diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6172 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9995724668661823,
+  "eval_steps": 500,
+  "global_step": 877,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003420265070542967,
+      "grad_norm": 10.156329759741077,
+      "learning_rate": 0.0,
+      "loss": 2.1575,
+      "step": 1
+    },
+    {
+      "epoch": 0.006840530141085934,
+      "grad_norm": 184.3295302770049,
+      "learning_rate": 1.41e-05,
+      "loss": 4.88,
+      "step": 2
+    },
+    {
+      "epoch": 0.010260795211628902,
+      "grad_norm": 182.65124398096702,
+      "learning_rate": 1.41e-05,
+      "loss": 4.8994,
+      "step": 3
+    },
+    {
+      "epoch": 0.013681060282171868,
+      "grad_norm": 30.32918049437796,
+      "learning_rate": 1.41e-05,
+      "loss": 3.5419,
+      "step": 4
+    },
+    {
+      "epoch": 0.017101325352714837,
+      "grad_norm": 16.444831424361816,
+      "learning_rate": 1.41e-05,
+      "loss": 2.7311,
+      "step": 5
+    },
+    {
+      "epoch": 0.020521590423257803,
+      "grad_norm": 6.633419850116959,
+      "learning_rate": 1.41e-05,
+      "loss": 2.2268,
+      "step": 6
+    },
+    {
+      "epoch": 0.02394185549380077,
+      "grad_norm": 9.401679436345786,
+      "learning_rate": 1.41e-05,
+      "loss": 2.1353,
+      "step": 7
+    },
+    {
+      "epoch": 0.027362120564343735,
+      "grad_norm": 7.95097945213086,
+      "learning_rate": 1.41e-05,
+      "loss": 2.0867,
+      "step": 8
+    },
+    {
+      "epoch": 0.030782385634886705,
+      "grad_norm": 3.9617869212350962,
+      "learning_rate": 1.41e-05,
+      "loss": 1.9823,
+      "step": 9
+    },
+    {
+      "epoch": 0.034202650705429674,
+      "grad_norm": 9.402637091305385,
+      "learning_rate": 1.41e-05,
+      "loss": 1.9361,
+      "step": 10
+    },
+    {
+      "epoch": 0.03762291577597264,
+      "grad_norm": 3.7983366071849862,
+      "learning_rate": 1.41e-05,
+      "loss": 1.9295,
+      "step": 11
+    },
+    {
+      "epoch": 0.041043180846515606,
+      "grad_norm": 4.558710902554969,
+      "learning_rate": 1.41e-05,
+      "loss": 1.8787,
+      "step": 12
+    },
+    {
+      "epoch": 0.04446344591705857,
+      "grad_norm": 3.9168701968301134,
+      "learning_rate": 1.41e-05,
+      "loss": 1.7936,
+      "step": 13
+    },
+    {
+      "epoch": 0.04788371098760154,
+      "grad_norm": 3.763989925966716,
+      "learning_rate": 1.41e-05,
+      "loss": 1.7617,
+      "step": 14
+    },
+    {
+      "epoch": 0.051303976058144504,
+      "grad_norm": 3.4821708929604007,
+      "learning_rate": 1.41e-05,
+      "loss": 1.6881,
+      "step": 15
+    },
+    {
+      "epoch": 0.05472424112868747,
+      "grad_norm": 2.1426267148325997,
+      "learning_rate": 1.41e-05,
+      "loss": 1.6905,
+      "step": 16
+    },
+    {
+      "epoch": 0.05814450619923044,
+      "grad_norm": 3.6919608929938255,
+      "learning_rate": 1.41e-05,
+      "loss": 1.6473,
+      "step": 17
+    },
+    {
+      "epoch": 0.06156477126977341,
+      "grad_norm": 2.725651684433171,
+      "learning_rate": 1.41e-05,
+      "loss": 1.7064,
+      "step": 18
+    },
+    {
+      "epoch": 0.06498503634031637,
+      "grad_norm": 2.9127756737882824,
+      "learning_rate": 1.41e-05,
+      "loss": 1.5901,
+      "step": 19
+    },
+    {
+      "epoch": 0.06840530141085935,
+      "grad_norm": 2.664503176902388,
+      "learning_rate": 1.41e-05,
+      "loss": 1.6196,
+      "step": 20
+    },
+    {
+      "epoch": 0.07182556648140231,
+      "grad_norm": 2.5256162285977077,
+      "learning_rate": 1.41e-05,
+      "loss": 1.4734,
+      "step": 21
+    },
+    {
+      "epoch": 0.07524583155194528,
+      "grad_norm": 2.232926984374301,
+      "learning_rate": 1.41e-05,
+      "loss": 1.4862,
+      "step": 22
+    },
+    {
+      "epoch": 0.07866609662248825,
+      "grad_norm": 1.8467296080556872,
+      "learning_rate": 1.41e-05,
+      "loss": 1.4919,
+      "step": 23
+    },
+    {
+      "epoch": 0.08208636169303121,
+      "grad_norm": 2.133405884576976,
+      "learning_rate": 1.41e-05,
+      "loss": 1.4962,
+      "step": 24
+    },
+    {
+      "epoch": 0.08550662676357418,
+      "grad_norm": 2.0723150961914256,
+      "learning_rate": 1.41e-05,
+      "loss": 1.3745,
+      "step": 25
+    },
+    {
+      "epoch": 0.08892689183411714,
+      "grad_norm": 2.2473775199610166,
+      "learning_rate": 1.41e-05,
+      "loss": 1.4678,
+      "step": 26
+    },
+    {
+      "epoch": 0.09234715690466011,
+      "grad_norm": 3.1366913970429366,
+      "learning_rate": 1.41e-05,
+      "loss": 1.3231,
+      "step": 27
+    },
+    {
+      "epoch": 0.09576742197520308,
+      "grad_norm": 2.2703890784758585,
+      "learning_rate": 1.41e-05,
+      "loss": 1.3328,
+      "step": 28
+    },
+    {
+      "epoch": 0.09918768704574604,
+      "grad_norm": 2.7735799657439943,
+      "learning_rate": 1.41e-05,
+      "loss": 1.1703,
+      "step": 29
+    },
+    {
+      "epoch": 0.10260795211628901,
+      "grad_norm": 2.5181744230002203,
+      "learning_rate": 1.41e-05,
+      "loss": 1.257,
+      "step": 30
+    },
+    {
+      "epoch": 0.10602821718683197,
+      "grad_norm": 2.5187681530121546,
+      "learning_rate": 1.41e-05,
+      "loss": 1.1321,
+      "step": 31
+    },
+    {
+      "epoch": 0.10944848225737494,
+      "grad_norm": 2.5085106975094678,
+      "learning_rate": 1.41e-05,
+      "loss": 1.0929,
+      "step": 32
+    },
+    {
+      "epoch": 0.11286874732791792,
+      "grad_norm": 2.1171154989402887,
+      "learning_rate": 1.41e-05,
+      "loss": 1.1053,
+      "step": 33
+    },
+    {
+      "epoch": 0.11628901239846089,
+      "grad_norm": 2.4016503905187725,
+      "learning_rate": 1.41e-05,
+      "loss": 0.9514,
+      "step": 34
+    },
+    {
+      "epoch": 0.11970927746900385,
+      "grad_norm": 4.505185827517653,
+      "learning_rate": 1.41e-05,
+      "loss": 1.1443,
+      "step": 35
+    },
+    {
+      "epoch": 0.12312954253954682,
+      "grad_norm": 2.283031446044169,
+      "learning_rate": 1.41e-05,
+      "loss": 1.1263,
+      "step": 36
+    },
+    {
+      "epoch": 0.12654980761008977,
+      "grad_norm": 2.474396323802316,
+      "learning_rate": 1.41e-05,
+      "loss": 0.9587,
+      "step": 37
+    },
+    {
+      "epoch": 0.12997007268063274,
+      "grad_norm": 3.3515325695241667,
+      "learning_rate": 1.41e-05,
+      "loss": 0.8727,
+      "step": 38
+    },
+    {
+      "epoch": 0.1333903377511757,
+      "grad_norm": 2.3607499413214694,
+      "learning_rate": 1.41e-05,
+      "loss": 0.929,
+      "step": 39
+    },
+    {
+      "epoch": 0.1368106028217187,
+      "grad_norm": 2.4723425187830537,
+      "learning_rate": 1.41e-05,
+      "loss": 1.0473,
+      "step": 40
+    },
+    {
+      "epoch": 0.14023086789226166,
+      "grad_norm": 2.183114877668465,
+      "learning_rate": 1.41e-05,
+      "loss": 0.854,
+      "step": 41
+    },
+    {
+      "epoch": 0.14365113296280463,
+      "grad_norm": 2.3660790712775555,
+      "learning_rate": 1.41e-05,
+      "loss": 0.8244,
+      "step": 42
+    },
+    {
+      "epoch": 0.1470713980333476,
+      "grad_norm": 1.9551361934244411,
+      "learning_rate": 1.41e-05,
+      "loss": 0.8276,
+      "step": 43
+    },
+    {
+      "epoch": 0.15049166310389056,
+      "grad_norm": 3.163303383692252,
+      "learning_rate": 1.41e-05,
+      "loss": 0.8468,
+      "step": 44
+    },
+    {
+      "epoch": 0.15391192817443353,
+      "grad_norm": 3.163426556149129,
+      "learning_rate": 1.41e-05,
+      "loss": 0.7213,
+      "step": 45
+    },
+    {
+      "epoch": 0.1573321932449765,
+      "grad_norm": 2.7934537634938783,
+      "learning_rate": 1.41e-05,
+      "loss": 0.7035,
+      "step": 46
+    },
+    {
+      "epoch": 0.16075245831551946,
+      "grad_norm": 1.8747838269439079,
+      "learning_rate": 1.41e-05,
+      "loss": 0.7371,
+      "step": 47
+    },
+    {
+      "epoch": 0.16417272338606242,
+      "grad_norm": 2.7812346413247484,
+      "learning_rate": 1.41e-05,
+      "loss": 0.7602,
+      "step": 48
+    },
+    {
+      "epoch": 0.1675929884566054,
+      "grad_norm": 2.378687443172595,
+      "learning_rate": 1.41e-05,
+      "loss": 0.8561,
+      "step": 49
+    },
+    {
+      "epoch": 0.17101325352714836,
+      "grad_norm": 2.3489013022006056,
+      "learning_rate": 1.41e-05,
+      "loss": 0.5867,
+      "step": 50
+    },
+    {
+      "epoch": 0.17443351859769132,
+      "grad_norm": 1.7892416152344746,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4853,
+      "step": 51
+    },
+    {
+      "epoch": 0.1778537836682343,
+      "grad_norm": 2.077063902741533,
+      "learning_rate": 1.41e-05,
+      "loss": 0.6375,
+      "step": 52
+    },
+    {
+      "epoch": 0.18127404873877725,
+      "grad_norm": 2.2312628788967412,
+      "learning_rate": 1.41e-05,
+      "loss": 0.6439,
+      "step": 53
+    },
+    {
+      "epoch": 0.18469431380932022,
+      "grad_norm": 2.996527586876452,
+      "learning_rate": 1.41e-05,
+      "loss": 0.5925,
+      "step": 54
+    },
+    {
+      "epoch": 0.1881145788798632,
+      "grad_norm": 1.8130852934136357,
+      "learning_rate": 1.41e-05,
+      "loss": 0.5575,
+      "step": 55
+    },
+    {
+      "epoch": 0.19153484395040615,
+      "grad_norm": 1.5587429488330231,
+      "learning_rate": 1.41e-05,
+      "loss": 0.5954,
+      "step": 56
+    },
+    {
+      "epoch": 0.19495510902094912,
+      "grad_norm": 1.7839250019187394,
+      "learning_rate": 1.41e-05,
+      "loss": 0.6196,
+      "step": 57
+    },
+    {
+      "epoch": 0.19837537409149208,
+      "grad_norm": 1.6933241820211493,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4969,
+      "step": 58
+    },
+    {
+      "epoch": 0.20179563916203505,
+      "grad_norm": 1.731887729628178,
+      "learning_rate": 1.41e-05,
+      "loss": 0.6548,
+      "step": 59
+    },
+    {
+      "epoch": 0.20521590423257802,
+      "grad_norm": 2.3987513930499995,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4595,
+      "step": 60
+    },
+    {
+      "epoch": 0.20863616930312098,
+      "grad_norm": 2.639090838839165,
+      "learning_rate": 1.41e-05,
+      "loss": 0.5341,
+      "step": 61
+    },
+    {
+      "epoch": 0.21205643437366395,
+      "grad_norm": 1.7474798765435617,
+      "learning_rate": 1.41e-05,
+      "loss": 0.6015,
+      "step": 62
+    },
+    {
+      "epoch": 0.21547669944420692,
+      "grad_norm": 3.571763528033457,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4851,
+      "step": 63
+    },
+    {
+      "epoch": 0.21889696451474988,
+      "grad_norm": 1.9011170860485487,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4911,
+      "step": 64
+    },
+    {
+      "epoch": 0.22231722958529285,
+      "grad_norm": 2.625704311611419,
+      "learning_rate": 1.41e-05,
+      "loss": 0.5247,
+      "step": 65
+    },
+    {
+      "epoch": 0.22573749465583584,
+      "grad_norm": 1.8272075489091404,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4123,
+      "step": 66
+    },
+    {
+      "epoch": 0.2291577597263788,
+      "grad_norm": 2.005801690308512,
+      "learning_rate": 1.41e-05,
+      "loss": 0.437,
+      "step": 67
+    },
+    {
+      "epoch": 0.23257802479692177,
+      "grad_norm": 2.009191411617363,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4553,
+      "step": 68
+    },
+    {
+      "epoch": 0.23599828986746474,
+      "grad_norm": 1.9856248091054856,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4911,
+      "step": 69
+    },
+    {
+      "epoch": 0.2394185549380077,
+      "grad_norm": 1.9582222549850024,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4345,
+      "step": 70
+    },
+    {
+      "epoch": 0.24283882000855067,
+      "grad_norm": 1.6336088110301379,
+      "learning_rate": 1.41e-05,
+      "loss": 0.4649,
+      "step": 71
+    },
+    {
+      "epoch": 0.24625908507909364,
+      "grad_norm": 1.946765368736228,
+      "learning_rate": 1.41e-05,
+      "loss": 0.3968,
+      "step": 72
+    },
+    {
+      "epoch": 0.2496793501496366,
+      "grad_norm": 1.9515164464490413,
+      "learning_rate": 1.41e-05,
+      "loss": 0.405,
+      "step": 73
+    },
+    {
+      "epoch": 0.25309961522017954,
+      "grad_norm": 1.5507200892542412,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2883,
+      "step": 74
+    },
+    {
+      "epoch": 0.25651988029072254,
+      "grad_norm": 1.8953946078798778,
+      "learning_rate": 1.41e-05,
+      "loss": 0.3697,
+      "step": 75
+    },
+    {
+      "epoch": 0.2599401453612655,
+      "grad_norm": 1.609567252811076,
+      "learning_rate": 1.41e-05,
+      "loss": 0.3912,
+      "step": 76
+    },
+    {
+      "epoch": 0.26336041043180847,
+      "grad_norm": 1.841628814862342,
+      "learning_rate": 1.41e-05,
+      "loss": 0.3753,
+      "step": 77
+    },
+    {
+      "epoch": 0.2667806755023514,
+      "grad_norm": 1.3866520663429722,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2978,
+      "step": 78
+    },
+    {
+      "epoch": 0.2702009405728944,
+      "grad_norm": 1.8662166292908873,
+      "learning_rate": 1.41e-05,
+      "loss": 0.325,
+      "step": 79
+    },
+    {
+      "epoch": 0.2736212056434374,
+      "grad_norm": 1.86990469712952,
+      "learning_rate": 1.41e-05,
+      "loss": 0.3608,
+      "step": 80
+    },
+    {
+      "epoch": 0.27704147071398033,
+      "grad_norm": 1.7776490861424512,
+      "learning_rate": 1.41e-05,
+      "loss": 0.3164,
+      "step": 81
+    },
+    {
+      "epoch": 0.2804617357845233,
+      "grad_norm": 1.5257018693275182,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2736,
+      "step": 82
+    },
+    {
+      "epoch": 0.28388200085506626,
+      "grad_norm": 1.2107786177311144,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2293,
+      "step": 83
+    },
+    {
+      "epoch": 0.28730226592560926,
+      "grad_norm": 1.7075082762966491,
+      "learning_rate": 1.41e-05,
+      "loss": 0.3312,
+      "step": 84
+    },
+    {
+      "epoch": 0.2907225309961522,
+      "grad_norm": 1.4271608707593932,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2731,
+      "step": 85
+    },
+    {
+      "epoch": 0.2941427960666952,
+      "grad_norm": 2.0611639591955466,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2708,
+      "step": 86
+    },
+    {
+      "epoch": 0.2975630611372381,
+      "grad_norm": 1.2754227728205574,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2301,
+      "step": 87
+    },
+    {
+      "epoch": 0.3009833262077811,
+      "grad_norm": 1.312428216280633,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2219,
+      "step": 88
+    },
+    {
+      "epoch": 0.30440359127832406,
+      "grad_norm": 1.36584246397569,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2761,
+      "step": 89
+    },
+    {
+      "epoch": 0.30782385634886705,
+      "grad_norm": 1.1768002468324985,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1884,
+      "step": 90
+    },
+    {
+      "epoch": 0.31124412141941,
+      "grad_norm": 1.402779753210287,
+      "learning_rate": 1.41e-05,
+      "loss": 0.24,
+      "step": 91
+    },
+    {
+      "epoch": 0.314664386489953,
+      "grad_norm": 1.2917943074724656,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2487,
+      "step": 92
+    },
+    {
+      "epoch": 0.3180846515604959,
+      "grad_norm": 1.4168579440960474,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2429,
+      "step": 93
+    },
+    {
+      "epoch": 0.3215049166310389,
+      "grad_norm": 1.1040710078361045,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1676,
+      "step": 94
+    },
+    {
+      "epoch": 0.32492518170158186,
+      "grad_norm": 1.6692824066389471,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2456,
+      "step": 95
+    },
+    {
+      "epoch": 0.32834544677212485,
+      "grad_norm": 1.4794418835245668,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2346,
+      "step": 96
+    },
+    {
+      "epoch": 0.3317657118426678,
+      "grad_norm": 1.2426804107195764,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2286,
+      "step": 97
+    },
+    {
+      "epoch": 0.3351859769132108,
+      "grad_norm": 1.5976518274152767,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2356,
+      "step": 98
+    },
+    {
+      "epoch": 0.3386062419837537,
+      "grad_norm": 1.3380817380576575,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2361,
+      "step": 99
+    },
+    {
+      "epoch": 0.3420265070542967,
+      "grad_norm": 1.8499907335313168,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2144,
+      "step": 100
+    },
+    {
+      "epoch": 0.34544677212483965,
+      "grad_norm": 1.5519650604852626,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2211,
+      "step": 101
+    },
+    {
+      "epoch": 0.34886703719538265,
+      "grad_norm": 1.4110088914262258,
+      "learning_rate": 1.41e-05,
+      "loss": 0.2422,
+      "step": 102
+    },
+    {
+      "epoch": 0.3522873022659256,
+      "grad_norm": 1.4052513404508866,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1939,
+      "step": 103
+    },
+    {
+      "epoch": 0.3557075673364686,
+      "grad_norm": 1.377026410169766,
+      "learning_rate": 1.41e-05,
+      "loss": 0.193,
+      "step": 104
+    },
+    {
+      "epoch": 0.3591278324070115,
+      "grad_norm": 1.0901968649772857,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1705,
+      "step": 105
+    },
+    {
+      "epoch": 0.3625480974775545,
+      "grad_norm": 1.2716891700474584,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1998,
+      "step": 106
+    },
+    {
+      "epoch": 0.3659683625480975,
+      "grad_norm": 1.269092339983086,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1845,
+      "step": 107
+    },
+    {
+      "epoch": 0.36938862761864044,
+      "grad_norm": 1.0134148691503204,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1457,
+      "step": 108
+    },
+    {
+      "epoch": 0.37280889268918344,
+      "grad_norm": 1.0973618617348355,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1829,
+      "step": 109
+    },
+    {
+      "epoch": 0.3762291577597264,
+      "grad_norm": 1.3476960032405298,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1936,
+      "step": 110
+    },
+    {
+      "epoch": 0.37964942283026937,
+      "grad_norm": 1.1243588113049168,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1763,
+      "step": 111
+    },
+    {
+      "epoch": 0.3830696879008123,
+      "grad_norm": 1.756776902549336,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1772,
+      "step": 112
+    },
+    {
+      "epoch": 0.3864899529713553,
+      "grad_norm": 1.0653616123456815,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1485,
+      "step": 113
+    },
+    {
+      "epoch": 0.38991021804189824,
+      "grad_norm": 1.322506116172751,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1511,
+      "step": 114
+    },
+    {
+      "epoch": 0.39333048311244123,
+      "grad_norm": 1.134939568700316,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1595,
+      "step": 115
+    },
+    {
+      "epoch": 0.39675074818298417,
+      "grad_norm": 1.2351108355732012,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1653,
+      "step": 116
+    },
+    {
+      "epoch": 0.40017101325352716,
+      "grad_norm": 1.291174811726909,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1414,
+      "step": 117
+    },
+    {
+      "epoch": 0.4035912783240701,
+      "grad_norm": 1.3110441911359814,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1602,
+      "step": 118
+    },
+    {
+      "epoch": 0.4070115433946131,
+      "grad_norm": 1.0324677945393839,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1589,
+      "step": 119
+    },
+    {
+      "epoch": 0.41043180846515603,
+      "grad_norm": 1.1528407686685203,
+      "learning_rate": 1.41e-05,
+      "loss": 0.142,
+      "step": 120
+    },
+    {
+      "epoch": 0.413852073535699,
+      "grad_norm": 1.259359755456077,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1671,
+      "step": 121
+    },
+    {
+      "epoch": 0.41727233860624197,
+      "grad_norm": 0.9342347863650876,
+      "learning_rate": 1.41e-05,
+      "loss": 0.123,
+      "step": 122
+    },
+    {
+      "epoch": 0.42069260367678496,
+      "grad_norm": 1.0262297732751853,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1379,
+      "step": 123
+    },
+    {
+      "epoch": 0.4241128687473279,
+      "grad_norm": 0.8528122922890459,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1115,
+      "step": 124
+    },
+    {
+      "epoch": 0.4275331338178709,
+      "grad_norm": 1.0773688633850005,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1368,
+      "step": 125
+    },
+    {
+      "epoch": 0.43095339888841383,
+      "grad_norm": 1.02543648603311,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1116,
+      "step": 126
+    },
+    {
+      "epoch": 0.4343736639589568,
+      "grad_norm": 0.8824964139082571,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1149,
+      "step": 127
+    },
+    {
+      "epoch": 0.43779392902949976,
+      "grad_norm": 1.1649358884512249,
+      "learning_rate": 1.41e-05,
+      "loss": 0.127,
+      "step": 128
+    },
+    {
+      "epoch": 0.44121419410004276,
+      "grad_norm": 1.0614948335318424,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1248,
+      "step": 129
+    },
+    {
+      "epoch": 0.4446344591705857,
+      "grad_norm": 0.7776250289623791,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1145,
+      "step": 130
+    },
+    {
+      "epoch": 0.4480547242411287,
+      "grad_norm": 0.8384984456348001,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0921,
+      "step": 131
+    },
+    {
+      "epoch": 0.4514749893116717,
+      "grad_norm": 1.0151085870429353,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1226,
+      "step": 132
+    },
+    {
+      "epoch": 0.4548952543822146,
+      "grad_norm": 0.7615407647681977,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0951,
+      "step": 133
+    },
+    {
+      "epoch": 0.4583155194527576,
+      "grad_norm": 0.9555465270405811,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1186,
+      "step": 134
+    },
+    {
+      "epoch": 0.46173578452330055,
+      "grad_norm": 0.9186643738206431,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1207,
+      "step": 135
+    },
+    {
+      "epoch": 0.46515604959384355,
+      "grad_norm": 0.8860210174403064,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1195,
+      "step": 136
+    },
+    {
+      "epoch": 0.4685763146643865,
+      "grad_norm": 0.8817179419098865,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0897,
+      "step": 137
+    },
+    {
+      "epoch": 0.4719965797349295,
+      "grad_norm": 0.8289163487116727,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1003,
+      "step": 138
+    },
+    {
+      "epoch": 0.4754168448054724,
+      "grad_norm": 0.904736501387183,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0924,
+      "step": 139
+    },
+    {
+      "epoch": 0.4788371098760154,
+      "grad_norm": 0.943604807746642,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1109,
+      "step": 140
+    },
+    {
+      "epoch": 0.48225737494655835,
+      "grad_norm": 0.9402997510757618,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1145,
+      "step": 141
+    },
+    {
+      "epoch": 0.48567764001710134,
+      "grad_norm": 0.7366551780058475,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0899,
+      "step": 142
+    },
+    {
+      "epoch": 0.4890979050876443,
+      "grad_norm": 0.9208800933128714,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1167,
+      "step": 143
+    },
+    {
+      "epoch": 0.4925181701581873,
+      "grad_norm": 0.750448679907497,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0809,
+      "step": 144
+    },
+    {
+      "epoch": 0.4959384352287302,
+      "grad_norm": 0.8092636444718945,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0908,
+      "step": 145
+    },
+    {
+      "epoch": 0.4993587002992732,
+      "grad_norm": 0.8625942305572842,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1031,
+      "step": 146
+    },
+    {
+      "epoch": 0.5027789653698161,
+      "grad_norm": 0.9198105373299891,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1113,
+      "step": 147
+    },
+    {
+      "epoch": 0.5061992304403591,
+      "grad_norm": 0.7848012242788545,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0865,
+      "step": 148
+    },
+    {
+      "epoch": 0.5096194955109021,
+      "grad_norm": 0.9324979893254519,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0971,
+      "step": 149
+    },
+    {
+      "epoch": 0.5130397605814451,
+      "grad_norm": 0.7143510040718761,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0795,
+      "step": 150
+    },
+    {
+      "epoch": 0.516460025651988,
+      "grad_norm": 1.0676584465757966,
+      "learning_rate": 1.41e-05,
+      "loss": 0.1135,
+      "step": 151
+    },
+    {
+      "epoch": 0.519880290722531,
+      "grad_norm": 0.6822736052229258,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0733,
+      "step": 152
+    },
+    {
+      "epoch": 0.523300555793074,
+      "grad_norm": 0.8374258007686984,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0907,
+      "step": 153
+    },
+    {
+      "epoch": 0.5267208208636169,
+      "grad_norm": 0.8341896688419776,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0939,
+      "step": 154
+    },
+    {
+      "epoch": 0.5301410859341599,
+      "grad_norm": 0.7611257220630606,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0869,
+      "step": 155
+    },
+    {
+      "epoch": 0.5335613510047028,
+      "grad_norm": 0.740325417856638,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0832,
+      "step": 156
+    },
+    {
+      "epoch": 0.5369816160752459,
+      "grad_norm": 0.7042263268553958,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0716,
+      "step": 157
+    },
+    {
+      "epoch": 0.5404018811457888,
+      "grad_norm": 0.7335797206103793,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0807,
+      "step": 158
+    },
+    {
+      "epoch": 0.5438221462163317,
+      "grad_norm": 0.8019722685081757,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0826,
+      "step": 159
+    },
+    {
+      "epoch": 0.5472424112868748,
+      "grad_norm": 0.6929468279193534,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0821,
+      "step": 160
+    },
+    {
+      "epoch": 0.5506626763574177,
+      "grad_norm": 0.6854252563729888,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0675,
+      "step": 161
+    },
+    {
+      "epoch": 0.5540829414279607,
+      "grad_norm": 0.6741146081895844,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0749,
+      "step": 162
+    },
+    {
+      "epoch": 0.5575032064985036,
+      "grad_norm": 0.6902694268516201,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0707,
+      "step": 163
+    },
+    {
+      "epoch": 0.5609234715690466,
+      "grad_norm": 0.6948144741970704,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0795,
+      "step": 164
+    },
+    {
+      "epoch": 0.5643437366395896,
+      "grad_norm": 0.7169974641783955,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0818,
+      "step": 165
+    },
+    {
+      "epoch": 0.5677640017101325,
+      "grad_norm": 0.6384211122986987,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0735,
+      "step": 166
+    },
+    {
+      "epoch": 0.5711842667806755,
+      "grad_norm": 0.7417356609403256,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0842,
+      "step": 167
+    },
+    {
+      "epoch": 0.5746045318512185,
+      "grad_norm": 0.66647458408933,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0814,
+      "step": 168
+    },
+    {
+      "epoch": 0.5780247969217615,
+      "grad_norm": 0.7378707726234803,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0746,
+      "step": 169
+    },
+    {
+      "epoch": 0.5814450619923044,
+      "grad_norm": 0.6812399388437269,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0828,
+      "step": 170
+    },
+    {
+      "epoch": 0.5848653270628473,
+      "grad_norm": 0.6793042032294047,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0724,
+      "step": 171
+    },
+    {
+      "epoch": 0.5882855921333904,
+      "grad_norm": 0.6062435025746024,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0649,
+      "step": 172
+    },
+    {
+      "epoch": 0.5917058572039333,
+      "grad_norm": 0.6813501327595134,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0711,
+      "step": 173
+    },
+    {
+      "epoch": 0.5951261222744763,
+      "grad_norm": 0.672907340612349,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0769,
+      "step": 174
+    },
+    {
+      "epoch": 0.5985463873450192,
+      "grad_norm": 0.5410302293555107,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0588,
+      "step": 175
+    },
+    {
+      "epoch": 0.6019666524155622,
+      "grad_norm": 0.6636988853462809,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0702,
+      "step": 176
+    },
+    {
+      "epoch": 0.6053869174861052,
+      "grad_norm": 0.5918082791413835,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0626,
+      "step": 177
+    },
+    {
+      "epoch": 0.6088071825566481,
+      "grad_norm": 0.5816411742116699,
+      "learning_rate": 1.41e-05,
+      "loss": 0.059,
+      "step": 178
+    },
+    {
+      "epoch": 0.6122274476271911,
+      "grad_norm": 0.6819230663398479,
+      "learning_rate": 1.41e-05,
+      "loss": 0.074,
+      "step": 179
+    },
+    {
+      "epoch": 0.6156477126977341,
+      "grad_norm": 0.6727834641069249,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0739,
+      "step": 180
+    },
+    {
+      "epoch": 0.619067977768277,
+      "grad_norm": 0.5791689945437091,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0553,
+      "step": 181
+    },
+    {
+      "epoch": 0.62248824283882,
+      "grad_norm": 0.6060693467771211,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0572,
+      "step": 182
+    },
+    {
+      "epoch": 0.6259085079093629,
+      "grad_norm": 0.6611378853926027,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0697,
+      "step": 183
+    },
+    {
+      "epoch": 0.629328772979906,
+      "grad_norm": 0.5887572335679387,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0603,
+      "step": 184
+    },
+    {
+      "epoch": 0.6327490380504489,
+      "grad_norm": 0.538930391171475,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0462,
+      "step": 185
+    },
+    {
+      "epoch": 0.6361693031209918,
+      "grad_norm": 0.5617056384496442,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0549,
+      "step": 186
+    },
+    {
+      "epoch": 0.6395895681915349,
+      "grad_norm": 0.5912693387471951,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0574,
+      "step": 187
+    },
+    {
+      "epoch": 0.6430098332620778,
+      "grad_norm": 0.5373216387973052,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0574,
+      "step": 188
+    },
+    {
+      "epoch": 0.6464300983326208,
+      "grad_norm": 0.6151671164129469,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0533,
+      "step": 189
+    },
+    {
+      "epoch": 0.6498503634031637,
+      "grad_norm": 0.5394742707247884,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0575,
+      "step": 190
+    },
+    {
+      "epoch": 0.6532706284737068,
+      "grad_norm": 0.5752514447611141,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0574,
+      "step": 191
+    },
+    {
+      "epoch": 0.6566908935442497,
+      "grad_norm": 0.5136422669182581,
+      "learning_rate": 1.41e-05,
+      "loss": 0.054,
+      "step": 192
+    },
+    {
+      "epoch": 0.6601111586147926,
+      "grad_norm": 0.6261951776293332,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0612,
+      "step": 193
+    },
+    {
+      "epoch": 0.6635314236853356,
+      "grad_norm": 0.5067466055288193,
+      "learning_rate": 1.41e-05,
+      "loss": 0.054,
+      "step": 194
+    },
+    {
+      "epoch": 0.6669516887558786,
+      "grad_norm": 0.5892942817895197,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0572,
+      "step": 195
+    },
+    {
+      "epoch": 0.6703719538264216,
+      "grad_norm": 0.5652390088377635,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0655,
+      "step": 196
+    },
+    {
+      "epoch": 0.6737922188969645,
+      "grad_norm": 0.5285822929717092,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0494,
+      "step": 197
+    },
+    {
+      "epoch": 0.6772124839675074,
+      "grad_norm": 0.5246191872665474,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0614,
+      "step": 198
+    },
+    {
+      "epoch": 0.6806327490380505,
+      "grad_norm": 0.5387443642901717,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0531,
+      "step": 199
+    },
+    {
+      "epoch": 0.6840530141085934,
+      "grad_norm": 0.5238421425157995,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0512,
+      "step": 200
+    },
+    {
+      "epoch": 0.6874732791791364,
+      "grad_norm": 0.4882221825014573,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0489,
+      "step": 201
+    },
+    {
+      "epoch": 0.6908935442496793,
+      "grad_norm": 0.5646516308611281,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0512,
+      "step": 202
+    },
+    {
+      "epoch": 0.6943138093202224,
+      "grad_norm": 0.5778234440206057,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0634,
+      "step": 203
+    },
+    {
+      "epoch": 0.6977340743907653,
+      "grad_norm": 0.5293741220834561,
+      "learning_rate": 1.41e-05,
+      "loss": 0.056,
+      "step": 204
+    },
+    {
+      "epoch": 0.7011543394613082,
+      "grad_norm": 0.527641578215631,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0561,
+      "step": 205
+    },
+    {
+      "epoch": 0.7045746045318512,
+      "grad_norm": 0.4000174523279179,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0401,
+      "step": 206
+    },
+    {
+      "epoch": 0.7079948696023942,
+      "grad_norm": 0.5298298097123045,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0609,
+      "step": 207
+    },
+    {
+      "epoch": 0.7114151346729372,
+      "grad_norm": 0.5349126180633413,
+      "learning_rate": 1.41e-05,
+      "loss": 0.052,
+      "step": 208
+    },
+    {
+      "epoch": 0.7148353997434801,
+      "grad_norm": 0.5748237411918808,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0651,
+      "step": 209
+    },
+    {
+      "epoch": 0.718255664814023,
+      "grad_norm": 0.4989630175919984,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0502,
+      "step": 210
+    },
+    {
+      "epoch": 0.7216759298845661,
+      "grad_norm": 0.5604803895822472,
+      "learning_rate": 1.41e-05,
+      "loss": 0.054,
+      "step": 211
+    },
+    {
+      "epoch": 0.725096194955109,
+      "grad_norm": 0.6062121706000653,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0644,
+      "step": 212
+    },
+    {
+      "epoch": 0.728516460025652,
+      "grad_norm": 0.49966896896386376,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0469,
+      "step": 213
+    },
+    {
+      "epoch": 0.731936725096195,
+      "grad_norm": 0.44454274508391683,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0428,
+      "step": 214
+    },
+    {
+      "epoch": 0.735356990166738,
+      "grad_norm": 0.563352438281472,
+      "learning_rate": 1.41e-05,
+      "loss": 0.055,
+      "step": 215
+    },
+    {
+      "epoch": 0.7387772552372809,
+      "grad_norm": 0.6191641486840816,
+      "learning_rate": 1.41e-05,
+      "loss": 0.063,
+      "step": 216
+    },
+    {
+      "epoch": 0.7421975203078238,
+      "grad_norm": 0.4827104401750106,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0496,
+      "step": 217
+    },
+    {
+      "epoch": 0.7456177853783669,
+      "grad_norm": 0.4656663916077845,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0528,
+      "step": 218
+    },
+    {
+      "epoch": 0.7490380504489098,
+      "grad_norm": 0.4785719764753492,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0438,
+      "step": 219
+    },
+    {
+      "epoch": 0.7524583155194527,
+      "grad_norm": 0.5448096052982832,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0422,
+      "step": 220
+    },
+    {
+      "epoch": 0.7558785805899957,
+      "grad_norm": 0.5065090698982245,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0489,
+      "step": 221
+    },
+    {
+      "epoch": 0.7592988456605387,
+      "grad_norm": 0.5059637979656728,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0485,
+      "step": 222
+    },
+    {
+      "epoch": 0.7627191107310817,
+      "grad_norm": 0.49583806553345944,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0525,
+      "step": 223
+    },
+    {
+      "epoch": 0.7661393758016246,
+      "grad_norm": 0.4012288566084756,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0384,
+      "step": 224
+    },
+    {
+      "epoch": 0.7695596408721675,
+      "grad_norm": 0.49449469823574593,
+      "learning_rate": 1.41e-05,
+      "loss": 0.05,
+      "step": 225
+    },
+    {
+      "epoch": 0.7729799059427106,
+      "grad_norm": 0.45672465215152086,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0479,
+      "step": 226
+    },
+    {
+      "epoch": 0.7764001710132535,
+      "grad_norm": 0.4909318818146214,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0499,
+      "step": 227
+    },
+    {
+      "epoch": 0.7798204360837965,
+      "grad_norm": 0.40392311014383353,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0406,
+      "step": 228
+    },
+    {
+      "epoch": 0.7832407011543394,
+      "grad_norm": 0.4131849222227493,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0388,
+      "step": 229
+    },
+    {
+      "epoch": 0.7866609662248825,
+      "grad_norm": 0.5611423817088044,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0594,
+      "step": 230
+    },
+    {
+      "epoch": 0.7900812312954254,
+      "grad_norm": 0.47364645404517464,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0522,
+      "step": 231
+    },
+    {
+      "epoch": 0.7935014963659683,
+      "grad_norm": 0.4545208834696141,
+      "learning_rate": 1.41e-05,
+      "loss": 0.048,
+      "step": 232
+    },
+    {
+      "epoch": 0.7969217614365113,
+      "grad_norm": 0.4113439560012879,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0393,
+      "step": 233
+    },
+    {
+      "epoch": 0.8003420265070543,
+      "grad_norm": 0.42334212760511825,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0404,
+      "step": 234
+    },
+    {
+      "epoch": 0.8037622915775973,
+      "grad_norm": 0.4768224768125407,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0479,
+      "step": 235
+    },
+    {
+      "epoch": 0.8071825566481402,
+      "grad_norm": 0.43678389875922824,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0442,
+      "step": 236
+    },
+    {
+      "epoch": 0.8106028217186833,
+      "grad_norm": 0.43277663891476426,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0409,
+      "step": 237
+    },
+    {
+      "epoch": 0.8140230867892262,
+      "grad_norm": 0.4443462520817696,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0465,
+      "step": 238
+    },
+    {
+      "epoch": 0.8174433518597691,
+      "grad_norm": 0.43839684525150946,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0415,
+      "step": 239
+    },
+    {
+      "epoch": 0.8208636169303121,
+      "grad_norm": 0.43297499486580887,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0445,
+      "step": 240
+    },
+    {
+      "epoch": 0.8242838820008551,
+      "grad_norm": 0.38283138235459124,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0404,
+      "step": 241
+    },
+    {
+      "epoch": 0.827704147071398,
+      "grad_norm": 0.40333504070789256,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0373,
+      "step": 242
+    },
+    {
+      "epoch": 0.831124412141941,
+      "grad_norm": 0.40950403615291003,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0423,
+      "step": 243
+    },
+    {
+      "epoch": 0.8345446772124839,
+      "grad_norm": 0.4314317792707956,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0424,
+      "step": 244
+    },
+    {
+      "epoch": 0.837964942283027,
+      "grad_norm": 0.4189847423801003,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0434,
+      "step": 245
+    },
+    {
+      "epoch": 0.8413852073535699,
+      "grad_norm": 0.4593355477826361,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0456,
+      "step": 246
+    },
+    {
+      "epoch": 0.8448054724241129,
+      "grad_norm": 0.43148788855113257,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0429,
+      "step": 247
+    },
+    {
+      "epoch": 0.8482257374946558,
+      "grad_norm": 0.41015663281431336,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0411,
+      "step": 248
+    },
+    {
+      "epoch": 0.8516460025651988,
+      "grad_norm": 0.39479744422344626,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0408,
+      "step": 249
+    },
+    {
+      "epoch": 0.8550662676357418,
+      "grad_norm": 0.49034951176740266,
+      "learning_rate": 1.41e-05,
+      "loss": 0.056,
+      "step": 250
+    },
+    {
+      "epoch": 0.8584865327062847,
+      "grad_norm": 0.4241786968876316,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0414,
+      "step": 251
+    },
+    {
+      "epoch": 0.8619067977768277,
+      "grad_norm": 0.40856481199229994,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0384,
+      "step": 252
+    },
+    {
+      "epoch": 0.8653270628473707,
+      "grad_norm": 0.44863803613683295,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0454,
+      "step": 253
+    },
+    {
+      "epoch": 0.8687473279179136,
+      "grad_norm": 0.4315245075802853,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0425,
+      "step": 254
+    },
+    {
+      "epoch": 0.8721675929884566,
+      "grad_norm": 0.4300931739828693,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0386,
+      "step": 255
+    },
+    {
+      "epoch": 0.8755878580589995,
+      "grad_norm": 0.41516523943726863,
+      "learning_rate": 1.41e-05,
+      "loss": 0.044,
+      "step": 256
+    },
+    {
+      "epoch": 0.8790081231295426,
+      "grad_norm": 0.41978138385172276,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0376,
+      "step": 257
+    },
+    {
+      "epoch": 0.8824283882000855,
+      "grad_norm": 0.38286432398443565,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0357,
+      "step": 258
+    },
+    {
+      "epoch": 0.8858486532706284,
+      "grad_norm": 1.0190380922512265,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0413,
+      "step": 259
+    },
+    {
+      "epoch": 0.8892689183411714,
+      "grad_norm": 0.46857802366186146,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0468,
+      "step": 260
+    },
+    {
+      "epoch": 0.8926891834117144,
+      "grad_norm": 0.38236119116632356,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0383,
+      "step": 261
+    },
+    {
+      "epoch": 0.8961094484822574,
+      "grad_norm": 0.4551658885616909,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0453,
+      "step": 262
+    },
+    {
+      "epoch": 0.8995297135528003,
+      "grad_norm": 0.4286695716016865,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0408,
+      "step": 263
+    },
+    {
+      "epoch": 0.9029499786233434,
+      "grad_norm": 0.43951997404486826,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0395,
+      "step": 264
+    },
+    {
+      "epoch": 0.9063702436938863,
+      "grad_norm": 0.4395704115846384,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0419,
+      "step": 265
+    },
+    {
+      "epoch": 0.9097905087644292,
+      "grad_norm": 0.3839941648623069,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0374,
+      "step": 266
+    },
+    {
+      "epoch": 0.9132107738349722,
+      "grad_norm": 0.922460042226267,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0372,
+      "step": 267
+    },
+    {
+      "epoch": 0.9166310389055152,
+      "grad_norm": 0.44811463225682846,
+      "learning_rate": 1.41e-05,
+      "loss": 0.043,
+      "step": 268
+    },
+    {
+      "epoch": 0.9200513039760582,
+      "grad_norm": 0.4172986063615246,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0365,
+      "step": 269
+    },
+    {
+      "epoch": 0.9234715690466011,
+      "grad_norm": 0.37219225103541853,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0377,
+      "step": 270
+    },
+    {
+      "epoch": 0.926891834117144,
+      "grad_norm": 0.3447899677572012,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0317,
+      "step": 271
+    },
+    {
+      "epoch": 0.9303120991876871,
+      "grad_norm": 0.3964175357323583,
+      "learning_rate": 1.41e-05,
+      "loss": 0.034,
+      "step": 272
+    },
+    {
+      "epoch": 0.93373236425823,
+      "grad_norm": 0.36028858741706155,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0351,
+      "step": 273
+    },
+    {
+      "epoch": 0.937152629328773,
+      "grad_norm": 0.46256283545817356,
+      "learning_rate": 1.41e-05,
+      "loss": 0.041,
+      "step": 274
+    },
+    {
+      "epoch": 0.9405728943993159,
+      "grad_norm": 0.3781578237183448,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0366,
+      "step": 275
+    },
+    {
+      "epoch": 0.943993159469859,
+      "grad_norm": 0.3818392999731623,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0321,
+      "step": 276
+    },
+    {
+      "epoch": 0.9474134245404019,
+      "grad_norm": 0.38568204387768645,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0336,
+      "step": 277
+    },
+    {
+      "epoch": 0.9508336896109448,
+      "grad_norm": 0.3905057560704898,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0349,
+      "step": 278
+    },
+    {
+      "epoch": 0.9542539546814878,
+      "grad_norm": 0.36210340527903007,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0351,
+      "step": 279
+    },
+    {
+      "epoch": 0.9576742197520308,
+      "grad_norm": 0.3784578196301756,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0382,
+      "step": 280
+    },
+    {
+      "epoch": 0.9610944848225738,
+      "grad_norm": 0.4415630974468222,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0417,
+      "step": 281
+    },
+    {
+      "epoch": 0.9645147498931167,
+      "grad_norm": 0.36220744727441767,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0336,
+      "step": 282
+    },
+    {
+      "epoch": 0.9679350149636596,
+      "grad_norm": 0.4266084587936558,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0384,
+      "step": 283
+    },
+    {
+      "epoch": 0.9713552800342027,
+      "grad_norm": 0.3599843427563046,
+      "learning_rate": 1.41e-05,
+      "loss": 0.035,
+      "step": 284
+    },
+    {
+      "epoch": 0.9747755451047456,
+      "grad_norm": 0.38722401771389997,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0348,
+      "step": 285
+    },
+    {
+      "epoch": 0.9781958101752886,
+      "grad_norm": 0.40946001056055625,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0373,
+      "step": 286
+    },
+    {
+      "epoch": 0.9816160752458315,
+      "grad_norm": 0.41550323303820474,
+      "learning_rate": 1.41e-05,
+      "loss": 0.04,
+      "step": 287
+    },
+    {
+      "epoch": 0.9850363403163745,
+      "grad_norm": 0.37048272944847027,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0309,
+      "step": 288
+    },
+    {
+      "epoch": 0.9884566053869175,
+      "grad_norm": 0.3620011070515116,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0341,
+      "step": 289
+    },
+    {
+      "epoch": 0.9918768704574604,
+      "grad_norm": 0.4132102658215653,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0383,
+      "step": 290
+    },
+    {
+      "epoch": 0.9952971355280035,
+      "grad_norm": 0.3852351778306183,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0357,
+      "step": 291
+    },
+    {
+      "epoch": 0.9987174005985464,
+      "grad_norm": 0.37414254856973184,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0348,
+      "step": 292
+    },
+    {
+      "epoch": 1.0021376656690895,
+      "grad_norm": 0.31841258787737364,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0308,
+      "step": 293
+    },
+    {
+      "epoch": 1.0055579307396323,
+      "grad_norm": 0.30923263652050564,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0263,
+      "step": 294
+    },
+    {
+      "epoch": 1.0089781958101753,
+      "grad_norm": 0.3118052788978316,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0243,
+      "step": 295
+    },
+    {
+      "epoch": 1.0123984608807182,
+      "grad_norm": 0.2908942931886208,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0239,
+      "step": 296
+    },
+    {
+      "epoch": 1.0158187259512612,
+      "grad_norm": 0.2904620406307295,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0222,
+      "step": 297
+    },
+    {
+      "epoch": 1.0192389910218043,
+      "grad_norm": 0.28692807204183246,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0225,
+      "step": 298
+    },
+    {
+      "epoch": 1.022659256092347,
+      "grad_norm": 0.32077856448530445,
+      "learning_rate": 1.41e-05,
+      "loss": 0.025,
+      "step": 299
+    },
+    {
+      "epoch": 1.0260795211628901,
+      "grad_norm": 0.28877834524497115,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0214,
+      "step": 300
+    },
+    {
+      "epoch": 1.0294997862334332,
+      "grad_norm": 0.27974144729840145,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0226,
+      "step": 301
+    },
+    {
+      "epoch": 1.032920051303976,
+      "grad_norm": 0.2976755981973845,
+      "learning_rate": 1.41e-05,
+      "loss": 0.025,
+      "step": 302
+    },
+    {
+      "epoch": 1.036340316374519,
+      "grad_norm": 0.2909288296995041,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0241,
+      "step": 303
+    },
+    {
+      "epoch": 1.039760581445062,
+      "grad_norm": 0.26580362507255084,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0227,
+      "step": 304
+    },
+    {
+      "epoch": 1.043180846515605,
+      "grad_norm": 0.3375634201266733,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0261,
+      "step": 305
+    },
+    {
+      "epoch": 1.046601111586148,
+      "grad_norm": 0.2797681439324777,
+      "learning_rate": 1.41e-05,
+      "loss": 0.023,
+      "step": 306
+    },
+    {
+      "epoch": 1.0500213766566908,
+      "grad_norm": 0.3197672217160271,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0258,
+      "step": 307
+    },
+    {
+      "epoch": 1.0534416417272339,
+      "grad_norm": 0.2834421342617412,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0231,
+      "step": 308
+    },
+    {
+      "epoch": 1.056861906797777,
+      "grad_norm": 0.30638464271397325,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0237,
+      "step": 309
+    },
+    {
+      "epoch": 1.0602821718683197,
+      "grad_norm": 0.2772969145815675,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0227,
+      "step": 310
+    },
+    {
+      "epoch": 1.0637024369388628,
+      "grad_norm": 0.27940000335528603,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0232,
+      "step": 311
+    },
+    {
+      "epoch": 1.0671227020094056,
+      "grad_norm": 0.2793001905583242,
+      "learning_rate": 1.41e-05,
+      "loss": 0.024,
+      "step": 312
+    },
+    {
+      "epoch": 1.0705429670799487,
+      "grad_norm": 0.2615333153159237,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0219,
+      "step": 313
+    },
+    {
+      "epoch": 1.0739632321504917,
+      "grad_norm": 0.2876480036224866,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0253,
+      "step": 314
+    },
+    {
+      "epoch": 1.0773834972210345,
+      "grad_norm": 0.30478384773242695,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0252,
+      "step": 315
+    },
+    {
+      "epoch": 1.0808037622915776,
+      "grad_norm": 0.2528846190302502,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0206,
+      "step": 316
+    },
+    {
+      "epoch": 1.0842240273621206,
+      "grad_norm": 0.2435449168695854,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0224,
+      "step": 317
+    },
+    {
+      "epoch": 1.0876442924326635,
+      "grad_norm": 0.3079236522189953,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0253,
+      "step": 318
+    },
+    {
+      "epoch": 1.0910645575032065,
+      "grad_norm": 0.2675432502865711,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0239,
+      "step": 319
+    },
+    {
+      "epoch": 1.0944848225737496,
+      "grad_norm": 0.3004587131742902,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0239,
+      "step": 320
+    },
+    {
+      "epoch": 1.0979050876442924,
+      "grad_norm": 0.29332378399603554,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0247,
+      "step": 321
+    },
+    {
+      "epoch": 1.1013253527148354,
+      "grad_norm": 0.23762802516720216,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0242,
+      "step": 322
+    },
+    {
+      "epoch": 1.1047456177853783,
+      "grad_norm": 0.26342688417785715,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0244,
+      "step": 323
+    },
+    {
+      "epoch": 1.1081658828559213,
+      "grad_norm": 0.3055498089807232,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0265,
+      "step": 324
+    },
+    {
+      "epoch": 1.1115861479264644,
+      "grad_norm": 0.23787182811525248,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0226,
+      "step": 325
+    },
+    {
+      "epoch": 1.1150064129970072,
+      "grad_norm": 0.2647940637873088,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0242,
+      "step": 326
+    },
+    {
+      "epoch": 1.1184266780675503,
+      "grad_norm": 0.2913294201237873,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0261,
+      "step": 327
+    },
+    {
+      "epoch": 1.1218469431380933,
+      "grad_norm": 0.2581915579915153,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0228,
+      "step": 328
+    },
+    {
+      "epoch": 1.1252672082086361,
+      "grad_norm": 0.2500171384316944,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0206,
+      "step": 329
+    },
+    {
+      "epoch": 1.1286874732791792,
+      "grad_norm": 0.2689440388389032,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0237,
+      "step": 330
+    },
+    {
+      "epoch": 1.1321077383497222,
+      "grad_norm": 0.2551497009606492,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0216,
+      "step": 331
+    },
+    {
+      "epoch": 1.135528003420265,
+      "grad_norm": 0.25757413049163996,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0224,
+      "step": 332
+    },
+    {
+      "epoch": 1.138948268490808,
+      "grad_norm": 0.2699506485343775,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0218,
+      "step": 333
+    },
+    {
+      "epoch": 1.142368533561351,
+      "grad_norm": 0.25767905743807257,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0227,
+      "step": 334
+    },
+    {
+      "epoch": 1.145788798631894,
+      "grad_norm": 0.2273460686067317,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0226,
+      "step": 335
+    },
+    {
+      "epoch": 1.149209063702437,
+      "grad_norm": 0.31200680170359746,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0196,
+      "step": 336
+    },
+    {
+      "epoch": 1.1526293287729799,
+      "grad_norm": 0.23994204361556526,
+      "learning_rate": 1.41e-05,
+      "loss": 0.022,
+      "step": 337
+    },
+    {
+      "epoch": 1.156049593843523,
+      "grad_norm": 0.2867059420441838,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0256,
+      "step": 338
+    },
+    {
+      "epoch": 1.1594698589140657,
+      "grad_norm": 0.2523475692407159,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0212,
+      "step": 339
+    },
+    {
+      "epoch": 1.1628901239846088,
+      "grad_norm": 0.24499307458764866,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0198,
+      "step": 340
+    },
+    {
+      "epoch": 1.1663103890551518,
+      "grad_norm": 0.2436283905820348,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0216,
+      "step": 341
+    },
+    {
+      "epoch": 1.1697306541256947,
+      "grad_norm": 0.25602523324809817,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0216,
+      "step": 342
+    },
+    {
+      "epoch": 1.1731509191962377,
+      "grad_norm": 0.25021177697750296,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0233,
+      "step": 343
+    },
+    {
+      "epoch": 1.1765711842667808,
+      "grad_norm": 0.28379855080073285,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0259,
+      "step": 344
+    },
+    {
+      "epoch": 1.1799914493373236,
+      "grad_norm": 0.25642024168179856,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0208,
+      "step": 345
+    },
+    {
+      "epoch": 1.1834117144078666,
+      "grad_norm": 0.24053576849839317,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0206,
+      "step": 346
+    },
+    {
+      "epoch": 1.1868319794784097,
+      "grad_norm": 0.26458302891901536,
+      "learning_rate": 1.41e-05,
+      "loss": 0.025,
+      "step": 347
+    },
+    {
+      "epoch": 1.1902522445489525,
+      "grad_norm": 0.25340954518832737,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0232,
+      "step": 348
+    },
+    {
+      "epoch": 1.1936725096194956,
+      "grad_norm": 0.2822910057628071,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0258,
+      "step": 349
+    },
+    {
+      "epoch": 1.1970927746900384,
+      "grad_norm": 0.24401825117116366,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0207,
+      "step": 350
+    },
+    {
+      "epoch": 1.2005130397605814,
+      "grad_norm": 0.25351547147013626,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0232,
+      "step": 351
+    },
+    {
+      "epoch": 1.2039333048311245,
+      "grad_norm": 0.2514277079230116,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0236,
+      "step": 352
+    },
+    {
+      "epoch": 1.2073535699016673,
+      "grad_norm": 0.2499624840027022,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0218,
+      "step": 353
+    },
+    {
+      "epoch": 1.2107738349722104,
+      "grad_norm": 0.25541034452698597,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0226,
+      "step": 354
+    },
+    {
+      "epoch": 1.2141941000427534,
+      "grad_norm": 0.26116617349172355,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0233,
+      "step": 355
+    },
+    {
+      "epoch": 1.2176143651132962,
+      "grad_norm": 0.3144220810007225,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0238,
+      "step": 356
+    },
+    {
+      "epoch": 1.2210346301838393,
+      "grad_norm": 0.2397648420075889,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0218,
+      "step": 357
+    },
+    {
+      "epoch": 1.2244548952543823,
+      "grad_norm": 0.23056631938102404,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0211,
+      "step": 358
+    },
+    {
+      "epoch": 1.2278751603249252,
+      "grad_norm": 0.6359133071495683,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0221,
+      "step": 359
+    },
+    {
+      "epoch": 1.2312954253954682,
+      "grad_norm": 0.2494306454740628,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0212,
+      "step": 360
+    },
+    {
+      "epoch": 1.234715690466011,
+      "grad_norm": 0.2661512294789303,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0254,
+      "step": 361
+    },
+    {
+      "epoch": 1.238135955536554,
+      "grad_norm": 0.7710920857502603,
+      "learning_rate": 1.41e-05,
+      "loss": 0.025,
+      "step": 362
+    },
+    {
+      "epoch": 1.2415562206070971,
+      "grad_norm": 0.3942637887185519,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0236,
+      "step": 363
+    },
+    {
+      "epoch": 1.24497648567764,
+      "grad_norm": 0.27087010860427124,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0252,
+      "step": 364
+    },
+    {
+      "epoch": 1.248396750748183,
+      "grad_norm": 0.24919842057781497,
+      "learning_rate": 1.41e-05,
+      "loss": 0.022,
+      "step": 365
+    },
+    {
+      "epoch": 1.2518170158187258,
+      "grad_norm": 0.247824008548709,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0235,
+      "step": 366
+    },
+    {
+      "epoch": 1.255237280889269,
+      "grad_norm": 0.2113113650016047,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0202,
+      "step": 367
+    },
+    {
+      "epoch": 1.258657545959812,
+      "grad_norm": 0.2259169472785117,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0192,
+      "step": 368
+    },
+    {
+      "epoch": 1.262077811030355,
+      "grad_norm": 0.2752050707801685,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0254,
+      "step": 369
+    },
+    {
+      "epoch": 1.2654980761008978,
+      "grad_norm": 0.23830392409535447,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0229,
+      "step": 370
+    },
+    {
+      "epoch": 1.2689183411714409,
+      "grad_norm": 0.24302497993054667,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0219,
+      "step": 371
+    },
+    {
+      "epoch": 1.2723386062419837,
+      "grad_norm": 0.24504347526605771,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0224,
+      "step": 372
+    },
+    {
+      "epoch": 1.2757588713125267,
+      "grad_norm": 0.2346626865644856,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0209,
+      "step": 373
+    },
+    {
+      "epoch": 1.2791791363830698,
+      "grad_norm": 0.24245589828220748,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0218,
+      "step": 374
+    },
+    {
+      "epoch": 1.2825994014536126,
+      "grad_norm": 0.282140360657227,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0234,
+      "step": 375
+    },
+    {
+      "epoch": 1.2860196665241557,
+      "grad_norm": 0.26692130820889887,
+      "learning_rate": 1.41e-05,
+      "loss": 0.02,
+      "step": 376
+    },
+    {
+      "epoch": 1.2894399315946985,
+      "grad_norm": 0.2691231899902035,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0221,
+      "step": 377
+    },
+    {
+      "epoch": 1.2928601966652415,
+      "grad_norm": 0.2281687474950334,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0222,
+      "step": 378
+    },
+    {
+      "epoch": 1.2962804617357846,
+      "grad_norm": 0.373488505328009,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0217,
+      "step": 379
+    },
+    {
+      "epoch": 1.2997007268063274,
+      "grad_norm": 0.2345889821941022,
+      "learning_rate": 1.41e-05,
+      "loss": 0.021,
+      "step": 380
+    },
+    {
+      "epoch": 1.3031209918768705,
+      "grad_norm": 0.2105687923684266,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0204,
+      "step": 381
+    },
+    {
+      "epoch": 1.3065412569474133,
+      "grad_norm": 0.23464935916200427,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0213,
+      "step": 382
+    },
+    {
+      "epoch": 1.3099615220179563,
+      "grad_norm": 0.23646699592209017,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0211,
+      "step": 383
+    },
+    {
+      "epoch": 1.3133817870884994,
+      "grad_norm": 0.2631733185003215,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0219,
+      "step": 384
+    },
+    {
+      "epoch": 1.3168020521590424,
+      "grad_norm": 0.25931322588284866,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0229,
+      "step": 385
+    },
+    {
+      "epoch": 1.3202223172295853,
+      "grad_norm": 0.28688530125524797,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0226,
+      "step": 386
+    },
+    {
+      "epoch": 1.3236425823001283,
+      "grad_norm": 0.25729658498109925,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0229,
+      "step": 387
+    },
+    {
+      "epoch": 1.3270628473706712,
+      "grad_norm": 0.25760420815172724,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0207,
+      "step": 388
+    },
+    {
+      "epoch": 1.3304831124412142,
+      "grad_norm": 0.32657940169239236,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0217,
+      "step": 389
+    },
+    {
+      "epoch": 1.3339033775117572,
+      "grad_norm": 0.22557643602886415,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0204,
+      "step": 390
+    },
+    {
+      "epoch": 1.3373236425823,
+      "grad_norm": 0.2661843657709089,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0243,
+      "step": 391
+    },
+    {
+      "epoch": 1.3407439076528431,
+      "grad_norm": 0.2974504589558172,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0241,
+      "step": 392
+    },
+    {
+      "epoch": 1.344164172723386,
+      "grad_norm": 0.26843823846639436,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0234,
+      "step": 393
+    },
+    {
+      "epoch": 1.347584437793929,
+      "grad_norm": 0.3218845205293079,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0221,
+      "step": 394
+    },
+    {
+      "epoch": 1.351004702864472,
+      "grad_norm": 0.2428094674581459,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0222,
+      "step": 395
+    },
+    {
+      "epoch": 1.354424967935015,
+      "grad_norm": 0.2455927243682846,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0211,
+      "step": 396
+    },
+    {
+      "epoch": 1.357845233005558,
+      "grad_norm": 0.2733123552071639,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0212,
+      "step": 397
+    },
+    {
+      "epoch": 1.361265498076101,
+      "grad_norm": 0.2659612332359629,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0203,
+      "step": 398
+    },
+    {
+      "epoch": 1.3646857631466438,
+      "grad_norm": 0.25254417828478626,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0201,
+      "step": 399
+    },
+    {
+      "epoch": 1.3681060282171869,
+      "grad_norm": 0.2598602588125554,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0217,
+      "step": 400
+    },
+    {
+      "epoch": 1.37152629328773,
+      "grad_norm": 0.25849620548333013,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0209,
+      "step": 401
+    },
+    {
+      "epoch": 1.3749465583582727,
+      "grad_norm": 0.3137023430125288,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0211,
+      "step": 402
+    },
+    {
+      "epoch": 1.3783668234288158,
+      "grad_norm": 0.2113162287336589,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0195,
+      "step": 403
+    },
+    {
+      "epoch": 1.3817870884993586,
+      "grad_norm": 0.2503171178420045,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0234,
+      "step": 404
+    },
+    {
+      "epoch": 1.3852073535699017,
+      "grad_norm": 0.23119898543579737,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0209,
+      "step": 405
+    },
+    {
+      "epoch": 1.3886276186404447,
+      "grad_norm": 0.26039590115583117,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0244,
+      "step": 406
+    },
+    {
+      "epoch": 1.3920478837109875,
+      "grad_norm": 0.23886269607593336,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0223,
+      "step": 407
+    },
+    {
+      "epoch": 1.3954681487815306,
+      "grad_norm": 0.2714710828662534,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0247,
+      "step": 408
+    },
+    {
+      "epoch": 1.3988884138520734,
+      "grad_norm": 0.27319555229782644,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0244,
+      "step": 409
+    },
+    {
+      "epoch": 1.4023086789226165,
+      "grad_norm": 0.24019773624911636,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0218,
+      "step": 410
+    },
+    {
+      "epoch": 1.4057289439931595,
+      "grad_norm": 0.2708671308101268,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0219,
+      "step": 411
+    },
+    {
+      "epoch": 1.4091492090637026,
+      "grad_norm": 0.22702757974948617,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0198,
+      "step": 412
+    },
+    {
+      "epoch": 1.4125694741342454,
+      "grad_norm": 0.25555770375998416,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0224,
+      "step": 413
+    },
+    {
+      "epoch": 1.4159897392047884,
+      "grad_norm": 0.23780595691689027,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0213,
+      "step": 414
+    },
+    {
+      "epoch": 1.4194100042753313,
+      "grad_norm": 0.26451555701259444,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0211,
+      "step": 415
+    },
+    {
+      "epoch": 1.4228302693458743,
+      "grad_norm": 0.24436230077463678,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0226,
+      "step": 416
+    },
+    {
+      "epoch": 1.4262505344164174,
+      "grad_norm": 0.2216762882841438,
+      "learning_rate": 1.41e-05,
+      "loss": 0.021,
+      "step": 417
+    },
+    {
+      "epoch": 1.4296707994869602,
+      "grad_norm": 0.26405628674467824,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0219,
+      "step": 418
+    },
+    {
+      "epoch": 1.4330910645575032,
+      "grad_norm": 0.23921814427898216,
+      "learning_rate": 1.41e-05,
+      "loss": 0.021,
+      "step": 419
+    },
+    {
+      "epoch": 1.436511329628046,
+      "grad_norm": 0.265453432302215,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0226,
+      "step": 420
+    },
+    {
+      "epoch": 1.4399315946985891,
+      "grad_norm": 0.24492268359315303,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0227,
+      "step": 421
+    },
+    {
+      "epoch": 1.4433518597691322,
+      "grad_norm": 0.25191935351384814,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0215,
+      "step": 422
+    },
+    {
+      "epoch": 1.4467721248396752,
+      "grad_norm": 0.27197993846345997,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0229,
+      "step": 423
+    },
+    {
+      "epoch": 1.450192389910218,
+      "grad_norm": 0.2075718815542033,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0183,
+      "step": 424
+    },
+    {
+      "epoch": 1.453612654980761,
+      "grad_norm": 0.26376818444878,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0221,
+      "step": 425
+    },
+    {
+      "epoch": 1.457032920051304,
+      "grad_norm": 0.26080564628665626,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0229,
+      "step": 426
+    },
+    {
+      "epoch": 1.460453185121847,
+      "grad_norm": 0.2434588148667514,
+      "learning_rate": 1.41e-05,
+      "loss": 0.022,
+      "step": 427
+    },
+    {
+      "epoch": 1.46387345019239,
+      "grad_norm": 0.20943782903685337,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0185,
+      "step": 428
+    },
+    {
+      "epoch": 1.4672937152629328,
+      "grad_norm": 0.24301387125104526,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0208,
+      "step": 429
+    },
+    {
+      "epoch": 1.470713980333476,
+      "grad_norm": 0.25766203422027834,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0237,
+      "step": 430
+    },
+    {
+      "epoch": 1.4741342454040187,
+      "grad_norm": 0.21556702000374744,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0213,
+      "step": 431
+    },
+    {
+      "epoch": 1.4775545104745618,
+      "grad_norm": 0.23784909822124217,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0207,
+      "step": 432
+    },
+    {
+      "epoch": 1.4809747755451048,
+      "grad_norm": 0.22414128247781562,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0198,
+      "step": 433
+    },
+    {
+      "epoch": 1.4843950406156476,
+      "grad_norm": 0.23386987857579358,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0201,
+      "step": 434
+    },
+    {
+      "epoch": 1.4878153056861907,
+      "grad_norm": 0.2853961947266083,
+      "learning_rate": 1.41e-05,
+      "loss": 0.023,
+      "step": 435
+    },
+    {
+      "epoch": 1.4912355707567335,
+      "grad_norm": 0.24974092416116495,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0209,
+      "step": 436
+    },
+    {
+      "epoch": 1.4946558358272766,
+      "grad_norm": 0.23615654630506394,
+      "learning_rate": 1.41e-05,
+      "loss": 0.022,
+      "step": 437
+    },
+    {
+      "epoch": 1.4980761008978196,
+      "grad_norm": 0.2284456001474283,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0195,
+      "step": 438
+    },
+    {
+      "epoch": 1.5014963659683627,
+      "grad_norm": 0.26063959990204033,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0258,
+      "step": 439
+    },
+    {
+      "epoch": 1.5049166310389055,
+      "grad_norm": 0.24423575575966894,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0215,
+      "step": 440
+    },
+    {
+      "epoch": 1.5083368961094483,
+      "grad_norm": 0.24270403980054853,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0198,
+      "step": 441
+    },
+    {
+      "epoch": 1.5117571611799914,
+      "grad_norm": 0.24660594248828135,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0208,
+      "step": 442
+    },
+    {
+      "epoch": 1.5151774262505344,
+      "grad_norm": 0.23764056229245814,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0185,
+      "step": 443
+    },
+    {
+      "epoch": 1.5185976913210775,
+      "grad_norm": 0.2327212499624357,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0205,
+      "step": 444
+    },
+    {
+      "epoch": 1.5220179563916203,
+      "grad_norm": 0.2644477025039793,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0222,
+      "step": 445
+    },
+    {
+      "epoch": 1.5254382214621633,
+      "grad_norm": 0.2385202884230959,
+      "learning_rate": 1.41e-05,
+      "loss": 0.023,
+      "step": 446
+    },
+    {
+      "epoch": 1.5288584865327062,
+      "grad_norm": 0.24308327876991292,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0198,
+      "step": 447
+    },
+    {
+      "epoch": 1.5322787516032492,
+      "grad_norm": 0.30634205660699,
+      "learning_rate": 1.41e-05,
+      "loss": 0.026,
+      "step": 448
+    },
+    {
+      "epoch": 1.5356990166737923,
+      "grad_norm": 0.21217473130594544,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0199,
+      "step": 449
+    },
+    {
+      "epoch": 1.5391192817443353,
+      "grad_norm": 0.2658497297922587,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0221,
+      "step": 450
+    },
+    {
+      "epoch": 1.5425395468148781,
+      "grad_norm": 0.21481338352800414,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0197,
+      "step": 451
+    },
+    {
+      "epoch": 1.545959811885421,
+      "grad_norm": 0.27821761611964035,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0246,
+      "step": 452
+    },
+    {
+      "epoch": 1.549380076955964,
+      "grad_norm": 0.2416104328737188,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0216,
+      "step": 453
+    },
+    {
+      "epoch": 1.552800342026507,
+      "grad_norm": 0.2322000952546936,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0219,
+      "step": 454
+    },
+    {
+      "epoch": 1.5562206070970501,
+      "grad_norm": 0.23752597721176905,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0218,
+      "step": 455
+    },
+    {
+      "epoch": 1.559640872167593,
+      "grad_norm": 0.24308738119455534,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0205,
+      "step": 456
+    },
+    {
+      "epoch": 1.563061137238136,
+      "grad_norm": 0.26073824102297316,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0247,
+      "step": 457
+    },
+    {
+      "epoch": 1.5664814023086788,
+      "grad_norm": 0.24050011455971732,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0223,
+      "step": 458
+    },
+    {
+      "epoch": 1.5699016673792219,
+      "grad_norm": 0.21610207139710053,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0192,
+      "step": 459
+    },
+    {
+      "epoch": 1.573321932449765,
+      "grad_norm": 0.25801793315676924,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0236,
+      "step": 460
+    },
+    {
+      "epoch": 1.576742197520308,
+      "grad_norm": 0.21228131497792052,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0191,
+      "step": 461
+    },
+    {
+      "epoch": 1.5801624625908508,
+      "grad_norm": 0.25197778439538576,
+      "learning_rate": 1.41e-05,
+      "loss": 0.022,
+      "step": 462
+    },
+    {
+      "epoch": 1.5835827276613936,
+      "grad_norm": 0.2565258961786049,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0219,
+      "step": 463
+    },
+    {
+      "epoch": 1.5870029927319367,
+      "grad_norm": 0.2559923791329211,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0217,
+      "step": 464
+    },
+    {
+      "epoch": 1.5904232578024797,
+      "grad_norm": 0.2535527705987336,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0221,
+      "step": 465
+    },
+    {
+      "epoch": 1.5938435228730228,
+      "grad_norm": 0.28495146888910583,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0272,
+      "step": 466
+    },
+    {
+      "epoch": 1.5972637879435656,
+      "grad_norm": 0.26151357261060504,
+      "learning_rate": 1.41e-05,
+      "loss": 0.022,
+      "step": 467
+    },
+    {
+      "epoch": 1.6006840530141084,
+      "grad_norm": 0.24889528453574525,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0218,
+      "step": 468
+    },
+    {
+      "epoch": 1.6041043180846515,
+      "grad_norm": 0.21219419933729458,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0176,
+      "step": 469
+    },
+    {
+      "epoch": 1.6075245831551945,
+      "grad_norm": 0.24995940249087767,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0207,
+      "step": 470
+    },
+    {
+      "epoch": 1.6109448482257376,
+      "grad_norm": 0.23890689663178338,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0204,
+      "step": 471
+    },
+    {
+      "epoch": 1.6143651132962804,
+      "grad_norm": 0.23090495982296502,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0209,
+      "step": 472
+    },
+    {
+      "epoch": 1.6177853783668235,
+      "grad_norm": 0.24984846656221682,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0191,
+      "step": 473
+    },
+    {
+      "epoch": 1.6212056434373663,
+      "grad_norm": 0.2864149291763989,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0243,
+      "step": 474
+    },
+    {
+      "epoch": 1.6246259085079093,
+      "grad_norm": 0.23371681508272213,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0197,
+      "step": 475
+    },
+    {
+      "epoch": 1.6280461735784524,
+      "grad_norm": 0.21807560914421245,
+      "learning_rate": 1.41e-05,
+      "loss": 0.019,
+      "step": 476
+    },
+    {
+      "epoch": 1.6314664386489954,
+      "grad_norm": 0.24325192491721911,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0212,
+      "step": 477
+    },
+    {
+      "epoch": 1.6348867037195383,
+      "grad_norm": 0.21802834032775958,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0183,
+      "step": 478
+    },
+    {
+      "epoch": 1.638306968790081,
+      "grad_norm": 0.25015961868430003,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0206,
+      "step": 479
+    },
+    {
+      "epoch": 1.6417272338606241,
+      "grad_norm": 0.2047993695964016,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0166,
+      "step": 480
+    },
+    {
+      "epoch": 1.6451474989311672,
+      "grad_norm": 0.24407449809464615,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0209,
+      "step": 481
+    },
+    {
+      "epoch": 1.6485677640017102,
+      "grad_norm": 0.22905836500660803,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0197,
+      "step": 482
+    },
+    {
+      "epoch": 1.651988029072253,
+      "grad_norm": 0.2302659533712275,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0184,
+      "step": 483
+    },
+    {
+      "epoch": 1.655408294142796,
+      "grad_norm": 0.22469778783055896,
+      "learning_rate": 1.41e-05,
+      "loss": 0.022,
+      "step": 484
+    },
+    {
+      "epoch": 1.658828559213339,
+      "grad_norm": 0.2403880470710616,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0184,
+      "step": 485
+    },
+    {
+      "epoch": 1.662248824283882,
+      "grad_norm": 0.21080166415704205,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0194,
+      "step": 486
+    },
+    {
+      "epoch": 1.665669089354425,
+      "grad_norm": 0.24366180394977457,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0212,
+      "step": 487
+    },
+    {
+      "epoch": 1.669089354424968,
+      "grad_norm": 0.2445725370169137,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0201,
+      "step": 488
+    },
+    {
+      "epoch": 1.672509619495511,
+      "grad_norm": 0.2400419596996237,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0217,
+      "step": 489
+    },
+    {
+      "epoch": 1.6759298845660537,
+      "grad_norm": 0.22845732992118456,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0193,
+      "step": 490
+    },
+    {
+      "epoch": 1.6793501496365968,
+      "grad_norm": 0.21963101467880122,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0205,
+      "step": 491
+    },
+    {
+      "epoch": 1.6827704147071398,
+      "grad_norm": 0.23832272298065416,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0208,
+      "step": 492
+    },
+    {
+      "epoch": 1.6861906797776829,
+      "grad_norm": 0.2701770424720032,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0245,
+      "step": 493
+    },
+    {
+      "epoch": 1.6896109448482257,
+      "grad_norm": 0.21999352850800136,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0191,
+      "step": 494
+    },
+    {
+      "epoch": 1.6930312099187685,
+      "grad_norm": 0.2303202011396172,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0206,
+      "step": 495
+    },
+    {
+      "epoch": 1.6964514749893116,
+      "grad_norm": 0.23207638496444852,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0196,
+      "step": 496
+    },
+    {
+      "epoch": 1.6998717400598546,
+      "grad_norm": 0.27476502715987833,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0243,
+      "step": 497
+    },
+    {
+      "epoch": 1.7032920051303977,
+      "grad_norm": 0.25943127784425596,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0209,
+      "step": 498
+    },
+    {
+      "epoch": 1.7067122702009405,
+      "grad_norm": 0.2582673258172837,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0213,
+      "step": 499
+    },
+    {
+      "epoch": 1.7101325352714836,
+      "grad_norm": 0.2270336558731971,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0207,
+      "step": 500
+    },
+    {
+      "epoch": 1.7135528003420264,
+      "grad_norm": 0.21716332073489333,
+      "learning_rate": 1.41e-05,
+      "loss": 0.017,
+      "step": 501
+    },
+    {
+      "epoch": 1.7169730654125694,
+      "grad_norm": 0.22361076006822656,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0198,
+      "step": 502
+    },
+    {
+      "epoch": 1.7203933304831125,
+      "grad_norm": 0.25453906115757163,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0215,
+      "step": 503
+    },
+    {
+      "epoch": 1.7238135955536555,
+      "grad_norm": 0.2639274732447762,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0211,
+      "step": 504
+    },
+    {
+      "epoch": 1.7272338606241984,
+      "grad_norm": 0.25322752853555974,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0228,
+      "step": 505
+    },
+    {
+      "epoch": 1.7306541256947412,
+      "grad_norm": 0.23285923741403275,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0205,
+      "step": 506
+    },
+    {
+      "epoch": 1.7340743907652842,
+      "grad_norm": 0.2323064412599781,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0195,
+      "step": 507
+    },
+    {
+      "epoch": 1.7374946558358273,
+      "grad_norm": 0.23653121619208753,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0211,
+      "step": 508
+    },
+    {
+      "epoch": 1.7409149209063703,
+      "grad_norm": 0.2576532470063006,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0216,
+      "step": 509
+    },
+    {
+      "epoch": 1.7443351859769132,
+      "grad_norm": 0.3031982152602418,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0212,
+      "step": 510
+    },
+    {
+      "epoch": 1.7477554510474562,
+      "grad_norm": 0.21353603367638405,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0187,
+      "step": 511
+    },
+    {
+      "epoch": 1.751175716117999,
+      "grad_norm": 0.23897983103928835,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0211,
+      "step": 512
+    },
+    {
+      "epoch": 1.754595981188542,
+      "grad_norm": 0.26759279857241974,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0225,
+      "step": 513
+    },
+    {
+      "epoch": 1.7580162462590851,
+      "grad_norm": 0.21204297476027956,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0182,
+      "step": 514
+    },
+    {
+      "epoch": 1.7614365113296282,
+      "grad_norm": 0.2522156014013123,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0224,
+      "step": 515
+    },
+    {
+      "epoch": 1.764856776400171,
+      "grad_norm": 0.2483681502861828,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0225,
+      "step": 516
+    },
+    {
+      "epoch": 1.7682770414707139,
+      "grad_norm": 0.2757743080951287,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0252,
+      "step": 517
+    },
+    {
+      "epoch": 1.771697306541257,
+      "grad_norm": 0.20783407363518988,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0189,
+      "step": 518
+    },
+    {
+      "epoch": 1.7751175716118,
+      "grad_norm": 0.25743244499425244,
+      "learning_rate": 1.41e-05,
+      "loss": 0.022,
+      "step": 519
+    },
+    {
+      "epoch": 1.778537836682343,
+      "grad_norm": 0.23017428673046214,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0195,
+      "step": 520
+    },
+    {
+      "epoch": 1.7819581017528858,
+      "grad_norm": 0.2315895066049246,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0178,
+      "step": 521
+    },
+    {
+      "epoch": 1.7853783668234287,
+      "grad_norm": 0.22782043412403868,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0215,
+      "step": 522
+    },
+    {
+      "epoch": 1.7887986318939717,
+      "grad_norm": 0.2724892881429743,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0224,
+      "step": 523
+    },
+    {
+      "epoch": 1.7922188969645148,
+      "grad_norm": 0.28330092514107247,
+      "learning_rate": 1.41e-05,
+      "loss": 0.02,
+      "step": 524
+    },
+    {
+      "epoch": 1.7956391620350578,
+      "grad_norm": 0.23561517254595046,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0199,
+      "step": 525
+    },
+    {
+      "epoch": 1.7990594271056009,
+      "grad_norm": 0.27080028676550905,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0213,
+      "step": 526
+    },
+    {
+      "epoch": 1.8024796921761437,
+      "grad_norm": 0.20908394348220563,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0181,
+      "step": 527
+    },
+    {
+      "epoch": 1.8058999572466865,
+      "grad_norm": 0.24764828335378633,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0208,
+      "step": 528
+    },
+    {
+      "epoch": 1.8093202223172296,
+      "grad_norm": 0.29733366903532993,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0218,
+      "step": 529
+    },
+    {
+      "epoch": 1.8127404873877726,
+      "grad_norm": 0.2708827452564214,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0226,
+      "step": 530
+    },
+    {
+      "epoch": 1.8161607524583157,
+      "grad_norm": 0.2887461777468973,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0226,
+      "step": 531
+    },
+    {
+      "epoch": 1.8195810175288585,
+      "grad_norm": 0.2774042332926017,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0246,
+      "step": 532
+    },
+    {
+      "epoch": 1.8230012825994013,
+      "grad_norm": 0.23877829686783267,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0202,
+      "step": 533
+    },
+    {
+      "epoch": 1.8264215476699444,
+      "grad_norm": 0.2544233995482896,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0204,
+      "step": 534
+    },
+    {
+      "epoch": 1.8298418127404874,
+      "grad_norm": 0.26365092898440745,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0223,
+      "step": 535
+    },
+    {
+      "epoch": 1.8332620778110305,
+      "grad_norm": 0.21178009134118125,
+      "learning_rate": 1.41e-05,
+      "loss": 0.018,
+      "step": 536
+    },
+    {
+      "epoch": 1.8366823428815733,
+      "grad_norm": 0.27115664081831004,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0211,
+      "step": 537
+    },
+    {
+      "epoch": 1.8401026079521163,
+      "grad_norm": 0.2653048117342068,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0212,
+      "step": 538
+    },
+    {
+      "epoch": 1.8435228730226592,
+      "grad_norm": 0.2757728122775812,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0204,
+      "step": 539
+    },
+    {
+      "epoch": 1.8469431380932022,
+      "grad_norm": 0.2664986190346045,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0207,
+      "step": 540
+    },
+    {
+      "epoch": 1.8503634031637453,
+      "grad_norm": 0.265518302380727,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0179,
+      "step": 541
+    },
+    {
+      "epoch": 1.8537836682342883,
+      "grad_norm": 0.2895623799588168,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0228,
+      "step": 542
+    },
+    {
+      "epoch": 1.8572039333048311,
+      "grad_norm": 0.23284728474850006,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0192,
+      "step": 543
+    },
+    {
+      "epoch": 1.860624198375374,
+      "grad_norm": 0.21864913411739378,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0178,
+      "step": 544
+    },
+    {
+      "epoch": 1.864044463445917,
+      "grad_norm": 0.25369925309969443,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0204,
+      "step": 545
+    },
+    {
+      "epoch": 1.86746472851646,
+      "grad_norm": 0.22162789215208448,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0214,
+      "step": 546
+    },
+    {
+      "epoch": 1.870884993587003,
+      "grad_norm": 0.228575157771765,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0215,
+      "step": 547
+    },
+    {
+      "epoch": 1.874305258657546,
+      "grad_norm": 0.29177202889082,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0231,
+      "step": 548
+    },
+    {
+      "epoch": 1.8777255237280888,
+      "grad_norm": 0.2526426700464145,
+      "learning_rate": 1.41e-05,
+      "loss": 0.022,
+      "step": 549
+    },
+    {
+      "epoch": 1.8811457887986318,
+      "grad_norm": 0.25824463173696005,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0238,
+      "step": 550
+    },
+    {
+      "epoch": 1.8845660538691749,
+      "grad_norm": 0.22931141277969364,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0195,
+      "step": 551
+    },
+    {
+      "epoch": 1.887986318939718,
+      "grad_norm": 0.20128540781283585,
+      "learning_rate": 1.41e-05,
+      "loss": 0.018,
+      "step": 552
+    },
+    {
+      "epoch": 1.891406584010261,
+      "grad_norm": 0.2726453682164885,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0215,
+      "step": 553
+    },
+    {
+      "epoch": 1.8948268490808038,
+      "grad_norm": 0.2379175413754306,
+      "learning_rate": 1.41e-05,
+      "loss": 0.02,
+      "step": 554
+    },
+    {
+      "epoch": 1.8982471141513466,
+      "grad_norm": 0.2657340317097815,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0215,
+      "step": 555
+    },
+    {
+      "epoch": 1.9016673792218897,
+      "grad_norm": 0.2469655254797632,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0218,
+      "step": 556
+    },
+    {
+      "epoch": 1.9050876442924327,
+      "grad_norm": 0.2306898046638619,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0206,
+      "step": 557
+    },
+    {
+      "epoch": 1.9085079093629758,
+      "grad_norm": 0.2501966986368517,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0208,
+      "step": 558
+    },
+    {
+      "epoch": 1.9119281744335186,
+      "grad_norm": 0.26337448659750273,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0224,
+      "step": 559
+    },
+    {
+      "epoch": 1.9153484395040614,
+      "grad_norm": 0.23788126235745702,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0207,
+      "step": 560
+    },
+    {
+      "epoch": 1.9187687045746045,
+      "grad_norm": 0.24400460810027635,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0199,
+      "step": 561
+    },
+    {
+      "epoch": 1.9221889696451475,
+      "grad_norm": 0.2524032823359455,
+      "learning_rate": 1.41e-05,
+      "loss": 0.02,
+      "step": 562
+    },
+    {
+      "epoch": 1.9256092347156906,
+      "grad_norm": 0.2288025108079512,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0201,
+      "step": 563
+    },
+    {
+      "epoch": 1.9290294997862334,
+      "grad_norm": 0.269452001004884,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0214,
+      "step": 564
+    },
+    {
+      "epoch": 1.9324497648567764,
+      "grad_norm": 0.2316242696085615,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0196,
+      "step": 565
+    },
+    {
+      "epoch": 1.9358700299273193,
+      "grad_norm": 0.221180291707861,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0177,
+      "step": 566
+    },
+    {
+      "epoch": 1.9392902949978623,
+      "grad_norm": 0.2607363855888593,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0192,
+      "step": 567
+    },
+    {
+      "epoch": 1.9427105600684054,
+      "grad_norm": 0.24514765767262625,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0182,
+      "step": 568
+    },
+    {
+      "epoch": 1.9461308251389484,
+      "grad_norm": 0.28916534222959817,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0207,
+      "step": 569
+    },
+    {
+      "epoch": 1.9495510902094912,
+      "grad_norm": 0.26414654167917073,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0204,
+      "step": 570
+    },
+    {
+      "epoch": 1.952971355280034,
+      "grad_norm": 0.2482098982197082,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0206,
+      "step": 571
+    },
+    {
+      "epoch": 1.9563916203505771,
+      "grad_norm": 0.2608562249261069,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0208,
+      "step": 572
+    },
+    {
+      "epoch": 1.9598118854211202,
+      "grad_norm": 0.25525358687825855,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0213,
+      "step": 573
+    },
+    {
+      "epoch": 1.9632321504916632,
+      "grad_norm": 0.23125518036524792,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0212,
+      "step": 574
+    },
+    {
+      "epoch": 1.966652415562206,
+      "grad_norm": 0.2432676298521834,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0187,
+      "step": 575
+    },
+    {
+      "epoch": 1.9700726806327489,
+      "grad_norm": 0.24502134522738972,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0194,
+      "step": 576
+    },
+    {
+      "epoch": 1.973492945703292,
+      "grad_norm": 0.26553777081307495,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0212,
+      "step": 577
+    },
+    {
+      "epoch": 1.976913210773835,
+      "grad_norm": 0.24919535132544937,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0208,
+      "step": 578
+    },
+    {
+      "epoch": 1.980333475844378,
+      "grad_norm": 0.2828767110434045,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0238,
+      "step": 579
+    },
+    {
+      "epoch": 1.983753740914921,
+      "grad_norm": 0.26780635861278174,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0225,
+      "step": 580
+    },
+    {
+      "epoch": 1.987174005985464,
+      "grad_norm": 0.26755832407143076,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0197,
+      "step": 581
+    },
+    {
+      "epoch": 1.9905942710560067,
+      "grad_norm": 0.23150201894075045,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0225,
+      "step": 582
+    },
+    {
+      "epoch": 1.9940145361265498,
+      "grad_norm": 0.2556938828383787,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0217,
+      "step": 583
+    },
+    {
+      "epoch": 1.9974348011970928,
+      "grad_norm": 0.24779362245147543,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0209,
+      "step": 584
+    },
+    {
+      "epoch": 2.000855066267636,
+      "grad_norm": 0.23220294742842815,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0205,
+      "step": 585
+    },
+    {
+      "epoch": 2.004275331338179,
+      "grad_norm": 0.19611669457431613,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0136,
+      "step": 586
+    },
+    {
+      "epoch": 2.0076955964087215,
+      "grad_norm": 0.18201675420540842,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 587
+    },
+    {
+      "epoch": 2.0111158614792646,
+      "grad_norm": 0.2111608412027396,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0165,
+      "step": 588
+    },
+    {
+      "epoch": 2.0145361265498076,
+      "grad_norm": 0.19725473414903413,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 589
+    },
+    {
+      "epoch": 2.0179563916203507,
+      "grad_norm": 0.21354984112177863,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0161,
+      "step": 590
+    },
+    {
+      "epoch": 2.0213766566908937,
+      "grad_norm": 0.22940862450456587,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0157,
+      "step": 591
+    },
+    {
+      "epoch": 2.0247969217614363,
+      "grad_norm": 0.18767782352642373,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0152,
+      "step": 592
+    },
+    {
+      "epoch": 2.0282171868319794,
+      "grad_norm": 0.20643159409018694,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 593
+    },
+    {
+      "epoch": 2.0316374519025224,
+      "grad_norm": 0.2037046870595368,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0138,
+      "step": 594
+    },
+    {
+      "epoch": 2.0350577169730655,
+      "grad_norm": 0.19705696567608205,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0162,
+      "step": 595
+    },
+    {
+      "epoch": 2.0384779820436085,
+      "grad_norm": 0.21000255959562664,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 596
+    },
+    {
+      "epoch": 2.041898247114151,
+      "grad_norm": 0.1909421240987611,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0136,
+      "step": 597
+    },
+    {
+      "epoch": 2.045318512184694,
+      "grad_norm": 0.22254593762119865,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 598
+    },
+    {
+      "epoch": 2.0487387772552372,
+      "grad_norm": 0.18039596470753322,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0147,
+      "step": 599
+    },
+    {
+      "epoch": 2.0521590423257803,
+      "grad_norm": 0.1836910759989102,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 600
+    },
+    {
+      "epoch": 2.0555793073963233,
+      "grad_norm": 0.2007119695404956,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 601
+    },
+    {
+      "epoch": 2.0589995724668664,
+      "grad_norm": 0.17881220811912174,
+      "learning_rate": 1.41e-05,
+      "loss": 0.016,
+      "step": 602
+    },
+    {
+      "epoch": 2.062419837537409,
+      "grad_norm": 0.18131752461639056,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0152,
+      "step": 603
+    },
+    {
+      "epoch": 2.065840102607952,
+      "grad_norm": 0.2091101968035948,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0147,
+      "step": 604
+    },
+    {
+      "epoch": 2.069260367678495,
+      "grad_norm": 0.1785580403127973,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0142,
+      "step": 605
+    },
+    {
+      "epoch": 2.072680632749038,
+      "grad_norm": 0.1952402053161801,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 606
+    },
+    {
+      "epoch": 2.076100897819581,
+      "grad_norm": 0.16849209697986273,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0135,
+      "step": 607
+    },
+    {
+      "epoch": 2.079521162890124,
+      "grad_norm": 0.18960749528358953,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0158,
+      "step": 608
+    },
+    {
+      "epoch": 2.082941427960667,
+      "grad_norm": 0.1822523628091137,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0157,
+      "step": 609
+    },
+    {
+      "epoch": 2.08636169303121,
+      "grad_norm": 0.22244344533378702,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0145,
+      "step": 610
+    },
+    {
+      "epoch": 2.089781958101753,
+      "grad_norm": 0.16677137506686548,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 611
+    },
+    {
+      "epoch": 2.093202223172296,
+      "grad_norm": 0.19008011213315038,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0149,
+      "step": 612
+    },
+    {
+      "epoch": 2.096622488242839,
+      "grad_norm": 0.1965146825157404,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0158,
+      "step": 613
+    },
+    {
+      "epoch": 2.1000427533133816,
+      "grad_norm": 0.17095730271612053,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 614
+    },
+    {
+      "epoch": 2.1034630183839247,
+      "grad_norm": 0.18376098918489842,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0133,
+      "step": 615
+    },
+    {
+      "epoch": 2.1068832834544677,
+      "grad_norm": 0.18859892344363186,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 616
+    },
+    {
+      "epoch": 2.110303548525011,
+      "grad_norm": 0.17678403587042746,
+      "learning_rate": 1.41e-05,
+      "loss": 0.013,
+      "step": 617
+    },
+    {
+      "epoch": 2.113723813595554,
+      "grad_norm": 0.18933984169673357,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0152,
+      "step": 618
+    },
+    {
+      "epoch": 2.1171440786660964,
+      "grad_norm": 0.20060664740874196,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0157,
+      "step": 619
+    },
+    {
+      "epoch": 2.1205643437366395,
+      "grad_norm": 0.1803038264349401,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 620
+    },
+    {
+      "epoch": 2.1239846088071825,
+      "grad_norm": 0.1910210164605926,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 621
+    },
+    {
+      "epoch": 2.1274048738777256,
+      "grad_norm": 0.1794365441256405,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0137,
+      "step": 622
+    },
+    {
+      "epoch": 2.1308251389482686,
+      "grad_norm": 0.15893205685478506,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0136,
+      "step": 623
+    },
+    {
+      "epoch": 2.1342454040188112,
+      "grad_norm": 0.19724991519919266,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 624
+    },
+    {
+      "epoch": 2.1376656690893543,
+      "grad_norm": 0.17458605756275755,
+      "learning_rate": 1.41e-05,
+      "loss": 0.016,
+      "step": 625
+    },
+    {
+      "epoch": 2.1410859341598973,
+      "grad_norm": 0.172521593573168,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 626
+    },
+    {
+      "epoch": 2.1445061992304404,
+      "grad_norm": 0.14557636073307398,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0135,
+      "step": 627
+    },
+    {
+      "epoch": 2.1479264643009834,
+      "grad_norm": 0.22843890765431576,
+      "learning_rate": 1.41e-05,
+      "loss": 0.016,
+      "step": 628
+    },
+    {
+      "epoch": 2.1513467293715265,
+      "grad_norm": 0.18196169286713992,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 629
+    },
+    {
+      "epoch": 2.154766994442069,
+      "grad_norm": 0.17272140560874683,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0135,
+      "step": 630
+    },
+    {
+      "epoch": 2.158187259512612,
+      "grad_norm": 0.15595539359263064,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 631
+    },
+    {
+      "epoch": 2.161607524583155,
+      "grad_norm": 0.16465865001486402,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0136,
+      "step": 632
+    },
+    {
+      "epoch": 2.1650277896536982,
+      "grad_norm": 0.17986632980817158,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 633
+    },
+    {
+      "epoch": 2.1684480547242413,
+      "grad_norm": 0.18929877257592365,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0163,
+      "step": 634
+    },
+    {
+      "epoch": 2.171868319794784,
+      "grad_norm": 0.19542505698465773,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0159,
+      "step": 635
+    },
+    {
+      "epoch": 2.175288584865327,
+      "grad_norm": 0.17114565781513597,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0145,
+      "step": 636
+    },
+    {
+      "epoch": 2.17870884993587,
+      "grad_norm": 0.18331913158118285,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 637
+    },
+    {
+      "epoch": 2.182129115006413,
+      "grad_norm": 0.16055196305467795,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 638
+    },
+    {
+      "epoch": 2.185549380076956,
+      "grad_norm": 0.15985440322290387,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 639
+    },
+    {
+      "epoch": 2.188969645147499,
+      "grad_norm": 0.16417247457666104,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0145,
+      "step": 640
+    },
+    {
+      "epoch": 2.1923899102180417,
+      "grad_norm": 0.16210688538374043,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 641
+    },
+    {
+      "epoch": 2.195810175288585,
+      "grad_norm": 0.16634144533077172,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 642
+    },
+    {
+      "epoch": 2.199230440359128,
+      "grad_norm": 0.18582606043218244,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 643
+    },
+    {
+      "epoch": 2.202650705429671,
+      "grad_norm": 0.1747184961523454,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0164,
+      "step": 644
+    },
+    {
+      "epoch": 2.206070970500214,
+      "grad_norm": 0.15610336933105856,
+      "learning_rate": 1.41e-05,
+      "loss": 0.013,
+      "step": 645
+    },
+    {
+      "epoch": 2.2094912355707566,
+      "grad_norm": 0.16920938302300378,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 646
+    },
+    {
+      "epoch": 2.2129115006412996,
+      "grad_norm": 0.17648344233280136,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0147,
+      "step": 647
+    },
+    {
+      "epoch": 2.2163317657118426,
+      "grad_norm": 0.17020885614859071,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 648
+    },
+    {
+      "epoch": 2.2197520307823857,
+      "grad_norm": 0.17716196526554184,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0142,
+      "step": 649
+    },
+    {
+      "epoch": 2.2231722958529287,
+      "grad_norm": 0.17375550168851311,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 650
+    },
+    {
+      "epoch": 2.2265925609234714,
+      "grad_norm": 0.15660012931864775,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0138,
+      "step": 651
+    },
+    {
+      "epoch": 2.2300128259940144,
+      "grad_norm": 0.192388171113886,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0151,
+      "step": 652
+    },
+    {
+      "epoch": 2.2334330910645575,
+      "grad_norm": 0.17112223174304642,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0137,
+      "step": 653
+    },
+    {
+      "epoch": 2.2368533561351005,
+      "grad_norm": 0.17046414663473125,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0163,
+      "step": 654
+    },
+    {
+      "epoch": 2.2402736212056436,
+      "grad_norm": 0.14723103129157328,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 655
+    },
+    {
+      "epoch": 2.2436938862761866,
+      "grad_norm": 0.16260020026720062,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0134,
+      "step": 656
+    },
+    {
+      "epoch": 2.247114151346729,
+      "grad_norm": 0.1605529186115587,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 657
+    },
+    {
+      "epoch": 2.2505344164172723,
+      "grad_norm": 0.1649326783596916,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 658
+    },
+    {
+      "epoch": 2.2539546814878153,
+      "grad_norm": 0.1664480794095166,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 659
+    },
+    {
+      "epoch": 2.2573749465583584,
+      "grad_norm": 0.16946179512795181,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 660
+    },
+    {
+      "epoch": 2.2607952116289014,
+      "grad_norm": 0.166434202624617,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0135,
+      "step": 661
+    },
+    {
+      "epoch": 2.2642154766994445,
+      "grad_norm": 0.1739998754952991,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0142,
+      "step": 662
+    },
+    {
+      "epoch": 2.267635741769987,
+      "grad_norm": 0.1809093145862042,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 663
+    },
+    {
+      "epoch": 2.27105600684053,
+      "grad_norm": 0.15727492334875873,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0137,
+      "step": 664
+    },
+    {
+      "epoch": 2.274476271911073,
+      "grad_norm": 0.18448939109027693,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0157,
+      "step": 665
+    },
+    {
+      "epoch": 2.277896536981616,
+      "grad_norm": 0.18236721526184443,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0166,
+      "step": 666
+    },
+    {
+      "epoch": 2.2813168020521593,
+      "grad_norm": 0.1695210468590141,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 667
+    },
+    {
+      "epoch": 2.284737067122702,
+      "grad_norm": 0.17157602170689792,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0153,
+      "step": 668
+    },
+    {
+      "epoch": 2.288157332193245,
+      "grad_norm": 0.16583571772207567,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0133,
+      "step": 669
+    },
+    {
+      "epoch": 2.291577597263788,
+      "grad_norm": 0.15309231682278962,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0134,
+      "step": 670
+    },
+    {
+      "epoch": 2.294997862334331,
+      "grad_norm": 0.18010491103515566,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 671
+    },
+    {
+      "epoch": 2.298418127404874,
+      "grad_norm": 0.165226793210629,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0134,
+      "step": 672
+    },
+    {
+      "epoch": 2.3018383924754167,
+      "grad_norm": 0.19017257841407276,
+      "learning_rate": 1.41e-05,
+      "loss": 0.016,
+      "step": 673
+    },
+    {
+      "epoch": 2.3052586575459597,
+      "grad_norm": 0.16105989698952458,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 674
+    },
+    {
+      "epoch": 2.3086789226165028,
+      "grad_norm": 0.17435397438208278,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0136,
+      "step": 675
+    },
+    {
+      "epoch": 2.312099187687046,
+      "grad_norm": 0.15489352805926157,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0149,
+      "step": 676
+    },
+    {
+      "epoch": 2.315519452757589,
+      "grad_norm": 0.15749574161575872,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0136,
+      "step": 677
+    },
+    {
+      "epoch": 2.3189397178281315,
+      "grad_norm": 0.17861788971942064,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 678
+    },
+    {
+      "epoch": 2.3223599828986745,
+      "grad_norm": 0.16852344878390402,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0152,
+      "step": 679
+    },
+    {
+      "epoch": 2.3257802479692176,
+      "grad_norm": 0.1903544466095535,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 680
+    },
+    {
+      "epoch": 2.3292005130397606,
+      "grad_norm": 0.1938655036641079,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0158,
+      "step": 681
+    },
+    {
+      "epoch": 2.3326207781103037,
+      "grad_norm": 0.20203642790445486,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 682
+    },
+    {
+      "epoch": 2.3360410431808463,
+      "grad_norm": 0.16328748761388734,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 683
+    },
+    {
+      "epoch": 2.3394613082513893,
+      "grad_norm": 0.17437303421082254,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0149,
+      "step": 684
+    },
+    {
+      "epoch": 2.3428815733219324,
+      "grad_norm": 0.17762818156916235,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0149,
+      "step": 685
+    },
+    {
+      "epoch": 2.3463018383924754,
+      "grad_norm": 0.18594818733099802,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0161,
+      "step": 686
+    },
+    {
+      "epoch": 2.3497221034630185,
+      "grad_norm": 0.16139308140544603,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 687
+    },
+    {
+      "epoch": 2.3531423685335615,
+      "grad_norm": 0.16108681383332696,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 688
+    },
+    {
+      "epoch": 2.3565626336041046,
+      "grad_norm": 0.15992689822195585,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 689
+    },
+    {
+      "epoch": 2.359982898674647,
+      "grad_norm": 0.19759610045563245,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0158,
+      "step": 690
+    },
+    {
+      "epoch": 2.36340316374519,
+      "grad_norm": 0.16149200857684542,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 691
+    },
+    {
+      "epoch": 2.3668234288157333,
+      "grad_norm": 0.15072989325955685,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 692
+    },
+    {
+      "epoch": 2.3702436938862763,
+      "grad_norm": 0.14913939759012662,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 693
+    },
+    {
+      "epoch": 2.3736639589568194,
+      "grad_norm": 0.15636976839856181,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0153,
+      "step": 694
+    },
+    {
+      "epoch": 2.377084224027362,
+      "grad_norm": 0.18599274943363073,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 695
+    },
+    {
+      "epoch": 2.380504489097905,
+      "grad_norm": 0.16568136382204926,
+      "learning_rate": 1.41e-05,
+      "loss": 0.013,
+      "step": 696
+    },
+    {
+      "epoch": 2.383924754168448,
+      "grad_norm": 0.17332302727716312,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0128,
+      "step": 697
+    },
+    {
+      "epoch": 2.387345019238991,
+      "grad_norm": 0.1879178833471297,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 698
+    },
+    {
+      "epoch": 2.390765284309534,
+      "grad_norm": 0.1726315547096704,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0136,
+      "step": 699
+    },
+    {
+      "epoch": 2.3941855493800768,
+      "grad_norm": 0.1720038600446686,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0158,
+      "step": 700
+    },
+    {
+      "epoch": 2.39760581445062,
+      "grad_norm": 0.17262771034044544,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 701
+    },
+    {
+      "epoch": 2.401026079521163,
+      "grad_norm": 0.18381965008540402,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0151,
+      "step": 702
+    },
+    {
+      "epoch": 2.404446344591706,
+      "grad_norm": 0.1943786246553059,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 703
+    },
+    {
+      "epoch": 2.407866609662249,
+      "grad_norm": 0.15285642974808178,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0133,
+      "step": 704
+    },
+    {
+      "epoch": 2.4112868747327916,
+      "grad_norm": 0.19330131469876405,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 705
+    },
+    {
+      "epoch": 2.4147071398033346,
+      "grad_norm": 0.16935279234922626,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 706
+    },
+    {
+      "epoch": 2.4181274048738777,
+      "grad_norm": 0.17130297774748693,
+      "learning_rate": 1.41e-05,
+      "loss": 0.016,
+      "step": 707
+    },
+    {
+      "epoch": 2.4215476699444207,
+      "grad_norm": 0.16649757139294383,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0135,
+      "step": 708
+    },
+    {
+      "epoch": 2.4249679350149638,
+      "grad_norm": 0.1509762981692361,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0131,
+      "step": 709
+    },
+    {
+      "epoch": 2.428388200085507,
+      "grad_norm": 0.16903612165580836,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 710
+    },
+    {
+      "epoch": 2.4318084651560494,
+      "grad_norm": 0.16965158831335697,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 711
+    },
+    {
+      "epoch": 2.4352287302265925,
+      "grad_norm": 0.17489335625033847,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0127,
+      "step": 712
+    },
+    {
+      "epoch": 2.4386489952971355,
+      "grad_norm": 0.17619733289193135,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 713
+    },
+    {
+      "epoch": 2.4420692603676786,
+      "grad_norm": 0.17617881507074173,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0152,
+      "step": 714
+    },
+    {
+      "epoch": 2.4454895254382216,
+      "grad_norm": 0.18048518098716906,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 715
+    },
+    {
+      "epoch": 2.4489097905087647,
+      "grad_norm": 0.2051952297924243,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 716
+    },
+    {
+      "epoch": 2.4523300555793073,
+      "grad_norm": 0.15664970563386238,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 717
+    },
+    {
+      "epoch": 2.4557503206498503,
+      "grad_norm": 0.16743723595706328,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0152,
+      "step": 718
+    },
+    {
+      "epoch": 2.4591705857203934,
+      "grad_norm": 0.17613442783841668,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0162,
+      "step": 719
+    },
+    {
+      "epoch": 2.4625908507909364,
+      "grad_norm": 0.1818706957568455,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0135,
+      "step": 720
+    },
+    {
+      "epoch": 2.4660111158614795,
+      "grad_norm": 0.17047836187664528,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 721
+    },
+    {
+      "epoch": 2.469431380932022,
+      "grad_norm": 0.17741467330052185,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0158,
+      "step": 722
+    },
+    {
+      "epoch": 2.472851646002565,
+      "grad_norm": 0.1800167905144302,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 723
+    },
+    {
+      "epoch": 2.476271911073108,
+      "grad_norm": 0.1520569062216661,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 724
+    },
+    {
+      "epoch": 2.4796921761436512,
+      "grad_norm": 0.19504143457216305,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 725
+    },
+    {
+      "epoch": 2.4831124412141943,
+      "grad_norm": 0.16406995475864966,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0152,
+      "step": 726
+    },
+    {
+      "epoch": 2.486532706284737,
+      "grad_norm": 0.1640747075139151,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0138,
+      "step": 727
+    },
+    {
+      "epoch": 2.48995297135528,
+      "grad_norm": 0.17515606314180554,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 728
+    },
+    {
+      "epoch": 2.493373236425823,
+      "grad_norm": 0.17908072157399882,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 729
+    },
+    {
+      "epoch": 2.496793501496366,
+      "grad_norm": 0.15664812551985538,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0147,
+      "step": 730
+    },
+    {
+      "epoch": 2.500213766566909,
+      "grad_norm": 0.165379248202876,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0151,
+      "step": 731
+    },
+    {
+      "epoch": 2.5036340316374517,
+      "grad_norm": 0.15423115964706519,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 732
+    },
+    {
+      "epoch": 2.5070542967079947,
+      "grad_norm": 0.16485219288254147,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0149,
+      "step": 733
+    },
+    {
+      "epoch": 2.510474561778538,
+      "grad_norm": 0.176637422447442,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 734
+    },
+    {
+      "epoch": 2.513894826849081,
+      "grad_norm": 0.17814754105281233,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 735
+    },
+    {
+      "epoch": 2.517315091919624,
+      "grad_norm": 0.17118656720070607,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 736
+    },
+    {
+      "epoch": 2.5207353569901665,
+      "grad_norm": 0.18260703508475823,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 737
+    },
+    {
+      "epoch": 2.52415562206071,
+      "grad_norm": 0.17080402514684412,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0136,
+      "step": 738
+    },
+    {
+      "epoch": 2.5275758871312526,
+      "grad_norm": 0.15709135265175292,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0134,
+      "step": 739
+    },
+    {
+      "epoch": 2.5309961522017956,
+      "grad_norm": 0.18246981305252055,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0142,
+      "step": 740
+    },
+    {
+      "epoch": 2.5344164172723387,
+      "grad_norm": 0.1808585737198185,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 741
+    },
+    {
+      "epoch": 2.5378366823428817,
+      "grad_norm": 0.16454795740570335,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0153,
+      "step": 742
+    },
+    {
+      "epoch": 2.541256947413425,
+      "grad_norm": 0.21032421935647805,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0165,
+      "step": 743
+    },
+    {
+      "epoch": 2.5446772124839674,
+      "grad_norm": 0.15249763976570654,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 744
+    },
+    {
+      "epoch": 2.5480974775545104,
+      "grad_norm": 0.1613635941460105,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0131,
+      "step": 745
+    },
+    {
+      "epoch": 2.5515177426250535,
+      "grad_norm": 0.1700060737892137,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0145,
+      "step": 746
+    },
+    {
+      "epoch": 2.5549380076955965,
+      "grad_norm": 0.20547449901345968,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0168,
+      "step": 747
+    },
+    {
+      "epoch": 2.5583582727661396,
+      "grad_norm": 0.16484966537277382,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0142,
+      "step": 748
+    },
+    {
+      "epoch": 2.561778537836682,
+      "grad_norm": 0.17889219393433903,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 749
+    },
+    {
+      "epoch": 2.5651988029072252,
+      "grad_norm": 0.17679721987917787,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0138,
+      "step": 750
+    },
+    {
+      "epoch": 2.5686190679777683,
+      "grad_norm": 0.16210612559541202,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0145,
+      "step": 751
+    },
+    {
+      "epoch": 2.5720393330483113,
+      "grad_norm": 0.17192503019428163,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 752
+    },
+    {
+      "epoch": 2.5754595981188544,
+      "grad_norm": 0.16114230533462182,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0151,
+      "step": 753
+    },
+    {
+      "epoch": 2.578879863189397,
+      "grad_norm": 0.16683326324078276,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 754
+    },
+    {
+      "epoch": 2.58230012825994,
+      "grad_norm": 0.15929030309974343,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0135,
+      "step": 755
+    },
+    {
+      "epoch": 2.585720393330483,
+      "grad_norm": 0.17728404681160545,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0134,
+      "step": 756
+    },
+    {
+      "epoch": 2.589140658401026,
+      "grad_norm": 0.16727441441267749,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 757
+    },
+    {
+      "epoch": 2.592560923471569,
+      "grad_norm": 0.21841845784314023,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0149,
+      "step": 758
+    },
+    {
+      "epoch": 2.595981188542112,
+      "grad_norm": 0.1770465088709539,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 759
+    },
+    {
+      "epoch": 2.599401453612655,
+      "grad_norm": 0.1737371304814951,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 760
+    },
+    {
+      "epoch": 2.602821718683198,
+      "grad_norm": 0.17069864594223258,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0134,
+      "step": 761
+    },
+    {
+      "epoch": 2.606241983753741,
+      "grad_norm": 0.16609359338004503,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0158,
+      "step": 762
+    },
+    {
+      "epoch": 2.609662248824284,
+      "grad_norm": 0.17378348121220705,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 763
+    },
+    {
+      "epoch": 2.6130825138948266,
+      "grad_norm": 0.17355912661147865,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0142,
+      "step": 764
+    },
+    {
+      "epoch": 2.61650277896537,
+      "grad_norm": 0.20077889987954006,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 765
+    },
+    {
+      "epoch": 2.6199230440359127,
+      "grad_norm": 0.16419074893195182,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 766
+    },
+    {
+      "epoch": 2.6233433091064557,
+      "grad_norm": 0.19476203022751112,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 767
+    },
+    {
+      "epoch": 2.626763574176999,
+      "grad_norm": 0.16568838103054007,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 768
+    },
+    {
+      "epoch": 2.630183839247542,
+      "grad_norm": 0.21099152906068663,
+      "learning_rate": 1.41e-05,
+      "loss": 0.017,
+      "step": 769
+    },
+    {
+      "epoch": 2.633604104318085,
+      "grad_norm": 0.17959731227982292,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 770
+    },
+    {
+      "epoch": 2.6370243693886275,
+      "grad_norm": 0.16972780919186436,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 771
+    },
+    {
+      "epoch": 2.6404446344591705,
+      "grad_norm": 0.1842651161264449,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0157,
+      "step": 772
+    },
+    {
+      "epoch": 2.6438648995297136,
+      "grad_norm": 0.17834439152890905,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0153,
+      "step": 773
+    },
+    {
+      "epoch": 2.6472851646002566,
+      "grad_norm": 0.18047152749420897,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 774
+    },
+    {
+      "epoch": 2.6507054296707997,
+      "grad_norm": 0.17154124492966955,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0147,
+      "step": 775
+    },
+    {
+      "epoch": 2.6541256947413423,
+      "grad_norm": 0.17549885521671557,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0161,
+      "step": 776
+    },
+    {
+      "epoch": 2.6575459598118853,
+      "grad_norm": 0.17861344190439962,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0151,
+      "step": 777
+    },
+    {
+      "epoch": 2.6609662248824284,
+      "grad_norm": 0.16730938712809765,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0145,
+      "step": 778
+    },
+    {
+      "epoch": 2.6643864899529714,
+      "grad_norm": 0.1775818865748014,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0159,
+      "step": 779
+    },
+    {
+      "epoch": 2.6678067550235145,
+      "grad_norm": 0.191333871334205,
+      "learning_rate": 1.41e-05,
+      "loss": 0.013,
+      "step": 780
+    },
+    {
+      "epoch": 2.671227020094057,
+      "grad_norm": 0.16907634204579375,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0145,
+      "step": 781
+    },
+    {
+      "epoch": 2.6746472851646,
+      "grad_norm": 0.18178434604030544,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0135,
+      "step": 782
+    },
+    {
+      "epoch": 2.678067550235143,
+      "grad_norm": 0.20503354023831713,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 783
+    },
+    {
+      "epoch": 2.6814878153056863,
+      "grad_norm": 0.16391814776977276,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 784
+    },
+    {
+      "epoch": 2.6849080803762293,
+      "grad_norm": 0.1839989787303562,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 785
+    },
+    {
+      "epoch": 2.688328345446772,
+      "grad_norm": 0.19236560848567033,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 786
+    },
+    {
+      "epoch": 2.691748610517315,
+      "grad_norm": 0.16811298067940522,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 787
+    },
+    {
+      "epoch": 2.695168875587858,
+      "grad_norm": 0.19975209346986805,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0159,
+      "step": 788
+    },
+    {
+      "epoch": 2.698589140658401,
+      "grad_norm": 0.16338526349657181,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0133,
+      "step": 789
+    },
+    {
+      "epoch": 2.702009405728944,
+      "grad_norm": 0.17351362318047445,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0159,
+      "step": 790
+    },
+    {
+      "epoch": 2.7054296707994867,
+      "grad_norm": 0.17661074926464723,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 791
+    },
+    {
+      "epoch": 2.70884993587003,
+      "grad_norm": 0.1762002497021315,
+      "learning_rate": 1.41e-05,
+      "loss": 0.016,
+      "step": 792
+    },
+    {
+      "epoch": 2.712270200940573,
+      "grad_norm": 0.19710675588819915,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0161,
+      "step": 793
+    },
+    {
+      "epoch": 2.715690466011116,
+      "grad_norm": 0.1692117872805568,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 794
+    },
+    {
+      "epoch": 2.719110731081659,
+      "grad_norm": 0.20731978857327257,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 795
+    },
+    {
+      "epoch": 2.722530996152202,
+      "grad_norm": 0.17806201725432017,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0147,
+      "step": 796
+    },
+    {
+      "epoch": 2.725951261222745,
+      "grad_norm": 0.16617443428776074,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 797
+    },
+    {
+      "epoch": 2.7293715262932876,
+      "grad_norm": 0.20059647446620907,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 798
+    },
+    {
+      "epoch": 2.7327917913638307,
+      "grad_norm": 0.1773142221832559,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0171,
+      "step": 799
+    },
+    {
+      "epoch": 2.7362120564343737,
+      "grad_norm": 0.1979846553461278,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 800
+    },
+    {
+      "epoch": 2.7396323215049168,
+      "grad_norm": 0.18511763263524342,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 801
+    },
+    {
+      "epoch": 2.74305258657546,
+      "grad_norm": 0.17699099585761474,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0118,
+      "step": 802
+    },
+    {
+      "epoch": 2.7464728516460024,
+      "grad_norm": 0.19352657405112964,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0163,
+      "step": 803
+    },
+    {
+      "epoch": 2.7498931167165455,
+      "grad_norm": 0.19135035210758095,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 804
+    },
+    {
+      "epoch": 2.7533133817870885,
+      "grad_norm": 0.1841148474869883,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 805
+    },
+    {
+      "epoch": 2.7567336468576316,
+      "grad_norm": 0.19008317628622337,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0152,
+      "step": 806
+    },
+    {
+      "epoch": 2.7601539119281746,
+      "grad_norm": 0.1949140281276627,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 807
+    },
+    {
+      "epoch": 2.763574176998717,
+      "grad_norm": 0.1944281661049677,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0145,
+      "step": 808
+    },
+    {
+      "epoch": 2.7669944420692603,
+      "grad_norm": 0.1674859606182937,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0162,
+      "step": 809
+    },
+    {
+      "epoch": 2.7704147071398033,
+      "grad_norm": 0.18850185666818006,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 810
+    },
+    {
+      "epoch": 2.7738349722103464,
+      "grad_norm": 0.18475272615726215,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0156,
+      "step": 811
+    },
+    {
+      "epoch": 2.7772552372808894,
+      "grad_norm": 0.15365265630124134,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0134,
+      "step": 812
+    },
+    {
+      "epoch": 2.780675502351432,
+      "grad_norm": 0.16024686744719197,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 813
+    },
+    {
+      "epoch": 2.784095767421975,
+      "grad_norm": 0.18228181650312342,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0142,
+      "step": 814
+    },
+    {
+      "epoch": 2.787516032492518,
+      "grad_norm": 0.15232575479299343,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 815
+    },
+    {
+      "epoch": 2.790936297563061,
+      "grad_norm": 0.19473439330433306,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 816
+    },
+    {
+      "epoch": 2.794356562633604,
+      "grad_norm": 0.20524717935488615,
+      "learning_rate": 1.41e-05,
+      "loss": 0.017,
+      "step": 817
+    },
+    {
+      "epoch": 2.797776827704147,
+      "grad_norm": 0.1908817371704934,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0147,
+      "step": 818
+    },
+    {
+      "epoch": 2.8011970927746903,
+      "grad_norm": 0.1817449147111606,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 819
+    },
+    {
+      "epoch": 2.804617357845233,
+      "grad_norm": 0.20849626412897376,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 820
+    },
+    {
+      "epoch": 2.808037622915776,
+      "grad_norm": 0.17439115442992892,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0165,
+      "step": 821
+    },
+    {
+      "epoch": 2.811457887986319,
+      "grad_norm": 0.184212269622675,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 822
+    },
+    {
+      "epoch": 2.814878153056862,
+      "grad_norm": 0.1742874984626096,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 823
+    },
+    {
+      "epoch": 2.818298418127405,
+      "grad_norm": 0.16854467961459083,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0145,
+      "step": 824
+    },
+    {
+      "epoch": 2.8217186831979477,
+      "grad_norm": 0.17468169666793745,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0149,
+      "step": 825
+    },
+    {
+      "epoch": 2.8251389482684908,
+      "grad_norm": 0.16936997795252084,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 826
+    },
+    {
+      "epoch": 2.828559213339034,
+      "grad_norm": 0.16572236927644424,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 827
+    },
+    {
+      "epoch": 2.831979478409577,
+      "grad_norm": 0.1651962024013233,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0127,
+      "step": 828
+    },
+    {
+      "epoch": 2.83539974348012,
+      "grad_norm": 0.1657376085150308,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0139,
+      "step": 829
+    },
+    {
+      "epoch": 2.8388200085506625,
+      "grad_norm": 0.17582133731054114,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0131,
+      "step": 830
+    },
+    {
+      "epoch": 2.8422402736212056,
+      "grad_norm": 0.20723879555852162,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0162,
+      "step": 831
+    },
+    {
+      "epoch": 2.8456605386917486,
+      "grad_norm": 0.6025905984134068,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0142,
+      "step": 832
+    },
+    {
+      "epoch": 2.8490808037622917,
+      "grad_norm": 0.1678623970594971,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0143,
+      "step": 833
+    },
+    {
+      "epoch": 2.8525010688328347,
+      "grad_norm": 0.21086844483510658,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0158,
+      "step": 834
+    },
+    {
+      "epoch": 2.8559213339033773,
+      "grad_norm": 0.18493480224421863,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0126,
+      "step": 835
+    },
+    {
+      "epoch": 2.8593415989739204,
+      "grad_norm": 0.1904246277238966,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 836
+    },
+    {
+      "epoch": 2.8627618640444634,
+      "grad_norm": 0.18774356330645445,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0174,
+      "step": 837
+    },
+    {
+      "epoch": 2.8661821291150065,
+      "grad_norm": 0.20481371594917794,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0132,
+      "step": 838
+    },
+    {
+      "epoch": 2.8696023941855495,
+      "grad_norm": 0.19259669092688833,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 839
+    },
+    {
+      "epoch": 2.873022659256092,
+      "grad_norm": 0.20161241494039436,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0159,
+      "step": 840
+    },
+    {
+      "epoch": 2.876442924326635,
+      "grad_norm": 0.19885243582338097,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 841
+    },
+    {
+      "epoch": 2.8798631893971782,
+      "grad_norm": 0.18089297372993837,
+      "learning_rate": 1.41e-05,
+      "loss": 0.015,
+      "step": 842
+    },
+    {
+      "epoch": 2.8832834544677213,
+      "grad_norm": 0.19064879861955514,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0132,
+      "step": 843
+    },
+    {
+      "epoch": 2.8867037195382643,
+      "grad_norm": 0.18907692558323141,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0146,
+      "step": 844
+    },
+    {
+      "epoch": 2.890123984608807,
+      "grad_norm": 0.19585899032623924,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0153,
+      "step": 845
+    },
+    {
+      "epoch": 2.8935442496793504,
+      "grad_norm": 0.22806745672139891,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0169,
+      "step": 846
+    },
+    {
+      "epoch": 2.896964514749893,
+      "grad_norm": 0.19336051551290398,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 847
+    },
+    {
+      "epoch": 2.900384779820436,
+      "grad_norm": 0.17663309403974298,
+      "learning_rate": 1.41e-05,
+      "loss": 0.013,
+      "step": 848
+    },
+    {
+      "epoch": 2.903805044890979,
+      "grad_norm": 0.1884422733573296,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0153,
+      "step": 849
+    },
+    {
+      "epoch": 2.907225309961522,
+      "grad_norm": 0.19574362737358483,
+      "learning_rate": 1.41e-05,
+      "loss": 0.016,
+      "step": 850
+    },
+    {
+      "epoch": 2.9106455750320652,
+      "grad_norm": 0.16218777133767662,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0135,
+      "step": 851
+    },
+    {
+      "epoch": 2.914065840102608,
+      "grad_norm": 0.1844053227394077,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0147,
+      "step": 852
+    },
+    {
+      "epoch": 2.917486105173151,
+      "grad_norm": 0.19292851510616071,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0149,
+      "step": 853
+    },
+    {
+      "epoch": 2.920906370243694,
+      "grad_norm": 0.2089035835617463,
+      "learning_rate": 1.41e-05,
+      "loss": 0.017,
+      "step": 854
+    },
+    {
+      "epoch": 2.924326635314237,
+      "grad_norm": 0.7055863227979324,
+      "learning_rate": 1.41e-05,
+      "loss": 0.016,
+      "step": 855
+    },
+    {
+      "epoch": 2.92774690038478,
+      "grad_norm": 0.18518793131577801,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 856
+    },
+    {
+      "epoch": 2.9311671654553226,
+      "grad_norm": 0.17688744692401304,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 857
+    },
+    {
+      "epoch": 2.9345874305258657,
+      "grad_norm": 0.20202970506605689,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0144,
+      "step": 858
+    },
+    {
+      "epoch": 2.9380076955964087,
+      "grad_norm": 0.19190788298961772,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0154,
+      "step": 859
+    },
+    {
+      "epoch": 2.941427960666952,
+      "grad_norm": 0.19843345575867563,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0167,
+      "step": 860
+    },
+    {
+      "epoch": 2.944848225737495,
+      "grad_norm": 0.18043514187233164,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0124,
+      "step": 861
+    },
+    {
+      "epoch": 2.9482684908080374,
+      "grad_norm": 0.1852741160641503,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0153,
+      "step": 862
+    },
+    {
+      "epoch": 2.9516887558785805,
+      "grad_norm": 0.19552493492009015,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0161,
+      "step": 863
+    },
+    {
+      "epoch": 2.9551090209491235,
+      "grad_norm": 0.1972068517674688,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0151,
+      "step": 864
+    },
+    {
+      "epoch": 2.9585292860196666,
+      "grad_norm": 0.2016593632101422,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0161,
+      "step": 865
+    },
+    {
+      "epoch": 2.9619495510902096,
+      "grad_norm": 0.18720575761526204,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 866
+    },
+    {
+      "epoch": 2.9653698161607522,
+      "grad_norm": 0.1948235931544798,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0142,
+      "step": 867
+    },
+    {
+      "epoch": 2.9687900812312953,
+      "grad_norm": 0.18580729685744493,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 868
+    },
+    {
+      "epoch": 2.9722103463018383,
+      "grad_norm": 0.1747215712068137,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0148,
+      "step": 869
+    },
+    {
+      "epoch": 2.9756306113723814,
+      "grad_norm": 0.21255162912353637,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0149,
+      "step": 870
+    },
+    {
+      "epoch": 2.9790508764429244,
+      "grad_norm": 0.18848472150934822,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0141,
+      "step": 871
+    },
+    {
+      "epoch": 2.982471141513467,
+      "grad_norm": 0.18016163548131123,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0147,
+      "step": 872
+    },
+    {
+      "epoch": 2.9858914065840105,
+      "grad_norm": 0.1795705522164676,
+      "learning_rate": 1.41e-05,
+      "loss": 0.014,
+      "step": 873
+    },
+    {
+      "epoch": 2.989311671654553,
+      "grad_norm": 0.17401025878895968,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0138,
+      "step": 874
+    },
+    {
+      "epoch": 2.992731936725096,
+      "grad_norm": 0.1783730795841426,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0132,
+      "step": 875
+    },
+    {
+      "epoch": 2.9961522017956392,
+      "grad_norm": 0.20051631261532152,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0166,
+      "step": 876
+    },
+    {
+      "epoch": 2.9995724668661823,
+      "grad_norm": 0.2056542584608787,
+      "learning_rate": 1.41e-05,
+      "loss": 0.0155,
+      "step": 877
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1460,
+  "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": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 367251916062720.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}