diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4592 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.984202967927238,
+  "eval_steps": 500,
+  "global_step": 650,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.007659167065581618,
+      "grad_norm": 5.921403529288989,
+      "learning_rate": 6.153846153846155e-07,
+      "loss": 0.9112,
+      "step": 1
+    },
+    {
+      "epoch": 0.015318334131163236,
+      "grad_norm": 6.161237327775742,
+      "learning_rate": 1.230769230769231e-06,
+      "loss": 0.9517,
+      "step": 2
+    },
+    {
+      "epoch": 0.022977501196744854,
+      "grad_norm": 6.058434736522049,
+      "learning_rate": 1.8461538461538465e-06,
+      "loss": 0.9295,
+      "step": 3
+    },
+    {
+      "epoch": 0.030636668262326472,
+      "grad_norm": 6.016130530129469,
+      "learning_rate": 2.461538461538462e-06,
+      "loss": 0.9235,
+      "step": 4
+    },
+    {
+      "epoch": 0.03829583532790809,
+      "grad_norm": 5.729487082326455,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.9035,
+      "step": 5
+    },
+    {
+      "epoch": 0.04595500239348971,
+      "grad_norm": 4.266411955281039,
+      "learning_rate": 3.692307692307693e-06,
+      "loss": 0.8544,
+      "step": 6
+    },
+    {
+      "epoch": 0.053614169459071326,
+      "grad_norm": 2.4685484993069466,
+      "learning_rate": 4.307692307692308e-06,
+      "loss": 0.8081,
+      "step": 7
+    },
+    {
+      "epoch": 0.061273336524652944,
+      "grad_norm": 2.2509637870140136,
+      "learning_rate": 4.923076923076924e-06,
+      "loss": 0.836,
+      "step": 8
+    },
+    {
+      "epoch": 0.06893250359023456,
+      "grad_norm": 3.6503885662831137,
+      "learning_rate": 5.538461538461539e-06,
+      "loss": 0.7788,
+      "step": 9
+    },
+    {
+      "epoch": 0.07659167065581618,
+      "grad_norm": 4.501975701703904,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.8309,
+      "step": 10
+    },
+    {
+      "epoch": 0.0842508377213978,
+      "grad_norm": 4.370930073810917,
+      "learning_rate": 6.76923076923077e-06,
+      "loss": 0.8224,
+      "step": 11
+    },
+    {
+      "epoch": 0.09191000478697942,
+      "grad_norm": 3.546183001564244,
+      "learning_rate": 7.384615384615386e-06,
+      "loss": 0.7941,
+      "step": 12
+    },
+    {
+      "epoch": 0.09956917185256103,
+      "grad_norm": 3.2331489380759555,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7635,
+      "step": 13
+    },
+    {
+      "epoch": 0.10722833891814265,
+      "grad_norm": 2.5015396395106966,
+      "learning_rate": 8.615384615384617e-06,
+      "loss": 0.7327,
+      "step": 14
+    },
+    {
+      "epoch": 0.11488750598372427,
+      "grad_norm": 1.8441949964024125,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.7094,
+      "step": 15
+    },
+    {
+      "epoch": 0.12254667304930589,
+      "grad_norm": 1.732314176486123,
+      "learning_rate": 9.846153846153848e-06,
+      "loss": 0.6849,
+      "step": 16
+    },
+    {
+      "epoch": 0.1302058401148875,
+      "grad_norm": 1.8608940893991852,
+      "learning_rate": 1.0461538461538463e-05,
+      "loss": 0.6811,
+      "step": 17
+    },
+    {
+      "epoch": 0.13786500718046912,
+      "grad_norm": 1.8077531019190372,
+      "learning_rate": 1.1076923076923079e-05,
+      "loss": 0.6745,
+      "step": 18
+    },
+    {
+      "epoch": 0.14552417424605074,
+      "grad_norm": 1.2577918457624666,
+      "learning_rate": 1.1692307692307694e-05,
+      "loss": 0.6578,
+      "step": 19
+    },
+    {
+      "epoch": 0.15318334131163236,
+      "grad_norm": 1.289096433457239,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.6754,
+      "step": 20
+    },
+    {
+      "epoch": 0.16084250837721398,
+      "grad_norm": 1.4570529134788588,
+      "learning_rate": 1.2923076923076925e-05,
+      "loss": 0.6655,
+      "step": 21
+    },
+    {
+      "epoch": 0.1685016754427956,
+      "grad_norm": 1.1952396245104895,
+      "learning_rate": 1.353846153846154e-05,
+      "loss": 0.64,
+      "step": 22
+    },
+    {
+      "epoch": 0.1761608425083772,
+      "grad_norm": 1.0573908695788548,
+      "learning_rate": 1.4153846153846156e-05,
+      "loss": 0.6226,
+      "step": 23
+    },
+    {
+      "epoch": 0.18382000957395883,
+      "grad_norm": 1.0531289868486842,
+      "learning_rate": 1.4769230769230772e-05,
+      "loss": 0.5999,
+      "step": 24
+    },
+    {
+      "epoch": 0.19147917663954045,
+      "grad_norm": 0.8767976703429683,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.6224,
+      "step": 25
+    },
+    {
+      "epoch": 0.19913834370512207,
+      "grad_norm": 1.0191023286379297,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6382,
+      "step": 26
+    },
+    {
+      "epoch": 0.20679751077070369,
+      "grad_norm": 0.8038756547216995,
+      "learning_rate": 1.6615384615384618e-05,
+      "loss": 0.6216,
+      "step": 27
+    },
+    {
+      "epoch": 0.2144566778362853,
+      "grad_norm": 0.958933004581782,
+      "learning_rate": 1.7230769230769234e-05,
+      "loss": 0.6154,
+      "step": 28
+    },
+    {
+      "epoch": 0.22211584490186692,
+      "grad_norm": 0.9664195885794775,
+      "learning_rate": 1.784615384615385e-05,
+      "loss": 0.6128,
+      "step": 29
+    },
+    {
+      "epoch": 0.22977501196744854,
+      "grad_norm": 0.9729481103689919,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.5937,
+      "step": 30
+    },
+    {
+      "epoch": 0.23743417903303016,
+      "grad_norm": 0.9246652793706185,
+      "learning_rate": 1.907692307692308e-05,
+      "loss": 0.6247,
+      "step": 31
+    },
+    {
+      "epoch": 0.24509334609861178,
+      "grad_norm": 0.9530006352752787,
+      "learning_rate": 1.9692307692307696e-05,
+      "loss": 0.5816,
+      "step": 32
+    },
+    {
+      "epoch": 0.2527525131641934,
+      "grad_norm": 0.8884934035836074,
+      "learning_rate": 2.0307692307692308e-05,
+      "loss": 0.6311,
+      "step": 33
+    },
+    {
+      "epoch": 0.260411680229775,
+      "grad_norm": 0.8458697710494548,
+      "learning_rate": 2.0923076923076927e-05,
+      "loss": 0.6026,
+      "step": 34
+    },
+    {
+      "epoch": 0.26807084729535663,
+      "grad_norm": 0.8241647035938812,
+      "learning_rate": 2.153846153846154e-05,
+      "loss": 0.5813,
+      "step": 35
+    },
+    {
+      "epoch": 0.27573001436093825,
+      "grad_norm": 0.6637335260284863,
+      "learning_rate": 2.2153846153846158e-05,
+      "loss": 0.5826,
+      "step": 36
+    },
+    {
+      "epoch": 0.28338918142651986,
+      "grad_norm": 0.9863751042251347,
+      "learning_rate": 2.276923076923077e-05,
+      "loss": 0.6061,
+      "step": 37
+    },
+    {
+      "epoch": 0.2910483484921015,
+      "grad_norm": 0.7725648338774505,
+      "learning_rate": 2.338461538461539e-05,
+      "loss": 0.5535,
+      "step": 38
+    },
+    {
+      "epoch": 0.2987075155576831,
+      "grad_norm": 0.9197682556877761,
+      "learning_rate": 2.4e-05,
+      "loss": 0.5891,
+      "step": 39
+    },
+    {
+      "epoch": 0.3063666826232647,
+      "grad_norm": 0.9877497296915406,
+      "learning_rate": 2.461538461538462e-05,
+      "loss": 0.575,
+      "step": 40
+    },
+    {
+      "epoch": 0.31402584968884634,
+      "grad_norm": 0.720541520642535,
+      "learning_rate": 2.523076923076923e-05,
+      "loss": 0.5779,
+      "step": 41
+    },
+    {
+      "epoch": 0.32168501675442795,
+      "grad_norm": 0.9225926491059669,
+      "learning_rate": 2.584615384615385e-05,
+      "loss": 0.5528,
+      "step": 42
+    },
+    {
+      "epoch": 0.3293441838200096,
+      "grad_norm": 0.8068833036305191,
+      "learning_rate": 2.6461538461538463e-05,
+      "loss": 0.5888,
+      "step": 43
+    },
+    {
+      "epoch": 0.3370033508855912,
+      "grad_norm": 0.957481417916251,
+      "learning_rate": 2.707692307692308e-05,
+      "loss": 0.5559,
+      "step": 44
+    },
+    {
+      "epoch": 0.3446625179511728,
+      "grad_norm": 0.78572658067857,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 0.5612,
+      "step": 45
+    },
+    {
+      "epoch": 0.3523216850167544,
+      "grad_norm": 0.9209329658610131,
+      "learning_rate": 2.8307692307692312e-05,
+      "loss": 0.584,
+      "step": 46
+    },
+    {
+      "epoch": 0.35998085208233604,
+      "grad_norm": 1.044590316803605,
+      "learning_rate": 2.8923076923076925e-05,
+      "loss": 0.5804,
+      "step": 47
+    },
+    {
+      "epoch": 0.36764001914791766,
+      "grad_norm": 1.172192022892308,
+      "learning_rate": 2.9538461538461543e-05,
+      "loss": 0.5518,
+      "step": 48
+    },
+    {
+      "epoch": 0.3752991862134993,
+      "grad_norm": 1.0766200831878052,
+      "learning_rate": 3.0153846153846155e-05,
+      "loss": 0.5846,
+      "step": 49
+    },
+    {
+      "epoch": 0.3829583532790809,
+      "grad_norm": 1.1041677201522513,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 0.5487,
+      "step": 50
+    },
+    {
+      "epoch": 0.3906175203446625,
+      "grad_norm": 1.0469042529413854,
+      "learning_rate": 3.1384615384615386e-05,
+      "loss": 0.5395,
+      "step": 51
+    },
+    {
+      "epoch": 0.39827668741024413,
+      "grad_norm": 1.0791323635301755,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.5657,
+      "step": 52
+    },
+    {
+      "epoch": 0.40593585447582575,
+      "grad_norm": 0.8078486400788046,
+      "learning_rate": 3.261538461538462e-05,
+      "loss": 0.5399,
+      "step": 53
+    },
+    {
+      "epoch": 0.41359502154140737,
+      "grad_norm": 1.191539346367779,
+      "learning_rate": 3.3230769230769236e-05,
+      "loss": 0.5377,
+      "step": 54
+    },
+    {
+      "epoch": 0.421254188606989,
+      "grad_norm": 0.8698447542598498,
+      "learning_rate": 3.384615384615385e-05,
+      "loss": 0.5551,
+      "step": 55
+    },
+    {
+      "epoch": 0.4289133556725706,
+      "grad_norm": 1.1800967382678582,
+      "learning_rate": 3.446153846153847e-05,
+      "loss": 0.5702,
+      "step": 56
+    },
+    {
+      "epoch": 0.4365725227381522,
+      "grad_norm": 1.220687522421987,
+      "learning_rate": 3.507692307692308e-05,
+      "loss": 0.5552,
+      "step": 57
+    },
+    {
+      "epoch": 0.44423168980373384,
+      "grad_norm": 0.9041645568135938,
+      "learning_rate": 3.56923076923077e-05,
+      "loss": 0.5688,
+      "step": 58
+    },
+    {
+      "epoch": 0.45189085686931546,
+      "grad_norm": 1.5855877495279507,
+      "learning_rate": 3.630769230769231e-05,
+      "loss": 0.5641,
+      "step": 59
+    },
+    {
+      "epoch": 0.4595500239348971,
+      "grad_norm": 0.8216109564331809,
+      "learning_rate": 3.692307692307693e-05,
+      "loss": 0.5536,
+      "step": 60
+    },
+    {
+      "epoch": 0.4672091910004787,
+      "grad_norm": 1.5857353518922366,
+      "learning_rate": 3.753846153846154e-05,
+      "loss": 0.5497,
+      "step": 61
+    },
+    {
+      "epoch": 0.4748683580660603,
+      "grad_norm": 1.0645534816694728,
+      "learning_rate": 3.815384615384616e-05,
+      "loss": 0.5611,
+      "step": 62
+    },
+    {
+      "epoch": 0.48252752513164193,
+      "grad_norm": 1.8572686093355073,
+      "learning_rate": 3.876923076923077e-05,
+      "loss": 0.5703,
+      "step": 63
+    },
+    {
+      "epoch": 0.49018669219722355,
+      "grad_norm": 1.6301218272572524,
+      "learning_rate": 3.938461538461539e-05,
+      "loss": 0.5468,
+      "step": 64
+    },
+    {
+      "epoch": 0.49784585926280517,
+      "grad_norm": 1.1854602191040988,
+      "learning_rate": 4e-05,
+      "loss": 0.564,
+      "step": 65
+    },
+    {
+      "epoch": 0.5055050263283868,
+      "grad_norm": 1.5458244211037195,
+      "learning_rate": 3.999971160550277e-05,
+      "loss": 0.5677,
+      "step": 66
+    },
+    {
+      "epoch": 0.5131641933939685,
+      "grad_norm": 1.2355216531245246,
+      "learning_rate": 3.999884643032821e-05,
+      "loss": 0.5606,
+      "step": 67
+    },
+    {
+      "epoch": 0.52082336045955,
+      "grad_norm": 1.0712313677482361,
+      "learning_rate": 3.9997404499427494e-05,
+      "loss": 0.551,
+      "step": 68
+    },
+    {
+      "epoch": 0.5284825275251317,
+      "grad_norm": 1.4808309855361168,
+      "learning_rate": 3.9995385854385124e-05,
+      "loss": 0.542,
+      "step": 69
+    },
+    {
+      "epoch": 0.5361416945907133,
+      "grad_norm": 0.8373088986931322,
+      "learning_rate": 3.999279055341771e-05,
+      "loss": 0.5255,
+      "step": 70
+    },
+    {
+      "epoch": 0.5438008616562949,
+      "grad_norm": 1.683135657978014,
+      "learning_rate": 3.9989618671372304e-05,
+      "loss": 0.5578,
+      "step": 71
+    },
+    {
+      "epoch": 0.5514600287218765,
+      "grad_norm": 1.1420245332567331,
+      "learning_rate": 3.998587029972423e-05,
+      "loss": 0.5729,
+      "step": 72
+    },
+    {
+      "epoch": 0.5591191957874582,
+      "grad_norm": 1.6242190582145548,
+      "learning_rate": 3.998154554657448e-05,
+      "loss": 0.5621,
+      "step": 73
+    },
+    {
+      "epoch": 0.5667783628530397,
+      "grad_norm": 1.4494865534227146,
+      "learning_rate": 3.997664453664654e-05,
+      "loss": 0.539,
+      "step": 74
+    },
+    {
+      "epoch": 0.5744375299186214,
+      "grad_norm": 1.2416402505339934,
+      "learning_rate": 3.9971167411282835e-05,
+      "loss": 0.5451,
+      "step": 75
+    },
+    {
+      "epoch": 0.582096696984203,
+      "grad_norm": 1.6591387835608409,
+      "learning_rate": 3.996511432844067e-05,
+      "loss": 0.5649,
+      "step": 76
+    },
+    {
+      "epoch": 0.5897558640497846,
+      "grad_norm": 1.2982089198354367,
+      "learning_rate": 3.9958485462687606e-05,
+      "loss": 0.5518,
+      "step": 77
+    },
+    {
+      "epoch": 0.5974150311153662,
+      "grad_norm": 1.3123456841661971,
+      "learning_rate": 3.9951281005196486e-05,
+      "loss": 0.5514,
+      "step": 78
+    },
+    {
+      "epoch": 0.6050741981809479,
+      "grad_norm": 1.189744218079718,
+      "learning_rate": 3.994350116373991e-05,
+      "loss": 0.5408,
+      "step": 79
+    },
+    {
+      "epoch": 0.6127333652465294,
+      "grad_norm": 1.4721204904424363,
+      "learning_rate": 3.9935146162684206e-05,
+      "loss": 0.5507,
+      "step": 80
+    },
+    {
+      "epoch": 0.6203925323121111,
+      "grad_norm": 1.1623645549927197,
+      "learning_rate": 3.9926216242983017e-05,
+      "loss": 0.5569,
+      "step": 81
+    },
+    {
+      "epoch": 0.6280516993776927,
+      "grad_norm": 1.325416617911315,
+      "learning_rate": 3.991671166217031e-05,
+      "loss": 0.5312,
+      "step": 82
+    },
+    {
+      "epoch": 0.6357108664432743,
+      "grad_norm": 1.1877951056068266,
+      "learning_rate": 3.990663269435298e-05,
+      "loss": 0.5644,
+      "step": 83
+    },
+    {
+      "epoch": 0.6433700335088559,
+      "grad_norm": 1.2286620885326376,
+      "learning_rate": 3.989597963020289e-05,
+      "loss": 0.5756,
+      "step": 84
+    },
+    {
+      "epoch": 0.6510292005744376,
+      "grad_norm": 1.1060602604243368,
+      "learning_rate": 3.9884752776948564e-05,
+      "loss": 0.5311,
+      "step": 85
+    },
+    {
+      "epoch": 0.6586883676400191,
+      "grad_norm": 1.0698164056511321,
+      "learning_rate": 3.9872952458366267e-05,
+      "loss": 0.5574,
+      "step": 86
+    },
+    {
+      "epoch": 0.6663475347056008,
+      "grad_norm": 1.1820578926976948,
+      "learning_rate": 3.986057901477069e-05,
+      "loss": 0.5513,
+      "step": 87
+    },
+    {
+      "epoch": 0.6740067017711824,
+      "grad_norm": 0.8169595363079514,
+      "learning_rate": 3.984763280300514e-05,
+      "loss": 0.5483,
+      "step": 88
+    },
+    {
+      "epoch": 0.681665868836764,
+      "grad_norm": 1.379920176811677,
+      "learning_rate": 3.983411419643125e-05,
+      "loss": 0.5892,
+      "step": 89
+    },
+    {
+      "epoch": 0.6893250359023456,
+      "grad_norm": 0.9681470926974101,
+      "learning_rate": 3.982002358491817e-05,
+      "loss": 0.5342,
+      "step": 90
+    },
+    {
+      "epoch": 0.6969842029679273,
+      "grad_norm": 1.119318426323717,
+      "learning_rate": 3.980536137483141e-05,
+      "loss": 0.5552,
+      "step": 91
+    },
+    {
+      "epoch": 0.7046433700335089,
+      "grad_norm": 1.4254561226777611,
+      "learning_rate": 3.9790127989021024e-05,
+      "loss": 0.5336,
+      "step": 92
+    },
+    {
+      "epoch": 0.7123025370990905,
+      "grad_norm": 1.0062543492174385,
+      "learning_rate": 3.9774323866809485e-05,
+      "loss": 0.5601,
+      "step": 93
+    },
+    {
+      "epoch": 0.7199617041646721,
+      "grad_norm": 1.705345700534379,
+      "learning_rate": 3.9757949463978975e-05,
+      "loss": 0.5659,
+      "step": 94
+    },
+    {
+      "epoch": 0.7276208712302538,
+      "grad_norm": 1.1181524317220572,
+      "learning_rate": 3.9741005252758255e-05,
+      "loss": 0.5204,
+      "step": 95
+    },
+    {
+      "epoch": 0.7352800382958353,
+      "grad_norm": 1.6689670685483973,
+      "learning_rate": 3.9723491721809076e-05,
+      "loss": 0.5498,
+      "step": 96
+    },
+    {
+      "epoch": 0.742939205361417,
+      "grad_norm": 1.4295009254142887,
+      "learning_rate": 3.970540937621201e-05,
+      "loss": 0.565,
+      "step": 97
+    },
+    {
+      "epoch": 0.7505983724269986,
+      "grad_norm": 1.4548186946761517,
+      "learning_rate": 3.9686758737451955e-05,
+      "loss": 0.55,
+      "step": 98
+    },
+    {
+      "epoch": 0.7582575394925802,
+      "grad_norm": 1.213694031644175,
+      "learning_rate": 3.966754034340308e-05,
+      "loss": 0.5348,
+      "step": 99
+    },
+    {
+      "epoch": 0.7659167065581618,
+      "grad_norm": 1.0580342997486925,
+      "learning_rate": 3.9647754748313294e-05,
+      "loss": 0.5242,
+      "step": 100
+    },
+    {
+      "epoch": 0.7735758736237435,
+      "grad_norm": 1.0821077174340752,
+      "learning_rate": 3.962740252278827e-05,
+      "loss": 0.542,
+      "step": 101
+    },
+    {
+      "epoch": 0.781235040689325,
+      "grad_norm": 1.2618844489347851,
+      "learning_rate": 3.960648425377499e-05,
+      "loss": 0.5465,
+      "step": 102
+    },
+    {
+      "epoch": 0.7888942077549067,
+      "grad_norm": 0.9576457209595916,
+      "learning_rate": 3.958500054454482e-05,
+      "loss": 0.5206,
+      "step": 103
+    },
+    {
+      "epoch": 0.7965533748204883,
+      "grad_norm": 1.6879319110317588,
+      "learning_rate": 3.9562952014676116e-05,
+      "loss": 0.5554,
+      "step": 104
+    },
+    {
+      "epoch": 0.8042125418860699,
+      "grad_norm": 1.3708595612461836,
+      "learning_rate": 3.954033930003634e-05,
+      "loss": 0.5745,
+      "step": 105
+    },
+    {
+      "epoch": 0.8118717089516515,
+      "grad_norm": 1.2986325728462578,
+      "learning_rate": 3.9517163052763756e-05,
+      "loss": 0.5413,
+      "step": 106
+    },
+    {
+      "epoch": 0.8195308760172332,
+      "grad_norm": 1.1532883579115862,
+      "learning_rate": 3.9493423941248564e-05,
+      "loss": 0.5367,
+      "step": 107
+    },
+    {
+      "epoch": 0.8271900430828147,
+      "grad_norm": 1.0800362940814041,
+      "learning_rate": 3.946912265011368e-05,
+      "loss": 0.5643,
+      "step": 108
+    },
+    {
+      "epoch": 0.8348492101483964,
+      "grad_norm": 1.0542424383185787,
+      "learning_rate": 3.944425988019498e-05,
+      "loss": 0.5434,
+      "step": 109
+    },
+    {
+      "epoch": 0.842508377213978,
+      "grad_norm": 0.9223839325666725,
+      "learning_rate": 3.9418836348521045e-05,
+      "loss": 0.5313,
+      "step": 110
+    },
+    {
+      "epoch": 0.8501675442795597,
+      "grad_norm": 0.9786970098640274,
+      "learning_rate": 3.9392852788292556e-05,
+      "loss": 0.5257,
+      "step": 111
+    },
+    {
+      "epoch": 0.8578267113451412,
+      "grad_norm": 0.8883004076978934,
+      "learning_rate": 3.936630994886109e-05,
+      "loss": 0.5315,
+      "step": 112
+    },
+    {
+      "epoch": 0.8654858784107229,
+      "grad_norm": 0.9210396895221461,
+      "learning_rate": 3.933920859570753e-05,
+      "loss": 0.5332,
+      "step": 113
+    },
+    {
+      "epoch": 0.8731450454763044,
+      "grad_norm": 1.0300033230225123,
+      "learning_rate": 3.931154951041998e-05,
+      "loss": 0.5364,
+      "step": 114
+    },
+    {
+      "epoch": 0.8808042125418861,
+      "grad_norm": 1.07072986303729,
+      "learning_rate": 3.928333349067125e-05,
+      "loss": 0.5581,
+      "step": 115
+    },
+    {
+      "epoch": 0.8884633796074677,
+      "grad_norm": 0.7856322271677167,
+      "learning_rate": 3.925456135019582e-05,
+      "loss": 0.5153,
+      "step": 116
+    },
+    {
+      "epoch": 0.8961225466730494,
+      "grad_norm": 1.3208925243839957,
+      "learning_rate": 3.922523391876638e-05,
+      "loss": 0.5397,
+      "step": 117
+    },
+    {
+      "epoch": 0.9037817137386309,
+      "grad_norm": 1.0321781910738392,
+      "learning_rate": 3.9195352042169924e-05,
+      "loss": 0.5434,
+      "step": 118
+    },
+    {
+      "epoch": 0.9114408808042126,
+      "grad_norm": 1.2263516987444574,
+      "learning_rate": 3.916491658218333e-05,
+      "loss": 0.5412,
+      "step": 119
+    },
+    {
+      "epoch": 0.9191000478697942,
+      "grad_norm": 1.0824446357123727,
+      "learning_rate": 3.913392841654851e-05,
+      "loss": 0.5345,
+      "step": 120
+    },
+    {
+      "epoch": 0.9267592149353758,
+      "grad_norm": 0.9492587544198882,
+      "learning_rate": 3.9102388438947104e-05,
+      "loss": 0.5374,
+      "step": 121
+    },
+    {
+      "epoch": 0.9344183820009574,
+      "grad_norm": 1.200654956363631,
+      "learning_rate": 3.907029755897473e-05,
+      "loss": 0.528,
+      "step": 122
+    },
+    {
+      "epoch": 0.9420775490665391,
+      "grad_norm": 0.8940904112935483,
+      "learning_rate": 3.903765670211469e-05,
+      "loss": 0.5432,
+      "step": 123
+    },
+    {
+      "epoch": 0.9497367161321206,
+      "grad_norm": 1.513206646677525,
+      "learning_rate": 3.9004466809711343e-05,
+      "loss": 0.55,
+      "step": 124
+    },
+    {
+      "epoch": 0.9573958831977023,
+      "grad_norm": 0.9181110535142029,
+      "learning_rate": 3.897072883894291e-05,
+      "loss": 0.5322,
+      "step": 125
+    },
+    {
+      "epoch": 0.9650550502632839,
+      "grad_norm": 1.493130538955249,
+      "learning_rate": 3.893644376279392e-05,
+      "loss": 0.5118,
+      "step": 126
+    },
+    {
+      "epoch": 0.9727142173288655,
+      "grad_norm": 1.199837488596111,
+      "learning_rate": 3.89016125700271e-05,
+      "loss": 0.5279,
+      "step": 127
+    },
+    {
+      "epoch": 0.9803733843944471,
+      "grad_norm": 1.7015944265286984,
+      "learning_rate": 3.8866236265154864e-05,
+      "loss": 0.5464,
+      "step": 128
+    },
+    {
+      "epoch": 0.9880325514600288,
+      "grad_norm": 1.4588454807007443,
+      "learning_rate": 3.88303158684104e-05,
+      "loss": 0.5544,
+      "step": 129
+    },
+    {
+      "epoch": 0.9956917185256103,
+      "grad_norm": 1.4514845707429012,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.5686,
+      "step": 130
+    },
+    {
+      "epoch": 1.0047869794159885,
+      "grad_norm": 2.2360445683138384,
+      "learning_rate": 3.875684695866409e-05,
+      "loss": 0.8604,
+      "step": 131
+    },
+    {
+      "epoch": 1.0124461464815702,
+      "grad_norm": 1.085982951452546,
+      "learning_rate": 3.871930056446518e-05,
+      "loss": 0.4109,
+      "step": 132
+    },
+    {
+      "epoch": 1.0201053135471518,
+      "grad_norm": 0.9862931565568734,
+      "learning_rate": 3.8681214315938786e-05,
+      "loss": 0.475,
+      "step": 133
+    },
+    {
+      "epoch": 1.0277644806127333,
+      "grad_norm": 1.33743772292278,
+      "learning_rate": 3.864258931147136e-05,
+      "loss": 0.4306,
+      "step": 134
+    },
+    {
+      "epoch": 1.035423647678315,
+      "grad_norm": 1.0921525549828976,
+      "learning_rate": 3.860342666498677e-05,
+      "loss": 0.4848,
+      "step": 135
+    },
+    {
+      "epoch": 1.0430828147438966,
+      "grad_norm": 1.2645353404433375,
+      "learning_rate": 3.856372750591419e-05,
+      "loss": 0.4485,
+      "step": 136
+    },
+    {
+      "epoch": 1.0507419818094783,
+      "grad_norm": 1.1984966128303223,
+      "learning_rate": 3.8523492979155534e-05,
+      "loss": 0.4808,
+      "step": 137
+    },
+    {
+      "epoch": 1.0584011488750598,
+      "grad_norm": 1.0500948118376512,
+      "learning_rate": 3.84827242450524e-05,
+      "loss": 0.4881,
+      "step": 138
+    },
+    {
+      "epoch": 1.0660603159406414,
+      "grad_norm": 0.746032471414876,
+      "learning_rate": 3.844142247935265e-05,
+      "loss": 0.4028,
+      "step": 139
+    },
+    {
+      "epoch": 1.073719483006223,
+      "grad_norm": 0.9623091350387484,
+      "learning_rate": 3.839958887317649e-05,
+      "loss": 0.4715,
+      "step": 140
+    },
+    {
+      "epoch": 1.0813786500718048,
+      "grad_norm": 0.7201641889161347,
+      "learning_rate": 3.835722463298208e-05,
+      "loss": 0.4606,
+      "step": 141
+    },
+    {
+      "epoch": 1.0890378171373862,
+      "grad_norm": 0.8943620531687253,
+      "learning_rate": 3.831433098053082e-05,
+      "loss": 0.4712,
+      "step": 142
+    },
+    {
+      "epoch": 1.096696984202968,
+      "grad_norm": 0.7046882034899636,
+      "learning_rate": 3.827090915285202e-05,
+      "loss": 0.4458,
+      "step": 143
+    },
+    {
+      "epoch": 1.1043561512685496,
+      "grad_norm": 0.6548369768890453,
+      "learning_rate": 3.8226960402207316e-05,
+      "loss": 0.3904,
+      "step": 144
+    },
+    {
+      "epoch": 1.1120153183341313,
+      "grad_norm": 0.8530160058379846,
+      "learning_rate": 3.818248599605448e-05,
+      "loss": 0.546,
+      "step": 145
+    },
+    {
+      "epoch": 1.1196744853997127,
+      "grad_norm": 0.8081896528154348,
+      "learning_rate": 3.813748721701091e-05,
+      "loss": 0.4651,
+      "step": 146
+    },
+    {
+      "epoch": 1.1273336524652944,
+      "grad_norm": 0.7600777048606053,
+      "learning_rate": 3.809196536281665e-05,
+      "loss": 0.3961,
+      "step": 147
+    },
+    {
+      "epoch": 1.134992819530876,
+      "grad_norm": 0.8170599343972957,
+      "learning_rate": 3.80459217462969e-05,
+      "loss": 0.4859,
+      "step": 148
+    },
+    {
+      "epoch": 1.1426519865964577,
+      "grad_norm": 0.5981126204379706,
+      "learning_rate": 3.799935769532425e-05,
+      "loss": 0.4112,
+      "step": 149
+    },
+    {
+      "epoch": 1.1503111536620392,
+      "grad_norm": 0.6918899342824092,
+      "learning_rate": 3.795227455278029e-05,
+      "loss": 0.445,
+      "step": 150
+    },
+    {
+      "epoch": 1.1579703207276208,
+      "grad_norm": 0.7615214743583981,
+      "learning_rate": 3.790467367651694e-05,
+      "loss": 0.5105,
+      "step": 151
+    },
+    {
+      "epoch": 1.1656294877932025,
+      "grad_norm": 0.5625993273766481,
+      "learning_rate": 3.785655643931728e-05,
+      "loss": 0.3587,
+      "step": 152
+    },
+    {
+      "epoch": 1.1732886548587842,
+      "grad_norm": 0.9402358775980364,
+      "learning_rate": 3.780792422885597e-05,
+      "loss": 0.5268,
+      "step": 153
+    },
+    {
+      "epoch": 1.1809478219243656,
+      "grad_norm": 1.0131842602903653,
+      "learning_rate": 3.7758778447659184e-05,
+      "loss": 0.4696,
+      "step": 154
+    },
+    {
+      "epoch": 1.1886069889899473,
+      "grad_norm": 0.7498386211674599,
+      "learning_rate": 3.7709120513064196e-05,
+      "loss": 0.4471,
+      "step": 155
+    },
+    {
+      "epoch": 1.196266156055529,
+      "grad_norm": 0.7182193922598371,
+      "learning_rate": 3.7658951857178544e-05,
+      "loss": 0.4344,
+      "step": 156
+    },
+    {
+      "epoch": 1.2039253231211107,
+      "grad_norm": 0.6854353371665008,
+      "learning_rate": 3.760827392683863e-05,
+      "loss": 0.4195,
+      "step": 157
+    },
+    {
+      "epoch": 1.2115844901866921,
+      "grad_norm": 1.1427693762223539,
+      "learning_rate": 3.755708818356809e-05,
+      "loss": 0.5111,
+      "step": 158
+    },
+    {
+      "epoch": 1.2192436572522738,
+      "grad_norm": 0.6625293413220111,
+      "learning_rate": 3.75053961035356e-05,
+      "loss": 0.4288,
+      "step": 159
+    },
+    {
+      "epoch": 1.2269028243178555,
+      "grad_norm": 0.9123675189917038,
+      "learning_rate": 3.745319917751229e-05,
+      "loss": 0.4656,
+      "step": 160
+    },
+    {
+      "epoch": 1.2345619913834371,
+      "grad_norm": 0.7341125406296108,
+      "learning_rate": 3.740049891082879e-05,
+      "loss": 0.425,
+      "step": 161
+    },
+    {
+      "epoch": 1.2422211584490186,
+      "grad_norm": 0.8631406479236612,
+      "learning_rate": 3.734729682333179e-05,
+      "loss": 0.4911,
+      "step": 162
+    },
+    {
+      "epoch": 1.2498803255146003,
+      "grad_norm": 0.7496345502817138,
+      "learning_rate": 3.729359444934022e-05,
+      "loss": 0.3874,
+      "step": 163
+    },
+    {
+      "epoch": 1.257539492580182,
+      "grad_norm": 0.7894969941658941,
+      "learning_rate": 3.723939333760099e-05,
+      "loss": 0.4578,
+      "step": 164
+    },
+    {
+      "epoch": 1.2651986596457636,
+      "grad_norm": 0.9223288305001206,
+      "learning_rate": 3.718469505124434e-05,
+      "loss": 0.523,
+      "step": 165
+    },
+    {
+      "epoch": 1.272857826711345,
+      "grad_norm": 0.7992363424810383,
+      "learning_rate": 3.712950116773875e-05,
+      "loss": 0.46,
+      "step": 166
+    },
+    {
+      "epoch": 1.2805169937769267,
+      "grad_norm": 0.8478230607138995,
+      "learning_rate": 3.707381327884545e-05,
+      "loss": 0.4749,
+      "step": 167
+    },
+    {
+      "epoch": 1.2881761608425084,
+      "grad_norm": 0.5454929496812748,
+      "learning_rate": 3.70176329905725e-05,
+      "loss": 0.4463,
+      "step": 168
+    },
+    {
+      "epoch": 1.2958353279080899,
+      "grad_norm": 0.750962890324772,
+      "learning_rate": 3.696096192312852e-05,
+      "loss": 0.4562,
+      "step": 169
+    },
+    {
+      "epoch": 1.3034944949736715,
+      "grad_norm": 0.6464014219567544,
+      "learning_rate": 3.69038017108759e-05,
+      "loss": 0.458,
+      "step": 170
+    },
+    {
+      "epoch": 1.3111536620392532,
+      "grad_norm": 0.6483521349636168,
+      "learning_rate": 3.6846154002283696e-05,
+      "loss": 0.4709,
+      "step": 171
+    },
+    {
+      "epoch": 1.3188128291048349,
+      "grad_norm": 0.7936957658021746,
+      "learning_rate": 3.678802045988012e-05,
+      "loss": 0.4581,
+      "step": 172
+    },
+    {
+      "epoch": 1.3264719961704166,
+      "grad_norm": 0.9116293942129717,
+      "learning_rate": 3.6729402760204535e-05,
+      "loss": 0.4367,
+      "step": 173
+    },
+    {
+      "epoch": 1.334131163235998,
+      "grad_norm": 0.8432492215433733,
+      "learning_rate": 3.667030259375915e-05,
+      "loss": 0.5148,
+      "step": 174
+    },
+    {
+      "epoch": 1.3417903303015797,
+      "grad_norm": 0.7438248026517662,
+      "learning_rate": 3.6610721664960236e-05,
+      "loss": 0.4653,
+      "step": 175
+    },
+    {
+      "epoch": 1.3494494973671614,
+      "grad_norm": 0.6889892205036577,
+      "learning_rate": 3.6550661692089e-05,
+      "loss": 0.4231,
+      "step": 176
+    },
+    {
+      "epoch": 1.3571086644327428,
+      "grad_norm": 0.7784374117922759,
+      "learning_rate": 3.6490124407242007e-05,
+      "loss": 0.5052,
+      "step": 177
+    },
+    {
+      "epoch": 1.3647678314983245,
+      "grad_norm": 0.8790420908019635,
+      "learning_rate": 3.642911155628124e-05,
+      "loss": 0.4415,
+      "step": 178
+    },
+    {
+      "epoch": 1.3724269985639062,
+      "grad_norm": 0.7174987210830999,
+      "learning_rate": 3.636762489878374e-05,
+      "loss": 0.4421,
+      "step": 179
+    },
+    {
+      "epoch": 1.3800861656294878,
+      "grad_norm": 0.8020599752892646,
+      "learning_rate": 3.6305666207990886e-05,
+      "loss": 0.4863,
+      "step": 180
+    },
+    {
+      "epoch": 1.3877453326950695,
+      "grad_norm": 0.7802012248909737,
+      "learning_rate": 3.624323727075723e-05,
+      "loss": 0.4426,
+      "step": 181
+    },
+    {
+      "epoch": 1.395404499760651,
+      "grad_norm": 0.671987752263627,
+      "learning_rate": 3.6180339887498953e-05,
+      "loss": 0.4077,
+      "step": 182
+    },
+    {
+      "epoch": 1.4030636668262326,
+      "grad_norm": 0.9587427415971348,
+      "learning_rate": 3.6116975872141984e-05,
+      "loss": 0.5101,
+      "step": 183
+    },
+    {
+      "epoch": 1.4107228338918143,
+      "grad_norm": 0.9118291287887855,
+      "learning_rate": 3.605314705206966e-05,
+      "loss": 0.4823,
+      "step": 184
+    },
+    {
+      "epoch": 1.4183820009573957,
+      "grad_norm": 0.8693454740841087,
+      "learning_rate": 3.598885526807003e-05,
+      "loss": 0.4712,
+      "step": 185
+    },
+    {
+      "epoch": 1.4260411680229774,
+      "grad_norm": 0.7409071346511643,
+      "learning_rate": 3.5924102374282754e-05,
+      "loss": 0.4422,
+      "step": 186
+    },
+    {
+      "epoch": 1.433700335088559,
+      "grad_norm": 0.9303869755499337,
+      "learning_rate": 3.5858890238145674e-05,
+      "loss": 0.4207,
+      "step": 187
+    },
+    {
+      "epoch": 1.4413595021541408,
+      "grad_norm": 0.852020974421498,
+      "learning_rate": 3.5793220740340904e-05,
+      "loss": 0.4927,
+      "step": 188
+    },
+    {
+      "epoch": 1.4490186692197224,
+      "grad_norm": 0.7715289615754756,
+      "learning_rate": 3.572709577474062e-05,
+      "loss": 0.4546,
+      "step": 189
+    },
+    {
+      "epoch": 1.456677836285304,
+      "grad_norm": 0.7648123067206034,
+      "learning_rate": 3.566051724835245e-05,
+      "loss": 0.4775,
+      "step": 190
+    },
+    {
+      "epoch": 1.4643370033508856,
+      "grad_norm": 0.6233854224617866,
+      "learning_rate": 3.559348708126445e-05,
+      "loss": 0.4428,
+      "step": 191
+    },
+    {
+      "epoch": 1.4719961704164672,
+      "grad_norm": 0.7266153094875722,
+      "learning_rate": 3.552600720658976e-05,
+      "loss": 0.4775,
+      "step": 192
+    },
+    {
+      "epoch": 1.4796553374820487,
+      "grad_norm": 0.6034805115192634,
+      "learning_rate": 3.545807957041084e-05,
+      "loss": 0.4353,
+      "step": 193
+    },
+    {
+      "epoch": 1.4873145045476304,
+      "grad_norm": 0.733938783383156,
+      "learning_rate": 3.538970613172332e-05,
+      "loss": 0.492,
+      "step": 194
+    },
+    {
+      "epoch": 1.494973671613212,
+      "grad_norm": 0.6587084782665951,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.4471,
+      "step": 195
+    },
+    {
+      "epoch": 1.5026328386787937,
+      "grad_norm": 0.7660206322795097,
+      "learning_rate": 3.525162974703174e-05,
+      "loss": 0.4278,
+      "step": 196
+    },
+    {
+      "epoch": 1.5102920057443754,
+      "grad_norm": 0.6137621338422651,
+      "learning_rate": 3.518193078307463e-05,
+      "loss": 0.4325,
+      "step": 197
+    },
+    {
+      "epoch": 1.517951172809957,
+      "grad_norm": 0.5768603184148955,
+      "learning_rate": 3.5111793980588006e-05,
+      "loss": 0.4529,
+      "step": 198
+    },
+    {
+      "epoch": 1.5256103398755385,
+      "grad_norm": 0.7888259317826345,
+      "learning_rate": 3.5041221362278644e-05,
+      "loss": 0.44,
+      "step": 199
+    },
+    {
+      "epoch": 1.5332695069411202,
+      "grad_norm": 0.5650383799794519,
+      "learning_rate": 3.497021496342203e-05,
+      "loss": 0.4585,
+      "step": 200
+    },
+    {
+      "epoch": 1.5409286740067016,
+      "grad_norm": 0.705765230853562,
+      "learning_rate": 3.489877683180362e-05,
+      "loss": 0.4616,
+      "step": 201
+    },
+    {
+      "epoch": 1.5485878410722833,
+      "grad_norm": 0.6873526187506311,
+      "learning_rate": 3.482690902765984e-05,
+      "loss": 0.4433,
+      "step": 202
+    },
+    {
+      "epoch": 1.556247008137865,
+      "grad_norm": 0.8007803368306625,
+      "learning_rate": 3.475461362361861e-05,
+      "loss": 0.5263,
+      "step": 203
+    },
+    {
+      "epoch": 1.5639061752034467,
+      "grad_norm": 0.6918325743424758,
+      "learning_rate": 3.468189270463959e-05,
+      "loss": 0.4325,
+      "step": 204
+    },
+    {
+      "epoch": 1.5715653422690283,
+      "grad_norm": 0.5677928436218839,
+      "learning_rate": 3.4608748367954064e-05,
+      "loss": 0.4654,
+      "step": 205
+    },
+    {
+      "epoch": 1.57922450933461,
+      "grad_norm": 0.7751869494065315,
+      "learning_rate": 3.4535182723004466e-05,
+      "loss": 0.461,
+      "step": 206
+    },
+    {
+      "epoch": 1.5868836764001915,
+      "grad_norm": 0.5725206135864617,
+      "learning_rate": 3.446119789138351e-05,
+      "loss": 0.4952,
+      "step": 207
+    },
+    {
+      "epoch": 1.5945428434657731,
+      "grad_norm": 0.7296776133168584,
+      "learning_rate": 3.438679600677303e-05,
+      "loss": 0.4313,
+      "step": 208
+    },
+    {
+      "epoch": 1.6022020105313546,
+      "grad_norm": 0.587167588865063,
+      "learning_rate": 3.431197921488242e-05,
+      "loss": 0.433,
+      "step": 209
+    },
+    {
+      "epoch": 1.6098611775969363,
+      "grad_norm": 0.8056341147184974,
+      "learning_rate": 3.423674967338681e-05,
+      "loss": 0.4514,
+      "step": 210
+    },
+    {
+      "epoch": 1.617520344662518,
+      "grad_norm": 0.7051970461447443,
+      "learning_rate": 3.416110955186477e-05,
+      "loss": 0.5431,
+      "step": 211
+    },
+    {
+      "epoch": 1.6251795117280996,
+      "grad_norm": 0.5607663565934765,
+      "learning_rate": 3.4085061031735794e-05,
+      "loss": 0.4408,
+      "step": 212
+    },
+    {
+      "epoch": 1.6328386787936813,
+      "grad_norm": 0.8246463721021458,
+      "learning_rate": 3.4008606306197336e-05,
+      "loss": 0.4864,
+      "step": 213
+    },
+    {
+      "epoch": 1.640497845859263,
+      "grad_norm": 0.6500034356382017,
+      "learning_rate": 3.393174758016161e-05,
+      "loss": 0.4215,
+      "step": 214
+    },
+    {
+      "epoch": 1.6481570129248444,
+      "grad_norm": 0.7062452977875294,
+      "learning_rate": 3.385448707019199e-05,
+      "loss": 0.4731,
+      "step": 215
+    },
+    {
+      "epoch": 1.655816179990426,
+      "grad_norm": 0.5942221349239608,
+      "learning_rate": 3.377682700443907e-05,
+      "loss": 0.4164,
+      "step": 216
+    },
+    {
+      "epoch": 1.6634753470560075,
+      "grad_norm": 0.7905876229344655,
+      "learning_rate": 3.3698769622576404e-05,
+      "loss": 0.5147,
+      "step": 217
+    },
+    {
+      "epoch": 1.6711345141215892,
+      "grad_norm": 0.5859170654910125,
+      "learning_rate": 3.3620317175735945e-05,
+      "loss": 0.4275,
+      "step": 218
+    },
+    {
+      "epoch": 1.6787936811871709,
+      "grad_norm": 0.6690808146840328,
+      "learning_rate": 3.3541471926443084e-05,
+      "loss": 0.4789,
+      "step": 219
+    },
+    {
+      "epoch": 1.6864528482527525,
+      "grad_norm": 0.6561200573249233,
+      "learning_rate": 3.34622361485514e-05,
+      "loss": 0.4991,
+      "step": 220
+    },
+    {
+      "epoch": 1.6941120153183342,
+      "grad_norm": 0.5562256326390935,
+      "learning_rate": 3.3382612127177166e-05,
+      "loss": 0.4343,
+      "step": 221
+    },
+    {
+      "epoch": 1.701771182383916,
+      "grad_norm": 0.7170323098042807,
+      "learning_rate": 3.330260215863332e-05,
+      "loss": 0.4486,
+      "step": 222
+    },
+    {
+      "epoch": 1.7094303494494973,
+      "grad_norm": 0.6014886240988541,
+      "learning_rate": 3.322220855036333e-05,
+      "loss": 0.4957,
+      "step": 223
+    },
+    {
+      "epoch": 1.717089516515079,
+      "grad_norm": 0.5278357930923132,
+      "learning_rate": 3.314143362087462e-05,
+      "loss": 0.442,
+      "step": 224
+    },
+    {
+      "epoch": 1.7247486835806605,
+      "grad_norm": 0.6035858404605409,
+      "learning_rate": 3.30602796996717e-05,
+      "loss": 0.4506,
+      "step": 225
+    },
+    {
+      "epoch": 1.7324078506462421,
+      "grad_norm": 0.6363716300860364,
+      "learning_rate": 3.297874912718902e-05,
+      "loss": 0.4663,
+      "step": 226
+    },
+    {
+      "epoch": 1.7400670177118238,
+      "grad_norm": 0.47737502299597623,
+      "learning_rate": 3.2896844254723414e-05,
+      "loss": 0.418,
+      "step": 227
+    },
+    {
+      "epoch": 1.7477261847774055,
+      "grad_norm": 0.8067471733349434,
+      "learning_rate": 3.281456744436634e-05,
+      "loss": 0.4957,
+      "step": 228
+    },
+    {
+      "epoch": 1.7553853518429872,
+      "grad_norm": 0.6089284826032468,
+      "learning_rate": 3.273192106893572e-05,
+      "loss": 0.4418,
+      "step": 229
+    },
+    {
+      "epoch": 1.7630445189085688,
+      "grad_norm": 0.5962202659179022,
+      "learning_rate": 3.2648907511907544e-05,
+      "loss": 0.4669,
+      "step": 230
+    },
+    {
+      "epoch": 1.7707036859741503,
+      "grad_norm": 0.5728200467314639,
+      "learning_rate": 3.256552916734713e-05,
+      "loss": 0.4678,
+      "step": 231
+    },
+    {
+      "epoch": 1.778362853039732,
+      "grad_norm": 0.541716837396919,
+      "learning_rate": 3.248178843984006e-05,
+      "loss": 0.3918,
+      "step": 232
+    },
+    {
+      "epoch": 1.7860220201053134,
+      "grad_norm": 0.6534786961268279,
+      "learning_rate": 3.239768774442281e-05,
+      "loss": 0.4922,
+      "step": 233
+    },
+    {
+      "epoch": 1.793681187170895,
+      "grad_norm": 0.8358591875149838,
+      "learning_rate": 3.2313229506513167e-05,
+      "loss": 0.4902,
+      "step": 234
+    },
+    {
+      "epoch": 1.8013403542364768,
+      "grad_norm": 0.591595847556571,
+      "learning_rate": 3.222841616184025e-05,
+      "loss": 0.4138,
+      "step": 235
+    },
+    {
+      "epoch": 1.8089995213020584,
+      "grad_norm": 0.6277241932218716,
+      "learning_rate": 3.2143250156374226e-05,
+      "loss": 0.4474,
+      "step": 236
+    },
+    {
+      "epoch": 1.81665868836764,
+      "grad_norm": 0.7278025713494423,
+      "learning_rate": 3.2057733946255844e-05,
+      "loss": 0.4709,
+      "step": 237
+    },
+    {
+      "epoch": 1.8243178554332218,
+      "grad_norm": 0.6410859956043007,
+      "learning_rate": 3.197186999772555e-05,
+      "loss": 0.4583,
+      "step": 238
+    },
+    {
+      "epoch": 1.8319770224988032,
+      "grad_norm": 0.6757146265800846,
+      "learning_rate": 3.188566078705235e-05,
+      "loss": 0.4704,
+      "step": 239
+    },
+    {
+      "epoch": 1.839636189564385,
+      "grad_norm": 0.7572042816661453,
+      "learning_rate": 3.1799108800462466e-05,
+      "loss": 0.4119,
+      "step": 240
+    },
+    {
+      "epoch": 1.8472953566299664,
+      "grad_norm": 0.590906796662085,
+      "learning_rate": 3.1712216534067536e-05,
+      "loss": 0.47,
+      "step": 241
+    },
+    {
+      "epoch": 1.854954523695548,
+      "grad_norm": 0.8241586342465295,
+      "learning_rate": 3.1624986493792735e-05,
+      "loss": 0.4624,
+      "step": 242
+    },
+    {
+      "epoch": 1.8626136907611297,
+      "grad_norm": 0.5604470218245483,
+      "learning_rate": 3.153742119530441e-05,
+      "loss": 0.4831,
+      "step": 243
+    },
+    {
+      "epoch": 1.8702728578267114,
+      "grad_norm": 0.8158955643638057,
+      "learning_rate": 3.144952316393758e-05,
+      "loss": 0.4204,
+      "step": 244
+    },
+    {
+      "epoch": 1.877932024892293,
+      "grad_norm": 0.5828232072086235,
+      "learning_rate": 3.136129493462312e-05,
+      "loss": 0.4646,
+      "step": 245
+    },
+    {
+      "epoch": 1.8855911919578747,
+      "grad_norm": 0.7063142478848118,
+      "learning_rate": 3.1272739051814594e-05,
+      "loss": 0.4576,
+      "step": 246
+    },
+    {
+      "epoch": 1.8932503590234562,
+      "grad_norm": 0.6560131266274781,
+      "learning_rate": 3.1183858069414936e-05,
+      "loss": 0.4537,
+      "step": 247
+    },
+    {
+      "epoch": 1.9009095260890378,
+      "grad_norm": 0.5709968540583878,
+      "learning_rate": 3.109465455070278e-05,
+      "loss": 0.4559,
+      "step": 248
+    },
+    {
+      "epoch": 1.9085686931546193,
+      "grad_norm": 0.6268646820645886,
+      "learning_rate": 3.1005131068258506e-05,
+      "loss": 0.3999,
+      "step": 249
+    },
+    {
+      "epoch": 1.916227860220201,
+      "grad_norm": 0.48362971083345346,
+      "learning_rate": 3.091529020389009e-05,
+      "loss": 0.4246,
+      "step": 250
+    },
+    {
+      "epoch": 1.9238870272857826,
+      "grad_norm": 0.6630981220865457,
+      "learning_rate": 3.082513454855863e-05,
+      "loss": 0.5091,
+      "step": 251
+    },
+    {
+      "epoch": 1.9315461943513643,
+      "grad_norm": 0.670450600935426,
+      "learning_rate": 3.073466670230361e-05,
+      "loss": 0.3812,
+      "step": 252
+    },
+    {
+      "epoch": 1.939205361416946,
+      "grad_norm": 0.552225236659151,
+      "learning_rate": 3.0643889274167926e-05,
+      "loss": 0.4504,
+      "step": 253
+    },
+    {
+      "epoch": 1.9468645284825277,
+      "grad_norm": 0.570779364405288,
+      "learning_rate": 3.055280488212266e-05,
+      "loss": 0.4444,
+      "step": 254
+    },
+    {
+      "epoch": 1.9545236955481091,
+      "grad_norm": 0.5961792678109743,
+      "learning_rate": 3.0461416152991555e-05,
+      "loss": 0.4763,
+      "step": 255
+    },
+    {
+      "epoch": 1.9621828626136908,
+      "grad_norm": 0.5968250026211483,
+      "learning_rate": 3.0369725722375274e-05,
+      "loss": 0.4754,
+      "step": 256
+    },
+    {
+      "epoch": 1.9698420296792722,
+      "grad_norm": 0.649954877349215,
+      "learning_rate": 3.0277736234575378e-05,
+      "loss": 0.4345,
+      "step": 257
+    },
+    {
+      "epoch": 1.977501196744854,
+      "grad_norm": 0.6643346772869196,
+      "learning_rate": 3.0185450342518075e-05,
+      "loss": 0.4532,
+      "step": 258
+    },
+    {
+      "epoch": 1.9851603638104356,
+      "grad_norm": 0.47332581564622644,
+      "learning_rate": 3.009287070767771e-05,
+      "loss": 0.4881,
+      "step": 259
+    },
+    {
+      "epoch": 1.9928195308760173,
+      "grad_norm": 0.5015320907864502,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4242,
+      "step": 260
+    },
+    {
+      "epoch": 2.0019147917663953,
+      "grad_norm": 0.8833235426774388,
+      "learning_rate": 2.990684089782507e-05,
+      "loss": 0.6896,
+      "step": 261
+    },
+    {
+      "epoch": 2.009573958831977,
+      "grad_norm": 0.8871508493864276,
+      "learning_rate": 2.9813396087810134e-05,
+      "loss": 0.3589,
+      "step": 262
+    },
+    {
+      "epoch": 2.0172331258975587,
+      "grad_norm": 1.0765160264598035,
+      "learning_rate": 2.971966826485212e-05,
+      "loss": 0.3823,
+      "step": 263
+    },
+    {
+      "epoch": 2.0248922929631403,
+      "grad_norm": 1.2741275889363188,
+      "learning_rate": 2.962566013200986e-05,
+      "loss": 0.3751,
+      "step": 264
+    },
+    {
+      "epoch": 2.032551460028722,
+      "grad_norm": 0.7162387782791261,
+      "learning_rate": 2.9531374400426158e-05,
+      "loss": 0.3232,
+      "step": 265
+    },
+    {
+      "epoch": 2.0402106270943037,
+      "grad_norm": 0.972447687697029,
+      "learning_rate": 2.943681378924964e-05,
+      "loss": 0.3447,
+      "step": 266
+    },
+    {
+      "epoch": 2.047869794159885,
+      "grad_norm": 0.7472088308314776,
+      "learning_rate": 2.934198102555631e-05,
+      "loss": 0.3502,
+      "step": 267
+    },
+    {
+      "epoch": 2.0555289612254666,
+      "grad_norm": 0.9904131833727945,
+      "learning_rate": 2.924687884427087e-05,
+      "loss": 0.3328,
+      "step": 268
+    },
+    {
+      "epoch": 2.0631881282910483,
+      "grad_norm": 0.8268717893678047,
+      "learning_rate": 2.9151509988087912e-05,
+      "loss": 0.323,
+      "step": 269
+    },
+    {
+      "epoch": 2.07084729535663,
+      "grad_norm": 1.0133251834372121,
+      "learning_rate": 2.9055877207392752e-05,
+      "loss": 0.3314,
+      "step": 270
+    },
+    {
+      "epoch": 2.0785064624222116,
+      "grad_norm": 0.9295212140684717,
+      "learning_rate": 2.8959983260182166e-05,
+      "loss": 0.3433,
+      "step": 271
+    },
+    {
+      "epoch": 2.0861656294877933,
+      "grad_norm": 0.7493423793907141,
+      "learning_rate": 2.886383091198483e-05,
+      "loss": 0.3174,
+      "step": 272
+    },
+    {
+      "epoch": 2.093824796553375,
+      "grad_norm": 1.1449594495047748,
+      "learning_rate": 2.876742293578155e-05,
+      "loss": 0.4027,
+      "step": 273
+    },
+    {
+      "epoch": 2.1014839636189566,
+      "grad_norm": 0.8495410462496846,
+      "learning_rate": 2.8670762111925313e-05,
+      "loss": 0.3762,
+      "step": 274
+    },
+    {
+      "epoch": 2.109143130684538,
+      "grad_norm": 0.7356070365983308,
+      "learning_rate": 2.8573851228061084e-05,
+      "loss": 0.3116,
+      "step": 275
+    },
+    {
+      "epoch": 2.1168022977501195,
+      "grad_norm": 0.7959446709248563,
+      "learning_rate": 2.8476693079045432e-05,
+      "loss": 0.3379,
+      "step": 276
+    },
+    {
+      "epoch": 2.124461464815701,
+      "grad_norm": 0.8100378872218936,
+      "learning_rate": 2.8379290466865906e-05,
+      "loss": 0.3305,
+      "step": 277
+    },
+    {
+      "epoch": 2.132120631881283,
+      "grad_norm": 0.8085735074400692,
+      "learning_rate": 2.828164620056024e-05,
+      "loss": 0.3598,
+      "step": 278
+    },
+    {
+      "epoch": 2.1397797989468645,
+      "grad_norm": 0.6344360858538197,
+      "learning_rate": 2.818376309613535e-05,
+      "loss": 0.302,
+      "step": 279
+    },
+    {
+      "epoch": 2.147438966012446,
+      "grad_norm": 0.57409154364413,
+      "learning_rate": 2.80856439764861e-05,
+      "loss": 0.3686,
+      "step": 280
+    },
+    {
+      "epoch": 2.155098133078028,
+      "grad_norm": 0.5845504651635646,
+      "learning_rate": 2.798729167131391e-05,
+      "loss": 0.3526,
+      "step": 281
+    },
+    {
+      "epoch": 2.1627573001436096,
+      "grad_norm": 0.4776801469885686,
+      "learning_rate": 2.7888709017045146e-05,
+      "loss": 0.3326,
+      "step": 282
+    },
+    {
+      "epoch": 2.170416467209191,
+      "grad_norm": 0.48684077839230894,
+      "learning_rate": 2.7789898856749297e-05,
+      "loss": 0.3342,
+      "step": 283
+    },
+    {
+      "epoch": 2.1780756342747725,
+      "grad_norm": 0.49773085707991915,
+      "learning_rate": 2.7690864040057023e-05,
+      "loss": 0.3051,
+      "step": 284
+    },
+    {
+      "epoch": 2.185734801340354,
+      "grad_norm": 0.4525197933418408,
+      "learning_rate": 2.7591607423077932e-05,
+      "loss": 0.3087,
+      "step": 285
+    },
+    {
+      "epoch": 2.193393968405936,
+      "grad_norm": 0.5639448536870705,
+      "learning_rate": 2.7492131868318247e-05,
+      "loss": 0.3503,
+      "step": 286
+    },
+    {
+      "epoch": 2.2010531354715175,
+      "grad_norm": 0.5607287965131402,
+      "learning_rate": 2.739244024459822e-05,
+      "loss": 0.3654,
+      "step": 287
+    },
+    {
+      "epoch": 2.208712302537099,
+      "grad_norm": 0.4515128468379325,
+      "learning_rate": 2.7292535426969436e-05,
+      "loss": 0.3149,
+      "step": 288
+    },
+    {
+      "epoch": 2.216371469602681,
+      "grad_norm": 0.6011149807368861,
+      "learning_rate": 2.7192420296631835e-05,
+      "loss": 0.3911,
+      "step": 289
+    },
+    {
+      "epoch": 2.2240306366682625,
+      "grad_norm": 0.4554658307562261,
+      "learning_rate": 2.7092097740850712e-05,
+      "loss": 0.3068,
+      "step": 290
+    },
+    {
+      "epoch": 2.231689803733844,
+      "grad_norm": 0.5520804650288836,
+      "learning_rate": 2.6991570652873357e-05,
+      "loss": 0.3681,
+      "step": 291
+    },
+    {
+      "epoch": 2.2393489707994254,
+      "grad_norm": 0.48661650666410344,
+      "learning_rate": 2.6890841931845674e-05,
+      "loss": 0.3022,
+      "step": 292
+    },
+    {
+      "epoch": 2.247008137865007,
+      "grad_norm": 0.5416624941038123,
+      "learning_rate": 2.6789914482728546e-05,
+      "loss": 0.359,
+      "step": 293
+    },
+    {
+      "epoch": 2.2546673049305888,
+      "grad_norm": 0.582685239343103,
+      "learning_rate": 2.6688791216214064e-05,
+      "loss": 0.3693,
+      "step": 294
+    },
+    {
+      "epoch": 2.2623264719961704,
+      "grad_norm": 0.4568416200895835,
+      "learning_rate": 2.6587475048641596e-05,
+      "loss": 0.32,
+      "step": 295
+    },
+    {
+      "epoch": 2.269985639061752,
+      "grad_norm": 0.6028222179058447,
+      "learning_rate": 2.6485968901913658e-05,
+      "loss": 0.3726,
+      "step": 296
+    },
+    {
+      "epoch": 2.2776448061273338,
+      "grad_norm": 0.5497203840683457,
+      "learning_rate": 2.6384275703411666e-05,
+      "loss": 0.3584,
+      "step": 297
+    },
+    {
+      "epoch": 2.2853039731929154,
+      "grad_norm": 0.4601068271433548,
+      "learning_rate": 2.6282398385911503e-05,
+      "loss": 0.3263,
+      "step": 298
+    },
+    {
+      "epoch": 2.2929631402584967,
+      "grad_norm": 0.5843782768173683,
+      "learning_rate": 2.618033988749895e-05,
+      "loss": 0.3854,
+      "step": 299
+    },
+    {
+      "epoch": 2.3006223073240784,
+      "grad_norm": 0.4668138387272922,
+      "learning_rate": 2.607810315148494e-05,
+      "loss": 0.3445,
+      "step": 300
+    },
+    {
+      "epoch": 2.30828147438966,
+      "grad_norm": 0.4966561419101829,
+      "learning_rate": 2.5975691126320678e-05,
+      "loss": 0.3486,
+      "step": 301
+    },
+    {
+      "epoch": 2.3159406414552417,
+      "grad_norm": 0.552195612826531,
+      "learning_rate": 2.587310676551262e-05,
+      "loss": 0.3478,
+      "step": 302
+    },
+    {
+      "epoch": 2.3235998085208234,
+      "grad_norm": 0.49951171616233425,
+      "learning_rate": 2.5770353027537276e-05,
+      "loss": 0.3494,
+      "step": 303
+    },
+    {
+      "epoch": 2.331258975586405,
+      "grad_norm": 0.4487518241170282,
+      "learning_rate": 2.5667432875755904e-05,
+      "loss": 0.3267,
+      "step": 304
+    },
+    {
+      "epoch": 2.3389181426519867,
+      "grad_norm": 0.47306289952945507,
+      "learning_rate": 2.5564349278329056e-05,
+      "loss": 0.3246,
+      "step": 305
+    },
+    {
+      "epoch": 2.3465773097175684,
+      "grad_norm": 0.4974553141384758,
+      "learning_rate": 2.5461105208130953e-05,
+      "loss": 0.2918,
+      "step": 306
+    },
+    {
+      "epoch": 2.35423647678315,
+      "grad_norm": 0.4809251554335802,
+      "learning_rate": 2.5357703642663766e-05,
+      "loss": 0.357,
+      "step": 307
+    },
+    {
+      "epoch": 2.3618956438487313,
+      "grad_norm": 0.5596662616004654,
+      "learning_rate": 2.525414756397174e-05,
+      "loss": 0.3287,
+      "step": 308
+    },
+    {
+      "epoch": 2.369554810914313,
+      "grad_norm": 0.46194307500162646,
+      "learning_rate": 2.5150439958555205e-05,
+      "loss": 0.323,
+      "step": 309
+    },
+    {
+      "epoch": 2.3772139779798946,
+      "grad_norm": 0.5354975145364615,
+      "learning_rate": 2.5046583817284437e-05,
+      "loss": 0.3641,
+      "step": 310
+    },
+    {
+      "epoch": 2.3848731450454763,
+      "grad_norm": 0.5273653633413932,
+      "learning_rate": 2.4942582135313393e-05,
+      "loss": 0.3553,
+      "step": 311
+    },
+    {
+      "epoch": 2.392532312111058,
+      "grad_norm": 0.49817920339367244,
+      "learning_rate": 2.4838437911993355e-05,
+      "loss": 0.3455,
+      "step": 312
+    },
+    {
+      "epoch": 2.4001914791766397,
+      "grad_norm": 0.515170607443925,
+      "learning_rate": 2.473415415078642e-05,
+      "loss": 0.3619,
+      "step": 313
+    },
+    {
+      "epoch": 2.4078506462422213,
+      "grad_norm": 0.4926042303192558,
+      "learning_rate": 2.4629733859178867e-05,
+      "loss": 0.3485,
+      "step": 314
+    },
+    {
+      "epoch": 2.4155098133078026,
+      "grad_norm": 0.42905408480479645,
+      "learning_rate": 2.4525180048594452e-05,
+      "loss": 0.3057,
+      "step": 315
+    },
+    {
+      "epoch": 2.4231689803733842,
+      "grad_norm": 0.4867004151742306,
+      "learning_rate": 2.4420495734307527e-05,
+      "loss": 0.3747,
+      "step": 316
+    },
+    {
+      "epoch": 2.430828147438966,
+      "grad_norm": 0.41338427949555484,
+      "learning_rate": 2.4315683935356127e-05,
+      "loss": 0.3475,
+      "step": 317
+    },
+    {
+      "epoch": 2.4384873145045476,
+      "grad_norm": 0.44888986646854223,
+      "learning_rate": 2.421074767445485e-05,
+      "loss": 0.303,
+      "step": 318
+    },
+    {
+      "epoch": 2.4461464815701293,
+      "grad_norm": 0.5491552096723435,
+      "learning_rate": 2.4105689977907722e-05,
+      "loss": 0.3979,
+      "step": 319
+    },
+    {
+      "epoch": 2.453805648635711,
+      "grad_norm": 0.4927516444394549,
+      "learning_rate": 2.4000513875520892e-05,
+      "loss": 0.3407,
+      "step": 320
+    },
+    {
+      "epoch": 2.4614648157012926,
+      "grad_norm": 0.466811974445456,
+      "learning_rate": 2.3895222400515282e-05,
+      "loss": 0.3425,
+      "step": 321
+    },
+    {
+      "epoch": 2.4691239827668743,
+      "grad_norm": 0.47821175171844904,
+      "learning_rate": 2.3789818589439094e-05,
+      "loss": 0.3098,
+      "step": 322
+    },
+    {
+      "epoch": 2.476783149832456,
+      "grad_norm": 0.4912242879477096,
+      "learning_rate": 2.3684305482080233e-05,
+      "loss": 0.3591,
+      "step": 323
+    },
+    {
+      "epoch": 2.484442316898037,
+      "grad_norm": 0.4811573112209665,
+      "learning_rate": 2.357868612137866e-05,
+      "loss": 0.3773,
+      "step": 324
+    },
+    {
+      "epoch": 2.492101483963619,
+      "grad_norm": 0.4870566155608247,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.3228,
+      "step": 325
+    },
+    {
+      "epoch": 2.4997606510292005,
+      "grad_norm": 0.5310392825279873,
+      "learning_rate": 2.3367140826940768e-05,
+      "loss": 0.3546,
+      "step": 326
+    },
+    {
+      "epoch": 2.507419818094782,
+      "grad_norm": 0.5483105250492419,
+      "learning_rate": 2.326122099405435e-05,
+      "loss": 0.3375,
+      "step": 327
+    },
+    {
+      "epoch": 2.515078985160364,
+      "grad_norm": 0.5003667793105148,
+      "learning_rate": 2.315520710934903e-05,
+      "loss": 0.3687,
+      "step": 328
+    },
+    {
+      "epoch": 2.5227381522259456,
+      "grad_norm": 0.533844789442384,
+      "learning_rate": 2.304910223020691e-05,
+      "loss": 0.3548,
+      "step": 329
+    },
+    {
+      "epoch": 2.5303973192915272,
+      "grad_norm": 0.5149064035903343,
+      "learning_rate": 2.2942909416634326e-05,
+      "loss": 0.3446,
+      "step": 330
+    },
+    {
+      "epoch": 2.5380564863571085,
+      "grad_norm": 0.4151169766555542,
+      "learning_rate": 2.2836631731173577e-05,
+      "loss": 0.3281,
+      "step": 331
+    },
+    {
+      "epoch": 2.54571565342269,
+      "grad_norm": 0.5156566437281681,
+      "learning_rate": 2.2730272238814636e-05,
+      "loss": 0.3479,
+      "step": 332
+    },
+    {
+      "epoch": 2.553374820488272,
+      "grad_norm": 0.536538863268432,
+      "learning_rate": 2.2623834006906732e-05,
+      "loss": 0.3468,
+      "step": 333
+    },
+    {
+      "epoch": 2.5610339875538535,
+      "grad_norm": 0.41537339391659595,
+      "learning_rate": 2.25173201050699e-05,
+      "loss": 0.311,
+      "step": 334
+    },
+    {
+      "epoch": 2.568693154619435,
+      "grad_norm": 0.5021990261148753,
+      "learning_rate": 2.2410733605106462e-05,
+      "loss": 0.3803,
+      "step": 335
+    },
+    {
+      "epoch": 2.576352321685017,
+      "grad_norm": 0.39621064090295866,
+      "learning_rate": 2.2304077580912423e-05,
+      "loss": 0.2936,
+      "step": 336
+    },
+    {
+      "epoch": 2.5840114887505985,
+      "grad_norm": 0.5372701642790138,
+      "learning_rate": 2.2197355108388835e-05,
+      "loss": 0.3184,
+      "step": 337
+    },
+    {
+      "epoch": 2.5916706558161797,
+      "grad_norm": 0.41383864714124363,
+      "learning_rate": 2.209056926535307e-05,
+      "loss": 0.3287,
+      "step": 338
+    },
+    {
+      "epoch": 2.599329822881762,
+      "grad_norm": 0.5114947611293079,
+      "learning_rate": 2.1983723131450088e-05,
+      "loss": 0.3683,
+      "step": 339
+    },
+    {
+      "epoch": 2.606988989947343,
+      "grad_norm": 0.4173762346729875,
+      "learning_rate": 2.1876819788063586e-05,
+      "loss": 0.3086,
+      "step": 340
+    },
+    {
+      "epoch": 2.6146481570129247,
+      "grad_norm": 0.5192212848345572,
+      "learning_rate": 2.176986231822717e-05,
+      "loss": 0.375,
+      "step": 341
+    },
+    {
+      "epoch": 2.6223073240785064,
+      "grad_norm": 0.5669533936852463,
+      "learning_rate": 2.166285380653541e-05,
+      "loss": 0.3432,
+      "step": 342
+    },
+    {
+      "epoch": 2.629966491144088,
+      "grad_norm": 0.4671450099921592,
+      "learning_rate": 2.1555797339054898e-05,
+      "loss": 0.3207,
+      "step": 343
+    },
+    {
+      "epoch": 2.6376256582096698,
+      "grad_norm": 0.5035698279484611,
+      "learning_rate": 2.1448696003235252e-05,
+      "loss": 0.3586,
+      "step": 344
+    },
+    {
+      "epoch": 2.6452848252752514,
+      "grad_norm": 0.49482528595990216,
+      "learning_rate": 2.1341552887820048e-05,
+      "loss": 0.3419,
+      "step": 345
+    },
+    {
+      "epoch": 2.652943992340833,
+      "grad_norm": 0.5124832493661444,
+      "learning_rate": 2.123437108275779e-05,
+      "loss": 0.3605,
+      "step": 346
+    },
+    {
+      "epoch": 2.6606031594064143,
+      "grad_norm": 0.3916326803311756,
+      "learning_rate": 2.112715367911275e-05,
+      "loss": 0.3102,
+      "step": 347
+    },
+    {
+      "epoch": 2.668262326471996,
+      "grad_norm": 0.41173103844867237,
+      "learning_rate": 2.1019903768975852e-05,
+      "loss": 0.3645,
+      "step": 348
+    },
+    {
+      "epoch": 2.6759214935375777,
+      "grad_norm": 0.48815468984239224,
+      "learning_rate": 2.0912624445375483e-05,
+      "loss": 0.3641,
+      "step": 349
+    },
+    {
+      "epoch": 2.6835806606031594,
+      "grad_norm": 0.40031092411352465,
+      "learning_rate": 2.0805318802188307e-05,
+      "loss": 0.3233,
+      "step": 350
+    },
+    {
+      "epoch": 2.691239827668741,
+      "grad_norm": 0.4801007606225588,
+      "learning_rate": 2.0697989934050025e-05,
+      "loss": 0.3492,
+      "step": 351
+    },
+    {
+      "epoch": 2.6988989947343227,
+      "grad_norm": 0.465959760136513,
+      "learning_rate": 2.0590640936266132e-05,
+      "loss": 0.3492,
+      "step": 352
+    },
+    {
+      "epoch": 2.7065581617999044,
+      "grad_norm": 0.432367828293715,
+      "learning_rate": 2.0483274904722647e-05,
+      "loss": 0.3478,
+      "step": 353
+    },
+    {
+      "epoch": 2.7142173288654856,
+      "grad_norm": 0.4510586932875561,
+      "learning_rate": 2.037589493579685e-05,
+      "loss": 0.3553,
+      "step": 354
+    },
+    {
+      "epoch": 2.7218764959310677,
+      "grad_norm": 0.467950300228232,
+      "learning_rate": 2.0268504126267952e-05,
+      "loss": 0.3653,
+      "step": 355
+    },
+    {
+      "epoch": 2.729535662996649,
+      "grad_norm": 0.5009397300600292,
+      "learning_rate": 2.0161105573227798e-05,
+      "loss": 0.36,
+      "step": 356
+    },
+    {
+      "epoch": 2.7371948300622306,
+      "grad_norm": 0.49408159207397456,
+      "learning_rate": 2.005370237399157e-05,
+      "loss": 0.3515,
+      "step": 357
+    },
+    {
+      "epoch": 2.7448539971278123,
+      "grad_norm": 0.35262515813271533,
+      "learning_rate": 1.9946297626008432e-05,
+      "loss": 0.2689,
+      "step": 358
+    },
+    {
+      "epoch": 2.752513164193394,
+      "grad_norm": 0.4827674543765931,
+      "learning_rate": 1.9838894426772205e-05,
+      "loss": 0.3706,
+      "step": 359
+    },
+    {
+      "epoch": 2.7601723312589757,
+      "grad_norm": 0.3780483362953392,
+      "learning_rate": 1.9731495873732055e-05,
+      "loss": 0.3135,
+      "step": 360
+    },
+    {
+      "epoch": 2.7678314983245573,
+      "grad_norm": 0.3811396663153698,
+      "learning_rate": 1.9624105064203157e-05,
+      "loss": 0.3576,
+      "step": 361
+    },
+    {
+      "epoch": 2.775490665390139,
+      "grad_norm": 0.45864928956878004,
+      "learning_rate": 1.951672509527736e-05,
+      "loss": 0.3321,
+      "step": 362
+    },
+    {
+      "epoch": 2.7831498324557202,
+      "grad_norm": 0.4507638328421475,
+      "learning_rate": 1.940935906373388e-05,
+      "loss": 0.3715,
+      "step": 363
+    },
+    {
+      "epoch": 2.790808999521302,
+      "grad_norm": 0.43134511596968567,
+      "learning_rate": 1.930201006594999e-05,
+      "loss": 0.3412,
+      "step": 364
+    },
+    {
+      "epoch": 2.7984681665868836,
+      "grad_norm": 0.52057087403904,
+      "learning_rate": 1.9194681197811703e-05,
+      "loss": 0.3849,
+      "step": 365
+    },
+    {
+      "epoch": 2.8061273336524652,
+      "grad_norm": 0.37380617464599,
+      "learning_rate": 1.9087375554624527e-05,
+      "loss": 0.3072,
+      "step": 366
+    },
+    {
+      "epoch": 2.813786500718047,
+      "grad_norm": 0.3665120285057917,
+      "learning_rate": 1.898009623102415e-05,
+      "loss": 0.3261,
+      "step": 367
+    },
+    {
+      "epoch": 2.8214456677836286,
+      "grad_norm": 0.45855248957112604,
+      "learning_rate": 1.887284632088725e-05,
+      "loss": 0.4029,
+      "step": 368
+    },
+    {
+      "epoch": 2.8291048348492103,
+      "grad_norm": 0.42598192016232783,
+      "learning_rate": 1.8765628917242213e-05,
+      "loss": 0.3588,
+      "step": 369
+    },
+    {
+      "epoch": 2.8367640019147915,
+      "grad_norm": 0.4155845765749022,
+      "learning_rate": 1.8658447112179952e-05,
+      "loss": 0.3039,
+      "step": 370
+    },
+    {
+      "epoch": 2.8444231689803736,
+      "grad_norm": 0.4601308365256824,
+      "learning_rate": 1.8551303996764755e-05,
+      "loss": 0.3836,
+      "step": 371
+    },
+    {
+      "epoch": 2.852082336045955,
+      "grad_norm": 0.4914229018003968,
+      "learning_rate": 1.8444202660945105e-05,
+      "loss": 0.3679,
+      "step": 372
+    },
+    {
+      "epoch": 2.8597415031115365,
+      "grad_norm": 0.46287822342404056,
+      "learning_rate": 1.8337146193464595e-05,
+      "loss": 0.288,
+      "step": 373
+    },
+    {
+      "epoch": 2.867400670177118,
+      "grad_norm": 0.49183098229163497,
+      "learning_rate": 1.8230137681772836e-05,
+      "loss": 0.3741,
+      "step": 374
+    },
+    {
+      "epoch": 2.8750598372427,
+      "grad_norm": 0.40911104249164226,
+      "learning_rate": 1.8123180211936417e-05,
+      "loss": 0.3425,
+      "step": 375
+    },
+    {
+      "epoch": 2.8827190043082815,
+      "grad_norm": 0.4656657585294656,
+      "learning_rate": 1.801627686854992e-05,
+      "loss": 0.3436,
+      "step": 376
+    },
+    {
+      "epoch": 2.890378171373863,
+      "grad_norm": 0.3955542489042361,
+      "learning_rate": 1.7909430734646936e-05,
+      "loss": 0.3142,
+      "step": 377
+    },
+    {
+      "epoch": 2.898037338439445,
+      "grad_norm": 0.3919921799832172,
+      "learning_rate": 1.780264489161117e-05,
+      "loss": 0.327,
+      "step": 378
+    },
+    {
+      "epoch": 2.905696505505026,
+      "grad_norm": 0.40565680454375597,
+      "learning_rate": 1.769592241908758e-05,
+      "loss": 0.3469,
+      "step": 379
+    },
+    {
+      "epoch": 2.913355672570608,
+      "grad_norm": 0.4612451296187882,
+      "learning_rate": 1.758926639489354e-05,
+      "loss": 0.398,
+      "step": 380
+    },
+    {
+      "epoch": 2.9210148396361895,
+      "grad_norm": 0.374323134300942,
+      "learning_rate": 1.748267989493011e-05,
+      "loss": 0.3077,
+      "step": 381
+    },
+    {
+      "epoch": 2.928674006701771,
+      "grad_norm": 0.447528401086559,
+      "learning_rate": 1.7376165993093278e-05,
+      "loss": 0.3791,
+      "step": 382
+    },
+    {
+      "epoch": 2.936333173767353,
+      "grad_norm": 0.3684289290106699,
+      "learning_rate": 1.7269727761185374e-05,
+      "loss": 0.3203,
+      "step": 383
+    },
+    {
+      "epoch": 2.9439923408329345,
+      "grad_norm": 0.3713011538957981,
+      "learning_rate": 1.7163368268826433e-05,
+      "loss": 0.3247,
+      "step": 384
+    },
+    {
+      "epoch": 2.951651507898516,
+      "grad_norm": 0.4288353235268095,
+      "learning_rate": 1.7057090583365678e-05,
+      "loss": 0.3714,
+      "step": 385
+    },
+    {
+      "epoch": 2.9593106749640974,
+      "grad_norm": 0.35222146097913287,
+      "learning_rate": 1.6950897769793093e-05,
+      "loss": 0.2943,
+      "step": 386
+    },
+    {
+      "epoch": 2.9669698420296795,
+      "grad_norm": 0.38643823068088784,
+      "learning_rate": 1.6844792890650976e-05,
+      "loss": 0.3443,
+      "step": 387
+    },
+    {
+      "epoch": 2.9746290090952607,
+      "grad_norm": 0.38938210740713053,
+      "learning_rate": 1.673877900594566e-05,
+      "loss": 0.3399,
+      "step": 388
+    },
+    {
+      "epoch": 2.9822881761608424,
+      "grad_norm": 0.43159107146354114,
+      "learning_rate": 1.6632859173059232e-05,
+      "loss": 0.344,
+      "step": 389
+    },
+    {
+      "epoch": 2.989947343226424,
+      "grad_norm": 0.38084254373242143,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.3455,
+      "step": 390
+    },
+    {
+      "epoch": 2.9976065102920058,
+      "grad_norm": 0.5938797732009524,
+      "learning_rate": 1.6421313878621344e-05,
+      "loss": 0.4699,
+      "step": 391
+    },
+    {
+      "epoch": 3.006701771182384,
+      "grad_norm": 0.6492454140958951,
+      "learning_rate": 1.631569451791977e-05,
+      "loss": 0.3313,
+      "step": 392
+    },
+    {
+      "epoch": 3.0143609382479655,
+      "grad_norm": 0.4219633362824025,
+      "learning_rate": 1.6210181410560912e-05,
+      "loss": 0.2324,
+      "step": 393
+    },
+    {
+      "epoch": 3.022020105313547,
+      "grad_norm": 0.5182168578367924,
+      "learning_rate": 1.610477759948472e-05,
+      "loss": 0.245,
+      "step": 394
+    },
+    {
+      "epoch": 3.029679272379129,
+      "grad_norm": 0.6377564861463809,
+      "learning_rate": 1.5999486124479115e-05,
+      "loss": 0.2515,
+      "step": 395
+    },
+    {
+      "epoch": 3.0373384394447105,
+      "grad_norm": 0.5098950195718823,
+      "learning_rate": 1.5894310022092288e-05,
+      "loss": 0.2364,
+      "step": 396
+    },
+    {
+      "epoch": 3.044997606510292,
+      "grad_norm": 0.5168454095271555,
+      "learning_rate": 1.5789252325545157e-05,
+      "loss": 0.2334,
+      "step": 397
+    },
+    {
+      "epoch": 3.052656773575874,
+      "grad_norm": 0.5240320586315429,
+      "learning_rate": 1.568431606464388e-05,
+      "loss": 0.2675,
+      "step": 398
+    },
+    {
+      "epoch": 3.060315940641455,
+      "grad_norm": 0.44117429526871266,
+      "learning_rate": 1.557950426569248e-05,
+      "loss": 0.2315,
+      "step": 399
+    },
+    {
+      "epoch": 3.0679751077070367,
+      "grad_norm": 0.48765308835045107,
+      "learning_rate": 1.547481995140556e-05,
+      "loss": 0.2437,
+      "step": 400
+    },
+    {
+      "epoch": 3.0756342747726184,
+      "grad_norm": 0.4520916642343294,
+      "learning_rate": 1.5370266140821143e-05,
+      "loss": 0.2341,
+      "step": 401
+    },
+    {
+      "epoch": 3.0832934418382,
+      "grad_norm": 0.4647186437171434,
+      "learning_rate": 1.5265845849213588e-05,
+      "loss": 0.2468,
+      "step": 402
+    },
+    {
+      "epoch": 3.0909526089037818,
+      "grad_norm": 0.46140225100955573,
+      "learning_rate": 1.5161562088006649e-05,
+      "loss": 0.2345,
+      "step": 403
+    },
+    {
+      "epoch": 3.0986117759693634,
+      "grad_norm": 0.4382412999538287,
+      "learning_rate": 1.5057417864686607e-05,
+      "loss": 0.2386,
+      "step": 404
+    },
+    {
+      "epoch": 3.106270943034945,
+      "grad_norm": 0.45590994359507603,
+      "learning_rate": 1.4953416182715566e-05,
+      "loss": 0.2443,
+      "step": 405
+    },
+    {
+      "epoch": 3.113930110100527,
+      "grad_norm": 0.44632890994369173,
+      "learning_rate": 1.4849560041444795e-05,
+      "loss": 0.2216,
+      "step": 406
+    },
+    {
+      "epoch": 3.121589277166108,
+      "grad_norm": 0.3690202710694795,
+      "learning_rate": 1.4745852436028262e-05,
+      "loss": 0.2289,
+      "step": 407
+    },
+    {
+      "epoch": 3.1292484442316897,
+      "grad_norm": 0.3987565576025537,
+      "learning_rate": 1.464229635733624e-05,
+      "loss": 0.2097,
+      "step": 408
+    },
+    {
+      "epoch": 3.1369076112972714,
+      "grad_norm": 0.3922785950430538,
+      "learning_rate": 1.4538894791869052e-05,
+      "loss": 0.2205,
+      "step": 409
+    },
+    {
+      "epoch": 3.144566778362853,
+      "grad_norm": 0.4019487118753477,
+      "learning_rate": 1.443565072167095e-05,
+      "loss": 0.2536,
+      "step": 410
+    },
+    {
+      "epoch": 3.1522259454284347,
+      "grad_norm": 0.37368100647725233,
+      "learning_rate": 1.43325671242441e-05,
+      "loss": 0.2514,
+      "step": 411
+    },
+    {
+      "epoch": 3.1598851124940164,
+      "grad_norm": 0.3884417771257955,
+      "learning_rate": 1.4229646972462732e-05,
+      "loss": 0.2456,
+      "step": 412
+    },
+    {
+      "epoch": 3.167544279559598,
+      "grad_norm": 0.3767796290523846,
+      "learning_rate": 1.412689323448739e-05,
+      "loss": 0.2237,
+      "step": 413
+    },
+    {
+      "epoch": 3.1752034466251793,
+      "grad_norm": 0.3674660595771459,
+      "learning_rate": 1.4024308873679327e-05,
+      "loss": 0.2591,
+      "step": 414
+    },
+    {
+      "epoch": 3.182862613690761,
+      "grad_norm": 0.3572633760098233,
+      "learning_rate": 1.3921896848515064e-05,
+      "loss": 0.2163,
+      "step": 415
+    },
+    {
+      "epoch": 3.1905217807563426,
+      "grad_norm": 0.381550127572975,
+      "learning_rate": 1.3819660112501054e-05,
+      "loss": 0.2442,
+      "step": 416
+    },
+    {
+      "epoch": 3.1981809478219243,
+      "grad_norm": 0.37631597527744115,
+      "learning_rate": 1.37176016140885e-05,
+      "loss": 0.2323,
+      "step": 417
+    },
+    {
+      "epoch": 3.205840114887506,
+      "grad_norm": 0.35956143405057833,
+      "learning_rate": 1.3615724296588342e-05,
+      "loss": 0.2044,
+      "step": 418
+    },
+    {
+      "epoch": 3.2134992819530876,
+      "grad_norm": 0.38268492786329705,
+      "learning_rate": 1.3514031098086349e-05,
+      "loss": 0.2402,
+      "step": 419
+    },
+    {
+      "epoch": 3.2211584490186693,
+      "grad_norm": 0.38994977473417125,
+      "learning_rate": 1.341252495135841e-05,
+      "loss": 0.2319,
+      "step": 420
+    },
+    {
+      "epoch": 3.228817616084251,
+      "grad_norm": 0.35677922752266494,
+      "learning_rate": 1.3311208783785945e-05,
+      "loss": 0.2341,
+      "step": 421
+    },
+    {
+      "epoch": 3.2364767831498327,
+      "grad_norm": 0.37528548994142574,
+      "learning_rate": 1.3210085517271459e-05,
+      "loss": 0.25,
+      "step": 422
+    },
+    {
+      "epoch": 3.244135950215414,
+      "grad_norm": 0.382737915829752,
+      "learning_rate": 1.3109158068154329e-05,
+      "loss": 0.2501,
+      "step": 423
+    },
+    {
+      "epoch": 3.2517951172809956,
+      "grad_norm": 0.3540547128374988,
+      "learning_rate": 1.3008429347126641e-05,
+      "loss": 0.2482,
+      "step": 424
+    },
+    {
+      "epoch": 3.2594542843465772,
+      "grad_norm": 0.3474058116435946,
+      "learning_rate": 1.2907902259149287e-05,
+      "loss": 0.2359,
+      "step": 425
+    },
+    {
+      "epoch": 3.267113451412159,
+      "grad_norm": 0.3903729184634132,
+      "learning_rate": 1.2807579703368162e-05,
+      "loss": 0.2122,
+      "step": 426
+    },
+    {
+      "epoch": 3.2747726184777406,
+      "grad_norm": 0.40188660683701266,
+      "learning_rate": 1.2707464573030572e-05,
+      "loss": 0.2884,
+      "step": 427
+    },
+    {
+      "epoch": 3.2824317855433223,
+      "grad_norm": 0.3471695792330223,
+      "learning_rate": 1.260755975540178e-05,
+      "loss": 0.217,
+      "step": 428
+    },
+    {
+      "epoch": 3.290090952608904,
+      "grad_norm": 0.36589717504907115,
+      "learning_rate": 1.250786813168176e-05,
+      "loss": 0.2379,
+      "step": 429
+    },
+    {
+      "epoch": 3.297750119674485,
+      "grad_norm": 0.3688472677382392,
+      "learning_rate": 1.2408392576922075e-05,
+      "loss": 0.2317,
+      "step": 430
+    },
+    {
+      "epoch": 3.305409286740067,
+      "grad_norm": 0.3822664640855971,
+      "learning_rate": 1.2309135959942986e-05,
+      "loss": 0.2376,
+      "step": 431
+    },
+    {
+      "epoch": 3.3130684538056485,
+      "grad_norm": 0.35469891768647493,
+      "learning_rate": 1.2210101143250708e-05,
+      "loss": 0.2183,
+      "step": 432
+    },
+    {
+      "epoch": 3.32072762087123,
+      "grad_norm": 0.3609442874261945,
+      "learning_rate": 1.211129098295486e-05,
+      "loss": 0.2372,
+      "step": 433
+    },
+    {
+      "epoch": 3.328386787936812,
+      "grad_norm": 0.40218147421255007,
+      "learning_rate": 1.2012708328686093e-05,
+      "loss": 0.2722,
+      "step": 434
+    },
+    {
+      "epoch": 3.3360459550023935,
+      "grad_norm": 0.34284386522492444,
+      "learning_rate": 1.1914356023513904e-05,
+      "loss": 0.2249,
+      "step": 435
+    },
+    {
+      "epoch": 3.343705122067975,
+      "grad_norm": 0.3398937321234149,
+      "learning_rate": 1.1816236903864656e-05,
+      "loss": 0.2314,
+      "step": 436
+    },
+    {
+      "epoch": 3.351364289133557,
+      "grad_norm": 0.3645220719375267,
+      "learning_rate": 1.1718353799439766e-05,
+      "loss": 0.2415,
+      "step": 437
+    },
+    {
+      "epoch": 3.3590234561991386,
+      "grad_norm": 0.35966116512097646,
+      "learning_rate": 1.1620709533134104e-05,
+      "loss": 0.2539,
+      "step": 438
+    },
+    {
+      "epoch": 3.36668262326472,
+      "grad_norm": 0.35249135244554275,
+      "learning_rate": 1.1523306920954571e-05,
+      "loss": 0.207,
+      "step": 439
+    },
+    {
+      "epoch": 3.3743417903303015,
+      "grad_norm": 0.350527349824925,
+      "learning_rate": 1.1426148771938915e-05,
+      "loss": 0.215,
+      "step": 440
+    },
+    {
+      "epoch": 3.382000957395883,
+      "grad_norm": 0.3645807698031509,
+      "learning_rate": 1.1329237888074691e-05,
+      "loss": 0.2294,
+      "step": 441
+    },
+    {
+      "epoch": 3.389660124461465,
+      "grad_norm": 0.3526979558756505,
+      "learning_rate": 1.123257706421845e-05,
+      "loss": 0.2357,
+      "step": 442
+    },
+    {
+      "epoch": 3.3973192915270465,
+      "grad_norm": 0.3298007139758826,
+      "learning_rate": 1.1136169088015177e-05,
+      "loss": 0.2196,
+      "step": 443
+    },
+    {
+      "epoch": 3.404978458592628,
+      "grad_norm": 0.3901583239274404,
+      "learning_rate": 1.1040016739817836e-05,
+      "loss": 0.2593,
+      "step": 444
+    },
+    {
+      "epoch": 3.41263762565821,
+      "grad_norm": 0.3312336548111748,
+      "learning_rate": 1.094412279260726e-05,
+      "loss": 0.2201,
+      "step": 445
+    },
+    {
+      "epoch": 3.420296792723791,
+      "grad_norm": 0.37712815006038436,
+      "learning_rate": 1.0848490011912096e-05,
+      "loss": 0.2353,
+      "step": 446
+    },
+    {
+      "epoch": 3.4279559597893727,
+      "grad_norm": 0.33427689171156727,
+      "learning_rate": 1.0753121155729133e-05,
+      "loss": 0.2247,
+      "step": 447
+    },
+    {
+      "epoch": 3.4356151268549544,
+      "grad_norm": 0.3403837414551559,
+      "learning_rate": 1.0658018974443692e-05,
+      "loss": 0.2334,
+      "step": 448
+    },
+    {
+      "epoch": 3.443274293920536,
+      "grad_norm": 0.3427488817401635,
+      "learning_rate": 1.056318621075036e-05,
+      "loss": 0.2394,
+      "step": 449
+    },
+    {
+      "epoch": 3.4509334609861178,
+      "grad_norm": 0.35721031062542025,
+      "learning_rate": 1.0468625599573842e-05,
+      "loss": 0.238,
+      "step": 450
+    },
+    {
+      "epoch": 3.4585926280516994,
+      "grad_norm": 0.35498625884545276,
+      "learning_rate": 1.037433986799015e-05,
+      "loss": 0.248,
+      "step": 451
+    },
+    {
+      "epoch": 3.466251795117281,
+      "grad_norm": 0.3686856880290374,
+      "learning_rate": 1.028033173514788e-05,
+      "loss": 0.2602,
+      "step": 452
+    },
+    {
+      "epoch": 3.4739109621828628,
+      "grad_norm": 0.32569736084610973,
+      "learning_rate": 1.0186603912189867e-05,
+      "loss": 0.2261,
+      "step": 453
+    },
+    {
+      "epoch": 3.4815701292484444,
+      "grad_norm": 0.35179935045391797,
+      "learning_rate": 1.0093159102174938e-05,
+      "loss": 0.228,
+      "step": 454
+    },
+    {
+      "epoch": 3.4892292963140257,
+      "grad_norm": 0.37412117087411056,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.2558,
+      "step": 455
+    },
+    {
+      "epoch": 3.4968884633796073,
+      "grad_norm": 0.3196396284614333,
+      "learning_rate": 9.907129292322298e-06,
+      "loss": 0.2016,
+      "step": 456
+    },
+    {
+      "epoch": 3.504547630445189,
+      "grad_norm": 0.367949651744729,
+      "learning_rate": 9.814549657481935e-06,
+      "loss": 0.2345,
+      "step": 457
+    },
+    {
+      "epoch": 3.5122067975107707,
+      "grad_norm": 0.36477967052612537,
+      "learning_rate": 9.722263765424628e-06,
+      "loss": 0.2276,
+      "step": 458
+    },
+    {
+      "epoch": 3.5198659645763524,
+      "grad_norm": 0.3332594226268887,
+      "learning_rate": 9.630274277624729e-06,
+      "loss": 0.23,
+      "step": 459
+    },
+    {
+      "epoch": 3.527525131641934,
+      "grad_norm": 0.3477521018845692,
+      "learning_rate": 9.538583847008452e-06,
+      "loss": 0.2352,
+      "step": 460
+    },
+    {
+      "epoch": 3.5351842987075157,
+      "grad_norm": 0.32086860442292786,
+      "learning_rate": 9.447195117877343e-06,
+      "loss": 0.2125,
+      "step": 461
+    },
+    {
+      "epoch": 3.542843465773097,
+      "grad_norm": 0.3678326552292364,
+      "learning_rate": 9.356110725832081e-06,
+      "loss": 0.2576,
+      "step": 462
+    },
+    {
+      "epoch": 3.550502632838679,
+      "grad_norm": 0.3656606530951392,
+      "learning_rate": 9.265333297696395e-06,
+      "loss": 0.2461,
+      "step": 463
+    },
+    {
+      "epoch": 3.5581617999042603,
+      "grad_norm": 0.3229274366638359,
+      "learning_rate": 9.174865451441375e-06,
+      "loss": 0.2298,
+      "step": 464
+    },
+    {
+      "epoch": 3.565820966969842,
+      "grad_norm": 0.32293232206607697,
+      "learning_rate": 9.084709796109907e-06,
+      "loss": 0.2166,
+      "step": 465
+    },
+    {
+      "epoch": 3.5734801340354236,
+      "grad_norm": 0.3461993603233461,
+      "learning_rate": 8.994868931741499e-06,
+      "loss": 0.2342,
+      "step": 466
+    },
+    {
+      "epoch": 3.5811393011010053,
+      "grad_norm": 0.31931497478868687,
+      "learning_rate": 8.905345449297223e-06,
+      "loss": 0.2245,
+      "step": 467
+    },
+    {
+      "epoch": 3.588798468166587,
+      "grad_norm": 0.34882777970263285,
+      "learning_rate": 8.816141930585067e-06,
+      "loss": 0.2412,
+      "step": 468
+    },
+    {
+      "epoch": 3.5964576352321687,
+      "grad_norm": 0.31810617065052815,
+      "learning_rate": 8.72726094818541e-06,
+      "loss": 0.2202,
+      "step": 469
+    },
+    {
+      "epoch": 3.6041168022977503,
+      "grad_norm": 0.36687350590070866,
+      "learning_rate": 8.638705065376887e-06,
+      "loss": 0.225,
+      "step": 470
+    },
+    {
+      "epoch": 3.6117759693633316,
+      "grad_norm": 0.35726654000876984,
+      "learning_rate": 8.550476836062419e-06,
+      "loss": 0.2343,
+      "step": 471
+    },
+    {
+      "epoch": 3.6194351364289132,
+      "grad_norm": 0.33724954427539455,
+      "learning_rate": 8.462578804695595e-06,
+      "loss": 0.2328,
+      "step": 472
+    },
+    {
+      "epoch": 3.627094303494495,
+      "grad_norm": 0.33977950988225286,
+      "learning_rate": 8.375013506207275e-06,
+      "loss": 0.2593,
+      "step": 473
+    },
+    {
+      "epoch": 3.6347534705600766,
+      "grad_norm": 0.3357461336976667,
+      "learning_rate": 8.287783465932466e-06,
+      "loss": 0.2162,
+      "step": 474
+    },
+    {
+      "epoch": 3.6424126376256583,
+      "grad_norm": 0.33938581871332074,
+      "learning_rate": 8.200891199537549e-06,
+      "loss": 0.2159,
+      "step": 475
+    },
+    {
+      "epoch": 3.65007180469124,
+      "grad_norm": 0.30850555682034936,
+      "learning_rate": 8.114339212947655e-06,
+      "loss": 0.2308,
+      "step": 476
+    },
+    {
+      "epoch": 3.6577309717568216,
+      "grad_norm": 0.33712030493831774,
+      "learning_rate": 8.028130002274459e-06,
+      "loss": 0.2617,
+      "step": 477
+    },
+    {
+      "epoch": 3.665390138822403,
+      "grad_norm": 0.33952468314921025,
+      "learning_rate": 7.942266053744155e-06,
+      "loss": 0.2277,
+      "step": 478
+    },
+    {
+      "epoch": 3.673049305887985,
+      "grad_norm": 0.32790624314094596,
+      "learning_rate": 7.856749843625777e-06,
+      "loss": 0.2459,
+      "step": 479
+    },
+    {
+      "epoch": 3.680708472953566,
+      "grad_norm": 0.30426795240880533,
+      "learning_rate": 7.771583838159756e-06,
+      "loss": 0.2373,
+      "step": 480
+    },
+    {
+      "epoch": 3.688367640019148,
+      "grad_norm": 0.32139830397764274,
+      "learning_rate": 7.686770493486835e-06,
+      "loss": 0.265,
+      "step": 481
+    },
+    {
+      "epoch": 3.6960268070847295,
+      "grad_norm": 0.30384977031133725,
+      "learning_rate": 7.602312255577193e-06,
+      "loss": 0.2166,
+      "step": 482
+    },
+    {
+      "epoch": 3.703685974150311,
+      "grad_norm": 0.33085512515068155,
+      "learning_rate": 7.518211560159949e-06,
+      "loss": 0.2511,
+      "step": 483
+    },
+    {
+      "epoch": 3.711345141215893,
+      "grad_norm": 0.3176611144031733,
+      "learning_rate": 7.434470832652865e-06,
+      "loss": 0.2084,
+      "step": 484
+    },
+    {
+      "epoch": 3.7190043082814745,
+      "grad_norm": 0.33574434095691075,
+      "learning_rate": 7.3510924880924575e-06,
+      "loss": 0.2308,
+      "step": 485
+    },
+    {
+      "epoch": 3.726663475347056,
+      "grad_norm": 0.2971615945594848,
+      "learning_rate": 7.268078931064293e-06,
+      "loss": 0.2217,
+      "step": 486
+    },
+    {
+      "epoch": 3.7343226424126374,
+      "grad_norm": 0.32961234977638115,
+      "learning_rate": 7.185432555633672e-06,
+      "loss": 0.2548,
+      "step": 487
+    },
+    {
+      "epoch": 3.741981809478219,
+      "grad_norm": 0.32814885991409504,
+      "learning_rate": 7.1031557452765934e-06,
+      "loss": 0.2439,
+      "step": 488
+    },
+    {
+      "epoch": 3.749640976543801,
+      "grad_norm": 0.34360104917258844,
+      "learning_rate": 7.021250872810983e-06,
+      "loss": 0.224,
+      "step": 489
+    },
+    {
+      "epoch": 3.7573001436093825,
+      "grad_norm": 0.3505861571637595,
+      "learning_rate": 6.939720300328303e-06,
+      "loss": 0.2459,
+      "step": 490
+    },
+    {
+      "epoch": 3.764959310674964,
+      "grad_norm": 0.33976097846578596,
+      "learning_rate": 6.858566379125389e-06,
+      "loss": 0.2329,
+      "step": 491
+    },
+    {
+      "epoch": 3.772618477740546,
+      "grad_norm": 0.32593443522992444,
+      "learning_rate": 6.777791449636681e-06,
+      "loss": 0.2295,
+      "step": 492
+    },
+    {
+      "epoch": 3.7802776448061275,
+      "grad_norm": 0.33560750807739526,
+      "learning_rate": 6.697397841366686e-06,
+      "loss": 0.2148,
+      "step": 493
+    },
+    {
+      "epoch": 3.7879368118717087,
+      "grad_norm": 0.3347449057734125,
+      "learning_rate": 6.617387872822842e-06,
+      "loss": 0.2197,
+      "step": 494
+    },
+    {
+      "epoch": 3.795595978937291,
+      "grad_norm": 0.3467626472846674,
+      "learning_rate": 6.537763851448593e-06,
+      "loss": 0.2391,
+      "step": 495
+    },
+    {
+      "epoch": 3.803255146002872,
+      "grad_norm": 0.32356879416863116,
+      "learning_rate": 6.458528073556925e-06,
+      "loss": 0.2268,
+      "step": 496
+    },
+    {
+      "epoch": 3.8109143130684537,
+      "grad_norm": 0.3283370951000147,
+      "learning_rate": 6.379682824264055e-06,
+      "loss": 0.2337,
+      "step": 497
+    },
+    {
+      "epoch": 3.8185734801340354,
+      "grad_norm": 0.30770811712569635,
+      "learning_rate": 6.301230377423595e-06,
+      "loss": 0.2224,
+      "step": 498
+    },
+    {
+      "epoch": 3.826232647199617,
+      "grad_norm": 0.3253183469854635,
+      "learning_rate": 6.223172995560935e-06,
+      "loss": 0.2193,
+      "step": 499
+    },
+    {
+      "epoch": 3.8338918142651988,
+      "grad_norm": 0.3348399208299024,
+      "learning_rate": 6.145512929808013e-06,
+      "loss": 0.241,
+      "step": 500
+    },
+    {
+      "epoch": 3.8415509813307804,
+      "grad_norm": 0.33895071979833463,
+      "learning_rate": 6.068252419838399e-06,
+      "loss": 0.2427,
+      "step": 501
+    },
+    {
+      "epoch": 3.849210148396362,
+      "grad_norm": 0.3343502843045349,
+      "learning_rate": 5.991393693802674e-06,
+      "loss": 0.2319,
+      "step": 502
+    },
+    {
+      "epoch": 3.8568693154619433,
+      "grad_norm": 0.32674599015773287,
+      "learning_rate": 5.9149389682642165e-06,
+      "loss": 0.2438,
+      "step": 503
+    },
+    {
+      "epoch": 3.864528482527525,
+      "grad_norm": 0.324625635133392,
+      "learning_rate": 5.838890448135228e-06,
+      "loss": 0.2464,
+      "step": 504
+    },
+    {
+      "epoch": 3.8721876495931067,
+      "grad_norm": 0.33427787078954374,
+      "learning_rate": 5.7632503266131925e-06,
+      "loss": 0.216,
+      "step": 505
+    },
+    {
+      "epoch": 3.8798468166586884,
+      "grad_norm": 0.2975052588803942,
+      "learning_rate": 5.688020785117581e-06,
+      "loss": 0.2193,
+      "step": 506
+    },
+    {
+      "epoch": 3.88750598372427,
+      "grad_norm": 0.35541630429900906,
+      "learning_rate": 5.613203993226981e-06,
+      "loss": 0.2416,
+      "step": 507
+    },
+    {
+      "epoch": 3.8951651507898517,
+      "grad_norm": 0.3667627536084296,
+      "learning_rate": 5.538802108616494e-06,
+      "loss": 0.2339,
+      "step": 508
+    },
+    {
+      "epoch": 3.9028243178554334,
+      "grad_norm": 0.32563123367734753,
+      "learning_rate": 5.46481727699554e-06,
+      "loss": 0.2259,
+      "step": 509
+    },
+    {
+      "epoch": 3.9104834849210146,
+      "grad_norm": 0.3147349932477261,
+      "learning_rate": 5.39125163204594e-06,
+      "loss": 0.2166,
+      "step": 510
+    },
+    {
+      "epoch": 3.9181426519865967,
+      "grad_norm": 0.3264794954593056,
+      "learning_rate": 5.318107295360424e-06,
+      "loss": 0.2191,
+      "step": 511
+    },
+    {
+      "epoch": 3.925801819052178,
+      "grad_norm": 0.3493610273060971,
+      "learning_rate": 5.245386376381398e-06,
+      "loss": 0.223,
+      "step": 512
+    },
+    {
+      "epoch": 3.9334609861177596,
+      "grad_norm": 0.31947742082366326,
+      "learning_rate": 5.17309097234016e-06,
+      "loss": 0.2286,
+      "step": 513
+    },
+    {
+      "epoch": 3.9411201531833413,
+      "grad_norm": 0.3273845550020943,
+      "learning_rate": 5.101223168196381e-06,
+      "loss": 0.2425,
+      "step": 514
+    },
+    {
+      "epoch": 3.948779320248923,
+      "grad_norm": 0.33071553976584445,
+      "learning_rate": 5.029785036577976e-06,
+      "loss": 0.2142,
+      "step": 515
+    },
+    {
+      "epoch": 3.9564384873145046,
+      "grad_norm": 0.29920602227346177,
+      "learning_rate": 4.958778637721364e-06,
+      "loss": 0.2255,
+      "step": 516
+    },
+    {
+      "epoch": 3.9640976543800863,
+      "grad_norm": 0.3220643902328496,
+      "learning_rate": 4.8882060194119985e-06,
+      "loss": 0.2316,
+      "step": 517
+    },
+    {
+      "epoch": 3.971756821445668,
+      "grad_norm": 0.3348880545917222,
+      "learning_rate": 4.8180692169253714e-06,
+      "loss": 0.2319,
+      "step": 518
+    },
+    {
+      "epoch": 3.9794159885112492,
+      "grad_norm": 0.3183864890443684,
+      "learning_rate": 4.74837025296826e-06,
+      "loss": 0.2398,
+      "step": 519
+    },
+    {
+      "epoch": 3.987075155576831,
+      "grad_norm": 0.3170966409917256,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.2235,
+      "step": 520
+    },
+    {
+      "epoch": 3.9947343226424126,
+      "grad_norm": 0.2929870337825558,
+      "learning_rate": 4.610293868276681e-06,
+      "loss": 0.2118,
+      "step": 521
+    },
+    {
+      "epoch": 4.003829583532791,
+      "grad_norm": 0.6366855376673984,
+      "learning_rate": 4.541920429589168e-06,
+      "loss": 0.3375,
+      "step": 522
+    },
+    {
+      "epoch": 4.011488750598373,
+      "grad_norm": 0.39332713233031513,
+      "learning_rate": 4.47399279341024e-06,
+      "loss": 0.1091,
+      "step": 523
+    },
+    {
+      "epoch": 4.019147917663954,
+      "grad_norm": 0.46392007812335173,
+      "learning_rate": 4.406512918735555e-06,
+      "loss": 0.2075,
+      "step": 524
+    },
+    {
+      "epoch": 4.026807084729536,
+      "grad_norm": 0.32163834207187486,
+      "learning_rate": 4.339482751647557e-06,
+      "loss": 0.1486,
+      "step": 525
+    },
+    {
+      "epoch": 4.034466251795117,
+      "grad_norm": 0.346990014252883,
+      "learning_rate": 4.272904225259387e-06,
+      "loss": 0.1689,
+      "step": 526
+    },
+    {
+      "epoch": 4.0421254188606985,
+      "grad_norm": 0.4470904005969031,
+      "learning_rate": 4.206779259659102e-06,
+      "loss": 0.1619,
+      "step": 527
+    },
+    {
+      "epoch": 4.049784585926281,
+      "grad_norm": 0.5423468490222553,
+      "learning_rate": 4.141109761854332e-06,
+      "loss": 0.1583,
+      "step": 528
+    },
+    {
+      "epoch": 4.057443752991862,
+      "grad_norm": 0.4455453744472251,
+      "learning_rate": 4.075897625717249e-06,
+      "loss": 0.1561,
+      "step": 529
+    },
+    {
+      "epoch": 4.065102920057444,
+      "grad_norm": 0.40070748734343326,
+      "learning_rate": 4.011144731929981e-06,
+      "loss": 0.1711,
+      "step": 530
+    },
+    {
+      "epoch": 4.072762087123025,
+      "grad_norm": 0.3857460690221848,
+      "learning_rate": 3.9468529479303445e-06,
+      "loss": 0.1874,
+      "step": 531
+    },
+    {
+      "epoch": 4.080421254188607,
+      "grad_norm": 0.3386030886136809,
+      "learning_rate": 3.883024127858017e-06,
+      "loss": 0.1676,
+      "step": 532
+    },
+    {
+      "epoch": 4.088080421254189,
+      "grad_norm": 0.3665495913561285,
+      "learning_rate": 3.819660112501053e-06,
+      "loss": 0.1605,
+      "step": 533
+    },
+    {
+      "epoch": 4.09573958831977,
+      "grad_norm": 0.35417549475087723,
+      "learning_rate": 3.756762729242773e-06,
+      "loss": 0.1504,
+      "step": 534
+    },
+    {
+      "epoch": 4.103398755385352,
+      "grad_norm": 0.3396123769965615,
+      "learning_rate": 3.694333792009115e-06,
+      "loss": 0.1375,
+      "step": 535
+    },
+    {
+      "epoch": 4.111057922450933,
+      "grad_norm": 0.32569257575060306,
+      "learning_rate": 3.632375101216259e-06,
+      "loss": 0.1518,
+      "step": 536
+    },
+    {
+      "epoch": 4.118717089516515,
+      "grad_norm": 0.3219593442840899,
+      "learning_rate": 3.5708884437187673e-06,
+      "loss": 0.168,
+      "step": 537
+    },
+    {
+      "epoch": 4.1263762565820965,
+      "grad_norm": 0.31324123070534954,
+      "learning_rate": 3.509875592757999e-06,
+      "loss": 0.1681,
+      "step": 538
+    },
+    {
+      "epoch": 4.134035423647679,
+      "grad_norm": 0.3836751303057977,
+      "learning_rate": 3.4493383079110054e-06,
+      "loss": 0.1898,
+      "step": 539
+    },
+    {
+      "epoch": 4.14169459071326,
+      "grad_norm": 0.3500886642993717,
+      "learning_rate": 3.3892783350397675e-06,
+      "loss": 0.1793,
+      "step": 540
+    },
+    {
+      "epoch": 4.149353757778842,
+      "grad_norm": 0.3188272991338731,
+      "learning_rate": 3.329697406240855e-06,
+      "loss": 0.1721,
+      "step": 541
+    },
+    {
+      "epoch": 4.157012924844423,
+      "grad_norm": 0.3012590925590462,
+      "learning_rate": 3.2705972397954655e-06,
+      "loss": 0.1537,
+      "step": 542
+    },
+    {
+      "epoch": 4.164672091910004,
+      "grad_norm": 0.3367792622931613,
+      "learning_rate": 3.211979540119883e-06,
+      "loss": 0.1797,
+      "step": 543
+    },
+    {
+      "epoch": 4.1723312589755865,
+      "grad_norm": 0.33133930012065166,
+      "learning_rate": 3.153845997716303e-06,
+      "loss": 0.1925,
+      "step": 544
+    },
+    {
+      "epoch": 4.179990426041168,
+      "grad_norm": 0.3094187164114957,
+      "learning_rate": 3.0961982891241083e-06,
+      "loss": 0.1481,
+      "step": 545
+    },
+    {
+      "epoch": 4.18764959310675,
+      "grad_norm": 0.317376469580326,
+      "learning_rate": 3.039038076871481e-06,
+      "loss": 0.1727,
+      "step": 546
+    },
+    {
+      "epoch": 4.195308760172331,
+      "grad_norm": 0.27650160786044625,
+      "learning_rate": 2.9823670094275e-06,
+      "loss": 0.1446,
+      "step": 547
+    },
+    {
+      "epoch": 4.202967927237913,
+      "grad_norm": 0.30890083357655423,
+      "learning_rate": 2.9261867211545603e-06,
+      "loss": 0.1622,
+      "step": 548
+    },
+    {
+      "epoch": 4.2106270943034945,
+      "grad_norm": 0.33112327548553416,
+      "learning_rate": 2.870498832261257e-06,
+      "loss": 0.1663,
+      "step": 549
+    },
+    {
+      "epoch": 4.218286261369076,
+      "grad_norm": 0.3120284953069467,
+      "learning_rate": 2.815304948755664e-06,
+      "loss": 0.1617,
+      "step": 550
+    },
+    {
+      "epoch": 4.225945428434658,
+      "grad_norm": 0.31255801847483555,
+      "learning_rate": 2.7606066623990145e-06,
+      "loss": 0.1673,
+      "step": 551
+    },
+    {
+      "epoch": 4.233604595500239,
+      "grad_norm": 0.3087850852760736,
+      "learning_rate": 2.7064055506597875e-06,
+      "loss": 0.1608,
+      "step": 552
+    },
+    {
+      "epoch": 4.241263762565821,
+      "grad_norm": 0.29550399939819244,
+      "learning_rate": 2.6527031766682142e-06,
+      "loss": 0.149,
+      "step": 553
+    },
+    {
+      "epoch": 4.248922929631402,
+      "grad_norm": 0.28827764887901935,
+      "learning_rate": 2.599501089171217e-06,
+      "loss": 0.1545,
+      "step": 554
+    },
+    {
+      "epoch": 4.2565820966969845,
+      "grad_norm": 0.31764263264427883,
+      "learning_rate": 2.546800822487714e-06,
+      "loss": 0.1581,
+      "step": 555
+    },
+    {
+      "epoch": 4.264241263762566,
+      "grad_norm": 0.2896451556809014,
+      "learning_rate": 2.494603896464405e-06,
+      "loss": 0.1432,
+      "step": 556
+    },
+    {
+      "epoch": 4.271900430828148,
+      "grad_norm": 0.2846656674626245,
+      "learning_rate": 2.4429118164319076e-06,
+      "loss": 0.1506,
+      "step": 557
+    },
+    {
+      "epoch": 4.279559597893729,
+      "grad_norm": 0.3167572946509413,
+      "learning_rate": 2.3917260731613733e-06,
+      "loss": 0.1681,
+      "step": 558
+    },
+    {
+      "epoch": 4.28721876495931,
+      "grad_norm": 0.2994943543117942,
+      "learning_rate": 2.3410481428214602e-06,
+      "loss": 0.1582,
+      "step": 559
+    },
+    {
+      "epoch": 4.294877932024892,
+      "grad_norm": 0.2958413618672273,
+      "learning_rate": 2.2908794869358044e-06,
+      "loss": 0.1629,
+      "step": 560
+    },
+    {
+      "epoch": 4.302537099090474,
+      "grad_norm": 0.2854040506798909,
+      "learning_rate": 2.2412215523408266e-06,
+      "loss": 0.1504,
+      "step": 561
+    },
+    {
+      "epoch": 4.310196266156056,
+      "grad_norm": 0.27706245243194494,
+      "learning_rate": 2.1920757711440354e-06,
+      "loss": 0.1335,
+      "step": 562
+    },
+    {
+      "epoch": 4.317855433221637,
+      "grad_norm": 0.3002281927963276,
+      "learning_rate": 2.143443560682721e-06,
+      "loss": 0.1656,
+      "step": 563
+    },
+    {
+      "epoch": 4.325514600287219,
+      "grad_norm": 0.3053710633138099,
+      "learning_rate": 2.0953263234830667e-06,
+      "loss": 0.1793,
+      "step": 564
+    },
+    {
+      "epoch": 4.3331737673528,
+      "grad_norm": 0.29373337337239025,
+      "learning_rate": 2.0477254472197237e-06,
+      "loss": 0.1607,
+      "step": 565
+    },
+    {
+      "epoch": 4.340832934418382,
+      "grad_norm": 0.29361686795928127,
+      "learning_rate": 2.0006423046757596e-06,
+      "loss": 0.1417,
+      "step": 566
+    },
+    {
+      "epoch": 4.348492101483964,
+      "grad_norm": 0.30321625536955094,
+      "learning_rate": 1.9540782537031045e-06,
+      "loss": 0.1561,
+      "step": 567
+    },
+    {
+      "epoch": 4.356151268549545,
+      "grad_norm": 0.294608355859984,
+      "learning_rate": 1.908034637183356e-06,
+      "loss": 0.1436,
+      "step": 568
+    },
+    {
+      "epoch": 4.363810435615127,
+      "grad_norm": 0.2981183902576262,
+      "learning_rate": 1.8625127829890922e-06,
+      "loss": 0.1829,
+      "step": 569
+    },
+    {
+      "epoch": 4.371469602680708,
+      "grad_norm": 0.2662376876124655,
+      "learning_rate": 1.817514003945524e-06,
+      "loss": 0.1321,
+      "step": 570
+    },
+    {
+      "epoch": 4.37912876974629,
+      "grad_norm": 0.2857595944219623,
+      "learning_rate": 1.7730395977926917e-06,
+      "loss": 0.1568,
+      "step": 571
+    },
+    {
+      "epoch": 4.386787936811872,
+      "grad_norm": 0.29444711124099954,
+      "learning_rate": 1.7290908471479805e-06,
+      "loss": 0.1619,
+      "step": 572
+    },
+    {
+      "epoch": 4.394447103877454,
+      "grad_norm": 0.299861629149527,
+      "learning_rate": 1.6856690194691872e-06,
+      "loss": 0.1765,
+      "step": 573
+    },
+    {
+      "epoch": 4.402106270943035,
+      "grad_norm": 0.2885973088406149,
+      "learning_rate": 1.6427753670179214e-06,
+      "loss": 0.1599,
+      "step": 574
+    },
+    {
+      "epoch": 4.409765438008616,
+      "grad_norm": 0.284486199205028,
+      "learning_rate": 1.6004111268235156e-06,
+      "loss": 0.1595,
+      "step": 575
+    },
+    {
+      "epoch": 4.417424605074198,
+      "grad_norm": 0.2941729164644181,
+      "learning_rate": 1.5585775206473508e-06,
+      "loss": 0.161,
+      "step": 576
+    },
+    {
+      "epoch": 4.4250837721397795,
+      "grad_norm": 0.29517993856011077,
+      "learning_rate": 1.5172757549476024e-06,
+      "loss": 0.1608,
+      "step": 577
+    },
+    {
+      "epoch": 4.432742939205362,
+      "grad_norm": 0.3009878509805784,
+      "learning_rate": 1.4765070208444732e-06,
+      "loss": 0.1657,
+      "step": 578
+    },
+    {
+      "epoch": 4.440402106270943,
+      "grad_norm": 0.2832295303483787,
+      "learning_rate": 1.4362724940858109e-06,
+      "loss": 0.1481,
+      "step": 579
+    },
+    {
+      "epoch": 4.448061273336525,
+      "grad_norm": 0.30017167488484514,
+      "learning_rate": 1.396573335013236e-06,
+      "loss": 0.1758,
+      "step": 580
+    },
+    {
+      "epoch": 4.455720440402106,
+      "grad_norm": 0.26948255003997973,
+      "learning_rate": 1.3574106885286465e-06,
+      "loss": 0.1506,
+      "step": 581
+    },
+    {
+      "epoch": 4.463379607467688,
+      "grad_norm": 0.2989296760767762,
+      "learning_rate": 1.3187856840612167e-06,
+      "loss": 0.1745,
+      "step": 582
+    },
+    {
+      "epoch": 4.47103877453327,
+      "grad_norm": 0.3057384139973513,
+      "learning_rate": 1.2806994355348224e-06,
+      "loss": 0.1721,
+      "step": 583
+    },
+    {
+      "epoch": 4.478697941598851,
+      "grad_norm": 0.2636716603605,
+      "learning_rate": 1.2431530413359138e-06,
+      "loss": 0.1367,
+      "step": 584
+    },
+    {
+      "epoch": 4.486357108664433,
+      "grad_norm": 0.29595905590077376,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.1651,
+      "step": 585
+    },
+    {
+      "epoch": 4.494016275730014,
+      "grad_norm": 0.28723857409594045,
+      "learning_rate": 1.169684131589608e-06,
+      "loss": 0.1686,
+      "step": 586
+    },
+    {
+      "epoch": 4.501675442795596,
+      "grad_norm": 0.28232920958455504,
+      "learning_rate": 1.1337637348451369e-06,
+      "loss": 0.1478,
+      "step": 587
+    },
+    {
+      "epoch": 4.5093346098611775,
+      "grad_norm": 0.2945977053685175,
+      "learning_rate": 1.0983874299729092e-06,
+      "loss": 0.1698,
+      "step": 588
+    },
+    {
+      "epoch": 4.516993776926759,
+      "grad_norm": 0.29635940814476264,
+      "learning_rate": 1.0635562372060825e-06,
+      "loss": 0.1684,
+      "step": 589
+    },
+    {
+      "epoch": 4.524652943992341,
+      "grad_norm": 0.28161502295558277,
+      "learning_rate": 1.0292711610570904e-06,
+      "loss": 0.1435,
+      "step": 590
+    },
+    {
+      "epoch": 4.532312111057922,
+      "grad_norm": 0.29973282189790135,
+      "learning_rate": 9.955331902886645e-07,
+      "loss": 0.1601,
+      "step": 591
+    },
+    {
+      "epoch": 4.539971278123504,
+      "grad_norm": 0.2912268259487591,
+      "learning_rate": 9.62343297885313e-07,
+      "loss": 0.1676,
+      "step": 592
+    },
+    {
+      "epoch": 4.547630445189085,
+      "grad_norm": 0.2804062252790609,
+      "learning_rate": 9.297024410252753e-07,
+      "loss": 0.1572,
+      "step": 593
+    },
+    {
+      "epoch": 4.5552896122546676,
+      "grad_norm": 0.2902402528058867,
+      "learning_rate": 8.976115610528957e-07,
+      "loss": 0.1665,
+      "step": 594
+    },
+    {
+      "epoch": 4.562948779320249,
+      "grad_norm": 0.2948360443924776,
+      "learning_rate": 8.660715834514977e-07,
+      "loss": 0.1798,
+      "step": 595
+    },
+    {
+      "epoch": 4.570607946385831,
+      "grad_norm": 0.30613400428544785,
+      "learning_rate": 8.350834178166755e-07,
+      "loss": 0.1708,
+      "step": 596
+    },
+    {
+      "epoch": 4.578267113451412,
+      "grad_norm": 0.29386467356756857,
+      "learning_rate": 8.046479578300803e-07,
+      "loss": 0.1674,
+      "step": 597
+    },
+    {
+      "epoch": 4.585926280516993,
+      "grad_norm": 0.26855112533508585,
+      "learning_rate": 7.747660812336221e-07,
+      "loss": 0.127,
+      "step": 598
+    },
+    {
+      "epoch": 4.5935854475825755,
+      "grad_norm": 0.30897844150919035,
+      "learning_rate": 7.454386498041865e-07,
+      "loss": 0.1832,
+      "step": 599
+    },
+    {
+      "epoch": 4.601244614648157,
+      "grad_norm": 0.26907534103375785,
+      "learning_rate": 7.166665093287539e-07,
+      "loss": 0.1514,
+      "step": 600
+    },
+    {
+      "epoch": 4.608903781713739,
+      "grad_norm": 0.2779300127385892,
+      "learning_rate": 6.884504895800237e-07,
+      "loss": 0.1596,
+      "step": 601
+    },
+    {
+      "epoch": 4.61656294877932,
+      "grad_norm": 0.283076309215598,
+      "learning_rate": 6.607914042924756e-07,
+      "loss": 0.1493,
+      "step": 602
+    },
+    {
+      "epoch": 4.624222115844902,
+      "grad_norm": 0.30512219679760166,
+      "learning_rate": 6.336900511389133e-07,
+      "loss": 0.1839,
+      "step": 603
+    },
+    {
+      "epoch": 4.631881282910483,
+      "grad_norm": 0.2784336379714144,
+      "learning_rate": 6.071472117074462e-07,
+      "loss": 0.1553,
+      "step": 604
+    },
+    {
+      "epoch": 4.6395404499760655,
+      "grad_norm": 0.2908490571653058,
+      "learning_rate": 5.811636514789598e-07,
+      "loss": 0.1735,
+      "step": 605
+    },
+    {
+      "epoch": 4.647199617041647,
+      "grad_norm": 0.2689077917871218,
+      "learning_rate": 5.557401198050327e-07,
+      "loss": 0.1406,
+      "step": 606
+    },
+    {
+      "epoch": 4.654858784107228,
+      "grad_norm": 0.30361924900865483,
+      "learning_rate": 5.308773498863251e-07,
+      "loss": 0.1806,
+      "step": 607
+    },
+    {
+      "epoch": 4.66251795117281,
+      "grad_norm": 0.2738298236426758,
+      "learning_rate": 5.065760587514446e-07,
+      "loss": 0.1555,
+      "step": 608
+    },
+    {
+      "epoch": 4.670177118238391,
+      "grad_norm": 0.28066674173939915,
+      "learning_rate": 4.828369472362493e-07,
+      "loss": 0.1456,
+      "step": 609
+    },
+    {
+      "epoch": 4.677836285303973,
+      "grad_norm": 0.28254726890674403,
+      "learning_rate": 4.5966069996365993e-07,
+      "loss": 0.1666,
+      "step": 610
+    },
+    {
+      "epoch": 4.685495452369555,
+      "grad_norm": 0.28356791946970256,
+      "learning_rate": 4.3704798532388624e-07,
+      "loss": 0.1623,
+      "step": 611
+    },
+    {
+      "epoch": 4.693154619435137,
+      "grad_norm": 0.27978608677096656,
+      "learning_rate": 4.1499945545518283e-07,
+      "loss": 0.1573,
+      "step": 612
+    },
+    {
+      "epoch": 4.700813786500718,
+      "grad_norm": 0.2807595614948837,
+      "learning_rate": 3.935157462250128e-07,
+      "loss": 0.1645,
+      "step": 613
+    },
+    {
+      "epoch": 4.7084729535663,
+      "grad_norm": 0.2903423678415902,
+      "learning_rate": 3.7259747721173134e-07,
+      "loss": 0.1697,
+      "step": 614
+    },
+    {
+      "epoch": 4.716132120631881,
+      "grad_norm": 0.27761326426161137,
+      "learning_rate": 3.522452516867048e-07,
+      "loss": 0.154,
+      "step": 615
+    },
+    {
+      "epoch": 4.723791287697463,
+      "grad_norm": 0.2867472262572132,
+      "learning_rate": 3.324596565969174e-07,
+      "loss": 0.1565,
+      "step": 616
+    },
+    {
+      "epoch": 4.731450454763045,
+      "grad_norm": 0.28465179062440776,
+      "learning_rate": 3.1324126254804524e-07,
+      "loss": 0.1401,
+      "step": 617
+    },
+    {
+      "epoch": 4.739109621828626,
+      "grad_norm": 0.2947996164689416,
+      "learning_rate": 2.9459062378799806e-07,
+      "loss": 0.1754,
+      "step": 618
+    },
+    {
+      "epoch": 4.746768788894208,
+      "grad_norm": 0.2863382862190152,
+      "learning_rate": 2.7650827819093005e-07,
+      "loss": 0.1599,
+      "step": 619
+    },
+    {
+      "epoch": 4.754427955959789,
+      "grad_norm": 0.2709108924581573,
+      "learning_rate": 2.5899474724174313e-07,
+      "loss": 0.1398,
+      "step": 620
+    },
+    {
+      "epoch": 4.7620871230253705,
+      "grad_norm": 0.30022047839665467,
+      "learning_rate": 2.4205053602103015e-07,
+      "loss": 0.1792,
+      "step": 621
+    },
+    {
+      "epoch": 4.769746290090953,
+      "grad_norm": 0.2981470620905006,
+      "learning_rate": 2.2567613319051997e-07,
+      "loss": 0.1449,
+      "step": 622
+    },
+    {
+      "epoch": 4.777405457156534,
+      "grad_norm": 0.2875561082245844,
+      "learning_rate": 2.0987201097897757e-07,
+      "loss": 0.1395,
+      "step": 623
+    },
+    {
+      "epoch": 4.785064624222116,
+      "grad_norm": 0.27525535566929177,
+      "learning_rate": 1.9463862516859277e-07,
+      "loss": 0.1436,
+      "step": 624
+    },
+    {
+      "epoch": 4.792723791287697,
+      "grad_norm": 0.2940141095677646,
+      "learning_rate": 1.799764150818306e-07,
+      "loss": 0.1655,
+      "step": 625
+    },
+    {
+      "epoch": 4.800382958353279,
+      "grad_norm": 0.3081953681996578,
+      "learning_rate": 1.658858035687594e-07,
+      "loss": 0.1667,
+      "step": 626
+    },
+    {
+      "epoch": 4.808042125418861,
+      "grad_norm": 0.26637485304480285,
+      "learning_rate": 1.5236719699486256e-07,
+      "loss": 0.1309,
+      "step": 627
+    },
+    {
+      "epoch": 4.815701292484443,
+      "grad_norm": 0.2705718541638546,
+      "learning_rate": 1.3942098522931491e-07,
+      "loss": 0.1613,
+      "step": 628
+    },
+    {
+      "epoch": 4.823360459550024,
+      "grad_norm": 0.3013254867336271,
+      "learning_rate": 1.2704754163374022e-07,
+      "loss": 0.1871,
+      "step": 629
+    },
+    {
+      "epoch": 4.831019626615605,
+      "grad_norm": 0.2781156531330604,
+      "learning_rate": 1.1524722305144231e-07,
+      "loss": 0.1583,
+      "step": 630
+    },
+    {
+      "epoch": 4.838678793681187,
+      "grad_norm": 0.27865348896247544,
+      "learning_rate": 1.0402036979711317e-07,
+      "loss": 0.1541,
+      "step": 631
+    },
+    {
+      "epoch": 4.8463379607467685,
+      "grad_norm": 0.3006123678070983,
+      "learning_rate": 9.336730564702745e-08,
+      "loss": 0.177,
+      "step": 632
+    },
+    {
+      "epoch": 4.853997127812351,
+      "grad_norm": 0.2933355244436529,
+      "learning_rate": 8.328833782969003e-08,
+      "loss": 0.1735,
+      "step": 633
+    },
+    {
+      "epoch": 4.861656294877932,
+      "grad_norm": 0.2612274384851061,
+      "learning_rate": 7.378375701698748e-08,
+      "loss": 0.142,
+      "step": 634
+    },
+    {
+      "epoch": 4.869315461943514,
+      "grad_norm": 0.27506364744141204,
+      "learning_rate": 6.485383731580142e-08,
+      "loss": 0.1599,
+      "step": 635
+    },
+    {
+      "epoch": 4.876974629009095,
+      "grad_norm": 0.2740746508966746,
+      "learning_rate": 5.649883626009933e-08,
+      "loss": 0.1515,
+      "step": 636
+    },
+    {
+      "epoch": 4.884633796074677,
+      "grad_norm": 0.2923776048680673,
+      "learning_rate": 4.871899480351605e-08,
+      "loss": 0.1614,
+      "step": 637
+    },
+    {
+      "epoch": 4.8922929631402585,
+      "grad_norm": 0.2563881864024306,
+      "learning_rate": 4.151453731239707e-08,
+      "loss": 0.1431,
+      "step": 638
+    },
+    {
+      "epoch": 4.89995213020584,
+      "grad_norm": 0.31290090906865226,
+      "learning_rate": 3.4885671559332645e-08,
+      "loss": 0.1929,
+      "step": 639
+    },
+    {
+      "epoch": 4.907611297271422,
+      "grad_norm": 0.2833641820727045,
+      "learning_rate": 2.8832588717164766e-08,
+      "loss": 0.1354,
+      "step": 640
+    },
+    {
+      "epoch": 4.915270464337003,
+      "grad_norm": 0.2886325657329948,
+      "learning_rate": 2.3355463353467168e-08,
+      "loss": 0.1543,
+      "step": 641
+    },
+    {
+      "epoch": 4.922929631402585,
+      "grad_norm": 0.27914356344167157,
+      "learning_rate": 1.8454453425527098e-08,
+      "loss": 0.1511,
+      "step": 642
+    },
+    {
+      "epoch": 4.9305887984681664,
+      "grad_norm": 0.3094563610476224,
+      "learning_rate": 1.4129700275771208e-08,
+      "loss": 0.1843,
+      "step": 643
+    },
+    {
+      "epoch": 4.938247965533749,
+      "grad_norm": 0.25323095165276716,
+      "learning_rate": 1.0381328627702136e-08,
+      "loss": 0.1128,
+      "step": 644
+    },
+    {
+      "epoch": 4.94590713259933,
+      "grad_norm": 0.2790763770971488,
+      "learning_rate": 7.209446582292501e-09,
+      "loss": 0.1528,
+      "step": 645
+    },
+    {
+      "epoch": 4.953566299664912,
+      "grad_norm": 0.294849164666762,
+      "learning_rate": 4.614145614876275e-09,
+      "loss": 0.1657,
+      "step": 646
+    },
+    {
+      "epoch": 4.961225466730493,
+      "grad_norm": 0.30506766088706516,
+      "learning_rate": 2.5955005725064597e-09,
+      "loss": 0.178,
+      "step": 647
+    },
+    {
+      "epoch": 4.968884633796074,
+      "grad_norm": 0.281891446902876,
+      "learning_rate": 1.1535696717945855e-09,
+      "loss": 0.1473,
+      "step": 648
+    },
+    {
+      "epoch": 4.9765438008616565,
+      "grad_norm": 0.2745560443089173,
+      "learning_rate": 2.8839449723205847e-10,
+      "loss": 0.1509,
+      "step": 649
+    },
+    {
+      "epoch": 4.984202967927238,
+      "grad_norm": 0.2816379591846223,
+      "learning_rate": 0.0,
+      "loss": 0.1471,
+      "step": 650
+    },
+    {
+      "epoch": 4.984202967927238,
+      "step": 650,
+      "total_flos": 1.3070689797343805e+18,
+      "train_loss": 0.36027532513325033,
+      "train_runtime": 68390.8162,
+      "train_samples_per_second": 1.222,
+      "train_steps_per_second": 0.01
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 650,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.3070689797343805e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}