      "learning_rate": 0.0001,
+      "loss": 1.6388,
+      "step": 44950
+    },
+    {
+      "epoch": 0.6189906326084266,
+      "grad_norm": 0.1121959462761879,
+      "learning_rate": 0.0001,
+      "loss": 1.6361,
+      "step": 45000
+    },
+    {
+      "epoch": 0.6196783999779915,
+      "grad_norm": 0.0968470573425293,
+      "learning_rate": 0.0001,
+      "loss": 1.6384,
+      "step": 45050
+    },
+    {
+      "epoch": 0.6203661673475563,
+      "grad_norm": 0.10296080261468887,
+      "learning_rate": 0.0001,
+      "loss": 1.6344,
+      "step": 45100
+    },
+    {
+      "epoch": 0.6210539347171212,
+      "grad_norm": 0.09767179936170578,
+      "learning_rate": 0.0001,
+      "loss": 1.635,
+      "step": 45150
+    },
+    {
+      "epoch": 0.6217417020866862,
+      "grad_norm": 0.09881862252950668,
+      "learning_rate": 0.0001,
+      "loss": 1.6338,
+      "step": 45200
+    },
+    {
+      "epoch": 0.6224294694562511,
+      "grad_norm": 0.11464895308017731,
+      "learning_rate": 0.0001,
+      "loss": 1.6344,
+      "step": 45250
+    },
+    {
+      "epoch": 0.623117236825816,
+      "grad_norm": 0.09954984486103058,
+      "learning_rate": 0.0001,
+      "loss": 1.6324,
+      "step": 45300
+    },
+    {
+      "epoch": 0.623805004195381,
+      "grad_norm": 0.10357275605201721,
+      "learning_rate": 0.0001,
+      "loss": 1.6382,
+      "step": 45350
+    },
+    {
+      "epoch": 0.6244927715649459,
+      "grad_norm": 0.09138759225606918,
+      "learning_rate": 0.0001,
+      "loss": 1.637,
+      "step": 45400
+    },
+    {
+      "epoch": 0.6251805389345108,
+      "grad_norm": 0.09229311347007751,
+      "learning_rate": 0.0001,
+      "loss": 1.6333,
+      "step": 45450
+    },
+    {
+      "epoch": 0.6258683063040757,
+      "grad_norm": 0.09420092403888702,
+      "learning_rate": 0.0001,
+      "loss": 1.6377,
+      "step": 45500
+    },
+    {
+      "epoch": 0.6265560736736406,
+      "grad_norm": 0.0874767154455185,
+      "learning_rate": 0.0001,
+      "loss": 1.635,
+      "step": 45550
+    },
+    {
+      "epoch": 0.6272438410432055,
+      "grad_norm": 0.09979282319545746,
+      "learning_rate": 0.0001,
+      "loss": 1.6356,
+      "step": 45600
+    },
+    {
+      "epoch": 0.6279316084127705,
+      "grad_norm": 0.105704665184021,
+      "learning_rate": 0.0001,
+      "loss": 1.6347,
+      "step": 45650
+    },
+    {
+      "epoch": 0.6286193757823354,
+      "grad_norm": 0.09676025807857513,
+      "learning_rate": 0.0001,
+      "loss": 1.6376,
+      "step": 45700
+    },
+    {
+      "epoch": 0.6293071431519003,
+      "grad_norm": 0.1145695224404335,
+      "learning_rate": 0.0001,
+      "loss": 1.6359,
+      "step": 45750
+    },
+    {
+      "epoch": 0.6299949105214652,
+      "grad_norm": 0.0872386172413826,
+      "learning_rate": 0.0001,
+      "loss": 1.6349,
+      "step": 45800
+    },
+    {
+      "epoch": 0.6306826778910302,
+      "grad_norm": 0.10765109956264496,
+      "learning_rate": 0.0001,
+      "loss": 1.6345,
+      "step": 45850
+    },
+    {
+      "epoch": 0.631370445260595,
+      "grad_norm": 0.09556550532579422,
+      "learning_rate": 0.0001,
+      "loss": 1.6346,
+      "step": 45900
+    },
+    {
+      "epoch": 0.6320582126301599,
+      "grad_norm": 0.09580112993717194,
+      "learning_rate": 0.0001,
+      "loss": 1.6357,
+      "step": 45950
+    },
+    {
+      "epoch": 0.6327459799997249,
+      "grad_norm": 0.09488419443368912,
+      "learning_rate": 0.0001,
+      "loss": 1.637,
+      "step": 46000
+    },
+    {
+      "epoch": 0.6334337473692898,
+      "grad_norm": 0.15395735204219818,
+      "learning_rate": 0.0001,
+      "loss": 1.6375,
+      "step": 46050
+    },
+    {
+      "epoch": 0.6341215147388547,
+      "grad_norm": 0.10091084241867065,
+      "learning_rate": 0.0001,
+      "loss": 1.6322,
+      "step": 46100
+    },
+    {
+      "epoch": 0.6348092821084197,
+      "grad_norm": 0.09612899273633957,
+      "learning_rate": 0.0001,
+      "loss": 1.6342,
+      "step": 46150
+    },
+    {
+      "epoch": 0.6354970494779846,
+      "grad_norm": 0.08732175827026367,
+      "learning_rate": 0.0001,
+      "loss": 1.6351,
+      "step": 46200
+    },
+    {
+      "epoch": 0.6361848168475495,
+      "grad_norm": 0.12121989578008652,
+      "learning_rate": 0.0001,
+      "loss": 1.6313,
+      "step": 46250
+    },
+    {
+      "epoch": 0.6368725842171145,
+      "grad_norm": 0.10441911965608597,
+      "learning_rate": 0.0001,
+      "loss": 1.6333,
+      "step": 46300
+    },
+    {
+      "epoch": 0.6375603515866793,
+      "grad_norm": 0.09910161048173904,
+      "learning_rate": 0.0001,
+      "loss": 1.6308,
+      "step": 46350
+    },
+    {
+      "epoch": 0.6382481189562442,
+      "grad_norm": 0.10073550045490265,
+      "learning_rate": 0.0001,
+      "loss": 1.6349,
+      "step": 46400
+    },
+    {
+      "epoch": 0.6389358863258091,
+      "grad_norm": 0.09164462983608246,
+      "learning_rate": 0.0001,
+      "loss": 1.6295,
+      "step": 46450
+    },
+    {
+      "epoch": 0.6396236536953741,
+      "grad_norm": 0.09485001116991043,
+      "learning_rate": 0.0001,
+      "loss": 1.6369,
+      "step": 46500
+    },
+    {
+      "epoch": 0.640311421064939,
+      "grad_norm": 0.09691156446933746,
+      "learning_rate": 0.0001,
+      "loss": 1.6345,
+      "step": 46550
+    },
+    {
+      "epoch": 0.6409991884345039,
+      "grad_norm": 0.0953722670674324,
+      "learning_rate": 0.0001,
+      "loss": 1.6308,
+      "step": 46600
+    },
+    {
+      "epoch": 0.6416869558040689,
+      "grad_norm": 0.09703515470027924,
+      "learning_rate": 0.0001,
+      "loss": 1.6349,
+      "step": 46650
+    },
+    {
+      "epoch": 0.6423747231736338,
+      "grad_norm": 0.09629255533218384,
+      "learning_rate": 0.0001,
+      "loss": 1.6346,
+      "step": 46700
+    },
+    {
+      "epoch": 0.6430624905431986,
+      "grad_norm": 0.09151779860258102,
+      "learning_rate": 0.0001,
+      "loss": 1.6338,
+      "step": 46750
+    },
+    {
+      "epoch": 0.6437502579127636,
+      "grad_norm": 0.0923495665192604,
+      "learning_rate": 0.0001,
+      "loss": 1.6315,
+      "step": 46800
+    },
+    {
+      "epoch": 0.6444380252823285,
+      "grad_norm": 0.09875286370515823,
+      "learning_rate": 0.0001,
+      "loss": 1.6301,
+      "step": 46850
+    },
+    {
+      "epoch": 0.6451257926518934,
+      "grad_norm": 0.09045711159706116,
+      "learning_rate": 0.0001,
+      "loss": 1.6336,
+      "step": 46900
+    },
+    {
+      "epoch": 0.6458135600214583,
+      "grad_norm": 0.10379324108362198,
+      "learning_rate": 0.0001,
+      "loss": 1.6338,
+      "step": 46950
+    },
+    {
+      "epoch": 0.6465013273910233,
+      "grad_norm": 0.11694531887769699,
+      "learning_rate": 0.0001,
+      "loss": 1.6351,
+      "step": 47000
+    },
+    {
+      "epoch": 0.6471890947605882,
+      "grad_norm": 0.11057207733392715,
+      "learning_rate": 0.0001,
+      "loss": 1.6355,
+      "step": 47050
+    },
+    {
+      "epoch": 0.647876862130153,
+      "grad_norm": 0.09550927579402924,
+      "learning_rate": 0.0001,
+      "loss": 1.6357,
+      "step": 47100
+    },
+    {
+      "epoch": 0.648564629499718,
+      "grad_norm": 0.09823106229305267,
+      "learning_rate": 0.0001,
+      "loss": 1.6314,
+      "step": 47150
+    },
+    {
+      "epoch": 0.6492523968692829,
+      "grad_norm": 0.10705850273370743,
+      "learning_rate": 0.0001,
+      "loss": 1.6336,
+      "step": 47200
+    },
+    {
+      "epoch": 0.6499401642388478,
+      "grad_norm": 0.10305814445018768,
+      "learning_rate": 0.0001,
+      "loss": 1.6323,
+      "step": 47250
+    },
+    {
+      "epoch": 0.6506279316084128,
+      "grad_norm": 0.09874992817640305,
+      "learning_rate": 0.0001,
+      "loss": 1.6292,
+      "step": 47300
+    },
+    {
+      "epoch": 0.6513156989779777,
+      "grad_norm": 0.09681331366300583,
+      "learning_rate": 0.0001,
+      "loss": 1.6377,
+      "step": 47350
+    },
+    {
+      "epoch": 0.6520034663475426,
+      "grad_norm": 0.10364115238189697,
+      "learning_rate": 0.0001,
+      "loss": 1.632,
+      "step": 47400
+    },
+    {
+      "epoch": 0.6526912337171076,
+      "grad_norm": 0.11742060631513596,
+      "learning_rate": 0.0001,
+      "loss": 1.6357,
+      "step": 47450
+    },
+    {
+      "epoch": 0.6533790010866725,
+      "grad_norm": 0.08997539430856705,
+      "learning_rate": 0.0001,
+      "loss": 1.6315,
+      "step": 47500
+    },
+    {
+      "epoch": 0.6540667684562373,
+      "grad_norm": 0.09857484698295593,
+      "learning_rate": 0.0001,
+      "loss": 1.6354,
+      "step": 47550
+    },
+    {
+      "epoch": 0.6547545358258022,
+      "grad_norm": 0.09228866547346115,
+      "learning_rate": 0.0001,
+      "loss": 1.6312,
+      "step": 47600
+    },
+    {
+      "epoch": 0.6554423031953672,
+      "grad_norm": 0.09897273778915405,
+      "learning_rate": 0.0001,
+      "loss": 1.6337,
+      "step": 47650
+    },
+    {
+      "epoch": 0.6561300705649321,
+      "grad_norm": 0.09947068989276886,
+      "learning_rate": 0.0001,
+      "loss": 1.6297,
+      "step": 47700
+    },
+    {
+      "epoch": 0.656817837934497,
+      "grad_norm": 0.11004633456468582,
+      "learning_rate": 0.0001,
+      "loss": 1.6314,
+      "step": 47750
+    },
+    {
+      "epoch": 0.657505605304062,
+      "grad_norm": 0.09651768952608109,
+      "learning_rate": 0.0001,
+      "loss": 1.6361,
+      "step": 47800
+    },
+    {
+      "epoch": 0.6581933726736269,
+      "grad_norm": 0.0905209481716156,
+      "learning_rate": 0.0001,
+      "loss": 1.6314,
+      "step": 47850
+    },
+    {
+      "epoch": 0.6588811400431918,
+      "grad_norm": 0.11071925610303879,
+      "learning_rate": 0.0001,
+      "loss": 1.6368,
+      "step": 47900
+    },
+    {
+      "epoch": 0.6595689074127568,
+      "grad_norm": 0.08634346723556519,
+      "learning_rate": 0.0001,
+      "loss": 1.6369,
+      "step": 47950
+    },
+    {
+      "epoch": 0.6602566747823216,
+      "grad_norm": 0.15600930154323578,
+      "learning_rate": 0.0001,
+      "loss": 1.6297,
+      "step": 48000
+    },
+    {
+      "epoch": 0.6609444421518865,
+      "grad_norm": 0.11416825652122498,
+      "learning_rate": 0.0001,
+      "loss": 1.6363,
+      "step": 48050
+    },
+    {
+      "epoch": 0.6616322095214515,
+      "grad_norm": 0.08934716135263443,
+      "learning_rate": 0.0001,
+      "loss": 1.6307,
+      "step": 48100
+    },
+    {
+      "epoch": 0.6623199768910164,
+      "grad_norm": 0.10360625386238098,
+      "learning_rate": 0.0001,
+      "loss": 1.6357,
+      "step": 48150
+    },
+    {
+      "epoch": 0.6630077442605813,
+      "grad_norm": 0.08972904086112976,
+      "learning_rate": 0.0001,
+      "loss": 1.6364,
+      "step": 48200
+    },
+    {
+      "epoch": 0.6636955116301462,
+      "grad_norm": 0.10486090183258057,
+      "learning_rate": 0.0001,
+      "loss": 1.6312,
+      "step": 48250
+    },
+    {
+      "epoch": 0.6643832789997112,
+      "grad_norm": 0.09509388357400894,
+      "learning_rate": 0.0001,
+      "loss": 1.632,
+      "step": 48300
+    },
+    {
+      "epoch": 0.665071046369276,
+      "grad_norm": 0.0978211984038353,
+      "learning_rate": 0.0001,
+      "loss": 1.6308,
+      "step": 48350
+    },
+    {
+      "epoch": 0.6657588137388409,
+      "grad_norm": 0.08715818077325821,
+      "learning_rate": 0.0001,
+      "loss": 1.6348,
+      "step": 48400
+    },
+    {
+      "epoch": 0.6664465811084059,
+      "grad_norm": 0.09371665865182877,
+      "learning_rate": 0.0001,
+      "loss": 1.6358,
+      "step": 48450
+    },
+    {
+      "epoch": 0.6671343484779708,
+      "grad_norm": 0.09343220293521881,
+      "learning_rate": 0.0001,
+      "loss": 1.6349,
+      "step": 48500
+    },
+    {
+      "epoch": 0.6678221158475357,
+      "grad_norm": 0.09670977294445038,
+      "learning_rate": 0.0001,
+      "loss": 1.6342,
+      "step": 48550
+    },
+    {
+      "epoch": 0.6685098832171007,
+      "grad_norm": 0.0940290242433548,
+      "learning_rate": 0.0001,
+      "loss": 1.6311,
+      "step": 48600
+    },
+    {
+      "epoch": 0.6691976505866656,
+      "grad_norm": 0.10598721355199814,
+      "learning_rate": 0.0001,
+      "loss": 1.6338,
+      "step": 48650
+    },
+    {
+      "epoch": 0.6698854179562305,
+      "grad_norm": 0.10703752934932709,
+      "learning_rate": 0.0001,
+      "loss": 1.6363,
+      "step": 48700
+    },
+    {
+      "epoch": 0.6705731853257954,
+      "grad_norm": 0.10036034137010574,
+      "learning_rate": 0.0001,
+      "loss": 1.6296,
+      "step": 48750
+    },
+    {
+      "epoch": 0.6712609526953603,
+      "grad_norm": 0.09091803431510925,
+      "learning_rate": 0.0001,
+      "loss": 1.6334,
+      "step": 48800
+    },
+    {
+      "epoch": 0.6719487200649252,
+      "grad_norm": 0.09322872757911682,
+      "learning_rate": 0.0001,
+      "loss": 1.6361,
+      "step": 48850
+    },
+    {
+      "epoch": 0.6726364874344901,
+      "grad_norm": 0.10486938804388046,
+      "learning_rate": 0.0001,
+      "loss": 1.6328,
+      "step": 48900
+    },
+    {
+      "epoch": 0.6733242548040551,
+      "grad_norm": 0.09341230988502502,
+      "learning_rate": 0.0001,
+      "loss": 1.6293,
+      "step": 48950
+    },
+    {
+      "epoch": 0.67401202217362,
+      "grad_norm": 0.09082542359828949,
+      "learning_rate": 0.0001,
+      "loss": 1.6335,
+      "step": 49000
+    },
+    {
+      "epoch": 0.6746997895431849,
+      "grad_norm": 0.10231276601552963,
+      "learning_rate": 0.0001,
+      "loss": 1.6319,
+      "step": 49050
+    },
+    {
+      "epoch": 0.6753875569127499,
+      "grad_norm": 0.08918558061122894,
+      "learning_rate": 0.0001,
+      "loss": 1.6386,
+      "step": 49100
+    },
+    {
+      "epoch": 0.6760753242823148,
+      "grad_norm": 0.09939314424991608,
+      "learning_rate": 0.0001,
+      "loss": 1.6343,
+      "step": 49150
+    },
+    {
+      "epoch": 0.6767630916518796,
+      "grad_norm": 0.09934555739164352,
+      "learning_rate": 0.0001,
+      "loss": 1.6348,
+      "step": 49200
+    },
+    {
+      "epoch": 0.6774508590214446,
+      "grad_norm": 0.09298571944236755,
+      "learning_rate": 0.0001,
+      "loss": 1.6317,
+      "step": 49250
+    },
+    {
+      "epoch": 0.6781386263910095,
+      "grad_norm": 0.09460543841123581,
+      "learning_rate": 0.0001,
+      "loss": 1.6298,
+      "step": 49300
+    },
+    {
+      "epoch": 0.6788263937605744,
+      "grad_norm": 0.08836425095796585,
+      "learning_rate": 0.0001,
+      "loss": 1.6328,
+      "step": 49350
+    },
+    {
+      "epoch": 0.6795141611301393,
+      "grad_norm": 0.11813092231750488,
+      "learning_rate": 0.0001,
+      "loss": 1.6351,
+      "step": 49400
+    },
+    {
+      "epoch": 0.6802019284997043,
+      "grad_norm": 0.09533282369375229,
+      "learning_rate": 0.0001,
+      "loss": 1.6338,
+      "step": 49450
+    },
+    {
+      "epoch": 0.6808896958692692,
+      "grad_norm": 0.09631752967834473,
+      "learning_rate": 0.0001,
+      "loss": 1.6309,
+      "step": 49500
+    },
+    {
+      "epoch": 0.6815774632388341,
+      "grad_norm": 0.10790685564279556,
+      "learning_rate": 0.0001,
+      "loss": 1.6299,
+      "step": 49550
+    },
+    {
+      "epoch": 0.682265230608399,
+      "grad_norm": 0.09565212577581406,
+      "learning_rate": 0.0001,
+      "loss": 1.6355,
+      "step": 49600
+    },
+    {
+      "epoch": 0.6829529979779639,
+      "grad_norm": 0.0955987349152565,
+      "learning_rate": 0.0001,
+      "loss": 1.6282,
+      "step": 49650
+    },
+    {
+      "epoch": 0.6836407653475288,
+      "grad_norm": 0.10549987852573395,
+      "learning_rate": 0.0001,
+      "loss": 1.6308,
+      "step": 49700
+    },
+    {
+      "epoch": 0.6843285327170938,
+      "grad_norm": 0.10708790272474289,
+      "learning_rate": 0.0001,
+      "loss": 1.6291,
+      "step": 49750
+    },
+    {
+      "epoch": 0.6850163000866587,
+      "grad_norm": 0.09704712778329849,
+      "learning_rate": 0.0001,
+      "loss": 1.6334,
+      "step": 49800
+    },
+    {
+      "epoch": 0.6857040674562236,
+      "grad_norm": 0.08443079143762589,
+      "learning_rate": 0.0001,
+      "loss": 1.629,
+      "step": 49850
+    },
+    {
+      "epoch": 0.6863918348257885,
+      "grad_norm": 0.09780169278383255,
+      "learning_rate": 0.0001,
+      "loss": 1.6333,
+      "step": 49900
+    },
+    {
+      "epoch": 0.6870796021953535,
+      "grad_norm": 0.09641287475824356,
+      "learning_rate": 0.0001,
+      "loss": 1.6351,
+      "step": 49950
+    },
+    {
+      "epoch": 0.6877673695649184,
+      "grad_norm": 0.10506296902894974,
+      "learning_rate": 0.0001,
+      "loss": 1.6337,
+      "step": 50000
+    },
+    {
+      "epoch": 0.6884551369344832,
+      "grad_norm": 0.08869095146656036,
+      "learning_rate": 0.0001,
+      "loss": 1.6316,
+      "step": 50050
+    },
+    {
+      "epoch": 0.6891429043040482,
+      "grad_norm": 0.0935509204864502,
+      "learning_rate": 0.0001,
+      "loss": 1.6354,
+      "step": 50100
+    },
+    {
+      "epoch": 0.6898306716736131,
+      "grad_norm": 0.09292100369930267,
+      "learning_rate": 0.0001,
+      "loss": 1.6294,
+      "step": 50150
+    },
+    {
+      "epoch": 0.690518439043178,
+      "grad_norm": 0.09042533487081528,
+      "learning_rate": 0.0001,
+      "loss": 1.6297,
+      "step": 50200
+    },
+    {
+      "epoch": 0.691206206412743,
+      "grad_norm": 0.09161574393510818,
+      "learning_rate": 0.0001,
+      "loss": 1.6329,
+      "step": 50250
+    },
+    {
+      "epoch": 0.6918939737823079,
+      "grad_norm": 0.0865265503525734,
+      "learning_rate": 0.0001,
+      "loss": 1.6328,
+      "step": 50300
+    },
+    {
+      "epoch": 0.6925817411518728,
+      "grad_norm": 0.09176287055015564,
+      "learning_rate": 0.0001,
+      "loss": 1.631,
+      "step": 50350
+    },
+    {
+      "epoch": 0.6932695085214378,
+      "grad_norm": 0.10909677296876907,
+      "learning_rate": 0.0001,
+      "loss": 1.6263,
+      "step": 50400
+    },
+    {
+      "epoch": 0.6939572758910026,
+      "grad_norm": 0.10558103770017624,
+      "learning_rate": 0.0001,
+      "loss": 1.6315,
+      "step": 50450
+    },
+    {
+      "epoch": 0.6946450432605675,
+      "grad_norm": 0.10107540339231491,
+      "learning_rate": 0.0001,
+      "loss": 1.6316,
+      "step": 50500
+    },
+    {
+      "epoch": 0.6953328106301324,
+      "grad_norm": 0.10059240460395813,
+      "learning_rate": 0.0001,
+      "loss": 1.6353,
+      "step": 50550
+    },
+    {
+      "epoch": 0.6960205779996974,
+      "grad_norm": 0.0953984409570694,
+      "learning_rate": 0.0001,
+      "loss": 1.6327,
+      "step": 50600
+    },
+    {
+      "epoch": 0.6967083453692623,
+      "grad_norm": 0.09249977767467499,
+      "learning_rate": 0.0001,
+      "loss": 1.6309,
+      "step": 50650
+    },
+    {
+      "epoch": 0.6973961127388272,
+      "grad_norm": 0.10159103572368622,
+      "learning_rate": 0.0001,
+      "loss": 1.6284,
+      "step": 50700
+    },
+    {
+      "epoch": 0.6980838801083922,
+      "grad_norm": 0.09958640486001968,
+      "learning_rate": 0.0001,
+      "loss": 1.6342,
+      "step": 50750
+    },
+    {
+      "epoch": 0.6987716474779571,
+      "grad_norm": 0.09354562312364578,
+      "learning_rate": 0.0001,
+      "loss": 1.6324,
+      "step": 50800
+    },
+    {
+      "epoch": 0.6994594148475219,
+      "grad_norm": 0.08785393089056015,
+      "learning_rate": 0.0001,
+      "loss": 1.6322,
+      "step": 50850
+    },
+    {
+      "epoch": 0.7001471822170869,
+      "grad_norm": 0.10028740018606186,
+      "learning_rate": 0.0001,
+      "loss": 1.6357,
+      "step": 50900
+    },
+    {
+      "epoch": 0.7008349495866518,
+      "grad_norm": 0.09850720316171646,
+      "learning_rate": 0.0001,
+      "loss": 1.628,
+      "step": 50950
+    },
+    {
+      "epoch": 0.7015227169562167,
+      "grad_norm": 0.0887802317738533,
+      "learning_rate": 0.0001,
+      "loss": 1.632,
+      "step": 51000
+    },
+    {
+      "epoch": 0.7022104843257817,
+      "grad_norm": 0.1057766005396843,
+      "learning_rate": 0.0001,
+      "loss": 1.6313,
+      "step": 51050
+    },
+    {
+      "epoch": 0.7028982516953466,
+      "grad_norm": 0.10256383568048477,
+      "learning_rate": 0.0001,
+      "loss": 1.6322,
+      "step": 51100
+    },
+    {
+      "epoch": 0.7035860190649115,
+      "grad_norm": 0.1051393449306488,
+      "learning_rate": 0.0001,
+      "loss": 1.6326,
+      "step": 51150
+    },
+    {
+      "epoch": 0.7042737864344764,
+      "grad_norm": 0.10550976544618607,
+      "learning_rate": 0.0001,
+      "loss": 1.6319,
+      "step": 51200
+    },
+    {
+      "epoch": 0.7049615538040414,
+      "grad_norm": 0.10621052980422974,
+      "learning_rate": 0.0001,
+      "loss": 1.6315,
+      "step": 51250
+    },
+    {
+      "epoch": 0.7056493211736062,
+      "grad_norm": 0.10684607177972794,
+      "learning_rate": 0.0001,
+      "loss": 1.6329,
+      "step": 51300
+    },
+    {
+      "epoch": 0.7063370885431711,
+      "grad_norm": 0.10426071286201477,
+      "learning_rate": 0.0001,
+      "loss": 1.6325,
+      "step": 51350
+    },
+    {
+      "epoch": 0.7070248559127361,
+      "grad_norm": 0.09892316907644272,
+      "learning_rate": 0.0001,
+      "loss": 1.6297,
+      "step": 51400
+    },
+    {
+      "epoch": 0.707712623282301,
+      "grad_norm": 0.11626065522432327,
+      "learning_rate": 0.0001,
+      "loss": 1.6345,
+      "step": 51450
+    },
+    {
+      "epoch": 0.7084003906518659,
+      "grad_norm": 0.10195190459489822,
+      "learning_rate": 0.0001,
+      "loss": 1.6279,
+      "step": 51500
+    },
+    {
+      "epoch": 0.7090881580214309,
+      "grad_norm": 0.10849519073963165,
+      "learning_rate": 0.0001,
+      "loss": 1.6299,
+      "step": 51550
+    },
+    {
+      "epoch": 0.7097759253909958,
+      "grad_norm": 0.10192925482988358,
+      "learning_rate": 0.0001,
+      "loss": 1.6327,
+      "step": 51600
+    },
+    {
+      "epoch": 0.7104636927605607,
+      "grad_norm": 0.09579839557409286,
+      "learning_rate": 0.0001,
+      "loss": 1.6357,
+      "step": 51650
+    },
+    {
+      "epoch": 0.7111514601301255,
+      "grad_norm": 0.09045730531215668,
+      "learning_rate": 0.0001,
+      "loss": 1.6338,
+      "step": 51700
+    },
+    {
+      "epoch": 0.7118392274996905,
+      "grad_norm": 0.09064441174268723,
+      "learning_rate": 0.0001,
+      "loss": 1.6333,
+      "step": 51750
+    },
+    {
+      "epoch": 0.7125269948692554,
+      "grad_norm": 0.09252890199422836,
+      "learning_rate": 0.0001,
+      "loss": 1.6323,
+      "step": 51800
+    },
+    {
+      "epoch": 0.7132147622388203,
+      "grad_norm": 0.09987664967775345,
+      "learning_rate": 0.0001,
+      "loss": 1.6347,
+      "step": 51850
+    },
+    {
+      "epoch": 0.7139025296083853,
+      "grad_norm": 0.09582144021987915,
+      "learning_rate": 0.0001,
+      "loss": 1.632,
+      "step": 51900
+    },
+    {
+      "epoch": 0.7145902969779502,
+      "grad_norm": 0.09969571232795715,
+      "learning_rate": 0.0001,
+      "loss": 1.6277,
+      "step": 51950
+    },
+    {
+      "epoch": 0.7152780643475151,
+      "grad_norm": 0.104249969124794,
+      "learning_rate": 0.0001,
+      "loss": 1.6321,
+      "step": 52000
+    },
+    {
+      "epoch": 0.7159658317170801,
+      "grad_norm": 0.1016249805688858,
+      "learning_rate": 0.0001,
+      "loss": 1.6331,
+      "step": 52050
+    },
+    {
+      "epoch": 0.7166535990866449,
+      "grad_norm": 0.10166680812835693,
+      "learning_rate": 0.0001,
+      "loss": 1.6281,
+      "step": 52100
+    },
+    {
+      "epoch": 0.7173413664562098,
+      "grad_norm": 0.08657228201627731,
+      "learning_rate": 0.0001,
+      "loss": 1.6267,
+      "step": 52150
+    },
+    {
+      "epoch": 0.7180291338257748,
+      "grad_norm": 0.09621983766555786,
+      "learning_rate": 0.0001,
+      "loss": 1.6301,
+      "step": 52200
+    },
+    {
+      "epoch": 0.7187169011953397,
+      "grad_norm": 0.10212218761444092,
+      "learning_rate": 0.0001,
+      "loss": 1.6283,
+      "step": 52250
+    },
+    {
+      "epoch": 0.7194046685649046,
+      "grad_norm": 0.09836029261350632,
+      "learning_rate": 0.0001,
+      "loss": 1.629,
+      "step": 52300
+    },
+    {
+      "epoch": 0.7200924359344695,
+      "grad_norm": 0.1043592095375061,
+      "learning_rate": 0.0001,
+      "loss": 1.6342,
+      "step": 52350
+    },
+    {
+      "epoch": 0.7207802033040345,
+      "grad_norm": 0.09003785252571106,
+      "learning_rate": 0.0001,
+      "loss": 1.631,
+      "step": 52400
+    },
+    {
+      "epoch": 0.7214679706735994,
+      "grad_norm": 0.09913681447505951,
+      "learning_rate": 0.0001,
+      "loss": 1.6303,
+      "step": 52450
+    },
+    {
+      "epoch": 0.7221557380431642,
+      "grad_norm": 0.0937609001994133,
+      "learning_rate": 0.0001,
+      "loss": 1.6339,
+      "step": 52500
+    },
+    {
+      "epoch": 0.7228435054127292,
+      "grad_norm": 0.10508369654417038,
+      "learning_rate": 0.0001,
+      "loss": 1.6315,
+      "step": 52550
+    },
+    {
+      "epoch": 0.7235312727822941,
+      "grad_norm": 0.08586754649877548,
+      "learning_rate": 0.0001,
+      "loss": 1.6355,
+      "step": 52600
+    },
+    {
+      "epoch": 0.724219040151859,
+      "grad_norm": 0.10271697491407394,
+      "learning_rate": 0.0001,
+      "loss": 1.6305,
+      "step": 52650
+    },
+    {
+      "epoch": 0.724906807521424,
+      "grad_norm": 0.09757651388645172,
+      "learning_rate": 0.0001,
+      "loss": 1.6299,
+      "step": 52700
+    },
+    {
+      "epoch": 0.7255945748909889,
+      "grad_norm": 0.09661528468132019,
+      "learning_rate": 0.0001,
+      "loss": 1.6346,
+      "step": 52750
+    },
+    {
+      "epoch": 0.7262823422605538,
+      "grad_norm": 0.10581836849451065,
+      "learning_rate": 0.0001,
+      "loss": 1.6296,
+      "step": 52800
+    },
+    {
+      "epoch": 0.7269701096301187,
+      "grad_norm": 0.08999978750944138,
+      "learning_rate": 0.0001,
+      "loss": 1.629,
+      "step": 52850
+    },
+    {
+      "epoch": 0.7276578769996837,
+      "grad_norm": 0.10840348899364471,
+      "learning_rate": 0.0001,
+      "loss": 1.6296,
+      "step": 52900
+    },
+    {
+      "epoch": 0.7283456443692485,
+      "grad_norm": 0.0947146937251091,
+      "learning_rate": 0.0001,
+      "loss": 1.6331,
+      "step": 52950
+    },
+    {
+      "epoch": 0.7290334117388134,
+      "grad_norm": 0.09926970303058624,
+      "learning_rate": 0.0001,
+      "loss": 1.633,
+      "step": 53000
+    },
+    {
+      "epoch": 0.7297211791083784,
+      "grad_norm": 0.10415767878293991,
+      "learning_rate": 0.0001,
+      "loss": 1.6298,
+      "step": 53050
+    },
+    {
+      "epoch": 0.7304089464779433,
+      "grad_norm": 0.11093425750732422,
+      "learning_rate": 0.0001,
+      "loss": 1.6296,
+      "step": 53100
+    },
+    {
+      "epoch": 0.7310967138475082,
+      "grad_norm": 0.11629913002252579,
+      "learning_rate": 0.0001,
+      "loss": 1.6299,
+      "step": 53150
+    },
+    {
+      "epoch": 0.7317844812170732,
+      "grad_norm": 0.09028749912977219,
+      "learning_rate": 0.0001,
+      "loss": 1.6296,
+      "step": 53200
+    },
+    {
+      "epoch": 0.7324722485866381,
+      "grad_norm": 0.09283788502216339,
+      "learning_rate": 0.0001,
+      "loss": 1.6316,
+      "step": 53250
+    },
+    {
+      "epoch": 0.733160015956203,
+      "grad_norm": 0.10100384056568146,
+      "learning_rate": 0.0001,
+      "loss": 1.6298,
+      "step": 53300
+    },
+    {
+      "epoch": 0.7338477833257679,
+      "grad_norm": 0.09549766778945923,
+      "learning_rate": 0.0001,
+      "loss": 1.6327,
+      "step": 53350
+    },
+    {
+      "epoch": 0.7345355506953328,
+      "grad_norm": 0.10129329562187195,
+      "learning_rate": 0.0001,
+      "loss": 1.6262,
+      "step": 53400
+    },
+    {
+      "epoch": 0.7352233180648977,
+      "grad_norm": 0.085689477622509,
+      "learning_rate": 0.0001,
+      "loss": 1.629,
+      "step": 53450
+    },
+    {
+      "epoch": 0.7359110854344626,
+      "grad_norm": 0.10251662135124207,
+      "learning_rate": 0.0001,
+      "loss": 1.6284,
+      "step": 53500
+    },
+    {
+      "epoch": 0.7365988528040276,
+      "grad_norm": 0.10034999996423721,
+      "learning_rate": 0.0001,
+      "loss": 1.6321,
+      "step": 53550
+    },
+    {
+      "epoch": 0.7372866201735925,
+      "grad_norm": 0.10141023993492126,
+      "learning_rate": 0.0001,
+      "loss": 1.6321,
+      "step": 53600
+    },
+    {
+      "epoch": 0.7379743875431574,
+      "grad_norm": 0.11427409946918488,
+      "learning_rate": 0.0001,
+      "loss": 1.6284,
+      "step": 53650
+    },
+    {
+      "epoch": 0.7386621549127224,
+      "grad_norm": 0.09212636202573776,
+      "learning_rate": 0.0001,
+      "loss": 1.6307,
+      "step": 53700
+    },
+    {
+      "epoch": 0.7393499222822872,
+      "grad_norm": 0.11293049156665802,
+      "learning_rate": 0.0001,
+      "loss": 1.6278,
+      "step": 53750
+    },
+    {
+      "epoch": 0.7400376896518521,
+      "grad_norm": 0.1038009375333786,
+      "learning_rate": 0.0001,
+      "loss": 1.6293,
+      "step": 53800
+    },
+    {
+      "epoch": 0.7407254570214171,
+      "grad_norm": 0.09193024039268494,
+      "learning_rate": 0.0001,
+      "loss": 1.6312,
+      "step": 53850
+    },
+    {
+      "epoch": 0.741413224390982,
+      "grad_norm": 0.10515929013490677,
+      "learning_rate": 0.0001,
+      "loss": 1.6303,
+      "step": 53900
+    },
+    {
+      "epoch": 0.7421009917605469,
+      "grad_norm": 0.09728860855102539,
+      "learning_rate": 0.0001,
+      "loss": 1.6324,
+      "step": 53950
+    },
+    {
+      "epoch": 0.7427887591301119,
+      "grad_norm": 0.10483654588460922,
+      "learning_rate": 0.0001,
+      "loss": 1.6336,
+      "step": 54000
+    },
+    {
+      "epoch": 0.7434765264996768,
+      "grad_norm": 0.09928806871175766,
+      "learning_rate": 0.0001,
+      "loss": 1.6302,
+      "step": 54050
+    },
+    {
+      "epoch": 0.7441642938692417,
+      "grad_norm": 0.10114587098360062,
+      "learning_rate": 0.0001,
+      "loss": 1.6239,
+      "step": 54100
+    },
+    {
+      "epoch": 0.7448520612388065,
+      "grad_norm": 0.09662097692489624,
+      "learning_rate": 0.0001,
+      "loss": 1.6305,
+      "step": 54150
+    },
+    {
+      "epoch": 0.7455398286083715,
+      "grad_norm": 0.1011410728096962,
+      "learning_rate": 0.0001,
+      "loss": 1.6293,
+      "step": 54200
+    },
+    {
+      "epoch": 0.7462275959779364,
+      "grad_norm": 0.09594978392124176,
+      "learning_rate": 0.0001,
+      "loss": 1.626,
+      "step": 54250
+    },
+    {
+      "epoch": 0.7469153633475013,
+      "grad_norm": 0.0986706018447876,
+      "learning_rate": 0.0001,
+      "loss": 1.6297,
+      "step": 54300
+    },
+    {
+      "epoch": 0.7476031307170663,
+      "grad_norm": 0.11080548167228699,
+      "learning_rate": 0.0001,
+      "loss": 1.629,
+      "step": 54350
+    },
+    {
+      "epoch": 0.7482908980866312,
+      "grad_norm": 0.10811348259449005,
+      "learning_rate": 0.0001,
+      "loss": 1.6266,
+      "step": 54400
+    },
+    {
+      "epoch": 0.7489786654561961,
+      "grad_norm": 0.09786169975996017,
+      "learning_rate": 0.0001,
+      "loss": 1.6303,
+      "step": 54450
+    },
+    {
+      "epoch": 0.7496664328257611,
+      "grad_norm": 0.09347623586654663,
+      "learning_rate": 0.0001,
+      "loss": 1.6292,
+      "step": 54500
+    },
+    {
+      "epoch": 0.750354200195326,
+      "grad_norm": 0.10388332605361938,
+      "learning_rate": 0.0001,
+      "loss": 1.6267,
+      "step": 54550
+    },
+    {
+      "epoch": 0.7510419675648908,
+      "grad_norm": 0.09866851568222046,
+      "learning_rate": 0.0001,
+      "loss": 1.629,
+      "step": 54600
+    },
+    {
+      "epoch": 0.7517297349344557,
+      "grad_norm": 0.10995014756917953,
+      "learning_rate": 0.0001,
+      "loss": 1.6312,
+      "step": 54650
+    },
+    {
+      "epoch": 0.7524175023040207,
+      "grad_norm": 0.09450124949216843,
+      "learning_rate": 0.0001,
+      "loss": 1.629,
+      "step": 54700
+    },
+    {
+      "epoch": 0.7531052696735856,
+      "grad_norm": 0.10181824117898941,
+      "learning_rate": 0.0001,
+      "loss": 1.6293,
+      "step": 54750
+    },
+    {
+      "epoch": 0.7537930370431505,
+      "grad_norm": 0.11469998210668564,
+      "learning_rate": 0.0001,
+      "loss": 1.6306,
+      "step": 54800
+    },
+    {
+      "epoch": 0.7544808044127155,
+      "grad_norm": 0.11170945316553116,
+      "learning_rate": 0.0001,
+      "loss": 1.6279,
+      "step": 54850
+    },
+    {
+      "epoch": 0.7551685717822804,
+      "grad_norm": 0.09184923022985458,
+      "learning_rate": 0.0001,
+      "loss": 1.6308,
+      "step": 54900
+    },
+    {
+      "epoch": 0.7558563391518452,
+      "grad_norm": 0.10830254852771759,
+      "learning_rate": 0.0001,
+      "loss": 1.6297,
+      "step": 54950
+    },
+    {
+      "epoch": 0.7565441065214102,
+      "grad_norm": 0.09561311453580856,
+      "learning_rate": 0.0001,
+      "loss": 1.6296,
+      "step": 55000
+    },
+    {
+      "epoch": 0.7572318738909751,
+      "grad_norm": 0.0922449454665184,
+      "learning_rate": 0.0001,
+      "loss": 1.6314,
+      "step": 55050
+    },
+    {
+      "epoch": 0.75791964126054,
+      "grad_norm": 0.09750694036483765,
+      "learning_rate": 0.0001,
+      "loss": 1.6263,
+      "step": 55100
+    },
+    {
+      "epoch": 0.758607408630105,
+      "grad_norm": 0.10356149077415466,
+      "learning_rate": 0.0001,
+      "loss": 1.6298,
+      "step": 55150
+    },
+    {
+      "epoch": 0.7592951759996699,
+      "grad_norm": 0.09972983598709106,
+      "learning_rate": 0.0001,
+      "loss": 1.6282,
+      "step": 55200
+    },
+    {
+      "epoch": 0.7599829433692348,
+      "grad_norm": 0.09673431515693665,
+      "learning_rate": 0.0001,
+      "loss": 1.6305,
+      "step": 55250
+    },
+    {
+      "epoch": 0.7606707107387997,
+      "grad_norm": 0.11159659922122955,
+      "learning_rate": 0.0001,
+      "loss": 1.6243,
+      "step": 55300
+    },
+    {
+      "epoch": 0.7613584781083647,
+      "grad_norm": 0.10829446464776993,
+      "learning_rate": 0.0001,
+      "loss": 1.6324,
+      "step": 55350
+    },
+    {
+      "epoch": 0.7620462454779295,
+      "grad_norm": 0.09917978942394257,
+      "learning_rate": 0.0001,
+      "loss": 1.6293,
+      "step": 55400
+    },
+    {
+      "epoch": 0.7627340128474944,
+      "grad_norm": 0.09947081655263901,
+      "learning_rate": 0.0001,
+      "loss": 1.6274,
+      "step": 55450
+    },
+    {
+      "epoch": 0.7634217802170594,
+      "grad_norm": 0.09545807540416718,
+      "learning_rate": 0.0001,
+      "loss": 1.6299,
+      "step": 55500
+    },
+    {
+      "epoch": 0.7641095475866243,
+      "grad_norm": 0.10184399038553238,
+      "learning_rate": 0.0001,
+      "loss": 1.631,
+      "step": 55550
+    },
+    {
+      "epoch": 0.7647973149561892,
+      "grad_norm": 0.0954732596874237,
+      "learning_rate": 0.0001,
+      "loss": 1.6285,
+      "step": 55600
+    },
+    {
+      "epoch": 0.7654850823257542,
+      "grad_norm": 0.1068364828824997,
+      "learning_rate": 0.0001,
+      "loss": 1.6289,
+      "step": 55650
+    },
+    {
+      "epoch": 0.7661728496953191,
+      "grad_norm": 0.09565538913011551,
+      "learning_rate": 0.0001,
+      "loss": 1.6293,
+      "step": 55700
+    },
+    {
+      "epoch": 0.766860617064884,
+      "grad_norm": 0.09793255478143692,
+      "learning_rate": 0.0001,
+      "loss": 1.6282,
+      "step": 55750
+    },
+    {
+      "epoch": 0.7675483844344488,
+      "grad_norm": 0.11079392582178116,
+      "learning_rate": 0.0001,
+      "loss": 1.6286,
+      "step": 55800
+    },
+    {
+      "epoch": 0.7682361518040138,
+      "grad_norm": 0.09202311187982559,
+      "learning_rate": 0.0001,
+      "loss": 1.6288,
+      "step": 55850
+    },
+    {
+      "epoch": 0.7689239191735787,
+      "grad_norm": 0.09076053649187088,
+      "learning_rate": 0.0001,
+      "loss": 1.6294,
+      "step": 55900
+    },
+    {
+      "epoch": 0.7696116865431436,
+      "grad_norm": 0.10245150327682495,
+      "learning_rate": 0.0001,
+      "loss": 1.6264,
+      "step": 55950
+    },
+    {
+      "epoch": 0.7702994539127086,
+      "grad_norm": 0.09228003025054932,
+      "learning_rate": 0.0001,
+      "loss": 1.6263,
+      "step": 56000
+    },
+    {
+      "epoch": 0.7709872212822735,
+      "grad_norm": 0.10882555693387985,
+      "learning_rate": 0.0001,
+      "loss": 1.632,
+      "step": 56050
+    },
+    {
+      "epoch": 0.7716749886518384,
+      "grad_norm": 0.11051888018846512,
+      "learning_rate": 0.0001,
+      "loss": 1.6292,
+      "step": 56100
+    },
+    {
+      "epoch": 0.7723627560214034,
+      "grad_norm": 0.11535434424877167,
+      "learning_rate": 0.0001,
+      "loss": 1.6254,
+      "step": 56150
+    },
+    {
+      "epoch": 0.7730505233909682,
+      "grad_norm": 0.09059146791696548,
+      "learning_rate": 0.0001,
+      "loss": 1.628,
+      "step": 56200
+    },
+    {
+      "epoch": 0.7737382907605331,
+      "grad_norm": 0.09102441370487213,
+      "learning_rate": 0.0001,
+      "loss": 1.6251,
+      "step": 56250
+    },
+    {
+      "epoch": 0.7744260581300981,
+      "grad_norm": 0.09455610811710358,
+      "learning_rate": 0.0001,
+      "loss": 1.6253,
+      "step": 56300
+    },
+    {
+      "epoch": 0.775113825499663,
+      "grad_norm": 0.09675318747758865,
+      "learning_rate": 0.0001,
+      "loss": 1.6273,
+      "step": 56350
+    },
+    {
+      "epoch": 0.7758015928692279,
+      "grad_norm": 0.08795568346977234,
+      "learning_rate": 0.0001,
+      "loss": 1.6272,
+      "step": 56400
+    },
+    {
+      "epoch": 0.7764893602387928,
+      "grad_norm": 0.1048702672123909,
+      "learning_rate": 0.0001,
+      "loss": 1.6242,
+      "step": 56450
+    },
+    {
+      "epoch": 0.7771771276083578,
+      "grad_norm": 0.10998234897851944,
+      "learning_rate": 0.0001,
+      "loss": 1.6259,
+      "step": 56500
+    },
+    {
+      "epoch": 0.7778648949779227,
+      "grad_norm": 0.09200160205364227,
+      "learning_rate": 0.0001,
+      "loss": 1.6266,
+      "step": 56550
+    },
+    {
+      "epoch": 0.7785526623474875,
+      "grad_norm": 0.12960132956504822,
+      "learning_rate": 0.0001,
+      "loss": 1.6278,
+      "step": 56600
+    },
+    {
+      "epoch": 0.7792404297170525,
+      "grad_norm": 0.09751798212528229,
+      "learning_rate": 0.0001,
+      "loss": 1.6267,
+      "step": 56650
+    },
+    {
+      "epoch": 0.7799281970866174,
+      "grad_norm": 0.09842194616794586,
+      "learning_rate": 0.0001,
+      "loss": 1.6299,
+      "step": 56700
+    },
+    {
+      "epoch": 0.7806159644561823,
+      "grad_norm": 0.09356516599655151,
+      "learning_rate": 0.0001,
+      "loss": 1.6272,
+      "step": 56750
+    },
+    {
+      "epoch": 0.7813037318257473,
+      "grad_norm": 0.10366610437631607,
+      "learning_rate": 0.0001,
+      "loss": 1.6282,
+      "step": 56800
+    },
+    {
+      "epoch": 0.7819914991953122,
+      "grad_norm": 0.10093358904123306,
+      "learning_rate": 0.0001,
+      "loss": 1.6288,
+      "step": 56850
+    },
+    {
+      "epoch": 0.7826792665648771,
+      "grad_norm": 0.11231804639101028,
+      "learning_rate": 0.0001,
+      "loss": 1.6279,
+      "step": 56900
+    },
+    {
+      "epoch": 0.7833670339344421,
+      "grad_norm": 0.12199211865663528,
+      "learning_rate": 0.0001,
+      "loss": 1.6276,
+      "step": 56950
+    },
+    {
+      "epoch": 0.784054801304007,
+      "grad_norm": 0.10776622593402863,
+      "learning_rate": 0.0001,
+      "loss": 1.6251,
+      "step": 57000
+    },
+    {
+      "epoch": 0.7847425686735718,
+      "grad_norm": 0.10442288219928741,
+      "learning_rate": 0.0001,
+      "loss": 1.6309,
+      "step": 57050
+    },
+    {
+      "epoch": 0.7854303360431367,
+      "grad_norm": 0.11845128238201141,
+      "learning_rate": 0.0001,
+      "loss": 1.6283,
+      "step": 57100
+    },
+    {
+      "epoch": 0.7861181034127017,
+      "grad_norm": 0.09980395436286926,
+      "learning_rate": 0.0001,
+      "loss": 1.6267,
+      "step": 57150
+    },
+    {
+      "epoch": 0.7868058707822666,
+      "grad_norm": 0.105705127120018,
+      "learning_rate": 0.0001,
+      "loss": 1.6281,
+      "step": 57200
+    },
+    {
+      "epoch": 0.7874936381518315,
+      "grad_norm": 0.09033431857824326,
+      "learning_rate": 0.0001,
+      "loss": 1.6317,
+      "step": 57250
+    },
+    {
+      "epoch": 0.7881814055213965,
+      "grad_norm": 0.10274254530668259,
+      "learning_rate": 0.0001,
+      "loss": 1.6286,
+      "step": 57300
+    },
+    {
+      "epoch": 0.7888691728909614,
+      "grad_norm": 0.09502615034580231,
+      "learning_rate": 0.0001,
+      "loss": 1.6265,
+      "step": 57350
+    },
+    {
+      "epoch": 0.7895569402605263,
+      "grad_norm": 0.10381683707237244,
+      "learning_rate": 0.0001,
+      "loss": 1.6306,
+      "step": 57400
+    },
+    {
+      "epoch": 0.7902447076300912,
+      "grad_norm": 0.10381805896759033,
+      "learning_rate": 0.0001,
+      "loss": 1.6248,
+      "step": 57450
+    },
+    {
+      "epoch": 0.7909324749996561,
+      "grad_norm": 0.0951840877532959,
+      "learning_rate": 0.0001,
+      "loss": 1.6263,
+      "step": 57500
+    },
+    {
+      "epoch": 0.791620242369221,
+      "grad_norm": 0.117914579808712,
+      "learning_rate": 0.0001,
+      "loss": 1.6316,
+      "step": 57550
+    },
+    {
+      "epoch": 0.7923080097387859,
+      "grad_norm": 0.10382041335105896,
+      "learning_rate": 0.0001,
+      "loss": 1.6268,
+      "step": 57600
+    },
+    {
+      "epoch": 0.7929957771083509,
+      "grad_norm": 0.11510676145553589,
+      "learning_rate": 0.0001,
+      "loss": 1.6255,
+      "step": 57650
+    },
+    {
+      "epoch": 0.7936835444779158,
+      "grad_norm": 0.09333761036396027,
+      "learning_rate": 0.0001,
+      "loss": 1.624,
+      "step": 57700
+    },
+    {
+      "epoch": 0.7943713118474807,
+      "grad_norm": 0.09227602928876877,
+      "learning_rate": 0.0001,
+      "loss": 1.631,
+      "step": 57750
+    },
+    {
+      "epoch": 0.7950590792170457,
+      "grad_norm": 0.10818013548851013,
+      "learning_rate": 0.0001,
+      "loss": 1.6254,
+      "step": 57800
+    },
+    {
+      "epoch": 0.7957468465866105,
+      "grad_norm": 0.10440421849489212,
+      "learning_rate": 0.0001,
+      "loss": 1.6308,
+      "step": 57850
+    },
+    {
+      "epoch": 0.7964346139561754,
+      "grad_norm": 0.10464609414339066,
+      "learning_rate": 0.0001,
+      "loss": 1.6291,
+      "step": 57900
+    },
+    {
+      "epoch": 0.7971223813257404,
+      "grad_norm": 0.11318383365869522,
+      "learning_rate": 0.0001,
+      "loss": 1.6283,
+      "step": 57950
+    },
+    {
+      "epoch": 0.7978101486953053,
+      "grad_norm": 0.1296527087688446,
+      "learning_rate": 0.0001,
+      "loss": 1.6254,
+      "step": 58000
+    },
+    {
+      "epoch": 0.7984979160648702,
+      "grad_norm": 0.09575483202934265,
+      "learning_rate": 0.0001,
+      "loss": 1.6288,
+      "step": 58050
+    },
+    {
+      "epoch": 0.7991856834344352,
+      "grad_norm": 0.11318851262331009,
+      "learning_rate": 0.0001,
+      "loss": 1.6287,
+      "step": 58100
+    },
+    {
+      "epoch": 0.7998734508040001,
+      "grad_norm": 0.10303325206041336,
+      "learning_rate": 0.0001,
+      "loss": 1.6272,
+      "step": 58150
+    },
+    {
+      "epoch": 0.800561218173565,
+      "grad_norm": 0.09992048889398575,
+      "learning_rate": 0.0001,
+      "loss": 1.6244,
+      "step": 58200
+    },
+    {
+      "epoch": 0.8012489855431298,
+      "grad_norm": 0.09654615819454193,
+      "learning_rate": 0.0001,
+      "loss": 1.6264,
+      "step": 58250
+    },
+    {
+      "epoch": 0.8019367529126948,
+      "grad_norm": 0.0984918400645256,
+      "learning_rate": 0.0001,
+      "loss": 1.6286,
+      "step": 58300
+    },
+    {
+      "epoch": 0.8026245202822597,
+      "grad_norm": 0.09900155663490295,
+      "learning_rate": 0.0001,
+      "loss": 1.6266,
+      "step": 58350
+    },
+    {
+      "epoch": 0.8033122876518246,
+      "grad_norm": 0.09438169002532959,
+      "learning_rate": 0.0001,
+      "loss": 1.6278,
+      "step": 58400
+    },
+    {
+      "epoch": 0.8040000550213896,
+      "grad_norm": 0.09372150897979736,
+      "learning_rate": 0.0001,
+      "loss": 1.6296,
+      "step": 58450
+    },
+    {
+      "epoch": 0.8046878223909545,
+      "grad_norm": 0.10187190771102905,
+      "learning_rate": 0.0001,
+      "loss": 1.6235,
+      "step": 58500
+    },
+    {
+      "epoch": 0.8053755897605194,
+      "grad_norm": 0.0914900004863739,
+      "learning_rate": 0.0001,
+      "loss": 1.6275,
+      "step": 58550
+    },
+    {
+      "epoch": 0.8060633571300844,
+      "grad_norm": 0.08726612478494644,
+      "learning_rate": 0.0001,
+      "loss": 1.6308,
+      "step": 58600
+    },
+    {
+      "epoch": 0.8067511244996493,
+      "grad_norm": 0.11641035974025726,
+      "learning_rate": 0.0001,
+      "loss": 1.6254,
+      "step": 58650
+    },
+    {
+      "epoch": 0.8074388918692141,
+      "grad_norm": 0.121851347386837,
+      "learning_rate": 0.0001,
+      "loss": 1.6297,
+      "step": 58700
+    },
+    {
+      "epoch": 0.808126659238779,
+      "grad_norm": 0.10153374820947647,
+      "learning_rate": 0.0001,
+      "loss": 1.6275,
+      "step": 58750
+    },
+    {
+      "epoch": 0.808814426608344,
+      "grad_norm": 0.10574651509523392,
+      "learning_rate": 0.0001,
+      "loss": 1.6269,
+      "step": 58800
+    },
+    {
+      "epoch": 0.8095021939779089,
+      "grad_norm": 0.11473171412944794,
+      "learning_rate": 0.0001,
+      "loss": 1.6268,
+      "step": 58850
+    },
+    {
+      "epoch": 0.8101899613474738,
+      "grad_norm": 0.09999868273735046,
+      "learning_rate": 0.0001,
+      "loss": 1.6232,
+      "step": 58900
+    },
+    {
+      "epoch": 0.8108777287170388,
+      "grad_norm": 0.10102663189172745,
+      "learning_rate": 0.0001,
+      "loss": 1.6234,
+      "step": 58950
+    },
+    {
+      "epoch": 0.8115654960866037,
+      "grad_norm": 0.09449347853660583,
+      "learning_rate": 0.0001,
+      "loss": 1.6305,
+      "step": 59000
+    },
+    {
+      "epoch": 0.8122532634561686,
+      "grad_norm": 0.1009218841791153,
+      "learning_rate": 0.0001,
+      "loss": 1.628,
+      "step": 59050
+    },
+    {
+      "epoch": 0.8129410308257335,
+      "grad_norm": 0.08553290367126465,
+      "learning_rate": 0.0001,
+      "loss": 1.63,
+      "step": 59100
+    },
+    {
+      "epoch": 0.8136287981952984,
+      "grad_norm": 0.1168675571680069,
+      "learning_rate": 0.0001,
+      "loss": 1.6277,
+      "step": 59150
+    },
+    {
+      "epoch": 0.8143165655648633,
+      "grad_norm": 0.10498977452516556,
+      "learning_rate": 0.0001,
+      "loss": 1.6236,
+      "step": 59200
+    },
+    {
+      "epoch": 0.8150043329344283,
+      "grad_norm": 0.09838738292455673,
+      "learning_rate": 0.0001,
+      "loss": 1.6263,
+      "step": 59250
+    },
+    {
+      "epoch": 0.8156921003039932,
+      "grad_norm": 0.1220153272151947,
+      "learning_rate": 0.0001,
+      "loss": 1.6248,
+      "step": 59300
+    },
+    {
+      "epoch": 0.8163798676735581,
+      "grad_norm": 0.12157600373029709,
+      "learning_rate": 0.0001,
+      "loss": 1.6253,
+      "step": 59350
+    },
+    {
+      "epoch": 0.817067635043123,
+      "grad_norm": 0.0998833104968071,
+      "learning_rate": 0.0001,
+      "loss": 1.6266,
+      "step": 59400
+    },
+    {
+      "epoch": 0.817755402412688,
+      "grad_norm": 0.10543464869260788,
+      "learning_rate": 0.0001,
+      "loss": 1.6267,
+      "step": 59450
+    },
+    {
+      "epoch": 0.8184431697822528,
+      "grad_norm": 0.09231004863977432,
+      "learning_rate": 0.0001,
+      "loss": 1.6249,
+      "step": 59500
+    },
+    {
+      "epoch": 0.8191309371518177,
+      "grad_norm": 0.10896420478820801,
+      "learning_rate": 0.0001,
+      "loss": 1.6251,
+      "step": 59550
+    },
+    {
+      "epoch": 0.8198187045213827,
+      "grad_norm": 0.09489674121141434,
+      "learning_rate": 0.0001,
+      "loss": 1.6268,
+      "step": 59600
+    },
+    {
+      "epoch": 0.8205064718909476,
+      "grad_norm": 0.10061246901750565,
+      "learning_rate": 0.0001,
+      "loss": 1.6294,
+      "step": 59650
+    },
+    {
+      "epoch": 0.8211942392605125,
+      "grad_norm": 0.10421578586101532,
+      "learning_rate": 0.0001,
+      "loss": 1.6244,
+      "step": 59700
+    },
+    {
+      "epoch": 0.8218820066300775,
+      "grad_norm": 0.10863976180553436,
+      "learning_rate": 0.0001,
+      "loss": 1.6268,
+      "step": 59750
+    },
+    {
+      "epoch": 0.8225697739996424,
+      "grad_norm": 0.10106450319290161,
+      "learning_rate": 0.0001,
+      "loss": 1.6247,
+      "step": 59800
+    },
+    {
+      "epoch": 0.8232575413692073,
+      "grad_norm": 0.1225091740489006,
+      "learning_rate": 0.0001,
+      "loss": 1.624,
+      "step": 59850
+    },
+    {
+      "epoch": 0.8239453087387723,
+      "grad_norm": 0.1194959282875061,
+      "learning_rate": 0.0001,
+      "loss": 1.6286,
+      "step": 59900
+    },
+    {
+      "epoch": 0.8246330761083371,
+      "grad_norm": 0.10718126595020294,
+      "learning_rate": 0.0001,
+      "loss": 1.6263,
+      "step": 59950
+    },
+    {
+      "epoch": 0.825320843477902,
+      "grad_norm": 0.09109228849411011,
+      "learning_rate": 0.0001,
+      "loss": 1.6279,
+      "step": 60000
+    },
+    {
+      "epoch": 0.8260086108474669,
+      "grad_norm": 0.12683629989624023,
+      "learning_rate": 0.0001,
+      "loss": 1.6303,
+      "step": 60050
+    },
+    {
+      "epoch": 0.8266963782170319,
+      "grad_norm": 0.09432388842105865,
+      "learning_rate": 0.0001,
+      "loss": 1.6248,
+      "step": 60100
+    },
+    {
+      "epoch": 0.8273841455865968,
+      "grad_norm": 0.10563753545284271,
+      "learning_rate": 0.0001,
+      "loss": 1.6247,
+      "step": 60150
+    },
+    {
+      "epoch": 0.8280719129561617,
+      "grad_norm": 0.09300840646028519,
+      "learning_rate": 0.0001,
+      "loss": 1.627,
+      "step": 60200
+    },
+    {
+      "epoch": 0.8287596803257267,
+      "grad_norm": 0.09419063478708267,
+      "learning_rate": 0.0001,
+      "loss": 1.6219,
+      "step": 60250
+    },
+    {
+      "epoch": 0.8294474476952916,
+      "grad_norm": 0.11976797133684158,
+      "learning_rate": 0.0001,
+      "loss": 1.6274,
+      "step": 60300
+    },
+    {
+      "epoch": 0.8301352150648564,
+      "grad_norm": 0.1019161194562912,
+      "learning_rate": 0.0001,
+      "loss": 1.6281,
+      "step": 60350
+    },
+    {
+      "epoch": 0.8308229824344214,
+      "grad_norm": 0.10402313619852066,
+      "learning_rate": 0.0001,
+      "loss": 1.6219,
+      "step": 60400
+    },
+    {
+      "epoch": 0.8315107498039863,
+      "grad_norm": 0.0936688631772995,
+      "learning_rate": 0.0001,
+      "loss": 1.6242,
+      "step": 60450
+    },
+    {
+      "epoch": 0.8321985171735512,
+      "grad_norm": 0.09794539213180542,
+      "learning_rate": 0.0001,
+      "loss": 1.6269,
+      "step": 60500
+    },
+    {
+      "epoch": 0.8328862845431161,
+      "grad_norm": 0.09901450574398041,
+      "learning_rate": 0.0001,
+      "loss": 1.6279,
+      "step": 60550
+    },
+    {
+      "epoch": 0.8335740519126811,
+      "grad_norm": 0.09567803889513016,
+      "learning_rate": 0.0001,
+      "loss": 1.6229,
+      "step": 60600
+    },
+    {
+      "epoch": 0.834261819282246,
+      "grad_norm": 0.12296677380800247,
+      "learning_rate": 0.0001,
+      "loss": 1.6262,
+      "step": 60650
+    },
+    {
+      "epoch": 0.8349495866518108,
+      "grad_norm": 0.10701528191566467,
+      "learning_rate": 0.0001,
+      "loss": 1.6297,
+      "step": 60700
+    },
+    {
+      "epoch": 0.8356373540213758,
+      "grad_norm": 0.09757701307535172,
+      "learning_rate": 0.0001,
+      "loss": 1.6246,
+      "step": 60750
+    },
+    {
+      "epoch": 0.8363251213909407,
+      "grad_norm": 0.11205516010522842,
+      "learning_rate": 0.0001,
+      "loss": 1.6232,
+      "step": 60800
+    },
+    {
+      "epoch": 0.8370128887605056,
+      "grad_norm": 0.09203255921602249,
+      "learning_rate": 0.0001,
+      "loss": 1.6262,
+      "step": 60850
+    },
+    {
+      "epoch": 0.8377006561300706,
+      "grad_norm": 0.10238678753376007,
+      "learning_rate": 0.0001,
+      "loss": 1.6256,
+      "step": 60900
+    },
+    {
+      "epoch": 0.8383884234996355,
+      "grad_norm": 0.0966443195939064,
+      "learning_rate": 0.0001,
+      "loss": 1.6282,
+      "step": 60950
+    },
+    {
+      "epoch": 0.8390761908692004,
+      "grad_norm": 0.09180071949958801,
+      "learning_rate": 0.0001,
+      "loss": 1.6252,
+      "step": 61000
+    },
+    {
+      "epoch": 0.8397639582387654,
+      "grad_norm": 0.09823059290647507,
+      "learning_rate": 0.0001,
+      "loss": 1.6264,
+      "step": 61050
+    },
+    {
+      "epoch": 0.8404517256083303,
+      "grad_norm": 0.11375342309474945,
+      "learning_rate": 0.0001,
+      "loss": 1.6276,
+      "step": 61100
+    },
+    {
+      "epoch": 0.8411394929778951,
+      "grad_norm": 0.10059309750795364,
+      "learning_rate": 0.0001,
+      "loss": 1.6278,
+      "step": 61150
+    },
+    {
+      "epoch": 0.84182726034746,
+      "grad_norm": 0.10789557546377182,
+      "learning_rate": 0.0001,
+      "loss": 1.6229,
+      "step": 61200
+    },
+    {
+      "epoch": 0.842515027717025,
+      "grad_norm": 0.09390353411436081,
+      "learning_rate": 0.0001,
+      "loss": 1.6251,
+      "step": 61250
+    },
+    {
+      "epoch": 0.8432027950865899,
+      "grad_norm": 0.09548452496528625,
+      "learning_rate": 0.0001,
+      "loss": 1.628,
+      "step": 61300
+    },
+    {
+      "epoch": 0.8438905624561548,
+      "grad_norm": 0.10145819932222366,
+      "learning_rate": 0.0001,
+      "loss": 1.6236,
+      "step": 61350
+    },
+    {
+      "epoch": 0.8445783298257198,
+      "grad_norm": 0.10150100290775299,
+      "learning_rate": 0.0001,
+      "loss": 1.6273,
+      "step": 61400
+    },
+    {
+      "epoch": 0.8452660971952847,
+      "grad_norm": 0.09277284890413284,
+      "learning_rate": 0.0001,
+      "loss": 1.6286,
+      "step": 61450
+    },
+    {
+      "epoch": 0.8459538645648496,
+      "grad_norm": 0.09942333400249481,
+      "learning_rate": 0.0001,
+      "loss": 1.628,
+      "step": 61500
+    },
+    {
+      "epoch": 0.8466416319344146,
+      "grad_norm": 0.0959252342581749,
+      "learning_rate": 0.0001,
+      "loss": 1.6214,
+      "step": 61550
+    },
+    {
+      "epoch": 0.8473293993039794,
+      "grad_norm": 0.12430740892887115,
+      "learning_rate": 0.0001,
+      "loss": 1.6248,
+      "step": 61600
+    },
+    {
+      "epoch": 0.8480171666735443,
+      "grad_norm": 0.09702388197183609,
+      "learning_rate": 0.0001,
+      "loss": 1.6234,
+      "step": 61650
+    },
+    {
+      "epoch": 0.8487049340431092,
+      "grad_norm": 0.09867244213819504,
+      "learning_rate": 0.0001,
+      "loss": 1.6236,
+      "step": 61700
+    },
+    {
+      "epoch": 0.8493927014126742,
+      "grad_norm": 0.10050085186958313,
+      "learning_rate": 0.0001,
+      "loss": 1.6254,
+      "step": 61750
+    },
+    {
+      "epoch": 0.8500804687822391,
+      "grad_norm": 0.09910161048173904,
+      "learning_rate": 0.0001,
+      "loss": 1.6253,
+      "step": 61800
+    },
+    {
+      "epoch": 0.850768236151804,
+      "grad_norm": 0.13913027942180634,
+      "learning_rate": 0.0001,
+      "loss": 1.6263,
+      "step": 61850
+    },
+    {
+      "epoch": 0.851456003521369,
+      "grad_norm": 0.11147304624319077,
+      "learning_rate": 0.0001,
+      "loss": 1.6249,
+      "step": 61900
+    },
+    {
+      "epoch": 0.8521437708909338,
+      "grad_norm": 0.09804028272628784,
+      "learning_rate": 0.0001,
+      "loss": 1.6278,
+      "step": 61950
+    },
+    {
+      "epoch": 0.8528315382604987,
+      "grad_norm": 0.09778320044279099,
+      "learning_rate": 0.0001,
+      "loss": 1.6242,
+      "step": 62000
+    },
+    {
+      "epoch": 0.8535193056300637,
+      "grad_norm": 0.098909392952919,
+      "learning_rate": 0.0001,
+      "loss": 1.6272,
+      "step": 62050
+    },
+    {
+      "epoch": 0.8542070729996286,
+      "grad_norm": 0.10008524358272552,
+      "learning_rate": 0.0001,
+      "loss": 1.6237,
+      "step": 62100
+    },
+    {
+      "epoch": 0.8548948403691935,
+      "grad_norm": 0.11636464297771454,
+      "learning_rate": 0.0001,
+      "loss": 1.6267,
+      "step": 62150
+    },
+    {
+      "epoch": 0.8555826077387585,
+      "grad_norm": 0.10067223757505417,
+      "learning_rate": 0.0001,
+      "loss": 1.6279,
+      "step": 62200
+    },
+    {
+      "epoch": 0.8562703751083234,
+      "grad_norm": 0.0951443538069725,
+      "learning_rate": 0.0001,
+      "loss": 1.6245,
+      "step": 62250
+    },
+    {
+      "epoch": 0.8569581424778883,
+      "grad_norm": 0.1106119304895401,
+      "learning_rate": 0.0001,
+      "loss": 1.6274,
+      "step": 62300
+    },
+    {
+      "epoch": 0.8576459098474531,
+      "grad_norm": 0.1256386637687683,
+      "learning_rate": 0.0001,
+      "loss": 1.6264,
+      "step": 62350
+    },
+    {
+      "epoch": 0.8583336772170181,
+      "grad_norm": 0.10102882981300354,
+      "learning_rate": 0.0001,
+      "loss": 1.6287,
+      "step": 62400
+    },
+    {
+      "epoch": 0.859021444586583,
+      "grad_norm": 0.0932788997888565,
+      "learning_rate": 0.0001,
+      "loss": 1.6264,
+      "step": 62450
+    },
+    {
+      "epoch": 0.8597092119561479,
+      "grad_norm": 0.10514089465141296,
+      "learning_rate": 0.0001,
+      "loss": 1.6265,
+      "step": 62500
+    },
+    {
+      "epoch": 0.8603969793257129,
+      "grad_norm": 0.11338023841381073,
+      "learning_rate": 0.0001,
+      "loss": 1.6291,
+      "step": 62550
+    },
+    {
+      "epoch": 0.8610847466952778,
+      "grad_norm": 0.11062069982290268,
+      "learning_rate": 0.0001,
+      "loss": 1.6243,
+      "step": 62600
+    },
+    {
+      "epoch": 0.8617725140648427,
+      "grad_norm": 0.11567613482475281,
+      "learning_rate": 0.0001,
+      "loss": 1.6247,
+      "step": 62650
+    },
+    {
+      "epoch": 0.8624602814344077,
+      "grad_norm": 0.10254676640033722,
+      "learning_rate": 0.0001,
+      "loss": 1.6267,
+      "step": 62700
+    },
+    {
+      "epoch": 0.8631480488039726,
+      "grad_norm": 0.10593708604574203,
+      "learning_rate": 0.0001,
+      "loss": 1.6245,
+      "step": 62750
+    },
+    {
+      "epoch": 0.8638358161735374,
+      "grad_norm": 0.10564611107110977,
+      "learning_rate": 0.0001,
+      "loss": 1.6256,
+      "step": 62800
+    },
+    {
+      "epoch": 0.8645235835431024,
+      "grad_norm": 0.10547158122062683,
+      "learning_rate": 0.0001,
+      "loss": 1.6227,
+      "step": 62850
+    },
+    {
+      "epoch": 0.8652113509126673,
+      "grad_norm": 0.09912673383951187,
+      "learning_rate": 0.0001,
+      "loss": 1.6269,
+      "step": 62900
+    },
+    {
+      "epoch": 0.8658991182822322,
+      "grad_norm": 0.09913560748100281,
+      "learning_rate": 0.0001,
+      "loss": 1.6212,
+      "step": 62950
+    },
+    {
+      "epoch": 0.8665868856517971,
+      "grad_norm": 0.10240307450294495,
+      "learning_rate": 0.0001,
+      "loss": 1.6221,
+      "step": 63000
+    },
+    {
+      "epoch": 0.8672746530213621,
+      "grad_norm": 0.11807160824537277,
+      "learning_rate": 0.0001,
+      "loss": 1.6249,
+      "step": 63050
+    },
+    {
+      "epoch": 0.867962420390927,
+      "grad_norm": 0.09863050282001495,
+      "learning_rate": 0.0001,
+      "loss": 1.6279,
+      "step": 63100
+    },
+    {
+      "epoch": 0.8686501877604919,
+      "grad_norm": 0.1057911291718483,
+      "learning_rate": 0.0001,
+      "loss": 1.6195,
+      "step": 63150
+    },
+    {
+      "epoch": 0.8693379551300568,
+      "grad_norm": 0.1161641925573349,
+      "learning_rate": 0.0001,
+      "loss": 1.626,
+      "step": 63200
+    },
+    {
+      "epoch": 0.8700257224996217,
+      "grad_norm": 0.10318218171596527,
+      "learning_rate": 0.0001,
+      "loss": 1.6245,
+      "step": 63250
+    },
+    {
+      "epoch": 0.8707134898691866,
+      "grad_norm": 0.09954388439655304,
+      "learning_rate": 0.0001,
+      "loss": 1.6231,
+      "step": 63300
+    },
+    {
+      "epoch": 0.8714012572387516,
+      "grad_norm": 0.09751243889331818,
+      "learning_rate": 0.0001,
+      "loss": 1.6186,
+      "step": 63350
+    },
+    {
+      "epoch": 0.8720890246083165,
+      "grad_norm": 0.09338164329528809,
+      "learning_rate": 0.0001,
+      "loss": 1.6249,
+      "step": 63400
+    },
+    {
+      "epoch": 0.8727767919778814,
+      "grad_norm": 0.09846915304660797,
+      "learning_rate": 0.0001,
+      "loss": 1.6262,
+      "step": 63450
+    },
+    {
+      "epoch": 0.8734645593474463,
+      "grad_norm": 0.10697762668132782,
+      "learning_rate": 0.0001,
+      "loss": 1.6249,
+      "step": 63500
+    },
+    {
+      "epoch": 0.8741523267170113,
+      "grad_norm": 0.10318434238433838,
+      "learning_rate": 0.0001,
+      "loss": 1.6209,
+      "step": 63550
+    },
+    {
+      "epoch": 0.8748400940865761,
+      "grad_norm": 0.10023129731416702,
+      "learning_rate": 0.0001,
+      "loss": 1.6227,
+      "step": 63600
+    },
+    {
+      "epoch": 0.875527861456141,
+      "grad_norm": 0.11498495936393738,
+      "learning_rate": 0.0001,
+      "loss": 1.6199,
+      "step": 63650
+    },
+    {
+      "epoch": 0.876215628825706,
+      "grad_norm": 0.09535204619169235,
+      "learning_rate": 0.0001,
+      "loss": 1.6255,
+      "step": 63700
+    },
+    {
+      "epoch": 0.8769033961952709,
+      "grad_norm": 0.10134591907262802,
+      "learning_rate": 0.0001,
+      "loss": 1.6245,
+      "step": 63750
+    },
+    {
+      "epoch": 0.8775911635648358,
+      "grad_norm": 0.11040335893630981,
+      "learning_rate": 0.0001,
+      "loss": 1.6282,
+      "step": 63800
+    },
+    {
+      "epoch": 0.8782789309344008,
+      "grad_norm": 0.09738776832818985,
+      "learning_rate": 0.0001,
+      "loss": 1.6254,
+      "step": 63850
+    },
+    {
+      "epoch": 0.8789666983039657,
+      "grad_norm": 0.09208226203918457,
+      "learning_rate": 0.0001,
+      "loss": 1.6227,
+      "step": 63900
+    },
+    {
+      "epoch": 0.8796544656735306,
+      "grad_norm": 0.1082238256931305,
+      "learning_rate": 0.0001,
+      "loss": 1.6241,
+      "step": 63950
+    },
+    {
+      "epoch": 0.8803422330430956,
+      "grad_norm": 0.10054194927215576,
+      "learning_rate": 0.0001,
+      "loss": 1.6202,
+      "step": 64000
+    },
+    {
+      "epoch": 0.8810300004126604,
+      "grad_norm": 0.10898640006780624,
+      "learning_rate": 0.0001,
+      "loss": 1.6235,
+      "step": 64050
+    },
+    {
+      "epoch": 0.8817177677822253,
+      "grad_norm": 0.10575822740793228,
+      "learning_rate": 0.0001,
+      "loss": 1.6215,
+      "step": 64100
+    },
+    {
+      "epoch": 0.8824055351517902,
+      "grad_norm": 0.0936664566397667,
+      "learning_rate": 0.0001,
+      "loss": 1.6224,
+      "step": 64150
+    },
+    {
+      "epoch": 0.8830933025213552,
+      "grad_norm": 0.08812412619590759,
+      "learning_rate": 0.0001,
+      "loss": 1.6228,
+      "step": 64200
+    },
+    {
+      "epoch": 0.8837810698909201,
+      "grad_norm": 0.10116361826658249,
+      "learning_rate": 0.0001,
+      "loss": 1.6207,
+      "step": 64250
+    },
+    {
+      "epoch": 0.884468837260485,
+      "grad_norm": 0.09832023084163666,
+      "learning_rate": 0.0001,
+      "loss": 1.6208,
+      "step": 64300
+    },
+    {
+      "epoch": 0.88515660463005,
+      "grad_norm": 0.10003996640443802,
+      "learning_rate": 0.0001,
+      "loss": 1.6268,
+      "step": 64350
+    },
+    {
+      "epoch": 0.8858443719996149,
+      "grad_norm": 0.10371046513319016,
+      "learning_rate": 0.0001,
+      "loss": 1.6211,
+      "step": 64400
+    },
+    {
+      "epoch": 0.8865321393691797,
+      "grad_norm": 0.11491281539201736,
+      "learning_rate": 0.0001,
+      "loss": 1.6216,
+      "step": 64450
+    },
+    {
+      "epoch": 0.8872199067387447,
+      "grad_norm": 0.10414805263280869,
+      "learning_rate": 0.0001,
+      "loss": 1.6198,
+      "step": 64500
+    },
+    {
+      "epoch": 0.8879076741083096,
+      "grad_norm": 0.09586089849472046,
+      "learning_rate": 0.0001,
+      "loss": 1.6254,
+      "step": 64550
+    },
+    {
+      "epoch": 0.8885954414778745,
+      "grad_norm": 0.09846927970647812,
+      "learning_rate": 0.0001,
+      "loss": 1.627,
+      "step": 64600
+    },
+    {
+      "epoch": 0.8892832088474394,
+      "grad_norm": 0.09198208153247833,
+      "learning_rate": 0.0001,
+      "loss": 1.6253,
+      "step": 64650
+    },
+    {
+      "epoch": 0.8899709762170044,
+      "grad_norm": 0.08919122070074081,
+      "learning_rate": 0.0001,
+      "loss": 1.6255,
+      "step": 64700
+    },
+    {
+      "epoch": 0.8906587435865693,
+      "grad_norm": 0.10351312905550003,
+      "learning_rate": 0.0001,
+      "loss": 1.6259,
+      "step": 64750
+    },
+    {
+      "epoch": 0.8913465109561342,
+      "grad_norm": 0.10056328028440475,
+      "learning_rate": 0.0001,
+      "loss": 1.6242,
+      "step": 64800
+    },
+    {
+      "epoch": 0.8920342783256991,
+      "grad_norm": 0.11205385625362396,
+      "learning_rate": 0.0001,
+      "loss": 1.6221,
+      "step": 64850
+    },
+    {
+      "epoch": 0.892722045695264,
+      "grad_norm": 0.11789285391569138,
+      "learning_rate": 0.0001,
+      "loss": 1.622,
+      "step": 64900
+    },
+    {
+      "epoch": 0.8934098130648289,
+      "grad_norm": 0.09621193259954453,
+      "learning_rate": 0.0001,
+      "loss": 1.629,
+      "step": 64950
+    },
+    {
+      "epoch": 0.8940975804343939,
+      "grad_norm": 0.11940276622772217,
+      "learning_rate": 0.0001,
+      "loss": 1.6276,
+      "step": 65000
+    },
+    {
+      "epoch": 0.8947853478039588,
+      "grad_norm": 0.0920642614364624,
+      "learning_rate": 0.0001,
+      "loss": 1.6247,
+      "step": 65050
+    },
+    {
+      "epoch": 0.8954731151735237,
+      "grad_norm": 0.11501994729042053,
+      "learning_rate": 0.0001,
+      "loss": 1.6215,
+      "step": 65100
+    },
+    {
+      "epoch": 0.8961608825430887,
+      "grad_norm": 0.09871424734592438,
+      "learning_rate": 0.0001,
+      "loss": 1.6208,
+      "step": 65150
+    },
+    {
+      "epoch": 0.8968486499126536,
+      "grad_norm": 0.11804469674825668,
+      "learning_rate": 0.0001,
+      "loss": 1.622,
+      "step": 65200
+    },
+    {
+      "epoch": 0.8975364172822184,
+      "grad_norm": 0.1077907532453537,
+      "learning_rate": 0.0001,
+      "loss": 1.6241,
+      "step": 65250
+    },
+    {
+      "epoch": 0.8982241846517833,
+      "grad_norm": 0.1031891405582428,
+      "learning_rate": 0.0001,
+      "loss": 1.6224,
+      "step": 65300
+    },
+    {
+      "epoch": 0.8989119520213483,
+      "grad_norm": 0.09871700406074524,
+      "learning_rate": 0.0001,
+      "loss": 1.6263,
+      "step": 65350
+    },
+    {
+      "epoch": 0.8995997193909132,
+      "grad_norm": 0.11097069084644318,
+      "learning_rate": 0.0001,
+      "loss": 1.6231,
+      "step": 65400
+    },
+    {
+      "epoch": 0.9002874867604781,
+      "grad_norm": 0.10109886527061462,
+      "learning_rate": 0.0001,
+      "loss": 1.6204,
+      "step": 65450
+    },
+    {
+      "epoch": 0.9009752541300431,
+      "grad_norm": 0.09359996765851974,
+      "learning_rate": 0.0001,
+      "loss": 1.6247,
+      "step": 65500
+    },
+    {
+      "epoch": 0.901663021499608,
+      "grad_norm": 0.09548566490411758,
+      "learning_rate": 0.0001,
+      "loss": 1.6262,
+      "step": 65550
+    },
+    {
+      "epoch": 0.9023507888691729,
+      "grad_norm": 0.09799979627132416,
+      "learning_rate": 0.0001,
+      "loss": 1.6221,
+      "step": 65600
+    },
+    {
+      "epoch": 0.9030385562387379,
+      "grad_norm": 0.10698045790195465,
+      "learning_rate": 0.0001,
+      "loss": 1.6245,
+      "step": 65650
+    },
+    {
+      "epoch": 0.9037263236083027,
+      "grad_norm": 0.11245743930339813,
+      "learning_rate": 0.0001,
+      "loss": 1.6265,
+      "step": 65700
+    },
+    {
+      "epoch": 0.9044140909778676,
+      "grad_norm": 0.0961054340004921,
+      "learning_rate": 0.0001,
+      "loss": 1.6214,
+      "step": 65750
+    },
+    {
+      "epoch": 0.9051018583474326,
+      "grad_norm": 0.10733941197395325,
+      "learning_rate": 0.0001,
+      "loss": 1.6311,
+      "step": 65800
+    },
+    {
+      "epoch": 0.9057896257169975,
+      "grad_norm": 0.11429203301668167,
+      "learning_rate": 0.0001,
+      "loss": 1.6216,
+      "step": 65850
+    },
+    {
+      "epoch": 0.9064773930865624,
+      "grad_norm": 0.10942267626523972,
+      "learning_rate": 0.0001,
+      "loss": 1.6232,
+      "step": 65900
+    },
+    {
+      "epoch": 0.9071651604561273,
+      "grad_norm": 0.11220332235097885,
+      "learning_rate": 0.0001,
+      "loss": 1.6235,
+      "step": 65950
+    },
+    {
+      "epoch": 0.9078529278256923,
+      "grad_norm": 0.10204115509986877,
+      "learning_rate": 0.0001,
+      "loss": 1.6238,
+      "step": 66000
+    },
+    {
+      "epoch": 0.9085406951952572,
+      "grad_norm": 0.09927280992269516,
+      "learning_rate": 0.0001,
+      "loss": 1.6244,
+      "step": 66050
+    },
+    {
+      "epoch": 0.909228462564822,
+      "grad_norm": 0.10095991939306259,
+      "learning_rate": 0.0001,
+      "loss": 1.6236,
+      "step": 66100
+    },
+    {
+      "epoch": 0.909916229934387,
+      "grad_norm": 0.0866779088973999,
+      "learning_rate": 0.0001,
+      "loss": 1.6235,
+      "step": 66150
+    },
+    {
+      "epoch": 0.9106039973039519,
+      "grad_norm": 0.10905203968286514,
+      "learning_rate": 0.0001,
+      "loss": 1.6211,
+      "step": 66200
+    },
+    {
+      "epoch": 0.9112917646735168,
+      "grad_norm": 0.08728614449501038,
+      "learning_rate": 0.0001,
+      "loss": 1.6218,
+      "step": 66250
+    },
+    {
+      "epoch": 0.9119795320430818,
+      "grad_norm": 0.11857550591230392,
+      "learning_rate": 0.0001,
+      "loss": 1.6195,
+      "step": 66300
+    },
+    {
+      "epoch": 0.9126672994126467,
+      "grad_norm": 0.11594309657812119,
+      "learning_rate": 0.0001,
+      "loss": 1.6198,
+      "step": 66350
+    },
+    {
+      "epoch": 0.9133550667822116,
+      "grad_norm": 0.09083879739046097,
+      "learning_rate": 0.0001,
+      "loss": 1.6239,
+      "step": 66400
+    },
+    {
+      "epoch": 0.9140428341517765,
+      "grad_norm": 0.10801181942224503,
+      "learning_rate": 0.0001,
+      "loss": 1.6216,
+      "step": 66450
+    },
+    {
+      "epoch": 0.9147306015213414,
+      "grad_norm": 0.09474363178014755,
+      "learning_rate": 0.0001,
+      "loss": 1.6207,
+      "step": 66500
+    },
+    {
+      "epoch": 0.9154183688909063,
+      "grad_norm": 0.11899229139089584,
+      "learning_rate": 0.0001,
+      "loss": 1.621,
+      "step": 66550
+    },
+    {
+      "epoch": 0.9161061362604712,
+      "grad_norm": 0.09917139261960983,
+      "learning_rate": 0.0001,
+      "loss": 1.6219,
+      "step": 66600
+    },
+    {
+      "epoch": 0.9167939036300362,
+      "grad_norm": 0.09271197021007538,
+      "learning_rate": 0.0001,
+      "loss": 1.616,
+      "step": 66650
+    },
+    {
+      "epoch": 0.9174816709996011,
+      "grad_norm": 0.10327056050300598,
+      "learning_rate": 0.0001,
+      "loss": 1.6216,
+      "step": 66700
+    },
+    {
+      "epoch": 0.918169438369166,
+      "grad_norm": 0.10798952728509903,
+      "learning_rate": 0.0001,
+      "loss": 1.6186,
+      "step": 66750
+    },
+    {
+      "epoch": 0.918857205738731,
+      "grad_norm": 0.09382636100053787,
+      "learning_rate": 0.0001,
+      "loss": 1.6209,
+      "step": 66800
+    },
+    {
+      "epoch": 0.9195449731082959,
+      "grad_norm": 0.09712707996368408,
+      "learning_rate": 0.0001,
+      "loss": 1.6218,
+      "step": 66850
+    },
+    {
+      "epoch": 0.9202327404778607,
+      "grad_norm": 0.09838785231113434,
+      "learning_rate": 0.0001,
+      "loss": 1.6232,
+      "step": 66900
+    },
+    {
+      "epoch": 0.9209205078474257,
+      "grad_norm": 0.09992516040802002,
+      "learning_rate": 0.0001,
+      "loss": 1.6248,
+      "step": 66950
+    },
+    {
+      "epoch": 0.9216082752169906,
+      "grad_norm": 0.10132128745317459,
+      "learning_rate": 0.0001,
+      "loss": 1.6241,
+      "step": 67000
+    },
+    {
+      "epoch": 0.9222960425865555,
+      "grad_norm": 0.11531495302915573,
+      "learning_rate": 0.0001,
+      "loss": 1.6196,
+      "step": 67050
+    },
+    {
+      "epoch": 0.9229838099561204,
+      "grad_norm": 0.1062740609049797,
+      "learning_rate": 0.0001,
+      "loss": 1.6244,
+      "step": 67100
+    },
+    {
+      "epoch": 0.9236715773256854,
+      "grad_norm": 0.10253230482339859,
+      "learning_rate": 0.0001,
+      "loss": 1.621,
+      "step": 67150
+    },
+    {
+      "epoch": 0.9243593446952503,
+      "grad_norm": 0.11413050442934036,
+      "learning_rate": 0.0001,
+      "loss": 1.6257,
+      "step": 67200
+    },
+    {
+      "epoch": 0.9250471120648152,
+      "grad_norm": 0.09013490378856659,
+      "learning_rate": 0.0001,
+      "loss": 1.6228,
+      "step": 67250
+    },
+    {
+      "epoch": 0.9257348794343802,
+      "grad_norm": 0.11905354261398315,
+      "learning_rate": 0.0001,
+      "loss": 1.6247,
+      "step": 67300
+    },
+    {
+      "epoch": 0.926422646803945,
+      "grad_norm": 0.11970172822475433,
+      "learning_rate": 0.0001,
+      "loss": 1.6198,
+      "step": 67350
+    },
+    {
+      "epoch": 0.9271104141735099,
+      "grad_norm": 0.0958150252699852,
+      "learning_rate": 0.0001,
+      "loss": 1.6212,
+      "step": 67400
+    },
+    {
+      "epoch": 0.9277981815430749,
+      "grad_norm": 0.11688381433486938,
+      "learning_rate": 0.0001,
+      "loss": 1.6195,
+      "step": 67450
+    },
+    {
+      "epoch": 0.9284859489126398,
+      "grad_norm": 0.09815829992294312,
+      "learning_rate": 0.0001,
+      "loss": 1.6292,
+      "step": 67500
+    },
+    {
+      "epoch": 0.9291737162822047,
+      "grad_norm": 0.10201198607683182,
+      "learning_rate": 0.0001,
+      "loss": 1.6211,
+      "step": 67550
+    },
+    {
+      "epoch": 0.9298614836517696,
+      "grad_norm": 0.12186852842569351,
+      "learning_rate": 0.0001,
+      "loss": 1.6221,
+      "step": 67600
+    },
+    {
+      "epoch": 0.9305492510213346,
+      "grad_norm": 0.1051778495311737,
+      "learning_rate": 0.0001,
+      "loss": 1.6218,
+      "step": 67650
+    },
+    {
+      "epoch": 0.9312370183908995,
+      "grad_norm": 0.09837380796670914,
+      "learning_rate": 0.0001,
+      "loss": 1.6181,
+      "step": 67700
+    },
+    {
+      "epoch": 0.9319247857604643,
+      "grad_norm": 0.10912145674228668,
+      "learning_rate": 0.0001,
+      "loss": 1.6226,
+      "step": 67750
+    },
+    {
+      "epoch": 0.9326125531300293,
+      "grad_norm": 0.10964906960725784,
+      "learning_rate": 0.0001,
+      "loss": 1.6218,
+      "step": 67800
+    },
+    {
+      "epoch": 0.9333003204995942,
+      "grad_norm": 0.11275553703308105,
+      "learning_rate": 0.0001,
+      "loss": 1.6192,
+      "step": 67850
+    },
+    {
+      "epoch": 0.9339880878691591,
+      "grad_norm": 0.1131332740187645,
+      "learning_rate": 0.0001,
+      "loss": 1.6233,
+      "step": 67900
+    },
+    {
+      "epoch": 0.9346758552387241,
+      "grad_norm": 0.09700637310743332,
+      "learning_rate": 0.0001,
+      "loss": 1.6222,
+      "step": 67950
+    },
+    {
+      "epoch": 0.935363622608289,
+      "grad_norm": 0.10155003517866135,
+      "learning_rate": 0.0001,
+      "loss": 1.6232,
+      "step": 68000
+    },
+    {
+      "epoch": 0.9360513899778539,
+      "grad_norm": 0.09441974759101868,
+      "learning_rate": 0.0001,
+      "loss": 1.6212,
+      "step": 68050
+    },
+    {
+      "epoch": 0.9367391573474189,
+      "grad_norm": 0.10332616418600082,
+      "learning_rate": 0.0001,
+      "loss": 1.6214,
+      "step": 68100
+    },
+    {
+      "epoch": 0.9374269247169837,
+      "grad_norm": 0.10507716983556747,
+      "learning_rate": 0.0001,
+      "loss": 1.6195,
+      "step": 68150
+    },
+    {
+      "epoch": 0.9381146920865486,
+      "grad_norm": 0.12356195598840714,
+      "learning_rate": 0.0001,
+      "loss": 1.6228,
+      "step": 68200
+    },
+    {
+      "epoch": 0.9388024594561135,
+      "grad_norm": 0.09743137657642365,
+      "learning_rate": 0.0001,
+      "loss": 1.6186,
+      "step": 68250
+    },
+    {
+      "epoch": 0.9394902268256785,
+      "grad_norm": 0.09538557380437851,
+      "learning_rate": 0.0001,
+      "loss": 1.6193,
+      "step": 68300
+    },
+    {
+      "epoch": 0.9401779941952434,
+      "grad_norm": 0.10625521838665009,
+      "learning_rate": 0.0001,
+      "loss": 1.6175,
+      "step": 68350
+    },
+    {
+      "epoch": 0.9408657615648083,
+      "grad_norm": 0.09300932288169861,
+      "learning_rate": 0.0001,
+      "loss": 1.6202,
+      "step": 68400
+    },
+    {
+      "epoch": 0.9415535289343733,
+      "grad_norm": 0.10895366966724396,
+      "learning_rate": 0.0001,
+      "loss": 1.6204,
+      "step": 68450
+    },
+    {
+      "epoch": 0.9422412963039382,
+      "grad_norm": 0.09512270987033844,
+      "learning_rate": 0.0001,
+      "loss": 1.6178,
+      "step": 68500
+    },
+    {
+      "epoch": 0.942929063673503,
+      "grad_norm": 0.09526942670345306,
+      "learning_rate": 0.0001,
+      "loss": 1.6204,
+      "step": 68550
+    },
+    {
+      "epoch": 0.943616831043068,
+      "grad_norm": 0.10735663026571274,
+      "learning_rate": 0.0001,
+      "loss": 1.6233,
+      "step": 68600
+    },
+    {
+      "epoch": 0.9443045984126329,
+      "grad_norm": 0.10702252388000488,
+      "learning_rate": 0.0001,
+      "loss": 1.621,
+      "step": 68650
+    },
+    {
+      "epoch": 0.9449923657821978,
+      "grad_norm": 0.12600167095661163,
+      "learning_rate": 0.0001,
+      "loss": 1.6183,
+      "step": 68700
+    },
+    {
+      "epoch": 0.9456801331517628,
+      "grad_norm": 0.10086681693792343,
+      "learning_rate": 0.0001,
+      "loss": 1.6184,
+      "step": 68750
+    },
+    {
+      "epoch": 0.9463679005213277,
+      "grad_norm": 0.10467182099819183,
+      "learning_rate": 0.0001,
+      "loss": 1.6204,
+      "step": 68800
+    },
+    {
+      "epoch": 0.9470556678908926,
+      "grad_norm": 0.10897420346736908,
+      "learning_rate": 0.0001,
+      "loss": 1.621,
+      "step": 68850
+    },
+    {
+      "epoch": 0.9477434352604575,
+      "grad_norm": 0.11620637029409409,
+      "learning_rate": 0.0001,
+      "loss": 1.6191,
+      "step": 68900
+    },
+    {
+      "epoch": 0.9484312026300225,
+      "grad_norm": 0.1035158559679985,
+      "learning_rate": 0.0001,
+      "loss": 1.6221,
+      "step": 68950
+    },
+    {
+      "epoch": 0.9491189699995873,
+      "grad_norm": 0.09212500602006912,
+      "learning_rate": 0.0001,
+      "loss": 1.62,
+      "step": 69000
+    },
+    {
+      "epoch": 0.9498067373691522,
+      "grad_norm": 0.10910465568304062,
+      "learning_rate": 0.0001,
+      "loss": 1.6216,
+      "step": 69050
+    },
+    {
+      "epoch": 0.9504945047387172,
+      "grad_norm": 0.09782394766807556,
+      "learning_rate": 0.0001,
+      "loss": 1.6205,
+      "step": 69100
+    },
+    {
+      "epoch": 0.9511822721082821,
+      "grad_norm": 0.11514250189065933,
+      "learning_rate": 0.0001,
+      "loss": 1.6189,
+      "step": 69150
+    },
+    {
+      "epoch": 0.951870039477847,
+      "grad_norm": 0.10694628953933716,
+      "learning_rate": 0.0001,
+      "loss": 1.6204,
+      "step": 69200
+    },
+    {
+      "epoch": 0.952557806847412,
+      "grad_norm": 0.11289383471012115,
+      "learning_rate": 0.0001,
+      "loss": 1.6208,
+      "step": 69250
+    },
+    {
+      "epoch": 0.9532455742169769,
+      "grad_norm": 0.11638884246349335,
+      "learning_rate": 0.0001,
+      "loss": 1.6226,
+      "step": 69300
+    },
+    {
+      "epoch": 0.9539333415865417,
+      "grad_norm": 0.09971887618303299,
+      "learning_rate": 0.0001,
+      "loss": 1.6199,
+      "step": 69350
+    },
+    {
+      "epoch": 0.9546211089561066,
+      "grad_norm": 0.09802468121051788,
+      "learning_rate": 0.0001,
+      "loss": 1.6177,
+      "step": 69400
+    },
+    {
+      "epoch": 0.9553088763256716,
+      "grad_norm": 0.10239792615175247,
+      "learning_rate": 0.0001,
+      "loss": 1.6176,
+      "step": 69450
+    },
+    {
+      "epoch": 0.9559966436952365,
+      "grad_norm": 0.10659025609493256,
+      "learning_rate": 0.0001,
+      "loss": 1.6243,
+      "step": 69500
+    },
+    {
+      "epoch": 0.9566844110648014,
+      "grad_norm": 0.09827960282564163,
+      "learning_rate": 0.0001,
+      "loss": 1.619,
+      "step": 69550
+    },
+    {
+      "epoch": 0.9573721784343664,
+      "grad_norm": 0.11827285587787628,
+      "learning_rate": 0.0001,
+      "loss": 1.6255,
+      "step": 69600
+    },
+    {
+      "epoch": 0.9580599458039313,
+      "grad_norm": 0.10839015990495682,
+      "learning_rate": 0.0001,
+      "loss": 1.6197,
+      "step": 69650
+    },
+    {
+      "epoch": 0.9587477131734962,
+      "grad_norm": 0.10128850489854813,
+      "learning_rate": 0.0001,
+      "loss": 1.6233,
+      "step": 69700
+    },
+    {
+      "epoch": 0.9594354805430612,
+      "grad_norm": 0.0996512919664383,
+      "learning_rate": 0.0001,
+      "loss": 1.6218,
+      "step": 69750
+    },
+    {
+      "epoch": 0.960123247912626,
+      "grad_norm": 0.11462216824293137,
+      "learning_rate": 0.0001,
+      "loss": 1.6196,
+      "step": 69800
+    },
+    {
+      "epoch": 0.9608110152821909,
+      "grad_norm": 0.10932575911283493,
+      "learning_rate": 0.0001,
+      "loss": 1.6199,
+      "step": 69850
+    },
+    {
+      "epoch": 0.9614987826517559,
+      "grad_norm": 0.12074451148509979,
+      "learning_rate": 0.0001,
+      "loss": 1.6217,
+      "step": 69900
+    },
+    {
+      "epoch": 0.9621865500213208,
+      "grad_norm": 0.10791147500276566,
+      "learning_rate": 0.0001,
+      "loss": 1.6195,
+      "step": 69950
+    },
+    {
+      "epoch": 0.9628743173908857,
+      "grad_norm": 0.09193072468042374,
+      "learning_rate": 0.0001,
+      "loss": 1.6236,
+      "step": 70000
+    },
+    {
+      "epoch": 0.9635620847604506,
+      "grad_norm": 0.09834913909435272,
+      "learning_rate": 0.0001,
+      "loss": 1.6217,
+      "step": 70050
+    },
+    {
+      "epoch": 0.9642498521300156,
+      "grad_norm": 0.10776387155056,
+      "learning_rate": 0.0001,
+      "loss": 1.6194,
+      "step": 70100
+    },
+    {
+      "epoch": 0.9649376194995805,
+      "grad_norm": 0.10709933191537857,
+      "learning_rate": 0.0001,
+      "loss": 1.623,
+      "step": 70150
+    },
+    {
+      "epoch": 0.9656253868691453,
+      "grad_norm": 0.09259042143821716,
+      "learning_rate": 0.0001,
+      "loss": 1.6195,
+      "step": 70200
+    },
+    {
+      "epoch": 0.9663131542387103,
+      "grad_norm": 0.09519805014133453,
+      "learning_rate": 0.0001,
+      "loss": 1.6216,
+      "step": 70250
+    },
+    {
+      "epoch": 0.9670009216082752,
+      "grad_norm": 0.10261819511651993,
+      "learning_rate": 0.0001,
+      "loss": 1.6196,
+      "step": 70300
+    },
+    {
+      "epoch": 0.9676886889778401,
+      "grad_norm": 0.09229016304016113,
+      "learning_rate": 0.0001,
+      "loss": 1.6206,
+      "step": 70350
+    },
+    {
+      "epoch": 0.9683764563474051,
+      "grad_norm": 0.10125798732042313,
+      "learning_rate": 0.0001,
+      "loss": 1.6225,
+      "step": 70400
+    },
+    {
+      "epoch": 0.96906422371697,
+      "grad_norm": 0.0942578986287117,
+      "learning_rate": 0.0001,
+      "loss": 1.6192,
+      "step": 70450
+    },
+    {
+      "epoch": 0.9697519910865349,
+      "grad_norm": 0.09841769933700562,
+      "learning_rate": 0.0001,
+      "loss": 1.6258,
+      "step": 70500
+    },
+    {
+      "epoch": 0.9704397584560998,
+      "grad_norm": 0.10628259927034378,
+      "learning_rate": 0.0001,
+      "loss": 1.6226,
+      "step": 70550
+    },
+    {
+      "epoch": 0.9711275258256648,
+      "grad_norm": 0.10877252370119095,
+      "learning_rate": 0.0001,
+      "loss": 1.6203,
+      "step": 70600
+    },
+    {
+      "epoch": 0.9718152931952296,
+      "grad_norm": 0.1018475666642189,
+      "learning_rate": 0.0001,
+      "loss": 1.6212,
+      "step": 70650
+    },
+    {
+      "epoch": 0.9725030605647945,
+      "grad_norm": 0.09722470492124557,
+      "learning_rate": 0.0001,
+      "loss": 1.6228,
+      "step": 70700
+    },
+    {
+      "epoch": 0.9731908279343595,
+      "grad_norm": 0.10455095767974854,
+      "learning_rate": 0.0001,
+      "loss": 1.6186,
+      "step": 70750
+    },
+    {
+      "epoch": 0.9738785953039244,
+      "grad_norm": 0.0981382504105568,
+      "learning_rate": 0.0001,
+      "loss": 1.6206,
+      "step": 70800
+    },
+    {
+      "epoch": 0.9745663626734893,
+      "grad_norm": 0.09375258535146713,
+      "learning_rate": 0.0001,
+      "loss": 1.6179,
+      "step": 70850
+    },
+    {
+      "epoch": 0.9752541300430543,
+      "grad_norm": 0.11002162843942642,
+      "learning_rate": 0.0001,
+      "loss": 1.6178,
+      "step": 70900
+    },
+    {
+      "epoch": 0.9759418974126192,
+      "grad_norm": 0.09439252316951752,
+      "learning_rate": 0.0001,
+      "loss": 1.6234,
+      "step": 70950
+    },
+    {
+      "epoch": 0.976629664782184,
+      "grad_norm": 0.08569147437810898,
+      "learning_rate": 0.0001,
+      "loss": 1.6218,
+      "step": 71000
+    },
+    {
+      "epoch": 0.977317432151749,
+      "grad_norm": 0.10502757132053375,
+      "learning_rate": 0.0001,
+      "loss": 1.619,
+      "step": 71050
+    },
+    {
+      "epoch": 0.9780051995213139,
+      "grad_norm": 0.09755092114210129,
+      "learning_rate": 0.0001,
+      "loss": 1.6179,
+      "step": 71100
+    },
+    {
+      "epoch": 0.9786929668908788,
+      "grad_norm": 0.09338153153657913,
+      "learning_rate": 0.0001,
+      "loss": 1.6235,
+      "step": 71150
+    },
+    {
+      "epoch": 0.9793807342604437,
+      "grad_norm": 0.13502080738544464,
+      "learning_rate": 0.0001,
+      "loss": 1.6214,
+      "step": 71200
+    },
+    {
+      "epoch": 0.9800685016300087,
+      "grad_norm": 0.10013073682785034,
+      "learning_rate": 0.0001,
+      "loss": 1.6229,
+      "step": 71250
+    },
+    {
+      "epoch": 0.9807562689995736,
+      "grad_norm": 0.0925469771027565,
+      "learning_rate": 0.0001,
+      "loss": 1.6171,
+      "step": 71300
+    },
+    {
+      "epoch": 0.9814440363691385,
+      "grad_norm": 0.09714153409004211,
+      "learning_rate": 0.0001,
+      "loss": 1.6193,
+      "step": 71350
+    },
+    {
+      "epoch": 0.9821318037387035,
+      "grad_norm": 0.09583668410778046,
+      "learning_rate": 0.0001,
+      "loss": 1.6189,
+      "step": 71400
+    },
+    {
+      "epoch": 0.9828195711082683,
+      "grad_norm": 0.10106085985898972,
+      "learning_rate": 0.0001,
+      "loss": 1.6215,
+      "step": 71450
+    },
+    {
+      "epoch": 0.9835073384778332,
+      "grad_norm": 0.122016541659832,
+      "learning_rate": 0.0001,
+      "loss": 1.6184,
+      "step": 71500
+    },
+    {
+      "epoch": 0.9841951058473982,
+      "grad_norm": 0.11579859256744385,
+      "learning_rate": 0.0001,
+      "loss": 1.6214,
+      "step": 71550
+    },
+    {
+      "epoch": 0.9848828732169631,
+      "grad_norm": 0.1063709408044815,
+      "learning_rate": 0.0001,
+      "loss": 1.6196,
+      "step": 71600
+    },
+    {
+      "epoch": 0.985570640586528,
+      "grad_norm": 0.10856927186250687,
+      "learning_rate": 0.0001,
+      "loss": 1.6234,
+      "step": 71650
+    },
+    {
+      "epoch": 0.986258407956093,
+      "grad_norm": 0.10665950924158096,
+      "learning_rate": 0.0001,
+      "loss": 1.6162,
+      "step": 71700
+    },
+    {
+      "epoch": 0.9869461753256579,
+      "grad_norm": 0.10389940440654755,
+      "learning_rate": 0.0001,
+      "loss": 1.6186,
+      "step": 71750
+    },
+    {
+      "epoch": 0.9876339426952228,
+      "grad_norm": 0.09887974709272385,
+      "learning_rate": 0.0001,
+      "loss": 1.6178,
+      "step": 71800
+    },
+    {
+      "epoch": 0.9883217100647876,
+      "grad_norm": 0.0978117361664772,
+      "learning_rate": 0.0001,
+      "loss": 1.6213,
+      "step": 71850
+    },
+    {
+      "epoch": 0.9890094774343526,
+      "grad_norm": 0.09448680281639099,
+      "learning_rate": 0.0001,
+      "loss": 1.6229,
+      "step": 71900
+    },
+    {
+      "epoch": 0.9896972448039175,
+      "grad_norm": 0.10754861682653427,
+      "learning_rate": 0.0001,
+      "loss": 1.6185,
+      "step": 71950
+    },
+    {
+      "epoch": 0.9903850121734824,
+      "grad_norm": 0.11369004845619202,
+      "learning_rate": 0.0001,
+      "loss": 1.6256,
+      "step": 72000
+    },
+    {
+      "epoch": 0.9910727795430474,
+      "grad_norm": 0.10211578011512756,
+      "learning_rate": 0.0001,
+      "loss": 1.6219,
+      "step": 72050
+    },
+    {
+      "epoch": 0.9917605469126123,
+      "grad_norm": 0.09252225607633591,
+      "learning_rate": 0.0001,
+      "loss": 1.6202,
+      "step": 72100
+    },
+    {
+      "epoch": 0.9924483142821772,
+      "grad_norm": 0.11989615112543106,
+      "learning_rate": 0.0001,
+      "loss": 1.624,
+      "step": 72150
+    },
+    {
+      "epoch": 0.9931360816517422,
+      "grad_norm": 0.09846003353595734,
+      "learning_rate": 0.0001,
+      "loss": 1.6173,
+      "step": 72200
+    },
+    {
+      "epoch": 0.993823849021307,
+      "grad_norm": 0.11069589853286743,
+      "learning_rate": 0.0001,
+      "loss": 1.6173,
+      "step": 72250
+    },
+    {
+      "epoch": 0.9945116163908719,
+      "grad_norm": 0.10039959102869034,
+      "learning_rate": 0.0001,
+      "loss": 1.6198,
+      "step": 72300
+    },
+    {
+      "epoch": 0.9951993837604368,
+      "grad_norm": 0.1280982345342636,
+      "learning_rate": 0.0001,
+      "loss": 1.6163,
+      "step": 72350
+    },
+    {
+      "epoch": 0.9958871511300018,
+      "grad_norm": 0.11330771446228027,
+      "learning_rate": 0.0001,
+      "loss": 1.6209,
+      "step": 72400
+    },
+    {
+      "epoch": 0.9965749184995667,
+      "grad_norm": 0.12211450189352036,
+      "learning_rate": 0.0001,
+      "loss": 1.6222,
+      "step": 72450
+    },
+    {
+      "epoch": 0.9972626858691316,
+      "grad_norm": 0.12727168202400208,
+      "learning_rate": 0.0001,
+      "loss": 1.6219,
+      "step": 72500
+    },
+    {
+      "epoch": 0.9979504532386966,
+      "grad_norm": 0.10834234952926636,
+      "learning_rate": 0.0001,
+      "loss": 1.6196,
+      "step": 72550
+    },
+    {
+      "epoch": 0.9986382206082615,
+      "grad_norm": 0.12034707516431808,
+      "learning_rate": 0.0001,
+      "loss": 1.6198,
+      "step": 72600
+    },
+    {
+      "epoch": 0.9993259879778263,
+      "grad_norm": 0.10049953311681747,
+      "learning_rate": 0.0001,
+      "loss": 1.6209,
+      "step": 72650
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 72699,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.156983764564035e+21,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}